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 | |
054f95b2 |
29 | #include "masm_tools.h" |
30 | |
69a3cdfc |
31 | enum op_flags { |
87bf6cec |
32 | OPF_RMD = (1 << 0), /* removed or optimized out */ |
33 | OPF_DATA = (1 << 1), /* data processing - writes to dst opr */ |
34 | OPF_FLAGS = (1 << 2), /* sets flags */ |
35 | OPF_JMP = (1 << 3), /* branches, ret and call */ |
36 | OPF_CC = (1 << 4), /* uses flags */ |
37 | OPF_TAIL = (1 << 5), /* ret or tail call */ |
7ba45c34 |
38 | OPF_RSAVE = (1 << 6), /* push/pop is local reg save/load */ |
39 | OPF_REP = (1 << 7), /* prefixed by rep */ |
40 | OPF_REPZ = (1 << 8), /* rep is repe/repz */ |
41 | OPF_REPNZ = (1 << 9), /* rep is repne/repnz */ |
de50b98b |
42 | OPF_FARG = (1 << 10), /* push collected as func arg (no reuse) */ |
a3684be1 |
43 | OPF_EBP_S = (1 << 11), /* ebp used as scratch, not BP */ |
c36e914d |
44 | }; |
45 | |
46 | enum op_op { |
47 | OP_INVAL, |
33c35af6 |
48 | OP_NOP, |
c36e914d |
49 | OP_PUSH, |
50 | OP_POP, |
51 | OP_MOV, |
850c9265 |
52 | OP_LEA, |
53 | OP_MOVZX, |
54 | OP_MOVSX, |
55 | OP_NOT, |
5101a5f9 |
56 | OP_CDQ, |
33c35af6 |
57 | OP_STOS, |
d4e3b5db |
58 | OP_MOVS, |
7ba45c34 |
59 | OP_CMPS, |
c36e914d |
60 | OP_RET, |
61 | OP_ADD, |
91977a1c |
62 | OP_SUB, |
850c9265 |
63 | OP_AND, |
64 | OP_OR, |
65 | OP_XOR, |
66 | OP_SHL, |
67 | OP_SHR, |
68 | OP_SAR, |
d4e3b5db |
69 | OP_ROL, |
70 | OP_ROR, |
69a3cdfc |
71 | OP_ADC, |
850c9265 |
72 | OP_SBB, |
73 | OP_INC, |
74 | OP_DEC, |
5101a5f9 |
75 | OP_NEG, |
850c9265 |
76 | OP_MUL, |
77 | OP_IMUL, |
5101a5f9 |
78 | OP_DIV, |
79 | OP_IDIV, |
c36e914d |
80 | OP_TEST, |
81 | OP_CMP, |
82 | OP_CALL, |
83 | OP_JMP, |
84 | OP_JO, |
85 | OP_JNO, |
86 | OP_JC, |
87 | OP_JNC, |
88 | OP_JZ, |
89 | OP_JNZ, |
90 | OP_JBE, |
91 | OP_JA, |
92 | OP_JS, |
93 | OP_JNS, |
94 | OP_JP, |
95 | OP_JNP, |
96 | OP_JL, |
97 | OP_JGE, |
98 | OP_JLE, |
99 | OP_JG, |
100 | }; |
101 | |
102 | enum opr_type { |
87bf6cec |
103 | OPT_UNSPEC, |
104 | OPT_REG, |
105 | OPT_REGMEM, |
106 | OPT_LABEL, |
850c9265 |
107 | OPT_OFFSET, |
87bf6cec |
108 | OPT_CONST, |
c36e914d |
109 | }; |
110 | |
2b43685d |
111 | // must be sorted (larger len must be further in enum) |
c36e914d |
112 | enum opr_lenmod { |
91977a1c |
113 | OPLM_UNSPEC, |
114 | OPLM_BYTE, |
115 | OPLM_WORD, |
116 | OPLM_DWORD, |
c36e914d |
117 | }; |
118 | |
850c9265 |
119 | #define MAX_OPERANDS 3 |
c36e914d |
120 | |
121 | struct parsed_opr { |
91977a1c |
122 | enum opr_type type; |
123 | enum opr_lenmod lmod; |
7ba45c34 |
124 | unsigned int is_ptr:1; // pointer in C |
125 | unsigned int is_array:1; // array in C |
a3684be1 |
126 | unsigned int type_from_var:1; // .. in header, sometimes wrong |
2b43685d |
127 | unsigned int size_mismatch:1; // type override differs from C |
128 | unsigned int size_lt:1; // type override is larger than C |
91977a1c |
129 | int reg; |
130 | unsigned int val; |
131 | char name[256]; |
c36e914d |
132 | }; |
133 | |
134 | struct parsed_op { |
91977a1c |
135 | enum op_op op; |
136 | struct parsed_opr operand[MAX_OPERANDS]; |
69a3cdfc |
137 | unsigned int flags; |
91977a1c |
138 | int operand_cnt; |
69a3cdfc |
139 | int regmask_src; // all referensed regs |
140 | int regmask_dst; |
940e8e66 |
141 | int pfomask; // flagop: parsed_flag_op that can't be delayed |
de50b98b |
142 | int argnum; // push: altered before call arg # |
940e8e66 |
143 | int cc_scratch; // scratch storage during analysis |
4c45fa73 |
144 | int bt_i; // branch target for branches |
145 | struct parsed_data *btj;// branch targets for jumptables |
91977a1c |
146 | void *datap; |
147 | }; |
148 | |
69a3cdfc |
149 | // datap: |
69a3cdfc |
150 | // OP_CALL - ptr to parsed_proto |
2b43685d |
151 | // (OPF_CC) - point to one of (OPF_FLAGS) that affects cc op |
69a3cdfc |
152 | |
91977a1c |
153 | struct parsed_equ { |
154 | char name[64]; |
155 | enum opr_lenmod lmod; |
156 | int offset; |
c36e914d |
157 | }; |
158 | |
4c45fa73 |
159 | struct parsed_data { |
160 | char label[256]; |
161 | enum opr_type type; |
162 | enum opr_lenmod lmod; |
163 | int count; |
164 | int count_alloc; |
165 | struct { |
166 | union { |
167 | char *label; |
63df67be |
168 | unsigned int val; |
4c45fa73 |
169 | } u; |
170 | int bt_i; |
171 | } *d; |
172 | }; |
173 | |
174 | struct label_ref { |
175 | int i; |
176 | struct label_ref *next; |
177 | }; |
178 | |
1bafb621 |
179 | enum ida_func_attr { |
180 | IDAFA_BP_FRAME = (1 << 0), |
181 | IDAFA_LIB_FUNC = (1 << 1), |
182 | IDAFA_STATIC = (1 << 2), |
183 | IDAFA_NORETURN = (1 << 3), |
184 | IDAFA_THUNK = (1 << 4), |
185 | IDAFA_FPD = (1 << 5), |
186 | }; |
187 | |
188 | #define MAX_OPS 4096 |
c36e914d |
189 | |
190 | static struct parsed_op ops[MAX_OPS]; |
91977a1c |
191 | static struct parsed_equ *g_eqs; |
192 | static int g_eqcnt; |
193 | static char g_labels[MAX_OPS][32]; |
4c45fa73 |
194 | static struct label_ref g_label_refs[MAX_OPS]; |
bd96f656 |
195 | static const struct parsed_proto *g_func_pp; |
4c45fa73 |
196 | static struct parsed_data *g_func_pd; |
197 | static int g_func_pd_cnt; |
91977a1c |
198 | static char g_func[256]; |
199 | static char g_comment[256]; |
200 | static int g_bp_frame; |
1bafb621 |
201 | static int g_sp_frame; |
a2c1d768 |
202 | static int g_stack_frame_used; |
1bafb621 |
203 | static int g_stack_fsz; |
4c45fa73 |
204 | static int g_ida_func_attr; |
91977a1c |
205 | #define ferr(op_, fmt, ...) do { \ |
63df67be |
206 | printf("error:%s:#%zd: '%s': " fmt, g_func, (op_) - ops, \ |
850c9265 |
207 | dump_op(op_), ##__VA_ARGS__); \ |
33c35af6 |
208 | fcloseall(); \ |
91977a1c |
209 | exit(1); \ |
210 | } while (0) |
de50b98b |
211 | #define fnote(op_, fmt, ...) \ |
212 | printf("error:%s:#%zd: '%s': " fmt, g_func, (op_) - ops, \ |
213 | dump_op(op_), ##__VA_ARGS__) |
91977a1c |
214 | |
215 | #define MAX_REGS 8 |
216 | |
217 | const char *regs_r32[] = { "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp" }; |
218 | const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" }; |
219 | const char *regs_r8l[] = { "al", "bl", "cl", "dl" }; |
220 | const char *regs_r8h[] = { "ah", "bh", "ch", "dh" }; |
221 | |
222 | enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP }; |
223 | |
69a3cdfc |
224 | // possible basic comparison types (without inversion) |
225 | enum parsed_flag_op { |
226 | PFO_O, // 0 OF=1 |
227 | PFO_C, // 2 CF=1 |
228 | PFO_Z, // 4 ZF=1 |
229 | PFO_BE, // 6 CF=1||ZF=1 |
230 | PFO_S, // 8 SF=1 |
231 | PFO_P, // a PF=1 |
232 | PFO_L, // c SF!=OF |
233 | PFO_LE, // e ZF=1||SF!=OF |
234 | }; |
235 | |
236 | static const char *parsed_flag_op_names[] = { |
237 | "o", "c", "z", "be", "s", "p", "l", "le" |
238 | }; |
239 | |
91977a1c |
240 | static int char_array_i(const char *array[], size_t len, const char *s) |
241 | { |
242 | int i; |
c36e914d |
243 | |
91977a1c |
244 | for (i = 0; i < len; i++) |
245 | if (IS(s, array[i])) |
246 | return i; |
c36e914d |
247 | |
91977a1c |
248 | return -1; |
249 | } |
250 | |
63df67be |
251 | static void printf_number(char *buf, size_t buf_size, |
252 | unsigned long number) |
91977a1c |
253 | { |
5101a5f9 |
254 | // output in C-friendly form |
255 | snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number); |
256 | } |
91977a1c |
257 | |
5101a5f9 |
258 | static int parse_reg(enum opr_lenmod *reg_lmod, const char *s) |
259 | { |
260 | int reg; |
91977a1c |
261 | |
5101a5f9 |
262 | reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s); |
263 | if (reg >= 0) { |
264 | *reg_lmod = OPLM_DWORD; |
265 | return reg; |
91977a1c |
266 | } |
5101a5f9 |
267 | reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s); |
268 | if (reg >= 0) { |
269 | *reg_lmod = OPLM_WORD; |
270 | return reg; |
271 | } |
272 | reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s); |
273 | if (reg >= 0) { |
274 | *reg_lmod = OPLM_BYTE; |
275 | return reg; |
276 | } |
277 | reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s); |
278 | if (reg >= 0) { |
279 | *reg_lmod = OPLM_BYTE; |
280 | return reg; |
850c9265 |
281 | } |
282 | |
283 | return -1; |
91977a1c |
284 | } |
285 | |
5101a5f9 |
286 | static int parse_indmode(char *name, int *regmask, int need_c_cvt) |
287 | { |
288 | enum opr_lenmod lmod; |
289 | char cvtbuf[256]; |
290 | char *d = cvtbuf; |
291 | char *s = name; |
292 | char w[64]; |
293 | long number; |
294 | int reg; |
295 | int c = 0; |
296 | |
297 | *d = 0; |
298 | |
299 | while (*s != 0) { |
300 | d += strlen(d); |
301 | while (my_isblank(*s)) |
302 | s++; |
303 | for (; my_issep(*s); d++, s++) |
304 | *d = *s; |
305 | while (my_isblank(*s)) |
306 | s++; |
307 | *d = 0; |
308 | |
87bf6cec |
309 | // skip 'ds:' prefix |
39b168b8 |
310 | if (IS_START(s, "ds:")) |
87bf6cec |
311 | s += 3; |
312 | |
5101a5f9 |
313 | s = next_idt(w, sizeof(w), s); |
314 | if (w[0] == 0) |
315 | break; |
316 | c++; |
317 | |
318 | reg = parse_reg(&lmod, w); |
319 | if (reg >= 0) { |
320 | *regmask |= 1 << reg; |
321 | goto pass; |
322 | } |
323 | |
324 | if ('0' <= w[0] && w[0] <= '9') { |
325 | number = parse_number(w); |
326 | printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number); |
327 | continue; |
328 | } |
329 | |
330 | // probably some label/identifier - pass |
331 | |
332 | pass: |
333 | snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w); |
334 | } |
335 | |
336 | if (need_c_cvt) |
337 | strcpy(name, cvtbuf); |
338 | |
339 | return c; |
340 | } |
341 | |
4c45fa73 |
342 | static int is_reg_in_str(const char *s) |
343 | { |
344 | int i; |
345 | |
346 | if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3]))) |
347 | return 0; |
348 | |
349 | for (i = 0; i < ARRAY_SIZE(regs_r32); i++) |
350 | if (!strncmp(s, regs_r32[i], 3)) |
351 | return 1; |
352 | |
353 | return 0; |
354 | } |
355 | |
356 | static const char *parse_stack_el(const char *name, char *extra_reg) |
d4e3b5db |
357 | { |
4c45fa73 |
358 | const char *p, *p2, *s; |
1bafb621 |
359 | char *endp = NULL; |
d4e3b5db |
360 | char buf[32]; |
1bafb621 |
361 | long val; |
d4e3b5db |
362 | int len; |
363 | |
4c45fa73 |
364 | p = name; |
365 | if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) { |
366 | p += 4; |
367 | if (extra_reg != NULL) { |
368 | strncpy(extra_reg, name, 3); |
369 | extra_reg[4] = 0; |
370 | } |
d4e3b5db |
371 | } |
4c45fa73 |
372 | |
373 | if (IS_START(p, "ebp+")) { |
374 | p += 4; |
375 | |
376 | p2 = strchr(p, '+'); |
377 | if (p2 != NULL && is_reg_in_str(p)) { |
378 | if (extra_reg != NULL) { |
379 | strncpy(extra_reg, p, p2 - p); |
380 | extra_reg[p2 - p] = 0; |
381 | } |
382 | p = p2 + 1; |
383 | } |
384 | |
385 | if (!('0' <= *p && *p <= '9')) |
386 | return p; |
387 | |
388 | return NULL; |
389 | } |
390 | |
39b168b8 |
391 | if (!IS_START(name, "esp+")) |
d4e3b5db |
392 | return NULL; |
393 | |
394 | p = strchr(name + 4, '+'); |
395 | if (p) { |
1bafb621 |
396 | // must be a number after esp+, already converted to 0x.. |
397 | s = name + 4; |
398 | if (!('0' <= *s && *s <= '9')) { |
399 | aerr("%s nan?\n", __func__); |
d4e3b5db |
400 | return NULL; |
1bafb621 |
401 | } |
402 | if (s[0] == '0' && s[1] == 'x') |
403 | s += 2; |
404 | len = p - s; |
d4e3b5db |
405 | if (len < sizeof(buf) - 1) { |
1bafb621 |
406 | strncpy(buf, s, len); |
d4e3b5db |
407 | buf[len] = 0; |
1bafb621 |
408 | val = strtol(buf, &endp, 16); |
409 | if (val == 0 || *endp != 0) { |
410 | aerr("%s num parse fail for '%s'\n", __func__, buf); |
411 | return NULL; |
412 | } |
d4e3b5db |
413 | } |
414 | p++; |
415 | } |
416 | else |
417 | p = name + 4; |
418 | |
419 | if ('0' <= *p && *p <= '9') |
420 | return NULL; |
421 | |
422 | return p; |
423 | } |
424 | |
850c9265 |
425 | static int guess_lmod_from_name(struct parsed_opr *opr) |
426 | { |
427 | if (!strncmp(opr->name, "dword_", 6)) { |
428 | opr->lmod = OPLM_DWORD; |
429 | return 1; |
430 | } |
431 | if (!strncmp(opr->name, "word_", 5)) { |
432 | opr->lmod = OPLM_WORD; |
433 | return 1; |
434 | } |
435 | if (!strncmp(opr->name, "byte_", 5)) { |
436 | opr->lmod = OPLM_BYTE; |
437 | return 1; |
438 | } |
439 | return 0; |
440 | } |
441 | |
3ebea2cf |
442 | static int guess_lmod_from_c_type(enum opr_lenmod *lmod, |
443 | const struct parsed_type *c_type) |
06c5d854 |
444 | { |
06c5d854 |
445 | static const char *dword_types[] = { |
a3684be1 |
446 | "int", "_DWORD", "UINT_PTR", |
447 | "DWORD", "HANDLE", "HWND", "HMODULE", |
840257f6 |
448 | "WPARAM", "LPARAM", "UINT", |
06c5d854 |
449 | }; |
450 | static const char *word_types[] = { |
2b43685d |
451 | "uint16_t", "int16_t", |
452 | "unsigned __int16", "__int16", |
06c5d854 |
453 | }; |
454 | static const char *byte_types[] = { |
2b43685d |
455 | "uint8_t", "int8_t", "char", |
456 | "unsigned __int8", "__int8", "BYTE", |
a3684be1 |
457 | "_UNKNOWN", |
06c5d854 |
458 | }; |
3ebea2cf |
459 | const char *n; |
06c5d854 |
460 | int i; |
461 | |
3ebea2cf |
462 | if (c_type->is_ptr) { |
463 | *lmod = OPLM_DWORD; |
06c5d854 |
464 | return 1; |
465 | } |
466 | |
3ebea2cf |
467 | n = skip_type_mod(c_type->name); |
06c5d854 |
468 | |
3ebea2cf |
469 | for (i = 0; i < ARRAY_SIZE(dword_types); i++) { |
470 | if (IS(n, dword_types[i])) { |
471 | *lmod = OPLM_DWORD; |
06c5d854 |
472 | return 1; |
473 | } |
474 | } |
475 | |
476 | for (i = 0; i < ARRAY_SIZE(word_types); i++) { |
3ebea2cf |
477 | if (IS(n, word_types[i])) { |
478 | *lmod = OPLM_WORD; |
06c5d854 |
479 | return 1; |
480 | } |
481 | } |
482 | |
483 | for (i = 0; i < ARRAY_SIZE(byte_types); i++) { |
3ebea2cf |
484 | if (IS(n, byte_types[i])) { |
485 | *lmod = OPLM_BYTE; |
06c5d854 |
486 | return 1; |
487 | } |
488 | } |
489 | |
06c5d854 |
490 | return 0; |
491 | } |
492 | |
4c45fa73 |
493 | static enum opr_type lmod_from_directive(const char *d) |
494 | { |
495 | if (IS(d, "dd")) |
496 | return OPLM_DWORD; |
497 | else if (IS(d, "dw")) |
498 | return OPLM_WORD; |
499 | else if (IS(d, "db")) |
500 | return OPLM_BYTE; |
501 | |
502 | aerr("unhandled directive: '%s'\n", d); |
503 | return OPLM_UNSPEC; |
504 | } |
505 | |
5101a5f9 |
506 | static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod, |
507 | int *regmask) |
508 | { |
509 | opr->type = OPT_REG; |
510 | opr->reg = reg; |
511 | opr->lmod = lmod; |
512 | *regmask |= 1 << reg; |
513 | } |
514 | |
39b168b8 |
515 | static struct parsed_equ *equ_find(struct parsed_op *po, const char *name, |
516 | int *extra_offs); |
87bf6cec |
517 | |
69a3cdfc |
518 | static int parse_operand(struct parsed_opr *opr, |
519 | int *regmask, int *regmask_indirect, |
1bafb621 |
520 | char words[16][256], int wordc, int w, unsigned int op_flags) |
c36e914d |
521 | { |
bd96f656 |
522 | const struct parsed_proto *pp; |
850c9265 |
523 | enum opr_lenmod tmplmod; |
63df67be |
524 | unsigned long number; |
850c9265 |
525 | int ret, len; |
1bafb621 |
526 | int wordc_in; |
39b168b8 |
527 | char *tmp; |
850c9265 |
528 | int i; |
c36e914d |
529 | |
1bafb621 |
530 | if (w >= wordc) |
531 | aerr("parse_operand w %d, wordc %d\n", w, wordc); |
c36e914d |
532 | |
1bafb621 |
533 | opr->reg = xUNSPEC; |
91977a1c |
534 | |
1bafb621 |
535 | for (i = w; i < wordc; i++) { |
536 | len = strlen(words[i]); |
537 | if (words[i][len - 1] == ',') { |
538 | words[i][len - 1] = 0; |
539 | wordc = i + 1; |
540 | break; |
541 | } |
542 | } |
c36e914d |
543 | |
1bafb621 |
544 | wordc_in = wordc - w; |
c36e914d |
545 | |
1bafb621 |
546 | if ((op_flags & OPF_JMP) && wordc_in > 0 |
547 | && !('0' <= words[w][0] && words[w][0] <= '9')) |
548 | { |
549 | const char *label = NULL; |
550 | |
551 | if (wordc_in == 3 && !strncmp(words[w], "near", 4) |
552 | && IS(words[w + 1], "ptr")) |
553 | label = words[w + 2]; |
554 | else if (wordc_in == 2 && IS(words[w], "short")) |
555 | label = words[w + 1]; |
556 | else if (wordc_in == 1 |
557 | && strchr(words[w], '[') == NULL |
558 | && parse_reg(&tmplmod, words[w]) < 0) |
559 | label = words[w]; |
560 | |
561 | if (label != NULL) { |
562 | opr->type = OPT_LABEL; |
39b168b8 |
563 | if (IS_START(label, "ds:")) |
564 | label += 3; |
1bafb621 |
565 | strcpy(opr->name, label); |
566 | return wordc; |
567 | } |
568 | } |
c36e914d |
569 | |
1bafb621 |
570 | if (wordc_in >= 3) { |
571 | if (IS(words[w + 1], "ptr")) { |
572 | if (IS(words[w], "dword")) |
573 | opr->lmod = OPLM_DWORD; |
574 | else if (IS(words[w], "word")) |
575 | opr->lmod = OPLM_WORD; |
576 | else if (IS(words[w], "byte")) |
577 | opr->lmod = OPLM_BYTE; |
578 | else |
579 | aerr("type parsing failed\n"); |
580 | w += 2; |
581 | wordc_in = wordc - w; |
582 | } |
583 | } |
584 | |
39b168b8 |
585 | if (wordc_in == 2) { |
586 | if (IS(words[w], "offset")) { |
587 | opr->type = OPT_OFFSET; |
588 | strcpy(opr->name, words[w + 1]); |
589 | return wordc; |
590 | } |
591 | if (IS(words[w], "(offset")) { |
592 | char *p = strchr(words[w + 1], ')'); |
593 | if (p == NULL) |
594 | aerr("parse of bracketed offset failed\n"); |
595 | *p = 0; |
596 | opr->type = OPT_OFFSET; |
597 | strcpy(opr->name, words[w + 1]); |
598 | return wordc; |
599 | } |
1bafb621 |
600 | } |
c36e914d |
601 | |
1bafb621 |
602 | if (wordc_in != 1) |
850c9265 |
603 | aerr("parse_operand 1 word expected\n"); |
c36e914d |
604 | |
39b168b8 |
605 | tmp = words[w]; |
606 | if (IS_START(tmp, "ds:")) |
607 | tmp += 3; |
608 | strcpy(opr->name, tmp); |
c36e914d |
609 | |
850c9265 |
610 | if (words[w][0] == '[') { |
611 | opr->type = OPT_REGMEM; |
612 | ret = sscanf(words[w], "[%[^]]]", opr->name); |
613 | if (ret != 1) |
614 | aerr("[] parse failure\n"); |
87bf6cec |
615 | |
5101a5f9 |
616 | parse_indmode(opr->name, regmask_indirect, 1); |
4c45fa73 |
617 | if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL)) { |
87bf6cec |
618 | // might be an equ |
d4e3b5db |
619 | struct parsed_equ *eq = |
4c45fa73 |
620 | equ_find(NULL, parse_stack_el(opr->name, NULL), &i); |
87bf6cec |
621 | if (eq) |
622 | opr->lmod = eq->lmod; |
623 | } |
850c9265 |
624 | return wordc; |
625 | } |
626 | else if (strchr(words[w], '[')) { |
627 | // label[reg] form |
628 | opr->type = OPT_REGMEM; |
629 | if (opr->lmod == OPLM_UNSPEC) |
630 | guess_lmod_from_name(opr); |
5101a5f9 |
631 | parse_indmode(strchr(words[w], '['), regmask_indirect, 0); |
850c9265 |
632 | return wordc; |
633 | } |
634 | else if (('0' <= words[w][0] && words[w][0] <= '9') |
635 | || words[w][0] == '-') |
636 | { |
5101a5f9 |
637 | number = parse_number(words[w]); |
91977a1c |
638 | opr->type = OPT_CONST; |
5101a5f9 |
639 | opr->val = number; |
640 | printf_number(opr->name, sizeof(opr->name), number); |
91977a1c |
641 | return wordc; |
850c9265 |
642 | } |
c36e914d |
643 | |
