notaz.gp2x.de / ia32rtools.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
translate: yet more work on sp tracking
[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"
9ea60b8d 16#include "common.h"
c36e914d 17
18#define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
19#define IS(w, y) !strcmp(w, y)
39b168b8 20#define IS_START(w, y) !strncmp(w, y, strlen(y))
c36e914d 21
22#include "protoparse.h"
23
06c5d854 24static const char *asmfn;
c36e914d 25static int asmln;
06c5d854 26static FILE *g_fhdr;
c36e914d 27
940e8e66 28#define anote(fmt, ...) \
29 printf("%s:%d: note: " fmt, asmfn, asmln, ##__VA_ARGS__)
c36e914d 30#define awarn(fmt, ...) \
940e8e66 31 printf("%s:%d: warning: " fmt, asmfn, asmln, ##__VA_ARGS__)
c36e914d 32#define aerr(fmt, ...) do { \
940e8e66 33 printf("%s:%d: error: " fmt, asmfn, asmln, ##__VA_ARGS__); \
33c35af6 34 fcloseall(); \
c36e914d 35 exit(1); \
36} while (0)
37
054f95b2 38#include "masm_tools.h"
39
69a3cdfc 40enum op_flags {
5e49b270 41 OPF_RMD = (1 << 0), /* removed from code generation */
87bf6cec 42 OPF_DATA = (1 << 1), /* data processing - writes to dst opr */
43 OPF_FLAGS = (1 << 2), /* sets flags */
5c024ef7 44 OPF_JMP = (1 << 3), /* branch, call */
45 OPF_CJMP = (1 << 4), /* cond. branch (cc or jecxz) */
46 OPF_CC = (1 << 5), /* uses flags */
47 OPF_TAIL = (1 << 6), /* ret or tail call */
48 OPF_RSAVE = (1 << 7), /* push/pop is local reg save/load */
49 OPF_REP = (1 << 8), /* prefixed by rep */
50 OPF_REPZ = (1 << 9), /* rep is repe/repz */
51 OPF_REPNZ = (1 << 10), /* rep is repne/repnz */
9af2d373 52 OPF_FARG = (1 << 11), /* push collected as func arg */
53 OPF_FARGNR = (1 << 12), /* push collected as func arg (no reuse) */
54 OPF_EBP_S = (1 << 13), /* ebp used as scratch here, not BP */
55 OPF_DF = (1 << 14), /* DF flag set */
56 OPF_ATAIL = (1 << 15), /* tail call with reused arg frame */
57 OPF_32BIT = (1 << 16), /* 32bit division */
58 OPF_LOCK = (1 << 17), /* op has lock prefix */
59 OPF_VAPUSH = (1 << 18), /* vararg ptr push (as call arg) */
26677139 60 OPF_DONE = (1 << 19), /* already fully handled by analysis */
25a330eb 61 OPF_PPUSH = (1 << 20), /* part of complex push-pop graph */
c36e914d 62};
63
64enum op_op {
65 OP_INVAL,
33c35af6 66 OP_NOP,
c36e914d 67 OP_PUSH,
68 OP_POP,
591721d7 69 OP_LEAVE,
c36e914d 70 OP_MOV,
850c9265 71 OP_LEA,
72 OP_MOVZX,
73 OP_MOVSX,
108e9fe3 74 OP_XCHG,
850c9265 75 OP_NOT,
5101a5f9 76 OP_CDQ,
092f64e1 77 OP_LODS,
33c35af6 78 OP_STOS,
d4e3b5db 79 OP_MOVS,
7ba45c34 80 OP_CMPS,
591721d7 81 OP_SCAS,
82 OP_STD,
83 OP_CLD,
c36e914d 84 OP_RET,
85 OP_ADD,
91977a1c 86 OP_SUB,
850c9265 87 OP_AND,
88 OP_OR,
89 OP_XOR,
90 OP_SHL,
91 OP_SHR,
92 OP_SAR,
3b2f4044 93 OP_SHRD,
d4e3b5db 94 OP_ROL,
95 OP_ROR,
cb090db0 96 OP_RCL,
97 OP_RCR,
69a3cdfc 98 OP_ADC,
850c9265 99 OP_SBB,
1f84f6b3 100 OP_BSF,
850c9265 101 OP_INC,
102 OP_DEC,
5101a5f9 103 OP_NEG,
850c9265 104 OP_MUL,
105 OP_IMUL,
5101a5f9 106 OP_DIV,
107 OP_IDIV,
c36e914d 108 OP_TEST,
109 OP_CMP,
110 OP_CALL,
111 OP_JMP,
5c024ef7 112 OP_JECXZ,
092f64e1 113 OP_JCC,
114 OP_SCC,
90307a99 115 // x87
116 // mmx
117 OP_EMMS,
30c8c549 118 // undefined
2c10ea1f 119 OP_UD2,
c36e914d 120};
121
122enum opr_type {
87bf6cec 123 OPT_UNSPEC,
124 OPT_REG,
125 OPT_REGMEM,
126 OPT_LABEL,
850c9265 127 OPT_OFFSET,
87bf6cec 128 OPT_CONST,
c36e914d 129};
130
2b43685d 131// must be sorted (larger len must be further in enum)
c36e914d 132enum opr_lenmod {
91977a1c 133 OPLM_UNSPEC,
134 OPLM_BYTE,
135 OPLM_WORD,
136 OPLM_DWORD,
90307a99 137 OPLM_QWORD,
c36e914d 138};
139
850c9265 140#define MAX_OPERANDS 3
92d715b6 141#define NAMELEN 112
c36e914d 142
143struct parsed_opr {
91977a1c 144 enum opr_type type;
145 enum opr_lenmod lmod;
7ba45c34 146 unsigned int is_ptr:1; // pointer in C
147 unsigned int is_array:1; // array in C
a3684be1 148 unsigned int type_from_var:1; // .. in header, sometimes wrong
2b43685d 149 unsigned int size_mismatch:1; // type override differs from C
150 unsigned int size_lt:1; // type override is larger than C
ddaf8bd7 151 unsigned int had_ds:1; // had ds: prefix
c7ed83dd 152 const struct parsed_proto *pp; // for OPT_LABEL
91977a1c 153 int reg;
154 unsigned int val;
92d715b6 155 char name[NAMELEN];
c36e914d 156};
157
158struct parsed_op {
91977a1c 159 enum op_op op;
160 struct parsed_opr operand[MAX_OPERANDS];
69a3cdfc 161 unsigned int flags;
092f64e1 162 unsigned char pfo;
163 unsigned char pfo_inv;
164 unsigned char operand_cnt;
3a5101d7 165 unsigned char p_argnum; // arg push: altered before call arg #
166 unsigned char p_arggrp; // arg push: arg group # for above
167 unsigned char p_argpass;// arg push: arg of host func
168 short p_argnext;// arg push: same arg pushed elsewhere or -1
69a3cdfc 169 int regmask_src; // all referensed regs
170 int regmask_dst;
940e8e66 171 int pfomask; // flagop: parsed_flag_op that can't be delayed
940e8e66 172 int cc_scratch; // scratch storage during analysis
4c45fa73 173 int bt_i; // branch target for branches
174 struct parsed_data *btj;// branch targets for jumptables
c7ed83dd 175 struct parsed_proto *pp;// parsed_proto for OP_CALL
91977a1c 176 void *datap;
8eb12e72 177 int asmln;
91977a1c 178};
179
69a3cdfc 180// datap:
25a330eb 181// OP_CALL - parser proto hint (str)
2c10ea1f 182// (OPF_CC) - points to one of (OPF_FLAGS) that affects cc op
25a330eb 183// OP_PUSH - points to OP_POP in complex push/pop graph
184// OP_POP - points to OP_PUSH in simple push/pop pair
69a3cdfc 185
91977a1c 186struct parsed_equ {
187 char name[64];
188 enum opr_lenmod lmod;
189 int offset;
c36e914d 190};
191
4c45fa73 192struct parsed_data {
193 char label[256];
194 enum opr_type type;
195 enum opr_lenmod lmod;
196 int count;
197 int count_alloc;
198 struct {
199 union {
200 char *label;
63df67be 201 unsigned int val;
4c45fa73 202 } u;
203 int bt_i;
204 } *d;
205};
206
207struct label_ref {
208 int i;
209 struct label_ref *next;
210};
211
1bafb621 212enum ida_func_attr {
213 IDAFA_BP_FRAME = (1 << 0),
214 IDAFA_LIB_FUNC = (1 << 1),
215 IDAFA_STATIC = (1 << 2),
216 IDAFA_NORETURN = (1 << 3),
217 IDAFA_THUNK = (1 << 4),
218 IDAFA_FPD = (1 << 5),
219};
220
3a5101d7 221// note: limited to 32k due to p_argnext
222#define MAX_OPS 4096
223#define MAX_ARG_GRP 2
c36e914d 224
225static struct parsed_op ops[MAX_OPS];
91977a1c 226static struct parsed_equ *g_eqs;
227static int g_eqcnt;
d7857c3a 228static char *g_labels[MAX_OPS];
4c45fa73 229static struct label_ref g_label_refs[MAX_OPS];
bd96f656 230static const struct parsed_proto *g_func_pp;
4c45fa73 231static struct parsed_data *g_func_pd;
232static int g_func_pd_cnt;
91977a1c 233static char g_func[256];
234static char g_comment[256];
235static int g_bp_frame;
1bafb621 236static int g_sp_frame;
a2c1d768 237static int g_stack_frame_used;
1bafb621 238static int g_stack_fsz;
4c45fa73 239static int g_ida_func_attr;
30c8c549 240static int g_skip_func;
89ff3147 241static int g_allow_regfunc;
92d715b6 242static int g_quiet_pp;
9af2d373 243static int g_header_mode;
92d715b6 244
91977a1c 245#define ferr(op_, fmt, ...) do { \
bfacdc83 246 printf("%s:%d: error %u: [%s] '%s': " fmt, asmfn, (op_)->asmln, \
247 __LINE__, g_func, dump_op(op_), ##__VA_ARGS__); \
33c35af6 248 fcloseall(); \
91977a1c 249 exit(1); \
250} while (0)
de50b98b 251#define fnote(op_, fmt, ...) \
8eb12e72 252 printf("%s:%d: note: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
de50b98b 253 dump_op(op_), ##__VA_ARGS__)
91977a1c 254
26677139 255#define ferr_assert(op_, cond) do { \
256 if (!(cond)) ferr(op_, "assertion '%s' failed on ln :%d\n", #cond, \
257 __LINE__); \
258} while (0)
259
90307a99 260const char *regs_r32[] = {
261 "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp",
262 // not r32, but list here for easy parsing and printing
263 "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7",
264};
91977a1c 265const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" };
266const char *regs_r8l[] = { "al", "bl", "cl", "dl" };
267const char *regs_r8h[] = { "ah", "bh", "ch", "dh" };
268
269enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP };
270
69a3cdfc 271// possible basic comparison types (without inversion)
272enum parsed_flag_op {
273 PFO_O, // 0 OF=1
274 PFO_C, // 2 CF=1
275 PFO_Z, // 4 ZF=1
276 PFO_BE, // 6 CF=1||ZF=1
277 PFO_S, // 8 SF=1
278 PFO_P, // a PF=1
279 PFO_L, // c SF!=OF
280 PFO_LE, // e ZF=1||SF!=OF
281};
282
90307a99 283#define PFOB_O (1 << PFO_O)
284#define PFOB_C (1 << PFO_C)
285#define PFOB_Z (1 << PFO_Z)
286#define PFOB_S (1 << PFO_S)
287
69a3cdfc 288static const char *parsed_flag_op_names[] = {
289 "o", "c", "z", "be", "s", "p", "l", "le"
290};
291
91977a1c 292static int char_array_i(const char *array[], size_t len, const char *s)
293{
294 int i;
c36e914d 295
91977a1c 296 for (i = 0; i < len; i++)
297 if (IS(s, array[i]))
298 return i;
c36e914d 299
91977a1c 300 return -1;
301}
302
63df67be 303static void printf_number(char *buf, size_t buf_size,
304 unsigned long number)
91977a1c 305{
5101a5f9 306 // output in C-friendly form
307 snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
308}
91977a1c 309
fdd5548a 310static int check_segment_prefix(const char *s)
311{
312 if (s[0] == 0 || s[1] != 's' || s[2] != ':')
313 return 0;
314
315 switch (s[0]) {
316 case 'c': return 1;
317 case 'd': return 2;
318 case 's': return 3;
319 case 'e': return 4;
320 case 'f': return 5;
321 case 'g': return 6;
322 default: return 0;
323 }
324}
325
5101a5f9 326static int parse_reg(enum opr_lenmod *reg_lmod, const char *s)
327{
328 int reg;
91977a1c 329
5101a5f9 330 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s);
90307a99 331 if (reg >= 8) {
332 *reg_lmod = OPLM_QWORD;
333 return reg;
334 }
5101a5f9 335 if (reg >= 0) {
336 *reg_lmod = OPLM_DWORD;
337 return reg;
91977a1c 338 }
5101a5f9 339 reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s);
340 if (reg >= 0) {
341 *reg_lmod = OPLM_WORD;
342 return reg;
343 }
344 reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s);
345 if (reg >= 0) {
346 *reg_lmod = OPLM_BYTE;
347 return reg;
348 }
349 reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s);
350 if (reg >= 0) {
351 *reg_lmod = OPLM_BYTE;
352 return reg;
850c9265 353 }
354
355 return -1;
91977a1c 356}
357
5101a5f9 358static int parse_indmode(char *name, int *regmask, int need_c_cvt)
359{
360 enum opr_lenmod lmod;
361 char cvtbuf[256];
362 char *d = cvtbuf;
363 char *s = name;
364 char w[64];
365 long number;
366 int reg;
367 int c = 0;
368
369 *d = 0;
370
371 while (*s != 0) {
372 d += strlen(d);
373 while (my_isblank(*s))
374 s++;
375 for (; my_issep(*s); d++, s++)
376 *d = *s;
377 while (my_isblank(*s))
378 s++;
379 *d = 0;
380
fdd5548a 381 // skip '?s:' prefixes
382 if (check_segment_prefix(s))
87bf6cec 383 s += 3;
384
5101a5f9 385 s = next_idt(w, sizeof(w), s);
386 if (w[0] == 0)
387 break;
388 c++;
389
390 reg = parse_reg(&lmod, w);
391 if (reg >= 0) {
392 *regmask |= 1 << reg;
393 goto pass;
394 }
395
396 if ('0' <= w[0] && w[0] <= '9') {
397 number = parse_number(w);
398 printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number);
399 continue;
400 }
401
402 // probably some label/identifier - pass
403
404pass:
405 snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w);
406 }
407
408 if (need_c_cvt)
409 strcpy(name, cvtbuf);
410
411 return c;
412}
413
4c45fa73 414static int is_reg_in_str(const char *s)
415{
416 int i;
417
418 if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
419 return 0;
420
421 for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
422 if (!strncmp(s, regs_r32[i], 3))
423 return 1;
424
425 return 0;
426}
427
09bc6fd5 428static const char *parse_stack_el(const char *name, char *extra_reg,
429 int early_try)
d4e3b5db 430{
4c45fa73 431 const char *p, *p2, *s;
1bafb621 432 char *endp = NULL;
d4e3b5db 433 char buf[32];
1bafb621 434 long val;
d4e3b5db 435 int len;
436
09bc6fd5 437 if (g_bp_frame || early_try)
438 {
439 p = name;
440 if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
441 p += 4;
442 if (extra_reg != NULL) {
443 strncpy(extra_reg, name, 3);
444 extra_reg[4] = 0;
445 }
4c45fa73 446 }
4c45fa73 447
09bc6fd5 448 if (IS_START(p, "ebp+")) {
449 p += 4;
4c45fa73 450
09bc6fd5 451 p2 = strchr(p, '+');
452 if (p2 != NULL && is_reg_in_str(p)) {
453 if (extra_reg != NULL) {
454 strncpy(extra_reg, p, p2 - p);
455 extra_reg[p2 - p] = 0;
456 }
457 p = p2 + 1;
4c45fa73 458 }
4c45fa73 459
09bc6fd5 460 if (!('0' <= *p && *p <= '9'))
461 return p;
4c45fa73 462
09bc6fd5 463 return NULL;
464 }
4c45fa73 465 }
466
39b168b8 467 if (!IS_START(name, "esp+"))
d4e3b5db 468 return NULL;
469
037f4971 470 s = name + 4;
471 p = strchr(s, '+');
d4e3b5db 472 if (p) {
037f4971 473 if (is_reg_in_str(s)) {
474 if (extra_reg != NULL) {
475 strncpy(extra_reg, s, p - s);
476 extra_reg[p - s] = 0;
477 }
478 s = p + 1;
479 p = strchr(s, '+');
480 if (p == NULL)
481 aerr("%s IDA stackvar not set?\n", __func__);
482 }
1bafb621 483 if (!('0' <= *s && *s <= '9')) {
037f4971 484 aerr("%s IDA stackvar offset not set?\n", __func__);
d4e3b5db 485 return NULL;
1bafb621 486 }
487 if (s[0] == '0' && s[1] == 'x')
488 s += 2;
489 len = p - s;
d4e3b5db 490 if (len < sizeof(buf) - 1) {
1bafb621 491 strncpy(buf, s, len);
d4e3b5db 492 buf[len] = 0;
1bafb621 493 val = strtol(buf, &endp, 16);
494 if (val == 0 || *endp != 0) {
495 aerr("%s num parse fail for '%s'\n", __func__, buf);
496 return NULL;
497 }
d4e3b5db 498 }
499 p++;
500 }
501 else
502 p = name + 4;
503
504 if ('0' <= *p && *p <= '9')
505 return NULL;
506
507 return p;
508}
509
850c9265 510static int guess_lmod_from_name(struct parsed_opr *opr)
511{
512 if (!strncmp(opr->name, "dword_", 6)) {
513 opr->lmod = OPLM_DWORD;
514 return 1;
515 }
516 if (!strncmp(opr->name, "word_", 5)) {
517 opr->lmod = OPLM_WORD;
518 return 1;
519 }
520 if (!strncmp(opr->name, "byte_", 5)) {
521 opr->lmod = OPLM_BYTE;
522 return 1;
523 }
90307a99 524 if (!strncmp(opr->name, "qword_", 6)) {
525 opr->lmod = OPLM_QWORD;
526 return 1;
527 }
850c9265 528 return 0;
529}
530
3ebea2cf 531static int guess_lmod_from_c_type(enum opr_lenmod *lmod,
532 const struct parsed_type *c_type)
06c5d854 533{
06c5d854 534 static const char *dword_types[] = {
c0de9015 535 "uint32_t", "int", "_DWORD", "UINT_PTR", "DWORD",
4741fdfe 536 "WPARAM", "LPARAM", "UINT", "__int32",
09bc6fd5 537 "LONG", "HIMC", "BOOL", "size_t",
feb0ee5d 538 "float",
06c5d854 539 };
540 static const char *word_types[] = {
fe18709a 541 "uint16_t", "int16_t", "_WORD", "WORD",
2b43685d 542 "unsigned __int16", "__int16",
06c5d854 543 };
544 static const char *byte_types[] = {
2b43685d 545 "uint8_t", "int8_t", "char",
89ff3147 546 "unsigned __int8", "__int8", "BYTE", "_BYTE",
4741fdfe 547 "CHAR", "_UNKNOWN",
b4878d2b 548 // structures.. deal the same as with _UNKNOWN for now
549 "CRITICAL_SECTION",
06c5d854 550 };
3ebea2cf 551 const char *n;
06c5d854 552 int i;
553
3ebea2cf 554 if (c_type->is_ptr) {
555 *lmod = OPLM_DWORD;
06c5d854 556 return 1;
557 }
558
3ebea2cf 559 n = skip_type_mod(c_type->name);
06c5d854 560
3ebea2cf 561 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
562 if (IS(n, dword_types[i])) {
563 *lmod = OPLM_DWORD;
06c5d854 564 return 1;
565 }
566 }
567
568 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
3ebea2cf 569 if (IS(n, word_types[i])) {
570 *lmod = OPLM_WORD;
06c5d854 571 return 1;
572 }
573 }
574
575 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
3ebea2cf 576 if (IS(n, byte_types[i])) {
577 *lmod = OPLM_BYTE;
06c5d854 578 return 1;
579 }
580 }
581
06c5d854 582 return 0;
583}
584
c7ed83dd 585static char *default_cast_to(char *buf, size_t buf_size,
586 struct parsed_opr *opr)
587{
588 buf[0] = 0;
589
590 if (!opr->is_ptr)
591 return buf;
592 if (opr->pp == NULL || opr->pp->type.name == NULL
593 || opr->pp->is_fptr)
594 {
595 snprintf(buf, buf_size, "%s", "(void *)");
596 return buf;
597 }
598
599 snprintf(buf, buf_size, "(%s)", opr->pp->type.name);
600 return buf;
601}
602
4c45fa73 603static enum opr_type lmod_from_directive(const char *d)
604{
605 if (IS(d, "dd"))
606 return OPLM_DWORD;
607 else if (IS(d, "dw"))
608 return OPLM_WORD;
609 else if (IS(d, "db"))
610 return OPLM_BYTE;
611
612 aerr("unhandled directive: '%s'\n", d);
613 return OPLM_UNSPEC;
614}
615
5101a5f9 616static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
617 int *regmask)
618{
619 opr->type = OPT_REG;
620 opr->reg = reg;
621 opr->lmod = lmod;
622 *regmask |= 1 << reg;
623}
624
39b168b8 625static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
626 int *extra_offs);
87bf6cec 627
69a3cdfc 628static int parse_operand(struct parsed_opr *opr,
629 int *regmask, int *regmask_indirect,
1bafb621 630 char words[16][256], int wordc, int w, unsigned int op_flags)
c36e914d 631{
92d715b6 632 const struct parsed_proto *pp = NULL;
850c9265 633 enum opr_lenmod tmplmod;
63df67be 634 unsigned long number;
89ff3147 635 char buf[256];
850c9265 636 int ret, len;
1bafb621 637 int wordc_in;
89ff3147 638 char *p;
850c9265 639 int i;
c36e914d 640
1bafb621 641 if (w >= wordc)
642 aerr("parse_operand w %d, wordc %d\n", w, wordc);
c36e914d 643
1bafb621 644 opr->reg = xUNSPEC;
91977a1c 645
1bafb621 646 for (i = w; i < wordc; i++) {
647 len = strlen(words[i]);
648 if (words[i][len - 1] == ',') {
649 words[i][len - 1] = 0;
650 wordc = i + 1;
651 break;
652 }
653 }
c36e914d 654
1bafb621 655 wordc_in = wordc - w;
c36e914d 656
1bafb621 657 if ((op_flags & OPF_JMP) && wordc_in > 0
658 && !('0' <= words[w][0] && words[w][0] <= '9'))
659 {
660 const char *label = NULL;
661
662 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
663 && IS(words[w + 1], "ptr"))
664 label = words[w + 2];
665 else if (wordc_in == 2 && IS(words[w], "short"))
666 label = words[w + 1];
667 else if (wordc_in == 1
668 && strchr(words[w], '[') == NULL
669 && parse_reg(&tmplmod, words[w]) < 0)
670 label = words[w];
671
672 if (label != NULL) {
673 opr->type = OPT_LABEL;
fdd5548a 674 ret = check_segment_prefix(label);
675 if (ret != 0) {
676 if (ret >= 5)
677 aerr("fs/gs used\n");
ddaf8bd7 678 opr->had_ds = 1;
39b168b8 679 label += 3;
ddaf8bd7 680 }
1bafb621 681 strcpy(opr->name, label);
682 return wordc;
683 }
684 }
c36e914d 685
1bafb621 686 if (wordc_in >= 3) {
687 if (IS(words[w + 1], "ptr")) {
688 if (IS(words[w], "dword"))
689 opr->lmod = OPLM_DWORD;
690 else if (IS(words[w], "word"))
691 opr->lmod = OPLM_WORD;
692 else if (IS(words[w], "byte"))
693 opr->lmod = OPLM_BYTE;
90307a99 694 else if (IS(words[w], "qword"))
695 opr->lmod = OPLM_QWORD;
1bafb621 696 else
697 aerr("type parsing failed\n");
698 w += 2;
699 wordc_in = wordc - w;
700 }
701 }
702
39b168b8 703 if (wordc_in == 2) {
704 if (IS(words[w], "offset")) {
705 opr->type = OPT_OFFSET;
27ebfaed 706 opr->lmod = OPLM_DWORD;
39b168b8 707 strcpy(opr->name, words[w + 1]);
27ebfaed 708 pp = proto_parse(g_fhdr, opr->name, 1);
709 goto do_label;
39b168b8 710 }
711 if (IS(words[w], "(offset")) {
89ff3147 712 p = strchr(words[w + 1], ')');
39b168b8 713 if (p == NULL)
714 aerr("parse of bracketed offset failed\n");
715 *p = 0;
716 opr->type = OPT_OFFSET;
717 strcpy(opr->name, words[w + 1]);
718 return wordc;
719 }
1bafb621 720 }
c36e914d 721
1bafb621 722 if (wordc_in != 1)
850c9265 723 aerr("parse_operand 1 word expected\n");
c36e914d 724
fdd5548a 725 ret = check_segment_prefix(words[w]);
726 if (ret != 0) {
727 if (ret >= 5)
728 aerr("fs/gs used\n");
ddaf8bd7 729 opr->had_ds = 1;
89ff3147 730 memmove(words[w], words[w] + 3, strlen(words[w]) - 2);
ddaf8bd7 731 }
89ff3147 732 strcpy(opr->name, words[w]);
c36e914d 733
850c9265 734 if (words[w][0] == '[') {
735 opr->type = OPT_REGMEM;
736 ret = sscanf(words[w], "[%[^]]]", opr->name);
737 if (ret != 1)
738 aerr("[] parse failure\n");
87bf6cec 739
5101a5f9 740 parse_indmode(opr->name, regmask_indirect, 1);
09bc6fd5 741 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL, 1))
742 {
87bf6cec 743 // might be an equ
d4e3b5db 744 struct parsed_equ *eq =
09bc6fd5 745 equ_find(NULL, parse_stack_el(opr->name, NULL, 1), &i);
87bf6cec 746 if (eq)
747 opr->lmod = eq->lmod;
748 }
850c9265 749 return wordc;
750 }
751 else if (strchr(words[w], '[')) {
752 // label[reg] form
89ff3147 753 p = strchr(words[w], '[');
850c9265 754 opr->type = OPT_REGMEM;
89ff3147 755 parse_indmode(p, regmask_indirect, 0);
756 strncpy(buf, words[w], p - words[w]);
757 buf[p - words[w]] = 0;
758 pp = proto_parse(g_fhdr, buf, 1);
759 goto do_label;
850c9265 760 }
761 else if (('0' <= words[w][0] && words[w][0] <= '9')
762 || words[w][0] == '-')
763 {
5101a5f9 764 number = parse_number(words[w]);
91977a1c 765 opr->type = OPT_CONST;
5101a5f9 766 opr->val = number;
767 printf_number(opr->name, sizeof(opr->name), number);
91977a1c 768 return wordc;
850c9265 769 }
c36e914d 770
5101a5f9 771 ret = parse_reg(&tmplmod, opr->name);
772 if (ret >= 0) {
773 setup_reg_opr(opr, ret, tmplmod, regmask);
850c9265 774 return wordc;
775 }
776
777 // most likely var in data segment
778 opr->type = OPT_LABEL;
92d715b6 779 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
89ff3147 780
781do_label:
bd96f656 782 if (pp != NULL) {
1cd4a663 783 if (pp->is_fptr || pp->is_func) {
06c5d854 784 opr->lmod = OPLM_DWORD;
785 opr->is_ptr = 1;
786 }
2b43685d 787 else {
840257f6 788 tmplmod = OPLM_UNSPEC;
bd96f656 789 if (!guess_lmod_from_c_type(&tmplmod, &pp->type))
2b43685d 790 anote("unhandled C type '%s' for '%s'\n",
bd96f656 791 pp->type.name, opr->name);
2b43685d 792
a3684be1 793 if (opr->lmod == OPLM_UNSPEC) {
2b43685d 794 opr->lmod = tmplmod;
a3684be1 795 opr->type_from_var = 1;
796 }
2b43685d 797 else if (opr->lmod != tmplmod) {
798 opr->size_mismatch = 1;
799 if (tmplmod < opr->lmod)
800 opr->size_lt = 1;
801 }
a652aa9f 802 opr->is_ptr = pp->type.is_ptr;
3ebea2cf 803 }
bd96f656 804 opr->is_array = pp->type.is_array;
06c5d854 805 }
c7ed83dd 806 opr->pp = pp;
06c5d854 807
850c9265 808 if (opr->lmod == OPLM_UNSPEC)
809 guess_lmod_from_name(opr);
850c9265 810 return wordc;
c36e914d 811}
812
33c35af6 813static const struct {
814 const char *name;
815 unsigned int flags;
816} pref_table[] = {
817 { "rep", OPF_REP },
7ba45c34 818 { "repe", OPF_REP|OPF_REPZ },
819 { "repz", OPF_REP|OPF_REPZ },
820 { "repne", OPF_REP|OPF_REPNZ },
821 { "repnz", OPF_REP|OPF_REPNZ },
037f4971 822 { "lock", OPF_LOCK }, // ignored for now..
33c35af6 823};
824
5c024ef7 825#define OPF_CJMP_CC (OPF_JMP|OPF_CJMP|OPF_CC)
826
c36e914d 827static const struct {
850c9265 828 const char *name;
829 enum op_op op;
092f64e1 830 unsigned short minopr;
831 unsigned short maxopr;
69a3cdfc 832 unsigned int flags;
092f64e1 833 unsigned char pfo;
834 unsigned char pfo_inv;
c36e914d 835} op_table[] = {
33c35af6 836 { "nop", OP_NOP, 0, 0, 0 },
69a3cdfc 837 { "push", OP_PUSH, 1, 1, 0 },
838 { "pop", OP_POP, 1, 1, OPF_DATA },
591721d7 839 { "leave",OP_LEAVE, 0, 0, OPF_DATA },
69a3cdfc 840 { "mov" , OP_MOV, 2, 2, OPF_DATA },
841 { "lea", OP_LEA, 2, 2, OPF_DATA },
842 { "movzx",OP_MOVZX, 2, 2, OPF_DATA },
843 { "movsx",OP_MOVSX, 2, 2, OPF_DATA },
108e9fe3 844 { "xchg", OP_XCHG, 2, 2, OPF_DATA },
69a3cdfc 845 { "not", OP_NOT, 1, 1, OPF_DATA },
5101a5f9 846 { "cdq", OP_CDQ, 0, 0, OPF_DATA },
092f64e1 847 { "lodsb",OP_LODS, 0, 0, OPF_DATA },
848 { "lodsw",OP_LODS, 0, 0, OPF_DATA },
849 { "lodsd",OP_LODS, 0, 0, OPF_DATA },
33c35af6 850 { "stosb",OP_STOS, 0, 0, OPF_DATA },
851 { "stosw",OP_STOS, 0, 0, OPF_DATA },
852 { "stosd",OP_STOS, 0, 0, OPF_DATA },
d4e3b5db 853 { "movsb",OP_MOVS, 0, 0, OPF_DATA },
854 { "movsw",OP_MOVS, 0, 0, OPF_DATA },
855 { "movsd",OP_MOVS, 0, 0, OPF_DATA },
7ba45c34 856 { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
857 { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
858 { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
591721d7 859 { "scasb",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
860 { "scasw",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
861 { "scasd",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
862 { "std", OP_STD, 0, 0, OPF_DATA }, // special flag
863 { "cld", OP_CLD, 0, 0, OPF_DATA },
69a3cdfc 864 { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
865 { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
866 { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
867 { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS },
868 { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS },
869 { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
870 { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
871 { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
872 { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
3b2f4044 873 { "shrd", OP_SHRD, 3, 3, OPF_DATA|OPF_FLAGS },
d4e3b5db 874 { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
875 { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
092f64e1 876 { "rcl", OP_RCL, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
877 { "rcr", OP_RCR, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
878 { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
879 { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
1f84f6b3 880 { "bsf", OP_BSF, 2, 2, OPF_DATA|OPF_FLAGS },
69a3cdfc 881 { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
882 { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS },
5101a5f9 883 { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS },
884 { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS },
69a3cdfc 885 { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS },
5101a5f9 886 { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS },
887 { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS },
69a3cdfc 888 { "test", OP_TEST, 2, 2, OPF_FLAGS },
889 { "cmp", OP_CMP, 2, 2, OPF_FLAGS },
a3684be1 890 { "retn", OP_RET, 0, 1, OPF_TAIL },
de50b98b 891 { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS },
69a3cdfc 892 { "jmp", OP_JMP, 1, 1, OPF_JMP },
5c024ef7 893 { "jecxz",OP_JECXZ, 1, 1, OPF_JMP|OPF_CJMP },
092f64e1 894 { "jo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 0 }, // 70 OF=1
895 { "jno", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 1 }, // 71 OF=0
896 { "jc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72 CF=1
897 { "jb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72
898 { "jnc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73 CF=0
899 { "jnb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
900 { "jae", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
901 { "jz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74 ZF=1
902 { "je", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74
903 { "jnz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75 ZF=0
904 { "jne", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75
905 { "jbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76 CF=1||ZF=1
906 { "jna", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76
907 { "ja", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77 CF=0&&ZF=0
908 { "jnbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77
909 { "js", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 0 }, // 78 SF=1
910 { "jns", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 1 }, // 79 SF=0
911 { "jp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a PF=1
912 { "jpe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a
913 { "jnp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b PF=0
914 { "jpo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b
915 { "jl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c SF!=OF
916 { "jnge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c
917 { "jge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d SF=OF
918 { "jnl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d
919 { "jle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e ZF=1||SF!=OF
920 { "jng", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e
921 { "jg", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f ZF=0&&SF=OF
922 { "jnle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f
923 { "seto", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 0 },
924 { "setno", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 1 },
925 { "setc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
926 { "setb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
927 { "setnc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
928 { "setae", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
929 { "setnb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
930 { "setz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
931 { "sete", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
932 { "setnz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
933 { "setne", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
934 { "setbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
935 { "setna", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
936 { "seta", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
937 { "setnbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
938 { "sets", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 0 },
939 { "setns", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 1 },
940 { "setp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
941 { "setpe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
942 { "setnp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
943 { "setpo", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
944 { "setl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
945 { "setnge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
946 { "setge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
947 { "setnl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
948 { "setle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
949 { "setng", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
950 { "setg", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
951 { "setnle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
90307a99 952 // x87
953 // mmx
954 { "emms", OP_EMMS, 0, 0, OPF_DATA },
955 { "movq", OP_MOV, 2, 2, OPF_DATA },
2c10ea1f 956 // must be last
957 { "ud2", OP_UD2 },
69a3cdfc 958};
c36e914d 959
960static void parse_op(struct parsed_op *op, char words[16][256], int wordc)
961{
bfa4a6ee 962 enum opr_lenmod lmod = OPLM_UNSPEC;
33c35af6 963 int prefix_flags = 0;
69a3cdfc 964 int regmask_ind;
965 int regmask;
33c35af6 966 int op_w = 0;
91977a1c 967 int opr = 0;
33c35af6 968 int w = 0;
91977a1c 969 int i;
c36e914d 970
33c35af6 971 for (i = 0; i < ARRAY_SIZE(pref_table); i++) {
972 if (IS(words[w], pref_table[i].name)) {
973 prefix_flags = pref_table[i].flags;
974 break;
975 }
976 }
977
978 if (prefix_flags) {
979 if (wordc <= 1)
980 aerr("lone prefix: '%s'\n", words[0]);
981 w++;
982 }
983
984 op_w = w;
91977a1c 985 for (i = 0; i < ARRAY_SIZE(op_table); i++) {
33c35af6 986 if (IS(words[w], op_table[i].name))
69a3cdfc 987 break;
988 }
c36e914d 989
2c10ea1f 990 if (i == ARRAY_SIZE(op_table)) {
30c8c549 991 if (!g_skip_func)
992 aerr("unhandled op: '%s'\n", words[0]);
2c10ea1f 993 i--; // OP_UD2
994 }
33c35af6 995 w++;
c36e914d 996
69a3cdfc 997 op->op = op_table[i].op;
33c35af6 998 op->flags = op_table[i].flags | prefix_flags;
092f64e1 999 op->pfo = op_table[i].pfo;
1000 op->pfo_inv = op_table[i].pfo_inv;
69a3cdfc 1001 op->regmask_src = op->regmask_dst = 0;
8eb12e72 1002 op->asmln = asmln;
69a3cdfc 1003
2c10ea1f 1004 if (op->op == OP_UD2)
1005 return;
1006
92d715b6 1007 for (opr = 0; opr < op_table[i].maxopr; opr++) {
1008 if (opr >= op_table[i].minopr && w >= wordc)
1009 break;
1010
69a3cdfc 1011 regmask = regmask_ind = 0;
1012 w = parse_operand(&op->operand[opr], &regmask, &regmask_ind,
1013 words, wordc, w, op->flags);
1014
1015 if (opr == 0 && (op->flags & OPF_DATA))
1016 op->regmask_dst = regmask;
92d715b6 1017 else
1018 op->regmask_src |= regmask;
1019 op->regmask_src |= regmask_ind;
91977a1c 1020 }
c36e914d 1021
91977a1c 1022 if (w < wordc)
1023 aerr("parse_op %s incomplete: %d/%d\n",
1024 words[0], w, wordc);
5101a5f9 1025
1026 // special cases
1027 op->operand_cnt = opr;
1028 if (!strncmp(op_table[i].name, "set", 3))
1029 op->operand[0].lmod = OPLM_BYTE;
1030
5101a5f9 1031 switch (op->op) {
92d715b6 1032 // first operand is not dst
1033 case OP_CMP:
1034 case OP_TEST:
1035 op->regmask_src |= op->regmask_dst;
1036 op->regmask_dst = 0;
1037 break;
1038
1039 // first operand is src too
1040 case OP_NOT:
1041 case OP_ADD:
1042 case OP_AND:
1043 case OP_OR:
1044 case OP_RCL:
1045 case OP_RCR:
1046 case OP_ADC:
1047 case OP_INC:
1048 case OP_DEC:
1049 case OP_NEG:
1050 // more below..
