notaz.gp2x.de / ia32rtools.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
cmpmrg_text: allow to skip actual comparison
[ia32rtools.git] / tools / translate.c
CommitLineData
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 15static const char *asmfn;
c36e914d 16static int asmln;
06c5d854 17static 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 31enum 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 */
2fe80fdb 46 OPF_ATAIL = (1 << 14), /* tail call with reused arg frame */
cb090db0 47 OPF_32BIT = (1 << 15), /* 32bit division */
c36e914d 48};
49
50enum op_op {
51 OP_INVAL,
33c35af6 52 OP_NOP,
c36e914d 53 OP_PUSH,
54 OP_POP,
591721d7 55 OP_LEAVE,
c36e914d 56 OP_MOV,
850c9265 57 OP_LEA,
58 OP_MOVZX,
59 OP_MOVSX,
108e9fe3 60 OP_XCHG,
850c9265 61 OP_NOT,
5101a5f9 62 OP_CDQ,
33c35af6 63 OP_STOS,
d4e3b5db 64 OP_MOVS,
7ba45c34 65 OP_CMPS,
591721d7 66 OP_SCAS,
67 OP_STD,
68 OP_CLD,
c36e914d 69 OP_RET,
70 OP_ADD,
91977a1c 71 OP_SUB,
850c9265 72 OP_AND,
73 OP_OR,
74 OP_XOR,
75 OP_SHL,
76 OP_SHR,
77 OP_SAR,
d4e3b5db 78 OP_ROL,
79 OP_ROR,
cb090db0 80 OP_RCL,
81 OP_RCR,
69a3cdfc 82 OP_ADC,
850c9265 83 OP_SBB,
84 OP_INC,
85 OP_DEC,
5101a5f9 86 OP_NEG,
850c9265 87 OP_MUL,
88 OP_IMUL,
5101a5f9 89 OP_DIV,
90 OP_IDIV,
c36e914d 91 OP_TEST,
92 OP_CMP,
93 OP_CALL,
94 OP_JMP,
5c024ef7 95 OP_JECXZ,
c36e914d 96 OP_JO,
97 OP_JNO,
98 OP_JC,
99 OP_JNC,
100 OP_JZ,
101 OP_JNZ,
102 OP_JBE,
103 OP_JA,
104 OP_JS,
105 OP_JNS,
106 OP_JP,
107 OP_JNP,
108 OP_JL,
109 OP_JGE,
110 OP_JLE,
111 OP_JG,
112};
113
114enum opr_type {
87bf6cec 115 OPT_UNSPEC,
116 OPT_REG,
117 OPT_REGMEM,
118 OPT_LABEL,
850c9265 119 OPT_OFFSET,
87bf6cec 120 OPT_CONST,
c36e914d 121};
122
2b43685d 123// must be sorted (larger len must be further in enum)
c36e914d 124enum opr_lenmod {
91977a1c 125 OPLM_UNSPEC,
126 OPLM_BYTE,
127 OPLM_WORD,
128 OPLM_DWORD,
c36e914d 129};
130
850c9265 131#define MAX_OPERANDS 3
c36e914d 132
133struct parsed_opr {
91977a1c 134 enum opr_type type;
135 enum opr_lenmod lmod;
7ba45c34 136 unsigned int is_ptr:1; // pointer in C
137 unsigned int is_array:1; // array in C
a3684be1 138 unsigned int type_from_var:1; // .. in header, sometimes wrong
2b43685d 139 unsigned int size_mismatch:1; // type override differs from C
140 unsigned int size_lt:1; // type override is larger than C
ddaf8bd7 141 unsigned int had_ds:1; // had ds: prefix
91977a1c 142 int reg;
143 unsigned int val;
144 char name[256];
c36e914d 145};
146
147struct parsed_op {
91977a1c 148 enum op_op op;
149 struct parsed_opr operand[MAX_OPERANDS];
69a3cdfc 150 unsigned int flags;
91977a1c 151 int operand_cnt;
69a3cdfc 152 int regmask_src; // all referensed regs
153 int regmask_dst;
940e8e66 154 int pfomask; // flagop: parsed_flag_op that can't be delayed
de50b98b 155 int argnum; // push: altered before call arg #
940e8e66 156 int cc_scratch; // scratch storage during analysis
4c45fa73 157 int bt_i; // branch target for branches
158 struct parsed_data *btj;// branch targets for jumptables
91977a1c 159 void *datap;
160};
161
69a3cdfc 162// datap:
ddaf8bd7 163// OP_CALL - parser proto hint (str), ptr to struct parsed_proto
2b43685d 164// (OPF_CC) - point to one of (OPF_FLAGS) that affects cc op
5c024ef7 165// OP_POP - point to OP_PUSH in push/pop pair
69a3cdfc 166
91977a1c 167struct parsed_equ {
168 char name[64];
169 enum opr_lenmod lmod;
170 int offset;
c36e914d 171};
172
4c45fa73 173struct parsed_data {
174 char label[256];
175 enum opr_type type;
176 enum opr_lenmod lmod;
177 int count;
178 int count_alloc;
179 struct {
180 union {
181 char *label;
63df67be 182 unsigned int val;
4c45fa73 183 } u;
184 int bt_i;
185 } *d;
186};
187
188struct label_ref {
189 int i;
190 struct label_ref *next;
191};
192
1bafb621 193enum ida_func_attr {
194 IDAFA_BP_FRAME = (1 << 0),
195 IDAFA_LIB_FUNC = (1 << 1),
196 IDAFA_STATIC = (1 << 2),
197 IDAFA_NORETURN = (1 << 3),
198 IDAFA_THUNK = (1 << 4),
199 IDAFA_FPD = (1 << 5),
200};
201
202#define MAX_OPS 4096
c36e914d 203
204static struct parsed_op ops[MAX_OPS];
91977a1c 205static struct parsed_equ *g_eqs;
206static int g_eqcnt;
207static char g_labels[MAX_OPS][32];
4c45fa73 208static struct label_ref g_label_refs[MAX_OPS];
bd96f656 209static const struct parsed_proto *g_func_pp;
4c45fa73 210static struct parsed_data *g_func_pd;
211static int g_func_pd_cnt;
91977a1c 212static char g_func[256];
213static char g_comment[256];
214static int g_bp_frame;
1bafb621 215static int g_sp_frame;
a2c1d768 216static int g_stack_frame_used;
1bafb621 217static int g_stack_fsz;
4c45fa73 218static int g_ida_func_attr;
89ff3147 219static int g_allow_regfunc;
91977a1c 220#define ferr(op_, fmt, ...) do { \
63df67be 221 printf("error:%s:#%zd: '%s': " fmt, g_func, (op_) - ops, \
850c9265 222 dump_op(op_), ##__VA_ARGS__); \
33c35af6 223 fcloseall(); \
91977a1c 224 exit(1); \
225} while (0)
de50b98b 226#define fnote(op_, fmt, ...) \
227 printf("error:%s:#%zd: '%s': " fmt, g_func, (op_) - ops, \
228 dump_op(op_), ##__VA_ARGS__)
91977a1c 229
230#define MAX_REGS 8
231
232const char *regs_r32[] = { "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp" };
233const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" };
234const char *regs_r8l[] = { "al", "bl", "cl", "dl" };
235const char *regs_r8h[] = { "ah", "bh", "ch", "dh" };
236
237enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP };
238
69a3cdfc 239// possible basic comparison types (without inversion)
240enum parsed_flag_op {
241 PFO_O, // 0 OF=1
242 PFO_C, // 2 CF=1
243 PFO_Z, // 4 ZF=1
244 PFO_BE, // 6 CF=1||ZF=1
245 PFO_S, // 8 SF=1
246 PFO_P, // a PF=1
247 PFO_L, // c SF!=OF
248 PFO_LE, // e ZF=1||SF!=OF
249};
250
251static const char *parsed_flag_op_names[] = {
252 "o", "c", "z", "be", "s", "p", "l", "le"
253};
254
91977a1c 255static int char_array_i(const char *array[], size_t len, const char *s)
256{
257 int i;
c36e914d 258
91977a1c 259 for (i = 0; i < len; i++)
260 if (IS(s, array[i]))
261 return i;
c36e914d 262
91977a1c 263 return -1;
264}
265
63df67be 266static void printf_number(char *buf, size_t buf_size,
267 unsigned long number)
91977a1c 268{
5101a5f9 269 // output in C-friendly form
270 snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
271}
91977a1c 272
5101a5f9 273static int parse_reg(enum opr_lenmod *reg_lmod, const char *s)
274{
275 int reg;
91977a1c 276
5101a5f9 277 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s);
278 if (reg >= 0) {
279 *reg_lmod = OPLM_DWORD;
280 return reg;
91977a1c 281 }
5101a5f9 282 reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s);
283 if (reg >= 0) {
284 *reg_lmod = OPLM_WORD;
285 return reg;
286 }
287 reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s);
288 if (reg >= 0) {
289 *reg_lmod = OPLM_BYTE;
290 return reg;
291 }
292 reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s);
293 if (reg >= 0) {
294 *reg_lmod = OPLM_BYTE;
295 return reg;
850c9265 296 }
297
298 return -1;
91977a1c 299}
300
5101a5f9 301static int parse_indmode(char *name, int *regmask, int need_c_cvt)
302{
303 enum opr_lenmod lmod;
304 char cvtbuf[256];
305 char *d = cvtbuf;
306 char *s = name;
307 char w[64];
308 long number;
309 int reg;
310 int c = 0;
311
312 *d = 0;
313
314 while (*s != 0) {
315 d += strlen(d);
316 while (my_isblank(*s))
317 s++;
318 for (; my_issep(*s); d++, s++)
319 *d = *s;
320 while (my_isblank(*s))
321 s++;
322 *d = 0;
323
87bf6cec 324 // skip 'ds:' prefix
39b168b8 325 if (IS_START(s, "ds:"))
87bf6cec 326 s += 3;
327
5101a5f9 328 s = next_idt(w, sizeof(w), s);
329 if (w[0] == 0)
330 break;
331 c++;
332
333 reg = parse_reg(&lmod, w);
334 if (reg >= 0) {
335 *regmask |= 1 << reg;
336 goto pass;
337 }
338
339 if ('0' <= w[0] && w[0] <= '9') {
340 number = parse_number(w);
341 printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number);
342 continue;
343 }
344
345 // probably some label/identifier - pass
346
347pass:
348 snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w);
349 }
350
351 if (need_c_cvt)
352 strcpy(name, cvtbuf);
353
354 return c;
355}
356
4c45fa73 357static int is_reg_in_str(const char *s)
358{
359 int i;
360
361 if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
362 return 0;
363
364 for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
365 if (!strncmp(s, regs_r32[i], 3))
366 return 1;
367
368 return 0;
369}
370
371static const char *parse_stack_el(const char *name, char *extra_reg)
d4e3b5db 372{
4c45fa73 373 const char *p, *p2, *s;
1bafb621 374 char *endp = NULL;
d4e3b5db 375 char buf[32];
1bafb621 376 long val;
d4e3b5db 377 int len;
378
4c45fa73 379 p = name;
380 if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
381 p += 4;
382 if (extra_reg != NULL) {
383 strncpy(extra_reg, name, 3);
384 extra_reg[4] = 0;
385 }
d4e3b5db 386 }
4c45fa73 387
388 if (IS_START(p, "ebp+")) {
389 p += 4;
390
391 p2 = strchr(p, '+');
392 if (p2 != NULL && is_reg_in_str(p)) {
393 if (extra_reg != NULL) {
394 strncpy(extra_reg, p, p2 - p);
395 extra_reg[p2 - p] = 0;
396 }
397 p = p2 + 1;
398 }
399
400 if (!('0' <= *p && *p <= '9'))
401 return p;
402
403 return NULL;
404 }
405
39b168b8 406 if (!IS_START(name, "esp+"))
d4e3b5db 407 return NULL;
408
409 p = strchr(name + 4, '+');
410 if (p) {
1bafb621 411 // must be a number after esp+, already converted to 0x..
412 s = name + 4;
413 if (!('0' <= *s && *s <= '9')) {
591721d7 414 aerr("%s nan?\n", __func__);
d4e3b5db 415 return NULL;
1bafb621 416 }
417 if (s[0] == '0' && s[1] == 'x')
418 s += 2;
419 len = p - s;
d4e3b5db 420 if (len < sizeof(buf) - 1) {
1bafb621 421 strncpy(buf, s, len);
d4e3b5db 422 buf[len] = 0;
1bafb621 423 val = strtol(buf, &endp, 16);
424 if (val == 0 || *endp != 0) {
425 aerr("%s num parse fail for '%s'\n", __func__, buf);
426 return NULL;
427 }
d4e3b5db 428 }
429 p++;
430 }
431 else
432 p = name + 4;
433
434 if ('0' <= *p && *p <= '9')
435 return NULL;
436
437 return p;
438}
439
850c9265 440static int guess_lmod_from_name(struct parsed_opr *opr)
441{
442 if (!strncmp(opr->name, "dword_", 6)) {
443 opr->lmod = OPLM_DWORD;
444 return 1;
445 }
446 if (!strncmp(opr->name, "word_", 5)) {
447 opr->lmod = OPLM_WORD;
448 return 1;
449 }
450 if (!strncmp(opr->name, "byte_", 5)) {
451 opr->lmod = OPLM_BYTE;
452 return 1;
453 }
454 return 0;
455}
456
3ebea2cf 457static int guess_lmod_from_c_type(enum opr_lenmod *lmod,
458 const struct parsed_type *c_type)
06c5d854 459{
06c5d854 460 static const char *dword_types[] = {
da87ae38 461 "int", "_DWORD", "UINT_PTR", "DWORD",
4741fdfe 462 "WPARAM", "LPARAM", "UINT", "__int32",
463 "LONG", "HIMC",
06c5d854 464 };
465 static const char *word_types[] = {
89ff3147 466 "uint16_t", "int16_t", "_WORD",
2b43685d 467 "unsigned __int16", "__int16",
06c5d854 468 };
469 static const char *byte_types[] = {
2b43685d 470 "uint8_t", "int8_t", "char",
89ff3147 471 "unsigned __int8", "__int8", "BYTE", "_BYTE",
4741fdfe 472 "CHAR", "_UNKNOWN",
06c5d854 473 };
3ebea2cf 474 const char *n;
06c5d854 475 int i;
476
3ebea2cf 477 if (c_type->is_ptr) {
478 *lmod = OPLM_DWORD;
06c5d854 479 return 1;
480 }
481
3ebea2cf 482 n = skip_type_mod(c_type->name);
06c5d854 483
3ebea2cf 484 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
485 if (IS(n, dword_types[i])) {
486 *lmod = OPLM_DWORD;
06c5d854 487 return 1;
488 }
489 }
490
491 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
3ebea2cf 492 if (IS(n, word_types[i])) {
493 *lmod = OPLM_WORD;
06c5d854 494 return 1;
495 }
496 }
497
498 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
3ebea2cf 499 if (IS(n, byte_types[i])) {
500 *lmod = OPLM_BYTE;
06c5d854 501 return 1;
502 }
503 }
504
06c5d854 505 return 0;
506}
507
4c45fa73 508static enum opr_type lmod_from_directive(const char *d)
509{
510 if (IS(d, "dd"))
511 return OPLM_DWORD;
512 else if (IS(d, "dw"))
513 return OPLM_WORD;
514 else if (IS(d, "db"))
515 return OPLM_BYTE;
516
517 aerr("unhandled directive: '%s'\n", d);
518 return OPLM_UNSPEC;
519}
520
5101a5f9 521static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
522 int *regmask)
523{
524 opr->type = OPT_REG;
525 opr->reg = reg;
526 opr->lmod = lmod;
527 *regmask |= 1 << reg;
528}
529
39b168b8 530static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
531 int *extra_offs);
87bf6cec 532
69a3cdfc 533static int parse_operand(struct parsed_opr *opr,
534 int *regmask, int *regmask_indirect,
1bafb621 535 char words[16][256], int wordc, int w, unsigned int op_flags)
c36e914d 536{
bd96f656 537 const struct parsed_proto *pp;
850c9265 538 enum opr_lenmod tmplmod;
63df67be 539 unsigned long number;
89ff3147 540 char buf[256];
850c9265 541 int ret, len;
1bafb621 542 int wordc_in;
89ff3147 543 char *p;
850c9265 544 int i;
c36e914d 545
1bafb621 546 if (w >= wordc)
547 aerr("parse_operand w %d, wordc %d\n", w, wordc);
c36e914d 548
1bafb621 549 opr->reg = xUNSPEC;
91977a1c 550
1bafb621 551 for (i = w; i < wordc; i++) {
552 len = strlen(words[i]);
553 if (words[i][len - 1] == ',') {
554 words[i][len - 1] = 0;
555 wordc = i + 1;
556 break;
557 }
558 }
c36e914d 559
1bafb621 560 wordc_in = wordc - w;
c36e914d 561
1bafb621 562 if ((op_flags & OPF_JMP) && wordc_in > 0
563 && !('0' <= words[w][0] && words[w][0] <= '9'))
564 {
565 const char *label = NULL;
566
567 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
568 && IS(words[w + 1], "ptr"))
569 label = words[w + 2];
570 else if (wordc_in == 2 && IS(words[w], "short"))
571 label = words[w + 1];
572 else if (wordc_in == 1
573 && strchr(words[w], '[') == NULL
574 && parse_reg(&tmplmod, words[w]) < 0)
575 label = words[w];
576
577 if (label != NULL) {
578 opr->type = OPT_LABEL;
ddaf8bd7 579 if (IS_START(label, "ds:")) {
580 opr->had_ds = 1;
39b168b8 581 label += 3;
ddaf8bd7 582 }
1bafb621 583 strcpy(opr->name, label);
584 return wordc;
585 }
586 }
c36e914d 587
1bafb621 588 if (wordc_in >= 3) {
589 if (IS(words[w + 1], "ptr")) {
590 if (IS(words[w], "dword"))
591 opr->lmod = OPLM_DWORD;
592 else if (IS(words[w], "word"))
593 opr->lmod = OPLM_WORD;
594 else if (IS(words[w], "byte"))
595 opr->lmod = OPLM_BYTE;
596 else
597 aerr("type parsing failed\n");
598 w += 2;
599 wordc_in = wordc - w;
600 }
601 }
602
39b168b8 603 if (wordc_in == 2) {
604 if (IS(words[w], "offset")) {
605 opr->type = OPT_OFFSET;
606 strcpy(opr->name, words[w + 1]);
607 return wordc;
608 }
609 if (IS(words[w], "(offset")) {
89ff3147 610 p = strchr(words[w + 1], ')');
39b168b8 611 if (p == NULL)
612 aerr("parse of bracketed offset failed\n");
613 *p = 0;
614 opr->type = OPT_OFFSET;
615 strcpy(opr->name, words[w + 1]);
616 return wordc;
617 }
1bafb621 618 }
c36e914d 619
1bafb621 620 if (wordc_in != 1)
850c9265 621 aerr("parse_operand 1 word expected\n");
c36e914d 622
ddaf8bd7 623 if (IS_START(words[w], "ds:")) {
624 opr->had_ds = 1;
89ff3147 625 memmove(words[w], words[w] + 3, strlen(words[w]) - 2);
ddaf8bd7 626 }
89ff3147 627 strcpy(opr->name, words[w]);
c36e914d 628
850c9265 629 if (words[w][0] == '[') {
630 opr->type = OPT_REGMEM;
631 ret = sscanf(words[w], "[%[^]]]", opr->name);
632 if (ret != 1)
633 aerr("[] parse failure\n");
87bf6cec 634
5101a5f9 635 parse_indmode(opr->name, regmask_indirect, 1);
4c45fa73 636 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL)) {
87bf6cec 637 // might be an equ
d4e3b5db 638 struct parsed_equ *eq =
4c45fa73 639 equ_find(NULL, parse_stack_el(opr->name, NULL), &i);
87bf6cec 640 if (eq)
641 opr->lmod = eq->lmod;
642 }
850c9265 643 return wordc;
644 }
645 else if (strchr(words[w], '[')) {
646 // label[reg] form
89ff3147 647 p = strchr(words[w], '[');
850c9265 648 opr->type = OPT_REGMEM;
89ff3147 649 parse_indmode(p, regmask_indirect, 0);
650 strncpy(buf, words[w], p - words[w]);
651 buf[p - words[w]] = 0;
652 pp = proto_parse(g_fhdr, buf, 1);
653 goto do_label;
850c9265 654 }
655 else if (('0' <= words[w][0] && words[w][0] <= '9')
656 || words[w][0] == '-')
657 {
5101a5f9 658 number = parse_number(words[w]);
91977a1c 659 opr->type = OPT_CONST;
5101a5f9 660 opr->val = number;
661 printf_number(opr->name, sizeof(opr->name), number);
91977a1c 662 return wordc;
850c9265 663 }
c36e914d 664
5101a5f9 665 ret = parse_reg(&tmplmod, opr->name);
666 if (ret >= 0) {
667 setup_reg_opr(opr, ret, tmplmod, regmask);
850c9265 668 return wordc;
669 }
670
671 // most likely var in data segment
672 opr->type = OPT_LABEL;
36595fd2 673 pp = proto_parse(g_fhdr, opr->name, 0);
89ff3147 674
675do_label:
bd96f656 676 if (pp != NULL) {
1cd4a663 677 if (pp->is_fptr || pp->is_func) {
06c5d854 678 opr->lmod = OPLM_DWORD;
679 opr->is_ptr = 1;
680 }
2b43685d 681 else {
840257f6 682 tmplmod = OPLM_UNSPEC;
bd96f656 683 if (!guess_lmod_from_c_type(&tmplmod, &pp->type))
2b43685d 684 anote("unhandled C type '%s' for '%s'\n",
bd96f656 685 pp->type.name, opr->name);
2b43685d 686
a3684be1 687 if (opr->lmod == OPLM_UNSPEC) {
2b43685d 688 opr->lmod = tmplmod;
a3684be1 689 opr->type_from_var = 1;
690 }
2b43685d 691 else if (opr->lmod != tmplmod) {
692 opr->size_mismatch = 1;
693 if (tmplmod < opr->lmod)
694 opr->size_lt = 1;
695 }
a652aa9f 696 opr->is_ptr = pp->type.is_ptr;
3ebea2cf 697 }
bd96f656 698 opr->is_array = pp->type.is_array;
06c5d854 699 }
700
850c9265 701 if (opr->lmod == OPLM_UNSPEC)
702 guess_lmod_from_name(opr);
850c9265 703 return wordc;
c36e914d 704}
705
33c35af6 706static const struct {
707 const char *name;
708 unsigned int flags;
709} pref_table[] = {
710 { "rep", OPF_REP },
7ba45c34 711 { "repe", OPF_REP|OPF_REPZ },
712 { "repz", OPF_REP|OPF_REPZ },
713 { "repne", OPF_REP|OPF_REPNZ },
714 { "repnz", OPF_REP|OPF_REPNZ },
33c35af6 715};
716
5c024ef7 717#define OPF_CJMP_CC (OPF_JMP|OPF_CJMP|OPF_CC)
718
c36e914d 719static const struct {
850c9265 720 const char *name;
721 enum op_op op;
69a3cdfc 722 unsigned int minopr;
723 unsigned int maxopr;
724 unsigned int flags;
c36e914d 725} op_table[] = {
33c35af6 726 { "nop", OP_NOP, 0, 0, 0 },
69a3cdfc 727 { "push", OP_PUSH, 1, 1, 0 },
728 { "pop", OP_POP, 1, 1, OPF_DATA },
591721d7 729 { "leave",OP_LEAVE, 0, 0, OPF_DATA },
69a3cdfc 730 { "mov" , OP_MOV, 2, 2, OPF_DATA },
731 { "lea", OP_LEA, 2, 2, OPF_DATA },
732 { "movzx",OP_MOVZX, 2, 2, OPF_DATA },
733 { "movsx",OP_MOVSX, 2, 2, OPF_DATA },
108e9fe3 734 { "xchg", OP_XCHG, 2, 2, OPF_DATA },
69a3cdfc 735 { "not", OP_NOT, 1, 1, OPF_DATA },
5101a5f9 736 { "cdq", OP_CDQ, 0, 0, OPF_DATA },
33c35af6 737 { "stosb",OP_STOS, 0, 0, OPF_DATA },
738 { "stosw",OP_STOS, 0, 0, OPF_DATA },
739 { "stosd",OP_STOS, 0, 0, OPF_DATA },
d4e3b5db 740 { "movsb",OP_MOVS, 0, 0, OPF_DATA },
741 { "movsw",OP_MOVS, 0, 0, OPF_DATA },
742 { "movsd",OP_MOVS, 0, 0, OPF_DATA },
7ba45c34 743 { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
744 { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
745 { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
591721d7 746 { "scasb",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
747 { "scasw",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
748 { "scasd",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
749 { "std", OP_STD, 0, 0, OPF_DATA }, // special flag
750 { "cld", OP_CLD, 0, 0, OPF_DATA },
69a3cdfc 751 { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
752 { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
753 { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
754 { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS },
755 { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS },
756 { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
757 { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
758 { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
759 { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
d4e3b5db 760 { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
761 { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
cb090db0 762 { "rcl", OP_RCL, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
763 { "rcr", OP_RCR, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
5101a5f9 764 { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
69a3cdfc 765 { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
766 { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
767 { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS },
5101a5f9 768 { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS },
