notaz.gp2x.de / ia32rtools.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
more stuff storm needs
[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 }
2860 fprintf(fout, ")");
2861 }
2862 else {
2863 fprintf(fout, "%s a%d", g_func_pp->arg[i].type.name, i + 1);
2864 }
89ff3147 2865 if (g_func_pp->arg[i].reg != NULL) {
2866 reg = char_array_i(regs_r32,
2867 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
2868 if (reg < 0)
2869 ferr(ops, "arg '%s' is not a reg?\n", g_func_pp->arg[i].reg);
2870 regmask_arg |= 1 << reg;
2871 }
91977a1c 2872 }
bd96f656 2873 if (g_func_pp->is_vararg) {
4f12f671 2874 if (i > 0)
2875 fprintf(fout, ", ");
2876 fprintf(fout, "...");
2877 }
54e763a1 2878
91977a1c 2879 fprintf(fout, ")\n{\n");
2880
2881 // pass1:
1bafb621 2882 // - handle ebp/esp frame, remove ops related to it
91977a1c 2883 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
2884 && ops[1].op == OP_MOV
2885 && IS(opr_name(&ops[1], 0), "ebp")
2886 && IS(opr_name(&ops[1], 1), "esp"))
2887 {
87bf6cec 2888 int ecx_push = 0;
2889
91977a1c 2890 g_bp_frame = 1;
69a3cdfc 2891 ops[0].flags |= OPF_RMD;
2892 ops[1].flags |= OPF_RMD;
4c45fa73 2893 i = 2;
91977a1c 2894
2895 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
1bafb621 2896 g_stack_fsz = opr_const(&ops[2], 1);
69a3cdfc 2897 ops[2].flags |= OPF_RMD;
4c45fa73 2898 i++;
91977a1c 2899 }
87bf6cec 2900 else {
2901 // another way msvc builds stack frame..
2902 i = 2;
2903 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
1bafb621 2904 g_stack_fsz += 4;
87bf6cec 2905 ops[i].flags |= OPF_RMD;
2906 ecx_push++;
2907 i++;
2908 }
4c45fa73 2909 // and another way..
2910 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
2911 && ops[i].operand[1].type == OPT_CONST
2912 && ops[i + 1].op == OP_CALL
2913 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
2914 {
2915 g_stack_fsz += ops[i].operand[1].val;
2916 ops[i].flags |= OPF_RMD;
2917 i++;
2918 ops[i].flags |= OPF_RMD;
2919 i++;
2920 }
87bf6cec 2921 }
91977a1c 2922
64c59faf 2923 found = 0;
91977a1c 2924 do {
2925 for (; i < opcnt; i++)
2926 if (ops[i].op == OP_RET)
2927 break;
2fe80fdb 2928 j = i - 1;
2929 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
2930 if (found) {
2931 if (ops[j].op == OP_JMP && !ops[j].operand[0].had_ds)
2932 // probably local branch back..
2933 break;
2934 if (ops[j].op == OP_CALL)
2935 // probably noreturn call..
2936 break;
2937 }
2938 j--;
2939 }
64c59faf 2940
2fe80fdb 2941 if ((ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ebp"))
2942 || ops[j].op == OP_LEAVE)
591721d7 2943 {
2fe80fdb 2944 ops[j].flags |= OPF_RMD;
591721d7 2945 }
e56ab892 2946 else if (!(g_ida_func_attr & IDAFA_NORETURN))
2fe80fdb 2947 ferr(&ops[j], "'pop ebp' expected\n");
91977a1c 2948
1bafb621 2949 if (g_stack_fsz != 0) {
2fe80fdb 2950 if (ops[j - 1].op == OP_MOV
2951 && IS(opr_name(&ops[j - 1], 0), "esp")
2952 && IS(opr_name(&ops[j - 1], 1), "ebp"))
91977a1c 2953 {
2fe80fdb 2954 ops[j - 1].flags |= OPF_RMD;
91977a1c 2955 }
2fe80fdb 2956 else if (ops[j].op != OP_LEAVE
591721d7 2957 && !(g_ida_func_attr & IDAFA_NORETURN))
2958 {
2fe80fdb 2959 ferr(&ops[j - 1], "esp restore expected\n");
591721d7 2960 }
87bf6cec 2961
2fe80fdb 2962 if (ecx_push && ops[j - 2].op == OP_POP
2963 && IS(opr_name(&ops[j - 2], 0), "ecx"))
87bf6cec 2964 {
2fe80fdb 2965 ferr(&ops[j - 2], "unexpected ecx pop\n");
87bf6cec 2966 }
91977a1c 2967 }
87bf6cec 2968
64c59faf 2969 found = 1;
91977a1c 2970 i++;
2971 } while (i < opcnt);
2972 }
1bafb621 2973 else {
2974 for (i = 0; i < opcnt; i++) {
2975 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
2976 break;
2977 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
2978 && ops[i].operand[1].type == OPT_CONST)
2979 {
2980 g_sp_frame = 1;
2981 break;
2982 }
2983 }
2984
2985 if (g_sp_frame)
2986 {
2987 g_stack_fsz = ops[i].operand[1].val;
2988 ops[i].flags |= OPF_RMD;
2989
2990 i++;
2991 do {
2992 for (; i < opcnt; i++)
2993 if (ops[i].op == OP_RET)
2994 break;
2995 if (ops[i - 1].op != OP_ADD
2996 || !IS(opr_name(&ops[i - 1], 0), "esp")
2997 || ops[i - 1].operand[1].type != OPT_CONST
2998 || ops[i - 1].operand[1].val != g_stack_fsz)
2999 ferr(&ops[i - 1], "'add esp' expected\n");
3000 ops[i - 1].flags |= OPF_RMD;
3001
3002 i++;
3003 } while (i < opcnt);
3004 }
3005 }
91977a1c 3006
3007 // pass2:
7ae48d73 3008 // - parse calls with labels
87bf6cec 3009 // - resolve all branches
de50b98b 3010 for (i = 0; i < opcnt; i++)
3011 {
87bf6cec 3012 po = &ops[i];
3013 po->bt_i = -1;
4c45fa73 3014 po->btj = NULL;
87bf6cec 3015
7ae48d73 3016 if (po->flags & OPF_RMD)
3017 continue;
3018
3019 if (po->op == OP_CALL) {
ddaf8bd7 3020 pp = NULL;
3021
7ae48d73 3022 if (po->operand[0].type == OPT_LABEL) {
3023 tmpname = opr_name(po, 0);
46411e6c 3024 if (IS_START(tmpname, "loc_"))
3025 ferr(po, "call to loc_*\n");
36595fd2 3026 pp_c = proto_parse(fhdr, tmpname, 0);
7ae48d73 3027 if (pp_c == NULL)
3028 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
da87ae38 3029
7ae48d73 3030 pp = proto_clone(pp_c);
3031 my_assert_not(pp, NULL);
ddaf8bd7 3032 }
3033 else if (po->datap != NULL) {
3034 pp = calloc(1, sizeof(*pp));
3035 my_assert_not(pp, NULL);
3036
3037 ret = parse_protostr(po->datap, pp);
3038 if (ret < 0)
3039 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
3040 free(po->datap);
3041 po->datap = NULL;
3042 }
3043
3044 if (pp != NULL) {
3045 if (pp->is_fptr)
3046 check_func_pp(po, pp, "fptr var call");
3047 if (pp->is_noreturn)
3048 po->flags |= OPF_TAIL;
7ae48d73 3049 po->datap = pp;
3050 }
3051 continue;
3052 }
3053
3054 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
87bf6cec 3055 continue;
3056
4c45fa73 3057 if (po->operand[0].type == OPT_REGMEM) {
3058 char *p = strchr(po->operand[0].name, '[');
3059 if (p == NULL)
89ff3147 3060 goto tailcall;
4c45fa73 3061 ret = p - po->operand[0].name;
3062 strncpy(buf1, po->operand[0].name, ret);
3063 buf1[ret] = 0;
3064
3065 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
3066 if (IS(g_func_pd[j].label, buf1)) {
3067 pd = &g_func_pd[j];
3068 break;
3069 }
3070 }
3071 if (pd == NULL)
840257f6 3072 //ferr(po, "label '%s' not parsed?\n", buf1);
3073 goto tailcall;
4c45fa73 3074 if (pd->type != OPT_OFFSET)
3075 ferr(po, "label '%s' with non-offset data?\n", buf1);
3076
3077 // find all labels, link
3078 for (j = 0; j < pd->count; j++) {
3079 for (l = 0; l < opcnt; l++) {
3080 if (g_labels[l][0] && IS(g_labels[l], pd->d[j].u.label)) {
3081 add_label_ref(&g_label_refs[l], i);
3082 pd->d[j].bt_i = l;
3083 break;
3084 }
3085 }
3086 }
3087
3088 po->btj = pd;
3089 continue;
3090 }
3091
3092 for (l = 0; l < opcnt; l++) {
3093 if (g_labels[l][0] && IS(po->operand[0].name, g_labels[l])) {
3094 add_label_ref(&g_label_refs[l], i);
3095 po->bt_i = l;
87bf6cec 3096 break;
3097 }
3098 }
3099
4c45fa73 3100 if (po->bt_i != -1)
3101 continue;
3102
840257f6 3103 if (po->operand[0].type == OPT_LABEL)
87bf6cec 3104 // assume tail call
840257f6 3105 goto tailcall;
4c45fa73 3106
3107 ferr(po, "unhandled branch\n");
840257f6 3108
3109tailcall:
3110 po->op = OP_CALL;
3111 po->flags |= OPF_TAIL;
2fe80fdb 3112 if (i > 0 && ops[i - 1].op == OP_POP)
3113 po->flags |= OPF_ATAIL;
7ae48d73 3114 i--; // reprocess
87bf6cec 3115 }
3116
3117 // pass3:
a2c1d768 3118 // - remove dead labels
91977a1c 3119 // - process calls
1bafb621 3120 for (i = 0; i < opcnt; i++)
3121 {
a2c1d768 3122 if (g_labels[i][0] != 0 && g_label_refs[i].i == -1)
3123 g_labels[i][0] = 0;
3124
69a3cdfc 3125 po = &ops[i];
3126 if (po->flags & OPF_RMD)
91977a1c 3127 continue;
850c9265 3128
d4e3b5db 3129 if (po->op == OP_CALL)
69a3cdfc 3130 {
1bafb621 3131 tmpname = opr_name(po, 0);
7ae48d73 3132 pp = po->datap;
3133 if (pp == NULL)
1bafb621 3134 {
7ae48d73 3135 // indirect call
89ff3147 3136 pp_c = resolve_icall(i, opcnt, &l);
3137 if (pp_c != NULL) {
1cd4a663 3138 pp = proto_clone(pp_c);
89ff3147 3139 my_assert_not(pp, NULL);
3140 if (l)
3141 // not resolved just to single func
3142 pp->is_fptr = 1;
3143 }
1cd4a663 3144 if (pp == NULL) {
3145 pp = calloc(1, sizeof(*pp));
3146 my_assert_not(pp, NULL);
89ff3147 3147 pp->is_fptr = 1;
4741fdfe 3148 ret = scan_for_esp_adjust(i + 1, opcnt, &j, &l);
89ff3147 3149 if (ret < 0) {
3150 if (!g_allow_regfunc)