5101a5f9 |
644 | ret = parse_reg(&tmplmod, opr->name); |
645 | if (ret >= 0) { |
646 | setup_reg_opr(opr, ret, tmplmod, regmask); |
850c9265 |
647 | return wordc; |
648 | } |
649 | |
650 | // most likely var in data segment |
651 | opr->type = OPT_LABEL; |
06c5d854 |
652 | |
bd96f656 |
653 | pp = proto_parse(g_fhdr, opr->name); |
654 | if (pp != NULL) { |
1cd4a663 |
655 | if (pp->is_fptr || pp->is_func) { |
06c5d854 |
656 | opr->lmod = OPLM_DWORD; |
657 | opr->is_ptr = 1; |
658 | } |
2b43685d |
659 | else { |
840257f6 |
660 | tmplmod = OPLM_UNSPEC; |
bd96f656 |
661 | if (!guess_lmod_from_c_type(&tmplmod, &pp->type)) |
2b43685d |
662 | anote("unhandled C type '%s' for '%s'\n", |
bd96f656 |
663 | pp->type.name, opr->name); |
2b43685d |
664 | |
a3684be1 |
665 | if (opr->lmod == OPLM_UNSPEC) { |
2b43685d |
666 | opr->lmod = tmplmod; |
a3684be1 |
667 | opr->type_from_var = 1; |
668 | } |
2b43685d |
669 | else if (opr->lmod != tmplmod) { |
670 | opr->size_mismatch = 1; |
671 | if (tmplmod < opr->lmod) |
672 | opr->size_lt = 1; |
673 | } |
a652aa9f |
674 | opr->is_ptr = pp->type.is_ptr; |
3ebea2cf |
675 | } |
bd96f656 |
676 | opr->is_array = pp->type.is_array; |
06c5d854 |
677 | } |
678 | |
850c9265 |
679 | if (opr->lmod == OPLM_UNSPEC) |
680 | guess_lmod_from_name(opr); |
850c9265 |
681 | return wordc; |
c36e914d |
682 | } |
683 | |
33c35af6 |
684 | static const struct { |
685 | const char *name; |
686 | unsigned int flags; |
687 | } pref_table[] = { |
688 | { "rep", OPF_REP }, |
7ba45c34 |
689 | { "repe", OPF_REP|OPF_REPZ }, |
690 | { "repz", OPF_REP|OPF_REPZ }, |
691 | { "repne", OPF_REP|OPF_REPNZ }, |
692 | { "repnz", OPF_REP|OPF_REPNZ }, |
33c35af6 |
693 | }; |
694 | |
c36e914d |
695 | static const struct { |
850c9265 |
696 | const char *name; |
697 | enum op_op op; |
69a3cdfc |
698 | unsigned int minopr; |
699 | unsigned int maxopr; |
700 | unsigned int flags; |
c36e914d |
701 | } op_table[] = { |
33c35af6 |
702 | { "nop", OP_NOP, 0, 0, 0 }, |
69a3cdfc |
703 | { "push", OP_PUSH, 1, 1, 0 }, |
704 | { "pop", OP_POP, 1, 1, OPF_DATA }, |
705 | { "mov" , OP_MOV, 2, 2, OPF_DATA }, |
706 | { "lea", OP_LEA, 2, 2, OPF_DATA }, |
707 | { "movzx",OP_MOVZX, 2, 2, OPF_DATA }, |
708 | { "movsx",OP_MOVSX, 2, 2, OPF_DATA }, |
709 | { "not", OP_NOT, 1, 1, OPF_DATA }, |
5101a5f9 |
710 | { "cdq", OP_CDQ, 0, 0, OPF_DATA }, |
33c35af6 |
711 | { "stosb",OP_STOS, 0, 0, OPF_DATA }, |
712 | { "stosw",OP_STOS, 0, 0, OPF_DATA }, |
713 | { "stosd",OP_STOS, 0, 0, OPF_DATA }, |
d4e3b5db |
714 | { "movsb",OP_MOVS, 0, 0, OPF_DATA }, |
715 | { "movsw",OP_MOVS, 0, 0, OPF_DATA }, |
716 | { "movsd",OP_MOVS, 0, 0, OPF_DATA }, |
7ba45c34 |
717 | { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS }, |
718 | { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS }, |
719 | { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS }, |
69a3cdfc |
720 | { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS }, |
721 | { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS }, |
722 | { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS }, |
723 | { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS }, |
724 | { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS }, |
725 | { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS }, |
726 | { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS }, |
727 | { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS }, |
728 | { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS }, |
d4e3b5db |
729 | { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS }, |
730 | { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS }, |
5101a5f9 |
731 | { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC }, |
69a3cdfc |
732 | { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC }, |
733 | { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS }, |
734 | { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS }, |
5101a5f9 |
735 | { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS }, |
736 | { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS }, |
69a3cdfc |
737 | { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS }, |
5101a5f9 |
738 | { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS }, |
739 | { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS }, |
69a3cdfc |
740 | { "test", OP_TEST, 2, 2, OPF_FLAGS }, |
741 | { "cmp", OP_CMP, 2, 2, OPF_FLAGS }, |
a3684be1 |
742 | { "retn", OP_RET, 0, 1, OPF_TAIL }, |
de50b98b |
743 | { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS }, |
69a3cdfc |
744 | { "jmp", OP_JMP, 1, 1, OPF_JMP }, |
745 | { "jo", OP_JO, 1, 1, OPF_JMP|OPF_CC }, // 70 OF=1 |
746 | { "jno", OP_JNO, 1, 1, OPF_JMP|OPF_CC }, // 71 OF=0 |
747 | { "jc", OP_JC, 1, 1, OPF_JMP|OPF_CC }, // 72 CF=1 |
748 | { "jb", OP_JC, 1, 1, OPF_JMP|OPF_CC }, // 72 |
749 | { "jnc", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 CF=0 |
33c35af6 |
750 | { "jnb", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 |
69a3cdfc |
751 | { "jae", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 |
752 | { "jz", OP_JZ, 1, 1, OPF_JMP|OPF_CC }, // 74 ZF=1 |
753 | { "je", OP_JZ, 1, 1, OPF_JMP|OPF_CC }, // 74 |
754 | { "jnz", OP_JNZ, 1, 1, OPF_JMP|OPF_CC }, // 75 ZF=0 |
755 | { "jne", OP_JNZ, 1, 1, OPF_JMP|OPF_CC }, // 75 |
756 | { "jbe", OP_JBE, 1, 1, OPF_JMP|OPF_CC }, // 76 CF=1 || ZF=1 |
757 | { "jna", OP_JBE, 1, 1, OPF_JMP|OPF_CC }, // 76 |
758 | { "ja", OP_JA, 1, 1, OPF_JMP|OPF_CC }, // 77 CF=0 && ZF=0 |
759 | { "jnbe", OP_JA, 1, 1, OPF_JMP|OPF_CC }, // 77 |
760 | { "js", OP_JS, 1, 1, OPF_JMP|OPF_CC }, // 78 SF=1 |
761 | { "jns", OP_JNS, 1, 1, OPF_JMP|OPF_CC }, // 79 SF=0 |
762 | { "jp", OP_JP, 1, 1, OPF_JMP|OPF_CC }, // 7a PF=1 |
763 | { "jpe", OP_JP, 1, 1, OPF_JMP|OPF_CC }, // 7a |
764 | { "jnp", OP_JNP, 1, 1, OPF_JMP|OPF_CC }, // 7b PF=0 |
765 | { "jpo", OP_JNP, 1, 1, OPF_JMP|OPF_CC }, // 7b |
766 | { "jl", OP_JL, 1, 1, OPF_JMP|OPF_CC }, // 7c SF!=OF |
767 | { "jnge", OP_JL, 1, 1, OPF_JMP|OPF_CC }, // 7c |
768 | { "jge", OP_JGE, 1, 1, OPF_JMP|OPF_CC }, // 7d SF=OF |
769 | { "jnl", OP_JGE, 1, 1, OPF_JMP|OPF_CC }, // 7d |
770 | { "jle", OP_JLE, 1, 1, OPF_JMP|OPF_CC }, // 7e ZF=1 || SF!=OF |
771 | { "jng", OP_JLE, 1, 1, OPF_JMP|OPF_CC }, // 7e |
772 | { "jg", OP_JG, 1, 1, OPF_JMP|OPF_CC }, // 7f ZF=0 && SF=OF |
773 | { "jnle", OP_JG, 1, 1, OPF_JMP|OPF_CC }, // 7f |
5101a5f9 |
774 | { "seto", OP_JO, 1, 1, OPF_DATA|OPF_CC }, |
775 | { "setno", OP_JNO, 1, 1, OPF_DATA|OPF_CC }, |
776 | { "setc", OP_JC, 1, 1, OPF_DATA|OPF_CC }, |
777 | { "setb", OP_JC, 1, 1, OPF_DATA|OPF_CC }, |
778 | { "setnc", OP_JNC, 1, 1, OPF_DATA|OPF_CC }, |
779 | { "setae", OP_JNC, 1, 1, OPF_DATA|OPF_CC }, |
780 | { "setz", OP_JZ, 1, 1, OPF_DATA|OPF_CC }, |
781 | { "sete", OP_JZ, 1, 1, OPF_DATA|OPF_CC }, |
782 | { "setnz", OP_JNZ, 1, 1, OPF_DATA|OPF_CC }, |
783 | { "setne", OP_JNZ, 1, 1, OPF_DATA|OPF_CC }, |
784 | { "setbe", OP_JBE, 1, 1, OPF_DATA|OPF_CC }, |
785 | { "setna", OP_JBE, 1, 1, OPF_DATA|OPF_CC }, |
786 | { "seta", OP_JA, 1, 1, OPF_DATA|OPF_CC }, |
787 | { "setnbe", OP_JA, 1, 1, OPF_DATA|OPF_CC }, |
788 | { "sets", OP_JS, 1, 1, OPF_DATA|OPF_CC }, |
789 | { "setns", OP_JNS, 1, 1, OPF_DATA|OPF_CC }, |
790 | { "setp", OP_JP, 1, 1, OPF_DATA|OPF_CC }, |
791 | { "setpe", OP_JP, 1, 1, OPF_DATA|OPF_CC }, |
792 | { "setnp", OP_JNP, 1, 1, OPF_DATA|OPF_CC }, |
793 | { "setpo", OP_JNP, 1, 1, OPF_DATA|OPF_CC }, |
794 | { "setl", OP_JL, 1, 1, OPF_DATA|OPF_CC }, |
795 | { "setnge", OP_JL, 1, 1, OPF_DATA|OPF_CC }, |
796 | { "setge", OP_JGE, 1, 1, OPF_DATA|OPF_CC }, |
797 | { "setnl", OP_JGE, 1, 1, OPF_DATA|OPF_CC }, |
798 | { "setle", OP_JLE, 1, 1, OPF_DATA|OPF_CC }, |
799 | { "setng", OP_JLE, 1, 1, OPF_DATA|OPF_CC }, |
800 | { "setg", OP_JG, 1, 1, OPF_DATA|OPF_CC }, |
801 | { "setnle", OP_JG, 1, 1, OPF_DATA|OPF_CC }, |
69a3cdfc |
802 | }; |
c36e914d |
803 | |
804 | static void parse_op(struct parsed_op *op, char words[16][256], int wordc) |
805 | { |
bfa4a6ee |
806 | enum opr_lenmod lmod = OPLM_UNSPEC; |
33c35af6 |
807 | int prefix_flags = 0; |
69a3cdfc |
808 | int regmask_ind; |
809 | int regmask; |
33c35af6 |
810 | int op_w = 0; |
91977a1c |
811 | int opr = 0; |
33c35af6 |
812 | int w = 0; |
91977a1c |
813 | int i; |
c36e914d |
814 | |
33c35af6 |
815 | for (i = 0; i < ARRAY_SIZE(pref_table); i++) { |
816 | if (IS(words[w], pref_table[i].name)) { |
817 | prefix_flags = pref_table[i].flags; |
818 | break; |
819 | } |
820 | } |
821 | |
822 | if (prefix_flags) { |
823 | if (wordc <= 1) |
824 | aerr("lone prefix: '%s'\n", words[0]); |
825 | w++; |
826 | } |
827 | |
828 | op_w = w; |
91977a1c |
829 | for (i = 0; i < ARRAY_SIZE(op_table); i++) { |
33c35af6 |
830 | if (IS(words[w], op_table[i].name)) |
69a3cdfc |
831 | break; |
832 | } |
c36e914d |
833 | |
69a3cdfc |
834 | if (i == ARRAY_SIZE(op_table)) |
835 | aerr("unhandled op: '%s'\n", words[0]); |
33c35af6 |
836 | w++; |
c36e914d |
837 | |
69a3cdfc |
838 | op->op = op_table[i].op; |
33c35af6 |
839 | op->flags = op_table[i].flags | prefix_flags; |
69a3cdfc |
840 | op->regmask_src = op->regmask_dst = 0; |
841 | |
842 | for (opr = 0; opr < op_table[i].minopr; opr++) { |
843 | regmask = regmask_ind = 0; |
844 | w = parse_operand(&op->operand[opr], ®mask, ®mask_ind, |
845 | words, wordc, w, op->flags); |
846 | |
847 | if (opr == 0 && (op->flags & OPF_DATA)) |
848 | op->regmask_dst = regmask; |
849 | // for now, mark dst as src too |
850 | op->regmask_src |= regmask | regmask_ind; |
851 | } |
c36e914d |
852 | |
69a3cdfc |
853 | for (; w < wordc && opr < op_table[i].maxopr; opr++) { |
854 | w = parse_operand(&op->operand[opr], |
855 | &op->regmask_src, &op->regmask_src, |
856 | words, wordc, w, op->flags); |
91977a1c |
857 | } |
c36e914d |
858 | |
91977a1c |
859 | if (w < wordc) |
860 | aerr("parse_op %s incomplete: %d/%d\n", |
861 | words[0], w, wordc); |
5101a5f9 |
862 | |
863 | // special cases |
864 | op->operand_cnt = opr; |
865 | if (!strncmp(op_table[i].name, "set", 3)) |
866 | op->operand[0].lmod = OPLM_BYTE; |
867 | |
868 | // ops with implicit argumets |
869 | switch (op->op) { |
870 | case OP_CDQ: |
871 | op->operand_cnt = 2; |
872 | setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst); |
873 | setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src); |
874 | break; |
875 | |
33c35af6 |
876 | case OP_STOS: |
877 | if (op->operand_cnt != 0) |
878 | break; |
879 | if (IS(words[op_w], "stosb")) |
880 | lmod = OPLM_BYTE; |
881 | else if (IS(words[op_w], "stosw")) |
882 | lmod = OPLM_WORD; |
883 | else if (IS(words[op_w], "stosd")) |
884 | lmod = OPLM_DWORD; |
885 | op->operand_cnt = 3; |
d4e3b5db |
886 | setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src); |
887 | setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src); |
888 | op->regmask_dst = op->regmask_src; |
33c35af6 |
889 | setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD, &op->regmask_src); |
890 | break; |
891 | |
d4e3b5db |
892 | case OP_MOVS: |
7ba45c34 |
893 | case OP_CMPS: |
d4e3b5db |
894 | if (op->operand_cnt != 0) |
895 | break; |
7ba45c34 |
896 | if (words[op_w][4] == 'b') |
d4e3b5db |
897 | lmod = OPLM_BYTE; |
7ba45c34 |
898 | else if (words[op_w][4] == 'w') |
d4e3b5db |
899 | lmod = OPLM_WORD; |
7ba45c34 |
900 | else if (words[op_w][4] == 'd') |
d4e3b5db |
901 | lmod = OPLM_DWORD; |
902 | op->operand_cnt = 3; |
903 | setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src); |
904 | setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src); |
905 | setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src); |
906 | op->regmask_dst = op->regmask_src; |
907 | break; |
908 | |
5101a5f9 |
909 | case OP_IMUL: |
910 | if (op->operand_cnt != 1) |
911 | break; |
912 | // fallthrough |
913 | case OP_MUL: |
914 | // singleop mul |
915 | op->regmask_dst = (1 << xDX) | (1 << xAX); |
916 | op->regmask_src |= (1 << xAX); |
917 | if (op->operand[0].lmod == OPLM_UNSPEC) |
918 | op->operand[0].lmod = OPLM_DWORD; |
919 | break; |
920 | |
921 | case OP_DIV: |
922 | case OP_IDIV: |
923 | // we could set up operands for edx:eax, but there is no real need to |
924 | // (see is_opr_modified()) |
925 | regmask = (1 << xDX) | (1 << xAX); |
926 | op->regmask_dst = regmask; |
927 | op->regmask_src |= regmask; |
928 | if (op->operand[0].lmod == OPLM_UNSPEC) |
929 | op->operand[0].lmod = OPLM_DWORD; |
930 | break; |
931 | |
932 | case OP_SHL: |
933 | case OP_SHR: |
934 | case OP_SAR: |
d4e3b5db |
935 | case OP_ROL: |
936 | case OP_ROR: |
5101a5f9 |
937 | if (op->operand[1].lmod == OPLM_UNSPEC) |
938 | op->operand[1].lmod = OPLM_BYTE; |
939 | break; |
940 | |
7ba45c34 |
941 | case OP_PUSH: |
942 | if (op->operand[0].lmod == OPLM_UNSPEC |
943 | && (op->operand[0].type == OPT_CONST |
944 | || op->operand[0].type == OPT_OFFSET |
945 | || op->operand[0].type == OPT_LABEL)) |
946 | op->operand[0].lmod = OPLM_DWORD; |
947 | break; |
948 | |
3ebea2cf |
949 | // alignment |
950 | case OP_MOV: |
951 | if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG |
952 | && op->operand[0].reg == xDI && op->operand[1].reg == xDI) |
953 | { |
954 | op->flags |= OPF_RMD; |
955 | } |
956 | break; |
957 | |
de50b98b |
958 | case OP_LEA: |
959 | if (op->operand[0].type == OPT_REG |
960 | && op->operand[1].type == OPT_REGMEM) |
961 | { |
962 | char buf[16]; |
963 | snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name); |
964 | if (IS(buf, op->operand[1].name)) |
965 | op->flags |= OPF_RMD; |
966 | } |
967 | break; |
968 | |
5101a5f9 |
969 | default: |
970 | break; |
971 | } |
c36e914d |
972 | } |
973 | |
850c9265 |
974 | static const char *op_name(enum op_op op) |
975 | { |
976 | int i; |
977 | |
978 | for (i = 0; i < ARRAY_SIZE(op_table); i++) |
979 | if (op_table[i].op == op) |
980 | return op_table[i].name; |
981 | |
982 | return "???"; |
983 | } |
984 | |
985 | // debug |
986 | static const char *dump_op(struct parsed_op *po) |
987 | { |
988 | static char out[128]; |
989 | char *p = out; |
990 | int i; |
991 | |
4c45fa73 |
992 | if (po == NULL) |
993 | return "???"; |
994 | |
850c9265 |
995 | snprintf(out, sizeof(out), "%s", op_name(po->op)); |
996 | for (i = 0; i < po->operand_cnt; i++) { |
997 | p += strlen(p); |
998 | if (i > 0) |
999 | *p++ = ','; |
1000 | snprintf(p, sizeof(out) - (p - out), |
1001 | po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s", |
1002 | po->operand[i].name); |
1003 | } |
1004 | |
1005 | return out; |
1006 | } |
1007 | |
4c45fa73 |
1008 | static const char *lmod_type_u(struct parsed_op *po, |
1009 | enum opr_lenmod lmod) |
1010 | { |
1011 | switch (lmod) { |
1012 | case OPLM_DWORD: |
1013 | return "u32"; |
1014 | case OPLM_WORD: |
1015 | return "u16"; |
1016 | case OPLM_BYTE: |
1017 | return "u8"; |
1018 | default: |
1019 | ferr(po, "invalid lmod: %d\n", lmod); |
1020 | return "(_invalid_)"; |
1021 | } |
1022 | } |
1023 | |
d4e3b5db |
1024 | static const char *lmod_cast_u(struct parsed_op *po, |
1025 | enum opr_lenmod lmod) |
1026 | { |
1027 | switch (lmod) { |
1028 | case OPLM_DWORD: |
1029 | return ""; |
1030 | case OPLM_WORD: |
1031 | return "(u16)"; |
1032 | case OPLM_BYTE: |
1033 | return "(u8)"; |
1034 | default: |
1035 | ferr(po, "invalid lmod: %d\n", lmod); |
1036 | return "(_invalid_)"; |
1037 | } |
1038 | } |
1039 | |
1040 | static const char *lmod_cast_u_ptr(struct parsed_op *po, |
1041 | enum opr_lenmod lmod) |
1042 | { |
1043 | switch (lmod) { |
1044 | case OPLM_DWORD: |
1045 | return "*(u32 *)"; |
1046 | case OPLM_WORD: |
1047 | return "*(u16 *)"; |
1048 | case OPLM_BYTE: |
1049 | return "*(u8 *)"; |
1050 | default: |
1051 | ferr(po, "invalid lmod: %d\n", lmod); |
1052 | return "(_invalid_)"; |
1053 | } |
1054 | } |
1055 | |
1056 | static const char *lmod_cast_s(struct parsed_op *po, |
1057 | enum opr_lenmod lmod) |
1058 | { |
1059 | switch (lmod) { |
1060 | case OPLM_DWORD: |
1061 | return "(s32)"; |
1062 | case OPLM_WORD: |
1063 | return "(s16)"; |
1064 | case OPLM_BYTE: |
1065 | return "(s8)"; |
1066 | default: |
1067 | ferr(po, "%s: invalid lmod: %d\n", __func__, lmod); |
1068 | return "(_invalid_)"; |
1069 | } |
1070 | } |
1071 | |
1072 | static const char *lmod_cast(struct parsed_op *po, |
1073 | enum opr_lenmod lmod, int is_signed) |
1074 | { |
1075 | return is_signed ? |
1076 | lmod_cast_s(po, lmod) : |
1077 | lmod_cast_u(po, lmod); |
1078 | } |
1079 | |
1080 | static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod) |
1081 | { |
1082 | switch (lmod) { |
1083 | case OPLM_DWORD: |
1084 | return 4; |
1085 | case OPLM_WORD: |
1086 | return 2; |
1087 | case OPLM_BYTE: |
1088 | return 1; |
1089 | default: |
1090 | ferr(po, "%s: invalid lmod: %d\n", __func__, lmod); |
1091 | return 0; |
1092 | } |
1093 | } |
1094 | |
91977a1c |
1095 | static const char *opr_name(struct parsed_op *po, int opr_num) |
c36e914d |
1096 | { |
91977a1c |
1097 | if (opr_num >= po->operand_cnt) |
1098 | ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt); |
1099 | return po->operand[opr_num].name; |
c36e914d |
1100 | } |
1101 | |
91977a1c |
1102 | static unsigned int opr_const(struct parsed_op *po, int opr_num) |
c36e914d |
1103 | { |
91977a1c |
1104 | if (opr_num >= po->operand_cnt) |
1105 | ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt); |
1106 | if (po->operand[opr_num].type != OPT_CONST) |
1107 | ferr(po, "opr %d: const expected\n", opr_num); |
1108 | return po->operand[opr_num].val; |
1109 | } |
c36e914d |
1110 | |
91977a1c |
1111 | static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr) |
1112 | { |
1113 | if ((unsigned int)popr->reg >= MAX_REGS) |
1114 | ferr(po, "invalid reg: %d\n", popr->reg); |
1115 | return regs_r32[popr->reg]; |
1116 | } |
c36e914d |
1117 | |
a2c1d768 |
1118 | // cast1 is the "final" cast |
1119 | static const char *simplify_cast(const char *cast1, const char *cast2) |
1120 | { |
1121 | static char buf[256]; |
1122 | |
1123 | if (cast1[0] == 0) |
1124 | return cast2; |
1125 | if (cast2[0] == 0) |
1126 | return cast1; |
1127 | if (IS(cast1, cast2)) |
1128 | return cast1; |
1129 | if (IS(cast1, "(s8)") && IS(cast2, "(u8)")) |
1130 | return cast1; |
1131 | if (IS(cast1, "(s16)") && IS(cast2, "(u16)")) |
1132 | return cast1; |
1133 | if (IS(cast1, "(u8)") && IS_START(cast2, "*(u8 *)")) |
1134 | return cast2; |
1135 | if (IS(cast1, "(u16)") && IS_START(cast2, "*(u16 *)")) |
1136 | return cast2; |
1cd4a663 |
1137 | if (strchr(cast1, '*') && IS_START(cast2, "(u32)")) |
1138 | return cast1; |
a2c1d768 |
1139 | |
1140 | snprintf(buf, sizeof(buf), "%s%s", cast1, cast2); |
1141 | return buf; |
1142 | } |
1143 | |
1144 | static const char *simplify_cast_num(const char *cast, unsigned int val) |
