The frange setter does all its work in trees. This incurs a penalty
for the real_value variants because they must wrap their arguments
into a tree and pass it to the tree setter, which will then do the
opposite. This is leftovers from the irange setter.
Even though the we still need constructors taking trees so we can
interact with the tree world, there's no sense penalizing the rest of
the implementation.
Tested on x86-64 Linux.
gcc/ChangeLog:
* value-range.cc (frange::set): Swap setters such that the one
accepting REAL_VALUE_TYPE does all the work.
We already disable all debug mode checks for C++11 and later, so we can
easily make everything constexpr. This fixes the FAIL results for the
new tests when using -D_GLIBCXX_DEBUG.
Also fix some other tests failing with non-default test flags.
libstdc++-v3/ChangeLog:
* include/debug/bitset (__debug::bitset): Add constexpr to all
member functions.
(operator&, operator|, operator^): Add inline and constexpr.
(operator>>, operator<<): Add inline.
* testsuite/20_util/bitset/access/constexpr.cc: Only check using
constexpr std::string for the cxx11 ABI.
* testsuite/20_util/bitset/cons/constexpr_c++23.cc: Likewise.
* testsuite/20_util/headers/bitset/synopsis.cc: Check constexpr
for C++23.
This makes to_string approximately twice as fast at any optimization
level. Instead of iterating through every bit, jump straight to the next
bit that is set, by using _Find_first and _Find_next.
libstdc++-v3/ChangeLog:
* include/std/bitset (bitset::_M_copy_to_string): Find set bits
instead of iterating over individual bits.
This patch adds -mcpu/-mtune support for the Arm Neoverse V2 core.
This updates the internal references to "demeter", but leaves "demeter" as an
accepted value to -mcpu/-mtune as it appears in the released GCC 12 series.
Bootstrapped and tested on aarch64-none-linux-gnu.
gcc/ChangeLog:
* config/aarch64/aarch64-cores.def (neoverse-v2): New entry.
(demeter): Update tunings to neoversev2.
* config/aarch64/aarch64-tune.md: Regenerate.
* config/aarch64/aarch64.cc (demeter_addrcost_table): Rename to
neoversev2_addrcost_table.
(demeter_regmove_cost): Rename to neoversev2_addrcost_table.
(demeter_advsimd_vector_cost): Rename to neoversev2_advsimd_vector_cost.
(demeter_sve_vector_cost): Rename to neoversev2_sve_vector_cost.
(demeter_scalar_issue_info): Rename to neoversev2_scalar_issue_info.
(demeter_advsimd_issue_info): Rename to neoversev2_advsimd_issue_info.
(demeter_sve_issue_info): Rename to neoversev2_sve_issue_info.
(demeter_vec_issue_info): Rename to neoversev2_vec_issue_info.
Update references to above.
(demeter_vector_cost): Rename to neoversev2_vector_cost.
(demeter_tunings): Rename to neoversev2_tunings.
(aarch64_vec_op_count::rename_cycles_per_iter): Use
neoversev2_sve_issue_info instead of demeter_sve_issue_info.
* doc/invoke.texi (AArch64 Options): Document neoverse-v2.
Often times when a check_function_body check fails it can be quite hard to
figure out why as no additional information is provided.
This changes it so that on failures it prints out the regex expression it's
using and the text it's comparing against to the verbose log.
This makes it much easier to figure out why a test has failed.
gcc/testsuite/ChangeLog:
* lib/scanasm.exp (check_function_body): Add debug output to verbose log
on failure.
Similarly to how we drop NANs to UNDEFINED when -ffinite-math-only, I
think we can drop the numbers outside of the min/max representable
numbers to the representable number.
This means the endpoings to VR_VARYING for -ffinite-math-only can now
be the min/max representable, instead of -INF and +INF.
Saturating in the setter means that the upcoming implementation for
binary operators no longer have to worry about doing the right
thing for -ffinite-math-only. If the range goes outside the limits,
it'll get chopped down.
Tested on x86-64 Linux.
gcc/ChangeLog:
* range-op-float.cc (build_le): Use vrp_val_*.
(build_lt): Same.
(build_ge): Same.
(build_gt): Same.
* value-range.cc (frange::set): Chop ranges outside of the
representable numbers for -ffinite-math-only.
