33c35af6 |
1 | #define _GNU_SOURCE |
c36e914d |
2 | #include <stdio.h> |
3 | #include <stdlib.h> |
4 | #include <string.h> |
5 | |
6 | #include "my_assert.h" |
7 | #include "my_str.h" |
8 | |
9 | #define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0])) |
10 | #define IS(w, y) !strcmp(w, y) |
39b168b8 |
11 | #define IS_START(w, y) !strncmp(w, y, strlen(y)) |
c36e914d |
12 | |
13 | #include "protoparse.h" |
14 | |
15 | const char *asmfn; |
16 | static int asmln; |
17 | |
940e8e66 |
18 | #define anote(fmt, ...) \ |
19 | printf("%s:%d: note: " fmt, asmfn, asmln, ##__VA_ARGS__) |
c36e914d |
20 | #define awarn(fmt, ...) \ |
940e8e66 |
21 | printf("%s:%d: warning: " fmt, asmfn, asmln, ##__VA_ARGS__) |
c36e914d |
22 | #define aerr(fmt, ...) do { \ |
940e8e66 |
23 | printf("%s:%d: error: " fmt, asmfn, asmln, ##__VA_ARGS__); \ |
33c35af6 |
24 | fcloseall(); \ |
c36e914d |
25 | exit(1); \ |
26 | } while (0) |
27 | |
69a3cdfc |
28 | enum op_flags { |
87bf6cec |
29 | OPF_RMD = (1 << 0), /* removed or optimized out */ |
30 | OPF_DATA = (1 << 1), /* data processing - writes to dst opr */ |
31 | OPF_FLAGS = (1 << 2), /* sets flags */ |
32 | OPF_JMP = (1 << 3), /* branches, ret and call */ |
33 | OPF_CC = (1 << 4), /* uses flags */ |
34 | OPF_TAIL = (1 << 5), /* ret or tail call */ |
33c35af6 |
35 | OPF_REP = (1 << 6), /* prefixed by rep */ |
d4e3b5db |
36 | OPF_RSAVE = (1 << 7), /* push/pop is local reg save/load */ |
c36e914d |
37 | }; |
38 | |
39 | enum op_op { |
40 | OP_INVAL, |
33c35af6 |
41 | OP_NOP, |
c36e914d |
42 | OP_PUSH, |
43 | OP_POP, |
44 | OP_MOV, |
850c9265 |
45 | OP_LEA, |
46 | OP_MOVZX, |
47 | OP_MOVSX, |
48 | OP_NOT, |
5101a5f9 |
49 | OP_CDQ, |
33c35af6 |
50 | OP_STOS, |
d4e3b5db |
51 | OP_MOVS, |
c36e914d |
52 | OP_RET, |
53 | OP_ADD, |
91977a1c |
54 | OP_SUB, |
850c9265 |
55 | OP_AND, |
56 | OP_OR, |
57 | OP_XOR, |
58 | OP_SHL, |
59 | OP_SHR, |
60 | OP_SAR, |
d4e3b5db |
61 | OP_ROL, |
62 | OP_ROR, |
69a3cdfc |
63 | OP_ADC, |
850c9265 |
64 | OP_SBB, |
65 | OP_INC, |
66 | OP_DEC, |
5101a5f9 |
67 | OP_NEG, |
850c9265 |
68 | OP_MUL, |
69 | OP_IMUL, |
5101a5f9 |
70 | OP_DIV, |
71 | OP_IDIV, |
c36e914d |
72 | OP_TEST, |
73 | OP_CMP, |
74 | OP_CALL, |
75 | OP_JMP, |
76 | OP_JO, |
77 | OP_JNO, |
78 | OP_JC, |
79 | OP_JNC, |
80 | OP_JZ, |
81 | OP_JNZ, |
82 | OP_JBE, |
83 | OP_JA, |
84 | OP_JS, |
85 | OP_JNS, |
86 | OP_JP, |
87 | OP_JNP, |
88 | OP_JL, |
89 | OP_JGE, |
90 | OP_JLE, |
91 | OP_JG, |
92 | }; |
93 | |
94 | enum opr_type { |
87bf6cec |
95 | OPT_UNSPEC, |
96 | OPT_REG, |
97 | OPT_REGMEM, |
98 | OPT_LABEL, |
850c9265 |
99 | OPT_OFFSET, |
87bf6cec |
100 | OPT_CONST, |
c36e914d |
101 | }; |
102 | |
103 | enum opr_lenmod { |
91977a1c |
104 | OPLM_UNSPEC, |
105 | OPLM_BYTE, |
106 | OPLM_WORD, |
107 | OPLM_DWORD, |
c36e914d |
108 | }; |
109 | |
850c9265 |
110 | #define MAX_OPERANDS 3 |
c36e914d |
111 | |
112 | struct parsed_opr { |
91977a1c |
113 | enum opr_type type; |
114 | enum opr_lenmod lmod; |
115 | int reg; |
116 | unsigned int val; |
117 | char name[256]; |
c36e914d |
118 | }; |
119 | |
120 | struct parsed_op { |
91977a1c |
121 | enum op_op op; |
122 | struct parsed_opr operand[MAX_OPERANDS]; |
69a3cdfc |
123 | unsigned int flags; |
91977a1c |
124 | int operand_cnt; |
69a3cdfc |
125 | int regmask_src; // all referensed regs |
126 | int regmask_dst; |
940e8e66 |
127 | int pfomask; // flagop: parsed_flag_op that can't be delayed |
128 | int argmask; // push: args that are altered before call |
129 | int cc_scratch; // scratch storage during analysis |
87bf6cec |
130 | int bt_i; // branch target (for branches) |
940e8e66 |
131 | struct parsed_op *lrl; // label reference list entry |
91977a1c |
132 | void *datap; |
133 | }; |
134 | |
69a3cdfc |
135 | // datap: |
69a3cdfc |
136 | // OP_CALL - ptr to parsed_proto |
137 | // (OPF_CC) - point to corresponding (OPF_FLAGS) |
138 | |
91977a1c |
139 | struct parsed_equ { |
140 | char name[64]; |
141 | enum opr_lenmod lmod; |
142 | int offset; |
c36e914d |
143 | }; |
144 | |
1bafb621 |
145 | enum ida_func_attr { |
146 | IDAFA_BP_FRAME = (1 << 0), |
147 | IDAFA_LIB_FUNC = (1 << 1), |
148 | IDAFA_STATIC = (1 << 2), |
149 | IDAFA_NORETURN = (1 << 3), |
150 | IDAFA_THUNK = (1 << 4), |
151 | IDAFA_FPD = (1 << 5), |
152 | }; |
153 | |
154 | #define MAX_OPS 4096 |
c36e914d |
155 | |
156 | static struct parsed_op ops[MAX_OPS]; |
91977a1c |
157 | static struct parsed_equ *g_eqs; |
158 | static int g_eqcnt; |
159 | static char g_labels[MAX_OPS][32]; |
940e8e66 |
160 | static struct parsed_op *g_label_refs[MAX_OPS]; |
91977a1c |
161 | static struct parsed_proto g_func_pp; |
162 | static char g_func[256]; |
163 | static char g_comment[256]; |
164 | static int g_bp_frame; |
1bafb621 |
165 | static int g_sp_frame; |
166 | static int g_stack_fsz; |
91977a1c |
167 | #define ferr(op_, fmt, ...) do { \ |
850c9265 |
168 | printf("error:%s:#%ld: '%s': " fmt, g_func, (op_) - ops, \ |
169 | dump_op(op_), ##__VA_ARGS__); \ |
33c35af6 |
170 | fcloseall(); \ |
91977a1c |
171 | exit(1); \ |
172 | } while (0) |
173 | |
174 | #define MAX_REGS 8 |
175 | |
176 | const char *regs_r32[] = { "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp" }; |
177 | const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" }; |
178 | const char *regs_r8l[] = { "al", "bl", "cl", "dl" }; |
179 | const char *regs_r8h[] = { "ah", "bh", "ch", "dh" }; |
180 | |
181 | enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP }; |
182 | |
69a3cdfc |
183 | // possible basic comparison types (without inversion) |
184 | enum parsed_flag_op { |
185 | PFO_O, // 0 OF=1 |
186 | PFO_C, // 2 CF=1 |
187 | PFO_Z, // 4 ZF=1 |
188 | PFO_BE, // 6 CF=1||ZF=1 |
189 | PFO_S, // 8 SF=1 |
190 | PFO_P, // a PF=1 |
191 | PFO_L, // c SF!=OF |
192 | PFO_LE, // e ZF=1||SF!=OF |
193 | }; |
194 | |
195 | static const char *parsed_flag_op_names[] = { |
196 | "o", "c", "z", "be", "s", "p", "l", "le" |
197 | }; |
198 | |
91977a1c |
199 | static int char_array_i(const char *array[], size_t len, const char *s) |
200 | { |
201 | int i; |
c36e914d |
202 | |
91977a1c |
203 | for (i = 0; i < len; i++) |
204 | if (IS(s, array[i])) |
205 | return i; |
c36e914d |
206 | |
91977a1c |
207 | return -1; |
208 | } |
209 | |
5101a5f9 |
210 | static void printf_number(char *buf, size_t buf_size, long number) |
91977a1c |
211 | { |
5101a5f9 |
212 | // output in C-friendly form |
213 | snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number); |
214 | } |
91977a1c |
215 | |
5101a5f9 |
216 | static int parse_reg(enum opr_lenmod *reg_lmod, const char *s) |
217 | { |
218 | int reg; |
91977a1c |
219 | |
5101a5f9 |
220 | reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s); |
221 | if (reg >= 0) { |
222 | *reg_lmod = OPLM_DWORD; |
223 | return reg; |
91977a1c |
224 | } |
5101a5f9 |
225 | reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s); |
226 | if (reg >= 0) { |
227 | *reg_lmod = OPLM_WORD; |
228 | return reg; |
229 | } |
230 | reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s); |
231 | if (reg >= 0) { |
232 | *reg_lmod = OPLM_BYTE; |
233 | return reg; |
234 | } |
235 | reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s); |
236 | if (reg >= 0) { |
237 | *reg_lmod = OPLM_BYTE; |
238 | return reg; |
850c9265 |
239 | } |
240 | |
241 | return -1; |
91977a1c |
242 | } |
243 | |
244 | static long parse_number(const char *number) |
245 | { |
246 | int len = strlen(number); |
247 | const char *p = number; |
248 | char *endp = NULL; |
249 | int neg = 0; |
250 | int bad; |
251 | long ret; |
252 | |
253 | if (*p == '-') { |
254 | neg = 1; |
255 | p++; |
256 | } |
257 | if (len > 1 && *p == '0') |
258 | p++; |
259 | if (number[len - 1] == 'h') { |
260 | ret = strtol(p, &endp, 16); |
261 | bad = (*endp != 'h'); |
262 | } |
263 | else { |
264 | ret = strtol(p, &endp, 10); |
265 | bad = (*endp != 0); |
266 | } |
267 | if (bad) |
268 | aerr("number parsing failed\n"); |
269 | if (neg) |
270 | ret = -ret; |
271 | return ret; |
272 | } |
273 | |
5101a5f9 |
274 | static int parse_indmode(char *name, int *regmask, int need_c_cvt) |
275 | { |
276 | enum opr_lenmod lmod; |
277 | char cvtbuf[256]; |
278 | char *d = cvtbuf; |
279 | char *s = name; |
280 | char w[64]; |
281 | long number; |
282 | int reg; |
283 | int c = 0; |
284 | |
285 | *d = 0; |
286 | |
287 | while (*s != 0) { |
288 | d += strlen(d); |
289 | while (my_isblank(*s)) |
290 | s++; |
291 | for (; my_issep(*s); d++, s++) |
292 | *d = *s; |
293 | while (my_isblank(*s)) |
294 | s++; |
295 | *d = 0; |
296 | |
87bf6cec |
297 | // skip 'ds:' prefix |
39b168b8 |
298 | if (IS_START(s, "ds:")) |
87bf6cec |
299 | s += 3; |
300 | |
5101a5f9 |
301 | s = next_idt(w, sizeof(w), s); |
302 | if (w[0] == 0) |
303 | break; |
304 | c++; |
305 | |
306 | reg = parse_reg(&lmod, w); |
307 | if (reg >= 0) { |
308 | *regmask |= 1 << reg; |
309 | goto pass; |
310 | } |
311 | |
312 | if ('0' <= w[0] && w[0] <= '9') { |
313 | number = parse_number(w); |
314 | printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number); |
315 | continue; |
316 | } |
317 | |
318 | // probably some label/identifier - pass |
319 | |
320 | pass: |
321 | snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w); |
322 | } |
323 | |
324 | if (need_c_cvt) |
325 | strcpy(name, cvtbuf); |
326 | |
327 | return c; |
328 | } |
329 | |
d4e3b5db |
330 | static const char *parse_stack_el(const char *name) |
331 | { |
1bafb621 |
332 | const char *p, *s; |
333 | char *endp = NULL; |
d4e3b5db |
334 | char buf[32]; |
1bafb621 |
335 | long val; |
d4e3b5db |
336 | int len; |
337 | |
39b168b8 |
338 | if (IS_START(name, "ebp+") |
d4e3b5db |
339 | && !('0' <= name[4] && name[4] <= '9')) |
340 | { |
341 | return name + 4; |
342 | } |
39b168b8 |
343 | if (!IS_START(name, "esp+")) |
d4e3b5db |
344 | return NULL; |
345 | |
346 | p = strchr(name + 4, '+'); |
347 | if (p) { |
1bafb621 |
348 | // must be a number after esp+, already converted to 0x.. |
349 | s = name + 4; |
350 | if (!('0' <= *s && *s <= '9')) { |
351 | aerr("%s nan?\n", __func__); |
d4e3b5db |
352 | return NULL; |
1bafb621 |
353 | } |
354 | if (s[0] == '0' && s[1] == 'x') |
355 | s += 2; |
356 | len = p - s; |
d4e3b5db |
357 | if (len < sizeof(buf) - 1) { |
1bafb621 |
358 | strncpy(buf, s, len); |
d4e3b5db |
359 | buf[len] = 0; |
1bafb621 |
360 | val = strtol(buf, &endp, 16); |
361 | if (val == 0 || *endp != 0) { |
362 | aerr("%s num parse fail for '%s'\n", __func__, buf); |
363 | return NULL; |
364 | } |
d4e3b5db |
365 | } |
366 | p++; |
367 | } |
368 | else |
369 | p = name + 4; |
370 | |
371 | if ('0' <= *p && *p <= '9') |
372 | return NULL; |
373 | |
374 | return p; |
375 | } |
376 | |
850c9265 |
377 | static int guess_lmod_from_name(struct parsed_opr *opr) |
378 | { |
379 | if (!strncmp(opr->name, "dword_", 6)) { |
380 | opr->lmod = OPLM_DWORD; |
381 | return 1; |
382 | } |
383 | if (!strncmp(opr->name, "word_", 5)) { |
384 | opr->lmod = OPLM_WORD; |
385 | return 1; |
386 | } |
387 | if (!strncmp(opr->name, "byte_", 5)) { |
388 | opr->lmod = OPLM_BYTE; |
389 | return 1; |
390 | } |
391 | return 0; |
392 | } |
393 | |
5101a5f9 |
394 | static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod, |
395 | int *regmask) |
396 | { |
397 | opr->type = OPT_REG; |
398 | opr->reg = reg; |
399 | opr->lmod = lmod; |
400 | *regmask |= 1 << reg; |
401 | } |
402 | |
39b168b8 |
403 | static struct parsed_equ *equ_find(struct parsed_op *po, const char *name, |
404 | int *extra_offs); |
87bf6cec |
405 | |
69a3cdfc |
406 | static int parse_operand(struct parsed_opr *opr, |
407 | int *regmask, int *regmask_indirect, |