1145 | { |
1146 | if (IS(cast, "(u8)") && val < 0x100) |
1147 | return ""; |
1148 | if (IS(cast, "(s8)") && val < 0x80) |
1149 | return ""; |
1150 | if (IS(cast, "(u16)") && val < 0x10000) |
1151 | return ""; |
1152 | if (IS(cast, "(s16)") && val < 0x8000) |
1153 | return ""; |
1154 | if (IS(cast, "(s32)") && val < 0x80000000) |
1155 | return ""; |
1156 | |
1157 | return cast; |
1158 | } |
1159 | |
39b168b8 |
1160 | static struct parsed_equ *equ_find(struct parsed_op *po, const char *name, |
1161 | int *extra_offs) |
91977a1c |
1162 | { |
39b168b8 |
1163 | const char *p; |
1164 | char *endp; |
1165 | int namelen; |
850c9265 |
1166 | int i; |
1167 | |
39b168b8 |
1168 | *extra_offs = 0; |
1169 | namelen = strlen(name); |
1170 | |
1171 | p = strchr(name, '+'); |
1172 | if (p != NULL) { |
1173 | namelen = p - name; |
1174 | if (namelen <= 0) |
1175 | ferr(po, "equ parse failed for '%s'\n", name); |
1176 | |
1177 | if (IS_START(p, "0x")) |
1178 | p += 2; |
1179 | *extra_offs = strtol(p, &endp, 16); |
1180 | if (*endp != 0) |
1181 | ferr(po, "equ parse failed for '%s'\n", name); |
1182 | } |
1183 | |
850c9265 |
1184 | for (i = 0; i < g_eqcnt; i++) |
39b168b8 |
1185 | if (strncmp(g_eqs[i].name, name, namelen) == 0 |
1186 | && g_eqs[i].name[namelen] == 0) |
850c9265 |
1187 | break; |
87bf6cec |
1188 | if (i >= g_eqcnt) { |
1189 | if (po != NULL) |
1190 | ferr(po, "unresolved equ name: '%s'\n", name); |
1191 | return NULL; |
1192 | } |
850c9265 |
1193 | |
1194 | return &g_eqs[i]; |
1195 | } |
1196 | |
1cd4a663 |
1197 | static int is_stack_access(struct parsed_op *po, |
1198 | const struct parsed_opr *popr) |
850c9265 |
1199 | { |
1cd4a663 |
1200 | return (parse_stack_el(popr->name, NULL) |
1201 | || (g_bp_frame && !(po->flags & OPF_EBP_S) |
1202 | && IS_START(popr->name, "ebp"))); |
1203 | } |
1204 | |
1205 | static void parse_stack_access(struct parsed_op *po, |
1206 | const char *name, char *ofs_reg, int *offset_out, |
1207 | int *stack_ra_out, const char **bp_arg_out) |
1208 | { |
1209 | const char *bp_arg = ""; |
1210 | const char *p = NULL; |
91977a1c |
1211 | struct parsed_equ *eq; |
4f12f671 |
1212 | char *endp = NULL; |
d4e3b5db |
1213 | int stack_ra = 0; |
39b168b8 |
1214 | int offset = 0; |
91977a1c |
1215 | |
1cd4a663 |
1216 | ofs_reg[0] = 0; |
a3684be1 |
1217 | |
1218 | if (IS_START(name, "ebp-") |
1219 | || (IS_START(name, "ebp+") && '0' <= name[4] && name[4] <= '9')) |
1220 | { |
1221 | p = name + 4; |
1222 | if (IS_START(p, "0x")) |
1223 | p += 2; |
1224 | offset = strtoul(p, &endp, 16); |
1225 | if (name[3] == '-') |
1226 | offset = -offset; |
1227 | if (*endp != 0) |
1228 | ferr(po, "ebp- parse of '%s' failed\n", name); |
1229 | } |
1230 | else { |
4f12f671 |
1231 | bp_arg = parse_stack_el(name, ofs_reg); |
1232 | snprintf(g_comment, sizeof(g_comment), "%s", bp_arg); |
1233 | eq = equ_find(po, bp_arg, &offset); |
1234 | if (eq == NULL) |
1235 | ferr(po, "detected but missing eq\n"); |
1236 | offset += eq->offset; |
1237 | } |
39b168b8 |
1238 | |
d4e3b5db |
1239 | if (!strncmp(name, "ebp", 3)) |
1240 | stack_ra = 4; |
1241 | |
94d447fb |
1242 | if (ofs_reg[0] == 0 && stack_ra <= offset && offset < stack_ra + 4) |
39b168b8 |
1243 | ferr(po, "reference to ra? %d %d\n", offset, stack_ra); |
d4e3b5db |
1244 | |
1cd4a663 |
1245 | *offset_out = offset; |
1246 | *stack_ra_out = stack_ra; |
1247 | if (bp_arg_out) |
1248 | *bp_arg_out = bp_arg; |
1249 | } |
1250 | |
1251 | static void stack_frame_access(struct parsed_op *po, |
1252 | struct parsed_opr *popr, char *buf, size_t buf_size, |
1253 | const char *name, const char *cast, int is_src, int is_lea) |
1254 | { |
1255 | enum opr_lenmod tmp_lmod = OPLM_UNSPEC; |
1256 | const char *prefix = ""; |
1257 | const char *bp_arg = NULL; |
1258 | char ofs_reg[16] = { 0, }; |
1259 | int i, arg_i, arg_s; |
1260 | int unaligned = 0; |
1261 | int stack_ra = 0; |
1262 | int offset = 0; |
1263 | int sf_ofs; |
1264 | int lim; |
1265 | |
1266 | if (po->flags & OPF_EBP_S) |
1267 | ferr(po, "stack_frame_access while ebp is scratch\n"); |
1268 | |
1269 | parse_stack_access(po, name, ofs_reg, &offset, &stack_ra, &bp_arg); |
1270 | |
4f12f671 |
1271 | if (offset > stack_ra) |
1272 | { |
39b168b8 |
1273 | arg_i = (offset - stack_ra - 4) / 4; |
bd96f656 |
1274 | if (arg_i < 0 || arg_i >= g_func_pp->argc_stack) |
4f12f671 |
1275 | { |
bd96f656 |
1276 | if (g_func_pp->is_vararg |
1277 | && arg_i == g_func_pp->argc_stack && is_lea) |
1278 | { |
4f12f671 |
1279 | // should be va_list |
1280 | if (cast[0] == 0) |
1281 | cast = "(u32)"; |
1282 | snprintf(buf, buf_size, "%sap", cast); |
1283 | return; |
1284 | } |
1285 | ferr(po, "offset %d (%s,%d) doesn't map to any arg\n", |
1286 | offset, bp_arg, arg_i); |
1287 | } |
4c45fa73 |
1288 | if (ofs_reg[0] != 0) |
7ba45c34 |
1289 | ferr(po, "offset reg on arg access?\n"); |
91977a1c |
1290 | |
bd96f656 |
1291 | for (i = arg_s = 0; i < g_func_pp->argc; i++) { |
1292 | if (g_func_pp->arg[i].reg != NULL) |
91977a1c |
1293 | continue; |
1294 | if (arg_s == arg_i) |
1295 | break; |
1296 | arg_s++; |
1297 | } |
bd96f656 |
1298 | if (i == g_func_pp->argc) |
91977a1c |
1299 | ferr(po, "arg %d not in prototype?\n", arg_i); |
850c9265 |
1300 | |
bd96f656 |
1301 | popr->is_ptr = g_func_pp->arg[i].type.is_ptr; |
de50b98b |
1302 | |
1303 | switch (popr->lmod) |
7ba45c34 |
1304 | { |
1305 | case OPLM_BYTE: |
4f12f671 |
1306 | if (is_lea) |
1307 | ferr(po, "lea/byte to arg?\n"); |
7ba45c34 |
1308 | if (is_src && (offset & 3) == 0) |
a2c1d768 |
1309 | snprintf(buf, buf_size, "%sa%d", |
1310 | simplify_cast(cast, "(u8)"), i + 1); |
7ba45c34 |
1311 | else |
a2c1d768 |
1312 | snprintf(buf, buf_size, "%sBYTE%d(a%d)", |
1313 | cast, offset & 3, i + 1); |
7ba45c34 |
1314 | break; |
1315 | |
1316 | case OPLM_WORD: |
4f12f671 |
1317 | if (is_lea) |
1318 | ferr(po, "lea/word to arg?\n"); |
a3684be1 |
1319 | if (offset & 1) { |
1320 | unaligned = 1; |
1321 | if (!is_src) { |
1322 | if (offset & 2) |
1323 | ferr(po, "problematic arg store\n"); |
a2c1d768 |
1324 | snprintf(buf, buf_size, "%s((char *)&a%d + 1)", |
1325 | simplify_cast(cast, "*(u16 *)"), i + 1); |
a3684be1 |
1326 | } |
1327 | else |
1328 | ferr(po, "unaligned arg word load\n"); |
1329 | } |
1330 | else if (is_src && (offset & 2) == 0) |
a2c1d768 |
1331 | snprintf(buf, buf_size, "%sa%d", |
1332 | simplify_cast(cast, "(u16)"), i + 1); |
7ba45c34 |
1333 | else |
a2c1d768 |
1334 | snprintf(buf, buf_size, "%s%sWORD(a%d)", |
1335 | cast, (offset & 2) ? "HI" : "LO", i + 1); |
7ba45c34 |
1336 | break; |
1337 | |
1338 | case OPLM_DWORD: |
3ebea2cf |
1339 | if (cast[0]) |
1340 | prefix = cast; |
1341 | else if (is_src) |
1342 | prefix = "(u32)"; |
a3684be1 |
1343 | |
de50b98b |
1344 | if (offset & 3) { |
a3684be1 |
1345 | unaligned = 1; |
2b43685d |
1346 | if (is_lea) |
1347 | snprintf(buf, buf_size, "(u32)&a%d + %d", |
1348 | i + 1, offset & 3); |
a3684be1 |
1349 | else if (!is_src) |
1350 | ferr(po, "unaligned arg store\n"); |
1351 | else { |
1352 | // mov edx, [ebp+arg_4+2]; movsx ecx, dx |
2b43685d |
1353 | snprintf(buf, buf_size, "%s(a%d >> %d)", |
1354 | prefix, i + 1, (offset & 3) * 8); |
a3684be1 |
1355 | } |
de50b98b |
1356 | } |
1357 | else { |
1358 | snprintf(buf, buf_size, "%s%sa%d", |
1359 | prefix, is_lea ? "&" : "", i + 1); |
1360 | } |
7ba45c34 |
1361 | break; |
1362 | |
1363 | default: |
de50b98b |
1364 | ferr(po, "bp_arg bad lmod: %d\n", popr->lmod); |
7ba45c34 |
1365 | } |
1366 | |
a3684be1 |
1367 | if (unaligned) |
1368 | snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg); |
1369 | |
7ba45c34 |
1370 | // common problem |
bd96f656 |
1371 | guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type); |
a2c1d768 |
1372 | if (tmp_lmod != OPLM_DWORD |
1373 | && (unaligned || (!is_src && tmp_lmod < popr->lmod))) |
1374 | { |
1375 | ferr(po, "bp_arg arg%d/w offset %d and type '%s' is too small\n", |
1376 | i + 1, offset, g_func_pp->arg[i].type.name); |
1377 | } |
46411e6c |
1378 | if (popr->is_ptr && popr->lmod != OPLM_DWORD) |
1379 | ferr(po, "bp_arg arg%d: non-dword ptr access\n", i + 1); |
91977a1c |
1380 | } |
4f12f671 |
1381 | else |
1382 | { |
1bafb621 |
1383 | if (g_stack_fsz == 0) |
1384 | ferr(po, "stack var access without stackframe\n"); |
a2c1d768 |
1385 | g_stack_frame_used = 1; |
850c9265 |
1386 | |
39b168b8 |
1387 | sf_ofs = g_stack_fsz + offset; |
de50b98b |
1388 | lim = (ofs_reg[0] != 0) ? -4 : 0; |
1389 | if (offset > 0 || sf_ofs < lim) |
39b168b8 |
1390 | ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz); |
850c9265 |
1391 | |
1392 | if (is_lea) |
33c35af6 |
1393 | prefix = "(u32)&"; |
3ebea2cf |
1394 | else |
1395 | prefix = cast; |
850c9265 |
1396 | |
de50b98b |
1397 | switch (popr->lmod) |
850c9265 |
1398 | { |
1399 | case OPLM_BYTE: |
7ba45c34 |
1400 | snprintf(buf, buf_size, "%ssf.b[%d%s%s]", |
1401 | prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg); |
850c9265 |
1402 | break; |
7ba45c34 |
1403 | |
850c9265 |
1404 | case OPLM_WORD: |
7ba45c34 |
1405 | if ((sf_ofs & 1) || ofs_reg[0] != 0) { |
1406 | // known unaligned or possibly unaligned |
1407 | strcat(g_comment, " unaligned"); |
1408 | if (prefix[0] == 0) |
1409 | prefix = "*(u16 *)&"; |
1410 | snprintf(buf, buf_size, "%ssf.b[%d%s%s]", |
1411 | prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg); |
1412 | break; |
1413 | } |
1414 | snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2); |
850c9265 |
1415 | break; |
7ba45c34 |
1416 | |
850c9265 |
1417 | case OPLM_DWORD: |
7ba45c34 |
1418 | if ((sf_ofs & 3) || ofs_reg[0] != 0) { |
1419 | // known unaligned or possibly unaligned |
1420 | strcat(g_comment, " unaligned"); |
1421 | if (prefix[0] == 0) |
1422 | prefix = "*(u32 *)&"; |
1423 | snprintf(buf, buf_size, "%ssf.b[%d%s%s]", |
1424 | prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg); |
1425 | break; |
1426 | } |
1427 | snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4); |
850c9265 |
1428 | break; |
7ba45c34 |
1429 | |
850c9265 |
1430 | default: |
de50b98b |
1431 | ferr(po, "bp_stack bad lmod: %d\n", popr->lmod); |
850c9265 |
1432 | } |
91977a1c |
1433 | } |
1434 | } |
c36e914d |
1435 | |
3ebea2cf |
1436 | static void check_label_read_ref(struct parsed_op *po, const char *name) |
1437 | { |
840257f6 |
1438 | const struct parsed_proto *pp; |
1439 | |
1440 | pp = proto_parse(g_fhdr, name); |
1441 | if (pp == NULL) |
1442 | ferr(po, "proto_parse failed for ref '%s'\n", name); |
1443 | |
1444 | // currently we can take __cdecl and __stdcall |
1445 | if (pp->is_func && pp->argc_reg != 0) |
1446 | ferr(po, "reg-arg func reference?\n"); |
3ebea2cf |
1447 | } |
1448 | |
91977a1c |
1449 | static char *out_src_opr(char *buf, size_t buf_size, |
3ebea2cf |
1450 | struct parsed_op *po, struct parsed_opr *popr, const char *cast, |
1451 | int is_lea) |
91977a1c |
1452 | { |
850c9265 |
1453 | char tmp1[256], tmp2[256]; |
1454 | char expr[256]; |
a2c1d768 |
1455 | char *p; |
850c9265 |
1456 | int ret; |
1457 | |
3ebea2cf |
1458 | if (cast == NULL) |
1459 | cast = ""; |
1460 | |
91977a1c |
1461 | switch (popr->type) { |
1462 | case OPT_REG: |
850c9265 |
1463 | if (is_lea) |
1464 | ferr(po, "lea from reg?\n"); |
1465 | |
91977a1c |
1466 | switch (popr->lmod) { |
1467 | case OPLM_DWORD: |
3ebea2cf |
1468 | snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr)); |
91977a1c |
1469 | break; |
850c9265 |
1470 | case OPLM_WORD: |
a2c1d768 |
1471 | snprintf(buf, buf_size, "%s%s", |
1472 | simplify_cast(cast, "(u16)"), opr_reg_p(po, popr)); |
850c9265 |
1473 | break; |
1474 | case OPLM_BYTE: |
5101a5f9 |
1475 | if (popr->name[1] == 'h') // XXX.. |
a2c1d768 |
1476 | snprintf(buf, buf_size, "%s(%s >> 8)", |
1477 | simplify_cast(cast, "(u8)"), opr_reg_p(po, popr)); |
5101a5f9 |
1478 | else |
a2c1d768 |
1479 | snprintf(buf, buf_size, "%s%s", |
1480 | simplify_cast(cast, "(u8)"), opr_reg_p(po, popr)); |
850c9265 |
1481 | break; |
91977a1c |
1482 | default: |
1483 | ferr(po, "invalid src lmod: %d\n", popr->lmod); |
1484 | } |
1485 | break; |
850c9265 |
1486 | |
91977a1c |
1487 | case OPT_REGMEM: |
1cd4a663 |
1488 | if (is_stack_access(po, popr)) { |
de50b98b |
1489 | stack_frame_access(po, popr, buf, buf_size, |
3ebea2cf |
1490 | popr->name, cast, 1, is_lea); |
91977a1c |
1491 | break; |
1492 | } |
850c9265 |
1493 | |
1494 | strcpy(expr, popr->name); |
1495 | if (strchr(expr, '[')) { |
1496 | // special case: '[' can only be left for label[reg] form |
1497 | ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2); |
1498 | if (ret != 2) |
1499 | ferr(po, "parse failure for '%s'\n", expr); |
a2c1d768 |
1500 | if (tmp1[0] == '(') { |
1501 | // (off_4FFF50+3)[eax] |
1502 | p = strchr(tmp1 + 1, ')'); |
1503 | if (p == NULL || p[1] != 0) |
1504 | ferr(po, "parse failure (2) for '%s'\n", expr); |
1505 | *p = 0; |
1506 | memmove(tmp1, tmp1 + 1, strlen(tmp1)); |
1507 | } |
33c35af6 |
1508 | snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2); |
850c9265 |
1509 | } |
1510 | |
1511 | // XXX: do we need more parsing? |
1512 | if (is_lea) { |
1513 | snprintf(buf, buf_size, "%s", expr); |
1514 | break; |
1515 | } |
1516 | |
a2c1d768 |
1517 | snprintf(buf, buf_size, "%s(%s)", |
1518 | simplify_cast(cast, lmod_cast_u_ptr(po, popr->lmod)), expr); |
91977a1c |
1519 | break; |
850c9265 |
1520 | |
91977a1c |
1521 | case OPT_LABEL: |
3ebea2cf |
1522 | check_label_read_ref(po, popr->name); |
1523 | if (cast[0] == 0 && popr->is_ptr) |
1524 | cast = "(u32)"; |
2b43685d |
1525 | |
850c9265 |
1526 | if (is_lea) |
1527 | snprintf(buf, buf_size, "(u32)&%s", popr->name); |
2b43685d |
1528 | else if (popr->size_lt) |
1529 | snprintf(buf, buf_size, "%s%s%s%s", cast, |
1530 | lmod_cast_u_ptr(po, popr->lmod), |
1531 | popr->is_array ? "" : "&", |
1532 | popr->name); |
850c9265 |
1533 | else |
3ebea2cf |
1534 | snprintf(buf, buf_size, "%s%s%s", cast, popr->name, |
7ba45c34 |
1535 | popr->is_array ? "[0]" : ""); |
850c9265 |
1536 | break; |
1537 | |
1538 | case OPT_OFFSET: |
3ebea2cf |
1539 | check_label_read_ref(po, popr->name); |
1540 | if (cast[0] == 0) |
1541 | cast = "(u32)"; |
850c9265 |
1542 | if (is_lea) |
1543 | ferr(po, "lea an offset?\n"); |
3ebea2cf |
1544 | snprintf(buf, buf_size, "%s&%s", cast, popr->name); |
91977a1c |
1545 | break; |
850c9265 |
1546 | |
91977a1c |
1547 | case OPT_CONST: |
850c9265 |
1548 | if (is_lea) |
1549 | ferr(po, "lea from const?\n"); |
1550 | |
a2c1d768 |
1551 | printf_number(tmp1, sizeof(tmp1), popr->val); |
1552 | snprintf(buf, buf_size, "%s%s", |
1553 | simplify_cast_num(cast, popr->val), tmp1); |
91977a1c |
1554 | break; |
850c9265 |
1555 | |
91977a1c |
1556 | default: |
1557 | ferr(po, "invalid src type: %d\n", popr->type); |
1558 | } |
1559 | |
1560 | return buf; |
1561 | } |
c36e914d |
1562 | |
de50b98b |
1563 | // note: may set is_ptr (we find that out late for ebp frame..) |
91977a1c |
1564 | static char *out_dst_opr(char *buf, size_t buf_size, |
1565 | struct parsed_op *po, struct parsed_opr *popr) |
1566 | { |
1567 | switch (popr->type) { |
1568 | case OPT_REG: |
1569 | switch (popr->lmod) { |
1570 | case OPLM_DWORD: |
1571 | snprintf(buf, buf_size, "%s", opr_reg_p(po, popr)); |
1572 | break; |
850c9265 |
1573 | case OPLM_WORD: |
1574 | // ugh.. |
1575 | snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr)); |
1576 | break; |
1577 | case OPLM_BYTE: |
1578 | // ugh.. |
5101a5f9 |
1579 | if (popr->name[1] == 'h') // XXX.. |
1580 | snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr)); |
1581 | else |
1582 | snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr)); |
850c9265 |
1583 | break; |
91977a1c |
1584 | default: |
1585 | ferr(po, "invalid dst lmod: %d\n", popr->lmod); |
1586 | } |
1587 | break; |
850c9265 |
1588 | |
1589 | case OPT_REGMEM: |
1cd4a663 |
1590 | if (is_stack_access(po, popr)) { |
de50b98b |
1591 | stack_frame_access(po, popr, buf, buf_size, |
3ebea2cf |
1592 | popr->name, "", 0, 0); |
850c9265 |
1593 | break; |
1594 | } |
1595 | |
3ebea2cf |
1596 | return out_src_opr(buf, buf_size, po, popr, NULL, 0); |
850c9265 |
1597 | |
bfa4a6ee |
1598 | case OPT_LABEL: |
2b43685d |
1599 | if (popr->size_mismatch) |
1600 | snprintf(buf, buf_size, "%s%s%s", |
1601 | lmod_cast_u_ptr(po, popr->lmod), |
1602 | popr->is_array ? "" : "&", popr->name); |
1603 | else |
1604 | snprintf(buf, buf_size, "%s%s", popr->name, |
1605 | popr->is_array ? "[0]" : ""); |
bfa4a6ee |
1606 | break; |
1607 | |
91977a1c |
1608 | default: |
1609 | ferr(po, "invalid dst type: %d\n", popr->type); |
1610 | } |
1611 | |
1612 | return buf; |
1613 | } |
c36e914d |
1614 | |
3ebea2cf |
1615 | static char *out_src_opr_u32(char *buf, size_t buf_size, |
1616 | struct parsed_op *po, struct parsed_opr *popr) |
1617 | { |
1618 | return out_src_opr(buf, buf_size, po, popr, NULL, 0); |
1619 | } |
1620 | |
69a3cdfc |
1621 | static enum parsed_flag_op split_cond(struct parsed_op *po, |
940e8e66 |
1622 | enum op_op op, int *is_inv) |
91977a1c |
1623 | { |
940e8e66 |
1624 | *is_inv = 0; |
91977a1c |
1625 | |
69a3cdfc |
1626 | switch (op) { |
1627 | case OP_JO: |
1628 | return PFO_O; |
1629 | case OP_JC: |
1630 | return PFO_C; |
1631 | case OP_JZ: |
1632 | return PFO_Z; |
1633 | case OP_JBE: |
1634 | return PFO_BE; |
1635 | case OP_JS: |
1636 | return PFO_S; |
1637 | case OP_JP: |
1638 | return PFO_P; |
1639 | case OP_JL: |
1640 | return PFO_L; |
1641 | case OP_JLE: |
1642 | return PFO_LE; |
1643 | |
91977a1c |
1644 | case OP_JNO: |
940e8e66 |
1645 | *is_inv = 1; |
69a3cdfc |
1646 | return PFO_O; |
91977a1c |
1647 | case OP_JNC: |
940e8e66 |
1648 | *is_inv = 1; |
69a3cdfc |
1649 | return PFO_C; |
91977a1c |
1650 | case OP_JNZ: |
940e8e66 |
1651 | *is_inv = 1; |
69a3cdfc |
1652 | return PFO_Z; |
1653 | case OP_JA: |
940e8e66 |
1654 | *is_inv = 1; |
69a3cdfc |
1655 | return PFO_BE; |
91977a1c |
1656 | case OP_JNS: |
940e8e66 |
1657 | *is_inv = 1; |
69a3cdfc |
1658 | return PFO_S; |
91977a1c |
1659 | case OP_JNP: |
940e8e66 |
1660 | *is_inv = 1; |
69a3cdfc |
1661 | return PFO_P; |
850c9265 |
1662 | case OP_JGE: |
940e8e66 |
1663 | *is_inv = 1; |
69a3cdfc |
1664 | return PFO_L; |
91977a1c |
1665 | case OP_JG: |
940e8e66 |
1666 | *is_inv = 1; |
69a3cdfc |
1667 | return PFO_LE; |
1668 | |
1669 | case OP_ADC: |
1670 | case OP_SBB: |
1671 | return PFO_C; |
1672 | |
91977a1c |
1673 | default: |
69a3cdfc |
1674 | ferr(po, "split_cond: bad op %d\n", op); |
1675 | return -1; |
91977a1c |
1676 | } |
1677 | } |
c36e914d |
1678 | |
91977a1c |
1679 | static void out_test_for_cc(char *buf, size_t buf_size, |
940e8e66 |
