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 | |
06c5d854 |
15 | static const char *asmfn; |
c36e914d |
16 | static int asmln; |
06c5d854 |
17 | static FILE *g_fhdr; |
c36e914d |
18 | |
940e8e66 |
19 | #define anote(fmt, ...) \ |
20 | printf("%s:%d: note: " fmt, asmfn, asmln, ##__VA_ARGS__) |
c36e914d |
21 | #define awarn(fmt, ...) \ |
940e8e66 |
22 | printf("%s:%d: warning: " fmt, asmfn, asmln, ##__VA_ARGS__) |
c36e914d |
23 | #define aerr(fmt, ...) do { \ |
940e8e66 |
24 | printf("%s:%d: error: " fmt, asmfn, asmln, ##__VA_ARGS__); \ |
33c35af6 |
25 | fcloseall(); \ |
c36e914d |
26 | exit(1); \ |
27 | } while (0) |
28 | |
69a3cdfc |
29 | enum op_flags { |
87bf6cec |
30 | OPF_RMD = (1 << 0), /* removed or optimized out */ |
31 | OPF_DATA = (1 << 1), /* data processing - writes to dst opr */ |
32 | OPF_FLAGS = (1 << 2), /* sets flags */ |
33 | OPF_JMP = (1 << 3), /* branches, ret and call */ |
34 | OPF_CC = (1 << 4), /* uses flags */ |
35 | OPF_TAIL = (1 << 5), /* ret or tail call */ |
33c35af6 |
36 | OPF_REP = (1 << 6), /* prefixed by rep */ |
d4e3b5db |
37 | OPF_RSAVE = (1 << 7), /* push/pop is local reg save/load */ |
c36e914d |
38 | }; |
39 | |
40 | enum op_op { |
41 | OP_INVAL, |
33c35af6 |
42 | OP_NOP, |
c36e914d |
43 | OP_PUSH, |
44 | OP_POP, |
45 | OP_MOV, |
850c9265 |
46 | OP_LEA, |
47 | OP_MOVZX, |
48 | OP_MOVSX, |
49 | OP_NOT, |
5101a5f9 |
50 | OP_CDQ, |
33c35af6 |
51 | OP_STOS, |
d4e3b5db |
52 | OP_MOVS, |
c36e914d |
53 | OP_RET, |
54 | OP_ADD, |
91977a1c |
55 | OP_SUB, |
850c9265 |
56 | OP_AND, |
57 | OP_OR, |
58 | OP_XOR, |
59 | OP_SHL, |
60 | OP_SHR, |
61 | OP_SAR, |
d4e3b5db |
62 | OP_ROL, |
63 | OP_ROR, |
69a3cdfc |
64 | OP_ADC, |
850c9265 |
65 | OP_SBB, |
66 | OP_INC, |
67 | OP_DEC, |
5101a5f9 |
68 | OP_NEG, |
850c9265 |
69 | OP_MUL, |
70 | OP_IMUL, |
5101a5f9 |
71 | OP_DIV, |
72 | OP_IDIV, |
c36e914d |
73 | OP_TEST, |
74 | OP_CMP, |
75 | OP_CALL, |
76 | OP_JMP, |
77 | OP_JO, |
78 | OP_JNO, |
79 | OP_JC, |
80 | OP_JNC, |
81 | OP_JZ, |
82 | OP_JNZ, |
83 | OP_JBE, |
84 | OP_JA, |
85 | OP_JS, |
86 | OP_JNS, |
87 | OP_JP, |
88 | OP_JNP, |
89 | OP_JL, |
90 | OP_JGE, |
91 | OP_JLE, |
92 | OP_JG, |
93 | }; |
94 | |
95 | enum opr_type { |
87bf6cec |
96 | OPT_UNSPEC, |
97 | OPT_REG, |
98 | OPT_REGMEM, |
99 | OPT_LABEL, |
850c9265 |
100 | OPT_OFFSET, |
87bf6cec |
101 | OPT_CONST, |
c36e914d |
102 | }; |
103 | |
104 | enum opr_lenmod { |
91977a1c |
105 | OPLM_UNSPEC, |
106 | OPLM_BYTE, |
107 | OPLM_WORD, |
108 | OPLM_DWORD, |
c36e914d |
109 | }; |
110 | |
850c9265 |
111 | #define MAX_OPERANDS 3 |
c36e914d |
112 | |
113 | struct parsed_opr { |
91977a1c |
114 | enum opr_type type; |
115 | enum opr_lenmod lmod; |
06c5d854 |
116 | unsigned int is_ptr:1; // pointer in C |
91977a1c |
117 | int reg; |
118 | unsigned int val; |
119 | char name[256]; |
c36e914d |
120 | }; |
121 | |
122 | struct parsed_op { |
91977a1c |
123 | enum op_op op; |
124 | struct parsed_opr operand[MAX_OPERANDS]; |
69a3cdfc |
125 | unsigned int flags; |
91977a1c |
126 | int operand_cnt; |
69a3cdfc |
127 | int regmask_src; // all referensed regs |
128 | int regmask_dst; |
940e8e66 |
129 | int pfomask; // flagop: parsed_flag_op that can't be delayed |
130 | int argmask; // push: args that are altered before call |
131 | int cc_scratch; // scratch storage during analysis |
87bf6cec |
132 | int bt_i; // branch target (for branches) |
940e8e66 |
133 | struct parsed_op *lrl; // label reference list entry |
91977a1c |
134 | void *datap; |
135 | }; |
136 | |
69a3cdfc |
137 | // datap: |
69a3cdfc |
138 | // OP_CALL - ptr to parsed_proto |
139 | // (OPF_CC) - point to corresponding (OPF_FLAGS) |
140 | |
91977a1c |
141 | struct parsed_equ { |
142 | char name[64]; |
143 | enum opr_lenmod lmod; |
144 | int offset; |
c36e914d |
145 | }; |
146 | |
1bafb621 |
147 | enum ida_func_attr { |
148 | IDAFA_BP_FRAME = (1 << 0), |
149 | IDAFA_LIB_FUNC = (1 << 1), |
150 | IDAFA_STATIC = (1 << 2), |
151 | IDAFA_NORETURN = (1 << 3), |
152 | IDAFA_THUNK = (1 << 4), |
153 | IDAFA_FPD = (1 << 5), |
154 | }; |
155 | |
156 | #define MAX_OPS 4096 |
c36e914d |
157 | |
158 | static struct parsed_op ops[MAX_OPS]; |
91977a1c |
159 | static struct parsed_equ *g_eqs; |
160 | static int g_eqcnt; |
161 | static char g_labels[MAX_OPS][32]; |
940e8e66 |
162 | static struct parsed_op *g_label_refs[MAX_OPS]; |
91977a1c |
163 | static struct parsed_proto g_func_pp; |
164 | static char g_func[256]; |
165 | static char g_comment[256]; |
166 | static int g_bp_frame; |
1bafb621 |
167 | static int g_sp_frame; |
168 | static int g_stack_fsz; |
91977a1c |
169 | #define ferr(op_, fmt, ...) do { \ |
850c9265 |
170 | printf("error:%s:#%ld: '%s': " fmt, g_func, (op_) - ops, \ |
171 | dump_op(op_), ##__VA_ARGS__); \ |
33c35af6 |
172 | fcloseall(); \ |
91977a1c |
173 | exit(1); \ |
174 | } while (0) |
175 | |
176 | #define MAX_REGS 8 |
177 | |
178 | const char *regs_r32[] = { "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp" }; |
179 | const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" }; |
180 | const char *regs_r8l[] = { "al", "bl", "cl", "dl" }; |
181 | const char *regs_r8h[] = { "ah", "bh", "ch", "dh" }; |
182 | |
183 | enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP }; |
184 | |
69a3cdfc |
185 | // possible basic comparison types (without inversion) |
186 | enum parsed_flag_op { |
187 | PFO_O, // 0 OF=1 |
188 | PFO_C, // 2 CF=1 |
189 | PFO_Z, // 4 ZF=1 |
190 | PFO_BE, // 6 CF=1||ZF=1 |
191 | PFO_S, // 8 SF=1 |
192 | PFO_P, // a PF=1 |
193 | PFO_L, // c SF!=OF |
194 | PFO_LE, // e ZF=1||SF!=OF |
195 | }; |
196 | |
197 | static const char *parsed_flag_op_names[] = { |
198 | "o", "c", "z", "be", "s", "p", "l", "le" |
199 | }; |
200 | |
91977a1c |
201 | static int char_array_i(const char *array[], size_t len, const char *s) |
202 | { |
203 | int i; |
c36e914d |
204 | |
91977a1c |
205 | for (i = 0; i < len; i++) |
206 | if (IS(s, array[i])) |
207 | return i; |
c36e914d |
208 | |
91977a1c |
209 | return -1; |
210 | } |
211 | |
5101a5f9 |
212 | static void printf_number(char *buf, size_t buf_size, long number) |
91977a1c |
213 | { |
5101a5f9 |
214 | // output in C-friendly form |
215 | snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number); |
216 | } |
91977a1c |
217 | |
5101a5f9 |
218 | static int parse_reg(enum opr_lenmod *reg_lmod, const char *s) |
219 | { |
220 | int reg; |
91977a1c |
221 | |
5101a5f9 |
222 | reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s); |
223 | if (reg >= 0) { |
224 | *reg_lmod = OPLM_DWORD; |
225 | return reg; |
91977a1c |
226 | } |
5101a5f9 |
227 | reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s); |
228 | if (reg >= 0) { |
229 | *reg_lmod = OPLM_WORD; |
230 | return reg; |
231 | } |
232 | reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s); |
233 | if (reg >= 0) { |
234 | *reg_lmod = OPLM_BYTE; |
235 | return reg; |
236 | } |
237 | reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s); |
238 | if (reg >= 0) { |
239 | *reg_lmod = OPLM_BYTE; |
240 | return reg; |
850c9265 |
241 | } |
242 | |
243 | return -1; |
91977a1c |
244 | } |
245 | |
246 | static long parse_number(const char *number) |
247 | { |
248 | int len = strlen(number); |
249 | const char *p = number; |
250 | char *endp = NULL; |
251 | int neg = 0; |
252 | int bad; |
253 | long ret; |
254 | |
255 | if (*p == '-') { |
256 | neg = 1; |
257 | p++; |
258 | } |
259 | if (len > 1 && *p == '0') |
260 | p++; |
261 | if (number[len - 1] == 'h') { |
262 | ret = strtol(p, &endp, 16); |
263 | bad = (*endp != 'h'); |
264 | } |
265 | else { |
266 | ret = strtol(p, &endp, 10); |
267 | bad = (*endp != 0); |
268 | } |
269 | if (bad) |
270 | aerr("number parsing failed\n"); |
271 | if (neg) |
272 | ret = -ret; |
273 | return ret; |
274 | } |
275 | |
5101a5f9 |
276 | static int parse_indmode(char *name, int *regmask, int need_c_cvt) |
277 | { |
278 | enum opr_lenmod lmod; |
279 | char cvtbuf[256]; |
280 | char *d = cvtbuf; |
281 | char *s = name; |
282 | char w[64]; |
283 | long number; |
284 | int reg; |
285 | int c = 0; |
286 | |
287 | *d = 0; |
288 | |
289 | while (*s != 0) { |
290 | d += strlen(d); |
291 | while (my_isblank(*s)) |
292 | s++; |
293 | for (; my_issep(*s); d++, s++) |
294 | *d = *s; |
295 | while (my_isblank(*s)) |
296 | s++; |
297 | *d = 0; |
298 | |
87bf6cec |
299 | // skip 'ds:' prefix |
39b168b8 |
300 | if (IS_START(s, "ds:")) |
87bf6cec |
301 | s += 3; |
302 | |
5101a5f9 |
303 | s = next_idt(w, sizeof(w), s); |
304 | if (w[0] == 0) |
305 | break; |
306 | c++; |
307 | |
308 | reg = parse_reg(&lmod, w); |
309 | if (reg >= 0) { |
310 | *regmask |= 1 << reg; |
311 | goto pass; |
312 | } |
313 | |
314 | if ('0' <= w[0] && w[0] <= '9') { |
315 | number = parse_number(w); |
316 | printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number); |
317 | continue; |
318 | } |
319 | |
320 | // probably some label/identifier - pass |
321 | |
322 | pass: |
323 | snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w); |
324 | } |
325 | |
326 | if (need_c_cvt) |
327 | strcpy(name, cvtbuf); |
328 | |
329 | return c; |
330 | } |
331 | |
d4e3b5db |
332 | static const char *parse_stack_el(const char *name) |
333 | { |
1bafb621 |
334 | const char *p, *s; |
335 | char *endp = NULL; |
d4e3b5db |
336 | char buf[32]; |
1bafb621 |
337 | long val; |
d4e3b5db |
338 | int len; |
339 | |
39b168b8 |
340 | if (IS_START(name, "ebp+") |
d4e3b5db |
341 | && !('0' <= name[4] && name[4] <= '9')) |
342 | { |
343 | return name + 4; |
344 | } |
39b168b8 |
345 | if (!IS_START(name, "esp+")) |
d4e3b5db |
346 | return NULL; |
347 | |
348 | p = strchr(name + 4, '+'); |
349 | if (p) { |
1bafb621 |
350 | // must be a number after esp+, already converted to 0x.. |
351 | s = name + 4; |
352 | if (!('0' <= *s && *s <= '9')) { |
353 | aerr("%s nan?\n", __func__); |
d4e3b5db |
354 | return NULL; |
1bafb621 |
355 | } |
356 | if (s[0] == '0' && s[1] == 'x') |
357 | s += 2; |
358 | len = p - s; |
d4e3b5db |
359 | if (len < sizeof(buf) - 1) { |
1bafb621 |
360 | strncpy(buf, s, len); |
d4e3b5db |
361 | buf[len] = 0; |
1bafb621 |
362 | val = strtol(buf, &endp, 16); |
363 | if (val == 0 || *endp != 0) { |
364 | aerr("%s num parse fail for '%s'\n", __func__, buf); |
365 | return NULL; |
366 | } |
d4e3b5db |
367 | } |
368 | p++; |
369 | } |
370 | else |
371 | p = name + 4; |
372 | |
373 | if ('0' <= *p && *p <= '9') |
374 | return NULL; |
375 | |
376 | return p; |
377 | } |
378 | |
850c9265 |
379 | static int guess_lmod_from_name(struct parsed_opr *opr) |
380 | { |
381 | if (!strncmp(opr->name, "dword_", 6)) { |
382 | opr->lmod = OPLM_DWORD; |
383 | return 1; |
384 | } |
385 | if (!strncmp(opr->name, "word_", 5)) { |
386 | opr->lmod = OPLM_WORD; |
387 | return 1; |
388 | } |
389 | if (!strncmp(opr->name, "byte_", 5)) { |
390 | opr->lmod = OPLM_BYTE; |
391 | return 1; |
392 | } |
393 | return 0; |
394 | } |
395 | |
06c5d854 |
396 | static int guess_lmod_from_c_type(struct parsed_opr *opr, const char *c_type) |
397 | { |
398 | static const char *ptr_types[] = { |
399 | "LPCSTR", |
400 | }; |
401 | static const char *dword_types[] = { |
402 | "int", "_DWORD", "DWORD", "HANDLE", "HWND", "HMODULE", |
403 | }; |
404 | static const char *word_types[] = { |
405 | "__int16", "unsigned __int16", |
406 | }; |
407 | static const char *byte_types[] = { |
408 | "char", "__int8", "unsigned __int8", "BYTE", |
409 | }; |
410 | int i; |
411 | |
412 | if (strchr(c_type, '*')) { |
413 | opr->lmod = OPLM_DWORD; |
414 | opr->is_ptr = 1; |
415 | return 1; |
416 | } |
417 | |
418 | for (i = 0; i < ARRAY_SIZE(dword_types); i++) { |
419 | if (IS(c_type, dword_types[i])) { |
420 | opr->lmod = OPLM_DWORD; |
421 | return 1; |
422 | } |
423 | } |
424 | |
425 | for (i = 0; i < ARRAY_SIZE(ptr_types); i++) { |
426 | if (IS(c_type, ptr_types[i])) { |
427 | opr->lmod = OPLM_DWORD; |
428 | opr->is_ptr = 1; |
429 | return 1; |
430 | } |
431 | } |
432 | |
433 | for (i = 0; i < ARRAY_SIZE(word_types); i++) { |
