notaz.gp2x.de / ia32rtools.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
translate: initial float support
[ia32rtools.git] / tools / translate.c
... / ...
CommitLineData
1/*
2 * ia32rtools
3 * (C) notaz, 2013-2015
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
9#define _GNU_SOURCE
10#include <stdio.h>
11#include <stdlib.h>
12#include <string.h>
13
14#include "my_assert.h"
15#include "my_str.h"
16#include "common.h"
17
18#define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
19#define IS(w, y) !strcmp(w, y)
20#define IS_START(w, y) !strncmp(w, y, strlen(y))
21
22#include "protoparse.h"
23
24static const char *asmfn;
25static int asmln;
26static FILE *g_fhdr;
27
28#define anote(fmt, ...) \
29 printf("%s:%d: note: " fmt, asmfn, asmln, ##__VA_ARGS__)
30#define awarn(fmt, ...) \
31 printf("%s:%d: warning: " fmt, asmfn, asmln, ##__VA_ARGS__)
32#define aerr(fmt, ...) do { \
33 printf("%s:%d: error: " fmt, asmfn, asmln, ##__VA_ARGS__); \
34 fcloseall(); \
35 exit(1); \
36} while (0)
37
38#include "masm_tools.h"
39
40enum op_flags {
41 OPF_RMD = (1 << 0), /* removed from code generation */
42 OPF_DATA = (1 << 1), /* data processing - writes to dst opr */
43 OPF_FLAGS = (1 << 2), /* sets flags */
44 OPF_JMP = (1 << 3), /* branch, call */
45 OPF_CJMP = (1 << 4), /* cond. branch (cc or jecxz/loop) */
46 OPF_CC = (1 << 5), /* uses flags */
47 OPF_TAIL = (1 << 6), /* ret or tail call */
48 OPF_RSAVE = (1 << 7), /* push/pop is local reg save/load */
49 OPF_REP = (1 << 8), /* prefixed by rep */
50 OPF_REPZ = (1 << 9), /* rep is repe/repz */
51 OPF_REPNZ = (1 << 10), /* rep is repne/repnz */
52 OPF_FARG = (1 << 11), /* push collected as func arg */
53 OPF_FARGNR = (1 << 12), /* push collected as func arg (no reuse) */
54 OPF_EBP_S = (1 << 13), /* ebp used as scratch here, not BP */
55 OPF_DF = (1 << 14), /* DF flag set */
56 OPF_ATAIL = (1 << 15), /* tail call with reused arg frame */
57 OPF_32BIT = (1 << 16), /* 32bit division */
58 OPF_LOCK = (1 << 17), /* op has lock prefix */
59 OPF_VAPUSH = (1 << 18), /* vararg ptr push (as call arg) */
60 OPF_DONE = (1 << 19), /* already fully handled by analysis */
61 OPF_PPUSH = (1 << 20), /* part of complex push-pop graph */
62 OPF_NOREGS = (1 << 21), /* don't track regs of this op */
63 OPF_FPUSH = (1 << 22), /* pushes x87 stack */
64 OPF_FPOP = (1 << 23), /* pops x87 stack */
65 OPF_FSHIFT = (1 << 24), /* x87 stack shift is actually needed */
66};
67
68enum op_op {
69 OP_INVAL,
70 OP_NOP,
71 OP_PUSH,
72 OP_POP,
73 OP_LEAVE,
74 OP_MOV,
75 OP_LEA,
76 OP_MOVZX,
77 OP_MOVSX,
78 OP_XCHG,
79 OP_NOT,
80 OP_XLAT,
81 OP_CDQ,
82 OP_LODS,
83 OP_STOS,
84 OP_MOVS,
85 OP_CMPS,
86 OP_SCAS,
87 OP_STD,
88 OP_CLD,
89 OP_RET,
90 OP_ADD,
91 OP_SUB,
92 OP_AND,
93 OP_OR,
94 OP_XOR,
95 OP_SHL,
96 OP_SHR,
97 OP_SAR,
98 OP_SHLD,
99 OP_SHRD,
100 OP_ROL,
101 OP_ROR,
102 OP_RCL,
103 OP_RCR,
104 OP_ADC,
105 OP_SBB,
106 OP_BSF,
107 OP_INC,
108 OP_DEC,
109 OP_NEG,
110 OP_MUL,
111 OP_IMUL,
112 OP_DIV,
113 OP_IDIV,
114 OP_TEST,
115 OP_CMP,
116 OP_CALL,
117 OP_JMP,
118 OP_JECXZ,
119 OP_LOOP,
120 OP_JCC,
121 OP_SCC,
122 // x87
123 OP_FLD,
124 OP_FILD,
125 OP_FLDc,
126 OP_FST,
127 OP_FADD,
128 OP_FDIV,
129 OP_FMUL,
130 OP_FSUB,
131 OP_FDIVR,
132 OP_FSUBR,
133 OP_FIADD,
134 OP_FIDIV,
135 OP_FIMUL,
136 OP_FISUB,
137 OP_FIDIVR,
138 OP_FISUBR,
139 // mmx
140 OP_EMMS,
141 // pseudo-ops for lib calls
142 OPP_FTOL,
143 // undefined
144 OP_UD2,
145};
146
147enum opr_type {
148 OPT_UNSPEC,
149 OPT_REG,
150 OPT_REGMEM,
151 OPT_LABEL,
152 OPT_OFFSET,
153 OPT_CONST,
154};
155
156// must be sorted (larger len must be further in enum)
157enum opr_lenmod {
158 OPLM_UNSPEC,
159 OPLM_BYTE,
160 OPLM_WORD,
161 OPLM_DWORD,
162 OPLM_QWORD,
163};
164
165#define MAX_EXITS 128
166
167#define MAX_OPERANDS 3
168#define NAMELEN 112
169
170#define OPR_INIT(type_, lmod_, reg_) \
171 { type_, lmod_, reg_, }
172
173struct parsed_opr {
174 enum opr_type type;
175 enum opr_lenmod lmod;
176 int reg;
177 unsigned int is_ptr:1; // pointer in C
178 unsigned int is_array:1; // array in C
179 unsigned int type_from_var:1; // .. in header, sometimes wrong
180 unsigned int size_mismatch:1; // type override differs from C
181 unsigned int size_lt:1; // type override is larger than C
182 unsigned int had_ds:1; // had ds: prefix
183 const struct parsed_proto *pp; // for OPT_LABEL
184 unsigned int val;
185 char name[NAMELEN];
186};
187
188struct parsed_op {
189 enum op_op op;
190 struct parsed_opr operand[MAX_OPERANDS];
191 unsigned int flags;
192 unsigned char pfo;
193 unsigned char pfo_inv;
194 unsigned char operand_cnt;
195 unsigned char p_argnum; // arg push: altered before call arg #
196 unsigned char p_arggrp; // arg push: arg group # for above
197 unsigned char p_argpass;// arg push: arg of host func
198 short p_argnext;// arg push: same arg pushed elsewhere or -1
199 int regmask_src; // all referensed regs
200 int regmask_dst;
201 int pfomask; // flagop: parsed_flag_op that can't be delayed
202 int cc_scratch; // scratch storage during analysis
203 int bt_i; // branch target for branches
204 struct parsed_data *btj;// branch targets for jumptables
205 struct parsed_proto *pp;// parsed_proto for OP_CALL
206 void *datap;
207 int asmln;
208};
209
210// datap:
211// OP_CALL - parser proto hint (str)
212// (OPF_CC) - points to one of (OPF_FLAGS) that affects cc op
213// OP_PUSH - points to OP_POP in complex push/pop graph
214// OP_POP - points to OP_PUSH in simple push/pop pair
215
216struct parsed_equ {
217 char name[64];
218 enum opr_lenmod lmod;
219 int offset;
220};
221
222struct parsed_data {
223 char label[256];
224 enum opr_type type;
225 enum opr_lenmod lmod;
226 int count;
227 int count_alloc;
228 struct {
229 union {
230 char *label;
231 unsigned int val;
232 } u;
233 int bt_i;
234 } *d;
235};
236
237struct label_ref {
238 int i;
239 struct label_ref *next;
240};
241
242enum ida_func_attr {
243 IDAFA_BP_FRAME = (1 << 0),
244 IDAFA_LIB_FUNC = (1 << 1),
245 IDAFA_STATIC = (1 << 2),
246 IDAFA_NORETURN = (1 << 3),
247 IDAFA_THUNK = (1 << 4),
248 IDAFA_FPD = (1 << 5),
249};
250
251enum x87_const {
252 X87_CONST_1 = 1,
253 X87_CONST_2T,
254 X87_CONST_2E,
255 X87_CONST_PI,
256 X87_CONST_LG2,
257 X87_CONST_LN2,
258 X87_CONST_Z,
259};
260
261// note: limited to 32k due to p_argnext
262#define MAX_OPS 4096
263#define MAX_ARG_GRP 2
264
265static struct parsed_op ops[MAX_OPS];
266static struct parsed_equ *g_eqs;
267static int g_eqcnt;
268static char *g_labels[MAX_OPS];
269static struct label_ref g_label_refs[MAX_OPS];
270static const struct parsed_proto *g_func_pp;
271static struct parsed_data *g_func_pd;
272static int g_func_pd_cnt;
273static int g_func_lmods;
274static char g_func[256];
275static char g_comment[256];
276static int g_bp_frame;
277static int g_sp_frame;
278static int g_stack_frame_used;
279static int g_stack_fsz;
280static int g_ida_func_attr;
281static int g_skip_func;
282static int g_allow_regfunc;
283static int g_quiet_pp;
284static int g_header_mode;
285
286#define ferr(op_, fmt, ...) do { \
287 printf("%s:%d: error %u: [%s] '%s': " fmt, asmfn, (op_)->asmln, \
288 __LINE__, g_func, dump_op(op_), ##__VA_ARGS__); \
289 fcloseall(); \
290 exit(1); \
291} while (0)
292#define fnote(op_, fmt, ...) \
293 printf("%s:%d: note: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
294 dump_op(op_), ##__VA_ARGS__)
295
296#define ferr_assert(op_, cond) do { \
297 if (!(cond)) ferr(op_, "assertion '%s' failed\n", #cond); \
298} while (0)
299
300const char *regs_r32[] = {
301 "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp",
302 // not r32, but list here for easy parsing and printing
303 "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7",
304 "st", "st(1)", "st(2)", "st(3)", "st(4)", "st(5)", "st(6)", "st(7)"
305};
306const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" };
307const char *regs_r8l[] = { "al", "bl", "cl", "dl" };
308const char *regs_r8h[] = { "ah", "bh", "ch", "dh" };
309
310enum x86_regs {
311 xUNSPEC = -1,
312 xAX, xBX, xCX, xDX,
313 xSI, xDI, xBP, xSP,
314 xMM0, xMM1, xMM2, xMM3, // mmx
315 xMM4, xMM5, xMM6, xMM7,
316 xST0, xST1, xST2, xST3, // x87
317 xST4, xST5, xST6, xST7,
318};
319
320#define mxAX (1 << xAX)
321#define mxDX (1 << xDX)
322#define mxST0 (1 << xST0)
323#define mxST1 (1 << xST1)
324
325// possible basic comparison types (without inversion)
326enum parsed_flag_op {
327 PFO_O, // 0 OF=1
328 PFO_C, // 2 CF=1
329 PFO_Z, // 4 ZF=1
330 PFO_BE, // 6 CF=1||ZF=1
331 PFO_S, // 8 SF=1
332 PFO_P, // a PF=1
333 PFO_L, // c SF!=OF
334 PFO_LE, // e ZF=1||SF!=OF
335};
336
337#define PFOB_O (1 << PFO_O)
338#define PFOB_C (1 << PFO_C)
339#define PFOB_Z (1 << PFO_Z)
340#define PFOB_S (1 << PFO_S)
341
342static const char *parsed_flag_op_names[] = {
343 "o", "c", "z", "be", "s", "p", "l", "le"
344};
345
346static int char_array_i(const char *array[], size_t len, const char *s)
347{
348 int i;
349
350 for (i = 0; i < len; i++)
351 if (IS(s, array[i]))
352 return i;
353
354 return -1;
355}
356
357static void printf_number(char *buf, size_t buf_size,
358 unsigned long number)
359{
360 // output in C-friendly form
361 snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
362}
363
364static int check_segment_prefix(const char *s)
365{
366 if (s[0] == 0 || s[1] != 's' || s[2] != ':')
367 return 0;
368
369 switch (s[0]) {
370 case 'c': return 1;
371 case 'd': return 2;
372 case 's': return 3;
373 case 'e': return 4;
374 case 'f': return 5;
375 case 'g': return 6;
376 default: return 0;
377 }
378}
379
380static int parse_reg(enum opr_lenmod *reg_lmod, const char *s)
381{
382 int reg;
383
384 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s);
385 if (reg >= 8) {
386 *reg_lmod = OPLM_QWORD;
387 return reg;
388 }
389 if (reg >= 0) {
390 *reg_lmod = OPLM_DWORD;
391 return reg;
392 }
393 reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s);
394 if (reg >= 0) {
395 *reg_lmod = OPLM_WORD;
396 return reg;
397 }
398 reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s);
399 if (reg >= 0) {
400 *reg_lmod = OPLM_BYTE;
401 return reg;
402 }
403 reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s);
404 if (reg >= 0) {
405 *reg_lmod = OPLM_BYTE;
406 return reg;
407 }
408
409 return -1;
410}
411
412static int parse_indmode(char *name, int *regmask, int need_c_cvt)
413{
414 enum opr_lenmod lmod;
415 char cvtbuf[256];
416 char *d = cvtbuf;
417 char *s = name;
418 char w[64];
419 long number;
420 int reg;
421 int c = 0;
422
423 *d = 0;
424
425 while (*s != 0) {
426 d += strlen(d);
427 while (my_isblank(*s))
428 s++;
429 for (; my_issep(*s); d++, s++)
430 *d = *s;
431 while (my_isblank(*s))
432 s++;
433 *d = 0;
434
435 // skip '?s:' prefixes
436 if (check_segment_prefix(s))
437 s += 3;
438
439 s = next_idt(w, sizeof(w), s);
440 if (w[0] == 0)
441 break;
442 c++;
443
444 reg = parse_reg(&lmod, w);
445 if (reg >= 0) {
446 *regmask |= 1 << reg;
447 goto pass;
448 }
449
450 if ('0' <= w[0] && w[0] <= '9') {
451 number = parse_number(w);
452 printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number);
453 continue;
454 }
455
456 // probably some label/identifier - pass
457
458pass:
459 snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w);
460 }
461
462 if (need_c_cvt)
463 strcpy(name, cvtbuf);
464
465 return c;
466}
467
468static int is_reg_in_str(const char *s)
469{
470 int i;
471
472 if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
473 return 0;
474
475 for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
476 if (!strncmp(s, regs_r32[i], 3))
477 return 1;
478
479 return 0;
480}
481
482static const char *parse_stack_el(const char *name, char *extra_reg,
483 int early_try)
484{
485 const char *p, *p2, *s;
486 char *endp = NULL;
487 char buf[32];
488 long val;
489 int len;
490
491 if (g_bp_frame || early_try)
492 {
493 p = name;
494 if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
495 p += 4;
496 if (extra_reg != NULL) {
497 strncpy(extra_reg, name, 3);
498 extra_reg[4] = 0;
499 }
500 }
501
502 if (IS_START(p, "ebp+")) {
503 p += 4;
504
505 p2 = strchr(p, '+');
506 if (p2 != NULL && is_reg_in_str(p)) {
507 if (extra_reg != NULL) {
508 strncpy(extra_reg, p, p2 - p);
509 extra_reg[p2 - p] = 0;
510 }
511 p = p2 + 1;
512 }
513
514 if (!('0' <= *p && *p <= '9'))
515 return p;
516
517 return NULL;
518 }
519 }
520
521 if (!IS_START(name, "esp+"))
522 return NULL;
523
524 s = name + 4;
525 p = strchr(s, '+');
526 if (p) {
527 if (is_reg_in_str(s)) {
528 if (extra_reg != NULL) {
529 strncpy(extra_reg, s, p - s);
530 extra_reg[p - s] = 0;
531 }
532 s = p + 1;
533 p = strchr(s, '+');
534 if (p == NULL)
535 aerr("%s IDA stackvar not set?\n", __func__);
536 }
537 if (!('0' <= *s && *s <= '9')) {
538 aerr("%s IDA stackvar offset not set?\n", __func__);
539 return NULL;
540 }
541 if (s[0] == '0' && s[1] == 'x')
542 s += 2;
543 len = p - s;
544 if (len < sizeof(buf) - 1) {
545 strncpy(buf, s, len);
546 buf[len] = 0;
547 val = strtol(buf, &endp, 16);
548 if (val == 0 || *endp != 0) {
549 aerr("%s num parse fail for '%s'\n", __func__, buf);
550 return NULL;
551 }
552 }
553 p++;
554 }
555 else
556 p = name + 4;
557
558 if ('0' <= *p && *p <= '9')
559 return NULL;
560
561 return p;
562}
563
564static int guess_lmod_from_name(struct parsed_opr *opr)
565{
566 if (IS_START(opr->name, "dword_") || IS_START(opr->name, "off_")) {
567 opr->lmod = OPLM_DWORD;
568 return 1;
569 }
570 if (IS_START(opr->name, "word_")) {
571 opr->lmod = OPLM_WORD;
572 return 1;
573 }
574 if (IS_START(opr->name, "byte_")) {
575 opr->lmod = OPLM_BYTE;
576 return 1;
577 }
578 if (IS_START(opr->name, "qword_")) {
579 opr->lmod = OPLM_QWORD;
580 return 1;
581 }
582 return 0;
583}
584
585static int guess_lmod_from_c_type(enum opr_lenmod *lmod,
586 const struct parsed_type *c_type)
587{
588 static const char *dword_types[] = {
589 "uint32_t", "int", "_DWORD", "UINT_PTR", "DWORD",
590 "WPARAM", "LPARAM", "UINT", "__int32",
591 "LONG", "HIMC", "BOOL", "size_t",
592 "float",
593 };
594 static const char *word_types[] = {
595 "uint16_t", "int16_t", "_WORD", "WORD",
596 "unsigned __int16", "__int16",
597 };
598 static const char *byte_types[] = {
599 "uint8_t", "int8_t", "char",
600 "unsigned __int8", "__int8", "BYTE", "_BYTE",
601 "CHAR", "_UNKNOWN",
602 // structures.. deal the same as with _UNKNOWN for now
603 "CRITICAL_SECTION",
604 };
605 const char *n;
606 int i;
607
608 if (c_type->is_ptr) {
609 *lmod = OPLM_DWORD;
610 return 1;
611 }
612
613 n = skip_type_mod(c_type->name);
614
615 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
616 if (IS(n, dword_types[i])) {
617 *lmod = OPLM_DWORD;
618 return 1;
619 }
620 }
621
622 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
623 if (IS(n, word_types[i])) {
624 *lmod = OPLM_WORD;
625 return 1;
626 }
627 }
628
629 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
630 if (IS(n, byte_types[i])) {
631 *lmod = OPLM_BYTE;
632 return 1;
633 }
634 }
635
636 return 0;
637}
638
639static char *default_cast_to(char *buf, size_t buf_size,
640 struct parsed_opr *opr)
641{
642 buf[0] = 0;
643
644 if (!opr->is_ptr)
645 return buf;
646 if (opr->pp == NULL || opr->pp->type.name == NULL
647 || opr->pp->is_fptr)
648 {
649 snprintf(buf, buf_size, "%s", "(void *)");
650 return buf;
651 }
652
653 snprintf(buf, buf_size, "(%s)", opr->pp->type.name);
654 return buf;
655}
656
657static enum opr_type lmod_from_directive(const char *d)
658{
659 if (IS(d, "dd"))
660 return OPLM_DWORD;
661 else if (IS(d, "dw"))
662 return OPLM_WORD;
663 else if (IS(d, "db"))
664 return OPLM_BYTE;
665
666 aerr("unhandled directive: '%s'\n", d);
667 return OPLM_UNSPEC;
668}
669
670static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
671 int *regmask)
672{
673 opr->type = OPT_REG;
674 opr->reg = reg;
675 opr->lmod = lmod;
676 *regmask |= 1 << reg;
677}
678
679static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
680 int *extra_offs);
