5 * This work is licensed under the terms of 3-clause BSD license.
6 * See COPYING file in the top-level directory.
13 unsigned int is_array:1;
14 unsigned int is_ptr:1;
15 unsigned int is_struct:1; // split for args
16 unsigned int is_retreg:1; // register to return to caller
17 unsigned int is_va_list:1;
18 unsigned int is_64bit:1;
19 unsigned int is_float:1; // float, double
22 struct parsed_proto_arg {
24 struct parsed_type type;
25 struct parsed_proto *pp; // fptr or struct
26 unsigned int is_saved:1; // not set here, for tool use
34 struct parsed_type ret_type;
35 struct parsed_type type;
37 struct parsed_proto_arg arg[32];
41 unsigned int is_func:1;
42 unsigned int is_stdcall:1;
43 unsigned int is_fastcall:1;
44 unsigned int is_vararg:1; // vararg func
45 unsigned int is_fptr:1;
46 unsigned int is_import:1; // data import
47 unsigned int is_noreturn:1;
48 unsigned int is_unresolved:1;
49 unsigned int is_guessed:1; // for extra checking
50 unsigned int is_userstack:1;
51 unsigned int is_include:1; // not from top-level header
52 unsigned int is_osinc:1; // OS/system library func
53 unsigned int is_cinc:1; // crt library func
54 unsigned int is_arg:1; // declared in some func arg
55 unsigned int has_structarg:1;
56 unsigned int has_retreg:1;
59 struct parsed_struct {
63 struct parsed_proto pp;
68 static const char *hdrfn;
69 static int hdrfline = 0;
71 static void pp_copy_arg(struct parsed_proto_arg *d,
72 const struct parsed_proto_arg *s);
74 static int b_pp_c_handler(char *proto, const char *fname,
75 int is_include, int is_osinc, int is_cinc);
76 static int struct_handler(FILE *fhdr, char *proto, int *line);
78 static int do_protostrs(FILE *fhdr, const char *fname, int is_include)
80 const char *finc_name;
81 const char *hdrfn_saved;
95 is_cinc = strstr(fname, "stdc.hlist") != NULL;
96 is_osinc = is_cinc || strstr(fname, "win32.hlist") != NULL;
98 while (fgets(protostr, sizeof(protostr), fhdr))
101 if (strncmp(protostr, "//#include ", 11) == 0) {
102 finc_name = protostr + 11;
103 p = strpbrk(finc_name, "\r\n ");
108 p = strrchr(hdrfn_saved, '/');
110 memcpy(path, hdrfn_saved,
111 p - hdrfn_saved + 1);
112 path[p - hdrfn_saved + 1] = 0;
114 snprintf(fname_inc, sizeof(fname_inc), "%s%s",
116 finc = fopen(fname_inc, "r");
118 printf("%s:%d: can't open '%s'\n",
119 fname_inc, line, finc_name);
122 ret = do_protostrs(finc, finc_name, 1);
128 if (strncmp(sskip(protostr), "//", 2) == 0)
131 p = protostr + strlen(protostr);
132 for (p--; p >= protostr && my_isblank(*p); --p)
139 if (!strncmp(protostr, "struct", 6)
140 && strchr(protostr, '{') != NULL)
141 ret = struct_handler(fhdr, protostr, &line);
143 ret = b_pp_c_handler(protostr, hdrfn,
144 is_include, is_osinc, is_cinc);
157 static int get_regparm(char *dst, size_t dlen, char *p, int *retreg)
172 for (o = 0; o < dlen; i++) {
184 static const char *known_type_mod[] = {
193 static const char *known_ptr_types[] = {
210 "HIMC", // DWORD in mingw, ptr in wine..