1bafb621 |
408 | char words[16][256], int wordc, int w, unsigned int op_flags) |
c36e914d |
409 | { |
850c9265 |
410 | enum opr_lenmod tmplmod; |
850c9265 |
411 | int ret, len; |
5101a5f9 |
412 | long number; |
1bafb621 |
413 | int wordc_in; |
39b168b8 |
414 | char *tmp; |
850c9265 |
415 | int i; |
c36e914d |
416 | |
1bafb621 |
417 | if (w >= wordc) |
418 | aerr("parse_operand w %d, wordc %d\n", w, wordc); |
c36e914d |
419 | |
1bafb621 |
420 | opr->reg = xUNSPEC; |
91977a1c |
421 | |
1bafb621 |
422 | for (i = w; i < wordc; i++) { |
423 | len = strlen(words[i]); |
424 | if (words[i][len - 1] == ',') { |
425 | words[i][len - 1] = 0; |
426 | wordc = i + 1; |
427 | break; |
428 | } |
429 | } |
c36e914d |
430 | |
1bafb621 |
431 | wordc_in = wordc - w; |
c36e914d |
432 | |
1bafb621 |
433 | if ((op_flags & OPF_JMP) && wordc_in > 0 |
434 | && !('0' <= words[w][0] && words[w][0] <= '9')) |
435 | { |
436 | const char *label = NULL; |
437 | |
438 | if (wordc_in == 3 && !strncmp(words[w], "near", 4) |
439 | && IS(words[w + 1], "ptr")) |
440 | label = words[w + 2]; |
441 | else if (wordc_in == 2 && IS(words[w], "short")) |
442 | label = words[w + 1]; |
443 | else if (wordc_in == 1 |
444 | && strchr(words[w], '[') == NULL |
445 | && parse_reg(&tmplmod, words[w]) < 0) |
446 | label = words[w]; |
447 | |
448 | if (label != NULL) { |
449 | opr->type = OPT_LABEL; |
39b168b8 |
450 | if (IS_START(label, "ds:")) |
451 | label += 3; |
1bafb621 |
452 | strcpy(opr->name, label); |
453 | return wordc; |
454 | } |
455 | } |
c36e914d |
456 | |
1bafb621 |
457 | if (wordc_in >= 3) { |
458 | if (IS(words[w + 1], "ptr")) { |
459 | if (IS(words[w], "dword")) |
460 | opr->lmod = OPLM_DWORD; |
461 | else if (IS(words[w], "word")) |
462 | opr->lmod = OPLM_WORD; |
463 | else if (IS(words[w], "byte")) |
464 | opr->lmod = OPLM_BYTE; |
465 | else |
466 | aerr("type parsing failed\n"); |
467 | w += 2; |
468 | wordc_in = wordc - w; |
469 | } |
470 | } |
471 | |
39b168b8 |
472 | if (wordc_in == 2) { |
473 | if (IS(words[w], "offset")) { |
474 | opr->type = OPT_OFFSET; |
475 | strcpy(opr->name, words[w + 1]); |
476 | return wordc; |
477 | } |
478 | if (IS(words[w], "(offset")) { |
479 | char *p = strchr(words[w + 1], ')'); |
480 | if (p == NULL) |
481 | aerr("parse of bracketed offset failed\n"); |
482 | *p = 0; |
483 | opr->type = OPT_OFFSET; |
484 | strcpy(opr->name, words[w + 1]); |
485 | return wordc; |
486 | } |
1bafb621 |
487 | } |
c36e914d |
488 | |
1bafb621 |
489 | if (wordc_in != 1) |
850c9265 |
490 | aerr("parse_operand 1 word expected\n"); |
c36e914d |
491 | |
39b168b8 |
492 | tmp = words[w]; |
493 | if (IS_START(tmp, "ds:")) |
494 | tmp += 3; |
495 | strcpy(opr->name, tmp); |
c36e914d |
496 | |
850c9265 |
497 | if (words[w][0] == '[') { |
498 | opr->type = OPT_REGMEM; |
499 | ret = sscanf(words[w], "[%[^]]]", opr->name); |
500 | if (ret != 1) |
501 | aerr("[] parse failure\n"); |
87bf6cec |
502 | |
5101a5f9 |
503 | parse_indmode(opr->name, regmask_indirect, 1); |
d4e3b5db |
504 | if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name)) { |
87bf6cec |
505 | // might be an equ |
d4e3b5db |
506 | struct parsed_equ *eq = |
39b168b8 |
507 | equ_find(NULL, parse_stack_el(opr->name), &i); |
87bf6cec |
508 | if (eq) |
509 | opr->lmod = eq->lmod; |
510 | } |
850c9265 |
511 | return wordc; |
512 | } |
513 | else if (strchr(words[w], '[')) { |
514 | // label[reg] form |
515 | opr->type = OPT_REGMEM; |
516 | if (opr->lmod == OPLM_UNSPEC) |
517 | guess_lmod_from_name(opr); |
5101a5f9 |
518 | parse_indmode(strchr(words[w], '['), regmask_indirect, 0); |
850c9265 |
519 | return wordc; |
520 | } |
521 | else if (('0' <= words[w][0] && words[w][0] <= '9') |
522 | || words[w][0] == '-') |
523 | { |
5101a5f9 |
524 | number = parse_number(words[w]); |
91977a1c |
525 | opr->type = OPT_CONST; |
5101a5f9 |
526 | opr->val = number; |
527 | printf_number(opr->name, sizeof(opr->name), number); |
91977a1c |
528 | return wordc; |
850c9265 |
529 | } |
c36e914d |
530 | |
5101a5f9 |
531 | ret = parse_reg(&tmplmod, opr->name); |
532 | if (ret >= 0) { |
533 | setup_reg_opr(opr, ret, tmplmod, regmask); |
850c9265 |
534 | return wordc; |
535 | } |
536 | |
537 | // most likely var in data segment |
538 | opr->type = OPT_LABEL; |
539 | if (opr->lmod == OPLM_UNSPEC) |
540 | guess_lmod_from_name(opr); |
541 | if (opr->lmod != OPLM_UNSPEC) |
542 | return wordc; |
91977a1c |
543 | |
850c9265 |
544 | // TODO: scan data seg to determine type? |
545 | return wordc; |
c36e914d |
546 | } |
547 | |
33c35af6 |
548 | static const struct { |
549 | const char *name; |
550 | unsigned int flags; |
551 | } pref_table[] = { |
552 | { "rep", OPF_REP }, |
553 | }; |
554 | |
c36e914d |
555 | static const struct { |
850c9265 |
556 | const char *name; |
557 | enum op_op op; |
69a3cdfc |
558 | unsigned int minopr; |
559 | unsigned int maxopr; |
560 | unsigned int flags; |
c36e914d |
561 | } op_table[] = { |
33c35af6 |
562 | { "nop", OP_NOP, 0, 0, 0 }, |
69a3cdfc |
563 | { "push", OP_PUSH, 1, 1, 0 }, |
564 | { "pop", OP_POP, 1, 1, OPF_DATA }, |
565 | { "mov" , OP_MOV, 2, 2, OPF_DATA }, |
566 | { "lea", OP_LEA, 2, 2, OPF_DATA }, |
567 | { "movzx",OP_MOVZX, 2, 2, OPF_DATA }, |
568 | { "movsx",OP_MOVSX, 2, 2, OPF_DATA }, |
569 | { "not", OP_NOT, 1, 1, OPF_DATA }, |
5101a5f9 |
570 | { "cdq", OP_CDQ, 0, 0, OPF_DATA }, |
33c35af6 |
571 | { "stosb",OP_STOS, 0, 0, OPF_DATA }, |
572 | { "stosw",OP_STOS, 0, 0, OPF_DATA }, |
573 | { "stosd",OP_STOS, 0, 0, OPF_DATA }, |
d4e3b5db |
574 | { "movsb",OP_MOVS, 0, 0, OPF_DATA }, |
575 | { "movsw",OP_MOVS, 0, 0, OPF_DATA }, |
576 | { "movsd",OP_MOVS, 0, 0, OPF_DATA }, |
69a3cdfc |
577 | { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS }, |
578 | { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS }, |
579 | { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS }, |
580 | { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS }, |
581 | { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS }, |
582 | { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS }, |
583 | { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS }, |
584 | { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS }, |
585 | { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS }, |
d4e3b5db |
586 | { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS }, |
587 | { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS }, |
5101a5f9 |
588 | { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC }, |
69a3cdfc |
589 | { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC }, |
590 | { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS }, |
591 | { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS }, |
5101a5f9 |
592 | { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS }, |
593 | { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS }, |
69a3cdfc |
594 | { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS }, |
5101a5f9 |
595 | { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS }, |
596 | { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS }, |
69a3cdfc |
597 | { "test", OP_TEST, 2, 2, OPF_FLAGS }, |
598 | { "cmp", OP_CMP, 2, 2, OPF_FLAGS }, |
87bf6cec |
599 | { "retn", OP_RET, 0, 1, OPF_JMP|OPF_TAIL }, |
600 | { "call", OP_CALL, 1, 1, OPF_JMP|OPF_FLAGS }, |
69a3cdfc |
601 | { "jmp", OP_JMP, 1, 1, OPF_JMP }, |
602 | { "jo", OP_JO, 1, 1, OPF_JMP|OPF_CC }, // 70 OF=1 |
603 | { "jno", OP_JNO, 1, 1, OPF_JMP|OPF_CC }, // 71 OF=0 |
604 | { "jc", OP_JC, 1, 1, OPF_JMP|OPF_CC }, // 72 CF=1 |
605 | { "jb", OP_JC, 1, 1, OPF_JMP|OPF_CC }, // 72 |
606 | { "jnc", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 CF=0 |
33c35af6 |
607 | { "jnb", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 |
69a3cdfc |
608 | { "jae", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 |
609 | { "jz", OP_JZ, 1, 1, OPF_JMP|OPF_CC }, // 74 ZF=1 |
610 | { "je", OP_JZ, 1, 1, OPF_JMP|OPF_CC }, // 74 |
611 | { "jnz", OP_JNZ, 1, 1, OPF_JMP|OPF_CC }, // 75 ZF=0 |
612 | { "jne", OP_JNZ, 1, 1, OPF_JMP|OPF_CC }, // 75 |
613 | { "jbe", OP_JBE, 1, 1, OPF_JMP|OPF_CC }, // 76 CF=1 || ZF=1 |
614 | { "jna", OP_JBE, 1, 1, OPF_JMP|OPF_CC }, // 76 |
615 | { "ja", OP_JA, 1, 1, OPF_JMP|OPF_CC }, // 77 CF=0 && ZF=0 |
616 | { "jnbe", OP_JA, 1, 1, OPF_JMP|OPF_CC }, // 77 |
617 | { "js", OP_JS, 1, 1, OPF_JMP|OPF_CC }, // 78 SF=1 |
618 | { "jns", OP_JNS, 1, 1, OPF_JMP|OPF_CC }, // 79 SF=0 |
619 | { "jp", OP_JP, 1, 1, OPF_JMP|OPF_CC }, // 7a PF=1 |
620 | { "jpe", OP_JP, 1, 1, OPF_JMP|OPF_CC }, // 7a |
621 | { "jnp", OP_JNP, 1, 1, OPF_JMP|OPF_CC }, // 7b PF=0 |
622 | { "jpo", OP_JNP, 1, 1, OPF_JMP|OPF_CC }, // 7b |
623 | { "jl", OP_JL, 1, 1, OPF_JMP|OPF_CC }, // 7c SF!=OF |
624 | { "jnge", OP_JL, 1, 1, OPF_JMP|OPF_CC }, // 7c |
625 | { "jge", OP_JGE, 1, 1, OPF_JMP|OPF_CC }, // 7d SF=OF |
626 | { "jnl", OP_JGE, 1, 1, OPF_JMP|OPF_CC }, // 7d |
627 | { "jle", OP_JLE, 1, 1, OPF_JMP|OPF_CC }, // 7e ZF=1 || SF!=OF |
628 | { "jng", OP_JLE, 1, 1, OPF_JMP|OPF_CC }, // 7e |
629 | { "jg", OP_JG, 1, 1, OPF_JMP|OPF_CC }, // 7f ZF=0 && SF=OF |
630 | { "jnle", OP_JG, 1, 1, OPF_JMP|OPF_CC }, // 7f |
5101a5f9 |
631 | { "seto", OP_JO, 1, 1, OPF_DATA|OPF_CC }, |
632 | { "setno", OP_JNO, 1, 1, OPF_DATA|OPF_CC }, |
633 | { "setc", OP_JC, 1, 1, OPF_DATA|OPF_CC }, |
634 | { "setb", OP_JC, 1, 1, OPF_DATA|OPF_CC }, |
635 | { "setnc", OP_JNC, 1, 1, OPF_DATA|OPF_CC }, |
636 | { "setae", OP_JNC, 1, 1, OPF_DATA|OPF_CC }, |
637 | { "setz", OP_JZ, 1, 1, OPF_DATA|OPF_CC }, |
638 | { "sete", OP_JZ, 1, 1, OPF_DATA|OPF_CC }, |
639 | { "setnz", OP_JNZ, 1, 1, OPF_DATA|OPF_CC }, |
640 | { "setne", OP_JNZ, 1, 1, OPF_DATA|OPF_CC }, |
641 | { "setbe", OP_JBE, 1, 1, OPF_DATA|OPF_CC }, |
642 | { "setna", OP_JBE, 1, 1, OPF_DATA|OPF_CC }, |
643 | { "seta", OP_JA, 1, 1, OPF_DATA|OPF_CC }, |
644 | { "setnbe", OP_JA, 1, 1, OPF_DATA|OPF_CC }, |
645 | { "sets", OP_JS, 1, 1, OPF_DATA|OPF_CC }, |
646 | { "setns", OP_JNS, 1, 1, OPF_DATA|OPF_CC }, |
647 | { "setp", OP_JP, 1, 1, OPF_DATA|OPF_CC }, |
648 | { "setpe", OP_JP, 1, 1, OPF_DATA|OPF_CC }, |
649 | { "setnp", OP_JNP, 1, 1, OPF_DATA|OPF_CC }, |
650 | { "setpo", OP_JNP, 1, 1, OPF_DATA|OPF_CC }, |
651 | { "setl", OP_JL, 1, 1, OPF_DATA|OPF_CC }, |
652 | { "setnge", OP_JL, 1, 1, OPF_DATA|OPF_CC }, |
653 | { "setge", OP_JGE, 1, 1, OPF_DATA|OPF_CC }, |
654 | { "setnl", OP_JGE, 1, 1, OPF_DATA|OPF_CC }, |
655 | { "setle", OP_JLE, 1, 1, OPF_DATA|OPF_CC }, |
656 | { "setng", OP_JLE, 1, 1, OPF_DATA|OPF_CC }, |
657 | { "setg", OP_JG, 1, 1, OPF_DATA|OPF_CC }, |
658 | { "setnle", OP_JG, 1, 1, OPF_DATA|OPF_CC }, |
69a3cdfc |
659 | }; |
c36e914d |
660 | |
661 | static void parse_op(struct parsed_op *op, char words[16][256], int wordc) |
662 | { |
bfa4a6ee |
663 | enum opr_lenmod lmod = OPLM_UNSPEC; |
33c35af6 |
664 | int prefix_flags = 0; |
69a3cdfc |
665 | int regmask_ind; |
666 | int regmask; |
33c35af6 |
667 | int op_w = 0; |
91977a1c |
668 | int opr = 0; |
33c35af6 |
669 | int w = 0; |
91977a1c |
670 | int i; |
c36e914d |
671 | |
33c35af6 |
672 | for (i = 0; i < ARRAY_SIZE(pref_table); i++) { |
673 | if (IS(words[w], pref_table[i].name)) { |
674 | prefix_flags = pref_table[i].flags; |
675 | break; |
676 | } |
677 | } |
678 | |
679 | if (prefix_flags) { |
680 | if (wordc <= 1) |
681 | aerr("lone prefix: '%s'\n", words[0]); |
682 | w++; |
683 | } |
684 | |
685 | op_w = w; |
91977a1c |
686 | for (i = 0; i < ARRAY_SIZE(op_table); i++) { |
33c35af6 |
687 | if (IS(words[w], op_table[i].name)) |
69a3cdfc |
688 | break; |
689 | } |
c36e914d |
690 | |
69a3cdfc |
691 | if (i == ARRAY_SIZE(op_table)) |
692 | aerr("unhandled op: '%s'\n", words[0]); |