681
682static int parse_operand(struct parsed_opr *opr,
683 int *regmask, int *regmask_indirect,
684 char words[16][256], int wordc, int w, unsigned int op_flags)
685{
686 const struct parsed_proto *pp = NULL;
687 enum opr_lenmod tmplmod;
688 unsigned long number;
689 char buf[256];
690 int ret, len;
691 int wordc_in;
692 char *p;
693 int i;
694
695 if (w >= wordc)
696 aerr("parse_operand w %d, wordc %d\n", w, wordc);
697
698 opr->reg = xUNSPEC;
699
700 for (i = w; i < wordc; i++) {
701 len = strlen(words[i]);
702 if (words[i][len - 1] == ',') {
703 words[i][len - 1] = 0;
704 wordc = i + 1;
705 break;
706 }
707 }
708
709 wordc_in = wordc - w;
710
711 if ((op_flags & OPF_JMP) && wordc_in > 0
712 && !('0' <= words[w][0] && words[w][0] <= '9'))
713 {
714 const char *label = NULL;
715
716 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
717 && IS(words[w + 1], "ptr"))
718 label = words[w + 2];
719 else if (wordc_in == 2 && IS(words[w], "short"))
720 label = words[w + 1];
721 else if (wordc_in == 1
722 && strchr(words[w], '[') == NULL
723 && parse_reg(&tmplmod, words[w]) < 0)
724 label = words[w];
725
726 if (label != NULL) {
727 opr->type = OPT_LABEL;
728 ret = check_segment_prefix(label);
729 if (ret != 0) {
730 if (ret >= 5)
731 aerr("fs/gs used\n");
732 opr->had_ds = 1;
733 label += 3;
734 }
735 strcpy(opr->name, label);
736 return wordc;
737 }
738 }
739
740 if (wordc_in >= 3) {
741 if (IS(words[w + 1], "ptr")) {
742 if (IS(words[w], "dword"))
743 opr->lmod = OPLM_DWORD;
744 else if (IS(words[w], "word"))
745 opr->lmod = OPLM_WORD;
746 else if (IS(words[w], "byte"))
747 opr->lmod = OPLM_BYTE;
748 else if (IS(words[w], "qword"))
749 opr->lmod = OPLM_QWORD;
750 else
751 aerr("type parsing failed\n");
752 w += 2;
753 wordc_in = wordc - w;
754 }
755 }
756
757 if (wordc_in == 2) {
758 if (IS(words[w], "offset")) {
759 opr->type = OPT_OFFSET;
760 opr->lmod = OPLM_DWORD;
761 strcpy(opr->name, words[w + 1]);
762 pp = proto_parse(g_fhdr, opr->name, 1);
763 goto do_label;
764 }
765 if (IS(words[w], "(offset")) {
766 p = strchr(words[w + 1], ')');
767 if (p == NULL)
768 aerr("parse of bracketed offset failed\n");
769 *p = 0;
770 opr->type = OPT_OFFSET;
771 strcpy(opr->name, words[w + 1]);
772 return wordc;
773 }
774 }
775
776 if (wordc_in != 1)
777 aerr("parse_operand 1 word expected\n");
778
779 ret = check_segment_prefix(words[w]);
780 if (ret != 0) {
781 if (ret >= 5)
782 aerr("fs/gs used\n");
783 opr->had_ds = 1;
784 memmove(words[w], words[w] + 3, strlen(words[w]) - 2);
785 }
786 strcpy(opr->name, words[w]);
787
788 if (words[w][0] == '[') {
789 opr->type = OPT_REGMEM;
790 ret = sscanf(words[w], "[%[^]]]", opr->name);
791 if (ret != 1)
792 aerr("[] parse failure\n");
793
794 parse_indmode(opr->name, regmask_indirect, 1);
795 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL, 1))
796 {
797 // might be an equ
798 struct parsed_equ *eq =
799 equ_find(NULL, parse_stack_el(opr->name, NULL, 1), &i);
800 if (eq)
801 opr->lmod = eq->lmod;
802
803 // might be unaligned access
804 g_func_lmods |= 1 << OPLM_BYTE;
805 }
806 return wordc;
807 }
808 else if (strchr(words[w], '[')) {
809 // label[reg] form
810 p = strchr(words[w], '[');
811 opr->type = OPT_REGMEM;
812 parse_indmode(p, regmask_indirect, 0);
813 strncpy(buf, words[w], p - words[w]);
814 buf[p - words[w]] = 0;
815 pp = proto_parse(g_fhdr, buf, 1);
816 goto do_label;
817 }
818 else if (('0' <= words[w][0] && words[w][0] <= '9')
819 || words[w][0] == '-')
820 {
821 number = parse_number(words[w]);
822 opr->type = OPT_CONST;
823 opr->val = number;
824 printf_number(opr->name, sizeof(opr->name), number);
825 return wordc;
826 }
827
828 ret = parse_reg(&tmplmod, opr->name);
829 if (ret >= 0) {
830 setup_reg_opr(opr, ret, tmplmod, regmask);
831 return wordc;
832 }
833
834 // most likely var in data segment
835 opr->type = OPT_LABEL;
836 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
837
838do_label:
839 if (pp != NULL) {
840 if (pp->is_fptr || pp->is_func) {
841 opr->lmod = OPLM_DWORD;
842 opr->is_ptr = 1;
843 }
844 else {
845 tmplmod = OPLM_UNSPEC;
846 if (!guess_lmod_from_c_type(&tmplmod, &pp->type))
847 anote("unhandled C type '%s' for '%s'\n",
848 pp->type.name, opr->name);
849
850 if (opr->lmod == OPLM_UNSPEC) {
851 opr->lmod = tmplmod;
852 opr->type_from_var = 1;
853 }
854 else if (opr->lmod != tmplmod) {
855 opr->size_mismatch = 1;
856 if (tmplmod < opr->lmod)
857 opr->size_lt = 1;
858 }
859 opr->is_ptr = pp->type.is_ptr;
860 }
861 opr->is_array = pp->type.is_array;
862 }
863 opr->pp = pp;
864
865 if (opr->lmod == OPLM_UNSPEC)
866 guess_lmod_from_name(opr);
867 return wordc;
868}
869
870static const struct {
871 const char *name;
872 unsigned int flags;
873} pref_table[] = {
874 { "rep", OPF_REP },
875 { "repe", OPF_REP|OPF_REPZ },
876 { "repz", OPF_REP|OPF_REPZ },
877 { "repne", OPF_REP|OPF_REPNZ },
878 { "repnz", OPF_REP|OPF_REPNZ },
879 { "lock", OPF_LOCK }, // ignored for now..
880};
881
882#define OPF_CJMP_CC (OPF_JMP|OPF_CJMP|OPF_CC)
883
884static const struct {
885 const char *name;
886 enum op_op op;
887 unsigned short minopr;
888 unsigned short maxopr;
889 unsigned int flags;
890 unsigned char pfo;
891 unsigned char pfo_inv;
892} op_table[] = {
893 { "nop", OP_NOP, 0, 0, 0 },
894 { "push", OP_PUSH, 1, 1, 0 },
895 { "pop", OP_POP, 1, 1, OPF_DATA },
896 { "leave",OP_LEAVE, 0, 0, OPF_DATA },
897 { "mov" , OP_MOV, 2, 2, OPF_DATA },
898 { "lea", OP_LEA, 2, 2, OPF_DATA },
899 { "movzx",OP_MOVZX, 2, 2, OPF_DATA },
900 { "movsx",OP_MOVSX, 2, 2, OPF_DATA },
901 { "xchg", OP_XCHG, 2, 2, OPF_DATA },
902 { "not", OP_NOT, 1, 1, OPF_DATA },
903 { "xlat", OP_XLAT, 0, 0, OPF_DATA },
904 { "cdq", OP_CDQ, 0, 0, OPF_DATA },
905 { "lodsb",OP_LODS, 0, 0, OPF_DATA },
906 { "lodsw",OP_LODS, 0, 0, OPF_DATA },
907 { "lodsd",OP_LODS, 0, 0, OPF_DATA },
908 { "stosb",OP_STOS, 0, 0, OPF_DATA },
909 { "stosw",OP_STOS, 0, 0, OPF_DATA },
910 { "stosd",OP_STOS, 0, 0, OPF_DATA },
911 { "movsb",OP_MOVS, 0, 0, OPF_DATA },
912 { "movsw",OP_MOVS, 0, 0, OPF_DATA },
913 { "movsd",OP_MOVS, 0, 0, OPF_DATA },
914 { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
915 { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
916 { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
917 { "scasb",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
918 { "scasw",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
919 { "scasd",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
920 { "std", OP_STD, 0, 0, OPF_DATA }, // special flag
921 { "cld", OP_CLD, 0, 0, OPF_DATA },
922 { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
923 { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
924 { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
925 { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS },
926 { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS },
927 { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
928 { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
929 { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
930 { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
931 { "shld", OP_SHLD, 3, 3, OPF_DATA|OPF_FLAGS },
932 { "shrd", OP_SHRD, 3, 3, OPF_DATA|OPF_FLAGS },
933 { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
934 { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
935 { "rcl", OP_RCL, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
936 { "rcr", OP_RCR, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
937 { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
938 { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
939 { "bsf", OP_BSF, 2, 2, OPF_DATA|OPF_FLAGS },
940 { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
941 { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS },
942 { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS },
943 { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS },
944 { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS },
945 { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS },
946 { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS },
947 { "test", OP_TEST, 2, 2, OPF_FLAGS },
948 { "cmp", OP_CMP, 2, 2, OPF_FLAGS },
949 { "retn", OP_RET, 0, 1, OPF_TAIL },
950 { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS },
951 { "jmp", OP_JMP, 1, 1, OPF_JMP },
952 { "jecxz",OP_JECXZ, 1, 1, OPF_JMP|OPF_CJMP },
953 { "loop", OP_LOOP, 1, 1, OPF_JMP|OPF_CJMP|OPF_DATA },
954 { "jo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 0 }, // 70 OF=1
955 { "jno", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 1 }, // 71 OF=0
956 { "jc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72 CF=1
957 { "jb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72
958 { "jnc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73 CF=0
959 { "jnb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
960 { "jae", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
961 { "jz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74 ZF=1
962 { "je", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74
963 { "jnz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75 ZF=0
964 { "jne", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75
965 { "jbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76 CF=1||ZF=1
966 { "jna", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76
967 { "ja", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77 CF=0&&ZF=0
968 { "jnbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77
969 { "js", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 0 }, // 78 SF=1
970 { "jns", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 1 }, // 79 SF=0
971 { "jp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a PF=1
972 { "jpe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a
973 { "jnp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b PF=0
974 { "jpo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b
975 { "jl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c SF!=OF
976 { "jnge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c
977 { "jge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d SF=OF
978 { "jnl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d
979 { "jle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e ZF=1||SF!=OF
980 { "jng", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e
981 { "jg", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f ZF=0&&SF=OF
982 { "jnle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f
983 { "seto", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 0 },
984 { "setno", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 1 },
985 { "setc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
986 { "setb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
987 { "setnc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
988 { "setae", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
989 { "setnb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
990 { "setz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
991 { "sete", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
992 { "setnz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
993 { "setne", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
994 { "setbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
995 { "setna", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
996 { "seta", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
997 { "setnbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
998 { "sets", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 0 },
999 { "setns", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 1 },
1000 { "setp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
1001 { "setpe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
1002 { "setnp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
1003 { "setpo", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
1004 { "setl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
1005 { "setnge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
1006 { "setge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
1007 { "setnl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
1008 { "setle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
1009 { "setng", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
1010 { "setg", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
1011 { "setnle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
1012 // x87
1013 { "fld", OP_FLD, 1, 1, OPF_FPUSH },
1014 { "fild", OP_FILD, 1, 1, OPF_FPUSH },
1015 { "fld1", OP_FLDc, 0, 0, OPF_FPUSH },
1016 { "fldz", OP_FLDc, 0, 0, OPF_FPUSH },
1017 { "fstp", OP_FST, 1, 1, OPF_FPOP },
1018 { "fst", OP_FST, 1, 1, 0 },
1019 { "fadd", OP_FADD, 0, 2, 0 },
1020 { "faddp", OP_FADD, 0, 2, OPF_FPOP },
1021 { "fdiv", OP_FDIV, 0, 2, 0 },
1022 { "fdivp", OP_FDIV, 0, 2, OPF_FPOP },
1023 { "fmul", OP_FMUL, 0, 2, 0 },
1024 { "fmulp", OP_FMUL, 0, 2, OPF_FPOP },
1025 { "fsub", OP_FSUB, 0, 2, 0 },
1026 { "fsubp", OP_FSUB, 0, 2, OPF_FPOP },
1027 { "fdivr", OP_FDIVR, 0, 2, 0 },
1028 { "fdivrp", OP_FDIVR, 0, 2, OPF_FPOP },
1029 { "fsubr", OP_FSUBR, 0, 2, 0 },
1030 { "fsubrp", OP_FSUBR, 0, 2, OPF_FPOP },
1031 { "fiadd", OP_FIADD, 1, 1, 0 },
1032 { "fidiv", OP_FIDIV, 1, 1, 0 },
1033 { "fimul", OP_FIMUL, 1, 1, 0 },
1034 { "fisub", OP_FISUB, 1, 1, 0 },
1035 { "fidivr", OP_FIDIVR, 1, 1, 0 },
1036 { "fisubr", OP_FISUBR, 1, 1, 0 },
1037 // mmx
1038 { "emms", OP_EMMS, 0, 0, OPF_DATA },
1039 { "movq", OP_MOV, 2, 2, OPF_DATA },
1040 // pseudo-ops for lib calls
1041 { "_ftol", OPP_FTOL },
1042 // must be last
1043 { "ud2", OP_UD2 },
1044};
1045
1046static void parse_op(struct parsed_op *op, char words[16][256], int wordc)
1047{
1048 enum opr_lenmod lmod = OPLM_UNSPEC;
1049 int prefix_flags = 0;
1050 int regmask_ind;
1051 int regmask;
1052 int op_w = 0;
1053 int opr = 0;
1054 int w = 0;
1055 int i, j;
1056
1057 for (i = 0; i < ARRAY_SIZE(pref_table); i++) {
1058 if (IS(words[w], pref_table[i].name)) {
1059 prefix_flags = pref_table[i].flags;
1060 break;
1061 }
1062 }
1063
1064 if (prefix_flags) {
1065 if (wordc <= 1)
1066 aerr("lone prefix: '%s'\n", words[0]);
1067 w++;
1068 }
1069
1070 op_w = w;
1071 for (i = 0; i < ARRAY_SIZE(op_table); i++) {
1072 if (IS(words[w], op_table[i].name))
1073 break;
1074 }
1075
1076 if (i == ARRAY_SIZE(op_table)) {
1077 if (!g_skip_func)
1078 aerr("unhandled op: '%s'\n", words[0]);
1079 i--; // OP_UD2
1080 }
1081 w++;
1082
1083 op->op = op_table[i].op;
1084 op->flags = op_table[i].flags | prefix_flags;
1085 op->pfo = op_table[i].pfo;
1086 op->pfo_inv = op_table[i].pfo_inv;
1087 op->regmask_src = op->regmask_dst = 0;
1088 op->asmln = asmln;
1089
1090 if (op->op == OP_UD2)
1091 return;
1092
1093 for (opr = 0; opr < op_table[i].maxopr; opr++) {
1094 if (opr >= op_table[i].minopr && w >= wordc)
1095 break;
1096
1097 regmask = regmask_ind = 0;
1098 w = parse_operand(&op->operand[opr], &regmask, &regmask_ind,
1099 words, wordc, w, op->flags);
1100
1101 if (opr == 0 && (op->flags & OPF_DATA))
1102 op->regmask_dst = regmask;
1103 else
1104 op->regmask_src |= regmask;
1105 op->regmask_src |= regmask_ind;
1106
1107 if (op->operand[opr].lmod != OPLM_UNSPEC)
1108 g_func_lmods |= 1 << op->operand[opr].lmod;
1109 }
1110
1111 if (w < wordc)
1112 aerr("parse_op %s incomplete: %d/%d\n",
1113 words[0], w, wordc);
1114
1115 // special cases
1116 op->operand_cnt = opr;
1117 if (!strncmp(op_table[i].name, "set", 3))
1118 op->operand[0].lmod = OPLM_BYTE;
1119
1120 switch (op->op) {
1121 // first operand is not dst
1122 case OP_CMP:
1123 case OP_TEST:
1124 op->regmask_src |= op->regmask_dst;
1125 op->regmask_dst = 0;
1126 break;
1127
1128 // first operand is src too
1129 case OP_NOT:
1130 case OP_ADD:
1131 case OP_AND:
1132 case OP_OR:
1133 case OP_RCL:
1134 case OP_RCR:
1135 case OP_ADC:
1136 case OP_INC:
1137 case OP_DEC:
1138 case OP_NEG:
1139 // more below..
