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 |
4c45fa73 |
132 | int bt_i; // branch target for branches |
133 | struct parsed_data *btj;// branch targets for jumptables |
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 | |
4c45fa73 |
147 | struct parsed_data { |
148 | char label[256]; |
149 | enum opr_type type; |
150 | enum opr_lenmod lmod; |
151 | int count; |
152 | int count_alloc; |
153 | struct { |
154 | union { |
155 | char *label; |
156 | int val; |
157 | } u; |
158 | int bt_i; |
159 | } *d; |
160 | }; |
161 | |
162 | struct label_ref { |
163 | int i; |
164 | struct label_ref *next; |
165 | }; |
166 | |
1bafb621 |
167 | enum ida_func_attr { |
168 | IDAFA_BP_FRAME = (1 << 0), |
169 | IDAFA_LIB_FUNC = (1 << 1), |
170 | IDAFA_STATIC = (1 << 2), |
171 | IDAFA_NORETURN = (1 << 3), |
172 | IDAFA_THUNK = (1 << 4), |
173 | IDAFA_FPD = (1 << 5), |
174 | }; |
175 | |
176 | #define MAX_OPS 4096 |
c36e914d |
177 | |
178 | static struct parsed_op ops[MAX_OPS]; |
91977a1c |
179 | static struct parsed_equ *g_eqs; |
180 | static int g_eqcnt; |
181 | static char g_labels[MAX_OPS][32]; |
4c45fa73 |
182 | static struct label_ref g_label_refs[MAX_OPS]; |
91977a1c |
183 | static struct parsed_proto g_func_pp; |
4c45fa73 |
184 | static struct parsed_data *g_func_pd; |
185 | static int g_func_pd_cnt; |
91977a1c |
186 | static char g_func[256]; |
187 | static char g_comment[256]; |
188 | static int g_bp_frame; |
1bafb621 |
189 | static int g_sp_frame; |
190 | static int g_stack_fsz; |
4c45fa73 |
191 | static int g_ida_func_attr; |
91977a1c |
192 | #define ferr(op_, fmt, ...) do { \ |
850c9265 |
193 | printf("error:%s:#%ld: '%s': " fmt, g_func, (op_) - ops, \ |
194 | dump_op(op_), ##__VA_ARGS__); \ |
33c35af6 |
195 | fcloseall(); \ |
91977a1c |
196 | exit(1); \ |
197 | } while (0) |
198 | |
199 | #define MAX_REGS 8 |
200 | |
201 | const char *regs_r32[] = { "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp" }; |
202 | const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" }; |
203 | const char *regs_r8l[] = { "al", "bl", "cl", "dl" }; |
204 | const char *regs_r8h[] = { "ah", "bh", "ch", "dh" }; |
205 | |
206 | enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP }; |
207 | |
69a3cdfc |
208 | // possible basic comparison types (without inversion) |
209 | enum parsed_flag_op { |
210 | PFO_O, // 0 OF=1 |
211 | PFO_C, // 2 CF=1 |
212 | PFO_Z, // 4 ZF=1 |
213 | PFO_BE, // 6 CF=1||ZF=1 |
214 | PFO_S, // 8 SF=1 |
215 | PFO_P, // a PF=1 |
216 | PFO_L, // c SF!=OF |
217 | PFO_LE, // e ZF=1||SF!=OF |
218 | }; |
219 | |
220 | static const char *parsed_flag_op_names[] = { |
221 | "o", "c", "z", "be", "s", "p", "l", "le" |
222 | }; |
223 | |
91977a1c |
224 | static int char_array_i(const char *array[], size_t len, const char *s) |
225 | { |
226 | int i; |
c36e914d |
227 | |
91977a1c |
228 | for (i = 0; i < len; i++) |
229 | if (IS(s, array[i])) |
230 | return i; |
c36e914d |
231 | |
91977a1c |
232 | return -1; |
233 | } |
234 | |
5101a5f9 |
235 | static void printf_number(char *buf, size_t buf_size, long number) |
91977a1c |
236 | { |
5101a5f9 |
237 | // output in C-friendly form |
238 | snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number); |
239 | } |
91977a1c |
240 | |
5101a5f9 |
241 | static int parse_reg(enum opr_lenmod *reg_lmod, const char *s) |
242 | { |
243 | int reg; |
91977a1c |
244 | |
5101a5f9 |
245 | reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s); |
246 | if (reg >= 0) { |
247 | *reg_lmod = OPLM_DWORD; |
248 | return reg; |
91977a1c |
249 | } |
5101a5f9 |
250 | reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s); |
251 | if (reg >= 0) { |
252 | *reg_lmod = OPLM_WORD; |
253 | return reg; |
254 | } |
255 | reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s); |
256 | if (reg >= 0) { |
257 | *reg_lmod = OPLM_BYTE; |
258 | return reg; |
259 | } |
260 | reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s); |
261 | if (reg >= 0) { |
262 | *reg_lmod = OPLM_BYTE; |
263 | return reg; |
850c9265 |
264 | } |
265 | |
266 | return -1; |
91977a1c |
267 | } |
268 | |
269 | static long parse_number(const char *number) |
270 | { |
271 | int len = strlen(number); |
272 | const char *p = number; |
273 | char *endp = NULL; |
274 | int neg = 0; |
275 | int bad; |
276 | long ret; |
277 | |
278 | if (*p == '-') { |
279 | neg = 1; |
280 | p++; |
281 | } |
282 | if (len > 1 && *p == '0') |
283 | p++; |
284 | if (number[len - 1] == 'h') { |
285 | ret = strtol(p, &endp, 16); |
286 | bad = (*endp != 'h'); |
287 | } |
288 | else { |
289 | ret = strtol(p, &endp, 10); |
290 | bad = (*endp != 0); |
291 | } |
292 | if (bad) |
293 | aerr("number parsing failed\n"); |
294 | if (neg) |
295 | ret = -ret; |
296 | return ret; |
297 | } |
298 | |
5101a5f9 |
299 | static int parse_indmode(char *name, int *regmask, int need_c_cvt) |
300 | { |
301 | enum opr_lenmod lmod; |
302 | char cvtbuf[256]; |
303 | char *d = cvtbuf; |
304 | char *s = name; |
305 | char w[64]; |
306 | long number; |
307 | int reg; |
308 | int c = 0; |
309 | |
310 | *d = 0; |
311 | |
312 | while (*s != 0) { |
313 | d += strlen(d); |
314 | while (my_isblank(*s)) |
315 | s++; |
316 | for (; my_issep(*s); d++, s++) |
317 | *d = *s; |
318 | while (my_isblank(*s)) |
319 | s++; |
320 | *d = 0; |
321 | |
87bf6cec |
322 | // skip 'ds:' prefix |
39b168b8 |
323 | if (IS_START(s, "ds:")) |
87bf6cec |
324 | s += 3; |
325 | |
5101a5f9 |
326 | s = next_idt(w, sizeof(w), s); |
327 | if (w[0] == 0) |
328 | break; |
329 | c++; |
330 | |
331 | reg = parse_reg(&lmod, w); |
332 | if (reg >= 0) { |
333 | *regmask |= 1 << reg; |
334 | goto pass; |
335 | } |
336 | |
337 | if ('0' <= w[0] && w[0] <= '9') { |
338 | number = parse_number(w); |
339 | printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number); |
340 | continue; |
341 | } |
342 | |
343 | // probably some label/identifier - pass |
344 | |
345 | pass: |
346 | snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w); |
347 | } |
348 | |
349 | if (need_c_cvt) |
350 | strcpy(name, cvtbuf); |
351 | |
352 | return c; |
353 | } |
354 | |
4c45fa73 |
355 | static int is_reg_in_str(const char *s) |
356 | { |
357 | int i; |
358 | |
359 | if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3]))) |
360 | return 0; |
361 | |
362 | for (i = 0; i < ARRAY_SIZE(regs_r32); i++) |
363 | if (!strncmp(s, regs_r32[i], 3)) |
364 | return 1; |
365 | |
366 | return 0; |
367 | } |
368 | |
369 | static const char *parse_stack_el(const char *name, char *extra_reg) |
d4e3b5db |
370 | { |
4c45fa73 |
371 | const char *p, *p2, *s; |
1bafb621 |
372 | char *endp = NULL; |
d4e3b5db |
373 | char buf[32]; |
1bafb621 |
374 | long val; |
d4e3b5db |
375 | int len; |
376 | |
4c45fa73 |
377 | p = name; |
378 | if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) { |
379 | p += 4; |
380 | if (extra_reg != NULL) { |
381 | strncpy(extra_reg, name, 3); |
382 | extra_reg[4] = 0; |
383 | } |
d4e3b5db |
384 | } |
4c45fa73 |
385 | |
386 | if (IS_START(p, "ebp+")) { |
387 | p += 4; |
388 | |
389 | p2 = strchr(p, '+'); |
390 | if (p2 != NULL && is_reg_in_str(p)) { |
391 | if (extra_reg != NULL) { |
392 | strncpy(extra_reg, p, p2 - p); |
393 | extra_reg[p2 - p] = 0; |
394 | } |
395 | p = p2 + 1; |
396 | } |
397 | |
398 | if (!('0' <= *p && *p <= '9')) |
399 | return p; |
400 | |
401 | return NULL; |
402 | } |
403 | |
39b168b8 |
404 | if (!IS_START(name, "esp+")) |
d4e3b5db |
405 | return NULL; |
406 | |
407 | p = strchr(name + 4, '+'); |
408 | if (p) { |
1bafb621 |
409 | // must be a number after esp+, already converted to 0x.. |
410 | s = name + 4; |
411 | if (!('0' <= *s && *s <= '9')) { |
412 | aerr("%s nan?\n", __func__); |
d4e3b5db |
413 | return NULL; |
1bafb621 |
414 | } |
415 | if (s[0] == '0' && s[1] == 'x') |
416 | s += 2; |
417 | len = p - s; |
d4e3b5db |
418 | if (len < sizeof(buf) - 1) { |
1bafb621 |
419 | strncpy(buf, s, len); |
d4e3b5db |
420 | buf[len] = 0; |
1bafb621 |
421 | val = strtol(buf, &endp, 16); |
422 | if (val == 0 || *endp != 0) { |
423 | aerr("%s num parse fail for '%s'\n", __func__, buf); |
424 | return NULL; |
425 | } |
d4e3b5db |
426 | } |
427 | p++; |
428 | } |
429 | else |
430 | p = name + 4; |
431 | |
432 | if ('0' <= *p && *p <= '9') |
433 | return NULL; |
434 | |
435 | return p; |
436 | } |
437 | |
850c9265 |
438 | static int guess_lmod_from_name(struct parsed_opr *opr) |
439 | { |
440 | if (!strncmp(opr->name, "dword_", 6)) { |
441 | opr->lmod = OPLM_DWORD; |
442 | return 1; |
443 | } |
444 | if (!strncmp(opr->name, "word_", 5)) { |
445 | opr->lmod = OPLM_WORD; |
446 | return 1; |
447 | } |
448 | if (!strncmp(opr->name, "byte_", 5)) { |
449 | opr->lmod = OPLM_BYTE; |
450 | return 1; |
451 | } |
452 | return 0; |
453 | } |
454 | |
06c5d854 |
455 | static int guess_lmod_from_c_type(struct parsed_opr *opr, const char *c_type) |
456 | { |
457 | static const char *ptr_types[] = { |
458 | "LPCSTR", |
459 | }; |
460 | static const char *dword_types[] = { |
461 | "int", "_DWORD", "DWORD", "HANDLE", "HWND", "HMODULE", |
462 | }; |
463 | static const char *word_types[] = { |
464 | "__int16", "unsigned __int16", |
465 | }; |
466 | static const char *byte_types[] = { |
467 | "char", "__int8", "unsigned __int8", "BYTE", |
468 | }; |
469 | int i; |
470 | |
471 | if (strchr(c_type, '*')) { |
472 | opr->lmod = OPLM_DWORD; |
473 | opr->is_ptr = 1; |
474 | return 1; |
475 | } |
476 | |
477 | for (i = 0; i < ARRAY_SIZE(dword_types); i++) { |
478 | if (IS(c_type, dword_types[i])) { |
479 | opr->lmod = OPLM_DWORD; |
480 | return 1; |
481 | } |
482 | } |
483 | |
484 | for (i = 0; i < ARRAY_SIZE(ptr_types); i++) { |
485 | if (IS(c_type, ptr_types[i])) { |
486 | opr->lmod = OPLM_DWORD; |
487 | opr->is_ptr = 1; |
488 | return 1; |
489 | } |
490 | } |
491 | |
492 | for (i = 0; i < ARRAY_SIZE(word_types); i++) { |
493 | if (IS(c_type, word_types[i])) { |
494 | opr->lmod = OPLM_WORD; |
495 | return 1; |
496 | } |
497 | } |
498 | |
499 | for (i = 0; i < ARRAY_SIZE(byte_types); i++) { |
500 | if (IS(c_type, byte_types[i])) { |
501 | opr->lmod = OPLM_BYTE; |
502 | return 1; |
503 | } |
504 | } |
505 | |
506 | anote("unhandled C type '%s' for '%s'\n", c_type, opr->name); |
507 | return 0; |
508 | } |
509 | |
4c45fa73 |
510 | static enum opr_type lmod_from_directive(const char *d) |
511 | { |
512 | if (IS(d, "dd")) |
513 | return OPLM_DWORD; |
514 | else if (IS(d, "dw")) |
515 | return OPLM_WORD; |
516 | else if (IS(d, "db")) |
517 | return OPLM_BYTE; |
518 | |
519 | aerr("unhandled directive: '%s'\n", d); |
520 | return OPLM_UNSPEC; |
521 | } |
522 | |
5101a5f9 |
523 | static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod, |
524 | int *regmask) |
525 | { |
526 | opr->type = OPT_REG; |
527 | opr->reg = reg; |
528 | opr->lmod = lmod; |
529 | *regmask |= 1 << reg; |
530 | } |
531 | |
39b168b8 |
532 | static struct parsed_equ *equ_find(struct parsed_op *po, const char *name, |
533 | int *extra_offs); |
87bf6cec |
534 | |
69a3cdfc |
535 | static int parse_operand(struct parsed_opr *opr, |
536 | int *regmask, int *regmask_indirect, |
1bafb621 |
537 | char words[16][256], int wordc, int w, unsigned int op_flags) |
c36e914d |
538 | { |
06c5d854 |
539 | struct parsed_proto pp; |
850c9265 |
540 | enum opr_lenmod tmplmod; |
850c9265 |
541 | int ret, len; |
5101a5f9 |
542 | long number; |
1bafb621 |
543 | int wordc_in; |
39b168b8 |
544 | char *tmp; |
850c9265 |
545 | int i; |
c36e914d |
546 | |
1bafb621 |
547 | if (w >= wordc) |
548 | aerr("parse_operand w %d, wordc %d\n", w, wordc); |
c36e914d |
549 | |
1bafb621 |
550 | opr->reg = xUNSPEC; |
91977a1c |
551 | |
1bafb621 |
552 | for (i = w; i < wordc; i++) { |
553 | len = strlen(words[i]); |
554 | if (words[i][len - 1] == ',') { |
555 | words[i][len - 1] = 0; |
556 | wordc = i + 1; |
557 | break; |
558 | } |
559 | } |
c36e914d |
560 | |
1bafb621 |
561 | wordc_in = wordc - w; |
c36e914d |
562 | |
1bafb621 |
563 | if ((op_flags & OPF_JMP) && wordc_in > 0 |
564 | && !('0' <= words[w][0] && words[w][0] <= '9')) |
565 | { |
566 | const char *label = NULL; |
567 | |
568 | if (wordc_in == 3 && !strncmp(words[w], "near", 4) |
569 | && IS(words[w + 1], "ptr")) |
570 | label = words[w + 2]; |
571 | else if (wordc_in == 2 && IS(words[w], "short")) |
572 | label = words[w + 1]; |
573 | else if (wordc_in == 1 |
574 | && strchr(words[w], '[') == NULL |
575 | && parse_reg(&tmplmod, words[w]) < 0) |
576 | label = words[w]; |
577 | |
578 | if (label != NULL) { |
579 | opr->type = OPT_LABEL; |
39b168b8 |
580 | if (IS_START(label, "ds:")) |
581 | label += 3; |
1bafb621 |
582 | strcpy(opr->name, label); |
583 | return wordc; |
584 | } |
585 | } |
c36e914d |
586 | |
1bafb621 |
587 | if (wordc_in >= 3) { |
588 | if (IS(words[w + 1], "ptr")) { |