33c35af6 |
693 | w++; |
c36e914d |
694 | |
69a3cdfc |
695 | op->op = op_table[i].op; |
33c35af6 |
696 | op->flags = op_table[i].flags | prefix_flags; |
69a3cdfc |
697 | op->regmask_src = op->regmask_dst = 0; |
698 | |
699 | for (opr = 0; opr < op_table[i].minopr; opr++) { |
700 | regmask = regmask_ind = 0; |
701 | w = parse_operand(&op->operand[opr], ®mask, ®mask_ind, |
702 | words, wordc, w, op->flags); |
703 | |
704 | if (opr == 0 && (op->flags & OPF_DATA)) |
705 | op->regmask_dst = regmask; |
706 | // for now, mark dst as src too |
707 | op->regmask_src |= regmask | regmask_ind; |
708 | } |
c36e914d |
709 | |
69a3cdfc |
710 | for (; w < wordc && opr < op_table[i].maxopr; opr++) { |
711 | w = parse_operand(&op->operand[opr], |
712 | &op->regmask_src, &op->regmask_src, |
713 | words, wordc, w, op->flags); |
91977a1c |
714 | } |
c36e914d |
715 | |
91977a1c |
716 | if (w < wordc) |
717 | aerr("parse_op %s incomplete: %d/%d\n", |
718 | words[0], w, wordc); |
5101a5f9 |
719 | |
720 | // special cases |
721 | op->operand_cnt = opr; |
722 | if (!strncmp(op_table[i].name, "set", 3)) |
723 | op->operand[0].lmod = OPLM_BYTE; |
724 | |
725 | // ops with implicit argumets |
726 | switch (op->op) { |
727 | case OP_CDQ: |
728 | op->operand_cnt = 2; |
729 | setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst); |
730 | setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src); |
731 | break; |
732 | |
33c35af6 |
733 | case OP_STOS: |
734 | if (op->operand_cnt != 0) |
735 | break; |
736 | if (IS(words[op_w], "stosb")) |
737 | lmod = OPLM_BYTE; |
738 | else if (IS(words[op_w], "stosw")) |
739 | lmod = OPLM_WORD; |
740 | else if (IS(words[op_w], "stosd")) |
741 | lmod = OPLM_DWORD; |
742 | op->operand_cnt = 3; |
d4e3b5db |
743 | setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src); |
744 | setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src); |
745 | op->regmask_dst = op->regmask_src; |
33c35af6 |
746 | setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD, &op->regmask_src); |
747 | break; |
748 | |
d4e3b5db |
749 | case OP_MOVS: |
750 | if (op->operand_cnt != 0) |
751 | break; |
752 | if (IS(words[op_w], "movsb")) |
753 | lmod = OPLM_BYTE; |
754 | else if (IS(words[op_w], "movsw")) |
755 | lmod = OPLM_WORD; |
756 | else if (IS(words[op_w], "movsd")) |
757 | lmod = OPLM_DWORD; |
758 | op->operand_cnt = 3; |
759 | setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src); |
760 | setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src); |
761 | setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src); |
762 | op->regmask_dst = op->regmask_src; |
763 | break; |
764 | |
5101a5f9 |
765 | case OP_IMUL: |
766 | if (op->operand_cnt != 1) |
767 | break; |
768 | // fallthrough |
769 | case OP_MUL: |
770 | // singleop mul |
771 | op->regmask_dst = (1 << xDX) | (1 << xAX); |
772 | op->regmask_src |= (1 << xAX); |
773 | if (op->operand[0].lmod == OPLM_UNSPEC) |
774 | op->operand[0].lmod = OPLM_DWORD; |
775 | break; |
776 | |
777 | case OP_DIV: |
778 | case OP_IDIV: |
779 | // we could set up operands for edx:eax, but there is no real need to |
780 | // (see is_opr_modified()) |
781 | regmask = (1 << xDX) | (1 << xAX); |
782 | op->regmask_dst = regmask; |
783 | op->regmask_src |= regmask; |
784 | if (op->operand[0].lmod == OPLM_UNSPEC) |
785 | op->operand[0].lmod = OPLM_DWORD; |
786 | break; |
787 | |
788 | case OP_SHL: |
789 | case OP_SHR: |
790 | case OP_SAR: |
d4e3b5db |
791 | case OP_ROL: |
792 | case OP_ROR: |
5101a5f9 |
793 | if (op->operand[1].lmod == OPLM_UNSPEC) |
794 | op->operand[1].lmod = OPLM_BYTE; |
795 | break; |
796 | |
797 | default: |
798 | break; |
799 | } |
c36e914d |
800 | } |
801 | |
850c9265 |
802 | static const char *op_name(enum op_op op) |
803 | { |
804 | int i; |
805 | |
806 | for (i = 0; i < ARRAY_SIZE(op_table); i++) |
807 | if (op_table[i].op == op) |
808 | return op_table[i].name; |
809 | |
810 | return "???"; |
811 | } |
812 | |
813 | // debug |
814 | static const char *dump_op(struct parsed_op *po) |
815 | { |
816 | static char out[128]; |
817 | char *p = out; |
818 | int i; |
819 | |
820 | snprintf(out, sizeof(out), "%s", op_name(po->op)); |
821 | for (i = 0; i < po->operand_cnt; i++) { |
822 | p += strlen(p); |
823 | if (i > 0) |
824 | *p++ = ','; |
825 | snprintf(p, sizeof(out) - (p - out), |
826 | po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s", |
827 | po->operand[i].name); |
828 | } |
829 | |
830 | return out; |
831 | } |
832 | |
d4e3b5db |
833 | static const char *lmod_cast_u(struct parsed_op *po, |
834 | enum opr_lenmod lmod) |
835 | { |
836 | switch (lmod) { |
837 | case OPLM_DWORD: |
838 | return ""; |
839 | case OPLM_WORD: |
840 | return "(u16)"; |
841 | case OPLM_BYTE: |
842 | return "(u8)"; |
843 | default: |
844 | ferr(po, "invalid lmod: %d\n", lmod); |
845 | return "(_invalid_)"; |
846 | } |
847 | } |
848 | |
849 | static const char *lmod_cast_u_ptr(struct parsed_op *po, |
850 | enum opr_lenmod lmod) |
851 | { |
852 | switch (lmod) { |
853 | case OPLM_DWORD: |
854 | return "*(u32 *)"; |
855 | case OPLM_WORD: |
856 | return "*(u16 *)"; |
857 | case OPLM_BYTE: |
858 | return "*(u8 *)"; |
859 | default: |
860 | ferr(po, "invalid lmod: %d\n", lmod); |
861 | return "(_invalid_)"; |
862 | } |
863 | } |
864 | |
865 | static const char *lmod_cast_s(struct parsed_op *po, |
866 | enum opr_lenmod lmod) |
867 | { |
868 | switch (lmod) { |
869 | case OPLM_DWORD: |
870 | return "(s32)"; |
871 | case OPLM_WORD: |
872 | return "(s16)"; |
873 | case OPLM_BYTE: |
874 | return "(s8)"; |
875 | default: |
876 | ferr(po, "%s: invalid lmod: %d\n", __func__, lmod); |
877 | return "(_invalid_)"; |
878 | } |
879 | } |
880 | |
881 | static const char *lmod_cast(struct parsed_op *po, |
882 | enum opr_lenmod lmod, int is_signed) |
883 | { |
884 | return is_signed ? |
885 | lmod_cast_s(po, lmod) : |
886 | lmod_cast_u(po, lmod); |
887 | } |
888 | |
889 | static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod) |
890 | { |
891 | switch (lmod) { |
892 | case OPLM_DWORD: |
893 | return 4; |
894 | case OPLM_WORD: |
895 | return 2; |
896 | case OPLM_BYTE: |
897 | return 1; |
898 | default: |
899 | ferr(po, "%s: invalid lmod: %d\n", __func__, lmod); |
900 | return 0; |
901 | } |
902 | } |
903 | |
91977a1c |
904 | static const char *opr_name(struct parsed_op *po, int opr_num) |
c36e914d |
905 | { |
91977a1c |
906 | if (opr_num >= po->operand_cnt) |
907 | ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt); |
908 | return po->operand[opr_num].name; |
c36e914d |
909 | } |
910 | |
91977a1c |
911 | static unsigned int opr_const(struct parsed_op *po, int opr_num) |
c36e914d |
912 | { |
91977a1c |
913 | if (opr_num >= po->operand_cnt) |
914 | ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt); |
915 | if (po->operand[opr_num].type != OPT_CONST) |
916 | ferr(po, "opr %d: const expected\n", opr_num); |
917 | return po->operand[opr_num].val; |
918 | } |
c36e914d |
919 | |
91977a1c |
920 | static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr) |
921 | { |
922 | if ((unsigned int)popr->reg >= MAX_REGS) |
923 | ferr(po, "invalid reg: %d\n", popr->reg); |
924 | return regs_r32[popr->reg]; |
925 | } |
c36e914d |
926 | |
39b168b8 |
927 | static struct parsed_equ *equ_find(struct parsed_op *po, const char *name, |
928 | int *extra_offs) |
91977a1c |
929 | { |
39b168b8 |
930 | const char *p; |
931 | char *endp; |
932 | int namelen; |
850c9265 |
933 | int i; |
934 | |
39b168b8 |
935 | *extra_offs = 0; |
936 | namelen = strlen(name); |
937 | |
938 | p = strchr(name, '+'); |
939 | if (p != NULL) { |
940 | namelen = p - name; |
941 | if (namelen <= 0) |
942 | ferr(po, "equ parse failed for '%s'\n", name); |
943 | |
944 | if (IS_START(p, "0x")) |
945 | p += 2; |
946 | *extra_offs = strtol(p, &endp, 16); |
947 | if (*endp != 0) |
948 | ferr(po, "equ parse failed for '%s'\n", name); |
949 | } |
950 | |
850c9265 |
951 | for (i = 0; i < g_eqcnt; i++) |
39b168b8 |
952 | if (strncmp(g_eqs[i].name, name, namelen) == 0 |
953 | && g_eqs[i].name[namelen] == 0) |
850c9265 |
954 | break; |
87bf6cec |
955 | if (i >= g_eqcnt) { |
956 | if (po != NULL) |
957 | ferr(po, "unresolved equ name: '%s'\n", name); |
958 | return NULL; |
959 | } |
850c9265 |
960 | |
961 | return &g_eqs[i]; |
962 | } |
963 | |
1bafb621 |
964 | static void stack_frame_access(struct parsed_op *po, |
965 | enum opr_lenmod lmod, char *buf, size_t buf_size, |
966 | const char *name, int is_src, int is_lea) |
850c9265 |
967 | { |
968 | const char *prefix = ""; |
91977a1c |
969 | struct parsed_equ *eq; |
970 | int i, arg_i, arg_s; |
d4e3b5db |
971 | const char *bp_arg; |
972 | int stack_ra = 0; |
39b168b8 |
973 | int offset = 0; |
850c9265 |
974 | int sf_ofs; |
91977a1c |
975 | |
d4e3b5db |
976 | bp_arg = parse_stack_el(name); |
91977a1c |
977 | snprintf(g_comment, sizeof(g_comment), "%s", bp_arg); |
39b168b8 |
978 | eq = equ_find(po, bp_arg, &offset); |
d4e3b5db |
979 | if (eq == NULL) |
980 | ferr(po, "detected but missing eq\n"); |
91977a1c |
981 | |
39b168b8 |
982 | offset += eq->offset; |
983 | |
d4e3b5db |
984 | if (!strncmp(name, "ebp", 3)) |
985 | stack_ra = 4; |
986 | |
39b168b8 |
987 | if (stack_ra <= offset && offset < stack_ra + 4) |
988 | ferr(po, "reference to ra? %d %d\n", offset, stack_ra); |
d4e3b5db |
989 | |
39b168b8 |
990 | if (offset > stack_ra) { |
991 | arg_i = (offset - stack_ra - 4) / 4; |
91977a1c |
992 | if (arg_i < 0 || arg_i >= g_func_pp.argc_stack) |
1bafb621 |
993 | ferr(po, "offset %d (%s) doesn't map to any arg\n", |
39b168b8 |
994 | offset, bp_arg); |
91977a1c |
995 | |
996 | for (i = arg_s = 0; i < g_func_pp.argc; i++) { |
997 | if (g_func_pp.arg[i].reg != NULL) |
998 | continue; |
999 | if (arg_s == arg_i) |
1000 | break; |
1001 | arg_s++; |
1002 | } |
1003 | if (i == g_func_pp.argc) |
1004 | ferr(po, "arg %d not in prototype?\n", arg_i); |
850c9265 |
1005 | if (is_lea) |
1006 | ferr(po, "lea to arg?\n"); |
1007 | |
91977a1c |
1008 | snprintf(buf, buf_size, "%sa%d", is_src ? "(u32)" : "", i + 1); |
1009 | } |
1010 | else { |
1bafb621 |
1011 | if (g_stack_fsz == 0) |
1012 | ferr(po, "stack var access without stackframe\n"); |
850c9265 |
1013 | |
39b168b8 |
1014 | sf_ofs = g_stack_fsz + offset; |
850c9265 |
1015 | if (sf_ofs < 0) |
39b168b8 |
1016 | ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz); |
850c9265 |
1017 | |
1018 | if (is_lea) |
33c35af6 |
1019 | prefix = "(u32)&"; |
850c9265 |
1020 | |
1021 | switch (lmod) |
1022 | { |
1023 | case OPLM_BYTE: |
1024 | snprintf(buf, buf_size, "%ssf.b[%d]", prefix, sf_ofs); |
1025 | break; |
1026 | case OPLM_WORD: |
1027 | snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2); |
1028 | break; |
1029 | case OPLM_DWORD: |
1030 | snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4); |
1031 | break; |
1032 | default: |
1033 | ferr(po, "bp_stack bad lmod: %d\n", lmod); |
1034 | } |
91977a1c |
1035 | } |
1036 | } |
c36e914d |
1037 | |
91977a1c |
1038 | static char *out_src_opr(char *buf, size_t buf_size, |
850c9265 |
1039 | struct parsed_op *po, struct parsed_opr *popr, int is_lea) |
91977a1c |
1040 | { |
850c9265 |
1041 | char tmp1[256], tmp2[256]; |
1042 | char expr[256]; |
1043 | int ret; |
1044 | |
91977a1c |
1045 | switch (popr->type) { |
1046 | case OPT_REG: |
850c9265 |
1047 | if (is_lea) |
1048 | ferr(po, "lea from reg?\n"); |
1049 | |
91977a1c |
1050 | switch (popr->lmod) { |
1051 | case OPLM_DWORD: |
1052 | snprintf(buf, buf_size, "%s", opr_reg_p(po, popr)); |
1053 | break; |
850c9265 |
1054 | case OPLM_WORD: |