234 "PMEMORY_BASIC_INFORMATION",
245 "SERVICE_STATUS_HANDLE",
254 static const char *ignored_keywords[] = {
262 static int typecmp(const char *n, const char *t)
264 for (; *t != 0; n++, t++) {
265 while (n[0] == ' ' && (n[1] == ' ' || n[1] == '*'))
267 while (t[0] == ' ' && (t[1] == ' ' || t[1] == '*'))
276 static const char *skip_type_mod(const char *n)
281 for (i = 0; i < ARRAY_SIZE(known_type_mod); i++) {
282 len = strlen(known_type_mod[i]);
283 if (strncmp(n, known_type_mod[i], len) != 0)
285 if (!my_isblank(n[len]))
289 while (my_isblank(*n))
297 static int check_type(const char *name, struct parsed_type *type)
303 n = skip_type_mod(name);
305 if (!strncmp(n, "struct", 6) && my_isblank(n[6])) {
309 while (my_isblank(*n))
313 for (i = 0; i < ARRAY_SIZE(known_ptr_types); i++) {
314 if (typecmp(n, known_ptr_types[i]))
321 if (n[0] == 'L' && n[1] == 'P' && strncmp(n, "LPARAM", 6))
324 // assume single word
325 while (!my_isblank(*n) && !my_issep(*n))
330 while (my_isblank(*n))
341 type->name = strndup(name, ret);
342 if (IS(type->name, "__VALIST") || IS(type->name, "va_list"))
343 type->is_va_list = 1;
344 if (IS(type->name, "VOID"))
345 memcpy(type->name, "void", 4);
350 /* args are always expanded to 32bit */
351 static const char *map_reg(const char *reg)
353 const char *regs_f[] = { "eax", "ebx", "ecx", "edx", "esi", "edi" };
354 const char *regs_w[] = { "ax", "bx", "cx", "dx", "si", "di" };
355 const char *regs_b[] = { "al", "bl", "cl", "dl" };
358 for (i = 0; i < ARRAY_SIZE(regs_w); i++)
359 if (IS(reg, regs_w[i]))
362 for (i = 0; i < ARRAY_SIZE(regs_b); i++)
363 if (IS(reg, regs_b[i]))
369 static int check_struct_arg(struct parsed_proto_arg *arg)
371 if (IS(arg->type.name, "POINT"))
377 static int parse_protostr(char *protostr, struct parsed_proto *pp);
379 static int parse_arg(char **p_, struct parsed_proto_arg *arg, int xarg)
386 arg->pp = calloc(1, sizeof(*arg->pp));
387 my_assert_not(arg->pp, NULL);
392 pe = strpbrk(pe, ",()");
395 if (*pe == ',' || *pe == ')')
397 pe = strchr(pe, ')');
403 if (pe - p > sizeof(buf) - 1)
405 memcpy(buf, p, pe - p);
408 ret = parse_protostr(buf, arg->pp);
412 if (IS_START(arg->pp->name, "guess"))
413 arg->pp->is_guessed = 1;
415 // we don't use actual names right now...
416 snprintf(arg->pp->name, sizeof(arg->pp->name), "a%d", xarg);
418 if (!arg->type.is_struct)
419 // we'll treat it as void * for non-calls
420 arg->type.name = strdup("void *");
421 arg->type.is_ptr = 1;
428 static int parse_protostr(char *protostr, struct parsed_proto *pp)
430 struct parsed_proto_arg *arg;
441 if (p[0] == '/' && p[1] == '/') {
442 printf("%s:%d: commented out?\n", hdrfn, hdrfline);
446 // allow start of line comment
447 if (p[0] == '/' && p[1] == '*') {
448 p = strstr(p + 2, "*/");
450 printf("%s:%d: multiline comments unsupported\n",
457 // we need remaining hints in comments, so strip / *
458 for (p1 = p; p1[0] != 0 && p1[1] != 0; p1++) {
459 if ((p1[0] == '/' && p1[1] == '*')
460 || (p1[0] == '*' && p1[1] == '/'))
464 if (!strncmp(p, "DECLSPEC_NORETURN ", 18)) {
469 for (i = 0; i < ARRAY_SIZE(ignored_keywords); i++) {
470 l = strlen(ignored_keywords[i]);
471 if (!strncmp(p, ignored_keywords[i], l) && my_isblank(p[l]))
472 p = sskip(p + l + 1);
475 if (IS_START(p, "DECL_IMPORT ")) {
480 ret = check_type(p, &pp->ret_type);
482 printf("%s:%d:%zd: unhandled return in '%s'\n",
483 hdrfn, hdrfline, (p - protostr) + 1, protostr);
488 if (!strncmp(p, "noreturn ", 9)) {
493 if (!strchr(p, ')')) {
494 p = next_idt(buf, sizeof(buf), p);
496 if (!pp->is_arg && buf[0] == 0) {
497 printf("%s:%d:%zd: var name is missing\n",
498 hdrfn, hdrfline, (p - protostr) + 1);
501 strcpy(pp->name, buf);
506 pp->ret_type.is_array = 1;
518 p = next_word(cconv, sizeof(cconv), p);
521 printf("%s:%d:%zd: cconv missing\n",
522 hdrfn, hdrfline, (p - protostr) + 1);
525 if (IS(cconv, "__cdecl"))
527 else if (IS(cconv, "__stdcall"))
529 else if (IS(cconv, "__fastcall")) {
531 pp->is_stdcall = 1; // sort of..