(frange::normalize_kind): Use vrp_val*.
(frange::verify_range): Same.
(frange::set_nonnegative): Same.
(range_tests_floats): Remove tests that depend on -INF and +INF.
* value-range.h (real_max_representable): Add prototype.
(real_min_representable): Same.
(vrp_val_max): Set max representable number for
-ffinite-math-only.
(vrp_val_min): Same but for min.
(frange::set_varying): Use vrp_val*.
We currently have no way of dumping REAL_VALUE_TYPEs when debugging.
Tested on a gdb session examining the real value 10.0:
(gdb) p min
$9 = {cl = 1, decimal = 0, sign = 0, signalling = 0, canonical = 0, uexp = 4, sig = {0, 0, 11529215046068469760}}
(gdb) p debug (min)
0x0.ap+4
gcc/ChangeLog:
* real.cc (debug): New.
It has been suggested that if we start bumping numbers by an ULP when
calculating open ranges (for example the numbers less than 3.0) that
dumping these will become increasingly harder to read, and instead we
should opt for the hex representation. I still find the floating
point representation easier to read for most numbers, but perhaps we
could have both?
With this patch this is the representation for [15.0, 20.0]:
[frange] float [1.5e+1 (0x0.fp+4), 2.0e+1 (0x0.ap+5)]
Would you find this useful, or should we stick to the hex
representation only?
Tested on x86-64 Linux.
gcc/ChangeLog:
* value-range-pretty-print.cc (vrange_printer::print_real_value): New.
(vrange_printer::visit): Call print_real_value.
* value-range-pretty-print.h: New print_real_value.
Fixes -flto-compression option:
- -flto-compression-level=<number> Use z Use zlib/zstd compression level <number> for IL.
+ -flto-compression-level=<0,19> Use zlib/zstd compression level <number> for IL.
gcc/ChangeLog:
* common.opt: Update -flto-compression-level documentation.
* opts.cc (print_filtered_help): Do not append range to an
option that uses \t syntax.
On Thu, Sep 22, 2022 at 01:10:08PM +0200, Richard Biener via Gcc-patches wrote:
> * g++.dg/tree-ssa/pr106922.C: Adjust.
> --- a/gcc/testsuite/g++.dg/tree-ssa/pr106922.C
> +++ b/gcc/testsuite/g++.dg/tree-ssa/pr106922.C
> @@ -87,5 +87,4 @@ void testfunctionfoo() {
> }
> }
>
> -// { dg-final { scan-tree-dump-times "Found fully redundant value" 4 "pre" { xfail { ! lp64 } } } }
> -// { dg-final { scan-tree-dump-not "m_initialized" "cddce3" { xfail { ! lp64 } } } }
> +// { dg-final { scan-tree-dump-not "m_initialized" "dce3" } }
I've noticed
+UNRESOLVED: g++.dg/tree-ssa/pr106922.C -std=gnu++20 scan-tree-dump-not dce3 "m_initialized"
+UNRESOLVED: g++.dg/tree-ssa/pr106922.C -std=gnu++2b scan-tree-dump-not dce3 "m_initialized"
with this change, both on x86_64 and i686.
The dump is still cddce3, additionally as the last reference to the pre
dump is gone, not sure it is worth creating that dump.
With the following patch, there aren't FAILs nor UNRESOLVED tests with
GXX_TESTSUITE_STDS=98,11,14,17,20,2b make check-g++ RUNTESTFLAGS="--target_board=unix\{-m32,-m64\} dg.exp='pr106922.C'"
2022-09-23 Jakub Jelinek <jakub@redhat.com>
PR tree-optimization/106922
* g++.dg/tree-ssa/pr106922.C: Scan in cddce3 dump rather than
dce3. Remove -fdump-tree-pre-details from dg-options.
When looking at the attribs code, I've noticed weird diagnostics
like
int a __attribute__((section ("foo", "bar")));
a.c:1:1: error: wrong number of arguments specified for ‘section’ attribute
1 | int a __attribute__((section ("foo", "bar")));
| ^~~
a.c:1:1: note: expected between 1 and 1, found 2
As roughly 50% of attributes that accept any arguments have
spec->min_length == spec->max_length, I think it is worth it to have
separate wording for such common case and just write simpler
a.c:1:1: note: expected 1, found 2
2022-09-23 Jakub Jelinek <jakub@redhat.com>
* attribs.cc (decl_attributes): Improve diagnostics, instead of
saying expected between 1 and 1, found 2 just say expected 1, found 2.