589 | if (IS(words[w], "dword")) |
590 | opr->lmod = OPLM_DWORD; |
591 | else if (IS(words[w], "word")) |
592 | opr->lmod = OPLM_WORD; |
593 | else if (IS(words[w], "byte")) |
594 | opr->lmod = OPLM_BYTE; |
595 | else |
596 | aerr("type parsing failed\n"); |
597 | w += 2; |
598 | wordc_in = wordc - w; |
599 | } |
600 | } |
601 | |
39b168b8 |
602 | if (wordc_in == 2) { |
603 | if (IS(words[w], "offset")) { |
604 | opr->type = OPT_OFFSET; |
605 | strcpy(opr->name, words[w + 1]); |
606 | return wordc; |
607 | } |
608 | if (IS(words[w], "(offset")) { |
609 | char *p = strchr(words[w + 1], ')'); |
610 | if (p == NULL) |
611 | aerr("parse of bracketed offset failed\n"); |
612 | *p = 0; |
613 | opr->type = OPT_OFFSET; |
614 | strcpy(opr->name, words[w + 1]); |
615 | return wordc; |
616 | } |
1bafb621 |
617 | } |
c36e914d |
618 | |
1bafb621 |
619 | if (wordc_in != 1) |
850c9265 |
620 | aerr("parse_operand 1 word expected\n"); |
c36e914d |
621 | |
39b168b8 |
622 | tmp = words[w]; |
623 | if (IS_START(tmp, "ds:")) |
624 | tmp += 3; |
625 | strcpy(opr->name, tmp); |
c36e914d |
626 | |
850c9265 |
627 | if (words[w][0] == '[') { |
628 | opr->type = OPT_REGMEM; |
629 | ret = sscanf(words[w], "[%[^]]]", opr->name); |
630 | if (ret != 1) |
631 | aerr("[] parse failure\n"); |
87bf6cec |
632 | |
5101a5f9 |
633 | parse_indmode(opr->name, regmask_indirect, 1); |
4c45fa73 |
634 | if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL)) { |
87bf6cec |
635 | // might be an equ |
d4e3b5db |
636 | struct parsed_equ *eq = |
4c45fa73 |
637 | equ_find(NULL, parse_stack_el(opr->name, NULL), &i); |
87bf6cec |
638 | if (eq) |
639 | opr->lmod = eq->lmod; |
640 | } |
850c9265 |
641 | return wordc; |
642 | } |
643 | else if (strchr(words[w], '[')) { |
644 | // label[reg] form |
645 | opr->type = OPT_REGMEM; |
646 | if (opr->lmod == OPLM_UNSPEC) |
647 | guess_lmod_from_name(opr); |
5101a5f9 |
648 | parse_indmode(strchr(words[w], '['), regmask_indirect, 0); |
850c9265 |
649 | return wordc; |
650 | } |
651 | else if (('0' <= words[w][0] && words[w][0] <= '9') |
652 | || words[w][0] == '-') |
653 | { |
5101a5f9 |
654 | number = parse_number(words[w]); |
91977a1c |
655 | opr->type = OPT_CONST; |
5101a5f9 |
656 | opr->val = number; |
657 | printf_number(opr->name, sizeof(opr->name), number); |
91977a1c |
658 | return wordc; |
850c9265 |
659 | } |
c36e914d |
660 | |
5101a5f9 |
661 | ret = parse_reg(&tmplmod, opr->name); |
662 | if (ret >= 0) { |
663 | setup_reg_opr(opr, ret, tmplmod, regmask); |
850c9265 |
664 | return wordc; |
665 | } |
666 | |
667 | // most likely var in data segment |
668 | opr->type = OPT_LABEL; |
06c5d854 |
669 | |
670 | ret = proto_parse(g_fhdr, opr->name, &pp); |
671 | if (ret == 0) { |
672 | if (pp.is_fptr) { |
673 | opr->lmod = OPLM_DWORD; |
674 | opr->is_ptr = 1; |
675 | } |
676 | else if (opr->lmod == OPLM_UNSPEC) |
677 | guess_lmod_from_c_type(opr, pp.ret_type); |
678 | } |
64c59faf |
679 | proto_release(&pp); |
06c5d854 |
680 | |
850c9265 |
681 | if (opr->lmod == OPLM_UNSPEC) |
682 | guess_lmod_from_name(opr); |
850c9265 |
683 | return wordc; |
c36e914d |
684 | } |
685 | |
33c35af6 |
686 | static const struct { |
687 | const char *name; |
688 | unsigned int flags; |
689 | } pref_table[] = { |
690 | { "rep", OPF_REP }, |
691 | }; |
692 | |
c36e914d |
693 | static const struct { |
850c9265 |
694 | const char *name; |
695 | enum op_op op; |
69a3cdfc |
696 | unsigned int minopr; |
697 | unsigned int maxopr; |
698 | unsigned int flags; |
c36e914d |
699 | } op_table[] = { |
33c35af6 |
700 | { "nop", OP_NOP, 0, 0, 0 }, |
69a3cdfc |
701 | { "push", OP_PUSH, 1, 1, 0 }, |
702 | { "pop", OP_POP, 1, 1, OPF_DATA }, |
703 | { "mov" , OP_MOV, 2, 2, OPF_DATA }, |
704 | { "lea", OP_LEA, 2, 2, OPF_DATA }, |
705 | { "movzx",OP_MOVZX, 2, 2, OPF_DATA }, |
706 | { "movsx",OP_MOVSX, 2, 2, OPF_DATA }, |
707 | { "not", OP_NOT, 1, 1, OPF_DATA }, |
5101a5f9 |
708 | { "cdq", OP_CDQ, 0, 0, OPF_DATA }, |
33c35af6 |
709 | { "stosb",OP_STOS, 0, 0, OPF_DATA }, |
710 | { "stosw",OP_STOS, 0, 0, OPF_DATA }, |
711 | { "stosd",OP_STOS, 0, 0, OPF_DATA }, |
d4e3b5db |
712 | { "movsb",OP_MOVS, 0, 0, OPF_DATA }, |
713 | { "movsw",OP_MOVS, 0, 0, OPF_DATA }, |
714 | { "movsd",OP_MOVS, 0, 0, OPF_DATA }, |
69a3cdfc |
715 | { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS }, |
716 | { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS }, |
717 | { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS }, |
718 | { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS }, |
719 | { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS }, |
720 | { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS }, |
721 | { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS }, |
722 | { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS }, |
723 | { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS }, |
d4e3b5db |
724 | { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS }, |
725 | { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS }, |
5101a5f9 |
726 | { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC }, |
69a3cdfc |
727 | { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC }, |
728 | { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS }, |
729 | { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS }, |
5101a5f9 |
730 | { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS }, |
731 | { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS }, |
69a3cdfc |
732 | { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS }, |
5101a5f9 |
733 | { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS }, |
734 | { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS }, |
69a3cdfc |
735 | { "test", OP_TEST, 2, 2, OPF_FLAGS }, |
736 | { "cmp", OP_CMP, 2, 2, OPF_FLAGS }, |
87bf6cec |
737 | { "retn", OP_RET, 0, 1, OPF_JMP|OPF_TAIL }, |
738 | { "call", OP_CALL, 1, 1, OPF_JMP|OPF_FLAGS }, |
69a3cdfc |
739 | { "jmp", OP_JMP, 1, 1, OPF_JMP }, |
740 | { "jo", OP_JO, 1, 1, OPF_JMP|OPF_CC }, // 70 OF=1 |
741 | { "jno", OP_JNO, 1, 1, OPF_JMP|OPF_CC }, // 71 OF=0 |
742 | { "jc", OP_JC, 1, 1, OPF_JMP|OPF_CC }, // 72 CF=1 |
743 | { "jb", OP_JC, 1, 1, OPF_JMP|OPF_CC }, // 72 |
744 | { "jnc", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 CF=0 |
33c35af6 |
745 | { "jnb", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 |
69a3cdfc |
746 | { "jae", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 |
747 | { "jz", OP_JZ, 1, 1, OPF_JMP|OPF_CC }, // 74 ZF=1 |
748 | { "je", OP_JZ, 1, 1, OPF_JMP|OPF_CC }, // 74 |
749 | { "jnz", OP_JNZ, 1, 1, OPF_JMP|OPF_CC }, // 75 ZF=0 |
750 | { "jne", OP_JNZ, 1, 1, OPF_JMP|OPF_CC }, // 75 |
751 | { "jbe", OP_JBE, 1, 1, OPF_JMP|OPF_CC }, // 76 CF=1 || ZF=1 |
752 | { "jna", OP_JBE, 1, 1, OPF_JMP|OPF_CC }, // 76 |
753 | { "ja", OP_JA, 1, 1, OPF_JMP|OPF_CC }, // 77 CF=0 && ZF=0 |
754 | { "jnbe", OP_JA, 1, 1, OPF_JMP|OPF_CC }, // 77 |
755 | { "js", OP_JS, 1, 1, OPF_JMP|OPF_CC }, // 78 SF=1 |
756 | { "jns", OP_JNS, 1, 1, OPF_JMP|OPF_CC }, // 79 SF=0 |
757 | { "jp", OP_JP, 1, 1, OPF_JMP|OPF_CC }, // 7a PF=1 |
758 | { "jpe", OP_JP, 1, 1, OPF_JMP|OPF_CC }, // 7a |
759 | { "jnp", OP_JNP, 1, 1, OPF_JMP|OPF_CC }, // 7b PF=0 |
760 | { "jpo", OP_JNP, 1, 1, OPF_JMP|OPF_CC }, // 7b |
761 | { "jl", OP_JL, 1, 1, OPF_JMP|OPF_CC }, // 7c SF!=OF |
762 | { "jnge", OP_JL, 1, 1, OPF_JMP|OPF_CC }, // 7c |
763 | { "jge", OP_JGE, 1, 1, OPF_JMP|OPF_CC }, // 7d SF=OF |
764 | { "jnl", OP_JGE, 1, 1, OPF_JMP|OPF_CC }, // 7d |
765 | { "jle", OP_JLE, 1, 1, OPF_JMP|OPF_CC }, // 7e ZF=1 || SF!=OF |
766 | { "jng", OP_JLE, 1, 1, OPF_JMP|OPF_CC }, // 7e |
767 | { "jg", OP_JG, 1, 1, OPF_JMP|OPF_CC }, // 7f ZF=0 && SF=OF |
768 | { "jnle", OP_JG, 1, 1, OPF_JMP|OPF_CC }, // 7f |
5101a5f9 |
769 | { "seto", OP_JO, 1, 1, OPF_DATA|OPF_CC }, |
770 | { "setno", OP_JNO, 1, 1, OPF_DATA|OPF_CC }, |
771 | { "setc", OP_JC, 1, 1, OPF_DATA|OPF_CC }, |
772 | { "setb", OP_JC, 1, 1, OPF_DATA|OPF_CC }, |
773 | { "setnc", OP_JNC, 1, 1, OPF_DATA|OPF_CC }, |
774 | { "setae", OP_JNC, 1, 1, OPF_DATA|OPF_CC }, |
775 | { "setz", OP_JZ, 1, 1, OPF_DATA|OPF_CC }, |
776 | { "sete", OP_JZ, 1, 1, OPF_DATA|OPF_CC }, |
777 | { "setnz", OP_JNZ, 1, 1, OPF_DATA|OPF_CC }, |
778 | { "setne", OP_JNZ, 1, 1, OPF_DATA|OPF_CC }, |
779 | { "setbe", OP_JBE, 1, 1, OPF_DATA|OPF_CC }, |
780 | { "setna", OP_JBE, 1, 1, OPF_DATA|OPF_CC }, |
781 | { "seta", OP_JA, 1, 1, OPF_DATA|OPF_CC }, |
782 | { "setnbe", OP_JA, 1, 1, OPF_DATA|OPF_CC }, |
783 | { "sets", OP_JS, 1, 1, OPF_DATA|OPF_CC }, |
784 | { "setns", OP_JNS, 1, 1, OPF_DATA|OPF_CC }, |
785 | { "setp", OP_JP, 1, 1, OPF_DATA|OPF_CC }, |
786 | { "setpe", OP_JP, 1, 1, OPF_DATA|OPF_CC }, |
787 | { "setnp", OP_JNP, 1, 1, OPF_DATA|OPF_CC }, |
788 | { "setpo", OP_JNP, 1, 1, OPF_DATA|OPF_CC }, |
789 | { "setl", OP_JL, 1, 1, OPF_DATA|OPF_CC }, |
790 | { "setnge", OP_JL, 1, 1, OPF_DATA|OPF_CC }, |
791 | { "setge", OP_JGE, 1, 1, OPF_DATA|OPF_CC }, |
792 | { "setnl", OP_JGE, 1, 1, OPF_DATA|OPF_CC }, |
793 | { "setle", OP_JLE, 1, 1, OPF_DATA|OPF_CC }, |
794 | { "setng", OP_JLE, 1, 1, OPF_DATA|OPF_CC }, |
795 | { "setg", OP_JG, 1, 1, OPF_DATA|OPF_CC }, |
796 | { "setnle", OP_JG, 1, 1, OPF_DATA|OPF_CC }, |
69a3cdfc |
797 | }; |
c36e914d |
798 | |
799 | static void parse_op(struct parsed_op *op, char words[16][256], int wordc) |
800 | { |
bfa4a6ee |
801 | enum opr_lenmod lmod = OPLM_UNSPEC; |
33c35af6 |
802 | int prefix_flags = 0; |
69a3cdfc |
803 | int regmask_ind; |
804 | int regmask; |
33c35af6 |
805 | int op_w = 0; |
91977a1c |
806 | int opr = 0; |
33c35af6 |
807 | int w = 0; |
91977a1c |
808 | int i; |
c36e914d |
809 | |
33c35af6 |
810 | for (i = 0; i < ARRAY_SIZE(pref_table); i++) { |
811 | if (IS(words[w], pref_table[i].name)) { |
812 | prefix_flags = pref_table[i].flags; |
813 | break; |
814 | } |
815 | } |
816 | |
817 | if (prefix_flags) { |
818 | if (wordc <= 1) |
819 | aerr("lone prefix: '%s'\n", words[0]); |
820 | w++; |
821 | } |
822 | |
823 | op_w = w; |
91977a1c |
824 | for (i = 0; i < ARRAY_SIZE(op_table); i++) { |
33c35af6 |
825 | if (IS(words[w], op_table[i].name)) |
69a3cdfc |
826 | break; |
827 | } |
c36e914d |
828 | |
69a3cdfc |
829 | if (i == ARRAY_SIZE(op_table)) |
830 | aerr("unhandled op: '%s'\n", words[0]); |
33c35af6 |
831 | w++; |
c36e914d |
832 | |
69a3cdfc |
833 | op->op = op_table[i].op; |
33c35af6 |
834 | op->flags = op_table[i].flags | prefix_flags; |
69a3cdfc |
835 | op->regmask_src = op->regmask_dst = 0; |
836 | |
837 | for (opr = 0; opr < op_table[i].minopr; opr++) { |
838 | regmask = regmask_ind = 0; |
839 | w = parse_operand(&op->operand[opr], ®mask, ®mask_ind, |
840 | words, wordc, w, op->flags); |
841 | |
842 | if (opr == 0 && (op->flags & OPF_DATA)) |
843 | op->regmask_dst = regmask; |
844 | // for now, mark dst as src too |
845 | op->regmask_src |= regmask | regmask_ind; |
846 | } |
c36e914d |
847 | |
69a3cdfc |
848 | for (; w < wordc && opr < op_table[i].maxopr; opr++) { |
849 | w = parse_operand(&op->operand[opr], |
850 | &op->regmask_src, &op->regmask_src, |
851 | words, wordc, w, op->flags); |
91977a1c |
852 | } |
c36e914d |
853 | |
91977a1c |
854 | if (w < wordc) |
855 | aerr("parse_op %s incomplete: %d/%d\n", |
856 | words[0], w, wordc); |
5101a5f9 |
857 | |
858 | // special cases |
859 | op->operand_cnt = opr; |
860 | if (!strncmp(op_table[i].name, "set", 3)) |
861 | op->operand[0].lmod = OPLM_BYTE; |
862 | |
863 | // ops with implicit argumets |
864 | switch (op->op) { |
865 | case OP_CDQ: |
866 | op->operand_cnt = 2; |
867 | setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst); |
868 | setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src); |
869 | break; |
870 | |
33c35af6 |
871 | case OP_STOS: |
872 | if (op->operand_cnt != 0) |
873 | break; |
874 | if (IS(words[op_w], "stosb")) |
875 | lmod = OPLM_BYTE; |
876 | else if (IS(words[op_w], "stosw")) |
877 | lmod = OPLM_WORD; |
878 | else if (IS(words[op_w], "stosd")) |
879 | lmod = OPLM_DWORD; |
880 | op->operand_cnt = 3; |
d4e3b5db |
881 | setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src); |
882 | setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src); |
883 | op->regmask_dst = op->regmask_src; |
33c35af6 |
884 | setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD, &op->regmask_src); |
885 | break; |
886 | |
d4e3b5db |
887 | case OP_MOVS: |
888 | if (op->operand_cnt != 0) |
889 | break; |
890 | if (IS(words[op_w], "movsb")) |
891 | lmod = OPLM_BYTE; |
892 | else if (IS(words[op_w], "movsw")) |
893 | lmod = OPLM_WORD; |
894 | else if (IS(words[op_w], "movsd")) |
895 | lmod = OPLM_DWORD; |
896 | op->operand_cnt = 3; |
897 | setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src); |
898 | setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src); |