1051 op->regmask_src |= op->regmask_dst;
1052 break;
1053
1054 // special
1055 case OP_XCHG:
1056 op->regmask_src |= op->regmask_dst;
1057 op->regmask_dst |= op->regmask_src;
1058 goto check_align;
1059
1060 case OP_SUB:
1061 case OP_SBB:
1062 case OP_XOR:
1063 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1064 && op->operand[0].lmod == op->operand[1].lmod
1065 && op->operand[0].reg == op->operand[1].reg
1066 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
1067 {
1068 op->regmask_src = 0;
1069 }
1070 else
1071 op->regmask_src |= op->regmask_dst;
1072 break;
1073
1074 // ops with implicit argumets
5101a5f9 1075 case OP_CDQ:
1076 op->operand_cnt = 2;
1077 setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst);
1078 setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src);
1079 break;
1080
092f64e1 1081 case OP_LODS:
33c35af6 1082 case OP_STOS:
591721d7 1083 case OP_SCAS:
33c35af6 1084 if (op->operand_cnt != 0)
1085 break;
591721d7 1086 if (words[op_w][4] == 'b')
33c35af6 1087 lmod = OPLM_BYTE;
591721d7 1088 else if (words[op_w][4] == 'w')
33c35af6 1089 lmod = OPLM_WORD;
591721d7 1090 else if (words[op_w][4] == 'd')
33c35af6 1091 lmod = OPLM_DWORD;
1092 op->operand_cnt = 3;
092f64e1 1093 setup_reg_opr(&op->operand[0], op->op == OP_LODS ? xSI : xDI,
1094 lmod, &op->regmask_src);
d4e3b5db 1095 setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src);
1096 op->regmask_dst = op->regmask_src;
092f64e1 1097 setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD,
1098 op->op == OP_LODS ? &op->regmask_dst : &op->regmask_src);
33c35af6 1099 break;
1100
d4e3b5db 1101 case OP_MOVS:
7ba45c34 1102 case OP_CMPS:
d4e3b5db 1103 if (op->operand_cnt != 0)
1104 break;
7ba45c34 1105 if (words[op_w][4] == 'b')
d4e3b5db 1106 lmod = OPLM_BYTE;
7ba45c34 1107 else if (words[op_w][4] == 'w')
d4e3b5db 1108 lmod = OPLM_WORD;
7ba45c34 1109 else if (words[op_w][4] == 'd')
d4e3b5db 1110 lmod = OPLM_DWORD;
1111 op->operand_cnt = 3;
1112 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
1113 setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src);
1114 setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src);
1115 op->regmask_dst = op->regmask_src;
1116 break;
1117
5c024ef7 1118 case OP_JECXZ:
1119 op->operand_cnt = 1;
1120 op->regmask_src = 1 << xCX;
1121 op->operand[0].type = OPT_REG;
1122 op->operand[0].reg = xCX;
1123 op->operand[0].lmod = OPLM_DWORD;
1124 break;
1125
5101a5f9 1126 case OP_IMUL:
1127 if (op->operand_cnt != 1)
1128 break;
1129 // fallthrough
1130 case OP_MUL:
1131 // singleop mul
92d715b6 1132 op->regmask_src |= op->regmask_dst;
5101a5f9 1133 op->regmask_dst = (1 << xDX) | (1 << xAX);
5101a5f9 1134 if (op->operand[0].lmod == OPLM_UNSPEC)
1135 op->operand[0].lmod = OPLM_DWORD;
1136 break;
1137
1138 case OP_DIV:
1139 case OP_IDIV:
1140 // we could set up operands for edx:eax, but there is no real need to
1141 // (see is_opr_modified())
92d715b6 1142 op->regmask_src |= op->regmask_dst;
1143 op->regmask_dst = (1 << xDX) | (1 << xAX);
5101a5f9 1144 if (op->operand[0].lmod == OPLM_UNSPEC)
1145 op->operand[0].lmod = OPLM_DWORD;
1146 break;
1147
1148 case OP_SHL:
1149 case OP_SHR:
1150 case OP_SAR:
d4e3b5db 1151 case OP_ROL:
1152 case OP_ROR:
92d715b6 1153 op->regmask_src |= op->regmask_dst;
5101a5f9 1154 if (op->operand[1].lmod == OPLM_UNSPEC)
1155 op->operand[1].lmod = OPLM_BYTE;
1156 break;
1157
3b2f4044 1158 case OP_SHRD:
92d715b6 1159 op->regmask_src |= op->regmask_dst;
3b2f4044 1160 if (op->operand[2].lmod == OPLM_UNSPEC)
1161 op->operand[2].lmod = OPLM_BYTE;
1162 break;
1163
7ba45c34 1164 case OP_PUSH:
92d715b6 1165 op->regmask_src |= op->regmask_dst;
1166 op->regmask_dst = 0;
7ba45c34 1167 if (op->operand[0].lmod == OPLM_UNSPEC
1168 && (op->operand[0].type == OPT_CONST
1169 || op->operand[0].type == OPT_OFFSET
1170 || op->operand[0].type == OPT_LABEL))
1171 op->operand[0].lmod = OPLM_DWORD;
1172 break;
1173
3ebea2cf 1174 // alignment
1175 case OP_MOV:
092f64e1 1176 check_align:
3ebea2cf 1177 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
092f64e1 1178 && op->operand[0].lmod == op->operand[1].lmod
1179 && op->operand[0].reg == op->operand[1].reg
92d715b6 1180 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
3ebea2cf 1181 {
5e49b270 1182 op->flags |= OPF_RMD | OPF_DONE;
092f64e1 1183 op->regmask_src = op->regmask_dst = 0;
3ebea2cf 1184 }
1185 break;
1186
de50b98b 1187 case OP_LEA:
1188 if (op->operand[0].type == OPT_REG
1189 && op->operand[1].type == OPT_REGMEM)
1190 {
1191 char buf[16];
1192 snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name);
1193 if (IS(buf, op->operand[1].name))
5e49b270 1194 op->flags |= OPF_RMD | OPF_DONE;
de50b98b 1195 }
1196 break;
1197
89ff3147 1198 case OP_CALL:
1199 // trashed regs must be explicitly detected later
1200 op->regmask_dst = 0;
1201 break;
1202
92d715b6 1203 case OP_LEAVE:
1204 op->regmask_dst = (1 << xBP) | (1 << xSP);
1205 op->regmask_src = 1 << xBP;
1206 break;
1207
5101a5f9 1208 default:
1209 break;
1210 }
c36e914d 1211}
1212
092f64e1 1213static const char *op_name(struct parsed_op *po)
850c9265 1214{
092f64e1 1215 static char buf[16];
1216 char *p;
850c9265 1217 int i;
1218
092f64e1 1219 if (po->op == OP_JCC || po->op == OP_SCC) {
1220 p = buf;
1221 *p++ = (po->op == OP_JCC) ? 'j' : 's';
1222 if (po->pfo_inv)
1223 *p++ = 'n';
1224 strcpy(p, parsed_flag_op_names[po->pfo]);
1225 return buf;
1226 }
1227
850c9265 1228 for (i = 0; i < ARRAY_SIZE(op_table); i++)
092f64e1 1229 if (op_table[i].op == po->op)
850c9265 1230 return op_table[i].name;
1231
1232 return "???";
1233}
1234
1235// debug
1236static const char *dump_op(struct parsed_op *po)
1237{
1238 static char out[128];
1239 char *p = out;
1240 int i;
1241
4c45fa73 1242 if (po == NULL)
1243 return "???";
1244
092f64e1 1245 snprintf(out, sizeof(out), "%s", op_name(po));
850c9265 1246 for (i = 0; i < po->operand_cnt; i++) {
1247 p += strlen(p);
1248 if (i > 0)
1249 *p++ = ',';
1250 snprintf(p, sizeof(out) - (p - out),
1251 po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s",
1252 po->operand[i].name);
1253 }
1254
1255 return out;
1256}
1257
4c45fa73 1258static const char *lmod_type_u(struct parsed_op *po,
1259 enum opr_lenmod lmod)
1260{
1261 switch (lmod) {
90307a99 1262 case OPLM_QWORD:
1263 return "u64";
4c45fa73 1264 case OPLM_DWORD:
1265 return "u32";
1266 case OPLM_WORD:
1267 return "u16";
1268 case OPLM_BYTE:
1269 return "u8";
1270 default:
1271 ferr(po, "invalid lmod: %d\n", lmod);
1272 return "(_invalid_)";
1273 }
1274}
1275
d4e3b5db 1276static const char *lmod_cast_u(struct parsed_op *po,
1277 enum opr_lenmod lmod)
1278{
1279 switch (lmod) {
90307a99 1280 case OPLM_QWORD:
1281 return "";
d4e3b5db 1282 case OPLM_DWORD:
1283 return "";
1284 case OPLM_WORD:
1285 return "(u16)";
1286 case OPLM_BYTE:
1287 return "(u8)";
1288 default:
1289 ferr(po, "invalid lmod: %d\n", lmod);
1290 return "(_invalid_)";
1291 }
1292}
1293
1294static const char *lmod_cast_u_ptr(struct parsed_op *po,
1295 enum opr_lenmod lmod)
1296{
1297 switch (lmod) {
90307a99 1298 case OPLM_QWORD:
1299 return "*(u64 *)";
d4e3b5db 1300 case OPLM_DWORD:
1301 return "*(u32 *)";
1302 case OPLM_WORD:
1303 return "*(u16 *)";
1304 case OPLM_BYTE:
1305 return "*(u8 *)";
1306 default:
1307 ferr(po, "invalid lmod: %d\n", lmod);
1308 return "(_invalid_)";
1309 }
1310}
1311
1312static const char *lmod_cast_s(struct parsed_op *po,
1313 enum opr_lenmod lmod)
1314{
1315 switch (lmod) {
90307a99 1316 case OPLM_QWORD:
1317 return "(s64)";
d4e3b5db 1318 case OPLM_DWORD:
1319 return "(s32)";
1320 case OPLM_WORD:
1321 return "(s16)";
1322 case OPLM_BYTE:
1323 return "(s8)";
1324 default:
1325 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1326 return "(_invalid_)";
1327 }
1328}
1329
1330static const char *lmod_cast(struct parsed_op *po,
1331 enum opr_lenmod lmod, int is_signed)
1332{
1333 return is_signed ?
1334 lmod_cast_s(po, lmod) :
1335 lmod_cast_u(po, lmod);
1336}
1337
1338static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
1339{
1340 switch (lmod) {
90307a99 1341 case OPLM_QWORD:
1342 return 8;
d4e3b5db 1343 case OPLM_DWORD:
1344 return 4;
1345 case OPLM_WORD:
1346 return 2;
1347 case OPLM_BYTE:
1348 return 1;
1349 default:
1350 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1351 return 0;
1352 }
1353}
1354
91977a1c 1355static const char *opr_name(struct parsed_op *po, int opr_num)
c36e914d 1356{
91977a1c 1357 if (opr_num >= po->operand_cnt)
1358 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1359 return po->operand[opr_num].name;
c36e914d 1360}
1361
91977a1c 1362static unsigned int opr_const(struct parsed_op *po, int opr_num)
c36e914d 1363{
91977a1c 1364 if (opr_num >= po->operand_cnt)
1365 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1366 if (po->operand[opr_num].type != OPT_CONST)
1367 ferr(po, "opr %d: const expected\n", opr_num);
1368 return po->operand[opr_num].val;
1369}
c36e914d 1370
91977a1c 1371static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr)
1372{
90307a99 1373 if ((unsigned int)popr->reg >= ARRAY_SIZE(regs_r32))
91977a1c 1374 ferr(po, "invalid reg: %d\n", popr->reg);
1375 return regs_r32[popr->reg];
1376}
c36e914d 1377
a2c1d768 1378// cast1 is the "final" cast
1379static const char *simplify_cast(const char *cast1, const char *cast2)
1380{
1381 static char buf[256];
1382
1383 if (cast1[0] == 0)
1384 return cast2;
1385 if (cast2[0] == 0)
1386 return cast1;
1387 if (IS(cast1, cast2))
1388 return cast1;
1389 if (IS(cast1, "(s8)") && IS(cast2, "(u8)"))
1390 return cast1;
1391 if (IS(cast1, "(s16)") && IS(cast2, "(u16)"))
1392 return cast1;
1393 if (IS(cast1, "(u8)") && IS_START(cast2, "*(u8 *)"))
1394 return cast2;
1395 if (IS(cast1, "(u16)") && IS_START(cast2, "*(u16 *)"))
1396 return cast2;
1cd4a663 1397 if (strchr(cast1, '*') && IS_START(cast2, "(u32)"))
1398 return cast1;
a2c1d768 1399
1400 snprintf(buf, sizeof(buf), "%s%s", cast1, cast2);
1401 return buf;
1402}
1403
1404static const char *simplify_cast_num(const char *cast, unsigned int val)
1405{
1406 if (IS(cast, "(u8)") && val < 0x100)
1407 return "";
1408 if (IS(cast, "(s8)") && val < 0x80)
1409 return "";
1410 if (IS(cast, "(u16)") && val < 0x10000)
1411 return "";
1412 if (IS(cast, "(s16)") && val < 0x8000)
1413 return "";
1414 if (IS(cast, "(s32)") && val < 0x80000000)
1415 return "";
1416
1417 return cast;
1418}
1419
39b168b8 1420static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
1421 int *extra_offs)
91977a1c 1422{
39b168b8 1423 const char *p;
1424 char *endp;
1425 int namelen;
850c9265 1426 int i;
1427
39b168b8 1428 *extra_offs = 0;
1429 namelen = strlen(name);
1430
1431 p = strchr(name, '+');
1432 if (p != NULL) {
1433 namelen = p - name;
1434 if (namelen <= 0)
1435 ferr(po, "equ parse failed for '%s'\n", name);
1436
1437 if (IS_START(p, "0x"))
1438 p += 2;
1439 *extra_offs = strtol(p, &endp, 16);
1440 if (*endp != 0)
1441 ferr(po, "equ parse failed for '%s'\n", name);
1442 }
1443
850c9265 1444 for (i = 0; i < g_eqcnt; i++)
39b168b8 1445 if (strncmp(g_eqs[i].name, name, namelen) == 0
1446 && g_eqs[i].name[namelen] == 0)
850c9265 1447 break;
87bf6cec 1448 if (i >= g_eqcnt) {
1449 if (po != NULL)
1450 ferr(po, "unresolved equ name: '%s'\n", name);
1451 return NULL;
1452 }
850c9265 1453
1454 return &g_eqs[i];
1455}
1456
1cd4a663 1457static int is_stack_access(struct parsed_op *po,
1458 const struct parsed_opr *popr)
850c9265 1459{
09bc6fd5 1460 return (parse_stack_el(popr->name, NULL, 0)
1cd4a663 1461 || (g_bp_frame && !(po->flags & OPF_EBP_S)
1462 && IS_START(popr->name, "ebp")));
1463}
1464
1465static void parse_stack_access(struct parsed_op *po,
1466 const char *name, char *ofs_reg, int *offset_out,
037f4971 1467 int *stack_ra_out, const char **bp_arg_out, int is_lea)
1cd4a663 1468{
1469 const char *bp_arg = "";
1470 const char *p = NULL;
91977a1c 1471 struct parsed_equ *eq;
4f12f671 1472 char *endp = NULL;
d4e3b5db 1473 int stack_ra = 0;
39b168b8 1474 int offset = 0;
91977a1c 1475
1cd4a663 1476 ofs_reg[0] = 0;
a3684be1 1477
1478 if (IS_START(name, "ebp-")
1479 || (IS_START(name, "ebp+") && '0' <= name[4] && name[4] <= '9'))
1480 {
1481 p = name + 4;
1482 if (IS_START(p, "0x"))
1483 p += 2;
1484 offset = strtoul(p, &endp, 16);
1485 if (name[3] == '-')
1486 offset = -offset;
1487 if (*endp != 0)
1488 ferr(po, "ebp- parse of '%s' failed\n", name);
1489 }
1490 else {
09bc6fd5 1491 bp_arg = parse_stack_el(name, ofs_reg, 0);
4f12f671 1492 snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
1493 eq = equ_find(po, bp_arg, &offset);
1494 if (eq == NULL)
1495 ferr(po, "detected but missing eq\n");
1496 offset += eq->offset;
1497 }
39b168b8 1498
d4e3b5db 1499 if (!strncmp(name, "ebp", 3))
1500 stack_ra = 4;
1501
037f4971 1502 // yes it sometimes LEAs ra for compares..
1503 if (!is_lea && ofs_reg[0] == 0
1504 && stack_ra <= offset && offset < stack_ra + 4)
1505 {
39b168b8 1506 ferr(po, "reference to ra? %d %d\n", offset, stack_ra);
037f4971 1507 }
d4e3b5db 1508
1cd4a663 1509 *offset_out = offset;
1510 *stack_ra_out = stack_ra;
1511 if (bp_arg_out)
1512 *bp_arg_out = bp_arg;
1513}
1514
5f70a34f 1515static int stack_frame_access(struct parsed_op *po,
1cd4a663 1516 struct parsed_opr *popr, char *buf, size_t buf_size,
1517 const char *name, const char *cast, int is_src, int is_lea)
1518{
1519 enum opr_lenmod tmp_lmod = OPLM_UNSPEC;
1520 const char *prefix = "";
1521 const char *bp_arg = NULL;
1522 char ofs_reg[16] = { 0, };
1523 int i, arg_i, arg_s;
1524 int unaligned = 0;
1525 int stack_ra = 0;
1526 int offset = 0;
5f70a34f 1527 int retval = -1;
1cd4a663 1528 int sf_ofs;
1529 int lim;
1530
1531 if (po->flags & OPF_EBP_S)
1532 ferr(po, "stack_frame_access while ebp is scratch\n");
1533
037f4971 1534 parse_stack_access(po, name, ofs_reg, &offset,
1535 &stack_ra, &bp_arg, is_lea);
1cd4a663 1536
4f12f671 1537 if (offset > stack_ra)
1538 {
39b168b8 1539 arg_i = (offset - stack_ra - 4) / 4;
bd96f656 1540 if (arg_i < 0 || arg_i >= g_func_pp->argc_stack)
4f12f671 1541 {
bd96f656 1542 if (g_func_pp->is_vararg
1543 && arg_i == g_func_pp->argc_stack && is_lea)
1544 {
4f12f671 1545 // should be va_list
1546 if (cast[0] == 0)
1547 cast = "(u32)";
1548 snprintf(buf, buf_size, "%sap", cast);
5f70a34f 1549 return -1;
4f12f671 1550 }
1551 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1552 offset, bp_arg, arg_i);
1553 }
4c45fa73 1554 if (ofs_reg[0] != 0)
7ba45c34 1555 ferr(po, "offset reg on arg access?\n");
91977a1c 1556
bd96f656 1557 for (i = arg_s = 0; i < g_func_pp->argc; i++) {
1558 if (g_func_pp->arg[i].reg != NULL)
91977a1c 1559 continue;
1560 if (arg_s == arg_i)
1561 break;
1562 arg_s++;
1563 }
bd96f656 1564 if (i == g_func_pp->argc)
91977a1c 1565 ferr(po, "arg %d not in prototype?\n", arg_i);
850c9265 1566
bd96f656 1567 popr->is_ptr = g_func_pp->arg[i].type.is_ptr;
5f70a34f 1568 retval = i;
de50b98b 1569
1570 switch (popr->lmod)
7ba45c34 1571 {
1572 case OPLM_BYTE:
4f12f671 1573 if (is_lea)
1574 ferr(po, "lea/byte to arg?\n");
7ba45c34 1575 if (is_src && (offset & 3) == 0)
a2c1d768 1576 snprintf(buf, buf_size, "%sa%d",
1577 simplify_cast(cast, "(u8)"), i + 1);
7ba45c34 1578 else
a2c1d768 1579 snprintf(buf, buf_size, "%sBYTE%d(a%d)",
1580 cast, offset & 3, i + 1);
7ba45c34 1581 break;
1582
1583 case OPLM_WORD:
4f12f671 1584 if (is_lea)
1585 ferr(po, "lea/word to arg?\n");
a3684be1 1586 if (offset & 1) {
1587 unaligned = 1;
1588 if (!is_src) {
1589 if (offset & 2)
1590 ferr(po, "problematic arg store\n");
a2c1d768 1591 snprintf(buf, buf_size, "%s((char *)&a%d + 1)",
1592 simplify_cast(cast, "*(u16 *)"), i + 1);
a3684be1 1593 }
1594 else
1595 ferr(po, "unaligned arg word load\n");
1596 }
1597 else if (is_src && (offset & 2) == 0)
a2c1d768 1598 snprintf(buf, buf_size, "%sa%d",
1599 simplify_cast(cast, "(u16)"), i + 1);
7ba45c34 1600 else
a2c1d768 1601 snprintf(buf, buf_size, "%s%sWORD(a%d)",
1602 cast, (offset & 2) ? "HI" : "LO", i + 1);
7ba45c34 1603 break;
1604
1605 case OPLM_DWORD:
3ebea2cf 1606 if (cast[0])
1607 prefix = cast;
1608 else if (is_src)
1609 prefix = "(u32)";
a3684be1 1610
de50b98b 1611 if (offset & 3) {
a3684be1 1612 unaligned = 1;
2b43685d 1613 if (is_lea)
1614 snprintf(buf, buf_size, "(u32)&a%d + %d",
1615 i + 1, offset & 3);
a3684be1 1616 else if (!is_src)
1617 ferr(po, "unaligned arg store\n");
1618 else {
1619 // mov edx, [ebp+arg_4+2]; movsx ecx, dx
2b43685d 1620 snprintf(buf, buf_size, "%s(a%d >> %d)",
1621 prefix, i + 1, (offset & 3) * 8);
a3684be1 1622 }
de50b98b 1623 }
1624 else {
1625 snprintf(buf, buf_size, "%s%sa%d",
1626 prefix, is_lea ? "&" : "", i + 1);
1627 }
7ba45c34 1628 break;
1629
1630 default:
de50b98b 1631 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
7ba45c34 1632 }
1633
a3684be1 1634 if (unaligned)
1635 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1636
7ba45c34 1637 // common problem
bd96f656 1638 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type);
a2c1d768 1639 if (tmp_lmod != OPLM_DWORD
de041e5b 1640 && (unaligned || (!is_src && lmod_bytes(po, tmp_lmod)
1641 < lmod_bytes(po, popr->lmod) + (offset & 3))))
a2c1d768 1642 {
1643 ferr(po, "bp_arg arg%d/w offset %d and type '%s' is too small\n",
1644 i + 1, offset, g_func_pp->arg[i].type.name);
1645 }
4741fdfe 1646 // can't check this because msvc likes to reuse
1647 // arg space for scratch..
1648 //if (popr->is_ptr && popr->lmod != OPLM_DWORD)
1649 // ferr(po, "bp_arg arg%d: non-dword ptr access\n", i + 1);
91977a1c 1650 }
4f12f671 1651 else
1652 {
1bafb621 1653 if (g_stack_fsz == 0)
1654 ferr(po, "stack var access without stackframe\n");
a2c1d768 1655 g_stack_frame_used = 1;
850c9265 1656
39b168b8 1657 sf_ofs = g_stack_fsz + offset;
de50b98b 1658 lim = (ofs_reg[0] != 0) ? -4 : 0;
1659 if (offset > 0 || sf_ofs < lim)
39b168b8 1660 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
850c9265 1661
1662 if (is_lea)
33c35af6 1663 prefix = "(u32)&";
3ebea2cf 1664 else
1665 prefix = cast;
850c9265 1666
de50b98b 1667 switch (popr->lmod)
850c9265 1668 {
1669 case OPLM_BYTE:
7ba45c34 1670 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1671 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
850c9265 1672 break;
7ba45c34 1673
850c9265 1674 case OPLM_WORD:
7ba45c34 1675 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1676 // known unaligned or possibly unaligned
1677 strcat(g_comment, " unaligned");
1678 if (prefix[0] == 0)
1679 prefix = "*(u16 *)&";
1680 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1681 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1682 break;
1683 }
1684 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
850c9265 1685 break;
7ba45c34 1686
850c9265 1687 case OPLM_DWORD:
7ba45c34 1688 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1689 // known unaligned or possibly unaligned
1690 strcat(g_comment, " unaligned");
1691 if (prefix[0] == 0)
1692 prefix = "*(u32 *)&";
1693 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1694 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1695 break;
1696 }
1697 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
850c9265 1698 break;
7ba45c34 1699
850c9265 1700 default:
de50b98b 1701 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
850c9265 1702 }
91977a1c 1703 }
5f70a34f 1704
1705 return retval;
91977a1c 1706}
c36e914d 1707
89ff3147 1708static void check_func_pp(struct parsed_op *po,
1709 const struct parsed_proto *pp, const char *pfx)
1710{
179b79a9 1711 enum opr_lenmod tmp_lmod;
b74c31e3 1712 char buf[256];
179b79a9 1713 int ret, i;
b74c31e3 1714
89ff3147 1715 if (pp->argc_reg != 0) {
b74c31e3 1716 if (/*!g_allow_regfunc &&*/ !pp->is_fastcall) {
1717 pp_print(buf, sizeof(buf), pp);
1718 ferr(po, "%s: unexpected reg arg in icall: %s\n", pfx, buf);
1719 }
89ff3147 1720 if (pp->argc_stack > 0 && pp->argc_reg != 2)
1721 ferr(po, "%s: %d reg arg(s) with %d stack arg(s)\n",
1722 pfx, pp->argc_reg, pp->argc_stack);
1723 }
179b79a9 1724
1725 // fptrs must use 32bit args, callsite might have no information and
1726 // lack a cast to smaller types, which results in incorrectly masked
1727 // args passed (callee may assume masked args, it does on ARM)
61e29183 1728 if (!pp->is_osinc) {
179b79a9 1729 for (i = 0; i < pp->argc; i++) {
1730 ret = guess_lmod_from_c_type(&tmp_lmod, &pp->arg[i].type);
1731 if (ret && tmp_lmod != OPLM_DWORD)
1732 ferr(po, "reference to %s with arg%d '%s'\n", pp->name,
1733 i + 1, pp->arg[i].type.name);
1734 }
1735 }
89ff3147 1736}
1737
7aca4698 1738static const char *check_label_read_ref(struct parsed_op *po,
1739 const char *name)
3ebea2cf 1740{
840257f6 1741 const struct parsed_proto *pp;
1742
36595fd2 1743 pp = proto_parse(g_fhdr, name, 0);
840257f6 1744 if (pp == NULL)
1745 ferr(po, "proto_parse failed for ref '%s'\n", name);
1746
89ff3147 1747 if (pp->is_func)
1748 check_func_pp(po, pp, "ref");
7aca4698 1749
1750 return pp->name;
3ebea2cf 1751}
1752
91977a1c 1753static char *out_src_opr(char *buf, size_t buf_size,
591721d7 1754 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
3ebea2cf 1755 int is_lea)
91977a1c 1756{
850c9265 1757 char tmp1[256], tmp2[256];
1758 char expr[256];
7aca4698 1759 const char *name;
a2c1d768 1760 char *p;
850c9265 1761 int ret;
1762
3ebea2cf 1763 if (cast == NULL)
1764 cast = "";
1765
91977a1c 1766 switch (popr->type) {
1767 case OPT_REG:
850c9265 1768 if (is_lea)
1769 ferr(po, "lea from reg?\n");
1770
91977a1c 1771 switch (popr->lmod) {
90307a99 1772 case OPLM_QWORD:
1773 snprintf(buf, buf_size, "%s%s.q", cast, opr_reg_p(po, popr));
1774 break;
91977a1c 1775 case OPLM_DWORD:
3ebea2cf 1776 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
91977a1c 1777 break;
850c9265 1778 case OPLM_WORD:
a2c1d768 1779 snprintf(buf, buf_size, "%s%s",
1780 simplify_cast(cast, "(u16)"), opr_reg_p(po, popr));
850c9265 1781 break;
1782 case OPLM_BYTE:
5101a5f9 1783 if (popr->name[1] == 'h') // XXX..
a2c1d768 1784 snprintf(buf, buf_size, "%s(%s >> 8)",
1785 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
5101a5f9 1786 else
a2c1d768 1787 snprintf(buf, buf_size, "%s%s",
1788 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
850c9265 1789 break;
91977a1c 1790 default:
1791 ferr(po, "invalid src lmod: %d\n", popr->lmod);
1792 }
1793 break;
850c9265 1794
91977a1c 1795 case OPT_REGMEM:
1cd4a663 1796 if (is_stack_access(po, popr)) {
de50b98b 1797 stack_frame_access(po, popr, buf, buf_size,
3ebea2cf 1798 popr->name, cast, 1, is_lea);
91977a1c 1799 break;
1800 }
850c9265 1801
1802 strcpy(expr, popr->name);
1803 if (strchr(expr, '[')) {
1804 // special case: '[' can only be left for label[reg] form
1805 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
1806 if (ret != 2)
1807 ferr(po, "parse failure for '%s'\n", expr);
a2c1d768 1808 if (tmp1[0] == '(') {
1809 // (off_4FFF50+3)[eax]
1810 p = strchr(tmp1 + 1, ')');
1811 if (p == NULL || p[1] != 0)
1812 ferr(po, "parse failure (2) for '%s'\n", expr);
1813 *p = 0;
1814 memmove(tmp1, tmp1 + 1, strlen(tmp1));
1815 }
33c35af6 1816 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
850c9265 1817 }
1818
1819 // XXX: do we need more parsing?
