X-Git-Url: https://notaz.gp2x.de/cgi-bin/gitweb.cgi?p=ia32rtools.git;a=blobdiff_plain;f=tools%2Fcvt_data.c;h=7c99a9172841b8010105f255368cd2032df2f8d8;hp=bf9840195c2da8fe2b130d0d4cd2632b833e7a6d;hb=75b4a70dafd7faf3c27b963fa8e67462538f91d4;hpb=b545ba7c32f3413fe3d1af1ed1a9fcc9c59b9687

diff --git a/tools/cvt_data.c b/tools/cvt_data.c
index bf98401..7c99a91 100644
--- a/tools/cvt_data.c
+++ b/tools/cvt_data.c
@@ -1,14 +1,21 @@
+/*
+ * ia32rtools
+ * (C) notaz, 2013-2015
+ *
+ * This work is licensed under the terms of 3-clause BSD license.
+ * See COPYING file in the top-level directory.
+ */
+
 #define _GNU_SOURCE
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <stdint.h>
+#include <inttypes.h>
 
 #include "my_assert.h"
 #include "my_str.h"
-
-#define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
-#define IS(w, y) !strcmp(w, y)
-#define IS_START(w, y) !strncmp(w, y, strlen(y))
+#include "common.h"
 
 #include "protoparse.h"
 
@@ -18,6 +25,8 @@ static int asmln;
 static const struct parsed_proto *g_func_sym_pp;
 static char g_comment[256];
 static int g_warn_cnt;
+static int g_cconv_novalidate;
+static int g_arm_mode;
 
 // note: must be in ascending order
 enum dx_type {
@@ -77,7 +86,7 @@ static void next_section(FILE *fasm, char *name)
 
   name[0] = 0;
 
-  while (fgets(line, sizeof(line), fasm))
+  while (my_fgets(line, sizeof(line), fasm))
   {
     wordc = 0;
     asmln++;
@@ -86,14 +95,8 @@ static void next_section(FILE *fasm, char *name)
     if (*p == 0)
       continue;
 
-    if (*p == ';') {
-      while (strlen(line) == sizeof(line) - 1) {
-        // one of those long comment lines..
-        if (!fgets(line, sizeof(line), fasm))
-          break;
-      }
+    if (*p == ';')
       continue;
-    }
 
     for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
       p = sskip(next_word(words[wordc], sizeof(words[0]), p));
@@ -136,7 +139,7 @@ static const char *type_name(enum dx_type type)
   case DXT_BYTE:
     return ".byte";
   case DXT_WORD:
-    return ".word";
+    return ".hword";
   case DXT_DWORD:
     return ".long";
   case DXT_QUAD:
@@ -190,7 +193,7 @@ static char *escape_string(char *s)
 
   for (; *s != 0; s++) {
     if (*s == '"') {
-      strcpy(t, "\\22");
+      strcpy(t, "\\\"");
       t += strlen(t);
       continue;
     }
@@ -201,32 +204,12 @@ static char *escape_string(char *s)
     }
     *t++ = *s;
   }
-  *t = *s;
+  *t++ = *s;
+  if (t - buf > sizeof(buf))
+    aerr("string is too long\n");
   return strcpy(s, buf);
 }
 
-static void sprint_pp(const struct parsed_proto *pp, char *buf,
-  size_t buf_size)
-{
-  size_t l;
-  int i;
-
-  snprintf(buf, buf_size, "%s %s(", pp->ret_type.name, pp->name);
-  l = strlen(buf);
-
-  for (i = 0; i < pp->argc_reg; i++) {
-    snprintf(buf + l, buf_size - l, "%s%s",
-      i == 0 ? "" : ", ", pp->arg[i].reg);
-    l = strlen(buf);
-  }
-  if (pp->argc_stack > 0) {
-    snprintf(buf + l, buf_size - l, "%s{%d stack}",
-      i == 0 ? "" : ", ", pp->argc_stack);
-    l = strlen(buf);
-  }
-  snprintf(buf + l, buf_size - l, ")");
-}
-
 static void sprint_pp_short(const struct parsed_proto *pp, char *buf,
   size_t buf_size)
 {
@@ -255,23 +238,26 @@ static void sprint_pp_short(const struct parsed_proto *pp, char *buf,
   snprintf(buf + l, buf_size - l, ")");
 }
 