1055 | snprintf(buf, buf_size, "(u16)%s", opr_reg_p(po, popr)); |
1056 | break; |
1057 | case OPLM_BYTE: |
5101a5f9 |
1058 | if (popr->name[1] == 'h') // XXX.. |
1059 | snprintf(buf, buf_size, "(u8)(%s >> 8)", opr_reg_p(po, popr)); |
1060 | else |
1061 | snprintf(buf, buf_size, "(u8)%s", opr_reg_p(po, popr)); |
850c9265 |
1062 | break; |
91977a1c |
1063 | default: |
1064 | ferr(po, "invalid src lmod: %d\n", popr->lmod); |
1065 | } |
1066 | break; |
850c9265 |
1067 | |
91977a1c |
1068 | case OPT_REGMEM: |
d4e3b5db |
1069 | if (parse_stack_el(popr->name)) { |
1bafb621 |
1070 | stack_frame_access(po, popr->lmod, buf, buf_size, |
d4e3b5db |
1071 | popr->name, 1, is_lea); |
91977a1c |
1072 | break; |
1073 | } |
850c9265 |
1074 | |
1075 | strcpy(expr, popr->name); |
1076 | if (strchr(expr, '[')) { |
1077 | // special case: '[' can only be left for label[reg] form |
1078 | ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2); |
1079 | if (ret != 2) |
1080 | ferr(po, "parse failure for '%s'\n", expr); |
33c35af6 |
1081 | snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2); |
850c9265 |
1082 | } |
1083 | |
1084 | // XXX: do we need more parsing? |
1085 | if (is_lea) { |
1086 | snprintf(buf, buf_size, "%s", expr); |
1087 | break; |
1088 | } |
1089 | |
d4e3b5db |
1090 | snprintf(buf, buf_size, "%s(%s)", |
1091 | lmod_cast_u_ptr(po, popr->lmod), expr); |
91977a1c |
1092 | break; |
850c9265 |
1093 | |
91977a1c |
1094 | case OPT_LABEL: |
850c9265 |
1095 | if (is_lea) |
1096 | snprintf(buf, buf_size, "(u32)&%s", popr->name); |
1097 | else |
87bf6cec |
1098 | snprintf(buf, buf_size, "(u32)%s", popr->name); |
850c9265 |
1099 | break; |
1100 | |
1101 | case OPT_OFFSET: |
1102 | if (is_lea) |
1103 | ferr(po, "lea an offset?\n"); |
1104 | snprintf(buf, buf_size, "(u32)&%s", popr->name); |
91977a1c |
1105 | break; |
850c9265 |
1106 | |
91977a1c |
1107 | case OPT_CONST: |
850c9265 |
1108 | if (is_lea) |
1109 | ferr(po, "lea from const?\n"); |
1110 | |
5101a5f9 |
1111 | printf_number(buf, buf_size, popr->val); |
91977a1c |
1112 | break; |
850c9265 |
1113 | |
91977a1c |
1114 | default: |
1115 | ferr(po, "invalid src type: %d\n", popr->type); |
1116 | } |
1117 | |
1118 | return buf; |
1119 | } |
c36e914d |
1120 | |
91977a1c |
1121 | static char *out_dst_opr(char *buf, size_t buf_size, |
1122 | struct parsed_op *po, struct parsed_opr *popr) |
1123 | { |
1124 | switch (popr->type) { |
1125 | case OPT_REG: |
1126 | switch (popr->lmod) { |
1127 | case OPLM_DWORD: |
1128 | snprintf(buf, buf_size, "%s", opr_reg_p(po, popr)); |
1129 | break; |
850c9265 |
1130 | case OPLM_WORD: |
1131 | // ugh.. |
1132 | snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr)); |
1133 | break; |
1134 | case OPLM_BYTE: |
1135 | // ugh.. |
5101a5f9 |
1136 | if (popr->name[1] == 'h') // XXX.. |
1137 | snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr)); |
1138 | else |
1139 | snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr)); |
850c9265 |
1140 | break; |
91977a1c |
1141 | default: |
1142 | ferr(po, "invalid dst lmod: %d\n", popr->lmod); |
1143 | } |
1144 | break; |
850c9265 |
1145 | |
1146 | case OPT_REGMEM: |
d4e3b5db |
1147 | if (parse_stack_el(popr->name)) { |
1bafb621 |
1148 | stack_frame_access(po, popr->lmod, buf, buf_size, |
d4e3b5db |
1149 | popr->name, 0, 0); |
850c9265 |
1150 | break; |
1151 | } |
1152 | |
1153 | return out_src_opr(buf, buf_size, po, popr, 0); |
1154 | |
bfa4a6ee |
1155 | case OPT_LABEL: |
1156 | snprintf(buf, buf_size, "%s", popr->name); |
1157 | break; |
1158 | |
91977a1c |
1159 | default: |
1160 | ferr(po, "invalid dst type: %d\n", popr->type); |
1161 | } |
1162 | |
1163 | return buf; |
1164 | } |
c36e914d |
1165 | |
69a3cdfc |
1166 | static enum parsed_flag_op split_cond(struct parsed_op *po, |
940e8e66 |
1167 | enum op_op op, int *is_inv) |
91977a1c |
1168 | { |
940e8e66 |
1169 | *is_inv = 0; |
91977a1c |
1170 | |
69a3cdfc |
1171 | switch (op) { |
1172 | case OP_JO: |
1173 | return PFO_O; |
1174 | case OP_JC: |
1175 | return PFO_C; |
1176 | case OP_JZ: |
1177 | return PFO_Z; |
1178 | case OP_JBE: |
1179 | return PFO_BE; |
1180 | case OP_JS: |
1181 | return PFO_S; |
1182 | case OP_JP: |
1183 | return PFO_P; |
1184 | case OP_JL: |
1185 | return PFO_L; |
1186 | case OP_JLE: |
1187 | return PFO_LE; |
1188 | |
91977a1c |
1189 | case OP_JNO: |
940e8e66 |
1190 | *is_inv = 1; |
69a3cdfc |
1191 | return PFO_O; |
91977a1c |
1192 | case OP_JNC: |
940e8e66 |
1193 | *is_inv = 1; |
69a3cdfc |
1194 | return PFO_C; |
91977a1c |
1195 | case OP_JNZ: |
940e8e66 |
1196 | *is_inv = 1; |
69a3cdfc |
1197 | return PFO_Z; |
1198 | case OP_JA: |
940e8e66 |
1199 | *is_inv = 1; |
69a3cdfc |
1200 | return PFO_BE; |
91977a1c |
1201 | case OP_JNS: |
940e8e66 |
1202 | *is_inv = 1; |
69a3cdfc |
1203 | return PFO_S; |
91977a1c |
1204 | case OP_JNP: |
940e8e66 |
1205 | *is_inv = 1; |
69a3cdfc |
1206 | return PFO_P; |
850c9265 |
1207 | case OP_JGE: |
940e8e66 |
1208 | *is_inv = 1; |
69a3cdfc |
1209 | return PFO_L; |
91977a1c |
1210 | case OP_JG: |
940e8e66 |
1211 | *is_inv = 1; |
69a3cdfc |
1212 | return PFO_LE; |
1213 | |
1214 | case OP_ADC: |
1215 | case OP_SBB: |
1216 | return PFO_C; |
1217 | |
91977a1c |
1218 | default: |
69a3cdfc |
1219 | ferr(po, "split_cond: bad op %d\n", op); |
1220 | return -1; |
91977a1c |
1221 | } |
1222 | } |
c36e914d |
1223 | |
91977a1c |
1224 | static void out_test_for_cc(char *buf, size_t buf_size, |
940e8e66 |
1225 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv, |
69a3cdfc |
1226 | enum opr_lenmod lmod, const char *expr) |
91977a1c |
1227 | { |
69a3cdfc |
1228 | const char *cast, *scast; |
91977a1c |
1229 | |
69a3cdfc |
1230 | cast = lmod_cast_u(po, lmod); |
1231 | scast = lmod_cast_s(po, lmod); |
1232 | |
1233 | switch (pfo) { |
1234 | case PFO_Z: |
87bf6cec |
1235 | case PFO_BE: // CF=1||ZF=1; CF=0 |
850c9265 |
1236 | snprintf(buf, buf_size, "(%s%s %s 0)", |
940e8e66 |
1237 | cast, expr, is_inv ? "!=" : "=="); |
91977a1c |
1238 | break; |
850c9265 |
1239 | |
5101a5f9 |
1240 | case PFO_S: |
1241 | case PFO_L: // SF!=OF; OF=0 |
1242 | snprintf(buf, buf_size, "(%s%s %s 0)", |
940e8e66 |
1243 | scast, expr, is_inv ? ">=" : "<"); |
5101a5f9 |
1244 | break; |
1245 | |
87bf6cec |
1246 | case PFO_LE: // ZF=1||SF!=OF; OF=0 |
69a3cdfc |
1247 | snprintf(buf, buf_size, "(%s%s %s 0)", |
940e8e66 |
1248 | scast, expr, is_inv ? ">" : "<="); |
850c9265 |
1249 | break; |
1250 | |
91977a1c |
1251 | default: |
69a3cdfc |
1252 | ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo); |
91977a1c |
1253 | } |
1254 | } |
c36e914d |
1255 | |
850c9265 |
1256 | static void out_cmp_for_cc(char *buf, size_t buf_size, |
940e8e66 |
1257 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv, |
69a3cdfc |
1258 | enum opr_lenmod lmod, const char *expr1, const char *expr2) |
850c9265 |
1259 | { |
69a3cdfc |
1260 | const char *cast, *scast; |
850c9265 |
1261 | |
69a3cdfc |
1262 | cast = lmod_cast_u(po, lmod); |
1263 | scast = lmod_cast_s(po, lmod); |
850c9265 |
1264 | |
69a3cdfc |
1265 | switch (pfo) { |
5101a5f9 |
1266 | case PFO_C: |
1267 | // note: must be unsigned compare |
1268 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
940e8e66 |
1269 | cast, expr1, is_inv ? ">=" : "<", cast, expr2); |
5101a5f9 |
1270 | break; |
1271 | |
69a3cdfc |
1272 | case PFO_Z: |
850c9265 |
1273 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
940e8e66 |
1274 | cast, expr1, is_inv ? "!=" : "==", cast, expr2); |
850c9265 |
1275 | break; |
1276 | |
5101a5f9 |
1277 | case PFO_BE: // !a |
850c9265 |
1278 | // note: must be unsigned compare |
1279 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
940e8e66 |
1280 | cast, expr1, is_inv ? ">" : "<=", cast, expr2); |
5101a5f9 |
1281 | break; |
1282 | |
1283 | // note: must be signed compare |
1284 | case PFO_S: |
1285 | snprintf(buf, buf_size, "(%s(%s - %s) %s 0)", |
940e8e66 |
1286 | scast, expr1, expr2, is_inv ? ">=" : "<"); |
850c9265 |
1287 | break; |
1288 | |
5101a5f9 |
1289 | case PFO_L: // !ge |
850c9265 |
1290 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
940e8e66 |
1291 | scast, expr1, is_inv ? ">=" : "<", scast, expr2); |
850c9265 |
1292 | break; |
1293 | |
5101a5f9 |
1294 | case PFO_LE: |
1295 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
940e8e66 |
1296 | scast, expr1, is_inv ? ">" : "<=", scast, expr2); |
5101a5f9 |
1297 | break; |
1298 | |
850c9265 |
1299 | default: |
69a3cdfc |
1300 | ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo); |
850c9265 |
1301 | } |
1302 | } |
1303 | |
69a3cdfc |
1304 | static void out_cmp_test(char *buf, size_t buf_size, |
940e8e66 |
1305 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv) |
69a3cdfc |
1306 | { |
1307 | char buf1[256], buf2[256], buf3[256]; |
1308 | |
1309 | if (po->op == OP_TEST) { |
1310 | if (IS(opr_name(po, 0), opr_name(po, 1))) { |
1311 | out_src_opr(buf3, sizeof(buf3), po, &po->operand[0], 0); |
1312 | } |
1313 | else { |
1314 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
1315 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0); |
1316 | snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2); |
1317 | } |
940e8e66 |
1318 | out_test_for_cc(buf, buf_size, po, pfo, is_inv, |
69a3cdfc |
1319 | po->operand[0].lmod, buf3); |
1320 | } |
1321 | else if (po->op == OP_CMP) { |
1322 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[0], 0); |
1323 | out_src_opr(buf3, sizeof(buf3), po, &po->operand[1], 0); |
940e8e66 |
1324 | out_cmp_for_cc(buf, buf_size, po, pfo, is_inv, |
69a3cdfc |
1325 | po->operand[0].lmod, buf2, buf3); |
1326 | } |
1327 | else |
1328 | ferr(po, "%s: unhandled op: %d\n", __func__, po->op); |
1329 | } |
1330 | |
850c9265 |
1331 | static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1, |
91977a1c |
1332 | struct parsed_opr *popr2) |
1333 | { |
1334 | if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC) |
1335 | ferr(po, "missing lmod for both operands\n"); |
1336 | |
1337 | if (popr1->lmod == OPLM_UNSPEC) |
1338 | popr1->lmod = popr2->lmod; |
1339 | else if (popr2->lmod == OPLM_UNSPEC) |
1340 | popr2->lmod = popr1->lmod; |
1341 | else if (popr1->lmod != popr2->lmod) |
1342 | ferr(po, "conflicting lmods: %d vs %d\n", popr1->lmod, popr2->lmod); |
1343 | } |
c36e914d |
1344 | |
850c9265 |
1345 | static const char *op_to_c(struct parsed_op *po) |
1346 | { |
1347 | switch (po->op) |
1348 | { |
1349 | case OP_ADD: |
5101a5f9 |
1350 | case OP_ADC: |
850c9265 |
1351 | return "+"; |
1352 | case OP_SUB: |
5101a5f9 |
1353 | case OP_SBB: |
850c9265 |
1354 | return "-"; |
1355 | case OP_AND: |
1356 | return "&"; |
1357 | case OP_OR: |
1358 | return "|"; |
1359 | case OP_XOR: |
1360 | return "^"; |
1361 | case OP_SHL: |
1362 | return "<<"; |
1363 | case OP_SHR: |
1364 | return ">>"; |
1365 | case OP_MUL: |
1366 | case OP_IMUL: |
1367 | return "*"; |
1368 | default: |
1369 | ferr(po, "op_to_c was supplied with %d\n", po->op); |
1370 | } |
1371 | } |
1372 | |
d4e3b5db |
1373 | static void set_flag_no_dup(struct parsed_op *po, enum op_flags flag, |
1374 | enum op_flags flag_check) |
1375 | { |
1376 | if (po->flags & flag) |
1377 | ferr(po, "flag %x already set\n", flag); |
1378 | if (po->flags & flag_check) |
1379 | ferr(po, "flag_check %x already set\n", flag_check); |
1380 | |
1381 | po->flags |= flag; |
1382 | } |
1383 | |
87bf6cec |
1384 | static int scan_for_pop(int i, int opcnt, const char *reg, |
d4e3b5db |
1385 | int magic, int depth, int *maxdepth, int do_flags) |
850c9265 |
1386 | { |
87bf6cec |
1387 | struct parsed_op *po; |
1388 | int ret = 0; |
1389 | |
850c9265 |
1390 | for (; i < opcnt; i++) { |
87bf6cec |
