notaz.gp2x.de / ia32rtools.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
translate: output complex headers from seed
[ia32rtools.git] / tools / translate.c
CommitLineData
7637b6cc 1/*
2 * ia32rtools
2c10ea1f 3 * (C) notaz, 2013-2015
7637b6cc 4 *
5 * This work is licensed under the terms of 3-clause BSD license.
6 * See COPYING file in the top-level directory.
7 */
8
33c35af6 9#define _GNU_SOURCE
c36e914d 10#include <stdio.h>
11#include <stdlib.h>
12#include <string.h>
13
14#include "my_assert.h"
15#include "my_str.h"
16
17#define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
18#define IS(w, y) !strcmp(w, y)
39b168b8 19#define IS_START(w, y) !strncmp(w, y, strlen(y))
c36e914d 20
21#include "protoparse.h"
22
06c5d854 23static const char *asmfn;
c36e914d 24static int asmln;
06c5d854 25static FILE *g_fhdr;
c36e914d 26
940e8e66 27#define anote(fmt, ...) \
28 printf("%s:%d: note: " fmt, asmfn, asmln, ##__VA_ARGS__)
c36e914d 29#define awarn(fmt, ...) \
940e8e66 30 printf("%s:%d: warning: " fmt, asmfn, asmln, ##__VA_ARGS__)
c36e914d 31#define aerr(fmt, ...) do { \
940e8e66 32 printf("%s:%d: error: " fmt, asmfn, asmln, ##__VA_ARGS__); \
33c35af6 33 fcloseall(); \
c36e914d 34 exit(1); \
35} while (0)
36
054f95b2 37#include "masm_tools.h"
38
69a3cdfc 39enum op_flags {
5e49b270 40 OPF_RMD = (1 << 0), /* removed from code generation */
87bf6cec 41 OPF_DATA = (1 << 1), /* data processing - writes to dst opr */
42 OPF_FLAGS = (1 << 2), /* sets flags */
5c024ef7 43 OPF_JMP = (1 << 3), /* branch, call */
44 OPF_CJMP = (1 << 4), /* cond. branch (cc or jecxz) */
45 OPF_CC = (1 << 5), /* uses flags */
46 OPF_TAIL = (1 << 6), /* ret or tail call */
47 OPF_RSAVE = (1 << 7), /* push/pop is local reg save/load */
48 OPF_REP = (1 << 8), /* prefixed by rep */
49 OPF_REPZ = (1 << 9), /* rep is repe/repz */
50 OPF_REPNZ = (1 << 10), /* rep is repne/repnz */
9af2d373 51 OPF_FARG = (1 << 11), /* push collected as func arg */
52 OPF_FARGNR = (1 << 12), /* push collected as func arg (no reuse) */
53 OPF_EBP_S = (1 << 13), /* ebp used as scratch here, not BP */
54 OPF_DF = (1 << 14), /* DF flag set */
55 OPF_ATAIL = (1 << 15), /* tail call with reused arg frame */
56 OPF_32BIT = (1 << 16), /* 32bit division */
57 OPF_LOCK = (1 << 17), /* op has lock prefix */
58 OPF_VAPUSH = (1 << 18), /* vararg ptr push (as call arg) */
26677139 59 OPF_DONE = (1 << 19), /* already fully handled by analysis */
c36e914d 60};
61
62enum op_op {
63 OP_INVAL,
33c35af6 64 OP_NOP,
c36e914d 65 OP_PUSH,
66 OP_POP,
591721d7 67 OP_LEAVE,
c36e914d 68 OP_MOV,
850c9265 69 OP_LEA,
70 OP_MOVZX,
71 OP_MOVSX,
108e9fe3 72 OP_XCHG,
850c9265 73 OP_NOT,
5101a5f9 74 OP_CDQ,
092f64e1 75 OP_LODS,
33c35af6 76 OP_STOS,
d4e3b5db 77 OP_MOVS,
7ba45c34 78 OP_CMPS,
591721d7 79 OP_SCAS,
80 OP_STD,
81 OP_CLD,
c36e914d 82 OP_RET,
83 OP_ADD,
91977a1c 84 OP_SUB,
850c9265 85 OP_AND,
86 OP_OR,
87 OP_XOR,
88 OP_SHL,
89 OP_SHR,
90 OP_SAR,
3b2f4044 91 OP_SHRD,
d4e3b5db 92 OP_ROL,
93 OP_ROR,
cb090db0 94 OP_RCL,
95 OP_RCR,
69a3cdfc 96 OP_ADC,
850c9265 97 OP_SBB,
1f84f6b3 98 OP_BSF,
850c9265 99 OP_INC,
100 OP_DEC,
5101a5f9 101 OP_NEG,
850c9265 102 OP_MUL,
103 OP_IMUL,
5101a5f9 104 OP_DIV,
105 OP_IDIV,
c36e914d 106 OP_TEST,
107 OP_CMP,
108 OP_CALL,
109 OP_JMP,
5c024ef7 110 OP_JECXZ,
092f64e1 111 OP_JCC,
112 OP_SCC,
90307a99 113 // x87
114 // mmx
115 OP_EMMS,
2c10ea1f 116 // mmx
117 OP_UD2,
c36e914d 118};
119
120enum opr_type {
87bf6cec 121 OPT_UNSPEC,
122 OPT_REG,
123 OPT_REGMEM,
124 OPT_LABEL,
850c9265 125 OPT_OFFSET,
87bf6cec 126 OPT_CONST,
c36e914d 127};
128
2b43685d 129// must be sorted (larger len must be further in enum)
c36e914d 130enum opr_lenmod {
91977a1c 131 OPLM_UNSPEC,
132 OPLM_BYTE,
133 OPLM_WORD,
134 OPLM_DWORD,
90307a99 135 OPLM_QWORD,
c36e914d 136};
137
850c9265 138#define MAX_OPERANDS 3
92d715b6 139#define NAMELEN 112
c36e914d 140
141struct parsed_opr {
91977a1c 142 enum opr_type type;
143 enum opr_lenmod lmod;
7ba45c34 144 unsigned int is_ptr:1; // pointer in C
145 unsigned int is_array:1; // array in C
a3684be1 146 unsigned int type_from_var:1; // .. in header, sometimes wrong
2b43685d 147 unsigned int size_mismatch:1; // type override differs from C
148 unsigned int size_lt:1; // type override is larger than C
ddaf8bd7 149 unsigned int had_ds:1; // had ds: prefix
c7ed83dd 150 const struct parsed_proto *pp; // for OPT_LABEL
91977a1c 151 int reg;
152 unsigned int val;
92d715b6 153 char name[NAMELEN];
c36e914d 154};
155
156struct parsed_op {
91977a1c 157 enum op_op op;
158 struct parsed_opr operand[MAX_OPERANDS];
69a3cdfc 159 unsigned int flags;
092f64e1 160 unsigned char pfo;
161 unsigned char pfo_inv;
162 unsigned char operand_cnt;
3a5101d7 163 unsigned char p_argnum; // arg push: altered before call arg #
164 unsigned char p_arggrp; // arg push: arg group # for above
165 unsigned char p_argpass;// arg push: arg of host func
166 short p_argnext;// arg push: same arg pushed elsewhere or -1
69a3cdfc 167 int regmask_src; // all referensed regs
168 int regmask_dst;
940e8e66 169 int pfomask; // flagop: parsed_flag_op that can't be delayed
940e8e66 170 int cc_scratch; // scratch storage during analysis
4c45fa73 171 int bt_i; // branch target for branches
172 struct parsed_data *btj;// branch targets for jumptables
c7ed83dd 173 struct parsed_proto *pp;// parsed_proto for OP_CALL
91977a1c 174 void *datap;
8eb12e72 175 int asmln;
91977a1c 176};
177
69a3cdfc 178// datap:
092f64e1 179// OP_CALL - parser proto hint (str)
2c10ea1f 180// (OPF_CC) - points to one of (OPF_FLAGS) that affects cc op
181// OP_POP - points to OP_PUSH in push/pop pair
69a3cdfc 182
91977a1c 183struct parsed_equ {
184 char name[64];
185 enum opr_lenmod lmod;
186 int offset;
c36e914d 187};
188
4c45fa73 189struct parsed_data {
190 char label[256];
191 enum opr_type type;
192 enum opr_lenmod lmod;
193 int count;
194 int count_alloc;
195 struct {
196 union {
197 char *label;
63df67be 198 unsigned int val;
4c45fa73 199 } u;
200 int bt_i;
201 } *d;
202};
203
204struct label_ref {
205 int i;
206 struct label_ref *next;
207};
208
1bafb621 209enum ida_func_attr {
210 IDAFA_BP_FRAME = (1 << 0),
211 IDAFA_LIB_FUNC = (1 << 1),
212 IDAFA_STATIC = (1 << 2),
213 IDAFA_NORETURN = (1 << 3),
214 IDAFA_THUNK = (1 << 4),
215 IDAFA_FPD = (1 << 5),
216};
217
3a5101d7 218// note: limited to 32k due to p_argnext
219#define MAX_OPS 4096
220#define MAX_ARG_GRP 2
c36e914d 221
222static struct parsed_op ops[MAX_OPS];
91977a1c 223static struct parsed_equ *g_eqs;
224static int g_eqcnt;
d7857c3a 225static char *g_labels[MAX_OPS];
4c45fa73 226static struct label_ref g_label_refs[MAX_OPS];
bd96f656 227static const struct parsed_proto *g_func_pp;
4c45fa73 228static struct parsed_data *g_func_pd;
229static int g_func_pd_cnt;
91977a1c 230static char g_func[256];
231static char g_comment[256];
232static int g_bp_frame;
1bafb621 233static int g_sp_frame;
a2c1d768 234static int g_stack_frame_used;
1bafb621 235static int g_stack_fsz;
4c45fa73 236static int g_ida_func_attr;
89ff3147 237static int g_allow_regfunc;
92d715b6 238static int g_quiet_pp;
9af2d373 239static int g_header_mode;
92d715b6 240
91977a1c 241#define ferr(op_, fmt, ...) do { \
8eb12e72 242 printf("%s:%d: error: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
850c9265 243 dump_op(op_), ##__VA_ARGS__); \
33c35af6 244 fcloseall(); \
91977a1c 245 exit(1); \
246} while (0)
de50b98b 247#define fnote(op_, fmt, ...) \
8eb12e72 248 printf("%s:%d: note: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
de50b98b 249 dump_op(op_), ##__VA_ARGS__)
91977a1c 250
26677139 251#define ferr_assert(op_, cond) do { \
252 if (!(cond)) ferr(op_, "assertion '%s' failed on ln :%d\n", #cond, \
253 __LINE__); \
254} while (0)
255
90307a99 256const char *regs_r32[] = {
257 "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp",
258 // not r32, but list here for easy parsing and printing
259 "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7",
260};
91977a1c 261const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" };
262const char *regs_r8l[] = { "al", "bl", "cl", "dl" };
263const char *regs_r8h[] = { "ah", "bh", "ch", "dh" };
264
265enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP };
266
69a3cdfc 267// possible basic comparison types (without inversion)
268enum parsed_flag_op {
269 PFO_O, // 0 OF=1
270 PFO_C, // 2 CF=1
271 PFO_Z, // 4 ZF=1
272 PFO_BE, // 6 CF=1||ZF=1
273 PFO_S, // 8 SF=1
274 PFO_P, // a PF=1
275 PFO_L, // c SF!=OF
276 PFO_LE, // e ZF=1||SF!=OF
277};
278
90307a99 279#define PFOB_O (1 << PFO_O)
280#define PFOB_C (1 << PFO_C)
281#define PFOB_Z (1 << PFO_Z)
282#define PFOB_S (1 << PFO_S)
283
69a3cdfc 284static const char *parsed_flag_op_names[] = {
285 "o", "c", "z", "be", "s", "p", "l", "le"
286};
287
91977a1c 288static int char_array_i(const char *array[], size_t len, const char *s)
289{
290 int i;
c36e914d 291
91977a1c 292 for (i = 0; i < len; i++)
293 if (IS(s, array[i]))
294 return i;
c36e914d 295
91977a1c 296 return -1;
297}
298
63df67be 299static void printf_number(char *buf, size_t buf_size,
300 unsigned long number)
91977a1c 301{
5101a5f9 302 // output in C-friendly form
303 snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
304}
91977a1c 305
fdd5548a 306static int check_segment_prefix(const char *s)
307{
308 if (s[0] == 0 || s[1] != 's' || s[2] != ':')
309 return 0;
310
311 switch (s[0]) {
312 case 'c': return 1;
313 case 'd': return 2;
314 case 's': return 3;
315 case 'e': return 4;
316 case 'f': return 5;
317 case 'g': return 6;
318 default: return 0;
319 }
320}
321
5101a5f9 322static int parse_reg(enum opr_lenmod *reg_lmod, const char *s)
323{
324 int reg;
91977a1c 325
5101a5f9 326 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s);
90307a99 327 if (reg >= 8) {
328 *reg_lmod = OPLM_QWORD;
329 return reg;
330 }
5101a5f9 331 if (reg >= 0) {
332 *reg_lmod = OPLM_DWORD;
333 return reg;
91977a1c 334 }
5101a5f9 335 reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s);
336 if (reg >= 0) {
337 *reg_lmod = OPLM_WORD;
338 return reg;
339 }
340 reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s);
341 if (reg >= 0) {
342 *reg_lmod = OPLM_BYTE;
343 return reg;
344 }
345 reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s);
346 if (reg >= 0) {
347 *reg_lmod = OPLM_BYTE;
348 return reg;
850c9265 349 }
350
351 return -1;
91977a1c 352}
353
5101a5f9 354static int parse_indmode(char *name, int *regmask, int need_c_cvt)
355{
356 enum opr_lenmod lmod;
357 char cvtbuf[256];
358 char *d = cvtbuf;
359 char *s = name;
360 char w[64];
361 long number;
362 int reg;
363 int c = 0;
364
365 *d = 0;
366
367 while (*s != 0) {
368 d += strlen(d);
369 while (my_isblank(*s))
370 s++;
371 for (; my_issep(*s); d++, s++)
372 *d = *s;
373 while (my_isblank(*s))
374 s++;
375 *d = 0;
376
fdd5548a 377 // skip '?s:' prefixes
378 if (check_segment_prefix(s))
87bf6cec 379 s += 3;
380
5101a5f9 381 s = next_idt(w, sizeof(w), s);
382 if (w[0] == 0)
383 break;
384 c++;
385
386 reg = parse_reg(&lmod, w);
387 if (reg >= 0) {
388 *regmask |= 1 << reg;
389 goto pass;
390 }
391
392 if ('0' <= w[0] && w[0] <= '9') {
393 number = parse_number(w);
394 printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number);
395 continue;
396 }
397
398 // probably some label/identifier - pass
399
400pass:
401 snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w);
402 }
403
404 if (need_c_cvt)
405 strcpy(name, cvtbuf);
406
407 return c;
408}
409
4c45fa73 410static int is_reg_in_str(const char *s)
411{
412 int i;
413
414 if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
415 return 0;
416
417 for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
418 if (!strncmp(s, regs_r32[i], 3))
419 return 1;
420
421 return 0;
422}
423
09bc6fd5 424static const char *parse_stack_el(const char *name, char *extra_reg,
425 int early_try)
d4e3b5db 426{
4c45fa73 427 const char *p, *p2, *s;
1bafb621 428 char *endp = NULL;
d4e3b5db 429 char buf[32];
1bafb621 430 long val;
d4e3b5db 431 int len;
432
09bc6fd5 433 if (g_bp_frame || early_try)
434 {
435 p = name;
436 if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
437 p += 4;
438 if (extra_reg != NULL) {
439 strncpy(extra_reg, name, 3);
440 extra_reg[4] = 0;
441 }
4c45fa73 442 }
4c45fa73 443
09bc6fd5 444 if (IS_START(p, "ebp+")) {
445 p += 4;
4c45fa73 446
09bc6fd5 447 p2 = strchr(p, '+');
448 if (p2 != NULL && is_reg_in_str(p)) {
449 if (extra_reg != NULL) {
450 strncpy(extra_reg, p, p2 - p);
451 extra_reg[p2 - p] = 0;
452 }
453 p = p2 + 1;
4c45fa73 454 }
4c45fa73 455
09bc6fd5 456 if (!('0' <= *p && *p <= '9'))
457 return p;
4c45fa73 458
09bc6fd5 459 return NULL;
460 }
4c45fa73 461 }
462
39b168b8 463 if (!IS_START(name, "esp+"))
d4e3b5db 464 return NULL;
465
037f4971 466 s = name + 4;
467 p = strchr(s, '+');
d4e3b5db 468 if (p) {
037f4971 469 if (is_reg_in_str(s)) {
470 if (extra_reg != NULL) {
471 strncpy(extra_reg, s, p - s);
472 extra_reg[p - s] = 0;
473 }
474 s = p + 1;
475 p = strchr(s, '+');
476 if (p == NULL)
477 aerr("%s IDA stackvar not set?\n", __func__);
478 }
1bafb621 479 if (!('0' <= *s && *s <= '9')) {
037f4971 480 aerr("%s IDA stackvar offset not set?\n", __func__);
d4e3b5db 481 return NULL;
1bafb621 482 }
483 if (s[0] == '0' && s[1] == 'x')
484 s += 2;
485 len = p - s;
d4e3b5db 486 if (len < sizeof(buf) - 1) {
1bafb621 487 strncpy(buf, s, len);
d4e3b5db 488 buf[len] = 0;
1bafb621 489 val = strtol(buf, &endp, 16);
490 if (val == 0 || *endp != 0) {
491 aerr("%s num parse fail for '%s'\n", __func__, buf);
492 return NULL;
493 }
d4e3b5db 494 }
495 p++;
496 }
497 else
498 p = name + 4;
499
500 if ('0' <= *p && *p <= '9')
501 return NULL;
502
503 return p;
504}
505
850c9265 506static int guess_lmod_from_name(struct parsed_opr *opr)
507{
508 if (!strncmp(opr->name, "dword_", 6)) {
509 opr->lmod = OPLM_DWORD;
510 return 1;
511 }
512 if (!strncmp(opr->name, "word_", 5)) {
513 opr->lmod = OPLM_WORD;
514 return 1;
515 }
516 if (!strncmp(opr->name, "byte_", 5)) {
517 opr->lmod = OPLM_BYTE;
518 return 1;
519 }
90307a99 520 if (!strncmp(opr->name, "qword_", 6)) {
521 opr->lmod = OPLM_QWORD;
522 return 1;
523 }
850c9265 524 return 0;
525}
526
3ebea2cf 527static int guess_lmod_from_c_type(enum opr_lenmod *lmod,
528 const struct parsed_type *c_type)
06c5d854 529{
06c5d854 530 static const char *dword_types[] = {
c0de9015 531 "uint32_t", "int", "_DWORD", "UINT_PTR", "DWORD",
4741fdfe 532 "WPARAM", "LPARAM", "UINT", "__int32",
09bc6fd5 533 "LONG", "HIMC", "BOOL", "size_t",
feb0ee5d 534 "float",
06c5d854 535 };
536 static const char *word_types[] = {
fe18709a 537 "uint16_t", "int16_t", "_WORD", "WORD",
2b43685d 538 "unsigned __int16", "__int16",
06c5d854 539 };
540 static const char *byte_types[] = {
2b43685d 541 "uint8_t", "int8_t", "char",
89ff3147 542 "unsigned __int8", "__int8", "BYTE", "_BYTE",
4741fdfe 543 "CHAR", "_UNKNOWN",
b4878d2b 544 // structures.. deal the same as with _UNKNOWN for now
545 "CRITICAL_SECTION",
06c5d854 546 };
3ebea2cf 547 const char *n;
06c5d854 548 int i;
549
3ebea2cf 550 if (c_type->is_ptr) {
551 *lmod = OPLM_DWORD;
06c5d854 552 return 1;
553 }
554
3ebea2cf 555 n = skip_type_mod(c_type->name);
06c5d854 556
3ebea2cf 557 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
558 if (IS(n, dword_types[i])) {
559 *lmod = OPLM_DWORD;
06c5d854 560 return 1;
561 }
562 }
563
564 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
3ebea2cf 565 if (IS(n, word_types[i])) {
566 *lmod = OPLM_WORD;
06c5d854 567 return 1;
568 }
569 }
570
571 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
3ebea2cf 572 if (IS(n, byte_types[i])) {
573 *lmod = OPLM_BYTE;
06c5d854 574 return 1;
575 }
576 }
577
06c5d854 578 return 0;
579}
580
c7ed83dd 581static char *default_cast_to(char *buf, size_t buf_size,
582 struct parsed_opr *opr)
583{
584 buf[0] = 0;
585
586 if (!opr->is_ptr)
587 return buf;
588 if (opr->pp == NULL || opr->pp->type.name == NULL
589 || opr->pp->is_fptr)
590 {
591 snprintf(buf, buf_size, "%s", "(void *)");
592 return buf;
593 }
594
595 snprintf(buf, buf_size, "(%s)", opr->pp->type.name);
596 return buf;
597}
598
4c45fa73 599static enum opr_type lmod_from_directive(const char *d)
600{
601 if (IS(d, "dd"))
602 return OPLM_DWORD;
603 else if (IS(d, "dw"))
604 return OPLM_WORD;
605 else if (IS(d, "db"))
606 return OPLM_BYTE;
607
608 aerr("unhandled directive: '%s'\n", d);
609 return OPLM_UNSPEC;
610}
611
5101a5f9 612static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
613 int *regmask)
614{
615 opr->type = OPT_REG;
616 opr->reg = reg;
617 opr->lmod = lmod;
618 *regmask |= 1 << reg;
619}
620
39b168b8 621static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
622 int *extra_offs);
87bf6cec 623
69a3cdfc 624static int parse_operand(struct parsed_opr *opr,
625 int *regmask, int *regmask_indirect,
1bafb621 626 char words[16][256], int wordc, int w, unsigned int op_flags)
c36e914d 627{
92d715b6 628 const struct parsed_proto *pp = NULL;
850c9265 629 enum opr_lenmod tmplmod;
63df67be 630 unsigned long number;
89ff3147 631 char buf[256];
850c9265 632 int ret, len;
1bafb621 633 int wordc_in;
89ff3147 634 char *p;
850c9265 635 int i;
c36e914d 636
1bafb621 637 if (w >= wordc)
638 aerr("parse_operand w %d, wordc %d\n", w, wordc);
c36e914d 639
1bafb621 640 opr->reg = xUNSPEC;
91977a1c 641
1bafb621 642 for (i = w; i < wordc; i++) {
643 len = strlen(words[i]);
644 if (words[i][len - 1] == ',') {
645 words[i][len - 1] = 0;
646 wordc = i + 1;
647 break;
648 }
649 }
c36e914d 650
1bafb621 651 wordc_in = wordc - w;
c36e914d 652
1bafb621 653 if ((op_flags & OPF_JMP) && wordc_in > 0
654 && !('0' <= words[w][0] && words[w][0] <= '9'))
655 {
656 const char *label = NULL;
657
658 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
659 && IS(words[w + 1], "ptr"))
660 label = words[w + 2];
661 else if (wordc_in == 2 && IS(words[w], "short"))
662 label = words[w + 1];
663 else if (wordc_in == 1
664 && strchr(words[w], '[') == NULL
665 && parse_reg(&tmplmod, words[w]) < 0)
666 label = words[w];
667
668 if (label != NULL) {
669 opr->type = OPT_LABEL;
fdd5548a 670 ret = check_segment_prefix(label);
671 if (ret != 0) {
672 if (ret >= 5)
673 aerr("fs/gs used\n");
ddaf8bd7 674 opr->had_ds = 1;
39b168b8 675 label += 3;
ddaf8bd7 676 }
1bafb621 677 strcpy(opr->name, label);
678 return wordc;
679 }
680 }
c36e914d 681
1bafb621 682 if (wordc_in >= 3) {
683 if (IS(words[w + 1], "ptr")) {
684 if (IS(words[w], "dword"))
685 opr->lmod = OPLM_DWORD;
686 else if (IS(words[w], "word"))
687 opr->lmod = OPLM_WORD;
688 else if (IS(words[w], "byte"))
689 opr->lmod = OPLM_BYTE;
90307a99 690 else if (IS(words[w], "qword"))
691 opr->lmod = OPLM_QWORD;
1bafb621 692 else
693 aerr("type parsing failed\n");
694 w += 2;
695 wordc_in = wordc - w;
696 }
697 }
698
39b168b8 699 if (wordc_in == 2) {
700 if (IS(words[w], "offset")) {
701 opr->type = OPT_OFFSET;
27ebfaed 702 opr->lmod = OPLM_DWORD;
39b168b8 703 strcpy(opr->name, words[w + 1]);
27ebfaed 704 pp = proto_parse(g_fhdr, opr->name, 1);
705 goto do_label;
39b168b8 706 }
707 if (IS(words[w], "(offset")) {
89ff3147 708 p = strchr(words[w + 1], ')');
39b168b8 709 if (p == NULL)
710 aerr("parse of bracketed offset failed\n");
711 *p = 0;
712 opr->type = OPT_OFFSET;
713 strcpy(opr->name, words[w + 1]);
714 return wordc;
715 }
1bafb621 716 }
c36e914d 717
1bafb621 718 if (wordc_in != 1)
850c9265 719 aerr("parse_operand 1 word expected\n");
c36e914d 720
fdd5548a 721 ret = check_segment_prefix(words[w]);
722 if (ret != 0) {
723 if (ret >= 5)
724 aerr("fs/gs used\n");
ddaf8bd7 725 opr->had_ds = 1;
89ff3147 726 memmove(words[w], words[w] + 3, strlen(words[w]) - 2);
ddaf8bd7 727 }
89ff3147 728 strcpy(opr->name, words[w]);
c36e914d 729
850c9265 730 if (words[w][0] == '[') {
731 opr->type = OPT_REGMEM;
732 ret = sscanf(words[w], "[%[^]]]", opr->name);
733 if (ret != 1)
734 aerr("[] parse failure\n");
87bf6cec 735
5101a5f9 736 parse_indmode(opr->name, regmask_indirect, 1);
09bc6fd5 737 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL, 1))
738 {
87bf6cec 739 // might be an equ
d4e3b5db 740 struct parsed_equ *eq =
09bc6fd5 741 equ_find(NULL, parse_stack_el(opr->name, NULL, 1), &i);
87bf6cec 742 if (eq)
743 opr->lmod = eq->lmod;
744 }
850c9265 745 return wordc;
746 }
747 else if (strchr(words[w], '[')) {
748 // label[reg] form
89ff3147 749 p = strchr(words[w], '[');
850c9265 750 opr->type = OPT_REGMEM;
89ff3147 751 parse_indmode(p, regmask_indirect, 0);
752 strncpy(buf, words[w], p - words[w]);
753 buf[p - words[w]] = 0;
754 pp = proto_parse(g_fhdr, buf, 1);
755 goto do_label;
850c9265 756 }
757 else if (('0' <= words[w][0] && words[w][0] <= '9')
758 || words[w][0] == '-')
759 {
5101a5f9 760 number = parse_number(words[w]);
91977a1c 761 opr->type = OPT_CONST;
5101a5f9 762 opr->val = number;
763 printf_number(opr->name, sizeof(opr->name), number);
91977a1c 764 return wordc;
850c9265 765 }
c36e914d 766
5101a5f9 767 ret = parse_reg(&tmplmod, opr->name);
768 if (ret >= 0) {
769 setup_reg_opr(opr, ret, tmplmod, regmask);
850c9265 770 return wordc;
771 }
772
773 // most likely var in data segment
774 opr->type = OPT_LABEL;
92d715b6 775 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
89ff3147 776
777do_label:
bd96f656 778 if (pp != NULL) {
1cd4a663 779 if (pp->is_fptr || pp->is_func) {
06c5d854 780 opr->lmod = OPLM_DWORD;
781 opr->is_ptr = 1;
782 }
2b43685d 783 else {
840257f6 784 tmplmod = OPLM_UNSPEC;
bd96f656 785 if (!guess_lmod_from_c_type(&tmplmod, &pp->type))
2b43685d 786 anote("unhandled C type '%s' for '%s'\n",
bd96f656 787 pp->type.name, opr->name);
2b43685d 788
a3684be1 789 if (opr->lmod == OPLM_UNSPEC) {
2b43685d 790 opr->lmod = tmplmod;
a3684be1 791 opr->type_from_var = 1;
792 }
2b43685d 793 else if (opr->lmod != tmplmod) {
794 opr->size_mismatch = 1;
795 if (tmplmod < opr->lmod)
796 opr->size_lt = 1;
797 }
a652aa9f 798 opr->is_ptr = pp->type.is_ptr;
3ebea2cf 799 }
bd96f656 800 opr->is_array = pp->type.is_array;
06c5d854 801 }
c7ed83dd 802 opr->pp = pp;
06c5d854 803
850c9265 804 if (opr->lmod == OPLM_UNSPEC)
805 guess_lmod_from_name(opr);
850c9265 806 return wordc;
c36e914d 807}
808
33c35af6 809static const struct {
810 const char *name;
811 unsigned int flags;
812} pref_table[] = {
813 { "rep", OPF_REP },
7ba45c34 814 { "repe", OPF_REP|OPF_REPZ },
815 { "repz", OPF_REP|OPF_REPZ },
816 { "repne", OPF_REP|OPF_REPNZ },
817 { "repnz", OPF_REP|OPF_REPNZ },
037f4971 818 { "lock", OPF_LOCK }, // ignored for now..