899 | setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src); |
900 | op->regmask_dst = op->regmask_src; |
901 | break; |
902 | |
5101a5f9 |
903 | case OP_IMUL: |
904 | if (op->operand_cnt != 1) |
905 | break; |
906 | // fallthrough |
907 | case OP_MUL: |
908 | // singleop mul |
909 | op->regmask_dst = (1 << xDX) | (1 << xAX); |
910 | op->regmask_src |= (1 << xAX); |
911 | if (op->operand[0].lmod == OPLM_UNSPEC) |
912 | op->operand[0].lmod = OPLM_DWORD; |
913 | break; |
914 | |
915 | case OP_DIV: |
916 | case OP_IDIV: |
917 | // we could set up operands for edx:eax, but there is no real need to |
918 | // (see is_opr_modified()) |
919 | regmask = (1 << xDX) | (1 << xAX); |
920 | op->regmask_dst = regmask; |
921 | op->regmask_src |= regmask; |
922 | if (op->operand[0].lmod == OPLM_UNSPEC) |
923 | op->operand[0].lmod = OPLM_DWORD; |
924 | break; |
925 | |
926 | case OP_SHL: |
927 | case OP_SHR: |
928 | case OP_SAR: |
d4e3b5db |
929 | case OP_ROL: |
930 | case OP_ROR: |
5101a5f9 |
931 | if (op->operand[1].lmod == OPLM_UNSPEC) |
932 | op->operand[1].lmod = OPLM_BYTE; |
933 | break; |
934 | |
935 | default: |
936 | break; |
937 | } |
c36e914d |
938 | } |
939 | |
850c9265 |
940 | static const char *op_name(enum op_op op) |
941 | { |
942 | int i; |
943 | |
944 | for (i = 0; i < ARRAY_SIZE(op_table); i++) |
945 | if (op_table[i].op == op) |
946 | return op_table[i].name; |
947 | |
948 | return "???"; |
949 | } |
950 | |
951 | // debug |
952 | static const char *dump_op(struct parsed_op *po) |
953 | { |
954 | static char out[128]; |
955 | char *p = out; |
956 | int i; |
957 | |
4c45fa73 |
958 | if (po == NULL) |
959 | return "???"; |
960 | |
850c9265 |
961 | snprintf(out, sizeof(out), "%s", op_name(po->op)); |
962 | for (i = 0; i < po->operand_cnt; i++) { |
963 | p += strlen(p); |
964 | if (i > 0) |
965 | *p++ = ','; |
966 | snprintf(p, sizeof(out) - (p - out), |
967 | po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s", |
968 | po->operand[i].name); |
969 | } |
970 | |
971 | return out; |
972 | } |
973 | |
4c45fa73 |
974 | static const char *lmod_type_u(struct parsed_op *po, |
975 | enum opr_lenmod lmod) |
976 | { |
977 | switch (lmod) { |
978 | case OPLM_DWORD: |
979 | return "u32"; |
980 | case OPLM_WORD: |
981 | return "u16"; |
982 | case OPLM_BYTE: |
983 | return "u8"; |
984 | default: |
985 | ferr(po, "invalid lmod: %d\n", lmod); |
986 | return "(_invalid_)"; |
987 | } |
988 | } |
989 | |
d4e3b5db |
990 | static const char *lmod_cast_u(struct parsed_op *po, |
991 | enum opr_lenmod lmod) |
992 | { |
993 | switch (lmod) { |
994 | case OPLM_DWORD: |
995 | return ""; |
996 | case OPLM_WORD: |
997 | return "(u16)"; |
998 | case OPLM_BYTE: |
999 | return "(u8)"; |
1000 | default: |
1001 | ferr(po, "invalid lmod: %d\n", lmod); |
1002 | return "(_invalid_)"; |
1003 | } |
1004 | } |
1005 | |
1006 | static const char *lmod_cast_u_ptr(struct parsed_op *po, |
1007 | enum opr_lenmod lmod) |
1008 | { |
1009 | switch (lmod) { |
1010 | case OPLM_DWORD: |
1011 | return "*(u32 *)"; |
1012 | case OPLM_WORD: |
1013 | return "*(u16 *)"; |
1014 | case OPLM_BYTE: |
1015 | return "*(u8 *)"; |
1016 | default: |
1017 | ferr(po, "invalid lmod: %d\n", lmod); |
1018 | return "(_invalid_)"; |
1019 | } |
1020 | } |
1021 | |
1022 | static const char *lmod_cast_s(struct parsed_op *po, |
1023 | enum opr_lenmod lmod) |
1024 | { |
1025 | switch (lmod) { |
1026 | case OPLM_DWORD: |
1027 | return "(s32)"; |
1028 | case OPLM_WORD: |
1029 | return "(s16)"; |
1030 | case OPLM_BYTE: |
1031 | return "(s8)"; |
1032 | default: |
1033 | ferr(po, "%s: invalid lmod: %d\n", __func__, lmod); |
1034 | return "(_invalid_)"; |
1035 | } |
1036 | } |
1037 | |
1038 | static const char *lmod_cast(struct parsed_op *po, |
1039 | enum opr_lenmod lmod, int is_signed) |
1040 | { |
1041 | return is_signed ? |
1042 | lmod_cast_s(po, lmod) : |
1043 | lmod_cast_u(po, lmod); |
1044 | } |
1045 | |
1046 | static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod) |
1047 | { |
1048 | switch (lmod) { |
1049 | case OPLM_DWORD: |
1050 | return 4; |
1051 | case OPLM_WORD: |
1052 | return 2; |
1053 | case OPLM_BYTE: |
1054 | return 1; |
1055 | default: |
1056 | ferr(po, "%s: invalid lmod: %d\n", __func__, lmod); |
1057 | return 0; |
1058 | } |
1059 | } |
1060 | |
91977a1c |
1061 | static const char *opr_name(struct parsed_op *po, int opr_num) |
c36e914d |
1062 | { |
91977a1c |
1063 | if (opr_num >= po->operand_cnt) |
1064 | ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt); |
1065 | return po->operand[opr_num].name; |
c36e914d |
1066 | } |
1067 | |
91977a1c |
1068 | static unsigned int opr_const(struct parsed_op *po, int opr_num) |
c36e914d |
1069 | { |
91977a1c |
1070 | if (opr_num >= po->operand_cnt) |
1071 | ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt); |
1072 | if (po->operand[opr_num].type != OPT_CONST) |
1073 | ferr(po, "opr %d: const expected\n", opr_num); |
1074 | return po->operand[opr_num].val; |
1075 | } |
c36e914d |
1076 | |
91977a1c |
1077 | static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr) |
1078 | { |
1079 | if ((unsigned int)popr->reg >= MAX_REGS) |
1080 | ferr(po, "invalid reg: %d\n", popr->reg); |
1081 | return regs_r32[popr->reg]; |
1082 | } |
c36e914d |
1083 | |
39b168b8 |
1084 | static struct parsed_equ *equ_find(struct parsed_op *po, const char *name, |
1085 | int *extra_offs) |
91977a1c |
1086 | { |
39b168b8 |
1087 | const char *p; |
1088 | char *endp; |
1089 | int namelen; |
850c9265 |
1090 | int i; |
1091 | |
39b168b8 |
1092 | *extra_offs = 0; |
1093 | namelen = strlen(name); |
1094 | |
1095 | p = strchr(name, '+'); |
1096 | if (p != NULL) { |
1097 | namelen = p - name; |
1098 | if (namelen <= 0) |
1099 | ferr(po, "equ parse failed for '%s'\n", name); |
1100 | |
1101 | if (IS_START(p, "0x")) |
1102 | p += 2; |
1103 | *extra_offs = strtol(p, &endp, 16); |
1104 | if (*endp != 0) |
1105 | ferr(po, "equ parse failed for '%s'\n", name); |
1106 | } |
1107 | |
850c9265 |
1108 | for (i = 0; i < g_eqcnt; i++) |
39b168b8 |
1109 | if (strncmp(g_eqs[i].name, name, namelen) == 0 |
1110 | && g_eqs[i].name[namelen] == 0) |
850c9265 |
1111 | break; |
87bf6cec |
1112 | if (i >= g_eqcnt) { |
1113 | if (po != NULL) |
1114 | ferr(po, "unresolved equ name: '%s'\n", name); |
1115 | return NULL; |
1116 | } |
850c9265 |
1117 | |
1118 | return &g_eqs[i]; |
1119 | } |
1120 | |
1bafb621 |
1121 | static void stack_frame_access(struct parsed_op *po, |
1122 | enum opr_lenmod lmod, char *buf, size_t buf_size, |
1123 | const char *name, int is_src, int is_lea) |
850c9265 |
1124 | { |
1125 | const char *prefix = ""; |
4c45fa73 |
1126 | char ofs_reg[16] = { 0, }; |
91977a1c |
1127 | struct parsed_equ *eq; |
1128 | int i, arg_i, arg_s; |
d4e3b5db |
1129 | const char *bp_arg; |
1130 | int stack_ra = 0; |
39b168b8 |
1131 | int offset = 0; |
850c9265 |
1132 | int sf_ofs; |
91977a1c |
1133 | |
4c45fa73 |
1134 | bp_arg = parse_stack_el(name, ofs_reg); |
91977a1c |
1135 | snprintf(g_comment, sizeof(g_comment), "%s", bp_arg); |
39b168b8 |
1136 | eq = equ_find(po, bp_arg, &offset); |
d4e3b5db |
1137 | if (eq == NULL) |
1138 | ferr(po, "detected but missing eq\n"); |
91977a1c |
1139 | |
39b168b8 |
1140 | offset += eq->offset; |
1141 | |
d4e3b5db |
1142 | if (!strncmp(name, "ebp", 3)) |
1143 | stack_ra = 4; |
1144 | |
39b168b8 |
1145 | if (stack_ra <= offset && offset < stack_ra + 4) |
1146 | ferr(po, "reference to ra? %d %d\n", offset, stack_ra); |
d4e3b5db |
1147 | |
39b168b8 |
1148 | if (offset > stack_ra) { |
1149 | arg_i = (offset - stack_ra - 4) / 4; |
91977a1c |
1150 | if (arg_i < 0 || arg_i >= g_func_pp.argc_stack) |
1bafb621 |
1151 | ferr(po, "offset %d (%s) doesn't map to any arg\n", |
39b168b8 |
1152 | offset, bp_arg); |
4c45fa73 |
1153 | if (ofs_reg[0] != 0) |
1154 | ferr(po, "offset reg on arg acecss?\n"); |
91977a1c |
1155 | |
1156 | for (i = arg_s = 0; i < g_func_pp.argc; i++) { |
1157 | if (g_func_pp.arg[i].reg != NULL) |
1158 | continue; |
1159 | if (arg_s == arg_i) |
1160 | break; |
1161 | arg_s++; |
1162 | } |
1163 | if (i == g_func_pp.argc) |
1164 | ferr(po, "arg %d not in prototype?\n", arg_i); |
850c9265 |
1165 | if (is_lea) |
1166 | ferr(po, "lea to arg?\n"); |
1167 | |
91977a1c |
1168 | snprintf(buf, buf_size, "%sa%d", is_src ? "(u32)" : "", i + 1); |
1169 | } |
1170 | else { |
1bafb621 |
1171 | if (g_stack_fsz == 0) |
1172 | ferr(po, "stack var access without stackframe\n"); |
850c9265 |
1173 | |
39b168b8 |
1174 | sf_ofs = g_stack_fsz + offset; |
850c9265 |
1175 | if (sf_ofs < 0) |
39b168b8 |
1176 | ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz); |
850c9265 |
1177 | |
1178 | if (is_lea) |
33c35af6 |
1179 | prefix = "(u32)&"; |
850c9265 |
1180 | |
1181 | switch (lmod) |
1182 | { |
1183 | case OPLM_BYTE: |
4c45fa73 |
1184 | if (ofs_reg[0] != 0) |
1185 | snprintf(buf, buf_size, "%ssf.b[%d+%s]", |
1186 | prefix, sf_ofs, ofs_reg); |
1187 | else |
1188 | snprintf(buf, buf_size, "%ssf.b[%d]", |
1189 | prefix, sf_ofs); |
850c9265 |
1190 | break; |
1191 | case OPLM_WORD: |
4c45fa73 |
1192 | if (ofs_reg[0] != 0) |
1193 | snprintf(buf, buf_size, "%ssf.w[%d+%s/2]", |
1194 | prefix, sf_ofs / 2, ofs_reg); |
1195 | else |
1196 | snprintf(buf, buf_size, "%ssf.w[%d]", |
1197 | prefix, sf_ofs / 2); |
850c9265 |
1198 | break; |
1199 | case OPLM_DWORD: |
4c45fa73 |
1200 | if (ofs_reg[0] != 0) |
1201 | snprintf(buf, buf_size, "%ssf.d[%d+%s/4]", |
1202 | prefix, sf_ofs / 4, ofs_reg); |
1203 | else |
1204 | snprintf(buf, buf_size, "%ssf.d[%d]", |
1205 | prefix, sf_ofs / 4); |
850c9265 |
1206 | break; |
1207 | default: |
1208 | ferr(po, "bp_stack bad lmod: %d\n", lmod); |
1209 | } |
91977a1c |
1210 | } |
1211 | } |
c36e914d |
1212 | |
91977a1c |
1213 | static char *out_src_opr(char *buf, size_t buf_size, |
850c9265 |
1214 | struct parsed_op *po, struct parsed_opr *popr, int is_lea) |
91977a1c |
1215 | { |
850c9265 |
1216 | char tmp1[256], tmp2[256]; |
1217 | char expr[256]; |
1218 | int ret; |
1219 | |
91977a1c |
1220 | switch (popr->type) { |
1221 | case OPT_REG: |
850c9265 |
1222 | if (is_lea) |
1223 | ferr(po, "lea from reg?\n"); |
1224 | |
91977a1c |
1225 | switch (popr->lmod) { |
1226 | case OPLM_DWORD: |
1227 | snprintf(buf, buf_size, "%s", opr_reg_p(po, popr)); |
1228 | break; |
850c9265 |
1229 | case OPLM_WORD: |
1230 | snprintf(buf, buf_size, "(u16)%s", opr_reg_p(po, popr)); |
1231 | break; |
1232 | case OPLM_BYTE: |
5101a5f9 |
1233 | if (popr->name[1] == 'h') // XXX.. |
1234 | snprintf(buf, buf_size, "(u8)(%s >> 8)", opr_reg_p(po, popr)); |
1235 | else |
1236 | snprintf(buf, buf_size, "(u8)%s", opr_reg_p(po, popr)); |
850c9265 |
1237 | break; |
91977a1c |
1238 | default: |
1239 | ferr(po, "invalid src lmod: %d\n", popr->lmod); |
1240 | } |
1241 | break; |
850c9265 |
1242 | |
91977a1c |
1243 | case OPT_REGMEM: |
4c45fa73 |
1244 | if (parse_stack_el(popr->name, NULL)) { |
1bafb621 |
1245 | stack_frame_access(po, popr->lmod, buf, buf_size, |
d4e3b5db |
1246 | popr->name, 1, is_lea); |
91977a1c |
1247 | break; |
1248 | } |
850c9265 |
1249 | |
1250 | strcpy(expr, popr->name); |
1251 | if (strchr(expr, '[')) { |
1252 | // special case: '[' can only be left for label[reg] form |
1253 | ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2); |
1254 | if (ret != 2) |
1255 | ferr(po, "parse failure for '%s'\n", expr); |
33c35af6 |
1256 | snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2); |
850c9265 |
1257 | } |
1258 | |
1259 | // XXX: do we need more parsing? |
1260 | if (is_lea) { |
1261 | snprintf(buf, buf_size, "%s", expr); |
1262 | break; |
1263 | } |
1264 | |
d4e3b5db |
1265 | snprintf(buf, buf_size, "%s(%s)", |
1266 | lmod_cast_u_ptr(po, popr->lmod), expr); |
91977a1c |
1267 | break; |
850c9265 |
1268 | |
91977a1c |
1269 | case OPT_LABEL: |
850c9265 |
1270 | if (is_lea) |
1271 | snprintf(buf, buf_size, "(u32)&%s", popr->name); |
1272 | else |
87bf6cec |
1273 | snprintf(buf, buf_size, "(u32)%s", popr->name); |
850c9265 |
1274 | break; |
1275 | |
1276 | case OPT_OFFSET: |
1277 | if (is_lea) |
1278 | ferr(po, "lea an offset?\n"); |
1279 | snprintf(buf, buf_size, "(u32)&%s", popr->name); |
91977a1c |
1280 | break; |
850c9265 |
1281 | |
91977a1c |
1282 | case OPT_CONST: |
850c9265 |
1283 | if (is_lea) |
1284 | ferr(po, "lea from const?\n"); |
1285 | |
5101a5f9 |
1286 | printf_number(buf, buf_size, popr->val); |
91977a1c |
1287 | break; |
850c9265 |
1288 | |
91977a1c |
1289 | default: |
1290 | ferr(po, "invalid src type: %d\n", popr->type); |
1291 | } |
1292 | |
1293 | return buf; |
1294 | } |
c36e914d |
1295 | |
91977a1c |
1296 | static char *out_dst_opr(char *buf, size_t buf_size, |
1297 | struct parsed_op *po, struct parsed_opr *popr) |
1298 | { |
1299 | switch (popr->type) { |