769 { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS },
69a3cdfc 770 { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS },
5101a5f9 771 { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS },
772 { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS },
69a3cdfc 773 { "test", OP_TEST, 2, 2, OPF_FLAGS },
774 { "cmp", OP_CMP, 2, 2, OPF_FLAGS },
a3684be1 775 { "retn", OP_RET, 0, 1, OPF_TAIL },
de50b98b 776 { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS },
69a3cdfc 777 { "jmp", OP_JMP, 1, 1, OPF_JMP },
5c024ef7 778 { "jecxz",OP_JECXZ, 1, 1, OPF_JMP|OPF_CJMP },
779 { "jo", OP_JO, 1, 1, OPF_CJMP_CC }, // 70 OF=1
780 { "jno", OP_JNO, 1, 1, OPF_CJMP_CC }, // 71 OF=0
781 { "jc", OP_JC, 1, 1, OPF_CJMP_CC }, // 72 CF=1
782 { "jb", OP_JC, 1, 1, OPF_CJMP_CC }, // 72
783 { "jnc", OP_JNC, 1, 1, OPF_CJMP_CC }, // 73 CF=0
784 { "jnb", OP_JNC, 1, 1, OPF_CJMP_CC }, // 73
785 { "jae", OP_JNC, 1, 1, OPF_CJMP_CC }, // 73
786 { "jz", OP_JZ, 1, 1, OPF_CJMP_CC }, // 74 ZF=1
787 { "je", OP_JZ, 1, 1, OPF_CJMP_CC }, // 74
788 { "jnz", OP_JNZ, 1, 1, OPF_CJMP_CC }, // 75 ZF=0
789 { "jne", OP_JNZ, 1, 1, OPF_CJMP_CC }, // 75
790 { "jbe", OP_JBE, 1, 1, OPF_CJMP_CC }, // 76 CF=1 || ZF=1
791 { "jna", OP_JBE, 1, 1, OPF_CJMP_CC }, // 76
792 { "ja", OP_JA, 1, 1, OPF_CJMP_CC }, // 77 CF=0 && ZF=0
793 { "jnbe", OP_JA, 1, 1, OPF_CJMP_CC }, // 77
794 { "js", OP_JS, 1, 1, OPF_CJMP_CC }, // 78 SF=1
795 { "jns", OP_JNS, 1, 1, OPF_CJMP_CC }, // 79 SF=0
796 { "jp", OP_JP, 1, 1, OPF_CJMP_CC }, // 7a PF=1
797 { "jpe", OP_JP, 1, 1, OPF_CJMP_CC }, // 7a
798 { "jnp", OP_JNP, 1, 1, OPF_CJMP_CC }, // 7b PF=0
799 { "jpo", OP_JNP, 1, 1, OPF_CJMP_CC }, // 7b
800 { "jl", OP_JL, 1, 1, OPF_CJMP_CC }, // 7c SF!=OF
801 { "jnge", OP_JL, 1, 1, OPF_CJMP_CC }, // 7c
802 { "jge", OP_JGE, 1, 1, OPF_CJMP_CC }, // 7d SF=OF
803 { "jnl", OP_JGE, 1, 1, OPF_CJMP_CC }, // 7d
804 { "jle", OP_JLE, 1, 1, OPF_CJMP_CC }, // 7e ZF=1 || SF!=OF
805 { "jng", OP_JLE, 1, 1, OPF_CJMP_CC }, // 7e
806 { "jg", OP_JG, 1, 1, OPF_CJMP_CC }, // 7f ZF=0 && SF=OF
807 { "jnle", OP_JG, 1, 1, OPF_CJMP_CC }, // 7f
5101a5f9 808 { "seto", OP_JO, 1, 1, OPF_DATA|OPF_CC },
809 { "setno", OP_JNO, 1, 1, OPF_DATA|OPF_CC },
810 { "setc", OP_JC, 1, 1, OPF_DATA|OPF_CC },
811 { "setb", OP_JC, 1, 1, OPF_DATA|OPF_CC },
812 { "setnc", OP_JNC, 1, 1, OPF_DATA|OPF_CC },
813 { "setae", OP_JNC, 1, 1, OPF_DATA|OPF_CC },
814 { "setz", OP_JZ, 1, 1, OPF_DATA|OPF_CC },
815 { "sete", OP_JZ, 1, 1, OPF_DATA|OPF_CC },
816 { "setnz", OP_JNZ, 1, 1, OPF_DATA|OPF_CC },
817 { "setne", OP_JNZ, 1, 1, OPF_DATA|OPF_CC },
818 { "setbe", OP_JBE, 1, 1, OPF_DATA|OPF_CC },
819 { "setna", OP_JBE, 1, 1, OPF_DATA|OPF_CC },
820 { "seta", OP_JA, 1, 1, OPF_DATA|OPF_CC },
821 { "setnbe", OP_JA, 1, 1, OPF_DATA|OPF_CC },
822 { "sets", OP_JS, 1, 1, OPF_DATA|OPF_CC },
823 { "setns", OP_JNS, 1, 1, OPF_DATA|OPF_CC },
824 { "setp", OP_JP, 1, 1, OPF_DATA|OPF_CC },
825 { "setpe", OP_JP, 1, 1, OPF_DATA|OPF_CC },
826 { "setnp", OP_JNP, 1, 1, OPF_DATA|OPF_CC },
827 { "setpo", OP_JNP, 1, 1, OPF_DATA|OPF_CC },
828 { "setl", OP_JL, 1, 1, OPF_DATA|OPF_CC },
829 { "setnge", OP_JL, 1, 1, OPF_DATA|OPF_CC },
830 { "setge", OP_JGE, 1, 1, OPF_DATA|OPF_CC },
831 { "setnl", OP_JGE, 1, 1, OPF_DATA|OPF_CC },
832 { "setle", OP_JLE, 1, 1, OPF_DATA|OPF_CC },
833 { "setng", OP_JLE, 1, 1, OPF_DATA|OPF_CC },
834 { "setg", OP_JG, 1, 1, OPF_DATA|OPF_CC },
835 { "setnle", OP_JG, 1, 1, OPF_DATA|OPF_CC },
69a3cdfc 836};
c36e914d 837
838static void parse_op(struct parsed_op *op, char words[16][256], int wordc)
839{
bfa4a6ee 840 enum opr_lenmod lmod = OPLM_UNSPEC;
33c35af6 841 int prefix_flags = 0;
69a3cdfc 842 int regmask_ind;
843 int regmask;
33c35af6 844 int op_w = 0;
91977a1c 845 int opr = 0;
33c35af6 846 int w = 0;
91977a1c 847 int i;
c36e914d 848
33c35af6 849 for (i = 0; i < ARRAY_SIZE(pref_table); i++) {
850 if (IS(words[w], pref_table[i].name)) {
851 prefix_flags = pref_table[i].flags;
852 break;
853 }
854 }
855
856 if (prefix_flags) {
857 if (wordc <= 1)
858 aerr("lone prefix: '%s'\n", words[0]);
859 w++;
860 }
861
862 op_w = w;
91977a1c 863 for (i = 0; i < ARRAY_SIZE(op_table); i++) {
33c35af6 864 if (IS(words[w], op_table[i].name))
69a3cdfc 865 break;
866 }
c36e914d 867
69a3cdfc 868 if (i == ARRAY_SIZE(op_table))
869 aerr("unhandled op: '%s'\n", words[0]);
33c35af6 870 w++;
c36e914d 871
69a3cdfc 872 op->op = op_table[i].op;
33c35af6 873 op->flags = op_table[i].flags | prefix_flags;
69a3cdfc 874 op->regmask_src = op->regmask_dst = 0;
875
876 for (opr = 0; opr < op_table[i].minopr; opr++) {
877 regmask = regmask_ind = 0;
878 w = parse_operand(&op->operand[opr], &regmask, &regmask_ind,
879 words, wordc, w, op->flags);
880
881 if (opr == 0 && (op->flags & OPF_DATA))
882 op->regmask_dst = regmask;
883 // for now, mark dst as src too
884 op->regmask_src |= regmask | regmask_ind;
885 }
c36e914d 886
69a3cdfc 887 for (; w < wordc && opr < op_table[i].maxopr; opr++) {
888 w = parse_operand(&op->operand[opr],
889 &op->regmask_src, &op->regmask_src,
890 words, wordc, w, op->flags);
91977a1c 891 }
c36e914d 892
91977a1c 893 if (w < wordc)
894 aerr("parse_op %s incomplete: %d/%d\n",
895 words[0], w, wordc);
5101a5f9 896
897 // special cases
898 op->operand_cnt = opr;
899 if (!strncmp(op_table[i].name, "set", 3))
900 op->operand[0].lmod = OPLM_BYTE;
901
902 // ops with implicit argumets
903 switch (op->op) {
904 case OP_CDQ:
905 op->operand_cnt = 2;
906 setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst);
907 setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src);
908 break;
909
33c35af6 910 case OP_STOS:
591721d7 911 case OP_SCAS:
33c35af6 912 if (op->operand_cnt != 0)
913 break;
591721d7 914 if (words[op_w][4] == 'b')
33c35af6 915 lmod = OPLM_BYTE;
591721d7 916 else if (words[op_w][4] == 'w')
33c35af6 917 lmod = OPLM_WORD;
591721d7 918 else if (words[op_w][4] == 'd')
33c35af6 919 lmod = OPLM_DWORD;
920 op->operand_cnt = 3;
d4e3b5db 921 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
922 setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src);
923 op->regmask_dst = op->regmask_src;
33c35af6 924 setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD, &op->regmask_src);
925 break;
926
d4e3b5db 927 case OP_MOVS:
7ba45c34 928 case OP_CMPS:
d4e3b5db 929 if (op->operand_cnt != 0)
930 break;
7ba45c34 931 if (words[op_w][4] == 'b')
d4e3b5db 932 lmod = OPLM_BYTE;
7ba45c34 933 else if (words[op_w][4] == 'w')
d4e3b5db 934 lmod = OPLM_WORD;
7ba45c34 935 else if (words[op_w][4] == 'd')
d4e3b5db 936 lmod = OPLM_DWORD;
937 op->operand_cnt = 3;
938 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
939 setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src);
940 setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src);
941 op->regmask_dst = op->regmask_src;
942 break;
943
108e9fe3 944 case OP_XCHG:
945 op->regmask_src |= op->regmask_dst;
946 op->regmask_dst |= op->regmask_src;
947 break;
948
5c024ef7 949 case OP_JECXZ:
950 op->operand_cnt = 1;
951 op->regmask_src = 1 << xCX;
952 op->operand[0].type = OPT_REG;
953 op->operand[0].reg = xCX;
954 op->operand[0].lmod = OPLM_DWORD;
955 break;
956
5101a5f9 957 case OP_IMUL:
958 if (op->operand_cnt != 1)
959 break;
960 // fallthrough
961 case OP_MUL:
962 // singleop mul
963 op->regmask_dst = (1 << xDX) | (1 << xAX);
964 op->regmask_src |= (1 << xAX);
965 if (op->operand[0].lmod == OPLM_UNSPEC)
966 op->operand[0].lmod = OPLM_DWORD;
967 break;
968
969 case OP_DIV:
970 case OP_IDIV:
971 // we could set up operands for edx:eax, but there is no real need to
972 // (see is_opr_modified())
973 regmask = (1 << xDX) | (1 << xAX);
974 op->regmask_dst = regmask;
975 op->regmask_src |= regmask;
976 if (op->operand[0].lmod == OPLM_UNSPEC)
977 op->operand[0].lmod = OPLM_DWORD;
978 break;
979
980 case OP_SHL:
981 case OP_SHR:
982 case OP_SAR:
d4e3b5db 983 case OP_ROL:
984 case OP_ROR:
5101a5f9 985 if (op->operand[1].lmod == OPLM_UNSPEC)
986 op->operand[1].lmod = OPLM_BYTE;
987 break;
988
7ba45c34 989 case OP_PUSH:
990 if (op->operand[0].lmod == OPLM_UNSPEC
991 && (op->operand[0].type == OPT_CONST
992 || op->operand[0].type == OPT_OFFSET
993 || op->operand[0].type == OPT_LABEL))
994 op->operand[0].lmod = OPLM_DWORD;
995 break;
996
3ebea2cf 997 // alignment
998 case OP_MOV:
999 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1000 && op->operand[0].reg == xDI && op->operand[1].reg == xDI)
1001 {
1002 op->flags |= OPF_RMD;
1003 }
1004 break;
1005
de50b98b 1006 case OP_LEA:
1007 if (op->operand[0].type == OPT_REG
1008 && op->operand[1].type == OPT_REGMEM)
1009 {
1010 char buf[16];
1011 snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name);
1012 if (IS(buf, op->operand[1].name))
1013 op->flags |= OPF_RMD;
1014 }
1015 break;
1016
89ff3147 1017 case OP_CALL:
1018 // trashed regs must be explicitly detected later
1019 op->regmask_dst = 0;
1020 break;
1021
5101a5f9 1022 default:
1023 break;
1024 }
c36e914d 1025}
1026
850c9265 1027static const char *op_name(enum op_op op)
1028{
1029 int i;
1030
1031 for (i = 0; i < ARRAY_SIZE(op_table); i++)
1032 if (op_table[i].op == op)
1033 return op_table[i].name;
1034
1035 return "???";
1036}
1037
1038// debug
1039static const char *dump_op(struct parsed_op *po)
1040{
1041 static char out[128];
1042 char *p = out;
1043 int i;
1044
4c45fa73 1045 if (po == NULL)
1046 return "???";
1047
850c9265 1048 snprintf(out, sizeof(out), "%s", op_name(po->op));
1049 for (i = 0; i < po->operand_cnt; i++) {
1050 p += strlen(p);
1051 if (i > 0)
1052 *p++ = ',';
1053 snprintf(p, sizeof(out) - (p - out),
1054 po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s",
1055 po->operand[i].name);
1056 }
1057
1058 return out;
1059}
1060
4c45fa73 1061static const char *lmod_type_u(struct parsed_op *po,
1062 enum opr_lenmod lmod)
1063{
1064 switch (lmod) {
1065 case OPLM_DWORD:
1066 return "u32";
1067 case OPLM_WORD:
1068 return "u16";
1069 case OPLM_BYTE:
1070 return "u8";
1071 default:
1072 ferr(po, "invalid lmod: %d\n", lmod);
1073 return "(_invalid_)";
1074 }
1075}
1076
d4e3b5db 1077static const char *lmod_cast_u(struct parsed_op *po,
1078 enum opr_lenmod lmod)
1079{
1080 switch (lmod) {
1081 case OPLM_DWORD:
1082 return "";
1083 case OPLM_WORD:
1084 return "(u16)";
1085 case OPLM_BYTE:
1086 return "(u8)";
1087 default:
1088 ferr(po, "invalid lmod: %d\n", lmod);
1089 return "(_invalid_)";
1090 }
1091}
1092
1093static const char *lmod_cast_u_ptr(struct parsed_op *po,
1094 enum opr_lenmod lmod)
1095{
1096 switch (lmod) {
1097 case OPLM_DWORD:
1098 return "*(u32 *)";
1099 case OPLM_WORD:
1100 return "*(u16 *)";
1101 case OPLM_BYTE:
1102 return "*(u8 *)";
1103 default:
1104 ferr(po, "invalid lmod: %d\n", lmod);
1105 return "(_invalid_)";
1106 }
1107}
1108
1109static const char *lmod_cast_s(struct parsed_op *po,
1110 enum opr_lenmod lmod)
1111{
1112 switch (lmod) {
1113 case OPLM_DWORD:
1114 return "(s32)";
1115 case OPLM_WORD:
1116 return "(s16)";
1117 case OPLM_BYTE:
1118 return "(s8)";
1119 default:
1120 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1121 return "(_invalid_)";
1122 }
1123}
1124
1125static const char *lmod_cast(struct parsed_op *po,
1126 enum opr_lenmod lmod, int is_signed)
1127{
1128 return is_signed ?
1129 lmod_cast_s(po, lmod) :
1130 lmod_cast_u(po, lmod);
1131}
1132
1133static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
1134{
1135 switch (lmod) {
1136 case OPLM_DWORD:
1137 return 4;
1138 case OPLM_WORD:
1139 return 2;
1140 case OPLM_BYTE:
1141 return 1;
1142 default:
1143 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1144 return 0;
1145 }
1146}
1147
91977a1c 1148static const char *opr_name(struct parsed_op *po, int opr_num)
c36e914d 1149{
91977a1c 1150 if (opr_num >= po->operand_cnt)
1151 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1152 return po->operand[opr_num].name;
c36e914d 1153}
1154
91977a1c 1155static unsigned int opr_const(struct parsed_op *po, int opr_num)
c36e914d 1156{
91977a1c 1157 if (opr_num >= po->operand_cnt)
1158 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1159 if (po->operand[opr_num].type != OPT_CONST)
1160 ferr(po, "opr %d: const expected\n", opr_num);
1161 return po->operand[opr_num].val;
1162}
c36e914d 1163
91977a1c 1164static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr)
1165{
1166 if ((unsigned int)popr->reg >= MAX_REGS)
1167 ferr(po, "invalid reg: %d\n", popr->reg);
1168 return regs_r32[popr->reg];
1169}
c36e914d 1170
a2c1d768 1171// cast1 is the "final" cast
1172static const char *simplify_cast(const char *cast1, const char *cast2)
1173{
1174 static char buf[256];
1175
1176 if (cast1[0] == 0)
1177 return cast2;
1178 if (cast2[0] == 0)
1179 return cast1;
1180 if (IS(cast1, cast2))
1181 return cast1;
1182 if (IS(cast1, "(s8)") && IS(cast2, "(u8)"))
1183 return cast1;
1184 if (IS(cast1, "(s16)") && IS(cast2, "(u16)"))
1185 return cast1;
1186 if (IS(cast1, "(u8)") && IS_START(cast2, "*(u8 *)"))
1187 return cast2;
1188 if (IS(cast1, "(u16)") && IS_START(cast2, "*(u16 *)"))
1189 return cast2;
1cd4a663 1190 if (strchr(cast1, '*') && IS_START(cast2, "(u32)"))
1191 return cast1;
a2c1d768 1192
1193 snprintf(buf, sizeof(buf), "%s%s", cast1, cast2);
1194 return buf;
1195}
1196
1197static const char *simplify_cast_num(const char *cast, unsigned int val)
1198{
1199 if (IS(cast, "(u8)") && val < 0x100)
1200 return "";
1201 if (IS(cast, "(s8)") && val < 0x80)
1202 return "";
1203 if (IS(cast, "(u16)") && val < 0x10000)
1204 return "";
1205 if (IS(cast, "(s16)") && val < 0x8000)
1206 return "";
1207 if (IS(cast, "(s32)") && val < 0x80000000)
1208 return "";
1209
1210 return cast;
1211}
1212
39b168b8 1213static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
1214 int *extra_offs)
91977a1c 1215{
39b168b8 1216 const char *p;
1217 char *endp;
1218 int namelen;
850c9265 1219 int i;
1220
39b168b8 1221 *extra_offs = 0;
1222 namelen = strlen(name);
1223
1224 p = strchr(name, '+');
1225 if (p != NULL) {
1226 namelen = p - name;
1227 if (namelen <= 0)
1228 ferr(po, "equ parse failed for '%s'\n", name);
1229
1230 if (IS_START(p, "0x"))
1231 p += 2;
1232 *extra_offs = strtol(p, &endp, 16);
1233 if (*endp != 0)
1234 ferr(po, "equ parse failed for '%s'\n", name);
1235 }
1236
850c9265 1237 for (i = 0; i < g_eqcnt; i++)
39b168b8 1238 if (strncmp(g_eqs[i].name, name, namelen) == 0
1239 && g_eqs[i].name[namelen] == 0)
850c9265 1240 break;
87bf6cec 1241 if (i >= g_eqcnt) {
1242 if (po != NULL)
1243 ferr(po, "unresolved equ name: '%s'\n", name);
1244 return NULL;
1245 }
850c9265 1246
1247 return &g_eqs[i];
1248}
1249
1cd4a663 1250static int is_stack_access(struct parsed_op *po,
1251 const struct parsed_opr *popr)
850c9265 1252{
1cd4a663 1253 return (parse_stack_el(popr->name, NULL)
1254 || (g_bp_frame && !(po->flags & OPF_EBP_S)
1255 && IS_START(popr->name, "ebp")));
1256}
1257
1258static void parse_stack_access(struct parsed_op *po,
1259 const char *name, char *ofs_reg, int *offset_out,
1260 int *stack_ra_out, const char **bp_arg_out)
1261{
1262 const char *bp_arg = "";
1263 const char *p = NULL;
91977a1c 1264 struct parsed_equ *eq;
4f12f671 1265 char *endp = NULL;
d4e3b5db 1266 int stack_ra = 0;
39b168b8 1267 int offset = 0;
91977a1c 1268
1cd4a663 1269 ofs_reg[0] = 0;
a3684be1 1270
1271 if (IS_START(name, "ebp-")
1272 || (IS_START(name, "ebp+") && '0' <= name[4] && name[4] <= '9'))
1273 {
1274 p = name + 4;
1275 if (IS_START(p, "0x"))
1276 p += 2;
1277 offset = strtoul(p, &endp, 16);
1278 if (name[3] == '-')
1279 offset = -offset;
1280 if (*endp != 0)
1281 ferr(po, "ebp- parse of '%s' failed\n", name);
1282 }
1283 else {
4f12f671 1284 bp_arg = parse_stack_el(name, ofs_reg);
1285 snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
1286 eq = equ_find(po, bp_arg, &offset);
1287 if (eq == NULL)
1288 ferr(po, "detected but missing eq\n");
1289 offset += eq->offset;
1290 }
39b168b8 1291
d4e3b5db 1292 if (!strncmp(name, "ebp", 3))
1293 stack_ra = 4;
1294
94d447fb 1295 if (ofs_reg[0] == 0 && stack_ra <= offset && offset < stack_ra + 4)
39b168b8 1296 ferr(po, "reference to ra? %d %d\n", offset, stack_ra);
d4e3b5db 1297
1cd4a663 1298 *offset_out = offset;
1299 *stack_ra_out = stack_ra;
1300 if (bp_arg_out)
1301 *bp_arg_out = bp_arg;
1302}
1303
1304static void stack_frame_access(struct parsed_op *po,
1305 struct parsed_opr *popr, char *buf, size_t buf_size,
1306 const char *name, const char *cast, int is_src, int is_lea)
1307{
1308 enum opr_lenmod tmp_lmod = OPLM_UNSPEC;
1309 const char *prefix = "";
1310 const char *bp_arg = NULL;
1311 char ofs_reg[16] = { 0, };
1312 int i, arg_i, arg_s;
1313 int unaligned = 0;
1314 int stack_ra = 0;
1315 int offset = 0;
1316 int sf_ofs;
1317 int lim;
1318
1319 if (po->flags & OPF_EBP_S)
1320 ferr(po, "stack_frame_access while ebp is scratch\n");
1321
1322 parse_stack_access(po, name, ofs_reg, &offset, &stack_ra, &bp_arg);
1323
4f12f671 1324 if (offset > stack_ra)
1325 {
39b168b8 1326 arg_i = (offset - stack_ra - 4) / 4;
bd96f656 1327 if (arg_i < 0 || arg_i >= g_func_pp->argc_stack)
4f12f671 1328 {
bd96f656 1329 if (g_func_pp->is_vararg
1330 && arg_i == g_func_pp->argc_stack && is_lea)
1331 {
4f12f671 1332 // should be va_list
1333 if (cast[0] == 0)
1334 cast = "(u32)";
1335 snprintf(buf, buf_size, "%sap", cast);
1336 return;
1337 }
1338 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1339 offset, bp_arg, arg_i);
1340 }
4c45fa73 1341 if (ofs_reg[0] != 0)
7ba45c34 1342 ferr(po, "offset reg on arg access?\n");
91977a1c 1343
bd96f656 1344 for (i = arg_s = 0; i < g_func_pp->argc; i++) {
1345 if (g_func_pp->arg[i].reg != NULL)
91977a1c 1346 continue;
1347 if (arg_s == arg_i)
1348 break;
1349 arg_s++;
1350 }
bd96f656 1351 if (i == g_func_pp->argc)
91977a1c 1352 ferr(po, "arg %d not in prototype?\n", arg_i);
850c9265 1353
bd96f656 1354 popr->is_ptr = g_func_pp->arg[i].type.is_ptr;
de50b98b 1355
1356 switch (popr->lmod)
7ba45c34 1357 {
1358 case OPLM_BYTE:
4f12f671 1359 if (is_lea)
1360 ferr(po, "lea/byte to arg?\n");
7ba45c34 1361 if (is_src && (offset & 3) == 0)
a2c1d768 1362 snprintf(buf, buf_size, "%sa%d",
1363 simplify_cast(cast, "(u8)"), i + 1);
7ba45c34 1364 else
a2c1d768 1365 snprintf(buf, buf_size, "%sBYTE%d(a%d)",
1366 cast, offset & 3, i + 1);
7ba45c34 1367 break;
1368
1369 case OPLM_WORD:
4f12f671 1370 if (is_lea)
1371 ferr(po, "lea/word to arg?\n");
a3684be1 1372 if (offset & 1) {
1373 unaligned = 1;
1374 if (!is_src) {
1375 if (offset & 2)
1376 ferr(po, "problematic arg store\n");
a2c1d768 1377 snprintf(buf, buf_size, "%s((char *)&a%d + 1)",
1378 simplify_cast(cast, "*(u16 *)"), i + 1);
a3684be1 1379 }
1380 else
1381 ferr(po, "unaligned arg word load\n");
1382 }
1383 else if (is_src && (offset & 2) == 0)
a2c1d768 1384 snprintf(buf, buf_size, "%sa%d",
1385 simplify_cast(cast, "(u16)"), i + 1);
7ba45c34 1386 else
a2c1d768 1387 snprintf(buf, buf_size, "%s%sWORD(a%d)",
1388 cast, (offset & 2) ? "HI" : "LO", i + 1);
7ba45c34 1389 break;
1390
1391 case OPLM_DWORD:
3ebea2cf 1392 if (cast[0])
1393 prefix = cast;
1394 else if (is_src)
1395 prefix = "(u32)";
a3684be1 1396
de50b98b 1397 if (offset & 3) {
a3684be1 1398 unaligned = 1;
2b43685d 1399 if (is_lea)
1400 snprintf(buf, buf_size, "(u32)&a%d + %d",
1401 i + 1, offset & 3);
a3684be1 1402 else if (!is_src)
1403 ferr(po, "unaligned arg store\n");
1404 else {
1405 // mov edx, [ebp+arg_4+2]; movsx ecx, dx
2b43685d 1406 snprintf(buf, buf_size, "%s(a%d >> %d)",
1407 prefix, i + 1, (offset & 3) * 8);
a3684be1 1408 }
de50b98b 1409 }
1410 else {
1411 snprintf(buf, buf_size, "%s%sa%d",
1412 prefix, is_lea ? "&" : "", i + 1);
1413 }
7ba45c34 1414 break;
1415
1416 default:
de50b98b 1417 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
7ba45c34 1418 }
1419
a3684be1 1420 if (unaligned)
1421 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1422
7ba45c34 1423 // common problem
bd96f656 1424 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type);
a2c1d768 1425 if (tmp_lmod != OPLM_DWORD
1426 && (unaligned || (!is_src && tmp_lmod < popr->lmod)))
1427 {
1428 ferr(po, "bp_arg arg%d/w offset %d and type '%s' is too small\n",
1429 i + 1, offset, g_func_pp->arg[i].type.name);
1430 }
4741fdfe 1431 // can't check this because msvc likes to reuse
1432 // arg space for scratch..