434 | if (IS(c_type, word_types[i])) { |
435 | opr->lmod = OPLM_WORD; |
436 | return 1; |
437 | } |
438 | } |
439 | |
440 | for (i = 0; i < ARRAY_SIZE(byte_types); i++) { |
441 | if (IS(c_type, byte_types[i])) { |
442 | opr->lmod = OPLM_BYTE; |
443 | return 1; |
444 | } |
445 | } |
446 | |
447 | anote("unhandled C type '%s' for '%s'\n", c_type, opr->name); |
448 | return 0; |
449 | } |
450 | |
5101a5f9 |
451 | static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod, |
452 | int *regmask) |
453 | { |
454 | opr->type = OPT_REG; |
455 | opr->reg = reg; |
456 | opr->lmod = lmod; |
457 | *regmask |= 1 << reg; |
458 | } |
459 | |
39b168b8 |
460 | static struct parsed_equ *equ_find(struct parsed_op *po, const char *name, |
461 | int *extra_offs); |
87bf6cec |
462 | |
69a3cdfc |
463 | static int parse_operand(struct parsed_opr *opr, |
464 | int *regmask, int *regmask_indirect, |
1bafb621 |
465 | char words[16][256], int wordc, int w, unsigned int op_flags) |
c36e914d |
466 | { |
06c5d854 |
467 | struct parsed_proto pp; |
850c9265 |
468 | enum opr_lenmod tmplmod; |
850c9265 |
469 | int ret, len; |
5101a5f9 |
470 | long number; |
1bafb621 |
471 | int wordc_in; |
39b168b8 |
472 | char *tmp; |
850c9265 |
473 | int i; |
c36e914d |
474 | |
1bafb621 |
475 | if (w >= wordc) |
476 | aerr("parse_operand w %d, wordc %d\n", w, wordc); |
c36e914d |
477 | |
1bafb621 |
478 | opr->reg = xUNSPEC; |
91977a1c |
479 | |
1bafb621 |
480 | for (i = w; i < wordc; i++) { |
481 | len = strlen(words[i]); |
482 | if (words[i][len - 1] == ',') { |
483 | words[i][len - 1] = 0; |
484 | wordc = i + 1; |
485 | break; |
486 | } |
487 | } |
c36e914d |
488 | |
1bafb621 |
489 | wordc_in = wordc - w; |
c36e914d |
490 | |
1bafb621 |
491 | if ((op_flags & OPF_JMP) && wordc_in > 0 |
492 | && !('0' <= words[w][0] && words[w][0] <= '9')) |
493 | { |
494 | const char *label = NULL; |
495 | |
496 | if (wordc_in == 3 && !strncmp(words[w], "near", 4) |
497 | && IS(words[w + 1], "ptr")) |
498 | label = words[w + 2]; |
499 | else if (wordc_in == 2 && IS(words[w], "short")) |
500 | label = words[w + 1]; |
501 | else if (wordc_in == 1 |
502 | && strchr(words[w], '[') == NULL |
503 | && parse_reg(&tmplmod, words[w]) < 0) |
504 | label = words[w]; |
505 | |
506 | if (label != NULL) { |
507 | opr->type = OPT_LABEL; |
39b168b8 |
508 | if (IS_START(label, "ds:")) |
509 | label += 3; |
1bafb621 |
510 | strcpy(opr->name, label); |
511 | return wordc; |
512 | } |
513 | } |
c36e914d |
514 | |
1bafb621 |
515 | if (wordc_in >= 3) { |
516 | if (IS(words[w + 1], "ptr")) { |
517 | if (IS(words[w], "dword")) |
518 | opr->lmod = OPLM_DWORD; |
519 | else if (IS(words[w], "word")) |
520 | opr->lmod = OPLM_WORD; |
521 | else if (IS(words[w], "byte")) |
522 | opr->lmod = OPLM_BYTE; |
523 | else |
524 | aerr("type parsing failed\n"); |
525 | w += 2; |
526 | wordc_in = wordc - w; |
527 | } |
528 | } |
529 | |
39b168b8 |
530 | if (wordc_in == 2) { |
531 | if (IS(words[w], "offset")) { |
532 | opr->type = OPT_OFFSET; |
533 | strcpy(opr->name, words[w + 1]); |
534 | return wordc; |
535 | } |
536 | if (IS(words[w], "(offset")) { |
537 | char *p = strchr(words[w + 1], ')'); |
538 | if (p == NULL) |
539 | aerr("parse of bracketed offset failed\n"); |
540 | *p = 0; |
541 | opr->type = OPT_OFFSET; |
542 | strcpy(opr->name, words[w + 1]); |
543 | return wordc; |
544 | } |
1bafb621 |
545 | } |
c36e914d |
546 | |
1bafb621 |
547 | if (wordc_in != 1) |
850c9265 |
548 | aerr("parse_operand 1 word expected\n"); |
c36e914d |
549 | |
39b168b8 |
550 | tmp = words[w]; |
551 | if (IS_START(tmp, "ds:")) |
552 | tmp += 3; |
553 | strcpy(opr->name, tmp); |
c36e914d |
554 | |
850c9265 |
555 | if (words[w][0] == '[') { |
556 | opr->type = OPT_REGMEM; |
557 | ret = sscanf(words[w], "[%[^]]]", opr->name); |
558 | if (ret != 1) |
559 | aerr("[] parse failure\n"); |
87bf6cec |
560 | |
5101a5f9 |
561 | parse_indmode(opr->name, regmask_indirect, 1); |
d4e3b5db |
562 | if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name)) { |
87bf6cec |
563 | // might be an equ |
d4e3b5db |
564 | struct parsed_equ *eq = |
39b168b8 |
565 | equ_find(NULL, parse_stack_el(opr->name), &i); |
87bf6cec |
566 | if (eq) |
567 | opr->lmod = eq->lmod; |
568 | } |
850c9265 |
569 | return wordc; |
570 | } |
571 | else if (strchr(words[w], '[')) { |
572 | // label[reg] form |
573 | opr->type = OPT_REGMEM; |
574 | if (opr->lmod == OPLM_UNSPEC) |
575 | guess_lmod_from_name(opr); |
5101a5f9 |
576 | parse_indmode(strchr(words[w], '['), regmask_indirect, 0); |
850c9265 |
577 | return wordc; |
578 | } |
579 | else if (('0' <= words[w][0] && words[w][0] <= '9') |
580 | || words[w][0] == '-') |
581 | { |
5101a5f9 |
582 | number = parse_number(words[w]); |
91977a1c |
583 | opr->type = OPT_CONST; |
5101a5f9 |
584 | opr->val = number; |
585 | printf_number(opr->name, sizeof(opr->name), number); |
91977a1c |
586 | return wordc; |
850c9265 |
587 | } |
c36e914d |
588 | |
5101a5f9 |
589 | ret = parse_reg(&tmplmod, opr->name); |
590 | if (ret >= 0) { |
591 | setup_reg_opr(opr, ret, tmplmod, regmask); |
850c9265 |
592 | return wordc; |
593 | } |
594 | |
595 | // most likely var in data segment |
596 | opr->type = OPT_LABEL; |
06c5d854 |
597 | |
598 | ret = proto_parse(g_fhdr, opr->name, &pp); |
599 | if (ret == 0) { |
600 | if (pp.is_fptr) { |
601 | opr->lmod = OPLM_DWORD; |
602 | opr->is_ptr = 1; |
603 | } |
604 | else if (opr->lmod == OPLM_UNSPEC) |
605 | guess_lmod_from_c_type(opr, pp.ret_type); |
606 | } |
64c59faf |
607 | proto_release(&pp); |
06c5d854 |
608 | |
850c9265 |
609 | if (opr->lmod == OPLM_UNSPEC) |
610 | guess_lmod_from_name(opr); |
850c9265 |
611 | return wordc; |
c36e914d |
612 | } |
613 | |
33c35af6 |
614 | static const struct { |
615 | const char *name; |
616 | unsigned int flags; |
617 | } pref_table[] = { |
618 | { "rep", OPF_REP }, |
619 | }; |
620 | |
c36e914d |
621 | static const struct { |
850c9265 |
622 | const char *name; |
623 | enum op_op op; |
69a3cdfc |
624 | unsigned int minopr; |
625 | unsigned int maxopr; |
626 | unsigned int flags; |
c36e914d |
627 | } op_table[] = { |
33c35af6 |
628 | { "nop", OP_NOP, 0, 0, 0 }, |
69a3cdfc |
629 | { "push", OP_PUSH, 1, 1, 0 }, |
630 | { "pop", OP_POP, 1, 1, OPF_DATA }, |
631 | { "mov" , OP_MOV, 2, 2, OPF_DATA }, |
632 | { "lea", OP_LEA, 2, 2, OPF_DATA }, |
633 | { "movzx",OP_MOVZX, 2, 2, OPF_DATA }, |
634 | { "movsx",OP_MOVSX, 2, 2, OPF_DATA }, |
635 | { "not", OP_NOT, 1, 1, OPF_DATA }, |
5101a5f9 |
636 | { "cdq", OP_CDQ, 0, 0, OPF_DATA }, |
33c35af6 |
637 | { "stosb",OP_STOS, 0, 0, OPF_DATA }, |
638 | { "stosw",OP_STOS, 0, 0, OPF_DATA }, |
639 | { "stosd",OP_STOS, 0, 0, OPF_DATA }, |
d4e3b5db |
640 | { "movsb",OP_MOVS, 0, 0, OPF_DATA }, |
641 | { "movsw",OP_MOVS, 0, 0, OPF_DATA }, |
642 | { "movsd",OP_MOVS, 0, 0, OPF_DATA }, |
69a3cdfc |
643 | { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS }, |
644 | { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS }, |
645 | { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS }, |
646 | { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS }, |
647 | { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS }, |
648 | { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS }, |
649 | { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS }, |
650 | { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS }, |
651 | { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS }, |
d4e3b5db |
652 | { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS }, |
653 | { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS }, |
5101a5f9 |
654 | { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC }, |
69a3cdfc |
655 | { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC }, |
656 | { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS }, |
657 | { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS }, |
5101a5f9 |
658 | { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS }, |
659 | { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS }, |
69a3cdfc |
660 | { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS }, |
5101a5f9 |
661 | { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS }, |
662 | { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS }, |
69a3cdfc |
663 | { "test", OP_TEST, 2, 2, OPF_FLAGS }, |
664 | { "cmp", OP_CMP, 2, 2, OPF_FLAGS }, |
87bf6cec |
665 | { "retn", OP_RET, 0, 1, OPF_JMP|OPF_TAIL }, |
666 | { "call", OP_CALL, 1, 1, OPF_JMP|OPF_FLAGS }, |
69a3cdfc |
667 | { "jmp", OP_JMP, 1, 1, OPF_JMP }, |
668 | { "jo", OP_JO, 1, 1, OPF_JMP|OPF_CC }, // 70 OF=1 |
669 | { "jno", OP_JNO, 1, 1, OPF_JMP|OPF_CC }, // 71 OF=0 |
670 | { "jc", OP_JC, 1, 1, OPF_JMP|OPF_CC }, // 72 CF=1 |
671 | { "jb", OP_JC, 1, 1, OPF_JMP|OPF_CC }, // 72 |
672 | { "jnc", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 CF=0 |
33c35af6 |
673 | { "jnb", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 |
69a3cdfc |
674 | { "jae", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 |
675 | { "jz", OP_JZ, 1, 1, OPF_JMP|OPF_CC }, // 74 ZF=1 |
676 | { "je", OP_JZ, 1, 1, OPF_JMP|OPF_CC }, // 74 |
677 | { "jnz", OP_JNZ, 1, 1, OPF_JMP|OPF_CC }, // 75 ZF=0 |
678 | { "jne", OP_JNZ, 1, 1, OPF_JMP|OPF_CC }, // 75 |
679 | { "jbe", OP_JBE, 1, 1, OPF_JMP|OPF_CC }, // 76 CF=1 || ZF=1 |
680 | { "jna", OP_JBE, 1, 1, OPF_JMP|OPF_CC }, // 76 |
681 | { "ja", OP_JA, 1, 1, OPF_JMP|OPF_CC }, // 77 CF=0 && ZF=0 |
682 | { "jnbe", OP_JA, 1, 1, OPF_JMP|OPF_CC }, // 77 |
683 | { "js", OP_JS, 1, 1, OPF_JMP|OPF_CC }, // 78 SF=1 |
684 | { "jns", OP_JNS, 1, 1, OPF_JMP|OPF_CC }, // 79 SF=0 |
685 | { "jp", OP_JP, 1, 1, OPF_JMP|OPF_CC }, // 7a PF=1 |
686 | { "jpe", OP_JP, 1, 1, OPF_JMP|OPF_CC }, // 7a |
687 | { "jnp", OP_JNP, 1, 1, OPF_JMP|OPF_CC }, // 7b PF=0 |
688 | { "jpo", OP_JNP, 1, 1, OPF_JMP|OPF_CC }, // 7b |
689 | { "jl", OP_JL, 1, 1, OPF_JMP|OPF_CC }, // 7c SF!=OF |
690 | { "jnge", OP_JL, 1, 1, OPF_JMP|OPF_CC }, // 7c |
691 | { "jge", OP_JGE, 1, 1, OPF_JMP|OPF_CC }, // 7d SF=OF |
692 | { "jnl", OP_JGE, 1, 1, OPF_JMP|OPF_CC }, // 7d |
693 | { "jle", OP_JLE, 1, 1, OPF_JMP|OPF_CC }, // 7e ZF=1 || SF!=OF |
694 | { "jng", OP_JLE, 1, 1, OPF_JMP|OPF_CC }, // 7e |
695 | { "jg", OP_JG, 1, 1, OPF_JMP|OPF_CC }, // 7f ZF=0 && SF=OF |
696 | { "jnle", OP_JG, 1, 1, OPF_JMP|OPF_CC }, // 7f |
5101a5f9 |
697 | { "seto", OP_JO, 1, 1, OPF_DATA|OPF_CC }, |
698 | { "setno", OP_JNO, 1, 1, OPF_DATA|OPF_CC }, |
699 | { "setc", OP_JC, 1, 1, OPF_DATA|OPF_CC }, |
700 | { "setb", OP_JC, 1, 1, OPF_DATA|OPF_CC }, |
701 | { "setnc", OP_JNC, 1, 1, OPF_DATA|OPF_CC }, |
702 | { "setae", OP_JNC, 1, 1, OPF_DATA|OPF_CC }, |
703 | { "setz", OP_JZ, 1, 1, OPF_DATA|OPF_CC }, |
704 | { "sete", OP_JZ, 1, 1, OPF_DATA|OPF_CC }, |
705 | { "setnz", OP_JNZ, 1, 1, OPF_DATA|OPF_CC }, |
706 | { "setne", OP_JNZ, 1, 1, OPF_DATA|OPF_CC }, |
707 | { "setbe", OP_JBE, 1, 1, OPF_DATA|OPF_CC }, |
708 | { "setna", OP_JBE, 1, 1, OPF_DATA|OPF_CC }, |
709 | { "seta", OP_JA, 1, 1, OPF_DATA|OPF_CC }, |
710 | { "setnbe", OP_JA, 1, 1, OPF_DATA|OPF_CC }, |
711 | { "sets", OP_JS, 1, 1, OPF_DATA|OPF_CC }, |
712 | { "setns", OP_JNS, 1, 1, OPF_DATA|OPF_CC }, |
713 | { "setp", OP_JP, 1, 1, OPF_DATA|OPF_CC }, |
714 | { "setpe", OP_JP, 1, 1, OPF_DATA|OPF_CC }, |
715 | { "setnp", OP_JNP, 1, 1, OPF_DATA|OPF_CC }, |
716 | { "setpo", OP_JNP, 1, 1, OPF_DATA|OPF_CC }, |
717 | { "setl", OP_JL, 1, 1, OPF_DATA|OPF_CC }, |
718 | { "setnge", OP_JL, 1, 1, OPF_DATA|OPF_CC }, |
719 | { "setge", OP_JGE, 1, 1, OPF_DATA|OPF_CC }, |
720 | { "setnl", OP_JGE, 1, 1, OPF_DATA|OPF_CC }, |
721 | { "setle", OP_JLE, 1, 1, OPF_DATA|OPF_CC }, |
722 | { "setng", OP_JLE, 1, 1, OPF_DATA|OPF_CC }, |
723 | { "setg", OP_JG, 1, 1, OPF_DATA|OPF_CC }, |
724 | { "setnle", OP_JG, 1, 1, OPF_DATA|OPF_CC }, |
69a3cdfc |
725 | }; |
c36e914d |
726 | |
727 | static void parse_op(struct parsed_op *op, char words[16][256], int wordc) |
728 | { |
bfa4a6ee |
729 | enum opr_lenmod lmod = OPLM_UNSPEC; |
33c35af6 |
730 | int prefix_flags = 0; |
69a3cdfc |