533 else if (IS(cconv, "__thiscall"))
535 else if (IS(cconv, "__userpurge"))
536 pp->is_stdcall = 1; // IDA
537 else if (IS(cconv, "__usercall"))
538 pp->is_stdcall = 0; // IDA
539 else if (IS(cconv, "__userstack")) {
540 pp->is_stdcall = 0; // custom
541 pp->is_userstack = 1;
543 else if (IS(cconv, "WINAPI") || IS(cconv, "PASCAL"))
546 printf("%s:%d:%zd: unhandled cconv: '%s'\n",
547 hdrfn, hdrfline, (p - protostr) + 1, cconv);
553 printf("%s:%d:%zd: '*' expected\n",
554 hdrfn, hdrfline, (p - protostr) + 1);
558 // XXX: skipping extra asterisks, for now
564 p = next_idt(buf, sizeof(buf), p);
567 //printf("%s:%d:%zd: func name missing\n",
568 // hdrfn, hdrfline, (p - protostr) + 1);
571 strcpy(pp->name, buf);
573 ret = get_regparm(regparm, sizeof(regparm), p, &is_retreg);
575 if (!IS(regparm, "eax") && !IS(regparm, "ax")
576 && !IS(regparm, "al") && !IS(regparm, "edx:eax"))
578 printf("%s:%d:%zd: bad regparm: %s\n",
579 hdrfn, hdrfline, (p - protostr) + 1, regparm);
588 // not really ret_type is array, but ohwell
589 pp->ret_type.is_array = 1;
590 p = strchr(p + 1, ']');
592 printf("%s:%d:%zd: ']' expected\n",
593 hdrfn, hdrfline, (p - protostr) + 1);
599 printf("%s:%d:%zd: ')' expected\n",
600 hdrfn, hdrfline, (p - protostr) + 1);
607 printf("%s:%d:%zd: '(' expected, got '%c'\n",
608 hdrfn, hdrfline, (p - protostr) + 1, *p);
615 if ((!strncmp(p, "void", 4) || !strncmp(p, "VOID", 4))
616 && *sskip(p + 4) == ')')
627 printf("%s:%d:%zd: ',' expected\n",
628 hdrfn, hdrfline, (p - protostr) + 1);
634 if (!strncmp(p, "...", 3)) {
641 printf("%s:%d:%zd: ')' expected\n",
642 hdrfn, hdrfline, (p - protostr) + 1);
646 if (xarg >= ARRAY_SIZE(pp->arg)) {
647 printf("%s:%d:%zd: too many args\n",
648 hdrfn, hdrfline, (p - protostr) + 1);
652 arg = &pp->arg[xarg];
656 ret = check_type(p, &arg->type);
658 printf("%s:%d:%zd: unhandled type for arg%d\n",
659 hdrfn, hdrfline, (p - protostr) + 1, xarg);
664 if (*p == '(' || arg->type.is_struct) {
665 // func ptr or struct
666 ret = parse_arg(&p1, arg, xarg);
668 printf("%s:%d:%zd: funcarg parse failed\n",
669 hdrfn, hdrfline, p1 - protostr);
675 p = next_idt(buf, sizeof(buf), p);
679 printf("%s:%d:%zd: idt missing for arg%d\n",
680 hdrfn, hdrfline, (p - protostr) + 1, xarg);
686 ret = get_regparm(regparm, sizeof(regparm), p, &is_retreg);
691 arg->reg = strdup(map_reg(regparm));
692 arg->type.is_retreg = is_retreg;
693 pp->has_retreg |= is_retreg;
696 if (IS(arg->type.name, "float")
697 || IS(arg->type.name, "double"))
699 arg->type.is_float = 1;
702 if (!arg->type.is_ptr && (strstr(arg->type.name, "int64")
703 || IS(arg->type.name, "double")))
705 arg->type.is_64bit = 1;
707 pp_copy_arg(&pp->arg[xarg], arg);
708 arg = &pp->arg[xarg];
710 free(arg->type.name);
711 arg->type.name = strdup("dummy");
714 ret = check_struct_arg(arg);
716 pp->has_structarg = 1;
717 arg->type.is_struct = 1;
718 free(arg->type.name);
719 arg->type.name = strdup("int");
720 for (l = 0; l < ret; l++) {
721 pp_copy_arg(&pp->arg[xarg], arg);
727 if (xarg > 0 && (IS(cconv, "__fastcall") || IS(cconv, "__thiscall"))) {
728 if (pp->arg[0].reg != NULL) {
729 printf("%s:%d: %s with arg1 spec %s?\n",
730 hdrfn, hdrfline, cconv, pp->arg[0].reg);
732 pp->arg[0].reg = strdup("ecx");
735 if (xarg > 1 && IS(cconv, "__fastcall")) {
736 if (pp->arg[1].reg != NULL) {