1433 //if (popr->is_ptr && popr->lmod != OPLM_DWORD)
1434 // ferr(po, "bp_arg arg%d: non-dword ptr access\n", i + 1);
91977a1c 1435 }
4f12f671 1436 else
1437 {
1bafb621 1438 if (g_stack_fsz == 0)
1439 ferr(po, "stack var access without stackframe\n");
a2c1d768 1440 g_stack_frame_used = 1;
850c9265 1441
39b168b8 1442 sf_ofs = g_stack_fsz + offset;
de50b98b 1443 lim = (ofs_reg[0] != 0) ? -4 : 0;
1444 if (offset > 0 || sf_ofs < lim)
39b168b8 1445 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
850c9265 1446
1447 if (is_lea)
33c35af6 1448 prefix = "(u32)&";
3ebea2cf 1449 else
1450 prefix = cast;
850c9265 1451
de50b98b 1452 switch (popr->lmod)
850c9265 1453 {
1454 case OPLM_BYTE:
7ba45c34 1455 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1456 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
850c9265 1457 break;
7ba45c34 1458
850c9265 1459 case OPLM_WORD:
7ba45c34 1460 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1461 // known unaligned or possibly unaligned
1462 strcat(g_comment, " unaligned");
1463 if (prefix[0] == 0)
1464 prefix = "*(u16 *)&";
1465 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1466 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1467 break;
1468 }
1469 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
850c9265 1470 break;
7ba45c34 1471
850c9265 1472 case OPLM_DWORD:
7ba45c34 1473 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1474 // known unaligned or possibly unaligned
1475 strcat(g_comment, " unaligned");
1476 if (prefix[0] == 0)
1477 prefix = "*(u32 *)&";
1478 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1479 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1480 break;
1481 }
1482 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
850c9265 1483 break;
7ba45c34 1484
850c9265 1485 default:
de50b98b 1486 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
850c9265 1487 }
91977a1c 1488 }
1489}
c36e914d 1490
89ff3147 1491static void check_func_pp(struct parsed_op *po,
1492 const struct parsed_proto *pp, const char *pfx)
1493{
b74c31e3 1494 char buf[256];
1495
89ff3147 1496 if (pp->argc_reg != 0) {
b74c31e3 1497 if (/*!g_allow_regfunc &&*/ !pp->is_fastcall) {
1498 pp_print(buf, sizeof(buf), pp);
1499 ferr(po, "%s: unexpected reg arg in icall: %s\n", pfx, buf);
1500 }
89ff3147 1501 if (pp->argc_stack > 0 && pp->argc_reg != 2)
1502 ferr(po, "%s: %d reg arg(s) with %d stack arg(s)\n",
1503 pfx, pp->argc_reg, pp->argc_stack);
1504 }
1505}
1506
3ebea2cf 1507static void check_label_read_ref(struct parsed_op *po, const char *name)
1508{
840257f6 1509 const struct parsed_proto *pp;
1510
36595fd2 1511 pp = proto_parse(g_fhdr, name, 0);
840257f6 1512 if (pp == NULL)
1513 ferr(po, "proto_parse failed for ref '%s'\n", name);
1514
89ff3147 1515 if (pp->is_func)
1516 check_func_pp(po, pp, "ref");
3ebea2cf 1517}
1518
91977a1c 1519static char *out_src_opr(char *buf, size_t buf_size,
591721d7 1520 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
3ebea2cf 1521 int is_lea)
91977a1c 1522{
850c9265 1523 char tmp1[256], tmp2[256];
1524 char expr[256];
a2c1d768 1525 char *p;
850c9265 1526 int ret;
1527
3ebea2cf 1528 if (cast == NULL)
1529 cast = "";
1530
91977a1c 1531 switch (popr->type) {
1532 case OPT_REG:
850c9265 1533 if (is_lea)
1534 ferr(po, "lea from reg?\n");
1535
91977a1c 1536 switch (popr->lmod) {
1537 case OPLM_DWORD:
3ebea2cf 1538 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
91977a1c 1539 break;
850c9265 1540 case OPLM_WORD:
a2c1d768 1541 snprintf(buf, buf_size, "%s%s",
1542 simplify_cast(cast, "(u16)"), opr_reg_p(po, popr));
850c9265 1543 break;
1544 case OPLM_BYTE:
5101a5f9 1545 if (popr->name[1] == 'h') // XXX..
a2c1d768 1546 snprintf(buf, buf_size, "%s(%s >> 8)",
1547 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
5101a5f9 1548 else
a2c1d768 1549 snprintf(buf, buf_size, "%s%s",
1550 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
850c9265 1551 break;
91977a1c 1552 default:
1553 ferr(po, "invalid src lmod: %d\n", popr->lmod);
1554 }
1555 break;
850c9265 1556
91977a1c 1557 case OPT_REGMEM:
1cd4a663 1558 if (is_stack_access(po, popr)) {
de50b98b 1559 stack_frame_access(po, popr, buf, buf_size,
3ebea2cf 1560 popr->name, cast, 1, is_lea);
91977a1c 1561 break;
1562 }
850c9265 1563
1564 strcpy(expr, popr->name);
1565 if (strchr(expr, '[')) {
1566 // special case: '[' can only be left for label[reg] form
1567 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
1568 if (ret != 2)
1569 ferr(po, "parse failure for '%s'\n", expr);
a2c1d768 1570 if (tmp1[0] == '(') {
1571 // (off_4FFF50+3)[eax]
1572 p = strchr(tmp1 + 1, ')');
1573 if (p == NULL || p[1] != 0)
1574 ferr(po, "parse failure (2) for '%s'\n", expr);
1575 *p = 0;
1576 memmove(tmp1, tmp1 + 1, strlen(tmp1));
1577 }
33c35af6 1578 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
850c9265 1579 }
1580
1581 // XXX: do we need more parsing?
1582 if (is_lea) {
1583 snprintf(buf, buf_size, "%s", expr);
1584 break;
1585 }
1586
a2c1d768 1587 snprintf(buf, buf_size, "%s(%s)",
1588 simplify_cast(cast, lmod_cast_u_ptr(po, popr->lmod)), expr);
91977a1c 1589 break;
850c9265 1590
91977a1c 1591 case OPT_LABEL:
3ebea2cf 1592 check_label_read_ref(po, popr->name);
1593 if (cast[0] == 0 && popr->is_ptr)
1594 cast = "(u32)";
2b43685d 1595
850c9265 1596 if (is_lea)
1597 snprintf(buf, buf_size, "(u32)&%s", popr->name);
2b43685d 1598 else if (popr->size_lt)
1599 snprintf(buf, buf_size, "%s%s%s%s", cast,
1600 lmod_cast_u_ptr(po, popr->lmod),
1601 popr->is_array ? "" : "&",
1602 popr->name);
850c9265 1603 else
3ebea2cf 1604 snprintf(buf, buf_size, "%s%s%s", cast, popr->name,
7ba45c34 1605 popr->is_array ? "[0]" : "");
850c9265 1606 break;
1607
1608 case OPT_OFFSET:
3ebea2cf 1609 check_label_read_ref(po, popr->name);
1610 if (cast[0] == 0)
1611 cast = "(u32)";
850c9265 1612 if (is_lea)
1613 ferr(po, "lea an offset?\n");
3ebea2cf 1614 snprintf(buf, buf_size, "%s&%s", cast, popr->name);
91977a1c 1615 break;
850c9265 1616
91977a1c 1617 case OPT_CONST:
850c9265 1618 if (is_lea)
1619 ferr(po, "lea from const?\n");
1620
a2c1d768 1621 printf_number(tmp1, sizeof(tmp1), popr->val);
ddaf8bd7 1622 if (popr->val == 0 && strchr(cast, '*'))
1623 snprintf(buf, buf_size, "NULL");
1624 else
1625 snprintf(buf, buf_size, "%s%s",
1626 simplify_cast_num(cast, popr->val), tmp1);
91977a1c 1627 break;
850c9265 1628
91977a1c 1629 default:
1630 ferr(po, "invalid src type: %d\n", popr->type);
1631 }
1632
1633 return buf;
1634}
c36e914d 1635
de50b98b 1636// note: may set is_ptr (we find that out late for ebp frame..)
91977a1c 1637static char *out_dst_opr(char *buf, size_t buf_size,
1638 struct parsed_op *po, struct parsed_opr *popr)
1639{
1640 switch (popr->type) {
1641 case OPT_REG:
1642 switch (popr->lmod) {
1643 case OPLM_DWORD:
1644 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
1645 break;
850c9265 1646 case OPLM_WORD:
1647 // ugh..
1648 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
1649 break;
1650 case OPLM_BYTE:
1651 // ugh..
5101a5f9 1652 if (popr->name[1] == 'h') // XXX..
1653 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
1654 else
1655 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
850c9265 1656 break;
91977a1c 1657 default:
1658 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
1659 }
1660 break;
850c9265 1661
1662 case OPT_REGMEM:
1cd4a663 1663 if (is_stack_access(po, popr)) {
de50b98b 1664 stack_frame_access(po, popr, buf, buf_size,
3ebea2cf 1665 popr->name, "", 0, 0);
850c9265 1666 break;
1667 }
1668
3ebea2cf 1669 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
850c9265 1670
bfa4a6ee 1671 case OPT_LABEL:
2b43685d 1672 if (popr->size_mismatch)
1673 snprintf(buf, buf_size, "%s%s%s",
1674 lmod_cast_u_ptr(po, popr->lmod),
1675 popr->is_array ? "" : "&", popr->name);
1676 else
1677 snprintf(buf, buf_size, "%s%s", popr->name,
1678 popr->is_array ? "[0]" : "");
bfa4a6ee 1679 break;
1680
91977a1c 1681 default:
1682 ferr(po, "invalid dst type: %d\n", popr->type);
1683 }
1684
1685 return buf;
1686}
c36e914d 1687
3ebea2cf 1688static char *out_src_opr_u32(char *buf, size_t buf_size,
1689 struct parsed_op *po, struct parsed_opr *popr)
1690{
1691 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1692}
1693
69a3cdfc 1694static enum parsed_flag_op split_cond(struct parsed_op *po,
940e8e66 1695 enum op_op op, int *is_inv)
91977a1c 1696{
940e8e66 1697 *is_inv = 0;
91977a1c 1698
69a3cdfc 1699 switch (op) {
1700 case OP_JO:
1701 return PFO_O;
1702 case OP_JC:
1703 return PFO_C;
1704 case OP_JZ:
1705 return PFO_Z;
1706 case OP_JBE:
1707 return PFO_BE;
1708 case OP_JS:
1709 return PFO_S;
1710 case OP_JP:
1711 return PFO_P;
1712 case OP_JL:
1713 return PFO_L;
1714 case OP_JLE:
1715 return PFO_LE;
1716
91977a1c 1717 case OP_JNO:
940e8e66 1718 *is_inv = 1;
69a3cdfc 1719 return PFO_O;
91977a1c 1720 case OP_JNC:
940e8e66 1721 *is_inv = 1;
69a3cdfc 1722 return PFO_C;
91977a1c 1723 case OP_JNZ:
940e8e66 1724 *is_inv = 1;
69a3cdfc 1725 return PFO_Z;
1726 case OP_JA:
940e8e66 1727 *is_inv = 1;
69a3cdfc 1728 return PFO_BE;
91977a1c 1729 case OP_JNS:
940e8e66 1730 *is_inv = 1;
69a3cdfc 1731 return PFO_S;
91977a1c 1732 case OP_JNP:
940e8e66 1733 *is_inv = 1;
69a3cdfc 1734 return PFO_P;
850c9265 1735 case OP_JGE:
940e8e66 1736 *is_inv = 1;
69a3cdfc 1737 return PFO_L;
91977a1c 1738 case OP_JG:
940e8e66 1739 *is_inv = 1;
69a3cdfc 1740 return PFO_LE;
1741
cb090db0 1742 case OP_RCL:
1743 case OP_RCR:
69a3cdfc 1744 case OP_ADC:
1745 case OP_SBB:
1746 return PFO_C;
1747
91977a1c 1748 default:
69a3cdfc 1749 ferr(po, "split_cond: bad op %d\n", op);
1750 return -1;
91977a1c 1751 }
1752}
c36e914d 1753
91977a1c 1754static void out_test_for_cc(char *buf, size_t buf_size,
940e8e66 1755 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
69a3cdfc 1756 enum opr_lenmod lmod, const char *expr)
91977a1c 1757{
69a3cdfc 1758 const char *cast, *scast;
91977a1c 1759
69a3cdfc 1760 cast = lmod_cast_u(po, lmod);
1761 scast = lmod_cast_s(po, lmod);
1762
1763 switch (pfo) {
1764 case PFO_Z:
87bf6cec 1765 case PFO_BE: // CF=1||ZF=1; CF=0
850c9265 1766 snprintf(buf, buf_size, "(%s%s %s 0)",
940e8e66 1767 cast, expr, is_inv ? "!=" : "==");
91977a1c 1768 break;
850c9265 1769
5101a5f9 1770 case PFO_S:
1771 case PFO_L: // SF!=OF; OF=0
1772 snprintf(buf, buf_size, "(%s%s %s 0)",
940e8e66 1773 scast, expr, is_inv ? ">=" : "<");
5101a5f9 1774 break;
1775
87bf6cec 1776 case PFO_LE: // ZF=1||SF!=OF; OF=0
69a3cdfc 1777 snprintf(buf, buf_size, "(%s%s %s 0)",
940e8e66 1778 scast, expr, is_inv ? ">" : "<=");
850c9265 1779 break;
1780
91977a1c 1781 default:
69a3cdfc 1782 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
91977a1c 1783 }
1784}
c36e914d 1785
850c9265 1786static void out_cmp_for_cc(char *buf, size_t buf_size,
a2c1d768 1787 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
850c9265 1788{
a2c1d768 1789 const char *cast, *scast, *cast_use;
1790 char buf1[256], buf2[256];
1791 enum opr_lenmod lmod;
1792
1793 if (po->operand[0].lmod != po->operand[1].lmod)
1794 ferr(po, "%s: lmod mismatch: %d %d\n", __func__,
1795 po->operand[0].lmod, po->operand[1].lmod);
1796 lmod = po->operand[0].lmod;
850c9265 1797
69a3cdfc 1798 cast = lmod_cast_u(po, lmod);
1799 scast = lmod_cast_s(po, lmod);
850c9265 1800
a2c1d768 1801 switch (pfo) {
1802 case PFO_C:
1803 case PFO_Z:
1804 case PFO_BE: // !a
1805 cast_use = cast;
1806 break;
1807
1808 case PFO_S:
1809 case PFO_L: // !ge
1810 case PFO_LE:
1811 cast_use = scast;
1812 break;
1813
1814 default:
1815 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1816 }
1817
1818 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], cast_use, 0);
1819 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], cast_use, 0);
1820
69a3cdfc 1821 switch (pfo) {
5101a5f9 1822 case PFO_C:
1823 // note: must be unsigned compare
a2c1d768 1824 snprintf(buf, buf_size, "(%s %s %s)",
1825 buf1, is_inv ? ">=" : "<", buf2);
5101a5f9 1826 break;
1827
69a3cdfc 1828 case PFO_Z:
a2c1d768 1829 snprintf(buf, buf_size, "(%s %s %s)",
1830 buf1, is_inv ? "!=" : "==", buf2);
850c9265 1831 break;
1832
5101a5f9 1833 case PFO_BE: // !a
850c9265 1834 // note: must be unsigned compare
a2c1d768 1835 snprintf(buf, buf_size, "(%s %s %s)",
1836 buf1, is_inv ? ">" : "<=", buf2);
1837
1838 // annoying case
1839 if (is_inv && lmod == OPLM_BYTE
1840 && po->operand[1].type == OPT_CONST
1841 && po->operand[1].val == 0xff)
1842 {
1843 snprintf(g_comment, sizeof(g_comment), "if %s", buf);
1844 snprintf(buf, buf_size, "(0)");
1845 }
5101a5f9 1846 break;
1847
1848 // note: must be signed compare
1849 case PFO_S:
1850 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
a2c1d768 1851 scast, buf1, buf2, is_inv ? ">=" : "<");
850c9265 1852 break;
1853
5101a5f9 1854 case PFO_L: // !ge
a2c1d768 1855 snprintf(buf, buf_size, "(%s %s %s)",
1856 buf1, is_inv ? ">=" : "<", buf2);
850c9265 1857 break;
1858
5101a5f9 1859 case PFO_LE:
a2c1d768 1860 snprintf(buf, buf_size, "(%s %s %s)",
1861 buf1, is_inv ? ">" : "<=", buf2);
5101a5f9 1862 break;
1863
850c9265 1864 default:
a2c1d768 1865 break;
850c9265 1866 }
1867}
1868
69a3cdfc 1869static void out_cmp_test(char *buf, size_t buf_size,
940e8e66 1870 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
69a3cdfc 1871{
1872 char buf1[256], buf2[256], buf3[256];
1873
1874 if (po->op == OP_TEST) {
1875 if (IS(opr_name(po, 0), opr_name(po, 1))) {
3ebea2cf 1876 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
69a3cdfc 1877 }
1878 else {
3ebea2cf 1879 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
1880 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
69a3cdfc 1881 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
1882 }
940e8e66 1883 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
69a3cdfc 1884 po->operand[0].lmod, buf3);
1885 }
1886 else if (po->op == OP_CMP) {
a2c1d768 1887 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv);
69a3cdfc 1888 }
1889 else
1890 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
1891}
1892
850c9265 1893static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
91977a1c 1894 struct parsed_opr *popr2)
1895{
1896 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
1897 ferr(po, "missing lmod for both operands\n");
1898
1899 if (popr1->lmod == OPLM_UNSPEC)
1900 popr1->lmod = popr2->lmod;
1901 else if (popr2->lmod == OPLM_UNSPEC)
1902 popr2->lmod = popr1->lmod;
a3684be1 1903 else if (popr1->lmod != popr2->lmod) {
1904 if (popr1->type_from_var) {
1905 popr1->size_mismatch = 1;
1906 if (popr1->lmod < popr2->lmod)
1907 popr1->size_lt = 1;
1908 popr1->lmod = popr2->lmod;
1909 }
1910 else if (popr2->type_from_var) {
1911 popr2->size_mismatch = 1;
1912 if (popr2->lmod < popr1->lmod)
1913 popr2->size_lt = 1;
1914 popr2->lmod = popr1->lmod;
1915 }
1916 else
1917 ferr(po, "conflicting lmods: %d vs %d\n",
1918 popr1->lmod, popr2->lmod);
1919 }
91977a1c 1920}
c36e914d 1921
850c9265 1922static const char *op_to_c(struct parsed_op *po)
1923{
1924 switch (po->op)
1925 {
1926 case OP_ADD:
5101a5f9 1927 case OP_ADC:
850c9265 1928 return "+";
1929 case OP_SUB:
5101a5f9 1930 case OP_SBB:
850c9265 1931 return "-";
1932 case OP_AND:
1933 return "&";
1934 case OP_OR:
1935 return "|";
1936 case OP_XOR:
1937 return "^";
1938 case OP_SHL:
1939 return "<<";
1940 case OP_SHR:
1941 return ">>";
1942 case OP_MUL:
1943 case OP_IMUL:
1944 return "*";
1945 default:
1946 ferr(po, "op_to_c was supplied with %d\n", po->op);
1947 }
1948}
1949
7ae48d73 1950static void op_set_clear_flag(struct parsed_op *po,
1951 enum op_flags flag_set, enum op_flags flag_clear)
d4e3b5db 1952{
7ae48d73 1953 po->flags |= flag_set;
1954 po->flags &= ~flag_clear;
d4e3b5db 1955}
1956
de50b98b 1957// last op in stream - unconditional branch or ret
1958#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
5c024ef7 1959 || (ops[_i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP)
de50b98b 1960
87bf6cec 1961static int scan_for_pop(int i, int opcnt, const char *reg,
d4e3b5db 1962 int magic, int depth, int *maxdepth, int do_flags)
850c9265 1963{
89ff3147 1964 const struct parsed_proto *pp;
87bf6cec 1965 struct parsed_op *po;
1966 int ret = 0;
4c45fa73 1967 int j;
87bf6cec 1968
850c9265 1969 for (; i < opcnt; i++) {
87bf6cec 1970 po = &ops[i];
1971 if (po->cc_scratch == magic)
1972 break; // already checked
1973 po->cc_scratch = magic;
1974
89ff3147 1975 if (po->flags & OPF_TAIL) {
1976 if (po->op == OP_CALL) {
1977 pp = proto_parse(g_fhdr, po->operand[0].name, 0);
1978 if (pp != NULL && pp->is_noreturn)
1979 // no stack cleanup for noreturn
1980 return ret;
1981 }
87bf6cec 1982 return -1; // deadend
89ff3147 1983 }
87bf6cec 1984
1bafb621 1985 if ((po->flags & OPF_RMD)
de50b98b 1986 || (po->op == OP_PUSH && po->argnum != 0)) // arg push
850c9265 1987 continue;
1988
87bf6cec 1989 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
4c45fa73 1990 if (po->btj != NULL) {
1991 // jumptable
da87ae38 1992 for (j = 0; j < po->btj->count; j++) {
4c45fa73 1993 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic,
1994 depth, maxdepth, do_flags);
1995 if (ret < 0)
1996 return ret; // dead end
1997 }
da87ae38 1998 return ret;
4c45fa73 1999 }
2000
87bf6cec 2001 if (po->bt_i < 0) {
2002 ferr(po, "dead branch\n");
2003 return -1;
2004 }
850c9265 2005
5c024ef7 2006 if (po->flags & OPF_CJMP) {
d4e3b5db 2007 ret |= scan_for_pop(po->bt_i, opcnt, reg, magic,
2008 depth, maxdepth, do_flags);
87bf6cec 2009 if (ret < 0)
2010 return ret; // dead end
2011 }
2012 else {
2013 i = po->bt_i - 1;
2014 }
2015 continue;
2016 }
2017
d4e3b5db 2018 if ((po->op == OP_POP || po->op == OP_PUSH)
2019 && po->operand[0].type == OPT_REG
87bf6cec 2020 && IS(po->operand[0].name, reg))
2021 {
da87ae38 2022 if (po->op == OP_PUSH && !(po->flags & OPF_FARG)) {
d4e3b5db 2023 depth++;
2024 if (depth > *maxdepth)
2025 *maxdepth = depth;
2026 if (do_flags)
7ae48d73 2027 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
d4e3b5db 2028 }
da87ae38 2029 else if (po->op == OP_POP) {
2030 if (depth == 0) {
2031 if (do_flags)
2032 op_set_clear_flag(po, OPF_RMD, OPF_RSAVE);
2033 return 1;
2034 }
2035 else {
2036 depth--;
2037 if (depth < 0) // should not happen
2038 ferr(po, "fail with depth\n");
2039 if (do_flags)
2040 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2041 }
d4e3b5db 2042 }
87bf6cec 2043 }
850c9265 2044 }
2045
87bf6cec 2046 return ret;
850c9265 2047}
2048
69a3cdfc 2049// scan for pop starting from 'ret' op (all paths)
d4e3b5db 2050static int scan_for_pop_ret(int i, int opcnt, const char *reg,
2051 int flag_set)
850c9265 2052{
2053 int found = 0;
2054 int j;
2055
2056 for (; i < opcnt; i++) {
87bf6cec 2057 if (!(ops[i].flags & OPF_TAIL))
850c9265 2058 continue;
2059
2060 for (j = i - 1; j >= 0; j--) {
69a3cdfc 2061 if (ops[j].flags & OPF_RMD)
2062 continue;
2063 if (ops[j].flags & OPF_JMP)
850c9265 2064 return -1;
2065
2066 if (ops[j].op == OP_POP && ops[j].operand[0].type == OPT_REG
2067 && IS(ops[j].operand[0].name, reg))
2068 {
2069 found = 1;
d4e3b5db 2070 ops[j].flags |= flag_set;
850c9265 2071 break;
2072 }
2073
2074 if (g_labels[j][0] != 0)
2075 return -1;
2076 }
2077 }
2078
2079 return found ? 0 : -1;
2080}
2081
591721d7 2082static void scan_propagate_df(int i, int opcnt)
2083{
2084 struct parsed_op *po = &ops[i];
2085 int j;
2086
2087 for (; i < opcnt; i++) {
2088 po = &ops[i];
2089 if (po->flags & OPF_DF)
2090 return; // already resolved
2091 po->flags |= OPF_DF;
2092
2093 if (po->op == OP_CALL)
2094 ferr(po, "call with DF set?\n");
2095
2096 if (po->flags & OPF_JMP) {
2097 if (po->btj != NULL) {
2098 // jumptable
2099 for (j = 0; j < po->btj->count; j++)
2100 scan_propagate_df(po->btj->d[j].bt_i, opcnt);
2101 return;
2102 }
2103
2104 if (po->bt_i < 0) {
2105 ferr(po, "dead branch\n");
2106 return;
2107 }
2108
5c024ef7 2109 if (po->flags & OPF_CJMP)
591721d7 2110 scan_propagate_df(po->bt_i, opcnt);
2111 else
2112 i = po->bt_i - 1;
2113 continue;
2114 }
2115
2116 if (po->flags & OPF_TAIL)
2117 break;
2118
2119 if (po->op == OP_CLD) {
2120 po->flags |= OPF_RMD;
2121 return;
2122 }
2123 }
2124
2125 ferr(po, "missing DF clear?\n");
2126}
2127
1cd4a663 2128// is operand 'opr' modified by parsed_op 'po'?