731 | int regmask_ind; |
732 | int regmask; |
33c35af6 |
733 | int op_w = 0; |
91977a1c |
734 | int opr = 0; |
33c35af6 |
735 | int w = 0; |
91977a1c |
736 | int i; |
c36e914d |
737 | |
33c35af6 |
738 | for (i = 0; i < ARRAY_SIZE(pref_table); i++) { |
739 | if (IS(words[w], pref_table[i].name)) { |
740 | prefix_flags = pref_table[i].flags; |
741 | break; |
742 | } |
743 | } |
744 | |
745 | if (prefix_flags) { |
746 | if (wordc <= 1) |
747 | aerr("lone prefix: '%s'\n", words[0]); |
748 | w++; |
749 | } |
750 | |
751 | op_w = w; |
91977a1c |
752 | for (i = 0; i < ARRAY_SIZE(op_table); i++) { |
33c35af6 |
753 | if (IS(words[w], op_table[i].name)) |
69a3cdfc |
754 | break; |
755 | } |
c36e914d |
756 | |
69a3cdfc |
757 | if (i == ARRAY_SIZE(op_table)) |
758 | aerr("unhandled op: '%s'\n", words[0]); |
33c35af6 |
759 | w++; |
c36e914d |
760 | |
69a3cdfc |
761 | op->op = op_table[i].op; |
33c35af6 |
762 | op->flags = op_table[i].flags | prefix_flags; |
69a3cdfc |
763 | op->regmask_src = op->regmask_dst = 0; |
764 | |
765 | for (opr = 0; opr < op_table[i].minopr; opr++) { |
766 | regmask = regmask_ind = 0; |
767 | w = parse_operand(&op->operand[opr], ®mask, ®mask_ind, |
768 | words, wordc, w, op->flags); |
769 | |
770 | if (opr == 0 && (op->flags & OPF_DATA)) |
771 | op->regmask_dst = regmask; |
772 | // for now, mark dst as src too |
773 | op->regmask_src |= regmask | regmask_ind; |
774 | } |
c36e914d |
775 | |
69a3cdfc |
776 | for (; w < wordc && opr < op_table[i].maxopr; opr++) { |
777 | w = parse_operand(&op->operand[opr], |
778 | &op->regmask_src, &op->regmask_src, |
779 | words, wordc, w, op->flags); |
91977a1c |
780 | } |
c36e914d |
781 | |
91977a1c |
782 | if (w < wordc) |
783 | aerr("parse_op %s incomplete: %d/%d\n", |
784 | words[0], w, wordc); |
5101a5f9 |
785 | |
786 | // special cases |
787 | op->operand_cnt = opr; |
788 | if (!strncmp(op_table[i].name, "set", 3)) |
789 | op->operand[0].lmod = OPLM_BYTE; |
790 | |
791 | // ops with implicit argumets |
792 | switch (op->op) { |
793 | case OP_CDQ: |
794 | op->operand_cnt = 2; |
795 | setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst); |
796 | setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src); |
797 | break; |
798 | |
33c35af6 |
799 | case OP_STOS: |
800 | if (op->operand_cnt != 0) |
801 | break; |
802 | if (IS(words[op_w], "stosb")) |
803 | lmod = OPLM_BYTE; |
804 | else if (IS(words[op_w], "stosw")) |
805 | lmod = OPLM_WORD; |
806 | else if (IS(words[op_w], "stosd")) |
807 | lmod = OPLM_DWORD; |
808 | op->operand_cnt = 3; |
d4e3b5db |
809 | setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src); |
810 | setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src); |
811 | op->regmask_dst = op->regmask_src; |
33c35af6 |
812 | setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD, &op->regmask_src); |
813 | break; |
814 | |
d4e3b5db |
815 | case OP_MOVS: |
816 | if (op->operand_cnt != 0) |
817 | break; |
818 | if (IS(words[op_w], "movsb")) |
819 | lmod = OPLM_BYTE; |
820 | else if (IS(words[op_w], "movsw")) |
821 | lmod = OPLM_WORD; |
822 | else if (IS(words[op_w], "movsd")) |
823 | lmod = OPLM_DWORD; |
824 | op->operand_cnt = 3; |
825 | setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src); |
826 | setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src); |
827 | setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src); |
828 | op->regmask_dst = op->regmask_src; |
829 | break; |
830 | |
5101a5f9 |
831 | case OP_IMUL: |
832 | if (op->operand_cnt != 1) |
833 | break; |
834 | // fallthrough |
835 | case OP_MUL: |
836 | // singleop mul |
837 | op->regmask_dst = (1 << xDX) | (1 << xAX); |
838 | op->regmask_src |= (1 << xAX); |
839 | if (op->operand[0].lmod == OPLM_UNSPEC) |
840 | op->operand[0].lmod = OPLM_DWORD; |
841 | break; |
842 | |
843 | case OP_DIV: |
844 | case OP_IDIV: |
845 | // we could set up operands for edx:eax, but there is no real need to |
846 | // (see is_opr_modified()) |
847 | regmask = (1 << xDX) | (1 << xAX); |
848 | op->regmask_dst = regmask; |
849 | op->regmask_src |= regmask; |
850 | if (op->operand[0].lmod == OPLM_UNSPEC) |
851 | op->operand[0].lmod = OPLM_DWORD; |
852 | break; |
853 | |
854 | case OP_SHL: |
855 | case OP_SHR: |
856 | case OP_SAR: |
d4e3b5db |
857 | case OP_ROL: |
858 | case OP_ROR: |
5101a5f9 |
859 | if (op->operand[1].lmod == OPLM_UNSPEC) |
860 | op->operand[1].lmod = OPLM_BYTE; |
861 | break; |
862 | |
863 | default: |
864 | break; |
865 | } |
c36e914d |
866 | } |
867 | |
850c9265 |
868 | static const char *op_name(enum op_op op) |
869 | { |
870 | int i; |
871 | |
872 | for (i = 0; i < ARRAY_SIZE(op_table); i++) |
873 | if (op_table[i].op == op) |
874 | return op_table[i].name; |
875 | |
876 | return "???"; |
877 | } |
878 | |
879 | // debug |
880 | static const char *dump_op(struct parsed_op *po) |
881 | { |
882 | static char out[128]; |
883 | char *p = out; |
884 | int i; |
885 | |
886 | snprintf(out, sizeof(out), "%s", op_name(po->op)); |
887 | for (i = 0; i < po->operand_cnt; i++) { |
888 | p += strlen(p); |
889 | if (i > 0) |
890 | *p++ = ','; |
891 | snprintf(p, sizeof(out) - (p - out), |
892 | po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s", |
893 | po->operand[i].name); |
894 | } |
895 | |
896 | return out; |
897 | } |
898 | |
d4e3b5db |
899 | static const char *lmod_cast_u(struct parsed_op *po, |
900 | enum opr_lenmod lmod) |
901 | { |
902 | switch (lmod) { |
903 | case OPLM_DWORD: |
904 | return ""; |
905 | case OPLM_WORD: |
906 | return "(u16)"; |
907 | case OPLM_BYTE: |
908 | return "(u8)"; |
909 | default: |
910 | ferr(po, "invalid lmod: %d\n", lmod); |
911 | return "(_invalid_)"; |
912 | } |
913 | } |
914 | |
915 | static const char *lmod_cast_u_ptr(struct parsed_op *po, |
916 | enum opr_lenmod lmod) |
917 | { |
918 | switch (lmod) { |
919 | case OPLM_DWORD: |
920 | return "*(u32 *)"; |
921 | case OPLM_WORD: |
922 | return "*(u16 *)"; |
923 | case OPLM_BYTE: |
924 | return "*(u8 *)"; |
925 | default: |
926 | ferr(po, "invalid lmod: %d\n", lmod); |
927 | return "(_invalid_)"; |
928 | } |
929 | } |
930 | |
931 | static const char *lmod_cast_s(struct parsed_op *po, |
932 | enum opr_lenmod lmod) |
933 | { |
934 | switch (lmod) { |
935 | case OPLM_DWORD: |
936 | return "(s32)"; |
937 | case OPLM_WORD: |
938 | return "(s16)"; |
939 | case OPLM_BYTE: |
940 | return "(s8)"; |
941 | default: |
942 | ferr(po, "%s: invalid lmod: %d\n", __func__, lmod); |
943 | return "(_invalid_)"; |
944 | } |
945 | } |
946 | |
947 | static const char *lmod_cast(struct parsed_op *po, |
948 | enum opr_lenmod lmod, int is_signed) |
949 | { |
950 | return is_signed ? |
951 | lmod_cast_s(po, lmod) : |
952 | lmod_cast_u(po, lmod); |
953 | } |
954 | |
955 | static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod) |
956 | { |
957 | switch (lmod) { |
958 | case OPLM_DWORD: |
959 | return 4; |
960 | case OPLM_WORD: |
961 | return 2; |
962 | case OPLM_BYTE: |
963 | return 1; |
964 | default: |
965 | ferr(po, "%s: invalid lmod: %d\n", __func__, lmod); |
966 | return 0; |
967 | } |
968 | } |
969 | |
91977a1c |
970 | static const char *opr_name(struct parsed_op *po, int opr_num) |
c36e914d |
971 | { |
91977a1c |
972 | if (opr_num >= po->operand_cnt) |
973 | ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt); |
974 | return po->operand[opr_num].name; |
c36e914d |
975 | } |
976 | |
91977a1c |
977 | static unsigned int opr_const(struct parsed_op *po, int opr_num) |
c36e914d |
978 | { |
91977a1c |
979 | if (opr_num >= po->operand_cnt) |
980 | ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt); |
981 | if (po->operand[opr_num].type != OPT_CONST) |
982 | ferr(po, "opr %d: const expected\n", opr_num); |
983 | return po->operand[opr_num].val; |
984 | } |
c36e914d |
985 | |
91977a1c |
986 | static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr) |
987 | { |
988 | if ((unsigned int)popr->reg >= MAX_REGS) |
989 | ferr(po, "invalid reg: %d\n", popr->reg); |
990 | return regs_r32[popr->reg]; |
991 | } |
c36e914d |
992 | |
39b168b8 |
993 | static struct parsed_equ *equ_find(struct parsed_op *po, const char *name, |
994 | int *extra_offs) |
91977a1c |
995 | { |
39b168b8 |
996 | const char *p; |
997 | char *endp; |
998 | int namelen; |
850c9265 |
999 | int i; |
1000 | |
39b168b8 |
1001 | *extra_offs = 0; |
1002 | namelen = strlen(name); |
1003 | |
1004 | p = strchr(name, '+'); |
1005 | if (p != NULL) { |
1006 | namelen = p - name; |
1007 | if (namelen <= 0) |
1008 | ferr(po, "equ parse failed for '%s'\n", name); |
1009 | |
1010 | if (IS_START(p, "0x")) |
1011 | p += 2; |
1012 | *extra_offs = strtol(p, &endp, 16); |
1013 | if (*endp != 0) |
1014 | ferr(po, "equ parse failed for '%s'\n", name); |
1015 | } |
1016 | |
850c9265 |
1017 | for (i = 0; i < g_eqcnt; i++) |
39b168b8 |
1018 | if (strncmp(g_eqs[i].name, name, namelen) == 0 |
1019 | && g_eqs[i].name[namelen] == 0) |
850c9265 |
1020 | break; |
87bf6cec |
1021 | if (i >= g_eqcnt) { |
1022 | if (po != NULL) |
1023 | ferr(po, "unresolved equ name: '%s'\n", name); |
1024 | return NULL; |
1025 | } |
850c9265 |
1026 | |
1027 | return &g_eqs[i]; |
1028 | } |
1029 | |
1bafb621 |
1030 | static void stack_frame_access(struct parsed_op *po, |
1031 | enum opr_lenmod lmod, char *buf, size_t buf_size, |
1032 | const char *name, int is_src, int is_lea) |
850c9265 |
1033 | { |
1034 | const char *prefix = ""; |
91977a1c |
1035 | struct parsed_equ *eq; |
1036 | int i, arg_i, arg_s; |
d4e3b5db |
1037 | const char *bp_arg; |
1038 | int stack_ra = 0; |
39b168b8 |
1039 | int offset = 0; |
850c9265 |
1040 | int sf_ofs; |
91977a1c |
1041 | |
d4e3b5db |
1042 | bp_arg = parse_stack_el(name); |
91977a1c |
1043 | snprintf(g_comment, sizeof(g_comment), "%s", bp_arg); |
39b168b8 |
1044 | eq = equ_find(po, bp_arg, &offset); |
d4e3b5db |
1045 | if (eq == NULL) |
1046 | ferr(po, "detected but missing eq\n"); |
91977a1c |
1047 | |
39b168b8 |
1048 | offset += eq->offset; |
1049 | |
d4e3b5db |
1050 | if (!strncmp(name, "ebp", 3)) |
1051 | stack_ra = 4; |
1052 | |
39b168b8 |
1053 | if (stack_ra <= offset && offset < stack_ra + 4) |
1054 | ferr(po, "reference to ra? %d %d\n", offset, stack_ra); |
d4e3b5db |
1055 | |
39b168b8 |
1056 | if (offset > stack_ra) { |
1057 | arg_i = (offset - stack_ra - 4) / 4; |
91977a1c |
1058 | if (arg_i < 0 || arg_i >= g_func_pp.argc_stack) |
1bafb621 |
1059 | ferr(po, "offset %d (%s) doesn't map to any arg\n", |
39b168b8 |
1060 | offset, bp_arg); |
91977a1c |
1061 | |
1062 | for (i = arg_s = 0; i < g_func_pp.argc; i++) { |
1063 | if (g_func_pp.arg[i].reg != NULL) |
1064 | continue; |
1065 | if (arg_s == arg_i) |
1066 | break; |
1067 | arg_s++; |
1068 | } |
1069 | if (i == g_func_pp.argc) |
1070 | ferr(po, "arg %d not in prototype?\n", arg_i); |
850c9265 |
1071 | if (is_lea) |
1072 | ferr(po, "lea to arg?\n"); |
1073 | |
91977a1c |
1074 | snprintf(buf, buf_size, "%sa%d", is_src ? "(u32)" : "", i + 1); |
1075 | } |
1076 | else { |
1bafb621 |
1077 | if (g_stack_fsz == 0) |
1078 | ferr(po, "stack var access without stackframe\n"); |
850c9265 |
1079 | |
39b168b8 |
1080 | sf_ofs = g_stack_fsz + offset; |
850c9265 |
1081 | if (sf_ofs < 0) |
39b168b8 |
1082 | ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz); |
850c9265 |
1083 | |
1084 | if (is_lea) |
33c35af6 |
1085 | prefix = "(u32)&"; |
850c9265 |
1086 | |
1087 | switch (lmod) |
1088 | { |
1089 | case OPLM_BYTE: |
1090 | snprintf(buf, buf_size, "%ssf.b[%d]", prefix, sf_ofs); |
1091 | break; |
1092 | case OPLM_WORD: |
1093 | snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2); |
1094 | break; |
1095 | case OPLM_DWORD: |
1096 | snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4); |
1097 | break; |
1098 | default: |
1099 | ferr(po, "bp_stack bad lmod: %d\n", lmod); |
1100 | } |
91977a1c |
1101 | } |
1102 | } |
c36e914d |
1103 | |
91977a1c |
1104 | static char *out_src_opr(char *buf, size_t buf_size, |
850c9265 |
1105 | struct parsed_op *po, struct parsed_opr *popr, int is_lea) |