33c35af6 819};
820
5c024ef7 821#define OPF_CJMP_CC (OPF_JMP|OPF_CJMP|OPF_CC)
822
c36e914d 823static const struct {
850c9265 824 const char *name;
825 enum op_op op;
092f64e1 826 unsigned short minopr;
827 unsigned short maxopr;
69a3cdfc 828 unsigned int flags;
092f64e1 829 unsigned char pfo;
830 unsigned char pfo_inv;
c36e914d 831} op_table[] = {
33c35af6 832 { "nop", OP_NOP, 0, 0, 0 },
69a3cdfc 833 { "push", OP_PUSH, 1, 1, 0 },
834 { "pop", OP_POP, 1, 1, OPF_DATA },
591721d7 835 { "leave",OP_LEAVE, 0, 0, OPF_DATA },
69a3cdfc 836 { "mov" , OP_MOV, 2, 2, OPF_DATA },
837 { "lea", OP_LEA, 2, 2, OPF_DATA },
838 { "movzx",OP_MOVZX, 2, 2, OPF_DATA },
839 { "movsx",OP_MOVSX, 2, 2, OPF_DATA },
108e9fe3 840 { "xchg", OP_XCHG, 2, 2, OPF_DATA },
69a3cdfc 841 { "not", OP_NOT, 1, 1, OPF_DATA },
5101a5f9 842 { "cdq", OP_CDQ, 0, 0, OPF_DATA },
092f64e1 843 { "lodsb",OP_LODS, 0, 0, OPF_DATA },
844 { "lodsw",OP_LODS, 0, 0, OPF_DATA },
845 { "lodsd",OP_LODS, 0, 0, OPF_DATA },
33c35af6 846 { "stosb",OP_STOS, 0, 0, OPF_DATA },
847 { "stosw",OP_STOS, 0, 0, OPF_DATA },
848 { "stosd",OP_STOS, 0, 0, OPF_DATA },
d4e3b5db 849 { "movsb",OP_MOVS, 0, 0, OPF_DATA },
850 { "movsw",OP_MOVS, 0, 0, OPF_DATA },
851 { "movsd",OP_MOVS, 0, 0, OPF_DATA },
7ba45c34 852 { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
853 { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
854 { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
591721d7 855 { "scasb",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
856 { "scasw",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
857 { "scasd",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
858 { "std", OP_STD, 0, 0, OPF_DATA }, // special flag
859 { "cld", OP_CLD, 0, 0, OPF_DATA },
69a3cdfc 860 { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
861 { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
862 { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
863 { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS },
864 { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS },
865 { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
866 { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
867 { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
868 { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
3b2f4044 869 { "shrd", OP_SHRD, 3, 3, OPF_DATA|OPF_FLAGS },
d4e3b5db 870 { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
871 { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
092f64e1 872 { "rcl", OP_RCL, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
873 { "rcr", OP_RCR, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
874 { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
875 { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
1f84f6b3 876 { "bsf", OP_BSF, 2, 2, OPF_DATA|OPF_FLAGS },
69a3cdfc 877 { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
878 { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS },
5101a5f9 879 { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS },
880 { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS },
69a3cdfc 881 { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS },
5101a5f9 882 { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS },
883 { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS },
69a3cdfc 884 { "test", OP_TEST, 2, 2, OPF_FLAGS },
885 { "cmp", OP_CMP, 2, 2, OPF_FLAGS },
a3684be1 886 { "retn", OP_RET, 0, 1, OPF_TAIL },
de50b98b 887 { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS },
69a3cdfc 888 { "jmp", OP_JMP, 1, 1, OPF_JMP },
5c024ef7 889 { "jecxz",OP_JECXZ, 1, 1, OPF_JMP|OPF_CJMP },
092f64e1 890 { "jo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 0 }, // 70 OF=1
891 { "jno", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 1 }, // 71 OF=0
892 { "jc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72 CF=1
893 { "jb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72
894 { "jnc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73 CF=0
895 { "jnb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
896 { "jae", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
897 { "jz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74 ZF=1
898 { "je", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74
899 { "jnz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75 ZF=0
900 { "jne", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75
901 { "jbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76 CF=1||ZF=1
902 { "jna", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76
903 { "ja", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77 CF=0&&ZF=0
904 { "jnbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77
905 { "js", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 0 }, // 78 SF=1
906 { "jns", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 1 }, // 79 SF=0
907 { "jp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a PF=1
908 { "jpe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a
909 { "jnp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b PF=0
910 { "jpo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b
911 { "jl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c SF!=OF
912 { "jnge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c
913 { "jge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d SF=OF
914 { "jnl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d
915 { "jle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e ZF=1||SF!=OF
916 { "jng", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e
917 { "jg", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f ZF=0&&SF=OF
918 { "jnle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f
919 { "seto", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 0 },
920 { "setno", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 1 },
921 { "setc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
922 { "setb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
923 { "setnc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
924 { "setae", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
925 { "setnb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
926 { "setz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
927 { "sete", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
928 { "setnz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
929 { "setne", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
930 { "setbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
931 { "setna", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
932 { "seta", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
933 { "setnbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
934 { "sets", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 0 },
935 { "setns", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 1 },
936 { "setp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
937 { "setpe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
938 { "setnp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
939 { "setpo", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
940 { "setl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
941 { "setnge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
942 { "setge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
943 { "setnl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
944 { "setle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
945 { "setng", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
946 { "setg", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
947 { "setnle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
90307a99 948 // x87
949 // mmx
950 { "emms", OP_EMMS, 0, 0, OPF_DATA },
951 { "movq", OP_MOV, 2, 2, OPF_DATA },
2c10ea1f 952 // must be last
953 { "ud2", OP_UD2 },
69a3cdfc 954};
c36e914d 955
956static void parse_op(struct parsed_op *op, char words[16][256], int wordc)
957{
bfa4a6ee 958 enum opr_lenmod lmod = OPLM_UNSPEC;
33c35af6 959 int prefix_flags = 0;
69a3cdfc 960 int regmask_ind;
961 int regmask;
33c35af6 962 int op_w = 0;
91977a1c 963 int opr = 0;
33c35af6 964 int w = 0;
91977a1c 965 int i;
c36e914d 966
33c35af6 967 for (i = 0; i < ARRAY_SIZE(pref_table); i++) {
968 if (IS(words[w], pref_table[i].name)) {
969 prefix_flags = pref_table[i].flags;
970 break;
971 }
972 }
973
974 if (prefix_flags) {
975 if (wordc <= 1)
976 aerr("lone prefix: '%s'\n", words[0]);
977 w++;
978 }
979
980 op_w = w;
91977a1c 981 for (i = 0; i < ARRAY_SIZE(op_table); i++) {
33c35af6 982 if (IS(words[w], op_table[i].name))
69a3cdfc 983 break;
984 }
c36e914d 985
2c10ea1f 986 if (i == ARRAY_SIZE(op_table)) {
987 anote("unhandled op: '%s'\n", words[0]);
988 i--; // OP_UD2
989 }
33c35af6 990 w++;
c36e914d 991
69a3cdfc 992 op->op = op_table[i].op;
33c35af6 993 op->flags = op_table[i].flags | prefix_flags;
092f64e1 994 op->pfo = op_table[i].pfo;
995 op->pfo_inv = op_table[i].pfo_inv;
69a3cdfc 996 op->regmask_src = op->regmask_dst = 0;
8eb12e72 997 op->asmln = asmln;
69a3cdfc 998
2c10ea1f 999 if (op->op == OP_UD2)
1000 return;
1001
92d715b6 1002 for (opr = 0; opr < op_table[i].maxopr; opr++) {
1003 if (opr >= op_table[i].minopr && w >= wordc)
1004 break;
1005
69a3cdfc 1006 regmask = regmask_ind = 0;
1007 w = parse_operand(&op->operand[opr], &regmask, &regmask_ind,
1008 words, wordc, w, op->flags);
1009
1010 if (opr == 0 && (op->flags & OPF_DATA))
1011 op->regmask_dst = regmask;
92d715b6 1012 else
1013 op->regmask_src |= regmask;
1014 op->regmask_src |= regmask_ind;
91977a1c 1015 }
c36e914d 1016
91977a1c 1017 if (w < wordc)
1018 aerr("parse_op %s incomplete: %d/%d\n",
1019 words[0], w, wordc);
5101a5f9 1020
1021 // special cases
1022 op->operand_cnt = opr;
1023 if (!strncmp(op_table[i].name, "set", 3))
1024 op->operand[0].lmod = OPLM_BYTE;
1025
5101a5f9 1026 switch (op->op) {
92d715b6 1027 // first operand is not dst
1028 case OP_CMP:
1029 case OP_TEST:
1030 op->regmask_src |= op->regmask_dst;
1031 op->regmask_dst = 0;
1032 break;
1033
1034 // first operand is src too
1035 case OP_NOT:
1036 case OP_ADD:
1037 case OP_AND:
1038 case OP_OR:
1039 case OP_RCL:
1040 case OP_RCR:
1041 case OP_ADC:
1042 case OP_INC:
1043 case OP_DEC:
1044 case OP_NEG:
1045 // more below..
1046 op->regmask_src |= op->regmask_dst;
1047 break;
1048
1049 // special
1050 case OP_XCHG:
1051 op->regmask_src |= op->regmask_dst;
1052 op->regmask_dst |= op->regmask_src;
1053 goto check_align;
1054
1055 case OP_SUB:
1056 case OP_SBB:
1057 case OP_XOR:
1058 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1059 && op->operand[0].lmod == op->operand[1].lmod
1060 && op->operand[0].reg == op->operand[1].reg
1061 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
1062 {
1063 op->regmask_src = 0;
1064 }
1065 else
1066 op->regmask_src |= op->regmask_dst;
1067 break;
1068
1069 // ops with implicit argumets
5101a5f9 1070 case OP_CDQ:
1071 op->operand_cnt = 2;
1072 setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst);
1073 setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src);
1074 break;
1075
092f64e1 1076 case OP_LODS:
33c35af6 1077 case OP_STOS:
591721d7 1078 case OP_SCAS:
33c35af6 1079 if (op->operand_cnt != 0)
1080 break;
591721d7 1081 if (words[op_w][4] == 'b')
33c35af6 1082 lmod = OPLM_BYTE;
591721d7 1083 else if (words[op_w][4] == 'w')
33c35af6 1084 lmod = OPLM_WORD;
591721d7 1085 else if (words[op_w][4] == 'd')
33c35af6 1086 lmod = OPLM_DWORD;
1087 op->operand_cnt = 3;
092f64e1 1088 setup_reg_opr(&op->operand[0], op->op == OP_LODS ? xSI : xDI,
1089 lmod, &op->regmask_src);
d4e3b5db 1090 setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src);
1091 op->regmask_dst = op->regmask_src;
092f64e1 1092 setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD,
1093 op->op == OP_LODS ? &op->regmask_dst : &op->regmask_src);
33c35af6 1094 break;
1095
d4e3b5db 1096 case OP_MOVS:
7ba45c34 1097 case OP_CMPS:
d4e3b5db 1098 if (op->operand_cnt != 0)
1099 break;
7ba45c34 1100 if (words[op_w][4] == 'b')
d4e3b5db 1101 lmod = OPLM_BYTE;
7ba45c34 1102 else if (words[op_w][4] == 'w')
d4e3b5db 1103 lmod = OPLM_WORD;
7ba45c34 1104 else if (words[op_w][4] == 'd')
d4e3b5db 1105 lmod = OPLM_DWORD;
1106 op->operand_cnt = 3;
1107 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
1108 setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src);
1109 setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src);
1110 op->regmask_dst = op->regmask_src;
1111 break;
1112
5c024ef7 1113 case OP_JECXZ:
1114 op->operand_cnt = 1;
1115 op->regmask_src = 1 << xCX;
1116 op->operand[0].type = OPT_REG;
1117 op->operand[0].reg = xCX;
1118 op->operand[0].lmod = OPLM_DWORD;
1119 break;
1120
5101a5f9 1121 case OP_IMUL:
1122 if (op->operand_cnt != 1)
1123 break;
1124 // fallthrough
1125 case OP_MUL:
1126 // singleop mul
92d715b6 1127 op->regmask_src |= op->regmask_dst;
5101a5f9 1128 op->regmask_dst = (1 << xDX) | (1 << xAX);
5101a5f9 1129 if (op->operand[0].lmod == OPLM_UNSPEC)
1130 op->operand[0].lmod = OPLM_DWORD;
1131 break;
1132
1133 case OP_DIV:
1134 case OP_IDIV:
1135 // we could set up operands for edx:eax, but there is no real need to
1136 // (see is_opr_modified())
92d715b6 1137 op->regmask_src |= op->regmask_dst;
1138 op->regmask_dst = (1 << xDX) | (1 << xAX);
5101a5f9 1139 if (op->operand[0].lmod == OPLM_UNSPEC)
1140 op->operand[0].lmod = OPLM_DWORD;
1141 break;
1142
1143 case OP_SHL:
1144 case OP_SHR:
1145 case OP_SAR:
d4e3b5db 1146 case OP_ROL:
1147 case OP_ROR:
92d715b6 1148 op->regmask_src |= op->regmask_dst;
5101a5f9 1149 if (op->operand[1].lmod == OPLM_UNSPEC)
1150 op->operand[1].lmod = OPLM_BYTE;
1151 break;
1152
3b2f4044 1153 case OP_SHRD:
92d715b6 1154 op->regmask_src |= op->regmask_dst;
3b2f4044 1155 if (op->operand[2].lmod == OPLM_UNSPEC)
1156 op->operand[2].lmod = OPLM_BYTE;
1157 break;
1158
7ba45c34 1159 case OP_PUSH:
92d715b6 1160 op->regmask_src |= op->regmask_dst;
1161 op->regmask_dst = 0;
7ba45c34 1162 if (op->operand[0].lmod == OPLM_UNSPEC
1163 && (op->operand[0].type == OPT_CONST
1164 || op->operand[0].type == OPT_OFFSET
1165 || op->operand[0].type == OPT_LABEL))
1166 op->operand[0].lmod = OPLM_DWORD;
1167 break;
1168
3ebea2cf 1169 // alignment
1170 case OP_MOV:
092f64e1 1171 check_align:
3ebea2cf 1172 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
092f64e1 1173 && op->operand[0].lmod == op->operand[1].lmod
1174 && op->operand[0].reg == op->operand[1].reg
92d715b6 1175 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
3ebea2cf 1176 {
5e49b270 1177 op->flags |= OPF_RMD | OPF_DONE;
092f64e1 1178 op->regmask_src = op->regmask_dst = 0;
3ebea2cf 1179 }
1180 break;
1181
de50b98b 1182 case OP_LEA:
1183 if (op->operand[0].type == OPT_REG
1184 && op->operand[1].type == OPT_REGMEM)
1185 {
1186 char buf[16];
1187 snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name);
1188 if (IS(buf, op->operand[1].name))
5e49b270 1189 op->flags |= OPF_RMD | OPF_DONE;
de50b98b 1190 }
1191 break;
1192
89ff3147 1193 case OP_CALL:
1194 // trashed regs must be explicitly detected later
1195 op->regmask_dst = 0;
1196 break;
1197
92d715b6 1198 case OP_LEAVE:
1199 op->regmask_dst = (1 << xBP) | (1 << xSP);
1200 op->regmask_src = 1 << xBP;
1201 break;
1202
5101a5f9 1203 default:
1204 break;
1205 }
c36e914d 1206}
1207
092f64e1 1208static const char *op_name(struct parsed_op *po)
850c9265 1209{
092f64e1 1210 static char buf[16];
1211 char *p;
850c9265 1212 int i;
1213
092f64e1 1214 if (po->op == OP_JCC || po->op == OP_SCC) {
1215 p = buf;
1216 *p++ = (po->op == OP_JCC) ? 'j' : 's';
1217 if (po->pfo_inv)
1218 *p++ = 'n';
1219 strcpy(p, parsed_flag_op_names[po->pfo]);
1220 return buf;
1221 }
1222
850c9265 1223 for (i = 0; i < ARRAY_SIZE(op_table); i++)
092f64e1 1224 if (op_table[i].op == po->op)
850c9265 1225 return op_table[i].name;
1226
1227 return "???";
1228}
1229
1230// debug
1231static const char *dump_op(struct parsed_op *po)
1232{
1233 static char out[128];
1234 char *p = out;
1235 int i;
1236
4c45fa73 1237 if (po == NULL)
1238 return "???";
1239
092f64e1 1240 snprintf(out, sizeof(out), "%s", op_name(po));
850c9265 1241 for (i = 0; i < po->operand_cnt; i++) {
1242 p += strlen(p);
1243 if (i > 0)
1244 *p++ = ',';
1245 snprintf(p, sizeof(out) - (p - out),
1246 po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s",
1247 po->operand[i].name);
1248 }
1249
1250 return out;
1251}
1252
4c45fa73 1253static const char *lmod_type_u(struct parsed_op *po,
1254 enum opr_lenmod lmod)
1255{
1256 switch (lmod) {
90307a99 1257 case OPLM_QWORD:
1258 return "u64";
4c45fa73 1259 case OPLM_DWORD:
1260 return "u32";
1261 case OPLM_WORD:
1262 return "u16";
1263 case OPLM_BYTE:
1264 return "u8";
1265 default:
1266 ferr(po, "invalid lmod: %d\n", lmod);
1267 return "(_invalid_)";
1268 }
1269}
1270
d4e3b5db 1271static const char *lmod_cast_u(struct parsed_op *po,
1272 enum opr_lenmod lmod)
1273{
1274 switch (lmod) {
90307a99 1275 case OPLM_QWORD:
1276 return "";
d4e3b5db 1277 case OPLM_DWORD:
1278 return "";
1279 case OPLM_WORD:
1280 return "(u16)";
1281 case OPLM_BYTE:
1282 return "(u8)";
1283 default:
1284 ferr(po, "invalid lmod: %d\n", lmod);
1285 return "(_invalid_)";
1286 }
1287}
1288
1289static const char *lmod_cast_u_ptr(struct parsed_op *po,
1290 enum opr_lenmod lmod)
1291{
1292 switch (lmod) {
90307a99 1293 case OPLM_QWORD:
1294 return "*(u64 *)";
d4e3b5db 1295 case OPLM_DWORD:
1296 return "*(u32 *)";
1297 case OPLM_WORD:
1298 return "*(u16 *)";
1299 case OPLM_BYTE:
1300 return "*(u8 *)";
1301 default:
1302 ferr(po, "invalid lmod: %d\n", lmod);
1303 return "(_invalid_)";
1304 }
1305}
1306
1307static const char *lmod_cast_s(struct parsed_op *po,
1308 enum opr_lenmod lmod)
1309{
1310 switch (lmod) {
90307a99 1311 case OPLM_QWORD:
1312 return "(s64)";
d4e3b5db 1313 case OPLM_DWORD:
1314 return "(s32)";
1315 case OPLM_WORD:
1316 return "(s16)";
1317 case OPLM_BYTE:
1318 return "(s8)";
1319 default:
1320 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1321 return "(_invalid_)";
1322 }
1323}
1324
1325static const char *lmod_cast(struct parsed_op *po,
1326 enum opr_lenmod lmod, int is_signed)
1327{
1328 return is_signed ?
1329 lmod_cast_s(po, lmod) :
1330 lmod_cast_u(po, lmod);
1331}
1332
1333static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
1334{
1335 switch (lmod) {
90307a99 1336 case OPLM_QWORD:
1337 return 8;
d4e3b5db 1338 case OPLM_DWORD:
1339 return 4;
1340 case OPLM_WORD:
1341 return 2;
1342 case OPLM_BYTE:
1343 return 1;
1344 default:
1345 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1346 return 0;
1347 }
1348}
1349
91977a1c 1350static const char *opr_name(struct parsed_op *po, int opr_num)
c36e914d 1351{
91977a1c 1352 if (opr_num >= po->operand_cnt)
1353 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1354 return po->operand[opr_num].name;
c36e914d 1355}
1356
91977a1c 1357static unsigned int opr_const(struct parsed_op *po, int opr_num)
c36e914d 1358{
91977a1c 1359 if (opr_num >= po->operand_cnt)
1360 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1361 if (po->operand[opr_num].type != OPT_CONST)
1362 ferr(po, "opr %d: const expected\n", opr_num);
1363 return po->operand[opr_num].val;
1364}
c36e914d 1365
91977a1c 1366static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr)
1367{
90307a99 1368 if ((unsigned int)popr->reg >= ARRAY_SIZE(regs_r32))
91977a1c 1369 ferr(po, "invalid reg: %d\n", popr->reg);
1370 return regs_r32[popr->reg];
1371}
c36e914d 1372
a2c1d768 1373// cast1 is the "final" cast
1374static const char *simplify_cast(const char *cast1, const char *cast2)
1375{
1376 static char buf[256];
1377
1378 if (cast1[0] == 0)
1379 return cast2;
1380 if (cast2[0] == 0)
1381 return cast1;
1382 if (IS(cast1, cast2))
1383 return cast1;
1384 if (IS(cast1, "(s8)") && IS(cast2, "(u8)"))
1385 return cast1;
1386 if (IS(cast1, "(s16)") && IS(cast2, "(u16)"))
1387 return cast1;
1388 if (IS(cast1, "(u8)") && IS_START(cast2, "*(u8 *)"))
1389 return cast2;
1390 if (IS(cast1, "(u16)") && IS_START(cast2, "*(u16 *)"))
1391 return cast2;
1cd4a663 1392 if (strchr(cast1, '*') && IS_START(cast2, "(u32)"))
1393 return cast1;
a2c1d768 1394
1395 snprintf(buf, sizeof(buf), "%s%s", cast1, cast2);
1396 return buf;
1397}
1398
1399static const char *simplify_cast_num(const char *cast, unsigned int val)
1400{
1401 if (IS(cast, "(u8)") && val < 0x100)
1402 return "";
1403 if (IS(cast, "(s8)") && val < 0x80)
1404 return "";
1405 if (IS(cast, "(u16)") && val < 0x10000)
1406 return "";
1407 if (IS(cast, "(s16)") && val < 0x8000)
1408 return "";
1409 if (IS(cast, "(s32)") && val < 0x80000000)
1410 return "";
1411
1412 return cast;
1413}
1414
39b168b8 1415static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
1416 int *extra_offs)
91977a1c 1417{
39b168b8 1418 const char *p;
1419 char *endp;
1420 int namelen;
850c9265 1421 int i;
1422
39b168b8 1423 *extra_offs = 0;
1424 namelen = strlen(name);
1425
1426 p = strchr(name, '+');
1427 if (p != NULL) {
1428 namelen = p - name;
1429 if (namelen <= 0)
1430 ferr(po, "equ parse failed for '%s'\n", name);
1431
1432 if (IS_START(p, "0x"))
1433 p += 2;
1434 *extra_offs = strtol(p, &endp, 16);
1435 if (*endp != 0)
1436 ferr(po, "equ parse failed for '%s'\n", name);
1437 }
1438
850c9265 1439 for (i = 0; i < g_eqcnt; i++)
39b168b8 1440 if (strncmp(g_eqs[i].name, name, namelen) == 0
1441 && g_eqs[i].name[namelen] == 0)
850c9265 1442 break;
87bf6cec 1443 if (i >= g_eqcnt) {
1444 if (po != NULL)
1445 ferr(po, "unresolved equ name: '%s'\n", name);
1446 return NULL;
1447 }
850c9265 1448
1449 return &g_eqs[i];
1450}
1451
1cd4a663 1452static int is_stack_access(struct parsed_op *po,
1453 const struct parsed_opr *popr)
850c9265 1454{
09bc6fd5 1455 return (parse_stack_el(popr->name, NULL, 0)
1cd4a663 1456 || (g_bp_frame && !(po->flags & OPF_EBP_S)
1457 && IS_START(popr->name, "ebp")));
1458}
1459
1460static void parse_stack_access(struct parsed_op *po,
1461 const char *name, char *ofs_reg, int *offset_out,
037f4971 1462 int *stack_ra_out, const char **bp_arg_out, int is_lea)
1cd4a663 1463{
1464 const char *bp_arg = "";
1465 const char *p = NULL;
91977a1c 1466 struct parsed_equ *eq;
4f12f671 1467 char *endp = NULL;
d4e3b5db 1468 int stack_ra = 0;
39b168b8 1469 int offset = 0;
91977a1c 1470
1cd4a663 1471 ofs_reg[0] = 0;
a3684be1 1472
1473 if (IS_START(name, "ebp-")
1474 || (IS_START(name, "ebp+") && '0' <= name[4] && name[4] <= '9'))
1475 {
1476 p = name + 4;
1477 if (IS_START(p, "0x"))
1478 p += 2;
1479 offset = strtoul(p, &endp, 16);
1480 if (name[3] == '-')
1481 offset = -offset;
1482 if (*endp != 0)
1483 ferr(po, "ebp- parse of '%s' failed\n", name);
1484 }
1485 else {
09bc6fd5 1486 bp_arg = parse_stack_el(name, ofs_reg, 0);
4f12f671 1487 snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
1488 eq = equ_find(po, bp_arg, &offset);
1489 if (eq == NULL)
1490 ferr(po, "detected but missing eq\n");
1491 offset += eq->offset;
1492 }
39b168b8 1493
d4e3b5db 1494 if (!strncmp(name, "ebp", 3))
1495 stack_ra = 4;
1496
037f4971 1497 // yes it sometimes LEAs ra for compares..
1498 if (!is_lea && ofs_reg[0] == 0
1499 && stack_ra <= offset && offset < stack_ra + 4)
1500 {
39b168b8 1501 ferr(po, "reference to ra? %d %d\n", offset, stack_ra);
037f4971 1502 }
d4e3b5db 1503
1cd4a663 1504 *offset_out = offset;
1505 *stack_ra_out = stack_ra;
1506 if (bp_arg_out)
1507 *bp_arg_out = bp_arg;
1508}
1509
5f70a34f 1510static int stack_frame_access(struct parsed_op *po,
1cd4a663 1511 struct parsed_opr *popr, char *buf, size_t buf_size,
1512 const char *name, const char *cast, int is_src, int is_lea)
1513{
1514 enum opr_lenmod tmp_lmod = OPLM_UNSPEC;
1515 const char *prefix = "";
1516 const char *bp_arg = NULL;
1517 char ofs_reg[16] = { 0, };
1518 int i, arg_i, arg_s;
1519 int unaligned = 0;
1520 int stack_ra = 0;
1521 int offset = 0;
5f70a34f 1522 int retval = -1;
1cd4a663 1523 int sf_ofs;
1524 int lim;
1525
1526 if (po->flags & OPF_EBP_S)
1527 ferr(po, "stack_frame_access while ebp is scratch\n");
1528
037f4971 1529 parse_stack_access(po, name, ofs_reg, &offset,
1530 &stack_ra, &bp_arg, is_lea);
1cd4a663 1531
4f12f671 1532 if (offset > stack_ra)
1533 {
39b168b8 1534 arg_i = (offset - stack_ra - 4) / 4;
bd96f656 1535 if (arg_i < 0 || arg_i >= g_func_pp->argc_stack)
4f12f671 1536 {
bd96f656 1537 if (g_func_pp->is_vararg
1538 && arg_i == g_func_pp->argc_stack && is_lea)
1539 {
4f12f671 1540 // should be va_list
1541 if (cast[0] == 0)
1542 cast = "(u32)";
1543 snprintf(buf, buf_size, "%sap", cast);
5f70a34f 1544 return -1;
4f12f671 1545 }
1546 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1547 offset, bp_arg, arg_i);
1548 }
4c45fa73 1549 if (ofs_reg[0] != 0)
7ba45c34 1550 ferr(po, "offset reg on arg access?\n");
91977a1c 1551
bd96f656 1552 for (i = arg_s = 0; i < g_func_pp->argc; i++) {
1553 if (g_func_pp->arg[i].reg != NULL)
91977a1c 1554 continue;
1555 if (arg_s == arg_i)
1556 break;
1557 arg_s++;
1558 }
bd96f656 1559 if (i == g_func_pp->argc)
91977a1c 1560 ferr(po, "arg %d not in prototype?\n", arg_i);
850c9265 1561
bd96f656 1562 popr->is_ptr = g_func_pp->arg[i].type.is_ptr;
5f70a34f 1563 retval = i;
de50b98b 1564
1565 switch (popr->lmod)
7ba45c34 1566 {
1567 case OPLM_BYTE:
4f12f671 1568 if (is_lea)
1569 ferr(po, "lea/byte to arg?\n");
7ba45c34 1570 if (is_src && (offset & 3) == 0)
a2c1d768 1571 snprintf(buf, buf_size, "%sa%d",
1572 simplify_cast(cast, "(u8)"), i + 1);
7ba45c34 1573 else
a2c1d768 1574 snprintf(buf, buf_size, "%sBYTE%d(a%d)",
1575 cast, offset & 3, i + 1);
7ba45c34 1576 break;
1577
1578 case OPLM_WORD:
4f12f671 1579 if (is_lea)
1580 ferr(po, "lea/word to arg?\n");
a3684be1 1581 if (offset & 1) {
1582 unaligned = 1;
1583 if (!is_src) {
1584 if (offset & 2)
1585 ferr(po, "problematic arg store\n");
a2c1d768 1586 snprintf(buf, buf_size, "%s((char *)&a%d + 1)",
1587 simplify_cast(cast, "*(u16 *)"), i + 1);
a3684be1 1588 }
1589 else
1590 ferr(po, "unaligned arg word load\n");
1591 }
1592 else if (is_src && (offset & 2) == 0)
a2c1d768 1593 snprintf(buf, buf_size, "%sa%d",
1594 simplify_cast(cast, "(u16)"), i + 1);
7ba45c34 1595 else
a2c1d768 1596 snprintf(buf, buf_size, "%s%sWORD(a%d)",
1597 cast, (offset & 2) ? "HI" : "LO", i + 1);
7ba45c34 1598 break;
1599
1600 case OPLM_DWORD:
3ebea2cf 1601 if (cast[0])
1602 prefix = cast;
1603 else if (is_src)
1604 prefix = "(u32)";
a3684be1 1605
de50b98b 1606 if (offset & 3) {
a3684be1 1607 unaligned = 1;
2b43685d 1608 if (is_lea)
1609 snprintf(buf, buf_size, "(u32)&a%d + %d",
1610 i + 1, offset & 3);
a3684be1 1611 else if (!is_src)
1612 ferr(po, "unaligned arg store\n");
1613 else {
1614 // mov edx, [ebp+arg_4+2]; movsx ecx, dx
2b43685d 1615 snprintf(buf, buf_size, "%s(a%d >> %d)",
1616 prefix, i + 1, (offset & 3) * 8);
a3684be1 1617 }
de50b98b 1618 }
1619 else {
1620 snprintf(buf, buf_size, "%s%sa%d",
1621 prefix, is_lea ? "&" : "", i + 1);
1622 }
7ba45c34 1623 break;
1624
1625 default:
de50b98b 1626 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
7ba45c34 1627 }
1628
a3684be1 1629 if (unaligned)
1630 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1631
7ba45c34 1632 // common problem
bd96f656 1633 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type);
a2c1d768 1634 if (tmp_lmod != OPLM_DWORD
de041e5b 1635 && (unaligned || (!is_src && lmod_bytes(po, tmp_lmod)
1636 < lmod_bytes(po, popr->lmod) + (offset & 3))))
a2c1d768 1637 {
1638 ferr(po, "bp_arg arg%d/w offset %d and type '%s' is too small\n",
1639 i + 1, offset, g_func_pp->arg[i].type.name);
1640 }
4741fdfe 1641 // can't check this because msvc likes to reuse
1642 // arg space for scratch..
1643 //if (popr->is_ptr && popr->lmod != OPLM_DWORD)
1644 // ferr(po, "bp_arg arg%d: non-dword ptr access\n", i + 1);
91977a1c 1645 }
4f12f671 1646 else
1647 {
1bafb621 1648 if (g_stack_fsz == 0)
1649 ferr(po, "stack var access without stackframe\n");
a2c1d768 1650 g_stack_frame_used = 1;
850c9265 1651
39b168b8 1652 sf_ofs = g_stack_fsz + offset;
de50b98b 1653 lim = (ofs_reg[0] != 0) ? -4 : 0;
1654 if (offset > 0 || sf_ofs < lim)
39b168b8 1655 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
850c9265 1656
1657 if (is_lea)
33c35af6 1658 prefix = "(u32)&";
3ebea2cf 1659 else
1660 prefix = cast;
850c9265 1661
de50b98b 1662 switch (popr->lmod)
850c9265 1663 {
1664 case OPLM_BYTE:
7ba45c34 1665 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1666 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
850c9265 1667 break;
7ba45c34 1668
850c9265 1669 case OPLM_WORD:
7ba45c34 1670 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1671 // known unaligned or possibly unaligned
1672 strcat(g_comment, " unaligned");
1673 if (prefix[0] == 0)
1674 prefix = "*(u16 *)&";
1675 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1676 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1677 break;
1678 }
1679 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
850c9265 1680 break;
7ba45c34 1681
850c9265 1682 case OPLM_DWORD:
7ba45c34 1683 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1684 // known unaligned or possibly unaligned
1685 strcat(g_comment, " unaligned");
1686 if (prefix[0] == 0)
1687 prefix = "*(u32 *)&";
1688 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1689 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1690 break;
1691 }
1692 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
850c9265 1693 break;
7ba45c34 1694
850c9265 1695 default:
de50b98b 1696 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
850c9265 1697 }
91977a1c 1698 }
5f70a34f 1699
1700 return retval;
91977a1c 1701}
c36e914d 1702
89ff3147 1703static void check_func_pp(struct parsed_op *po,
1704 const struct parsed_proto *pp, const char *pfx)
1705{
179b79a9 1706 enum opr_lenmod tmp_lmod;
b74c31e3 1707 char buf[256];
179b79a9 1708 int ret, i;
b74c31e3 1709
89ff3147 1710 if (pp->argc_reg != 0) {
b74c31e3 1711 if (/*!g_allow_regfunc &&*/ !pp->is_fastcall) {
1712 pp_print(buf, sizeof(buf), pp);
1713 ferr(po, "%s: unexpected reg arg in icall: %s\n", pfx, buf);
1714 }
89ff3147 1715 if (pp->argc_stack > 0 && pp->argc_reg != 2)
1716 ferr(po, "%s: %d reg arg(s) with %d stack arg(s)\n",
1717 pfx, pp->argc_reg, pp->argc_stack);
1718 }
179b79a9 1719
1720 // fptrs must use 32bit args, callsite might have no information and
1721 // lack a cast to smaller types, which results in incorrectly masked
1722 // args passed (callee may assume masked args, it does on ARM)
61e29183 1723 if (!pp->is_osinc) {
179b79a9 1724 for (i = 0; i < pp->argc; i++) {
1725 ret = guess_lmod_from_c_type(&tmp_lmod, &pp->arg[i].type);
1726 if (ret && tmp_lmod != OPLM_DWORD)
1727 ferr(po, "reference to %s with arg%d '%s'\n", pp->name,
1728 i + 1, pp->arg[i].type.name);
1729 }
1730 }
89ff3147 1731}
1732
7aca4698 1733static const char *check_label_read_ref(struct parsed_op *po,
1734 const char *name)
3ebea2cf 1735{
840257f6 1736 const struct parsed_proto *pp;
1737
36595fd2 1738 pp = proto_parse(g_fhdr, name, 0);
840257f6 1739 if (pp == NULL)
1740 ferr(po, "proto_parse failed for ref '%s'\n", name);
1741
89ff3147 1742 if (pp->is_func)
1743 check_func_pp(po, pp, "ref");
7aca4698 1744
1745 return pp->name;
3ebea2cf 1746}
1747
91977a1c 1748static char *out_src_opr(char *buf, size_t buf_size,
591721d7 1749 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
3ebea2cf 1750 int is_lea)
91977a1c 1751{
850c9265 1752 char tmp1[256], tmp2[256];
1753 char expr[256];
7aca4698 1754 const char *name;
a2c1d768 1755 char *p;
850c9265 1756 int ret;
1757
3ebea2cf 1758 if (cast == NULL)
1759 cast = "";
1760
91977a1c 1761 switch (popr->type) {
1762 case OPT_REG:
850c9265 1763 if (is_lea)
1764 ferr(po, "lea from reg?\n");
1765
91977a1c 1766 switch (popr->lmod) {
90307a99 1767 case OPLM_QWORD:
1768 snprintf(buf, buf_size, "%s%s.q", cast, opr_reg_p(po, popr));
1769 break;
91977a1c 1770 case OPLM_DWORD:
3ebea2cf 1771 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
91977a1c 1772 break;
850c9265 1773 case OPLM_WORD:
a2c1d768 1774 snprintf(buf, buf_size, "%s%s",
1775 simplify_cast(cast, "(u16)"), opr_reg_p(po, popr));
850c9265 1776 break;
1777 case OPLM_BYTE:
5101a5f9 1778 if (popr->name[1] == 'h') // XXX..
a2c1d768 1779 snprintf(buf, buf_size, "%s(%s >> 8)",
1780 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
5101a5f9 1781 else
a2c1d768 1782 snprintf(buf, buf_size, "%s%s",
1783 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
850c9265 1784 break;
91977a1c 1785 default:
1786 ferr(po, "invalid src lmod: %d\n", popr->lmod);
1787 }
1788 break;
850c9265 1789
91977a1c 1790 case OPT_REGMEM:
1cd4a663 1791 if (is_stack_access(po, popr)) {
de50b98b 1792 stack_frame_access(po, popr, buf, buf_size,
3ebea2cf 1793 popr->name, cast, 1, is_lea);
91977a1c 1794 break;
1795 }
850c9265 1796
1797 strcpy(expr, popr->name);
1798 if (strchr(expr, '[')) {
1799 // special case: '[' can only be left for label[reg] form
1800 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
1801 if (ret != 2)
1802 ferr(po, "parse failure for '%s'\n", expr);
a2c1d768 1803 if (tmp1[0] == '(') {
1804 // (off_4FFF50+3)[eax]
1805 p = strchr(tmp1 + 1, ')');
1806 if (p == NULL || p[1] != 0)
1807 ferr(po, "parse failure (2) for '%s'\n", expr);
1808 *p = 0;
1809 memmove(tmp1, tmp1 + 1, strlen(tmp1));
1810 }
33c35af6 1811 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
850c9265 1812 }
1813
1814 // XXX: do we need more parsing?