1680 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv, |
69a3cdfc |
1681 | enum opr_lenmod lmod, const char *expr) |
91977a1c |
1682 | { |
69a3cdfc |
1683 | const char *cast, *scast; |
91977a1c |
1684 | |
69a3cdfc |
1685 | cast = lmod_cast_u(po, lmod); |
1686 | scast = lmod_cast_s(po, lmod); |
1687 | |
1688 | switch (pfo) { |
1689 | case PFO_Z: |
87bf6cec |
1690 | case PFO_BE: // CF=1||ZF=1; CF=0 |
850c9265 |
1691 | snprintf(buf, buf_size, "(%s%s %s 0)", |
940e8e66 |
1692 | cast, expr, is_inv ? "!=" : "=="); |
91977a1c |
1693 | break; |
850c9265 |
1694 | |
5101a5f9 |
1695 | case PFO_S: |
1696 | case PFO_L: // SF!=OF; OF=0 |
1697 | snprintf(buf, buf_size, "(%s%s %s 0)", |
940e8e66 |
1698 | scast, expr, is_inv ? ">=" : "<"); |
5101a5f9 |
1699 | break; |
1700 | |
87bf6cec |
1701 | case PFO_LE: // ZF=1||SF!=OF; OF=0 |
69a3cdfc |
1702 | snprintf(buf, buf_size, "(%s%s %s 0)", |
940e8e66 |
1703 | scast, expr, is_inv ? ">" : "<="); |
850c9265 |
1704 | break; |
1705 | |
91977a1c |
1706 | default: |
69a3cdfc |
1707 | ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo); |
91977a1c |
1708 | } |
1709 | } |
c36e914d |
1710 | |
850c9265 |
1711 | static void out_cmp_for_cc(char *buf, size_t buf_size, |
a2c1d768 |
1712 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv) |
850c9265 |
1713 | { |
a2c1d768 |
1714 | const char *cast, *scast, *cast_use; |
1715 | char buf1[256], buf2[256]; |
1716 | enum opr_lenmod lmod; |
1717 | |
1718 | if (po->operand[0].lmod != po->operand[1].lmod) |
1719 | ferr(po, "%s: lmod mismatch: %d %d\n", __func__, |
1720 | po->operand[0].lmod, po->operand[1].lmod); |
1721 | lmod = po->operand[0].lmod; |
850c9265 |
1722 | |
69a3cdfc |
1723 | cast = lmod_cast_u(po, lmod); |
1724 | scast = lmod_cast_s(po, lmod); |
850c9265 |
1725 | |
a2c1d768 |
1726 | switch (pfo) { |
1727 | case PFO_C: |
1728 | case PFO_Z: |
1729 | case PFO_BE: // !a |
1730 | cast_use = cast; |
1731 | break; |
1732 | |
1733 | case PFO_S: |
1734 | case PFO_L: // !ge |
1735 | case PFO_LE: |
1736 | cast_use = scast; |
1737 | break; |
1738 | |
1739 | default: |
1740 | ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo); |
1741 | } |
1742 | |
1743 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], cast_use, 0); |
1744 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], cast_use, 0); |
1745 | |
69a3cdfc |
1746 | switch (pfo) { |
5101a5f9 |
1747 | case PFO_C: |
1748 | // note: must be unsigned compare |
a2c1d768 |
1749 | snprintf(buf, buf_size, "(%s %s %s)", |
1750 | buf1, is_inv ? ">=" : "<", buf2); |
5101a5f9 |
1751 | break; |
1752 | |
69a3cdfc |
1753 | case PFO_Z: |
a2c1d768 |
1754 | snprintf(buf, buf_size, "(%s %s %s)", |
1755 | buf1, is_inv ? "!=" : "==", buf2); |
850c9265 |
1756 | break; |
1757 | |
5101a5f9 |
1758 | case PFO_BE: // !a |
850c9265 |
1759 | // note: must be unsigned compare |
a2c1d768 |
1760 | snprintf(buf, buf_size, "(%s %s %s)", |
1761 | buf1, is_inv ? ">" : "<=", buf2); |
1762 | |
1763 | // annoying case |
1764 | if (is_inv && lmod == OPLM_BYTE |
1765 | && po->operand[1].type == OPT_CONST |
1766 | && po->operand[1].val == 0xff) |
1767 | { |
1768 | snprintf(g_comment, sizeof(g_comment), "if %s", buf); |
1769 | snprintf(buf, buf_size, "(0)"); |
1770 | } |
5101a5f9 |
1771 | break; |
1772 | |
1773 | // note: must be signed compare |
1774 | case PFO_S: |
1775 | snprintf(buf, buf_size, "(%s(%s - %s) %s 0)", |
a2c1d768 |
1776 | scast, buf1, buf2, is_inv ? ">=" : "<"); |
850c9265 |
1777 | break; |
1778 | |
5101a5f9 |
1779 | case PFO_L: // !ge |
a2c1d768 |
1780 | snprintf(buf, buf_size, "(%s %s %s)", |
1781 | buf1, is_inv ? ">=" : "<", buf2); |
850c9265 |
1782 | break; |
1783 | |
5101a5f9 |
1784 | case PFO_LE: |
a2c1d768 |
1785 | snprintf(buf, buf_size, "(%s %s %s)", |
1786 | buf1, is_inv ? ">" : "<=", buf2); |
5101a5f9 |
1787 | break; |
1788 | |
850c9265 |
1789 | default: |
a2c1d768 |
1790 | break; |
850c9265 |
1791 | } |
1792 | } |
1793 | |
69a3cdfc |
1794 | static void out_cmp_test(char *buf, size_t buf_size, |
940e8e66 |
1795 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv) |
69a3cdfc |
1796 | { |
1797 | char buf1[256], buf2[256], buf3[256]; |
1798 | |
1799 | if (po->op == OP_TEST) { |
1800 | if (IS(opr_name(po, 0), opr_name(po, 1))) { |
3ebea2cf |
1801 | out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]); |
69a3cdfc |
1802 | } |
1803 | else { |
3ebea2cf |
1804 | out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]); |
1805 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]); |
69a3cdfc |
1806 | snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2); |
1807 | } |
940e8e66 |
1808 | out_test_for_cc(buf, buf_size, po, pfo, is_inv, |
69a3cdfc |
1809 | po->operand[0].lmod, buf3); |
1810 | } |
1811 | else if (po->op == OP_CMP) { |
a2c1d768 |
1812 | out_cmp_for_cc(buf, buf_size, po, pfo, is_inv); |
69a3cdfc |
1813 | } |
1814 | else |
1815 | ferr(po, "%s: unhandled op: %d\n", __func__, po->op); |
1816 | } |
1817 | |
850c9265 |
1818 | static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1, |
91977a1c |
1819 | struct parsed_opr *popr2) |
1820 | { |
1821 | if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC) |
1822 | ferr(po, "missing lmod for both operands\n"); |
1823 | |
1824 | if (popr1->lmod == OPLM_UNSPEC) |
1825 | popr1->lmod = popr2->lmod; |
1826 | else if (popr2->lmod == OPLM_UNSPEC) |
1827 | popr2->lmod = popr1->lmod; |
a3684be1 |
1828 | else if (popr1->lmod != popr2->lmod) { |
1829 | if (popr1->type_from_var) { |
1830 | popr1->size_mismatch = 1; |
1831 | if (popr1->lmod < popr2->lmod) |
1832 | popr1->size_lt = 1; |
1833 | popr1->lmod = popr2->lmod; |
1834 | } |
1835 | else if (popr2->type_from_var) { |
1836 | popr2->size_mismatch = 1; |
1837 | if (popr2->lmod < popr1->lmod) |
1838 | popr2->size_lt = 1; |
1839 | popr2->lmod = popr1->lmod; |
1840 | } |
1841 | else |
1842 | ferr(po, "conflicting lmods: %d vs %d\n", |
1843 | popr1->lmod, popr2->lmod); |
1844 | } |
91977a1c |
1845 | } |
c36e914d |
1846 | |
850c9265 |
1847 | static const char *op_to_c(struct parsed_op *po) |
1848 | { |
1849 | switch (po->op) |
1850 | { |
1851 | case OP_ADD: |
5101a5f9 |
1852 | case OP_ADC: |
850c9265 |
1853 | return "+"; |
1854 | case OP_SUB: |
5101a5f9 |
1855 | case OP_SBB: |
850c9265 |
1856 | return "-"; |
1857 | case OP_AND: |
1858 | return "&"; |
1859 | case OP_OR: |
1860 | return "|"; |
1861 | case OP_XOR: |
1862 | return "^"; |
1863 | case OP_SHL: |
1864 | return "<<"; |
1865 | case OP_SHR: |
1866 | return ">>"; |
1867 | case OP_MUL: |
1868 | case OP_IMUL: |
1869 | return "*"; |
1870 | default: |
1871 | ferr(po, "op_to_c was supplied with %d\n", po->op); |
1872 | } |
1873 | } |
1874 | |
7ae48d73 |
1875 | static void op_set_clear_flag(struct parsed_op *po, |
1876 | enum op_flags flag_set, enum op_flags flag_clear) |
d4e3b5db |
1877 | { |
7ae48d73 |
1878 | po->flags |= flag_set; |
1879 | po->flags &= ~flag_clear; |
d4e3b5db |
1880 | } |
1881 | |
de50b98b |
1882 | // last op in stream - unconditional branch or ret |
1883 | #define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \ |
1884 | || (ops[_i].flags & (OPF_JMP|OPF_CC)) == OPF_JMP) |
1885 | |
87bf6cec |
1886 | static int scan_for_pop(int i, int opcnt, const char *reg, |
d4e3b5db |
1887 | int magic, int depth, int *maxdepth, int do_flags) |
850c9265 |
1888 | { |
87bf6cec |
1889 | struct parsed_op *po; |
1890 | int ret = 0; |
4c45fa73 |
1891 | int j; |
87bf6cec |
1892 | |
850c9265 |
1893 | for (; i < opcnt; i++) { |
87bf6cec |
1894 | po = &ops[i]; |
1895 | if (po->cc_scratch == magic) |
1896 | break; // already checked |
1897 | po->cc_scratch = magic; |
1898 | |
1899 | if (po->flags & OPF_TAIL) |
1900 | return -1; // deadend |
1901 | |
1bafb621 |
1902 | if ((po->flags & OPF_RMD) |
de50b98b |
1903 | || (po->op == OP_PUSH && po->argnum != 0)) // arg push |
850c9265 |
1904 | continue; |
1905 | |
87bf6cec |
1906 | if ((po->flags & OPF_JMP) && po->op != OP_CALL) { |
4c45fa73 |
1907 | if (po->btj != NULL) { |
1908 | // jumptable |
1909 | for (j = 0; j < po->btj->count - 1; j++) { |
1910 | ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic, |
1911 | depth, maxdepth, do_flags); |
1912 | if (ret < 0) |
1913 | return ret; // dead end |
1914 | } |
1915 | // follow last jumptable entry |
1916 | i = po->btj->d[j].bt_i - 1; |
1917 | continue; |
1918 | } |
1919 | |
87bf6cec |
1920 | if (po->bt_i < 0) { |
1921 | ferr(po, "dead branch\n"); |
1922 | return -1; |
1923 | } |
850c9265 |
1924 | |
87bf6cec |
1925 | if (po->flags & OPF_CC) { |
d4e3b5db |
1926 | ret |= scan_for_pop(po->bt_i, opcnt, reg, magic, |
1927 | depth, maxdepth, do_flags); |
87bf6cec |
1928 | if (ret < 0) |
1929 | return ret; // dead end |
1930 | } |
1931 | else { |
1932 | i = po->bt_i - 1; |
1933 | } |
1934 | continue; |
1935 | } |
1936 | |
d4e3b5db |
1937 | if ((po->op == OP_POP || po->op == OP_PUSH) |
1938 | && po->operand[0].type == OPT_REG |
87bf6cec |
1939 | && IS(po->operand[0].name, reg)) |
1940 | { |
d4e3b5db |
1941 | if (po->op == OP_PUSH) { |
1942 | depth++; |
1943 | if (depth > *maxdepth) |
1944 | *maxdepth = depth; |
1945 | if (do_flags) |
7ae48d73 |
1946 | op_set_clear_flag(po, OPF_RSAVE, OPF_RMD); |
d4e3b5db |
1947 | } |
1948 | else if (depth == 0) { |
1949 | if (do_flags) |
7ae48d73 |
1950 | op_set_clear_flag(po, OPF_RMD, OPF_RSAVE); |
d4e3b5db |
1951 | return 1; |
1952 | } |
1953 | else { |
1954 | depth--; |
1955 | if (depth < 0) // should not happen |
1956 | ferr(po, "fail with depth\n"); |
1957 | if (do_flags) |
7ae48d73 |
1958 | op_set_clear_flag(po, OPF_RSAVE, OPF_RMD); |
d4e3b5db |
1959 | } |
87bf6cec |
1960 | } |
850c9265 |
1961 | } |
1962 | |
87bf6cec |
1963 | return ret; |
850c9265 |
1964 | } |
1965 | |
69a3cdfc |
1966 | // scan for pop starting from 'ret' op (all paths) |
d4e3b5db |
1967 | static int scan_for_pop_ret(int i, int opcnt, const char *reg, |
1968 | int flag_set) |
850c9265 |
1969 | { |
1970 | int found = 0; |
1971 | int j; |
1972 | |
1973 | for (; i < opcnt; i++) { |
87bf6cec |
1974 | if (!(ops[i].flags & OPF_TAIL)) |
850c9265 |
1975 | continue; |
1976 | |
1977 | for (j = i - 1; j >= 0; j--) { |
69a3cdfc |
1978 | if (ops[j].flags & OPF_RMD) |
1979 | continue; |
1980 | if (ops[j].flags & OPF_JMP) |
850c9265 |
1981 | return -1; |
1982 | |
1983 | if (ops[j].op == OP_POP && ops[j].operand[0].type == OPT_REG |
1984 | && IS(ops[j].operand[0].name, reg)) |
1985 | { |
1986 | found = 1; |
d4e3b5db |
1987 | ops[j].flags |= flag_set; |
850c9265 |
1988 | break; |
1989 | } |
1990 | |
1991 | if (g_labels[j][0] != 0) |
1992 | return -1; |
1993 | } |
1994 | } |
1995 | |
1996 | return found ? 0 : -1; |
1997 | } |
1998 | |
1cd4a663 |
1999 | // is operand 'opr' modified by parsed_op 'po'? |
5101a5f9 |
2000 | static int is_opr_modified(const struct parsed_opr *opr, |
69a3cdfc |
2001 | const struct parsed_op *po) |
2002 | { |
2003 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA)) |
2004 | return 0; |
2005 | |
2006 | if (opr->type == OPT_REG && po->operand[0].type == OPT_REG) { |
2007 | if (po->regmask_dst & (1 << opr->reg)) |
2008 | return 1; |
2009 | else |
2010 | return 0; |
2011 | } |
2012 | |
2013 | return IS(po->operand[0].name, opr->name); |
2014 | } |
2015 | |
5101a5f9 |
2016 | // is any operand of parsed_op 'po_test' modified by parsed_op 'po'? |
2017 | static int is_any_opr_modified(const struct parsed_op *po_test, |
2018 | const struct parsed_op *po) |
2019 | { |
2020 | int i; |
2021 | |
2022 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA)) |
2023 | return 0; |
2024 | |
de50b98b |
2025 | if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST) |
2026 | return 0; |
2027 | |
2028 | if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst) |
2029 | return 1; |
2030 | |
2031 | // in reality, it can wreck any register, but in decompiled C |
2b43685d |
2032 | // version it can only overwrite eax or edx:eax |
de50b98b |
2033 | if (po->op == OP_CALL |
2b43685d |
2034 | && ((po_test->regmask_src | po_test->regmask_dst) |
2035 | & ((1 << xAX)|(1 << xDX)))) |
5101a5f9 |
2036 | return 1; |
2037 | |
2038 | for (i = 0; i < po_test->operand_cnt; i++) |
2039 | if (IS(po_test->operand[i].name, po->operand[0].name)) |
2040 | return 1; |
2041 | |
2042 | return 0; |
2043 | } |
2044 | |
940e8e66 |
2045 | // scan for any po_test operand modification in range given |
5101a5f9 |
2046 | static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt) |
69a3cdfc |
2047 | { |
2b43685d |
2048 | if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST) |
2049 | return -1; |
2050 | |
69a3cdfc |
2051 | for (; i < opcnt; i++) { |
5101a5f9 |
2052 | if (is_any_opr_modified(po_test, &ops[i])) |
69a3cdfc |
2053 | return i; |
2054 | } |
2055 | |
2056 | return -1; |
2057 | } |
2058 | |
940e8e66 |
2059 | // scan for po_test operand[0] modification in range given |
2060 | static int scan_for_mod_opr0(struct parsed_op *po_test, |
2061 | int i, int opcnt) |
2062 | { |
2063 | for (; i < opcnt; i++) { |
2064 | if (is_opr_modified(&po_test->operand[0], &ops[i])) |
2065 | return i; |
2066 | } |
2067 | |
2068 | return -1; |
2069 | } |
2070 | |
04f8a628 |
2071 | static int scan_for_flag_set(int i, int magic, int *branched, |
2072 | int *setters, int *setter_cnt) |
69a3cdfc |
2073 | { |
04f8a628 |
2074 | struct label_ref *lr; |
2b43685d |
2075 | int ret; |
de50b98b |
2076 | |
2077 | while (i >= 0) { |
04f8a628 |
2078 | if (ops[i].cc_scratch == magic) { |
2079 | ferr(&ops[i], "%s looped\n", __func__); |
2080 | return -1; |
2081 | } |
2082 | ops[i].cc_scratch = magic; |
2083 | |
de50b98b |
2084 | if (g_labels[i][0] != 0) { |
2b43685d |
2085 | *branched = 1; |
04f8a628 |
2086 | |
2087 | lr = &g_label_refs[i]; |
2088 | for (; lr->next; lr = lr->next) { |
2089 | ret = scan_for_flag_set(lr->i, magic, |
2090 | branched, setters, setter_cnt); |
2091 | if (ret < 0) |
2092 | return ret; |
2093 | } |
2094 | |
de50b98b |
2095 | if (i > 0 && LAST_OP(i - 1)) { |
94d447fb |
2096 | i = lr->i; |
de50b98b |
2097 | continue; |
2098 | } |
04f8a628 |
2099 | ret = scan_for_flag_set(lr->i, magic, |
2100 | branched, setters, setter_cnt); |
2b43685d |
2101 | if (ret < 0) |
2102 | return ret; |
de50b98b |
2103 | } |
2104 | i--; |
2105 | |
2b43685d |
2106 | if (ops[i].flags & OPF_FLAGS) { |
2107 | setters[*setter_cnt] = i; |
2108 | (*setter_cnt)++; |
2109 | return 0; |
2110 | } |
69a3cdfc |
2111 | |
2112 | if ((ops[i].flags & OPF_JMP) && !(ops[i].flags & OPF_CC)) |
2113 | return -1; |
69a3cdfc |
2114 | } |
2115 | |
2116 | return -1; |
2117 | } |
2118 | |
5101a5f9 |
2119 | // scan back for cdq, if anything modifies edx, fail |
2120 | static int scan_for_cdq_edx(int i) |
2121 | { |
cdfaeed7 |
2122 | while (i >= 0) { |
2123 | if (g_labels[i][0] != 0) { |
2124 | if (g_label_refs[i].next != NULL) |
2125 | return -1; |
2126 | if (i > 0 && LAST_OP(i - 1)) { |
2127 | i = g_label_refs[i].i; |
2128 | continue; |
2129 | } |
2130 | return -1; |
2131 | } |
2132 | i--; |
2133 | |
5101a5f9 |
2134 | if (ops[i].op == OP_CDQ) |
2135 | return i; |
2136 | |
2137 | if (ops[i].regmask_dst & (1 << xDX)) |
2138 | return -1; |
2139 | if (g_labels[i][0] != 0) |
2140 | return -1; |
2141 | } |
2142 | |
2143 | return -1; |
2144 | } |
2145 | |
64c59faf |
2146 | static int scan_for_reg_clear(int i, int reg) |
2147 | { |
cdfaeed7 |
2148 | while (i >= 0) { |
2149 | if (g_labels[i][0] != 0) { |
2150 | if (g_label_refs[i].next != NULL) |
2151 | return -1; |
2152 | if (i > 0 && LAST_OP(i - 1)) { |
2153 | i = g_label_refs[i].i; |
2154 | continue; |
2155 | } |
2156 | return -1; |
2157 | } |
2158 | i--; |
2159 | |
64c59faf |
2160 | if (ops[i].op == OP_XOR |
2161 | && ops[i].operand[0].lmod == OPLM_DWORD |
2162 | && ops[i].operand[0].reg == ops[i].operand[1].reg |
2163 | && ops[i].operand[0].reg == reg) |
2164 | return i; |
2165 | |
2166 | if (ops[i].regmask_dst & (1 << reg)) |
2167 | return -1; |
2168 | if (g_labels[i][0] != 0) |
2169 | return -1; |
2170 | } |
2171 | |
2172 | return -1; |
2173 | } |
2174 | |
1bafb621 |
2175 | // scan for positive, constant esp adjust |
2176 | static int scan_for_esp_adjust(int i, int opcnt, int *adj) |
2177 | { |
7ba45c34 |
2178 | struct parsed_op *po; |
46411e6c |
2179 | int first_pop = -1; |
7ba45c34 |
2180 | *adj = 0; |
2181 | |
1bafb621 |
2182 | for (; i < opcnt; i++) { |
7ba45c34 |
2183 | po = &ops[i]; |
2184 | |
a2c1d768 |
2185 | if (g_labels[i][0] != 0) |
2186 | break; |
2187 | |
7ba45c34 |
2188 | if (po->op == OP_ADD && po->operand[0].reg == xSP) { |
2189 | if (po->operand[1].type != OPT_CONST) |
1bafb621 |
2190 | ferr(&ops[i], "non-const esp adjust?\n"); |
7ba45c34 |
2191 | *adj += po->operand[1].val; |
1bafb621 |
2192 | if (*adj & 3) |
2193 | ferr(&ops[i], "unaligned esp adjust: %x\n", *adj); |
2194 | return i; |
2195 | } |
46411e6c |
2196 | else if (po->op == OP_PUSH) { |
2197 | if (first_pop == -1) |
2198 | first_pop = -2; // none |
7ba45c34 |
2199 | *adj -= lmod_bytes(po, po->operand[0].lmod); |
46411e6c |
2200 | } |
2201 | else if (po->op == OP_POP) { |
2202 | if (first_pop == -1) |
2203 | first_pop = i; |
7ba45c34 |
2204 | *adj += lmod_bytes(po, po->operand[0].lmod); |
46411e6c |
2205 | } |
e56ab892 |
2206 | else if (po->flags & (OPF_JMP|OPF_TAIL)) { |
2207 | if (po->op != OP_CALL) |
a2c1d768 |
2208 | break; |
e56ab892 |
2209 | if (po->operand[0].type != OPT_LABEL) |
a2c1d768 |
2210 | break; |
e56ab892 |
2211 | // TODO: should only allow combining __cdecl calls.. |
2212 | } |
a2c1d768 |
2213 | } |
7ba45c34 |
2214 | |
46411e6c |
2215 | if (*adj == 4 && first_pop >= 0 && ops[first_pop].op == OP_POP |
2216 | && ops[first_pop].operand[0].type == OPT_REG |
2217 | && ops[first_pop].operand[0].reg == xCX) |
a2c1d768 |
2218 | { |
2219 | // probably 'pop ecx' was used.. |
46411e6c |
2220 | return first_pop; |
1bafb621 |
2221 | } |
2222 | |
2223 | return -1; |
2224 | } |
2225 | |
a3684be1 |
2226 | static void scan_fwd_set_flags(int i, int opcnt, int magic, int flags) |
2227 | { |
2228 | struct parsed_op *po; |
2229 | int j; |
2230 | |
2231 | if (i < 0) |
2232 | ferr(ops, "%s: followed bad branch?\n", __func__); |
2233 | |
2234 | for (; i < opcnt; i++) { |
2235 | po = &ops[i]; |
2236 | if (po->cc_scratch == magic) |
2237 | return; |
2238 | po->cc_scratch = magic; |
2239 | po->flags |= flags; |
2240 | |
2241 | if ((po->flags & OPF_JMP) && po->op != OP_CALL) { |
2242 | if (po->btj != NULL) { |
2243 | // jumptable |
2244 | for (j = 0; j < po->btj->count; j++) |
2245 | scan_fwd_set_flags(po->btj->d[j].bt_i, opcnt, magic, flags); |
2246 | return; |
2247 | } |
2248 | |
2249 | scan_fwd_set_flags(po->bt_i, opcnt, magic, flags); |