1820 if (is_lea) {
1821 snprintf(buf, buf_size, "%s", expr);
1822 break;
1823 }
1824
a2c1d768 1825 snprintf(buf, buf_size, "%s(%s)",
1826 simplify_cast(cast, lmod_cast_u_ptr(po, popr->lmod)), expr);
91977a1c 1827 break;
850c9265 1828
91977a1c 1829 case OPT_LABEL:
7aca4698 1830 name = check_label_read_ref(po, popr->name);
3ebea2cf 1831 if (cast[0] == 0 && popr->is_ptr)
1832 cast = "(u32)";
2b43685d 1833
850c9265 1834 if (is_lea)
7aca4698 1835 snprintf(buf, buf_size, "(u32)&%s", name);
2b43685d 1836 else if (popr->size_lt)
1837 snprintf(buf, buf_size, "%s%s%s%s", cast,
1838 lmod_cast_u_ptr(po, popr->lmod),
7aca4698 1839 popr->is_array ? "" : "&", name);
850c9265 1840 else
7aca4698 1841 snprintf(buf, buf_size, "%s%s%s", cast, name,
7ba45c34 1842 popr->is_array ? "[0]" : "");
850c9265 1843 break;
1844
1845 case OPT_OFFSET:
7aca4698 1846 name = check_label_read_ref(po, popr->name);
3ebea2cf 1847 if (cast[0] == 0)
1848 cast = "(u32)";
850c9265 1849 if (is_lea)
1850 ferr(po, "lea an offset?\n");
7aca4698 1851 snprintf(buf, buf_size, "%s&%s", cast, name);
91977a1c 1852 break;
850c9265 1853
91977a1c 1854 case OPT_CONST:
850c9265 1855 if (is_lea)
1856 ferr(po, "lea from const?\n");
1857
a2c1d768 1858 printf_number(tmp1, sizeof(tmp1), popr->val);
ddaf8bd7 1859 if (popr->val == 0 && strchr(cast, '*'))
1860 snprintf(buf, buf_size, "NULL");
1861 else
1862 snprintf(buf, buf_size, "%s%s",
1863 simplify_cast_num(cast, popr->val), tmp1);
91977a1c 1864 break;
850c9265 1865
91977a1c 1866 default:
1867 ferr(po, "invalid src type: %d\n", popr->type);
1868 }
1869
1870 return buf;
1871}
c36e914d 1872
de50b98b 1873// note: may set is_ptr (we find that out late for ebp frame..)
91977a1c 1874static char *out_dst_opr(char *buf, size_t buf_size,
1875 struct parsed_op *po, struct parsed_opr *popr)
1876{
1877 switch (popr->type) {
1878 case OPT_REG:
1879 switch (popr->lmod) {
90307a99 1880 case OPLM_QWORD:
1881 snprintf(buf, buf_size, "%s.q", opr_reg_p(po, popr));
1882 break;
91977a1c 1883 case OPLM_DWORD:
1884 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
1885 break;
850c9265 1886 case OPLM_WORD:
1887 // ugh..
1888 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
1889 break;
1890 case OPLM_BYTE:
1891 // ugh..
5101a5f9 1892 if (popr->name[1] == 'h') // XXX..
1893 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
1894 else
1895 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
850c9265 1896 break;
91977a1c 1897 default:
1898 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
1899 }
1900 break;
850c9265 1901
1902 case OPT_REGMEM:
1cd4a663 1903 if (is_stack_access(po, popr)) {
de50b98b 1904 stack_frame_access(po, popr, buf, buf_size,
3ebea2cf 1905 popr->name, "", 0, 0);
850c9265 1906 break;
1907 }
1908
3ebea2cf 1909 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
850c9265 1910
bfa4a6ee 1911 case OPT_LABEL:
2b43685d 1912 if (popr->size_mismatch)
1913 snprintf(buf, buf_size, "%s%s%s",
1914 lmod_cast_u_ptr(po, popr->lmod),
1915 popr->is_array ? "" : "&", popr->name);
1916 else
1917 snprintf(buf, buf_size, "%s%s", popr->name,
1918 popr->is_array ? "[0]" : "");
bfa4a6ee 1919 break;
1920
91977a1c 1921 default:
1922 ferr(po, "invalid dst type: %d\n", popr->type);
1923 }
1924
1925 return buf;
1926}
c36e914d 1927
3ebea2cf 1928static char *out_src_opr_u32(char *buf, size_t buf_size,
1929 struct parsed_op *po, struct parsed_opr *popr)
1930{
1931 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1932}
1933
91977a1c 1934static void out_test_for_cc(char *buf, size_t buf_size,
940e8e66 1935 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
69a3cdfc 1936 enum opr_lenmod lmod, const char *expr)
91977a1c 1937{
69a3cdfc 1938 const char *cast, *scast;
91977a1c 1939
69a3cdfc 1940 cast = lmod_cast_u(po, lmod);
1941 scast = lmod_cast_s(po, lmod);
1942
1943 switch (pfo) {
1944 case PFO_Z:
87bf6cec 1945 case PFO_BE: // CF=1||ZF=1; CF=0
850c9265 1946 snprintf(buf, buf_size, "(%s%s %s 0)",
940e8e66 1947 cast, expr, is_inv ? "!=" : "==");
91977a1c 1948 break;
850c9265 1949
5101a5f9 1950 case PFO_S:
1951 case PFO_L: // SF!=OF; OF=0
1952 snprintf(buf, buf_size, "(%s%s %s 0)",
940e8e66 1953 scast, expr, is_inv ? ">=" : "<");
5101a5f9 1954 break;
1955
87bf6cec 1956 case PFO_LE: // ZF=1||SF!=OF; OF=0
69a3cdfc 1957 snprintf(buf, buf_size, "(%s%s %s 0)",
940e8e66 1958 scast, expr, is_inv ? ">" : "<=");
850c9265 1959 break;
1960
91977a1c 1961 default:
69a3cdfc 1962 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
91977a1c 1963 }
1964}
c36e914d 1965
850c9265 1966static void out_cmp_for_cc(char *buf, size_t buf_size,
a2c1d768 1967 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
850c9265 1968{
a2c1d768 1969 const char *cast, *scast, *cast_use;
1970 char buf1[256], buf2[256];
1971 enum opr_lenmod lmod;
1972
90307a99 1973 if (po->op != OP_DEC && po->operand[0].lmod != po->operand[1].lmod)
a2c1d768 1974 ferr(po, "%s: lmod mismatch: %d %d\n", __func__,
1975 po->operand[0].lmod, po->operand[1].lmod);
1976 lmod = po->operand[0].lmod;
850c9265 1977
69a3cdfc 1978 cast = lmod_cast_u(po, lmod);
1979 scast = lmod_cast_s(po, lmod);
850c9265 1980
a2c1d768 1981 switch (pfo) {
1982 case PFO_C:
1983 case PFO_Z:
1984 case PFO_BE: // !a
1985 cast_use = cast;
1986 break;
1987
1988 case PFO_S:
1989 case PFO_L: // !ge
1990 case PFO_LE:
1991 cast_use = scast;
1992 break;
1993
1994 default:
1995 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1996 }
1997
1998 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], cast_use, 0);
90307a99 1999 if (po->op == OP_DEC)
2000 snprintf(buf2, sizeof(buf2), "1");
2001 else
2002 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], cast_use, 0);
a2c1d768 2003
69a3cdfc 2004 switch (pfo) {
5101a5f9 2005 case PFO_C:
2006 // note: must be unsigned compare
a2c1d768 2007 snprintf(buf, buf_size, "(%s %s %s)",
2008 buf1, is_inv ? ">=" : "<", buf2);
5101a5f9 2009 break;
2010
69a3cdfc 2011 case PFO_Z:
a2c1d768 2012 snprintf(buf, buf_size, "(%s %s %s)",
2013 buf1, is_inv ? "!=" : "==", buf2);
850c9265 2014 break;
2015
5101a5f9 2016 case PFO_BE: // !a
850c9265 2017 // note: must be unsigned compare
a2c1d768 2018 snprintf(buf, buf_size, "(%s %s %s)",
2019 buf1, is_inv ? ">" : "<=", buf2);
2020
2021 // annoying case
2022 if (is_inv && lmod == OPLM_BYTE
2023 && po->operand[1].type == OPT_CONST
2024 && po->operand[1].val == 0xff)
2025 {
2026 snprintf(g_comment, sizeof(g_comment), "if %s", buf);
2027 snprintf(buf, buf_size, "(0)");
2028 }
5101a5f9 2029 break;
2030
2031 // note: must be signed compare
2032 case PFO_S:
2033 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
a2c1d768 2034 scast, buf1, buf2, is_inv ? ">=" : "<");
850c9265 2035 break;
2036
5101a5f9 2037 case PFO_L: // !ge
a2c1d768 2038 snprintf(buf, buf_size, "(%s %s %s)",
2039 buf1, is_inv ? ">=" : "<", buf2);
850c9265 2040 break;
2041
90307a99 2042 case PFO_LE: // !g
a2c1d768 2043 snprintf(buf, buf_size, "(%s %s %s)",
2044 buf1, is_inv ? ">" : "<=", buf2);
5101a5f9 2045 break;
2046
850c9265 2047 default:
a2c1d768 2048 break;
850c9265 2049 }
2050}
2051
69a3cdfc 2052static void out_cmp_test(char *buf, size_t buf_size,
940e8e66 2053 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
69a3cdfc 2054{
2055 char buf1[256], buf2[256], buf3[256];
2056
2057 if (po->op == OP_TEST) {
2058 if (IS(opr_name(po, 0), opr_name(po, 1))) {
3ebea2cf 2059 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
69a3cdfc 2060 }
2061 else {
3ebea2cf 2062 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
2063 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
69a3cdfc 2064 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
2065 }
940e8e66 2066 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
69a3cdfc 2067 po->operand[0].lmod, buf3);
2068 }
2069 else if (po->op == OP_CMP) {
a2c1d768 2070 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv);
69a3cdfc 2071 }
2072 else
2073 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
2074}
2075
850c9265 2076static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
91977a1c 2077 struct parsed_opr *popr2)
2078{
2079 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
2080 ferr(po, "missing lmod for both operands\n");
2081
2082 if (popr1->lmod == OPLM_UNSPEC)
2083 popr1->lmod = popr2->lmod;
2084 else if (popr2->lmod == OPLM_UNSPEC)
2085 popr2->lmod = popr1->lmod;
a3684be1 2086 else if (popr1->lmod != popr2->lmod) {
2087 if (popr1->type_from_var) {
2088 popr1->size_mismatch = 1;
2089 if (popr1->lmod < popr2->lmod)
2090 popr1->size_lt = 1;
2091 popr1->lmod = popr2->lmod;
2092 }
2093 else if (popr2->type_from_var) {
2094 popr2->size_mismatch = 1;
2095 if (popr2->lmod < popr1->lmod)
2096 popr2->size_lt = 1;
2097 popr2->lmod = popr1->lmod;
2098 }
2099 else
2100 ferr(po, "conflicting lmods: %d vs %d\n",
2101 popr1->lmod, popr2->lmod);
2102 }
91977a1c 2103}
c36e914d 2104
850c9265 2105static const char *op_to_c(struct parsed_op *po)
2106{
2107 switch (po->op)
2108 {
2109 case OP_ADD:
5101a5f9 2110 case OP_ADC:
850c9265 2111 return "+";
2112 case OP_SUB:
5101a5f9 2113 case OP_SBB:
850c9265 2114 return "-";
2115 case OP_AND:
2116 return "&";
2117 case OP_OR:
2118 return "|";
2119 case OP_XOR:
2120 return "^";
2121 case OP_SHL:
2122 return "<<";
2123 case OP_SHR:
2124 return ">>";
2125 case OP_MUL:
2126 case OP_IMUL:
2127 return "*";
2128 default:
2129 ferr(po, "op_to_c was supplied with %d\n", po->op);
2130 }
2131}
2132
7ae48d73 2133static void op_set_clear_flag(struct parsed_op *po,
2134 enum op_flags flag_set, enum op_flags flag_clear)
d4e3b5db 2135{
7ae48d73 2136 po->flags |= flag_set;
2137 po->flags &= ~flag_clear;
d4e3b5db 2138}
2139
de50b98b 2140// last op in stream - unconditional branch or ret
2141#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
092f64e1 2142 || ((ops[_i].flags & (OPF_JMP|OPF_CJMP|OPF_RMD)) == OPF_JMP \
037f4971 2143 && ops[_i].op != OP_CALL))
de50b98b 2144
db63af51 2145#define check_i(po, i) \
2146 if ((i) < 0) \
2147 ferr(po, "bad " #i ": %d\n", i)
2148
87bf6cec 2149static int scan_for_pop(int i, int opcnt, const char *reg,
d4e3b5db 2150 int magic, int depth, int *maxdepth, int do_flags)
850c9265 2151{
89ff3147 2152 const struct parsed_proto *pp;
87bf6cec 2153 struct parsed_op *po;
2154 int ret = 0;
4c45fa73 2155 int j;
87bf6cec 2156
850c9265 2157 for (; i < opcnt; i++) {
87bf6cec 2158 po = &ops[i];
2159 if (po->cc_scratch == magic)
2160 break; // already checked
2161 po->cc_scratch = magic;
2162
89ff3147 2163 if (po->flags & OPF_TAIL) {
2164 if (po->op == OP_CALL) {
92d715b6 2165 pp = proto_parse(g_fhdr, po->operand[0].name, g_quiet_pp);
89ff3147 2166 if (pp != NULL && pp->is_noreturn)
2167 // no stack cleanup for noreturn
2168 return ret;
2169 }
87bf6cec 2170 return -1; // deadend
89ff3147 2171 }
87bf6cec 2172
5e49b270 2173 if ((po->flags & (OPF_RMD|OPF_DONE))
5f70a34f 2174 || (po->op == OP_PUSH && po->p_argnum != 0)) // arg push
850c9265 2175 continue;
2176
87bf6cec 2177 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
4c45fa73 2178 if (po->btj != NULL) {
2179 // jumptable
da87ae38 2180 for (j = 0; j < po->btj->count; j++) {
db63af51 2181 check_i(po, po->btj->d[j].bt_i);
4c45fa73 2182 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic,
2183 depth, maxdepth, do_flags);
2184 if (ret < 0)
2185 return ret; // dead end
2186 }
da87ae38 2187 return ret;
4c45fa73 2188 }
2189
db63af51 2190 check_i(po, po->bt_i);
5c024ef7 2191 if (po->flags & OPF_CJMP) {
d4e3b5db 2192 ret |= scan_for_pop(po->bt_i, opcnt, reg, magic,
2193 depth, maxdepth, do_flags);
87bf6cec 2194 if (ret < 0)
2195 return ret; // dead end
2196 }
2197 else {
2198 i = po->bt_i - 1;
2199 }
2200 continue;
2201 }
2202
d4e3b5db 2203 if ((po->op == OP_POP || po->op == OP_PUSH)
2204 && po->operand[0].type == OPT_REG
87bf6cec 2205 && IS(po->operand[0].name, reg))
2206 {
9af2d373 2207 if (po->op == OP_PUSH && !(po->flags & OPF_FARGNR)) {
d4e3b5db 2208 depth++;
2209 if (depth > *maxdepth)
2210 *maxdepth = depth;
2211 if (do_flags)
7ae48d73 2212 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
d4e3b5db 2213 }
da87ae38 2214 else if (po->op == OP_POP) {
2215 if (depth == 0) {
2216 if (do_flags)
2217 op_set_clear_flag(po, OPF_RMD, OPF_RSAVE);
2218 return 1;
2219 }
2220 else {
2221 depth--;
2222 if (depth < 0) // should not happen
2223 ferr(po, "fail with depth\n");
2224 if (do_flags)
2225 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2226 }
d4e3b5db 2227 }
87bf6cec 2228 }
850c9265 2229 }
2230
87bf6cec 2231 return ret;
850c9265 2232}
2233
69a3cdfc 2234// scan for pop starting from 'ret' op (all paths)
d4e3b5db 2235static int scan_for_pop_ret(int i, int opcnt, const char *reg,
2236 int flag_set)
850c9265 2237{
2238 int found = 0;
2239 int j;
2240
2241 for (; i < opcnt; i++) {
87bf6cec 2242 if (!(ops[i].flags & OPF_TAIL))
850c9265 2243 continue;
2244
2245 for (j = i - 1; j >= 0; j--) {
5e49b270 2246 if (ops[j].flags & (OPF_RMD|OPF_DONE))
69a3cdfc 2247 continue;
2248 if (ops[j].flags & OPF_JMP)
850c9265 2249 return -1;
2250
5e49b270 2251 if (ops[j].op == OP_POP && ops[j].datap == NULL
2252 && ops[j].operand[0].type == OPT_REG
850c9265 2253 && IS(ops[j].operand[0].name, reg))
2254 {
2255 found = 1;
d4e3b5db 2256 ops[j].flags |= flag_set;
850c9265 2257 break;
2258 }
2259
d7857c3a 2260 if (g_labels[j] != NULL)
850c9265 2261 return -1;
2262 }
2263 }
2264
2265 return found ? 0 : -1;
2266}
2267
25a330eb 2268static void scan_for_pop_const(int i, int opcnt, int *regmask_pp)
9af2d373 2269{
25a330eb 2270 struct parsed_op *po;
2271 int is_multipath = 0;
9af2d373 2272 int j;
2273
2274 for (j = i + 1; j < opcnt; j++) {
25a330eb 2275 po = &ops[j];
2276
2277 if (po->op == OP_JMP && po->btj == NULL) {
2278 ferr_assert(po, po->bt_i >= 0);
2279 j = po->bt_i - 1;
2280 continue;
2281 }
2282
2283 if ((po->flags & (OPF_JMP|OPF_TAIL|OPF_RSAVE))
2284 || po->op == OP_PUSH)
9af2d373 2285 {
2286 break;
2287 }
2288
25a330eb 2289 if (g_labels[j] != NULL)
2290 is_multipath = 1;
2291
2292 if (po->op == OP_POP && !(po->flags & OPF_RMD))
9af2d373 2293 {
25a330eb 2294 is_multipath |= !!(po->flags & OPF_PPUSH);
2295 if (is_multipath) {
2296 ops[i].flags |= OPF_PPUSH | OPF_DONE;
2297 ops[i].datap = po;
2298 po->flags |= OPF_PPUSH | OPF_DONE;
2299 *regmask_pp |= 1 << po->operand[0].reg;
2300 }
2301 else {
2302 ops[i].flags |= OPF_RMD | OPF_DONE;
2303 po->flags |= OPF_DONE;
2304 po->datap = &ops[i];
2305 }
9af2d373 2306 break;
2307 }
2308 }
2309}
2310
591721d7 2311static void scan_propagate_df(int i, int opcnt)
2312{
2313 struct parsed_op *po = &ops[i];
2314 int j;
2315
2316 for (; i < opcnt; i++) {
2317 po = &ops[i];
2318 if (po->flags & OPF_DF)
2319 return; // already resolved
2320 po->flags |= OPF_DF;
2321
2322 if (po->op == OP_CALL)
2323 ferr(po, "call with DF set?\n");
2324
2325 if (po->flags & OPF_JMP) {
2326 if (po->btj != NULL) {
2327 // jumptable
db63af51 2328 for (j = 0; j < po->btj->count; j++) {
2329 check_i(po, po->btj->d[j].bt_i);
591721d7 2330 scan_propagate_df(po->btj->d[j].bt_i, opcnt);
db63af51 2331 }
591721d7 2332 return;
2333 }
2334
db63af51 2335 check_i(po, po->bt_i);
5c024ef7 2336 if (po->flags & OPF_CJMP)
591721d7 2337 scan_propagate_df(po->bt_i, opcnt);
2338 else
2339 i = po->bt_i - 1;
2340 continue;
2341 }
2342
2343 if (po->flags & OPF_TAIL)
2344 break;
2345
2346 if (po->op == OP_CLD) {
5e49b270 2347 po->flags |= OPF_RMD | OPF_DONE;
591721d7 2348 return;
2349 }
2350 }
2351
2352 ferr(po, "missing DF clear?\n");
2353}
2354
db63af51 2355// is operand 'opr' referenced by parsed_op 'po'?
2356static int is_opr_referenced(const struct parsed_opr *opr,
2357 const struct parsed_op *po)
2358{
2359 int i, mask;
2360
2361 if (opr->type == OPT_REG) {
2362 mask = po->regmask_dst | po->regmask_src;
2363 if (po->op == OP_CALL)
2364 mask |= (1 << xAX) | (1 << xCX) | (1 << xDX);
2365 if ((1 << opr->reg) & mask)
2366 return 1;
2367 else
2368 return 0;
2369 }
2370
2371 for (i = 0; i < po->operand_cnt; i++)
2372 if (IS(po->operand[0].name, opr->name))
2373 return 1;
2374
2375 return 0;
2376}
2377
2378// is operand 'opr' read by parsed_op 'po'?
2379static int is_opr_read(const struct parsed_opr *opr,
2380 const struct parsed_op *po)
2381{
2382 int mask;
2383
2384 if (opr->type == OPT_REG) {
2385 mask = po->regmask_src;
2386 if (po->op == OP_CALL)
2387 // assume worst case
2388 mask |= (1 << xAX) | (1 << xCX) | (1 << xDX);
2389 if ((1 << opr->reg) & mask)
2390 return 1;
2391 else
2392 return 0;
2393 }
2394
2395 // yes I'm lazy
2396 return 0;
2397}
2398
1cd4a663 2399// is operand 'opr' modified by parsed_op 'po'?
5101a5f9 2400static int is_opr_modified(const struct parsed_opr *opr,
69a3cdfc 2401 const struct parsed_op *po)
2402{
89ff3147 2403 int mask;
2404
db63af51 2405 if (!(po->flags & OPF_DATA))
69a3cdfc 2406 return 0;
2407
89ff3147 2408 if (opr->type == OPT_REG) {
2409 if (po->op == OP_CALL) {
2410 mask = (1 << xAX) | (1 << xCX) | (1 << xDX);
2411 if ((1 << opr->reg) & mask)
2412 return 1;
2413 else
2414 return 0;
2415 }
2416
2417 if (po->operand[0].type == OPT_REG) {
2418 if (po->regmask_dst & (1 << opr->reg))
2419 return 1;
2420 else
2421 return 0;
2422 }
69a3cdfc 2423 }
2424
2425 return IS(po->operand[0].name, opr->name);
2426}
2427
5101a5f9 2428// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
2429static int is_any_opr_modified(const struct parsed_op *po_test,
89ff3147 2430 const struct parsed_op *po, int c_mode)
5101a5f9 2431{
89ff3147 2432 int mask;
5101a5f9 2433 int i;
2434
2435 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2436 return 0;
2437
de50b98b 2438 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2439 return 0;
2440
2441 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
2442 return 1;
2443
2444 // in reality, it can wreck any register, but in decompiled C
2b43685d 2445 // version it can only overwrite eax or edx:eax
89ff3147 2446 mask = (1 << xAX) | (1 << xDX);
2447 if (!c_mode)
2448 mask |= 1 << xCX;
2449
de50b98b 2450 if (po->op == OP_CALL
89ff3147 2451 && ((po_test->regmask_src | po_test->regmask_dst) & mask))
5101a5f9 2452 return 1;
2453
2454 for (i = 0; i < po_test->operand_cnt; i++)
2455 if (IS(po_test->operand[i].name, po->operand[0].name))
2456 return 1;
2457
2458 return 0;
2459}
2460
940e8e66 2461// scan for any po_test operand modification in range given
89ff3147 2462static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt,
2463 int c_mode)
69a3cdfc 2464{
2b43685d 2465 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2466 return -1;
2467
69a3cdfc 2468 for (; i < opcnt; i++) {
89ff3147 2469 if (is_any_opr_modified(po_test, &ops[i], c_mode))
69a3cdfc 2470 return i;
2471 }
2472
2473 return -1;
2474}
2475
940e8e66 2476// scan for po_test operand[0] modification in range given
2477static int scan_for_mod_opr0(struct parsed_op *po_test,
2478 int i, int opcnt)
2479{
2480 for (; i < opcnt; i++) {
2481 if (is_opr_modified(&po_test->operand[0], &ops[i]))
2482 return i;
2483 }
2484
2485 return -1;
2486}
2487
04f8a628 2488static int scan_for_flag_set(int i, int magic, int *branched,
2489 int *setters, int *setter_cnt)
69a3cdfc 2490{
04f8a628 2491 struct label_ref *lr;
2b43685d 2492 int ret;
de50b98b 2493
2494 while (i >= 0) {
04f8a628 2495 if (ops[i].cc_scratch == magic) {
2496 ferr(&ops[i], "%s looped\n", __func__);
2497 return -1;
2498 }
2499 ops[i].cc_scratch = magic;
2500
d7857c3a 2501 if (g_labels[i] != NULL) {
2b43685d 2502 *branched = 1;
04f8a628 2503
2504 lr = &g_label_refs[i];
2505 for (; lr->next; lr = lr->next) {
92d715b6 2506 check_i(&ops[i], lr->i);
04f8a628 2507 ret = scan_for_flag_set(lr->i, magic,
2508 branched, setters, setter_cnt);
2509 if (ret < 0)
2510 return ret;
2511 }
2512
92d715b6 2513 check_i(&ops[i], lr->i);
de50b98b 2514 if (i > 0 && LAST_OP(i - 1)) {
94d447fb 2515 i = lr->i;
de50b98b 2516 continue;
2517 }
04f8a628 2518 ret = scan_for_flag_set(lr->i, magic,
2519 branched, setters, setter_cnt);
2b43685d 2520 if (ret < 0)
2521 return ret;
de50b98b 2522 }
2523 i--;
2524
2b43685d 2525 if (ops[i].flags & OPF_FLAGS) {
2526 setters[*setter_cnt] = i;
2527 (*setter_cnt)++;
2528 return 0;
2529 }
69a3cdfc 2530
5c024ef7 2531 if ((ops[i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP)
69a3cdfc 2532 return -1;
69a3cdfc 2533 }
2534
2535 return -1;
2536}
2537
5101a5f9 2538// scan back for cdq, if anything modifies edx, fail
2539static int scan_for_cdq_edx(int i)
2540{
cdfaeed7 2541 while (i >= 0) {
d7857c3a 2542 if (g_labels[i] != NULL) {
cdfaeed7 2543 if (g_label_refs[i].next != NULL)
2544 return -1;
2545 if (i > 0 && LAST_OP(i - 1)) {
2546 i = g_label_refs[i].i;
2547 continue;
2548 }
2549 return -1;
2550 }
2551 i--;
2552
5101a5f9 2553 if (ops[i].op == OP_CDQ)
2554 return i;
2555
2556 if (ops[i].regmask_dst & (1 << xDX))
2557 return -1;
5101a5f9 2558 }
2559
2560 return -1;
2561}
2562
64c59faf 2563static int scan_for_reg_clear(int i, int reg)
2564{
cdfaeed7 2565 while (i >= 0) {
d7857c3a 2566 if (g_labels[i] != NULL) {
cdfaeed7 2567 if (g_label_refs[i].next != NULL)
2568 return -1;
2569 if (i > 0 && LAST_OP(i - 1)) {
2570 i = g_label_refs[i].i;
2571 continue;
2572 }
2573 return -1;
2574 }
2575 i--;
2576
64c59faf 2577 if (ops[i].op == OP_XOR
2578 && ops[i].operand[0].lmod == OPLM_DWORD
2579 && ops[i].operand[0].reg == ops[i].operand[1].reg
2580 && ops[i].operand[0].reg == reg)
2581 return i;
2582
2583 if (ops[i].regmask_dst & (1 << reg))
2584 return -1;
64c59faf 2585 }
2586
2587 return -1;
2588}
2589
ee2361b9 2590static void patch_esp_adjust(struct parsed_op *po, int adj)
2591{
2592 ferr_assert(po, po->op == OP_ADD);
2593 ferr_assert(po, IS(opr_name(po, 0), "esp"));
2594 ferr_assert(po, po->operand[1].type == OPT_CONST);
2595
2596 // this is a bit of a hack, but deals with use of
2597 // single adj for multiple calls
2598 po->operand[1].val -= adj;
2599 po->flags |= OPF_RMD;
2600 if (po->operand[1].val == 0)
2601 po->flags |= OPF_DONE;
2602 ferr_assert(po, (int)po->operand[1].val >= 0);
2603}
2604
1bafb621 2605// scan for positive, constant esp adjust
ee2361b9 2606// multipath case is preliminary
9af2d373 2607static int scan_for_esp_adjust(int i, int opcnt,
ee2361b9 2608 int adj_expect, int *adj, int *is_multipath, int do_update)
1bafb621 2609{
bfacdc83 2610 int adj_expect_unknown = 0;
7ba45c34 2611 struct parsed_op *po;
46411e6c 2612 int first_pop = -1;
bfacdc83 2613 int adj_best = 0;
4741fdfe 2614
ee2361b9 2615 *adj = *is_multipath = 0;
bfacdc83 2616 if (adj_expect < 0) {
2617 adj_expect_unknown = 1;
2618 adj_expect = 32 * 4; // enough?
2619 }
7ba45c34 2620
9af2d373 2621 for (; i < opcnt && *adj < adj_expect; i++) {
d7857c3a 2622 if (g_labels[i] != NULL)
ee2361b9 2623 *is_multipath = 1;
a2c1d768 2624
5e49b270 2625 po = &ops[i];
2626 if (po->flags & OPF_DONE)
2627 continue;
2628
7ba45c34 2629 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
2630 if (po->operand[1].type != OPT_CONST)
1bafb621 2631 ferr(&ops[i], "non-const esp adjust?\n");
7ba45c34 2632 *adj += po->operand[1].val;
1bafb621 2633 if (*adj & 3)
2634 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
ee2361b9 2635 if (do_update) {
2636 if (!*is_multipath)
2637 patch_esp_adjust(po, adj_expect);
2638 else
2639 po->flags |= OPF_RMD;
2640 }
1bafb621 2641 return i;
2642 }
5e49b270 2643 else if (po->op == OP_PUSH) {
5c024ef7 2644 //if (first_pop == -1)
2645 // first_pop = -2; // none
7ba45c34 2646 *adj -= lmod_bytes(po, po->operand[0].lmod);
46411e6c 2647 }
5e49b270 2648 else if (po->op == OP_POP) {
91ca764a 2649 if (!(po->flags & OPF_DONE)) {
2650 // seems like msvc only uses 'pop ecx' for stack realignment..
2651 if (po->operand[0].type != OPT_REG || po->operand[0].reg != xCX)
2652 break;
2653 if (first_pop == -1 && *adj >= 0)
2654 first_pop = i;
2655 }
ee2361b9 2656 if (do_update && *adj >= 0) {
2657 po->flags |= OPF_RMD;
2658 if (!*is_multipath)
2659 po->flags |= OPF_DONE;
2660 }
2661
7ba45c34 2662 *adj += lmod_bytes(po, po->operand[0].lmod);
bfacdc83 2663 if (*adj > adj_best)
2664 adj_best = *adj;
46411e6c 2665 }
e56ab892 2666 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
4741fdfe 2667 if (po->op == OP_JMP && po->btj == NULL) {
5e49b270 2668 if (po->bt_i <= i)
2669 break;
4741fdfe 2670 i = po->bt_i - 1;
2671 continue;
2672 }
e56ab892 2673 if (po->op != OP_CALL)
a2c1d768 2674 break;
e56ab892 2675 if (po->operand[0].type != OPT_LABEL)
a2c1d768 2676 break;
91ca764a 2677 if (po->pp != NULL && po->pp->is_stdcall)
89ff3147 2678 break;
bfacdc83 2679 if (adj_expect_unknown && first_pop >= 0)
2680 break;
91ca764a 2681 // assume it's another cdecl call
e56ab892 2682 }
a2c1d768 2683 }
7ba45c34 2684
108e9fe3 2685 if (first_pop >= 0) {
bfacdc83 2686 // probably only 'pop ecx' was used
2687 *adj = adj_best;
46411e6c 2688 return first_pop;
1bafb621 2689 }
2690
2691 return -1;
2692}
2693
a3684be1 2694static void scan_fwd_set_flags(int i, int opcnt, int magic, int flags)
2695{
2696 struct parsed_op *po;
2697 int j;
2698
2699 if (i < 0)
2700 ferr(ops, "%s: followed bad branch?\n", __func__);
2701
2702 for (; i < opcnt; i++) {
2703 po = &ops[i];
2704 if (po->cc_scratch == magic)
2705 return;
2706 po->cc_scratch = magic;
2707 po->flags |= flags;
2708
2709 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2710 if (po->btj != NULL) {
2711 // jumptable
2712 for (j = 0; j < po->btj->count; j++)
2713 scan_fwd_set_flags(po->btj->d[j].bt_i, opcnt, magic, flags);
2714 return;
2715 }
2716
2717 scan_fwd_set_flags(po->bt_i, opcnt, magic, flags);
5c024ef7 2718 if (!(po->flags & OPF_CJMP))
a3684be1 2719 return;
2720 }
2721 if (po->flags & OPF_TAIL)
2722 return;
2723 }
2724}
2725
1cd4a663 2726static const struct parsed_proto *try_recover_pp(
da87ae38 2727 struct parsed_op *po, const struct parsed_opr *opr, int *search_instead)
1cd4a663 2728{
2729 const struct parsed_proto *pp = NULL;
89ff3147 2730 char buf[256];
2731 char *p;
1cd4a663 2732
2733 // maybe an arg of g_func?