1815 if (is_lea) {
1816 snprintf(buf, buf_size, "%s", expr);
1817 break;
1818 }
1819
a2c1d768 1820 snprintf(buf, buf_size, "%s(%s)",
1821 simplify_cast(cast, lmod_cast_u_ptr(po, popr->lmod)), expr);
91977a1c 1822 break;
850c9265 1823
91977a1c 1824 case OPT_LABEL:
7aca4698 1825 name = check_label_read_ref(po, popr->name);
3ebea2cf 1826 if (cast[0] == 0 && popr->is_ptr)
1827 cast = "(u32)";
2b43685d 1828
850c9265 1829 if (is_lea)
7aca4698 1830 snprintf(buf, buf_size, "(u32)&%s", name);
2b43685d 1831 else if (popr->size_lt)
1832 snprintf(buf, buf_size, "%s%s%s%s", cast,
1833 lmod_cast_u_ptr(po, popr->lmod),
7aca4698 1834 popr->is_array ? "" : "&", name);
850c9265 1835 else
7aca4698 1836 snprintf(buf, buf_size, "%s%s%s", cast, name,
7ba45c34 1837 popr->is_array ? "[0]" : "");
850c9265 1838 break;
1839
1840 case OPT_OFFSET:
7aca4698 1841 name = check_label_read_ref(po, popr->name);
3ebea2cf 1842 if (cast[0] == 0)
1843 cast = "(u32)";
850c9265 1844 if (is_lea)
1845 ferr(po, "lea an offset?\n");
7aca4698 1846 snprintf(buf, buf_size, "%s&%s", cast, name);
91977a1c 1847 break;
850c9265 1848
91977a1c 1849 case OPT_CONST:
850c9265 1850 if (is_lea)
1851 ferr(po, "lea from const?\n");
1852
a2c1d768 1853 printf_number(tmp1, sizeof(tmp1), popr->val);
ddaf8bd7 1854 if (popr->val == 0 && strchr(cast, '*'))
1855 snprintf(buf, buf_size, "NULL");
1856 else
1857 snprintf(buf, buf_size, "%s%s",
1858 simplify_cast_num(cast, popr->val), tmp1);
91977a1c 1859 break;
850c9265 1860
91977a1c 1861 default:
1862 ferr(po, "invalid src type: %d\n", popr->type);
1863 }
1864
1865 return buf;
1866}
c36e914d 1867
de50b98b 1868// note: may set is_ptr (we find that out late for ebp frame..)
91977a1c 1869static char *out_dst_opr(char *buf, size_t buf_size,
1870 struct parsed_op *po, struct parsed_opr *popr)
1871{
1872 switch (popr->type) {
1873 case OPT_REG:
1874 switch (popr->lmod) {
90307a99 1875 case OPLM_QWORD:
1876 snprintf(buf, buf_size, "%s.q", opr_reg_p(po, popr));
1877 break;
91977a1c 1878 case OPLM_DWORD:
1879 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
1880 break;
850c9265 1881 case OPLM_WORD:
1882 // ugh..
1883 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
1884 break;
1885 case OPLM_BYTE:
1886 // ugh..
5101a5f9 1887 if (popr->name[1] == 'h') // XXX..
1888 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
1889 else
1890 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
850c9265 1891 break;
91977a1c 1892 default:
1893 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
1894 }
1895 break;
850c9265 1896
1897 case OPT_REGMEM:
1cd4a663 1898 if (is_stack_access(po, popr)) {
de50b98b 1899 stack_frame_access(po, popr, buf, buf_size,
3ebea2cf 1900 popr->name, "", 0, 0);
850c9265 1901 break;
1902 }
1903
3ebea2cf 1904 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
850c9265 1905
bfa4a6ee 1906 case OPT_LABEL:
2b43685d 1907 if (popr->size_mismatch)
1908 snprintf(buf, buf_size, "%s%s%s",
1909 lmod_cast_u_ptr(po, popr->lmod),
1910 popr->is_array ? "" : "&", popr->name);
1911 else
1912 snprintf(buf, buf_size, "%s%s", popr->name,
1913 popr->is_array ? "[0]" : "");
bfa4a6ee 1914 break;
1915
91977a1c 1916 default:
1917 ferr(po, "invalid dst type: %d\n", popr->type);
1918 }
1919
1920 return buf;
1921}
c36e914d 1922
3ebea2cf 1923static char *out_src_opr_u32(char *buf, size_t buf_size,
1924 struct parsed_op *po, struct parsed_opr *popr)
1925{
1926 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1927}
1928
91977a1c 1929static void out_test_for_cc(char *buf, size_t buf_size,
940e8e66 1930 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
69a3cdfc 1931 enum opr_lenmod lmod, const char *expr)
91977a1c 1932{
69a3cdfc 1933 const char *cast, *scast;
91977a1c 1934
69a3cdfc 1935 cast = lmod_cast_u(po, lmod);
1936 scast = lmod_cast_s(po, lmod);
1937
1938 switch (pfo) {
1939 case PFO_Z:
87bf6cec 1940 case PFO_BE: // CF=1||ZF=1; CF=0
850c9265 1941 snprintf(buf, buf_size, "(%s%s %s 0)",
940e8e66 1942 cast, expr, is_inv ? "!=" : "==");
91977a1c 1943 break;
850c9265 1944
5101a5f9 1945 case PFO_S:
1946 case PFO_L: // SF!=OF; OF=0
1947 snprintf(buf, buf_size, "(%s%s %s 0)",
940e8e66 1948 scast, expr, is_inv ? ">=" : "<");
5101a5f9 1949 break;
1950
87bf6cec 1951 case PFO_LE: // ZF=1||SF!=OF; OF=0
69a3cdfc 1952 snprintf(buf, buf_size, "(%s%s %s 0)",
940e8e66 1953 scast, expr, is_inv ? ">" : "<=");
850c9265 1954 break;
1955
91977a1c 1956 default:
69a3cdfc 1957 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
91977a1c 1958 }
1959}
c36e914d 1960
850c9265 1961static void out_cmp_for_cc(char *buf, size_t buf_size,
a2c1d768 1962 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
850c9265 1963{
a2c1d768 1964 const char *cast, *scast, *cast_use;
1965 char buf1[256], buf2[256];
1966 enum opr_lenmod lmod;
1967
90307a99 1968 if (po->op != OP_DEC && po->operand[0].lmod != po->operand[1].lmod)
a2c1d768 1969 ferr(po, "%s: lmod mismatch: %d %d\n", __func__,
1970 po->operand[0].lmod, po->operand[1].lmod);
1971 lmod = po->operand[0].lmod;
850c9265 1972
69a3cdfc 1973 cast = lmod_cast_u(po, lmod);
1974 scast = lmod_cast_s(po, lmod);
850c9265 1975
a2c1d768 1976 switch (pfo) {
1977 case PFO_C:
1978 case PFO_Z:
1979 case PFO_BE: // !a
1980 cast_use = cast;
1981 break;
1982
1983 case PFO_S:
1984 case PFO_L: // !ge
1985 case PFO_LE:
1986 cast_use = scast;
1987 break;
1988
1989 default:
1990 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1991 }
1992
1993 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], cast_use, 0);
90307a99 1994 if (po->op == OP_DEC)
1995 snprintf(buf2, sizeof(buf2), "1");
1996 else
1997 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], cast_use, 0);
a2c1d768 1998
69a3cdfc 1999 switch (pfo) {
5101a5f9 2000 case PFO_C:
2001 // note: must be unsigned compare
a2c1d768 2002 snprintf(buf, buf_size, "(%s %s %s)",
2003 buf1, is_inv ? ">=" : "<", buf2);
5101a5f9 2004 break;
2005
69a3cdfc 2006 case PFO_Z:
a2c1d768 2007 snprintf(buf, buf_size, "(%s %s %s)",
2008 buf1, is_inv ? "!=" : "==", buf2);
850c9265 2009 break;
2010
5101a5f9 2011 case PFO_BE: // !a
850c9265 2012 // note: must be unsigned compare
a2c1d768 2013 snprintf(buf, buf_size, "(%s %s %s)",
2014 buf1, is_inv ? ">" : "<=", buf2);
2015
2016 // annoying case
2017 if (is_inv && lmod == OPLM_BYTE
2018 && po->operand[1].type == OPT_CONST
2019 && po->operand[1].val == 0xff)
2020 {
2021 snprintf(g_comment, sizeof(g_comment), "if %s", buf);
2022 snprintf(buf, buf_size, "(0)");
2023 }
5101a5f9 2024 break;
2025
2026 // note: must be signed compare
2027 case PFO_S:
2028 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
a2c1d768 2029 scast, buf1, buf2, is_inv ? ">=" : "<");
850c9265 2030 break;
2031
5101a5f9 2032 case PFO_L: // !ge
a2c1d768 2033 snprintf(buf, buf_size, "(%s %s %s)",
2034 buf1, is_inv ? ">=" : "<", buf2);
850c9265 2035 break;
2036
90307a99 2037 case PFO_LE: // !g
a2c1d768 2038 snprintf(buf, buf_size, "(%s %s %s)",
2039 buf1, is_inv ? ">" : "<=", buf2);
5101a5f9 2040 break;
2041
850c9265 2042 default:
a2c1d768 2043 break;
850c9265 2044 }
2045}
2046
69a3cdfc 2047static void out_cmp_test(char *buf, size_t buf_size,
940e8e66 2048 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
69a3cdfc 2049{
2050 char buf1[256], buf2[256], buf3[256];
2051
2052 if (po->op == OP_TEST) {
2053 if (IS(opr_name(po, 0), opr_name(po, 1))) {
3ebea2cf 2054 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
69a3cdfc 2055 }
2056 else {
3ebea2cf 2057 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
2058 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
69a3cdfc 2059 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
2060 }
940e8e66 2061 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
69a3cdfc 2062 po->operand[0].lmod, buf3);
2063 }
2064 else if (po->op == OP_CMP) {
a2c1d768 2065 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv);
69a3cdfc 2066 }
2067 else
2068 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
2069}
2070
850c9265 2071static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
91977a1c 2072 struct parsed_opr *popr2)
2073{
2074 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
2075 ferr(po, "missing lmod for both operands\n");
2076
2077 if (popr1->lmod == OPLM_UNSPEC)
2078 popr1->lmod = popr2->lmod;
2079 else if (popr2->lmod == OPLM_UNSPEC)
2080 popr2->lmod = popr1->lmod;
a3684be1 2081 else if (popr1->lmod != popr2->lmod) {
2082 if (popr1->type_from_var) {
2083 popr1->size_mismatch = 1;
2084 if (popr1->lmod < popr2->lmod)
2085 popr1->size_lt = 1;
2086 popr1->lmod = popr2->lmod;
2087 }
2088 else if (popr2->type_from_var) {
2089 popr2->size_mismatch = 1;
2090 if (popr2->lmod < popr1->lmod)
2091 popr2->size_lt = 1;
2092 popr2->lmod = popr1->lmod;
2093 }
2094 else
2095 ferr(po, "conflicting lmods: %d vs %d\n",
2096 popr1->lmod, popr2->lmod);
2097 }
91977a1c 2098}
c36e914d 2099
850c9265 2100static const char *op_to_c(struct parsed_op *po)
2101{
2102 switch (po->op)
2103 {
2104 case OP_ADD:
5101a5f9 2105 case OP_ADC:
850c9265 2106 return "+";
2107 case OP_SUB:
5101a5f9 2108 case OP_SBB:
850c9265 2109 return "-";
2110 case OP_AND:
2111 return "&";
2112 case OP_OR:
2113 return "|";
2114 case OP_XOR:
2115 return "^";
2116 case OP_SHL:
2117 return "<<";
2118 case OP_SHR:
2119 return ">>";
2120 case OP_MUL:
2121 case OP_IMUL:
2122 return "*";
2123 default:
2124 ferr(po, "op_to_c was supplied with %d\n", po->op);
2125 }
2126}
2127
7ae48d73 2128static void op_set_clear_flag(struct parsed_op *po,
2129 enum op_flags flag_set, enum op_flags flag_clear)
d4e3b5db 2130{
7ae48d73 2131 po->flags |= flag_set;
2132 po->flags &= ~flag_clear;
d4e3b5db 2133}
2134
de50b98b 2135// last op in stream - unconditional branch or ret
2136#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
092f64e1 2137 || ((ops[_i].flags & (OPF_JMP|OPF_CJMP|OPF_RMD)) == OPF_JMP \
037f4971 2138 && ops[_i].op != OP_CALL))
de50b98b 2139
87bf6cec 2140static int scan_for_pop(int i, int opcnt, const char *reg,
d4e3b5db 2141 int magic, int depth, int *maxdepth, int do_flags)
850c9265 2142{
89ff3147 2143 const struct parsed_proto *pp;
87bf6cec 2144 struct parsed_op *po;
2145 int ret = 0;
4c45fa73 2146 int j;
87bf6cec 2147
850c9265 2148 for (; i < opcnt; i++) {
87bf6cec 2149 po = &ops[i];
2150 if (po->cc_scratch == magic)
2151 break; // already checked
2152 po->cc_scratch = magic;
2153
89ff3147 2154 if (po->flags & OPF_TAIL) {
2155 if (po->op == OP_CALL) {
92d715b6 2156 pp = proto_parse(g_fhdr, po->operand[0].name, g_quiet_pp);
89ff3147 2157 if (pp != NULL && pp->is_noreturn)
2158 // no stack cleanup for noreturn
2159 return ret;
2160 }
87bf6cec 2161 return -1; // deadend
89ff3147 2162 }
87bf6cec 2163
5e49b270 2164 if ((po->flags & (OPF_RMD|OPF_DONE))
5f70a34f 2165 || (po->op == OP_PUSH && po->p_argnum != 0)) // arg push
850c9265 2166 continue;
2167
87bf6cec 2168 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
4c45fa73 2169 if (po->btj != NULL) {
2170 // jumptable
da87ae38 2171 for (j = 0; j < po->btj->count; j++) {
4c45fa73 2172 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic,
2173 depth, maxdepth, do_flags);
2174 if (ret < 0)
2175 return ret; // dead end
2176 }
da87ae38 2177 return ret;
4c45fa73 2178 }
2179
87bf6cec 2180 if (po->bt_i < 0) {
2181 ferr(po, "dead branch\n");
2182 return -1;
2183 }
850c9265 2184
5c024ef7 2185 if (po->flags & OPF_CJMP) {
d4e3b5db 2186 ret |= scan_for_pop(po->bt_i, opcnt, reg, magic,
2187 depth, maxdepth, do_flags);
87bf6cec 2188 if (ret < 0)
2189 return ret; // dead end
2190 }
2191 else {
2192 i = po->bt_i - 1;
2193 }
2194 continue;
2195 }
2196
d4e3b5db 2197 if ((po->op == OP_POP || po->op == OP_PUSH)
2198 && po->operand[0].type == OPT_REG
87bf6cec 2199 && IS(po->operand[0].name, reg))
2200 {
9af2d373 2201 if (po->op == OP_PUSH && !(po->flags & OPF_FARGNR)) {
d4e3b5db 2202 depth++;
2203 if (depth > *maxdepth)
2204 *maxdepth = depth;
2205 if (do_flags)
7ae48d73 2206 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
d4e3b5db 2207 }
da87ae38 2208 else if (po->op == OP_POP) {
2209 if (depth == 0) {
2210 if (do_flags)
2211 op_set_clear_flag(po, OPF_RMD, OPF_RSAVE);
2212 return 1;
2213 }
2214 else {
2215 depth--;
2216 if (depth < 0) // should not happen
2217 ferr(po, "fail with depth\n");
2218 if (do_flags)
2219 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2220 }
d4e3b5db 2221 }
87bf6cec 2222 }
850c9265 2223 }
2224
87bf6cec 2225 return ret;
850c9265 2226}
2227
69a3cdfc 2228// scan for pop starting from 'ret' op (all paths)
d4e3b5db 2229static int scan_for_pop_ret(int i, int opcnt, const char *reg,
2230 int flag_set)
850c9265 2231{
2232 int found = 0;
2233 int j;
2234
2235 for (; i < opcnt; i++) {
87bf6cec 2236 if (!(ops[i].flags & OPF_TAIL))
850c9265 2237 continue;
2238
2239 for (j = i - 1; j >= 0; j--) {
5e49b270 2240 if (ops[j].flags & (OPF_RMD|OPF_DONE))
69a3cdfc 2241 continue;
2242 if (ops[j].flags & OPF_JMP)
850c9265 2243 return -1;
2244
5e49b270 2245 if (ops[j].op == OP_POP && ops[j].datap == NULL
2246 && ops[j].operand[0].type == OPT_REG
850c9265 2247 && IS(ops[j].operand[0].name, reg))
2248 {
2249 found = 1;
d4e3b5db 2250 ops[j].flags |= flag_set;
850c9265 2251 break;
2252 }
2253
d7857c3a 2254 if (g_labels[j] != NULL)
850c9265 2255 return -1;
2256 }
2257 }
2258
2259 return found ? 0 : -1;
2260}
2261
9af2d373 2262static void scan_for_pop_const(int i, int opcnt)
2263{
2264 int j;
2265
2266 for (j = i + 1; j < opcnt; j++) {
2267 if ((ops[j].flags & (OPF_JMP|OPF_TAIL|OPF_RSAVE))
2268 || ops[j].op == OP_PUSH || g_labels[i] != NULL)
2269 {
2270 break;
2271 }
2272
5e49b270 2273 if (ops[j].op == OP_POP && !(ops[j].flags & (OPF_RMD|OPF_DONE)))
9af2d373 2274 {
5e49b270 2275 ops[i].flags |= OPF_RMD | OPF_DONE;
2276 ops[j].flags |= OPF_DONE;
9af2d373 2277 ops[j].datap = &ops[i];
2278 break;
2279 }
2280 }
2281}
2282
591721d7 2283static void scan_propagate_df(int i, int opcnt)
2284{
2285 struct parsed_op *po = &ops[i];
2286 int j;
2287
2288 for (; i < opcnt; i++) {
2289 po = &ops[i];
2290 if (po->flags & OPF_DF)
2291 return; // already resolved
2292 po->flags |= OPF_DF;
2293
2294 if (po->op == OP_CALL)
2295 ferr(po, "call with DF set?\n");
2296
2297 if (po->flags & OPF_JMP) {
2298 if (po->btj != NULL) {
2299 // jumptable
2300 for (j = 0; j < po->btj->count; j++)
2301 scan_propagate_df(po->btj->d[j].bt_i, opcnt);
2302 return;
2303 }
2304
2305 if (po->bt_i < 0) {
2306 ferr(po, "dead branch\n");
2307 return;
2308 }
2309
5c024ef7 2310 if (po->flags & OPF_CJMP)
591721d7 2311 scan_propagate_df(po->bt_i, opcnt);
2312 else
2313 i = po->bt_i - 1;
2314 continue;
2315 }
2316
2317 if (po->flags & OPF_TAIL)
2318 break;
2319
2320 if (po->op == OP_CLD) {
5e49b270 2321 po->flags |= OPF_RMD | OPF_DONE;
591721d7 2322 return;
2323 }
2324 }
2325
2326 ferr(po, "missing DF clear?\n");
2327}
2328
1cd4a663 2329// is operand 'opr' modified by parsed_op 'po'?
5101a5f9 2330static int is_opr_modified(const struct parsed_opr *opr,
69a3cdfc 2331 const struct parsed_op *po)
2332{
89ff3147 2333 int mask;
2334
69a3cdfc 2335 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2336 return 0;
2337
89ff3147 2338 if (opr->type == OPT_REG) {
2339 if (po->op == OP_CALL) {
2340 mask = (1 << xAX) | (1 << xCX) | (1 << xDX);
2341 if ((1 << opr->reg) & mask)
2342 return 1;
2343 else
2344 return 0;
2345 }
2346
2347 if (po->operand[0].type == OPT_REG) {
2348 if (po->regmask_dst & (1 << opr->reg))
2349 return 1;
2350 else
2351 return 0;
2352 }
69a3cdfc 2353 }
2354
2355 return IS(po->operand[0].name, opr->name);
2356}
2357
5101a5f9 2358// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
2359static int is_any_opr_modified(const struct parsed_op *po_test,
89ff3147 2360 const struct parsed_op *po, int c_mode)
5101a5f9 2361{
89ff3147 2362 int mask;
5101a5f9 2363 int i;
2364
2365 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2366 return 0;
2367
de50b98b 2368 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2369 return 0;
2370
2371 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
2372 return 1;
2373
2374 // in reality, it can wreck any register, but in decompiled C
2b43685d 2375 // version it can only overwrite eax or edx:eax
89ff3147 2376 mask = (1 << xAX) | (1 << xDX);
2377 if (!c_mode)
2378 mask |= 1 << xCX;
2379
de50b98b 2380 if (po->op == OP_CALL
89ff3147 2381 && ((po_test->regmask_src | po_test->regmask_dst) & mask))
5101a5f9 2382 return 1;
2383
2384 for (i = 0; i < po_test->operand_cnt; i++)
2385 if (IS(po_test->operand[i].name, po->operand[0].name))
2386 return 1;
2387
2388 return 0;
2389}
2390
940e8e66 2391// scan for any po_test operand modification in range given
89ff3147 2392static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt,
2393 int c_mode)
69a3cdfc 2394{
2b43685d 2395 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2396 return -1;
2397
69a3cdfc 2398 for (; i < opcnt; i++) {
89ff3147 2399 if (is_any_opr_modified(po_test, &ops[i], c_mode))
69a3cdfc 2400 return i;
2401 }
2402
2403 return -1;
2404}
2405
940e8e66 2406// scan for po_test operand[0] modification in range given
2407static int scan_for_mod_opr0(struct parsed_op *po_test,
2408 int i, int opcnt)
2409{
2410 for (; i < opcnt; i++) {
2411 if (is_opr_modified(&po_test->operand[0], &ops[i]))
2412 return i;
2413 }
2414
2415 return -1;
2416}
2417
92d715b6 2418#define check_i(po, i) \
2419 if ((i) < 0) \
2420 ferr(po, "bad " #i ": %d\n", i)
2421
04f8a628 2422static int scan_for_flag_set(int i, int magic, int *branched,
2423 int *setters, int *setter_cnt)
69a3cdfc 2424{
04f8a628 2425 struct label_ref *lr;
2b43685d 2426 int ret;
de50b98b 2427
2428 while (i >= 0) {
04f8a628 2429 if (ops[i].cc_scratch == magic) {
2430 ferr(&ops[i], "%s looped\n", __func__);
2431 return -1;
2432 }
2433 ops[i].cc_scratch = magic;
2434
d7857c3a 2435 if (g_labels[i] != NULL) {
2b43685d 2436 *branched = 1;
04f8a628 2437
2438 lr = &g_label_refs[i];
2439 for (; lr->next; lr = lr->next) {
92d715b6 2440 check_i(&ops[i], lr->i);
04f8a628 2441 ret = scan_for_flag_set(lr->i, magic,
2442 branched, setters, setter_cnt);
2443 if (ret < 0)
2444 return ret;
2445 }
2446
92d715b6 2447 check_i(&ops[i], lr->i);
de50b98b 2448 if (i > 0 && LAST_OP(i - 1)) {
94d447fb 2449 i = lr->i;
de50b98b 2450 continue;
2451 }
04f8a628 2452 ret = scan_for_flag_set(lr->i, magic,
2453 branched, setters, setter_cnt);
2b43685d 2454 if (ret < 0)
2455 return ret;
de50b98b 2456 }
2457 i--;
2458
2b43685d 2459 if (ops[i].flags & OPF_FLAGS) {
2460 setters[*setter_cnt] = i;
2461 (*setter_cnt)++;
2462 return 0;
2463 }
69a3cdfc 2464
5c024ef7 2465 if ((ops[i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP)
69a3cdfc 2466 return -1;
69a3cdfc 2467 }
2468
2469 return -1;
2470}
2471
5101a5f9 2472// scan back for cdq, if anything modifies edx, fail
2473static int scan_for_cdq_edx(int i)
2474{
cdfaeed7 2475 while (i >= 0) {
d7857c3a 2476 if (g_labels[i] != NULL) {
cdfaeed7 2477 if (g_label_refs[i].next != NULL)
2478 return -1;
2479 if (i > 0 && LAST_OP(i - 1)) {
2480 i = g_label_refs[i].i;
2481 continue;
2482 }
2483 return -1;
2484 }
2485 i--;
2486
5101a5f9 2487 if (ops[i].op == OP_CDQ)
2488 return i;
2489
2490 if (ops[i].regmask_dst & (1 << xDX))
2491 return -1;
5101a5f9 2492 }
2493
2494 return -1;
2495}
2496
64c59faf 2497static int scan_for_reg_clear(int i, int reg)
2498{
cdfaeed7 2499 while (i >= 0) {
d7857c3a 2500 if (g_labels[i] != NULL) {
cdfaeed7 2501 if (g_label_refs[i].next != NULL)
2502 return -1;
2503 if (i > 0 && LAST_OP(i - 1)) {
2504 i = g_label_refs[i].i;
2505 continue;
2506 }
2507 return -1;
2508 }
2509 i--;
2510
64c59faf 2511 if (ops[i].op == OP_XOR
2512 && ops[i].operand[0].lmod == OPLM_DWORD
2513 && ops[i].operand[0].reg == ops[i].operand[1].reg
2514 && ops[i].operand[0].reg == reg)
2515 return i;
2516
2517 if (ops[i].regmask_dst & (1 << reg))
2518 return -1;
64c59faf 2519 }
2520
2521 return -1;
2522}
2523
ee2361b9 2524static void patch_esp_adjust(struct parsed_op *po, int adj)
2525{
2526 ferr_assert(po, po->op == OP_ADD);
2527 ferr_assert(po, IS(opr_name(po, 0), "esp"));
2528 ferr_assert(po, po->operand[1].type == OPT_CONST);
2529
2530 // this is a bit of a hack, but deals with use of
2531 // single adj for multiple calls
2532 po->operand[1].val -= adj;
2533 po->flags |= OPF_RMD;
2534 if (po->operand[1].val == 0)
2535 po->flags |= OPF_DONE;
2536 ferr_assert(po, (int)po->operand[1].val >= 0);
2537}
2538
1bafb621 2539// scan for positive, constant esp adjust
ee2361b9 2540// multipath case is preliminary
9af2d373 2541static int scan_for_esp_adjust(int i, int opcnt,
ee2361b9 2542 int adj_expect, int *adj, int *is_multipath, int do_update)
1bafb621 2543{
7ba45c34 2544 struct parsed_op *po;
46411e6c 2545 int first_pop = -1;
4741fdfe 2546
ee2361b9 2547 *adj = *is_multipath = 0;
7ba45c34 2548
9af2d373 2549 for (; i < opcnt && *adj < adj_expect; i++) {
d7857c3a 2550 if (g_labels[i] != NULL)
ee2361b9 2551 *is_multipath = 1;
a2c1d768 2552
5e49b270 2553 po = &ops[i];
2554 if (po->flags & OPF_DONE)
2555 continue;
2556
7ba45c34 2557 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
2558 if (po->operand[1].type != OPT_CONST)
1bafb621 2559 ferr(&ops[i], "non-const esp adjust?\n");
7ba45c34 2560 *adj += po->operand[1].val;
1bafb621 2561 if (*adj & 3)
2562 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
ee2361b9 2563 if (do_update) {
2564 if (!*is_multipath)
2565 patch_esp_adjust(po, adj_expect);
2566 else
2567 po->flags |= OPF_RMD;
2568 }
1bafb621 2569 return i;
2570 }
5e49b270 2571 else if (po->op == OP_PUSH) {
5c024ef7 2572 //if (first_pop == -1)
2573 // first_pop = -2; // none
7ba45c34 2574 *adj -= lmod_bytes(po, po->operand[0].lmod);
46411e6c 2575 }
5e49b270 2576 else if (po->op == OP_POP) {
91ca764a 2577 if (!(po->flags & OPF_DONE)) {
2578 // seems like msvc only uses 'pop ecx' for stack realignment..
2579 if (po->operand[0].type != OPT_REG || po->operand[0].reg != xCX)
2580 break;
2581 if (first_pop == -1 && *adj >= 0)
2582 first_pop = i;
2583 }
ee2361b9 2584 if (do_update && *adj >= 0) {
2585 po->flags |= OPF_RMD;
2586 if (!*is_multipath)
2587 po->flags |= OPF_DONE;
2588 }
2589
7ba45c34 2590 *adj += lmod_bytes(po, po->operand[0].lmod);
46411e6c 2591 }
e56ab892 2592 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
4741fdfe 2593 if (po->op == OP_JMP && po->btj == NULL) {
5e49b270 2594 if (po->bt_i <= i)
2595 break;
4741fdfe 2596 i = po->bt_i - 1;
2597 continue;
2598 }
e56ab892 2599 if (po->op != OP_CALL)
a2c1d768 2600 break;
e56ab892 2601 if (po->operand[0].type != OPT_LABEL)
a2c1d768 2602 break;
91ca764a 2603 if (po->pp != NULL && po->pp->is_stdcall)
89ff3147 2604 break;
91ca764a 2605 // assume it's another cdecl call
e56ab892 2606 }
a2c1d768 2607 }
7ba45c34 2608
108e9fe3 2609 if (first_pop >= 0) {
a2c1d768 2610 // probably 'pop ecx' was used..
46411e6c 2611 return first_pop;
1bafb621 2612 }
2613
2614 return -1;
2615}
2616
a3684be1 2617static void scan_fwd_set_flags(int i, int opcnt, int magic, int flags)
2618{
2619 struct parsed_op *po;
2620 int j;
2621
2622 if (i < 0)
2623 ferr(ops, "%s: followed bad branch?\n", __func__);
2624
2625 for (; i < opcnt; i++) {
2626 po = &ops[i];
2627 if (po->cc_scratch == magic)
2628 return;
2629 po->cc_scratch = magic;
2630 po->flags |= flags;
2631
2632 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2633 if (po->btj != NULL) {
2634 // jumptable
2635 for (j = 0; j < po->btj->count; j++)
2636 scan_fwd_set_flags(po->btj->d[j].bt_i, opcnt, magic, flags);
2637 return;
2638 }
2639
2640 scan_fwd_set_flags(po->bt_i, opcnt, magic, flags);
5c024ef7 2641 if (!(po->flags & OPF_CJMP))
a3684be1 2642 return;
2643 }
2644 if (po->flags & OPF_TAIL)
2645 return;
2646 }
2647}
2648
1cd4a663 2649static const struct parsed_proto *try_recover_pp(
da87ae38 2650 struct parsed_op *po, const struct parsed_opr *opr, int *search_instead)
1cd4a663 2651{
2652 const struct parsed_proto *pp = NULL;
89ff3147 2653 char buf[256];
2654 char *p;
1cd4a663 2655
2656 // maybe an arg of g_func?