91977a1c |
1106 | { |
850c9265 |
1107 | char tmp1[256], tmp2[256]; |
1108 | char expr[256]; |
1109 | int ret; |
1110 | |
91977a1c |
1111 | switch (popr->type) { |
1112 | case OPT_REG: |
850c9265 |
1113 | if (is_lea) |
1114 | ferr(po, "lea from reg?\n"); |
1115 | |
91977a1c |
1116 | switch (popr->lmod) { |
1117 | case OPLM_DWORD: |
1118 | snprintf(buf, buf_size, "%s", opr_reg_p(po, popr)); |
1119 | break; |
850c9265 |
1120 | case OPLM_WORD: |
1121 | snprintf(buf, buf_size, "(u16)%s", opr_reg_p(po, popr)); |
1122 | break; |
1123 | case OPLM_BYTE: |
5101a5f9 |
1124 | if (popr->name[1] == 'h') // XXX.. |
1125 | snprintf(buf, buf_size, "(u8)(%s >> 8)", opr_reg_p(po, popr)); |
1126 | else |
1127 | snprintf(buf, buf_size, "(u8)%s", opr_reg_p(po, popr)); |
850c9265 |
1128 | break; |
91977a1c |
1129 | default: |
1130 | ferr(po, "invalid src lmod: %d\n", popr->lmod); |
1131 | } |
1132 | break; |
850c9265 |
1133 | |
91977a1c |
1134 | case OPT_REGMEM: |
d4e3b5db |
1135 | if (parse_stack_el(popr->name)) { |
1bafb621 |
1136 | stack_frame_access(po, popr->lmod, buf, buf_size, |
d4e3b5db |
1137 | popr->name, 1, is_lea); |
91977a1c |
1138 | break; |
1139 | } |
850c9265 |
1140 | |
1141 | strcpy(expr, popr->name); |
1142 | if (strchr(expr, '[')) { |
1143 | // special case: '[' can only be left for label[reg] form |
1144 | ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2); |
1145 | if (ret != 2) |
1146 | ferr(po, "parse failure for '%s'\n", expr); |
33c35af6 |
1147 | snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2); |
850c9265 |
1148 | } |
1149 | |
1150 | // XXX: do we need more parsing? |
1151 | if (is_lea) { |
1152 | snprintf(buf, buf_size, "%s", expr); |
1153 | break; |
1154 | } |
1155 | |
d4e3b5db |
1156 | snprintf(buf, buf_size, "%s(%s)", |
1157 | lmod_cast_u_ptr(po, popr->lmod), expr); |
91977a1c |
1158 | break; |
850c9265 |
1159 | |
91977a1c |
1160 | case OPT_LABEL: |
850c9265 |
1161 | if (is_lea) |
1162 | snprintf(buf, buf_size, "(u32)&%s", popr->name); |
1163 | else |
87bf6cec |
1164 | snprintf(buf, buf_size, "(u32)%s", popr->name); |
850c9265 |
1165 | break; |
1166 | |
1167 | case OPT_OFFSET: |
1168 | if (is_lea) |
1169 | ferr(po, "lea an offset?\n"); |
1170 | snprintf(buf, buf_size, "(u32)&%s", popr->name); |
91977a1c |
1171 | break; |
850c9265 |
1172 | |
91977a1c |
1173 | case OPT_CONST: |
850c9265 |
1174 | if (is_lea) |
1175 | ferr(po, "lea from const?\n"); |
1176 | |
5101a5f9 |
1177 | printf_number(buf, buf_size, popr->val); |
91977a1c |
1178 | break; |
850c9265 |
1179 | |
91977a1c |
1180 | default: |
1181 | ferr(po, "invalid src type: %d\n", popr->type); |
1182 | } |
1183 | |
1184 | return buf; |
1185 | } |
c36e914d |
1186 | |
91977a1c |
1187 | static char *out_dst_opr(char *buf, size_t buf_size, |
1188 | struct parsed_op *po, struct parsed_opr *popr) |
1189 | { |
1190 | switch (popr->type) { |
1191 | case OPT_REG: |
1192 | switch (popr->lmod) { |
1193 | case OPLM_DWORD: |
1194 | snprintf(buf, buf_size, "%s", opr_reg_p(po, popr)); |
1195 | break; |
850c9265 |
1196 | case OPLM_WORD: |
1197 | // ugh.. |
1198 | snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr)); |
1199 | break; |
1200 | case OPLM_BYTE: |
1201 | // ugh.. |
5101a5f9 |
1202 | if (popr->name[1] == 'h') // XXX.. |
1203 | snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr)); |
1204 | else |
1205 | snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr)); |
850c9265 |
1206 | break; |
91977a1c |
1207 | default: |
1208 | ferr(po, "invalid dst lmod: %d\n", popr->lmod); |
1209 | } |
1210 | break; |
850c9265 |
1211 | |
1212 | case OPT_REGMEM: |
d4e3b5db |
1213 | if (parse_stack_el(popr->name)) { |
1bafb621 |
1214 | stack_frame_access(po, popr->lmod, buf, buf_size, |
d4e3b5db |
1215 | popr->name, 0, 0); |
850c9265 |
1216 | break; |
1217 | } |
1218 | |
1219 | return out_src_opr(buf, buf_size, po, popr, 0); |
1220 | |
bfa4a6ee |
1221 | case OPT_LABEL: |
1222 | snprintf(buf, buf_size, "%s", popr->name); |
1223 | break; |
1224 | |
91977a1c |
1225 | default: |
1226 | ferr(po, "invalid dst type: %d\n", popr->type); |
1227 | } |
1228 | |
1229 | return buf; |
1230 | } |
c36e914d |
1231 | |
69a3cdfc |
1232 | static enum parsed_flag_op split_cond(struct parsed_op *po, |
940e8e66 |
1233 | enum op_op op, int *is_inv) |
91977a1c |
1234 | { |
940e8e66 |
1235 | *is_inv = 0; |
91977a1c |
1236 | |
69a3cdfc |
1237 | switch (op) { |
1238 | case OP_JO: |
1239 | return PFO_O; |
1240 | case OP_JC: |
1241 | return PFO_C; |
1242 | case OP_JZ: |
1243 | return PFO_Z; |
1244 | case OP_JBE: |
1245 | return PFO_BE; |
1246 | case OP_JS: |
1247 | return PFO_S; |
1248 | case OP_JP: |
1249 | return PFO_P; |
1250 | case OP_JL: |
1251 | return PFO_L; |
1252 | case OP_JLE: |
1253 | return PFO_LE; |
1254 | |
91977a1c |
1255 | case OP_JNO: |
940e8e66 |
1256 | *is_inv = 1; |
69a3cdfc |
1257 | return PFO_O; |
91977a1c |
1258 | case OP_JNC: |
940e8e66 |
1259 | *is_inv = 1; |
69a3cdfc |
1260 | return PFO_C; |
91977a1c |
1261 | case OP_JNZ: |
940e8e66 |
1262 | *is_inv = 1; |
69a3cdfc |
1263 | return PFO_Z; |
1264 | case OP_JA: |
940e8e66 |
1265 | *is_inv = 1; |
69a3cdfc |
1266 | return PFO_BE; |
91977a1c |
1267 | case OP_JNS: |
940e8e66 |
1268 | *is_inv = 1; |
69a3cdfc |
1269 | return PFO_S; |
91977a1c |
1270 | case OP_JNP: |
940e8e66 |
1271 | *is_inv = 1; |
69a3cdfc |
1272 | return PFO_P; |
850c9265 |
1273 | case OP_JGE: |
940e8e66 |
1274 | *is_inv = 1; |
69a3cdfc |
1275 | return PFO_L; |
91977a1c |
1276 | case OP_JG: |
940e8e66 |
1277 | *is_inv = 1; |
69a3cdfc |
1278 | return PFO_LE; |
1279 | |
1280 | case OP_ADC: |
1281 | case OP_SBB: |
1282 | return PFO_C; |
1283 | |
91977a1c |
1284 | default: |
69a3cdfc |
1285 | ferr(po, "split_cond: bad op %d\n", op); |
1286 | return -1; |
91977a1c |
1287 | } |
1288 | } |
c36e914d |
1289 | |
91977a1c |
1290 | static void out_test_for_cc(char *buf, size_t buf_size, |
940e8e66 |
1291 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv, |
69a3cdfc |
1292 | enum opr_lenmod lmod, const char *expr) |
91977a1c |
1293 | { |
69a3cdfc |
1294 | const char *cast, *scast; |
91977a1c |
1295 | |
69a3cdfc |
1296 | cast = lmod_cast_u(po, lmod); |
1297 | scast = lmod_cast_s(po, lmod); |
1298 | |
1299 | switch (pfo) { |
1300 | case PFO_Z: |
87bf6cec |
1301 | case PFO_BE: // CF=1||ZF=1; CF=0 |
850c9265 |
1302 | snprintf(buf, buf_size, "(%s%s %s 0)", |
940e8e66 |
1303 | cast, expr, is_inv ? "!=" : "=="); |
91977a1c |
1304 | break; |
850c9265 |
1305 | |
5101a5f9 |
1306 | case PFO_S: |
1307 | case PFO_L: // SF!=OF; OF=0 |
1308 | snprintf(buf, buf_size, "(%s%s %s 0)", |
940e8e66 |
1309 | scast, expr, is_inv ? ">=" : "<"); |
5101a5f9 |
1310 | break; |
1311 | |
87bf6cec |
1312 | case PFO_LE: // ZF=1||SF!=OF; OF=0 |
69a3cdfc |
1313 | snprintf(buf, buf_size, "(%s%s %s 0)", |
940e8e66 |
1314 | scast, expr, is_inv ? ">" : "<="); |
850c9265 |
1315 | break; |
1316 | |
91977a1c |
1317 | default: |
69a3cdfc |
1318 | ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo); |
91977a1c |
1319 | } |
1320 | } |
c36e914d |
1321 | |
850c9265 |
1322 | static void out_cmp_for_cc(char *buf, size_t buf_size, |
940e8e66 |
1323 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv, |
69a3cdfc |
1324 | enum opr_lenmod lmod, const char *expr1, const char *expr2) |
850c9265 |
1325 | { |
69a3cdfc |
1326 | const char *cast, *scast; |
850c9265 |
1327 | |
69a3cdfc |
1328 | cast = lmod_cast_u(po, lmod); |
1329 | scast = lmod_cast_s(po, lmod); |
850c9265 |
1330 | |
69a3cdfc |
1331 | switch (pfo) { |
5101a5f9 |
1332 | case PFO_C: |
1333 | // note: must be unsigned compare |
1334 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
940e8e66 |
1335 | cast, expr1, is_inv ? ">=" : "<", cast, expr2); |
5101a5f9 |
1336 | break; |
1337 | |
69a3cdfc |
1338 | case PFO_Z: |
850c9265 |
1339 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
940e8e66 |
1340 | cast, expr1, is_inv ? "!=" : "==", cast, expr2); |
850c9265 |
1341 | break; |
1342 | |
5101a5f9 |
1343 | case PFO_BE: // !a |
850c9265 |
1344 | // note: must be unsigned compare |
1345 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
940e8e66 |
1346 | cast, expr1, is_inv ? ">" : "<=", cast, expr2); |
5101a5f9 |
1347 | break; |
1348 | |
1349 | // note: must be signed compare |
1350 | case PFO_S: |
1351 | snprintf(buf, buf_size, "(%s(%s - %s) %s 0)", |
940e8e66 |
1352 | scast, expr1, expr2, is_inv ? ">=" : "<"); |
850c9265 |
1353 | break; |
1354 | |
5101a5f9 |
1355 | case PFO_L: // !ge |
850c9265 |
1356 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
940e8e66 |
1357 | scast, expr1, is_inv ? ">=" : "<", scast, expr2); |
850c9265 |
1358 | break; |
1359 | |
5101a5f9 |
1360 | case PFO_LE: |
1361 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
940e8e66 |
1362 | scast, expr1, is_inv ? ">" : "<=", scast, expr2); |
5101a5f9 |
1363 | break; |
1364 | |
850c9265 |
1365 | default: |
69a3cdfc |
1366 | ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo); |
850c9265 |
1367 | } |
1368 | } |
1369 | |
69a3cdfc |
1370 | static void out_cmp_test(char *buf, size_t buf_size, |
940e8e66 |
1371 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv) |
69a3cdfc |
1372 | { |
1373 | char buf1[256], buf2[256], buf3[256]; |
1374 | |
1375 | if (po->op == OP_TEST) { |
1376 | if (IS(opr_name(po, 0), opr_name(po, 1))) { |
1377 | out_src_opr(buf3, sizeof(buf3), po, &po->operand[0], 0); |
1378 | } |
1379 | else { |
1380 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
1381 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0); |
1382 | snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2); |
1383 | } |
940e8e66 |
1384 | out_test_for_cc(buf, buf_size, po, pfo, is_inv, |
69a3cdfc |
1385 | po->operand[0].lmod, buf3); |
1386 | } |
1387 | else if (po->op == OP_CMP) { |
1388 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[0], 0); |
1389 | out_src_opr(buf3, sizeof(buf3), po, &po->operand[1], 0); |
940e8e66 |
1390 | out_cmp_for_cc(buf, buf_size, po, pfo, is_inv, |
69a3cdfc |
1391 | po->operand[0].lmod, buf2, buf3); |
1392 | } |
1393 | else |
1394 | ferr(po, "%s: unhandled op: %d\n", __func__, po->op); |
1395 | } |
1396 | |
850c9265 |
1397 | static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1, |
91977a1c |
1398 | struct parsed_opr *popr2) |
1399 | { |
1400 | if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC) |
1401 | ferr(po, "missing lmod for both operands\n"); |
1402 | |
1403 | if (popr1->lmod == OPLM_UNSPEC) |
1404 | popr1->lmod = popr2->lmod; |
1405 | else if (popr2->lmod == OPLM_UNSPEC) |
1406 | popr2->lmod = popr1->lmod; |
1407 | else if (popr1->lmod != popr2->lmod) |
1408 | ferr(po, "conflicting lmods: %d vs %d\n", popr1->lmod, popr2->lmod); |
1409 | } |
c36e914d |
1410 | |
850c9265 |
1411 | static const char *op_to_c(struct parsed_op *po) |
1412 | { |
1413 | switch (po->op) |
1414 | { |
1415 | case OP_ADD: |
5101a5f9 |
1416 | case OP_ADC: |
850c9265 |
1417 | return "+"; |
1418 | case OP_SUB: |
5101a5f9 |
1419 | case OP_SBB: |
850c9265 |
1420 | return "-"; |
1421 | case OP_AND: |
1422 | return "&"; |
1423 | case OP_OR: |
1424 | return "|"; |
1425 | case OP_XOR: |
1426 | return "^"; |
1427 | case OP_SHL: |
1428 | return "<<"; |
1429 | case OP_SHR: |
1430 | return ">>"; |
1431 | case OP_MUL: |
1432 | case OP_IMUL: |
1433 | return "*"; |
1434 | default: |
1435 | ferr(po, "op_to_c was supplied with %d\n", po->op); |
1436 | } |
1437 | } |
1438 | |
d4e3b5db |
1439 | static void set_flag_no_dup(struct parsed_op *po, enum op_flags flag, |
1440 | enum op_flags flag_check) |
1441 | { |
1442 | if (po->flags & flag) |
1443 | ferr(po, "flag %x already set\n", flag); |
1444 | if (po->flags & flag_check) |
1445 | ferr(po, "flag_check %x already set\n", flag_check); |
1446 | |
1447 | po->flags |= flag; |
1448 | } |
1449 | |
87bf6cec |
1450 | static int scan_for_pop(int i, int opcnt, const char *reg, |
d4e3b5db |
1451 | int magic, int depth, int *maxdepth, int do_flags) |
850c9265 |
1452 | { |
87bf6cec |
1453 | struct parsed_op *po; |