1391 | po = &ops[i]; |
1392 | if (po->cc_scratch == magic) |
1393 | break; // already checked |
1394 | po->cc_scratch = magic; |
1395 | |
1396 | if (po->flags & OPF_TAIL) |
1397 | return -1; // deadend |
1398 | |
1bafb621 |
1399 | if ((po->flags & OPF_RMD) |
1400 | || (po->op == OP_PUSH && po->argmask)) // arg push |
850c9265 |
1401 | continue; |
1402 | |
87bf6cec |
1403 | if ((po->flags & OPF_JMP) && po->op != OP_CALL) { |
1404 | if (po->bt_i < 0) { |
1405 | ferr(po, "dead branch\n"); |
1406 | return -1; |
1407 | } |
850c9265 |
1408 | |
87bf6cec |
1409 | if (po->flags & OPF_CC) { |
d4e3b5db |
1410 | ret |= scan_for_pop(po->bt_i, opcnt, reg, magic, |
1411 | depth, maxdepth, do_flags); |
87bf6cec |
1412 | if (ret < 0) |
1413 | return ret; // dead end |
1414 | } |
1415 | else { |
1416 | i = po->bt_i - 1; |
1417 | } |
1418 | continue; |
1419 | } |
1420 | |
d4e3b5db |
1421 | if ((po->op == OP_POP || po->op == OP_PUSH) |
1422 | && po->operand[0].type == OPT_REG |
87bf6cec |
1423 | && IS(po->operand[0].name, reg)) |
1424 | { |
d4e3b5db |
1425 | if (po->op == OP_PUSH) { |
1426 | depth++; |
1427 | if (depth > *maxdepth) |
1428 | *maxdepth = depth; |
1429 | if (do_flags) |
1430 | set_flag_no_dup(po, OPF_RSAVE, OPF_RMD); |
1431 | } |
1432 | else if (depth == 0) { |
1433 | if (do_flags) |
1434 | set_flag_no_dup(po, OPF_RMD, OPF_RSAVE); |
1435 | return 1; |
1436 | } |
1437 | else { |
1438 | depth--; |
1439 | if (depth < 0) // should not happen |
1440 | ferr(po, "fail with depth\n"); |
1441 | if (do_flags) |
1442 | set_flag_no_dup(po, OPF_RSAVE, OPF_RMD); |
1443 | } |
87bf6cec |
1444 | } |
850c9265 |
1445 | } |
1446 | |
87bf6cec |
1447 | return ret; |
850c9265 |
1448 | } |
1449 | |
69a3cdfc |
1450 | // scan for pop starting from 'ret' op (all paths) |
d4e3b5db |
1451 | static int scan_for_pop_ret(int i, int opcnt, const char *reg, |
1452 | int flag_set) |
850c9265 |
1453 | { |
1454 | int found = 0; |
1455 | int j; |
1456 | |
1457 | for (; i < opcnt; i++) { |
87bf6cec |
1458 | if (!(ops[i].flags & OPF_TAIL)) |
850c9265 |
1459 | continue; |
1460 | |
1461 | for (j = i - 1; j >= 0; j--) { |
69a3cdfc |
1462 | if (ops[j].flags & OPF_RMD) |
1463 | continue; |
1464 | if (ops[j].flags & OPF_JMP) |
850c9265 |
1465 | return -1; |
1466 | |
1467 | if (ops[j].op == OP_POP && ops[j].operand[0].type == OPT_REG |
1468 | && IS(ops[j].operand[0].name, reg)) |
1469 | { |
1470 | found = 1; |
d4e3b5db |
1471 | ops[j].flags |= flag_set; |
850c9265 |
1472 | break; |
1473 | } |
1474 | |
1475 | if (g_labels[j][0] != 0) |
1476 | return -1; |
1477 | } |
1478 | } |
1479 | |
1480 | return found ? 0 : -1; |
1481 | } |
1482 | |
5101a5f9 |
1483 | // is operand 'opr modified' by parsed_op 'po'? |
1484 | static int is_opr_modified(const struct parsed_opr *opr, |
69a3cdfc |
1485 | const struct parsed_op *po) |
1486 | { |
1487 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA)) |
1488 | return 0; |
1489 | |
1490 | if (opr->type == OPT_REG && po->operand[0].type == OPT_REG) { |
1491 | if (po->regmask_dst & (1 << opr->reg)) |
1492 | return 1; |
1493 | else |
1494 | return 0; |
1495 | } |
1496 | |
1497 | return IS(po->operand[0].name, opr->name); |
1498 | } |
1499 | |
5101a5f9 |
1500 | // is any operand of parsed_op 'po_test' modified by parsed_op 'po'? |
1501 | static int is_any_opr_modified(const struct parsed_op *po_test, |
1502 | const struct parsed_op *po) |
1503 | { |
1504 | int i; |
1505 | |
1506 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA)) |
1507 | return 0; |
1508 | |
1509 | if (po_test->regmask_src & po->regmask_dst) |
1510 | return 1; |
1511 | |
1512 | for (i = 0; i < po_test->operand_cnt; i++) |
1513 | if (IS(po_test->operand[i].name, po->operand[0].name)) |
1514 | return 1; |
1515 | |
1516 | return 0; |
1517 | } |
1518 | |
940e8e66 |
1519 | // scan for any po_test operand modification in range given |
5101a5f9 |
1520 | static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt) |
69a3cdfc |
1521 | { |
1522 | for (; i < opcnt; i++) { |
5101a5f9 |
1523 | if (is_any_opr_modified(po_test, &ops[i])) |
69a3cdfc |
1524 | return i; |
1525 | } |
1526 | |
1527 | return -1; |
1528 | } |
1529 | |
940e8e66 |
1530 | // scan for po_test operand[0] modification in range given |
1531 | static int scan_for_mod_opr0(struct parsed_op *po_test, |
1532 | int i, int opcnt) |
1533 | { |
1534 | for (; i < opcnt; i++) { |
1535 | if (is_opr_modified(&po_test->operand[0], &ops[i])) |
1536 | return i; |
1537 | } |
1538 | |
1539 | return -1; |
1540 | } |
1541 | |
5101a5f9 |
1542 | static int scan_for_flag_set(int i) |
69a3cdfc |
1543 | { |
1544 | for (; i >= 0; i--) { |
1545 | if (ops[i].flags & OPF_FLAGS) |
1546 | return i; |
1547 | |
1548 | if ((ops[i].flags & OPF_JMP) && !(ops[i].flags & OPF_CC)) |
1549 | return -1; |
1550 | if (g_labels[i][0] != 0) |
1551 | return -1; |
1552 | } |
1553 | |
1554 | return -1; |
1555 | } |
1556 | |
5101a5f9 |
1557 | // scan back for cdq, if anything modifies edx, fail |
1558 | static int scan_for_cdq_edx(int i) |
1559 | { |
1560 | for (; i >= 0; i--) { |
1561 | if (ops[i].op == OP_CDQ) |
1562 | return i; |
1563 | |
1564 | if (ops[i].regmask_dst & (1 << xDX)) |
1565 | return -1; |
1566 | if (g_labels[i][0] != 0) |
1567 | return -1; |
1568 | } |
1569 | |
1570 | return -1; |
1571 | } |
1572 | |
1bafb621 |
1573 | // scan for positive, constant esp adjust |
1574 | static int scan_for_esp_adjust(int i, int opcnt, int *adj) |
1575 | { |
1576 | for (; i < opcnt; i++) { |
1577 | if (ops[i].op == OP_ADD && ops[i].operand[0].reg == xSP) { |
1578 | if (ops[i].operand[1].type != OPT_CONST) |
1579 | ferr(&ops[i], "non-const esp adjust?\n"); |
1580 | *adj = ops[i].operand[1].val; |
1581 | if (*adj & 3) |
1582 | ferr(&ops[i], "unaligned esp adjust: %x\n", *adj); |
1583 | return i; |
1584 | } |
1585 | |
1586 | if ((ops[i].flags & (OPF_JMP|OPF_TAIL)) |
1587 | || ops[i].op == OP_PUSH || ops[i].op == OP_POP) |
1588 | return -1; |
1589 | if (g_labels[i][0] != 0) |
1590 | return -1; |
1591 | } |
1592 | |
1593 | return -1; |
1594 | } |
1595 | |
91977a1c |
1596 | static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt) |
1597 | { |
69a3cdfc |
1598 | struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op; |
850c9265 |
1599 | struct parsed_opr *last_arith_dst = NULL; |
91977a1c |
1600 | char buf1[256], buf2[256], buf3[256]; |
940e8e66 |
1601 | struct parsed_proto *pp, *pp_tmp; |
91977a1c |
1602 | const char *tmpname; |
940e8e66 |
1603 | enum parsed_flag_op pfo; |
69a3cdfc |
1604 | int save_arg_vars = 0; |
1605 | int cmp_result_vars = 0; |
87bf6cec |
1606 | int need_mul_var = 0; |
91977a1c |
1607 | int had_decl = 0; |
d4e3b5db |
1608 | int regmask_save = 0; |
91977a1c |
1609 | int regmask_arg = 0; |
1610 | int regmask = 0; |
940e8e66 |
1611 | int pfomask = 0; |
d4e3b5db |
1612 | int depth = 0; |
91977a1c |
1613 | int no_output; |
69a3cdfc |
1614 | int dummy; |
91977a1c |
1615 | int arg; |
1616 | int i, j; |
1617 | int reg; |
1618 | int ret; |
1619 | |
1bafb621 |
1620 | g_bp_frame = g_sp_frame = g_stack_fsz = 0; |
91977a1c |
1621 | |
1622 | ret = proto_parse(fhdr, funcn, &g_func_pp); |
1623 | if (ret) |
1624 | ferr(ops, "proto_parse failed for '%s'\n", funcn); |
1625 | |
1626 | fprintf(fout, "%s %s(", g_func_pp.ret_type, funcn); |
1627 | for (i = 0; i < g_func_pp.argc; i++) { |
1628 | if (i > 0) |
1629 | fprintf(fout, ", "); |
1630 | fprintf(fout, "%s a%d", g_func_pp.arg[i].type, i + 1); |
1631 | } |
1632 | fprintf(fout, ")\n{\n"); |
1633 | |
1634 | // pass1: |
1bafb621 |
1635 | // - handle ebp/esp frame, remove ops related to it |
91977a1c |
1636 | if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp") |
1637 | && ops[1].op == OP_MOV |
1638 | && IS(opr_name(&ops[1], 0), "ebp") |
1639 | && IS(opr_name(&ops[1], 1), "esp")) |
1640 | { |
87bf6cec |
1641 | int ecx_push = 0; |
1642 | |
91977a1c |
1643 | g_bp_frame = 1; |
69a3cdfc |
1644 | ops[0].flags |= OPF_RMD; |
1645 | ops[1].flags |= OPF_RMD; |
91977a1c |
1646 | |
1647 | if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) { |
1bafb621 |
1648 | g_stack_fsz = opr_const(&ops[2], 1); |
69a3cdfc |
1649 | ops[2].flags |= OPF_RMD; |
91977a1c |
1650 | } |
87bf6cec |
1651 | else { |
1652 | // another way msvc builds stack frame.. |
1653 | i = 2; |
1654 | while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) { |
1bafb621 |
1655 | g_stack_fsz += 4; |
87bf6cec |
1656 | ops[i].flags |= OPF_RMD; |
1657 | ecx_push++; |
1658 | i++; |
1659 | } |
1660 | } |
91977a1c |
1661 | |
1662 | i = 2; |
1663 | do { |
1664 | for (; i < opcnt; i++) |
1665 | if (ops[i].op == OP_RET) |
1666 | break; |
1667 | if (ops[i - 1].op != OP_POP || !IS(opr_name(&ops[i - 1], 0), "ebp")) |
1668 | ferr(&ops[i - 1], "'pop ebp' expected\n"); |
69a3cdfc |
1669 | ops[i - 1].flags |= OPF_RMD; |
91977a1c |
1670 | |
1bafb621 |
1671 | if (g_stack_fsz != 0) { |
91977a1c |
1672 | if (ops[i - 2].op != OP_MOV |
1673 | || !IS(opr_name(&ops[i - 2], 0), "esp") |
1674 | || !IS(opr_name(&ops[i - 2], 1), "ebp")) |
1675 | { |
1676 | ferr(&ops[i - 2], "esp restore expected\n"); |
1677 | } |
69a3cdfc |
1678 | ops[i - 2].flags |= OPF_RMD; |
87bf6cec |
1679 | |
1680 | if (ecx_push && ops[i - 3].op == OP_POP |
1681 | && IS(opr_name(&ops[i - 3], 0), "ecx")) |
1682 | { |
1683 | ferr(&ops[i - 3], "unexpected ecx pop\n"); |
1684 | } |
91977a1c |
1685 | } |
87bf6cec |
1686 | |
91977a1c |
1687 | i++; |
1688 | } while (i < opcnt); |
1689 | } |
1bafb621 |
1690 | else { |
1691 | for (i = 0; i < opcnt; i++) { |
1692 | if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL))) |
1693 | break; |
1694 | if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP |
1695 | && ops[i].operand[1].type == OPT_CONST) |
1696 | { |
1697 | g_sp_frame = 1; |
1698 | break; |
1699 | } |
1700 | } |
1701 | |
1702 | if (g_sp_frame) |
1703 | { |
1704 | g_stack_fsz = ops[i].operand[1].val; |
1705 | ops[i].flags |= OPF_RMD; |
1706 | |
1707 | i++; |
1708 | do { |
1709 | for (; i < opcnt; i++) |
1710 | if (ops[i].op == OP_RET) |
1711 | break; |
1712 | if (ops[i - 1].op != OP_ADD |
1713 | || !IS(opr_name(&ops[i - 1], 0), "esp") |
1714 | || ops[i - 1].operand[1].type != OPT_CONST |
1715 | || ops[i - 1].operand[1].val != g_stack_fsz) |
1716 | ferr(&ops[i - 1], "'add esp' expected\n"); |
1717 | ops[i - 1].flags |= OPF_RMD; |
1718 | |
1719 | i++; |
1720 | } while (i < opcnt); |
1721 | } |
1722 | } |
91977a1c |
1723 | |
1724 | // pass2: |
87bf6cec |
1725 | // - resolve all branches |
1726 | for (i = 0; i < opcnt; i++) { |
1727 | po = &ops[i]; |
1728 | po->bt_i = -1; |
1729 | |
1730 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_JMP) |
1731 | || po->op == OP_CALL || po->op == OP_RET) |
1732 | continue; |
1733 | |
1734 | for (j = 0; j < opcnt; j++) { |
1735 | if (g_labels[j][0] && IS(po->operand[0].name, g_labels[j])) { |
1736 | po->bt_i = j; |
940e8e66 |
1737 | po->lrl = g_label_refs[j]; |
1738 | g_label_refs[j] = po; |
87bf6cec |
1739 | break; |
1740 | } |
1741 | } |
1742 | |
1743 | if (po->bt_i == -1) { |
1744 | // assume tail call |
1745 | po->op = OP_CALL; |
1746 | po->flags |= OPF_TAIL; |
1747 | } |
1748 | } |
1749 | |
1750 | // pass3: |
91977a1c |
1751 | // - process calls |
1bafb621 |
1752 | for (i = 0; i < opcnt; i++) |
1753 | { |
69a3cdfc |
1754 | po = &ops[i]; |
1755 | if (po->flags & OPF_RMD) |
91977a1c |
1756 | continue; |
850c9265 |
1757 | |
d4e3b5db |
1758 | if (po->op == OP_CALL) |
69a3cdfc |
1759 | { |
1bafb621 |
1760 | pp = calloc(1, sizeof(*pp)); |
91977a1c |
1761 | my_assert_not(pp, NULL); |
1bafb621 |
1762 | tmpname = opr_name(po, 0); |
1763 | if (po->operand[0].type != OPT_LABEL) |