737 printf("%s:%d: %s with arg2 spec %s?\n",
738 hdrfn, hdrfline, cconv, pp->arg[1].reg);
740 pp->arg[1].reg = strdup("edx");
745 for (i = 0; i < pp->argc; i++) {
746 if (pp->arg[i].reg == NULL)
752 if (pp->argc == 1 && pp->arg[0].reg != NULL
753 && IS(pp->arg[0].reg, "ecx"))
757 else if (pp->argc_reg == 2
758 && pp->arg[0].reg != NULL && IS(pp->arg[0].reg, "ecx")
759 && pp->arg[1].reg != NULL && IS(pp->arg[1].reg, "edx"))
764 if (pp->is_vararg && (pp->is_stdcall || pp->is_fastcall)) {
765 printf("%s:%d: vararg %s?\n", hdrfn, hdrfline, cconv);
772 static int pp_name_cmp(const void *p1, const void *p2)
774 const struct parsed_proto *pp1 = p1, *pp2 = p2;
775 return strcmp(pp1->name, pp2->name);
778 static int ps_name_cmp(const void *p1, const void *p2)
780 const struct parsed_struct *ps1 = p1, *ps2 = p2;
781 return strcmp(ps1->name, ps2->name);
784 // parsed struct cache
785 static struct parsed_struct *ps_cache;
786 static int ps_cache_size;
787 static int ps_cache_alloc;
789 static int struct_handler(FILE *fhdr, char *proto, int *line)
791 struct parsed_struct *ps;
797 if (ps_cache_size >= ps_cache_alloc) {
798 ps_cache_alloc = ps_cache_alloc * 2 + 64;
799 ps_cache = realloc(ps_cache, ps_cache_alloc
800 * sizeof(ps_cache[0]));
801 my_assert_not(ps_cache, NULL);
802 memset(ps_cache + ps_cache_size, 0,
803 (ps_cache_alloc - ps_cache_size)
804 * sizeof(ps_cache[0]));
807 ps = &ps_cache[ps_cache_size++];
808 ret = sscanf(proto, "struct %255s {", ps->name);
810 printf("%s:%d: struct parse failed\n", hdrfn, *line);
814 while (fgets(lstr, sizeof(lstr), fhdr))
819 if (p[0] == '/' && p[1] == '/')
824 if (m >= ARRAY_SIZE(ps->members)) {
825 printf("%s:%d: too many struct members\n",
831 ret = parse_protostr(p, &ps->members[m].pp);
833 printf("%s:%d: struct member #%d/%02x "
834 "doesn't parse\n", hdrfn, *line,
838 ps->members[m].offset = offset;
843 ps->member_count = m;
848 // parsed proto cache
849 static struct parsed_proto *pp_cache;
850 static int pp_cache_size;
851 static int pp_cache_alloc;
853 static int b_pp_c_handler(char *proto, const char *fname,
854 int is_include, int is_osinc, int is_cinc)
858 if (pp_cache_size >= pp_cache_alloc) {
859 pp_cache_alloc = pp_cache_alloc * 2 + 64;
860 pp_cache = realloc(pp_cache, pp_cache_alloc
861 * sizeof(pp_cache[0]));
862 my_assert_not(pp_cache, NULL);
863 memset(pp_cache + pp_cache_size, 0,
864 (pp_cache_alloc - pp_cache_size)
865 * sizeof(pp_cache[0]));
868 ret = parse_protostr(proto, &pp_cache[pp_cache_size]);
872 pp_cache[pp_cache_size].is_include = is_include;
873 pp_cache[pp_cache_size].is_osinc = is_osinc;
874 pp_cache[pp_cache_size].is_cinc = is_cinc;
879 static void build_caches(FILE *fhdr)
887 ret = do_protostrs(fhdr, hdrfn, 0);
891 qsort(pp_cache, pp_cache_size, sizeof(pp_cache[0]), pp_name_cmp);
892 qsort(ps_cache, ps_cache_size, sizeof(ps_cache[0]), ps_name_cmp);
893 fseek(fhdr, pos, SEEK_SET);
896 static const struct parsed_proto *proto_parse(FILE *fhdr, const char *sym,
899 const struct parsed_proto *pp_ret;
900 struct parsed_proto pp_search;
903 if (pp_cache == NULL)
907 if (sym[0] == '_' && !IS_START(sym, "__W"))
910 strcpy(pp_search.name, sym);
911 p = strchr(pp_search.name, '@');
915 pp_ret = bsearch(&pp_search, pp_cache, pp_cache_size,
916 sizeof(pp_cache[0]), pp_name_cmp);
