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 */ |
5c024ef7 |
35 | OPF_JMP = (1 << 3), /* branch, call */ |
36 | OPF_CJMP = (1 << 4), /* cond. branch (cc or jecxz) */ |
37 | OPF_CC = (1 << 5), /* uses flags */ |
38 | OPF_TAIL = (1 << 6), /* ret or tail call */ |
39 | OPF_RSAVE = (1 << 7), /* push/pop is local reg save/load */ |
40 | OPF_REP = (1 << 8), /* prefixed by rep */ |
41 | OPF_REPZ = (1 << 9), /* rep is repe/repz */ |
42 | OPF_REPNZ = (1 << 10), /* rep is repne/repnz */ |
43 | OPF_FARG = (1 << 11), /* push collected as func arg (no reuse) */ |
44 | OPF_EBP_S = (1 << 12), /* ebp used as scratch here, not BP */ |
45 | OPF_DF = (1 << 13), /* DF flag set */ |
c36e914d |
46 | }; |
47 | |
48 | enum op_op { |
49 | OP_INVAL, |
33c35af6 |
50 | OP_NOP, |
c36e914d |
51 | OP_PUSH, |
52 | OP_POP, |
591721d7 |
53 | OP_LEAVE, |
c36e914d |
54 | OP_MOV, |
850c9265 |
55 | OP_LEA, |
56 | OP_MOVZX, |
57 | OP_MOVSX, |
58 | OP_NOT, |
5101a5f9 |
59 | OP_CDQ, |
33c35af6 |
60 | OP_STOS, |
d4e3b5db |
61 | OP_MOVS, |
7ba45c34 |
62 | OP_CMPS, |
591721d7 |
63 | OP_SCAS, |
64 | OP_STD, |
65 | OP_CLD, |
c36e914d |
66 | OP_RET, |
67 | OP_ADD, |
91977a1c |
68 | OP_SUB, |
850c9265 |
69 | OP_AND, |
70 | OP_OR, |
71 | OP_XOR, |
72 | OP_SHL, |
73 | OP_SHR, |
74 | OP_SAR, |
d4e3b5db |
75 | OP_ROL, |
76 | OP_ROR, |
69a3cdfc |
77 | OP_ADC, |
850c9265 |
78 | OP_SBB, |
79 | OP_INC, |
80 | OP_DEC, |
5101a5f9 |
81 | OP_NEG, |
850c9265 |
82 | OP_MUL, |
83 | OP_IMUL, |
5101a5f9 |
84 | OP_DIV, |
85 | OP_IDIV, |
c36e914d |
86 | OP_TEST, |
87 | OP_CMP, |
88 | OP_CALL, |
89 | OP_JMP, |
5c024ef7 |
90 | OP_JECXZ, |
c36e914d |
91 | OP_JO, |
92 | OP_JNO, |
93 | OP_JC, |
94 | OP_JNC, |
95 | OP_JZ, |
96 | OP_JNZ, |
97 | OP_JBE, |
98 | OP_JA, |
99 | OP_JS, |
100 | OP_JNS, |
101 | OP_JP, |
102 | OP_JNP, |
103 | OP_JL, |
104 | OP_JGE, |
105 | OP_JLE, |
106 | OP_JG, |
107 | }; |
108 | |
109 | enum opr_type { |
87bf6cec |
110 | OPT_UNSPEC, |
111 | OPT_REG, |
112 | OPT_REGMEM, |
113 | OPT_LABEL, |
850c9265 |
114 | OPT_OFFSET, |
87bf6cec |
115 | OPT_CONST, |
c36e914d |
116 | }; |
117 | |
2b43685d |
118 | // must be sorted (larger len must be further in enum) |
c36e914d |
119 | enum opr_lenmod { |
91977a1c |
120 | OPLM_UNSPEC, |
121 | OPLM_BYTE, |
122 | OPLM_WORD, |
123 | OPLM_DWORD, |
c36e914d |
124 | }; |
125 | |
850c9265 |
126 | #define MAX_OPERANDS 3 |
c36e914d |
127 | |
128 | struct parsed_opr { |
91977a1c |
129 | enum opr_type type; |
130 | enum opr_lenmod lmod; |
7ba45c34 |
131 | unsigned int is_ptr:1; // pointer in C |
132 | unsigned int is_array:1; // array in C |
a3684be1 |
133 | unsigned int type_from_var:1; // .. in header, sometimes wrong |
2b43685d |
134 | unsigned int size_mismatch:1; // type override differs from C |
135 | unsigned int size_lt:1; // type override is larger than C |
ddaf8bd7 |
136 | unsigned int had_ds:1; // had ds: prefix |
91977a1c |
137 | int reg; |
138 | unsigned int val; |
139 | char name[256]; |
c36e914d |
140 | }; |
141 | |
142 | struct parsed_op { |
91977a1c |
143 | enum op_op op; |
144 | struct parsed_opr operand[MAX_OPERANDS]; |
69a3cdfc |
145 | unsigned int flags; |
91977a1c |
146 | int operand_cnt; |
69a3cdfc |
147 | int regmask_src; // all referensed regs |
148 | int regmask_dst; |
940e8e66 |
149 | int pfomask; // flagop: parsed_flag_op that can't be delayed |
de50b98b |
150 | int argnum; // push: altered before call arg # |
940e8e66 |
151 | int cc_scratch; // scratch storage during analysis |
4c45fa73 |
152 | int bt_i; // branch target for branches |
153 | struct parsed_data *btj;// branch targets for jumptables |
91977a1c |
154 | void *datap; |
155 | }; |
156 | |
69a3cdfc |
157 | // datap: |
ddaf8bd7 |
158 | // OP_CALL - parser proto hint (str), ptr to struct parsed_proto |
2b43685d |
159 | // (OPF_CC) - point to one of (OPF_FLAGS) that affects cc op |
5c024ef7 |
160 | // OP_POP - point to OP_PUSH in push/pop pair |
69a3cdfc |
161 | |
91977a1c |
162 | struct parsed_equ { |
163 | char name[64]; |
164 | enum opr_lenmod lmod; |
165 | int offset; |
c36e914d |
166 | }; |
167 | |
4c45fa73 |
168 | struct parsed_data { |
169 | char label[256]; |
170 | enum opr_type type; |
171 | enum opr_lenmod lmod; |
172 | int count; |
173 | int count_alloc; |
174 | struct { |
175 | union { |
176 | char *label; |
63df67be |
177 | unsigned int val; |
4c45fa73 |
178 | } u; |
179 | int bt_i; |
180 | } *d; |
181 | }; |
182 | |
183 | struct label_ref { |
184 | int i; |
185 | struct label_ref *next; |
186 | }; |
187 | |
1bafb621 |
188 | enum ida_func_attr { |
189 | IDAFA_BP_FRAME = (1 << 0), |
190 | IDAFA_LIB_FUNC = (1 << 1), |
191 | IDAFA_STATIC = (1 << 2), |
192 | IDAFA_NORETURN = (1 << 3), |
193 | IDAFA_THUNK = (1 << 4), |
194 | IDAFA_FPD = (1 << 5), |
195 | }; |
196 | |
197 | #define MAX_OPS 4096 |
c36e914d |
198 | |
199 | static struct parsed_op ops[MAX_OPS]; |
91977a1c |
200 | static struct parsed_equ *g_eqs; |
201 | static int g_eqcnt; |
202 | static char g_labels[MAX_OPS][32]; |
4c45fa73 |
203 | static struct label_ref g_label_refs[MAX_OPS]; |
bd96f656 |
204 | static const struct parsed_proto *g_func_pp; |
4c45fa73 |
205 | static struct parsed_data *g_func_pd; |
206 | static int g_func_pd_cnt; |
91977a1c |
207 | static char g_func[256]; |
208 | static char g_comment[256]; |
209 | static int g_bp_frame; |
1bafb621 |
210 | static int g_sp_frame; |
a2c1d768 |
211 | static int g_stack_frame_used; |
1bafb621 |
212 | static int g_stack_fsz; |
4c45fa73 |
213 | static int g_ida_func_attr; |
89ff3147 |
214 | static int g_allow_regfunc; |
91977a1c |
215 | #define ferr(op_, fmt, ...) do { \ |
63df67be |
216 | printf("error:%s:#%zd: '%s': " fmt, g_func, (op_) - ops, \ |
850c9265 |
217 | dump_op(op_), ##__VA_ARGS__); \ |
33c35af6 |
218 | fcloseall(); \ |
91977a1c |
219 | exit(1); \ |
220 | } while (0) |
de50b98b |
221 | #define fnote(op_, fmt, ...) \ |
222 | printf("error:%s:#%zd: '%s': " fmt, g_func, (op_) - ops, \ |
223 | dump_op(op_), ##__VA_ARGS__) |
91977a1c |
224 | |
225 | #define MAX_REGS 8 |
226 | |
227 | const char *regs_r32[] = { "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp" }; |
228 | const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" }; |
229 | const char *regs_r8l[] = { "al", "bl", "cl", "dl" }; |
230 | const char *regs_r8h[] = { "ah", "bh", "ch", "dh" }; |
231 | |
232 | enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP }; |
233 | |
69a3cdfc |
234 | // possible basic comparison types (without inversion) |
235 | enum parsed_flag_op { |
236 | PFO_O, // 0 OF=1 |
237 | PFO_C, // 2 CF=1 |
238 | PFO_Z, // 4 ZF=1 |
239 | PFO_BE, // 6 CF=1||ZF=1 |
240 | PFO_S, // 8 SF=1 |
241 | PFO_P, // a PF=1 |
242 | PFO_L, // c SF!=OF |
243 | PFO_LE, // e ZF=1||SF!=OF |
244 | }; |
245 | |
246 | static const char *parsed_flag_op_names[] = { |
247 | "o", "c", "z", "be", "s", "p", "l", "le" |
248 | }; |
249 | |
91977a1c |
250 | static int char_array_i(const char *array[], size_t len, const char *s) |
251 | { |
252 | int i; |
c36e914d |
253 | |
91977a1c |
254 | for (i = 0; i < len; i++) |
255 | if (IS(s, array[i])) |
256 | return i; |
c36e914d |
257 | |
91977a1c |
258 | return -1; |
259 | } |
260 | |
63df67be |
261 | static void printf_number(char *buf, size_t buf_size, |
262 | unsigned long number) |
91977a1c |
263 | { |
5101a5f9 |
264 | // output in C-friendly form |
265 | snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number); |
266 | } |
91977a1c |
267 | |
5101a5f9 |
268 | static int parse_reg(enum opr_lenmod *reg_lmod, const char *s) |
269 | { |
270 | int reg; |
91977a1c |
271 | |
5101a5f9 |
272 | reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s); |
273 | if (reg >= 0) { |
274 | *reg_lmod = OPLM_DWORD; |
275 | return reg; |
91977a1c |
276 | } |
5101a5f9 |
277 | reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s); |
278 | if (reg >= 0) { |
279 | *reg_lmod = OPLM_WORD; |
280 | return reg; |
281 | } |
282 | reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s); |
283 | if (reg >= 0) { |
284 | *reg_lmod = OPLM_BYTE; |
285 | return reg; |
286 | } |
287 | reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s); |
288 | if (reg >= 0) { |
289 | *reg_lmod = OPLM_BYTE; |
290 | return reg; |
850c9265 |
291 | } |
292 | |
293 | return -1; |
91977a1c |
294 | } |
295 | |
5101a5f9 |
296 | static int parse_indmode(char *name, int *regmask, int need_c_cvt) |
297 | { |
298 | enum opr_lenmod lmod; |
299 | char cvtbuf[256]; |
300 | char *d = cvtbuf; |
301 | char *s = name; |
302 | char w[64]; |
303 | long number; |
304 | int reg; |
305 | int c = 0; |
306 | |
307 | *d = 0; |
308 | |
309 | while (*s != 0) { |
310 | d += strlen(d); |
311 | while (my_isblank(*s)) |
312 | s++; |
313 | for (; my_issep(*s); d++, s++) |
314 | *d = *s; |
315 | while (my_isblank(*s)) |
316 | s++; |
317 | *d = 0; |
318 | |
87bf6cec |
319 | // skip 'ds:' prefix |
39b168b8 |
320 | if (IS_START(s, "ds:")) |
87bf6cec |
321 | s += 3; |
322 | |
5101a5f9 |
323 | s = next_idt(w, sizeof(w), s); |
324 | if (w[0] == 0) |
325 | break; |
326 | c++; |
327 | |
328 | reg = parse_reg(&lmod, w); |
329 | if (reg >= 0) { |
330 | *regmask |= 1 << reg; |
331 | goto pass; |
332 | } |
333 | |
334 | if ('0' <= w[0] && w[0] <= '9') { |
335 | number = parse_number(w); |
336 | printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number); |
337 | continue; |
338 | } |
339 | |
340 | // probably some label/identifier - pass |
341 | |
342 | pass: |
343 | snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w); |
344 | } |
345 | |
346 | if (need_c_cvt) |
347 | strcpy(name, cvtbuf); |
348 | |
349 | return c; |
350 | } |
351 | |
4c45fa73 |
352 | static int is_reg_in_str(const char *s) |
353 | { |
354 | int i; |
355 | |
356 | if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3]))) |
357 | return 0; |
358 | |
359 | for (i = 0; i < ARRAY_SIZE(regs_r32); i++) |
360 | if (!strncmp(s, regs_r32[i], 3)) |
361 | return 1; |
362 | |
363 | return 0; |
364 | } |
365 | |
366 | static const char *parse_stack_el(const char *name, char *extra_reg) |
d4e3b5db |
367 | { |
4c45fa73 |
368 | const char *p, *p2, *s; |
1bafb621 |
369 | char *endp = NULL; |
d4e3b5db |
370 | char buf[32]; |
1bafb621 |
371 | long val; |
d4e3b5db |
372 | int len; |
373 | |
4c45fa73 |
374 | p = name; |
375 | if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) { |
376 | p += 4; |
377 | if (extra_reg != NULL) { |
378 | strncpy(extra_reg, name, 3); |
379 | extra_reg[4] = 0; |
380 | } |
d4e3b5db |
381 | } |
4c45fa73 |
382 | |
383 | if (IS_START(p, "ebp+")) { |
384 | p += 4; |
385 | |
386 | p2 = strchr(p, '+'); |
387 | if (p2 != NULL && is_reg_in_str(p)) { |
388 | if (extra_reg != NULL) { |
389 | strncpy(extra_reg, p, p2 - p); |
390 | extra_reg[p2 - p] = 0; |
391 | } |
392 | p = p2 + 1; |
393 | } |
394 | |
395 | if (!('0' <= *p && *p <= '9')) |
396 | return p; |
397 | |
398 | return NULL; |
399 | } |
400 | |
39b168b8 |
401 | if (!IS_START(name, "esp+")) |
d4e3b5db |
402 | return NULL; |
403 | |
404 | p = strchr(name + 4, '+'); |
405 | if (p) { |
1bafb621 |
406 | // must be a number after esp+, already converted to 0x.. |
407 | s = name + 4; |
408 | if (!('0' <= *s && *s <= '9')) { |
591721d7 |
409 | aerr("%s nan?\n", __func__); |
d4e3b5db |
410 | return NULL; |
1bafb621 |
411 | } |
412 | if (s[0] == '0' && s[1] == 'x') |
413 | s += 2; |
414 | len = p - s; |
d4e3b5db |
415 | if (len < sizeof(buf) - 1) { |
1bafb621 |
416 | strncpy(buf, s, len); |
d4e3b5db |
417 | buf[len] = 0; |
1bafb621 |
418 | val = strtol(buf, &endp, 16); |
419 | if (val == 0 || *endp != 0) { |
420 | aerr("%s num parse fail for '%s'\n", __func__, buf); |
421 | return NULL; |
422 | } |
d4e3b5db |
423 | } |
424 | p++; |
425 | } |
426 | else |
427 | p = name + 4; |
428 | |
429 | if ('0' <= *p && *p <= '9') |
430 | return NULL; |
431 | |
432 | return p; |
433 | } |
434 | |
850c9265 |
435 | static int guess_lmod_from_name(struct parsed_opr *opr) |
436 | { |
437 | if (!strncmp(opr->name, "dword_", 6)) { |
438 | opr->lmod = OPLM_DWORD; |
439 | return 1; |
440 | } |
441 | if (!strncmp(opr->name, "word_", 5)) { |
442 | opr->lmod = OPLM_WORD; |
443 | return 1; |
444 | } |
445 | if (!strncmp(opr->name, "byte_", 5)) { |
446 | opr->lmod = OPLM_BYTE; |
447 | return 1; |
448 | } |
449 | return 0; |
450 | } |
451 | |
3ebea2cf |
452 | static int guess_lmod_from_c_type(enum opr_lenmod *lmod, |
453 | const struct parsed_type *c_type) |
06c5d854 |
454 | { |
06c5d854 |
455 | static const char *dword_types[] = { |
da87ae38 |
456 | "int", "_DWORD", "UINT_PTR", "DWORD", |
840257f6 |
457 | "WPARAM", "LPARAM", "UINT", |
da87ae38 |
458 | "HIMC", |
06c5d854 |
459 | }; |
460 | static const char *word_types[] = { |
89ff3147 |
461 | "uint16_t", "int16_t", "_WORD", |
2b43685d |
462 | "unsigned __int16", "__int16", |
06c5d854 |
463 | }; |
464 | static const char *byte_types[] = { |
2b43685d |
465 | "uint8_t", "int8_t", "char", |
89ff3147 |
466 | "unsigned __int8", "__int8", "BYTE", "_BYTE", |
a3684be1 |
467 | "_UNKNOWN", |
06c5d854 |
468 | }; |
3ebea2cf |
469 | const char *n; |
06c5d854 |
470 | int i; |
471 | |
3ebea2cf |
472 | if (c_type->is_ptr) { |
473 | *lmod = OPLM_DWORD; |
06c5d854 |
474 | return 1; |
475 | } |
476 | |
3ebea2cf |
477 | n = skip_type_mod(c_type->name); |
06c5d854 |
478 | |
3ebea2cf |
479 | for (i = 0; i < ARRAY_SIZE(dword_types); i++) { |
480 | if (IS(n, dword_types[i])) { |
481 | *lmod = OPLM_DWORD; |
06c5d854 |
482 | return 1; |
483 | } |
484 | } |
485 | |
486 | for (i = 0; i < ARRAY_SIZE(word_types); i++) { |
3ebea2cf |
487 | if (IS(n, word_types[i])) { |
488 | *lmod = OPLM_WORD; |
06c5d854 |
489 | return 1; |
490 | } |
491 | } |
492 | |
493 | for (i = 0; i < ARRAY_SIZE(byte_types); i++) { |
3ebea2cf |
494 | if (IS(n, byte_types[i])) { |
495 | *lmod = OPLM_BYTE; |
06c5d854 |
496 | return 1; |
497 | } |
498 | } |
499 | |
06c5d854 |
500 | return 0; |
501 | } |
502 | |
4c45fa73 |
503 | static enum opr_type lmod_from_directive(const char *d) |
504 | { |
505 | if (IS(d, "dd")) |
506 | return OPLM_DWORD; |
507 | else if (IS(d, "dw")) |
508 | return OPLM_WORD; |
509 | else if (IS(d, "db")) |
510 | return OPLM_BYTE; |
511 | |
512 | aerr("unhandled directive: '%s'\n", d); |
513 | return OPLM_UNSPEC; |
514 | } |
515 | |
5101a5f9 |
516 | static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod, |
517 | int *regmask) |
518 | { |
519 | opr->type = OPT_REG; |
520 | opr->reg = reg; |
521 | opr->lmod = lmod; |
522 | *regmask |= 1 << reg; |
523 | } |
524 | |
39b168b8 |
525 | static struct parsed_equ *equ_find(struct parsed_op *po, const char *name, |
526 | int *extra_offs); |
87bf6cec |
527 | |
69a3cdfc |
528 | static int parse_operand(struct parsed_opr *opr, |
529 | int *regmask, int *regmask_indirect, |
1bafb621 |
530 | char words[16][256], int wordc, int w, unsigned int op_flags) |
c36e914d |
531 | { |
bd96f656 |
532 | const struct parsed_proto *pp; |
850c9265 |
533 | enum opr_lenmod tmplmod; |
63df67be |
534 | unsigned long number; |
89ff3147 |
535 | char buf[256]; |
850c9265 |
536 | int ret, len; |
1bafb621 |
537 | int wordc_in; |
89ff3147 |
538 | char *p; |
850c9265 |
539 | int i; |
c36e914d |
540 | |
1bafb621 |
541 | if (w >= wordc) |
542 | aerr("parse_operand w %d, wordc %d\n", w, wordc); |
c36e914d |
543 | |
1bafb621 |
544 | opr->reg = xUNSPEC; |
91977a1c |
545 | |
1bafb621 |
546 | for (i = w; i < wordc; i++) { |
547 | len = strlen(words[i]); |
548 | if (words[i][len - 1] == ',') { |
549 | words[i][len - 1] = 0; |
550 | wordc = i + 1; |
551 | break; |
552 | } |
553 | } |
c36e914d |
554 | |
1bafb621 |
555 | wordc_in = wordc - w; |
c36e914d |
556 | |
1bafb621 |
557 | if ((op_flags & OPF_JMP) && wordc_in > 0 |
558 | && !('0' <= words[w][0] && words[w][0] <= '9')) |
559 | { |
560 | const char *label = NULL; |
561 | |
562 | if (wordc_in == 3 && !strncmp(words[w], "near", 4) |
563 | && IS(words[w + 1], "ptr")) |
564 | label = words[w + 2]; |
565 | else if (wordc_in == 2 && IS(words[w], "short")) |
566 | label = words[w + 1]; |
567 | else if (wordc_in == 1 |
568 | && strchr(words[w], '[') == NULL |
569 | && parse_reg(&tmplmod, words[w]) < 0) |
570 | label = words[w]; |
571 | |
572 | if (label != NULL) { |
573 | opr->type = OPT_LABEL; |
ddaf8bd7 |
574 | if (IS_START(label, "ds:")) { |
575 | opr->had_ds = 1; |
39b168b8 |
576 | label += 3; |
ddaf8bd7 |
577 | } |
1bafb621 |
578 | strcpy(opr->name, label); |
579 | return wordc; |
580 | } |
581 | } |
c36e914d |
582 | |
1bafb621 |
583 | if (wordc_in >= 3) { |
584 | if (IS(words[w + 1], "ptr")) { |
585 | if (IS(words[w], "dword")) |
586 | opr->lmod = OPLM_DWORD; |
587 | else if (IS(words[w], "word")) |
588 | opr->lmod = OPLM_WORD; |
589 | else if (IS(words[w], "byte")) |
590 | opr->lmod = OPLM_BYTE; |
591 | else |
592 | aerr("type parsing failed\n"); |
593 | w += 2; |
594 | wordc_in = wordc - w; |
595 | } |
596 | } |
597 | |
39b168b8 |
598 | if (wordc_in == 2) { |
599 | if (IS(words[w], "offset")) { |
600 | opr->type = OPT_OFFSET; |
601 | strcpy(opr->name, words[w + 1]); |
602 | return wordc; |
603 | } |
604 | if (IS(words[w], "(offset")) { |
89ff3147 |
605 | p = strchr(words[w + 1], ')'); |
39b168b8 |
606 | if (p == NULL) |
607 | aerr("parse of bracketed offset failed\n"); |
608 | *p = 0; |
609 | opr->type = OPT_OFFSET; |
610 | strcpy(opr->name, words[w + 1]); |
611 | return wordc; |
612 | } |
1bafb621 |
613 | } |
c36e914d |
614 | |
1bafb621 |
615 | if (wordc_in != 1) |
850c9265 |
616 | aerr("parse_operand 1 word expected\n"); |
c36e914d |
617 | |
ddaf8bd7 |
618 | if (IS_START(words[w], "ds:")) { |
619 | opr->had_ds = 1; |
89ff3147 |
620 | memmove(words[w], words[w] + 3, strlen(words[w]) - 2); |
ddaf8bd7 |
621 | } |
89ff3147 |
622 | strcpy(opr->name, words[w]); |
c36e914d |
623 | |
850c9265 |
624 | if (words[w][0] == '[') { |
625 | opr->type = OPT_REGMEM; |
626 | ret = sscanf(words[w], "[%[^]]]", opr->name); |
627 | if (ret != 1) |
628 | aerr("[] parse failure\n"); |
87bf6cec |
629 | |
5101a5f9 |
630 | parse_indmode(opr->name, regmask_indirect, 1); |
4c45fa73 |
631 | if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL)) { |
87bf6cec |
632 | // might be an equ |
d4e3b5db |
633 | struct parsed_equ *eq = |
4c45fa73 |
634 | equ_find(NULL, parse_stack_el(opr->name, NULL), &i); |
87bf6cec |
635 | if (eq) |
636 | opr->lmod = eq->lmod; |
637 | } |
850c9265 |
638 | return wordc; |
639 | } |
640 | else if (strchr(words[w], '[')) { |
641 | // label[reg] form |
89ff3147 |
642 | p = strchr(words[w], '['); |
850c9265 |
643 | opr->type = OPT_REGMEM; |
89ff3147 |
644 | parse_indmode(p, regmask_indirect, 0); |
645 | strncpy(buf, words[w], p - words[w]); |
646 | buf[p - words[w]] = 0; |
647 | pp = proto_parse(g_fhdr, buf, 1); |
648 | goto do_label; |
850c9265 |
649 | } |
650 | else if (('0' <= words[w][0] && words[w][0] <= '9') |
651 | || words[w][0] == '-') |
652 | { |
5101a5f9 |
653 | number = parse_number(words[w]); |
91977a1c |
654 | opr->type = OPT_CONST; |
5101a5f9 |
655 | opr->val = number; |
656 | printf_number(opr->name, sizeof(opr->name), number); |
91977a1c |
657 | return wordc; |
850c9265 |
658 | } |
c36e914d |
659 | |
5101a5f9 |
660 | ret = parse_reg(&tmplmod, opr->name); |
661 | if (ret >= 0) { |
662 | setup_reg_opr(opr, ret, tmplmod, regmask); |
850c9265 |
663 | return wordc; |
664 | } |
665 | |
666 | // most likely var in data segment |
667 | opr->type = OPT_LABEL; |
36595fd2 |
668 | pp = proto_parse(g_fhdr, opr->name, 0); |
89ff3147 |
669 | |
670 | do_label: |
bd96f656 |
671 | if (pp != NULL) { |
1cd4a663 |
672 | if (pp->is_fptr || pp->is_func) { |
06c5d854 |
673 | opr->lmod = OPLM_DWORD; |
674 | opr->is_ptr = 1; |
675 | } |
2b43685d |
676 | else { |
840257f6 |
677 | tmplmod = OPLM_UNSPEC; |
bd96f656 |
678 | if (!guess_lmod_from_c_type(&tmplmod, &pp->type)) |
2b43685d |
679 | anote("unhandled C type '%s' for '%s'\n", |
bd96f656 |
680 | pp->type.name, opr->name); |
2b43685d |
681 | |
a3684be1 |
682 | if (opr->lmod == OPLM_UNSPEC) { |
2b43685d |
683 | opr->lmod = tmplmod; |
a3684be1 |
684 | opr->type_from_var = 1; |
685 | } |
2b43685d |
686 | else if (opr->lmod != tmplmod) { |
687 | opr->size_mismatch = 1; |
688 | if (tmplmod < opr->lmod) |
689 | opr->size_lt = 1; |
690 | } |
a652aa9f |
691 | opr->is_ptr = pp->type.is_ptr; |
3ebea2cf |
692 | } |
bd96f656 |
693 | opr->is_array = pp->type.is_array; |
06c5d854 |
694 | } |
695 | |
850c9265 |
696 | if (opr->lmod == OPLM_UNSPEC) |
697 | guess_lmod_from_name(opr); |
850c9265 |
698 | return wordc; |
c36e914d |
699 | } |
700 | |
33c35af6 |
701 | static const struct { |
702 | const char *name; |
703 | unsigned int flags; |
704 | } pref_table[] = { |
705 | { "rep", OPF_REP }, |
7ba45c34 |
