scas, DF, more pop stack adjust, etc..
[ia32rtools.git] / tools / asmproc.c
1 #include <stdio.h>
2 #include <stdlib.h>
3 #include <string.h>
4
5 #include "my_assert.h"
6 #include "my_str.h"
7
8 struct sl_item {
9         char *name;
10         unsigned int callsites:1;
11         unsigned int found:1;
12 };
13
14 static int cmp_sym(const void *p1_, const void *p2_)
15 {
16         const struct sl_item *p1 = p1_, *p2 = p2_;
17         const char *s1 = p1->name, *s2 = p2->name;
18         int i;
19
20         for (i = 0; ; i++) {
21                 if ((s1[i] | s2[i]) == 0)
22                         break;
23
24                 if (s1[i] == s2[i])
25                         continue;
26
27                 if (s1[i] ==  0  && s2[i] == '@')
28                         break;
29                 if (s1[i] == '@' && s2[i] ==  0)
30                         break;
31
32                 return s1[i] - s2[i];
33         }
34
35         return 0;
36 }
37
38 static int cmp_sym_sort(const void *p1_, const void *p2_)
39 {
40         const struct sl_item *p1 = p1_, *p2 = p2_;
41         const char *s1 = p1->name, *s2 = p2->name;
42         int ret;
43         
44         ret = cmp_sym(p1_, p2_);
45         if (ret == 0) {
46                 printf("%s: dupe sym: '%s' '%s'\n", __func__, s1, s2);
47                 exit(1);
48         }
49         return ret;
50 }
51
52 void read_list(struct sl_item **sl_in, int *cnt, int *alloc,
53         FILE *f, int callsites)
54 {
55         struct sl_item *sl = *sl_in;
56         int c = *cnt;
57         char line[256];
58         char word[256];
59
60         while (fgets(line, sizeof(line), f) != NULL) {
61                 next_word(word, sizeof(word), line);
62                 if (word[0] == 0 || word[0] == ';' || word[0] == '#')
63                         continue;
64
65                 sl[c].name = strdup(word);
66                 sl[c].callsites = callsites;
67                 sl[c].found = 0;
68                 c++;
69
70                 if (c >= *alloc) {
71                         *alloc *= 2;
72                         sl = realloc(sl, *alloc * sizeof(sl[0]));
73                         my_assert_not(sl, NULL);
74                         memset(sl + c, 0, (*alloc - c) * sizeof(sl[0]));
75                 }
76         }
77
78         *sl_in = sl;
79         *cnt = c;
80 }
81
82 const char *sym_use(const struct sl_item *sym)
83 {
84         static char buf[256+3];
85         int ret;
86
87         ret = snprintf(buf, sizeof(buf), "rm_%s", sym->name);
88         if (ret >= sizeof(buf)) {
89                 printf("truncation detected: '%s'\n", buf);
90                 exit(1);
91         }
92
93         return buf;
94 }
95
96 #define IS(w, y) !strcasecmp(w, y)
97 #define IS_OR2(w, x, y) (IS(w, x) || IS(w, y))
98 #define IS_OR3(w, x, y, z) (IS(w, x) || IS(w, y) || IS(w, z))
99
100 int main(int argc, char *argv[])
101 {
102         struct sl_item *symlist, *sym, ssym = { NULL, };
103         int patch_callsites = 0;
104         FILE *fout, *fin, *f;
105         int symlist_alloc;
106         int symlist_cnt;
107         char line[256];
108         char word[256];
109         char word2[256];
110         char word3[256];
111         char word4[256];
112         char word5[256];
113         char word6[256];
114         char func[256];
115         char *p, *p2;
116         int i;
117
118         if (argc < 4) {
119                 // -c - patch callsites
120                 printf("usage:\n%s <asmf_out> <asmf_in> [[-c] <listf>]*>\n",
121                         argv[0]);
122                 return 1;
123         }
124
125         symlist_alloc = 16;
126         symlist_cnt = 0;
127         symlist = calloc(symlist_alloc, sizeof(symlist[0]));
128         my_assert_not(symlist, NULL);
129
130         for (i = 3; i < argc; i++) {
131                 if (strcmp(argv[i], "-c") == 0) {
132                         patch_callsites = 1;
133                         continue;
134                 }
135
136                 f = fopen(argv[i], "r");
137                 my_assert_not(f, NULL);
138                 read_list(&symlist, &symlist_cnt, &symlist_alloc,
139                         f, patch_callsites);
140                 fclose(f);
141
142                 patch_callsites = 0;
143         }
144
145         qsort(symlist, symlist_cnt, sizeof(symlist[0]), cmp_sym_sort);
146
147 #if 0
148         printf("symlist:\n");
149         for (i = 0; i < symlist_cnt; i++)
150                 printf("%d '%s'\n", symlist[i].callsites, symlist[i].name);
151 #endif
152
153         fin = fopen(argv[2], "r");
154         my_assert_not(fin, NULL);
155
156         fout = fopen(argv[1], "w");
157         my_assert_not(fout, NULL);
158
159         while (fgets(line, sizeof(line), fin))
160         {
161                 p = sskip(line);
162                 if (*p == 0 || *p == ';')
163                         goto pass;
164
165                 p = sskip(next_word(word, sizeof(word), p));
166                 if (*p == 0 || *p == ';')
167                         goto pass; // need at least 2 words
168
169                 p = next_word(word2, sizeof(word2), p);
170
171                 if (IS_OR2(word2, "proc", "endp")) {
172                         if (IS(word2, "proc"))
173                                 strcpy(func, word);
174                         else
175                                 func[0] = 0;
176
177                         ssym.name = word;
178                         sym = bsearch(&ssym, symlist, symlist_cnt,
179                                 sizeof(symlist[0]), cmp_sym);
180                         if (sym != NULL) {
181                                 sym->found = 1;
182                                 fprintf(fout, "rm_%s\t%s%s", word, word2, p);
183                                 continue;
184                         }
185                 }
186