1300 | case OPT_REG: |
1301 | switch (popr->lmod) { |
1302 | case OPLM_DWORD: |
1303 | snprintf(buf, buf_size, "%s", opr_reg_p(po, popr)); |
1304 | break; |
850c9265 |
1305 | case OPLM_WORD: |
1306 | // ugh.. |
1307 | snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr)); |
1308 | break; |
1309 | case OPLM_BYTE: |
1310 | // ugh.. |
5101a5f9 |
1311 | if (popr->name[1] == 'h') // XXX.. |
1312 | snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr)); |
1313 | else |
1314 | snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr)); |
850c9265 |
1315 | break; |
91977a1c |
1316 | default: |
1317 | ferr(po, "invalid dst lmod: %d\n", popr->lmod); |
1318 | } |
1319 | break; |
850c9265 |
1320 | |
1321 | case OPT_REGMEM: |
4c45fa73 |
1322 | if (parse_stack_el(popr->name, NULL)) { |
1bafb621 |
1323 | stack_frame_access(po, popr->lmod, buf, buf_size, |
d4e3b5db |
1324 | popr->name, 0, 0); |
850c9265 |
1325 | break; |
1326 | } |
1327 | |
1328 | return out_src_opr(buf, buf_size, po, popr, 0); |
1329 | |
bfa4a6ee |
1330 | case OPT_LABEL: |
1331 | snprintf(buf, buf_size, "%s", popr->name); |
1332 | break; |
1333 | |
91977a1c |
1334 | default: |
1335 | ferr(po, "invalid dst type: %d\n", popr->type); |
1336 | } |
1337 | |
1338 | return buf; |
1339 | } |
c36e914d |
1340 | |
69a3cdfc |
1341 | static enum parsed_flag_op split_cond(struct parsed_op *po, |
940e8e66 |
1342 | enum op_op op, int *is_inv) |
91977a1c |
1343 | { |
940e8e66 |
1344 | *is_inv = 0; |
91977a1c |
1345 | |
69a3cdfc |
1346 | switch (op) { |
1347 | case OP_JO: |
1348 | return PFO_O; |
1349 | case OP_JC: |
1350 | return PFO_C; |
1351 | case OP_JZ: |
1352 | return PFO_Z; |
1353 | case OP_JBE: |
1354 | return PFO_BE; |
1355 | case OP_JS: |
1356 | return PFO_S; |
1357 | case OP_JP: |
1358 | return PFO_P; |
1359 | case OP_JL: |
1360 | return PFO_L; |
1361 | case OP_JLE: |
1362 | return PFO_LE; |
1363 | |
91977a1c |
1364 | case OP_JNO: |
940e8e66 |
1365 | *is_inv = 1; |
69a3cdfc |
1366 | return PFO_O; |
91977a1c |
1367 | case OP_JNC: |
940e8e66 |
1368 | *is_inv = 1; |
69a3cdfc |
1369 | return PFO_C; |
91977a1c |
1370 | case OP_JNZ: |
940e8e66 |
1371 | *is_inv = 1; |
69a3cdfc |
1372 | return PFO_Z; |
1373 | case OP_JA: |
940e8e66 |
1374 | *is_inv = 1; |
69a3cdfc |
1375 | return PFO_BE; |
91977a1c |
1376 | case OP_JNS: |
940e8e66 |
1377 | *is_inv = 1; |
69a3cdfc |
1378 | return PFO_S; |
91977a1c |
1379 | case OP_JNP: |
940e8e66 |
1380 | *is_inv = 1; |
69a3cdfc |
1381 | return PFO_P; |
850c9265 |
1382 | case OP_JGE: |
940e8e66 |
1383 | *is_inv = 1; |
69a3cdfc |
1384 | return PFO_L; |
91977a1c |
1385 | case OP_JG: |
940e8e66 |
1386 | *is_inv = 1; |
69a3cdfc |
1387 | return PFO_LE; |
1388 | |
1389 | case OP_ADC: |
1390 | case OP_SBB: |
1391 | return PFO_C; |
1392 | |
91977a1c |
1393 | default: |
69a3cdfc |
1394 | ferr(po, "split_cond: bad op %d\n", op); |
1395 | return -1; |
91977a1c |
1396 | } |
1397 | } |
c36e914d |
1398 | |
91977a1c |
1399 | static void out_test_for_cc(char *buf, size_t buf_size, |
940e8e66 |
1400 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv, |
69a3cdfc |
1401 | enum opr_lenmod lmod, const char *expr) |
91977a1c |
1402 | { |
69a3cdfc |
1403 | const char *cast, *scast; |
91977a1c |
1404 | |
69a3cdfc |
1405 | cast = lmod_cast_u(po, lmod); |
1406 | scast = lmod_cast_s(po, lmod); |
1407 | |
1408 | switch (pfo) { |
1409 | case PFO_Z: |
87bf6cec |
1410 | case PFO_BE: // CF=1||ZF=1; CF=0 |
850c9265 |
1411 | snprintf(buf, buf_size, "(%s%s %s 0)", |
940e8e66 |
1412 | cast, expr, is_inv ? "!=" : "=="); |
91977a1c |
1413 | break; |
850c9265 |
1414 | |
5101a5f9 |
1415 | case PFO_S: |
1416 | case PFO_L: // SF!=OF; OF=0 |
1417 | snprintf(buf, buf_size, "(%s%s %s 0)", |
940e8e66 |
1418 | scast, expr, is_inv ? ">=" : "<"); |
5101a5f9 |
1419 | break; |
1420 | |
87bf6cec |
1421 | case PFO_LE: // ZF=1||SF!=OF; OF=0 |
69a3cdfc |
1422 | snprintf(buf, buf_size, "(%s%s %s 0)", |
940e8e66 |
1423 | scast, expr, is_inv ? ">" : "<="); |
850c9265 |
1424 | break; |
1425 | |
91977a1c |
1426 | default: |
69a3cdfc |
1427 | ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo); |
91977a1c |
1428 | } |
1429 | } |
c36e914d |
1430 | |
850c9265 |
1431 | static void out_cmp_for_cc(char *buf, size_t buf_size, |
940e8e66 |
1432 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv, |
69a3cdfc |
1433 | enum opr_lenmod lmod, const char *expr1, const char *expr2) |
850c9265 |
1434 | { |
69a3cdfc |
1435 | const char *cast, *scast; |
850c9265 |
1436 | |
69a3cdfc |
1437 | cast = lmod_cast_u(po, lmod); |
1438 | scast = lmod_cast_s(po, lmod); |
850c9265 |
1439 | |
69a3cdfc |
1440 | switch (pfo) { |
5101a5f9 |
1441 | case PFO_C: |
1442 | // note: must be unsigned compare |
1443 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
940e8e66 |
1444 | cast, expr1, is_inv ? ">=" : "<", cast, expr2); |
5101a5f9 |
1445 | break; |
1446 | |
69a3cdfc |
1447 | case PFO_Z: |
850c9265 |
1448 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
940e8e66 |
1449 | cast, expr1, is_inv ? "!=" : "==", cast, expr2); |
850c9265 |
1450 | break; |
1451 | |
5101a5f9 |
1452 | case PFO_BE: // !a |
850c9265 |
1453 | // note: must be unsigned compare |
1454 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
940e8e66 |
1455 | cast, expr1, is_inv ? ">" : "<=", cast, expr2); |
5101a5f9 |
1456 | break; |
1457 | |
1458 | // note: must be signed compare |
1459 | case PFO_S: |
1460 | snprintf(buf, buf_size, "(%s(%s - %s) %s 0)", |
940e8e66 |
1461 | scast, expr1, expr2, is_inv ? ">=" : "<"); |
850c9265 |
1462 | break; |
1463 | |
5101a5f9 |
1464 | case PFO_L: // !ge |
850c9265 |
1465 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
940e8e66 |
1466 | scast, expr1, is_inv ? ">=" : "<", scast, expr2); |
850c9265 |
1467 | break; |
1468 | |
5101a5f9 |
1469 | case PFO_LE: |
1470 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
940e8e66 |
1471 | scast, expr1, is_inv ? ">" : "<=", scast, expr2); |
5101a5f9 |
1472 | break; |
1473 | |
850c9265 |
1474 | default: |
69a3cdfc |
1475 | ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo); |
850c9265 |
1476 | } |
1477 | } |
1478 | |
69a3cdfc |
1479 | static void out_cmp_test(char *buf, size_t buf_size, |
940e8e66 |
1480 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv) |
69a3cdfc |
1481 | { |
1482 | char buf1[256], buf2[256], buf3[256]; |
1483 | |
1484 | if (po->op == OP_TEST) { |
1485 | if (IS(opr_name(po, 0), opr_name(po, 1))) { |
1486 | out_src_opr(buf3, sizeof(buf3), po, &po->operand[0], 0); |
1487 | } |
1488 | else { |
1489 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
1490 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0); |
1491 | snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2); |
1492 | } |
940e8e66 |
1493 | out_test_for_cc(buf, buf_size, po, pfo, is_inv, |
69a3cdfc |
1494 | po->operand[0].lmod, buf3); |
1495 | } |
1496 | else if (po->op == OP_CMP) { |
1497 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[0], 0); |
1498 | out_src_opr(buf3, sizeof(buf3), po, &po->operand[1], 0); |
940e8e66 |
1499 | out_cmp_for_cc(buf, buf_size, po, pfo, is_inv, |
69a3cdfc |
1500 | po->operand[0].lmod, buf2, buf3); |
1501 | } |
1502 | else |
1503 | ferr(po, "%s: unhandled op: %d\n", __func__, po->op); |
1504 | } |
1505 | |
850c9265 |
1506 | static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1, |
91977a1c |
1507 | struct parsed_opr *popr2) |
1508 | { |
1509 | if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC) |
1510 | ferr(po, "missing lmod for both operands\n"); |
1511 | |
1512 | if (popr1->lmod == OPLM_UNSPEC) |
1513 | popr1->lmod = popr2->lmod; |
1514 | else if (popr2->lmod == OPLM_UNSPEC) |
1515 | popr2->lmod = popr1->lmod; |
1516 | else if (popr1->lmod != popr2->lmod) |
1517 | ferr(po, "conflicting lmods: %d vs %d\n", popr1->lmod, popr2->lmod); |
1518 | } |
c36e914d |
1519 | |
850c9265 |
1520 | static const char *op_to_c(struct parsed_op *po) |
1521 | { |
1522 | switch (po->op) |
1523 | { |
1524 | case OP_ADD: |
5101a5f9 |
1525 | case OP_ADC: |
850c9265 |
1526 | return "+"; |
1527 | case OP_SUB: |
5101a5f9 |
1528 | case OP_SBB: |
850c9265 |
1529 | return "-"; |
1530 | case OP_AND: |
1531 | return "&"; |
1532 | case OP_OR: |
1533 | return "|"; |
1534 | case OP_XOR: |
1535 | return "^"; |
1536 | case OP_SHL: |
1537 | return "<<"; |
1538 | case OP_SHR: |
1539 | return ">>"; |
1540 | case OP_MUL: |
1541 | case OP_IMUL: |
1542 | return "*"; |
1543 | default: |
1544 | ferr(po, "op_to_c was supplied with %d\n", po->op); |
1545 | } |
1546 | } |
1547 | |
d4e3b5db |
1548 | static void set_flag_no_dup(struct parsed_op *po, enum op_flags flag, |
1549 | enum op_flags flag_check) |
1550 | { |
1551 | if (po->flags & flag) |
1552 | ferr(po, "flag %x already set\n", flag); |
1553 | if (po->flags & flag_check) |
1554 | ferr(po, "flag_check %x already set\n", flag_check); |
1555 | |
1556 | po->flags |= flag; |
1557 | } |
1558 | |
87bf6cec |
1559 | static int scan_for_pop(int i, int opcnt, const char *reg, |
d4e3b5db |
1560 | int magic, int depth, int *maxdepth, int do_flags) |
850c9265 |
1561 | { |
87bf6cec |
1562 | struct parsed_op *po; |
1563 | int ret = 0; |
4c45fa73 |
1564 | int j; |
87bf6cec |
1565 | |
850c9265 |
1566 | for (; i < opcnt; i++) { |
87bf6cec |
1567 | po = &ops[i]; |
1568 | if (po->cc_scratch == magic) |
1569 | break; // already checked |
1570 | po->cc_scratch = magic; |
1571 | |
1572 | if (po->flags & OPF_TAIL) |
1573 | return -1; // deadend |
1574 | |
1bafb621 |
1575 | if ((po->flags & OPF_RMD) |
1576 | || (po->op == OP_PUSH && po->argmask)) // arg push |
850c9265 |
1577 | continue; |
1578 | |
87bf6cec |
1579 | if ((po->flags & OPF_JMP) && po->op != OP_CALL) { |
4c45fa73 |
1580 | if (po->btj != NULL) { |
1581 | // jumptable |
1582 | for (j = 0; j < po->btj->count - 1; j++) { |
1583 | ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic, |
1584 | depth, maxdepth, do_flags); |
1585 | if (ret < 0) |
1586 | return ret; // dead end |
1587 | } |
1588 | // follow last jumptable entry |
1589 | i = po->btj->d[j].bt_i - 1; |
1590 | continue; |
1591 | } |
1592 | |
87bf6cec |
1593 | if (po->bt_i < 0) { |
1594 | ferr(po, "dead branch\n"); |
1595 | return -1; |
1596 | } |
850c9265 |
1597 | |
87bf6cec |
1598 | if (po->flags & OPF_CC) { |
d4e3b5db |
1599 | ret |= scan_for_pop(po->bt_i, opcnt, reg, magic, |
1600 | depth, maxdepth, do_flags); |
87bf6cec |
1601 | if (ret < 0) |
1602 | return ret; // dead end |
1603 | } |
1604 | else { |
1605 | i = po->bt_i - 1; |
1606 | } |
1607 | continue; |
1608 | } |
1609 | |
d4e3b5db |
1610 | if ((po->op == OP_POP || po->op == OP_PUSH) |
1611 | && po->operand[0].type == OPT_REG |
87bf6cec |
1612 | && IS(po->operand[0].name, reg)) |
1613 | { |
d4e3b5db |
1614 | if (po->op == OP_PUSH) { |
1615 | depth++; |
1616 | if (depth > *maxdepth) |
1617 | *maxdepth = depth; |
1618 | if (do_flags) |
1619 | set_flag_no_dup(po, OPF_RSAVE, OPF_RMD); |
1620 | } |
1621 | else if (depth == 0) { |
1622 | if (do_flags) |
1623 | set_flag_no_dup(po, OPF_RMD, OPF_RSAVE); |
1624 | return 1; |
1625 | } |
1626 | else { |
1627 | depth--; |
1628 | if (depth < 0) // should not happen |
1629 | ferr(po, "fail with depth\n"); |
1630 | if (do_flags) |
1631 | set_flag_no_dup(po, OPF_RSAVE, OPF_RMD); |
1632 | } |
87bf6cec |
1633 | } |
850c9265 |
1634 | } |
1635 | |
87bf6cec |
1636 | return ret; |
850c9265 |
1637 | } |
1638 | |
69a3cdfc |
1639 | // scan for pop starting from 'ret' op (all paths) |
d4e3b5db |
1640 | static int scan_for_pop_ret(int i, int opcnt, const char *reg, |
1641 | int flag_set) |
850c9265 |
1642 | { |
1643 | int found = 0; |
1644 | int j; |
1645 | |
1646 | for (; i < opcnt; i++) { |
87bf6cec |
1647 | if (!(ops[i].flags & OPF_TAIL)) |
850c9265 |
1648 | continue; |
1649 | |
1650 | for (j = i - 1; j >= 0; j--) { |
69a3cdfc |
1651 | if (ops[j].flags & OPF_RMD) |
1652 | continue; |
1653 | if (ops[j].flags & OPF_JMP) |
850c9265 |
1654 | return -1; |
1655 | |
1656 | if (ops[j].op == OP_POP && ops[j].operand[0].type == OPT_REG |
1657 | && IS(ops[j].operand[0].name, reg)) |
1658 | { |
1659 | found = 1; |
d4e3b5db |
1660 | ops[j].flags |= flag_set; |
850c9265 |
1661 | break; |
1662 | } |
1663 | |
1664 | if (g_labels[j][0] != 0) |
1665 | return -1; |
1666 | } |
1667 | } |
1668 | |
1669 | return found ? 0 : -1; |
1670 | } |
1671 | |
5101a5f9 |
1672 | // is operand 'opr modified' by parsed_op 'po'? |
1673 | static int is_opr_modified(const struct parsed_opr *opr, |
69a3cdfc |
1674 | const struct parsed_op *po) |
1675 | { |
1676 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA)) |
1677 | return 0; |
1678 | |
1679 | if (opr->type == OPT_REG && po->operand[0].type == OPT_REG) { |
1680 | if (po->regmask_dst & (1 << opr->reg)) |
1681 | return 1; |
1682 | else |
1683 | return 0; |