We ICE in the code added in r12-7117: type_build_dtor_call gets
the error_mark_node because the type of 'prev' wasn't declared.
PR c++/106983
gcc/cp/ChangeLog:
* typeck2.cc (split_nonconstant_init_1): Check TYPE_P.
gcc/testsuite/ChangeLog:
* g++.dg/other/error36.C: New test.
In some scenarios (e.g., when mixing gcc and clang code), it can
happen that frames are deregistered after the lookup structure
has already been destroyed. That in itself would be fine, but
it triggers an assert in __deregister_frame_info_bases that
expects to find the frame.
To avoid that, we now remember that the btree as already been
destroyed and disable the assert in that case.
libgcc/ChangeLog:
* unwind-dw2-fde.c: (release_register_frames) Remember
when the btree has been destroyed.
(__deregister_frame_info_bases) Disable the assert when
shutting down.
Also, as the last builtin remaining, also remove the builtin infrastrucure
routines from fold_using_range.
* gimple-range-fold.cc (range_of_range_op): Handle no operands.
(range_of_call): Do not check for builtins.
(fold_using_range::range_of_builtin_call): Delete.
(fold_using_range::range_of_builtin_int_call): Delete.
* gimple-range-fold.h: Adjust prototypes.
* gimple-range-op.cc (class cfn_parity): New.
(gimple_range_op_handler::maybe_builtin_call): Set arguments.
* gimple-range-fold.cc (range_of_builtin_int_call): Remove case
for CFN_GOACC_DIM_*.
* gimple-range-op.cc (class cfn_goacc_dim): New.
(gimple_range_op_handler::maybe_builtin_call): Set arguments.
* gimple-range-fold.cc (range_of_builtin_int_call): Remove case
for CFN_BUILT_IN_STRLEN.
* gimple-range-op.cc (class cfn_strlen): New.
(gimple_range_op_handler::maybe_builtin_call): Set arguments.
* gimple-range-fold.cc (range_of_builtin_int_call): Remove case
for CFN_BUILT_IN_CLRSB.
* gimple-range-op.cc (class cfn_clrsb): New.
(gimple_range_op_handler::maybe_builtin_call): Set arguments.
* gimple-range-fold.cc (range_of_builtin_int_call): Remove case
for CFN_CTZ.
* gimple-range-op.cc (class cfn_ctz): New.
(gimple_range_op_handler::maybe_builtin_call): Set arguments.
* gimple-range-fold.cc (range_of_builtin_int_call): Remove case
for CFN_CLZ.
* gimple-range-op.cc (class cfn_clz): New.
(gimple_range_op_handler::maybe_builtin_call): Set arguments.
* gimple-range-fold.cc (range_of_builtin_int_call): Remove case
for CFN_FFS and CFN_POPCOUNT.
* gimple-range-op.cc (class cfn_pocount): New.
(gimple_range_op_handler::maybe_builtin_call): Set arguments.
* gimple-range-fold.cc (get_letter_range): Move to new class.
(range_of_builtin_int_call): Remove case for CFN_BUILT_IN_TOUPPER
and CFN_BUILT_IN_TOLOWER.
* gimple-range-op.cc (class cfn_toupper_tolower): New.
(gimple_range_op_handler::maybe_builtin_call): Set arguments.
* gimple-range-fold.cc (range_of_builtin_int_call): Remove case
for CFN_BUILT_IN_SIGNBIT.
* gimple-range-op.cc (class cfn_signbit): New.
(gimple_range_op_handler::maybe_builtin_call): Set arguments.
Convert CFN_BUILT_IN_CONSTANT_P as first POC.
* gimple-range-fold.cc
(fold_using_range::range_of_builtin_int_call): Remove case for
CFN_BUILT_IN_CONSTANT_P.
* gimple-range-op.cc (gimple_range_op_handler::supported_p):
Check if a call also creates a range-op object.
(gimple_range_op_handler): Also check builtin calls.