3151 ferr(po, "non-__cdecl indirect call unhandled yet\n");
3152 pp->is_unresolved = 1;
3153 j = 0;
3154 }
1cd4a663 3155 j /= 4;
3156 if (j > ARRAY_SIZE(pp->arg))
89ff3147 3157 ferr(po, "esp adjust too large: %d\n", j);
1cd4a663 3158 pp->ret_type.name = strdup("int");
3159 pp->argc = pp->argc_stack = j;
3160 for (arg = 0; arg < pp->argc; arg++)
3161 pp->arg[arg].type.name = strdup("int");
3162 }
7ae48d73 3163 po->datap = pp;
1bafb621 3164 }
3165
7ba45c34 3166 // look for and make use of esp adjust
3167 ret = -1;
e56ab892 3168 if (!pp->is_stdcall && pp->argc_stack > 0)
4741fdfe 3169 ret = scan_for_esp_adjust(i + 1, opcnt, &j, &l);
1bafb621 3170 if (ret >= 0) {
7ba45c34 3171 if (pp->is_vararg) {
3172 if (j / 4 < pp->argc_stack)
3173 ferr(po, "esp adjust is too small: %x < %x\n",
3174 j, pp->argc_stack * 4);
3175 // modify pp to make it have varargs as normal args
3176 arg = pp->argc;
3177 pp->argc += j / 4 - pp->argc_stack;
3178 for (; arg < pp->argc; arg++) {
3ebea2cf 3179 pp->arg[arg].type.name = strdup("int");
7ba45c34 3180 pp->argc_stack++;
3181 }
3182 if (pp->argc > ARRAY_SIZE(pp->arg))
e56ab892 3183 ferr(po, "too many args for '%s'\n", tmpname);
7ba45c34 3184 }
1bafb621 3185 if (pp->argc_stack != j / 4)
e56ab892 3186 ferr(po, "stack tracking failed for '%s': %x %x\n",
3187 tmpname, pp->argc_stack * 4, j);
7ba45c34 3188
1bafb621 3189 ops[ret].flags |= OPF_RMD;
591721d7 3190 if (ops[ret].op == OP_POP && j > 4) {
3191 // deal with multi-pop stack adjust
3192 j = pp->argc_stack;
3193 while (ops[ret].op == OP_POP && j > 0 && ret < opcnt) {
3194 ops[ret].flags |= OPF_RMD;
3195 j--;
3196 ret++;
3197 }
3198 }
4741fdfe 3199 else if (!l) {
591721d7 3200 // a bit of a hack, but deals with use of
3201 // single adj for multiple calls
3202 ops[ret].operand[1].val -= j;
3203 }
1bafb621 3204 }
7ba45c34 3205 else if (pp->is_vararg)
3206 ferr(po, "missing esp_adjust for vararg func '%s'\n",
3207 pp->name);
91977a1c 3208
2fe80fdb 3209 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
89ff3147 3210 // since we know the args, collect them
3211 collect_call_args(po, i, pp, &regmask, &save_arg_vars,
3212 i + opcnt * 2);
3213 }
2b43685d 3214
3215 if (strstr(pp->ret_type.name, "int64"))
2fe80fdb 3216 need_tmp64 = 1;
91977a1c 3217 }
d4e3b5db 3218 }
3219
3220 // pass4:
3221 // - find POPs for PUSHes, rm both
591721d7 3222 // - scan for STD/CLD, propagate DF
d4e3b5db 3223 // - scan for all used registers
3224 // - find flag set ops for their users
89ff3147 3225 // - do unreselved calls
1bafb621 3226 // - declare indirect functions
d4e3b5db 3227 for (i = 0; i < opcnt; i++) {
3228 po = &ops[i];
3229 if (po->flags & OPF_RMD)
3230 continue;
3231
7ae48d73 3232 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE)) {
3233 reg = po->operand[0].reg;
3234 if (!(regmask & (1 << reg)))
3235 // not a reg save after all, rerun scan_for_pop
3236 po->flags &= ~OPF_RSAVE;
3237 else
3238 regmask_save |= 1 << reg;
3239 }
3240
5c024ef7 3241 if (po->op == OP_PUSH && po->argnum == 0
3242 && !(po->flags & OPF_RSAVE))
d4e3b5db 3243 {
5c024ef7 3244 if (po->operand[0].type == OPT_REG)
3245 {
3246 reg = po->operand[0].reg;
3247 if (reg < 0)
3248 ferr(po, "reg not set for push?\n");
3249
3250 depth = 0;
3251 ret = scan_for_pop(i + 1, opcnt,
3252 po->operand[0].name, i + opcnt * 3, 0, &depth, 0);
3253 if (ret == 1) {
3254 if (depth > 1)
3255 ferr(po, "too much depth: %d\n", depth);
3256
3257 po->flags |= OPF_RMD;
3258 scan_for_pop(i + 1, opcnt, po->operand[0].name,
3259 i + opcnt * 4, 0, &depth, 1);
3260 continue;
3261 }
3262 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
3263 if (ret == 0) {
3264 arg = OPF_RMD;
3265 if (regmask & (1 << reg)) {
3266 if (regmask_save & (1 << reg))
3267 ferr(po, "%s already saved?\n", po->operand[0].name);
3268 arg = OPF_RSAVE;
3269 }
3270 po->flags |= arg;
3271 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg);
3272 continue;
3273 }
d4e3b5db 3274 }
5c024ef7 3275 else if (po->operand[0].type == OPT_CONST) {
3276 for (j = i + 1; j < opcnt; j++) {
3277 if ((ops[j].flags & (OPF_JMP|OPF_TAIL|OPF_RSAVE))
3278 || ops[j].op == OP_PUSH || g_labels[i][0] != 0)
3279 {
3280 break;
3281 }
3282
3283 if (!(ops[j].flags & OPF_RMD) && ops[j].op == OP_POP)
3284 {
3285 po->flags |= OPF_RMD;
3286 ops[j].datap = po;
3287 break;
3288 }
d4e3b5db 3289 }
d4e3b5db 3290 }
3291 }
3292
591721d7 3293 if (po->op == OP_STD) {
3294 po->flags |= OPF_DF | OPF_RMD;
3295 scan_propagate_df(i + 1, opcnt);
3296 }
3297
a3684be1 3298 regmask_now = po->regmask_src | po->regmask_dst;
3299 if (regmask_now & (1 << xBP)) {
3300 if (g_bp_frame && !(po->flags & OPF_EBP_S)) {
3301 if (po->regmask_dst & (1 << xBP))
3302 // compiler decided to drop bp frame and use ebp as scratch
3303 scan_fwd_set_flags(i, opcnt, i + opcnt * 5, OPF_EBP_S);
3304 else
3305 regmask_now &= ~(1 << xBP);
3306 }
3307 }
3308
3309 regmask |= regmask_now;
d4e3b5db 3310
3311 if (po->flags & OPF_CC)
3312 {
2b43685d 3313 int setters[16], cnt = 0, branched = 0;
3314
04f8a628 3315 ret = scan_for_flag_set(i, i + opcnt * 6,
3316 &branched, setters, &cnt);
2b43685d 3317 if (ret < 0 || cnt <= 0)
3318 ferr(po, "unable to trace flag setter(s)\n");
3319 if (cnt > ARRAY_SIZE(setters))
3320 ferr(po, "too many flag setters\n");
d4e3b5db 3321
d4e3b5db 3322 pfo = split_cond(po, po->op, &dummy);
2b43685d 3323 for (j = 0; j < cnt; j++)
3324 {
3325 tmp_op = &ops[setters[j]]; // flag setter
3326 pfomask = 0;
3327
3328 // to get nicer code, we try to delay test and cmp;
3329 // if we can't because of operand modification, or if we
591721d7 3330 // have arith op, or branch, make it calculate flags explicitly
3331 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP)
3332 {
89ff3147 3333 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i, 0) >= 0)
2b43685d 3334 pfomask = 1 << pfo;
3335 }
4741fdfe 3336 else if (tmp_op->op == OP_CMPS || tmp_op->op == OP_SCAS) {
591721d7 3337 pfomask = 1 << pfo;
3338 }
2b43685d 3339 else {
04f8a628 3340 // see if we'll be able to handle based on op result
3341 if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR
3342 && pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P)
3343 || branched
2b43685d 3344 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
3345 pfomask = 1 << pfo;
2fe80fdb 3346
3347 if (tmp_op->op == OP_ADD && pfo == PFO_C)
3348 need_tmp64 = 1;
2b43685d 3349 }
3350 if (pfomask) {
3351 tmp_op->pfomask |= pfomask;
cb090db0 3352 cond_vars |= pfomask;
2b43685d 3353 }
04f8a628 3354 // note: may overwrite, currently not a problem
3355 po->datap = tmp_op;
d4e3b5db 3356 }
3357
cb090db0 3358 if (po->op == OP_RCL || po->op == OP_RCR
3359 || po->op == OP_ADC || po->op == OP_SBB)
3360 cond_vars |= 1 << PFO_C;
d4e3b5db 3361 }
591721d7 3362 else if (po->op == OP_CMPS || po->op == OP_SCAS) {
cb090db0 3363 cond_vars |= 1 << PFO_Z;
591721d7 3364 }
87bf6cec 3365 else if (po->op == OP_MUL
3366 || (po->op == OP_IMUL && po->operand_cnt == 1))
3367 {
2fe80fdb 3368 need_tmp64 = 1;
87bf6cec 3369 }
89ff3147 3370 else if (po->op == OP_CALL) {
1bafb621 3371 pp = po->datap;
89ff3147 3372 if (pp == NULL)
3373 ferr(po, "NULL pp\n");
3374
3375 if (pp->is_unresolved) {
ddaf8bd7 3376 int regmask_stack = 0;
b74c31e3 3377 collect_call_args(po, i, pp, &regmask, &save_arg_vars,
89ff3147 3378 i + opcnt * 2);
3379
b74c31e3 3380 // this is pretty rough guess:
3381 // see ecx and edx were pushed (and not their saved versions)
3382 for (arg = 0; arg < pp->argc; arg++) {
3383 if (pp->arg[arg].reg != NULL)
3384 continue;
3385
3386 tmp_op = pp->arg[arg].datap;
3387 if (tmp_op == NULL)
3388 ferr(po, "parsed_op missing for arg%d\n", arg);
3389 if (tmp_op->argnum == 0 && tmp_op->operand[0].type == OPT_REG)
3390 regmask_stack |= 1 << tmp_op->operand[0].reg;
3391 }
3392
ddaf8bd7 3393 if (!((regmask_stack & (1 << xCX))
3394 && (regmask_stack & (1 << xDX))))
89ff3147 3395 {
3396 if (pp->argc_stack != 0
c0050df6 3397 || ((regmask | regmask_arg) & ((1 << xCX)|(1 << xDX))))
89ff3147 3398 {
3399 pp_insert_reg_arg(pp, "ecx");
c0050df6 3400 pp->is_fastcall = 1;
ddaf8bd7 3401 regmask_init |= 1 << xCX;
89ff3147 3402 regmask |= 1 << xCX;
3403 }
3404 if (pp->argc_stack != 0
3405 || ((regmask | regmask_arg) & (1 << xDX)))
3406 {
3407 pp_insert_reg_arg(pp, "edx");
ddaf8bd7 3408 regmask_init |= 1 << xDX;