1140 op->regmask_src |= op->regmask_dst;
1141 break;
1142
1143 // special
1144 case OP_XCHG:
1145 op->regmask_src |= op->regmask_dst;
1146 op->regmask_dst |= op->regmask_src;
1147 goto check_align;
1148
1149 case OP_SUB:
1150 case OP_SBB:
1151 case OP_XOR:
1152 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1153 && op->operand[0].lmod == op->operand[1].lmod
1154 && op->operand[0].reg == op->operand[1].reg
1155 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
1156 {
1157 op->regmask_src = 0;
1158 }
1159 else
1160 op->regmask_src |= op->regmask_dst;
1161 break;
1162
1163 // ops with implicit argumets
1164 case OP_XLAT:
1165 op->operand_cnt = 2;
1166 setup_reg_opr(&op->operand[0], xAX, OPLM_BYTE, &op->regmask_src);
1167 op->regmask_dst = op->regmask_src;
1168 setup_reg_opr(&op->operand[1], xBX, OPLM_DWORD, &op->regmask_src);
1169 break;
1170
1171 case OP_CDQ:
1172 op->operand_cnt = 2;
1173 setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst);
1174 setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src);
1175 break;
1176
1177 case OP_LODS:
1178 case OP_STOS:
1179 case OP_SCAS:
1180 if (words[op_w][4] == 'b')
1181 lmod = OPLM_BYTE;
1182 else if (words[op_w][4] == 'w')
1183 lmod = OPLM_WORD;
1184 else if (words[op_w][4] == 'd')
1185 lmod = OPLM_DWORD;
1186 j = 0;
1187 op->regmask_src = 0;
1188 setup_reg_opr(&op->operand[j++], op->op == OP_LODS ? xSI : xDI,
1189 OPLM_DWORD, &op->regmask_src);
1190 op->regmask_dst = op->regmask_src;
1191 setup_reg_opr(&op->operand[j++], xAX, lmod,
1192 op->op == OP_LODS ? &op->regmask_dst : &op->regmask_src);
1193 if (op->flags & OPF_REP) {
1194 setup_reg_opr(&op->operand[j++], xCX, OPLM_DWORD, &op->regmask_src);
1195 op->regmask_dst |= 1 << xCX;
1196 }
1197 op->operand_cnt = j;
1198 break;
1199
1200 case OP_MOVS:
1201 case OP_CMPS:
1202 if (words[op_w][4] == 'b')
1203 lmod = OPLM_BYTE;
1204 else if (words[op_w][4] == 'w')
1205 lmod = OPLM_WORD;
1206 else if (words[op_w][4] == 'd')
1207 lmod = OPLM_DWORD;
1208 j = 0;
1209 op->regmask_src = 0;
1210 // note: lmod is not correct, don't have where to place it
1211 setup_reg_opr(&op->operand[j++], xDI, lmod, &op->regmask_src);
1212 setup_reg_opr(&op->operand[j++], xSI, OPLM_DWORD, &op->regmask_src);
1213 if (op->flags & OPF_REP)
1214 setup_reg_opr(&op->operand[j++], xCX, OPLM_DWORD, &op->regmask_src);
1215 op->operand_cnt = j;
1216 op->regmask_dst = op->regmask_src;
1217 break;
1218
1219 case OP_LOOP:
1220 op->regmask_dst = 1 << xCX;
1221 // fallthrough
1222 case OP_JECXZ:
1223 op->operand_cnt = 2;
1224 op->regmask_src = 1 << xCX;
1225 op->operand[1].type = OPT_REG;
1226 op->operand[1].reg = xCX;
1227 op->operand[1].lmod = OPLM_DWORD;
1228 break;
1229
1230 case OP_IMUL:
1231 if (op->operand_cnt == 2) {
1232 if (op->operand[0].type != OPT_REG)
1233 aerr("reg expected\n");
1234 op->regmask_src |= 1 << op->operand[0].reg;
1235 }
1236 if (op->operand_cnt != 1)
1237 break;
1238 // fallthrough
1239 case OP_MUL:
1240 // singleop mul
1241 op->regmask_src |= op->regmask_dst;
1242 op->regmask_dst = (1 << xDX) | (1 << xAX);
1243 if (op->operand[0].lmod == OPLM_UNSPEC)
1244 op->operand[0].lmod = OPLM_DWORD;
1245 break;
1246
1247 case OP_DIV:
1248 case OP_IDIV:
1249 // we could set up operands for edx:eax, but there is no real need to
1250 // (see is_opr_modified())
1251 op->regmask_src |= op->regmask_dst;
1252 op->regmask_dst = (1 << xDX) | (1 << xAX);
1253 if (op->operand[0].lmod == OPLM_UNSPEC)
1254 op->operand[0].lmod = OPLM_DWORD;
1255 break;
1256
1257 case OP_SHL:
1258 case OP_SHR:
1259 case OP_SAR:
1260 case OP_ROL:
1261 case OP_ROR:
1262 op->regmask_src |= op->regmask_dst;
1263 if (op->operand[1].lmod == OPLM_UNSPEC)
1264 op->operand[1].lmod = OPLM_BYTE;
1265 break;
1266
1267 case OP_SHLD:
1268 case OP_SHRD:
1269 op->regmask_src |= op->regmask_dst;
1270 if (op->operand[2].lmod == OPLM_UNSPEC)
1271 op->operand[2].lmod = OPLM_BYTE;
1272 break;
1273
1274 case OP_PUSH:
1275 op->regmask_src |= op->regmask_dst;
1276 op->regmask_dst = 0;
1277 if (op->operand[0].lmod == OPLM_UNSPEC
1278 && (op->operand[0].type == OPT_CONST
1279 || op->operand[0].type == OPT_OFFSET
1280 || op->operand[0].type == OPT_LABEL))
1281 op->operand[0].lmod = OPLM_DWORD;
1282 break;
1283
1284 // alignment
1285 case OP_MOV:
1286 check_align:
1287 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1288 && op->operand[0].lmod == op->operand[1].lmod
1289 && op->operand[0].reg == op->operand[1].reg
1290 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
1291 {
1292 op->flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
1293 op->regmask_src = op->regmask_dst = 0;
1294 }
1295 break;
1296
1297 case OP_LEA:
1298 if (op->operand[0].type == OPT_REG
1299 && op->operand[1].type == OPT_REGMEM)
1300 {
1301 char buf[16];
1302 snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name);
1303 if (IS(buf, op->operand[1].name))
1304 op->flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
1305 }
1306 break;
1307
1308 case OP_CALL:
1309 // trashed regs must be explicitly detected later
1310 op->regmask_dst = 0;
1311 break;
1312
1313 case OP_LEAVE:
1314 op->regmask_dst = (1 << xBP) | (1 << xSP);
1315 op->regmask_src = 1 << xBP;
1316 break;
1317
1318 case OP_FLD:
1319 case OP_FILD:
1320 op->regmask_dst |= mxST0;
1321 break;
1322
1323 case OP_FLDc:
1324 op->regmask_dst |= mxST0;
1325 if (IS(words[op_w] + 3, "1"))
1326 op->operand[0].val = X87_CONST_1;
1327 else if (IS(words[op_w] + 3, "z"))
1328 op->operand[0].val = X87_CONST_Z;
1329 else
1330 aerr("TODO\n");
1331 break;
1332
1333 case OP_FST:
1334 op->regmask_src |= mxST0;
1335 break;
1336
1337 case OP_FADD:
1338 case OP_FDIV:
1339 case OP_FMUL:
1340 case OP_FSUB:
1341 case OP_FDIVR:
1342 case OP_FSUBR:
1343 op->regmask_src |= mxST0;
1344 if (op->operand_cnt == 2)
1345 op->regmask_src |= op->regmask_dst;
1346 else if (op->operand_cnt == 1) {
1347 memcpy(&op->operand[1], &op->operand[0], sizeof(op->operand[1]));
1348 op->operand[0].type = OPT_REG;
1349 op->operand[0].lmod = OPLM_QWORD;
1350 op->operand[0].reg = xST0;
1351 op->regmask_dst |= mxST0;
1352 }
1353 else
1354 // IDA doesn't use this
1355 aerr("no operands?\n");
1356 break;
1357
1358 case OP_FIADD:
1359 case OP_FIDIV:
1360 case OP_FIMUL:
1361 case OP_FISUB:
1362 case OP_FIDIVR:
1363 case OP_FISUBR:
1364 op->regmask_src |= mxST0;
1365 op->regmask_dst |= mxST0;
1366 break;
1367
1368 default:
1369 break;
1370 }
1371
1372 if (op->operand[0].type == OPT_REG
1373 && op->operand[1].type == OPT_CONST)
1374 {
1375 struct parsed_opr *op1 = &op->operand[1];
1376 if ((op->op == OP_AND && op1->val == 0)
1377 || (op->op == OP_OR
1378 && (op1->val == ~0
1379 || (op->operand[0].lmod == OPLM_WORD && op1->val == 0xffff)
1380 || (op->operand[0].lmod == OPLM_BYTE && op1->val == 0xff))))
1381 {
1382 op->regmask_src = 0;
1383 }
1384 }
1385}
1386
1387static const char *op_name(struct parsed_op *po)
1388{
1389 static char buf[16];
1390 char *p;
1391 int i;
1392
1393 if (po->op == OP_JCC || po->op == OP_SCC) {
1394 p = buf;
1395 *p++ = (po->op == OP_JCC) ? 'j' : 's';
1396 if (po->pfo_inv)
1397 *p++ = 'n';
1398 strcpy(p, parsed_flag_op_names[po->pfo]);
1399 return buf;
1400 }
1401
1402 for (i = 0; i < ARRAY_SIZE(op_table); i++)
1403 if (op_table[i].op == po->op)
1404 return op_table[i].name;
1405
1406 return "???";
1407}
1408
1409// debug
1410static const char *dump_op(struct parsed_op *po)
1411{
1412 static char out[128];
1413 char *p = out;
1414 int i;
1415
1416 if (po == NULL)
1417 return "???";
1418
1419 snprintf(out, sizeof(out), "%s", op_name(po));
1420 for (i = 0; i < po->operand_cnt; i++) {
1421 p += strlen(p);
1422 if (i > 0)
1423 *p++ = ',';
1424 snprintf(p, sizeof(out) - (p - out),
1425 po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s",
1426 po->operand[i].name);
1427 }
1428
1429 return out;
1430}
1431
1432static const char *lmod_type_u(struct parsed_op *po,
1433 enum opr_lenmod lmod)
1434{
1435 switch (lmod) {
1436 case OPLM_QWORD:
1437 return "u64";
1438 case OPLM_DWORD:
1439 return "u32";
1440 case OPLM_WORD:
1441 return "u16";
1442 case OPLM_BYTE:
1443 return "u8";
1444 default:
1445 ferr(po, "invalid lmod: %d\n", lmod);
1446 return "(_invalid_)";
1447 }
1448}
1449
1450static const char *lmod_cast_u(struct parsed_op *po,
1451 enum opr_lenmod lmod)
1452{
1453 switch (lmod) {
1454 case OPLM_QWORD:
1455 return "";
1456 case OPLM_DWORD:
1457 return "";
1458 case OPLM_WORD:
1459 return "(u16)";
1460 case OPLM_BYTE:
1461 return "(u8)";
1462 default:
1463 ferr(po, "invalid lmod: %d\n", lmod);
1464 return "(_invalid_)";
1465 }
1466}
1467
1468static const char *lmod_cast_u_ptr(struct parsed_op *po,
1469 enum opr_lenmod lmod)
1470{
1471 switch (lmod) {
1472 case OPLM_QWORD:
1473 return "*(u64 *)";
1474 case OPLM_DWORD:
1475 return "*(u32 *)";
1476 case OPLM_WORD:
1477 return "*(u16 *)";
1478 case OPLM_BYTE:
1479 return "*(u8 *)";
1480 default:
1481 ferr(po, "invalid lmod: %d\n", lmod);
1482 return "(_invalid_)";
1483 }
1484}
1485
1486static const char *lmod_cast_s(struct parsed_op *po,
1487 enum opr_lenmod lmod)
1488{
1489 switch (lmod) {
1490 case OPLM_QWORD:
1491 return "(s64)";
1492 case OPLM_DWORD:
1493 return "(s32)";
1494 case OPLM_WORD:
1495 return "(s16)";
1496 case OPLM_BYTE:
1497 return "(s8)";
1498 default:
1499 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1500 return "(_invalid_)";
1501 }
1502}
1503
1504static const char *lmod_cast(struct parsed_op *po,
1505 enum opr_lenmod lmod, int is_signed)
1506{
1507 return is_signed ?
1508 lmod_cast_s(po, lmod) :
1509 lmod_cast_u(po, lmod);
1510}
1511
1512static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
1513{
1514 switch (lmod) {
1515 case OPLM_QWORD:
1516 return 8;
1517 case OPLM_DWORD:
1518 return 4;
1519 case OPLM_WORD:
1520 return 2;
1521 case OPLM_BYTE:
1522 return 1;
1523 default:
1524 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1525 return 0;
1526 }
1527}
1528
1529static const char *opr_name(struct parsed_op *po, int opr_num)
1530{
1531 if (opr_num >= po->operand_cnt)
1532 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1533 return po->operand[opr_num].name;
1534}
1535
1536static unsigned int opr_const(struct parsed_op *po, int opr_num)
1537{
1538 if (opr_num >= po->operand_cnt)
1539 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1540 if (po->operand[opr_num].type != OPT_CONST)
1541 ferr(po, "opr %d: const expected\n", opr_num);
1542 return po->operand[opr_num].val;
1543}
1544
1545static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr)
1546{
1547 if ((unsigned int)popr->reg >= ARRAY_SIZE(regs_r32))
1548 ferr(po, "invalid reg: %d\n", popr->reg);
1549 return regs_r32[popr->reg];
1550}
1551
1552static int check_simple_cast(const char *cast, int *bits, int *is_signed)
1553{
1554 if (IS_START(cast, "(s8)") || IS_START(cast, "(u8)"))
1555 *bits = 8;
1556 else if (IS_START(cast, "(s16)") || IS_START(cast, "(u16)"))
1557 *bits = 16;
1558 else if (IS_START(cast, "(s32)") || IS_START(cast, "(u32)"))
1559 *bits = 32;
1560 else if (IS_START(cast, "(s64)") || IS_START(cast, "(u64)"))
1561 *bits = 64;
1562 else
1563 return -1;
1564
1565 *is_signed = cast[1] == 's' ? 1 : 0;
1566 return 0;
1567}
1568
1569static int check_deref_cast(const char *cast, int *bits)
1570{
1571 if (IS_START(cast, "*(u8 *)"))
1572 *bits = 8;
1573 else if (IS_START(cast, "*(u16 *)"))
1574 *bits = 16;
1575 else if (IS_START(cast, "*(u32 *)"))
1576 *bits = 32;
1577 else if (IS_START(cast, "*(u64 *)"))
1578 *bits = 64;
1579 else
1580 return -1;
1581
1582 return 0;
1583}
1584
1585// cast1 is the "final" cast
1586static const char *simplify_cast(const char *cast1, const char *cast2)
1587{
1588 static char buf[256];
1589 int bits1, bits2;
1590 int s1, s2;
1591
1592 if (cast1[0] == 0)
1593 return cast2;
1594 if (cast2[0] == 0)
1595 return cast1;
1596 if (IS(cast1, cast2))
1597 return cast1;
1598
1599 if (check_simple_cast(cast1, &bits1, &s1) == 0
1600 && check_simple_cast(cast2, &bits2, &s2) == 0)
1601 {
1602 if (bits1 <= bits2)
1603 return cast1;
1604 }
1605 if (check_simple_cast(cast1, &bits1, &s1) == 0
1606 && check_deref_cast(cast2, &bits2) == 0)
1607 {
1608 if (bits1 == bits2) {
1609 snprintf(buf, sizeof(buf), "*(%c%d *)", s1 ? 's' : 'u', bits1);
1610 return buf;
1611 }
1612 }
1613
1614 if (strchr(cast1, '*') && IS_START(cast2, "(u32)"))
1615 return cast1;
1616
1617 snprintf(buf, sizeof(buf), "%s%s", cast1, cast2);
1618 return buf;
1619}
1620
1621static const char *simplify_cast_num(const char *cast, unsigned int val)
1622{
1623 if (IS(cast, "(u8)") && val < 0x100)
1624 return "";
1625 if (IS(cast, "(s8)") && val < 0x80)
1626 return "";
1627 if (IS(cast, "(u16)") && val < 0x10000)
1628 return "";
1629 if (IS(cast, "(s16)") && val < 0x8000)
1630 return "";
1631 if (IS(cast, "(s32)") && val < 0x80000000)
1632 return "";
1633
1634 return cast;
1635}
1636
1637static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
1638 int *extra_offs)
1639{
1640 const char *p;
1641 char *endp;
1642 int namelen;
1643 int i;
1644
1645 *extra_offs = 0;
1646 namelen = strlen(name);
1647
1648 p = strchr(name, '+');
1649 if (p != NULL) {
1650 namelen = p - name;
1651 if (namelen <= 0)
1652 ferr(po, "equ parse failed for '%s'\n", name);
1653
1654 if (IS_START(p, "0x"))
1655 p += 2;
1656 *extra_offs = strtol(p, &endp, 16);
1657 if (*endp != 0)
1658 ferr(po, "equ parse failed for '%s'\n", name);
1659 }
1660
1661 for (i = 0; i < g_eqcnt; i++)
1662 if (strncmp(g_eqs[i].name, name, namelen) == 0
1663 && g_eqs[i].name[namelen] == 0)
1664 break;
1665 if (i >= g_eqcnt) {
1666 if (po != NULL)
1667 ferr(po, "unresolved equ name: '%s'\n", name);
1668 return NULL;
1669 }
1670
1671 return &g_eqs[i];
1672}
1673
1674static int is_stack_access(struct parsed_op *po,
1675 const struct parsed_opr *popr)
1676{
1677 return (parse_stack_el(popr->name, NULL, 0)
1678 || (g_bp_frame && !(po->flags & OPF_EBP_S)
1679 && IS_START(popr->name, "ebp")));
1680}
1681
1682static void parse_stack_access(struct parsed_op *po,
1683 const char *name, char *ofs_reg, int *offset_out,
1684 int *stack_ra_out, const char **bp_arg_out, int is_lea)
1685{
1686 const char *bp_arg = "";
1687 const char *p = NULL;
1688 struct parsed_equ *eq;
1689 char *endp = NULL;
1690 int stack_ra = 0;
1691 int offset = 0;
1692
1693 ofs_reg[0] = 0;
1694
1695 if (IS_START(name, "ebp-")
1696 || (IS_START(name, "ebp+") && '0' <= name[4] && name[4] <= '9'))
1697 {
1698 p = name + 4;
1699 if (IS_START(p, "0x"))
1700 p += 2;
1701 offset = strtoul(p, &endp, 16);
1702 if (name[3] == '-')
1703 offset = -offset;
1704 if (*endp != 0)
1705 ferr(po, "ebp- parse of '%s' failed\n", name);
1706 }
1707 else {
1708 bp_arg = parse_stack_el(name, ofs_reg, 0);
1709 snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
1710 eq = equ_find(po, bp_arg, &offset);
1711 if (eq == NULL)
1712 ferr(po, "detected but missing eq\n");
1713 offset += eq->offset;
1714 }
1715
1716 if (!strncmp(name, "ebp", 3))
1717 stack_ra = 4;
1718
1719 // yes it sometimes LEAs ra for compares..