5101a5f9 2129static int is_opr_modified(const struct parsed_opr *opr,
69a3cdfc 2130 const struct parsed_op *po)
2131{
89ff3147 2132 int mask;
2133
69a3cdfc 2134 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2135 return 0;
2136
89ff3147 2137 if (opr->type == OPT_REG) {
2138 if (po->op == OP_CALL) {
2139 mask = (1 << xAX) | (1 << xCX) | (1 << xDX);
2140 if ((1 << opr->reg) & mask)
2141 return 1;
2142 else
2143 return 0;
2144 }
2145
2146 if (po->operand[0].type == OPT_REG) {
2147 if (po->regmask_dst & (1 << opr->reg))
2148 return 1;
2149 else
2150 return 0;
2151 }
69a3cdfc 2152 }
2153
2154 return IS(po->operand[0].name, opr->name);
2155}
2156
5101a5f9 2157// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
2158static int is_any_opr_modified(const struct parsed_op *po_test,
89ff3147 2159 const struct parsed_op *po, int c_mode)
5101a5f9 2160{
89ff3147 2161 int mask;
5101a5f9 2162 int i;
2163
2164 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2165 return 0;
2166
de50b98b 2167 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2168 return 0;
2169
2170 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
2171 return 1;
2172
2173 // in reality, it can wreck any register, but in decompiled C
2b43685d 2174 // version it can only overwrite eax or edx:eax
89ff3147 2175 mask = (1 << xAX) | (1 << xDX);
2176 if (!c_mode)
2177 mask |= 1 << xCX;
2178
de50b98b 2179 if (po->op == OP_CALL
89ff3147 2180 && ((po_test->regmask_src | po_test->regmask_dst) & mask))
5101a5f9 2181 return 1;
2182
2183 for (i = 0; i < po_test->operand_cnt; i++)
2184 if (IS(po_test->operand[i].name, po->operand[0].name))
2185 return 1;
2186
2187 return 0;
2188}
2189
940e8e66 2190// scan for any po_test operand modification in range given
89ff3147 2191static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt,
2192 int c_mode)
69a3cdfc 2193{
2b43685d 2194 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2195 return -1;
2196
69a3cdfc 2197 for (; i < opcnt; i++) {
89ff3147 2198 if (is_any_opr_modified(po_test, &ops[i], c_mode))
69a3cdfc 2199 return i;
2200 }
2201
2202 return -1;
2203}
2204
940e8e66 2205// scan for po_test operand[0] modification in range given
2206static int scan_for_mod_opr0(struct parsed_op *po_test,
2207 int i, int opcnt)
2208{
2209 for (; i < opcnt; i++) {
2210 if (is_opr_modified(&po_test->operand[0], &ops[i]))
2211 return i;
2212 }
2213
2214 return -1;
2215}
2216
04f8a628 2217static int scan_for_flag_set(int i, int magic, int *branched,
2218 int *setters, int *setter_cnt)
69a3cdfc 2219{
04f8a628 2220 struct label_ref *lr;
2b43685d 2221 int ret;
de50b98b 2222
2223 while (i >= 0) {
04f8a628 2224 if (ops[i].cc_scratch == magic) {
2225 ferr(&ops[i], "%s looped\n", __func__);
2226 return -1;
2227 }
2228 ops[i].cc_scratch = magic;
2229
de50b98b 2230 if (g_labels[i][0] != 0) {
2b43685d 2231 *branched = 1;
04f8a628 2232
2233 lr = &g_label_refs[i];
2234 for (; lr->next; lr = lr->next) {
2235 ret = scan_for_flag_set(lr->i, magic,
2236 branched, setters, setter_cnt);
2237 if (ret < 0)
2238 return ret;
2239 }
2240
de50b98b 2241 if (i > 0 && LAST_OP(i - 1)) {
94d447fb 2242 i = lr->i;
de50b98b 2243 continue;
2244 }
04f8a628 2245 ret = scan_for_flag_set(lr->i, magic,
2246 branched, setters, setter_cnt);
2b43685d 2247 if (ret < 0)
2248 return ret;
de50b98b 2249 }
2250 i--;
2251
2b43685d 2252 if (ops[i].flags & OPF_FLAGS) {
2253 setters[*setter_cnt] = i;
2254 (*setter_cnt)++;
2255 return 0;
2256 }
69a3cdfc 2257
5c024ef7 2258 if ((ops[i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP)
69a3cdfc 2259 return -1;
69a3cdfc 2260 }
2261
2262 return -1;
2263}
2264
5101a5f9 2265// scan back for cdq, if anything modifies edx, fail
2266static int scan_for_cdq_edx(int i)
2267{
cdfaeed7 2268 while (i >= 0) {
2269 if (g_labels[i][0] != 0) {
2270 if (g_label_refs[i].next != NULL)
2271 return -1;
2272 if (i > 0 && LAST_OP(i - 1)) {
2273 i = g_label_refs[i].i;
2274 continue;
2275 }
2276 return -1;
2277 }
2278 i--;
2279
5101a5f9 2280 if (ops[i].op == OP_CDQ)
2281 return i;
2282
2283 if (ops[i].regmask_dst & (1 << xDX))
2284 return -1;
5101a5f9 2285 }
2286
2287 return -1;
2288}
2289
64c59faf 2290static int scan_for_reg_clear(int i, int reg)
2291{
cdfaeed7 2292 while (i >= 0) {
2293 if (g_labels[i][0] != 0) {
2294 if (g_label_refs[i].next != NULL)
2295 return -1;
2296 if (i > 0 && LAST_OP(i - 1)) {
2297 i = g_label_refs[i].i;
2298 continue;
2299 }
2300 return -1;
2301 }
2302 i--;
2303
64c59faf 2304 if (ops[i].op == OP_XOR
2305 && ops[i].operand[0].lmod == OPLM_DWORD
2306 && ops[i].operand[0].reg == ops[i].operand[1].reg
2307 && ops[i].operand[0].reg == reg)
2308 return i;
2309
2310 if (ops[i].regmask_dst & (1 << reg))
2311 return -1;
64c59faf 2312 }
2313
2314 return -1;
2315}
2316
1bafb621 2317// scan for positive, constant esp adjust
4741fdfe 2318static int scan_for_esp_adjust(int i, int opcnt, int *adj,
2319 int *multipath)
1bafb621 2320{
89ff3147 2321 const struct parsed_proto *pp;
7ba45c34 2322 struct parsed_op *po;
46411e6c 2323 int first_pop = -1;
4741fdfe 2324
2325 *adj = *multipath = 0;
7ba45c34 2326
1bafb621 2327 for (; i < opcnt; i++) {
7ba45c34 2328 po = &ops[i];
2329
a2c1d768 2330 if (g_labels[i][0] != 0)
4741fdfe 2331 *multipath = 1;
a2c1d768 2332
7ba45c34 2333 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
2334 if (po->operand[1].type != OPT_CONST)
1bafb621 2335 ferr(&ops[i], "non-const esp adjust?\n");
7ba45c34 2336 *adj += po->operand[1].val;
1bafb621 2337 if (*adj & 3)
2338 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
2339 return i;
2340 }
108e9fe3 2341 else if (po->op == OP_PUSH && !(po->flags & OPF_RMD)) {
5c024ef7 2342 //if (first_pop == -1)
2343 // first_pop = -2; // none
7ba45c34 2344 *adj -= lmod_bytes(po, po->operand[0].lmod);
46411e6c 2345 }
5c024ef7 2346 else if (po->op == OP_POP && !(po->flags & OPF_RMD)) {
108e9fe3 2347 // seems like msvc only uses 'pop ecx' for stack realignment..
2348 if (po->operand[0].type != OPT_REG || po->operand[0].reg != xCX)
2349 break;
5c024ef7 2350 if (first_pop == -1 && *adj >= 0)
46411e6c 2351 first_pop = i;
7ba45c34 2352 *adj += lmod_bytes(po, po->operand[0].lmod);
46411e6c 2353 }
e56ab892 2354 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
4741fdfe 2355 if (po->op == OP_JMP && po->btj == NULL) {
2356 i = po->bt_i - 1;
2357 continue;
2358 }
e56ab892 2359 if (po->op != OP_CALL)
a2c1d768 2360 break;
e56ab892 2361 if (po->operand[0].type != OPT_LABEL)
a2c1d768 2362 break;
89ff3147 2363 pp = po->datap;
2364 if (pp != NULL && pp->is_stdcall)
2365 break;
e56ab892 2366 }
a2c1d768 2367 }
7ba45c34 2368
108e9fe3 2369 if (first_pop >= 0) {
a2c1d768 2370 // probably 'pop ecx' was used..
46411e6c 2371 return first_pop;
1bafb621 2372 }
2373
2374 return -1;
2375}
2376
a3684be1 2377static void scan_fwd_set_flags(int i, int opcnt, int magic, int flags)
2378{
2379 struct parsed_op *po;
2380 int j;
2381
2382 if (i < 0)
2383 ferr(ops, "%s: followed bad branch?\n", __func__);
2384
2385 for (; i < opcnt; i++) {
2386 po = &ops[i];
2387 if (po->cc_scratch == magic)
2388 return;
2389 po->cc_scratch = magic;
2390 po->flags |= flags;
2391
2392 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2393 if (po->btj != NULL) {
2394 // jumptable
2395 for (j = 0; j < po->btj->count; j++)
2396 scan_fwd_set_flags(po->btj->d[j].bt_i, opcnt, magic, flags);
2397 return;
2398 }
2399
2400 scan_fwd_set_flags(po->bt_i, opcnt, magic, flags);
5c024ef7 2401 if (!(po->flags & OPF_CJMP))
a3684be1 2402 return;
2403 }
2404 if (po->flags & OPF_TAIL)
2405 return;
2406 }
2407}
2408
1cd4a663 2409static const struct parsed_proto *try_recover_pp(
da87ae38 2410 struct parsed_op *po, const struct parsed_opr *opr, int *search_instead)
1cd4a663 2411{
2412 const struct parsed_proto *pp = NULL;
89ff3147 2413 char buf[256];
2414 char *p;
1cd4a663 2415
2416 // maybe an arg of g_func?
2417 if (opr->type == OPT_REGMEM && is_stack_access(po, opr))
2418 {
2419 char ofs_reg[16] = { 0, };
2420 int arg, arg_s, arg_i;
2421 int stack_ra = 0;
2422 int offset = 0;
2423
2424 parse_stack_access(po, opr->name, ofs_reg,
2425 &offset, &stack_ra, NULL);
2426 if (ofs_reg[0] != 0)
2427 ferr(po, "offset reg on arg access?\n");
da87ae38 2428 if (offset <= stack_ra) {
2429 // search who set the stack var instead
2430 if (search_instead != NULL)
2431 *search_instead = 1;
2432 return NULL;
2433 }
1cd4a663 2434
2435 arg_i = (offset - stack_ra - 4) / 4;
2436 for (arg = arg_s = 0; arg < g_func_pp->argc; arg++) {
2437 if (g_func_pp->arg[arg].reg != NULL)
2438 continue;
2439 if (arg_s == arg_i)
2440 break;
2441 arg_s++;
2442 }
2443 if (arg == g_func_pp->argc)
2444 ferr(po, "stack arg %d not in prototype?\n", arg_i);
2445
2446 pp = g_func_pp->arg[arg].fptr;
2447 if (pp == NULL)
46411e6c 2448 ferr(po, "icall sa: arg%d is not a fptr?\n", arg + 1);
89ff3147 2449 check_func_pp(po, pp, "icall arg");
2450 }
2451 else if (opr->type == OPT_REGMEM && strchr(opr->name + 1, '[')) {
2452 // label[index]
2453 p = strchr(opr->name + 1, '[');
2454 memcpy(buf, opr->name, p - opr->name);
2455 buf[p - opr->name] = 0;
2456 pp = proto_parse(g_fhdr, buf, 0);
1cd4a663 2457 }
2458 else if (opr->type == OPT_OFFSET || opr->type == OPT_LABEL) {
36595fd2 2459 pp = proto_parse(g_fhdr, opr->name, 0);
1cd4a663 2460 if (pp == NULL)
2461 ferr(po, "proto_parse failed for icall from '%s'\n", opr->name);
89ff3147 2462 check_func_pp(po, pp, "reg-fptr ref");
1cd4a663 2463 }
2464
2465 return pp;
2466}
2467
da87ae38 2468static void scan_for_call_type(int i, const struct parsed_opr *opr,
89ff3147 2469 int magic, const struct parsed_proto **pp_found, int *multi)
1cd4a663 2470{
2471 const struct parsed_proto *pp = NULL;
2472 struct parsed_op *po;
2473 struct label_ref *lr;
2474
2475 while (i >= 0) {
2476 if (ops[i].cc_scratch == magic)
2477 return;
2478 ops[i].cc_scratch = magic;
2479
2480 if (g_labels[i][0] != 0) {
2481 lr = &g_label_refs[i];
2482 for (; lr != NULL; lr = lr->next)
89ff3147 2483 scan_for_call_type(lr->i, opr, magic, pp_found, multi);
46411e6c 2484 if (i > 0 && LAST_OP(i - 1))
2485 return;
1cd4a663 2486 }
2487 i--;
2488
2489 if (!(ops[i].flags & OPF_DATA))
2490 continue;
2491 if (!is_opr_modified(opr, &ops[i]))
2492 continue;
2493 if (ops[i].op != OP_MOV && ops[i].op != OP_LEA) {
2494 // most probably trashed by some processing
2495 *pp_found = NULL;
2496 return;
2497 }
2498
2499 opr = &ops[i].operand[1];
2500 if (opr->type != OPT_REG)
2501 break;
2502 }
2503
2504 po = (i >= 0) ? &ops[i] : ops;
2505
2506 if (i < 0) {
2507 // reached the top - can only be an arg-reg
2508 if (opr->type != OPT_REG)
2509 return;
2510
2511 for (i = 0; i < g_func_pp->argc; i++) {
2512 if (g_func_pp->arg[i].reg == NULL)
2513 continue;
2514 if (IS(opr->name, g_func_pp->arg[i].reg))
2515 break;
2516 }
2517 if (i == g_func_pp->argc)
2518 return;
2519 pp = g_func_pp->arg[i].fptr;
2520 if (pp == NULL)
46411e6c 2521 ferr(po, "icall: arg%d (%s) is not a fptr?\n",
2522 i + 1, g_func_pp->arg[i].reg);
89ff3147 2523 check_func_pp(po, pp, "icall reg-arg");
1cd4a663 2524 }
2525 else
da87ae38 2526 pp = try_recover_pp(po, opr, NULL);
1cd4a663 2527
89ff3147 2528 if (*pp_found != NULL && pp != NULL && *pp_found != pp) {
1cd4a663 2529 if (!IS((*pp_found)->ret_type.name, pp->ret_type.name)
2530 || (*pp_found)->is_stdcall != pp->is_stdcall
89ff3147 2531 || (*pp_found)->is_fptr != pp->is_fptr
1cd4a663 2532 || (*pp_found)->argc != pp->argc
2533 || (*pp_found)->argc_reg != pp->argc_reg
2534 || (*pp_found)->argc_stack != pp->argc_stack)
2535 {
2536 ferr(po, "icall: parsed_proto mismatch\n");
2537 }
89ff3147 2538 *multi = 1;
1cd4a663 2539 }
2540 if (pp != NULL)
2541 *pp_found = pp;
2542}
2543
89ff3147 2544static const struct parsed_proto *resolve_icall(int i, int opcnt,
2545 int *multi_src)
1cd4a663 2546{
2547 const struct parsed_proto *pp = NULL;
da87ae38 2548 int search_advice = 0;
1cd4a663 2549
89ff3147 2550 *multi_src = 0;
2551
1cd4a663 2552 switch (ops[i].operand[0].type) {
2553 case OPT_REGMEM:
2554 case OPT_LABEL:
2555 case OPT_OFFSET:
da87ae38 2556 pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice);
2557 if (!search_advice)
2558 break;
2559 // fallthrough
1cd4a663 2560 default:
89ff3147 2561 scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp,
2562 multi_src);
1cd4a663 2563 break;
2564 }
2565
2566 return pp;
2567}
2568
89ff3147 2569static int collect_call_args_r(struct parsed_op *po, int i,
75ad0378 2570 struct parsed_proto *pp, int *regmask, int *save_arg_vars, int arg,
a3684be1 2571 int magic, int need_op_saving, int may_reuse)
e56ab892 2572{
2573 struct parsed_proto *pp_tmp;
2574 struct label_ref *lr;
2b43685d 2575 int need_to_save_current;
e56ab892 2576 int ret = 0;
2577 int j;
2578
a3684be1 2579 if (i < 0) {
a2c1d768 2580 ferr(po, "dead label encountered\n");
a3684be1 2581 return -1;
2582 }
e56ab892 2583
de50b98b 2584 for (; arg < pp->argc; arg++)
e56ab892 2585 if (pp->arg[arg].reg == NULL)
2586 break;
a3684be1 2587 magic = (magic & 0xffffff) | (arg << 24);
e56ab892 2588
89ff3147 2589 for (j = i; j >= 0 && (arg < pp->argc || pp->is_unresolved); )
e56ab892 2590 {
a3684be1 2591 if (((ops[j].cc_scratch ^ magic) & 0xffffff) == 0) {
2592 if (ops[j].cc_scratch != magic) {
2593 ferr(&ops[j], "arg collect hit same path with diff args for %s\n",
2594 pp->name);
2595 return -1;
2596 }
2597 // ok: have already been here
2598 return 0;
2599 }
2600 ops[j].cc_scratch = magic;
2601
a2c1d768 2602 if (g_labels[j][0] != 0 && g_label_refs[j].i != -1) {
e56ab892 2603 lr = &g_label_refs[j];
2604 if (lr->next != NULL)
2605 need_op_saving = 1;
a652aa9f 2606 for (; lr->next; lr = lr->next) {
5c024ef7 2607 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
a652aa9f 2608 may_reuse = 1;
89ff3147 2609 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
a3684be1 2610 arg, magic, need_op_saving, may_reuse);
2611 if (ret < 0)
2612 return ret;
a652aa9f 2613 }
e56ab892 2614
5c024ef7 2615 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
a652aa9f 2616 may_reuse = 1;
de50b98b 2617 if (j > 0 && LAST_OP(j - 1)) {
e56ab892 2618 // follow last branch in reverse
2619 j = lr->i;
2620 continue;
2621 }
2622 need_op_saving = 1;
89ff3147 2623 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
a3684be1 2624 arg, magic, need_op_saving, may_reuse);
2625 if (ret < 0)
2626 return ret;
e56ab892 2627 }
2628 j--;
2629
2630 if (ops[j].op == OP_CALL)
2631 {
89ff3147 2632 if (pp->is_unresolved)
2633 break;
2634
e56ab892 2635 pp_tmp = ops[j].datap;
2636 if (pp_tmp == NULL)
7ae48d73 2637 ferr(po, "arg collect hit unparsed call '%s'\n",
2638 ops[j].operand[0].name);
a652aa9f 2639 if (may_reuse && pp_tmp->argc_stack > 0)
de50b98b 2640 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
2641 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
e56ab892 2642 }
de50b98b 2643 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP) {
89ff3147 2644 if (pp->is_unresolved)
2645 break;
2646
de50b98b 2647 ferr(po, "arg collect %d/%d hit esp adjust\n",
2648 arg, pp->argc);
2649 }
2650 else if (ops[j].op == OP_POP) {
89ff3147 2651 if (pp->is_unresolved)
2652 break;
2653
de50b98b 2654 ferr(po, "arg collect %d/%d hit pop\n", arg, pp->argc);
2655 }
5c024ef7 2656 else if (ops[j].flags & OPF_CJMP)
de50b98b 2657 {
89ff3147 2658 if (pp->is_unresolved)
2659 break;
2660
a652aa9f 2661 may_reuse = 1;
de50b98b 2662 }
2663 else if (ops[j].op == OP_PUSH && !(ops[j].flags & OPF_FARG))
e56ab892 2664 {
89ff3147 2665 if (pp->is_unresolved && (ops[j].flags & OPF_RMD))
2666 break;
2667
e56ab892 2668 pp->arg[arg].datap = &ops[j];
2b43685d 2669 need_to_save_current = 0;
e56ab892 2670 if (!need_op_saving) {
89ff3147 2671 ret = scan_for_mod(&ops[j], j + 1, i, 1);
2b43685d 2672 need_to_save_current = (ret >= 0);
e56ab892 2673 }
2b43685d 2674 if (need_op_saving || need_to_save_current) {
e56ab892 2675 // mark this push as one that needs operand saving
2676 ops[j].flags &= ~OPF_RMD;
de50b98b 2677 if (ops[j].argnum == 0) {
2678 ops[j].argnum = arg + 1;
2679 *save_arg_vars |= 1 << arg;
2680 }
2681 else if (ops[j].argnum < arg + 1)
2682 ferr(&ops[j], "argnum conflict (%d<%d) for '%s'\n",
2683 ops[j].argnum, arg + 1, pp->name);
e56ab892 2684 }
de50b98b 2685 else if (ops[j].argnum == 0)
e56ab892 2686 ops[j].flags |= OPF_RMD;
2687
a652aa9f 2688 // some PUSHes are reused by different calls on other branches,
de50b98b 2689 // but that can't happen if we didn't branch, so they
2690 // can be removed from future searches (handles nested calls)
a652aa9f 2691 if (!may_reuse)
de50b98b 2692 ops[j].flags |= OPF_FARG;
2693
da87ae38 2694 ops[j].flags &= ~OPF_RSAVE;
2695
89ff3147 2696 arg++;
2697 if (!pp->is_unresolved) {
2698 // next arg
2699 for (; arg < pp->argc; arg++)
2700 if (pp->arg[arg].reg == NULL)
2701 break;
2702 }
a3684be1 2703 magic = (magic & 0xffffff) | (arg << 24);
75ad0378 2704
2705 // tracking reg usage
2706 if (ops[j].operand[0].type == OPT_REG)
2707 *regmask |= 1 << ops[j].operand[0].reg;
e56ab892 2708 }
2709 }
2710
2711 if (arg < pp->argc) {
2712 ferr(po, "arg collect failed for '%s': %d/%d\n",
2713 pp->name, arg, pp->argc);