-static void check_var(FILE *fhdr, const char *sym, const char *varname)
+static const struct parsed_proto *check_var(FILE *fhdr,
+  const char *sym, const char *varname, int is_export)
 {
   const struct parsed_proto *pp, *pp_sym;
-  char fp_sym[256], fp_var[256];
-  int i, bad = 0;
+  char fp_sym[256], fp_var[256], *p;
+  int i;
 
   pp = proto_parse(fhdr, varname, 1);
   if (pp == NULL) {
     if (IS_START(varname, "sub_"))
       awarn("sub_ sym missing proto: '%s'\n", varname);
-    return;
+    return NULL;
   }
 
+  if (is_export)
+    return NULL;
   if (!pp->is_func && !pp->is_fptr)
-    return;
+    return NULL;
 
-  sprint_pp(pp, fp_var, sizeof(fp_var));
+  pp_print(fp_var, sizeof(fp_var), pp);
 
   if (pp->argc_reg == 0)
     goto check_sym;
@@ -280,21 +266,36 @@ static void check_var(FILE *fhdr, const char *sym, const char *varname)
   {
     goto check_sym;
   }
-  if (pp->argc_reg != 2
-    || !IS(pp->arg[0].reg, "ecx")
-    || !IS(pp->arg[1].reg, "edx"))
+  if (!g_cconv_novalidate
+    && (pp->argc_reg != 2
+       || !IS(pp->arg[0].reg, "ecx")
+       || !IS(pp->arg[1].reg, "edx")))
   {
     awarn("unhandled reg call: %s\n", fp_var);
   }
 
 check_sym:
+  // fptrs must use 32bit args, callsite might have no information and
+  // lack a cast to smaller types, which results in incorrectly masked
+  // args passed (callee may assume masked args, it does on ARM)
+  for (i = 0; i < pp->argc; i++) {
+    if (pp->arg[i].type.is_ptr)
+      continue;
+    p = pp->arg[i].type.name;
+    if (strstr(p, "int8") || strstr(p, "int16")
+      || strstr(p, "char") || strstr(p, "short"))
+    {
+      awarn("reference to %s with arg%d '%s'\n", pp->name, i + 1, p);
+    }
+  }
+
   sprint_pp_short(pp, g_comment, sizeof(g_comment));
 
   if (sym != NULL) {
     g_func_sym_pp = NULL;
     pp_sym = proto_parse(fhdr, sym, 1);
     if (pp_sym == NULL)
-      return;
+      return pp;
     if (!pp_sym->is_fptr)
       aerr("func ptr data, but label '%s' !is_fptr\n", pp_sym->name);
     g_func_sym_pp = pp_sym;
@@ -302,32 +303,39 @@ check_sym:
   else {
     pp_sym = g_func_sym_pp;
     if (pp_sym == NULL)
-      return;
+      return pp;
   }
 
-  if (pp->argc != pp_sym->argc || pp->argc_reg != pp_sym->argc_reg)
-    bad = 1;
-  else {
-    for (i = 0; i < pp->argc; i++) {
-      if ((pp->arg[i].reg != NULL) != (pp_sym->arg[i].reg != NULL)) {
-        bad = 1;
-        break;
-      }
-      if ((pp->arg[i].reg != NULL)
-        && !IS(pp->arg[i].reg, pp_sym->arg[i].reg))
-      {
-        bad = 1;
-        break;
-      }
-    }
+  if (!pp_compatible_func(pp_sym, pp)) {
+    pp_print(fp_sym, sizeof(fp_sym), pp_sym);
+    anote("entry: %s\n", fp_var);
+    anote("label: %s\n", fp_sym);
+    awarn("^ mismatch\n");
   }
 