706 | { "repe", OPF_REP|OPF_REPZ }, |
707 | { "repz", OPF_REP|OPF_REPZ }, |
708 | { "repne", OPF_REP|OPF_REPNZ }, |
709 | { "repnz", OPF_REP|OPF_REPNZ }, |
33c35af6 |
710 | }; |
711 | |
5c024ef7 |
712 | #define OPF_CJMP_CC (OPF_JMP|OPF_CJMP|OPF_CC) |
713 | |
c36e914d |
714 | static const struct { |
850c9265 |
715 | const char *name; |
716 | enum op_op op; |
69a3cdfc |
717 | unsigned int minopr; |
718 | unsigned int maxopr; |
719 | unsigned int flags; |
c36e914d |
720 | } op_table[] = { |
33c35af6 |
721 | { "nop", OP_NOP, 0, 0, 0 }, |
69a3cdfc |
722 | { "push", OP_PUSH, 1, 1, 0 }, |
723 | { "pop", OP_POP, 1, 1, OPF_DATA }, |
591721d7 |
724 | { "leave",OP_LEAVE, 0, 0, OPF_DATA }, |
69a3cdfc |
725 | { "mov" , OP_MOV, 2, 2, OPF_DATA }, |
726 | { "lea", OP_LEA, 2, 2, OPF_DATA }, |
727 | { "movzx",OP_MOVZX, 2, 2, OPF_DATA }, |
728 | { "movsx",OP_MOVSX, 2, 2, OPF_DATA }, |
729 | { "not", OP_NOT, 1, 1, OPF_DATA }, |
5101a5f9 |
730 | { "cdq", OP_CDQ, 0, 0, OPF_DATA }, |
33c35af6 |
731 | { "stosb",OP_STOS, 0, 0, OPF_DATA }, |
732 | { "stosw",OP_STOS, 0, 0, OPF_DATA }, |
733 | { "stosd",OP_STOS, 0, 0, OPF_DATA }, |
d4e3b5db |
734 | { "movsb",OP_MOVS, 0, 0, OPF_DATA }, |
735 | { "movsw",OP_MOVS, 0, 0, OPF_DATA }, |
736 | { "movsd",OP_MOVS, 0, 0, OPF_DATA }, |
7ba45c34 |
737 | { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS }, |
738 | { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS }, |
739 | { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS }, |
591721d7 |
740 | { "scasb",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS }, |
741 | { "scasw",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS }, |
742 | { "scasd",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS }, |
743 | { "std", OP_STD, 0, 0, OPF_DATA }, // special flag |
744 | { "cld", OP_CLD, 0, 0, OPF_DATA }, |
69a3cdfc |
745 | { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS }, |
746 | { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS }, |
747 | { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS }, |
748 | { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS }, |
749 | { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS }, |
750 | { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS }, |
751 | { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS }, |
752 | { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS }, |
753 | { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS }, |
d4e3b5db |
754 | { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS }, |
755 | { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS }, |
5101a5f9 |
756 | { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC }, |
69a3cdfc |
757 | { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC }, |
758 | { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS }, |
759 | { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS }, |
5101a5f9 |
760 | { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS }, |
761 | { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS }, |
69a3cdfc |
762 | { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS }, |
5101a5f9 |
763 | { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS }, |
764 | { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS }, |
69a3cdfc |
765 | { "test", OP_TEST, 2, 2, OPF_FLAGS }, |
766 | { "cmp", OP_CMP, 2, 2, OPF_FLAGS }, |
a3684be1 |
767 | { "retn", OP_RET, 0, 1, OPF_TAIL }, |
de50b98b |
768 | { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS }, |
69a3cdfc |
769 | { "jmp", OP_JMP, 1, 1, OPF_JMP }, |
5c024ef7 |
770 | { "jecxz",OP_JECXZ, 1, 1, OPF_JMP|OPF_CJMP }, |
771 | { "jo", OP_JO, 1, 1, OPF_CJMP_CC }, // 70 OF=1 |
772 | { "jno", OP_JNO, 1, 1, OPF_CJMP_CC }, // 71 OF=0 |
773 | { "jc", OP_JC, 1, 1, OPF_CJMP_CC }, // 72 CF=1 |
774 | { "jb", OP_JC, 1, 1, OPF_CJMP_CC }, // 72 |
775 | { "jnc", OP_JNC, 1, 1, OPF_CJMP_CC }, // 73 CF=0 |
776 | { "jnb", OP_JNC, 1, 1, OPF_CJMP_CC }, // 73 |
777 | { "jae", OP_JNC, 1, 1, OPF_CJMP_CC }, // 73 |
778 | { "jz", OP_JZ, 1, 1, OPF_CJMP_CC }, // 74 ZF=1 |
779 | { "je", OP_JZ, 1, 1, OPF_CJMP_CC }, // 74 |
780 | { "jnz", OP_JNZ, 1, 1, OPF_CJMP_CC }, // 75 ZF=0 |
781 | { "jne", OP_JNZ, 1, 1, OPF_CJMP_CC }, // 75 |
782 | { "jbe", OP_JBE, 1, 1, OPF_CJMP_CC }, // 76 CF=1 || ZF=1 |
783 | { "jna", OP_JBE, 1, 1, OPF_CJMP_CC }, // 76 |
784 | { "ja", OP_JA, 1, 1, OPF_CJMP_CC }, // 77 CF=0 && ZF=0 |
785 | { "jnbe", OP_JA, 1, 1, OPF_CJMP_CC }, // 77 |
786 | { "js", OP_JS, 1, 1, OPF_CJMP_CC }, // 78 SF=1 |
787 | { "jns", OP_JNS, 1, 1, OPF_CJMP_CC }, // 79 SF=0 |
788 | { "jp", OP_JP, 1, 1, OPF_CJMP_CC }, // 7a PF=1 |
789 | { "jpe", OP_JP, 1, 1, OPF_CJMP_CC }, // 7a |
790 | { "jnp", OP_JNP, 1, 1, OPF_CJMP_CC }, // 7b PF=0 |
791 | { "jpo", OP_JNP, 1, 1, OPF_CJMP_CC }, // 7b |
792 | { "jl", OP_JL, 1, 1, OPF_CJMP_CC }, // 7c SF!=OF |
793 | { "jnge", OP_JL, 1, 1, OPF_CJMP_CC }, // 7c |
794 | { "jge", OP_JGE, 1, 1, OPF_CJMP_CC }, // 7d SF=OF |
795 | { "jnl", OP_JGE, 1, 1, OPF_CJMP_CC }, // 7d |
796 | { "jle", OP_JLE, 1, 1, OPF_CJMP_CC }, // 7e ZF=1 || SF!=OF |
797 | { "jng", OP_JLE, 1, 1, OPF_CJMP_CC }, // 7e |
798 | { "jg", OP_JG, 1, 1, OPF_CJMP_CC }, // 7f ZF=0 && SF=OF |
799 | { "jnle", OP_JG, 1, 1, OPF_CJMP_CC }, // 7f |
5101a5f9 |
800 | { "seto", OP_JO, 1, 1, OPF_DATA|OPF_CC }, |
801 | { "setno", OP_JNO, 1, 1, OPF_DATA|OPF_CC }, |
802 | { "setc", OP_JC, 1, 1, OPF_DATA|OPF_CC }, |
803 | { "setb", OP_JC, 1, 1, OPF_DATA|OPF_CC }, |
804 | { "setnc", OP_JNC, 1, 1, OPF_DATA|OPF_CC }, |
805 | { "setae", OP_JNC, 1, 1, OPF_DATA|OPF_CC }, |
806 | { "setz", OP_JZ, 1, 1, OPF_DATA|OPF_CC }, |
807 | { "sete", OP_JZ, 1, 1, OPF_DATA|OPF_CC }, |
808 | { "setnz", OP_JNZ, 1, 1, OPF_DATA|OPF_CC }, |
809 | { "setne", OP_JNZ, 1, 1, OPF_DATA|OPF_CC }, |
810 | { "setbe", OP_JBE, 1, 1, OPF_DATA|OPF_CC }, |
811 | { "setna", OP_JBE, 1, 1, OPF_DATA|OPF_CC }, |
812 | { "seta", OP_JA, 1, 1, OPF_DATA|OPF_CC }, |
813 | { "setnbe", OP_JA, 1, 1, OPF_DATA|OPF_CC }, |
814 | { "sets", OP_JS, 1, 1, OPF_DATA|OPF_CC }, |
815 | { "setns", OP_JNS, 1, 1, OPF_DATA|OPF_CC }, |
816 | { "setp", OP_JP, 1, 1, OPF_DATA|OPF_CC }, |
817 | { "setpe", OP_JP, 1, 1, OPF_DATA|OPF_CC }, |
818 | { "setnp", OP_JNP, 1, 1, OPF_DATA|OPF_CC }, |
819 | { "setpo", OP_JNP, 1, 1, OPF_DATA|OPF_CC }, |
820 | { "setl", OP_JL, 1, 1, OPF_DATA|OPF_CC }, |
821 | { "setnge", OP_JL, 1, 1, OPF_DATA|OPF_CC }, |
822 | { "setge", OP_JGE, 1, 1, OPF_DATA|OPF_CC }, |
823 | { "setnl", OP_JGE, 1, 1, OPF_DATA|OPF_CC }, |
824 | { "setle", OP_JLE, 1, 1, OPF_DATA|OPF_CC }, |
825 | { "setng", OP_JLE, 1, 1, OPF_DATA|OPF_CC }, |
826 | { "setg", OP_JG, 1, 1, OPF_DATA|OPF_CC }, |
827 | { "setnle", OP_JG, 1, 1, OPF_DATA|OPF_CC }, |
69a3cdfc |
828 | }; |
c36e914d |
829 | |
830 | static void parse_op(struct parsed_op *op, char words[16][256], int wordc) |
831 | { |
bfa4a6ee |
832 | enum opr_lenmod lmod = OPLM_UNSPEC; |
33c35af6 |
833 | int prefix_flags = 0; |
69a3cdfc |
834 | int regmask_ind; |
835 | int regmask; |
33c35af6 |
836 | int op_w = 0; |
91977a1c |
837 | int opr = 0; |
33c35af6 |
838 | int w = 0; |
91977a1c |
839 | int i; |
c36e914d |
840 | |
33c35af6 |
841 | for (i = 0; i < ARRAY_SIZE(pref_table); i++) { |
842 | if (IS(words[w], pref_table[i].name)) { |
843 | prefix_flags = pref_table[i].flags; |
844 | break; |
845 | } |
846 | } |
847 | |
848 | if (prefix_flags) { |
849 | if (wordc <= 1) |
850 | aerr("lone prefix: '%s'\n", words[0]); |
851 | w++; |
852 | } |
853 | |
854 | op_w = w; |
91977a1c |
855 | for (i = 0; i < ARRAY_SIZE(op_table); i++) { |
33c35af6 |
856 | if (IS(words[w], op_table[i].name)) |
69a3cdfc |
857 | break; |
858 | } |
c36e914d |
859 | |
69a3cdfc |
860 | if (i == ARRAY_SIZE(op_table)) |
861 | aerr("unhandled op: '%s'\n", words[0]); |
33c35af6 |
862 | w++; |
c36e914d |
863 | |
69a3cdfc |
864 | op->op = op_table[i].op; |
33c35af6 |
865 | op->flags = op_table[i].flags | prefix_flags; |
69a3cdfc |
866 | op->regmask_src = op->regmask_dst = 0; |
867 | |
868 | for (opr = 0; opr < op_table[i].minopr; opr++) { |
869 | regmask = regmask_ind = 0; |
870 | w = parse_operand(&op->operand[opr], ®mask, ®mask_ind, |
871 | words, wordc, w, op->flags); |
872 | |
873 | if (opr == 0 && (op->flags & OPF_DATA)) |
874 | op->regmask_dst = regmask; |
875 | // for now, mark dst as src too |
876 | op->regmask_src |= regmask | regmask_ind; |
877 | } |
c36e914d |
878 | |
69a3cdfc |
879 | for (; w < wordc && opr < op_table[i].maxopr; opr++) { |
880 | w = parse_operand(&op->operand[opr], |
881 | &op->regmask_src, &op->regmask_src, |
882 | words, wordc, w, op->flags); |
91977a1c |
883 | } |
c36e914d |
884 | |
91977a1c |
885 | if (w < wordc) |
886 | aerr("parse_op %s incomplete: %d/%d\n", |
887 | words[0], w, wordc); |
5101a5f9 |
888 | |
889 | // special cases |
890 | op->operand_cnt = opr; |
891 | if (!strncmp(op_table[i].name, "set", 3)) |
892 | op->operand[0].lmod = OPLM_BYTE; |
893 | |
894 | // ops with implicit argumets |
895 | switch (op->op) { |
896 | case OP_CDQ: |
897 | op->operand_cnt = 2; |
898 | setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst); |
899 | setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src); |
900 | break; |
901 | |
33c35af6 |
902 | case OP_STOS: |
591721d7 |
903 | case OP_SCAS: |
33c35af6 |
904 | if (op->operand_cnt != 0) |
905 | break; |
591721d7 |
906 | if (words[op_w][4] == 'b') |
33c35af6 |
907 | lmod = OPLM_BYTE; |
591721d7 |
908 | else if (words[op_w][4] == 'w') |
33c35af6 |
909 | lmod = OPLM_WORD; |
591721d7 |
910 | else if (words[op_w][4] == 'd') |
33c35af6 |
911 | lmod = OPLM_DWORD; |
912 | op->operand_cnt = 3; |
d4e3b5db |
913 | setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src); |
914 | setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src); |
915 | op->regmask_dst = op->regmask_src; |
33c35af6 |
916 | setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD, &op->regmask_src); |
917 | break; |
918 | |
d4e3b5db |
919 | case OP_MOVS: |
7ba45c34 |
920 | case OP_CMPS: |
d4e3b5db |
921 | if (op->operand_cnt != 0) |
922 | break; |
7ba45c34 |
923 | if (words[op_w][4] == 'b') |
d4e3b5db |
924 | lmod = OPLM_BYTE; |
7ba45c34 |
925 | else if (words[op_w][4] == 'w') |
d4e3b5db |
926 | lmod = OPLM_WORD; |
7ba45c34 |
927 | else if (words[op_w][4] == 'd') |
d4e3b5db |
928 | lmod = OPLM_DWORD; |
929 | op->operand_cnt = 3; |
930 | setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src); |
931 | setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src); |
932 | setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src); |
933 | op->regmask_dst = op->regmask_src; |
934 | break; |
935 | |
5c024ef7 |
936 | case OP_JECXZ: |
937 | op->operand_cnt = 1; |
938 | op->regmask_src = 1 << xCX; |
939 | op->operand[0].type = OPT_REG; |
940 | op->operand[0].reg = xCX; |
941 | op->operand[0].lmod = OPLM_DWORD; |
942 | break; |
943 | |
5101a5f9 |
944 | case OP_IMUL: |
945 | if (op->operand_cnt != 1) |
946 | break; |
947 | // fallthrough |
948 | case OP_MUL: |
949 | // singleop mul |
950 | op->regmask_dst = (1 << xDX) | (1 << xAX); |
951 | op->regmask_src |= (1 << xAX); |
952 | if (op->operand[0].lmod == OPLM_UNSPEC) |
953 | op->operand[0].lmod = OPLM_DWORD; |
954 | break; |
955 | |
956 | case OP_DIV: |
957 | case OP_IDIV: |
958 | // we could set up operands for edx:eax, but there is no real need to |
959 | // (see is_opr_modified()) |
960 | regmask = (1 << xDX) | (1 << xAX); |
961 | op->regmask_dst = regmask; |
962 | op->regmask_src |= regmask; |
963 | if (op->operand[0].lmod == OPLM_UNSPEC) |
964 | op->operand[0].lmod = OPLM_DWORD; |
965 | break; |
966 | |
967 | case OP_SHL: |
968 | case OP_SHR: |
969 | case OP_SAR: |
d4e3b5db |
970 | case OP_ROL: |
971 | case OP_ROR: |
5101a5f9 |
972 | if (op->operand[1].lmod == OPLM_UNSPEC) |
973 | op->operand[1].lmod = OPLM_BYTE; |
974 | break; |
975 | |
7ba45c34 |
976 | case OP_PUSH: |
977 | if (op->operand[0].lmod == OPLM_UNSPEC |
978 | && (op->operand[0].type == OPT_CONST |
979 | || op->operand[0].type == OPT_OFFSET |
980 | || op->operand[0].type == OPT_LABEL)) |
981 | op->operand[0].lmod = OPLM_DWORD; |
982 | break; |
983 | |
3ebea2cf |
984 | // alignment |
985 | case OP_MOV: |
986 | if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG |
987 | && op->operand[0].reg == xDI && op->operand[1].reg == xDI) |
988 | { |
989 | op->flags |= OPF_RMD; |
990 | } |
991 | break; |
992 | |
de50b98b |
993 | case OP_LEA: |
994 | if (op->operand[0].type == OPT_REG |
995 | && op->operand[1].type == OPT_REGMEM) |
996 | { |
997 | char buf[16]; |
998 | snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name); |
999 | if (IS(buf, op->operand[1].name)) |
1000 | op->flags |= OPF_RMD; |
1001 | } |
1002 | break; |
1003 | |
89ff3147 |
1004 | case OP_CALL: |
1005 | // trashed regs must be explicitly detected later |
1006 | op->regmask_dst = 0; |
1007 | break; |
1008 | |
5101a5f9 |
1009 | default: |
1010 | break; |
1011 | } |
c36e914d |
1012 | } |
1013 | |
850c9265 |
1014 | static const char *op_name(enum op_op op) |
1015 | { |
1016 | int i; |
1017 | |
1018 | for (i = 0; i < ARRAY_SIZE(op_table); i++) |
1019 | if (op_table[i].op == op) |
1020 | return op_table[i].name; |
1021 | |
1022 | return "???"; |
1023 | } |
1024 | |
1025 | // debug |
1026 | static const char *dump_op(struct parsed_op *po) |
1027 | { |
1028 | static char out[128]; |
1029 | char *p = out; |
1030 | int i; |
1031 | |
4c45fa73 |
1032 | if (po == NULL) |
1033 | return "???"; |
1034 | |
850c9265 |
1035 | snprintf(out, sizeof(out), "%s", op_name(po->op)); |
1036 | for (i = 0; i < po->operand_cnt; i++) { |
1037 | p += strlen(p); |
1038 | if (i > 0) |
1039 | *p++ = ','; |
1040 | snprintf(p, sizeof(out) - (p - out), |
1041 | po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s", |
1042 | po->operand[i].name); |
1043 | } |
1044 | |
1045 | return out; |
1046 | } |
1047 | |
4c45fa73 |
1048 | static const char *lmod_type_u(struct parsed_op *po, |
1049 | enum opr_lenmod lmod) |
1050 | { |
1051 | switch (lmod) { |
1052 | case OPLM_DWORD: |
1053 | return "u32"; |
1054 | case OPLM_WORD: |
1055 | return "u16"; |
1056 | case OPLM_BYTE: |
1057 | return "u8"; |
1058 | default: |
1059 | ferr(po, "invalid lmod: %d\n", lmod); |
1060 | return "(_invalid_)"; |
1061 | } |
1062 | } |
1063 | |
d4e3b5db |
1064 | static const char *lmod_cast_u(struct parsed_op *po, |
1065 | enum opr_lenmod lmod) |
1066 | { |
1067 | switch (lmod) { |
1068 | case OPLM_DWORD: |
1069 | return ""; |
1070 | case OPLM_WORD: |
1071 | return "(u16)"; |
1072 | case OPLM_BYTE: |
1073 | return "(u8)"; |
1074 | default: |
1075 | ferr(po, "invalid lmod: %d\n", lmod); |
1076 | return "(_invalid_)"; |
1077 | } |
1078 | } |
1079 | |
1080 | static const char *lmod_cast_u_ptr(struct parsed_op *po, |
1081 | enum opr_lenmod lmod) |
1082 | { |
1083 | switch (lmod) { |
1084 | case OPLM_DWORD: |
1085 | return "*(u32 *)"; |
1086 | case OPLM_WORD: |
1087 | return "*(u16 *)"; |
1088 | case OPLM_BYTE: |
1089 | return "*(u8 *)"; |
1090 | default: |
1091 | ferr(po, "invalid lmod: %d\n", lmod); |
1092 | return "(_invalid_)"; |
1093 | } |
1094 | } |
1095 | |
1096 | static const char *lmod_cast_s(struct parsed_op *po, |
1097 | enum opr_lenmod lmod) |
1098 | { |
1099 | switch (lmod) { |
1100 | case OPLM_DWORD: |
1101 | return "(s32)"; |
1102 | case OPLM_WORD: |
1103 | return "(s16)"; |
1104 | case OPLM_BYTE: |
1105 | return "(s8)"; |
1106 | default: |
1107 | ferr(po, "%s: invalid lmod: %d\n", __func__, lmod); |
1108 | return "(_invalid_)"; |
1109 | } |
1110 | } |
1111 | |
1112 | static const char *lmod_cast(struct parsed_op *po, |
1113 | enum opr_lenmod lmod, int is_signed) |
1114 | { |
1115 | return is_signed ? |
1116 | lmod_cast_s(po, lmod) : |
1117 | lmod_cast_u(po, lmod); |
1118 | } |
1119 | |
1120 | static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod) |
1121 | { |
1122 | switch (lmod) { |
1123 | case OPLM_DWORD: |
1124 | return 4; |
1125 | case OPLM_WORD: |
1126 | return 2; |
1127 | case OPLM_BYTE: |
1128 | return 1; |
1129 | default: |
1130 | ferr(po, "%s: invalid lmod: %d\n", __func__, lmod); |
1131 | return 0; |
1132 | } |
1133 | } |
1134 | |
91977a1c |
1135 | static const char *opr_name(struct parsed_op *po, int opr_num) |
c36e914d |
1136 | { |
91977a1c |
1137 | if (opr_num >= po->operand_cnt) |
1138 | ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt); |
1139 | return po->operand[opr_num].name; |
c36e914d |
1140 | } |
1141 | |
91977a1c |
1142 | static unsigned int opr_const(struct parsed_op *po, int opr_num) |
c36e914d |
1143 | { |
91977a1c |
1144 | if (opr_num >= po->operand_cnt) |
1145 | ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt); |
1146 | if (po->operand[opr_num].type != OPT_CONST) |
1147 | ferr(po, "opr %d: const expected\n", opr_num); |
1148 | return po->operand[opr_num].val; |
1149 | } |
c36e914d |
1150 | |
91977a1c |
1151 | static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr) |
1152 | { |
1153 | if ((unsigned int)popr->reg >= MAX_REGS) |
1154 | ferr(po, "invalid reg: %d\n", popr->reg); |
1155 | return regs_r32[popr->reg]; |
1156 | } |
c36e914d |
1157 | |
a2c1d768 |
1158 | // cast1 is the "final" cast |
1159 | static const char *simplify_cast(const char *cast1, const char *cast2) |
1160 | { |
1161 | static char buf[256]; |
1162 | |
1163 | if (cast1[0] == 0) |
1164 | return cast2; |
1165 | if (cast2[0] == 0) |
1166 | return cast1; |
1167 | if (IS(cast1, cast2)) |
1168 | return cast1; |
1169 | if (IS(cast1, "(s8)") && IS(cast2, "(u8)")) |
1170 | return cast1; |
1171 | if (IS(cast1, "(s16)") && IS(cast2, "(u16)")) |
1172 | return cast1; |
1173 | if (IS(cast1, "(u8)") && IS_START(cast2, "*(u8 *)")) |
1174 | return cast2; |
1175 | if (IS(cast1, "(u16)") && IS_START(cast2, "*(u16 *)")) |
1176 | return cast2; |
1cd4a663 |
1177 | if (strchr(cast1, '*') && IS_START(cast2, "(u32)")) |
1178 | return cast1; |
a2c1d768 |
1179 | |
1180 | snprintf(buf, sizeof(buf), "%s%s", cast1, cast2); |
1181 | return buf; |
1182 | } |
1183 | |
1184 | static const char *simplify_cast_num(const char *cast, unsigned int val) |
1185 | { |
1186 | if (IS(cast, "(u8)") && val < 0x100) |
1187 | return ""; |
1188 | if (IS(cast, "(s8)") && val < 0x80) |
1189 | return ""; |
1190 | if (IS(cast, "(u16)") && val < 0x10000) |
1191 | return ""; |
1192 | if (IS(cast, "(s16)") && val < 0x8000) |
1193 | return ""; |
1194 | if (IS(cast, "(s32)") && val < 0x80000000) |
1195 | return ""; |
1196 | |
1197 | return cast; |
1198 | } |
1199 | |
39b168b8 |
1200 | static struct parsed_equ *equ_find(struct parsed_op *po, const char *name, |
1201 | int *extra_offs) |
91977a1c |
1202 | { |
39b168b8 |
1203 | const char *p; |
1204 | char *endp; |
1205 | int namelen; |
850c9265 |
1206 | int i; |
1207 | |
39b168b8 |
1208 | *extra_offs = 0; |
1209 | namelen = strlen(name); |
1210 | |
1211 | p = strchr(name, '+'); |
1212 | if (p != NULL) { |
1213 | namelen = p - name; |
1214 | if (namelen <= 0) |
1215 | ferr(po, "equ parse failed for '%s'\n", name); |
1216 | |
1217 | if (IS_START(p, "0x")) |
1218 | p += 2; |
1219 | *extra_offs = strtol(p, &endp, 16); |
1220 | if (*endp != 0) |
1221 | ferr(po, "equ parse failed for '%s'\n", name); |
1222 | } |
1223 | |
850c9265 |
1224 | for (i = 0; i < g_eqcnt; i++) |
39b168b8 |
1225 | if (strncmp(g_eqs[i].name, name, namelen) == 0 |
1226 | && g_eqs[i].name[namelen] == 0) |
850c9265 |
1227 | break; |
87bf6cec |
1228 | if (i >= g_eqcnt) { |
1229 | if (po != NULL) |
1230 | ferr(po, "unresolved equ name: '%s'\n", name); |
1231 | return NULL; |
1232 | } |
850c9265 |
1233 | |
1234 | return &g_eqs[i]; |
1235 | } |
1236 | |
1cd4a663 |
1237 | static int is_stack_access(struct parsed_op *po, |
1238 | const struct parsed_opr *popr) |
850c9265 |
1239 | { |
1cd4a663 |
1240 | return (parse_stack_el(popr->name, NULL) |
1241 | || (g_bp_frame && !(po->flags & OPF_EBP_S) |
1242 | && IS_START(popr->name, "ebp"))); |
1243 | } |
1244 | |
1245 | static void parse_stack_access(struct parsed_op *po, |
1246 | const char *name, char *ofs_reg, int *offset_out, |
1247 | int *stack_ra_out, const char **bp_arg_out) |
1248 | { |
1249 | const char *bp_arg = ""; |
1250 | const char *p = NULL; |
91977a1c |
1251 | struct parsed_equ *eq; |
4f12f671 |
1252 | char *endp = NULL; |
d4e3b5db |
1253 | int stack_ra = 0; |
39b168b8 |
1254 | int offset = 0; |
91977a1c |
1255 | |
1cd4a663 |
1256 | ofs_reg[0] = 0; |
a3684be1 |
1257 | |
1258 | if (IS_START(name, "ebp-") |
1259 | || (IS_START(name, "ebp+") && '0' <= name[4] && name[4] <= '9')) |
1260 | { |
1261 | p = name + 4; |
1262 | if (IS_START(p, "0x")) |
1263 | p += 2; |
1264 | offset = strtoul(p, &endp, 16); |
1265 | if (name[3] == '-') |