1764 | { |
1765 | ret = scan_for_esp_adjust(i + 1, opcnt, &j); |
1766 | if (ret < 0) |
1767 | ferr(po, "non-__cdecl indirect call unhandled yet\n"); |
1768 | j /= 4; |
1769 | if (j > ARRAY_SIZE(pp->arg)) |
1770 | ferr(po, "esp adjust too large?\n"); |
1771 | pp->ret_type = "int"; |
1772 | pp->argc = pp->argc_stack = j; |
1773 | for (arg = 0; arg < pp->argc; arg++) |
1774 | pp->arg[arg].type = "int"; |
1775 | } |
1776 | else { |
1777 | ret = proto_parse(fhdr, tmpname, pp); |
1778 | if (ret) |
1779 | ferr(po, "proto_parse failed for call '%s'\n", tmpname); |
1780 | } |
1781 | |
1782 | ret = scan_for_esp_adjust(i + 1, opcnt, &j); |
1783 | if (ret >= 0) { |
1784 | if (pp->argc_stack != j / 4) |
1785 | ferr(po, "stack tracking failed: %x %x\n", |
1786 | pp->argc_stack, j); |
1787 | ops[ret].flags |= OPF_RMD; |
1788 | } |
91977a1c |
1789 | |
39b168b8 |
1790 | // can't call functions with non-__cdecl callbacks yet |
1791 | for (arg = 0; arg < pp->argc; arg++) { |
1792 | if (pp->arg[arg].fptr != NULL) { |
1793 | pp_tmp = pp->arg[arg].fptr; |
1794 | if (pp_tmp->is_stdcall || pp_tmp->argc != pp_tmp->argc_stack) |
1795 | ferr(po, "'%s' has a non-__cdecl callback\n", tmpname); |
1796 | } |
1797 | } |
1798 | |
91977a1c |
1799 | for (arg = 0; arg < pp->argc; arg++) |
1800 | if (pp->arg[arg].reg == NULL) |
1801 | break; |
1802 | |
1bafb621 |
1803 | for (j = i; j >= 0 && arg < pp->argc; ) |
1804 | { |
1805 | if (g_labels[j][0] != 0) { |
1806 | if (j > 0 && ((ops[j - 1].flags & OPF_TAIL) |
1807 | || (ops[j - 1].flags & (OPF_JMP|OPF_CC)) == OPF_JMP)) |
1808 | { |
1809 | // follow the branch in reverse |
1810 | if (g_label_refs[j] == NULL) |
1811 | ferr(po, "no refs for '%s'?\n", g_labels[j]); |
1812 | if (g_label_refs[j]->lrl != NULL) |
1813 | ferr(po, "unhandled multiple fefs to '%s'\n", g_labels[j]); |
1814 | j = (g_label_refs[j] - ops) + 1; |
1815 | continue; |
1816 | } |
1817 | break; |
1818 | } |
1819 | j--; |
1820 | |
1821 | if (ops[j].op == OP_CALL) |
1822 | { |
940e8e66 |
1823 | pp_tmp = ops[j].datap; |
1824 | if (pp_tmp == NULL) |
1825 | ferr(po, "arg collect hit unparsed call\n"); |
1826 | if (pp_tmp->argc_stack > 0) |
1827 | ferr(po, "arg collect hit '%s' with %d stack args\n", |
1828 | opr_name(&ops[j], 0), pp_tmp->argc_stack); |
1829 | } |
1830 | else if ((ops[j].flags & OPF_TAIL) |
1831 | || (ops[j].flags & (OPF_JMP|OPF_CC)) == OPF_JMP) |
1832 | { |
1833 | break; |
69a3cdfc |
1834 | } |
1bafb621 |
1835 | else if (ops[j].op == OP_PUSH) |
1836 | { |
940e8e66 |
1837 | pp->arg[arg].datap = &ops[j]; |
1838 | ret = scan_for_mod(&ops[j], j + 1, i); |
1839 | if (ret >= 0) { |
1840 | // mark this push as one that needs operand saving |
1bafb621 |
1841 | ops[j].flags &= ~OPF_RMD; |
940e8e66 |
1842 | ops[j].argmask |= 1 << arg; |
1843 | save_arg_vars |= 1 << arg; |
1844 | } |
1845 | else |
1846 | ops[j].flags |= OPF_RMD; |
1847 | |
1848 | // next arg |
1849 | for (arg++; arg < pp->argc; arg++) |
1850 | if (pp->arg[arg].reg == NULL) |
1851 | break; |
1852 | } |
91977a1c |
1853 | } |
1854 | if (arg < pp->argc) |
69a3cdfc |
1855 | ferr(po, "arg collect failed for '%s'\n", tmpname); |
1856 | po->datap = pp; |
91977a1c |
1857 | } |
d4e3b5db |
1858 | } |
1859 | |
1860 | // pass4: |
1861 | // - find POPs for PUSHes, rm both |
1862 | // - scan for all used registers |
1863 | // - find flag set ops for their users |
1bafb621 |
1864 | // - declare indirect functions |
d4e3b5db |
1865 | for (i = 0; i < opcnt; i++) { |
1866 | po = &ops[i]; |
1867 | if (po->flags & OPF_RMD) |
1868 | continue; |
1869 | |
1870 | if (po->op == OP_PUSH |
1871 | && po->argmask == 0 && !(po->flags & OPF_RSAVE) |
1872 | && po->operand[0].type == OPT_REG) |
1873 | { |
1874 | reg = po->operand[0].reg; |
1875 | if (reg < 0) |
1876 | ferr(po, "reg not set for push?\n"); |
1877 | |
1878 | depth = 0; |
1879 | ret = scan_for_pop(i + 1, opcnt, |
1880 | po->operand[0].name, i + opcnt, 0, &depth, 0); |
1881 | if (ret == 1) { |
1882 | if (depth > 1) |
1883 | ferr(po, "too much depth: %d\n", depth); |
1884 | if (depth > 0) |
1885 | regmask_save |= 1 << reg; |
1886 | |
1887 | po->flags |= OPF_RMD; |
1888 | scan_for_pop(i + 1, opcnt, po->operand[0].name, |
1889 | i + opcnt * 2, 0, &depth, 1); |
1890 | continue; |
1891 | } |
1892 | ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0); |
1893 | if (ret == 0) { |
1894 | arg = OPF_RMD; |
1895 | if (regmask & (1 << reg)) { |
1896 | if (regmask_save & (1 << reg)) |
1897 | ferr(po, "%s already saved?\n", po->operand[0].name); |
1898 | arg = OPF_RSAVE; |
1899 | } |
1900 | po->flags |= arg; |
1901 | scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg); |
1902 | continue; |
1903 | } |
1904 | } |
1905 | |
1906 | regmask |= po->regmask_src | po->regmask_dst; |
1907 | |
1908 | if (po->flags & OPF_CC) |
1909 | { |
1910 | ret = scan_for_flag_set(i - 1); |
1911 | if (ret < 0) |
1912 | ferr(po, "unable to trace flag setter\n"); |
1913 | |
1914 | tmp_op = &ops[ret]; // flag setter |
1915 | pfo = split_cond(po, po->op, &dummy); |
1916 | pfomask = 0; |
1917 | |
1918 | // to get nicer code, we try to delay test and cmp; |
1919 | // if we can't because of operand modification, or if we |
1920 | // have math op, make it calculate flags explicitly |
1921 | if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP) { |
1922 | if (scan_for_mod(tmp_op, ret + 1, i) >= 0) |
1923 | pfomask = 1 << pfo; |
1924 | } |
1925 | else { |
1926 | if ((pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P) |
1927 | || scan_for_mod_opr0(tmp_op, ret + 1, i) >= 0) |
1928 | pfomask = 1 << pfo; |
1929 | } |
1930 | if (pfomask) { |
1931 | tmp_op->pfomask |= pfomask; |
1932 | cmp_result_vars |= pfomask; |
1933 | po->datap = tmp_op; |
1934 | } |
1935 | |
1936 | if (po->op == OP_ADC || po->op == OP_SBB) |
1937 | cmp_result_vars |= 1 << PFO_C; |
1938 | } |
87bf6cec |
1939 | else if (po->op == OP_MUL |
1940 | || (po->op == OP_IMUL && po->operand_cnt == 1)) |
1941 | { |
1942 | need_mul_var = 1; |
1943 | } |
1bafb621 |
1944 | else if (po->op == OP_CALL && po->operand[0].type != OPT_LABEL) { |
1945 | pp = po->datap; |
1946 | my_assert_not(pp, NULL); |
1947 | fprintf(fout, " %s (*icall%d)(", pp->ret_type, i); |
1948 | for (j = 0; j < pp->argc; j++) { |
1949 | if (j > 0) |
1950 | fprintf(fout, ", "); |
1951 | fprintf(fout, "%s a%d", pp->arg[j].type, j + 1); |
1952 | } |
1953 | fprintf(fout, ");\n"); |
1954 | } |
91977a1c |
1955 | } |
1956 | |
850c9265 |
1957 | // declare stack frame |
1bafb621 |
1958 | if (g_stack_fsz) |
850c9265 |
1959 | fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n", |
1bafb621 |
1960 | (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz); |
850c9265 |
1961 | |
940e8e66 |
1962 | // declare arg-registers |
91977a1c |
1963 | for (i = 0; i < g_func_pp.argc; i++) { |
1964 | if (g_func_pp.arg[i].reg != NULL) { |
1965 | reg = char_array_i(regs_r32, |
1966 | ARRAY_SIZE(regs_r32), g_func_pp.arg[i].reg); |
1967 | if (reg < 0) |
1968 | ferr(ops, "arg '%s' is not a reg?\n", g_func_pp.arg[i].reg); |
1969 | |
1970 | regmask_arg |= 1 << reg; |
1971 | fprintf(fout, " u32 %s = (u32)a%d;\n", |
850c9265 |
1972 | g_func_pp.arg[i].reg, i + 1); |
91977a1c |
1973 | had_decl = 1; |
1974 | } |
1975 | } |
1976 | |
940e8e66 |
1977 | // declare other regs - special case for eax |
91977a1c |
1978 | if (!((regmask | regmask_arg) & 1) && !IS(g_func_pp.ret_type, "void")) { |
1979 | fprintf(fout, " u32 eax = 0;\n"); |
1980 | had_decl = 1; |
1981 | } |
1982 | |
1983 | regmask &= ~regmask_arg; |
d4e3b5db |
1984 | regmask &= ~(1 << xSP); |
91977a1c |
1985 | if (g_bp_frame) |
1986 | regmask &= ~(1 << xBP); |
1987 | if (regmask) { |
1988 | for (reg = 0; reg < 8; reg++) { |
1989 | if (regmask & (1 << reg)) { |
1990 | fprintf(fout, " u32 %s;\n", regs_r32[reg]); |
1991 | had_decl = 1; |
1992 | } |
1993 | } |
1994 | } |
1995 | |
d4e3b5db |
1996 | if (regmask_save) { |
1997 | for (reg = 0; reg < 8; reg++) { |
1998 | if (regmask_save & (1 << reg)) { |
1999 | fprintf(fout, " u32 s_%s;\n", regs_r32[reg]); |
2000 | had_decl = 1; |
2001 | } |
2002 | } |
2003 | } |
2004 | |
69a3cdfc |
2005 | if (save_arg_vars) { |
2006 | for (reg = 0; reg < 32; reg++) { |
2007 | if (save_arg_vars & (1 << reg)) { |
2008 | fprintf(fout, " u32 s_a%d;\n", reg + 1); |
2009 | had_decl = 1; |
2010 | } |
2011 | } |
2012 | } |
2013 | |
2014 | if (cmp_result_vars) { |
2015 | for (i = 0; i < 8; i++) { |
2016 | if (cmp_result_vars & (1 << i)) { |
2017 | fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]); |
2018 | had_decl = 1; |
2019 | } |
2020 | } |
2021 | } |
2022 | |
87bf6cec |
2023 | if (need_mul_var) { |
2024 | fprintf(fout, " u64 mul_tmp;\n"); |
2025 | had_decl = 1; |
2026 | } |
2027 | |
91977a1c |
2028 | if (had_decl) |
2029 | fprintf(fout, "\n"); |
2030 | |
2031 | // output ops |
69a3cdfc |
2032 | for (i = 0; i < opcnt; i++) |
2033 | { |
d4e3b5db |
2034 | if (g_labels[i][0] != 0 && g_label_refs[i] != NULL) |
91977a1c |
2035 | fprintf(fout, "\n%s:\n", g_labels[i]); |
2036 | |
69a3cdfc |
2037 | po = &ops[i]; |
2038 | if (po->flags & OPF_RMD) |
91977a1c |
2039 | continue; |
2040 | |
2041 | no_output = 0; |
2042 | |
91977a1c |
2043 | #define assert_operand_cnt(n_) \ |
850c9265 |
2044 | if (po->operand_cnt != n_) \ |
2045 | ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_) |
2046 | |
69a3cdfc |
2047 | // conditional/flag using op? |
2048 | if (po->flags & OPF_CC) |
850c9265 |
2049 | { |
940e8e66 |
2050 | int is_delayed = 0; |
2051 | int is_inv = 0; |
69a3cdfc |
2052 | |
940e8e66 |
2053 | pfo = split_cond(po, po->op, &is_inv); |
850c9265 |
2054 | |
69a3cdfc |
2055 | // we go through all this trouble to avoid using parsed_flag_op, |
2056 | // which makes generated code much nicer |
2057 | if (delayed_flag_op != NULL) |
850c9265 |
2058 | { |
940e8e66 |
2059 | out_cmp_test(buf1, sizeof(buf1), delayed_flag_op, pfo, is_inv); |
2060 | is_delayed = 1; |
91977a1c |
2061 | } |
850c9265 |
2062 | else if (last_arith_dst != NULL |
69a3cdfc |
2063 | && (pfo == PFO_Z || pfo == PFO_S || pfo == PFO_P)) |
850c9265 |
2064 | { |
2065 | out_src_opr(buf3, sizeof(buf3), po, last_arith_dst, 0); |
940e8e66 |
2066 | out_test_for_cc(buf1, sizeof(buf1), po, pfo, is_inv, |
850c9265 |
2067 | last_arith_dst->lmod, buf3); |
940e8e66 |
2068 | is_delayed = 1; |
850c9265 |
2069 | } |
69a3cdfc |
2070 | else if (po->datap != NULL) { |
2071 | // use preprocessed results |
2072 | tmp_op = po->datap; |
2073 | if (!tmp_op || !(tmp_op->pfomask & (1 << pfo))) |
2074 | ferr(po, "not prepared for pfo %d\n", pfo); |
2075 | |
940e8e66 |
2076 | // note: is_inv was not yet applied |
69a3cdfc |
2077 | snprintf(buf1, sizeof(buf1), "(%scond_%s)", |
940e8e66 |
2078 | is_inv ? "!" : "", parsed_flag_op_names[pfo]); |
69a3cdfc |
2079 | } |
2080 | else { |
2081 | ferr(po, "all methods of finding comparison failed\n"); |
2082 | } |
850c9265 |
2083 | |
69a3cdfc |
2084 | if (po->flags & OPF_JMP) { |
850c9265 |
2085 | fprintf(fout, " if %s\n", buf1); |
2086 | } |
5101a5f9 |
2087 | else if (po->op == OP_ADC || po->op == OP_SBB) { |
940e8e66 |
2088 | if (is_delayed) |
2089 | fprintf(fout, " cond_%s = %s;\n", |
2090 | parsed_flag_op_names[pfo], buf1); |
850c9265 |
2091 | } |
5101a5f9 |
2092 | else if (po->flags & OPF_DATA) { // SETcc |
850c9265 |
2093 | out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]); |
2094 | fprintf(fout, " %s = %s;", buf2, buf1); |
91977a1c |
2095 | } |
69a3cdfc |
2096 | else { |
2097 | ferr(po, "unhandled conditional op\n"); |
2098 | } |
91977a1c |
2099 | } |
2100 | |
940e8e66 |
2101 | pfomask = po->pfomask; |
2102 | |
850c9265 |
2103 | switch (po->op) |
