notaz.gp2x.de / ia32rtools.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
translate: simplify seed passthrough
[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,
30c8c549 116 // undefined
2c10ea1f 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;
30c8c549 237static int g_skip_func;
89ff3147 238static int g_allow_regfunc;
92d715b6 239static int g_quiet_pp;
9af2d373 240static int g_header_mode;
92d715b6 241
91977a1c 242#define ferr(op_, fmt, ...) do { \
8eb12e72 243 printf("%s:%d: error: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
850c9265 244 dump_op(op_), ##__VA_ARGS__); \
33c35af6 245 fcloseall(); \
91977a1c 246 exit(1); \
247} while (0)
de50b98b 248#define fnote(op_, fmt, ...) \
8eb12e72 249 printf("%s:%d: note: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
de50b98b 250 dump_op(op_), ##__VA_ARGS__)
91977a1c 251
26677139 252#define ferr_assert(op_, cond) do { \
253 if (!(cond)) ferr(op_, "assertion '%s' failed on ln :%d\n", #cond, \
254 __LINE__); \
255} while (0)
256
90307a99 257const char *regs_r32[] = {
258 "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp",
259 // not r32, but list here for easy parsing and printing
260 "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7",
261};
91977a1c 262const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" };
263const char *regs_r8l[] = { "al", "bl", "cl", "dl" };
264const char *regs_r8h[] = { "ah", "bh", "ch", "dh" };
265
266enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP };
267
69a3cdfc 268// possible basic comparison types (without inversion)
269enum parsed_flag_op {
270 PFO_O, // 0 OF=1
271 PFO_C, // 2 CF=1
272 PFO_Z, // 4 ZF=1
273 PFO_BE, // 6 CF=1||ZF=1
274 PFO_S, // 8 SF=1
275 PFO_P, // a PF=1
276 PFO_L, // c SF!=OF
277 PFO_LE, // e ZF=1||SF!=OF
278};
279
90307a99 280#define PFOB_O (1 << PFO_O)
281#define PFOB_C (1 << PFO_C)
282#define PFOB_Z (1 << PFO_Z)
283#define PFOB_S (1 << PFO_S)
284
69a3cdfc 285static const char *parsed_flag_op_names[] = {
286 "o", "c", "z", "be", "s", "p", "l", "le"
287};
288
91977a1c 289static int char_array_i(const char *array[], size_t len, const char *s)
290{
291 int i;
c36e914d 292
91977a1c 293 for (i = 0; i < len; i++)
294 if (IS(s, array[i]))
295 return i;
c36e914d 296
91977a1c 297 return -1;
298}
299
63df67be 300static void printf_number(char *buf, size_t buf_size,
301 unsigned long number)
91977a1c 302{
5101a5f9 303 // output in C-friendly form
304 snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
305}
91977a1c 306
fdd5548a 307static int check_segment_prefix(const char *s)
308{
309 if (s[0] == 0 || s[1] != 's' || s[2] != ':')
310 return 0;
311
312 switch (s[0]) {
313 case 'c': return 1;
314 case 'd': return 2;
315 case 's': return 3;
316 case 'e': return 4;
317 case 'f': return 5;
318 case 'g': return 6;
319 default: return 0;
320 }
321}
322
5101a5f9 323static int parse_reg(enum opr_lenmod *reg_lmod, const char *s)
324{
325 int reg;
91977a1c 326
5101a5f9 327 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s);
90307a99 328 if (reg >= 8) {
329 *reg_lmod = OPLM_QWORD;
330 return reg;
331 }
5101a5f9 332 if (reg >= 0) {
333 *reg_lmod = OPLM_DWORD;
334 return reg;
91977a1c 335 }
5101a5f9 336 reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s);
337 if (reg >= 0) {
338 *reg_lmod = OPLM_WORD;
339 return reg;
340 }
341 reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s);
342 if (reg >= 0) {
343 *reg_lmod = OPLM_BYTE;
344 return reg;
345 }
346 reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s);
347 if (reg >= 0) {
348 *reg_lmod = OPLM_BYTE;
349 return reg;
850c9265 350 }
351
352 return -1;
91977a1c 353}
354
5101a5f9 355static int parse_indmode(char *name, int *regmask, int need_c_cvt)
356{
357 enum opr_lenmod lmod;
358 char cvtbuf[256];
359 char *d = cvtbuf;
360 char *s = name;
361 char w[64];
362 long number;
363 int reg;
364 int c = 0;
365
366 *d = 0;
367
368 while (*s != 0) {
369 d += strlen(d);
370 while (my_isblank(*s))
371 s++;
372 for (; my_issep(*s); d++, s++)
373 *d = *s;
374 while (my_isblank(*s))
375 s++;
376 *d = 0;
377
fdd5548a 378 // skip '?s:' prefixes
379 if (check_segment_prefix(s))
87bf6cec 380 s += 3;
381
5101a5f9 382 s = next_idt(w, sizeof(w), s);
383 if (w[0] == 0)
384 break;
385 c++;
386
387 reg = parse_reg(&lmod, w);
388 if (reg >= 0) {
389 *regmask |= 1 << reg;
390 goto pass;
391 }
392
393 if ('0' <= w[0] && w[0] <= '9') {
394 number = parse_number(w);
395 printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number);
396 continue;
397 }
398
399 // probably some label/identifier - pass
400
401pass:
402 snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w);
403 }
404
405 if (need_c_cvt)
406 strcpy(name, cvtbuf);
407
408 return c;
409}
410
4c45fa73 411static int is_reg_in_str(const char *s)
412{
413 int i;
414
415 if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
416 return 0;
417
418 for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
419 if (!strncmp(s, regs_r32[i], 3))
420 return 1;
421
422 return 0;
423}
424
09bc6fd5 425static const char *parse_stack_el(const char *name, char *extra_reg,
426 int early_try)
d4e3b5db 427{
4c45fa73 428 const char *p, *p2, *s;
1bafb621 429 char *endp = NULL;
d4e3b5db 430 char buf[32];
1bafb621 431 long val;
d4e3b5db 432 int len;
433
09bc6fd5 434 if (g_bp_frame || early_try)
435 {
436 p = name;
437 if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
438 p += 4;
439 if (extra_reg != NULL) {
440 strncpy(extra_reg, name, 3);
441 extra_reg[4] = 0;
442 }
4c45fa73 443 }
4c45fa73 444
09bc6fd5 445 if (IS_START(p, "ebp+")) {
446 p += 4;
4c45fa73 447
09bc6fd5 448 p2 = strchr(p, '+');
449 if (p2 != NULL && is_reg_in_str(p)) {
450 if (extra_reg != NULL) {
451 strncpy(extra_reg, p, p2 - p);
452 extra_reg[p2 - p] = 0;
453 }
454 p = p2 + 1;
4c45fa73 455 }
4c45fa73 456
09bc6fd5 457 if (!('0' <= *p && *p <= '9'))
458 return p;
4c45fa73 459
09bc6fd5 460 return NULL;
461 }
4c45fa73 462 }
463
39b168b8 464 if (!IS_START(name, "esp+"))
d4e3b5db 465 return NULL;
466
037f4971 467 s = name + 4;
468 p = strchr(s, '+');
d4e3b5db 469 if (p) {
037f4971 470 if (is_reg_in_str(s)) {
471 if (extra_reg != NULL) {
472 strncpy(extra_reg, s, p - s);
473 extra_reg[p - s] = 0;
474 }
475 s = p + 1;
476 p = strchr(s, '+');
477 if (p == NULL)
478 aerr("%s IDA stackvar not set?\n", __func__);
479 }
1bafb621 480 if (!('0' <= *s && *s <= '9')) {
037f4971 481 aerr("%s IDA stackvar offset not set?\n", __func__);
d4e3b5db 482 return NULL;
1bafb621 483 }
484 if (s[0] == '0' && s[1] == 'x')
485 s += 2;
486 len = p - s;
d4e3b5db 487 if (len < sizeof(buf) - 1) {
1bafb621 488 strncpy(buf, s, len);
d4e3b5db 489 buf[len] = 0;
1bafb621 490 val = strtol(buf, &endp, 16);
491 if (val == 0 || *endp != 0) {
492 aerr("%s num parse fail for '%s'\n", __func__, buf);
493 return NULL;
494 }
d4e3b5db 495 }
496 p++;
497 }
498 else
499 p = name + 4;
500
501 if ('0' <= *p && *p <= '9')
502 return NULL;
503
504 return p;
505}
506
850c9265 507static int guess_lmod_from_name(struct parsed_opr *opr)
508{
509 if (!strncmp(opr->name, "dword_", 6)) {
510 opr->lmod = OPLM_DWORD;
511 return 1;
512 }
513 if (!strncmp(opr->name, "word_", 5)) {
514 opr->lmod = OPLM_WORD;
515 return 1;
516 }
517 if (!strncmp(opr->name, "byte_", 5)) {
518 opr->lmod = OPLM_BYTE;
519 return 1;
520 }
90307a99 521 if (!strncmp(opr->name, "qword_", 6)) {
522 opr->lmod = OPLM_QWORD;
523 return 1;
524 }
850c9265 525 return 0;
526}
527
3ebea2cf 528static int guess_lmod_from_c_type(enum opr_lenmod *lmod,
529 const struct parsed_type *c_type)
06c5d854 530{
06c5d854 531 static const char *dword_types[] = {
c0de9015 532 "uint32_t", "int", "_DWORD", "UINT_PTR", "DWORD",
4741fdfe 533 "WPARAM", "LPARAM", "UINT", "__int32",
09bc6fd5 534 "LONG", "HIMC", "BOOL", "size_t",
feb0ee5d 535 "float",
06c5d854 536 };
537 static const char *word_types[] = {
fe18709a 538 "uint16_t", "int16_t", "_WORD", "WORD",
2b43685d 539 "unsigned __int16", "__int16",
06c5d854 540 };
541 static const char *byte_types[] = {
2b43685d 542 "uint8_t", "int8_t", "char",
89ff3147 543 "unsigned __int8", "__int8", "BYTE", "_BYTE",
4741fdfe 544 "CHAR", "_UNKNOWN",
b4878d2b 545 // structures.. deal the same as with _UNKNOWN for now
546 "CRITICAL_SECTION",
06c5d854 547 };
3ebea2cf 548 const char *n;
06c5d854 549 int i;
550
3ebea2cf 551 if (c_type->is_ptr) {
552 *lmod = OPLM_DWORD;
06c5d854 553 return 1;
554 }
555
3ebea2cf 556 n = skip_type_mod(c_type->name);
06c5d854 557
3ebea2cf 558 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
559 if (IS(n, dword_types[i])) {
560 *lmod = OPLM_DWORD;
06c5d854 561 return 1;
562 }
563 }
564
565 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
3ebea2cf 566 if (IS(n, word_types[i])) {
567 *lmod = OPLM_WORD;
06c5d854 568 return 1;
569 }
570 }
571
572 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
3ebea2cf 573 if (IS(n, byte_types[i])) {
574 *lmod = OPLM_BYTE;
06c5d854 575 return 1;
576 }
577 }
578
06c5d854 579 return 0;
580}
581
c7ed83dd 582static char *default_cast_to(char *buf, size_t buf_size,
583 struct parsed_opr *opr)
584{
585 buf[0] = 0;
586
587 if (!opr->is_ptr)
588 return buf;
589 if (opr->pp == NULL || opr->pp->type.name == NULL
590 || opr->pp->is_fptr)
591 {
592 snprintf(buf, buf_size, "%s", "(void *)");
593 return buf;
594 }
595
596 snprintf(buf, buf_size, "(%s)", opr->pp->type.name);
597 return buf;
598}
599
4c45fa73 600static enum opr_type lmod_from_directive(const char *d)
601{
602 if (IS(d, "dd"))
603 return OPLM_DWORD;
604 else if (IS(d, "dw"))
605 return OPLM_WORD;
606 else if (IS(d, "db"))
607 return OPLM_BYTE;
608
609 aerr("unhandled directive: '%s'\n", d);
610 return OPLM_UNSPEC;
611}
612
5101a5f9 613static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
614 int *regmask)
615{
616 opr->type = OPT_REG;
617 opr->reg = reg;
618 opr->lmod = lmod;
619 *regmask |= 1 << reg;
620}
621
39b168b8 622static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
623 int *extra_offs);
87bf6cec 624
69a3cdfc 625static int parse_operand(struct parsed_opr *opr,
626 int *regmask, int *regmask_indirect,
1bafb621 627 char words[16][256], int wordc, int w, unsigned int op_flags)
c36e914d 628{
92d715b6 629 const struct parsed_proto *pp = NULL;
850c9265 630 enum opr_lenmod tmplmod;
63df67be 631 unsigned long number;
89ff3147 632 char buf[256];
850c9265 633 int ret, len;
1bafb621 634 int wordc_in;
89ff3147 635 char *p;
850c9265 636 int i;
c36e914d 637
1bafb621 638 if (w >= wordc)
639 aerr("parse_operand w %d, wordc %d\n", w, wordc);
c36e914d 640
1bafb621 641 opr->reg = xUNSPEC;
91977a1c 642
1bafb621 643 for (i = w; i < wordc; i++) {
644 len = strlen(words[i]);
645 if (words[i][len - 1] == ',') {
646 words[i][len - 1] = 0;
647 wordc = i + 1;
648 break;
649 }
650 }
c36e914d 651
1bafb621 652 wordc_in = wordc - w;
c36e914d 653
1bafb621 654 if ((op_flags & OPF_JMP) && wordc_in > 0
655 && !('0' <= words[w][0] && words[w][0] <= '9'))
656 {
657 const char *label = NULL;
658
659 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
660 && IS(words[w + 1], "ptr"))
661 label = words[w + 2];
662 else if (wordc_in == 2 && IS(words[w], "short"))
663 label = words[w + 1];
664 else if (wordc_in == 1
665 && strchr(words[w], '[') == NULL
666 && parse_reg(&tmplmod, words[w]) < 0)
667 label = words[w];
668
669 if (label != NULL) {
670 opr->type = OPT_LABEL;
fdd5548a 671 ret = check_segment_prefix(label);
672 if (ret != 0) {
673 if (ret >= 5)
674 aerr("fs/gs used\n");
ddaf8bd7 675 opr->had_ds = 1;
39b168b8 676 label += 3;
ddaf8bd7 677 }
1bafb621 678 strcpy(opr->name, label);
679 return wordc;
680 }
681 }
c36e914d 682
1bafb621 683 if (wordc_in >= 3) {
684 if (IS(words[w + 1], "ptr")) {
685 if (IS(words[w], "dword"))
686 opr->lmod = OPLM_DWORD;
687 else if (IS(words[w], "word"))
688 opr->lmod = OPLM_WORD;
689 else if (IS(words[w], "byte"))
690 opr->lmod = OPLM_BYTE;
90307a99 691 else if (IS(words[w], "qword"))
692 opr->lmod = OPLM_QWORD;
1bafb621 693 else
694 aerr("type parsing failed\n");
695 w += 2;
696 wordc_in = wordc - w;
697 }
698 }
699
39b168b8 700 if (wordc_in == 2) {
701 if (IS(words[w], "offset")) {
702 opr->type = OPT_OFFSET;
27ebfaed 703 opr->lmod = OPLM_DWORD;
39b168b8 704 strcpy(opr->name, words[w + 1]);
27ebfaed 705 pp = proto_parse(g_fhdr, opr->name, 1);
706 goto do_label;
39b168b8 707 }
708 if (IS(words[w], "(offset")) {
89ff3147 709 p = strchr(words[w + 1], ')');
39b168b8 710 if (p == NULL)
711 aerr("parse of bracketed offset failed\n");
712 *p = 0;
713 opr->type = OPT_OFFSET;
714 strcpy(opr->name, words[w + 1]);
715 return wordc;
716 }
1bafb621 717 }
c36e914d 718
1bafb621 719 if (wordc_in != 1)
850c9265 720 aerr("parse_operand 1 word expected\n");
c36e914d 721
fdd5548a 722 ret = check_segment_prefix(words[w]);
723 if (ret != 0) {
724 if (ret >= 5)
725 aerr("fs/gs used\n");
ddaf8bd7 726 opr->had_ds = 1;
89ff3147 727 memmove(words[w], words[w] + 3, strlen(words[w]) - 2);
ddaf8bd7 728 }
89ff3147 729 strcpy(opr->name, words[w]);
c36e914d 730
850c9265 731 if (words[w][0] == '[') {
732 opr->type = OPT_REGMEM;
733 ret = sscanf(words[w], "[%[^]]]", opr->name);
734 if (ret != 1)
735 aerr("[] parse failure\n");
87bf6cec 736
5101a5f9 737 parse_indmode(opr->name, regmask_indirect, 1);
09bc6fd5 738 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL, 1))
739 {
87bf6cec 740 // might be an equ
d4e3b5db 741 struct parsed_equ *eq =
09bc6fd5 742 equ_find(NULL, parse_stack_el(opr->name, NULL, 1), &i);
87bf6cec 743 if (eq)
744 opr->lmod = eq->lmod;
745 }
850c9265 746 return wordc;
747 }
748 else if (strchr(words[w], '[')) {
749 // label[reg] form
89ff3147 750 p = strchr(words[w], '[');
850c9265 751 opr->type = OPT_REGMEM;
89ff3147 752 parse_indmode(p, regmask_indirect, 0);
753 strncpy(buf, words[w], p - words[w]);
754 buf[p - words[w]] = 0;
755 pp = proto_parse(g_fhdr, buf, 1);
756 goto do_label;
850c9265 757 }
758 else if (('0' <= words[w][0] && words[w][0] <= '9')
759 || words[w][0] == '-')
760 {
5101a5f9 761 number = parse_number(words[w]);
91977a1c 762 opr->type = OPT_CONST;
5101a5f9 763 opr->val = number;
764 printf_number(opr->name, sizeof(opr->name), number);
91977a1c 765 return wordc;
850c9265 766 }
c36e914d 767
5101a5f9 768 ret = parse_reg(&tmplmod, opr->name);
769 if (ret >= 0) {
770 setup_reg_opr(opr, ret, tmplmod, regmask);
850c9265 771 return wordc;
772 }
773
774 // most likely var in data segment
775 opr->type = OPT_LABEL;
92d715b6 776 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
89ff3147 777
778do_label:
bd96f656 779 if (pp != NULL) {
1cd4a663 780 if (pp->is_fptr || pp->is_func) {
06c5d854 781 opr->lmod = OPLM_DWORD;
782 opr->is_ptr = 1;
783 }
2b43685d 784 else {
840257f6 785 tmplmod = OPLM_UNSPEC;
bd96f656 786 if (!guess_lmod_from_c_type(&tmplmod, &pp->type))
2b43685d 787 anote("unhandled C type '%s' for '%s'\n",
bd96f656 788 pp->type.name, opr->name);
2b43685d 789
a3684be1 790 if (opr->lmod == OPLM_UNSPEC) {
2b43685d 791 opr->lmod = tmplmod;
a3684be1 792 opr->type_from_var = 1;
793 }
2b43685d 794 else if (opr->lmod != tmplmod) {
795 opr->size_mismatch = 1;
796 if (tmplmod < opr->lmod)
797 opr->size_lt = 1;
798 }
a652aa9f 799 opr->is_ptr = pp->type.is_ptr;
3ebea2cf 800 }
bd96f656 801 opr->is_array = pp->type.is_array;
06c5d854 802 }
c7ed83dd 803 opr->pp = pp;
06c5d854 804
850c9265 805 if (opr->lmod == OPLM_UNSPEC)
806 guess_lmod_from_name(opr);
850c9265 807 return wordc;
c36e914d 808}
809
33c35af6 810static const struct {
811 const char *name;
812 unsigned int flags;
813} pref_table[] = {
814 { "rep", OPF_REP },
7ba45c34 815 { "repe", OPF_REP|OPF_REPZ },
816 { "repz", OPF_REP|OPF_REPZ },
817 { "repne", OPF_REP|OPF_REPNZ },
818 { "repnz", OPF_REP|OPF_REPNZ },
037f4971 819 { "lock", OPF_LOCK }, // ignored for now..
33c35af6 820};
821
5c024ef7 822#define OPF_CJMP_CC (OPF_JMP|OPF_CJMP|OPF_CC)
823
c36e914d 824static const struct {
850c9265 825 const char *name;
826 enum op_op op;
092f64e1 827 unsigned short minopr;
828 unsigned short maxopr;
69a3cdfc 829 unsigned int flags;
092f64e1 830 unsigned char pfo;
831 unsigned char pfo_inv;
c36e914d 832} op_table[] = {
33c35af6 833 { "nop", OP_NOP, 0, 0, 0 },
69a3cdfc 834 { "push", OP_PUSH, 1, 1, 0 },
835 { "pop", OP_POP, 1, 1, OPF_DATA },
591721d7 836 { "leave",OP_LEAVE, 0, 0, OPF_DATA },
69a3cdfc 837 { "mov" , OP_MOV, 2, 2, OPF_DATA },
838 { "lea", OP_LEA, 2, 2, OPF_DATA },
839 { "movzx",OP_MOVZX, 2, 2, OPF_DATA },
840 { "movsx",OP_MOVSX, 2, 2, OPF_DATA },
108e9fe3 841 { "xchg", OP_XCHG, 2, 2, OPF_DATA },
69a3cdfc 842 { "not", OP_NOT, 1, 1, OPF_DATA },
5101a5f9 843 { "cdq", OP_CDQ, 0, 0, OPF_DATA },
092f64e1 844 { "lodsb",OP_LODS, 0, 0, OPF_DATA },
845 { "lodsw",OP_LODS, 0, 0, OPF_DATA },
846 { "lodsd",OP_LODS, 0, 0, OPF_DATA },
33c35af6 847 { "stosb",OP_STOS, 0, 0, OPF_DATA },
848 { "stosw",OP_STOS, 0, 0, OPF_DATA },
849 { "stosd",OP_STOS, 0, 0, OPF_DATA },
d4e3b5db 850 { "movsb",OP_MOVS, 0, 0, OPF_DATA },
851 { "movsw",OP_MOVS, 0, 0, OPF_DATA },
852 { "movsd",OP_MOVS, 0, 0, OPF_DATA },
7ba45c34 853 { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
854 { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
855 { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
591721d7 856 { "scasb",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
857 { "scasw",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
858 { "scasd",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
859 { "std", OP_STD, 0, 0, OPF_DATA }, // special flag
860 { "cld", OP_CLD, 0, 0, OPF_DATA },
69a3cdfc 861 { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
862 { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
863 { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
864 { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS },
865 { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS },
866 { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
867 { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
868 { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
869 { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
3b2f4044 870 { "shrd", OP_SHRD, 3, 3, OPF_DATA|OPF_FLAGS },
d4e3b5db 871 { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
872 { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
092f64e1 873 { "rcl", OP_RCL, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
874 { "rcr", OP_RCR, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
875 { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
876 { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
1f84f6b3 877 { "bsf", OP_BSF, 2, 2, OPF_DATA|OPF_FLAGS },
69a3cdfc 878 { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
879 { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS },
5101a5f9 880 { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS },
881 { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS },
69a3cdfc 882 { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS },
5101a5f9 883 { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS },
884 { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS },
69a3cdfc 885 { "test", OP_TEST, 2, 2, OPF_FLAGS },
886 { "cmp", OP_CMP, 2, 2, OPF_FLAGS },
a3684be1 887 { "retn", OP_RET, 0, 1, OPF_TAIL },
de50b98b 888 { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS },
69a3cdfc 889 { "jmp", OP_JMP, 1, 1, OPF_JMP },
5c024ef7 890 { "jecxz",OP_JECXZ, 1, 1, OPF_JMP|OPF_CJMP },
092f64e1 891 { "jo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 0 }, // 70 OF=1
892 { "jno", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 1 }, // 71 OF=0
893 { "jc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72 CF=1
894 { "jb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72
895 { "jnc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73 CF=0
896 { "jnb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
897 { "jae", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
898 { "jz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74 ZF=1
899 { "je", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74
900 { "jnz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75 ZF=0
901 { "jne", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75
902 { "jbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76 CF=1||ZF=1
903 { "jna", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76
904 { "ja", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77 CF=0&&ZF=0
905 { "jnbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77
906 { "js", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 0 }, // 78 SF=1
907 { "jns", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 1 }, // 79 SF=0
908 { "jp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a PF=1
909 { "jpe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a
910 { "jnp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b PF=0
911 { "jpo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b
912 { "jl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c SF!=OF
913 { "jnge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c
914 { "jge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d SF=OF
915 { "jnl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d
916 { "jle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e ZF=1||SF!=OF
917 { "jng", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e
918 { "jg", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f ZF=0&&SF=OF
919 { "jnle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f
920 { "seto", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 0 },
921 { "setno", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 1 },
922 { "setc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
923 { "setb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
924 { "setnc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
925 { "setae", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
926 { "setnb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
927 { "setz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
928 { "sete", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
929 { "setnz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
930 { "setne", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
931 { "setbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
932 { "setna", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
933 { "seta", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
934 { "setnbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
935 { "sets", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 0 },
936 { "setns", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 1 },
937 { "setp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
938 { "setpe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
939 { "setnp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
940 { "setpo", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
941 { "setl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
942 { "setnge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
943 { "setge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
944 { "setnl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
945 { "setle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
946 { "setng", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
947 { "setg", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
948 { "setnle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
90307a99 949 // x87
950 // mmx
951 { "emms", OP_EMMS, 0, 0, OPF_DATA },
952 { "movq", OP_MOV, 2, 2, OPF_DATA },
2c10ea1f 953 // must be last
954 { "ud2", OP_UD2 },
69a3cdfc 955};
c36e914d 956
957static void parse_op(struct parsed_op *op, char words[16][256], int wordc)
958{
bfa4a6ee 959 enum opr_lenmod lmod = OPLM_UNSPEC;
33c35af6 960 int prefix_flags = 0;
69a3cdfc 961 int regmask_ind;
962 int regmask;
33c35af6 963 int op_w = 0;
91977a1c 964 int opr = 0;
33c35af6 965 int w = 0;
91977a1c 966 int i;
c36e914d 967
33c35af6 968 for (i = 0; i < ARRAY_SIZE(pref_table); i++) {
969 if (IS(words[w], pref_table[i].name)) {
970 prefix_flags = pref_table[i].flags;
971 break;
972 }
973 }
974
975 if (prefix_flags) {
976 if (wordc <= 1)
977 aerr("lone prefix: '%s'\n", words[0]);
978 w++;
979 }
980
981 op_w = w;
91977a1c 982 for (i = 0; i < ARRAY_SIZE(op_table); i++) {
33c35af6 983 if (IS(words[w], op_table[i].name))
69a3cdfc 984 break;
985 }
c36e914d 986
2c10ea1f 987 if (i == ARRAY_SIZE(op_table)) {
30c8c549 988 if (!g_skip_func)
989 aerr("unhandled op: '%s'\n", words[0]);
2c10ea1f 990 i--; // OP_UD2
991 }
33c35af6 992 w++;
c36e914d 993
69a3cdfc 994 op->op = op_table[i].op;
33c35af6 995 op->flags = op_table[i].flags | prefix_flags;
092f64e1 996 op->pfo = op_table[i].pfo;
997 op->pfo_inv = op_table[i].pfo_inv;
69a3cdfc 998 op->regmask_src = op->regmask_dst = 0;
8eb12e72 999 op->asmln = asmln;
69a3cdfc 1000
2c10ea1f 1001 if (op->op == OP_UD2)
1002 return;
1003
92d715b6 1004 for (opr = 0; opr < op_table[i].maxopr; opr++) {
1005 if (opr >= op_table[i].minopr && w >= wordc)
1006 break;
1007
69a3cdfc 1008 regmask = regmask_ind = 0;
1009 w = parse_operand(&op->operand[opr], &regmask, &regmask_ind,
1010 words, wordc, w, op->flags);
1011
1012 if (opr == 0 && (op->flags & OPF_DATA))
1013 op->regmask_dst = regmask;
92d715b6 1014 else
1015 op->regmask_src |= regmask;
1016 op->regmask_src |= regmask_ind;
91977a1c 1017 }
c36e914d 1018
91977a1c 1019 if (w < wordc)
1020 aerr("parse_op %s incomplete: %d/%d\n",
1021 words[0], w, wordc);
5101a5f9 1022
1023 // special cases
1024 op->operand_cnt = opr;
1025 if (!strncmp(op_table[i].name, "set", 3))
1026 op->operand[0].lmod = OPLM_BYTE;
1027
5101a5f9 1028 switch (op->op) {
92d715b6 1029 // first operand is not dst
1030 case OP_CMP:
1031 case OP_TEST:
1032 op->regmask_src |= op->regmask_dst;
1033 op->regmask_dst = 0;
1034 break;
1035
1036 // first operand is src too
1037 case OP_NOT:
1038 case OP_ADD:
1039 case OP_AND:
1040 case OP_OR:
1041 case OP_RCL:
1042 case OP_RCR:
1043 case OP_ADC:
1044 case OP_INC:
1045 case OP_DEC:
1046 case OP_NEG:
1047 // more below..