1266 | offset = -offset; |
1267 | if (*endp != 0) |
1268 | ferr(po, "ebp- parse of '%s' failed\n", name); |
1269 | } |
1270 | else { |
4f12f671 |
1271 | bp_arg = parse_stack_el(name, ofs_reg); |
1272 | snprintf(g_comment, sizeof(g_comment), "%s", bp_arg); |
1273 | eq = equ_find(po, bp_arg, &offset); |
1274 | if (eq == NULL) |
1275 | ferr(po, "detected but missing eq\n"); |
1276 | offset += eq->offset; |
1277 | } |
39b168b8 |
1278 | |
d4e3b5db |
1279 | if (!strncmp(name, "ebp", 3)) |
1280 | stack_ra = 4; |
1281 | |
94d447fb |
1282 | if (ofs_reg[0] == 0 && stack_ra <= offset && offset < stack_ra + 4) |
39b168b8 |
1283 | ferr(po, "reference to ra? %d %d\n", offset, stack_ra); |
d4e3b5db |
1284 | |
1cd4a663 |
1285 | *offset_out = offset; |
1286 | *stack_ra_out = stack_ra; |
1287 | if (bp_arg_out) |
1288 | *bp_arg_out = bp_arg; |
1289 | } |
1290 | |
1291 | static void stack_frame_access(struct parsed_op *po, |
1292 | struct parsed_opr *popr, char *buf, size_t buf_size, |
1293 | const char *name, const char *cast, int is_src, int is_lea) |
1294 | { |
1295 | enum opr_lenmod tmp_lmod = OPLM_UNSPEC; |
1296 | const char *prefix = ""; |
1297 | const char *bp_arg = NULL; |
1298 | char ofs_reg[16] = { 0, }; |
1299 | int i, arg_i, arg_s; |
1300 | int unaligned = 0; |
1301 | int stack_ra = 0; |
1302 | int offset = 0; |
1303 | int sf_ofs; |
1304 | int lim; |
1305 | |
1306 | if (po->flags & OPF_EBP_S) |
1307 | ferr(po, "stack_frame_access while ebp is scratch\n"); |
1308 | |
1309 | parse_stack_access(po, name, ofs_reg, &offset, &stack_ra, &bp_arg); |
1310 | |
4f12f671 |
1311 | if (offset > stack_ra) |
1312 | { |
39b168b8 |
1313 | arg_i = (offset - stack_ra - 4) / 4; |
bd96f656 |
1314 | if (arg_i < 0 || arg_i >= g_func_pp->argc_stack) |
4f12f671 |
1315 | { |
bd96f656 |
1316 | if (g_func_pp->is_vararg |
1317 | && arg_i == g_func_pp->argc_stack && is_lea) |
1318 | { |
4f12f671 |
1319 | // should be va_list |
1320 | if (cast[0] == 0) |
1321 | cast = "(u32)"; |
1322 | snprintf(buf, buf_size, "%sap", cast); |
1323 | return; |
1324 | } |
1325 | ferr(po, "offset %d (%s,%d) doesn't map to any arg\n", |
1326 | offset, bp_arg, arg_i); |
1327 | } |
4c45fa73 |
1328 | if (ofs_reg[0] != 0) |
7ba45c34 |
1329 | ferr(po, "offset reg on arg access?\n"); |
91977a1c |
1330 | |
bd96f656 |
1331 | for (i = arg_s = 0; i < g_func_pp->argc; i++) { |
1332 | if (g_func_pp->arg[i].reg != NULL) |
91977a1c |
1333 | continue; |
1334 | if (arg_s == arg_i) |
1335 | break; |
1336 | arg_s++; |
1337 | } |
bd96f656 |
1338 | if (i == g_func_pp->argc) |
91977a1c |
1339 | ferr(po, "arg %d not in prototype?\n", arg_i); |
850c9265 |
1340 | |
bd96f656 |
1341 | popr->is_ptr = g_func_pp->arg[i].type.is_ptr; |
de50b98b |
1342 | |
1343 | switch (popr->lmod) |
7ba45c34 |
1344 | { |
1345 | case OPLM_BYTE: |
4f12f671 |
1346 | if (is_lea) |
1347 | ferr(po, "lea/byte to arg?\n"); |
7ba45c34 |
1348 | if (is_src && (offset & 3) == 0) |
a2c1d768 |
1349 | snprintf(buf, buf_size, "%sa%d", |
1350 | simplify_cast(cast, "(u8)"), i + 1); |
7ba45c34 |
1351 | else |
a2c1d768 |
1352 | snprintf(buf, buf_size, "%sBYTE%d(a%d)", |
1353 | cast, offset & 3, i + 1); |
7ba45c34 |
1354 | break; |
1355 | |
1356 | case OPLM_WORD: |
4f12f671 |
1357 | if (is_lea) |
1358 | ferr(po, "lea/word to arg?\n"); |
a3684be1 |
1359 | if (offset & 1) { |
1360 | unaligned = 1; |
1361 | if (!is_src) { |
1362 | if (offset & 2) |
1363 | ferr(po, "problematic arg store\n"); |
a2c1d768 |
1364 | snprintf(buf, buf_size, "%s((char *)&a%d + 1)", |
1365 | simplify_cast(cast, "*(u16 *)"), i + 1); |
a3684be1 |
1366 | } |
1367 | else |
1368 | ferr(po, "unaligned arg word load\n"); |
1369 | } |
1370 | else if (is_src && (offset & 2) == 0) |
a2c1d768 |
1371 | snprintf(buf, buf_size, "%sa%d", |
1372 | simplify_cast(cast, "(u16)"), i + 1); |
7ba45c34 |
1373 | else |
a2c1d768 |
1374 | snprintf(buf, buf_size, "%s%sWORD(a%d)", |
1375 | cast, (offset & 2) ? "HI" : "LO", i + 1); |
7ba45c34 |
1376 | break; |
1377 | |
1378 | case OPLM_DWORD: |
3ebea2cf |
1379 | if (cast[0]) |
1380 | prefix = cast; |
1381 | else if (is_src) |
1382 | prefix = "(u32)"; |
a3684be1 |
1383 | |
de50b98b |
1384 | if (offset & 3) { |
a3684be1 |
1385 | unaligned = 1; |
2b43685d |
1386 | if (is_lea) |
1387 | snprintf(buf, buf_size, "(u32)&a%d + %d", |
1388 | i + 1, offset & 3); |
a3684be1 |
1389 | else if (!is_src) |
1390 | ferr(po, "unaligned arg store\n"); |
1391 | else { |
1392 | // mov edx, [ebp+arg_4+2]; movsx ecx, dx |
2b43685d |
1393 | snprintf(buf, buf_size, "%s(a%d >> %d)", |
1394 | prefix, i + 1, (offset & 3) * 8); |
a3684be1 |
1395 | } |
de50b98b |
1396 | } |
1397 | else { |
1398 | snprintf(buf, buf_size, "%s%sa%d", |
1399 | prefix, is_lea ? "&" : "", i + 1); |
1400 | } |
7ba45c34 |
1401 | break; |
1402 | |
1403 | default: |
de50b98b |
1404 | ferr(po, "bp_arg bad lmod: %d\n", popr->lmod); |
7ba45c34 |
1405 | } |
1406 | |
a3684be1 |
1407 | if (unaligned) |
1408 | snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg); |
1409 | |
7ba45c34 |
1410 | // common problem |
bd96f656 |
1411 | guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type); |
a2c1d768 |
1412 | if (tmp_lmod != OPLM_DWORD |
1413 | && (unaligned || (!is_src && tmp_lmod < popr->lmod))) |
1414 | { |
1415 | ferr(po, "bp_arg arg%d/w offset %d and type '%s' is too small\n", |
1416 | i + 1, offset, g_func_pp->arg[i].type.name); |
1417 | } |
46411e6c |
1418 | if (popr->is_ptr && popr->lmod != OPLM_DWORD) |
1419 | ferr(po, "bp_arg arg%d: non-dword ptr access\n", i + 1); |
91977a1c |
1420 | } |
4f12f671 |
1421 | else |
1422 | { |
1bafb621 |
1423 | if (g_stack_fsz == 0) |
1424 | ferr(po, "stack var access without stackframe\n"); |
a2c1d768 |
1425 | g_stack_frame_used = 1; |
850c9265 |
1426 | |
39b168b8 |
1427 | sf_ofs = g_stack_fsz + offset; |
de50b98b |
1428 | lim = (ofs_reg[0] != 0) ? -4 : 0; |
1429 | if (offset > 0 || sf_ofs < lim) |
39b168b8 |
1430 | ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz); |
850c9265 |
1431 | |
1432 | if (is_lea) |
33c35af6 |
1433 | prefix = "(u32)&"; |
3ebea2cf |
1434 | else |
1435 | prefix = cast; |
850c9265 |
1436 | |
de50b98b |
1437 | switch (popr->lmod) |
850c9265 |
1438 | { |
1439 | case OPLM_BYTE: |
7ba45c34 |
1440 | snprintf(buf, buf_size, "%ssf.b[%d%s%s]", |
1441 | prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg); |
850c9265 |
1442 | break; |
7ba45c34 |
1443 | |
850c9265 |
1444 | case OPLM_WORD: |
7ba45c34 |
1445 | if ((sf_ofs & 1) || ofs_reg[0] != 0) { |
1446 | // known unaligned or possibly unaligned |
1447 | strcat(g_comment, " unaligned"); |
1448 | if (prefix[0] == 0) |
1449 | prefix = "*(u16 *)&"; |
1450 | snprintf(buf, buf_size, "%ssf.b[%d%s%s]", |
1451 | prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg); |
1452 | break; |
1453 | } |
1454 | snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2); |
850c9265 |
1455 | break; |
7ba45c34 |
1456 | |
850c9265 |
1457 | case OPLM_DWORD: |
7ba45c34 |
1458 | if ((sf_ofs & 3) || ofs_reg[0] != 0) { |
1459 | // known unaligned or possibly unaligned |
1460 | strcat(g_comment, " unaligned"); |
1461 | if (prefix[0] == 0) |
1462 | prefix = "*(u32 *)&"; |
1463 | snprintf(buf, buf_size, "%ssf.b[%d%s%s]", |
1464 | prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg); |
1465 | break; |
1466 | } |
1467 | snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4); |
850c9265 |
1468 | break; |
7ba45c34 |
1469 | |
850c9265 |
1470 | default: |
de50b98b |
1471 | ferr(po, "bp_stack bad lmod: %d\n", popr->lmod); |
850c9265 |
1472 | } |
91977a1c |
1473 | } |
1474 | } |
c36e914d |
1475 | |
89ff3147 |
1476 | static void check_func_pp(struct parsed_op *po, |
1477 | const struct parsed_proto *pp, const char *pfx) |
1478 | { |
b74c31e3 |
1479 | char buf[256]; |
1480 | |
89ff3147 |
1481 | if (pp->argc_reg != 0) { |
b74c31e3 |
1482 | if (/*!g_allow_regfunc &&*/ !pp->is_fastcall) { |
1483 | pp_print(buf, sizeof(buf), pp); |
1484 | ferr(po, "%s: unexpected reg arg in icall: %s\n", pfx, buf); |
1485 | } |
89ff3147 |
1486 | if (pp->argc_stack > 0 && pp->argc_reg != 2) |
1487 | ferr(po, "%s: %d reg arg(s) with %d stack arg(s)\n", |
1488 | pfx, pp->argc_reg, pp->argc_stack); |
1489 | } |
1490 | } |
1491 | |
3ebea2cf |
1492 | static void check_label_read_ref(struct parsed_op *po, const char *name) |
1493 | { |
840257f6 |
1494 | const struct parsed_proto *pp; |
1495 | |
36595fd2 |
1496 | pp = proto_parse(g_fhdr, name, 0); |
840257f6 |
1497 | if (pp == NULL) |
1498 | ferr(po, "proto_parse failed for ref '%s'\n", name); |
1499 | |
89ff3147 |
1500 | if (pp->is_func) |
1501 | check_func_pp(po, pp, "ref"); |
3ebea2cf |
1502 | } |
1503 | |
91977a1c |
1504 | static char *out_src_opr(char *buf, size_t buf_size, |
591721d7 |
1505 | struct parsed_op *po, struct parsed_opr *popr, const char *cast, |
3ebea2cf |
1506 | int is_lea) |
91977a1c |
1507 | { |
850c9265 |
1508 | char tmp1[256], tmp2[256]; |
1509 | char expr[256]; |
a2c1d768 |
1510 | char *p; |
850c9265 |
1511 | int ret; |
1512 | |
3ebea2cf |
1513 | if (cast == NULL) |
1514 | cast = ""; |
1515 | |
91977a1c |
1516 | switch (popr->type) { |
1517 | case OPT_REG: |
850c9265 |
1518 | if (is_lea) |
1519 | ferr(po, "lea from reg?\n"); |
1520 | |
91977a1c |
1521 | switch (popr->lmod) { |
1522 | case OPLM_DWORD: |
3ebea2cf |
1523 | snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr)); |
91977a1c |
1524 | break; |
850c9265 |
1525 | case OPLM_WORD: |
a2c1d768 |
1526 | snprintf(buf, buf_size, "%s%s", |
1527 | simplify_cast(cast, "(u16)"), opr_reg_p(po, popr)); |
850c9265 |
1528 | break; |
1529 | case OPLM_BYTE: |
5101a5f9 |
1530 | if (popr->name[1] == 'h') // XXX.. |
a2c1d768 |
1531 | snprintf(buf, buf_size, "%s(%s >> 8)", |
1532 | simplify_cast(cast, "(u8)"), opr_reg_p(po, popr)); |
5101a5f9 |
1533 | else |
a2c1d768 |
1534 | snprintf(buf, buf_size, "%s%s", |
1535 | simplify_cast(cast, "(u8)"), opr_reg_p(po, popr)); |
850c9265 |
1536 | break; |
91977a1c |
1537 | default: |
1538 | ferr(po, "invalid src lmod: %d\n", popr->lmod); |
1539 | } |
1540 | break; |
850c9265 |
1541 | |
91977a1c |
1542 | case OPT_REGMEM: |
1cd4a663 |
1543 | if (is_stack_access(po, popr)) { |
de50b98b |
1544 | stack_frame_access(po, popr, buf, buf_size, |
3ebea2cf |
1545 | popr->name, cast, 1, is_lea); |
91977a1c |
1546 | break; |
1547 | } |
850c9265 |
1548 | |
1549 | strcpy(expr, popr->name); |
1550 | if (strchr(expr, '[')) { |
1551 | // special case: '[' can only be left for label[reg] form |
1552 | ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2); |
1553 | if (ret != 2) |
1554 | ferr(po, "parse failure for '%s'\n", expr); |
a2c1d768 |
1555 | if (tmp1[0] == '(') { |
1556 | // (off_4FFF50+3)[eax] |
1557 | p = strchr(tmp1 + 1, ')'); |
1558 | if (p == NULL || p[1] != 0) |
1559 | ferr(po, "parse failure (2) for '%s'\n", expr); |
1560 | *p = 0; |
1561 | memmove(tmp1, tmp1 + 1, strlen(tmp1)); |
1562 | } |
33c35af6 |
1563 | snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2); |
850c9265 |
1564 | } |
1565 | |
1566 | // XXX: do we need more parsing? |
1567 | if (is_lea) { |
1568 | snprintf(buf, buf_size, "%s", expr); |
1569 | break; |
1570 | } |
1571 | |
a2c1d768 |
1572 | snprintf(buf, buf_size, "%s(%s)", |
1573 | simplify_cast(cast, lmod_cast_u_ptr(po, popr->lmod)), expr); |
91977a1c |
1574 | break; |
850c9265 |
1575 | |
91977a1c |
1576 | case OPT_LABEL: |
3ebea2cf |
1577 | check_label_read_ref(po, popr->name); |
1578 | if (cast[0] == 0 && popr->is_ptr) |
1579 | cast = "(u32)"; |
2b43685d |
1580 | |
850c9265 |
1581 | if (is_lea) |
1582 | snprintf(buf, buf_size, "(u32)&%s", popr->name); |
2b43685d |
1583 | else if (popr->size_lt) |
1584 | snprintf(buf, buf_size, "%s%s%s%s", cast, |
1585 | lmod_cast_u_ptr(po, popr->lmod), |
1586 | popr->is_array ? "" : "&", |
1587 | popr->name); |
850c9265 |
1588 | else |
3ebea2cf |
1589 | snprintf(buf, buf_size, "%s%s%s", cast, popr->name, |
7ba45c34 |
1590 | popr->is_array ? "[0]" : ""); |
850c9265 |
1591 | break; |
1592 | |
1593 | case OPT_OFFSET: |
3ebea2cf |
1594 | check_label_read_ref(po, popr->name); |
1595 | if (cast[0] == 0) |
1596 | cast = "(u32)"; |
850c9265 |
1597 | if (is_lea) |
1598 | ferr(po, "lea an offset?\n"); |
3ebea2cf |
1599 | snprintf(buf, buf_size, "%s&%s", cast, popr->name); |
91977a1c |
1600 | break; |
850c9265 |
1601 | |
91977a1c |
1602 | case OPT_CONST: |
850c9265 |
1603 | if (is_lea) |
1604 | ferr(po, "lea from const?\n"); |
1605 | |
a2c1d768 |
1606 | printf_number(tmp1, sizeof(tmp1), popr->val); |
ddaf8bd7 |
1607 | if (popr->val == 0 && strchr(cast, '*')) |
1608 | snprintf(buf, buf_size, "NULL"); |
1609 | else |
1610 | snprintf(buf, buf_size, "%s%s", |
1611 | simplify_cast_num(cast, popr->val), tmp1); |
91977a1c |
1612 | break; |
850c9265 |
1613 | |
91977a1c |
1614 | default: |
1615 | ferr(po, "invalid src type: %d\n", popr->type); |
1616 | } |
1617 | |
1618 | return buf; |
1619 | } |
c36e914d |
1620 | |
de50b98b |
1621 | // note: may set is_ptr (we find that out late for ebp frame..) |
91977a1c |
1622 | static char *out_dst_opr(char *buf, size_t buf_size, |
1623 | struct parsed_op *po, struct parsed_opr *popr) |
1624 | { |
1625 | switch (popr->type) { |
1626 | case OPT_REG: |
1627 | switch (popr->lmod) { |
1628 | case OPLM_DWORD: |
1629 | snprintf(buf, buf_size, "%s", opr_reg_p(po, popr)); |
1630 | break; |
850c9265 |
1631 | case OPLM_WORD: |
1632 | // ugh.. |
1633 | snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr)); |
1634 | break; |
1635 | case OPLM_BYTE: |
1636 | // ugh.. |
5101a5f9 |
1637 | if (popr->name[1] == 'h') // XXX.. |
1638 | snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr)); |
1639 | else |
1640 | snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr)); |
850c9265 |
1641 | break; |
91977a1c |
1642 | default: |
1643 | ferr(po, "invalid dst lmod: %d\n", popr->lmod); |
1644 | } |
1645 | break; |
850c9265 |
1646 | |
1647 | case OPT_REGMEM: |
1cd4a663 |
1648 | if (is_stack_access(po, popr)) { |
de50b98b |
1649 | stack_frame_access(po, popr, buf, buf_size, |
3ebea2cf |
1650 | popr->name, "", 0, 0); |
850c9265 |
1651 | break; |
1652 | } |
1653 | |
3ebea2cf |
1654 | return out_src_opr(buf, buf_size, po, popr, NULL, 0); |
850c9265 |
1655 | |
bfa4a6ee |
1656 | case OPT_LABEL: |
2b43685d |
1657 | if (popr->size_mismatch) |
1658 | snprintf(buf, buf_size, "%s%s%s", |
1659 | lmod_cast_u_ptr(po, popr->lmod), |
1660 | popr->is_array ? "" : "&", popr->name); |
1661 | else |
1662 | snprintf(buf, buf_size, "%s%s", popr->name, |
1663 | popr->is_array ? "[0]" : ""); |
bfa4a6ee |
1664 | break; |
1665 | |
91977a1c |
1666 | default: |
1667 | ferr(po, "invalid dst type: %d\n", popr->type); |
1668 | } |
1669 | |
1670 | return buf; |
1671 | } |
c36e914d |
1672 | |
3ebea2cf |
1673 | static char *out_src_opr_u32(char *buf, size_t buf_size, |
1674 | struct parsed_op *po, struct parsed_opr *popr) |
1675 | { |
1676 | return out_src_opr(buf, buf_size, po, popr, NULL, 0); |
1677 | } |
1678 | |
69a3cdfc |
1679 | static enum parsed_flag_op split_cond(struct parsed_op *po, |
940e8e66 |
1680 | enum op_op op, int *is_inv) |
91977a1c |
1681 | { |
940e8e66 |
1682 | *is_inv = 0; |
91977a1c |
1683 | |
69a3cdfc |
1684 | switch (op) { |
1685 | case OP_JO: |
1686 | return PFO_O; |
1687 | case OP_JC: |
1688 | return PFO_C; |
1689 | case OP_JZ: |
1690 | return PFO_Z; |
1691 | case OP_JBE: |
1692 | return PFO_BE; |
1693 | case OP_JS: |
1694 | return PFO_S; |
1695 | case OP_JP: |
1696 | return PFO_P; |
1697 | case OP_JL: |
1698 | return PFO_L; |
1699 | case OP_JLE: |
1700 | return PFO_LE; |
1701 | |
91977a1c |
1702 | case OP_JNO: |
940e8e66 |
1703 | *is_inv = 1; |
69a3cdfc |
1704 | return PFO_O; |
91977a1c |
1705 | case OP_JNC: |
940e8e66 |
1706 | *is_inv = 1; |
69a3cdfc |
1707 | return PFO_C; |
91977a1c |
1708 | case OP_JNZ: |
940e8e66 |
1709 | *is_inv = 1; |
69a3cdfc |
1710 | return PFO_Z; |
1711 | case OP_JA: |
940e8e66 |
1712 | *is_inv = 1; |
69a3cdfc |
1713 | return PFO_BE; |
91977a1c |
1714 | case OP_JNS: |
940e8e66 |
1715 | *is_inv = 1; |
69a3cdfc |
1716 | return PFO_S; |
91977a1c |
1717 | case OP_JNP: |
940e8e66 |
1718 | *is_inv = 1; |
69a3cdfc |
1719 | return PFO_P; |
850c9265 |
1720 | case OP_JGE: |
940e8e66 |
1721 | *is_inv = 1; |
69a3cdfc |
1722 | return PFO_L; |
91977a1c |
1723 | case OP_JG: |
940e8e66 |
1724 | *is_inv = 1; |
69a3cdfc |
1725 | return PFO_LE; |
1726 | |
1727 | case OP_ADC: |
1728 | case OP_SBB: |
1729 | return PFO_C; |
1730 | |
91977a1c |
1731 | default: |
69a3cdfc |
1732 | ferr(po, "split_cond: bad op %d\n", op); |
1733 | return -1; |
91977a1c |
1734 | } |
1735 | } |
c36e914d |
1736 | |
91977a1c |
1737 | static void out_test_for_cc(char *buf, size_t buf_size, |
940e8e66 |
1738 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv, |
69a3cdfc |
1739 | enum opr_lenmod lmod, const char *expr) |
91977a1c |
1740 | { |
69a3cdfc |
1741 | const char *cast, *scast; |
91977a1c |
1742 | |
69a3cdfc |
1743 | cast = lmod_cast_u(po, lmod); |
1744 | scast = lmod_cast_s(po, lmod); |
1745 | |
1746 | switch (pfo) { |
1747 | case PFO_Z: |
87bf6cec |
1748 | case PFO_BE: // CF=1||ZF=1; CF=0 |
850c9265 |
1749 | snprintf(buf, buf_size, "(%s%s %s 0)", |
940e8e66 |
1750 | cast, expr, is_inv ? "!=" : "=="); |
91977a1c |
1751 | break; |
850c9265 |
1752 | |
5101a5f9 |
1753 | case PFO_S: |
1754 | case PFO_L: // SF!=OF; OF=0 |
1755 | snprintf(buf, buf_size, "(%s%s %s 0)", |
940e8e66 |
1756 | scast, expr, is_inv ? ">=" : "<"); |
5101a5f9 |
1757 | break; |
1758 | |
87bf6cec |
1759 | case PFO_LE: // ZF=1||SF!=OF; OF=0 |
69a3cdfc |
1760 | snprintf(buf, buf_size, "(%s%s %s 0)", |
940e8e66 |
1761 | scast, expr, is_inv ? ">" : "<="); |
850c9265 |
1762 | break; |
1763 | |
91977a1c |
1764 | default: |
69a3cdfc |
1765 | ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo); |
91977a1c |
1766 | } |
1767 | } |
c36e914d |
1768 | |
850c9265 |
1769 | static void out_cmp_for_cc(char *buf, size_t buf_size, |
a2c1d768 |
1770 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv) |
850c9265 |
1771 | { |
a2c1d768 |
1772 | const char *cast, *scast, *cast_use; |
1773 | char buf1[256], buf2[256]; |
1774 | enum opr_lenmod lmod; |
1775 | |
1776 | if (po->operand[0].lmod != po->operand[1].lmod) |
1777 | ferr(po, "%s: lmod mismatch: %d %d\n", __func__, |
1778 | po->operand[0].lmod, po->operand[1].lmod); |
1779 | lmod = po->operand[0].lmod; |
850c9265 |
1780 | |
69a3cdfc |
1781 | cast = lmod_cast_u(po, lmod); |
1782 | scast = lmod_cast_s(po, lmod); |
850c9265 |
1783 | |
a2c1d768 |
1784 | switch (pfo) { |
1785 | case PFO_C: |
1786 | case PFO_Z: |
1787 | case PFO_BE: // !a |
1788 | cast_use = cast; |
1789 | break; |
1790 | |
1791 | case PFO_S: |
1792 | case PFO_L: // !ge |
1793 | case PFO_LE: |
1794 | cast_use = scast; |
1795 | break; |
1796 | |
1797 | default: |
1798 | ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo); |
1799 | } |
1800 | |
1801 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], cast_use, 0); |
1802 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], cast_use, 0); |
1803 | |
69a3cdfc |
1804 | switch (pfo) { |
5101a5f9 |
1805 | case PFO_C: |
1806 | // note: must be unsigned compare |
a2c1d768 |
1807 | snprintf(buf, buf_size, "(%s %s %s)", |
1808 | buf1, is_inv ? ">=" : "<", buf2); |
5101a5f9 |
1809 | break; |
1810 | |
69a3cdfc |
1811 | case PFO_Z: |
a2c1d768 |
1812 | snprintf(buf, buf_size, "(%s %s %s)", |
1813 | buf1, is_inv ? "!=" : "==", buf2); |
850c9265 |
1814 | break; |
1815 | |
5101a5f9 |
1816 | case PFO_BE: // !a |
850c9265 |
1817 | // note: must be unsigned compare |
a2c1d768 |
1818 | snprintf(buf, buf_size, "(%s %s %s)", |
1819 | buf1, is_inv ? ">" : "<=", buf2); |
1820 | |
1821 | // annoying case |
1822 | if (is_inv && lmod == OPLM_BYTE |
1823 | && po->operand[1].type == OPT_CONST |
1824 | && po->operand[1].val == 0xff) |
1825 | { |
1826 | snprintf(g_comment, sizeof(g_comment), "if %s", buf); |
1827 | snprintf(buf, buf_size, "(0)"); |
1828 | } |
5101a5f9 |
1829 | break; |
1830 | |
1831 | // note: must be signed compare |
1832 | case PFO_S: |
1833 | snprintf(buf, buf_size, "(%s(%s - %s) %s 0)", |
a2c1d768 |
1834 | scast, buf1, buf2, is_inv ? ">=" : "<"); |
850c9265 |
1835 | break; |
1836 | |
5101a5f9 |
1837 | case PFO_L: // !ge |
a2c1d768 |
1838 | snprintf(buf, buf_size, "(%s %s %s)", |
1839 | buf1, is_inv ? ">=" : "<", buf2); |
850c9265 |
1840 | break; |
1841 | |
5101a5f9 |
1842 | case PFO_LE: |
a2c1d768 |
1843 | snprintf(buf, buf_size, "(%s %s %s)", |
1844 | buf1, is_inv ? ">" : "<=", buf2); |
5101a5f9 |
1845 | break; |
1846 | |
850c9265 |
1847 | default: |
a2c1d768 |
1848 | break; |
850c9265 |
1849 | } |
1850 | } |
1851 | |
69a3cdfc |