89ff3147 2714 return -1;
e56ab892 2715 }
89ff3147 2716
2717 return arg;
2718}
2719
2720static int collect_call_args(struct parsed_op *po, int i,
2721 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
2722 int magic)
2723{
2724 int ret;
2725 int a;
2726
2727 ret = collect_call_args_r(po, i, pp, regmask, save_arg_vars,
2728 0, magic, 0, 0);
2729 if (ret < 0)
2730 return ret;
2731
2732 if (pp->is_unresolved) {
2733 pp->argc += ret;
2734 pp->argc_stack += ret;
2735 for (a = 0; a < pp->argc; a++)
2736 if (pp->arg[a].type.name == NULL)
2737 pp->arg[a].type.name = strdup("int");
2738 }
2739
e56ab892 2740 return ret;
2741}
2742
89ff3147 2743static void pp_insert_reg_arg(struct parsed_proto *pp, const char *reg)
2744{
2745 int i;
2746
2747 for (i = 0; i < pp->argc; i++)
2748 if (pp->arg[i].reg == NULL)
2749 break;
2750
2751 if (pp->argc_stack)
2752 memmove(&pp->arg[i + 1], &pp->arg[i],
2753 sizeof(pp->arg[0]) * pp->argc_stack);
2754 memset(&pp->arg[i], 0, sizeof(pp->arg[i]));
2755 pp->arg[i].reg = strdup(reg);
2756 pp->arg[i].type.name = strdup("int");
2757 pp->argc++;
2758 pp->argc_reg++;
2759}
2760
4c45fa73 2761static void add_label_ref(struct label_ref *lr, int op_i)
2762{
2763 struct label_ref *lr_new;
2764
2765 if (lr->i == -1) {
2766 lr->i = op_i;
2767 return;
2768 }
2769
2770 lr_new = calloc(1, sizeof(*lr_new));
2771 lr_new->i = op_i;
a652aa9f 2772 lr_new->next = lr->next;
4c45fa73 2773 lr->next = lr_new;
2774}
2775
04f8a628 2776static void output_std_flags(FILE *fout, struct parsed_op *po,
2777 int *pfomask, const char *dst_opr_text)
2778{
2779 if (*pfomask & (1 << PFO_Z)) {
2780 fprintf(fout, "\n cond_z = (%s%s == 0);",
2781 lmod_cast_u(po, po->operand[0].lmod), dst_opr_text);
2782 *pfomask &= ~(1 << PFO_Z);
2783 }
2784 if (*pfomask & (1 << PFO_S)) {
2785 fprintf(fout, "\n cond_s = (%s%s < 0);",
2786 lmod_cast_s(po, po->operand[0].lmod), dst_opr_text);
2787 *pfomask &= ~(1 << PFO_S);
2788 }
2789}
2790
c0050df6 2791static void output_pp_attrs(FILE *fout, const struct parsed_proto *pp,
2792 int is_noreturn)
2793{
2794 if (pp->is_fastcall)
2795 fprintf(fout, "__fastcall ");
2796 else if (pp->is_stdcall && pp->argc_reg == 0)
2797 fprintf(fout, "__stdcall ");
2798 if (pp->is_noreturn || is_noreturn)
2799 fprintf(fout, "noreturn ");
2800}
2801
91977a1c 2802static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
2803{
69a3cdfc 2804 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
850c9265 2805 struct parsed_opr *last_arith_dst = NULL;
3ebea2cf 2806 char buf1[256], buf2[256], buf3[256], cast[64];
bd96f656 2807 const struct parsed_proto *pp_c;
ddaf8bd7 2808 struct parsed_proto *pp, *pp_tmp;
4c45fa73 2809 struct parsed_data *pd;
91977a1c 2810 const char *tmpname;
940e8e66 2811 enum parsed_flag_op pfo;
69a3cdfc 2812 int save_arg_vars = 0;
cb090db0 2813 int cond_vars = 0;
108e9fe3 2814 int need_tmp_var = 0;
2fe80fdb 2815 int need_tmp64 = 0;
91977a1c 2816 int had_decl = 0;
3ebea2cf 2817 int label_pending = 0;
d4e3b5db 2818 int regmask_save = 0;
91977a1c 2819 int regmask_arg = 0;
a3684be1 2820 int regmask_now = 0;
ddaf8bd7 2821 int regmask_init = 0;
91977a1c 2822 int regmask = 0;
940e8e66 2823 int pfomask = 0;
64c59faf 2824 int found = 0;
d4e3b5db 2825 int depth = 0;
91977a1c 2826 int no_output;
4c45fa73 2827 int i, j, l;
69a3cdfc 2828 int dummy;
91977a1c 2829 int arg;
91977a1c 2830 int reg;
2831 int ret;
2832
1bafb621 2833 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
a2c1d768 2834 g_stack_frame_used = 0;
91977a1c 2835
36595fd2 2836 g_func_pp = proto_parse(fhdr, funcn, 0);
bd96f656 2837 if (g_func_pp == NULL)
91977a1c 2838 ferr(ops, "proto_parse failed for '%s'\n", funcn);
2839
bd96f656 2840 fprintf(fout, "%s ", g_func_pp->ret_type.name);
c0050df6 2841 output_pp_attrs(fout, g_func_pp, g_ida_func_attr & IDAFA_NORETURN);
cb090db0 2842 fprintf(fout, "%s(", g_func_pp->name);
54e763a1 2843
bd96f656 2844 for (i = 0; i < g_func_pp->argc; i++) {
91977a1c 2845 if (i > 0)
2846 fprintf(fout, ", ");
54e763a1 2847 if (g_func_pp->arg[i].fptr != NULL) {
2848 // func pointer..
2849 pp = g_func_pp->arg[i].fptr;
2850 fprintf(fout, "%s (", pp->ret_type.name);
c0050df6 2851 output_pp_attrs(fout, pp, 0);
54e763a1 2852 fprintf(fout, "*a%d)(", i + 1);
2853 for (j = 0; j < pp->argc; j++) {
2854 if (j > 0)
2855 fprintf(fout, ", ");
2856 if (pp->arg[j].fptr)
2857 ferr(ops, "nested fptr\n");
2858 fprintf(fout, "%s", pp->arg[j].type.name);
2859 }
efea2951 2860 if (pp->is_vararg) {
2861 if (j > 0)
2862 fprintf(fout, ", ");
2863 fprintf(fout, "...");
2864 }
54e763a1 2865 fprintf(fout, ")");
2866 }
2867 else {
2868 fprintf(fout, "%s a%d", g_func_pp->arg[i].type.name, i + 1);
2869 }
89ff3147 2870 if (g_func_pp->arg[i].reg != NULL) {
2871 reg = char_array_i(regs_r32,
2872 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
2873 if (reg < 0)
2874 ferr(ops, "arg '%s' is not a reg?\n", g_func_pp->arg[i].reg);
2875 regmask_arg |= 1 << reg;
2876 }
91977a1c 2877 }
bd96f656 2878 if (g_func_pp->is_vararg) {
4f12f671 2879 if (i > 0)
2880 fprintf(fout, ", ");
2881 fprintf(fout, "...");
2882 }
54e763a1 2883
91977a1c 2884 fprintf(fout, ")\n{\n");
2885
2886 // pass1:
1bafb621 2887 // - handle ebp/esp frame, remove ops related to it
91977a1c 2888 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
2889 && ops[1].op == OP_MOV
2890 && IS(opr_name(&ops[1], 0), "ebp")
2891 && IS(opr_name(&ops[1], 1), "esp"))
2892 {
87bf6cec 2893 int ecx_push = 0;
2894
91977a1c 2895 g_bp_frame = 1;
69a3cdfc 2896 ops[0].flags |= OPF_RMD;
2897 ops[1].flags |= OPF_RMD;
4c45fa73 2898 i = 2;
91977a1c 2899
2900 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
1bafb621 2901 g_stack_fsz = opr_const(&ops[2], 1);
69a3cdfc 2902 ops[2].flags |= OPF_RMD;
4c45fa73 2903 i++;
91977a1c 2904 }
87bf6cec 2905 else {
2906 // another way msvc builds stack frame..
2907 i = 2;
2908 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
1bafb621 2909 g_stack_fsz += 4;
87bf6cec 2910 ops[i].flags |= OPF_RMD;
2911 ecx_push++;
2912 i++;
2913 }
4c45fa73 2914 // and another way..
2915 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
2916 && ops[i].operand[1].type == OPT_CONST
2917 && ops[i + 1].op == OP_CALL
2918 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
2919 {
2920 g_stack_fsz += ops[i].operand[1].val;
2921 ops[i].flags |= OPF_RMD;
2922 i++;
2923 ops[i].flags |= OPF_RMD;
2924 i++;
2925 }
87bf6cec 2926 }
91977a1c 2927
64c59faf 2928 found = 0;
91977a1c 2929 do {
2930 for (; i < opcnt; i++)
2931 if (ops[i].op == OP_RET)
2932 break;
2fe80fdb 2933 j = i - 1;
2934 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
2935 if (found) {
2936 if (ops[j].op == OP_JMP && !ops[j].operand[0].had_ds)
2937 // probably local branch back..
2938 break;
2939 if (ops[j].op == OP_CALL)
2940 // probably noreturn call..
2941 break;
2942 }
2943 j--;
2944 }
64c59faf 2945
2fe80fdb 2946 if ((ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ebp"))
2947 || ops[j].op == OP_LEAVE)
591721d7 2948 {
2fe80fdb 2949 ops[j].flags |= OPF_RMD;
591721d7 2950 }
e56ab892 2951 else if (!(g_ida_func_attr & IDAFA_NORETURN))
2fe80fdb 2952 ferr(&ops[j], "'pop ebp' expected\n");
91977a1c 2953
1bafb621 2954 if (g_stack_fsz != 0) {
2fe80fdb 2955 if (ops[j - 1].op == OP_MOV
2956 && IS(opr_name(&ops[j - 1], 0), "esp")
2957 && IS(opr_name(&ops[j - 1], 1), "ebp"))
91977a1c 2958 {
2fe80fdb 2959 ops[j - 1].flags |= OPF_RMD;
91977a1c 2960 }
2fe80fdb 2961 else if (ops[j].op != OP_LEAVE
591721d7 2962 && !(g_ida_func_attr & IDAFA_NORETURN))
2963 {
2fe80fdb 2964 ferr(&ops[j - 1], "esp restore expected\n");
591721d7 2965 }
87bf6cec 2966
2fe80fdb 2967 if (ecx_push && ops[j - 2].op == OP_POP
2968 && IS(opr_name(&ops[j - 2], 0), "ecx"))
87bf6cec 2969 {
2fe80fdb 2970 ferr(&ops[j - 2], "unexpected ecx pop\n");
87bf6cec 2971 }
91977a1c 2972 }
87bf6cec 2973
64c59faf 2974 found = 1;
91977a1c 2975 i++;
2976 } while (i < opcnt);
2977 }
1bafb621 2978 else {
2979 for (i = 0; i < opcnt; i++) {
2980 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
2981 break;
2982 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
2983 && ops[i].operand[1].type == OPT_CONST)
2984 {
2985 g_sp_frame = 1;
2986 break;
2987 }
2988 }
2989
2990 if (g_sp_frame)
2991 {
2992 g_stack_fsz = ops[i].operand[1].val;
2993 ops[i].flags |= OPF_RMD;
2994
2995 i++;
2996 do {
2997 for (; i < opcnt; i++)
2998 if (ops[i].op == OP_RET)
2999 break;
3000 if (ops[i - 1].op != OP_ADD
3001 || !IS(opr_name(&ops[i - 1], 0), "esp")
3002 || ops[i - 1].operand[1].type != OPT_CONST
3003 || ops[i - 1].operand[1].val != g_stack_fsz)
3004 ferr(&ops[i - 1], "'add esp' expected\n");
3005 ops[i - 1].flags |= OPF_RMD;
3006
3007 i++;
3008 } while (i < opcnt);
3009 }
3010 }
91977a1c 3011
3012 // pass2:
7ae48d73 3013 // - parse calls with labels
87bf6cec 3014 // - resolve all branches
de50b98b 3015 for (i = 0; i < opcnt; i++)
3016 {
87bf6cec 3017 po = &ops[i];
3018 po->bt_i = -1;
4c45fa73 3019 po->btj = NULL;
87bf6cec 3020
7ae48d73 3021 if (po->flags & OPF_RMD)
3022 continue;
3023
3024 if (po->op == OP_CALL) {
ddaf8bd7 3025 pp = NULL;
3026
7ae48d73 3027 if (po->operand[0].type == OPT_LABEL) {
3028 tmpname = opr_name(po, 0);
46411e6c 3029 if (IS_START(tmpname, "loc_"))
3030 ferr(po, "call to loc_*\n");
36595fd2 3031 pp_c = proto_parse(fhdr, tmpname, 0);
7ae48d73 3032 if (pp_c == NULL)
3033 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
da87ae38 3034
7ae48d73 3035 pp = proto_clone(pp_c);
3036 my_assert_not(pp, NULL);
ddaf8bd7 3037 }
3038 else if (po->datap != NULL) {
3039 pp = calloc(1, sizeof(*pp));
3040 my_assert_not(pp, NULL);
3041
3042 ret = parse_protostr(po->datap, pp);
3043 if (ret < 0)
3044 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
3045 free(po->datap);
3046 po->datap = NULL;
3047 }
3048
3049 if (pp != NULL) {
3050 if (pp->is_fptr)
3051 check_func_pp(po, pp, "fptr var call");
3052 if (pp->is_noreturn)
3053 po->flags |= OPF_TAIL;
7ae48d73 3054 po->datap = pp;
3055 }
3056 continue;
3057 }
3058
3059 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
87bf6cec 3060 continue;
3061
4c45fa73 3062 if (po->operand[0].type == OPT_REGMEM) {
3063 char *p = strchr(po->operand[0].name, '[');
3064 if (p == NULL)
89ff3147 3065 goto tailcall;
4c45fa73 3066 ret = p - po->operand[0].name;
3067 strncpy(buf1, po->operand[0].name, ret);
3068 buf1[ret] = 0;
3069
3070 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
3071 if (IS(g_func_pd[j].label, buf1)) {
3072 pd = &g_func_pd[j];
3073 break;
3074 }
3075 }
3076 if (pd == NULL)
840257f6 3077 //ferr(po, "label '%s' not parsed?\n", buf1);
3078 goto tailcall;
4c45fa73 3079 if (pd->type != OPT_OFFSET)
3080 ferr(po, "label '%s' with non-offset data?\n", buf1);
3081
3082 // find all labels, link
3083 for (j = 0; j < pd->count; j++) {
3084 for (l = 0; l < opcnt; l++) {
3085 if (g_labels[l][0] && IS(g_labels[l], pd->d[j].u.label)) {
3086 add_label_ref(&g_label_refs[l], i);
3087 pd->d[j].bt_i = l;
3088 break;
3089 }
3090 }
3091 }
3092
3093 po->btj = pd;
3094 continue;
3095 }
3096
3097 for (l = 0; l < opcnt; l++) {
3098 if (g_labels[l][0] && IS(po->operand[0].name, g_labels[l])) {
3099 add_label_ref(&g_label_refs[l], i);
3100 po->bt_i = l;
87bf6cec 3101 break;
3102 }
3103 }
3104
4c45fa73 3105 if (po->bt_i != -1)
3106 continue;
3107
840257f6 3108 if (po->operand[0].type == OPT_LABEL)
87bf6cec 3109 // assume tail call
840257f6 3110 goto tailcall;
4c45fa73 3111
3112 ferr(po, "unhandled branch\n");
840257f6 3113
3114tailcall:
3115 po->op = OP_CALL;
3116 po->flags |= OPF_TAIL;
2fe80fdb 3117 if (i > 0 && ops[i - 1].op == OP_POP)
3118 po->flags |= OPF_ATAIL;
7ae48d73 3119 i--; // reprocess
87bf6cec 3120 }
3121
3122 // pass3:
a2c1d768 3123 // - remove dead labels
91977a1c 3124 // - process calls
1bafb621 3125 for (i = 0; i < opcnt; i++)
3126 {
a2c1d768 3127 if (g_labels[i][0] != 0 && g_label_refs[i].i == -1)
3128 g_labels[i][0] = 0;
3129
69a3cdfc 3130 po = &ops[i];
3131 if (po->flags & OPF_RMD)
91977a1c 3132 continue;
850c9265 3133
d4e3b5db 3134 if (po->op == OP_CALL)
69a3cdfc 3135 {
1bafb621 3136 tmpname = opr_name(po, 0);
7ae48d73 3137 pp = po->datap;
3138 if (pp == NULL)
1bafb621 3139 {
7ae48d73 3140 // indirect call
89ff3147 3141 pp_c = resolve_icall(i, opcnt, &l);
3142 if (pp_c != NULL) {
1cd4a663 3143 pp = proto_clone(pp_c);
89ff3147 3144 my_assert_not(pp, NULL);
3145 if (l)
3146 // not resolved just to single func
3147 pp->is_fptr = 1;
3148 }
1cd4a663 3149 if (pp == NULL) {
3150 pp = calloc(1, sizeof(*pp));
3151 my_assert_not(pp, NULL);
89ff3147 3152 pp->is_fptr = 1;
4741fdfe 3153 ret = scan_for_esp_adjust(i + 1, opcnt, &j, &l);
89ff3147 3154 if (ret < 0) {
3155 if (!g_allow_regfunc)
3156 ferr(po, "non-__cdecl indirect call unhandled yet\n");
3157 pp->is_unresolved = 1;
3158 j = 0;
3159 }
1cd4a663 3160 j /= 4;
3161 if (j > ARRAY_SIZE(pp->arg))
89ff3147 3162 ferr(po, "esp adjust too large: %d\n", j);
1cd4a663 3163 pp->ret_type.name = strdup("int");
3164 pp->argc = pp->argc_stack = j;
3165 for (arg = 0; arg < pp->argc; arg++)
3166 pp->arg[arg].type.name = strdup("int");
3167 }
7ae48d73 3168 po->datap = pp;
1bafb621 3169 }
3170
7ba45c34 3171 // look for and make use of esp adjust
3172 ret = -1;
e56ab892 3173 if (!pp->is_stdcall && pp->argc_stack > 0)
4741fdfe 3174 ret = scan_for_esp_adjust(i + 1, opcnt, &j, &l);
1bafb621 3175 if (ret >= 0) {
7ba45c34 3176 if (pp->is_vararg) {
3177 if (j / 4 < pp->argc_stack)
3178 ferr(po, "esp adjust is too small: %x < %x\n",
3179 j, pp->argc_stack * 4);
3180 // modify pp to make it have varargs as normal args
3181 arg = pp->argc;
3182 pp->argc += j / 4 - pp->argc_stack;
3183 for (; arg < pp->argc; arg++) {
3ebea2cf 3184 pp->arg[arg].type.name = strdup("int");
7ba45c34 3185 pp->argc_stack++;
3186 }
3187 if (pp->argc > ARRAY_SIZE(pp->arg))
e56ab892 3188 ferr(po, "too many args for '%s'\n", tmpname);
7ba45c34 3189 }
1bafb621 3190 if (pp->argc_stack != j / 4)
e56ab892 3191 ferr(po, "stack tracking failed for '%s': %x %x\n",
3192 tmpname, pp->argc_stack * 4, j);
7ba45c34 3193
1bafb621 3194 ops[ret].flags |= OPF_RMD;
591721d7 3195 if (ops[ret].op == OP_POP && j > 4) {
3196 // deal with multi-pop stack adjust
3197 j = pp->argc_stack;
3198 while (ops[ret].op == OP_POP && j > 0 && ret < opcnt) {
3199 ops[ret].flags |= OPF_RMD;
3200 j--;
3201 ret++;
3202 }
3203 }
4741fdfe 3204 else if (!l) {
591721d7 3205 // a bit of a hack, but deals with use of
3206 // single adj for multiple calls
3207 ops[ret].operand[1].val -= j;
3208 }
1bafb621 3209 }
7ba45c34 3210 else if (pp->is_vararg)
3211 ferr(po, "missing esp_adjust for vararg func '%s'\n",
3212 pp->name);
91977a1c 3213
2fe80fdb 3214 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
89ff3147 3215 // since we know the args, collect them
3216 collect_call_args(po, i, pp, &regmask, &save_arg_vars,
3217 i + opcnt * 2);
3218 }
2b43685d 3219
3220 if (strstr(pp->ret_type.name, "int64"))
2fe80fdb 3221 need_tmp64 = 1;
91977a1c 3222 }
d4e3b5db 3223 }
3224
3225 // pass4:
3226 // - find POPs for PUSHes, rm both
591721d7 3227 // - scan for STD/CLD, propagate DF
d4e3b5db 3228 // - scan for all used registers
3229 // - find flag set ops for their users
89ff3147 3230 // - do unreselved calls
1bafb621 3231 // - declare indirect functions
d4e3b5db 3232 for (i = 0; i < opcnt; i++) {
3233 po = &ops[i];
3234 if (po->flags & OPF_RMD)
3235 continue;
3236
7ae48d73 3237 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE)) {
3238 reg = po->operand[0].reg;
3239 if (!(regmask & (1 << reg)))
3240 // not a reg save after all, rerun scan_for_pop
3241 po->flags &= ~OPF_RSAVE;
3242 else
3243 regmask_save |= 1 << reg;
3244 }
3245
5c024ef7 3246 if (po->op == OP_PUSH && po->argnum == 0
3247 && !(po->flags & OPF_RSAVE))
d4e3b5db 3248 {
5c024ef7 3249 if (po->operand[0].type == OPT_REG)
3250 {
3251 reg = po->operand[0].reg;
3252 if (reg < 0)
3253 ferr(po, "reg not set for push?\n");
3254
3255 depth = 0;
3256 ret = scan_for_pop(i + 1, opcnt,
3257 po->operand[0].name, i + opcnt * 3, 0, &depth, 0);
3258 if (ret == 1) {
3259 if (depth > 1)
3260 ferr(po, "too much depth: %d\n", depth);
3261
3262 po->flags |= OPF_RMD;
3263 scan_for_pop(i + 1, opcnt, po->operand[0].name,
3264 i + opcnt * 4, 0, &depth, 1);
3265 continue;
3266 }
3267 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
3268 if (ret == 0) {
3269 arg = OPF_RMD;
3270 if (regmask & (1 << reg)) {
3271 if (regmask_save & (1 << reg))
3272 ferr(po, "%s already saved?\n", po->operand[0].name);
3273 arg = OPF_RSAVE;
3274 }
3275 po->flags |= arg;
3276 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg);
3277 continue;
3278 }
d4e3b5db 3279 }
5c024ef7 3280 else if (po->operand[0].type == OPT_CONST) {
3281 for (j = i + 1; j < opcnt; j++) {
3282 if ((ops[j].flags & (OPF_JMP|OPF_TAIL|OPF_RSAVE))
3283 || ops[j].op == OP_PUSH || g_labels[i][0] != 0)
3284 {
3285 break;
3286 }
3287
3288 if (!(ops[j].flags & OPF_RMD) && ops[j].op == OP_POP)
3289 {
3290 po->flags |= OPF_RMD;
3291 ops[j].datap = po;
3292 break;
3293 }
d4e3b5db 3294 }
d4e3b5db 3295 }
3296 }
3297
591721d7 3298 if (po->op == OP_STD) {
3299 po->flags |= OPF_DF | OPF_RMD;