1048 op->regmask_src |= op->regmask_dst;
1049 break;
1050
1051 // special
1052 case OP_XCHG:
1053 op->regmask_src |= op->regmask_dst;
1054 op->regmask_dst |= op->regmask_src;
1055 goto check_align;
1056
1057 case OP_SUB:
1058 case OP_SBB:
1059 case OP_XOR:
1060 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1061 && op->operand[0].lmod == op->operand[1].lmod
1062 && op->operand[0].reg == op->operand[1].reg
1063 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
1064 {
1065 op->regmask_src = 0;
1066 }
1067 else
1068 op->regmask_src |= op->regmask_dst;
1069 break;
1070
1071 // ops with implicit argumets
5101a5f9 1072 case OP_CDQ:
1073 op->operand_cnt = 2;
1074 setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst);
1075 setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src);
1076 break;
1077
092f64e1 1078 case OP_LODS:
33c35af6 1079 case OP_STOS:
591721d7 1080 case OP_SCAS:
33c35af6 1081 if (op->operand_cnt != 0)
1082 break;
591721d7 1083 if (words[op_w][4] == 'b')
33c35af6 1084 lmod = OPLM_BYTE;
591721d7 1085 else if (words[op_w][4] == 'w')
33c35af6 1086 lmod = OPLM_WORD;
591721d7 1087 else if (words[op_w][4] == 'd')
33c35af6 1088 lmod = OPLM_DWORD;
1089 op->operand_cnt = 3;
092f64e1 1090 setup_reg_opr(&op->operand[0], op->op == OP_LODS ? xSI : xDI,
1091 lmod, &op->regmask_src);
d4e3b5db 1092 setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src);
1093 op->regmask_dst = op->regmask_src;
092f64e1 1094 setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD,
1095 op->op == OP_LODS ? &op->regmask_dst : &op->regmask_src);
33c35af6 1096 break;
1097
d4e3b5db 1098 case OP_MOVS:
7ba45c34 1099 case OP_CMPS:
d4e3b5db 1100 if (op->operand_cnt != 0)
1101 break;
7ba45c34 1102 if (words[op_w][4] == 'b')
d4e3b5db 1103 lmod = OPLM_BYTE;
7ba45c34 1104 else if (words[op_w][4] == 'w')
d4e3b5db 1105 lmod = OPLM_WORD;
7ba45c34 1106 else if (words[op_w][4] == 'd')
d4e3b5db 1107 lmod = OPLM_DWORD;
1108 op->operand_cnt = 3;
1109 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
1110 setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src);
1111 setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src);
1112 op->regmask_dst = op->regmask_src;
1113 break;
1114
5c024ef7 1115 case OP_JECXZ:
1116 op->operand_cnt = 1;
1117 op->regmask_src = 1 << xCX;
1118 op->operand[0].type = OPT_REG;
1119 op->operand[0].reg = xCX;
1120 op->operand[0].lmod = OPLM_DWORD;
1121 break;
1122
5101a5f9 1123 case OP_IMUL:
1124 if (op->operand_cnt != 1)
1125 break;
1126 // fallthrough
1127 case OP_MUL:
1128 // singleop mul
92d715b6 1129 op->regmask_src |= op->regmask_dst;
5101a5f9 1130 op->regmask_dst = (1 << xDX) | (1 << xAX);
5101a5f9 1131 if (op->operand[0].lmod == OPLM_UNSPEC)
1132 op->operand[0].lmod = OPLM_DWORD;
1133 break;
1134
1135 case OP_DIV:
1136 case OP_IDIV:
1137 // we could set up operands for edx:eax, but there is no real need to
1138 // (see is_opr_modified())
92d715b6 1139 op->regmask_src |= op->regmask_dst;
1140 op->regmask_dst = (1 << xDX) | (1 << xAX);
5101a5f9 1141 if (op->operand[0].lmod == OPLM_UNSPEC)
1142 op->operand[0].lmod = OPLM_DWORD;
1143 break;
1144
1145 case OP_SHL:
1146 case OP_SHR:
1147 case OP_SAR:
d4e3b5db 1148 case OP_ROL:
1149 case OP_ROR:
92d715b6 1150 op->regmask_src |= op->regmask_dst;
5101a5f9 1151 if (op->operand[1].lmod == OPLM_UNSPEC)
1152 op->operand[1].lmod = OPLM_BYTE;
1153 break;
1154
3b2f4044 1155 case OP_SHRD:
92d715b6 1156 op->regmask_src |= op->regmask_dst;
3b2f4044 1157 if (op->operand[2].lmod == OPLM_UNSPEC)
1158 op->operand[2].lmod = OPLM_BYTE;
1159 break;
1160
7ba45c34 1161 case OP_PUSH:
92d715b6 1162 op->regmask_src |= op->regmask_dst;
1163 op->regmask_dst = 0;
7ba45c34 1164 if (op->operand[0].lmod == OPLM_UNSPEC
1165 && (op->operand[0].type == OPT_CONST
1166 || op->operand[0].type == OPT_OFFSET
1167 || op->operand[0].type == OPT_LABEL))
1168 op->operand[0].lmod = OPLM_DWORD;
1169 break;
1170
3ebea2cf 1171 // alignment
1172 case OP_MOV:
092f64e1 1173 check_align:
3ebea2cf 1174 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
092f64e1 1175 && op->operand[0].lmod == op->operand[1].lmod
1176 && op->operand[0].reg == op->operand[1].reg
92d715b6 1177 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
3ebea2cf 1178 {
5e49b270 1179 op->flags |= OPF_RMD | OPF_DONE;
092f64e1 1180 op->regmask_src = op->regmask_dst = 0;
3ebea2cf 1181 }
1182 break;
1183
de50b98b 1184 case OP_LEA:
1185 if (op->operand[0].type == OPT_REG
1186 && op->operand[1].type == OPT_REGMEM)
1187 {
1188 char buf[16];
1189 snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name);
1190 if (IS(buf, op->operand[1].name))
5e49b270 1191 op->flags |= OPF_RMD | OPF_DONE;
de50b98b 1192 }
1193 break;
1194
89ff3147 1195 case OP_CALL:
1196 // trashed regs must be explicitly detected later
1197 op->regmask_dst = 0;
1198 break;
1199
92d715b6 1200 case OP_LEAVE:
1201 op->regmask_dst = (1 << xBP) | (1 << xSP);
1202 op->regmask_src = 1 << xBP;
1203 break;
1204
5101a5f9 1205 default:
1206 break;
1207 }
c36e914d 1208}
1209
092f64e1 1210static const char *op_name(struct parsed_op *po)
850c9265 1211{
092f64e1 1212 static char buf[16];
1213 char *p;
850c9265 1214 int i;
1215
092f64e1 1216 if (po->op == OP_JCC || po->op == OP_SCC) {
1217 p = buf;
1218 *p++ = (po->op == OP_JCC) ? 'j' : 's';
1219 if (po->pfo_inv)
1220 *p++ = 'n';
1221 strcpy(p, parsed_flag_op_names[po->pfo]);
1222 return buf;
1223 }
1224
850c9265 1225 for (i = 0; i < ARRAY_SIZE(op_table); i++)
092f64e1 1226 if (op_table[i].op == po->op)
850c9265 1227 return op_table[i].name;
1228
1229 return "???";
1230}
1231
1232// debug
1233static const char *dump_op(struct parsed_op *po)
1234{
1235 static char out[128];
1236 char *p = out;
1237 int i;
1238
4c45fa73 1239 if (po == NULL)
1240 return "???";
1241
092f64e1 1242 snprintf(out, sizeof(out), "%s", op_name(po));
850c9265 1243 for (i = 0; i < po->operand_cnt; i++) {
1244 p += strlen(p);
1245 if (i > 0)
1246 *p++ = ',';
1247 snprintf(p, sizeof(out) - (p - out),
1248 po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s",
1249 po->operand[i].name);
1250 }
1251
1252 return out;
1253}
1254
4c45fa73 1255static const char *lmod_type_u(struct parsed_op *po,
1256 enum opr_lenmod lmod)
1257{
1258 switch (lmod) {
90307a99 1259 case OPLM_QWORD:
1260 return "u64";
4c45fa73 1261 case OPLM_DWORD:
1262 return "u32";
1263 case OPLM_WORD:
1264 return "u16";
1265 case OPLM_BYTE:
1266 return "u8";
1267 default:
1268 ferr(po, "invalid lmod: %d\n", lmod);
1269 return "(_invalid_)";
1270 }
1271}
1272
d4e3b5db 1273static const char *lmod_cast_u(struct parsed_op *po,
1274 enum opr_lenmod lmod)
1275{
1276 switch (lmod) {
90307a99 1277 case OPLM_QWORD:
1278 return "";
d4e3b5db 1279 case OPLM_DWORD:
1280 return "";
1281 case OPLM_WORD:
1282 return "(u16)";
1283 case OPLM_BYTE:
1284 return "(u8)";
1285 default:
1286 ferr(po, "invalid lmod: %d\n", lmod);
1287 return "(_invalid_)";
1288 }
1289}
1290
1291static const char *lmod_cast_u_ptr(struct parsed_op *po,
1292 enum opr_lenmod lmod)
1293{
1294 switch (lmod) {
90307a99 1295 case OPLM_QWORD:
1296 return "*(u64 *)";
d4e3b5db 1297 case OPLM_DWORD:
1298 return "*(u32 *)";
1299 case OPLM_WORD:
1300 return "*(u16 *)";
1301 case OPLM_BYTE:
1302 return "*(u8 *)";
1303 default:
1304 ferr(po, "invalid lmod: %d\n", lmod);
1305 return "(_invalid_)";
1306 }
1307}
1308
1309static const char *lmod_cast_s(struct parsed_op *po,
1310 enum opr_lenmod lmod)
1311{
1312 switch (lmod) {
90307a99 1313 case OPLM_QWORD:
1314 return "(s64)";
d4e3b5db 1315 case OPLM_DWORD:
1316 return "(s32)";
1317 case OPLM_WORD:
1318 return "(s16)";
1319 case OPLM_BYTE:
1320 return "(s8)";
1321 default:
1322 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1323 return "(_invalid_)";
1324 }
1325}
1326
1327static const char *lmod_cast(struct parsed_op *po,
1328 enum opr_lenmod lmod, int is_signed)
1329{
1330 return is_signed ?
1331 lmod_cast_s(po, lmod) :
1332 lmod_cast_u(po, lmod);
1333}
1334
1335static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
1336{
1337 switch (lmod) {
90307a99 1338 case OPLM_QWORD:
1339 return 8;
d4e3b5db 1340 case OPLM_DWORD:
1341 return 4;
1342 case OPLM_WORD:
1343 return 2;
1344 case OPLM_BYTE:
1345 return 1;
1346 default:
1347 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1348 return 0;
1349 }
1350}
1351
91977a1c 1352static const char *opr_name(struct parsed_op *po, int opr_num)
c36e914d 1353{
91977a1c 1354 if (opr_num >= po->operand_cnt)
1355 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1356 return po->operand[opr_num].name;
c36e914d 1357}
1358
91977a1c 1359static unsigned int opr_const(struct parsed_op *po, int opr_num)
c36e914d 1360{
91977a1c 1361 if (opr_num >= po->operand_cnt)
1362 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1363 if (po->operand[opr_num].type != OPT_CONST)
1364 ferr(po, "opr %d: const expected\n", opr_num);
1365 return po->operand[opr_num].val;
1366}
c36e914d 1367
91977a1c 1368static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr)
1369{
90307a99 1370 if ((unsigned int)popr->reg >= ARRAY_SIZE(regs_r32))
91977a1c 1371 ferr(po, "invalid reg: %d\n", popr->reg);
1372 return regs_r32[popr->reg];
1373}
c36e914d 1374
a2c1d768 1375// cast1 is the "final" cast
1376static const char *simplify_cast(const char *cast1, const char *cast2)
1377{
1378 static char buf[256];
1379
1380 if (cast1[0] == 0)
1381 return cast2;
1382 if (cast2[0] == 0)
1383 return cast1;
1384 if (IS(cast1, cast2))
1385 return cast1;
1386 if (IS(cast1, "(s8)") && IS(cast2, "(u8)"))
1387 return cast1;
1388 if (IS(cast1, "(s16)") && IS(cast2, "(u16)"))
1389 return cast1;
1390 if (IS(cast1, "(u8)") && IS_START(cast2, "*(u8 *)"))
1391 return cast2;
1392 if (IS(cast1, "(u16)") && IS_START(cast2, "*(u16 *)"))
1393 return cast2;
1cd4a663 1394 if (strchr(cast1, '*') && IS_START(cast2, "(u32)"))
1395 return cast1;
a2c1d768 1396
1397 snprintf(buf, sizeof(buf), "%s%s", cast1, cast2);
1398 return buf;
1399}
1400
1401static const char *simplify_cast_num(const char *cast, unsigned int val)
1402{
1403 if (IS(cast, "(u8)") && val < 0x100)
1404 return "";
1405 if (IS(cast, "(s8)") && val < 0x80)
1406 return "";
1407 if (IS(cast, "(u16)") && val < 0x10000)
1408 return "";
1409 if (IS(cast, "(s16)") && val < 0x8000)
1410 return "";
1411 if (IS(cast, "(s32)") && val < 0x80000000)
1412 return "";
1413
1414 return cast;
1415}
1416
39b168b8 1417static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
1418 int *extra_offs)
91977a1c 1419{
39b168b8 1420 const char *p;
1421 char *endp;
1422 int namelen;
850c9265 1423 int i;
1424
39b168b8 1425 *extra_offs = 0;
1426 namelen = strlen(name);
1427
1428 p = strchr(name, '+');
1429 if (p != NULL) {
1430 namelen = p - name;
1431 if (namelen <= 0)
1432 ferr(po, "equ parse failed for '%s'\n", name);
1433
1434 if (IS_START(p, "0x"))
1435 p += 2;
1436 *extra_offs = strtol(p, &endp, 16);
1437 if (*endp != 0)
1438 ferr(po, "equ parse failed for '%s'\n", name);
1439 }
1440
850c9265 1441 for (i = 0; i < g_eqcnt; i++)
39b168b8 1442 if (strncmp(g_eqs[i].name, name, namelen) == 0
1443 && g_eqs[i].name[namelen] == 0)
850c9265 1444 break;
87bf6cec 1445 if (i >= g_eqcnt) {
1446 if (po != NULL)
1447 ferr(po, "unresolved equ name: '%s'\n", name);
1448 return NULL;
1449 }
850c9265 1450
1451 return &g_eqs[i];
1452}
1453
1cd4a663 1454static int is_stack_access(struct parsed_op *po,
1455 const struct parsed_opr *popr)
850c9265 1456{
09bc6fd5 1457 return (parse_stack_el(popr->name, NULL, 0)
1cd4a663 1458 || (g_bp_frame && !(po->flags & OPF_EBP_S)
1459 && IS_START(popr->name, "ebp")));
1460}
1461
1462static void parse_stack_access(struct parsed_op *po,
1463 const char *name, char *ofs_reg, int *offset_out,
037f4971 1464 int *stack_ra_out, const char **bp_arg_out, int is_lea)
1cd4a663 1465{
1466 const char *bp_arg = "";
1467 const char *p = NULL;
91977a1c 1468 struct parsed_equ *eq;
4f12f671 1469 char *endp = NULL;
d4e3b5db 1470 int stack_ra = 0;
39b168b8 1471 int offset = 0;
91977a1c 1472
1cd4a663 1473 ofs_reg[0] = 0;
a3684be1 1474
1475 if (IS_START(name, "ebp-")
1476 || (IS_START(name, "ebp+") && '0' <= name[4] && name[4] <= '9'))
1477 {
1478 p = name + 4;
1479 if (IS_START(p, "0x"))
1480 p += 2;
1481 offset = strtoul(p, &endp, 16);
1482 if (name[3] == '-')
1483 offset = -offset;
1484 if (*endp != 0)
1485 ferr(po, "ebp- parse of '%s' failed\n", name);
1486 }
1487 else {
09bc6fd5 1488 bp_arg = parse_stack_el(name, ofs_reg, 0);
4f12f671 1489 snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
1490 eq = equ_find(po, bp_arg, &offset);
1491 if (eq == NULL)
1492 ferr(po, "detected but missing eq\n");
1493 offset += eq->offset;
1494 }
39b168b8 1495
d4e3b5db 1496 if (!strncmp(name, "ebp", 3))
1497 stack_ra = 4;
1498
037f4971 1499 // yes it sometimes LEAs ra for compares..
1500 if (!is_lea && ofs_reg[0] == 0
1501 && stack_ra <= offset && offset < stack_ra + 4)
1502 {
39b168b8 1503 ferr(po, "reference to ra? %d %d\n", offset, stack_ra);
037f4971 1504 }
d4e3b5db 1505
1cd4a663 1506 *offset_out = offset;
1507 *stack_ra_out = stack_ra;
1508 if (bp_arg_out)
1509 *bp_arg_out = bp_arg;
1510}
1511
5f70a34f 1512static int stack_frame_access(struct parsed_op *po,
1cd4a663 1513 struct parsed_opr *popr, char *buf, size_t buf_size,
1514 const char *name, const char *cast, int is_src, int is_lea)
1515{
1516 enum opr_lenmod tmp_lmod = OPLM_UNSPEC;
1517 const char *prefix = "";
1518 const char *bp_arg = NULL;
1519 char ofs_reg[16] = { 0, };
1520 int i, arg_i, arg_s;
1521 int unaligned = 0;
1522 int stack_ra = 0;
1523 int offset = 0;
5f70a34f 1524 int retval = -1;
1cd4a663 1525 int sf_ofs;
1526 int lim;
1527
1528 if (po->flags & OPF_EBP_S)
1529 ferr(po, "stack_frame_access while ebp is scratch\n");
1530
037f4971 1531 parse_stack_access(po, name, ofs_reg, &offset,
1532 &stack_ra, &bp_arg, is_lea);
1cd4a663 1533
4f12f671 1534 if (offset > stack_ra)
1535 {
39b168b8 1536 arg_i = (offset - stack_ra - 4) / 4;
bd96f656 1537 if (arg_i < 0 || arg_i >= g_func_pp->argc_stack)
4f12f671 1538 {
bd96f656 1539 if (g_func_pp->is_vararg
1540 && arg_i == g_func_pp->argc_stack && is_lea)
1541 {
4f12f671 1542 // should be va_list
1543 if (cast[0] == 0)
1544 cast = "(u32)";
1545 snprintf(buf, buf_size, "%sap", cast);
5f70a34f 1546 return -1;
4f12f671 1547 }
1548 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1549 offset, bp_arg, arg_i);
1550 }
4c45fa73 1551 if (ofs_reg[0] != 0)
7ba45c34 1552 ferr(po, "offset reg on arg access?\n");
91977a1c 1553
bd96f656 1554 for (i = arg_s = 0; i < g_func_pp->argc; i++) {
1555 if (g_func_pp->arg[i].reg != NULL)
91977a1c 1556 continue;
1557 if (arg_s == arg_i)
1558 break;
1559 arg_s++;
1560 }
bd96f656 1561 if (i == g_func_pp->argc)
91977a1c 1562 ferr(po, "arg %d not in prototype?\n", arg_i);
850c9265 1563
bd96f656 1564 popr->is_ptr = g_func_pp->arg[i].type.is_ptr;
5f70a34f 1565 retval = i;
de50b98b 1566
1567 switch (popr->lmod)
7ba45c34 1568 {
1569 case OPLM_BYTE:
4f12f671 1570 if (is_lea)
1571 ferr(po, "lea/byte to arg?\n");
7ba45c34 1572 if (is_src && (offset & 3) == 0)
a2c1d768 1573 snprintf(buf, buf_size, "%sa%d",
1574 simplify_cast(cast, "(u8)"), i + 1);
7ba45c34 1575 else
a2c1d768 1576 snprintf(buf, buf_size, "%sBYTE%d(a%d)",
1577 cast, offset & 3, i + 1);
7ba45c34 1578 break;
1579
1580 case OPLM_WORD:
4f12f671 1581 if (is_lea)
1582 ferr(po, "lea/word to arg?\n");
a3684be1 1583 if (offset & 1) {
1584 unaligned = 1;
1585 if (!is_src) {
1586 if (offset & 2)
1587 ferr(po, "problematic arg store\n");
a2c1d768 1588 snprintf(buf, buf_size, "%s((char *)&a%d + 1)",
1589 simplify_cast(cast, "*(u16 *)"), i + 1);
a3684be1 1590 }
1591 else
1592 ferr(po, "unaligned arg word load\n");
1593 }
1594 else if (is_src && (offset & 2) == 0)
a2c1d768 1595 snprintf(buf, buf_size, "%sa%d",
1596 simplify_cast(cast, "(u16)"), i + 1);
7ba45c34 1597 else
a2c1d768 1598 snprintf(buf, buf_size, "%s%sWORD(a%d)",
1599 cast, (offset & 2) ? "HI" : "LO", i + 1);
7ba45c34 1600 break;
1601
1602 case OPLM_DWORD:
3ebea2cf 1603 if (cast[0])
1604 prefix = cast;
1605 else if (is_src)
1606 prefix = "(u32)";
a3684be1 1607
de50b98b 1608 if (offset & 3) {
a3684be1 1609 unaligned = 1;
2b43685d 1610 if (is_lea)
1611 snprintf(buf, buf_size, "(u32)&a%d + %d",
1612 i + 1, offset & 3);
a3684be1 1613 else if (!is_src)
1614 ferr(po, "unaligned arg store\n");
1615 else {
1616 // mov edx, [ebp+arg_4+2]; movsx ecx, dx
2b43685d 1617 snprintf(buf, buf_size, "%s(a%d >> %d)",
1618 prefix, i + 1, (offset & 3) * 8);
a3684be1 1619 }
de50b98b 1620 }
1621 else {
1622 snprintf(buf, buf_size, "%s%sa%d",
1623 prefix, is_lea ? "&" : "", i + 1);
1624 }
7ba45c34 1625 break;
1626
1627 default:
de50b98b 1628 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
7ba45c34 1629 }
1630
a3684be1 1631 if (unaligned)
1632 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1633
7ba45c34 1634 // common problem
bd96f656 1635 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type);
a2c1d768 1636 if (tmp_lmod != OPLM_DWORD
de041e5b 1637 && (unaligned || (!is_src && lmod_bytes(po, tmp_lmod)
1638 < lmod_bytes(po, popr->lmod) + (offset & 3))))
a2c1d768 1639 {
1640 ferr(po, "bp_arg arg%d/w offset %d and type '%s' is too small\n",
1641 i + 1, offset, g_func_pp->arg[i].type.name);
1642 }
4741fdfe 1643 // can't check this because msvc likes to reuse
1644 // arg space for scratch..
1645 //if (popr->is_ptr && popr->lmod != OPLM_DWORD)
1646 // ferr(po, "bp_arg arg%d: non-dword ptr access\n", i + 1);
91977a1c 1647 }
4f12f671 1648 else
1649 {
1bafb621 1650 if (g_stack_fsz == 0)
1651 ferr(po, "stack var access without stackframe\n");
a2c1d768 1652 g_stack_frame_used = 1;
850c9265 1653
39b168b8 1654 sf_ofs = g_stack_fsz + offset;
de50b98b 1655 lim = (ofs_reg[0] != 0) ? -4 : 0;
1656 if (offset > 0 || sf_ofs < lim)
39b168b8 1657 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
850c9265 1658
1659 if (is_lea)
33c35af6 1660 prefix = "(u32)&";
3ebea2cf 1661 else
1662 prefix = cast;
850c9265 1663
de50b98b 1664 switch (popr->lmod)
850c9265 1665 {
1666 case OPLM_BYTE:
7ba45c34 1667 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1668 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
850c9265 1669 break;
7ba45c34 1670
850c9265 1671 case OPLM_WORD:
7ba45c34 1672 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1673 // known unaligned or possibly unaligned
1674 strcat(g_comment, " unaligned");
1675 if (prefix[0] == 0)
1676 prefix = "*(u16 *)&";
1677 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1678 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1679 break;
1680 }
1681 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
850c9265 1682 break;
7ba45c34 1683
850c9265 1684 case OPLM_DWORD:
7ba45c34 1685 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1686 // known unaligned or possibly unaligned
1687 strcat(g_comment, " unaligned");
1688 if (prefix[0] == 0)
1689 prefix = "*(u32 *)&";
1690 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1691 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1692 break;
1693 }
1694 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
850c9265 1695 break;
7ba45c34 1696
850c9265 1697 default:
de50b98b 1698 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
850c9265 1699 }
91977a1c 1700 }
5f70a34f 1701
1702 return retval;
91977a1c 1703}
c36e914d 1704
89ff3147 1705static void check_func_pp(struct parsed_op *po,
1706 const struct parsed_proto *pp, const char *pfx)
1707{
179b79a9 1708 enum opr_lenmod tmp_lmod;
b74c31e3 1709 char buf[256];
179b79a9 1710 int ret, i;
b74c31e3 1711
89ff3147 1712 if (pp->argc_reg != 0) {
b74c31e3 1713 if (/*!g_allow_regfunc &&*/ !pp->is_fastcall) {
1714 pp_print(buf, sizeof(buf), pp);
1715 ferr(po, "%s: unexpected reg arg in icall: %s\n", pfx, buf);
1716 }
89ff3147 1717 if (pp->argc_stack > 0 && pp->argc_reg != 2)
1718 ferr(po, "%s: %d reg arg(s) with %d stack arg(s)\n",
1719 pfx, pp->argc_reg, pp->argc_stack);
1720 }
179b79a9 1721
1722 // fptrs must use 32bit args, callsite might have no information and
1723 // lack a cast to smaller types, which results in incorrectly masked
1724 // args passed (callee may assume masked args, it does on ARM)
61e29183 1725 if (!pp->is_osinc) {
179b79a9 1726 for (i = 0; i < pp->argc; i++) {
1727 ret = guess_lmod_from_c_type(&tmp_lmod, &pp->arg[i].type);
1728 if (ret && tmp_lmod != OPLM_DWORD)
1729 ferr(po, "reference to %s with arg%d '%s'\n", pp->name,
1730 i + 1, pp->arg[i].type.name);
1731 }
1732 }
89ff3147 1733}
1734
7aca4698 1735static const char *check_label_read_ref(struct parsed_op *po,
1736 const char *name)
3ebea2cf 1737{
840257f6 1738 const struct parsed_proto *pp;
1739
36595fd2 1740 pp = proto_parse(g_fhdr, name, 0);
840257f6 1741 if (pp == NULL)
1742 ferr(po, "proto_parse failed for ref '%s'\n", name);
1743
89ff3147 1744 if (pp->is_func)
1745 check_func_pp(po, pp, "ref");
7aca4698 1746
1747 return pp->name;
3ebea2cf 1748}
1749
91977a1c 1750static char *out_src_opr(char *buf, size_t buf_size,
591721d7 1751 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
3ebea2cf 1752 int is_lea)
91977a1c 1753{
850c9265 1754 char tmp1[256], tmp2[256];
1755 char expr[256];
7aca4698 1756 const char *name;
a2c1d768 1757 char *p;
850c9265 1758 int ret;
1759
3ebea2cf 1760 if (cast == NULL)
1761 cast = "";
1762
91977a1c 1763 switch (popr->type) {
1764 case OPT_REG:
850c9265 1765 if (is_lea)
1766 ferr(po, "lea from reg?\n");
1767
91977a1c 1768 switch (popr->lmod) {
90307a99 1769 case OPLM_QWORD:
1770 snprintf(buf, buf_size, "%s%s.q", cast, opr_reg_p(po, popr));
1771 break;
91977a1c 1772 case OPLM_DWORD:
3ebea2cf 1773 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
91977a1c 1774 break;
850c9265 1775 case OPLM_WORD:
a2c1d768 1776 snprintf(buf, buf_size, "%s%s",
1777 simplify_cast(cast, "(u16)"), opr_reg_p(po, popr));
850c9265 1778 break;
1779 case OPLM_BYTE:
5101a5f9 1780 if (popr->name[1] == 'h') // XXX..
a2c1d768 1781 snprintf(buf, buf_size, "%s(%s >> 8)",
1782 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
5101a5f9 1783 else
a2c1d768 1784 snprintf(buf, buf_size, "%s%s",
1785 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
850c9265 1786 break;
91977a1c 1787 default:
1788 ferr(po, "invalid src lmod: %d\n", popr->lmod);
1789 }
1790 break;
850c9265 1791
91977a1c 1792 case OPT_REGMEM:
1cd4a663 1793 if (is_stack_access(po, popr)) {
de50b98b 1794 stack_frame_access(po, popr, buf, buf_size,
3ebea2cf 1795 popr->name, cast, 1, is_lea);
91977a1c 1796 break;
1797 }
850c9265 1798
1799 strcpy(expr, popr->name);
1800 if (strchr(expr, '[')) {
1801 // special case: '[' can only be left for label[reg] form
1802 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
1803 if (ret != 2)
1804 ferr(po, "parse failure for '%s'\n", expr);
a2c1d768 1805 if (tmp1[0] == '(') {
1806 // (off_4FFF50+3)[eax]
1807 p = strchr(tmp1 + 1, ')');
1808 if (p == NULL || p[1] != 0)
1809 ferr(po, "parse failure (2) for '%s'\n", expr);
1810 *p = 0;
1811 memmove(tmp1, tmp1 + 1, strlen(tmp1));
1812 }
33c35af6 1813 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
850c9265 1814 }
1815
1816 // XXX: do we need more parsing?
1817 if (is_lea) {
1818 snprintf(buf, buf_size, "%s", expr);
1819 break;
1820 }
1821
a2c1d768 1822 snprintf(buf, buf_size, "%s(%s)",
1823 simplify_cast(cast, lmod_cast_u_ptr(po, popr->lmod)), expr);
91977a1c 1824 break;
850c9265 1825
91977a1c 1826 case OPT_LABEL:
7aca4698 1827 name = check_label_read_ref(po, popr->name);
3ebea2cf 1828 if (cast[0] == 0 && popr->is_ptr)
1829 cast = "(u32)";
2b43685d 1830
850c9265 1831 if (is_lea)
7aca4698 1832 snprintf(buf, buf_size, "(u32)&%s", name);
2b43685d 1833 else if (popr->size_lt)
1834 snprintf(buf, buf_size, "%s%s%s%s", cast,
1835 lmod_cast_u_ptr(po, popr->lmod),
7aca4698 1836 popr->is_array ? "" : "&", name);
850c9265 1837 else
7aca4698 1838 snprintf(buf, buf_size, "%s%s%s", cast, name,
7ba45c34 1839 popr->is_array ? "[0]" : "");
850c9265 1840 break;
1841
1842 case OPT_OFFSET:
7aca4698 1843 name = check_label_read_ref(po, popr->name);
3ebea2cf 1844 if (cast[0] == 0)
1845 cast = "(u32)";
850c9265 1846 if (is_lea)
1847 ferr(po, "lea an offset?\n");
7aca4698 1848 snprintf(buf, buf_size, "%s&%s", cast, name);
91977a1c 1849 break;
850c9265 1850
91977a1c 1851 case OPT_CONST:
850c9265 1852 if (is_lea)
1853 ferr(po, "lea from const?\n");
1854
a2c1d768 1855 printf_number(tmp1, sizeof(tmp1), popr->val);
ddaf8bd7 1856 if (popr->val == 0 && strchr(cast, '*'))
1857 snprintf(buf, buf_size, "NULL");
1858 else
1859 snprintf(buf, buf_size, "%s%s",
1860 simplify_cast_num(cast, popr->val), tmp1);
91977a1c 1861 break;
850c9265 1862
91977a1c 1863 default:
1864 ferr(po, "invalid src type: %d\n", popr->type);
1865 }
1866
1867 return buf;
1868}
c36e914d 1869
de50b98b 1870// note: may set is_ptr (we find that out late for ebp frame..)
91977a1c 1871static char *out_dst_opr(char *buf, size_t buf_size,
1872 struct parsed_op *po, struct parsed_opr *popr)
1873{
1874 switch (popr->type) {
1875 case OPT_REG:
1876 switch (popr->lmod) {
90307a99 1877 case OPLM_QWORD:
1878 snprintf(buf, buf_size, "%s.q", opr_reg_p(po, popr));
1879 break;
91977a1c 1880 case OPLM_DWORD:
1881 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
1882 break;
850c9265 1883 case OPLM_WORD:
1884 // ugh..
1885 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
1886 break;
1887 case OPLM_BYTE:
1888 // ugh..
5101a5f9 1889 if (popr->name[1] == 'h') // XXX..