2250 | if (!(po->flags & OPF_CC)) |
2251 | return; |
2252 | } |
2253 | if (po->flags & OPF_TAIL) |
2254 | return; |
2255 | } |
2256 | } |
2257 | |
1cd4a663 |
2258 | static const struct parsed_proto *try_recover_pp( |
2259 | struct parsed_op *po, const struct parsed_opr *opr) |
2260 | { |
2261 | const struct parsed_proto *pp = NULL; |
2262 | |
2263 | // maybe an arg of g_func? |
2264 | if (opr->type == OPT_REGMEM && is_stack_access(po, opr)) |
2265 | { |
2266 | char ofs_reg[16] = { 0, }; |
2267 | int arg, arg_s, arg_i; |
2268 | int stack_ra = 0; |
2269 | int offset = 0; |
2270 | |
2271 | parse_stack_access(po, opr->name, ofs_reg, |
2272 | &offset, &stack_ra, NULL); |
2273 | if (ofs_reg[0] != 0) |
2274 | ferr(po, "offset reg on arg access?\n"); |
2275 | if (offset <= stack_ra) |
2276 | ferr(po, "stack var call unhandled yet\n"); |
2277 | |
2278 | arg_i = (offset - stack_ra - 4) / 4; |
2279 | for (arg = arg_s = 0; arg < g_func_pp->argc; arg++) { |
2280 | if (g_func_pp->arg[arg].reg != NULL) |
2281 | continue; |
2282 | if (arg_s == arg_i) |
2283 | break; |
2284 | arg_s++; |
2285 | } |
2286 | if (arg == g_func_pp->argc) |
2287 | ferr(po, "stack arg %d not in prototype?\n", arg_i); |
2288 | |
2289 | pp = g_func_pp->arg[arg].fptr; |
2290 | if (pp == NULL) |
46411e6c |
2291 | ferr(po, "icall sa: arg%d is not a fptr?\n", arg + 1); |
1cd4a663 |
2292 | if (pp->argc_reg != 0) |
46411e6c |
2293 | ferr(po, "icall sa: reg arg in arg-call unhandled yet\n"); |
1cd4a663 |
2294 | } |
2295 | else if (opr->type == OPT_OFFSET || opr->type == OPT_LABEL) { |
2296 | pp = proto_parse(g_fhdr, opr->name); |
2297 | if (pp == NULL) |
2298 | ferr(po, "proto_parse failed for icall from '%s'\n", opr->name); |
2299 | if (pp->argc_reg != 0) |
2300 | ferr(po, "arg-call unhandled yet for '%s'\n", opr->name); |
2301 | } |
2302 | |
2303 | return pp; |
2304 | } |
2305 | |
2306 | static void scan_for_call_type(int i, struct parsed_opr *opr, |
2307 | int magic, const struct parsed_proto **pp_found) |
2308 | { |
2309 | const struct parsed_proto *pp = NULL; |
2310 | struct parsed_op *po; |
2311 | struct label_ref *lr; |
2312 | |
2313 | while (i >= 0) { |
2314 | if (ops[i].cc_scratch == magic) |
2315 | return; |
2316 | ops[i].cc_scratch = magic; |
2317 | |
2318 | if (g_labels[i][0] != 0) { |
2319 | lr = &g_label_refs[i]; |
2320 | for (; lr != NULL; lr = lr->next) |
2321 | scan_for_call_type(lr->i, opr, magic, pp_found); |
46411e6c |
2322 | if (i > 0 && LAST_OP(i - 1)) |
2323 | return; |
1cd4a663 |
2324 | } |
2325 | i--; |
2326 | |
2327 | if (!(ops[i].flags & OPF_DATA)) |
2328 | continue; |
2329 | if (!is_opr_modified(opr, &ops[i])) |
2330 | continue; |
2331 | if (ops[i].op != OP_MOV && ops[i].op != OP_LEA) { |
2332 | // most probably trashed by some processing |
2333 | *pp_found = NULL; |
2334 | return; |
2335 | } |
2336 | |
2337 | opr = &ops[i].operand[1]; |
2338 | if (opr->type != OPT_REG) |
2339 | break; |
2340 | } |
2341 | |
2342 | po = (i >= 0) ? &ops[i] : ops; |
2343 | |
2344 | if (i < 0) { |
2345 | // reached the top - can only be an arg-reg |
2346 | if (opr->type != OPT_REG) |
2347 | return; |
2348 | |
2349 | for (i = 0; i < g_func_pp->argc; i++) { |
2350 | if (g_func_pp->arg[i].reg == NULL) |
2351 | continue; |
2352 | if (IS(opr->name, g_func_pp->arg[i].reg)) |
2353 | break; |
2354 | } |
2355 | if (i == g_func_pp->argc) |
2356 | return; |
2357 | pp = g_func_pp->arg[i].fptr; |
2358 | if (pp == NULL) |
46411e6c |
2359 | ferr(po, "icall: arg%d (%s) is not a fptr?\n", |
2360 | i + 1, g_func_pp->arg[i].reg); |
1cd4a663 |
2361 | if (pp->argc_reg != 0) |
2362 | ferr(po, "icall: reg arg in arg-call unhandled yet\n"); |
2363 | } |
2364 | else |
2365 | pp = try_recover_pp(po, opr); |
2366 | |
2367 | if (*pp_found != NULL && pp != NULL) { |
2368 | if (!IS((*pp_found)->ret_type.name, pp->ret_type.name) |
2369 | || (*pp_found)->is_stdcall != pp->is_stdcall |
2370 | || (*pp_found)->argc != pp->argc |
2371 | || (*pp_found)->argc_reg != pp->argc_reg |
2372 | || (*pp_found)->argc_stack != pp->argc_stack) |
2373 | { |
2374 | ferr(po, "icall: parsed_proto mismatch\n"); |
2375 | } |
2376 | } |
2377 | if (pp != NULL) |
2378 | *pp_found = pp; |
2379 | } |
2380 | |
2381 | static const struct parsed_proto *resolve_call(int i, int opcnt) |
2382 | { |
2383 | const struct parsed_proto *pp = NULL; |
2384 | |
2385 | switch (ops[i].operand[0].type) { |
2386 | case OPT_REGMEM: |
2387 | case OPT_LABEL: |
2388 | case OPT_OFFSET: |
2389 | pp = try_recover_pp(&ops[i], &ops[i].operand[0]); |
2390 | break; |
2391 | default: |
2392 | scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp); |
2393 | break; |
2394 | } |
2395 | |
2396 | return pp; |
2397 | } |
2398 | |
e56ab892 |
2399 | static int collect_call_args(struct parsed_op *po, int i, |
2400 | struct parsed_proto *pp, int *save_arg_vars, int arg, |
a3684be1 |
2401 | int magic, int need_op_saving, int may_reuse) |
e56ab892 |
2402 | { |
2403 | struct parsed_proto *pp_tmp; |
2404 | struct label_ref *lr; |
2b43685d |
2405 | int need_to_save_current; |
e56ab892 |
2406 | int ret = 0; |
2407 | int j; |
2408 | |
a3684be1 |
2409 | if (i < 0) { |
a2c1d768 |
2410 | ferr(po, "dead label encountered\n"); |
a3684be1 |
2411 | return -1; |
2412 | } |
e56ab892 |
2413 | |
de50b98b |
2414 | for (; arg < pp->argc; arg++) |
e56ab892 |
2415 | if (pp->arg[arg].reg == NULL) |
2416 | break; |
a3684be1 |
2417 | magic = (magic & 0xffffff) | (arg << 24); |
e56ab892 |
2418 | |
2419 | for (j = i; j >= 0 && arg < pp->argc; ) |
2420 | { |
a3684be1 |
2421 | if (((ops[j].cc_scratch ^ magic) & 0xffffff) == 0) { |
2422 | if (ops[j].cc_scratch != magic) { |
2423 | ferr(&ops[j], "arg collect hit same path with diff args for %s\n", |
2424 | pp->name); |
2425 | return -1; |
2426 | } |
2427 | // ok: have already been here |
2428 | return 0; |
2429 | } |
2430 | ops[j].cc_scratch = magic; |
2431 | |
a2c1d768 |
2432 | if (g_labels[j][0] != 0 && g_label_refs[j].i != -1) { |
e56ab892 |
2433 | lr = &g_label_refs[j]; |
2434 | if (lr->next != NULL) |
2435 | need_op_saving = 1; |
a652aa9f |
2436 | for (; lr->next; lr = lr->next) { |
2437 | if ((ops[lr->i].flags & (OPF_JMP|OPF_CC)) != OPF_JMP) |
2438 | may_reuse = 1; |
a3684be1 |
2439 | ret = collect_call_args(po, lr->i, pp, save_arg_vars, |
2440 | arg, magic, need_op_saving, may_reuse); |
2441 | if (ret < 0) |
2442 | return ret; |
a652aa9f |
2443 | } |
e56ab892 |
2444 | |
a652aa9f |
2445 | if ((ops[lr->i].flags & (OPF_JMP|OPF_CC)) != OPF_JMP) |
2446 | may_reuse = 1; |
de50b98b |
2447 | if (j > 0 && LAST_OP(j - 1)) { |
e56ab892 |
2448 | // follow last branch in reverse |
2449 | j = lr->i; |
2450 | continue; |
2451 | } |
2452 | need_op_saving = 1; |
a3684be1 |
2453 | ret = collect_call_args(po, lr->i, pp, save_arg_vars, |
2454 | arg, magic, need_op_saving, may_reuse); |
2455 | if (ret < 0) |
2456 | return ret; |
e56ab892 |
2457 | } |
2458 | j--; |
2459 | |
2460 | if (ops[j].op == OP_CALL) |
2461 | { |
2462 | pp_tmp = ops[j].datap; |
2463 | if (pp_tmp == NULL) |
7ae48d73 |
2464 | ferr(po, "arg collect hit unparsed call '%s'\n", |
2465 | ops[j].operand[0].name); |
a652aa9f |
2466 | if (may_reuse && pp_tmp->argc_stack > 0) |
de50b98b |
2467 | ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n", |
2468 | arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack); |
e56ab892 |
2469 | } |
de50b98b |
2470 | else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP) { |
2471 | ferr(po, "arg collect %d/%d hit esp adjust\n", |
2472 | arg, pp->argc); |
2473 | } |
2474 | else if (ops[j].op == OP_POP) { |
2475 | ferr(po, "arg collect %d/%d hit pop\n", arg, pp->argc); |
2476 | } |
de50b98b |
2477 | else if ((ops[j].flags & (OPF_JMP|OPF_CC)) == (OPF_JMP|OPF_CC)) |
2478 | { |
a652aa9f |
2479 | may_reuse = 1; |
de50b98b |
2480 | } |
2481 | else if (ops[j].op == OP_PUSH && !(ops[j].flags & OPF_FARG)) |
e56ab892 |
2482 | { |
2483 | pp->arg[arg].datap = &ops[j]; |
2b43685d |
2484 | need_to_save_current = 0; |
e56ab892 |
2485 | if (!need_op_saving) { |
2486 | ret = scan_for_mod(&ops[j], j + 1, i); |
2b43685d |
2487 | need_to_save_current = (ret >= 0); |
e56ab892 |
2488 | } |
2b43685d |
2489 | if (need_op_saving || need_to_save_current) { |
e56ab892 |
2490 | // mark this push as one that needs operand saving |
2491 | ops[j].flags &= ~OPF_RMD; |
de50b98b |
2492 | if (ops[j].argnum == 0) { |
2493 | ops[j].argnum = arg + 1; |
2494 | *save_arg_vars |= 1 << arg; |
2495 | } |
2496 | else if (ops[j].argnum < arg + 1) |
2497 | ferr(&ops[j], "argnum conflict (%d<%d) for '%s'\n", |
2498 | ops[j].argnum, arg + 1, pp->name); |
e56ab892 |
2499 | } |
de50b98b |
2500 | else if (ops[j].argnum == 0) |
e56ab892 |
2501 | ops[j].flags |= OPF_RMD; |
2502 | |
a652aa9f |
2503 | // some PUSHes are reused by different calls on other branches, |
de50b98b |
2504 | // but that can't happen if we didn't branch, so they |
2505 | // can be removed from future searches (handles nested calls) |
a652aa9f |
2506 | if (!may_reuse) |
de50b98b |
2507 | ops[j].flags |= OPF_FARG; |
2508 | |
e56ab892 |
2509 | // next arg |
2510 | for (arg++; arg < pp->argc; arg++) |
2511 | if (pp->arg[arg].reg == NULL) |
2512 | break; |
a3684be1 |
2513 | magic = (magic & 0xffffff) | (arg << 24); |
e56ab892 |
2514 | } |
2515 | } |
2516 | |
2517 | if (arg < pp->argc) { |
2518 | ferr(po, "arg collect failed for '%s': %d/%d\n", |
2519 | pp->name, arg, pp->argc); |
2520 | ret = -1; |
2521 | } |
2522 | return ret; |
2523 | } |
2524 | |
4c45fa73 |
2525 | static void add_label_ref(struct label_ref *lr, int op_i) |
2526 | { |
2527 | struct label_ref *lr_new; |
2528 | |
2529 | if (lr->i == -1) { |
2530 | lr->i = op_i; |
2531 | return; |
2532 | } |
2533 | |
2534 | lr_new = calloc(1, sizeof(*lr_new)); |
2535 | lr_new->i = op_i; |
a652aa9f |
2536 | lr_new->next = lr->next; |
4c45fa73 |
2537 | lr->next = lr_new; |
2538 | } |
2539 | |
04f8a628 |
2540 | static void output_std_flags(FILE *fout, struct parsed_op *po, |
2541 | int *pfomask, const char *dst_opr_text) |
2542 | { |
2543 | if (*pfomask & (1 << PFO_Z)) { |
2544 | fprintf(fout, "\n cond_z = (%s%s == 0);", |
2545 | lmod_cast_u(po, po->operand[0].lmod), dst_opr_text); |
2546 | *pfomask &= ~(1 << PFO_Z); |
2547 | } |
2548 | if (*pfomask & (1 << PFO_S)) { |
2549 | fprintf(fout, "\n cond_s = (%s%s < 0);", |
2550 | lmod_cast_s(po, po->operand[0].lmod), dst_opr_text); |
2551 | *pfomask &= ~(1 << PFO_S); |
2552 | } |
2553 | } |
2554 | |
91977a1c |
2555 | static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt) |
2556 | { |
69a3cdfc |
2557 | struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op; |
850c9265 |
2558 | struct parsed_opr *last_arith_dst = NULL; |
3ebea2cf |
2559 | char buf1[256], buf2[256], buf3[256], cast[64]; |
bd96f656 |
2560 | const struct parsed_proto *pp_c; |
940e8e66 |
2561 | struct parsed_proto *pp, *pp_tmp; |
4c45fa73 |
2562 | struct parsed_data *pd; |
91977a1c |
2563 | const char *tmpname; |
940e8e66 |
2564 | enum parsed_flag_op pfo; |
69a3cdfc |
2565 | int save_arg_vars = 0; |
2566 | int cmp_result_vars = 0; |
87bf6cec |
2567 | int need_mul_var = 0; |
a3684be1 |
2568 | int have_func_ret = 0; |
a2c1d768 |
2569 | int have_normal_ret = 0; |
91977a1c |
2570 | int had_decl = 0; |
3ebea2cf |
2571 | int label_pending = 0; |
d4e3b5db |
2572 | int regmask_save = 0; |
91977a1c |
2573 | int regmask_arg = 0; |
a3684be1 |
2574 | int regmask_now = 0; |
91977a1c |
2575 | int regmask = 0; |
940e8e66 |
2576 | int pfomask = 0; |
64c59faf |
2577 | int found = 0; |
d4e3b5db |
2578 | int depth = 0; |
91977a1c |
2579 | int no_output; |
4c45fa73 |
2580 | int i, j, l; |
69a3cdfc |
2581 | int dummy; |
91977a1c |
2582 | int arg; |
91977a1c |
2583 | int reg; |
2584 | int ret; |
2585 | |
1bafb621 |
2586 | g_bp_frame = g_sp_frame = g_stack_fsz = 0; |
a2c1d768 |
2587 | g_stack_frame_used = 0; |
91977a1c |
2588 | |
bd96f656 |
2589 | g_func_pp = proto_parse(fhdr, funcn); |
2590 | if (g_func_pp == NULL) |
91977a1c |
2591 | ferr(ops, "proto_parse failed for '%s'\n", funcn); |
2592 | |
bd96f656 |
2593 | fprintf(fout, "%s ", g_func_pp->ret_type.name); |
54e763a1 |
2594 | if (g_func_pp->is_stdcall && g_func_pp->argc_reg == 0) |
2595 | fprintf(fout, "__stdcall "); |
e56ab892 |
2596 | if (g_ida_func_attr & IDAFA_NORETURN) |
2597 | fprintf(fout, "noreturn "); |
2598 | fprintf(fout, "%s(", funcn); |
54e763a1 |
2599 | |
bd96f656 |
2600 | for (i = 0; i < g_func_pp->argc; i++) { |
91977a1c |
2601 | if (i > 0) |
2602 | fprintf(fout, ", "); |
54e763a1 |
2603 | if (g_func_pp->arg[i].fptr != NULL) { |
2604 | // func pointer.. |
2605 | pp = g_func_pp->arg[i].fptr; |
2606 | fprintf(fout, "%s (", pp->ret_type.name); |
2607 | if (pp->is_stdcall && pp->argc_reg == 0) |
2608 | fprintf(fout, "__stdcall "); |
2609 | fprintf(fout, "*a%d)(", i + 1); |
2610 | for (j = 0; j < pp->argc; j++) { |
2611 | if (j > 0) |
2612 | fprintf(fout, ", "); |
2613 | if (pp->arg[j].fptr) |
2614 | ferr(ops, "nested fptr\n"); |
2615 | fprintf(fout, "%s", pp->arg[j].type.name); |
2616 | } |
2617 | fprintf(fout, ")"); |
2618 | } |
2619 | else { |
2620 | fprintf(fout, "%s a%d", g_func_pp->arg[i].type.name, i + 1); |
2621 | } |
91977a1c |
2622 | } |
bd96f656 |
2623 | if (g_func_pp->is_vararg) { |
4f12f671 |
2624 | if (i > 0) |
2625 | fprintf(fout, ", "); |
2626 | fprintf(fout, "..."); |
2627 | } |
54e763a1 |
2628 | |
91977a1c |
2629 | fprintf(fout, ")\n{\n"); |
2630 | |
2631 | // pass1: |
1bafb621 |
2632 | // - handle ebp/esp frame, remove ops related to it |
91977a1c |
2633 | if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp") |
2634 | && ops[1].op == OP_MOV |
2635 | && IS(opr_name(&ops[1], 0), "ebp") |
2636 | && IS(opr_name(&ops[1], 1), "esp")) |
2637 | { |
87bf6cec |
2638 | int ecx_push = 0; |
2639 | |
91977a1c |
2640 | g_bp_frame = 1; |
69a3cdfc |
2641 | ops[0].flags |= OPF_RMD; |
2642 | ops[1].flags |= OPF_RMD; |
4c45fa73 |
2643 | i = 2; |
91977a1c |
2644 | |
2645 | if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) { |
1bafb621 |
2646 | g_stack_fsz = opr_const(&ops[2], 1); |
69a3cdfc |
2647 | ops[2].flags |= OPF_RMD; |
4c45fa73 |
2648 | i++; |
91977a1c |
2649 | } |
87bf6cec |
2650 | else { |
2651 | // another way msvc builds stack frame.. |
2652 | i = 2; |
2653 | while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) { |
1bafb621 |
2654 | g_stack_fsz += 4; |
87bf6cec |
2655 | ops[i].flags |= OPF_RMD; |
2656 | ecx_push++; |
2657 | i++; |
2658 | } |
4c45fa73 |
2659 | // and another way.. |
2660 | if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX |
2661 | && ops[i].operand[1].type == OPT_CONST |
2662 | && ops[i + 1].op == OP_CALL |
2663 | && IS(opr_name(&ops[i + 1], 0), "__alloca_probe")) |
2664 | { |
2665 | g_stack_fsz += ops[i].operand[1].val; |
2666 | ops[i].flags |= OPF_RMD; |
2667 | i++; |
2668 | ops[i].flags |= OPF_RMD; |
2669 | i++; |
2670 | } |
87bf6cec |
2671 | } |
91977a1c |
2672 | |
64c59faf |
2673 | found = 0; |
91977a1c |
2674 | do { |
2675 | for (; i < opcnt; i++) |
2676 | if (ops[i].op == OP_RET) |
2677 | break; |
64c59faf |
2678 | if (i == opcnt && (ops[i - 1].flags & OPF_JMP) && found) |
2679 | break; |
2680 | |
e56ab892 |
2681 | if (ops[i - 1].op == OP_POP && IS(opr_name(&ops[i - 1], 0), "ebp")) |
2682 | ops[i - 1].flags |= OPF_RMD; |
2683 | else if (!(g_ida_func_attr & IDAFA_NORETURN)) |
91977a1c |
2684 | ferr(&ops[i - 1], "'pop ebp' expected\n"); |
91977a1c |
2685 | |
1bafb621 |
2686 | if (g_stack_fsz != 0) { |
e56ab892 |
2687 | if (ops[i - 2].op == OP_MOV |
2688 | && IS(opr_name(&ops[i - 2], 0), "esp") |
2689 | && IS(opr_name(&ops[i - 2], 1), "ebp")) |
91977a1c |
2690 | { |
e56ab892 |
2691 | ops[i - 2].flags |= OPF_RMD; |
91977a1c |
2692 | } |
e56ab892 |
2693 | else if (!(g_ida_func_attr & IDAFA_NORETURN)) |
2694 | ferr(&ops[i - 2], "esp restore expected\n"); |
87bf6cec |
2695 | |
2696 | if (ecx_push && ops[i - 3].op == OP_POP |
2697 | && IS(opr_name(&ops[i - 3], 0), "ecx")) |
2698 | { |
2699 | ferr(&ops[i - 3], "unexpected ecx pop\n"); |
2700 | } |
91977a1c |
2701 | } |
87bf6cec |
2702 | |
64c59faf |
2703 | found = 1; |
91977a1c |
2704 | i++; |
2705 | } while (i < opcnt); |
2706 | } |
1bafb621 |
2707 | else { |
2708 | for (i = 0; i < opcnt; i++) { |
2709 | if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL))) |
2710 | break; |
2711 | if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP |
2712 | && ops[i].operand[1].type == OPT_CONST) |
2713 | { |
2714 | g_sp_frame = 1; |
2715 | break; |
2716 | } |
2717 | } |
2718 | |
2719 | if (g_sp_frame) |
2720 | { |
2721 | g_stack_fsz = ops[i].operand[1].val; |
2722 | ops[i].flags |= OPF_RMD; |
2723 | |
2724 | i++; |
2725 | do { |
2726 | for (; i < opcnt; i++) |
2727 | if (ops[i].op == OP_RET) |
2728 | break; |
2729 | if (ops[i - 1].op != OP_ADD |
2730 | || !IS(opr_name(&ops[i - 1], 0), "esp") |
2731 | || ops[i - 1].operand[1].type != OPT_CONST |
2732 | || ops[i - 1].operand[1].val != g_stack_fsz) |
2733 | ferr(&ops[i - 1], "'add esp' expected\n"); |
2734 | ops[i - 1].flags |= OPF_RMD; |
2735 | |
2736 | i++; |
2737 | } while (i < opcnt); |
2738 | } |
2739 | } |
91977a1c |
2740 | |
2741 | // pass2: |
7ae48d73 |
2742 | // - parse calls with labels |
87bf6cec |
2743 | // - resolve all branches |
de50b98b |
2744 | for (i = 0; i < opcnt; i++) |
2745 | { |
87bf6cec |
2746 | po = &ops[i]; |
2747 | po->bt_i = -1; |
4c45fa73 |
2748 | po->btj = NULL; |
87bf6cec |
2749 | |
7ae48d73 |
2750 | if (po->flags & OPF_RMD) |
2751 | continue; |
2752 | |
2753 | if (po->op == OP_CALL) { |
2754 | if (po->operand[0].type == OPT_LABEL) { |
2755 | tmpname = opr_name(po, 0); |
46411e6c |
2756 | if (IS_START(tmpname, "loc_")) |
2757 | ferr(po, "call to loc_*\n"); |
7ae48d73 |
2758 | pp_c = proto_parse(fhdr, tmpname); |
2759 | if (pp_c == NULL) |
2760 | ferr(po, "proto_parse failed for call '%s'\n", tmpname); |
2761 | if (pp_c->is_fptr && pp_c->argc_reg != 0) |
2762 | ferr(po, "fptr call with reg arg\n"); |
2763 | pp = proto_clone(pp_c); |
2764 | my_assert_not(pp, NULL); |
2765 | po->datap = pp; |
2766 | } |
2767 | continue; |
2768 | } |
2769 | |
2770 | if (!(po->flags & OPF_JMP) || po->op == OP_RET) |
87bf6cec |
2771 | continue; |
2772 | |
4c45fa73 |
2773 | if (po->operand[0].type == OPT_REGMEM) { |
2774 | char *p = strchr(po->operand[0].name, '['); |
2775 | if (p == NULL) |
2776 | ferr(po, "unhandled indirect branch\n"); |
2777 | ret = p - po->operand[0].name; |
2778 | strncpy(buf1, po->operand[0].name, ret); |
2779 | buf1[ret] = 0; |
2780 | |
2781 | for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) { |
2782 | if (IS(g_func_pd[j].label, buf1)) { |
2783 | pd = &g_func_pd[j]; |