1454 | int ret = 0; |
1455 | |
850c9265 |
1456 | for (; i < opcnt; i++) { |
87bf6cec |
1457 | po = &ops[i]; |
1458 | if (po->cc_scratch == magic) |
1459 | break; // already checked |
1460 | po->cc_scratch = magic; |
1461 | |
1462 | if (po->flags & OPF_TAIL) |
1463 | return -1; // deadend |
1464 | |
1bafb621 |
1465 | if ((po->flags & OPF_RMD) |
1466 | || (po->op == OP_PUSH && po->argmask)) // arg push |
850c9265 |
1467 | continue; |
1468 | |
87bf6cec |
1469 | if ((po->flags & OPF_JMP) && po->op != OP_CALL) { |
1470 | if (po->bt_i < 0) { |
1471 | ferr(po, "dead branch\n"); |
1472 | return -1; |
1473 | } |
850c9265 |
1474 | |
87bf6cec |
1475 | if (po->flags & OPF_CC) { |
d4e3b5db |
1476 | ret |= scan_for_pop(po->bt_i, opcnt, reg, magic, |
1477 | depth, maxdepth, do_flags); |
87bf6cec |
1478 | if (ret < 0) |
1479 | return ret; // dead end |
1480 | } |
1481 | else { |
1482 | i = po->bt_i - 1; |
1483 | } |
1484 | continue; |
1485 | } |
1486 | |
d4e3b5db |
1487 | if ((po->op == OP_POP || po->op == OP_PUSH) |
1488 | && po->operand[0].type == OPT_REG |
87bf6cec |
1489 | && IS(po->operand[0].name, reg)) |
1490 | { |
d4e3b5db |
1491 | if (po->op == OP_PUSH) { |
1492 | depth++; |
1493 | if (depth > *maxdepth) |
1494 | *maxdepth = depth; |
1495 | if (do_flags) |
1496 | set_flag_no_dup(po, OPF_RSAVE, OPF_RMD); |
1497 | } |
1498 | else if (depth == 0) { |
1499 | if (do_flags) |
1500 | set_flag_no_dup(po, OPF_RMD, OPF_RSAVE); |
1501 | return 1; |
1502 | } |
1503 | else { |
1504 | depth--; |
1505 | if (depth < 0) // should not happen |
1506 | ferr(po, "fail with depth\n"); |
1507 | if (do_flags) |
1508 | set_flag_no_dup(po, OPF_RSAVE, OPF_RMD); |
1509 | } |
87bf6cec |
1510 | } |
850c9265 |
1511 | } |
1512 | |
87bf6cec |
1513 | return ret; |
850c9265 |
1514 | } |
1515 | |
69a3cdfc |
1516 | // scan for pop starting from 'ret' op (all paths) |
d4e3b5db |
1517 | static int scan_for_pop_ret(int i, int opcnt, const char *reg, |
1518 | int flag_set) |
850c9265 |
1519 | { |
1520 | int found = 0; |
1521 | int j; |
1522 | |
1523 | for (; i < opcnt; i++) { |
87bf6cec |
1524 | if (!(ops[i].flags & OPF_TAIL)) |
850c9265 |
1525 | continue; |
1526 | |
1527 | for (j = i - 1; j >= 0; j--) { |
69a3cdfc |
1528 | if (ops[j].flags & OPF_RMD) |
1529 | continue; |
1530 | if (ops[j].flags & OPF_JMP) |
850c9265 |
1531 | return -1; |
1532 | |
1533 | if (ops[j].op == OP_POP && ops[j].operand[0].type == OPT_REG |
1534 | && IS(ops[j].operand[0].name, reg)) |
1535 | { |
1536 | found = 1; |
d4e3b5db |
1537 | ops[j].flags |= flag_set; |
850c9265 |
1538 | break; |
1539 | } |
1540 | |
1541 | if (g_labels[j][0] != 0) |
1542 | return -1; |
1543 | } |
1544 | } |
1545 | |
1546 | return found ? 0 : -1; |
1547 | } |
1548 | |
5101a5f9 |
1549 | // is operand 'opr modified' by parsed_op 'po'? |
1550 | static int is_opr_modified(const struct parsed_opr *opr, |
69a3cdfc |
1551 | const struct parsed_op *po) |
1552 | { |
1553 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA)) |
1554 | return 0; |
1555 | |
1556 | if (opr->type == OPT_REG && po->operand[0].type == OPT_REG) { |
1557 | if (po->regmask_dst & (1 << opr->reg)) |
1558 | return 1; |
1559 | else |
1560 | return 0; |
1561 | } |
1562 | |
1563 | return IS(po->operand[0].name, opr->name); |
1564 | } |
1565 | |
5101a5f9 |
1566 | // is any operand of parsed_op 'po_test' modified by parsed_op 'po'? |
1567 | static int is_any_opr_modified(const struct parsed_op *po_test, |
1568 | const struct parsed_op *po) |
1569 | { |
1570 | int i; |
1571 | |
1572 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA)) |
1573 | return 0; |
1574 | |
1575 | if (po_test->regmask_src & po->regmask_dst) |
1576 | return 1; |
1577 | |
1578 | for (i = 0; i < po_test->operand_cnt; i++) |
1579 | if (IS(po_test->operand[i].name, po->operand[0].name)) |
1580 | return 1; |
1581 | |
1582 | return 0; |
1583 | } |
1584 | |
940e8e66 |
1585 | // scan for any po_test operand modification in range given |
5101a5f9 |
1586 | static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt) |
69a3cdfc |
1587 | { |
1588 | for (; i < opcnt; i++) { |
5101a5f9 |
1589 | if (is_any_opr_modified(po_test, &ops[i])) |
69a3cdfc |
1590 | return i; |
1591 | } |
1592 | |
1593 | return -1; |
1594 | } |
1595 | |
940e8e66 |
1596 | // scan for po_test operand[0] modification in range given |
1597 | static int scan_for_mod_opr0(struct parsed_op *po_test, |
1598 | int i, int opcnt) |
1599 | { |
1600 | for (; i < opcnt; i++) { |
1601 | if (is_opr_modified(&po_test->operand[0], &ops[i])) |
1602 | return i; |
1603 | } |
1604 | |
1605 | return -1; |
1606 | } |
1607 | |
5101a5f9 |
1608 | static int scan_for_flag_set(int i) |
69a3cdfc |
1609 | { |
1610 | for (; i >= 0; i--) { |
1611 | if (ops[i].flags & OPF_FLAGS) |
1612 | return i; |
1613 | |
1614 | if ((ops[i].flags & OPF_JMP) && !(ops[i].flags & OPF_CC)) |
1615 | return -1; |
1616 | if (g_labels[i][0] != 0) |
1617 | return -1; |
1618 | } |
1619 | |
1620 | return -1; |
1621 | } |
1622 | |
5101a5f9 |
1623 | // scan back for cdq, if anything modifies edx, fail |
1624 | static int scan_for_cdq_edx(int i) |
1625 | { |
1626 | for (; i >= 0; i--) { |
1627 | if (ops[i].op == OP_CDQ) |
1628 | return i; |
1629 | |
1630 | if (ops[i].regmask_dst & (1 << xDX)) |
1631 | return -1; |
1632 | if (g_labels[i][0] != 0) |
1633 | return -1; |
1634 | } |
1635 | |
1636 | return -1; |
1637 | } |
1638 | |
64c59faf |
1639 | static int scan_for_reg_clear(int i, int reg) |
1640 | { |
1641 | for (; i >= 0; i--) { |
1642 | if (ops[i].op == OP_XOR |
1643 | && ops[i].operand[0].lmod == OPLM_DWORD |
1644 | && ops[i].operand[0].reg == ops[i].operand[1].reg |
1645 | && ops[i].operand[0].reg == reg) |
1646 | return i; |
1647 | |
1648 | if (ops[i].regmask_dst & (1 << reg)) |
1649 | return -1; |
1650 | if (g_labels[i][0] != 0) |
1651 | return -1; |
1652 | } |
1653 | |
1654 | return -1; |
1655 | } |
1656 | |
1bafb621 |
1657 | // scan for positive, constant esp adjust |
1658 | static int scan_for_esp_adjust(int i, int opcnt, int *adj) |
1659 | { |
1660 | for (; i < opcnt; i++) { |
1661 | if (ops[i].op == OP_ADD && ops[i].operand[0].reg == xSP) { |
1662 | if (ops[i].operand[1].type != OPT_CONST) |
1663 | ferr(&ops[i], "non-const esp adjust?\n"); |
1664 | *adj = ops[i].operand[1].val; |
1665 | if (*adj & 3) |
1666 | ferr(&ops[i], "unaligned esp adjust: %x\n", *adj); |
1667 | return i; |
1668 | } |
1669 | |
1670 | if ((ops[i].flags & (OPF_JMP|OPF_TAIL)) |
1671 | || ops[i].op == OP_PUSH || ops[i].op == OP_POP) |
1672 | return -1; |
1673 | if (g_labels[i][0] != 0) |
1674 | return -1; |
1675 | } |
1676 | |
1677 | return -1; |
1678 | } |
1679 | |
91977a1c |
1680 | static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt) |
1681 | { |
69a3cdfc |
1682 | struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op; |
850c9265 |
1683 | struct parsed_opr *last_arith_dst = NULL; |
91977a1c |
1684 | char buf1[256], buf2[256], buf3[256]; |
940e8e66 |
1685 | struct parsed_proto *pp, *pp_tmp; |
91977a1c |
1686 | const char *tmpname; |
940e8e66 |
1687 | enum parsed_flag_op pfo; |
69a3cdfc |
1688 | int save_arg_vars = 0; |
1689 | int cmp_result_vars = 0; |
87bf6cec |
1690 | int need_mul_var = 0; |
91977a1c |
1691 | int had_decl = 0; |
d4e3b5db |
1692 | int regmask_save = 0; |
91977a1c |
1693 | int regmask_arg = 0; |
1694 | int regmask = 0; |
940e8e66 |
1695 | int pfomask = 0; |
64c59faf |
1696 | int found = 0; |
d4e3b5db |
1697 | int depth = 0; |
91977a1c |
1698 | int no_output; |
69a3cdfc |
1699 | int dummy; |
91977a1c |
1700 | int arg; |
1701 | int i, j; |
1702 | int reg; |
1703 | int ret; |
1704 | |
1bafb621 |
1705 | g_bp_frame = g_sp_frame = g_stack_fsz = 0; |
91977a1c |
1706 | |
1707 | ret = proto_parse(fhdr, funcn, &g_func_pp); |
1708 | if (ret) |
1709 | ferr(ops, "proto_parse failed for '%s'\n", funcn); |
1710 | |
1711 | fprintf(fout, "%s %s(", g_func_pp.ret_type, funcn); |
1712 | for (i = 0; i < g_func_pp.argc; i++) { |
1713 | if (i > 0) |
1714 | fprintf(fout, ", "); |
1715 | fprintf(fout, "%s a%d", g_func_pp.arg[i].type, i + 1); |
1716 | } |
1717 | fprintf(fout, ")\n{\n"); |
1718 | |
1719 | // pass1: |
1bafb621 |
1720 | // - handle ebp/esp frame, remove ops related to it |
91977a1c |
1721 | if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp") |
1722 | && ops[1].op == OP_MOV |
1723 | && IS(opr_name(&ops[1], 0), "ebp") |
1724 | && IS(opr_name(&ops[1], 1), "esp")) |
1725 | { |
87bf6cec |
1726 | int ecx_push = 0; |
1727 | |
91977a1c |
1728 | g_bp_frame = 1; |
69a3cdfc |
1729 | ops[0].flags |= OPF_RMD; |
1730 | ops[1].flags |= OPF_RMD; |
91977a1c |
1731 | |
1732 | if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) { |
1bafb621 |
1733 | g_stack_fsz = opr_const(&ops[2], 1); |
69a3cdfc |
1734 | ops[2].flags |= OPF_RMD; |
91977a1c |
1735 | } |
87bf6cec |
1736 | else { |
1737 | // another way msvc builds stack frame.. |
1738 | i = 2; |
1739 | while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) { |
1bafb621 |
1740 | g_stack_fsz += 4; |
87bf6cec |
1741 | ops[i].flags |= OPF_RMD; |
1742 | ecx_push++; |
1743 | i++; |
1744 | } |
1745 | } |
91977a1c |
1746 | |
64c59faf |
1747 | found = 0; |
91977a1c |
1748 | i = 2; |
1749 | do { |
1750 | for (; i < opcnt; i++) |
1751 | if (ops[i].op == OP_RET) |
1752 | break; |
64c59faf |
1753 | if (i == opcnt && (ops[i - 1].flags & OPF_JMP) && found) |
1754 | break; |
1755 | |
91977a1c |
1756 | if (ops[i - 1].op != OP_POP || !IS(opr_name(&ops[i - 1], 0), "ebp")) |
1757 | ferr(&ops[i - 1], "'pop ebp' expected\n"); |
69a3cdfc |
1758 | ops[i - 1].flags |= OPF_RMD; |
91977a1c |
1759 | |
1bafb621 |
1760 | if (g_stack_fsz != 0) { |
91977a1c |
1761 | if (ops[i - 2].op != OP_MOV |
1762 | || !IS(opr_name(&ops[i - 2], 0), "esp") |
1763 | || !IS(opr_name(&ops[i - 2], 1), "ebp")) |
1764 | { |
1765 | ferr(&ops[i - 2], "esp restore expected\n"); |
1766 | } |
69a3cdfc |
1767 | ops[i - 2].flags |= OPF_RMD; |
87bf6cec |
1768 | |
1769 | if (ecx_push && ops[i - 3].op == OP_POP |
1770 | && IS(opr_name(&ops[i - 3], 0), "ecx")) |
1771 | { |
1772 | ferr(&ops[i - 3], "unexpected ecx pop\n"); |
1773 | } |
91977a1c |
1774 | } |
87bf6cec |
1775 | |
64c59faf |
1776 | found = 1; |
91977a1c |
1777 | i++; |
1778 | } while (i < opcnt); |
1779 | } |
1bafb621 |
1780 | else { |
1781 | for (i = 0; i < opcnt; i++) { |
1782 | if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL))) |
1783 | break; |
1784 | if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP |
1785 | && ops[i].operand[1].type == OPT_CONST) |
1786 | { |
1787 | g_sp_frame = 1; |
1788 | break; |
1789 | } |
1790 | } |
1791 | |
1792 | if (g_sp_frame) |
1793 | { |
1794 | g_stack_fsz = ops[i].operand[1].val; |
1795 | ops[i].flags |= OPF_RMD; |
1796 | |
1797 | i++; |
1798 | do { |
1799 | for (; i < opcnt; i++) |
1800 | if (ops[i].op == OP_RET) |
1801 | break; |
1802 | if (ops[i - 1].op != OP_ADD |
1803 | || !IS(opr_name(&ops[i - 1], 0), "esp") |
1804 | || ops[i - 1].operand[1].type != OPT_CONST |
1805 | || ops[i - 1].operand[1].val != g_stack_fsz) |
1806 | ferr(&ops[i - 1], "'add esp' expected\n"); |
1807 | ops[i - 1].flags |= OPF_RMD; |
1808 | |
1809 | i++; |
1810 | } while (i < opcnt); |
1811 | } |
1812 | } |
91977a1c |
1813 | |
1814 | // pass2: |
87bf6cec |
1815 | // - resolve all branches |
1816 | for (i = 0; i < opcnt; i++) { |
1817 | po = &ops[i]; |
1818 | po->bt_i = -1; |
1819 | |
1820 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_JMP) |
1821 | || po->op == OP_CALL || po->op == OP_RET) |
1822 | continue; |
1823 | |
1824 | for (j = 0; j < opcnt; j++) { |
1825 | if (g_labels[j][0] && IS(po->operand[0].name, g_labels[j])) { |
1826 | po->bt_i = j; |
940e8e66 |
1827 | po->lrl = g_label_refs[j]; |
1828 | g_label_refs[j] = po; |
87bf6cec |
1829 | break; |
1830 | } |
1831 | } |
1832 | |
1833 | if (po->bt_i == -1) { |
1834 | // assume tail call |
1835 | po->op = OP_CALL; |
1836 | po->flags |= OPF_TAIL; |
1837 | } |
1838 | } |
1839 | |
1840 | // pass3: |
91977a1c |
1841 | // - process calls |