2734 if (opr->type == OPT_REGMEM && is_stack_access(po, opr))
2735 {
2736 char ofs_reg[16] = { 0, };
2737 int arg, arg_s, arg_i;
2738 int stack_ra = 0;
2739 int offset = 0;
2740
9af2d373 2741 if (g_header_mode)
2742 return NULL;
2743
1cd4a663 2744 parse_stack_access(po, opr->name, ofs_reg,
037f4971 2745 &offset, &stack_ra, NULL, 0);
1cd4a663 2746 if (ofs_reg[0] != 0)
2747 ferr(po, "offset reg on arg access?\n");
da87ae38 2748 if (offset <= stack_ra) {
2749 // search who set the stack var instead
2750 if (search_instead != NULL)
2751 *search_instead = 1;
2752 return NULL;
2753 }
1cd4a663 2754
2755 arg_i = (offset - stack_ra - 4) / 4;
2756 for (arg = arg_s = 0; arg < g_func_pp->argc; arg++) {
2757 if (g_func_pp->arg[arg].reg != NULL)
2758 continue;
2759 if (arg_s == arg_i)
2760 break;
2761 arg_s++;
2762 }
2763 if (arg == g_func_pp->argc)
2764 ferr(po, "stack arg %d not in prototype?\n", arg_i);
2765
2766 pp = g_func_pp->arg[arg].fptr;
2767 if (pp == NULL)
46411e6c 2768 ferr(po, "icall sa: arg%d is not a fptr?\n", arg + 1);
89ff3147 2769 check_func_pp(po, pp, "icall arg");
2770 }
2771 else if (opr->type == OPT_REGMEM && strchr(opr->name + 1, '[')) {
2772 // label[index]
2773 p = strchr(opr->name + 1, '[');
2774 memcpy(buf, opr->name, p - opr->name);
2775 buf[p - opr->name] = 0;
92d715b6 2776 pp = proto_parse(g_fhdr, buf, g_quiet_pp);
1cd4a663 2777 }
2778 else if (opr->type == OPT_OFFSET || opr->type == OPT_LABEL) {
9af2d373 2779 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
2780 if (pp == NULL) {
2781 if (!g_header_mode)
2782 ferr(po, "proto_parse failed for icall to '%s'\n", opr->name);
2783 }
2784 else
2785 check_func_pp(po, pp, "reg-fptr ref");
1cd4a663 2786 }
2787
2788 return pp;
2789}
2790
da87ae38 2791static void scan_for_call_type(int i, const struct parsed_opr *opr,
db63af51 2792 int magic, const struct parsed_proto **pp_found, int *pp_i,
2793 int *multi)
1cd4a663 2794{
2795 const struct parsed_proto *pp = NULL;
2796 struct parsed_op *po;
2797 struct label_ref *lr;
2798
037f4971 2799 ops[i].cc_scratch = magic;
1cd4a663 2800
037f4971 2801 while (1) {
d7857c3a 2802 if (g_labels[i] != NULL) {
1cd4a663 2803 lr = &g_label_refs[i];
92d715b6 2804 for (; lr != NULL; lr = lr->next) {
2805 check_i(&ops[i], lr->i);
db63af51 2806 scan_for_call_type(lr->i, opr, magic, pp_found, pp_i, multi);
92d715b6 2807 }
46411e6c 2808 if (i > 0 && LAST_OP(i - 1))
2809 return;
1cd4a663 2810 }
037f4971 2811
1cd4a663 2812 i--;
037f4971 2813 if (i < 0)
2814 break;
2815
2816 if (ops[i].cc_scratch == magic)
2817 return;
2818 ops[i].cc_scratch = magic;
1cd4a663 2819
2820 if (!(ops[i].flags & OPF_DATA))
2821 continue;
2822 if (!is_opr_modified(opr, &ops[i]))
2823 continue;
2824 if (ops[i].op != OP_MOV && ops[i].op != OP_LEA) {
2825 // most probably trashed by some processing
2826 *pp_found = NULL;
2827 return;
2828 }
2829
2830 opr = &ops[i].operand[1];
2831 if (opr->type != OPT_REG)
2832 break;
2833 }
2834
2835 po = (i >= 0) ? &ops[i] : ops;
2836
2837 if (i < 0) {
2838 // reached the top - can only be an arg-reg
2839 if (opr->type != OPT_REG)
2840 return;
2841
2842 for (i = 0; i < g_func_pp->argc; i++) {
2843 if (g_func_pp->arg[i].reg == NULL)
2844 continue;
2845 if (IS(opr->name, g_func_pp->arg[i].reg))
2846 break;
2847 }
2848 if (i == g_func_pp->argc)
2849 return;
2850 pp = g_func_pp->arg[i].fptr;
2851 if (pp == NULL)
46411e6c 2852 ferr(po, "icall: arg%d (%s) is not a fptr?\n",
2853 i + 1, g_func_pp->arg[i].reg);
89ff3147 2854 check_func_pp(po, pp, "icall reg-arg");
1cd4a663 2855 }
2856 else
da87ae38 2857 pp = try_recover_pp(po, opr, NULL);
1cd4a663 2858
89ff3147 2859 if (*pp_found != NULL && pp != NULL && *pp_found != pp) {
1cd4a663 2860 if (!IS((*pp_found)->ret_type.name, pp->ret_type.name)
2861 || (*pp_found)->is_stdcall != pp->is_stdcall
89ff3147 2862 || (*pp_found)->is_fptr != pp->is_fptr
1cd4a663 2863 || (*pp_found)->argc != pp->argc
2864 || (*pp_found)->argc_reg != pp->argc_reg
2865 || (*pp_found)->argc_stack != pp->argc_stack)
2866 {
2867 ferr(po, "icall: parsed_proto mismatch\n");
2868 }
89ff3147 2869 *multi = 1;
1cd4a663 2870 }
db63af51 2871 if (pp != NULL) {
1cd4a663 2872 *pp_found = pp;
db63af51 2873 *pp_i = po - ops;
2874 }
1cd4a663 2875}
2876
66bdb2b0 2877static void add_label_ref(struct label_ref *lr, int op_i)
9af2d373 2878{
66bdb2b0 2879 struct label_ref *lr_new;
9af2d373 2880
66bdb2b0 2881 if (lr->i == -1) {
2882 lr->i = op_i;
2883 return;
2884 }
9af2d373 2885
66bdb2b0 2886 lr_new = calloc(1, sizeof(*lr_new));
2887 lr_new->i = op_i;
2888 lr_new->next = lr->next;
2889 lr->next = lr_new;
2890}
2891
2892static struct parsed_data *try_resolve_jumptab(int i, int opcnt)
2893{
2894 struct parsed_op *po = &ops[i];
2895 struct parsed_data *pd;
2896 char label[NAMELEN], *p;
2897 int len, j, l;
2898
2899 p = strchr(po->operand[0].name, '[');
2900 if (p == NULL)
2901 return NULL;
2902
2903 len = p - po->operand[0].name;
2904 strncpy(label, po->operand[0].name, len);
2905 label[len] = 0;
2906
2907 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
2908 if (IS(g_func_pd[j].label, label)) {
2909 pd = &g_func_pd[j];
2910 break;
2911 }
2912 }
2913 if (pd == NULL)
2914 //ferr(po, "label '%s' not parsed?\n", label);
2915 return NULL;
2916
2917 if (pd->type != OPT_OFFSET)
2918 ferr(po, "label '%s' with non-offset data?\n", label);
2919
2920 // find all labels, link
2921 for (j = 0; j < pd->count; j++) {
2922 for (l = 0; l < opcnt; l++) {
2923 if (g_labels[l] != NULL && IS(g_labels[l], pd->d[j].u.label)) {
2924 add_label_ref(&g_label_refs[l], i);
2925 pd->d[j].bt_i = l;
2926 break;
2927 }
2928 }
2929 }
2930
2931 return pd;
2932}
2933
2934static void clear_labels(int count)
2935{
2936 int i;
2937
2938 for (i = 0; i < count; i++) {
2939 if (g_labels[i] != NULL) {
2940 free(g_labels[i]);
2941 g_labels[i] = NULL;
2942 }
2943 }
2944}
2945
2946static void resolve_branches_parse_calls(int opcnt)
2947{
2948 const struct parsed_proto *pp_c;
2949 struct parsed_proto *pp;
2950 struct parsed_data *pd;
2951 struct parsed_op *po;
2952 const char *tmpname;
2953 int i, l, ret;
2954
2955 for (i = 0; i < opcnt; i++)
2956 {
2957 po = &ops[i];
2958 po->bt_i = -1;
2959 po->btj = NULL;
2960
2961 if (po->op == OP_CALL) {
2962 pp = NULL;
2963
2964 if (po->operand[0].type == OPT_LABEL) {
2965 tmpname = opr_name(po, 0);
2966 if (IS_START(tmpname, "loc_"))
2967 ferr(po, "call to loc_*\n");
2968 pp_c = proto_parse(g_fhdr, tmpname, g_header_mode);
2969 if (!g_header_mode && pp_c == NULL)
2970 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
2971
2972 if (pp_c != NULL) {
2973 pp = proto_clone(pp_c);
2974 my_assert_not(pp, NULL);
2975 }
2976 }
2977 else if (po->datap != NULL) {
2978 pp = calloc(1, sizeof(*pp));
2979 my_assert_not(pp, NULL);
2980
2981 ret = parse_protostr(po->datap, pp);
2982 if (ret < 0)
2983 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
2984 free(po->datap);
2985 po->datap = NULL;
2986 }
2987
2988 if (pp != NULL) {
2989 if (pp->is_fptr)
2990 check_func_pp(po, pp, "fptr var call");
2991 if (pp->is_noreturn)
2992 po->flags |= OPF_TAIL;
2993 }
2994 po->pp = pp;
2995 continue;
2996 }
2997
2998 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
2999 continue;
3000
3001 if (po->operand[0].type == OPT_REGMEM) {
3002 pd = try_resolve_jumptab(i, opcnt);
3003 if (pd == NULL)
3004 goto tailcall;
3005
3006 po->btj = pd;
3007 continue;
3008 }
3009
3010 for (l = 0; l < opcnt; l++) {
3011 if (g_labels[l] != NULL
3012 && IS(po->operand[0].name, g_labels[l]))
3013 {
3014 if (l == i + 1 && po->op == OP_JMP) {
3015 // yet another alignment type..
3016 po->flags |= OPF_RMD|OPF_DONE;
3017 break;
3018 }
3019 add_label_ref(&g_label_refs[l], i);
3020 po->bt_i = l;
3021 break;
3022 }
3023 }
3024
3025 if (po->bt_i != -1 || (po->flags & OPF_RMD))
3026 continue;
3027
3028 if (po->operand[0].type == OPT_LABEL)
3029 // assume tail call
3030 goto tailcall;
3031
3032 ferr(po, "unhandled branch\n");
3033
3034tailcall:
3035 po->op = OP_CALL;
3036 po->flags |= OPF_TAIL;
3037 if (i > 0 && ops[i - 1].op == OP_POP)
3038 po->flags |= OPF_ATAIL;
3039 i--; // reprocess
3040 }
9af2d373 3041}
3042
3043static void scan_prologue_epilogue(int opcnt)
3044{
3045 int ecx_push = 0, esp_sub = 0;
3046 int found;
3047 int i, j, l;
3048
3049 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
3050 && ops[1].op == OP_MOV
3051 && IS(opr_name(&ops[1], 0), "ebp")
3052 && IS(opr_name(&ops[1], 1), "esp"))
3053 {
3054 g_bp_frame = 1;
5e49b270 3055 ops[0].flags |= OPF_RMD | OPF_DONE;
3056 ops[1].flags |= OPF_RMD | OPF_DONE;
9af2d373 3057 i = 2;
3058
3059 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
3060 g_stack_fsz = opr_const(&ops[2], 1);
5e49b270 3061 ops[2].flags |= OPF_RMD | OPF_DONE;
9af2d373 3062 i++;
3063 }
3064 else {
3065 // another way msvc builds stack frame..
3066 i = 2;
3067 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
3068 g_stack_fsz += 4;
5e49b270 3069 ops[i].flags |= OPF_RMD | OPF_DONE;
9af2d373 3070 ecx_push++;
3071 i++;
3072 }
3073 // and another way..
3074 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
3075 && ops[i].operand[1].type == OPT_CONST
3076 && ops[i + 1].op == OP_CALL
3077 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
3078 {
3079 g_stack_fsz += ops[i].operand[1].val;
5e49b270 3080 ops[i].flags |= OPF_RMD | OPF_DONE;
9af2d373 3081 i++;
5e49b270 3082 ops[i].flags |= OPF_RMD | OPF_DONE;
9af2d373 3083 i++;
3084 }
3085 }
3086
3087 found = 0;
3088 do {
3089 for (; i < opcnt; i++)
66bdb2b0 3090 if (ops[i].flags & OPF_TAIL)
9af2d373 3091 break;
3092 j = i - 1;
3093 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
66bdb2b0 3094 if (ops[j].bt_i != -1 || ops[j].btj != NULL)
3095 break;
3096 i--;
9af2d373 3097 j--;
3098 }
3099
3100 if ((ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ebp"))
3101 || ops[j].op == OP_LEAVE)
3102 {
5e49b270 3103 ops[j].flags |= OPF_RMD | OPF_DONE;
9af2d373 3104 }
66bdb2b0 3105 else if (ops[i].op == OP_CALL && ops[i].pp != NULL
3106 && ops[i].pp->is_noreturn)
3107 {
3108 // on noreturn, msvc sometimes cleans stack, sometimes not
3109 i++;
3110 found = 1;
3111 continue;
3112 }
9af2d373 3113 else if (!(g_ida_func_attr & IDAFA_NORETURN))
3114 ferr(&ops[j], "'pop ebp' expected\n");
3115
3116 if (g_stack_fsz != 0) {
bfacdc83 3117 if (ops[j].op == OP_LEAVE)
3118 j--;
3119 else if (ops[j].op == OP_POP
3120 && ops[j - 1].op == OP_MOV
9af2d373 3121 && IS(opr_name(&ops[j - 1], 0), "esp")
3122 && IS(opr_name(&ops[j - 1], 1), "ebp"))
3123 {
5e49b270 3124 ops[j - 1].flags |= OPF_RMD | OPF_DONE;
bfacdc83 3125 j -= 2;
9af2d373 3126 }
bfacdc83 3127 else if (!(g_ida_func_attr & IDAFA_NORETURN))
9af2d373 3128 {
bfacdc83 3129 ferr(&ops[j], "esp restore expected\n");
9af2d373 3130 }
3131
bfacdc83 3132 if (ecx_push && j >= 0 && ops[j].op == OP_POP
3133 && IS(opr_name(&ops[j], 0), "ecx"))
9af2d373 3134 {
bfacdc83 3135 ferr(&ops[j], "unexpected ecx pop\n");
9af2d373 3136 }
3137 }
3138
3139 found = 1;
3140 i++;
3141 } while (i < opcnt);
3142
66bdb2b0 3143 if (!found)
3144 ferr(ops, "missing ebp epilogue\n");
9af2d373 3145 return;
3146 }
3147
3148 // non-bp frame
3149 i = 0;
3150 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
5e49b270 3151 ops[i].flags |= OPF_RMD | OPF_DONE;
9af2d373 3152 g_stack_fsz += 4;
3153 ecx_push++;
3154 i++;
3155 }
3156
3157 for (; i < opcnt; i++) {
3158 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
3159 break;
3160 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
3161 && ops[i].operand[1].type == OPT_CONST)
3162 {
3163 g_stack_fsz = ops[i].operand[1].val;
5e49b270 3164 ops[i].flags |= OPF_RMD | OPF_DONE;
9af2d373 3165 esp_sub = 1;
3166 break;
3167 }
3168 }
3169
3170 if (ecx_push && !esp_sub) {
3171 // could actually be args for a call..
3172 for (; i < opcnt; i++)
3173 if (ops[i].op != OP_PUSH)
3174 break;
3175
3176 if (ops[i].op == OP_CALL && ops[i].operand[0].type == OPT_LABEL) {
3177 const struct parsed_proto *pp;
3178 pp = proto_parse(g_fhdr, opr_name(&ops[i], 0), 1);
3179 j = pp ? pp->argc_stack : 0;
3180 while (i > 0 && j > 0) {
3181 i--;
3182 if (ops[i].op == OP_PUSH) {
5e49b270 3183 ops[i].flags &= ~(OPF_RMD | OPF_DONE);
9af2d373 3184 j--;
3185 }
3186 }
3187 if (j != 0)
3188 ferr(&ops[i], "unhandled prologue\n");
3189
3190 // recheck
3191 i = g_stack_fsz = ecx_push = 0;
3192 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
3193 if (!(ops[i].flags & OPF_RMD))
3194 break;
3195 g_stack_fsz += 4;
3196 ecx_push++;
3197 i++;
3198 }
3199 }
3200 }
3201
3202 found = 0;
3203 if (ecx_push || esp_sub)
3204 {
3205 g_sp_frame = 1;
3206
3207 i++;
3208 do {
3209 for (; i < opcnt; i++)
66bdb2b0 3210 if (ops[i].flags & OPF_TAIL)
9af2d373 3211 break;
3212 j = i - 1;
3213 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
66bdb2b0 3214 if (ops[j].bt_i != -1 || ops[j].btj != NULL)
3215 break;
3216 i--;
9af2d373 3217 j--;
3218 }
3219
3220 if (ecx_push > 0) {
3221 for (l = 0; l < ecx_push; l++) {
3222 if (ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ecx"))
3223 /* pop ecx */;
3224 else if (ops[j].op == OP_ADD
3225 && IS(opr_name(&ops[j], 0), "esp")
3226 && ops[j].operand[1].type == OPT_CONST)
3227 {
3228 /* add esp, N */
5e49b270 3229 l += ops[j].operand[1].val / 4 - 1;
9af2d373 3230 }
3231 else
3232 ferr(&ops[j], "'pop ecx' expected\n");
3233
5e49b270 3234 ops[j].flags |= OPF_RMD | OPF_DONE;
9af2d373 3235 j--;
3236 }
3237 if (l != ecx_push)
3238 ferr(&ops[j], "epilogue scan failed\n");
3239
3240 found = 1;
3241 }
3242
3243 if (esp_sub) {
3244 if (ops[j].op != OP_ADD
3245 || !IS(opr_name(&ops[j], 0), "esp")
3246 || ops[j].operand[1].type != OPT_CONST
3247 || ops[j].operand[1].val != g_stack_fsz)
3248 ferr(&ops[j], "'add esp' expected\n");
3249
5e49b270 3250 ops[j].flags |= OPF_RMD | OPF_DONE;
9af2d373 3251 ops[j].operand[1].val = 0; // hack for stack arg scanner
3252 found = 1;
3253 }
3254
3255 i++;
3256 } while (i < opcnt);
66bdb2b0 3257
3258 if (!found)
3259 ferr(ops, "missing esp epilogue\n");
9af2d373 3260 }
3261}
3262
89ff3147 3263static const struct parsed_proto *resolve_icall(int i, int opcnt,
db63af51 3264 int *pp_i, int *multi_src)
1cd4a663 3265{
3266 const struct parsed_proto *pp = NULL;
da87ae38 3267 int search_advice = 0;
1cd4a663 3268
89ff3147 3269 *multi_src = 0;
db63af51 3270 *pp_i = -1;
89ff3147 3271
1cd4a663 3272 switch (ops[i].operand[0].type) {
3273 case OPT_REGMEM:
3274 case OPT_LABEL:
3275 case OPT_OFFSET:
da87ae38 3276 pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice);
3277 if (!search_advice)
3278 break;
3279 // fallthrough
1cd4a663 3280 default:
89ff3147 3281 scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp,
db63af51 3282 pp_i, multi_src);
1cd4a663 3283 break;
3284 }
3285
3286 return pp;
3287}
3288
23fd0b11 3289// find an instruction that changed opr before i op
db63af51 3290// *op_i must be set to -1 by the caller
92d715b6 3291// *entry is set to 1 if one source is determined to be the caller
3292// returns 1 if found, *op_i is then set to origin
23fd0b11 3293static int resolve_origin(int i, const struct parsed_opr *opr,
92d715b6 3294 int magic, int *op_i, int *is_caller)
1f84f6b3 3295{
3296 struct label_ref *lr;
3297 int ret = 0;
3298
92d715b6 3299 if (ops[i].cc_scratch == magic)
3300 return 0;
1f84f6b3 3301 ops[i].cc_scratch = magic;
3302
3303 while (1) {
d7857c3a 3304 if (g_labels[i] != NULL) {
1f84f6b3 3305 lr = &g_label_refs[i];
92d715b6 3306 for (; lr != NULL; lr = lr->next) {
3307 check_i(&ops[i], lr->i);
3308 ret |= resolve_origin(lr->i, opr, magic, op_i, is_caller);
3309 }
1f84f6b3 3310 if (i > 0 && LAST_OP(i - 1))
3311 return ret;
3312 }
3313
3314 i--;
92d715b6 3315 if (i < 0) {
3316 if (is_caller != NULL)
3317 *is_caller = 1;
1f84f6b3 3318 return -1;
92d715b6 3319 }
1f84f6b3 3320
3321 if (ops[i].cc_scratch == magic)
ebc4dc43 3322 return ret;
1f84f6b3 3323 ops[i].cc_scratch = magic;
3324
3325 if (!(ops[i].flags & OPF_DATA))
3326 continue;
3327 if (!is_opr_modified(opr, &ops[i]))
3328 continue;
23fd0b11 3329
3330 if (*op_i >= 0) {
3331 if (*op_i == i)
ebc4dc43 3332 return ret | 1;
3333
23fd0b11 3334 // XXX: could check if the other op does the same
3335 return -1;
3336 }
3337
3338 *op_i = i;
ebc4dc43 3339 return ret | 1;
23fd0b11 3340 }
3341}
3342
db63af51 3343// find an instruction that previously referenced opr
3344// if multiple results are found - fail
3345// *op_i must be set to -1 by the caller
3346// returns 1 if found, *op_i is then set to referencer insn
3347static int resolve_last_ref(int i, const struct parsed_opr *opr,
3348 int magic, int *op_i)
3349{
3350 struct label_ref *lr;
3351 int ret = 0;
3352
3353 if (ops[i].cc_scratch == magic)
3354 return 0;
3355 ops[i].cc_scratch = magic;
3356
3357 while (1) {
3358 if (g_labels[i] != NULL) {
3359 lr = &g_label_refs[i];
3360 for (; lr != NULL; lr = lr->next) {
3361 check_i(&ops[i], lr->i);
3362 ret |= resolve_last_ref(lr->i, opr, magic, op_i);
3363 }
3364 if (i > 0 && LAST_OP(i - 1))
3365 return ret;
3366 }
3367
3368 i--;
3369 if (i < 0)
3370 return -1;
3371
3372 if (ops[i].cc_scratch == magic)
3373 return 0;
3374 ops[i].cc_scratch = magic;
3375
3376 if (!is_opr_referenced(opr, &ops[i]))
3377 continue;
3378
3379 if (*op_i >= 0)
3380 return -1;
3381
3382 *op_i = i;
3383 return 1;
3384 }
3385}
3386
3387// find next instruction that reads opr
3388// if multiple results are found - fail
3389// *op_i must be set to -1 by the caller
3390// returns 1 if found, *op_i is then set to referencer insn
3391static int find_next_read(int i, int opcnt,
3392 const struct parsed_opr *opr, int magic, int *op_i)
3393{
3394 struct parsed_op *po;
3395 int j, ret = 0;
3396
3397 for (; i < opcnt; i++)
3398 {
3399 if (ops[i].cc_scratch == magic)
3400 return 0;
3401 ops[i].cc_scratch = magic;
3402
3403 po = &ops[i];
3404 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
3405 if (po->btj != NULL) {
3406 // jumptable
3407 for (j = 0; j < po->btj->count; j++) {
3408 check_i(po, po->btj->d[j].bt_i);
3409 ret |= find_next_read(po->btj->d[j].bt_i, opcnt, opr,
3410 magic, op_i);
3411 }
3412 return ret;
3413 }
3414
3415 if (po->flags & OPF_RMD)
3416 continue;
3417 check_i(po, po->bt_i);
3418 if (po->flags & OPF_CJMP) {
3419 ret = find_next_read(po->bt_i, opcnt, opr, magic, op_i);
3420 if (ret < 0)
3421 return ret;
3422 }
3423
3424 i = po->bt_i - 1;
3425 continue;
3426 }
3427
3428 if (!is_opr_read(opr, po)) {
3429 if (is_opr_modified(opr, po))
3430 // it's overwritten
3431 return 0;
3432 if (po->flags & OPF_TAIL)
3433 return 0;
3434 continue;
3435 }
3436
3437 if (*op_i >= 0)
3438 return -1;
3439
3440 *op_i = i;
3441 return 1;
3442 }
3443
3444 return 0;
3445}
3446
23fd0b11 3447static int try_resolve_const(int i, const struct parsed_opr *opr,
3448 int magic, unsigned int *val)
3449{
3450 int s_i = -1;
92d715b6 3451 int ret;
23fd0b11 3452
92d715b6 3453 ret = resolve_origin(i, opr, magic, &s_i, NULL);
23fd0b11 3454 if (ret == 1) {
3455 i = s_i;
1f84f6b3 3456 if (ops[i].op != OP_MOV && ops[i].operand[1].type != OPT_CONST)
3457 return -1;
3458
3459 *val = ops[i].operand[1].val;
3460 return 1;
3461 }
23fd0b11 3462
3463 return -1;
1f84f6b3 3464}
3465
26677139 3466static struct parsed_proto *process_call_early(int i, int opcnt,
3467 int *adj_i)
3468{
3469 struct parsed_op *po = &ops[i];
3470 struct parsed_proto *pp;
3471 int multipath = 0;
3472 int adj = 0;
bfacdc83 3473 int j, ret;
26677139 3474
3475 pp = po->pp;
3476 if (pp == NULL || pp->is_vararg || pp->argc_reg != 0)
3477 // leave for later
3478 return NULL;
3479
3480 // look for and make use of esp adjust
3481 *adj_i = ret = -1;
3482 if (!pp->is_stdcall && pp->argc_stack > 0)
3483 ret = scan_for_esp_adjust(i + 1, opcnt,
ee2361b9 3484 pp->argc_stack * 4, &adj, &multipath, 0);
26677139 3485 if (ret >= 0) {
3486 if (pp->argc_stack > adj / 4)
3487 return NULL;
3488 if (multipath)
3489 return NULL;
bfacdc83 3490 if (ops[ret].op == OP_POP) {
3491 for (j = 1; j < adj / 4; j++) {
3492 if (ops[ret + j].op != OP_POP
3493 || ops[ret + j].operand[0].reg != xCX)
3494 {
3495 return NULL;
3496 }
3497 }
3498 }
26677139 3499 }
3500
3501 *adj_i = ret;
3502 return pp;
3503}
3504
9af2d373 3505static struct parsed_proto *process_call(int i, int opcnt)
3506{
3507 struct parsed_op *po = &ops[i];
3508 const struct parsed_proto *pp_c;
3509 struct parsed_proto *pp;
3510 const char *tmpname;
db63af51 3511 int call_i = -1, ref_i = -1;
26677139 3512 int adj = 0, multipath = 0;
9af2d373 3513 int ret, arg;
3514
3515 tmpname = opr_name(po, 0);
3516 pp = po->pp;
3517 if (pp == NULL)
3518 {
3519 // indirect call
db63af51 3520 pp_c = resolve_icall(i, opcnt, &call_i, &multipath);
9af2d373 3521 if (pp_c != NULL) {
3522 if (!pp_c->is_func && !pp_c->is_fptr)
3523 ferr(po, "call to non-func: %s\n", pp_c->name);
3524 pp = proto_clone(pp_c);
3525 my_assert_not(pp, NULL);
26677139 3526 if (multipath)
9af2d373 3527 // not resolved just to single func
3528 pp->is_fptr = 1;
3529
3530 switch (po->operand[0].type) {
3531 case OPT_REG:
3532 // we resolved this call and no longer need the register
3533 po->regmask_src &= ~(1 << po->operand[0].reg);
db63af51 3534
3535 if (!multipath && i != call_i && ops[call_i].op == OP_MOV
3536 && ops[call_i].operand[1].type == OPT_LABEL)
3537 {
3538 // no other source users?
3539 ret = resolve_last_ref(i, &po->operand[0], opcnt * 10,
3540 &ref_i);
3541 if (ret == 1 && call_i == ref_i) {
3542 // and nothing uses it after us?
3543 ref_i = -1;
3544 ret = find_next_read(i + 1, opcnt, &po->operand[0],
3545 opcnt * 11, &ref_i);
3546 if (ret != 1)
3547 // then also don't need the source mov
3548 ops[call_i].flags |= OPF_RMD;
3549 }
3550 }
9af2d373 3551 break;
3552 case OPT_REGMEM:
3553 pp->is_fptr = 1;
3554 break;
3555 default:
3556 break;
3557 }
3558 }
3559 if (pp == NULL) {
3560 pp = calloc(1, sizeof(*pp));
3561 my_assert_not(pp, NULL);
3562
3563 pp->is_fptr = 1;
ee2361b9 3564 ret = scan_for_esp_adjust(i + 1, opcnt,
bfacdc83 3565 -1, &adj, &multipath, 0);
26677139 3566 if (ret < 0 || adj < 0) {
9af2d373 3567 if (!g_allow_regfunc)
3568 ferr(po, "non-__cdecl indirect call unhandled yet\n");
3569 pp->is_unresolved = 1;
26677139 3570 adj = 0;
9af2d373 3571 }
26677139 3572 adj /= 4;
3573 if (adj > ARRAY_SIZE(pp->arg))
3574 ferr(po, "esp adjust too large: %d\n", adj);
9af2d373 3575 pp->ret_type.name = strdup("int");
26677139 3576 pp->argc = pp->argc_stack = adj;
9af2d373 3577 for (arg = 0; arg < pp->argc; arg++)
3578 pp->arg[arg].type.name = strdup("int");
3579 }
3580 po->pp = pp;
3581 }
3582
3583 // look for and make use of esp adjust
91ca764a 3584 multipath = 0;
9af2d373 3585 ret = -1;
bfacdc83 3586 if (!pp->is_stdcall && pp->argc_stack > 0) {
3587 int adj_expect = pp->is_vararg ? -1 : pp->argc_stack * 4;
9af2d373 3588 ret = scan_for_esp_adjust(i + 1, opcnt,
bfacdc83 3589 adj_expect, &adj, &multipath, 0);
3590 }
9af2d373 3591 if (ret >= 0) {
3592 if (pp->is_vararg) {
26677139 3593 if (adj / 4 < pp->argc_stack) {
3594 fnote(po, "(this call)\n");
3595 ferr(&ops[ret], "esp adjust is too small: %x < %x\n",
3596 adj, pp->argc_stack * 4);
3597 }
9af2d373 3598 // modify pp to make it have varargs as normal args
3599 arg = pp->argc;
26677139 3600 pp->argc += adj / 4 - pp->argc_stack;
9af2d373 3601 for (; arg < pp->argc; arg++) {
3602 pp->arg[arg].type.name = strdup("int");
3603 pp->argc_stack++;
3604 }
3605 if (pp->argc > ARRAY_SIZE(pp->arg))
3606 ferr(po, "too many args for '%s'\n", tmpname);
3607 }
26677139 3608 if (pp->argc_stack > adj / 4) {
9af2d373 3609 fnote(po, "(this call)\n");
3610 ferr(&ops[ret], "stack tracking failed for '%s': %x %x\n",
26677139 3611 tmpname, pp->argc_stack * 4, adj);
9af2d373 3612 }
3613
ee2361b9 3614 scan_for_esp_adjust(i + 1, opcnt,
3615 pp->argc_stack * 4, &adj, &multipath, 1);
9af2d373 3616 }
3617 else if (pp->is_vararg)
3618 ferr(po, "missing esp_adjust for vararg func '%s'\n",
3619 pp->name);
3620
3621 return pp;
3622}
3623
26677139 3624static int collect_call_args_early(struct parsed_op *po, int i,
3625 struct parsed_proto *pp, int *regmask)
3626{
3627 int arg, ret;
3628 int j;
3629
3630 for (arg = 0; arg < pp->argc; arg++)
3631 if (pp->arg[arg].reg == NULL)
3632 break;
3633
3634 // first see if it can be easily done
3635 for (j = i; j > 0 && arg < pp->argc; )
3636 {
3637 if (g_labels[j] != NULL)
3638 return -1;
3639 j--;
3640
3641 if (ops[j].op == OP_CALL)
3642 return -1;
3643 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP)
3644 return -1;
3645 else if (ops[j].op == OP_POP)
3646 return -1;
3647 else if (ops[j].flags & OPF_CJMP)
3648 return -1;
3649 else if (ops[j].op == OP_PUSH) {
3650 if (ops[j].flags & (OPF_FARG|OPF_FARGNR))
3651 return -1;
3652 ret = scan_for_mod(&ops[j], j + 1, i, 1);
3653 if (ret >= 0)
3654 return -1;
3655
3656 if (pp->arg[arg].type.is_va_list)
3657 return -1;
3658
3659 // next arg
3660 for (arg++; arg < pp->argc; arg++)
3661 if (pp->arg[arg].reg == NULL)
3662 break;
3663 }
3664 }
3665
3666 if (arg < pp->argc)
3667 return -1;
3668
3669 // now do it
3670 for (arg = 0; arg < pp->argc; arg++)
3671 if (pp->arg[arg].reg == NULL)
3672 break;
3673
3674 for (j = i; j > 0 && arg < pp->argc; )
3675 {
3676 j--;
3677
3678 if (ops[j].op == OP_PUSH)
3679 {
3680 ops[j].p_argnext = -1;
3681 ferr_assert(&ops[j], pp->arg[arg].datap == NULL);
3682 pp->arg[arg].datap = &ops[j];
3683
3684 if (ops[j].operand[0].type == OPT_REG)
3685 *regmask |= 1 << ops[j].operand[0].reg;
3686
5e49b270 3687 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_FARGNR | OPF_FARG;
26677139 3688 ops[j].flags &= ~OPF_RSAVE;
3689
3690 // next arg
3691 for (arg++; arg < pp->argc; arg++)
3692 if (pp->arg[arg].reg == NULL)
3693 break;
3694 }
3695 }
3696
3697 return 0;
3698}
3699
89ff3147 3700static int collect_call_args_r(struct parsed_op *po, int i,
3a5101d7 3701 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
3702 int *arg_grp, int arg, int magic, int need_op_saving, int may_reuse)
e56ab892 3703{
3704 struct parsed_proto *pp_tmp;
3a5101d7 3705 struct parsed_op *po_tmp;
e56ab892 3706 struct label_ref *lr;
2b43685d 3707 int need_to_save_current;
3a5101d7 3708 int arg_grp_current = 0;
3709 int save_args_seen = 0;
23fd0b11 3710 int save_args;
e56ab892 3711 int ret = 0;
5f70a34f 3712 int reg;
23fd0b11 3713 char buf[32];
3714 int j, k;
e56ab892 3715
a3684be1 3716 if (i < 0) {
a2c1d768 3717 ferr(po, "dead label encountered\n");
a3684be1 3718 return -1;
3719 }
e56ab892 3720
de50b98b 3721 for (; arg < pp->argc; arg++)
e56ab892 3722 if (pp->arg[arg].reg == NULL)
3723 break;
a3684be1 3724 magic = (magic & 0xffffff) | (arg << 24);
e56ab892 3725
89ff3147 3726 for (j = i; j >= 0 && (arg < pp->argc || pp->is_unresolved); )
e56ab892 3727 {
a3684be1 3728 if (((ops[j].cc_scratch ^ magic) & 0xffffff) == 0) {
3729 if (ops[j].cc_scratch != magic) {
3730 ferr(&ops[j], "arg collect hit same path with diff args for %s\n",
3731 pp->name);
3732 return -1;
3733 }
3734 // ok: have already been here
3735 return 0;
3736 }
3737 ops[j].cc_scratch = magic;
3738
d7857c3a 3739 if (g_labels[j] != NULL && g_label_refs[j].i != -1) {
e56ab892 3740 lr = &g_label_refs[j];
3741 if (lr->next != NULL)
3742 need_op_saving = 1;
a652aa9f 3743 for (; lr->next; lr = lr->next) {
92d715b6 3744 check_i(&ops[j], lr->i);
5c024ef7 3745 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
a652aa9f 3746 may_reuse = 1;
89ff3147 3747 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
3a5101d7 3748 arg_grp, arg, magic, need_op_saving, may_reuse);
a3684be1 3749 if (ret < 0)
3750 return ret;
a652aa9f 3751 }
e56ab892 3752
92d715b6 3753 check_i(&ops[j], lr->i);
5c024ef7 3754 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
a652aa9f 3755 may_reuse = 1;
de50b98b 3756 if (j > 0 && LAST_OP(j - 1)) {
e56ab892 3757 // follow last branch in reverse
3758 j = lr->i;
3759 continue;
3760 }
3761 need_op_saving = 1;
89ff3147 3762 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
3a5101d7 3763 arg_grp, arg, magic, need_op_saving, may_reuse);
a3684be1 3764 if (ret < 0)
3765 return ret;
e56ab892 3766 }
3767 j--;
3768
3769 if (ops[j].op == OP_CALL)
3770 {
89ff3147 3771 if (pp->is_unresolved)
3772 break;
3773
092f64e1 3774 pp_tmp = ops[j].pp;
e56ab892 3775 if (pp_tmp == NULL)
7ae48d73 3776 ferr(po, "arg collect hit unparsed call '%s'\n",
3777 ops[j].operand[0].name);
a652aa9f 3778 if (may_reuse && pp_tmp->argc_stack > 0)
de50b98b 3779 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
3780 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
e56ab892 3781 }
fdd5548a 3782 // esp adjust of 0 means we collected it before
3783 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP
3784 && (ops[j].operand[1].type != OPT_CONST
3785 || ops[j].operand[1].val != 0))
3786 {
89ff3147 3787 if (pp->is_unresolved)
3788 break;
3789
fdd5548a 3790 ferr(po, "arg collect %d/%d hit esp adjust of %d\n",
3791 arg, pp->argc, ops[j].operand[1].val);
de50b98b 3792 }
5e49b270 3793 else if (ops[j].op == OP_POP && !(ops[j].flags & OPF_DONE))
9af2d373 3794 {
89ff3147 3795 if (pp->is_unresolved)
3796 break;
3797
de50b98b 3798 ferr(po, "arg collect %d/%d hit pop\n", arg, pp->argc);
3799 }
5c024ef7 3800 else if (ops[j].flags & OPF_CJMP)
de50b98b 3801 {
89ff3147 3802 if (pp->is_unresolved)
3803 break;
3804
a652aa9f 3805 may_reuse = 1;
de50b98b 3806 }
91ca764a 3807 else if (ops[j].op == OP_PUSH
3808 && !(ops[j].flags & (OPF_FARGNR|OPF_DONE)))
e56ab892 3809 {
89ff3147 3810 if (pp->is_unresolved && (ops[j].flags & OPF_RMD))
3811 break;
3812
3a5101d7 3813 ops[j].p_argnext = -1;
3814 po_tmp = pp->arg[arg].datap;
3815 if (po_tmp != NULL)
3816 ops[j].p_argnext = po_tmp - ops;
e56ab892 3817 pp->arg[arg].datap = &ops[j];
3a5101d7 3818
2b43685d 3819 need_to_save_current = 0;
23fd0b11 3820 save_args = 0;
5f70a34f 3821 reg = -1;
3822 if (ops[j].operand[0].type == OPT_REG)
3823 reg = ops[j].operand[0].reg;
3824
e56ab892 3825 if (!need_op_saving) {
89ff3147 3826 ret = scan_for_mod(&ops[j], j + 1, i, 1);
2b43685d 3827 need_to_save_current = (ret >= 0);
e56ab892 3828 }
2b43685d 3829 if (need_op_saving || need_to_save_current) {
e56ab892 3830 // mark this push as one that needs operand saving
3831 ops[j].flags &= ~OPF_RMD;
5f70a34f 3832 if (ops[j].p_argnum == 0) {
3833 ops[j].p_argnum = arg + 1;
23fd0b11 3834 save_args |= 1 << arg;
de50b98b 3835 }
fdd5548a 3836 else if (ops[j].p_argnum < arg + 1) {
3837 // XXX: might kill valid var..
