gcc/libcpp/files.cc

2616 lines
73 KiB
C++
Raw Normal View History

Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Part of CPP library. File handling.
2024-01-03 12:19:35 +01:00
Copyright (C) 1986-2024 Free Software Foundation, Inc.
[multiple changes] 1998-10-28 16:10 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * c-lang.c: Declare extern char *yy_cur if USE_CPPLIB. (lang_init): Call check_newline always. * c-lex.c (init_parse) [USE_CPPLIB=1]: After calling cpp_start_read, set yy_cur and yy_lim to read from parse_in.token_buffer, so that we'll see the first #line directive. * cpplib.c (cpp_start_read): finclude the main input file before processing -include/-imacros. Process -imacros and -include separately, and handle -include by stacking a buffer for the file in question as if it'd been #included. * toplev.c (documented_lang_options) Recognize -H when USE_CPPLIB is on. 1998-10-28 16:09 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * cpplib.c: Merge do_once into do_pragma. Break file handling code out of do_include. Move append_include_chain, deps_output, file_cleanup, redundant_include_p, import_hash, lookup_import, add_import, read_filename_string, read_name_map, open_include_file, finclude, safe_read to cppfiles.c. Move prototypes for deps_output, append_include_chain, finclude to cpplib.h. Move definition of struct file_name_list there also. * cppfiles.c: New file. Contains all the above functions broken out of cpplib.c; also hack_vms_include_specification from cccp.c and find_include_file, a new function broken out of do_include. * Makefile.in (cppmain): Depend on cppfiles.o. (fix-header): Likewise. (cppfiles.o): New target. * configure.in (--enable-c-cpplib): Add cppfiles.o to extra_c_objs. Add ../cppfiles.o to extra_cxx_objs. From-SVN: r23424
1998-10-29 11:54:13 +00:00
Written by Per Bothner, 1994.
Based on CCCP program by Paul Rubin, June 1986
Adapted to ANSI C, Richard Stallman, Jan 1987
Split out of cpplib.c, Zack Weinberg, Oct 1998
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
Reimplemented, Neil Booth, Jul 2003
[multiple changes] 1998-10-28 16:10 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * c-lang.c: Declare extern char *yy_cur if USE_CPPLIB. (lang_init): Call check_newline always. * c-lex.c (init_parse) [USE_CPPLIB=1]: After calling cpp_start_read, set yy_cur and yy_lim to read from parse_in.token_buffer, so that we'll see the first #line directive. * cpplib.c (cpp_start_read): finclude the main input file before processing -include/-imacros. Process -imacros and -include separately, and handle -include by stacking a buffer for the file in question as if it'd been #included. * toplev.c (documented_lang_options) Recognize -H when USE_CPPLIB is on. 1998-10-28 16:09 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * cpplib.c: Merge do_once into do_pragma. Break file handling code out of do_include. Move append_include_chain, deps_output, file_cleanup, redundant_include_p, import_hash, lookup_import, add_import, read_filename_string, read_name_map, open_include_file, finclude, safe_read to cppfiles.c. Move prototypes for deps_output, append_include_chain, finclude to cpplib.h. Move definition of struct file_name_list there also. * cppfiles.c: New file. Contains all the above functions broken out of cpplib.c; also hack_vms_include_specification from cccp.c and find_include_file, a new function broken out of do_include. * Makefile.in (cppmain): Depend on cppfiles.o. (fix-header): Likewise. (cppfiles.o): New target. * configure.in (--enable-c-cpplib): Add cppfiles.o to extra_c_objs. Add ../cppfiles.o to extra_cxx_objs. From-SVN: r23424
1998-10-29 11:54:13 +00:00
This program is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
Free Software Foundation; either version 3, or (at your option) any
[multiple changes] 1998-10-28 16:10 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * c-lang.c: Declare extern char *yy_cur if USE_CPPLIB. (lang_init): Call check_newline always. * c-lex.c (init_parse) [USE_CPPLIB=1]: After calling cpp_start_read, set yy_cur and yy_lim to read from parse_in.token_buffer, so that we'll see the first #line directive. * cpplib.c (cpp_start_read): finclude the main input file before processing -include/-imacros. Process -imacros and -include separately, and handle -include by stacking a buffer for the file in question as if it'd been #included. * toplev.c (documented_lang_options) Recognize -H when USE_CPPLIB is on. 1998-10-28 16:09 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * cpplib.c: Merge do_once into do_pragma. Break file handling code out of do_include. Move append_include_chain, deps_output, file_cleanup, redundant_include_p, import_hash, lookup_import, add_import, read_filename_string, read_name_map, open_include_file, finclude, safe_read to cppfiles.c. Move prototypes for deps_output, append_include_chain, finclude to cpplib.h. Move definition of struct file_name_list there also. * cppfiles.c: New file. Contains all the above functions broken out of cpplib.c; also hack_vms_include_specification from cccp.c and find_include_file, a new function broken out of do_include. * Makefile.in (cppmain): Depend on cppfiles.o. (fix-header): Likewise. (cppfiles.o): New target. * configure.in (--enable-c-cpplib): Add cppfiles.o to extra_c_objs. Add ../cppfiles.o to extra_cxx_objs. From-SVN: r23424
1998-10-29 11:54:13 +00:00
later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; see the file COPYING3. If not see
<http://www.gnu.org/licenses/>. */
[multiple changes] 1998-10-28 16:10 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * c-lang.c: Declare extern char *yy_cur if USE_CPPLIB. (lang_init): Call check_newline always. * c-lex.c (init_parse) [USE_CPPLIB=1]: After calling cpp_start_read, set yy_cur and yy_lim to read from parse_in.token_buffer, so that we'll see the first #line directive. * cpplib.c (cpp_start_read): finclude the main input file before processing -include/-imacros. Process -imacros and -include separately, and handle -include by stacking a buffer for the file in question as if it'd been #included. * toplev.c (documented_lang_options) Recognize -H when USE_CPPLIB is on. 1998-10-28 16:09 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * cpplib.c: Merge do_once into do_pragma. Break file handling code out of do_include. Move append_include_chain, deps_output, file_cleanup, redundant_include_p, import_hash, lookup_import, add_import, read_filename_string, read_name_map, open_include_file, finclude, safe_read to cppfiles.c. Move prototypes for deps_output, append_include_chain, finclude to cpplib.h. Move definition of struct file_name_list there also. * cppfiles.c: New file. Contains all the above functions broken out of cpplib.c; also hack_vms_include_specification from cccp.c and find_include_file, a new function broken out of do_include. * Makefile.in (cppmain): Depend on cppfiles.o. (fix-header): Likewise. (cppfiles.o): New target. * configure.in (--enable-c-cpplib): Add cppfiles.o to extra_c_objs. Add ../cppfiles.o to extra_cxx_objs. From-SVN: r23424
1998-10-29 11:54:13 +00:00
#include "config.h"
#include "system.h"
#include "cpplib.h"
Makefile.def (host_modules): add libcpp. ChangeLog: 2004-05-23 Paolo Bonzini <bonzini@gnu.org> * Makefile.def (host_modules): add libcpp. * Makefile.tpl: Add dependencies on and for libcpp. * Makefile.in: Regenerate. * configure.in: Add libcpp host module. * configure: Regenerate. config/ChangeLog: 2004-05-23 Paolo Bonzini <bonzini@gnu.org> * acx.m4 (ACX_HEADER_STDBOOL, ACX_HEADER_STRING): From gcc. gcc/ChangeLog: 2004-05-23 Paolo Bonzini <bonzini@gnu.org> Move libcpp to the toplevel. * Makefile.in: Remove references to libcpp files, use CPPLIBS instead of libcpp.a. Define SYMTAB_H and change hashtable.h to that. * aclocal.m4 (gcc_AC_HEADER_STDBOOL, gcc_AC_HEADER_STRING, gcc_AC_C__BOOL): Remove. * configure.ac (gcc_AC_C__BOOL, HAVE_UCHAR): Remove tests. * configure: Regenerate. * config.in: Regenerate. * c-ppoutput.c: Include ../libcpp/internal.h instead of cpphash.h. * cppcharset.c: Removed. * cpperror.c: Removed. * cppexp.c: Removed. * cppfiles.c: Removed. * cpphash.c: Removed. * cpphash.h: Removed. * cppinit.c: Removed. * cpplex.c: Removed. * cpplib.c: Removed. * cpplib.h: Removed. * cppmacro.c: Removed. * cpppch.c: Removed. * cpptrad.c: Removed. * cppucnid.h: Removed. * cppucnid.pl: Removed. * cppucnid.tab: Removed. * hashtable.c: Removed. * hashtable.h: Removed. * line-map.c: Removed. * line-map.h: Removed. * mkdeps.c: Removed. * mkdeps.h: Removed. * stringpool.h: Include symtab.h instead of hashtable.h. * tree.h: Include symtab.h instead of hashtable.h. * system.h (O_NONBLOCK, O_NOCTTY): Do not define. gcc/cp/ChangeLog: 2004-05-23 Paolo Bonzini <bonzini@gnu.org> * Make-lang.in: No need to specify $(LIBCPP). gcc/java/ChangeLog: 2004-05-23 Paolo Bonzini <bonzini@gnu.org> * Make-lang.in: Link in $(LIBCPP) instead of mkdeps.o. libcpp/ChangeLog: 2004-05-23 Paolo Bonzini <bonzini@gnu.org> Moved libcpp from the gcc subdirectory to the toplevel. * Makefile.am: New file. * Makefile.in: Regenerate. * configure.ac: New file. * configure: Regenerate. * config.in: Regenerate. * charset.c: Moved from gcc/cppcharset.c. Add note about brokenness of input charset detection. Adjust for change in name of cppucnid.h. * errors.c: Moved from gcc/cpperror.c. Do not include intl.h. * expr.c: Moved from gcc/cppexp.c. * files.c: Moved from gcc/cppfiles.c. Do not include intl.h. Remove #define of O_BINARY, it is in system.h. * identifiers.c: Moved from gcc/cpphash.c. * internal.h: Moved from gcc/cpphash.h. Change header guard name. All other files adjusted to match name change. * init.c: Moved from gcc/cppinit.c. (init_library) [ENABLE_NLS]: Call bindtextdomain. * lex.c: Moved from gcc/cpplex.c. * directives.c: Moved from gcc/cpplib.c. * macro.c: Moved from gcc/cppmacro.c. * pch.c: Moved from gcc/cpppch.c. Do not include intl.h. * traditional.c: Moved from gcc/cpptrad.c. * ucnid.h: Moved from gcc/cppucnid.h. Change header guard name. * ucnid.pl: Moved from gcc/cppucnid.pl. * ucnid.tab: Moved from gcc/cppucnid.tab. Change header guard name. * symtab.c: Moved from gcc/hashtable.c. * line-map.c: Moved from gcc. Do not include intl.h. * mkdeps.c: Moved from gcc. * system.h: New file. libcpp/include/ChangeLog: 2004-05-23 Paolo Bonzini <bonzini@gnu.org> * cpplib.h: Moved from gcc. Change header guard name. * line-map.h: Moved from gcc. Change header guard name. * mkdeps.h: Moved from gcc. Change header guard name. * symtab.h: Moved from gcc/hashtable.h. Change header guard name. libcpp/po/ChangeLog: 2004-05-23 Paolo Bonzini <bonzini@gnu.org> * be.po: Extracted from gcc/po/be.po. * ca.po: Extracted from gcc/po/ca.po. * da.po: Extracted from gcc/po/da.po. * de.po: Extracted from gcc/po/de.po. * el.po: Extracted from gcc/po/el.po. * es.po: Extracted from gcc/po/es.po. * fr.po: Extracted from gcc/po/fr.po. * ja.po: Extracted from gcc/po/ja.po. * nl.po: Extracted from gcc/po/nl.po. * sv.po: Extracted from gcc/po/sv.po. * tr.po: Extracted from gcc/po/tr.po. From-SVN: r82199
2004-05-24 10:50:45 +00:00
#include "internal.h"
cppfiles.c: Include mkdeps.h. * cppfiles.c: Include mkdeps.h. (find_include_file, read_include_file): Remove _cpp_ prefix from name, make static. (_cpp_execute_include): New function, broken out of do_include. * cpplib.c: Don't include mkdeps.h. (struct directive): Remove type field. Reorder entries. The function takes only one argument. (struct if_stack): Make type field an int. (directive_table): Rename to dtable. Generate it, the prototypes of the directive handlers, and the enum for the directive numbers, from a template macro. (do_ifndef, do_include_next, do_import): New functions. (do_define, do_include, do_endif, do_ifdef, do_if, do_else, do_undef, do_line, do_elif, do_error, do_pragma, do_warning, do_ident, do_assert, do_unassert, do_sccs): Take only one argument. (do_sccs): Define always, but alter behavior based on SCCS_DIRECTIVE. (_cpp_handle_directive, consider_directive_while_skipping): Restructure for new directive table layout. (pass_thru_directive): Take a directive number, not a pointer to a struct directive. (parse_include): New function, broken out of do_include. (do_include, do_import, do_include_next): Use parse_include and _cpp_execute_include. (do_elif, do_else): Test for T_ELSE specifically when checking for #elif/#else after #else. (parse_ifdef): New function, broken out of do_ifdef. (validate_else): Expect a name arg without a leading #. (if_directive_name): Delete. (cpp_define, cpp_assert, cpp_undef, cpp_unassert): Call directive handlers with only one argument. * cpphash.h: Update prototypes. (enum node_type): Remove entries for directives. * Makefile.in: Update dependencies. From-SVN: r32519
2000-03-14 06:34:11 +00:00
#include "mkdeps.h"
#include "obstack.h"
#include "hashtab.h"
#include "md5.h"
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
#include <dirent.h>
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Variable length record files on VMS will have a stat size that includes
record control characters that won't be included in the read size. */
#ifdef VMS
# define FAB_C_VAR 2 /* variable length records (see Starlet fabdef.h) */
# define STAT_SIZE_RELIABLE(ST) ((ST).st_fab_rfm != FAB_C_VAR)
#else
# define STAT_SIZE_RELIABLE(ST) true
#endif
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
#ifdef __DJGPP__
#include <io.h>
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* For DJGPP redirected input is opened in text mode. */
# define set_stdin_to_binary_mode() \
if (! isatty (0)) setmode (0, O_BINARY)
#else
# define set_stdin_to_binary_mode() /* Nothing */
#endif
/* This structure represents a file searched for by CPP, whether it
exists or not. An instance may be pointed to by more than one
cpp_file_hash_entry; at present no reference count is kept. */
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
struct _cpp_file
{
/* Filename as given to #include or command line switch. */
const char *name;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* The full path used to find the file. */
const char *path;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* The full path of the pch file. */
const char *pchname;
[multiple changes] 1998-10-28 16:10 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * c-lang.c: Declare extern char *yy_cur if USE_CPPLIB. (lang_init): Call check_newline always. * c-lex.c (init_parse) [USE_CPPLIB=1]: After calling cpp_start_read, set yy_cur and yy_lim to read from parse_in.token_buffer, so that we'll see the first #line directive. * cpplib.c (cpp_start_read): finclude the main input file before processing -include/-imacros. Process -imacros and -include separately, and handle -include by stacking a buffer for the file in question as if it'd been #included. * toplev.c (documented_lang_options) Recognize -H when USE_CPPLIB is on. 1998-10-28 16:09 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * cpplib.c: Merge do_once into do_pragma. Break file handling code out of do_include. Move append_include_chain, deps_output, file_cleanup, redundant_include_p, import_hash, lookup_import, add_import, read_filename_string, read_name_map, open_include_file, finclude, safe_read to cppfiles.c. Move prototypes for deps_output, append_include_chain, finclude to cpplib.h. Move definition of struct file_name_list there also. * cppfiles.c: New file. Contains all the above functions broken out of cpplib.c; also hack_vms_include_specification from cccp.c and find_include_file, a new function broken out of do_include. * Makefile.in (cppmain): Depend on cppfiles.o. (fix-header): Likewise. (cppfiles.o): New target. * configure.in (--enable-c-cpplib): Add cppfiles.o to extra_c_objs. Add ../cppfiles.o to extra_cxx_objs. From-SVN: r23424
1998-10-29 11:54:13 +00:00
/* The file's path with the basename stripped. NULL if it hasn't
been calculated yet. */
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
const char *dir_name;
cpplex.c: Don't include sys/mman.h. toplevel: * cpplex.c: Don't include sys/mman.h. (cpp_push_buffer, cpp_pop_buffer): Moved to cpplib.c. * cpplib.c: Include sys/mman.h and obstack.h. (cpp_push_buffer): Moved from cpplex.c; allocate buffers on an obstack. (cpp_pop_buffer): Moved from cpplex.c; free buffers from an obstack. (_cpp_unwind_if_stack): Now static, unwind_if_stack. Don't bother freeing if stack entries (they will be freed with their buffer). (do_endif): Free if stack entries from the buffer obstack. (push_conditional): Allocate if stack entries from the buffer obstack. (find_answer): Rename to _cpp_find_answer. (do_assert, do_unassert): Update. * cpphash.h: Update prototypes. (xobnew): New convenience macro. * cpplib.h (struct cpp_reader): Add hash_ob and buffer_ob fields. Update comments. (struct cpp_hashnode): Remove disabled field. * cppinit.c: Don't include hashtab.h or splay-tree.h. (report_missing_guard): Moved to cppfiles.c. (cpp_reader_init): Call cpp_init_stacks, cpp_init_macros, cpp_init_includes. (cpp_cleanup): Call cpp_cleanup_stacks, cpp_cleanup_macros, cpp_cleanup_includes. Don't destroy hashtab or all_include_files here. (cpp_finish): Use _cpp_report_missing_guards. * cppfiles.c (report_missing_guard): Moved from cppinit.c. (_cpp_init_include_table): Rename _cpp_init_includes. (_cpp_cleanup_includes, _cpp_report_missing_guards): New. * cppexp.c (parse_assertion): Update for new name of find_answer. * Makefile.in (cpplib.o, cpphash.o, cppinit.o): Update deps. * cpplib.c (do_ident): s/VSPACE/EOF/ testsuite: * gcc.dg/cpp/ident.c: New test. From-SVN: r34870
2000-07-05 05:33:57 +00:00
/* Chain through all files. */
struct _cpp_file *next_file;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* The contents of NAME after calling read_file(). */
const uchar *buffer;
/* Pointer to the real start of BUFFER. read_file() might increment
BUFFER; when freeing, this this pointer must be used instead. */
const uchar *buffer_start;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* The macro, if any, preventing re-inclusion. */
const cpp_hashnode *cmacro;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* The directory in the search path where FILE was found. Used for
#include_next and determining whether a header is a system
header. */
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
cpp_dir *dir;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* As filled in by stat(2) for the file. */
struct stat st;
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
/* Size for #embed, perhaps smaller than st.st_size. */
size_t limit;
/* Offset for #embed. */
off_t offset;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* File descriptor. Invalid if -1, otherwise open. */
int fd;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Zero if this file was successfully opened and stat()-ed,
otherwise errno obtained from failure. */
int err_no;
[multiple changes] 1998-10-28 16:10 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * c-lang.c: Declare extern char *yy_cur if USE_CPPLIB. (lang_init): Call check_newline always. * c-lex.c (init_parse) [USE_CPPLIB=1]: After calling cpp_start_read, set yy_cur and yy_lim to read from parse_in.token_buffer, so that we'll see the first #line directive. * cpplib.c (cpp_start_read): finclude the main input file before processing -include/-imacros. Process -imacros and -include separately, and handle -include by stacking a buffer for the file in question as if it'd been #included. * toplev.c (documented_lang_options) Recognize -H when USE_CPPLIB is on. 1998-10-28 16:09 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * cpplib.c: Merge do_once into do_pragma. Break file handling code out of do_include. Move append_include_chain, deps_output, file_cleanup, redundant_include_p, import_hash, lookup_import, add_import, read_filename_string, read_name_map, open_include_file, finclude, safe_read to cppfiles.c. Move prototypes for deps_output, append_include_chain, finclude to cpplib.h. Move definition of struct file_name_list there also. * cppfiles.c: New file. Contains all the above functions broken out of cpplib.c; also hack_vms_include_specification from cccp.c and find_include_file, a new function broken out of do_include. * Makefile.in (cppmain): Depend on cppfiles.o. (fix-header): Likewise. (cppfiles.o): New target. * configure.in (--enable-c-cpplib): Add cppfiles.o to extra_c_objs. Add ../cppfiles.o to extra_cxx_objs. From-SVN: r23424
1998-10-29 11:54:13 +00:00
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Number of times the file has been stacked for preprocessing. */
unsigned short stack_count;
/* If opened with #import or contains #pragma once. */
bool once_only : 1;
[multiple changes] 1998-10-28 16:10 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * c-lang.c: Declare extern char *yy_cur if USE_CPPLIB. (lang_init): Call check_newline always. * c-lex.c (init_parse) [USE_CPPLIB=1]: After calling cpp_start_read, set yy_cur and yy_lim to read from parse_in.token_buffer, so that we'll see the first #line directive. * cpplib.c (cpp_start_read): finclude the main input file before processing -include/-imacros. Process -imacros and -include separately, and handle -include by stacking a buffer for the file in question as if it'd been #included. * toplev.c (documented_lang_options) Recognize -H when USE_CPPLIB is on. 1998-10-28 16:09 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * cpplib.c: Merge do_once into do_pragma. Break file handling code out of do_include. Move append_include_chain, deps_output, file_cleanup, redundant_include_p, import_hash, lookup_import, add_import, read_filename_string, read_name_map, open_include_file, finclude, safe_read to cppfiles.c. Move prototypes for deps_output, append_include_chain, finclude to cpplib.h. Move definition of struct file_name_list there also. * cppfiles.c: New file. Contains all the above functions broken out of cpplib.c; also hack_vms_include_specification from cccp.c and find_include_file, a new function broken out of do_include. * Makefile.in (cppmain): Depend on cppfiles.o. (fix-header): Likewise. (cppfiles.o): New target. * configure.in (--enable-c-cpplib): Add cppfiles.o to extra_c_objs. Add ../cppfiles.o to extra_cxx_objs. From-SVN: r23424
1998-10-29 11:54:13 +00:00
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* If read() failed before. */
bool dont_read : 1;
[multiple changes] 1998-10-28 16:10 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * c-lang.c: Declare extern char *yy_cur if USE_CPPLIB. (lang_init): Call check_newline always. * c-lex.c (init_parse) [USE_CPPLIB=1]: After calling cpp_start_read, set yy_cur and yy_lim to read from parse_in.token_buffer, so that we'll see the first #line directive. * cpplib.c (cpp_start_read): finclude the main input file before processing -include/-imacros. Process -imacros and -include separately, and handle -include by stacking a buffer for the file in question as if it'd been #included. * toplev.c (documented_lang_options) Recognize -H when USE_CPPLIB is on. 1998-10-28 16:09 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * cpplib.c: Merge do_once into do_pragma. Break file handling code out of do_include. Move append_include_chain, deps_output, file_cleanup, redundant_include_p, import_hash, lookup_import, add_import, read_filename_string, read_name_map, open_include_file, finclude, safe_read to cppfiles.c. Move prototypes for deps_output, append_include_chain, finclude to cpplib.h. Move definition of struct file_name_list there also. * cppfiles.c: New file. Contains all the above functions broken out of cpplib.c; also hack_vms_include_specification from cccp.c and find_include_file, a new function broken out of do_include. * Makefile.in (cppmain): Depend on cppfiles.o. (fix-header): Likewise. (cppfiles.o): New target. * configure.in (--enable-c-cpplib): Add cppfiles.o to extra_c_objs. Add ../cppfiles.o to extra_cxx_objs. From-SVN: r23424
1998-10-29 11:54:13 +00:00
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* If BUFFER above contains the true contents of the file. */
bool buffer_valid : 1;
gcc: * config.gcc (*-*-linux* | frv-*-*linux* | *-*-kfreebsd*-gnu | *-*-knetbsd*-gnu | *-*-gnu* | *-*-kopensolaris*-gnu): Use glibc-c.o in c_target_objs and cxx_target_objs. Use t-glibc in tmake_file. Set target_has_targetcm. (tilegx-*-linux*, tilepro-*-linux*): Append to c_target_objs and cxx_target_objs rather than overriding previous value. * config/glibc-c.c, config/t-glibc: New. * doc/tm.texi.in (TARGET_C_PREINCLUDE): New @hook. * doc/tm.texi: Regenerate. * hooks.c (hook_constcharptr_void_null): New. * hooks.h (hook_constcharptr_void_null): Declare. gcc/c-family: * c-common.h (pch_cpp_save_state): Declare. * c-target.def (c_preinclude): New hook. * c-opts.c (done_preinclude): New. (push_command_line_include): Handle default preincluded header. (cb_file_change): Call pch_cpp_save_state when calling push_command_line_include. * c-pch.c (pch_ready_to_save_cpp_state, pch_cpp_state_saved) (pch_cpp_save_state): New. (pch_init): Call pch_cpp_save_state conditionally, instead of calling cpp_save_state. gcc/testsuite: * gcc.dg/c99-predef-1.c: New test. * gcc.dg/cpp/cmdlne-dU-1.c, gcc.dg/cpp/cmdlne-dU-2.c, gcc.dg/cpp/cmdlne-dU-3.c, gcc.dg/cpp/cmdlne-dU-4.c, gcc.dg/cpp/cmdlne-dU-5.c, gcc.dg/cpp/cmdlne-dU-6.c, gcc.dg/cpp/cmdlne-dU-7.c, gcc.dg/cpp/cmdlne-dU-8.c, gcc.dg/cpp/cmdlne-dU-9.c, gcc.dg/cpp/cmdlne-dU-10.c, gcc.dg/cpp/cmdlne-dU-11.c, gcc.dg/cpp/cmdlne-dU-12.c, gcc.dg/cpp/cmdlne-dU-13.c, gcc.dg/cpp/cmdlne-dU-14.c, gcc.dg/cpp/cmdlne-dU-15.c, gcc.dg/cpp/cmdlne-dU-16.c, gcc.dg/cpp/cmdlne-dU-17.c, gcc.dg/cpp/cmdlne-dU-18.c, gcc.dg/cpp/cmdlne-dU-19.c, gcc.dg/cpp/cmdlne-dU-20.c, gcc.dg/cpp/cmdlne-dU-21.c, gcc.dg/cpp/cmdlne-dU-22.c, gcc.dg/cpp/mi5.c, gcc.dg/cpp/multiline.c: Add -nostdinc to dg-options. libcpp: * files.c (struct _cpp_file): Add implicit_preinclude. (pch_open_file): Allow a previously opened implicitly included file. (_cpp_find_file): Add implicit_preinclude argument. Free file and do not call open_file_failed if implicit_preinclude. Store implicit_preinclude value. (_cpp_stack_include, _cpp_fake_include, _cpp_compare_file_date): Update calls to _cpp_find_file. (_cpp_stack_include): Handle IT_DEFAULT. (cpp_push_default_include): New. * include/cpplib.h (cpp_push_default_include): Declare. * init.c (cpp_read_main_file): Update call to _cpp_find_file. * internal.h (enum include_type): Add IT_DEFAULT. (_cpp_find_file): Update prototype. From-SVN: r192715
2012-10-23 15:55:55 +01:00
/* If this file is implicitly preincluded. */
bool implicit_preinclude : 1;
/* Set if a header wasn't found with __has_include or __has_include_next
and error should be emitted if it is included normally. */
bool deferred_error : 1;
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
/* File loaded from #embed. */
bool embed : 1;
/* > 0: Known C++ Module header unit, <0: known not. ==0, unknown */
int header_unit : 2;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
};
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* A singly-linked list for all searches for a given file name, with
its head pointed to by a slot in FILE_HASH. The file name is what
appeared between the quotes in a #include directive; it can be
determined implicitly from the hash table location or explicitly
from FILE->name.
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
FILE is a structure containing details about the file that was
found with that search, or details of how the search failed.
START_DIR is the starting location of the search in the include
chain. The current directories for "" includes are also hashed in
the hash table and therefore unique. Files that are looked up
without using a search path, such as absolute filenames and file
names from the command line share a special starting directory so
they don't cause cache hits with normal include-chain lookups.
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
If START_DIR is NULL then the entry is for a directory, not a file,
and the directory is in DIR. Since the starting point in a file
lookup chain is never NULL, this means that simple pointer
comparisons against START_DIR can be made to determine cache hits
in file lookups.
If a cache lookup fails because of e.g. an extra "./" in the path,
then nothing will break. It is just less efficient as CPP will
have to do more work re-preprocessing the file, and/or comparing
its contents against earlier once-only files.
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
*/
struct cpp_file_hash_entry
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
{
struct cpp_file_hash_entry *next;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
cpp_dir *start_dir;
Eliminate source_location in favor of location_t Historically GCC used location_t, while libcpp used source_location. This inconsistency has been annoying me for a while, so this patch removes source_location in favor of location_t throughout (as the latter is shorter). gcc/ChangeLog: * builtins.c: Replace "source_location" with "location_t". * diagnostic-show-locus.c: Likewise. * diagnostic.c: Likewise. * dumpfile.c: Likewise. * gcc-rich-location.h: Likewise. * genmatch.c: Likewise. * gimple.h: Likewise. * gimplify.c: Likewise. * input.c: Likewise. * input.h: Likewise. Eliminate the typedef. * omp-expand.c: Likewise. * selftest.h: Likewise. * substring-locations.h (get_source_location_for_substring): Rename to.. (get_location_within_string): ...this. * tree-cfg.c: Replace "source_location" with "location_t". * tree-cfgcleanup.c: Likewise. * tree-diagnostic.c: Likewise. * tree-into-ssa.c: Likewise. * tree-outof-ssa.c: Likewise. * tree-parloops.c: Likewise. * tree-phinodes.c: Likewise. * tree-phinodes.h: Likewise. * tree-ssa-loop-ivopts.c: Likewise. * tree-ssa-loop-manip.c: Likewise. * tree-ssa-phiopt.c: Likewise. * tree-ssa-phiprop.c: Likewise. * tree-ssa-threadupdate.c: Likewise. * tree-ssa.c: Likewise. * tree-ssa.h: Likewise. * tree-vect-loop-manip.c: Likewise. gcc/c-family/ChangeLog: * c-common.c (c_get_substring_location): Update for renaming of get_source_location_for_substring to get_location_within_string. * c-lex.c: Replace "source_location" with "location_t". * c-opts.c: Likewise. * c-ppoutput.c: Likewise. gcc/c/ChangeLog: * c-decl.c: Replace "source_location" with "location_t". * c-tree.h: Likewise. * c-typeck.c: Likewise. * gimple-parser.c: Likewise. gcc/cp/ChangeLog: * call.c: Replace "source_location" with "location_t". * cp-tree.h: Likewise. * cvt.c: Likewise. * name-lookup.c: Likewise. * parser.c: Likewise. * typeck.c: Likewise. gcc/fortran/ChangeLog: * cpp.c: Replace "source_location" with "location_t". * gfortran.h: Likewise. gcc/go/ChangeLog: * go-gcc-diagnostics.cc: Replace "source_location" with "location_t". * go-gcc.cc: Likewise. * go-linemap.cc: Likewise. * go-location.h: Likewise. * gofrontend/README: Likewise. gcc/jit/ChangeLog: * jit-playback.c: Replace "source_location" with "location_t". gcc/testsuite/ChangeLog: * g++.dg/plugin/comment_plugin.c: Replace "source_location" with "location_t". * gcc.dg/plugin/diagnostic_plugin_test_show_locus.c: Likewise. libcc1/ChangeLog: * libcc1plugin.cc: Replace "source_location" with "location_t". (plugin_context::get_source_location): Rename to... (plugin_context::get_location_t): ...this. * libcp1plugin.cc: Likewise. libcpp/ChangeLog: * charset.c: Replace "source_location" with "location_t". * directives-only.c: Likewise. * directives.c: Likewise. * errors.c: Likewise. * expr.c: Likewise. * files.c: Likewise. * include/cpplib.h: Likewise. Rename MAX_SOURCE_LOCATION to MAX_LOCATION_T. * include/line-map.h: Likewise. * init.c: Likewise. * internal.h: Likewise. * lex.c: Likewise. * line-map.c: Likewise. * location-example.txt: Likewise. * macro.c: Likewise. * pch.c: Likewise. * traditional.c: Likewise. From-SVN: r266085
2018-11-13 20:05:03 +00:00
location_t location;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
union
{
_cpp_file *file;
cpp_dir *dir;
} u;
};
/* Number of entries to put in a cpp_file_hash_entry pool. */
#define FILE_HASH_POOL_SIZE 127
/* A file hash entry pool. We allocate cpp_file_hash_entry object from
one of these. */
struct file_hash_entry_pool
{
/* Number of entries used from this pool. */
unsigned int file_hash_entries_used;
/* Next pool in the chain; used when freeing. */
struct file_hash_entry_pool *next;
/* The memory pool. */
struct cpp_file_hash_entry pool[FILE_HASH_POOL_SIZE];
};
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
static bool open_file (_cpp_file *file);
static bool pch_open_file (cpp_reader *pfile, _cpp_file *file,
bool *invalid_pch);
static bool find_file_in_dir (cpp_reader *pfile, _cpp_file *file,
Eliminate source_location in favor of location_t Historically GCC used location_t, while libcpp used source_location. This inconsistency has been annoying me for a while, so this patch removes source_location in favor of location_t throughout (as the latter is shorter). gcc/ChangeLog: * builtins.c: Replace "source_location" with "location_t". * diagnostic-show-locus.c: Likewise. * diagnostic.c: Likewise. * dumpfile.c: Likewise. * gcc-rich-location.h: Likewise. * genmatch.c: Likewise. * gimple.h: Likewise. * gimplify.c: Likewise. * input.c: Likewise. * input.h: Likewise. Eliminate the typedef. * omp-expand.c: Likewise. * selftest.h: Likewise. * substring-locations.h (get_source_location_for_substring): Rename to.. (get_location_within_string): ...this. * tree-cfg.c: Replace "source_location" with "location_t". * tree-cfgcleanup.c: Likewise. * tree-diagnostic.c: Likewise. * tree-into-ssa.c: Likewise. * tree-outof-ssa.c: Likewise. * tree-parloops.c: Likewise. * tree-phinodes.c: Likewise. * tree-phinodes.h: Likewise. * tree-ssa-loop-ivopts.c: Likewise. * tree-ssa-loop-manip.c: Likewise. * tree-ssa-phiopt.c: Likewise. * tree-ssa-phiprop.c: Likewise. * tree-ssa-threadupdate.c: Likewise. * tree-ssa.c: Likewise. * tree-ssa.h: Likewise. * tree-vect-loop-manip.c: Likewise. gcc/c-family/ChangeLog: * c-common.c (c_get_substring_location): Update for renaming of get_source_location_for_substring to get_location_within_string. * c-lex.c: Replace "source_location" with "location_t". * c-opts.c: Likewise. * c-ppoutput.c: Likewise. gcc/c/ChangeLog: * c-decl.c: Replace "source_location" with "location_t". * c-tree.h: Likewise. * c-typeck.c: Likewise. * gimple-parser.c: Likewise. gcc/cp/ChangeLog: * call.c: Replace "source_location" with "location_t". * cp-tree.h: Likewise. * cvt.c: Likewise. * name-lookup.c: Likewise. * parser.c: Likewise. * typeck.c: Likewise. gcc/fortran/ChangeLog: * cpp.c: Replace "source_location" with "location_t". * gfortran.h: Likewise. gcc/go/ChangeLog: * go-gcc-diagnostics.cc: Replace "source_location" with "location_t". * go-gcc.cc: Likewise. * go-linemap.cc: Likewise. * go-location.h: Likewise. * gofrontend/README: Likewise. gcc/jit/ChangeLog: * jit-playback.c: Replace "source_location" with "location_t". gcc/testsuite/ChangeLog: * g++.dg/plugin/comment_plugin.c: Replace "source_location" with "location_t". * gcc.dg/plugin/diagnostic_plugin_test_show_locus.c: Likewise. libcc1/ChangeLog: * libcc1plugin.cc: Replace "source_location" with "location_t". (plugin_context::get_source_location): Rename to... (plugin_context::get_location_t): ...this. * libcp1plugin.cc: Likewise. libcpp/ChangeLog: * charset.c: Replace "source_location" with "location_t". * directives-only.c: Likewise. * directives.c: Likewise. * errors.c: Likewise. * expr.c: Likewise. * files.c: Likewise. * include/cpplib.h: Likewise. Rename MAX_SOURCE_LOCATION to MAX_LOCATION_T. * include/line-map.h: Likewise. * init.c: Likewise. * internal.h: Likewise. * lex.c: Likewise. * line-map.c: Likewise. * location-example.txt: Likewise. * macro.c: Likewise. * pch.c: Likewise. * traditional.c: Likewise. From-SVN: r266085
2018-11-13 20:05:03 +00:00
bool *invalid_pch, location_t loc);
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
static bool read_file_guts (cpp_reader *pfile, _cpp_file *file,
diagnostics: Support for -finput-charset [PR93067] Adds the logic to handle -finput-charset in layout_get_source_line(), so that source lines are converted from their input encodings prior to being output by diagnostics machinery. Also adds the ability to strip a UTF-8 BOM similarly. gcc/c-family/ChangeLog: PR other/93067 * c-opts.c (c_common_input_charset_cb): New function. (c_common_post_options): Call new function diagnostic_initialize_input_context(). gcc/d/ChangeLog: PR other/93067 * d-lang.cc (d_input_charset_callback): New function. (d_init): Call new function diagnostic_initialize_input_context(). gcc/fortran/ChangeLog: PR other/93067 * cpp.c (gfc_cpp_post_options): Call new function diagnostic_initialize_input_context(). gcc/ChangeLog: PR other/93067 * coretypes.h (typedef diagnostic_input_charset_callback): Declare. * diagnostic.c (diagnostic_initialize_input_context): New function. * diagnostic.h (diagnostic_initialize_input_context): Declare. * input.c (default_charset_callback): New function. (file_cache::initialize_input_context): New function. (file_cache_slot::create): Added ability to convert the input according to the input context. (file_cache::file_cache): Initialize the new input context. (class file_cache_slot): Added new m_alloc_offset member. (file_cache_slot::file_cache_slot): Initialize the new member. (file_cache_slot::~file_cache_slot): Handle potentially offset buffer. (file_cache_slot::maybe_grow): Likewise. (file_cache_slot::needs_read_p): Handle NULL fp, which is now possible. (file_cache_slot::get_next_line): Likewise. * input.h (class file_cache): Added input context member. libcpp/ChangeLog: PR other/93067 * charset.c (init_iconv_desc): Adapt to permit PFILE argument to be NULL. (_cpp_convert_input): Likewise. Also move UTF-8 BOM logic to... (cpp_check_utf8_bom): ...here. New function. (cpp_input_conversion_is_trivial): New function. * files.c (read_file_guts): Allow PFILE argument to be NULL. Add INPUT_CHARSET argument as an alternate source of this information. (read_file): Pass the new argument to read_file_guts. (cpp_get_converted_source): New function. * include/cpplib.h (struct cpp_converted_source): Declare. (cpp_get_converted_source): Declare. (cpp_input_conversion_is_trivial): Declare. (cpp_check_utf8_bom): Declare. gcc/testsuite/ChangeLog: PR other/93067 * gcc.dg/diagnostic-input-charset-1.c: New test. * gcc.dg/diagnostic-input-utf8-bom.c: New test.
2021-08-24 19:30:44 -04:00
location_t loc, const char *input_charset);
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
static bool read_file (cpp_reader *pfile, _cpp_file *file,
Eliminate source_location in favor of location_t Historically GCC used location_t, while libcpp used source_location. This inconsistency has been annoying me for a while, so this patch removes source_location in favor of location_t throughout (as the latter is shorter). gcc/ChangeLog: * builtins.c: Replace "source_location" with "location_t". * diagnostic-show-locus.c: Likewise. * diagnostic.c: Likewise. * dumpfile.c: Likewise. * gcc-rich-location.h: Likewise. * genmatch.c: Likewise. * gimple.h: Likewise. * gimplify.c: Likewise. * input.c: Likewise. * input.h: Likewise. Eliminate the typedef. * omp-expand.c: Likewise. * selftest.h: Likewise. * substring-locations.h (get_source_location_for_substring): Rename to.. (get_location_within_string): ...this. * tree-cfg.c: Replace "source_location" with "location_t". * tree-cfgcleanup.c: Likewise. * tree-diagnostic.c: Likewise. * tree-into-ssa.c: Likewise. * tree-outof-ssa.c: Likewise. * tree-parloops.c: Likewise. * tree-phinodes.c: Likewise. * tree-phinodes.h: Likewise. * tree-ssa-loop-ivopts.c: Likewise. * tree-ssa-loop-manip.c: Likewise. * tree-ssa-phiopt.c: Likewise. * tree-ssa-phiprop.c: Likewise. * tree-ssa-threadupdate.c: Likewise. * tree-ssa.c: Likewise. * tree-ssa.h: Likewise. * tree-vect-loop-manip.c: Likewise. gcc/c-family/ChangeLog: * c-common.c (c_get_substring_location): Update for renaming of get_source_location_for_substring to get_location_within_string. * c-lex.c: Replace "source_location" with "location_t". * c-opts.c: Likewise. * c-ppoutput.c: Likewise. gcc/c/ChangeLog: * c-decl.c: Replace "source_location" with "location_t". * c-tree.h: Likewise. * c-typeck.c: Likewise. * gimple-parser.c: Likewise. gcc/cp/ChangeLog: * call.c: Replace "source_location" with "location_t". * cp-tree.h: Likewise. * cvt.c: Likewise. * name-lookup.c: Likewise. * parser.c: Likewise. * typeck.c: Likewise. gcc/fortran/ChangeLog: * cpp.c: Replace "source_location" with "location_t". * gfortran.h: Likewise. gcc/go/ChangeLog: * go-gcc-diagnostics.cc: Replace "source_location" with "location_t". * go-gcc.cc: Likewise. * go-linemap.cc: Likewise. * go-location.h: Likewise. * gofrontend/README: Likewise. gcc/jit/ChangeLog: * jit-playback.c: Replace "source_location" with "location_t". gcc/testsuite/ChangeLog: * g++.dg/plugin/comment_plugin.c: Replace "source_location" with "location_t". * gcc.dg/plugin/diagnostic_plugin_test_show_locus.c: Likewise. libcc1/ChangeLog: * libcc1plugin.cc: Replace "source_location" with "location_t". (plugin_context::get_source_location): Rename to... (plugin_context::get_location_t): ...this. * libcp1plugin.cc: Likewise. libcpp/ChangeLog: * charset.c: Replace "source_location" with "location_t". * directives-only.c: Likewise. * directives.c: Likewise. * errors.c: Likewise. * expr.c: Likewise. * files.c: Likewise. * include/cpplib.h: Likewise. Rename MAX_SOURCE_LOCATION to MAX_LOCATION_T. * include/line-map.h: Likewise. * init.c: Likewise. * internal.h: Likewise. * lex.c: Likewise. * line-map.c: Likewise. * location-example.txt: Likewise. * macro.c: Likewise. * pch.c: Likewise. * traditional.c: Likewise. From-SVN: r266085
2018-11-13 20:05:03 +00:00
location_t loc);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
static struct cpp_dir *search_path_head (cpp_reader *, const char *fname,
int angle_brackets, enum include_type,
bool suppress_diagnostic = false);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
static const char *dir_name_of_file (_cpp_file *file);
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
static void open_file_failed (cpp_reader *pfile, _cpp_file *file, int,
Eliminate source_location in favor of location_t Historically GCC used location_t, while libcpp used source_location. This inconsistency has been annoying me for a while, so this patch removes source_location in favor of location_t throughout (as the latter is shorter). gcc/ChangeLog: * builtins.c: Replace "source_location" with "location_t". * diagnostic-show-locus.c: Likewise. * diagnostic.c: Likewise. * dumpfile.c: Likewise. * gcc-rich-location.h: Likewise. * genmatch.c: Likewise. * gimple.h: Likewise. * gimplify.c: Likewise. * input.c: Likewise. * input.h: Likewise. Eliminate the typedef. * omp-expand.c: Likewise. * selftest.h: Likewise. * substring-locations.h (get_source_location_for_substring): Rename to.. (get_location_within_string): ...this. * tree-cfg.c: Replace "source_location" with "location_t". * tree-cfgcleanup.c: Likewise. * tree-diagnostic.c: Likewise. * tree-into-ssa.c: Likewise. * tree-outof-ssa.c: Likewise. * tree-parloops.c: Likewise. * tree-phinodes.c: Likewise. * tree-phinodes.h: Likewise. * tree-ssa-loop-ivopts.c: Likewise. * tree-ssa-loop-manip.c: Likewise. * tree-ssa-phiopt.c: Likewise. * tree-ssa-phiprop.c: Likewise. * tree-ssa-threadupdate.c: Likewise. * tree-ssa.c: Likewise. * tree-ssa.h: Likewise. * tree-vect-loop-manip.c: Likewise. gcc/c-family/ChangeLog: * c-common.c (c_get_substring_location): Update for renaming of get_source_location_for_substring to get_location_within_string. * c-lex.c: Replace "source_location" with "location_t". * c-opts.c: Likewise. * c-ppoutput.c: Likewise. gcc/c/ChangeLog: * c-decl.c: Replace "source_location" with "location_t". * c-tree.h: Likewise. * c-typeck.c: Likewise. * gimple-parser.c: Likewise. gcc/cp/ChangeLog: * call.c: Replace "source_location" with "location_t". * cp-tree.h: Likewise. * cvt.c: Likewise. * name-lookup.c: Likewise. * parser.c: Likewise. * typeck.c: Likewise. gcc/fortran/ChangeLog: * cpp.c: Replace "source_location" with "location_t". * gfortran.h: Likewise. gcc/go/ChangeLog: * go-gcc-diagnostics.cc: Replace "source_location" with "location_t". * go-gcc.cc: Likewise. * go-linemap.cc: Likewise. * go-location.h: Likewise. * gofrontend/README: Likewise. gcc/jit/ChangeLog: * jit-playback.c: Replace "source_location" with "location_t". gcc/testsuite/ChangeLog: * g++.dg/plugin/comment_plugin.c: Replace "source_location" with "location_t". * gcc.dg/plugin/diagnostic_plugin_test_show_locus.c: Likewise. libcc1/ChangeLog: * libcc1plugin.cc: Replace "source_location" with "location_t". (plugin_context::get_source_location): Rename to... (plugin_context::get_location_t): ...this. * libcp1plugin.cc: Likewise. libcpp/ChangeLog: * charset.c: Replace "source_location" with "location_t". * directives-only.c: Likewise. * directives.c: Likewise. * errors.c: Likewise. * expr.c: Likewise. * files.c: Likewise. * include/cpplib.h: Likewise. Rename MAX_SOURCE_LOCATION to MAX_LOCATION_T. * include/line-map.h: Likewise. * init.c: Likewise. * internal.h: Likewise. * lex.c: Likewise. * line-map.c: Likewise. * location-example.txt: Likewise. * macro.c: Likewise. * pch.c: Likewise. * traditional.c: Likewise. From-SVN: r266085
2018-11-13 20:05:03 +00:00
location_t);
static struct cpp_file_hash_entry *search_cache (struct cpp_file_hash_entry *head,
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
const cpp_dir *start_dir,
bool is_embed);
static _cpp_file *make_cpp_file (cpp_dir *, const char *fname);
static void destroy_cpp_file (_cpp_file *);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
static cpp_dir *make_cpp_dir (cpp_reader *, const char *dir_name, int sysp);
static void allocate_file_hash_entries (cpp_reader *pfile);
static struct cpp_file_hash_entry *new_file_hash_entry (cpp_reader *pfile);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
static int report_missing_guard (void **slot, void *b);
static hashval_t file_hash_hash (const void *p);
static int file_hash_eq (const void *p, const void *q);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
static char *read_filename_string (int ch, FILE *f);
static void read_name_map (cpp_dir *dir);
static char *remap_filename (cpp_reader *pfile, _cpp_file *file);
static char *append_file_to_dir (const char *fname, cpp_dir *dir);
static bool validate_pch (cpp_reader *, _cpp_file *file, const char *pchname);
static int pchf_save_compare (const void *e1, const void *e2);
static int pchf_compare (const void *d_p, const void *e_p);
static bool check_file_against_entries (cpp_reader *, _cpp_file *, bool);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Given a filename in FILE->PATH, with the empty string interpreted
as <stdin>, open it.
On success FILE contains an open file descriptor and stat
information for the file. On failure the file descriptor is -1 and
the appropriate errno is also stored in FILE. Returns TRUE iff
successful.
We used to open files in nonblocking mode, but that caused more
problems than it solved. Do take care not to acquire a controlling
terminal by mistake (this can't happen on sane systems, but
paranoia is a virtue).
Use the three-argument form of open even though we aren't
specifying O_CREAT, to defend against broken system headers.
O_BINARY tells some runtime libraries (notably DJGPP) not to do
newline translation; we can handle DOS line breaks just fine
ourselves. */
static bool
open_file (_cpp_file *file)
{
if (file->path[0] == '\0')
{
file->fd = 0;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
set_stdin_to_binary_mode ();
}
cpplex.c (cpp_output_tokens, [...]): New public interfaces. * cpplex.c (cpp_output_tokens, cpp_scan_buffer_nooutput): New public interfaces. (safe_fwrite, output_line_command): New static functions. (cpp_expand_to_buffer): Now private to cpplib. (cpp_scan_buffer): Take a printer. * cpphash.h: Update prototypes. * cpplib.h: Update prototypes. (cpp_printer): New. (cpp_buffer): Remove last_nominal_fname. (cpp_reader): Remove lineno. * cppmain.c: Use a cpp_printer. * fix-header.c: No need to inhibit line commands. Call cpp_start_read with no printer. * cpperror.c (cpp_notice_from_errno): Provide default name. * cppfiles.c (make_IHASH, _cpp_fake_ihash): New functions. (find_include_file, cpp_read_file): Use make_IHASH. (file_cleanup): Set control_macro and clear input_stack_listing_current here. (_cpp_execute_include): Don't output entering-file marker. * cpphash.c (special_symbol): Look for the line number in the buffer, not the reader. (_cpp_macroexpand): No need to disable line commands. (_cpp_dump_definition): No need to generate line commands. (dump_hash_helper): Remove excess newline from output. * cppinit.c (dump_special_to_buffer): No need to generate line commands. (cpp_printer_init): New. (cpp_start_read): Take a printer, and start it up if it's not NULL. No need to generate line commands. (cpp_finish): Expect no buffers stacked at all. Take a printer argument, and flush the output buffer if it's not NULL. * cpplex.c (_cpp_lex_token): Return EOF if there's no buffer. Don't put two hashes at the beginning of an assertion. (cpp_get_token): Don't increment pfile->lineno or emit line commands here. Return EOF if there's no buffer when we get EOF. * cpplib.c (do_define, skip_if_group): No need to disable line commands. (_cpp_output_line_command): Delete function. (do_line): Don't emit line commands here, but set things up so they will be emitted if necessary. Use _cpp_fake_ihash to make unique nominal_fnames if necessary. (do_elif, do_else, _cpp_handle_eof): Call cpp_error_with_line with 0 for column, not -1. (_cpp_handle_eof): Don't set the control macro here. Don't clear input_stack_listing_current here. Don't emit line commands. From-SVN: r33159
2000-04-14 23:29:45 +00:00
else
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
file->fd = open (file->path, O_RDONLY | O_NOCTTY | O_BINARY, 0666);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (file->fd != -1)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (fstat (file->fd, &file->st) == 0)
{
if (!S_ISDIR (file->st.st_mode))
{
file->err_no = 0;
return true;
}
/* Ignore a directory and continue the search. The file we're
looking for may be elsewhere in the search path. */
errno = ENOENT;
}
close (file->fd);
file->fd = -1;
}
#if defined(_WIN32) && !defined(__CYGWIN__)
else if (errno == EACCES)
{
/* On most UNIX systems, open succeeds on a directory. Above,
we check if we have opened a directory and if so, set errno
to ENOENT. However, on Windows, opening a directory
fails with EACCES. We want to return ENOENT in that
case too. */
if (stat (file->path, &file->st) == 0
&& S_ISDIR (file->st.st_mode))
errno = ENOENT;
else
/* The call to stat may have reset errno. */
errno = EACCES;
}
#endif
else if (errno == ENOTDIR)
errno = ENOENT;
file->err_no = errno;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
return false;
}
/* Temporary PCH intercept of opening a file. Try to find a PCH file
based on FILE->name and FILE->dir, and test those found for
validity using PFILE->cb.valid_pch. Return true iff a valid file is
found. Set *INVALID_PCH if a PCH file is found but wasn't valid. */
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
static bool
pch_open_file (cpp_reader *pfile, _cpp_file *file, bool *invalid_pch)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
static const char extension[] = ".gch";
const char *path = file->path;
size_t len, flen;
char *pchname;
struct stat st;
bool valid = false;
/* No PCH on <stdin> or if not requested. */
if (file->name[0] == '\0' || !pfile->cb.valid_pch)
return false;
/* If the file is not included as first include from either the toplevel
file or the command-line it is not a valid use of PCH. */
for (_cpp_file *f = pfile->all_files; f; f = f->next_file)
if (f->implicit_preinclude)
continue;
else if (pfile->main_file == f)
break;
else
return false;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
flen = strlen (path);
len = flen + sizeof (extension);
configure.ac: Check declarations for asprintf and vasprintf. * configure.ac: Check declarations for asprintf and vasprintf. * config.in: Regenerate. * configure: Likewise. * charset.c (conversion_loop): Use XRESIZEVEC. (convert_no_conversion): Likewise. (convert_using_iconv): Likewise. (init_iconv_desc): Cast return value of alloca. (cpp_host_to_exec_charset): Use XNEWVEC. (emit_numeric_escape): Use XRESIZEVEC. (cpp_interpret_string): Use XNEWVEC. (cpp_interpret_string): Use XRESIZEVEC. (_cpp_interpret_identifier): Cast return value of alloca. (_cpp_convert_input): Use XNEWVEC and XRESIZEVEC. * directives.c (glue_header_name): Use XNEWVEC and XRESIZEVEC. (parse_include): Use XNEWVEC. (insert_pragma_entry): Rename local variable "new" to "new_entry". (save_registered_pragmas): Cast return value of xmemdup. (destringize_and_run): Same for alloca. (parse_assertion): Likewise. (do_assert): Cast allocated storage to proper type. (cpp_define): Likewise. (_cpp_define_builtin): Likewise. (cpp_undef): Likewise. (handle_assertion): Likewise. (cpp_push_buffer): Rename local variable "new" to "new_buffer". * expr.c (CPP_UPLUS): Cast value to type cpp_ttype. (CPP_UMINUS): Likewise. (struct cpp_operator): Rename from struct operator. (_cpp_expand_op_stack): Use XRESIZEVEC. * files.c (pch_open_file): Use XNEWVEC. (pch_open_file): Use XRESIZEVEC. (read_file_guts): Use XNEWVEC and XRESIZEVEC. (dir_name_of_file): Use XNEWVEC. (make_cpp_file): Use XCNEW. (make_cpp_dir): Likewise. (allocate_file_hash_entries): USE XNEWVEC. (cpp_included): Cast return value of htab_find_with_hash. (append_file_to_dir): Use XNEWVEC. (read_filename_string): Likewise. Use XRESIZEVEC too. (read_name_map): Cast return value of alloca. Use XRESIZEVEC. (remap_filename): Use XNEWVEC. (struct pchf_entry): Move definition out of struct pchf_data. (_cpp_save_file_entries): Use XCNEWVAR. (_cpp_read_file_entries): Use XNEWVAR. * identifiers.c (alloc_node): Use XOBNEW. * init.c (cpp_create_reader): Use XCNEW. (cpp_init_builtins): Cast of b->value to enum builtin_type. (read_original_directory): Cast return value of alloca. * lex.c (add_line_note): Use XRESIZEVEC. (warn_about_normalization): Use XNEWVEC. (_cpp_lex_direct): Cast node->directive_index to (enum cpp_ttype). (new_buff): Use XNEWVEC. * line-map.c (linemap_add): Use XRESIZEVEC. * macro.c (builtin_macro): Cast return value of alloca. (paste_tokens): Likewise. (expand_arg): Use XNEWVEC and XRESIZEVEC. (_cpp_save_parameter): Use XRESIZEVEC. (create_iso_definition): Cast allocated storage to proper type. (_cpp_create_definition): Likewise. (cpp_macro_definition): Use XRESIZEVEC. * makedepend.c (add_clm): Use XNEW. (add_dir): Likewise. * mkdeps.c (munge): Use XNEWVEC. (deps_init): Use XCNEW. (deps_add_target): Use XRESIZEVEC. (deps_add_default_target): Cast return value of alloca. (deps_add_dep): Use XRESIZEVEC. (deps_add_vpath): Likewise. Use XNEWVEC too. (deps_restore): Likewise. * pch.c (save_idents): Use XNEW and XNEWVEC. (cpp_save_state): Use XNEW. (count_defs): Cast return value of htab_find. (write_defs): Likewise. (cpp_write_pch_deps): Use XNEWVEC. (collect_ht_nodes): Use XRESIZEVEC. (cpp_valid_state): Use XNEWVEC. (save_macros): Use XRESIZEVEC. Cast return value of xmemdup. * symtab.c (ht_create): Use XCNEW. (ht_lookup_with_hash): Cast return value of obstack_copy0. (ht_expand): Use XCNEWVEC. * system.h (HAVE_DESIGNATED_INITIALIZERS): False if __cplusplus. (bool): Do not define if __cplusplus. From-SVN: r100295
2005-05-28 15:52:48 +00:00
pchname = XNEWVEC (char, len);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
memcpy (pchname, path, flen);
memcpy (pchname + flen, extension, sizeof (extension));
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (stat (pchname, &st) == 0)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
DIR *pchdir;
struct dirent *d;
size_t dlen, plen = len;
if (!S_ISDIR (st.st_mode))
valid = validate_pch (pfile, file, pchname);
else if ((pchdir = opendir (pchname)) != NULL)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
pchname[plen - 1] = '/';
while ((d = readdir (pchdir)) != NULL)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
dlen = strlen (d->d_name) + 1;
if ((strcmp (d->d_name, ".") == 0)
|| (strcmp (d->d_name, "..") == 0))
continue;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (dlen + plen > len)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
len += dlen + 64;
configure.ac: Check declarations for asprintf and vasprintf. * configure.ac: Check declarations for asprintf and vasprintf. * config.in: Regenerate. * configure: Likewise. * charset.c (conversion_loop): Use XRESIZEVEC. (convert_no_conversion): Likewise. (convert_using_iconv): Likewise. (init_iconv_desc): Cast return value of alloca. (cpp_host_to_exec_charset): Use XNEWVEC. (emit_numeric_escape): Use XRESIZEVEC. (cpp_interpret_string): Use XNEWVEC. (cpp_interpret_string): Use XRESIZEVEC. (_cpp_interpret_identifier): Cast return value of alloca. (_cpp_convert_input): Use XNEWVEC and XRESIZEVEC. * directives.c (glue_header_name): Use XNEWVEC and XRESIZEVEC. (parse_include): Use XNEWVEC. (insert_pragma_entry): Rename local variable "new" to "new_entry". (save_registered_pragmas): Cast return value of xmemdup. (destringize_and_run): Same for alloca. (parse_assertion): Likewise. (do_assert): Cast allocated storage to proper type. (cpp_define): Likewise. (_cpp_define_builtin): Likewise. (cpp_undef): Likewise. (handle_assertion): Likewise. (cpp_push_buffer): Rename local variable "new" to "new_buffer". * expr.c (CPP_UPLUS): Cast value to type cpp_ttype. (CPP_UMINUS): Likewise. (struct cpp_operator): Rename from struct operator. (_cpp_expand_op_stack): Use XRESIZEVEC. * files.c (pch_open_file): Use XNEWVEC. (pch_open_file): Use XRESIZEVEC. (read_file_guts): Use XNEWVEC and XRESIZEVEC. (dir_name_of_file): Use XNEWVEC. (make_cpp_file): Use XCNEW. (make_cpp_dir): Likewise. (allocate_file_hash_entries): USE XNEWVEC. (cpp_included): Cast return value of htab_find_with_hash. (append_file_to_dir): Use XNEWVEC. (read_filename_string): Likewise. Use XRESIZEVEC too. (read_name_map): Cast return value of alloca. Use XRESIZEVEC. (remap_filename): Use XNEWVEC. (struct pchf_entry): Move definition out of struct pchf_data. (_cpp_save_file_entries): Use XCNEWVAR. (_cpp_read_file_entries): Use XNEWVAR. * identifiers.c (alloc_node): Use XOBNEW. * init.c (cpp_create_reader): Use XCNEW. (cpp_init_builtins): Cast of b->value to enum builtin_type. (read_original_directory): Cast return value of alloca. * lex.c (add_line_note): Use XRESIZEVEC. (warn_about_normalization): Use XNEWVEC. (_cpp_lex_direct): Cast node->directive_index to (enum cpp_ttype). (new_buff): Use XNEWVEC. * line-map.c (linemap_add): Use XRESIZEVEC. * macro.c (builtin_macro): Cast return value of alloca. (paste_tokens): Likewise. (expand_arg): Use XNEWVEC and XRESIZEVEC. (_cpp_save_parameter): Use XRESIZEVEC. (create_iso_definition): Cast allocated storage to proper type. (_cpp_create_definition): Likewise. (cpp_macro_definition): Use XRESIZEVEC. * makedepend.c (add_clm): Use XNEW. (add_dir): Likewise. * mkdeps.c (munge): Use XNEWVEC. (deps_init): Use XCNEW. (deps_add_target): Use XRESIZEVEC. (deps_add_default_target): Cast return value of alloca. (deps_add_dep): Use XRESIZEVEC. (deps_add_vpath): Likewise. Use XNEWVEC too. (deps_restore): Likewise. * pch.c (save_idents): Use XNEW and XNEWVEC. (cpp_save_state): Use XNEW. (count_defs): Cast return value of htab_find. (write_defs): Likewise. (cpp_write_pch_deps): Use XNEWVEC. (collect_ht_nodes): Use XRESIZEVEC. (cpp_valid_state): Use XNEWVEC. (save_macros): Use XRESIZEVEC. Cast return value of xmemdup. * symtab.c (ht_create): Use XCNEW. (ht_lookup_with_hash): Cast return value of obstack_copy0. (ht_expand): Use XCNEWVEC. * system.h (HAVE_DESIGNATED_INITIALIZERS): False if __cplusplus. (bool): Do not define if __cplusplus. From-SVN: r100295
2005-05-28 15:52:48 +00:00
pchname = XRESIZEVEC (char, pchname, len);
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
memcpy (pchname + plen, d->d_name, dlen);
valid = validate_pch (pfile, file, pchname);
if (valid)
break;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
closedir (pchdir);
}
if (!valid)
*invalid_pch = true;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (valid)
file->pchname = pchname;
else
free (pchname);
return valid;
}
/* Canonicalize the path to FILE. Return the canonical form if it is
shorter, otherwise return NULL. This function does NOT free the
memory pointed by FILE. */
static char *
maybe_shorter_path (const char * file)
{
char * file2 = lrealpath (file);
if (file2 && strlen (file2) < strlen (file))
{
return file2;
}
else
{
free (file2);
return NULL;
}
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Try to open the path FILE->name appended to FILE->dir. This is
where remap and PCH intercept the file lookup process. Return true
[multiple changes] 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * c-opts.c (c_common_handle_option): Add -finput-charset. * c.opt: Ditto. * cppcharset.c (one_iso88591_to_utf8): Remove. (convert_iso88591_utf8): Ditto. (conversion_tab): Remove 8859-1 converter. (_cpp_input_to_utf8): Remove. (_cpp_init_iconv_buffer): Ditto. (_cpp_close_iconv_buffer): Ditto. (_cpp_convert_input): New function. (_cpp_default_encoding): Ditto. * cpphash.h: Add/remove prototypes for above. * cppfiles.c (read_file_guts): Use _cpp_convert_input. * cppinit.c (cpp_create_reader): Use _cpp_default_encoding for narrow execution and input character sets. * cpplib.c (cpp_push_buffer): Delete uses of removed functions. * doc/cppopts.texi: Document -finput-charset. 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * gcc.c-torture/execute/wchar_t-1.c: Add -finput-charset. 2004-01-29 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * testsuite/22_locale/collate/compare/wchar_t/2.cc: Remove xfail. Use -finput-charset. * testsuite/22_locale/collate/compare/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/compare/wchar_t/wrapped_locale.cc: Ditto * testsuite/22_locale/collate/hash/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_locale.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_locale.cc: Ditto. From-SVN: r77136
2004-02-02 20:20:58 +00:00
if the file was found, whether or not the open was successful.