-  if (bad) {
-    sprint_pp(pp_sym, fp_sym, sizeof(fp_sym));
-    anote("var: %s\n", fp_var);
-    anote("sym: %s\n", fp_sym);
-    awarn("^ mismatch\n");
+  return pp;
+}
+
+static void output_decorated_pp(FILE *fout,
+  const struct parsed_proto *pp)
+{
+  if (pp->name[0] != '_')
+    fprintf(fout, pp->is_fastcall ? "@" : "_");
+  fprintf(fout, "%s", pp->name);
+  if (pp->is_stdcall && pp->argc > 0)
+    fprintf(fout, "@%d", pp->argc * 4);
+}
+
+static int align_value(int src_val)
+{
+  if (src_val <= 0) {
+    awarn("bad align: %d\n", src_val);
+    src_val = 1;
   }
+  if (!g_arm_mode)
+    return src_val;
+
+  return __builtin_ffs(src_val) - 1;
 }
 
 static int cmpstringp(const void *p1, const void *p2)
@@ -335,15 +343,57 @@ static int cmpstringp(const void *p1, const void *p2)
   return strcmp(*(char * const *)p1, *(char * const *)p2);
 }
 
+/* XXX: maybe move to external file? */
+static const char *unwanted_syms[] = {
+  "aRuntimeError",
+  "aTlossError",
+  "aSingError",
+  "aDomainError",
+  "aR6029ThisAppli",
+  "aR6028UnableToI",
+  "aR6027NotEnough",
+  "aR6026NotEnough",
+  "aR6025PureVirtu",
+  "aR6024NotEnough",
+  "aR6019UnableToO",
+  "aR6018Unexpecte",
+  "aR6017Unexpecte",
+  "aR6016NotEnough",
+  "aAbnormalProgra",
+  "aR6009NotEnough",
+  "aR6008NotEnough",
+  "aR6002FloatingP",
+  "aMicrosoftVisua",
+  "aRuntimeErrorPr",
+  "aThisApplicatio",
+  "aMicrosoftFindF",
+  "aMicrosoftOffic",
+};
+
+static int is_unwanted_sym(const char *sym)
+{
+  return bsearch(&sym, unwanted_syms, ARRAY_SIZE(unwanted_syms),
+    sizeof(unwanted_syms[0]), cmpstringp) != NULL;
+}
+
 int main(int argc, char *argv[])
 {
-  FILE *fout, *fasm, *fhdr, *frlist;
+  FILE *fout, *fasm, *fhdr = NULL, *frlist;
   const struct parsed_proto *pp;
+  int no_decorations = 0;
+  int header_mode = 0;
+  int maybe_func_table;
+  int in_export_table;
+  int rm_labels_lines;
+  int is_zero_val;
+  char comment_char = '#';
   char words[20][256];
   char word[256];
   char line[256];
+  char last_sym[32];
   unsigned long val;
   unsigned long cnt;
+  uint64_t val64;
   const char *sym;
   enum dx_type type;
   char **pub_syms;
@@ -352,6 +402,7 @@ int main(int argc, char *argv[])
   char **rlist;
   int rlist_cnt = 0;
   int rlist_alloc;
+  int is_ro = 0;
   int is_label;
   int is_bss;
   int wordc;
@@ -364,20 +415,39 @@ int main(int argc, char *argv[])
   char *p2;
 
   if (argc < 4) {
-    printf("usage:\n%s <.s> <.asm> <hdrf> [rlist]*\n",
-      argv[0]);
+    // -nd: no symbol decorations
+    printf("usage:\n%s [-nd] [-i] [-a] <.s> <.asm> <hdrf> [rlist]*\n"
+           "%s -hdr <.h> <.asm>\n",
+      argv[0], argv[0]);
     return 1;
   }
 
+  for (arg = 1; arg < argc; arg++) {
+    if (IS(argv[arg], "-nd"))
+      no_decorations = 1;
+    else if (IS(argv[arg], "-i"))
+      g_cconv_novalidate = 1;
+    else if (IS(argv[arg], "-a")) {
+      comment_char = '@';
+      g_arm_mode = 1;
+    }
+    else if (IS(argv[arg], "-hdr"))
+      header_mode = 1;
+    else
+      break;
+  }
+
   arg_out = arg++;
 
   asmfn = argv[arg++];
   fasm = fopen(asmfn, "r");
   my_assert_not(fasm, NULL);
 