89ff3147 3409 regmask |= 1 << xDX;
3410 }
3411 }
c0050df6 3412
3413 // note: __cdecl doesn't fall into is_unresolved category
3414 if (pp->argc_stack > 0)
3415 pp->is_stdcall = 1;
ddaf8bd7 3416 }
3417
3418 for (arg = 0; arg < pp->argc; arg++) {
3419 if (pp->arg[arg].reg != NULL) {
3420 reg = char_array_i(regs_r32,
3421 ARRAY_SIZE(regs_r32), pp->arg[arg].reg);
3422 if (reg < 0)
3423 ferr(ops, "arg '%s' is not a reg?\n", pp->arg[arg].reg);
3424 if (!(regmask & (1 << reg))) {
3425 regmask_init |= 1 << reg;
3426 regmask |= 1 << reg;
3427 }
3428 }
89ff3147 3429 }
3430
3431 if (pp->is_fptr) {
ddaf8bd7 3432 if (pp->name[0] != 0) {
3433 memmove(pp->name + 2, pp->name, strlen(pp->name) + 1);
3434 memcpy(pp->name, "i_", 2);
3435
3436 // might be declared already
3437 found = 0;
3438 for (j = 0; j < i; j++) {
3439 if (ops[j].op == OP_CALL && (pp_tmp = ops[j].datap)) {
3440 if (pp_tmp->is_fptr && IS(pp->name, pp_tmp->name)) {
3441 found = 1;
3442 break;
3443 }
3444 }
3445 }
3446 if (found)
3447 continue;
3448 }
3449 else
3450 snprintf(pp->name, sizeof(pp->name), "icall%d", i);
3451
89ff3147 3452 fprintf(fout, " %s (", pp->ret_type.name);
c0050df6 3453 output_pp_attrs(fout, pp, 0);
ddaf8bd7 3454 fprintf(fout, "*%s)(", pp->name);
89ff3147 3455 for (j = 0; j < pp->argc; j++) {
3456 if (j > 0)
3457 fprintf(fout, ", ");
3458 fprintf(fout, "%s a%d", pp->arg[j].type.name, j + 1);
3459 }
3460 fprintf(fout, ");\n");
1bafb621 3461 }
1bafb621 3462 }
75ad0378 3463 else if (po->op == OP_RET && !IS(g_func_pp->ret_type.name, "void"))
3464 regmask |= 1 << xAX;
cb090db0 3465 else if (po->op == OP_DIV || po->op == OP_IDIV) {
3466 // 32bit division is common, look for it
3467 if (po->op == OP_DIV)
3468 ret = scan_for_reg_clear(i, xDX);
3469 else
3470 ret = scan_for_cdq_edx(i);
3471 if (ret >= 0)
3472 po->flags |= OPF_32BIT;
3473 else
3474 need_tmp64 = 1;
3475 }
3476
3477 if (po->op == OP_RCL || po->op == OP_RCR || po->op == OP_XCHG) {
3478 need_tmp_var = 1;
3479 }
91977a1c 3480 }
3481
da87ae38 3482 // pass4:
3483 // - confirm regmask_save, it might have been reduced
3484 if (regmask_save != 0)
3485 {
3486 regmask_save = 0;
3487 for (i = 0; i < opcnt; i++) {
3488 po = &ops[i];
3489 if (po->flags & OPF_RMD)
3490 continue;
3491
3492 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE))
3493 regmask_save |= 1 << po->operand[0].reg;
3494 }
3495 }
3496
3497
4c45fa73 3498 // output LUTs/jumptables
3499 for (i = 0; i < g_func_pd_cnt; i++) {
3500 pd = &g_func_pd[i];
3501 fprintf(fout, " static const ");
3502 if (pd->type == OPT_OFFSET) {
3503 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
3504
3505 for (j = 0; j < pd->count; j++) {
3506 if (j > 0)
3507 fprintf(fout, ", ");
3508 fprintf(fout, "&&%s", pd->d[j].u.label);
3509 }
3510 }
3511 else {
3512 fprintf(fout, "%s %s[] =\n { ",
3513 lmod_type_u(ops, pd->lmod), pd->label);
3514
3515 for (j = 0; j < pd->count; j++) {
3516 if (j > 0)
3517 fprintf(fout, ", ");
3518 fprintf(fout, "%u", pd->d[j].u.val);
3519 }
3520 }
3521 fprintf(fout, " };\n");
3522 }
3523
4f12f671 3524 // declare stack frame, va_arg
1bafb621 3525 if (g_stack_fsz)
850c9265 3526 fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
1bafb621 3527 (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz);
850c9265 3528
bd96f656 3529 if (g_func_pp->is_vararg)
4f12f671 3530 fprintf(fout, " va_list ap;\n");
3531
940e8e66 3532 // declare arg-registers
bd96f656 3533 for (i = 0; i < g_func_pp->argc; i++) {
3534 if (g_func_pp->arg[i].reg != NULL) {
91977a1c 3535 reg = char_array_i(regs_r32,
bd96f656 3536 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
75ad0378 3537 if (regmask & (1 << reg)) {
3538 fprintf(fout, " u32 %s = (u32)a%d;\n",
3539 g_func_pp->arg[i].reg, i + 1);
3540 }
3541 else
3542 fprintf(fout, " // %s = a%d; // unused\n",
3543 g_func_pp->arg[i].reg, i + 1);
91977a1c 3544 had_decl = 1;
3545 }
3546 }
3547
75ad0378 3548 regmask_now = regmask & ~regmask_arg;
3549 regmask_now &= ~(1 << xSP);
3550 if (regmask_now) {
91977a1c 3551 for (reg = 0; reg < 8; reg++) {
75ad0378 3552 if (regmask_now & (1 << reg)) {
ddaf8bd7 3553 fprintf(fout, " u32 %s", regs_r32[reg]);
3554 if (regmask_init & (1 << reg))
3555 fprintf(fout, " = 0");
3556 fprintf(fout, ";\n");
91977a1c 3557 had_decl = 1;
3558 }
3559 }
3560 }
3561
d4e3b5db 3562 if (regmask_save) {
3563 for (reg = 0; reg < 8; reg++) {
3564 if (regmask_save & (1 << reg)) {
3565 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
3566 had_decl = 1;
3567 }
3568 }
3569 }
3570
69a3cdfc 3571 if (save_arg_vars) {
3572 for (reg = 0; reg < 32; reg++) {
3573 if (save_arg_vars & (1 << reg)) {
3574 fprintf(fout, " u32 s_a%d;\n", reg + 1);
3575 had_decl = 1;
3576 }
3577 }
3578 }
3579
cb090db0 3580 if (cond_vars) {
69a3cdfc 3581 for (i = 0; i < 8; i++) {
cb090db0 3582 if (cond_vars & (1 << i)) {
69a3cdfc 3583 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
3584 had_decl = 1;
3585 }
3586 }
3587 }
3588
108e9fe3 3589 if (need_tmp_var) {
3590 fprintf(fout, " u32 tmp;\n");
3591 had_decl = 1;
3592 }
3593
2fe80fdb 3594 if (need_tmp64) {
3595 fprintf(fout, " u64 tmp64;\n");
87bf6cec 3596 had_decl = 1;
3597 }
3598
91977a1c 3599 if (had_decl)
3600 fprintf(fout, "\n");
3601
bd96f656 3602 if (g_func_pp->is_vararg) {
3603 if (g_func_pp->argc_stack == 0)
4f12f671 3604 ferr(ops, "vararg func without stack args?\n");
bd96f656 3605 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
4f12f671 3606 }
3607
91977a1c 3608 // output ops
69a3cdfc 3609 for (i = 0; i < opcnt; i++)
3610 {
a2c1d768 3611 if (g_labels[i][0] != 0) {
91977a1c 3612 fprintf(fout, "\n%s:\n", g_labels[i]);
3ebea2cf 3613 label_pending = 1;
2b43685d 3614
3615 delayed_flag_op = NULL;
3616 last_arith_dst = NULL;
3ebea2cf 3617 }
91977a1c 3618
69a3cdfc 3619 po = &ops[i];
3620 if (po->flags & OPF_RMD)
91977a1c 3621 continue;
3622
3623 no_output = 0;
3624
91977a1c 3625 #define assert_operand_cnt(n_) \
850c9265 3626 if (po->operand_cnt != n_) \
3627 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
3628
69a3cdfc 3629 // conditional/flag using op?
3630 if (po->flags & OPF_CC)
850c9265 3631 {
940e8e66 3632 int is_delayed = 0;
3633 int is_inv = 0;
69a3cdfc 3634
940e8e66 3635 pfo = split_cond(po, po->op, &is_inv);
04f8a628 3636 tmp_op = po->datap;
850c9265 3637
69a3cdfc 3638 // we go through all this trouble to avoid using parsed_flag_op,
3639 // which makes generated code much nicer
3640 if (delayed_flag_op != NULL)
850c9265 3641 {
940e8e66 3642 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op, pfo, is_inv);
3643 is_delayed = 1;
91977a1c 3644 }
850c9265 3645 else if (last_arith_dst != NULL
04f8a628 3646 && (pfo == PFO_Z || pfo == PFO_S || pfo == PFO_P
3647 || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR))
3648 ))
850c9265 3649 {
3ebea2cf 3650 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
940e8e66 3651 out_test_for_cc(buf1, sizeof(buf1), po, pfo, is_inv,
850c9265 3652 last_arith_dst->lmod, buf3);
940e8e66 3653 is_delayed = 1;
850c9265 3654 }
04f8a628 3655 else if (tmp_op != NULL) {
7ba45c34 3656 // use preprocessed flag calc results
04f8a628 3657 if (!(tmp_op->pfomask & (1 << pfo)))
69a3cdfc 3658 ferr(po, "not prepared for pfo %d\n", pfo);
3659
940e8e66 3660 // note: is_inv was not yet applied
69a3cdfc 3661 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
940e8e66 3662 is_inv ? "!" : "", parsed_flag_op_names[pfo]);
69a3cdfc 3663 }
3664 else {
3665 ferr(po, "all methods of finding comparison failed\n");
3666 }
850c9265 3667
69a3cdfc 3668 if (po->flags & OPF_JMP) {
850c9265 3669 fprintf(fout, " if %s\n", buf1);
3670 }
cb090db0 3671 else if (po->op == OP_RCL || po->op == OP_RCR
3672 || po->op == OP_ADC || po->op == OP_SBB)
3673 {
940e8e66 3674 if (is_delayed)
3675 fprintf(fout, " cond_%s = %s;\n",
3676 parsed_flag_op_names[pfo], buf1);
850c9265 3677 }
5101a5f9 3678 else if (po->flags & OPF_DATA) { // SETcc
850c9265 3679 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
3680 fprintf(fout, " %s = %s;", buf2, buf1);
91977a1c 3681 }
69a3cdfc 3682 else {
3683 ferr(po, "unhandled conditional op\n");
3684 }
91977a1c 3685 }
3686
940e8e66 3687 pfomask = po->pfomask;
3688
4741fdfe 3689 if (po->flags & (OPF_REPZ|OPF_REPNZ)) {
3690 // we need initial flags for ecx=0 case..