(class cfn_constant_float_p): New. Float CFN_BUILT_IN_CONSTANT_P.
(class cfn_constant_p): New. Integral CFN_BUILT_IN_CONSTANT_P.
(gimple_range_op_handler::maybe_builtin_call): Set arguments and
handler for supported built-in calls.
* gimple-range-op.h (maybe_builtin_call): New prototype.
The fold_range routine in range-ops returns FALSE if the operation
fails. There are a few places which assume the operation was
successful. Fix those.
* gimple-range-fold.cc (range_of_range_op): Set result to
VARYING if the call to fold_range fails.
* tree-data-ref.cc (compute_distributive_range): Ditto.
* tree-vrp.cc (range_fold_binary_expr): Ditto.
(range_fold_unary_expr): Ditto.
* value-query.cc (range_query::get_tree_range): Ditto.
Unary operations require op2 to be the range of the type of the LHS.
This is so the type for the LHS can be properly set.
* range-op-float.cc (range_operator_float::fold_range): New base
method for "int = float op int".
* range-op.cc (range_op_handler::fold_range): New case.
* range-op.h: Update prototypes.
Unary operations pass the type of operand 1 into op1_range. If that
range is undefined, the routine blindly picks the type of operand 2,
which in the case of a unary op, does not exist and traps.
* gimple-range-op.cc (gimple_range_op_handler::calc_op1): Use
operand 1 for second range if there is no operand 2.
Range-ops is meant to be IL independent. Some gimple processing has
be placed in range-ops, and some is located in gori. Split it all into
a file and isolate it in a new class gimple_range_op_handler.
* Makefile.in (OBJS): Add gimple-range-op.o.
* gimple-range-edge.cc (gimple_outgoing_range_stmt_p): Use
gimple_range_op_handler.
* gimple-range-fold.cc (gimple_range_base_of_assignment): Move
to a method in gimple_range_op_handler.
(gimple_range_operand1): Ditto.
(gimple_range_operand2): Ditto.
(fold_using_range::fold_stmt): Use gimple_range_op_handler.
(fold_using_range::range_of_range_op): Ditto.
(fold_using_range::relation_fold_and_or): Ditto.
(fur_source::register_outgoing_edges): Ditto.
(gimple_range_ssa_names): Relocate to gimple-range-op.cc.
* gimple-range-fold.h: Adjust prototypes.
* gimple-range-gori.cc (gimple_range_calc_op1): Move
to a method in gimple_range_op_handler.
(gimple_range_calc_op2): Ditto.
(gori_compute::compute_operand_range): Use
gimple_range_op_handler.
(gori_compute::compute_logical_operands): Ditto.
(compute_operand1_range): Ditto.
(gori_compute::compute_operand2_range): Ditto.
(gori_compute::compute_operand1_and_operand2_range): Ditto.
* gimple-range-gori.h: Adjust protoypes.
* gimple-range-op.cc: New. Supply gimple_range_op_handler methods.
* gimple-range-op.h: New. Supply gimple_range_op_handler class.
* gimple-range.cc (gimple_ranger::prefill_name): Use
gimple_range_op_handler.
(gimple_ranger::prefill_stmt_dependencies): Ditto.
* gimple-range.h: Include gimple-range-op.h.
* range-op.cc (range_op_handler::range_op_handler): Adjust and
remove gimple * parameter option.
* range-op.h: Adjust prototypes.
Range_op_handler currently stores a tree code and a type. It defers
checking to see if there is a valid handler until asked.
This change checks at constuctor time and store a pointer to
the handler if there is one.
* range-op.cc (range_op_handler::set_op_handler): Set new fields.
(ange_op_handler::range_op_handler): Likewise.
(range_op_handler::operator bool): Remove.
(range_op_handler::fold_range): Use appropriate handler.
(range_op_handler::op1_range): Likewise.
(range_op_handler::op2_range): Likewise.
(range_op_handler::lhs_op1_relation): Likewise.
(range_op_handler::lhs_op2_relation): Likewise.
(range_op_handler::op1_op2_relation): Likewise.
* range-op.h (class range_op_handler): Store handler pointers.
(range_op_handler:: operator bool): Inline.
When the original patch was applied, I missed a spot which could
also be rewritten to use gimple_range_ssa_names.