2784 | break; |
2785 | } |
2786 | } |
2787 | if (pd == NULL) |
840257f6 |
2788 | //ferr(po, "label '%s' not parsed?\n", buf1); |
2789 | goto tailcall; |
4c45fa73 |
2790 | if (pd->type != OPT_OFFSET) |
2791 | ferr(po, "label '%s' with non-offset data?\n", buf1); |
2792 | |
2793 | // find all labels, link |
2794 | for (j = 0; j < pd->count; j++) { |
2795 | for (l = 0; l < opcnt; l++) { |
2796 | if (g_labels[l][0] && IS(g_labels[l], pd->d[j].u.label)) { |
2797 | add_label_ref(&g_label_refs[l], i); |
2798 | pd->d[j].bt_i = l; |
2799 | break; |
2800 | } |
2801 | } |
2802 | } |
2803 | |
2804 | po->btj = pd; |
2805 | continue; |
2806 | } |
2807 | |
2808 | for (l = 0; l < opcnt; l++) { |
2809 | if (g_labels[l][0] && IS(po->operand[0].name, g_labels[l])) { |
2810 | add_label_ref(&g_label_refs[l], i); |
2811 | po->bt_i = l; |
87bf6cec |
2812 | break; |
2813 | } |
2814 | } |
2815 | |
4c45fa73 |
2816 | if (po->bt_i != -1) |
2817 | continue; |
2818 | |
840257f6 |
2819 | if (po->operand[0].type == OPT_LABEL) |
87bf6cec |
2820 | // assume tail call |
840257f6 |
2821 | goto tailcall; |
4c45fa73 |
2822 | |
2823 | ferr(po, "unhandled branch\n"); |
840257f6 |
2824 | |
2825 | tailcall: |
2826 | po->op = OP_CALL; |
2827 | po->flags |= OPF_TAIL; |
7ae48d73 |
2828 | i--; // reprocess |
87bf6cec |
2829 | } |
2830 | |
2831 | // pass3: |
a2c1d768 |
2832 | // - remove dead labels |
91977a1c |
2833 | // - process calls |
1bafb621 |
2834 | for (i = 0; i < opcnt; i++) |
2835 | { |
a2c1d768 |
2836 | if (g_labels[i][0] != 0 && g_label_refs[i].i == -1) |
2837 | g_labels[i][0] = 0; |
2838 | |
69a3cdfc |
2839 | po = &ops[i]; |
2840 | if (po->flags & OPF_RMD) |
91977a1c |
2841 | continue; |
850c9265 |
2842 | |
d4e3b5db |
2843 | if (po->op == OP_CALL) |
69a3cdfc |
2844 | { |
1bafb621 |
2845 | tmpname = opr_name(po, 0); |
7ae48d73 |
2846 | pp = po->datap; |
2847 | if (pp == NULL) |
1bafb621 |
2848 | { |
7ae48d73 |
2849 | // indirect call |
1cd4a663 |
2850 | pp_c = resolve_call(i, opcnt); |
2851 | if (pp_c != NULL) |
2852 | pp = proto_clone(pp_c); |
2853 | if (pp == NULL) { |
2854 | pp = calloc(1, sizeof(*pp)); |
2855 | my_assert_not(pp, NULL); |
2856 | ret = scan_for_esp_adjust(i + 1, opcnt, &j); |
2857 | if (ret < 0) |
2858 | ferr(po, "non-__cdecl indirect call unhandled yet\n"); |
2859 | j /= 4; |
2860 | if (j > ARRAY_SIZE(pp->arg)) |
2861 | ferr(po, "esp adjust too large?\n"); |
2862 | pp->ret_type.name = strdup("int"); |
2863 | pp->argc = pp->argc_stack = j; |
2864 | for (arg = 0; arg < pp->argc; arg++) |
2865 | pp->arg[arg].type.name = strdup("int"); |
2866 | } |
7ae48d73 |
2867 | po->datap = pp; |
1bafb621 |
2868 | } |
2869 | |
7ba45c34 |
2870 | // look for and make use of esp adjust |
2871 | ret = -1; |
e56ab892 |
2872 | if (!pp->is_stdcall && pp->argc_stack > 0) |
7ba45c34 |
2873 | ret = scan_for_esp_adjust(i + 1, opcnt, &j); |
1bafb621 |
2874 | if (ret >= 0) { |
7ba45c34 |
2875 | if (pp->is_vararg) { |
2876 | if (j / 4 < pp->argc_stack) |
2877 | ferr(po, "esp adjust is too small: %x < %x\n", |
2878 | j, pp->argc_stack * 4); |
2879 | // modify pp to make it have varargs as normal args |
2880 | arg = pp->argc; |
2881 | pp->argc += j / 4 - pp->argc_stack; |
2882 | for (; arg < pp->argc; arg++) { |
3ebea2cf |
2883 | pp->arg[arg].type.name = strdup("int"); |
7ba45c34 |
2884 | pp->argc_stack++; |
2885 | } |
2886 | if (pp->argc > ARRAY_SIZE(pp->arg)) |
e56ab892 |
2887 | ferr(po, "too many args for '%s'\n", tmpname); |
7ba45c34 |
2888 | } |
1bafb621 |
2889 | if (pp->argc_stack != j / 4) |
e56ab892 |
2890 | ferr(po, "stack tracking failed for '%s': %x %x\n", |
2891 | tmpname, pp->argc_stack * 4, j); |
7ba45c34 |
2892 | |
1bafb621 |
2893 | ops[ret].flags |= OPF_RMD; |
7ba45c34 |
2894 | // a bit of a hack, but deals with use of |
2895 | // single adj for multiple calls |
2896 | ops[ret].operand[1].val -= j; |
1bafb621 |
2897 | } |
7ba45c34 |
2898 | else if (pp->is_vararg) |
2899 | ferr(po, "missing esp_adjust for vararg func '%s'\n", |
2900 | pp->name); |
91977a1c |
2901 | |
39b168b8 |
2902 | // can't call functions with non-__cdecl callbacks yet |
2903 | for (arg = 0; arg < pp->argc; arg++) { |
2904 | if (pp->arg[arg].fptr != NULL) { |
2905 | pp_tmp = pp->arg[arg].fptr; |
1cd4a663 |
2906 | if (pp_tmp->argc_reg != 0) |
2907 | ferr(po, "'%s' has a callback with reg-args\n", tmpname); |
39b168b8 |
2908 | } |
2909 | } |
2910 | |
a3684be1 |
2911 | collect_call_args(po, i, pp, &save_arg_vars, |
2912 | 0, i + opcnt * 2, 0, 0); |
2b43685d |
2913 | |
2914 | if (strstr(pp->ret_type.name, "int64")) |
2915 | need_mul_var = 1; |
a3684be1 |
2916 | if (!(po->flags & OPF_TAIL) && !IS(pp->ret_type.name, "void")) |
2917 | have_func_ret = 1; |
91977a1c |
2918 | } |
d4e3b5db |
2919 | } |
2920 | |
2921 | // pass4: |
2922 | // - find POPs for PUSHes, rm both |
2923 | // - scan for all used registers |
2924 | // - find flag set ops for their users |
1bafb621 |
2925 | // - declare indirect functions |
d4e3b5db |
2926 | for (i = 0; i < opcnt; i++) { |
2927 | po = &ops[i]; |
2928 | if (po->flags & OPF_RMD) |
2929 | continue; |
2930 | |
7ae48d73 |
2931 | if (po->op == OP_PUSH && (po->flags & OPF_RSAVE)) { |
2932 | reg = po->operand[0].reg; |
2933 | if (!(regmask & (1 << reg))) |
2934 | // not a reg save after all, rerun scan_for_pop |
2935 | po->flags &= ~OPF_RSAVE; |
2936 | else |
2937 | regmask_save |= 1 << reg; |
2938 | } |
2939 | |
d4e3b5db |
2940 | if (po->op == OP_PUSH |
de50b98b |
2941 | && po->argnum == 0 && !(po->flags & OPF_RSAVE) |
d4e3b5db |
2942 | && po->operand[0].type == OPT_REG) |
2943 | { |
2944 | reg = po->operand[0].reg; |
2945 | if (reg < 0) |
2946 | ferr(po, "reg not set for push?\n"); |
2947 | |
2948 | depth = 0; |
2949 | ret = scan_for_pop(i + 1, opcnt, |
a3684be1 |
2950 | po->operand[0].name, i + opcnt * 3, 0, &depth, 0); |
d4e3b5db |
2951 | if (ret == 1) { |
2952 | if (depth > 1) |
2953 | ferr(po, "too much depth: %d\n", depth); |
d4e3b5db |
2954 | |
2955 | po->flags |= OPF_RMD; |
2956 | scan_for_pop(i + 1, opcnt, po->operand[0].name, |
a3684be1 |
2957 | i + opcnt * 4, 0, &depth, 1); |
d4e3b5db |
2958 | continue; |
2959 | } |
2960 | ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0); |
2961 | if (ret == 0) { |
2962 | arg = OPF_RMD; |
2963 | if (regmask & (1 << reg)) { |
2964 | if (regmask_save & (1 << reg)) |
2965 | ferr(po, "%s already saved?\n", po->operand[0].name); |
2966 | arg = OPF_RSAVE; |
2967 | } |
2968 | po->flags |= arg; |
2969 | scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg); |
2970 | continue; |
2971 | } |
2972 | } |
2973 | |
a3684be1 |
2974 | regmask_now = po->regmask_src | po->regmask_dst; |
2975 | if (regmask_now & (1 << xBP)) { |
2976 | if (g_bp_frame && !(po->flags & OPF_EBP_S)) { |
2977 | if (po->regmask_dst & (1 << xBP)) |
2978 | // compiler decided to drop bp frame and use ebp as scratch |
2979 | scan_fwd_set_flags(i, opcnt, i + opcnt * 5, OPF_EBP_S); |
2980 | else |
2981 | regmask_now &= ~(1 << xBP); |
2982 | } |
2983 | } |
2984 | |
2985 | regmask |= regmask_now; |
d4e3b5db |
2986 | |
2987 | if (po->flags & OPF_CC) |
2988 | { |
2b43685d |
2989 | int setters[16], cnt = 0, branched = 0; |
2990 | |
04f8a628 |
2991 | ret = scan_for_flag_set(i, i + opcnt * 6, |
2992 | &branched, setters, &cnt); |
2b43685d |
2993 | if (ret < 0 || cnt <= 0) |
2994 | ferr(po, "unable to trace flag setter(s)\n"); |
2995 | if (cnt > ARRAY_SIZE(setters)) |
2996 | ferr(po, "too many flag setters\n"); |
d4e3b5db |
2997 | |
d4e3b5db |
2998 | pfo = split_cond(po, po->op, &dummy); |
2b43685d |
2999 | for (j = 0; j < cnt; j++) |
3000 | { |
3001 | tmp_op = &ops[setters[j]]; // flag setter |
3002 | pfomask = 0; |
3003 | |
3004 | // to get nicer code, we try to delay test and cmp; |
3005 | // if we can't because of operand modification, or if we |
3006 | // have math op, or branch, make it calculate flags explicitly |
3007 | if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP) { |
3008 | if (branched || scan_for_mod(tmp_op, setters[j] + 1, i) >= 0) |
3009 | pfomask = 1 << pfo; |
3010 | } |
3011 | else if (tmp_op->op == OP_CMPS) { |
3012 | pfomask = 1 << PFO_Z; |
3013 | } |
3014 | else { |
04f8a628 |
3015 | // see if we'll be able to handle based on op result |
3016 | if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR |
3017 | && pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P) |
3018 | || branched |
2b43685d |
3019 | || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0) |
3020 | pfomask = 1 << pfo; |
3021 | } |
3022 | if (pfomask) { |
3023 | tmp_op->pfomask |= pfomask; |
3024 | cmp_result_vars |= pfomask; |
2b43685d |
3025 | } |
04f8a628 |
3026 | // note: may overwrite, currently not a problem |
3027 | po->datap = tmp_op; |
d4e3b5db |
3028 | } |
3029 | |
3030 | if (po->op == OP_ADC || po->op == OP_SBB) |
3031 | cmp_result_vars |= 1 << PFO_C; |
3032 | } |
87bf6cec |
3033 | else if (po->op == OP_MUL |
3034 | || (po->op == OP_IMUL && po->operand_cnt == 1)) |
3035 | { |
3036 | need_mul_var = 1; |
3037 | } |
1bafb621 |
3038 | else if (po->op == OP_CALL && po->operand[0].type != OPT_LABEL) { |
3039 | pp = po->datap; |
3040 | my_assert_not(pp, NULL); |
1cd4a663 |
3041 | fprintf(fout, " %s (", pp->ret_type.name); |
3042 | if (pp->is_stdcall && pp->argc_reg == 0) |
3043 | fprintf(fout, "__stdcall "); |
3044 | fprintf(fout, "*icall%d)(", i); |
1bafb621 |
3045 | for (j = 0; j < pp->argc; j++) { |
3046 | if (j > 0) |
3047 | fprintf(fout, ", "); |
3ebea2cf |
3048 | fprintf(fout, "%s a%d", pp->arg[j].type.name, j + 1); |
1bafb621 |
3049 | } |
3050 | fprintf(fout, ");\n"); |
3051 | } |
a2c1d768 |
3052 | else if (po->op == OP_RET) |
3053 | have_normal_ret = 1; |
91977a1c |
3054 | } |
3055 | |
4c45fa73 |
3056 | // output LUTs/jumptables |
3057 | for (i = 0; i < g_func_pd_cnt; i++) { |
3058 | pd = &g_func_pd[i]; |
3059 | fprintf(fout, " static const "); |
3060 | if (pd->type == OPT_OFFSET) { |
3061 | fprintf(fout, "void *jt_%s[] =\n { ", pd->label); |
3062 | |
3063 | for (j = 0; j < pd->count; j++) { |
3064 | if (j > 0) |
3065 | fprintf(fout, ", "); |
3066 | fprintf(fout, "&&%s", pd->d[j].u.label); |
3067 | } |
3068 | } |
3069 | else { |
3070 | fprintf(fout, "%s %s[] =\n { ", |
3071 | lmod_type_u(ops, pd->lmod), pd->label); |
3072 | |
3073 | for (j = 0; j < pd->count; j++) { |
3074 | if (j > 0) |
3075 | fprintf(fout, ", "); |
3076 | fprintf(fout, "%u", pd->d[j].u.val); |
3077 | } |
3078 | } |
3079 | fprintf(fout, " };\n"); |
3080 | } |
3081 | |
4f12f671 |
3082 | // declare stack frame, va_arg |
1bafb621 |
3083 | if (g_stack_fsz) |
850c9265 |
3084 | fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n", |
1bafb621 |
3085 | (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz); |
850c9265 |
3086 | |
bd96f656 |
3087 | if (g_func_pp->is_vararg) |
4f12f671 |
3088 | fprintf(fout, " va_list ap;\n"); |
3089 | |
940e8e66 |
3090 | // declare arg-registers |
bd96f656 |
3091 | for (i = 0; i < g_func_pp->argc; i++) { |
3092 | if (g_func_pp->arg[i].reg != NULL) { |
91977a1c |
3093 | reg = char_array_i(regs_r32, |
bd96f656 |
3094 | ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg); |
91977a1c |
3095 | if (reg < 0) |
bd96f656 |
3096 | ferr(ops, "arg '%s' is not a reg?\n", g_func_pp->arg[i].reg); |
91977a1c |
3097 | |
3098 | regmask_arg |= 1 << reg; |
3099 | fprintf(fout, " u32 %s = (u32)a%d;\n", |
bd96f656 |
3100 | g_func_pp->arg[i].reg, i + 1); |
91977a1c |
3101 | had_decl = 1; |
3102 | } |
3103 | } |
3104 | |
940e8e66 |
3105 | // declare other regs - special case for eax |
a2c1d768 |
3106 | if (!((regmask | regmask_arg) & 1) |
3107 | && (have_func_ret || have_normal_ret) |
bd96f656 |
3108 | && !IS(g_func_pp->ret_type.name, "void")) |
3ebea2cf |
3109 | { |
91977a1c |
3110 | fprintf(fout, " u32 eax = 0;\n"); |
3111 | had_decl = 1; |
3112 | } |
3113 | |
3114 | regmask &= ~regmask_arg; |
d4e3b5db |
3115 | regmask &= ~(1 << xSP); |
91977a1c |
3116 | if (regmask) { |
3117 | for (reg = 0; reg < 8; reg++) { |
3118 | if (regmask & (1 << reg)) { |
3119 | fprintf(fout, " u32 %s;\n", regs_r32[reg]); |
3120 | had_decl = 1; |
3121 | } |
3122 | } |
3123 | } |
3124 | |
d4e3b5db |
3125 | if (regmask_save) { |
3126 | for (reg = 0; reg < 8; reg++) { |
3127 | if (regmask_save & (1 << reg)) { |
3128 | fprintf(fout, " u32 s_%s;\n", regs_r32[reg]); |
3129 | had_decl = 1; |
3130 | } |
3131 | } |
3132 | } |
3133 | |
69a3cdfc |
3134 | if (save_arg_vars) { |
3135 | for (reg = 0; reg < 32; reg++) { |
3136 | if (save_arg_vars & (1 << reg)) { |
3137 | fprintf(fout, " u32 s_a%d;\n", reg + 1); |
3138 | had_decl = 1; |
3139 | } |
3140 | } |
3141 | } |
3142 | |
3143 | if (cmp_result_vars) { |
3144 | for (i = 0; i < 8; i++) { |
3145 | if (cmp_result_vars & (1 << i)) { |
3146 | fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]); |
3147 | had_decl = 1; |
3148 | } |
3149 | } |
3150 | } |
3151 | |
87bf6cec |
3152 | if (need_mul_var) { |
3153 | fprintf(fout, " u64 mul_tmp;\n"); |
3154 | had_decl = 1; |
3155 | } |
3156 | |
91977a1c |
3157 | if (had_decl) |
3158 | fprintf(fout, "\n"); |
3159 | |
bd96f656 |
3160 | if (g_func_pp->is_vararg) { |
3161 | if (g_func_pp->argc_stack == 0) |
4f12f671 |
3162 | ferr(ops, "vararg func without stack args?\n"); |
bd96f656 |
3163 | fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc); |
4f12f671 |
3164 | } |
3165 | |
91977a1c |
3166 | // output ops |
69a3cdfc |
3167 | for (i = 0; i < opcnt; i++) |
3168 | { |
a2c1d768 |
3169 | if (g_labels[i][0] != 0) { |
91977a1c |
3170 | fprintf(fout, "\n%s:\n", g_labels[i]); |
3ebea2cf |
3171 | label_pending = 1; |
2b43685d |
3172 | |
3173 | delayed_flag_op = NULL; |
3174 | last_arith_dst = NULL; |
3ebea2cf |
3175 | } |
91977a1c |
3176 | |
69a3cdfc |
3177 | po = &ops[i]; |
3178 | if (po->flags & OPF_RMD) |
91977a1c |
3179 | continue; |
3180 | |
3181 | no_output = 0; |
3182 | |
91977a1c |
3183 | #define assert_operand_cnt(n_) \ |
850c9265 |
3184 | if (po->operand_cnt != n_) \ |
3185 | ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_) |
3186 | |
69a3cdfc |
3187 | // conditional/flag using op? |
3188 | if (po->flags & OPF_CC) |
850c9265 |
3189 | { |
940e8e66 |
3190 | int is_delayed = 0; |
3191 | int is_inv = 0; |
69a3cdfc |
3192 | |
940e8e66 |
3193 | pfo = split_cond(po, po->op, &is_inv); |
04f8a628 |
3194 | tmp_op = po->datap; |
850c9265 |
3195 | |
69a3cdfc |
3196 | // we go through all this trouble to avoid using parsed_flag_op, |
3197 | // which makes generated code much nicer |
3198 | if (delayed_flag_op != NULL) |
850c9265 |
3199 | { |
940e8e66 |
3200 | out_cmp_test(buf1, sizeof(buf1), delayed_flag_op, pfo, is_inv); |
3201 | is_delayed = 1; |
91977a1c |
3202 | } |
850c9265 |
3203 | else if (last_arith_dst != NULL |
04f8a628 |
3204 | && (pfo == PFO_Z || pfo == PFO_S || pfo == PFO_P |
3205 | || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR)) |
3206 | )) |
850c9265 |
3207 | { |
3ebea2cf |
3208 | out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst); |
940e8e66 |
3209 | out_test_for_cc(buf1, sizeof(buf1), po, pfo, is_inv, |
850c9265 |
3210 | last_arith_dst->lmod, buf3); |
940e8e66 |
3211 | is_delayed = 1; |
850c9265 |
3212 | } |
04f8a628 |
3213 | else if (tmp_op != NULL) { |
7ba45c34 |
3214 | // use preprocessed flag calc results |
04f8a628 |
3215 | if (!(tmp_op->pfomask & (1 << pfo))) |
69a3cdfc |
3216 | ferr(po, "not prepared for pfo %d\n", pfo); |
3217 | |
940e8e66 |
3218 | // note: is_inv was not yet applied |
69a3cdfc |
3219 | snprintf(buf1, sizeof(buf1), "(%scond_%s)", |
940e8e66 |
3220 | is_inv ? "!" : "", parsed_flag_op_names[pfo]); |
69a3cdfc |
3221 | } |
3222 | else { |
3223 | ferr(po, "all methods of finding comparison failed\n"); |
3224 | } |
850c9265 |
3225 | |
69a3cdfc |
3226 | if (po->flags & OPF_JMP) { |
850c9265 |
3227 | fprintf(fout, " if %s\n", buf1); |
3228 | } |
5101a5f9 |
3229 | else if (po->op == OP_ADC || po->op == OP_SBB) { |
940e8e66 |
3230 | if (is_delayed) |
3231 | fprintf(fout, " cond_%s = %s;\n", |
3232 | parsed_flag_op_names[pfo], buf1); |
850c9265 |
3233 | } |
5101a5f9 |
3234 | else if (po->flags & OPF_DATA) { // SETcc |
850c9265 |
3235 | out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]); |
3236 | fprintf(fout, " %s = %s;", buf2, buf1); |
91977a1c |
3237 | } |
69a3cdfc |
3238 | else { |
3239 | ferr(po, "unhandled conditional op\n"); |
3240 | } |
91977a1c |
3241 | } |
3242 | |
940e8e66 |
3243 | pfomask = po->pfomask; |
3244 | |
850c9265 |
3245 | switch (po->op) |
91977a1c |
3246 | { |
3247 | case OP_MOV: |
3248 | assert_operand_cnt(2); |
850c9265 |
3249 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
de50b98b |
3250 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
3251 | fprintf(fout, " %s = %s;", buf1, |
3ebea2cf |
3252 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], |
3253 | po->operand[0].is_ptr ? "(void *)" : "", 0)); |
850c9265 |
3254 | break; |
3255 | |
3256 | case OP_LEA: |
3257 | assert_operand_cnt(2); |
87bf6cec |
3258 | po->operand[1].lmod = OPLM_DWORD; // always |
850c9265 |
3259 | fprintf(fout, " %s = %s;", |
3260 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
3ebea2cf |
3261 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], |
3262 | NULL, 1)); |
850c9265 |
3263 | break; |
3264 | |
3265 | case OP_MOVZX: |
3266 | assert_operand_cnt(2); |
91977a1c |
3267 | fprintf(fout, " %s = %s;", |
850c9265 |
3268 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
3ebea2cf |
3269 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1])); |
850c9265 |
3270 | break; |
3271 | |
3272 | case OP_MOVSX: |
3273 | assert_operand_cnt(2); |
3274 | switch (po->operand[1].lmod) { |
3275 | case OPLM_BYTE: |
3276 | strcpy(buf3, "(s8)"); |
3277 | break; |
3278 | case OPLM_WORD: |
3279 | strcpy(buf3, "(s16)"); |
3280 | break; |
3281 | default: |
3282 | ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod); |
3283 | } |
a2c1d768 |
3284 | fprintf(fout, " %s = %s;", |
850c9265 |
3285 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
a2c1d768 |
3286 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], |
3287 | buf3, 0)); |
850c9265 |
3288 | break; |
3289 | |
3290 | case OP_NOT: |
3291 | assert_operand_cnt(1); |
3292 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
3293 | fprintf(fout, " %s = ~%s;", buf1, buf1); |
3294 | break; |
3295 | |
5101a5f9 |
3296 | case OP_CDQ: |
3297 | assert_operand_cnt(2); |
3298 | fprintf(fout, " %s = (s32)%s >> 31;", |
3299 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
3ebea2cf |
3300 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1])); |
5101a5f9 |
3301 | strcpy(g_comment, "cdq"); |