3838 //*save_arg_vars &= ~(1 << (ops[j].p_argnum - 1));
3839 ops[j].p_argnum = arg + 1;
3840 save_args |= 1 << arg;
3841 }
3a5101d7 3842
3843 if (save_args_seen & (1 << (ops[j].p_argnum - 1))) {
3844 save_args_seen = 0;
3845 arg_grp_current++;
3846 if (arg_grp_current >= MAX_ARG_GRP)
3847 ferr(&ops[j], "out of arg groups (arg%d), f %s\n",
3848 ops[j].p_argnum, pp->name);
3849 }
e56ab892 3850 }
5f70a34f 3851 else if (ops[j].p_argnum == 0)
e56ab892 3852 ops[j].flags |= OPF_RMD;
3853
a652aa9f 3854 // some PUSHes are reused by different calls on other branches,
de50b98b 3855 // but that can't happen if we didn't branch, so they
3856 // can be removed from future searches (handles nested calls)
a652aa9f 3857 if (!may_reuse)
9af2d373 3858 ops[j].flags |= OPF_FARGNR;
de50b98b 3859
9af2d373 3860 ops[j].flags |= OPF_FARG;
da87ae38 3861 ops[j].flags &= ~OPF_RSAVE;
3862
23fd0b11 3863 // check for __VALIST
5f70a34f 3864 if (!pp->is_unresolved && pp->arg[arg].type.is_va_list) {
23fd0b11 3865 k = -1;
92d715b6 3866 ret = resolve_origin(j, &ops[j].operand[0],
3867 magic + 1, &k, NULL);
5f70a34f 3868 if (ret == 1 && k >= 0)
23fd0b11 3869 {
5f70a34f 3870 if (ops[k].op == OP_LEA) {
3871 snprintf(buf, sizeof(buf), "arg_%X",
3872 g_func_pp->argc_stack * 4);
3873 if (!g_func_pp->is_vararg
3874 || strstr(ops[k].operand[1].name, buf))
3875 {
5e49b270 3876 ops[k].flags |= OPF_RMD | OPF_DONE;
5f70a34f 3877 ops[j].flags |= OPF_RMD | OPF_VAPUSH;
3878 save_args &= ~(1 << arg);
3879 reg = -1;
3880 }
3881 else
3882 ferr(&ops[j], "lea va_list used, but no vararg?\n");
3883 }
3884 // check for va_list from g_func_pp arg too
3885 else if (ops[k].op == OP_MOV
3886 && is_stack_access(&ops[k], &ops[k].operand[1]))
3887 {
3888 ret = stack_frame_access(&ops[k], &ops[k].operand[1],
3889 buf, sizeof(buf), ops[k].operand[1].name, "", 1, 0);
3890 if (ret >= 0) {
5e49b270 3891 ops[k].flags |= OPF_RMD | OPF_DONE;
5f70a34f 3892 ops[j].flags |= OPF_RMD;
3893 ops[j].p_argpass = ret + 1;
3894 save_args &= ~(1 << arg);
3895 reg = -1;
3896 }
3897 }
23fd0b11 3898 }
3899 }
3900
3901 *save_arg_vars |= save_args;
3902
3903 // tracking reg usage
5f70a34f 3904 if (reg >= 0)
3905 *regmask |= 1 << reg;
23fd0b11 3906
89ff3147 3907 arg++;
3908 if (!pp->is_unresolved) {
3909 // next arg
3910 for (; arg < pp->argc; arg++)
3911 if (pp->arg[arg].reg == NULL)
3912 break;
3913 }
a3684be1 3914 magic = (magic & 0xffffff) | (arg << 24);
e56ab892 3915 }
3a5101d7 3916
3917 if (ops[j].p_arggrp > arg_grp_current) {
3918 save_args_seen = 0;
3919 arg_grp_current = ops[j].p_arggrp;
3920 }
3921 if (ops[j].p_argnum > 0)
3922 save_args_seen |= 1 << (ops[j].p_argnum - 1);
e56ab892 3923 }
3924
3925 if (arg < pp->argc) {
3926 ferr(po, "arg collect failed for '%s': %d/%d\n",
3927 pp->name, arg, pp->argc);
89ff3147 3928 return -1;
e56ab892 3929 }
89ff3147 3930
3a5101d7 3931 if (arg_grp_current > *arg_grp)
3932 *arg_grp = arg_grp_current;
3933
89ff3147 3934 return arg;
3935}
3936
3937static int collect_call_args(struct parsed_op *po, int i,
3938 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
3939 int magic)
3940{
3a5101d7 3941 // arg group is for cases when pushes for
3942 // multiple funcs are going on
3943 struct parsed_op *po_tmp;
3944 int save_arg_vars_current = 0;
3945 int arg_grp = 0;
89ff3147 3946 int ret;
3947 int a;
3948
3a5101d7 3949 ret = collect_call_args_r(po, i, pp, regmask,
3950 &save_arg_vars_current, &arg_grp, 0, magic, 0, 0);
89ff3147 3951 if (ret < 0)
3952 return ret;
3953
3a5101d7 3954 if (arg_grp != 0) {
3955 // propagate arg_grp
3956 for (a = 0; a < pp->argc; a++) {
3957 if (pp->arg[a].reg != NULL)
3958 continue;
3959
3960 po_tmp = pp->arg[a].datap;
3961 while (po_tmp != NULL) {
3962 po_tmp->p_arggrp = arg_grp;
3963 if (po_tmp->p_argnext > 0)
3964 po_tmp = &ops[po_tmp->p_argnext];
3965 else
3966 po_tmp = NULL;
3967 }
3968 }
3969 }
3970 save_arg_vars[arg_grp] |= save_arg_vars_current;
3971
89ff3147 3972 if (pp->is_unresolved) {
3973 pp->argc += ret;
3974 pp->argc_stack += ret;
3975 for (a = 0; a < pp->argc; a++)
3976 if (pp->arg[a].type.name == NULL)
3977 pp->arg[a].type.name = strdup("int");
3978 }
3979
e56ab892 3980 return ret;
3981}
3982
89ff3147 3983static void pp_insert_reg_arg(struct parsed_proto *pp, const char *reg)
3984{
3985 int i;
3986
3987 for (i = 0; i < pp->argc; i++)
3988 if (pp->arg[i].reg == NULL)
3989 break;
3990
3991 if (pp->argc_stack)
3992 memmove(&pp->arg[i + 1], &pp->arg[i],
3993 sizeof(pp->arg[0]) * pp->argc_stack);
3994 memset(&pp->arg[i], 0, sizeof(pp->arg[i]));
3995 pp->arg[i].reg = strdup(reg);
3996 pp->arg[i].type.name = strdup("int");
3997 pp->argc++;
3998 pp->argc_reg++;
3999}
4000
04f8a628 4001static void output_std_flags(FILE *fout, struct parsed_op *po,
4002 int *pfomask, const char *dst_opr_text)
4003{
4004 if (*pfomask & (1 << PFO_Z)) {
4005 fprintf(fout, "\n cond_z = (%s%s == 0);",
4006 lmod_cast_u(po, po->operand[0].lmod), dst_opr_text);
4007 *pfomask &= ~(1 << PFO_Z);
4008 }
4009 if (*pfomask & (1 << PFO_S)) {
4010 fprintf(fout, "\n cond_s = (%s%s < 0);",
4011 lmod_cast_s(po, po->operand[0].lmod), dst_opr_text);
4012 *pfomask &= ~(1 << PFO_S);
4013 }
4014}
4015
c0de9015 4016enum {
4017 OPP_FORCE_NORETURN = (1 << 0),
4018 OPP_SIMPLE_ARGS = (1 << 1),
4019 OPP_ALIGN = (1 << 2),
4020};
4021
c0050df6 4022static void output_pp_attrs(FILE *fout, const struct parsed_proto *pp,
c0de9015 4023 int flags)
c0050df6 4024{
c0de9015 4025 const char *cconv = "";
4026
c0050df6 4027 if (pp->is_fastcall)
c0de9015 4028 cconv = "__fastcall ";
c0050df6 4029 else if (pp->is_stdcall && pp->argc_reg == 0)
c0de9015 4030 cconv = "__stdcall ";
4031
4032 fprintf(fout, (flags & OPP_ALIGN) ? "%-16s" : "%s", cconv);
4033
4034 if (pp->is_noreturn || (flags & OPP_FORCE_NORETURN))
c0050df6 4035 fprintf(fout, "noreturn ");
4036}
4037
c0de9015 4038static void output_pp(FILE *fout, const struct parsed_proto *pp,
4039 int flags)
4040{
4041 int i;
4042
4043 fprintf(fout, (flags & OPP_ALIGN) ? "%-5s" : "%s ",
4044 pp->ret_type.name);
4045 if (pp->is_fptr)
4046 fprintf(fout, "(");
4047 output_pp_attrs(fout, pp, flags);
4048 if (pp->is_fptr)
4049 fprintf(fout, "*");
4050 fprintf(fout, "%s", pp->name);
4051 if (pp->is_fptr)
4052 fprintf(fout, ")");
4053
4054 fprintf(fout, "(");
4055 for (i = 0; i < pp->argc; i++) {
4056 if (i > 0)
4057 fprintf(fout, ", ");
4058 if (pp->arg[i].fptr != NULL && !(flags & OPP_SIMPLE_ARGS)) {
4059 // func pointer
4060 output_pp(fout, pp->arg[i].fptr, 0);
4061 }
4062 else if (pp->arg[i].type.is_retreg) {
4063 fprintf(fout, "u32 *r_%s", pp->arg[i].reg);
4064 }
4065 else {
4066 fprintf(fout, "%s", pp->arg[i].type.name);
4067 if (!pp->is_fptr)
4068 fprintf(fout, " a%d", i + 1);
4069 }
4070 }
4071 if (pp->is_vararg) {
4072 if (i > 0)
4073 fprintf(fout, ", ");
4074 fprintf(fout, "...");
4075 }
4076 fprintf(fout, ")");
4077}
4078
9af2d373 4079static int get_pp_arg_regmask(const struct parsed_proto *pp)
4080{
4081 int regmask = 0;
4082 int i, reg;
4083
4084 for (i = 0; i < pp->argc; i++) {
4085 if (pp->arg[i].reg != NULL) {
4086 reg = char_array_i(regs_r32,
4087 ARRAY_SIZE(regs_r32), pp->arg[i].reg);
4088 if (reg < 0)
4089 ferr(ops, "arg '%s' of func '%s' is not a reg?\n",
4090 pp->arg[i].reg, pp->name);
4091 regmask |= 1 << reg;
4092 }
4093 }
4094
4095 return regmask;
4096}
4097
3a5101d7 4098static char *saved_arg_name(char *buf, size_t buf_size, int grp, int num)
4099{
4100 char buf1[16];
4101
4102 buf1[0] = 0;
4103 if (grp > 0)
4104 snprintf(buf1, sizeof(buf1), "%d", grp);
4105 snprintf(buf, buf_size, "s%s_a%d", buf1, num);
4106
4107 return buf;
4108}
4109
9af2d373 4110static void gen_x_cleanup(int opcnt);
4111
91977a1c 4112static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
4113{
69a3cdfc 4114 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
850c9265 4115 struct parsed_opr *last_arith_dst = NULL;
3ebea2cf 4116 char buf1[256], buf2[256], buf3[256], cast[64];
ddaf8bd7 4117 struct parsed_proto *pp, *pp_tmp;
4c45fa73 4118 struct parsed_data *pd;
1f84f6b3 4119 unsigned int uval;
3a5101d7 4120 int save_arg_vars[MAX_ARG_GRP] = { 0, };
cb090db0 4121 int cond_vars = 0;
108e9fe3 4122 int need_tmp_var = 0;
2fe80fdb 4123 int need_tmp64 = 0;
91977a1c 4124 int had_decl = 0;
3ebea2cf 4125 int label_pending = 0;
25a330eb 4126 int regmask_save = 0; // regs saved/restored in this func
4127 int regmask_arg = 0; // regs carrying function args (fastcall, etc)
4128 int regmask_now; // temp
4129 int regmask_init = 0; // regs that need zero initialization
4130 int regmask_pp = 0; // regs used in complex push-pop graph
4131 int regmask = 0; // used regs
940e8e66 4132 int pfomask = 0;
64c59faf 4133 int found = 0;
d4e3b5db 4134 int depth = 0;
91977a1c 4135 int no_output;
4c45fa73 4136 int i, j, l;
91977a1c 4137 int arg;
91977a1c 4138 int reg;
4139 int ret;
4140
1bafb621 4141 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
a2c1d768 4142 g_stack_frame_used = 0;
91977a1c 4143
36595fd2 4144 g_func_pp = proto_parse(fhdr, funcn, 0);
bd96f656 4145 if (g_func_pp == NULL)
91977a1c 4146 ferr(ops, "proto_parse failed for '%s'\n", funcn);
4147
9af2d373 4148 regmask_arg = get_pp_arg_regmask(g_func_pp);
91977a1c 4149
4150 // pass1:
87bf6cec 4151 // - resolve all branches
66bdb2b0 4152 // - parse calls with labels
4153 resolve_branches_parse_calls(opcnt);
840257f6 4154
66bdb2b0 4155 // pass2:
4156 // - handle ebp/esp frame, remove ops related to it
4157 scan_prologue_epilogue(opcnt);
87bf6cec 4158
4159 // pass3:
a2c1d768 4160 // - remove dead labels
1bafb621 4161 for (i = 0; i < opcnt; i++)
4162 {
d7857c3a 4163 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
4164 free(g_labels[i]);
4165 g_labels[i] = NULL;
4166 }
66bdb2b0 4167 }
a2c1d768 4168
66bdb2b0 4169 // pass4:
4170 // - process trivial calls
4171 for (i = 0; i < opcnt; i++)
4172 {
69a3cdfc 4173 po = &ops[i];
5e49b270 4174 if (po->flags & (OPF_RMD|OPF_DONE))
91977a1c 4175 continue;
850c9265 4176
d4e3b5db 4177 if (po->op == OP_CALL)
26677139 4178 {
4179 pp = process_call_early(i, opcnt, &j);
4180 if (pp != NULL) {
4181 if (!(po->flags & OPF_ATAIL))
4182 // since we know the args, try to collect them
4183 if (collect_call_args_early(po, i, pp, &regmask) != 0)
4184 pp = NULL;
4185 }
4186
4187 if (pp != NULL) {
4188 if (j >= 0) {
4189 // commit esp adjust
5e49b270 4190 if (ops[j].op != OP_POP)
4191 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
bfacdc83 4192 else {
4193 for (l = 0; l < pp->argc_stack; l++)
4194 ops[j + l].flags |= OPF_DONE | OPF_RMD;
4195 }
26677139 4196 }
4197
4198 if (strstr(pp->ret_type.name, "int64"))
4199 need_tmp64 = 1;
4200
4201 po->flags |= OPF_DONE;
4202 }
4203 }
4204 }
4205
66bdb2b0 4206 // pass5:
26677139 4207 // - process calls
ee2361b9 4208 // - handle push <const>/pop pairs
26677139 4209 for (i = 0; i < opcnt; i++)
4210 {
4211 po = &ops[i];
5e49b270 4212 if (po->flags & (OPF_RMD|OPF_DONE))
26677139 4213 continue;
4214
4215 if (po->op == OP_CALL && !(po->flags & OPF_DONE))
69a3cdfc 4216 {
9af2d373 4217 pp = process_call(i, opcnt);
91977a1c 4218
2fe80fdb 4219 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
89ff3147 4220 // since we know the args, collect them
3a5101d7 4221 collect_call_args(po, i, pp, &regmask, save_arg_vars,
89ff3147 4222 i + opcnt * 2);
4223 }
2b43685d 4224
4225 if (strstr(pp->ret_type.name, "int64"))
2fe80fdb 4226 need_tmp64 = 1;
91977a1c 4227 }
ee2361b9 4228 else if (po->op == OP_PUSH && !(po->flags & OPF_FARG)
4229 && !(po->flags & OPF_RSAVE) && po->operand[0].type == OPT_CONST)
25a330eb 4230 scan_for_pop_const(i, opcnt, &regmask_pp);
d4e3b5db 4231 }
4232
66bdb2b0 4233 // pass6:
d4e3b5db 4234 // - find POPs for PUSHes, rm both
591721d7 4235 // - scan for STD/CLD, propagate DF
d4e3b5db 4236 // - scan for all used registers
4237 // - find flag set ops for their users
89ff3147 4238 // - do unreselved calls
1bafb621 4239 // - declare indirect functions
26677139 4240 for (i = 0; i < opcnt; i++)
4241 {
d4e3b5db 4242 po = &ops[i];
5e49b270 4243 if (po->flags & (OPF_RMD|OPF_DONE))
d4e3b5db 4244 continue;
4245
7ae48d73 4246 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE)) {
4247 reg = po->operand[0].reg;
4248 if (!(regmask & (1 << reg)))
4249 // not a reg save after all, rerun scan_for_pop
4250 po->flags &= ~OPF_RSAVE;
4251 else
4252 regmask_save |= 1 << reg;
4253 }
4254
9af2d373 4255 if (po->op == OP_PUSH && !(po->flags & OPF_FARG)
1f84f6b3 4256 && !(po->flags & OPF_RSAVE) && !g_func_pp->is_userstack)
d4e3b5db 4257 {
5c024ef7 4258 if (po->operand[0].type == OPT_REG)
4259 {
4260 reg = po->operand[0].reg;
4261 if (reg < 0)
4262 ferr(po, "reg not set for push?\n");
4263
4264 depth = 0;
4265 ret = scan_for_pop(i + 1, opcnt,
4266 po->operand[0].name, i + opcnt * 3, 0, &depth, 0);
4267 if (ret == 1) {
4268 if (depth > 1)
4269 ferr(po, "too much depth: %d\n", depth);
4270
4271 po->flags |= OPF_RMD;
4272 scan_for_pop(i + 1, opcnt, po->operand[0].name,
4273 i + opcnt * 4, 0, &depth, 1);
4274 continue;
4275 }
4276 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
4277 if (ret == 0) {
4278 arg = OPF_RMD;
4279 if (regmask & (1 << reg)) {
4280 if (regmask_save & (1 << reg))
4281 ferr(po, "%s already saved?\n", po->operand[0].name);
4282 arg = OPF_RSAVE;
4283 }
4284 po->flags |= arg;
4285 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg);
4286 continue;
4287 }
d4e3b5db 4288 }
d4e3b5db 4289 }
4290
591721d7 4291 if (po->op == OP_STD) {
5e49b270 4292 po->flags |= OPF_DF | OPF_RMD | OPF_DONE;
591721d7 4293 scan_propagate_df(i + 1, opcnt);
4294 }
4295
a3684be1 4296 regmask_now = po->regmask_src | po->regmask_dst;
4297 if (regmask_now & (1 << xBP)) {
4298 if (g_bp_frame && !(po->flags & OPF_EBP_S)) {
4299 if (po->regmask_dst & (1 << xBP))
4300 // compiler decided to drop bp frame and use ebp as scratch
037f4971 4301 scan_fwd_set_flags(i + 1, opcnt, i + opcnt * 5, OPF_EBP_S);
a3684be1 4302 else
4303 regmask_now &= ~(1 << xBP);
4304 }
4305 }
4306
4307 regmask |= regmask_now;
d4e3b5db 4308
4309 if (po->flags & OPF_CC)
4310 {
2b43685d 4311 int setters[16], cnt = 0, branched = 0;
4312
04f8a628 4313 ret = scan_for_flag_set(i, i + opcnt * 6,
4314 &branched, setters, &cnt);
2b43685d 4315 if (ret < 0 || cnt <= 0)
4316 ferr(po, "unable to trace flag setter(s)\n");
4317 if (cnt > ARRAY_SIZE(setters))
4318 ferr(po, "too many flag setters\n");
d4e3b5db 4319
2b43685d 4320 for (j = 0; j < cnt; j++)
4321 {
4322 tmp_op = &ops[setters[j]]; // flag setter
4323 pfomask = 0;
4324
4325 // to get nicer code, we try to delay test and cmp;
4326 // if we can't because of operand modification, or if we
591721d7 4327 // have arith op, or branch, make it calculate flags explicitly
4328 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP)
4329 {
89ff3147 4330 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i, 0) >= 0)
092f64e1 4331 pfomask = 1 << po->pfo;
2b43685d 4332 }
4741fdfe 4333 else if (tmp_op->op == OP_CMPS || tmp_op->op == OP_SCAS) {
092f64e1 4334 pfomask = 1 << po->pfo;
591721d7 4335 }
2b43685d 4336 else {
04f8a628 4337 // see if we'll be able to handle based on op result
4338 if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR
092f64e1 4339 && po->pfo != PFO_Z && po->pfo != PFO_S
4340 && po->pfo != PFO_P)
04f8a628 4341 || branched
2b43685d 4342 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
092f64e1 4343 {
4344 pfomask = 1 << po->pfo;
4345 }
2fe80fdb 4346
c8dbc5be 4347 if (tmp_op->op == OP_ADD && po->pfo == PFO_C) {
4348 propagate_lmod(tmp_op, &tmp_op->operand[0],
4349 &tmp_op->operand[1]);
4350 if (tmp_op->operand[0].lmod == OPLM_DWORD)
4351 need_tmp64 = 1;
4352 }
2b43685d 4353 }
4354 if (pfomask) {
4355 tmp_op->pfomask |= pfomask;
cb090db0 4356 cond_vars |= pfomask;
2b43685d 4357 }
04f8a628 4358 // note: may overwrite, currently not a problem
4359 po->datap = tmp_op;
d4e3b5db 4360 }
4361
cb090db0 4362 if (po->op == OP_RCL || po->op == OP_RCR
4363 || po->op == OP_ADC || po->op == OP_SBB)
4364 cond_vars |= 1 << PFO_C;
d4e3b5db 4365 }
092f64e1 4366
4367 if (po->op == OP_CMPS || po->op == OP_SCAS) {
cb090db0 4368 cond_vars |= 1 << PFO_Z;
591721d7 4369 }
87bf6cec 4370 else if (po->op == OP_MUL
4371 || (po->op == OP_IMUL && po->operand_cnt == 1))
4372 {
c8dbc5be 4373 if (po->operand[0].lmod == OPLM_DWORD)
4374 need_tmp64 = 1;
87bf6cec 4375 }
89ff3147 4376 else if (po->op == OP_CALL) {
26677139 4377 // note: resolved non-reg calls are OPF_DONE already
092f64e1 4378 pp = po->pp;
89ff3147 4379 if (pp == NULL)
4380 ferr(po, "NULL pp\n");
4381
4382 if (pp->is_unresolved) {
ddaf8bd7 4383 int regmask_stack = 0;
3a5101d7 4384 collect_call_args(po, i, pp, &regmask, save_arg_vars,
89ff3147 4385 i + opcnt * 2);
4386
b74c31e3 4387 // this is pretty rough guess:
4388 // see ecx and edx were pushed (and not their saved versions)
4389 for (arg = 0; arg < pp->argc; arg++) {
4390 if (pp->arg[arg].reg != NULL)
4391 continue;
4392
4393 tmp_op = pp->arg[arg].datap;
4394 if (tmp_op == NULL)
4395 ferr(po, "parsed_op missing for arg%d\n", arg);
5f70a34f 4396 if (tmp_op->p_argnum == 0 && tmp_op->operand[0].type == OPT_REG)
b74c31e3 4397 regmask_stack |= 1 << tmp_op->operand[0].reg;
4398 }
4399
ddaf8bd7 4400 if (!((regmask_stack & (1 << xCX))
4401 && (regmask_stack & (1 << xDX))))
89ff3147 4402 {
4403 if (pp->argc_stack != 0
c0050df6 4404 || ((regmask | regmask_arg) & ((1 << xCX)|(1 << xDX))))
89ff3147 4405 {
4406 pp_insert_reg_arg(pp, "ecx");
c0050df6 4407 pp->is_fastcall = 1;
ddaf8bd7 4408 regmask_init |= 1 << xCX;
89ff3147 4409 regmask |= 1 << xCX;
4410 }
4411 if (pp->argc_stack != 0
4412 || ((regmask | regmask_arg) & (1 << xDX)))
4413 {
4414 pp_insert_reg_arg(pp, "edx");
ddaf8bd7 4415 regmask_init |= 1 << xDX;
89ff3147 4416 regmask |= 1 << xDX;
4417 }
4418 }
c0050df6 4419
4420 // note: __cdecl doesn't fall into is_unresolved category
4421 if (pp->argc_stack > 0)
4422 pp->is_stdcall = 1;
ddaf8bd7 4423 }
4424
4425 for (arg = 0; arg < pp->argc; arg++) {
4426 if (pp->arg[arg].reg != NULL) {
4427 reg = char_array_i(regs_r32,
4428 ARRAY_SIZE(regs_r32), pp->arg[arg].reg);
4429 if (reg < 0)
4430 ferr(ops, "arg '%s' is not a reg?\n", pp->arg[arg].reg);
4431 if (!(regmask & (1 << reg))) {
4432 regmask_init |= 1 << reg;
4433 regmask |= 1 << reg;
4434 }
4435 }
89ff3147 4436 }
1bafb621 4437 }
27ebfaed 4438 else if (po->op == OP_MOV && po->operand[0].pp != NULL
4439 && po->operand[1].pp != NULL)
4440 {
4441 // <var> = offset <something>
4442 if ((po->operand[1].pp->is_func || po->operand[1].pp->is_fptr)
4443 && !IS_START(po->operand[1].name, "off_"))
4444 {
4445 if (!po->operand[0].pp->is_fptr)
4446 ferr(po, "%s not declared as fptr when it should be\n",
4447 po->operand[0].name);
4448 if (pp_cmp_func(po->operand[0].pp, po->operand[1].pp)) {
4449 pp_print(buf1, sizeof(buf1), po->operand[0].pp);
4450 pp_print(buf2, sizeof(buf2), po->operand[1].pp);
4451 fnote(po, "var: %s\n", buf1);
4452 fnote(po, "func: %s\n", buf2);
4453 ferr(po, "^ mismatch\n");
4454 }
4455 }
4456 }
75ad0378 4457 else if (po->op == OP_RET && !IS(g_func_pp->ret_type.name, "void"))
4458 regmask |= 1 << xAX;
cb090db0 4459 else if (po->op == OP_DIV || po->op == OP_IDIV) {
4460 // 32bit division is common, look for it
4461 if (po->op == OP_DIV)
4462 ret = scan_for_reg_clear(i, xDX);
4463 else
4464 ret = scan_for_cdq_edx(i);
4465 if (ret >= 0)
4466 po->flags |= OPF_32BIT;
4467 else
4468 need_tmp64 = 1;
4469 }
037f4971 4470 else if (po->op == OP_CLD)
5e49b270 4471 po->flags |= OPF_RMD | OPF_DONE;
cb090db0 4472
4473 if (po->op == OP_RCL || po->op == OP_RCR || po->op == OP_XCHG) {
4474 need_tmp_var = 1;
4475 }
91977a1c 4476 }
4477
66bdb2b0 4478 // pass7:
da87ae38 4479 // - confirm regmask_save, it might have been reduced
4480 if (regmask_save != 0)
4481 {
4482 regmask_save = 0;
4483 for (i = 0; i < opcnt; i++) {
4484 po = &ops[i];
4485 if (po->flags & OPF_RMD)
4486 continue;
4487
4488 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE))
4489 regmask_save |= 1 << po->operand[0].reg;
4490 }
4491 }
4492
60fe410c 4493 // output starts here
4494
4495 // define userstack size
4496 if (g_func_pp->is_userstack) {
4497 fprintf(fout, "#ifndef US_SZ_%s\n", g_func_pp->name);
4498 fprintf(fout, "#define US_SZ_%s USERSTACK_SIZE\n", g_func_pp->name);
4499 fprintf(fout, "#endif\n");
4500 }
4501
4502 // the function itself
c0de9015 4503 ferr_assert(ops, !g_func_pp->is_fptr);
4504 output_pp(fout, g_func_pp,
4505 (g_ida_func_attr & IDAFA_NORETURN) ? OPP_FORCE_NORETURN : 0);
4506 fprintf(fout, "\n{\n");
60fe410c 4507
4508 // declare indirect functions
4509 for (i = 0; i < opcnt; i++) {
4510 po = &ops[i];
4511 if (po->flags & OPF_RMD)
4512 continue;
4513
4514 if (po->op == OP_CALL) {
4515 pp = po->pp;
4516 if (pp == NULL)
4517 ferr(po, "NULL pp\n");
4518
4519 if (pp->is_fptr && !(pp->name[0] != 0 && pp->is_arg)) {
4520 if (pp->name[0] != 0) {
4521 memmove(pp->name + 2, pp->name, strlen(pp->name) + 1);
4522 memcpy(pp->name, "i_", 2);
4523
4524 // might be declared already
4525 found = 0;
4526 for (j = 0; j < i; j++) {
4527 if (ops[j].op == OP_CALL && (pp_tmp = ops[j].pp)) {
4528 if (pp_tmp->is_fptr && IS(pp->name, pp_tmp->name)) {
4529 found = 1;
4530 break;
4531 }
4532 }
4533 }
4534 if (found)
4535 continue;
4536 }
4537 else
4538 snprintf(pp->name, sizeof(pp->name), "icall%d", i);
4539
c0de9015 4540 fprintf(fout, " ");
4541 output_pp(fout, pp, OPP_SIMPLE_ARGS);
4542 fprintf(fout, ";\n");
60fe410c 4543 }
4544 }
4545 }
da87ae38 4546
4c45fa73 4547 // output LUTs/jumptables
4548 for (i = 0; i < g_func_pd_cnt; i++) {
4549 pd = &g_func_pd[i];
4550 fprintf(fout, " static const ");
4551 if (pd->type == OPT_OFFSET) {
4552 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
4553
4554 for (j = 0; j < pd->count; j++) {
4555 if (j > 0)
4556 fprintf(fout, ", ");
4557 fprintf(fout, "&&%s", pd->d[j].u.label);
4558 }
4559 }
4560 else {
4561 fprintf(fout, "%s %s[] =\n { ",
4562 lmod_type_u(ops, pd->lmod), pd->label);
4563
4564 for (j = 0; j < pd->count; j++) {
4565 if (j > 0)
4566 fprintf(fout, ", ");
4567 fprintf(fout, "%u", pd->d[j].u.val);
4568 }
4569 }
4570 fprintf(fout, " };\n");
1f84f6b3 4571 had_decl = 1;
4c45fa73 4572 }
4573
4f12f671 4574 // declare stack frame, va_arg
1f84f6b3 4575 if (g_stack_fsz) {
850c9265 4576 fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
1bafb621 4577 (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz);
1f84f6b3 4578 had_decl = 1;
4579 }
4580
4581 if (g_func_pp->is_userstack) {
60fe410c 4582 fprintf(fout, " u32 fake_sf[US_SZ_%s / 4];\n", g_func_pp->name);
4583 fprintf(fout, " u32 *esp = &fake_sf[sizeof(fake_sf) / 4];\n");
1f84f6b3 4584 had_decl = 1;
4585 }
850c9265 4586
1f84f6b3 4587 if (g_func_pp->is_vararg) {
4f12f671 4588 fprintf(fout, " va_list ap;\n");
1f84f6b3 4589 had_decl = 1;
4590 }
4f12f671 4591
940e8e66 4592 // declare arg-registers
bd96f656 4593 for (i = 0; i < g_func_pp->argc; i++) {
4594 if (g_func_pp->arg[i].reg != NULL) {
91977a1c 4595 reg = char_array_i(regs_r32,
bd96f656 4596 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
75ad0378 4597 if (regmask & (1 << reg)) {
1f84f6b3 4598 if (g_func_pp->arg[i].type.is_retreg)
4599 fprintf(fout, " u32 %s = *r_%s;\n",
4600 g_func_pp->arg[i].reg, g_func_pp->arg[i].reg);
4601 else
4602 fprintf(fout, " u32 %s = (u32)a%d;\n",
4603 g_func_pp->arg[i].reg, i + 1);
75ad0378 4604 }
1f84f6b3 4605 else {
4606 if (g_func_pp->arg[i].type.is_retreg)
4607 ferr(ops, "retreg '%s' is unused?\n",
4608 g_func_pp->arg[i].reg);
75ad0378 4609 fprintf(fout, " // %s = a%d; // unused\n",
4610 g_func_pp->arg[i].reg, i + 1);
1f84f6b3 4611 }
91977a1c 4612 had_decl = 1;
4613 }
4614 }
4615
25a330eb 4616 // declare normal registers
75ad0378 4617 regmask_now = regmask & ~regmask_arg;
4618 regmask_now &= ~(1 << xSP);
90307a99 4619 if (regmask_now & 0x00ff) {
91977a1c 4620 for (reg = 0; reg < 8; reg++) {
75ad0378 4621 if (regmask_now & (1 << reg)) {
ddaf8bd7 4622 fprintf(fout, " u32 %s", regs_r32[reg]);
4623 if (regmask_init & (1 << reg))
4624 fprintf(fout, " = 0");
4625 fprintf(fout, ";\n");
91977a1c 4626 had_decl = 1;
4627 }
4628 }
4629 }
90307a99 4630 if (regmask_now & 0xff00) {
4631 for (reg = 8; reg < 16; reg++) {
4632 if (regmask_now & (1 << reg)) {
4633 fprintf(fout, " mmxr %s", regs_r32[reg]);
4634 if (regmask_init & (1 << reg))
4635 fprintf(fout, " = { 0, }");
4636 fprintf(fout, ";\n");
4637 had_decl = 1;
4638 }
4639 }
4640 }
91977a1c 4641
d4e3b5db 4642 if (regmask_save) {
4643 for (reg = 0; reg < 8; reg++) {
4644 if (regmask_save & (1 << reg)) {
4645 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
4646 had_decl = 1;
4647 }
4648 }
4649 }
4650
3a5101d7 4651 for (i = 0; i < ARRAY_SIZE(save_arg_vars); i++) {
4652 if (save_arg_vars[i] == 0)
4653 continue;
69a3cdfc 4654 for (reg = 0; reg < 32; reg++) {
3a5101d7 4655 if (save_arg_vars[i] & (1 << reg)) {
4656 fprintf(fout, " u32 %s;\n",
4657 saved_arg_name(buf1, sizeof(buf1), i, reg + 1));
69a3cdfc 4658 had_decl = 1;
4659 }
4660 }
4661 }
4662
25a330eb 4663 // declare push-pop temporaries
4664 if (regmask_pp) {
4665 for (reg = 0; reg < 8; reg++) {
4666 if (regmask_pp & (1 << reg)) {
4667 fprintf(fout, " u32 pp_%s;\n", regs_r32[reg]);
4668 had_decl = 1;
4669 }
4670 }
4671 }
4672
cb090db0 4673 if (cond_vars) {
69a3cdfc 4674 for (i = 0; i < 8; i++) {
cb090db0 4675 if (cond_vars & (1 << i)) {
69a3cdfc 4676 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
4677 had_decl = 1;
4678 }
4679 }
4680 }
4681
108e9fe3 4682 if (need_tmp_var) {
4683 fprintf(fout, " u32 tmp;\n");
4684 had_decl = 1;
4685 }
4686
2fe80fdb 4687 if (need_tmp64) {
4688 fprintf(fout, " u64 tmp64;\n");
87bf6cec 4689 had_decl = 1;
4690 }
4691
91977a1c 4692 if (had_decl)
4693 fprintf(fout, "\n");
4694
bd96f656 4695 if (g_func_pp->is_vararg) {
4696 if (g_func_pp->argc_stack == 0)
4f12f671 4697 ferr(ops, "vararg func without stack args?\n");
bd96f656 4698 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
4f12f671 4699 }
4700
91977a1c 4701 // output ops
69a3cdfc 4702 for (i = 0; i < opcnt; i++)
4703 {
d7857c3a 4704 if (g_labels[i] != NULL) {
91977a1c 4705 fprintf(fout, "\n%s:\n", g_labels[i]);
3ebea2cf 4706 label_pending = 1;
2b43685d 4707
4708 delayed_flag_op = NULL;
4709 last_arith_dst = NULL;
3ebea2cf 4710 }
91977a1c 4711
69a3cdfc 4712 po = &ops[i];
4713 if (po->flags & OPF_RMD)
91977a1c 4714 continue;
4715
4716 no_output = 0;
4717
91977a1c 4718 #define assert_operand_cnt(n_) \
850c9265 4719 if (po->operand_cnt != n_) \
4720 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
4721
69a3cdfc 4722 // conditional/flag using op?