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
Set *INVALID_PCH to true if a PCH file is found but wasn't valid.
Use LOC when emitting any diagnostics. */
static bool
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
find_file_in_dir (cpp_reader *pfile, _cpp_file *file, bool *invalid_pch,
Eliminate source_location in favor of location_t Historically GCC used location_t, while libcpp used source_location. This inconsistency has been annoying me for a while, so this patch removes source_location in favor of location_t throughout (as the latter is shorter). gcc/ChangeLog: * builtins.c: Replace "source_location" with "location_t". * diagnostic-show-locus.c: Likewise. * diagnostic.c: Likewise. * dumpfile.c: Likewise. * gcc-rich-location.h: Likewise. * genmatch.c: Likewise. * gimple.h: Likewise. * gimplify.c: Likewise. * input.c: Likewise. * input.h: Likewise. Eliminate the typedef. * omp-expand.c: Likewise. * selftest.h: Likewise. * substring-locations.h (get_source_location_for_substring): Rename to.. (get_location_within_string): ...this. * tree-cfg.c: Replace "source_location" with "location_t". * tree-cfgcleanup.c: Likewise. * tree-diagnostic.c: Likewise. * tree-into-ssa.c: Likewise. * tree-outof-ssa.c: Likewise. * tree-parloops.c: Likewise. * tree-phinodes.c: Likewise. * tree-phinodes.h: Likewise. * tree-ssa-loop-ivopts.c: Likewise. * tree-ssa-loop-manip.c: Likewise. * tree-ssa-phiopt.c: Likewise. * tree-ssa-phiprop.c: Likewise. * tree-ssa-threadupdate.c: Likewise. * tree-ssa.c: Likewise. * tree-ssa.h: Likewise. * tree-vect-loop-manip.c: Likewise. gcc/c-family/ChangeLog: * c-common.c (c_get_substring_location): Update for renaming of get_source_location_for_substring to get_location_within_string. * c-lex.c: Replace "source_location" with "location_t". * c-opts.c: Likewise. * c-ppoutput.c: Likewise. gcc/c/ChangeLog: * c-decl.c: Replace "source_location" with "location_t". * c-tree.h: Likewise. * c-typeck.c: Likewise. * gimple-parser.c: Likewise. gcc/cp/ChangeLog: * call.c: Replace "source_location" with "location_t". * cp-tree.h: Likewise. * cvt.c: Likewise. * name-lookup.c: Likewise. * parser.c: Likewise. * typeck.c: Likewise. gcc/fortran/ChangeLog: * cpp.c: Replace "source_location" with "location_t". * gfortran.h: Likewise. gcc/go/ChangeLog: * go-gcc-diagnostics.cc: Replace "source_location" with "location_t". * go-gcc.cc: Likewise. * go-linemap.cc: Likewise. * go-location.h: Likewise. * gofrontend/README: Likewise. gcc/jit/ChangeLog: * jit-playback.c: Replace "source_location" with "location_t". gcc/testsuite/ChangeLog: * g++.dg/plugin/comment_plugin.c: Replace "source_location" with "location_t". * gcc.dg/plugin/diagnostic_plugin_test_show_locus.c: Likewise. libcc1/ChangeLog: * libcc1plugin.cc: Replace "source_location" with "location_t". (plugin_context::get_source_location): Rename to... (plugin_context::get_location_t): ...this. * libcp1plugin.cc: Likewise. libcpp/ChangeLog: * charset.c: Replace "source_location" with "location_t". * directives-only.c: Likewise. * directives.c: Likewise. * errors.c: Likewise. * expr.c: Likewise. * files.c: Likewise. * include/cpplib.h: Likewise. Rename MAX_SOURCE_LOCATION to MAX_LOCATION_T. * include/line-map.h: Likewise. * init.c: Likewise. * internal.h: Likewise. * lex.c: Likewise. * line-map.c: Likewise. * location-example.txt: Likewise. * macro.c: Likewise. * pch.c: Likewise. * traditional.c: Likewise. From-SVN: r266085
2018-11-13 20:05:03 +00:00
location_t loc)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
char *path;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (CPP_OPTION (pfile, remap) && (path = remap_filename (pfile, file)))
;
else
Add framework support for darwin. * c-incpath.c: Include target.h and machmode.h. (add_path): Use a consistent style for cpp_dir. Initialize p->construct to 0. (add_cpp_dir_path): New. (register_include_chains): Add use of extra_includes callback. (hook_void_int): Add. (target_c_incpath): Add. * c-incpath.h (add_cpp_dir_path): New. (target_c_incpath_s): Add. (target_c_incpath): Add. (C_INCPATH_INIT): Add. * c-opts.c (c_common_missing_argument, c_common_handle_option): Add -F argument processing. * c.opt: Add -F argument processing. * gcc.c (trad_capable_cpp): Add -F argument processing. * cppfiles.c (find_file_in_dir): Update to use construct callback. (search_path_exhausted, cpp_get_path, cpp_get_buffer, cpp_get_prev): New. (_cpp_find_file): Use search_path_exhausted. (make_cpp_dir): Initialize construct to 0. * cpplib.h (missing_header_cb cpp_get_path, cpp_get_buffer, cpp_get_file, cpp_get_prev): New. (cpp_callbacks): Add missing_header (cpp_dir): Add construct. * target-def.h: (TARGET_OPTF): New. * hooks.c (hook_void_int, hook_void_charptr): Add. * hooks.h (hook_void_int, hook_void_charptr): Add. * Makefile.in (c-incpath.o) : Add $(TARGET_H) and $(MACHMODE_H) dependencies. * doc/invoke.texi (Darwin Options): Document -F. * doc/tm.texi (TARGET_EXTRA_INCLUDES): Add. (TARGET_OPTF): Add. * fix-header.c (target_c_incpath): Add. * config/darwin-c.c: Add c-incpath.h include. (using_frameworks, find_subframework_file, find_subframework_header, add_system_framework_path, frameworks_in_use, num_frameworks, max_frameworks, add_framework, find_framework, struct framework_header, framework_header_dirs, framework_construct_pathname, find_subframework_file, add_system_framework_path, add_framework_path, framework_defaults, darwin_register_frameworks, find_subframework_header): Add. * config/darwin.h (TARGET_EXTRA_INCLUDES, TARGET_OPTF): New. (TARGET_OPTION_TRANSLATE_TABLE): Add -framework support. (CPP_SPEC): Add __APPLE_CC__ support. * t-darwin (darwin-c.o): Add c-incpath.h dependency. From-SVN: r78875
2004-03-04 00:18:54 +00:00
if (file->dir->construct)
path = file->dir->construct (file->name, file->dir);
else
path = append_file_to_dir (file->name, file->dir);
Add framework support for darwin. * c-incpath.c: Include target.h and machmode.h. (add_path): Use a consistent style for cpp_dir. Initialize p->construct to 0. (add_cpp_dir_path): New. (register_include_chains): Add use of extra_includes callback. (hook_void_int): Add. (target_c_incpath): Add. * c-incpath.h (add_cpp_dir_path): New. (target_c_incpath_s): Add. (target_c_incpath): Add. (C_INCPATH_INIT): Add. * c-opts.c (c_common_missing_argument, c_common_handle_option): Add -F argument processing. * c.opt: Add -F argument processing. * gcc.c (trad_capable_cpp): Add -F argument processing. * cppfiles.c (find_file_in_dir): Update to use construct callback. (search_path_exhausted, cpp_get_path, cpp_get_buffer, cpp_get_prev): New. (_cpp_find_file): Use search_path_exhausted. (make_cpp_dir): Initialize construct to 0. * cpplib.h (missing_header_cb cpp_get_path, cpp_get_buffer, cpp_get_file, cpp_get_prev): New. (cpp_callbacks): Add missing_header (cpp_dir): Add construct. * target-def.h: (TARGET_OPTF): New. * hooks.c (hook_void_int, hook_void_charptr): Add. * hooks.h (hook_void_int, hook_void_charptr): Add. * Makefile.in (c-incpath.o) : Add $(TARGET_H) and $(MACHMODE_H) dependencies. * doc/invoke.texi (Darwin Options): Document -F. * doc/tm.texi (TARGET_EXTRA_INCLUDES): Add. (TARGET_OPTF): Add. * fix-header.c (target_c_incpath): Add. * config/darwin-c.c: Add c-incpath.h include. (using_frameworks, find_subframework_file, find_subframework_header, add_system_framework_path, frameworks_in_use, num_frameworks, max_frameworks, add_framework, find_framework, struct framework_header, framework_header_dirs, framework_construct_pathname, find_subframework_file, add_system_framework_path, add_framework_path, framework_defaults, darwin_register_frameworks, find_subframework_header): Add. * config/darwin.h (TARGET_EXTRA_INCLUDES, TARGET_OPTF): New. (TARGET_OPTION_TRANSLATE_TABLE): Add -framework support. (CPP_SPEC): Add __APPLE_CC__ support. * t-darwin (darwin-c.o): Add c-incpath.h dependency. From-SVN: r78875
2004-03-04 00:18:54 +00:00
if (path)
{
hashval_t hv;
char *copy;
void **pp;
/* We try to canonicalize system headers. For DOS based file
* system, we always try to shorten non-system headers, as DOS
* has a tighter constraint on max path length. */
if ((CPP_OPTION (pfile, canonical_system_headers) && file->dir->sysp)
#ifdef HAVE_DOS_BASED_FILE_SYSTEM
|| !file->dir->sysp
#endif
)
{
char * canonical_path = maybe_shorter_path (path);
if (canonical_path)
{
/* The canonical path was newly allocated. Let's free the
non-canonical one. */
free (path);
path = canonical_path;
}
}
hv = htab_hash_string (path);
if (htab_find_with_hash (pfile->nonexistent_file_hash, path, hv) != NULL)
{
file->err_no = ENOENT;
return false;
}
Add framework support for darwin. * c-incpath.c: Include target.h and machmode.h. (add_path): Use a consistent style for cpp_dir. Initialize p->construct to 0. (add_cpp_dir_path): New. (register_include_chains): Add use of extra_includes callback. (hook_void_int): Add. (target_c_incpath): Add. * c-incpath.h (add_cpp_dir_path): New. (target_c_incpath_s): Add. (target_c_incpath): Add. (C_INCPATH_INIT): Add. * c-opts.c (c_common_missing_argument, c_common_handle_option): Add -F argument processing. * c.opt: Add -F argument processing. * gcc.c (trad_capable_cpp): Add -F argument processing. * cppfiles.c (find_file_in_dir): Update to use construct callback. (search_path_exhausted, cpp_get_path, cpp_get_buffer, cpp_get_prev): New. (_cpp_find_file): Use search_path_exhausted. (make_cpp_dir): Initialize construct to 0. * cpplib.h (missing_header_cb cpp_get_path, cpp_get_buffer, cpp_get_file, cpp_get_prev): New. (cpp_callbacks): Add missing_header (cpp_dir): Add construct. * target-def.h: (TARGET_OPTF): New. * hooks.c (hook_void_int, hook_void_charptr): Add. * hooks.h (hook_void_int, hook_void_charptr): Add. * Makefile.in (c-incpath.o) : Add $(TARGET_H) and $(MACHMODE_H) dependencies. * doc/invoke.texi (Darwin Options): Document -F. * doc/tm.texi (TARGET_EXTRA_INCLUDES): Add. (TARGET_OPTF): Add. * fix-header.c (target_c_incpath): Add. * config/darwin-c.c: Add c-incpath.h include. (using_frameworks, find_subframework_file, find_subframework_header, add_system_framework_path, frameworks_in_use, num_frameworks, max_frameworks, add_framework, find_framework, struct framework_header, framework_header_dirs, framework_construct_pathname, find_subframework_file, add_system_framework_path, add_framework_path, framework_defaults, darwin_register_frameworks, find_subframework_header): Add. * config/darwin.h (TARGET_EXTRA_INCLUDES, TARGET_OPTF): New. (TARGET_OPTION_TRANSLATE_TABLE): Add -framework support. (CPP_SPEC): Add __APPLE_CC__ support. * t-darwin (darwin-c.o): Add c-incpath.h dependency. From-SVN: r78875
2004-03-04 00:18:54 +00:00
file->path = path;
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
if (!file->embed && pch_open_file (pfile, file, invalid_pch))
Add framework support for darwin. * c-incpath.c: Include target.h and machmode.h. (add_path): Use a consistent style for cpp_dir. Initialize p->construct to 0. (add_cpp_dir_path): New. (register_include_chains): Add use of extra_includes callback. (hook_void_int): Add. (target_c_incpath): Add. * c-incpath.h (add_cpp_dir_path): New. (target_c_incpath_s): Add. (target_c_incpath): Add. (C_INCPATH_INIT): Add. * c-opts.c (c_common_missing_argument, c_common_handle_option): Add -F argument processing. * c.opt: Add -F argument processing. * gcc.c (trad_capable_cpp): Add -F argument processing. * cppfiles.c (find_file_in_dir): Update to use construct callback. (search_path_exhausted, cpp_get_path, cpp_get_buffer, cpp_get_prev): New. (_cpp_find_file): Use search_path_exhausted. (make_cpp_dir): Initialize construct to 0. * cpplib.h (missing_header_cb cpp_get_path, cpp_get_buffer, cpp_get_file, cpp_get_prev): New. (cpp_callbacks): Add missing_header (cpp_dir): Add construct. * target-def.h: (TARGET_OPTF): New. * hooks.c (hook_void_int, hook_void_charptr): Add. * hooks.h (hook_void_int, hook_void_charptr): Add. * Makefile.in (c-incpath.o) : Add $(TARGET_H) and $(MACHMODE_H) dependencies. * doc/invoke.texi (Darwin Options): Document -F. * doc/tm.texi (TARGET_EXTRA_INCLUDES): Add. (TARGET_OPTF): Add. * fix-header.c (target_c_incpath): Add. * config/darwin-c.c: Add c-incpath.h include. (using_frameworks, find_subframework_file, find_subframework_header, add_system_framework_path, frameworks_in_use, num_frameworks, max_frameworks, add_framework, find_framework, struct framework_header, framework_header_dirs, framework_construct_pathname, find_subframework_file, add_system_framework_path, add_framework_path, framework_defaults, darwin_register_frameworks, find_subframework_header): Add. * config/darwin.h (TARGET_EXTRA_INCLUDES, TARGET_OPTF): New. (TARGET_OPTION_TRANSLATE_TABLE): Add -framework support. (CPP_SPEC): Add __APPLE_CC__ support. * t-darwin (darwin-c.o): Add c-incpath.h dependency. From-SVN: r78875
2004-03-04 00:18:54 +00:00
return true;
Add framework support for darwin. * c-incpath.c: Include target.h and machmode.h. (add_path): Use a consistent style for cpp_dir. Initialize p->construct to 0. (add_cpp_dir_path): New. (register_include_chains): Add use of extra_includes callback. (hook_void_int): Add. (target_c_incpath): Add. * c-incpath.h (add_cpp_dir_path): New. (target_c_incpath_s): Add. (target_c_incpath): Add. (C_INCPATH_INIT): Add. * c-opts.c (c_common_missing_argument, c_common_handle_option): Add -F argument processing. * c.opt: Add -F argument processing. * gcc.c (trad_capable_cpp): Add -F argument processing. * cppfiles.c (find_file_in_dir): Update to use construct callback. (search_path_exhausted, cpp_get_path, cpp_get_buffer, cpp_get_prev): New. (_cpp_find_file): Use search_path_exhausted. (make_cpp_dir): Initialize construct to 0. * cpplib.h (missing_header_cb cpp_get_path, cpp_get_buffer, cpp_get_file, cpp_get_prev): New. (cpp_callbacks): Add missing_header (cpp_dir): Add construct. * target-def.h: (TARGET_OPTF): New. * hooks.c (hook_void_int, hook_void_charptr): Add. * hooks.h (hook_void_int, hook_void_charptr): Add. * Makefile.in (c-incpath.o) : Add $(TARGET_H) and $(MACHMODE_H) dependencies. * doc/invoke.texi (Darwin Options): Document -F. * doc/tm.texi (TARGET_EXTRA_INCLUDES): Add. (TARGET_OPTF): Add. * fix-header.c (target_c_incpath): Add. * config/darwin-c.c: Add c-incpath.h include. (using_frameworks, find_subframework_file, find_subframework_header, add_system_framework_path, frameworks_in_use, num_frameworks, max_frameworks, add_framework, find_framework, struct framework_header, framework_header_dirs, framework_construct_pathname, find_subframework_file, add_system_framework_path, add_framework_path, framework_defaults, darwin_register_frameworks, find_subframework_header): Add. * config/darwin.h (TARGET_EXTRA_INCLUDES, TARGET_OPTF): New. (TARGET_OPTION_TRANSLATE_TABLE): Add -framework support. (CPP_SPEC): Add __APPLE_CC__ support. * t-darwin (darwin-c.o): Add c-incpath.h dependency. From-SVN: r78875
2004-03-04 00:18:54 +00:00
if (open_file (file))
return true;
if (file->err_no != ENOENT)
{
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
open_file_failed (pfile, file, 0, loc);
Add framework support for darwin. * c-incpath.c: Include target.h and machmode.h. (add_path): Use a consistent style for cpp_dir. Initialize p->construct to 0. (add_cpp_dir_path): New. (register_include_chains): Add use of extra_includes callback. (hook_void_int): Add. (target_c_incpath): Add. * c-incpath.h (add_cpp_dir_path): New. (target_c_incpath_s): Add. (target_c_incpath): Add. (C_INCPATH_INIT): Add. * c-opts.c (c_common_missing_argument, c_common_handle_option): Add -F argument processing. * c.opt: Add -F argument processing. * gcc.c (trad_capable_cpp): Add -F argument processing. * cppfiles.c (find_file_in_dir): Update to use construct callback. (search_path_exhausted, cpp_get_path, cpp_get_buffer, cpp_get_prev): New. (_cpp_find_file): Use search_path_exhausted. (make_cpp_dir): Initialize construct to 0. * cpplib.h (missing_header_cb cpp_get_path, cpp_get_buffer, cpp_get_file, cpp_get_prev): New. (cpp_callbacks): Add missing_header (cpp_dir): Add construct. * target-def.h: (TARGET_OPTF): New. * hooks.c (hook_void_int, hook_void_charptr): Add. * hooks.h (hook_void_int, hook_void_charptr): Add. * Makefile.in (c-incpath.o) : Add $(TARGET_H) and $(MACHMODE_H) dependencies. * doc/invoke.texi (Darwin Options): Document -F. * doc/tm.texi (TARGET_EXTRA_INCLUDES): Add. (TARGET_OPTF): Add. * fix-header.c (target_c_incpath): Add. * config/darwin-c.c: Add c-incpath.h include. (using_frameworks, find_subframework_file, find_subframework_header, add_system_framework_path, frameworks_in_use, num_frameworks, max_frameworks, add_framework, find_framework, struct framework_header, framework_header_dirs, framework_construct_pathname, find_subframework_file, add_system_framework_path, add_framework_path, framework_defaults, darwin_register_frameworks, find_subframework_header): Add. * config/darwin.h (TARGET_EXTRA_INCLUDES, TARGET_OPTF): New. (TARGET_OPTION_TRANSLATE_TABLE): Add -framework support. (CPP_SPEC): Add __APPLE_CC__ support. * t-darwin (darwin-c.o): Add c-incpath.h dependency. From-SVN: r78875
2004-03-04 00:18:54 +00:00
return true;
}
/* We copy the path name onto an obstack partly so that we don't
leak the memory, but mostly so that we don't fragment the
heap. */
copy = (char *) obstack_copy0 (&pfile->nonexistent_file_ob, path,
strlen (path));
Add framework support for darwin. * c-incpath.c: Include target.h and machmode.h. (add_path): Use a consistent style for cpp_dir. Initialize p->construct to 0. (add_cpp_dir_path): New. (register_include_chains): Add use of extra_includes callback. (hook_void_int): Add. (target_c_incpath): Add. * c-incpath.h (add_cpp_dir_path): New. (target_c_incpath_s): Add. (target_c_incpath): Add. (C_INCPATH_INIT): Add. * c-opts.c (c_common_missing_argument, c_common_handle_option): Add -F argument processing. * c.opt: Add -F argument processing. * gcc.c (trad_capable_cpp): Add -F argument processing. * cppfiles.c (find_file_in_dir): Update to use construct callback. (search_path_exhausted, cpp_get_path, cpp_get_buffer, cpp_get_prev): New. (_cpp_find_file): Use search_path_exhausted. (make_cpp_dir): Initialize construct to 0. * cpplib.h (missing_header_cb cpp_get_path, cpp_get_buffer, cpp_get_file, cpp_get_prev): New. (cpp_callbacks): Add missing_header (cpp_dir): Add construct. * target-def.h: (TARGET_OPTF): New. * hooks.c (hook_void_int, hook_void_charptr): Add. * hooks.h (hook_void_int, hook_void_charptr): Add. * Makefile.in (c-incpath.o) : Add $(TARGET_H) and $(MACHMODE_H) dependencies. * doc/invoke.texi (Darwin Options): Document -F. * doc/tm.texi (TARGET_EXTRA_INCLUDES): Add. (TARGET_OPTF): Add. * fix-header.c (target_c_incpath): Add. * config/darwin-c.c: Add c-incpath.h include. (using_frameworks, find_subframework_file, find_subframework_header, add_system_framework_path, frameworks_in_use, num_frameworks, max_frameworks, add_framework, find_framework, struct framework_header, framework_header_dirs, framework_construct_pathname, find_subframework_file, add_system_framework_path, add_framework_path, framework_defaults, darwin_register_frameworks, find_subframework_header): Add. * config/darwin.h (TARGET_EXTRA_INCLUDES, TARGET_OPTF): New. (TARGET_OPTION_TRANSLATE_TABLE): Add -framework support. (CPP_SPEC): Add __APPLE_CC__ support. * t-darwin (darwin-c.o): Add c-incpath.h dependency. From-SVN: r78875
2004-03-04 00:18:54 +00:00
free (path);
pp = htab_find_slot_with_hash (pfile->nonexistent_file_hash,
copy, hv, INSERT);
*pp = copy;
Add framework support for darwin. * c-incpath.c: Include target.h and machmode.h. (add_path): Use a consistent style for cpp_dir. Initialize p->construct to 0. (add_cpp_dir_path): New. (register_include_chains): Add use of extra_includes callback. (hook_void_int): Add. (target_c_incpath): Add. * c-incpath.h (add_cpp_dir_path): New. (target_c_incpath_s): Add. (target_c_incpath): Add. (C_INCPATH_INIT): Add. * c-opts.c (c_common_missing_argument, c_common_handle_option): Add -F argument processing. * c.opt: Add -F argument processing. * gcc.c (trad_capable_cpp): Add -F argument processing. * cppfiles.c (find_file_in_dir): Update to use construct callback. (search_path_exhausted, cpp_get_path, cpp_get_buffer, cpp_get_prev): New. (_cpp_find_file): Use search_path_exhausted. (make_cpp_dir): Initialize construct to 0. * cpplib.h (missing_header_cb cpp_get_path, cpp_get_buffer, cpp_get_file, cpp_get_prev): New. (cpp_callbacks): Add missing_header (cpp_dir): Add construct. * target-def.h: (TARGET_OPTF): New. * hooks.c (hook_void_int, hook_void_charptr): Add. * hooks.h (hook_void_int, hook_void_charptr): Add. * Makefile.in (c-incpath.o) : Add $(TARGET_H) and $(MACHMODE_H) dependencies. * doc/invoke.texi (Darwin Options): Document -F. * doc/tm.texi (TARGET_EXTRA_INCLUDES): Add. (TARGET_OPTF): Add. * fix-header.c (target_c_incpath): Add. * config/darwin-c.c: Add c-incpath.h include. (using_frameworks, find_subframework_file, find_subframework_header, add_system_framework_path, frameworks_in_use, num_frameworks, max_frameworks, add_framework, find_framework, struct framework_header, framework_header_dirs, framework_construct_pathname, find_subframework_file, add_system_framework_path, add_framework_path, framework_defaults, darwin_register_frameworks, find_subframework_header): Add. * config/darwin.h (TARGET_EXTRA_INCLUDES, TARGET_OPTF): New. (TARGET_OPTION_TRANSLATE_TABLE): Add -framework support. (CPP_SPEC): Add __APPLE_CC__ support. * t-darwin (darwin-c.o): Add c-incpath.h dependency. From-SVN: r78875
2004-03-04 00:18:54 +00:00
file->path = file->name;
}
else
{
file->err_no = ENOENT;
file->path = NULL;
}
return false;
}
/* Return true iff the missing_header callback found the given HEADER. */
Add framework support for darwin. * c-incpath.c: Include target.h and machmode.h. (add_path): Use a consistent style for cpp_dir. Initialize p->construct to 0. (add_cpp_dir_path): New. (register_include_chains): Add use of extra_includes callback. (hook_void_int): Add. (target_c_incpath): Add. * c-incpath.h (add_cpp_dir_path): New. (target_c_incpath_s): Add. (target_c_incpath): Add. (C_INCPATH_INIT): Add. * c-opts.c (c_common_missing_argument, c_common_handle_option): Add -F argument processing. * c.opt: Add -F argument processing. * gcc.c (trad_capable_cpp): Add -F argument processing. * cppfiles.c (find_file_in_dir): Update to use construct callback. (search_path_exhausted, cpp_get_path, cpp_get_buffer, cpp_get_prev): New. (_cpp_find_file): Use search_path_exhausted. (make_cpp_dir): Initialize construct to 0. * cpplib.h (missing_header_cb cpp_get_path, cpp_get_buffer, cpp_get_file, cpp_get_prev): New. (cpp_callbacks): Add missing_header (cpp_dir): Add construct. * target-def.h: (TARGET_OPTF): New. * hooks.c (hook_void_int, hook_void_charptr): Add. * hooks.h (hook_void_int, hook_void_charptr): Add. * Makefile.in (c-incpath.o) : Add $(TARGET_H) and $(MACHMODE_H) dependencies. * doc/invoke.texi (Darwin Options): Document -F. * doc/tm.texi (TARGET_EXTRA_INCLUDES): Add. (TARGET_OPTF): Add. * fix-header.c (target_c_incpath): Add. * config/darwin-c.c: Add c-incpath.h include. (using_frameworks, find_subframework_file, find_subframework_header, add_system_framework_path, frameworks_in_use, num_frameworks, max_frameworks, add_framework, find_framework, struct framework_header, framework_header_dirs, framework_construct_pathname, find_subframework_file, add_system_framework_path, add_framework_path, framework_defaults, darwin_register_frameworks, find_subframework_header): Add. * config/darwin.h (TARGET_EXTRA_INCLUDES, TARGET_OPTF): New. (TARGET_OPTION_TRANSLATE_TABLE): Add -framework support. (CPP_SPEC): Add __APPLE_CC__ support. * t-darwin (darwin-c.o): Add c-incpath.h dependency. From-SVN: r78875
2004-03-04 00:18:54 +00:00
static bool
search_path_exhausted (cpp_reader *pfile, const char *header, _cpp_file *file)
{
missing_header_cb func = pfile->cb.missing_header;
/* When the regular search path doesn't work, try context dependent
headers search paths. */
if (func
&& file->dir == NULL)
{
if ((file->path = func (pfile, header, &file->dir)) != NULL)
Add framework support for darwin. * c-incpath.c: Include target.h and machmode.h. (add_path): Use a consistent style for cpp_dir. Initialize p->construct to 0. (add_cpp_dir_path): New. (register_include_chains): Add use of extra_includes callback. (hook_void_int): Add. (target_c_incpath): Add. * c-incpath.h (add_cpp_dir_path): New. (target_c_incpath_s): Add. (target_c_incpath): Add. (C_INCPATH_INIT): Add. * c-opts.c (c_common_missing_argument, c_common_handle_option): Add -F argument processing. * c.opt: Add -F argument processing. * gcc.c (trad_capable_cpp): Add -F argument processing. * cppfiles.c (find_file_in_dir): Update to use construct callback. (search_path_exhausted, cpp_get_path, cpp_get_buffer, cpp_get_prev): New. (_cpp_find_file): Use search_path_exhausted. (make_cpp_dir): Initialize construct to 0. * cpplib.h (missing_header_cb cpp_get_path, cpp_get_buffer, cpp_get_file, cpp_get_prev): New. (cpp_callbacks): Add missing_header (cpp_dir): Add construct. * target-def.h: (TARGET_OPTF): New. * hooks.c (hook_void_int, hook_void_charptr): Add. * hooks.h (hook_void_int, hook_void_charptr): Add. * Makefile.in (c-incpath.o) : Add $(TARGET_H) and $(MACHMODE_H) dependencies. * doc/invoke.texi (Darwin Options): Document -F. * doc/tm.texi (TARGET_EXTRA_INCLUDES): Add. (TARGET_OPTF): Add. * fix-header.c (target_c_incpath): Add. * config/darwin-c.c: Add c-incpath.h include. (using_frameworks, find_subframework_file, find_subframework_header, add_system_framework_path, frameworks_in_use, num_frameworks, max_frameworks, add_framework, find_framework, struct framework_header, framework_header_dirs, framework_construct_pathname, find_subframework_file, add_system_framework_path, add_framework_path, framework_defaults, darwin_register_frameworks, find_subframework_header): Add. * config/darwin.h (TARGET_EXTRA_INCLUDES, TARGET_OPTF): New. (TARGET_OPTION_TRANSLATE_TABLE): Add -framework support. (CPP_SPEC): Add __APPLE_CC__ support. * t-darwin (darwin-c.o): Add c-incpath.h dependency. From-SVN: r78875
2004-03-04 00:18:54 +00:00
{
if (open_file (file))
return true;
free ((void *)file->path);
}
file->path = file->name;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
return false;
}
bool
_cpp_find_failed (_cpp_file *file)
{
return file->err_no != 0;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Given a filename FNAME search for such a file in the include path
starting from START_DIR. If FNAME is the empty string it is
interpreted as STDIN if START_DIR is PFILE->no_search_path.
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
If the file is not found in the file cache fall back to the O/S and
add the result to our cache.
If the file was not found in the filesystem, or there was an error
opening it, then ERR_NO is nonzero and FD is -1. If the file was
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
found, then ERR_NO is zero and FD could be -1 or an open file
descriptor. FD can be -1 if the file was found in the cache and
had previously been closed. To open it again pass the return value
to open_file().
gcc: * config.gcc (*-*-linux* | frv-*-*linux* | *-*-kfreebsd*-gnu | *-*-knetbsd*-gnu | *-*-gnu* | *-*-kopensolaris*-gnu): Use glibc-c.o in c_target_objs and cxx_target_objs. Use t-glibc in tmake_file. Set target_has_targetcm. (tilegx-*-linux*, tilepro-*-linux*): Append to c_target_objs and cxx_target_objs rather than overriding previous value. * config/glibc-c.c, config/t-glibc: New. * doc/tm.texi.in (TARGET_C_PREINCLUDE): New @hook. * doc/tm.texi: Regenerate. * hooks.c (hook_constcharptr_void_null): New. * hooks.h (hook_constcharptr_void_null): Declare. gcc/c-family: * c-common.h (pch_cpp_save_state): Declare. * c-target.def (c_preinclude): New hook. * c-opts.c (done_preinclude): New. (push_command_line_include): Handle default preincluded header. (cb_file_change): Call pch_cpp_save_state when calling push_command_line_include. * c-pch.c (pch_ready_to_save_cpp_state, pch_cpp_state_saved) (pch_cpp_save_state): New. (pch_init): Call pch_cpp_save_state conditionally, instead of calling cpp_save_state. gcc/testsuite: * gcc.dg/c99-predef-1.c: New test. * gcc.dg/cpp/cmdlne-dU-1.c, gcc.dg/cpp/cmdlne-dU-2.c, gcc.dg/cpp/cmdlne-dU-3.c, gcc.dg/cpp/cmdlne-dU-4.c, gcc.dg/cpp/cmdlne-dU-5.c, gcc.dg/cpp/cmdlne-dU-6.c, gcc.dg/cpp/cmdlne-dU-7.c, gcc.dg/cpp/cmdlne-dU-8.c, gcc.dg/cpp/cmdlne-dU-9.c, gcc.dg/cpp/cmdlne-dU-10.c, gcc.dg/cpp/cmdlne-dU-11.c, gcc.dg/cpp/cmdlne-dU-12.c, gcc.dg/cpp/cmdlne-dU-13.c, gcc.dg/cpp/cmdlne-dU-14.c, gcc.dg/cpp/cmdlne-dU-15.c, gcc.dg/cpp/cmdlne-dU-16.c, gcc.dg/cpp/cmdlne-dU-17.c, gcc.dg/cpp/cmdlne-dU-18.c, gcc.dg/cpp/cmdlne-dU-19.c, gcc.dg/cpp/cmdlne-dU-20.c, gcc.dg/cpp/cmdlne-dU-21.c, gcc.dg/cpp/cmdlne-dU-22.c, gcc.dg/cpp/mi5.c, gcc.dg/cpp/multiline.c: Add -nostdinc to dg-options. libcpp: * files.c (struct _cpp_file): Add implicit_preinclude. (pch_open_file): Allow a previously opened implicitly included file. (_cpp_find_file): Add implicit_preinclude argument. Free file and do not call open_file_failed if implicit_preinclude. Store implicit_preinclude value. (_cpp_stack_include, _cpp_fake_include, _cpp_compare_file_date): Update calls to _cpp_find_file. (_cpp_stack_include): Handle IT_DEFAULT. (cpp_push_default_include): New. * include/cpplib.h (cpp_push_default_include): Declare. * init.c (cpp_read_main_file): Update call to _cpp_find_file. * internal.h (enum include_type): Add IT_DEFAULT. (_cpp_find_file): Update prototype. From-SVN: r192715
2012-10-23 15:55:55 +01:00
If KIND is _cpp_FFK_PRE_INCLUDE then it is OK for the file to be
missing. If present, it is OK for a precompiled header to be
included after it.
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
Use LOC as the location for any errors. */
_cpp_file *
gcc: * config.gcc (*-*-linux* | frv-*-*linux* | *-*-kfreebsd*-gnu | *-*-knetbsd*-gnu | *-*-gnu* | *-*-kopensolaris*-gnu): Use glibc-c.o in c_target_objs and cxx_target_objs. Use t-glibc in tmake_file. Set target_has_targetcm. (tilegx-*-linux*, tilepro-*-linux*): Append to c_target_objs and cxx_target_objs rather than overriding previous value. * config/glibc-c.c, config/t-glibc: New. * doc/tm.texi.in (TARGET_C_PREINCLUDE): New @hook. * doc/tm.texi: Regenerate. * hooks.c (hook_constcharptr_void_null): New. * hooks.h (hook_constcharptr_void_null): Declare. gcc/c-family: * c-common.h (pch_cpp_save_state): Declare. * c-target.def (c_preinclude): New hook. * c-opts.c (done_preinclude): New. (push_command_line_include): Handle default preincluded header. (cb_file_change): Call pch_cpp_save_state when calling push_command_line_include. * c-pch.c (pch_ready_to_save_cpp_state, pch_cpp_state_saved) (pch_cpp_save_state): New. (pch_init): Call pch_cpp_save_state conditionally, instead of calling cpp_save_state. gcc/testsuite: * gcc.dg/c99-predef-1.c: New test. * gcc.dg/cpp/cmdlne-dU-1.c, gcc.dg/cpp/cmdlne-dU-2.c, gcc.dg/cpp/cmdlne-dU-3.c, gcc.dg/cpp/cmdlne-dU-4.c, gcc.dg/cpp/cmdlne-dU-5.c, gcc.dg/cpp/cmdlne-dU-6.c, gcc.dg/cpp/cmdlne-dU-7.c, gcc.dg/cpp/cmdlne-dU-8.c, gcc.dg/cpp/cmdlne-dU-9.c, gcc.dg/cpp/cmdlne-dU-10.c, gcc.dg/cpp/cmdlne-dU-11.c, gcc.dg/cpp/cmdlne-dU-12.c, gcc.dg/cpp/cmdlne-dU-13.c, gcc.dg/cpp/cmdlne-dU-14.c, gcc.dg/cpp/cmdlne-dU-15.c, gcc.dg/cpp/cmdlne-dU-16.c, gcc.dg/cpp/cmdlne-dU-17.c, gcc.dg/cpp/cmdlne-dU-18.c, gcc.dg/cpp/cmdlne-dU-19.c, gcc.dg/cpp/cmdlne-dU-20.c, gcc.dg/cpp/cmdlne-dU-21.c, gcc.dg/cpp/cmdlne-dU-22.c, gcc.dg/cpp/mi5.c, gcc.dg/cpp/multiline.c: Add -nostdinc to dg-options. libcpp: * files.c (struct _cpp_file): Add implicit_preinclude. (pch_open_file): Allow a previously opened implicitly included file. (_cpp_find_file): Add implicit_preinclude argument. Free file and do not call open_file_failed if implicit_preinclude. Store implicit_preinclude value. (_cpp_stack_include, _cpp_fake_include, _cpp_compare_file_date): Update calls to _cpp_find_file. (_cpp_stack_include): Handle IT_DEFAULT. (cpp_push_default_include): New. * include/cpplib.h (cpp_push_default_include): Declare. * init.c (cpp_read_main_file): Update call to _cpp_find_file. * internal.h (enum include_type): Add IT_DEFAULT. (_cpp_find_file): Update prototype. From-SVN: r192715
2012-10-23 15:55:55 +01:00
_cpp_find_file (cpp_reader *pfile, const char *fname, cpp_dir *start_dir,
int angle_brackets, _cpp_find_file_kind kind, location_t loc)
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
{
bool invalid_pch = false;
bool saw_bracket_include = false;
bool saw_quote_include = false;
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
bool saw_embed_include = false;
struct cpp_dir *found_in_cache = NULL;
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
bool is_embed = kind == _cpp_FFK_EMBED || kind == _cpp_FFK_HAS_EMBED;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Ensure we get no confusion between cached files and directories. */
if (start_dir == NULL)
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
cpp_error_at (pfile, CPP_DL_ICE, loc, "NULL directory in find_file");
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
void **hash_slot
= htab_find_slot_with_hash (pfile->file_hash, fname,
htab_hash_string (fname), INSERT);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* First check the cache before we resort to memory allocation. */
cpp_file_hash_entry *entry
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
= search_cache ((struct cpp_file_hash_entry *) *hash_slot, start_dir,
is_embed);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (entry)
{
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
if (entry->u.file->deferred_error
&& (kind == _cpp_FFK_NORMAL || kind == _cpp_FFK_EMBED))
{
open_file_failed (pfile, entry->u.file, angle_brackets, loc);
entry->u.file->deferred_error = false;
}
return entry->u.file;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
_cpp_file *file = make_cpp_file (start_dir, fname);
file->implicit_preinclude
= (kind == _cpp_FFK_PRE_INCLUDE
|| (pfile->buffer && pfile->buffer->file->implicit_preinclude));
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
file->embed = is_embed;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (kind == _cpp_FFK_FAKE)
file->dont_read = true;
else
/* Try each path in the include chain. */
for (;;)
{
if (find_file_in_dir (pfile, file, &invalid_pch, loc))
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
break;
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
if (is_embed
&& file->dir == start_dir
&& start_dir != pfile->embed_include
&& start_dir != &pfile->no_search_path)
file->dir = pfile->embed_include;
else
file->dir = file->dir->next;
if (file->dir == NULL)
{
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
if (!is_embed
&& search_path_exhausted (pfile, fname, file))
{
/* Although this file must not go in the cache,
because the file found might depend on things (like
the current file) that aren't represented in the
cache, it still has to go in the list of all files
so that #import works. */
file->next_file = pfile->all_files;
pfile->all_files = file;
if (*hash_slot == NULL)
{
/* If *hash_slot is NULL, the above
htab_find_slot_with_hash call just created the
Add -Wuse-after-free [PR80532]. gcc/c-family/ChangeLog PR tree-optimization/80532 * c.opt (-Wuse-after-free): New options. gcc/ChangeLog: PR tree-optimization/80532 * common.opt (-Wuse-after-free): New options. * diagnostic-spec.c (nowarn_spec_t::nowarn_spec_t): Handle OPT_Wreturn_local_addr and OPT_Wuse_after_free_. * diagnostic-spec.h (NW_DANGLING): New enumerator. * doc/invoke.texi (-Wuse-after-free): Document new option. * gimple-ssa-warn-access.cc (pass_waccess::check_call): Rename... (pass_waccess::check_call_access): ...to this. (pass_waccess::check): Rename... (pass_waccess::check_block): ...to this. (pass_waccess::check_pointer_uses): New function. (pass_waccess::gimple_call_return_arg): New function. (pass_waccess::warn_invalid_pointer): New function. (pass_waccess::check_builtin): Handle free and realloc. (gimple_use_after_inval_p): New function. (get_realloc_lhs): New function. (maybe_warn_mismatched_realloc): New function. (pointers_related_p): New function. (pass_waccess::check_call): Call check_pointer_uses. (pass_waccess::execute): Compute and free dominance info. libcpp/ChangeLog: * files.c (_cpp_find_file): Substitute a valid pointer for an invalid one to avoid -Wuse-after-free. libiberty/ChangeLog: * regex.c: Suppress -Wuse-after-free. gcc/testsuite/ChangeLog: PR tree-optimization/80532 * gcc.dg/Wmismatched-dealloc-2.c: Avoid -Wuse-after-free. * gcc.dg/Wmismatched-dealloc-3.c: Same. * gcc.dg/analyzer/file-1.c: Prune expected warning. * gcc.dg/analyzer/file-2.c: Same. * gcc.dg/attr-alloc_size-6.c: Disable -Wuse-after-free. * gcc.dg/attr-alloc_size-7.c: Same. * c-c++-common/Wuse-after-free-2.c: New test. * c-c++-common/Wuse-after-free-3.c: New test. * c-c++-common/Wuse-after-free-4.c: New test. * c-c++-common/Wuse-after-free-5.c: New test. * c-c++-common/Wuse-after-free-6.c: New test. * c-c++-common/Wuse-after-free-7.c: New test. * c-c++-common/Wuse-after-free.c: New test. * g++.dg/warn/Wmismatched-dealloc-3.C: New test. * g++.dg/warn/Wuse-after-free.C: New test.
2022-01-15 16:37:54 -07:00
slot, but we aren't going to store there anything
of use, so need to remove the newly created entry.
htab_clear_slot requires that it is non-NULL, so
store some non-NULL but valid pointer there,
htab_clear_slot will immediately overwrite it. */
*hash_slot = file;
htab_clear_slot (pfile->file_hash, hash_slot);
}
return file;
}
if (invalid_pch)
{
cpp_error (pfile, CPP_DL_ERROR,
"one or more PCH files were found,"
" but they were invalid");
if (!cpp_get_options (pfile)->warn_invalid_pch)
cpp_error (pfile, CPP_DL_NOTE,
"use -Winvalid-pch for more information");
}
if (kind == _cpp_FFK_PRE_INCLUDE)
{
free ((char *) file->name);
free (file);
if (*hash_slot == NULL)
{
/* See comment on the above htab_clear_slot call. */
Add -Wuse-after-free [PR80532]. gcc/c-family/ChangeLog PR tree-optimization/80532 * c.opt (-Wuse-after-free): New options. gcc/ChangeLog: PR tree-optimization/80532 * common.opt (-Wuse-after-free): New options. * diagnostic-spec.c (nowarn_spec_t::nowarn_spec_t): Handle OPT_Wreturn_local_addr and OPT_Wuse_after_free_. * diagnostic-spec.h (NW_DANGLING): New enumerator. * doc/invoke.texi (-Wuse-after-free): Document new option. * gimple-ssa-warn-access.cc (pass_waccess::check_call): Rename... (pass_waccess::check_call_access): ...to this. (pass_waccess::check): Rename... (pass_waccess::check_block): ...to this. (pass_waccess::check_pointer_uses): New function. (pass_waccess::gimple_call_return_arg): New function. (pass_waccess::warn_invalid_pointer): New function. (pass_waccess::check_builtin): Handle free and realloc. (gimple_use_after_inval_p): New function. (get_realloc_lhs): New function. (maybe_warn_mismatched_realloc): New function. (pointers_related_p): New function. (pass_waccess::check_call): Call check_pointer_uses. (pass_waccess::execute): Compute and free dominance info. libcpp/ChangeLog: * files.c (_cpp_find_file): Substitute a valid pointer for an invalid one to avoid -Wuse-after-free. libiberty/ChangeLog: * regex.c: Suppress -Wuse-after-free. gcc/testsuite/ChangeLog: PR tree-optimization/80532 * gcc.dg/Wmismatched-dealloc-2.c: Avoid -Wuse-after-free. * gcc.dg/Wmismatched-dealloc-3.c: Same. * gcc.dg/analyzer/file-1.c: Prune expected warning. * gcc.dg/analyzer/file-2.c: Same. * gcc.dg/attr-alloc_size-6.c: Disable -Wuse-after-free. * gcc.dg/attr-alloc_size-7.c: Same. * c-c++-common/Wuse-after-free-2.c: New test. * c-c++-common/Wuse-after-free-3.c: New test. * c-c++-common/Wuse-after-free-4.c: New test. * c-c++-common/Wuse-after-free-5.c: New test. * c-c++-common/Wuse-after-free-6.c: New test. * c-c++-common/Wuse-after-free-7.c: New test. * c-c++-common/Wuse-after-free.c: New test. * g++.dg/warn/Wmismatched-dealloc-3.C: New test. * g++.dg/warn/Wuse-after-free.C: New test.
2022-01-15 16:37:54 -07:00
*hash_slot = &hash_slot;
htab_clear_slot (pfile->file_hash, hash_slot);
}
return NULL;
}
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
if (kind != _cpp_FFK_HAS_INCLUDE && kind != _cpp_FFK_HAS_EMBED)
[PR 80005] Fix __has_include __has_include is funky in that it is macro-like from the POV of #ifdef and friends, but lexes its parenthesize argument #include-like. We were failing the second part of that, because we used a forwarding macro to an internal name, and hence always lexed the argument in macro-parameter context. We componded that by not setting the right flag when lexing, so it didn't even know. Mostly users got lucky. This reimplements the handline. 1) Remove the forwarding, but declare object-like macros that expand to themselves. This satisfies the #ifdef requirement 2) Correctly set angled_brackets when lexing the parameter. This tells the lexer (a) <...> is a header name and (b) "..." is too (not a string). 3) Remove the in__has_include lexer state, just tell find_file that that's what's happenning, so it doesn't emit an error. We lose the (undocumented) ability to #undef __has_include. That may well have been an accident of implementation. There are no tests for it. We gain __has_include behaviour for all users of the preprocessors -- not just the C-family ones that defined a forwarding macro. libcpp/ PR preprocessor/80005 * include/cpplib.h (BT_HAS_ATTRIBUTE): Fix comment. * internal.h (struct lexer_state): Delete in__has_include field. (struct spec_nodes): Rename n__has_include{,_next}__ fields. (_cpp_defined_macro_p): New. (_cpp_find_file): Add has_include parm. * directives.c (lex_macro_node): Combine defined, __has_inline{,_next} checking. (do_ifdef, do_ifndef): Use _cpp_defined_macro_p. (_cpp_init_directives): Refactor. * expr.c (parse_defined): Use _cpp_defined_macro_p. (eval_token): Adjust parse_has_include calls. (parse_has_include): Add OP parameter. Reimplement. * files.c (_cpp_find_file): Add HAS_INCLUDE parm. Use it to inhibit error message. (_cpp_stack_include): Adjust _cpp_find_file call. (_cpp_fake_include, _cpp_compare_file_date): Likewise. (open_file_failed): Remove in__has_include check. (_cpp_has_header): Adjust _cpp_find_file call. * identifiers.c (_cpp_init_hashtable): Don't init __has_include{,_next} here ... * init.c (cpp_init_builtins): ... init them here. Define as macros. (cpp_read_main_file): Adjust _cpp_find_file call. * pch.c (cpp_read_state): Adjust __has_include{,_next} access. * traditional.c (_cpp_scan_out_locgical_line): Likewise. gcc/c-family/ PR preprocessor/80005 * c-cppbuiltins.c (c_cpp_builtins): Don't define __has_include{,_next}. gcc/testsuite/ PR preprocessor/80005 * g++.dg/cpp1y/feat-cxx14.C: Adjust. * g++.dg/cpp1z/feat-cxx17.C: Adjust. * g++.dg/cpp2a/feat-cxx2a.C: Adjust. * g++.dg/cpp/pr80005.C: New.
2020-01-20 05:39:59 -08:00
open_file_failed (pfile, file, angle_brackets, loc);
else
file->deferred_error = true;
break;
}
/* Only check the cache for the starting location (done above)
and the quote and bracket chain heads because there are no
other possible starting points for searches. */
if (file->dir == pfile->bracket_include)
saw_bracket_include = true;
else if (file->dir == pfile->quote_include)
saw_quote_include = true;
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
else if (file->dir == pfile->embed_include)
saw_embed_include = true;
else
continue;
entry
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
= search_cache ((struct cpp_file_hash_entry *) *hash_slot,
file->dir, is_embed);
if (entry)
{
found_in_cache = file->dir;
break;
}
}
Makefile.am: Add makedepend. libcpp: * Makefile.am: Add makedepend. * Makefile.in, aclocal.m4: Regenerate. * charset.c: Insert a space to avoid a warning. * directives.c: Include mkdeps.h. (_cpp_handle_directive): Reenable macro expander if appropriate. (undefine_macros): Inline body of _cpp_free_definition for speed. Do not call undef callback or _cpp_warn_if_unused_macro. (cpp_get_deps): New interface. * files.c (search_cache): Add pfile argument. Check for file that would be found by "" or <> search here... (_cpp_find_file): ...not here. Correct recorded start_dir of files found by directory-of-current-file search that would be found by "" or <> search. * init.c (cpp_add_dependency_target): Delete. * internal.h (struct lexer_state): Add discarding_output flag. * lex.c (lex_identifier): Compute hash function while scanning. * macro.c (cpp_scan_nooutput): Disable macro expansion outside directives. * makedepend.c: New file. * mkdeps.c (struct deps): Add vpath vector. (apply_vpath, deps_add_vpath): New function. (deps_free): Free vpath vector. (deps_add_dep, deps_add_target): Use apply_vpath. * symtab.c (calc_hash): Use HT_HASHSTEP and HT_FINISH. (ht_lookup_with_hash): New function. * cpplib.h, mkdeps.h: Update prototypes. * symtab.h: Update prototypes. (HT_HASHSTEP, HT_FINISH): New macros. gcc: * Makefile.in (MKDEPS_H): New shorthand. (c-opts.o): Update dependencies. * c-opts.c: Include mkdeps.h. (handle_deferred_opts): Use cpp_get_deps and deps_add_target, not cpp_add_dependency_target. From-SVN: r82654
2004-06-05 20:58:06 +00:00
if (entry)
{
/* Cache for START_DIR too, sharing the _cpp_file structure. */
free ((char *) file->name);
free (file);
file = entry->u.file;
}
else
{
/* This is a new file; put it in the list. */
file->next_file = pfile->all_files;
pfile->all_files = file;
Makefile.am: Add makedepend. libcpp: * Makefile.am: Add makedepend. * Makefile.in, aclocal.m4: Regenerate. * charset.c: Insert a space to avoid a warning. * directives.c: Include mkdeps.h. (_cpp_handle_directive): Reenable macro expander if appropriate. (undefine_macros): Inline body of _cpp_free_definition for speed. Do not call undef callback or _cpp_warn_if_unused_macro. (cpp_get_deps): New interface. * files.c (search_cache): Add pfile argument. Check for file that would be found by "" or <> search here... (_cpp_find_file): ...not here. Correct recorded start_dir of files found by directory-of-current-file search that would be found by "" or <> search. * init.c (cpp_add_dependency_target): Delete. * internal.h (struct lexer_state): Add discarding_output flag. * lex.c (lex_identifier): Compute hash function while scanning. * macro.c (cpp_scan_nooutput): Disable macro expansion outside directives. * makedepend.c: New file. * mkdeps.c (struct deps): Add vpath vector. (apply_vpath, deps_add_vpath): New function. (deps_free): Free vpath vector. (deps_add_dep, deps_add_target): Use apply_vpath. * symtab.c (calc_hash): Use HT_HASHSTEP and HT_FINISH. (ht_lookup_with_hash): New function. * cpplib.h, mkdeps.h: Update prototypes. * symtab.h: Update prototypes. (HT_HASHSTEP, HT_FINISH): New macros. gcc: * Makefile.in (MKDEPS_H): New shorthand. (c-opts.o): Update dependencies. * c-opts.c: Include mkdeps.h. (handle_deferred_opts): Use cpp_get_deps and deps_add_target, not cpp_add_dependency_target. From-SVN: r82654
2004-06-05 20:58:06 +00:00
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Store this new result in the hash table. */
entry = new_file_hash_entry (pfile);
entry->next = (struct cpp_file_hash_entry *) *hash_slot;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
entry->start_dir = start_dir;
entry->location = loc;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
entry->u.file = file;
*hash_slot = (void *) entry;
/* If we passed the quote or bracket chain heads, cache them also.
This speeds up processing if there are lots of -I options. */
if (saw_bracket_include
&& pfile->bracket_include != start_dir
&& found_in_cache != pfile->bracket_include)
{
entry = new_file_hash_entry (pfile);
entry->next = (struct cpp_file_hash_entry *) *hash_slot;
entry->start_dir = pfile->bracket_include;
entry->location = loc;
entry->u.file = file;
*hash_slot = (void *) entry;
}
if (saw_quote_include
&& pfile->quote_include != start_dir
&& found_in_cache != pfile->quote_include)
{
entry = new_file_hash_entry (pfile);
entry->next = (struct cpp_file_hash_entry *) *hash_slot;
entry->start_dir = pfile->quote_include;
entry->location = loc;
entry->u.file = file;
*hash_slot = (void *) entry;
}
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
if (saw_embed_include
&& pfile->embed_include != start_dir
&& found_in_cache != pfile->embed_include)
{
entry = new_file_hash_entry (pfile);
entry->next = (struct cpp_file_hash_entry *) *hash_slot;
entry->start_dir = pfile->embed_include;
entry->location = loc;
entry->u.file = file;
*hash_slot = (void *) entry;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
return file;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Read a file into FILE->buffer, returning true on success.
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
If FILE->fd is something weird, like a block device, we don't want
to read it at all. Don't even try to figure out what something is,
except for plain files and block devices, since there is no
reliable portable way of doing this.
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
Use LOC for any diagnostics.
diagnostics: Support for -finput-charset [PR93067] Adds the logic to handle -finput-charset in layout_get_source_line(), so that source lines are converted from their input encodings prior to being output by diagnostics machinery. Also adds the ability to strip a UTF-8 BOM similarly. gcc/c-family/ChangeLog: PR other/93067 * c-opts.c (c_common_input_charset_cb): New function. (c_common_post_options): Call new function diagnostic_initialize_input_context(). gcc/d/ChangeLog: PR other/93067 * d-lang.cc (d_input_charset_callback): New function. (d_init): Call new function diagnostic_initialize_input_context(). gcc/fortran/ChangeLog: PR other/93067 * cpp.c (gfc_cpp_post_options): Call new function diagnostic_initialize_input_context(). gcc/ChangeLog: PR other/93067 * coretypes.h (typedef diagnostic_input_charset_callback): Declare. * diagnostic.c (diagnostic_initialize_input_context): New function. * diagnostic.h (diagnostic_initialize_input_context): Declare. * input.c (default_charset_callback): New function. (file_cache::initialize_input_context): New function. (file_cache_slot::create): Added ability to convert the input according to the input context. (file_cache::file_cache): Initialize the new input context. (class file_cache_slot): Added new m_alloc_offset member. (file_cache_slot::file_cache_slot): Initialize the new member. (file_cache_slot::~file_cache_slot): Handle potentially offset buffer. (file_cache_slot::maybe_grow): Likewise. (file_cache_slot::needs_read_p): Handle NULL fp, which is now possible. (file_cache_slot::get_next_line): Likewise. * input.h (class file_cache): Added input context member. libcpp/ChangeLog: PR other/93067 * charset.c (init_iconv_desc): Adapt to permit PFILE argument to be NULL. (_cpp_convert_input): Likewise. Also move UTF-8 BOM logic to... (cpp_check_utf8_bom): ...here. New function. (cpp_input_conversion_is_trivial): New function. * files.c (read_file_guts): Allow PFILE argument to be NULL. Add INPUT_CHARSET argument as an alternate source of this information. (read_file): Pass the new argument to read_file_guts. (cpp_get_converted_source): New function. * include/cpplib.h (struct cpp_converted_source): Declare. (cpp_get_converted_source): Declare. (cpp_input_conversion_is_trivial): Declare. (cpp_check_utf8_bom): Declare. gcc/testsuite/ChangeLog: PR other/93067 * gcc.dg/diagnostic-input-charset-1.c: New test. * gcc.dg/diagnostic-input-utf8-bom.c: New test.
2021-08-24 19:30:44 -04:00
PFILE may be NULL. In this case, no diagnostics are issued.