1720 if (!is_lea && ofs_reg[0] == 0
1721 && stack_ra <= offset && offset < stack_ra + 4)
1722 {
1723 ferr(po, "reference to ra? %d %d\n", offset, stack_ra);
1724 }
1725
1726 *offset_out = offset;
1727 *stack_ra_out = stack_ra;
1728 if (bp_arg_out)
1729 *bp_arg_out = bp_arg;
1730}
1731
1732static int stack_frame_access(struct parsed_op *po,
1733 struct parsed_opr *popr, char *buf, size_t buf_size,
1734 const char *name, const char *cast, int is_src, int is_lea)
1735{
1736 enum opr_lenmod tmp_lmod = OPLM_UNSPEC;
1737 const char *prefix = "";
1738 const char *bp_arg = NULL;
1739 char ofs_reg[16] = { 0, };
1740 int i, arg_i, arg_s;
1741 int unaligned = 0;
1742 int stack_ra = 0;
1743 int offset = 0;
1744 int retval = -1;
1745 int sf_ofs;
1746 int lim;
1747
1748 if (po->flags & OPF_EBP_S)
1749 ferr(po, "stack_frame_access while ebp is scratch\n");
1750
1751 parse_stack_access(po, name, ofs_reg, &offset,
1752 &stack_ra, &bp_arg, is_lea);
1753
1754 if (offset > stack_ra)
1755 {
1756 arg_i = (offset - stack_ra - 4) / 4;
1757 if (arg_i < 0 || arg_i >= g_func_pp->argc_stack)
1758 {
1759 if (g_func_pp->is_vararg
1760 && arg_i == g_func_pp->argc_stack && is_lea)
1761 {
1762 // should be va_list
1763 if (cast[0] == 0)
1764 cast = "(u32)";
1765 snprintf(buf, buf_size, "%sap", cast);
1766 return -1;
1767 }
1768 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1769 offset, bp_arg, arg_i);
1770 }
1771 if (ofs_reg[0] != 0)
1772 ferr(po, "offset reg on arg access?\n");
1773
1774 for (i = arg_s = 0; i < g_func_pp->argc; i++) {
1775 if (g_func_pp->arg[i].reg != NULL)
1776 continue;
1777 if (arg_s == arg_i)
1778 break;
1779 arg_s++;
1780 }
1781 if (i == g_func_pp->argc)
1782 ferr(po, "arg %d not in prototype?\n", arg_i);
1783
1784 popr->is_ptr = g_func_pp->arg[i].type.is_ptr;
1785 retval = i;
1786
1787 switch (popr->lmod)
1788 {
1789 case OPLM_BYTE:
1790 if (is_lea)
1791 ferr(po, "lea/byte to arg?\n");
1792 if (is_src && (offset & 3) == 0)
1793 snprintf(buf, buf_size, "%sa%d",
1794 simplify_cast(cast, "(u8)"), i + 1);
1795 else
1796 snprintf(buf, buf_size, "%sBYTE%d(a%d)",
1797 cast, offset & 3, i + 1);
1798 break;
1799
1800 case OPLM_WORD:
1801 if (is_lea)
1802 ferr(po, "lea/word to arg?\n");
1803 if (offset & 1) {
1804 unaligned = 1;
1805 if (!is_src) {
1806 if (offset & 2)
1807 ferr(po, "problematic arg store\n");
1808 snprintf(buf, buf_size, "%s((char *)&a%d + 1)",
1809 simplify_cast(cast, "*(u16 *)"), i + 1);
1810 }
1811 else
1812 ferr(po, "unaligned arg word load\n");
1813 }
1814 else if (is_src && (offset & 2) == 0)
1815 snprintf(buf, buf_size, "%sa%d",
1816 simplify_cast(cast, "(u16)"), i + 1);
1817 else
1818 snprintf(buf, buf_size, "%s%sWORD(a%d)",
1819 cast, (offset & 2) ? "HI" : "LO", i + 1);
1820 break;
1821
1822 case OPLM_DWORD:
1823 if (cast[0])
1824 prefix = cast;
1825 else if (is_src)
1826 prefix = "(u32)";
1827
1828 if (offset & 3) {
1829 unaligned = 1;
1830 if (is_lea)
1831 snprintf(buf, buf_size, "(u32)&a%d + %d",
1832 i + 1, offset & 3);
1833 else if (!is_src)
1834 ferr(po, "unaligned arg store\n");
1835 else {
1836 // mov edx, [ebp+arg_4+2]; movsx ecx, dx
1837 snprintf(buf, buf_size, "%s(a%d >> %d)",
1838 prefix, i + 1, (offset & 3) * 8);
1839 }
1840 }
1841 else {
1842 snprintf(buf, buf_size, "%s%sa%d",
1843 prefix, is_lea ? "&" : "", i + 1);
1844 }
1845 break;
1846
1847 default:
1848 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
1849 }
1850
1851 if (unaligned)
1852 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1853
1854 // common problem
1855 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type);
1856 if (tmp_lmod != OPLM_DWORD
1857 && (unaligned || (!is_src && lmod_bytes(po, tmp_lmod)
1858 < lmod_bytes(po, popr->lmod) + (offset & 3))))
1859 {
1860 ferr(po, "bp_arg arg%d/w offset %d and type '%s' is too small\n",
1861 i + 1, offset, g_func_pp->arg[i].type.name);
1862 }
1863 // can't check this because msvc likes to reuse
1864 // arg space for scratch..
1865 //if (popr->is_ptr && popr->lmod != OPLM_DWORD)
1866 // ferr(po, "bp_arg arg%d: non-dword ptr access\n", i + 1);
1867 }
1868 else
1869 {
1870 if (g_stack_fsz == 0)
1871 ferr(po, "stack var access without stackframe\n");
1872 g_stack_frame_used = 1;
1873
1874 sf_ofs = g_stack_fsz + offset;
1875 lim = (ofs_reg[0] != 0) ? -4 : 0;
1876 if (offset > 0 || sf_ofs < lim)
1877 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
1878
1879 if (is_lea)
1880 prefix = "(u32)&";
1881 else
1882 prefix = cast;
1883
1884 switch (popr->lmod)
1885 {
1886 case OPLM_BYTE:
1887 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1888 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1889 break;
1890
1891 case OPLM_WORD:
1892 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1893 // known unaligned or possibly unaligned
1894 strcat(g_comment, " unaligned");
1895 if (prefix[0] == 0)
1896 prefix = "*(u16 *)&";
1897 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1898 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1899 break;
1900 }
1901 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
1902 break;
1903
1904 case OPLM_DWORD:
1905 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1906 // known unaligned or possibly unaligned
1907 strcat(g_comment, " unaligned");
1908 if (prefix[0] == 0)
1909 prefix = "*(u32 *)&";
1910 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1911 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1912 break;
1913 }
1914 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
1915 break;
1916
1917 case OPLM_QWORD:
1918 ferr_assert(po, !(sf_ofs & 7));
1919 ferr_assert(po, ofs_reg[0] == 0);
1920 // float callers set is_lea
1921 ferr_assert(po, is_lea);
1922 snprintf(buf, buf_size, "%ssf.q[%d]", prefix, sf_ofs / 8);
1923 break;
1924
1925 default:
1926 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
1927 }
1928 }
1929
1930 return retval;
1931}
1932
1933static void check_func_pp(struct parsed_op *po,
1934 const struct parsed_proto *pp, const char *pfx)
1935{
1936 enum opr_lenmod tmp_lmod;
1937 char buf[256];
1938 int ret, i;
1939
1940 if (pp->argc_reg != 0) {
1941 if (/*!g_allow_regfunc &&*/ !pp->is_fastcall) {
1942 pp_print(buf, sizeof(buf), pp);
1943 ferr(po, "%s: unexpected reg arg in icall: %s\n", pfx, buf);
1944 }
1945 if (pp->argc_stack > 0 && pp->argc_reg != 2)
1946 ferr(po, "%s: %d reg arg(s) with %d stack arg(s)\n",
1947 pfx, pp->argc_reg, pp->argc_stack);
1948 }
1949
1950 // fptrs must use 32bit args, callsite might have no information and
1951 // lack a cast to smaller types, which results in incorrectly masked
1952 // args passed (callee may assume masked args, it does on ARM)
1953 if (!pp->is_osinc) {
1954 for (i = 0; i < pp->argc; i++) {
1955 ret = guess_lmod_from_c_type(&tmp_lmod, &pp->arg[i].type);
1956 if (ret && tmp_lmod != OPLM_DWORD)
1957 ferr(po, "reference to %s with arg%d '%s'\n", pp->name,
1958 i + 1, pp->arg[i].type.name);
1959 }
1960 }
1961}
1962
1963static const char *check_label_read_ref(struct parsed_op *po,
1964 const char *name)
1965{
1966 const struct parsed_proto *pp;
1967
1968 pp = proto_parse(g_fhdr, name, 0);
1969 if (pp == NULL)
1970 ferr(po, "proto_parse failed for ref '%s'\n", name);
1971
1972 if (pp->is_func)
1973 check_func_pp(po, pp, "ref");
1974
1975 return pp->name;
1976}
1977
1978static char *out_src_opr(char *buf, size_t buf_size,
1979 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
1980 int is_lea)
1981{
1982 char tmp1[256], tmp2[256];
1983 char expr[256];
1984 const char *name;
1985 char *p;
1986 int ret;
1987
1988 if (cast == NULL)
1989 cast = "";
1990
1991 switch (popr->type) {
1992 case OPT_REG:
1993 if (is_lea)
1994 ferr(po, "lea from reg?\n");
1995
1996 switch (popr->lmod) {
1997 case OPLM_QWORD:
1998 snprintf(buf, buf_size, "%s%s.q", cast, opr_reg_p(po, popr));
1999 break;
2000 case OPLM_DWORD:
2001 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
2002 break;
2003 case OPLM_WORD:
2004 snprintf(buf, buf_size, "%s%s",
2005 simplify_cast(cast, "(u16)"), opr_reg_p(po, popr));
2006 break;
2007 case OPLM_BYTE:
2008 if (popr->name[1] == 'h') // XXX..
2009 snprintf(buf, buf_size, "%s(%s >> 8)",
2010 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
2011 else
2012 snprintf(buf, buf_size, "%s%s",
2013 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
2014 break;
2015 default:
2016 ferr(po, "invalid src lmod: %d\n", popr->lmod);
2017 }
2018 break;
2019
2020 case OPT_REGMEM:
2021 if (is_stack_access(po, popr)) {
2022 stack_frame_access(po, popr, buf, buf_size,
2023 popr->name, cast, 1, is_lea);
2024 break;
2025 }
2026
2027 strcpy(expr, popr->name);
2028 if (strchr(expr, '[')) {
2029 // special case: '[' can only be left for label[reg] form
2030 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
2031 if (ret != 2)
2032 ferr(po, "parse failure for '%s'\n", expr);
2033 if (tmp1[0] == '(') {
2034 // (off_4FFF50+3)[eax]
2035 p = strchr(tmp1 + 1, ')');
2036 if (p == NULL || p[1] != 0)
2037 ferr(po, "parse failure (2) for '%s'\n", expr);
2038 *p = 0;
2039 memmove(tmp1, tmp1 + 1, strlen(tmp1));
2040 }
2041 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
2042 }
2043
2044 // XXX: do we need more parsing?
2045 if (is_lea) {
2046 snprintf(buf, buf_size, "%s", expr);
2047 break;
2048 }
2049
2050 snprintf(buf, buf_size, "%s(%s)",
2051 simplify_cast(cast, lmod_cast_u_ptr(po, popr->lmod)), expr);
2052 break;
2053
2054 case OPT_LABEL:
2055 name = check_label_read_ref(po, popr->name);
2056 if (cast[0] == 0 && popr->is_ptr)
2057 cast = "(u32)";
2058
2059 if (is_lea)
2060 snprintf(buf, buf_size, "(u32)&%s", name);
2061 else if (popr->size_lt)
2062 snprintf(buf, buf_size, "%s%s%s%s", cast,
2063 lmod_cast_u_ptr(po, popr->lmod),
2064 popr->is_array ? "" : "&", name);
2065 else
2066 snprintf(buf, buf_size, "%s%s%s", cast, name,
2067 popr->is_array ? "[0]" : "");
2068 break;
2069
2070 case OPT_OFFSET:
2071 name = check_label_read_ref(po, popr->name);
2072 if (cast[0] == 0)
2073 cast = "(u32)";
2074 if (is_lea)
2075 ferr(po, "lea an offset?\n");
2076 snprintf(buf, buf_size, "%s&%s", cast, name);
2077 break;
2078
2079 case OPT_CONST:
2080 if (is_lea)
2081 ferr(po, "lea from const?\n");
2082
2083 printf_number(tmp1, sizeof(tmp1), popr->val);
2084 if (popr->val == 0 && strchr(cast, '*'))
2085 snprintf(buf, buf_size, "NULL");
2086 else
2087 snprintf(buf, buf_size, "%s%s",
2088 simplify_cast_num(cast, popr->val), tmp1);
2089 break;
2090
2091 default:
2092 ferr(po, "invalid src type: %d\n", popr->type);
2093 }
2094
2095 return buf;
2096}
2097
2098// note: may set is_ptr (we find that out late for ebp frame..)
2099static char *out_dst_opr(char *buf, size_t buf_size,
2100 struct parsed_op *po, struct parsed_opr *popr)
2101{
2102 switch (popr->type) {
2103 case OPT_REG:
2104 switch (popr->lmod) {
2105 case OPLM_QWORD:
2106 snprintf(buf, buf_size, "%s.q", opr_reg_p(po, popr));
2107 break;
2108 case OPLM_DWORD:
2109 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
2110 break;
2111 case OPLM_WORD:
2112 // ugh..
2113 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
2114 break;
2115 case OPLM_BYTE:
2116 // ugh..
2117 if (popr->name[1] == 'h') // XXX..
2118 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
2119 else
2120 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
2121 break;
2122 default:
2123 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
2124 }
2125 break;
2126
2127 case OPT_REGMEM:
2128 if (is_stack_access(po, popr)) {
2129 stack_frame_access(po, popr, buf, buf_size,
2130 popr->name, "", 0, 0);
2131 break;
2132 }
2133
2134 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
2135
2136 case OPT_LABEL:
2137 if (popr->size_mismatch)
2138 snprintf(buf, buf_size, "%s%s%s",
2139 lmod_cast_u_ptr(po, popr->lmod),
2140 popr->is_array ? "" : "&", popr->name);
2141 else
2142 snprintf(buf, buf_size, "%s%s", popr->name,
2143 popr->is_array ? "[0]" : "");
2144 break;
2145
2146 default:
2147 ferr(po, "invalid dst type: %d\n", popr->type);
2148 }
2149
2150 return buf;
2151}
2152
2153static char *out_src_opr_u32(char *buf, size_t buf_size,
2154 struct parsed_op *po, struct parsed_opr *popr)
2155{
2156 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
2157}
2158
2159static char *out_src_opr_float(char *buf, size_t buf_size,
2160 struct parsed_op *po, struct parsed_opr *popr)
2161{
2162 const char *cast = NULL;
2163 char tmp[256];
2164
2165 switch (popr->type) {
2166 case OPT_REG:
2167 if (popr->reg < xST0 || popr->reg > xST7)
2168 ferr(po, "bad reg: %d\n", popr->reg);
2169
2170 snprintf(buf, buf_size, "f_st%d", popr->reg - xST0);
2171 break;
2172
2173 case OPT_REGMEM:
2174 case OPT_LABEL:
2175 case OPT_OFFSET:
2176 switch (popr->lmod) {
2177 case OPLM_QWORD:
2178 cast = "double";
2179 break;
2180 case OPLM_DWORD:
2181 cast = "float";
2182 break;
2183 default:
2184 ferr(po, "unhandled lmod: %d\n", popr->lmod);
2185 break;
2186 }
2187 out_src_opr(tmp, sizeof(tmp), po, popr, "", 1);
2188 snprintf(buf, buf_size, "*((%s *)%s)", cast, tmp);
2189 break;
2190
2191 default:
2192 ferr(po, "invalid float type: %d\n", popr->type);
2193 }
2194
2195 return buf;
2196}
2197
2198static char *out_dst_opr_float(char *buf, size_t buf_size,
2199 struct parsed_op *po, struct parsed_opr *popr)
2200{
2201 // same?