2657 if (opr->type == OPT_REGMEM && is_stack_access(po, opr))
2658 {
2659 char ofs_reg[16] = { 0, };
2660 int arg, arg_s, arg_i;
2661 int stack_ra = 0;
2662 int offset = 0;
2663
9af2d373 2664 if (g_header_mode)
2665 return NULL;
2666
1cd4a663 2667 parse_stack_access(po, opr->name, ofs_reg,
037f4971 2668 &offset, &stack_ra, NULL, 0);
1cd4a663 2669 if (ofs_reg[0] != 0)
2670 ferr(po, "offset reg on arg access?\n");
da87ae38 2671 if (offset <= stack_ra) {
2672 // search who set the stack var instead
2673 if (search_instead != NULL)
2674 *search_instead = 1;
2675 return NULL;
2676 }
1cd4a663 2677
2678 arg_i = (offset - stack_ra - 4) / 4;
2679 for (arg = arg_s = 0; arg < g_func_pp->argc; arg++) {
2680 if (g_func_pp->arg[arg].reg != NULL)
2681 continue;
2682 if (arg_s == arg_i)
2683 break;
2684 arg_s++;
2685 }
2686 if (arg == g_func_pp->argc)
2687 ferr(po, "stack arg %d not in prototype?\n", arg_i);
2688
2689 pp = g_func_pp->arg[arg].fptr;
2690 if (pp == NULL)
46411e6c 2691 ferr(po, "icall sa: arg%d is not a fptr?\n", arg + 1);
89ff3147 2692 check_func_pp(po, pp, "icall arg");
2693 }
2694 else if (opr->type == OPT_REGMEM && strchr(opr->name + 1, '[')) {
2695 // label[index]
2696 p = strchr(opr->name + 1, '[');
2697 memcpy(buf, opr->name, p - opr->name);
2698 buf[p - opr->name] = 0;
92d715b6 2699 pp = proto_parse(g_fhdr, buf, g_quiet_pp);
1cd4a663 2700 }
2701 else if (opr->type == OPT_OFFSET || opr->type == OPT_LABEL) {
9af2d373 2702 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
2703 if (pp == NULL) {
2704 if (!g_header_mode)
2705 ferr(po, "proto_parse failed for icall to '%s'\n", opr->name);
2706 }
2707 else
2708 check_func_pp(po, pp, "reg-fptr ref");
1cd4a663 2709 }
2710
2711 return pp;
2712}
2713
da87ae38 2714static void scan_for_call_type(int i, const struct parsed_opr *opr,
89ff3147 2715 int magic, const struct parsed_proto **pp_found, int *multi)
1cd4a663 2716{
2717 const struct parsed_proto *pp = NULL;
2718 struct parsed_op *po;
2719 struct label_ref *lr;
2720
037f4971 2721 ops[i].cc_scratch = magic;
1cd4a663 2722
037f4971 2723 while (1) {
d7857c3a 2724 if (g_labels[i] != NULL) {
1cd4a663 2725 lr = &g_label_refs[i];
92d715b6 2726 for (; lr != NULL; lr = lr->next) {
2727 check_i(&ops[i], lr->i);
89ff3147 2728 scan_for_call_type(lr->i, opr, magic, pp_found, multi);
92d715b6 2729 }
46411e6c 2730 if (i > 0 && LAST_OP(i - 1))
2731 return;
1cd4a663 2732 }
037f4971 2733
1cd4a663 2734 i--;
037f4971 2735 if (i < 0)
2736 break;
2737
2738 if (ops[i].cc_scratch == magic)
2739 return;
2740 ops[i].cc_scratch = magic;
1cd4a663 2741
2742 if (!(ops[i].flags & OPF_DATA))
2743 continue;
2744 if (!is_opr_modified(opr, &ops[i]))
2745 continue;
2746 if (ops[i].op != OP_MOV && ops[i].op != OP_LEA) {
2747 // most probably trashed by some processing
2748 *pp_found = NULL;
2749 return;
2750 }
2751
2752 opr = &ops[i].operand[1];
2753 if (opr->type != OPT_REG)
2754 break;
2755 }
2756
2757 po = (i >= 0) ? &ops[i] : ops;
2758
2759 if (i < 0) {
2760 // reached the top - can only be an arg-reg
2761 if (opr->type != OPT_REG)
2762 return;
2763
2764 for (i = 0; i < g_func_pp->argc; i++) {
2765 if (g_func_pp->arg[i].reg == NULL)
2766 continue;
2767 if (IS(opr->name, g_func_pp->arg[i].reg))
2768 break;
2769 }
2770 if (i == g_func_pp->argc)
2771 return;
2772 pp = g_func_pp->arg[i].fptr;
2773 if (pp == NULL)
46411e6c 2774 ferr(po, "icall: arg%d (%s) is not a fptr?\n",
2775 i + 1, g_func_pp->arg[i].reg);
89ff3147 2776 check_func_pp(po, pp, "icall reg-arg");
1cd4a663 2777 }
2778 else
da87ae38 2779 pp = try_recover_pp(po, opr, NULL);
1cd4a663 2780
89ff3147 2781 if (*pp_found != NULL && pp != NULL && *pp_found != pp) {
1cd4a663 2782 if (!IS((*pp_found)->ret_type.name, pp->ret_type.name)
2783 || (*pp_found)->is_stdcall != pp->is_stdcall
89ff3147 2784 || (*pp_found)->is_fptr != pp->is_fptr
1cd4a663 2785 || (*pp_found)->argc != pp->argc
2786 || (*pp_found)->argc_reg != pp->argc_reg
2787 || (*pp_found)->argc_stack != pp->argc_stack)
2788 {
2789 ferr(po, "icall: parsed_proto mismatch\n");
2790 }
89ff3147 2791 *multi = 1;
1cd4a663 2792 }
2793 if (pp != NULL)
2794 *pp_found = pp;
2795}
2796
9af2d373 2797// early check for tail call or branch back
2798static int is_like_tailjmp(int j)
2799{
2800 if (!(ops[j].flags & OPF_JMP))
2801 return 0;
2802
2803 if (ops[j].op == OP_JMP && !ops[j].operand[0].had_ds)
2804 // probably local branch back..
2805 return 1;
2806 if (ops[j].op == OP_CALL)
2807 // probably noreturn call..
2808 return 1;
2809
2810 return 0;
2811}
2812
2813static void scan_prologue_epilogue(int opcnt)
2814{
2815 int ecx_push = 0, esp_sub = 0;
2816 int found;
2817 int i, j, l;
2818
2819 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
2820 && ops[1].op == OP_MOV
2821 && IS(opr_name(&ops[1], 0), "ebp")
2822 && IS(opr_name(&ops[1], 1), "esp"))
2823 {
2824 g_bp_frame = 1;
5e49b270 2825 ops[0].flags |= OPF_RMD | OPF_DONE;
2826 ops[1].flags |= OPF_RMD | OPF_DONE;
9af2d373 2827 i = 2;
2828
2829 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
2830 g_stack_fsz = opr_const(&ops[2], 1);
5e49b270 2831 ops[2].flags |= OPF_RMD | OPF_DONE;
9af2d373 2832 i++;
2833 }
2834 else {
2835 // another way msvc builds stack frame..
2836 i = 2;
2837 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
2838 g_stack_fsz += 4;
5e49b270 2839 ops[i].flags |= OPF_RMD | OPF_DONE;
9af2d373 2840 ecx_push++;
2841 i++;
2842 }
2843 // and another way..
2844 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
2845 && ops[i].operand[1].type == OPT_CONST
2846 && ops[i + 1].op == OP_CALL
2847 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
2848 {
2849 g_stack_fsz += ops[i].operand[1].val;
5e49b270 2850 ops[i].flags |= OPF_RMD | OPF_DONE;
9af2d373 2851 i++;
5e49b270 2852 ops[i].flags |= OPF_RMD | OPF_DONE;
9af2d373 2853 i++;
2854 }
2855 }
2856
2857 found = 0;
2858 do {
2859 for (; i < opcnt; i++)
2860 if (ops[i].op == OP_RET)
2861 break;
2862 j = i - 1;
2863 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
2864 if (found && is_like_tailjmp(j))
2865 break;
2866 j--;
2867 }
2868
2869 if ((ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ebp"))
2870 || ops[j].op == OP_LEAVE)
2871 {
5e49b270 2872 ops[j].flags |= OPF_RMD | OPF_DONE;
9af2d373 2873 }
2874 else if (!(g_ida_func_attr & IDAFA_NORETURN))
2875 ferr(&ops[j], "'pop ebp' expected\n");
2876
2877 if (g_stack_fsz != 0) {
2878 if (ops[j - 1].op == OP_MOV
2879 && IS(opr_name(&ops[j - 1], 0), "esp")
2880 && IS(opr_name(&ops[j - 1], 1), "ebp"))
2881 {
5e49b270 2882 ops[j - 1].flags |= OPF_RMD | OPF_DONE;
9af2d373 2883 }
2884 else if (ops[j].op != OP_LEAVE
2885 && !(g_ida_func_attr & IDAFA_NORETURN))
2886 {
2887 ferr(&ops[j - 1], "esp restore expected\n");
2888 }
2889
2890 if (ecx_push && ops[j - 2].op == OP_POP
2891 && IS(opr_name(&ops[j - 2], 0), "ecx"))
2892 {
2893 ferr(&ops[j - 2], "unexpected ecx pop\n");
2894 }
2895 }
2896
2897 found = 1;
2898 i++;
2899 } while (i < opcnt);
2900
2901 return;
2902 }
2903
2904 // non-bp frame
2905 i = 0;
2906 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
5e49b270 2907 ops[i].flags |= OPF_RMD | OPF_DONE;
9af2d373 2908 g_stack_fsz += 4;
2909 ecx_push++;
2910 i++;
2911 }
2912
2913 for (; i < opcnt; i++) {
2914 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
2915 break;
2916 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
2917 && ops[i].operand[1].type == OPT_CONST)
2918 {
2919 g_stack_fsz = ops[i].operand[1].val;
5e49b270 2920 ops[i].flags |= OPF_RMD | OPF_DONE;
9af2d373 2921 esp_sub = 1;
2922 break;
2923 }
2924 }
2925
2926 if (ecx_push && !esp_sub) {
2927 // could actually be args for a call..
2928 for (; i < opcnt; i++)
2929 if (ops[i].op != OP_PUSH)
2930 break;
2931
2932 if (ops[i].op == OP_CALL && ops[i].operand[0].type == OPT_LABEL) {
2933 const struct parsed_proto *pp;
2934 pp = proto_parse(g_fhdr, opr_name(&ops[i], 0), 1);
2935 j = pp ? pp->argc_stack : 0;
2936 while (i > 0 && j > 0) {
2937 i--;
2938 if (ops[i].op == OP_PUSH) {
5e49b270 2939 ops[i].flags &= ~(OPF_RMD | OPF_DONE);
9af2d373 2940 j--;
2941 }
2942 }
2943 if (j != 0)
2944 ferr(&ops[i], "unhandled prologue\n");
2945
2946 // recheck
2947 i = g_stack_fsz = ecx_push = 0;
2948 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
2949 if (!(ops[i].flags & OPF_RMD))
2950 break;
2951 g_stack_fsz += 4;
2952 ecx_push++;
2953 i++;
2954 }
2955 }
2956 }
2957
2958 found = 0;
2959 if (ecx_push || esp_sub)
2960 {
2961 g_sp_frame = 1;
2962
2963 i++;
2964 do {
2965 for (; i < opcnt; i++)
2966 if (ops[i].op == OP_RET)
2967 break;
2968 j = i - 1;
2969 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
2970 if (found && is_like_tailjmp(j))
2971 break;
2972 j--;
2973 }
2974
2975 if (ecx_push > 0) {
2976 for (l = 0; l < ecx_push; l++) {
2977 if (ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ecx"))
2978 /* pop ecx */;
2979 else if (ops[j].op == OP_ADD
2980 && IS(opr_name(&ops[j], 0), "esp")
2981 && ops[j].operand[1].type == OPT_CONST)
2982 {
2983 /* add esp, N */
5e49b270 2984 l += ops[j].operand[1].val / 4 - 1;
9af2d373 2985 }
2986 else
2987 ferr(&ops[j], "'pop ecx' expected\n");
2988
5e49b270 2989 ops[j].flags |= OPF_RMD | OPF_DONE;
9af2d373 2990 j--;
2991 }
2992 if (l != ecx_push)
2993 ferr(&ops[j], "epilogue scan failed\n");
2994
2995 found = 1;
2996 }
2997
2998 if (esp_sub) {
2999 if (ops[j].op != OP_ADD
3000 || !IS(opr_name(&ops[j], 0), "esp")
3001 || ops[j].operand[1].type != OPT_CONST
3002 || ops[j].operand[1].val != g_stack_fsz)
3003 ferr(&ops[j], "'add esp' expected\n");
3004
5e49b270 3005 ops[j].flags |= OPF_RMD | OPF_DONE;
9af2d373 3006 ops[j].operand[1].val = 0; // hack for stack arg scanner
3007 found = 1;
3008 }
3009
3010 i++;
3011 } while (i < opcnt);
3012 }
3013}
3014
89ff3147 3015static const struct parsed_proto *resolve_icall(int i, int opcnt,
3016 int *multi_src)
1cd4a663 3017{
3018 const struct parsed_proto *pp = NULL;
da87ae38 3019 int search_advice = 0;
1cd4a663 3020
89ff3147 3021 *multi_src = 0;
3022
1cd4a663 3023 switch (ops[i].operand[0].type) {
3024 case OPT_REGMEM:
3025 case OPT_LABEL:
3026 case OPT_OFFSET:
da87ae38 3027 pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice);
3028 if (!search_advice)
3029 break;
3030 // fallthrough
1cd4a663 3031 default:
89ff3147 3032 scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp,
3033 multi_src);
1cd4a663 3034 break;
3035 }
3036
3037 return pp;
3038}
3039
23fd0b11 3040// find an instruction that changed opr before i op
92d715b6 3041// *op_i must be set to -1 by caller
3042// *entry is set to 1 if one source is determined to be the caller
3043// returns 1 if found, *op_i is then set to origin
23fd0b11 3044static int resolve_origin(int i, const struct parsed_opr *opr,
92d715b6 3045 int magic, int *op_i, int *is_caller)
1f84f6b3 3046{
3047 struct label_ref *lr;
3048 int ret = 0;
3049
92d715b6 3050 if (ops[i].cc_scratch == magic)
3051 return 0;
1f84f6b3 3052 ops[i].cc_scratch = magic;
3053
3054 while (1) {
d7857c3a 3055 if (g_labels[i] != NULL) {
1f84f6b3 3056 lr = &g_label_refs[i];
92d715b6 3057 for (; lr != NULL; lr = lr->next) {
3058 check_i(&ops[i], lr->i);
3059 ret |= resolve_origin(lr->i, opr, magic, op_i, is_caller);
3060 }
1f84f6b3 3061 if (i > 0 && LAST_OP(i - 1))
3062 return ret;
3063 }
3064
3065 i--;
92d715b6 3066 if (i < 0) {
3067 if (is_caller != NULL)
3068 *is_caller = 1;
1f84f6b3 3069 return -1;
92d715b6 3070 }
1f84f6b3 3071
3072 if (ops[i].cc_scratch == magic)
3073 return 0;
3074 ops[i].cc_scratch = magic;
3075
3076 if (!(ops[i].flags & OPF_DATA))
3077 continue;
3078 if (!is_opr_modified(opr, &ops[i]))
3079 continue;
23fd0b11 3080
3081 if (*op_i >= 0) {
3082 if (*op_i == i)
3083 return 1;
3084 // XXX: could check if the other op does the same
3085 return -1;
3086 }
3087
3088 *op_i = i;
3089 return 1;
3090 }
3091}
3092
3093static int try_resolve_const(int i, const struct parsed_opr *opr,
3094 int magic, unsigned int *val)
3095{
3096 int s_i = -1;
92d715b6 3097 int ret;
23fd0b11 3098
92d715b6 3099 ret = resolve_origin(i, opr, magic, &s_i, NULL);
23fd0b11 3100 if (ret == 1) {
3101 i = s_i;
1f84f6b3 3102 if (ops[i].op != OP_MOV && ops[i].operand[1].type != OPT_CONST)
3103 return -1;
3104
3105 *val = ops[i].operand[1].val;
3106 return 1;
3107 }
23fd0b11 3108
3109 return -1;
1f84f6b3 3110}
3111
26677139 3112static struct parsed_proto *process_call_early(int i, int opcnt,
3113 int *adj_i)
3114{
3115 struct parsed_op *po = &ops[i];
3116 struct parsed_proto *pp;
3117 int multipath = 0;
3118 int adj = 0;
3119 int ret;
3120
3121 pp = po->pp;
3122 if (pp == NULL || pp->is_vararg || pp->argc_reg != 0)
3123 // leave for later
3124 return NULL;
3125
3126 // look for and make use of esp adjust
3127 *adj_i = ret = -1;
3128 if (!pp->is_stdcall && pp->argc_stack > 0)
3129 ret = scan_for_esp_adjust(i + 1, opcnt,
ee2361b9 3130 pp->argc_stack * 4, &adj, &multipath, 0);
26677139 3131 if (ret >= 0) {
3132 if (pp->argc_stack > adj / 4)
3133 return NULL;
3134 if (multipath)
3135 return NULL;
3136 if (ops[ret].op == OP_POP && adj != 4)
3137 return NULL;
3138 }
3139
3140 *adj_i = ret;
3141 return pp;
3142}
3143
9af2d373 3144static struct parsed_proto *process_call(int i, int opcnt)
3145{
3146 struct parsed_op *po = &ops[i];
3147 const struct parsed_proto *pp_c;
3148 struct parsed_proto *pp;
3149 const char *tmpname;
26677139 3150 int adj = 0, multipath = 0;
9af2d373 3151 int ret, arg;
3152
3153 tmpname = opr_name(po, 0);
3154 pp = po->pp;
3155 if (pp == NULL)
3156 {
3157 // indirect call
26677139 3158 pp_c = resolve_icall(i, opcnt, &multipath);
9af2d373 3159 if (pp_c != NULL) {
3160 if (!pp_c->is_func && !pp_c->is_fptr)
3161 ferr(po, "call to non-func: %s\n", pp_c->name);
3162 pp = proto_clone(pp_c);
3163 my_assert_not(pp, NULL);
26677139 3164 if (multipath)
9af2d373 3165 // not resolved just to single func
3166 pp->is_fptr = 1;
3167
3168 switch (po->operand[0].type) {
3169 case OPT_REG:
3170 // we resolved this call and no longer need the register
3171 po->regmask_src &= ~(1 << po->operand[0].reg);
3172 break;
3173 case OPT_REGMEM:
3174 pp->is_fptr = 1;
3175 break;
3176 default:
3177 break;
3178 }
3179 }
3180 if (pp == NULL) {
3181 pp = calloc(1, sizeof(*pp));
3182 my_assert_not(pp, NULL);
3183
3184 pp->is_fptr = 1;
ee2361b9 3185 ret = scan_for_esp_adjust(i + 1, opcnt,
3186 32*4, &adj, &multipath, 0);
26677139 3187 if (ret < 0 || adj < 0) {
9af2d373 3188 if (!g_allow_regfunc)
3189 ferr(po, "non-__cdecl indirect call unhandled yet\n");
3190 pp->is_unresolved = 1;
26677139 3191 adj = 0;
9af2d373 3192 }
26677139 3193 adj /= 4;
3194 if (adj > ARRAY_SIZE(pp->arg))
3195 ferr(po, "esp adjust too large: %d\n", adj);
9af2d373 3196 pp->ret_type.name = strdup("int");
26677139 3197 pp->argc = pp->argc_stack = adj;
9af2d373 3198 for (arg = 0; arg < pp->argc; arg++)
3199 pp->arg[arg].type.name = strdup("int");
3200 }
3201 po->pp = pp;
3202 }
3203
3204 // look for and make use of esp adjust
91ca764a 3205 multipath = 0;
9af2d373 3206 ret = -1;
3207 if (!pp->is_stdcall && pp->argc_stack > 0)
3208 ret = scan_for_esp_adjust(i + 1, opcnt,
ee2361b9 3209 pp->argc_stack * 4, &adj, &multipath, 0);
9af2d373 3210 if (ret >= 0) {
3211 if (pp->is_vararg) {
26677139 3212 if (adj / 4 < pp->argc_stack) {
3213 fnote(po, "(this call)\n");
3214 ferr(&ops[ret], "esp adjust is too small: %x < %x\n",
3215 adj, pp->argc_stack * 4);
3216 }
9af2d373 3217 // modify pp to make it have varargs as normal args
3218 arg = pp->argc;
26677139 3219 pp->argc += adj / 4 - pp->argc_stack;
9af2d373 3220 for (; arg < pp->argc; arg++) {
3221 pp->arg[arg].type.name = strdup("int");
3222 pp->argc_stack++;
3223 }
3224 if (pp->argc > ARRAY_SIZE(pp->arg))
3225 ferr(po, "too many args for '%s'\n", tmpname);
3226 }
26677139 3227 if (pp->argc_stack > adj / 4) {
9af2d373 3228 fnote(po, "(this call)\n");
3229 ferr(&ops[ret], "stack tracking failed for '%s': %x %x\n",
26677139 3230 tmpname, pp->argc_stack * 4, adj);
9af2d373 3231 }
3232
ee2361b9 3233 scan_for_esp_adjust(i + 1, opcnt,
3234 pp->argc_stack * 4, &adj, &multipath, 1);
9af2d373 3235 }
3236 else if (pp->is_vararg)
3237 ferr(po, "missing esp_adjust for vararg func '%s'\n",
3238 pp->name);
3239
3240 return pp;
3241}
3242
26677139 3243static int collect_call_args_early(struct parsed_op *po, int i,
3244 struct parsed_proto *pp, int *regmask)
3245{
3246 int arg, ret;
3247 int j;
3248
3249 for (arg = 0; arg < pp->argc; arg++)
3250 if (pp->arg[arg].reg == NULL)
3251 break;
3252
3253 // first see if it can be easily done
3254 for (j = i; j > 0 && arg < pp->argc; )
3255 {
3256 if (g_labels[j] != NULL)
3257 return -1;
3258 j--;
3259
3260 if (ops[j].op == OP_CALL)
3261 return -1;
3262 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP)
3263 return -1;
3264 else if (ops[j].op == OP_POP)
3265 return -1;
3266 else if (ops[j].flags & OPF_CJMP)
3267 return -1;
3268 else if (ops[j].op == OP_PUSH) {
3269 if (ops[j].flags & (OPF_FARG|OPF_FARGNR))
3270 return -1;
3271 ret = scan_for_mod(&ops[j], j + 1, i, 1);
3272 if (ret >= 0)
3273 return -1;
3274
3275 if (pp->arg[arg].type.is_va_list)
3276 return -1;
3277
3278 // next arg
3279 for (arg++; arg < pp->argc; arg++)
3280 if (pp->arg[arg].reg == NULL)
3281 break;
3282 }
3283 }
3284
3285 if (arg < pp->argc)
3286 return -1;
3287
3288 // now do it
3289 for (arg = 0; arg < pp->argc; arg++)
3290 if (pp->arg[arg].reg == NULL)
3291 break;
3292
3293 for (j = i; j > 0 && arg < pp->argc; )
3294 {
3295 j--;
3296
3297 if (ops[j].op == OP_PUSH)
3298 {
3299 ops[j].p_argnext = -1;
3300 ferr_assert(&ops[j], pp->arg[arg].datap == NULL);
3301 pp->arg[arg].datap = &ops[j];
3302
3303 if (ops[j].operand[0].type == OPT_REG)
3304 *regmask |= 1 << ops[j].operand[0].reg;
3305
5e49b270 3306 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_FARGNR | OPF_FARG;
26677139 3307 ops[j].flags &= ~OPF_RSAVE;
3308
3309 // next arg
3310 for (arg++; arg < pp->argc; arg++)
3311 if (pp->arg[arg].reg == NULL)
3312 break;
3313 }
3314 }
3315
3316 return 0;
3317}
3318
89ff3147 3319static int collect_call_args_r(struct parsed_op *po, int i,
3a5101d7 3320 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
3321 int *arg_grp, int arg, int magic, int need_op_saving, int may_reuse)
e56ab892 3322{
3323 struct parsed_proto *pp_tmp;
3a5101d7 3324 struct parsed_op *po_tmp;
e56ab892 3325 struct label_ref *lr;
2b43685d 3326 int need_to_save_current;
3a5101d7 3327 int arg_grp_current = 0;
3328 int save_args_seen = 0;
23fd0b11 3329 int save_args;
e56ab892 3330 int ret = 0;
5f70a34f 3331 int reg;
23fd0b11 3332 char buf[32];
3333 int j, k;
e56ab892 3334
a3684be1 3335 if (i < 0) {
a2c1d768 3336 ferr(po, "dead label encountered\n");
a3684be1 3337 return -1;
3338 }
e56ab892 3339
de50b98b 3340 for (; arg < pp->argc; arg++)
e56ab892 3341 if (pp->arg[arg].reg == NULL)
3342 break;
a3684be1 3343 magic = (magic & 0xffffff) | (arg << 24);
e56ab892 3344
89ff3147 3345 for (j = i; j >= 0 && (arg < pp->argc || pp->is_unresolved); )
e56ab892 3346 {
a3684be1 3347 if (((ops[j].cc_scratch ^ magic) & 0xffffff) == 0) {
3348 if (ops[j].cc_scratch != magic) {
3349 ferr(&ops[j], "arg collect hit same path with diff args for %s\n",
3350 pp->name);
3351 return -1;
3352 }
3353 // ok: have already been here
3354 return 0;
3355 }
3356 ops[j].cc_scratch = magic;
3357
d7857c3a 3358 if (g_labels[j] != NULL && g_label_refs[j].i != -1) {
e56ab892 3359 lr = &g_label_refs[j];
3360 if (lr->next != NULL)
3361 need_op_saving = 1;
a652aa9f 3362 for (; lr->next; lr = lr->next) {
92d715b6 3363 check_i(&ops[j], lr->i);
5c024ef7 3364 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
a652aa9f 3365 may_reuse = 1;
89ff3147 3366 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
3a5101d7 3367 arg_grp, arg, magic, need_op_saving, may_reuse);
a3684be1 3368 if (ret < 0)
3369 return ret;
a652aa9f 3370 }
e56ab892 3371
92d715b6 3372 check_i(&ops[j], lr->i);
5c024ef7 3373 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
a652aa9f 3374 may_reuse = 1;
de50b98b 3375 if (j > 0 && LAST_OP(j - 1)) {
e56ab892 3376 // follow last branch in reverse
3377 j = lr->i;
3378 continue;
3379 }
3380 need_op_saving = 1;
89ff3147 3381 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
3a5101d7 3382 arg_grp, arg, magic, need_op_saving, may_reuse);
a3684be1 3383 if (ret < 0)
3384 return ret;
e56ab892 3385 }
3386 j--;
3387
3388 if (ops[j].op == OP_CALL)
3389 {
89ff3147 3390 if (pp->is_unresolved)
3391 break;
3392
092f64e1 3393 pp_tmp = ops[j].pp;
e56ab892 3394 if (pp_tmp == NULL)
7ae48d73 3395 ferr(po, "arg collect hit unparsed call '%s'\n",
3396 ops[j].operand[0].name);
a652aa9f 3397 if (may_reuse && pp_tmp->argc_stack > 0)
de50b98b 3398 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
3399 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
e56ab892 3400 }
fdd5548a 3401 // esp adjust of 0 means we collected it before
3402 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP
3403 && (ops[j].operand[1].type != OPT_CONST
3404 || ops[j].operand[1].val != 0))
3405 {
89ff3147 3406 if (pp->is_unresolved)
3407 break;
3408
fdd5548a 3409 ferr(po, "arg collect %d/%d hit esp adjust of %d\n",
3410 arg, pp->argc, ops[j].operand[1].val);
de50b98b 3411 }
5e49b270 3412 else if (ops[j].op == OP_POP && !(ops[j].flags & OPF_DONE))
9af2d373 3413 {
89ff3147 3414 if (pp->is_unresolved)
3415 break;
3416
de50b98b 3417 ferr(po, "arg collect %d/%d hit pop\n", arg, pp->argc);
3418 }
5c024ef7 3419 else if (ops[j].flags & OPF_CJMP)
de50b98b 3420 {
89ff3147 3421 if (pp->is_unresolved)
3422 break;
3423
a652aa9f 3424 may_reuse = 1;
de50b98b 3425 }
91ca764a 3426 else if (ops[j].op == OP_PUSH
3427 && !(ops[j].flags & (OPF_FARGNR|OPF_DONE)))
e56ab892 3428 {
89ff3147 3429 if (pp->is_unresolved && (ops[j].flags & OPF_RMD))
3430 break;
3431
3a5101d7 3432 ops[j].p_argnext = -1;
3433 po_tmp = pp->arg[arg].datap;
3434 if (po_tmp != NULL)
3435 ops[j].p_argnext = po_tmp - ops;
e56ab892 3436 pp->arg[arg].datap = &ops[j];
3a5101d7 3437
2b43685d 3438 need_to_save_current = 0;
23fd0b11 3439 save_args = 0;
5f70a34f 3440 reg = -1;
3441 if (ops[j].operand[0].type == OPT_REG)
3442 reg = ops[j].operand[0].reg;
3443
e56ab892 3444 if (!need_op_saving) {
89ff3147 3445 ret = scan_for_mod(&ops[j], j + 1, i, 1);
2b43685d 3446 need_to_save_current = (ret >= 0);
e56ab892 3447 }
2b43685d 3448 if (need_op_saving || need_to_save_current) {
e56ab892 3449 // mark this push as one that needs operand saving
3450 ops[j].flags &= ~OPF_RMD;
5f70a34f 3451 if (ops[j].p_argnum == 0) {
3452 ops[j].p_argnum = arg + 1;
23fd0b11 3453 save_args |= 1 << arg;
de50b98b 3454 }
fdd5548a 3455 else if (ops[j].p_argnum < arg + 1) {
3456 // XXX: might kill valid var..