-  hdrfn = argv[arg++];
-  fhdr = fopen(hdrfn, "r");
-  my_assert_not(fhdr, NULL);
+  if (!header_mode) {
+    hdrfn = argv[arg++];
+    fhdr = fopen(hdrfn, "r");
+    my_assert_not(fhdr, NULL);
+  }
 
   fout = fopen(argv[arg_out], "w");
   my_assert_not(fout, NULL);
@@ -394,9 +464,9 @@ int main(int argc, char *argv[])
     frlist = fopen(argv[arg], "r");
     my_assert_not(frlist, NULL);
 
-    while (fgets(line, sizeof(line), frlist)) {
+    while (my_fgets(line, sizeof(line), frlist)) {
       p = sskip(line);
-      if (*p == 0 || *p == ';')
+      if (*p == 0 || *p == ';' || *p == '#')
         continue;
 
       p = next_word(words[0], sizeof(words[0]), p);
@@ -418,31 +488,62 @@ int main(int argc, char *argv[])
   if (rlist_cnt > 0)
     qsort(rlist, rlist_cnt, sizeof(rlist[0]), cmpstringp);
 
+  qsort(unwanted_syms, ARRAY_SIZE(unwanted_syms),
+    sizeof(unwanted_syms[0]), cmpstringp);
+
   while (1) {
+    last_sym[0] = 0;
+    g_func_sym_pp = NULL;
+    maybe_func_table = 0;
+    in_export_table = 0;
+    rm_labels_lines = 0;
+
     next_section(fasm, line);
     if (feof(fasm))
       break;
     if (IS(line + 1, "text"))
       continue;
 
-    if (IS(line + 1, "rdata"))
-      fprintf(fout, "\n.section .rodata\n");
-    else if (IS(line + 1, "data"))
-      fprintf(fout, "\n.data\n");
+    if (IS(line + 1, "rdata")) {
+      is_ro = 1;
+      if (!header_mode)
+        fprintf(fout, "\n.section .rodata\n");
+    }
+    else if (IS(line + 1, "data")) {
+      is_ro = 0;
+      if (!header_mode)
+        fprintf(fout, "\n.data\n");
+    }
     else
       aerr("unhandled section: '%s'\n", line);
 
-    fprintf(fout, ".align 4\n");
+    if (!header_mode)
+      fprintf(fout, ".align %d\n", align_value(4));
 
-    while (fgets(line, sizeof(line), fasm))
+    while (my_fgets(line, sizeof(line), fasm))
     {
+      is_zero_val = 0;
       sym = NULL;
       asmln++;
 
       p = sskip(line);
-      if (*p == 0 || *p == ';')
+      if (*p == 0)
         continue;
 
+      if (*p == ';') {
+        if (IS_START(p, ";org") && sscanf(p + 5, "%Xh", &i) == 1) {
+          // ;org is only seen at section start, so assume . addr 0
+          i &= 0xfff;
+          if (i != 0 && !header_mode)
+            fprintf(fout, "\t\t  .skip 0x%x\n", i);
+        }
+        else if (IS_START(p, "; Export Address"))
+          in_export_table = 1;
+        else if (IS_START(p, "; Export"))
+          in_export_table = 0;
+        continue;
+      }
+
       for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
         p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
         if (*p == 0 || *p == ';') {
@@ -456,8 +557,10 @@ int main(int argc, char *argv[])
 
       if (*p == ';') {
         p = sskip(p + 1);
-        if (IS_START(p, "sctclrtype"))
+        if (IS_START(p, "sctclrtype")) {
+          maybe_func_table = 0;
           g_func_sym_pp = NULL;
+        }
       }
 
       if (wordc == 2 && IS(words[1], "ends"))
@@ -472,11 +575,19 @@ int main(int argc, char *argv[])
         continue;
 
       if (IS(words[0], "align")) {
-        val = parse_number(words[1]);
-        fprintf(fout, "\t\t  .align %ld", val);
+        if (header_mode)
+          continue;
+
+        val = parse_number(words[1], 0);
+        fprintf(fout, "\t\t  .align %d", align_value(val));
         goto fin;
       }
 