1852 | static void out_cmp_test(char *buf, size_t buf_size, |
940e8e66 |
1853 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv) |
69a3cdfc |
1854 | { |
1855 | char buf1[256], buf2[256], buf3[256]; |
1856 | |
1857 | if (po->op == OP_TEST) { |
1858 | if (IS(opr_name(po, 0), opr_name(po, 1))) { |
3ebea2cf |
1859 | out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]); |
69a3cdfc |
1860 | } |
1861 | else { |
3ebea2cf |
1862 | out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]); |
1863 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]); |
69a3cdfc |
1864 | snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2); |
1865 | } |
940e8e66 |
1866 | out_test_for_cc(buf, buf_size, po, pfo, is_inv, |
69a3cdfc |
1867 | po->operand[0].lmod, buf3); |
1868 | } |
1869 | else if (po->op == OP_CMP) { |
a2c1d768 |
1870 | out_cmp_for_cc(buf, buf_size, po, pfo, is_inv); |
69a3cdfc |
1871 | } |
1872 | else |
1873 | ferr(po, "%s: unhandled op: %d\n", __func__, po->op); |
1874 | } |
1875 | |
850c9265 |
1876 | static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1, |
91977a1c |
1877 | struct parsed_opr *popr2) |
1878 | { |
1879 | if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC) |
1880 | ferr(po, "missing lmod for both operands\n"); |
1881 | |
1882 | if (popr1->lmod == OPLM_UNSPEC) |
1883 | popr1->lmod = popr2->lmod; |
1884 | else if (popr2->lmod == OPLM_UNSPEC) |
1885 | popr2->lmod = popr1->lmod; |
a3684be1 |
1886 | else if (popr1->lmod != popr2->lmod) { |
1887 | if (popr1->type_from_var) { |
1888 | popr1->size_mismatch = 1; |
1889 | if (popr1->lmod < popr2->lmod) |
1890 | popr1->size_lt = 1; |
1891 | popr1->lmod = popr2->lmod; |
1892 | } |
1893 | else if (popr2->type_from_var) { |
1894 | popr2->size_mismatch = 1; |
1895 | if (popr2->lmod < popr1->lmod) |
1896 | popr2->size_lt = 1; |
1897 | popr2->lmod = popr1->lmod; |
1898 | } |
1899 | else |
1900 | ferr(po, "conflicting lmods: %d vs %d\n", |
1901 | popr1->lmod, popr2->lmod); |
1902 | } |
91977a1c |
1903 | } |
c36e914d |
1904 | |
850c9265 |
1905 | static const char *op_to_c(struct parsed_op *po) |
1906 | { |
1907 | switch (po->op) |
1908 | { |
1909 | case OP_ADD: |
5101a5f9 |
1910 | case OP_ADC: |
850c9265 |
1911 | return "+"; |
1912 | case OP_SUB: |
5101a5f9 |
1913 | case OP_SBB: |
850c9265 |
1914 | return "-"; |
1915 | case OP_AND: |
1916 | return "&"; |
1917 | case OP_OR: |
1918 | return "|"; |
1919 | case OP_XOR: |
1920 | return "^"; |
1921 | case OP_SHL: |
1922 | return "<<"; |
1923 | case OP_SHR: |
1924 | return ">>"; |
1925 | case OP_MUL: |
1926 | case OP_IMUL: |
1927 | return "*"; |
1928 | default: |
1929 | ferr(po, "op_to_c was supplied with %d\n", po->op); |
1930 | } |
1931 | } |
1932 | |
7ae48d73 |
1933 | static void op_set_clear_flag(struct parsed_op *po, |
1934 | enum op_flags flag_set, enum op_flags flag_clear) |
d4e3b5db |
1935 | { |
7ae48d73 |
1936 | po->flags |= flag_set; |
1937 | po->flags &= ~flag_clear; |
d4e3b5db |
1938 | } |
1939 | |
de50b98b |
1940 | // last op in stream - unconditional branch or ret |
1941 | #define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \ |
5c024ef7 |
1942 | || (ops[_i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP) |
de50b98b |
1943 | |
87bf6cec |
1944 | static int scan_for_pop(int i, int opcnt, const char *reg, |
d4e3b5db |
1945 | int magic, int depth, int *maxdepth, int do_flags) |
850c9265 |
1946 | { |
89ff3147 |
1947 | const struct parsed_proto *pp; |
87bf6cec |
1948 | struct parsed_op *po; |
1949 | int ret = 0; |
4c45fa73 |
1950 | int j; |
87bf6cec |
1951 | |
850c9265 |
1952 | for (; i < opcnt; i++) { |
87bf6cec |
1953 | po = &ops[i]; |
1954 | if (po->cc_scratch == magic) |
1955 | break; // already checked |
1956 | po->cc_scratch = magic; |
1957 | |
89ff3147 |
1958 | if (po->flags & OPF_TAIL) { |
1959 | if (po->op == OP_CALL) { |
1960 | pp = proto_parse(g_fhdr, po->operand[0].name, 0); |
1961 | if (pp != NULL && pp->is_noreturn) |
1962 | // no stack cleanup for noreturn |
1963 | return ret; |
1964 | } |
87bf6cec |
1965 | return -1; // deadend |
89ff3147 |
1966 | } |
87bf6cec |
1967 | |
1bafb621 |
1968 | if ((po->flags & OPF_RMD) |
de50b98b |
1969 | || (po->op == OP_PUSH && po->argnum != 0)) // arg push |
850c9265 |
1970 | continue; |
1971 | |
87bf6cec |
1972 | if ((po->flags & OPF_JMP) && po->op != OP_CALL) { |
4c45fa73 |
1973 | if (po->btj != NULL) { |
1974 | // jumptable |
da87ae38 |
1975 | for (j = 0; j < po->btj->count; j++) { |
4c45fa73 |
1976 | ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic, |
1977 | depth, maxdepth, do_flags); |
1978 | if (ret < 0) |
1979 | return ret; // dead end |
1980 | } |
da87ae38 |
1981 | return ret; |
4c45fa73 |
1982 | } |
1983 | |
87bf6cec |
1984 | if (po->bt_i < 0) { |
1985 | ferr(po, "dead branch\n"); |
1986 | return -1; |
1987 | } |
850c9265 |
1988 | |
5c024ef7 |
1989 | if (po->flags & OPF_CJMP) { |
d4e3b5db |
1990 | ret |= scan_for_pop(po->bt_i, opcnt, reg, magic, |
1991 | depth, maxdepth, do_flags); |
87bf6cec |
1992 | if (ret < 0) |
1993 | return ret; // dead end |
1994 | } |
1995 | else { |
1996 | i = po->bt_i - 1; |
1997 | } |
1998 | continue; |
1999 | } |
2000 | |
d4e3b5db |
2001 | if ((po->op == OP_POP || po->op == OP_PUSH) |
2002 | && po->operand[0].type == OPT_REG |
87bf6cec |
2003 | && IS(po->operand[0].name, reg)) |
2004 | { |
da87ae38 |
2005 | if (po->op == OP_PUSH && !(po->flags & OPF_FARG)) { |
d4e3b5db |
2006 | depth++; |
2007 | if (depth > *maxdepth) |
2008 | *maxdepth = depth; |
2009 | if (do_flags) |
7ae48d73 |
2010 | op_set_clear_flag(po, OPF_RSAVE, OPF_RMD); |
d4e3b5db |
2011 | } |
da87ae38 |
2012 | else if (po->op == OP_POP) { |
2013 | if (depth == 0) { |
2014 | if (do_flags) |
2015 | op_set_clear_flag(po, OPF_RMD, OPF_RSAVE); |
2016 | return 1; |
2017 | } |
2018 | else { |
2019 | depth--; |
2020 | if (depth < 0) // should not happen |
2021 | ferr(po, "fail with depth\n"); |
2022 | if (do_flags) |
2023 | op_set_clear_flag(po, OPF_RSAVE, OPF_RMD); |
2024 | } |
d4e3b5db |
2025 | } |
87bf6cec |
2026 | } |
850c9265 |
2027 | } |
2028 | |
87bf6cec |
2029 | return ret; |
850c9265 |
2030 | } |
2031 | |
69a3cdfc |
2032 | // scan for pop starting from 'ret' op (all paths) |
d4e3b5db |
2033 | static int scan_for_pop_ret(int i, int opcnt, const char *reg, |
2034 | int flag_set) |
850c9265 |
2035 | { |
2036 | int found = 0; |
2037 | int j; |
2038 | |
2039 | for (; i < opcnt; i++) { |
87bf6cec |
2040 | if (!(ops[i].flags & OPF_TAIL)) |
850c9265 |
2041 | continue; |
2042 | |
2043 | for (j = i - 1; j >= 0; j--) { |
69a3cdfc |
2044 | if (ops[j].flags & OPF_RMD) |
2045 | continue; |
2046 | if (ops[j].flags & OPF_JMP) |
850c9265 |
2047 | return -1; |
2048 | |
2049 | if (ops[j].op == OP_POP && ops[j].operand[0].type == OPT_REG |
2050 | && IS(ops[j].operand[0].name, reg)) |
2051 | { |
2052 | found = 1; |
d4e3b5db |
2053 | ops[j].flags |= flag_set; |
850c9265 |
2054 | break; |
2055 | } |
2056 | |
2057 | if (g_labels[j][0] != 0) |
2058 | return -1; |
2059 | } |
2060 | } |
2061 | |
2062 | return found ? 0 : -1; |
2063 | } |
2064 | |
591721d7 |
2065 | static void scan_propagate_df(int i, int opcnt) |
2066 | { |
2067 | struct parsed_op *po = &ops[i]; |
2068 | int j; |
2069 | |
2070 | for (; i < opcnt; i++) { |
2071 | po = &ops[i]; |
2072 | if (po->flags & OPF_DF) |
2073 | return; // already resolved |
2074 | po->flags |= OPF_DF; |
2075 | |
2076 | if (po->op == OP_CALL) |
2077 | ferr(po, "call with DF set?\n"); |
2078 | |
2079 | if (po->flags & OPF_JMP) { |
2080 | if (po->btj != NULL) { |
2081 | // jumptable |
2082 | for (j = 0; j < po->btj->count; j++) |
2083 | scan_propagate_df(po->btj->d[j].bt_i, opcnt); |
2084 | return; |
2085 | } |
2086 | |
2087 | if (po->bt_i < 0) { |
2088 | ferr(po, "dead branch\n"); |
2089 | return; |
2090 | } |
2091 | |
5c024ef7 |
2092 | if (po->flags & OPF_CJMP) |
591721d7 |
2093 | scan_propagate_df(po->bt_i, opcnt); |
2094 | else |
2095 | i = po->bt_i - 1; |
2096 | continue; |
2097 | } |
2098 | |
2099 | if (po->flags & OPF_TAIL) |
2100 | break; |
2101 | |
2102 | if (po->op == OP_CLD) { |
2103 | po->flags |= OPF_RMD; |
2104 | return; |
2105 | } |
2106 | } |
2107 | |
2108 | ferr(po, "missing DF clear?\n"); |
2109 | } |
2110 | |
1cd4a663 |
2111 | // is operand 'opr' modified by parsed_op 'po'? |
5101a5f9 |
2112 | static int is_opr_modified(const struct parsed_opr *opr, |
69a3cdfc |
2113 | const struct parsed_op *po) |
2114 | { |
89ff3147 |
2115 | int mask; |
2116 | |
69a3cdfc |
2117 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA)) |
2118 | return 0; |
2119 | |
89ff3147 |
2120 | if (opr->type == OPT_REG) { |
2121 | if (po->op == OP_CALL) { |
2122 | mask = (1 << xAX) | (1 << xCX) | (1 << xDX); |
2123 | if ((1 << opr->reg) & mask) |
2124 | return 1; |
2125 | else |
2126 | return 0; |
2127 | } |
2128 | |
2129 | if (po->operand[0].type == OPT_REG) { |
2130 | if (po->regmask_dst & (1 << opr->reg)) |
2131 | return 1; |
2132 | else |
2133 | return 0; |
2134 | } |
69a3cdfc |
2135 | } |
2136 | |
2137 | return IS(po->operand[0].name, opr->name); |
2138 | } |
2139 | |
5101a5f9 |
2140 | // is any operand of parsed_op 'po_test' modified by parsed_op 'po'? |
2141 | static int is_any_opr_modified(const struct parsed_op *po_test, |
89ff3147 |
2142 | const struct parsed_op *po, int c_mode) |
5101a5f9 |
2143 | { |
89ff3147 |
2144 | int mask; |
5101a5f9 |
2145 | int i; |
2146 | |
2147 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA)) |
2148 | return 0; |
2149 | |
de50b98b |
2150 | if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST) |
2151 | return 0; |
2152 | |
2153 | if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst) |
2154 | return 1; |
2155 | |
2156 | // in reality, it can wreck any register, but in decompiled C |
2b43685d |
2157 | // version it can only overwrite eax or edx:eax |
89ff3147 |
2158 | mask = (1 << xAX) | (1 << xDX); |
2159 | if (!c_mode) |
2160 | mask |= 1 << xCX; |
2161 | |
de50b98b |
2162 | if (po->op == OP_CALL |
89ff3147 |
2163 | && ((po_test->regmask_src | po_test->regmask_dst) & mask)) |
5101a5f9 |
2164 | return 1; |
2165 | |
2166 | for (i = 0; i < po_test->operand_cnt; i++) |
2167 | if (IS(po_test->operand[i].name, po->operand[0].name)) |
2168 | return 1; |
2169 | |
2170 | return 0; |
2171 | } |
2172 | |
940e8e66 |
2173 | // scan for any po_test operand modification in range given |
89ff3147 |
2174 | static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt, |
2175 | int c_mode) |
69a3cdfc |
2176 | { |
2b43685d |
2177 | if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST) |
2178 | return -1; |
2179 | |
69a3cdfc |
2180 | for (; i < opcnt; i++) { |
89ff3147 |
2181 | if (is_any_opr_modified(po_test, &ops[i], c_mode)) |
69a3cdfc |
2182 | return i; |
2183 | } |
2184 | |
2185 | return -1; |
2186 | } |
2187 | |
940e8e66 |
2188 | // scan for po_test operand[0] modification in range given |
2189 | static int scan_for_mod_opr0(struct parsed_op *po_test, |
2190 | int i, int opcnt) |
2191 | { |
2192 | for (; i < opcnt; i++) { |
2193 | if (is_opr_modified(&po_test->operand[0], &ops[i])) |
2194 | return i; |
2195 | } |
2196 | |
2197 | return -1; |
2198 | } |
2199 | |
04f8a628 |
2200 | static int scan_for_flag_set(int i, int magic, int *branched, |
2201 | int *setters, int *setter_cnt) |
69a3cdfc |
2202 | { |
04f8a628 |
2203 | struct label_ref *lr; |
2b43685d |
2204 | int ret; |
de50b98b |
2205 | |
2206 | while (i >= 0) { |
04f8a628 |
2207 | if (ops[i].cc_scratch == magic) { |
2208 | ferr(&ops[i], "%s looped\n", __func__); |
2209 | return -1; |
2210 | } |
2211 | ops[i].cc_scratch = magic; |
2212 | |
de50b98b |
2213 | if (g_labels[i][0] != 0) { |
2b43685d |
2214 | *branched = 1; |
04f8a628 |
2215 | |
2216 | lr = &g_label_refs[i]; |
2217 | for (; lr->next; lr = lr->next) { |
2218 | ret = scan_for_flag_set(lr->i, magic, |
2219 | branched, setters, setter_cnt); |
2220 | if (ret < 0) |
2221 | return ret; |
2222 | } |
2223 | |
de50b98b |
2224 | if (i > 0 && LAST_OP(i - 1)) { |
94d447fb |
2225 | i = lr->i; |
de50b98b |
2226 | continue; |
2227 | } |
04f8a628 |
2228 | ret = scan_for_flag_set(lr->i, magic, |
2229 | branched, setters, setter_cnt); |
2b43685d |
2230 | if (ret < 0) |
2231 | return ret; |
de50b98b |
2232 | } |
2233 | i--; |
2234 | |
2b43685d |
2235 | if (ops[i].flags & OPF_FLAGS) { |
2236 | setters[*setter_cnt] = i; |
2237 | (*setter_cnt)++; |
2238 | return 0; |
2239 | } |
69a3cdfc |
2240 | |
5c024ef7 |
2241 | if ((ops[i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP) |
69a3cdfc |
2242 | return -1; |
69a3cdfc |
2243 | } |
2244 | |
2245 | return -1; |
2246 | } |
2247 | |
5101a5f9 |
2248 | // scan back for cdq, if anything modifies edx, fail |
2249 | static int scan_for_cdq_edx(int i) |
2250 | { |
cdfaeed7 |
2251 | while (i >= 0) { |
2252 | if (g_labels[i][0] != 0) { |
2253 | if (g_label_refs[i].next != NULL) |
2254 | return -1; |
2255 | if (i > 0 && LAST_OP(i - 1)) { |
2256 | i = g_label_refs[i].i; |
2257 | continue; |
2258 | } |
2259 | return -1; |
2260 | } |
2261 | i--; |
2262 | |
5101a5f9 |
2263 | if (ops[i].op == OP_CDQ) |
2264 | return i; |
2265 | |
2266 | if (ops[i].regmask_dst & (1 << xDX)) |
2267 | return -1; |
5101a5f9 |
2268 | } |
2269 | |
2270 | return -1; |
2271 | } |
2272 | |
64c59faf |
2273 | static int scan_for_reg_clear(int i, int reg) |
2274 | { |
cdfaeed7 |
2275 | while (i >= 0) { |
2276 | if (g_labels[i][0] != 0) { |
2277 | if (g_label_refs[i].next != NULL) |
2278 | return -1; |
2279 | if (i > 0 && LAST_OP(i - 1)) { |
2280 | i = g_label_refs[i].i; |
2281 | continue; |
2282 | } |
2283 | return -1; |
2284 | } |
2285 | i--; |
2286 | |
64c59faf |
2287 | if (ops[i].op == OP_XOR |
2288 | && ops[i].operand[0].lmod == OPLM_DWORD |
2289 | && ops[i].operand[0].reg == ops[i].operand[1].reg |
2290 | && ops[i].operand[0].reg == reg) |
2291 | return i; |
2292 | |
2293 | if (ops[i].regmask_dst & (1 << reg)) |
2294 | return -1; |
64c59faf |
2295 | } |
2296 | |
2297 | return -1; |
2298 | } |
2299 | |
1bafb621 |
2300 | // scan for positive, constant esp adjust |
2301 | static int scan_for_esp_adjust(int i, int opcnt, int *adj) |
2302 | { |
89ff3147 |
2303 | const struct parsed_proto *pp; |
7ba45c34 |
2304 | struct parsed_op *po; |
46411e6c |
2305 | int first_pop = -1; |
7ba45c34 |
2306 | *adj = 0; |
2307 | |
1bafb621 |
2308 | for (; i < opcnt; i++) { |
7ba45c34 |
2309 | po = &ops[i]; |
2310 | |
a2c1d768 |
2311 | if (g_labels[i][0] != 0) |
2312 | break; |
2313 | |
7ba45c34 |
2314 | if (po->op == OP_ADD && po->operand[0].reg == xSP) { |
2315 | if (po->operand[1].type != OPT_CONST) |
1bafb621 |
2316 | ferr(&ops[i], "non-const esp adjust?\n"); |
7ba45c34 |
2317 | *adj += po->operand[1].val; |
1bafb621 |
2318 | if (*adj & 3) |
2319 | ferr(&ops[i], "unaligned esp adjust: %x\n", *adj); |
2320 | return i; |
2321 | } |
46411e6c |
2322 | else if (po->op == OP_PUSH) { |
5c024ef7 |
2323 | //if (first_pop == -1) |
2324 | // first_pop = -2; // none |
7ba45c34 |
2325 | *adj -= lmod_bytes(po, po->operand[0].lmod); |
46411e6c |
2326 | } |
5c024ef7 |
2327 | else if (po->op == OP_POP && !(po->flags & OPF_RMD)) { |
2328 | if (first_pop == -1 && *adj >= 0) |
46411e6c |
2329 | first_pop = i; |
7ba45c34 |
2330 | *adj += lmod_bytes(po, po->operand[0].lmod); |
46411e6c |
2331 | } |
e56ab892 |
2332 | else if (po->flags & (OPF_JMP|OPF_TAIL)) { |
2333 | if (po->op != OP_CALL) |
a2c1d768 |
2334 | break; |
e56ab892 |
2335 | if (po->operand[0].type != OPT_LABEL) |
a2c1d768 |
2336 | break; |
89ff3147 |
2337 | pp = po->datap; |
2338 | if (pp != NULL && pp->is_stdcall) |
2339 | break; |
e56ab892 |
2340 | } |
a2c1d768 |
2341 | } |
7ba45c34 |
2342 | |
591721d7 |
2343 | if (*adj <= 8 && first_pop >= 0 && ops[first_pop].op == OP_POP |
46411e6c |
2344 | && ops[first_pop].operand[0].type == OPT_REG |
2345 | && ops[first_pop].operand[0].reg == xCX) |
a2c1d768 |
2346 | { |
2347 | // probably 'pop ecx' was used.. |
46411e6c |
2348 | return first_pop; |
1bafb621 |
2349 | } |
2350 | |
2351 | return -1; |
2352 | } |
2353 | |
a3684be1 |
2354 | static void scan_fwd_set_flags(int i, int opcnt, int magic, int flags) |
2355 | { |
2356 | struct parsed_op *po; |
2357 | int j; |
2358 | |
2359 | if (i < 0) |
2360 | ferr(ops, "%s: followed bad branch?\n", __func__); |
2361 | |
2362 | for (; i < opcnt; i++) { |
2363 | po = &ops[i]; |
2364 | if (po->cc_scratch == magic) |
2365 | return; |
2366 | po->cc_scratch = magic; |
2367 | po->flags |= flags; |
2368 | |
2369 | if ((po->flags & OPF_JMP) && po->op != OP_CALL) { |
2370 | if (po->btj != NULL) { |
2371 | // jumptable |
2372 | for (j = 0; j < po->btj->count; j++) |
2373 | scan_fwd_set_flags(po->btj->d[j].bt_i, opcnt, magic, flags); |
2374 | return; |
2375 | } |
2376 | |
2377 | scan_fwd_set_flags(po->bt_i, opcnt, magic, flags); |
5c024ef7 |
2378 | if (!(po->flags & OPF_CJMP)) |
a3684be1 |
2379 | return; |
2380 | } |
2381 | if (po->flags & OPF_TAIL) |
2382 | return; |
2383 | } |
2384 | } |
2385 | |
1cd4a663 |
2386 | static const struct parsed_proto *try_recover_pp( |
da87ae38 |
2387 | struct parsed_op *po, const struct parsed_opr *opr, int *search_instead) |
1cd4a663 |
2388 | { |
2389 | const struct parsed_proto *pp = NULL; |
89ff3147 |
2390 | char buf[256]; |
2391 | char *p; |
1cd4a663 |
2392 | |
2393 | // maybe an arg of g_func? |
2394 | if (opr->type == OPT_REGMEM && is_stack_access(po, opr)) |
2395 | { |
2396 | char ofs_reg[16] = { 0, }; |
2397 | int arg, arg_s, arg_i; |
2398 | int stack_ra = 0; |
2399 | int offset = 0; |
2400 | |
2401 | parse_stack_access(po, opr->name, ofs_reg, |
2402 | &offset, &stack_ra, NULL); |
2403 | if (ofs_reg[0] != 0) |
2404 | ferr(po, "offset reg on arg access?\n"); |
da87ae38 |
2405 | if (offset <= stack_ra) { |
2406 | // search who set the stack var instead |
2407 | if (search_instead != NULL) |
2408 | *search_instead = 1; |
2409 | return NULL; |
2410 | } |
1cd4a663 |
2411 | |
2412 | arg_i = (offset - stack_ra - 4) / 4; |
2413 | for (arg = arg_s = 0; arg < g_func_pp->argc; arg++) { |
2414 | if (g_func_pp->arg[arg].reg != NULL) |
2415 | continue; |
2416 | if (arg_s == arg_i) |
2417 | break; |
2418 | arg_s++; |
2419 | } |
2420 | if (arg == g_func_pp->argc) |
2421 | ferr(po, "stack arg %d not in prototype?\n", arg_i); |
2422 | |
2423 | pp = g_func_pp->arg[arg].fptr; |
2424 | if (pp == NULL) |
46411e6c |
2425 | ferr(po, "icall sa: arg%d is not a fptr?\n", arg + 1); |
89ff3147 |
2426 | check_func_pp(po, pp, "icall arg"); |
2427 | } |
2428 | else if (opr->type == OPT_REGMEM && strchr(opr->name + 1, '[')) { |
2429 | // label[index] |
2430 | p = strchr(opr->name + 1, '['); |
2431 | memcpy(buf, opr->name, p - opr->name); |
2432 | buf[p - opr->name] = 0; |
2433 | pp = proto_parse(g_fhdr, buf, 0); |
1cd4a663 |
2434 | } |
2435 | else if (opr->type == OPT_OFFSET || opr->type == OPT_LABEL) { |
36595fd2 |
2436 | pp = proto_parse(g_fhdr, opr->name, 0); |
1cd4a663 |
2437 | if (pp == NULL) |
2438 | ferr(po, "proto_parse failed for icall from '%s'\n", opr->name); |
89ff3147 |
2439 | check_func_pp(po, pp, "reg-fptr ref"); |
1cd4a663 |
2440 | } |
2441 | |
2442 | return pp; |
2443 | } |
2444 | |
da87ae38 |
2445 | static void scan_for_call_type(int i, const struct parsed_opr *opr, |
89ff3147 |
2446 | int magic, const struct parsed_proto **pp_found, int *multi) |
1cd4a663 |
2447 | { |
2448 | const struct parsed_proto *pp = NULL; |
2449 | struct parsed_op *po; |
2450 | struct label_ref *lr; |
2451 | |
2452 | while (i >= 0) { |
2453 | if (ops[i].cc_scratch == magic) |
2454 | return; |
2455 | ops[i].cc_scratch = magic; |
2456 | |
2457 | if (g_labels[i][0] != 0) { |
2458 | lr = &g_label_refs[i]; |
2459 | for (; lr != NULL; lr = lr->next) |
89ff3147 |
2460 | scan_for_call_type(lr->i, opr, magic, pp_found, multi); |
46411e6c |
2461 | if (i > 0 && LAST_OP(i - 1)) |
2462 | return; |
1cd4a663 |
2463 | } |
2464 | i--; |
2465 | |
2466 | if (!(ops[i].flags & OPF_DATA)) |
2467 | continue; |
2468 | if (!is_opr_modified(opr, &ops[i])) |
2469 | continue; |
2470 | if (ops[i].op != OP_MOV && ops[i].op != OP_LEA) { |
2471 | // most probably trashed by some processing |
2472 | *pp_found = NULL; |
2473 | return; |
2474 | } |
2475 | |
2476 | opr = &ops[i].operand[1]; |
2477 | if (opr->type != OPT_REG) |
2478 | break; |
2479 | } |
2480 | |
2481 | po = (i >= 0) ? &ops[i] : ops; |
2482 | |
2483 | if (i < 0) { |
2484 | // reached the top - can only be an arg-reg |
2485 | if (opr->type != OPT_REG) |
2486 | return; |
2487 | |
2488 | for (i = 0; i < g_func_pp->argc; i++) { |
2489 | if (g_func_pp->arg[i].reg == NULL) |
2490 | continue; |
2491 | if (IS(opr->name, g_func_pp->arg[i].reg)) |
2492 | break; |
2493 | } |
2494 | if (i == g_func_pp->argc) |