3457 //*save_arg_vars &= ~(1 << (ops[j].p_argnum - 1));
3458 ops[j].p_argnum = arg + 1;
3459 save_args |= 1 << arg;
3460 }
3a5101d7 3461
3462 if (save_args_seen & (1 << (ops[j].p_argnum - 1))) {
3463 save_args_seen = 0;
3464 arg_grp_current++;
3465 if (arg_grp_current >= MAX_ARG_GRP)
3466 ferr(&ops[j], "out of arg groups (arg%d), f %s\n",
3467 ops[j].p_argnum, pp->name);
3468 }
e56ab892 3469 }
5f70a34f 3470 else if (ops[j].p_argnum == 0)
e56ab892 3471 ops[j].flags |= OPF_RMD;
3472
a652aa9f 3473 // some PUSHes are reused by different calls on other branches,
de50b98b 3474 // but that can't happen if we didn't branch, so they
3475 // can be removed from future searches (handles nested calls)
a652aa9f 3476 if (!may_reuse)
9af2d373 3477 ops[j].flags |= OPF_FARGNR;
de50b98b 3478
9af2d373 3479 ops[j].flags |= OPF_FARG;
da87ae38 3480 ops[j].flags &= ~OPF_RSAVE;
3481
23fd0b11 3482 // check for __VALIST
5f70a34f 3483 if (!pp->is_unresolved && pp->arg[arg].type.is_va_list) {
23fd0b11 3484 k = -1;
92d715b6 3485 ret = resolve_origin(j, &ops[j].operand[0],
3486 magic + 1, &k, NULL);
5f70a34f 3487 if (ret == 1 && k >= 0)
23fd0b11 3488 {
5f70a34f 3489 if (ops[k].op == OP_LEA) {
3490 snprintf(buf, sizeof(buf), "arg_%X",
3491 g_func_pp->argc_stack * 4);
3492 if (!g_func_pp->is_vararg
3493 || strstr(ops[k].operand[1].name, buf))
3494 {
5e49b270 3495 ops[k].flags |= OPF_RMD | OPF_DONE;
5f70a34f 3496 ops[j].flags |= OPF_RMD | OPF_VAPUSH;
3497 save_args &= ~(1 << arg);
3498 reg = -1;
3499 }
3500 else
3501 ferr(&ops[j], "lea va_list used, but no vararg?\n");
3502 }
3503 // check for va_list from g_func_pp arg too
3504 else if (ops[k].op == OP_MOV
3505 && is_stack_access(&ops[k], &ops[k].operand[1]))
3506 {
3507 ret = stack_frame_access(&ops[k], &ops[k].operand[1],
3508 buf, sizeof(buf), ops[k].operand[1].name, "", 1, 0);
3509 if (ret >= 0) {
5e49b270 3510 ops[k].flags |= OPF_RMD | OPF_DONE;
5f70a34f 3511 ops[j].flags |= OPF_RMD;
3512 ops[j].p_argpass = ret + 1;
3513 save_args &= ~(1 << arg);
3514 reg = -1;
3515 }
3516 }
23fd0b11 3517 }
3518 }
3519
3520 *save_arg_vars |= save_args;
3521
3522 // tracking reg usage
5f70a34f 3523 if (reg >= 0)
3524 *regmask |= 1 << reg;
23fd0b11 3525
89ff3147 3526 arg++;
3527 if (!pp->is_unresolved) {
3528 // next arg
3529 for (; arg < pp->argc; arg++)
3530 if (pp->arg[arg].reg == NULL)
3531 break;
3532 }
a3684be1 3533 magic = (magic & 0xffffff) | (arg << 24);
e56ab892 3534 }
3a5101d7 3535
3536 if (ops[j].p_arggrp > arg_grp_current) {
3537 save_args_seen = 0;
3538 arg_grp_current = ops[j].p_arggrp;
3539 }
3540 if (ops[j].p_argnum > 0)
3541 save_args_seen |= 1 << (ops[j].p_argnum - 1);
e56ab892 3542 }
3543
3544 if (arg < pp->argc) {
3545 ferr(po, "arg collect failed for '%s': %d/%d\n",
3546 pp->name, arg, pp->argc);
89ff3147 3547 return -1;
e56ab892 3548 }
89ff3147 3549
3a5101d7 3550 if (arg_grp_current > *arg_grp)
3551 *arg_grp = arg_grp_current;
3552
89ff3147 3553 return arg;
3554}
3555
3556static int collect_call_args(struct parsed_op *po, int i,
3557 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
3558 int magic)
3559{
3a5101d7 3560 // arg group is for cases when pushes for
3561 // multiple funcs are going on
3562 struct parsed_op *po_tmp;
3563 int save_arg_vars_current = 0;
3564 int arg_grp = 0;
89ff3147 3565 int ret;
3566 int a;
3567
3a5101d7 3568 ret = collect_call_args_r(po, i, pp, regmask,
3569 &save_arg_vars_current, &arg_grp, 0, magic, 0, 0);
89ff3147 3570 if (ret < 0)
3571 return ret;
3572
3a5101d7 3573 if (arg_grp != 0) {
3574 // propagate arg_grp
3575 for (a = 0; a < pp->argc; a++) {
3576 if (pp->arg[a].reg != NULL)
3577 continue;
3578
3579 po_tmp = pp->arg[a].datap;
3580 while (po_tmp != NULL) {
3581 po_tmp->p_arggrp = arg_grp;
3582 if (po_tmp->p_argnext > 0)
3583 po_tmp = &ops[po_tmp->p_argnext];
3584 else
3585 po_tmp = NULL;
3586 }
3587 }
3588 }
3589 save_arg_vars[arg_grp] |= save_arg_vars_current;
3590
89ff3147 3591 if (pp->is_unresolved) {
3592 pp->argc += ret;
3593 pp->argc_stack += ret;
3594 for (a = 0; a < pp->argc; a++)
3595 if (pp->arg[a].type.name == NULL)
3596 pp->arg[a].type.name = strdup("int");
3597 }
3598
e56ab892 3599 return ret;
3600}
3601
89ff3147 3602static void pp_insert_reg_arg(struct parsed_proto *pp, const char *reg)
3603{
3604 int i;
3605
3606 for (i = 0; i < pp->argc; i++)
3607 if (pp->arg[i].reg == NULL)
3608 break;
3609
3610 if (pp->argc_stack)
3611 memmove(&pp->arg[i + 1], &pp->arg[i],
3612 sizeof(pp->arg[0]) * pp->argc_stack);
3613 memset(&pp->arg[i], 0, sizeof(pp->arg[i]));
3614 pp->arg[i].reg = strdup(reg);
3615 pp->arg[i].type.name = strdup("int");
3616 pp->argc++;
3617 pp->argc_reg++;
3618}
3619
4c45fa73 3620static void add_label_ref(struct label_ref *lr, int op_i)
3621{
3622 struct label_ref *lr_new;
3623
3624 if (lr->i == -1) {
3625 lr->i = op_i;
3626 return;
3627 }
3628
3629 lr_new = calloc(1, sizeof(*lr_new));
3630 lr_new->i = op_i;
a652aa9f 3631 lr_new->next = lr->next;
4c45fa73 3632 lr->next = lr_new;
3633}
3634
92d715b6 3635static struct parsed_data *try_resolve_jumptab(int i, int opcnt)
3636{
3637 struct parsed_op *po = &ops[i];
3638 struct parsed_data *pd;
3639 char label[NAMELEN], *p;
3640 int len, j, l;
3641
3642 p = strchr(po->operand[0].name, '[');
3643 if (p == NULL)
3644 return NULL;
3645
3646 len = p - po->operand[0].name;
3647 strncpy(label, po->operand[0].name, len);
3648 label[len] = 0;
3649
3650 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
3651 if (IS(g_func_pd[j].label, label)) {
3652 pd = &g_func_pd[j];
3653 break;
3654 }
3655 }
3656 if (pd == NULL)
3657 //ferr(po, "label '%s' not parsed?\n", label);
3658 return NULL;
3659
3660 if (pd->type != OPT_OFFSET)
3661 ferr(po, "label '%s' with non-offset data?\n", label);
3662
3663 // find all labels, link
3664 for (j = 0; j < pd->count; j++) {
3665 for (l = 0; l < opcnt; l++) {
d7857c3a 3666 if (g_labels[l] != NULL && IS(g_labels[l], pd->d[j].u.label)) {
92d715b6 3667 add_label_ref(&g_label_refs[l], i);
3668 pd->d[j].bt_i = l;
3669 break;
3670 }
3671 }
3672 }
3673
3674 return pd;
3675}
3676
d7857c3a 3677static void clear_labels(int count)
3678{
3679 int i;
3680
3681 for (i = 0; i < count; i++) {
3682 if (g_labels[i] != NULL) {
3683 free(g_labels[i]);
3684 g_labels[i] = NULL;
3685 }
3686 }
3687}
3688
04f8a628 3689static void output_std_flags(FILE *fout, struct parsed_op *po,
3690 int *pfomask, const char *dst_opr_text)
3691{
3692 if (*pfomask & (1 << PFO_Z)) {
3693 fprintf(fout, "\n cond_z = (%s%s == 0);",
3694 lmod_cast_u(po, po->operand[0].lmod), dst_opr_text);
3695 *pfomask &= ~(1 << PFO_Z);
3696 }
3697 if (*pfomask & (1 << PFO_S)) {
3698 fprintf(fout, "\n cond_s = (%s%s < 0);",
3699 lmod_cast_s(po, po->operand[0].lmod), dst_opr_text);
3700 *pfomask &= ~(1 << PFO_S);
3701 }
3702}
3703
c0de9015 3704enum {
3705 OPP_FORCE_NORETURN = (1 << 0),
3706 OPP_SIMPLE_ARGS = (1 << 1),
3707 OPP_ALIGN = (1 << 2),
3708};
3709
c0050df6 3710static void output_pp_attrs(FILE *fout, const struct parsed_proto *pp,
c0de9015 3711 int flags)
c0050df6 3712{
c0de9015 3713 const char *cconv = "";
3714
c0050df6 3715 if (pp->is_fastcall)
c0de9015 3716 cconv = "__fastcall ";
c0050df6 3717 else if (pp->is_stdcall && pp->argc_reg == 0)
c0de9015 3718 cconv = "__stdcall ";
3719
3720 fprintf(fout, (flags & OPP_ALIGN) ? "%-16s" : "%s", cconv);
3721
3722 if (pp->is_noreturn || (flags & OPP_FORCE_NORETURN))
c0050df6 3723 fprintf(fout, "noreturn ");
3724}
3725
c0de9015 3726static void output_pp(FILE *fout, const struct parsed_proto *pp,
3727 int flags)
3728{
3729 int i;
3730
3731 fprintf(fout, (flags & OPP_ALIGN) ? "%-5s" : "%s ",
3732 pp->ret_type.name);
3733 if (pp->is_fptr)
3734 fprintf(fout, "(");
3735 output_pp_attrs(fout, pp, flags);
3736 if (pp->is_fptr)
3737 fprintf(fout, "*");
3738 fprintf(fout, "%s", pp->name);
3739 if (pp->is_fptr)
3740 fprintf(fout, ")");
3741
3742 fprintf(fout, "(");
3743 for (i = 0; i < pp->argc; i++) {
3744 if (i > 0)
3745 fprintf(fout, ", ");
3746 if (pp->arg[i].fptr != NULL && !(flags & OPP_SIMPLE_ARGS)) {
3747 // func pointer
3748 output_pp(fout, pp->arg[i].fptr, 0);
3749 }
3750 else if (pp->arg[i].type.is_retreg) {
3751 fprintf(fout, "u32 *r_%s", pp->arg[i].reg);
3752 }
3753 else {
3754 fprintf(fout, "%s", pp->arg[i].type.name);
3755 if (!pp->is_fptr)
3756 fprintf(fout, " a%d", i + 1);
3757 }
3758 }
3759 if (pp->is_vararg) {
3760 if (i > 0)
3761 fprintf(fout, ", ");
3762 fprintf(fout, "...");
3763 }
3764 fprintf(fout, ")");
3765}
3766
9af2d373 3767static int get_pp_arg_regmask(const struct parsed_proto *pp)
3768{
3769 int regmask = 0;
3770 int i, reg;
3771
3772 for (i = 0; i < pp->argc; i++) {
3773 if (pp->arg[i].reg != NULL) {
3774 reg = char_array_i(regs_r32,
3775 ARRAY_SIZE(regs_r32), pp->arg[i].reg);
3776 if (reg < 0)
3777 ferr(ops, "arg '%s' of func '%s' is not a reg?\n",
3778 pp->arg[i].reg, pp->name);
3779 regmask |= 1 << reg;
3780 }
3781 }
3782
3783 return regmask;
3784}
3785
3a5101d7 3786static char *saved_arg_name(char *buf, size_t buf_size, int grp, int num)
3787{
3788 char buf1[16];
3789
3790 buf1[0] = 0;
3791 if (grp > 0)
3792 snprintf(buf1, sizeof(buf1), "%d", grp);
3793 snprintf(buf, buf_size, "s%s_a%d", buf1, num);
3794
3795 return buf;
3796}
3797
9af2d373 3798static void gen_x_cleanup(int opcnt);
3799
91977a1c 3800static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
3801{
69a3cdfc 3802 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
850c9265 3803 struct parsed_opr *last_arith_dst = NULL;
3ebea2cf 3804 char buf1[256], buf2[256], buf3[256], cast[64];
bd96f656 3805 const struct parsed_proto *pp_c;
ddaf8bd7 3806 struct parsed_proto *pp, *pp_tmp;
4c45fa73 3807 struct parsed_data *pd;
91977a1c 3808 const char *tmpname;
1f84f6b3 3809 unsigned int uval;
3a5101d7 3810 int save_arg_vars[MAX_ARG_GRP] = { 0, };
cb090db0 3811 int cond_vars = 0;
108e9fe3 3812 int need_tmp_var = 0;
2fe80fdb 3813 int need_tmp64 = 0;
91977a1c 3814 int had_decl = 0;
3ebea2cf 3815 int label_pending = 0;
d4e3b5db 3816 int regmask_save = 0;
91977a1c 3817 int regmask_arg = 0;
a3684be1 3818 int regmask_now = 0;
ddaf8bd7 3819 int regmask_init = 0;
91977a1c 3820 int regmask = 0;
940e8e66 3821 int pfomask = 0;
64c59faf 3822 int found = 0;
d4e3b5db 3823 int depth = 0;
91977a1c 3824 int no_output;
4c45fa73 3825 int i, j, l;
91977a1c 3826 int arg;
91977a1c 3827 int reg;
3828 int ret;
3829
1bafb621 3830 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
a2c1d768 3831 g_stack_frame_used = 0;
91977a1c 3832
36595fd2 3833 g_func_pp = proto_parse(fhdr, funcn, 0);
bd96f656 3834 if (g_func_pp == NULL)
91977a1c 3835 ferr(ops, "proto_parse failed for '%s'\n", funcn);
3836
9af2d373 3837 regmask_arg = get_pp_arg_regmask(g_func_pp);
91977a1c 3838
3839 // pass1:
1bafb621 3840 // - handle ebp/esp frame, remove ops related to it
9af2d373 3841 scan_prologue_epilogue(opcnt);
91977a1c 3842
3843 // pass2:
7ae48d73 3844 // - parse calls with labels
87bf6cec 3845 // - resolve all branches
de50b98b 3846 for (i = 0; i < opcnt; i++)
3847 {
87bf6cec 3848 po = &ops[i];
3849 po->bt_i = -1;
4c45fa73 3850 po->btj = NULL;
87bf6cec 3851
5e49b270 3852 if (po->flags & (OPF_RMD|OPF_DONE))
7ae48d73 3853 continue;
3854
3855 if (po->op == OP_CALL) {
ddaf8bd7 3856 pp = NULL;
3857
7ae48d73 3858 if (po->operand[0].type == OPT_LABEL) {
3859 tmpname = opr_name(po, 0);
46411e6c 3860 if (IS_START(tmpname, "loc_"))
3861 ferr(po, "call to loc_*\n");
36595fd2 3862 pp_c = proto_parse(fhdr, tmpname, 0);
7ae48d73 3863 if (pp_c == NULL)
3864 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
da87ae38 3865
7ae48d73 3866 pp = proto_clone(pp_c);
3867 my_assert_not(pp, NULL);
ddaf8bd7 3868 }
3869 else if (po->datap != NULL) {
3870 pp = calloc(1, sizeof(*pp));
3871 my_assert_not(pp, NULL);
3872
3873 ret = parse_protostr(po->datap, pp);
3874 if (ret < 0)
3875 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
3876 free(po->datap);
3877 po->datap = NULL;
3878 }
3879
3880 if (pp != NULL) {
3881 if (pp->is_fptr)
3882 check_func_pp(po, pp, "fptr var call");
3883 if (pp->is_noreturn)
3884 po->flags |= OPF_TAIL;
7ae48d73 3885 }
092f64e1 3886 po->pp = pp;
7ae48d73 3887 continue;
3888 }
3889
3890 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
87bf6cec 3891 continue;
3892
4c45fa73 3893 if (po->operand[0].type == OPT_REGMEM) {
92d715b6 3894 pd = try_resolve_jumptab(i, opcnt);
4c45fa73 3895 if (pd == NULL)
840257f6 3896 goto tailcall;
4c45fa73 3897
3898 po->btj = pd;
3899 continue;
3900 }
3901
3902 for (l = 0; l < opcnt; l++) {
d7857c3a 3903 if (g_labels[l] != NULL
3904 && IS(po->operand[0].name, g_labels[l]))
3905 {
092f64e1 3906 if (l == i + 1 && po->op == OP_JMP) {
3907 // yet another alignment type..
5e49b270 3908 po->flags |= OPF_RMD|OPF_DONE;
092f64e1 3909 break;
3910 }
4c45fa73 3911 add_label_ref(&g_label_refs[l], i);
3912 po->bt_i = l;
87bf6cec 3913 break;
3914 }
3915 }
3916
092f64e1 3917 if (po->bt_i != -1 || (po->flags & OPF_RMD))
4c45fa73 3918 continue;
3919
840257f6 3920 if (po->operand[0].type == OPT_LABEL)
87bf6cec 3921 // assume tail call
840257f6 3922 goto tailcall;
4c45fa73 3923
3924 ferr(po, "unhandled branch\n");
840257f6 3925
3926tailcall:
3927 po->op = OP_CALL;
3928 po->flags |= OPF_TAIL;
2fe80fdb 3929 if (i > 0 && ops[i - 1].op == OP_POP)
3930 po->flags |= OPF_ATAIL;
7ae48d73 3931 i--; // reprocess
87bf6cec 3932 }
3933
3934 // pass3:
a2c1d768 3935 // - remove dead labels
26677139 3936 // - process trivial calls
1bafb621 3937 for (i = 0; i < opcnt; i++)
3938 {
d7857c3a 3939 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
3940 free(g_labels[i]);
3941 g_labels[i] = NULL;
3942 }
a2c1d768 3943
69a3cdfc 3944 po = &ops[i];
5e49b270 3945 if (po->flags & (OPF_RMD|OPF_DONE))
91977a1c 3946 continue;
850c9265 3947
d4e3b5db 3948 if (po->op == OP_CALL)
26677139 3949 {
3950 pp = process_call_early(i, opcnt, &j);
3951 if (pp != NULL) {
3952 if (!(po->flags & OPF_ATAIL))
3953 // since we know the args, try to collect them
3954 if (collect_call_args_early(po, i, pp, &regmask) != 0)
3955 pp = NULL;
3956 }
3957
3958 if (pp != NULL) {
3959 if (j >= 0) {
3960 // commit esp adjust
3961 ops[j].flags |= OPF_RMD;
5e49b270 3962 if (ops[j].op != OP_POP)
3963 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
3964 else
3965 ops[j].flags |= OPF_DONE;
26677139 3966 }
3967
3968 if (strstr(pp->ret_type.name, "int64"))
3969 need_tmp64 = 1;
3970
3971 po->flags |= OPF_DONE;
3972 }
3973 }
3974 }
3975
3976 // pass4:
3977 // - process calls
ee2361b9 3978 // - handle push <const>/pop pairs
26677139 3979 for (i = 0; i < opcnt; i++)
3980 {
3981 po = &ops[i];
5e49b270 3982 if (po->flags & (OPF_RMD|OPF_DONE))
26677139 3983 continue;
3984
3985 if (po->op == OP_CALL && !(po->flags & OPF_DONE))
69a3cdfc 3986 {
9af2d373 3987 pp = process_call(i, opcnt);
91977a1c 3988
2fe80fdb 3989 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
89ff3147 3990 // since we know the args, collect them
3a5101d7 3991 collect_call_args(po, i, pp, &regmask, save_arg_vars,
89ff3147 3992 i + opcnt * 2);
3993 }
2b43685d 3994
3995 if (strstr(pp->ret_type.name, "int64"))
2fe80fdb 3996 need_tmp64 = 1;
91977a1c 3997 }
ee2361b9 3998 else if (po->op == OP_PUSH && !(po->flags & OPF_FARG)
3999 && !(po->flags & OPF_RSAVE) && po->operand[0].type == OPT_CONST)
4000 scan_for_pop_const(i, opcnt);
d4e3b5db 4001 }
4002
26677139 4003 // pass5:
d4e3b5db 4004 // - find POPs for PUSHes, rm both
591721d7 4005 // - scan for STD/CLD, propagate DF
d4e3b5db 4006 // - scan for all used registers
4007 // - find flag set ops for their users
89ff3147 4008 // - do unreselved calls
1bafb621 4009 // - declare indirect functions
26677139 4010 for (i = 0; i < opcnt; i++)
4011 {
d4e3b5db 4012 po = &ops[i];
5e49b270 4013 if (po->flags & (OPF_RMD|OPF_DONE))
d4e3b5db 4014 continue;
4015
7ae48d73 4016 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE)) {
4017 reg = po->operand[0].reg;
4018 if (!(regmask & (1 << reg)))
4019 // not a reg save after all, rerun scan_for_pop
4020 po->flags &= ~OPF_RSAVE;
4021 else
4022 regmask_save |= 1 << reg;
4023 }
4024
9af2d373 4025 if (po->op == OP_PUSH && !(po->flags & OPF_FARG)
1f84f6b3 4026 && !(po->flags & OPF_RSAVE) && !g_func_pp->is_userstack)
d4e3b5db 4027 {
5c024ef7 4028 if (po->operand[0].type == OPT_REG)
4029 {
4030 reg = po->operand[0].reg;
4031 if (reg < 0)
4032 ferr(po, "reg not set for push?\n");
4033
4034 depth = 0;
4035 ret = scan_for_pop(i + 1, opcnt,
4036 po->operand[0].name, i + opcnt * 3, 0, &depth, 0);
4037 if (ret == 1) {
4038 if (depth > 1)
4039 ferr(po, "too much depth: %d\n", depth);
4040
4041 po->flags |= OPF_RMD;
4042 scan_for_pop(i + 1, opcnt, po->operand[0].name,
4043 i + opcnt * 4, 0, &depth, 1);
4044 continue;
4045 }
4046 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
4047 if (ret == 0) {
4048 arg = OPF_RMD;
4049 if (regmask & (1 << reg)) {
4050 if (regmask_save & (1 << reg))
4051 ferr(po, "%s already saved?\n", po->operand[0].name);
4052 arg = OPF_RSAVE;
4053 }
4054 po->flags |= arg;
4055 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg);
4056 continue;
4057 }
d4e3b5db 4058 }
d4e3b5db 4059 }
4060
591721d7 4061 if (po->op == OP_STD) {
5e49b270 4062 po->flags |= OPF_DF | OPF_RMD | OPF_DONE;
591721d7 4063 scan_propagate_df(i + 1, opcnt);
4064 }
4065
a3684be1 4066 regmask_now = po->regmask_src | po->regmask_dst;
4067 if (regmask_now & (1 << xBP)) {
4068 if (g_bp_frame && !(po->flags & OPF_EBP_S)) {
4069 if (po->regmask_dst & (1 << xBP))
4070 // compiler decided to drop bp frame and use ebp as scratch
037f4971 4071 scan_fwd_set_flags(i + 1, opcnt, i + opcnt * 5, OPF_EBP_S);
a3684be1 4072 else
4073 regmask_now &= ~(1 << xBP);
4074 }
4075 }
4076
4077 regmask |= regmask_now;
d4e3b5db 4078
4079 if (po->flags & OPF_CC)
4080 {
2b43685d 4081 int setters[16], cnt = 0, branched = 0;
4082
04f8a628 4083 ret = scan_for_flag_set(i, i + opcnt * 6,
4084 &branched, setters, &cnt);
2b43685d 4085 if (ret < 0 || cnt <= 0)
4086 ferr(po, "unable to trace flag setter(s)\n");
4087 if (cnt > ARRAY_SIZE(setters))
4088 ferr(po, "too many flag setters\n");
d4e3b5db 4089
2b43685d 4090 for (j = 0; j < cnt; j++)
4091 {
4092 tmp_op = &ops[setters[j]]; // flag setter
4093 pfomask = 0;
4094
4095 // to get nicer code, we try to delay test and cmp;
4096 // if we can't because of operand modification, or if we
591721d7 4097 // have arith op, or branch, make it calculate flags explicitly
4098 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP)
4099 {
89ff3147 4100 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i, 0) >= 0)
092f64e1 4101 pfomask = 1 << po->pfo;
2b43685d 4102 }
4741fdfe 4103 else if (tmp_op->op == OP_CMPS || tmp_op->op == OP_SCAS) {
092f64e1 4104 pfomask = 1 << po->pfo;
591721d7 4105 }
2b43685d 4106 else {
04f8a628 4107 // see if we'll be able to handle based on op result
4108 if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR
092f64e1 4109 && po->pfo != PFO_Z && po->pfo != PFO_S
4110 && po->pfo != PFO_P)
04f8a628 4111 || branched
2b43685d 4112 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
092f64e1 4113 {
4114 pfomask = 1 << po->pfo;
4115 }
2fe80fdb 4116
c8dbc5be 4117 if (tmp_op->op == OP_ADD && po->pfo == PFO_C) {
4118 propagate_lmod(tmp_op, &tmp_op->operand[0],
4119 &tmp_op->operand[1]);
4120 if (tmp_op->operand[0].lmod == OPLM_DWORD)
4121 need_tmp64 = 1;
4122 }
2b43685d 4123 }
4124 if (pfomask) {
4125 tmp_op->pfomask |= pfomask;
cb090db0 4126 cond_vars |= pfomask;
2b43685d 4127 }
04f8a628 4128 // note: may overwrite, currently not a problem
4129 po->datap = tmp_op;
d4e3b5db 4130 }
4131
cb090db0 4132 if (po->op == OP_RCL || po->op == OP_RCR
4133 || po->op == OP_ADC || po->op == OP_SBB)
4134 cond_vars |= 1 << PFO_C;
d4e3b5db 4135 }
092f64e1 4136
4137 if (po->op == OP_CMPS || po->op == OP_SCAS) {
cb090db0 4138 cond_vars |= 1 << PFO_Z;
591721d7 4139 }
87bf6cec 4140 else if (po->op == OP_MUL
4141 || (po->op == OP_IMUL && po->operand_cnt == 1))
4142 {
c8dbc5be 4143 if (po->operand[0].lmod == OPLM_DWORD)
4144 need_tmp64 = 1;
87bf6cec 4145 }
89ff3147 4146 else if (po->op == OP_CALL) {
26677139 4147 // note: resolved non-reg calls are OPF_DONE already
092f64e1 4148 pp = po->pp;
89ff3147 4149 if (pp == NULL)
4150 ferr(po, "NULL pp\n");
4151
4152 if (pp->is_unresolved) {
ddaf8bd7 4153 int regmask_stack = 0;
3a5101d7 4154 collect_call_args(po, i, pp, &regmask, save_arg_vars,
89ff3147 4155 i + opcnt * 2);
4156
b74c31e3 4157 // this is pretty rough guess:
4158 // see ecx and edx were pushed (and not their saved versions)
4159 for (arg = 0; arg < pp->argc; arg++) {
4160 if (pp->arg[arg].reg != NULL)
4161 continue;
4162
4163 tmp_op = pp->arg[arg].datap;
4164 if (tmp_op == NULL)
4165 ferr(po, "parsed_op missing for arg%d\n", arg);
5f70a34f 4166 if (tmp_op->p_argnum == 0 && tmp_op->operand[0].type == OPT_REG)
b74c31e3 4167 regmask_stack |= 1 << tmp_op->operand[0].reg;
4168 }
4169
ddaf8bd7 4170 if (!((regmask_stack & (1 << xCX))
4171 && (regmask_stack & (1 << xDX))))
89ff3147 4172 {
4173 if (pp->argc_stack != 0
c0050df6 4174 || ((regmask | regmask_arg) & ((1 << xCX)|(1 << xDX))))
89ff3147 4175 {
4176 pp_insert_reg_arg(pp, "ecx");
c0050df6 4177 pp->is_fastcall = 1;
ddaf8bd7 4178 regmask_init |= 1 << xCX;
89ff3147 4179 regmask |= 1 << xCX;
4180 }
4181 if (pp->argc_stack != 0
4182 || ((regmask | regmask_arg) & (1 << xDX)))
4183 {
4184 pp_insert_reg_arg(pp, "edx");
ddaf8bd7 4185 regmask_init |= 1 << xDX;
89ff3147 4186 regmask |= 1 << xDX;
4187 }
4188 }
c0050df6 4189
4190 // note: __cdecl doesn't fall into is_unresolved category
4191 if (pp->argc_stack > 0)
4192 pp->is_stdcall = 1;
ddaf8bd7 4193 }
4194
4195 for (arg = 0; arg < pp->argc; arg++) {
4196 if (pp->arg[arg].reg != NULL) {
4197 reg = char_array_i(regs_r32,
4198 ARRAY_SIZE(regs_r32), pp->arg[arg].reg);
4199 if (reg < 0)
4200 ferr(ops, "arg '%s' is not a reg?\n", pp->arg[arg].reg);
4201 if (!(regmask & (1 << reg))) {
4202 regmask_init |= 1 << reg;
4203 regmask |= 1 << reg;
4204 }
4205 }
89ff3147 4206 }
1bafb621 4207 }
27ebfaed 4208 else if (po->op == OP_MOV && po->operand[0].pp != NULL
4209 && po->operand[1].pp != NULL)
4210 {
4211 // <var> = offset <something>
4212 if ((po->operand[1].pp->is_func || po->operand[1].pp->is_fptr)
4213 && !IS_START(po->operand[1].name, "off_"))
4214 {
4215 if (!po->operand[0].pp->is_fptr)
4216 ferr(po, "%s not declared as fptr when it should be\n",
4217 po->operand[0].name);
4218 if (pp_cmp_func(po->operand[0].pp, po->operand[1].pp)) {
4219 pp_print(buf1, sizeof(buf1), po->operand[0].pp);
4220 pp_print(buf2, sizeof(buf2), po->operand[1].pp);
4221 fnote(po, "var: %s\n", buf1);
4222 fnote(po, "func: %s\n", buf2);
4223 ferr(po, "^ mismatch\n");
4224 }
4225 }
4226 }
75ad0378 4227 else if (po->op == OP_RET && !IS(g_func_pp->ret_type.name, "void"))
4228 regmask |= 1 << xAX;
cb090db0 4229 else if (po->op == OP_DIV || po->op == OP_IDIV) {
4230 // 32bit division is common, look for it
4231 if (po->op == OP_DIV)
4232 ret = scan_for_reg_clear(i, xDX);
4233 else
4234 ret = scan_for_cdq_edx(i);
4235 if (ret >= 0)
4236 po->flags |= OPF_32BIT;
4237 else
4238 need_tmp64 = 1;
4239 }
037f4971 4240 else if (po->op == OP_CLD)
5e49b270 4241 po->flags |= OPF_RMD | OPF_DONE;
cb090db0 4242
4243 if (po->op == OP_RCL || po->op == OP_RCR || po->op == OP_XCHG) {
4244 need_tmp_var = 1;
4245 }
91977a1c 4246 }
4247
26677139 4248 // pass6:
da87ae38 4249 // - confirm regmask_save, it might have been reduced
4250 if (regmask_save != 0)
4251 {
4252 regmask_save = 0;
4253 for (i = 0; i < opcnt; i++) {
4254 po = &ops[i];
4255 if (po->flags & OPF_RMD)
4256 continue;
4257
4258 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE))
4259 regmask_save |= 1 << po->operand[0].reg;
4260 }
4261 }
4262
60fe410c 4263 // output starts here
4264
4265 // define userstack size
4266 if (g_func_pp->is_userstack) {
4267 fprintf(fout, "#ifndef US_SZ_%s\n", g_func_pp->name);
4268 fprintf(fout, "#define US_SZ_%s USERSTACK_SIZE\n", g_func_pp->name);
4269 fprintf(fout, "#endif\n");
4270 }
4271
4272 // the function itself
c0de9015 4273 ferr_assert(ops, !g_func_pp->is_fptr);
4274 output_pp(fout, g_func_pp,
4275 (g_ida_func_attr & IDAFA_NORETURN) ? OPP_FORCE_NORETURN : 0);
4276 fprintf(fout, "\n{\n");
60fe410c 4277
4278 // declare indirect functions
4279 for (i = 0; i < opcnt; i++) {
4280 po = &ops[i];
4281 if (po->flags & OPF_RMD)
4282 continue;
4283
4284 if (po->op == OP_CALL) {
4285 pp = po->pp;
4286 if (pp == NULL)
4287 ferr(po, "NULL pp\n");
4288
4289 if (pp->is_fptr && !(pp->name[0] != 0 && pp->is_arg)) {
4290 if (pp->name[0] != 0) {
4291 memmove(pp->name + 2, pp->name, strlen(pp->name) + 1);
4292 memcpy(pp->name, "i_", 2);
4293
4294 // might be declared already
4295 found = 0;
4296 for (j = 0; j < i; j++) {
4297 if (ops[j].op == OP_CALL && (pp_tmp = ops[j].pp)) {
4298 if (pp_tmp->is_fptr && IS(pp->name, pp_tmp->name)) {
4299 found = 1;
4300 break;
4301 }
4302 }
4303 }
4304 if (found)
4305 continue;
4306 }
4307 else
4308 snprintf(pp->name, sizeof(pp->name), "icall%d", i);
4309
c0de9015 4310 fprintf(fout, " ");
4311 output_pp(fout, pp, OPP_SIMPLE_ARGS);
4312 fprintf(fout, ";\n");
60fe410c 4313 }
4314 }
4315 }
da87ae38 4316
4c45fa73 4317 // output LUTs/jumptables
4318 for (i = 0; i < g_func_pd_cnt; i++) {
4319 pd = &g_func_pd[i];
4320 fprintf(fout, " static const ");
4321 if (pd->type == OPT_OFFSET) {
4322 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
4323
4324 for (j = 0; j < pd->count; j++) {
4325 if (j > 0)
4326 fprintf(fout, ", ");
4327 fprintf(fout, "&&%s", pd->d[j].u.label);
4328 }
4329 }
4330 else {
4331 fprintf(fout, "%s %s[] =\n { ",
4332 lmod_type_u(ops, pd->lmod), pd->label);
4333
4334 for (j = 0; j < pd->count; j++) {
4335 if (j > 0)
4336 fprintf(fout, ", ");
4337 fprintf(fout, "%u", pd->d[j].u.val);
4338 }
4339 }
4340 fprintf(fout, " };\n");
1f84f6b3 4341 had_decl = 1;
4c45fa73 4342 }
4343
4f12f671 4344 // declare stack frame, va_arg
1f84f6b3 4345 if (g_stack_fsz) {
850c9265 4346 fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
1bafb621 4347 (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz);
1f84f6b3 4348 had_decl = 1;
4349 }
4350
4351 if (g_func_pp->is_userstack) {
60fe410c 4352 fprintf(fout, " u32 fake_sf[US_SZ_%s / 4];\n", g_func_pp->name);
4353 fprintf(fout, " u32 *esp = &fake_sf[sizeof(fake_sf) / 4];\n");
1f84f6b3 4354 had_decl = 1;
4355 }
850c9265 4356
1f84f6b3 4357 if (g_func_pp->is_vararg) {
4f12f671 4358 fprintf(fout, " va_list ap;\n");
1f84f6b3 4359 had_decl = 1;
4360 }
4f12f671 4361
940e8e66 4362 // declare arg-registers
bd96f656 4363 for (i = 0; i < g_func_pp->argc; i++) {
4364 if (g_func_pp->arg[i].reg != NULL) {
91977a1c 4365 reg = char_array_i(regs_r32,
bd96f656 4366 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
75ad0378 4367 if (regmask & (1 << reg)) {
1f84f6b3 4368 if (g_func_pp->arg[i].type.is_retreg)
4369 fprintf(fout, " u32 %s = *r_%s;\n",
4370 g_func_pp->arg[i].reg, g_func_pp->arg[i].reg);
4371 else
4372 fprintf(fout, " u32 %s = (u32)a%d;\n",
4373 g_func_pp->arg[i].reg, i + 1);
75ad0378 4374 }
1f84f6b3 4375 else {
4376 if (g_func_pp->arg[i].type.is_retreg)
4377 ferr(ops, "retreg '%s' is unused?\n",
4378 g_func_pp->arg[i].reg);
75ad0378 4379 fprintf(fout, " // %s = a%d; // unused\n",
4380 g_func_pp->arg[i].reg, i + 1);
1f84f6b3 4381 }
91977a1c 4382 had_decl = 1;
4383 }
4384 }
4385
75ad0378 4386 regmask_now = regmask & ~regmask_arg;
4387 regmask_now &= ~(1 << xSP);
90307a99 4388 if (regmask_now & 0x00ff) {
91977a1c 4389 for (reg = 0; reg < 8; reg++) {
75ad0378 4390 if (regmask_now & (1 << reg)) {
ddaf8bd7 4391 fprintf(fout, " u32 %s", regs_r32[reg]);
4392 if (regmask_init & (1 << reg))
4393 fprintf(fout, " = 0");
4394 fprintf(fout, ";\n");
91977a1c 4395 had_decl = 1;
4396 }
4397 }
4398 }
90307a99 4399 if (regmask_now & 0xff00) {
4400 for (reg = 8; reg < 16; reg++) {
4401 if (regmask_now & (1 << reg)) {
4402 fprintf(fout, " mmxr %s", regs_r32[reg]);
4403 if (regmask_init & (1 << reg))
4404 fprintf(fout, " = { 0, }");
4405 fprintf(fout, ";\n");
4406 had_decl = 1;
4407 }
4408 }
4409 }
91977a1c 4410
d4e3b5db 4411 if (regmask_save) {
4412 for (reg = 0; reg < 8; reg++) {
4413 if (regmask_save & (1 << reg)) {
4414 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
4415 had_decl = 1;
4416 }
4417 }
4418 }
4419
3a5101d7 4420 for (i = 0; i < ARRAY_SIZE(save_arg_vars); i++) {
4421 if (save_arg_vars[i] == 0)
4422 continue;
69a3cdfc 4423 for (reg = 0; reg < 32; reg++) {
3a5101d7 4424 if (save_arg_vars[i] & (1 << reg)) {
4425 fprintf(fout, " u32 %s;\n",
4426 saved_arg_name(buf1, sizeof(buf1), i, reg + 1));
69a3cdfc 4427 had_decl = 1;
4428 }
4429 }
4430 }
4431
cb090db0 4432 if (cond_vars) {
69a3cdfc 4433 for (i = 0; i < 8; i++) {
cb090db0 4434 if (cond_vars & (1 << i)) {
69a3cdfc 4435 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
4436 had_decl = 1;
4437 }
4438 }
4439 }
4440
108e9fe3 4441 if (need_tmp_var) {
4442 fprintf(fout, " u32 tmp;\n");
4443 had_decl = 1;
4444 }
4445
2fe80fdb 4446 if (need_tmp64) {
4447 fprintf(fout, " u64 tmp64;\n");
87bf6cec 4448 had_decl = 1;
4449 }
4450
91977a1c 4451 if (had_decl)
4452 fprintf(fout, "\n");
4453
bd96f656 4454 if (g_func_pp->is_vararg) {
4455 if (g_func_pp->argc_stack == 0)
4f12f671 4456 ferr(ops, "vararg func without stack args?\n");
bd96f656 4457 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
4f12f671 4458 }
4459
91977a1c 4460 // output ops
69a3cdfc 4461 for (i = 0; i < opcnt; i++)
4462 {
d7857c3a 4463 if (g_labels[i] != NULL) {
91977a1c 4464 fprintf(fout, "\n%s:\n", g_labels[i]);
3ebea2cf 4465 label_pending = 1;
2b43685d 4466
4467 delayed_flag_op = NULL;
4468 last_arith_dst = NULL;
3ebea2cf 4469 }
91977a1c 4470
69a3cdfc 4471 po = &ops[i];
4472 if (po->flags & OPF_RMD)
91977a1c 4473 continue;
4474
4475 no_output = 0;
4476
91977a1c 4477 #define assert_operand_cnt(n_) \
850c9265 4478 if (po->operand_cnt != n_) \
4479 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
4480
69a3cdfc 4481 // conditional/flag using op?