91977a1c |
2104 | { |
2105 | case OP_MOV: |
2106 | assert_operand_cnt(2); |
850c9265 |
2107 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
2108 | fprintf(fout, " %s = %s;", |
2109 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2110 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2111 | break; |
2112 | |
2113 | case OP_LEA: |
2114 | assert_operand_cnt(2); |
87bf6cec |
2115 | po->operand[1].lmod = OPLM_DWORD; // always |
850c9265 |
2116 | fprintf(fout, " %s = %s;", |
2117 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2118 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 1)); |
2119 | break; |
2120 | |
2121 | case OP_MOVZX: |
2122 | assert_operand_cnt(2); |
91977a1c |
2123 | fprintf(fout, " %s = %s;", |
850c9265 |
2124 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2125 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2126 | break; |
2127 | |
2128 | case OP_MOVSX: |
2129 | assert_operand_cnt(2); |
2130 | switch (po->operand[1].lmod) { |
2131 | case OPLM_BYTE: |
2132 | strcpy(buf3, "(s8)"); |
2133 | break; |
2134 | case OPLM_WORD: |
2135 | strcpy(buf3, "(s16)"); |
2136 | break; |
2137 | default: |
2138 | ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod); |
2139 | } |
2140 | fprintf(fout, " %s = %s%s;", |
2141 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2142 | buf3, |
2143 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2144 | break; |
2145 | |
2146 | case OP_NOT: |
2147 | assert_operand_cnt(1); |
2148 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
2149 | fprintf(fout, " %s = ~%s;", buf1, buf1); |
2150 | break; |
2151 | |
5101a5f9 |
2152 | case OP_CDQ: |
2153 | assert_operand_cnt(2); |
2154 | fprintf(fout, " %s = (s32)%s >> 31;", |
2155 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2156 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2157 | strcpy(g_comment, "cdq"); |
2158 | break; |
2159 | |
33c35af6 |
2160 | case OP_STOS: |
2161 | // assumes DF=0 |
2162 | assert_operand_cnt(3); |
2163 | if (po->flags & OPF_REP) { |
2164 | fprintf(fout, " for (; ecx != 0; ecx--, edi += %d)\n", |
2165 | lmod_bytes(po, po->operand[0].lmod)); |
d4e3b5db |
2166 | fprintf(fout, " %sedi = eax;", |
2167 | lmod_cast_u_ptr(po, po->operand[0].lmod)); |
33c35af6 |
2168 | strcpy(g_comment, "rep stos"); |
2169 | } |
2170 | else { |
d4e3b5db |
2171 | fprintf(fout, " %sedi = eax; edi += %d;", |
2172 | lmod_cast_u_ptr(po, po->operand[0].lmod), |
33c35af6 |
2173 | lmod_bytes(po, po->operand[0].lmod)); |
2174 | strcpy(g_comment, "stos"); |
2175 | } |
2176 | break; |
2177 | |
d4e3b5db |
2178 | case OP_MOVS: |
2179 | // assumes DF=0 |
2180 | assert_operand_cnt(3); |
2181 | j = lmod_bytes(po, po->operand[0].lmod); |
2182 | strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod)); |
2183 | if (po->flags & OPF_REP) { |
2184 | fprintf(fout, |
2185 | " for (; ecx != 0; ecx--, edi += %d, esi += %d)\n", |
2186 | j, j); |
2187 | fprintf(fout, |
2188 | " %sedi = %sesi;", buf1, buf1); |
2189 | strcpy(g_comment, "rep movs"); |
2190 | } |
2191 | else { |
2192 | fprintf(fout, " %sedi = %sesi; edi += %d; esi += %d;", |
2193 | buf1, buf1, j, j); |
2194 | strcpy(g_comment, "movs"); |
2195 | } |
2196 | break; |
2197 | |
850c9265 |
2198 | // arithmetic w/flags |
2199 | case OP_ADD: |
2200 | case OP_SUB: |
2201 | case OP_AND: |
2202 | case OP_OR: |
5101a5f9 |
2203 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
2204 | // fallthrough |
850c9265 |
2205 | case OP_SHL: |
2206 | case OP_SHR: |
2207 | dualop_arith: |
2208 | assert_operand_cnt(2); |
850c9265 |
2209 | fprintf(fout, " %s %s= %s;", |
2210 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2211 | op_to_c(po), |
2212 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2213 | last_arith_dst = &po->operand[0]; |
69a3cdfc |
2214 | delayed_flag_op = NULL; |
850c9265 |
2215 | break; |
2216 | |
d4e3b5db |
2217 | case OP_SAR: |
2218 | assert_operand_cnt(2); |
2219 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
2220 | fprintf(fout, " %s = %s%s >> %s;", buf1, |
2221 | lmod_cast_s(po, po->operand[0].lmod), buf1, |
2222 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2223 | last_arith_dst = &po->operand[0]; |
2224 | delayed_flag_op = NULL; |
2225 | break; |
2226 | |
2227 | case OP_ROL: |
2228 | case OP_ROR: |
2229 | assert_operand_cnt(2); |
2230 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
2231 | if (po->operand[1].type == OPT_CONST) { |
2232 | j = po->operand[1].val; |
2233 | j %= lmod_bytes(po, po->operand[0].lmod) * 8; |
2234 | fprintf(fout, po->op == OP_ROL ? |
2235 | " %s = (%s << %d) | (%s >> %d);" : |
2236 | " %s = (%s >> %d) | (%s << %d);", |
2237 | buf1, buf1, j, buf1, |
2238 | lmod_bytes(po, po->operand[0].lmod) * 8 - j); |
2239 | } |
2240 | else |
2241 | ferr(po, "TODO\n"); |
2242 | last_arith_dst = &po->operand[0]; |
2243 | delayed_flag_op = NULL; |
2244 | break; |
2245 | |
5101a5f9 |
2246 | case OP_XOR: |
850c9265 |
2247 | assert_operand_cnt(2); |
2248 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
5101a5f9 |
2249 | if (IS(opr_name(po, 0), opr_name(po, 1))) { |
2250 | // special case for XOR |
2251 | fprintf(fout, " %s = 0;", |
2252 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0])); |
2253 | last_arith_dst = &po->operand[0]; |
2254 | delayed_flag_op = NULL; |
850c9265 |
2255 | break; |
850c9265 |
2256 | } |
5101a5f9 |
2257 | goto dualop_arith; |
2258 | |
5101a5f9 |
2259 | case OP_ADC: |
850c9265 |
2260 | case OP_SBB: |
5101a5f9 |
2261 | assert_operand_cnt(2); |
2262 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
2263 | fprintf(fout, " %s %s= %s + cond_c;", |
2264 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2265 | op_to_c(po), |
2266 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2267 | last_arith_dst = &po->operand[0]; |
2268 | delayed_flag_op = NULL; |
850c9265 |
2269 | break; |
2270 | |
2271 | case OP_INC: |
2272 | case OP_DEC: |
2273 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
1bafb621 |
2274 | if (po->operand[0].type == OPT_REG) { |
2275 | strcpy(buf2, po->op == OP_INC ? "++" : "--"); |
2276 | fprintf(fout, " %s%s;", buf1, buf2); |
2277 | } |
2278 | else { |
2279 | strcpy(buf2, po->op == OP_INC ? "+" : "-"); |
2280 | fprintf(fout, " %s %s= 1;", buf1, buf2); |
2281 | } |
5101a5f9 |
2282 | last_arith_dst = &po->operand[0]; |
2283 | delayed_flag_op = NULL; |
2284 | break; |
2285 | |
2286 | case OP_NEG: |
2287 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
2288 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[0], 0); |
2289 | fprintf(fout, " %s = -%s%s;", buf1, |
2290 | lmod_cast_s(po, po->operand[0].lmod), buf2); |
850c9265 |
2291 | last_arith_dst = &po->operand[0]; |
69a3cdfc |
2292 | delayed_flag_op = NULL; |
940e8e66 |
2293 | if (pfomask & (1 << PFO_C)) { |
2294 | fprintf(fout, "\n cond_c = (%s != 0);", buf1); |
2295 | pfomask &= ~(1 << PFO_C); |
2296 | } |
850c9265 |
2297 | break; |
2298 | |
2299 | case OP_IMUL: |
2300 | if (po->operand_cnt == 2) |
2301 | goto dualop_arith; |
87bf6cec |
2302 | if (po->operand_cnt == 3) |
2303 | ferr(po, "TODO imul3\n"); |
2304 | // fallthrough |
2305 | case OP_MUL: |
2306 | assert_operand_cnt(1); |
2307 | strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)"); |
2308 | fprintf(fout, " mul_tmp = %seax * %s%s;\n", buf1, buf1, |
2309 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[0], 0)); |
2310 | fprintf(fout, " edx = mul_tmp >> 32;\n"); |
2311 | fprintf(fout, " eax = mul_tmp;"); |
2312 | last_arith_dst = NULL; |
69a3cdfc |
2313 | delayed_flag_op = NULL; |
91977a1c |
2314 | break; |
2315 | |
5101a5f9 |
2316 | case OP_DIV: |
2317 | case OP_IDIV: |
2318 | assert_operand_cnt(1); |
2319 | if (po->operand[0].lmod != OPLM_DWORD) |
2320 | ferr(po, "unhandled lmod %d\n", po->operand[0].lmod); |
2321 | |
2322 | // 32bit division is common, look for it |
2323 | if (scan_for_cdq_edx(i - 1) >= 0) { |
2324 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
2325 | strcpy(buf2, lmod_cast(po, po->operand[0].lmod, |
2326 | po->op == OP_IDIV)); |
2327 | fprintf(fout, " edx = %seax %% %s%s;\n", buf2, buf2, buf1); |
2328 | fprintf(fout, " eax = %seax / %s%s;", buf2, buf2, buf1); |
2329 | } |
2330 | else |
2331 | ferr(po, "TODO 64bit divident\n"); |
87bf6cec |
2332 | last_arith_dst = NULL; |
2333 | delayed_flag_op = NULL; |
5101a5f9 |
2334 | break; |
2335 | |
91977a1c |
2336 | case OP_TEST: |
2337 | case OP_CMP: |
850c9265 |
2338 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
940e8e66 |
2339 | if (pfomask != 0) { |
69a3cdfc |
2340 | for (j = 0; j < 8; j++) { |
940e8e66 |
2341 | if (pfomask & (1 << j)) { |
69a3cdfc |
2342 | out_cmp_test(buf1, sizeof(buf1), po, j, 0); |
2343 | fprintf(fout, " cond_%s = %s;", |
2344 | parsed_flag_op_names[j], buf1); |
2345 | } |
2346 | } |
940e8e66 |
2347 | pfomask = 0; |
69a3cdfc |
2348 | } |
2349 | else |
2350 | no_output = 1; |
2351 | delayed_flag_op = po; |
91977a1c |
2352 | break; |
2353 | |
69a3cdfc |
2354 | // note: we reuse OP_Jcc for SETcc, only flags differ |
91977a1c |
2355 | case OP_JO ... OP_JG: |
5101a5f9 |
2356 | if (po->flags & OPF_JMP) |
850c9265 |
2357 | fprintf(fout, " goto %s;", po->operand[0].name); |
5101a5f9 |
2358 | // else SETcc - should already be handled |
850c9265 |
2359 | break; |
2360 | |
2361 | case OP_JMP: |
87bf6cec |
2362 | assert_operand_cnt(1); |
2363 | if (po->operand[0].type != OPT_LABEL) |
2364 | ferr(po, "unhandled call type\n"); |
2365 | |
850c9265 |
2366 | fprintf(fout, " goto %s;", po->operand[0].name); |
91977a1c |
2367 | break; |
2368 | |
2369 | case OP_CALL: |
5101a5f9 |
2370 | assert_operand_cnt(1); |
850c9265 |
2371 | pp = po->datap; |
91977a1c |
2372 | if (pp == NULL) |
850c9265 |
2373 | ferr(po, "NULL pp\n"); |
91977a1c |
2374 | |
1bafb621 |
2375 | if (po->operand[0].type != OPT_LABEL) |
2376 | fprintf(fout, " icall%d = (void *)%s;\n", i, |
2377 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0)); |
2378 | |
91977a1c |
2379 | fprintf(fout, " "); |
2380 | if (!IS(pp->ret_type, "void")) { |
87bf6cec |
2381 | if (po->flags & OPF_TAIL) |
2382 | fprintf(fout, "return "); |
2383 | else |
2384 | fprintf(fout, "eax = "); |
91977a1c |
2385 | if (strchr(pp->ret_type, '*')) |
2386 | fprintf(fout, "(u32)"); |
2387 | } |
87bf6cec |
2388 | |
39b168b8 |
2389 | if (po->operand[0].type != OPT_LABEL) { |
1bafb621 |
2390 | fprintf(fout, "icall%d(", i); |
39b168b8 |
2391 | } |
2392 | else { |
2393 | if (pp->name[0] == 0) |
2394 | ferr(po, "missing pp->name\n"); |
2395 | fprintf(fout, "%s(", pp->name); |
2396 | } |
2397 | |
91977a1c |
2398 | for (arg = 0; arg < pp->argc; arg++) { |
2399 | if (arg > 0) |
2400 | fprintf(fout, ", "); |
87bf6cec |
2401 | |
2402 | if (strchr(pp->arg[arg].type, '*')) |
2403 | fprintf(fout, "(%s)", pp->arg[arg].type); |
2404 | |
91977a1c |
2405 | if (pp->arg[arg].reg != NULL) { |
850c9265 |
2406 | fprintf(fout, "%s", pp->arg[arg].reg); |
91977a1c |
2407 | continue; |
2408 | } |
2409 | |
2410 | // stack arg |
2411 | tmp_op = pp->arg[arg].datap; |
2412 | if (tmp_op == NULL) |
850c9265 |
2413 | ferr(po, "parsed_op missing for arg%d\n", arg); |
940e8e66 |
2414 | if (tmp_op->argmask) { |
2415 | fprintf(fout, "s_a%d", arg + 1); |
69a3cdfc |
2416 | } |
2417 | else { |
2418 | fprintf(fout, "%s", |
2419 | out_src_opr(buf1, sizeof(buf1), |
2420 | tmp_op, &tmp_op->operand[0], 0)); |
2421 | } |
91977a1c |
2422 | } |
2423 | fprintf(fout, ");"); |
87bf6cec |
2424 | |
2425 | if (po->flags & OPF_TAIL) { |
2426 | strcpy(g_comment, "tailcall"); |
2427 | if (IS(pp->ret_type, "void")) |
2428 | fprintf(fout, "\n return;"); |
2429 | } |