1684 | } |
1685 | |
1686 | return IS(po->operand[0].name, opr->name); |
1687 | } |
1688 | |
5101a5f9 |
1689 | // is any operand of parsed_op 'po_test' modified by parsed_op 'po'? |
1690 | static int is_any_opr_modified(const struct parsed_op *po_test, |
1691 | const struct parsed_op *po) |
1692 | { |
1693 | int i; |
1694 | |
1695 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA)) |
1696 | return 0; |
1697 | |
1698 | if (po_test->regmask_src & po->regmask_dst) |
1699 | return 1; |
1700 | |
1701 | for (i = 0; i < po_test->operand_cnt; i++) |
1702 | if (IS(po_test->operand[i].name, po->operand[0].name)) |
1703 | return 1; |
1704 | |
1705 | return 0; |
1706 | } |
1707 | |
940e8e66 |
1708 | // scan for any po_test operand modification in range given |
5101a5f9 |
1709 | static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt) |
69a3cdfc |
1710 | { |
1711 | for (; i < opcnt; i++) { |
5101a5f9 |
1712 | if (is_any_opr_modified(po_test, &ops[i])) |
69a3cdfc |
1713 | return i; |
1714 | } |
1715 | |
1716 | return -1; |
1717 | } |
1718 | |
940e8e66 |
1719 | // scan for po_test operand[0] modification in range given |
1720 | static int scan_for_mod_opr0(struct parsed_op *po_test, |
1721 | int i, int opcnt) |
1722 | { |
1723 | for (; i < opcnt; i++) { |
1724 | if (is_opr_modified(&po_test->operand[0], &ops[i])) |
1725 | return i; |
1726 | } |
1727 | |
1728 | return -1; |
1729 | } |
1730 | |
5101a5f9 |
1731 | static int scan_for_flag_set(int i) |
69a3cdfc |
1732 | { |
1733 | for (; i >= 0; i--) { |
1734 | if (ops[i].flags & OPF_FLAGS) |
1735 | return i; |
1736 | |
1737 | if ((ops[i].flags & OPF_JMP) && !(ops[i].flags & OPF_CC)) |
1738 | return -1; |
1739 | if (g_labels[i][0] != 0) |
1740 | return -1; |
1741 | } |
1742 | |
1743 | return -1; |
1744 | } |
1745 | |
5101a5f9 |
1746 | // scan back for cdq, if anything modifies edx, fail |
1747 | static int scan_for_cdq_edx(int i) |
1748 | { |
1749 | for (; i >= 0; i--) { |
1750 | if (ops[i].op == OP_CDQ) |
1751 | return i; |
1752 | |
1753 | if (ops[i].regmask_dst & (1 << xDX)) |
1754 | return -1; |
1755 | if (g_labels[i][0] != 0) |
1756 | return -1; |
1757 | } |
1758 | |
1759 | return -1; |
1760 | } |
1761 | |
64c59faf |
1762 | static int scan_for_reg_clear(int i, int reg) |
1763 | { |
1764 | for (; i >= 0; i--) { |
1765 | if (ops[i].op == OP_XOR |
1766 | && ops[i].operand[0].lmod == OPLM_DWORD |
1767 | && ops[i].operand[0].reg == ops[i].operand[1].reg |
1768 | && ops[i].operand[0].reg == reg) |
1769 | return i; |
1770 | |
1771 | if (ops[i].regmask_dst & (1 << reg)) |
1772 | return -1; |
1773 | if (g_labels[i][0] != 0) |
1774 | return -1; |
1775 | } |
1776 | |
1777 | return -1; |
1778 | } |
1779 | |
1bafb621 |
1780 | // scan for positive, constant esp adjust |
1781 | static int scan_for_esp_adjust(int i, int opcnt, int *adj) |
1782 | { |
1783 | for (; i < opcnt; i++) { |
1784 | if (ops[i].op == OP_ADD && ops[i].operand[0].reg == xSP) { |
1785 | if (ops[i].operand[1].type != OPT_CONST) |
1786 | ferr(&ops[i], "non-const esp adjust?\n"); |
1787 | *adj = ops[i].operand[1].val; |
1788 | if (*adj & 3) |
1789 | ferr(&ops[i], "unaligned esp adjust: %x\n", *adj); |
1790 | return i; |
1791 | } |
1792 | |
1793 | if ((ops[i].flags & (OPF_JMP|OPF_TAIL)) |
1794 | || ops[i].op == OP_PUSH || ops[i].op == OP_POP) |
1795 | return -1; |
1796 | if (g_labels[i][0] != 0) |
1797 | return -1; |
1798 | } |
1799 | |
1800 | return -1; |
1801 | } |
1802 | |
4c45fa73 |
1803 | static void add_label_ref(struct label_ref *lr, int op_i) |
1804 | { |
1805 | struct label_ref *lr_new; |
1806 | |
1807 | if (lr->i == -1) { |
1808 | lr->i = op_i; |
1809 | return; |
1810 | } |
1811 | |
1812 | lr_new = calloc(1, sizeof(*lr_new)); |
1813 | lr_new->i = op_i; |
1814 | lr->next = lr_new; |
1815 | } |
1816 | |
91977a1c |
1817 | static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt) |
1818 | { |
69a3cdfc |
1819 | struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op; |
850c9265 |
1820 | struct parsed_opr *last_arith_dst = NULL; |
91977a1c |
1821 | char buf1[256], buf2[256], buf3[256]; |
940e8e66 |
1822 | struct parsed_proto *pp, *pp_tmp; |
4c45fa73 |
1823 | struct parsed_data *pd; |
91977a1c |
1824 | const char *tmpname; |
940e8e66 |
1825 | enum parsed_flag_op pfo; |
69a3cdfc |
1826 | int save_arg_vars = 0; |
1827 | int cmp_result_vars = 0; |
87bf6cec |
1828 | int need_mul_var = 0; |
91977a1c |
1829 | int had_decl = 0; |
d4e3b5db |
1830 | int regmask_save = 0; |
91977a1c |
1831 | int regmask_arg = 0; |
1832 | int regmask = 0; |
940e8e66 |
1833 | int pfomask = 0; |
64c59faf |
1834 | int found = 0; |
d4e3b5db |
1835 | int depth = 0; |
91977a1c |
1836 | int no_output; |
4c45fa73 |
1837 | int i, j, l; |
69a3cdfc |
1838 | int dummy; |
91977a1c |
1839 | int arg; |
91977a1c |
1840 | int reg; |
1841 | int ret; |
1842 | |
1bafb621 |
1843 | g_bp_frame = g_sp_frame = g_stack_fsz = 0; |
91977a1c |
1844 | |
1845 | ret = proto_parse(fhdr, funcn, &g_func_pp); |
1846 | if (ret) |
1847 | ferr(ops, "proto_parse failed for '%s'\n", funcn); |
1848 | |
1849 | fprintf(fout, "%s %s(", g_func_pp.ret_type, funcn); |
1850 | for (i = 0; i < g_func_pp.argc; i++) { |
1851 | if (i > 0) |
1852 | fprintf(fout, ", "); |
1853 | fprintf(fout, "%s a%d", g_func_pp.arg[i].type, i + 1); |
1854 | } |
1855 | fprintf(fout, ")\n{\n"); |
1856 | |
1857 | // pass1: |
1bafb621 |
1858 | // - handle ebp/esp frame, remove ops related to it |
91977a1c |
1859 | if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp") |
1860 | && ops[1].op == OP_MOV |
1861 | && IS(opr_name(&ops[1], 0), "ebp") |
1862 | && IS(opr_name(&ops[1], 1), "esp")) |
1863 | { |
87bf6cec |
1864 | int ecx_push = 0; |
1865 | |
91977a1c |
1866 | g_bp_frame = 1; |
69a3cdfc |
1867 | ops[0].flags |= OPF_RMD; |
1868 | ops[1].flags |= OPF_RMD; |
4c45fa73 |
1869 | i = 2; |
91977a1c |
1870 | |
1871 | if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) { |
1bafb621 |
1872 | g_stack_fsz = opr_const(&ops[2], 1); |
69a3cdfc |
1873 | ops[2].flags |= OPF_RMD; |
4c45fa73 |
1874 | i++; |
91977a1c |
1875 | } |
87bf6cec |
1876 | else { |
1877 | // another way msvc builds stack frame.. |
1878 | i = 2; |
1879 | while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) { |
1bafb621 |
1880 | g_stack_fsz += 4; |
87bf6cec |
1881 | ops[i].flags |= OPF_RMD; |
1882 | ecx_push++; |
1883 | i++; |
1884 | } |
4c45fa73 |
1885 | // and another way.. |
1886 | if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX |
1887 | && ops[i].operand[1].type == OPT_CONST |
1888 | && ops[i + 1].op == OP_CALL |
1889 | && IS(opr_name(&ops[i + 1], 0), "__alloca_probe")) |
1890 | { |
1891 | g_stack_fsz += ops[i].operand[1].val; |
1892 | ops[i].flags |= OPF_RMD; |
1893 | i++; |
1894 | ops[i].flags |= OPF_RMD; |
1895 | i++; |
1896 | } |
87bf6cec |
1897 | } |
91977a1c |
1898 | |
64c59faf |
1899 | found = 0; |
91977a1c |
1900 | do { |
1901 | for (; i < opcnt; i++) |
1902 | if (ops[i].op == OP_RET) |
1903 | break; |
64c59faf |
1904 | if (i == opcnt && (ops[i - 1].flags & OPF_JMP) && found) |
1905 | break; |
1906 | |
91977a1c |
1907 | if (ops[i - 1].op != OP_POP || !IS(opr_name(&ops[i - 1], 0), "ebp")) |
1908 | ferr(&ops[i - 1], "'pop ebp' expected\n"); |
69a3cdfc |
1909 | ops[i - 1].flags |= OPF_RMD; |
91977a1c |
1910 | |
1bafb621 |
1911 | if (g_stack_fsz != 0) { |
91977a1c |
1912 | if (ops[i - 2].op != OP_MOV |
1913 | || !IS(opr_name(&ops[i - 2], 0), "esp") |
1914 | || !IS(opr_name(&ops[i - 2], 1), "ebp")) |
1915 | { |
1916 | ferr(&ops[i - 2], "esp restore expected\n"); |
1917 | } |
69a3cdfc |
1918 | ops[i - 2].flags |= OPF_RMD; |
87bf6cec |
1919 | |
1920 | if (ecx_push && ops[i - 3].op == OP_POP |
1921 | && IS(opr_name(&ops[i - 3], 0), "ecx")) |
1922 | { |
1923 | ferr(&ops[i - 3], "unexpected ecx pop\n"); |
1924 | } |
91977a1c |
1925 | } |
87bf6cec |
1926 | |
64c59faf |
1927 | found = 1; |
91977a1c |
1928 | i++; |
1929 | } while (i < opcnt); |
1930 | } |
1bafb621 |
1931 | else { |
1932 | for (i = 0; i < opcnt; i++) { |
1933 | if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL))) |
1934 | break; |
1935 | if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP |
1936 | && ops[i].operand[1].type == OPT_CONST) |
1937 | { |
1938 | g_sp_frame = 1; |
1939 | break; |
1940 | } |
1941 | } |
1942 | |
1943 | if (g_sp_frame) |
1944 | { |
1945 | g_stack_fsz = ops[i].operand[1].val; |
1946 | ops[i].flags |= OPF_RMD; |
1947 | |
1948 | i++; |
1949 | do { |
1950 | for (; i < opcnt; i++) |
1951 | if (ops[i].op == OP_RET) |
1952 | break; |
1953 | if (ops[i - 1].op != OP_ADD |
1954 | || !IS(opr_name(&ops[i - 1], 0), "esp") |
1955 | || ops[i - 1].operand[1].type != OPT_CONST |
1956 | || ops[i - 1].operand[1].val != g_stack_fsz) |
1957 | ferr(&ops[i - 1], "'add esp' expected\n"); |
1958 | ops[i - 1].flags |= OPF_RMD; |
1959 | |
1960 | i++; |
1961 | } while (i < opcnt); |
1962 | } |
1963 | } |
91977a1c |
1964 | |
1965 | // pass2: |
87bf6cec |
1966 | // - resolve all branches |
1967 | for (i = 0; i < opcnt; i++) { |
1968 | po = &ops[i]; |
1969 | po->bt_i = -1; |
4c45fa73 |
1970 | po->btj = NULL; |
87bf6cec |
1971 | |
1972 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_JMP) |
1973 | || po->op == OP_CALL || po->op == OP_RET) |
1974 | continue; |
1975 | |
4c45fa73 |
1976 | if (po->operand[0].type == OPT_REGMEM) { |
1977 | char *p = strchr(po->operand[0].name, '['); |
1978 | if (p == NULL) |
1979 | ferr(po, "unhandled indirect branch\n"); |
1980 | ret = p - po->operand[0].name; |
1981 | strncpy(buf1, po->operand[0].name, ret); |
1982 | buf1[ret] = 0; |
1983 | |
1984 | for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) { |
1985 | if (IS(g_func_pd[j].label, buf1)) { |
1986 | pd = &g_func_pd[j]; |
1987 | break; |
1988 | } |
1989 | } |
1990 | if (pd == NULL) |
1991 | ferr(po, "label '%s' not parsed?\n", buf1); |
1992 | if (pd->type != OPT_OFFSET) |
1993 | ferr(po, "label '%s' with non-offset data?\n", buf1); |
1994 | |
1995 | // find all labels, link |
1996 | for (j = 0; j < pd->count; j++) { |
1997 | for (l = 0; l < opcnt; l++) { |
1998 | if (g_labels[l][0] && IS(g_labels[l], pd->d[j].u.label)) { |
1999 | add_label_ref(&g_label_refs[l], i); |
2000 | pd->d[j].bt_i = l; |
2001 | break; |
2002 | } |
2003 | } |
2004 | } |
2005 | |
2006 | po->btj = pd; |
2007 | continue; |
2008 | } |
2009 | |
2010 | for (l = 0; l < opcnt; l++) { |
2011 | if (g_labels[l][0] && IS(po->operand[0].name, g_labels[l])) { |
2012 | add_label_ref(&g_label_refs[l], i); |
2013 | po->bt_i = l; |
87bf6cec |
2014 | break; |
2015 | } |
2016 | } |
2017 | |
4c45fa73 |
2018 | if (po->bt_i != -1) |
2019 | continue; |
2020 | |
2021 | if (po->operand[0].type == OPT_LABEL) { |
87bf6cec |
2022 | // assume tail call |
2023 | po->op = OP_CALL; |
2024 | po->flags |= OPF_TAIL; |
4c45fa73 |
2025 | continue; |
87bf6cec |
2026 | } |
4c45fa73 |
2027 | |
2028 | ferr(po, "unhandled branch\n"); |
87bf6cec |
2029 | } |
2030 | |
2031 | // pass3: |
91977a1c |
2032 | // - process calls |
1bafb621 |
2033 | for (i = 0; i < opcnt; i++) |
2034 | { |
69a3cdfc |
2035 | po = &ops[i]; |
2036 | if (po->flags & OPF_RMD) |
91977a1c |
2037 | continue; |
850c9265 |
2038 | |
d4e3b5db |
2039 | if (po->op == OP_CALL) |
69a3cdfc |
2040 | { |
1bafb621 |
2041 | pp = calloc(1, sizeof(*pp)); |
91977a1c |
2042 | my_assert_not(pp, NULL); |
1bafb621 |
2043 | tmpname = opr_name(po, 0); |
2044 | if (po->operand[0].type != OPT_LABEL) |
2045 | { |
2046 | ret = scan_for_esp_adjust(i + 1, opcnt, &j); |
2047 | if (ret < 0) |
2048 | ferr(po, "non-__cdecl indirect call unhandled yet\n"); |
2049 | j /= 4; |
2050 | if (j > ARRAY_SIZE(pp->arg)) |
2051 | ferr(po, "esp adjust too large?\n"); |
64c59faf |
2052 | pp->ret_type = strdup("int"); |
1bafb621 |
2053 | pp->argc = pp->argc_stack = j; |
2054 | for (arg = 0; arg < pp->argc; arg++) |
64c59faf |
2055 | pp->arg[arg].type = strdup("int"); |
1bafb621 |
2056 | } |
2057 | else { |
2058 | ret = proto_parse(fhdr, tmpname, pp); |
2059 | if (ret) |
2060 | ferr(po, "proto_parse failed for call '%s'\n", tmpname); |
2061 | } |
2062 | |
2063 | ret = scan_for_esp_adjust(i + 1, opcnt, &j); |
2064 | if (ret >= 0) { |
2065 | if (pp->argc_stack != j / 4) |
2066 | ferr(po, "stack tracking failed: %x %x\n", |
2067 | pp->argc_stack, j); |
2068 | ops[ret].flags |= OPF_RMD; |
2069 | } |
91977a1c |
2070 | |
39b168b8 |
2071 | // can't call functions with non-__cdecl callbacks yet |
2072 | for (arg = 0; arg < pp->argc; arg++) { |
2073 | if (pp->arg[arg].fptr != NULL) { |
2074 | pp_tmp = pp->arg[arg].fptr; |
2075 | if (pp_tmp->is_stdcall || pp_tmp->argc != pp_tmp->argc_stack) |
2076 | ferr(po, "'%s' has a non-__cdecl callback\n", tmpname); |
2077 | } |
2078 | } |
2079 | |
4c45fa73 |
2080 | // collect all stack args |
91977a1c |