3691 if (i > 0 && ops[i - 1].op == OP_XOR
3692 && IS(ops[i - 1].operand[0].name,
3693 ops[i - 1].operand[1].name))
3694 {
3695 fprintf(fout, " cond_z = ");
3696 if (pfomask & (1 << PFO_C))
3697 fprintf(fout, "cond_c = ");
3698 fprintf(fout, "0;\n");
3699 }
3700 else if (last_arith_dst != NULL) {
3701 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
3702 out_test_for_cc(buf1, sizeof(buf1), po, PFO_Z, 0,
3703 last_arith_dst->lmod, buf3);
3704 fprintf(fout, " cond_z = %s;\n", buf1);
3705 }
3706 else
3707 ferr(po, "missing initial ZF\n");
3708 }
3709
850c9265 3710 switch (po->op)
91977a1c 3711 {
3712 case OP_MOV:
3713 assert_operand_cnt(2);
850c9265 3714 propagate_lmod(po, &po->operand[0], &po->operand[1]);
de50b98b 3715 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3716 fprintf(fout, " %s = %s;", buf1,
3ebea2cf 3717 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
3718 po->operand[0].is_ptr ? "(void *)" : "", 0));
850c9265 3719 break;
3720
3721 case OP_LEA:
3722 assert_operand_cnt(2);
87bf6cec 3723 po->operand[1].lmod = OPLM_DWORD; // always
850c9265 3724 fprintf(fout, " %s = %s;",
3725 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3ebea2cf 3726 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
3727 NULL, 1));
850c9265 3728 break;
3729
3730 case OP_MOVZX:
3731 assert_operand_cnt(2);
91977a1c 3732 fprintf(fout, " %s = %s;",
850c9265 3733 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3ebea2cf 3734 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
850c9265 3735 break;
3736
3737 case OP_MOVSX:
3738 assert_operand_cnt(2);
3739 switch (po->operand[1].lmod) {
3740 case OPLM_BYTE:
3741 strcpy(buf3, "(s8)");
3742 break;
3743 case OPLM_WORD:
3744 strcpy(buf3, "(s16)");
3745 break;
3746 default:
3747 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
3748 }
a2c1d768 3749 fprintf(fout, " %s = %s;",
850c9265 3750 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
a2c1d768 3751 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
3752 buf3, 0));
850c9265 3753 break;
3754
108e9fe3 3755 case OP_XCHG:
3756 assert_operand_cnt(2);
3757 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3758 fprintf(fout, " tmp = %s;",
3759 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], "", 0));
3760 fprintf(fout, " %s = %s;",
3761 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3762 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], "", 0));
3763 fprintf(fout, " %s = tmp;",
3764 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]));
3765 snprintf(g_comment, sizeof(g_comment), "xchg");
3766 break;
3767
850c9265 3768 case OP_NOT:
3769 assert_operand_cnt(1);
3770 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3771 fprintf(fout, " %s = ~%s;", buf1, buf1);
3772 break;
3773
5101a5f9 3774 case OP_CDQ:
3775 assert_operand_cnt(2);
3776 fprintf(fout, " %s = (s32)%s >> 31;",
3777 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3ebea2cf 3778 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5101a5f9 3779 strcpy(g_comment, "cdq");
3780 break;
3781
33c35af6 3782 case OP_STOS:
33c35af6 3783 assert_operand_cnt(3);
3784 if (po->flags & OPF_REP) {
591721d7 3785 fprintf(fout, " for (; ecx != 0; ecx--, edi %c= %d)\n",
3786 (po->flags & OPF_DF) ? '-' : '+',
33c35af6 3787 lmod_bytes(po, po->operand[0].lmod));
d4e3b5db 3788 fprintf(fout, " %sedi = eax;",
3789 lmod_cast_u_ptr(po, po->operand[0].lmod));
33c35af6 3790 strcpy(g_comment, "rep stos");
3791 }
3792 else {
591721d7 3793 fprintf(fout, " %sedi = eax; edi %c= %d;",
d4e3b5db 3794 lmod_cast_u_ptr(po, po->operand[0].lmod),
591721d7 3795 (po->flags & OPF_DF) ? '-' : '+',
33c35af6 3796 lmod_bytes(po, po->operand[0].lmod));
3797 strcpy(g_comment, "stos");
3798 }
3799 break;
3800
d4e3b5db 3801 case OP_MOVS:
d4e3b5db 3802 assert_operand_cnt(3);
3803 j = lmod_bytes(po, po->operand[0].lmod);
3804 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
591721d7 3805 l = (po->flags & OPF_DF) ? '-' : '+';
d4e3b5db 3806 if (po->flags & OPF_REP) {
3807 fprintf(fout,
591721d7 3808 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n",
3809 l, j, l, j);
d4e3b5db 3810 fprintf(fout,
3811 " %sedi = %sesi;", buf1, buf1);
3812 strcpy(g_comment, "rep movs");
3813 }
3814 else {
591721d7 3815 fprintf(fout, " %sedi = %sesi; edi %c= %d; esi %c= %d;",
3816 buf1, buf1, l, j, l, j);
d4e3b5db 3817 strcpy(g_comment, "movs");
3818 }
3819 break;
3820
7ba45c34 3821 case OP_CMPS:
7ba45c34 3822 // repe ~ repeat while ZF=1
3823 assert_operand_cnt(3);
3824 j = lmod_bytes(po, po->operand[0].lmod);
3825 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
591721d7 3826 l = (po->flags & OPF_DF) ? '-' : '+';
7ba45c34 3827 if (po->flags & OPF_REP) {
3828 fprintf(fout,
4741fdfe 3829 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d) {\n",
591721d7 3830 l, j, l, j);
4741fdfe 3831 if (pfomask & (1 << PFO_C)) {
3832 // ugh..
3833 fprintf(fout,
3834 " cond_c = %sedi < %sesi;\n", buf1, buf1);
3835 pfomask &= ~(1 << PFO_C);
3836 }
7ba45c34 3837 fprintf(fout,
3838 " if ((cond_z = (%sedi == %sesi)) %s 0)\n",
3839 buf1, buf1, (po->flags & OPF_REPZ) ? "==" : "!=");
3840 fprintf(fout,
4741fdfe 3841 " break;\n"
3842 " }");
7ba45c34 3843 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
3844 (po->flags & OPF_REPZ) ? "e" : "ne");
3845 }
3846 else {
3847 fprintf(fout,
591721d7 3848 " cond_z = (%sedi = %sesi); edi %c= %d; esi %c= %d;",
3849 buf1, buf1, l, j, l, j);
7ba45c34 3850 strcpy(g_comment, "cmps");
3851 }
3852 pfomask &= ~(1 << PFO_Z);
3853 last_arith_dst = NULL;
3854 delayed_flag_op = NULL;
3855 break;
3856
591721d7 3857 case OP_SCAS:
3858 // only does ZF (for now)
3859 // repe ~ repeat while ZF=1
3860 assert_operand_cnt(3);
3861 j = lmod_bytes(po, po->operand[0].lmod);
3862 l = (po->flags & OPF_DF) ? '-' : '+';
3863 if (po->flags & OPF_REP) {
3864 fprintf(fout,
3865 " for (; ecx != 0; ecx--, edi %c= %d)\n", l, j);
3866 fprintf(fout,
3867 " if ((cond_z = (%seax == %sedi)) %s 0)\n",
3868 lmod_cast_u(po, po->operand[0].lmod),
3869 lmod_cast_u_ptr(po, po->operand[0].lmod),
3870 (po->flags & OPF_REPZ) ? "==" : "!=");
3871 fprintf(fout,
3872 " break;");
3873 snprintf(g_comment, sizeof(g_comment), "rep%s scas",
3874 (po->flags & OPF_REPZ) ? "e" : "ne");
3875 }
3876 else {
3877 fprintf(fout, " cond_z = (%seax = %sedi); edi %c= %d;",
3878 lmod_cast_u(po, po->operand[0].lmod),
3879 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
3880 strcpy(g_comment, "scas");
3881 }
3882 pfomask &= ~(1 << PFO_Z);
3883 last_arith_dst = NULL;
3884 delayed_flag_op = NULL;
3885 break;
3886
850c9265 3887 // arithmetic w/flags
850c9265 3888 case OP_AND:
3889 case OP_OR:
5101a5f9 3890 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3891 // fallthrough
850c9265 3892 dualop_arith:
3893 assert_operand_cnt(2);
850c9265 3894 fprintf(fout, " %s %s= %s;",
3895 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3896 op_to_c(po),
3ebea2cf 3897 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
04f8a628 3898 output_std_flags(fout, po, &pfomask, buf1);
3899 last_arith_dst = &po->operand[0];
3900 delayed_flag_op = NULL;
3901 break;
3902
3903 case OP_SHL:
3904 case OP_SHR:
3905 assert_operand_cnt(2);
3906 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3907 if (pfomask & (1 << PFO_C)) {
3908 if (po->operand[1].type == OPT_CONST) {
3909 l = lmod_bytes(po, po->operand[0].lmod) * 8;
3910 j = po->operand[1].val;
3911 j %= l;
3912 if (j != 0) {
3913 if (po->op == OP_SHL)
3914 j = l - j;
3915 else
3916 j -= 1;
cb090db0 3917 fprintf(fout, " cond_c = (%s >> %d) & 1;\n",
3918 buf1, j);
04f8a628 3919 }
3920 else
3921 ferr(po, "zero shift?\n");
3922 }
3923 else
3924 ferr(po, "TODO\n");
3925 pfomask &= ~(1 << PFO_C);
840257f6 3926 }
04f8a628 3927 fprintf(fout, " %s %s= %s;", buf1, op_to_c(po),
3928 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
3929 output_std_flags(fout, po, &pfomask, buf1);
850c9265 3930 last_arith_dst = &po->operand[0];
69a3cdfc 3931 delayed_flag_op = NULL;
850c9265 3932 break;
3933
d4e3b5db 3934 case OP_SAR:
3935 assert_operand_cnt(2);
3936 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3937 fprintf(fout, " %s = %s%s >> %s;", buf1,
3938 lmod_cast_s(po, po->operand[0].lmod), buf1,
3ebea2cf 3939 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
04f8a628 3940 output_std_flags(fout, po, &pfomask, buf1);
d4e3b5db 3941 last_arith_dst = &po->operand[0];
3942 delayed_flag_op = NULL;
3943 break;
3944
3945 case OP_ROL:
3946 case OP_ROR:
3947 assert_operand_cnt(2);
3948 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3949 if (po->operand[1].type == OPT_CONST) {
3950 j = po->operand[1].val;
3951 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
3952 fprintf(fout, po->op == OP_ROL ?