2495 | return; |
2496 | pp = g_func_pp->arg[i].fptr; |
2497 | if (pp == NULL) |
46411e6c |
2498 | ferr(po, "icall: arg%d (%s) is not a fptr?\n", |
2499 | i + 1, g_func_pp->arg[i].reg); |
89ff3147 |
2500 | check_func_pp(po, pp, "icall reg-arg"); |
1cd4a663 |
2501 | } |
2502 | else |
da87ae38 |
2503 | pp = try_recover_pp(po, opr, NULL); |
1cd4a663 |
2504 | |
89ff3147 |
2505 | if (*pp_found != NULL && pp != NULL && *pp_found != pp) { |
1cd4a663 |
2506 | if (!IS((*pp_found)->ret_type.name, pp->ret_type.name) |
2507 | || (*pp_found)->is_stdcall != pp->is_stdcall |
89ff3147 |
2508 | || (*pp_found)->is_fptr != pp->is_fptr |
1cd4a663 |
2509 | || (*pp_found)->argc != pp->argc |
2510 | || (*pp_found)->argc_reg != pp->argc_reg |
2511 | || (*pp_found)->argc_stack != pp->argc_stack) |
2512 | { |
2513 | ferr(po, "icall: parsed_proto mismatch\n"); |
2514 | } |
89ff3147 |
2515 | *multi = 1; |
1cd4a663 |
2516 | } |
2517 | if (pp != NULL) |
2518 | *pp_found = pp; |
2519 | } |
2520 | |
89ff3147 |
2521 | static const struct parsed_proto *resolve_icall(int i, int opcnt, |
2522 | int *multi_src) |
1cd4a663 |
2523 | { |
2524 | const struct parsed_proto *pp = NULL; |
da87ae38 |
2525 | int search_advice = 0; |
1cd4a663 |
2526 | |
89ff3147 |
2527 | *multi_src = 0; |
2528 | |
1cd4a663 |
2529 | switch (ops[i].operand[0].type) { |
2530 | case OPT_REGMEM: |
2531 | case OPT_LABEL: |
2532 | case OPT_OFFSET: |
da87ae38 |
2533 | pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice); |
2534 | if (!search_advice) |
2535 | break; |
2536 | // fallthrough |
1cd4a663 |
2537 | default: |
89ff3147 |
2538 | scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp, |
2539 | multi_src); |
1cd4a663 |
2540 | break; |
2541 | } |
2542 | |
2543 | return pp; |
2544 | } |
2545 | |
89ff3147 |
2546 | static int collect_call_args_r(struct parsed_op *po, int i, |
75ad0378 |
2547 | struct parsed_proto *pp, int *regmask, int *save_arg_vars, int arg, |
a3684be1 |
2548 | int magic, int need_op_saving, int may_reuse) |
e56ab892 |
2549 | { |
2550 | struct parsed_proto *pp_tmp; |
2551 | struct label_ref *lr; |
2b43685d |
2552 | int need_to_save_current; |
e56ab892 |
2553 | int ret = 0; |
2554 | int j; |
2555 | |
a3684be1 |
2556 | if (i < 0) { |
a2c1d768 |
2557 | ferr(po, "dead label encountered\n"); |
a3684be1 |
2558 | return -1; |
2559 | } |
e56ab892 |
2560 | |
de50b98b |
2561 | for (; arg < pp->argc; arg++) |
e56ab892 |
2562 | if (pp->arg[arg].reg == NULL) |
2563 | break; |
a3684be1 |
2564 | magic = (magic & 0xffffff) | (arg << 24); |
e56ab892 |
2565 | |
89ff3147 |
2566 | for (j = i; j >= 0 && (arg < pp->argc || pp->is_unresolved); ) |
e56ab892 |
2567 | { |
a3684be1 |
2568 | if (((ops[j].cc_scratch ^ magic) & 0xffffff) == 0) { |
2569 | if (ops[j].cc_scratch != magic) { |
2570 | ferr(&ops[j], "arg collect hit same path with diff args for %s\n", |
2571 | pp->name); |
2572 | return -1; |
2573 | } |
2574 | // ok: have already been here |
2575 | return 0; |
2576 | } |
2577 | ops[j].cc_scratch = magic; |
2578 | |
a2c1d768 |
2579 | if (g_labels[j][0] != 0 && g_label_refs[j].i != -1) { |
e56ab892 |
2580 | lr = &g_label_refs[j]; |
2581 | if (lr->next != NULL) |
2582 | need_op_saving = 1; |
a652aa9f |
2583 | for (; lr->next; lr = lr->next) { |
5c024ef7 |
2584 | if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP) |
a652aa9f |
2585 | may_reuse = 1; |
89ff3147 |
2586 | ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars, |
a3684be1 |
2587 | arg, magic, need_op_saving, may_reuse); |
2588 | if (ret < 0) |
2589 | return ret; |
a652aa9f |
2590 | } |
e56ab892 |
2591 | |
5c024ef7 |
2592 | if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP) |
a652aa9f |
2593 | may_reuse = 1; |
de50b98b |
2594 | if (j > 0 && LAST_OP(j - 1)) { |
e56ab892 |
2595 | // follow last branch in reverse |
2596 | j = lr->i; |
2597 | continue; |
2598 | } |
2599 | need_op_saving = 1; |
89ff3147 |
2600 | ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars, |
a3684be1 |
2601 | arg, magic, need_op_saving, may_reuse); |
2602 | if (ret < 0) |
2603 | return ret; |
e56ab892 |
2604 | } |
2605 | j--; |
2606 | |
2607 | if (ops[j].op == OP_CALL) |
2608 | { |
89ff3147 |
2609 | if (pp->is_unresolved) |
2610 | break; |
2611 | |
e56ab892 |
2612 | pp_tmp = ops[j].datap; |
2613 | if (pp_tmp == NULL) |
7ae48d73 |
2614 | ferr(po, "arg collect hit unparsed call '%s'\n", |
2615 | ops[j].operand[0].name); |
a652aa9f |
2616 | if (may_reuse && pp_tmp->argc_stack > 0) |
de50b98b |
2617 | ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n", |
2618 | arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack); |
e56ab892 |
2619 | } |
de50b98b |
2620 | else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP) { |
89ff3147 |
2621 | if (pp->is_unresolved) |
2622 | break; |
2623 | |
de50b98b |
2624 | ferr(po, "arg collect %d/%d hit esp adjust\n", |
2625 | arg, pp->argc); |
2626 | } |
2627 | else if (ops[j].op == OP_POP) { |
89ff3147 |
2628 | if (pp->is_unresolved) |
2629 | break; |
2630 | |
de50b98b |
2631 | ferr(po, "arg collect %d/%d hit pop\n", arg, pp->argc); |
2632 | } |
5c024ef7 |
2633 | else if (ops[j].flags & OPF_CJMP) |
de50b98b |
2634 | { |
89ff3147 |
2635 | if (pp->is_unresolved) |
2636 | break; |
2637 | |
a652aa9f |
2638 | may_reuse = 1; |
de50b98b |
2639 | } |
2640 | else if (ops[j].op == OP_PUSH && !(ops[j].flags & OPF_FARG)) |
e56ab892 |
2641 | { |
89ff3147 |
2642 | if (pp->is_unresolved && (ops[j].flags & OPF_RMD)) |
2643 | break; |
2644 | |
e56ab892 |
2645 | pp->arg[arg].datap = &ops[j]; |
2b43685d |
2646 | need_to_save_current = 0; |
e56ab892 |
2647 | if (!need_op_saving) { |
89ff3147 |
2648 | ret = scan_for_mod(&ops[j], j + 1, i, 1); |
2b43685d |
2649 | need_to_save_current = (ret >= 0); |
e56ab892 |
2650 | } |
2b43685d |
2651 | if (need_op_saving || need_to_save_current) { |
e56ab892 |
2652 | // mark this push as one that needs operand saving |
2653 | ops[j].flags &= ~OPF_RMD; |
de50b98b |
2654 | if (ops[j].argnum == 0) { |
2655 | ops[j].argnum = arg + 1; |
2656 | *save_arg_vars |= 1 << arg; |
2657 | } |
2658 | else if (ops[j].argnum < arg + 1) |
2659 | ferr(&ops[j], "argnum conflict (%d<%d) for '%s'\n", |
2660 | ops[j].argnum, arg + 1, pp->name); |
e56ab892 |
2661 | } |
de50b98b |
2662 | else if (ops[j].argnum == 0) |
e56ab892 |
2663 | ops[j].flags |= OPF_RMD; |
2664 | |
a652aa9f |
2665 | // some PUSHes are reused by different calls on other branches, |
de50b98b |
2666 | // but that can't happen if we didn't branch, so they |
2667 | // can be removed from future searches (handles nested calls) |
a652aa9f |
2668 | if (!may_reuse) |
de50b98b |
2669 | ops[j].flags |= OPF_FARG; |
2670 | |
da87ae38 |
2671 | ops[j].flags &= ~OPF_RSAVE; |
2672 | |
89ff3147 |
2673 | arg++; |
2674 | if (!pp->is_unresolved) { |
2675 | // next arg |
2676 | for (; arg < pp->argc; arg++) |
2677 | if (pp->arg[arg].reg == NULL) |
2678 | break; |
2679 | } |
a3684be1 |
2680 | magic = (magic & 0xffffff) | (arg << 24); |
75ad0378 |
2681 | |
2682 | // tracking reg usage |
2683 | if (ops[j].operand[0].type == OPT_REG) |
2684 | *regmask |= 1 << ops[j].operand[0].reg; |
e56ab892 |
2685 | } |
2686 | } |
2687 | |
2688 | if (arg < pp->argc) { |
2689 | ferr(po, "arg collect failed for '%s': %d/%d\n", |
2690 | pp->name, arg, pp->argc); |
89ff3147 |
2691 | return -1; |
e56ab892 |
2692 | } |
89ff3147 |
2693 | |
2694 | return arg; |
2695 | } |
2696 | |
2697 | static int collect_call_args(struct parsed_op *po, int i, |
2698 | struct parsed_proto *pp, int *regmask, int *save_arg_vars, |
2699 | int magic) |
2700 | { |
2701 | int ret; |
2702 | int a; |
2703 | |
2704 | ret = collect_call_args_r(po, i, pp, regmask, save_arg_vars, |
2705 | 0, magic, 0, 0); |
2706 | if (ret < 0) |
2707 | return ret; |
2708 | |
2709 | if (pp->is_unresolved) { |
2710 | pp->argc += ret; |
2711 | pp->argc_stack += ret; |
2712 | for (a = 0; a < pp->argc; a++) |
2713 | if (pp->arg[a].type.name == NULL) |
2714 | pp->arg[a].type.name = strdup("int"); |
2715 | } |
2716 | |
e56ab892 |
2717 | return ret; |
2718 | } |
2719 | |
89ff3147 |
2720 | static void pp_insert_reg_arg(struct parsed_proto *pp, const char *reg) |
2721 | { |
2722 | int i; |
2723 | |
2724 | for (i = 0; i < pp->argc; i++) |
2725 | if (pp->arg[i].reg == NULL) |
2726 | break; |
2727 | |
2728 | if (pp->argc_stack) |
2729 | memmove(&pp->arg[i + 1], &pp->arg[i], |
2730 | sizeof(pp->arg[0]) * pp->argc_stack); |
2731 | memset(&pp->arg[i], 0, sizeof(pp->arg[i])); |
2732 | pp->arg[i].reg = strdup(reg); |
2733 | pp->arg[i].type.name = strdup("int"); |
2734 | pp->argc++; |
2735 | pp->argc_reg++; |
2736 | } |
2737 | |
4c45fa73 |
2738 | static void add_label_ref(struct label_ref *lr, int op_i) |
2739 | { |
2740 | struct label_ref *lr_new; |
2741 | |
2742 | if (lr->i == -1) { |
2743 | lr->i = op_i; |
2744 | return; |
2745 | } |
2746 | |
2747 | lr_new = calloc(1, sizeof(*lr_new)); |
2748 | lr_new->i = op_i; |
a652aa9f |
2749 | lr_new->next = lr->next; |
4c45fa73 |
2750 | lr->next = lr_new; |
2751 | } |
2752 | |
04f8a628 |
2753 | static void output_std_flags(FILE *fout, struct parsed_op *po, |
2754 | int *pfomask, const char *dst_opr_text) |
2755 | { |
2756 | if (*pfomask & (1 << PFO_Z)) { |
2757 | fprintf(fout, "\n cond_z = (%s%s == 0);", |
2758 | lmod_cast_u(po, po->operand[0].lmod), dst_opr_text); |
2759 | *pfomask &= ~(1 << PFO_Z); |
2760 | } |
2761 | if (*pfomask & (1 << PFO_S)) { |
2762 | fprintf(fout, "\n cond_s = (%s%s < 0);", |
2763 | lmod_cast_s(po, po->operand[0].lmod), dst_opr_text); |
2764 | *pfomask &= ~(1 << PFO_S); |
2765 | } |
2766 | } |
2767 | |
c0050df6 |
2768 | static void output_pp_attrs(FILE *fout, const struct parsed_proto *pp, |
2769 | int is_noreturn) |
2770 | { |
2771 | if (pp->is_fastcall) |
2772 | fprintf(fout, "__fastcall "); |
2773 | else if (pp->is_stdcall && pp->argc_reg == 0) |
2774 | fprintf(fout, "__stdcall "); |
2775 | if (pp->is_noreturn || is_noreturn) |
2776 | fprintf(fout, "noreturn "); |
2777 | } |
2778 | |
91977a1c |
2779 | static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt) |
2780 | { |
69a3cdfc |
2781 | struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op; |
850c9265 |
2782 | struct parsed_opr *last_arith_dst = NULL; |
3ebea2cf |
2783 | char buf1[256], buf2[256], buf3[256], cast[64]; |
bd96f656 |
2784 | const struct parsed_proto *pp_c; |
ddaf8bd7 |
2785 | struct parsed_proto *pp, *pp_tmp; |
4c45fa73 |
2786 | struct parsed_data *pd; |
91977a1c |
2787 | const char *tmpname; |
940e8e66 |
2788 | enum parsed_flag_op pfo; |
69a3cdfc |
2789 | int save_arg_vars = 0; |
2790 | int cmp_result_vars = 0; |
87bf6cec |
2791 | int need_mul_var = 0; |
91977a1c |
2792 | int had_decl = 0; |
3ebea2cf |
2793 | int label_pending = 0; |
d4e3b5db |
2794 | int regmask_save = 0; |
91977a1c |
2795 | int regmask_arg = 0; |
a3684be1 |
2796 | int regmask_now = 0; |
ddaf8bd7 |
2797 | int regmask_init = 0; |
91977a1c |
2798 | int regmask = 0; |
940e8e66 |
2799 | int pfomask = 0; |
64c59faf |
2800 | int found = 0; |
d4e3b5db |
2801 | int depth = 0; |
91977a1c |
2802 | int no_output; |
4c45fa73 |
2803 | int i, j, l; |
69a3cdfc |
2804 | int dummy; |
91977a1c |
2805 | int arg; |
91977a1c |
2806 | int reg; |
2807 | int ret; |
2808 | |
1bafb621 |
2809 | g_bp_frame = g_sp_frame = g_stack_fsz = 0; |
a2c1d768 |
2810 | g_stack_frame_used = 0; |
91977a1c |
2811 | |
36595fd2 |
2812 | g_func_pp = proto_parse(fhdr, funcn, 0); |
bd96f656 |
2813 | if (g_func_pp == NULL) |
91977a1c |
2814 | ferr(ops, "proto_parse failed for '%s'\n", funcn); |
2815 | |
bd96f656 |
2816 | fprintf(fout, "%s ", g_func_pp->ret_type.name); |
c0050df6 |
2817 | output_pp_attrs(fout, g_func_pp, g_ida_func_attr & IDAFA_NORETURN); |
e56ab892 |
2818 | fprintf(fout, "%s(", funcn); |
54e763a1 |
2819 | |
bd96f656 |
2820 | for (i = 0; i < g_func_pp->argc; i++) { |
91977a1c |
2821 | if (i > 0) |
2822 | fprintf(fout, ", "); |
54e763a1 |
2823 | if (g_func_pp->arg[i].fptr != NULL) { |
2824 | // func pointer.. |
2825 | pp = g_func_pp->arg[i].fptr; |
2826 | fprintf(fout, "%s (", pp->ret_type.name); |
c0050df6 |
2827 | output_pp_attrs(fout, pp, 0); |
54e763a1 |
2828 | fprintf(fout, "*a%d)(", i + 1); |
2829 | for (j = 0; j < pp->argc; j++) { |
2830 | if (j > 0) |
2831 | fprintf(fout, ", "); |
2832 | if (pp->arg[j].fptr) |
2833 | ferr(ops, "nested fptr\n"); |
2834 | fprintf(fout, "%s", pp->arg[j].type.name); |
2835 | } |
2836 | fprintf(fout, ")"); |
2837 | } |
2838 | else { |
2839 | fprintf(fout, "%s a%d", g_func_pp->arg[i].type.name, i + 1); |
2840 | } |
89ff3147 |
2841 | if (g_func_pp->arg[i].reg != NULL) { |
2842 | reg = char_array_i(regs_r32, |
2843 | ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg); |
2844 | if (reg < 0) |
2845 | ferr(ops, "arg '%s' is not a reg?\n", g_func_pp->arg[i].reg); |
2846 | regmask_arg |= 1 << reg; |
2847 | } |
91977a1c |
2848 | } |
bd96f656 |
2849 | if (g_func_pp->is_vararg) { |
4f12f671 |
2850 | if (i > 0) |
2851 | fprintf(fout, ", "); |
2852 | fprintf(fout, "..."); |
2853 | } |
54e763a1 |
2854 | |
91977a1c |
2855 | fprintf(fout, ")\n{\n"); |
2856 | |
2857 | // pass1: |
1bafb621 |
2858 | // - handle ebp/esp frame, remove ops related to it |
91977a1c |
2859 | if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp") |
2860 | && ops[1].op == OP_MOV |
2861 | && IS(opr_name(&ops[1], 0), "ebp") |
2862 | && IS(opr_name(&ops[1], 1), "esp")) |
2863 | { |
87bf6cec |
2864 | int ecx_push = 0; |
2865 | |
91977a1c |
2866 | g_bp_frame = 1; |
69a3cdfc |
2867 | ops[0].flags |= OPF_RMD; |
2868 | ops[1].flags |= OPF_RMD; |
4c45fa73 |
2869 | i = 2; |
91977a1c |
2870 | |
2871 | if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) { |
1bafb621 |
2872 | g_stack_fsz = opr_const(&ops[2], 1); |
69a3cdfc |
2873 | ops[2].flags |= OPF_RMD; |
4c45fa73 |
2874 | i++; |
91977a1c |
2875 | } |
87bf6cec |
2876 | else { |
2877 | // another way msvc builds stack frame.. |
2878 | i = 2; |
2879 | while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) { |
1bafb621 |
2880 | g_stack_fsz += 4; |
87bf6cec |
2881 | ops[i].flags |= OPF_RMD; |
2882 | ecx_push++; |
2883 | i++; |
2884 | } |
4c45fa73 |
2885 | // and another way.. |
2886 | if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX |
2887 | && ops[i].operand[1].type == OPT_CONST |
2888 | && ops[i + 1].op == OP_CALL |
2889 | && IS(opr_name(&ops[i + 1], 0), "__alloca_probe")) |
2890 | { |
2891 | g_stack_fsz += ops[i].operand[1].val; |
2892 | ops[i].flags |= OPF_RMD; |
2893 | i++; |
2894 | ops[i].flags |= OPF_RMD; |
2895 | i++; |
2896 | } |
87bf6cec |
2897 | } |
91977a1c |
2898 | |
64c59faf |
2899 | found = 0; |
91977a1c |
2900 | do { |
2901 | for (; i < opcnt; i++) |
2902 | if (ops[i].op == OP_RET) |
2903 | break; |
64c59faf |
2904 | if (i == opcnt && (ops[i - 1].flags & OPF_JMP) && found) |
2905 | break; |
2906 | |
591721d7 |
2907 | if ((ops[i - 1].op == OP_POP && IS(opr_name(&ops[i - 1], 0), "ebp")) |
2908 | || ops[i - 1].op == OP_LEAVE) |
2909 | { |
e56ab892 |
2910 | ops[i - 1].flags |= OPF_RMD; |
591721d7 |
2911 | } |
e56ab892 |
2912 | else if (!(g_ida_func_attr & IDAFA_NORETURN)) |
91977a1c |
2913 | ferr(&ops[i - 1], "'pop ebp' expected\n"); |
91977a1c |
2914 | |
1bafb621 |
2915 | if (g_stack_fsz != 0) { |
e56ab892 |
2916 | if (ops[i - 2].op == OP_MOV |
2917 | && IS(opr_name(&ops[i - 2], 0), "esp") |
2918 | && IS(opr_name(&ops[i - 2], 1), "ebp")) |
91977a1c |
2919 | { |
e56ab892 |
2920 | ops[i - 2].flags |= OPF_RMD; |
91977a1c |
2921 | } |
591721d7 |
2922 | else if (ops[i - 1].op != OP_LEAVE |
2923 | && !(g_ida_func_attr & IDAFA_NORETURN)) |
2924 | { |
e56ab892 |
2925 | ferr(&ops[i - 2], "esp restore expected\n"); |
591721d7 |
2926 | } |
87bf6cec |
2927 | |
2928 | if (ecx_push && ops[i - 3].op == OP_POP |
2929 | && IS(opr_name(&ops[i - 3], 0), "ecx")) |
2930 | { |
2931 | ferr(&ops[i - 3], "unexpected ecx pop\n"); |
2932 | } |
91977a1c |
2933 | } |
87bf6cec |
2934 | |
64c59faf |
2935 | found = 1; |
91977a1c |
2936 | i++; |
2937 | } while (i < opcnt); |
2938 | } |
1bafb621 |
2939 | else { |
2940 | for (i = 0; i < opcnt; i++) { |
2941 | if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL))) |
2942 | break; |
2943 | if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP |
2944 | && ops[i].operand[1].type == OPT_CONST) |
2945 | { |
2946 | g_sp_frame = 1; |
2947 | break; |
2948 | } |
2949 | } |
2950 | |
2951 | if (g_sp_frame) |
2952 | { |
2953 | g_stack_fsz = ops[i].operand[1].val; |
2954 | ops[i].flags |= OPF_RMD; |
2955 | |
2956 | i++; |
2957 | do { |
2958 | for (; i < opcnt; i++) |
2959 | if (ops[i].op == OP_RET) |
2960 | break; |
2961 | if (ops[i - 1].op != OP_ADD |
2962 | || !IS(opr_name(&ops[i - 1], 0), "esp") |
2963 | || ops[i - 1].operand[1].type != OPT_CONST |
2964 | || ops[i - 1].operand[1].val != g_stack_fsz) |
2965 | ferr(&ops[i - 1], "'add esp' expected\n"); |
2966 | ops[i - 1].flags |= OPF_RMD; |
2967 | |
2968 | i++; |
2969 | } while (i < opcnt); |
2970 | } |
2971 | } |
91977a1c |
2972 | |
2973 | // pass2: |
7ae48d73 |
2974 | // - parse calls with labels |
87bf6cec |
2975 | // - resolve all branches |
de50b98b |
2976 | for (i = 0; i < opcnt; i++) |
2977 | { |
87bf6cec |
2978 | po = &ops[i]; |
2979 | po->bt_i = -1; |
4c45fa73 |
2980 | po->btj = NULL; |
87bf6cec |
2981 | |
7ae48d73 |
2982 | if (po->flags & OPF_RMD) |
2983 | continue; |
2984 | |
2985 | if (po->op == OP_CALL) { |
ddaf8bd7 |
2986 | pp = NULL; |
2987 | |
7ae48d73 |
2988 | if (po->operand[0].type == OPT_LABEL) { |
2989 | tmpname = opr_name(po, 0); |
46411e6c |
2990 | if (IS_START(tmpname, "loc_")) |
2991 | ferr(po, "call to loc_*\n"); |
36595fd2 |
2992 | pp_c = proto_parse(fhdr, tmpname, 0); |
7ae48d73 |
2993 | if (pp_c == NULL) |
2994 | ferr(po, "proto_parse failed for call '%s'\n", tmpname); |
da87ae38 |
2995 | |
7ae48d73 |
2996 | pp = proto_clone(pp_c); |
2997 | my_assert_not(pp, NULL); |
ddaf8bd7 |
2998 | } |
2999 | else if (po->datap != NULL) { |
3000 | pp = calloc(1, sizeof(*pp)); |
3001 | my_assert_not(pp, NULL); |
3002 | |
3003 | ret = parse_protostr(po->datap, pp); |
3004 | if (ret < 0) |
3005 | ferr(po, "bad protostr supplied: %s\n", (char *)po->datap); |
3006 | free(po->datap); |
3007 | po->datap = NULL; |
3008 | } |
3009 | |
3010 | if (pp != NULL) { |
3011 | if (pp->is_fptr) |
3012 | check_func_pp(po, pp, "fptr var call"); |
3013 | if (pp->is_noreturn) |
3014 | po->flags |= OPF_TAIL; |
7ae48d73 |
3015 | po->datap = pp; |
3016 | } |
3017 | continue; |
3018 | } |
3019 | |
3020 | if (!(po->flags & OPF_JMP) || po->op == OP_RET) |
87bf6cec |
3021 | continue; |
3022 | |
4c45fa73 |
3023 | if (po->operand[0].type == OPT_REGMEM) { |
3024 | char *p = strchr(po->operand[0].name, '['); |
3025 | if (p == NULL) |
89ff3147 |
3026 | goto tailcall; |
4c45fa73 |
3027 | ret = p - po->operand[0].name; |
3028 | strncpy(buf1, po->operand[0].name, ret); |
3029 | buf1[ret] = 0; |
3030 | |
3031 | for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) { |
3032 | if (IS(g_func_pd[j].label, buf1)) { |
3033 | pd = &g_func_pd[j]; |
3034 | break; |
3035 | } |
3036 | } |
3037 | if (pd == NULL) |
840257f6 |
3038 | //ferr(po, "label '%s' not parsed?\n", buf1); |
3039 | goto tailcall; |
4c45fa73 |
3040 | if (pd->type != OPT_OFFSET) |
3041 | ferr(po, "label '%s' with non-offset data?\n", buf1); |
3042 | |
3043 | // find all labels, link |
3044 | for (j = 0; j < pd->count; j++) { |
3045 | for (l = 0; l < opcnt; l++) { |
3046 | if (g_labels[l][0] && IS(g_labels[l], pd->d[j].u.label)) { |
3047 | add_label_ref(&g_label_refs[l], i); |
3048 | pd->d[j].bt_i = l; |
3049 | break; |
3050 | } |
3051 | } |
3052 | } |
3053 | |
3054 | po->btj = pd; |
3055 | continue; |
3056 | } |
3057 | |
3058 | for (l = 0; l < opcnt; l++) { |
3059 | if (g_labels[l][0] && IS(po->operand[0].name, g_labels[l])) { |
3060 | add_label_ref(&g_label_refs[l], i); |
3061 | po->bt_i = l; |
87bf6cec |
3062 | break; |
3063 | } |
3064 | } |
3065 | |
4c45fa73 |
3066 | if (po->bt_i != -1) |
3067 | continue; |
3068 | |
840257f6 |
3069 | if (po->operand[0].type == OPT_LABEL) |
87bf6cec |
3070 | // assume tail call |
840257f6 |
3071 | goto tailcall; |
4c45fa73 |
3072 | |
3073 | ferr(po, "unhandled branch\n"); |
840257f6 |
3074 | |
3075 | tailcall: |
3076 | po->op = OP_CALL; |
3077 | po->flags |= OPF_TAIL; |
7ae48d73 |
3078 | i--; // reprocess |
87bf6cec |
3079 | } |
3080 | |
3081 | // pass3: |
a2c1d768 |
3082 | // - remove dead labels |
91977a1c |
3083 | // - process calls |
1bafb621 |
3084 | for (i = 0; i < opcnt; i++) |
3085 | { |
a2c1d768 |
3086 | if (g_labels[i][0] != 0 && g_label_refs[i].i == -1) |
3087 | g_labels[i][0] = 0; |
3088 | |
69a3cdfc |
3089 | po = &ops[i]; |