1bafb621 |
1842 | for (i = 0; i < opcnt; i++) |
1843 | { |
69a3cdfc |
1844 | po = &ops[i]; |
1845 | if (po->flags & OPF_RMD) |
91977a1c |
1846 | continue; |
850c9265 |
1847 | |
d4e3b5db |
1848 | if (po->op == OP_CALL) |
69a3cdfc |
1849 | { |
1bafb621 |
1850 | pp = calloc(1, sizeof(*pp)); |
91977a1c |
1851 | my_assert_not(pp, NULL); |
1bafb621 |
1852 | tmpname = opr_name(po, 0); |
1853 | if (po->operand[0].type != OPT_LABEL) |
1854 | { |
1855 | ret = scan_for_esp_adjust(i + 1, opcnt, &j); |
1856 | if (ret < 0) |
1857 | ferr(po, "non-__cdecl indirect call unhandled yet\n"); |
1858 | j /= 4; |
1859 | if (j > ARRAY_SIZE(pp->arg)) |
1860 | ferr(po, "esp adjust too large?\n"); |
64c59faf |
1861 | pp->ret_type = strdup("int"); |
1bafb621 |
1862 | pp->argc = pp->argc_stack = j; |
1863 | for (arg = 0; arg < pp->argc; arg++) |
64c59faf |
1864 | pp->arg[arg].type = strdup("int"); |
1bafb621 |
1865 | } |
1866 | else { |
1867 | ret = proto_parse(fhdr, tmpname, pp); |
1868 | if (ret) |
1869 | ferr(po, "proto_parse failed for call '%s'\n", tmpname); |
1870 | } |
1871 | |
1872 | ret = scan_for_esp_adjust(i + 1, opcnt, &j); |
1873 | if (ret >= 0) { |
1874 | if (pp->argc_stack != j / 4) |
1875 | ferr(po, "stack tracking failed: %x %x\n", |
1876 | pp->argc_stack, j); |
1877 | ops[ret].flags |= OPF_RMD; |
1878 | } |
91977a1c |
1879 | |
39b168b8 |
1880 | // can't call functions with non-__cdecl callbacks yet |
1881 | for (arg = 0; arg < pp->argc; arg++) { |
1882 | if (pp->arg[arg].fptr != NULL) { |
1883 | pp_tmp = pp->arg[arg].fptr; |
1884 | if (pp_tmp->is_stdcall || pp_tmp->argc != pp_tmp->argc_stack) |
1885 | ferr(po, "'%s' has a non-__cdecl callback\n", tmpname); |
1886 | } |
1887 | } |
1888 | |
91977a1c |
1889 | for (arg = 0; arg < pp->argc; arg++) |
1890 | if (pp->arg[arg].reg == NULL) |
1891 | break; |
1892 | |
1bafb621 |
1893 | for (j = i; j >= 0 && arg < pp->argc; ) |
1894 | { |
1895 | if (g_labels[j][0] != 0) { |
1896 | if (j > 0 && ((ops[j - 1].flags & OPF_TAIL) |
1897 | || (ops[j - 1].flags & (OPF_JMP|OPF_CC)) == OPF_JMP)) |
1898 | { |
1899 | // follow the branch in reverse |
1900 | if (g_label_refs[j] == NULL) |
1901 | ferr(po, "no refs for '%s'?\n", g_labels[j]); |
1902 | if (g_label_refs[j]->lrl != NULL) |
1903 | ferr(po, "unhandled multiple fefs to '%s'\n", g_labels[j]); |
1904 | j = (g_label_refs[j] - ops) + 1; |
1905 | continue; |
1906 | } |
1907 | break; |
1908 | } |
1909 | j--; |
1910 | |
1911 | if (ops[j].op == OP_CALL) |
1912 | { |
940e8e66 |
1913 | pp_tmp = ops[j].datap; |
1914 | if (pp_tmp == NULL) |
1915 | ferr(po, "arg collect hit unparsed call\n"); |
1916 | if (pp_tmp->argc_stack > 0) |
1917 | ferr(po, "arg collect hit '%s' with %d stack args\n", |
1918 | opr_name(&ops[j], 0), pp_tmp->argc_stack); |
1919 | } |
1920 | else if ((ops[j].flags & OPF_TAIL) |
1921 | || (ops[j].flags & (OPF_JMP|OPF_CC)) == OPF_JMP) |
1922 | { |
1923 | break; |
69a3cdfc |
1924 | } |
1bafb621 |
1925 | else if (ops[j].op == OP_PUSH) |
1926 | { |
940e8e66 |
1927 | pp->arg[arg].datap = &ops[j]; |
1928 | ret = scan_for_mod(&ops[j], j + 1, i); |
1929 | if (ret >= 0) { |
1930 | // mark this push as one that needs operand saving |
1bafb621 |
1931 | ops[j].flags &= ~OPF_RMD; |
940e8e66 |
1932 | ops[j].argmask |= 1 << arg; |
1933 | save_arg_vars |= 1 << arg; |
1934 | } |
1935 | else |
1936 | ops[j].flags |= OPF_RMD; |
1937 | |
1938 | // next arg |
1939 | for (arg++; arg < pp->argc; arg++) |
1940 | if (pp->arg[arg].reg == NULL) |
1941 | break; |
1942 | } |
91977a1c |
1943 | } |
1944 | if (arg < pp->argc) |
64c59faf |
1945 | ferr(po, "arg collect failed for '%s': %d/%d\n", |
1946 | tmpname, arg, pp->argc); |
69a3cdfc |
1947 | po->datap = pp; |
91977a1c |
1948 | } |
d4e3b5db |
1949 | } |
1950 | |
1951 | // pass4: |
1952 | // - find POPs for PUSHes, rm both |
1953 | // - scan for all used registers |
1954 | // - find flag set ops for their users |
1bafb621 |
1955 | // - declare indirect functions |
d4e3b5db |
1956 | for (i = 0; i < opcnt; i++) { |
1957 | po = &ops[i]; |
1958 | if (po->flags & OPF_RMD) |
1959 | continue; |
1960 | |
1961 | if (po->op == OP_PUSH |
1962 | && po->argmask == 0 && !(po->flags & OPF_RSAVE) |
1963 | && po->operand[0].type == OPT_REG) |
1964 | { |
1965 | reg = po->operand[0].reg; |
1966 | if (reg < 0) |
1967 | ferr(po, "reg not set for push?\n"); |
1968 | |
1969 | depth = 0; |
1970 | ret = scan_for_pop(i + 1, opcnt, |
1971 | po->operand[0].name, i + opcnt, 0, &depth, 0); |
1972 | if (ret == 1) { |
1973 | if (depth > 1) |
1974 | ferr(po, "too much depth: %d\n", depth); |
1975 | if (depth > 0) |
1976 | regmask_save |= 1 << reg; |
1977 | |
1978 | po->flags |= OPF_RMD; |
1979 | scan_for_pop(i + 1, opcnt, po->operand[0].name, |
1980 | i + opcnt * 2, 0, &depth, 1); |
1981 | continue; |
1982 | } |
1983 | ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0); |
1984 | if (ret == 0) { |
1985 | arg = OPF_RMD; |
1986 | if (regmask & (1 << reg)) { |
1987 | if (regmask_save & (1 << reg)) |
1988 | ferr(po, "%s already saved?\n", po->operand[0].name); |
1989 | arg = OPF_RSAVE; |
1990 | } |
1991 | po->flags |= arg; |
1992 | scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg); |
1993 | continue; |
1994 | } |
1995 | } |
1996 | |
1997 | regmask |= po->regmask_src | po->regmask_dst; |
1998 | |
1999 | if (po->flags & OPF_CC) |
2000 | { |
2001 | ret = scan_for_flag_set(i - 1); |
2002 | if (ret < 0) |
2003 | ferr(po, "unable to trace flag setter\n"); |
2004 | |
2005 | tmp_op = &ops[ret]; // flag setter |
2006 | pfo = split_cond(po, po->op, &dummy); |
2007 | pfomask = 0; |
2008 | |
2009 | // to get nicer code, we try to delay test and cmp; |
2010 | // if we can't because of operand modification, or if we |
2011 | // have math op, make it calculate flags explicitly |
2012 | if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP) { |
2013 | if (scan_for_mod(tmp_op, ret + 1, i) >= 0) |
2014 | pfomask = 1 << pfo; |
2015 | } |
2016 | else { |
2017 | if ((pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P) |
2018 | || scan_for_mod_opr0(tmp_op, ret + 1, i) >= 0) |
2019 | pfomask = 1 << pfo; |
2020 | } |
2021 | if (pfomask) { |
2022 | tmp_op->pfomask |= pfomask; |
2023 | cmp_result_vars |= pfomask; |
2024 | po->datap = tmp_op; |
2025 | } |
2026 | |
2027 | if (po->op == OP_ADC || po->op == OP_SBB) |
2028 | cmp_result_vars |= 1 << PFO_C; |
2029 | } |
87bf6cec |
2030 | else if (po->op == OP_MUL |
2031 | || (po->op == OP_IMUL && po->operand_cnt == 1)) |
2032 | { |
2033 | need_mul_var = 1; |
2034 | } |
1bafb621 |
2035 | else if (po->op == OP_CALL && po->operand[0].type != OPT_LABEL) { |
2036 | pp = po->datap; |
2037 | my_assert_not(pp, NULL); |
2038 | fprintf(fout, " %s (*icall%d)(", pp->ret_type, i); |
2039 | for (j = 0; j < pp->argc; j++) { |
2040 | if (j > 0) |
2041 | fprintf(fout, ", "); |
2042 | fprintf(fout, "%s a%d", pp->arg[j].type, j + 1); |
2043 | } |
2044 | fprintf(fout, ");\n"); |
2045 | } |
91977a1c |
2046 | } |
2047 | |
850c9265 |
2048 | // declare stack frame |
1bafb621 |
2049 | if (g_stack_fsz) |
850c9265 |
2050 | fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n", |
1bafb621 |
2051 | (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz); |
850c9265 |
2052 | |
940e8e66 |
2053 | // declare arg-registers |
91977a1c |
2054 | for (i = 0; i < g_func_pp.argc; i++) { |
2055 | if (g_func_pp.arg[i].reg != NULL) { |
2056 | reg = char_array_i(regs_r32, |
2057 | ARRAY_SIZE(regs_r32), g_func_pp.arg[i].reg); |
2058 | if (reg < 0) |
2059 | ferr(ops, "arg '%s' is not a reg?\n", g_func_pp.arg[i].reg); |
2060 | |
2061 | regmask_arg |= 1 << reg; |
2062 | fprintf(fout, " u32 %s = (u32)a%d;\n", |
850c9265 |
2063 | g_func_pp.arg[i].reg, i + 1); |
91977a1c |
2064 | had_decl = 1; |
2065 | } |
2066 | } |
2067 | |
940e8e66 |
2068 | // declare other regs - special case for eax |
91977a1c |
2069 | if (!((regmask | regmask_arg) & 1) && !IS(g_func_pp.ret_type, "void")) { |
2070 | fprintf(fout, " u32 eax = 0;\n"); |
2071 | had_decl = 1; |
2072 | } |
2073 | |
2074 | regmask &= ~regmask_arg; |
d4e3b5db |
2075 | regmask &= ~(1 << xSP); |
91977a1c |
2076 | if (g_bp_frame) |
2077 | regmask &= ~(1 << xBP); |
2078 | if (regmask) { |
2079 | for (reg = 0; reg < 8; reg++) { |
2080 | if (regmask & (1 << reg)) { |
2081 | fprintf(fout, " u32 %s;\n", regs_r32[reg]); |
2082 | had_decl = 1; |
2083 | } |
2084 | } |
2085 | } |
2086 | |
d4e3b5db |
2087 | if (regmask_save) { |
2088 | for (reg = 0; reg < 8; reg++) { |
2089 | if (regmask_save & (1 << reg)) { |
2090 | fprintf(fout, " u32 s_%s;\n", regs_r32[reg]); |
2091 | had_decl = 1; |
2092 | } |
2093 | } |
2094 | } |
2095 | |
69a3cdfc |
2096 | if (save_arg_vars) { |
2097 | for (reg = 0; reg < 32; reg++) { |
2098 | if (save_arg_vars & (1 << reg)) { |
2099 | fprintf(fout, " u32 s_a%d;\n", reg + 1); |
2100 | had_decl = 1; |
2101 | } |
2102 | } |
2103 | } |
2104 | |
2105 | if (cmp_result_vars) { |
2106 | for (i = 0; i < 8; i++) { |
2107 | if (cmp_result_vars & (1 << i)) { |
2108 | fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]); |
2109 | had_decl = 1; |
2110 | } |
2111 | } |
2112 | } |
2113 | |
87bf6cec |
2114 | if (need_mul_var) { |
2115 | fprintf(fout, " u64 mul_tmp;\n"); |
2116 | had_decl = 1; |
2117 | } |
2118 | |
91977a1c |
2119 | if (had_decl) |
2120 | fprintf(fout, "\n"); |
2121 | |
2122 | // output ops |
69a3cdfc |
2123 | for (i = 0; i < opcnt; i++) |
2124 | { |
d4e3b5db |
2125 | if (g_labels[i][0] != 0 && g_label_refs[i] != NULL) |
91977a1c |
2126 | fprintf(fout, "\n%s:\n", g_labels[i]); |
2127 | |
69a3cdfc |
2128 | po = &ops[i]; |
2129 | if (po->flags & OPF_RMD) |
91977a1c |
2130 | continue; |
2131 | |
2132 | no_output = 0; |
2133 | |
91977a1c |
2134 | #define assert_operand_cnt(n_) \ |
850c9265 |
2135 | if (po->operand_cnt != n_) \ |
2136 | ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_) |
2137 | |
69a3cdfc |
2138 | // conditional/flag using op? |
2139 | if (po->flags & OPF_CC) |
850c9265 |
2140 | { |
940e8e66 |
2141 | int is_delayed = 0; |
2142 | int is_inv = 0; |
69a3cdfc |
2143 | |
940e8e66 |
2144 | pfo = split_cond(po, po->op, &is_inv); |
850c9265 |
2145 | |
69a3cdfc |
2146 | // we go through all this trouble to avoid using parsed_flag_op, |
2147 | // which makes generated code much nicer |
2148 | if (delayed_flag_op != NULL) |
850c9265 |
2149 | { |
940e8e66 |
2150 | out_cmp_test(buf1, sizeof(buf1), delayed_flag_op, pfo, is_inv); |
2151 | is_delayed = 1; |
91977a1c |
2152 | } |
850c9265 |
2153 | else if (last_arith_dst != NULL |
69a3cdfc |
2154 | && (pfo == PFO_Z || pfo == PFO_S || pfo == PFO_P)) |
850c9265 |
2155 | { |
2156 | out_src_opr(buf3, sizeof(buf3), po, last_arith_dst, 0); |
940e8e66 |
2157 | out_test_for_cc(buf1, sizeof(buf1), po, pfo, is_inv, |
850c9265 |
2158 | last_arith_dst->lmod, buf3); |
940e8e66 |
2159 | is_delayed = 1; |
850c9265 |
2160 | } |
69a3cdfc |
2161 | else if (po->datap != NULL) { |
2162 | // use preprocessed results |
2163 | tmp_op = po->datap; |
2164 | if (!tmp_op || !(tmp_op->pfomask & (1 << pfo))) |
2165 | ferr(po, "not prepared for pfo %d\n", pfo); |
2166 | |
940e8e66 |
2167 | // note: is_inv was not yet applied |
69a3cdfc |
2168 | snprintf(buf1, sizeof(buf1), "(%scond_%s)", |
940e8e66 |
2169 | is_inv ? "!" : "", parsed_flag_op_names[pfo]); |
69a3cdfc |
2170 | } |
2171 | else { |
2172 | ferr(po, "all methods of finding comparison failed\n"); |
2173 | } |
850c9265 |
2174 | |
69a3cdfc |
2175 | if (po->flags & OPF_JMP) { |
850c9265 |
2176 | fprintf(fout, " if %s\n", buf1); |
2177 | } |
5101a5f9 |
2178 | else if (po->op == OP_ADC || po->op == OP_SBB) { |
940e8e66 |
2179 | if (is_delayed) |
2180 | fprintf(fout, " cond_%s = %s;\n", |
2181 | parsed_flag_op_names[pfo], buf1); |
850c9265 |
2182 | } |
5101a5f9 |
2183 | else if (po->flags & OPF_DATA) { // SETcc |
850c9265 |
2184 | out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]); |
2185 | fprintf(fout, " %s = %s;", buf2, buf1); |
91977a1c |
2186 | } |
69a3cdfc |
2187 | else { |