2202 return out_src_opr_float(buf, buf_size, po, popr);
2203}
2204
2205static void out_test_for_cc(char *buf, size_t buf_size,
2206 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
2207 enum opr_lenmod lmod, const char *expr)
2208{
2209 const char *cast, *scast;
2210
2211 cast = lmod_cast_u(po, lmod);
2212 scast = lmod_cast_s(po, lmod);
2213
2214 switch (pfo) {
2215 case PFO_Z:
2216 case PFO_BE: // CF=1||ZF=1; CF=0
2217 snprintf(buf, buf_size, "(%s%s %s 0)",
2218 cast, expr, is_inv ? "!=" : "==");
2219 break;
2220
2221 case PFO_S:
2222 case PFO_L: // SF!=OF; OF=0
2223 snprintf(buf, buf_size, "(%s%s %s 0)",
2224 scast, expr, is_inv ? ">=" : "<");
2225 break;
2226
2227 case PFO_LE: // ZF=1||SF!=OF; OF=0
2228 snprintf(buf, buf_size, "(%s%s %s 0)",
2229 scast, expr, is_inv ? ">" : "<=");
2230 break;
2231
2232 default:
2233 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
2234 }
2235}
2236
2237static void out_cmp_for_cc(char *buf, size_t buf_size,
2238 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
2239{
2240 const char *cast, *scast, *cast_use;
2241 char buf1[256], buf2[256];
2242 enum opr_lenmod lmod;
2243
2244 if (po->op != OP_DEC && po->operand[0].lmod != po->operand[1].lmod)
2245 ferr(po, "%s: lmod mismatch: %d %d\n", __func__,
2246 po->operand[0].lmod, po->operand[1].lmod);
2247 lmod = po->operand[0].lmod;
2248
2249 cast = lmod_cast_u(po, lmod);
2250 scast = lmod_cast_s(po, lmod);
2251
2252 switch (pfo) {
2253 case PFO_C:
2254 case PFO_Z:
2255 case PFO_BE: // !a
2256 cast_use = cast;
2257 break;
2258
2259 case PFO_S:
2260 case PFO_L: // !ge
2261 case PFO_LE:
2262 cast_use = scast;
2263 break;
2264
2265 default:
2266 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
2267 }
2268
2269 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], cast_use, 0);
2270 if (po->op == OP_DEC)
2271 snprintf(buf2, sizeof(buf2), "1");
2272 else
2273 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], cast_use, 0);
2274
2275 switch (pfo) {
2276 case PFO_C:
2277 // note: must be unsigned compare
2278 snprintf(buf, buf_size, "(%s %s %s)",
2279 buf1, is_inv ? ">=" : "<", buf2);
2280 break;
2281
2282 case PFO_Z:
2283 snprintf(buf, buf_size, "(%s %s %s)",
2284 buf1, is_inv ? "!=" : "==", buf2);
2285 break;
2286
2287 case PFO_BE: // !a
2288 // note: must be unsigned compare
2289 snprintf(buf, buf_size, "(%s %s %s)",
2290 buf1, is_inv ? ">" : "<=", buf2);
2291
2292 // annoying case
2293 if (is_inv && lmod == OPLM_BYTE
2294 && po->operand[1].type == OPT_CONST
2295 && po->operand[1].val == 0xff)
2296 {
2297 snprintf(g_comment, sizeof(g_comment), "if %s", buf);
2298 snprintf(buf, buf_size, "(0)");
2299 }
2300 break;
2301
2302 // note: must be signed compare
2303 case PFO_S:
2304 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
2305 scast, buf1, buf2, is_inv ? ">=" : "<");
2306 break;
2307
2308 case PFO_L: // !ge
2309 snprintf(buf, buf_size, "(%s %s %s)",
2310 buf1, is_inv ? ">=" : "<", buf2);
2311 break;
2312
2313 case PFO_LE: // !g
2314 snprintf(buf, buf_size, "(%s %s %s)",
2315 buf1, is_inv ? ">" : "<=", buf2);
2316 break;
2317
2318 default:
2319 break;
2320 }
2321}
2322
2323static void out_cmp_test(char *buf, size_t buf_size,
2324 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
2325{
2326 char buf1[256], buf2[256], buf3[256];
2327
2328 if (po->op == OP_TEST) {
2329 if (IS(opr_name(po, 0), opr_name(po, 1))) {
2330 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
2331 }
2332 else {
2333 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
2334 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
2335 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
2336 }
2337 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
2338 po->operand[0].lmod, buf3);
2339 }
2340 else if (po->op == OP_CMP) {
2341 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv);
2342 }
2343 else
2344 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
2345}
2346
2347static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
2348 struct parsed_opr *popr2)
2349{
2350 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
2351 ferr(po, "missing lmod for both operands\n");
2352
2353 if (popr1->lmod == OPLM_UNSPEC)
2354 popr1->lmod = popr2->lmod;
2355 else if (popr2->lmod == OPLM_UNSPEC)
2356 popr2->lmod = popr1->lmod;
2357 else if (popr1->lmod != popr2->lmod) {
2358 if (popr1->type_from_var) {
2359 popr1->size_mismatch = 1;
2360 if (popr1->lmod < popr2->lmod)
2361 popr1->size_lt = 1;
2362 popr1->lmod = popr2->lmod;
2363 }
2364 else if (popr2->type_from_var) {
2365 popr2->size_mismatch = 1;
2366 if (popr2->lmod < popr1->lmod)
2367 popr2->size_lt = 1;
2368 popr2->lmod = popr1->lmod;
2369 }
2370 else
2371 ferr(po, "conflicting lmods: %d vs %d\n",
2372 popr1->lmod, popr2->lmod);
2373 }
2374}
2375
2376static const char *op_to_c(struct parsed_op *po)
2377{
2378 switch (po->op)
2379 {
2380 case OP_ADD:
2381 case OP_ADC:
2382 return "+";
2383 case OP_SUB:
2384 case OP_SBB:
2385 return "-";
2386 case OP_AND:
2387 return "&";
2388 case OP_OR:
2389 return "|";
2390 case OP_XOR:
2391 return "^";
2392 case OP_SHL:
2393 return "<<";
2394 case OP_SHR:
2395 return ">>";
2396 case OP_MUL:
2397 case OP_IMUL:
2398 return "*";
2399 default:
2400 ferr(po, "op_to_c was supplied with %d\n", po->op);
2401 }
2402}
2403
2404// last op in stream - unconditional branch or ret
2405#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
2406 || ((ops[_i].flags & (OPF_JMP|OPF_CJMP|OPF_RMD)) == OPF_JMP \
2407 && ops[_i].op != OP_CALL))
2408
2409#define check_i(po, i) \
2410 if ((i) < 0) \
2411 ferr(po, "bad " #i ": %d\n", i)
2412
2413// note: this skips over calls and rm'd stuff assuming they're handled
2414// so it's intended to use at one of final passes
2415static int scan_for_pop(int i, int opcnt, int magic, int reg,
2416 int depth, int flags_set)
2417{
2418 struct parsed_op *po;
2419 int relevant;
2420 int ret = 0;
2421 int j;
2422
2423 for (; i < opcnt; i++) {
2424 po = &ops[i];
2425 if (po->cc_scratch == magic)
2426 return ret; // already checked
2427 po->cc_scratch = magic;
2428
2429 if (po->flags & OPF_TAIL) {
2430 if (po->op == OP_CALL) {
2431 if (po->pp != NULL && po->pp->is_noreturn)
2432 // assume no stack cleanup for noreturn
2433 return 1;
2434 }
2435 return -1; // deadend
2436 }
2437
2438 if (po->flags & (OPF_RMD|OPF_DONE|OPF_FARG))
2439 continue;
2440
2441 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2442 if (po->btj != NULL) {
2443 // jumptable
2444 for (j = 0; j < po->btj->count; j++) {
2445 check_i(po, po->btj->d[j].bt_i);
2446 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, magic, reg,
2447 depth, flags_set);
2448 if (ret < 0)
2449 return ret; // dead end
2450 }
2451 return ret;
2452 }
2453
2454 check_i(po, po->bt_i);
2455 if (po->flags & OPF_CJMP) {
2456 ret |= scan_for_pop(po->bt_i, opcnt, magic, reg,
2457 depth, flags_set);
2458 if (ret < 0)
2459 return ret; // dead end
2460 }
2461 else {
2462 i = po->bt_i - 1;
2463 }
2464 continue;
2465 }
2466
2467 relevant = 0;
2468 if ((po->op == OP_POP || po->op == OP_PUSH)
2469 && po->operand[0].type == OPT_REG && po->operand[0].reg == reg)
2470 {
2471 relevant = 1;
2472 }
2473
2474 if (po->op == OP_PUSH) {
2475 depth++;
2476 }
2477 else if (po->op == OP_POP) {
2478 if (relevant && depth == 0) {
2479 po->flags |= flags_set;
2480 return 1;
2481 }
2482 depth--;
2483 }
2484 }
2485
2486 return -1;
2487}
2488
2489// scan for 'reg' pop backwards starting from i
2490// intended to use for register restore search, so other reg
2491// references are considered an error
2492static int scan_for_rsave_pop_reg(int i, int magic, int reg, int set_flags)
2493{
2494 struct parsed_op *po;
2495 struct label_ref *lr;
2496 int ret = 0;
2497
2498 ops[i].cc_scratch = magic;
2499
2500 while (1)
2501 {
2502 if (g_labels[i] != NULL) {
2503 lr = &g_label_refs[i];
2504 for (; lr != NULL; lr = lr->next) {
2505 check_i(&ops[i], lr->i);
2506 ret |= scan_for_rsave_pop_reg(lr->i, magic, reg, set_flags);
2507 if (ret < 0)
2508 return ret;
2509 }
2510 if (i > 0 && LAST_OP(i - 1))
2511 return ret;
2512 }
2513
2514 i--;
2515 if (i < 0)
2516 break;
2517
2518 if (ops[i].cc_scratch == magic)
2519 return ret;
2520 ops[i].cc_scratch = magic;
2521
2522 po = &ops[i];
2523 if (po->op == OP_POP && po->operand[0].reg == reg) {
2524 if (po->flags & (OPF_RMD|OPF_DONE))
2525 return -1;
2526
2527 po->flags |= set_flags;
2528 return 1;
2529 }
2530
2531 // this also covers the case where we reach corresponding push
2532 if ((po->regmask_dst | po->regmask_src) & (1 << reg))
2533 return -1;
2534 }
2535
2536 // nothing interesting on this path
2537 return 0;
2538}
2539
2540static void find_reachable_exits(int i, int opcnt, int magic,
2541 int *exits, int *exit_count)
2542{
2543 struct parsed_op *po;
2544 int j;
2545
2546 for (; i < opcnt; i++)
2547 {
2548 po = &ops[i];
2549 if (po->cc_scratch == magic)
2550 return;
2551 po->cc_scratch = magic;
2552
2553 if (po->flags & OPF_TAIL) {
2554 ferr_assert(po, *exit_count < MAX_EXITS);
2555 exits[*exit_count] = i;
2556 (*exit_count)++;
2557 return;
2558 }
2559
2560 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2561 if (po->flags & OPF_RMD)
2562 continue;
2563
2564 if (po->btj != NULL) {
2565 for (j = 0; j < po->btj->count; j++) {
2566 check_i(po, po->btj->d[j].bt_i);
2567 find_reachable_exits(po->btj->d[j].bt_i, opcnt, magic,
2568 exits, exit_count);
2569 }
2570 return;
2571 }
2572
2573 check_i(po, po->bt_i);
2574 if (po->flags & OPF_CJMP)
2575 find_reachable_exits(po->bt_i, opcnt, magic, exits, exit_count);
2576 else
2577 i = po->bt_i - 1;
2578 continue;
2579 }
2580 }
2581}
2582
2583// scan for 'reg' pop backwards starting from exits (all paths)
2584static int scan_for_pop_ret(int i, int opcnt, int reg, int set_flags)
2585{
2586 static int exits[MAX_EXITS];
2587 static int exit_count;
2588 int j, ret;
2589
2590 if (!set_flags) {
2591 exit_count = 0;
2592 find_reachable_exits(i, opcnt, i + opcnt * 15, exits,
2593 &exit_count);
2594 ferr_assert(&ops[i], exit_count > 0);
2595 }
2596
2597 for (j = 0; j < exit_count; j++) {
2598 ret = scan_for_rsave_pop_reg(exits[j], i + opcnt * 16 + set_flags,
2599 reg, set_flags);
2600 if (ret == -1)
2601 return -1;
2602 }
2603
2604 return 1;
2605}
2606
2607// scan for one or more pop of push <const>
2608static int scan_for_pop_const_r(int i, int opcnt, int magic,
2609 int push_i, int is_probe)
2610{
2611 struct parsed_op *po;
2612 struct label_ref *lr;
2613 int ret = 0;
2614 int j;
2615
2616 for (; i < opcnt; i++)
2617 {
2618 po = &ops[i];
2619 if (po->cc_scratch == magic)
2620 return ret; // already checked
2621 po->cc_scratch = magic;
2622
2623 if (po->flags & OPF_JMP) {
2624 if (po->flags & OPF_RMD)
2625 continue;
2626 if (po->op == OP_CALL)
2627 return -1;
2628
2629 if (po->btj != NULL) {
2630 for (j = 0; j < po->btj->count; j++) {
2631 check_i(po, po->btj->d[j].bt_i);
2632 ret |= scan_for_pop_const_r(po->btj->d[j].bt_i, opcnt, magic,
2633 push_i, is_probe);
2634 if (ret < 0)
2635 return ret;
2636 }
2637 return ret;
2638 }
2639
2640 check_i(po, po->bt_i);
2641 if (po->flags & OPF_CJMP) {
2642 ret |= scan_for_pop_const_r(po->bt_i, opcnt, magic, push_i,
2643 is_probe);
2644 if (ret < 0)
2645 return ret;
2646 }
2647 else {
2648 i = po->bt_i - 1;
2649 }
2650 continue;
2651 }
2652
2653 if ((po->flags & (OPF_TAIL|OPF_RSAVE)) || po->op == OP_PUSH)
2654 return -1;
2655
2656 if (g_labels[i] != NULL) {
2657 // all refs must be visited
2658 lr = &g_label_refs[i];
2659 for (; lr != NULL; lr = lr->next) {
2660 check_i(po, lr->i);
2661 if (ops[lr->i].cc_scratch != magic)
2662 return -1;
2663 }
2664 if (i > 0 && !LAST_OP(i - 1) && ops[i - 1].cc_scratch != magic)
2665 return -1;
2666 }
2667
2668 if (po->op == OP_POP)
2669 {
2670 if (po->flags & (OPF_RMD|OPF_DONE))
2671 return -1;
2672
2673 if (!is_probe) {
2674 po->flags |= OPF_DONE;
2675 po->datap = &ops[push_i];
2676 }
2677 return 1;
2678 }
2679 }
2680
2681 return -1;
2682}
2683
2684static void scan_for_pop_const(int i, int opcnt, int magic)
2685{
2686 int ret;
2687
2688 ret = scan_for_pop_const_r(i + 1, opcnt, magic, i, 1);
2689 if (ret == 1) {
2690 ops[i].flags |= OPF_RMD | OPF_DONE;
2691 scan_for_pop_const_r(i + 1, opcnt, magic + 1, i, 0);
2692 }
2693}
2694
2695// check if all branch targets within a marked path are also marked
2696// note: the path checked must not be empty or end with a branch
2697static int check_path_branches(int opcnt, int magic)
2698{
2699 struct parsed_op *po;
2700 int i, j;
2701
2702 for (i = 0; i < opcnt; i++) {
2703 po = &ops[i];
2704 if (po->cc_scratch != magic)
2705 continue;
2706
2707 if (po->flags & OPF_JMP) {
2708 if ((po->flags & OPF_RMD) || po->op == OP_CALL)
2709 continue;
2710
2711 if (po->btj != NULL) {
2712 for (j = 0; j < po->btj->count; j++) {
2713 check_i(po, po->btj->d[j].bt_i);
2714 if (ops[po->btj->d[j].bt_i].cc_scratch != magic)
2715 return 0;
2716 }
2717 }
2718
2719 check_i(po, po->bt_i);
2720 if (ops[po->bt_i].cc_scratch != magic)
2721 return 0;
2722 if ((po->flags & OPF_CJMP) && ops[i + 1].cc_scratch != magic)
2723 return 0;
2724 }
2725 }
2726
2727 return 1;
2728}
2729
2730// scan for multiple pushes for given pop
2731static int scan_pushes_for_pop_r(int i, int magic, int pop_i,
2732 int is_probe)
2733{
2734 int reg = ops[pop_i].operand[0].reg;
2735 struct parsed_op *po;
2736 struct label_ref *lr;
2737 int ret = 0;
2738
2739 ops[i].cc_scratch = magic;
2740
2741 while (1)
2742 {
2743 if (g_labels[i] != NULL) {
2744 lr = &g_label_refs[i];
2745 for (; lr != NULL; lr = lr->next) {
2746 check_i(&ops[i], lr->i);
2747 ret |= scan_pushes_for_pop_r(lr->i, magic, pop_i, is_probe);
2748 if (ret < 0)
2749 return ret;
2750 }
2751 if (i > 0 && LAST_OP(i - 1))
2752 return ret;
2753 }
2754
2755 i--;
2756 if (i < 0)
2757 break;
2758
2759 if (ops[i].cc_scratch == magic)
2760 return ret;
2761 ops[i].cc_scratch = magic;
2762
2763 po = &ops[i];
2764 if (po->op == OP_CALL)
2765 return -1;
2766 if ((po->flags & (OPF_TAIL|OPF_RSAVE)) || po->op == OP_POP)
2767 return -1;
2768
2769 if (po->op == OP_PUSH)
2770 {
2771 if (po->datap != NULL)
2772 return -1;
2773 if (po->operand[0].type == OPT_REG && po->operand[0].reg == reg)
2774 // leave this case for reg save/restore handlers
2775 return -1;
2776
2777 if (!is_probe) {
2778 po->flags |= OPF_PPUSH | OPF_DONE;
2779 po->datap = &ops[pop_i];
2780 }
2781 return 1;
2782 }
2783 }
2784
2785 return -1;
2786}
2787
2788static void scan_pushes_for_pop(int i, int opcnt, int *regmask_pp)
2789{
2790 int magic = i + opcnt * 14;
2791 int ret;
2792
2793 ret = scan_pushes_for_pop_r(i, magic, i, 1);
2794 if (ret == 1) {
2795 ret = check_path_branches(opcnt, magic);
2796 if (ret == 1) {
2797 ops[i].flags |= OPF_PPUSH | OPF_DONE;
2798 *regmask_pp |= 1 << ops[i].operand[0].reg;
2799 scan_pushes_for_pop_r(i, magic + 1, i, 0);
2800 }
2801 }
2802}
2803
2804static void scan_propagate_df(int i, int opcnt)
2805{
2806 struct parsed_op *po = &ops[i];
2807 int j;
2808
2809 for (; i < opcnt; i++) {
2810 po = &ops[i];
2811 if (po->flags & OPF_DF)
2812 return; // already resolved
2813 po->flags |= OPF_DF;
2814
2815 if (po->op == OP_CALL)
2816 ferr(po, "call with DF set?\n");
2817
2818 if (po->flags & OPF_JMP) {
2819 if (po->btj != NULL) {
2820 // jumptable
2821 for (j = 0; j < po->btj->count; j++) {
2822 check_i(po, po->btj->d[j].bt_i);
2823 scan_propagate_df(po->btj->d[j].bt_i, opcnt);
2824 }
2825 return;
2826 }
2827
2828 if (po->flags & OPF_RMD)
2829 continue;
2830 check_i(po, po->bt_i);
2831 if (po->flags & OPF_CJMP)
2832 scan_propagate_df(po->bt_i, opcnt);
2833 else
2834 i = po->bt_i - 1;
2835 continue;
2836 }
2837
2838 if (po->flags & OPF_TAIL)
2839 break;
2840
2841 if (po->op == OP_CLD) {
2842 po->flags |= OPF_RMD | OPF_DONE;
2843 return;
2844 }
2845 }
2846
2847 ferr(po, "missing DF clear?\n");
2848}
2849
2850// is operand 'opr' referenced by parsed_op 'po'?