1890 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
1891 else
1892 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
850c9265 1893 break;
91977a1c 1894 default:
1895 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
1896 }
1897 break;
850c9265 1898
1899 case OPT_REGMEM:
1cd4a663 1900 if (is_stack_access(po, popr)) {
de50b98b 1901 stack_frame_access(po, popr, buf, buf_size,
3ebea2cf 1902 popr->name, "", 0, 0);
850c9265 1903 break;
1904 }
1905
3ebea2cf 1906 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
850c9265 1907
bfa4a6ee 1908 case OPT_LABEL:
2b43685d 1909 if (popr->size_mismatch)
1910 snprintf(buf, buf_size, "%s%s%s",
1911 lmod_cast_u_ptr(po, popr->lmod),
1912 popr->is_array ? "" : "&", popr->name);
1913 else
1914 snprintf(buf, buf_size, "%s%s", popr->name,
1915 popr->is_array ? "[0]" : "");
bfa4a6ee 1916 break;
1917
91977a1c 1918 default:
1919 ferr(po, "invalid dst type: %d\n", popr->type);
1920 }
1921
1922 return buf;
1923}
c36e914d 1924
3ebea2cf 1925static char *out_src_opr_u32(char *buf, size_t buf_size,
1926 struct parsed_op *po, struct parsed_opr *popr)
1927{
1928 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1929}
1930
91977a1c 1931static void out_test_for_cc(char *buf, size_t buf_size,
940e8e66 1932 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
69a3cdfc 1933 enum opr_lenmod lmod, const char *expr)
91977a1c 1934{
69a3cdfc 1935 const char *cast, *scast;
91977a1c 1936
69a3cdfc 1937 cast = lmod_cast_u(po, lmod);
1938 scast = lmod_cast_s(po, lmod);
1939
1940 switch (pfo) {
1941 case PFO_Z:
87bf6cec 1942 case PFO_BE: // CF=1||ZF=1; CF=0
850c9265 1943 snprintf(buf, buf_size, "(%s%s %s 0)",
940e8e66 1944 cast, expr, is_inv ? "!=" : "==");
91977a1c 1945 break;
850c9265 1946
5101a5f9 1947 case PFO_S:
1948 case PFO_L: // SF!=OF; OF=0
1949 snprintf(buf, buf_size, "(%s%s %s 0)",
940e8e66 1950 scast, expr, is_inv ? ">=" : "<");
5101a5f9 1951 break;
1952
87bf6cec 1953 case PFO_LE: // ZF=1||SF!=OF; OF=0
69a3cdfc 1954 snprintf(buf, buf_size, "(%s%s %s 0)",
940e8e66 1955 scast, expr, is_inv ? ">" : "<=");
850c9265 1956 break;
1957
91977a1c 1958 default:
69a3cdfc 1959 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
91977a1c 1960 }
1961}
c36e914d 1962
850c9265 1963static void out_cmp_for_cc(char *buf, size_t buf_size,
a2c1d768 1964 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
850c9265 1965{
a2c1d768 1966 const char *cast, *scast, *cast_use;
1967 char buf1[256], buf2[256];
1968 enum opr_lenmod lmod;
1969
90307a99 1970 if (po->op != OP_DEC && po->operand[0].lmod != po->operand[1].lmod)
a2c1d768 1971 ferr(po, "%s: lmod mismatch: %d %d\n", __func__,
1972 po->operand[0].lmod, po->operand[1].lmod);
1973 lmod = po->operand[0].lmod;
850c9265 1974
69a3cdfc 1975 cast = lmod_cast_u(po, lmod);
1976 scast = lmod_cast_s(po, lmod);
850c9265 1977
a2c1d768 1978 switch (pfo) {
1979 case PFO_C:
1980 case PFO_Z:
1981 case PFO_BE: // !a
1982 cast_use = cast;
1983 break;
1984
1985 case PFO_S:
1986 case PFO_L: // !ge
1987 case PFO_LE:
1988 cast_use = scast;
1989 break;
1990
1991 default:
1992 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1993 }
1994
1995 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], cast_use, 0);
90307a99 1996 if (po->op == OP_DEC)
1997 snprintf(buf2, sizeof(buf2), "1");
1998 else
1999 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], cast_use, 0);
a2c1d768 2000
69a3cdfc 2001 switch (pfo) {
5101a5f9 2002 case PFO_C:
2003 // note: must be unsigned compare
a2c1d768 2004 snprintf(buf, buf_size, "(%s %s %s)",
2005 buf1, is_inv ? ">=" : "<", buf2);
5101a5f9 2006 break;
2007
69a3cdfc 2008 case PFO_Z:
a2c1d768 2009 snprintf(buf, buf_size, "(%s %s %s)",
2010 buf1, is_inv ? "!=" : "==", buf2);
850c9265 2011 break;
2012
5101a5f9 2013 case PFO_BE: // !a
850c9265 2014 // note: must be unsigned compare
a2c1d768 2015 snprintf(buf, buf_size, "(%s %s %s)",
2016 buf1, is_inv ? ">" : "<=", buf2);
2017
2018 // annoying case
2019 if (is_inv && lmod == OPLM_BYTE
2020 && po->operand[1].type == OPT_CONST
2021 && po->operand[1].val == 0xff)
2022 {
2023 snprintf(g_comment, sizeof(g_comment), "if %s", buf);
2024 snprintf(buf, buf_size, "(0)");
2025 }
5101a5f9 2026 break;
2027
2028 // note: must be signed compare
2029 case PFO_S:
2030 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
a2c1d768 2031 scast, buf1, buf2, is_inv ? ">=" : "<");
850c9265 2032 break;
2033
5101a5f9 2034 case PFO_L: // !ge
a2c1d768 2035 snprintf(buf, buf_size, "(%s %s %s)",
2036 buf1, is_inv ? ">=" : "<", buf2);
850c9265 2037 break;
2038
90307a99 2039 case PFO_LE: // !g
a2c1d768 2040 snprintf(buf, buf_size, "(%s %s %s)",
2041 buf1, is_inv ? ">" : "<=", buf2);
5101a5f9 2042 break;
2043
850c9265 2044 default:
a2c1d768 2045 break;
850c9265 2046 }
2047}
2048
69a3cdfc 2049static void out_cmp_test(char *buf, size_t buf_size,
940e8e66 2050 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
69a3cdfc 2051{
2052 char buf1[256], buf2[256], buf3[256];
2053
2054 if (po->op == OP_TEST) {
2055 if (IS(opr_name(po, 0), opr_name(po, 1))) {
3ebea2cf 2056 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
69a3cdfc 2057 }
2058 else {
3ebea2cf 2059 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
2060 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
69a3cdfc 2061 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
2062 }
940e8e66 2063 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
69a3cdfc 2064 po->operand[0].lmod, buf3);
2065 }
2066 else if (po->op == OP_CMP) {
a2c1d768 2067 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv);
69a3cdfc 2068 }
2069 else
2070 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
2071}
2072
850c9265 2073static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
91977a1c 2074 struct parsed_opr *popr2)
2075{
2076 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
2077 ferr(po, "missing lmod for both operands\n");
2078
2079 if (popr1->lmod == OPLM_UNSPEC)
2080 popr1->lmod = popr2->lmod;
2081 else if (popr2->lmod == OPLM_UNSPEC)
2082 popr2->lmod = popr1->lmod;
a3684be1 2083 else if (popr1->lmod != popr2->lmod) {
2084 if (popr1->type_from_var) {
2085 popr1->size_mismatch = 1;
2086 if (popr1->lmod < popr2->lmod)
2087 popr1->size_lt = 1;
2088 popr1->lmod = popr2->lmod;
2089 }
2090 else if (popr2->type_from_var) {
2091 popr2->size_mismatch = 1;
2092 if (popr2->lmod < popr1->lmod)
2093 popr2->size_lt = 1;
2094 popr2->lmod = popr1->lmod;
2095 }
2096 else
2097 ferr(po, "conflicting lmods: %d vs %d\n",
2098 popr1->lmod, popr2->lmod);
2099 }
91977a1c 2100}
c36e914d 2101
850c9265 2102static const char *op_to_c(struct parsed_op *po)
2103{
2104 switch (po->op)
2105 {
2106 case OP_ADD:
5101a5f9 2107 case OP_ADC:
850c9265 2108 return "+";
2109 case OP_SUB:
5101a5f9 2110 case OP_SBB:
850c9265 2111 return "-";
2112 case OP_AND:
2113 return "&";
2114 case OP_OR:
2115 return "|";
2116 case OP_XOR:
2117 return "^";
2118 case OP_SHL:
2119 return "<<";
2120 case OP_SHR:
2121 return ">>";
2122 case OP_MUL:
2123 case OP_IMUL:
2124 return "*";
2125 default:
2126 ferr(po, "op_to_c was supplied with %d\n", po->op);
2127 }
2128}
2129
7ae48d73 2130static void op_set_clear_flag(struct parsed_op *po,
2131 enum op_flags flag_set, enum op_flags flag_clear)
d4e3b5db 2132{
7ae48d73 2133 po->flags |= flag_set;
2134 po->flags &= ~flag_clear;
d4e3b5db 2135}
2136
de50b98b 2137// last op in stream - unconditional branch or ret
2138#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
092f64e1 2139 || ((ops[_i].flags & (OPF_JMP|OPF_CJMP|OPF_RMD)) == OPF_JMP \
037f4971 2140 && ops[_i].op != OP_CALL))
de50b98b 2141
87bf6cec 2142static int scan_for_pop(int i, int opcnt, const char *reg,
d4e3b5db 2143 int magic, int depth, int *maxdepth, int do_flags)
850c9265 2144{
89ff3147 2145 const struct parsed_proto *pp;
87bf6cec 2146 struct parsed_op *po;
2147 int ret = 0;
4c45fa73 2148 int j;
87bf6cec 2149
850c9265 2150 for (; i < opcnt; i++) {
87bf6cec 2151 po = &ops[i];
2152 if (po->cc_scratch == magic)
2153 break; // already checked
2154 po->cc_scratch = magic;
2155
89ff3147 2156 if (po->flags & OPF_TAIL) {
2157 if (po->op == OP_CALL) {
92d715b6 2158 pp = proto_parse(g_fhdr, po->operand[0].name, g_quiet_pp);
89ff3147 2159 if (pp != NULL && pp->is_noreturn)
2160 // no stack cleanup for noreturn
2161 return ret;
2162 }
87bf6cec 2163 return -1; // deadend
89ff3147 2164 }
87bf6cec 2165
5e49b270 2166 if ((po->flags & (OPF_RMD|OPF_DONE))
5f70a34f 2167 || (po->op == OP_PUSH && po->p_argnum != 0)) // arg push
850c9265 2168 continue;
2169
87bf6cec 2170 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
4c45fa73 2171 if (po->btj != NULL) {
2172 // jumptable
da87ae38 2173 for (j = 0; j < po->btj->count; j++) {
4c45fa73 2174 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic,
2175 depth, maxdepth, do_flags);
2176 if (ret < 0)
2177 return ret; // dead end
2178 }
da87ae38 2179 return ret;
4c45fa73 2180 }
2181
87bf6cec 2182 if (po->bt_i < 0) {
2183 ferr(po, "dead branch\n");
2184 return -1;
2185 }
850c9265 2186
5c024ef7 2187 if (po->flags & OPF_CJMP) {
d4e3b5db 2188 ret |= scan_for_pop(po->bt_i, opcnt, reg, magic,
2189 depth, maxdepth, do_flags);
87bf6cec 2190 if (ret < 0)
2191 return ret; // dead end
2192 }
2193 else {
2194 i = po->bt_i - 1;
2195 }
2196 continue;
2197 }
2198
d4e3b5db 2199 if ((po->op == OP_POP || po->op == OP_PUSH)
2200 && po->operand[0].type == OPT_REG
87bf6cec 2201 && IS(po->operand[0].name, reg))
2202 {
9af2d373 2203 if (po->op == OP_PUSH && !(po->flags & OPF_FARGNR)) {
d4e3b5db 2204 depth++;
2205 if (depth > *maxdepth)
2206 *maxdepth = depth;
2207 if (do_flags)
7ae48d73 2208 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
d4e3b5db 2209 }
da87ae38 2210 else if (po->op == OP_POP) {
2211 if (depth == 0) {
2212 if (do_flags)
2213 op_set_clear_flag(po, OPF_RMD, OPF_RSAVE);
2214 return 1;
2215 }
2216 else {
2217 depth--;
2218 if (depth < 0) // should not happen
2219 ferr(po, "fail with depth\n");
2220 if (do_flags)
2221 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2222 }
d4e3b5db 2223 }
87bf6cec 2224 }
850c9265 2225 }
2226
87bf6cec 2227 return ret;
850c9265 2228}
2229
69a3cdfc 2230// scan for pop starting from 'ret' op (all paths)
d4e3b5db 2231static int scan_for_pop_ret(int i, int opcnt, const char *reg,
2232 int flag_set)
850c9265 2233{
2234 int found = 0;
2235 int j;
2236
2237 for (; i < opcnt; i++) {
87bf6cec 2238 if (!(ops[i].flags & OPF_TAIL))
850c9265 2239 continue;
2240
2241 for (j = i - 1; j >= 0; j--) {
5e49b270 2242 if (ops[j].flags & (OPF_RMD|OPF_DONE))
69a3cdfc 2243 continue;
2244 if (ops[j].flags & OPF_JMP)
850c9265 2245 return -1;
2246
5e49b270 2247 if (ops[j].op == OP_POP && ops[j].datap == NULL
2248 && ops[j].operand[0].type == OPT_REG
850c9265 2249 && IS(ops[j].operand[0].name, reg))
2250 {
2251 found = 1;
d4e3b5db 2252 ops[j].flags |= flag_set;
850c9265 2253 break;
2254 }
2255
d7857c3a 2256 if (g_labels[j] != NULL)
850c9265 2257 return -1;
2258 }
2259 }
2260
2261 return found ? 0 : -1;
2262}
2263
9af2d373 2264static void scan_for_pop_const(int i, int opcnt)
2265{
2266 int j;
2267
2268 for (j = i + 1; j < opcnt; j++) {
2269 if ((ops[j].flags & (OPF_JMP|OPF_TAIL|OPF_RSAVE))
2270 || ops[j].op == OP_PUSH || g_labels[i] != NULL)
2271 {
2272 break;
2273 }
2274
5e49b270 2275 if (ops[j].op == OP_POP && !(ops[j].flags & (OPF_RMD|OPF_DONE)))
9af2d373 2276 {
5e49b270 2277 ops[i].flags |= OPF_RMD | OPF_DONE;
2278 ops[j].flags |= OPF_DONE;
9af2d373 2279 ops[j].datap = &ops[i];
2280 break;
2281 }
2282 }
2283}
2284
591721d7 2285static void scan_propagate_df(int i, int opcnt)
2286{
2287 struct parsed_op *po = &ops[i];
2288 int j;
2289
2290 for (; i < opcnt; i++) {
2291 po = &ops[i];
2292 if (po->flags & OPF_DF)
2293 return; // already resolved
2294 po->flags |= OPF_DF;
2295
2296 if (po->op == OP_CALL)
2297 ferr(po, "call with DF set?\n");
2298
2299 if (po->flags & OPF_JMP) {
2300 if (po->btj != NULL) {
2301 // jumptable
2302 for (j = 0; j < po->btj->count; j++)
2303 scan_propagate_df(po->btj->d[j].bt_i, opcnt);
2304 return;
2305 }
2306
2307 if (po->bt_i < 0) {
2308 ferr(po, "dead branch\n");
2309 return;
2310 }
2311
5c024ef7 2312 if (po->flags & OPF_CJMP)
591721d7 2313 scan_propagate_df(po->bt_i, opcnt);
2314 else
2315 i = po->bt_i - 1;
2316 continue;
2317 }
2318
2319 if (po->flags & OPF_TAIL)
2320 break;
2321
2322 if (po->op == OP_CLD) {
5e49b270 2323 po->flags |= OPF_RMD | OPF_DONE;
591721d7 2324 return;
2325 }
2326 }
2327
2328 ferr(po, "missing DF clear?\n");
2329}
2330
1cd4a663 2331// is operand 'opr' modified by parsed_op 'po'?
5101a5f9 2332static int is_opr_modified(const struct parsed_opr *opr,
69a3cdfc 2333 const struct parsed_op *po)
2334{
89ff3147 2335 int mask;
2336
69a3cdfc 2337 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2338 return 0;
2339
89ff3147 2340 if (opr->type == OPT_REG) {
2341 if (po->op == OP_CALL) {
2342 mask = (1 << xAX) | (1 << xCX) | (1 << xDX);
2343 if ((1 << opr->reg) & mask)
2344 return 1;
2345 else
2346 return 0;
2347 }
2348
2349 if (po->operand[0].type == OPT_REG) {
2350 if (po->regmask_dst & (1 << opr->reg))
2351 return 1;
2352 else
2353 return 0;
2354 }
69a3cdfc 2355 }
2356
2357 return IS(po->operand[0].name, opr->name);
2358}
2359
5101a5f9 2360// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
2361static int is_any_opr_modified(const struct parsed_op *po_test,
89ff3147 2362 const struct parsed_op *po, int c_mode)
5101a5f9 2363{
89ff3147 2364 int mask;
5101a5f9 2365 int i;
2366
2367 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2368 return 0;
2369
de50b98b 2370 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2371 return 0;
2372
2373 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
2374 return 1;
2375
2376 // in reality, it can wreck any register, but in decompiled C
2b43685d 2377 // version it can only overwrite eax or edx:eax
89ff3147 2378 mask = (1 << xAX) | (1 << xDX);
2379 if (!c_mode)
2380 mask |= 1 << xCX;
2381
de50b98b 2382 if (po->op == OP_CALL
89ff3147 2383 && ((po_test->regmask_src | po_test->regmask_dst) & mask))
5101a5f9 2384 return 1;
2385
2386 for (i = 0; i < po_test->operand_cnt; i++)
2387 if (IS(po_test->operand[i].name, po->operand[0].name))
2388 return 1;
2389
2390 return 0;
2391}
2392
940e8e66 2393// scan for any po_test operand modification in range given
89ff3147 2394static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt,
2395 int c_mode)
69a3cdfc 2396{
2b43685d 2397 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2398 return -1;
2399
69a3cdfc 2400 for (; i < opcnt; i++) {
89ff3147 2401 if (is_any_opr_modified(po_test, &ops[i], c_mode))
69a3cdfc 2402 return i;
2403 }
2404
2405 return -1;
2406}
2407
940e8e66 2408// scan for po_test operand[0] modification in range given
2409static int scan_for_mod_opr0(struct parsed_op *po_test,
2410 int i, int opcnt)
2411{
2412 for (; i < opcnt; i++) {
2413 if (is_opr_modified(&po_test->operand[0], &ops[i]))
2414 return i;
2415 }
2416
2417 return -1;
2418}
2419
92d715b6 2420#define check_i(po, i) \
2421 if ((i) < 0) \
2422 ferr(po, "bad " #i ": %d\n", i)
2423
04f8a628 2424static int scan_for_flag_set(int i, int magic, int *branched,
2425 int *setters, int *setter_cnt)
69a3cdfc 2426{
04f8a628 2427 struct label_ref *lr;
2b43685d 2428 int ret;
de50b98b 2429
2430 while (i >= 0) {
04f8a628 2431 if (ops[i].cc_scratch == magic) {
2432 ferr(&ops[i], "%s looped\n", __func__);
2433 return -1;
2434 }
2435 ops[i].cc_scratch = magic;
2436
d7857c3a 2437 if (g_labels[i] != NULL) {
2b43685d 2438 *branched = 1;
04f8a628 2439
2440 lr = &g_label_refs[i];
2441 for (; lr->next; lr = lr->next) {
92d715b6 2442 check_i(&ops[i], lr->i);
04f8a628 2443 ret = scan_for_flag_set(lr->i, magic,
2444 branched, setters, setter_cnt);
2445 if (ret < 0)
2446 return ret;
2447 }
2448
92d715b6 2449 check_i(&ops[i], lr->i);
de50b98b 2450 if (i > 0 && LAST_OP(i - 1)) {
94d447fb 2451 i = lr->i;
de50b98b 2452 continue;
2453 }
04f8a628 2454 ret = scan_for_flag_set(lr->i, magic,
2455 branched, setters, setter_cnt);
2b43685d 2456 if (ret < 0)
2457 return ret;
de50b98b 2458 }
2459 i--;
2460
2b43685d 2461 if (ops[i].flags & OPF_FLAGS) {
2462 setters[*setter_cnt] = i;
2463 (*setter_cnt)++;
2464 return 0;
2465 }
69a3cdfc 2466
5c024ef7 2467 if ((ops[i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP)
69a3cdfc 2468 return -1;
69a3cdfc 2469 }
2470
2471 return -1;
2472}
2473
5101a5f9 2474// scan back for cdq, if anything modifies edx, fail
2475static int scan_for_cdq_edx(int i)
2476{
cdfaeed7 2477 while (i >= 0) {
d7857c3a 2478 if (g_labels[i] != NULL) {
cdfaeed7 2479 if (g_label_refs[i].next != NULL)
2480 return -1;
2481 if (i > 0 && LAST_OP(i - 1)) {
2482 i = g_label_refs[i].i;
2483 continue;
2484 }
2485 return -1;
2486 }
2487 i--;
2488
5101a5f9 2489 if (ops[i].op == OP_CDQ)
2490 return i;
2491
2492 if (ops[i].regmask_dst & (1 << xDX))
2493 return -1;
5101a5f9 2494 }
2495
2496 return -1;
2497}
2498
64c59faf 2499static int scan_for_reg_clear(int i, int reg)
2500{
cdfaeed7 2501 while (i >= 0) {
d7857c3a 2502 if (g_labels[i] != NULL) {
cdfaeed7 2503 if (g_label_refs[i].next != NULL)
2504 return -1;
2505 if (i > 0 && LAST_OP(i - 1)) {
2506 i = g_label_refs[i].i;
2507 continue;
2508 }
2509 return -1;
2510 }
2511 i--;
2512
64c59faf 2513 if (ops[i].op == OP_XOR
2514 && ops[i].operand[0].lmod == OPLM_DWORD
2515 && ops[i].operand[0].reg == ops[i].operand[1].reg
2516 && ops[i].operand[0].reg == reg)
2517 return i;
2518
2519 if (ops[i].regmask_dst & (1 << reg))
2520 return -1;
64c59faf 2521 }
2522
2523 return -1;
2524}
2525
ee2361b9 2526static void patch_esp_adjust(struct parsed_op *po, int adj)
2527{
2528 ferr_assert(po, po->op == OP_ADD);
2529 ferr_assert(po, IS(opr_name(po, 0), "esp"));
2530 ferr_assert(po, po->operand[1].type == OPT_CONST);
2531
2532 // this is a bit of a hack, but deals with use of
2533 // single adj for multiple calls
2534 po->operand[1].val -= adj;
2535 po->flags |= OPF_RMD;
2536 if (po->operand[1].val == 0)
2537 po->flags |= OPF_DONE;
2538 ferr_assert(po, (int)po->operand[1].val >= 0);
2539}
2540
1bafb621 2541// scan for positive, constant esp adjust
ee2361b9 2542// multipath case is preliminary
9af2d373 2543static int scan_for_esp_adjust(int i, int opcnt,
ee2361b9 2544 int adj_expect, int *adj, int *is_multipath, int do_update)
1bafb621 2545{
7ba45c34 2546 struct parsed_op *po;
46411e6c 2547 int first_pop = -1;
4741fdfe 2548
ee2361b9 2549 *adj = *is_multipath = 0;
7ba45c34 2550
9af2d373 2551 for (; i < opcnt && *adj < adj_expect; i++) {
d7857c3a 2552 if (g_labels[i] != NULL)
ee2361b9 2553 *is_multipath = 1;
a2c1d768 2554
5e49b270 2555 po = &ops[i];
2556 if (po->flags & OPF_DONE)
2557 continue;
2558
7ba45c34 2559 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
2560 if (po->operand[1].type != OPT_CONST)
1bafb621 2561 ferr(&ops[i], "non-const esp adjust?\n");
7ba45c34 2562 *adj += po->operand[1].val;
1bafb621 2563 if (*adj & 3)
2564 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
ee2361b9 2565 if (do_update) {
2566 if (!*is_multipath)
2567 patch_esp_adjust(po, adj_expect);
2568 else
2569 po->flags |= OPF_RMD;
2570 }
1bafb621 2571 return i;
2572 }
5e49b270 2573 else if (po->op == OP_PUSH) {
5c024ef7 2574 //if (first_pop == -1)
2575 // first_pop = -2; // none
7ba45c34 2576 *adj -= lmod_bytes(po, po->operand[0].lmod);
46411e6c 2577 }
5e49b270 2578 else if (po->op == OP_POP) {
91ca764a 2579 if (!(po->flags & OPF_DONE)) {
2580 // seems like msvc only uses 'pop ecx' for stack realignment..
2581 if (po->operand[0].type != OPT_REG || po->operand[0].reg != xCX)
2582 break;
2583 if (first_pop == -1 && *adj >= 0)
2584 first_pop = i;
2585 }
ee2361b9 2586 if (do_update && *adj >= 0) {
2587 po->flags |= OPF_RMD;
2588 if (!*is_multipath)
2589 po->flags |= OPF_DONE;
2590 }
2591
7ba45c34 2592 *adj += lmod_bytes(po, po->operand[0].lmod);
46411e6c 2593 }
e56ab892 2594 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
4741fdfe 2595 if (po->op == OP_JMP && po->btj == NULL) {
5e49b270 2596 if (po->bt_i <= i)
2597 break;
4741fdfe 2598 i = po->bt_i - 1;
2599 continue;
2600 }
e56ab892 2601 if (po->op != OP_CALL)
a2c1d768 2602 break;
e56ab892 2603 if (po->operand[0].type != OPT_LABEL)
a2c1d768 2604 break;
91ca764a 2605 if (po->pp != NULL && po->pp->is_stdcall)
89ff3147 2606 break;
91ca764a 2607 // assume it's another cdecl call
e56ab892 2608 }
a2c1d768 2609 }
7ba45c34 2610
108e9fe3 2611 if (first_pop >= 0) {
a2c1d768 2612 // probably 'pop ecx' was used..
46411e6c 2613 return first_pop;
1bafb621 2614 }
2615
2616 return -1;
2617}
2618
a3684be1 2619static void scan_fwd_set_flags(int i, int opcnt, int magic, int flags)
2620{
2621 struct parsed_op *po;
2622 int j;
2623
2624 if (i < 0)
2625 ferr(ops, "%s: followed bad branch?\n", __func__);
2626
2627 for (; i < opcnt; i++) {
2628 po = &ops[i];
2629 if (po->cc_scratch == magic)
2630 return;
2631 po->cc_scratch = magic;
2632 po->flags |= flags;
2633
2634 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2635 if (po->btj != NULL) {
2636 // jumptable
2637 for (j = 0; j < po->btj->count; j++)
2638 scan_fwd_set_flags(po->btj->d[j].bt_i, opcnt, magic, flags);
2639 return;
2640 }
2641
2642 scan_fwd_set_flags(po->bt_i, opcnt, magic, flags);
5c024ef7 2643 if (!(po->flags & OPF_CJMP))
a3684be1 2644 return;
2645 }
2646 if (po->flags & OPF_TAIL)
2647 return;
2648 }
2649}
2650
1cd4a663 2651static const struct parsed_proto *try_recover_pp(
da87ae38 2652 struct parsed_op *po, const struct parsed_opr *opr, int *search_instead)
1cd4a663 2653{
2654 const struct parsed_proto *pp = NULL;
89ff3147 2655 char buf[256];
2656 char *p;
1cd4a663 2657
2658 // maybe an arg of g_func?
2659 if (opr->type == OPT_REGMEM && is_stack_access(po, opr))
2660 {
2661 char ofs_reg[16] = { 0, };
2662 int arg, arg_s, arg_i;
2663 int stack_ra = 0;
2664 int offset = 0;
2665
9af2d373 2666 if (g_header_mode)
2667 return NULL;
2668
1cd4a663 2669 parse_stack_access(po, opr->name, ofs_reg,
037f4971 2670 &offset, &stack_ra, NULL, 0);
1cd4a663 2671 if (ofs_reg[0] != 0)
2672 ferr(po, "offset reg on arg access?\n");
da87ae38 2673 if (offset <= stack_ra) {
2674 // search who set the stack var instead
2675 if (search_instead != NULL)
2676 *search_instead = 1;
2677 return NULL;
2678 }
1cd4a663 2679
2680 arg_i = (offset - stack_ra - 4) / 4;
2681 for (arg = arg_s = 0; arg < g_func_pp->argc; arg++) {
2682 if (g_func_pp->arg[arg].reg != NULL)
2683 continue;
2684 if (arg_s == arg_i)
2685 break;
2686 arg_s++;
2687 }
2688 if (arg == g_func_pp->argc)
2689 ferr(po, "stack arg %d not in prototype?\n", arg_i);
2690
2691 pp = g_func_pp->arg[arg].fptr;
2692 if (pp == NULL)
46411e6c 2693 ferr(po, "icall sa: arg%d is not a fptr?\n", arg + 1);
89ff3147 2694 check_func_pp(po, pp, "icall arg");
2695 }
2696 else if (opr->type == OPT_REGMEM && strchr(opr->name + 1, '[')) {
2697 // label[index]
2698 p = strchr(opr->name + 1, '[');
2699 memcpy(buf, opr->name, p - opr->name);
2700 buf[p - opr->name] = 0;
92d715b6 2701 pp = proto_parse(g_fhdr, buf, g_quiet_pp);
1cd4a663 2702 }
2703 else if (opr->type == OPT_OFFSET || opr->type == OPT_LABEL) {
9af2d373 2704 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
2705 if (pp == NULL) {
2706 if (!g_header_mode)
2707 ferr(po, "proto_parse failed for icall to '%s'\n", opr->name);
2708 }
2709 else
2710 check_func_pp(po, pp, "reg-fptr ref");
1cd4a663 2711 }
2712
2713 return pp;
2714}
2715
da87ae38 2716static void scan_for_call_type(int i, const struct parsed_opr *opr,
89ff3147 2717 int magic, const struct parsed_proto **pp_found, int *multi)
1cd4a663 2718{
2719 const struct parsed_proto *pp = NULL;
2720 struct parsed_op *po;
2721 struct label_ref *lr;
2722
037f4971 2723 ops[i].cc_scratch = magic;
1cd4a663 2724
037f4971 2725 while (1) {
d7857c3a 2726 if (g_labels[i] != NULL) {
1cd4a663 2727 lr = &g_label_refs[i];
92d715b6 2728 for (; lr != NULL; lr = lr->next) {
2729 check_i(&ops[i], lr->i);
89ff3147 2730 scan_for_call_type(lr->i, opr, magic, pp_found, multi);
92d715b6 2731 }
46411e6c 2732 if (i > 0 && LAST_OP(i - 1))
2733 return;
1cd4a663 2734 }
037f4971 2735
1cd4a663 2736 i--;
037f4971 2737 if (i < 0)
2738 break;
2739
2740 if (ops[i].cc_scratch == magic)
2741 return;
2742 ops[i].cc_scratch = magic;
1cd4a663 2743
2744 if (!(ops[i].flags & OPF_DATA))
2745 continue;
2746 if (!is_opr_modified(opr, &ops[i]))
2747 continue;
2748 if (ops[i].op != OP_MOV && ops[i].op != OP_LEA) {
2749 // most probably trashed by some processing
2750 *pp_found = NULL;
2751 return;
2752 }
2753
2754 opr = &ops[i].operand[1];
2755 if (opr->type != OPT_REG)
2756 break;
2757 }
2758
2759 po = (i >= 0) ? &ops[i] : ops;
2760
2761 if (i < 0) {
2762 // reached the top - can only be an arg-reg
2763 if (opr->type != OPT_REG)
2764 return;
2765
2766 for (i = 0; i < g_func_pp->argc; i++) {
2767 if (g_func_pp->arg[i].reg == NULL)
2768 continue;
2769 if (IS(opr->name, g_func_pp->arg[i].reg))
2770 break;
2771 }
2772 if (i == g_func_pp->argc)
2773 return;
2774 pp = g_func_pp->arg[i].fptr;
2775 if (pp == NULL)
46411e6c 2776 ferr(po, "icall: arg%d (%s) is not a fptr?\n",
2777 i + 1, g_func_pp->arg[i].reg);
89ff3147 2778 check_func_pp(po, pp, "icall reg-arg");
1cd4a663 2779 }
2780 else
da87ae38 2781 pp = try_recover_pp(po, opr, NULL);
1cd4a663 2782
89ff3147 2783 if (*pp_found != NULL && pp != NULL && *pp_found != pp) {
1cd4a663 2784 if (!IS((*pp_found)->ret_type.name, pp->ret_type.name)
2785 || (*pp_found)->is_stdcall != pp->is_stdcall
89ff3147 2786 || (*pp_found)->is_fptr != pp->is_fptr
1cd4a663 2787 || (*pp_found)->argc != pp->argc
2788 || (*pp_found)->argc_reg != pp->argc_reg
2789 || (*pp_found)->argc_stack != pp->argc_stack)
2790 {
2791 ferr(po, "icall: parsed_proto mismatch\n");
2792 }
89ff3147 2793 *multi = 1;
1cd4a663 2794 }
2795 if (pp != NULL)
2796 *pp_found = pp;
2797}
2798
9af2d373 2799// early check for tail call or branch back
2800static int is_like_tailjmp(int j)
2801{
2802 if (!(ops[j].flags & OPF_JMP))
2803 return 0;
2804
2805 if (ops[j].op == OP_JMP && !ops[j].operand[0].had_ds)
2806 // probably local branch back..
2807 return 1;
2808 if (ops[j].op == OP_CALL)
2809 // probably noreturn call..
2810 return 1;
2811
2812 return 0;
2813}
2814
2815static void scan_prologue_epilogue(int opcnt)
2816{
2817 int ecx_push = 0, esp_sub = 0;
2818 int found;
2819 int i, j, l;
2820
2821 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
2822 && ops[1].op == OP_MOV
2823 && IS(opr_name(&ops[1], 0), "ebp")
2824 && IS(opr_name(&ops[1], 1), "esp"))
2825 {
2826 g_bp_frame = 1;
5e49b270 2827 ops[0].flags |= OPF_RMD | OPF_DONE;
2828 ops[1].flags |= OPF_RMD | OPF_DONE;
9af2d373 2829 i = 2;
2830
2831 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
2832 g_stack_fsz = opr_const(&ops[2], 1);
5e49b270 2833 ops[2].flags |= OPF_RMD | OPF_DONE;
9af2d373 2834 i++;
2835 }
2836 else {
2837 // another way msvc builds stack frame..