* tree-ssa-threadbackward.cc
(back_threader::find_paths_to_names): Replace sequence with
a call to gimple_range_ssa_names.
Also add _GLIBCXX_HOSTED checks to simplify making <bitset>
freestanding in the near future.
libstdc++-v3/ChangeLog:
* include/std/bitset (bitset): Add constexpr for C++23. Guard
members using std::string with _GLIBCXX_HOSTED.
* include/std/version (__cpp_lib_constexpr_bitset): Define.
* testsuite/20_util/bitset/access/constexpr.cc: New test.
* testsuite/20_util/bitset/cons/constexpr_c++23.cc: New test.
* testsuite/20_util/bitset/count/constexpr.cc: New test.
* testsuite/20_util/bitset/ext/constexpr.cc: New test.
* testsuite/20_util/bitset/operations/constexpr_c++23.cc: New test.
* testsuite/20_util/bitset/version.cc: New test.
In C++03 std::bitset was in the Container clause, but since C++11 it has
been in the Utilties clause. This moves the tests to the 20_util
directory, where most people probably expect to find them.
Also create 'access', 'observers', and 'io' subdirectories and group
some tests under there, rather than having one directory per function
name, and only a single test in that directory.
libstdc++-v3/ChangeLog:
* testsuite/23_containers/bitset/18604.cc: Moved to...
* testsuite/20_util/bitset/18604.cc: ...here.
* testsuite/23_containers/bitset/45713.cc: Moved to...
* testsuite/20_util/bitset/45713.cc: ...here.
* testsuite/23_containers/bitset/to_string/dr396.cc: Moved to...
* testsuite/20_util/bitset/access/dr396.cc: ...here.
* testsuite/23_containers/bitset/to_string/1.cc: Moved to...
* testsuite/20_util/bitset/access/to_string.cc: ...here.
* testsuite/23_containers/bitset/to_ullong/1.cc: Moved to...
* testsuite/20_util/bitset/access/to_ullong.cc: ...here.
* testsuite/23_containers/bitset/to_ulong/1.cc: Moved to...
* testsuite/20_util/bitset/access/to_ulong.cc: ...here.
* testsuite/23_containers/bitset/cons/1.cc: Moved to...
* testsuite/20_util/bitset/cons/1.cc: ...here.
* testsuite/23_containers/bitset/cons/16020.cc: Moved to...
* testsuite/20_util/bitset/cons/16020.cc: ...here.
* testsuite/23_containers/bitset/cons/2.cc: Moved to...
* testsuite/20_util/bitset/cons/2.cc: ...here.
* testsuite/23_containers/bitset/cons/3.cc: Moved to...
* testsuite/20_util/bitset/cons/3.cc: ...here.
* testsuite/23_containers/bitset/cons/38244.cc: Moved to...
* testsuite/20_util/bitset/cons/38244.cc: ...here.
* testsuite/23_containers/bitset/cons/50268.cc: Moved to...
* testsuite/20_util/bitset/cons/50268.cc: ...here.
* testsuite/23_containers/bitset/cons/6282.cc: Moved to...
* testsuite/20_util/bitset/cons/6282.cc: ...here.
* testsuite/23_containers/bitset/cons/constexpr.cc: Moved to...
* testsuite/20_util/bitset/cons/constexpr.cc: ...here.
* testsuite/23_containers/bitset/cons/dr1325-1.cc: Moved to...
* testsuite/20_util/bitset/cons/dr1325-1.cc: ...here.
* testsuite/23_containers/bitset/cons/dr1325-2.cc: Moved to...
* testsuite/20_util/bitset/cons/dr1325-2.cc: ...here.
* testsuite/23_containers/bitset/cons/dr396.cc: Moved to...
* testsuite/20_util/bitset/cons/dr396.cc: ...here.
* testsuite/23_containers/bitset/debug/invalidation/1.cc: Moved to...
* testsuite/20_util/bitset/debug/invalidation/1.cc: ...here.
* testsuite/23_containers/bitset/ext/15361.cc: Moved to...
* testsuite/20_util/bitset/ext/15361.cc: ...here.
* testsuite/23_containers/bitset/hash/1.cc: Moved to...
* testsuite/20_util/bitset/hash/1.cc: ...here.