2188 | ferr(po, "unhandled conditional op\n"); |
2189 | } |
91977a1c |
2190 | } |
2191 | |
940e8e66 |
2192 | pfomask = po->pfomask; |
2193 | |
850c9265 |
2194 | switch (po->op) |
91977a1c |
2195 | { |
2196 | case OP_MOV: |
2197 | assert_operand_cnt(2); |
850c9265 |
2198 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
06c5d854 |
2199 | fprintf(fout, " %s = %s%s;", |
850c9265 |
2200 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
06c5d854 |
2201 | po->operand[0].is_ptr ? "(void *)" : "", |
850c9265 |
2202 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2203 | break; |
2204 | |
2205 | case OP_LEA: |
2206 | assert_operand_cnt(2); |
87bf6cec |
2207 | po->operand[1].lmod = OPLM_DWORD; // always |
850c9265 |
2208 | fprintf(fout, " %s = %s;", |
2209 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2210 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 1)); |
2211 | break; |
2212 | |
2213 | case OP_MOVZX: |
2214 | assert_operand_cnt(2); |
91977a1c |
2215 | fprintf(fout, " %s = %s;", |
850c9265 |
2216 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2217 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2218 | break; |
2219 | |
2220 | case OP_MOVSX: |
2221 | assert_operand_cnt(2); |
2222 | switch (po->operand[1].lmod) { |
2223 | case OPLM_BYTE: |
2224 | strcpy(buf3, "(s8)"); |
2225 | break; |
2226 | case OPLM_WORD: |
2227 | strcpy(buf3, "(s16)"); |
2228 | break; |
2229 | default: |
2230 | ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod); |
2231 | } |
2232 | fprintf(fout, " %s = %s%s;", |
2233 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2234 | buf3, |
2235 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2236 | break; |
2237 | |
2238 | case OP_NOT: |
2239 | assert_operand_cnt(1); |
2240 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
2241 | fprintf(fout, " %s = ~%s;", buf1, buf1); |
2242 | break; |
2243 | |
5101a5f9 |
2244 | case OP_CDQ: |
2245 | assert_operand_cnt(2); |
2246 | fprintf(fout, " %s = (s32)%s >> 31;", |
2247 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2248 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2249 | strcpy(g_comment, "cdq"); |
2250 | break; |
2251 | |
33c35af6 |
2252 | case OP_STOS: |
2253 | // assumes DF=0 |
2254 | assert_operand_cnt(3); |
2255 | if (po->flags & OPF_REP) { |
2256 | fprintf(fout, " for (; ecx != 0; ecx--, edi += %d)\n", |
2257 | lmod_bytes(po, po->operand[0].lmod)); |
d4e3b5db |
2258 | fprintf(fout, " %sedi = eax;", |
2259 | lmod_cast_u_ptr(po, po->operand[0].lmod)); |
33c35af6 |
2260 | strcpy(g_comment, "rep stos"); |
2261 | } |
2262 | else { |
d4e3b5db |
2263 | fprintf(fout, " %sedi = eax; edi += %d;", |
2264 | lmod_cast_u_ptr(po, po->operand[0].lmod), |
33c35af6 |
2265 | lmod_bytes(po, po->operand[0].lmod)); |
2266 | strcpy(g_comment, "stos"); |
2267 | } |
2268 | break; |
2269 | |
d4e3b5db |
2270 | case OP_MOVS: |
2271 | // assumes DF=0 |
2272 | assert_operand_cnt(3); |
2273 | j = lmod_bytes(po, po->operand[0].lmod); |
2274 | strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod)); |
2275 | if (po->flags & OPF_REP) { |
2276 | fprintf(fout, |
2277 | " for (; ecx != 0; ecx--, edi += %d, esi += %d)\n", |
2278 | j, j); |
2279 | fprintf(fout, |
2280 | " %sedi = %sesi;", buf1, buf1); |
2281 | strcpy(g_comment, "rep movs"); |
2282 | } |
2283 | else { |
2284 | fprintf(fout, " %sedi = %sesi; edi += %d; esi += %d;", |
2285 | buf1, buf1, j, j); |
2286 | strcpy(g_comment, "movs"); |
2287 | } |
2288 | break; |
2289 | |
850c9265 |
2290 | // arithmetic w/flags |
2291 | case OP_ADD: |
2292 | case OP_SUB: |
2293 | case OP_AND: |
2294 | case OP_OR: |
5101a5f9 |
2295 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
2296 | // fallthrough |
850c9265 |
2297 | case OP_SHL: |
2298 | case OP_SHR: |
2299 | dualop_arith: |
2300 | assert_operand_cnt(2); |
850c9265 |
2301 | fprintf(fout, " %s %s= %s;", |
2302 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2303 | op_to_c(po), |
2304 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2305 | last_arith_dst = &po->operand[0]; |
69a3cdfc |
2306 | delayed_flag_op = NULL; |
850c9265 |
2307 | break; |
2308 | |
d4e3b5db |
2309 | case OP_SAR: |
2310 | assert_operand_cnt(2); |
2311 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
2312 | fprintf(fout, " %s = %s%s >> %s;", buf1, |
2313 | lmod_cast_s(po, po->operand[0].lmod), buf1, |
2314 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2315 | last_arith_dst = &po->operand[0]; |
2316 | delayed_flag_op = NULL; |
2317 | break; |
2318 | |
2319 | case OP_ROL: |
2320 | case OP_ROR: |
2321 | assert_operand_cnt(2); |
2322 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
2323 | if (po->operand[1].type == OPT_CONST) { |
2324 | j = po->operand[1].val; |
2325 | j %= lmod_bytes(po, po->operand[0].lmod) * 8; |
2326 | fprintf(fout, po->op == OP_ROL ? |
2327 | " %s = (%s << %d) | (%s >> %d);" : |
2328 | " %s = (%s >> %d) | (%s << %d);", |
2329 | buf1, buf1, j, buf1, |
2330 | lmod_bytes(po, po->operand[0].lmod) * 8 - j); |
2331 | } |
2332 | else |
2333 | ferr(po, "TODO\n"); |
2334 | last_arith_dst = &po->operand[0]; |
2335 | delayed_flag_op = NULL; |
2336 | break; |
2337 | |
5101a5f9 |
2338 | case OP_XOR: |
850c9265 |
2339 | assert_operand_cnt(2); |
2340 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
5101a5f9 |
2341 | if (IS(opr_name(po, 0), opr_name(po, 1))) { |
2342 | // special case for XOR |
2343 | fprintf(fout, " %s = 0;", |
2344 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0])); |
2345 | last_arith_dst = &po->operand[0]; |
2346 | delayed_flag_op = NULL; |
850c9265 |
2347 | break; |
850c9265 |
2348 | } |
5101a5f9 |
2349 | goto dualop_arith; |
2350 | |
5101a5f9 |
2351 | case OP_ADC: |
850c9265 |
2352 | case OP_SBB: |
5101a5f9 |
2353 | assert_operand_cnt(2); |
2354 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
2355 | fprintf(fout, " %s %s= %s + cond_c;", |
2356 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2357 | op_to_c(po), |
2358 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2359 | last_arith_dst = &po->operand[0]; |
2360 | delayed_flag_op = NULL; |
850c9265 |
2361 | break; |
2362 | |
2363 | case OP_INC: |
2364 | case OP_DEC: |
2365 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
1bafb621 |
2366 | if (po->operand[0].type == OPT_REG) { |
2367 | strcpy(buf2, po->op == OP_INC ? "++" : "--"); |
2368 | fprintf(fout, " %s%s;", buf1, buf2); |
2369 | } |
2370 | else { |
2371 | strcpy(buf2, po->op == OP_INC ? "+" : "-"); |
2372 | fprintf(fout, " %s %s= 1;", buf1, buf2); |
2373 | } |
5101a5f9 |
2374 | last_arith_dst = &po->operand[0]; |
2375 | delayed_flag_op = NULL; |
2376 | break; |
2377 | |
2378 | case OP_NEG: |
2379 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
2380 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[0], 0); |
2381 | fprintf(fout, " %s = -%s%s;", buf1, |
2382 | lmod_cast_s(po, po->operand[0].lmod), buf2); |
850c9265 |
2383 | last_arith_dst = &po->operand[0]; |
69a3cdfc |
2384 | delayed_flag_op = NULL; |
940e8e66 |
2385 | if (pfomask & (1 << PFO_C)) { |
2386 | fprintf(fout, "\n cond_c = (%s != 0);", buf1); |
2387 | pfomask &= ~(1 << PFO_C); |
2388 | } |
850c9265 |
2389 | break; |
2390 | |
2391 | case OP_IMUL: |
2392 | if (po->operand_cnt == 2) |
2393 | goto dualop_arith; |
87bf6cec |
2394 | if (po->operand_cnt == 3) |
2395 | ferr(po, "TODO imul3\n"); |
2396 | // fallthrough |
2397 | case OP_MUL: |
2398 | assert_operand_cnt(1); |
2399 | strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)"); |
2400 | fprintf(fout, " mul_tmp = %seax * %s%s;\n", buf1, buf1, |
2401 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[0], 0)); |
2402 | fprintf(fout, " edx = mul_tmp >> 32;\n"); |
2403 | fprintf(fout, " eax = mul_tmp;"); |
2404 | last_arith_dst = NULL; |
69a3cdfc |
2405 | delayed_flag_op = NULL; |
91977a1c |
2406 | break; |
2407 | |
5101a5f9 |
2408 | case OP_DIV: |
2409 | case OP_IDIV: |
2410 | assert_operand_cnt(1); |
2411 | if (po->operand[0].lmod != OPLM_DWORD) |
2412 | ferr(po, "unhandled lmod %d\n", po->operand[0].lmod); |
2413 | |
2414 | // 32bit division is common, look for it |
64c59faf |
2415 | if (po->op == OP_DIV) |
2416 | ret = scan_for_reg_clear(i - 1, xDX); |
2417 | else |
2418 | ret = scan_for_cdq_edx(i - 1); |
2419 | if (ret >= 0) { |
5101a5f9 |
2420 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
2421 | strcpy(buf2, lmod_cast(po, po->operand[0].lmod, |
2422 | po->op == OP_IDIV)); |
2423 | fprintf(fout, " edx = %seax %% %s%s;\n", buf2, buf2, buf1); |
2424 | fprintf(fout, " eax = %seax / %s%s;", buf2, buf2, buf1); |
2425 | } |
2426 | else |
2427 | ferr(po, "TODO 64bit divident\n"); |
87bf6cec |
2428 | last_arith_dst = NULL; |
2429 | delayed_flag_op = NULL; |
5101a5f9 |
2430 | break; |
2431 | |
91977a1c |
2432 | case OP_TEST: |
2433 | case OP_CMP: |
850c9265 |
2434 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
940e8e66 |
2435 | if (pfomask != 0) { |
69a3cdfc |
2436 | for (j = 0; j < 8; j++) { |
940e8e66 |
2437 | if (pfomask & (1 << j)) { |
69a3cdfc |
2438 | out_cmp_test(buf1, sizeof(buf1), po, j, 0); |
2439 | fprintf(fout, " cond_%s = %s;", |
2440 | parsed_flag_op_names[j], buf1); |
2441 | } |
2442 | } |
940e8e66 |
2443 | pfomask = 0; |
69a3cdfc |
2444 | } |
2445 | else |
2446 | no_output = 1; |
2447 | delayed_flag_op = po; |
91977a1c |
2448 | break; |
2449 | |
69a3cdfc |
2450 | // note: we reuse OP_Jcc for SETcc, only flags differ |
91977a1c |
2451 | case OP_JO ... OP_JG: |
5101a5f9 |
2452 | if (po->flags & OPF_JMP) |
850c9265 |
2453 | fprintf(fout, " goto %s;", po->operand[0].name); |
5101a5f9 |
2454 | // else SETcc - should already be handled |
850c9265 |
2455 | break; |
2456 | |
2457 | case OP_JMP: |
87bf6cec |
2458 | assert_operand_cnt(1); |
2459 | if (po->operand[0].type != OPT_LABEL) |
2460 | ferr(po, "unhandled call type\n"); |
2461 | |
850c9265 |
2462 | fprintf(fout, " goto %s;", po->operand[0].name); |
91977a1c |
2463 | break; |
2464 | |
2465 | case OP_CALL: |
5101a5f9 |
2466 | assert_operand_cnt(1); |
850c9265 |
2467 | pp = po->datap; |
91977a1c |
2468 | if (pp == NULL) |
850c9265 |
2469 | ferr(po, "NULL pp\n"); |
91977a1c |
2470 | |
1bafb621 |
2471 | if (po->operand[0].type != OPT_LABEL) |
2472 | fprintf(fout, " icall%d = (void *)%s;\n", i, |
2473 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0)); |
2474 | |
91977a1c |
2475 | fprintf(fout, " "); |
2476 | if (!IS(pp->ret_type, "void")) { |
87bf6cec |
2477 | if (po->flags & OPF_TAIL) |
2478 | fprintf(fout, "return "); |
2479 | else |
2480 | fprintf(fout, "eax = "); |
91977a1c |
2481 | if (strchr(pp->ret_type, '*')) |
2482 | fprintf(fout, "(u32)"); |
2483 | } |
87bf6cec |
2484 | |
39b168b8 |
2485 | if (po->operand[0].type != OPT_LABEL) { |
1bafb621 |
2486 | fprintf(fout, "icall%d(", i); |
39b168b8 |
2487 | } |
2488 | else { |
2489 | if (pp->name[0] == 0) |
2490 | ferr(po, "missing pp->name\n"); |
2491 | fprintf(fout, "%s(", pp->name); |
2492 | } |
2493 | |
91977a1c |
2494 | for (arg = 0; arg < pp->argc; arg++) { |
2495 | if (arg > 0) |
2496 | fprintf(fout, ", "); |
87bf6cec |
2497 | |
2498 | if (strchr(pp->arg[arg].type, '*')) |
2499 | fprintf(fout, "(%s)", pp->arg[arg].type); |
2500 | |
91977a1c |
2501 | if (pp->arg[arg].reg != NULL) { |
850c9265 |
2502 | fprintf(fout, "%s", pp->arg[arg].reg); |
91977a1c |
2503 | continue; |
2504 | } |
2505 | |
2506 | // stack arg |
2507 | tmp_op = pp->arg[arg].datap; |
2508 | if (tmp_op == NULL) |
850c9265 |
2509 | ferr(po, "parsed_op missing for arg%d\n", arg); |
940e8e66 |
2510 | if (tmp_op->argmask) { |
2511 | fprintf(fout, "s_a%d", arg + 1); |
69a3cdfc |
2512 | } |
2513 | else { |
2514 | fprintf(fout, "%s", |
2515 | out_src_opr(buf1, sizeof(buf1), |
2516 | tmp_op, &tmp_op->operand[0], 0)); |
2517 | } |
91977a1c |
2518 | } |
2519 | fprintf(fout, ");"); |
87bf6cec |
2520 | |
2521 | if (po->flags & OPF_TAIL) { |
2522 | strcpy(g_comment, "tailcall"); |
2523 | if (IS(pp->ret_type, "void")) |
2524 | fprintf(fout, "\n return;"); |