FIXME: Flush file cache and try again if we run out of memory. */
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
static bool
diagnostics: Support for -finput-charset [PR93067] Adds the logic to handle -finput-charset in layout_get_source_line(), so that source lines are converted from their input encodings prior to being output by diagnostics machinery. Also adds the ability to strip a UTF-8 BOM similarly. gcc/c-family/ChangeLog: PR other/93067 * c-opts.c (c_common_input_charset_cb): New function. (c_common_post_options): Call new function diagnostic_initialize_input_context(). gcc/d/ChangeLog: PR other/93067 * d-lang.cc (d_input_charset_callback): New function. (d_init): Call new function diagnostic_initialize_input_context(). gcc/fortran/ChangeLog: PR other/93067 * cpp.c (gfc_cpp_post_options): Call new function diagnostic_initialize_input_context(). gcc/ChangeLog: PR other/93067 * coretypes.h (typedef diagnostic_input_charset_callback): Declare. * diagnostic.c (diagnostic_initialize_input_context): New function. * diagnostic.h (diagnostic_initialize_input_context): Declare. * input.c (default_charset_callback): New function. (file_cache::initialize_input_context): New function. (file_cache_slot::create): Added ability to convert the input according to the input context. (file_cache::file_cache): Initialize the new input context. (class file_cache_slot): Added new m_alloc_offset member. (file_cache_slot::file_cache_slot): Initialize the new member. (file_cache_slot::~file_cache_slot): Handle potentially offset buffer. (file_cache_slot::maybe_grow): Likewise. (file_cache_slot::needs_read_p): Handle NULL fp, which is now possible. (file_cache_slot::get_next_line): Likewise. * input.h (class file_cache): Added input context member. libcpp/ChangeLog: PR other/93067 * charset.c (init_iconv_desc): Adapt to permit PFILE argument to be NULL. (_cpp_convert_input): Likewise. Also move UTF-8 BOM logic to... (cpp_check_utf8_bom): ...here. New function. (cpp_input_conversion_is_trivial): New function. * files.c (read_file_guts): Allow PFILE argument to be NULL. Add INPUT_CHARSET argument as an alternate source of this information. (read_file): Pass the new argument to read_file_guts. (cpp_get_converted_source): New function. * include/cpplib.h (struct cpp_converted_source): Declare. (cpp_get_converted_source): Declare. (cpp_input_conversion_is_trivial): Declare. (cpp_check_utf8_bom): Declare. gcc/testsuite/ChangeLog: PR other/93067 * gcc.dg/diagnostic-input-charset-1.c: New test. * gcc.dg/diagnostic-input-utf8-bom.c: New test.
2021-08-24 19:30:44 -04:00
read_file_guts (cpp_reader *pfile, _cpp_file *file, location_t loc,
const char *input_charset)
{
ssize_t size, pad, total, count;
uchar *buf;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
bool regular;
[multiple changes] 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * c-opts.c (c_common_handle_option): Add -finput-charset. * c.opt: Ditto. * cppcharset.c (one_iso88591_to_utf8): Remove. (convert_iso88591_utf8): Ditto. (conversion_tab): Remove 8859-1 converter. (_cpp_input_to_utf8): Remove. (_cpp_init_iconv_buffer): Ditto. (_cpp_close_iconv_buffer): Ditto. (_cpp_convert_input): New function. (_cpp_default_encoding): Ditto. * cpphash.h: Add/remove prototypes for above. * cppfiles.c (read_file_guts): Use _cpp_convert_input. * cppinit.c (cpp_create_reader): Use _cpp_default_encoding for narrow execution and input character sets. * cpplib.c (cpp_push_buffer): Delete uses of removed functions. * doc/cppopts.texi: Document -finput-charset. 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * gcc.c-torture/execute/wchar_t-1.c: Add -finput-charset. 2004-01-29 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * testsuite/22_locale/collate/compare/wchar_t/2.cc: Remove xfail. Use -finput-charset. * testsuite/22_locale/collate/compare/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/compare/wchar_t/wrapped_locale.cc: Ditto * testsuite/22_locale/collate/hash/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_locale.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_locale.cc: Ditto. From-SVN: r77136
2004-02-02 20:20:58 +00:00
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (S_ISBLK (file->st.st_mode))
{
diagnostics: Support for -finput-charset [PR93067] Adds the logic to handle -finput-charset in layout_get_source_line(), so that source lines are converted from their input encodings prior to being output by diagnostics machinery. Also adds the ability to strip a UTF-8 BOM similarly. gcc/c-family/ChangeLog: PR other/93067 * c-opts.c (c_common_input_charset_cb): New function. (c_common_post_options): Call new function diagnostic_initialize_input_context(). gcc/d/ChangeLog: PR other/93067 * d-lang.cc (d_input_charset_callback): New function. (d_init): Call new function diagnostic_initialize_input_context(). gcc/fortran/ChangeLog: PR other/93067 * cpp.c (gfc_cpp_post_options): Call new function diagnostic_initialize_input_context(). gcc/ChangeLog: PR other/93067 * coretypes.h (typedef diagnostic_input_charset_callback): Declare. * diagnostic.c (diagnostic_initialize_input_context): New function. * diagnostic.h (diagnostic_initialize_input_context): Declare. * input.c (default_charset_callback): New function. (file_cache::initialize_input_context): New function. (file_cache_slot::create): Added ability to convert the input according to the input context. (file_cache::file_cache): Initialize the new input context. (class file_cache_slot): Added new m_alloc_offset member. (file_cache_slot::file_cache_slot): Initialize the new member. (file_cache_slot::~file_cache_slot): Handle potentially offset buffer. (file_cache_slot::maybe_grow): Likewise. (file_cache_slot::needs_read_p): Handle NULL fp, which is now possible. (file_cache_slot::get_next_line): Likewise. * input.h (class file_cache): Added input context member. libcpp/ChangeLog: PR other/93067 * charset.c (init_iconv_desc): Adapt to permit PFILE argument to be NULL. (_cpp_convert_input): Likewise. Also move UTF-8 BOM logic to... (cpp_check_utf8_bom): ...here. New function. (cpp_input_conversion_is_trivial): New function. * files.c (read_file_guts): Allow PFILE argument to be NULL. Add INPUT_CHARSET argument as an alternate source of this information. (read_file): Pass the new argument to read_file_guts. (cpp_get_converted_source): New function. * include/cpplib.h (struct cpp_converted_source): Declare. (cpp_get_converted_source): Declare. (cpp_input_conversion_is_trivial): Declare. (cpp_check_utf8_bom): Declare. gcc/testsuite/ChangeLog: PR other/93067 * gcc.dg/diagnostic-input-charset-1.c: New test. * gcc.dg/diagnostic-input-utf8-bom.c: New test.
2021-08-24 19:30:44 -04:00
if (pfile)
cpp_error_at (pfile, CPP_DL_ERROR, loc,
"%s is a block device", file->path);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
return false;
}
regular = S_ISREG (file->st.st_mode) != 0;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (regular)
cpplex.c (cpp_output_tokens, [...]): New public interfaces. * cpplex.c (cpp_output_tokens, cpp_scan_buffer_nooutput): New public interfaces. (safe_fwrite, output_line_command): New static functions. (cpp_expand_to_buffer): Now private to cpplib. (cpp_scan_buffer): Take a printer. * cpphash.h: Update prototypes. * cpplib.h: Update prototypes. (cpp_printer): New. (cpp_buffer): Remove last_nominal_fname. (cpp_reader): Remove lineno. * cppmain.c: Use a cpp_printer. * fix-header.c: No need to inhibit line commands. Call cpp_start_read with no printer. * cpperror.c (cpp_notice_from_errno): Provide default name. * cppfiles.c (make_IHASH, _cpp_fake_ihash): New functions. (find_include_file, cpp_read_file): Use make_IHASH. (file_cleanup): Set control_macro and clear input_stack_listing_current here. (_cpp_execute_include): Don't output entering-file marker. * cpphash.c (special_symbol): Look for the line number in the buffer, not the reader. (_cpp_macroexpand): No need to disable line commands. (_cpp_dump_definition): No need to generate line commands. (dump_hash_helper): Remove excess newline from output. * cppinit.c (dump_special_to_buffer): No need to generate line commands. (cpp_printer_init): New. (cpp_start_read): Take a printer, and start it up if it's not NULL. No need to generate line commands. (cpp_finish): Expect no buffers stacked at all. Take a printer argument, and flush the output buffer if it's not NULL. * cpplex.c (_cpp_lex_token): Return EOF if there's no buffer. Don't put two hashes at the beginning of an assertion. (cpp_get_token): Don't increment pfile->lineno or emit line commands here. Return EOF if there's no buffer when we get EOF. * cpplib.c (do_define, skip_if_group): No need to disable line commands. (_cpp_output_line_command): Delete function. (do_line): Don't emit line commands here, but set things up so they will be emitted if necessary. Use _cpp_fake_ihash to make unique nominal_fnames if necessary. (do_elif, do_else, _cpp_handle_eof): Call cpp_error_with_line with 0 for column, not -1. (_cpp_handle_eof): Don't set the control macro here. Don't clear input_stack_listing_current here. Don't emit line commands. From-SVN: r33159
2000-04-14 23:29:45 +00:00
{
/* off_t might have a wider range than ssize_t - in other words,
the max size of a file might be bigger than the address
space. We can't handle a file that large. (Anyone with
a single source file bigger than 2GB needs to rethink
their coding style.) Some systems (e.g. AIX 4.1) define
SSIZE_MAX to be much smaller than the actual range of the
type. Use INTTYPE_MAXIMUM unconditionally to ensure this
does not bite us. */
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (file->st.st_size > INTTYPE_MAXIMUM (ssize_t))
cpplex.c (cpp_output_tokens, [...]): New public interfaces. * cpplex.c (cpp_output_tokens, cpp_scan_buffer_nooutput): New public interfaces. (safe_fwrite, output_line_command): New static functions. (cpp_expand_to_buffer): Now private to cpplib. (cpp_scan_buffer): Take a printer. * cpphash.h: Update prototypes. * cpplib.h: Update prototypes. (cpp_printer): New. (cpp_buffer): Remove last_nominal_fname. (cpp_reader): Remove lineno. * cppmain.c: Use a cpp_printer. * fix-header.c: No need to inhibit line commands. Call cpp_start_read with no printer. * cpperror.c (cpp_notice_from_errno): Provide default name. * cppfiles.c (make_IHASH, _cpp_fake_ihash): New functions. (find_include_file, cpp_read_file): Use make_IHASH. (file_cleanup): Set control_macro and clear input_stack_listing_current here. (_cpp_execute_include): Don't output entering-file marker. * cpphash.c (special_symbol): Look for the line number in the buffer, not the reader. (_cpp_macroexpand): No need to disable line commands. (_cpp_dump_definition): No need to generate line commands. (dump_hash_helper): Remove excess newline from output. * cppinit.c (dump_special_to_buffer): No need to generate line commands. (cpp_printer_init): New. (cpp_start_read): Take a printer, and start it up if it's not NULL. No need to generate line commands. (cpp_finish): Expect no buffers stacked at all. Take a printer argument, and flush the output buffer if it's not NULL. * cpplex.c (_cpp_lex_token): Return EOF if there's no buffer. Don't put two hashes at the beginning of an assertion. (cpp_get_token): Don't increment pfile->lineno or emit line commands here. Return EOF if there's no buffer when we get EOF. * cpplib.c (do_define, skip_if_group): No need to disable line commands. (_cpp_output_line_command): Delete function. (do_line): Don't emit line commands here, but set things up so they will be emitted if necessary. Use _cpp_fake_ihash to make unique nominal_fnames if necessary. (do_elif, do_else, _cpp_handle_eof): Call cpp_error_with_line with 0 for column, not -1. (_cpp_handle_eof): Don't set the control macro here. Don't clear input_stack_listing_current here. Don't emit line commands. From-SVN: r33159
2000-04-14 23:29:45 +00:00
{
diagnostics: Support for -finput-charset [PR93067] Adds the logic to handle -finput-charset in layout_get_source_line(), so that source lines are converted from their input encodings prior to being output by diagnostics machinery. Also adds the ability to strip a UTF-8 BOM similarly. gcc/c-family/ChangeLog: PR other/93067 * c-opts.c (c_common_input_charset_cb): New function. (c_common_post_options): Call new function diagnostic_initialize_input_context(). gcc/d/ChangeLog: PR other/93067 * d-lang.cc (d_input_charset_callback): New function. (d_init): Call new function diagnostic_initialize_input_context(). gcc/fortran/ChangeLog: PR other/93067 * cpp.c (gfc_cpp_post_options): Call new function diagnostic_initialize_input_context(). gcc/ChangeLog: PR other/93067 * coretypes.h (typedef diagnostic_input_charset_callback): Declare. * diagnostic.c (diagnostic_initialize_input_context): New function. * diagnostic.h (diagnostic_initialize_input_context): Declare. * input.c (default_charset_callback): New function. (file_cache::initialize_input_context): New function. (file_cache_slot::create): Added ability to convert the input according to the input context. (file_cache::file_cache): Initialize the new input context. (class file_cache_slot): Added new m_alloc_offset member. (file_cache_slot::file_cache_slot): Initialize the new member. (file_cache_slot::~file_cache_slot): Handle potentially offset buffer. (file_cache_slot::maybe_grow): Likewise. (file_cache_slot::needs_read_p): Handle NULL fp, which is now possible. (file_cache_slot::get_next_line): Likewise. * input.h (class file_cache): Added input context member. libcpp/ChangeLog: PR other/93067 * charset.c (init_iconv_desc): Adapt to permit PFILE argument to be NULL. (_cpp_convert_input): Likewise. Also move UTF-8 BOM logic to... (cpp_check_utf8_bom): ...here. New function. (cpp_input_conversion_is_trivial): New function. * files.c (read_file_guts): Allow PFILE argument to be NULL. Add INPUT_CHARSET argument as an alternate source of this information. (read_file): Pass the new argument to read_file_guts. (cpp_get_converted_source): New function. * include/cpplib.h (struct cpp_converted_source): Declare. (cpp_get_converted_source): Declare. (cpp_input_conversion_is_trivial): Declare. (cpp_check_utf8_bom): Declare. gcc/testsuite/ChangeLog: PR other/93067 * gcc.dg/diagnostic-input-charset-1.c: New test. * gcc.dg/diagnostic-input-utf8-bom.c: New test.
2021-08-24 19:30:44 -04:00
if (pfile)
cpp_error_at (pfile, CPP_DL_ERROR, loc,
"%s is too large", file->path);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
return false;
cpplex.c (cpp_output_tokens, [...]): New public interfaces. * cpplex.c (cpp_output_tokens, cpp_scan_buffer_nooutput): New public interfaces. (safe_fwrite, output_line_command): New static functions. (cpp_expand_to_buffer): Now private to cpplib. (cpp_scan_buffer): Take a printer. * cpphash.h: Update prototypes. * cpplib.h: Update prototypes. (cpp_printer): New. (cpp_buffer): Remove last_nominal_fname. (cpp_reader): Remove lineno. * cppmain.c: Use a cpp_printer. * fix-header.c: No need to inhibit line commands. Call cpp_start_read with no printer. * cpperror.c (cpp_notice_from_errno): Provide default name. * cppfiles.c (make_IHASH, _cpp_fake_ihash): New functions. (find_include_file, cpp_read_file): Use make_IHASH. (file_cleanup): Set control_macro and clear input_stack_listing_current here. (_cpp_execute_include): Don't output entering-file marker. * cpphash.c (special_symbol): Look for the line number in the buffer, not the reader. (_cpp_macroexpand): No need to disable line commands. (_cpp_dump_definition): No need to generate line commands. (dump_hash_helper): Remove excess newline from output. * cppinit.c (dump_special_to_buffer): No need to generate line commands. (cpp_printer_init): New. (cpp_start_read): Take a printer, and start it up if it's not NULL. No need to generate line commands. (cpp_finish): Expect no buffers stacked at all. Take a printer argument, and flush the output buffer if it's not NULL. * cpplex.c (_cpp_lex_token): Return EOF if there's no buffer. Don't put two hashes at the beginning of an assertion. (cpp_get_token): Don't increment pfile->lineno or emit line commands here. Return EOF if there's no buffer when we get EOF. * cpplib.c (do_define, skip_if_group): No need to disable line commands. (_cpp_output_line_command): Delete function. (do_line): Don't emit line commands here, but set things up so they will be emitted if necessary. Use _cpp_fake_ihash to make unique nominal_fnames if necessary. (do_elif, do_else, _cpp_handle_eof): Call cpp_error_with_line with 0 for column, not -1. (_cpp_handle_eof): Don't set the control macro here. Don't clear input_stack_listing_current here. Don't emit line commands. From-SVN: r33159
2000-04-14 23:29:45 +00:00
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
size = file->st.st_size;
}
else
/* 8 kilobytes is a sensible starting size. It ought to be bigger
than the kernel pipe buffer, and it's definitely bigger than
the majority of C source files. */
size = 8 * 1024;
pad = CPP_BUFFER_PADDING;
/* The '+ PAD' here is space for the final '\n' and PAD-1 bytes of padding,
allowing search_line_fast to use (possibly misaligned) vector loads. */
buf = XNEWVEC (uchar, size + pad);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
total = 0;
while ((count = read (file->fd, buf + total, size - total)) > 0)
{
total += count;
if (total == size)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (regular)
break;
size *= 2;
buf = XRESIZEVEC (uchar, buf, size + pad);
}
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (count < 0)
{
diagnostics: Support for -finput-charset [PR93067] Adds the logic to handle -finput-charset in layout_get_source_line(), so that source lines are converted from their input encodings prior to being output by diagnostics machinery. Also adds the ability to strip a UTF-8 BOM similarly. gcc/c-family/ChangeLog: PR other/93067 * c-opts.c (c_common_input_charset_cb): New function. (c_common_post_options): Call new function diagnostic_initialize_input_context(). gcc/d/ChangeLog: PR other/93067 * d-lang.cc (d_input_charset_callback): New function. (d_init): Call new function diagnostic_initialize_input_context(). gcc/fortran/ChangeLog: PR other/93067 * cpp.c (gfc_cpp_post_options): Call new function diagnostic_initialize_input_context(). gcc/ChangeLog: PR other/93067 * coretypes.h (typedef diagnostic_input_charset_callback): Declare. * diagnostic.c (diagnostic_initialize_input_context): New function. * diagnostic.h (diagnostic_initialize_input_context): Declare. * input.c (default_charset_callback): New function. (file_cache::initialize_input_context): New function. (file_cache_slot::create): Added ability to convert the input according to the input context. (file_cache::file_cache): Initialize the new input context. (class file_cache_slot): Added new m_alloc_offset member. (file_cache_slot::file_cache_slot): Initialize the new member. (file_cache_slot::~file_cache_slot): Handle potentially offset buffer. (file_cache_slot::maybe_grow): Likewise. (file_cache_slot::needs_read_p): Handle NULL fp, which is now possible. (file_cache_slot::get_next_line): Likewise. * input.h (class file_cache): Added input context member. libcpp/ChangeLog: PR other/93067 * charset.c (init_iconv_desc): Adapt to permit PFILE argument to be NULL. (_cpp_convert_input): Likewise. Also move UTF-8 BOM logic to... (cpp_check_utf8_bom): ...here. New function. (cpp_input_conversion_is_trivial): New function. * files.c (read_file_guts): Allow PFILE argument to be NULL. Add INPUT_CHARSET argument as an alternate source of this information. (read_file): Pass the new argument to read_file_guts. (cpp_get_converted_source): New function. * include/cpplib.h (struct cpp_converted_source): Declare. (cpp_get_converted_source): Declare. (cpp_input_conversion_is_trivial): Declare. (cpp_check_utf8_bom): Declare. gcc/testsuite/ChangeLog: PR other/93067 * gcc.dg/diagnostic-input-charset-1.c: New test. * gcc.dg/diagnostic-input-utf8-bom.c: New test.
2021-08-24 19:30:44 -04:00
if (pfile)
cpp_errno_filename (pfile, CPP_DL_ERROR, file->path, loc);
free (buf);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
return false;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
diagnostics: Support for -finput-charset [PR93067] Adds the logic to handle -finput-charset in layout_get_source_line(), so that source lines are converted from their input encodings prior to being output by diagnostics machinery. Also adds the ability to strip a UTF-8 BOM similarly. gcc/c-family/ChangeLog: PR other/93067 * c-opts.c (c_common_input_charset_cb): New function. (c_common_post_options): Call new function diagnostic_initialize_input_context(). gcc/d/ChangeLog: PR other/93067 * d-lang.cc (d_input_charset_callback): New function. (d_init): Call new function diagnostic_initialize_input_context(). gcc/fortran/ChangeLog: PR other/93067 * cpp.c (gfc_cpp_post_options): Call new function diagnostic_initialize_input_context(). gcc/ChangeLog: PR other/93067 * coretypes.h (typedef diagnostic_input_charset_callback): Declare. * diagnostic.c (diagnostic_initialize_input_context): New function. * diagnostic.h (diagnostic_initialize_input_context): Declare. * input.c (default_charset_callback): New function. (file_cache::initialize_input_context): New function. (file_cache_slot::create): Added ability to convert the input according to the input context. (file_cache::file_cache): Initialize the new input context. (class file_cache_slot): Added new m_alloc_offset member. (file_cache_slot::file_cache_slot): Initialize the new member. (file_cache_slot::~file_cache_slot): Handle potentially offset buffer. (file_cache_slot::maybe_grow): Likewise. (file_cache_slot::needs_read_p): Handle NULL fp, which is now possible. (file_cache_slot::get_next_line): Likewise. * input.h (class file_cache): Added input context member. libcpp/ChangeLog: PR other/93067 * charset.c (init_iconv_desc): Adapt to permit PFILE argument to be NULL. (_cpp_convert_input): Likewise. Also move UTF-8 BOM logic to... (cpp_check_utf8_bom): ...here. New function. (cpp_input_conversion_is_trivial): New function. * files.c (read_file_guts): Allow PFILE argument to be NULL. Add INPUT_CHARSET argument as an alternate source of this information. (read_file): Pass the new argument to read_file_guts. (cpp_get_converted_source): New function. * include/cpplib.h (struct cpp_converted_source): Declare. (cpp_get_converted_source): Declare. (cpp_input_conversion_is_trivial): Declare. (cpp_check_utf8_bom): Declare. gcc/testsuite/ChangeLog: PR other/93067 * gcc.dg/diagnostic-input-charset-1.c: New test. * gcc.dg/diagnostic-input-utf8-bom.c: New test.
2021-08-24 19:30:44 -04:00
if (pfile && regular && total != size && STAT_SIZE_RELIABLE (file->st))
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
cpp_error_at (pfile, CPP_DL_WARNING, loc,
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
"%s is shorter than expected", file->path);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
file->buffer = _cpp_convert_input (pfile,
diagnostics: Support for -finput-charset [PR93067] Adds the logic to handle -finput-charset in layout_get_source_line(), so that source lines are converted from their input encodings prior to being output by diagnostics machinery. Also adds the ability to strip a UTF-8 BOM similarly. gcc/c-family/ChangeLog: PR other/93067 * c-opts.c (c_common_input_charset_cb): New function. (c_common_post_options): Call new function diagnostic_initialize_input_context(). gcc/d/ChangeLog: PR other/93067 * d-lang.cc (d_input_charset_callback): New function. (d_init): Call new function diagnostic_initialize_input_context(). gcc/fortran/ChangeLog: PR other/93067 * cpp.c (gfc_cpp_post_options): Call new function diagnostic_initialize_input_context(). gcc/ChangeLog: PR other/93067 * coretypes.h (typedef diagnostic_input_charset_callback): Declare. * diagnostic.c (diagnostic_initialize_input_context): New function. * diagnostic.h (diagnostic_initialize_input_context): Declare. * input.c (default_charset_callback): New function. (file_cache::initialize_input_context): New function. (file_cache_slot::create): Added ability to convert the input according to the input context. (file_cache::file_cache): Initialize the new input context. (class file_cache_slot): Added new m_alloc_offset member. (file_cache_slot::file_cache_slot): Initialize the new member. (file_cache_slot::~file_cache_slot): Handle potentially offset buffer. (file_cache_slot::maybe_grow): Likewise. (file_cache_slot::needs_read_p): Handle NULL fp, which is now possible. (file_cache_slot::get_next_line): Likewise. * input.h (class file_cache): Added input context member. libcpp/ChangeLog: PR other/93067 * charset.c (init_iconv_desc): Adapt to permit PFILE argument to be NULL. (_cpp_convert_input): Likewise. Also move UTF-8 BOM logic to... (cpp_check_utf8_bom): ...here. New function. (cpp_input_conversion_is_trivial): New function. * files.c (read_file_guts): Allow PFILE argument to be NULL. Add INPUT_CHARSET argument as an alternate source of this information. (read_file): Pass the new argument to read_file_guts. (cpp_get_converted_source): New function. * include/cpplib.h (struct cpp_converted_source): Declare. (cpp_get_converted_source): Declare. (cpp_input_conversion_is_trivial): Declare. (cpp_check_utf8_bom): Declare. gcc/testsuite/ChangeLog: PR other/93067 * gcc.dg/diagnostic-input-charset-1.c: New test. * gcc.dg/diagnostic-input-utf8-bom.c: New test.
2021-08-24 19:30:44 -04:00
input_charset,
buf, size + pad, total,
&file->buffer_start,
&file->st.st_size);
diagnostics: Support for -finput-charset [PR93067] Adds the logic to handle -finput-charset in layout_get_source_line(), so that source lines are converted from their input encodings prior to being output by diagnostics machinery. Also adds the ability to strip a UTF-8 BOM similarly. gcc/c-family/ChangeLog: PR other/93067 * c-opts.c (c_common_input_charset_cb): New function. (c_common_post_options): Call new function diagnostic_initialize_input_context(). gcc/d/ChangeLog: PR other/93067 * d-lang.cc (d_input_charset_callback): New function. (d_init): Call new function diagnostic_initialize_input_context(). gcc/fortran/ChangeLog: PR other/93067 * cpp.c (gfc_cpp_post_options): Call new function diagnostic_initialize_input_context(). gcc/ChangeLog: PR other/93067 * coretypes.h (typedef diagnostic_input_charset_callback): Declare. * diagnostic.c (diagnostic_initialize_input_context): New function. * diagnostic.h (diagnostic_initialize_input_context): Declare. * input.c (default_charset_callback): New function. (file_cache::initialize_input_context): New function. (file_cache_slot::create): Added ability to convert the input according to the input context. (file_cache::file_cache): Initialize the new input context. (class file_cache_slot): Added new m_alloc_offset member. (file_cache_slot::file_cache_slot): Initialize the new member. (file_cache_slot::~file_cache_slot): Handle potentially offset buffer. (file_cache_slot::maybe_grow): Likewise. (file_cache_slot::needs_read_p): Handle NULL fp, which is now possible. (file_cache_slot::get_next_line): Likewise. * input.h (class file_cache): Added input context member. libcpp/ChangeLog: PR other/93067 * charset.c (init_iconv_desc): Adapt to permit PFILE argument to be NULL. (_cpp_convert_input): Likewise. Also move UTF-8 BOM logic to... (cpp_check_utf8_bom): ...here. New function. (cpp_input_conversion_is_trivial): New function. * files.c (read_file_guts): Allow PFILE argument to be NULL. Add INPUT_CHARSET argument as an alternate source of this information. (read_file): Pass the new argument to read_file_guts. (cpp_get_converted_source): New function. * include/cpplib.h (struct cpp_converted_source): Declare. (cpp_get_converted_source): Declare. (cpp_input_conversion_is_trivial): Declare. (cpp_check_utf8_bom): Declare. gcc/testsuite/ChangeLog: PR other/93067 * gcc.dg/diagnostic-input-charset-1.c: New test. * gcc.dg/diagnostic-input-utf8-bom.c: New test.
2021-08-24 19:30:44 -04:00
file->buffer_valid = file->buffer;
return file->buffer_valid;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
}
/* Convenience wrapper around read_file_guts that opens the file if
necessary and closes the file descriptor after reading. FILE must
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
have been passed through find_file() at some stage. Use LOC for
diagnostics: Support for -finput-charset [PR93067] Adds the logic to handle -finput-charset in layout_get_source_line(), so that source lines are converted from their input encodings prior to being output by diagnostics machinery. Also adds the ability to strip a UTF-8 BOM similarly. gcc/c-family/ChangeLog: PR other/93067 * c-opts.c (c_common_input_charset_cb): New function. (c_common_post_options): Call new function diagnostic_initialize_input_context(). gcc/d/ChangeLog: PR other/93067 * d-lang.cc (d_input_charset_callback): New function. (d_init): Call new function diagnostic_initialize_input_context(). gcc/fortran/ChangeLog: PR other/93067 * cpp.c (gfc_cpp_post_options): Call new function diagnostic_initialize_input_context(). gcc/ChangeLog: PR other/93067 * coretypes.h (typedef diagnostic_input_charset_callback): Declare. * diagnostic.c (diagnostic_initialize_input_context): New function. * diagnostic.h (diagnostic_initialize_input_context): Declare. * input.c (default_charset_callback): New function. (file_cache::initialize_input_context): New function. (file_cache_slot::create): Added ability to convert the input according to the input context. (file_cache::file_cache): Initialize the new input context. (class file_cache_slot): Added new m_alloc_offset member. (file_cache_slot::file_cache_slot): Initialize the new member. (file_cache_slot::~file_cache_slot): Handle potentially offset buffer. (file_cache_slot::maybe_grow): Likewise. (file_cache_slot::needs_read_p): Handle NULL fp, which is now possible. (file_cache_slot::get_next_line): Likewise. * input.h (class file_cache): Added input context member. libcpp/ChangeLog: PR other/93067 * charset.c (init_iconv_desc): Adapt to permit PFILE argument to be NULL. (_cpp_convert_input): Likewise. Also move UTF-8 BOM logic to... (cpp_check_utf8_bom): ...here. New function. (cpp_input_conversion_is_trivial): New function. * files.c (read_file_guts): Allow PFILE argument to be NULL. Add INPUT_CHARSET argument as an alternate source of this information. (read_file): Pass the new argument to read_file_guts. (cpp_get_converted_source): New function. * include/cpplib.h (struct cpp_converted_source): Declare. (cpp_get_converted_source): Declare. (cpp_input_conversion_is_trivial): Declare. (cpp_check_utf8_bom): Declare. gcc/testsuite/ChangeLog: PR other/93067 * gcc.dg/diagnostic-input-charset-1.c: New test. * gcc.dg/diagnostic-input-utf8-bom.c: New test.
2021-08-24 19:30:44 -04:00
any diagnostics. Unlike read_file_guts(), PFILE may not be NULL. */
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
static bool
Eliminate source_location in favor of location_t Historically GCC used location_t, while libcpp used source_location. This inconsistency has been annoying me for a while, so this patch removes source_location in favor of location_t throughout (as the latter is shorter). gcc/ChangeLog: * builtins.c: Replace "source_location" with "location_t". * diagnostic-show-locus.c: Likewise. * diagnostic.c: Likewise. * dumpfile.c: Likewise. * gcc-rich-location.h: Likewise. * genmatch.c: Likewise. * gimple.h: Likewise. * gimplify.c: Likewise. * input.c: Likewise. * input.h: Likewise. Eliminate the typedef. * omp-expand.c: Likewise. * selftest.h: Likewise. * substring-locations.h (get_source_location_for_substring): Rename to.. (get_location_within_string): ...this. * tree-cfg.c: Replace "source_location" with "location_t". * tree-cfgcleanup.c: Likewise. * tree-diagnostic.c: Likewise. * tree-into-ssa.c: Likewise. * tree-outof-ssa.c: Likewise. * tree-parloops.c: Likewise. * tree-phinodes.c: Likewise. * tree-phinodes.h: Likewise. * tree-ssa-loop-ivopts.c: Likewise. * tree-ssa-loop-manip.c: Likewise. * tree-ssa-phiopt.c: Likewise. * tree-ssa-phiprop.c: Likewise. * tree-ssa-threadupdate.c: Likewise. * tree-ssa.c: Likewise. * tree-ssa.h: Likewise. * tree-vect-loop-manip.c: Likewise. gcc/c-family/ChangeLog: * c-common.c (c_get_substring_location): Update for renaming of get_source_location_for_substring to get_location_within_string. * c-lex.c: Replace "source_location" with "location_t". * c-opts.c: Likewise. * c-ppoutput.c: Likewise. gcc/c/ChangeLog: * c-decl.c: Replace "source_location" with "location_t". * c-tree.h: Likewise. * c-typeck.c: Likewise. * gimple-parser.c: Likewise. gcc/cp/ChangeLog: * call.c: Replace "source_location" with "location_t". * cp-tree.h: Likewise. * cvt.c: Likewise. * name-lookup.c: Likewise. * parser.c: Likewise. * typeck.c: Likewise. gcc/fortran/ChangeLog: * cpp.c: Replace "source_location" with "location_t". * gfortran.h: Likewise. gcc/go/ChangeLog: * go-gcc-diagnostics.cc: Replace "source_location" with "location_t". * go-gcc.cc: Likewise. * go-linemap.cc: Likewise. * go-location.h: Likewise. * gofrontend/README: Likewise. gcc/jit/ChangeLog: * jit-playback.c: Replace "source_location" with "location_t". gcc/testsuite/ChangeLog: * g++.dg/plugin/comment_plugin.c: Replace "source_location" with "location_t". * gcc.dg/plugin/diagnostic_plugin_test_show_locus.c: Likewise. libcc1/ChangeLog: * libcc1plugin.cc: Replace "source_location" with "location_t". (plugin_context::get_source_location): Rename to... (plugin_context::get_location_t): ...this. * libcp1plugin.cc: Likewise. libcpp/ChangeLog: * charset.c: Replace "source_location" with "location_t". * directives-only.c: Likewise. * directives.c: Likewise. * errors.c: Likewise. * expr.c: Likewise. * files.c: Likewise. * include/cpplib.h: Likewise. Rename MAX_SOURCE_LOCATION to MAX_LOCATION_T. * include/line-map.h: Likewise. * init.c: Likewise. * internal.h: Likewise. * lex.c: Likewise. * line-map.c: Likewise. * location-example.txt: Likewise. * macro.c: Likewise. * pch.c: Likewise. * traditional.c: Likewise. From-SVN: r266085
2018-11-13 20:05:03 +00:00
read_file (cpp_reader *pfile, _cpp_file *file, location_t loc)
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
{
/* If we already have its contents in memory, succeed immediately. */
if (file->buffer_valid)
return true;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* If an earlier read failed for some reason don't try again. */
if (file->dont_read || file->err_no)
return false;
cppfiles.c (stack_include_file): Don't optimize zero-length files. * cppfiles.c (stack_include_file): Don't optimize zero-length files. (read_include_file): NUL-terminate read files. * cpplex.c (handle_newline, skip_escaped_newlines, get_effective_char, skip_whitespace, parse_identifier, parse_identifier_slow, parse_number, parse_string, _cpp_lex_direct): Optimize for the fact that buffers are guaranteed NUL-terminated. * cpplib.c (destringize_and_run, cpp_define, handle_assertion): Be sure buffers are NUL terminated. * cppmacro.c (warn_of_redefinition): Kill compile warning. * c-common.c: Include tree-inline.h. (c_language): Move separate definitions here. (c_common_init_options, c_common_post_options): New. (c_common_lang_init): Rename c_common_init. * c-common.h (c_common_lang_init): Similarly. (c_common_init_options, c_common_post_options): New. * c-lang.c (c_post_options): Move body to c_common_post_options. (c_init_options): Use c_common_init_options. (c_init): Update. * langhooks.def: Rearrange. * langhooks.h: Rearrange, and improve comments. * toplev.c (do_compile): New function. (toplev_main): Use it. (lang_independent_f_options, parse_options_and_default_flags, process_options): Remove trailing periods. * Makefile.in: Update. cp: * decl2.c (c_language): Move to c-common.c. * lex.c (cxx_post_options, cxx_init_options): Use c-common.c functions. (cxx_init): Update. objc: * objc-act.c (objc_post_options, objc_init_options): Use c-common.c functions. (ojbc_init): Update. From-SVN: r47362
2001-11-26 23:44:54 +00:00
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (file->fd == -1 && !open_file (file))
{
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
open_file_failed (pfile, file, 0, loc);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
return false;
cpplex.c (cpp_output_tokens, [...]): New public interfaces. * cpplex.c (cpp_output_tokens, cpp_scan_buffer_nooutput): New public interfaces. (safe_fwrite, output_line_command): New static functions. (cpp_expand_to_buffer): Now private to cpplib. (cpp_scan_buffer): Take a printer. * cpphash.h: Update prototypes. * cpplib.h: Update prototypes. (cpp_printer): New. (cpp_buffer): Remove last_nominal_fname. (cpp_reader): Remove lineno. * cppmain.c: Use a cpp_printer. * fix-header.c: No need to inhibit line commands. Call cpp_start_read with no printer. * cpperror.c (cpp_notice_from_errno): Provide default name. * cppfiles.c (make_IHASH, _cpp_fake_ihash): New functions. (find_include_file, cpp_read_file): Use make_IHASH. (file_cleanup): Set control_macro and clear input_stack_listing_current here. (_cpp_execute_include): Don't output entering-file marker. * cpphash.c (special_symbol): Look for the line number in the buffer, not the reader. (_cpp_macroexpand): No need to disable line commands. (_cpp_dump_definition): No need to generate line commands. (dump_hash_helper): Remove excess newline from output. * cppinit.c (dump_special_to_buffer): No need to generate line commands. (cpp_printer_init): New. (cpp_start_read): Take a printer, and start it up if it's not NULL. No need to generate line commands. (cpp_finish): Expect no buffers stacked at all. Take a printer argument, and flush the output buffer if it's not NULL. * cpplex.c (_cpp_lex_token): Return EOF if there's no buffer. Don't put two hashes at the beginning of an assertion. (cpp_get_token): Don't increment pfile->lineno or emit line commands here. Return EOF if there's no buffer when we get EOF. * cpplib.c (do_define, skip_if_group): No need to disable line commands. (_cpp_output_line_command): Delete function. (do_line): Don't emit line commands here, but set things up so they will be emitted if necessary. Use _cpp_fake_ihash to make unique nominal_fnames if necessary. (do_elif, do_else, _cpp_handle_eof): Call cpp_error_with_line with 0 for column, not -1. (_cpp_handle_eof): Don't set the control macro here. Don't clear input_stack_listing_current here. Don't emit line commands. From-SVN: r33159
2000-04-14 23:29:45 +00:00
}
diagnostics: Support for -finput-charset [PR93067] Adds the logic to handle -finput-charset in layout_get_source_line(), so that source lines are converted from their input encodings prior to being output by diagnostics machinery. Also adds the ability to strip a UTF-8 BOM similarly. gcc/c-family/ChangeLog: PR other/93067 * c-opts.c (c_common_input_charset_cb): New function. (c_common_post_options): Call new function diagnostic_initialize_input_context(). gcc/d/ChangeLog: PR other/93067 * d-lang.cc (d_input_charset_callback): New function. (d_init): Call new function diagnostic_initialize_input_context(). gcc/fortran/ChangeLog: PR other/93067 * cpp.c (gfc_cpp_post_options): Call new function diagnostic_initialize_input_context(). gcc/ChangeLog: PR other/93067 * coretypes.h (typedef diagnostic_input_charset_callback): Declare. * diagnostic.c (diagnostic_initialize_input_context): New function. * diagnostic.h (diagnostic_initialize_input_context): Declare. * input.c (default_charset_callback): New function. (file_cache::initialize_input_context): New function. (file_cache_slot::create): Added ability to convert the input according to the input context. (file_cache::file_cache): Initialize the new input context. (class file_cache_slot): Added new m_alloc_offset member. (file_cache_slot::file_cache_slot): Initialize the new member. (file_cache_slot::~file_cache_slot): Handle potentially offset buffer. (file_cache_slot::maybe_grow): Likewise. (file_cache_slot::needs_read_p): Handle NULL fp, which is now possible. (file_cache_slot::get_next_line): Likewise. * input.h (class file_cache): Added input context member. libcpp/ChangeLog: PR other/93067 * charset.c (init_iconv_desc): Adapt to permit PFILE argument to be NULL. (_cpp_convert_input): Likewise. Also move UTF-8 BOM logic to... (cpp_check_utf8_bom): ...here. New function. (cpp_input_conversion_is_trivial): New function. * files.c (read_file_guts): Allow PFILE argument to be NULL. Add INPUT_CHARSET argument as an alternate source of this information. (read_file): Pass the new argument to read_file_guts. (cpp_get_converted_source): New function. * include/cpplib.h (struct cpp_converted_source): Declare. (cpp_get_converted_source): Declare. (cpp_input_conversion_is_trivial): Declare. (cpp_check_utf8_bom): Declare. gcc/testsuite/ChangeLog: PR other/93067 * gcc.dg/diagnostic-input-charset-1.c: New test. * gcc.dg/diagnostic-input-utf8-bom.c: New test.
2021-08-24 19:30:44 -04:00
file->dont_read = !read_file_guts (pfile, file, loc,
CPP_OPTION (pfile, input_charset));
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
close (file->fd);
file->fd = -1;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
return !file->dont_read;
}
/* Returns TRUE if FILE is already known to be idempotent, and should
therefore not be read again. */
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
static bool
is_known_idempotent_file (cpp_reader *pfile, _cpp_file *file, bool import)
{
/* Skip once-only files. */
if (file->once_only)
return true;
[multiple changes] 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * c-opts.c (c_common_handle_option): Add -finput-charset. * c.opt: Ditto. * cppcharset.c (one_iso88591_to_utf8): Remove. (convert_iso88591_utf8): Ditto. (conversion_tab): Remove 8859-1 converter. (_cpp_input_to_utf8): Remove. (_cpp_init_iconv_buffer): Ditto. (_cpp_close_iconv_buffer): Ditto. (_cpp_convert_input): New function. (_cpp_default_encoding): Ditto. * cpphash.h: Add/remove prototypes for above. * cppfiles.c (read_file_guts): Use _cpp_convert_input. * cppinit.c (cpp_create_reader): Use _cpp_default_encoding for narrow execution and input character sets. * cpplib.c (cpp_push_buffer): Delete uses of removed functions. * doc/cppopts.texi: Document -finput-charset. 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * gcc.c-torture/execute/wchar_t-1.c: Add -finput-charset. 2004-01-29 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * testsuite/22_locale/collate/compare/wchar_t/2.cc: Remove xfail. Use -finput-charset. * testsuite/22_locale/collate/compare/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/compare/wchar_t/wrapped_locale.cc: Ditto * testsuite/22_locale/collate/hash/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_locale.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_locale.cc: Ditto. From-SVN: r77136
2004-02-02 20:20:58 +00:00
/* We must mark the file once-only if #import now, before header
guard checks. Otherwise, undefining the header guard might
cause the file to be re-stacked. */
if (import)
{
_cpp_mark_file_once_only (pfile, file);
/* Don't stack files that have been stacked before. */
if (file->stack_count)
return true;
}
/* Skip if the file had a header guard and the macro is defined.
PCH relies on this appearing before the PCH handler below. */
[PATCH] CPP Macro predicates https://gcc.gnu.org/ml/gcc-patches/2018-08/msg00897.html libcpp/ * include/cpplib.h (cpp_user_macro_p, cpp_builtin_macro_p) (cpp_macro_p): New inlines. * directives.c (do_pragma_poison): Use cpp_macro_p. (do_ifdef, do_ifndef): Likewise. Use _cpp_maybe_notify_macro_use. (cpp_pop_definition): Use cpp_macro_p. Move _cpp_free_definition earlier. Don't zap node directly. * expr.c (parse_defined): Use _cpp_maybe_notify_macro_use & cpp_macro_p. * files.c (should_stack_file): Use cpp_macro_p. * identifiers.c (cpp_defined): Likewise. * internal.h (_cpp_mark_macro): Use cpp_user_macro_p. (_cpp_notify_macro_use): Declare. (_cpp_maybe_notify_macro_use): New inline. * lex.c (is_macro): Use cpp_macro_p. * macro.c (_cpp_warn_if_unused_macro): Use cpp_user_macro_p. (enter_macro_context): Likewise. (_cpp_create_definition): Use cpp_builtin_macro_p, cpp_user_macro_p. Move _cpp_free_definition earlier. (_cpp_notify_macro_use): New, broken out of multiple call sites. * traditional.c (fun_like_macro_p): Use cpp_builtin_macro_p. (maybe_start_funlike, _cpp_scan_out_logical_line) (push_replacement_text): Likewise. gcc/c-family/ * c-ada-spec.c (count_ada_macro): Use cpp_user_macro_p. (store_ada_macro): Likewise. * c-ppoutput.c (cb_used_define, dump_macro): Likewise. * c-spellcheck.cc (should-suggest_as_macro_p): Likewise, gcc/ * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p. * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise. gcc/cp/ * name-lookup.c (lookup_name_fuzzy): Likewise. gcc/fortran/ * cpp.c (dump_macro): Use cpp_user_macro_p. From-SVN: r263587
2018-08-16 13:51:38 +00:00
if (file->cmacro && cpp_macro_p (file->cmacro))
return true;
/* Handle PCH files immediately; don't stack them. */
if (file->pchname)
{
pfile->cb.read_pch (pfile, file->pchname, file->fd, file->path);
file->fd = -1;
free ((void *) file->pchname);
file->pchname = NULL;
return true;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
return false;
}
/* Return TRUE if file has unique contents, so we should read process
it. The file's contents must already have been read. */
static bool
has_unique_contents (cpp_reader *pfile, _cpp_file *file, bool import,
location_t loc)
{
/* Check the file against the PCH file. This is done before
checking against files we've already seen, since it may save on
I/O. */
if (check_file_against_entries (pfile, file, import))
{
/* If this isn't a #import, but yet we can't include the file,
that means that it was #import-ed in the PCH file,
so we can never include it again. */
if (! import)
_cpp_mark_file_once_only (pfile, file);
return false;
}
/* Now we've read the file's contents, we can stack it if there
are no once-only files. */
if (!pfile->seen_once_only)
return true;
/* We may have read the file under a different name. Look
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
for likely candidates and compare file contents to be sure. */
for (_cpp_file *f = pfile->all_files; f; f = f->next_file)
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
{
if (f == file)
continue; /* It'sa me! */
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
if (f->embed)
continue;
if ((import || f->once_only)
&& f->err_no == 0
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
&& f->st.st_mtime == file->st.st_mtime
&& f->st.st_size == file->st.st_size)
{
_cpp_file *ref_file;
if (f->buffer && !f->buffer_valid)
{
/* We already have a buffer but it is not valid, because
the file is still stacked. Make a new one. */
ref_file = make_cpp_file (f->dir, f->name);
ref_file->path = f->path;
}
else
/* The file is not stacked anymore. We can reuse it. */
ref_file = f;
bool same_file_p = (read_file (pfile, ref_file, loc)
/* Size might have changed in read_file(). */
&& ref_file->st.st_size == file->st.st_size
&& !memcmp (ref_file->buffer, file->buffer,
file->st.st_size));
if (f->buffer && !f->buffer_valid)
{
ref_file->path = 0;
destroy_cpp_file (ref_file);
}
if (same_file_p)
/* Already seen under a different name. */
return false;
}
cpplib.h: Delete struct import_file. Wed Nov 25 14:54:46 1998 Zack Weinberg <zack@rabi.phys.columbia.edu> * cpplib.h: Delete struct import_file. Add ihash element to struct cpp_buffer. Delete dont_repeat_files and import_hash_table elements from cpp_reader; change all_include_files to a hash table. Delete all foobar_include / last_foobar_include elements from struct cpp_options; put back four such: quote_include, bracket_include, system_include, after_include. Redo struct file_name_list completely. Add new structure type include_hash. Add prototypes for merge_include_chains and include_hash. Change prototypes for finclude, find_include_file, and append_include_chain to match changes below. * cppfiles.c (simplify_pathname, include_hash, remap_filename, merge_include_chains): New functions. (add_import, lookup_import, open_include_file): Removed. (INO_T_EQ): Define this (copied from cccp.c). (hack_vms_include_specification): Remove all calls and #if 0 out the definition. It was being called incorrectly and at the wrong times. Until a VMSie can look at this, it's better to not pretend to support it. (append_include_chain): Change calling convention; now takes only one directory at a time, and sets up the data structure itself. (redundant_include_p): Rewritten - this is now used for all include redundancy, whether by #ifndef, #import, or #pragma once. Looks up things in the include hash table. (file_cleanup): Decrement pfile->system_include_depth here if it's >0. (find_include_file): Calling convention changed; now passes around a struct include_hash instead of 3 separate parameters. Guts ripped out and replaced with new include_hash mechanism. (finclude): Calling convention changed as for find_include_file. Error exits pulled out-of-line. Reformat. (safe_read): Return a long, not an int. (deps_output): Don't recurse. * cpplib.c (is_system_include): Deleted. (path_include): Fix up call to append_include_chain. (do_include): Fix up calls to find_include_file and finclude. Clean up dependency output a bit. Shorten obnoxiously lengthy #import warning message. Don't decrement pfile->system_include_depth here. (do_pragma): Understand the include_hash structure. Reformat. (do_endif): Correct handling of control macros. Understand the include_hash. (cpp_start_read): Fix up calls to finclude. Call merge_include_chains. (cpp_handle_option): Fix up calls to append_include_chain. Understand the four partial include chains. (cpp_finish): Add debugging code (#if 0-ed out) for the include_hash. (cpp_cleanup): Free the include_hash, not the import hash and the all_include and dont_repeat lists which no longer exist. From-SVN: r23857
1998-11-25 11:56:54 +00:00
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
return true;
}
/* Place the file referenced by FILE into a new buffer on the buffer
stack if possible. Returns true if a buffer is stacked. Use LOC
for any diagnostics. */
bool
_cpp_stack_file (cpp_reader *pfile, _cpp_file *file, include_type type,
Eliminate source_location in favor of location_t Historically GCC used location_t, while libcpp used source_location. This inconsistency has been annoying me for a while, so this patch removes source_location in favor of location_t throughout (as the latter is shorter). gcc/ChangeLog: * builtins.c: Replace "source_location" with "location_t". * diagnostic-show-locus.c: Likewise. * diagnostic.c: Likewise. * dumpfile.c: Likewise. * gcc-rich-location.h: Likewise. * genmatch.c: Likewise. * gimple.h: Likewise. * gimplify.c: Likewise. * input.c: Likewise. * input.h: Likewise. Eliminate the typedef. * omp-expand.c: Likewise. * selftest.h: Likewise. * substring-locations.h (get_source_location_for_substring): Rename to.. (get_location_within_string): ...this. * tree-cfg.c: Replace "source_location" with "location_t". * tree-cfgcleanup.c: Likewise. * tree-diagnostic.c: Likewise. * tree-into-ssa.c: Likewise. * tree-outof-ssa.c: Likewise. * tree-parloops.c: Likewise. * tree-phinodes.c: Likewise. * tree-phinodes.h: Likewise. * tree-ssa-loop-ivopts.c: Likewise. * tree-ssa-loop-manip.c: Likewise. * tree-ssa-phiopt.c: Likewise. * tree-ssa-phiprop.c: Likewise. * tree-ssa-threadupdate.c: Likewise. * tree-ssa.c: Likewise. * tree-ssa.h: Likewise. * tree-vect-loop-manip.c: Likewise. gcc/c-family/ChangeLog: * c-common.c (c_get_substring_location): Update for renaming of get_source_location_for_substring to get_location_within_string. * c-lex.c: Replace "source_location" with "location_t". * c-opts.c: Likewise. * c-ppoutput.c: Likewise. gcc/c/ChangeLog: * c-decl.c: Replace "source_location" with "location_t". * c-tree.h: Likewise. * c-typeck.c: Likewise. * gimple-parser.c: Likewise. gcc/cp/ChangeLog: * call.c: Replace "source_location" with "location_t". * cp-tree.h: Likewise. * cvt.c: Likewise. * name-lookup.c: Likewise. * parser.c: Likewise. * typeck.c: Likewise. gcc/fortran/ChangeLog: * cpp.c: Replace "source_location" with "location_t". * gfortran.h: Likewise. gcc/go/ChangeLog: * go-gcc-diagnostics.cc: Replace "source_location" with "location_t". * go-gcc.cc: Likewise. * go-linemap.cc: Likewise. * go-location.h: Likewise. * gofrontend/README: Likewise. gcc/jit/ChangeLog: * jit-playback.c: Replace "source_location" with "location_t". gcc/testsuite/ChangeLog: * g++.dg/plugin/comment_plugin.c: Replace "source_location" with "location_t". * gcc.dg/plugin/diagnostic_plugin_test_show_locus.c: Likewise. libcc1/ChangeLog: * libcc1plugin.cc: Replace "source_location" with "location_t". (plugin_context::get_source_location): Rename to... (plugin_context::get_location_t): ...this. * libcp1plugin.cc: Likewise. libcpp/ChangeLog: * charset.c: Replace "source_location" with "location_t". * directives-only.c: Likewise. * directives.c: Likewise. * errors.c: Likewise. * expr.c: Likewise. * files.c: Likewise. * include/cpplib.h: Likewise. Rename MAX_SOURCE_LOCATION to MAX_LOCATION_T. * include/line-map.h: Likewise. * init.c: Likewise. * internal.h: Likewise. * lex.c: Likewise. * line-map.c: Likewise. * location-example.txt: Likewise. * macro.c: Likewise. * pch.c: Likewise. * traditional.c: Likewise. From-SVN: r266085
2018-11-13 20:05:03 +00:00
location_t loc)
{
if (is_known_idempotent_file (pfile, file, type == IT_IMPORT))
return false;
int sysp = 0;
char *buf = nullptr;
/* Check C++ module include translation. */
if (!file->header_unit && type < IT_HEADER_HWM
/* Do not include translate include-next. */
&& type != IT_INCLUDE_NEXT
&& pfile->cb.translate_include)
buf = (pfile->cb.translate_include
(pfile, pfile->line_table, loc, file->path));
if (buf)
{
/* We don't increment the line number at the end of a buffer,
because we don't usually need that location (we're popping an
include file). However in this case we do want to do the
increment. So push a writable buffer of two newlines to acheive
that. (We also need an extra newline, so this looks like a regular
file, which we do that to to make sure we don't fall off the end in the
middle of a line. */
static uchar newlines[] = "\n\n\n";
cpp_push_buffer (pfile, newlines, 2, true);
size_t len = strlen (buf);
buf[len] = '\n'; /* See above */
cpp_buffer *buffer
= cpp_push_buffer (pfile, reinterpret_cast<unsigned char *> (buf),
len, true);
buffer->to_free = buffer->buf;
file->header_unit = +1;
_cpp_mark_file_once_only (pfile, file);
}
else
{
/* Not a header unit, and we know it. */
file->header_unit = -1;
if (!read_file (pfile, file, loc))
return false;
if (!has_unique_contents (pfile, file, type == IT_IMPORT, loc))
return false;
if (pfile->buffer && file->dir)
sysp = MAX (pfile->buffer->sysp, file->dir->sysp);
/* Add the file to the dependencies on its first inclusion. */
if (CPP_OPTION (pfile, deps.style) > (sysp != 0)
&& !file->stack_count
&& file->path[0]
&& !(pfile->main_file == file
&& CPP_OPTION (pfile, deps.ignore_main_file)))
deps_add_dep (pfile->deps, file->path);
/* Clear buffer_valid since _cpp_clean_line messes it up. */
file->buffer_valid = false;
file->stack_count++;
/* Stack the buffer. */
cpp_buffer *buffer
= cpp_push_buffer (pfile, file->buffer, file->st.st_size,
CPP_OPTION (pfile, preprocessed)
&& !CPP_OPTION (pfile, directives_only));
buffer->file = file;
buffer->sysp = sysp;
buffer->to_free = file->buffer_start;
/* Initialize controlling macro state. */
pfile->mi_valid = true;
pfile->mi_cmacro = 0;
}
/* In the case of a normal #include, we're now at the start of the
line *following* the #include. A separate location_t for this
location makes no sense, until we do the LC_LEAVE.
