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