2081 | for (arg = 0; arg < pp->argc; arg++) |
2082 | if (pp->arg[arg].reg == NULL) |
2083 | break; |
2084 | |
1bafb621 |
2085 | for (j = i; j >= 0 && arg < pp->argc; ) |
2086 | { |
2087 | if (g_labels[j][0] != 0) { |
2088 | if (j > 0 && ((ops[j - 1].flags & OPF_TAIL) |
2089 | || (ops[j - 1].flags & (OPF_JMP|OPF_CC)) == OPF_JMP)) |
2090 | { |
2091 | // follow the branch in reverse |
4c45fa73 |
2092 | if (g_label_refs[j].i == -1) |
1bafb621 |
2093 | ferr(po, "no refs for '%s'?\n", g_labels[j]); |
4c45fa73 |
2094 | if (g_label_refs[j].next != NULL) |
2095 | ferr(po, "unhandled multiple refs to '%s'\n", g_labels[j]); |
2096 | j = g_label_refs[j].i + 1; |
1bafb621 |
2097 | continue; |
2098 | } |
2099 | break; |
2100 | } |
2101 | j--; |
2102 | |
2103 | if (ops[j].op == OP_CALL) |
2104 | { |
940e8e66 |
2105 | pp_tmp = ops[j].datap; |
2106 | if (pp_tmp == NULL) |
2107 | ferr(po, "arg collect hit unparsed call\n"); |
2108 | if (pp_tmp->argc_stack > 0) |
2109 | ferr(po, "arg collect hit '%s' with %d stack args\n", |
2110 | opr_name(&ops[j], 0), pp_tmp->argc_stack); |
2111 | } |
2112 | else if ((ops[j].flags & OPF_TAIL) |
2113 | || (ops[j].flags & (OPF_JMP|OPF_CC)) == OPF_JMP) |
2114 | { |
2115 | break; |
69a3cdfc |
2116 | } |
1bafb621 |
2117 | else if (ops[j].op == OP_PUSH) |
2118 | { |
940e8e66 |
2119 | pp->arg[arg].datap = &ops[j]; |
2120 | ret = scan_for_mod(&ops[j], j + 1, i); |
2121 | if (ret >= 0) { |
2122 | // mark this push as one that needs operand saving |
1bafb621 |
2123 | ops[j].flags &= ~OPF_RMD; |
940e8e66 |
2124 | ops[j].argmask |= 1 << arg; |
2125 | save_arg_vars |= 1 << arg; |
2126 | } |
2127 | else |
2128 | ops[j].flags |= OPF_RMD; |
2129 | |
2130 | // next arg |
2131 | for (arg++; arg < pp->argc; arg++) |
2132 | if (pp->arg[arg].reg == NULL) |
2133 | break; |
2134 | } |
91977a1c |
2135 | } |
2136 | if (arg < pp->argc) |
64c59faf |
2137 | ferr(po, "arg collect failed for '%s': %d/%d\n", |
2138 | tmpname, arg, pp->argc); |
69a3cdfc |
2139 | po->datap = pp; |
91977a1c |
2140 | } |
d4e3b5db |
2141 | } |
2142 | |
2143 | // pass4: |
2144 | // - find POPs for PUSHes, rm both |
2145 | // - scan for all used registers |
2146 | // - find flag set ops for their users |
1bafb621 |
2147 | // - declare indirect functions |
d4e3b5db |
2148 | for (i = 0; i < opcnt; i++) { |
2149 | po = &ops[i]; |
2150 | if (po->flags & OPF_RMD) |
2151 | continue; |
2152 | |
2153 | if (po->op == OP_PUSH |
2154 | && po->argmask == 0 && !(po->flags & OPF_RSAVE) |
2155 | && po->operand[0].type == OPT_REG) |
2156 | { |
2157 | reg = po->operand[0].reg; |
2158 | if (reg < 0) |
2159 | ferr(po, "reg not set for push?\n"); |
2160 | |
2161 | depth = 0; |
2162 | ret = scan_for_pop(i + 1, opcnt, |
2163 | po->operand[0].name, i + opcnt, 0, &depth, 0); |
2164 | if (ret == 1) { |
2165 | if (depth > 1) |
2166 | ferr(po, "too much depth: %d\n", depth); |
2167 | if (depth > 0) |
2168 | regmask_save |= 1 << reg; |
2169 | |
2170 | po->flags |= OPF_RMD; |
2171 | scan_for_pop(i + 1, opcnt, po->operand[0].name, |
2172 | i + opcnt * 2, 0, &depth, 1); |
2173 | continue; |
2174 | } |
2175 | ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0); |
2176 | if (ret == 0) { |
2177 | arg = OPF_RMD; |
2178 | if (regmask & (1 << reg)) { |
2179 | if (regmask_save & (1 << reg)) |
2180 | ferr(po, "%s already saved?\n", po->operand[0].name); |
2181 | arg = OPF_RSAVE; |
2182 | } |
2183 | po->flags |= arg; |
2184 | scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg); |
2185 | continue; |
2186 | } |
2187 | } |
2188 | |
2189 | regmask |= po->regmask_src | po->regmask_dst; |
2190 | |
2191 | if (po->flags & OPF_CC) |
2192 | { |
2193 | ret = scan_for_flag_set(i - 1); |
2194 | if (ret < 0) |
2195 | ferr(po, "unable to trace flag setter\n"); |
2196 | |
2197 | tmp_op = &ops[ret]; // flag setter |
2198 | pfo = split_cond(po, po->op, &dummy); |
2199 | pfomask = 0; |
2200 | |
2201 | // to get nicer code, we try to delay test and cmp; |
2202 | // if we can't because of operand modification, or if we |
2203 | // have math op, make it calculate flags explicitly |
2204 | if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP) { |
2205 | if (scan_for_mod(tmp_op, ret + 1, i) >= 0) |
2206 | pfomask = 1 << pfo; |
2207 | } |
2208 | else { |
2209 | if ((pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P) |
2210 | || scan_for_mod_opr0(tmp_op, ret + 1, i) >= 0) |
2211 | pfomask = 1 << pfo; |
2212 | } |
2213 | if (pfomask) { |
2214 | tmp_op->pfomask |= pfomask; |
2215 | cmp_result_vars |= pfomask; |
2216 | po->datap = tmp_op; |
2217 | } |
2218 | |
2219 | if (po->op == OP_ADC || po->op == OP_SBB) |
2220 | cmp_result_vars |= 1 << PFO_C; |
2221 | } |
87bf6cec |
2222 | else if (po->op == OP_MUL |
2223 | || (po->op == OP_IMUL && po->operand_cnt == 1)) |
2224 | { |
2225 | need_mul_var = 1; |
2226 | } |
1bafb621 |
2227 | else if (po->op == OP_CALL && po->operand[0].type != OPT_LABEL) { |
2228 | pp = po->datap; |
2229 | my_assert_not(pp, NULL); |
2230 | fprintf(fout, " %s (*icall%d)(", pp->ret_type, i); |
2231 | for (j = 0; j < pp->argc; j++) { |
2232 | if (j > 0) |
2233 | fprintf(fout, ", "); |
2234 | fprintf(fout, "%s a%d", pp->arg[j].type, j + 1); |
2235 | } |
2236 | fprintf(fout, ");\n"); |
2237 | } |
91977a1c |
2238 | } |
2239 | |
4c45fa73 |
2240 | // output LUTs/jumptables |
2241 | for (i = 0; i < g_func_pd_cnt; i++) { |
2242 | pd = &g_func_pd[i]; |
2243 | fprintf(fout, " static const "); |
2244 | if (pd->type == OPT_OFFSET) { |
2245 | fprintf(fout, "void *jt_%s[] =\n { ", pd->label); |
2246 | |
2247 | for (j = 0; j < pd->count; j++) { |
2248 | if (j > 0) |
2249 | fprintf(fout, ", "); |
2250 | fprintf(fout, "&&%s", pd->d[j].u.label); |
2251 | } |
2252 | } |
2253 | else { |
2254 | fprintf(fout, "%s %s[] =\n { ", |
2255 | lmod_type_u(ops, pd->lmod), pd->label); |
2256 | |
2257 | for (j = 0; j < pd->count; j++) { |
2258 | if (j > 0) |
2259 | fprintf(fout, ", "); |
2260 | fprintf(fout, "%u", pd->d[j].u.val); |
2261 | } |
2262 | } |
2263 | fprintf(fout, " };\n"); |
2264 | } |
2265 | |
850c9265 |
2266 | // declare stack frame |
1bafb621 |
2267 | if (g_stack_fsz) |
850c9265 |
2268 | fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n", |
1bafb621 |
2269 | (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz); |
850c9265 |
2270 | |
940e8e66 |
2271 | // declare arg-registers |
91977a1c |
2272 | for (i = 0; i < g_func_pp.argc; i++) { |
2273 | if (g_func_pp.arg[i].reg != NULL) { |
2274 | reg = char_array_i(regs_r32, |
2275 | ARRAY_SIZE(regs_r32), g_func_pp.arg[i].reg); |
2276 | if (reg < 0) |
2277 | ferr(ops, "arg '%s' is not a reg?\n", g_func_pp.arg[i].reg); |
2278 | |
2279 | regmask_arg |= 1 << reg; |
2280 | fprintf(fout, " u32 %s = (u32)a%d;\n", |
850c9265 |
2281 | g_func_pp.arg[i].reg, i + 1); |
91977a1c |
2282 | had_decl = 1; |
2283 | } |
2284 | } |
2285 | |
940e8e66 |
2286 | // declare other regs - special case for eax |
91977a1c |
2287 | if (!((regmask | regmask_arg) & 1) && !IS(g_func_pp.ret_type, "void")) { |
2288 | fprintf(fout, " u32 eax = 0;\n"); |
2289 | had_decl = 1; |
2290 | } |
2291 | |
2292 | regmask &= ~regmask_arg; |
d4e3b5db |
2293 | regmask &= ~(1 << xSP); |
91977a1c |
2294 | if (g_bp_frame) |
2295 | regmask &= ~(1 << xBP); |
2296 | if (regmask) { |
2297 | for (reg = 0; reg < 8; reg++) { |
2298 | if (regmask & (1 << reg)) { |
2299 | fprintf(fout, " u32 %s;\n", regs_r32[reg]); |
2300 | had_decl = 1; |
2301 | } |
2302 | } |
2303 | } |
2304 | |
d4e3b5db |
2305 | if (regmask_save) { |
2306 | for (reg = 0; reg < 8; reg++) { |
2307 | if (regmask_save & (1 << reg)) { |
2308 | fprintf(fout, " u32 s_%s;\n", regs_r32[reg]); |
2309 | had_decl = 1; |
2310 | } |
2311 | } |
2312 | } |
2313 | |
69a3cdfc |
2314 | if (save_arg_vars) { |
2315 | for (reg = 0; reg < 32; reg++) { |
2316 | if (save_arg_vars & (1 << reg)) { |
2317 | fprintf(fout, " u32 s_a%d;\n", reg + 1); |
2318 | had_decl = 1; |
2319 | } |
2320 | } |
2321 | } |
2322 | |
2323 | if (cmp_result_vars) { |
2324 | for (i = 0; i < 8; i++) { |
2325 | if (cmp_result_vars & (1 << i)) { |
2326 | fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]); |
2327 | had_decl = 1; |
2328 | } |
2329 | } |
2330 | } |
2331 | |
87bf6cec |
2332 | if (need_mul_var) { |
2333 | fprintf(fout, " u64 mul_tmp;\n"); |
2334 | had_decl = 1; |
2335 | } |
2336 | |
91977a1c |
2337 | if (had_decl) |
2338 | fprintf(fout, "\n"); |
2339 | |
2340 | // output ops |
69a3cdfc |
2341 | for (i = 0; i < opcnt; i++) |
2342 | { |
4c45fa73 |
2343 | if (g_labels[i][0] != 0 && g_label_refs[i].i != -1) |
91977a1c |
2344 | fprintf(fout, "\n%s:\n", g_labels[i]); |
2345 | |
69a3cdfc |
2346 | po = &ops[i]; |
2347 | if (po->flags & OPF_RMD) |
91977a1c |
2348 | continue; |
2349 | |
2350 | no_output = 0; |
2351 | |
91977a1c |
2352 | #define assert_operand_cnt(n_) \ |
850c9265 |
2353 | if (po->operand_cnt != n_) \ |
2354 | ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_) |
2355 | |
69a3cdfc |
2356 | // conditional/flag using op? |
2357 | if (po->flags & OPF_CC) |
850c9265 |
2358 | { |
940e8e66 |
2359 | int is_delayed = 0; |
2360 | int is_inv = 0; |
69a3cdfc |
2361 | |
940e8e66 |
2362 | pfo = split_cond(po, po->op, &is_inv); |
850c9265 |
2363 | |
69a3cdfc |
2364 | // we go through all this trouble to avoid using parsed_flag_op, |
2365 | // which makes generated code much nicer |
2366 | if (delayed_flag_op != NULL) |
850c9265 |
2367 | { |
940e8e66 |
2368 | out_cmp_test(buf1, sizeof(buf1), delayed_flag_op, pfo, is_inv); |
2369 | is_delayed = 1; |
91977a1c |
2370 | } |
850c9265 |
2371 | else if (last_arith_dst != NULL |
69a3cdfc |
2372 | && (pfo == PFO_Z || pfo == PFO_S || pfo == PFO_P)) |
850c9265 |
2373 | { |
2374 | out_src_opr(buf3, sizeof(buf3), po, last_arith_dst, 0); |
940e8e66 |
2375 | out_test_for_cc(buf1, sizeof(buf1), po, pfo, is_inv, |
850c9265 |
2376 | last_arith_dst->lmod, buf3); |
940e8e66 |
2377 | is_delayed = 1; |
850c9265 |
2378 | } |
69a3cdfc |
2379 | else if (po->datap != NULL) { |
2380 | // use preprocessed results |
2381 | tmp_op = po->datap; |
2382 | if (!tmp_op || !(tmp_op->pfomask & (1 << pfo))) |
2383 | ferr(po, "not prepared for pfo %d\n", pfo); |
2384 | |
940e8e66 |
2385 | // note: is_inv was not yet applied |
69a3cdfc |
2386 | snprintf(buf1, sizeof(buf1), "(%scond_%s)", |
940e8e66 |
2387 | is_inv ? "!" : "", parsed_flag_op_names[pfo]); |
69a3cdfc |
2388 | } |
2389 | else { |
2390 | ferr(po, "all methods of finding comparison failed\n"); |
2391 | } |
850c9265 |
2392 | |
69a3cdfc |
2393 | if (po->flags & OPF_JMP) { |
850c9265 |
2394 | fprintf(fout, " if %s\n", buf1); |
2395 | } |
5101a5f9 |
2396 | else if (po->op == OP_ADC || po->op == OP_SBB) { |
940e8e66 |
2397 | if (is_delayed) |
2398 | fprintf(fout, " cond_%s = %s;\n", |
2399 | parsed_flag_op_names[pfo], buf1); |
850c9265 |
2400 | } |
5101a5f9 |
2401 | else if (po->flags & OPF_DATA) { // SETcc |
850c9265 |
2402 | out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]); |
2403 | fprintf(fout, " %s = %s;", buf2, buf1); |
91977a1c |
2404 | } |
69a3cdfc |
2405 | else { |
2406 | ferr(po, "unhandled conditional op\n"); |
2407 | } |
91977a1c |
2408 | } |
2409 | |
940e8e66 |
2410 | pfomask = po->pfomask; |
2411 | |
850c9265 |
2412 | switch (po->op) |
91977a1c |
2413 | { |
2414 | case OP_MOV: |
2415 | assert_operand_cnt(2); |
850c9265 |
2416 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
06c5d854 |
2417 | fprintf(fout, " %s = %s%s;", |
850c9265 |
2418 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
06c5d854 |
2419 | po->operand[0].is_ptr ? "(void *)" : "", |
850c9265 |
2420 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2421 | break; |
2422 | |
2423 | case OP_LEA: |
2424 | assert_operand_cnt(2); |
87bf6cec |
2425 | po->operand[1].lmod = OPLM_DWORD; // always |
850c9265 |
2426 | fprintf(fout, " %s = %s;", |
2427 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2428 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 1)); |
2429 | break; |
2430 | |
2431 | case OP_MOVZX: |
2432 | assert_operand_cnt(2); |
91977a1c |
2433 | fprintf(fout, " %s = %s;", |
850c9265 |
2434 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2435 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2436 | break; |
2437 | |
2438 | case OP_MOVSX: |
2439 | assert_operand_cnt(2); |
2440 | switch (po->operand[1].lmod) { |
2441 | case OPLM_BYTE: |
2442 | strcpy(buf3, "(s8)"); |
2443 | break; |
2444 | case OPLM_WORD: |
2445 | strcpy(buf3, "(s16)"); |
2446 | break; |
2447 | default: |
2448 | ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod); |
2449 | } |
2450 | fprintf(fout, " %s = %s%s;", |
2451 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2452 | buf3, |
2453 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2454 | break; |
2455 | |