3090 | if (po->flags & OPF_RMD) |
91977a1c |
3091 | continue; |
850c9265 |
3092 | |
d4e3b5db |
3093 | if (po->op == OP_CALL) |
69a3cdfc |
3094 | { |
1bafb621 |
3095 | tmpname = opr_name(po, 0); |
7ae48d73 |
3096 | pp = po->datap; |
3097 | if (pp == NULL) |
1bafb621 |
3098 | { |
7ae48d73 |
3099 | // indirect call |
89ff3147 |
3100 | pp_c = resolve_icall(i, opcnt, &l); |
3101 | if (pp_c != NULL) { |
1cd4a663 |
3102 | pp = proto_clone(pp_c); |
89ff3147 |
3103 | my_assert_not(pp, NULL); |
3104 | if (l) |
3105 | // not resolved just to single func |
3106 | pp->is_fptr = 1; |
3107 | } |
1cd4a663 |
3108 | if (pp == NULL) { |
3109 | pp = calloc(1, sizeof(*pp)); |
3110 | my_assert_not(pp, NULL); |
89ff3147 |
3111 | pp->is_fptr = 1; |
1cd4a663 |
3112 | ret = scan_for_esp_adjust(i + 1, opcnt, &j); |
89ff3147 |
3113 | if (ret < 0) { |
3114 | if (!g_allow_regfunc) |
3115 | ferr(po, "non-__cdecl indirect call unhandled yet\n"); |
3116 | pp->is_unresolved = 1; |
3117 | j = 0; |
3118 | } |
1cd4a663 |
3119 | j /= 4; |
3120 | if (j > ARRAY_SIZE(pp->arg)) |
89ff3147 |
3121 | ferr(po, "esp adjust too large: %d\n", j); |
1cd4a663 |
3122 | pp->ret_type.name = strdup("int"); |
3123 | pp->argc = pp->argc_stack = j; |
3124 | for (arg = 0; arg < pp->argc; arg++) |
3125 | pp->arg[arg].type.name = strdup("int"); |
3126 | } |
7ae48d73 |
3127 | po->datap = pp; |
1bafb621 |
3128 | } |
3129 | |
7ba45c34 |
3130 | // look for and make use of esp adjust |
3131 | ret = -1; |
e56ab892 |
3132 | if (!pp->is_stdcall && pp->argc_stack > 0) |
7ba45c34 |
3133 | ret = scan_for_esp_adjust(i + 1, opcnt, &j); |
1bafb621 |
3134 | if (ret >= 0) { |
7ba45c34 |
3135 | if (pp->is_vararg) { |
3136 | if (j / 4 < pp->argc_stack) |
3137 | ferr(po, "esp adjust is too small: %x < %x\n", |
3138 | j, pp->argc_stack * 4); |
3139 | // modify pp to make it have varargs as normal args |
3140 | arg = pp->argc; |
3141 | pp->argc += j / 4 - pp->argc_stack; |
3142 | for (; arg < pp->argc; arg++) { |
3ebea2cf |
3143 | pp->arg[arg].type.name = strdup("int"); |
7ba45c34 |
3144 | pp->argc_stack++; |
3145 | } |
3146 | if (pp->argc > ARRAY_SIZE(pp->arg)) |
e56ab892 |
3147 | ferr(po, "too many args for '%s'\n", tmpname); |
7ba45c34 |
3148 | } |
1bafb621 |
3149 | if (pp->argc_stack != j / 4) |
e56ab892 |
3150 | ferr(po, "stack tracking failed for '%s': %x %x\n", |
3151 | tmpname, pp->argc_stack * 4, j); |
7ba45c34 |
3152 | |
1bafb621 |
3153 | ops[ret].flags |= OPF_RMD; |
591721d7 |
3154 | if (ops[ret].op == OP_POP && j > 4) { |
3155 | // deal with multi-pop stack adjust |
3156 | j = pp->argc_stack; |
3157 | while (ops[ret].op == OP_POP && j > 0 && ret < opcnt) { |
3158 | ops[ret].flags |= OPF_RMD; |
3159 | j--; |
3160 | ret++; |
3161 | } |
3162 | } |
3163 | else { |
3164 | // a bit of a hack, but deals with use of |
3165 | // single adj for multiple calls |
3166 | ops[ret].operand[1].val -= j; |
3167 | } |
1bafb621 |
3168 | } |
7ba45c34 |
3169 | else if (pp->is_vararg) |
3170 | ferr(po, "missing esp_adjust for vararg func '%s'\n", |
3171 | pp->name); |
91977a1c |
3172 | |
89ff3147 |
3173 | if (!pp->is_unresolved) { |
3174 | // since we know the args, collect them |
3175 | collect_call_args(po, i, pp, ®mask, &save_arg_vars, |
3176 | i + opcnt * 2); |
3177 | } |
2b43685d |
3178 | |
3179 | if (strstr(pp->ret_type.name, "int64")) |
3180 | need_mul_var = 1; |
91977a1c |
3181 | } |
d4e3b5db |
3182 | } |
3183 | |
3184 | // pass4: |
3185 | // - find POPs for PUSHes, rm both |
591721d7 |
3186 | // - scan for STD/CLD, propagate DF |
d4e3b5db |
3187 | // - scan for all used registers |
3188 | // - find flag set ops for their users |
89ff3147 |
3189 | // - do unreselved calls |
1bafb621 |
3190 | // - declare indirect functions |
d4e3b5db |
3191 | for (i = 0; i < opcnt; i++) { |
3192 | po = &ops[i]; |
3193 | if (po->flags & OPF_RMD) |
3194 | continue; |
3195 | |
7ae48d73 |
3196 | if (po->op == OP_PUSH && (po->flags & OPF_RSAVE)) { |
3197 | reg = po->operand[0].reg; |
3198 | if (!(regmask & (1 << reg))) |
3199 | // not a reg save after all, rerun scan_for_pop |
3200 | po->flags &= ~OPF_RSAVE; |
3201 | else |
3202 | regmask_save |= 1 << reg; |
3203 | } |
3204 | |
5c024ef7 |
3205 | if (po->op == OP_PUSH && po->argnum == 0 |
3206 | && !(po->flags & OPF_RSAVE)) |
d4e3b5db |
3207 | { |
5c024ef7 |
3208 | if (po->operand[0].type == OPT_REG) |
3209 | { |
3210 | reg = po->operand[0].reg; |
3211 | if (reg < 0) |
3212 | ferr(po, "reg not set for push?\n"); |
3213 | |
3214 | depth = 0; |
3215 | ret = scan_for_pop(i + 1, opcnt, |
3216 | po->operand[0].name, i + opcnt * 3, 0, &depth, 0); |
3217 | if (ret == 1) { |
3218 | if (depth > 1) |
3219 | ferr(po, "too much depth: %d\n", depth); |
3220 | |
3221 | po->flags |= OPF_RMD; |
3222 | scan_for_pop(i + 1, opcnt, po->operand[0].name, |
3223 | i + opcnt * 4, 0, &depth, 1); |
3224 | continue; |
3225 | } |
3226 | ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0); |
3227 | if (ret == 0) { |
3228 | arg = OPF_RMD; |
3229 | if (regmask & (1 << reg)) { |
3230 | if (regmask_save & (1 << reg)) |
3231 | ferr(po, "%s already saved?\n", po->operand[0].name); |
3232 | arg = OPF_RSAVE; |
3233 | } |
3234 | po->flags |= arg; |
3235 | scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg); |
3236 | continue; |
3237 | } |
d4e3b5db |
3238 | } |
5c024ef7 |
3239 | else if (po->operand[0].type == OPT_CONST) { |
3240 | for (j = i + 1; j < opcnt; j++) { |
3241 | if ((ops[j].flags & (OPF_JMP|OPF_TAIL|OPF_RSAVE)) |
3242 | || ops[j].op == OP_PUSH || g_labels[i][0] != 0) |
3243 | { |
3244 | break; |
3245 | } |
3246 | |
3247 | if (!(ops[j].flags & OPF_RMD) && ops[j].op == OP_POP) |
3248 | { |
3249 | po->flags |= OPF_RMD; |
3250 | ops[j].datap = po; |
3251 | break; |
3252 | } |
d4e3b5db |
3253 | } |
d4e3b5db |
3254 | } |
3255 | } |
3256 | |
591721d7 |
3257 | if (po->op == OP_STD) { |
3258 | po->flags |= OPF_DF | OPF_RMD; |
3259 | scan_propagate_df(i + 1, opcnt); |
3260 | } |
3261 | |
a3684be1 |
3262 | regmask_now = po->regmask_src | po->regmask_dst; |
3263 | if (regmask_now & (1 << xBP)) { |
3264 | if (g_bp_frame && !(po->flags & OPF_EBP_S)) { |
3265 | if (po->regmask_dst & (1 << xBP)) |
3266 | // compiler decided to drop bp frame and use ebp as scratch |
3267 | scan_fwd_set_flags(i, opcnt, i + opcnt * 5, OPF_EBP_S); |
3268 | else |
3269 | regmask_now &= ~(1 << xBP); |
3270 | } |
3271 | } |
3272 | |
3273 | regmask |= regmask_now; |
d4e3b5db |
3274 | |
3275 | if (po->flags & OPF_CC) |
3276 | { |
2b43685d |
3277 | int setters[16], cnt = 0, branched = 0; |
3278 | |
04f8a628 |
3279 | ret = scan_for_flag_set(i, i + opcnt * 6, |
3280 | &branched, setters, &cnt); |
2b43685d |
3281 | if (ret < 0 || cnt <= 0) |
3282 | ferr(po, "unable to trace flag setter(s)\n"); |
3283 | if (cnt > ARRAY_SIZE(setters)) |
3284 | ferr(po, "too many flag setters\n"); |
d4e3b5db |
3285 | |
d4e3b5db |
3286 | pfo = split_cond(po, po->op, &dummy); |
2b43685d |
3287 | for (j = 0; j < cnt; j++) |
3288 | { |
3289 | tmp_op = &ops[setters[j]]; // flag setter |
3290 | pfomask = 0; |
3291 | |
3292 | // to get nicer code, we try to delay test and cmp; |
3293 | // if we can't because of operand modification, or if we |
591721d7 |
3294 | // have arith op, or branch, make it calculate flags explicitly |
3295 | if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP) |
3296 | { |
89ff3147 |
3297 | if (branched || scan_for_mod(tmp_op, setters[j] + 1, i, 0) >= 0) |
2b43685d |
3298 | pfomask = 1 << pfo; |
3299 | } |
591721d7 |
3300 | else if (tmp_op->op == OP_SCAS) { |
3301 | pfomask = 1 << pfo; |
3302 | } |
2b43685d |
3303 | else if (tmp_op->op == OP_CMPS) { |
3304 | pfomask = 1 << PFO_Z; |
3305 | } |
3306 | else { |
04f8a628 |
3307 | // see if we'll be able to handle based on op result |
3308 | if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR |
3309 | && pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P) |
3310 | || branched |
2b43685d |
3311 | || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0) |
3312 | pfomask = 1 << pfo; |
3313 | } |
3314 | if (pfomask) { |
3315 | tmp_op->pfomask |= pfomask; |
3316 | cmp_result_vars |= pfomask; |
2b43685d |
3317 | } |
04f8a628 |
3318 | // note: may overwrite, currently not a problem |
3319 | po->datap = tmp_op; |
d4e3b5db |
3320 | } |
3321 | |
3322 | if (po->op == OP_ADC || po->op == OP_SBB) |
3323 | cmp_result_vars |= 1 << PFO_C; |
3324 | } |
591721d7 |
3325 | else if (po->op == OP_CMPS || po->op == OP_SCAS) { |
3326 | cmp_result_vars |= 1 << PFO_Z; |
3327 | } |
87bf6cec |
3328 | else if (po->op == OP_MUL |
3329 | || (po->op == OP_IMUL && po->operand_cnt == 1)) |
3330 | { |
3331 | need_mul_var = 1; |
3332 | } |
89ff3147 |
3333 | else if (po->op == OP_CALL) { |
1bafb621 |
3334 | pp = po->datap; |
89ff3147 |
3335 | if (pp == NULL) |
3336 | ferr(po, "NULL pp\n"); |
3337 | |
3338 | if (pp->is_unresolved) { |
ddaf8bd7 |
3339 | int regmask_stack = 0; |
b74c31e3 |
3340 | collect_call_args(po, i, pp, ®mask, &save_arg_vars, |
89ff3147 |
3341 | i + opcnt * 2); |
3342 | |
b74c31e3 |
3343 | // this is pretty rough guess: |
3344 | // see ecx and edx were pushed (and not their saved versions) |
3345 | for (arg = 0; arg < pp->argc; arg++) { |
3346 | if (pp->arg[arg].reg != NULL) |
3347 | continue; |
3348 | |
3349 | tmp_op = pp->arg[arg].datap; |
3350 | if (tmp_op == NULL) |
3351 | ferr(po, "parsed_op missing for arg%d\n", arg); |
3352 | if (tmp_op->argnum == 0 && tmp_op->operand[0].type == OPT_REG) |
3353 | regmask_stack |= 1 << tmp_op->operand[0].reg; |
3354 | } |
3355 | |
ddaf8bd7 |
3356 | if (!((regmask_stack & (1 << xCX)) |
3357 | && (regmask_stack & (1 << xDX)))) |
89ff3147 |
3358 | { |
3359 | if (pp->argc_stack != 0 |
c0050df6 |
3360 | || ((regmask | regmask_arg) & ((1 << xCX)|(1 << xDX)))) |
89ff3147 |
3361 | { |
3362 | pp_insert_reg_arg(pp, "ecx"); |
c0050df6 |
3363 | pp->is_fastcall = 1; |
ddaf8bd7 |
3364 | regmask_init |= 1 << xCX; |
89ff3147 |
3365 | regmask |= 1 << xCX; |
3366 | } |
3367 | if (pp->argc_stack != 0 |
3368 | || ((regmask | regmask_arg) & (1 << xDX))) |
3369 | { |
3370 | pp_insert_reg_arg(pp, "edx"); |
ddaf8bd7 |
3371 | regmask_init |= 1 << xDX; |
89ff3147 |
3372 | regmask |= 1 << xDX; |
3373 | } |
3374 | } |
c0050df6 |
3375 | |
3376 | // note: __cdecl doesn't fall into is_unresolved category |
3377 | if (pp->argc_stack > 0) |
3378 | pp->is_stdcall = 1; |
ddaf8bd7 |
3379 | } |
3380 | |
3381 | for (arg = 0; arg < pp->argc; arg++) { |
3382 | if (pp->arg[arg].reg != NULL) { |
3383 | reg = char_array_i(regs_r32, |
3384 | ARRAY_SIZE(regs_r32), pp->arg[arg].reg); |
3385 | if (reg < 0) |
3386 | ferr(ops, "arg '%s' is not a reg?\n", pp->arg[arg].reg); |
3387 | if (!(regmask & (1 << reg))) { |
3388 | regmask_init |= 1 << reg; |
3389 | regmask |= 1 << reg; |
3390 | } |
3391 | } |
89ff3147 |
3392 | } |
3393 | |
3394 | if (pp->is_fptr) { |
ddaf8bd7 |
3395 | if (pp->name[0] != 0) { |
3396 | memmove(pp->name + 2, pp->name, strlen(pp->name) + 1); |
3397 | memcpy(pp->name, "i_", 2); |
3398 | |
3399 | // might be declared already |
3400 | found = 0; |
3401 | for (j = 0; j < i; j++) { |
3402 | if (ops[j].op == OP_CALL && (pp_tmp = ops[j].datap)) { |
3403 | if (pp_tmp->is_fptr && IS(pp->name, pp_tmp->name)) { |
3404 | found = 1; |
3405 | break; |
3406 | } |
3407 | } |
3408 | } |
3409 | if (found) |
3410 | continue; |
3411 | } |
3412 | else |
3413 | snprintf(pp->name, sizeof(pp->name), "icall%d", i); |
3414 | |
89ff3147 |
3415 | fprintf(fout, " %s (", pp->ret_type.name); |
c0050df6 |
3416 | output_pp_attrs(fout, pp, 0); |
ddaf8bd7 |
3417 | fprintf(fout, "*%s)(", pp->name); |
89ff3147 |
3418 | for (j = 0; j < pp->argc; j++) { |
3419 | if (j > 0) |
3420 | fprintf(fout, ", "); |
3421 | fprintf(fout, "%s a%d", pp->arg[j].type.name, j + 1); |
3422 | } |
3423 | fprintf(fout, ");\n"); |
1bafb621 |
3424 | } |
1bafb621 |
3425 | } |
75ad0378 |
3426 | else if (po->op == OP_RET && !IS(g_func_pp->ret_type.name, "void")) |
3427 | regmask |= 1 << xAX; |
91977a1c |
3428 | } |
3429 | |
da87ae38 |
3430 | // pass4: |
3431 | // - confirm regmask_save, it might have been reduced |
3432 | if (regmask_save != 0) |
3433 | { |
3434 | regmask_save = 0; |
3435 | for (i = 0; i < opcnt; i++) { |
3436 | po = &ops[i]; |
3437 | if (po->flags & OPF_RMD) |
3438 | continue; |
3439 | |
3440 | if (po->op == OP_PUSH && (po->flags & OPF_RSAVE)) |
3441 | regmask_save |= 1 << po->operand[0].reg; |
3442 | } |
3443 | } |
3444 | |
3445 | |
4c45fa73 |
3446 | // output LUTs/jumptables |
3447 | for (i = 0; i < g_func_pd_cnt; i++) { |
3448 | pd = &g_func_pd[i]; |
3449 | fprintf(fout, " static const "); |
3450 | if (pd->type == OPT_OFFSET) { |
3451 | fprintf(fout, "void *jt_%s[] =\n { ", pd->label); |
3452 | |
3453 | for (j = 0; j < pd->count; j++) { |
3454 | if (j > 0) |
3455 | fprintf(fout, ", "); |
3456 | fprintf(fout, "&&%s", pd->d[j].u.label); |
3457 | } |
3458 | } |
3459 | else { |
3460 | fprintf(fout, "%s %s[] =\n { ", |
3461 | lmod_type_u(ops, pd->lmod), pd->label); |
3462 | |
3463 | for (j = 0; j < pd->count; j++) { |
3464 | if (j > 0) |
3465 | fprintf(fout, ", "); |
3466 | fprintf(fout, "%u", pd->d[j].u.val); |
3467 | } |
3468 | } |
3469 | fprintf(fout, " };\n"); |
3470 | } |
3471 | |
4f12f671 |
3472 | // declare stack frame, va_arg |
1bafb621 |
3473 | if (g_stack_fsz) |
850c9265 |
3474 | fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n", |
1bafb621 |
3475 | (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz); |
850c9265 |
3476 | |
bd96f656 |
3477 | if (g_func_pp->is_vararg) |
4f12f671 |
3478 | fprintf(fout, " va_list ap;\n"); |
3479 | |
940e8e66 |
3480 | // declare arg-registers |
bd96f656 |
3481 | for (i = 0; i < g_func_pp->argc; i++) { |
3482 | if (g_func_pp->arg[i].reg != NULL) { |
91977a1c |
3483 | reg = char_array_i(regs_r32, |
bd96f656 |
3484 | ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg); |
75ad0378 |
3485 | if (regmask & (1 << reg)) { |
3486 | fprintf(fout, " u32 %s = (u32)a%d;\n", |
3487 | g_func_pp->arg[i].reg, i + 1); |
3488 | } |
3489 | else |
3490 | fprintf(fout, " // %s = a%d; // unused\n", |
3491 | g_func_pp->arg[i].reg, i + 1); |
91977a1c |
3492 | had_decl = 1; |
3493 | } |
3494 | } |
3495 | |
75ad0378 |
3496 | regmask_now = regmask & ~regmask_arg; |
3497 | regmask_now &= ~(1 << xSP); |
3498 | if (regmask_now) { |
91977a1c |
3499 | for (reg = 0; reg < 8; reg++) { |
75ad0378 |
3500 | if (regmask_now & (1 << reg)) { |
ddaf8bd7 |
3501 | fprintf(fout, " u32 %s", regs_r32[reg]); |
3502 | if (regmask_init & (1 << reg)) |
3503 | fprintf(fout, " = 0"); |
3504 | fprintf(fout, ";\n"); |
91977a1c |
3505 | had_decl = 1; |
3506 | } |
3507 | } |
3508 | } |
3509 | |
d4e3b5db |
3510 | if (regmask_save) { |
3511 | for (reg = 0; reg < 8; reg++) { |
3512 | if (regmask_save & (1 << reg)) { |
3513 | fprintf(fout, " u32 s_%s;\n", regs_r32[reg]); |
3514 | had_decl = 1; |
3515 | } |
3516 | } |
3517 | } |
3518 | |
69a3cdfc |
3519 | if (save_arg_vars) { |
3520 | for (reg = 0; reg < 32; reg++) { |
3521 | if (save_arg_vars & (1 << reg)) { |
3522 | fprintf(fout, " u32 s_a%d;\n", reg + 1); |
3523 | had_decl = 1; |
3524 | } |
3525 | } |
3526 | } |
3527 | |
3528 | if (cmp_result_vars) { |
3529 | for (i = 0; i < 8; i++) { |
3530 | if (cmp_result_vars & (1 << i)) { |
3531 | fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]); |
3532 | had_decl = 1; |
3533 | } |
3534 | } |
3535 | } |
3536 | |
87bf6cec |
3537 | if (need_mul_var) { |
3538 | fprintf(fout, " u64 mul_tmp;\n"); |
3539 | had_decl = 1; |
3540 | } |
3541 | |
91977a1c |
3542 | if (had_decl) |
3543 | fprintf(fout, "\n"); |
3544 | |
bd96f656 |
3545 | if (g_func_pp->is_vararg) { |
3546 | if (g_func_pp->argc_stack == 0) |
4f12f671 |
3547 | ferr(ops, "vararg func without stack args?\n"); |
bd96f656 |
3548 | fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc); |
4f12f671 |
3549 | } |
3550 | |
91977a1c |
3551 | // output ops |
69a3cdfc |
3552 | for (i = 0; i < opcnt; i++) |
3553 | { |
a2c1d768 |
3554 | if (g_labels[i][0] != 0) { |
91977a1c |
3555 | fprintf(fout, "\n%s:\n", g_labels[i]); |
3ebea2cf |
3556 | label_pending = 1; |
2b43685d |
3557 | |
3558 | delayed_flag_op = NULL; |
3559 | last_arith_dst = NULL; |
3ebea2cf |
3560 | } |
91977a1c |
3561 | |
69a3cdfc |
3562 | po = &ops[i]; |
3563 | if (po->flags & OPF_RMD) |
91977a1c |
3564 | continue; |
3565 | |
3566 | no_output = 0; |
3567 | |
91977a1c |
3568 | #define assert_operand_cnt(n_) \ |
850c9265 |
3569 | if (po->operand_cnt != n_) \ |
3570 | ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_) |
3571 | |
69a3cdfc |
3572 | // conditional/flag using op? |
3573 | if (po->flags & OPF_CC) |
850c9265 |
3574 | { |
940e8e66 |
3575 | int is_delayed = 0; |
3576 | int is_inv = 0; |
69a3cdfc |
3577 | |
940e8e66 |
3578 | pfo = split_cond(po, po->op, &is_inv); |
04f8a628 |
3579 | tmp_op = po->datap; |
850c9265 |
3580 | |
69a3cdfc |
3581 | // we go through all this trouble to avoid using parsed_flag_op, |
3582 | // which makes generated code much nicer |
3583 | if (delayed_flag_op != NULL) |
850c9265 |
3584 | { |
940e8e66 |
3585 | out_cmp_test(buf1, sizeof(buf1), delayed_flag_op, pfo, is_inv); |
3586 | is_delayed = 1; |
91977a1c |
3587 | } |
850c9265 |
3588 | else if (last_arith_dst != NULL |
04f8a628 |
3589 | && (pfo == PFO_Z || pfo == PFO_S || pfo == PFO_P |
3590 | || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR)) |
3591 | )) |
850c9265 |
3592 | { |
3ebea2cf |
3593 | out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst); |
940e8e66 |
3594 | out_test_for_cc(buf1, sizeof(buf1), po, pfo, is_inv, |
850c9265 |
3595 | last_arith_dst->lmod, buf3); |
940e8e66 |
3596 | is_delayed = 1; |
850c9265 |
3597 | } |
04f8a628 |
3598 | else if (tmp_op != NULL) { |
7ba45c34 |
3599 | // use preprocessed flag calc results |
04f8a628 |
3600 | if (!(tmp_op->pfomask & (1 << pfo))) |
69a3cdfc |
3601 | ferr(po, "not prepared for pfo %d\n", pfo); |
3602 | |
940e8e66 |
3603 | // note: is_inv was not yet applied |
69a3cdfc |
3604 | snprintf(buf1, sizeof(buf1), "(%scond_%s)", |
940e8e66 |
3605 | is_inv ? "!" : "", parsed_flag_op_names[pfo]); |
69a3cdfc |
3606 | } |
3607 | else { |
3608 | ferr(po, "all methods of finding comparison failed\n"); |
3609 | } |
850c9265 |
3610 | |
69a3cdfc |
3611 | if (po->flags & OPF_JMP) { |
850c9265 |
3612 | fprintf(fout, " if %s\n", buf1); |
3613 | } |
5101a5f9 |
3614 | else if (po->op == OP_ADC || po->op == OP_SBB) { |
940e8e66 |
3615 | if (is_delayed) |
3616 | fprintf(fout, " cond_%s = %s;\n", |
3617 | parsed_flag_op_names[pfo], buf1); |
850c9265 |
3618 | } |
5101a5f9 |
3619 | else if (po->flags & OPF_DATA) { // SETcc |
850c9265 |
3620 | out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]); |
3621 | fprintf(fout, " %s = %s;", buf2, buf1); |
91977a1c |
3622 | } |
69a3cdfc |
3623 | else { |
3624 | ferr(po, "unhandled conditional op\n"); |
3625 | } |
91977a1c |
3626 | } |
3627 | |
940e8e66 |
3628 | pfomask = po->pfomask; |
3629 | |
850c9265 |
3630 | switch (po->op) |
91977a1c |
3631 | { |
3632 | case OP_MOV: |
3633 | assert_operand_cnt(2); |
850c9265 |
3634 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
de50b98b |
3635 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
3636 | fprintf(fout, " %s = %s;", buf1, |
3ebea2cf |
3637 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], |
3638 | po->operand[0].is_ptr ? "(void *)" : "", 0)); |
850c9265 |
3639 | break; |
3640 | |
3641 | case OP_LEA: |
3642 | assert_operand_cnt(2); |
87bf6cec |
3643 | po->operand[1].lmod = OPLM_DWORD; // always |
850c9265 |
3644 | fprintf(fout, " %s = %s;", |
3645 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
3ebea2cf |
3646 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], |
3647 | NULL, 1)); |
850c9265 |
3648 | break; |
3649 | |
3650 | case OP_MOVZX: |
3651 | assert_operand_cnt(2); |
91977a1c |
3652 | fprintf(fout, " %s = %s;", |
850c9265 |
3653 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
3ebea2cf |
3654 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1])); |
850c9265 |
3655 | break; |
3656 | |
3657 | case OP_MOVSX: |
3658 | assert_operand_cnt(2); |
3659 | switch (po->operand[1].lmod) { |
3660 | case OPLM_BYTE: |
3661 | strcpy(buf3, "(s8)"); |
3662 | break; |
3663 | case OPLM_WORD: |
3664 | strcpy(buf3, "(s16)"); |
3665 | break; |
3666 | default: |
3667 | ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod); |
3668 | } |
a2c1d768 |
3669 | fprintf(fout, " %s = %s;", |
850c9265 |
3670 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
a2c1d768 |
3671 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], |
3672 | buf3, 0)); |
850c9265 |
3673 | break; |
3674 | |
3675 | case OP_NOT: |