This does not apply if we found a PCH file, we're not a regular
include, or we ran out of locations. */
bool decrement = (file->pchname == NULL
&& type < IT_DIRECTIVE_HWM
&& (pfile->line_table->highest_location
!= LINE_MAP_MAX_LOCATION - 1));
if (decrement)
pfile->line_table->highest_location--;
/* Normally a header unit becomes an __import directive in the current file,
but with -include we need something to LC_LEAVE to trigger the file_change
hook and continue to the next -include or the main source file. */
if (file->header_unit <= 0 || type == IT_CMDLINE)
/* Add line map and do callbacks. */
_cpp_do_file_change (pfile, LC_ENTER, file->path,
/* With preamble injection, start on line zero,
so the preamble doesn't appear to have been
included from line 1. Likewise when
starting preprocessed, we expect an initial
locating line. */
type == IT_PRE_MAIN ? 0 : 1, sysp);
else if (decrement)
{
/* Adjust the line back one so we appear on the #include line itself. */
const line_map_ordinary *map
= LINEMAPS_LAST_ORDINARY_MAP (pfile->line_table);
linenum_type line = SOURCE_LINE (map, pfile->line_table->highest_line);
linemap_line_start (pfile->line_table, line - 1, 0);
}
return true;
[multiple changes] 1998-10-28 16:10 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * c-lang.c: Declare extern char *yy_cur if USE_CPPLIB. (lang_init): Call check_newline always. * c-lex.c (init_parse) [USE_CPPLIB=1]: After calling cpp_start_read, set yy_cur and yy_lim to read from parse_in.token_buffer, so that we'll see the first #line directive. * cpplib.c (cpp_start_read): finclude the main input file before processing -include/-imacros. Process -imacros and -include separately, and handle -include by stacking a buffer for the file in question as if it'd been #included. * toplev.c (documented_lang_options) Recognize -H when USE_CPPLIB is on. 1998-10-28 16:09 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * cpplib.c: Merge do_once into do_pragma. Break file handling code out of do_include. Move append_include_chain, deps_output, file_cleanup, redundant_include_p, import_hash, lookup_import, add_import, read_filename_string, read_name_map, open_include_file, finclude, safe_read to cppfiles.c. Move prototypes for deps_output, append_include_chain, finclude to cpplib.h. Move definition of struct file_name_list there also. * cppfiles.c: New file. Contains all the above functions broken out of cpplib.c; also hack_vms_include_specification from cccp.c and find_include_file, a new function broken out of do_include. * Makefile.in (cppmain): Depend on cppfiles.o. (fix-header): Likewise. (cppfiles.o): New target. * configure.in (--enable-c-cpplib): Add cppfiles.o to extra_c_objs. Add ../cppfiles.o to extra_cxx_objs. From-SVN: r23424
1998-10-29 11:54:13 +00:00
}
/* Mark FILE to be included once only. */
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
void
_cpp_mark_file_once_only (cpp_reader *pfile, _cpp_file *file)
cpplex.c (cpp_output_tokens, [...]): New public interfaces. * cpplex.c (cpp_output_tokens, cpp_scan_buffer_nooutput): New public interfaces. (safe_fwrite, output_line_command): New static functions. (cpp_expand_to_buffer): Now private to cpplib. (cpp_scan_buffer): Take a printer. * cpphash.h: Update prototypes. * cpplib.h: Update prototypes. (cpp_printer): New. (cpp_buffer): Remove last_nominal_fname. (cpp_reader): Remove lineno. * cppmain.c: Use a cpp_printer. * fix-header.c: No need to inhibit line commands. Call cpp_start_read with no printer. * cpperror.c (cpp_notice_from_errno): Provide default name. * cppfiles.c (make_IHASH, _cpp_fake_ihash): New functions. (find_include_file, cpp_read_file): Use make_IHASH. (file_cleanup): Set control_macro and clear input_stack_listing_current here. (_cpp_execute_include): Don't output entering-file marker. * cpphash.c (special_symbol): Look for the line number in the buffer, not the reader. (_cpp_macroexpand): No need to disable line commands. (_cpp_dump_definition): No need to generate line commands. (dump_hash_helper): Remove excess newline from output. * cppinit.c (dump_special_to_buffer): No need to generate line commands. (cpp_printer_init): New. (cpp_start_read): Take a printer, and start it up if it's not NULL. No need to generate line commands. (cpp_finish): Expect no buffers stacked at all. Take a printer argument, and flush the output buffer if it's not NULL. * cpplex.c (_cpp_lex_token): Return EOF if there's no buffer. Don't put two hashes at the beginning of an assertion. (cpp_get_token): Don't increment pfile->lineno or emit line commands here. Return EOF if there's no buffer when we get EOF. * cpplib.c (do_define, skip_if_group): No need to disable line commands. (_cpp_output_line_command): Delete function. (do_line): Don't emit line commands here, but set things up so they will be emitted if necessary. Use _cpp_fake_ihash to make unique nominal_fnames if necessary. (do_elif, do_else, _cpp_handle_eof): Call cpp_error_with_line with 0 for column, not -1. (_cpp_handle_eof): Don't set the control macro here. Don't clear input_stack_listing_current here. Don't emit line commands. From-SVN: r33159
2000-04-14 23:29:45 +00:00
{
pfile->seen_once_only = true;
file->once_only = true;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
}
/* Return the directory from which searching for FNAME should start,
considering the directive TYPE and ANGLE_BRACKETS. If there is
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
nothing left in the path, returns NULL. */
static struct cpp_dir *
search_path_head (cpp_reader *pfile, const char *fname, int angle_brackets,
enum include_type type, bool suppress_diagnostic)
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
{
cpp_dir *dir;
_cpp_file *file;
if (IS_ABSOLUTE_PATH (fname))
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
return &pfile->no_search_path;
/* pfile->buffer is NULL when processing an -include command-line flag. */
file = pfile->buffer == NULL ? pfile->main_file : pfile->buffer->file;
/* For #include_next, skip in the search path past the dir in which
the current file was found, but if it was found via an absolute
path use the normal search logic. */
if (type == IT_INCLUDE_NEXT && file->dir
&& file->dir != &pfile->no_search_path)
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
dir = file->dir->next;
else if (angle_brackets)
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
dir = type == IT_EMBED ? pfile->embed_include : pfile->bracket_include;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
else if (type == IT_CMDLINE)
/* -include and -imacros use the #include "" chain with the
preprocessor's cwd prepended. */
return make_cpp_dir (pfile, "./", false);
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
else if (pfile->quote_ignores_source_dir && type != IT_EMBED)
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
dir = pfile->quote_include;
else
Represent column numbers using line-map's source_location. The "next available source_location" is now managed internally by line-maps.c rather than by clients. * line-map.h (struct line_map): New field column_bits. <from_line>: Rename field to start_location. (struct line_maps): New fields highest_location and max_column_hint. (linemap_check_files_exited): New declaration. (linemap_line_start): New declaration. (linemap_add): Remove from_line parameter; use highest_location field. (SOURCE_LINE, LAST_SOURCE_LINE): Modify to use column_bits. (SOURCE_COLUMN, LAST_SOURCE_LINE_LOCATION): New macros. (CURRENT_LINE_MAP): Remove macro. (linemap_position_for_column): New inline function. * line-map.c (linemap_init): Clear new fields. (linemap_check_files_exited): New function, extracted from ... (linemap_free): Use linemap_check_files_exited. (linemap_add): Remove from_line parameter. Various updates. (linemap_line_start): New function. (linemap_lookeup): Update for new field names. * cpphash.h (struct cpp_reader) <map>: Field removed. Because linemap_position_for_column may unpredictably change the current map, it is cleaner and simpler for us to not cache it in cpp_reader. (struct cpp_buffer): New sysp field. Changed warned_cplusplus_comments and from_stage3 to bitfields. * cppinit.c (cpp_read_min_file): pfile->map no longer exists. * cpplib.c (do_line, do_linemarker, _cpp_do_file_change): Get current map using linemap_lookup. (do_linemarker): Also set buffer's sysp field. (destringize_and_run): No longer need to decrement current line. * cppfiles.c (_cpp_stack_file): Set sysp from and in buffer. (search_path_head, open_file_failed): Use buffer's sysp. (cpp_make_system_header): Get current map using linemap_lookup. Also set buffer's sysp flag. * cppmacro.c (_cpp_builtin_macro_text): Likewise use linemap_lookup. * cpphash.h (CPP_INCREMENT_LINE): New macro. (struct cpp_buffer): Moved fields saved_cur, saved_rlimit to ... (struct cpp_reader): ... and adding saved_line_base field. * cpptrad.c (_cpp_overlay_buffer, _cpp_remove_overlay): Update accordingly. Don't adjust line. (_cpp_scan_out_logical_line): Use CPP_INCREMENT_LINE. * cpphash.c (CPP_IN_SYSTEM_HEADER): Replaced macro by ... (cpp_in_system_header): ... new inline function, using buffer's sysp. * cpperror.c (_cpp_begin_message): Update to use cpp_in_system_header. * cpplex.c (_cpp_lex_direct): Likewise. * cppmacro.c (_cpp_builtin_macro_text): Likewise. * cppmacro.c (_cpp_create_definition): Use buffer's sysp field. * cpplib.h (struct cpp_token): Rename line field to src_loc. Remove col field as it is now subsumed by src_loc. * cpperror.c: Update various field, parameter, and macro names. (print_location): If col==0, try SOURCE_COLUMN of line. (cpp_error): Use cur_token's src_loc field, rather than line+col. * cpplib.c (do_diagnostic): Token's src_loc fields replaces line+col. * cpplex.c (_cpp_process_line_notes, _cpp_lex_direct, _cpp_skip_block_comment): Use CPP_INCREMENT_LINE. (_cpp_temp_token): Replace cpp_token's line+col fields by src_loc. (_cpp_get_fresh_line): Don't need to adjust line for missing newline. (_cpp_lex_direct): Use linemap_position_for_column. * c-ppoutput.c (maybe_print_line, print_line): Don't take map parameter. Instead get it from the line_table global. Adjust callers. (print): Remove map field. Replace line field to src_line. (init_pp_output, account_for_newlines, maybe_print_line): Adjust. (cb_line_change): Use SOURCE_COLUMN. Minor optimizations. (pp_file_change): Use MAIN_FILE_P since we cannot checked print.map. Use LAST_SOURCE_LINE_LOCATION to "catch up" after #include. * cpptrad.c (copy_comment): Rename variable. * c-lex.c (map): Remove static variable, for same reason we removed cpp_reader's map field. (cb_line_change, cb_def_pragma, cb_define, cb_undef): Hence we need to call linemap_lookup. (cb_line_change): Token's line field replaced by src_loc. (fe_file_change): Use MAINFILE_P and LAST_SOURCE_LINE macros. Don't save new_map. * cpphash.h, cpperror.c, cpplib.h: Some renames of fileline to source_location. From-SVN: r77663
2004-02-11 07:29:30 -08:00
return make_cpp_dir (pfile, dir_name_of_file (file),
pfile->buffer ? pfile->buffer->sysp : 0);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (dir == NULL && !suppress_diagnostic)
cpp_error (pfile, CPP_DL_ERROR,
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
"no include path in which to search for %s", fname);
return dir;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Strip the basename from the file's path. It ends with a slash if
of nonzero length. Note that this procedure also works for
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
<stdin>, which is represented by the empty string. */
static const char *
dir_name_of_file (_cpp_file *file)
{
if (!file->dir_name)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
size_t len = lbasename (file->path) - file->path;
configure.ac: Check declarations for asprintf and vasprintf. * configure.ac: Check declarations for asprintf and vasprintf. * config.in: Regenerate. * configure: Likewise. * charset.c (conversion_loop): Use XRESIZEVEC. (convert_no_conversion): Likewise. (convert_using_iconv): Likewise. (init_iconv_desc): Cast return value of alloca. (cpp_host_to_exec_charset): Use XNEWVEC. (emit_numeric_escape): Use XRESIZEVEC. (cpp_interpret_string): Use XNEWVEC. (cpp_interpret_string): Use XRESIZEVEC. (_cpp_interpret_identifier): Cast return value of alloca. (_cpp_convert_input): Use XNEWVEC and XRESIZEVEC. * directives.c (glue_header_name): Use XNEWVEC and XRESIZEVEC. (parse_include): Use XNEWVEC. (insert_pragma_entry): Rename local variable "new" to "new_entry". (save_registered_pragmas): Cast return value of xmemdup. (destringize_and_run): Same for alloca. (parse_assertion): Likewise. (do_assert): Cast allocated storage to proper type. (cpp_define): Likewise. (_cpp_define_builtin): Likewise. (cpp_undef): Likewise. (handle_assertion): Likewise. (cpp_push_buffer): Rename local variable "new" to "new_buffer". * expr.c (CPP_UPLUS): Cast value to type cpp_ttype. (CPP_UMINUS): Likewise. (struct cpp_operator): Rename from struct operator. (_cpp_expand_op_stack): Use XRESIZEVEC. * files.c (pch_open_file): Use XNEWVEC. (pch_open_file): Use XRESIZEVEC. (read_file_guts): Use XNEWVEC and XRESIZEVEC. (dir_name_of_file): Use XNEWVEC. (make_cpp_file): Use XCNEW. (make_cpp_dir): Likewise. (allocate_file_hash_entries): USE XNEWVEC. (cpp_included): Cast return value of htab_find_with_hash. (append_file_to_dir): Use XNEWVEC. (read_filename_string): Likewise. Use XRESIZEVEC too. (read_name_map): Cast return value of alloca. Use XRESIZEVEC. (remap_filename): Use XNEWVEC. (struct pchf_entry): Move definition out of struct pchf_data. (_cpp_save_file_entries): Use XCNEWVAR. (_cpp_read_file_entries): Use XNEWVAR. * identifiers.c (alloc_node): Use XOBNEW. * init.c (cpp_create_reader): Use XCNEW. (cpp_init_builtins): Cast of b->value to enum builtin_type. (read_original_directory): Cast return value of alloca. * lex.c (add_line_note): Use XRESIZEVEC. (warn_about_normalization): Use XNEWVEC. (_cpp_lex_direct): Cast node->directive_index to (enum cpp_ttype). (new_buff): Use XNEWVEC. * line-map.c (linemap_add): Use XRESIZEVEC. * macro.c (builtin_macro): Cast return value of alloca. (paste_tokens): Likewise. (expand_arg): Use XNEWVEC and XRESIZEVEC. (_cpp_save_parameter): Use XRESIZEVEC. (create_iso_definition): Cast allocated storage to proper type. (_cpp_create_definition): Likewise. (cpp_macro_definition): Use XRESIZEVEC. * makedepend.c (add_clm): Use XNEW. (add_dir): Likewise. * mkdeps.c (munge): Use XNEWVEC. (deps_init): Use XCNEW. (deps_add_target): Use XRESIZEVEC. (deps_add_default_target): Cast return value of alloca. (deps_add_dep): Use XRESIZEVEC. (deps_add_vpath): Likewise. Use XNEWVEC too. (deps_restore): Likewise. * pch.c (save_idents): Use XNEW and XNEWVEC. (cpp_save_state): Use XNEW. (count_defs): Cast return value of htab_find. (write_defs): Likewise. (cpp_write_pch_deps): Use XNEWVEC. (collect_ht_nodes): Use XRESIZEVEC. (cpp_valid_state): Use XNEWVEC. (save_macros): Use XRESIZEVEC. Cast return value of xmemdup. * symtab.c (ht_create): Use XCNEW. (ht_lookup_with_hash): Cast return value of obstack_copy0. (ht_expand): Use XCNEWVEC. * system.h (HAVE_DESIGNATED_INITIALIZERS): False if __cplusplus. (bool): Do not define if __cplusplus. From-SVN: r100295
2005-05-28 15:52:48 +00:00
char *dir_name = XNEWVEC (char, len + 1);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
memcpy (dir_name, file->path, len);
dir_name[len] = '\0';
file->dir_name = dir_name;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
return file->dir_name;
}
/* Handles #include-family directives (distinguished by TYPE),
including HEADER, and the command line -imacros and -include.
Returns true if a buffer was stacked. */
bool
_cpp_stack_include (cpp_reader *pfile, const char *fname, int angle_brackets,
Eliminate source_location in favor of location_t Historically GCC used location_t, while libcpp used source_location. This inconsistency has been annoying me for a while, so this patch removes source_location in favor of location_t throughout (as the latter is shorter). gcc/ChangeLog: * builtins.c: Replace "source_location" with "location_t". * diagnostic-show-locus.c: Likewise. * diagnostic.c: Likewise. * dumpfile.c: Likewise. * gcc-rich-location.h: Likewise. * genmatch.c: Likewise. * gimple.h: Likewise. * gimplify.c: Likewise. * input.c: Likewise. * input.h: Likewise. Eliminate the typedef. * omp-expand.c: Likewise. * selftest.h: Likewise. * substring-locations.h (get_source_location_for_substring): Rename to.. (get_location_within_string): ...this. * tree-cfg.c: Replace "source_location" with "location_t". * tree-cfgcleanup.c: Likewise. * tree-diagnostic.c: Likewise. * tree-into-ssa.c: Likewise. * tree-outof-ssa.c: Likewise. * tree-parloops.c: Likewise. * tree-phinodes.c: Likewise. * tree-phinodes.h: Likewise. * tree-ssa-loop-ivopts.c: Likewise. * tree-ssa-loop-manip.c: Likewise. * tree-ssa-phiopt.c: Likewise. * tree-ssa-phiprop.c: Likewise. * tree-ssa-threadupdate.c: Likewise. * tree-ssa.c: Likewise. * tree-ssa.h: Likewise. * tree-vect-loop-manip.c: Likewise. gcc/c-family/ChangeLog: * c-common.c (c_get_substring_location): Update for renaming of get_source_location_for_substring to get_location_within_string. * c-lex.c: Replace "source_location" with "location_t". * c-opts.c: Likewise. * c-ppoutput.c: Likewise. gcc/c/ChangeLog: * c-decl.c: Replace "source_location" with "location_t". * c-tree.h: Likewise. * c-typeck.c: Likewise. * gimple-parser.c: Likewise. gcc/cp/ChangeLog: * call.c: Replace "source_location" with "location_t". * cp-tree.h: Likewise. * cvt.c: Likewise. * name-lookup.c: Likewise. * parser.c: Likewise. * typeck.c: Likewise. gcc/fortran/ChangeLog: * cpp.c: Replace "source_location" with "location_t". * gfortran.h: Likewise. gcc/go/ChangeLog: * go-gcc-diagnostics.cc: Replace "source_location" with "location_t". * go-gcc.cc: Likewise. * go-linemap.cc: Likewise. * go-location.h: Likewise. * gofrontend/README: Likewise. gcc/jit/ChangeLog: * jit-playback.c: Replace "source_location" with "location_t". gcc/testsuite/ChangeLog: * g++.dg/plugin/comment_plugin.c: Replace "source_location" with "location_t". * gcc.dg/plugin/diagnostic_plugin_test_show_locus.c: Likewise. libcc1/ChangeLog: * libcc1plugin.cc: Replace "source_location" with "location_t". (plugin_context::get_source_location): Rename to... (plugin_context::get_location_t): ...this. * libcp1plugin.cc: Likewise. libcpp/ChangeLog: * charset.c: Replace "source_location" with "location_t". * directives-only.c: Likewise. * directives.c: Likewise. * errors.c: Likewise. * expr.c: Likewise. * files.c: Likewise. * include/cpplib.h: Likewise. Rename MAX_SOURCE_LOCATION to MAX_LOCATION_T. * include/line-map.h: Likewise. * init.c: Likewise. * internal.h: Likewise. * lex.c: Likewise. * line-map.c: Likewise. * location-example.txt: Likewise. * macro.c: Likewise. * pch.c: Likewise. * traditional.c: Likewise. From-SVN: r266085
2018-11-13 20:05:03 +00:00
enum include_type type, location_t loc)
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
{
/* For -include command-line flags we have type == IT_CMDLINE.
When the first -include file is processed we have the case, where
pfile->cur_token == pfile->cur_run->base, we are directly called up
by the front end. However in the case of the second -include file,
we are called from _cpp_lex_token -> _cpp_get_fresh_line ->
cpp_push_include, with pfile->cur_token != pfile->cur_run->base,
and pfile->cur_token[-1].src_loc not (yet) initialized.
However, when the include file cannot be found, we need src_loc to
be initialized to some safe value: 0 means UNKNOWN_LOCATION. */
if (type == IT_CMDLINE && pfile->cur_token != pfile->cur_run->base)
pfile->cur_token[-1].src_loc = 0;
cpp_dir *dir = search_path_head (pfile, fname, angle_brackets, type);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (!dir)
return false;
[PR 80005] Fix __has_include __has_include is funky in that it is macro-like from the POV of #ifdef and friends, but lexes its parenthesize argument #include-like. We were failing the second part of that, because we used a forwarding macro to an internal name, and hence always lexed the argument in macro-parameter context. We componded that by not setting the right flag when lexing, so it didn't even know. Mostly users got lucky. This reimplements the handline. 1) Remove the forwarding, but declare object-like macros that expand to themselves. This satisfies the #ifdef requirement 2) Correctly set angled_brackets when lexing the parameter. This tells the lexer (a) <...> is a header name and (b) "..." is too (not a string). 3) Remove the in__has_include lexer state, just tell find_file that that's what's happenning, so it doesn't emit an error. We lose the (undocumented) ability to #undef __has_include. That may well have been an accident of implementation. There are no tests for it. We gain __has_include behaviour for all users of the preprocessors -- not just the C-family ones that defined a forwarding macro. libcpp/ PR preprocessor/80005 * include/cpplib.h (BT_HAS_ATTRIBUTE): Fix comment. * internal.h (struct lexer_state): Delete in__has_include field. (struct spec_nodes): Rename n__has_include{,_next}__ fields. (_cpp_defined_macro_p): New. (_cpp_find_file): Add has_include parm. * directives.c (lex_macro_node): Combine defined, __has_inline{,_next} checking. (do_ifdef, do_ifndef): Use _cpp_defined_macro_p. (_cpp_init_directives): Refactor. * expr.c (parse_defined): Use _cpp_defined_macro_p. (eval_token): Adjust parse_has_include calls. (parse_has_include): Add OP parameter. Reimplement. * files.c (_cpp_find_file): Add HAS_INCLUDE parm. Use it to inhibit error message. (_cpp_stack_include): Adjust _cpp_find_file call. (_cpp_fake_include, _cpp_compare_file_date): Likewise. (open_file_failed): Remove in__has_include check. (_cpp_has_header): Adjust _cpp_find_file call. * identifiers.c (_cpp_init_hashtable): Don't init __has_include{,_next} here ... * init.c (cpp_init_builtins): ... init them here. Define as macros. (cpp_read_main_file): Adjust _cpp_find_file call. * pch.c (cpp_read_state): Adjust __has_include{,_next} access. * traditional.c (_cpp_scan_out_locgical_line): Likewise. gcc/c-family/ PR preprocessor/80005 * c-cppbuiltins.c (c_cpp_builtins): Don't define __has_include{,_next}. gcc/testsuite/ PR preprocessor/80005 * g++.dg/cpp1y/feat-cxx14.C: Adjust. * g++.dg/cpp1z/feat-cxx17.C: Adjust. * g++.dg/cpp2a/feat-cxx2a.C: Adjust. * g++.dg/cpp/pr80005.C: New.
2020-01-20 05:39:59 -08:00
_cpp_file *file = _cpp_find_file (pfile, fname, dir, angle_brackets,
type == IT_DEFAULT ? _cpp_FFK_PRE_INCLUDE
: _cpp_FFK_NORMAL, loc);
gcc: * config.gcc (*-*-linux* | frv-*-*linux* | *-*-kfreebsd*-gnu | *-*-knetbsd*-gnu | *-*-gnu* | *-*-kopensolaris*-gnu): Use glibc-c.o in c_target_objs and cxx_target_objs. Use t-glibc in tmake_file. Set target_has_targetcm. (tilegx-*-linux*, tilepro-*-linux*): Append to c_target_objs and cxx_target_objs rather than overriding previous value. * config/glibc-c.c, config/t-glibc: New. * doc/tm.texi.in (TARGET_C_PREINCLUDE): New @hook. * doc/tm.texi: Regenerate. * hooks.c (hook_constcharptr_void_null): New. * hooks.h (hook_constcharptr_void_null): Declare. gcc/c-family: * c-common.h (pch_cpp_save_state): Declare. * c-target.def (c_preinclude): New hook. * c-opts.c (done_preinclude): New. (push_command_line_include): Handle default preincluded header. (cb_file_change): Call pch_cpp_save_state when calling push_command_line_include. * c-pch.c (pch_ready_to_save_cpp_state, pch_cpp_state_saved) (pch_cpp_save_state): New. (pch_init): Call pch_cpp_save_state conditionally, instead of calling cpp_save_state. gcc/testsuite: * gcc.dg/c99-predef-1.c: New test. * gcc.dg/cpp/cmdlne-dU-1.c, gcc.dg/cpp/cmdlne-dU-2.c, gcc.dg/cpp/cmdlne-dU-3.c, gcc.dg/cpp/cmdlne-dU-4.c, gcc.dg/cpp/cmdlne-dU-5.c, gcc.dg/cpp/cmdlne-dU-6.c, gcc.dg/cpp/cmdlne-dU-7.c, gcc.dg/cpp/cmdlne-dU-8.c, gcc.dg/cpp/cmdlne-dU-9.c, gcc.dg/cpp/cmdlne-dU-10.c, gcc.dg/cpp/cmdlne-dU-11.c, gcc.dg/cpp/cmdlne-dU-12.c, gcc.dg/cpp/cmdlne-dU-13.c, gcc.dg/cpp/cmdlne-dU-14.c, gcc.dg/cpp/cmdlne-dU-15.c, gcc.dg/cpp/cmdlne-dU-16.c, gcc.dg/cpp/cmdlne-dU-17.c, gcc.dg/cpp/cmdlne-dU-18.c, gcc.dg/cpp/cmdlne-dU-19.c, gcc.dg/cpp/cmdlne-dU-20.c, gcc.dg/cpp/cmdlne-dU-21.c, gcc.dg/cpp/cmdlne-dU-22.c, gcc.dg/cpp/mi5.c, gcc.dg/cpp/multiline.c: Add -nostdinc to dg-options. libcpp: * files.c (struct _cpp_file): Add implicit_preinclude. (pch_open_file): Allow a previously opened implicitly included file. (_cpp_find_file): Add implicit_preinclude argument. Free file and do not call open_file_failed if implicit_preinclude. Store implicit_preinclude value. (_cpp_stack_include, _cpp_fake_include, _cpp_compare_file_date): Update calls to _cpp_find_file. (_cpp_stack_include): Handle IT_DEFAULT. (cpp_push_default_include): New. * include/cpplib.h (cpp_push_default_include): Declare. * init.c (cpp_read_main_file): Update call to _cpp_find_file. * internal.h (enum include_type): Add IT_DEFAULT. (_cpp_find_file): Update prototype. From-SVN: r192715
2012-10-23 15:55:55 +01:00
if (type == IT_DEFAULT && file == NULL)
return false;
return _cpp_stack_file (pfile, file, type, loc);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
}
c++: header-unit build capability [PR 99023] This defect really required building header-units and include translation of pieces of the standard library. This adds smarts to the modules test harness to do that -- accept .X files as the source file, but provide '-x c++-system-header $HDR' in the options. The .X file will be considered by the driver to be a linker script and ignored (with a warning). Using this we can add 2 tests that end up building list_initializer and iostream, along with a test that iostream's build include-translates list_initializer's #include. That discovered a set of issues with the -flang-info-include-translate=HDR handling, also fixed and documented here. PR c++/99023 gcc/cp/ * module.cc (canonicalize_header_name): Use cpp_probe_header_unit. (maybe_translate_include): Fix note_includes comparison. (init_modules): Fix note_includes string termination. libcpp/ * include/cpplib.h (cpp_find_header_unit): Rename to ... (cpp_probe_header_unit): ... this. * internal.h (_cp_find_header_unit): Declare. * files.c (cpp_find_header_unit): Break apart to .. (test_header_unit): ... this, and ... (_cpp_find_header_unit): ... and, or and ... (cpp_probe_header_unit): ... this. * macro.c (cpp_get_token_1): Call _cpp_find_header_unit. gcc/ * doc/invoke.texi (flang-info-include-translate): Document header lookup behaviour. gcc/testsuite/ * g++.dg/modules/modules.exp: Bail on cross-testing. Add support for .X files. * g++.dg/modules/pr99023_a.X: New. * g++.dg/modules/pr99023_b.X: New.
2021-02-18 12:46:25 -08:00
/* NAME is a header file name, find the _cpp_file, if any. */
c++: header-unit build capability [PR 99023] This defect really required building header-units and include translation of pieces of the standard library. This adds smarts to the modules test harness to do that -- accept .X files as the source file, but provide '-x c++-system-header $HDR' in the options. The .X file will be considered by the driver to be a linker script and ignored (with a warning). Using this we can add 2 tests that end up building list_initializer and iostream, along with a test that iostream's build include-translates list_initializer's #include. That discovered a set of issues with the -flang-info-include-translate=HDR handling, also fixed and documented here. PR c++/99023 gcc/cp/ * module.cc (canonicalize_header_name): Use cpp_probe_header_unit. (maybe_translate_include): Fix note_includes comparison. (init_modules): Fix note_includes string termination. libcpp/ * include/cpplib.h (cpp_find_header_unit): Rename to ... (cpp_probe_header_unit): ... this. * internal.h (_cp_find_header_unit): Declare. * files.c (cpp_find_header_unit): Break apart to .. (test_header_unit): ... this, and ... (_cpp_find_header_unit): ... and, or and ... (cpp_probe_header_unit): ... this. * macro.c (cpp_get_token_1): Call _cpp_find_header_unit. gcc/ * doc/invoke.texi (flang-info-include-translate): Document header lookup behaviour. gcc/testsuite/ * g++.dg/modules/modules.exp: Bail on cross-testing. Add support for .X files. * g++.dg/modules/pr99023_a.X: New. * g++.dg/modules/pr99023_b.X: New.
2021-02-18 12:46:25 -08:00
static _cpp_file *
test_header_unit (cpp_reader *pfile, const char *name, bool angle,
location_t loc)
{
c++: header-unit build capability [PR 99023] This defect really required building header-units and include translation of pieces of the standard library. This adds smarts to the modules test harness to do that -- accept .X files as the source file, but provide '-x c++-system-header $HDR' in the options. The .X file will be considered by the driver to be a linker script and ignored (with a warning). Using this we can add 2 tests that end up building list_initializer and iostream, along with a test that iostream's build include-translates list_initializer's #include. That discovered a set of issues with the -flang-info-include-translate=HDR handling, also fixed and documented here. PR c++/99023 gcc/cp/ * module.cc (canonicalize_header_name): Use cpp_probe_header_unit. (maybe_translate_include): Fix note_includes comparison. (init_modules): Fix note_includes string termination. libcpp/ * include/cpplib.h (cpp_find_header_unit): Rename to ... (cpp_probe_header_unit): ... this. * internal.h (_cp_find_header_unit): Declare. * files.c (cpp_find_header_unit): Break apart to .. (test_header_unit): ... this, and ... (_cpp_find_header_unit): ... and, or and ... (cpp_probe_header_unit): ... this. * macro.c (cpp_get_token_1): Call _cpp_find_header_unit. gcc/ * doc/invoke.texi (flang-info-include-translate): Document header lookup behaviour. gcc/testsuite/ * g++.dg/modules/modules.exp: Bail on cross-testing. Add support for .X files. * g++.dg/modules/pr99023_a.X: New. * g++.dg/modules/pr99023_b.X: New.
2021-02-18 12:46:25 -08:00
if (cpp_dir *dir = search_path_head (pfile, name, angle, IT_INCLUDE))
return _cpp_find_file (pfile, name, dir, angle, _cpp_FFK_NORMAL, loc);
c++: header-unit build capability [PR 99023] This defect really required building header-units and include translation of pieces of the standard library. This adds smarts to the modules test harness to do that -- accept .X files as the source file, but provide '-x c++-system-header $HDR' in the options. The .X file will be considered by the driver to be a linker script and ignored (with a warning). Using this we can add 2 tests that end up building list_initializer and iostream, along with a test that iostream's build include-translates list_initializer's #include. That discovered a set of issues with the -flang-info-include-translate=HDR handling, also fixed and documented here. PR c++/99023 gcc/cp/ * module.cc (canonicalize_header_name): Use cpp_probe_header_unit. (maybe_translate_include): Fix note_includes comparison. (init_modules): Fix note_includes string termination. libcpp/ * include/cpplib.h (cpp_find_header_unit): Rename to ... (cpp_probe_header_unit): ... this. * internal.h (_cp_find_header_unit): Declare. * files.c (cpp_find_header_unit): Break apart to .. (test_header_unit): ... this, and ... (_cpp_find_header_unit): ... and, or and ... (cpp_probe_header_unit): ... this. * macro.c (cpp_get_token_1): Call _cpp_find_header_unit. gcc/ * doc/invoke.texi (flang-info-include-translate): Document header lookup behaviour. gcc/testsuite/ * g++.dg/modules/modules.exp: Bail on cross-testing. Add support for .X files. * g++.dg/modules/pr99023_a.X: New. * g++.dg/modules/pr99023_b.X: New.
2021-02-18 12:46:25 -08:00
return nullptr;
}
/* NAME is a header file name, find the path we'll use to open it and infer that
it is a header-unit. */
c++: header-unit build capability [PR 99023] This defect really required building header-units and include translation of pieces of the standard library. This adds smarts to the modules test harness to do that -- accept .X files as the source file, but provide '-x c++-system-header $HDR' in the options. The .X file will be considered by the driver to be a linker script and ignored (with a warning). Using this we can add 2 tests that end up building list_initializer and iostream, along with a test that iostream's build include-translates list_initializer's #include. That discovered a set of issues with the -flang-info-include-translate=HDR handling, also fixed and documented here. PR c++/99023 gcc/cp/ * module.cc (canonicalize_header_name): Use cpp_probe_header_unit. (maybe_translate_include): Fix note_includes comparison. (init_modules): Fix note_includes string termination. libcpp/ * include/cpplib.h (cpp_find_header_unit): Rename to ... (cpp_probe_header_unit): ... this. * internal.h (_cp_find_header_unit): Declare. * files.c (cpp_find_header_unit): Break apart to .. (test_header_unit): ... this, and ... (_cpp_find_header_unit): ... and, or and ... (cpp_probe_header_unit): ... this. * macro.c (cpp_get_token_1): Call _cpp_find_header_unit. gcc/ * doc/invoke.texi (flang-info-include-translate): Document header lookup behaviour. gcc/testsuite/ * g++.dg/modules/modules.exp: Bail on cross-testing. Add support for .X files. * g++.dg/modules/pr99023_a.X: New. * g++.dg/modules/pr99023_b.X: New.
2021-02-18 12:46:25 -08:00
const char *
_cpp_find_header_unit (cpp_reader *pfile, const char *name, bool angle,
location_t loc)
{
if (_cpp_file *file = test_header_unit (pfile, name, angle, loc))
{
c++: header-unit build capability [PR 99023] This defect really required building header-units and include translation of pieces of the standard library. This adds smarts to the modules test harness to do that -- accept .X files as the source file, but provide '-x c++-system-header $HDR' in the options. The .X file will be considered by the driver to be a linker script and ignored (with a warning). Using this we can add 2 tests that end up building list_initializer and iostream, along with a test that iostream's build include-translates list_initializer's #include. That discovered a set of issues with the -flang-info-include-translate=HDR handling, also fixed and documented here. PR c++/99023 gcc/cp/ * module.cc (canonicalize_header_name): Use cpp_probe_header_unit. (maybe_translate_include): Fix note_includes comparison. (init_modules): Fix note_includes string termination. libcpp/ * include/cpplib.h (cpp_find_header_unit): Rename to ... (cpp_probe_header_unit): ... this. * internal.h (_cp_find_header_unit): Declare. * files.c (cpp_find_header_unit): Break apart to .. (test_header_unit): ... this, and ... (_cpp_find_header_unit): ... and, or and ... (cpp_probe_header_unit): ... this. * macro.c (cpp_get_token_1): Call _cpp_find_header_unit. gcc/ * doc/invoke.texi (flang-info-include-translate): Document header lookup behaviour. gcc/testsuite/ * g++.dg/modules/modules.exp: Bail on cross-testing. Add support for .X files. * g++.dg/modules/pr99023_a.X: New. * g++.dg/modules/pr99023_b.X: New.
2021-02-18 12:46:25 -08:00
if (file->fd > 0)
{
/* Don't leave it open. */
close (file->fd);
file->fd = 0;
}
file->header_unit = +1;
_cpp_mark_file_once_only (pfile, file);
return file->path;
}
c++: header-unit build capability [PR 99023] This defect really required building header-units and include translation of pieces of the standard library. This adds smarts to the modules test harness to do that -- accept .X files as the source file, but provide '-x c++-system-header $HDR' in the options. The .X file will be considered by the driver to be a linker script and ignored (with a warning). Using this we can add 2 tests that end up building list_initializer and iostream, along with a test that iostream's build include-translates list_initializer's #include. That discovered a set of issues with the -flang-info-include-translate=HDR handling, also fixed and documented here. PR c++/99023 gcc/cp/ * module.cc (canonicalize_header_name): Use cpp_probe_header_unit. (maybe_translate_include): Fix note_includes comparison. (init_modules): Fix note_includes string termination. libcpp/ * include/cpplib.h (cpp_find_header_unit): Rename to ... (cpp_probe_header_unit): ... this. * internal.h (_cp_find_header_unit): Declare. * files.c (cpp_find_header_unit): Break apart to .. (test_header_unit): ... this, and ... (_cpp_find_header_unit): ... and, or and ... (cpp_probe_header_unit): ... this. * macro.c (cpp_get_token_1): Call _cpp_find_header_unit. gcc/ * doc/invoke.texi (flang-info-include-translate): Document header lookup behaviour. gcc/testsuite/ * g++.dg/modules/modules.exp: Bail on cross-testing. Add support for .X files. * g++.dg/modules/pr99023_a.X: New. * g++.dg/modules/pr99023_b.X: New.
2021-02-18 12:46:25 -08:00
return nullptr;
}
/* NAME is a header file name, find the path we'll use to open it. But do not
infer it is a header unit. */
const char *
cpp_probe_header_unit (cpp_reader *pfile, const char *name, bool angle,
location_t loc)
{
if (_cpp_file *file = test_header_unit (pfile, name, angle, loc))
return file->path;
return nullptr;
}
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
/* Try to load FNAME with #embed/__has_embed parameters PARAMS.
If !PARAMS->has_embed, return new token in pfile->directive_result
(first token) and rest in a pushed non-macro context.
Returns 0 for not found/errors, 1 for non-empty resource and 2
for empty resource. */
int
_cpp_stack_embed (cpp_reader *pfile, const char *fname, bool angle,
struct cpp_embed_params *params)
{
cpp_dir *dir = search_path_head (pfile, fname, angle, IT_EMBED,
params->has_embed);
if (!dir)
return 0;
_cpp_file *file = _cpp_find_file (pfile, fname, dir, angle,
params->has_embed
? _cpp_FFK_HAS_EMBED : _cpp_FFK_EMBED,
params->loc);
if (!file)
return 0;
if (file->dont_read || file->err_no)
return 0;
_cpp_file *orig_file = file;
if (file->buffer_valid
&& (!S_ISREG (file->st.st_mode)
|| file->offset + (cpp_num_part) 0 > params->offset
|| (file->limit < file->st.st_size - file->offset + (size_t) 0
&& (params->offset - file->offset > (cpp_num_part) file->limit
|| file->limit - (params->offset
- file->offset) < params->limit))))
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
{
bool found = false;
if (S_ISREG (file->st.st_mode))
{
while (file->next_file
&& file->next_file->embed
&& file->next_file->buffer_valid
&& file->next_file->dir == file->dir
&& strcmp (file->name, file->next_file->name) == 0
&& strcmp (file->path, file->next_file->path) == 0)
{
file = file->next_file;
if (file->offset + (cpp_num_part) 0 <= params->offset
&& (file->limit >= (file->st.st_size - file->offset
+ (size_t) 0)
|| (params->offset
- file->offset <= (cpp_num_part) file->limit
&& file->limit - (params->offset
- file->offset) >= params->limit)))
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
{
found = true;
break;
}
}
}
if (!found)
{
_cpp_file *file2 = make_cpp_file (file->dir, file->name);
file2->path = xstrdup (file->path);
file2->next_file = file->next_file;
file2->embed = true;
file->next_file = file2;
file = file2;
}
}
if (!file->buffer_valid)
{
if (file->fd == -1 && !open_file (file))
{
if (params->has_embed)
file->deferred_error = true;
else
open_file_failed (pfile, file, 0, params->loc);
return 0;
}
if (S_ISBLK (file->st.st_mode))
{
if (params->has_embed)
{
close (file->fd);
file->fd = -1;
return 0;
}
cpp_error_at (pfile, CPP_DL_ERROR, params->loc,
"%s is a block device", file->path);
fail:
close (file->fd);
file->fd = -1;
file->dont_read = true;
return 0;
}
if (CPP_OPTION (pfile, deps.style)
&& !params->has_embed
&& file == orig_file
&& file->path[0])
deps_add_dep (pfile->deps, file->path);
bool regular = S_ISREG (file->st.st_mode) != 0;
ssize_t size, total, count;
uchar *buf;
if (regular)
{
cpp_num_part limit;
if (file->st.st_size + (cpp_num_part) 0 < params->offset)
limit = 0;
else if (file->st.st_size - params->offset < params->limit)
limit = file->st.st_size - params->offset;
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
else
limit = params->limit;
if (params->has_embed)
return limit != 0 ? 1 : 2;
if (limit > INTTYPE_MAXIMUM (ssize_t))
{
cpp_error_at (pfile, CPP_DL_ERROR, params->loc,
"%s is too large", file->path);
goto fail;
}
if (lseek (file->fd, params->offset, SEEK_CUR)
!= (off_t) params->offset)
{
cpp_errno_filename (pfile, CPP_DL_ERROR, file->path,
params->loc);
goto fail;
}
file->offset = params->offset;
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
file->limit = limit;
size = limit;
}
else if (params->has_embed)
return 2;
else if (params->limit > 8 * 1024)
size = 8 * 1024;
else
size = params->limit;
buf = XNEWVEC (uchar, size ? size : 1);
total = 0;
if (!regular && params->offset)
{
uchar *buf2 = buf;
ssize_t size2 = size;
cpp_num_part total2 = params->offset;
if (params->offset > 8 * 1024 && size < 8 * 1024)
{
size2 = 32 * 1024;
buf2 = XNEWVEC (uchar, size2);
}
do
{
if ((cpp_num_part) size2 > total2)
size2 = total2;
count = read (file->fd, buf2, size2);
if (count < 0)
{
cpp_errno_filename (pfile, CPP_DL_ERROR, file->path,
params->loc);
if (buf2 != buf)
free (buf2);
free (buf);
goto fail;
}
total2 -= count;
}
while (total2);
if (buf2 != buf)
free (buf2);
}
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
while ((count = read (file->fd, buf + total, size - total)) > 0)
{
total += count;
if (total == size)
{
if (regular || size + (cpp_num_part) 0 == params->limit)
break;
size = (size_t) size * 2;
if (size < 0)
{
if (params->limit <= INTTYPE_MAXIMUM (ssize_t))
size = params->limit;
else
{
cpp_error_at (pfile, CPP_DL_ERROR, params->loc,
"%s is too large", file->path);
free (buf);
goto fail;
}
}
else if (size + (cpp_num_part) 0 > params->limit)
size = params->limit;
buf = XRESIZEVEC (uchar, buf, size);
}
}
if (count < 0)
{
cpp_errno_filename (pfile, CPP_DL_ERROR, file->path, params->loc);
free (buf);
goto fail;
}
if (regular && total != size && STAT_SIZE_RELIABLE (file->st))
{
cpp_error_at (pfile, CPP_DL_WARNING, params->loc,
"%s is shorter than expected", file->path);
file->limit = total;
}
else if (!regular)
{
file->offset = params->offset;
file->limit = total;
}
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
file->buffer_start = buf;
file->buffer = buf;
file->buffer_valid = 1;
close (file->fd);
file->fd = -1;
}
else if (params->has_embed)
{
if (params->offset - file->offset > file->limit)
return 2;
size_t limit = file->limit - (params->offset - file->offset);
return limit && params->limit ? 1 : 2;
}
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
const uchar *buffer = file->buffer;
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
size_t limit = file->limit;
if (params->offset - file->offset > limit)
limit = 0;
else
{
buffer += params->offset - file->offset;
limit -= params->offset - file->offset;
}
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
if (params->limit < limit)
limit = params->limit;
/* For sizes larger than say 64 bytes, this is just a temporary
solution, we should emit a single new token which the FEs will
handle as an optimization. */
size_t max = INTTYPE_MAXIMUM (size_t) / sizeof (cpp_token);
if (limit > max / 2
|| (limit
? (params->prefix.count > max
|| params->suffix.count > max
|| (limit * 2 + params->prefix.count
+ params->suffix.count > max))
: params->if_empty.count > max))
{
cpp_error_at (pfile, CPP_DL_ERROR, params->loc,
"%s is too large", file->path);
return 0;
}
size_t len = 0;
for (size_t i = 0; i < limit; ++i)
{
if (buffer[i] < 10)
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
len += 2;
else if (buffer[i] < 100)
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
len += 3;
#if UCHAR_MAX == 255
else
len += 4;
#else
else if (buffer[i] < 1000)
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
len += 4;
else
{
char buf[64];
len += sprintf (buf, "%d", buffer[i]) + 1;
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
}
#endif
if (len > INTTYPE_MAXIMUM (ssize_t))
{
cpp_error_at (pfile, CPP_DL_ERROR, params->loc,
"%s is too large", file->path);
return 0;
}
}
uchar *s = len ? _cpp_unaligned_alloc (pfile, len) : NULL;
_cpp_buff *tok_buff = NULL;
cpp_token *toks = NULL, *tok = &pfile->directive_result;
size_t count = 0;
if (limit)
count = (params->prefix.count + limit * 2 - 1
+ params->suffix.count) - 1;
else if (params->if_empty.count)
count = params->if_empty.count - 1;
if (count)
{
tok_buff = _cpp_get_buff (pfile, count * sizeof (cpp_token));
toks = (cpp_token *) tok_buff->base;
}
cpp_embed_params_tokens *prefix
= limit ? &params->prefix : &params->if_empty;
if (prefix->count)
{
*tok = *prefix->base_run.base;
tok = toks;
tokenrun *cur_run = &prefix->base_run;
while (cur_run)
{
size_t cnt = (cur_run->next ? cur_run->limit
: prefix->cur_token) - cur_run->base;
cpp_token *t = cur_run->base;
if (cur_run == &prefix->base_run)
{
t++;
cnt--;
}
memcpy (tok, t, cnt * sizeof (cpp_token));
tok += cnt;
cur_run = cur_run->next;
}
}
for (size_t i = 0; i < limit; ++i)
{
tok->src_loc = params->loc;
tok->type = CPP_NUMBER;
tok->flags = NO_EXPAND;
if (i == 0)
tok->flags |= PREV_WHITE;
tok->val.str.text = s;
tok->val.str.len = sprintf ((char *) s, "%d", buffer[i]);
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
s += tok->val.str.len + 1;
if (tok == &pfile->directive_result)
tok = toks;
else
tok++;
if (i < limit - 1)
{
tok->src_loc = params->loc;
tok->type = CPP_COMMA;
tok->flags = NO_EXPAND;
tok++;
}
}
if (limit && params->suffix.count)
{
tokenrun *cur_run = &params->suffix.base_run;
cpp_token *orig_tok = tok;
while (cur_run)
{
size_t cnt = (cur_run->next ? cur_run->limit
: params->suffix.cur_token) - cur_run->base;
cpp_token *t = cur_run->base;
memcpy (tok, t, cnt * sizeof (cpp_token));
tok += cnt;
cur_run = cur_run->next;
}
orig_tok->flags |= PREV_WHITE;
}
pfile->directive_result.flags |= PREV_WHITE;
if (count)
{
_cpp_push_token_context (pfile, NULL, toks, count);
pfile->context->buff = tok_buff;
}
return limit ? 1 : 2;
}
/* Retrofit the just-entered main file asif it was an include. This
will permit correct include_next use, and mark it as a system
header if that's where it resides. We use filesystem-appropriate
prefix matching of the include path to locate the main file. */
void
cpp_retrofit_as_include (cpp_reader *pfile)
{
/* We should be the outermost. */
gcc_assert (!pfile->buffer->prev);
if (const char *name = pfile->main_file->name)
{
/* Locate name on the include dir path, using a prefix match. */
size_t name_len = strlen (name);
for (cpp_dir *dir = pfile->quote_include; dir; dir = dir->next)
if (dir->len < name_len
&& IS_DIR_SEPARATOR (name[dir->len])
&& !filename_ncmp (name, dir->name, dir->len))
{
pfile->main_file->dir = dir;
if (dir->sysp)
cpp_make_system_header (pfile, 1, 0);
break;
}
}
/* Initialize controlling macro state. */
pfile->mi_valid = true;
pfile->mi_cmacro = 0;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Could not open FILE. The complication is dependency output. */
static void
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
open_file_failed (cpp_reader *pfile, _cpp_file *file, int angle_brackets,
Eliminate source_location in favor of location_t Historically GCC used location_t, while libcpp used source_location. This inconsistency has been annoying me for a while, so this patch removes source_location in favor of location_t throughout (as the latter is shorter). gcc/ChangeLog: * builtins.c: Replace "source_location" with "location_t". * diagnostic-show-locus.c: Likewise. * diagnostic.c: Likewise. * dumpfile.c: Likewise. * gcc-rich-location.h: Likewise. * genmatch.c: Likewise. * gimple.h: Likewise. * gimplify.c: Likewise. * input.c: Likewise. * input.h: Likewise. Eliminate the typedef. * omp-expand.c: Likewise. * selftest.h: Likewise. * substring-locations.h (get_source_location_for_substring): Rename to.. (get_location_within_string): ...this. * tree-cfg.c: Replace "source_location" with "location_t". * tree-cfgcleanup.c: Likewise. * tree-diagnostic.c: Likewise. * tree-into-ssa.c: Likewise. * tree-outof-ssa.c: Likewise. * tree-parloops.c: Likewise. * tree-phinodes.c: Likewise. * tree-phinodes.h: Likewise. * tree-ssa-loop-ivopts.c: Likewise. * tree-ssa-loop-manip.c: Likewise. * tree-ssa-phiopt.c: Likewise. * tree-ssa-phiprop.c: Likewise. * tree-ssa-threadupdate.c: Likewise. * tree-ssa.c: Likewise. * tree-ssa.h: Likewise. * tree-vect-loop-manip.c: Likewise. gcc/c-family/ChangeLog: * c-common.c (c_get_substring_location): Update for renaming of get_source_location_for_substring to get_location_within_string. * c-lex.c: Replace "source_location" with "location_t". * c-opts.c: Likewise. * c-ppoutput.c: Likewise. gcc/c/ChangeLog: * c-decl.c: Replace "source_location" with "location_t". * c-tree.h: Likewise. * c-typeck.c: Likewise. * gimple-parser.c: Likewise. gcc/cp/ChangeLog: * call.c: Replace "source_location" with "location_t". * cp-tree.h: Likewise. * cvt.c: Likewise. * name-lookup.c: Likewise. * parser.c: Likewise. * typeck.c: Likewise. gcc/fortran/ChangeLog: * cpp.c: Replace "source_location" with "location_t". * gfortran.h: Likewise. gcc/go/ChangeLog: * go-gcc-diagnostics.cc: Replace "source_location" with "location_t". * go-gcc.cc: Likewise. * go-linemap.cc: Likewise. * go-location.h: Likewise. * gofrontend/README: Likewise. gcc/jit/ChangeLog: * jit-playback.c: Replace "source_location" with "location_t". gcc/testsuite/ChangeLog: * g++.dg/plugin/comment_plugin.c: Replace "source_location" with "location_t". * gcc.dg/plugin/diagnostic_plugin_test_show_locus.c: Likewise. libcc1/ChangeLog: * libcc1plugin.cc: Replace "source_location" with "location_t". (plugin_context::get_source_location): Rename to... (plugin_context::get_location_t): ...this. * libcp1plugin.cc: Likewise. libcpp/ChangeLog: * charset.c: Replace "source_location" with "location_t". * directives-only.c: Likewise. * directives.c: Likewise. * errors.c: Likewise. * expr.c: Likewise. * files.c: Likewise. * include/cpplib.h: Likewise. Rename MAX_SOURCE_LOCATION to MAX_LOCATION_T. * include/line-map.h: Likewise. * init.c: Likewise. * internal.h: Likewise. * lex.c: Likewise. * line-map.c: Likewise. * location-example.txt: Likewise. * macro.c: Likewise. * pch.c: Likewise. * traditional.c: Likewise. From-SVN: r266085
2018-11-13 20:05:03 +00:00
location_t loc)
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
{
line-map.h (struct line_maps): New field highest_line. * line-map.h (struct line_maps): New field highest_line. (linemap_position_for_column): Make non-inline function. (LINEMAP_POSITION_FOR_COLUMN): New macro. * line-map.c (linemap_init): Clear highest_line field. (linemap_add): Set highest_line field. (linemap_line_start): Minor optimization - use highest_line field. Reduce maximum column hint to 10000. Update highest_line field. (linemap_position_for_column): Moved from line-map.h. Optimize a bit. * cpphash.h (struct cpp_reader): Remove line field - instead use line_table->highest_line. (saved_line): Remove unused field. (CPP_INCREMENT_FILE): Don't do linemap_lookup - just use newest map. Use line_table's highest_line field instead of cpp_reader's line. * cpplib.c (start_directive): Likewise use highest_line field. (do_line, do_linemarker): Likewise just use newest map. (_cpp_do_file_change): Don't need to set cpp_reader's line field. * cpperror.c (cpp_error): Likewise use highest_line field. * cppfiles.c (open_file_failed: Likewise. (cpp_make_system_header): Likewise use newest map and highest_line. * cppinit.c (cpp_create_reader): Don't initialize removed field. * cpplex.c (_cpp_process_line_notes, _cpp_skip_block_comment, skip_line_comment, skip_whitespace, _cpp_get_fresh_line, _cpp_lex_direct): Likewise use highest_line. (_cpp_lex_direct): Use new LINEMAP_POSITION_FOR_COLUMN macro. * cppmacro.c (_cpp_builtin_macro_text): Likewise use highest_line, and use newest map. * cpppch.c (cpp_read_state): Don't save+restore cpp_reader's line. * cpptrad.c (_cpp_overlay_buffer): Don't save cpp_reader's line. (copy_comment, _cpp_scan_out_logical_line): Likewise use highest_line. From-SVN: r81074
2004-04-22 19:22:27 -07:00
int sysp = pfile->line_table->highest_line > 1 && pfile->buffer ? pfile->buffer->sysp : 0;
bool print_dep = CPP_OPTION (pfile, deps.style) > (angle_brackets || !!sysp);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
errno = file->err_no;
if (print_dep && CPP_OPTION (pfile, deps.missing_files) && errno == ENOENT)
{
deps_add_dep (pfile->deps, file->name);
/* If the preprocessor output (other than dependency information) is
being used, we must also flag an error. */
if (CPP_OPTION (pfile, deps.need_preprocessor_output))
cpp_errno_filename (pfile, CPP_DL_FATAL,
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
file->path ? file->path : file->name,
loc);
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
else
[multiple changes] 1998-10-28 16:10 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * c-lang.c: Declare extern char *yy_cur if USE_CPPLIB. (lang_init): Call check_newline always. * c-lex.c (init_parse) [USE_CPPLIB=1]: After calling cpp_start_read, set yy_cur and yy_lim to read from parse_in.token_buffer, so that we'll see the first #line directive. * cpplib.c (cpp_start_read): finclude the main input file before processing -include/-imacros. Process -imacros and -include separately, and handle -include by stacking a buffer for the file in question as if it'd been #included. * toplev.c (documented_lang_options) Recognize -H when USE_CPPLIB is on. 1998-10-28 16:09 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * cpplib.c: Merge do_once into do_pragma. Break file handling code out of do_include. Move append_include_chain, deps_output, file_cleanup, redundant_include_p, import_hash, lookup_import, add_import, read_filename_string, read_name_map, open_include_file, finclude, safe_read to cppfiles.c. Move prototypes for deps_output, append_include_chain, finclude to cpplib.h. Move definition of struct file_name_list there also. * cppfiles.c: New file. Contains all the above functions broken out of cpplib.c; also hack_vms_include_specification from cccp.c and find_include_file, a new function broken out of do_include. * Makefile.in (cppmain): Depend on cppfiles.o. (fix-header): Likewise. (cppfiles.o): New target. * configure.in (--enable-c-cpplib): Add cppfiles.o to extra_c_objs. Add ../cppfiles.o to extra_cxx_objs. From-SVN: r23424
1998-10-29 11:54:13 +00:00
{
/* If we are not outputting dependencies, or if we are and dependencies
were requested for this file, or if preprocessor output is needed
in addition to dependency information, this is an error.
Otherwise (outputting dependencies but not for this file, and not
using the preprocessor output), we can still produce correct output
so it's only a warning. */
if (CPP_OPTION (pfile, deps.style) == DEPS_NONE
|| print_dep
|| CPP_OPTION (pfile, deps.need_preprocessor_output))
cpp_errno_filename (pfile, CPP_DL_FATAL,
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
file->path ? file->path : file->name,
loc);
else
cpp_errno_filename (pfile, CPP_DL_WARNING,
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
file->path ? file->path : file->name,
loc);
[multiple changes] 1998-10-28 16:10 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * c-lang.c: Declare extern char *yy_cur if USE_CPPLIB. (lang_init): Call check_newline always. * c-lex.c (init_parse) [USE_CPPLIB=1]: After calling cpp_start_read, set yy_cur and yy_lim to read from parse_in.token_buffer, so that we'll see the first #line directive. * cpplib.c (cpp_start_read): finclude the main input file before processing -include/-imacros. Process -imacros and -include separately, and handle -include by stacking a buffer for the file in question as if it'd been #included. * toplev.c (documented_lang_options) Recognize -H when USE_CPPLIB is on. 1998-10-28 16:09 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * cpplib.c: Merge do_once into do_pragma. Break file handling code out of do_include. Move append_include_chain, deps_output, file_cleanup, redundant_include_p, import_hash, lookup_import, add_import, read_filename_string, read_name_map, open_include_file, finclude, safe_read to cppfiles.c. Move prototypes for deps_output, append_include_chain, finclude to cpplib.h. Move definition of struct file_name_list there also. * cppfiles.c: New file. Contains all the above functions broken out of cpplib.c; also hack_vms_include_specification from cccp.c and find_include_file, a new function broken out of do_include. * Makefile.in (cppmain): Depend on cppfiles.o. (fix-header): Likewise. (cppfiles.o): New target. * configure.in (--enable-c-cpplib): Add cppfiles.o to extra_c_objs. Add ../cppfiles.o to extra_cxx_objs. From-SVN: r23424
1998-10-29 11:54:13 +00:00
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Search in the chain beginning at HEAD for a file whose search path
started at START_DIR != NULL. */
static struct cpp_file_hash_entry *
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
search_cache (struct cpp_file_hash_entry *head, const cpp_dir *start_dir,
bool is_embed)
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
{
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
while (head && (head->start_dir != start_dir
|| head->u.file->embed != is_embed))
head = head->next;
return head;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
}
/* Allocate a new _cpp_file structure. */
static _cpp_file *
make_cpp_file (cpp_dir *dir, const char *fname)
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
{
_cpp_file *file = XCNEW (_cpp_file);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
file->fd = -1;
file->dir = dir;
file->name = xstrdup (fname);
return file;
}
/* Release a _cpp_file structure. */
static void
destroy_cpp_file (_cpp_file *file)
{
free ((void *) file->buffer_start);
free ((void *) file->name);
input.c/libcpp: fix lifetimes of path buffers Running "make selftest-valgrind" showed various leaks of the form: 408 bytes in 24 blocks are definitely lost in loss record 572 of 679 at 0x4A0645D: malloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so) by 0x1B0D057: xmalloc (xmalloc.c:148) by 0x1ACCAA1: append_file_to_dir(char const*, cpp_dir*) [clone .isra.3] (files.c:1567) by 0x1ACD56F: _cpp_find_file (files.c:390) by 0x1ACF8FB: cpp_read_main_file(cpp_reader*, char const*) (init.c:632) by 0x1AB3D97: selftest::lexer_test::lexer_test(selftest::line_table_case const&, char const*, selftest::lexer_test_options*) (input.c:2014) by 0x1AB792B: selftest::test_lexer_string_locations_u8(selftest::line_table_case const&) (input.c:2713) by 0x1ABA22A: selftest::for_each_line_table_case(void (*)(selftest::line_table_case const&)) (input.c:3227) by 0x1ABA381: selftest::input_c_tests() (input.c:3260) by 0x1A295F1: selftest::run_tests() (selftest-run-tests.c:62) by 0xF20DC4: toplev::run_self_tests() (toplev.c:2076) by 0xF20FCD: toplev::main(int, char**) (toplev.c:2153) Fix the leak by freeing the file->path in destroy_cpp_file. However, doing so would lead to a use-after-free in input.c's file cache since the filenames in this cache are the libcpp file->path buffers. Hence we need to ensure that any references to the file in the input.c cache are purged before cleaning up file->path. This is normally done by the temp_source_file dtor. Hence we need to reorder things to that the temp_source_file dtor runs before cleaning up the cpp_parser. The patch does this by introducing a wrapper class around cpp_parser *, so that the dtor can run after the dtor for temp_source_file. gcc/ChangeLog: * input.c (fcache::file_patch): Add comment about lifetime. (selftest::cpp_reader_ptr): New class. (selftest::lexer_test): Convert m_parser from cpp_reader * to a cpp_reader_ptr, and move m_tempfile to after it. (selftest::lexer_test::lexer_test): Update for above reordering. (lexer_test::~lexer_test): Move cleanup of m_parser to cpp_reader_ptr's dtor. libcpp/ChangeLog: * files.c (destroy_cpp_file): Free file->path. From-SVN: r241536
2016-10-25 19:24:01 +00:00
free ((void *) file->path);
free (file);
}
/* Release all the files allocated by this reader. */
static void
destroy_all_cpp_files (cpp_reader *pfile)
{
_cpp_file *iter = pfile->all_files;
while (iter)
{
_cpp_file *next = iter->next_file;
destroy_cpp_file (iter);
iter = next;
}
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* A hash of directory names. The directory names are the path names
of files which contain a #include "", the included file name is
appended to this directories.