4723 if (po->flags & OPF_CC)
850c9265 4724 {
940e8e66 4725 int is_delayed = 0;
69a3cdfc 4726
04f8a628 4727 tmp_op = po->datap;
850c9265 4728
69a3cdfc 4729 // we go through all this trouble to avoid using parsed_flag_op,
4730 // which makes generated code much nicer
4731 if (delayed_flag_op != NULL)
850c9265 4732 {
092f64e1 4733 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op,
4734 po->pfo, po->pfo_inv);
940e8e66 4735 is_delayed = 1;
91977a1c 4736 }
850c9265 4737 else if (last_arith_dst != NULL
092f64e1 4738 && (po->pfo == PFO_Z || po->pfo == PFO_S || po->pfo == PFO_P
04f8a628 4739 || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR))
4740 ))
850c9265 4741 {
3ebea2cf 4742 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
092f64e1 4743 out_test_for_cc(buf1, sizeof(buf1), po, po->pfo, po->pfo_inv,
850c9265 4744 last_arith_dst->lmod, buf3);
940e8e66 4745 is_delayed = 1;
850c9265 4746 }
04f8a628 4747 else if (tmp_op != NULL) {
7ba45c34 4748 // use preprocessed flag calc results
092f64e1 4749 if (!(tmp_op->pfomask & (1 << po->pfo)))
4750 ferr(po, "not prepared for pfo %d\n", po->pfo);
69a3cdfc 4751
092f64e1 4752 // note: pfo_inv was not yet applied
69a3cdfc 4753 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
092f64e1 4754 po->pfo_inv ? "!" : "", parsed_flag_op_names[po->pfo]);
69a3cdfc 4755 }
4756 else {
4757 ferr(po, "all methods of finding comparison failed\n");
4758 }
850c9265 4759
69a3cdfc 4760 if (po->flags & OPF_JMP) {
092f64e1 4761 fprintf(fout, " if %s", buf1);
850c9265 4762 }
cb090db0 4763 else if (po->op == OP_RCL || po->op == OP_RCR
4764 || po->op == OP_ADC || po->op == OP_SBB)
4765 {
940e8e66 4766 if (is_delayed)
4767 fprintf(fout, " cond_%s = %s;\n",
092f64e1 4768 parsed_flag_op_names[po->pfo], buf1);
850c9265 4769 }
5101a5f9 4770 else if (po->flags & OPF_DATA) { // SETcc
850c9265 4771 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
4772 fprintf(fout, " %s = %s;", buf2, buf1);
91977a1c 4773 }
69a3cdfc 4774 else {
4775 ferr(po, "unhandled conditional op\n");
4776 }
91977a1c 4777 }
4778
940e8e66 4779 pfomask = po->pfomask;
4780
4741fdfe 4781 if (po->flags & (OPF_REPZ|OPF_REPNZ)) {
1f84f6b3 4782 struct parsed_opr opr = {0,};
4783 opr.type = OPT_REG;
4784 opr.reg = xCX;
4785 opr.lmod = OPLM_DWORD;
4786 ret = try_resolve_const(i, &opr, opcnt * 7 + i, &uval);
4787
4788 if (ret != 1 || uval == 0) {
4789 // we need initial flags for ecx=0 case..
4790 if (i > 0 && ops[i - 1].op == OP_XOR
4791 && IS(ops[i - 1].operand[0].name,
4792 ops[i - 1].operand[1].name))
4793 {
4794 fprintf(fout, " cond_z = ");
4795 if (pfomask & (1 << PFO_C))
4796 fprintf(fout, "cond_c = ");
4797 fprintf(fout, "0;\n");
4798 }
4799 else if (last_arith_dst != NULL) {
4800 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
4801 out_test_for_cc(buf1, sizeof(buf1), po, PFO_Z, 0,
4802 last_arith_dst->lmod, buf3);
4803 fprintf(fout, " cond_z = %s;\n", buf1);
4804 }
4805 else
4806 ferr(po, "missing initial ZF\n");
4741fdfe 4807 }
4741fdfe 4808 }
4809
850c9265 4810 switch (po->op)
91977a1c 4811 {
4812 case OP_MOV:
4813 assert_operand_cnt(2);
850c9265 4814 propagate_lmod(po, &po->operand[0], &po->operand[1]);
de50b98b 4815 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
c7ed83dd 4816 default_cast_to(buf3, sizeof(buf3), &po->operand[0]);
de50b98b 4817 fprintf(fout, " %s = %s;", buf1,
3ebea2cf 4818 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
c7ed83dd 4819 buf3, 0));
850c9265 4820 break;
4821
4822 case OP_LEA:
4823 assert_operand_cnt(2);
87bf6cec 4824 po->operand[1].lmod = OPLM_DWORD; // always
850c9265 4825 fprintf(fout, " %s = %s;",
4826 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3ebea2cf 4827 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4828 NULL, 1));
850c9265 4829 break;
4830
4831 case OP_MOVZX:
4832 assert_operand_cnt(2);
91977a1c 4833 fprintf(fout, " %s = %s;",
850c9265 4834 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3ebea2cf 4835 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
850c9265 4836 break;
4837
4838 case OP_MOVSX:
4839 assert_operand_cnt(2);
4840 switch (po->operand[1].lmod) {
4841 case OPLM_BYTE:
4842 strcpy(buf3, "(s8)");
4843 break;
4844 case OPLM_WORD:
4845 strcpy(buf3, "(s16)");
4846 break;
4847 default:
4848 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
4849 }
a2c1d768 4850 fprintf(fout, " %s = %s;",
850c9265 4851 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
a2c1d768 4852 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4853 buf3, 0));
850c9265 4854 break;
4855
108e9fe3 4856 case OP_XCHG:
4857 assert_operand_cnt(2);
4858 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4859 fprintf(fout, " tmp = %s;",
4860 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], "", 0));
4861 fprintf(fout, " %s = %s;",
4862 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
c7ed83dd 4863 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4864 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
4865 fprintf(fout, " %s = %stmp;",
4866 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]),
4867 default_cast_to(buf3, sizeof(buf3), &po->operand[1]));
108e9fe3 4868 snprintf(g_comment, sizeof(g_comment), "xchg");
4869 break;
4870
850c9265 4871 case OP_NOT:
4872 assert_operand_cnt(1);
4873 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4874 fprintf(fout, " %s = ~%s;", buf1, buf1);
4875 break;
4876
5101a5f9 4877 case OP_CDQ:
4878 assert_operand_cnt(2);
4879 fprintf(fout, " %s = (s32)%s >> 31;",
4880 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3ebea2cf 4881 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5101a5f9 4882 strcpy(g_comment, "cdq");
4883 break;
4884
092f64e1 4885 case OP_LODS:
4886 assert_operand_cnt(3);
4887 if (po->flags & OPF_REP) {
4888 // hmh..
4889 ferr(po, "TODO\n");
4890 }
4891 else {
05381e4a 4892 fprintf(fout, " eax = %sesi; esi %c= %d;",
092f64e1 4893 lmod_cast_u_ptr(po, po->operand[0].lmod),
4894 (po->flags & OPF_DF) ? '-' : '+',
4895 lmod_bytes(po, po->operand[0].lmod));
4896 strcpy(g_comment, "lods");
4897 }
4898 break;
4899
33c35af6 4900 case OP_STOS:
33c35af6 4901 assert_operand_cnt(3);
4902 if (po->flags & OPF_REP) {
591721d7 4903 fprintf(fout, " for (; ecx != 0; ecx--, edi %c= %d)\n",
4904 (po->flags & OPF_DF) ? '-' : '+',
33c35af6 4905 lmod_bytes(po, po->operand[0].lmod));
d4e3b5db 4906 fprintf(fout, " %sedi = eax;",
4907 lmod_cast_u_ptr(po, po->operand[0].lmod));
33c35af6 4908 strcpy(g_comment, "rep stos");
4909 }
4910 else {
092f64e1 4911 fprintf(fout, " %sedi = eax; edi %c= %d;",
d4e3b5db 4912 lmod_cast_u_ptr(po, po->operand[0].lmod),
591721d7 4913 (po->flags & OPF_DF) ? '-' : '+',
33c35af6 4914 lmod_bytes(po, po->operand[0].lmod));
4915 strcpy(g_comment, "stos");
4916 }
4917 break;
4918
d4e3b5db 4919 case OP_MOVS:
d4e3b5db 4920 assert_operand_cnt(3);
4921 j = lmod_bytes(po, po->operand[0].lmod);
4922 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
591721d7 4923 l = (po->flags & OPF_DF) ? '-' : '+';
d4e3b5db 4924 if (po->flags & OPF_REP) {
4925 fprintf(fout,
591721d7 4926 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n",
4927 l, j, l, j);
d4e3b5db 4928 fprintf(fout,
4929 " %sedi = %sesi;", buf1, buf1);
4930 strcpy(g_comment, "rep movs");
4931 }
4932 else {
092f64e1 4933 fprintf(fout, " %sedi = %sesi; edi %c= %d; esi %c= %d;",
591721d7 4934 buf1, buf1, l, j, l, j);
d4e3b5db 4935 strcpy(g_comment, "movs");
4936 }
4937 break;
4938
7ba45c34 4939 case OP_CMPS:
7ba45c34 4940 // repe ~ repeat while ZF=1
4941 assert_operand_cnt(3);
4942 j = lmod_bytes(po, po->operand[0].lmod);
4943 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
591721d7 4944 l = (po->flags & OPF_DF) ? '-' : '+';
7ba45c34 4945 if (po->flags & OPF_REP) {
4946 fprintf(fout,
1f84f6b3 4947 " for (; ecx != 0; ecx--) {\n");
4741fdfe 4948 if (pfomask & (1 << PFO_C)) {
4949 // ugh..
4950 fprintf(fout,
05381e4a 4951 " cond_c = %sesi < %sedi;\n", buf1, buf1);
4741fdfe 4952 pfomask &= ~(1 << PFO_C);
4953 }
7ba45c34 4954 fprintf(fout,
05381e4a 4955 " cond_z = (%sesi == %sedi); esi %c= %d, edi %c= %d;\n",
1f84f6b3 4956 buf1, buf1, l, j, l, j);
4957 fprintf(fout,
4958 " if (cond_z %s 0) break;\n",
4959 (po->flags & OPF_REPZ) ? "==" : "!=");
7ba45c34 4960 fprintf(fout,
4741fdfe 4961 " }");
7ba45c34 4962 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
4963 (po->flags & OPF_REPZ) ? "e" : "ne");
4964 }
4965 else {
4966 fprintf(fout,
05381e4a 4967 " cond_z = (%sesi == %sedi); esi %c= %d; edi %c= %d;",
591721d7 4968 buf1, buf1, l, j, l, j);
7ba45c34 4969 strcpy(g_comment, "cmps");
4970 }
4971 pfomask &= ~(1 << PFO_Z);
4972 last_arith_dst = NULL;
4973 delayed_flag_op = NULL;
4974 break;
4975
591721d7 4976 case OP_SCAS:
4977 // only does ZF (for now)
4978 // repe ~ repeat while ZF=1
4979 assert_operand_cnt(3);
4980 j = lmod_bytes(po, po->operand[0].lmod);
4981 l = (po->flags & OPF_DF) ? '-' : '+';
4982 if (po->flags & OPF_REP) {
4983 fprintf(fout,
1f84f6b3 4984 " for (; ecx != 0; ecx--) {\n");
591721d7 4985 fprintf(fout,
1f84f6b3 4986 " cond_z = (%seax == %sedi); edi %c= %d;\n",
591721d7 4987 lmod_cast_u(po, po->operand[0].lmod),
1f84f6b3 4988 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
4989 fprintf(fout,
4990 " if (cond_z %s 0) break;\n",
591721d7 4991 (po->flags & OPF_REPZ) ? "==" : "!=");
4992 fprintf(fout,
1f84f6b3 4993 " }");
591721d7 4994 snprintf(g_comment, sizeof(g_comment), "rep%s scas",
4995 (po->flags & OPF_REPZ) ? "e" : "ne");
4996 }
4997 else {
05381e4a 4998 fprintf(fout, " cond_z = (%seax == %sedi); edi %c= %d;",
591721d7 4999 lmod_cast_u(po, po->operand[0].lmod),
5000 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
5001 strcpy(g_comment, "scas");
5002 }
5003 pfomask &= ~(1 << PFO_Z);
5004 last_arith_dst = NULL;
5005 delayed_flag_op = NULL;
5006 break;
5007
850c9265 5008 // arithmetic w/flags
850c9265 5009 case OP_AND:
66bdb2b0 5010 if (po->operand[1].type == OPT_CONST && !po->operand[1].val) {
5011 // deal with complex dst clear
5012 assert_operand_cnt(2);
5013 fprintf(fout, " %s = %s;",
5014 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5015 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
5016 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
5017 output_std_flags(fout, po, &pfomask, buf1);
5018 last_arith_dst = &po->operand[0];
5019 delayed_flag_op = NULL;
5020 break;
5021 }
5022 // fallthrough
850c9265 5023 case OP_OR:
5101a5f9 5024 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5025 // fallthrough
850c9265 5026 dualop_arith:
5027 assert_operand_cnt(2);
850c9265 5028 fprintf(fout, " %s %s= %s;",
5029 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5030 op_to_c(po),
3ebea2cf 5031 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
04f8a628 5032 output_std_flags(fout, po, &pfomask, buf1);
5033 last_arith_dst = &po->operand[0];
5034 delayed_flag_op = NULL;
5035 break;
5036
5037 case OP_SHL:
5038 case OP_SHR:
5039 assert_operand_cnt(2);
5040 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5041 if (pfomask & (1 << PFO_C)) {
5042 if (po->operand[1].type == OPT_CONST) {
5043 l = lmod_bytes(po, po->operand[0].lmod) * 8;
5044 j = po->operand[1].val;
5045 j %= l;
5046 if (j != 0) {
5047 if (po->op == OP_SHL)
5048 j = l - j;
5049 else
5050 j -= 1;
cb090db0 5051 fprintf(fout, " cond_c = (%s >> %d) & 1;\n",
5052 buf1, j);
04f8a628 5053 }
5054 else
5055 ferr(po, "zero shift?\n");
5056 }
5057 else
5058 ferr(po, "TODO\n");
5059 pfomask &= ~(1 << PFO_C);
840257f6 5060 }
04f8a628 5061 fprintf(fout, " %s %s= %s;", buf1, op_to_c(po),
5062 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5063 output_std_flags(fout, po, &pfomask, buf1);
850c9265 5064 last_arith_dst = &po->operand[0];
69a3cdfc 5065 delayed_flag_op = NULL;
850c9265 5066 break;
5067
d4e3b5db 5068 case OP_SAR:
5069 assert_operand_cnt(2);
5070 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5071 fprintf(fout, " %s = %s%s >> %s;", buf1,
5072 lmod_cast_s(po, po->operand[0].lmod), buf1,
3ebea2cf 5073 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
04f8a628 5074 output_std_flags(fout, po, &pfomask, buf1);
d4e3b5db 5075 last_arith_dst = &po->operand[0];
5076 delayed_flag_op = NULL;
5077 break;
5078
3b2f4044 5079 case OP_SHRD:
5080 assert_operand_cnt(3);
5081 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5082 l = lmod_bytes(po, po->operand[0].lmod) * 8;
5083 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5084 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
5085 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[2]);
5086 fprintf(fout, " %s >>= %s; %s |= %s << (%d - %s);",
5087 buf1, buf3, buf1, buf2, l, buf3);
5088 strcpy(g_comment, "shrd");
5089 output_std_flags(fout, po, &pfomask, buf1);
5090 last_arith_dst = &po->operand[0];
5091 delayed_flag_op = NULL;
5092 break;
5093
d4e3b5db 5094 case OP_ROL:
5095 case OP_ROR:
5096 assert_operand_cnt(2);
5097 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5098 if (po->operand[1].type == OPT_CONST) {
5099 j = po->operand[1].val;
5100 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
5101 fprintf(fout, po->op == OP_ROL ?
5102 " %s = (%s << %d) | (%s >> %d);" :
5103 " %s = (%s >> %d) | (%s << %d);",
5104 buf1, buf1, j, buf1,
5105 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
5106 }
5107 else
5108 ferr(po, "TODO\n");
04f8a628 5109 output_std_flags(fout, po, &pfomask, buf1);
d4e3b5db 5110 last_arith_dst = &po->operand[0];
5111 delayed_flag_op = NULL;
5112 break;
5113
cb090db0 5114 case OP_RCL:
5115 case OP_RCR:
5116 assert_operand_cnt(2);
5117 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5118 l = lmod_bytes(po, po->operand[0].lmod) * 8;
5119 if (po->operand[1].type == OPT_CONST) {
5120 j = po->operand[1].val % l;
5121 if (j == 0)
5122 ferr(po, "zero rotate\n");
5123 fprintf(fout, " tmp = (%s >> %d) & 1;\n",
5124 buf1, (po->op == OP_RCL) ? (l - j) : (j - 1));
5125 if (po->op == OP_RCL) {
5126 fprintf(fout,
5127 " %s = (%s << %d) | (cond_c << %d)",
5128 buf1, buf1, j, j - 1);
5129 if (j != 1)
5130 fprintf(fout, " | (%s >> %d)", buf1, l + 1 - j);
5131 }
5132 else {
5133 fprintf(fout,
5134 " %s = (%s >> %d) | (cond_c << %d)",
5135 buf1, buf1, j, l - j);
5136 if (j != 1)
5137 fprintf(fout, " | (%s << %d)", buf1, l + 1 - j);
5138 }
5139 fprintf(fout, ";\n");
5140 fprintf(fout, " cond_c = tmp;");
5141 }
5142 else
5143 ferr(po, "TODO\n");
5144 strcpy(g_comment, (po->op == OP_RCL) ? "rcl" : "rcr");
5145 output_std_flags(fout, po, &pfomask, buf1);
5146 last_arith_dst = &po->operand[0];
5147 delayed_flag_op = NULL;
5148 break;
5149
5101a5f9 5150 case OP_XOR:
850c9265 5151 assert_operand_cnt(2);
5152 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5101a5f9 5153 if (IS(opr_name(po, 0), opr_name(po, 1))) {
5154 // special case for XOR
2fe80fdb 5155 if (pfomask & (1 << PFO_BE)) { // weird, but it happens..
5156 fprintf(fout, " cond_be = 1;\n");
5157 pfomask &= ~(1 << PFO_BE);
5158 }
5101a5f9 5159 fprintf(fout, " %s = 0;",
5160 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
5161 last_arith_dst = &po->operand[0];
5162 delayed_flag_op = NULL;
850c9265 5163 break;
850c9265 5164 }
5101a5f9 5165 goto dualop_arith;
5166
2fe80fdb 5167 case OP_ADD:
5168 assert_operand_cnt(2);
5169 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5170 if (pfomask & (1 << PFO_C)) {
c8dbc5be 5171 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
5172 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
5173 if (po->operand[0].lmod == OPLM_DWORD) {
5174 fprintf(fout, " tmp64 = (u64)%s + %s;\n", buf1, buf2);
5175 fprintf(fout, " cond_c = tmp64 >> 32;\n");
5176 fprintf(fout, " %s = (u32)tmp64;",
5177 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
5178 strcat(g_comment, "add64");
5179 }
5180 else {
5181 fprintf(fout, " cond_c = ((u32)%s + %s) >> %d;\n",
5182 buf1, buf2, lmod_bytes(po, po->operand[0].lmod) * 8);
5183 fprintf(fout, " %s += %s;",
5184 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5185 buf2);
5186 }
2fe80fdb 5187 pfomask &= ~(1 << PFO_C);
5188 output_std_flags(fout, po, &pfomask, buf1);
5189 last_arith_dst = &po->operand[0];
5190 delayed_flag_op = NULL;
5191 break;
5192 }
5193 goto dualop_arith;
5194
5195 case OP_SUB:
5196 assert_operand_cnt(2);
5197 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3b2f4044 5198 if (pfomask & ~((1 << PFO_Z) | (1 << PFO_S))) {
5199 for (j = 0; j <= PFO_LE; j++) {
5200 if (!(pfomask & (1 << j)))
5201 continue;
5202 if (j == PFO_Z || j == PFO_S)
5203 continue;
5204
5205 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
5206 fprintf(fout, " cond_%s = %s;\n",
5207 parsed_flag_op_names[j], buf1);
5208 pfomask &= ~(1 << j);
5209 }
2fe80fdb 5210 }
5211 goto dualop_arith;
5212
5101a5f9 5213 case OP_ADC:
850c9265 5214 case OP_SBB:
5101a5f9 5215 assert_operand_cnt(2);
5216 propagate_lmod(po, &po->operand[0], &po->operand[1]);
a2c1d768 5217 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
840257f6 5218 if (po->op == OP_SBB
5219 && IS(po->operand[0].name, po->operand[1].name))
5220 {
5221 // avoid use of unitialized var
a2c1d768 5222 fprintf(fout, " %s = -cond_c;", buf1);
94d447fb 5223 // carry remains what it was
5224 pfomask &= ~(1 << PFO_C);
840257f6 5225 }
5226 else {
a2c1d768 5227 fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po),
3ebea2cf 5228 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
840257f6 5229 }
a2c1d768 5230 output_std_flags(fout, po, &pfomask, buf1);
5101a5f9 5231 last_arith_dst = &po->operand[0];
5232 delayed_flag_op = NULL;
850c9265 5233 break;
5234
1f84f6b3 5235 case OP_BSF:
5236 assert_operand_cnt(2);
5237 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
5238 fprintf(fout, " %s = %s ? __builtin_ffs(%s) - 1 : 0;",
5239 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5240 buf2, buf2);
5241 output_std_flags(fout, po, &pfomask, buf1);
5242 last_arith_dst = &po->operand[0];
5243 delayed_flag_op = NULL;
5244 strcat(g_comment, "bsf");
5245 break;
5246
850c9265 5247 case OP_DEC:
90307a99 5248 if (pfomask & ~(PFOB_S | PFOB_S | PFOB_C)) {
5249 for (j = 0; j <= PFO_LE; j++) {
5250 if (!(pfomask & (1 << j)))
5251 continue;
5252 if (j == PFO_Z || j == PFO_S || j == PFO_C)
5253 continue;
5254
5255 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
5256 fprintf(fout, " cond_%s = %s;\n",
5257 parsed_flag_op_names[j], buf1);
5258 pfomask &= ~(1 << j);
5259 }
5260 }
5261 // fallthrough
5262
5263 case OP_INC:
5264 if (pfomask & (1 << PFO_C))
5265 // carry is unaffected by inc/dec.. wtf?