3953 " %s = (%s << %d) | (%s >> %d);" :
3954 " %s = (%s >> %d) | (%s << %d);",
3955 buf1, buf1, j, buf1,
3956 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
3957 }
3958 else
3959 ferr(po, "TODO\n");
04f8a628 3960 output_std_flags(fout, po, &pfomask, buf1);
d4e3b5db 3961 last_arith_dst = &po->operand[0];
3962 delayed_flag_op = NULL;
3963 break;
3964
cb090db0 3965 case OP_RCL:
3966 case OP_RCR:
3967 assert_operand_cnt(2);
3968 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3969 l = lmod_bytes(po, po->operand[0].lmod) * 8;
3970 if (po->operand[1].type == OPT_CONST) {
3971 j = po->operand[1].val % l;
3972 if (j == 0)
3973 ferr(po, "zero rotate\n");
3974 fprintf(fout, " tmp = (%s >> %d) & 1;\n",
3975 buf1, (po->op == OP_RCL) ? (l - j) : (j - 1));
3976 if (po->op == OP_RCL) {
3977 fprintf(fout,
3978 " %s = (%s << %d) | (cond_c << %d)",
3979 buf1, buf1, j, j - 1);
3980 if (j != 1)
3981 fprintf(fout, " | (%s >> %d)", buf1, l + 1 - j);
3982 }
3983 else {
3984 fprintf(fout,
3985 " %s = (%s >> %d) | (cond_c << %d)",
3986 buf1, buf1, j, l - j);
3987 if (j != 1)
3988 fprintf(fout, " | (%s << %d)", buf1, l + 1 - j);
3989 }
3990 fprintf(fout, ";\n");
3991 fprintf(fout, " cond_c = tmp;");
3992 }
3993 else
3994 ferr(po, "TODO\n");
3995 strcpy(g_comment, (po->op == OP_RCL) ? "rcl" : "rcr");
3996 output_std_flags(fout, po, &pfomask, buf1);
3997 last_arith_dst = &po->operand[0];
3998 delayed_flag_op = NULL;
3999 break;
4000
5101a5f9 4001 case OP_XOR:
850c9265 4002 assert_operand_cnt(2);
4003 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5101a5f9 4004 if (IS(opr_name(po, 0), opr_name(po, 1))) {
4005 // special case for XOR
2fe80fdb 4006 if (pfomask & (1 << PFO_BE)) { // weird, but it happens..
4007 fprintf(fout, " cond_be = 1;\n");
4008 pfomask &= ~(1 << PFO_BE);
4009 }
5101a5f9 4010 fprintf(fout, " %s = 0;",
4011 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4012 last_arith_dst = &po->operand[0];
4013 delayed_flag_op = NULL;
850c9265 4014 break;
850c9265 4015 }
5101a5f9 4016 goto dualop_arith;
4017
2fe80fdb 4018 case OP_ADD:
4019 assert_operand_cnt(2);
4020 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4021 if (pfomask & (1 << PFO_C)) {
4022 fprintf(fout, " tmp64 = (u64)%s + %s;\n",
4023 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]),
4024 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4025 fprintf(fout, " cond_c = tmp64 >> 32;\n");
4026 fprintf(fout, " %s = (u32)tmp64;",
4027 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4028 strcat(g_comment, "add64");
4029 pfomask &= ~(1 << PFO_C);
4030 output_std_flags(fout, po, &pfomask, buf1);
4031 last_arith_dst = &po->operand[0];
4032 delayed_flag_op = NULL;
4033 break;
4034 }
4035 goto dualop_arith;
4036
4037 case OP_SUB:
4038 assert_operand_cnt(2);
4039 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4040 if (pfomask & (1 << PFO_C)) {
4041 fprintf(fout, " cond_c = %s < %s;\n",
4042 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]),
4043 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4044 pfomask &= ~(1 << PFO_C);
4045 }
4046 goto dualop_arith;
4047
5101a5f9 4048 case OP_ADC:
850c9265 4049 case OP_SBB:
5101a5f9 4050 assert_operand_cnt(2);
4051 propagate_lmod(po, &po->operand[0], &po->operand[1]);
a2c1d768 4052 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
840257f6 4053 if (po->op == OP_SBB
4054 && IS(po->operand[0].name, po->operand[1].name))
4055 {
4056 // avoid use of unitialized var
a2c1d768 4057 fprintf(fout, " %s = -cond_c;", buf1);
94d447fb 4058 // carry remains what it was
4059 pfomask &= ~(1 << PFO_C);
840257f6 4060 }
4061 else {
a2c1d768 4062 fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po),
3ebea2cf 4063 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
840257f6 4064 }
a2c1d768 4065 output_std_flags(fout, po, &pfomask, buf1);
5101a5f9 4066 last_arith_dst = &po->operand[0];
4067 delayed_flag_op = NULL;
850c9265 4068 break;
4069
4070 case OP_INC:
4071 case OP_DEC:
4072 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
1bafb621 4073 if (po->operand[0].type == OPT_REG) {
4074 strcpy(buf2, po->op == OP_INC ? "++" : "--");
4075 fprintf(fout, " %s%s;", buf1, buf2);
4076 }
4077 else {
4078 strcpy(buf2, po->op == OP_INC ? "+" : "-");
4079 fprintf(fout, " %s %s= 1;", buf1, buf2);
4080 }
a2c1d768 4081 output_std_flags(fout, po, &pfomask, buf1);
5101a5f9 4082 last_arith_dst = &po->operand[0];
4083 delayed_flag_op = NULL;
4084 break;
4085
4086 case OP_NEG:
4087 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3ebea2cf 4088 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
5101a5f9 4089 fprintf(fout, " %s = -%s%s;", buf1,
4090 lmod_cast_s(po, po->operand[0].lmod), buf2);
850c9265 4091 last_arith_dst = &po->operand[0];
69a3cdfc 4092 delayed_flag_op = NULL;
940e8e66 4093 if (pfomask & (1 << PFO_C)) {
4094 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
4095 pfomask &= ~(1 << PFO_C);
4096 }
850c9265 4097 break;
4098
4099 case OP_IMUL:
de50b98b 4100 if (po->operand_cnt == 2) {
4101 propagate_lmod(po, &po->operand[0], &po->operand[1]);
850c9265 4102 goto dualop_arith;
de50b98b 4103 }
87bf6cec 4104 if (po->operand_cnt == 3)
4105 ferr(po, "TODO imul3\n");
4106 // fallthrough
4107 case OP_MUL:
4108 assert_operand_cnt(1);
4109 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
2fe80fdb 4110 fprintf(fout, " tmp64 = %seax * %s%s;\n", buf1, buf1,
3ebea2cf 4111 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
2fe80fdb 4112 fprintf(fout, " edx = tmp64 >> 32;\n");
4113 fprintf(fout, " eax = tmp64;");
87bf6cec 4114 last_arith_dst = NULL;
69a3cdfc 4115 delayed_flag_op = NULL;
91977a1c 4116 break;
4117
5101a5f9 4118 case OP_DIV:
4119 case OP_IDIV:
4120 assert_operand_cnt(1);
4121 if (po->operand[0].lmod != OPLM_DWORD)
4122 ferr(po, "unhandled lmod %d\n", po->operand[0].lmod);
4123
cb090db0 4124 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4125 strcpy(buf2, lmod_cast(po, po->operand[0].lmod,
4126 po->op == OP_IDIV));
4127 switch (po->operand[0].lmod) {
4128 case OPLM_DWORD:
4129 if (po->flags & OPF_32BIT)
4130 snprintf(buf3, sizeof(buf3), "%seax", buf2);
4131 else {
4132 fprintf(fout, " tmp64 = ((u64)edx << 32) | eax;\n");
4133 snprintf(buf3, sizeof(buf3), "%stmp64",
4134 (po->op == OP_IDIV) ? "(s64)" : "");
4135 }
4136 if (po->operand[0].type == OPT_REG
4137 && po->operand[0].reg == xDX)
4138 {
4139 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
4140 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
4141 }
4142 else {
4143 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
4144 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
4145 }
4146 break;
4147 default:
4148 ferr(po, "unhandled division type\n");
5101a5f9 4149 }
87bf6cec 4150 last_arith_dst = NULL;
4151 delayed_flag_op = NULL;
5101a5f9 4152 break;
4153
91977a1c 4154 case OP_TEST:
4155 case OP_CMP:
850c9265 4156 propagate_lmod(po, &po->operand[0], &po->operand[1]);
940e8e66 4157 if (pfomask != 0) {
69a3cdfc 4158 for (j = 0; j < 8; j++) {
940e8e66 4159 if (pfomask & (1 << j)) {
69a3cdfc 4160 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
4161 fprintf(fout, " cond_%s = %s;",
4162 parsed_flag_op_names[j], buf1);
4163 }
4164 }
940e8e66 4165 pfomask = 0;
69a3cdfc 4166 }
4167 else
4168 no_output = 1;
7ba45c34 4169 last_arith_dst = NULL;
69a3cdfc 4170 delayed_flag_op = po;
91977a1c 4171 break;
4172
69a3cdfc 4173 // note: we reuse OP_Jcc for SETcc, only flags differ
91977a1c 4174 case OP_JO ... OP_JG:
5101a5f9 4175 if (po->flags & OPF_JMP)
850c9265 4176 fprintf(fout, " goto %s;", po->operand[0].name);
5101a5f9 4177 // else SETcc - should already be handled
850c9265 4178 break;
4179
5c024ef7 4180 case OP_JECXZ:
4181 fprintf(fout, " if (ecx == 0)\n");
4182 fprintf(fout, " goto %s;", po->operand[0].name);
4183 strcat(g_comment, "jecxz");
4184 break;
4185
850c9265 4186 case OP_JMP:
87bf6cec 4187 assert_operand_cnt(1);
de50b98b 4188 last_arith_dst = NULL;
4189 delayed_flag_op = NULL;
4190
4c45fa73 4191 if (po->operand[0].type == OPT_REGMEM) {
4192 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
4193 buf1, buf2);
4194 if (ret != 2)
4195 ferr(po, "parse failure for jmp '%s'\n",
4196 po->operand[0].name);
4197 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
4198 break;
4199 }
4200 else if (po->operand[0].type != OPT_LABEL)
4201 ferr(po, "unhandled jmp type\n");
87bf6cec 4202
850c9265 4203 fprintf(fout, " goto %s;", po->operand[0].name);
91977a1c 4204 break;
4205
4206 case OP_CALL:
5101a5f9 4207 assert_operand_cnt(1);
850c9265 4208 pp = po->datap;
89ff3147 4209 my_assert_not(pp, NULL);
91977a1c 4210
89ff3147 4211 if (pp->is_fptr)