2838 i = 2;
2839 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
2840 g_stack_fsz += 4;
5e49b270 2841 ops[i].flags |= OPF_RMD | OPF_DONE;
9af2d373 2842 ecx_push++;
2843 i++;
2844 }
2845 // and another way..
2846 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
2847 && ops[i].operand[1].type == OPT_CONST
2848 && ops[i + 1].op == OP_CALL
2849 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
2850 {
2851 g_stack_fsz += ops[i].operand[1].val;
5e49b270 2852 ops[i].flags |= OPF_RMD | OPF_DONE;
9af2d373 2853 i++;
5e49b270 2854 ops[i].flags |= OPF_RMD | OPF_DONE;
9af2d373 2855 i++;
2856 }
2857 }
2858
2859 found = 0;
2860 do {
2861 for (; i < opcnt; i++)
2862 if (ops[i].op == OP_RET)
2863 break;
2864 j = i - 1;
2865 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
2866 if (found && is_like_tailjmp(j))
2867 break;
2868 j--;
2869 }
2870
2871 if ((ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ebp"))
2872 || ops[j].op == OP_LEAVE)
2873 {
5e49b270 2874 ops[j].flags |= OPF_RMD | OPF_DONE;
9af2d373 2875 }
2876 else if (!(g_ida_func_attr & IDAFA_NORETURN))
2877 ferr(&ops[j], "'pop ebp' expected\n");
2878
2879 if (g_stack_fsz != 0) {
2880 if (ops[j - 1].op == OP_MOV
2881 && IS(opr_name(&ops[j - 1], 0), "esp")
2882 && IS(opr_name(&ops[j - 1], 1), "ebp"))
2883 {
5e49b270 2884 ops[j - 1].flags |= OPF_RMD | OPF_DONE;
9af2d373 2885 }
2886 else if (ops[j].op != OP_LEAVE
2887 && !(g_ida_func_attr & IDAFA_NORETURN))
2888 {
2889 ferr(&ops[j - 1], "esp restore expected\n");
2890 }
2891
2892 if (ecx_push && ops[j - 2].op == OP_POP
2893 && IS(opr_name(&ops[j - 2], 0), "ecx"))
2894 {
2895 ferr(&ops[j - 2], "unexpected ecx pop\n");
2896 }
2897 }
2898
2899 found = 1;
2900 i++;
2901 } while (i < opcnt);
2902
2903 return;
2904 }
2905
2906 // non-bp frame
2907 i = 0;
2908 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
5e49b270 2909 ops[i].flags |= OPF_RMD | OPF_DONE;
9af2d373 2910 g_stack_fsz += 4;
2911 ecx_push++;
2912 i++;
2913 }
2914
2915 for (; i < opcnt; i++) {
2916 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
2917 break;
2918 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
2919 && ops[i].operand[1].type == OPT_CONST)
2920 {
2921 g_stack_fsz = ops[i].operand[1].val;
5e49b270 2922 ops[i].flags |= OPF_RMD | OPF_DONE;
9af2d373 2923 esp_sub = 1;
2924 break;
2925 }
2926 }
2927
2928 if (ecx_push && !esp_sub) {
2929 // could actually be args for a call..
2930 for (; i < opcnt; i++)
2931 if (ops[i].op != OP_PUSH)
2932 break;
2933
2934 if (ops[i].op == OP_CALL && ops[i].operand[0].type == OPT_LABEL) {
2935 const struct parsed_proto *pp;
2936 pp = proto_parse(g_fhdr, opr_name(&ops[i], 0), 1);
2937 j = pp ? pp->argc_stack : 0;
2938 while (i > 0 && j > 0) {
2939 i--;
2940 if (ops[i].op == OP_PUSH) {
5e49b270 2941 ops[i].flags &= ~(OPF_RMD | OPF_DONE);
9af2d373 2942 j--;
2943 }
2944 }
2945 if (j != 0)
2946 ferr(&ops[i], "unhandled prologue\n");
2947
2948 // recheck
2949 i = g_stack_fsz = ecx_push = 0;
2950 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
2951 if (!(ops[i].flags & OPF_RMD))
2952 break;
2953 g_stack_fsz += 4;
2954 ecx_push++;
2955 i++;
2956 }
2957 }
2958 }
2959
2960 found = 0;
2961 if (ecx_push || esp_sub)
2962 {
2963 g_sp_frame = 1;
2964
2965 i++;
2966 do {
2967 for (; i < opcnt; i++)
2968 if (ops[i].op == OP_RET)
2969 break;
2970 j = i - 1;
2971 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
2972 if (found && is_like_tailjmp(j))
2973 break;
2974 j--;
2975 }
2976
2977 if (ecx_push > 0) {
2978 for (l = 0; l < ecx_push; l++) {
2979 if (ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ecx"))
2980 /* pop ecx */;
2981 else if (ops[j].op == OP_ADD
2982 && IS(opr_name(&ops[j], 0), "esp")
2983 && ops[j].operand[1].type == OPT_CONST)
2984 {
2985 /* add esp, N */
5e49b270 2986 l += ops[j].operand[1].val / 4 - 1;
9af2d373 2987 }
2988 else
2989 ferr(&ops[j], "'pop ecx' expected\n");
2990
5e49b270 2991 ops[j].flags |= OPF_RMD | OPF_DONE;
9af2d373 2992 j--;
2993 }
2994 if (l != ecx_push)
2995 ferr(&ops[j], "epilogue scan failed\n");
2996
2997 found = 1;
2998 }
2999
3000 if (esp_sub) {
3001 if (ops[j].op != OP_ADD
3002 || !IS(opr_name(&ops[j], 0), "esp")
3003 || ops[j].operand[1].type != OPT_CONST
3004 || ops[j].operand[1].val != g_stack_fsz)
3005 ferr(&ops[j], "'add esp' expected\n");
3006
5e49b270 3007 ops[j].flags |= OPF_RMD | OPF_DONE;
9af2d373 3008 ops[j].operand[1].val = 0; // hack for stack arg scanner
3009 found = 1;
3010 }
3011
3012 i++;
3013 } while (i < opcnt);
3014 }
3015}
3016
89ff3147 3017static const struct parsed_proto *resolve_icall(int i, int opcnt,
3018 int *multi_src)
1cd4a663 3019{
3020 const struct parsed_proto *pp = NULL;
da87ae38 3021 int search_advice = 0;
1cd4a663 3022
89ff3147 3023 *multi_src = 0;
3024
1cd4a663 3025 switch (ops[i].operand[0].type) {
3026 case OPT_REGMEM:
3027 case OPT_LABEL:
3028 case OPT_OFFSET:
da87ae38 3029 pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice);
3030 if (!search_advice)
3031 break;
3032 // fallthrough
1cd4a663 3033 default:
89ff3147 3034 scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp,
3035 multi_src);
1cd4a663 3036 break;
3037 }
3038
3039 return pp;
3040}
3041
23fd0b11 3042// find an instruction that changed opr before i op
92d715b6 3043// *op_i must be set to -1 by caller
3044// *entry is set to 1 if one source is determined to be the caller
3045// returns 1 if found, *op_i is then set to origin
23fd0b11 3046static int resolve_origin(int i, const struct parsed_opr *opr,
92d715b6 3047 int magic, int *op_i, int *is_caller)
1f84f6b3 3048{
3049 struct label_ref *lr;
3050 int ret = 0;
3051
92d715b6 3052 if (ops[i].cc_scratch == magic)
3053 return 0;
1f84f6b3 3054 ops[i].cc_scratch = magic;
3055
3056 while (1) {
d7857c3a 3057 if (g_labels[i] != NULL) {
1f84f6b3 3058 lr = &g_label_refs[i];
92d715b6 3059 for (; lr != NULL; lr = lr->next) {
3060 check_i(&ops[i], lr->i);
3061 ret |= resolve_origin(lr->i, opr, magic, op_i, is_caller);
3062 }
1f84f6b3 3063 if (i > 0 && LAST_OP(i - 1))
3064 return ret;
3065 }
3066
3067 i--;
92d715b6 3068 if (i < 0) {
3069 if (is_caller != NULL)
3070 *is_caller = 1;
1f84f6b3 3071 return -1;
92d715b6 3072 }
1f84f6b3 3073
3074 if (ops[i].cc_scratch == magic)
3075 return 0;
3076 ops[i].cc_scratch = magic;
3077
3078 if (!(ops[i].flags & OPF_DATA))
3079 continue;
3080 if (!is_opr_modified(opr, &ops[i]))
3081 continue;
23fd0b11 3082
3083 if (*op_i >= 0) {
3084 if (*op_i == i)
3085 return 1;
3086 // XXX: could check if the other op does the same
3087 return -1;
3088 }
3089
3090 *op_i = i;
3091 return 1;
3092 }
3093}
3094
3095static int try_resolve_const(int i, const struct parsed_opr *opr,
3096 int magic, unsigned int *val)
3097{
3098 int s_i = -1;
92d715b6 3099 int ret;
23fd0b11 3100
92d715b6 3101 ret = resolve_origin(i, opr, magic, &s_i, NULL);
23fd0b11 3102 if (ret == 1) {
3103 i = s_i;
1f84f6b3 3104 if (ops[i].op != OP_MOV && ops[i].operand[1].type != OPT_CONST)
3105 return -1;
3106
3107 *val = ops[i].operand[1].val;
3108 return 1;
3109 }
23fd0b11 3110
3111 return -1;
1f84f6b3 3112}
3113
26677139 3114static struct parsed_proto *process_call_early(int i, int opcnt,
3115 int *adj_i)
3116{
3117 struct parsed_op *po = &ops[i];
3118 struct parsed_proto *pp;
3119 int multipath = 0;
3120 int adj = 0;
3121 int ret;
3122
3123 pp = po->pp;
3124 if (pp == NULL || pp->is_vararg || pp->argc_reg != 0)
3125 // leave for later
3126 return NULL;
3127
3128 // look for and make use of esp adjust
3129 *adj_i = ret = -1;
3130 if (!pp->is_stdcall && pp->argc_stack > 0)
3131 ret = scan_for_esp_adjust(i + 1, opcnt,
ee2361b9 3132 pp->argc_stack * 4, &adj, &multipath, 0);
26677139 3133 if (ret >= 0) {
3134 if (pp->argc_stack > adj / 4)
3135 return NULL;
3136 if (multipath)
3137 return NULL;
3138 if (ops[ret].op == OP_POP && adj != 4)
3139 return NULL;
3140 }
3141
3142 *adj_i = ret;
3143 return pp;
3144}
3145
9af2d373 3146static struct parsed_proto *process_call(int i, int opcnt)
3147{
3148 struct parsed_op *po = &ops[i];
3149 const struct parsed_proto *pp_c;
3150 struct parsed_proto *pp;
3151 const char *tmpname;
26677139 3152 int adj = 0, multipath = 0;
9af2d373 3153 int ret, arg;
3154
3155 tmpname = opr_name(po, 0);
3156 pp = po->pp;
3157 if (pp == NULL)
3158 {
3159 // indirect call
26677139 3160 pp_c = resolve_icall(i, opcnt, &multipath);
9af2d373 3161 if (pp_c != NULL) {
3162 if (!pp_c->is_func && !pp_c->is_fptr)
3163 ferr(po, "call to non-func: %s\n", pp_c->name);
3164 pp = proto_clone(pp_c);
3165 my_assert_not(pp, NULL);
26677139 3166 if (multipath)
9af2d373 3167 // not resolved just to single func
3168 pp->is_fptr = 1;
3169
3170 switch (po->operand[0].type) {
3171 case OPT_REG:
3172 // we resolved this call and no longer need the register
3173 po->regmask_src &= ~(1 << po->operand[0].reg);
3174 break;
3175 case OPT_REGMEM:
3176 pp->is_fptr = 1;
3177 break;
3178 default:
3179 break;
3180 }
3181 }
3182 if (pp == NULL) {
3183 pp = calloc(1, sizeof(*pp));
3184 my_assert_not(pp, NULL);
3185
3186 pp->is_fptr = 1;
ee2361b9 3187 ret = scan_for_esp_adjust(i + 1, opcnt,
3188 32*4, &adj, &multipath, 0);
26677139 3189 if (ret < 0 || adj < 0) {
9af2d373 3190 if (!g_allow_regfunc)
3191 ferr(po, "non-__cdecl indirect call unhandled yet\n");
3192 pp->is_unresolved = 1;
26677139 3193 adj = 0;
9af2d373 3194 }
26677139 3195 adj /= 4;
3196 if (adj > ARRAY_SIZE(pp->arg))
3197 ferr(po, "esp adjust too large: %d\n", adj);
9af2d373 3198 pp->ret_type.name = strdup("int");
26677139 3199 pp->argc = pp->argc_stack = adj;
9af2d373 3200 for (arg = 0; arg < pp->argc; arg++)
3201 pp->arg[arg].type.name = strdup("int");
3202 }
3203 po->pp = pp;
3204 }
3205
3206 // look for and make use of esp adjust
91ca764a 3207 multipath = 0;
9af2d373 3208 ret = -1;
3209 if (!pp->is_stdcall && pp->argc_stack > 0)
3210 ret = scan_for_esp_adjust(i + 1, opcnt,
ee2361b9 3211 pp->argc_stack * 4, &adj, &multipath, 0);
9af2d373 3212 if (ret >= 0) {
3213 if (pp->is_vararg) {
26677139 3214 if (adj / 4 < pp->argc_stack) {
3215 fnote(po, "(this call)\n");
3216 ferr(&ops[ret], "esp adjust is too small: %x < %x\n",
3217 adj, pp->argc_stack * 4);
3218 }
9af2d373 3219 // modify pp to make it have varargs as normal args
3220 arg = pp->argc;
26677139 3221 pp->argc += adj / 4 - pp->argc_stack;
9af2d373 3222 for (; arg < pp->argc; arg++) {
3223 pp->arg[arg].type.name = strdup("int");
3224 pp->argc_stack++;
3225 }
3226 if (pp->argc > ARRAY_SIZE(pp->arg))
3227 ferr(po, "too many args for '%s'\n", tmpname);
3228 }
26677139 3229 if (pp->argc_stack > adj / 4) {
9af2d373 3230 fnote(po, "(this call)\n");
3231 ferr(&ops[ret], "stack tracking failed for '%s': %x %x\n",
26677139 3232 tmpname, pp->argc_stack * 4, adj);
9af2d373 3233 }
3234
ee2361b9 3235 scan_for_esp_adjust(i + 1, opcnt,
3236 pp->argc_stack * 4, &adj, &multipath, 1);
9af2d373 3237 }
3238 else if (pp->is_vararg)
3239 ferr(po, "missing esp_adjust for vararg func '%s'\n",
3240 pp->name);
3241
3242 return pp;
3243}
3244
26677139 3245static int collect_call_args_early(struct parsed_op *po, int i,
3246 struct parsed_proto *pp, int *regmask)
3247{
3248 int arg, ret;
3249 int j;
3250
3251 for (arg = 0; arg < pp->argc; arg++)
3252 if (pp->arg[arg].reg == NULL)
3253 break;
3254
3255 // first see if it can be easily done
3256 for (j = i; j > 0 && arg < pp->argc; )
3257 {
3258 if (g_labels[j] != NULL)
3259 return -1;
3260 j--;
3261
3262 if (ops[j].op == OP_CALL)
3263 return -1;
3264 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP)
3265 return -1;
3266 else if (ops[j].op == OP_POP)
3267 return -1;
3268 else if (ops[j].flags & OPF_CJMP)
3269 return -1;
3270 else if (ops[j].op == OP_PUSH) {
3271 if (ops[j].flags & (OPF_FARG|OPF_FARGNR))
3272 return -1;
3273 ret = scan_for_mod(&ops[j], j + 1, i, 1);
3274 if (ret >= 0)
3275 return -1;
3276
3277 if (pp->arg[arg].type.is_va_list)
3278 return -1;
3279
3280 // next arg
3281 for (arg++; arg < pp->argc; arg++)
3282 if (pp->arg[arg].reg == NULL)
3283 break;
3284 }
3285 }
3286
3287 if (arg < pp->argc)
3288 return -1;
3289
3290 // now do it
3291 for (arg = 0; arg < pp->argc; arg++)
3292 if (pp->arg[arg].reg == NULL)
3293 break;
3294
3295 for (j = i; j > 0 && arg < pp->argc; )
3296 {
3297 j--;
3298
3299 if (ops[j].op == OP_PUSH)
3300 {
3301 ops[j].p_argnext = -1;
3302 ferr_assert(&ops[j], pp->arg[arg].datap == NULL);
3303 pp->arg[arg].datap = &ops[j];
3304
3305 if (ops[j].operand[0].type == OPT_REG)
3306 *regmask |= 1 << ops[j].operand[0].reg;
3307
5e49b270 3308 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_FARGNR | OPF_FARG;
26677139 3309 ops[j].flags &= ~OPF_RSAVE;
3310
3311 // next arg
3312 for (arg++; arg < pp->argc; arg++)
3313 if (pp->arg[arg].reg == NULL)
3314 break;
3315 }
3316 }
3317
3318 return 0;
3319}
3320
89ff3147 3321static int collect_call_args_r(struct parsed_op *po, int i,
3a5101d7 3322 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
3323 int *arg_grp, int arg, int magic, int need_op_saving, int may_reuse)
e56ab892 3324{
3325 struct parsed_proto *pp_tmp;
3a5101d7 3326 struct parsed_op *po_tmp;
e56ab892 3327 struct label_ref *lr;
2b43685d 3328 int need_to_save_current;
3a5101d7 3329 int arg_grp_current = 0;
3330 int save_args_seen = 0;
23fd0b11 3331 int save_args;
e56ab892 3332 int ret = 0;
5f70a34f 3333 int reg;
23fd0b11 3334 char buf[32];
3335 int j, k;
e56ab892 3336
a3684be1 3337 if (i < 0) {
a2c1d768 3338 ferr(po, "dead label encountered\n");
a3684be1 3339 return -1;
3340 }
e56ab892 3341
de50b98b 3342 for (; arg < pp->argc; arg++)
e56ab892 3343 if (pp->arg[arg].reg == NULL)
3344 break;
a3684be1 3345 magic = (magic & 0xffffff) | (arg << 24);
e56ab892 3346
89ff3147 3347 for (j = i; j >= 0 && (arg < pp->argc || pp->is_unresolved); )
e56ab892 3348 {
a3684be1 3349 if (((ops[j].cc_scratch ^ magic) & 0xffffff) == 0) {
3350 if (ops[j].cc_scratch != magic) {
3351 ferr(&ops[j], "arg collect hit same path with diff args for %s\n",
3352 pp->name);
3353 return -1;
3354 }
3355 // ok: have already been here
3356 return 0;
3357 }
3358 ops[j].cc_scratch = magic;
3359
d7857c3a 3360 if (g_labels[j] != NULL && g_label_refs[j].i != -1) {
e56ab892 3361 lr = &g_label_refs[j];
3362 if (lr->next != NULL)
3363 need_op_saving = 1;
a652aa9f 3364 for (; lr->next; lr = lr->next) {
92d715b6 3365 check_i(&ops[j], lr->i);
5c024ef7 3366 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
a652aa9f 3367 may_reuse = 1;
89ff3147 3368 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
3a5101d7 3369 arg_grp, arg, magic, need_op_saving, may_reuse);
a3684be1 3370 if (ret < 0)
3371 return ret;
a652aa9f 3372 }
e56ab892 3373
92d715b6 3374 check_i(&ops[j], lr->i);
5c024ef7 3375 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
a652aa9f 3376 may_reuse = 1;
de50b98b 3377 if (j > 0 && LAST_OP(j - 1)) {
e56ab892 3378 // follow last branch in reverse
3379 j = lr->i;
3380 continue;
3381 }
3382 need_op_saving = 1;
89ff3147 3383 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
3a5101d7 3384 arg_grp, arg, magic, need_op_saving, may_reuse);
a3684be1 3385 if (ret < 0)
3386 return ret;
e56ab892 3387 }
3388 j--;
3389
3390 if (ops[j].op == OP_CALL)
3391 {
89ff3147 3392 if (pp->is_unresolved)
3393 break;
3394
092f64e1 3395 pp_tmp = ops[j].pp;
e56ab892 3396 if (pp_tmp == NULL)
7ae48d73 3397 ferr(po, "arg collect hit unparsed call '%s'\n",
3398 ops[j].operand[0].name);
a652aa9f 3399 if (may_reuse && pp_tmp->argc_stack > 0)
de50b98b 3400 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
3401 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
e56ab892 3402 }
fdd5548a 3403 // esp adjust of 0 means we collected it before
3404 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP
3405 && (ops[j].operand[1].type != OPT_CONST
3406 || ops[j].operand[1].val != 0))
3407 {
89ff3147 3408 if (pp->is_unresolved)
3409 break;
3410
fdd5548a 3411 ferr(po, "arg collect %d/%d hit esp adjust of %d\n",
3412 arg, pp->argc, ops[j].operand[1].val);
de50b98b 3413 }
5e49b270 3414 else if (ops[j].op == OP_POP && !(ops[j].flags & OPF_DONE))
9af2d373 3415 {
89ff3147 3416 if (pp->is_unresolved)
3417 break;
3418
de50b98b 3419 ferr(po, "arg collect %d/%d hit pop\n", arg, pp->argc);
3420 }
5c024ef7 3421 else if (ops[j].flags & OPF_CJMP)
de50b98b 3422 {
89ff3147 3423 if (pp->is_unresolved)
3424 break;
3425
a652aa9f 3426 may_reuse = 1;
de50b98b 3427 }
91ca764a 3428 else if (ops[j].op == OP_PUSH
3429 && !(ops[j].flags & (OPF_FARGNR|OPF_DONE)))
e56ab892 3430 {
89ff3147 3431 if (pp->is_unresolved && (ops[j].flags & OPF_RMD))
3432 break;
3433
3a5101d7 3434 ops[j].p_argnext = -1;
3435 po_tmp = pp->arg[arg].datap;
3436 if (po_tmp != NULL)
3437 ops[j].p_argnext = po_tmp - ops;
e56ab892 3438 pp->arg[arg].datap = &ops[j];
3a5101d7 3439
2b43685d 3440 need_to_save_current = 0;
23fd0b11 3441 save_args = 0;
5f70a34f 3442 reg = -1;
3443 if (ops[j].operand[0].type == OPT_REG)
3444 reg = ops[j].operand[0].reg;
3445
e56ab892 3446 if (!need_op_saving) {
89ff3147 3447 ret = scan_for_mod(&ops[j], j + 1, i, 1);
2b43685d 3448 need_to_save_current = (ret >= 0);
e56ab892 3449 }
2b43685d 3450 if (need_op_saving || need_to_save_current) {
e56ab892 3451 // mark this push as one that needs operand saving
3452 ops[j].flags &= ~OPF_RMD;
5f70a34f 3453 if (ops[j].p_argnum == 0) {
3454 ops[j].p_argnum = arg + 1;
23fd0b11 3455 save_args |= 1 << arg;
de50b98b 3456 }
fdd5548a 3457 else if (ops[j].p_argnum < arg + 1) {
3458 // XXX: might kill valid var..