5266 ferr(po, "carry propagation needed\n");
5267
850c9265 5268 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
1bafb621 5269 if (po->operand[0].type == OPT_REG) {
5270 strcpy(buf2, po->op == OP_INC ? "++" : "--");
5271 fprintf(fout, " %s%s;", buf1, buf2);
5272 }
5273 else {
5274 strcpy(buf2, po->op == OP_INC ? "+" : "-");
5275 fprintf(fout, " %s %s= 1;", buf1, buf2);
5276 }
a2c1d768 5277 output_std_flags(fout, po, &pfomask, buf1);
5101a5f9 5278 last_arith_dst = &po->operand[0];
5279 delayed_flag_op = NULL;
5280 break;
5281
5282 case OP_NEG:
5283 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3ebea2cf 5284 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
5101a5f9 5285 fprintf(fout, " %s = -%s%s;", buf1,
5286 lmod_cast_s(po, po->operand[0].lmod), buf2);
850c9265 5287 last_arith_dst = &po->operand[0];
69a3cdfc 5288 delayed_flag_op = NULL;
940e8e66 5289 if (pfomask & (1 << PFO_C)) {
5290 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
5291 pfomask &= ~(1 << PFO_C);
5292 }
850c9265 5293 break;
5294
5295 case OP_IMUL:
de50b98b 5296 if (po->operand_cnt == 2) {
5297 propagate_lmod(po, &po->operand[0], &po->operand[1]);
850c9265 5298 goto dualop_arith;
de50b98b 5299 }
87bf6cec 5300 if (po->operand_cnt == 3)
5301 ferr(po, "TODO imul3\n");
5302 // fallthrough
5303 case OP_MUL:
5304 assert_operand_cnt(1);
c8dbc5be 5305 switch (po->operand[0].lmod) {
5306 case OPLM_DWORD:
5307 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
5308 fprintf(fout, " tmp64 = %seax * %s%s;\n", buf1, buf1,
5309 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
5310 fprintf(fout, " edx = tmp64 >> 32;\n");
5311 fprintf(fout, " eax = tmp64;");
5312 break;
5313 case OPLM_BYTE:
5314 strcpy(buf1, po->op == OP_IMUL ? "(s16)(s8)" : "(u16)(u8)");
5315 fprintf(fout, " LOWORD(eax) = %seax * %s;", buf1,
5316 out_src_opr(buf2, sizeof(buf2), po, &po->operand[0],
5317 buf1, 0));
5318 break;
5319 default:
5320 ferr(po, "TODO: unhandled mul type\n");
5321 break;
5322 }
87bf6cec 5323 last_arith_dst = NULL;
69a3cdfc 5324 delayed_flag_op = NULL;
91977a1c 5325 break;
5326
5101a5f9 5327 case OP_DIV:
5328 case OP_IDIV:
5329 assert_operand_cnt(1);
5330 if (po->operand[0].lmod != OPLM_DWORD)
5331 ferr(po, "unhandled lmod %d\n", po->operand[0].lmod);
5332
cb090db0 5333 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
5334 strcpy(buf2, lmod_cast(po, po->operand[0].lmod,
5335 po->op == OP_IDIV));
5336 switch (po->operand[0].lmod) {
5337 case OPLM_DWORD:
5338 if (po->flags & OPF_32BIT)
5339 snprintf(buf3, sizeof(buf3), "%seax", buf2);
5340 else {
5341 fprintf(fout, " tmp64 = ((u64)edx << 32) | eax;\n");
5342 snprintf(buf3, sizeof(buf3), "%stmp64",
5343 (po->op == OP_IDIV) ? "(s64)" : "");
5344 }
5345 if (po->operand[0].type == OPT_REG
5346 && po->operand[0].reg == xDX)
5347 {
5348 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
5349 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
5350 }
5351 else {
5352 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
5353 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
5354 }
5355 break;
5356 default:
5357 ferr(po, "unhandled division type\n");
5101a5f9 5358 }
87bf6cec 5359 last_arith_dst = NULL;
5360 delayed_flag_op = NULL;
5101a5f9 5361 break;
5362
91977a1c 5363 case OP_TEST:
5364 case OP_CMP:
850c9265 5365 propagate_lmod(po, &po->operand[0], &po->operand[1]);
940e8e66 5366 if (pfomask != 0) {
69a3cdfc 5367 for (j = 0; j < 8; j++) {
940e8e66 5368 if (pfomask & (1 << j)) {
69a3cdfc 5369 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
5370 fprintf(fout, " cond_%s = %s;",
5371 parsed_flag_op_names[j], buf1);
5372 }
5373 }
940e8e66 5374 pfomask = 0;
69a3cdfc 5375 }
5376 else
5377 no_output = 1;
7ba45c34 5378 last_arith_dst = NULL;
69a3cdfc 5379 delayed_flag_op = po;
91977a1c 5380 break;
5381
092f64e1 5382 case OP_SCC:
5383 // SETcc - should already be handled
5384 break;
5385
69a3cdfc 5386 // note: we reuse OP_Jcc for SETcc, only flags differ
092f64e1 5387 case OP_JCC:
5388 fprintf(fout, "\n goto %s;", po->operand[0].name);
850c9265 5389 break;
5390
5c024ef7 5391 case OP_JECXZ:
5392 fprintf(fout, " if (ecx == 0)\n");
5393 fprintf(fout, " goto %s;", po->operand[0].name);
5394 strcat(g_comment, "jecxz");
5395 break;
5396
850c9265 5397 case OP_JMP:
87bf6cec 5398 assert_operand_cnt(1);
de50b98b 5399 last_arith_dst = NULL;
5400 delayed_flag_op = NULL;
5401
4c45fa73 5402 if (po->operand[0].type == OPT_REGMEM) {
5403 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
5404 buf1, buf2);
5405 if (ret != 2)
5406 ferr(po, "parse failure for jmp '%s'\n",
5407 po->operand[0].name);
5408 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
5409 break;
5410 }
5411 else if (po->operand[0].type != OPT_LABEL)
5412 ferr(po, "unhandled jmp type\n");
87bf6cec 5413
850c9265 5414 fprintf(fout, " goto %s;", po->operand[0].name);
91977a1c 5415 break;
5416
5417 case OP_CALL:
5101a5f9 5418 assert_operand_cnt(1);
092f64e1 5419 pp = po->pp;
89ff3147 5420 my_assert_not(pp, NULL);
91977a1c 5421
092f64e1 5422 strcpy(buf3, " ");
5423 if (po->flags & OPF_CC) {
5424 // we treat conditional branch to another func
5425 // (yes such code exists..) as conditional tailcall
5426 strcat(buf3, " ");
5427 fprintf(fout, " {\n");
5428 }
5429
8eb12e72 5430 if (pp->is_fptr && !pp->is_arg) {
092f64e1 5431 fprintf(fout, "%s%s = %s;\n", buf3, pp->name,
1cd4a663 5432 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
5433 "(void *)", 0));
8eb12e72 5434 if (pp->is_unresolved)
5435 fprintf(fout, "%sunresolved_call(\"%s:%d\", %s);\n",
5436 buf3, asmfn, po->asmln, pp->name);
5437 }
1bafb621 5438
092f64e1 5439 fprintf(fout, "%s", buf3);
2b43685d 5440 if (strstr(pp->ret_type.name, "int64")) {
87bf6cec 5441 if (po->flags & OPF_TAIL)
2b43685d 5442 ferr(po, "int64 and tail?\n");
2fe80fdb 5443 fprintf(fout, "tmp64 = ");
2b43685d 5444 }
5445 else if (!IS(pp->ret_type.name, "void")) {
5446 if (po->flags & OPF_TAIL) {
840257f6 5447 if (!IS(g_func_pp->ret_type.name, "void")) {
5448 fprintf(fout, "return ");
5449 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
5450 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
5451 }
2b43685d 5452 }
75ad0378 5453 else if (regmask & (1 << xAX)) {
87bf6cec 5454 fprintf(fout, "eax = ");
2b43685d 5455 if (pp->ret_type.is_ptr)
5456 fprintf(fout, "(u32)");
5457 }
91977a1c 5458 }
87bf6cec 5459
ddaf8bd7 5460 if (pp->name[0] == 0)
5461 ferr(po, "missing pp->name\n");
5462 fprintf(fout, "%s%s(", pp->name,
5463 pp->has_structarg ? "_sa" : "");
39b168b8 5464
2fe80fdb 5465 if (po->flags & OPF_ATAIL) {
5466 if (pp->argc_stack != g_func_pp->argc_stack
5467 || (pp->argc_stack > 0
5468 && pp->is_stdcall != g_func_pp->is_stdcall))
5469 ferr(po, "incompatible tailcall\n");
1f84f6b3 5470 if (g_func_pp->has_retreg)
5471 ferr(po, "TODO: retreg+tailcall\n");
87bf6cec 5472
2fe80fdb 5473 for (arg = j = 0; arg < pp->argc; arg++) {
5474 if (arg > 0)
5475 fprintf(fout, ", ");
87bf6cec 5476
2fe80fdb 5477 cast[0] = 0;
5478 if (pp->arg[arg].type.is_ptr)
5479 snprintf(cast, sizeof(cast), "(%s)",
5480 pp->arg[arg].type.name);
91977a1c 5481
2fe80fdb 5482 if (pp->arg[arg].reg != NULL) {
5483 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
5484 continue;
5485 }
5486 // stack arg
5487 for (; j < g_func_pp->argc; j++)
5488 if (g_func_pp->arg[j].reg == NULL)
5489 break;
5490 fprintf(fout, "%sa%d", cast, j + 1);
5491 j++;
69a3cdfc 5492 }
2fe80fdb 5493 }
5494 else {
5495 for (arg = 0; arg < pp->argc; arg++) {
5496 if (arg > 0)
5497 fprintf(fout, ", ");
5498
5499 cast[0] = 0;
5500 if (pp->arg[arg].type.is_ptr)
5501 snprintf(cast, sizeof(cast), "(%s)",
5502 pp->arg[arg].type.name);
5503
5504 if (pp->arg[arg].reg != NULL) {
1f84f6b3 5505 if (pp->arg[arg].type.is_retreg)
5506 fprintf(fout, "&%s", pp->arg[arg].reg);
5507 else
5508 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
2fe80fdb 5509 continue;
5510 }
5511
5512 // stack arg
5513 tmp_op = pp->arg[arg].datap;
5514 if (tmp_op == NULL)
5515 ferr(po, "parsed_op missing for arg%d\n", arg);
23fd0b11 5516
5517 if (tmp_op->flags & OPF_VAPUSH) {
5518 fprintf(fout, "ap");
5519 }
5f70a34f 5520 else if (tmp_op->p_argpass != 0) {
5521 fprintf(fout, "a%d", tmp_op->p_argpass);
5522 }
5523 else if (tmp_op->p_argnum != 0) {
3a5101d7 5524 fprintf(fout, "%s%s", cast,
5525 saved_arg_name(buf1, sizeof(buf1),
5526 tmp_op->p_arggrp, tmp_op->p_argnum));
2fe80fdb 5527 }
5528 else {
5529 fprintf(fout, "%s",
5530 out_src_opr(buf1, sizeof(buf1),
5531 tmp_op, &tmp_op->operand[0], cast, 0));
5532 }
69a3cdfc 5533 }
91977a1c 5534 }
5535 fprintf(fout, ");");
87bf6cec 5536
2b43685d 5537 if (strstr(pp->ret_type.name, "int64")) {
5538 fprintf(fout, "\n");
092f64e1 5539 fprintf(fout, "%sedx = tmp64 >> 32;\n", buf3);
5540 fprintf(fout, "%seax = tmp64;", buf3);
2b43685d 5541 }
5542
89ff3147 5543 if (pp->is_unresolved) {
8eb12e72 5544 snprintf(buf2, sizeof(buf2), " unresolved %dreg",
89ff3147 5545 pp->argc_reg);
092f64e1 5546 strcat(g_comment, buf2);
89ff3147 5547 }
5548
87bf6cec 5549 if (po->flags & OPF_TAIL) {
840257f6 5550 ret = 0;
ddaf8bd7 5551 if (i == opcnt - 1 || pp->is_noreturn)
840257f6 5552 ret = 0;
5553 else if (IS(pp->ret_type.name, "void"))
5554 ret = 1;
5555 else if (IS(g_func_pp->ret_type.name, "void"))
5556 ret = 1;
5557 // else already handled as 'return f()'
5558
5559 if (ret) {
da87ae38 5560 if (!IS(g_func_pp->ret_type.name, "void")) {
ddaf8bd7 5561 ferr(po, "int func -> void func tailcall?\n");
da87ae38 5562 }
5563 else {
092f64e1 5564 fprintf(fout, "\n%sreturn;", buf3);
ddaf8bd7 5565 strcat(g_comment, " ^ tailcall");
da87ae38 5566 }
3ebea2cf 5567 }
89ff3147 5568 else
ddaf8bd7 5569 strcat(g_comment, " tailcall");
87bf6cec 5570 }
ddaf8bd7 5571 if (pp->is_noreturn)
5572 strcat(g_comment, " noreturn");
2fe80fdb 5573 if ((po->flags & OPF_ATAIL) && pp->argc_stack > 0)
5574 strcat(g_comment, " argframe");
092f64e1 5575 if (po->flags & OPF_CC)
5576 strcat(g_comment, " cond");
5577
5578 if (po->flags & OPF_CC)
5579 fprintf(fout, "\n }");
5580
87bf6cec 5581 delayed_flag_op = NULL;
5582 last_arith_dst = NULL;
91977a1c 5583 break;
5584
5585 case OP_RET:
bd96f656 5586 if (g_func_pp->is_vararg)
4f12f671 5587 fprintf(fout, " va_end(ap);\n");
1f84f6b3 5588 if (g_func_pp->has_retreg) {
5589 for (arg = 0; arg < g_func_pp->argc; arg++)
5590 if (g_func_pp->arg[arg].type.is_retreg)
5591 fprintf(fout, " *r_%s = %s;\n",
5592 g_func_pp->arg[arg].reg, g_func_pp->arg[arg].reg);
5593 }
4f12f671 5594
bd96f656 5595 if (IS(g_func_pp->ret_type.name, "void")) {
3ebea2cf 5596 if (i != opcnt - 1 || label_pending)
5597 fprintf(fout, " return;");
5598 }
bd96f656 5599 else if (g_func_pp->ret_type.is_ptr) {
d4e3b5db 5600 fprintf(fout, " return (%s)eax;",
bd96f656 5601 g_func_pp->ret_type.name);
3ebea2cf 5602 }
2fe80fdb 5603 else if (IS(g_func_pp->ret_type.name, "__int64"))
5604 fprintf(fout, " return ((u64)edx << 32) | eax;");
91977a1c 5605 else
5606 fprintf(fout, " return eax;");
de50b98b 5607
5608 last_arith_dst = NULL;
5609 delayed_flag_op = NULL;
91977a1c 5610 break;
5611
5612 case OP_PUSH:
1f84f6b3 5613 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
5f70a34f 5614 if (po->p_argnum != 0) {
69a3cdfc 5615 // special case - saved func arg
3a5101d7 5616 fprintf(fout, " %s = %s;",
5617 saved_arg_name(buf2, sizeof(buf2),
5618 po->p_arggrp, po->p_argnum), buf1);
69a3cdfc 5619 break;
5620 }
d4e3b5db 5621 else if (po->flags & OPF_RSAVE) {
d4e3b5db 5622 fprintf(fout, " s_%s = %s;", buf1, buf1);
5623 break;
5624 }
25a330eb 5625 else if (po->flags & OPF_PPUSH) {
5626 tmp_op = po->datap;
5627 ferr_assert(po, tmp_op != NULL);
5628 out_dst_opr(buf2, sizeof(buf2), po, &tmp_op->operand[0]);
5629 fprintf(fout, " pp_%s = %s;", buf2, buf1);
5630 break;
5631 }
1f84f6b3 5632 else if (g_func_pp->is_userstack) {
5633 fprintf(fout, " *(--esp) = %s;", buf1);
5634 break;
5635 }
e56ab892 5636 if (!(g_ida_func_attr & IDAFA_NORETURN))
5637 ferr(po, "stray push encountered\n");
5638 no_output = 1;
91977a1c 5639 break;
5640
5641 case OP_POP:
25a330eb 5642 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
d4e3b5db 5643 if (po->flags & OPF_RSAVE) {
d4e3b5db 5644 fprintf(fout, " %s = s_%s;", buf1, buf1);
5645 break;
5646 }
25a330eb 5647 else if (po->flags & OPF_PPUSH) {
5648 // push/pop graph
5649 ferr_assert(po, po->datap == NULL);
5650 fprintf(fout, " %s = pp_%s;", buf1, buf1);
5651 break;
5652 }
5c024ef7 5653 else if (po->datap != NULL) {
5654 // push/pop pair
5655 tmp_op = po->datap;
5c024ef7 5656 fprintf(fout, " %s = %s;", buf1,
5657 out_src_opr(buf2, sizeof(buf2),
5658 tmp_op, &tmp_op->operand[0],
c7ed83dd 5659 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
5c024ef7 5660 break;
5661 }
1f84f6b3 5662 else if (g_func_pp->is_userstack) {
25a330eb 5663 fprintf(fout, " %s = *esp++;", buf1);
1f84f6b3 5664 break;
5665 }
5666 else
5667 ferr(po, "stray pop encountered\n");
91977a1c 5668 break;
5669
33c35af6 5670 case OP_NOP:
2b43685d 5671 no_output = 1;
33c35af6 5672 break;
5673
90307a99 5674 // mmx
5675 case OP_EMMS:
5676 strcpy(g_comment, "(emms)");
5677 break;
5678
91977a1c 5679 default:
5680 no_output = 1;
69a3cdfc 5681 ferr(po, "unhandled op type %d, flags %x\n",
5682 po->op, po->flags);
91977a1c 5683 break;
5684 }
5685
5686 if (g_comment[0] != 0) {
ddaf8bd7 5687 char *p = g_comment;
5688 while (my_isblank(*p))
5689 p++;
5690 fprintf(fout, " // %s", p);
91977a1c 5691 g_comment[0] = 0;
5692 no_output = 0;
5693 }
5694 if (!no_output)
5695 fprintf(fout, "\n");
5101a5f9 5696
2b43685d 5697 // some sanity checking
591721d7 5698 if (po->flags & OPF_REP) {
5699 if (po->op != OP_STOS && po->op != OP_MOVS
5700 && po->op != OP_CMPS && po->op != OP_SCAS)
5701 ferr(po, "unexpected rep\n");
5702 if (!(po->flags & (OPF_REPZ|OPF_REPNZ))
5703 && (po->op == OP_CMPS || po->op == OP_SCAS))
5704 ferr(po, "cmps/scas with plain rep\n");
5705 }
5706 if ((po->flags & (OPF_REPZ|OPF_REPNZ))
5707 && po->op != OP_CMPS && po->op != OP_SCAS)
2b43685d 5708 ferr(po, "unexpected repz/repnz\n");
5709
940e8e66 5710 if (pfomask != 0)
7ba45c34 5711 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
940e8e66 5712
5101a5f9 5713 // see is delayed flag stuff is still valid
5714 if (delayed_flag_op != NULL && delayed_flag_op != po) {
89ff3147 5715 if (is_any_opr_modified(delayed_flag_op, po, 0))
5101a5f9 5716 delayed_flag_op = NULL;
5717 }
5718
5719 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
5720 if (is_opr_modified(last_arith_dst, po))
5721 last_arith_dst = NULL;
5722 }
3ebea2cf 5723
5724 label_pending = 0;
91977a1c 5725 }
5726
a2c1d768 5727 if (g_stack_fsz && !g_stack_frame_used)
5728 fprintf(fout, " (void)sf;\n");
5729
91977a1c 5730 fprintf(fout, "}\n\n");
5731
9af2d373 5732 gen_x_cleanup(opcnt);
5733}
5734
5735static void gen_x_cleanup(int opcnt)
5736{
5737 int i;
5738
91977a1c 5739 for (i = 0; i < opcnt; i++) {
4c45fa73 5740 struct label_ref *lr, *lr_del;
5741
5742 lr = g_label_refs[i].next;
5743 while (lr != NULL) {
5744 lr_del = lr;
5745 lr = lr->next;
5746 free(lr_del);
5747 }
5748 g_label_refs[i].i = -1;
5749 g_label_refs[i].next = NULL;
5750
91977a1c 5751 if (ops[i].op == OP_CALL) {
092f64e1 5752 if (ops[i].pp)
5753 proto_release(ops[i].pp);
91977a1c 5754 }
5755 }
bd96f656 5756 g_func_pp = NULL;
91977a1c 5757}
c36e914d 5758
92d715b6 5759struct func_proto_dep;
5760
5761struct func_prototype {
5762 char name[NAMELEN];
5763 int id;
5764 int argc_stack;
5765 int regmask_dep;
91ca764a 5766 int has_ret:3; // -1, 0, 1: unresolved, no, yes
92d715b6 5767 unsigned int dep_resolved:1;
5768 unsigned int is_stdcall:1;
5769 struct func_proto_dep *dep_func;
5770 int dep_func_cnt;
91ca764a 5771 const struct parsed_proto *pp; // seed pp, if any
92d715b6 5772};
5773
5774struct func_proto_dep {
5775 char *name;
5776 struct func_prototype *proto;
5777 int regmask_live; // .. at the time of call
5778 unsigned int ret_dep:1; // return from this is caller's return
5779};
5780
5781static struct func_prototype *hg_fp;
5782static int hg_fp_cnt;
5783
5fa1256f 5784static struct scanned_var {
5785 char name[NAMELEN];
5786 enum opr_lenmod lmod;
5787 unsigned int is_seeded:1;
61e29183 5788 unsigned int is_c_str:1;
c0de9015 5789 const struct parsed_proto *pp; // seed pp, if any
5fa1256f 5790} *hg_vars;
5791static int hg_var_cnt;
5792
9af2d373 5793static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
5794 int count);
5795
ebc4dc43 5796struct func_prototype *hg_fp_add(const char *funcn)
5797{
5798 struct func_prototype *fp;
5799
5800 if ((hg_fp_cnt & 0xff) == 0) {
5801 hg_fp = realloc(hg_fp, sizeof(hg_fp[0]) * (hg_fp_cnt + 0x100));
5802 my_assert_not(hg_fp, NULL);
5803 memset(hg_fp + hg_fp_cnt, 0, sizeof(hg_fp[0]) * 0x100);
5804 }
5805
5806 fp = &hg_fp[hg_fp_cnt];
5807 snprintf(fp->name, sizeof(fp->name), "%s", funcn);
5808 fp->id = hg_fp_cnt;
5809 fp->argc_stack = -1;
5810 hg_fp_cnt++;
5811
5812 return fp;
5813}
5814
92d715b6 5815static struct func_proto_dep *hg_fp_find_dep(struct func_prototype *fp,
5816 const char *name)
5817{
5818 int i;
5819
5820 for (i = 0; i < fp->dep_func_cnt; i++)
5821 if (IS(fp->dep_func[i].name, name))
5822 return &fp->dep_func[i];
5823
5824 return NULL;
5825}
5826
5827static void hg_fp_add_dep(struct func_prototype *fp, const char *name)
5828{
5829 // is it a dupe?
5830 if (hg_fp_find_dep(fp, name))
5831 return;
5832
5833 if ((fp->dep_func_cnt & 0xff) == 0) {
5834 fp->dep_func = realloc(fp->dep_func,
5835 sizeof(fp->dep_func[0]) * (fp->dep_func_cnt + 0x100));
5836 my_assert_not(fp->dep_func, NULL);
5837 memset(&fp->dep_func[fp->dep_func_cnt], 0,
5838 sizeof(fp->dep_func[0]) * 0x100);
5839 }
5840 fp->dep_func[fp->dep_func_cnt].name = strdup(name);
5841 fp->dep_func_cnt++;
5842}
5843
5844static int hg_fp_cmp_name(const void *p1_, const void *p2_)
5845{
5846 const struct func_prototype *p1 = p1_, *p2 = p2_;
5847 return strcmp(p1->name, p2->name);
5848}
5849
5850#if 0
5851static int hg_fp_cmp_id(const void *p1_, const void *p2_)
5852{
5853 const struct func_prototype *p1 = p1_, *p2 = p2_;
5854 return p1->id - p2->id;
5855}
5856#endif
5857
91ca764a 5858// recursive register dep pass
5859// - track saved regs (part 2)
5860// - try to figure out arg-regs
5861// - calculate reg deps
5862static void gen_hdr_dep_pass(int i, int opcnt, unsigned char *cbits,
5863 struct func_prototype *fp, int regmask_save, int regmask_dst,
5864 int *regmask_dep, int *has_ret)
5865{
5866 struct func_proto_dep *dep;
5867 struct parsed_op *po;
5868 int from_caller = 0;
5869 int depth;
5870 int j, l;
5871 int reg;
5872 int ret;
5873
5874 for (; i < opcnt; i++)
5875 {
5876 if (cbits[i >> 3] & (1 << (i & 7)))
5877 return;
5878 cbits[i >> 3] |= (1 << (i & 7));
5879
5880 po = &ops[i];
5881
5882 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
5883 if (po->btj != NULL) {
5884 // jumptable
5885 for (j = 0; j < po->btj->count; j++) {
db63af51 5886 check_i(po, po->btj->d[j].bt_i);
91ca764a 5887 gen_hdr_dep_pass(po->btj->d[j].bt_i, opcnt, cbits, fp,
5888 regmask_save, regmask_dst, regmask_dep, has_ret);
5889 }
5890 return;
5891 }
5892
db63af51 5893 check_i(po, po->bt_i);
91ca764a 5894 if (po->flags & OPF_CJMP) {
5895 gen_hdr_dep_pass(po->bt_i, opcnt, cbits, fp,
5896 regmask_save, regmask_dst, regmask_dep, has_ret);
5897 }
5898 else {
5899 i = po->bt_i - 1;
5900 }
5901 continue;
5902 }
5903
5904 if (po->flags & OPF_FARG)
5905 /* (just calculate register deps) */;
5906 else if (po->op == OP_PUSH && po->operand[0].type == OPT_REG)
5907 {
5908 reg = po->operand[0].reg;
5909 if (reg < 0)
5910 ferr(po, "reg not set for push?\n");
5911
5912 if (po->flags & OPF_RSAVE) {
5913 regmask_save |= 1 << reg;
5914 continue;
5915 }
5916 if (po->flags & OPF_DONE)
5917 continue;
5918
5919 depth = 0;
5920 ret = scan_for_pop(i + 1, opcnt,
5921 po->operand[0].name, i + opcnt * 2, 0, &depth, 0);
5922 if (ret == 1) {
5923 regmask_save |= 1 << reg;
5924 po->flags |= OPF_RMD;
5925 scan_for_pop(i + 1, opcnt,
5926 po->operand[0].name, i + opcnt * 3, 0, &depth, 1);
5927 continue;
5928 }
5929 }
5930 else if (po->flags & OPF_RMD)
5931 continue;
5932 else if (po->op == OP_CALL) {
5933 po->regmask_dst |= 1 << xAX;
5934
5935 dep = hg_fp_find_dep(fp, po->operand[0].name);
5936 if (dep != NULL)
5937 dep->regmask_live = regmask_save | regmask_dst;
5938 }
5939 else if (po->op == OP_RET) {
5940 if (po->operand_cnt > 0) {
5941 fp->is_stdcall = 1;
5942 if (fp->argc_stack >= 0
5943 && fp->argc_stack != po->operand[0].val / 4)
5944 ferr(po, "ret mismatch? (%d)\n", fp->argc_stack * 4);
5945 fp->argc_stack = po->operand[0].val / 4;
5946 }
5947 }
5948
ebc4dc43 5949 // if has_ret is 0, there is uninitialized eax path,
5950 // which means it's most likely void func
91ca764a 5951 if (*has_ret != 0 && (po->flags & OPF_TAIL)) {
5952 if (po->op == OP_CALL) {
5953 j = i;
5954 ret = 1;
5955 }
5956 else {
5957 struct parsed_opr opr = { 0, };
5958 opr.type = OPT_REG;
5959 opr.reg = xAX;
5960 j = -1;
5961 from_caller = 0;
5962 ret = resolve_origin(i, &opr, i + opcnt * 4, &j, &from_caller);
5963 }
5964
ebc4dc43 5965 if (ret != 1 && from_caller) {
91ca764a 5966 // unresolved eax - probably void func
5967 *has_ret = 0;
5968 }
5969 else {
ebc4dc43 5970 if (j >= 0 && ops[j].op == OP_CALL) {
5971 dep = hg_fp_find_dep(fp, ops[j].operand[0].name);
91ca764a 5972 if (dep != NULL)
5973 dep->ret_dep = 1;
5974 else
5975 *has_ret = 1;
5976 }
5977 else
5978 *has_ret = 1;
5979 }
5980 }
5981
5982 l = regmask_save | regmask_dst;
5983 if (g_bp_frame && !(po->flags & OPF_EBP_S))
5984 l |= 1 << xBP;
5985
5986 l = po->regmask_src & ~l;
5987#if 0
5988 if (l)
5989 fnote(po, "dep |= %04x, dst %04x, save %04x (f %x)\n",
5990 l, regmask_dst, regmask_save, po->flags);
5991#endif
5992 *regmask_dep |= l;
5993 regmask_dst |= po->regmask_dst;
5994
5995 if (po->flags & OPF_TAIL)
5996 return;
5997 }
5998}
5999
92d715b6 6000static void gen_hdr(const char *funcn, int opcnt)
6001{
26677139 6002 int save_arg_vars[MAX_ARG_GRP] = { 0, };
91ca764a 6003 unsigned char cbits[MAX_OPS / 8];
ebc4dc43 6004 const struct parsed_proto *pp_c;
9af2d373 6005 struct parsed_proto *pp;
92d715b6 6006 struct func_prototype *fp;
92d715b6 6007 struct parsed_op *po;
26677139 6008 int regmask_dummy = 0;
91ca764a 6009 int regmask_dep;
92d715b6 6010 int max_bp_offset = 0;
91ca764a 6011 int has_ret;
bfacdc83 6012 int i, j, l;
6013 int ret;
92d715b6 6014
ebc4dc43 6015 pp_c = proto_parse(g_fhdr, funcn, 1);
6016 if (pp_c != NULL)
6017 // already in seed, will add to hg_fp later
9af2d373 6018 return;
ebc4dc43 6019
6020 fp = hg_fp_add(funcn);
9af2d373 6021
6022 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
6023 g_stack_frame_used = 0;
6024
92d715b6 6025 // pass1:
66bdb2b0 6026 // - resolve all branches
6027 // - parse calls with labels
6028 resolve_branches_parse_calls(opcnt);
6029
6030 // pass2:
9af2d373 6031 // - handle ebp/esp frame, remove ops related to it
6032 scan_prologue_epilogue(opcnt);
6033
66bdb2b0 6034 // pass3:
6035 // - remove dead labels
92d715b6 6036 // - collect calls
92d715b6 6037 for (i = 0; i < opcnt; i++)
6038 {
66bdb2b0 6039 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
6040 free(g_labels[i]);
6041 g_labels[i] = NULL;
6042 }
92d715b6 6043
66bdb2b0 6044 po = &ops[i];
5e49b270 6045 if (po->flags & (OPF_RMD|OPF_DONE))
92d715b6 6046 continue;
6047
6048 if (po->op == OP_CALL) {
66bdb2b0 6049 if (po->operand[0].type == OPT_LABEL)
6050 hg_fp_add_dep(fp, opr_name(po, 0));
6051 else if (po->pp != NULL)
6052 hg_fp_add_dep(fp, po->pp->name);
92d715b6 6053 }
92d715b6 6054 }
6055
66bdb2b0 6056 // pass4:
92d715b6 6057 // - remove dead labels
9af2d373 6058 // - handle push <const>/pop pairs
92d715b6 6059 for (i = 0; i < opcnt; i++)
6060 {
d7857c3a 6061 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
6062 free(g_labels[i]);
6063 g_labels[i] = NULL;
6064 }
9af2d373 6065
91ca764a 6066 po = &ops[i];
6067 if (po->flags & (OPF_RMD|OPF_DONE))
6068 continue;
6069
6070 if (po->op == OP_PUSH && po->operand[0].type == OPT_CONST)
25a330eb 6071 scan_for_pop_const(i, opcnt, &regmask_dummy);
91ca764a 6072 }
6073
66bdb2b0 6074 // pass5:
91ca764a 6075 // - process trivial calls
6076 for (i = 0; i < opcnt; i++)
6077 {
9af2d373 6078 po = &ops[i];
5e49b270 6079 if (po->flags & (OPF_RMD|OPF_DONE))
9af2d373 6080 continue;
6081
26677139 6082 if (po->op == OP_CALL)
6083 {
6084 pp = process_call_early(i, opcnt, &j);
6085 if (pp != NULL) {
6086 if (!(po->flags & OPF_ATAIL))
6087 // since we know the args, try to collect them
6088 if (collect_call_args_early(po, i, pp, &regmask_dummy) != 0)
6089 pp = NULL;
6090 }
6091
6092 if (pp != NULL) {
6093 if (j >= 0) {
6094 // commit esp adjust
5e49b270 6095 if (ops[j].op != OP_POP)
6096 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
bfacdc83 6097 else {
6098 for (l = 0; l < pp->argc_stack; l++)
6099 ops[j + l].flags |= OPF_DONE | OPF_RMD;
6100 }
26677139 6101 }
6102
6103 po->flags |= OPF_DONE;
6104 }
6105 }
26677139 6106 }
6107
66bdb2b0 6108 // pass6:
5e49b270 6109 // - track saved regs (simple)
26677139 6110 // - process calls
6111 for (i = 0; i < opcnt; i++)
6112 {
6113 po = &ops[i];
5e49b270 6114 if (po->flags & (OPF_RMD|OPF_DONE))
26677139 6115 continue;
6116
5e49b270 6117 if (po->op == OP_PUSH && po->operand[0].type == OPT_REG)
6118 {
6119 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
6120 if (ret == 0) {
91ca764a 6121 // regmask_save |= 1 << po->operand[0].reg; // do it later
6122 po->flags |= OPF_RSAVE | OPF_RMD | OPF_DONE;
5e49b270 6123 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, OPF_RMD);
6124 }
6125 }
6126 else if (po->op == OP_CALL && !(po->flags & OPF_DONE))
26677139 6127 {
9af2d373 6128 pp = process_call(i, opcnt);
6129
6130 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
9af2d373 6131 // since we know the args, collect them
91ca764a 6132 ret = collect_call_args(po, i, pp, &regmask_dummy, save_arg_vars,
5e49b270 6133 i + opcnt * 1);
9af2d373 6134 }
6135 }
92d715b6 6136 }
6137
66bdb2b0 6138 // pass7
91ca764a 6139 memset(cbits, 0, sizeof(cbits));
6140 regmask_dep = 0;
6141 has_ret = -1;
6142
6143 gen_hdr_dep_pass(0, opcnt, cbits, fp, 0, 0, &regmask_dep, &has_ret);
6144
6145 // find unreachable code - must be fixed in IDA
92d715b6 6146 for (i = 0; i < opcnt; i++)
6147 {
91ca764a 6148 if (cbits[i >> 3] & (1 << (i & 7)))
9af2d373 6149 continue;
92d715b6 6150
91ca764a 6151 if (ops[i].op != OP_NOP)
6152 ferr(&ops[i], "unreachable code\n");
92d715b6 6153 }
6154
9af2d373 6155 for (i = 0; i < g_eqcnt; i++) {
92d715b6 6156 if (g_eqs[i].offset > max_bp_offset && g_eqs[i].offset < 4*32)
6157 max_bp_offset = g_eqs[i].offset;
9af2d373 6158 }
92d715b6 6159
9af2d373 6160 if (fp->argc_stack < 0) {
92d715b6 6161 max_bp_offset = (max_bp_offset + 3) & ~3;
9af2d373 6162 fp->argc_stack = max_bp_offset / 4;
6163 if ((g_ida_func_attr & IDAFA_BP_FRAME) && fp->argc_stack > 0)
92d715b6 6164 fp->argc_stack--;
6165 }
6166
6167 fp->regmask_dep = regmask_dep & ~(1 << xSP);
6168 fp->has_ret = has_ret;
91ca764a 6169#if 0
6170 printf("// has_ret %d, regmask_dep %x\n",
6171 fp->has_ret, fp->regmask_dep);
6172 output_hdr_fp(stdout, fp, 1);
ebc4dc43 6173 if (IS(funcn, "sub_10007F72")) exit(1);
91ca764a 6174#endif
9af2d373 6175
6176 gen_x_cleanup(opcnt);
92d715b6 6177}
6178
6179static void hg_fp_resolve_deps(struct func_prototype *fp)
6180{
6181 struct func_prototype fp_s;
91ca764a 6182 int dep;
92d715b6 6183 int i;
6184
6185 // this thing is recursive, so mark first..