4482 if (po->flags & OPF_CC)
850c9265 4483 {
940e8e66 4484 int is_delayed = 0;
69a3cdfc 4485
04f8a628 4486 tmp_op = po->datap;
850c9265 4487
69a3cdfc 4488 // we go through all this trouble to avoid using parsed_flag_op,
4489 // which makes generated code much nicer
4490 if (delayed_flag_op != NULL)
850c9265 4491 {
092f64e1 4492 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op,
4493 po->pfo, po->pfo_inv);
940e8e66 4494 is_delayed = 1;
91977a1c 4495 }
850c9265 4496 else if (last_arith_dst != NULL
092f64e1 4497 && (po->pfo == PFO_Z || po->pfo == PFO_S || po->pfo == PFO_P
04f8a628 4498 || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR))
4499 ))
850c9265 4500 {
3ebea2cf 4501 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
092f64e1 4502 out_test_for_cc(buf1, sizeof(buf1), po, po->pfo, po->pfo_inv,
850c9265 4503 last_arith_dst->lmod, buf3);
940e8e66 4504 is_delayed = 1;
850c9265 4505 }
04f8a628 4506 else if (tmp_op != NULL) {
7ba45c34 4507 // use preprocessed flag calc results
092f64e1 4508 if (!(tmp_op->pfomask & (1 << po->pfo)))
4509 ferr(po, "not prepared for pfo %d\n", po->pfo);
69a3cdfc 4510
092f64e1 4511 // note: pfo_inv was not yet applied
69a3cdfc 4512 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
092f64e1 4513 po->pfo_inv ? "!" : "", parsed_flag_op_names[po->pfo]);
69a3cdfc 4514 }
4515 else {
4516 ferr(po, "all methods of finding comparison failed\n");
4517 }
850c9265 4518
69a3cdfc 4519 if (po->flags & OPF_JMP) {
092f64e1 4520 fprintf(fout, " if %s", buf1);
850c9265 4521 }
cb090db0 4522 else if (po->op == OP_RCL || po->op == OP_RCR
4523 || po->op == OP_ADC || po->op == OP_SBB)
4524 {
940e8e66 4525 if (is_delayed)
4526 fprintf(fout, " cond_%s = %s;\n",
092f64e1 4527 parsed_flag_op_names[po->pfo], buf1);
850c9265 4528 }
5101a5f9 4529 else if (po->flags & OPF_DATA) { // SETcc
850c9265 4530 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
4531 fprintf(fout, " %s = %s;", buf2, buf1);
91977a1c 4532 }
69a3cdfc 4533 else {
4534 ferr(po, "unhandled conditional op\n");
4535 }
91977a1c 4536 }
4537
940e8e66 4538 pfomask = po->pfomask;
4539
4741fdfe 4540 if (po->flags & (OPF_REPZ|OPF_REPNZ)) {
1f84f6b3 4541 struct parsed_opr opr = {0,};
4542 opr.type = OPT_REG;
4543 opr.reg = xCX;
4544 opr.lmod = OPLM_DWORD;
4545 ret = try_resolve_const(i, &opr, opcnt * 7 + i, &uval);
4546
4547 if (ret != 1 || uval == 0) {
4548 // we need initial flags for ecx=0 case..
4549 if (i > 0 && ops[i - 1].op == OP_XOR
4550 && IS(ops[i - 1].operand[0].name,
4551 ops[i - 1].operand[1].name))
4552 {
4553 fprintf(fout, " cond_z = ");
4554 if (pfomask & (1 << PFO_C))
4555 fprintf(fout, "cond_c = ");
4556 fprintf(fout, "0;\n");
4557 }
4558 else if (last_arith_dst != NULL) {
4559 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
4560 out_test_for_cc(buf1, sizeof(buf1), po, PFO_Z, 0,
4561 last_arith_dst->lmod, buf3);
4562 fprintf(fout, " cond_z = %s;\n", buf1);
4563 }
4564 else
4565 ferr(po, "missing initial ZF\n");
4741fdfe 4566 }
4741fdfe 4567 }
4568
850c9265 4569 switch (po->op)
91977a1c 4570 {
4571 case OP_MOV:
4572 assert_operand_cnt(2);
850c9265 4573 propagate_lmod(po, &po->operand[0], &po->operand[1]);
de50b98b 4574 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
c7ed83dd 4575 default_cast_to(buf3, sizeof(buf3), &po->operand[0]);
de50b98b 4576 fprintf(fout, " %s = %s;", buf1,
3ebea2cf 4577 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
c7ed83dd 4578 buf3, 0));
850c9265 4579 break;
4580
4581 case OP_LEA:
4582 assert_operand_cnt(2);
87bf6cec 4583 po->operand[1].lmod = OPLM_DWORD; // always
850c9265 4584 fprintf(fout, " %s = %s;",
4585 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3ebea2cf 4586 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4587 NULL, 1));
850c9265 4588 break;
4589
4590 case OP_MOVZX:
4591 assert_operand_cnt(2);
91977a1c 4592 fprintf(fout, " %s = %s;",
850c9265 4593 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3ebea2cf 4594 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
850c9265 4595 break;
4596
4597 case OP_MOVSX:
4598 assert_operand_cnt(2);
4599 switch (po->operand[1].lmod) {
4600 case OPLM_BYTE:
4601 strcpy(buf3, "(s8)");
4602 break;
4603 case OPLM_WORD:
4604 strcpy(buf3, "(s16)");
4605 break;
4606 default:
4607 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
4608 }
a2c1d768 4609 fprintf(fout, " %s = %s;",
850c9265 4610 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
a2c1d768 4611 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4612 buf3, 0));
850c9265 4613 break;
4614
108e9fe3 4615 case OP_XCHG:
4616 assert_operand_cnt(2);
4617 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4618 fprintf(fout, " tmp = %s;",
4619 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], "", 0));
4620 fprintf(fout, " %s = %s;",
4621 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
c7ed83dd 4622 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4623 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
4624 fprintf(fout, " %s = %stmp;",
4625 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]),
4626 default_cast_to(buf3, sizeof(buf3), &po->operand[1]));
108e9fe3 4627 snprintf(g_comment, sizeof(g_comment), "xchg");
4628 break;
4629
850c9265 4630 case OP_NOT:
4631 assert_operand_cnt(1);
4632 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4633 fprintf(fout, " %s = ~%s;", buf1, buf1);
4634 break;
4635
5101a5f9 4636 case OP_CDQ:
4637 assert_operand_cnt(2);
4638 fprintf(fout, " %s = (s32)%s >> 31;",
4639 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3ebea2cf 4640 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5101a5f9 4641 strcpy(g_comment, "cdq");
4642 break;
4643
092f64e1 4644 case OP_LODS:
4645 assert_operand_cnt(3);
4646 if (po->flags & OPF_REP) {
4647 // hmh..
4648 ferr(po, "TODO\n");
4649 }
4650 else {
05381e4a 4651 fprintf(fout, " eax = %sesi; esi %c= %d;",
092f64e1 4652 lmod_cast_u_ptr(po, po->operand[0].lmod),
4653 (po->flags & OPF_DF) ? '-' : '+',
4654 lmod_bytes(po, po->operand[0].lmod));
4655 strcpy(g_comment, "lods");
4656 }
4657 break;
4658
33c35af6 4659 case OP_STOS:
33c35af6 4660 assert_operand_cnt(3);
4661 if (po->flags & OPF_REP) {
591721d7 4662 fprintf(fout, " for (; ecx != 0; ecx--, edi %c= %d)\n",
4663 (po->flags & OPF_DF) ? '-' : '+',
33c35af6 4664 lmod_bytes(po, po->operand[0].lmod));
d4e3b5db 4665 fprintf(fout, " %sedi = eax;",
4666 lmod_cast_u_ptr(po, po->operand[0].lmod));
33c35af6 4667 strcpy(g_comment, "rep stos");
4668 }
4669 else {
092f64e1 4670 fprintf(fout, " %sedi = eax; edi %c= %d;",
d4e3b5db 4671 lmod_cast_u_ptr(po, po->operand[0].lmod),
591721d7 4672 (po->flags & OPF_DF) ? '-' : '+',
33c35af6 4673 lmod_bytes(po, po->operand[0].lmod));
4674 strcpy(g_comment, "stos");
4675 }
4676 break;
4677
d4e3b5db 4678 case OP_MOVS:
d4e3b5db 4679 assert_operand_cnt(3);
4680 j = lmod_bytes(po, po->operand[0].lmod);
4681 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
591721d7 4682 l = (po->flags & OPF_DF) ? '-' : '+';
d4e3b5db 4683 if (po->flags & OPF_REP) {
4684 fprintf(fout,
591721d7 4685 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n",
4686 l, j, l, j);
d4e3b5db 4687 fprintf(fout,
4688 " %sedi = %sesi;", buf1, buf1);
4689 strcpy(g_comment, "rep movs");
4690 }
4691 else {
092f64e1 4692 fprintf(fout, " %sedi = %sesi; edi %c= %d; esi %c= %d;",
591721d7 4693 buf1, buf1, l, j, l, j);
d4e3b5db 4694 strcpy(g_comment, "movs");
4695 }
4696 break;
4697
7ba45c34 4698 case OP_CMPS:
7ba45c34 4699 // repe ~ repeat while ZF=1
4700 assert_operand_cnt(3);
4701 j = lmod_bytes(po, po->operand[0].lmod);
4702 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
591721d7 4703 l = (po->flags & OPF_DF) ? '-' : '+';
7ba45c34 4704 if (po->flags & OPF_REP) {
4705 fprintf(fout,
1f84f6b3 4706 " for (; ecx != 0; ecx--) {\n");
4741fdfe 4707 if (pfomask & (1 << PFO_C)) {
4708 // ugh..
4709 fprintf(fout,
05381e4a 4710 " cond_c = %sesi < %sedi;\n", buf1, buf1);
4741fdfe 4711 pfomask &= ~(1 << PFO_C);
4712 }
7ba45c34 4713 fprintf(fout,
05381e4a 4714 " cond_z = (%sesi == %sedi); esi %c= %d, edi %c= %d;\n",
1f84f6b3 4715 buf1, buf1, l, j, l, j);
4716 fprintf(fout,
4717 " if (cond_z %s 0) break;\n",
4718 (po->flags & OPF_REPZ) ? "==" : "!=");
7ba45c34 4719 fprintf(fout,
4741fdfe 4720 " }");
7ba45c34 4721 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
4722 (po->flags & OPF_REPZ) ? "e" : "ne");
4723 }
4724 else {
4725 fprintf(fout,
05381e4a 4726 " cond_z = (%sesi == %sedi); esi %c= %d; edi %c= %d;",
591721d7 4727 buf1, buf1, l, j, l, j);
7ba45c34 4728 strcpy(g_comment, "cmps");
4729 }
4730 pfomask &= ~(1 << PFO_Z);
4731 last_arith_dst = NULL;
4732 delayed_flag_op = NULL;
4733 break;
4734
591721d7 4735 case OP_SCAS:
4736 // only does ZF (for now)
4737 // repe ~ repeat while ZF=1
4738 assert_operand_cnt(3);
4739 j = lmod_bytes(po, po->operand[0].lmod);
4740 l = (po->flags & OPF_DF) ? '-' : '+';
4741 if (po->flags & OPF_REP) {
4742 fprintf(fout,
1f84f6b3 4743 " for (; ecx != 0; ecx--) {\n");
591721d7 4744 fprintf(fout,
1f84f6b3 4745 " cond_z = (%seax == %sedi); edi %c= %d;\n",
591721d7 4746 lmod_cast_u(po, po->operand[0].lmod),
1f84f6b3 4747 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
4748 fprintf(fout,
4749 " if (cond_z %s 0) break;\n",
591721d7 4750 (po->flags & OPF_REPZ) ? "==" : "!=");
4751 fprintf(fout,
1f84f6b3 4752 " }");
591721d7 4753 snprintf(g_comment, sizeof(g_comment), "rep%s scas",
4754 (po->flags & OPF_REPZ) ? "e" : "ne");
4755 }
4756 else {
05381e4a 4757 fprintf(fout, " cond_z = (%seax == %sedi); edi %c= %d;",
591721d7 4758 lmod_cast_u(po, po->operand[0].lmod),
4759 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
4760 strcpy(g_comment, "scas");
4761 }
4762 pfomask &= ~(1 << PFO_Z);
4763 last_arith_dst = NULL;
4764 delayed_flag_op = NULL;
4765 break;
4766
850c9265 4767 // arithmetic w/flags
850c9265 4768 case OP_AND:
4769 case OP_OR:
5101a5f9 4770 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4771 // fallthrough
850c9265 4772 dualop_arith:
4773 assert_operand_cnt(2);
850c9265 4774 fprintf(fout, " %s %s= %s;",
4775 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4776 op_to_c(po),
3ebea2cf 4777 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
04f8a628 4778 output_std_flags(fout, po, &pfomask, buf1);
4779 last_arith_dst = &po->operand[0];
4780 delayed_flag_op = NULL;
4781 break;
4782
4783 case OP_SHL:
4784 case OP_SHR:
4785 assert_operand_cnt(2);
4786 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4787 if (pfomask & (1 << PFO_C)) {
4788 if (po->operand[1].type == OPT_CONST) {
4789 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4790 j = po->operand[1].val;
4791 j %= l;
4792 if (j != 0) {
4793 if (po->op == OP_SHL)
4794 j = l - j;
4795 else
4796 j -= 1;
cb090db0 4797 fprintf(fout, " cond_c = (%s >> %d) & 1;\n",
4798 buf1, j);
04f8a628 4799 }
4800 else
4801 ferr(po, "zero shift?\n");
4802 }
4803 else
4804 ferr(po, "TODO\n");
4805 pfomask &= ~(1 << PFO_C);
840257f6 4806 }
04f8a628 4807 fprintf(fout, " %s %s= %s;", buf1, op_to_c(po),
4808 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4809 output_std_flags(fout, po, &pfomask, buf1);
850c9265 4810 last_arith_dst = &po->operand[0];
69a3cdfc 4811 delayed_flag_op = NULL;
850c9265 4812 break;
4813
d4e3b5db 4814 case OP_SAR:
4815 assert_operand_cnt(2);
4816 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4817 fprintf(fout, " %s = %s%s >> %s;", buf1,
4818 lmod_cast_s(po, po->operand[0].lmod), buf1,
3ebea2cf 4819 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
04f8a628 4820 output_std_flags(fout, po, &pfomask, buf1);
d4e3b5db 4821 last_arith_dst = &po->operand[0];
4822 delayed_flag_op = NULL;
4823 break;
4824
3b2f4044 4825 case OP_SHRD:
4826 assert_operand_cnt(3);
4827 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4828 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4829 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4830 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4831 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[2]);
4832 fprintf(fout, " %s >>= %s; %s |= %s << (%d - %s);",
4833 buf1, buf3, buf1, buf2, l, buf3);
4834 strcpy(g_comment, "shrd");
4835 output_std_flags(fout, po, &pfomask, buf1);
4836 last_arith_dst = &po->operand[0];
4837 delayed_flag_op = NULL;
4838 break;
4839
d4e3b5db 4840 case OP_ROL:
4841 case OP_ROR:
4842 assert_operand_cnt(2);
4843 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4844 if (po->operand[1].type == OPT_CONST) {
4845 j = po->operand[1].val;
4846 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
4847 fprintf(fout, po->op == OP_ROL ?
4848 " %s = (%s << %d) | (%s >> %d);" :
4849 " %s = (%s >> %d) | (%s << %d);",
4850 buf1, buf1, j, buf1,
4851 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
4852 }
4853 else
4854 ferr(po, "TODO\n");
04f8a628 4855 output_std_flags(fout, po, &pfomask, buf1);
d4e3b5db 4856 last_arith_dst = &po->operand[0];
4857 delayed_flag_op = NULL;
4858 break;
4859
cb090db0 4860 case OP_RCL:
4861 case OP_RCR:
4862 assert_operand_cnt(2);
4863 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4864 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4865 if (po->operand[1].type == OPT_CONST) {
4866 j = po->operand[1].val % l;
4867 if (j == 0)
4868 ferr(po, "zero rotate\n");
4869 fprintf(fout, " tmp = (%s >> %d) & 1;\n",
4870 buf1, (po->op == OP_RCL) ? (l - j) : (j - 1));
4871 if (po->op == OP_RCL) {
4872 fprintf(fout,
4873 " %s = (%s << %d) | (cond_c << %d)",
4874 buf1, buf1, j, j - 1);
4875 if (j != 1)
4876 fprintf(fout, " | (%s >> %d)", buf1, l + 1 - j);
4877 }
4878 else {
4879 fprintf(fout,
4880 " %s = (%s >> %d) | (cond_c << %d)",
4881 buf1, buf1, j, l - j);
4882 if (j != 1)
4883 fprintf(fout, " | (%s << %d)", buf1, l + 1 - j);
4884 }
4885 fprintf(fout, ";\n");
4886 fprintf(fout, " cond_c = tmp;");
4887 }
4888 else
4889 ferr(po, "TODO\n");
4890 strcpy(g_comment, (po->op == OP_RCL) ? "rcl" : "rcr");
4891 output_std_flags(fout, po, &pfomask, buf1);
4892 last_arith_dst = &po->operand[0];
4893 delayed_flag_op = NULL;
4894 break;
4895
5101a5f9 4896 case OP_XOR:
850c9265 4897 assert_operand_cnt(2);
4898 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5101a5f9 4899 if (IS(opr_name(po, 0), opr_name(po, 1))) {
4900 // special case for XOR
2fe80fdb 4901 if (pfomask & (1 << PFO_BE)) { // weird, but it happens..
4902 fprintf(fout, " cond_be = 1;\n");
4903 pfomask &= ~(1 << PFO_BE);
4904 }
5101a5f9 4905 fprintf(fout, " %s = 0;",
4906 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4907 last_arith_dst = &po->operand[0];
4908 delayed_flag_op = NULL;
850c9265 4909 break;
850c9265 4910 }
5101a5f9 4911 goto dualop_arith;
4912
2fe80fdb 4913 case OP_ADD:
4914 assert_operand_cnt(2);
4915 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4916 if (pfomask & (1 << PFO_C)) {
c8dbc5be 4917 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4918 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4919 if (po->operand[0].lmod == OPLM_DWORD) {
4920 fprintf(fout, " tmp64 = (u64)%s + %s;\n", buf1, buf2);
4921 fprintf(fout, " cond_c = tmp64 >> 32;\n");
4922 fprintf(fout, " %s = (u32)tmp64;",
4923 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4924 strcat(g_comment, "add64");
4925 }
4926 else {
4927 fprintf(fout, " cond_c = ((u32)%s + %s) >> %d;\n",
4928 buf1, buf2, lmod_bytes(po, po->operand[0].lmod) * 8);
4929 fprintf(fout, " %s += %s;",
4930 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4931 buf2);
4932 }
2fe80fdb 4933 pfomask &= ~(1 << PFO_C);
4934 output_std_flags(fout, po, &pfomask, buf1);
4935 last_arith_dst = &po->operand[0];
4936 delayed_flag_op = NULL;
4937 break;
4938 }
4939 goto dualop_arith;
4940
4941 case OP_SUB:
4942 assert_operand_cnt(2);
4943 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3b2f4044 4944 if (pfomask & ~((1 << PFO_Z) | (1 << PFO_S))) {
4945 for (j = 0; j <= PFO_LE; j++) {
4946 if (!(pfomask & (1 << j)))
4947 continue;
4948 if (j == PFO_Z || j == PFO_S)
4949 continue;
4950
4951 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
4952 fprintf(fout, " cond_%s = %s;\n",
4953 parsed_flag_op_names[j], buf1);
4954 pfomask &= ~(1 << j);
4955 }
2fe80fdb 4956 }
4957 goto dualop_arith;
4958
5101a5f9 4959 case OP_ADC:
850c9265 4960 case OP_SBB:
5101a5f9 4961 assert_operand_cnt(2);
4962 propagate_lmod(po, &po->operand[0], &po->operand[1]);
a2c1d768 4963 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
840257f6 4964 if (po->op == OP_SBB
4965 && IS(po->operand[0].name, po->operand[1].name))
4966 {
4967 // avoid use of unitialized var
a2c1d768 4968 fprintf(fout, " %s = -cond_c;", buf1);
94d447fb 4969 // carry remains what it was
4970 pfomask &= ~(1 << PFO_C);
840257f6 4971 }
4972 else {
a2c1d768 4973 fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po),
3ebea2cf 4974 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
840257f6 4975 }
a2c1d768 4976 output_std_flags(fout, po, &pfomask, buf1);
5101a5f9 4977 last_arith_dst = &po->operand[0];
4978 delayed_flag_op = NULL;
850c9265 4979 break;
4980
1f84f6b3 4981 case OP_BSF:
4982 assert_operand_cnt(2);
4983 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4984 fprintf(fout, " %s = %s ? __builtin_ffs(%s) - 1 : 0;",
4985 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4986 buf2, buf2);
4987 output_std_flags(fout, po, &pfomask, buf1);
4988 last_arith_dst = &po->operand[0];
4989 delayed_flag_op = NULL;
4990 strcat(g_comment, "bsf");
4991 break;
4992
850c9265 4993 case OP_DEC:
90307a99 4994 if (pfomask & ~(PFOB_S | PFOB_S | PFOB_C)) {
4995 for (j = 0; j <= PFO_LE; j++) {
4996 if (!(pfomask & (1 << j)))
4997 continue;
4998 if (j == PFO_Z || j == PFO_S || j == PFO_C)
4999 continue;
5000
5001 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
5002 fprintf(fout, " cond_%s = %s;\n",
5003 parsed_flag_op_names[j], buf1);
5004 pfomask &= ~(1 << j);
5005 }
5006 }
5007 // fallthrough
5008
5009 case OP_INC:
5010 if (pfomask & (1 << PFO_C))
5011 // carry is unaffected by inc/dec.. wtf?