To avoid duplicate entries we follow the convention that all
non-empty directory names should end in a '/'. DIR_NAME must be
stored in permanently allocated memory. */
static cpp_dir *
make_cpp_dir (cpp_reader *pfile, const char *dir_name, int sysp)
{
struct cpp_file_hash_entry *entry, **hash_slot;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
cpp_dir *dir;
hash_slot = (struct cpp_file_hash_entry **)
htab_find_slot_with_hash (pfile->dir_hash, dir_name,
htab_hash_string (dir_name),
INSERT);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Have we already hashed this directory? */
for (entry = *hash_slot; entry; entry = entry->next)
if (entry->start_dir == NULL)
return entry->u.dir;
configure.ac: Check declarations for asprintf and vasprintf. * configure.ac: Check declarations for asprintf and vasprintf. * config.in: Regenerate. * configure: Likewise. * charset.c (conversion_loop): Use XRESIZEVEC. (convert_no_conversion): Likewise. (convert_using_iconv): Likewise. (init_iconv_desc): Cast return value of alloca. (cpp_host_to_exec_charset): Use XNEWVEC. (emit_numeric_escape): Use XRESIZEVEC. (cpp_interpret_string): Use XNEWVEC. (cpp_interpret_string): Use XRESIZEVEC. (_cpp_interpret_identifier): Cast return value of alloca. (_cpp_convert_input): Use XNEWVEC and XRESIZEVEC. * directives.c (glue_header_name): Use XNEWVEC and XRESIZEVEC. (parse_include): Use XNEWVEC. (insert_pragma_entry): Rename local variable "new" to "new_entry". (save_registered_pragmas): Cast return value of xmemdup. (destringize_and_run): Same for alloca. (parse_assertion): Likewise. (do_assert): Cast allocated storage to proper type. (cpp_define): Likewise. (_cpp_define_builtin): Likewise. (cpp_undef): Likewise. (handle_assertion): Likewise. (cpp_push_buffer): Rename local variable "new" to "new_buffer". * expr.c (CPP_UPLUS): Cast value to type cpp_ttype. (CPP_UMINUS): Likewise. (struct cpp_operator): Rename from struct operator. (_cpp_expand_op_stack): Use XRESIZEVEC. * files.c (pch_open_file): Use XNEWVEC. (pch_open_file): Use XRESIZEVEC. (read_file_guts): Use XNEWVEC and XRESIZEVEC. (dir_name_of_file): Use XNEWVEC. (make_cpp_file): Use XCNEW. (make_cpp_dir): Likewise. (allocate_file_hash_entries): USE XNEWVEC. (cpp_included): Cast return value of htab_find_with_hash. (append_file_to_dir): Use XNEWVEC. (read_filename_string): Likewise. Use XRESIZEVEC too. (read_name_map): Cast return value of alloca. Use XRESIZEVEC. (remap_filename): Use XNEWVEC. (struct pchf_entry): Move definition out of struct pchf_data. (_cpp_save_file_entries): Use XCNEWVAR. (_cpp_read_file_entries): Use XNEWVAR. * identifiers.c (alloc_node): Use XOBNEW. * init.c (cpp_create_reader): Use XCNEW. (cpp_init_builtins): Cast of b->value to enum builtin_type. (read_original_directory): Cast return value of alloca. * lex.c (add_line_note): Use XRESIZEVEC. (warn_about_normalization): Use XNEWVEC. (_cpp_lex_direct): Cast node->directive_index to (enum cpp_ttype). (new_buff): Use XNEWVEC. * line-map.c (linemap_add): Use XRESIZEVEC. * macro.c (builtin_macro): Cast return value of alloca. (paste_tokens): Likewise. (expand_arg): Use XNEWVEC and XRESIZEVEC. (_cpp_save_parameter): Use XRESIZEVEC. (create_iso_definition): Cast allocated storage to proper type. (_cpp_create_definition): Likewise. (cpp_macro_definition): Use XRESIZEVEC. * makedepend.c (add_clm): Use XNEW. (add_dir): Likewise. * mkdeps.c (munge): Use XNEWVEC. (deps_init): Use XCNEW. (deps_add_target): Use XRESIZEVEC. (deps_add_default_target): Cast return value of alloca. (deps_add_dep): Use XRESIZEVEC. (deps_add_vpath): Likewise. Use XNEWVEC too. (deps_restore): Likewise. * pch.c (save_idents): Use XNEW and XNEWVEC. (cpp_save_state): Use XNEW. (count_defs): Cast return value of htab_find. (write_defs): Likewise. (cpp_write_pch_deps): Use XNEWVEC. (collect_ht_nodes): Use XRESIZEVEC. (cpp_valid_state): Use XNEWVEC. (save_macros): Use XRESIZEVEC. Cast return value of xmemdup. * symtab.c (ht_create): Use XCNEW. (ht_lookup_with_hash): Cast return value of obstack_copy0. (ht_expand): Use XCNEWVEC. * system.h (HAVE_DESIGNATED_INITIALIZERS): False if __cplusplus. (bool): Do not define if __cplusplus. From-SVN: r100295
2005-05-28 15:52:48 +00:00
dir = XCNEW (cpp_dir);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
dir->next = pfile->quote_include;
dir->name = (char *) dir_name;
dir->len = strlen (dir_name);
dir->sysp = sysp;
Add framework support for darwin. * c-incpath.c: Include target.h and machmode.h. (add_path): Use a consistent style for cpp_dir. Initialize p->construct to 0. (add_cpp_dir_path): New. (register_include_chains): Add use of extra_includes callback. (hook_void_int): Add. (target_c_incpath): Add. * c-incpath.h (add_cpp_dir_path): New. (target_c_incpath_s): Add. (target_c_incpath): Add. (C_INCPATH_INIT): Add. * c-opts.c (c_common_missing_argument, c_common_handle_option): Add -F argument processing. * c.opt: Add -F argument processing. * gcc.c (trad_capable_cpp): Add -F argument processing. * cppfiles.c (find_file_in_dir): Update to use construct callback. (search_path_exhausted, cpp_get_path, cpp_get_buffer, cpp_get_prev): New. (_cpp_find_file): Use search_path_exhausted. (make_cpp_dir): Initialize construct to 0. * cpplib.h (missing_header_cb cpp_get_path, cpp_get_buffer, cpp_get_file, cpp_get_prev): New. (cpp_callbacks): Add missing_header (cpp_dir): Add construct. * target-def.h: (TARGET_OPTF): New. * hooks.c (hook_void_int, hook_void_charptr): Add. * hooks.h (hook_void_int, hook_void_charptr): Add. * Makefile.in (c-incpath.o) : Add $(TARGET_H) and $(MACHMODE_H) dependencies. * doc/invoke.texi (Darwin Options): Document -F. * doc/tm.texi (TARGET_EXTRA_INCLUDES): Add. (TARGET_OPTF): Add. * fix-header.c (target_c_incpath): Add. * config/darwin-c.c: Add c-incpath.h include. (using_frameworks, find_subframework_file, find_subframework_header, add_system_framework_path, frameworks_in_use, num_frameworks, max_frameworks, add_framework, find_framework, struct framework_header, framework_header_dirs, framework_construct_pathname, find_subframework_file, add_system_framework_path, add_framework_path, framework_defaults, darwin_register_frameworks, find_subframework_header): Add. * config/darwin.h (TARGET_EXTRA_INCLUDES, TARGET_OPTF): New. (TARGET_OPTION_TRANSLATE_TABLE): Add -framework support. (CPP_SPEC): Add __APPLE_CC__ support. * t-darwin (darwin-c.o): Add c-incpath.h dependency. From-SVN: r78875
2004-03-04 00:18:54 +00:00
dir->construct = 0;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Store this new result in the hash table. */
entry = new_file_hash_entry (pfile);
entry->next = *hash_slot;
entry->start_dir = NULL;
entry->location = pfile->line_table->highest_location;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
entry->u.dir = dir;
*hash_slot = entry;
return dir;
}
/* Create a new block of memory for file hash entries. */
static void
allocate_file_hash_entries (cpp_reader *pfile)
{
struct file_hash_entry_pool *pool = XNEW (struct file_hash_entry_pool);
pool->file_hash_entries_used = 0;
pool->next = pfile->file_hash_entries;
pfile->file_hash_entries = pool;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
}
/* Return a new file hash entry. */
static struct cpp_file_hash_entry *
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
new_file_hash_entry (cpp_reader *pfile)
{
unsigned int idx;
if (pfile->file_hash_entries->file_hash_entries_used == FILE_HASH_POOL_SIZE)
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
allocate_file_hash_entries (pfile);
idx = pfile->file_hash_entries->file_hash_entries_used++;
return &pfile->file_hash_entries->pool[idx];
}
/* Free the file hash entry pools. */
static void
free_file_hash_entries (cpp_reader *pfile)
{
struct file_hash_entry_pool *iter = pfile->file_hash_entries;
while (iter)
{
struct file_hash_entry_pool *next = iter->next;
free (iter);
iter = next;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
}
/* Returns TRUE if a file FNAME has ever been successfully opened.
This routine is not intended to correctly handle filenames aliased
by links or redundant . or .. traversals etc. */
bool
cpp_included (cpp_reader *pfile, const char *fname)
{
struct cpp_file_hash_entry *entry;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
entry = (struct cpp_file_hash_entry *)
configure.ac: Check declarations for asprintf and vasprintf. * configure.ac: Check declarations for asprintf and vasprintf. * config.in: Regenerate. * configure: Likewise. * charset.c (conversion_loop): Use XRESIZEVEC. (convert_no_conversion): Likewise. (convert_using_iconv): Likewise. (init_iconv_desc): Cast return value of alloca. (cpp_host_to_exec_charset): Use XNEWVEC. (emit_numeric_escape): Use XRESIZEVEC. (cpp_interpret_string): Use XNEWVEC. (cpp_interpret_string): Use XRESIZEVEC. (_cpp_interpret_identifier): Cast return value of alloca. (_cpp_convert_input): Use XNEWVEC and XRESIZEVEC. * directives.c (glue_header_name): Use XNEWVEC and XRESIZEVEC. (parse_include): Use XNEWVEC. (insert_pragma_entry): Rename local variable "new" to "new_entry". (save_registered_pragmas): Cast return value of xmemdup. (destringize_and_run): Same for alloca. (parse_assertion): Likewise. (do_assert): Cast allocated storage to proper type. (cpp_define): Likewise. (_cpp_define_builtin): Likewise. (cpp_undef): Likewise. (handle_assertion): Likewise. (cpp_push_buffer): Rename local variable "new" to "new_buffer". * expr.c (CPP_UPLUS): Cast value to type cpp_ttype. (CPP_UMINUS): Likewise. (struct cpp_operator): Rename from struct operator. (_cpp_expand_op_stack): Use XRESIZEVEC. * files.c (pch_open_file): Use XNEWVEC. (pch_open_file): Use XRESIZEVEC. (read_file_guts): Use XNEWVEC and XRESIZEVEC. (dir_name_of_file): Use XNEWVEC. (make_cpp_file): Use XCNEW. (make_cpp_dir): Likewise. (allocate_file_hash_entries): USE XNEWVEC. (cpp_included): Cast return value of htab_find_with_hash. (append_file_to_dir): Use XNEWVEC. (read_filename_string): Likewise. Use XRESIZEVEC too. (read_name_map): Cast return value of alloca. Use XRESIZEVEC. (remap_filename): Use XNEWVEC. (struct pchf_entry): Move definition out of struct pchf_data. (_cpp_save_file_entries): Use XCNEWVAR. (_cpp_read_file_entries): Use XNEWVAR. * identifiers.c (alloc_node): Use XOBNEW. * init.c (cpp_create_reader): Use XCNEW. (cpp_init_builtins): Cast of b->value to enum builtin_type. (read_original_directory): Cast return value of alloca. * lex.c (add_line_note): Use XRESIZEVEC. (warn_about_normalization): Use XNEWVEC. (_cpp_lex_direct): Cast node->directive_index to (enum cpp_ttype). (new_buff): Use XNEWVEC. * line-map.c (linemap_add): Use XRESIZEVEC. * macro.c (builtin_macro): Cast return value of alloca. (paste_tokens): Likewise. (expand_arg): Use XNEWVEC and XRESIZEVEC. (_cpp_save_parameter): Use XRESIZEVEC. (create_iso_definition): Cast allocated storage to proper type. (_cpp_create_definition): Likewise. (cpp_macro_definition): Use XRESIZEVEC. * makedepend.c (add_clm): Use XNEW. (add_dir): Likewise. * mkdeps.c (munge): Use XNEWVEC. (deps_init): Use XCNEW. (deps_add_target): Use XRESIZEVEC. (deps_add_default_target): Cast return value of alloca. (deps_add_dep): Use XRESIZEVEC. (deps_add_vpath): Likewise. Use XNEWVEC too. (deps_restore): Likewise. * pch.c (save_idents): Use XNEW and XNEWVEC. (cpp_save_state): Use XNEW. (count_defs): Cast return value of htab_find. (write_defs): Likewise. (cpp_write_pch_deps): Use XNEWVEC. (collect_ht_nodes): Use XRESIZEVEC. (cpp_valid_state): Use XNEWVEC. (save_macros): Use XRESIZEVEC. Cast return value of xmemdup. * symtab.c (ht_create): Use XCNEW. (ht_lookup_with_hash): Cast return value of obstack_copy0. (ht_expand): Use XCNEWVEC. * system.h (HAVE_DESIGNATED_INITIALIZERS): False if __cplusplus. (bool): Do not define if __cplusplus. From-SVN: r100295
2005-05-28 15:52:48 +00:00
htab_find_with_hash (pfile->file_hash, fname, htab_hash_string (fname));
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
while (entry && (entry->start_dir == NULL || entry->u.file->err_no))
entry = entry->next;
return entry != NULL;
}
/* Returns TRUE if a file FNAME has ever been successfully opened
before LOCATION. This routine is not intended to correctly handle
filenames aliased by links or redundant . or .. traversals etc. */
bool
cpp_included_before (cpp_reader *pfile, const char *fname,
Eliminate source_location in favor of location_t Historically GCC used location_t, while libcpp used source_location. This inconsistency has been annoying me for a while, so this patch removes source_location in favor of location_t throughout (as the latter is shorter). gcc/ChangeLog: * builtins.c: Replace "source_location" with "location_t". * diagnostic-show-locus.c: Likewise. * diagnostic.c: Likewise. * dumpfile.c: Likewise. * gcc-rich-location.h: Likewise. * genmatch.c: Likewise. * gimple.h: Likewise. * gimplify.c: Likewise. * input.c: Likewise. * input.h: Likewise. Eliminate the typedef. * omp-expand.c: Likewise. * selftest.h: Likewise. * substring-locations.h (get_source_location_for_substring): Rename to.. (get_location_within_string): ...this. * tree-cfg.c: Replace "source_location" with "location_t". * tree-cfgcleanup.c: Likewise. * tree-diagnostic.c: Likewise. * tree-into-ssa.c: Likewise. * tree-outof-ssa.c: Likewise. * tree-parloops.c: Likewise. * tree-phinodes.c: Likewise. * tree-phinodes.h: Likewise. * tree-ssa-loop-ivopts.c: Likewise. * tree-ssa-loop-manip.c: Likewise. * tree-ssa-phiopt.c: Likewise. * tree-ssa-phiprop.c: Likewise. * tree-ssa-threadupdate.c: Likewise. * tree-ssa.c: Likewise. * tree-ssa.h: Likewise. * tree-vect-loop-manip.c: Likewise. gcc/c-family/ChangeLog: * c-common.c (c_get_substring_location): Update for renaming of get_source_location_for_substring to get_location_within_string. * c-lex.c: Replace "source_location" with "location_t". * c-opts.c: Likewise. * c-ppoutput.c: Likewise. gcc/c/ChangeLog: * c-decl.c: Replace "source_location" with "location_t". * c-tree.h: Likewise. * c-typeck.c: Likewise. * gimple-parser.c: Likewise. gcc/cp/ChangeLog: * call.c: Replace "source_location" with "location_t". * cp-tree.h: Likewise. * cvt.c: Likewise. * name-lookup.c: Likewise. * parser.c: Likewise. * typeck.c: Likewise. gcc/fortran/ChangeLog: * cpp.c: Replace "source_location" with "location_t". * gfortran.h: Likewise. gcc/go/ChangeLog: * go-gcc-diagnostics.cc: Replace "source_location" with "location_t". * go-gcc.cc: Likewise. * go-linemap.cc: Likewise. * go-location.h: Likewise. * gofrontend/README: Likewise. gcc/jit/ChangeLog: * jit-playback.c: Replace "source_location" with "location_t". gcc/testsuite/ChangeLog: * g++.dg/plugin/comment_plugin.c: Replace "source_location" with "location_t". * gcc.dg/plugin/diagnostic_plugin_test_show_locus.c: Likewise. libcc1/ChangeLog: * libcc1plugin.cc: Replace "source_location" with "location_t". (plugin_context::get_source_location): Rename to... (plugin_context::get_location_t): ...this. * libcp1plugin.cc: Likewise. libcpp/ChangeLog: * charset.c: Replace "source_location" with "location_t". * directives-only.c: Likewise. * directives.c: Likewise. * errors.c: Likewise. * expr.c: Likewise. * files.c: Likewise. * include/cpplib.h: Likewise. Rename MAX_SOURCE_LOCATION to MAX_LOCATION_T. * include/line-map.h: Likewise. * init.c: Likewise. * internal.h: Likewise. * lex.c: Likewise. * line-map.c: Likewise. * location-example.txt: Likewise. * macro.c: Likewise. * pch.c: Likewise. * traditional.c: Likewise. From-SVN: r266085
2018-11-13 20:05:03 +00:00
location_t location)
{
struct cpp_file_hash_entry *entry
= (struct cpp_file_hash_entry *)
htab_find_with_hash (pfile->file_hash, fname, htab_hash_string (fname));
if (IS_ADHOC_LOC (location))
location = get_location_from_adhoc_loc (pfile->line_table, location);
while (entry && (entry->start_dir == NULL || entry->u.file->err_no
|| entry->location > location))
entry = entry->next;
return entry != NULL;
}
/* Calculate the hash value of a file hash entry P. */
static hashval_t
file_hash_hash (const void *p)
{
struct cpp_file_hash_entry *entry = (struct cpp_file_hash_entry *) p;
const char *hname;
if (entry->start_dir)
hname = entry->u.file->name;
else
hname = entry->u.dir->name;
return htab_hash_string (hname);
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Compare a string Q against a file hash entry P. */
static int
file_hash_eq (const void *p, const void *q)
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
{
struct cpp_file_hash_entry *entry = (struct cpp_file_hash_entry *) p;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
const char *fname = (const char *) q;
const char *hname;
if (entry->start_dir)
hname = entry->u.file->name;
else
hname = entry->u.dir->name;
return filename_cmp (hname, fname) == 0;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
}
/* Compare entries in the nonexistent file hash table. These are just
strings. */
static int
nonexistent_file_hash_eq (const void *p, const void *q)
{
return filename_cmp ((const char *) p, (const char *) q) == 0;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Initialize everything in this source file. */
void
_cpp_init_files (cpp_reader *pfile)
{
pfile->file_hash = htab_create_alloc (127, file_hash_hash, file_hash_eq,
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
NULL, xcalloc, free);
pfile->dir_hash = htab_create_alloc (127, file_hash_hash, file_hash_eq,
NULL, xcalloc, free);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
allocate_file_hash_entries (pfile);
pfile->nonexistent_file_hash = htab_create_alloc (127, htab_hash_string,
nonexistent_file_hash_eq,
NULL, xcalloc, free);
obstack_specify_allocation (&pfile->nonexistent_file_ob, 0, 0,
xmalloc, free);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
}
/* Finalize everything in this source file. */
void
_cpp_cleanup_files (cpp_reader *pfile)
{
htab_delete (pfile->file_hash);
htab_delete (pfile->dir_hash);
htab_delete (pfile->nonexistent_file_hash);
obstack_free (&pfile->nonexistent_file_ob, 0);
free_file_hash_entries (pfile);
destroy_all_cpp_files (pfile);
}
/* Make the parser forget about files it has seen. This can be useful
for resetting the parser to start another run. */
void
cpp_clear_file_cache (cpp_reader *pfile)
{
_cpp_cleanup_files (pfile);
pfile->file_hash_entries = NULL;
pfile->all_files = NULL;
_cpp_init_files (pfile);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
}
/* Enter a file name in the hash for the sake of cpp_included. */
void
_cpp_fake_include (cpp_reader *pfile, const char *fname)
{
/* It does not matter what are the contents of fake_source_dir, it will never
be inspected; we just use its address to uniquely signify that this file
was added as a fake include, so a later call to _cpp_find_file (to include
the file for real) won't find the fake one in the hash table. */
static cpp_dir fake_source_dir;
_cpp_find_file (pfile, fname, &fake_source_dir, 0, _cpp_FFK_FAKE, 0);
[multiple changes] 1998-10-28 16:10 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * c-lang.c: Declare extern char *yy_cur if USE_CPPLIB. (lang_init): Call check_newline always. * c-lex.c (init_parse) [USE_CPPLIB=1]: After calling cpp_start_read, set yy_cur and yy_lim to read from parse_in.token_buffer, so that we'll see the first #line directive. * cpplib.c (cpp_start_read): finclude the main input file before processing -include/-imacros. Process -imacros and -include separately, and handle -include by stacking a buffer for the file in question as if it'd been #included. * toplev.c (documented_lang_options) Recognize -H when USE_CPPLIB is on. 1998-10-28 16:09 -0500 Zack Weinberg <zack@rabi.phys.columbia.edu> * cpplib.c: Merge do_once into do_pragma. Break file handling code out of do_include. Move append_include_chain, deps_output, file_cleanup, redundant_include_p, import_hash, lookup_import, add_import, read_filename_string, read_name_map, open_include_file, finclude, safe_read to cppfiles.c. Move prototypes for deps_output, append_include_chain, finclude to cpplib.h. Move definition of struct file_name_list there also. * cppfiles.c: New file. Contains all the above functions broken out of cpplib.c; also hack_vms_include_specification from cccp.c and find_include_file, a new function broken out of do_include. * Makefile.in (cppmain): Depend on cppfiles.o. (fix-header): Likewise. (cppfiles.o): New target. * configure.in (--enable-c-cpplib): Add cppfiles.o to extra_c_objs. Add ../cppfiles.o to extra_cxx_objs. From-SVN: r23424
1998-10-29 11:54:13 +00:00
}
/* Not everyone who wants to set system-header-ness on a buffer can
see the details of a buffer. This is an exported interface because
fix-header needs it. */
void
cpp_make_system_header (cpp_reader *pfile, int syshdr, int externc)
{
int flags = 0;
PR c++/61339 - add mismatch between struct and class [-Wmismatched-tags] to non-bugs gcc/c/ChangeLog: PR c++/61339 * c-decl.c (xref_tag): Change class-key of PODs to struct and others to class. (field_decl_cmp): Same. * c-parser.c (c_parser_struct_or_union_specifier): Same. * c-tree.h: Same. * gimple-parser.c (c_parser_gimple_compound_statement): Same. gcc/c-family/ChangeLog: PR c++/61339 * c-opts.c (handle_deferred_opts): : Change class-key of PODs to struct and others to class. * c-pretty-print.h: Same. gcc/cp/ChangeLog: PR c++/61339 * cp-tree.h: Change class-key of PODs to struct and others to class. * search.c: Same. * semantics.c (finalize_nrv_r): Same. gcc/lto/ChangeLog: PR c++/61339 * lto-common.c (lto_splay_tree_new): : Change class-key of PODs to struct and others to class. (mentions_vars_p): Same. (register_resolution): Same. (lto_register_var_decl_in_symtab): Same. (lto_register_function_decl_in_symtab): Same. (cmp_tree): Same. (lto_read_decls): Same. gcc/ChangeLog: PR c++/61339 * auto-profile.c: Change class-key of PODs to struct and others to class. * basic-block.h: Same. * bitmap.c (bitmap_alloc): Same. * bitmap.h: Same. * builtins.c (expand_builtin_prefetch): Same. (expand_builtin_interclass_mathfn): Same. (expand_builtin_strlen): Same. (expand_builtin_mempcpy_args): Same. (expand_cmpstr): Same. (expand_builtin___clear_cache): Same. (expand_ifn_atomic_bit_test_and): Same. (expand_builtin_thread_pointer): Same. (expand_builtin_set_thread_pointer): Same. * caller-save.c (setup_save_areas): Same. (replace_reg_with_saved_mem): Same. (insert_restore): Same. (insert_save): Same. (add_used_regs): Same. * cfg.c (get_bb_copy): Same. (set_loop_copy): Same. * cfg.h: Same. * cfganal.h: Same. * cfgexpand.c (alloc_stack_frame_space): Same. (add_stack_var): Same. (add_stack_var_conflict): Same. (add_scope_conflicts_1): Same. (update_alias_info_with_stack_vars): Same. (expand_used_vars): Same. * cfghooks.c (redirect_edge_and_branch_force): Same. (delete_basic_block): Same. (split_edge): Same. (make_forwarder_block): Same. (force_nonfallthru): Same. (duplicate_block): Same. (lv_flush_pending_stmts): Same. * cfghooks.h: Same. * cfgloop.c (flow_loops_cfg_dump): Same. (flow_loop_nested_p): Same. (superloop_at_depth): Same. (get_loop_latch_edges): Same. (flow_loop_dump): Same. (flow_loops_dump): Same. (flow_loops_free): Same. (flow_loop_nodes_find): Same. (establish_preds): Same. (flow_loop_tree_node_add): Same. (flow_loop_tree_node_remove): Same. (flow_loops_find): Same. (find_subloop_latch_edge_by_profile): Same. (find_subloop_latch_edge_by_ivs): Same. (mfb_redirect_edges_in_set): Same. (form_subloop): Same. (merge_latch_edges): Same. (disambiguate_multiple_latches): Same. (disambiguate_loops_with_multiple_latches): Same. (flow_bb_inside_loop_p): Same. (glb_enum_p): Same. (get_loop_body_with_size): Same. (get_loop_body): Same. (fill_sons_in_loop): Same. (get_loop_body_in_dom_order): Same. (get_loop_body_in_custom_order): Same. (release_recorded_exits): Same. (get_loop_exit_edges): Same. (num_loop_branches): Same. (remove_bb_from_loops): Same. (find_common_loop): Same. (delete_loop): Same. (cancel_loop): Same. (verify_loop_structure): Same. (loop_preheader_edge): Same. (loop_exit_edge_p): Same. (single_exit): Same. (loop_exits_to_bb_p): Same. (loop_exits_from_bb_p): Same. (get_loop_location): Same. (record_niter_bound): Same. (get_estimated_loop_iterations_int): Same. (max_stmt_executions_int): Same. (likely_max_stmt_executions_int): Same. (get_estimated_loop_iterations): Same. (get_max_loop_iterations): Same. (get_max_loop_iterations_int): Same. (get_likely_max_loop_iterations): Same. * cfgloop.h (simple_loop_desc): Same. (get_loop): Same. (loop_depth): Same. (loop_outer): Same. (loop_iterator::next): Same. (loop_outermost): Same. * cfgloopanal.c (mark_irreducible_loops): Same. (num_loop_insns): Same. (average_num_loop_insns): Same. (expected_loop_iterations_unbounded): Same. (expected_loop_iterations): Same. (mark_loop_exit_edges): Same. (single_likely_exit): Same. * cfgloopmanip.c (fix_bb_placement): Same. (fix_bb_placements): Same. (remove_path): Same. (place_new_loop): Same. (add_loop): Same. (scale_loop_frequencies): Same. (scale_loop_profile): Same. (create_empty_if_region_on_edge): Same. (create_empty_loop_on_edge): Same. (loopify): Same. (unloop): Same. (fix_loop_placements): Same. (copy_loop_info): Same. (duplicate_loop): Same. (duplicate_subloops): Same. (loop_redirect_edge): Same. (can_duplicate_loop_p): Same. (duplicate_loop_to_header_edge): Same. (mfb_keep_just): Same. (has_preds_from_loop): Same. (create_preheader): Same. (create_preheaders): Same. (lv_adjust_loop_entry_edge): Same. (loop_version): Same. * cfgloopmanip.h: Same. * cgraph.h: Same. * cgraphbuild.c: Same. * combine.c (make_extraction): Same. * config/i386/i386-features.c: Same. * config/i386/i386-features.h: Same. * config/i386/i386.c (ix86_emit_outlined_ms2sysv_save): Same. (ix86_emit_outlined_ms2sysv_restore): Same. (ix86_noce_conversion_profitable_p): Same. (ix86_init_cost): Same. (ix86_simd_clone_usable): Same. * configure.ac: Same. * coretypes.h: Same. * data-streamer-in.c (string_for_index): Same. (streamer_read_indexed_string): Same. (streamer_read_string): Same. (bp_unpack_indexed_string): Same. (bp_unpack_string): Same. (streamer_read_uhwi): Same. (streamer_read_hwi): Same. (streamer_read_gcov_count): Same. (streamer_read_wide_int): Same. * data-streamer.h (streamer_write_bitpack): Same. (bp_unpack_value): Same. (streamer_write_char_stream): Same. (streamer_write_hwi_in_range): Same. (streamer_write_record_start): Same. * ddg.c (create_ddg_dep_from_intra_loop_link): Same. (add_cross_iteration_register_deps): Same. (build_intra_loop_deps): Same. * df-core.c (df_analyze): Same. (loop_post_order_compute): Same. (loop_inverted_post_order_compute): Same. * df-problems.c (df_rd_alloc): Same. (df_rd_simulate_one_insn): Same. (df_rd_local_compute): Same. (df_rd_init_solution): Same. (df_rd_confluence_n): Same. (df_rd_transfer_function): Same. (df_rd_free): Same. (df_rd_dump_defs_set): Same. (df_rd_top_dump): Same. (df_lr_alloc): Same. (df_lr_reset): Same. (df_lr_local_compute): Same. (df_lr_init): Same. (df_lr_confluence_n): Same. (df_lr_free): Same. (df_lr_top_dump): Same. (df_lr_verify_transfer_functions): Same. (df_live_alloc): Same. (df_live_reset): Same. (df_live_init): Same. (df_live_confluence_n): Same. (df_live_finalize): Same. (df_live_free): Same. (df_live_top_dump): Same. (df_live_verify_transfer_functions): Same. (df_mir_alloc): Same. (df_mir_reset): Same. (df_mir_init): Same. (df_mir_confluence_n): Same. (df_mir_free): Same. (df_mir_top_dump): Same. (df_word_lr_alloc): Same. (df_word_lr_reset): Same. (df_word_lr_init): Same. (df_word_lr_confluence_n): Same. (df_word_lr_free): Same. (df_word_lr_top_dump): Same. (df_md_alloc): Same. (df_md_simulate_one_insn): Same. (df_md_reset): Same. (df_md_init): Same. (df_md_free): Same. (df_md_top_dump): Same. * df-scan.c (df_insn_delete): Same. (df_insn_rescan): Same. (df_notes_rescan): Same. (df_sort_and_compress_mws): Same. (df_install_mws): Same. (df_refs_add_to_chains): Same. (df_ref_create_structure): Same. (df_ref_record): Same. (df_def_record_1): Same. (df_find_hard_reg_defs): Same. (df_uses_record): Same. (df_get_conditional_uses): Same. (df_get_call_refs): Same. (df_recompute_luids): Same. (df_get_entry_block_def_set): Same. (df_entry_block_defs_collect): Same. (df_get_exit_block_use_set): Same. (df_exit_block_uses_collect): Same. (df_mws_verify): Same. (df_bb_verify): Same. * df.h (df_scan_get_bb_info): Same. * doc/tm.texi: Same. * dse.c (record_store): Same. * dumpfile.h: Same. * emit-rtl.c (const_fixed_hasher::equal): Same. (set_mem_attributes_minus_bitpos): Same. (change_address): Same. (adjust_address_1): Same. (offset_address): Same. * emit-rtl.h: Same. * except.c (dw2_build_landing_pads): Same. (sjlj_emit_dispatch_table): Same. * explow.c (allocate_dynamic_stack_space): Same. (emit_stack_probe): Same. (probe_stack_range): Same. * expmed.c (store_bit_field_using_insv): Same. (store_bit_field_1): Same. (store_integral_bit_field): Same. (extract_bit_field_using_extv): Same. (extract_bit_field_1): Same. (emit_cstore): Same. * expr.c (emit_block_move_via_cpymem): Same. (expand_cmpstrn_or_cmpmem): Same. (set_storage_via_setmem): Same. (emit_single_push_insn_1): Same. (expand_assignment): Same. (store_constructor): Same. (expand_expr_real_2): Same. (expand_expr_real_1): Same. (try_casesi): Same. * flags.h: Same. * function.c (try_fit_stack_local): Same. (assign_stack_local_1): Same. (assign_stack_local): Same. (cut_slot_from_list): Same. (insert_slot_to_list): Same. (max_slot_level): Same. (move_slot_to_level): Same. (temp_address_hasher::equal): Same. (remove_unused_temp_slot_addresses): Same. (assign_temp): Same. (combine_temp_slots): Same. (update_temp_slot_address): Same. (preserve_temp_slots): Same. * function.h: Same. * fwprop.c: Same. * gcc-rich-location.h: Same. * gcov.c: Same. * genattrtab.c (check_attr_test): Same. (check_attr_value): Same. (convert_set_attr_alternative): Same. (convert_set_attr): Same. (check_defs): Same. (copy_boolean): Same. (get_attr_value): Same. (expand_delays): Same. (make_length_attrs): Same. (min_fn): Same. (make_alternative_compare): Same. (simplify_test_exp): Same. (tests_attr_p): Same. (get_attr_order): Same. (clear_struct_flag): Same. (gen_attr): Same. (compares_alternatives_p): Same. (gen_insn): Same. (gen_delay): Same. (find_attrs_to_cache): Same. (write_test_expr): Same. (walk_attr_value): Same. (write_attr_get): Same. (eliminate_known_true): Same. (write_insn_cases): Same. (write_attr_case): Same. (write_attr_valueq): Same. (write_attr_value): Same. (write_dummy_eligible_delay): Same. (next_comma_elt): Same. (find_attr): Same. (make_internal_attr): Same. (copy_rtx_unchanging): Same. (gen_insn_reserv): Same. (check_tune_attr): Same. (make_automaton_attrs): Same. (handle_arg): Same. * genextract.c (gen_insn): Same. (VEC_char_to_string): Same. * genmatch.c (print_operand): Same. (lower): Same. (parser::parse_operation): Same. (parser::parse_capture): Same. (parser::parse_c_expr): Same. (parser::parse_simplify): Same. (main): Same. * genoutput.c (output_operand_data): Same. (output_get_insn_name): Same. (compare_operands): Same. (place_operands): Same. (process_template): Same. (validate_insn_alternatives): Same. (validate_insn_operands): Same. (gen_expand): Same. (note_constraint): Same. * genpreds.c (write_one_predicate_function): Same. (add_constraint): Same. (process_define_register_constraint): Same. (write_lookup_constraint_1): Same. (write_lookup_constraint_array): Same. (write_insn_constraint_len): Same. (write_reg_class_for_constraint_1): Same. (write_constraint_satisfied_p_array): Same. * genrecog.c (optimize_subroutine_group): Same. * gensupport.c (process_define_predicate): Same. (queue_pattern): Same. (remove_from_queue): Same. (process_rtx): Same. (is_predicable): Same. (change_subst_attribute): Same. (subst_pattern_match): Same. (alter_constraints): Same. (alter_attrs_for_insn): Same. (shift_output_template): Same. (alter_output_for_subst_insn): Same. (process_one_cond_exec): Same. (subst_dup): Same. (process_define_cond_exec): Same. (mnemonic_htab_callback): Same. (gen_mnemonic_attr): Same. (read_md_rtx): Same. * ggc-page.c: Same. * gimple-loop-interchange.cc (dump_reduction): Same. (dump_induction): Same. (loop_cand::~loop_cand): Same. (free_data_refs_with_aux): Same. (tree_loop_interchange::interchange_loops): Same. (tree_loop_interchange::map_inductions_to_loop): Same. (tree_loop_interchange::move_code_to_inner_loop): Same. (compute_access_stride): Same. (compute_access_strides): Same. (proper_loop_form_for_interchange): Same. (tree_loop_interchange_compute_ddrs): Same. (prune_datarefs_not_in_loop): Same. (prepare_data_references): Same. (pass_linterchange::execute): Same. * gimple-loop-jam.c (bb_prevents_fusion_p): Same. (unroll_jam_possible_p): Same. (fuse_loops): Same. (adjust_unroll_factor): Same. (tree_loop_unroll_and_jam): Same. * gimple-loop-versioning.cc (loop_versioning::~loop_versioning): Same. (loop_versioning::expensive_stmt_p): Same. (loop_versioning::version_for_unity): Same. (loop_versioning::dump_inner_likelihood): Same. (loop_versioning::find_per_loop_multiplication): Same. (loop_versioning::analyze_term_using_scevs): Same. (loop_versioning::record_address_fragment): Same. (loop_versioning::analyze_expr): Same. (loop_versioning::analyze_blocks): Same. (loop_versioning::prune_conditions): Same. (loop_versioning::merge_loop_info): Same. (loop_versioning::add_loop_to_queue): Same. (loop_versioning::decide_whether_loop_is_versionable): Same. (loop_versioning::make_versioning_decisions): Same. (loop_versioning::implement_versioning_decisions): Same. * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::record_ranges_from_phis): Same. * gimple-ssa-store-merging.c (split_store::split_store): Same. (count_multiple_uses): Same. (split_group): Same. (imm_store_chain_info::output_merged_store): Same. (pass_store_merging::process_store): Same. * gimple-ssa-strength-reduction.c (slsr_process_phi): Same. * gimple-ssa-warn-alloca.c (adjusted_warn_limit): Same. (is_max): Same. (alloca_call_type): Same. (pass_walloca::execute): Same. * gimple-streamer-in.c (input_phi): Same. (input_gimple_stmt): Same. * gimple-streamer.h: Same. * godump.c (go_force_record_alignment): Same. (go_format_type): Same. (go_output_type): Same. (go_output_fndecl): Same. (go_output_typedef): Same. (keyword_hash_init): Same. (find_dummy_types): Same. * graph.c (draw_cfg_nodes_no_loops): Same. (draw_cfg_nodes_for_loop): Same. * hard-reg-set.h (hard_reg_set_iter_next): Same. * hsa-brig.c: Same. * hsa-common.h (hsa_internal_fn_hasher::equal): Same. * hsa-dump.c (dump_hsa_cfun): Same. * hsa-gen.c (gen_function_def_parameters): Same. * hsa-regalloc.c (dump_hsa_cfun_regalloc): Same. * input.c (dump_line_table_statistics): Same. (test_lexer): Same. * input.h: Same. * internal-fn.c (get_multi_vector_move): Same. (expand_load_lanes_optab_fn): Same. (expand_GOMP_SIMT_ENTER_ALLOC): Same. (expand_GOMP_SIMT_EXIT): Same. (expand_GOMP_SIMT_LAST_LANE): Same. (expand_GOMP_SIMT_ORDERED_PRED): Same. (expand_GOMP_SIMT_VOTE_ANY): Same. (expand_GOMP_SIMT_XCHG_BFLY): Same. (expand_GOMP_SIMT_XCHG_IDX): Same. (expand_addsub_overflow): Same. (expand_neg_overflow): Same. (expand_mul_overflow): Same. (expand_call_mem_ref): Same. (expand_mask_load_optab_fn): Same. (expand_scatter_store_optab_fn): Same. (expand_gather_load_optab_fn): Same. * ipa-cp.c (ipa_get_parm_lattices): Same. (print_all_lattices): Same. (ignore_edge_p): Same. (build_toporder_info): Same. (free_toporder_info): Same. (push_node_to_stack): Same. (ipcp_lattice<valtype>::set_contains_variable): Same. (set_agg_lats_to_bottom): Same. (ipcp_bits_lattice::meet_with): Same. (set_single_call_flag): Same. (initialize_node_lattices): Same. (ipa_get_jf_ancestor_result): Same. (ipcp_verify_propagated_values): Same. (propagate_scalar_across_jump_function): Same. (propagate_context_across_jump_function): Same. (propagate_bits_across_jump_function): Same. (ipa_vr_operation_and_type_effects): Same. (propagate_vr_across_jump_function): Same. (set_check_aggs_by_ref): Same. (set_chain_of_aglats_contains_variable): Same. (merge_aggregate_lattices): Same. (agg_pass_through_permissible_p): Same. (propagate_aggs_across_jump_function): Same. (call_passes_through_thunk_p): Same. (propagate_constants_across_call): Same. (devirtualization_time_bonus): Same. (good_cloning_opportunity_p): Same. (context_independent_aggregate_values): Same. (gather_context_independent_values): Same. (perform_estimation_of_a_value): Same. (estimate_local_effects): Same. (value_topo_info<valtype>::add_val): Same. (add_all_node_vals_to_toposort): Same. (value_topo_info<valtype>::propagate_effects): Same. (ipcp_propagate_stage): Same. (ipcp_discover_new_direct_edges): Same. (same_node_or_its_all_contexts_clone_p): Same. (cgraph_edge_brings_value_p): Same. (gather_edges_for_value): Same. (create_specialized_node): Same. (find_more_scalar_values_for_callers_subset): Same. (find_more_contexts_for_caller_subset): Same. (copy_plats_to_inter): Same. (intersect_aggregates_with_edge): Same. (find_aggregate_values_for_callers_subset): Same. (cgraph_edge_brings_all_agg_vals_for_node): Same. (decide_about_value): Same. (decide_whether_version_node): Same. (spread_undeadness): Same. (identify_dead_nodes): Same. (ipcp_store_vr_results): Same. * ipa-devirt.c (final_warning_record::grow_type_warnings): Same. * ipa-fnsummary.c (ipa_fn_summary::account_size_time): Same. (redirect_to_unreachable): Same. (edge_set_predicate): Same. (evaluate_conditions_for_known_args): Same. (evaluate_properties_for_edge): Same. (ipa_fn_summary_t::duplicate): Same. (ipa_call_summary_t::duplicate): Same. (dump_ipa_call_summary): Same. (ipa_dump_fn_summary): Same. (eliminated_by_inlining_prob): Same. (set_cond_stmt_execution_predicate): Same. (set_switch_stmt_execution_predicate): Same. (compute_bb_predicates): Same. (will_be_nonconstant_expr_predicate): Same. (phi_result_unknown_predicate): Same. (analyze_function_body): Same. (compute_fn_summary): Same. (estimate_edge_devirt_benefit): Same. (estimate_edge_size_and_time): Same. (estimate_calls_size_and_time): Same. (estimate_node_size_and_time): Same. (remap_edge_change_prob): Same. (remap_edge_summaries): Same. (ipa_merge_fn_summary_after_inlining): Same. (ipa_fn_summary_generate): Same. (inline_read_section): Same. (ipa_fn_summary_read): Same. (ipa_fn_summary_write): Same. * ipa-fnsummary.h: Same. * ipa-hsa.c (ipa_hsa_read_section): Same. * ipa-icf-gimple.c (func_checker::compare_loops): Same. * ipa-icf.c (sem_function::param_used_p): Same. * ipa-inline-analysis.c (do_estimate_edge_time): Same. * ipa-inline.c (edge_badness): Same. (inline_small_functions): Same. * ipa-polymorphic-call.c (ipa_polymorphic_call_context::stream_out): Same. * ipa-predicate.c (predicate::remap_after_duplication): Same. (predicate::remap_after_inlining): Same. (predicate::stream_out): Same. * ipa-predicate.h: Same. * ipa-profile.c (ipa_profile_read_summary): Same. * ipa-prop.c (ipa_get_param_decl_index_1): Same. (count_formal_params): Same. (ipa_dump_param): Same. (ipa_alloc_node_params): Same. (ipa_print_node_jump_functions_for_edge): Same. (ipa_print_node_jump_functions): Same. (ipa_load_from_parm_agg): Same. (get_ancestor_addr_info): Same. (ipa_compute_jump_functions_for_edge): Same. (ipa_analyze_virtual_call_uses): Same. (ipa_analyze_stmt_uses): Same. (ipa_analyze_params_uses_in_bb): Same. (update_jump_functions_after_inlining): Same. (try_decrement_rdesc_refcount): Same. (ipa_impossible_devirt_target): Same. (update_indirect_edges_after_inlining): Same. (combine_controlled_uses_counters): Same. (ipa_edge_args_sum_t::duplicate): Same. (ipa_write_jump_function): Same. (ipa_write_indirect_edge_info): Same. (ipa_write_node_info): Same. (ipa_read_edge_info): Same. (ipa_prop_read_section): Same. (read_replacements_section): Same. * ipa-prop.h (ipa_get_param_count): Same. (ipa_get_param): Same. (ipa_get_type): Same. (ipa_get_param_move_cost): Same. (ipa_set_param_used): Same. (ipa_get_controlled_uses): Same. (ipa_set_controlled_uses): Same. (ipa_get_cs_argument_count): Same. * ipa-pure-const.c (analyze_function): Same. (pure_const_read_summary): Same. * ipa-ref.h: Same. * ipa-reference.c (ipa_reference_read_optimization_summary): Same. * ipa-split.c (test_nonssa_use): Same. (dump_split_point): Same. (dominated_by_forbidden): Same. (split_part_set_ssa_name_p): Same. (find_split_points): Same. * ira-build.c (finish_loop_tree_nodes): Same. (low_pressure_loop_node_p): Same. * ira-color.c (ira_reuse_stack_slot): Same. * ira-int.h: Same. * ira.c (setup_reg_equiv): Same. (print_insn_chain): Same. (ira): Same. * loop-doloop.c (doloop_condition_get): Same. (add_test): Same. (record_reg_sets): Same. (doloop_optimize): Same. * loop-init.c (loop_optimizer_init): Same. (fix_loop_structure): Same. * loop-invariant.c (merge_identical_invariants): Same. (compute_always_reached): Same. (find_exits): Same. (may_assign_reg_p): Same. (find_invariants_bb): Same. (find_invariants_body): Same. (replace_uses): Same. (can_move_invariant_reg): Same. (free_inv_motion_data): Same. (move_single_loop_invariants): Same. (change_pressure): Same. (mark_ref_regs): Same. (calculate_loop_reg_pressure): Same. * loop-iv.c (biv_entry_hasher::equal): Same. (iv_extend_to_rtx_code): Same. (check_iv_ref_table_size): Same. (clear_iv_info): Same. (latch_dominating_def): Same. (iv_get_reaching_def): Same. (iv_constant): Same. (iv_subreg): Same. (iv_extend): Same. (iv_neg): Same. (iv_add): Same. (iv_mult): Same. (get_biv_step): Same. (record_iv): Same. (analyzed_for_bivness_p): Same. (record_biv): Same. (iv_analyze_biv): Same. (iv_analyze_expr): Same. (iv_analyze_def): Same. (iv_analyze_op): Same. (iv_analyze): Same. (iv_analyze_result): Same. (biv_p): Same. (eliminate_implied_conditions): Same. (simplify_using_initial_values): Same. (shorten_into_mode): Same. (canonicalize_iv_subregs): Same. (determine_max_iter): Same. (check_simple_exit): Same. (find_simple_exit): Same. (get_simple_loop_desc): Same. * loop-unroll.c (report_unroll): Same. (decide_unrolling): Same. (unroll_loops): Same. (loop_exit_at_end_p): Same. (decide_unroll_constant_iterations): Same. (unroll_loop_constant_iterations): Same. (compare_and_jump_seq): Same. (unroll_loop_runtime_iterations): Same. (decide_unroll_stupid): Same. (unroll_loop_stupid): Same. (referenced_in_one_insn_in_loop_p): Same. (reset_debug_uses_in_loop): Same. (analyze_iv_to_split_insn): Same. * lra-eliminations.c (lra_debug_elim_table): Same. (setup_can_eliminate): Same. (form_sum): Same. (lra_get_elimination_hard_regno): Same. (lra_eliminate_regs_1): Same. (eliminate_regs_in_insn): Same. (update_reg_eliminate): Same. (init_elimination): Same. (lra_eliminate): Same. * lra-int.h: Same. * lra-lives.c (initiate_live_solver): Same. * lra-remat.c (create_remat_bb_data): Same. * lra-spills.c (lra_spill): Same. * lra.c (lra_set_insn_recog_data): Same. (lra_set_used_insn_alternative_by_uid): Same. (init_reg_info): Same. (expand_reg_info): Same. * lto-cgraph.c (output_symtab): Same. (read_identifier): Same. (get_alias_symbol): Same. (input_node): Same. (input_varpool_node): Same. (input_ref): Same. (input_edge): Same. (input_cgraph_1): Same. (input_refs): Same. (input_symtab): Same. (input_offload_tables): Same. (output_cgraph_opt_summary): Same. (input_edge_opt_summary): Same. (input_cgraph_opt_section): Same. * lto-section-in.c (lto_free_raw_section_data): Same. (lto_create_simple_input_block): Same. (lto_free_function_in_decl_state_for_node): Same. * lto-streamer-in.c (lto_tag_check_set): Same. (lto_location_cache::revert_location_cache): Same. (lto_location_cache::input_location): Same. (lto_input_location): Same. (stream_input_location_now): Same. (lto_input_tree_ref): Same. (lto_input_eh_catch_list): Same. (input_eh_region): Same. (lto_init_eh): Same. (make_new_block): Same. (input_cfg): Same. (fixup_call_stmt_edges): Same. (input_struct_function_base): Same. (input_function): Same. (lto_read_body_or_constructor): Same. (lto_read_tree_1): Same. (lto_read_tree): Same. (lto_input_scc): Same. (lto_input_tree_1): Same. (lto_input_toplevel_asms): Same. (lto_input_mode_table): Same. (lto_reader_init): Same. (lto_data_in_create): Same. * lto-streamer-out.c (output_cfg): Same. * lto-streamer.h: Same. * modulo-sched.c (duplicate_insns_of_cycles): Same. (generate_prolog_epilog): Same. (mark_loop_unsched): Same. (dump_insn_location): Same. (loop_canon_p): Same. (sms_schedule): Same. * omp-expand.c (expand_omp_for_ordered_loops): Same. (expand_omp_for_generic): Same. (expand_omp_for_static_nochunk): Same. (expand_omp_for_static_chunk): Same. (expand_omp_simd): Same. (expand_omp_taskloop_for_inner): Same. (expand_oacc_for): Same. (expand_omp_atomic_pipeline): Same. (mark_loops_in_oacc_kernels_region): Same. * omp-offload.c (oacc_xform_loop): Same. * omp-simd-clone.c (simd_clone_adjust): Same. * optabs-query.c (get_traditional_extraction_insn): Same. * optabs.c (expand_vector_broadcast): Same. (expand_binop_directly): Same. (expand_twoval_unop): Same. (expand_twoval_binop): Same. (expand_unop_direct): Same. (emit_indirect_jump): Same. (emit_conditional_move): Same. (emit_conditional_neg_or_complement): Same. (emit_conditional_add): Same. (vector_compare_rtx): Same. (expand_vec_perm_1): Same. (expand_vec_perm_const): Same. (expand_vec_cond_expr): Same. (expand_vec_series_expr): Same. (maybe_emit_atomic_exchange): Same. (maybe_emit_sync_lock_test_and_set): Same. (expand_atomic_compare_and_swap): Same. (expand_atomic_load): Same. (expand_atomic_store): Same. (maybe_emit_op): Same. (valid_multiword_target_p): Same. (create_integer_operand): Same. (maybe_legitimize_operand_same_code): Same. (maybe_legitimize_operand): Same. (create_convert_operand_from_type): Same. (can_reuse_operands_p): Same. (maybe_legitimize_operands): Same. (maybe_gen_insn): Same. (maybe_expand_insn): Same. (maybe_expand_jump_insn): Same. (expand_insn): Same. * optabs.h (create_expand_operand): Same. (create_fixed_operand): Same. (create_output_operand): Same. (create_input_operand): Same. (create_convert_operand_to): Same. (create_convert_operand_from): Same. * optinfo.h: Same. * poly-int.h: Same. * predict.c (optimize_insn_for_speed_p): Same. (optimize_loop_for_size_p): Same. (optimize_loop_for_speed_p): Same. (optimize_loop_nest_for_speed_p): Same. (get_base_value): Same. (predicted_by_loop_heuristics_p): Same. (predict_extra_loop_exits): Same. (predict_loops): Same. (predict_paths_for_bb): Same. (predict_paths_leading_to): Same. (propagate_freq): Same. (pass_profile::execute): Same. * predict.h: Same. * profile-count.c (profile_count::differs_from_p): Same. (profile_probability::differs_lot_from_p): Same. * profile-count.h: Same. * profile.c (branch_prob): Same. * regrename.c (free_chain_data): Same. (mark_conflict): Same. (create_new_chain): Same. (merge_overlapping_regs): Same. (init_rename_info): Same. (merge_chains): Same. (regrename_analyze): Same. (regrename_do_replace): Same. (scan_rtx_reg): Same. (record_out_operands): Same. (build_def_use): Same. * regrename.h: Same. * reload.h: Same. * reload1.c (init_reload): Same. (maybe_fix_stack_asms): Same. (copy_reloads): Same. (count_pseudo): Same. (count_spilled_pseudo): Same. (find_reg): Same. (find_reload_regs): Same. (select_reload_regs): Same. (spill_hard_reg): Same. (fixup_eh_region_note): Same. (set_reload_reg): Same. (allocate_reload_reg): Same. (compute_reload_subreg_offset): Same. (reload_adjust_reg_for_icode): Same. (emit_input_reload_insns): Same. (emit_output_reload_insns): Same. (do_input_reload): Same. (inherit_piecemeal_p): Same. * rtl.h: Same. * sanopt.c (maybe_get_dominating_check): Same. (maybe_optimize_ubsan_ptr_ifn): Same. (can_remove_asan_check): Same. (maybe_optimize_asan_check_ifn): Same. (sanopt_optimize_walker): Same. * sched-deps.c (add_dependence_list): Same. (chain_to_prev_insn): Same. (add_insn_mem_dependence): Same. (create_insn_reg_set): Same. (maybe_extend_reg_info_p): Same. (sched_analyze_reg): Same. (sched_analyze_1): Same. (get_implicit_reg_pending_clobbers): Same. (chain_to_prev_insn_p): Same. (deps_analyze_insn): Same. (deps_start_bb): Same. (sched_free_deps): Same. (init_deps): Same. (init_deps_reg_last): Same. (free_deps): Same. * sched-ebb.c: Same. * sched-int.h: Same. * sched-rgn.c (add_branch_dependences): Same. (concat_insn_mem_list): Same. (deps_join): Same. (sched_rgn_compute_dependencies): Same. * sel-sched-ir.c (reset_target_context): Same. (copy_deps_context): Same. (init_id_from_df): Same. (has_dependence_p): Same. (change_loops_latches): Same. (bb_top_order_comparator): Same. (make_region_from_loop_preheader): Same. (sel_init_pipelining): Same. (get_loop_nest_for_rgn): Same. (make_regions_from_the_rest): Same. (sel_is_loop_preheader_p): Same. * sel-sched-ir.h (inner_loop_header_p): Same. (get_all_loop_exits): Same. * selftest.h: Same. * sese.c (sese_build_liveouts): Same. (sese_insert_phis_for_liveouts): Same. * sese.h (defined_in_sese_p): Same. * sreal.c (sreal::stream_out): Same. * sreal.h: Same. * streamer-hooks.h: Same. * target-globals.c (save_target_globals): Same. * target-globals.h: Same. * target.def: Same. * target.h: Same. * targhooks.c (default_has_ifunc_p): Same. (default_empty_mask_is_expensive): Same. (default_init_cost): Same. * targhooks.h: Same. * toplev.c: Same. * tree-affine.c (aff_combination_mult): Same. (aff_combination_expand): Same. (aff_combination_constant_multiple_p): Same. * tree-affine.h: Same. * tree-cfg.c (build_gimple_cfg): Same. (replace_loop_annotate_in_block): Same. (replace_uses_by): Same. (remove_bb): Same. (dump_cfg_stats): Same. (gimple_duplicate_sese_region): Same. (gimple_duplicate_sese_tail): Same. (move_block_to_fn): Same. (replace_block_vars_by_duplicates): Same. (move_sese_region_to_fn): Same. (print_loops_bb): Same. (print_loop): Same. (print_loops): Same. (debug): Same. (debug_loops): Same. * tree-cfg.h: Same. * tree-chrec.c (chrec_fold_plus_poly_poly): Same. (chrec_fold_multiply_poly_poly): Same. (chrec_evaluate): Same. (chrec_component_in_loop_num): Same. (reset_evolution_in_loop): Same. (is_multivariate_chrec): Same. (chrec_contains_symbols): Same. (nb_vars_in_chrec): Same. (chrec_convert_1): Same. (chrec_convert_aggressive): Same. * tree-chrec.h: Same. * tree-core.h: Same. * tree-data-ref.c (dump_data_dependence_relation): Same. (canonicalize_base_object_address): Same. (data_ref_compare_tree): Same. (prune_runtime_alias_test_list): Same. (get_segment_min_max): Same. (create_intersect_range_checks): Same. (conflict_fn_no_dependence): Same. (object_address_invariant_in_loop_p): Same. (analyze_ziv_subscript): Same. (analyze_siv_subscript_cst_affine): Same. (analyze_miv_subscript): Same. (analyze_overlapping_iterations): Same. (build_classic_dist_vector_1): Same. (add_other_self_distances): Same. (same_access_functions): Same. (build_classic_dir_vector): Same. (subscript_dependence_tester_1): Same. (subscript_dependence_tester): Same. (access_functions_are_affine_or_constant_p): Same. (get_references_in_stmt): Same. (loop_nest_has_data_refs): Same. (graphite_find_data_references_in_stmt): Same. (find_data_references_in_bb): Same. (get_base_for_alignment): Same. (find_loop_nest_1): Same. (find_loop_nest): Same. * tree-data-ref.h (dr_alignment): Same. (ddr_dependence_level): Same. * tree-if-conv.c (fold_build_cond_expr): Same. (add_to_predicate_list): Same. (add_to_dst_predicate_list): Same. (phi_convertible_by_degenerating_args): Same. (idx_within_array_bound): Same. (all_preds_critical_p): Same. (pred_blocks_visited_p): Same. (predicate_bbs): Same. (build_region): Same. (if_convertible_loop_p_1): Same. (is_cond_scalar_reduction): Same. (predicate_scalar_phi): Same. (remove_conditions_and_labels): Same. (combine_blocks): Same. (version_loop_for_if_conversion): Same. (versionable_outer_loop_p): Same. (ifcvt_local_dce): Same. (tree_if_conversion): Same. (pass_if_conversion::gate): Same. * tree-if-conv.h: Same. * tree-inline.c (maybe_move_debug_stmts_to_successors): Same. * tree-loop-distribution.c (bb_top_order_cmp): Same. (free_rdg): Same. (stmt_has_scalar_dependences_outside_loop): Same. (copy_loop_before): Same. (create_bb_after_loop): Same. (const_with_all_bytes_same): Same. (generate_memset_builtin): Same. (generate_memcpy_builtin): Same. (destroy_loop): Same. (build_rdg_partition_for_vertex): Same. (compute_access_range): Same. (data_ref_segment_size): Same. (latch_dominated_by_data_ref): Same. (compute_alias_check_pairs): Same. (fuse_memset_builtins): Same. (finalize_partitions): Same. (find_seed_stmts_for_distribution): Same. (prepare_perfect_loop_nest): Same. * tree-parloops.c (lambda_transform_legal_p): Same. (loop_parallel_p): Same. (reduc_stmt_res): Same. (add_field_for_name): Same. (create_call_for_reduction_1): Same. (replace_uses_in_bb_by): Same. (transform_to_exit_first_loop_alt): Same. (try_transform_to_exit_first_loop_alt): Same. (transform_to_exit_first_loop): Same. (num_phis): Same. (gen_parallel_loop): Same. (gather_scalar_reductions): Same. (get_omp_data_i_param): Same. (try_create_reduction_list): Same. (oacc_entry_exit_single_gang): Same. (parallelize_loops): Same. * tree-pass.h: Same. * tree-predcom.c (determine_offset): Same. (last_always_executed_block): Same. (split_data_refs_to_components): Same. (suitable_component_p): Same. (valid_initializer_p): Same. (find_looparound_phi): Same. (insert_looparound_copy): Same. (add_looparound_copies): Same. (determine_roots_comp): Same. (predcom_tmp_var): Same. (initialize_root_vars): Same. (initialize_root_vars_store_elim_1): Same. (initialize_root_vars_store_elim_2): Same. (finalize_eliminated_stores): Same. (initialize_root_vars_lm): Same. (remove_stmt): Same. (determine_unroll_factor): Same. (execute_pred_commoning_cbck): Same. (base_names_in_chain_on): Same. (combine_chains): Same. (pcom_stmt_dominates_stmt_p): Same. (try_combine_chains): Same. (prepare_initializers_chain_store_elim): Same. (prepare_initializers_chain): Same. (prepare_initializers): Same. (prepare_finalizers_chain): Same. (prepare_finalizers): Same. (insert_init_seqs): Same. * tree-scalar-evolution.c (loop_phi_node_p): Same. (compute_overall_effect_of_inner_loop): Same. (add_to_evolution_1): Same. (add_to_evolution): Same. (follow_ssa_edge_binary): Same. (follow_ssa_edge_expr): Same. (backedge_phi_arg_p): Same. (follow_ssa_edge_in_condition_phi_branch): Same. (follow_ssa_edge_in_condition_phi): Same. (follow_ssa_edge_inner_loop_phi): Same. (follow_ssa_edge): Same. (analyze_evolution_in_loop): Same. (analyze_initial_condition): Same. (interpret_loop_phi): Same. (interpret_condition_phi): Same. (interpret_rhs_expr): Same. (interpret_expr): Same. (interpret_gimple_assign): Same. (analyze_scalar_evolution_1): Same. (analyze_scalar_evolution): Same. (analyze_scalar_evolution_for_address_of): Same. (get_instantiated_value_entry): Same. (loop_closed_phi_def): Same. (instantiate_scev_name): Same. (instantiate_scev_poly): Same. (instantiate_scev_binary): Same. (instantiate_scev_convert): Same. (instantiate_scev_not): Same. (instantiate_scev_r): Same. (instantiate_scev): Same. (resolve_mixers): Same. (initialize_scalar_evolutions_analyzer): Same. (scev_reset_htab): Same. (scev_reset): Same. (derive_simple_iv_with_niters): Same. (simple_iv_with_niters): Same. (expression_expensive_p): Same. (final_value_replacement_loop): Same. * tree-scalar-evolution.h (block_before_loop): Same. * tree-ssa-address.h: Same. * tree-ssa-dce.c (find_obviously_necessary_stmts): Same. * tree-ssa-dom.c (edge_info::record_simple_equiv): Same. (record_edge_info): Same. * tree-ssa-live.c (var_map_base_fini): Same. (remove_unused_locals): Same. * tree-ssa-live.h: Same. * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Same. (pass_ch_vect::execute): Same. (pass_ch::process_loop_p): Same. * tree-ssa-loop-im.c (mem_ref_hasher::hash): Same. (movement_possibility): Same. (outermost_invariant_loop): Same. (stmt_cost): Same. (determine_max_movement): Same. (invariantness_dom_walker::before_dom_children): Same. (move_computations): Same. (may_move_till): Same. (force_move_till_op): Same. (force_move_till): Same. (memref_free): Same. (record_mem_ref_loc): Same. (set_ref_stored_in_loop): Same. (mark_ref_stored): Same. (sort_bbs_in_loop_postorder_cmp): Same. (sort_locs_in_loop_postorder_cmp): Same. (analyze_memory_references): Same. (mem_refs_may_alias_p): Same. (find_ref_loc_in_loop_cmp): Same. (rewrite_mem_ref_loc::operator): Same. (first_mem_ref_loc_1::operator): Same. (sm_set_flag_if_changed::operator): Same. (execute_sm_if_changed_flag_set): Same. (execute_sm): Same. (hoist_memory_references): Same. (ref_always_accessed::operator): Same. (refs_independent_p): Same. (record_dep_loop): Same. (ref_indep_loop_p_1): Same. (ref_indep_loop_p): Same. (can_sm_ref_p): Same. (find_refs_for_sm): Same. (loop_suitable_for_sm): Same. (store_motion_loop): Same. (store_motion): Same. (fill_always_executed_in): Same. * tree-ssa-loop-ivcanon.c (constant_after_peeling): Same. (estimated_unrolled_size): Same. (loop_edge_to_cancel): Same. (remove_exits_and_undefined_stmts): Same. (remove_redundant_iv_tests): Same. (unloop_loops): Same. (estimated_peeled_sequence_size): Same. (try_peel_loop): Same. (canonicalize_loop_induction_variables): Same. (canonicalize_induction_variables): Same. * tree-ssa-loop-ivopts.c (iv_inv_expr_hasher::equal): Same. (name_info): Same. (stmt_after_inc_pos): Same. (contains_abnormal_ssa_name_p): Same. (niter_for_exit): Same. (find_bivs): Same. (mark_bivs): Same. (find_givs_in_bb): Same. (find_induction_variables): Same. (find_interesting_uses_cond): Same. (outermost_invariant_loop_for_expr): Same. (idx_find_step): Same. (add_candidate_1): Same. (add_iv_candidate_derived_from_uses): Same. (alloc_use_cost_map): Same. (prepare_decl_rtl): Same. (generic_predict_doloop_p): Same. (computation_cost): Same. (determine_common_wider_type): Same. (get_computation_aff_1): Same. (get_use_type): Same. (determine_group_iv_cost_address): Same. (iv_period): Same. (difference_cannot_overflow_p): Same. (may_eliminate_iv): Same. (determine_set_costs): Same. (cheaper_cost_pair): Same. (compare_cost_pair): Same. (iv_ca_cand_for_group): Same. (iv_ca_recount_cost): Same. (iv_ca_set_remove_invs): Same. (iv_ca_set_no_cp): Same. (iv_ca_set_add_invs): Same. (iv_ca_set_cp): Same. (iv_ca_add_group): Same. (iv_ca_cost): Same. (iv_ca_compare_deps): Same. (iv_ca_delta_reverse): Same. (iv_ca_delta_commit): Same. (iv_ca_cand_used_p): Same. (iv_ca_delta_free): Same. (iv_ca_new): Same. (iv_ca_free): Same. (iv_ca_dump): Same. (iv_ca_extend): Same. (iv_ca_narrow): Same. (iv_ca_prune): Same. (cheaper_cost_with_cand): Same. (iv_ca_replace): Same. (try_add_cand_for): Same. (get_initial_solution): Same. (try_improve_iv_set): Same. (find_optimal_iv_set_1): Same. (create_new_iv): Same. (rewrite_use_compare): Same. (remove_unused_ivs): Same. (determine_scaling_factor): Same. * tree-ssa-loop-ivopts.h: Same. * tree-ssa-loop-manip.c (create_iv): Same. (compute_live_loop_exits): Same. (add_exit_phi): Same. (add_exit_phis): Same. (find_uses_to_rename_use): Same. (find_uses_to_rename_def): Same. (find_uses_to_rename_in_loop): Same. (rewrite_into_loop_closed_ssa): Same. (check_loop_closed_ssa_bb): Same. (split_loop_exit_edge): Same. (ip_end_pos): Same. (ip_normal_pos): Same. (copy_phi_node_args): Same. (gimple_duplicate_loop_to_header_edge): Same. (can_unroll_loop_p): Same. (determine_exit_conditions): Same. (scale_dominated_blocks_in_loop): Same. (niter_for_unrolled_loop): Same. (tree_transform_and_unroll_loop): Same. (rewrite_all_phi_nodes_with_iv): Same. * tree-ssa-loop-manip.h: Same. * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Same. (number_of_iterations_ne): Same. (assert_no_overflow_lt): Same. (assert_loop_rolls_lt): Same. (number_of_iterations_lt): Same. (adjust_cond_for_loop_until_wrap): Same. (tree_simplify_using_condition): Same. (simplify_using_initial_conditions): Same. (simplify_using_outer_evolutions): Same. (loop_only_exit_p): Same. (ssa_defined_by_minus_one_stmt_p): Same. (number_of_iterations_popcount): Same. (number_of_iterations_exit): Same. (find_loop_niter): Same. (finite_loop_p): Same. (chain_of_csts_start): Same. (get_val_for): Same. (loop_niter_by_eval): Same. (derive_constant_upper_bound_ops): Same. (do_warn_aggressive_loop_optimizations): Same. (record_estimate): Same. (get_cst_init_from_scev): Same. (record_nonwrapping_iv): Same. (idx_infer_loop_bounds): Same. (infer_loop_bounds_from_ref): Same. (infer_loop_bounds_from_array): Same. (infer_loop_bounds_from_pointer_arith): Same. (infer_loop_bounds_from_signedness): Same. (bound_index): Same. (discover_iteration_bound_by_body_walk): Same. (maybe_lower_iteration_bound): Same. (estimate_numbers_of_iterations): Same. (estimated_loop_iterations): Same. (estimated_loop_iterations_int): Same. (max_loop_iterations): Same. (max_loop_iterations_int): Same. (likely_max_loop_iterations): Same. (likely_max_loop_iterations_int): Same. (estimated_stmt_executions_int): Same. (max_stmt_executions): Same. (likely_max_stmt_executions): Same. (estimated_stmt_executions): Same. (stmt_dominates_stmt_p): Same. (nowrap_type_p): Same. (loop_exits_before_overflow): Same. (scev_var_range_cant_overflow): Same. (scev_probably_wraps_p): Same. (free_numbers_of_iterations_estimates): Same. * tree-ssa-loop-niter.h: Same. * tree-ssa-loop-prefetch.c (release_mem_refs): Same. (idx_analyze_ref): Same. (analyze_ref): Same. (gather_memory_references_ref): Same. (mark_nontemporal_store): Same. (emit_mfence_after_loop): Same. (may_use_storent_in_loop_p): Same. (mark_nontemporal_stores): Same. (should_unroll_loop_p): Same. (volume_of_dist_vector): Same. (add_subscript_strides): Same. (self_reuse_distance): Same. (insn_to_prefetch_ratio_too_small_p): Same. * tree-ssa-loop-split.c (split_at_bb_p): Same. (patch_loop_exit): Same. (find_or_create_guard_phi): Same. (easy_exit_values): Same. (connect_loop_phis): Same. (connect_loops): Same. (compute_new_first_bound): Same. (split_loop): Same. (tree_ssa_split_loops): Same. * tree-ssa-loop-unswitch.c (tree_ssa_unswitch_loops): Same. (is_maybe_undefined): Same. (tree_may_unswitch_on): Same. (simplify_using_entry_checks): Same. (tree_unswitch_single_loop): Same. (tree_unswitch_loop): Same. (tree_unswitch_outer_loop): Same. (empty_bb_without_guard_p): Same. (used_outside_loop_p): Same. (get_vop_from_header): Same. (hoist_guard): Same. * tree-ssa-loop.c (gate_oacc_kernels): Same. (get_lsm_tmp_name): Same. * tree-ssa-loop.h: Same. * tree-ssa-reassoc.c (add_repeat_to_ops_vec): Same. (build_and_add_sum): Same. (no_side_effect_bb): Same. (get_ops): Same. (linearize_expr): Same. (should_break_up_subtract): Same. (linearize_expr_tree): Same. * tree-ssa-scopedtables.c: Same. * tree-ssa-scopedtables.h: Same. * tree-ssa-structalias.c (condense_visit): Same. (label_visit): Same. (dump_pred_graph): Same. (perform_var_substitution): Same. (move_complex_constraints): Same. (remove_preds_and_fake_succs): Same. * tree-ssa-threadupdate.c (dbds_continue_enumeration_p): Same. (determine_bb_domination_status): Same. (duplicate_thread_path): Same. (thread_through_all_blocks): Same. * tree-ssa-threadupdate.h: Same. * tree-streamer-in.c (streamer_read_string_cst): Same. (input_identifier): Same. (unpack_ts_type_common_value_fields): Same. (unpack_ts_block_value_fields): Same. (unpack_ts_translation_unit_decl_value_fields): Same. (unpack_ts_omp_clause_value_fields): Same. (streamer_read_tree_bitfields): Same. (streamer_alloc_tree): Same. (lto_input_ts_common_tree_pointers): Same. (lto_input_ts_vector_tree_pointers): Same. (lto_input_ts_poly_tree_pointers): Same. (lto_input_ts_complex_tree_pointers): Same. (lto_input_ts_decl_minimal_tree_pointers): Same. (lto_input_ts_decl_common_tree_pointers): Same. (lto_input_ts_decl_non_common_tree_pointers): Same. (lto_input_ts_decl_with_vis_tree_pointers): Same. (lto_input_ts_field_decl_tree_pointers): Same. (lto_input_ts_function_decl_tree_pointers): Same. (lto_input_ts_type_common_tree_pointers): Same. (lto_input_ts_type_non_common_tree_pointers): Same. (lto_input_ts_list_tree_pointers): Same. (lto_input_ts_vec_tree_pointers): Same. (lto_input_ts_exp_tree_pointers): Same. (lto_input_ts_block_tree_pointers): Same. (lto_input_ts_binfo_tree_pointers): Same. (lto_input_ts_constructor_tree_pointers): Same. (lto_input_ts_omp_clause_tree_pointers): Same. (streamer_read_tree_body): Same. * tree-streamer.h: Same. * tree-switch-conversion.c (bit_test_cluster::is_beneficial): Same. * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Same. (vect_analyze_possibly_independent_ddr): Same. (vect_analyze_data_ref_dependence): Same. (vect_compute_data_ref_alignment): Same. (vect_enhance_data_refs_alignment): Same. (vect_analyze_data_ref_access): Same. (vect_check_gather_scatter): Same. (vect_find_stmt_data_reference): Same. (vect_create_addr_base_for_vector_ref): Same. (vect_setup_realignment): Same. (vect_supportable_dr_alignment): Same. * tree-vect-loop-manip.c (rename_variables_in_bb): Same. (adjust_phi_and_debug_stmts): Same. (vect_set_loop_mask): Same. (add_preheader_seq): Same. (vect_maybe_permute_loop_masks): Same. (vect_set_loop_masks_directly): Same. (vect_set_loop_condition_masked): Same. (vect_set_loop_condition_unmasked): Same. (slpeel_duplicate_current_defs_from_edges): Same. (slpeel_add_loop_guard): Same. (slpeel_can_duplicate_loop_p): Same. (create_lcssa_for_virtual_phi): Same. (iv_phi_p): Same. (vect_update_ivs_after_vectorizer): Same. (vect_gen_vector_loop_niters_mult_vf): Same. (slpeel_update_phi_nodes_for_loops): Same. (slpeel_update_phi_nodes_for_guard1): Same. (find_guard_arg): Same. (slpeel_update_phi_nodes_for_guard2): Same. (slpeel_update_phi_nodes_for_lcssa): Same. (vect_do_peeling): Same. (vect_create_cond_for_alias_checks): Same. (vect_loop_versioning): Same. * tree-vect-loop.c (vect_determine_vf_for_stmt): Same. (vect_inner_phi_in_double_reduction_p): Same. (vect_analyze_scalar_cycles_1): Same. (vect_fixup_scalar_cycles_with_patterns): Same. (vect_get_loop_niters): Same. (bb_in_loop_p): Same. (vect_get_max_nscalars_per_iter): Same. (vect_verify_full_masking): Same. (vect_compute_single_scalar_iteration_cost): Same. (vect_analyze_loop_form_1): Same. (vect_analyze_loop_form): Same. (vect_active_double_reduction_p): Same. (vect_analyze_loop_operations): Same. (neutral_op_for_slp_reduction): Same. (vect_is_simple_reduction): Same. (vect_model_reduction_cost): Same. (get_initial_def_for_reduction): Same. (get_initial_defs_for_reduction): Same. (vect_create_epilog_for_reduction): Same. (vectorize_fold_left_reduction): Same. (vectorizable_reduction): Same. (vectorizable_induction): Same. (vectorizable_live_operation): Same. (loop_niters_no_overflow): Same. (vect_get_loop_mask): Same. (vect_transform_loop_stmt): Same. (vect_transform_loop): Same. * tree-vect-patterns.c (vect_reassociating_reduction_p): Same. (vect_determine_precisions): Same. (vect_pattern_recog_1): Same. * tree-vect-slp.c (vect_analyze_slp_instance): Same. * tree-vect-stmts.c (stmt_vectype): Same. (process_use): Same. (vect_init_vector_1): Same. (vect_truncate_gather_scatter_offset): Same. (get_group_load_store_type): Same. (vect_build_gather_load_calls): Same. (vect_get_strided_load_store_ops): Same. (vectorizable_simd_clone_call): Same. (vectorizable_store): Same. (permute_vec_elements): Same. (vectorizable_load): Same. (vect_transform_stmt): Same. (supportable_widening_operation): Same. * tree-vectorizer.c (vec_info::replace_stmt): Same. (vec_info::free_stmt_vec_info): Same. (vect_free_loop_info_assumptions): Same. (vect_loop_vectorized_call): Same. (set_uid_loop_bbs): Same. (vectorize_loops): Same. * tree-vectorizer.h (STMT_VINFO_BB_VINFO): Same. * tree.c (add_tree_to_fld_list): Same. (fld_type_variant_equal_p): Same. (fld_decl_context): Same. (fld_incomplete_type_of): Same. (free_lang_data_in_binfo): Same. (need_assembler_name_p): Same. (find_decls_types_r): Same. (get_eh_types_for_runtime): Same. (find_decls_types_in_eh_region): Same. (find_decls_types_in_node): Same. (assign_assembler_name_if_needed): Same. * value-prof.c (stream_out_histogram_value): Same. * value-prof.h: Same. * var-tracking.c (use_narrower_mode): Same. (prepare_call_arguments): Same. (vt_expand_loc_callback): Same. (resolve_expansions_pending_recursion): Same. (vt_expand_loc): Same. * varasm.c (const_hash_1): Same. (compare_constant): Same. (tree_output_constant_def): Same. (simplify_subtraction): Same. (get_pool_constant): Same. (output_constant_pool_2): Same. (output_constant_pool_1): Same. (mark_constants_in_pattern): Same. (mark_constant_pool): Same. (get_section_anchor): Same. * vr-values.c (compare_range_with_value): Same. (vr_values::extract_range_from_phi_node): Same. * vr-values.h: Same. * web.c (unionfind_union): Same. * wide-int.h: Same. From-SVN: r273311
2019-07-09 18:32:49 +00:00
const class line_maps *line_table = pfile->line_table;
Replace line_map union with C++ class hierarchy gcc/ChangeLog: * diagnostic.c (diagnostic_report_current_module): Strengthen local "new_map" from const line_map * to const line_map_ordinary *. * genmatch.c (error_cb): Likewise for local "map". (output_line_directive): Likewise for local "map". * input.c (expand_location_1): Likewise for local "map". Pass NULL rather than &map to linemap_unwind_to_first_non_reserved_loc, since the value is never read from there, and the value written back not read from here. (is_location_from_builtin_token): Strengthen local "map" from const line_map * to const line_map_ordinary *. (dump_location_info): Strengthen locals "map" from line_map *, one to const line_map_ordinary *, the other to const line_map_macro *. * tree-diagnostic.c (loc_map_pair): Strengthen field "map" from const line_map * to const line_map_macro *. (maybe_unwind_expanded_macro_loc): Add a call to linemap_check_macro when writing to the "map" field of the loc_map_pair. Introduce local const line_map_ordinary * "ord_map", using it in place of "map" in the part of the function where we know we have an ordinary map. Strengthen local "m" from const line_map * to const line_map_ordinary *. gcc/ada/ChangeLog: * gcc-interface/trans.c (Sloc_to_locus1): Strenghthen local "map" from line_map * to line_map_ordinary *. gcc/c-family/ChangeLog: * c-common.h (fe_file_change): Strengthen param from const line_map * to const line_map_ordinary *. (pp_file_change): Likewise. * c-lex.c (fe_file_change): Likewise. (cb_define): Use linemap_check_ordinary when invoking SOURCE_LINE. (cb_undef): Likewise. * c-opts.c (c_finish_options): Use linemap_check_ordinary when invoking cb_file_change. (c_finish_options): Likewise. (push_command_line_include): Likewise. (cb_file_change): Strengthen param "new_map" from const line_map * to const line_map_ordinary *. * c-ppoutput.c (cb_define): Likewise for local "map". (pp_file_change): Likewise for param "map" and local "from". gcc/fortran/ChangeLog: * cpp.c (maybe_print_line): Strengthen local "map" from const line_map * to const line_map_ordinary *. (cb_file_change): Likewise for param "map" and local "from". (cb_line_change): Likewise for local "map". libcpp/ChangeLog: * directives.c (do_line): Strengthen local "map" from const line_map * to const line_map_ordinary *. (do_linemarker): Likewise. (_cpp_do_file_change): Assert that we're not dealing with a macro map. Introduce local "ord_map" via a call to linemap_check_ordinary, guarded within the check for non-NULL. Use it for typesafety. * files.c (cpp_make_system_header): Strengthen local "map" from const line_map * to const line_map_ordinary *. * include/cpplib.h (struct cpp_callbacks): Likewise for second parameter of "file_change" callback. * include/line-map.h (struct line_map): Convert from a struct containing a union to a base class. (struct line_map_ordinary): Convert to a subclass of line_map. (struct line_map_macro): Likewise. (linemap_check_ordinary): Strengthen return type from line_map * to line_map_ordinary *, and add a const-variant. (linemap_check_macro): New pair of functions. (ORDINARY_MAP_STARTING_LINE_NUMBER): Strengthen param from const line_map * to const line_map_ordinary *, eliminating call to linemap_check_ordinary. Likewise for the non-const variant. (ORDINARY_MAP_INCLUDER_FILE_INDEX): Likewise. (ORDINARY_MAP_IN_SYSTEM_HEADER_P): Likewise. (ORDINARY_MAP_NUMBER_OF_COLUMN_BITS): Likewise. (ORDINARY_MAP_FILE_NAME): Likewise. (MACRO_MAP_MACRO): Strengthen param from const line_map * to const line_map_macro *. Likewise for the non-const variant. (MACRO_MAP_NUM_MACRO_TOKENS): Likewise. (MACRO_MAP_LOCATIONS): Likewise. (MACRO_MAP_EXPANSION_POINT_LOCATION): Likewise. (struct maps_info): Replace with... (struct maps_info_ordinary):...this and... (struct maps_info_macro): ...this. (struct line_maps): Convert fields "info_ordinary" and "info_macro" to the above new structs. (LINEMAPS_MAP_INFO): Delete both functions. (LINEMAPS_MAPS): Likewise. (LINEMAPS_ALLOCATED): Rewrite both variants to avoid using LINEMAPS_MAP_INFO. (LINEMAPS_USED): Likewise. (LINEMAPS_CACHE): Likewise. (LINEMAPS_MAP_AT): Likewise. (LINEMAPS_ORDINARY_MAPS): Strengthen return type from line_map * to line_map_ordinary *. (LINEMAPS_ORDINARY_MAP_AT): Likewise. (LINEMAPS_LAST_ORDINARY_MAP): Likewise. (LINEMAPS_LAST_ALLOCATED_ORDINARY_MAP): Likewise. (LINEMAPS_MACRO_MAPS): Strengthen return type from line_map * to line_map_macro *. (LINEMAPS_MACRO_MAP_AT): Likewise. (LINEMAPS_LAST_MACRO_MAP): Likewise. (LINEMAPS_LAST_ALLOCATED_MACRO_MAP): Likewise. (linemap_map_get_macro_name): Strengthen param from const line_map * to const line_map_macro *. (SOURCE_LINE): Strengthen first param from const line_map * to const line_map_ordinary *, removing call to linemap_check_ordinary. (SOURCE_COLUMN): Likewise. (LAST_SOURCE_LINE_LOCATION): Likewise. (LAST_SOURCE_LINE): Strengthen first param from const line_map * to const line_map_ordinary *. (LAST_SOURCE_COLUMN): Likewise. (INCLUDED_FROM): Strengthen return type from line_map * to line_map_ordinary *., and second param from const line_map * to const line_map_ordinary *, removing call to linemap_check_ordinary. (MAIN_FILE_P): Strengthen param from const line_map * to const line_map_ordinary *, removing call to linemap_check_ordinary. (linemap_position_for_line_and_column): Strengthen param from const line_map * to const line_map_ordinary *. (LINEMAP_FILE): Strengthen param from const line_map * to const line_map_ordinary *, removing call to linemap_check_ordinary. (LINEMAP_LINE): Likewise. (LINEMAP_SYSP): Likewise. (linemap_resolve_location): Strengthen final param from const line_map ** to const line_map_ordinary **. * internal.h (CPP_INCREMENT_LINE): Likewise for local "map". (linemap_enter_macro): Strengthen return type from const line_map * to const line_map_macro *. (linemap_add_macro_token): Likewise for first param. * line-map.c (linemap_check_files_exited): Strengthen local "map" from const line_map * to const line_map_ordinary *. (new_linemap): Introduce local "map_size" and use it when calculating how large the buffer should be. Rewrite based on change of info_macro and info_ordinary into distinct types. (linemap_add): Strengthen locals "map" and "from" from line_map * to line_map_ordinary *. (linemap_enter_macro): Strengthen return type from const line_map * to const line_map_macro *, and local "map" from line_map * to line_map_macro *. (linemap_add_macro_token): Strengthen param "map" from const line_map * to const line_map_macro *. (linemap_line_start): Strengthen local "map" from line_map * to line_map_ordinary *. (linemap_position_for_column): Likewise. (linemap_position_for_line_and_column): Strengthen first param from const line_map * to const line_map_ordinary *. (linemap_position_for_loc_and_offset): Strengthen local "map" from const line_map * to const line_map_ordinary *. (linemap_ordinary_map_lookup): Likewise for return type and locals "cached" and "result". (linemap_macro_map_lookup): Strengthen return type and locals "cached" and "result" from const line_map * to const line_map_macro *. (linemap_macro_map_loc_to_exp_point): Likewise for param "map". (linemap_macro_map_loc_to_def_point): Likewise. (linemap_macro_map_loc_unwind_toward_spelling): Likewise. (linemap_get_expansion_line): Strengthen local "map" from const line_map * to const line_map_ordinary *. (linemap_get_expansion_filename): Likewise. (linemap_map_get_macro_name): Strengthen param from const line_map * to const line_map_macro *. (linemap_location_in_system_header_p): Add call to linemap_check_ordinary in region guarded by !linemap_macro_expansion_map_p. Introduce local "macro_map" via linemap_check_macro in other region, using it in place of "map" for typesafety. (first_map_in_common_1): Add calls to linemap_check_macro. (trace_include): Strengthen param "map" from const line_map * to const line_map_ordinary *. (linemap_macro_loc_to_spelling_point): Strengthen final param from const line_map ** to const line_map_ordinary **. Replace a C-style cast with a const_cast, and add calls to linemap_check_macro and linemap_check_ordinary. (linemap_macro_loc_to_def_point): Likewise. (linemap_macro_loc_to_exp_point): Likewise. (linemap_resolve_location): Strengthen final param from const line_map ** to const line_map_ordinary **. (linemap_unwind_toward_expansion): Introduce local "macro_map" via a checked cast and use it in place of *map. (linemap_unwind_to_first_non_reserved_loc): Strengthen local "map1" from const line_map * to const line_map_ordinary *. (linemap_expand_location): Introduce local "ord_map" via a checked cast and use it in place of map. (linemap_dump): Make local "map" const. Strengthen local "includer_map" from line_map * to const line_map_ordinary *. Introduce locals "ord_map" and "macro_map" via checked casts and use them in place of "map" for typesafety. (linemap_dump_location): Strengthen local "map" from const line_map * to const line_map_ordinary *. (linemap_get_file_highest_location): Update for elimination of union. (linemap_get_statistics): Strengthen local "cur_map" from line_map * to const line_map_macro *. Update uses of sizeof to use the appropriate line_map subclasses. * macro.c (_cpp_warn_if_unused_macro): Add call to linemap_check_ordinary. (builtin_macro): Strengthen local "map" from const line_map * to const line_map_macro *. (enter_macro_context): Likewise. (replace_args): Likewise. (tokens_buff_put_token_to): Likewise for param "map". (tokens_buff_add_token): Likewise. From-SVN: r223365
2015-05-19 13:18:01 +00:00
const line_map_ordinary *map = LINEMAPS_LAST_ORDINARY_MAP (line_table);
/* 1 = system header, 2 = system header to be treated as C. */
if (syshdr)
flags = 1 + (externc != 0);
Represent column numbers using line-map's source_location. The "next available source_location" is now managed internally by line-maps.c rather than by clients. * line-map.h (struct line_map): New field column_bits. <from_line>: Rename field to start_location. (struct line_maps): New fields highest_location and max_column_hint. (linemap_check_files_exited): New declaration. (linemap_line_start): New declaration. (linemap_add): Remove from_line parameter; use highest_location field. (SOURCE_LINE, LAST_SOURCE_LINE): Modify to use column_bits. (SOURCE_COLUMN, LAST_SOURCE_LINE_LOCATION): New macros. (CURRENT_LINE_MAP): Remove macro. (linemap_position_for_column): New inline function. * line-map.c (linemap_init): Clear new fields. (linemap_check_files_exited): New function, extracted from ... (linemap_free): Use linemap_check_files_exited. (linemap_add): Remove from_line parameter. Various updates. (linemap_line_start): New function. (linemap_lookeup): Update for new field names. * cpphash.h (struct cpp_reader) <map>: Field removed. Because linemap_position_for_column may unpredictably change the current map, it is cleaner and simpler for us to not cache it in cpp_reader. (struct cpp_buffer): New sysp field. Changed warned_cplusplus_comments and from_stage3 to bitfields. * cppinit.c (cpp_read_min_file): pfile->map no longer exists. * cpplib.c (do_line, do_linemarker, _cpp_do_file_change): Get current map using linemap_lookup. (do_linemarker): Also set buffer's sysp field. (destringize_and_run): No longer need to decrement current line. * cppfiles.c (_cpp_stack_file): Set sysp from and in buffer. (search_path_head, open_file_failed): Use buffer's sysp. (cpp_make_system_header): Get current map using linemap_lookup. Also set buffer's sysp flag. * cppmacro.c (_cpp_builtin_macro_text): Likewise use linemap_lookup. * cpphash.h (CPP_INCREMENT_LINE): New macro. (struct cpp_buffer): Moved fields saved_cur, saved_rlimit to ... (struct cpp_reader): ... and adding saved_line_base field. * cpptrad.c (_cpp_overlay_buffer, _cpp_remove_overlay): Update accordingly. Don't adjust line. (_cpp_scan_out_logical_line): Use CPP_INCREMENT_LINE. * cpphash.c (CPP_IN_SYSTEM_HEADER): Replaced macro by ... (cpp_in_system_header): ... new inline function, using buffer's sysp. * cpperror.c (_cpp_begin_message): Update to use cpp_in_system_header. * cpplex.c (_cpp_lex_direct): Likewise. * cppmacro.c (_cpp_builtin_macro_text): Likewise. * cppmacro.c (_cpp_create_definition): Use buffer's sysp field. * cpplib.h (struct cpp_token): Rename line field to src_loc. Remove col field as it is now subsumed by src_loc. * cpperror.c: Update various field, parameter, and macro names. (print_location): If col==0, try SOURCE_COLUMN of line. (cpp_error): Use cur_token's src_loc field, rather than line+col. * cpplib.c (do_diagnostic): Token's src_loc fields replaces line+col. * cpplex.c (_cpp_process_line_notes, _cpp_lex_direct, _cpp_skip_block_comment): Use CPP_INCREMENT_LINE. (_cpp_temp_token): Replace cpp_token's line+col fields by src_loc. (_cpp_get_fresh_line): Don't need to adjust line for missing newline. (_cpp_lex_direct): Use linemap_position_for_column. * c-ppoutput.c (maybe_print_line, print_line): Don't take map parameter. Instead get it from the line_table global. Adjust callers. (print): Remove map field. Replace line field to src_line. (init_pp_output, account_for_newlines, maybe_print_line): Adjust. (cb_line_change): Use SOURCE_COLUMN. Minor optimizations. (pp_file_change): Use MAIN_FILE_P since we cannot checked print.map. Use LAST_SOURCE_LINE_LOCATION to "catch up" after #include. * cpptrad.c (copy_comment): Rename variable. * c-lex.c (map): Remove static variable, for same reason we removed cpp_reader's map field. (cb_line_change, cb_def_pragma, cb_define, cb_undef): Hence we need to call linemap_lookup. (cb_line_change): Token's line field replaced by src_loc. (fe_file_change): Use MAINFILE_P and LAST_SOURCE_LINE macros. Don't save new_map. * cpphash.h, cpperror.c, cpplib.h: Some renames of fileline to source_location. From-SVN: r77663
2004-02-11 07:29:30 -08:00
pfile->buffer->sysp = flags;
Linemap infrastructure for virtual locations This is the first instalment of a set which goal is to track locations of tokens across macro expansions. Tom Tromey did the original work and attached the patch to PR preprocessor/7263. This opus is a derivative of that original work. This patch modifies the linemap module of libcpp to add virtual locations support. A virtual location is a mapped location that can resolve to several different physical locations. It can always resolve to the spelling location of a token. For tokens resulting from macro expansion it can resolve to: - either the location of the expansion point of the macro. - or the location of the token in the definition of the macro - or, if the token is an argument of a function-like macro, the location of the use of the matching macro parameter in the definition of the macro The patch creates a new type of line map called a macro map. For every single macro expansion, there is a macro map that generates a virtual location for every single resulting token of the expansion. The good old type of line map we all know is now called an ordinary map. That one still encodes spelling locations as it has always had. As a result linemap_lookup as been extended to return a macro map when given a virtual location resulting from a macro expansion. The layout of structs line_map has changed to support this new type of map. So did the layout of struct line_maps. Accessor macros have been introduced to avoid messing with the implementation details of these datastructures directly. This helped already as we have been testing different ways of arranging these datastructure. Having to constantly adjust client code that is too tied with the internals of line_map and line_maps would have been even more painful. Of course, many new public functions have been added to the linemap module to handle the resolution of virtual locations. This patch introduces the infrastructure but no part of the compiler uses virtual locations yet. However the client code of the linemap data structures has been adjusted as per the changes. E.g, it's not anymore reliable for a client code to manipulate struct line_map directly if it just wants to deal with spelling locations, because struct line_map can now represent a macro map as well. In that case, it's better to use the convenient API to resolve the initial (possibly virtual) location to a spelling location (or to an ordinary map) and use that. This is the reason why the patch adjusts the Java, Ada and Fortran front ends. Also, note that virtual locations are not supposed to be ordered for relations '<' and '>' anymore. To test if a virtual location appears "before" another one, one has to use a new operator exposed by the line map interface. The patch updates the only spot (in the diagnostics module) I have found that was making the assumption that locations were ordered for these relations. This is the only change that introduces a use of the new line map API in this patch, so I am adding a regression test for it only. From-SVN: r180081
2011-10-17 11:58:56 +02:00
_cpp_do_file_change (pfile, LC_RENAME, ORDINARY_MAP_FILE_NAME (map),
SOURCE_LINE (map, pfile->line_table->highest_line),
flags);
}
/* Allow the client to change the current file. Used by the front end
to achieve pseudo-file names like <built-in>.
If REASON is LC_LEAVE, then NEW_NAME must be NULL. */
void
cpp_change_file (cpp_reader *pfile, enum lc_reason reason,
const char *new_name)
{
_cpp_do_file_change (pfile, reason, new_name, 1, 0);
}
struct report_missing_guard_data
{
cpp_reader *pfile;
const char **paths;
size_t count;
};
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Callback function for htab_traverse. */
cpplex.c: Don't include sys/mman.h. toplevel: * cpplex.c: Don't include sys/mman.h. (cpp_push_buffer, cpp_pop_buffer): Moved to cpplib.c. * cpplib.c: Include sys/mman.h and obstack.h. (cpp_push_buffer): Moved from cpplex.c; allocate buffers on an obstack. (cpp_pop_buffer): Moved from cpplex.c; free buffers from an obstack. (_cpp_unwind_if_stack): Now static, unwind_if_stack. Don't bother freeing if stack entries (they will be freed with their buffer). (do_endif): Free if stack entries from the buffer obstack. (push_conditional): Allocate if stack entries from the buffer obstack. (find_answer): Rename to _cpp_find_answer. (do_assert, do_unassert): Update. * cpphash.h: Update prototypes. (xobnew): New convenience macro. * cpplib.h (struct cpp_reader): Add hash_ob and buffer_ob fields. Update comments. (struct cpp_hashnode): Remove disabled field. * cppinit.c: Don't include hashtab.h or splay-tree.h. (report_missing_guard): Moved to cppfiles.c. (cpp_reader_init): Call cpp_init_stacks, cpp_init_macros, cpp_init_includes. (cpp_cleanup): Call cpp_cleanup_stacks, cpp_cleanup_macros, cpp_cleanup_includes. Don't destroy hashtab or all_include_files here. (cpp_finish): Use _cpp_report_missing_guards. * cppfiles.c (report_missing_guard): Moved from cppinit.c. (_cpp_init_include_table): Rename _cpp_init_includes. (_cpp_cleanup_includes, _cpp_report_missing_guards): New. * cppexp.c (parse_assertion): Update for new name of find_answer. * Makefile.in (cpplib.o, cpphash.o, cppinit.o): Update deps. * cpplib.c (do_ident): s/VSPACE/EOF/ testsuite: * gcc.dg/cpp/ident.c: New test. From-SVN: r34870
2000-07-05 05:33:57 +00:00
static int
report_missing_guard (void **slot, void *d)
cpplex.c: Don't include sys/mman.h. toplevel: * cpplex.c: Don't include sys/mman.h. (cpp_push_buffer, cpp_pop_buffer): Moved to cpplib.c. * cpplib.c: Include sys/mman.h and obstack.h. (cpp_push_buffer): Moved from cpplex.c; allocate buffers on an obstack. (cpp_pop_buffer): Moved from cpplex.c; free buffers from an obstack. (_cpp_unwind_if_stack): Now static, unwind_if_stack. Don't bother freeing if stack entries (they will be freed with their buffer). (do_endif): Free if stack entries from the buffer obstack. (push_conditional): Allocate if stack entries from the buffer obstack. (find_answer): Rename to _cpp_find_answer. (do_assert, do_unassert): Update. * cpphash.h: Update prototypes. (xobnew): New convenience macro. * cpplib.h (struct cpp_reader): Add hash_ob and buffer_ob fields. Update comments. (struct cpp_hashnode): Remove disabled field. * cppinit.c: Don't include hashtab.h or splay-tree.h. (report_missing_guard): Moved to cppfiles.c. (cpp_reader_init): Call cpp_init_stacks, cpp_init_macros, cpp_init_includes. (cpp_cleanup): Call cpp_cleanup_stacks, cpp_cleanup_macros, cpp_cleanup_includes. Don't destroy hashtab or all_include_files here. (cpp_finish): Use _cpp_report_missing_guards. * cppfiles.c (report_missing_guard): Moved from cppinit.c. (_cpp_init_include_table): Rename _cpp_init_includes. (_cpp_cleanup_includes, _cpp_report_missing_guards): New. * cppexp.c (parse_assertion): Update for new name of find_answer. * Makefile.in (cpplib.o, cpphash.o, cppinit.o): Update deps. * cpplib.c (do_ident): s/VSPACE/EOF/ testsuite: * gcc.dg/cpp/ident.c: New test. From-SVN: r34870
2000-07-05 05:33:57 +00:00
{
struct cpp_file_hash_entry *entry = (struct cpp_file_hash_entry *) *slot;
struct report_missing_guard_data *data
= (struct report_missing_guard_data *) d;
cpplex.c: Don't include sys/mman.h. toplevel: * cpplex.c: Don't include sys/mman.h. (cpp_push_buffer, cpp_pop_buffer): Moved to cpplib.c. * cpplib.c: Include sys/mman.h and obstack.h. (cpp_push_buffer): Moved from cpplex.c; allocate buffers on an obstack. (cpp_pop_buffer): Moved from cpplex.c; free buffers from an obstack. (_cpp_unwind_if_stack): Now static, unwind_if_stack. Don't bother freeing if stack entries (they will be freed with their buffer). (do_endif): Free if stack entries from the buffer obstack. (push_conditional): Allocate if stack entries from the buffer obstack. (find_answer): Rename to _cpp_find_answer. (do_assert, do_unassert): Update. * cpphash.h: Update prototypes. (xobnew): New convenience macro. * cpplib.h (struct cpp_reader): Add hash_ob and buffer_ob fields. Update comments. (struct cpp_hashnode): Remove disabled field. * cppinit.c: Don't include hashtab.h or splay-tree.h. (report_missing_guard): Moved to cppfiles.c. (cpp_reader_init): Call cpp_init_stacks, cpp_init_macros, cpp_init_includes. (cpp_cleanup): Call cpp_cleanup_stacks, cpp_cleanup_macros, cpp_cleanup_includes. Don't destroy hashtab or all_include_files here. (cpp_finish): Use _cpp_report_missing_guards. * cppfiles.c (report_missing_guard): Moved from cppinit.c. (_cpp_init_include_table): Rename _cpp_init_includes. (_cpp_cleanup_includes, _cpp_report_missing_guards): New. * cppexp.c (parse_assertion): Update for new name of find_answer. * Makefile.in (cpplib.o, cpphash.o, cppinit.o): Update deps. * cpplib.c (do_ident): s/VSPACE/EOF/ testsuite: * gcc.dg/cpp/ident.c: New test. From-SVN: r34870
2000-07-05 05:33:57 +00:00
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Skip directories. */
if (entry->start_dir != NULL)
cpplex.c: Don't include sys/mman.h. toplevel: * cpplex.c: Don't include sys/mman.h. (cpp_push_buffer, cpp_pop_buffer): Moved to cpplib.c. * cpplib.c: Include sys/mman.h and obstack.h. (cpp_push_buffer): Moved from cpplex.c; allocate buffers on an obstack. (cpp_pop_buffer): Moved from cpplex.c; free buffers from an obstack. (_cpp_unwind_if_stack): Now static, unwind_if_stack. Don't bother freeing if stack entries (they will be freed with their buffer). (do_endif): Free if stack entries from the buffer obstack. (push_conditional): Allocate if stack entries from the buffer obstack. (find_answer): Rename to _cpp_find_answer. (do_assert, do_unassert): Update. * cpphash.h: Update prototypes. (xobnew): New convenience macro. * cpplib.h (struct cpp_reader): Add hash_ob and buffer_ob fields. Update comments. (struct cpp_hashnode): Remove disabled field. * cppinit.c: Don't include hashtab.h or splay-tree.h. (report_missing_guard): Moved to cppfiles.c. (cpp_reader_init): Call cpp_init_stacks, cpp_init_macros, cpp_init_includes. (cpp_cleanup): Call cpp_cleanup_stacks, cpp_cleanup_macros, cpp_cleanup_includes. Don't destroy hashtab or all_include_files here. (cpp_finish): Use _cpp_report_missing_guards. * cppfiles.c (report_missing_guard): Moved from cppinit.c. (_cpp_init_include_table): Rename _cpp_init_includes. (_cpp_cleanup_includes, _cpp_report_missing_guards): New. * cppexp.c (parse_assertion): Update for new name of find_answer. * Makefile.in (cpplib.o, cpphash.o, cppinit.o): Update deps. * cpplib.c (do_ident): s/VSPACE/EOF/ testsuite: * gcc.dg/cpp/ident.c: New test. From-SVN: r34870
2000-07-05 05:33:57 +00:00
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
_cpp_file *file = entry->u.file;
/* We don't want MI guard advice for the main file. */
if (!file->once_only
&& file->cmacro == NULL
&& file->stack_count == 1
&& data->pfile->main_file != file)
cpplex.c: Don't include sys/mman.h. toplevel: * cpplex.c: Don't include sys/mman.h. (cpp_push_buffer, cpp_pop_buffer): Moved to cpplib.c. * cpplib.c: Include sys/mman.h and obstack.h. (cpp_push_buffer): Moved from cpplex.c; allocate buffers on an obstack. (cpp_pop_buffer): Moved from cpplex.c; free buffers from an obstack. (_cpp_unwind_if_stack): Now static, unwind_if_stack. Don't bother freeing if stack entries (they will be freed with their buffer). (do_endif): Free if stack entries from the buffer obstack. (push_conditional): Allocate if stack entries from the buffer obstack. (find_answer): Rename to _cpp_find_answer. (do_assert, do_unassert): Update. * cpphash.h: Update prototypes. (xobnew): New convenience macro. * cpplib.h (struct cpp_reader): Add hash_ob and buffer_ob fields. Update comments. (struct cpp_hashnode): Remove disabled field. * cppinit.c: Don't include hashtab.h or splay-tree.h. (report_missing_guard): Moved to cppfiles.c. (cpp_reader_init): Call cpp_init_stacks, cpp_init_macros, cpp_init_includes. (cpp_cleanup): Call cpp_cleanup_stacks, cpp_cleanup_macros, cpp_cleanup_includes. Don't destroy hashtab or all_include_files here. (cpp_finish): Use _cpp_report_missing_guards. * cppfiles.c (report_missing_guard): Moved from cppinit.c. (_cpp_init_include_table): Rename _cpp_init_includes. (_cpp_cleanup_includes, _cpp_report_missing_guards): New. * cppexp.c (parse_assertion): Update for new name of find_answer. * Makefile.in (cpplib.o, cpphash.o, cppinit.o): Update deps. * cpplib.c (do_ident): s/VSPACE/EOF/ testsuite: * gcc.dg/cpp/ident.c: New test. From-SVN: r34870
2000-07-05 05:33:57 +00:00
{
if (data->paths == NULL)
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
{
data->paths = XCNEWVEC (const char *, data->count);
data->count = 0;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
}
data->paths[data->count++] = file->path;
cpplex.c: Don't include sys/mman.h. toplevel: * cpplex.c: Don't include sys/mman.h. (cpp_push_buffer, cpp_pop_buffer): Moved to cpplib.c. * cpplib.c: Include sys/mman.h and obstack.h. (cpp_push_buffer): Moved from cpplex.c; allocate buffers on an obstack. (cpp_pop_buffer): Moved from cpplex.c; free buffers from an obstack. (_cpp_unwind_if_stack): Now static, unwind_if_stack. Don't bother freeing if stack entries (they will be freed with their buffer). (do_endif): Free if stack entries from the buffer obstack. (push_conditional): Allocate if stack entries from the buffer obstack. (find_answer): Rename to _cpp_find_answer. (do_assert, do_unassert): Update. * cpphash.h: Update prototypes. (xobnew): New convenience macro. * cpplib.h (struct cpp_reader): Add hash_ob and buffer_ob fields. Update comments. (struct cpp_hashnode): Remove disabled field. * cppinit.c: Don't include hashtab.h or splay-tree.h. (report_missing_guard): Moved to cppfiles.c. (cpp_reader_init): Call cpp_init_stacks, cpp_init_macros, cpp_init_includes. (cpp_cleanup): Call cpp_cleanup_stacks, cpp_cleanup_macros, cpp_cleanup_includes. Don't destroy hashtab or all_include_files here. (cpp_finish): Use _cpp_report_missing_guards. * cppfiles.c (report_missing_guard): Moved from cppinit.c. (_cpp_init_include_table): Rename _cpp_init_includes. (_cpp_cleanup_includes, _cpp_report_missing_guards): New. * cppexp.c (parse_assertion): Update for new name of find_answer. * Makefile.in (cpplib.o, cpphash.o, cppinit.o): Update deps. * cpplib.c (do_ident): s/VSPACE/EOF/ testsuite: * gcc.dg/cpp/ident.c: New test. From-SVN: r34870
2000-07-05 05:33:57 +00:00
}
}
/* Keep traversing the hash table. */
return 1;
}
/* Comparison function for qsort. */
static int
report_missing_guard_cmp (const void *p1, const void *p2)
{
return strcmp (*(const char *const *) p1, *(const char *const *) p2);
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Report on all files that might benefit from a multiple include guard.
Triggered by -H. */
void
_cpp_report_missing_guards (cpp_reader *pfile)
cppfiles.c: Include mkdeps.h. * cppfiles.c: Include mkdeps.h. (find_include_file, read_include_file): Remove _cpp_ prefix from name, make static. (_cpp_execute_include): New function, broken out of do_include. * cpplib.c: Don't include mkdeps.h. (struct directive): Remove type field. Reorder entries. The function takes only one argument. (struct if_stack): Make type field an int. (directive_table): Rename to dtable. Generate it, the prototypes of the directive handlers, and the enum for the directive numbers, from a template macro. (do_ifndef, do_include_next, do_import): New functions. (do_define, do_include, do_endif, do_ifdef, do_if, do_else, do_undef, do_line, do_elif, do_error, do_pragma, do_warning, do_ident, do_assert, do_unassert, do_sccs): Take only one argument. (do_sccs): Define always, but alter behavior based on SCCS_DIRECTIVE. (_cpp_handle_directive, consider_directive_while_skipping): Restructure for new directive table layout. (pass_thru_directive): Take a directive number, not a pointer to a struct directive. (parse_include): New function, broken out of do_include. (do_include, do_import, do_include_next): Use parse_include and _cpp_execute_include. (do_elif, do_else): Test for T_ELSE specifically when checking for #elif/#else after #else. (parse_ifdef): New function, broken out of do_ifdef. (validate_else): Expect a name arg without a leading #. (if_directive_name): Delete. (cpp_define, cpp_assert, cpp_undef, cpp_unassert): Call directive handlers with only one argument. * cpphash.h: Update prototypes. (enum node_type): Remove entries for directives. * Makefile.in: Update dependencies. From-SVN: r32519
2000-03-14 06:34:11 +00:00
{
struct report_missing_guard_data data;
data.pfile = pfile;
data.paths = NULL;
data.count = htab_elements (pfile->file_hash);
htab_traverse (pfile->file_hash, report_missing_guard, &data);
if (data.paths != NULL)
{
size_t i;
/* Sort the paths to avoid outputting them in hash table
order. */
qsort (data.paths, data.count, sizeof (const char *),
report_missing_guard_cmp);
fputs (_("Multiple include guards may be useful for:\n"),
stderr);
for (i = 0; i < data.count; i++)
{
fputs (data.paths[i], stderr);
putc ('\n', stderr);
}
free (data.paths);
}
cppfiles.c: Include mkdeps.h. * cppfiles.c: Include mkdeps.h. (find_include_file, read_include_file): Remove _cpp_ prefix from name, make static. (_cpp_execute_include): New function, broken out of do_include. * cpplib.c: Don't include mkdeps.h. (struct directive): Remove type field. Reorder entries. The function takes only one argument. (struct if_stack): Make type field an int. (directive_table): Rename to dtable. Generate it, the prototypes of the directive handlers, and the enum for the directive numbers, from a template macro. (do_ifndef, do_include_next, do_import): New functions. (do_define, do_include, do_endif, do_ifdef, do_if, do_else, do_undef, do_line, do_elif, do_error, do_pragma, do_warning, do_ident, do_assert, do_unassert, do_sccs): Take only one argument. (do_sccs): Define always, but alter behavior based on SCCS_DIRECTIVE. (_cpp_handle_directive, consider_directive_while_skipping): Restructure for new directive table layout. (pass_thru_directive): Take a directive number, not a pointer to a struct directive. (parse_include): New function, broken out of do_include. (do_include, do_import, do_include_next): Use parse_include and _cpp_execute_include. (do_elif, do_else): Test for T_ELSE specifically when checking for #elif/#else after #else. (parse_ifdef): New function, broken out of do_ifdef. (validate_else): Expect a name arg without a leading #. (if_directive_name): Delete. (cpp_define, cpp_assert, cpp_undef, cpp_unassert): Call directive handlers with only one argument. * cpphash.h: Update prototypes. (enum node_type): Remove entries for directives. * Makefile.in: Update dependencies. From-SVN: r32519
2000-03-14 06:34:11 +00:00
}
/* Locate HEADER, and determine whether it is newer than the current
file. If it cannot be located or dated, return -1, if it is
newer, return 1, otherwise 0. */
int
_cpp_compare_file_date (cpp_reader *pfile, const char *fname,
int angle_brackets)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
_cpp_file *file;
struct cpp_dir *dir;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
dir = search_path_head (pfile, fname, angle_brackets, IT_INCLUDE);
if (!dir)
return -1;
file = _cpp_find_file (pfile, fname, dir, angle_brackets, _cpp_FFK_NORMAL, 0);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (file->err_no)
return -1;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (file->fd != -1)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
close (file->fd);
file->fd = -1;
}
top level: 2000-07-03 Zack Weinberg <zack@wolery.cumb.org> * fix-header.c (struct partial_proto): Remove unnecessary fields. (recognized_extern, recognized_function, read_scan_file): Update for new scheme. (check_protection): It's still a multiple include guard even if it doesn't always trigger. * scan-decls.c (skip_to_closing_brace, scan_decls): Update for new scheme. * scan.h: Declare struct cpp_token. Update prototypes. 2000-07-03 Neil Booth <neilb@earthling.net> Zack Weinberg <zack@wolery.cumb.org> Complete overhaul of the lexer and macro expander. * cpphash.c (object_defn, funct_defn, push_macro_expansion, arg, arglist, argdata, reflist, collect_objlike_expansion, collect_funlike_expansion, collect_params, warn_trad_stringify, trad_stringify, duplicate_arg_p, add_pat, unsafe_chars, macarg, compare_defs, special_symbol, scan_arguments, stringify, funlike_macroexpand, _cpp_quote_string, monthnames): Delete. (cpp_lookup, _cpp_free_definition, dump_funlike_macro, _cpp_create_definition, _cpp_dump_definition, dump_hash_helper): Adjust. (find_param, count_params, parse_define, var_args_str, check_macro_redefinition, save_expansion): New. * cpplex.c (skip_block_comment, skip_line_comment, parse_name, parse_string, output_line_command, trigraph_replace, lex_line, cpp_push_buffer, cpp_pop_buffer, cpp_output_tokens, cpp_scan_buffer_nooutput, cpp_scan_buffer, cpp_free_toklist, cpp_idcmp, _cpp_get_directive_token, _cpp_init_input_buffer, _cpp_skip_rest_of_line): Modify. (maybe_macroexpand, skip_comment, copy_comment, skip_string, find_position, null_warning, bump_column, expand_name_space, pedantic_whitespace, _cpp_output_list, _cpp_slice_toklist, _cpp_squeeze_toklist, _cpp_scan_until, _cpp_skip_hspace, _cpp_parse_name, _cpp_lex_token, cpp_get_non_space_token, _cpp_prescan): Delete. (dump_param_spelling, process_directive, lex_next, is_macro_disabled, stringify_arg, expand_context_stack, output_token, make_string_token, alloc_number_token, special_symbol, duplicate_token, maybe_paste_with_next, can_paste, prevent_macro_expansion, restore_macro_expansion, get_temp_token, release_temp_tokens, quote_string, token_names, token_spellings, _cpp_expand_name_space, _cpp_glue_header_name, _cpp_reserve_name_space, digraph_spellings, trigraph_ok, skip_whitespace, save_comment, placemarker_token, eof_token, cpp_context, macro_args, get_raw_token, parse_arg, parse_args, save_token, push_arg_context, push_macro_context, pop_context, do_pop_context, free_macro_args, _cpp_get_line, _cpp_run_directive): New. * cpplib.c (validate_else, parse_include, push_conditional, pass_thru_directive, read_line_number, parse_ifdef, detect_if_not_defined, _cpp_check_directive, do_define, do_undef, do_include, do_import, do_include_next, do_error, do_warning, do_ident, do_pragma, pragma_dispatch, gcc_pragmas, top_pragmas, do_pragma_gcc, do_pragma_implementation, do_pragma_poison, do_pragma_system_header, do_pragma_dependency, do_sccs, do_ifdef, do_ifndef, do_else, dl_elif, do_endif, _cpp_unwind_if_stack, do_assert, do_unassert, cpp_define, cpp_undef, cpp_assert, cpp_unassert, cpp_defined): Update for new scheme. (strtoul_for_line, get_define_node, dump_macro_name, _cpp_check_linemarker, _cpp_parse_assertion): New. (_cpp_handle_directive, do_pragma_default): Delete. * cpphash.h (struct predicate): Now struct answer. (enum spell_type, struct token_spelling, struct directive, directive_handler): New. Update prototypes. Remove unused macros. * cpplib.h: Update prototypes. Remove unused macros, structure definitions, and fields. * cpperror.c (print_containing_files, v_message): Adjust. * cppexp.c (parse_assertion, lex, parse_escape, _cpp_parse_expr): Adjust. * cppfiles.c (open_include_file, _cpp_execute_include, _cpp_compare_file_date, cpp_read_file, read_include_file): Adjust. * cppinit.c (dump_special_to_buffer): Delete. (append_include_chain, merge_include_chains, cpp_reader_init, cpp_cleanup, initialize_builtins, builtin_array, cpp_start_read, cpp_finish, handle_option, print_help): Adjust. * cppmain.c (main): Adjust. testsuite: 2000-07-03 Zack Weinberg <zack@wolery.cumb.org> * testsuite/gcc.dg/cpp/19951025-1.c: Adjust regexps. * testsuite/gcc.dg/cpp/19990703-1.c: Likewise. * testsuite/gcc.dg/cpp/20000625-1.c: Likewise. * testsuite/gcc.dg/cpp/20000625-2.c: Likewise. * testsuite/gcc.dg/cpp/macro1.c, testsuite/gcc.dg/cpp/paste1.c, testsuite/gcc.dg/cpp/paste2.c, testsuite/gcc.dg/cpp/paste3.c, testsuite/gcc.dg/cpp/paste4.c, testsuite/gcc.dg/cpp/strify1.c, testsuite/gcc.dg/cpp/strify2.c: New tests. From-SVN: r34859
2000-07-04 01:58:21 +00:00
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
return file->st.st_mtime > pfile->buffer->file->st.st_mtime;
}
/* Pushes the given file onto the buffer stack. Returns nonzero if
successful. */
bool
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
cpp_push_include (cpp_reader *pfile, const char *fname)
{
preprocessor: Better line info for <builtin> & <command-line> With C++ module header units it becomes important to distinguish between macros defined in forced headers (& commandline & builtins) from those defined in the header file being processed. We weren't making that easy because we treated the builtins and command-line locations somewhat file-like, with incrementing line numbers, and showing them as included from line 1 of the main file. This patch does 3 things: 0) extend the idiom that 'line 0' of a file means 'the file as a whole' 1) builtins and command-line macros are shown as-if included from line zero. 2) when emitting preprocessed output we keep resetting the line number so that re-reading that preprocessed output will get the same set of locations for the command line etc. For instance the new c-c++-common/cpp/line-2.c test, now emits In file included from <command-line>: ./line-2.h:4:2: error: #error wrong 4 | #error wrong | ^~~~~ line-2.c:3:11: error: macro "bill" passed 1 arguments, but takes just 0 3 | int bill(1); | ^ In file included from <command-line>: ./line-2.h:3: note: macro "bill" defined here 3 | #define bill() 2 | Before it told you about including from <command-line>:31. the preprocessed output looks like: ... (There's a new optimization in do_line_marker to stop each of these line markers causing a new line map. We can simply rewind the location, and keep using the same line map.) libcpp/ * directives.c (do_linemarker): Optimize rewinding to line zero. * files.c (_cpp_stack_file): Start on line zero when about to inject headers. (cpp_push_include, cpp_push_default_include): Use highest_line as the location. * include/cpplib.h (cpp_read_main_file): Add injecting parm. * init.c (cpp_read_main_file): Likewise, inform _cpp_stack_file. * internal.h (enum include_type): Add IT_MAIN_INJECT. gcc/c-family/ * c-opts.c (c_common_post_options): Add 'injecting' arg to cpp_read_main_file. (c_finish_options): Add linemap_line_start calls for builtin and cmd maps. Force token position to line_table's highest line. * c-ppoutput.c (print_line_1): Refactor, print line zero. (cb_define): Always increment source line. gcc/testsuite/ * c-c++-common/cpp/line-2.c: New. * c-c++-common/cpp/line-2.h: New. * c-c++-common/cpp/line-3.c: New. * c-c++-common/cpp/line-4.c: New. * c-c++-common/cpp/line-4.h: New.
2020-07-07 11:28:59 -07:00
return _cpp_stack_include (pfile, fname, false, IT_CMDLINE,
pfile->line_table->highest_line);
}
gcc: * config.gcc (*-*-linux* | frv-*-*linux* | *-*-kfreebsd*-gnu | *-*-knetbsd*-gnu | *-*-gnu* | *-*-kopensolaris*-gnu): Use glibc-c.o in c_target_objs and cxx_target_objs. Use t-glibc in tmake_file. Set target_has_targetcm. (tilegx-*-linux*, tilepro-*-linux*): Append to c_target_objs and cxx_target_objs rather than overriding previous value. * config/glibc-c.c, config/t-glibc: New. * doc/tm.texi.in (TARGET_C_PREINCLUDE): New @hook. * doc/tm.texi: Regenerate. * hooks.c (hook_constcharptr_void_null): New. * hooks.h (hook_constcharptr_void_null): Declare. gcc/c-family: * c-common.h (pch_cpp_save_state): Declare. * c-target.def (c_preinclude): New hook. * c-opts.c (done_preinclude): New. (push_command_line_include): Handle default preincluded header. (cb_file_change): Call pch_cpp_save_state when calling push_command_line_include. * c-pch.c (pch_ready_to_save_cpp_state, pch_cpp_state_saved) (pch_cpp_save_state): New. (pch_init): Call pch_cpp_save_state conditionally, instead of calling cpp_save_state. gcc/testsuite: * gcc.dg/c99-predef-1.c: New test. * gcc.dg/cpp/cmdlne-dU-1.c, gcc.dg/cpp/cmdlne-dU-2.c, gcc.dg/cpp/cmdlne-dU-3.c, gcc.dg/cpp/cmdlne-dU-4.c, gcc.dg/cpp/cmdlne-dU-5.c, gcc.dg/cpp/cmdlne-dU-6.c, gcc.dg/cpp/cmdlne-dU-7.c, gcc.dg/cpp/cmdlne-dU-8.c, gcc.dg/cpp/cmdlne-dU-9.c, gcc.dg/cpp/cmdlne-dU-10.c, gcc.dg/cpp/cmdlne-dU-11.c, gcc.dg/cpp/cmdlne-dU-12.c, gcc.dg/cpp/cmdlne-dU-13.c, gcc.dg/cpp/cmdlne-dU-14.c, gcc.dg/cpp/cmdlne-dU-15.c, gcc.dg/cpp/cmdlne-dU-16.c, gcc.dg/cpp/cmdlne-dU-17.c, gcc.dg/cpp/cmdlne-dU-18.c, gcc.dg/cpp/cmdlne-dU-19.c, gcc.dg/cpp/cmdlne-dU-20.c, gcc.dg/cpp/cmdlne-dU-21.c, gcc.dg/cpp/cmdlne-dU-22.c, gcc.dg/cpp/mi5.c, gcc.dg/cpp/multiline.c: Add -nostdinc to dg-options. libcpp: * files.c (struct _cpp_file): Add implicit_preinclude. (pch_open_file): Allow a previously opened implicitly included file. (_cpp_find_file): Add implicit_preinclude argument. Free file and do not call open_file_failed if implicit_preinclude. Store implicit_preinclude value. (_cpp_stack_include, _cpp_fake_include, _cpp_compare_file_date): Update calls to _cpp_find_file. (_cpp_stack_include): Handle IT_DEFAULT. (cpp_push_default_include): New. * include/cpplib.h (cpp_push_default_include): Declare. * init.c (cpp_read_main_file): Update call to _cpp_find_file. * internal.h (enum include_type): Add IT_DEFAULT. (_cpp_find_file): Update prototype. From-SVN: r192715
2012-10-23 15:55:55 +01:00
/* Pushes the given file, implicitly included at the start of a
compilation, onto the buffer stack but without any errors if the
file is not found. Returns nonzero if successful. */
bool
cpp_push_default_include (cpp_reader *pfile, const char *fname)
{
preprocessor: Better line info for <builtin> & <command-line> With C++ module header units it becomes important to distinguish between macros defined in forced headers (& commandline & builtins) from those defined in the header file being processed. We weren't making that easy because we treated the builtins and command-line locations somewhat file-like, with incrementing line numbers, and showing them as included from line 1 of the main file. This patch does 3 things: 0) extend the idiom that 'line 0' of a file means 'the file as a whole' 1) builtins and command-line macros are shown as-if included from line zero. 2) when emitting preprocessed output we keep resetting the line number so that re-reading that preprocessed output will get the same set of locations for the command line etc. For instance the new c-c++-common/cpp/line-2.c test, now emits In file included from <command-line>: ./line-2.h:4:2: error: #error wrong 4 | #error wrong | ^~~~~ line-2.c:3:11: error: macro "bill" passed 1 arguments, but takes just 0 3 | int bill(1); | ^ In file included from <command-line>: ./line-2.h:3: note: macro "bill" defined here 3 | #define bill() 2 | Before it told you about including from <command-line>:31. the preprocessed output looks like: ... (There's a new optimization in do_line_marker to stop each of these line markers causing a new line map. We can simply rewind the location, and keep using the same line map.) libcpp/ * directives.c (do_linemarker): Optimize rewinding to line zero. * files.c (_cpp_stack_file): Start on line zero when about to inject headers. (cpp_push_include, cpp_push_default_include): Use highest_line as the location. * include/cpplib.h (cpp_read_main_file): Add injecting parm. * init.c (cpp_read_main_file): Likewise, inform _cpp_stack_file. * internal.h (enum include_type): Add IT_MAIN_INJECT. gcc/c-family/ * c-opts.c (c_common_post_options): Add 'injecting' arg to cpp_read_main_file. (c_finish_options): Add linemap_line_start calls for builtin and cmd maps. Force token position to line_table's highest line. * c-ppoutput.c (print_line_1): Refactor, print line zero. (cb_define): Always increment source line. gcc/testsuite/ * c-c++-common/cpp/line-2.c: New. * c-c++-common/cpp/line-2.h: New. * c-c++-common/cpp/line-3.c: New. * c-c++-common/cpp/line-4.c: New. * c-c++-common/cpp/line-4.h: New.