3676 | assert_operand_cnt(1); |
3677 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
3678 | fprintf(fout, " %s = ~%s;", buf1, buf1); |
3679 | break; |
3680 | |
5101a5f9 |
3681 | case OP_CDQ: |
3682 | assert_operand_cnt(2); |
3683 | fprintf(fout, " %s = (s32)%s >> 31;", |
3684 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
3ebea2cf |
3685 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1])); |
5101a5f9 |
3686 | strcpy(g_comment, "cdq"); |
3687 | break; |
3688 | |
33c35af6 |
3689 | case OP_STOS: |
33c35af6 |
3690 | assert_operand_cnt(3); |
3691 | if (po->flags & OPF_REP) { |
591721d7 |
3692 | fprintf(fout, " for (; ecx != 0; ecx--, edi %c= %d)\n", |
3693 | (po->flags & OPF_DF) ? '-' : '+', |
33c35af6 |
3694 | lmod_bytes(po, po->operand[0].lmod)); |
d4e3b5db |
3695 | fprintf(fout, " %sedi = eax;", |
3696 | lmod_cast_u_ptr(po, po->operand[0].lmod)); |
33c35af6 |
3697 | strcpy(g_comment, "rep stos"); |
3698 | } |
3699 | else { |
591721d7 |
3700 | fprintf(fout, " %sedi = eax; edi %c= %d;", |
d4e3b5db |
3701 | lmod_cast_u_ptr(po, po->operand[0].lmod), |
591721d7 |
3702 | (po->flags & OPF_DF) ? '-' : '+', |
33c35af6 |
3703 | lmod_bytes(po, po->operand[0].lmod)); |
3704 | strcpy(g_comment, "stos"); |
3705 | } |
3706 | break; |
3707 | |
d4e3b5db |
3708 | case OP_MOVS: |
d4e3b5db |
3709 | assert_operand_cnt(3); |
3710 | j = lmod_bytes(po, po->operand[0].lmod); |
3711 | strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod)); |
591721d7 |
3712 | l = (po->flags & OPF_DF) ? '-' : '+'; |
d4e3b5db |
3713 | if (po->flags & OPF_REP) { |
3714 | fprintf(fout, |
591721d7 |
3715 | " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n", |
3716 | l, j, l, j); |
d4e3b5db |
3717 | fprintf(fout, |
3718 | " %sedi = %sesi;", buf1, buf1); |
3719 | strcpy(g_comment, "rep movs"); |
3720 | } |
3721 | else { |
591721d7 |
3722 | fprintf(fout, " %sedi = %sesi; edi %c= %d; esi %c= %d;", |
3723 | buf1, buf1, l, j, l, j); |
d4e3b5db |
3724 | strcpy(g_comment, "movs"); |
3725 | } |
3726 | break; |
3727 | |
7ba45c34 |
3728 | case OP_CMPS: |
7ba45c34 |
3729 | // repe ~ repeat while ZF=1 |
3730 | assert_operand_cnt(3); |
3731 | j = lmod_bytes(po, po->operand[0].lmod); |
3732 | strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod)); |
591721d7 |
3733 | l = (po->flags & OPF_DF) ? '-' : '+'; |
7ba45c34 |
3734 | if (po->flags & OPF_REP) { |
3735 | fprintf(fout, |
591721d7 |
3736 | " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n", |
3737 | l, j, l, j); |
7ba45c34 |
3738 | fprintf(fout, |
3739 | " if ((cond_z = (%sedi == %sesi)) %s 0)\n", |
3740 | buf1, buf1, (po->flags & OPF_REPZ) ? "==" : "!="); |
3741 | fprintf(fout, |
3742 | " break;"); |
3743 | snprintf(g_comment, sizeof(g_comment), "rep%s cmps", |
3744 | (po->flags & OPF_REPZ) ? "e" : "ne"); |
3745 | } |
3746 | else { |
3747 | fprintf(fout, |
591721d7 |
3748 | " cond_z = (%sedi = %sesi); edi %c= %d; esi %c= %d;", |
3749 | buf1, buf1, l, j, l, j); |
7ba45c34 |
3750 | strcpy(g_comment, "cmps"); |
3751 | } |
3752 | pfomask &= ~(1 << PFO_Z); |
3753 | last_arith_dst = NULL; |
3754 | delayed_flag_op = NULL; |
3755 | break; |
3756 | |
591721d7 |
3757 | case OP_SCAS: |
3758 | // only does ZF (for now) |
3759 | // repe ~ repeat while ZF=1 |
3760 | assert_operand_cnt(3); |
3761 | j = lmod_bytes(po, po->operand[0].lmod); |
3762 | l = (po->flags & OPF_DF) ? '-' : '+'; |
3763 | if (po->flags & OPF_REP) { |
3764 | fprintf(fout, |
3765 | " for (; ecx != 0; ecx--, edi %c= %d)\n", l, j); |
3766 | fprintf(fout, |
3767 | " if ((cond_z = (%seax == %sedi)) %s 0)\n", |
3768 | lmod_cast_u(po, po->operand[0].lmod), |
3769 | lmod_cast_u_ptr(po, po->operand[0].lmod), |
3770 | (po->flags & OPF_REPZ) ? "==" : "!="); |
3771 | fprintf(fout, |
3772 | " break;"); |
3773 | snprintf(g_comment, sizeof(g_comment), "rep%s scas", |
3774 | (po->flags & OPF_REPZ) ? "e" : "ne"); |
3775 | } |
3776 | else { |
3777 | fprintf(fout, " cond_z = (%seax = %sedi); edi %c= %d;", |
3778 | lmod_cast_u(po, po->operand[0].lmod), |
3779 | lmod_cast_u_ptr(po, po->operand[0].lmod), l, j); |
3780 | strcpy(g_comment, "scas"); |
3781 | } |
3782 | pfomask &= ~(1 << PFO_Z); |
3783 | last_arith_dst = NULL; |
3784 | delayed_flag_op = NULL; |
3785 | break; |
3786 | |
850c9265 |
3787 | // arithmetic w/flags |
3788 | case OP_ADD: |
3789 | case OP_SUB: |
3790 | case OP_AND: |
3791 | case OP_OR: |
5101a5f9 |
3792 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
3793 | // fallthrough |
850c9265 |
3794 | dualop_arith: |
3795 | assert_operand_cnt(2); |
850c9265 |
3796 | fprintf(fout, " %s %s= %s;", |
3797 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
3798 | op_to_c(po), |
3ebea2cf |
3799 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1])); |
04f8a628 |
3800 | output_std_flags(fout, po, &pfomask, buf1); |
3801 | last_arith_dst = &po->operand[0]; |
3802 | delayed_flag_op = NULL; |
3803 | break; |
3804 | |
3805 | case OP_SHL: |
3806 | case OP_SHR: |
3807 | assert_operand_cnt(2); |
3808 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
3809 | if (pfomask & (1 << PFO_C)) { |
3810 | if (po->operand[1].type == OPT_CONST) { |
3811 | l = lmod_bytes(po, po->operand[0].lmod) * 8; |
3812 | j = po->operand[1].val; |
3813 | j %= l; |
3814 | if (j != 0) { |
3815 | if (po->op == OP_SHL) |
3816 | j = l - j; |
3817 | else |
3818 | j -= 1; |
3819 | fprintf(fout, " cond_c = (%s & 0x%02x) ? 1 : 0;\n", |
3820 | buf1, 1 << j); |
3821 | } |
3822 | else |
3823 | ferr(po, "zero shift?\n"); |
3824 | } |
3825 | else |
3826 | ferr(po, "TODO\n"); |
3827 | pfomask &= ~(1 << PFO_C); |
840257f6 |
3828 | } |
04f8a628 |
3829 | fprintf(fout, " %s %s= %s;", buf1, op_to_c(po), |
3830 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1])); |
3831 | output_std_flags(fout, po, &pfomask, buf1); |
850c9265 |
3832 | last_arith_dst = &po->operand[0]; |
69a3cdfc |
3833 | delayed_flag_op = NULL; |
850c9265 |
3834 | break; |
3835 | |
d4e3b5db |
3836 | case OP_SAR: |
3837 | assert_operand_cnt(2); |
3838 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
3839 | fprintf(fout, " %s = %s%s >> %s;", buf1, |
3840 | lmod_cast_s(po, po->operand[0].lmod), buf1, |
3ebea2cf |
3841 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1])); |
04f8a628 |
3842 | output_std_flags(fout, po, &pfomask, buf1); |
d4e3b5db |
3843 | last_arith_dst = &po->operand[0]; |
3844 | delayed_flag_op = NULL; |
3845 | break; |
3846 | |
3847 | case OP_ROL: |
3848 | case OP_ROR: |
3849 | assert_operand_cnt(2); |
3850 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
3851 | if (po->operand[1].type == OPT_CONST) { |
3852 | j = po->operand[1].val; |
3853 | j %= lmod_bytes(po, po->operand[0].lmod) * 8; |
3854 | fprintf(fout, po->op == OP_ROL ? |
3855 | " %s = (%s << %d) | (%s >> %d);" : |
3856 | " %s = (%s >> %d) | (%s << %d);", |
3857 | buf1, buf1, j, buf1, |
3858 | lmod_bytes(po, po->operand[0].lmod) * 8 - j); |
3859 | } |
3860 | else |
3861 | ferr(po, "TODO\n"); |
04f8a628 |
3862 | output_std_flags(fout, po, &pfomask, buf1); |
d4e3b5db |
3863 | last_arith_dst = &po->operand[0]; |
3864 | delayed_flag_op = NULL; |
3865 | break; |
3866 | |
5101a5f9 |
3867 | case OP_XOR: |
850c9265 |
3868 | assert_operand_cnt(2); |
3869 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
5101a5f9 |
3870 | if (IS(opr_name(po, 0), opr_name(po, 1))) { |
3871 | // special case for XOR |
3872 | fprintf(fout, " %s = 0;", |
3873 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0])); |
3874 | last_arith_dst = &po->operand[0]; |
3875 | delayed_flag_op = NULL; |
850c9265 |
3876 | break; |
850c9265 |
3877 | } |
5101a5f9 |
3878 | goto dualop_arith; |
3879 | |
5101a5f9 |
3880 | case OP_ADC: |
850c9265 |
3881 | case OP_SBB: |
5101a5f9 |
3882 | assert_operand_cnt(2); |
3883 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
a2c1d768 |
3884 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
840257f6 |
3885 | if (po->op == OP_SBB |
3886 | && IS(po->operand[0].name, po->operand[1].name)) |
3887 | { |
3888 | // avoid use of unitialized var |
a2c1d768 |
3889 | fprintf(fout, " %s = -cond_c;", buf1); |
94d447fb |
3890 | // carry remains what it was |
3891 | pfomask &= ~(1 << PFO_C); |
840257f6 |
3892 | } |
3893 | else { |
a2c1d768 |
3894 | fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po), |
3ebea2cf |
3895 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1])); |
840257f6 |
3896 | } |
a2c1d768 |
3897 | output_std_flags(fout, po, &pfomask, buf1); |
5101a5f9 |
3898 | last_arith_dst = &po->operand[0]; |
3899 | delayed_flag_op = NULL; |
850c9265 |
3900 | break; |
3901 | |
3902 | case OP_INC: |
3903 | case OP_DEC: |
3904 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
1bafb621 |
3905 | if (po->operand[0].type == OPT_REG) { |
3906 | strcpy(buf2, po->op == OP_INC ? "++" : "--"); |
3907 | fprintf(fout, " %s%s;", buf1, buf2); |
3908 | } |
3909 | else { |
3910 | strcpy(buf2, po->op == OP_INC ? "+" : "-"); |
3911 | fprintf(fout, " %s %s= 1;", buf1, buf2); |
3912 | } |
a2c1d768 |
3913 | output_std_flags(fout, po, &pfomask, buf1); |
5101a5f9 |
3914 | last_arith_dst = &po->operand[0]; |
3915 | delayed_flag_op = NULL; |
3916 | break; |
3917 | |
3918 | case OP_NEG: |
3919 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
3ebea2cf |
3920 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]); |
5101a5f9 |
3921 | fprintf(fout, " %s = -%s%s;", buf1, |
3922 | lmod_cast_s(po, po->operand[0].lmod), buf2); |
850c9265 |
3923 | last_arith_dst = &po->operand[0]; |
69a3cdfc |
3924 | delayed_flag_op = NULL; |
940e8e66 |
3925 | if (pfomask & (1 << PFO_C)) { |
3926 | fprintf(fout, "\n cond_c = (%s != 0);", buf1); |
3927 | pfomask &= ~(1 << PFO_C); |
3928 | } |
850c9265 |
3929 | break; |
3930 | |
3931 | case OP_IMUL: |
de50b98b |
3932 | if (po->operand_cnt == 2) { |
3933 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
850c9265 |
3934 | goto dualop_arith; |
de50b98b |
3935 | } |
87bf6cec |
3936 | if (po->operand_cnt == 3) |
3937 | ferr(po, "TODO imul3\n"); |
3938 | // fallthrough |
3939 | case OP_MUL: |
3940 | assert_operand_cnt(1); |
3941 | strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)"); |
3942 | fprintf(fout, " mul_tmp = %seax * %s%s;\n", buf1, buf1, |
3ebea2cf |
3943 | out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0])); |
87bf6cec |
3944 | fprintf(fout, " edx = mul_tmp >> 32;\n"); |
3945 | fprintf(fout, " eax = mul_tmp;"); |
3946 | last_arith_dst = NULL; |
69a3cdfc |
3947 | delayed_flag_op = NULL; |
91977a1c |
3948 | break; |
3949 | |
5101a5f9 |
3950 | case OP_DIV: |
3951 | case OP_IDIV: |
3952 | assert_operand_cnt(1); |
3953 | if (po->operand[0].lmod != OPLM_DWORD) |
3954 | ferr(po, "unhandled lmod %d\n", po->operand[0].lmod); |
3955 | |
3956 | // 32bit division is common, look for it |
64c59faf |
3957 | if (po->op == OP_DIV) |
cdfaeed7 |
3958 | ret = scan_for_reg_clear(i, xDX); |
64c59faf |
3959 | else |
cdfaeed7 |
3960 | ret = scan_for_cdq_edx(i); |
64c59faf |
3961 | if (ret >= 0) { |
3ebea2cf |
3962 | out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]); |
5101a5f9 |
3963 | strcpy(buf2, lmod_cast(po, po->operand[0].lmod, |
3964 | po->op == OP_IDIV)); |
3965 | fprintf(fout, " edx = %seax %% %s%s;\n", buf2, buf2, buf1); |
3966 | fprintf(fout, " eax = %seax / %s%s;", buf2, buf2, buf1); |
3967 | } |
3968 | else |
3969 | ferr(po, "TODO 64bit divident\n"); |
87bf6cec |
3970 | last_arith_dst = NULL; |
3971 | delayed_flag_op = NULL; |
5101a5f9 |
3972 | break; |
3973 | |
91977a1c |
3974 | case OP_TEST: |
3975 | case OP_CMP: |
850c9265 |
3976 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
940e8e66 |
3977 | if (pfomask != 0) { |
69a3cdfc |
3978 | for (j = 0; j < 8; j++) { |
940e8e66 |
3979 | if (pfomask & (1 << j)) { |
69a3cdfc |
3980 | out_cmp_test(buf1, sizeof(buf1), po, j, 0); |
3981 | fprintf(fout, " cond_%s = %s;", |
3982 | parsed_flag_op_names[j], buf1); |
3983 | } |
3984 | } |
940e8e66 |
3985 | pfomask = 0; |
69a3cdfc |
3986 | } |
3987 | else |
3988 | no_output = 1; |
7ba45c34 |
3989 | last_arith_dst = NULL; |
69a3cdfc |
3990 | delayed_flag_op = po; |
91977a1c |
3991 | break; |
3992 | |
69a3cdfc |
3993 | // note: we reuse OP_Jcc for SETcc, only flags differ |
91977a1c |
3994 | case OP_JO ... OP_JG: |
5101a5f9 |
3995 | if (po->flags & OPF_JMP) |
850c9265 |
3996 | fprintf(fout, " goto %s;", po->operand[0].name); |
5101a5f9 |
3997 | // else SETcc - should already be handled |
850c9265 |
3998 | break; |
3999 | |
5c024ef7 |
4000 | case OP_JECXZ: |
4001 | fprintf(fout, " if (ecx == 0)\n"); |
4002 | fprintf(fout, " goto %s;", po->operand[0].name); |
4003 | strcat(g_comment, "jecxz"); |
4004 | break; |
4005 | |
850c9265 |
4006 | case OP_JMP: |
87bf6cec |
4007 | assert_operand_cnt(1); |
de50b98b |
4008 | last_arith_dst = NULL; |
4009 | delayed_flag_op = NULL; |
4010 | |
4c45fa73 |
4011 | if (po->operand[0].type == OPT_REGMEM) { |
4012 | ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]", |
4013 | buf1, buf2); |
4014 | if (ret != 2) |
4015 | ferr(po, "parse failure for jmp '%s'\n", |
4016 | po->operand[0].name); |
4017 | fprintf(fout, " goto *jt_%s[%s];", buf1, buf2); |
4018 | break; |
4019 | } |
4020 | else if (po->operand[0].type != OPT_LABEL) |
4021 | ferr(po, "unhandled jmp type\n"); |
87bf6cec |
4022 | |
850c9265 |
4023 | fprintf(fout, " goto %s;", po->operand[0].name); |
91977a1c |
4024 | break; |
4025 | |
4026 | case OP_CALL: |
5101a5f9 |
4027 | assert_operand_cnt(1); |
850c9265 |
4028 | pp = po->datap; |
89ff3147 |
4029 | my_assert_not(pp, NULL); |
91977a1c |
4030 | |
89ff3147 |
4031 | if (pp->is_fptr) |
ddaf8bd7 |
4032 | fprintf(fout, " %s = %s;\n", pp->name, |
1cd4a663 |
4033 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], |
4034 | "(void *)", 0)); |
1bafb621 |
4035 | |
91977a1c |
4036 | fprintf(fout, " "); |
2b43685d |
4037 | if (strstr(pp->ret_type.name, "int64")) { |
87bf6cec |
4038 | if (po->flags & OPF_TAIL) |
2b43685d |
4039 | ferr(po, "int64 and tail?\n"); |
4040 | fprintf(fout, "mul_tmp = "); |
4041 | } |
4042 | else if (!IS(pp->ret_type.name, "void")) { |
4043 | if (po->flags & OPF_TAIL) { |
840257f6 |
4044 | if (!IS(g_func_pp->ret_type.name, "void")) { |
4045 | fprintf(fout, "return "); |
4046 | if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr) |
4047 | fprintf(fout, "(%s)", g_func_pp->ret_type.name); |
4048 | } |
2b43685d |
4049 | } |
75ad0378 |
4050 | else if (regmask & (1 << xAX)) { |
87bf6cec |
4051 | fprintf(fout, "eax = "); |
2b43685d |
4052 | if (pp->ret_type.is_ptr) |
4053 | fprintf(fout, "(u32)"); |
4054 | } |
91977a1c |
4055 | } |
87bf6cec |
4056 | |
ddaf8bd7 |
4057 | if (pp->name[0] == 0) |
4058 | ferr(po, "missing pp->name\n"); |
4059 | fprintf(fout, "%s%s(", pp->name, |
4060 | pp->has_structarg ? "_sa" : ""); |
39b168b8 |
4061 | |
91977a1c |
4062 | for (arg = 0; arg < pp->argc; arg++) { |
4063 | if (arg > 0) |
4064 | fprintf(fout, ", "); |
87bf6cec |
4065 | |
3ebea2cf |
4066 | cast[0] = 0; |
4067 | if (pp->arg[arg].type.is_ptr) |
4068 | snprintf(cast, sizeof(cast), "(%s)", pp->arg[arg].type.name); |
87bf6cec |
4069 | |
91977a1c |
4070 | if (pp->arg[arg].reg != NULL) { |
3ebea2cf |
4071 | fprintf(fout, "%s%s", cast, pp->arg[arg].reg); |
91977a1c |
4072 | continue; |
4073 | } |
4074 | |
4075 | // stack arg |
4076 | tmp_op = pp->arg[arg].datap; |
4077 | if (tmp_op == NULL) |
850c9265 |
4078 | ferr(po, "parsed_op missing for arg%d\n", arg); |
de50b98b |
4079 | if (tmp_op->argnum != 0) { |
4080 | fprintf(fout, "%ss_a%d", cast, tmp_op->argnum); |
69a3cdfc |
4081 | } |
4082 | else { |
4083 | fprintf(fout, "%s", |
4084 | out_src_opr(buf1, sizeof(buf1), |
3ebea2cf |
4085 | tmp_op, &tmp_op->operand[0], cast, 0)); |
69a3cdfc |
4086 | } |
91977a1c |
4087 | } |
4088 | fprintf(fout, ");"); |
87bf6cec |
4089 | |
2b43685d |
4090 | if (strstr(pp->ret_type.name, "int64")) { |
4091 | fprintf(fout, "\n"); |
4092 | fprintf(fout, " edx = mul_tmp >> 32;\n"); |
4093 | fprintf(fout, " eax = mul_tmp;"); |
4094 | } |
4095 | |
89ff3147 |
4096 | if (pp->is_unresolved) { |
ddaf8bd7 |
4097 | snprintf(buf3, sizeof(buf3), " unresoved %dreg", |
89ff3147 |
4098 | pp->argc_reg); |
4099 | strcat(g_comment, buf3); |
4100 | } |
4101 | |
87bf6cec |
4102 | if (po->flags & OPF_TAIL) { |
840257f6 |
4103 | ret = 0; |
ddaf8bd7 |
4104 | if (i == opcnt - 1 || pp->is_noreturn) |
840257f6 |
4105 | ret = 0; |
4106 | else if (IS(pp->ret_type.name, "void")) |
4107 | ret = 1; |
4108 | else if (IS(g_func_pp->ret_type.name, "void")) |
4109 | ret = 1; |
4110 | // else already handled as 'return f()' |
4111 | |
4112 | if (ret) { |
da87ae38 |
4113 | if (!IS(g_func_pp->ret_type.name, "void")) { |
ddaf8bd7 |
4114 | ferr(po, "int func -> void func tailcall?\n"); |
da87ae38 |
4115 | } |
4116 | else { |
4117 | fprintf(fout, "\n return;"); |
ddaf8bd7 |
4118 | strcat(g_comment, " ^ tailcall"); |
da87ae38 |
4119 | } |
3ebea2cf |
4120 | } |
89ff3147 |
4121 | else |
ddaf8bd7 |
4122 | strcat(g_comment, " tailcall"); |
87bf6cec |
4123 | } |
ddaf8bd7 |
4124 | if (pp->is_noreturn) |
4125 | strcat(g_comment, " noreturn"); |
87bf6cec |
4126 | delayed_flag_op = NULL; |
4127 | last_arith_dst = NULL; |
91977a1c |
4128 | break; |
4129 | |
4130 | case OP_RET: |
bd96f656 |
4131 | if (g_func_pp->is_vararg) |
4f12f671 |
4132 | fprintf(fout, " va_end(ap);\n"); |
4133 | |
bd96f656 |
4134 | if (IS(g_func_pp->ret_type.name, "void")) { |
3ebea2cf |
4135 | if (i != opcnt - 1 || label_pending) |
4136 | fprintf(fout, " return;"); |
4137 | } |
bd96f656 |
4138 | else if (g_func_pp->ret_type.is_ptr) { |
d4e3b5db |
4139 | fprintf(fout, " return (%s)eax;", |
bd96f656 |
4140 | g_func_pp->ret_type.name); |
3ebea2cf |
4141 | } |
91977a1c |
4142 | else |
4143 | fprintf(fout, " return eax;"); |
de50b98b |
4144 | |
4145 | last_arith_dst = NULL; |
4146 | delayed_flag_op = NULL; |
91977a1c |
4147 | break; |
4148 | |
4149 | case OP_PUSH: |
de50b98b |
4150 | if (po->argnum != 0) { |
69a3cdfc |
4151 | // special case - saved func arg |
3ebea2cf |
4152 | out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]); |
de50b98b |
4153 | fprintf(fout, " s_a%d = %s;", po->argnum, buf1); |
69a3cdfc |
4154 | break; |
4155 | } |
d4e3b5db |
4156 | else if (po->flags & OPF_RSAVE) { |
3ebea2cf |
4157 | out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]); |
d4e3b5db |
4158 | fprintf(fout, " s_%s = %s;", buf1, buf1); |
4159 | break; |
4160 | } |
e56ab892 |
4161 | if (!(g_ida_func_attr & IDAFA_NORETURN)) |
4162 | ferr(po, "stray push encountered\n"); |
4163 | no_output = 1; |
91977a1c |
4164 | break; |
4165 | |
4166 | case OP_POP: |
d4e3b5db |
4167 | if (po->flags & OPF_RSAVE) { |
3ebea2cf |
4168 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
d4e3b5db |
4169 | fprintf(fout, " %s = s_%s;", buf1, buf1); |
4170 | break; |
4171 | } |
5c024ef7 |
4172 | else if (po->datap != NULL) { |
4173 | // push/pop pair |
4174 | tmp_op = po->datap; |
4175 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
4176 | fprintf(fout, " %s = %s;", buf1, |
4177 | out_src_opr(buf2, sizeof(buf2), |
4178 | tmp_op, &tmp_op->operand[0], |
4179 | po->operand[0].is_ptr ? "(void *)" : "", 0)); |
4180 | break; |
4181 | } |
d4e3b5db |
4182 | ferr(po, "stray pop encountered\n"); |
91977a1c |
4183 | break; |
4184 | |
33c35af6 |
4185 | case OP_NOP: |
2b43685d |
4186 | no_output = 1; |
33c35af6 |
4187 | break; |
4188 | |
91977a1c |
4189 | default: |
4190 | no_output = 1; |
69a3cdfc |
4191 | ferr(po, "unhandled op type %d, flags %x\n", |
4192 | po->op, po->flags); |
91977a1c |
4193 | break; |
4194 | } |
4195 | |
4196 | if (g_comment[0] != 0) { |
ddaf8bd7 |
4197 | char *p = g_comment; |
4198 | while (my_isblank(*p)) |
4199 | p++; |
4200 | fprintf(fout, " // %s", p); |
91977a1c |
4201 | g_comment[0] = 0; |
4202 | no_output = 0; |
4203 | } |
4204 | if (!no_output) |
4205 | fprintf(fout, "\n"); |
5101a5f9 |
4206 | |
2b43685d |
4207 | // some sanity checking |
591721d7 |
4208 | if (po->flags & OPF_REP) { |
4209 | if (po->op != OP_STOS && po->op != OP_MOVS |
4210 | && po->op != OP_CMPS && po->op != OP_SCAS) |
4211 | ferr(po, "unexpected rep\n"); |
4212 | if (!(po->flags & (OPF_REPZ|OPF_REPNZ)) |
4213 | && (po->op == OP_CMPS || po->op == OP_SCAS)) |
4214 | ferr(po, "cmps/scas with plain rep\n"); |
4215 | } |
4216 | if ((po->flags & (OPF_REPZ|OPF_REPNZ)) |
4217 | && po->op != OP_CMPS && po->op != OP_SCAS) |
2b43685d |
4218 | ferr(po, "unexpected repz/repnz\n"); |
4219 | |
940e8e66 |
4220 | if (pfomask != 0) |
7ba45c34 |
4221 | ferr(po, "missed flag calc, pfomask=%x\n", pfomask); |
940e8e66 |
4222 | |
5101a5f9 |
4223 | // see is delayed flag stuff is still valid |
4224 | if (delayed_flag_op != NULL && delayed_flag_op != po) { |