* testsuite/23_containers/bitset/input/1.cc: Moved to...
* testsuite/20_util/bitset/io/input.cc: ...here.
* testsuite/23_containers/bitset/count/6124.cc: Moved to...
* testsuite/20_util/bitset/observers/6124.cc: ...here.
* testsuite/23_containers/bitset/all/1.cc: Moved to...
* testsuite/20_util/bitset/observers/all.cc: ...here.
* testsuite/23_containers/bitset/test/1.cc: Moved to...
* testsuite/20_util/bitset/observers/test.cc: ...here.
* testsuite/23_containers/bitset/operations/1.cc: Moved to...
* testsuite/20_util/bitset/operations/1.cc: ...here.
* testsuite/23_containers/bitset/operations/13838.cc: Moved to...
* testsuite/20_util/bitset/operations/13838.cc: ...here.
* testsuite/23_containers/bitset/operations/2.cc: Moved to...
* testsuite/20_util/bitset/operations/2.cc: ...here.
* testsuite/23_containers/bitset/operations/96303.cc: Moved to...
* testsuite/20_util/bitset/operations/96303.cc: ...here.
* testsuite/23_containers/bitset/operations/constexpr-2.cc: Moved to...
* testsuite/20_util/bitset/operations/constexpr-2.cc: ...here.
* testsuite/23_containers/bitset/operations/constexpr.cc: Moved to...
* testsuite/20_util/bitset/operations/constexpr.cc: ...here.
* testsuite/23_containers/bitset/requirements/constexpr_functions.cc: Moved to...
* testsuite/20_util/bitset/requirements/constexpr_functions.cc: ...here.
* testsuite/23_containers/bitset/requirements/explicit_instantiation/1.cc: Moved to...
* testsuite/20_util/bitset/requirements/explicit_instantiation/1.cc: ...here.
* testsuite/23_containers/bitset/requirements/explicit_instantiation/1_c++0x.cc: Moved to...
* testsuite/20_util/bitset/requirements/explicit_instantiation/1_c++0x.cc: ...here.
* testsuite/23_containers/headers/bitset/synopsis.cc: Moved to...
* testsuite/20_util/headers/bitset/synopsis.cc: ...here.
With partial variable template specializations, it looks like we
stream the VAR_DECL (i.e. the DECL_TEMPLATE_RESULT of the corresponding
TEMPLATE_DECL) since process_partial_specialization adds it to the
specializations table, but we end up never streaming the corresponding
TEMPLATE_DECL itself that's reachable only from the primary template's
DECL_TEMPLATE_SPECIALIZATIONS list, which leads to this list being
incomplete on stream-in.
The modules machinery already has special logic for streaming partial
specializations of class templates; this patch attempts to generalize
it to handle those of variable templates as well.
PR c++/106826
gcc/cp/ChangeLog:
* module.cc (trees_out::decl_value): Use get_template_info in
the MK_partial case to handle both VAR_DECL and TYPE_DECL.
(trees_out::key_mergeable): Likewise.
(trees_in::key_mergeable): Likewise.
(has_definition): Consider DECL_INITIAL of a partial variable
template specialization.
(depset:#️⃣:make_dependency): Handle partial variable template
specializations too.
gcc/testsuite/ChangeLog:
* g++.dg/modules/partial-2_a.C: New test.
* g++.dg/modules/partial-2_b.C: New test.
The option value is legacy and probably not used at all,
thus ignore it.
gcc/ChangeLog:
* configure: Regenerate.
* configure.ac: Simplify to gcc_cv_ld_compress_debug={0,1}
and gcc_cv_as_compress_debug={0,1}.
* doc/invoke.texi: Document the removal.
* gcc.cc (LINK_COMPRESS_DEBUG_SPEC): Simplify and ignore
zlib-gnu.
(ASM_COMPRESS_DEBUG_SPEC): Likewise.
Co-Authored-By: Fangrui Song <i@maskray.me>
I added c_expr::m_decimal in r13-2386-gbedfca647a9e9c1a as part of the
implementation of -Wxor-used-as-pow, but I missed various places where
the field needed to be initialized.