6186 fp->dep_resolved = 1;
6187
6188 for (i = 0; i < fp->dep_func_cnt; i++) {
6189 strcpy(fp_s.name, fp->dep_func[i].name);
6190 fp->dep_func[i].proto = bsearch(&fp_s, hg_fp, hg_fp_cnt,
6191 sizeof(hg_fp[0]), hg_fp_cmp_name);
6192 if (fp->dep_func[i].proto != NULL) {
6193 if (!fp->dep_func[i].proto->dep_resolved)
6194 hg_fp_resolve_deps(fp->dep_func[i].proto);
6195
91ca764a 6196 dep = ~fp->dep_func[i].regmask_live
6197 & fp->dep_func[i].proto->regmask_dep;
6198 fp->regmask_dep |= dep;
6199 // printf("dep %s %s |= %x\n", fp->name,
6200 // fp->dep_func[i].name, dep);
92d715b6 6201
ebc4dc43 6202 if (fp->has_ret == -1 && fp->dep_func[i].ret_dep)
92d715b6 6203 fp->has_ret = fp->dep_func[i].proto->has_ret;
6204 }
6205 }
6206}
6207
9af2d373 6208static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
6209 int count)
92d715b6 6210{
61e29183 6211 const struct parsed_proto *pp;
6212 char *p, namebuf[NAMELEN];
6213 const char *name;
92d715b6 6214 int regmask_dep;
6215 int argc_stack;
9af2d373 6216 int j, arg;
92d715b6 6217
9af2d373 6218 for (; count > 0; count--, fp++) {
92d715b6 6219 if (fp->has_ret == -1)
6220 fprintf(fout, "// ret unresolved\n");
6221#if 0
6222 fprintf(fout, "// dep:");
6223 for (j = 0; j < fp->dep_func_cnt; j++) {
6224 fprintf(fout, " %s/", fp->dep_func[j].name);
6225 if (fp->dep_func[j].proto != NULL)
6226 fprintf(fout, "%04x/%d", fp->dep_func[j].proto->regmask_dep,
6227 fp->dep_func[j].proto->has_ret);
6228 }
6229 fprintf(fout, "\n");
6230#endif
6231
61e29183 6232 p = strchr(fp->name, '@');
6233 if (p != NULL) {
6234 memcpy(namebuf, fp->name, p - fp->name);
6235 namebuf[p - fp->name] = 0;
6236 name = namebuf;
6237 }
6238 else
6239 name = fp->name;
6240 if (name[0] == '_')
6241 name++;
6242
6243 pp = proto_parse(g_fhdr, name, 1);
6244 if (pp != NULL && pp->is_include)
6245 continue;
6246
c0de9015 6247 if (fp->pp != NULL) {
4e81a3a2 6248 // part of seed, output later
6249 continue;
c0de9015 6250 }
6251
92d715b6 6252 regmask_dep = fp->regmask_dep;
6253 argc_stack = fp->argc_stack;
6254
91ca764a 6255 fprintf(fout, "%-5s", fp->pp ? fp->pp->ret_type.name :
6256 (fp->has_ret ? "int" : "void"));
92d715b6 6257 if (regmask_dep && (fp->is_stdcall || argc_stack == 0)
6258 && (regmask_dep & ~((1 << xCX) | (1 << xDX))) == 0)
6259 {
9af2d373 6260 fprintf(fout, " __fastcall ");
92d715b6 6261 if (!(regmask_dep & (1 << xDX)) && argc_stack == 0)
6262 argc_stack = 1;
6263 else
6264 argc_stack += 2;
6265 regmask_dep = 0;
6266 }
6267 else if (regmask_dep && !fp->is_stdcall) {
6268 fprintf(fout, "/*__usercall*/ ");
92d715b6 6269 }
6270 else if (regmask_dep) {
6271 fprintf(fout, "/*__userpurge*/ ");
92d715b6 6272 }
6273 else if (fp->is_stdcall)
9af2d373 6274 fprintf(fout, " __stdcall ");
92d715b6 6275 else
9af2d373 6276 fprintf(fout, " __cdecl ");
92d715b6 6277
61e29183 6278 fprintf(fout, "%s(", name);
92d715b6 6279
6280 arg = 0;
6281 for (j = 0; j < xSP; j++) {
6282 if (regmask_dep & (1 << j)) {
6283 arg++;
6284 if (arg != 1)
6285 fprintf(fout, ", ");
91ca764a 6286 if (fp->pp != NULL)
6287 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
6288 else
6289 fprintf(fout, "int");
6290 fprintf(fout, " a%d/*<%s>*/", arg, regs_r32[j]);
92d715b6 6291 }
6292 }
6293
6294 for (j = 0; j < argc_stack; j++) {
6295 arg++;
6296 if (arg != 1)
6297 fprintf(fout, ", ");
91ca764a 6298 if (fp->pp != NULL) {
6299 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
6300 if (!fp->pp->arg[arg - 1].type.is_ptr)
6301 fprintf(fout, " ");
6302 }
6303 else
6304 fprintf(fout, "int ");
6305 fprintf(fout, "a%d", arg);
92d715b6 6306 }
6307
6308 fprintf(fout, ");\n");
6309 }
6310}
6311
9af2d373 6312static void output_hdr(FILE *fout)
6313{
5fa1256f 6314 static const char *lmod_c_names[] = {
6315 [OPLM_UNSPEC] = "???",
6316 [OPLM_BYTE] = "uint8_t",
6317 [OPLM_WORD] = "uint16_t",
6318 [OPLM_DWORD] = "uint32_t",
6319 [OPLM_QWORD] = "uint64_t",
6320 };
6321 const struct scanned_var *var;
ebc4dc43 6322 struct func_prototype *fp;
c0de9015 6323 char line[256] = { 0, };
ebc4dc43 6324 char name[256];
9af2d373 6325 int i;
6326
ebc4dc43 6327 // add stuff from headers
6328 for (i = 0; i < pp_cache_size; i++) {
6329 if (pp_cache[i].is_cinc && !pp_cache[i].is_stdcall)
6330 snprintf(name, sizeof(name), "_%s", pp_cache[i].name);
6331 else
6332 snprintf(name, sizeof(name), "%s", pp_cache[i].name);
6333 fp = hg_fp_add(name);
6334 fp->pp = &pp_cache[i];
6335 fp->argc_stack = fp->pp->argc_stack;
6336 fp->is_stdcall = fp->pp->is_stdcall;
6337 fp->regmask_dep = get_pp_arg_regmask(fp->pp);
6338 fp->has_ret = !IS(fp->pp->ret_type.name, "void");
6339 }
6340
9af2d373 6341 // resolve deps
6342 qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_name);
6343 for (i = 0; i < hg_fp_cnt; i++)
6344 hg_fp_resolve_deps(&hg_fp[i]);
6345
6346 // note: messes up .proto ptr, don't use
6347 //qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_id);
6348
5fa1256f 6349 // output variables
6350 for (i = 0; i < hg_var_cnt; i++) {
6351 var = &hg_vars[i];
6352
4e81a3a2 6353 if (var->pp != NULL)
6354 // part of seed
6355 continue;
c0de9015 6356 else if (var->is_c_str)
61e29183 6357 fprintf(fout, "extern %-8s %s[];", "char", var->name);
6358 else
6359 fprintf(fout, "extern %-8s %s;",
6360 lmod_c_names[var->lmod], var->name);
5fa1256f 6361
6362 if (var->is_seeded)
6363 fprintf(fout, " // seeded");
6364 fprintf(fout, "\n");
6365 }
6366
6367 fprintf(fout, "\n");
6368
6369 // output function prototypes
9af2d373 6370 output_hdr_fp(fout, hg_fp, hg_fp_cnt);
c0de9015 6371
4e81a3a2 6372 // seed passthrough
6373 fprintf(fout, "\n// - seed -\n");
c0de9015 6374
6375 rewind(g_fhdr);
4e81a3a2 6376 while (fgets(line, sizeof(line), g_fhdr))
6377 fwrite(line, 1, strlen(line), fout);
9af2d373 6378}
6379
61e29183 6380// '=' needs special treatment
6381// also ' quote
bfa4a6ee 6382static char *next_word_s(char *w, size_t wsize, char *s)
6383{
61e29183 6384 size_t i;
bfa4a6ee 6385
61e29183 6386 s = sskip(s);
bfa4a6ee 6387
61e29183 6388 i = 0;
6389 if (*s == '\'') {
6390 w[0] = s[0];
6391 for (i = 1; i < wsize - 1; i++) {
6392 if (s[i] == 0) {
6393 printf("warning: missing closing quote: \"%s\"\n", s);
6394 break;
6395 }
6396 if (s[i] == '\'')
6397 break;
6398 w[i] = s[i];
6399 }
6400 }
bfa4a6ee 6401
61e29183 6402 for (; i < wsize - 1; i++) {
6403 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
6404 break;
6405 w[i] = s[i];
6406 }
6407 w[i] = 0;
6408
6409 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
6410 printf("warning: '%s' truncated\n", w);
bfa4a6ee 6411
61e29183 6412 return s + i;
bfa4a6ee 6413}
6414
5fa1256f 6415static void scan_variables(FILE *fasm)
6416{
5fa1256f 6417 struct scanned_var *var;
6418 char line[256] = { 0, };
61e29183 6419 char words[3][256];
5fa1256f 6420 char *p = NULL;
6421 int wordc;
61e29183 6422 int l;
5fa1256f 6423
6424 while (!feof(fasm))
6425 {
6426 // skip to next data section
6427 while (my_fgets(line, sizeof(line), fasm))
6428 {
6429 asmln++;
6430
6431 p = sskip(line);
6432 if (*p == 0 || *p == ';')
6433 continue;
6434
6435 p = sskip(next_word_s(words[0], sizeof(words[0]), p));
6436 if (*p == 0 || *p == ';')
6437 continue;
6438
6439 if (*p != 's' || !IS_START(p, "segment para public"))
6440 continue;
6441
6442 break;
6443 }
6444
6445 if (p == NULL || !IS_START(p, "segment para public"))
6446 break;
6447 p = sskip(p + 19);
6448
6449 if (!IS_START(p, "'DATA'"))
6450 continue;
6451
6452 // now process it
6453 while (my_fgets(line, sizeof(line), fasm))
6454 {
6455 asmln++;
6456
6457 p = line;
6458 if (my_isblank(*p))
6459 continue;
6460
6461 p = sskip(p);
6462 if (*p == 0 || *p == ';')
6463 continue;
6464
6465 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
6466 words[wordc][0] = 0;
6467 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
6468 if (*p == 0 || *p == ';') {
6469 wordc++;
6470 break;
6471 }
6472 }
6473
6474 if (wordc == 2 && IS(words[1], "ends"))
6475 break;
61e29183 6476 if (wordc < 2)
6477 continue;
5fa1256f 6478
9ea60b8d 6479 if (IS_START(words[0], "__IMPORT_DESCRIPTOR_")) {
6480 // when this starts, we don't need anything from this section
6481 break;
6482 }
6483
5fa1256f 6484 if ((hg_var_cnt & 0xff) == 0) {
6485 hg_vars = realloc(hg_vars, sizeof(hg_vars[0])
6486 * (hg_var_cnt + 0x100));
6487 my_assert_not(hg_vars, NULL);
6488 memset(hg_vars + hg_var_cnt, 0, sizeof(hg_vars[0]) * 0x100);
6489 }
6490
6491 var = &hg_vars[hg_var_cnt++];
6492 snprintf(var->name, sizeof(var->name), "%s", words[0]);
6493
6494 // maybe already in seed header?
c0de9015 6495 var->pp = proto_parse(g_fhdr, var->name, 1);
6496 if (var->pp != NULL) {
6497 if (var->pp->is_fptr) {
5fa1256f 6498 var->lmod = OPLM_DWORD;
6499 //var->is_ptr = 1;
6500 }
c0de9015 6501 else if (var->pp->is_func)
6502 aerr("func?\n");
6503 else if (!guess_lmod_from_c_type(&var->lmod, &var->pp->type))
5fa1256f 6504 aerr("unhandled C type '%s' for '%s'\n",
c0de9015 6505 var->pp->type.name, var->name);
5fa1256f 6506
6507 var->is_seeded = 1;
6508 continue;
6509 }
6510
6511 if (IS(words[1], "dd"))
6512 var->lmod = OPLM_DWORD;
6513 else if (IS(words[1], "dw"))
6514 var->lmod = OPLM_WORD;
61e29183 6515 else if (IS(words[1], "db")) {
5fa1256f 6516 var->lmod = OPLM_BYTE;
61e29183 6517 if (wordc >= 3 && (l = strlen(words[2])) > 4) {
6518 if (words[2][0] == '\'' && IS(words[2] + l - 2, ",0"))
6519 var->is_c_str = 1;
6520 }
6521 }
5fa1256f 6522 else if (IS(words[1], "dq"))
6523 var->lmod = OPLM_QWORD;
6524 //else if (IS(words[1], "dt"))
6525 else
6526 aerr("type '%s' not known\n", words[1]);
6527 }
6528 }
6529
6530 rewind(fasm);
6531 asmln = 0;
6532}
6533
6534static void set_label(int i, const char *name)
6535{
6536 const char *p;
6537 int len;
6538
6539 len = strlen(name);
6540 p = strchr(name, ':');
6541 if (p != NULL)
6542 len = p - name;
6543
6544 if (g_labels[i] != NULL && !IS_START(g_labels[i], "algn_"))
6545 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
6546 g_labels[i] = realloc(g_labels[i], len + 1);
6547 my_assert_not(g_labels[i], NULL);
6548 memcpy(g_labels[i], name, len);
6549 g_labels[i][len] = 0;
6550}
6551
e56ab892 6552struct chunk_item {
6553 char *name;
6554 long fptr;
de50b98b 6555 int asmln;
e56ab892 6556};
6557
cdfaeed7 6558static struct chunk_item *func_chunks;
6559static int func_chunk_cnt;
6560static int func_chunk_alloc;
6561
6562static void add_func_chunk(FILE *fasm, const char *name, int line)
6563{
6564 if (func_chunk_cnt >= func_chunk_alloc) {
6565 func_chunk_alloc *= 2;
6566 func_chunks = realloc(func_chunks,
6567 func_chunk_alloc * sizeof(func_chunks[0]));
6568 my_assert_not(func_chunks, NULL);
6569 }
6570 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
6571 func_chunks[func_chunk_cnt].name = strdup(name);
6572 func_chunks[func_chunk_cnt].asmln = line;
6573 func_chunk_cnt++;
6574}
6575
e56ab892 6576static int cmp_chunks(const void *p1, const void *p2)
6577{
6578 const struct chunk_item *c1 = p1, *c2 = p2;
6579 return strcmp(c1->name, c2->name);
6580}
6581
bfa4a6ee 6582static int cmpstringp(const void *p1, const void *p2)
6583{
6584 return strcmp(*(char * const *)p1, *(char * const *)p2);
6585}
6586
cdfaeed7 6587static void scan_ahead(FILE *fasm)
6588{
6589 char words[2][256];
6590 char line[256];
6591 long oldpos;
6592 int oldasmln;
6593 int wordc;
6594 char *p;
6595 int i;
6596
6597 oldpos = ftell(fasm);
6598 oldasmln = asmln;
6599
5fa1256f 6600 while (my_fgets(line, sizeof(line), fasm))
cdfaeed7 6601 {
6602 wordc = 0;
6603 asmln++;
6604
6605 p = sskip(line);
6606 if (*p == 0)
6607 continue;
6608
6609 if (*p == ';')
6610 {
6611 // get rid of random tabs
6612 for (i = 0; line[i] != 0; i++)
6613 if (line[i] == '\t')
6614 line[i] = ' ';
6615
6616 if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
6617 {
6618 p += 30;
6619 next_word(words[0], sizeof(words[0]), p);
6620 if (words[0][0] == 0)
6621 aerr("missing name for func chunk?\n");
6622
6623 add_func_chunk(fasm, words[0], asmln);
6624 }
46b388c2 6625 else if (IS_START(p, "; sctend"))
6626 break;
6627
cdfaeed7 6628 continue;
6629 } // *p == ';'
6630
6631 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
6632 words[wordc][0] = 0;
6633 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
6634 if (*p == 0 || *p == ';') {
6635 wordc++;
6636 break;
6637 }
6638 }
6639
6640 if (wordc == 2 && IS(words[1], "ends"))
6641 break;
6642 }
6643
6644 fseek(fasm, oldpos, SEEK_SET);
6645 asmln = oldasmln;
6646}
6647
91977a1c 6648int main(int argc, char *argv[])
6649{
06c5d854 6650 FILE *fout, *fasm, *frlist;
4c45fa73 6651 struct parsed_data *pd = NULL;
6652 int pd_alloc = 0;
6653 char **rlist = NULL;
6654 int rlist_len = 0;
6655 int rlist_alloc = 0;
e56ab892 6656 int func_chunks_used = 0;
6657 int func_chunks_sorted = 0;
e56ab892 6658 int func_chunk_i = -1;
6659 long func_chunk_ret = 0;
de50b98b 6660 int func_chunk_ret_ln = 0;
cdfaeed7 6661 int scanned_ahead = 0;
91977a1c 6662 char line[256];
a2c1d768 6663 char words[20][256];
4c45fa73 6664 enum opr_lenmod lmod;
ddaf8bd7 6665 char *sctproto = NULL;
91977a1c 6666 int in_func = 0;
4c45fa73 6667 int pending_endp = 0;
bfa4a6ee 6668 int skip_func = 0;
940e8e66 6669 int skip_warned = 0;
91977a1c 6670 int eq_alloc;
bfa4a6ee 6671 int verbose = 0;
1f84f6b3 6672 int multi_seg = 0;
46b388c2 6673 int end = 0;
bfa4a6ee 6674 int arg_out;
89ff3147 6675 int arg;
91977a1c 6676 int pi = 0;
e56ab892 6677 int i, j;
6678 int ret, len;
91977a1c 6679 char *p;
6680 int wordc;
6681
89ff3147 6682 for (arg = 1; arg < argc; arg++) {
6683 if (IS(argv[arg], "-v"))
6684 verbose = 1;
6685 else if (IS(argv[arg], "-rf"))
6686 g_allow_regfunc = 1;
1f84f6b3 6687 else if (IS(argv[arg], "-m"))
6688 multi_seg = 1;
92d715b6 6689 else if (IS(argv[arg], "-hdr"))
9af2d373 6690 g_header_mode = g_quiet_pp = g_allow_regfunc = 1;
89ff3147 6691 else
6692 break;
bfa4a6ee 6693 }
6694
6695 if (argc < arg + 3) {
315b77eb 6696 printf("usage:\n%s [-v] [-rf] [-m] <.c> <.asm> <hdr.h> [rlist]*\n"
6697 "%s -hdr <out.h> <.asm> <seed.h> [rlist]*\n"
6698 "options:\n"
6699 " -hdr - header generation mode\n"
6700 " -rf - allow unannotated indirect calls\n"
6701 " -m - allow multiple .text sections\n"
6702 "[rlist] is a file with function names to skip,"
6703 " one per line\n",
92d715b6 6704 argv[0], argv[0]);
91977a1c 6705 return 1;
6706 }
6707
bfa4a6ee 6708 arg_out = arg++;
91977a1c 6709
bfa4a6ee 6710 asmfn = argv[arg++];
91977a1c 6711 fasm = fopen(asmfn, "r");
6712 my_assert_not(fasm, NULL);
6713
bfa4a6ee 6714 hdrfn = argv[arg++];
06c5d854 6715 g_fhdr = fopen(hdrfn, "r");
6716 my_assert_not(g_fhdr, NULL);
bfa4a6ee 6717
6718 rlist_alloc = 64;
6719 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
6720 my_assert_not(rlist, NULL);
6721 // needs special handling..
6722 rlist[rlist_len++] = "__alloca_probe";
6723
e56ab892 6724 func_chunk_alloc = 32;
6725 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
6726 my_assert_not(func_chunks, NULL);
6727
a2c1d768 6728 memset(words, 0, sizeof(words));
6729
bfa4a6ee 6730 for (; arg < argc; arg++) {
6731 frlist = fopen(argv[arg], "r");
6732 my_assert_not(frlist, NULL);
6733
5fa1256f 6734 while (my_fgets(line, sizeof(line), frlist)) {
bfa4a6ee 6735 p = sskip(line);
1cd4a663 6736 if (*p == 0 || *p == ';')
6737 continue;
6738 if (*p == '#') {
89ff3147 6739 if (IS_START(p, "#if 0")
6740 || (g_allow_regfunc && IS_START(p, "#if NO_REGFUNC")))
6741 {
1cd4a663 6742 skip_func = 1;
89ff3147 6743 }
1cd4a663 6744 else if (IS_START(p, "#endif"))
6745 skip_func = 0;
6746 continue;
6747 }
6748 if (skip_func)
bfa4a6ee 6749 continue;
6750
6751 p = next_word(words[0], sizeof(words[0]), p);
6752 if (words[0][0] == 0)
6753 continue;
6754
6755 if (rlist_len >= rlist_alloc) {
6756 rlist_alloc = rlist_alloc * 2 + 64;
6757 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
6758 my_assert_not(rlist, NULL);
6759 }
6760 rlist[rlist_len++] = strdup(words[0]);
6761 }
1cd4a663 6762 skip_func = 0;
bfa4a6ee 6763
6764 fclose(frlist);
6765 frlist = NULL;
6766 }
6767
6768 if (rlist_len > 0)
6769 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
6770
6771 fout = fopen(argv[arg_out], "w");
91977a1c 6772 my_assert_not(fout, NULL);
6773
6774 eq_alloc = 128;
6775 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
6776 my_assert_not(g_eqs, NULL);
6777
4c45fa73 6778 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
6779 g_label_refs[i].i = -1;
6780 g_label_refs[i].next = NULL;
6781 }
6782
5fa1256f 6783 if (g_header_mode)
6784 scan_variables(fasm);
6785
6786 while (my_fgets(line, sizeof(line), fasm))
91977a1c 6787 {
4c45fa73 6788 wordc = 0;
91977a1c 6789 asmln++;
6790
6791 p = sskip(line);
1bafb621 6792 if (*p == 0)
91977a1c 6793 continue;
6794
de50b98b 6795 // get rid of random tabs
6796 for (i = 0; line[i] != 0; i++)
6797 if (line[i] == '\t')
6798 line[i] = ' ';
6799
e56ab892 6800 if (*p == ';')
6801 {
e56ab892 6802 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
6803 goto do_pending_endp; // eww..
6804
6805 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
6806 {
6807 static const char *attrs[] = {
6808 "bp-based frame",
6809 "library function",
6810 "static",
6811 "noreturn",
6812 "thunk",
6813 "fpd=",
6814 };
6815
6816 // parse IDA's attribute-list comment
6817 g_ida_func_attr = 0;
6818 p = sskip(p + 13);
6819
6820 for (; *p != 0; p = sskip(p)) {
6821 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
6822 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
6823 g_ida_func_attr |= 1 << i;
6824 p += strlen(attrs[i]);
6825 break;
6826 }
6827 }
6828 if (i == ARRAY_SIZE(attrs)) {
6829 anote("unparsed IDA attr: %s\n", p);
1bafb621 6830 break;
6831 }
e56ab892 6832 if (IS(attrs[i], "fpd=")) {
6833 p = next_word(words[0], sizeof(words[0]), p);
6834 // ignore for now..
6835 }
1bafb621 6836 }
e56ab892 6837 }
6838 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
6839 {
6840 p += 30;
6841 next_word(words[0], sizeof(words[0]), p);
6842 if (words[0][0] == 0)
cdfaeed7 6843 aerr("missing name for func chunk?\n");
6844
6845 if (!scanned_ahead) {
6846 add_func_chunk(fasm, words[0], asmln);
6847 func_chunks_sorted = 0;
e56ab892 6848 }
e56ab892 6849 }
6850 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
6851 {
6852 if (func_chunk_i >= 0) {
6853 if (func_chunk_i < func_chunk_cnt
6854 && IS(func_chunks[func_chunk_i].name, g_func))
6855 {
6856 // move on to next chunk
6857 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
6858 if (ret)
6859 aerr("seek failed for '%s' chunk #%d\n",
6860 g_func, func_chunk_i);
de50b98b 6861 asmln = func_chunks[func_chunk_i].asmln;
e56ab892 6862 func_chunk_i++;
6863 }
6864 else {
6865 if (func_chunk_ret == 0)
6866 aerr("no return from chunk?\n");
6867 fseek(fasm, func_chunk_ret, SEEK_SET);
de50b98b 6868 asmln = func_chunk_ret_ln;
e56ab892 6869 func_chunk_ret = 0;
6870 pending_endp = 1;
6871 }
1bafb621 6872 }
e56ab892 6873 }
6874 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
6875 func_chunks_used = 1;
6876 p += 20;
6877 if (IS_START(g_func, "sub_")) {
6878 unsigned long addr = strtoul(p, NULL, 16);
6879 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
cdfaeed7 6880 if (addr > f_addr && !scanned_ahead) {
94d447fb 6881 anote("scan_ahead caused by '%s', addr %lx\n",
6882 g_func, addr);
cdfaeed7 6883 scan_ahead(fasm);
6884 scanned_ahead = 1;
6885 func_chunks_sorted = 0;
6886 }
1bafb621 6887 }
6888 }
6889 continue;
e56ab892 6890 } // *p == ';'
1bafb621 6891
06c5d854 6892parse_words:
a2c1d768 6893 for (i = wordc; i < ARRAY_SIZE(words); i++)
6894 words[i][0] = 0;
cdfaeed7 6895 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
bfa4a6ee 6896 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
91977a1c 6897 if (*p == 0 || *p == ';') {
6898 wordc++;
6899 break;
6900 }
6901 }
a2c1d768 6902 if (*p != 0 && *p != ';')
6903 aerr("too many words\n");
91977a1c 6904
06c5d854 6905 // alow asm patches in comments
ddaf8bd7 6906 if (*p == ';') {
6907 if (IS_START(p, "; sctpatch:")) {
6908 p = sskip(p + 11);
6909 if (*p == 0 || *p == ';')
6910 continue;
6911 goto parse_words; // lame
6912 }
6913 if (IS_START(p, "; sctproto:")) {
6914 sctproto = strdup(p + 11);
6915 }
46b388c2 6916 else if (IS_START(p, "; sctend")) {
6917 end = 1;
6918 if (!pending_endp)
6919 break;
6920 }
06c5d854 6921 }
6922
91977a1c 6923 if (wordc == 0) {
6924 // shouldn't happen
6925 awarn("wordc == 0?\n");
6926 continue;
6927 }
6928
6929 // don't care about this:
6930 if (words[0][0] == '.'
6931 || IS(words[0], "include")
6932 || IS(words[0], "assume") || IS(words[1], "segment")
6933 || IS(words[0], "align"))
6934 {
6935 continue;
6936 }
6937
4c45fa73 6938do_pending_endp:
6939 // do delayed endp processing to collect switch jumptables
6940 if (pending_endp) {
30c8c549 6941 if (in_func && !g_skip_func && !end && wordc >= 2
4c45fa73 6942 && ((words[0][0] == 'd' && words[0][2] == 0)
6943 || (words[1][0] == 'd' && words[1][2] == 0)))
6944 {
6945 i = 1;
6946 if (words[1][0] == 'd' && words[1][2] == 0) {
6947 // label
6948 if (g_func_pd_cnt >= pd_alloc) {
6949 pd_alloc = pd_alloc * 2 + 16;
6950 g_func_pd = realloc(g_func_pd,
6951 sizeof(g_func_pd[0]) * pd_alloc);
6952 my_assert_not(g_func_pd, NULL);
6953 }
6954 pd = &g_func_pd[g_func_pd_cnt];
6955 g_func_pd_cnt++;
6956 memset(pd, 0, sizeof(*pd));
6957 strcpy(pd->label, words[0]);
6958 pd->type = OPT_CONST;
6959 pd->lmod = lmod_from_directive(words[1]);
6960 i = 2;
6961 }
6962 else {
da87ae38 6963 if (pd == NULL) {
6964 if (verbose)
6965 anote("skipping alignment byte?\n");
6966 continue;
6967 }
4c45fa73 6968 lmod = lmod_from_directive(words[0]);
6969 if (lmod != pd->lmod)
6970 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
6971 }
6972
6973 if (pd->count_alloc < pd->count + wordc) {
6974 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
6975 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
6976 my_assert_not(pd->d, NULL);
6977 }
6978 for (; i < wordc; i++) {
6979 if (IS(words[i], "offset")) {
6980 pd->type = OPT_OFFSET;
6981 i++;
6982 }
6983 p = strchr(words[i], ',');
6984 if (p != NULL)
6985 *p = 0;
6986 if (pd->type == OPT_OFFSET)
6987 pd->d[pd->count].u.label = strdup(words[i]);
6988 else
6989 pd->d[pd->count].u.val = parse_number(words[i]);
6990 pd->d[pd->count].bt_i = -1;
6991 pd->count++;
6992 }
6993 continue;
6994 }
6995
30c8c549 6996 if (in_func && !g_skip_func) {
9af2d373 6997 if (g_header_mode)
92d715b6 6998 gen_hdr(g_func, pi);
6999 else
7000 gen_func(fout, g_fhdr, g_func, pi);
7001 }
4c45fa73 7002
7003 pending_endp = 0;
7004 in_func = 0;
7005 g_ida_func_attr = 0;
7006 skip_warned = 0;
30c8c549 7007 g_skip_func = 0;
4c45fa73 7008 g_func[0] = 0;
e56ab892 7009 func_chunks_used = 0;
7010 func_chunk_i = -1;
4c45fa73 7011 if (pi != 0) {
7012 memset(&ops, 0, pi * sizeof(ops[0]));
d7857c3a 7013 clear_labels(pi);
4c45fa73 7014 pi = 0;
7015 }
7016 g_eqcnt = 0;
7017 for (i = 0; i < g_func_pd_cnt; i++) {
7018 pd = &g_func_pd[i];
7019 if (pd->type == OPT_OFFSET) {
7020 for (j = 0; j < pd->count; j++)
7021 free(pd->d[j].u.label);
7022 }
7023 free(pd->d);
7024 pd->d = NULL;
7025 }
7026 g_func_pd_cnt = 0;
7027 pd = NULL;
46b388c2 7028
7029 if (end)
7030 break;
4c45fa73 7031 if (wordc == 0)
7032 continue;
7033 }
7034
91977a1c 7035 if (IS(words[1], "proc")) {
7036 if (in_func)
7037 aerr("proc '%s' while in_func '%s'?\n",
7038 words[0], g_func);
bfa4a6ee 7039 p = words[0];
ddaf8bd7 7040 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
30c8c549 7041 g_skip_func = 1;
91977a1c 7042 strcpy(g_func, words[0]);
d4e3b5db 7043 set_label(0, words[0]);
91977a1c 7044 in_func = 1;
7045 continue;
7046 }
7047
e56ab892 7048 if (IS(words[1], "endp"))
7049 {
91977a1c 7050 if (!in_func)
7051 aerr("endp '%s' while not in_func?\n", words[0]);
7052 if (!IS(g_func, words[0]))
7053 aerr("endp '%s' while in_func '%s'?\n",
7054 words[0], g_func);
bfa4a6ee 7055
ddaf8bd7 7056 if ((g_ida_func_attr & IDAFA_THUNK) && pi == 1
7057 && ops[0].op == OP_JMP && ops[0].operand[0].had_ds)
7058 {
7059 // import jump
30c8c549 7060 g_skip_func = 1;
ddaf8bd7 7061 }
7062
30c8c549 7063 if (!g_skip_func && func_chunks_used) {
e56ab892 7064 // start processing chunks
7065 struct chunk_item *ci, key = { g_func, 0 };
7066
7067 func_chunk_ret = ftell(fasm);
de50b98b 7068 func_chunk_ret_ln = asmln;
e56ab892 7069 if (!func_chunks_sorted) {
7070 qsort(func_chunks, func_chunk_cnt,
7071 sizeof(func_chunks[0]), cmp_chunks);
7072 func_chunks_sorted = 1;
7073 }
7074 ci = bsearch(&key, func_chunks, func_chunk_cnt,
7075 sizeof(func_chunks[0]), cmp_chunks);
7076 if (ci == NULL)
7077 aerr("'%s' needs chunks, but none found\n", g_func);
7078 func_chunk_i = ci - func_chunks;
7079 for (; func_chunk_i > 0; func_chunk_i--)
7080 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
7081 break;
7082
7083 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
7084 if (ret)
7085 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
de50b98b 7086 asmln = func_chunks[func_chunk_i].asmln;
e56ab892 7087 func_chunk_i++;
7088 continue;
7089 }
4c45fa73 7090 pending_endp = 1;
91977a1c 7091 continue;
7092 }
7093
1f84f6b3 7094 if (wordc == 2 && IS(words[1], "ends")) {
46b388c2 7095 if (!multi_seg) {
7096 end = 1;
7097 if (pending_endp)
7098 goto do_pending_endp;
1f84f6b3 7099 break;
46b388c2 7100 }
1f84f6b3 7101
7102 // scan for next text segment
5fa1256f 7103 while (my_fgets(line, sizeof(line), fasm)) {
1f84f6b3 7104 asmln++;
7105 p = sskip(line);
7106 if (*p == 0 || *p == ';')
7107 continue;
7108
7109 if (strstr(p, "segment para public 'CODE' use32"))
7110 break;
7111 }
7112
7113 continue;
7114 }
a2c1d768 7115
bfa4a6ee 7116 p = strchr(words[0], ':');
7117 if (p != NULL) {
d4e3b5db 7118 set_label(pi, words[0]);
bfa4a6ee 7119 continue;
7120 }
7121
30c8c549 7122 if (!in_func || g_skip_func) {
7123 if (!skip_warned && !g_skip_func && g_labels[pi] != NULL) {
bfa4a6ee 7124 if (verbose)
7125 anote("skipping from '%s'\n", g_labels[pi]);
7126 skip_warned = 1;
7127 }
d7857c3a 7128 free(g_labels[pi]);
7129 g_labels[pi] = NULL;
bfa4a6ee 7130 continue;
7131 }
7132
ddaf8bd7 7133 if (wordc > 1 && IS(words[1], "="))
7134 {
91977a1c 7135 if (wordc != 5)
7136 aerr("unhandled equ, wc=%d\n", wordc);
7137 if (g_eqcnt >= eq_alloc) {
7138 eq_alloc *= 2;
7139 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
7140 my_assert_not(g_eqs, NULL);
7141 }
7142
7143 len = strlen(words[0]);
7144 if (len > sizeof(g_eqs[0].name) - 1)
7145 aerr("equ name too long: %d\n", len);
7146 strcpy(g_eqs[g_eqcnt].name, words[0]);
7147
7148 if (!IS(words[3], "ptr"))
7149 aerr("unhandled equ\n");
7150 if (IS(words[2], "dword"))
7151 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
7152 else if (IS(words[2], "word"))
7153 g_eqs[g_eqcnt].lmod = OPLM_WORD;
7154 else if (IS(words[2], "byte"))
7155 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
90307a99 7156 else if (IS(words[2], "qword"))
7157 g_eqs[g_eqcnt].lmod = OPLM_QWORD;
91977a1c 7158 else
7159 aerr("bad lmod: '%s'\n", words[2]);
7160
7161 g_eqs[g_eqcnt].offset = parse_number(words[4]);
7162 g_eqcnt++;
7163 continue;
7164 }
7165
7166 if (pi >= ARRAY_SIZE(ops))
7167 aerr("too many ops\n");
7168
91977a1c 7169 parse_op(&ops[pi], words, wordc);
ddaf8bd7 7170
7171 if (sctproto != NULL) {
8eb12e72 7172 if (ops[pi].op == OP_CALL || ops[pi].op == OP_JMP)
ddaf8bd7 7173 ops[pi].datap = sctproto;
7174 sctproto = NULL;
7175 }
91977a1c 7176 pi++;
91977a1c 7177 }
7178
9af2d373 7179 if (g_header_mode)
92d715b6 7180 output_hdr(fout);
7181
91977a1c 7182 fclose(fout);
7183 fclose(fasm);
06c5d854 7184 fclose(g_fhdr);
91977a1c 7185
7186 return 0;
c36e914d 7187}
91977a1c 7188
7189// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.