3300 scan_propagate_df(i + 1, opcnt);
3301 }
3302
a3684be1 3303 regmask_now = po->regmask_src | po->regmask_dst;
3304 if (regmask_now & (1 << xBP)) {
3305 if (g_bp_frame && !(po->flags & OPF_EBP_S)) {
3306 if (po->regmask_dst & (1 << xBP))
3307 // compiler decided to drop bp frame and use ebp as scratch
3308 scan_fwd_set_flags(i, opcnt, i + opcnt * 5, OPF_EBP_S);
3309 else
3310 regmask_now &= ~(1 << xBP);
3311 }
3312 }
3313
3314 regmask |= regmask_now;
d4e3b5db 3315
3316 if (po->flags & OPF_CC)
3317 {
2b43685d 3318 int setters[16], cnt = 0, branched = 0;
3319
04f8a628 3320 ret = scan_for_flag_set(i, i + opcnt * 6,
3321 &branched, setters, &cnt);
2b43685d 3322 if (ret < 0 || cnt <= 0)
3323 ferr(po, "unable to trace flag setter(s)\n");
3324 if (cnt > ARRAY_SIZE(setters))
3325 ferr(po, "too many flag setters\n");
d4e3b5db 3326
d4e3b5db 3327 pfo = split_cond(po, po->op, &dummy);
2b43685d 3328 for (j = 0; j < cnt; j++)
3329 {
3330 tmp_op = &ops[setters[j]]; // flag setter
3331 pfomask = 0;
3332
3333 // to get nicer code, we try to delay test and cmp;
3334 // if we can't because of operand modification, or if we
591721d7 3335 // have arith op, or branch, make it calculate flags explicitly
3336 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP)
3337 {
89ff3147 3338 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i, 0) >= 0)
2b43685d 3339 pfomask = 1 << pfo;
3340 }
4741fdfe 3341 else if (tmp_op->op == OP_CMPS || tmp_op->op == OP_SCAS) {
591721d7 3342 pfomask = 1 << pfo;
3343 }
2b43685d 3344 else {
04f8a628 3345 // see if we'll be able to handle based on op result
3346 if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR
3347 && pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P)
3348 || branched
2b43685d 3349 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
3350 pfomask = 1 << pfo;
2fe80fdb 3351
3352 if (tmp_op->op == OP_ADD && pfo == PFO_C)
3353 need_tmp64 = 1;
2b43685d 3354 }
3355 if (pfomask) {
3356 tmp_op->pfomask |= pfomask;
cb090db0 3357 cond_vars |= pfomask;
2b43685d 3358 }
04f8a628 3359 // note: may overwrite, currently not a problem
3360 po->datap = tmp_op;
d4e3b5db 3361 }
3362
cb090db0 3363 if (po->op == OP_RCL || po->op == OP_RCR
3364 || po->op == OP_ADC || po->op == OP_SBB)
3365 cond_vars |= 1 << PFO_C;
d4e3b5db 3366 }
591721d7 3367 else if (po->op == OP_CMPS || po->op == OP_SCAS) {
cb090db0 3368 cond_vars |= 1 << PFO_Z;
591721d7 3369 }
87bf6cec 3370 else if (po->op == OP_MUL
3371 || (po->op == OP_IMUL && po->operand_cnt == 1))
3372 {
2fe80fdb 3373 need_tmp64 = 1;
87bf6cec 3374 }
89ff3147 3375 else if (po->op == OP_CALL) {
1bafb621 3376 pp = po->datap;
89ff3147 3377 if (pp == NULL)
3378 ferr(po, "NULL pp\n");
3379
3380 if (pp->is_unresolved) {
ddaf8bd7 3381 int regmask_stack = 0;
b74c31e3 3382 collect_call_args(po, i, pp, &regmask, &save_arg_vars,
89ff3147 3383 i + opcnt * 2);
3384
b74c31e3 3385 // this is pretty rough guess:
3386 // see ecx and edx were pushed (and not their saved versions)
3387 for (arg = 0; arg < pp->argc; arg++) {
3388 if (pp->arg[arg].reg != NULL)
3389 continue;
3390
3391 tmp_op = pp->arg[arg].datap;
3392 if (tmp_op == NULL)
3393 ferr(po, "parsed_op missing for arg%d\n", arg);
3394 if (tmp_op->argnum == 0 && tmp_op->operand[0].type == OPT_REG)
3395 regmask_stack |= 1 << tmp_op->operand[0].reg;
3396 }
3397
ddaf8bd7 3398 if (!((regmask_stack & (1 << xCX))
3399 && (regmask_stack & (1 << xDX))))
89ff3147 3400 {
3401 if (pp->argc_stack != 0
c0050df6 3402 || ((regmask | regmask_arg) & ((1 << xCX)|(1 << xDX))))
89ff3147 3403 {
3404 pp_insert_reg_arg(pp, "ecx");
c0050df6 3405 pp->is_fastcall = 1;
ddaf8bd7 3406 regmask_init |= 1 << xCX;
89ff3147 3407 regmask |= 1 << xCX;
3408 }
3409 if (pp->argc_stack != 0
3410 || ((regmask | regmask_arg) & (1 << xDX)))
3411 {
3412 pp_insert_reg_arg(pp, "edx");
ddaf8bd7 3413 regmask_init |= 1 << xDX;
89ff3147 3414 regmask |= 1 << xDX;
3415 }
3416 }
c0050df6 3417
3418 // note: __cdecl doesn't fall into is_unresolved category
3419 if (pp->argc_stack > 0)
3420 pp->is_stdcall = 1;
ddaf8bd7 3421 }
3422
3423 for (arg = 0; arg < pp->argc; arg++) {
3424 if (pp->arg[arg].reg != NULL) {
3425 reg = char_array_i(regs_r32,
3426 ARRAY_SIZE(regs_r32), pp->arg[arg].reg);
3427 if (reg < 0)
3428 ferr(ops, "arg '%s' is not a reg?\n", pp->arg[arg].reg);
3429 if (!(regmask & (1 << reg))) {
3430 regmask_init |= 1 << reg;
3431 regmask |= 1 << reg;
3432 }
3433 }
89ff3147 3434 }
3435
efea2951 3436 // declare indirect funcs
3437 if (pp->is_fptr && !(pp->name[0] != 0 && pp->is_arg)) {
ddaf8bd7 3438 if (pp->name[0] != 0) {
3439 memmove(pp->name + 2, pp->name, strlen(pp->name) + 1);
3440 memcpy(pp->name, "i_", 2);
3441
3442 // might be declared already
3443 found = 0;
3444 for (j = 0; j < i; j++) {
3445 if (ops[j].op == OP_CALL && (pp_tmp = ops[j].datap)) {
3446 if (pp_tmp->is_fptr && IS(pp->name, pp_tmp->name)) {
3447 found = 1;
3448 break;
3449 }
3450 }
3451 }
3452 if (found)
3453 continue;
3454 }
3455 else
3456 snprintf(pp->name, sizeof(pp->name), "icall%d", i);
3457
89ff3147 3458 fprintf(fout, " %s (", pp->ret_type.name);
c0050df6 3459 output_pp_attrs(fout, pp, 0);
ddaf8bd7 3460 fprintf(fout, "*%s)(", pp->name);
89ff3147 3461 for (j = 0; j < pp->argc; j++) {
3462 if (j > 0)
3463 fprintf(fout, ", ");
3464 fprintf(fout, "%s a%d", pp->arg[j].type.name, j + 1);
3465 }
3466 fprintf(fout, ");\n");
1bafb621 3467 }
1bafb621 3468 }
75ad0378 3469 else if (po->op == OP_RET && !IS(g_func_pp->ret_type.name, "void"))
3470 regmask |= 1 << xAX;
cb090db0 3471 else if (po->op == OP_DIV || po->op == OP_IDIV) {
3472 // 32bit division is common, look for it
3473 if (po->op == OP_DIV)
3474 ret = scan_for_reg_clear(i, xDX);
3475 else
3476 ret = scan_for_cdq_edx(i);
3477 if (ret >= 0)
3478 po->flags |= OPF_32BIT;
3479 else
3480 need_tmp64 = 1;
3481 }
3482
3483 if (po->op == OP_RCL || po->op == OP_RCR || po->op == OP_XCHG) {
3484 need_tmp_var = 1;
3485 }
91977a1c 3486 }
3487
da87ae38 3488 // pass4:
3489 // - confirm regmask_save, it might have been reduced
3490 if (regmask_save != 0)
3491 {
3492 regmask_save = 0;
3493 for (i = 0; i < opcnt; i++) {
3494 po = &ops[i];
3495 if (po->flags & OPF_RMD)
3496 continue;
3497
3498 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE))
3499 regmask_save |= 1 << po->operand[0].reg;
3500 }
3501 }
3502
3503
4c45fa73 3504 // output LUTs/jumptables
3505 for (i = 0; i < g_func_pd_cnt; i++) {
3506 pd = &g_func_pd[i];
3507 fprintf(fout, " static const ");
3508 if (pd->type == OPT_OFFSET) {
3509 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
3510
3511 for (j = 0; j < pd->count; j++) {
3512 if (j > 0)
3513 fprintf(fout, ", ");
3514 fprintf(fout, "&&%s", pd->d[j].u.label);
3515 }
3516 }
3517 else {
3518 fprintf(fout, "%s %s[] =\n { ",
3519 lmod_type_u(ops, pd->lmod), pd->label);
3520
3521 for (j = 0; j < pd->count; j++) {
3522 if (j > 0)
3523 fprintf(fout, ", ");
3524 fprintf(fout, "%u", pd->d[j].u.val);
3525 }
3526 }
3527 fprintf(fout, " };\n");
3528 }
3529
4f12f671 3530 // declare stack frame, va_arg
1bafb621 3531 if (g_stack_fsz)
850c9265 3532 fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
1bafb621 3533 (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz);
850c9265 3534
bd96f656 3535 if (g_func_pp->is_vararg)
4f12f671 3536 fprintf(fout, " va_list ap;\n");
3537
940e8e66 3538 // declare arg-registers
bd96f656 3539 for (i = 0; i < g_func_pp->argc; i++) {
3540 if (g_func_pp->arg[i].reg != NULL) {
91977a1c 3541 reg = char_array_i(regs_r32,
bd96f656 3542 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
75ad0378 3543 if (regmask & (1 << reg)) {
3544 fprintf(fout, " u32 %s = (u32)a%d;\n",
3545 g_func_pp->arg[i].reg, i + 1);
3546 }
3547 else
3548 fprintf(fout, " // %s = a%d; // unused\n",
3549 g_func_pp->arg[i].reg, i + 1);
91977a1c 3550 had_decl = 1;
3551 }
3552 }
3553
75ad0378 3554 regmask_now = regmask & ~regmask_arg;
3555 regmask_now &= ~(1 << xSP);
3556 if (regmask_now) {
91977a1c 3557 for (reg = 0; reg < 8; reg++) {
75ad0378 3558 if (regmask_now & (1 << reg)) {
ddaf8bd7 3559 fprintf(fout, " u32 %s", regs_r32[reg]);
3560 if (regmask_init & (1 << reg))
3561 fprintf(fout, " = 0");
3562 fprintf(fout, ";\n");
91977a1c 3563 had_decl = 1;
3564 }
3565 }
3566 }
3567
d4e3b5db 3568 if (regmask_save) {
3569 for (reg = 0; reg < 8; reg++) {
3570 if (regmask_save & (1 << reg)) {
3571 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
3572 had_decl = 1;
3573 }
3574 }
3575 }
3576
69a3cdfc 3577 if (save_arg_vars) {
3578 for (reg = 0; reg < 32; reg++) {
3579 if (save_arg_vars & (1 << reg)) {
3580 fprintf(fout, " u32 s_a%d;\n", reg + 1);
3581 had_decl = 1;
3582 }
3583 }
3584 }
3585
cb090db0 3586 if (cond_vars) {
69a3cdfc 3587 for (i = 0; i < 8; i++) {
cb090db0 3588 if (cond_vars & (1 << i)) {
69a3cdfc 3589 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
3590 had_decl = 1;
3591 }
3592 }
3593 }
3594
108e9fe3 3595 if (need_tmp_var) {
3596 fprintf(fout, " u32 tmp;\n");
3597 had_decl = 1;
3598 }
3599
2fe80fdb 3600 if (need_tmp64) {
3601 fprintf(fout, " u64 tmp64;\n");
87bf6cec 3602 had_decl = 1;
3603 }
3604
91977a1c 3605 if (had_decl)
3606 fprintf(fout, "\n");
3607
bd96f656 3608 if (g_func_pp->is_vararg) {
3609 if (g_func_pp->argc_stack == 0)
4f12f671 3610 ferr(ops, "vararg func without stack args?\n");
bd96f656 3611 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
4f12f671 3612 }
3613
91977a1c 3614 // output ops
69a3cdfc 3615 for (i = 0; i < opcnt; i++)
3616 {
a2c1d768 3617 if (g_labels[i][0] != 0) {
91977a1c 3618 fprintf(fout, "\n%s:\n", g_labels[i]);
3ebea2cf 3619 label_pending = 1;
2b43685d 3620
3621 delayed_flag_op = NULL;
3622 last_arith_dst = NULL;
3ebea2cf 3623 }
91977a1c 3624
69a3cdfc 3625 po = &ops[i];
3626 if (po->flags & OPF_RMD)
91977a1c 3627 continue;
3628
3629 no_output = 0;
3630
91977a1c 3631 #define assert_operand_cnt(n_) \
850c9265 3632 if (po->operand_cnt != n_) \
3633 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
3634
69a3cdfc 3635 // conditional/flag using op?
3636 if (po->flags & OPF_CC)
850c9265 3637 {
940e8e66 3638 int is_delayed = 0;
3639 int is_inv = 0;
69a3cdfc 3640
940e8e66 3641 pfo = split_cond(po, po->op, &is_inv);
04f8a628 3642 tmp_op = po->datap;
850c9265 3643
69a3cdfc 3644 // we go through all this trouble to avoid using parsed_flag_op,
3645 // which makes generated code much nicer
3646 if (delayed_flag_op != NULL)
850c9265 3647 {
940e8e66 3648 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op, pfo, is_inv);
3649 is_delayed = 1;
91977a1c 3650 }
850c9265 3651 else if (last_arith_dst != NULL
04f8a628 3652 && (pfo == PFO_Z || pfo == PFO_S || pfo == PFO_P
3653 || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR))
3654 ))
850c9265 3655 {
3ebea2cf 3656 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
940e8e66 3657 out_test_for_cc(buf1, sizeof(buf1), po, pfo, is_inv,
850c9265 3658 last_arith_dst->lmod, buf3);
940e8e66 3659 is_delayed = 1;
850c9265 3660 }
04f8a628 3661 else if (tmp_op != NULL) {
7ba45c34 3662 // use preprocessed flag calc results
04f8a628 3663 if (!(tmp_op->pfomask & (1 << pfo)))
69a3cdfc 3664 ferr(po, "not prepared for pfo %d\n", pfo);
3665
940e8e66 3666 // note: is_inv was not yet applied
69a3cdfc 3667 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
940e8e66 3668 is_inv ? "!" : "", parsed_flag_op_names[pfo]);
69a3cdfc 3669 }
3670 else {
3671 ferr(po, "all methods of finding comparison failed\n");
3672 }
850c9265 3673
69a3cdfc 3674 if (po->flags & OPF_JMP) {
850c9265 3675 fprintf(fout, " if %s\n", buf1);
3676 }
cb090db0 3677 else if (po->op == OP_RCL || po->op == OP_RCR
3678 || po->op == OP_ADC || po->op == OP_SBB)
3679 {
940e8e66 3680 if (is_delayed)
3681 fprintf(fout, " cond_%s = %s;\n",
3682 parsed_flag_op_names[pfo], buf1);
850c9265 3683 }
5101a5f9 3684 else if (po->flags & OPF_DATA) { // SETcc
850c9265 3685 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
3686 fprintf(fout, " %s = %s;", buf2, buf1);
91977a1c 3687 }
69a3cdfc 3688 else {
3689 ferr(po, "unhandled conditional op\n");
3690 }
91977a1c 3691 }
3692
940e8e66 3693 pfomask = po->pfomask;
3694
4741fdfe 3695 if (po->flags & (OPF_REPZ|OPF_REPNZ)) {
3696 // we need initial flags for ecx=0 case..
3697 if (i > 0 && ops[i - 1].op == OP_XOR
3698 && IS(ops[i - 1].operand[0].name,
3699 ops[i - 1].operand[1].name))
3700 {
3701 fprintf(fout, " cond_z = ");
3702 if (pfomask & (1 << PFO_C))
3703 fprintf(fout, "cond_c = ");
3704 fprintf(fout, "0;\n");
3705 }
3706 else if (last_arith_dst != NULL) {
3707 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
3708 out_test_for_cc(buf1, sizeof(buf1), po, PFO_Z, 0,
3709 last_arith_dst->lmod, buf3);
3710 fprintf(fout, " cond_z = %s;\n", buf1);
3711 }
3712 else
3713 ferr(po, "missing initial ZF\n");
3714 }
3715
850c9265 3716 switch (po->op)
91977a1c 3717 {
3718 case OP_MOV:
3719 assert_operand_cnt(2);
850c9265 3720 propagate_lmod(po, &po->operand[0], &po->operand[1]);
de50b98b 3721 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3722 fprintf(fout, " %s = %s;", buf1,
3ebea2cf 3723 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
3724 po->operand[0].is_ptr ? "(void *)" : "", 0));
850c9265 3725 break;
3726
3727 case OP_LEA:
3728 assert_operand_cnt(2);
87bf6cec 3729 po->operand[1].lmod = OPLM_DWORD; // always
850c9265 3730 fprintf(fout, " %s = %s;",
3731 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3ebea2cf 3732 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
3733 NULL, 1));
850c9265 3734 break;
3735
3736 case OP_MOVZX:
3737 assert_operand_cnt(2);
91977a1c 3738 fprintf(fout, " %s = %s;",
850c9265 3739 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3ebea2cf 3740 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
850c9265 3741 break;
3742
3743 case OP_MOVSX:
3744 assert_operand_cnt(2);
3745 switch (po->operand[1].lmod) {
3746 case OPLM_BYTE:
3747 strcpy(buf3, "(s8)");
3748 break;
3749 case OPLM_WORD:
3750 strcpy(buf3, "(s16)");
3751 break;
3752 default:
3753 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
3754 }
a2c1d768 3755 fprintf(fout, " %s = %s;",
850c9265 3756 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
a2c1d768 3757 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
3758 buf3, 0));
850c9265 3759 break;
3760
108e9fe3 3761 case OP_XCHG:
3762 assert_operand_cnt(2);
3763 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3764 fprintf(fout, " tmp = %s;",
3765 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], "", 0));
3766 fprintf(fout, " %s = %s;",
3767 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3768 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], "", 0));
3769 fprintf(fout, " %s = tmp;",
3770 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]));
3771 snprintf(g_comment, sizeof(g_comment), "xchg");
3772 break;
3773
850c9265 3774 case OP_NOT:
3775 assert_operand_cnt(1);
3776 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3777 fprintf(fout, " %s = ~%s;", buf1, buf1);
3778 break;
3779
5101a5f9 3780 case OP_CDQ:
3781 assert_operand_cnt(2);
3782 fprintf(fout, " %s = (s32)%s >> 31;",
3783 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3ebea2cf 3784 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5101a5f9 3785 strcpy(g_comment, "cdq");
3786 break;
3787
33c35af6 3788 case OP_STOS:
33c35af6 3789 assert_operand_cnt(3);
3790 if (po->flags & OPF_REP) {
591721d7 3791 fprintf(fout, " for (; ecx != 0; ecx--, edi %c= %d)\n",
3792 (po->flags & OPF_DF) ? '-' : '+',
33c35af6 3793 lmod_bytes(po, po->operand[0].lmod));
d4e3b5db 3794 fprintf(fout, " %sedi = eax;",
3795 lmod_cast_u_ptr(po, po->operand[0].lmod));
33c35af6 3796 strcpy(g_comment, "rep stos");
3797 }
3798 else {
591721d7 3799 fprintf(fout, " %sedi = eax; edi %c= %d;",
d4e3b5db 3800 lmod_cast_u_ptr(po, po->operand[0].lmod),
591721d7 3801 (po->flags & OPF_DF) ? '-' : '+',
33c35af6 3802 lmod_bytes(po, po->operand[0].lmod));
3803 strcpy(g_comment, "stos");
3804 }
3805 break;
3806
d4e3b5db 3807 case OP_MOVS:
d4e3b5db 3808 assert_operand_cnt(3);
3809 j = lmod_bytes(po, po->operand[0].lmod);
3810 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
591721d7 3811 l = (po->flags & OPF_DF) ? '-' : '+';
d4e3b5db 3812 if (po->flags & OPF_REP) {
3813 fprintf(fout,
591721d7 3814 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n",
3815 l, j, l, j);
d4e3b5db 3816 fprintf(fout,
3817 " %sedi = %sesi;", buf1, buf1);
3818 strcpy(g_comment, "rep movs");
3819 }
3820 else {
591721d7 3821 fprintf(fout, " %sedi = %sesi; edi %c= %d; esi %c= %d;",
3822 buf1, buf1, l, j, l, j);
d4e3b5db 3823 strcpy(g_comment, "movs");
3824 }
3825 break;
3826
7ba45c34 3827 case OP_CMPS:
7ba45c34 3828 // repe ~ repeat while ZF=1
3829 assert_operand_cnt(3);
3830 j = lmod_bytes(po, po->operand[0].lmod);
3831 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
591721d7 3832 l = (po->flags & OPF_DF) ? '-' : '+';
7ba45c34 3833 if (po->flags & OPF_REP) {
3834 fprintf(fout,
4741fdfe 3835 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d) {\n",
591721d7 3836 l, j, l, j);
4741fdfe 3837 if (pfomask & (1 << PFO_C)) {
3838 // ugh..