Fixed thusly.
gcc/c-family/ChangeLog:
PR c/106830
* c-warn.cc (check_for_xor_used_as_pow): Don't try checking
values that don't fit in uhwi.
gcc/c/ChangeLog:
PR c/106830
* c-parser.cc (c_parser_initelt): Initialize m_decimal.
(c_parser_cast_expression): Likewise.
(c_parser_alignof_expression): Likewise.
(c_parser_postfix_expression_after_paren_type): Likewise.
(c_parser_postfix_expression_after_primary): Likewise.
(c_parser_expression): Likewise.
(c_parser_omp_variable_list): Likewise.
(c_parser_transaction_expression): Likewise.
* c-tree.h (c_expr::set_error): Likewise.
* c-typeck.cc (c_expr_sizeof_expr): Likewise.
(parser_build_unary_op): Likewise.
(parser_build_binary_op): Likewise.
(digest_init): Likewise.
(pop_init_level): Likewise.
* gimple-parser.cc (c_parser_gimple_call_internal): Likewise.
gcc/testsuite/ChangeLog:
PR c/106830
* gcc.dg/Wxor-used-as-pow-pr106830.c: New test.
Signed-off-by: David Malcolm <dmalcolm@redhat.com>
The following enhances the store-with-same-value trick in
vn_reference_lookup_3 by not only looking for
a = val;
*ptr = val;
.. = a;
but also
*ptr = val;
other = x;
.. = a;
where the earlier store is more than one hop away. It does this
by queueing the actual value to compare until after the walk but
as disadvantage only allows a single such skipped store from a
constant value.
Unfortunately we cannot handle defs from non-constants this way
since we're prone to pick up values from the past loop iteration
this way and we have no good way to identify values that are
invariant in the currently iterated cycle. That's why we keep
the single-hop lookup for those cases. gcc.dg/tree-ssa/pr87126.c
would be a testcase that's un-XFAILed when we'd handle those
as well.
PR tree-optimization/106922
* tree-ssa-sccvn.cc (vn_walk_cb_data::same_val): New member.
(vn_walk_cb_data::finish): Perform delayed verification of
a skipped may-alias.
(vn_reference_lookup_pieces): Likewise.
(vn_reference_lookup): Likewise.
(vn_reference_lookup_3): When skipping stores of the same
value also handle constant stores that are more than a
single VDEF away by delaying the verification.
* gcc.dg/tree-ssa/ssa-fre-100.c: New testcase.
* g++.dg/tree-ssa/pr106922.C: Adjust.
gcc/
* config/xtensa/xtensa.cc (xtensa_can_output_mi_thunk)
(xtensa_output_mi_thunk): New functions.
(TARGET_ASM_CAN_OUTPUT_MI_THUNK)
(TARGET_ASM_OUTPUT_MI_THUNK): New macro definitions.
(xtensa_prepare_expand_call): Use fixed register a8 as temporary
when called with reload_completed set to 1.
The following resolves the issue that DSE cannot handle references
with variable offsets well when identifying possible uses of a store.
Instead of just relying on ref_maybe_used_by_stmt_p we use data-ref
analysis, making sure to perform that at most once per stmt. The
new mode is only exercised by the DSE pass before loop optimization
as specified by a new pass parameter and when expensive optimizations
are enabled, so it's disabled below -O2.
PR tree-optimization/99407
* tree-ssa-dse.cc (dse_stmt_to_dr_map): New global.
(dse_classify_store): Use data-ref analysis to disambiguate more uses.
(pass_dse::use_dr_analysis_p): New pass parameter.
(pass_dse::set_pass_param): Implement.
(pass_dse::execute): Allocate and deallocate dse_stmt_to_dr_map.
* passes.def: Allow DR analysis for the DSE pass before loop.
* gcc.dg/vect/tsvc/vect-tsvc-s243.c: Remove XFAIL.
The following was prompted by review of the patch introducing
equivalences to VN.
* tree-ssa-sccvn.cc (can_track_predicate_on_edge): New
function split out from ...
(vn_nary_op_insert_pieces_predicated): ... here.
gcc/ChangeLog:
PR target/106994
* config/i386/mmx.md (floorv2sf2): Fix typo, use
register_operand instead of vector_operand for operands[1].
gcc/testsuite/ChangeLog:
* gcc.target/i386/pr106994.c: New test.