3459 //*save_arg_vars &= ~(1 << (ops[j].p_argnum - 1));
3460 ops[j].p_argnum = arg + 1;
3461 save_args |= 1 << arg;
3462 }
3a5101d7 3463
3464 if (save_args_seen & (1 << (ops[j].p_argnum - 1))) {
3465 save_args_seen = 0;
3466 arg_grp_current++;
3467 if (arg_grp_current >= MAX_ARG_GRP)
3468 ferr(&ops[j], "out of arg groups (arg%d), f %s\n",
3469 ops[j].p_argnum, pp->name);
3470 }
e56ab892 3471 }
5f70a34f 3472 else if (ops[j].p_argnum == 0)
e56ab892 3473 ops[j].flags |= OPF_RMD;
3474
a652aa9f 3475 // some PUSHes are reused by different calls on other branches,
de50b98b 3476 // but that can't happen if we didn't branch, so they
3477 // can be removed from future searches (handles nested calls)
a652aa9f 3478 if (!may_reuse)
9af2d373 3479 ops[j].flags |= OPF_FARGNR;
de50b98b 3480
9af2d373 3481 ops[j].flags |= OPF_FARG;
da87ae38 3482 ops[j].flags &= ~OPF_RSAVE;
3483
23fd0b11 3484 // check for __VALIST
5f70a34f 3485 if (!pp->is_unresolved && pp->arg[arg].type.is_va_list) {
23fd0b11 3486 k = -1;
92d715b6 3487 ret = resolve_origin(j, &ops[j].operand[0],
3488 magic + 1, &k, NULL);
5f70a34f 3489 if (ret == 1 && k >= 0)
23fd0b11 3490 {
5f70a34f 3491 if (ops[k].op == OP_LEA) {
3492 snprintf(buf, sizeof(buf), "arg_%X",
3493 g_func_pp->argc_stack * 4);
3494 if (!g_func_pp->is_vararg
3495 || strstr(ops[k].operand[1].name, buf))
3496 {
5e49b270 3497 ops[k].flags |= OPF_RMD | OPF_DONE;
5f70a34f 3498 ops[j].flags |= OPF_RMD | OPF_VAPUSH;
3499 save_args &= ~(1 << arg);
3500 reg = -1;
3501 }
3502 else
3503 ferr(&ops[j], "lea va_list used, but no vararg?\n");
3504 }
3505 // check for va_list from g_func_pp arg too
3506 else if (ops[k].op == OP_MOV
3507 && is_stack_access(&ops[k], &ops[k].operand[1]))
3508 {
3509 ret = stack_frame_access(&ops[k], &ops[k].operand[1],
3510 buf, sizeof(buf), ops[k].operand[1].name, "", 1, 0);
3511 if (ret >= 0) {
5e49b270 3512 ops[k].flags |= OPF_RMD | OPF_DONE;
5f70a34f 3513 ops[j].flags |= OPF_RMD;
3514 ops[j].p_argpass = ret + 1;
3515 save_args &= ~(1 << arg);
3516 reg = -1;
3517 }
3518 }
23fd0b11 3519 }
3520 }
3521
3522 *save_arg_vars |= save_args;
3523
3524 // tracking reg usage
5f70a34f 3525 if (reg >= 0)
3526 *regmask |= 1 << reg;
23fd0b11 3527
89ff3147 3528 arg++;
3529 if (!pp->is_unresolved) {
3530 // next arg
3531 for (; arg < pp->argc; arg++)
3532 if (pp->arg[arg].reg == NULL)
3533 break;
3534 }
a3684be1 3535 magic = (magic & 0xffffff) | (arg << 24);
e56ab892 3536 }
3a5101d7 3537
3538 if (ops[j].p_arggrp > arg_grp_current) {
3539 save_args_seen = 0;
3540 arg_grp_current = ops[j].p_arggrp;
3541 }
3542 if (ops[j].p_argnum > 0)
3543 save_args_seen |= 1 << (ops[j].p_argnum - 1);
e56ab892 3544 }
3545
3546 if (arg < pp->argc) {
3547 ferr(po, "arg collect failed for '%s': %d/%d\n",
3548 pp->name, arg, pp->argc);
89ff3147 3549 return -1;
e56ab892 3550 }
89ff3147 3551
3a5101d7 3552 if (arg_grp_current > *arg_grp)
3553 *arg_grp = arg_grp_current;
3554
89ff3147 3555 return arg;
3556}
3557
3558static int collect_call_args(struct parsed_op *po, int i,
3559 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
3560 int magic)
3561{
3a5101d7 3562 // arg group is for cases when pushes for
3563 // multiple funcs are going on
3564 struct parsed_op *po_tmp;
3565 int save_arg_vars_current = 0;
3566 int arg_grp = 0;
89ff3147 3567 int ret;
3568 int a;
3569
3a5101d7 3570 ret = collect_call_args_r(po, i, pp, regmask,
3571 &save_arg_vars_current, &arg_grp, 0, magic, 0, 0);
89ff3147 3572 if (ret < 0)
3573 return ret;
3574
3a5101d7 3575 if (arg_grp != 0) {
3576 // propagate arg_grp
3577 for (a = 0; a < pp->argc; a++) {
3578 if (pp->arg[a].reg != NULL)
3579 continue;
3580
3581 po_tmp = pp->arg[a].datap;
3582 while (po_tmp != NULL) {
3583 po_tmp->p_arggrp = arg_grp;
3584 if (po_tmp->p_argnext > 0)
3585 po_tmp = &ops[po_tmp->p_argnext];
3586 else
3587 po_tmp = NULL;
3588 }
3589 }
3590 }
3591 save_arg_vars[arg_grp] |= save_arg_vars_current;
3592
89ff3147 3593 if (pp->is_unresolved) {
3594 pp->argc += ret;
3595 pp->argc_stack += ret;
3596 for (a = 0; a < pp->argc; a++)
3597 if (pp->arg[a].type.name == NULL)
3598 pp->arg[a].type.name = strdup("int");
3599 }
3600
e56ab892 3601 return ret;
3602}
3603
89ff3147 3604static void pp_insert_reg_arg(struct parsed_proto *pp, const char *reg)
3605{
3606 int i;
3607
3608 for (i = 0; i < pp->argc; i++)
3609 if (pp->arg[i].reg == NULL)
3610 break;
3611
3612 if (pp->argc_stack)
3613 memmove(&pp->arg[i + 1], &pp->arg[i],
3614 sizeof(pp->arg[0]) * pp->argc_stack);
3615 memset(&pp->arg[i], 0, sizeof(pp->arg[i]));
3616 pp->arg[i].reg = strdup(reg);
3617 pp->arg[i].type.name = strdup("int");
3618 pp->argc++;
3619 pp->argc_reg++;
3620}
3621
4c45fa73 3622static void add_label_ref(struct label_ref *lr, int op_i)
3623{
3624 struct label_ref *lr_new;
3625
3626 if (lr->i == -1) {
3627 lr->i = op_i;
3628 return;
3629 }
3630
3631 lr_new = calloc(1, sizeof(*lr_new));
3632 lr_new->i = op_i;
a652aa9f 3633 lr_new->next = lr->next;
4c45fa73 3634 lr->next = lr_new;
3635}
3636
92d715b6 3637static struct parsed_data *try_resolve_jumptab(int i, int opcnt)
3638{
3639 struct parsed_op *po = &ops[i];
3640 struct parsed_data *pd;
3641 char label[NAMELEN], *p;
3642 int len, j, l;
3643
3644 p = strchr(po->operand[0].name, '[');
3645 if (p == NULL)
3646 return NULL;
3647
3648 len = p - po->operand[0].name;
3649 strncpy(label, po->operand[0].name, len);
3650 label[len] = 0;
3651
3652 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
3653 if (IS(g_func_pd[j].label, label)) {
3654 pd = &g_func_pd[j];
3655 break;
3656 }
3657 }
3658 if (pd == NULL)
3659 //ferr(po, "label '%s' not parsed?\n", label);
3660 return NULL;
3661
3662 if (pd->type != OPT_OFFSET)
3663 ferr(po, "label '%s' with non-offset data?\n", label);
3664
3665 // find all labels, link
3666 for (j = 0; j < pd->count; j++) {
3667 for (l = 0; l < opcnt; l++) {
d7857c3a 3668 if (g_labels[l] != NULL && IS(g_labels[l], pd->d[j].u.label)) {
92d715b6 3669 add_label_ref(&g_label_refs[l], i);
3670 pd->d[j].bt_i = l;
3671 break;
3672 }
3673 }
3674 }
3675
3676 return pd;
3677}
3678
d7857c3a 3679static void clear_labels(int count)
3680{
3681 int i;
3682
3683 for (i = 0; i < count; i++) {
3684 if (g_labels[i] != NULL) {
3685 free(g_labels[i]);
3686 g_labels[i] = NULL;
3687 }
3688 }
3689}
3690
04f8a628 3691static void output_std_flags(FILE *fout, struct parsed_op *po,
3692 int *pfomask, const char *dst_opr_text)
3693{
3694 if (*pfomask & (1 << PFO_Z)) {
3695 fprintf(fout, "\n cond_z = (%s%s == 0);",
3696 lmod_cast_u(po, po->operand[0].lmod), dst_opr_text);
3697 *pfomask &= ~(1 << PFO_Z);
3698 }
3699 if (*pfomask & (1 << PFO_S)) {
3700 fprintf(fout, "\n cond_s = (%s%s < 0);",
3701 lmod_cast_s(po, po->operand[0].lmod), dst_opr_text);
3702 *pfomask &= ~(1 << PFO_S);
3703 }
3704}
3705
c0de9015 3706enum {
3707 OPP_FORCE_NORETURN = (1 << 0),
3708 OPP_SIMPLE_ARGS = (1 << 1),
3709 OPP_ALIGN = (1 << 2),
3710};
3711
c0050df6 3712static void output_pp_attrs(FILE *fout, const struct parsed_proto *pp,
c0de9015 3713 int flags)
c0050df6 3714{
c0de9015 3715 const char *cconv = "";
3716
c0050df6 3717 if (pp->is_fastcall)
c0de9015 3718 cconv = "__fastcall ";
c0050df6 3719 else if (pp->is_stdcall && pp->argc_reg == 0)
c0de9015 3720 cconv = "__stdcall ";
3721
3722 fprintf(fout, (flags & OPP_ALIGN) ? "%-16s" : "%s", cconv);
3723
3724 if (pp->is_noreturn || (flags & OPP_FORCE_NORETURN))
c0050df6 3725 fprintf(fout, "noreturn ");
3726}
3727
c0de9015 3728static void output_pp(FILE *fout, const struct parsed_proto *pp,
3729 int flags)
3730{
3731 int i;
3732
3733 fprintf(fout, (flags & OPP_ALIGN) ? "%-5s" : "%s ",
3734 pp->ret_type.name);
3735 if (pp->is_fptr)
3736 fprintf(fout, "(");
3737 output_pp_attrs(fout, pp, flags);
3738 if (pp->is_fptr)
3739 fprintf(fout, "*");
3740 fprintf(fout, "%s", pp->name);
3741 if (pp->is_fptr)
3742 fprintf(fout, ")");
3743
3744 fprintf(fout, "(");
3745 for (i = 0; i < pp->argc; i++) {
3746 if (i > 0)
3747 fprintf(fout, ", ");
3748 if (pp->arg[i].fptr != NULL && !(flags & OPP_SIMPLE_ARGS)) {
3749 // func pointer
3750 output_pp(fout, pp->arg[i].fptr, 0);
3751 }
3752 else if (pp->arg[i].type.is_retreg) {
3753 fprintf(fout, "u32 *r_%s", pp->arg[i].reg);
3754 }
3755 else {
3756 fprintf(fout, "%s", pp->arg[i].type.name);
3757 if (!pp->is_fptr)
3758 fprintf(fout, " a%d", i + 1);
3759 }
3760 }
3761 if (pp->is_vararg) {
3762 if (i > 0)
3763 fprintf(fout, ", ");
3764 fprintf(fout, "...");
3765 }
3766 fprintf(fout, ")");
3767}
3768
9af2d373 3769static int get_pp_arg_regmask(const struct parsed_proto *pp)
3770{
3771 int regmask = 0;
3772 int i, reg;
3773
3774 for (i = 0; i < pp->argc; i++) {
3775 if (pp->arg[i].reg != NULL) {
3776 reg = char_array_i(regs_r32,
3777 ARRAY_SIZE(regs_r32), pp->arg[i].reg);
3778 if (reg < 0)
3779 ferr(ops, "arg '%s' of func '%s' is not a reg?\n",
3780 pp->arg[i].reg, pp->name);
3781 regmask |= 1 << reg;
3782 }
3783 }
3784
3785 return regmask;
3786}
3787
3a5101d7 3788static char *saved_arg_name(char *buf, size_t buf_size, int grp, int num)
3789{
3790 char buf1[16];
3791
3792 buf1[0] = 0;
3793 if (grp > 0)
3794 snprintf(buf1, sizeof(buf1), "%d", grp);
3795 snprintf(buf, buf_size, "s%s_a%d", buf1, num);
3796
3797 return buf;
3798}
3799
9af2d373 3800static void gen_x_cleanup(int opcnt);
3801
91977a1c 3802static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
3803{
69a3cdfc 3804 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
850c9265 3805 struct parsed_opr *last_arith_dst = NULL;
3ebea2cf 3806 char buf1[256], buf2[256], buf3[256], cast[64];
bd96f656 3807 const struct parsed_proto *pp_c;
ddaf8bd7 3808 struct parsed_proto *pp, *pp_tmp;
4c45fa73 3809 struct parsed_data *pd;
91977a1c 3810 const char *tmpname;
1f84f6b3 3811 unsigned int uval;
3a5101d7 3812 int save_arg_vars[MAX_ARG_GRP] = { 0, };
cb090db0 3813 int cond_vars = 0;
108e9fe3 3814 int need_tmp_var = 0;
2fe80fdb 3815 int need_tmp64 = 0;
91977a1c 3816 int had_decl = 0;
3ebea2cf 3817 int label_pending = 0;
d4e3b5db 3818 int regmask_save = 0;
91977a1c 3819 int regmask_arg = 0;
a3684be1 3820 int regmask_now = 0;
ddaf8bd7 3821 int regmask_init = 0;
91977a1c 3822 int regmask = 0;
940e8e66 3823 int pfomask = 0;
64c59faf 3824 int found = 0;
d4e3b5db 3825 int depth = 0;
91977a1c 3826 int no_output;
4c45fa73 3827 int i, j, l;
91977a1c 3828 int arg;
91977a1c 3829 int reg;
3830 int ret;
3831
1bafb621 3832 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
a2c1d768 3833 g_stack_frame_used = 0;
91977a1c 3834
36595fd2 3835 g_func_pp = proto_parse(fhdr, funcn, 0);
bd96f656 3836 if (g_func_pp == NULL)
91977a1c 3837 ferr(ops, "proto_parse failed for '%s'\n", funcn);
3838
9af2d373 3839 regmask_arg = get_pp_arg_regmask(g_func_pp);
91977a1c 3840
3841 // pass1:
1bafb621 3842 // - handle ebp/esp frame, remove ops related to it
9af2d373 3843 scan_prologue_epilogue(opcnt);
91977a1c 3844
3845 // pass2:
7ae48d73 3846 // - parse calls with labels
87bf6cec 3847 // - resolve all branches
de50b98b 3848 for (i = 0; i < opcnt; i++)
3849 {
87bf6cec 3850 po = &ops[i];
3851 po->bt_i = -1;
4c45fa73 3852 po->btj = NULL;
87bf6cec 3853
5e49b270 3854 if (po->flags & (OPF_RMD|OPF_DONE))
7ae48d73 3855 continue;
3856
3857 if (po->op == OP_CALL) {
ddaf8bd7 3858 pp = NULL;
3859
7ae48d73 3860 if (po->operand[0].type == OPT_LABEL) {
3861 tmpname = opr_name(po, 0);
46411e6c 3862 if (IS_START(tmpname, "loc_"))
3863 ferr(po, "call to loc_*\n");
36595fd2 3864 pp_c = proto_parse(fhdr, tmpname, 0);
7ae48d73 3865 if (pp_c == NULL)
3866 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
da87ae38 3867
7ae48d73 3868 pp = proto_clone(pp_c);
3869 my_assert_not(pp, NULL);
ddaf8bd7 3870 }
3871 else if (po->datap != NULL) {
3872 pp = calloc(1, sizeof(*pp));
3873 my_assert_not(pp, NULL);
3874
3875 ret = parse_protostr(po->datap, pp);
3876 if (ret < 0)
3877 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
3878 free(po->datap);
3879 po->datap = NULL;
3880 }
3881
3882 if (pp != NULL) {
3883 if (pp->is_fptr)
3884 check_func_pp(po, pp, "fptr var call");
3885 if (pp->is_noreturn)
3886 po->flags |= OPF_TAIL;
7ae48d73 3887 }
092f64e1 3888 po->pp = pp;
7ae48d73 3889 continue;
3890 }
3891
3892 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
87bf6cec 3893 continue;
3894
4c45fa73 3895 if (po->operand[0].type == OPT_REGMEM) {
92d715b6 3896 pd = try_resolve_jumptab(i, opcnt);
4c45fa73 3897 if (pd == NULL)
840257f6 3898 goto tailcall;
4c45fa73 3899
3900 po->btj = pd;
3901 continue;
3902 }
3903
3904 for (l = 0; l < opcnt; l++) {
d7857c3a 3905 if (g_labels[l] != NULL
3906 && IS(po->operand[0].name, g_labels[l]))
3907 {
092f64e1 3908 if (l == i + 1 && po->op == OP_JMP) {
3909 // yet another alignment type..
5e49b270 3910 po->flags |= OPF_RMD|OPF_DONE;
092f64e1 3911 break;
3912 }
4c45fa73 3913 add_label_ref(&g_label_refs[l], i);
3914 po->bt_i = l;
87bf6cec 3915 break;
3916 }
3917 }
3918
092f64e1 3919 if (po->bt_i != -1 || (po->flags & OPF_RMD))
4c45fa73 3920 continue;
3921
840257f6 3922 if (po->operand[0].type == OPT_LABEL)
87bf6cec 3923 // assume tail call
840257f6 3924 goto tailcall;
4c45fa73 3925
3926 ferr(po, "unhandled branch\n");
840257f6 3927
3928tailcall:
3929 po->op = OP_CALL;
3930 po->flags |= OPF_TAIL;
2fe80fdb 3931 if (i > 0 && ops[i - 1].op == OP_POP)
3932 po->flags |= OPF_ATAIL;
7ae48d73 3933 i--; // reprocess
87bf6cec 3934 }
3935
3936 // pass3:
a2c1d768 3937 // - remove dead labels
26677139 3938 // - process trivial calls
1bafb621 3939 for (i = 0; i < opcnt; i++)
3940 {
d7857c3a 3941 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
3942 free(g_labels[i]);
3943 g_labels[i] = NULL;
3944 }
a2c1d768 3945
69a3cdfc 3946 po = &ops[i];
5e49b270 3947 if (po->flags & (OPF_RMD|OPF_DONE))
91977a1c 3948 continue;
850c9265 3949
d4e3b5db 3950 if (po->op == OP_CALL)
26677139 3951 {
3952 pp = process_call_early(i, opcnt, &j);
3953 if (pp != NULL) {
3954 if (!(po->flags & OPF_ATAIL))
3955 // since we know the args, try to collect them
3956 if (collect_call_args_early(po, i, pp, &regmask) != 0)
3957 pp = NULL;
3958 }
3959
3960 if (pp != NULL) {
3961 if (j >= 0) {
3962 // commit esp adjust
3963 ops[j].flags |= OPF_RMD;
5e49b270 3964 if (ops[j].op != OP_POP)
3965 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
3966 else
3967 ops[j].flags |= OPF_DONE;
26677139 3968 }
3969
3970 if (strstr(pp->ret_type.name, "int64"))
3971 need_tmp64 = 1;
3972
3973 po->flags |= OPF_DONE;
3974 }
3975 }
3976 }
3977
3978 // pass4:
3979 // - process calls
ee2361b9 3980 // - handle push <const>/pop pairs
26677139 3981 for (i = 0; i < opcnt; i++)
3982 {
3983 po = &ops[i];
5e49b270 3984 if (po->flags & (OPF_RMD|OPF_DONE))
26677139 3985 continue;
3986
3987 if (po->op == OP_CALL && !(po->flags & OPF_DONE))
69a3cdfc 3988 {
9af2d373 3989 pp = process_call(i, opcnt);
91977a1c 3990
2fe80fdb 3991 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
89ff3147 3992 // since we know the args, collect them
3a5101d7 3993 collect_call_args(po, i, pp, &regmask, save_arg_vars,
89ff3147 3994 i + opcnt * 2);
3995 }
2b43685d 3996
3997 if (strstr(pp->ret_type.name, "int64"))
2fe80fdb 3998 need_tmp64 = 1;
91977a1c 3999 }
ee2361b9 4000 else if (po->op == OP_PUSH && !(po->flags & OPF_FARG)
4001 && !(po->flags & OPF_RSAVE) && po->operand[0].type == OPT_CONST)
4002 scan_for_pop_const(i, opcnt);
d4e3b5db 4003 }
4004
26677139 4005 // pass5:
d4e3b5db 4006 // - find POPs for PUSHes, rm both
591721d7 4007 // - scan for STD/CLD, propagate DF
d4e3b5db 4008 // - scan for all used registers
4009 // - find flag set ops for their users
89ff3147 4010 // - do unreselved calls
1bafb621 4011 // - declare indirect functions
26677139 4012 for (i = 0; i < opcnt; i++)
4013 {
d4e3b5db 4014 po = &ops[i];
5e49b270 4015 if (po->flags & (OPF_RMD|OPF_DONE))
d4e3b5db 4016 continue;
4017
7ae48d73 4018 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE)) {
4019 reg = po->operand[0].reg;
4020 if (!(regmask & (1 << reg)))
4021 // not a reg save after all, rerun scan_for_pop
4022 po->flags &= ~OPF_RSAVE;
4023 else
4024 regmask_save |= 1 << reg;
4025 }
4026
9af2d373 4027 if (po->op == OP_PUSH && !(po->flags & OPF_FARG)
1f84f6b3 4028 && !(po->flags & OPF_RSAVE) && !g_func_pp->is_userstack)
d4e3b5db 4029 {
5c024ef7 4030 if (po->operand[0].type == OPT_REG)
4031 {
4032 reg = po->operand[0].reg;
4033 if (reg < 0)
4034 ferr(po, "reg not set for push?\n");
4035
4036 depth = 0;
4037 ret = scan_for_pop(i + 1, opcnt,
4038 po->operand[0].name, i + opcnt * 3, 0, &depth, 0);
4039 if (ret == 1) {
4040 if (depth > 1)
4041 ferr(po, "too much depth: %d\n", depth);
4042
4043 po->flags |= OPF_RMD;
4044 scan_for_pop(i + 1, opcnt, po->operand[0].name,
4045 i + opcnt * 4, 0, &depth, 1);
4046 continue;
4047 }
4048 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
4049 if (ret == 0) {
4050 arg = OPF_RMD;
4051 if (regmask & (1 << reg)) {
4052 if (regmask_save & (1 << reg))
4053 ferr(po, "%s already saved?\n", po->operand[0].name);
4054 arg = OPF_RSAVE;
4055 }
4056 po->flags |= arg;
4057 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg);
4058 continue;
4059 }
d4e3b5db 4060 }
d4e3b5db 4061 }
4062
591721d7 4063 if (po->op == OP_STD) {
5e49b270 4064 po->flags |= OPF_DF | OPF_RMD | OPF_DONE;
591721d7 4065 scan_propagate_df(i + 1, opcnt);
4066 }
4067
a3684be1 4068 regmask_now = po->regmask_src | po->regmask_dst;
4069 if (regmask_now & (1 << xBP)) {
4070 if (g_bp_frame && !(po->flags & OPF_EBP_S)) {
4071 if (po->regmask_dst & (1 << xBP))
4072 // compiler decided to drop bp frame and use ebp as scratch
037f4971 4073 scan_fwd_set_flags(i + 1, opcnt, i + opcnt * 5, OPF_EBP_S);
a3684be1 4074 else
4075 regmask_now &= ~(1 << xBP);
4076 }
4077 }
4078
4079 regmask |= regmask_now;
d4e3b5db 4080
4081 if (po->flags & OPF_CC)
4082 {
2b43685d 4083 int setters[16], cnt = 0, branched = 0;
4084
04f8a628 4085 ret = scan_for_flag_set(i, i + opcnt * 6,
4086 &branched, setters, &cnt);
2b43685d 4087 if (ret < 0 || cnt <= 0)
4088 ferr(po, "unable to trace flag setter(s)\n");
4089 if (cnt > ARRAY_SIZE(setters))
4090 ferr(po, "too many flag setters\n");
d4e3b5db 4091
2b43685d 4092 for (j = 0; j < cnt; j++)
4093 {
4094 tmp_op = &ops[setters[j]]; // flag setter
4095 pfomask = 0;
4096
4097 // to get nicer code, we try to delay test and cmp;
4098 // if we can't because of operand modification, or if we
591721d7 4099 // have arith op, or branch, make it calculate flags explicitly
4100 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP)
4101 {
89ff3147 4102 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i, 0) >= 0)
092f64e1 4103 pfomask = 1 << po->pfo;
2b43685d 4104 }
4741fdfe 4105 else if (tmp_op->op == OP_CMPS || tmp_op->op == OP_SCAS) {
092f64e1 4106 pfomask = 1 << po->pfo;
591721d7 4107 }
2b43685d 4108 else {
04f8a628 4109 // see if we'll be able to handle based on op result
4110 if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR
092f64e1 4111 && po->pfo != PFO_Z && po->pfo != PFO_S
4112 && po->pfo != PFO_P)
04f8a628 4113 || branched
2b43685d 4114 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
092f64e1 4115 {
4116 pfomask = 1 << po->pfo;
4117 }
2fe80fdb 4118
c8dbc5be 4119 if (tmp_op->op == OP_ADD && po->pfo == PFO_C) {
4120 propagate_lmod(tmp_op, &tmp_op->operand[0],
4121 &tmp_op->operand[1]);
4122 if (tmp_op->operand[0].lmod == OPLM_DWORD)
4123 need_tmp64 = 1;
4124 }
2b43685d 4125 }
4126 if (pfomask) {
4127 tmp_op->pfomask |= pfomask;
cb090db0 4128 cond_vars |= pfomask;
2b43685d 4129 }
04f8a628 4130 // note: may overwrite, currently not a problem
4131 po->datap = tmp_op;
d4e3b5db 4132 }
4133
cb090db0 4134 if (po->op == OP_RCL || po->op == OP_RCR
4135 || po->op == OP_ADC || po->op == OP_SBB)
4136 cond_vars |= 1 << PFO_C;
d4e3b5db 4137 }
092f64e1 4138
4139 if (po->op == OP_CMPS || po->op == OP_SCAS) {
cb090db0 4140 cond_vars |= 1 << PFO_Z;
591721d7 4141 }
87bf6cec 4142 else if (po->op == OP_MUL
4143 || (po->op == OP_IMUL && po->operand_cnt == 1))
4144 {
c8dbc5be 4145 if (po->operand[0].lmod == OPLM_DWORD)
4146 need_tmp64 = 1;
87bf6cec 4147 }
89ff3147 4148 else if (po->op == OP_CALL) {
26677139 4149 // note: resolved non-reg calls are OPF_DONE already
092f64e1 4150 pp = po->pp;
89ff3147 4151 if (pp == NULL)
4152 ferr(po, "NULL pp\n");
4153
4154 if (pp->is_unresolved) {
ddaf8bd7 4155 int regmask_stack = 0;
3a5101d7 4156 collect_call_args(po, i, pp, &regmask, save_arg_vars,
89ff3147 4157 i + opcnt * 2);
4158
b74c31e3 4159 // this is pretty rough guess:
4160 // see ecx and edx were pushed (and not their saved versions)
4161 for (arg = 0; arg < pp->argc; arg++) {
4162 if (pp->arg[arg].reg != NULL)
4163 continue;
4164
4165 tmp_op = pp->arg[arg].datap;
4166 if (tmp_op == NULL)
4167 ferr(po, "parsed_op missing for arg%d\n", arg);
5f70a34f 4168 if (tmp_op->p_argnum == 0 && tmp_op->operand[0].type == OPT_REG)
b74c31e3 4169 regmask_stack |= 1 << tmp_op->operand[0].reg;
4170 }
4171
ddaf8bd7 4172 if (!((regmask_stack & (1 << xCX))
4173 && (regmask_stack & (1 << xDX))))
89ff3147 4174 {
4175 if (pp->argc_stack != 0
c0050df6 4176 || ((regmask | regmask_arg) & ((1 << xCX)|(1 << xDX))))
89ff3147 4177 {
4178 pp_insert_reg_arg(pp, "ecx");
c0050df6 4179 pp->is_fastcall = 1;
ddaf8bd7 4180 regmask_init |= 1 << xCX;
89ff3147 4181 regmask |= 1 << xCX;
4182 }
4183 if (pp->argc_stack != 0
4184 || ((regmask | regmask_arg) & (1 << xDX)))
4185 {
4186 pp_insert_reg_arg(pp, "edx");
ddaf8bd7 4187 regmask_init |= 1 << xDX;
89ff3147 4188 regmask |= 1 << xDX;
4189 }
4190 }
c0050df6 4191
4192 // note: __cdecl doesn't fall into is_unresolved category
4193 if (pp->argc_stack > 0)
4194 pp->is_stdcall = 1;
ddaf8bd7 4195 }
4196
4197 for (arg = 0; arg < pp->argc; arg++) {
4198 if (pp->arg[arg].reg != NULL) {
4199 reg = char_array_i(regs_r32,
4200 ARRAY_SIZE(regs_r32), pp->arg[arg].reg);
4201 if (reg < 0)
4202 ferr(ops, "arg '%s' is not a reg?\n", pp->arg[arg].reg);
4203 if (!(regmask & (1 << reg))) {
4204 regmask_init |= 1 << reg;
4205 regmask |= 1 << reg;
4206 }
4207 }
89ff3147 4208 }
1bafb621 4209 }
27ebfaed 4210 else if (po->op == OP_MOV && po->operand[0].pp != NULL
4211 && po->operand[1].pp != NULL)
4212 {
4213 // <var> = offset <something>
4214 if ((po->operand[1].pp->is_func || po->operand[1].pp->is_fptr)
4215 && !IS_START(po->operand[1].name, "off_"))
4216 {
4217 if (!po->operand[0].pp->is_fptr)
4218 ferr(po, "%s not declared as fptr when it should be\n",
4219 po->operand[0].name);
4220 if (pp_cmp_func(po->operand[0].pp, po->operand[1].pp)) {
4221 pp_print(buf1, sizeof(buf1), po->operand[0].pp);
4222 pp_print(buf2, sizeof(buf2), po->operand[1].pp);
4223 fnote(po, "var: %s\n", buf1);
4224 fnote(po, "func: %s\n", buf2);
4225 ferr(po, "^ mismatch\n");
4226 }
4227 }
4228 }
75ad0378 4229 else if (po->op == OP_RET && !IS(g_func_pp->ret_type.name, "void"))
4230 regmask |= 1 << xAX;
cb090db0 4231 else if (po->op == OP_DIV || po->op == OP_IDIV) {
4232 // 32bit division is common, look for it
4233 if (po->op == OP_DIV)
4234 ret = scan_for_reg_clear(i, xDX);
4235 else
4236 ret = scan_for_cdq_edx(i);
4237 if (ret >= 0)
4238 po->flags |= OPF_32BIT;
4239 else
4240 need_tmp64 = 1;
4241 }
037f4971 4242 else if (po->op == OP_CLD)
5e49b270 4243 po->flags |= OPF_RMD | OPF_DONE;
cb090db0 4244
4245 if (po->op == OP_RCL || po->op == OP_RCR || po->op == OP_XCHG) {
4246 need_tmp_var = 1;
4247 }
91977a1c 4248 }
4249
26677139 4250 // pass6:
da87ae38 4251 // - confirm regmask_save, it might have been reduced
4252 if (regmask_save != 0)
4253 {
4254 regmask_save = 0;
4255 for (i = 0; i < opcnt; i++) {
4256 po = &ops[i];
4257 if (po->flags & OPF_RMD)
4258 continue;
4259
4260 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE))
4261 regmask_save |= 1 << po->operand[0].reg;
4262 }
4263 }
4264
60fe410c 4265 // output starts here
4266
4267 // define userstack size
4268 if (g_func_pp->is_userstack) {
4269 fprintf(fout, "#ifndef US_SZ_%s\n", g_func_pp->name);
4270 fprintf(fout, "#define US_SZ_%s USERSTACK_SIZE\n", g_func_pp->name);
4271 fprintf(fout, "#endif\n");
4272 }
4273
4274 // the function itself
c0de9015 4275 ferr_assert(ops, !g_func_pp->is_fptr);
4276 output_pp(fout, g_func_pp,
4277 (g_ida_func_attr & IDAFA_NORETURN) ? OPP_FORCE_NORETURN : 0);
4278 fprintf(fout, "\n{\n");
60fe410c 4279
4280 // declare indirect functions
4281 for (i = 0; i < opcnt; i++) {
4282 po = &ops[i];
4283 if (po->flags & OPF_RMD)
4284 continue;
4285
4286 if (po->op == OP_CALL) {
4287 pp = po->pp;
4288 if (pp == NULL)
4289 ferr(po, "NULL pp\n");
4290
4291 if (pp->is_fptr && !(pp->name[0] != 0 && pp->is_arg)) {
4292 if (pp->name[0] != 0) {
4293 memmove(pp->name + 2, pp->name, strlen(pp->name) + 1);
4294 memcpy(pp->name, "i_", 2);
4295
4296 // might be declared already
4297 found = 0;
4298 for (j = 0; j < i; j++) {
4299 if (ops[j].op == OP_CALL && (pp_tmp = ops[j].pp)) {
4300 if (pp_tmp->is_fptr && IS(pp->name, pp_tmp->name)) {
4301 found = 1;
4302 break;
4303 }
4304 }
4305 }
4306 if (found)
4307 continue;
4308 }
4309 else
4310 snprintf(pp->name, sizeof(pp->name), "icall%d", i);
4311
c0de9015 4312 fprintf(fout, " ");
4313 output_pp(fout, pp, OPP_SIMPLE_ARGS);
4314 fprintf(fout, ";\n");
60fe410c 4315 }
4316 }
4317 }
da87ae38 4318
4c45fa73 4319 // output LUTs/jumptables
4320 for (i = 0; i < g_func_pd_cnt; i++) {
4321 pd = &g_func_pd[i];
4322 fprintf(fout, " static const ");
4323 if (pd->type == OPT_OFFSET) {
4324 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
4325
4326 for (j = 0; j < pd->count; j++) {
4327 if (j > 0)
4328 fprintf(fout, ", ");
4329 fprintf(fout, "&&%s", pd->d[j].u.label);
4330 }
4331 }
4332 else {
4333 fprintf(fout, "%s %s[] =\n { ",
4334 lmod_type_u(ops, pd->lmod), pd->label);
4335
4336 for (j = 0; j < pd->count; j++) {
4337 if (j > 0)
4338 fprintf(fout, ", ");
4339 fprintf(fout, "%u", pd->d[j].u.val);
4340 }
4341 }
4342 fprintf(fout, " };\n");
1f84f6b3 4343 had_decl = 1;
4c45fa73 4344 }
4345
4f12f671 4346 // declare stack frame, va_arg
1f84f6b3 4347 if (g_stack_fsz) {
850c9265 4348 fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
1bafb621 4349 (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz);
1f84f6b3 4350 had_decl = 1;
4351 }
4352
4353 if (g_func_pp->is_userstack) {
60fe410c 4354 fprintf(fout, " u32 fake_sf[US_SZ_%s / 4];\n", g_func_pp->name);
4355 fprintf(fout, " u32 *esp = &fake_sf[sizeof(fake_sf) / 4];\n");
1f84f6b3 4356 had_decl = 1;
4357 }
850c9265 4358
1f84f6b3 4359 if (g_func_pp->is_vararg) {
4f12f671 4360 fprintf(fout, " va_list ap;\n");
1f84f6b3 4361 had_decl = 1;
4362 }
4f12f671 4363
940e8e66 4364 // declare arg-registers
bd96f656 4365 for (i = 0; i < g_func_pp->argc; i++) {
4366 if (g_func_pp->arg[i].reg != NULL) {
91977a1c 4367 reg = char_array_i(regs_r32,
bd96f656 4368 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
75ad0378 4369 if (regmask & (1 << reg)) {
1f84f6b3 4370 if (g_func_pp->arg[i].type.is_retreg)
4371 fprintf(fout, " u32 %s = *r_%s;\n",
4372 g_func_pp->arg[i].reg, g_func_pp->arg[i].reg);
4373 else
4374 fprintf(fout, " u32 %s = (u32)a%d;\n",
4375 g_func_pp->arg[i].reg, i + 1);
75ad0378 4376 }
1f84f6b3 4377 else {
4378 if (g_func_pp->arg[i].type.is_retreg)
4379 ferr(ops, "retreg '%s' is unused?\n",
4380 g_func_pp->arg[i].reg);
75ad0378 4381 fprintf(fout, " // %s = a%d; // unused\n",
4382 g_func_pp->arg[i].reg, i + 1);
1f84f6b3 4383 }
91977a1c 4384 had_decl = 1;
4385 }
4386 }
4387
75ad0378 4388 regmask_now = regmask & ~regmask_arg;
4389 regmask_now &= ~(1 << xSP);
90307a99 4390 if (regmask_now & 0x00ff) {
91977a1c 4391 for (reg = 0; reg < 8; reg++) {
75ad0378 4392 if (regmask_now & (1 << reg)) {
ddaf8bd7 4393 fprintf(fout, " u32 %s", regs_r32[reg]);
4394 if (regmask_init & (1 << reg))
4395 fprintf(fout, " = 0");
4396 fprintf(fout, ";\n");
91977a1c 4397 had_decl = 1;
4398 }
4399 }
4400 }
90307a99 4401 if (regmask_now & 0xff00) {
4402 for (reg = 8; reg < 16; reg++) {
4403 if (regmask_now & (1 << reg)) {
4404 fprintf(fout, " mmxr %s", regs_r32[reg]);
4405 if (regmask_init & (1 << reg))
4406 fprintf(fout, " = { 0, }");
4407 fprintf(fout, ";\n");
4408 had_decl = 1;
4409 }
4410 }
4411 }
91977a1c 4412
d4e3b5db 4413 if (regmask_save) {
4414 for (reg = 0; reg < 8; reg++) {
4415 if (regmask_save & (1 << reg)) {
4416 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
4417 had_decl = 1;
4418 }
4419 }
4420 }
4421
3a5101d7 4422 for (i = 0; i < ARRAY_SIZE(save_arg_vars); i++) {
4423 if (save_arg_vars[i] == 0)
4424 continue;
69a3cdfc 4425 for (reg = 0; reg < 32; reg++) {
3a5101d7 4426 if (save_arg_vars[i] & (1 << reg)) {
4427 fprintf(fout, " u32 %s;\n",
4428 saved_arg_name(buf1, sizeof(buf1), i, reg + 1));
69a3cdfc 4429 had_decl = 1;
4430 }
4431 }
4432 }
4433
cb090db0 4434 if (cond_vars) {
69a3cdfc 4435 for (i = 0; i < 8; i++) {
cb090db0 4436 if (cond_vars & (1 << i)) {
69a3cdfc 4437 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
4438 had_decl = 1;
4439 }
4440 }
4441 }
4442
108e9fe3 4443 if (need_tmp_var) {
4444 fprintf(fout, " u32 tmp;\n");
4445 had_decl = 1;
4446 }
4447
2fe80fdb 4448 if (need_tmp64) {
4449 fprintf(fout, " u64 tmp64;\n");
87bf6cec 4450 had_decl = 1;
4451 }
4452
91977a1c 4453 if (had_decl)
4454 fprintf(fout, "\n");
4455
bd96f656 4456 if (g_func_pp->is_vararg) {
4457 if (g_func_pp->argc_stack == 0)
4f12f671 4458 ferr(ops, "vararg func without stack args?\n");
bd96f656 4459 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
4f12f671 4460 }
4461
91977a1c 4462 // output ops
69a3cdfc 4463 for (i = 0; i < opcnt; i++)
4464 {
d7857c3a 4465 if (g_labels[i] != NULL) {
91977a1c 4466 fprintf(fout, "\n%s:\n", g_labels[i]);
3ebea2cf 4467 label_pending = 1;
2b43685d 4468
4469 delayed_flag_op = NULL;
4470 last_arith_dst = NULL;
3ebea2cf 4471 }
91977a1c 4472
69a3cdfc 4473 po = &ops[i];
4474 if (po->flags & OPF_RMD)
91977a1c 4475 continue;
4476
4477 no_output = 0;
4478
91977a1c 4479 #define assert_operand_cnt(n_) \
850c9265 4480 if (po->operand_cnt != n_) \
4481 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
4482
69a3cdfc 4483 // conditional/flag using op?