2851static int is_opr_referenced(const struct parsed_opr *opr,
2852 const struct parsed_op *po)
2853{
2854 int i, mask;
2855
2856 if (opr->type == OPT_REG) {
2857 mask = po->regmask_dst | po->regmask_src;
2858 if (po->op == OP_CALL)
2859 mask |= (1 << xAX) | (1 << xCX) | (1 << xDX);
2860 if ((1 << opr->reg) & mask)
2861 return 1;
2862 else
2863 return 0;
2864 }
2865
2866 for (i = 0; i < po->operand_cnt; i++)
2867 if (IS(po->operand[0].name, opr->name))
2868 return 1;
2869
2870 return 0;
2871}
2872
2873// is operand 'opr' read by parsed_op 'po'?
2874static int is_opr_read(const struct parsed_opr *opr,
2875 const struct parsed_op *po)
2876{
2877 if (opr->type == OPT_REG) {
2878 if (po->regmask_src & (1 << opr->reg))
2879 return 1;
2880 else
2881 return 0;
2882 }
2883
2884 // yes I'm lazy
2885 return 0;
2886}
2887
2888// is operand 'opr' modified by parsed_op 'po'?
2889static int is_opr_modified(const struct parsed_opr *opr,
2890 const struct parsed_op *po)
2891{
2892 int mask;
2893
2894 if (opr->type == OPT_REG) {
2895 if (po->op == OP_CALL) {
2896 mask = po->regmask_dst;
2897 mask |= (1 << xAX) | (1 << xCX) | (1 << xDX); // ?
2898 if (mask & (1 << opr->reg))
2899 return 1;
2900 else
2901 return 0;
2902 }
2903
2904 if (po->regmask_dst & (1 << opr->reg))
2905 return 1;
2906 else
2907 return 0;
2908 }
2909
2910 return IS(po->operand[0].name, opr->name);
2911}
2912
2913// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
2914static int is_any_opr_modified(const struct parsed_op *po_test,
2915 const struct parsed_op *po, int c_mode)
2916{
2917 int mask;
2918 int i;
2919
2920 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2921 return 0;
2922
2923 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2924 return 0;
2925
2926 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
2927 return 1;
2928
2929 // in reality, it can wreck any register, but in decompiled C
2930 // version it can only overwrite eax or edx:eax
2931 mask = (1 << xAX) | (1 << xDX);
2932 if (!c_mode)
2933 mask |= 1 << xCX;
2934
2935 if (po->op == OP_CALL
2936 && ((po_test->regmask_src | po_test->regmask_dst) & mask))
2937 return 1;
2938
2939 for (i = 0; i < po_test->operand_cnt; i++)
2940 if (IS(po_test->operand[i].name, po->operand[0].name))
2941 return 1;
2942
2943 return 0;
2944}
2945
2946// scan for any po_test operand modification in range given
2947static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt,
2948 int c_mode)
2949{
2950 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2951 return -1;
2952
2953 for (; i < opcnt; i++) {
2954 if (is_any_opr_modified(po_test, &ops[i], c_mode))
2955 return i;
2956 }
2957
2958 return -1;
2959}
2960
2961// scan for po_test operand[0] modification in range given
2962static int scan_for_mod_opr0(struct parsed_op *po_test,
2963 int i, int opcnt)
2964{
2965 for (; i < opcnt; i++) {
2966 if (is_opr_modified(&po_test->operand[0], &ops[i]))
2967 return i;
2968 }
2969
2970 return -1;
2971}
2972
2973static int scan_for_flag_set(int i, int magic, int *branched,
2974 int *setters, int *setter_cnt)
2975{
2976 struct label_ref *lr;
2977 int ret;
2978
2979 while (i >= 0) {
2980 if (ops[i].cc_scratch == magic) {
2981 // is this a problem?
2982 //ferr(&ops[i], "%s looped\n", __func__);
2983 return 0;
2984 }
2985 ops[i].cc_scratch = magic;
2986
2987 if (g_labels[i] != NULL) {
2988 *branched = 1;
2989
2990 lr = &g_label_refs[i];
2991 for (; lr->next; lr = lr->next) {
2992 check_i(&ops[i], lr->i);
2993 ret = scan_for_flag_set(lr->i, magic,
2994 branched, setters, setter_cnt);
2995 if (ret < 0)
2996 return ret;
2997 }
2998
2999 check_i(&ops[i], lr->i);
3000 if (i > 0 && LAST_OP(i - 1)) {
3001 i = lr->i;
3002 continue;
3003 }
3004 ret = scan_for_flag_set(lr->i, magic,
3005 branched, setters, setter_cnt);
3006 if (ret < 0)
3007 return ret;
3008 }
3009 i--;
3010
3011 if (ops[i].flags & OPF_FLAGS) {
3012 setters[*setter_cnt] = i;
3013 (*setter_cnt)++;
3014 return 0;
3015 }
3016
3017 if ((ops[i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP)
3018 return -1;
3019 }
3020
3021 return -1;
3022}
3023
3024// scan back for cdq, if anything modifies edx, fail
3025static int scan_for_cdq_edx(int i)
3026{
3027 while (i >= 0) {
3028 if (g_labels[i] != NULL) {
3029 if (g_label_refs[i].next != NULL)
3030 return -1;
3031 if (i > 0 && LAST_OP(i - 1)) {
3032 i = g_label_refs[i].i;
3033 continue;
3034 }
3035 return -1;
3036 }
3037 i--;
3038
3039 if (ops[i].op == OP_CDQ)
3040 return i;
3041
3042 if (ops[i].regmask_dst & (1 << xDX))
3043 return -1;
3044 }
3045
3046 return -1;
3047}
3048
3049static int scan_for_reg_clear(int i, int reg)
3050{
3051 while (i >= 0) {
3052 if (g_labels[i] != NULL) {
3053 if (g_label_refs[i].next != NULL)
3054 return -1;
3055 if (i > 0 && LAST_OP(i - 1)) {
3056 i = g_label_refs[i].i;
3057 continue;
3058 }
3059 return -1;
3060 }
3061 i--;
3062
3063 if (ops[i].op == OP_XOR
3064 && ops[i].operand[0].lmod == OPLM_DWORD
3065 && ops[i].operand[0].reg == ops[i].operand[1].reg
3066 && ops[i].operand[0].reg == reg)
3067 return i;
3068
3069 if (ops[i].regmask_dst & (1 << reg))
3070 return -1;
3071 }
3072
3073 return -1;
3074}
3075
3076static void patch_esp_adjust(struct parsed_op *po, int adj)
3077{
3078 ferr_assert(po, po->op == OP_ADD);
3079 ferr_assert(po, IS(opr_name(po, 0), "esp"));
3080 ferr_assert(po, po->operand[1].type == OPT_CONST);
3081
3082 // this is a bit of a hack, but deals with use of
3083 // single adj for multiple calls
3084 po->operand[1].val -= adj;
3085 po->flags |= OPF_RMD;
3086 if (po->operand[1].val == 0)
3087 po->flags |= OPF_DONE;
3088 ferr_assert(po, (int)po->operand[1].val >= 0);
3089}
3090
3091// scan for positive, constant esp adjust
3092// multipath case is preliminary
3093static int scan_for_esp_adjust(int i, int opcnt,
3094 int adj_expect, int *adj, int *is_multipath, int do_update)
3095{
3096 int adj_expect_unknown = 0;
3097 struct parsed_op *po;
3098 int first_pop = -1;
3099 int adj_best = 0;
3100
3101 *adj = *is_multipath = 0;
3102 if (adj_expect < 0) {
3103 adj_expect_unknown = 1;
3104 adj_expect = 32 * 4; // enough?
3105 }
3106
3107 for (; i < opcnt && *adj < adj_expect; i++) {
3108 if (g_labels[i] != NULL)
3109 *is_multipath = 1;
3110
3111 po = &ops[i];
3112 if (po->flags & OPF_DONE)
3113 continue;
3114
3115 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
3116 if (po->operand[1].type != OPT_CONST)
3117 ferr(&ops[i], "non-const esp adjust?\n");
3118 *adj += po->operand[1].val;
3119 if (*adj & 3)
3120 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
3121 if (do_update) {
3122 if (!*is_multipath)
3123 patch_esp_adjust(po, adj_expect);
3124 else
3125 po->flags |= OPF_RMD;
3126 }
3127 return i;
3128 }
3129 else if (po->op == OP_PUSH) {
3130 //if (first_pop == -1)
3131 // first_pop = -2; // none
3132 *adj -= lmod_bytes(po, po->operand[0].lmod);
3133 }
3134 else if (po->op == OP_POP) {
3135 if (!(po->flags & OPF_DONE)) {
3136 // seems like msvc only uses 'pop ecx' for stack realignment..
3137 if (po->operand[0].type != OPT_REG || po->operand[0].reg != xCX)
3138 break;
3139 if (first_pop == -1 && *adj >= 0)
3140 first_pop = i;
3141 }
3142 if (do_update && *adj >= 0) {
3143 po->flags |= OPF_RMD;
3144 if (!*is_multipath)
3145 po->flags |= OPF_DONE | OPF_NOREGS;
3146 }
3147
3148 *adj += lmod_bytes(po, po->operand[0].lmod);
3149 if (*adj > adj_best)
3150 adj_best = *adj;
3151 }
3152 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
3153 if (po->op == OP_JMP && po->btj == NULL) {
3154 if (po->bt_i <= i)
3155 break;
3156 i = po->bt_i - 1;
3157 continue;
3158 }
3159 if (po->op != OP_CALL)
3160 break;
3161 if (po->operand[0].type != OPT_LABEL)
3162 break;
3163 if (po->pp != NULL && po->pp->is_stdcall)
3164 break;
3165 if (adj_expect_unknown && first_pop >= 0)
3166 break;
3167 // assume it's another cdecl call
3168 }
3169 }
3170
3171 if (first_pop >= 0) {
3172 // probably only 'pop ecx' was used
3173 *adj = adj_best;
3174 return first_pop;
3175 }
3176
3177 return -1;
3178}
3179
3180static void scan_fwd_set_flags(int i, int opcnt, int magic, int flags)
3181{
3182 struct parsed_op *po;
3183 int j;
3184
3185 if (i < 0)
3186 ferr(ops, "%s: followed bad branch?\n", __func__);
3187
3188 for (; i < opcnt; i++) {
3189 po = &ops[i];
3190 if (po->cc_scratch == magic)
3191 return;
3192 po->cc_scratch = magic;
3193 po->flags |= flags;
3194
3195 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
3196 if (po->btj != NULL) {
3197 // jumptable
3198 for (j = 0; j < po->btj->count; j++)
3199 scan_fwd_set_flags(po->btj->d[j].bt_i, opcnt, magic, flags);
3200 return;
3201 }
3202
3203 scan_fwd_set_flags(po->bt_i, opcnt, magic, flags);
3204 if (!(po->flags & OPF_CJMP))
3205 return;
3206 }
3207 if (po->flags & OPF_TAIL)
3208 return;
3209 }
3210}
3211
3212static const struct parsed_proto *try_recover_pp(
3213 struct parsed_op *po, const struct parsed_opr *opr, int *search_instead)
3214{
3215 const struct parsed_proto *pp = NULL;
3216 char buf[256];
3217 char *p;
3218
3219 // maybe an arg of g_func?