3302 | break; |
3303 | |
33c35af6 |
3304 | case OP_STOS: |
3305 | // assumes DF=0 |
3306 | assert_operand_cnt(3); |
3307 | if (po->flags & OPF_REP) { |
3308 | fprintf(fout, " for (; ecx != 0; ecx--, edi += %d)\n", |
3309 | lmod_bytes(po, po->operand[0].lmod)); |
d4e3b5db |
3310 | fprintf(fout, " %sedi = eax;", |
3311 | lmod_cast_u_ptr(po, po->operand[0].lmod)); |
33c35af6 |
3312 | strcpy(g_comment, "rep stos"); |
3313 | } |
3314 | else { |
d4e3b5db |
3315 | fprintf(fout, " %sedi = eax; edi += %d;", |
3316 | lmod_cast_u_ptr(po, po->operand[0].lmod), |
33c35af6 |
3317 | lmod_bytes(po, po->operand[0].lmod)); |
3318 | strcpy(g_comment, "stos"); |
3319 | } |
3320 | break; |
3321 | |
d4e3b5db |
3322 | case OP_MOVS: |
3323 | // assumes DF=0 |
3324 | assert_operand_cnt(3); |
3325 | j = lmod_bytes(po, po->operand[0].lmod); |
3326 | strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod)); |
3327 | if (po->flags & OPF_REP) { |
3328 | fprintf(fout, |
3329 | " for (; ecx != 0; ecx--, edi += %d, esi += %d)\n", |
3330 | j, j); |
3331 | fprintf(fout, |
3332 | " %sedi = %sesi;", buf1, buf1); |
3333 | strcpy(g_comment, "rep movs"); |
3334 | } |
3335 | else { |
3336 | fprintf(fout, " %sedi = %sesi; edi += %d; esi += %d;", |
3337 | buf1, buf1, j, j); |
3338 | strcpy(g_comment, "movs"); |
3339 | } |
3340 | break; |
3341 | |
7ba45c34 |
3342 | case OP_CMPS: |
3343 | // assumes DF=0 |
3344 | // repe ~ repeat while ZF=1 |
3345 | assert_operand_cnt(3); |
3346 | j = lmod_bytes(po, po->operand[0].lmod); |
3347 | strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod)); |
3348 | if (po->flags & OPF_REP) { |
3349 | fprintf(fout, |
3350 | " for (; ecx != 0; ecx--, edi += %d, esi += %d)\n", |
3351 | j, j); |
3352 | fprintf(fout, |
3353 | " if ((cond_z = (%sedi == %sesi)) %s 0)\n", |
3354 | buf1, buf1, (po->flags & OPF_REPZ) ? "==" : "!="); |
3355 | fprintf(fout, |
3356 | " break;"); |
3357 | snprintf(g_comment, sizeof(g_comment), "rep%s cmps", |
3358 | (po->flags & OPF_REPZ) ? "e" : "ne"); |
3359 | } |
3360 | else { |
3361 | fprintf(fout, |
3362 | " cond_z = (%sedi = %sesi); edi += %d; esi += %d;", |
3363 | buf1, buf1, j, j); |
3364 | strcpy(g_comment, "cmps"); |
3365 | } |
3366 | pfomask &= ~(1 << PFO_Z); |
3367 | last_arith_dst = NULL; |
3368 | delayed_flag_op = NULL; |
3369 | break; |
3370 | |
850c9265 |
3371 | // arithmetic w/flags |
3372 | case OP_ADD: |
3373 | case OP_SUB: |
3374 | case OP_AND: |
3375 | case OP_OR: |
5101a5f9 |
3376 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
3377 | // fallthrough |
850c9265 |
3378 | dualop_arith: |
3379 | assert_operand_cnt(2); |
850c9265 |
3380 | fprintf(fout, " %s %s= %s;", |
3381 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
3382 | op_to_c(po), |
3ebea2cf |
3383 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1])); |
04f8a628 |
3384 | output_std_flags(fout, po, &pfomask, buf1); |
3385 | last_arith_dst = &po->operand[0]; |
3386 | delayed_flag_op = NULL; |
3387 | break; |
3388 | |
3389 | case OP_SHL: |
3390 | case OP_SHR: |
3391 | assert_operand_cnt(2); |
3392 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
3393 | if (pfomask & (1 << PFO_C)) { |
3394 | if (po->operand[1].type == OPT_CONST) { |
3395 | l = lmod_bytes(po, po->operand[0].lmod) * 8; |
3396 | j = po->operand[1].val; |
3397 | j %= l; |
3398 | if (j != 0) { |
3399 | if (po->op == OP_SHL) |
3400 | j = l - j; |
3401 | else |
3402 | j -= 1; |
3403 | fprintf(fout, " cond_c = (%s & 0x%02x) ? 1 : 0;\n", |
3404 | buf1, 1 << j); |
3405 | } |
3406 | else |
3407 | ferr(po, "zero shift?\n"); |
3408 | } |
3409 | else |
3410 | ferr(po, "TODO\n"); |
3411 | pfomask &= ~(1 << PFO_C); |
840257f6 |
3412 | } |
04f8a628 |
3413 | fprintf(fout, " %s %s= %s;", buf1, op_to_c(po), |
3414 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1])); |
3415 | output_std_flags(fout, po, &pfomask, buf1); |
850c9265 |
3416 | last_arith_dst = &po->operand[0]; |
69a3cdfc |
3417 | delayed_flag_op = NULL; |
850c9265 |
3418 | break; |
3419 | |
d4e3b5db |
3420 | case OP_SAR: |
3421 | assert_operand_cnt(2); |
3422 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
3423 | fprintf(fout, " %s = %s%s >> %s;", buf1, |
3424 | lmod_cast_s(po, po->operand[0].lmod), buf1, |
3ebea2cf |
3425 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1])); |
04f8a628 |
3426 | output_std_flags(fout, po, &pfomask, buf1); |
d4e3b5db |
3427 | last_arith_dst = &po->operand[0]; |
3428 | delayed_flag_op = NULL; |
3429 | break; |
3430 | |
3431 | case OP_ROL: |
3432 | case OP_ROR: |
3433 | assert_operand_cnt(2); |
3434 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
3435 | if (po->operand[1].type == OPT_CONST) { |
3436 | j = po->operand[1].val; |
3437 | j %= lmod_bytes(po, po->operand[0].lmod) * 8; |
3438 | fprintf(fout, po->op == OP_ROL ? |
3439 | " %s = (%s << %d) | (%s >> %d);" : |
3440 | " %s = (%s >> %d) | (%s << %d);", |
3441 | buf1, buf1, j, buf1, |
3442 | lmod_bytes(po, po->operand[0].lmod) * 8 - j); |
3443 | } |
3444 | else |
3445 | ferr(po, "TODO\n"); |
04f8a628 |
3446 | output_std_flags(fout, po, &pfomask, buf1); |
d4e3b5db |
3447 | last_arith_dst = &po->operand[0]; |
3448 | delayed_flag_op = NULL; |
3449 | break; |
3450 | |
5101a5f9 |
3451 | case OP_XOR: |
850c9265 |
3452 | assert_operand_cnt(2); |
3453 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
5101a5f9 |
3454 | if (IS(opr_name(po, 0), opr_name(po, 1))) { |
3455 | // special case for XOR |
3456 | fprintf(fout, " %s = 0;", |
3457 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0])); |
3458 | last_arith_dst = &po->operand[0]; |
3459 | delayed_flag_op = NULL; |
850c9265 |
3460 | break; |
850c9265 |
3461 | } |
5101a5f9 |
3462 | goto dualop_arith; |
3463 | |
5101a5f9 |
3464 | case OP_ADC: |
850c9265 |
3465 | case OP_SBB: |
5101a5f9 |
3466 | assert_operand_cnt(2); |
3467 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
a2c1d768 |
3468 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
840257f6 |
3469 | if (po->op == OP_SBB |
3470 | && IS(po->operand[0].name, po->operand[1].name)) |
3471 | { |
3472 | // avoid use of unitialized var |
a2c1d768 |
3473 | fprintf(fout, " %s = -cond_c;", buf1); |
94d447fb |
3474 | // carry remains what it was |
3475 | pfomask &= ~(1 << PFO_C); |
840257f6 |
3476 | } |
3477 | else { |
a2c1d768 |
3478 | fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po), |
3ebea2cf |
3479 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1])); |
840257f6 |
3480 | } |
a2c1d768 |
3481 | output_std_flags(fout, po, &pfomask, buf1); |
5101a5f9 |
3482 | last_arith_dst = &po->operand[0]; |
3483 | delayed_flag_op = NULL; |
850c9265 |
3484 | break; |
3485 | |
3486 | case OP_INC: |
3487 | case OP_DEC: |
3488 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
1bafb621 |
3489 | if (po->operand[0].type == OPT_REG) { |
3490 | strcpy(buf2, po->op == OP_INC ? "++" : "--"); |
3491 | fprintf(fout, " %s%s;", buf1, buf2); |
3492 | } |
3493 | else { |
3494 | strcpy(buf2, po->op == OP_INC ? "+" : "-"); |
3495 | fprintf(fout, " %s %s= 1;", buf1, buf2); |
3496 | } |
a2c1d768 |
3497 | output_std_flags(fout, po, &pfomask, buf1); |
5101a5f9 |
3498 | last_arith_dst = &po->operand[0]; |
3499 | delayed_flag_op = NULL; |
3500 | break; |
3501 | |
3502 | case OP_NEG: |
3503 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
3ebea2cf |
3504 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]); |
5101a5f9 |
3505 | fprintf(fout, " %s = -%s%s;", buf1, |
3506 | lmod_cast_s(po, po->operand[0].lmod), buf2); |
850c9265 |
3507 | last_arith_dst = &po->operand[0]; |
69a3cdfc |
3508 | delayed_flag_op = NULL; |
940e8e66 |
3509 | if (pfomask & (1 << PFO_C)) { |
3510 | fprintf(fout, "\n cond_c = (%s != 0);", buf1); |
3511 | pfomask &= ~(1 << PFO_C); |
3512 | } |
850c9265 |
3513 | break; |
3514 | |
3515 | case OP_IMUL: |
de50b98b |
3516 | if (po->operand_cnt == 2) { |
3517 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
850c9265 |
3518 | goto dualop_arith; |
de50b98b |
3519 | } |
87bf6cec |
3520 | if (po->operand_cnt == 3) |
3521 | ferr(po, "TODO imul3\n"); |
3522 | // fallthrough |
3523 | case OP_MUL: |
3524 | assert_operand_cnt(1); |
3525 | strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)"); |
3526 | fprintf(fout, " mul_tmp = %seax * %s%s;\n", buf1, buf1, |
3ebea2cf |
3527 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0])); |
87bf6cec |
3528 | fprintf(fout, " edx = mul_tmp >> 32;\n"); |
3529 | fprintf(fout, " eax = mul_tmp;"); |
3530 | last_arith_dst = NULL; |
69a3cdfc |
3531 | delayed_flag_op = NULL; |
91977a1c |
3532 | break; |
3533 | |
5101a5f9 |
3534 | case OP_DIV: |
3535 | case OP_IDIV: |
3536 | assert_operand_cnt(1); |
3537 | if (po->operand[0].lmod != OPLM_DWORD) |
3538 | ferr(po, "unhandled lmod %d\n", po->operand[0].lmod); |
3539 | |
3540 | // 32bit division is common, look for it |
64c59faf |
3541 | if (po->op == OP_DIV) |
cdfaeed7 |
3542 | ret = scan_for_reg_clear(i, xDX); |
64c59faf |
3543 | else |
cdfaeed7 |
3544 | ret = scan_for_cdq_edx(i); |
64c59faf |
3545 | if (ret >= 0) { |
3ebea2cf |
3546 | out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]); |
5101a5f9 |
3547 | strcpy(buf2, lmod_cast(po, po->operand[0].lmod, |
3548 | po->op == OP_IDIV)); |
3549 | fprintf(fout, " edx = %seax %% %s%s;\n", buf2, buf2, buf1); |
3550 | fprintf(fout, " eax = %seax / %s%s;", buf2, buf2, buf1); |
3551 | } |
3552 | else |
3553 | ferr(po, "TODO 64bit divident\n"); |
87bf6cec |
3554 | last_arith_dst = NULL; |
3555 | delayed_flag_op = NULL; |
5101a5f9 |
3556 | break; |
3557 | |
91977a1c |
3558 | case OP_TEST: |
3559 | case OP_CMP: |
850c9265 |
3560 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
940e8e66 |
3561 | if (pfomask != 0) { |
69a3cdfc |
3562 | for (j = 0; j < 8; j++) { |
940e8e66 |
3563 | if (pfomask & (1 << j)) { |
69a3cdfc |
3564 | out_cmp_test(buf1, sizeof(buf1), po, j, 0); |
3565 | fprintf(fout, " cond_%s = %s;", |
3566 | parsed_flag_op_names[j], buf1); |
3567 | } |
3568 | } |
940e8e66 |
3569 | pfomask = 0; |
69a3cdfc |
3570 | } |
3571 | else |
3572 | no_output = 1; |
7ba45c34 |
3573 | last_arith_dst = NULL; |
69a3cdfc |
3574 | delayed_flag_op = po; |
91977a1c |
3575 | break; |
3576 | |
69a3cdfc |
3577 | // note: we reuse OP_Jcc for SETcc, only flags differ |
91977a1c |
3578 | case OP_JO ... OP_JG: |
5101a5f9 |
3579 | if (po->flags & OPF_JMP) |
850c9265 |
3580 | fprintf(fout, " goto %s;", po->operand[0].name); |
5101a5f9 |
3581 | // else SETcc - should already be handled |
850c9265 |
3582 | break; |
3583 | |
3584 | case OP_JMP: |
87bf6cec |
3585 | assert_operand_cnt(1); |
de50b98b |
3586 | last_arith_dst = NULL; |
3587 | delayed_flag_op = NULL; |
3588 | |
4c45fa73 |
3589 | if (po->operand[0].type == OPT_REGMEM) { |
3590 | ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]", |
3591 | buf1, buf2); |
3592 | if (ret != 2) |
3593 | ferr(po, "parse failure for jmp '%s'\n", |
3594 | po->operand[0].name); |
3595 | fprintf(fout, " goto *jt_%s[%s];", buf1, buf2); |
3596 | break; |
3597 | } |
3598 | else if (po->operand[0].type != OPT_LABEL) |
3599 | ferr(po, "unhandled jmp type\n"); |
87bf6cec |
3600 | |
850c9265 |
3601 | fprintf(fout, " goto %s;", po->operand[0].name); |
91977a1c |
3602 | break; |
3603 | |
3604 | case OP_CALL: |
5101a5f9 |
3605 | assert_operand_cnt(1); |
850c9265 |
3606 | pp = po->datap; |
91977a1c |
3607 | if (pp == NULL) |
850c9265 |
3608 | ferr(po, "NULL pp\n"); |
91977a1c |
3609 | |
1bafb621 |
3610 | if (po->operand[0].type != OPT_LABEL) |
1cd4a663 |
3611 | fprintf(fout, " icall%d = %s;\n", i, |
3612 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], |
3613 | "(void *)", 0)); |
1bafb621 |
3614 | |
91977a1c |
3615 | fprintf(fout, " "); |
2b43685d |
3616 | if (strstr(pp->ret_type.name, "int64")) { |
87bf6cec |
3617 | if (po->flags & OPF_TAIL) |
2b43685d |
3618 | ferr(po, "int64 and tail?\n"); |
3619 | fprintf(fout, "mul_tmp = "); |
3620 | } |
3621 | else if (!IS(pp->ret_type.name, "void")) { |
3622 | if (po->flags & OPF_TAIL) { |
840257f6 |
3623 | if (!IS(g_func_pp->ret_type.name, "void")) { |
3624 | fprintf(fout, "return "); |
3625 | if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr) |
3626 | fprintf(fout, "(%s)", g_func_pp->ret_type.name); |
3627 | } |
2b43685d |
3628 | } |
3629 | else { |
87bf6cec |
3630 | fprintf(fout, "eax = "); |
2b43685d |
3631 | if (pp->ret_type.is_ptr) |
3632 | fprintf(fout, "(u32)"); |
3633 | } |
91977a1c |
3634 | } |
87bf6cec |
3635 | |
39b168b8 |
3636 | if (po->operand[0].type != OPT_LABEL) { |
1bafb621 |
3637 | fprintf(fout, "icall%d(", i); |
39b168b8 |
3638 | } |
3639 | else { |
3640 | if (pp->name[0] == 0) |
3641 | ferr(po, "missing pp->name\n"); |
a652aa9f |
3642 | fprintf(fout, "%s%s(", pp->name, |
3643 | pp->has_structarg ? "_sa" : ""); |
39b168b8 |
3644 | } |
3645 | |
91977a1c |
3646 | for (arg = 0; arg < pp->argc; arg++) { |
3647 | if (arg > 0) |
3648 | fprintf(fout, ", "); |
87bf6cec |
3649 | |
3ebea2cf |
3650 | cast[0] = 0; |
3651 | if (pp->arg[arg].type.is_ptr) |
3652 | snprintf(cast, sizeof(cast), "(%s)", pp->arg[arg].type.name); |
87bf6cec |
3653 | |
91977a1c |
3654 | if (pp->arg[arg].reg != NULL) { |
3ebea2cf |
3655 | fprintf(fout, "%s%s", cast, pp->arg[arg].reg); |
91977a1c |
3656 | continue; |
3657 | } |
3658 | |
3659 | // stack arg |
3660 | tmp_op = pp->arg[arg].datap; |
3661 | if (tmp_op == NULL) |
850c9265 |
3662 | ferr(po, "parsed_op missing for arg%d\n", arg); |
de50b98b |
3663 | if (tmp_op->argnum != 0) { |
3664 | fprintf(fout, "%ss_a%d", cast, tmp_op->argnum); |
69a3cdfc |
3665 | } |
3666 | else { |
3667 | fprintf(fout, "%s", |
3668 | out_src_opr(buf1, sizeof(buf1), |
3ebea2cf |
3669 | tmp_op, &tmp_op->operand[0], cast, 0)); |
69a3cdfc |
3670 | } |
91977a1c |
3671 | } |
3672 | fprintf(fout, ");"); |
87bf6cec |
3673 | |
2b43685d |
3674 | if (strstr(pp->ret_type.name, "int64")) { |
3675 | fprintf(fout, "\n"); |
3676 | fprintf(fout, " edx = mul_tmp >> 32;\n"); |
3677 | fprintf(fout, " eax = mul_tmp;"); |
3678 | } |
3679 | |
87bf6cec |
3680 | if (po->flags & OPF_TAIL) { |
3681 | strcpy(g_comment, "tailcall"); |
840257f6 |
3682 | ret = 0; |
3683 | if (i == opcnt - 1) |
3684 | ret = 0; |
3685 | else if (IS(pp->ret_type.name, "void")) |
3686 | ret = 1; |
3687 | else if (IS(g_func_pp->ret_type.name, "void")) |
3688 | ret = 1; |
3689 | // else already handled as 'return f()' |
3690 | |
3691 | if (ret) { |
a2c1d768 |
3692 | if (!IS(g_func_pp->ret_type.name, "void")) |
3693 | ferr(po, "int func -> void func tailcall?\n"); |
87bf6cec |
3694 | fprintf(fout, "\n return;"); |
3ebea2cf |
3695 | strcpy(g_comment, "^ tailcall"); |
3696 | } |
87bf6cec |
3697 | } |
3698 | delayed_flag_op = NULL; |
3699 | last_arith_dst = NULL; |
91977a1c |
3700 | break; |
3701 | |
3702 | case OP_RET: |
bd96f656 |
3703 | if (g_func_pp->is_vararg) |
4f12f671 |
3704 | fprintf(fout, " va_end(ap);\n"); |
3705 | |
bd96f656 |
3706 | if (IS(g_func_pp->ret_type.name, "void")) { |
3ebea2cf |
3707 | if (i != opcnt - 1 || label_pending) |
3708 | fprintf(fout, " return;"); |
3709 | } |
bd96f656 |
3710 | else if (g_func_pp->ret_type.is_ptr) { |
d4e3b5db |
3711 | fprintf(fout, " return (%s)eax;", |
bd96f656 |
3712 | g_func_pp->ret_type.name); |
3ebea2cf |
3713 | } |
91977a1c |
3714 | else |
3715 | fprintf(fout, " return eax;"); |
de50b98b |
3716 | |
3717 | last_arith_dst = NULL; |
3718 | delayed_flag_op = NULL; |
91977a1c |
3719 | break; |
3720 | |
3721 | case OP_PUSH: |
de50b98b |
3722 | if (po->argnum != 0) { |
69a3cdfc |
3723 | // special case - saved func arg |
3ebea2cf |
3724 | out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]); |
de50b98b |
3725 | fprintf(fout, " s_a%d = %s;", po->argnum, buf1); |
69a3cdfc |
3726 | break; |
3727 | } |
d4e3b5db |
3728 | else if (po->flags & OPF_RSAVE) { |
3ebea2cf |
3729 | out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]); |
d4e3b5db |
3730 | fprintf(fout, " s_%s = %s;", buf1, buf1); |
3731 | break; |
3732 | } |
e56ab892 |
3733 | if (!(g_ida_func_attr & IDAFA_NORETURN)) |
3734 | ferr(po, "stray push encountered\n"); |
3735 | no_output = 1; |
91977a1c |
3736 | break; |
3737 | |
3738 | case OP_POP: |
d4e3b5db |
3739 | if (po->flags & OPF_RSAVE) { |
3ebea2cf |
3740 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
d4e3b5db |
3741 | fprintf(fout, " %s = s_%s;", buf1, buf1); |
3742 | break; |
3743 | } |
3744 | ferr(po, "stray pop encountered\n"); |
91977a1c |
3745 | break; |
3746 | |
33c35af6 |
3747 | case OP_NOP: |
2b43685d |
3748 | no_output = 1; |
33c35af6 |
3749 | break; |
3750 | |
91977a1c |
3751 | default: |
3752 | no_output = 1; |
69a3cdfc |
3753 | ferr(po, "unhandled op type %d, flags %x\n", |
3754 | po->op, po->flags); |
91977a1c |
3755 | break; |
3756 | } |
3757 | |
3758 | if (g_comment[0] != 0) { |
3759 | fprintf(fout, " // %s", g_comment); |
3760 | g_comment[0] = 0; |
3761 | no_output = 0; |
3762 | } |
3763 | if (!no_output) |
3764 | fprintf(fout, "\n"); |
5101a5f9 |
3765 | |
2b43685d |
3766 | // some sanity checking |
3767 | if ((po->flags & OPF_REP) && po->op != OP_STOS |
3768 | && po->op != OP_MOVS && po->op != OP_CMPS) |
3769 | ferr(po, "unexpected rep\n"); |
3770 | if ((po->flags & (OPF_REPZ|OPF_REPNZ)) && po->op != OP_CMPS) |
3771 | ferr(po, "unexpected repz/repnz\n"); |
3772 | |
940e8e66 |
3773 | if (pfomask != 0) |
7ba45c34 |
3774 | ferr(po, "missed flag calc, pfomask=%x\n", pfomask); |
940e8e66 |
3775 | |
5101a5f9 |
3776 | // see is delayed flag stuff is still valid |
3777 | if (delayed_flag_op != NULL && delayed_flag_op != po) { |
3778 | if (is_any_opr_modified(delayed_flag_op, po)) |
3779 | delayed_flag_op = NULL; |
3780 | } |
3781 | |
3782 | if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) { |
3783 | if (is_opr_modified(last_arith_dst, po)) |
3784 | last_arith_dst = NULL; |
3785 | } |
3ebea2cf |
3786 | |
3787 | label_pending = 0; |
91977a1c |
3788 | } |
3789 | |
a2c1d768 |
3790 | if (g_stack_fsz && !g_stack_frame_used) |
3791 | fprintf(fout, " (void)sf;\n"); |
3792 | |
91977a1c |
3793 | fprintf(fout, "}\n\n"); |
3794 | |
3795 | // cleanup |
3796 | for (i = 0; i < opcnt; i++) { |
4c45fa73 |
3797 | struct label_ref *lr, *lr_del; |
3798 | |
3799 | lr = g_label_refs[i].next; |
3800 | while (lr != NULL) { |
3801 | lr_del = lr; |
3802 | lr = lr->next; |
3803 | free(lr_del); |
3804 | } |
3805 | g_label_refs[i].i = -1; |
3806 | g_label_refs[i].next = NULL; |
3807 | |
91977a1c |
3808 | if (ops[i].op == OP_CALL) { |
3809 | pp = ops[i].datap; |
bd96f656 |
3810 | if (pp) |
91977a1c |
3811 | proto_release(pp); |
91977a1c |
3812 | } |
3813 | } |
bd96f656 |
3814 | g_func_pp = NULL; |
91977a1c |