ddaf8bd7 4212 fprintf(fout, " %s = %s;\n", pp->name,
1cd4a663 4213 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
4214 "(void *)", 0));
1bafb621 4215
91977a1c 4216 fprintf(fout, " ");
2b43685d 4217 if (strstr(pp->ret_type.name, "int64")) {
87bf6cec 4218 if (po->flags & OPF_TAIL)
2b43685d 4219 ferr(po, "int64 and tail?\n");
2fe80fdb 4220 fprintf(fout, "tmp64 = ");
2b43685d 4221 }
4222 else if (!IS(pp->ret_type.name, "void")) {
4223 if (po->flags & OPF_TAIL) {
840257f6 4224 if (!IS(g_func_pp->ret_type.name, "void")) {
4225 fprintf(fout, "return ");
4226 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
4227 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
4228 }
2b43685d 4229 }
75ad0378 4230 else if (regmask & (1 << xAX)) {
87bf6cec 4231 fprintf(fout, "eax = ");
2b43685d 4232 if (pp->ret_type.is_ptr)
4233 fprintf(fout, "(u32)");
4234 }
91977a1c 4235 }
87bf6cec 4236
ddaf8bd7 4237 if (pp->name[0] == 0)
4238 ferr(po, "missing pp->name\n");
4239 fprintf(fout, "%s%s(", pp->name,
4240 pp->has_structarg ? "_sa" : "");
39b168b8 4241
2fe80fdb 4242 if (po->flags & OPF_ATAIL) {
4243 if (pp->argc_stack != g_func_pp->argc_stack
4244 || (pp->argc_stack > 0
4245 && pp->is_stdcall != g_func_pp->is_stdcall))
4246 ferr(po, "incompatible tailcall\n");
87bf6cec 4247
2fe80fdb 4248 for (arg = j = 0; arg < pp->argc; arg++) {
4249 if (arg > 0)
4250 fprintf(fout, ", ");
87bf6cec 4251
2fe80fdb 4252 cast[0] = 0;
4253 if (pp->arg[arg].type.is_ptr)
4254 snprintf(cast, sizeof(cast), "(%s)",
4255 pp->arg[arg].type.name);
91977a1c 4256
2fe80fdb 4257 if (pp->arg[arg].reg != NULL) {
4258 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
4259 continue;
4260 }
4261 // stack arg
4262 for (; j < g_func_pp->argc; j++)
4263 if (g_func_pp->arg[j].reg == NULL)
4264 break;
4265 fprintf(fout, "%sa%d", cast, j + 1);
4266 j++;
69a3cdfc 4267 }
2fe80fdb 4268 }
4269 else {
4270 for (arg = 0; arg < pp->argc; arg++) {
4271 if (arg > 0)
4272 fprintf(fout, ", ");
4273
4274 cast[0] = 0;
4275 if (pp->arg[arg].type.is_ptr)
4276 snprintf(cast, sizeof(cast), "(%s)",
4277 pp->arg[arg].type.name);
4278
4279 if (pp->arg[arg].reg != NULL) {
4280 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
4281 continue;
4282 }
4283
4284 // stack arg
4285 tmp_op = pp->arg[arg].datap;
4286 if (tmp_op == NULL)
4287 ferr(po, "parsed_op missing for arg%d\n", arg);
4288 if (tmp_op->argnum != 0) {
4289 fprintf(fout, "%ss_a%d", cast, tmp_op->argnum);
4290 }
4291 else {
4292 fprintf(fout, "%s",
4293 out_src_opr(buf1, sizeof(buf1),
4294 tmp_op, &tmp_op->operand[0], cast, 0));
4295 }
69a3cdfc 4296 }
91977a1c 4297 }
4298 fprintf(fout, ");");
87bf6cec 4299
2b43685d 4300 if (strstr(pp->ret_type.name, "int64")) {
4301 fprintf(fout, "\n");
2fe80fdb 4302 fprintf(fout, " edx = tmp64 >> 32;\n");
4303 fprintf(fout, " eax = tmp64;");
2b43685d 4304 }
4305
89ff3147 4306 if (pp->is_unresolved) {
ddaf8bd7 4307 snprintf(buf3, sizeof(buf3), " unresoved %dreg",
89ff3147 4308 pp->argc_reg);
4309 strcat(g_comment, buf3);
4310 }
4311
87bf6cec 4312 if (po->flags & OPF_TAIL) {
840257f6 4313 ret = 0;
ddaf8bd7 4314 if (i == opcnt - 1 || pp->is_noreturn)
840257f6 4315 ret = 0;
4316 else if (IS(pp->ret_type.name, "void"))
4317 ret = 1;
4318 else if (IS(g_func_pp->ret_type.name, "void"))
4319 ret = 1;
4320 // else already handled as 'return f()'
4321
4322 if (ret) {
da87ae38 4323 if (!IS(g_func_pp->ret_type.name, "void")) {
ddaf8bd7 4324 ferr(po, "int func -> void func tailcall?\n");
da87ae38 4325 }
4326 else {
4327 fprintf(fout, "\n return;");
ddaf8bd7 4328 strcat(g_comment, " ^ tailcall");
da87ae38 4329 }
3ebea2cf 4330 }
89ff3147 4331 else
ddaf8bd7 4332 strcat(g_comment, " tailcall");
87bf6cec 4333 }
ddaf8bd7 4334 if (pp->is_noreturn)
4335 strcat(g_comment, " noreturn");
2fe80fdb 4336 if ((po->flags & OPF_ATAIL) && pp->argc_stack > 0)
4337 strcat(g_comment, " argframe");
87bf6cec 4338 delayed_flag_op = NULL;
4339 last_arith_dst = NULL;
91977a1c 4340 break;
4341
4342 case OP_RET:
bd96f656 4343 if (g_func_pp->is_vararg)
4f12f671 4344 fprintf(fout, " va_end(ap);\n");
4345
bd96f656 4346 if (IS(g_func_pp->ret_type.name, "void")) {
3ebea2cf 4347 if (i != opcnt - 1 || label_pending)
4348 fprintf(fout, " return;");
4349 }
bd96f656 4350 else if (g_func_pp->ret_type.is_ptr) {
d4e3b5db 4351 fprintf(fout, " return (%s)eax;",
bd96f656 4352 g_func_pp->ret_type.name);
3ebea2cf 4353 }
2fe80fdb 4354 else if (IS(g_func_pp->ret_type.name, "__int64"))
4355 fprintf(fout, " return ((u64)edx << 32) | eax;");
91977a1c 4356 else
4357 fprintf(fout, " return eax;");
de50b98b 4358
4359 last_arith_dst = NULL;
4360 delayed_flag_op = NULL;
91977a1c 4361 break;
4362
4363 case OP_PUSH:
de50b98b 4364 if (po->argnum != 0) {
69a3cdfc 4365 // special case - saved func arg
3ebea2cf 4366 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
de50b98b 4367 fprintf(fout, " s_a%d = %s;", po->argnum, buf1);
69a3cdfc 4368 break;
4369 }
d4e3b5db 4370 else if (po->flags & OPF_RSAVE) {
3ebea2cf 4371 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
d4e3b5db 4372 fprintf(fout, " s_%s = %s;", buf1, buf1);
4373 break;
4374 }
e56ab892 4375 if (!(g_ida_func_attr & IDAFA_NORETURN))
4376 ferr(po, "stray push encountered\n");
4377 no_output = 1;
91977a1c 4378 break;
4379
4380 case OP_POP:
d4e3b5db 4381 if (po->flags & OPF_RSAVE) {
3ebea2cf 4382 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
d4e3b5db 4383 fprintf(fout, " %s = s_%s;", buf1, buf1);
4384 break;
4385 }
5c024ef7 4386 else if (po->datap != NULL) {
4387 // push/pop pair
4388 tmp_op = po->datap;
4389 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4390 fprintf(fout, " %s = %s;", buf1,
4391 out_src_opr(buf2, sizeof(buf2),
4392 tmp_op, &tmp_op->operand[0],
4393 po->operand[0].is_ptr ? "(void *)" : "", 0));
4394 break;
4395 }
d4e3b5db 4396 ferr(po, "stray pop encountered\n");
91977a1c 4397 break;
4398
33c35af6 4399 case OP_NOP:
2b43685d 4400 no_output = 1;
33c35af6 4401 break;
4402
91977a1c 4403 default:
4404 no_output = 1;
69a3cdfc 4405 ferr(po, "unhandled op type %d, flags %x\n",
4406 po->op, po->flags);
91977a1c 4407 break;
4408 }
4409
4410 if (g_comment[0] != 0) {
ddaf8bd7 4411 char *p = g_comment;
4412 while (my_isblank(*p))
4413 p++;
4414 fprintf(fout, " // %s", p);
91977a1c 4415 g_comment[0] = 0;
4416 no_output = 0;
4417 }
4418 if (!no_output)
4419 fprintf(fout, "\n");
5101a5f9 4420
2b43685d 4421 // some sanity checking
591721d7 4422 if (po->flags & OPF_REP) {
4423 if (po->op != OP_STOS && po->op != OP_MOVS
4424 && po->op != OP_CMPS && po->op != OP_SCAS)
4425 ferr(po, "unexpected rep\n");
4426 if (!(po->flags & (OPF_REPZ|OPF_REPNZ))
4427 && (po->op == OP_CMPS || po->op == OP_SCAS))
4428 ferr(po, "cmps/scas with plain rep\n");
4429 }
4430 if ((po->flags & (OPF_REPZ|OPF_REPNZ))
4431 && po->op != OP_CMPS && po->op != OP_SCAS)
2b43685d 4432 ferr(po, "unexpected repz/repnz\n");
4433
940e8e66 4434 if (pfomask != 0)
7ba45c34 4435 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
940e8e66 4436
5101a5f9 4437 // see is delayed flag stuff is still valid
4438 if (delayed_flag_op != NULL && delayed_flag_op != po) {
89ff3147 4439 if (is_any_opr_modified(delayed_flag_op, po, 0))
5101a5f9 4440 delayed_flag_op = NULL;
4441 }
4442
4443 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
4444 if (is_opr_modified(last_arith_dst, po))
4445 last_arith_dst = NULL;
4446 }
3ebea2cf 4447
4448 label_pending = 0;
91977a1c 4449 }
4450
a2c1d768 4451 if (g_stack_fsz && !g_stack_frame_used)
4452 fprintf(fout, " (void)sf;\n");
4453
91977a1c 4454 fprintf(fout, "}\n\n");
4455
4456 // cleanup
4457 for (i = 0; i < opcnt; i++) {
4c45fa73 4458 struct label_ref *lr, *lr_del;
4459
4460 lr = g_label_refs[i].next;
4461 while (lr != NULL) {
4462 lr_del = lr;
4463 lr = lr->next;
4464 free(lr_del);
4465 }
4466 g_label_refs[i].i = -1;
4467 g_label_refs[i].next = NULL;
4468
91977a1c 4469 if (ops[i].op == OP_CALL) {
4470 pp = ops[i].datap;
bd96f656 4471 if (pp)
91977a1c 4472 proto_release(pp);
91977a1c 4473 }
4474 }
bd96f656 4475 g_func_pp = NULL;
91977a1c 4476}
c36e914d 4477
d4e3b5db 4478static void set_label(int i, const char *name)
4479{
4480 const char *p;
4481 int len;
4482
4483 len = strlen(name);
4484 p = strchr(name, ':');
4485 if (p != NULL)
4486 len = p - name;
4487
4488 if (len > sizeof(g_labels[0]) - 1)
4489 aerr("label '%s' too long: %d\n", name, len);
e56ab892 4490 if (g_labels[i][0] != 0 && !IS_START(g_labels[i], "algn_"))
4491 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
d4e3b5db 4492 memcpy(g_labels[i], name, len);
4493 g_labels[i][len] = 0;
4494}
4495
bfa4a6ee 4496// '=' needs special treatment..