2020-07-07 11:28:59 -07:00
return _cpp_stack_include (pfile, fname, true, IT_DEFAULT,
pfile->line_table->highest_line);
gcc: * config.gcc (*-*-linux* | frv-*-*linux* | *-*-kfreebsd*-gnu | *-*-knetbsd*-gnu | *-*-gnu* | *-*-kopensolaris*-gnu): Use glibc-c.o in c_target_objs and cxx_target_objs. Use t-glibc in tmake_file. Set target_has_targetcm. (tilegx-*-linux*, tilepro-*-linux*): Append to c_target_objs and cxx_target_objs rather than overriding previous value. * config/glibc-c.c, config/t-glibc: New. * doc/tm.texi.in (TARGET_C_PREINCLUDE): New @hook. * doc/tm.texi: Regenerate. * hooks.c (hook_constcharptr_void_null): New. * hooks.h (hook_constcharptr_void_null): Declare. gcc/c-family: * c-common.h (pch_cpp_save_state): Declare. * c-target.def (c_preinclude): New hook. * c-opts.c (done_preinclude): New. (push_command_line_include): Handle default preincluded header. (cb_file_change): Call pch_cpp_save_state when calling push_command_line_include. * c-pch.c (pch_ready_to_save_cpp_state, pch_cpp_state_saved) (pch_cpp_save_state): New. (pch_init): Call pch_cpp_save_state conditionally, instead of calling cpp_save_state. gcc/testsuite: * gcc.dg/c99-predef-1.c: New test. * gcc.dg/cpp/cmdlne-dU-1.c, gcc.dg/cpp/cmdlne-dU-2.c, gcc.dg/cpp/cmdlne-dU-3.c, gcc.dg/cpp/cmdlne-dU-4.c, gcc.dg/cpp/cmdlne-dU-5.c, gcc.dg/cpp/cmdlne-dU-6.c, gcc.dg/cpp/cmdlne-dU-7.c, gcc.dg/cpp/cmdlne-dU-8.c, gcc.dg/cpp/cmdlne-dU-9.c, gcc.dg/cpp/cmdlne-dU-10.c, gcc.dg/cpp/cmdlne-dU-11.c, gcc.dg/cpp/cmdlne-dU-12.c, gcc.dg/cpp/cmdlne-dU-13.c, gcc.dg/cpp/cmdlne-dU-14.c, gcc.dg/cpp/cmdlne-dU-15.c, gcc.dg/cpp/cmdlne-dU-16.c, gcc.dg/cpp/cmdlne-dU-17.c, gcc.dg/cpp/cmdlne-dU-18.c, gcc.dg/cpp/cmdlne-dU-19.c, gcc.dg/cpp/cmdlne-dU-20.c, gcc.dg/cpp/cmdlne-dU-21.c, gcc.dg/cpp/cmdlne-dU-22.c, gcc.dg/cpp/mi5.c, gcc.dg/cpp/multiline.c: Add -nostdinc to dg-options. libcpp: * files.c (struct _cpp_file): Add implicit_preinclude. (pch_open_file): Allow a previously opened implicitly included file. (_cpp_find_file): Add implicit_preinclude argument. Free file and do not call open_file_failed if implicit_preinclude. Store implicit_preinclude value. (_cpp_stack_include, _cpp_fake_include, _cpp_compare_file_date): Update calls to _cpp_find_file. (_cpp_stack_include): Handle IT_DEFAULT. (cpp_push_default_include): New. * include/cpplib.h (cpp_push_default_include): Declare. * init.c (cpp_read_main_file): Update call to _cpp_find_file. * internal.h (enum include_type): Add IT_DEFAULT. (_cpp_find_file): Update prototype. From-SVN: r192715
2012-10-23 15:55:55 +01:00
}
/* Do appropriate cleanup when a file INC's buffer is popped off the
input stack. */
void
_cpp_pop_file_buffer (cpp_reader *pfile, _cpp_file *file,
const unsigned char *to_free)
{
/* Record the inclusion-preventing macro, which could be NULL
meaning no controlling macro. */
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (pfile->mi_valid && file->cmacro == NULL)
file->cmacro = pfile->mi_cmacro;
New macro expander. 2000-10-28 Neil Booth <neilb@earthling.net> New macro expander. * cpplib.c (struct answer): New. (struct if_stack): Use cpp_lexer_pos rather than line and col. Rename cmacro mi_cmacro. (struct directive, KANDR, STDC89, EXTENSION, COND, IF_COND, INCL, IN_I): New directive and flags. (skip_rest_of_line, check_eol, run_directive, glue_header_name, parse_answer, parse_assertion, find_answer): New functions. (parse_ifdef, detect_if_not_defined, validate_else): Remove. (lex_macro_node): New function to replace parse_ifdef and get_define_node. (_cpp_handle_directive): New function, combines _cpp_check_directive and _cpp_check_linemarker. (do_define, do_undef, parse_include, do_include, do_import, do_include_next, read_line_number, do_line, do_ident, do_pragma, do_pragma_once, do_pragma_poison, do_pragma_dependency): Update for new token getting interface. (do_ifdef, do_ifndef, do_if, do_else, do_endif, push_conditional) : Update for new multiple-include optimisation technique. (do_elif): Don't forget to invalidate controlling macros. (unwind_if_stack, cpp_defined, cpp_push_buffer, cpp_pop_buffer): Update. (parse_assertion, parse_answer, find_answer, _cpp_test_assertion): Functions to handle assertions with the new token interface. (do_assert, do_unassert): Use them. (cpp_define, _cpp_define_builtin, cpp_undef, cpp_assert, cpp_unassert): Use run_directive. (_cpp_init_stacks): Register directive names. Don't register special nodes. * cpperror.c (print_containing_files, _cpp_begin_message): Update to new position recording regime. (cpp_ice, cpp_fatal, cpp_error, cpp_error_with_line, cpp_warning, cpp_warning_with_line, cpp_pedwarn, cpp_pedwarn_with_line, cpp_pedwarn_with_file_and_line): Update for _cpp_begin_message changes. (cpp_type2name): Move to cpplex.c. * cppexp.c (parse_charconst): spec_nodes is no longer a pointer. (parse_defined): Update to handle new multiple include optimisation method. Remove poisoned identifier warning. (parse_assertion, TYPE_NAME): Delete. (lex): Update for multiple include optimisation, removal of CPP_DEFINED, to use _cpp_test_assertion for assertions and cpp_token_as_text. (_cpp_parse_expr): Update for MI optimisation, and to use op_as_text. (op_as_text): New function, to wrap cpp_token_as_text. * cppfiles.c (stack_include_file, _cpp_pop_file_buffer): Update for MI optimisation. (_cpp_execute_include): Take a token rather than 3 arguments. Fix segfault on diagnostic. (_cpp_compare_file_date): Take a token rather than 3 args. (cpp_read_file): Work correctly for zero-length files. * cpphash.c (_cpp_init_macros, _cpp_cleanup_macros): Rename _cpp_init_hashtable and _cpp_cleanup_hashtable. (cpp_lookup): Place identifiers at front of identifier pool for _cpp_lookup_with_hash. (_cpp_lookup_with_hash): Require identifiers to be at the front of the identifier pool. Commit the memory if not already in the hash table. * cppinit.c (cpp_reader_init): Move cpp_init_completed test to top. Initialise various members of cpp_reader, memory pools, and the special nodes. (cpp_printer_init): Delete. (cpp_cleanup): Update. (struct builtin, builtin_array, initialize_builtins): Update for new hashnode definition and builtin handling. (cpp_start_read, cpp_finish): Don't take or initialise a printer. Update. * cpplib.h (cpp_printer, cpp_toklist, CPP_DEFINED, BOL, PASTED, VAR_ARGS, BEG_OF_FILE, IN_DIRECTIVE, KNOWN_DIRECTIVE, T_VOID, T_SPECLINE, T_DATE, T_FILE, T_BASE_FILE, T_INCLUDE_LEVEL, T_TIME, T_STDC, T_OPERATOR, T_POISON, T_MACRO, T_ASSERTION): Delete. (struct cpp_pool, struct cpp_macro, struct cpp_lexer_pos, struct cpp_lookahead, CPP_DHASH, enum mi_state, enum mi_ind, NO_EXPAND, VARARGS_FIRST, struct cpp_token_with_pos, struct toklist, struct cpp_context, struct specnodes, TOKEN_LOOKAHEAD, TOKEN_BUFFSIZE, NODE_OPERATOR, NODE_POISONED, NODE_BUILTIN, NODE_DIAGNOSTIC, NT_VOID, NT_MACRO, NT_ASSERTION, enum builtin_type, cpp_can_paste): New. (struct cpp_token): Delete line and col members. (struct cpp_buffer): New member output_lineno. (struct lexer_state): Delete indented, in_lex_line, seen_dot. Add va_args_ok, poisoned_ok, prevent_expansion, parsing_args. (struct cpp_reader): New members lexer_pos, macro_pos, directive_pos, ident_pool, temp_string_pool, macro_pool, argument_pool, string_pool, base_context, context, directive, mi_state, mi_if_not_defined, mi_lexed, mi_cmacro, mi_ind_cmacro, la_read, la_write, la_unused, mlstring_pos, macro_buffer, macro_buffer_len. Delete members mls_line, mls_column, token_list, potential_control_macro, temp_tokens, temp_cap, temp_alloced, temp_used, first_directive_token, context_cap, cur_context, no_expand_level, paste_level, contexts, args, save_parameter_spellings, need_newline, . Change type of date, time and spec_nodes members. Change prototypes for include and ident callbacks. (struct cpp_hashnode): Change type of name. Remove union members expansion and code. Add members macro, operator and builtin. (cpp_token_len, cpp_token_as_text, cpp_spell_token, cpp_start_read, cpp_finish, cpp_avoid_paste, cpp_get_token, cpp_get_line, cpp_get_output_line, cpp_macro_definition, cpp_start_lookahead, cpp_stop_lookahead): New prototypes. (cpp_printer_init, cpp_dump_definition): Delete prototypes. (U_CHAR, U, ustrcmp, ustrncmp, ustrlen, uxstrdup, ustrchr, ufputs): Move from cpphash.h. * cpphash.h (U_CHAR, U, ustrcmp, ustrncmp, ustrlen, uxstrdup, ustrchr, ufputs): Move to cpplib.h. (enum spell_type, struct token_spelling, _cpp_token_spellings, TOKEN_SPELL, TOKEN_NAME, struct answer, FREE_ANSWER, KANDR, STDC89, EXTENSION, COND, EXPAND, INCL, COMMENTS, IN_I, struct directive, directive_handler, struct spec_nodes, _cpp_digraph_spellings, _cpp_free_temp_tokens, _cpp_init_input_buffer, _cpp_grow_token_buffer, _cpp_init_toklist, _cpp_clear_toklist, _cpp_expand_token_space, _cpp_expand_name_space, _cpp_equiv_tokens, _cpp_equiv_toklists, _cpp_process_directive, _cpp_run_directive, _cpp_get_line, _cpp_get_raw_token, _cpp_glue_header_name, _cpp_can_paste, _cpp_check_directive, _cpp_check_linemarker, _cpp_parse_assertion, _cpp_find_answer): Delete. (VALID_SIGN, ALIGN, POOL_FRONT, POOL_LIMIT, POOL_BASE, POOL_SIZE, POOL_USED, POOL_COMMIT, struct cpp_chunk, _cpp_lex_token, _cpp_init_pool, _cpp_free_pool, _cpp_pool_reserve, _cpp_pool_alloc, _cpp_next_chunk, _cpp_lock_pool, _cpp_unlock_pool, _cpp_test_assertion, _cpp_handle_directive, DSC): New. (struct include_file): New member defined. (DO_NOT_REREAD, _cpp_begin_message, _cpp_execute_include, _cpp_compare_file_date): Update. (_cpp_pop_context, _cpp_get_token, _cpp_free_lookaheads, _cpp_push_token): New. (_cpp_init_macros, _cpp_cleanup_macros): Rename to _cpp_init_hashtable, _cpp_cleanup_hashtable. * Makefile.in: Remove cppoutput.c. * cppoutput.c: Delete * fixheader.c (read_scan_file): Update for new cpp_get_token prototype. (recognized_function): New argument LINE. * scan-decls.c (skip_to_closing_brace, scan_decls): Update for new cpp_get_token prototype. * scan.h (recognized_function): Update prototype. * po/POTFILES.in: Remove cppoutput.c. From-SVN: r37098
2000-10-28 17:59:06 +00:00
/* Invalidate control macros in the #including file. */
pfile->mi_valid = false;
if (to_free)
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
{
if (to_free == file->buffer_start)
{
file->buffer_start = NULL;
file->buffer = NULL;
file->buffer_valid = false;
}
free ((void *) to_free);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
}
}
/* Return the file name associated with FILE. */
const char *
_cpp_get_file_name (_cpp_file *file)
{
return file->name;
}
/* Inteface to file statistics record in _cpp_file structure. */
struct stat *
_cpp_get_file_stat (_cpp_file *file)
{
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
return &file->st;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Set the include chain for "" to QUOTE, for <> to BRACKET. If
QUOTE_IGNORES_SOURCE_DIR, then "" includes do not look in the
directory of the including file.
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
If BRACKET does not lie in the QUOTE chain, it is set to QUOTE.
EMBED is include chain for #embed <>. */
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
void
cpp_set_include_chains (cpp_reader *pfile, cpp_dir *quote, cpp_dir *bracket,
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
cpp_dir *embed, int quote_ignores_source_dir)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
pfile->quote_include = quote;
pfile->bracket_include = quote;
pfile->quote_ignores_source_dir = quote_ignores_source_dir;
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
pfile->embed_include = embed;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
for (; quote; quote = quote->next)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
quote->name_map = NULL;
quote->len = strlen (quote->name);
if (quote == bracket)
pfile->bracket_include = bracket;
}
libcpp, c-family: Add (dumb) C23 N3017 #embed support [PR105863] The following patch implements the C23 N3017 "#embed - a scannable, tooling-friendly binary resource inclusion mechanism" paper. The implementation is intentionally dumb, in that it doesn't significantly speed up compilation of larger initializers and doesn't make it possible to use huge #embeds (like several gigabytes large, that is compile time and memory still infeasible). There are 2 reasons for this. One is that I think like it is implemented now in the patch is how we should use it for the smaller #embed sizes, dunno with which boundary, whether 32 bytes or 64 or something like that, certainly handling the single byte cases which is something that can appear anywhere in the source where constant integer literal can appear is desirable and I think for a few bytes it isn't worth it to come up with something smarter and users would like to e.g. see it in -E readably as well (perhaps the slow vs. fast boundary should be determined by command line option). And the other one is to be able to more easily find regressions in behavior caused by the optimizations, so we have something to get back in git to compare against. I'm definitely willing to work on the optimizations (likely introduce a new CPP_* token type to refer to a range of libcpp owned memory (start + size) and similarly some tree which can do the same, and can be at any time e.g. split into 2 subparts + say INTEGER_CST in between if needed say for const unsigned char d[] = { #embed "2GB.dat" prefix (0, 0, ) suffix (, [0x40000000] = 42) }; still without having to copy around huge amounts of data; STRING_CST owns the memory it points to and can be only 2GB in size), but would like to do that incrementally. And would like to first include some extensions also not included in this patch, like gnu::offset (off) parameter to allow to skip certain constant amount of bytes at the start of the files, plus gnu::base64 ("base64_encoded_data") parameter to add something which can store more efficiently large amounts of the #embed data in preprocessed source. I've been cross-checking all the tests also against the LLVM implementation https://github.com/llvm/llvm-project/pull/68620 which has been for a few hours even committed to LLVM trunk but reverted afterwards. LLVM now has the support committed and I admit I haven't rechecked whether the behavior on the below mentioned spots have been fixed in it already or not yet. The patch uses --embed-dir= option that clang plans to add above and doesn't use other variants on the search directories yet, plus there are no default directories at least for the time being where to search for embed files. So, #embed "..." works if it is found in the same directory (or relative to the current file's directory) and #embed "/..." or #embed </...> work always, but relative #embed <...> doesn't unless at least one --embed-dir= is specified. There is no reason to differentiate between system and non-system directories, so we don't need -isystem like counterpart, perhaps -iquote like counterpart could be useful in the future, dunno what else. It has --embed-directory=dir and --embed-directory dir as aliases. There are some differences beyond clang ICEs, so I'd like to point them out to make sure there is agreement on the choices in the patch. They are also mentioned in the comments of the llvm pull request. The most important is that the GCC patch (as well as the original thephd.dev LLVM branch on godbolt) expands #embed (or acts as if it is expanded) into a mere sequence of numbers like 123,2,35,26 rather then what clang effectively treats as (unsigned char)123,(unsigned char)2,(unsigned char)35,(unsigned char)26 but only does that when using integrated preprocessor, not when using -save-temps where it acts as GCC. JeanHeyd as the original author agrees that is how it is currently worded in C23. Another difference (not tested in the testsuite, not sure how to check for effective target /dev/urandom nor am sure it is desirable to check that during testsuite) is how to treat character devices, named pipes etc. (block devices are errored on). The original paper uses /dev/urandom in various examples and seems to assume that unlike regular files the devices aren't really cached, so #embed </dev/urandom> limit(1) prefix(int a = ) suffix(;) #embed </dev/urandom> limit(1) prefix(int b = ) suffix(;) usually results in a != b. That is what the godbolt thephd.dev branch implements too and what this patch does as well, but clang actually seems to just go from st.st_size == 0, ergo it must be zero-sized resource and so just copies over if_empty if present. It is really questionable what to do about the character devices/named pipes with __has_embed, for regular files the patch doesn't read anything from them, relies on st.st_size + limit for whether it is empty or non-empty. But I don't know of a way to check if read on say a character device would read anything or not (the </dev/null> limit (1) vs. </dev/zero> limit (1) cases), and if we read something, that would be better cached for later because #embed later if it reads again could read no further data even when it first read something. So, the patch currently for __has_embed just always returns 2 on the non-regular files, like the thephd.dev branch does as well and like the clang pull request as well. A question is also what to do for gnu::offset on the non-regular files even for #embed, those aren't seekable and do we want to just read and throw away the offset bytes each time we see it used? clang also chokes on the #if __has_embed (__FILE__ __limit__ (1) __prefix__ () suffix (1 / 0) \ __if_empty__ ((({{[0[0{0{0(0(0)1)1}1}]]}})))) != __STDC_EMBED_FOUND__ #error "__has_embed fail" #endif in embed-1.c, but thephd.dev branch accepts it and I don't see why it shouldn't, (({{[0[0{0{0(0(0)1)1}1}]]}}))) is a balanced token sequence and the file isn't empty, so it should just be parsed and discarded. clang also IMHO mishandles const unsigned char w[] = { #embed __FILE__ prefix([0] = 42, [15] =) limit(32) }; but again only without -save-temps, seems like it treats it as [0] = 42, [15] = (99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98) rather than [0] = 42, [15] = 99,111,110,115,116,32,117,110,115,105,103,110,101,100, 32,99,104,97,114,32,119,91,93,32,61,32,123,10,35,101,109,98 and warns on it for -Wunused-value and just compiles it as [0] = 42, [15] = 98 And also void foo (int, int, int, int); void bar (void) { foo ( #embed __FILE__ limit (4) prefix (172 + ) suffix (+ 2) ); } is treated as 172 + (118, 111, 105, 100) + 2 rather than 172 + 118, 111, 105, 100 + 2 which clang -save-temps or GCC treats it like, so results in just one argument passed rather than 4. if (!strstr ((const char *) magna_carta, "imprisonétur")) abort (); in the testcase fails as well, but in that case calling it in gdb succeeds: p ((char *(*)(char *, char *))__strstr_sse2) (magna_carta, "imprisonétur") $2 = 0x555555558d3c <magna_carta+11564> "imprisonétur aut disseisiátur"... so I guess they are just trying to constant evaluate strstr and do it incorrectly. They started with making the optimizations together in the initial patch set, so they don't have the luxury to compare if it is just because of the optimization they are trying to do or because that is how the feature works for them. At least unless they use -save-temps for now. There is also different behavior between clang and gcc on -M or other dependency generating options. Seems clang includes the __has_embed searched files in dependencies, while my patch doesn't. But so does clang for __has_include and GCC doesn't. Emitting a hard dependency on some header just because there was __has_include/__has_embed for it seems wrong to me, because (at least when properly written) the source likely doesn't mind if the file is missing, it will do something else, so a hard error from make because of it doesn't seem right. Does make have some weaker dependencies, such that if some file can be remade it is but if it doesn't exist, it isn't fatal? I wonder whether #embed <non-existent-file> really needs to be fatal or whether we could simply after diagnosing it pretend the file exists and is empty. For #include I think fatal errors make tons of sense, but perhaps for #embed which is more localized we'd get better error reporting if we didn't bail out immediately. Note, both GCC and clang currently treat those as fatal errors. clang also added -dE option which with -E instead of preprocessing the #embed directives keeps them as is, but the preprocessed source then isn't self-contained. That option looks more harmful than useful to me. Also, it isn't clear to me from C23 whether it is possible to have __has_include/__has_c_attribute/__has_embed expressions inside of the limit #embed/__has_embed argument. 6.10.3.2/2 says that defined should not appear there (and the patch diagnoses it and testsuite tests), but for __has_include/__has_embed etc. 6.10.1/11 says: "The identifiers __has_include, __has_embed, and __has_c_attribute shall not appear in any context not mentioned in this subclause." If that subclause in that case means 6.10.1, then it presumably shouldn't appear in #embed in 6.10.3, but __has_embed is in 6.10.1... But 6.10.3.2/3 says that it should be parsed according to the 6.10.1 rules. Haven't included tests like #if __has_embed (__FILE__ limit (__has_embed (__FILE__ limit (1)))) or #embed __FILE__ limit (__has_include (__FILE__)) into the testsuite because of the doubts but I think the patch should handle those right now. The reason I've used Magna Carta text in some of the testcases is that I hope it shouldn't be copyrighted after the centuries and I'd strongly prefer not to have binary blobs in git after the xz backdoor lesson and wanted something larger which doesn't change all the time. Oh, BTW, I see in C23 draft 6.10.3.2 in Example 4 if (f_source == NULL); return 1; (note the spurious semicolon after closing paren), has that been fixed already? Like the thephd.dev and clang implementations, the patch always macro expands the whole #embed and __has_embed directives except for the embed keyword. That is most likely not what C23 says, my limited understanding right now is that in #embed one needs to parse the whole directive line with macro expansion disabled and check if it satisfies the grammar, if not, the whole directive is macro expanded, if yes, only the limit parameter argument is macro expanded and the prefix/suffix/if_empty arguments are maybe macro expanded when actually used (and not at all if unused). And I think __has_embed macro expansion has conflicting rules. 2024-09-12 Jakub Jelinek <jakub@redhat.com> PR c/105863 libcpp/ * include/cpplib.h: Implement C23 N3017 #embed - a scannable, tooling-friendly binary resource inclusion mechanism paper. (struct cpp_options): Add embed member. (enum cpp_builtin_type): Add BT_HAS_EMBED. (cpp_set_include_chains): Add another cpp_dir * argument to the declaration. * internal.h (enum include_type): Add IT_EMBED. (struct cpp_reader): Add embed_include member. (struct cpp_embed_params_tokens): New type. (struct cpp_embed_params): New type. (_cpp_get_token_no_padding): Declare. (enum _cpp_find_file_kind): Add _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (_cpp_stack_embed): Declare. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. (_cpp_parse_embed_params): Declare. * directives.cc (DIRECTIVE_TABLE): Add embed entry. (end_directive): Don't call skip_rest_of_line for T_EMBED directive. (_cpp_handle_directive): Return 2 rather than 1 for T_EMBED in directives-only mode. (parse_include): Don't Call check_eol for T_EMBED directive. (skip_balanced_token_seq): New function. (EMBED_PARAMS): Define. (enum embed_param_kind): New type. (embed_params): New variable. (_cpp_parse_embed_params): New function. (do_embed): New function. (do_if): Adjust _cpp_parse_expr caller. (do_elif): Likewise. * expr.cc (parse_defined): Diagnose defined in #embed or __has_embed parameters. (_cpp_parse_expr): Change return type to cpp_num_part instead of bool, change second argument from bool to const char * and add third argument. Adjust function comment. For #embed/__has_embed parameters add an artificial CPP_OPEN_PAREN. Use the second argument DIR directly instead of string literals conditional on IS_IF. For #embed/__has_embed parameter, stop on reaching CPP_CLOSE_PAREN matching the artificial one. Diagnose negative or too large embed parameter operands. (num_binary_op): Use #embed instead of #if for diagnostics if inside #embed/__has_embed parameter. (num_div_op): Likewise. * files.cc (struct _cpp_file): Add limit member and embed bitfield. (search_cache): Add IS_EMBED argument, formatting fix. Skip over files with different file->embed from the argument. (find_file_in_dir): Don't call pch_open_file if file->embed. (_cpp_find_file): Handle _cpp_FFK_EMBED and _cpp_FFK_HAS_EMBED. (read_file_guts): Formatting fix. (has_unique_contents): Ignore file->embed files. (search_path_head): Handle IT_EMBED type. (_cpp_stack_embed): New function. (_cpp_get_file_stat): Formatting fix. (cpp_set_include_chains): Add embed argument, save it to pfile->embed_include and compute lens for the chain. * init.cc (struct lang_flags): Add embed member. (lang_defaults): Add embed initializers. (cpp_set_lang): Initialize CPP_OPTION (pfile, embed). (builtin_array): Add __has_embed entry. (cpp_init_builtins): Predefine __STDC_EMBED_NOT_FOUND__, __STDC_EMBED_FOUND__ and __STDC_EMBED_EMPTY__. * lex.cc (cpp_directive_only_process): Handle #embed. * macro.cc (cpp_get_token_no_padding): Rename to ... (_cpp_get_token_no_padding): ... this. No longer static. (builtin_has_include_1): New function. (builtin_has_include): Use it. Use _cpp_get_token_no_padding instead of cpp_get_token_no_padding. (builtin_has_embed): New function. (_cpp_builtin_macro_text): Handle BT_HAS_EMBED. gcc/ * doc/cppdiropts.texi (--embed-dir=): Document. * doc/cpp.texi (Binary Resource Inclusion): New chapter. (__has_embed): Document. * doc/invoke.texi (Directory Options): Mention --embed-dir=. * gcc.cc (cpp_unique_options): Add %{-embed*}. * genmatch.cc (main): Adjust cpp_set_include_chains caller. * incpath.h (enum incpath_kind): Add INC_EMBED. * incpath.cc (merge_include_chains): Handle INC_EMBED. (register_include_chains): Adjust cpp_set_include_chains caller. gcc/c-family/ * c.opt (-embed-dir=): New option. (-embed-directory): New alias. (-embed-directory=): New alias. * c-opts.cc (c_common_handle_option): Handle OPT__embed_dir_. gcc/testsuite/ * c-c++-common/cpp/embed-1.c: New test. * c-c++-common/cpp/embed-2.c: New test. * c-c++-common/cpp/embed-3.c: New test. * c-c++-common/cpp/embed-4.c: New test. * c-c++-common/cpp/embed-5.c: New test. * c-c++-common/cpp/embed-6.c: New test. * c-c++-common/cpp/embed-7.c: New test. * c-c++-common/cpp/embed-8.c: New test. * c-c++-common/cpp/embed-9.c: New test. * c-c++-common/cpp/embed-10.c: New test. * c-c++-common/cpp/embed-11.c: New test. * c-c++-common/cpp/embed-12.c: New test. * c-c++-common/cpp/embed-13.c: New test. * c-c++-common/cpp/embed-14.c: New test. * c-c++-common/cpp/embed-25.c: New test. * c-c++-common/cpp/embed-26.c: New test. * c-c++-common/cpp/embed-dir/embed-1.inc: New test. * c-c++-common/cpp/embed-dir/embed-3.c: New test. * c-c++-common/cpp/embed-dir/embed-4.c: New test. * c-c++-common/cpp/embed-dir/magna-carta.txt: New test. * gcc.dg/cpp/embed-1.c: New test. * gcc.dg/cpp/embed-2.c: New test. * gcc.dg/cpp/embed-3.c: New test. * gcc.dg/cpp/embed-4.c: New test. * g++.dg/cpp/embed-1.C: New test. * g++.dg/cpp/embed-2.C: New test. * g++.dg/cpp/embed-3.C: New test.
2024-09-12 11:15:38 +02:00
for (; embed; embed = embed->next)
{
embed->name_map = NULL;
embed->len = strlen (embed->name);
}
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Append the file name to the directory to create the path, but don't
turn / into // or // into ///; // may be a namespace escape. */
static char *
append_file_to_dir (const char *fname, cpp_dir *dir)
{
size_t dlen, flen;
char *path;
dlen = dir->len;
flen = strlen (fname);
configure.ac: Check declarations for asprintf and vasprintf. * configure.ac: Check declarations for asprintf and vasprintf. * config.in: Regenerate. * configure: Likewise. * charset.c (conversion_loop): Use XRESIZEVEC. (convert_no_conversion): Likewise. (convert_using_iconv): Likewise. (init_iconv_desc): Cast return value of alloca. (cpp_host_to_exec_charset): Use XNEWVEC. (emit_numeric_escape): Use XRESIZEVEC. (cpp_interpret_string): Use XNEWVEC. (cpp_interpret_string): Use XRESIZEVEC. (_cpp_interpret_identifier): Cast return value of alloca. (_cpp_convert_input): Use XNEWVEC and XRESIZEVEC. * directives.c (glue_header_name): Use XNEWVEC and XRESIZEVEC. (parse_include): Use XNEWVEC. (insert_pragma_entry): Rename local variable "new" to "new_entry". (save_registered_pragmas): Cast return value of xmemdup. (destringize_and_run): Same for alloca. (parse_assertion): Likewise. (do_assert): Cast allocated storage to proper type. (cpp_define): Likewise. (_cpp_define_builtin): Likewise. (cpp_undef): Likewise. (handle_assertion): Likewise. (cpp_push_buffer): Rename local variable "new" to "new_buffer". * expr.c (CPP_UPLUS): Cast value to type cpp_ttype. (CPP_UMINUS): Likewise. (struct cpp_operator): Rename from struct operator. (_cpp_expand_op_stack): Use XRESIZEVEC. * files.c (pch_open_file): Use XNEWVEC. (pch_open_file): Use XRESIZEVEC. (read_file_guts): Use XNEWVEC and XRESIZEVEC. (dir_name_of_file): Use XNEWVEC. (make_cpp_file): Use XCNEW. (make_cpp_dir): Likewise. (allocate_file_hash_entries): USE XNEWVEC. (cpp_included): Cast return value of htab_find_with_hash. (append_file_to_dir): Use XNEWVEC. (read_filename_string): Likewise. Use XRESIZEVEC too. (read_name_map): Cast return value of alloca. Use XRESIZEVEC. (remap_filename): Use XNEWVEC. (struct pchf_entry): Move definition out of struct pchf_data. (_cpp_save_file_entries): Use XCNEWVAR. (_cpp_read_file_entries): Use XNEWVAR. * identifiers.c (alloc_node): Use XOBNEW. * init.c (cpp_create_reader): Use XCNEW. (cpp_init_builtins): Cast of b->value to enum builtin_type. (read_original_directory): Cast return value of alloca. * lex.c (add_line_note): Use XRESIZEVEC. (warn_about_normalization): Use XNEWVEC. (_cpp_lex_direct): Cast node->directive_index to (enum cpp_ttype). (new_buff): Use XNEWVEC. * line-map.c (linemap_add): Use XRESIZEVEC. * macro.c (builtin_macro): Cast return value of alloca. (paste_tokens): Likewise. (expand_arg): Use XNEWVEC and XRESIZEVEC. (_cpp_save_parameter): Use XRESIZEVEC. (create_iso_definition): Cast allocated storage to proper type. (_cpp_create_definition): Likewise. (cpp_macro_definition): Use XRESIZEVEC. * makedepend.c (add_clm): Use XNEW. (add_dir): Likewise. * mkdeps.c (munge): Use XNEWVEC. (deps_init): Use XCNEW. (deps_add_target): Use XRESIZEVEC. (deps_add_default_target): Cast return value of alloca. (deps_add_dep): Use XRESIZEVEC. (deps_add_vpath): Likewise. Use XNEWVEC too. (deps_restore): Likewise. * pch.c (save_idents): Use XNEW and XNEWVEC. (cpp_save_state): Use XNEW. (count_defs): Cast return value of htab_find. (write_defs): Likewise. (cpp_write_pch_deps): Use XNEWVEC. (collect_ht_nodes): Use XRESIZEVEC. (cpp_valid_state): Use XNEWVEC. (save_macros): Use XRESIZEVEC. Cast return value of xmemdup. * symtab.c (ht_create): Use XCNEW. (ht_lookup_with_hash): Cast return value of obstack_copy0. (ht_expand): Use XCNEWVEC. * system.h (HAVE_DESIGNATED_INITIALIZERS): False if __cplusplus. (bool): Do not define if __cplusplus. From-SVN: r100295
2005-05-28 15:52:48 +00:00
path = XNEWVEC (char, dlen + 1 + flen + 1);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
memcpy (path, dir->name, dlen);
if (dlen && !IS_DIR_SEPARATOR (path[dlen - 1]))
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
path[dlen++] = '/';
memcpy (&path[dlen], fname, flen + 1);
return path;
}
/* Read a space delimited string of unlimited length from a stdio
file F. */
static char *
read_filename_string (int ch, FILE *f)
{
char *alloc, *set;
int len;
len = 20;
configure.ac: Check declarations for asprintf and vasprintf. * configure.ac: Check declarations for asprintf and vasprintf. * config.in: Regenerate. * configure: Likewise. * charset.c (conversion_loop): Use XRESIZEVEC. (convert_no_conversion): Likewise. (convert_using_iconv): Likewise. (init_iconv_desc): Cast return value of alloca. (cpp_host_to_exec_charset): Use XNEWVEC. (emit_numeric_escape): Use XRESIZEVEC. (cpp_interpret_string): Use XNEWVEC. (cpp_interpret_string): Use XRESIZEVEC. (_cpp_interpret_identifier): Cast return value of alloca. (_cpp_convert_input): Use XNEWVEC and XRESIZEVEC. * directives.c (glue_header_name): Use XNEWVEC and XRESIZEVEC. (parse_include): Use XNEWVEC. (insert_pragma_entry): Rename local variable "new" to "new_entry". (save_registered_pragmas): Cast return value of xmemdup. (destringize_and_run): Same for alloca. (parse_assertion): Likewise. (do_assert): Cast allocated storage to proper type. (cpp_define): Likewise. (_cpp_define_builtin): Likewise. (cpp_undef): Likewise. (handle_assertion): Likewise. (cpp_push_buffer): Rename local variable "new" to "new_buffer". * expr.c (CPP_UPLUS): Cast value to type cpp_ttype. (CPP_UMINUS): Likewise. (struct cpp_operator): Rename from struct operator. (_cpp_expand_op_stack): Use XRESIZEVEC. * files.c (pch_open_file): Use XNEWVEC. (pch_open_file): Use XRESIZEVEC. (read_file_guts): Use XNEWVEC and XRESIZEVEC. (dir_name_of_file): Use XNEWVEC. (make_cpp_file): Use XCNEW. (make_cpp_dir): Likewise. (allocate_file_hash_entries): USE XNEWVEC. (cpp_included): Cast return value of htab_find_with_hash. (append_file_to_dir): Use XNEWVEC. (read_filename_string): Likewise. Use XRESIZEVEC too. (read_name_map): Cast return value of alloca. Use XRESIZEVEC. (remap_filename): Use XNEWVEC. (struct pchf_entry): Move definition out of struct pchf_data. (_cpp_save_file_entries): Use XCNEWVAR. (_cpp_read_file_entries): Use XNEWVAR. * identifiers.c (alloc_node): Use XOBNEW. * init.c (cpp_create_reader): Use XCNEW. (cpp_init_builtins): Cast of b->value to enum builtin_type. (read_original_directory): Cast return value of alloca. * lex.c (add_line_note): Use XRESIZEVEC. (warn_about_normalization): Use XNEWVEC. (_cpp_lex_direct): Cast node->directive_index to (enum cpp_ttype). (new_buff): Use XNEWVEC. * line-map.c (linemap_add): Use XRESIZEVEC. * macro.c (builtin_macro): Cast return value of alloca. (paste_tokens): Likewise. (expand_arg): Use XNEWVEC and XRESIZEVEC. (_cpp_save_parameter): Use XRESIZEVEC. (create_iso_definition): Cast allocated storage to proper type. (_cpp_create_definition): Likewise. (cpp_macro_definition): Use XRESIZEVEC. * makedepend.c (add_clm): Use XNEW. (add_dir): Likewise. * mkdeps.c (munge): Use XNEWVEC. (deps_init): Use XCNEW. (deps_add_target): Use XRESIZEVEC. (deps_add_default_target): Cast return value of alloca. (deps_add_dep): Use XRESIZEVEC. (deps_add_vpath): Likewise. Use XNEWVEC too. (deps_restore): Likewise. * pch.c (save_idents): Use XNEW and XNEWVEC. (cpp_save_state): Use XNEW. (count_defs): Cast return value of htab_find. (write_defs): Likewise. (cpp_write_pch_deps): Use XNEWVEC. (collect_ht_nodes): Use XRESIZEVEC. (cpp_valid_state): Use XNEWVEC. (save_macros): Use XRESIZEVEC. Cast return value of xmemdup. * symtab.c (ht_create): Use XCNEW. (ht_lookup_with_hash): Cast return value of obstack_copy0. (ht_expand): Use XCNEWVEC. * system.h (HAVE_DESIGNATED_INITIALIZERS): False if __cplusplus. (bool): Do not define if __cplusplus. From-SVN: r100295
2005-05-28 15:52:48 +00:00
set = alloc = XNEWVEC (char, len + 1);
if (! is_space (ch))
{
*set++ = ch;
while ((ch = getc (f)) != EOF && ! is_space (ch))
{
if (set - alloc == len)
{
len *= 2;
configure.ac: Check declarations for asprintf and vasprintf. * configure.ac: Check declarations for asprintf and vasprintf. * config.in: Regenerate. * configure: Likewise. * charset.c (conversion_loop): Use XRESIZEVEC. (convert_no_conversion): Likewise. (convert_using_iconv): Likewise. (init_iconv_desc): Cast return value of alloca. (cpp_host_to_exec_charset): Use XNEWVEC. (emit_numeric_escape): Use XRESIZEVEC. (cpp_interpret_string): Use XNEWVEC. (cpp_interpret_string): Use XRESIZEVEC. (_cpp_interpret_identifier): Cast return value of alloca. (_cpp_convert_input): Use XNEWVEC and XRESIZEVEC. * directives.c (glue_header_name): Use XNEWVEC and XRESIZEVEC. (parse_include): Use XNEWVEC. (insert_pragma_entry): Rename local variable "new" to "new_entry". (save_registered_pragmas): Cast return value of xmemdup. (destringize_and_run): Same for alloca. (parse_assertion): Likewise. (do_assert): Cast allocated storage to proper type. (cpp_define): Likewise. (_cpp_define_builtin): Likewise. (cpp_undef): Likewise. (handle_assertion): Likewise. (cpp_push_buffer): Rename local variable "new" to "new_buffer". * expr.c (CPP_UPLUS): Cast value to type cpp_ttype. (CPP_UMINUS): Likewise. (struct cpp_operator): Rename from struct operator. (_cpp_expand_op_stack): Use XRESIZEVEC. * files.c (pch_open_file): Use XNEWVEC. (pch_open_file): Use XRESIZEVEC. (read_file_guts): Use XNEWVEC and XRESIZEVEC. (dir_name_of_file): Use XNEWVEC. (make_cpp_file): Use XCNEW. (make_cpp_dir): Likewise. (allocate_file_hash_entries): USE XNEWVEC. (cpp_included): Cast return value of htab_find_with_hash. (append_file_to_dir): Use XNEWVEC. (read_filename_string): Likewise. Use XRESIZEVEC too. (read_name_map): Cast return value of alloca. Use XRESIZEVEC. (remap_filename): Use XNEWVEC. (struct pchf_entry): Move definition out of struct pchf_data. (_cpp_save_file_entries): Use XCNEWVAR. (_cpp_read_file_entries): Use XNEWVAR. * identifiers.c (alloc_node): Use XOBNEW. * init.c (cpp_create_reader): Use XCNEW. (cpp_init_builtins): Cast of b->value to enum builtin_type. (read_original_directory): Cast return value of alloca. * lex.c (add_line_note): Use XRESIZEVEC. (warn_about_normalization): Use XNEWVEC. (_cpp_lex_direct): Cast node->directive_index to (enum cpp_ttype). (new_buff): Use XNEWVEC. * line-map.c (linemap_add): Use XRESIZEVEC. * macro.c (builtin_macro): Cast return value of alloca. (paste_tokens): Likewise. (expand_arg): Use XNEWVEC and XRESIZEVEC. (_cpp_save_parameter): Use XRESIZEVEC. (create_iso_definition): Cast allocated storage to proper type. (_cpp_create_definition): Likewise. (cpp_macro_definition): Use XRESIZEVEC. * makedepend.c (add_clm): Use XNEW. (add_dir): Likewise. * mkdeps.c (munge): Use XNEWVEC. (deps_init): Use XCNEW. (deps_add_target): Use XRESIZEVEC. (deps_add_default_target): Cast return value of alloca. (deps_add_dep): Use XRESIZEVEC. (deps_add_vpath): Likewise. Use XNEWVEC too. (deps_restore): Likewise. * pch.c (save_idents): Use XNEW and XNEWVEC. (cpp_save_state): Use XNEW. (count_defs): Cast return value of htab_find. (write_defs): Likewise. (cpp_write_pch_deps): Use XNEWVEC. (collect_ht_nodes): Use XRESIZEVEC. (cpp_valid_state): Use XNEWVEC. (save_macros): Use XRESIZEVEC. Cast return value of xmemdup. * symtab.c (ht_create): Use XCNEW. (ht_lookup_with_hash): Cast return value of obstack_copy0. (ht_expand): Use XCNEWVEC. * system.h (HAVE_DESIGNATED_INITIALIZERS): False if __cplusplus. (bool): Do not define if __cplusplus. From-SVN: r100295
2005-05-28 15:52:48 +00:00
alloc = XRESIZEVEC (char, alloc, len + 1);
set = alloc + len / 2;
}
*set++ = ch;
}
}
*set = '\0';
ungetc (ch, f);
return alloc;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Read the file name map file for DIR. */
static void
read_name_map (cpp_dir *dir)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
static const char FILE_NAME_MAP_FILE[] = "header.gcc";
char *name;
FILE *f;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
size_t len, count = 0, room = 9;
len = dir->len;
configure.ac: Check declarations for asprintf and vasprintf. * configure.ac: Check declarations for asprintf and vasprintf. * config.in: Regenerate. * configure: Likewise. * charset.c (conversion_loop): Use XRESIZEVEC. (convert_no_conversion): Likewise. (convert_using_iconv): Likewise. (init_iconv_desc): Cast return value of alloca. (cpp_host_to_exec_charset): Use XNEWVEC. (emit_numeric_escape): Use XRESIZEVEC. (cpp_interpret_string): Use XNEWVEC. (cpp_interpret_string): Use XRESIZEVEC. (_cpp_interpret_identifier): Cast return value of alloca. (_cpp_convert_input): Use XNEWVEC and XRESIZEVEC. * directives.c (glue_header_name): Use XNEWVEC and XRESIZEVEC. (parse_include): Use XNEWVEC. (insert_pragma_entry): Rename local variable "new" to "new_entry". (save_registered_pragmas): Cast return value of xmemdup. (destringize_and_run): Same for alloca. (parse_assertion): Likewise. (do_assert): Cast allocated storage to proper type. (cpp_define): Likewise. (_cpp_define_builtin): Likewise. (cpp_undef): Likewise. (handle_assertion): Likewise. (cpp_push_buffer): Rename local variable "new" to "new_buffer". * expr.c (CPP_UPLUS): Cast value to type cpp_ttype. (CPP_UMINUS): Likewise. (struct cpp_operator): Rename from struct operator. (_cpp_expand_op_stack): Use XRESIZEVEC. * files.c (pch_open_file): Use XNEWVEC. (pch_open_file): Use XRESIZEVEC. (read_file_guts): Use XNEWVEC and XRESIZEVEC. (dir_name_of_file): Use XNEWVEC. (make_cpp_file): Use XCNEW. (make_cpp_dir): Likewise. (allocate_file_hash_entries): USE XNEWVEC. (cpp_included): Cast return value of htab_find_with_hash. (append_file_to_dir): Use XNEWVEC. (read_filename_string): Likewise. Use XRESIZEVEC too. (read_name_map): Cast return value of alloca. Use XRESIZEVEC. (remap_filename): Use XNEWVEC. (struct pchf_entry): Move definition out of struct pchf_data. (_cpp_save_file_entries): Use XCNEWVAR. (_cpp_read_file_entries): Use XNEWVAR. * identifiers.c (alloc_node): Use XOBNEW. * init.c (cpp_create_reader): Use XCNEW. (cpp_init_builtins): Cast of b->value to enum builtin_type. (read_original_directory): Cast return value of alloca. * lex.c (add_line_note): Use XRESIZEVEC. (warn_about_normalization): Use XNEWVEC. (_cpp_lex_direct): Cast node->directive_index to (enum cpp_ttype). (new_buff): Use XNEWVEC. * line-map.c (linemap_add): Use XRESIZEVEC. * macro.c (builtin_macro): Cast return value of alloca. (paste_tokens): Likewise. (expand_arg): Use XNEWVEC and XRESIZEVEC. (_cpp_save_parameter): Use XRESIZEVEC. (create_iso_definition): Cast allocated storage to proper type. (_cpp_create_definition): Likewise. (cpp_macro_definition): Use XRESIZEVEC. * makedepend.c (add_clm): Use XNEW. (add_dir): Likewise. * mkdeps.c (munge): Use XNEWVEC. (deps_init): Use XCNEW. (deps_add_target): Use XRESIZEVEC. (deps_add_default_target): Cast return value of alloca. (deps_add_dep): Use XRESIZEVEC. (deps_add_vpath): Likewise. Use XNEWVEC too. (deps_restore): Likewise. * pch.c (save_idents): Use XNEW and XNEWVEC. (cpp_save_state): Use XNEW. (count_defs): Cast return value of htab_find. (write_defs): Likewise. (cpp_write_pch_deps): Use XNEWVEC. (collect_ht_nodes): Use XRESIZEVEC. (cpp_valid_state): Use XNEWVEC. (save_macros): Use XRESIZEVEC. Cast return value of xmemdup. * symtab.c (ht_create): Use XCNEW. (ht_lookup_with_hash): Cast return value of obstack_copy0. (ht_expand): Use XCNEWVEC. * system.h (HAVE_DESIGNATED_INITIALIZERS): False if __cplusplus. (bool): Do not define if __cplusplus. From-SVN: r100295
2005-05-28 15:52:48 +00:00
name = (char *) alloca (len + sizeof (FILE_NAME_MAP_FILE) + 1);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
memcpy (name, dir->name, len);
if (len && !IS_DIR_SEPARATOR (name[len - 1]))
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
name[len++] = '/';
strcpy (name + len, FILE_NAME_MAP_FILE);
f = fopen (name, "r");
configure.ac: Check declarations for asprintf and vasprintf. * configure.ac: Check declarations for asprintf and vasprintf. * config.in: Regenerate. * configure: Likewise. * charset.c (conversion_loop): Use XRESIZEVEC. (convert_no_conversion): Likewise. (convert_using_iconv): Likewise. (init_iconv_desc): Cast return value of alloca. (cpp_host_to_exec_charset): Use XNEWVEC. (emit_numeric_escape): Use XRESIZEVEC. (cpp_interpret_string): Use XNEWVEC. (cpp_interpret_string): Use XRESIZEVEC. (_cpp_interpret_identifier): Cast return value of alloca. (_cpp_convert_input): Use XNEWVEC and XRESIZEVEC. * directives.c (glue_header_name): Use XNEWVEC and XRESIZEVEC. (parse_include): Use XNEWVEC. (insert_pragma_entry): Rename local variable "new" to "new_entry". (save_registered_pragmas): Cast return value of xmemdup. (destringize_and_run): Same for alloca. (parse_assertion): Likewise. (do_assert): Cast allocated storage to proper type. (cpp_define): Likewise. (_cpp_define_builtin): Likewise. (cpp_undef): Likewise. (handle_assertion): Likewise. (cpp_push_buffer): Rename local variable "new" to "new_buffer". * expr.c (CPP_UPLUS): Cast value to type cpp_ttype. (CPP_UMINUS): Likewise. (struct cpp_operator): Rename from struct operator. (_cpp_expand_op_stack): Use XRESIZEVEC. * files.c (pch_open_file): Use XNEWVEC. (pch_open_file): Use XRESIZEVEC. (read_file_guts): Use XNEWVEC and XRESIZEVEC. (dir_name_of_file): Use XNEWVEC. (make_cpp_file): Use XCNEW. (make_cpp_dir): Likewise. (allocate_file_hash_entries): USE XNEWVEC. (cpp_included): Cast return value of htab_find_with_hash. (append_file_to_dir): Use XNEWVEC. (read_filename_string): Likewise. Use XRESIZEVEC too. (read_name_map): Cast return value of alloca. Use XRESIZEVEC. (remap_filename): Use XNEWVEC. (struct pchf_entry): Move definition out of struct pchf_data. (_cpp_save_file_entries): Use XCNEWVAR. (_cpp_read_file_entries): Use XNEWVAR. * identifiers.c (alloc_node): Use XOBNEW. * init.c (cpp_create_reader): Use XCNEW. (cpp_init_builtins): Cast of b->value to enum builtin_type. (read_original_directory): Cast return value of alloca. * lex.c (add_line_note): Use XRESIZEVEC. (warn_about_normalization): Use XNEWVEC. (_cpp_lex_direct): Cast node->directive_index to (enum cpp_ttype). (new_buff): Use XNEWVEC. * line-map.c (linemap_add): Use XRESIZEVEC. * macro.c (builtin_macro): Cast return value of alloca. (paste_tokens): Likewise. (expand_arg): Use XNEWVEC and XRESIZEVEC. (_cpp_save_parameter): Use XRESIZEVEC. (create_iso_definition): Cast allocated storage to proper type. (_cpp_create_definition): Likewise. (cpp_macro_definition): Use XRESIZEVEC. * makedepend.c (add_clm): Use XNEW. (add_dir): Likewise. * mkdeps.c (munge): Use XNEWVEC. (deps_init): Use XCNEW. (deps_add_target): Use XRESIZEVEC. (deps_add_default_target): Cast return value of alloca. (deps_add_dep): Use XRESIZEVEC. (deps_add_vpath): Likewise. Use XNEWVEC too. (deps_restore): Likewise. * pch.c (save_idents): Use XNEW and XNEWVEC. (cpp_save_state): Use XNEW. (count_defs): Cast return value of htab_find. (write_defs): Likewise. (cpp_write_pch_deps): Use XNEWVEC. (collect_ht_nodes): Use XRESIZEVEC. (cpp_valid_state): Use XNEWVEC. (save_macros): Use XRESIZEVEC. Cast return value of xmemdup. * symtab.c (ht_create): Use XCNEW. (ht_lookup_with_hash): Cast return value of obstack_copy0. (ht_expand): Use XCNEWVEC. * system.h (HAVE_DESIGNATED_INITIALIZERS): False if __cplusplus. (bool): Do not define if __cplusplus. From-SVN: r100295
2005-05-28 15:52:48 +00:00
dir->name_map = XNEWVEC (const char *, room);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Silently return NULL if we cannot open. */
if (f)
{
int ch;
while ((ch = getc (f)) != EOF)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
char *to;
if (is_space (ch))
continue;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (count + 2 > room)
{
room += 8;
configure.ac: Check declarations for asprintf and vasprintf. * configure.ac: Check declarations for asprintf and vasprintf. * config.in: Regenerate. * configure: Likewise. * charset.c (conversion_loop): Use XRESIZEVEC. (convert_no_conversion): Likewise. (convert_using_iconv): Likewise. (init_iconv_desc): Cast return value of alloca. (cpp_host_to_exec_charset): Use XNEWVEC. (emit_numeric_escape): Use XRESIZEVEC. (cpp_interpret_string): Use XNEWVEC. (cpp_interpret_string): Use XRESIZEVEC. (_cpp_interpret_identifier): Cast return value of alloca. (_cpp_convert_input): Use XNEWVEC and XRESIZEVEC. * directives.c (glue_header_name): Use XNEWVEC and XRESIZEVEC. (parse_include): Use XNEWVEC. (insert_pragma_entry): Rename local variable "new" to "new_entry". (save_registered_pragmas): Cast return value of xmemdup. (destringize_and_run): Same for alloca. (parse_assertion): Likewise. (do_assert): Cast allocated storage to proper type. (cpp_define): Likewise. (_cpp_define_builtin): Likewise. (cpp_undef): Likewise. (handle_assertion): Likewise. (cpp_push_buffer): Rename local variable "new" to "new_buffer". * expr.c (CPP_UPLUS): Cast value to type cpp_ttype. (CPP_UMINUS): Likewise. (struct cpp_operator): Rename from struct operator. (_cpp_expand_op_stack): Use XRESIZEVEC. * files.c (pch_open_file): Use XNEWVEC. (pch_open_file): Use XRESIZEVEC. (read_file_guts): Use XNEWVEC and XRESIZEVEC. (dir_name_of_file): Use XNEWVEC. (make_cpp_file): Use XCNEW. (make_cpp_dir): Likewise. (allocate_file_hash_entries): USE XNEWVEC. (cpp_included): Cast return value of htab_find_with_hash. (append_file_to_dir): Use XNEWVEC. (read_filename_string): Likewise. Use XRESIZEVEC too. (read_name_map): Cast return value of alloca. Use XRESIZEVEC. (remap_filename): Use XNEWVEC. (struct pchf_entry): Move definition out of struct pchf_data. (_cpp_save_file_entries): Use XCNEWVAR. (_cpp_read_file_entries): Use XNEWVAR. * identifiers.c (alloc_node): Use XOBNEW. * init.c (cpp_create_reader): Use XCNEW. (cpp_init_builtins): Cast of b->value to enum builtin_type. (read_original_directory): Cast return value of alloca. * lex.c (add_line_note): Use XRESIZEVEC. (warn_about_normalization): Use XNEWVEC. (_cpp_lex_direct): Cast node->directive_index to (enum cpp_ttype). (new_buff): Use XNEWVEC. * line-map.c (linemap_add): Use XRESIZEVEC. * macro.c (builtin_macro): Cast return value of alloca. (paste_tokens): Likewise. (expand_arg): Use XNEWVEC and XRESIZEVEC. (_cpp_save_parameter): Use XRESIZEVEC. (create_iso_definition): Cast allocated storage to proper type. (_cpp_create_definition): Likewise. (cpp_macro_definition): Use XRESIZEVEC. * makedepend.c (add_clm): Use XNEW. (add_dir): Likewise. * mkdeps.c (munge): Use XNEWVEC. (deps_init): Use XCNEW. (deps_add_target): Use XRESIZEVEC. (deps_add_default_target): Cast return value of alloca. (deps_add_dep): Use XRESIZEVEC. (deps_add_vpath): Likewise. Use XNEWVEC too. (deps_restore): Likewise. * pch.c (save_idents): Use XNEW and XNEWVEC. (cpp_save_state): Use XNEW. (count_defs): Cast return value of htab_find. (write_defs): Likewise. (cpp_write_pch_deps): Use XNEWVEC. (collect_ht_nodes): Use XRESIZEVEC. (cpp_valid_state): Use XNEWVEC. (save_macros): Use XRESIZEVEC. Cast return value of xmemdup. * symtab.c (ht_create): Use XCNEW. (ht_lookup_with_hash): Cast return value of obstack_copy0. (ht_expand): Use XCNEWVEC. * system.h (HAVE_DESIGNATED_INITIALIZERS): False if __cplusplus. (bool): Do not define if __cplusplus. From-SVN: r100295
2005-05-28 15:52:48 +00:00
dir->name_map = XRESIZEVEC (const char *, dir->name_map, room);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
}
dir->name_map[count] = read_filename_string (ch, f);
while ((ch = getc (f)) != EOF && is_hspace (ch))
;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
to = read_filename_string (ch, f);
if (IS_ABSOLUTE_PATH (to))
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
dir->name_map[count + 1] = to;
else
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
dir->name_map[count + 1] = append_file_to_dir (to, dir);
free (to);
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
count += 2;
while ((ch = getc (f)) != '\n')
if (ch == EOF)
break;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
fclose (f);
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Terminate the list of maps. */
dir->name_map[count] = NULL;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Remap a FILE's name based on the file_name_map, if any, for
FILE->dir. If the file name has any directory separators,
recursively check those directories too. */
static char *
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
remap_filename (cpp_reader *pfile, _cpp_file *file)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
const char *fname, *p;
char *new_dir, *p3;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
cpp_dir *dir;
size_t index, len;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
dir = file->dir;
fname = file->name;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
for (;;)
{
if (!dir->name_map)
read_name_map (dir);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
for (index = 0; dir->name_map[index]; index += 2)
if (!filename_cmp (dir->name_map[index], fname))
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
return xstrdup (dir->name_map[index + 1]);
if (IS_ABSOLUTE_PATH (fname))
return NULL;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
p = strchr (fname, '/');
#ifdef HAVE_DOS_BASED_FILE_SYSTEM
{
const char *p2 = strchr (fname, '\\');
if (!p || (p2 && p > p2))
p = p2;
}
#endif
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (!p || p == fname)
return NULL;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
len = dir->len + (p - fname + 1);
new_dir = XNEWVEC (char, len + 2);
p3 = new_dir + dir->len;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
memcpy (new_dir, dir->name, dir->len);
if (dir->len && !IS_DIR_SEPARATOR (dir->name[dir->len - 1]))
{
*p3++ = '/';
len++;
}
memcpy (p3, fname, p - fname + 1);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
new_dir[len] = '\0';
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
dir = make_cpp_dir (pfile, new_dir, dir->sysp);
fname = p + 1;
cpplib.h: Delete struct import_file. Wed Nov 25 14:54:46 1998 Zack Weinberg <zack@rabi.phys.columbia.edu> * cpplib.h: Delete struct import_file. Add ihash element to struct cpp_buffer. Delete dont_repeat_files and import_hash_table elements from cpp_reader; change all_include_files to a hash table. Delete all foobar_include / last_foobar_include elements from struct cpp_options; put back four such: quote_include, bracket_include, system_include, after_include. Redo struct file_name_list completely. Add new structure type include_hash. Add prototypes for merge_include_chains and include_hash. Change prototypes for finclude, find_include_file, and append_include_chain to match changes below. * cppfiles.c (simplify_pathname, include_hash, remap_filename, merge_include_chains): New functions. (add_import, lookup_import, open_include_file): Removed. (INO_T_EQ): Define this (copied from cccp.c). (hack_vms_include_specification): Remove all calls and #if 0 out the definition. It was being called incorrectly and at the wrong times. Until a VMSie can look at this, it's better to not pretend to support it. (append_include_chain): Change calling convention; now takes only one directory at a time, and sets up the data structure itself. (redundant_include_p): Rewritten - this is now used for all include redundancy, whether by #ifndef, #import, or #pragma once. Looks up things in the include hash table. (file_cleanup): Decrement pfile->system_include_depth here if it's >0. (find_include_file): Calling convention changed; now passes around a struct include_hash instead of 3 separate parameters. Guts ripped out and replaced with new include_hash mechanism. (finclude): Calling convention changed as for find_include_file. Error exits pulled out-of-line. Reformat. (safe_read): Return a long, not an int. (deps_output): Don't recurse. * cpplib.c (is_system_include): Deleted. (path_include): Fix up call to append_include_chain. (do_include): Fix up calls to find_include_file and finclude. Clean up dependency output a bit. Shorten obnoxiously lengthy #import warning message. Don't decrement pfile->system_include_depth here. (do_pragma): Understand the include_hash structure. Reformat. (do_endif): Correct handling of control macros. Understand the include_hash. (cpp_start_read): Fix up calls to finclude. Call merge_include_chains. (cpp_handle_option): Fix up calls to append_include_chain. Understand the four partial include chains. (cpp_finish): Add debugging code (#if 0-ed out) for the include_hash. (cpp_cleanup): Free the include_hash, not the import hash and the all_include and dont_repeat lists which no longer exist. From-SVN: r23857
1998-11-25 11:56:54 +00:00
}
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
/* Returns true if PCHNAME is a valid PCH file for FILE. */
static bool
validate_pch (cpp_reader *pfile, _cpp_file *file, const char *pchname)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
const char *saved_path = file->path;
bool valid = false;
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
file->path = pchname;
if (open_file (file))
{
valid = 1 & pfile->cb.valid_pch (pfile, pchname, file->fd);
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (!valid)
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
close (file->fd);
file->fd = -1;
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
if (CPP_OPTION (pfile, print_include_names))
{
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
unsigned int i;
2004-01-19 23:35:21 -08:00
for (i = 1; i < pfile->line_table->depth; i++)
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
putc ('.', stderr);
fprintf (stderr, "%c %s\n",
valid ? '!' : 'x', pchname);
}
}
Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * Makefile.in (LIBCPP_DEPS): Add HASHTAB_H. * cppfiles.c: Completely rewritten. * c-incpath.c (free_path, remove_duplicates, heads, tails, add_path): struct cpp_path is now struct cpp_dir. (remove_duplicates): Don't simplify path names. * c-opts.c (c_common_parse_file): cpp_read_next_file renamed cpp_stack_file. * cpphash.h: Include hashtab.h. (_cpp_file): Declare. (struct cpp_buffer): struct include_file is now struct _cpp_file, and struct cpp_path is now struct cpp_dir. Rename members. (struct cpp_reader): Similarly. New members once_only_files, file_hash, file_hash_entries, quote_ignores_source_dir, no_search_path, saw_pragma_once. Remove all_include_files and max_include_len. Make some members bool. (_cpp_mark_only_only): Renamed from _cpp_never_reread. (_cpp_stack_file): Renamed from _cpp_read_file. (_cpp_stack_include): Renamed from _cpp_execute_include. (_cpp_init_files): Renamed from _cpp_init_includes. (_cpp_cleanup_files): Renamed from _cpp_cleanup_includes. * cppinit.c (cpp_create_reader): Initialize no_search_path. Update. (cpp_read_next_file): Rename and move to cppfiles.c. (cpp_read_main_file): Update. * cpplib.c (run_directive): Update for renamed members. (do_include_common, _cpp_pop_buffer): Update. (do_import): Undeprecate #import. (do_pragma_once): Undeprecate. Use _cpp_mark_file_once_only. * cpplib.h: Remove file_name_map_list. (cpp_options): Remove map_list. (cpp_dir): Rename from cpp_path. New datatype for name_map. (cpp_set_include_chains, cpp_stack_file, cpp_included): Update. testsuite: * gcc.dg/cpp/include2.c: Only expect one message. From-SVN: r69942
2003-07-29 22:26:13 +00:00
file->path = saved_path;
return valid;
}
Add framework support for darwin. * c-incpath.c: Include target.h and machmode.h. (add_path): Use a consistent style for cpp_dir. Initialize p->construct to 0. (add_cpp_dir_path): New. (register_include_chains): Add use of extra_includes callback. (hook_void_int): Add. (target_c_incpath): Add. * c-incpath.h (add_cpp_dir_path): New. (target_c_incpath_s): Add. (target_c_incpath): Add. (C_INCPATH_INIT): Add. * c-opts.c (c_common_missing_argument, c_common_handle_option): Add -F argument processing. * c.opt: Add -F argument processing. * gcc.c (trad_capable_cpp): Add -F argument processing. * cppfiles.c (find_file_in_dir): Update to use construct callback. (search_path_exhausted, cpp_get_path, cpp_get_buffer, cpp_get_prev): New. (_cpp_find_file): Use search_path_exhausted. (make_cpp_dir): Initialize construct to 0. * cpplib.h (missing_header_cb cpp_get_path, cpp_get_buffer, cpp_get_file, cpp_get_prev): New. (cpp_callbacks): Add missing_header (cpp_dir): Add construct. * target-def.h: (TARGET_OPTF): New. * hooks.c (hook_void_int, hook_void_charptr): Add. * hooks.h (hook_void_int, hook_void_charptr): Add. * Makefile.in (c-incpath.o) : Add $(TARGET_H) and $(MACHMODE_H) dependencies. * doc/invoke.texi (Darwin Options): Document -F. * doc/tm.texi (TARGET_EXTRA_INCLUDES): Add. (TARGET_OPTF): Add. * fix-header.c (target_c_incpath): Add. * config/darwin-c.c: Add c-incpath.h include. (using_frameworks, find_subframework_file, find_subframework_header, add_system_framework_path, frameworks_in_use, num_frameworks, max_frameworks, add_framework, find_framework, struct framework_header, framework_header_dirs, framework_construct_pathname, find_subframework_file, add_system_framework_path, add_framework_path, framework_defaults, darwin_register_frameworks, find_subframework_header): Add. * config/darwin.h (TARGET_EXTRA_INCLUDES, TARGET_OPTF): New. (TARGET_OPTION_TRANSLATE_TABLE): Add -framework support. (CPP_SPEC): Add __APPLE_CC__ support. * t-darwin (darwin-c.o): Add c-incpath.h dependency. From-SVN: r78875
2004-03-04 00:18:54 +00:00
/* Get the path associated with the _cpp_file F. The path includes
the base name from the include directive and the directory it was
found in via the search path. */
const char *
cpp_get_path (struct _cpp_file *f)
{
return f->path;
}
/* Get the directory associated with the _cpp_file F. */
cpp_dir *
cpp_get_dir (struct _cpp_file *f)
{
return f->dir;
}
Add framework support for darwin. * c-incpath.c: Include target.h and machmode.h. (add_path): Use a consistent style for cpp_dir. Initialize p->construct to 0. (add_cpp_dir_path): New. (register_include_chains): Add use of extra_includes callback. (hook_void_int): Add. (target_c_incpath): Add. * c-incpath.h (add_cpp_dir_path): New. (target_c_incpath_s): Add. (target_c_incpath): Add. (C_INCPATH_INIT): Add. * c-opts.c (c_common_missing_argument, c_common_handle_option): Add -F argument processing. * c.opt: Add -F argument processing. * gcc.c (trad_capable_cpp): Add -F argument processing. * cppfiles.c (find_file_in_dir): Update to use construct callback. (search_path_exhausted, cpp_get_path, cpp_get_buffer, cpp_get_prev): New. (_cpp_find_file): Use search_path_exhausted. (make_cpp_dir): Initialize construct to 0. * cpplib.h (missing_header_cb cpp_get_path, cpp_get_buffer, cpp_get_file, cpp_get_prev): New. (cpp_callbacks): Add missing_header (cpp_dir): Add construct. * target-def.h: (TARGET_OPTF): New. * hooks.c (hook_void_int, hook_void_charptr): Add. * hooks.h (hook_void_int, hook_void_charptr): Add. * Makefile.in (c-incpath.o) : Add $(TARGET_H) and $(MACHMODE_H) dependencies. * doc/invoke.texi (Darwin Options): Document -F. * doc/tm.texi (TARGET_EXTRA_INCLUDES): Add. (TARGET_OPTF): Add. * fix-header.c (target_c_incpath): Add. * config/darwin-c.c: Add c-incpath.h include. (using_frameworks, find_subframework_file, find_subframework_header, add_system_framework_path, frameworks_in_use, num_frameworks, max_frameworks, add_framework, find_framework, struct framework_header, framework_header_dirs, framework_construct_pathname, find_subframework_file, add_system_framework_path, add_framework_path, framework_defaults, darwin_register_frameworks, find_subframework_header): Add. * config/darwin.h (TARGET_EXTRA_INCLUDES, TARGET_OPTF): New. (TARGET_OPTION_TRANSLATE_TABLE): Add -framework support. (CPP_SPEC): Add __APPLE_CC__ support. * t-darwin (darwin-c.o): Add c-incpath.h dependency. From-SVN: r78875
2004-03-04 00:18:54 +00:00
/* Get the cpp_buffer currently associated with the cpp_reader
PFILE. */
cpp_buffer *
cpp_get_buffer (cpp_reader *pfile)
{
return pfile->buffer;
}
/* Get the _cpp_file associated with the cpp_buffer B. */
_cpp_file *
cpp_get_file (cpp_buffer *b)
{
return b->file;
}
/* Get the previous cpp_buffer given a cpp_buffer B. The previous
buffer is the buffer that included the given buffer. */
cpp_buffer *
cpp_get_prev (cpp_buffer *b)
{
return b->prev;
}
/* This data structure holds the list of header files that were seen
while the PCH was being built. The 'entries' field is kept sorted
in memcmp() order; yes, this means that on little-endian systems,
it's sorted initially by the least-significant byte of 'size', but
that's OK. The code does rely on having entries with the same size
next to each other. */
configure.ac: Check declarations for asprintf and vasprintf. * configure.ac: Check declarations for asprintf and vasprintf. * config.in: Regenerate. * configure: Likewise. * charset.c (conversion_loop): Use XRESIZEVEC. (convert_no_conversion): Likewise. (convert_using_iconv): Likewise. (init_iconv_desc): Cast return value of alloca. (cpp_host_to_exec_charset): Use XNEWVEC. (emit_numeric_escape): Use XRESIZEVEC. (cpp_interpret_string): Use XNEWVEC. (cpp_interpret_string): Use XRESIZEVEC. (_cpp_interpret_identifier): Cast return value of alloca. (_cpp_convert_input): Use XNEWVEC and XRESIZEVEC. * directives.c (glue_header_name): Use XNEWVEC and XRESIZEVEC. (parse_include): Use XNEWVEC. (insert_pragma_entry): Rename local variable "new" to "new_entry". (save_registered_pragmas): Cast return value of xmemdup. (destringize_and_run): Same for alloca. (parse_assertion): Likewise. (do_assert): Cast allocated storage to proper type. (cpp_define): Likewise. (_cpp_define_builtin): Likewise. (cpp_undef): Likewise. (handle_assertion): Likewise. (cpp_push_buffer): Rename local variable "new" to "new_buffer". * expr.c (CPP_UPLUS): Cast value to type cpp_ttype. (CPP_UMINUS): Likewise. (struct cpp_operator): Rename from struct operator. (_cpp_expand_op_stack): Use XRESIZEVEC. * files.c (pch_open_file): Use XNEWVEC. (pch_open_file): Use XRESIZEVEC. (read_file_guts): Use XNEWVEC and XRESIZEVEC. (dir_name_of_file): Use XNEWVEC. (make_cpp_file): Use XCNEW. (make_cpp_dir): Likewise. (allocate_file_hash_entries): USE XNEWVEC. (cpp_included): Cast return value of htab_find_with_hash. (append_file_to_dir): Use XNEWVEC. (read_filename_string): Likewise. Use XRESIZEVEC too. (read_name_map): Cast return value of alloca. Use XRESIZEVEC. (remap_filename): Use XNEWVEC. (struct pchf_entry): Move definition out of struct pchf_data. (_cpp_save_file_entries): Use XCNEWVAR. (_cpp_read_file_entries): Use XNEWVAR. * identifiers.c (alloc_node): Use XOBNEW. * init.c (cpp_create_reader): Use XCNEW. (cpp_init_builtins): Cast of b->value to enum builtin_type. (read_original_directory): Cast return value of alloca. * lex.c (add_line_note): Use XRESIZEVEC. (warn_about_normalization): Use XNEWVEC. (_cpp_lex_direct): Cast node->directive_index to (enum cpp_ttype). (new_buff): Use XNEWVEC. * line-map.c (linemap_add): Use XRESIZEVEC. * macro.c (builtin_macro): Cast return value of alloca. (paste_tokens): Likewise. (expand_arg): Use XNEWVEC and XRESIZEVEC. (_cpp_save_parameter): Use XRESIZEVEC. (create_iso_definition): Cast allocated storage to proper type. (_cpp_create_definition): Likewise. (cpp_macro_definition): Use XRESIZEVEC. * makedepend.c (add_clm): Use XNEW. (add_dir): Likewise. * mkdeps.c (munge): Use XNEWVEC. (deps_init): Use XCNEW. (deps_add_target): Use XRESIZEVEC. (deps_add_default_target): Cast return value of alloca. (deps_add_dep): Use XRESIZEVEC. (deps_add_vpath): Likewise. Use XNEWVEC too. (deps_restore): Likewise. * pch.c (save_idents): Use XNEW and XNEWVEC. (cpp_save_state): Use XNEW. (count_defs): Cast return value of htab_find. (write_defs): Likewise. (cpp_write_pch_deps): Use XNEWVEC. (collect_ht_nodes): Use XRESIZEVEC. (cpp_valid_state): Use XNEWVEC. (save_macros): Use XRESIZEVEC. Cast return value of xmemdup. * symtab.c (ht_create): Use XCNEW. (ht_lookup_with_hash): Cast return value of obstack_copy0. (ht_expand): Use XCNEWVEC. * system.h (HAVE_DESIGNATED_INITIALIZERS): False if __cplusplus. (bool): Do not define if __cplusplus. From-SVN: r100295
2005-05-28 15:52:48 +00:00
struct pchf_entry {
/* The size of this file. This is used to save running a MD5 checksum
if the sizes don't match. */
off_t size;
/* The MD5 checksum of this file. */
unsigned char sum[16];
/* Is this file to be included only once? */
bool once_only;
};
struct pchf_data {
/* Number of pchf_entry structures. */
size_t count;
/* Are there any values with once_only set?
