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