4497static char *next_word_s(char *w, size_t wsize, char *s)
4498{
4499 size_t i;
4500
4501 s = sskip(s);
4502
4503 for (i = 0; i < wsize - 1; i++) {
4504 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
4505 break;
4506 w[i] = s[i];
4507 }
4508 w[i] = 0;
4509
4510 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
4511 printf("warning: '%s' truncated\n", w);
4512
4513 return s + i;
4514}
4515
e56ab892 4516struct chunk_item {
4517 char *name;
4518 long fptr;
de50b98b 4519 int asmln;
e56ab892 4520};
4521
cdfaeed7 4522static struct chunk_item *func_chunks;
4523static int func_chunk_cnt;
4524static int func_chunk_alloc;
4525
4526static void add_func_chunk(FILE *fasm, const char *name, int line)
4527{
4528 if (func_chunk_cnt >= func_chunk_alloc) {
4529 func_chunk_alloc *= 2;
4530 func_chunks = realloc(func_chunks,
4531 func_chunk_alloc * sizeof(func_chunks[0]));
4532 my_assert_not(func_chunks, NULL);
4533 }
4534 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
4535 func_chunks[func_chunk_cnt].name = strdup(name);
4536 func_chunks[func_chunk_cnt].asmln = line;
4537 func_chunk_cnt++;
4538}
4539
e56ab892 4540static int cmp_chunks(const void *p1, const void *p2)
4541{
4542 const struct chunk_item *c1 = p1, *c2 = p2;
4543 return strcmp(c1->name, c2->name);
4544}
4545
bfa4a6ee 4546static int cmpstringp(const void *p1, const void *p2)
4547{
4548 return strcmp(*(char * const *)p1, *(char * const *)p2);
4549}
4550
cdfaeed7 4551static void scan_ahead(FILE *fasm)
4552{
4553 char words[2][256];
4554 char line[256];
4555 long oldpos;
4556 int oldasmln;
4557 int wordc;
4558 char *p;
4559 int i;
4560
4561 oldpos = ftell(fasm);
4562 oldasmln = asmln;
4563
4564 while (fgets(line, sizeof(line), fasm))
4565 {
4566 wordc = 0;
4567 asmln++;
4568
4569 p = sskip(line);
4570 if (*p == 0)
4571 continue;
4572
4573 if (*p == ';')
4574 {
4575 // get rid of random tabs
4576 for (i = 0; line[i] != 0; i++)
4577 if (line[i] == '\t')
4578 line[i] = ' ';
4579
4580 if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
4581 {
4582 p += 30;
4583 next_word(words[0], sizeof(words[0]), p);
4584 if (words[0][0] == 0)
4585 aerr("missing name for func chunk?\n");
4586
4587 add_func_chunk(fasm, words[0], asmln);
4588 }
4589 continue;
4590 } // *p == ';'
4591
4592 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
4593 words[wordc][0] = 0;
4594 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
4595 if (*p == 0 || *p == ';') {
4596 wordc++;
4597 break;
4598 }
4599 }
4600
4601 if (wordc == 2 && IS(words[1], "ends"))
4602 break;
4603 }
4604
4605 fseek(fasm, oldpos, SEEK_SET);
4606 asmln = oldasmln;
4607}
4608
91977a1c 4609int main(int argc, char *argv[])
4610{
06c5d854 4611 FILE *fout, *fasm, *frlist;
4c45fa73 4612 struct parsed_data *pd = NULL;
4613 int pd_alloc = 0;
4614 char **rlist = NULL;
4615 int rlist_len = 0;
4616 int rlist_alloc = 0;
e56ab892 4617 int func_chunks_used = 0;
4618 int func_chunks_sorted = 0;
e56ab892 4619 int func_chunk_i = -1;
4620 long func_chunk_ret = 0;
de50b98b 4621 int func_chunk_ret_ln = 0;
cdfaeed7 4622 int scanned_ahead = 0;
91977a1c 4623 char line[256];
a2c1d768 4624 char words[20][256];
4c45fa73 4625 enum opr_lenmod lmod;
ddaf8bd7 4626 char *sctproto = NULL;
91977a1c 4627 int in_func = 0;
4c45fa73 4628 int pending_endp = 0;
bfa4a6ee 4629 int skip_func = 0;
940e8e66 4630 int skip_warned = 0;
91977a1c 4631 int eq_alloc;
bfa4a6ee 4632 int verbose = 0;
4633 int arg_out;
89ff3147 4634 int arg;
91977a1c 4635 int pi = 0;
e56ab892 4636 int i, j;
4637 int ret, len;
91977a1c 4638 char *p;
4639 int wordc;
4640
89ff3147 4641 for (arg = 1; arg < argc; arg++) {
4642 if (IS(argv[arg], "-v"))
4643 verbose = 1;
4644 else if (IS(argv[arg], "-rf"))
4645 g_allow_regfunc = 1;
4646 else
4647 break;
bfa4a6ee 4648 }
4649
4650 if (argc < arg + 3) {
89ff3147 4651 printf("usage:\n%s [-v] [-rf] <.c> <.asm> <hdrf> [rlist]*\n",
91977a1c 4652 argv[0]);
4653 return 1;
4654 }
4655
bfa4a6ee 4656 arg_out = arg++;
91977a1c 4657
bfa4a6ee 4658 asmfn = argv[arg++];
91977a1c 4659 fasm = fopen(asmfn, "r");
4660 my_assert_not(fasm, NULL);
4661
bfa4a6ee 4662 hdrfn = argv[arg++];
06c5d854 4663 g_fhdr = fopen(hdrfn, "r");
4664 my_assert_not(g_fhdr, NULL);
bfa4a6ee 4665
4666 rlist_alloc = 64;
4667 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
4668 my_assert_not(rlist, NULL);
4669 // needs special handling..
4670 rlist[rlist_len++] = "__alloca_probe";
4671
e56ab892 4672 func_chunk_alloc = 32;
4673 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
4674 my_assert_not(func_chunks, NULL);
4675
a2c1d768 4676 memset(words, 0, sizeof(words));
4677
bfa4a6ee 4678 for (; arg < argc; arg++) {
4679 frlist = fopen(argv[arg], "r");
4680 my_assert_not(frlist, NULL);
4681
4682 while (fgets(line, sizeof(line), frlist)) {
4683 p = sskip(line);
1cd4a663 4684 if (*p == 0 || *p == ';')
4685 continue;
4686 if (*p == '#') {
89ff3147 4687 if (IS_START(p, "#if 0")
4688 || (g_allow_regfunc && IS_START(p, "#if NO_REGFUNC")))
4689 {
1cd4a663 4690 skip_func = 1;
89ff3147 4691 }
1cd4a663 4692 else if (IS_START(p, "#endif"))
4693 skip_func = 0;
4694 continue;
4695 }
4696 if (skip_func)
bfa4a6ee 4697 continue;
4698
4699 p = next_word(words[0], sizeof(words[0]), p);
4700 if (words[0][0] == 0)
4701 continue;
4702
4703 if (rlist_len >= rlist_alloc) {
4704 rlist_alloc = rlist_alloc * 2 + 64;
4705 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
4706 my_assert_not(rlist, NULL);
4707 }
4708 rlist[rlist_len++] = strdup(words[0]);
4709 }
1cd4a663 4710 skip_func = 0;
bfa4a6ee 4711
4712 fclose(frlist);
4713 frlist = NULL;
4714 }
4715
4716 if (rlist_len > 0)
4717 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
4718
4719 fout = fopen(argv[arg_out], "w");
91977a1c 4720 my_assert_not(fout, NULL);
4721
4722 eq_alloc = 128;
4723 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
4724 my_assert_not(g_eqs, NULL);
4725
4c45fa73 4726 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
4727 g_label_refs[i].i = -1;
4728 g_label_refs[i].next = NULL;
4729 }
4730
91977a1c 4731 while (fgets(line, sizeof(line), fasm))
4732 {
4c45fa73 4733 wordc = 0;
91977a1c 4734 asmln++;
4735
4736 p = sskip(line);
1bafb621 4737 if (*p == 0)
91977a1c 4738 continue;
4739
de50b98b 4740 // get rid of random tabs
4741 for (i = 0; line[i] != 0; i++)
4742 if (line[i] == '\t')
4743 line[i] = ' ';
4744
e56ab892 4745 if (*p == ';')
4746 {
e56ab892 4747 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
4748 goto do_pending_endp; // eww..
4749
4750 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
4751 {
4752 static const char *attrs[] = {
4753 "bp-based frame",
4754 "library function",
4755 "static",
4756 "noreturn",
4757 "thunk",
4758 "fpd=",
4759 };
4760
4761 // parse IDA's attribute-list comment
4762 g_ida_func_attr = 0;
4763 p = sskip(p + 13);
4764
4765 for (; *p != 0; p = sskip(p)) {
4766 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
4767 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
4768 g_ida_func_attr |= 1 << i;
4769 p += strlen(attrs[i]);
4770 break;
4771 }
4772 }
4773 if (i == ARRAY_SIZE(attrs)) {
4774 anote("unparsed IDA attr: %s\n", p);
1bafb621 4775 break;
4776 }
e56ab892 4777 if (IS(attrs[i], "fpd=")) {
4778 p = next_word(words[0], sizeof(words[0]), p);
4779 // ignore for now..