5012 ferr(po, "carry propagation needed\n");
5013
850c9265 5014 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
1bafb621 5015 if (po->operand[0].type == OPT_REG) {
5016 strcpy(buf2, po->op == OP_INC ? "++" : "--");
5017 fprintf(fout, " %s%s;", buf1, buf2);
5018 }
5019 else {
5020 strcpy(buf2, po->op == OP_INC ? "+" : "-");
5021 fprintf(fout, " %s %s= 1;", buf1, buf2);
5022 }
a2c1d768 5023 output_std_flags(fout, po, &pfomask, buf1);
5101a5f9 5024 last_arith_dst = &po->operand[0];
5025 delayed_flag_op = NULL;
5026 break;
5027
5028 case OP_NEG:
5029 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3ebea2cf 5030 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
5101a5f9 5031 fprintf(fout, " %s = -%s%s;", buf1,
5032 lmod_cast_s(po, po->operand[0].lmod), buf2);
850c9265 5033 last_arith_dst = &po->operand[0];
69a3cdfc 5034 delayed_flag_op = NULL;
940e8e66 5035 if (pfomask & (1 << PFO_C)) {
5036 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
5037 pfomask &= ~(1 << PFO_C);
5038 }
850c9265 5039 break;
5040
5041 case OP_IMUL:
de50b98b 5042 if (po->operand_cnt == 2) {
5043 propagate_lmod(po, &po->operand[0], &po->operand[1]);
850c9265 5044 goto dualop_arith;
de50b98b 5045 }
87bf6cec 5046 if (po->operand_cnt == 3)
5047 ferr(po, "TODO imul3\n");
5048 // fallthrough
5049 case OP_MUL:
5050 assert_operand_cnt(1);
c8dbc5be 5051 switch (po->operand[0].lmod) {
5052 case OPLM_DWORD:
5053 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
5054 fprintf(fout, " tmp64 = %seax * %s%s;\n", buf1, buf1,
5055 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
5056 fprintf(fout, " edx = tmp64 >> 32;\n");
5057 fprintf(fout, " eax = tmp64;");
5058 break;
5059 case OPLM_BYTE:
5060 strcpy(buf1, po->op == OP_IMUL ? "(s16)(s8)" : "(u16)(u8)");
5061 fprintf(fout, " LOWORD(eax) = %seax * %s;", buf1,
5062 out_src_opr(buf2, sizeof(buf2), po, &po->operand[0],
5063 buf1, 0));
5064 break;
5065 default:
5066 ferr(po, "TODO: unhandled mul type\n");
5067 break;
5068 }
87bf6cec 5069 last_arith_dst = NULL;
69a3cdfc 5070 delayed_flag_op = NULL;
91977a1c 5071 break;
5072
5101a5f9 5073 case OP_DIV:
5074 case OP_IDIV:
5075 assert_operand_cnt(1);
5076 if (po->operand[0].lmod != OPLM_DWORD)
5077 ferr(po, "unhandled lmod %d\n", po->operand[0].lmod);
5078
cb090db0 5079 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
5080 strcpy(buf2, lmod_cast(po, po->operand[0].lmod,
5081 po->op == OP_IDIV));
5082 switch (po->operand[0].lmod) {
5083 case OPLM_DWORD:
5084 if (po->flags & OPF_32BIT)
5085 snprintf(buf3, sizeof(buf3), "%seax", buf2);
5086 else {
5087 fprintf(fout, " tmp64 = ((u64)edx << 32) | eax;\n");
5088 snprintf(buf3, sizeof(buf3), "%stmp64",
5089 (po->op == OP_IDIV) ? "(s64)" : "");
5090 }
5091 if (po->operand[0].type == OPT_REG
5092 && po->operand[0].reg == xDX)
5093 {
5094 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
5095 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
5096 }
5097 else {
5098 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
5099 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
5100 }
5101 break;
5102 default:
5103 ferr(po, "unhandled division type\n");
5101a5f9 5104 }
87bf6cec 5105 last_arith_dst = NULL;
5106 delayed_flag_op = NULL;
5101a5f9 5107 break;
5108
91977a1c 5109 case OP_TEST:
5110 case OP_CMP:
850c9265 5111 propagate_lmod(po, &po->operand[0], &po->operand[1]);
940e8e66 5112 if (pfomask != 0) {
69a3cdfc 5113 for (j = 0; j < 8; j++) {
940e8e66 5114 if (pfomask & (1 << j)) {
69a3cdfc 5115 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
5116 fprintf(fout, " cond_%s = %s;",
5117 parsed_flag_op_names[j], buf1);
5118 }
5119 }
940e8e66 5120 pfomask = 0;
69a3cdfc 5121 }
5122 else
5123 no_output = 1;
7ba45c34 5124 last_arith_dst = NULL;
69a3cdfc 5125 delayed_flag_op = po;
91977a1c 5126 break;
5127
092f64e1 5128 case OP_SCC:
5129 // SETcc - should already be handled
5130 break;
5131
69a3cdfc 5132 // note: we reuse OP_Jcc for SETcc, only flags differ
092f64e1 5133 case OP_JCC:
5134 fprintf(fout, "\n goto %s;", po->operand[0].name);
850c9265 5135 break;
5136
5c024ef7 5137 case OP_JECXZ:
5138 fprintf(fout, " if (ecx == 0)\n");
5139 fprintf(fout, " goto %s;", po->operand[0].name);
5140 strcat(g_comment, "jecxz");
5141 break;
5142
850c9265 5143 case OP_JMP:
87bf6cec 5144 assert_operand_cnt(1);
de50b98b 5145 last_arith_dst = NULL;
5146 delayed_flag_op = NULL;
5147
4c45fa73 5148 if (po->operand[0].type == OPT_REGMEM) {
5149 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
5150 buf1, buf2);
5151 if (ret != 2)
5152 ferr(po, "parse failure for jmp '%s'\n",
5153 po->operand[0].name);
5154 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
5155 break;
5156 }
5157 else if (po->operand[0].type != OPT_LABEL)
5158 ferr(po, "unhandled jmp type\n");
87bf6cec 5159
850c9265 5160 fprintf(fout, " goto %s;", po->operand[0].name);
91977a1c 5161 break;
5162
5163 case OP_CALL:
5101a5f9 5164 assert_operand_cnt(1);
092f64e1 5165 pp = po->pp;
89ff3147 5166 my_assert_not(pp, NULL);
91977a1c 5167
092f64e1 5168 strcpy(buf3, " ");
5169 if (po->flags & OPF_CC) {
5170 // we treat conditional branch to another func
5171 // (yes such code exists..) as conditional tailcall
5172 strcat(buf3, " ");
5173 fprintf(fout, " {\n");
5174 }
5175
8eb12e72 5176 if (pp->is_fptr && !pp->is_arg) {
092f64e1 5177 fprintf(fout, "%s%s = %s;\n", buf3, pp->name,
1cd4a663 5178 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
5179 "(void *)", 0));
8eb12e72 5180 if (pp->is_unresolved)
5181 fprintf(fout, "%sunresolved_call(\"%s:%d\", %s);\n",
5182 buf3, asmfn, po->asmln, pp->name);
5183 }
1bafb621 5184
092f64e1 5185 fprintf(fout, "%s", buf3);
2b43685d 5186 if (strstr(pp->ret_type.name, "int64")) {
87bf6cec 5187 if (po->flags & OPF_TAIL)
2b43685d 5188 ferr(po, "int64 and tail?\n");
2fe80fdb 5189 fprintf(fout, "tmp64 = ");
2b43685d 5190 }
5191 else if (!IS(pp->ret_type.name, "void")) {
5192 if (po->flags & OPF_TAIL) {
840257f6 5193 if (!IS(g_func_pp->ret_type.name, "void")) {
5194 fprintf(fout, "return ");
5195 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
5196 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
5197 }
2b43685d 5198 }
75ad0378 5199 else if (regmask & (1 << xAX)) {
87bf6cec 5200 fprintf(fout, "eax = ");
2b43685d 5201 if (pp->ret_type.is_ptr)
5202 fprintf(fout, "(u32)");
5203 }
91977a1c 5204 }
87bf6cec 5205
ddaf8bd7 5206 if (pp->name[0] == 0)
5207 ferr(po, "missing pp->name\n");
5208 fprintf(fout, "%s%s(", pp->name,
5209 pp->has_structarg ? "_sa" : "");
39b168b8 5210
2fe80fdb 5211 if (po->flags & OPF_ATAIL) {
5212 if (pp->argc_stack != g_func_pp->argc_stack
5213 || (pp->argc_stack > 0
5214 && pp->is_stdcall != g_func_pp->is_stdcall))
5215 ferr(po, "incompatible tailcall\n");
1f84f6b3 5216 if (g_func_pp->has_retreg)
5217 ferr(po, "TODO: retreg+tailcall\n");
87bf6cec 5218
2fe80fdb 5219 for (arg = j = 0; arg < pp->argc; arg++) {
5220 if (arg > 0)
5221 fprintf(fout, ", ");
87bf6cec 5222
2fe80fdb 5223 cast[0] = 0;
5224 if (pp->arg[arg].type.is_ptr)
5225 snprintf(cast, sizeof(cast), "(%s)",
5226 pp->arg[arg].type.name);
91977a1c 5227
2fe80fdb 5228 if (pp->arg[arg].reg != NULL) {
5229 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
5230 continue;
5231 }
5232 // stack arg
5233 for (; j < g_func_pp->argc; j++)
5234 if (g_func_pp->arg[j].reg == NULL)
5235 break;
5236 fprintf(fout, "%sa%d", cast, j + 1);
5237 j++;
69a3cdfc 5238 }
2fe80fdb 5239 }
5240 else {
5241 for (arg = 0; arg < pp->argc; arg++) {
5242 if (arg > 0)
5243 fprintf(fout, ", ");
5244
5245 cast[0] = 0;
5246 if (pp->arg[arg].type.is_ptr)
5247 snprintf(cast, sizeof(cast), "(%s)",
5248 pp->arg[arg].type.name);
5249
5250 if (pp->arg[arg].reg != NULL) {
1f84f6b3 5251 if (pp->arg[arg].type.is_retreg)
5252 fprintf(fout, "&%s", pp->arg[arg].reg);
5253 else
5254 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
2fe80fdb 5255 continue;
5256 }
5257
5258 // stack arg
5259 tmp_op = pp->arg[arg].datap;
5260 if (tmp_op == NULL)
5261 ferr(po, "parsed_op missing for arg%d\n", arg);
23fd0b11 5262
5263 if (tmp_op->flags & OPF_VAPUSH) {
5264 fprintf(fout, "ap");
5265 }
5f70a34f 5266 else if (tmp_op->p_argpass != 0) {
5267 fprintf(fout, "a%d", tmp_op->p_argpass);
5268 }
5269 else if (tmp_op->p_argnum != 0) {
3a5101d7 5270 fprintf(fout, "%s%s", cast,
5271 saved_arg_name(buf1, sizeof(buf1),
5272 tmp_op->p_arggrp, tmp_op->p_argnum));
2fe80fdb 5273 }
5274 else {
5275 fprintf(fout, "%s",
5276 out_src_opr(buf1, sizeof(buf1),
5277 tmp_op, &tmp_op->operand[0], cast, 0));
5278 }
69a3cdfc 5279 }
91977a1c 5280 }
5281 fprintf(fout, ");");
87bf6cec 5282
2b43685d 5283 if (strstr(pp->ret_type.name, "int64")) {
5284 fprintf(fout, "\n");
092f64e1 5285 fprintf(fout, "%sedx = tmp64 >> 32;\n", buf3);
5286 fprintf(fout, "%seax = tmp64;", buf3);
2b43685d 5287 }
5288
89ff3147 5289 if (pp->is_unresolved) {
8eb12e72 5290 snprintf(buf2, sizeof(buf2), " unresolved %dreg",
89ff3147 5291 pp->argc_reg);
092f64e1 5292 strcat(g_comment, buf2);
89ff3147 5293 }
5294
87bf6cec 5295 if (po->flags & OPF_TAIL) {
840257f6 5296 ret = 0;
ddaf8bd7 5297 if (i == opcnt - 1 || pp->is_noreturn)
840257f6 5298 ret = 0;
5299 else if (IS(pp->ret_type.name, "void"))
5300 ret = 1;
5301 else if (IS(g_func_pp->ret_type.name, "void"))
5302 ret = 1;
5303 // else already handled as 'return f()'
5304
5305 if (ret) {
da87ae38 5306 if (!IS(g_func_pp->ret_type.name, "void")) {
ddaf8bd7 5307 ferr(po, "int func -> void func tailcall?\n");
da87ae38 5308 }
5309 else {
092f64e1 5310 fprintf(fout, "\n%sreturn;", buf3);
ddaf8bd7 5311 strcat(g_comment, " ^ tailcall");
da87ae38 5312 }
3ebea2cf 5313 }
89ff3147 5314 else
ddaf8bd7 5315 strcat(g_comment, " tailcall");
87bf6cec 5316 }
ddaf8bd7 5317 if (pp->is_noreturn)
5318 strcat(g_comment, " noreturn");
2fe80fdb 5319 if ((po->flags & OPF_ATAIL) && pp->argc_stack > 0)
5320 strcat(g_comment, " argframe");
092f64e1 5321 if (po->flags & OPF_CC)
5322 strcat(g_comment, " cond");
5323
5324 if (po->flags & OPF_CC)
5325 fprintf(fout, "\n }");
5326
87bf6cec 5327 delayed_flag_op = NULL;
5328 last_arith_dst = NULL;
91977a1c 5329 break;
5330
5331 case OP_RET:
bd96f656 5332 if (g_func_pp->is_vararg)
4f12f671 5333 fprintf(fout, " va_end(ap);\n");
1f84f6b3 5334 if (g_func_pp->has_retreg) {
5335 for (arg = 0; arg < g_func_pp->argc; arg++)
5336 if (g_func_pp->arg[arg].type.is_retreg)
5337 fprintf(fout, " *r_%s = %s;\n",
5338 g_func_pp->arg[arg].reg, g_func_pp->arg[arg].reg);
5339 }
4f12f671 5340
bd96f656 5341 if (IS(g_func_pp->ret_type.name, "void")) {
3ebea2cf 5342 if (i != opcnt - 1 || label_pending)
5343 fprintf(fout, " return;");
5344 }
bd96f656 5345 else if (g_func_pp->ret_type.is_ptr) {
d4e3b5db 5346 fprintf(fout, " return (%s)eax;",
bd96f656 5347 g_func_pp->ret_type.name);
3ebea2cf 5348 }
2fe80fdb 5349 else if (IS(g_func_pp->ret_type.name, "__int64"))
5350 fprintf(fout, " return ((u64)edx << 32) | eax;");
91977a1c 5351 else
5352 fprintf(fout, " return eax;");
de50b98b 5353
5354 last_arith_dst = NULL;
5355 delayed_flag_op = NULL;
91977a1c 5356 break;
5357
5358 case OP_PUSH:
1f84f6b3 5359 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
5f70a34f 5360 if (po->p_argnum != 0) {
69a3cdfc 5361 // special case - saved func arg
3a5101d7 5362 fprintf(fout, " %s = %s;",
5363 saved_arg_name(buf2, sizeof(buf2),
5364 po->p_arggrp, po->p_argnum), buf1);
69a3cdfc 5365 break;
5366 }
d4e3b5db 5367 else if (po->flags & OPF_RSAVE) {
d4e3b5db 5368 fprintf(fout, " s_%s = %s;", buf1, buf1);
5369 break;
5370 }
1f84f6b3 5371 else if (g_func_pp->is_userstack) {
5372 fprintf(fout, " *(--esp) = %s;", buf1);
5373 break;
5374 }
e56ab892 5375 if (!(g_ida_func_attr & IDAFA_NORETURN))
5376 ferr(po, "stray push encountered\n");
5377 no_output = 1;
91977a1c 5378 break;
5379
5380 case OP_POP:
d4e3b5db 5381 if (po->flags & OPF_RSAVE) {
3ebea2cf 5382 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
d4e3b5db 5383 fprintf(fout, " %s = s_%s;", buf1, buf1);
5384 break;
5385 }
5c024ef7 5386 else if (po->datap != NULL) {
5387 // push/pop pair
5388 tmp_op = po->datap;
5389 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5390 fprintf(fout, " %s = %s;", buf1,
5391 out_src_opr(buf2, sizeof(buf2),
5392 tmp_op, &tmp_op->operand[0],
c7ed83dd 5393 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
5c024ef7 5394 break;
5395 }
1f84f6b3 5396 else if (g_func_pp->is_userstack) {
5397 fprintf(fout, " %s = *esp++;",
5398 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
5399 break;
5400 }
5401 else
5402 ferr(po, "stray pop encountered\n");
91977a1c 5403 break;
5404
33c35af6 5405 case OP_NOP:
2b43685d 5406 no_output = 1;
33c35af6 5407 break;
5408
90307a99 5409 // mmx
5410 case OP_EMMS:
5411 strcpy(g_comment, "(emms)");
5412 break;
5413
91977a1c 5414 default:
5415 no_output = 1;
69a3cdfc 5416 ferr(po, "unhandled op type %d, flags %x\n",
5417 po->op, po->flags);
91977a1c 5418 break;
5419 }
5420
5421 if (g_comment[0] != 0) {
ddaf8bd7 5422 char *p = g_comment;
5423 while (my_isblank(*p))
5424 p++;
5425 fprintf(fout, " // %s", p);
91977a1c 5426 g_comment[0] = 0;
5427 no_output = 0;
5428 }
5429 if (!no_output)
5430 fprintf(fout, "\n");
5101a5f9 5431
2b43685d 5432 // some sanity checking
591721d7 5433 if (po->flags & OPF_REP) {
5434 if (po->op != OP_STOS && po->op != OP_MOVS
5435 && po->op != OP_CMPS && po->op != OP_SCAS)
5436 ferr(po, "unexpected rep\n");
5437 if (!(po->flags & (OPF_REPZ|OPF_REPNZ))
5438 && (po->op == OP_CMPS || po->op == OP_SCAS))
5439 ferr(po, "cmps/scas with plain rep\n");
5440 }
5441 if ((po->flags & (OPF_REPZ|OPF_REPNZ))
5442 && po->op != OP_CMPS && po->op != OP_SCAS)
2b43685d 5443 ferr(po, "unexpected repz/repnz\n");
5444
940e8e66 5445 if (pfomask != 0)
7ba45c34 5446 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
940e8e66 5447
5101a5f9 5448 // see is delayed flag stuff is still valid
5449 if (delayed_flag_op != NULL && delayed_flag_op != po) {
89ff3147 5450 if (is_any_opr_modified(delayed_flag_op, po, 0))
5101a5f9 5451 delayed_flag_op = NULL;
5452 }
5453
5454 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
5455 if (is_opr_modified(last_arith_dst, po))
5456 last_arith_dst = NULL;
5457 }
3ebea2cf 5458
5459 label_pending = 0;
91977a1c 5460 }
5461
a2c1d768 5462 if (g_stack_fsz && !g_stack_frame_used)
5463 fprintf(fout, " (void)sf;\n");
5464
91977a1c 5465 fprintf(fout, "}\n\n");
5466
9af2d373 5467 gen_x_cleanup(opcnt);
5468}
5469
5470static void gen_x_cleanup(int opcnt)
5471{
5472 int i;
5473
91977a1c 5474 for (i = 0; i < opcnt; i++) {
4c45fa73 5475 struct label_ref *lr, *lr_del;
5476
5477 lr = g_label_refs[i].next;
5478 while (lr != NULL) {
5479 lr_del = lr;
5480 lr = lr->next;
5481 free(lr_del);
5482 }
5483 g_label_refs[i].i = -1;
5484 g_label_refs[i].next = NULL;
5485
91977a1c 5486 if (ops[i].op == OP_CALL) {
092f64e1 5487 if (ops[i].pp)
5488 proto_release(ops[i].pp);
91977a1c 5489 }
5490 }
bd96f656 5491 g_func_pp = NULL;
91977a1c 5492}
c36e914d 5493
92d715b6 5494struct func_proto_dep;
5495
5496struct func_prototype {
5497 char name[NAMELEN];
5498 int id;
5499 int argc_stack;
5500 int regmask_dep;
91ca764a 5501 int has_ret:3; // -1, 0, 1: unresolved, no, yes
92d715b6 5502 unsigned int dep_resolved:1;
5503 unsigned int is_stdcall:1;
5504 struct func_proto_dep *dep_func;
5505 int dep_func_cnt;
91ca764a 5506 const struct parsed_proto *pp; // seed pp, if any
92d715b6 5507};
5508
5509struct func_proto_dep {
5510 char *name;
5511 struct func_prototype *proto;
5512 int regmask_live; // .. at the time of call
5513 unsigned int ret_dep:1; // return from this is caller's return
5514};
5515
5516static struct func_prototype *hg_fp;
5517static int hg_fp_cnt;
5518
5fa1256f 5519static struct scanned_var {
5520 char name[NAMELEN];
5521 enum opr_lenmod lmod;
5522 unsigned int is_seeded:1;
61e29183 5523 unsigned int is_c_str:1;
c0de9015 5524 const struct parsed_proto *pp; // seed pp, if any
5fa1256f 5525} *hg_vars;
5526static int hg_var_cnt;
5527
9af2d373 5528static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
5529 int count);
5530
92d715b6 5531static struct func_proto_dep *hg_fp_find_dep(struct func_prototype *fp,
5532 const char *name)
5533{
5534 int i;
5535
5536 for (i = 0; i < fp->dep_func_cnt; i++)
5537 if (IS(fp->dep_func[i].name, name))
5538 return &fp->dep_func[i];
5539
5540 return NULL;
5541}
5542
5543static void hg_fp_add_dep(struct func_prototype *fp, const char *name)
5544{
5545 // is it a dupe?
5546 if (hg_fp_find_dep(fp, name))
5547 return;
5548
5549 if ((fp->dep_func_cnt & 0xff) == 0) {
5550 fp->dep_func = realloc(fp->dep_func,
5551 sizeof(fp->dep_func[0]) * (fp->dep_func_cnt + 0x100));
5552 my_assert_not(fp->dep_func, NULL);
5553 memset(&fp->dep_func[fp->dep_func_cnt], 0,
5554 sizeof(fp->dep_func[0]) * 0x100);
5555 }
5556 fp->dep_func[fp->dep_func_cnt].name = strdup(name);
5557 fp->dep_func_cnt++;
5558}
5559
5560static int hg_fp_cmp_name(const void *p1_, const void *p2_)
5561{
5562 const struct func_prototype *p1 = p1_, *p2 = p2_;
5563 return strcmp(p1->name, p2->name);
5564}
5565
5566#if 0
5567static int hg_fp_cmp_id(const void *p1_, const void *p2_)
5568{
5569 const struct func_prototype *p1 = p1_, *p2 = p2_;
5570 return p1->id - p2->id;
5571}
5572#endif
5573
91ca764a 5574// recursive register dep pass
5575// - track saved regs (part 2)
5576// - try to figure out arg-regs
5577// - calculate reg deps
5578static void gen_hdr_dep_pass(int i, int opcnt, unsigned char *cbits,
5579 struct func_prototype *fp, int regmask_save, int regmask_dst,
5580 int *regmask_dep, int *has_ret)
5581{
5582 struct func_proto_dep *dep;
5583 struct parsed_op *po;
5584 int from_caller = 0;
5585 int depth;
5586 int j, l;
5587 int reg;
5588 int ret;
5589
5590 for (; i < opcnt; i++)
5591 {
5592 if (cbits[i >> 3] & (1 << (i & 7)))
5593 return;
5594 cbits[i >> 3] |= (1 << (i & 7));
5595
5596 po = &ops[i];
5597
5598 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
5599 if (po->btj != NULL) {
5600 // jumptable
5601 for (j = 0; j < po->btj->count; j++) {
5602 gen_hdr_dep_pass(po->btj->d[j].bt_i, opcnt, cbits, fp,
5603 regmask_save, regmask_dst, regmask_dep, has_ret);
5604 }
5605 return;
5606 }
5607
5608 if (po->bt_i < 0) {
5609 ferr(po, "dead branch\n");
5610 return;
5611 }
5612
5613 if (po->flags & OPF_CJMP) {
5614 gen_hdr_dep_pass(po->bt_i, opcnt, cbits, fp,
5615 regmask_save, regmask_dst, regmask_dep, has_ret);
5616 }
5617 else {
5618 i = po->bt_i - 1;
5619 }
5620 continue;
5621 }
5622
5623 if (po->flags & OPF_FARG)
5624 /* (just calculate register deps) */;
5625 else if (po->op == OP_PUSH && po->operand[0].type == OPT_REG)
5626 {
5627 reg = po->operand[0].reg;
5628 if (reg < 0)
5629 ferr(po, "reg not set for push?\n");
5630
5631 if (po->flags & OPF_RSAVE) {
5632 regmask_save |= 1 << reg;
5633 continue;
5634 }
5635 if (po->flags & OPF_DONE)
5636 continue;
5637
5638 depth = 0;
5639 ret = scan_for_pop(i + 1, opcnt,
5640 po->operand[0].name, i + opcnt * 2, 0, &depth, 0);
5641 if (ret == 1) {
5642 regmask_save |= 1 << reg;
5643 po->flags |= OPF_RMD;
5644 scan_for_pop(i + 1, opcnt,
5645 po->operand[0].name, i + opcnt * 3, 0, &depth, 1);
5646 continue;
5647 }
5648 }
5649 else if (po->flags & OPF_RMD)
5650 continue;
5651 else if (po->op == OP_CALL) {
5652 po->regmask_dst |= 1 << xAX;
5653
5654 dep = hg_fp_find_dep(fp, po->operand[0].name);
5655 if (dep != NULL)
5656 dep->regmask_live = regmask_save | regmask_dst;
5657 }
5658 else if (po->op == OP_RET) {
5659 if (po->operand_cnt > 0) {
5660 fp->is_stdcall = 1;
5661 if (fp->argc_stack >= 0
5662 && fp->argc_stack != po->operand[0].val / 4)
5663 ferr(po, "ret mismatch? (%d)\n", fp->argc_stack * 4);
5664 fp->argc_stack = po->operand[0].val / 4;
5665 }
5666 }
5667
5668 if (*has_ret != 0 && (po->flags & OPF_TAIL)) {
5669 if (po->op == OP_CALL) {
5670 j = i;
5671 ret = 1;
5672 }
5673 else {
5674 struct parsed_opr opr = { 0, };
5675 opr.type = OPT_REG;
5676 opr.reg = xAX;
5677 j = -1;
5678 from_caller = 0;
5679 ret = resolve_origin(i, &opr, i + opcnt * 4, &j, &from_caller);
5680 }
5681
5682 if (ret == -1 && from_caller) {
5683 // unresolved eax - probably void func
5684 *has_ret = 0;
5685 }
5686 else {
5687 if (ops[j].op == OP_CALL) {
5688 dep = hg_fp_find_dep(fp, po->operand[0].name);
5689 if (dep != NULL)
5690 dep->ret_dep = 1;
5691 else
5692 *has_ret = 1;
5693 }
5694 else
5695 *has_ret = 1;
5696 }
5697 }
5698
5699 l = regmask_save | regmask_dst;
5700 if (g_bp_frame && !(po->flags & OPF_EBP_S))
5701 l |= 1 << xBP;
5702
5703 l = po->regmask_src & ~l;
5704#if 0
5705 if (l)
5706 fnote(po, "dep |= %04x, dst %04x, save %04x (f %x)\n",
5707 l, regmask_dst, regmask_save, po->flags);
5708#endif
5709 *regmask_dep |= l;
5710 regmask_dst |= po->regmask_dst;
5711
5712 if (po->flags & OPF_TAIL)
5713 return;
5714 }
5715}
5716
92d715b6 5717static void gen_hdr(const char *funcn, int opcnt)
5718{
26677139 5719 int save_arg_vars[MAX_ARG_GRP] = { 0, };
91ca764a 5720 unsigned char cbits[MAX_OPS / 8];
9af2d373 5721 const struct parsed_proto *pp_c;
5722 struct parsed_proto *pp;
92d715b6 5723 struct func_prototype *fp;
92d715b6 5724 struct parsed_data *pd;
5725 struct parsed_op *po;
9af2d373 5726 const char *tmpname;
26677139 5727 int regmask_dummy = 0;
91ca764a 5728 int regmask_dep;
92d715b6 5729 int max_bp_offset = 0;
91ca764a 5730 int has_ret;
92d715b6 5731 int i, j, l, ret;
92d715b6 5732
5733 if ((hg_fp_cnt & 0xff) == 0) {
5734 hg_fp = realloc(hg_fp, sizeof(hg_fp[0]) * (hg_fp_cnt + 0x100));
5735 my_assert_not(hg_fp, NULL);
5736 memset(hg_fp + hg_fp_cnt, 0, sizeof(hg_fp[0]) * 0x100);
5737 }
5738
5739 fp = &hg_fp[hg_fp_cnt];
5740 snprintf(fp->name, sizeof(fp->name), "%s", funcn);
5741 fp->id = hg_fp_cnt;
9af2d373 5742 fp->argc_stack = -1;
92d715b6 5743 hg_fp_cnt++;
5744
9af2d373 5745 // perhaps already in seed header?
91ca764a 5746 fp->pp = proto_parse(g_fhdr, funcn, 1);
5747 if (fp->pp != NULL) {
5748 fp->argc_stack = fp->pp->argc_stack;
5749 fp->is_stdcall = fp->pp->is_stdcall;
5750 fp->regmask_dep = get_pp_arg_regmask(fp->pp);
5751 fp->has_ret = !IS(fp->pp->ret_type.name, "void");
9af2d373 5752 return;
5753 }
5754
5755 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
5756 g_stack_frame_used = 0;
5757
92d715b6 5758 // pass1:
9af2d373 5759 // - handle ebp/esp frame, remove ops related to it
5760 scan_prologue_epilogue(opcnt);
5761
5762 // pass2:
92d715b6 5763 // - collect calls
5764 // - resolve all branches
5765 for (i = 0; i < opcnt; i++)
5766 {
5767 po = &ops[i];
5768 po->bt_i = -1;
5769 po->btj = NULL;
5770
5e49b270 5771 if (po->flags & (OPF_RMD|OPF_DONE))
92d715b6 5772 continue;
5773
5774 if (po->op == OP_CALL) {
9af2d373 5775 tmpname = opr_name(po, 0);
5776 pp = NULL;
5777 if (po->operand[0].type == OPT_LABEL) {
5778 hg_fp_add_dep(fp, tmpname);
5779
5780 // perhaps a call to already known func?
5781 pp_c = proto_parse(g_fhdr, tmpname, 1);
5782 if (pp_c != NULL)
5783 pp = proto_clone(pp_c);
5784 }
5785 else if (po->datap != NULL) {
5786 pp = calloc(1, sizeof(*pp));
5787 my_assert_not(pp, NULL);
92d715b6 5788
9af2d373 5789 ret = parse_protostr(po->datap, pp);
5790 if (ret < 0)
5791 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
5792 free(po->datap);
5793 po->datap = NULL;
5794 }
5795 if (pp != NULL && pp->is_noreturn)
5796 po->flags |= OPF_TAIL;
5797
5798 po->pp = pp;
92d715b6 5799 continue;
5800 }
5801
5802 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
5803 continue;
5804
5805 if (po->operand[0].type == OPT_REGMEM) {
5806 pd = try_resolve_jumptab(i, opcnt);
5807 if (pd == NULL)
5808 goto tailcall;
5809
5810 po->btj = pd;
5811 continue;
5812 }
5813
5814 for (l = 0; l < opcnt; l++) {
d7857c3a 5815 if (g_labels[l] != NULL
5816 && IS(po->operand[0].name, g_labels[l]))
5817 {
92d715b6 5818 add_label_ref(&g_label_refs[l], i);
5819 po->bt_i = l;
5820 break;
5821 }
5822 }
5823
5824 if (po->bt_i != -1 || (po->flags & OPF_RMD))
5825 continue;
5826
5827 if (po->operand[0].type == OPT_LABEL)
5828 // assume tail call
5829 goto tailcall;
5830
5831 ferr(po, "unhandled branch\n");
5832
5833tailcall:
5834 po->op = OP_CALL;
5835 po->flags |= OPF_TAIL;
5836 if (i > 0 && ops[i - 1].op == OP_POP)
5837 po->flags |= OPF_ATAIL;
5838 i--; // reprocess
5839 }
5840
9af2d373 5841 // pass3:
92d715b6 5842 // - remove dead labels
9af2d373 5843 // - handle push <const>/pop pairs
92d715b6 5844 for (i = 0; i < opcnt; i++)
5845 {
d7857c3a 5846 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
5847 free(g_labels[i]);
5848 g_labels[i] = NULL;
5849 }
9af2d373 5850
91ca764a 5851 po = &ops[i];
5852 if (po->flags & (OPF_RMD|OPF_DONE))
5853 continue;
5854
5855 if (po->op == OP_PUSH && po->operand[0].type == OPT_CONST)
5856 scan_for_pop_const(i, opcnt);
5857 }
5858
5859 // pass4:
5860 // - process trivial calls
5861 for (i = 0; i < opcnt; i++)
5862 {
9af2d373 5863 po = &ops[i];
5e49b270 5864 if (po->flags & (OPF_RMD|OPF_DONE))
9af2d373 5865 continue;
5866
26677139 5867 if (po->op == OP_CALL)
5868 {
5869 pp = process_call_early(i, opcnt, &j);
5870 if (pp != NULL) {
5871 if (!(po->flags & OPF_ATAIL))
5872 // since we know the args, try to collect them
5873 if (collect_call_args_early(po, i, pp, &regmask_dummy) != 0)
5874 pp = NULL;
5875 }
5876
5877 if (pp != NULL) {
5878 if (j >= 0) {
5879 // commit esp adjust
5880 ops[j].flags |= OPF_RMD;
5e49b270 5881 if (ops[j].op != OP_POP)
5882 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
5883 else
5884 ops[j].flags |= OPF_DONE;
26677139 5885 }
5886
5887 po->flags |= OPF_DONE;
5888 }
5889 }
26677139 5890 }
5891
91ca764a 5892 // pass5:
5e49b270 5893 // - track saved regs (simple)
26677139 5894 // - process calls
5895 for (i = 0; i < opcnt; i++)
5896 {
5897 po = &ops[i];
5e49b270 5898 if (po->flags & (OPF_RMD|OPF_DONE))
26677139 5899 continue;
5900
5e49b270 5901 if (po->op == OP_PUSH && po->operand[0].type == OPT_REG)
5902 {
5903 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
5904 if (ret == 0) {
91ca764a 5905 // regmask_save |= 1 << po->operand[0].reg; // do it later
5906 po->flags |= OPF_RSAVE | OPF_RMD | OPF_DONE;
5e49b270 5907 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, OPF_RMD);
5908 }
5909 }
5910 else if (po->op == OP_CALL && !(po->flags & OPF_DONE))
26677139 5911 {
9af2d373 5912 pp = process_call(i, opcnt);
5913
5914 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
9af2d373 5915 // since we know the args, collect them
91ca764a 5916 ret = collect_call_args(po, i, pp, &regmask_dummy, save_arg_vars,
5e49b270 5917 i + opcnt * 1);
9af2d373 5918 }
5919 }
92d715b6 5920 }
5921
91ca764a 5922 // pass6
5923 memset(cbits, 0, sizeof(cbits));
5924 regmask_dep = 0;
5925 has_ret = -1;
5926
5927 gen_hdr_dep_pass(0, opcnt, cbits, fp, 0, 0, &regmask_dep, &has_ret);
5928
5929 // find unreachable code - must be fixed in IDA
92d715b6 5930 for (i = 0; i < opcnt; i++)
5931 {
91ca764a 5932 if (cbits[i >> 3] & (1 << (i & 7)))
9af2d373 5933 continue;
92d715b6 5934
91ca764a 5935 if (ops[i].op != OP_NOP)
5936 ferr(&ops[i], "unreachable code\n");
92d715b6 5937 }
5938
5939 if (has_ret == -1 && (regmask_dep & (1 << xAX)))
5940 has_ret = 1;
5941
9af2d373 5942 for (i = 0; i < g_eqcnt; i++) {
92d715b6 5943 if (g_eqs[i].offset > max_bp_offset && g_eqs[i].offset < 4*32)
5944 max_bp_offset = g_eqs[i].offset;
9af2d373 5945 }
92d715b6 5946
9af2d373 5947 if (fp->argc_stack < 0) {
92d715b6 5948 max_bp_offset = (max_bp_offset + 3) & ~3;
9af2d373 5949 fp->argc_stack = max_bp_offset / 4;
5950 if ((g_ida_func_attr & IDAFA_BP_FRAME) && fp->argc_stack > 0)
92d715b6 5951 fp->argc_stack--;
5952 }
5953
5954 fp->regmask_dep = regmask_dep & ~(1 << xSP);
5955 fp->has_ret = has_ret;
91ca764a 5956#if 0
5957 printf("// has_ret %d, regmask_dep %x\n",
5958 fp->has_ret, fp->regmask_dep);
5959 output_hdr_fp(stdout, fp, 1);
5960 if (IS(funcn, "sub_100073FD")) exit(1);
5961#endif
9af2d373 5962
5963 gen_x_cleanup(opcnt);
92d715b6 5964}
5965
5966static void hg_fp_resolve_deps(struct func_prototype *fp)
5967{
5968 struct func_prototype fp_s;
91ca764a 5969 int dep;
92d715b6 5970 int i;
5971
5972 // this thing is recursive, so mark first..