3815 | } |
c36e914d |
3816 | |
d4e3b5db |
3817 | static void set_label(int i, const char *name) |
3818 | { |
3819 | const char *p; |
3820 | int len; |
3821 | |
3822 | len = strlen(name); |
3823 | p = strchr(name, ':'); |
3824 | if (p != NULL) |
3825 | len = p - name; |
3826 | |
3827 | if (len > sizeof(g_labels[0]) - 1) |
3828 | aerr("label '%s' too long: %d\n", name, len); |
e56ab892 |
3829 | if (g_labels[i][0] != 0 && !IS_START(g_labels[i], "algn_")) |
3830 | aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]); |
d4e3b5db |
3831 | memcpy(g_labels[i], name, len); |
3832 | g_labels[i][len] = 0; |
3833 | } |
3834 | |
bfa4a6ee |
3835 | // '=' needs special treatment.. |
3836 | static char *next_word_s(char *w, size_t wsize, char *s) |
3837 | { |
3838 | size_t i; |
3839 | |
3840 | s = sskip(s); |
3841 | |
3842 | for (i = 0; i < wsize - 1; i++) { |
3843 | if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0)) |
3844 | break; |
3845 | w[i] = s[i]; |
3846 | } |
3847 | w[i] = 0; |
3848 | |
3849 | if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=') |
3850 | printf("warning: '%s' truncated\n", w); |
3851 | |
3852 | return s + i; |
3853 | } |
3854 | |
e56ab892 |
3855 | struct chunk_item { |
3856 | char *name; |
3857 | long fptr; |
de50b98b |
3858 | int asmln; |
e56ab892 |
3859 | }; |
3860 | |
cdfaeed7 |
3861 | static struct chunk_item *func_chunks; |
3862 | static int func_chunk_cnt; |
3863 | static int func_chunk_alloc; |
3864 | |
3865 | static void add_func_chunk(FILE *fasm, const char *name, int line) |
3866 | { |
3867 | if (func_chunk_cnt >= func_chunk_alloc) { |
3868 | func_chunk_alloc *= 2; |
3869 | func_chunks = realloc(func_chunks, |
3870 | func_chunk_alloc * sizeof(func_chunks[0])); |
3871 | my_assert_not(func_chunks, NULL); |
3872 | } |
3873 | func_chunks[func_chunk_cnt].fptr = ftell(fasm); |
3874 | func_chunks[func_chunk_cnt].name = strdup(name); |
3875 | func_chunks[func_chunk_cnt].asmln = line; |
3876 | func_chunk_cnt++; |
3877 | } |
3878 | |
e56ab892 |
3879 | static int cmp_chunks(const void *p1, const void *p2) |
3880 | { |
3881 | const struct chunk_item *c1 = p1, *c2 = p2; |
3882 | return strcmp(c1->name, c2->name); |
3883 | } |
3884 | |
bfa4a6ee |
3885 | static int cmpstringp(const void *p1, const void *p2) |
3886 | { |
3887 | return strcmp(*(char * const *)p1, *(char * const *)p2); |
3888 | } |
3889 | |
cdfaeed7 |
3890 | static void scan_ahead(FILE *fasm) |
3891 | { |
3892 | char words[2][256]; |
3893 | char line[256]; |
3894 | long oldpos; |
3895 | int oldasmln; |
3896 | int wordc; |
3897 | char *p; |
3898 | int i; |
3899 | |
3900 | oldpos = ftell(fasm); |
3901 | oldasmln = asmln; |
3902 | |
3903 | while (fgets(line, sizeof(line), fasm)) |
3904 | { |
3905 | wordc = 0; |
3906 | asmln++; |
3907 | |
3908 | p = sskip(line); |
3909 | if (*p == 0) |
3910 | continue; |
3911 | |
3912 | if (*p == ';') |
3913 | { |
3914 | // get rid of random tabs |
3915 | for (i = 0; line[i] != 0; i++) |
3916 | if (line[i] == '\t') |
3917 | line[i] = ' '; |
3918 | |
3919 | if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR ")) |
3920 | { |
3921 | p += 30; |
3922 | next_word(words[0], sizeof(words[0]), p); |
3923 | if (words[0][0] == 0) |
3924 | aerr("missing name for func chunk?\n"); |
3925 | |
3926 | add_func_chunk(fasm, words[0], asmln); |
3927 | } |
3928 | continue; |
3929 | } // *p == ';' |
3930 | |
3931 | for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) { |
3932 | words[wordc][0] = 0; |
3933 | p = sskip(next_word_s(words[wordc], sizeof(words[0]), p)); |
3934 | if (*p == 0 || *p == ';') { |
3935 | wordc++; |
3936 | break; |
3937 | } |
3938 | } |
3939 | |
3940 | if (wordc == 2 && IS(words[1], "ends")) |
3941 | break; |
3942 | } |
3943 | |
3944 | fseek(fasm, oldpos, SEEK_SET); |
3945 | asmln = oldasmln; |
3946 | } |
3947 | |
91977a1c |
3948 | int main(int argc, char *argv[]) |
3949 | { |
06c5d854 |
3950 | FILE *fout, *fasm, *frlist; |
4c45fa73 |
3951 | struct parsed_data *pd = NULL; |
3952 | int pd_alloc = 0; |
3953 | char **rlist = NULL; |
3954 | int rlist_len = 0; |
3955 | int rlist_alloc = 0; |
e56ab892 |
3956 | int func_chunks_used = 0; |
3957 | int func_chunks_sorted = 0; |
e56ab892 |
3958 | int func_chunk_i = -1; |
3959 | long func_chunk_ret = 0; |
de50b98b |
3960 | int func_chunk_ret_ln = 0; |
cdfaeed7 |
3961 | int scanned_ahead = 0; |
91977a1c |
3962 | char line[256]; |
a2c1d768 |
3963 | char words[20][256]; |
4c45fa73 |
3964 | enum opr_lenmod lmod; |
91977a1c |
3965 | int in_func = 0; |
4c45fa73 |
3966 | int pending_endp = 0; |
bfa4a6ee |
3967 | int skip_func = 0; |
940e8e66 |
3968 | int skip_warned = 0; |
91977a1c |
3969 | int eq_alloc; |
bfa4a6ee |
3970 | int verbose = 0; |
3971 | int arg_out; |
3972 | int arg = 1; |
91977a1c |
3973 | int pi = 0; |
e56ab892 |
3974 | int i, j; |
3975 | int ret, len; |
91977a1c |
3976 | char *p; |
3977 | int wordc; |
3978 | |
bfa4a6ee |
3979 | if (argv[1] && IS(argv[1], "-v")) { |
3980 | verbose = 1; |
3981 | arg++; |
3982 | } |
3983 | |
3984 | if (argc < arg + 3) { |
3985 | printf("usage:\n%s [-v] <.c> <.asm> <hdrf> [rlist]*\n", |
91977a1c |
3986 | argv[0]); |
3987 | return 1; |
3988 | } |
3989 | |
bfa4a6ee |
3990 | arg_out = arg++; |
91977a1c |
3991 | |
bfa4a6ee |
3992 | asmfn = argv[arg++]; |
91977a1c |
3993 | fasm = fopen(asmfn, "r"); |
3994 | my_assert_not(fasm, NULL); |
3995 | |
bfa4a6ee |
3996 | hdrfn = argv[arg++]; |
06c5d854 |
3997 | g_fhdr = fopen(hdrfn, "r"); |
3998 | my_assert_not(g_fhdr, NULL); |
bfa4a6ee |
3999 | |
4000 | rlist_alloc = 64; |
4001 | rlist = malloc(rlist_alloc * sizeof(rlist[0])); |
4002 | my_assert_not(rlist, NULL); |
4003 | // needs special handling.. |
4004 | rlist[rlist_len++] = "__alloca_probe"; |
4005 | |
e56ab892 |
4006 | func_chunk_alloc = 32; |
4007 | func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0])); |
4008 | my_assert_not(func_chunks, NULL); |
4009 | |
a2c1d768 |
4010 | memset(words, 0, sizeof(words)); |
4011 | |
bfa4a6ee |
4012 | for (; arg < argc; arg++) { |
4013 | frlist = fopen(argv[arg], "r"); |
4014 | my_assert_not(frlist, NULL); |
4015 | |
4016 | while (fgets(line, sizeof(line), frlist)) { |
4017 | p = sskip(line); |
1cd4a663 |
4018 | if (*p == 0 || *p == ';') |
4019 | continue; |
4020 | if (*p == '#') { |
4021 | if (IS_START(p, "#if 0")) |
4022 | skip_func = 1; |
4023 | else if (IS_START(p, "#endif")) |
4024 | skip_func = 0; |
4025 | continue; |
4026 | } |
4027 | if (skip_func) |
bfa4a6ee |
4028 | continue; |
4029 | |
4030 | p = next_word(words[0], sizeof(words[0]), p); |
4031 | if (words[0][0] == 0) |
4032 | continue; |
4033 | |
4034 | if (rlist_len >= rlist_alloc) { |
4035 | rlist_alloc = rlist_alloc * 2 + 64; |
4036 | rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0])); |
4037 | my_assert_not(rlist, NULL); |
4038 | } |
4039 | rlist[rlist_len++] = strdup(words[0]); |
4040 | } |
1cd4a663 |
4041 | skip_func = 0; |
bfa4a6ee |
4042 | |
4043 | fclose(frlist); |
4044 | frlist = NULL; |
4045 | } |
4046 | |
4047 | if (rlist_len > 0) |
4048 | qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp); |
4049 | |
4050 | fout = fopen(argv[arg_out], "w"); |
91977a1c |
4051 | my_assert_not(fout, NULL); |
4052 | |
4053 | eq_alloc = 128; |
4054 | g_eqs = malloc(eq_alloc * sizeof(g_eqs[0])); |
4055 | my_assert_not(g_eqs, NULL); |
4056 | |
4c45fa73 |
4057 | for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) { |
4058 | g_label_refs[i].i = -1; |
4059 | g_label_refs[i].next = NULL; |
4060 | } |
4061 | |
91977a1c |
4062 | while (fgets(line, sizeof(line), fasm)) |
4063 | { |
4c45fa73 |
4064 | wordc = 0; |
91977a1c |
4065 | asmln++; |
4066 | |
4067 | p = sskip(line); |
1bafb621 |
4068 | if (*p == 0) |
91977a1c |
4069 | continue; |
4070 | |
de50b98b |
4071 | // get rid of random tabs |
4072 | for (i = 0; line[i] != 0; i++) |
4073 | if (line[i] == '\t') |
4074 | line[i] = ' '; |
4075 | |
e56ab892 |
4076 | if (*p == ';') |
4077 | { |
e56ab892 |
4078 | if (p[2] == '=' && IS_START(p, "; =============== S U B")) |
4079 | goto do_pending_endp; // eww.. |
4080 | |
4081 | if (p[2] == 'A' && IS_START(p, "; Attributes:")) |
4082 | { |
4083 | static const char *attrs[] = { |
4084 | "bp-based frame", |
4085 | "library function", |
4086 | "static", |
4087 | "noreturn", |
4088 | "thunk", |
4089 | "fpd=", |
4090 | }; |
4091 | |
4092 | // parse IDA's attribute-list comment |
4093 | g_ida_func_attr = 0; |
4094 | p = sskip(p + 13); |
4095 | |
4096 | for (; *p != 0; p = sskip(p)) { |
4097 | for (i = 0; i < ARRAY_SIZE(attrs); i++) { |
4098 | if (!strncmp(p, attrs[i], strlen(attrs[i]))) { |
4099 | g_ida_func_attr |= 1 << i; |
4100 | p += strlen(attrs[i]); |
4101 | break; |
4102 | } |
4103 | } |
4104 | if (i == ARRAY_SIZE(attrs)) { |
4105 | anote("unparsed IDA attr: %s\n", p); |
1bafb621 |
4106 | break; |
4107 | } |
e56ab892 |
4108 | if (IS(attrs[i], "fpd=")) { |
4109 | p = next_word(words[0], sizeof(words[0]), p); |
4110 | // ignore for now.. |
4111 | } |
1bafb621 |
4112 | } |
e56ab892 |
4113 | } |
4114 | else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR ")) |
4115 | { |
4116 | p += 30; |
4117 | next_word(words[0], sizeof(words[0]), p); |
4118 | if (words[0][0] == 0) |
cdfaeed7 |
4119 | aerr("missing name for func chunk?\n"); |
4120 | |
4121 | if (!scanned_ahead) { |
4122 | add_func_chunk(fasm, words[0], asmln); |
4123 | func_chunks_sorted = 0; |
e56ab892 |
4124 | } |
e56ab892 |
4125 | } |
4126 | else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK")) |
4127 | { |
4128 | if (func_chunk_i >= 0) { |
4129 | if (func_chunk_i < func_chunk_cnt |
4130 | && IS(func_chunks[func_chunk_i].name, g_func)) |
4131 | { |
4132 | // move on to next chunk |
4133 | ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET); |
4134 | if (ret) |
4135 | aerr("seek failed for '%s' chunk #%d\n", |
4136 | g_func, func_chunk_i); |
de50b98b |
4137 | asmln = func_chunks[func_chunk_i].asmln; |
e56ab892 |
4138 | func_chunk_i++; |
4139 | } |
4140 | else { |
4141 | if (func_chunk_ret == 0) |
4142 | aerr("no return from chunk?\n"); |
4143 | fseek(fasm, func_chunk_ret, SEEK_SET); |
de50b98b |
4144 | asmln = func_chunk_ret_ln; |
e56ab892 |
4145 | func_chunk_ret = 0; |
4146 | pending_endp = 1; |
4147 | } |
1bafb621 |
4148 | } |
e56ab892 |
4149 | } |
4150 | else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) { |
4151 | func_chunks_used = 1; |
4152 | p += 20; |
4153 | if (IS_START(g_func, "sub_")) { |
4154 | unsigned long addr = strtoul(p, NULL, 16); |
4155 | unsigned long f_addr = strtoul(g_func + 4, NULL, 16); |
cdfaeed7 |
4156 | if (addr > f_addr && !scanned_ahead) { |
94d447fb |
4157 | anote("scan_ahead caused by '%s', addr %lx\n", |
4158 | g_func, addr); |
cdfaeed7 |
4159 | scan_ahead(fasm); |
4160 | scanned_ahead = 1; |
4161 | func_chunks_sorted = 0; |
4162 | } |
1bafb621 |
4163 | } |
4164 | } |
4165 | continue; |
e56ab892 |
4166 | } // *p == ';' |
1bafb621 |
4167 | |
06c5d854 |
4168 | parse_words: |
a2c1d768 |
4169 | for (i = wordc; i < ARRAY_SIZE(words); i++) |
4170 | words[i][0] = 0; |
cdfaeed7 |
4171 | for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) { |
bfa4a6ee |
4172 | p = sskip(next_word_s(words[wordc], sizeof(words[0]), p)); |
91977a1c |
4173 | if (*p == 0 || *p == ';') { |
4174 | wordc++; |
4175 | break; |
4176 | } |
4177 | } |
a2c1d768 |
4178 | if (*p != 0 && *p != ';') |
4179 | aerr("too many words\n"); |
91977a1c |
4180 | |
06c5d854 |
4181 | // alow asm patches in comments |
3ebea2cf |
4182 | if (*p == ';' && IS_START(p, "; sctpatch:")) { |
4183 | p = sskip(p + 11); |
06c5d854 |
4184 | if (*p == 0 || *p == ';') |
4185 | continue; |
4186 | goto parse_words; // lame |
4187 | } |
4188 | |
91977a1c |
4189 | if (wordc == 0) { |
4190 | // shouldn't happen |
4191 | awarn("wordc == 0?\n"); |
4192 | continue; |
4193 | } |
4194 | |
4195 | // don't care about this: |
4196 | if (words[0][0] == '.' |
4197 | || IS(words[0], "include") |
4198 | || IS(words[0], "assume") || IS(words[1], "segment") |
4199 | || IS(words[0], "align")) |
4200 | { |
4201 | continue; |
4202 | } |
4203 | |
4c45fa73 |
4204 | do_pending_endp: |
4205 | // do delayed endp processing to collect switch jumptables |
4206 | if (pending_endp) { |
4207 | if (in_func && !skip_func && wordc >= 2 |
4208 | && ((words[0][0] == 'd' && words[0][2] == 0) |
4209 | || (words[1][0] == 'd' && words[1][2] == 0))) |
4210 | { |
4211 | i = 1; |
4212 | if (words[1][0] == 'd' && words[1][2] == 0) { |
4213 | // label |
4214 | if (g_func_pd_cnt >= pd_alloc) { |
4215 | pd_alloc = pd_alloc * 2 + 16; |
4216 | g_func_pd = realloc(g_func_pd, |
4217 | sizeof(g_func_pd[0]) * pd_alloc); |
4218 | my_assert_not(g_func_pd, NULL); |
4219 | } |
4220 | pd = &g_func_pd[g_func_pd_cnt]; |
4221 | g_func_pd_cnt++; |
4222 | memset(pd, 0, sizeof(*pd)); |
4223 | strcpy(pd->label, words[0]); |
4224 | pd->type = OPT_CONST; |
4225 | pd->lmod = lmod_from_directive(words[1]); |
4226 | i = 2; |
4227 | } |
4228 | else { |
4229 | lmod = lmod_from_directive(words[0]); |
4230 | if (lmod != pd->lmod) |
4231 | aerr("lmod change? %d->%d\n", pd->lmod, lmod); |
4232 | } |
4233 | |
4234 | if (pd->count_alloc < pd->count + wordc) { |
4235 | pd->count_alloc = pd->count_alloc * 2 + 14 + wordc; |
4236 | pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc); |
4237 | my_assert_not(pd->d, NULL); |
4238 | } |
4239 | for (; i < wordc; i++) { |
4240 | if (IS(words[i], "offset")) { |
4241 | pd->type = OPT_OFFSET; |
4242 | i++; |
4243 | } |
4244 | p = strchr(words[i], ','); |
4245 | if (p != NULL) |
4246 | *p = 0; |
4247 | if (pd->type == OPT_OFFSET) |
4248 | pd->d[pd->count].u.label = strdup(words[i]); |
4249 | else |
4250 | pd->d[pd->count].u.val = parse_number(words[i]); |
4251 | pd->d[pd->count].bt_i = -1; |
4252 | pd->count++; |
4253 | } |
4254 | continue; |
4255 | } |
4256 | |
4257 | if (in_func && !skip_func) |
4258 | gen_func(fout, g_fhdr, g_func, pi); |
4259 | |
4260 | pending_endp = 0; |
4261 | in_func = 0; |
4262 | g_ida_func_attr = 0; |
4263 | skip_warned = 0; |
4264 | skip_func = 0; |
4265 | g_func[0] = 0; |
e56ab892 |
4266 | func_chunks_used = 0; |
4267 | func_chunk_i = -1; |
4c45fa73 |
4268 | if (pi != 0) { |
4269 | memset(&ops, 0, pi * sizeof(ops[0])); |
4270 | memset(g_labels, 0, pi * sizeof(g_labels[0])); |
4271 | pi = 0; |
4272 | } |
4273 | g_eqcnt = 0; |
4274 | for (i = 0; i < g_func_pd_cnt; i++) { |
4275 | pd = &g_func_pd[i]; |
4276 | if (pd->type == OPT_OFFSET) { |
4277 | for (j = 0; j < pd->count; j++) |
4278 | free(pd->d[j].u.label); |
4279 | } |
4280 | free(pd->d); |
4281 | pd->d = NULL; |
4282 | } |
4283 | g_func_pd_cnt = 0; |
4284 | pd = NULL; |
4285 | if (wordc == 0) |
4286 | continue; |
4287 | } |
4288 | |
91977a1c |
4289 | if (IS(words[1], "proc")) { |
4290 | if (in_func) |
4291 | aerr("proc '%s' while in_func '%s'?\n", |
4292 | words[0], g_func); |
bfa4a6ee |
4293 | p = words[0]; |
4c45fa73 |
4294 | if ((g_ida_func_attr & IDAFA_THUNK) |
1bafb621 |
4295 | || bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp)) |
bfa4a6ee |
4296 | skip_func = 1; |
91977a1c |
4297 | strcpy(g_func, words[0]); |
d4e3b5db |
4298 | set_label(0, words[0]); |
91977a1c |
4299 | in_func = 1; |
4300 | continue; |
4301 | } |
4302 | |
e56ab892 |
4303 | if (IS(words[1], "endp")) |
4304 | { |
91977a1c |
4305 | if (!in_func) |
4306 | aerr("endp '%s' while not in_func?\n", words[0]); |
4307 | if (!IS(g_func, words[0])) |
4308 | aerr("endp '%s' while in_func '%s'?\n", |
4309 | words[0], g_func); |
bfa4a6ee |
4310 | |
e56ab892 |
4311 | if (!skip_func && func_chunks_used) { |
4312 | // start processing chunks |
4313 | struct chunk_item *ci, key = { g_func, 0 }; |
4314 | |
4315 | func_chunk_ret = ftell(fasm); |
de50b98b |
4316 | func_chunk_ret_ln = asmln; |
e56ab892 |
4317 | if (!func_chunks_sorted) { |
4318 | qsort(func_chunks, func_chunk_cnt, |
4319 | sizeof(func_chunks[0]), cmp_chunks); |
4320 | func_chunks_sorted = 1; |
4321 | } |
4322 | ci = bsearch(&key, func_chunks, func_chunk_cnt, |
4323 | sizeof(func_chunks[0]), cmp_chunks); |
4324 | if (ci == NULL) |
4325 | aerr("'%s' needs chunks, but none found\n", g_func); |
4326 | func_chunk_i = ci - func_chunks; |
4327 | for (; func_chunk_i > 0; func_chunk_i--) |
4328 | if (!IS(func_chunks[func_chunk_i - 1].name, g_func)) |
4329 | break; |
4330 | |
4331 | ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET); |
4332 | if (ret) |
4333 | aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i); |
de50b98b |
4334 | asmln = func_chunks[func_chunk_i].asmln; |
e56ab892 |
4335 | func_chunk_i++; |
4336 | continue; |
4337 | } |
4c45fa73 |
4338 | pending_endp = 1; |
91977a1c |
4339 | continue; |
4340 | } |
4341 | |
a2c1d768 |
4342 | if (wordc == 2 && IS(words[1], "ends")) |
4343 | break; |
4344 | |
bfa4a6ee |
4345 | p = strchr(words[0], ':'); |
4346 | if (p != NULL) { |
d4e3b5db |
4347 | set_label(pi, words[0]); |
bfa4a6ee |
4348 | continue; |
4349 | } |
4350 | |
4351 | if (!in_func || skip_func) { |
4352 | if (!skip_warned && !skip_func && g_labels[pi][0] != 0) { |
4353 | if (verbose) |
4354 | anote("skipping from '%s'\n", g_labels[pi]); |
4355 | skip_warned = 1; |
4356 | } |
4357 | g_labels[pi][0] = 0; |
4358 | continue; |
4359 | } |
4360 | |
4c45fa73 |
4361 | if (wordc > 1 && IS(words[1], "=")) { |
91977a1c |
4362 | if (wordc != 5) |
4363 | aerr("unhandled equ, wc=%d\n", wordc); |
4364 | if (g_eqcnt >= eq_alloc) { |
4365 | eq_alloc *= 2; |
4366 | g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0])); |
4367 | my_assert_not(g_eqs, NULL); |
4368 | } |
4369 | |
4370 | len = strlen(words[0]); |
4371 | if (len > sizeof(g_eqs[0].name) - 1) |
4372 | aerr("equ name too long: %d\n", len); |
4373 | strcpy(g_eqs[g_eqcnt].name, words[0]); |
4374 | |
4375 | if (!IS(words[3], "ptr")) |
4376 | aerr("unhandled equ\n"); |
4377 | if (IS(words[2], "dword")) |
4378 | g_eqs[g_eqcnt].lmod = OPLM_DWORD; |
4379 | else if (IS(words[2], "word")) |
4380 | g_eqs[g_eqcnt].lmod = OPLM_WORD; |
4381 | else if (IS(words[2], "byte")) |
4382 | g_eqs[g_eqcnt].lmod = OPLM_BYTE; |
4383 | else |
4384 | aerr("bad lmod: '%s'\n", words[2]); |
4385 | |
4386 | g_eqs[g_eqcnt].offset = parse_number(words[4]); |
4387 | g_eqcnt++; |
4388 | continue; |
4389 | } |
4390 | |
4391 | if (pi >= ARRAY_SIZE(ops)) |
4392 | aerr("too many ops\n"); |
4393 | |
91977a1c |
4394 | parse_op(&ops[pi], words, wordc); |
4395 | pi++; |
91977a1c |
4396 | } |
4397 | |
4398 | fclose(fout); |
4399 | fclose(fasm); |
06c5d854 |
4400 | fclose(g_fhdr); |
91977a1c |
4401 | |
4402 | return 0; |
c36e914d |
4403 | } |
91977a1c |
4404 | |
4405 | // vim:ts=2:shiftwidth=2:expandtab |