89ff3147 |
4225 | if (is_any_opr_modified(delayed_flag_op, po, 0)) |
5101a5f9 |
4226 | delayed_flag_op = NULL; |
4227 | } |
4228 | |
4229 | if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) { |
4230 | if (is_opr_modified(last_arith_dst, po)) |
4231 | last_arith_dst = NULL; |
4232 | } |
3ebea2cf |
4233 | |
4234 | label_pending = 0; |
91977a1c |
4235 | } |
4236 | |
a2c1d768 |
4237 | if (g_stack_fsz && !g_stack_frame_used) |
4238 | fprintf(fout, " (void)sf;\n"); |
4239 | |
91977a1c |
4240 | fprintf(fout, "}\n\n"); |
4241 | |
4242 | // cleanup |
4243 | for (i = 0; i < opcnt; i++) { |
4c45fa73 |
4244 | struct label_ref *lr, *lr_del; |
4245 | |
4246 | lr = g_label_refs[i].next; |
4247 | while (lr != NULL) { |
4248 | lr_del = lr; |
4249 | lr = lr->next; |
4250 | free(lr_del); |
4251 | } |
4252 | g_label_refs[i].i = -1; |
4253 | g_label_refs[i].next = NULL; |
4254 | |
91977a1c |
4255 | if (ops[i].op == OP_CALL) { |
4256 | pp = ops[i].datap; |
bd96f656 |
4257 | if (pp) |
91977a1c |
4258 | proto_release(pp); |
91977a1c |
4259 | } |
4260 | } |
bd96f656 |
4261 | g_func_pp = NULL; |
91977a1c |
4262 | } |
c36e914d |
4263 | |
d4e3b5db |
4264 | static void set_label(int i, const char *name) |
4265 | { |
4266 | const char *p; |
4267 | int len; |
4268 | |
4269 | len = strlen(name); |
4270 | p = strchr(name, ':'); |
4271 | if (p != NULL) |
4272 | len = p - name; |
4273 | |
4274 | if (len > sizeof(g_labels[0]) - 1) |
4275 | aerr("label '%s' too long: %d\n", name, len); |
e56ab892 |
4276 | if (g_labels[i][0] != 0 && !IS_START(g_labels[i], "algn_")) |
4277 | aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]); |
d4e3b5db |
4278 | memcpy(g_labels[i], name, len); |
4279 | g_labels[i][len] = 0; |
4280 | } |
4281 | |
bfa4a6ee |
4282 | // '=' needs special treatment.. |
4283 | static char *next_word_s(char *w, size_t wsize, char *s) |
4284 | { |
4285 | size_t i; |
4286 | |
4287 | s = sskip(s); |
4288 | |
4289 | for (i = 0; i < wsize - 1; i++) { |
4290 | if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0)) |
4291 | break; |
4292 | w[i] = s[i]; |
4293 | } |
4294 | w[i] = 0; |
4295 | |
4296 | if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=') |
4297 | printf("warning: '%s' truncated\n", w); |
4298 | |
4299 | return s + i; |
4300 | } |
4301 | |
e56ab892 |
4302 | struct chunk_item { |
4303 | char *name; |
4304 | long fptr; |
de50b98b |
4305 | int asmln; |
e56ab892 |
4306 | }; |
4307 | |
cdfaeed7 |
4308 | static struct chunk_item *func_chunks; |
4309 | static int func_chunk_cnt; |
4310 | static int func_chunk_alloc; |
4311 | |
4312 | static void add_func_chunk(FILE *fasm, const char *name, int line) |
4313 | { |
4314 | if (func_chunk_cnt >= func_chunk_alloc) { |
4315 | func_chunk_alloc *= 2; |
4316 | func_chunks = realloc(func_chunks, |
4317 | func_chunk_alloc * sizeof(func_chunks[0])); |
4318 | my_assert_not(func_chunks, NULL); |
4319 | } |
4320 | func_chunks[func_chunk_cnt].fptr = ftell(fasm); |
4321 | func_chunks[func_chunk_cnt].name = strdup(name); |
4322 | func_chunks[func_chunk_cnt].asmln = line; |
4323 | func_chunk_cnt++; |
4324 | } |
4325 | |
e56ab892 |
4326 | static int cmp_chunks(const void *p1, const void *p2) |
4327 | { |
4328 | const struct chunk_item *c1 = p1, *c2 = p2; |
4329 | return strcmp(c1->name, c2->name); |
4330 | } |
4331 | |
bfa4a6ee |
4332 | static int cmpstringp(const void *p1, const void *p2) |
4333 | { |
4334 | return strcmp(*(char * const *)p1, *(char * const *)p2); |
4335 | } |
4336 | |
cdfaeed7 |
4337 | static void scan_ahead(FILE *fasm) |
4338 | { |
4339 | char words[2][256]; |
4340 | char line[256]; |
4341 | long oldpos; |
4342 | int oldasmln; |
4343 | int wordc; |
4344 | char *p; |
4345 | int i; |
4346 | |
4347 | oldpos = ftell(fasm); |
4348 | oldasmln = asmln; |
4349 | |
4350 | while (fgets(line, sizeof(line), fasm)) |
4351 | { |
4352 | wordc = 0; |
4353 | asmln++; |
4354 | |
4355 | p = sskip(line); |
4356 | if (*p == 0) |
4357 | continue; |
4358 | |
4359 | if (*p == ';') |
4360 | { |
4361 | // get rid of random tabs |
4362 | for (i = 0; line[i] != 0; i++) |
4363 | if (line[i] == '\t') |
4364 | line[i] = ' '; |
4365 | |
4366 | if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR ")) |
4367 | { |
4368 | p += 30; |
4369 | next_word(words[0], sizeof(words[0]), p); |
4370 | if (words[0][0] == 0) |
4371 | aerr("missing name for func chunk?\n"); |
4372 | |
4373 | add_func_chunk(fasm, words[0], asmln); |
4374 | } |
4375 | continue; |
4376 | } // *p == ';' |
4377 | |
4378 | for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) { |
4379 | words[wordc][0] = 0; |
4380 | p = sskip(next_word_s(words[wordc], sizeof(words[0]), p)); |
4381 | if (*p == 0 || *p == ';') { |
4382 | wordc++; |
4383 | break; |
4384 | } |
4385 | } |
4386 | |
4387 | if (wordc == 2 && IS(words[1], "ends")) |
4388 | break; |
4389 | } |
4390 | |
4391 | fseek(fasm, oldpos, SEEK_SET); |
4392 | asmln = oldasmln; |
4393 | } |
4394 | |
91977a1c |
4395 | int main(int argc, char *argv[]) |
4396 | { |
06c5d854 |
4397 | FILE *fout, *fasm, *frlist; |
4c45fa73 |
4398 | struct parsed_data *pd = NULL; |
4399 | int pd_alloc = 0; |
4400 | char **rlist = NULL; |
4401 | int rlist_len = 0; |
4402 | int rlist_alloc = 0; |
e56ab892 |
4403 | int func_chunks_used = 0; |
4404 | int func_chunks_sorted = 0; |
e56ab892 |
4405 | int func_chunk_i = -1; |
4406 | long func_chunk_ret = 0; |
de50b98b |
4407 | int func_chunk_ret_ln = 0; |
cdfaeed7 |
4408 | int scanned_ahead = 0; |
91977a1c |
4409 | char line[256]; |
a2c1d768 |
4410 | char words[20][256]; |
4c45fa73 |
4411 | enum opr_lenmod lmod; |
ddaf8bd7 |
4412 | char *sctproto = NULL; |
91977a1c |
4413 | int in_func = 0; |
4c45fa73 |
4414 | int pending_endp = 0; |
bfa4a6ee |
4415 | int skip_func = 0; |
940e8e66 |
4416 | int skip_warned = 0; |
91977a1c |
4417 | int eq_alloc; |
bfa4a6ee |
4418 | int verbose = 0; |
4419 | int arg_out; |
89ff3147 |
4420 | int arg; |
91977a1c |
4421 | int pi = 0; |
e56ab892 |
4422 | int i, j; |
4423 | int ret, len; |
91977a1c |
4424 | char *p; |
4425 | int wordc; |
4426 | |
89ff3147 |
4427 | for (arg = 1; arg < argc; arg++) { |
4428 | if (IS(argv[arg], "-v")) |
4429 | verbose = 1; |
4430 | else if (IS(argv[arg], "-rf")) |
4431 | g_allow_regfunc = 1; |
4432 | else |
4433 | break; |
bfa4a6ee |
4434 | } |
4435 | |
4436 | if (argc < arg + 3) { |
89ff3147 |
4437 | printf("usage:\n%s [-v] [-rf] <.c> <.asm> <hdrf> [rlist]*\n", |
91977a1c |
4438 | argv[0]); |
4439 | return 1; |
4440 | } |
4441 | |
bfa4a6ee |
4442 | arg_out = arg++; |
91977a1c |
4443 | |
bfa4a6ee |
4444 | asmfn = argv[arg++]; |
91977a1c |
4445 | fasm = fopen(asmfn, "r"); |
4446 | my_assert_not(fasm, NULL); |
4447 | |
bfa4a6ee |
4448 | hdrfn = argv[arg++]; |
06c5d854 |
4449 | g_fhdr = fopen(hdrfn, "r"); |
4450 | my_assert_not(g_fhdr, NULL); |
bfa4a6ee |
4451 | |
4452 | rlist_alloc = 64; |
4453 | rlist = malloc(rlist_alloc * sizeof(rlist[0])); |
4454 | my_assert_not(rlist, NULL); |
4455 | // needs special handling.. |
4456 | rlist[rlist_len++] = "__alloca_probe"; |
4457 | |
e56ab892 |
4458 | func_chunk_alloc = 32; |
4459 | func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0])); |
4460 | my_assert_not(func_chunks, NULL); |
4461 | |
a2c1d768 |
4462 | memset(words, 0, sizeof(words)); |
4463 | |
bfa4a6ee |
4464 | for (; arg < argc; arg++) { |
4465 | frlist = fopen(argv[arg], "r"); |
4466 | my_assert_not(frlist, NULL); |
4467 | |
4468 | while (fgets(line, sizeof(line), frlist)) { |
4469 | p = sskip(line); |
1cd4a663 |
4470 | if (*p == 0 || *p == ';') |
4471 | continue; |
4472 | if (*p == '#') { |
89ff3147 |
4473 | if (IS_START(p, "#if 0") |
4474 | || (g_allow_regfunc && IS_START(p, "#if NO_REGFUNC"))) |
4475 | { |
1cd4a663 |
4476 | skip_func = 1; |
89ff3147 |
4477 | } |
1cd4a663 |
4478 | else if (IS_START(p, "#endif")) |
4479 | skip_func = 0; |
4480 | continue; |
4481 | } |
4482 | if (skip_func) |
bfa4a6ee |
4483 | continue; |
4484 | |
4485 | p = next_word(words[0], sizeof(words[0]), p); |
4486 | if (words[0][0] == 0) |
4487 | continue; |
4488 | |
4489 | if (rlist_len >= rlist_alloc) { |
4490 | rlist_alloc = rlist_alloc * 2 + 64; |
4491 | rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0])); |
4492 | my_assert_not(rlist, NULL); |
4493 | } |
4494 | rlist[rlist_len++] = strdup(words[0]); |
4495 | } |
1cd4a663 |
4496 | skip_func = 0; |
bfa4a6ee |
4497 | |
4498 | fclose(frlist); |
4499 | frlist = NULL; |
4500 | } |
4501 | |
4502 | if (rlist_len > 0) |
4503 | qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp); |
4504 | |
4505 | fout = fopen(argv[arg_out], "w"); |
91977a1c |
4506 | my_assert_not(fout, NULL); |
4507 | |
4508 | eq_alloc = 128; |
4509 | g_eqs = malloc(eq_alloc * sizeof(g_eqs[0])); |
4510 | my_assert_not(g_eqs, NULL); |
4511 | |
4c45fa73 |
4512 | for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) { |
4513 | g_label_refs[i].i = -1; |
4514 | g_label_refs[i].next = NULL; |
4515 | } |
4516 | |
91977a1c |
4517 | while (fgets(line, sizeof(line), fasm)) |
4518 | { |
4c45fa73 |
4519 | wordc = 0; |
91977a1c |
4520 | asmln++; |
4521 | |
4522 | p = sskip(line); |
1bafb621 |
4523 | if (*p == 0) |
91977a1c |
4524 | continue; |
4525 | |
de50b98b |
4526 | // get rid of random tabs |
4527 | for (i = 0; line[i] != 0; i++) |
4528 | if (line[i] == '\t') |
4529 | line[i] = ' '; |
4530 | |
e56ab892 |
4531 | if (*p == ';') |
4532 | { |
e56ab892 |
4533 | if (p[2] == '=' && IS_START(p, "; =============== S U B")) |
4534 | goto do_pending_endp; // eww.. |
4535 | |
4536 | if (p[2] == 'A' && IS_START(p, "; Attributes:")) |
4537 | { |
4538 | static const char *attrs[] = { |
4539 | "bp-based frame", |
4540 | "library function", |
4541 | "static", |
4542 | "noreturn", |
4543 | "thunk", |
4544 | "fpd=", |
4545 | }; |
4546 | |
4547 | // parse IDA's attribute-list comment |
4548 | g_ida_func_attr = 0; |
4549 | p = sskip(p + 13); |
4550 | |
4551 | for (; *p != 0; p = sskip(p)) { |
4552 | for (i = 0; i < ARRAY_SIZE(attrs); i++) { |
4553 | if (!strncmp(p, attrs[i], strlen(attrs[i]))) { |
4554 | g_ida_func_attr |= 1 << i; |
4555 | p += strlen(attrs[i]); |
4556 | break; |
4557 | } |
4558 | } |
4559 | if (i == ARRAY_SIZE(attrs)) { |
4560 | anote("unparsed IDA attr: %s\n", p); |
1bafb621 |
4561 | break; |
4562 | } |
e56ab892 |
4563 | if (IS(attrs[i], "fpd=")) { |
4564 | p = next_word(words[0], sizeof(words[0]), p); |
4565 | // ignore for now.. |
4566 | } |
1bafb621 |
4567 | } |
e56ab892 |
4568 | } |
4569 | else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR ")) |
4570 | { |
4571 | p += 30; |
4572 | next_word(words[0], sizeof(words[0]), p); |
4573 | if (words[0][0] == 0) |
cdfaeed7 |
4574 | aerr("missing name for func chunk?\n"); |
4575 | |
4576 | if (!scanned_ahead) { |
4577 | add_func_chunk(fasm, words[0], asmln); |
4578 | func_chunks_sorted = 0; |
e56ab892 |
4579 | } |
e56ab892 |
4580 | } |
4581 | else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK")) |
4582 | { |
4583 | if (func_chunk_i >= 0) { |
4584 | if (func_chunk_i < func_chunk_cnt |
4585 | && IS(func_chunks[func_chunk_i].name, g_func)) |
4586 | { |
4587 | // move on to next chunk |
4588 | ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET); |
4589 | if (ret) |
4590 | aerr("seek failed for '%s' chunk #%d\n", |
4591 | g_func, func_chunk_i); |
de50b98b |
4592 | asmln = func_chunks[func_chunk_i].asmln; |
e56ab892 |
4593 | func_chunk_i++; |
4594 | } |
4595 | else { |
4596 | if (func_chunk_ret == 0) |
4597 | aerr("no return from chunk?\n"); |
4598 | fseek(fasm, func_chunk_ret, SEEK_SET); |
de50b98b |
4599 | asmln = func_chunk_ret_ln; |
e56ab892 |
4600 | func_chunk_ret = 0; |
4601 | pending_endp = 1; |
4602 | } |
1bafb621 |
4603 | } |
e56ab892 |
4604 | } |
4605 | else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) { |
4606 | func_chunks_used = 1; |
4607 | p += 20; |
4608 | if (IS_START(g_func, "sub_")) { |
4609 | unsigned long addr = strtoul(p, NULL, 16); |
4610 | unsigned long f_addr = strtoul(g_func + 4, NULL, 16); |
cdfaeed7 |
4611 | if (addr > f_addr && !scanned_ahead) { |
94d447fb |
4612 | anote("scan_ahead caused by '%s', addr %lx\n", |
4613 | g_func, addr); |
cdfaeed7 |
4614 | scan_ahead(fasm); |
4615 | scanned_ahead = 1; |
4616 | func_chunks_sorted = 0; |
4617 | } |
1bafb621 |
4618 | } |
4619 | } |
4620 | continue; |
e56ab892 |
4621 | } // *p == ';' |
1bafb621 |
4622 | |
06c5d854 |
4623 | parse_words: |
a2c1d768 |
4624 | for (i = wordc; i < ARRAY_SIZE(words); i++) |
4625 | words[i][0] = 0; |
cdfaeed7 |
4626 | for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) { |
bfa4a6ee |
4627 | p = sskip(next_word_s(words[wordc], sizeof(words[0]), p)); |
91977a1c |
4628 | if (*p == 0 || *p == ';') { |
4629 | wordc++; |
4630 | break; |
4631 | } |
4632 | } |
a2c1d768 |
4633 | if (*p != 0 && *p != ';') |
4634 | aerr("too many words\n"); |
91977a1c |
4635 | |
06c5d854 |
4636 | // alow asm patches in comments |
ddaf8bd7 |
4637 | if (*p == ';') { |
4638 | if (IS_START(p, "; sctpatch:")) { |
4639 | p = sskip(p + 11); |
4640 | if (*p == 0 || *p == ';') |
4641 | continue; |
4642 | goto parse_words; // lame |
4643 | } |
4644 | if (IS_START(p, "; sctproto:")) { |
4645 | sctproto = strdup(p + 11); |
4646 | } |
06c5d854 |
4647 | } |
4648 | |
91977a1c |
4649 | if (wordc == 0) { |
4650 | // shouldn't happen |
4651 | awarn("wordc == 0?\n"); |
4652 | continue; |
4653 | } |
4654 | |
4655 | // don't care about this: |
4656 | if (words[0][0] == '.' |
4657 | || IS(words[0], "include") |
4658 | || IS(words[0], "assume") || IS(words[1], "segment") |
4659 | || IS(words[0], "align")) |
4660 | { |
4661 | continue; |
4662 | } |
4663 | |
4c45fa73 |
4664 | do_pending_endp: |
4665 | // do delayed endp processing to collect switch jumptables |
4666 | if (pending_endp) { |
4667 | if (in_func && !skip_func && wordc >= 2 |
4668 | && ((words[0][0] == 'd' && words[0][2] == 0) |
4669 | || (words[1][0] == 'd' && words[1][2] == 0))) |
4670 | { |
4671 | i = 1; |
4672 | if (words[1][0] == 'd' && words[1][2] == 0) { |
4673 | // label |
4674 | if (g_func_pd_cnt >= pd_alloc) { |
4675 | pd_alloc = pd_alloc * 2 + 16; |
4676 | g_func_pd = realloc(g_func_pd, |
4677 | sizeof(g_func_pd[0]) * pd_alloc); |
4678 | my_assert_not(g_func_pd, NULL); |
4679 | } |
4680 | pd = &g_func_pd[g_func_pd_cnt]; |
4681 | g_func_pd_cnt++; |
4682 | memset(pd, 0, sizeof(*pd)); |
4683 | strcpy(pd->label, words[0]); |
4684 | pd->type = OPT_CONST; |
4685 | pd->lmod = lmod_from_directive(words[1]); |
4686 | i = 2; |
4687 | } |
4688 | else { |
da87ae38 |
4689 | if (pd == NULL) { |
4690 | if (verbose) |
4691 | anote("skipping alignment byte?\n"); |
4692 | continue; |
4693 | } |
4c45fa73 |
4694 | lmod = lmod_from_directive(words[0]); |
4695 | if (lmod != pd->lmod) |
4696 | aerr("lmod change? %d->%d\n", pd->lmod, lmod); |
4697 | } |
4698 | |
4699 | if (pd->count_alloc < pd->count + wordc) { |
4700 | pd->count_alloc = pd->count_alloc * 2 + 14 + wordc; |
4701 | pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc); |
4702 | my_assert_not(pd->d, NULL); |
4703 | } |
4704 | for (; i < wordc; i++) { |
4705 | if (IS(words[i], "offset")) { |
4706 | pd->type = OPT_OFFSET; |
4707 | i++; |
4708 | } |
4709 | p = strchr(words[i], ','); |
4710 | if (p != NULL) |
4711 | *p = 0; |
4712 | if (pd->type == OPT_OFFSET) |
4713 | pd->d[pd->count].u.label = strdup(words[i]); |
4714 | else |
4715 | pd->d[pd->count].u.val = parse_number(words[i]); |
4716 | pd->d[pd->count].bt_i = -1; |
4717 | pd->count++; |
4718 | } |
4719 | continue; |
4720 | } |
4721 | |
4722 | if (in_func && !skip_func) |
4723 | gen_func(fout, g_fhdr, g_func, pi); |
4724 | |
4725 | pending_endp = 0; |
4726 | in_func = 0; |
4727 | g_ida_func_attr = 0; |
4728 | skip_warned = 0; |
4729 | skip_func = 0; |
4730 | g_func[0] = 0; |
e56ab892 |
4731 | func_chunks_used = 0; |
4732 | func_chunk_i = -1; |
4c45fa73 |
4733 | if (pi != 0) { |
4734 | memset(&ops, 0, pi * sizeof(ops[0])); |
4735 | memset(g_labels, 0, pi * sizeof(g_labels[0])); |
4736 | pi = 0; |
4737 | } |
4738 | g_eqcnt = 0; |
4739 | for (i = 0; i < g_func_pd_cnt; i++) { |
4740 | pd = &g_func_pd[i]; |
4741 | if (pd->type == OPT_OFFSET) { |
4742 | for (j = 0; j < pd->count; j++) |
4743 | free(pd->d[j].u.label); |
4744 | } |
4745 | free(pd->d); |
4746 | pd->d = NULL; |
4747 | } |
4748 | g_func_pd_cnt = 0; |
4749 | pd = NULL; |
4750 | if (wordc == 0) |
4751 | continue; |
4752 | } |
4753 | |
91977a1c |
4754 | if (IS(words[1], "proc")) { |
4755 | if (in_func) |
4756 | aerr("proc '%s' while in_func '%s'?\n", |
4757 | words[0], g_func); |
bfa4a6ee |
4758 | p = words[0]; |
ddaf8bd7 |
4759 | if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp)) |
bfa4a6ee |
4760 | skip_func = 1; |
91977a1c |
4761 | strcpy(g_func, words[0]); |
d4e3b5db |
4762 | set_label(0, words[0]); |
91977a1c |
4763 | in_func = 1; |
4764 | continue; |
4765 | } |
4766 | |
e56ab892 |
4767 | if (IS(words[1], "endp")) |
4768 | { |
91977a1c |
4769 | if (!in_func) |
4770 | aerr("endp '%s' while not in_func?\n", words[0]); |
4771 | if (!IS(g_func, words[0])) |
4772 | aerr("endp '%s' while in_func '%s'?\n", |
4773 | words[0], g_func); |
bfa4a6ee |
4774 | |
ddaf8bd7 |
4775 | if ((g_ida_func_attr & IDAFA_THUNK) && pi == 1 |
4776 | && ops[0].op == OP_JMP && ops[0].operand[0].had_ds) |
4777 | { |
4778 | // import jump |
4779 | skip_func = 1; |
4780 | } |
4781 | |
e56ab892 |
4782 | if (!skip_func && func_chunks_used) { |
4783 | // start processing chunks |
4784 | struct chunk_item *ci, key = { g_func, 0 }; |
4785 | |
4786 | func_chunk_ret = ftell(fasm); |
de50b98b |
4787 | func_chunk_ret_ln = asmln; |
e56ab892 |
4788 | if (!func_chunks_sorted) { |
4789 | qsort(func_chunks, func_chunk_cnt, |
4790 | sizeof(func_chunks[0]), cmp_chunks); |
4791 | func_chunks_sorted = 1; |
4792 | } |
4793 | ci = bsearch(&key, func_chunks, func_chunk_cnt, |
4794 | sizeof(func_chunks[0]), cmp_chunks); |
4795 | if (ci == NULL) |
4796 | aerr("'%s' needs chunks, but none found\n", g_func); |
4797 | func_chunk_i = ci - func_chunks; |
4798 | for (; func_chunk_i > 0; func_chunk_i--) |
4799 | if (!IS(func_chunks[func_chunk_i - 1].name, g_func)) |
4800 | break; |
4801 | |
4802 | ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET); |
4803 | if (ret) |
4804 | aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i); |
de50b98b |
4805 | asmln = func_chunks[func_chunk_i].asmln; |
e56ab892 |
4806 | func_chunk_i++; |
4807 | continue; |
4808 | } |
4c45fa73 |
4809 | pending_endp = 1; |
91977a1c |
4810 | continue; |
4811 | } |
4812 | |
a2c1d768 |
4813 | if (wordc == 2 && IS(words[1], "ends")) |
4814 | break; |
4815 | |
bfa4a6ee |
4816 | p = strchr(words[0], ':'); |
4817 | if (p != NULL) { |
d4e3b5db |
4818 | set_label(pi, words[0]); |
bfa4a6ee |
4819 | continue; |
4820 | } |
4821 | |
4822 | if (!in_func || skip_func) { |
4823 | if (!skip_warned && !skip_func && g_labels[pi][0] != 0) { |
4824 | if (verbose) |
4825 | anote("skipping from '%s'\n", g_labels[pi]); |
4826 | skip_warned = 1; |
4827 | } |
4828 | g_labels[pi][0] = 0; |
4829 | continue; |
4830 | } |
4831 | |
ddaf8bd7 |
4832 | if (wordc > 1 && IS(words[1], "=")) |
4833 | { |
91977a1c |
4834 | if (wordc != 5) |
4835 | aerr("unhandled equ, wc=%d\n", wordc); |
4836 | if (g_eqcnt >= eq_alloc) { |
4837 | eq_alloc *= 2; |
4838 | g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0])); |
4839 | my_assert_not(g_eqs, NULL); |
4840 | } |
4841 | |
4842 | len = strlen(words[0]); |
4843 | if (len > sizeof(g_eqs[0].name) - 1) |
4844 | aerr("equ name too long: %d\n", len); |
4845 | strcpy(g_eqs[g_eqcnt].name, words[0]); |
4846 | |
4847 | if (!IS(words[3], "ptr")) |
4848 | aerr("unhandled equ\n"); |
4849 | if (IS(words[2], "dword")) |
4850 | g_eqs[g_eqcnt].lmod = OPLM_DWORD; |
4851 | else if (IS(words[2], "word")) |
4852 | g_eqs[g_eqcnt].lmod = OPLM_WORD; |
4853 | else if (IS(words[2], "byte")) |
4854 | g_eqs[g_eqcnt].lmod = OPLM_BYTE; |
4855 | else |
4856 | aerr("bad lmod: '%s'\n", words[2]); |
4857 | |
4858 | g_eqs[g_eqcnt].offset = parse_number(words[4]); |
4859 | g_eqcnt++; |
4860 | continue; |
4861 | } |
4862 | |
4863 | if (pi >= ARRAY_SIZE(ops)) |
4864 | aerr("too many ops\n"); |
4865 | |
91977a1c |
4866 | parse_op(&ops[pi], words, wordc); |
ddaf8bd7 |
4867 | |
4868 | if (sctproto != NULL) { |
4869 | if (ops[pi].op == OP_CALL) |
4870 | ops[pi].datap = sctproto; |
4871 | sctproto = NULL; |
4872 | } |
91977a1c |
4873 | pi++; |
91977a1c |
4874 | } |
4875 | |
4876 | fclose(fout); |
4877 | fclose(fasm); |
06c5d854 |
4878 | fclose(g_fhdr); |
91977a1c |
4879 | |
4880 | return 0; |
c36e914d |
4881 | } |
91977a1c |
4882 | |
4883 | // vim:ts=2:shiftwidth=2:expandtab |