3839 fprintf(fout,
3840 " cond_c = %sedi < %sesi;\n", buf1, buf1);
3841 pfomask &= ~(1 << PFO_C);
3842 }
7ba45c34 3843 fprintf(fout,
3844 " if ((cond_z = (%sedi == %sesi)) %s 0)\n",
3845 buf1, buf1, (po->flags & OPF_REPZ) ? "==" : "!=");
3846 fprintf(fout,
4741fdfe 3847 " break;\n"
3848 " }");
7ba45c34 3849 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
3850 (po->flags & OPF_REPZ) ? "e" : "ne");
3851 }
3852 else {
3853 fprintf(fout,
591721d7 3854 " cond_z = (%sedi = %sesi); edi %c= %d; esi %c= %d;",
3855 buf1, buf1, l, j, l, j);
7ba45c34 3856 strcpy(g_comment, "cmps");
3857 }
3858 pfomask &= ~(1 << PFO_Z);
3859 last_arith_dst = NULL;
3860 delayed_flag_op = NULL;
3861 break;
3862
591721d7 3863 case OP_SCAS:
3864 // only does ZF (for now)
3865 // repe ~ repeat while ZF=1
3866 assert_operand_cnt(3);
3867 j = lmod_bytes(po, po->operand[0].lmod);
3868 l = (po->flags & OPF_DF) ? '-' : '+';
3869 if (po->flags & OPF_REP) {
3870 fprintf(fout,
3871 " for (; ecx != 0; ecx--, edi %c= %d)\n", l, j);
3872 fprintf(fout,
3873 " if ((cond_z = (%seax == %sedi)) %s 0)\n",
3874 lmod_cast_u(po, po->operand[0].lmod),
3875 lmod_cast_u_ptr(po, po->operand[0].lmod),
3876 (po->flags & OPF_REPZ) ? "==" : "!=");
3877 fprintf(fout,
3878 " break;");
3879 snprintf(g_comment, sizeof(g_comment), "rep%s scas",
3880 (po->flags & OPF_REPZ) ? "e" : "ne");
3881 }
3882 else {
3883 fprintf(fout, " cond_z = (%seax = %sedi); edi %c= %d;",
3884 lmod_cast_u(po, po->operand[0].lmod),
3885 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
3886 strcpy(g_comment, "scas");
3887 }
3888 pfomask &= ~(1 << PFO_Z);
3889 last_arith_dst = NULL;
3890 delayed_flag_op = NULL;
3891 break;
3892
850c9265 3893 // arithmetic w/flags
850c9265 3894 case OP_AND:
3895 case OP_OR:
5101a5f9 3896 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3897 // fallthrough
850c9265 3898 dualop_arith:
3899 assert_operand_cnt(2);
850c9265 3900 fprintf(fout, " %s %s= %s;",
3901 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3902 op_to_c(po),
3ebea2cf 3903 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
04f8a628 3904 output_std_flags(fout, po, &pfomask, buf1);
3905 last_arith_dst = &po->operand[0];
3906 delayed_flag_op = NULL;
3907 break;
3908
3909 case OP_SHL:
3910 case OP_SHR:
3911 assert_operand_cnt(2);
3912 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3913 if (pfomask & (1 << PFO_C)) {
3914 if (po->operand[1].type == OPT_CONST) {
3915 l = lmod_bytes(po, po->operand[0].lmod) * 8;
3916 j = po->operand[1].val;
3917 j %= l;
3918 if (j != 0) {
3919 if (po->op == OP_SHL)
3920 j = l - j;
3921 else
3922 j -= 1;
cb090db0 3923 fprintf(fout, " cond_c = (%s >> %d) & 1;\n",
3924 buf1, j);
04f8a628 3925 }
3926 else
3927 ferr(po, "zero shift?\n");
3928 }
3929 else
3930 ferr(po, "TODO\n");
3931 pfomask &= ~(1 << PFO_C);
840257f6 3932 }
04f8a628 3933 fprintf(fout, " %s %s= %s;", buf1, op_to_c(po),
3934 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
3935 output_std_flags(fout, po, &pfomask, buf1);
850c9265 3936 last_arith_dst = &po->operand[0];
69a3cdfc 3937 delayed_flag_op = NULL;
850c9265 3938 break;
3939
d4e3b5db 3940 case OP_SAR:
3941 assert_operand_cnt(2);
3942 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3943 fprintf(fout, " %s = %s%s >> %s;", buf1,
3944 lmod_cast_s(po, po->operand[0].lmod), buf1,
3ebea2cf 3945 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
04f8a628 3946 output_std_flags(fout, po, &pfomask, buf1);
d4e3b5db 3947 last_arith_dst = &po->operand[0];
3948 delayed_flag_op = NULL;
3949 break;
3950
3951 case OP_ROL:
3952 case OP_ROR:
3953 assert_operand_cnt(2);
3954 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3955 if (po->operand[1].type == OPT_CONST) {
3956 j = po->operand[1].val;
3957 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
3958 fprintf(fout, po->op == OP_ROL ?
3959 " %s = (%s << %d) | (%s >> %d);" :
3960 " %s = (%s >> %d) | (%s << %d);",
3961 buf1, buf1, j, buf1,
3962 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
3963 }
3964 else
3965 ferr(po, "TODO\n");
04f8a628 3966 output_std_flags(fout, po, &pfomask, buf1);
d4e3b5db 3967 last_arith_dst = &po->operand[0];
3968 delayed_flag_op = NULL;
3969 break;
3970
cb090db0 3971 case OP_RCL:
3972 case OP_RCR:
3973 assert_operand_cnt(2);
3974 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3975 l = lmod_bytes(po, po->operand[0].lmod) * 8;
3976 if (po->operand[1].type == OPT_CONST) {
3977 j = po->operand[1].val % l;
3978 if (j == 0)
3979 ferr(po, "zero rotate\n");
3980 fprintf(fout, " tmp = (%s >> %d) & 1;\n",
3981 buf1, (po->op == OP_RCL) ? (l - j) : (j - 1));
3982 if (po->op == OP_RCL) {
3983 fprintf(fout,
3984 " %s = (%s << %d) | (cond_c << %d)",
3985 buf1, buf1, j, j - 1);
3986 if (j != 1)
3987 fprintf(fout, " | (%s >> %d)", buf1, l + 1 - j);
3988 }
3989 else {
3990 fprintf(fout,
3991 " %s = (%s >> %d) | (cond_c << %d)",
3992 buf1, buf1, j, l - j);
3993 if (j != 1)
3994 fprintf(fout, " | (%s << %d)", buf1, l + 1 - j);
3995 }
3996 fprintf(fout, ";\n");
3997 fprintf(fout, " cond_c = tmp;");
3998 }
3999 else
4000 ferr(po, "TODO\n");
4001 strcpy(g_comment, (po->op == OP_RCL) ? "rcl" : "rcr");
4002 output_std_flags(fout, po, &pfomask, buf1);
4003 last_arith_dst = &po->operand[0];
4004 delayed_flag_op = NULL;
4005 break;
4006
5101a5f9 4007 case OP_XOR:
850c9265 4008 assert_operand_cnt(2);
4009 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5101a5f9 4010 if (IS(opr_name(po, 0), opr_name(po, 1))) {
4011 // special case for XOR
2fe80fdb 4012 if (pfomask & (1 << PFO_BE)) { // weird, but it happens..
4013 fprintf(fout, " cond_be = 1;\n");
4014 pfomask &= ~(1 << PFO_BE);
4015 }
5101a5f9 4016 fprintf(fout, " %s = 0;",
4017 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4018 last_arith_dst = &po->operand[0];
4019 delayed_flag_op = NULL;
850c9265 4020 break;
850c9265 4021 }
5101a5f9 4022 goto dualop_arith;
4023
2fe80fdb 4024 case OP_ADD:
4025 assert_operand_cnt(2);
4026 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4027 if (pfomask & (1 << PFO_C)) {
4028 fprintf(fout, " tmp64 = (u64)%s + %s;\n",
4029 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]),
4030 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4031 fprintf(fout, " cond_c = tmp64 >> 32;\n");
4032 fprintf(fout, " %s = (u32)tmp64;",
4033 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4034 strcat(g_comment, "add64");
4035 pfomask &= ~(1 << PFO_C);
4036 output_std_flags(fout, po, &pfomask, buf1);
4037 last_arith_dst = &po->operand[0];
4038 delayed_flag_op = NULL;
4039 break;
4040 }
4041 goto dualop_arith;
4042
4043 case OP_SUB:
4044 assert_operand_cnt(2);
4045 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4046 if (pfomask & (1 << PFO_C)) {
4047 fprintf(fout, " cond_c = %s < %s;\n",
4048 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]),
4049 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4050 pfomask &= ~(1 << PFO_C);
4051 }
4052 goto dualop_arith;
4053
5101a5f9 4054 case OP_ADC:
850c9265 4055 case OP_SBB:
5101a5f9 4056 assert_operand_cnt(2);
4057 propagate_lmod(po, &po->operand[0], &po->operand[1]);
a2c1d768 4058 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
840257f6 4059 if (po->op == OP_SBB
4060 && IS(po->operand[0].name, po->operand[1].name))
4061 {
4062 // avoid use of unitialized var
a2c1d768 4063 fprintf(fout, " %s = -cond_c;", buf1);
94d447fb 4064 // carry remains what it was
4065 pfomask &= ~(1 << PFO_C);
840257f6 4066 }
4067 else {
a2c1d768 4068 fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po),
3ebea2cf 4069 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
840257f6 4070 }
a2c1d768 4071 output_std_flags(fout, po, &pfomask, buf1);
5101a5f9 4072 last_arith_dst = &po->operand[0];
4073 delayed_flag_op = NULL;
850c9265 4074 break;
4075
4076 case OP_INC:
4077 case OP_DEC:
4078 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
1bafb621 4079 if (po->operand[0].type == OPT_REG) {
4080 strcpy(buf2, po->op == OP_INC ? "++" : "--");
4081 fprintf(fout, " %s%s;", buf1, buf2);
4082 }
4083 else {
4084 strcpy(buf2, po->op == OP_INC ? "+" : "-");
4085 fprintf(fout, " %s %s= 1;", buf1, buf2);
4086 }
a2c1d768 4087 output_std_flags(fout, po, &pfomask, buf1);
5101a5f9 4088 last_arith_dst = &po->operand[0];
4089 delayed_flag_op = NULL;
4090 break;
4091
4092 case OP_NEG:
4093 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3ebea2cf 4094 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
5101a5f9 4095 fprintf(fout, " %s = -%s%s;", buf1,
4096 lmod_cast_s(po, po->operand[0].lmod), buf2);
850c9265 4097 last_arith_dst = &po->operand[0];
69a3cdfc 4098 delayed_flag_op = NULL;
940e8e66 4099 if (pfomask & (1 << PFO_C)) {
4100 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
4101 pfomask &= ~(1 << PFO_C);
4102 }
850c9265 4103 break;
4104
4105 case OP_IMUL:
de50b98b 4106 if (po->operand_cnt == 2) {
4107 propagate_lmod(po, &po->operand[0], &po->operand[1]);
850c9265 4108 goto dualop_arith;
de50b98b 4109 }
87bf6cec 4110 if (po->operand_cnt == 3)
4111 ferr(po, "TODO imul3\n");
4112 // fallthrough
4113 case OP_MUL:
4114 assert_operand_cnt(1);
4115 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
2fe80fdb 4116 fprintf(fout, " tmp64 = %seax * %s%s;\n", buf1, buf1,
3ebea2cf 4117 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
2fe80fdb 4118 fprintf(fout, " edx = tmp64 >> 32;\n");
4119 fprintf(fout, " eax = tmp64;");
87bf6cec 4120 last_arith_dst = NULL;
69a3cdfc 4121 delayed_flag_op = NULL;
91977a1c 4122 break;
4123
5101a5f9 4124 case OP_DIV:
4125 case OP_IDIV:
4126 assert_operand_cnt(1);
4127 if (po->operand[0].lmod != OPLM_DWORD)
4128 ferr(po, "unhandled lmod %d\n", po->operand[0].lmod);
4129
cb090db0 4130 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4131 strcpy(buf2, lmod_cast(po, po->operand[0].lmod,
4132 po->op == OP_IDIV));
4133 switch (po->operand[0].lmod) {
4134 case OPLM_DWORD:
4135 if (po->flags & OPF_32BIT)
4136 snprintf(buf3, sizeof(buf3), "%seax", buf2);
4137 else {
4138 fprintf(fout, " tmp64 = ((u64)edx << 32) | eax;\n");
4139 snprintf(buf3, sizeof(buf3), "%stmp64",
4140 (po->op == OP_IDIV) ? "(s64)" : "");
4141 }
4142 if (po->operand[0].type == OPT_REG
4143 && po->operand[0].reg == xDX)
4144 {
4145 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
4146 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
4147 }
4148 else {
4149 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
4150 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
4151 }
4152 break;
4153 default:
4154 ferr(po, "unhandled division type\n");
5101a5f9 4155 }
87bf6cec 4156 last_arith_dst = NULL;
4157 delayed_flag_op = NULL;
5101a5f9 4158 break;
4159
91977a1c 4160 case OP_TEST:
4161 case OP_CMP:
850c9265 4162 propagate_lmod(po, &po->operand[0], &po->operand[1]);
940e8e66 4163 if (pfomask != 0) {
69a3cdfc 4164 for (j = 0; j < 8; j++) {
940e8e66 4165 if (pfomask & (1 << j)) {
69a3cdfc 4166 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
4167 fprintf(fout, " cond_%s = %s;",
4168 parsed_flag_op_names[j], buf1);
4169 }
4170 }
940e8e66 4171 pfomask = 0;
69a3cdfc 4172 }
4173 else
4174 no_output = 1;
7ba45c34 4175 last_arith_dst = NULL;
69a3cdfc 4176 delayed_flag_op = po;
91977a1c 4177 break;
4178
69a3cdfc 4179 // note: we reuse OP_Jcc for SETcc, only flags differ
91977a1c 4180 case OP_JO ... OP_JG:
5101a5f9 4181 if (po->flags & OPF_JMP)
850c9265 4182 fprintf(fout, " goto %s;", po->operand[0].name);
5101a5f9 4183 // else SETcc - should already be handled
850c9265 4184 break;
4185
5c024ef7 4186 case OP_JECXZ:
4187 fprintf(fout, " if (ecx == 0)\n");
4188 fprintf(fout, " goto %s;", po->operand[0].name);
4189 strcat(g_comment, "jecxz");
4190 break;
4191
850c9265 4192 case OP_JMP:
87bf6cec 4193 assert_operand_cnt(1);
de50b98b 4194 last_arith_dst = NULL;
4195 delayed_flag_op = NULL;
4196
4c45fa73 4197 if (po->operand[0].type == OPT_REGMEM) {
4198 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
4199 buf1, buf2);
4200 if (ret != 2)
4201 ferr(po, "parse failure for jmp '%s'\n",
4202 po->operand[0].name);
4203 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
4204 break;
4205 }
4206 else if (po->operand[0].type != OPT_LABEL)
4207 ferr(po, "unhandled jmp type\n");
87bf6cec 4208
850c9265 4209 fprintf(fout, " goto %s;", po->operand[0].name);
91977a1c 4210 break;
4211
4212 case OP_CALL:
5101a5f9 4213 assert_operand_cnt(1);
850c9265 4214 pp = po->datap;
89ff3147 4215 my_assert_not(pp, NULL);
91977a1c 4216
efea2951 4217 if (pp->is_fptr && !pp->is_arg)
ddaf8bd7 4218 fprintf(fout, " %s = %s;\n", pp->name,
1cd4a663 4219 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
4220 "(void *)", 0));
1bafb621 4221
91977a1c 4222 fprintf(fout, " ");
2b43685d 4223 if (strstr(pp->ret_type.name, "int64")) {
87bf6cec 4224 if (po->flags & OPF_TAIL)
2b43685d 4225 ferr(po, "int64 and tail?\n");
2fe80fdb 4226 fprintf(fout, "tmp64 = ");
2b43685d 4227 }
4228 else if (!IS(pp->ret_type.name, "void")) {
4229 if (po->flags & OPF_TAIL) {
840257f6 4230 if (!IS(g_func_pp->ret_type.name, "void")) {
4231 fprintf(fout, "return ");
4232 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
4233 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
4234 }
2b43685d 4235 }
75ad0378 4236 else if (regmask & (1 << xAX)) {
87bf6cec 4237 fprintf(fout, "eax = ");
2b43685d 4238 if (pp->ret_type.is_ptr)
4239 fprintf(fout, "(u32)");
4240 }
91977a1c 4241 }
87bf6cec 4242
ddaf8bd7 4243 if (pp->name[0] == 0)
4244 ferr(po, "missing pp->name\n");
4245 fprintf(fout, "%s%s(", pp->name,
4246 pp->has_structarg ? "_sa" : "");
39b168b8 4247
2fe80fdb 4248 if (po->flags & OPF_ATAIL) {
4249 if (pp->argc_stack != g_func_pp->argc_stack
4250 || (pp->argc_stack > 0
4251 && pp->is_stdcall != g_func_pp->is_stdcall))
4252 ferr(po, "incompatible tailcall\n");
87bf6cec 4253
2fe80fdb 4254 for (arg = j = 0; arg < pp->argc; arg++) {
4255 if (arg > 0)
4256 fprintf(fout, ", ");
87bf6cec 4257
2fe80fdb 4258 cast[0] = 0;
4259 if (pp->arg[arg].type.is_ptr)
4260 snprintf(cast, sizeof(cast), "(%s)",
4261 pp->arg[arg].type.name);
91977a1c 4262
2fe80fdb 4263 if (pp->arg[arg].reg != NULL) {
4264 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
4265 continue;
4266 }
4267 // stack arg
4268 for (; j < g_func_pp->argc; j++)
4269 if (g_func_pp->arg[j].reg == NULL)
4270 break;
4271 fprintf(fout, "%sa%d", cast, j + 1);
4272 j++;
69a3cdfc 4273 }
2fe80fdb 4274 }
4275 else {
4276 for (arg = 0; arg < pp->argc; arg++) {
4277 if (arg > 0)
4278 fprintf(fout, ", ");
4279
4280 cast[0] = 0;
4281 if (pp->arg[arg].type.is_ptr)
4282 snprintf(cast, sizeof(cast), "(%s)",
4283 pp->arg[arg].type.name);
4284
4285 if (pp->arg[arg].reg != NULL) {
4286 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
4287 continue;
4288 }
4289
4290 // stack arg
4291 tmp_op = pp->arg[arg].datap;
4292 if (tmp_op == NULL)
4293 ferr(po, "parsed_op missing for arg%d\n", arg);
4294 if (tmp_op->argnum != 0) {
4295 fprintf(fout, "%ss_a%d", cast, tmp_op->argnum);
4296 }
4297 else {
4298 fprintf(fout, "%s",
4299 out_src_opr(buf1, sizeof(buf1),
4300 tmp_op, &tmp_op->operand[0], cast, 0));
4301 }
69a3cdfc 4302 }
91977a1c 4303 }
4304 fprintf(fout, ");");
87bf6cec 4305
2b43685d 4306 if (strstr(pp->ret_type.name, "int64")) {
4307 fprintf(fout, "\n");
2fe80fdb 4308 fprintf(fout, " edx = tmp64 >> 32;\n");
4309 fprintf(fout, " eax = tmp64;");
2b43685d 4310 }
4311
89ff3147 4312 if (pp->is_unresolved) {
ddaf8bd7 4313 snprintf(buf3, sizeof(buf3), " unresoved %dreg",
89ff3147 4314 pp->argc_reg);
4315 strcat(g_comment, buf3);
4316 }
4317
87bf6cec 4318 if (po->flags & OPF_TAIL) {
840257f6 4319 ret = 0;
ddaf8bd7 4320 if (i == opcnt - 1 || pp->is_noreturn)
840257f6 4321 ret = 0;
4322 else if (IS(pp->ret_type.name, "void"))
4323 ret = 1;
4324 else if (IS(g_func_pp->ret_type.name, "void"))
4325 ret = 1;
4326 // else already handled as 'return f()'
4327
4328 if (ret) {
da87ae38 4329 if (!IS(g_func_pp->ret_type.name, "void")) {
ddaf8bd7 4330 ferr(po, "int func -> void func tailcall?\n");
da87ae38 4331 }
4332 else {
4333 fprintf(fout, "\n return;");
ddaf8bd7 4334 strcat(g_comment, " ^ tailcall");
da87ae38 4335 }
3ebea2cf 4336 }
89ff3147 4337 else
ddaf8bd7 4338 strcat(g_comment, " tailcall");
87bf6cec 4339 }
ddaf8bd7 4340 if (pp->is_noreturn)
4341 strcat(g_comment, " noreturn");
2fe80fdb 4342 if ((po->flags & OPF_ATAIL) && pp->argc_stack > 0)
4343 strcat(g_comment, " argframe");
87bf6cec 4344 delayed_flag_op = NULL;
4345 last_arith_dst = NULL;
91977a1c 4346 break;
4347
4348 case OP_RET:
bd96f656 4349 if (g_func_pp->is_vararg)
4f12f671 4350 fprintf(fout, " va_end(ap);\n");
4351
bd96f656 4352 if (IS(g_func_pp->ret_type.name, "void")) {
3ebea2cf 4353 if (i != opcnt - 1 || label_pending)
4354 fprintf(fout, " return;");
4355 }
bd96f656 4356 else if (g_func_pp->ret_type.is_ptr) {
d4e3b5db 4357 fprintf(fout, " return (%s)eax;",
bd96f656 4358 g_func_pp->ret_type.name);
3ebea2cf 4359 }
2fe80fdb 4360 else if (IS(g_func_pp->ret_type.name, "__int64"))
4361 fprintf(fout, " return ((u64)edx << 32) | eax;");
91977a1c 4362 else
4363 fprintf(fout, " return eax;");
de50b98b 4364
4365 last_arith_dst = NULL;
4366 delayed_flag_op = NULL;
91977a1c 4367 break;
4368
4369 case OP_PUSH:
de50b98b 4370 if (po->argnum != 0) {
69a3cdfc 4371 // special case - saved func arg
3ebea2cf 4372 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
de50b98b 4373 fprintf(fout, " s_a%d = %s;", po->argnum, buf1);
69a3cdfc 4374 break;
4375 }
d4e3b5db 4376 else if (po->flags & OPF_RSAVE) {
3ebea2cf 4377 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
d4e3b5db 4378 fprintf(fout, " s_%s = %s;", buf1, buf1);
4379 break;
4380 }
e56ab892 4381 if (!(g_ida_func_attr & IDAFA_NORETURN))
4382 ferr(po, "stray push encountered\n");
4383 no_output = 1;
91977a1c 4384 break;
4385
4386 case OP_POP:
d4e3b5db 4387 if (po->flags & OPF_RSAVE) {
3ebea2cf 4388 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
d4e3b5db 4389 fprintf(fout, " %s = s_%s;", buf1, buf1);
4390 break;
4391 }
5c024ef7 4392 else if (po->datap != NULL) {
4393 // push/pop pair
4394 tmp_op = po->datap;
4395 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4396 fprintf(fout, " %s = %s;", buf1,
4397 out_src_opr(buf2, sizeof(buf2),
4398 tmp_op, &tmp_op->operand[0],
4399 po->operand[0].is_ptr ? "(void *)" : "", 0));
4400 break;
4401 }
d4e3b5db 4402 ferr(po, "stray pop encountered\n");
91977a1c 4403 break;
4404
33c35af6 4405 case OP_NOP:
2b43685d 4406 no_output = 1;
33c35af6 4407 break;
4408
91977a1c 4409 default:
4410 no_output = 1;
69a3cdfc 4411 ferr(po, "unhandled op type %d, flags %x\n",
4412 po->op, po->flags);
91977a1c 4413 break;
4414 }
4415
4416 if (g_comment[0] != 0) {
ddaf8bd7 4417 char *p = g_comment;
4418 while (my_isblank(*p))
4419 p++;
4420 fprintf(fout, " // %s", p);
91977a1c 4421 g_comment[0] = 0;
4422 no_output = 0;
4423 }
4424 if (!no_output)
4425 fprintf(fout, "\n");
5101a5f9 4426
2b43685d 4427 // some sanity checking
591721d7 4428 if (po->flags & OPF_REP) {
4429 if (po->op != OP_STOS && po->op != OP_MOVS
4430 && po->op != OP_CMPS && po->op != OP_SCAS)
4431 ferr(po, "unexpected rep\n");
4432 if (!(po->flags & (OPF_REPZ|OPF_REPNZ))
4433 && (po->op == OP_CMPS || po->op == OP_SCAS))
4434 ferr(po, "cmps/scas with plain rep\n");
4435 }
4436 if ((po->flags & (OPF_REPZ|OPF_REPNZ))
4437 && po->op != OP_CMPS && po->op != OP_SCAS)
2b43685d 4438 ferr(po, "unexpected repz/repnz\n");
4439
940e8e66 4440 if (pfomask != 0)
7ba45c34 4441 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
940e8e66 4442
5101a5f9 4443 // see is delayed flag stuff is still valid
4444 if (delayed_flag_op != NULL && delayed_flag_op != po) {
89ff3147 4445 if (is_any_opr_modified(delayed_flag_op, po, 0))
5101a5f9 4446 delayed_flag_op = NULL;
4447 }
4448
4449 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
4450 if (is_opr_modified(last_arith_dst, po))
4451 last_arith_dst = NULL;
4452 }
3ebea2cf 4453
4454 label_pending = 0;
91977a1c 4455 }
4456
a2c1d768 4457 if (g_stack_fsz && !g_stack_frame_used)
4458 fprintf(fout, " (void)sf;\n");
4459
91977a1c 4460 fprintf(fout, "}\n\n");
4461
4462 // cleanup
4463 for (i = 0; i < opcnt; i++) {
4c45fa73 4464 struct label_ref *lr, *lr_del;
4465
4466 lr = g_label_refs[i].next;
4467 while (lr != NULL) {
4468 lr_del = lr;
4469 lr = lr->next;
4470 free(lr_del);
4471 }
4472 g_label_refs[i].i = -1;
4473 g_label_refs[i].next = NULL;
4474
91977a1c 4475 if (ops[i].op == OP_CALL) {
4476 pp = ops[i].datap;
bd96f656 4477 if (pp)
91977a1c 4478 proto_release(pp);
91977a1c 4479 }
4480 }
bd96f656 4481 g_func_pp = NULL;
91977a1c 4482}
c36e914d 4483
d4e3b5db 4484static void set_label(int i, const char *name)
4485{
4486 const char *p;
4487 int len;
4488
4489 len = strlen(name);
4490 p = strchr(name, ':');
4491 if (p != NULL)
4492 len = p - name;
4493
4494 if (len > sizeof(g_labels[0]) - 1)
4495 aerr("label '%s' too long: %d\n", name, len);
e56ab892 4496 if (g_labels[i][0] != 0 && !IS_START(g_labels[i], "algn_"))
4497 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
d4e3b5db 4498 memcpy(g_labels[i], name, len);
4499 g_labels[i][len] = 0;
4500}
4501
bfa4a6ee 4502// '=' needs special treatment..
