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