3220 if (opr->type == OPT_REGMEM && is_stack_access(po, opr))
3221 {
3222 char ofs_reg[16] = { 0, };
3223 int arg, arg_s, arg_i;
3224 int stack_ra = 0;
3225 int offset = 0;
3226
3227 if (g_header_mode)
3228 return NULL;
3229
3230 parse_stack_access(po, opr->name, ofs_reg,
3231 &offset, &stack_ra, NULL, 0);
3232 if (ofs_reg[0] != 0)
3233 ferr(po, "offset reg on arg access?\n");
3234 if (offset <= stack_ra) {
3235 // search who set the stack var instead
3236 if (search_instead != NULL)
3237 *search_instead = 1;
3238 return NULL;
3239 }
3240
3241 arg_i = (offset - stack_ra - 4) / 4;
3242 for (arg = arg_s = 0; arg < g_func_pp->argc; arg++) {
3243 if (g_func_pp->arg[arg].reg != NULL)
3244 continue;
3245 if (arg_s == arg_i)
3246 break;
3247 arg_s++;
3248 }
3249 if (arg == g_func_pp->argc)
3250 ferr(po, "stack arg %d not in prototype?\n", arg_i);
3251
3252 pp = g_func_pp->arg[arg].fptr;
3253 if (pp == NULL)
3254 ferr(po, "icall sa: arg%d is not a fptr?\n", arg + 1);
3255 check_func_pp(po, pp, "icall arg");
3256 }
3257 else if (opr->type == OPT_REGMEM && strchr(opr->name + 1, '[')) {
3258 // label[index]
3259 p = strchr(opr->name + 1, '[');
3260 memcpy(buf, opr->name, p - opr->name);
3261 buf[p - opr->name] = 0;
3262 pp = proto_parse(g_fhdr, buf, g_quiet_pp);
3263 }
3264 else if (opr->type == OPT_OFFSET || opr->type == OPT_LABEL) {
3265 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
3266 if (pp == NULL) {
3267 if (!g_header_mode)
3268 ferr(po, "proto_parse failed for icall to '%s'\n", opr->name);
3269 }
3270 else
3271 check_func_pp(po, pp, "reg-fptr ref");
3272 }
3273
3274 return pp;
3275}
3276
3277static void scan_for_call_type(int i, const struct parsed_opr *opr,
3278 int magic, const struct parsed_proto **pp_found, int *pp_i,
3279 int *multi)
3280{
3281 const struct parsed_proto *pp = NULL;
3282 struct parsed_op *po;
3283 struct label_ref *lr;
3284
3285 ops[i].cc_scratch = magic;
3286
3287 while (1) {
3288 if (g_labels[i] != NULL) {
3289 lr = &g_label_refs[i];
3290 for (; lr != NULL; lr = lr->next) {
3291 check_i(&ops[i], lr->i);
3292 scan_for_call_type(lr->i, opr, magic, pp_found, pp_i, multi);
3293 }
3294 if (i > 0 && LAST_OP(i - 1))
3295 return;
3296 }
3297
3298 i--;
3299 if (i < 0)
3300 break;
3301
3302 if (ops[i].cc_scratch == magic)
3303 return;
3304 ops[i].cc_scratch = magic;
3305
3306 if (!(ops[i].flags & OPF_DATA))
3307 continue;
3308 if (!is_opr_modified(opr, &ops[i]))
3309 continue;
3310 if (ops[i].op != OP_MOV && ops[i].op != OP_LEA) {
3311 // most probably trashed by some processing
3312 *pp_found = NULL;
3313 return;
3314 }
3315
3316 opr = &ops[i].operand[1];
3317 if (opr->type != OPT_REG)
3318 break;
3319 }
3320
3321 po = (i >= 0) ? &ops[i] : ops;
3322
3323 if (i < 0) {
3324 // reached the top - can only be an arg-reg
3325 if (opr->type != OPT_REG || g_func_pp == NULL)
3326 return;
3327
3328 for (i = 0; i < g_func_pp->argc; i++) {
3329 if (g_func_pp->arg[i].reg == NULL)
3330 continue;
3331 if (IS(opr->name, g_func_pp->arg[i].reg))
3332 break;
3333 }
3334 if (i == g_func_pp->argc)
3335 return;
3336 pp = g_func_pp->arg[i].fptr;
3337 if (pp == NULL)
3338 ferr(po, "icall: arg%d (%s) is not a fptr?\n",
3339 i + 1, g_func_pp->arg[i].reg);
3340 check_func_pp(po, pp, "icall reg-arg");
3341 }
3342 else
3343 pp = try_recover_pp(po, opr, NULL);
3344
3345 if (*pp_found != NULL && pp != NULL && *pp_found != pp) {
3346 if (!IS((*pp_found)->ret_type.name, pp->ret_type.name)
3347 || (*pp_found)->is_stdcall != pp->is_stdcall
3348 || (*pp_found)->is_fptr != pp->is_fptr
3349 || (*pp_found)->argc != pp->argc
3350 || (*pp_found)->argc_reg != pp->argc_reg
3351 || (*pp_found)->argc_stack != pp->argc_stack)
3352 {
3353 ferr(po, "icall: parsed_proto mismatch\n");
3354 }
3355 *multi = 1;
3356 }
3357 if (pp != NULL) {
3358 *pp_found = pp;
3359 *pp_i = po - ops;
3360 }
3361}
3362
3363static void add_label_ref(struct label_ref *lr, int op_i)
3364{
3365 struct label_ref *lr_new;
3366
3367 if (lr->i == -1) {
3368 lr->i = op_i;
3369 return;
3370 }
3371
3372 lr_new = calloc(1, sizeof(*lr_new));
3373 lr_new->i = op_i;
3374 lr_new->next = lr->next;
3375 lr->next = lr_new;
3376}
3377
3378static struct parsed_data *try_resolve_jumptab(int i, int opcnt)
3379{
3380 struct parsed_op *po = &ops[i];
3381 struct parsed_data *pd;
3382 char label[NAMELEN], *p;
3383 int len, j, l;
3384
3385 p = strchr(po->operand[0].name, '[');
3386 if (p == NULL)
3387 return NULL;
3388
3389 len = p - po->operand[0].name;
3390 strncpy(label, po->operand[0].name, len);
3391 label[len] = 0;
3392
3393 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
3394 if (IS(g_func_pd[j].label, label)) {
3395 pd = &g_func_pd[j];
3396 break;
3397 }
3398 }
3399 if (pd == NULL)
3400 //ferr(po, "label '%s' not parsed?\n", label);
3401 return NULL;
3402
3403 if (pd->type != OPT_OFFSET)
3404 ferr(po, "label '%s' with non-offset data?\n", label);
3405
3406 // find all labels, link
3407 for (j = 0; j < pd->count; j++) {
3408 for (l = 0; l < opcnt; l++) {
3409 if (g_labels[l] != NULL && IS(g_labels[l], pd->d[j].u.label)) {
3410 add_label_ref(&g_label_refs[l], i);
3411 pd->d[j].bt_i = l;
3412 break;
3413 }
3414 }
3415 }
3416
3417 return pd;
3418}
3419
3420static void clear_labels(int count)
3421{
3422 int i;
3423
3424 for (i = 0; i < count; i++) {
3425 if (g_labels[i] != NULL) {
3426 free(g_labels[i]);
3427 g_labels[i] = NULL;
3428 }
3429 }
3430}
3431
3432static int get_pp_arg_regmask_src(const struct parsed_proto *pp)
3433{
3434 int regmask = 0;
3435 int i, reg;
3436
3437 for (i = 0; i < pp->argc; i++) {
3438 if (pp->arg[i].reg != NULL) {
3439 reg = char_array_i(regs_r32,
3440 ARRAY_SIZE(regs_r32), pp->arg[i].reg);
3441 if (reg < 0)
3442 ferr(ops, "arg '%s' of func '%s' is not a reg?\n",
3443 pp->arg[i].reg, pp->name);
3444 regmask |= 1 << reg;
3445 }
3446 }
3447
3448 return regmask;
3449}
3450
3451static int get_pp_arg_regmask_dst(const struct parsed_proto *pp)
3452{
3453 if (strstr(pp->ret_type.name, "int64"))
3454 return (1 << xAX) | (1 << xDX);
3455 if (IS(pp->ret_type.name, "float")
3456 || IS(pp->ret_type.name, "double"))
3457 {
3458 return mxST0;
3459 }
3460 if (strcasecmp(pp->ret_type.name, "void") == 0)
3461 return 0;
3462
3463 return mxAX;
3464}
3465
3466static void resolve_branches_parse_calls(int opcnt)
3467{
3468 static const struct {
3469 const char *name;
3470 enum op_op op;
3471 unsigned int flags;
3472 unsigned int regmask_src;
3473 unsigned int regmask_dst;
3474 } pseudo_ops[] = {
3475 { "__ftol", OPP_FTOL, OPF_FPOP, mxST0, mxAX | mxDX },
3476 };
3477 const struct parsed_proto *pp_c;
3478 struct parsed_proto *pp;
3479 struct parsed_data *pd;
3480 struct parsed_op *po;
3481 const char *tmpname;
3482 int i, l;
3483 int ret;
3484
3485 for (i = 0; i < opcnt; i++)
3486 {
3487 po = &ops[i];
3488 po->bt_i = -1;
3489 po->btj = NULL;
3490
3491 if (po->op == OP_CALL) {
3492 pp = NULL;
3493
3494 if (po->operand[0].type == OPT_LABEL) {
3495 tmpname = opr_name(po, 0);
3496 if (IS_START(tmpname, "loc_"))
3497 ferr(po, "call to loc_*\n");
3498
3499 // convert some calls to pseudo-ops
3500 for (l = 0; l < ARRAY_SIZE(pseudo_ops); l++) {
3501 if (!IS(tmpname, pseudo_ops[l].name))
3502 continue;
3503
3504 po->op = pseudo_ops[l].op;
3505 po->operand_cnt = 0;
3506 po->regmask_src = pseudo_ops[l].regmask_src;
3507 po->regmask_dst = pseudo_ops[l].regmask_dst;
3508 po->flags = pseudo_ops[l].flags;
3509 po->flags |= po->regmask_dst ? OPF_DATA : 0;
3510 break;
3511 }
3512 if (l < ARRAY_SIZE(pseudo_ops))
3513 continue;
3514
3515 pp_c = proto_parse(g_fhdr, tmpname, g_header_mode);
3516 if (!g_header_mode && pp_c == NULL)
3517 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
3518
3519 if (pp_c != NULL) {
3520 pp = proto_clone(pp_c);
3521 my_assert_not(pp, NULL);
3522 }
3523 }
3524 else if (po->datap != NULL) {
3525 pp = calloc(1, sizeof(*pp));
3526 my_assert_not(pp, NULL);
3527
3528 ret = parse_protostr(po->datap, pp);
3529 if (ret < 0)
3530 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
3531 free(po->datap);
3532 po->datap = NULL;
3533 }
3534
3535 if (pp != NULL) {
3536 if (pp->is_fptr)
3537 check_func_pp(po, pp, "fptr var call");
3538 if (pp->is_noreturn)
3539 po->flags |= OPF_TAIL;
3540 }
3541 po->pp = pp;
3542 continue;
3543 }
3544
3545 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
3546 continue;
3547
3548 if (po->operand[0].type == OPT_REGMEM) {
3549 pd = try_resolve_jumptab(i, opcnt);
3550 if (pd == NULL)
3551 goto tailcall;
3552
3553 po->btj = pd;
3554 continue;
3555 }
3556
3557 for (l = 0; l < opcnt; l++) {
3558 if (g_labels[l] != NULL
3559 && IS(po->operand[0].name, g_labels[l]))
3560 {
3561 if (l == i + 1 && po->op == OP_JMP) {
3562 // yet another alignment type..
3563 po->flags |= OPF_RMD|OPF_DONE;
3564 break;
3565 }
3566 add_label_ref(&g_label_refs[l], i);
3567 po->bt_i = l;
3568 break;
3569 }
3570 }
3571
3572 if (po->bt_i != -1 || (po->flags & OPF_RMD))
3573 continue;
3574
3575 if (po->operand[0].type == OPT_LABEL)
3576 // assume tail call
3577 goto tailcall;
3578
3579 ferr(po, "unhandled branch\n");
3580
3581tailcall:
3582 po->op = OP_CALL;
3583 po->flags |= OPF_TAIL;
3584 if (i > 0 && ops[i - 1].op == OP_POP)
3585 po->flags |= OPF_ATAIL;
3586 i--; // reprocess
3587 }
3588}
3589
3590static void scan_prologue_epilogue(int opcnt)
3591{
3592 int ecx_push = 0, esp_sub = 0;
3593 int found;
3594 int i, j, l;
3595
3596 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
3597 && ops[1].op == OP_MOV
3598 && IS(opr_name(&ops[1], 0), "ebp")
3599 && IS(opr_name(&ops[1], 1), "esp"))
3600 {
3601 g_bp_frame = 1;
3602 ops[0].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3603 ops[1].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3604 i = 2;
3605
3606 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
3607 g_stack_fsz = opr_const(&ops[2], 1);
3608 ops[2].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3609 i++;
3610 }
3611 else {
3612 // another way msvc builds stack frame..
3613 i = 2;
3614 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
3615 g_stack_fsz += 4;
3616 ops[i].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3617 ecx_push++;
3618 i++;
3619 }
3620 // and another way..
3621 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
3622 && ops[i].operand[1].type == OPT_CONST
3623 && ops[i + 1].op == OP_CALL
3624 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
3625 {
3626 g_stack_fsz += ops[i].operand[1].val;
3627 ops[i].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3628 i++;
3629 ops[i].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3630 i++;
3631 }
3632 }
3633
3634 found = 0;
3635 do {
3636 for (; i < opcnt; i++)
3637 if (ops[i].flags & OPF_TAIL)
3638 break;
3639 j = i - 1;
3640 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
3641 if (ops[j].bt_i != -1 || ops[j].btj != NULL)
3642 break;
3643 i--;
3644 j--;
3645 }
3646
3647 if ((ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ebp"))
3648 || ops[j].op == OP_LEAVE)
3649 {
3650 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3651 }
3652 else if (ops[i].op == OP_CALL && ops[i].pp != NULL
3653 && ops[i].pp->is_noreturn)
3654 {
3655 // on noreturn, msvc sometimes cleans stack, sometimes not
3656 i++;
3657 found = 1;
3658 continue;
3659 }
3660 else if (!(g_ida_func_attr & IDAFA_NORETURN))
3661 ferr(&ops[j], "'pop ebp' expected\n");
3662
3663 if (g_stack_fsz != 0) {
3664 if (ops[j].op == OP_LEAVE)
3665 j--;
3666 else if (ops[j].op == OP_POP
3667 && ops[j - 1].op == OP_MOV
3668 && IS(opr_name(&ops[j - 1], 0), "esp")
3669 && IS(opr_name(&ops[j - 1], 1), "ebp"))
3670 {
3671 ops[j - 1].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3672 j -= 2;
3673 }
3674 else if (!(g_ida_func_attr & IDAFA_NORETURN))
3675 {
3676 ferr(&ops[j], "esp restore expected\n");
3677 }
3678
3679 if (ecx_push && j >= 0 && ops[j].op == OP_POP
3680 && IS(opr_name(&ops[j], 0), "ecx"))
3681 {
3682 ferr(&ops[j], "unexpected ecx pop\n");
3683 }
3684 }
3685
3686 found = 1;
3687 i++;
3688 } while (i < opcnt);
3689
3690 if (!found)
3691 ferr(ops, "missing ebp epilogue\n");
3692 return;
3693 }
3694
3695 // non-bp frame
3696 i = 0;
3697 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
3698 ops[i].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3699 g_stack_fsz += 4;
3700 ecx_push++;
3701 i++;
3702 }
3703
3704 for (; i < opcnt; i++) {
3705 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
3706 break;
3707 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
3708 && ops[i].operand[1].type == OPT_CONST)
3709 {
3710 g_stack_fsz = ops[i].operand[1].val;
3711 ops[i].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3712 esp_sub = 1;
3713 break;
3714 }
3715 }
3716
3717 if (ecx_push && !esp_sub) {
3718 // could actually be args for a call..
3719 for (; i < opcnt; i++)
3720 if (ops[i].op != OP_PUSH)
3721 break;
3722
3723 if (ops[i].op == OP_CALL && ops[i].operand[0].type == OPT_LABEL) {
3724 const struct parsed_proto *pp;
3725 pp = proto_parse(g_fhdr, opr_name(&ops[i], 0), 1);
3726 j = pp ? pp->argc_stack : 0;
3727 while (i > 0 && j > 0) {
3728 i--;
3729 if (ops[i].op == OP_PUSH) {
3730 ops[i].flags &= ~(OPF_RMD | OPF_DONE | OPF_NOREGS);
3731 j--;
3732 }
3733 }
3734 if (j != 0)
3735 ferr(&ops[i], "unhandled prologue\n");
3736
3737 // recheck
3738 i = g_stack_fsz = ecx_push = 0;
3739 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
3740 if (!(ops[i].flags & OPF_RMD))
3741 break;
3742 g_stack_fsz += 4;
3743 ecx_push++;
3744 i++;
3745 }
3746 }
3747 }
3748
3749 found = 0;
3750 if (ecx_push || esp_sub)
3751 {
3752 g_sp_frame = 1;
3753
3754 i++;
3755 do {
3756 for (; i < opcnt; i++)
3757 if (ops[i].flags & OPF_TAIL)
3758 break;
3759 j = i - 1;
3760 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
3761 if (ops[j].bt_i != -1 || ops[j].btj != NULL)
3762 break;
3763 i--;
3764 j--;
3765 }
3766
3767 if (ecx_push > 0) {
3768 for (l = 0; l < ecx_push; l++) {
3769 if (ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ecx"))
3770 /* pop ecx */;
3771 else if (ops[j].op == OP_ADD
3772 && IS(opr_name(&ops[j], 0), "esp")
3773 && ops[j].operand[1].type == OPT_CONST)
3774 {
3775 /* add esp, N */
3776 l += ops[j].operand[1].val / 4 - 1;
3777 }
3778 else
3779 ferr(&ops[j], "'pop ecx' expected\n");
3780
3781 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3782 j--;
3783 }
3784 if (l != ecx_push)
3785 ferr(&ops[j], "epilogue scan failed\n");
3786
3787 found = 1;
3788 }
3789
3790 if (esp_sub) {
3791 if (ops[j].op != OP_ADD
3792 || !IS(opr_name(&ops[j], 0), "esp")
3793 || ops[j].operand[1].type != OPT_CONST
3794 || ops[j].operand[1].val != g_stack_fsz)
3795 ferr(&ops[j], "'add esp' expected\n");
3796
3797 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3798 ops[j].operand[1].val = 0; // hack for stack arg scanner
3799 found = 1;
3800 }
3801
3802 i++;
3803 } while (i < opcnt);
3804
3805 if (!found)
3806 ferr(ops, "missing esp epilogue\n");
3807 }
3808}
3809
3810static const struct parsed_proto *resolve_icall(int i, int opcnt,
3811 int *pp_i, int *multi_src)
3812{
3813 const struct parsed_proto *pp = NULL;
3814 int search_advice = 0;
3815
3816 *multi_src = 0;
3817 *pp_i = -1;
3818
3819 switch (ops[i].operand[0].type) {
3820 case OPT_REGMEM:
3821 case OPT_LABEL:
3822 case OPT_OFFSET:
3823 pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice);
3824 if (!search_advice)
3825 break;
3826 // fallthrough
3827 default:
3828 scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp,
3829 pp_i, multi_src);
3830 break;
3831 }
3832
3833 return pp;
3834}
3835