4503static char *next_word_s(char *w, size_t wsize, char *s)
4504{
4505 size_t i;
4506
4507 s = sskip(s);
4508
4509 for (i = 0; i < wsize - 1; i++) {
4510 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
4511 break;
4512 w[i] = s[i];
4513 }
4514 w[i] = 0;
4515
4516 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
4517 printf("warning: '%s' truncated\n", w);
4518
4519 return s + i;
4520}
4521
e56ab892 4522struct chunk_item {
4523 char *name;
4524 long fptr;
de50b98b 4525 int asmln;
e56ab892 4526};
4527
cdfaeed7 4528static struct chunk_item *func_chunks;
4529static int func_chunk_cnt;
4530static int func_chunk_alloc;
4531
4532static void add_func_chunk(FILE *fasm, const char *name, int line)
4533{
4534 if (func_chunk_cnt >= func_chunk_alloc) {
4535 func_chunk_alloc *= 2;
4536 func_chunks = realloc(func_chunks,
4537 func_chunk_alloc * sizeof(func_chunks[0]));
4538 my_assert_not(func_chunks, NULL);
4539 }
4540 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
4541 func_chunks[func_chunk_cnt].name = strdup(name);
4542 func_chunks[func_chunk_cnt].asmln = line;
4543 func_chunk_cnt++;
4544}
4545
e56ab892 4546static int cmp_chunks(const void *p1, const void *p2)
4547{
4548 const struct chunk_item *c1 = p1, *c2 = p2;
4549 return strcmp(c1->name, c2->name);
4550}
4551
bfa4a6ee 4552static int cmpstringp(const void *p1, const void *p2)
4553{
4554 return strcmp(*(char * const *)p1, *(char * const *)p2);
4555}
4556
cdfaeed7 4557static void scan_ahead(FILE *fasm)
4558{
4559 char words[2][256];
4560 char line[256];
4561 long oldpos;
4562 int oldasmln;
4563 int wordc;
4564 char *p;
4565 int i;
4566
4567 oldpos = ftell(fasm);
4568 oldasmln = asmln;
4569
4570 while (fgets(line, sizeof(line), fasm))
4571 {
4572 wordc = 0;
4573 asmln++;
4574
4575 p = sskip(line);
4576 if (*p == 0)
4577 continue;
4578
4579 if (*p == ';')
4580 {
4581 // get rid of random tabs
4582 for (i = 0; line[i] != 0; i++)
4583 if (line[i] == '\t')
4584 line[i] = ' ';
4585
4586 if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
4587 {
4588 p += 30;
4589 next_word(words[0], sizeof(words[0]), p);
4590 if (words[0][0] == 0)
4591 aerr("missing name for func chunk?\n");
4592
4593 add_func_chunk(fasm, words[0], asmln);
4594 }
4595 continue;
4596 } // *p == ';'
4597
4598 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
4599 words[wordc][0] = 0;
4600 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
4601 if (*p == 0 || *p == ';') {
4602 wordc++;
4603 break;
4604 }
4605 }
4606
4607 if (wordc == 2 && IS(words[1], "ends"))
4608 break;
4609 }
4610
4611 fseek(fasm, oldpos, SEEK_SET);
4612 asmln = oldasmln;
4613}
4614
91977a1c 4615int main(int argc, char *argv[])
4616{
06c5d854 4617 FILE *fout, *fasm, *frlist;
4c45fa73 4618 struct parsed_data *pd = NULL;
4619 int pd_alloc = 0;
4620 char **rlist = NULL;
4621 int rlist_len = 0;
4622 int rlist_alloc = 0;
e56ab892 4623 int func_chunks_used = 0;
4624 int func_chunks_sorted = 0;
e56ab892 4625 int func_chunk_i = -1;
4626 long func_chunk_ret = 0;
de50b98b 4627 int func_chunk_ret_ln = 0;
cdfaeed7 4628 int scanned_ahead = 0;
91977a1c 4629 char line[256];
a2c1d768 4630 char words[20][256];
4c45fa73 4631 enum opr_lenmod lmod;
ddaf8bd7 4632 char *sctproto = NULL;
91977a1c 4633 int in_func = 0;
4c45fa73 4634 int pending_endp = 0;
bfa4a6ee 4635 int skip_func = 0;
940e8e66 4636 int skip_warned = 0;
91977a1c 4637 int eq_alloc;
bfa4a6ee 4638 int verbose = 0;
4639 int arg_out;
89ff3147 4640 int arg;
91977a1c 4641 int pi = 0;
e56ab892 4642 int i, j;
4643 int ret, len;
91977a1c 4644 char *p;
4645 int wordc;
4646
89ff3147 4647 for (arg = 1; arg < argc; arg++) {
4648 if (IS(argv[arg], "-v"))
4649 verbose = 1;
4650 else if (IS(argv[arg], "-rf"))
4651 g_allow_regfunc = 1;
4652 else
4653 break;
bfa4a6ee 4654 }
4655
4656 if (argc < arg + 3) {
89ff3147 4657 printf("usage:\n%s [-v] [-rf] <.c> <.asm> <hdrf> [rlist]*\n",
91977a1c 4658 argv[0]);
4659 return 1;
4660 }
4661
bfa4a6ee 4662 arg_out = arg++;
91977a1c 4663
bfa4a6ee 4664 asmfn = argv[arg++];
91977a1c 4665 fasm = fopen(asmfn, "r");
4666 my_assert_not(fasm, NULL);
4667
bfa4a6ee 4668 hdrfn = argv[arg++];
06c5d854 4669 g_fhdr = fopen(hdrfn, "r");
4670 my_assert_not(g_fhdr, NULL);
bfa4a6ee 4671
4672 rlist_alloc = 64;
4673 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
4674 my_assert_not(rlist, NULL);
4675 // needs special handling..
4676 rlist[rlist_len++] = "__alloca_probe";
4677
e56ab892 4678 func_chunk_alloc = 32;
4679 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
4680 my_assert_not(func_chunks, NULL);
4681
a2c1d768 4682 memset(words, 0, sizeof(words));
4683
bfa4a6ee 4684 for (; arg < argc; arg++) {
4685 frlist = fopen(argv[arg], "r");
4686 my_assert_not(frlist, NULL);
4687
4688 while (fgets(line, sizeof(line), frlist)) {
4689 p = sskip(line);
1cd4a663 4690 if (*p == 0 || *p == ';')
4691 continue;
4692 if (*p == '#') {
89ff3147 4693 if (IS_START(p, "#if 0")
4694 || (g_allow_regfunc && IS_START(p, "#if NO_REGFUNC")))
4695 {
1cd4a663 4696 skip_func = 1;
89ff3147 4697 }
1cd4a663 4698 else if (IS_START(p, "#endif"))
4699 skip_func = 0;
4700 continue;
4701 }
4702 if (skip_func)
bfa4a6ee 4703 continue;
4704
4705 p = next_word(words[0], sizeof(words[0]), p);
4706 if (words[0][0] == 0)
4707 continue;
4708
4709 if (rlist_len >= rlist_alloc) {
4710 rlist_alloc = rlist_alloc * 2 + 64;
4711 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
4712 my_assert_not(rlist, NULL);
4713 }
4714 rlist[rlist_len++] = strdup(words[0]);
4715 }
1cd4a663 4716 skip_func = 0;
bfa4a6ee 4717
4718 fclose(frlist);
4719 frlist = NULL;
4720 }
4721
4722 if (rlist_len > 0)
4723 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
4724
4725 fout = fopen(argv[arg_out], "w");
91977a1c 4726 my_assert_not(fout, NULL);
4727
4728 eq_alloc = 128;
4729 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
4730 my_assert_not(g_eqs, NULL);
4731
4c45fa73 4732 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
4733 g_label_refs[i].i = -1;
4734 g_label_refs[i].next = NULL;
4735 }
4736
91977a1c 4737 while (fgets(line, sizeof(line), fasm))
4738 {
4c45fa73 4739 wordc = 0;
91977a1c 4740 asmln++;
4741
4742 p = sskip(line);
1bafb621 4743 if (*p == 0)
91977a1c 4744 continue;
4745
de50b98b 4746 // get rid of random tabs
4747 for (i = 0; line[i] != 0; i++)
4748 if (line[i] == '\t')
4749 line[i] = ' ';
4750
e56ab892 4751 if (*p == ';')
4752 {
e56ab892 4753 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
4754 goto do_pending_endp; // eww..
4755
4756 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
4757 {
4758 static const char *attrs[] = {
4759 "bp-based frame",
4760 "library function",
4761 "static",
4762 "noreturn",
4763 "thunk",
4764 "fpd=",
4765 };
4766
4767 // parse IDA's attribute-list comment
4768 g_ida_func_attr = 0;
4769 p = sskip(p + 13);
4770
4771 for (; *p != 0; p = sskip(p)) {
4772 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
4773 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
4774 g_ida_func_attr |= 1 << i;
4775 p += strlen(attrs[i]);
4776 break;
4777 }
4778 }
4779 if (i == ARRAY_SIZE(attrs)) {
4780 anote("unparsed IDA attr: %s\n", p);
1bafb621 4781 break;
4782 }
e56ab892 4783 if (IS(attrs[i], "fpd=")) {
4784 p = next_word(words[0], sizeof(words[0]), p);
4785 // ignore for now..
4786 }
1bafb621 4787 }
e56ab892 4788 }
4789 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
4790 {
4791 p += 30;
4792 next_word(words[0], sizeof(words[0]), p);
4793 if (words[0][0] == 0)
cdfaeed7 4794 aerr("missing name for func chunk?\n");
4795
4796 if (!scanned_ahead) {
4797 add_func_chunk(fasm, words[0], asmln);
4798 func_chunks_sorted = 0;
e56ab892 4799 }
e56ab892 4800 }
4801 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
4802 {
4803 if (func_chunk_i >= 0) {
4804 if (func_chunk_i < func_chunk_cnt
4805 && IS(func_chunks[func_chunk_i].name, g_func))
4806 {
4807 // move on to next chunk
4808 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
4809 if (ret)
4810 aerr("seek failed for '%s' chunk #%d\n",
4811 g_func, func_chunk_i);
de50b98b 4812 asmln = func_chunks[func_chunk_i].asmln;
e56ab892 4813 func_chunk_i++;
4814 }
4815 else {
4816 if (func_chunk_ret == 0)
4817 aerr("no return from chunk?\n");
4818 fseek(fasm, func_chunk_ret, SEEK_SET);
de50b98b 4819 asmln = func_chunk_ret_ln;
e56ab892 4820 func_chunk_ret = 0;
4821 pending_endp = 1;
4822 }
1bafb621 4823 }
e56ab892 4824 }
4825 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
4826 func_chunks_used = 1;
4827 p += 20;
4828 if (IS_START(g_func, "sub_")) {
4829 unsigned long addr = strtoul(p, NULL, 16);
4830 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
cdfaeed7 4831 if (addr > f_addr && !scanned_ahead) {
94d447fb 4832 anote("scan_ahead caused by '%s', addr %lx\n",
4833 g_func, addr);
cdfaeed7 4834 scan_ahead(fasm);
4835 scanned_ahead = 1;
4836 func_chunks_sorted = 0;
4837 }
1bafb621 4838 }
4839 }
4840 continue;
e56ab892 4841 } // *p == ';'
1bafb621 4842
06c5d854 4843parse_words:
a2c1d768 4844 for (i = wordc; i < ARRAY_SIZE(words); i++)
4845 words[i][0] = 0;
cdfaeed7 4846 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
bfa4a6ee 4847 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
91977a1c 4848 if (*p == 0 || *p == ';') {
4849 wordc++;
4850 break;
4851 }
4852 }
a2c1d768 4853 if (*p != 0 && *p != ';')
4854 aerr("too many words\n");
91977a1c 4855
06c5d854 4856 // alow asm patches in comments
ddaf8bd7 4857 if (*p == ';') {
4858 if (IS_START(p, "; sctpatch:")) {
4859 p = sskip(p + 11);
4860 if (*p == 0 || *p == ';')
4861 continue;
4862 goto parse_words; // lame
4863 }
4864 if (IS_START(p, "; sctproto:")) {
4865 sctproto = strdup(p + 11);
4866 }
06c5d854 4867 }
4868
91977a1c 4869 if (wordc == 0) {
4870 // shouldn't happen
4871 awarn("wordc == 0?\n");
4872 continue;
4873 }
4874
4875 // don't care about this:
4876 if (words[0][0] == '.'
4877 || IS(words[0], "include")
4878 || IS(words[0], "assume") || IS(words[1], "segment")
4879 || IS(words[0], "align"))
4880 {
4881 continue;
4882 }
4883
4c45fa73 4884do_pending_endp:
4885 // do delayed endp processing to collect switch jumptables
4886 if (pending_endp) {
4887 if (in_func && !skip_func && wordc >= 2
4888 && ((words[0][0] == 'd' && words[0][2] == 0)
4889 || (words[1][0] == 'd' && words[1][2] == 0)))
4890 {
4891 i = 1;
4892 if (words[1][0] == 'd' && words[1][2] == 0) {
4893 // label
4894 if (g_func_pd_cnt >= pd_alloc) {
4895 pd_alloc = pd_alloc * 2 + 16;
4896 g_func_pd = realloc(g_func_pd,
4897 sizeof(g_func_pd[0]) * pd_alloc);
4898 my_assert_not(g_func_pd, NULL);
4899 }
4900 pd = &g_func_pd[g_func_pd_cnt];
4901 g_func_pd_cnt++;
4902 memset(pd, 0, sizeof(*pd));
4903 strcpy(pd->label, words[0]);
4904 pd->type = OPT_CONST;
4905 pd->lmod = lmod_from_directive(words[1]);
4906 i = 2;
4907 }
4908 else {
da87ae38 4909 if (pd == NULL) {
4910 if (verbose)
4911 anote("skipping alignment byte?\n");
4912 continue;
4913 }
4c45fa73 4914 lmod = lmod_from_directive(words[0]);
4915 if (lmod != pd->lmod)
4916 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
4917 }
4918
4919 if (pd->count_alloc < pd->count + wordc) {
4920 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
4921 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
4922 my_assert_not(pd->d, NULL);
4923 }
4924 for (; i < wordc; i++) {
4925 if (IS(words[i], "offset")) {
4926 pd->type = OPT_OFFSET;
4927 i++;
4928 }
4929 p = strchr(words[i], ',');
4930 if (p != NULL)
4931 *p = 0;
4932 if (pd->type == OPT_OFFSET)
4933 pd->d[pd->count].u.label = strdup(words[i]);
4934 else
4935 pd->d[pd->count].u.val = parse_number(words[i]);
4936 pd->d[pd->count].bt_i = -1;
4937 pd->count++;
4938 }
4939 continue;
4940 }
4941
4942 if (in_func && !skip_func)
4943 gen_func(fout, g_fhdr, g_func, pi);
4944
4945 pending_endp = 0;
4946 in_func = 0;
4947 g_ida_func_attr = 0;
4948 skip_warned = 0;
4949 skip_func = 0;
4950 g_func[0] = 0;
e56ab892 4951 func_chunks_used = 0;
4952 func_chunk_i = -1;
4c45fa73 4953 if (pi != 0) {
4954 memset(&ops, 0, pi * sizeof(ops[0]));
4955 memset(g_labels, 0, pi * sizeof(g_labels[0]));
4956 pi = 0;
4957 }
4958 g_eqcnt = 0;
4959 for (i = 0; i < g_func_pd_cnt; i++) {
4960 pd = &g_func_pd[i];
4961 if (pd->type == OPT_OFFSET) {
4962 for (j = 0; j < pd->count; j++)
4963 free(pd->d[j].u.label);
4964 }
4965 free(pd->d);
4966 pd->d = NULL;
4967 }
4968 g_func_pd_cnt = 0;
4969 pd = NULL;
4970 if (wordc == 0)
4971 continue;
4972 }
4973
91977a1c 4974 if (IS(words[1], "proc")) {
4975 if (in_func)
4976 aerr("proc '%s' while in_func '%s'?\n",
4977 words[0], g_func);
bfa4a6ee 4978 p = words[0];
ddaf8bd7 4979 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
bfa4a6ee 4980 skip_func = 1;
91977a1c 4981 strcpy(g_func, words[0]);
d4e3b5db 4982 set_label(0, words[0]);
91977a1c 4983 in_func = 1;
4984 continue;
4985 }
4986
e56ab892 4987 if (IS(words[1], "endp"))
4988 {
91977a1c 4989 if (!in_func)
4990 aerr("endp '%s' while not in_func?\n", words[0]);
4991 if (!IS(g_func, words[0]))
4992 aerr("endp '%s' while in_func '%s'?\n",
4993 words[0], g_func);
bfa4a6ee 4994
ddaf8bd7 4995 if ((g_ida_func_attr & IDAFA_THUNK) && pi == 1
4996 && ops[0].op == OP_JMP && ops[0].operand[0].had_ds)
4997 {
4998 // import jump
4999 skip_func = 1;
5000 }
5001
e56ab892 5002 if (!skip_func && func_chunks_used) {
5003 // start processing chunks
5004 struct chunk_item *ci, key = { g_func, 0 };
5005
5006 func_chunk_ret = ftell(fasm);
de50b98b 5007 func_chunk_ret_ln = asmln;
e56ab892 5008 if (!func_chunks_sorted) {
5009 qsort(func_chunks, func_chunk_cnt,
5010 sizeof(func_chunks[0]), cmp_chunks);
5011 func_chunks_sorted = 1;
5012 }
5013 ci = bsearch(&key, func_chunks, func_chunk_cnt,
5014 sizeof(func_chunks[0]), cmp_chunks);
5015 if (ci == NULL)
5016 aerr("'%s' needs chunks, but none found\n", g_func);
5017 func_chunk_i = ci - func_chunks;
5018 for (; func_chunk_i > 0; func_chunk_i--)
5019 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
5020 break;
5021
5022 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
5023 if (ret)
5024 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
de50b98b 5025 asmln = func_chunks[func_chunk_i].asmln;
e56ab892 5026 func_chunk_i++;
5027 continue;
5028 }
4c45fa73 5029 pending_endp = 1;
91977a1c 5030 continue;
5031 }
5032
a2c1d768 5033 if (wordc == 2 && IS(words[1], "ends"))
5034 break;
5035
bfa4a6ee 5036 p = strchr(words[0], ':');
5037 if (p != NULL) {
d4e3b5db 5038 set_label(pi, words[0]);
bfa4a6ee 5039 continue;
5040 }
5041
5042 if (!in_func || skip_func) {
5043 if (!skip_warned && !skip_func && g_labels[pi][0] != 0) {
5044 if (verbose)
5045 anote("skipping from '%s'\n", g_labels[pi]);
5046 skip_warned = 1;
5047 }
5048 g_labels[pi][0] = 0;
5049 continue;
5050 }
5051
ddaf8bd7 5052 if (wordc > 1 && IS(words[1], "="))
5053 {
91977a1c 5054 if (wordc != 5)
5055 aerr("unhandled equ, wc=%d\n", wordc);
5056 if (g_eqcnt >= eq_alloc) {
5057 eq_alloc *= 2;
5058 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
5059 my_assert_not(g_eqs, NULL);
5060 }
5061
5062 len = strlen(words[0]);
5063 if (len > sizeof(g_eqs[0].name) - 1)
5064 aerr("equ name too long: %d\n", len);
5065 strcpy(g_eqs[g_eqcnt].name, words[0]);
5066
5067 if (!IS(words[3], "ptr"))
5068 aerr("unhandled equ\n");
5069 if (IS(words[2], "dword"))
5070 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
5071 else if (IS(words[2], "word"))
5072 g_eqs[g_eqcnt].lmod = OPLM_WORD;
5073 else if (IS(words[2], "byte"))
5074 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
5075 else
5076 aerr("bad lmod: '%s'\n", words[2]);
5077
5078 g_eqs[g_eqcnt].offset = parse_number(words[4]);
5079 g_eqcnt++;
5080 continue;
5081 }
5082
5083 if (pi >= ARRAY_SIZE(ops))
5084 aerr("too many ops\n");
5085
91977a1c 5086 parse_op(&ops[pi], words, wordc);
ddaf8bd7 5087
5088 if (sctproto != NULL) {
5089 if (ops[pi].op == OP_CALL)
5090 ops[pi].datap = sctproto;
5091 sctproto = NULL;
5092 }
91977a1c 5093 pi++;
91977a1c 5094 }
5095
5096 fclose(fout);
5097 fclose(fasm);
06c5d854 5098 fclose(g_fhdr);
91977a1c 5099
5100 return 0;
c36e914d 5101}
91977a1c 5102
5103// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.