+      if (IS(words[0], "public")) {
+        // skip, sym should appear in header anyway
+        continue;
+      }
+
       w = 1;
       type = parse_dx_directive(words[0]);
       if (type == DXT_UNSPEC) {
@@ -487,21 +598,65 @@ int main(int argc, char *argv[])
       if (type == DXT_UNSPEC)
         aerr("unhandled decl: '%s %s'\n", words[0], words[1]);
 
-      if (sym != NULL) {
-        // public/global name
-        if (pub_sym_cnt >= pub_sym_alloc) {
-          pub_sym_alloc *= 2;
-          pub_syms = realloc(pub_syms, pub_sym_alloc * sizeof(pub_syms[0]));
-          my_assert_not(pub_syms, NULL);
+      if (sym != NULL)
+      {
+        if (header_mode) {
+          int is_str = 0;
+
+          fprintf(fout, "extern ");
+          if (is_ro)
+            fprintf(fout, "const ");
+
+          switch (type) {
+          case DXT_BYTE:
+            for (i = w; i < wordc; i++)
+              if (words[i][0] == '\'')
+                is_str = 1;
+            if (is_str)
+              fprintf(fout, "char     %s[];\n", sym);
+            else
+              fprintf(fout, "uint8_t  %s;\n", sym);
+            break;
+
+          case DXT_WORD:
+            fprintf(fout, "uint16_t %s;\n", sym);
+            break;
+
+          case DXT_DWORD:
+            fprintf(fout, "uint32_t %s;\n", sym);
+            break;
+
+          default:
+            fprintf(fout, "_UNKNOWN %s;\n", sym);
+            break;
+          }
+
+          continue;
+        }
+
+        snprintf(last_sym, sizeof(last_sym), "%s", sym);
+        maybe_func_table = type == DXT_DWORD;
+
+        if (IS_START(sym, "__IMPORT_DESCRIPTOR_")) {
+          rm_labels_lines = 5;
+          maybe_func_table = 0;
         }
-        pub_syms[pub_sym_cnt++] = strdup(sym);
 
         pp = proto_parse(fhdr, sym, 1);
-        if (pp != NULL)
+        if (pp != NULL) {
           g_func_sym_pp = NULL;
 
+          // public/global name
+          if (pub_sym_cnt >= pub_sym_alloc) {
+            pub_sym_alloc *= 2;
+            pub_syms = realloc(pub_syms, pub_sym_alloc * sizeof(pub_syms[0]));
+            my_assert_not(pub_syms, NULL);
+          }
+          pub_syms[pub_sym_cnt++] = strdup(sym);
+        }
+
         len = strlen(sym);
-        fprintf(fout, "_%s:", sym);
+        fprintf(fout, "%s%s:", no_decorations ? "" : "_", sym);
 
         len += 2;
         if (len < 8)
@@ -514,10 +669,35 @@ int main(int argc, char *argv[])
           fprintf(fout, " ");
       }
       else {
+        if (header_mode)
+          continue;
+
         fprintf(fout, "\t\t  ");
       }
 
-      if (type == DXT_BYTE && words[w][0] == '\'') {
+      // fill out some unwanted strings with zeroes..
+      if (type == DXT_BYTE && words[w][0] == '\''
+        && is_unwanted_sym(last_sym))
+      {
+        len = 0;
+        for (; w < wordc; w++) {
+          if (words[w][0] == '\'') {
+            p = words[w] + 1;
+            for (; *p && *p != '\''; p++)
+              len++;
+          }
+          else {
+            // assume encoded byte
+            len++;
+          }
+        }
+        fprintf(fout, ".skip %d", len);
+        goto fin;
+      }
+      else if (type == DXT_BYTE
+        && (words[w][0] == '\''
+            || (w + 1 < wordc && words[w + 1][0] == '\'')))
+      {
         // string; use asciz for most common case
         if (w == wordc - 2 && IS(words[w + 1], "0")) {
           fprintf(fout, ".asciz \"");
@@ -537,10 +717,12 @@ int main(int argc, char *argv[])
             fprintf(fout, "%s", escape_string(word));
           }
           else {
-            val = parse_number(words[w]);
+            val = parse_number(words[w], 0);
             if (val & ~0xff)
               aerr("bad string trailing byte?\n");
-            fprintf(fout, "\\x%02lx", val);
+            // unfortunately \xHH is unusable - gas interprets
+            // things like \x27b as 0x7b, so have to use octal here
+            fprintf(fout, "\\%03lo", val);
           }
         }
         fprintf(fout, "\"");
@@ -549,7 +731,7 @@ int main(int argc, char *argv[])
 