4780 }
1bafb621 4781 }
e56ab892 4782 }
4783 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
4784 {
4785 p += 30;
4786 next_word(words[0], sizeof(words[0]), p);
4787 if (words[0][0] == 0)
cdfaeed7 4788 aerr("missing name for func chunk?\n");
4789
4790 if (!scanned_ahead) {
4791 add_func_chunk(fasm, words[0], asmln);
4792 func_chunks_sorted = 0;
e56ab892 4793 }
e56ab892 4794 }
4795 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
4796 {
4797 if (func_chunk_i >= 0) {
4798 if (func_chunk_i < func_chunk_cnt
4799 && IS(func_chunks[func_chunk_i].name, g_func))
4800 {
4801 // move on to next chunk
4802 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
4803 if (ret)
4804 aerr("seek failed for '%s' chunk #%d\n",
4805 g_func, func_chunk_i);
de50b98b 4806 asmln = func_chunks[func_chunk_i].asmln;
e56ab892 4807 func_chunk_i++;
4808 }
4809 else {
4810 if (func_chunk_ret == 0)
4811 aerr("no return from chunk?\n");
4812 fseek(fasm, func_chunk_ret, SEEK_SET);
de50b98b 4813 asmln = func_chunk_ret_ln;
e56ab892 4814 func_chunk_ret = 0;
4815 pending_endp = 1;
4816 }
1bafb621 4817 }
e56ab892 4818 }
4819 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
4820 func_chunks_used = 1;
4821 p += 20;
4822 if (IS_START(g_func, "sub_")) {
4823 unsigned long addr = strtoul(p, NULL, 16);
4824 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
cdfaeed7 4825 if (addr > f_addr && !scanned_ahead) {
94d447fb 4826 anote("scan_ahead caused by '%s', addr %lx\n",
4827 g_func, addr);
cdfaeed7 4828 scan_ahead(fasm);
4829 scanned_ahead = 1;
4830 func_chunks_sorted = 0;
4831 }
1bafb621 4832 }
4833 }
4834 continue;
e56ab892 4835 } // *p == ';'
1bafb621 4836
06c5d854 4837parse_words:
a2c1d768 4838 for (i = wordc; i < ARRAY_SIZE(words); i++)
4839 words[i][0] = 0;
cdfaeed7 4840 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
bfa4a6ee 4841 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
91977a1c 4842 if (*p == 0 || *p == ';') {
4843 wordc++;
4844 break;
4845 }
4846 }
a2c1d768 4847 if (*p != 0 && *p != ';')
4848 aerr("too many words\n");
91977a1c 4849
06c5d854 4850 // alow asm patches in comments
ddaf8bd7 4851 if (*p == ';') {
4852 if (IS_START(p, "; sctpatch:")) {
4853 p = sskip(p + 11);
4854 if (*p == 0 || *p == ';')
4855 continue;
4856 goto parse_words; // lame
4857 }
4858 if (IS_START(p, "; sctproto:")) {
4859 sctproto = strdup(p + 11);
4860 }
06c5d854 4861 }
4862
91977a1c 4863 if (wordc == 0) {
4864 // shouldn't happen
4865 awarn("wordc == 0?\n");
4866 continue;
4867 }
4868
4869 // don't care about this:
4870 if (words[0][0] == '.'
4871 || IS(words[0], "include")
4872 || IS(words[0], "assume") || IS(words[1], "segment")
4873 || IS(words[0], "align"))
4874 {
4875 continue;
4876 }
4877
4c45fa73 4878do_pending_endp:
4879 // do delayed endp processing to collect switch jumptables
4880 if (pending_endp) {
4881 if (in_func && !skip_func && wordc >= 2
4882 && ((words[0][0] == 'd' && words[0][2] == 0)
4883 || (words[1][0] == 'd' && words[1][2] == 0)))
4884 {
4885 i = 1;
4886 if (words[1][0] == 'd' && words[1][2] == 0) {
4887 // label
4888 if (g_func_pd_cnt >= pd_alloc) {
4889 pd_alloc = pd_alloc * 2 + 16;
4890 g_func_pd = realloc(g_func_pd,
4891 sizeof(g_func_pd[0]) * pd_alloc);
4892 my_assert_not(g_func_pd, NULL);
4893 }
4894 pd = &g_func_pd[g_func_pd_cnt];
4895 g_func_pd_cnt++;
4896 memset(pd, 0, sizeof(*pd));
4897 strcpy(pd->label, words[0]);
4898 pd->type = OPT_CONST;
4899 pd->lmod = lmod_from_directive(words[1]);
4900 i = 2;
4901 }
4902 else {
da87ae38 4903 if (pd == NULL) {
4904 if (verbose)
4905 anote("skipping alignment byte?\n");
4906 continue;
4907 }
4c45fa73 4908 lmod = lmod_from_directive(words[0]);
4909 if (lmod != pd->lmod)
4910 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
4911 }
4912
4913 if (pd->count_alloc < pd->count + wordc) {
4914 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
4915 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
4916 my_assert_not(pd->d, NULL);
4917 }
4918 for (; i < wordc; i++) {
4919 if (IS(words[i], "offset")) {
4920 pd->type = OPT_OFFSET;
4921 i++;
4922 }
4923 p = strchr(words[i], ',');
4924 if (p != NULL)
4925 *p = 0;
4926 if (pd->type == OPT_OFFSET)
4927 pd->d[pd->count].u.label = strdup(words[i]);
4928 else
4929 pd->d[pd->count].u.val = parse_number(words[i]);
4930 pd->d[pd->count].bt_i = -1;
4931 pd->count++;
4932 }
4933 continue;
4934 }
4935
4936 if (in_func && !skip_func)
4937 gen_func(fout, g_fhdr, g_func, pi);
4938
4939 pending_endp = 0;
4940 in_func = 0;
4941 g_ida_func_attr = 0;
4942 skip_warned = 0;
4943 skip_func = 0;
4944 g_func[0] = 0;
e56ab892 4945 func_chunks_used = 0;
4946 func_chunk_i = -1;
4c45fa73 4947 if (pi != 0) {
4948 memset(&ops, 0, pi * sizeof(ops[0]));
4949 memset(g_labels, 0, pi * sizeof(g_labels[0]));
4950 pi = 0;
4951 }
4952 g_eqcnt = 0;
4953 for (i = 0; i < g_func_pd_cnt; i++) {
4954 pd = &g_func_pd[i];
4955 if (pd->type == OPT_OFFSET) {
4956 for (j = 0; j < pd->count; j++)
4957 free(pd->d[j].u.label);
4958 }
4959 free(pd->d);
4960 pd->d = NULL;
4961 }
4962 g_func_pd_cnt = 0;
4963 pd = NULL;
4964 if (wordc == 0)
4965 continue;
4966 }
4967
91977a1c 4968 if (IS(words[1], "proc")) {
4969 if (in_func)
4970 aerr("proc '%s' while in_func '%s'?\n",
4971 words[0], g_func);
bfa4a6ee 4972 p = words[0];
ddaf8bd7 4973 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
bfa4a6ee 4974 skip_func = 1;
91977a1c 4975 strcpy(g_func, words[0]);
d4e3b5db 4976 set_label(0, words[0]);
91977a1c 4977 in_func = 1;
4978 continue;
4979 }
4980
e56ab892 4981 if (IS(words[1], "endp"))
4982 {
91977a1c 4983 if (!in_func)
4984 aerr("endp '%s' while not in_func?\n", words[0]);
4985 if (!IS(g_func, words[0]))
4986 aerr("endp '%s' while in_func '%s'?\n",
4987 words[0], g_func);
bfa4a6ee 4988
ddaf8bd7 4989 if ((g_ida_func_attr & IDAFA_THUNK) && pi == 1
4990 && ops[0].op == OP_JMP && ops[0].operand[0].had_ds)
4991 {
4992 // import jump
4993 skip_func = 1;
4994 }
4995
e56ab892 4996 if (!skip_func && func_chunks_used) {
4997 // start processing chunks
4998 struct chunk_item *ci, key = { g_func, 0 };
4999
5000 func_chunk_ret = ftell(fasm);
de50b98b 5001 func_chunk_ret_ln = asmln;
e56ab892 5002 if (!func_chunks_sorted) {
5003 qsort(func_chunks, func_chunk_cnt,
5004 sizeof(func_chunks[0]), cmp_chunks);
5005 func_chunks_sorted = 1;
5006 }
5007 ci = bsearch(&key, func_chunks, func_chunk_cnt,
5008 sizeof(func_chunks[0]), cmp_chunks);
5009 if (ci == NULL)
5010 aerr("'%s' needs chunks, but none found\n", g_func);
5011 func_chunk_i = ci - func_chunks;
5012 for (; func_chunk_i > 0; func_chunk_i--)
5013 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
5014 break;
5015
5016 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
5017 if (ret)
5018 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
de50b98b 5019 asmln = func_chunks[func_chunk_i].asmln;
e56ab892 5020 func_chunk_i++;
5021 continue;
5022 }
4c45fa73 5023 pending_endp = 1;
91977a1c 5024 continue;
5025 }
5026
a2c1d768 5027 if (wordc == 2 && IS(words[1], "ends"))
5028 break;
5029
bfa4a6ee 5030 p = strchr(words[0], ':');
5031 if (p != NULL) {
d4e3b5db 5032 set_label(pi, words[0]);
bfa4a6ee 5033 continue;
5034 }
5035
5036 if (!in_func || skip_func) {
5037 if (!skip_warned && !skip_func && g_labels[pi][0] != 0) {
5038 if (verbose)
5039 anote("skipping from '%s'\n", g_labels[pi]);
5040 skip_warned = 1;
5041 }
5042 g_labels[pi][0] = 0;
5043 continue;
5044 }
5045
ddaf8bd7 5046 if (wordc > 1 && IS(words[1], "="))
5047 {
91977a1c 5048 if (wordc != 5)
5049 aerr("unhandled equ, wc=%d\n", wordc);
5050 if (g_eqcnt >= eq_alloc) {
5051 eq_alloc *= 2;
5052 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
5053 my_assert_not(g_eqs, NULL);
5054 }
5055
5056 len = strlen(words[0]);
5057 if (len > sizeof(g_eqs[0].name) - 1)
5058 aerr("equ name too long: %d\n", len);
5059 strcpy(g_eqs[g_eqcnt].name, words[0]);
5060
5061 if (!IS(words[3], "ptr"))
5062 aerr("unhandled equ\n");
5063 if (IS(words[2], "dword"))
5064 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
5065 else if (IS(words[2], "word"))
5066 g_eqs[g_eqcnt].lmod = OPLM_WORD;
5067 else if (IS(words[2], "byte"))
5068 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
5069 else
5070 aerr("bad lmod: '%s'\n", words[2]);
5071
5072 g_eqs[g_eqcnt].offset = parse_number(words[4]);
5073 g_eqcnt++;
5074 continue;
5075 }
5076
5077 if (pi >= ARRAY_SIZE(ops))
5078 aerr("too many ops\n");
5079
91977a1c 5080 parse_op(&ops[pi], words, wordc);
ddaf8bd7 5081
5082 if (sctproto != NULL) {
5083 if (ops[pi].op == OP_CALL)
5084 ops[pi].datap = sctproto;
5085 sctproto = NULL;
5086 }
91977a1c 5087 pi++;
91977a1c 5088 }
5089
5090 fclose(fout);
5091 fclose(fasm);
06c5d854 5092 fclose(g_fhdr);
91977a1c 5093
5094 return 0;
c36e914d 5095}
91977a1c 5096
5097// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.