917 if (pp_ret == NULL && !quiet)
918 printf("%s: sym '%s' is missing\n", hdrfn, sym);
923 static const struct parsed_proto *proto_lookup_struct(FILE *fhdr,
924 const char *type, int offset)
926 struct parsed_struct ps_search, *ps;
929 if (pp_cache == NULL)
931 if (ps_cache_size == 0)
934 while (my_isblank(*type))
936 if (!strncmp(type, "struct", 6) && my_isblank(type[6]))
939 if (sscanf(type, "%255s", ps_search.name) != 1)
942 ps = bsearch(&ps_search, ps_cache, ps_cache_size,
943 sizeof(ps_cache[0]), ps_name_cmp);
945 printf("%s: struct '%s' is missing\n",
946 hdrfn, ps_search.name);
950 for (m = 0; m < ps->member_count; m++) {
951 if (ps->members[m].offset == offset)
952 return &ps->members[m].pp;
958 static void pp_copy_arg(struct parsed_proto_arg *d,
959 const struct parsed_proto_arg *s)
961 memcpy(d, s, sizeof(*d));
963 if (s->reg != NULL) {
964 d->reg = strdup(s->reg);
965 my_assert_not(d->reg, NULL);
967 if (s->type.name != NULL) {
968 d->type.name = strdup(s->type.name);
969 my_assert_not(d->type.name, NULL);
972 d->pp = malloc(sizeof(*d->pp));
973 my_assert_not(d->pp, NULL);
974 memcpy(d->pp, s->pp, sizeof(*d->pp));
978 struct parsed_proto *proto_clone(const struct parsed_proto *pp_c)
980 struct parsed_proto *pp;
983 pp = malloc(sizeof(*pp));
984 my_assert_not(pp, NULL);
985 memcpy(pp, pp_c, sizeof(*pp)); // lazy..
987 // do the actual deep copy..
988 for (i = 0; i < pp_c->argc; i++)
989 pp_copy_arg(&pp->arg[i], &pp_c->arg[i]);
990 if (pp_c->ret_type.name != NULL)
991 pp->ret_type.name = strdup(pp_c->ret_type.name);
997 static inline int pp_cmp_func(const struct parsed_proto *pp1,
998 const struct parsed_proto *pp2)
1002 if (pp1->argc != pp2->argc || pp1->argc_reg != pp2->argc_reg)
1004 if (pp1->is_stdcall != pp2->is_stdcall)
1007 // because of poor void return detection, return is not
1008 // checked for now to avoid heaps of false positives
1010 for (i = 0; i < pp1->argc; i++) {
1011 if ((pp1->arg[i].reg != NULL) != (pp2->arg[i].reg != NULL))
1014 if ((pp1->arg[i].reg != NULL)
1015 && !IS(pp1->arg[i].reg, pp2->arg[i].reg))
1024 static inline int pp_compatible_func(
1025 const struct parsed_proto *pp_site,
1026 const struct parsed_proto *pp_callee)
1028 if (pp_cmp_func(pp_site, pp_callee) == 0)
1031 if (pp_site->argc_stack == 0 && pp_site->is_fastcall
1032 && pp_callee->argc_stack == 0
1033 && (pp_callee->is_fastcall || pp_callee->argc_reg == 0)
1034 && pp_site->argc_reg > pp_callee->argc_reg)
1035 /* fascall compatible callee doesn't use all args -> ok */
1041 static inline void pp_print(char *buf, size_t buf_size,
1042 const struct parsed_proto *pp)
1047 snprintf(buf, buf_size, "%s %s(", pp->ret_type.name, pp->name);
1050 for (i = 0; i < pp->argc_reg; i++) {
1051 snprintf(buf + l, buf_size - l, "%s%s",
1052 i == 0 ? "" : ", ", pp->arg[i].reg);
1055 if (pp->argc_stack > 0) {
1056 snprintf(buf + l, buf_size - l, "%s{%d stack}",
1057 i == 0 ? "" : ", ", pp->argc_stack);
1060 snprintf(buf + l, buf_size - l, ")");
1063 static inline void proto_release(struct parsed_proto *pp)
1067 for (i = 0; i < pp->argc; i++) {
1068 free(pp->arg[i].reg);
1069 free(pp->arg[i].type.name);
1070 free(pp->arg[i].pp);
1071 free(pp->arg[i].push_refs);
1073 if (pp->ret_type.name != NULL)
1074 free(pp->ret_type.name);
1077 (void)proto_lookup_struct;