       if (w == wordc - 2) {
         if (IS_START(words[w + 1], "dup(")) {
-          cnt = parse_number(words[w]);
+          cnt = parse_number(words[w], 0);
           p = words[w + 1] + 4;
           p2 = strchr(p, ')');
           if (p2 == NULL)
@@ -559,7 +741,7 @@ int main(int argc, char *argv[])
 
           val = 0;
           if (!IS(word, "?"))
-            val = parse_number(word);
+            val = parse_number(word, 0);
 
           fprintf(fout, ".fill 0x%02lx,%d,0x%02lx",
             cnt, type_size(type), val);
@@ -585,7 +767,13 @@ int main(int argc, char *argv[])
         if (w != wordc - 1)
           aerr("TODO\n");
 
-        fprintf(fout, "%s %s", type_name_float(type), words[w]);
+        if (g_arm_mode && type == DXT_TEN) {
+          fprintf(fout, ".fill 10");
+          snprintf(g_comment, sizeof(g_comment), "%s %s",
+            type_name_float(type), words[w]);
+        }
+        else
+          fprintf(fout, "%s %s", type_name_float(type), words[w]);
         goto fin;
       }
 
@@ -616,7 +804,9 @@ int main(int argc, char *argv[])
         }
         else if (is_label) {
           p = words[w];
-          if (IS_START(p, "loc_") || strchr(p, '?') || strchr(p, '@')
+          if (IS_START(p, "loc_") || IS_START(p, "__imp")
+             || strchr(p, '?') || strchr(p, '@')
+             || rm_labels_lines > 0
              || bsearch(&p, rlist, rlist_cnt, sizeof(rlist[0]),
                   cmpstringp))
           {
@@ -624,24 +814,43 @@ int main(int argc, char *argv[])
             snprintf(g_comment, sizeof(g_comment), "%s", p);
           }
           else {
-            check_var(fhdr, sym, p);
-            fprintf(fout, "_%s", p);
+            const char *f_sym = maybe_func_table ? last_sym : NULL;
+
+            pp = check_var(fhdr, f_sym, p, in_export_table);
+            if (pp == NULL) {
+              fprintf(fout, "%s%s",
+                (no_decorations || p[0] == '_') ? "" : "_", p);
+            }
+            else {
+              if (no_decorations)
+                fprintf(fout, "%s", pp->name);
+              else
+                output_decorated_pp(fout, pp);
+            }
           }
         }
         else {
-          val = parse_number(words[w]);
-          if (val < 10)
-            fprintf(fout, "%ld", val);
+          val64 = parse_number(words[w], 1);
+          if (val64 < 10)
+            fprintf(fout, "%d", (int)val64);
           else
-            fprintf(fout, "0x%lx", val);
+            fprintf(fout, "0x%" PRIx64, val64);
+
+          is_zero_val = val64 == 0;
         }
 
         first = 0;
       }
 
 fin:
+      if (!is_zero_val)
+        maybe_func_table = 0;
+
+      if (rm_labels_lines > 0)
+        rm_labels_lines--;
+
       if (g_comment[0] != 0) {
-        fprintf(fout, "\t\t# %s", g_comment);
+        fprintf(fout, "\t\t%c %s", comment_char, g_comment);
         g_comment[0] = 0;
       }
       fprintf(fout, "\n");
@@ -652,11 +861,13 @@ fin:
 
   // dump public syms
   for (i = 0; i < pub_sym_cnt; i++)
-    fprintf(fout, ".global _%s\n", pub_syms[i]);
+    fprintf(fout, ".global %s%s\n",
+      no_decorations ? "" : "_", pub_syms[i]);
 
   fclose(fout);
   fclose(fasm);
-  fclose(fhdr);
+  if (fhdr != NULL)
+    fclose(fhdr);
 
   return 0;
 }