2525 | } |
2526 | delayed_flag_op = NULL; |
2527 | last_arith_dst = NULL; |
91977a1c |
2528 | break; |
2529 | |
2530 | case OP_RET: |
2531 | if (IS(g_func_pp.ret_type, "void")) |
2532 | fprintf(fout, " return;"); |
d4e3b5db |
2533 | else if (strchr(g_func_pp.ret_type, '*')) |
2534 | fprintf(fout, " return (%s)eax;", |
2535 | g_func_pp.ret_type); |
91977a1c |
2536 | else |
2537 | fprintf(fout, " return eax;"); |
2538 | break; |
2539 | |
2540 | case OP_PUSH: |
940e8e66 |
2541 | if (po->argmask) { |
69a3cdfc |
2542 | // special case - saved func arg |
d4e3b5db |
2543 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
940e8e66 |
2544 | for (j = 0; j < 32; j++) { |
d4e3b5db |
2545 | if (po->argmask & (1 << j)) |
2546 | fprintf(fout, " s_a%d = %s;", j + 1, buf1); |
940e8e66 |
2547 | } |
69a3cdfc |
2548 | break; |
2549 | } |
d4e3b5db |
2550 | else if (po->flags & OPF_RSAVE) { |
2551 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
2552 | fprintf(fout, " s_%s = %s;", buf1, buf1); |
2553 | break; |
2554 | } |
2555 | ferr(po, "stray push encountered\n"); |
91977a1c |
2556 | break; |
2557 | |
2558 | case OP_POP: |
d4e3b5db |
2559 | if (po->flags & OPF_RSAVE) { |
2560 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
2561 | fprintf(fout, " %s = s_%s;", buf1, buf1); |
2562 | break; |
2563 | } |
2564 | ferr(po, "stray pop encountered\n"); |
91977a1c |
2565 | break; |
2566 | |
33c35af6 |
2567 | case OP_NOP: |
2568 | break; |
2569 | |
91977a1c |
2570 | default: |
2571 | no_output = 1; |
69a3cdfc |
2572 | ferr(po, "unhandled op type %d, flags %x\n", |
2573 | po->op, po->flags); |
91977a1c |
2574 | break; |
2575 | } |
2576 | |
2577 | if (g_comment[0] != 0) { |
2578 | fprintf(fout, " // %s", g_comment); |
2579 | g_comment[0] = 0; |
2580 | no_output = 0; |
2581 | } |
2582 | if (!no_output) |
2583 | fprintf(fout, "\n"); |
5101a5f9 |
2584 | |
940e8e66 |
2585 | if (pfomask != 0) |
2586 | ferr(po, "missed flag calc, pfomask=%x\n", pfomask); |
2587 | |
5101a5f9 |
2588 | // see is delayed flag stuff is still valid |
2589 | if (delayed_flag_op != NULL && delayed_flag_op != po) { |
2590 | if (is_any_opr_modified(delayed_flag_op, po)) |
2591 | delayed_flag_op = NULL; |
2592 | } |
2593 | |
2594 | if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) { |
2595 | if (is_opr_modified(last_arith_dst, po)) |
2596 | last_arith_dst = NULL; |
2597 | } |
91977a1c |
2598 | } |
2599 | |
2600 | fprintf(fout, "}\n\n"); |
2601 | |
2602 | // cleanup |
2603 | for (i = 0; i < opcnt; i++) { |
2604 | if (ops[i].op == OP_CALL) { |
2605 | pp = ops[i].datap; |
2606 | if (pp) { |
2607 | proto_release(pp); |
2608 | free(pp); |
2609 | } |
2610 | } |
2611 | } |
2612 | proto_release(&g_func_pp); |
2613 | } |
c36e914d |
2614 | |
d4e3b5db |
2615 | static void set_label(int i, const char *name) |
2616 | { |
2617 | const char *p; |
2618 | int len; |
2619 | |
2620 | len = strlen(name); |
2621 | p = strchr(name, ':'); |
2622 | if (p != NULL) |
2623 | len = p - name; |
2624 | |
2625 | if (len > sizeof(g_labels[0]) - 1) |
2626 | aerr("label '%s' too long: %d\n", name, len); |
2627 | if (g_labels[i][0] != 0) |
2628 | aerr("dupe label '%s'?\n", name); |
2629 | memcpy(g_labels[i], name, len); |
2630 | g_labels[i][len] = 0; |
2631 | } |
2632 | |
bfa4a6ee |
2633 | // '=' needs special treatment.. |
2634 | static char *next_word_s(char *w, size_t wsize, char *s) |
2635 | { |
2636 | size_t i; |
2637 | |
2638 | s = sskip(s); |
2639 | |
2640 | for (i = 0; i < wsize - 1; i++) { |
2641 | if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0)) |
2642 | break; |
2643 | w[i] = s[i]; |
2644 | } |
2645 | w[i] = 0; |
2646 | |
2647 | if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=') |
2648 | printf("warning: '%s' truncated\n", w); |
2649 | |
2650 | return s + i; |
2651 | } |
2652 | |
2653 | static int cmpstringp(const void *p1, const void *p2) |
2654 | { |
2655 | return strcmp(*(char * const *)p1, *(char * const *)p2); |
2656 | } |
2657 | |
91977a1c |
2658 | int main(int argc, char *argv[]) |
2659 | { |
06c5d854 |
2660 | FILE *fout, *fasm, *frlist; |
91977a1c |
2661 | char line[256]; |
2662 | char words[16][256]; |
1bafb621 |
2663 | int ida_func_attr = 0; |
91977a1c |
2664 | int in_func = 0; |
bfa4a6ee |
2665 | int skip_func = 0; |
940e8e66 |
2666 | int skip_warned = 0; |
91977a1c |
2667 | int eq_alloc; |
bfa4a6ee |
2668 | char **rlist = NULL; |
2669 | int rlist_len = 0; |
2670 | int rlist_alloc = 0; |
2671 | int verbose = 0; |
2672 | int arg_out; |
2673 | int arg = 1; |
91977a1c |
2674 | int pi = 0; |
1bafb621 |
2675 | int i, len; |
91977a1c |
2676 | char *p; |
2677 | int wordc; |
2678 | |
bfa4a6ee |
2679 | if (argv[1] && IS(argv[1], "-v")) { |
2680 | verbose = 1; |
2681 | arg++; |
2682 | } |
2683 | |
2684 | if (argc < arg + 3) { |
2685 | printf("usage:\n%s [-v] <.c> <.asm> <hdrf> [rlist]*\n", |
91977a1c |
2686 | argv[0]); |
2687 | return 1; |
2688 | } |
2689 | |
bfa4a6ee |
2690 | arg_out = arg++; |
91977a1c |
2691 | |
bfa4a6ee |
2692 | asmfn = argv[arg++]; |
91977a1c |
2693 | fasm = fopen(asmfn, "r"); |
2694 | my_assert_not(fasm, NULL); |
2695 | |
bfa4a6ee |
2696 | hdrfn = argv[arg++]; |
06c5d854 |
2697 | g_fhdr = fopen(hdrfn, "r"); |
2698 | my_assert_not(g_fhdr, NULL); |
bfa4a6ee |
2699 | |
2700 | rlist_alloc = 64; |
2701 | rlist = malloc(rlist_alloc * sizeof(rlist[0])); |
2702 | my_assert_not(rlist, NULL); |
2703 | // needs special handling.. |
2704 | rlist[rlist_len++] = "__alloca_probe"; |
2705 | |
2706 | for (; arg < argc; arg++) { |
2707 | frlist = fopen(argv[arg], "r"); |
2708 | my_assert_not(frlist, NULL); |
2709 | |
2710 | while (fgets(line, sizeof(line), frlist)) { |
2711 | p = sskip(line); |
2712 | if (*p == 0 || *p == ';' || *p == '#') |
2713 | continue; |
2714 | |
2715 | p = next_word(words[0], sizeof(words[0]), p); |
2716 | if (words[0][0] == 0) |
2717 | continue; |
2718 | |
2719 | if (rlist_len >= rlist_alloc) { |
2720 | rlist_alloc = rlist_alloc * 2 + 64; |
2721 | rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0])); |
2722 | my_assert_not(rlist, NULL); |
2723 | } |
2724 | rlist[rlist_len++] = strdup(words[0]); |
2725 | } |
2726 | |
2727 | fclose(frlist); |
2728 | frlist = NULL; |
2729 | } |
2730 | |
2731 | if (rlist_len > 0) |
2732 | qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp); |
2733 | |
2734 | fout = fopen(argv[arg_out], "w"); |
91977a1c |
2735 | my_assert_not(fout, NULL); |
2736 | |
2737 | eq_alloc = 128; |
2738 | g_eqs = malloc(eq_alloc * sizeof(g_eqs[0])); |
2739 | my_assert_not(g_eqs, NULL); |
2740 | |
2741 | while (fgets(line, sizeof(line), fasm)) |
2742 | { |
2743 | asmln++; |
2744 | |
2745 | p = sskip(line); |
1bafb621 |
2746 | if (*p == 0) |
91977a1c |
2747 | continue; |
2748 | |
1bafb621 |
2749 | if (*p == ';') { |
2750 | static const char *attrs[] = { |
2751 | "bp-based frame", |
2752 | "library function", |
2753 | "static", |
2754 | "noreturn", |
2755 | "thunk", |
2756 | "fpd=", |
2757 | }; |
2758 | if (p[2] != 'A' || strncmp(p, "; Attributes:", 13) != 0) |
2759 | continue; |
2760 | |
2761 | // parse IDA's attribute-list comment |
2762 | ida_func_attr = 0; |
2763 | p = sskip(p + 13); |
2764 | // get rid of random tabs |
2765 | for (i = 0; p[i] != 0; i++) |
2766 | if (p[i] == '\t') |
2767 | p[i] = ' '; |
2768 | |
2769 | for (; *p != 0; p = sskip(p)) { |
2770 | for (i = 0; i < ARRAY_SIZE(attrs); i++) { |
2771 | if (!strncmp(p, attrs[i], strlen(attrs[i]))) { |
2772 | ida_func_attr |= 1 << i; |
2773 | p += strlen(attrs[i]); |
2774 | break; |
2775 | } |
2776 | } |
2777 | if (i == ARRAY_SIZE(attrs)) { |
2778 | anote("unparsed IDA attr: %s\n", p); |
2779 | break; |
2780 | } |
2781 | if (IS(attrs[i], "fpd=")) { |
2782 | p = next_word(words[0], sizeof(words[0]), p); |
2783 | // ignore for now.. |
2784 | } |
2785 | } |
2786 | continue; |
2787 | } |
2788 | |
06c5d854 |
2789 | parse_words: |
91977a1c |
2790 | memset(words, 0, sizeof(words)); |
2791 | for (wordc = 0; wordc < 16; wordc++) { |
bfa4a6ee |
2792 | p = sskip(next_word_s(words[wordc], sizeof(words[0]), p)); |
91977a1c |
2793 | if (*p == 0 || *p == ';') { |
2794 | wordc++; |
2795 | break; |
2796 | } |
2797 | } |
2798 | |
06c5d854 |
2799 | // alow asm patches in comments |
2800 | if (*p == ';' && IS_START(p, "; sctpatch: ")) { |
2801 | p = sskip(p + 12); |
2802 | if (*p == 0 || *p == ';') |
2803 | continue; |
2804 | goto parse_words; // lame |
2805 | } |
2806 | |
91977a1c |
2807 | if (wordc == 0) { |
2808 | // shouldn't happen |
2809 | awarn("wordc == 0?\n"); |
2810 | continue; |
2811 | } |
2812 | |
2813 | // don't care about this: |
2814 | if (words[0][0] == '.' |
2815 | || IS(words[0], "include") |
2816 | || IS(words[0], "assume") || IS(words[1], "segment") |
2817 | || IS(words[0], "align")) |
2818 | { |
2819 | continue; |
2820 | } |
2821 | |
2822 | if (IS(words[1], "proc")) { |
2823 | if (in_func) |
2824 | aerr("proc '%s' while in_func '%s'?\n", |
2825 | words[0], g_func); |
bfa4a6ee |
2826 | p = words[0]; |
1bafb621 |
2827 | if ((ida_func_attr & IDAFA_THUNK) |
2828 | || bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp)) |
bfa4a6ee |
2829 | skip_func = 1; |
91977a1c |
2830 | strcpy(g_func, words[0]); |
d4e3b5db |
2831 | set_label(0, words[0]); |
91977a1c |
2832 | in_func = 1; |
2833 | continue; |
2834 | } |
2835 | |
2836 | if (IS(words[1], "endp")) { |
2837 | if (!in_func) |
2838 | aerr("endp '%s' while not in_func?\n", words[0]); |
2839 | if (!IS(g_func, words[0])) |
2840 | aerr("endp '%s' while in_func '%s'?\n", |
2841 | words[0], g_func); |
bfa4a6ee |
2842 | |
2843 | if (in_func && !skip_func) |
06c5d854 |
2844 | gen_func(fout, g_fhdr, g_func, pi); |
bfa4a6ee |
2845 | |
91977a1c |
2846 | in_func = 0; |
940e8e66 |
2847 | skip_warned = 0; |
bfa4a6ee |
2848 | skip_func = 0; |
91977a1c |
2849 | g_func[0] = 0; |
2850 | if (pi != 0) { |
2851 | memset(&ops, 0, pi * sizeof(ops[0])); |
2852 | memset(g_labels, 0, pi * sizeof(g_labels[0])); |
940e8e66 |
2853 | memset(g_label_refs, 0, pi * sizeof(g_label_refs[0])); |
91977a1c |
2854 | pi = 0; |
2855 | } |
2856 | g_eqcnt = 0; |
1bafb621 |
2857 | ida_func_attr = 0; |
91977a1c |
2858 | continue; |
2859 | } |
2860 | |
bfa4a6ee |
2861 | p = strchr(words[0], ':'); |
2862 | if (p != NULL) { |
d4e3b5db |
2863 | set_label(pi, words[0]); |
bfa4a6ee |
2864 | continue; |
2865 | } |
2866 | |
2867 | if (!in_func || skip_func) { |
2868 | if (!skip_warned && !skip_func && g_labels[pi][0] != 0) { |
2869 | if (verbose) |
2870 | anote("skipping from '%s'\n", g_labels[pi]); |
2871 | skip_warned = 1; |
2872 | } |
2873 | g_labels[pi][0] = 0; |
2874 | continue; |
2875 | } |
2876 | |
91977a1c |
2877 | if (IS(words[1], "=")) { |
2878 | if (wordc != 5) |
2879 | aerr("unhandled equ, wc=%d\n", wordc); |
2880 | if (g_eqcnt >= eq_alloc) { |
2881 | eq_alloc *= 2; |
2882 | g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0])); |
2883 | my_assert_not(g_eqs, NULL); |
2884 | } |
2885 | |
2886 | len = strlen(words[0]); |
2887 | if (len > sizeof(g_eqs[0].name) - 1) |
2888 | aerr("equ name too long: %d\n", len); |
2889 | strcpy(g_eqs[g_eqcnt].name, words[0]); |
2890 | |
2891 | if (!IS(words[3], "ptr")) |
2892 | aerr("unhandled equ\n"); |
2893 | if (IS(words[2], "dword")) |
2894 | g_eqs[g_eqcnt].lmod = OPLM_DWORD; |
2895 | else if (IS(words[2], "word")) |
2896 | g_eqs[g_eqcnt].lmod = OPLM_WORD; |
2897 | else if (IS(words[2], "byte")) |
2898 | g_eqs[g_eqcnt].lmod = OPLM_BYTE; |
2899 | else |
2900 | aerr("bad lmod: '%s'\n", words[2]); |
2901 | |
2902 | g_eqs[g_eqcnt].offset = parse_number(words[4]); |
2903 | g_eqcnt++; |
2904 | continue; |
2905 | } |
2906 | |
2907 | if (pi >= ARRAY_SIZE(ops)) |
2908 | aerr("too many ops\n"); |
2909 | |
91977a1c |
2910 | parse_op(&ops[pi], words, wordc); |
2911 | pi++; |
91977a1c |
2912 | } |
2913 | |
2914 | fclose(fout); |
2915 | fclose(fasm); |
06c5d854 |
2916 | fclose(g_fhdr); |
91977a1c |
2917 | |
2918 | return 0; |
c36e914d |
2919 | } |
91977a1c |
2920 | |
2921 | // vim:ts=2:shiftwidth=2:expandtab |