187                 if (IS_OR3(word, "call", "jmp", "public")) {
188                         ssym.name = word2;
189                         sym = bsearch(&ssym, symlist, symlist_cnt,
190                                 sizeof(symlist[0]), cmp_sym);
191                         if (sym != NULL
192                             && (sym->callsites || IS(word2, func)))
193                         {
194                                 fprintf(fout, "\t\t%s\t%s%s", word,
195                                         sym_use(sym), p);
196                                 continue;
197                         }
198                 }
199
200                 p = sskip(p);
201                 if (*p == 0 || *p == ';')
202                         goto pass; // need at least 3 words
203
204                 p = next_word(word3, sizeof(word3), p);
205
206                 // push offset <sym>
207                 // jcc short <sym>
208                 if ( (IS(word, "push") && IS(word2, "offset"))
209                   || (word[0] == 'j' && IS(word2, "short")) ) {
210                         ssym.name = word3;
211                         sym = bsearch(&ssym, symlist, symlist_cnt,
212                                 sizeof(symlist[0]), cmp_sym);
213                         if (sym != NULL
214                             && (sym->callsites || IS(word3, func)))
215                         {
216                                 fprintf(fout, "\t\t%s %s %s%s",
217                                         word, word2, sym_use(sym), p);
218                                 continue;
219                         }
220                 }
221
222                 // dd offset <sym>
223                 if (IS(word, "dd") && IS(word2, "offset")) {
224                         fprintf(fout, "\t\tdd");
225                         strcpy(word, word3);
226                         goto offset_loop;
227                 }
228
229                 p = sskip(p);
230                 if (*p == 0 || *p == ';')
231                         goto pass; // need at least 4 words
232
233                 p = next_word(word4, sizeof(word4), p);
234
235                 // <name> dd offset <sym>
236                 if (IS(word2, "dd") && IS(word3, "offset")) {
237                         fprintf(fout, "%s\tdd", word);
238                         strcpy(word, word4);
239                         goto offset_loop;
240                 }
241
242                 // mov <something>, offset <sym>
243                 // jcc <some> ptr <sym>
244                 if ( (IS(word, "mov") && IS(word3, "offset"))
245                   || (word[0] == 'j' && IS(word3, "ptr")) ) {
246                         ssym.name = word4;
247                         sym = bsearch(&ssym, symlist, symlist_cnt,
248                                 sizeof(symlist[0]), cmp_sym);
249                         if (sym != NULL && sym->callsites) {
250                                 fprintf(fout, "\t\t%s\t%s %s %s%s",
251                                         word, word2, word3,
252                                         sym_use(sym), p);
253                                 continue;
254                         }
255                 }
256
257                 p = sskip(p);
258                 if (*p == 0 || *p == ';')
259                         goto pass; // need at least 5 words
260
261                 p = next_word(word5, sizeof(word5), p);
262
263                 p = sskip(p);
264                 if (*p == 0 || *p == ';')
265                         goto pass; // need at least 6 words
266
267                 p = next_word(word6, sizeof(word6), p);
268
269                 // <op> dword ptr <something>, offset <sym>
270                 if ( IS(word2, "dword") && IS(word3, "ptr")
271                   && IS(word5, "offset") ) {
272                         ssym.name = word6;
273                         sym = bsearch(&ssym, symlist, symlist_cnt,
274                                 sizeof(symlist[0]), cmp_sym);
275                         if (sym != NULL && sym->callsites) {
276                                 fprintf(fout, "\t\t%s\tdword ptr %s offset %s%s",
277                                         word, word4, sym_use(sym), p);
278                                 continue;
279                         }
280                 }
281
282 pass:
283                 fwrite(line, 1, strlen(line), fout);
284                 continue;
285
286 offset_loop:
287                 while (1) {
288                         p2 = strchr(word, ',');
289                         if (p2)
290                                 *p2 = 0;
291
292                         ssym.name = word;
293                         sym = bsearch(&ssym, symlist, symlist_cnt,
294                                 sizeof(symlist[0]), cmp_sym);
295                         fprintf(fout, " offset %s%s",
296                                 (sym != NULL && sym->callsites) ? sym_use(sym) : word,
297                                 p2 ? "," : "");
298
299                         p2 = next_word(word, sizeof(word), p);
300                         if (word[0] == 0 || word[0] == ';') {
301                                 break;
302                         }
303                         if (!IS(word, "offset")) {
304                                 printf("could not handle offset array\n");
305                                 break;
306                         }
307                         p = next_word(word, sizeof(word), p2);
308                 }
309                 fprintf(fout, "%s", p);
310                 continue;
311         }
312
313         for (i = 0; i < symlist_cnt; i++) {
314                 if (!symlist[i].found)
315                         printf("warning: sym '%s' not found\n", symlist[i].name);
316         }
317
318         fclose(fin);
319         fclose(fout);
320
321         return 0;
322 }
323
324 // vim:ts=2:shiftwidth=2