New backend interface - assembler internals #1

This provides the first pass of assembler internals for a new, richer,
backend interface.  It also reduces the amount of data carried in
function parameters inside the assembler.

The new interface is in the form of a structure, which will contain
substantially more information that the previous interface.  This will
allow things like ELF GOT32X and Mach-O X86_64_RELOC_BRANCH
relocations to be reliably emitted.

This provides the initial set of structures.  Future additions should
include:

1. Pass down the base symbol, if any.
2. Segments should be a structure rather than a number, and the
   horrible hack of adding one for a segment reference should be
   removed (OUT_SEGMENT replaces it.)

Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>

asm/listing.c

@@ -166,7 +166,7 @@ static void list_cleanup(void)
     fclose(listfp);
 }
 
-static void list_out(int32_t offset, char *str)
+static void list_out(int64_t offset, char *str)
 {
     if (strlen(listdata) + strlen(str) > LIST_HEXBIT) {
         strcat(listdata, "-");
@@ -177,7 +177,7 @@ static void list_out(int32_t offset, char *str)
     strcat(listdata, str);
 }
 
-static void list_address(int32_t offset, const char *brackets,
+static void list_address(int64_t offset, const char *brackets,
 			 int64_t addr, int size)
 {
     char q[20];
@@ -196,21 +196,22 @@ static void list_address(int32_t offset, const char *brackets,
     list_out(offset, q);
 }
 
-static void list_output(int32_t offset, const void *data,
-			enum out_type type, uint64_t size)
+static void list_output(const struct out_data *data)
 {
     char q[20];
+    uint64_t size = data->size;
+    uint64_t offset = data->offset;
 
     if (!listp || suppress || user_nolist)
         return;
 
-    switch (type) {
+    switch (data->type) {
     case OUT_RAWDATA:
     {
-        uint8_t const *p = data;
+        const uint8_t *p = data->data;
 
 	if (size == 0 && !listdata[0])
-	    listoffset = offset;
+	    listoffset = data->offset;
         while (size--) {
             HEX(q, *p);
             q[2] = '\0';
@@ -220,19 +221,11 @@ static void list_output(int32_t offset, const void *data,
 	break;
     }
     case OUT_ADDRESS:
-      list_address(offset, "[]", *(int64_t *)data, abs((int)size));
+    case OUT_SEGMENT:
+      list_address(offset, "[]", data->toffset, size);
 	break;
-    case OUT_REL1ADR:
-	list_address(offset, "()", *(int64_t *)data, 1);
-	break;
-    case OUT_REL2ADR:
-	list_address(offset, "()", *(int64_t *)data, 2);
-	break;
-    case OUT_REL4ADR:
-	list_address(offset, "()", *(int64_t *)data, 4);
-	break;
-    case OUT_REL8ADR:
-	list_address(offset, "()", *(int64_t *)data, 8);
+    case OUT_RELADDR:
+	list_address(offset, "()", data->toffset, size);
 	break;
     case OUT_RESERVE:
     {
@@ -240,6 +233,8 @@ static void list_output(int32_t offset, const void *data,
         list_out(offset, q);
 	break;
     }
+    default:
+        panic();
     }
 }
 
@@ -324,6 +319,10 @@ static void list_error(int severity, const char *pfx, const char *msg)
 	list_emit();
 }
 