4484 if (po->flags & OPF_CC)
850c9265 4485 {
940e8e66 4486 int is_delayed = 0;
69a3cdfc 4487
04f8a628 4488 tmp_op = po->datap;
850c9265 4489
69a3cdfc 4490 // we go through all this trouble to avoid using parsed_flag_op,
4491 // which makes generated code much nicer
4492 if (delayed_flag_op != NULL)
850c9265 4493 {
092f64e1 4494 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op,
4495 po->pfo, po->pfo_inv);
940e8e66 4496 is_delayed = 1;
91977a1c 4497 }
850c9265 4498 else if (last_arith_dst != NULL
092f64e1 4499 && (po->pfo == PFO_Z || po->pfo == PFO_S || po->pfo == PFO_P
04f8a628 4500 || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR))
4501 ))
850c9265 4502 {
3ebea2cf 4503 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
092f64e1 4504 out_test_for_cc(buf1, sizeof(buf1), po, po->pfo, po->pfo_inv,
850c9265 4505 last_arith_dst->lmod, buf3);
940e8e66 4506 is_delayed = 1;
850c9265 4507 }
04f8a628 4508 else if (tmp_op != NULL) {
7ba45c34 4509 // use preprocessed flag calc results
092f64e1 4510 if (!(tmp_op->pfomask & (1 << po->pfo)))
4511 ferr(po, "not prepared for pfo %d\n", po->pfo);
69a3cdfc 4512
092f64e1 4513 // note: pfo_inv was not yet applied
69a3cdfc 4514 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
092f64e1 4515 po->pfo_inv ? "!" : "", parsed_flag_op_names[po->pfo]);
69a3cdfc 4516 }
4517 else {
4518 ferr(po, "all methods of finding comparison failed\n");
4519 }
850c9265 4520
69a3cdfc 4521 if (po->flags & OPF_JMP) {
092f64e1 4522 fprintf(fout, " if %s", buf1);
850c9265 4523 }
cb090db0 4524 else if (po->op == OP_RCL || po->op == OP_RCR
4525 || po->op == OP_ADC || po->op == OP_SBB)
4526 {
940e8e66 4527 if (is_delayed)
4528 fprintf(fout, " cond_%s = %s;\n",
092f64e1 4529 parsed_flag_op_names[po->pfo], buf1);
850c9265 4530 }
5101a5f9 4531 else if (po->flags & OPF_DATA) { // SETcc
850c9265 4532 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
4533 fprintf(fout, " %s = %s;", buf2, buf1);
91977a1c 4534 }
69a3cdfc 4535 else {
4536 ferr(po, "unhandled conditional op\n");
4537 }
91977a1c 4538 }
4539
940e8e66 4540 pfomask = po->pfomask;
4541
4741fdfe 4542 if (po->flags & (OPF_REPZ|OPF_REPNZ)) {
1f84f6b3 4543 struct parsed_opr opr = {0,};
4544 opr.type = OPT_REG;
4545 opr.reg = xCX;
4546 opr.lmod = OPLM_DWORD;
4547 ret = try_resolve_const(i, &opr, opcnt * 7 + i, &uval);
4548
4549 if (ret != 1 || uval == 0) {
4550 // we need initial flags for ecx=0 case..
4551 if (i > 0 && ops[i - 1].op == OP_XOR
4552 && IS(ops[i - 1].operand[0].name,
4553 ops[i - 1].operand[1].name))
4554 {
4555 fprintf(fout, " cond_z = ");
4556 if (pfomask & (1 << PFO_C))
4557 fprintf(fout, "cond_c = ");
4558 fprintf(fout, "0;\n");
4559 }
4560 else if (last_arith_dst != NULL) {
4561 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
4562 out_test_for_cc(buf1, sizeof(buf1), po, PFO_Z, 0,
4563 last_arith_dst->lmod, buf3);
4564 fprintf(fout, " cond_z = %s;\n", buf1);
4565 }
4566 else
4567 ferr(po, "missing initial ZF\n");
4741fdfe 4568 }
4741fdfe 4569 }
4570
850c9265 4571 switch (po->op)
91977a1c 4572 {
4573 case OP_MOV:
4574 assert_operand_cnt(2);
850c9265 4575 propagate_lmod(po, &po->operand[0], &po->operand[1]);
de50b98b 4576 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
c7ed83dd 4577 default_cast_to(buf3, sizeof(buf3), &po->operand[0]);
de50b98b 4578 fprintf(fout, " %s = %s;", buf1,
3ebea2cf 4579 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
c7ed83dd 4580 buf3, 0));
850c9265 4581 break;
4582
4583 case OP_LEA:
4584 assert_operand_cnt(2);
87bf6cec 4585 po->operand[1].lmod = OPLM_DWORD; // always
850c9265 4586 fprintf(fout, " %s = %s;",
4587 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3ebea2cf 4588 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4589 NULL, 1));
850c9265 4590 break;
4591
4592 case OP_MOVZX:
4593 assert_operand_cnt(2);
91977a1c 4594 fprintf(fout, " %s = %s;",
850c9265 4595 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3ebea2cf 4596 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
850c9265 4597 break;
4598
4599 case OP_MOVSX:
4600 assert_operand_cnt(2);
4601 switch (po->operand[1].lmod) {
4602 case OPLM_BYTE:
4603 strcpy(buf3, "(s8)");
4604 break;
4605 case OPLM_WORD:
4606 strcpy(buf3, "(s16)");
4607 break;
4608 default:
4609 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
4610 }
a2c1d768 4611 fprintf(fout, " %s = %s;",
850c9265 4612 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
a2c1d768 4613 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4614 buf3, 0));
850c9265 4615 break;
4616
108e9fe3 4617 case OP_XCHG:
4618 assert_operand_cnt(2);
4619 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4620 fprintf(fout, " tmp = %s;",
4621 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], "", 0));
4622 fprintf(fout, " %s = %s;",
4623 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
c7ed83dd 4624 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4625 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
4626 fprintf(fout, " %s = %stmp;",
4627 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]),
4628 default_cast_to(buf3, sizeof(buf3), &po->operand[1]));
108e9fe3 4629 snprintf(g_comment, sizeof(g_comment), "xchg");
4630 break;
4631
850c9265 4632 case OP_NOT:
4633 assert_operand_cnt(1);
4634 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4635 fprintf(fout, " %s = ~%s;", buf1, buf1);
4636 break;
4637
5101a5f9 4638 case OP_CDQ:
4639 assert_operand_cnt(2);
4640 fprintf(fout, " %s = (s32)%s >> 31;",
4641 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3ebea2cf 4642 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5101a5f9 4643 strcpy(g_comment, "cdq");
4644 break;
4645
092f64e1 4646 case OP_LODS:
4647 assert_operand_cnt(3);
4648 if (po->flags & OPF_REP) {
4649 // hmh..
4650 ferr(po, "TODO\n");
4651 }
4652 else {
05381e4a 4653 fprintf(fout, " eax = %sesi; esi %c= %d;",
092f64e1 4654 lmod_cast_u_ptr(po, po->operand[0].lmod),
4655 (po->flags & OPF_DF) ? '-' : '+',
4656 lmod_bytes(po, po->operand[0].lmod));
4657 strcpy(g_comment, "lods");
4658 }
4659 break;
4660
33c35af6 4661 case OP_STOS:
33c35af6 4662 assert_operand_cnt(3);
4663 if (po->flags & OPF_REP) {
591721d7 4664 fprintf(fout, " for (; ecx != 0; ecx--, edi %c= %d)\n",
4665 (po->flags & OPF_DF) ? '-' : '+',
33c35af6 4666 lmod_bytes(po, po->operand[0].lmod));
d4e3b5db 4667 fprintf(fout, " %sedi = eax;",
4668 lmod_cast_u_ptr(po, po->operand[0].lmod));
33c35af6 4669 strcpy(g_comment, "rep stos");
4670 }
4671 else {
092f64e1 4672 fprintf(fout, " %sedi = eax; edi %c= %d;",
d4e3b5db 4673 lmod_cast_u_ptr(po, po->operand[0].lmod),
591721d7 4674 (po->flags & OPF_DF) ? '-' : '+',
33c35af6 4675 lmod_bytes(po, po->operand[0].lmod));
4676 strcpy(g_comment, "stos");
4677 }
4678 break;
4679
d4e3b5db 4680 case OP_MOVS:
d4e3b5db 4681 assert_operand_cnt(3);
4682 j = lmod_bytes(po, po->operand[0].lmod);
4683 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
591721d7 4684 l = (po->flags & OPF_DF) ? '-' : '+';
d4e3b5db 4685 if (po->flags & OPF_REP) {
4686 fprintf(fout,
591721d7 4687 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n",
4688 l, j, l, j);
d4e3b5db 4689 fprintf(fout,
4690 " %sedi = %sesi;", buf1, buf1);
4691 strcpy(g_comment, "rep movs");
4692 }
4693 else {
092f64e1 4694 fprintf(fout, " %sedi = %sesi; edi %c= %d; esi %c= %d;",
591721d7 4695 buf1, buf1, l, j, l, j);
d4e3b5db 4696 strcpy(g_comment, "movs");
4697 }
4698 break;
4699
7ba45c34 4700 case OP_CMPS:
7ba45c34 4701 // repe ~ repeat while ZF=1
4702 assert_operand_cnt(3);
4703 j = lmod_bytes(po, po->operand[0].lmod);
4704 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
591721d7 4705 l = (po->flags & OPF_DF) ? '-' : '+';
7ba45c34 4706 if (po->flags & OPF_REP) {
4707 fprintf(fout,
1f84f6b3 4708 " for (; ecx != 0; ecx--) {\n");
4741fdfe 4709 if (pfomask & (1 << PFO_C)) {
4710 // ugh..
4711 fprintf(fout,
05381e4a 4712 " cond_c = %sesi < %sedi;\n", buf1, buf1);
4741fdfe 4713 pfomask &= ~(1 << PFO_C);
4714 }
7ba45c34 4715 fprintf(fout,
05381e4a 4716 " cond_z = (%sesi == %sedi); esi %c= %d, edi %c= %d;\n",
1f84f6b3 4717 buf1, buf1, l, j, l, j);
4718 fprintf(fout,
4719 " if (cond_z %s 0) break;\n",
4720 (po->flags & OPF_REPZ) ? "==" : "!=");
7ba45c34 4721 fprintf(fout,
4741fdfe 4722 " }");
7ba45c34 4723 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
4724 (po->flags & OPF_REPZ) ? "e" : "ne");
4725 }
4726 else {
4727 fprintf(fout,
05381e4a 4728 " cond_z = (%sesi == %sedi); esi %c= %d; edi %c= %d;",
591721d7 4729 buf1, buf1, l, j, l, j);
7ba45c34 4730 strcpy(g_comment, "cmps");
4731 }
4732 pfomask &= ~(1 << PFO_Z);
4733 last_arith_dst = NULL;
4734 delayed_flag_op = NULL;
4735 break;
4736
591721d7 4737 case OP_SCAS:
4738 // only does ZF (for now)
4739 // repe ~ repeat while ZF=1
4740 assert_operand_cnt(3);
4741 j = lmod_bytes(po, po->operand[0].lmod);
4742 l = (po->flags & OPF_DF) ? '-' : '+';
4743 if (po->flags & OPF_REP) {
4744 fprintf(fout,
1f84f6b3 4745 " for (; ecx != 0; ecx--) {\n");
591721d7 4746 fprintf(fout,
1f84f6b3 4747 " cond_z = (%seax == %sedi); edi %c= %d;\n",
591721d7 4748 lmod_cast_u(po, po->operand[0].lmod),
1f84f6b3 4749 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
4750 fprintf(fout,
4751 " if (cond_z %s 0) break;\n",
591721d7 4752 (po->flags & OPF_REPZ) ? "==" : "!=");
4753 fprintf(fout,
1f84f6b3 4754 " }");
591721d7 4755 snprintf(g_comment, sizeof(g_comment), "rep%s scas",
4756 (po->flags & OPF_REPZ) ? "e" : "ne");
4757 }
4758 else {
05381e4a 4759 fprintf(fout, " cond_z = (%seax == %sedi); edi %c= %d;",
591721d7 4760 lmod_cast_u(po, po->operand[0].lmod),
4761 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
4762 strcpy(g_comment, "scas");
4763 }
4764 pfomask &= ~(1 << PFO_Z);
4765 last_arith_dst = NULL;
4766 delayed_flag_op = NULL;
4767 break;
4768
850c9265 4769 // arithmetic w/flags
850c9265 4770 case OP_AND:
4771 case OP_OR:
5101a5f9 4772 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4773 // fallthrough
850c9265 4774 dualop_arith:
4775 assert_operand_cnt(2);
850c9265 4776 fprintf(fout, " %s %s= %s;",
4777 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4778 op_to_c(po),
3ebea2cf 4779 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
04f8a628 4780 output_std_flags(fout, po, &pfomask, buf1);
4781 last_arith_dst = &po->operand[0];
4782 delayed_flag_op = NULL;
4783 break;
4784
4785 case OP_SHL:
4786 case OP_SHR:
4787 assert_operand_cnt(2);
4788 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4789 if (pfomask & (1 << PFO_C)) {
4790 if (po->operand[1].type == OPT_CONST) {
4791 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4792 j = po->operand[1].val;
4793 j %= l;
4794 if (j != 0) {
4795 if (po->op == OP_SHL)
4796 j = l - j;
4797 else
4798 j -= 1;
cb090db0 4799 fprintf(fout, " cond_c = (%s >> %d) & 1;\n",
4800 buf1, j);
04f8a628 4801 }
4802 else
4803 ferr(po, "zero shift?\n");
4804 }
4805 else
4806 ferr(po, "TODO\n");
4807 pfomask &= ~(1 << PFO_C);
840257f6 4808 }
04f8a628 4809 fprintf(fout, " %s %s= %s;", buf1, op_to_c(po),
4810 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4811 output_std_flags(fout, po, &pfomask, buf1);
850c9265 4812 last_arith_dst = &po->operand[0];
69a3cdfc 4813 delayed_flag_op = NULL;
850c9265 4814 break;
4815
d4e3b5db 4816 case OP_SAR:
4817 assert_operand_cnt(2);
4818 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4819 fprintf(fout, " %s = %s%s >> %s;", buf1,
4820 lmod_cast_s(po, po->operand[0].lmod), buf1,
3ebea2cf 4821 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
04f8a628 4822 output_std_flags(fout, po, &pfomask, buf1);
d4e3b5db 4823 last_arith_dst = &po->operand[0];
4824 delayed_flag_op = NULL;
4825 break;
4826
3b2f4044 4827 case OP_SHRD:
4828 assert_operand_cnt(3);
4829 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4830 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4831 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4832 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4833 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[2]);
4834 fprintf(fout, " %s >>= %s; %s |= %s << (%d - %s);",
4835 buf1, buf3, buf1, buf2, l, buf3);
4836 strcpy(g_comment, "shrd");
4837 output_std_flags(fout, po, &pfomask, buf1);
4838 last_arith_dst = &po->operand[0];
4839 delayed_flag_op = NULL;
4840 break;
4841
d4e3b5db 4842 case OP_ROL:
4843 case OP_ROR:
4844 assert_operand_cnt(2);
4845 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4846 if (po->operand[1].type == OPT_CONST) {
4847 j = po->operand[1].val;
4848 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
4849 fprintf(fout, po->op == OP_ROL ?
4850 " %s = (%s << %d) | (%s >> %d);" :
4851 " %s = (%s >> %d) | (%s << %d);",
4852 buf1, buf1, j, buf1,
4853 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
4854 }
4855 else
4856 ferr(po, "TODO\n");
04f8a628 4857 output_std_flags(fout, po, &pfomask, buf1);
d4e3b5db 4858 last_arith_dst = &po->operand[0];
4859 delayed_flag_op = NULL;
4860 break;
4861
cb090db0 4862 case OP_RCL:
4863 case OP_RCR:
4864 assert_operand_cnt(2);
4865 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4866 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4867 if (po->operand[1].type == OPT_CONST) {
4868 j = po->operand[1].val % l;
4869 if (j == 0)
4870 ferr(po, "zero rotate\n");
4871 fprintf(fout, " tmp = (%s >> %d) & 1;\n",
4872 buf1, (po->op == OP_RCL) ? (l - j) : (j - 1));
4873 if (po->op == OP_RCL) {
4874 fprintf(fout,
4875 " %s = (%s << %d) | (cond_c << %d)",
4876 buf1, buf1, j, j - 1);
4877 if (j != 1)
4878 fprintf(fout, " | (%s >> %d)", buf1, l + 1 - j);
4879 }
4880 else {
4881 fprintf(fout,
4882 " %s = (%s >> %d) | (cond_c << %d)",
4883 buf1, buf1, j, l - j);
4884 if (j != 1)
4885 fprintf(fout, " | (%s << %d)", buf1, l + 1 - j);
4886 }
4887 fprintf(fout, ";\n");
4888 fprintf(fout, " cond_c = tmp;");
4889 }
4890 else
4891 ferr(po, "TODO\n");
4892 strcpy(g_comment, (po->op == OP_RCL) ? "rcl" : "rcr");
4893 output_std_flags(fout, po, &pfomask, buf1);
4894 last_arith_dst = &po->operand[0];
4895 delayed_flag_op = NULL;
4896 break;
4897
5101a5f9 4898 case OP_XOR:
850c9265 4899 assert_operand_cnt(2);
4900 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5101a5f9 4901 if (IS(opr_name(po, 0), opr_name(po, 1))) {
4902 // special case for XOR
2fe80fdb 4903 if (pfomask & (1 << PFO_BE)) { // weird, but it happens..
4904 fprintf(fout, " cond_be = 1;\n");
4905 pfomask &= ~(1 << PFO_BE);
4906 }
5101a5f9 4907 fprintf(fout, " %s = 0;",
4908 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4909 last_arith_dst = &po->operand[0];
4910 delayed_flag_op = NULL;
850c9265 4911 break;
850c9265 4912 }
5101a5f9 4913 goto dualop_arith;
4914
2fe80fdb 4915 case OP_ADD:
4916 assert_operand_cnt(2);
4917 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4918 if (pfomask & (1 << PFO_C)) {
c8dbc5be 4919 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4920 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4921 if (po->operand[0].lmod == OPLM_DWORD) {
4922 fprintf(fout, " tmp64 = (u64)%s + %s;\n", buf1, buf2);
4923 fprintf(fout, " cond_c = tmp64 >> 32;\n");
4924 fprintf(fout, " %s = (u32)tmp64;",
4925 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4926 strcat(g_comment, "add64");
4927 }
4928 else {
4929 fprintf(fout, " cond_c = ((u32)%s + %s) >> %d;\n",
4930 buf1, buf2, lmod_bytes(po, po->operand[0].lmod) * 8);
4931 fprintf(fout, " %s += %s;",
4932 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4933 buf2);
4934 }
2fe80fdb 4935 pfomask &= ~(1 << PFO_C);
4936 output_std_flags(fout, po, &pfomask, buf1);
4937 last_arith_dst = &po->operand[0];
4938 delayed_flag_op = NULL;
4939 break;
4940 }
4941 goto dualop_arith;
4942
4943 case OP_SUB:
4944 assert_operand_cnt(2);
4945 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3b2f4044 4946 if (pfomask & ~((1 << PFO_Z) | (1 << PFO_S))) {
4947 for (j = 0; j <= PFO_LE; j++) {
4948 if (!(pfomask & (1 << j)))
4949 continue;
4950 if (j == PFO_Z || j == PFO_S)
4951 continue;
4952
4953 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
4954 fprintf(fout, " cond_%s = %s;\n",
4955 parsed_flag_op_names[j], buf1);
4956 pfomask &= ~(1 << j);
4957 }
2fe80fdb 4958 }
4959 goto dualop_arith;
4960
5101a5f9 4961 case OP_ADC:
850c9265 4962 case OP_SBB:
5101a5f9 4963 assert_operand_cnt(2);
4964 propagate_lmod(po, &po->operand[0], &po->operand[1]);
a2c1d768 4965 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
840257f6 4966 if (po->op == OP_SBB
4967 && IS(po->operand[0].name, po->operand[1].name))
4968 {
4969 // avoid use of unitialized var
a2c1d768 4970 fprintf(fout, " %s = -cond_c;", buf1);
94d447fb 4971 // carry remains what it was
4972 pfomask &= ~(1 << PFO_C);
840257f6 4973 }
4974 else {
a2c1d768 4975 fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po),
3ebea2cf 4976 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
840257f6 4977 }
a2c1d768 4978 output_std_flags(fout, po, &pfomask, buf1);
5101a5f9 4979 last_arith_dst = &po->operand[0];
4980 delayed_flag_op = NULL;
850c9265 4981 break;
4982
1f84f6b3 4983 case OP_BSF:
4984 assert_operand_cnt(2);
4985 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4986 fprintf(fout, " %s = %s ? __builtin_ffs(%s) - 1 : 0;",
4987 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4988 buf2, buf2);
4989 output_std_flags(fout, po, &pfomask, buf1);
4990 last_arith_dst = &po->operand[0];
4991 delayed_flag_op = NULL;
4992 strcat(g_comment, "bsf");
4993 break;
4994
850c9265 4995 case OP_DEC:
90307a99 4996 if (pfomask & ~(PFOB_S | PFOB_S | PFOB_C)) {
4997 for (j = 0; j <= PFO_LE; j++) {
4998 if (!(pfomask & (1 << j)))
4999 continue;
5000 if (j == PFO_Z || j == PFO_S || j == PFO_C)
5001 continue;
5002
5003 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
5004 fprintf(fout, " cond_%s = %s;\n",
5005 parsed_flag_op_names[j], buf1);
5006 pfomask &= ~(1 << j);
5007 }
5008 }
5009 // fallthrough
5010
5011 case OP_INC:
5012 if (pfomask & (1 << PFO_C))
5013 // carry is unaffected by inc/dec.. wtf?