2430 | delayed_flag_op = NULL; |
2431 | last_arith_dst = NULL; |
91977a1c |
2432 | break; |
2433 | |
2434 | case OP_RET: |
2435 | if (IS(g_func_pp.ret_type, "void")) |
2436 | fprintf(fout, " return;"); |
d4e3b5db |
2437 | else if (strchr(g_func_pp.ret_type, '*')) |
2438 | fprintf(fout, " return (%s)eax;", |
2439 | g_func_pp.ret_type); |
91977a1c |
2440 | else |
2441 | fprintf(fout, " return eax;"); |
2442 | break; |
2443 | |
2444 | case OP_PUSH: |
940e8e66 |
2445 | if (po->argmask) { |
69a3cdfc |
2446 | // special case - saved func arg |
d4e3b5db |
2447 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
940e8e66 |
2448 | for (j = 0; j < 32; j++) { |
d4e3b5db |
2449 | if (po->argmask & (1 << j)) |
2450 | fprintf(fout, " s_a%d = %s;", j + 1, buf1); |
940e8e66 |
2451 | } |
69a3cdfc |
2452 | break; |
2453 | } |
d4e3b5db |
2454 | else if (po->flags & OPF_RSAVE) { |
2455 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
2456 | fprintf(fout, " s_%s = %s;", buf1, buf1); |
2457 | break; |
2458 | } |
2459 | ferr(po, "stray push encountered\n"); |
91977a1c |
2460 | break; |
2461 | |
2462 | case OP_POP: |
d4e3b5db |
2463 | if (po->flags & OPF_RSAVE) { |
2464 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
2465 | fprintf(fout, " %s = s_%s;", buf1, buf1); |
2466 | break; |
2467 | } |
2468 | ferr(po, "stray pop encountered\n"); |
91977a1c |
2469 | break; |
2470 | |
33c35af6 |
2471 | case OP_NOP: |
2472 | break; |
2473 | |
91977a1c |
2474 | default: |
2475 | no_output = 1; |
69a3cdfc |
2476 | ferr(po, "unhandled op type %d, flags %x\n", |
2477 | po->op, po->flags); |
91977a1c |
2478 | break; |
2479 | } |
2480 | |
2481 | if (g_comment[0] != 0) { |
2482 | fprintf(fout, " // %s", g_comment); |
2483 | g_comment[0] = 0; |
2484 | no_output = 0; |
2485 | } |
2486 | if (!no_output) |
2487 | fprintf(fout, "\n"); |
5101a5f9 |
2488 | |
940e8e66 |
2489 | if (pfomask != 0) |
2490 | ferr(po, "missed flag calc, pfomask=%x\n", pfomask); |
2491 | |
5101a5f9 |
2492 | // see is delayed flag stuff is still valid |
2493 | if (delayed_flag_op != NULL && delayed_flag_op != po) { |
2494 | if (is_any_opr_modified(delayed_flag_op, po)) |
2495 | delayed_flag_op = NULL; |
2496 | } |
2497 | |
2498 | if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) { |
2499 | if (is_opr_modified(last_arith_dst, po)) |
2500 | last_arith_dst = NULL; |
2501 | } |
91977a1c |
2502 | } |
2503 | |
2504 | fprintf(fout, "}\n\n"); |
2505 | |
2506 | // cleanup |
2507 | for (i = 0; i < opcnt; i++) { |
2508 | if (ops[i].op == OP_CALL) { |
2509 | pp = ops[i].datap; |
2510 | if (pp) { |
2511 | proto_release(pp); |
2512 | free(pp); |
2513 | } |
2514 | } |
2515 | } |
2516 | proto_release(&g_func_pp); |
2517 | } |
c36e914d |
2518 | |
d4e3b5db |
2519 | static void set_label(int i, const char *name) |
2520 | { |
2521 | const char *p; |
2522 | int len; |
2523 | |
2524 | len = strlen(name); |
2525 | p = strchr(name, ':'); |
2526 | if (p != NULL) |
2527 | len = p - name; |
2528 | |
2529 | if (len > sizeof(g_labels[0]) - 1) |
2530 | aerr("label '%s' too long: %d\n", name, len); |
2531 | if (g_labels[i][0] != 0) |
2532 | aerr("dupe label '%s'?\n", name); |
2533 | memcpy(g_labels[i], name, len); |
2534 | g_labels[i][len] = 0; |
2535 | } |
2536 | |
bfa4a6ee |
2537 | // '=' needs special treatment.. |
2538 | static char *next_word_s(char *w, size_t wsize, char *s) |
2539 | { |
2540 | size_t i; |
2541 | |
2542 | s = sskip(s); |
2543 | |
2544 | for (i = 0; i < wsize - 1; i++) { |
2545 | if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0)) |
2546 | break; |
2547 | w[i] = s[i]; |
2548 | } |
2549 | w[i] = 0; |
2550 | |
2551 | if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=') |
2552 | printf("warning: '%s' truncated\n", w); |
2553 | |
2554 | return s + i; |
2555 | } |
2556 | |
2557 | static int cmpstringp(const void *p1, const void *p2) |
2558 | { |
2559 | return strcmp(*(char * const *)p1, *(char * const *)p2); |
2560 | } |
2561 | |
91977a1c |
2562 | int main(int argc, char *argv[]) |
2563 | { |
bfa4a6ee |
2564 | FILE *fout, *fasm, *fhdr, *frlist; |
91977a1c |
2565 | char line[256]; |
2566 | char words[16][256]; |
1bafb621 |
2567 | int ida_func_attr = 0; |
91977a1c |
2568 | int in_func = 0; |
bfa4a6ee |
2569 | int skip_func = 0; |
940e8e66 |
2570 | int skip_warned = 0; |
91977a1c |
2571 | int eq_alloc; |
bfa4a6ee |
2572 | char **rlist = NULL; |
2573 | int rlist_len = 0; |
2574 | int rlist_alloc = 0; |
2575 | int verbose = 0; |
2576 | int arg_out; |
2577 | int arg = 1; |
91977a1c |
2578 | int pi = 0; |
1bafb621 |
2579 | int i, len; |
91977a1c |
2580 | char *p; |
2581 | int wordc; |
2582 | |
bfa4a6ee |
2583 | if (argv[1] && IS(argv[1], "-v")) { |
2584 | verbose = 1; |
2585 | arg++; |
2586 | } |
2587 | |
2588 | if (argc < arg + 3) { |
2589 | printf("usage:\n%s [-v] <.c> <.asm> <hdrf> [rlist]*\n", |
91977a1c |
2590 | argv[0]); |
2591 | return 1; |
2592 | } |
2593 | |
bfa4a6ee |
2594 | arg_out = arg++; |
91977a1c |
2595 | |
bfa4a6ee |
2596 | asmfn = argv[arg++]; |
91977a1c |
2597 | fasm = fopen(asmfn, "r"); |
2598 | my_assert_not(fasm, NULL); |
2599 | |
bfa4a6ee |
2600 | hdrfn = argv[arg++]; |
2601 | fhdr = fopen(hdrfn, "r"); |
2602 | my_assert_not(fhdr, NULL); |
2603 | |
2604 | rlist_alloc = 64; |
2605 | rlist = malloc(rlist_alloc * sizeof(rlist[0])); |
2606 | my_assert_not(rlist, NULL); |
2607 | // needs special handling.. |
2608 | rlist[rlist_len++] = "__alloca_probe"; |
2609 | |
2610 | for (; arg < argc; arg++) { |
2611 | frlist = fopen(argv[arg], "r"); |
2612 | my_assert_not(frlist, NULL); |
2613 | |
2614 | while (fgets(line, sizeof(line), frlist)) { |
2615 | p = sskip(line); |
2616 | if (*p == 0 || *p == ';' || *p == '#') |
2617 | continue; |
2618 | |
2619 | p = next_word(words[0], sizeof(words[0]), p); |
2620 | if (words[0][0] == 0) |
2621 | continue; |
2622 | |
2623 | if (rlist_len >= rlist_alloc) { |
2624 | rlist_alloc = rlist_alloc * 2 + 64; |
2625 | rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0])); |
2626 | my_assert_not(rlist, NULL); |
2627 | } |
2628 | rlist[rlist_len++] = strdup(words[0]); |
2629 | } |
2630 | |
2631 | fclose(frlist); |
2632 | frlist = NULL; |
2633 | } |
2634 | |
2635 | if (rlist_len > 0) |
2636 | qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp); |
2637 | |
2638 | fout = fopen(argv[arg_out], "w"); |
91977a1c |
2639 | my_assert_not(fout, NULL); |
2640 | |
2641 | eq_alloc = 128; |
2642 | g_eqs = malloc(eq_alloc * sizeof(g_eqs[0])); |
2643 | my_assert_not(g_eqs, NULL); |
2644 | |
2645 | while (fgets(line, sizeof(line), fasm)) |
2646 | { |
2647 | asmln++; |
2648 | |
2649 | p = sskip(line); |
1bafb621 |
2650 | if (*p == 0) |
91977a1c |
2651 | continue; |
2652 | |
1bafb621 |
2653 | if (*p == ';') { |
2654 | static const char *attrs[] = { |
2655 | "bp-based frame", |
2656 | "library function", |
2657 | "static", |
2658 | "noreturn", |
2659 | "thunk", |
2660 | "fpd=", |
2661 | }; |
2662 | if (p[2] != 'A' || strncmp(p, "; Attributes:", 13) != 0) |
2663 | continue; |
2664 | |
2665 | // parse IDA's attribute-list comment |
2666 | ida_func_attr = 0; |
2667 | p = sskip(p + 13); |
2668 | // get rid of random tabs |
2669 | for (i = 0; p[i] != 0; i++) |
2670 | if (p[i] == '\t') |
2671 | p[i] = ' '; |
2672 | |
2673 | for (; *p != 0; p = sskip(p)) { |
2674 | for (i = 0; i < ARRAY_SIZE(attrs); i++) { |
2675 | if (!strncmp(p, attrs[i], strlen(attrs[i]))) { |
2676 | ida_func_attr |= 1 << i; |
2677 | p += strlen(attrs[i]); |
2678 | break; |
2679 | } |
2680 | } |
2681 | if (i == ARRAY_SIZE(attrs)) { |
2682 | anote("unparsed IDA attr: %s\n", p); |
2683 | break; |
2684 | } |
2685 | if (IS(attrs[i], "fpd=")) { |
2686 | p = next_word(words[0], sizeof(words[0]), p); |
2687 | // ignore for now.. |
2688 | } |
2689 | } |
2690 | continue; |
2691 | } |
2692 | |
91977a1c |
2693 | memset(words, 0, sizeof(words)); |
2694 | for (wordc = 0; wordc < 16; wordc++) { |
bfa4a6ee |
2695 | p = sskip(next_word_s(words[wordc], sizeof(words[0]), p)); |
91977a1c |
2696 | if (*p == 0 || *p == ';') { |
2697 | wordc++; |
2698 | break; |
2699 | } |
2700 | } |
2701 | |
2702 | if (wordc == 0) { |
2703 | // shouldn't happen |
2704 | awarn("wordc == 0?\n"); |
2705 | continue; |
2706 | } |
2707 | |
2708 | // don't care about this: |
2709 | if (words[0][0] == '.' |
2710 | || IS(words[0], "include") |
2711 | || IS(words[0], "assume") || IS(words[1], "segment") |
2712 | || IS(words[0], "align")) |
2713 | { |
2714 | continue; |
2715 | } |
2716 | |
2717 | if (IS(words[1], "proc")) { |
2718 | if (in_func) |
2719 | aerr("proc '%s' while in_func '%s'?\n", |
2720 | words[0], g_func); |
bfa4a6ee |
2721 | p = words[0]; |
1bafb621 |
2722 | if ((ida_func_attr & IDAFA_THUNK) |
2723 | || bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp)) |
bfa4a6ee |
2724 | skip_func = 1; |
91977a1c |
2725 | strcpy(g_func, words[0]); |
d4e3b5db |
2726 | set_label(0, words[0]); |
91977a1c |
2727 | in_func = 1; |
2728 | continue; |
2729 | } |
2730 | |
2731 | if (IS(words[1], "endp")) { |
2732 | if (!in_func) |
2733 | aerr("endp '%s' while not in_func?\n", words[0]); |
2734 | if (!IS(g_func, words[0])) |
2735 | aerr("endp '%s' while in_func '%s'?\n", |
2736 | words[0], g_func); |
bfa4a6ee |
2737 | |
2738 | if (in_func && !skip_func) |
2739 | gen_func(fout, fhdr, g_func, pi); |
2740 | |
91977a1c |
2741 | in_func = 0; |
940e8e66 |
2742 | skip_warned = 0; |
bfa4a6ee |
2743 | skip_func = 0; |
91977a1c |
2744 | g_func[0] = 0; |
2745 | if (pi != 0) { |
2746 | memset(&ops, 0, pi * sizeof(ops[0])); |
2747 | memset(g_labels, 0, pi * sizeof(g_labels[0])); |
940e8e66 |
2748 | memset(g_label_refs, 0, pi * sizeof(g_label_refs[0])); |
91977a1c |
2749 | pi = 0; |
2750 | } |
2751 | g_eqcnt = 0; |
1bafb621 |
2752 | ida_func_attr = 0; |
91977a1c |
2753 | continue; |
2754 | } |
2755 | |
bfa4a6ee |
2756 | p = strchr(words[0], ':'); |
2757 | if (p != NULL) { |
d4e3b5db |
2758 | set_label(pi, words[0]); |
bfa4a6ee |
2759 | continue; |
2760 | } |
2761 | |
2762 | if (!in_func || skip_func) { |
2763 | if (!skip_warned && !skip_func && g_labels[pi][0] != 0) { |
2764 | if (verbose) |
2765 | anote("skipping from '%s'\n", g_labels[pi]); |
2766 | skip_warned = 1; |
2767 | } |
2768 | g_labels[pi][0] = 0; |
2769 | continue; |
2770 | } |
2771 | |
91977a1c |
2772 | if (IS(words[1], "=")) { |
2773 | if (wordc != 5) |
2774 | aerr("unhandled equ, wc=%d\n", wordc); |
2775 | if (g_eqcnt >= eq_alloc) { |
2776 | eq_alloc *= 2; |
2777 | g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0])); |
2778 | my_assert_not(g_eqs, NULL); |
2779 | } |
2780 | |
2781 | len = strlen(words[0]); |
2782 | if (len > sizeof(g_eqs[0].name) - 1) |
2783 | aerr("equ name too long: %d\n", len); |
2784 | strcpy(g_eqs[g_eqcnt].name, words[0]); |
2785 | |
2786 | if (!IS(words[3], "ptr")) |
2787 | aerr("unhandled equ\n"); |
2788 | if (IS(words[2], "dword")) |
2789 | g_eqs[g_eqcnt].lmod = OPLM_DWORD; |
2790 | else if (IS(words[2], "word")) |
2791 | g_eqs[g_eqcnt].lmod = OPLM_WORD; |
2792 | else if (IS(words[2], "byte")) |
2793 | g_eqs[g_eqcnt].lmod = OPLM_BYTE; |
2794 | else |
2795 | aerr("bad lmod: '%s'\n", words[2]); |
2796 | |
2797 | g_eqs[g_eqcnt].offset = parse_number(words[4]); |
2798 | g_eqcnt++; |
2799 | continue; |
2800 | } |
2801 | |
2802 | if (pi >= ARRAY_SIZE(ops)) |
2803 | aerr("too many ops\n"); |
2804 | |
91977a1c |
2805 | parse_op(&ops[pi], words, wordc); |
2806 | pi++; |
91977a1c |
2807 | } |
2808 | |
2809 | fclose(fout); |
2810 | fclose(fasm); |
2811 | fclose(fhdr); |
2812 | |
2813 | return 0; |
c36e914d |
2814 | } |
91977a1c |
2815 | |
2816 | // vim:ts=2:shiftwidth=2:expandtab |