+static void list_set_offset(uint64_t offset)
+{
+    listoffset = offset;
+}
 
 static const struct lfmt nasm_list = {
     list_init,
@@ -332,7 +331,8 @@ static const struct lfmt nasm_list = {
     list_line,
     list_uplevel,
     list_downlevel,
-    list_error
+    list_error,
+    list_set_offset
 };
 
 const struct lfmt *lfmt = &nasm_list;

asm/listing.h

@@ -1,6 +1,6 @@
 /* ----------------------------------------------------------------------- *
  *   
- *   Copyright 1996-2009 The NASM Authors - All Rights Reserved
+ *   Copyright 1996-2016 The NASM Authors - All Rights Reserved
  *   See the file AUTHORS included with the NASM distribution for
  *   the specific copyright holders.
  *
@@ -61,12 +61,9 @@ struct lfmt {
      * output-format interface, only OUT_ADDRESS will _always_ be
      * displayed as if it's relocatable, so ensure that any non-
      * relocatable address has been converted to OUT_RAWDATA by
-     * then. Note that OUT_RAWDATA,0 is a valid data type, and is a
-     * dummy call used to give the listing generator an offset to
-     * work with when doing things like uplevel(LIST_TIMES) or
-     * uplevel(LIST_INCBIN).
+     * then.
      */
-    void (*output)(int32_t offset, const void *data, enum out_type type, uint64_t size);
+    void (*output)(const struct out_data *data);
 
     /*
      * Called to send a text line to the listing generator. The
@@ -100,6 +97,14 @@ struct lfmt {
      * Called on a warning or error, with the error message.
      */
     void (*error)(int severity, const char *pfx, const char *msg);
+
+    /*
+     * Update the current offset.  Used to give the listing generator
+     * an offset to work with when doing things like
+     * uplevel(LIST_TIMES) or uplevel(LIST_INCBIN); see
+     * list_set_offset();
+     */
+    void (*set_offset)(uint64_t offset);
 };
 
 extern const struct lfmt *lfmt;

asm/nasm.c

@@ -1687,6 +1687,7 @@ static void assemble_file(char *fname, StrList **depend_ptr)
 
                         int64_t l = insn_size(location.segment, offs, sb, cpu,
 					      &output_ins);
+                        l *= output_ins.times;
 
                         /* if (using_debug_info)  && output_ins.opcode != -1) */
                         if (using_debug_info)

include/nasm.h

@@ -100,12 +100,43 @@ struct ofmt;
  */
 enum out_type {
     OUT_RAWDATA,    /* Plain bytes */
-    OUT_ADDRESS,    /* An address (symbol value) */
     OUT_RESERVE,    /* Reserved bytes (RESB et al) */
-    OUT_REL1ADR,    /* 1-byte relative address */
-    OUT_REL2ADR,    /* 2-byte relative address */
-    OUT_REL4ADR,    /* 4-byte relative address */
-    OUT_REL8ADR     /* 8-byte relative address */
+    OUT_ADDRESS,    /* An address (symbol value) */
+    OUT_RELADDR,    /* A relative address (relative to instruction end) */
+    OUT_SEGMENT,    /* A segment number */
+
+    /* These are temporary until the backend change */
+    OUT_REL1ADR,
+    OUT_REL2ADR,
+    OUT_REL4ADR,
+    OUT_REL8ADR
+};
+
+enum out_sign {
+    OUT_WRAP,                   /* Undefined signedness (wraps) */
+    OUT_SIGNED,                 /* Value is signed */
+    OUT_UNSIGNED                /* Value is unsigned */
+};
+
+/*
+ * The data we send down to the backend.
+ * XXX: We still want to push down the base address symbol if
+ * available, and replace the segment numbers with a structure.
+ */
+struct out_data {
+    int64_t offset;             /* Offset within segment */
+    int32_t segment;            /* Segment written to */
+    enum out_type type;         /* See above */
+    enum out_sign sign;         /* See above */
+    int inslen;                 /* Length of instruction */
+    int insoffs;                /* Offset inside instruction */
+    int bits;                   /* Bits mode of compilation */
+    uint64_t size;              /* Size of output */
+    const struct itemplate *itemp; /* Instruction template */
+    const void *data;           /* Data for OUT_RAWDATA */
+    uint64_t toffset;           /* Target address offset for relocation */
+    int32_t tsegment;           /* Target segment for relocation */
+    int32_t twrt;               /* Relocation with respect to */
 };
 
 /*