5014 ferr(po, "carry propagation needed\n");
5015
850c9265 5016 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
1bafb621 5017 if (po->operand[0].type == OPT_REG) {
5018 strcpy(buf2, po->op == OP_INC ? "++" : "--");
5019 fprintf(fout, " %s%s;", buf1, buf2);
5020 }
5021 else {
5022 strcpy(buf2, po->op == OP_INC ? "+" : "-");
5023 fprintf(fout, " %s %s= 1;", buf1, buf2);
5024 }
a2c1d768 5025 output_std_flags(fout, po, &pfomask, buf1);
5101a5f9 5026 last_arith_dst = &po->operand[0];
5027 delayed_flag_op = NULL;
5028 break;
5029
5030 case OP_NEG:
5031 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3ebea2cf 5032 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
5101a5f9 5033 fprintf(fout, " %s = -%s%s;", buf1,
5034 lmod_cast_s(po, po->operand[0].lmod), buf2);
850c9265 5035 last_arith_dst = &po->operand[0];
69a3cdfc 5036 delayed_flag_op = NULL;
940e8e66 5037 if (pfomask & (1 << PFO_C)) {
5038 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
5039 pfomask &= ~(1 << PFO_C);
5040 }
850c9265 5041 break;
5042
5043 case OP_IMUL:
de50b98b 5044 if (po->operand_cnt == 2) {
5045 propagate_lmod(po, &po->operand[0], &po->operand[1]);
850c9265 5046 goto dualop_arith;
de50b98b 5047 }
87bf6cec 5048 if (po->operand_cnt == 3)
5049 ferr(po, "TODO imul3\n");
5050 // fallthrough
5051 case OP_MUL:
5052 assert_operand_cnt(1);
c8dbc5be 5053 switch (po->operand[0].lmod) {
5054 case OPLM_DWORD:
5055 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
5056 fprintf(fout, " tmp64 = %seax * %s%s;\n", buf1, buf1,
5057 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
5058 fprintf(fout, " edx = tmp64 >> 32;\n");
5059 fprintf(fout, " eax = tmp64;");
5060 break;
5061 case OPLM_BYTE:
5062 strcpy(buf1, po->op == OP_IMUL ? "(s16)(s8)" : "(u16)(u8)");
5063 fprintf(fout, " LOWORD(eax) = %seax * %s;", buf1,
5064 out_src_opr(buf2, sizeof(buf2), po, &po->operand[0],
5065 buf1, 0));
5066 break;
5067 default:
5068 ferr(po, "TODO: unhandled mul type\n");
5069 break;
5070 }
87bf6cec 5071 last_arith_dst = NULL;
69a3cdfc 5072 delayed_flag_op = NULL;
91977a1c 5073 break;
5074
5101a5f9 5075 case OP_DIV:
5076 case OP_IDIV:
5077 assert_operand_cnt(1);
5078 if (po->operand[0].lmod != OPLM_DWORD)
5079 ferr(po, "unhandled lmod %d\n", po->operand[0].lmod);
5080
cb090db0 5081 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
5082 strcpy(buf2, lmod_cast(po, po->operand[0].lmod,
5083 po->op == OP_IDIV));
5084 switch (po->operand[0].lmod) {
5085 case OPLM_DWORD:
5086 if (po->flags & OPF_32BIT)
5087 snprintf(buf3, sizeof(buf3), "%seax", buf2);
5088 else {
5089 fprintf(fout, " tmp64 = ((u64)edx << 32) | eax;\n");
5090 snprintf(buf3, sizeof(buf3), "%stmp64",
5091 (po->op == OP_IDIV) ? "(s64)" : "");
5092 }
5093 if (po->operand[0].type == OPT_REG
5094 && po->operand[0].reg == xDX)
5095 {
5096 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
5097 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
5098 }
5099 else {
5100 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
5101 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
5102 }
5103 break;
5104 default:
5105 ferr(po, "unhandled division type\n");
5101a5f9 5106 }
87bf6cec 5107 last_arith_dst = NULL;
5108 delayed_flag_op = NULL;
5101a5f9 5109 break;
5110
91977a1c 5111 case OP_TEST:
5112 case OP_CMP:
850c9265 5113 propagate_lmod(po, &po->operand[0], &po->operand[1]);
940e8e66 5114 if (pfomask != 0) {
69a3cdfc 5115 for (j = 0; j < 8; j++) {
940e8e66 5116 if (pfomask & (1 << j)) {
69a3cdfc 5117 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
5118 fprintf(fout, " cond_%s = %s;",
5119 parsed_flag_op_names[j], buf1);
5120 }
5121 }
940e8e66 5122 pfomask = 0;
69a3cdfc 5123 }
5124 else
5125 no_output = 1;
7ba45c34 5126 last_arith_dst = NULL;
69a3cdfc 5127 delayed_flag_op = po;
91977a1c 5128 break;
5129
092f64e1 5130 case OP_SCC:
5131 // SETcc - should already be handled
5132 break;
5133
69a3cdfc 5134 // note: we reuse OP_Jcc for SETcc, only flags differ
092f64e1 5135 case OP_JCC:
5136 fprintf(fout, "\n goto %s;", po->operand[0].name);
850c9265 5137 break;
5138
5c024ef7 5139 case OP_JECXZ:
5140 fprintf(fout, " if (ecx == 0)\n");
5141 fprintf(fout, " goto %s;", po->operand[0].name);
5142 strcat(g_comment, "jecxz");
5143 break;
5144
850c9265 5145 case OP_JMP:
87bf6cec 5146 assert_operand_cnt(1);
de50b98b 5147 last_arith_dst = NULL;
5148 delayed_flag_op = NULL;
5149
4c45fa73 5150 if (po->operand[0].type == OPT_REGMEM) {
5151 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
5152 buf1, buf2);
5153 if (ret != 2)
5154 ferr(po, "parse failure for jmp '%s'\n",
5155 po->operand[0].name);
5156 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
5157 break;
5158 }
5159 else if (po->operand[0].type != OPT_LABEL)
5160 ferr(po, "unhandled jmp type\n");
87bf6cec 5161
850c9265 5162 fprintf(fout, " goto %s;", po->operand[0].name);
91977a1c 5163 break;
5164
5165 case OP_CALL:
5101a5f9 5166 assert_operand_cnt(1);
092f64e1 5167 pp = po->pp;
89ff3147 5168 my_assert_not(pp, NULL);
91977a1c 5169
092f64e1 5170 strcpy(buf3, " ");
5171 if (po->flags & OPF_CC) {
5172 // we treat conditional branch to another func
5173 // (yes such code exists..) as conditional tailcall
5174 strcat(buf3, " ");
5175 fprintf(fout, " {\n");
5176 }
5177
8eb12e72 5178 if (pp->is_fptr && !pp->is_arg) {
092f64e1 5179 fprintf(fout, "%s%s = %s;\n", buf3, pp->name,
1cd4a663 5180 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
5181 "(void *)", 0));
8eb12e72 5182 if (pp->is_unresolved)
5183 fprintf(fout, "%sunresolved_call(\"%s:%d\", %s);\n",
5184 buf3, asmfn, po->asmln, pp->name);
5185 }
1bafb621 5186
092f64e1 5187 fprintf(fout, "%s", buf3);
2b43685d 5188 if (strstr(pp->ret_type.name, "int64")) {
87bf6cec 5189 if (po->flags & OPF_TAIL)
2b43685d 5190 ferr(po, "int64 and tail?\n");
2fe80fdb 5191 fprintf(fout, "tmp64 = ");
2b43685d 5192 }
5193 else if (!IS(pp->ret_type.name, "void")) {
5194 if (po->flags & OPF_TAIL) {
840257f6 5195 if (!IS(g_func_pp->ret_type.name, "void")) {
5196 fprintf(fout, "return ");
5197 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
5198 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
5199 }
2b43685d 5200 }
75ad0378 5201 else if (regmask & (1 << xAX)) {
87bf6cec 5202 fprintf(fout, "eax = ");
2b43685d 5203 if (pp->ret_type.is_ptr)
5204 fprintf(fout, "(u32)");
5205 }
91977a1c 5206 }
87bf6cec 5207
ddaf8bd7 5208 if (pp->name[0] == 0)
5209 ferr(po, "missing pp->name\n");
5210 fprintf(fout, "%s%s(", pp->name,
5211 pp->has_structarg ? "_sa" : "");
39b168b8 5212
2fe80fdb 5213 if (po->flags & OPF_ATAIL) {
5214 if (pp->argc_stack != g_func_pp->argc_stack
5215 || (pp->argc_stack > 0
5216 && pp->is_stdcall != g_func_pp->is_stdcall))
5217 ferr(po, "incompatible tailcall\n");
1f84f6b3 5218 if (g_func_pp->has_retreg)
5219 ferr(po, "TODO: retreg+tailcall\n");
87bf6cec 5220
2fe80fdb 5221 for (arg = j = 0; arg < pp->argc; arg++) {
5222 if (arg > 0)
5223 fprintf(fout, ", ");
87bf6cec 5224
2fe80fdb 5225 cast[0] = 0;
5226 if (pp->arg[arg].type.is_ptr)
5227 snprintf(cast, sizeof(cast), "(%s)",
5228 pp->arg[arg].type.name);
91977a1c 5229
2fe80fdb 5230 if (pp->arg[arg].reg != NULL) {
5231 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
5232 continue;
5233 }
5234 // stack arg
5235 for (; j < g_func_pp->argc; j++)
5236 if (g_func_pp->arg[j].reg == NULL)
5237 break;
5238 fprintf(fout, "%sa%d", cast, j + 1);
5239 j++;
69a3cdfc 5240 }
2fe80fdb 5241 }
5242 else {
5243 for (arg = 0; arg < pp->argc; arg++) {
5244 if (arg > 0)
5245 fprintf(fout, ", ");
5246
5247 cast[0] = 0;
5248 if (pp->arg[arg].type.is_ptr)
5249 snprintf(cast, sizeof(cast), "(%s)",
5250 pp->arg[arg].type.name);
5251
5252 if (pp->arg[arg].reg != NULL) {
1f84f6b3 5253 if (pp->arg[arg].type.is_retreg)
5254 fprintf(fout, "&%s", pp->arg[arg].reg);
5255 else
5256 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
2fe80fdb 5257 continue;
5258 }
5259
5260 // stack arg
5261 tmp_op = pp->arg[arg].datap;
5262 if (tmp_op == NULL)
5263 ferr(po, "parsed_op missing for arg%d\n", arg);
23fd0b11 5264
5265 if (tmp_op->flags & OPF_VAPUSH) {
5266 fprintf(fout, "ap");
5267 }
5f70a34f 5268 else if (tmp_op->p_argpass != 0) {
5269 fprintf(fout, "a%d", tmp_op->p_argpass);
5270 }
5271 else if (tmp_op->p_argnum != 0) {
3a5101d7 5272 fprintf(fout, "%s%s", cast,
5273 saved_arg_name(buf1, sizeof(buf1),
5274 tmp_op->p_arggrp, tmp_op->p_argnum));
2fe80fdb 5275 }
5276 else {
5277 fprintf(fout, "%s",
5278 out_src_opr(buf1, sizeof(buf1),
5279 tmp_op, &tmp_op->operand[0], cast, 0));
5280 }
69a3cdfc 5281 }
91977a1c 5282 }
5283 fprintf(fout, ");");
87bf6cec 5284
2b43685d 5285 if (strstr(pp->ret_type.name, "int64")) {
5286 fprintf(fout, "\n");
092f64e1 5287 fprintf(fout, "%sedx = tmp64 >> 32;\n", buf3);
5288 fprintf(fout, "%seax = tmp64;", buf3);
2b43685d 5289 }
5290
89ff3147 5291 if (pp->is_unresolved) {
8eb12e72 5292 snprintf(buf2, sizeof(buf2), " unresolved %dreg",
89ff3147 5293 pp->argc_reg);
092f64e1 5294 strcat(g_comment, buf2);
89ff3147 5295 }
5296
87bf6cec 5297 if (po->flags & OPF_TAIL) {
840257f6 5298 ret = 0;
ddaf8bd7 5299 if (i == opcnt - 1 || pp->is_noreturn)
840257f6 5300 ret = 0;
5301 else if (IS(pp->ret_type.name, "void"))
5302 ret = 1;
5303 else if (IS(g_func_pp->ret_type.name, "void"))
5304 ret = 1;
5305 // else already handled as 'return f()'
5306
5307 if (ret) {
da87ae38 5308 if (!IS(g_func_pp->ret_type.name, "void")) {
ddaf8bd7 5309 ferr(po, "int func -> void func tailcall?\n");
da87ae38 5310 }
5311 else {
092f64e1 5312 fprintf(fout, "\n%sreturn;", buf3);
ddaf8bd7 5313 strcat(g_comment, " ^ tailcall");
da87ae38 5314 }
3ebea2cf 5315 }
89ff3147 5316 else
ddaf8bd7 5317 strcat(g_comment, " tailcall");
87bf6cec 5318 }
ddaf8bd7 5319 if (pp->is_noreturn)
5320 strcat(g_comment, " noreturn");
2fe80fdb 5321 if ((po->flags & OPF_ATAIL) && pp->argc_stack > 0)
5322 strcat(g_comment, " argframe");
092f64e1 5323 if (po->flags & OPF_CC)
5324 strcat(g_comment, " cond");
5325
5326 if (po->flags & OPF_CC)
5327 fprintf(fout, "\n }");
5328
87bf6cec 5329 delayed_flag_op = NULL;
5330 last_arith_dst = NULL;
91977a1c 5331 break;
5332
5333 case OP_RET:
bd96f656 5334 if (g_func_pp->is_vararg)
4f12f671 5335 fprintf(fout, " va_end(ap);\n");
1f84f6b3 5336 if (g_func_pp->has_retreg) {
5337 for (arg = 0; arg < g_func_pp->argc; arg++)
5338 if (g_func_pp->arg[arg].type.is_retreg)
5339 fprintf(fout, " *r_%s = %s;\n",
5340 g_func_pp->arg[arg].reg, g_func_pp->arg[arg].reg);
5341 }
4f12f671 5342
bd96f656 5343 if (IS(g_func_pp->ret_type.name, "void")) {
3ebea2cf 5344 if (i != opcnt - 1 || label_pending)
5345 fprintf(fout, " return;");
5346 }
bd96f656 5347 else if (g_func_pp->ret_type.is_ptr) {
d4e3b5db 5348 fprintf(fout, " return (%s)eax;",
bd96f656 5349 g_func_pp->ret_type.name);
3ebea2cf 5350 }
2fe80fdb 5351 else if (IS(g_func_pp->ret_type.name, "__int64"))
5352 fprintf(fout, " return ((u64)edx << 32) | eax;");
91977a1c 5353 else
5354 fprintf(fout, " return eax;");
de50b98b 5355
5356 last_arith_dst = NULL;
5357 delayed_flag_op = NULL;
91977a1c 5358 break;
5359
5360 case OP_PUSH:
1f84f6b3 5361 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
5f70a34f 5362 if (po->p_argnum != 0) {
69a3cdfc 5363 // special case - saved func arg
3a5101d7 5364 fprintf(fout, " %s = %s;",
5365 saved_arg_name(buf2, sizeof(buf2),
5366 po->p_arggrp, po->p_argnum), buf1);
69a3cdfc 5367 break;
5368 }
d4e3b5db 5369 else if (po->flags & OPF_RSAVE) {
d4e3b5db 5370 fprintf(fout, " s_%s = %s;", buf1, buf1);
5371 break;
5372 }
1f84f6b3 5373 else if (g_func_pp->is_userstack) {
5374 fprintf(fout, " *(--esp) = %s;", buf1);
5375 break;
5376 }
e56ab892 5377 if (!(g_ida_func_attr & IDAFA_NORETURN))
5378 ferr(po, "stray push encountered\n");
5379 no_output = 1;
91977a1c 5380 break;
5381
5382 case OP_POP:
d4e3b5db 5383 if (po->flags & OPF_RSAVE) {
3ebea2cf 5384 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
d4e3b5db 5385 fprintf(fout, " %s = s_%s;", buf1, buf1);
5386 break;
5387 }
5c024ef7 5388 else if (po->datap != NULL) {
5389 // push/pop pair
5390 tmp_op = po->datap;
5391 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5392 fprintf(fout, " %s = %s;", buf1,
5393 out_src_opr(buf2, sizeof(buf2),
5394 tmp_op, &tmp_op->operand[0],
c7ed83dd 5395 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
5c024ef7 5396 break;
5397 }
1f84f6b3 5398 else if (g_func_pp->is_userstack) {
5399 fprintf(fout, " %s = *esp++;",
5400 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
5401 break;
5402 }
5403 else
5404 ferr(po, "stray pop encountered\n");
91977a1c 5405 break;
5406
33c35af6 5407 case OP_NOP:
2b43685d 5408 no_output = 1;
33c35af6 5409 break;
5410
90307a99 5411 // mmx
5412 case OP_EMMS:
5413 strcpy(g_comment, "(emms)");
5414 break;
5415
91977a1c 5416 default:
5417 no_output = 1;
69a3cdfc 5418 ferr(po, "unhandled op type %d, flags %x\n",
5419 po->op, po->flags);
91977a1c 5420 break;
5421 }
5422
5423 if (g_comment[0] != 0) {
ddaf8bd7 5424 char *p = g_comment;
5425 while (my_isblank(*p))
5426 p++;
5427 fprintf(fout, " // %s", p);
91977a1c 5428 g_comment[0] = 0;
5429 no_output = 0;
5430 }
5431 if (!no_output)
5432 fprintf(fout, "\n");
5101a5f9 5433
2b43685d 5434 // some sanity checking
591721d7 5435 if (po->flags & OPF_REP) {
5436 if (po->op != OP_STOS && po->op != OP_MOVS
5437 && po->op != OP_CMPS && po->op != OP_SCAS)
5438 ferr(po, "unexpected rep\n");
5439 if (!(po->flags & (OPF_REPZ|OPF_REPNZ))
5440 && (po->op == OP_CMPS || po->op == OP_SCAS))
5441 ferr(po, "cmps/scas with plain rep\n");
5442 }
5443 if ((po->flags & (OPF_REPZ|OPF_REPNZ))
5444 && po->op != OP_CMPS && po->op != OP_SCAS)
2b43685d 5445 ferr(po, "unexpected repz/repnz\n");
5446
940e8e66 5447 if (pfomask != 0)
7ba45c34 5448 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
940e8e66 5449
5101a5f9 5450 // see is delayed flag stuff is still valid
5451 if (delayed_flag_op != NULL && delayed_flag_op != po) {
89ff3147 5452 if (is_any_opr_modified(delayed_flag_op, po, 0))
5101a5f9 5453 delayed_flag_op = NULL;
5454 }
5455
5456 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
5457 if (is_opr_modified(last_arith_dst, po))
5458 last_arith_dst = NULL;
5459 }
3ebea2cf 5460
5461 label_pending = 0;
91977a1c 5462 }
5463
a2c1d768 5464 if (g_stack_fsz && !g_stack_frame_used)
5465 fprintf(fout, " (void)sf;\n");
5466
91977a1c 5467 fprintf(fout, "}\n\n");
5468
9af2d373 5469 gen_x_cleanup(opcnt);
5470}
5471
5472static void gen_x_cleanup(int opcnt)
5473{
5474 int i;
5475
91977a1c 5476 for (i = 0; i < opcnt; i++) {
4c45fa73 5477 struct label_ref *lr, *lr_del;
5478
5479 lr = g_label_refs[i].next;
5480 while (lr != NULL) {
5481 lr_del = lr;
5482 lr = lr->next;
5483 free(lr_del);
5484 }
5485 g_label_refs[i].i = -1;
5486 g_label_refs[i].next = NULL;
5487
91977a1c 5488 if (ops[i].op == OP_CALL) {
092f64e1 5489 if (ops[i].pp)
5490 proto_release(ops[i].pp);
91977a1c 5491 }
5492 }
bd96f656 5493 g_func_pp = NULL;
91977a1c 5494}
c36e914d 5495
92d715b6 5496struct func_proto_dep;
5497
5498struct func_prototype {
5499 char name[NAMELEN];
5500 int id;
5501 int argc_stack;
5502 int regmask_dep;
91ca764a 5503 int has_ret:3; // -1, 0, 1: unresolved, no, yes
92d715b6 5504 unsigned int dep_resolved:1;
5505 unsigned int is_stdcall:1;
5506 struct func_proto_dep *dep_func;
5507 int dep_func_cnt;
91ca764a 5508 const struct parsed_proto *pp; // seed pp, if any
92d715b6 5509};
5510
5511struct func_proto_dep {
5512 char *name;
5513 struct func_prototype *proto;
5514 int regmask_live; // .. at the time of call
5515 unsigned int ret_dep:1; // return from this is caller's return
5516};
5517
5518static struct func_prototype *hg_fp;
5519static int hg_fp_cnt;
5520
5fa1256f 5521static struct scanned_var {
5522 char name[NAMELEN];
5523 enum opr_lenmod lmod;
5524 unsigned int is_seeded:1;
61e29183 5525 unsigned int is_c_str:1;
c0de9015 5526 const struct parsed_proto *pp; // seed pp, if any
5fa1256f 5527} *hg_vars;
5528static int hg_var_cnt;
5529
9af2d373 5530static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
5531 int count);
5532
92d715b6 5533static struct func_proto_dep *hg_fp_find_dep(struct func_prototype *fp,
5534 const char *name)
5535{
5536 int i;
5537
5538 for (i = 0; i < fp->dep_func_cnt; i++)
5539 if (IS(fp->dep_func[i].name, name))
5540 return &fp->dep_func[i];
5541
5542 return NULL;
5543}
5544
5545static void hg_fp_add_dep(struct func_prototype *fp, const char *name)
5546{
5547 // is it a dupe?
5548 if (hg_fp_find_dep(fp, name))
5549 return;
5550
5551 if ((fp->dep_func_cnt & 0xff) == 0) {
5552 fp->dep_func = realloc(fp->dep_func,
5553 sizeof(fp->dep_func[0]) * (fp->dep_func_cnt + 0x100));
5554 my_assert_not(fp->dep_func, NULL);
5555 memset(&fp->dep_func[fp->dep_func_cnt], 0,
5556 sizeof(fp->dep_func[0]) * 0x100);
5557 }
5558 fp->dep_func[fp->dep_func_cnt].name = strdup(name);
5559 fp->dep_func_cnt++;
5560}
5561
5562static int hg_fp_cmp_name(const void *p1_, const void *p2_)
5563{
5564 const struct func_prototype *p1 = p1_, *p2 = p2_;
5565 return strcmp(p1->name, p2->name);
5566}
5567
5568#if 0
5569static int hg_fp_cmp_id(const void *p1_, const void *p2_)
5570{
5571 const struct func_prototype *p1 = p1_, *p2 = p2_;
5572 return p1->id - p2->id;
5573}
5574#endif
5575
91ca764a 5576// recursive register dep pass
5577// - track saved regs (part 2)
5578// - try to figure out arg-regs
5579// - calculate reg deps
5580static void gen_hdr_dep_pass(int i, int opcnt, unsigned char *cbits,
5581 struct func_prototype *fp, int regmask_save, int regmask_dst,
5582 int *regmask_dep, int *has_ret)
5583{
5584 struct func_proto_dep *dep;
5585 struct parsed_op *po;
5586 int from_caller = 0;
5587 int depth;
5588 int j, l;
5589 int reg;
5590 int ret;
5591
5592 for (; i < opcnt; i++)
5593 {
5594 if (cbits[i >> 3] & (1 << (i & 7)))
5595 return;
5596 cbits[i >> 3] |= (1 << (i & 7));
5597
5598 po = &ops[i];
5599
5600 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
5601 if (po->btj != NULL) {
5602 // jumptable
5603 for (j = 0; j < po->btj->count; j++) {
5604 gen_hdr_dep_pass(po->btj->d[j].bt_i, opcnt, cbits, fp,
5605 regmask_save, regmask_dst, regmask_dep, has_ret);
5606 }
5607 return;
5608 }
5609
5610 if (po->bt_i < 0) {
5611 ferr(po, "dead branch\n");
5612 return;
5613 }
5614
5615 if (po->flags & OPF_CJMP) {
5616 gen_hdr_dep_pass(po->bt_i, opcnt, cbits, fp,
5617 regmask_save, regmask_dst, regmask_dep, has_ret);
5618 }
5619 else {
5620 i = po->bt_i - 1;
5621 }
5622 continue;
5623 }
5624
5625 if (po->flags & OPF_FARG)
5626 /* (just calculate register deps) */;
5627 else if (po->op == OP_PUSH && po->operand[0].type == OPT_REG)
5628 {
5629 reg = po->operand[0].reg;
5630 if (reg < 0)
5631 ferr(po, "reg not set for push?\n");
5632
5633 if (po->flags & OPF_RSAVE) {
5634 regmask_save |= 1 << reg;
5635 continue;
5636 }
5637 if (po->flags & OPF_DONE)
5638 continue;
5639
5640 depth = 0;
5641 ret = scan_for_pop(i + 1, opcnt,
5642 po->operand[0].name, i + opcnt * 2, 0, &depth, 0);
5643 if (ret == 1) {
5644 regmask_save |= 1 << reg;
5645 po->flags |= OPF_RMD;
5646 scan_for_pop(i + 1, opcnt,
5647 po->operand[0].name, i + opcnt * 3, 0, &depth, 1);
5648 continue;
5649 }
5650 }
5651 else if (po->flags & OPF_RMD)
5652 continue;
5653 else if (po->op == OP_CALL) {
5654 po->regmask_dst |= 1 << xAX;
5655
5656 dep = hg_fp_find_dep(fp, po->operand[0].name);
5657 if (dep != NULL)
5658 dep->regmask_live = regmask_save | regmask_dst;
5659 }
5660 else if (po->op == OP_RET) {
5661 if (po->operand_cnt > 0) {
5662 fp->is_stdcall = 1;
5663 if (fp->argc_stack >= 0
5664 && fp->argc_stack != po->operand[0].val / 4)
5665 ferr(po, "ret mismatch? (%d)\n", fp->argc_stack * 4);
5666 fp->argc_stack = po->operand[0].val / 4;
5667 }
5668 }
5669
5670 if (*has_ret != 0 && (po->flags & OPF_TAIL)) {
5671 if (po->op == OP_CALL) {
5672 j = i;
5673 ret = 1;
5674 }
5675 else {
5676 struct parsed_opr opr = { 0, };
5677 opr.type = OPT_REG;
5678 opr.reg = xAX;
5679 j = -1;
5680 from_caller = 0;
5681 ret = resolve_origin(i, &opr, i + opcnt * 4, &j, &from_caller);
5682 }
5683
5684 if (ret == -1 && from_caller) {
5685 // unresolved eax - probably void func
5686 *has_ret = 0;
5687 }
5688 else {
5689 if (ops[j].op == OP_CALL) {
5690 dep = hg_fp_find_dep(fp, po->operand[0].name);
5691 if (dep != NULL)
5692 dep->ret_dep = 1;
5693 else
5694 *has_ret = 1;
5695 }
5696 else
5697 *has_ret = 1;
5698 }
5699 }
5700
5701 l = regmask_save | regmask_dst;
5702 if (g_bp_frame && !(po->flags & OPF_EBP_S))
5703 l |= 1 << xBP;
5704
5705 l = po->regmask_src & ~l;
5706#if 0
5707 if (l)
5708 fnote(po, "dep |= %04x, dst %04x, save %04x (f %x)\n",
5709 l, regmask_dst, regmask_save, po->flags);
5710#endif
5711 *regmask_dep |= l;
5712 regmask_dst |= po->regmask_dst;
5713
5714 if (po->flags & OPF_TAIL)
5715 return;
5716 }
5717}
5718
92d715b6 5719static void gen_hdr(const char *funcn, int opcnt)
5720{
26677139 5721 int save_arg_vars[MAX_ARG_GRP] = { 0, };
91ca764a 5722 unsigned char cbits[MAX_OPS / 8];
9af2d373 5723 const struct parsed_proto *pp_c;
5724 struct parsed_proto *pp;
92d715b6 5725 struct func_prototype *fp;
92d715b6 5726 struct parsed_data *pd;
5727 struct parsed_op *po;
9af2d373 5728 const char *tmpname;
26677139 5729 int regmask_dummy = 0;
91ca764a 5730 int regmask_dep;
92d715b6 5731 int max_bp_offset = 0;
91ca764a 5732 int has_ret;
92d715b6 5733 int i, j, l, ret;
92d715b6 5734
5735 if ((hg_fp_cnt & 0xff) == 0) {
5736 hg_fp = realloc(hg_fp, sizeof(hg_fp[0]) * (hg_fp_cnt + 0x100));
5737 my_assert_not(hg_fp, NULL);
5738 memset(hg_fp + hg_fp_cnt, 0, sizeof(hg_fp[0]) * 0x100);
5739 }
5740
5741 fp = &hg_fp[hg_fp_cnt];
5742 snprintf(fp->name, sizeof(fp->name), "%s", funcn);
5743 fp->id = hg_fp_cnt;
9af2d373 5744 fp->argc_stack = -1;
92d715b6 5745 hg_fp_cnt++;
5746
9af2d373 5747 // perhaps already in seed header?
91ca764a 5748 fp->pp = proto_parse(g_fhdr, funcn, 1);
5749 if (fp->pp != NULL) {
5750 fp->argc_stack = fp->pp->argc_stack;
5751 fp->is_stdcall = fp->pp->is_stdcall;
5752 fp->regmask_dep = get_pp_arg_regmask(fp->pp);
5753 fp->has_ret = !IS(fp->pp->ret_type.name, "void");
9af2d373 5754 return;
5755 }
5756
5757 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
5758 g_stack_frame_used = 0;
5759
92d715b6 5760 // pass1:
9af2d373 5761 // - handle ebp/esp frame, remove ops related to it
5762 scan_prologue_epilogue(opcnt);
5763
5764 // pass2:
92d715b6 5765 // - collect calls
5766 // - resolve all branches
5767 for (i = 0; i < opcnt; i++)
5768 {
5769 po = &ops[i];
5770 po->bt_i = -1;
5771 po->btj = NULL;
5772
5e49b270 5773 if (po->flags & (OPF_RMD|OPF_DONE))
92d715b6 5774 continue;
5775
5776 if (po->op == OP_CALL) {
9af2d373 5777 tmpname = opr_name(po, 0);
5778 pp = NULL;
5779 if (po->operand[0].type == OPT_LABEL) {
5780 hg_fp_add_dep(fp, tmpname);
5781
5782 // perhaps a call to already known func?
5783 pp_c = proto_parse(g_fhdr, tmpname, 1);
5784 if (pp_c != NULL)
5785 pp = proto_clone(pp_c);
5786 }
5787 else if (po->datap != NULL) {
5788 pp = calloc(1, sizeof(*pp));
5789 my_assert_not(pp, NULL);
92d715b6 5790
9af2d373 5791 ret = parse_protostr(po->datap, pp);
5792 if (ret < 0)
5793 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
5794 free(po->datap);
5795 po->datap = NULL;
5796 }
5797 if (pp != NULL && pp->is_noreturn)
5798 po->flags |= OPF_TAIL;
5799
5800 po->pp = pp;
92d715b6 5801 continue;
5802 }
5803
5804 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
5805 continue;
5806
5807 if (po->operand[0].type == OPT_REGMEM) {
5808 pd = try_resolve_jumptab(i, opcnt);
5809 if (pd == NULL)
5810 goto tailcall;
5811
5812 po->btj = pd;
5813 continue;
5814 }
5815
5816 for (l = 0; l < opcnt; l++) {
d7857c3a 5817 if (g_labels[l] != NULL
5818 && IS(po->operand[0].name, g_labels[l]))
5819 {
92d715b6 5820 add_label_ref(&g_label_refs[l], i);
5821 po->bt_i = l;
5822 break;
5823 }
5824 }
5825
5826 if (po->bt_i != -1 || (po->flags & OPF_RMD))
5827 continue;
5828
5829 if (po->operand[0].type == OPT_LABEL)
5830 // assume tail call
5831 goto tailcall;
5832
5833 ferr(po, "unhandled branch\n");
5834
5835tailcall:
5836 po->op = OP_CALL;
5837 po->flags |= OPF_TAIL;
5838 if (i > 0 && ops[i - 1].op == OP_POP)
5839 po->flags |= OPF_ATAIL;
5840 i--; // reprocess
5841 }
5842
9af2d373 5843 // pass3:
92d715b6 5844 // - remove dead labels
9af2d373 5845 // - handle push <const>/pop pairs
92d715b6 5846 for (i = 0; i < opcnt; i++)
5847 {
d7857c3a 5848 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
5849 free(g_labels[i]);
5850 g_labels[i] = NULL;
5851 }
9af2d373 5852
91ca764a 5853 po = &ops[i];
5854 if (po->flags & (OPF_RMD|OPF_DONE))
5855 continue;
5856
5857 if (po->op == OP_PUSH && po->operand[0].type == OPT_CONST)
5858 scan_for_pop_const(i, opcnt);
5859 }
5860
5861 // pass4:
5862 // - process trivial calls
5863 for (i = 0; i < opcnt; i++)
5864 {
9af2d373 5865 po = &ops[i];
5e49b270 5866 if (po->flags & (OPF_RMD|OPF_DONE))
9af2d373 5867 continue;
5868
26677139 5869 if (po->op == OP_CALL)
5870 {
5871 pp = process_call_early(i, opcnt, &j);
5872 if (pp != NULL) {
5873 if (!(po->flags & OPF_ATAIL))
5874 // since we know the args, try to collect them
5875 if (collect_call_args_early(po, i, pp, &regmask_dummy) != 0)
5876 pp = NULL;
5877 }
5878
5879 if (pp != NULL) {
5880 if (j >= 0) {
5881 // commit esp adjust
5882 ops[j].flags |= OPF_RMD;
5e49b270 5883 if (ops[j].op != OP_POP)
5884 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
5885 else
5886 ops[j].flags |= OPF_DONE;
26677139 5887 }
5888
5889 po->flags |= OPF_DONE;
5890 }
5891 }
26677139 5892 }
5893
91ca764a 5894 // pass5:
5e49b270 5895 // - track saved regs (simple)
26677139 5896 // - process calls
5897 for (i = 0; i < opcnt; i++)
5898 {
5899 po = &ops[i];
5e49b270 5900 if (po->flags & (OPF_RMD|OPF_DONE))
26677139 5901 continue;
5902
5e49b270 5903 if (po->op == OP_PUSH && po->operand[0].type == OPT_REG)
5904 {
5905 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
5906 if (ret == 0) {
91ca764a 5907 // regmask_save |= 1 << po->operand[0].reg; // do it later
5908 po->flags |= OPF_RSAVE | OPF_RMD | OPF_DONE;
5e49b270 5909 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, OPF_RMD);
5910 }
5911 }
5912 else if (po->op == OP_CALL && !(po->flags & OPF_DONE))
26677139 5913 {
9af2d373 5914 pp = process_call(i, opcnt);
5915
5916 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
9af2d373 5917 // since we know the args, collect them
91ca764a 5918 ret = collect_call_args(po, i, pp, &regmask_dummy, save_arg_vars,
5e49b270 5919 i + opcnt * 1);
9af2d373 5920 }
5921 }
92d715b6 5922 }
5923
91ca764a 5924 // pass6
5925 memset(cbits, 0, sizeof(cbits));
5926 regmask_dep = 0;
5927 has_ret = -1;
5928
5929 gen_hdr_dep_pass(0, opcnt, cbits, fp, 0, 0, &regmask_dep, &has_ret);
5930
5931 // find unreachable code - must be fixed in IDA
92d715b6 5932 for (i = 0; i < opcnt; i++)
5933 {
91ca764a 5934 if (cbits[i >> 3] & (1 << (i & 7)))
9af2d373 5935 continue;
92d715b6 5936
91ca764a 5937 if (ops[i].op != OP_NOP)
5938 ferr(&ops[i], "unreachable code\n");
92d715b6 5939 }
5940
5941 if (has_ret == -1 && (regmask_dep & (1 << xAX)))
5942 has_ret = 1;
5943
9af2d373 5944 for (i = 0; i < g_eqcnt; i++) {
92d715b6 5945 if (g_eqs[i].offset > max_bp_offset && g_eqs[i].offset < 4*32)
5946 max_bp_offset = g_eqs[i].offset;
9af2d373 5947 }
92d715b6 5948
9af2d373 5949 if (fp->argc_stack < 0) {
92d715b6 5950 max_bp_offset = (max_bp_offset + 3) & ~3;
9af2d373 5951 fp->argc_stack = max_bp_offset / 4;
5952 if ((g_ida_func_attr & IDAFA_BP_FRAME) && fp->argc_stack > 0)
92d715b6 5953 fp->argc_stack--;
5954 }
5955
5956 fp->regmask_dep = regmask_dep & ~(1 << xSP);
5957 fp->has_ret = has_ret;
91ca764a 5958#if 0
5959 printf("// has_ret %d, regmask_dep %x\n",
5960 fp->has_ret, fp->regmask_dep);
5961 output_hdr_fp(stdout, fp, 1);
5962 if (IS(funcn, "sub_100073FD")) exit(1);
5963#endif
9af2d373 5964
5965 gen_x_cleanup(opcnt);
92d715b6 5966}
5967
5968static void hg_fp_resolve_deps(struct func_prototype *fp)
5969{
5970 struct func_prototype fp_s;
91ca764a 5971 int dep;
92d715b6 5972 int i;
5973
5974 // this thing is recursive, so mark first..