2456 | case OP_NOT: |
2457 | assert_operand_cnt(1); |
2458 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
2459 | fprintf(fout, " %s = ~%s;", buf1, buf1); |
2460 | break; |
2461 | |
5101a5f9 |
2462 | case OP_CDQ: |
2463 | assert_operand_cnt(2); |
2464 | fprintf(fout, " %s = (s32)%s >> 31;", |
2465 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2466 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2467 | strcpy(g_comment, "cdq"); |
2468 | break; |
2469 | |
33c35af6 |
2470 | case OP_STOS: |
2471 | // assumes DF=0 |
2472 | assert_operand_cnt(3); |
2473 | if (po->flags & OPF_REP) { |
2474 | fprintf(fout, " for (; ecx != 0; ecx--, edi += %d)\n", |
2475 | lmod_bytes(po, po->operand[0].lmod)); |
d4e3b5db |
2476 | fprintf(fout, " %sedi = eax;", |
2477 | lmod_cast_u_ptr(po, po->operand[0].lmod)); |
33c35af6 |
2478 | strcpy(g_comment, "rep stos"); |
2479 | } |
2480 | else { |
d4e3b5db |
2481 | fprintf(fout, " %sedi = eax; edi += %d;", |
2482 | lmod_cast_u_ptr(po, po->operand[0].lmod), |
33c35af6 |
2483 | lmod_bytes(po, po->operand[0].lmod)); |
2484 | strcpy(g_comment, "stos"); |
2485 | } |
2486 | break; |
2487 | |
d4e3b5db |
2488 | case OP_MOVS: |
2489 | // assumes DF=0 |
2490 | assert_operand_cnt(3); |
2491 | j = lmod_bytes(po, po->operand[0].lmod); |
2492 | strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod)); |
2493 | if (po->flags & OPF_REP) { |
2494 | fprintf(fout, |
2495 | " for (; ecx != 0; ecx--, edi += %d, esi += %d)\n", |
2496 | j, j); |
2497 | fprintf(fout, |
2498 | " %sedi = %sesi;", buf1, buf1); |
2499 | strcpy(g_comment, "rep movs"); |
2500 | } |
2501 | else { |
2502 | fprintf(fout, " %sedi = %sesi; edi += %d; esi += %d;", |
2503 | buf1, buf1, j, j); |
2504 | strcpy(g_comment, "movs"); |
2505 | } |
2506 | break; |
2507 | |
850c9265 |
2508 | // arithmetic w/flags |
2509 | case OP_ADD: |
2510 | case OP_SUB: |
2511 | case OP_AND: |
2512 | case OP_OR: |
5101a5f9 |
2513 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
2514 | // fallthrough |
850c9265 |
2515 | case OP_SHL: |
2516 | case OP_SHR: |
2517 | dualop_arith: |
2518 | assert_operand_cnt(2); |
850c9265 |
2519 | fprintf(fout, " %s %s= %s;", |
2520 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2521 | op_to_c(po), |
2522 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2523 | last_arith_dst = &po->operand[0]; |
69a3cdfc |
2524 | delayed_flag_op = NULL; |
850c9265 |
2525 | break; |
2526 | |
d4e3b5db |
2527 | case OP_SAR: |
2528 | assert_operand_cnt(2); |
2529 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
2530 | fprintf(fout, " %s = %s%s >> %s;", buf1, |
2531 | lmod_cast_s(po, po->operand[0].lmod), buf1, |
2532 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2533 | last_arith_dst = &po->operand[0]; |
2534 | delayed_flag_op = NULL; |
2535 | break; |
2536 | |
2537 | case OP_ROL: |
2538 | case OP_ROR: |
2539 | assert_operand_cnt(2); |
2540 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
2541 | if (po->operand[1].type == OPT_CONST) { |
2542 | j = po->operand[1].val; |
2543 | j %= lmod_bytes(po, po->operand[0].lmod) * 8; |
2544 | fprintf(fout, po->op == OP_ROL ? |
2545 | " %s = (%s << %d) | (%s >> %d);" : |
2546 | " %s = (%s >> %d) | (%s << %d);", |
2547 | buf1, buf1, j, buf1, |
2548 | lmod_bytes(po, po->operand[0].lmod) * 8 - j); |
2549 | } |
2550 | else |
2551 | ferr(po, "TODO\n"); |
2552 | last_arith_dst = &po->operand[0]; |
2553 | delayed_flag_op = NULL; |
2554 | break; |
2555 | |
5101a5f9 |
2556 | case OP_XOR: |
850c9265 |
2557 | assert_operand_cnt(2); |
2558 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
5101a5f9 |
2559 | if (IS(opr_name(po, 0), opr_name(po, 1))) { |
2560 | // special case for XOR |
2561 | fprintf(fout, " %s = 0;", |
2562 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0])); |
2563 | last_arith_dst = &po->operand[0]; |
2564 | delayed_flag_op = NULL; |
850c9265 |
2565 | break; |
850c9265 |
2566 | } |
5101a5f9 |
2567 | goto dualop_arith; |
2568 | |
5101a5f9 |
2569 | case OP_ADC: |
850c9265 |
2570 | case OP_SBB: |
5101a5f9 |
2571 | assert_operand_cnt(2); |
2572 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
2573 | fprintf(fout, " %s %s= %s + cond_c;", |
2574 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
2575 | op_to_c(po), |
2576 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
2577 | last_arith_dst = &po->operand[0]; |
2578 | delayed_flag_op = NULL; |
850c9265 |
2579 | break; |
2580 | |
2581 | case OP_INC: |
2582 | case OP_DEC: |
2583 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
1bafb621 |
2584 | if (po->operand[0].type == OPT_REG) { |
2585 | strcpy(buf2, po->op == OP_INC ? "++" : "--"); |
2586 | fprintf(fout, " %s%s;", buf1, buf2); |
2587 | } |
2588 | else { |
2589 | strcpy(buf2, po->op == OP_INC ? "+" : "-"); |
2590 | fprintf(fout, " %s %s= 1;", buf1, buf2); |
2591 | } |
5101a5f9 |
2592 | last_arith_dst = &po->operand[0]; |
2593 | delayed_flag_op = NULL; |
2594 | break; |
2595 | |
2596 | case OP_NEG: |
2597 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
2598 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[0], 0); |
2599 | fprintf(fout, " %s = -%s%s;", buf1, |
2600 | lmod_cast_s(po, po->operand[0].lmod), buf2); |
850c9265 |
2601 | last_arith_dst = &po->operand[0]; |
69a3cdfc |
2602 | delayed_flag_op = NULL; |
940e8e66 |
2603 | if (pfomask & (1 << PFO_C)) { |
2604 | fprintf(fout, "\n cond_c = (%s != 0);", buf1); |
2605 | pfomask &= ~(1 << PFO_C); |
2606 | } |
850c9265 |
2607 | break; |
2608 | |
2609 | case OP_IMUL: |
2610 | if (po->operand_cnt == 2) |
2611 | goto dualop_arith; |
87bf6cec |
2612 | if (po->operand_cnt == 3) |
2613 | ferr(po, "TODO imul3\n"); |
2614 | // fallthrough |
2615 | case OP_MUL: |
2616 | assert_operand_cnt(1); |
2617 | strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)"); |
2618 | fprintf(fout, " mul_tmp = %seax * %s%s;\n", buf1, buf1, |
2619 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[0], 0)); |
2620 | fprintf(fout, " edx = mul_tmp >> 32;\n"); |
2621 | fprintf(fout, " eax = mul_tmp;"); |
2622 | last_arith_dst = NULL; |
69a3cdfc |
2623 | delayed_flag_op = NULL; |
91977a1c |
2624 | break; |
2625 | |
5101a5f9 |
2626 | case OP_DIV: |
2627 | case OP_IDIV: |
2628 | assert_operand_cnt(1); |
2629 | if (po->operand[0].lmod != OPLM_DWORD) |
2630 | ferr(po, "unhandled lmod %d\n", po->operand[0].lmod); |
2631 | |
2632 | // 32bit division is common, look for it |
64c59faf |
2633 | if (po->op == OP_DIV) |
2634 | ret = scan_for_reg_clear(i - 1, xDX); |
2635 | else |
2636 | ret = scan_for_cdq_edx(i - 1); |
2637 | if (ret >= 0) { |
5101a5f9 |
2638 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
2639 | strcpy(buf2, lmod_cast(po, po->operand[0].lmod, |
2640 | po->op == OP_IDIV)); |
2641 | fprintf(fout, " edx = %seax %% %s%s;\n", buf2, buf2, buf1); |
2642 | fprintf(fout, " eax = %seax / %s%s;", buf2, buf2, buf1); |
2643 | } |
2644 | else |
2645 | ferr(po, "TODO 64bit divident\n"); |
87bf6cec |
2646 | last_arith_dst = NULL; |
2647 | delayed_flag_op = NULL; |
5101a5f9 |
2648 | break; |
2649 | |
91977a1c |
2650 | case OP_TEST: |
2651 | case OP_CMP: |
850c9265 |
2652 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
940e8e66 |
2653 | if (pfomask != 0) { |
69a3cdfc |
2654 | for (j = 0; j < 8; j++) { |
940e8e66 |
2655 | if (pfomask & (1 << j)) { |
69a3cdfc |
2656 | out_cmp_test(buf1, sizeof(buf1), po, j, 0); |
2657 | fprintf(fout, " cond_%s = %s;", |
2658 | parsed_flag_op_names[j], buf1); |
2659 | } |
2660 | } |
940e8e66 |
2661 | pfomask = 0; |
69a3cdfc |
2662 | } |
2663 | else |
2664 | no_output = 1; |
2665 | delayed_flag_op = po; |
91977a1c |
2666 | break; |
2667 | |
69a3cdfc |
2668 | // note: we reuse OP_Jcc for SETcc, only flags differ |
91977a1c |
2669 | case OP_JO ... OP_JG: |
5101a5f9 |
2670 | if (po->flags & OPF_JMP) |
850c9265 |
2671 | fprintf(fout, " goto %s;", po->operand[0].name); |
5101a5f9 |
2672 | // else SETcc - should already be handled |
850c9265 |
2673 | break; |
2674 | |
2675 | case OP_JMP: |
87bf6cec |
2676 | assert_operand_cnt(1); |
4c45fa73 |
2677 | if (po->operand[0].type == OPT_REGMEM) { |
2678 | ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]", |
2679 | buf1, buf2); |
2680 | if (ret != 2) |
2681 | ferr(po, "parse failure for jmp '%s'\n", |
2682 | po->operand[0].name); |
2683 | fprintf(fout, " goto *jt_%s[%s];", buf1, buf2); |
2684 | break; |
2685 | } |
2686 | else if (po->operand[0].type != OPT_LABEL) |
2687 | ferr(po, "unhandled jmp type\n"); |
87bf6cec |
2688 | |
850c9265 |
2689 | fprintf(fout, " goto %s;", po->operand[0].name); |
91977a1c |
2690 | break; |
2691 | |
2692 | case OP_CALL: |
5101a5f9 |
2693 | assert_operand_cnt(1); |
850c9265 |
2694 | pp = po->datap; |
91977a1c |
2695 | if (pp == NULL) |
850c9265 |
2696 | ferr(po, "NULL pp\n"); |
91977a1c |
2697 | |
1bafb621 |
2698 | if (po->operand[0].type != OPT_LABEL) |
2699 | fprintf(fout, " icall%d = (void *)%s;\n", i, |
2700 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0)); |
2701 | |
91977a1c |
2702 | fprintf(fout, " "); |
2703 | if (!IS(pp->ret_type, "void")) { |
87bf6cec |
2704 | if (po->flags & OPF_TAIL) |
2705 | fprintf(fout, "return "); |
2706 | else |
2707 | fprintf(fout, "eax = "); |
91977a1c |
2708 | if (strchr(pp->ret_type, '*')) |
2709 | fprintf(fout, "(u32)"); |
2710 | } |
87bf6cec |
2711 | |
39b168b8 |
2712 | if (po->operand[0].type != OPT_LABEL) { |
1bafb621 |
2713 | fprintf(fout, "icall%d(", i); |
39b168b8 |
2714 | } |
2715 | else { |
2716 | if (pp->name[0] == 0) |
2717 | ferr(po, "missing pp->name\n"); |
2718 | fprintf(fout, "%s(", pp->name); |
2719 | } |
2720 | |
91977a1c |
2721 | for (arg = 0; arg < pp->argc; arg++) { |
2722 | if (arg > 0) |
2723 | fprintf(fout, ", "); |
87bf6cec |
2724 | |
2725 | if (strchr(pp->arg[arg].type, '*')) |
2726 | fprintf(fout, "(%s)", pp->arg[arg].type); |
2727 | |
91977a1c |
2728 | if (pp->arg[arg].reg != NULL) { |
850c9265 |
2729 | fprintf(fout, "%s", pp->arg[arg].reg); |
91977a1c |
2730 | continue; |
2731 | } |
2732 | |
2733 | // stack arg |
2734 | tmp_op = pp->arg[arg].datap; |
2735 | if (tmp_op == NULL) |
850c9265 |
2736 | ferr(po, "parsed_op missing for arg%d\n", arg); |
940e8e66 |
2737 | if (tmp_op->argmask) { |
2738 | fprintf(fout, "s_a%d", arg + 1); |
69a3cdfc |
2739 | } |
2740 | else { |
2741 | fprintf(fout, "%s", |
2742 | out_src_opr(buf1, sizeof(buf1), |
2743 | tmp_op, &tmp_op->operand[0], 0)); |
2744 | } |
91977a1c |
2745 | } |
2746 | fprintf(fout, ");"); |
87bf6cec |
2747 | |
2748 | if (po->flags & OPF_TAIL) { |
2749 | strcpy(g_comment, "tailcall"); |
2750 | if (IS(pp->ret_type, "void")) |
2751 | fprintf(fout, "\n return;"); |
2752 | } |
2753 | delayed_flag_op = NULL; |
2754 | last_arith_dst = NULL; |
91977a1c |
2755 | break; |
2756 | |
2757 | case OP_RET: |
2758 | if (IS(g_func_pp.ret_type, "void")) |
2759 | fprintf(fout, " return;"); |
d4e3b5db |
2760 | else if (strchr(g_func_pp.ret_type, '*')) |
2761 | fprintf(fout, " return (%s)eax;", |
2762 | g_func_pp.ret_type); |
91977a1c |
2763 | else |
2764 | fprintf(fout, " return eax;"); |
2765 | break; |
2766 | |
2767 | case OP_PUSH: |
940e8e66 |
2768 | if (po->argmask) { |
69a3cdfc |
2769 | // special case - saved func arg |
d4e3b5db |
2770 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
940e8e66 |
2771 | for (j = 0; j < 32; j++) { |
d4e3b5db |
2772 | if (po->argmask & (1 << j)) |
2773 | fprintf(fout, " s_a%d = %s;", j + 1, buf1); |
940e8e66 |
2774 | } |
69a3cdfc |
2775 | break; |
2776 | } |
d4e3b5db |
2777 | else if (po->flags & OPF_RSAVE) { |
2778 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
2779 | fprintf(fout, " s_%s = %s;", buf1, buf1); |
2780 | break; |
2781 | } |
2782 | ferr(po, "stray push encountered\n"); |
91977a1c |
2783 | break; |
2784 | |
2785 | case OP_POP: |
d4e3b5db |
2786 | if (po->flags & OPF_RSAVE) { |
2787 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
2788 | fprintf(fout, " %s = s_%s;", buf1, buf1); |
2789 | break; |
2790 | } |
2791 | ferr(po, "stray pop encountered\n"); |
91977a1c |
2792 | break; |
2793 | |
33c35af6 |
2794 | case OP_NOP: |
2795 | break; |
2796 | |
91977a1c |
2797 | default: |
2798 | no_output = 1; |
69a3cdfc |
2799 | ferr(po, "unhandled op type %d, flags %x\n", |
2800 | po->op, po->flags); |
91977a1c |
2801 | break; |
2802 | } |
2803 | |
2804 | if (g_comment[0] != 0) { |
2805 | fprintf(fout, " // %s", g_comment); |
2806 | g_comment[0] = 0; |
2807 | no_output = 0; |
2808 | } |
2809 | if (!no_output) |
2810 | fprintf(fout, "\n"); |
5101a5f9 |
2811 | |
940e8e66 |
2812 | if (pfomask != 0) |
2813 | ferr(po, "missed flag calc, pfomask=%x\n", pfomask); |
2814 | |
5101a5f9 |
2815 | // see is delayed flag stuff is still valid |
2816 | if (delayed_flag_op != NULL && delayed_flag_op != po) { |