5973 fp->dep_resolved = 1;
5974
5975 for (i = 0; i < fp->dep_func_cnt; i++) {
5976 strcpy(fp_s.name, fp->dep_func[i].name);
5977 fp->dep_func[i].proto = bsearch(&fp_s, hg_fp, hg_fp_cnt,
5978 sizeof(hg_fp[0]), hg_fp_cmp_name);
5979 if (fp->dep_func[i].proto != NULL) {
5980 if (!fp->dep_func[i].proto->dep_resolved)
5981 hg_fp_resolve_deps(fp->dep_func[i].proto);
5982
91ca764a 5983 dep = ~fp->dep_func[i].regmask_live
5984 & fp->dep_func[i].proto->regmask_dep;
5985 fp->regmask_dep |= dep;
5986 // printf("dep %s %s |= %x\n", fp->name,
5987 // fp->dep_func[i].name, dep);
92d715b6 5988
5989 if (fp->has_ret == -1)
5990 fp->has_ret = fp->dep_func[i].proto->has_ret;
5991 }
5992 }
5993}
5994
9af2d373 5995static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
5996 int count)
92d715b6 5997{
61e29183 5998 const struct parsed_proto *pp;
5999 char *p, namebuf[NAMELEN];
6000 const char *name;
92d715b6 6001 int regmask_dep;
6002 int argc_stack;
9af2d373 6003 int j, arg;
92d715b6 6004
9af2d373 6005 for (; count > 0; count--, fp++) {
92d715b6 6006 if (fp->has_ret == -1)
6007 fprintf(fout, "// ret unresolved\n");
6008#if 0
6009 fprintf(fout, "// dep:");
6010 for (j = 0; j < fp->dep_func_cnt; j++) {
6011 fprintf(fout, " %s/", fp->dep_func[j].name);
6012 if (fp->dep_func[j].proto != NULL)
6013 fprintf(fout, "%04x/%d", fp->dep_func[j].proto->regmask_dep,
6014 fp->dep_func[j].proto->has_ret);
6015 }
6016 fprintf(fout, "\n");
6017#endif
6018
61e29183 6019 p = strchr(fp->name, '@');
6020 if (p != NULL) {
6021 memcpy(namebuf, fp->name, p - fp->name);
6022 namebuf[p - fp->name] = 0;
6023 name = namebuf;
6024 }
6025 else
6026 name = fp->name;
6027 if (name[0] == '_')
6028 name++;
6029
6030 pp = proto_parse(g_fhdr, name, 1);
6031 if (pp != NULL && pp->is_include)
6032 continue;
6033
c0de9015 6034 if (fp->pp != NULL) {
6035 // prefer fp for common style,
6036 // only use output_pp if args are complex
6037 for (j = 0; j < fp->pp->argc; j++) {
6038 if (fp->pp->arg[j].fptr != NULL)
6039 break;
6040 }
6041 if (j != fp->pp->argc) {
6042 output_pp(fout, fp->pp, OPP_ALIGN);
6043 fprintf(fout, ";\n");
6044 continue;
6045 }
6046 }
6047
92d715b6 6048 regmask_dep = fp->regmask_dep;
6049 argc_stack = fp->argc_stack;
6050
91ca764a 6051 fprintf(fout, "%-5s", fp->pp ? fp->pp->ret_type.name :
6052 (fp->has_ret ? "int" : "void"));
92d715b6 6053 if (regmask_dep && (fp->is_stdcall || argc_stack == 0)
6054 && (regmask_dep & ~((1 << xCX) | (1 << xDX))) == 0)
6055 {
9af2d373 6056 fprintf(fout, " __fastcall ");
92d715b6 6057 if (!(regmask_dep & (1 << xDX)) && argc_stack == 0)
6058 argc_stack = 1;
6059 else
6060 argc_stack += 2;
6061 regmask_dep = 0;
6062 }
6063 else if (regmask_dep && !fp->is_stdcall) {
6064 fprintf(fout, "/*__usercall*/ ");
92d715b6 6065 }
6066 else if (regmask_dep) {
6067 fprintf(fout, "/*__userpurge*/ ");
92d715b6 6068 }
6069 else if (fp->is_stdcall)
9af2d373 6070 fprintf(fout, " __stdcall ");
92d715b6 6071 else
9af2d373 6072 fprintf(fout, " __cdecl ");
92d715b6 6073
61e29183 6074 fprintf(fout, "%s(", name);
92d715b6 6075
6076 arg = 0;
6077 for (j = 0; j < xSP; j++) {
6078 if (regmask_dep & (1 << j)) {
6079 arg++;
6080 if (arg != 1)
6081 fprintf(fout, ", ");
91ca764a 6082 if (fp->pp != NULL)
6083 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
6084 else
6085 fprintf(fout, "int");
6086 fprintf(fout, " a%d/*<%s>*/", arg, regs_r32[j]);
92d715b6 6087 }
6088 }
6089
6090 for (j = 0; j < argc_stack; j++) {
6091 arg++;
6092 if (arg != 1)
6093 fprintf(fout, ", ");
91ca764a 6094 if (fp->pp != NULL) {
6095 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
6096 if (!fp->pp->arg[arg - 1].type.is_ptr)
6097 fprintf(fout, " ");
6098 }
6099 else
6100 fprintf(fout, "int ");
6101 fprintf(fout, "a%d", arg);
92d715b6 6102 }
6103
6104 fprintf(fout, ");\n");
6105 }
6106}
6107
9af2d373 6108static void output_hdr(FILE *fout)
6109{
5fa1256f 6110 static const char *lmod_c_names[] = {
6111 [OPLM_UNSPEC] = "???",
6112 [OPLM_BYTE] = "uint8_t",
6113 [OPLM_WORD] = "uint16_t",
6114 [OPLM_DWORD] = "uint32_t",
6115 [OPLM_QWORD] = "uint64_t",
6116 };
6117 const struct scanned_var *var;
c0de9015 6118 char line[256] = { 0, };
9af2d373 6119 int i;
6120
6121 // resolve deps
6122 qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_name);
6123 for (i = 0; i < hg_fp_cnt; i++)
6124 hg_fp_resolve_deps(&hg_fp[i]);
6125
6126 // note: messes up .proto ptr, don't use
6127 //qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_id);
6128
5fa1256f 6129 // output variables
6130 for (i = 0; i < hg_var_cnt; i++) {
6131 var = &hg_vars[i];
6132
c0de9015 6133 if (var->pp != NULL && var->pp->is_fptr) {
6134 fprintf(fout, "extern ");
6135 output_pp(fout, var->pp, 0);
6136 fprintf(fout, ";");
6137 }
6138 else if (var->is_c_str)
61e29183 6139 fprintf(fout, "extern %-8s %s[];", "char", var->name);
6140 else
6141 fprintf(fout, "extern %-8s %s;",
6142 lmod_c_names[var->lmod], var->name);
5fa1256f 6143
6144 if (var->is_seeded)
6145 fprintf(fout, " // seeded");
6146 fprintf(fout, "\n");
6147 }
6148
6149 fprintf(fout, "\n");
6150
6151 // output function prototypes
9af2d373 6152 output_hdr_fp(fout, hg_fp, hg_fp_cnt);
c0de9015 6153
6154 // include passthrough
6155 fprintf(fout, "\n// for translate\n");
6156
6157 rewind(g_fhdr);
6158 while (fgets(line, sizeof(line), g_fhdr)) {
6159 if (IS_START(line, "//#"))
6160 fwrite(line, 1, strlen(line), fout);
6161 }
9af2d373 6162}
6163
5fa1256f 6164// read a line, truncating it if it doesn't fit
6165static char *my_fgets(char *s, size_t size, FILE *stream)
d4e3b5db 6166{
5fa1256f 6167 char *ret, *ret2;
6168 char buf[64];
6169 int p;
d4e3b5db 6170
5fa1256f 6171 p = size - 2;
6172 if (p >= 0)
6173 s[p] = 0;
d4e3b5db 6174
5fa1256f 6175 ret = fgets(s, size, stream);
6176 if (ret != NULL && p >= 0 && s[p] != 0 && s[p] != '\n') {
6177 p = sizeof(buf) - 2;
6178 do {
6179 buf[p] = 0;
6180 ret2 = fgets(buf, sizeof(buf), stream);
6181 }
6182 while (ret2 != NULL && buf[p] != 0 && buf[p] != '\n');
6183 }
6184
6185 return ret;
d4e3b5db 6186}
6187
61e29183 6188// '=' needs special treatment
6189// also ' quote
bfa4a6ee 6190static char *next_word_s(char *w, size_t wsize, char *s)
6191{
61e29183 6192 size_t i;
bfa4a6ee 6193
61e29183 6194 s = sskip(s);
bfa4a6ee 6195
61e29183 6196 i = 0;
6197 if (*s == '\'') {
6198 w[0] = s[0];
6199 for (i = 1; i < wsize - 1; i++) {
6200 if (s[i] == 0) {
6201 printf("warning: missing closing quote: \"%s\"\n", s);
6202 break;
6203 }
6204 if (s[i] == '\'')
6205 break;
6206 w[i] = s[i];
6207 }
6208 }
bfa4a6ee 6209
61e29183 6210 for (; i < wsize - 1; i++) {
6211 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
6212 break;
6213 w[i] = s[i];
6214 }
6215 w[i] = 0;
6216
6217 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
6218 printf("warning: '%s' truncated\n", w);
bfa4a6ee 6219
61e29183 6220 return s + i;
bfa4a6ee 6221}
6222
5fa1256f 6223static void scan_variables(FILE *fasm)
6224{
5fa1256f 6225 struct scanned_var *var;
6226 char line[256] = { 0, };
61e29183 6227 char words[3][256];
5fa1256f 6228 char *p = NULL;
6229 int wordc;
61e29183 6230 int l;
5fa1256f 6231
6232 while (!feof(fasm))
6233 {
6234 // skip to next data section
6235 while (my_fgets(line, sizeof(line), fasm))
6236 {
6237 asmln++;
6238
6239 p = sskip(line);
6240 if (*p == 0 || *p == ';')
6241 continue;
6242
6243 p = sskip(next_word_s(words[0], sizeof(words[0]), p));
6244 if (*p == 0 || *p == ';')
6245 continue;
6246
6247 if (*p != 's' || !IS_START(p, "segment para public"))
6248 continue;
6249
6250 break;
6251 }
6252
6253 if (p == NULL || !IS_START(p, "segment para public"))
6254 break;
6255 p = sskip(p + 19);
6256
6257 if (!IS_START(p, "'DATA'"))
6258 continue;
6259
6260 // now process it
6261 while (my_fgets(line, sizeof(line), fasm))
6262 {
6263 asmln++;
6264
6265 p = line;
6266 if (my_isblank(*p))
6267 continue;
6268
6269 p = sskip(p);
6270 if (*p == 0 || *p == ';')
6271 continue;
6272
6273 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
6274 words[wordc][0] = 0;
6275 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
6276 if (*p == 0 || *p == ';') {
6277 wordc++;
6278 break;
6279 }
6280 }
6281
6282 if (wordc == 2 && IS(words[1], "ends"))
6283 break;
61e29183 6284 if (wordc < 2)
6285 continue;
5fa1256f 6286
6287 if ((hg_var_cnt & 0xff) == 0) {
6288 hg_vars = realloc(hg_vars, sizeof(hg_vars[0])
6289 * (hg_var_cnt + 0x100));
6290 my_assert_not(hg_vars, NULL);
6291 memset(hg_vars + hg_var_cnt, 0, sizeof(hg_vars[0]) * 0x100);
6292 }
6293
6294 var = &hg_vars[hg_var_cnt++];
6295 snprintf(var->name, sizeof(var->name), "%s", words[0]);
6296
6297 // maybe already in seed header?
c0de9015 6298 var->pp = proto_parse(g_fhdr, var->name, 1);
6299 if (var->pp != NULL) {
6300 if (var->pp->is_fptr) {
5fa1256f 6301 var->lmod = OPLM_DWORD;
6302 //var->is_ptr = 1;
6303 }
c0de9015 6304 else if (var->pp->is_func)
6305 aerr("func?\n");
6306 else if (!guess_lmod_from_c_type(&var->lmod, &var->pp->type))
5fa1256f 6307 aerr("unhandled C type '%s' for '%s'\n",
c0de9015 6308 var->pp->type.name, var->name);
5fa1256f 6309
6310 var->is_seeded = 1;
6311 continue;
6312 }
6313
6314 if (IS(words[1], "dd"))
6315 var->lmod = OPLM_DWORD;
6316 else if (IS(words[1], "dw"))
6317 var->lmod = OPLM_WORD;
61e29183 6318 else if (IS(words[1], "db")) {
5fa1256f 6319 var->lmod = OPLM_BYTE;
61e29183 6320 if (wordc >= 3 && (l = strlen(words[2])) > 4) {
6321 if (words[2][0] == '\'' && IS(words[2] + l - 2, ",0"))
6322 var->is_c_str = 1;
6323 }
6324 }
5fa1256f 6325 else if (IS(words[1], "dq"))
6326 var->lmod = OPLM_QWORD;
6327 //else if (IS(words[1], "dt"))
6328 else
6329 aerr("type '%s' not known\n", words[1]);
6330 }
6331 }
6332
6333 rewind(fasm);
6334 asmln = 0;
6335}
6336
6337static void set_label(int i, const char *name)
6338{
6339 const char *p;
6340 int len;
6341
6342 len = strlen(name);
6343 p = strchr(name, ':');
6344 if (p != NULL)
6345 len = p - name;
6346
6347 if (g_labels[i] != NULL && !IS_START(g_labels[i], "algn_"))
6348 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
6349 g_labels[i] = realloc(g_labels[i], len + 1);
6350 my_assert_not(g_labels[i], NULL);
6351 memcpy(g_labels[i], name, len);
6352 g_labels[i][len] = 0;
6353}
6354
e56ab892 6355struct chunk_item {
6356 char *name;
6357 long fptr;
de50b98b 6358 int asmln;
e56ab892 6359};
6360
cdfaeed7 6361static struct chunk_item *func_chunks;
6362static int func_chunk_cnt;
6363static int func_chunk_alloc;
6364
6365static void add_func_chunk(FILE *fasm, const char *name, int line)
6366{
6367 if (func_chunk_cnt >= func_chunk_alloc) {
6368 func_chunk_alloc *= 2;
6369 func_chunks = realloc(func_chunks,
6370 func_chunk_alloc * sizeof(func_chunks[0]));
6371 my_assert_not(func_chunks, NULL);
6372 }
6373 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
6374 func_chunks[func_chunk_cnt].name = strdup(name);
6375 func_chunks[func_chunk_cnt].asmln = line;
6376 func_chunk_cnt++;
6377}
6378
e56ab892 6379static int cmp_chunks(const void *p1, const void *p2)
6380{
6381 const struct chunk_item *c1 = p1, *c2 = p2;
6382 return strcmp(c1->name, c2->name);
6383}
6384
bfa4a6ee 6385static int cmpstringp(const void *p1, const void *p2)
6386{
6387 return strcmp(*(char * const *)p1, *(char * const *)p2);
6388}
6389
cdfaeed7 6390static void scan_ahead(FILE *fasm)
6391{
6392 char words[2][256];
6393 char line[256];
6394 long oldpos;
6395 int oldasmln;
6396 int wordc;
6397 char *p;
6398 int i;
6399
6400 oldpos = ftell(fasm);
6401 oldasmln = asmln;
6402
5fa1256f 6403 while (my_fgets(line, sizeof(line), fasm))
cdfaeed7 6404 {
6405 wordc = 0;
6406 asmln++;
6407
6408 p = sskip(line);
6409 if (*p == 0)
6410 continue;
6411
6412 if (*p == ';')
6413 {
6414 // get rid of random tabs
6415 for (i = 0; line[i] != 0; i++)
6416 if (line[i] == '\t')
6417 line[i] = ' ';
6418
6419 if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
6420 {
6421 p += 30;
6422 next_word(words[0], sizeof(words[0]), p);
6423 if (words[0][0] == 0)
6424 aerr("missing name for func chunk?\n");
6425
6426 add_func_chunk(fasm, words[0], asmln);
6427 }
46b388c2 6428 else if (IS_START(p, "; sctend"))
6429 break;
6430
cdfaeed7 6431 continue;
6432 } // *p == ';'
6433
6434 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
6435 words[wordc][0] = 0;
6436 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
6437 if (*p == 0 || *p == ';') {
6438 wordc++;
6439 break;
6440 }
6441 }
6442
6443 if (wordc == 2 && IS(words[1], "ends"))
6444 break;
6445 }
6446
6447 fseek(fasm, oldpos, SEEK_SET);
6448 asmln = oldasmln;
6449}
6450
91977a1c 6451int main(int argc, char *argv[])
6452{
06c5d854 6453 FILE *fout, *fasm, *frlist;
4c45fa73 6454 struct parsed_data *pd = NULL;
6455 int pd_alloc = 0;
6456 char **rlist = NULL;
6457 int rlist_len = 0;
6458 int rlist_alloc = 0;
e56ab892 6459 int func_chunks_used = 0;
6460 int func_chunks_sorted = 0;
e56ab892 6461 int func_chunk_i = -1;
6462 long func_chunk_ret = 0;
de50b98b 6463 int func_chunk_ret_ln = 0;
cdfaeed7 6464 int scanned_ahead = 0;
91977a1c 6465 char line[256];
a2c1d768 6466 char words[20][256];
4c45fa73 6467 enum opr_lenmod lmod;
ddaf8bd7 6468 char *sctproto = NULL;
91977a1c 6469 int in_func = 0;
4c45fa73 6470 int pending_endp = 0;
bfa4a6ee 6471 int skip_func = 0;
940e8e66 6472 int skip_warned = 0;
91977a1c 6473 int eq_alloc;
bfa4a6ee 6474 int verbose = 0;
1f84f6b3 6475 int multi_seg = 0;
46b388c2 6476 int end = 0;
bfa4a6ee 6477 int arg_out;
89ff3147 6478 int arg;
91977a1c 6479 int pi = 0;
e56ab892 6480 int i, j;
6481 int ret, len;
91977a1c 6482 char *p;
6483 int wordc;
6484
89ff3147 6485 for (arg = 1; arg < argc; arg++) {
6486 if (IS(argv[arg], "-v"))
6487 verbose = 1;
6488 else if (IS(argv[arg], "-rf"))
6489 g_allow_regfunc = 1;
1f84f6b3 6490 else if (IS(argv[arg], "-m"))
6491 multi_seg = 1;
92d715b6 6492 else if (IS(argv[arg], "-hdr"))
9af2d373 6493 g_header_mode = g_quiet_pp = g_allow_regfunc = 1;
89ff3147 6494 else
6495 break;
bfa4a6ee 6496 }
6497
6498 if (argc < arg + 3) {
315b77eb 6499 printf("usage:\n%s [-v] [-rf] [-m] <.c> <.asm> <hdr.h> [rlist]*\n"
6500 "%s -hdr <out.h> <.asm> <seed.h> [rlist]*\n"
6501 "options:\n"
6502 " -hdr - header generation mode\n"
6503 " -rf - allow unannotated indirect calls\n"
6504 " -m - allow multiple .text sections\n"
6505 "[rlist] is a file with function names to skip,"
6506 " one per line\n",
92d715b6 6507 argv[0], argv[0]);
91977a1c 6508 return 1;
6509 }
6510
bfa4a6ee 6511 arg_out = arg++;
91977a1c 6512
bfa4a6ee 6513 asmfn = argv[arg++];
91977a1c 6514 fasm = fopen(asmfn, "r");
6515 my_assert_not(fasm, NULL);
6516
bfa4a6ee 6517 hdrfn = argv[arg++];
06c5d854 6518 g_fhdr = fopen(hdrfn, "r");
6519 my_assert_not(g_fhdr, NULL);
bfa4a6ee 6520
6521 rlist_alloc = 64;
6522 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
6523 my_assert_not(rlist, NULL);
6524 // needs special handling..
6525 rlist[rlist_len++] = "__alloca_probe";
6526
e56ab892 6527 func_chunk_alloc = 32;
6528 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
6529 my_assert_not(func_chunks, NULL);
6530
a2c1d768 6531 memset(words, 0, sizeof(words));
6532
bfa4a6ee 6533 for (; arg < argc; arg++) {
6534 frlist = fopen(argv[arg], "r");
6535 my_assert_not(frlist, NULL);
6536
5fa1256f 6537 while (my_fgets(line, sizeof(line), frlist)) {
bfa4a6ee 6538 p = sskip(line);
1cd4a663 6539 if (*p == 0 || *p == ';')
6540 continue;
6541 if (*p == '#') {
89ff3147 6542 if (IS_START(p, "#if 0")
6543 || (g_allow_regfunc && IS_START(p, "#if NO_REGFUNC")))
6544 {
1cd4a663 6545 skip_func = 1;
89ff3147 6546 }
1cd4a663 6547 else if (IS_START(p, "#endif"))
6548 skip_func = 0;
6549 continue;
6550 }
6551 if (skip_func)
bfa4a6ee 6552 continue;
6553
6554 p = next_word(words[0], sizeof(words[0]), p);
6555 if (words[0][0] == 0)
6556 continue;
6557
6558 if (rlist_len >= rlist_alloc) {
6559 rlist_alloc = rlist_alloc * 2 + 64;
6560 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
6561 my_assert_not(rlist, NULL);
6562 }
6563 rlist[rlist_len++] = strdup(words[0]);
6564 }
1cd4a663 6565 skip_func = 0;
bfa4a6ee 6566
6567 fclose(frlist);
6568 frlist = NULL;
6569 }
6570
6571 if (rlist_len > 0)
6572 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
6573
6574 fout = fopen(argv[arg_out], "w");
91977a1c 6575 my_assert_not(fout, NULL);
6576
6577 eq_alloc = 128;
6578 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
6579 my_assert_not(g_eqs, NULL);
6580
4c45fa73 6581 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
6582 g_label_refs[i].i = -1;
6583 g_label_refs[i].next = NULL;
6584 }
6585
5fa1256f 6586 if (g_header_mode)
6587 scan_variables(fasm);
6588
6589 while (my_fgets(line, sizeof(line), fasm))
91977a1c 6590 {
4c45fa73 6591 wordc = 0;
91977a1c 6592 asmln++;
6593
6594 p = sskip(line);
1bafb621 6595 if (*p == 0)
91977a1c 6596 continue;
6597
de50b98b 6598 // get rid of random tabs
6599 for (i = 0; line[i] != 0; i++)
6600 if (line[i] == '\t')
6601 line[i] = ' ';
6602
e56ab892 6603 if (*p == ';')
6604 {
e56ab892 6605 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
6606 goto do_pending_endp; // eww..
6607
6608 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
6609 {
6610 static const char *attrs[] = {
6611 "bp-based frame",
6612 "library function",
6613 "static",
6614 "noreturn",
6615 "thunk",
6616 "fpd=",
6617 };
6618
6619 // parse IDA's attribute-list comment
6620 g_ida_func_attr = 0;
6621 p = sskip(p + 13);
6622
6623 for (; *p != 0; p = sskip(p)) {
6624 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
6625 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
6626 g_ida_func_attr |= 1 << i;
6627 p += strlen(attrs[i]);
6628 break;
6629 }
6630 }
6631 if (i == ARRAY_SIZE(attrs)) {
6632 anote("unparsed IDA attr: %s\n", p);
1bafb621 6633 break;
6634 }
e56ab892 6635 if (IS(attrs[i], "fpd=")) {
6636 p = next_word(words[0], sizeof(words[0]), p);
6637 // ignore for now..
6638 }
1bafb621 6639 }
e56ab892 6640 }
6641 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
6642 {
6643 p += 30;
6644 next_word(words[0], sizeof(words[0]), p);
6645 if (words[0][0] == 0)
cdfaeed7 6646 aerr("missing name for func chunk?\n");
6647
6648 if (!scanned_ahead) {
6649 add_func_chunk(fasm, words[0], asmln);
6650 func_chunks_sorted = 0;
e56ab892 6651 }
e56ab892 6652 }
6653 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
6654 {
6655 if (func_chunk_i >= 0) {
6656 if (func_chunk_i < func_chunk_cnt
6657 && IS(func_chunks[func_chunk_i].name, g_func))
6658 {
6659 // move on to next chunk
6660 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
6661 if (ret)
6662 aerr("seek failed for '%s' chunk #%d\n",
6663 g_func, func_chunk_i);
de50b98b 6664 asmln = func_chunks[func_chunk_i].asmln;
e56ab892 6665 func_chunk_i++;
6666 }
6667 else {
6668 if (func_chunk_ret == 0)
6669 aerr("no return from chunk?\n");
6670 fseek(fasm, func_chunk_ret, SEEK_SET);
de50b98b 6671 asmln = func_chunk_ret_ln;
e56ab892 6672 func_chunk_ret = 0;
6673 pending_endp = 1;
6674 }
1bafb621 6675 }
e56ab892 6676 }
6677 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
6678 func_chunks_used = 1;
6679 p += 20;
6680 if (IS_START(g_func, "sub_")) {
6681 unsigned long addr = strtoul(p, NULL, 16);
6682 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
cdfaeed7 6683 if (addr > f_addr && !scanned_ahead) {
94d447fb 6684 anote("scan_ahead caused by '%s', addr %lx\n",
6685 g_func, addr);
cdfaeed7 6686 scan_ahead(fasm);
6687 scanned_ahead = 1;
6688 func_chunks_sorted = 0;
6689 }
1bafb621 6690 }
6691 }
6692 continue;
e56ab892 6693 } // *p == ';'
1bafb621 6694
06c5d854 6695parse_words:
a2c1d768 6696 for (i = wordc; i < ARRAY_SIZE(words); i++)
6697 words[i][0] = 0;
cdfaeed7 6698 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
bfa4a6ee 6699 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
91977a1c 6700 if (*p == 0 || *p == ';') {
6701 wordc++;
6702 break;
6703 }
6704 }
a2c1d768 6705 if (*p != 0 && *p != ';')
6706 aerr("too many words\n");
91977a1c 6707
06c5d854 6708 // alow asm patches in comments
ddaf8bd7 6709 if (*p == ';') {
6710 if (IS_START(p, "; sctpatch:")) {
6711 p = sskip(p + 11);
6712 if (*p == 0 || *p == ';')
6713 continue;
6714 goto parse_words; // lame
6715 }
6716 if (IS_START(p, "; sctproto:")) {
6717 sctproto = strdup(p + 11);
6718 }
46b388c2 6719 else if (IS_START(p, "; sctend")) {
6720 end = 1;
6721 if (!pending_endp)
6722 break;
6723 }
06c5d854 6724 }
6725
91977a1c 6726 if (wordc == 0) {
6727 // shouldn't happen
6728 awarn("wordc == 0?\n");
6729 continue;
6730 }
6731
6732 // don't care about this:
6733 if (words[0][0] == '.'
6734 || IS(words[0], "include")
6735 || IS(words[0], "assume") || IS(words[1], "segment")
6736 || IS(words[0], "align"))
6737 {
6738 continue;
6739 }
6740
4c45fa73 6741do_pending_endp:
6742 // do delayed endp processing to collect switch jumptables
6743 if (pending_endp) {
46b388c2 6744 if (in_func && !skip_func && !end && wordc >= 2
4c45fa73 6745 && ((words[0][0] == 'd' && words[0][2] == 0)
6746 || (words[1][0] == 'd' && words[1][2] == 0)))
6747 {
6748 i = 1;
6749 if (words[1][0] == 'd' && words[1][2] == 0) {
6750 // label
6751 if (g_func_pd_cnt >= pd_alloc) {
6752 pd_alloc = pd_alloc * 2 + 16;
6753 g_func_pd = realloc(g_func_pd,
6754 sizeof(g_func_pd[0]) * pd_alloc);
6755 my_assert_not(g_func_pd, NULL);
6756 }
6757 pd = &g_func_pd[g_func_pd_cnt];
6758 g_func_pd_cnt++;
6759 memset(pd, 0, sizeof(*pd));
6760 strcpy(pd->label, words[0]);
6761 pd->type = OPT_CONST;
6762 pd->lmod = lmod_from_directive(words[1]);
6763 i = 2;
6764 }
6765 else {
da87ae38 6766 if (pd == NULL) {
6767 if (verbose)
6768 anote("skipping alignment byte?\n");
6769 continue;
6770 }
4c45fa73 6771 lmod = lmod_from_directive(words[0]);
6772 if (lmod != pd->lmod)
6773 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
6774 }
6775
6776 if (pd->count_alloc < pd->count + wordc) {
6777 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
6778 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
6779 my_assert_not(pd->d, NULL);
6780 }
6781 for (; i < wordc; i++) {
6782 if (IS(words[i], "offset")) {
6783 pd->type = OPT_OFFSET;
6784 i++;
6785 }
6786 p = strchr(words[i], ',');
6787 if (p != NULL)
6788 *p = 0;
6789 if (pd->type == OPT_OFFSET)
6790 pd->d[pd->count].u.label = strdup(words[i]);
6791 else
6792 pd->d[pd->count].u.val = parse_number(words[i]);
6793 pd->d[pd->count].bt_i = -1;
6794 pd->count++;
6795 }
6796 continue;
6797 }
6798
92d715b6 6799 if (in_func && !skip_func) {
9af2d373 6800 if (g_header_mode)
92d715b6 6801 gen_hdr(g_func, pi);
6802 else
6803 gen_func(fout, g_fhdr, g_func, pi);
6804 }
4c45fa73 6805
6806 pending_endp = 0;
6807 in_func = 0;
6808 g_ida_func_attr = 0;
6809 skip_warned = 0;
6810 skip_func = 0;
6811 g_func[0] = 0;
e56ab892 6812 func_chunks_used = 0;
6813 func_chunk_i = -1;
4c45fa73 6814 if (pi != 0) {
6815 memset(&ops, 0, pi * sizeof(ops[0]));
d7857c3a 6816 clear_labels(pi);
4c45fa73 6817 pi = 0;
6818 }
6819 g_eqcnt = 0;
6820 for (i = 0; i < g_func_pd_cnt; i++) {
6821 pd = &g_func_pd[i];
6822 if (pd->type == OPT_OFFSET) {
6823 for (j = 0; j < pd->count; j++)
6824 free(pd->d[j].u.label);
6825 }
6826 free(pd->d);
6827 pd->d = NULL;
6828 }
6829 g_func_pd_cnt = 0;
6830 pd = NULL;
46b388c2 6831
6832 if (end)
6833 break;
4c45fa73 6834 if (wordc == 0)
6835 continue;
6836 }
6837
91977a1c 6838 if (IS(words[1], "proc")) {
6839 if (in_func)
6840 aerr("proc '%s' while in_func '%s'?\n",
6841 words[0], g_func);
bfa4a6ee 6842 p = words[0];
ddaf8bd7 6843 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
bfa4a6ee 6844 skip_func = 1;
91977a1c 6845 strcpy(g_func, words[0]);
d4e3b5db 6846 set_label(0, words[0]);
91977a1c 6847 in_func = 1;
6848 continue;
6849 }
6850
e56ab892 6851 if (IS(words[1], "endp"))
6852 {
91977a1c 6853 if (!in_func)
6854 aerr("endp '%s' while not in_func?\n", words[0]);
6855 if (!IS(g_func, words[0]))
6856 aerr("endp '%s' while in_func '%s'?\n",
6857 words[0], g_func);
bfa4a6ee 6858
ddaf8bd7 6859 if ((g_ida_func_attr & IDAFA_THUNK) && pi == 1
6860 && ops[0].op == OP_JMP && ops[0].operand[0].had_ds)
6861 {
6862 // import jump
6863 skip_func = 1;
6864 }
6865
e56ab892 6866 if (!skip_func && func_chunks_used) {
6867 // start processing chunks
6868 struct chunk_item *ci, key = { g_func, 0 };
6869
6870 func_chunk_ret = ftell(fasm);
de50b98b 6871 func_chunk_ret_ln = asmln;
e56ab892 6872 if (!func_chunks_sorted) {
6873 qsort(func_chunks, func_chunk_cnt,
6874 sizeof(func_chunks[0]), cmp_chunks);
6875 func_chunks_sorted = 1;
6876 }
6877 ci = bsearch(&key, func_chunks, func_chunk_cnt,
6878 sizeof(func_chunks[0]), cmp_chunks);
6879 if (ci == NULL)
6880 aerr("'%s' needs chunks, but none found\n", g_func);
6881 func_chunk_i = ci - func_chunks;
6882 for (; func_chunk_i > 0; func_chunk_i--)
6883 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
6884 break;
6885
6886 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
6887 if (ret)
6888 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
de50b98b 6889 asmln = func_chunks[func_chunk_i].asmln;
e56ab892 6890 func_chunk_i++;
6891 continue;
6892 }
4c45fa73 6893 pending_endp = 1;
91977a1c 6894 continue;
6895 }
6896
1f84f6b3 6897 if (wordc == 2 && IS(words[1], "ends")) {
46b388c2 6898 if (!multi_seg) {
6899 end = 1;
6900 if (pending_endp)
6901 goto do_pending_endp;
1f84f6b3 6902 break;
46b388c2 6903 }
1f84f6b3 6904
6905 // scan for next text segment
5fa1256f 6906 while (my_fgets(line, sizeof(line), fasm)) {
1f84f6b3 6907 asmln++;
6908 p = sskip(line);
6909 if (*p == 0 || *p == ';')
6910 continue;
6911
6912 if (strstr(p, "segment para public 'CODE' use32"))
6913 break;
6914 }
6915
6916 continue;
6917 }
a2c1d768 6918
bfa4a6ee 6919 p = strchr(words[0], ':');
6920 if (p != NULL) {
d4e3b5db 6921 set_label(pi, words[0]);
bfa4a6ee 6922 continue;
6923 }
6924
6925 if (!in_func || skip_func) {
d7857c3a 6926 if (!skip_warned && !skip_func && g_labels[pi] != NULL) {
bfa4a6ee 6927 if (verbose)
6928 anote("skipping from '%s'\n", g_labels[pi]);
6929 skip_warned = 1;
6930 }
d7857c3a 6931 free(g_labels[pi]);
6932 g_labels[pi] = NULL;
bfa4a6ee 6933 continue;
6934 }
6935
ddaf8bd7 6936 if (wordc > 1 && IS(words[1], "="))
6937 {
91977a1c 6938 if (wordc != 5)
6939 aerr("unhandled equ, wc=%d\n", wordc);
6940 if (g_eqcnt >= eq_alloc) {
6941 eq_alloc *= 2;
6942 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
6943 my_assert_not(g_eqs, NULL);
6944 }
6945
6946 len = strlen(words[0]);
6947 if (len > sizeof(g_eqs[0].name) - 1)
6948 aerr("equ name too long: %d\n", len);
6949 strcpy(g_eqs[g_eqcnt].name, words[0]);
6950
6951 if (!IS(words[3], "ptr"))
6952 aerr("unhandled equ\n");
6953 if (IS(words[2], "dword"))
6954 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
6955 else if (IS(words[2], "word"))
6956 g_eqs[g_eqcnt].lmod = OPLM_WORD;
6957 else if (IS(words[2], "byte"))
6958 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
90307a99 6959 else if (IS(words[2], "qword"))
6960 g_eqs[g_eqcnt].lmod = OPLM_QWORD;
91977a1c 6961 else
6962 aerr("bad lmod: '%s'\n", words[2]);
6963
6964 g_eqs[g_eqcnt].offset = parse_number(words[4]);
6965 g_eqcnt++;
6966 continue;
6967 }
6968
6969 if (pi >= ARRAY_SIZE(ops))
6970 aerr("too many ops\n");
6971
91977a1c 6972 parse_op(&ops[pi], words, wordc);
ddaf8bd7 6973
6974 if (sctproto != NULL) {
8eb12e72 6975 if (ops[pi].op == OP_CALL || ops[pi].op == OP_JMP)
ddaf8bd7 6976 ops[pi].datap = sctproto;
6977 sctproto = NULL;
6978 }
91977a1c 6979 pi++;
91977a1c 6980 }
6981
9af2d373 6982 if (g_header_mode)
92d715b6 6983 output_hdr(fout);
6984
91977a1c 6985 fclose(fout);
6986 fclose(fasm);
06c5d854 6987 fclose(g_fhdr);
91977a1c 6988
6989 return 0;
c36e914d 6990}
91977a1c 6991
6992// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.