5975 fp->dep_resolved = 1;
5976
5977 for (i = 0; i < fp->dep_func_cnt; i++) {
5978 strcpy(fp_s.name, fp->dep_func[i].name);
5979 fp->dep_func[i].proto = bsearch(&fp_s, hg_fp, hg_fp_cnt,
5980 sizeof(hg_fp[0]), hg_fp_cmp_name);
5981 if (fp->dep_func[i].proto != NULL) {
5982 if (!fp->dep_func[i].proto->dep_resolved)
5983 hg_fp_resolve_deps(fp->dep_func[i].proto);
5984
91ca764a 5985 dep = ~fp->dep_func[i].regmask_live
5986 & fp->dep_func[i].proto->regmask_dep;
5987 fp->regmask_dep |= dep;
5988 // printf("dep %s %s |= %x\n", fp->name,
5989 // fp->dep_func[i].name, dep);
92d715b6 5990
5991 if (fp->has_ret == -1)
5992 fp->has_ret = fp->dep_func[i].proto->has_ret;
5993 }
5994 }
5995}
5996
9af2d373 5997static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
5998 int count)
92d715b6 5999{
61e29183 6000 const struct parsed_proto *pp;
6001 char *p, namebuf[NAMELEN];
6002 const char *name;
92d715b6 6003 int regmask_dep;
6004 int argc_stack;
9af2d373 6005 int j, arg;
92d715b6 6006
9af2d373 6007 for (; count > 0; count--, fp++) {
92d715b6 6008 if (fp->has_ret == -1)
6009 fprintf(fout, "// ret unresolved\n");
6010#if 0
6011 fprintf(fout, "// dep:");
6012 for (j = 0; j < fp->dep_func_cnt; j++) {
6013 fprintf(fout, " %s/", fp->dep_func[j].name);
6014 if (fp->dep_func[j].proto != NULL)
6015 fprintf(fout, "%04x/%d", fp->dep_func[j].proto->regmask_dep,
6016 fp->dep_func[j].proto->has_ret);
6017 }
6018 fprintf(fout, "\n");
6019#endif
6020
61e29183 6021 p = strchr(fp->name, '@');
6022 if (p != NULL) {
6023 memcpy(namebuf, fp->name, p - fp->name);
6024 namebuf[p - fp->name] = 0;
6025 name = namebuf;
6026 }
6027 else
6028 name = fp->name;
6029 if (name[0] == '_')
6030 name++;
6031
6032 pp = proto_parse(g_fhdr, name, 1);
6033 if (pp != NULL && pp->is_include)
6034 continue;
6035
c0de9015 6036 if (fp->pp != NULL) {
4e81a3a2 6037 // part of seed, output later
6038 continue;
c0de9015 6039 }
6040
92d715b6 6041 regmask_dep = fp->regmask_dep;
6042 argc_stack = fp->argc_stack;
6043
91ca764a 6044 fprintf(fout, "%-5s", fp->pp ? fp->pp->ret_type.name :
6045 (fp->has_ret ? "int" : "void"));
92d715b6 6046 if (regmask_dep && (fp->is_stdcall || argc_stack == 0)
6047 && (regmask_dep & ~((1 << xCX) | (1 << xDX))) == 0)
6048 {
9af2d373 6049 fprintf(fout, " __fastcall ");
92d715b6 6050 if (!(regmask_dep & (1 << xDX)) && argc_stack == 0)
6051 argc_stack = 1;
6052 else
6053 argc_stack += 2;
6054 regmask_dep = 0;
6055 }
6056 else if (regmask_dep && !fp->is_stdcall) {
6057 fprintf(fout, "/*__usercall*/ ");
92d715b6 6058 }
6059 else if (regmask_dep) {
6060 fprintf(fout, "/*__userpurge*/ ");
92d715b6 6061 }
6062 else if (fp->is_stdcall)
9af2d373 6063 fprintf(fout, " __stdcall ");
92d715b6 6064 else
9af2d373 6065 fprintf(fout, " __cdecl ");
92d715b6 6066
61e29183 6067 fprintf(fout, "%s(", name);
92d715b6 6068
6069 arg = 0;
6070 for (j = 0; j < xSP; j++) {
6071 if (regmask_dep & (1 << j)) {
6072 arg++;
6073 if (arg != 1)
6074 fprintf(fout, ", ");
91ca764a 6075 if (fp->pp != NULL)
6076 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
6077 else
6078 fprintf(fout, "int");
6079 fprintf(fout, " a%d/*<%s>*/", arg, regs_r32[j]);
92d715b6 6080 }
6081 }
6082
6083 for (j = 0; j < argc_stack; j++) {
6084 arg++;
6085 if (arg != 1)
6086 fprintf(fout, ", ");
91ca764a 6087 if (fp->pp != NULL) {
6088 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
6089 if (!fp->pp->arg[arg - 1].type.is_ptr)
6090 fprintf(fout, " ");
6091 }
6092 else
6093 fprintf(fout, "int ");
6094 fprintf(fout, "a%d", arg);
92d715b6 6095 }
6096
6097 fprintf(fout, ");\n");
6098 }
6099}
6100
9af2d373 6101static void output_hdr(FILE *fout)
6102{
5fa1256f 6103 static const char *lmod_c_names[] = {
6104 [OPLM_UNSPEC] = "???",
6105 [OPLM_BYTE] = "uint8_t",
6106 [OPLM_WORD] = "uint16_t",
6107 [OPLM_DWORD] = "uint32_t",
6108 [OPLM_QWORD] = "uint64_t",
6109 };
6110 const struct scanned_var *var;
c0de9015 6111 char line[256] = { 0, };
9af2d373 6112 int i;
6113
6114 // resolve deps
6115 qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_name);
6116 for (i = 0; i < hg_fp_cnt; i++)
6117 hg_fp_resolve_deps(&hg_fp[i]);
6118
6119 // note: messes up .proto ptr, don't use
6120 //qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_id);
6121
5fa1256f 6122 // output variables
6123 for (i = 0; i < hg_var_cnt; i++) {
6124 var = &hg_vars[i];
6125
4e81a3a2 6126 if (var->pp != NULL)
6127 // part of seed
6128 continue;
c0de9015 6129 else if (var->is_c_str)
61e29183 6130 fprintf(fout, "extern %-8s %s[];", "char", var->name);
6131 else
6132 fprintf(fout, "extern %-8s %s;",
6133 lmod_c_names[var->lmod], var->name);
5fa1256f 6134
6135 if (var->is_seeded)
6136 fprintf(fout, " // seeded");
6137 fprintf(fout, "\n");
6138 }
6139
6140 fprintf(fout, "\n");
6141
6142 // output function prototypes
9af2d373 6143 output_hdr_fp(fout, hg_fp, hg_fp_cnt);
c0de9015 6144
4e81a3a2 6145 // seed passthrough
6146 fprintf(fout, "\n// - seed -\n");
c0de9015 6147
6148 rewind(g_fhdr);
4e81a3a2 6149 while (fgets(line, sizeof(line), g_fhdr))
6150 fwrite(line, 1, strlen(line), fout);
9af2d373 6151}
6152
5fa1256f 6153// read a line, truncating it if it doesn't fit
6154static char *my_fgets(char *s, size_t size, FILE *stream)
d4e3b5db 6155{
5fa1256f 6156 char *ret, *ret2;
6157 char buf[64];
6158 int p;
d4e3b5db 6159
5fa1256f 6160 p = size - 2;
6161 if (p >= 0)
6162 s[p] = 0;
d4e3b5db 6163
5fa1256f 6164 ret = fgets(s, size, stream);
6165 if (ret != NULL && p >= 0 && s[p] != 0 && s[p] != '\n') {
6166 p = sizeof(buf) - 2;
6167 do {
6168 buf[p] = 0;
6169 ret2 = fgets(buf, sizeof(buf), stream);
6170 }
6171 while (ret2 != NULL && buf[p] != 0 && buf[p] != '\n');
6172 }
6173
6174 return ret;
d4e3b5db 6175}
6176
61e29183 6177// '=' needs special treatment
6178// also ' quote
bfa4a6ee 6179static char *next_word_s(char *w, size_t wsize, char *s)
6180{
61e29183 6181 size_t i;
bfa4a6ee 6182
61e29183 6183 s = sskip(s);
bfa4a6ee 6184
61e29183 6185 i = 0;
6186 if (*s == '\'') {
6187 w[0] = s[0];
6188 for (i = 1; i < wsize - 1; i++) {
6189 if (s[i] == 0) {
6190 printf("warning: missing closing quote: \"%s\"\n", s);
6191 break;
6192 }
6193 if (s[i] == '\'')
6194 break;
6195 w[i] = s[i];
6196 }
6197 }
bfa4a6ee 6198
61e29183 6199 for (; i < wsize - 1; i++) {
6200 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
6201 break;
6202 w[i] = s[i];
6203 }
6204 w[i] = 0;
6205
6206 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
6207 printf("warning: '%s' truncated\n", w);
bfa4a6ee 6208
61e29183 6209 return s + i;
bfa4a6ee 6210}
6211
5fa1256f 6212static void scan_variables(FILE *fasm)
6213{
5fa1256f 6214 struct scanned_var *var;
6215 char line[256] = { 0, };
61e29183 6216 char words[3][256];
5fa1256f 6217 char *p = NULL;
6218 int wordc;
61e29183 6219 int l;
5fa1256f 6220
6221 while (!feof(fasm))
6222 {
6223 // skip to next data section
6224 while (my_fgets(line, sizeof(line), fasm))
6225 {
6226 asmln++;
6227
6228 p = sskip(line);
6229 if (*p == 0 || *p == ';')
6230 continue;
6231
6232 p = sskip(next_word_s(words[0], sizeof(words[0]), p));
6233 if (*p == 0 || *p == ';')
6234 continue;
6235
6236 if (*p != 's' || !IS_START(p, "segment para public"))
6237 continue;
6238
6239 break;
6240 }
6241
6242 if (p == NULL || !IS_START(p, "segment para public"))
6243 break;
6244 p = sskip(p + 19);
6245
6246 if (!IS_START(p, "'DATA'"))
6247 continue;
6248
6249 // now process it
6250 while (my_fgets(line, sizeof(line), fasm))
6251 {
6252 asmln++;
6253
6254 p = line;
6255 if (my_isblank(*p))
6256 continue;
6257
6258 p = sskip(p);
6259 if (*p == 0 || *p == ';')
6260 continue;
6261
6262 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
6263 words[wordc][0] = 0;
6264 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
6265 if (*p == 0 || *p == ';') {
6266 wordc++;
6267 break;
6268 }
6269 }
6270
6271 if (wordc == 2 && IS(words[1], "ends"))
6272 break;
61e29183 6273 if (wordc < 2)
6274 continue;
5fa1256f 6275
6276 if ((hg_var_cnt & 0xff) == 0) {
6277 hg_vars = realloc(hg_vars, sizeof(hg_vars[0])
6278 * (hg_var_cnt + 0x100));
6279 my_assert_not(hg_vars, NULL);
6280 memset(hg_vars + hg_var_cnt, 0, sizeof(hg_vars[0]) * 0x100);
6281 }
6282
6283 var = &hg_vars[hg_var_cnt++];
6284 snprintf(var->name, sizeof(var->name), "%s", words[0]);
6285
6286 // maybe already in seed header?
c0de9015 6287 var->pp = proto_parse(g_fhdr, var->name, 1);
6288 if (var->pp != NULL) {
6289 if (var->pp->is_fptr) {
5fa1256f 6290 var->lmod = OPLM_DWORD;
6291 //var->is_ptr = 1;
6292 }
c0de9015 6293 else if (var->pp->is_func)
6294 aerr("func?\n");
6295 else if (!guess_lmod_from_c_type(&var->lmod, &var->pp->type))
5fa1256f 6296 aerr("unhandled C type '%s' for '%s'\n",
c0de9015 6297 var->pp->type.name, var->name);
5fa1256f 6298
6299 var->is_seeded = 1;
6300 continue;
6301 }
6302
6303 if (IS(words[1], "dd"))
6304 var->lmod = OPLM_DWORD;
6305 else if (IS(words[1], "dw"))
6306 var->lmod = OPLM_WORD;
61e29183 6307 else if (IS(words[1], "db")) {
5fa1256f 6308 var->lmod = OPLM_BYTE;
61e29183 6309 if (wordc >= 3 && (l = strlen(words[2])) > 4) {
6310 if (words[2][0] == '\'' && IS(words[2] + l - 2, ",0"))
6311 var->is_c_str = 1;
6312 }
6313 }
5fa1256f 6314 else if (IS(words[1], "dq"))
6315 var->lmod = OPLM_QWORD;
6316 //else if (IS(words[1], "dt"))
6317 else
6318 aerr("type '%s' not known\n", words[1]);
6319 }
6320 }
6321
6322 rewind(fasm);
6323 asmln = 0;
6324}
6325
6326static void set_label(int i, const char *name)
6327{
6328 const char *p;
6329 int len;
6330
6331 len = strlen(name);
6332 p = strchr(name, ':');
6333 if (p != NULL)
6334 len = p - name;
6335
6336 if (g_labels[i] != NULL && !IS_START(g_labels[i], "algn_"))
6337 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
6338 g_labels[i] = realloc(g_labels[i], len + 1);
6339 my_assert_not(g_labels[i], NULL);
6340 memcpy(g_labels[i], name, len);
6341 g_labels[i][len] = 0;
6342}
6343
e56ab892 6344struct chunk_item {
6345 char *name;
6346 long fptr;
de50b98b 6347 int asmln;
e56ab892 6348};
6349
cdfaeed7 6350static struct chunk_item *func_chunks;
6351static int func_chunk_cnt;
6352static int func_chunk_alloc;
6353
6354static void add_func_chunk(FILE *fasm, const char *name, int line)
6355{
6356 if (func_chunk_cnt >= func_chunk_alloc) {
6357 func_chunk_alloc *= 2;
6358 func_chunks = realloc(func_chunks,
6359 func_chunk_alloc * sizeof(func_chunks[0]));
6360 my_assert_not(func_chunks, NULL);
6361 }
6362 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
6363 func_chunks[func_chunk_cnt].name = strdup(name);
6364 func_chunks[func_chunk_cnt].asmln = line;
6365 func_chunk_cnt++;
6366}
6367
e56ab892 6368static int cmp_chunks(const void *p1, const void *p2)
6369{
6370 const struct chunk_item *c1 = p1, *c2 = p2;
6371 return strcmp(c1->name, c2->name);
6372}
6373
bfa4a6ee 6374static int cmpstringp(const void *p1, const void *p2)
6375{
6376 return strcmp(*(char * const *)p1, *(char * const *)p2);
6377}
6378
cdfaeed7 6379static void scan_ahead(FILE *fasm)
6380{
6381 char words[2][256];
6382 char line[256];
6383 long oldpos;
6384 int oldasmln;
6385 int wordc;
6386 char *p;
6387 int i;
6388
6389 oldpos = ftell(fasm);
6390 oldasmln = asmln;
6391
5fa1256f 6392 while (my_fgets(line, sizeof(line), fasm))
cdfaeed7 6393 {
6394 wordc = 0;
6395 asmln++;
6396
6397 p = sskip(line);
6398 if (*p == 0)
6399 continue;
6400
6401 if (*p == ';')
6402 {
6403 // get rid of random tabs
6404 for (i = 0; line[i] != 0; i++)
6405 if (line[i] == '\t')
6406 line[i] = ' ';
6407
6408 if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
6409 {
6410 p += 30;
6411 next_word(words[0], sizeof(words[0]), p);
6412 if (words[0][0] == 0)
6413 aerr("missing name for func chunk?\n");
6414
6415 add_func_chunk(fasm, words[0], asmln);
6416 }
46b388c2 6417 else if (IS_START(p, "; sctend"))
6418 break;
6419
cdfaeed7 6420 continue;
6421 } // *p == ';'
6422
6423 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
6424 words[wordc][0] = 0;
6425 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
6426 if (*p == 0 || *p == ';') {
6427 wordc++;
6428 break;
6429 }
6430 }
6431
6432 if (wordc == 2 && IS(words[1], "ends"))
6433 break;
6434 }
6435
6436 fseek(fasm, oldpos, SEEK_SET);
6437 asmln = oldasmln;
6438}
6439
91977a1c 6440int main(int argc, char *argv[])
6441{
06c5d854 6442 FILE *fout, *fasm, *frlist;
4c45fa73 6443 struct parsed_data *pd = NULL;
6444 int pd_alloc = 0;
6445 char **rlist = NULL;
6446 int rlist_len = 0;
6447 int rlist_alloc = 0;
e56ab892 6448 int func_chunks_used = 0;
6449 int func_chunks_sorted = 0;
e56ab892 6450 int func_chunk_i = -1;
6451 long func_chunk_ret = 0;
de50b98b 6452 int func_chunk_ret_ln = 0;
cdfaeed7 6453 int scanned_ahead = 0;
91977a1c 6454 char line[256];
a2c1d768 6455 char words[20][256];
4c45fa73 6456 enum opr_lenmod lmod;
ddaf8bd7 6457 char *sctproto = NULL;
91977a1c 6458 int in_func = 0;
4c45fa73 6459 int pending_endp = 0;
bfa4a6ee 6460 int skip_func = 0;
940e8e66 6461 int skip_warned = 0;
91977a1c 6462 int eq_alloc;
bfa4a6ee 6463 int verbose = 0;
1f84f6b3 6464 int multi_seg = 0;
46b388c2 6465 int end = 0;
bfa4a6ee 6466 int arg_out;
89ff3147 6467 int arg;
91977a1c 6468 int pi = 0;
e56ab892 6469 int i, j;
6470 int ret, len;
91977a1c 6471 char *p;
6472 int wordc;
6473
89ff3147 6474 for (arg = 1; arg < argc; arg++) {
6475 if (IS(argv[arg], "-v"))
6476 verbose = 1;
6477 else if (IS(argv[arg], "-rf"))
6478 g_allow_regfunc = 1;
1f84f6b3 6479 else if (IS(argv[arg], "-m"))
6480 multi_seg = 1;
92d715b6 6481 else if (IS(argv[arg], "-hdr"))
9af2d373 6482 g_header_mode = g_quiet_pp = g_allow_regfunc = 1;
89ff3147 6483 else
6484 break;
bfa4a6ee 6485 }
6486
6487 if (argc < arg + 3) {
315b77eb 6488 printf("usage:\n%s [-v] [-rf] [-m] <.c> <.asm> <hdr.h> [rlist]*\n"
6489 "%s -hdr <out.h> <.asm> <seed.h> [rlist]*\n"
6490 "options:\n"
6491 " -hdr - header generation mode\n"
6492 " -rf - allow unannotated indirect calls\n"
6493 " -m - allow multiple .text sections\n"
6494 "[rlist] is a file with function names to skip,"
6495 " one per line\n",
92d715b6 6496 argv[0], argv[0]);
91977a1c 6497 return 1;
6498 }
6499
bfa4a6ee 6500 arg_out = arg++;
91977a1c 6501
bfa4a6ee 6502 asmfn = argv[arg++];
91977a1c 6503 fasm = fopen(asmfn, "r");
6504 my_assert_not(fasm, NULL);
6505
bfa4a6ee 6506 hdrfn = argv[arg++];
06c5d854 6507 g_fhdr = fopen(hdrfn, "r");
6508 my_assert_not(g_fhdr, NULL);
bfa4a6ee 6509
6510 rlist_alloc = 64;
6511 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
6512 my_assert_not(rlist, NULL);
6513 // needs special handling..
6514 rlist[rlist_len++] = "__alloca_probe";
6515
e56ab892 6516 func_chunk_alloc = 32;
6517 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
6518 my_assert_not(func_chunks, NULL);
6519
a2c1d768 6520 memset(words, 0, sizeof(words));
6521
bfa4a6ee 6522 for (; arg < argc; arg++) {
6523 frlist = fopen(argv[arg], "r");
6524 my_assert_not(frlist, NULL);
6525
5fa1256f 6526 while (my_fgets(line, sizeof(line), frlist)) {
bfa4a6ee 6527 p = sskip(line);
1cd4a663 6528 if (*p == 0 || *p == ';')
6529 continue;
6530 if (*p == '#') {
89ff3147 6531 if (IS_START(p, "#if 0")
6532 || (g_allow_regfunc && IS_START(p, "#if NO_REGFUNC")))
6533 {
1cd4a663 6534 skip_func = 1;
89ff3147 6535 }
1cd4a663 6536 else if (IS_START(p, "#endif"))
6537 skip_func = 0;
6538 continue;
6539 }
6540 if (skip_func)
bfa4a6ee 6541 continue;
6542
6543 p = next_word(words[0], sizeof(words[0]), p);
6544 if (words[0][0] == 0)
6545 continue;
6546
6547 if (rlist_len >= rlist_alloc) {
6548 rlist_alloc = rlist_alloc * 2 + 64;
6549 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
6550 my_assert_not(rlist, NULL);
6551 }
6552 rlist[rlist_len++] = strdup(words[0]);
6553 }
1cd4a663 6554 skip_func = 0;
bfa4a6ee 6555
6556 fclose(frlist);
6557 frlist = NULL;
6558 }
6559
6560 if (rlist_len > 0)
6561 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
6562
6563 fout = fopen(argv[arg_out], "w");
91977a1c 6564 my_assert_not(fout, NULL);
6565
6566 eq_alloc = 128;
6567 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
6568 my_assert_not(g_eqs, NULL);
6569
4c45fa73 6570 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
6571 g_label_refs[i].i = -1;
6572 g_label_refs[i].next = NULL;
6573 }
6574
5fa1256f 6575 if (g_header_mode)
6576 scan_variables(fasm);
6577
6578 while (my_fgets(line, sizeof(line), fasm))
91977a1c 6579 {
4c45fa73 6580 wordc = 0;
91977a1c 6581 asmln++;
6582
6583 p = sskip(line);
1bafb621 6584 if (*p == 0)
91977a1c 6585 continue;
6586
de50b98b 6587 // get rid of random tabs
6588 for (i = 0; line[i] != 0; i++)
6589 if (line[i] == '\t')
6590 line[i] = ' ';
6591
e56ab892 6592 if (*p == ';')
6593 {
e56ab892 6594 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
6595 goto do_pending_endp; // eww..
6596
6597 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
6598 {
6599 static const char *attrs[] = {
6600 "bp-based frame",
6601 "library function",
6602 "static",
6603 "noreturn",
6604 "thunk",
6605 "fpd=",
6606 };
6607
6608 // parse IDA's attribute-list comment
6609 g_ida_func_attr = 0;
6610 p = sskip(p + 13);
6611
6612 for (; *p != 0; p = sskip(p)) {
6613 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
6614 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
6615 g_ida_func_attr |= 1 << i;
6616 p += strlen(attrs[i]);
6617 break;
6618 }
6619 }
6620 if (i == ARRAY_SIZE(attrs)) {
6621 anote("unparsed IDA attr: %s\n", p);
1bafb621 6622 break;
6623 }
e56ab892 6624 if (IS(attrs[i], "fpd=")) {
6625 p = next_word(words[0], sizeof(words[0]), p);
6626 // ignore for now..
6627 }
1bafb621 6628 }
e56ab892 6629 }
6630 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
6631 {
6632 p += 30;
6633 next_word(words[0], sizeof(words[0]), p);
6634 if (words[0][0] == 0)
cdfaeed7 6635 aerr("missing name for func chunk?\n");
6636
6637 if (!scanned_ahead) {
6638 add_func_chunk(fasm, words[0], asmln);
6639 func_chunks_sorted = 0;
e56ab892 6640 }
e56ab892 6641 }
6642 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
6643 {
6644 if (func_chunk_i >= 0) {
6645 if (func_chunk_i < func_chunk_cnt
6646 && IS(func_chunks[func_chunk_i].name, g_func))
6647 {
6648 // move on to next chunk
6649 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
6650 if (ret)
6651 aerr("seek failed for '%s' chunk #%d\n",
6652 g_func, func_chunk_i);
de50b98b 6653 asmln = func_chunks[func_chunk_i].asmln;
e56ab892 6654 func_chunk_i++;
6655 }
6656 else {
6657 if (func_chunk_ret == 0)
6658 aerr("no return from chunk?\n");
6659 fseek(fasm, func_chunk_ret, SEEK_SET);
de50b98b 6660 asmln = func_chunk_ret_ln;
e56ab892 6661 func_chunk_ret = 0;
6662 pending_endp = 1;
6663 }
1bafb621 6664 }
e56ab892 6665 }
6666 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
6667 func_chunks_used = 1;
6668 p += 20;
6669 if (IS_START(g_func, "sub_")) {
6670 unsigned long addr = strtoul(p, NULL, 16);
6671 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
cdfaeed7 6672 if (addr > f_addr && !scanned_ahead) {
94d447fb 6673 anote("scan_ahead caused by '%s', addr %lx\n",
6674 g_func, addr);
cdfaeed7 6675 scan_ahead(fasm);
6676 scanned_ahead = 1;
6677 func_chunks_sorted = 0;
6678 }
1bafb621 6679 }
6680 }
6681 continue;
e56ab892 6682 } // *p == ';'
1bafb621 6683
06c5d854 6684parse_words:
a2c1d768 6685 for (i = wordc; i < ARRAY_SIZE(words); i++)
6686 words[i][0] = 0;
cdfaeed7 6687 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
bfa4a6ee 6688 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
91977a1c 6689 if (*p == 0 || *p == ';') {
6690 wordc++;
6691 break;
6692 }
6693 }
a2c1d768 6694 if (*p != 0 && *p != ';')
6695 aerr("too many words\n");
91977a1c 6696
06c5d854 6697 // alow asm patches in comments
ddaf8bd7 6698 if (*p == ';') {
6699 if (IS_START(p, "; sctpatch:")) {
6700 p = sskip(p + 11);
6701 if (*p == 0 || *p == ';')
6702 continue;
6703 goto parse_words; // lame
6704 }
6705 if (IS_START(p, "; sctproto:")) {
6706 sctproto = strdup(p + 11);
6707 }
46b388c2 6708 else if (IS_START(p, "; sctend")) {
6709 end = 1;
6710 if (!pending_endp)
6711 break;
6712 }
06c5d854 6713 }
6714
91977a1c 6715 if (wordc == 0) {
6716 // shouldn't happen
6717 awarn("wordc == 0?\n");
6718 continue;
6719 }
6720
6721 // don't care about this:
6722 if (words[0][0] == '.'
6723 || IS(words[0], "include")
6724 || IS(words[0], "assume") || IS(words[1], "segment")
6725 || IS(words[0], "align"))
6726 {
6727 continue;
6728 }
6729
4c45fa73 6730do_pending_endp:
6731 // do delayed endp processing to collect switch jumptables
6732 if (pending_endp) {
30c8c549 6733 if (in_func && !g_skip_func && !end && wordc >= 2
4c45fa73 6734 && ((words[0][0] == 'd' && words[0][2] == 0)
6735 || (words[1][0] == 'd' && words[1][2] == 0)))
6736 {
6737 i = 1;
6738 if (words[1][0] == 'd' && words[1][2] == 0) {
6739 // label
6740 if (g_func_pd_cnt >= pd_alloc) {
6741 pd_alloc = pd_alloc * 2 + 16;
6742 g_func_pd = realloc(g_func_pd,
6743 sizeof(g_func_pd[0]) * pd_alloc);
6744 my_assert_not(g_func_pd, NULL);
6745 }
6746 pd = &g_func_pd[g_func_pd_cnt];
6747 g_func_pd_cnt++;
6748 memset(pd, 0, sizeof(*pd));
6749 strcpy(pd->label, words[0]);
6750 pd->type = OPT_CONST;
6751 pd->lmod = lmod_from_directive(words[1]);
6752 i = 2;
6753 }
6754 else {
da87ae38 6755 if (pd == NULL) {
6756 if (verbose)
6757 anote("skipping alignment byte?\n");
6758 continue;
6759 }
4c45fa73 6760 lmod = lmod_from_directive(words[0]);
6761 if (lmod != pd->lmod)
6762 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
6763 }
6764
6765 if (pd->count_alloc < pd->count + wordc) {
6766 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
6767 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
6768 my_assert_not(pd->d, NULL);
6769 }
6770 for (; i < wordc; i++) {
6771 if (IS(words[i], "offset")) {
6772 pd->type = OPT_OFFSET;
6773 i++;
6774 }
6775 p = strchr(words[i], ',');
6776 if (p != NULL)
6777 *p = 0;
6778 if (pd->type == OPT_OFFSET)
6779 pd->d[pd->count].u.label = strdup(words[i]);
6780 else
6781 pd->d[pd->count].u.val = parse_number(words[i]);
6782 pd->d[pd->count].bt_i = -1;
6783 pd->count++;
6784 }
6785 continue;
6786 }
6787
30c8c549 6788 if (in_func && !g_skip_func) {
9af2d373 6789 if (g_header_mode)
92d715b6 6790 gen_hdr(g_func, pi);
6791 else
6792 gen_func(fout, g_fhdr, g_func, pi);
6793 }
4c45fa73 6794
6795 pending_endp = 0;
6796 in_func = 0;
6797 g_ida_func_attr = 0;
6798 skip_warned = 0;
30c8c549 6799 g_skip_func = 0;
4c45fa73 6800 g_func[0] = 0;
e56ab892 6801 func_chunks_used = 0;
6802 func_chunk_i = -1;
4c45fa73 6803 if (pi != 0) {
6804 memset(&ops, 0, pi * sizeof(ops[0]));
d7857c3a 6805 clear_labels(pi);
4c45fa73 6806 pi = 0;
6807 }
6808 g_eqcnt = 0;
6809 for (i = 0; i < g_func_pd_cnt; i++) {
6810 pd = &g_func_pd[i];
6811 if (pd->type == OPT_OFFSET) {
6812 for (j = 0; j < pd->count; j++)
6813 free(pd->d[j].u.label);
6814 }
6815 free(pd->d);
6816 pd->d = NULL;
6817 }
6818 g_func_pd_cnt = 0;
6819 pd = NULL;
46b388c2 6820
6821 if (end)
6822 break;
4c45fa73 6823 if (wordc == 0)
6824 continue;
6825 }
6826
91977a1c 6827 if (IS(words[1], "proc")) {
6828 if (in_func)
6829 aerr("proc '%s' while in_func '%s'?\n",
6830 words[0], g_func);
bfa4a6ee 6831 p = words[0];
ddaf8bd7 6832 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
30c8c549 6833 g_skip_func = 1;
91977a1c 6834 strcpy(g_func, words[0]);
d4e3b5db 6835 set_label(0, words[0]);
91977a1c 6836 in_func = 1;
6837 continue;
6838 }
6839
e56ab892 6840 if (IS(words[1], "endp"))
6841 {
91977a1c 6842 if (!in_func)
6843 aerr("endp '%s' while not in_func?\n", words[0]);
6844 if (!IS(g_func, words[0]))
6845 aerr("endp '%s' while in_func '%s'?\n",
6846 words[0], g_func);
bfa4a6ee 6847
ddaf8bd7 6848 if ((g_ida_func_attr & IDAFA_THUNK) && pi == 1
6849 && ops[0].op == OP_JMP && ops[0].operand[0].had_ds)
6850 {
6851 // import jump
30c8c549 6852 g_skip_func = 1;
ddaf8bd7 6853 }
6854
30c8c549 6855 if (!g_skip_func && func_chunks_used) {
e56ab892 6856 // start processing chunks
6857 struct chunk_item *ci, key = { g_func, 0 };
6858
6859 func_chunk_ret = ftell(fasm);
de50b98b 6860 func_chunk_ret_ln = asmln;
e56ab892 6861 if (!func_chunks_sorted) {
6862 qsort(func_chunks, func_chunk_cnt,
6863 sizeof(func_chunks[0]), cmp_chunks);
6864 func_chunks_sorted = 1;
6865 }
6866 ci = bsearch(&key, func_chunks, func_chunk_cnt,
6867 sizeof(func_chunks[0]), cmp_chunks);
6868 if (ci == NULL)
6869 aerr("'%s' needs chunks, but none found\n", g_func);
6870 func_chunk_i = ci - func_chunks;
6871 for (; func_chunk_i > 0; func_chunk_i--)
6872 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
6873 break;
6874
6875 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
6876 if (ret)
6877 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
de50b98b 6878 asmln = func_chunks[func_chunk_i].asmln;
e56ab892 6879 func_chunk_i++;
6880 continue;
6881 }
4c45fa73 6882 pending_endp = 1;
91977a1c 6883 continue;
6884 }
6885
1f84f6b3 6886 if (wordc == 2 && IS(words[1], "ends")) {
46b388c2 6887 if (!multi_seg) {
6888 end = 1;
6889 if (pending_endp)
6890 goto do_pending_endp;
1f84f6b3 6891 break;
46b388c2 6892 }
1f84f6b3 6893
6894 // scan for next text segment
5fa1256f 6895 while (my_fgets(line, sizeof(line), fasm)) {
1f84f6b3 6896 asmln++;
6897 p = sskip(line);
6898 if (*p == 0 || *p == ';')
6899 continue;
6900
6901 if (strstr(p, "segment para public 'CODE' use32"))
6902 break;
6903 }
6904
6905 continue;
6906 }
a2c1d768 6907
bfa4a6ee 6908 p = strchr(words[0], ':');
6909 if (p != NULL) {
d4e3b5db 6910 set_label(pi, words[0]);
bfa4a6ee 6911 continue;
6912 }
6913
30c8c549 6914 if (!in_func || g_skip_func) {
6915 if (!skip_warned && !g_skip_func && g_labels[pi] != NULL) {
bfa4a6ee 6916 if (verbose)
6917 anote("skipping from '%s'\n", g_labels[pi]);
6918 skip_warned = 1;
6919 }
d7857c3a 6920 free(g_labels[pi]);
6921 g_labels[pi] = NULL;
bfa4a6ee 6922 continue;
6923 }
6924
ddaf8bd7 6925 if (wordc > 1 && IS(words[1], "="))
6926 {
91977a1c 6927 if (wordc != 5)
6928 aerr("unhandled equ, wc=%d\n", wordc);
6929 if (g_eqcnt >= eq_alloc) {
6930 eq_alloc *= 2;
6931 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
6932 my_assert_not(g_eqs, NULL);
6933 }
6934
6935 len = strlen(words[0]);
6936 if (len > sizeof(g_eqs[0].name) - 1)
6937 aerr("equ name too long: %d\n", len);
6938 strcpy(g_eqs[g_eqcnt].name, words[0]);
6939
6940 if (!IS(words[3], "ptr"))
6941 aerr("unhandled equ\n");
6942 if (IS(words[2], "dword"))
6943 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
6944 else if (IS(words[2], "word"))
6945 g_eqs[g_eqcnt].lmod = OPLM_WORD;
6946 else if (IS(words[2], "byte"))
6947 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
90307a99 6948 else if (IS(words[2], "qword"))
6949 g_eqs[g_eqcnt].lmod = OPLM_QWORD;
91977a1c 6950 else
6951 aerr("bad lmod: '%s'\n", words[2]);
6952
6953 g_eqs[g_eqcnt].offset = parse_number(words[4]);
6954 g_eqcnt++;
6955 continue;
6956 }
6957
6958 if (pi >= ARRAY_SIZE(ops))
6959 aerr("too many ops\n");
6960
91977a1c 6961 parse_op(&ops[pi], words, wordc);
ddaf8bd7 6962
6963 if (sctproto != NULL) {
8eb12e72 6964 if (ops[pi].op == OP_CALL || ops[pi].op == OP_JMP)
ddaf8bd7 6965 ops[pi].datap = sctproto;
6966 sctproto = NULL;
6967 }
91977a1c 6968 pi++;
91977a1c 6969 }
6970
9af2d373 6971 if (g_header_mode)
92d715b6 6972 output_hdr(fout);
6973
91977a1c 6974 fclose(fout);
6975 fclose(fasm);
06c5d854 6976 fclose(g_fhdr);
91977a1c 6977
6978 return 0;
c36e914d 6979}
91977a1c 6980
6981// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.