2817 | if (is_any_opr_modified(delayed_flag_op, po)) |
2818 | delayed_flag_op = NULL; |
2819 | } |
2820 | |
2821 | if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) { |
2822 | if (is_opr_modified(last_arith_dst, po)) |
2823 | last_arith_dst = NULL; |
2824 | } |
91977a1c |
2825 | } |
2826 | |
2827 | fprintf(fout, "}\n\n"); |
2828 | |
2829 | // cleanup |
2830 | for (i = 0; i < opcnt; i++) { |
4c45fa73 |
2831 | struct label_ref *lr, *lr_del; |
2832 | |
2833 | lr = g_label_refs[i].next; |
2834 | while (lr != NULL) { |
2835 | lr_del = lr; |
2836 | lr = lr->next; |
2837 | free(lr_del); |
2838 | } |
2839 | g_label_refs[i].i = -1; |
2840 | g_label_refs[i].next = NULL; |
2841 | |
91977a1c |
2842 | if (ops[i].op == OP_CALL) { |
2843 | pp = ops[i].datap; |
2844 | if (pp) { |
2845 | proto_release(pp); |
2846 | free(pp); |
2847 | } |
2848 | } |
2849 | } |
2850 | proto_release(&g_func_pp); |
2851 | } |
c36e914d |
2852 | |
d4e3b5db |
2853 | static void set_label(int i, const char *name) |
2854 | { |
2855 | const char *p; |
2856 | int len; |
2857 | |
2858 | len = strlen(name); |
2859 | p = strchr(name, ':'); |
2860 | if (p != NULL) |
2861 | len = p - name; |
2862 | |
2863 | if (len > sizeof(g_labels[0]) - 1) |
2864 | aerr("label '%s' too long: %d\n", name, len); |
2865 | if (g_labels[i][0] != 0) |
2866 | aerr("dupe label '%s'?\n", name); |
2867 | memcpy(g_labels[i], name, len); |
2868 | g_labels[i][len] = 0; |
2869 | } |
2870 | |
bfa4a6ee |
2871 | // '=' needs special treatment.. |
2872 | static char *next_word_s(char *w, size_t wsize, char *s) |
2873 | { |
2874 | size_t i; |
2875 | |
2876 | s = sskip(s); |
2877 | |
2878 | for (i = 0; i < wsize - 1; i++) { |
2879 | if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0)) |
2880 | break; |
2881 | w[i] = s[i]; |
2882 | } |
2883 | w[i] = 0; |
2884 | |
2885 | if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=') |
2886 | printf("warning: '%s' truncated\n", w); |
2887 | |
2888 | return s + i; |
2889 | } |
2890 | |
2891 | static int cmpstringp(const void *p1, const void *p2) |
2892 | { |
2893 | return strcmp(*(char * const *)p1, *(char * const *)p2); |
2894 | } |
2895 | |
91977a1c |
2896 | int main(int argc, char *argv[]) |
2897 | { |
06c5d854 |
2898 | FILE *fout, *fasm, *frlist; |
4c45fa73 |
2899 | struct parsed_data *pd = NULL; |
2900 | int pd_alloc = 0; |
2901 | char **rlist = NULL; |
2902 | int rlist_len = 0; |
2903 | int rlist_alloc = 0; |
91977a1c |
2904 | char line[256]; |
2905 | char words[16][256]; |
4c45fa73 |
2906 | enum opr_lenmod lmod; |
91977a1c |
2907 | int in_func = 0; |
4c45fa73 |
2908 | int pending_endp = 0; |
bfa4a6ee |
2909 | int skip_func = 0; |
940e8e66 |
2910 | int skip_warned = 0; |
91977a1c |
2911 | int eq_alloc; |
bfa4a6ee |
2912 | int verbose = 0; |
2913 | int arg_out; |
2914 | int arg = 1; |
91977a1c |
2915 | int pi = 0; |
4c45fa73 |
2916 | int i, j, len; |
91977a1c |
2917 | char *p; |
2918 | int wordc; |
2919 | |
bfa4a6ee |
2920 | if (argv[1] && IS(argv[1], "-v")) { |
2921 | verbose = 1; |
2922 | arg++; |
2923 | } |
2924 | |
2925 | if (argc < arg + 3) { |
2926 | printf("usage:\n%s [-v] <.c> <.asm> <hdrf> [rlist]*\n", |
91977a1c |
2927 | argv[0]); |
2928 | return 1; |
2929 | } |
2930 | |
bfa4a6ee |
2931 | arg_out = arg++; |
91977a1c |
2932 | |
bfa4a6ee |
2933 | asmfn = argv[arg++]; |
91977a1c |
2934 | fasm = fopen(asmfn, "r"); |
2935 | my_assert_not(fasm, NULL); |
2936 | |
bfa4a6ee |
2937 | hdrfn = argv[arg++]; |
06c5d854 |
2938 | g_fhdr = fopen(hdrfn, "r"); |
2939 | my_assert_not(g_fhdr, NULL); |
bfa4a6ee |
2940 | |
2941 | rlist_alloc = 64; |
2942 | rlist = malloc(rlist_alloc * sizeof(rlist[0])); |
2943 | my_assert_not(rlist, NULL); |
2944 | // needs special handling.. |
2945 | rlist[rlist_len++] = "__alloca_probe"; |
2946 | |
2947 | for (; arg < argc; arg++) { |
2948 | frlist = fopen(argv[arg], "r"); |
2949 | my_assert_not(frlist, NULL); |
2950 | |
2951 | while (fgets(line, sizeof(line), frlist)) { |
2952 | p = sskip(line); |
2953 | if (*p == 0 || *p == ';' || *p == '#') |
2954 | continue; |
2955 | |
2956 | p = next_word(words[0], sizeof(words[0]), p); |
2957 | if (words[0][0] == 0) |
2958 | continue; |
2959 | |
2960 | if (rlist_len >= rlist_alloc) { |
2961 | rlist_alloc = rlist_alloc * 2 + 64; |
2962 | rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0])); |
2963 | my_assert_not(rlist, NULL); |
2964 | } |
2965 | rlist[rlist_len++] = strdup(words[0]); |
2966 | } |
2967 | |
2968 | fclose(frlist); |
2969 | frlist = NULL; |
2970 | } |
2971 | |
2972 | if (rlist_len > 0) |
2973 | qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp); |
2974 | |
2975 | fout = fopen(argv[arg_out], "w"); |
91977a1c |
2976 | my_assert_not(fout, NULL); |
2977 | |
2978 | eq_alloc = 128; |
2979 | g_eqs = malloc(eq_alloc * sizeof(g_eqs[0])); |
2980 | my_assert_not(g_eqs, NULL); |
2981 | |
4c45fa73 |
2982 | for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) { |
2983 | g_label_refs[i].i = -1; |
2984 | g_label_refs[i].next = NULL; |
2985 | } |
2986 | |
91977a1c |
2987 | while (fgets(line, sizeof(line), fasm)) |
2988 | { |
4c45fa73 |
2989 | wordc = 0; |
91977a1c |
2990 | asmln++; |
2991 | |
2992 | p = sskip(line); |
1bafb621 |
2993 | if (*p == 0) |
91977a1c |
2994 | continue; |
2995 | |
1bafb621 |
2996 | if (*p == ';') { |
2997 | static const char *attrs[] = { |
2998 | "bp-based frame", |
2999 | "library function", |
3000 | "static", |
3001 | "noreturn", |
3002 | "thunk", |
3003 | "fpd=", |
3004 | }; |
4c45fa73 |
3005 | if (p[2] == '=' && IS_START(p, "; =============== S U B")) |
3006 | goto do_pending_endp; // eww.. |
3007 | |
3008 | if (p[2] != 'A' || !IS_START(p, "; Attributes:")) |
1bafb621 |
3009 | continue; |
3010 | |
3011 | // parse IDA's attribute-list comment |
4c45fa73 |
3012 | g_ida_func_attr = 0; |
1bafb621 |
3013 | p = sskip(p + 13); |
3014 | // get rid of random tabs |
3015 | for (i = 0; p[i] != 0; i++) |
3016 | if (p[i] == '\t') |
3017 | p[i] = ' '; |
3018 | |
3019 | for (; *p != 0; p = sskip(p)) { |
3020 | for (i = 0; i < ARRAY_SIZE(attrs); i++) { |
3021 | if (!strncmp(p, attrs[i], strlen(attrs[i]))) { |
4c45fa73 |
3022 | g_ida_func_attr |= 1 << i; |
1bafb621 |
3023 | p += strlen(attrs[i]); |
3024 | break; |
3025 | } |
3026 | } |
3027 | if (i == ARRAY_SIZE(attrs)) { |
3028 | anote("unparsed IDA attr: %s\n", p); |
3029 | break; |
3030 | } |
3031 | if (IS(attrs[i], "fpd=")) { |
3032 | p = next_word(words[0], sizeof(words[0]), p); |
3033 | // ignore for now.. |
3034 | } |
3035 | } |
3036 | continue; |
3037 | } |
3038 | |
06c5d854 |
3039 | parse_words: |
91977a1c |
3040 | memset(words, 0, sizeof(words)); |
3041 | for (wordc = 0; wordc < 16; wordc++) { |
bfa4a6ee |
3042 | p = sskip(next_word_s(words[wordc], sizeof(words[0]), p)); |
91977a1c |
3043 | if (*p == 0 || *p == ';') { |
3044 | wordc++; |
3045 | break; |
3046 | } |
3047 | } |
3048 | |
06c5d854 |
3049 | // alow asm patches in comments |
3050 | if (*p == ';' && IS_START(p, "; sctpatch: ")) { |
3051 | p = sskip(p + 12); |
3052 | if (*p == 0 || *p == ';') |
3053 | continue; |
3054 | goto parse_words; // lame |
3055 | } |
3056 | |
91977a1c |
3057 | if (wordc == 0) { |
3058 | // shouldn't happen |
3059 | awarn("wordc == 0?\n"); |
3060 | continue; |
3061 | } |
3062 | |
3063 | // don't care about this: |
3064 | if (words[0][0] == '.' |
3065 | || IS(words[0], "include") |
3066 | || IS(words[0], "assume") || IS(words[1], "segment") |
3067 | || IS(words[0], "align")) |
3068 | { |
3069 | continue; |
3070 | } |
3071 | |
4c45fa73 |
3072 | do_pending_endp: |
3073 | // do delayed endp processing to collect switch jumptables |
3074 | if (pending_endp) { |
3075 | if (in_func && !skip_func && wordc >= 2 |
3076 | && ((words[0][0] == 'd' && words[0][2] == 0) |
3077 | || (words[1][0] == 'd' && words[1][2] == 0))) |
3078 | { |
3079 | i = 1; |
3080 | if (words[1][0] == 'd' && words[1][2] == 0) { |
3081 | // label |
3082 | if (g_func_pd_cnt >= pd_alloc) { |
3083 | pd_alloc = pd_alloc * 2 + 16; |
3084 | g_func_pd = realloc(g_func_pd, |
3085 | sizeof(g_func_pd[0]) * pd_alloc); |
3086 | my_assert_not(g_func_pd, NULL); |
3087 | } |
3088 | pd = &g_func_pd[g_func_pd_cnt]; |
3089 | g_func_pd_cnt++; |
3090 | memset(pd, 0, sizeof(*pd)); |
3091 | strcpy(pd->label, words[0]); |
3092 | pd->type = OPT_CONST; |
3093 | pd->lmod = lmod_from_directive(words[1]); |
3094 | i = 2; |
3095 | } |
3096 | else { |
3097 | lmod = lmod_from_directive(words[0]); |
3098 | if (lmod != pd->lmod) |
3099 | aerr("lmod change? %d->%d\n", pd->lmod, lmod); |
3100 | } |
3101 | |
3102 | if (pd->count_alloc < pd->count + wordc) { |
3103 | pd->count_alloc = pd->count_alloc * 2 + 14 + wordc; |
3104 | pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc); |
3105 | my_assert_not(pd->d, NULL); |
3106 | } |
3107 | for (; i < wordc; i++) { |
3108 | if (IS(words[i], "offset")) { |
3109 | pd->type = OPT_OFFSET; |
3110 | i++; |
3111 | } |
3112 | p = strchr(words[i], ','); |
3113 | if (p != NULL) |
3114 | *p = 0; |
3115 | if (pd->type == OPT_OFFSET) |
3116 | pd->d[pd->count].u.label = strdup(words[i]); |
3117 | else |
3118 | pd->d[pd->count].u.val = parse_number(words[i]); |
3119 | pd->d[pd->count].bt_i = -1; |
3120 | pd->count++; |
3121 | } |
3122 | continue; |
3123 | } |
3124 | |
3125 | if (in_func && !skip_func) |
3126 | gen_func(fout, g_fhdr, g_func, pi); |
3127 | |
3128 | pending_endp = 0; |
3129 | in_func = 0; |
3130 | g_ida_func_attr = 0; |
3131 | skip_warned = 0; |
3132 | skip_func = 0; |
3133 | g_func[0] = 0; |
3134 | if (pi != 0) { |
3135 | memset(&ops, 0, pi * sizeof(ops[0])); |
3136 | memset(g_labels, 0, pi * sizeof(g_labels[0])); |
3137 | pi = 0; |
3138 | } |
3139 | g_eqcnt = 0; |
3140 | for (i = 0; i < g_func_pd_cnt; i++) { |
3141 | pd = &g_func_pd[i]; |
3142 | if (pd->type == OPT_OFFSET) { |
3143 | for (j = 0; j < pd->count; j++) |
3144 | free(pd->d[j].u.label); |
3145 | } |
3146 | free(pd->d); |
3147 | pd->d = NULL; |
3148 | } |
3149 | g_func_pd_cnt = 0; |
3150 | pd = NULL; |
3151 | if (wordc == 0) |
3152 | continue; |
3153 | } |
3154 | |
91977a1c |
3155 | if (IS(words[1], "proc")) { |
3156 | if (in_func) |
3157 | aerr("proc '%s' while in_func '%s'?\n", |
3158 | words[0], g_func); |
bfa4a6ee |
3159 | p = words[0]; |
4c45fa73 |
3160 | if ((g_ida_func_attr & IDAFA_THUNK) |
1bafb621 |
3161 | || bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp)) |
bfa4a6ee |
3162 | skip_func = 1; |
91977a1c |
3163 | strcpy(g_func, words[0]); |
d4e3b5db |
3164 | set_label(0, words[0]); |
91977a1c |
3165 | in_func = 1; |
3166 | continue; |
3167 | } |
3168 | |
3169 | if (IS(words[1], "endp")) { |
3170 | if (!in_func) |
3171 | aerr("endp '%s' while not in_func?\n", words[0]); |
3172 | if (!IS(g_func, words[0])) |
3173 | aerr("endp '%s' while in_func '%s'?\n", |
3174 | words[0], g_func); |
bfa4a6ee |
3175 | |
4c45fa73 |
3176 | pending_endp = 1; |
91977a1c |
3177 | continue; |
3178 | } |
3179 | |
bfa4a6ee |
3180 | p = strchr(words[0], ':'); |
3181 | if (p != NULL) { |
d4e3b5db |
3182 | set_label(pi, words[0]); |
bfa4a6ee |
3183 | continue; |
3184 | } |
3185 | |
3186 | if (!in_func || skip_func) { |
3187 | if (!skip_warned && !skip_func && g_labels[pi][0] != 0) { |
3188 | if (verbose) |
3189 | anote("skipping from '%s'\n", g_labels[pi]); |
3190 | skip_warned = 1; |
3191 | } |
3192 | g_labels[pi][0] = 0; |
3193 | continue; |
3194 | } |
3195 | |
4c45fa73 |
3196 | if (wordc > 1 && IS(words[1], "=")) { |
91977a1c |
3197 | if (wordc != 5) |
3198 | aerr("unhandled equ, wc=%d\n", wordc); |
3199 | if (g_eqcnt >= eq_alloc) { |
3200 | eq_alloc *= 2; |
3201 | g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0])); |
3202 | my_assert_not(g_eqs, NULL); |
3203 | } |
3204 | |
3205 | len = strlen(words[0]); |
3206 | if (len > sizeof(g_eqs[0].name) - 1) |
3207 | aerr("equ name too long: %d\n", len); |
3208 | strcpy(g_eqs[g_eqcnt].name, words[0]); |
3209 | |
3210 | if (!IS(words[3], "ptr")) |
3211 | aerr("unhandled equ\n"); |
3212 | if (IS(words[2], "dword")) |
3213 | g_eqs[g_eqcnt].lmod = OPLM_DWORD; |
3214 | else if (IS(words[2], "word")) |
3215 | g_eqs[g_eqcnt].lmod = OPLM_WORD; |
3216 | else if (IS(words[2], "byte")) |
3217 | g_eqs[g_eqcnt].lmod = OPLM_BYTE; |
3218 | else |
3219 | aerr("bad lmod: '%s'\n", words[2]); |
3220 | |
3221 | g_eqs[g_eqcnt].offset = parse_number(words[4]); |
3222 | g_eqcnt++; |
3223 | continue; |
3224 | } |
3225 | |
3226 | if (pi >= ARRAY_SIZE(ops)) |
3227 | aerr("too many ops\n"); |
3228 | |
91977a1c |
3229 | parse_op(&ops[pi], words, wordc); |
3230 | pi++; |
91977a1c |
3231 | } |
3232 | |
3233 | fclose(fout); |
3234 | fclose(fasm); |
06c5d854 |
3235 | fclose(g_fhdr); |
91977a1c |
3236 | |
3237 | return 0; |
c36e914d |
3238 | } |
91977a1c |
3239 | |
3240 | // vim:ts=2:shiftwidth=2:expandtab |