This is used as an optimisation, it means we don't have to search
the structure if we're processing a regular #include. */
bool have_once_only;
[multiple changes] 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * c-opts.c (c_common_handle_option): Add -finput-charset. * c.opt: Ditto. * cppcharset.c (one_iso88591_to_utf8): Remove. (convert_iso88591_utf8): Ditto. (conversion_tab): Remove 8859-1 converter. (_cpp_input_to_utf8): Remove. (_cpp_init_iconv_buffer): Ditto. (_cpp_close_iconv_buffer): Ditto. (_cpp_convert_input): New function. (_cpp_default_encoding): Ditto. * cpphash.h: Add/remove prototypes for above. * cppfiles.c (read_file_guts): Use _cpp_convert_input. * cppinit.c (cpp_create_reader): Use _cpp_default_encoding for narrow execution and input character sets. * cpplib.c (cpp_push_buffer): Delete uses of removed functions. * doc/cppopts.texi: Document -finput-charset. 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * gcc.c-torture/execute/wchar_t-1.c: Add -finput-charset. 2004-01-29 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * testsuite/22_locale/collate/compare/wchar_t/2.cc: Remove xfail. Use -finput-charset. * testsuite/22_locale/collate/compare/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/compare/wchar_t/wrapped_locale.cc: Ditto * testsuite/22_locale/collate/hash/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_locale.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_locale.cc: Ditto. From-SVN: r77136
2004-02-02 20:20:58 +00:00
configure.ac: Check declarations for asprintf and vasprintf. * configure.ac: Check declarations for asprintf and vasprintf. * config.in: Regenerate. * configure: Likewise. * charset.c (conversion_loop): Use XRESIZEVEC. (convert_no_conversion): Likewise. (convert_using_iconv): Likewise. (init_iconv_desc): Cast return value of alloca. (cpp_host_to_exec_charset): Use XNEWVEC. (emit_numeric_escape): Use XRESIZEVEC. (cpp_interpret_string): Use XNEWVEC. (cpp_interpret_string): Use XRESIZEVEC. (_cpp_interpret_identifier): Cast return value of alloca. (_cpp_convert_input): Use XNEWVEC and XRESIZEVEC. * directives.c (glue_header_name): Use XNEWVEC and XRESIZEVEC. (parse_include): Use XNEWVEC. (insert_pragma_entry): Rename local variable "new" to "new_entry". (save_registered_pragmas): Cast return value of xmemdup. (destringize_and_run): Same for alloca. (parse_assertion): Likewise. (do_assert): Cast allocated storage to proper type. (cpp_define): Likewise. (_cpp_define_builtin): Likewise. (cpp_undef): Likewise. (handle_assertion): Likewise. (cpp_push_buffer): Rename local variable "new" to "new_buffer". * expr.c (CPP_UPLUS): Cast value to type cpp_ttype. (CPP_UMINUS): Likewise. (struct cpp_operator): Rename from struct operator. (_cpp_expand_op_stack): Use XRESIZEVEC. * files.c (pch_open_file): Use XNEWVEC. (pch_open_file): Use XRESIZEVEC. (read_file_guts): Use XNEWVEC and XRESIZEVEC. (dir_name_of_file): Use XNEWVEC. (make_cpp_file): Use XCNEW. (make_cpp_dir): Likewise. (allocate_file_hash_entries): USE XNEWVEC. (cpp_included): Cast return value of htab_find_with_hash. (append_file_to_dir): Use XNEWVEC. (read_filename_string): Likewise. Use XRESIZEVEC too. (read_name_map): Cast return value of alloca. Use XRESIZEVEC. (remap_filename): Use XNEWVEC. (struct pchf_entry): Move definition out of struct pchf_data. (_cpp_save_file_entries): Use XCNEWVAR. (_cpp_read_file_entries): Use XNEWVAR. * identifiers.c (alloc_node): Use XOBNEW. * init.c (cpp_create_reader): Use XCNEW. (cpp_init_builtins): Cast of b->value to enum builtin_type. (read_original_directory): Cast return value of alloca. * lex.c (add_line_note): Use XRESIZEVEC. (warn_about_normalization): Use XNEWVEC. (_cpp_lex_direct): Cast node->directive_index to (enum cpp_ttype). (new_buff): Use XNEWVEC. * line-map.c (linemap_add): Use XRESIZEVEC. * macro.c (builtin_macro): Cast return value of alloca. (paste_tokens): Likewise. (expand_arg): Use XNEWVEC and XRESIZEVEC. (_cpp_save_parameter): Use XRESIZEVEC. (create_iso_definition): Cast allocated storage to proper type. (_cpp_create_definition): Likewise. (cpp_macro_definition): Use XRESIZEVEC. * makedepend.c (add_clm): Use XNEW. (add_dir): Likewise. * mkdeps.c (munge): Use XNEWVEC. (deps_init): Use XCNEW. (deps_add_target): Use XRESIZEVEC. (deps_add_default_target): Cast return value of alloca. (deps_add_dep): Use XRESIZEVEC. (deps_add_vpath): Likewise. Use XNEWVEC too. (deps_restore): Likewise. * pch.c (save_idents): Use XNEW and XNEWVEC. (cpp_save_state): Use XNEW. (count_defs): Cast return value of htab_find. (write_defs): Likewise. (cpp_write_pch_deps): Use XNEWVEC. (collect_ht_nodes): Use XRESIZEVEC. (cpp_valid_state): Use XNEWVEC. (save_macros): Use XRESIZEVEC. Cast return value of xmemdup. * symtab.c (ht_create): Use XCNEW. (ht_lookup_with_hash): Cast return value of obstack_copy0. (ht_expand): Use XCNEWVEC. * system.h (HAVE_DESIGNATED_INITIALIZERS): False if __cplusplus. (bool): Do not define if __cplusplus. From-SVN: r100295
2005-05-28 15:52:48 +00:00
struct pchf_entry entries[1];
};
static struct pchf_data *pchf;
/* A qsort ordering function for pchf_entry structures. */
static int
pchf_save_compare (const void *e1, const void *e2)
{
return memcmp (e1, e2, sizeof (struct pchf_entry));
}
/* Create and write to F a pchf_data structure. */
bool
_cpp_save_file_entries (cpp_reader *pfile, FILE *fp)
{
size_t count = 0;
struct pchf_data *result;
size_t result_size;
_cpp_file *f;
bool ret;
for (f = pfile->all_files; f; f = f->next_file)
++count;
result_size = (sizeof (struct pchf_data)
+ sizeof (struct pchf_entry) * (count - 1));
configure.ac: Check declarations for asprintf and vasprintf. * configure.ac: Check declarations for asprintf and vasprintf. * config.in: Regenerate. * configure: Likewise. * charset.c (conversion_loop): Use XRESIZEVEC. (convert_no_conversion): Likewise. (convert_using_iconv): Likewise. (init_iconv_desc): Cast return value of alloca. (cpp_host_to_exec_charset): Use XNEWVEC. (emit_numeric_escape): Use XRESIZEVEC. (cpp_interpret_string): Use XNEWVEC. (cpp_interpret_string): Use XRESIZEVEC. (_cpp_interpret_identifier): Cast return value of alloca. (_cpp_convert_input): Use XNEWVEC and XRESIZEVEC. * directives.c (glue_header_name): Use XNEWVEC and XRESIZEVEC. (parse_include): Use XNEWVEC. (insert_pragma_entry): Rename local variable "new" to "new_entry". (save_registered_pragmas): Cast return value of xmemdup. (destringize_and_run): Same for alloca. (parse_assertion): Likewise. (do_assert): Cast allocated storage to proper type. (cpp_define): Likewise. (_cpp_define_builtin): Likewise. (cpp_undef): Likewise. (handle_assertion): Likewise. (cpp_push_buffer): Rename local variable "new" to "new_buffer". * expr.c (CPP_UPLUS): Cast value to type cpp_ttype. (CPP_UMINUS): Likewise. (struct cpp_operator): Rename from struct operator. (_cpp_expand_op_stack): Use XRESIZEVEC. * files.c (pch_open_file): Use XNEWVEC. (pch_open_file): Use XRESIZEVEC. (read_file_guts): Use XNEWVEC and XRESIZEVEC. (dir_name_of_file): Use XNEWVEC. (make_cpp_file): Use XCNEW. (make_cpp_dir): Likewise. (allocate_file_hash_entries): USE XNEWVEC. (cpp_included): Cast return value of htab_find_with_hash. (append_file_to_dir): Use XNEWVEC. (read_filename_string): Likewise. Use XRESIZEVEC too. (read_name_map): Cast return value of alloca. Use XRESIZEVEC. (remap_filename): Use XNEWVEC. (struct pchf_entry): Move definition out of struct pchf_data. (_cpp_save_file_entries): Use XCNEWVAR. (_cpp_read_file_entries): Use XNEWVAR. * identifiers.c (alloc_node): Use XOBNEW. * init.c (cpp_create_reader): Use XCNEW. (cpp_init_builtins): Cast of b->value to enum builtin_type. (read_original_directory): Cast return value of alloca. * lex.c (add_line_note): Use XRESIZEVEC. (warn_about_normalization): Use XNEWVEC. (_cpp_lex_direct): Cast node->directive_index to (enum cpp_ttype). (new_buff): Use XNEWVEC. * line-map.c (linemap_add): Use XRESIZEVEC. * macro.c (builtin_macro): Cast return value of alloca. (paste_tokens): Likewise. (expand_arg): Use XNEWVEC and XRESIZEVEC. (_cpp_save_parameter): Use XRESIZEVEC. (create_iso_definition): Cast allocated storage to proper type. (_cpp_create_definition): Likewise. (cpp_macro_definition): Use XRESIZEVEC. * makedepend.c (add_clm): Use XNEW. (add_dir): Likewise. * mkdeps.c (munge): Use XNEWVEC. (deps_init): Use XCNEW. (deps_add_target): Use XRESIZEVEC. (deps_add_default_target): Cast return value of alloca. (deps_add_dep): Use XRESIZEVEC. (deps_add_vpath): Likewise. Use XNEWVEC too. (deps_restore): Likewise. * pch.c (save_idents): Use XNEW and XNEWVEC. (cpp_save_state): Use XNEW. (count_defs): Cast return value of htab_find. (write_defs): Likewise. (cpp_write_pch_deps): Use XNEWVEC. (collect_ht_nodes): Use XRESIZEVEC. (cpp_valid_state): Use XNEWVEC. (save_macros): Use XRESIZEVEC. Cast return value of xmemdup. * symtab.c (ht_create): Use XCNEW. (ht_lookup_with_hash): Cast return value of obstack_copy0. (ht_expand): Use XCNEWVEC. * system.h (HAVE_DESIGNATED_INITIALIZERS): False if __cplusplus. (bool): Do not define if __cplusplus. From-SVN: r100295
2005-05-28 15:52:48 +00:00
result = XCNEWVAR (struct pchf_data, result_size);
result->count = 0;
result->have_once_only = false;
for (f = pfile->all_files; f; f = f->next_file)
{
size_t count;
/* This should probably never happen, since if a read error occurred
the PCH file shouldn't be written... */
if (f->dont_read || f->err_no)
continue;
if (f->stack_count == 0)
continue;
[multiple changes] 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * c-opts.c (c_common_handle_option): Add -finput-charset. * c.opt: Ditto. * cppcharset.c (one_iso88591_to_utf8): Remove. (convert_iso88591_utf8): Ditto. (conversion_tab): Remove 8859-1 converter. (_cpp_input_to_utf8): Remove. (_cpp_init_iconv_buffer): Ditto. (_cpp_close_iconv_buffer): Ditto. (_cpp_convert_input): New function. (_cpp_default_encoding): Ditto. * cpphash.h: Add/remove prototypes for above. * cppfiles.c (read_file_guts): Use _cpp_convert_input. * cppinit.c (cpp_create_reader): Use _cpp_default_encoding for narrow execution and input character sets. * cpplib.c (cpp_push_buffer): Delete uses of removed functions. * doc/cppopts.texi: Document -finput-charset. 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * gcc.c-torture/execute/wchar_t-1.c: Add -finput-charset. 2004-01-29 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * testsuite/22_locale/collate/compare/wchar_t/2.cc: Remove xfail. Use -finput-charset. * testsuite/22_locale/collate/compare/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/compare/wchar_t/wrapped_locale.cc: Ditto * testsuite/22_locale/collate/hash/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_locale.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_locale.cc: Ditto. From-SVN: r77136
2004-02-02 20:20:58 +00:00
count = result->count++;
result->entries[count].once_only = f->once_only;
/* |= is avoided in the next line because of an HP C compiler bug */
result->have_once_only = result->have_once_only | f->once_only;
if (f->buffer_valid)
md5_buffer ((const char *)f->buffer,
f->st.st_size, result->entries[count].sum);
else
{
FILE *ff;
int oldfd = f->fd;
if (!open_file (f))
{
libcpp: Tweak to missing #include source location This patch tweaks the error message location for missing header files. Previously these read: test.c:1:17: fatal error: 404.h: No such file or directory #include "404.h" ^ compilation terminated. With this patch, the pertinent string is underlined: test.c:1:10: fatal error: 404.h: No such file or directory #include "404.h" ^~~~~~~ compilation terminated. gcc/testsuite/ChangeLog: * c-c++-common/missing-header-1.c: New test case. * c-c++-common/missing-header-2.c: New test case. * c-c++-common/missing-header-3.c: New test case. * c-c++-common/missing-header-4.c: New test case. libcpp/ChangeLog: * directives.c (do_include_common): Pass on "location" to _cpp_stack_include. * errors.c (cpp_diagnostic): Reimplement in terms of... (cpp_diagnostic_at): New function. (cpp_error_at): New function. (cpp_errno_filename): Add "loc" param and use it by using cpp_error_at rather than cpp_error. * files.c (find_file_in_dir): Add "loc" param and pass it to open_file_failed. (_cpp_find_file): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at, and pass it to find_file_in_dir and open_file_failed. (read_file_guts): Add "loc" param. Use it to convert calls to cpp_error to cpp_error_at. Pass it to cpp_errno_filename. (read_file): Add "loc" param. Pass it to open_file_failed and read_file_guts. (should_stack_file): Add "loc" param. Pass it to read_file. (_cpp_stack_file): Add "loc" param. Pass it to should_stack_file. (_cpp_stack_include): Add "loc" param. Pass it to _cpp_find_file and _cpp_stack_file. (open_file_failed): Add "loc" param. Pass it to cpp_errno_filename. (_cpp_fake_include): Add 0 as a source_location in call to _cpp_find_file. (_cpp_compare_file_date): Likewise. (cpp_push_include): Likewise for call to _cpp_stack_include. (cpp_push_default_include): Likewise. (_cpp_save_file_entries): Likewise for call to open_file_failed. (_cpp_has_header): Likewise for call to _cpp_find_file. * include/cpplib.h (cpp_errno_filename): Add source_location param. (cpp_error_at): New declaration. * init.c (cpp_read_main_file): Add 0 as a source_location in calls to _cpp_find_file and _cpp_stack_file. * internal.h (_cpp_find_file): Add source_location param. (_cpp_stack_file): Likewise. (_cpp_stack_include): Likewise. From-SVN: r237715
2016-06-22 15:29:21 +00:00
open_file_failed (pfile, f, 0, 0);
free (result);
return false;
}
ff = fdopen (f->fd, "rb");
md5_stream (ff, result->entries[count].sum);
fclose (ff);
f->fd = oldfd;
}
result->entries[count].size = f->st.st_size;
}
result_size = (sizeof (struct pchf_data)
+ sizeof (struct pchf_entry) * (result->count - 1));
[multiple changes] 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * c-opts.c (c_common_handle_option): Add -finput-charset. * c.opt: Ditto. * cppcharset.c (one_iso88591_to_utf8): Remove. (convert_iso88591_utf8): Ditto. (conversion_tab): Remove 8859-1 converter. (_cpp_input_to_utf8): Remove. (_cpp_init_iconv_buffer): Ditto. (_cpp_close_iconv_buffer): Ditto. (_cpp_convert_input): New function. (_cpp_default_encoding): Ditto. * cpphash.h: Add/remove prototypes for above. * cppfiles.c (read_file_guts): Use _cpp_convert_input. * cppinit.c (cpp_create_reader): Use _cpp_default_encoding for narrow execution and input character sets. * cpplib.c (cpp_push_buffer): Delete uses of removed functions. * doc/cppopts.texi: Document -finput-charset. 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * gcc.c-torture/execute/wchar_t-1.c: Add -finput-charset. 2004-01-29 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * testsuite/22_locale/collate/compare/wchar_t/2.cc: Remove xfail. Use -finput-charset. * testsuite/22_locale/collate/compare/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/compare/wchar_t/wrapped_locale.cc: Ditto * testsuite/22_locale/collate/hash/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_locale.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_locale.cc: Ditto. From-SVN: r77136
2004-02-02 20:20:58 +00:00
qsort (result->entries, result->count, sizeof (struct pchf_entry),
pchf_save_compare);
ret = fwrite (result, result_size, 1, fp) == 1;
free (result);
return ret;
}
/* Read the pchf_data structure from F. */
bool
_cpp_read_file_entries (cpp_reader *pfile ATTRIBUTE_UNUSED, FILE *f)
{
struct pchf_data d;
[multiple changes] 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * c-opts.c (c_common_handle_option): Add -finput-charset. * c.opt: Ditto. * cppcharset.c (one_iso88591_to_utf8): Remove. (convert_iso88591_utf8): Ditto. (conversion_tab): Remove 8859-1 converter. (_cpp_input_to_utf8): Remove. (_cpp_init_iconv_buffer): Ditto. (_cpp_close_iconv_buffer): Ditto. (_cpp_convert_input): New function. (_cpp_default_encoding): Ditto. * cpphash.h: Add/remove prototypes for above. * cppfiles.c (read_file_guts): Use _cpp_convert_input. * cppinit.c (cpp_create_reader): Use _cpp_default_encoding for narrow execution and input character sets. * cpplib.c (cpp_push_buffer): Delete uses of removed functions. * doc/cppopts.texi: Document -finput-charset. 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * gcc.c-torture/execute/wchar_t-1.c: Add -finput-charset. 2004-01-29 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * testsuite/22_locale/collate/compare/wchar_t/2.cc: Remove xfail. Use -finput-charset. * testsuite/22_locale/collate/compare/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/compare/wchar_t/wrapped_locale.cc: Ditto * testsuite/22_locale/collate/hash/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_locale.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_locale.cc: Ditto. From-SVN: r77136
2004-02-02 20:20:58 +00:00
if (fread (&d, sizeof (struct pchf_data) - sizeof (struct pchf_entry), 1, f)
!= 1)
return false;
[multiple changes] 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * c-opts.c (c_common_handle_option): Add -finput-charset. * c.opt: Ditto. * cppcharset.c (one_iso88591_to_utf8): Remove. (convert_iso88591_utf8): Ditto. (conversion_tab): Remove 8859-1 converter. (_cpp_input_to_utf8): Remove. (_cpp_init_iconv_buffer): Ditto. (_cpp_close_iconv_buffer): Ditto. (_cpp_convert_input): New function. (_cpp_default_encoding): Ditto. * cpphash.h: Add/remove prototypes for above. * cppfiles.c (read_file_guts): Use _cpp_convert_input. * cppinit.c (cpp_create_reader): Use _cpp_default_encoding for narrow execution and input character sets. * cpplib.c (cpp_push_buffer): Delete uses of removed functions. * doc/cppopts.texi: Document -finput-charset. 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * gcc.c-torture/execute/wchar_t-1.c: Add -finput-charset. 2004-01-29 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * testsuite/22_locale/collate/compare/wchar_t/2.cc: Remove xfail. Use -finput-charset. * testsuite/22_locale/collate/compare/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/compare/wchar_t/wrapped_locale.cc: Ditto * testsuite/22_locale/collate/hash/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_locale.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_locale.cc: Ditto. From-SVN: r77136
2004-02-02 20:20:58 +00:00
configure.ac: Check declarations for asprintf and vasprintf. * configure.ac: Check declarations for asprintf and vasprintf. * config.in: Regenerate. * configure: Likewise. * charset.c (conversion_loop): Use XRESIZEVEC. (convert_no_conversion): Likewise. (convert_using_iconv): Likewise. (init_iconv_desc): Cast return value of alloca. (cpp_host_to_exec_charset): Use XNEWVEC. (emit_numeric_escape): Use XRESIZEVEC. (cpp_interpret_string): Use XNEWVEC. (cpp_interpret_string): Use XRESIZEVEC. (_cpp_interpret_identifier): Cast return value of alloca. (_cpp_convert_input): Use XNEWVEC and XRESIZEVEC. * directives.c (glue_header_name): Use XNEWVEC and XRESIZEVEC. (parse_include): Use XNEWVEC. (insert_pragma_entry): Rename local variable "new" to "new_entry". (save_registered_pragmas): Cast return value of xmemdup. (destringize_and_run): Same for alloca. (parse_assertion): Likewise. (do_assert): Cast allocated storage to proper type. (cpp_define): Likewise. (_cpp_define_builtin): Likewise. (cpp_undef): Likewise. (handle_assertion): Likewise. (cpp_push_buffer): Rename local variable "new" to "new_buffer". * expr.c (CPP_UPLUS): Cast value to type cpp_ttype. (CPP_UMINUS): Likewise. (struct cpp_operator): Rename from struct operator. (_cpp_expand_op_stack): Use XRESIZEVEC. * files.c (pch_open_file): Use XNEWVEC. (pch_open_file): Use XRESIZEVEC. (read_file_guts): Use XNEWVEC and XRESIZEVEC. (dir_name_of_file): Use XNEWVEC. (make_cpp_file): Use XCNEW. (make_cpp_dir): Likewise. (allocate_file_hash_entries): USE XNEWVEC. (cpp_included): Cast return value of htab_find_with_hash. (append_file_to_dir): Use XNEWVEC. (read_filename_string): Likewise. Use XRESIZEVEC too. (read_name_map): Cast return value of alloca. Use XRESIZEVEC. (remap_filename): Use XNEWVEC. (struct pchf_entry): Move definition out of struct pchf_data. (_cpp_save_file_entries): Use XCNEWVAR. (_cpp_read_file_entries): Use XNEWVAR. * identifiers.c (alloc_node): Use XOBNEW. * init.c (cpp_create_reader): Use XCNEW. (cpp_init_builtins): Cast of b->value to enum builtin_type. (read_original_directory): Cast return value of alloca. * lex.c (add_line_note): Use XRESIZEVEC. (warn_about_normalization): Use XNEWVEC. (_cpp_lex_direct): Cast node->directive_index to (enum cpp_ttype). (new_buff): Use XNEWVEC. * line-map.c (linemap_add): Use XRESIZEVEC. * macro.c (builtin_macro): Cast return value of alloca. (paste_tokens): Likewise. (expand_arg): Use XNEWVEC and XRESIZEVEC. (_cpp_save_parameter): Use XRESIZEVEC. (create_iso_definition): Cast allocated storage to proper type. (_cpp_create_definition): Likewise. (cpp_macro_definition): Use XRESIZEVEC. * makedepend.c (add_clm): Use XNEW. (add_dir): Likewise. * mkdeps.c (munge): Use XNEWVEC. (deps_init): Use XCNEW. (deps_add_target): Use XRESIZEVEC. (deps_add_default_target): Cast return value of alloca. (deps_add_dep): Use XRESIZEVEC. (deps_add_vpath): Likewise. Use XNEWVEC too. (deps_restore): Likewise. * pch.c (save_idents): Use XNEW and XNEWVEC. (cpp_save_state): Use XNEW. (count_defs): Cast return value of htab_find. (write_defs): Likewise. (cpp_write_pch_deps): Use XNEWVEC. (collect_ht_nodes): Use XRESIZEVEC. (cpp_valid_state): Use XNEWVEC. (save_macros): Use XRESIZEVEC. Cast return value of xmemdup. * symtab.c (ht_create): Use XCNEW. (ht_lookup_with_hash): Cast return value of obstack_copy0. (ht_expand): Use XCNEWVEC. * system.h (HAVE_DESIGNATED_INITIALIZERS): False if __cplusplus. (bool): Do not define if __cplusplus. From-SVN: r100295
2005-05-28 15:52:48 +00:00
pchf = XNEWVAR (struct pchf_data, sizeof (struct pchf_data)
+ sizeof (struct pchf_entry) * (d.count - 1));
memcpy (pchf, &d, sizeof (struct pchf_data) - sizeof (struct pchf_entry));
if (fread (pchf->entries, sizeof (struct pchf_entry), d.count, f)
!= d.count)
return false;
return true;
}
/* The parameters for pchf_compare. */
struct pchf_compare_data
{
/* The size of the file we're looking for. */
off_t size;
/* The MD5 checksum of the file, if it's been computed. */
unsigned char sum[16];
/* Is SUM valid? */
bool sum_computed;
/* Do we need to worry about entries that don't have ONCE_ONLY set? */
bool check_included;
[multiple changes] 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * c-opts.c (c_common_handle_option): Add -finput-charset. * c.opt: Ditto. * cppcharset.c (one_iso88591_to_utf8): Remove. (convert_iso88591_utf8): Ditto. (conversion_tab): Remove 8859-1 converter. (_cpp_input_to_utf8): Remove. (_cpp_init_iconv_buffer): Ditto. (_cpp_close_iconv_buffer): Ditto. (_cpp_convert_input): New function. (_cpp_default_encoding): Ditto. * cpphash.h: Add/remove prototypes for above. * cppfiles.c (read_file_guts): Use _cpp_convert_input. * cppinit.c (cpp_create_reader): Use _cpp_default_encoding for narrow execution and input character sets. * cpplib.c (cpp_push_buffer): Delete uses of removed functions. * doc/cppopts.texi: Document -finput-charset. 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * gcc.c-torture/execute/wchar_t-1.c: Add -finput-charset. 2004-01-29 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * testsuite/22_locale/collate/compare/wchar_t/2.cc: Remove xfail. Use -finput-charset. * testsuite/22_locale/collate/compare/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/compare/wchar_t/wrapped_locale.cc: Ditto * testsuite/22_locale/collate/hash/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_locale.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_locale.cc: Ditto. From-SVN: r77136
2004-02-02 20:20:58 +00:00
/* The file that we're searching for. */
_cpp_file *f;
};
/* bsearch comparison function; look for D_P in E_P. */
static int
pchf_compare (const void *d_p, const void *e_p)
{
const struct pchf_entry *e = (const struct pchf_entry *)e_p;
struct pchf_compare_data *d = (struct pchf_compare_data *)d_p;
int result;
[multiple changes] 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * c-opts.c (c_common_handle_option): Add -finput-charset. * c.opt: Ditto. * cppcharset.c (one_iso88591_to_utf8): Remove. (convert_iso88591_utf8): Ditto. (conversion_tab): Remove 8859-1 converter. (_cpp_input_to_utf8): Remove. (_cpp_init_iconv_buffer): Ditto. (_cpp_close_iconv_buffer): Ditto. (_cpp_convert_input): New function. (_cpp_default_encoding): Ditto. * cpphash.h: Add/remove prototypes for above. * cppfiles.c (read_file_guts): Use _cpp_convert_input. * cppinit.c (cpp_create_reader): Use _cpp_default_encoding for narrow execution and input character sets. * cpplib.c (cpp_push_buffer): Delete uses of removed functions. * doc/cppopts.texi: Document -finput-charset. 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * gcc.c-torture/execute/wchar_t-1.c: Add -finput-charset. 2004-01-29 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * testsuite/22_locale/collate/compare/wchar_t/2.cc: Remove xfail. Use -finput-charset. * testsuite/22_locale/collate/compare/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/compare/wchar_t/wrapped_locale.cc: Ditto * testsuite/22_locale/collate/hash/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_locale.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_locale.cc: Ditto. From-SVN: r77136
2004-02-02 20:20:58 +00:00
result = memcmp (&d->size, &e->size, sizeof (off_t));
if (result != 0)
return result;
[multiple changes] 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * c-opts.c (c_common_handle_option): Add -finput-charset. * c.opt: Ditto. * cppcharset.c (one_iso88591_to_utf8): Remove. (convert_iso88591_utf8): Ditto. (conversion_tab): Remove 8859-1 converter. (_cpp_input_to_utf8): Remove. (_cpp_init_iconv_buffer): Ditto. (_cpp_close_iconv_buffer): Ditto. (_cpp_convert_input): New function. (_cpp_default_encoding): Ditto. * cpphash.h: Add/remove prototypes for above. * cppfiles.c (read_file_guts): Use _cpp_convert_input. * cppinit.c (cpp_create_reader): Use _cpp_default_encoding for narrow execution and input character sets. * cpplib.c (cpp_push_buffer): Delete uses of removed functions. * doc/cppopts.texi: Document -finput-charset. 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * gcc.c-torture/execute/wchar_t-1.c: Add -finput-charset. 2004-01-29 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * testsuite/22_locale/collate/compare/wchar_t/2.cc: Remove xfail. Use -finput-charset. * testsuite/22_locale/collate/compare/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/compare/wchar_t/wrapped_locale.cc: Ditto * testsuite/22_locale/collate/hash/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_locale.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_locale.cc: Ditto. From-SVN: r77136
2004-02-02 20:20:58 +00:00
if (! d->sum_computed)
{
_cpp_file *const f = d->f;
[multiple changes] 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * c-opts.c (c_common_handle_option): Add -finput-charset. * c.opt: Ditto. * cppcharset.c (one_iso88591_to_utf8): Remove. (convert_iso88591_utf8): Ditto. (conversion_tab): Remove 8859-1 converter. (_cpp_input_to_utf8): Remove. (_cpp_init_iconv_buffer): Ditto. (_cpp_close_iconv_buffer): Ditto. (_cpp_convert_input): New function. (_cpp_default_encoding): Ditto. * cpphash.h: Add/remove prototypes for above. * cppfiles.c (read_file_guts): Use _cpp_convert_input. * cppinit.c (cpp_create_reader): Use _cpp_default_encoding for narrow execution and input character sets. * cpplib.c (cpp_push_buffer): Delete uses of removed functions. * doc/cppopts.texi: Document -finput-charset. 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * gcc.c-torture/execute/wchar_t-1.c: Add -finput-charset. 2004-01-29 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * testsuite/22_locale/collate/compare/wchar_t/2.cc: Remove xfail. Use -finput-charset. * testsuite/22_locale/collate/compare/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/compare/wchar_t/wrapped_locale.cc: Ditto * testsuite/22_locale/collate/hash/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_locale.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_locale.cc: Ditto. From-SVN: r77136
2004-02-02 20:20:58 +00:00
md5_buffer ((const char *)f->buffer, f->st.st_size, d->sum);
d->sum_computed = true;
}
result = memcmp (d->sum, e->sum, 16);
if (result != 0)
return result;
if (d->check_included || e->once_only)
return 0;
else
return 1;
}
[multiple changes] 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * c-opts.c (c_common_handle_option): Add -finput-charset. * c.opt: Ditto. * cppcharset.c (one_iso88591_to_utf8): Remove. (convert_iso88591_utf8): Ditto. (conversion_tab): Remove 8859-1 converter. (_cpp_input_to_utf8): Remove. (_cpp_init_iconv_buffer): Ditto. (_cpp_close_iconv_buffer): Ditto. (_cpp_convert_input): New function. (_cpp_default_encoding): Ditto. * cpphash.h: Add/remove prototypes for above. * cppfiles.c (read_file_guts): Use _cpp_convert_input. * cppinit.c (cpp_create_reader): Use _cpp_default_encoding for narrow execution and input character sets. * cpplib.c (cpp_push_buffer): Delete uses of removed functions. * doc/cppopts.texi: Document -finput-charset. 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * gcc.c-torture/execute/wchar_t-1.c: Add -finput-charset. 2004-01-29 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * testsuite/22_locale/collate/compare/wchar_t/2.cc: Remove xfail. Use -finput-charset. * testsuite/22_locale/collate/compare/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/compare/wchar_t/wrapped_locale.cc: Ditto * testsuite/22_locale/collate/hash/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_locale.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_locale.cc: Ditto. From-SVN: r77136
2004-02-02 20:20:58 +00:00
/* Check that F is not in a list read from a PCH file (if any).
Assumes that f->buffer_valid is true. Return TRUE if the file
should not be read. */
static bool
check_file_against_entries (cpp_reader *pfile ATTRIBUTE_UNUSED,
_cpp_file *f,
bool check_included)
{
struct pchf_compare_data d;
[multiple changes] 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * c-opts.c (c_common_handle_option): Add -finput-charset. * c.opt: Ditto. * cppcharset.c (one_iso88591_to_utf8): Remove. (convert_iso88591_utf8): Ditto. (conversion_tab): Remove 8859-1 converter. (_cpp_input_to_utf8): Remove. (_cpp_init_iconv_buffer): Ditto. (_cpp_close_iconv_buffer): Ditto. (_cpp_convert_input): New function. (_cpp_default_encoding): Ditto. * cpphash.h: Add/remove prototypes for above. * cppfiles.c (read_file_guts): Use _cpp_convert_input. * cppinit.c (cpp_create_reader): Use _cpp_default_encoding for narrow execution and input character sets. * cpplib.c (cpp_push_buffer): Delete uses of removed functions. * doc/cppopts.texi: Document -finput-charset. 2004-02-02 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * gcc.c-torture/execute/wchar_t-1.c: Add -finput-charset. 2004-01-29 Eric Christopher <echristo@redhat.com> Zack Weinberg <zack@codesourcery.com> * testsuite/22_locale/collate/compare/wchar_t/2.cc: Remove xfail. Use -finput-charset. * testsuite/22_locale/collate/compare/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/compare/wchar_t/wrapped_locale.cc: Ditto * testsuite/22_locale/collate/hash/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/hash/wchar_t/wrapped_locale.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/2.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_env.cc: Ditto. * testsuite/22_locale/collate/transform/wchar_t/wrapped_locale.cc: Ditto. From-SVN: r77136
2004-02-02 20:20:58 +00:00
if (pchf == NULL
|| (! check_included && ! pchf->have_once_only))
return false;
d.size = f->st.st_size;
d.sum_computed = false;
d.f = f;
d.check_included = check_included;
return bsearch (&d, pchf->entries, pchf->count, sizeof (struct pchf_entry),
pchf_compare) != NULL;
}
Implement SD-6: SG10 Feature Test Recommendations 2014-10-01 Edward Smith-Rowland <3dw4rd@verizon.net> Implement SD-6: SG10 Feature Test Recommendations * internal.h (lexer_state, spec_nodes): Add in__has_include__. * directives.c: Support __has_include__ builtin. * expr.c (parse_has_include): New function to parse __has_include__ builtin; (eval_token()): Use it. * files.c (_cpp_has_header()): New funtion to look for header; (open_file_failed()): Not an error to not find a header file for __has_include__. * identifiers.c (_cpp_init_hashtable()): Add entry for __has_include__. * pch.c (cpp_read_state): Lookup __has_include__. * traditional.c (enum ls, _cpp_scan_out_logical_line()): Walk through __has_include__ statements. 2014-10-01 Edward Smith-Rowland <3dw4rd@verizon.net> Implement SD-6: SG10 Feature Test Recommendations * c-cppbuiltin.c (c_cpp_builtins()): Define language feature macros and the __has_header macro. 2014-10-01 Edward Smith-Rowland <3dw4rd@verizon.net> Implement SD-6: SG10 Feature Test Recommendations * include/bits/basic_string.h: Add __cpp_lib feature test macro. * include/bits/stl_algobase.h: Ditto. * include/bits/stl_function.h: Ditto. * include/bits/unique_ptr.h: Ditto. * include/std/chrono: Ditto. * include/std/complex: Ditto. * include/std/iomanip: Ditto. * include/std/shared_mutex: Ditto. * include/std/tuple: Ditto. * include/std/type_traits: Ditto. * include/std/utility: Ditto. * testsuite/experimental/feat-cxx14.cc: New. * testsuite/experimental/feat-lib-fund.cc: New. * testsuite/20_util/declval/requirements/1_neg.cc: Adjust. * testsuite/20_util/duration/literals/range.cc: Adjust. * testsuite/20_util/duration/requirements/typedefs_neg1.cc: Adjust. * testsuite/20_util/duration/requirements/typedefs_neg2.cc: Adjust. * testsuite/20_util/duration/requirements/typedefs_neg3.cc: Adjust. * testsuite/20_util/make_signed/requirements/typedefs_neg.cc: Adjust. * testsuite/20_util/make_unsigned/requirements/typedefs_neg.cc: Adjust. * testsuite/23_containers/array/tuple_interface/get_neg.cc: Adjust. * testsuite/23_containers/array/tuple_interface/tuple_element_neg.cc: Adjust. 2014-10-01 Edward Smith-Rowland <3dw4rd@verizon.net> Implement SD-6: SG10 Feature Test Recommendations * g++.dg/cpp1y/feat-cxx11-neg.C: New. * g++.dg/cpp1y/feat-cxx11.C: New. * g++.dg/cpp1y/feat-cxx14.C: New. * g++.dg/cpp1y/feat-cxx98.C: New. * g++.dg/cpp1y/feat-cxx98-neg.C: New. * g++.dg/cpp1y/phoobhar.h: New. * g++.dg/cpp1y/testinc/phoobhar.h: New. From-SVN: r215752
2014-10-01 11:49:23 +00:00
/* Return true if the file FNAME is found in the appropriate include file path
as indicated by ANGLE_BRACKETS. */
bool
_cpp_has_header (cpp_reader *pfile, const char *fname, int angle_brackets,
enum include_type type)
{
cpp_dir *start_dir = search_path_head (pfile, fname, angle_brackets, type,
/* suppress_diagnostic = */ true);
if (!start_dir)
return false;
[PR 80005] Fix __has_include __has_include is funky in that it is macro-like from the POV of #ifdef and friends, but lexes its parenthesize argument #include-like. We were failing the second part of that, because we used a forwarding macro to an internal name, and hence always lexed the argument in macro-parameter context. We componded that by not setting the right flag when lexing, so it didn't even know. Mostly users got lucky. This reimplements the handline. 1) Remove the forwarding, but declare object-like macros that expand to themselves. This satisfies the #ifdef requirement 2) Correctly set angled_brackets when lexing the parameter. This tells the lexer (a) <...> is a header name and (b) "..." is too (not a string). 3) Remove the in__has_include lexer state, just tell find_file that that's what's happenning, so it doesn't emit an error. We lose the (undocumented) ability to #undef __has_include. That may well have been an accident of implementation. There are no tests for it. We gain __has_include behaviour for all users of the preprocessors -- not just the C-family ones that defined a forwarding macro. libcpp/ PR preprocessor/80005 * include/cpplib.h (BT_HAS_ATTRIBUTE): Fix comment. * internal.h (struct lexer_state): Delete in__has_include field. (struct spec_nodes): Rename n__has_include{,_next}__ fields. (_cpp_defined_macro_p): New. (_cpp_find_file): Add has_include parm. * directives.c (lex_macro_node): Combine defined, __has_inline{,_next} checking. (do_ifdef, do_ifndef): Use _cpp_defined_macro_p. (_cpp_init_directives): Refactor. * expr.c (parse_defined): Use _cpp_defined_macro_p. (eval_token): Adjust parse_has_include calls. (parse_has_include): Add OP parameter. Reimplement. * files.c (_cpp_find_file): Add HAS_INCLUDE parm. Use it to inhibit error message. (_cpp_stack_include): Adjust _cpp_find_file call. (_cpp_fake_include, _cpp_compare_file_date): Likewise. (open_file_failed): Remove in__has_include check. (_cpp_has_header): Adjust _cpp_find_file call. * identifiers.c (_cpp_init_hashtable): Don't init __has_include{,_next} here ... * init.c (cpp_init_builtins): ... init them here. Define as macros. (cpp_read_main_file): Adjust _cpp_find_file call. * pch.c (cpp_read_state): Adjust __has_include{,_next} access. * traditional.c (_cpp_scan_out_locgical_line): Likewise. gcc/c-family/ PR preprocessor/80005 * c-cppbuiltins.c (c_cpp_builtins): Don't define __has_include{,_next}. gcc/testsuite/ PR preprocessor/80005 * g++.dg/cpp1y/feat-cxx14.C: Adjust. * g++.dg/cpp1z/feat-cxx17.C: Adjust. * g++.dg/cpp2a/feat-cxx2a.C: Adjust. * g++.dg/cpp/pr80005.C: New.
2020-01-20 05:39:59 -08:00
_cpp_file *file = _cpp_find_file (pfile, fname, start_dir, angle_brackets,
_cpp_FFK_HAS_INCLUDE, 0);
Implement SD-6: SG10 Feature Test Recommendations 2014-10-01 Edward Smith-Rowland <3dw4rd@verizon.net> Implement SD-6: SG10 Feature Test Recommendations * internal.h (lexer_state, spec_nodes): Add in__has_include__. * directives.c: Support __has_include__ builtin. * expr.c (parse_has_include): New function to parse __has_include__ builtin; (eval_token()): Use it. * files.c (_cpp_has_header()): New funtion to look for header; (open_file_failed()): Not an error to not find a header file for __has_include__. * identifiers.c (_cpp_init_hashtable()): Add entry for __has_include__. * pch.c (cpp_read_state): Lookup __has_include__. * traditional.c (enum ls, _cpp_scan_out_logical_line()): Walk through __has_include__ statements. 2014-10-01 Edward Smith-Rowland <3dw4rd@verizon.net> Implement SD-6: SG10 Feature Test Recommendations * c-cppbuiltin.c (c_cpp_builtins()): Define language feature macros and the __has_header macro. 2014-10-01 Edward Smith-Rowland <3dw4rd@verizon.net> Implement SD-6: SG10 Feature Test Recommendations * include/bits/basic_string.h: Add __cpp_lib feature test macro. * include/bits/stl_algobase.h: Ditto. * include/bits/stl_function.h: Ditto. * include/bits/unique_ptr.h: Ditto. * include/std/chrono: Ditto. * include/std/complex: Ditto. * include/std/iomanip: Ditto. * include/std/shared_mutex: Ditto. * include/std/tuple: Ditto. * include/std/type_traits: Ditto. * include/std/utility: Ditto. * testsuite/experimental/feat-cxx14.cc: New. * testsuite/experimental/feat-lib-fund.cc: New. * testsuite/20_util/declval/requirements/1_neg.cc: Adjust. * testsuite/20_util/duration/literals/range.cc: Adjust. * testsuite/20_util/duration/requirements/typedefs_neg1.cc: Adjust. * testsuite/20_util/duration/requirements/typedefs_neg2.cc: Adjust. * testsuite/20_util/duration/requirements/typedefs_neg3.cc: Adjust. * testsuite/20_util/make_signed/requirements/typedefs_neg.cc: Adjust. * testsuite/20_util/make_unsigned/requirements/typedefs_neg.cc: Adjust. * testsuite/23_containers/array/tuple_interface/get_neg.cc: Adjust. * testsuite/23_containers/array/tuple_interface/tuple_element_neg.cc: Adjust. 2014-10-01 Edward Smith-Rowland <3dw4rd@verizon.net> Implement SD-6: SG10 Feature Test Recommendations * g++.dg/cpp1y/feat-cxx11-neg.C: New. * g++.dg/cpp1y/feat-cxx11.C: New. * g++.dg/cpp1y/feat-cxx14.C: New. * g++.dg/cpp1y/feat-cxx98.C: New. * g++.dg/cpp1y/feat-cxx98-neg.C: New. * g++.dg/cpp1y/phoobhar.h: New. * g++.dg/cpp1y/testinc/phoobhar.h: New. From-SVN: r215752
2014-10-01 11:49:23 +00:00
return file->err_no != ENOENT;
}
diagnostics: Support for -finput-charset [PR93067] Adds the logic to handle -finput-charset in layout_get_source_line(), so that source lines are converted from their input encodings prior to being output by diagnostics machinery. Also adds the ability to strip a UTF-8 BOM similarly. gcc/c-family/ChangeLog: PR other/93067 * c-opts.c (c_common_input_charset_cb): New function. (c_common_post_options): Call new function diagnostic_initialize_input_context(). gcc/d/ChangeLog: PR other/93067 * d-lang.cc (d_input_charset_callback): New function. (d_init): Call new function diagnostic_initialize_input_context(). gcc/fortran/ChangeLog: PR other/93067 * cpp.c (gfc_cpp_post_options): Call new function diagnostic_initialize_input_context(). gcc/ChangeLog: PR other/93067 * coretypes.h (typedef diagnostic_input_charset_callback): Declare. * diagnostic.c (diagnostic_initialize_input_context): New function. * diagnostic.h (diagnostic_initialize_input_context): Declare. * input.c (default_charset_callback): New function. (file_cache::initialize_input_context): New function. (file_cache_slot::create): Added ability to convert the input according to the input context. (file_cache::file_cache): Initialize the new input context. (class file_cache_slot): Added new m_alloc_offset member. (file_cache_slot::file_cache_slot): Initialize the new member. (file_cache_slot::~file_cache_slot): Handle potentially offset buffer. (file_cache_slot::maybe_grow): Likewise. (file_cache_slot::needs_read_p): Handle NULL fp, which is now possible. (file_cache_slot::get_next_line): Likewise. * input.h (class file_cache): Added input context member. libcpp/ChangeLog: PR other/93067 * charset.c (init_iconv_desc): Adapt to permit PFILE argument to be NULL. (_cpp_convert_input): Likewise. Also move UTF-8 BOM logic to... (cpp_check_utf8_bom): ...here. New function. (cpp_input_conversion_is_trivial): New function. * files.c (read_file_guts): Allow PFILE argument to be NULL. Add INPUT_CHARSET argument as an alternate source of this information. (read_file): Pass the new argument to read_file_guts. (cpp_get_converted_source): New function. * include/cpplib.h (struct cpp_converted_source): Declare. (cpp_get_converted_source): Declare. (cpp_input_conversion_is_trivial): Declare. (cpp_check_utf8_bom): Declare. gcc/testsuite/ChangeLog: PR other/93067 * gcc.dg/diagnostic-input-charset-1.c: New test. * gcc.dg/diagnostic-input-utf8-bom.c: New test.
2021-08-24 19:30:44 -04:00
/* Read a file and convert to input charset, the same as if it were being read
by a cpp_reader. */
cpp_converted_source
cpp_get_converted_source (const char *fname, const char *input_charset)
{
cpp_converted_source res = {};
_cpp_file file = {};
file.fd = -1;
file.name = lbasename (fname);
file.path = fname;
if (!open_file (&file))
return res;
const bool ok = read_file_guts (NULL, &file, 0, input_charset);
close (file.fd);
if (!ok)
return res;
res.to_free = (char *) file.buffer_start;
res.data = (char *) file.buffer;
res.len = file.st.st_size;
return res;
}