+/*
+ * 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"
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 {
name[0] = 0;
- while (fgets(line, sizeof(line), fasm))
+ while (my_fgets(line, sizeof(line), fasm))
{
wordc = 0;
asmln++;
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));
case DXT_BYTE:
return ".byte";
case DXT_WORD:
- return ".word";
+ return ".hword";
case DXT_DWORD:
return ".long";
case DXT_QUAD:
for (; *s != 0; s++) {
if (*s == '"') {
- strcpy(t, "\\22");
+ strcpy(t, "\\\"");
t += strlen(t);
continue;
}
}
*t++ = *s;
}
- *t = *s;
+ *t++ = *s;
+ if (t - buf > sizeof(buf))
+ aerr("string is too long\n");
return strcpy(s, buf);
}
}
static const struct parsed_proto *check_var(FILE *fhdr,
- const char *sym, const char *varname)
+ 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) {
return NULL;
}
+ if (is_export)
+ return NULL;
if (!pp->is_func && !pp->is_fptr)
return NULL;
{
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) {
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 (bad) {
+ if (!pp_compatible_func(pp_sym, pp)) {
pp_print(fp_sym, sizeof(fp_sym), pp_sym);
- anote("var: %s\n", fp_var);
- anote("sym: %s\n", fp_sym);
+ anote("entry: %s\n", fp_var);
+ anote("label: %s\n", fp_sym);
awarn("^ mismatch\n");
}
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)
{
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 in_export_table = 0;
+ int rm_labels_lines = 0;
+ 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;
char **rlist;
int rlist_cnt = 0;
int rlist_alloc;
+ int header_mode = 0;
+ int is_ro = 0;
int is_label;
int is_bss;
int wordc;
if (argc < 4) {
// -nd: no symbol decorations
- printf("usage:\n%s [-nd] <.s> <.asm> <hdrf> [rlist]*\n",
- argv[0]);
+ 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;
}
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);
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);
if (rlist_cnt > 0)
qsort(rlist, rlist_cnt, sizeof(rlist[0]), cmpstringp);
+ qsort(unwanted_syms, ARRAY_SIZE(unwanted_syms),
+ sizeof(unwanted_syms[0]), cmpstringp);
+
+ last_sym[0] = 0;
+
while (1) {
next_section(fasm, line);
if (feof(fasm))
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))
{
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 == ';') {
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) {
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;
}
- pub_syms[pub_sym_cnt++] = strdup(sym);
+
+ snprintf(last_sym, sizeof(last_sym), "%s", sym);
+ if (IS_START(sym, "__IMPORT_DESCRIPTOR_"))
+ rm_labels_lines = 5;
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%s:", no_decorations ? "" : "_", sym);
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 \"");
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, "\"");
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)
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);
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;
}
p = words[w];
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))
{
snprintf(g_comment, sizeof(g_comment), "%s", p);
}
else {
- pp = check_var(fhdr, sym, p);
+ pp = check_var(fhdr, sym, p, in_export_table);
if (pp == NULL) {
fprintf(fout, "%s%s",
(no_decorations || p[0] == '_') ? "" : "_", p);
}
}
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);
}
first = 0;
}
fin:
+ 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");
fclose(fout);
fclose(fasm);
- fclose(fhdr);
+ if (fhdr != NULL)
+ fclose(fhdr);
return 0;
}