3836// find an instruction that changed opr before i op
3837// *op_i must be set to -1 by the caller
3838// *entry is set to 1 if one source is determined to be the caller
3839// returns 1 if found, *op_i is then set to origin
3840static int resolve_origin(int i, const struct parsed_opr *opr,
3841 int magic, int *op_i, int *is_caller)
3842{
3843 struct label_ref *lr;
3844 int ret = 0;
3845
3846 if (ops[i].cc_scratch == magic)
3847 return 0;
3848 ops[i].cc_scratch = magic;
3849
3850 while (1) {
3851 if (g_labels[i] != NULL) {
3852 lr = &g_label_refs[i];
3853 for (; lr != NULL; lr = lr->next) {
3854 check_i(&ops[i], lr->i);
3855 ret |= resolve_origin(lr->i, opr, magic, op_i, is_caller);
3856 }
3857 if (i > 0 && LAST_OP(i - 1))
3858 return ret;
3859 }
3860
3861 i--;
3862 if (i < 0) {
3863 if (is_caller != NULL)
3864 *is_caller = 1;
3865 return -1;
3866 }
3867
3868 if (ops[i].cc_scratch == magic)
3869 return ret;
3870 ops[i].cc_scratch = magic;
3871
3872 if (!(ops[i].flags & OPF_DATA))
3873 continue;
3874 if (!is_opr_modified(opr, &ops[i]))
3875 continue;
3876
3877 if (*op_i >= 0) {
3878 if (*op_i == i)
3879 return ret | 1;
3880
3881 // XXX: could check if the other op does the same
3882 return -1;
3883 }
3884
3885 *op_i = i;
3886 return ret | 1;
3887 }
3888}
3889
3890// find an instruction that previously referenced opr
3891// if multiple results are found - fail
3892// *op_i must be set to -1 by the caller
3893// returns 1 if found, *op_i is then set to referencer insn
3894static int resolve_last_ref(int i, const struct parsed_opr *opr,
3895 int magic, int *op_i)
3896{
3897 struct label_ref *lr;
3898 int ret = 0;
3899
3900 if (ops[i].cc_scratch == magic)
3901 return 0;
3902 ops[i].cc_scratch = magic;
3903
3904 while (1) {
3905 if (g_labels[i] != NULL) {
3906 lr = &g_label_refs[i];
3907 for (; lr != NULL; lr = lr->next) {
3908 check_i(&ops[i], lr->i);
3909 ret |= resolve_last_ref(lr->i, opr, magic, op_i);
3910 }
3911 if (i > 0 && LAST_OP(i - 1))
3912 return ret;
3913 }
3914
3915 i--;
3916 if (i < 0)
3917 return -1;
3918
3919 if (ops[i].cc_scratch == magic)
3920 return 0;
3921 ops[i].cc_scratch = magic;
3922
3923 if (!is_opr_referenced(opr, &ops[i]))
3924 continue;
3925
3926 if (*op_i >= 0)
3927 return -1;
3928
3929 *op_i = i;
3930 return 1;
3931 }
3932}
3933
3934// find next instruction that reads opr
3935// *op_i must be set to -1 by the caller
3936// on return, *op_i is set to first referencer insn
3937// returns 1 if exactly 1 referencer is found
3938static int find_next_read(int i, int opcnt,
3939 const struct parsed_opr *opr, int magic, int *op_i)
3940{
3941 struct parsed_op *po;
3942 int j, ret = 0;
3943
3944 for (; i < opcnt; i++)
3945 {
3946 if (ops[i].cc_scratch == magic)
3947 return ret;
3948 ops[i].cc_scratch = magic;
3949
3950 po = &ops[i];
3951 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
3952 if (po->btj != NULL) {
3953 // jumptable
3954 for (j = 0; j < po->btj->count; j++) {
3955 check_i(po, po->btj->d[j].bt_i);
3956 ret |= find_next_read(po->btj->d[j].bt_i, opcnt, opr,
3957 magic, op_i);
3958 }
3959 return ret;
3960 }
3961
3962 if (po->flags & OPF_RMD)
3963 continue;
3964 check_i(po, po->bt_i);
3965 if (po->flags & OPF_CJMP) {
3966 ret |= find_next_read(po->bt_i, opcnt, opr, magic, op_i);
3967 if (ret < 0)
3968 return ret;
3969 }
3970 else
3971 i = po->bt_i - 1;
3972 continue;
3973 }
3974
3975 if (!is_opr_read(opr, po)) {
3976 if (is_opr_modified(opr, po)
3977 && (po->op == OP_CALL
3978 || ((po->flags & OPF_DATA)
3979 && po->operand[0].lmod == OPLM_DWORD)))
3980 {
3981 // it's overwritten
3982 return ret;
3983 }
3984 if (po->flags & OPF_TAIL)
3985 return ret;
3986 continue;
3987 }
3988
3989 if (*op_i >= 0)
3990 return -1;
3991
3992 *op_i = i;
3993 return 1;
3994 }
3995
3996 return 0;
3997}
3998
3999static int try_resolve_const(int i, const struct parsed_opr *opr,
4000 int magic, unsigned int *val)
4001{
4002 int s_i = -1;
4003 int ret;
4004
4005 ret = resolve_origin(i, opr, magic, &s_i, NULL);
4006 if (ret == 1) {
4007 i = s_i;
4008 if (ops[i].op != OP_MOV && ops[i].operand[1].type != OPT_CONST)
4009 return -1;
4010
4011 *val = ops[i].operand[1].val;
4012 return 1;
4013 }
4014
4015 return -1;
4016}
4017
4018static struct parsed_proto *process_call_early(int i, int opcnt,
4019 int *adj_i)
4020{
4021 struct parsed_op *po = &ops[i];
4022 struct parsed_proto *pp;
4023 int multipath = 0;
4024 int adj = 0;
4025 int j, ret;
4026
4027 pp = po->pp;
4028 if (pp == NULL || pp->is_vararg || pp->argc_reg != 0)
4029 // leave for later
4030 return NULL;
4031
4032 // look for and make use of esp adjust
4033 *adj_i = ret = -1;
4034 if (!pp->is_stdcall && pp->argc_stack > 0)
4035 ret = scan_for_esp_adjust(i + 1, opcnt,
4036 pp->argc_stack * 4, &adj, &multipath, 0);
4037 if (ret >= 0) {
4038 if (pp->argc_stack > adj / 4)
4039 return NULL;
4040 if (multipath)
4041 return NULL;
4042 if (ops[ret].op == OP_POP) {
4043 for (j = 1; j < adj / 4; j++) {
4044 if (ops[ret + j].op != OP_POP
4045 || ops[ret + j].operand[0].reg != xCX)
4046 {
4047 return NULL;
4048 }
4049 }
4050 }
4051 }
4052
4053 *adj_i = ret;
4054 return pp;
4055}
4056
4057static struct parsed_proto *process_call(int i, int opcnt)
4058{
4059 struct parsed_op *po = &ops[i];
4060 const struct parsed_proto *pp_c;
4061 struct parsed_proto *pp;
4062 const char *tmpname;
4063 int call_i = -1, ref_i = -1;
4064 int adj = 0, multipath = 0;
4065 int ret, arg;
4066
4067 tmpname = opr_name(po, 0);
4068 pp = po->pp;
4069 if (pp == NULL)
4070 {
4071 // indirect call
4072 pp_c = resolve_icall(i, opcnt, &call_i, &multipath);
4073 if (pp_c != NULL) {
4074 if (!pp_c->is_func && !pp_c->is_fptr)
4075 ferr(po, "call to non-func: %s\n", pp_c->name);
4076 pp = proto_clone(pp_c);
4077 my_assert_not(pp, NULL);
4078 if (multipath)
4079 // not resolved just to single func
4080 pp->is_fptr = 1;
4081
4082 switch (po->operand[0].type) {
4083 case OPT_REG:
4084 // we resolved this call and no longer need the register
4085 po->regmask_src &= ~(1 << po->operand[0].reg);
4086
4087 if (!multipath && i != call_i && ops[call_i].op == OP_MOV
4088 && ops[call_i].operand[1].type == OPT_LABEL)
4089 {
4090 // no other source users?
4091 ret = resolve_last_ref(i, &po->operand[0], i + opcnt * 10,
4092 &ref_i);
4093 if (ret == 1 && call_i == ref_i) {
4094 // and nothing uses it after us?
4095 ref_i = -1;
4096 find_next_read(i + 1, opcnt, &po->operand[0],
4097 i + opcnt * 11, &ref_i);
4098 if (ref_i == -1)
4099 // then also don't need the source mov
4100 ops[call_i].flags |= OPF_RMD | OPF_NOREGS;
4101 }
4102 }
4103 break;
4104 case OPT_REGMEM:
4105 pp->is_fptr = 1;
4106 break;
4107 default:
4108 break;
4109 }
4110 }
4111 if (pp == NULL) {
4112 pp = calloc(1, sizeof(*pp));
4113 my_assert_not(pp, NULL);
4114
4115 pp->is_fptr = 1;
4116 ret = scan_for_esp_adjust(i + 1, opcnt,
4117 -1, &adj, &multipath, 0);
4118 if (ret < 0 || adj < 0) {
4119 if (!g_allow_regfunc)
4120 ferr(po, "non-__cdecl indirect call unhandled yet\n");
4121 pp->is_unresolved = 1;
4122 adj = 0;
4123 }
4124 adj /= 4;
4125 if (adj > ARRAY_SIZE(pp->arg))
4126 ferr(po, "esp adjust too large: %d\n", adj);
4127 pp->ret_type.name = strdup("int");
4128 pp->argc = pp->argc_stack = adj;
4129 for (arg = 0; arg < pp->argc; arg++)
4130 pp->arg[arg].type.name = strdup("int");
4131 }
4132 po->pp = pp;
4133 }
4134
4135 // look for and make use of esp adjust
4136 multipath = 0;
4137 ret = -1;
4138 if (!pp->is_stdcall && pp->argc_stack > 0) {
4139 int adj_expect = pp->is_vararg ? -1 : pp->argc_stack * 4;
4140 ret = scan_for_esp_adjust(i + 1, opcnt,
4141 adj_expect, &adj, &multipath, 0);
4142 }
4143 if (ret >= 0) {
4144 if (pp->is_vararg) {
4145 if (adj / 4 < pp->argc_stack) {
4146 fnote(po, "(this call)\n");
4147 ferr(&ops[ret], "esp adjust is too small: %x < %x\n",
4148 adj, pp->argc_stack * 4);
4149 }
4150 // modify pp to make it have varargs as normal args
4151 arg = pp->argc;
4152 pp->argc += adj / 4 - pp->argc_stack;
4153 for (; arg < pp->argc; arg++) {
4154 pp->arg[arg].type.name = strdup("int");
4155 pp->argc_stack++;
4156 }
4157 if (pp->argc > ARRAY_SIZE(pp->arg))
4158 ferr(po, "too many args for '%s'\n", tmpname);
4159 }
4160 if (pp->argc_stack > adj / 4) {
4161 fnote(po, "(this call)\n");
4162 ferr(&ops[ret], "stack tracking failed for '%s': %x %x\n",
4163 tmpname, pp->argc_stack * 4, adj);
4164 }
4165
4166 scan_for_esp_adjust(i + 1, opcnt,
4167 pp->argc_stack * 4, &adj, &multipath, 1);
4168 }
4169 else if (pp->is_vararg)
4170 ferr(po, "missing esp_adjust for vararg func '%s'\n",
4171 pp->name);
4172
4173 return pp;
4174}
4175
4176static int collect_call_args_early(struct parsed_op *po, int i,
4177 struct parsed_proto *pp, int *regmask)
4178{
4179 int arg, ret;
4180 int j;
4181
4182 for (arg = 0; arg < pp->argc; arg++)
4183 if (pp->arg[arg].reg == NULL)
4184 break;
4185
4186 // first see if it can be easily done
4187 for (j = i; j > 0 && arg < pp->argc; )
4188 {
4189 if (g_labels[j] != NULL)
4190 return -1;
4191 j--;
4192
4193 if (ops[j].op == OP_CALL)
4194 return -1;
4195 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP)
4196 return -1;
4197 else if (ops[j].op == OP_POP)
4198 return -1;
4199 else if (ops[j].flags & OPF_CJMP)
4200 return -1;
4201 else if (ops[j].op == OP_PUSH) {
4202 if (ops[j].flags & (OPF_FARG|OPF_FARGNR))
4203 return -1;
4204 ret = scan_for_mod(&ops[j], j + 1, i, 1);
4205 if (ret >= 0)
4206 return -1;
4207
4208 if (pp->arg[arg].type.is_va_list)
4209 return -1;
4210
4211 // next arg
4212 for (arg++; arg < pp->argc; arg++)
4213 if (pp->arg[arg].reg == NULL)
4214 break;
4215 }
4216 }
4217
4218 if (arg < pp->argc)
4219 return -1;
4220
4221 // now do it
4222 for (arg = 0; arg < pp->argc; arg++)
4223 if (pp->arg[arg].reg == NULL)
4224 break;
4225
4226 for (j = i; j > 0 && arg < pp->argc; )
4227 {
4228 j--;
4229
4230 if (ops[j].op == OP_PUSH)
4231 {
4232 ops[j].p_argnext = -1;
4233 ferr_assert(&ops[j], pp->arg[arg].datap == NULL);
4234 pp->arg[arg].datap = &ops[j];
4235
4236 if (ops[j].operand[0].type == OPT_REG)
4237 *regmask |= 1 << ops[j].operand[0].reg;
4238
4239 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_FARGNR | OPF_FARG;
4240 ops[j].flags &= ~OPF_RSAVE;
4241
4242 // next arg
4243 for (arg++; arg < pp->argc; arg++)
4244 if (pp->arg[arg].reg == NULL)
4245 break;
4246 }
4247 }
4248
4249 return 0;
4250}
4251
4252static int collect_call_args_r(struct parsed_op *po, int i,
4253 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
4254 int *arg_grp, int arg, int magic, int need_op_saving, int may_reuse)
4255{
4256 struct parsed_proto *pp_tmp;
4257 struct parsed_op *po_tmp;
4258 struct label_ref *lr;
4259 int need_to_save_current;
4260 int arg_grp_current = 0;
4261 int save_args_seen = 0;
4262 int save_args;
4263 int ret = 0;
4264 int reg;
4265 char buf[32];
4266 int j, k;
4267
4268 if (i < 0) {
4269 ferr(po, "dead label encountered\n");
4270 return -1;
4271 }
4272
4273 for (; arg < pp->argc; arg++)
4274 if (pp->arg[arg].reg == NULL)
4275 break;
4276 magic = (magic & 0xffffff) | (arg << 24);
4277
4278 for (j = i; j >= 0 && (arg < pp->argc || pp->is_unresolved); )
4279 {
4280 if (((ops[j].cc_scratch ^ magic) & 0xffffff) == 0) {
4281 if (ops[j].cc_scratch != magic) {
4282 ferr(&ops[j], "arg collect hit same path with diff args for %s\n",
4283 pp->name);
4284 return -1;
4285 }
4286 // ok: have already been here
4287 return 0;
4288 }
4289 ops[j].cc_scratch = magic;
4290
4291 if (g_labels[j] != NULL && g_label_refs[j].i != -1) {
4292 lr = &g_label_refs[j];
4293 if (lr->next != NULL)
4294 need_op_saving = 1;
4295 for (; lr->next; lr = lr->next) {
4296 check_i(&ops[j], lr->i);
4297 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
4298 may_reuse = 1;
4299 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
4300 arg_grp, arg, magic, need_op_saving, may_reuse);
4301 if (ret < 0)
4302 return ret;
4303 }
4304
4305 check_i(&ops[j], lr->i);
4306 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
4307 may_reuse = 1;
4308 if (j > 0 && LAST_OP(j - 1)) {
4309 // follow last branch in reverse
4310 j = lr->i;
4311 continue;
4312 }
4313 need_op_saving = 1;
4314 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
4315 arg_grp, arg, magic, need_op_saving, may_reuse);
4316 if (ret < 0)
4317 return ret;
4318 }
4319 j--;
4320
4321 if (ops[j].op == OP_CALL)
4322 {
4323 if (pp->is_unresolved)
4324 break;
4325
4326 pp_tmp = ops[j].pp;
4327 if (pp_tmp == NULL)
4328 ferr(po, "arg collect hit unparsed call '%s'\n",
4329 ops[j].operand[0].name);
4330 if (may_reuse && pp_tmp->argc_stack > 0)
4331 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
4332 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
4333 }
4334 // esp adjust of 0 means we collected it before
4335 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP
4336 && (ops[j].operand[1].type != OPT_CONST
4337 || ops[j].operand[1].val != 0))
4338 {
4339 if (pp->is_unresolved)
4340 break;
4341
4342 fnote(po, "(this call)\n");
4343 ferr(&ops[j], "arg collect %d/%d hit esp adjust of %d\n",
4344 arg, pp->argc, ops[j].operand[1].val);
4345 }
4346 else if (ops[j].op == OP_POP && !(ops[j].flags & OPF_DONE))
4347 {
4348 if (pp->is_unresolved)
4349 break;
4350
4351 fnote(po, "(this call)\n");
4352 ferr(&ops[j], "arg collect %d/%d hit pop\n", arg, pp->argc);
4353 }
4354 else if (ops[j].flags & OPF_CJMP)
4355 {
4356 if (pp->is_unresolved)
4357 break;
4358
4359 may_reuse = 1;
4360 }
4361 else if (ops[j].op == OP_PUSH
4362 && !(ops[j].flags & (OPF_FARGNR|OPF_DONE)))
4363 {
4364 if (pp->is_unresolved && (ops[j].flags & OPF_RMD))
4365 break;
4366
4367 ops[j].p_argnext = -1;
4368 po_tmp = pp->arg[arg].datap;
4369 if (po_tmp != NULL)
4370 ops[j].p_argnext = po_tmp - ops;
4371 pp->arg[arg].datap = &ops[j];
4372
4373 need_to_save_current = 0;
4374 save_args = 0;
4375 reg = -1;
4376 if (ops[j].operand[0].type == OPT_REG)
4377 reg = ops[j].operand[0].reg;
4378
4379 if (!need_op_saving) {
4380 ret = scan_for_mod(&ops[j], j + 1, i, 1);
4381 need_to_save_current = (ret >= 0);
4382 }
4383 if (need_op_saving || need_to_save_current) {
4384 // mark this push as one that needs operand saving
4385 ops[j].flags &= ~OPF_RMD;
4386 if (ops[j].p_argnum == 0) {
4387 ops[j].p_argnum = arg + 1;
4388 save_args |= 1 << arg;
4389 }
4390 else if (ops[j].p_argnum < arg + 1) {
4391 // XXX: might kill valid var..
4392 //*save_arg_vars &= ~(1 << (ops[j].p_argnum - 1));
4393 ops[j].p_argnum = arg + 1;
4394 save_args |= 1 << arg;
4395 }
4396
4397 if (save_args_seen & (1 << (ops[j].p_argnum - 1))) {
4398 save_args_seen = 0;
4399 arg_grp_current++;
4400 if (arg_grp_current >= MAX_ARG_GRP)
4401 ferr(&ops[j], "out of arg groups (arg%d), f %s\n",
4402 ops[j].p_argnum, pp->name);
4403 }
4404 }
4405 else if (ops[j].p_argnum == 0)
4406 ops[j].flags |= OPF_RMD;
4407
4408 // some PUSHes are reused by different calls on other branches,
4409 // but that can't happen if we didn't branch, so they
4410 // can be removed from future searches (handles nested calls)
4411 if (!may_reuse)
4412 ops[j].flags |= OPF_FARGNR;
4413
4414 ops[j].flags |= OPF_FARG;
4415 ops[j].flags &= ~OPF_RSAVE;
4416
4417 // check for __VALIST
4418 if (!pp->is_unresolved && g_func_pp != NULL
4419 && pp->arg[arg].type.is_va_list)
4420 {
4421 k = -1;
4422 ret = resolve_origin(j, &ops[j].operand[0],
4423 magic + 1, &k, NULL);
4424 if (ret == 1 && k >= 0)
4425 {
4426 if (ops[k].op == OP_LEA) {
4427 if (!g_func_pp->is_vararg)
4428 ferr(&ops[k], "lea <arg> used, but %s is not vararg?\n",
4429 g_func_pp->name);
4430
4431 snprintf(buf, sizeof(buf), "arg_%X",
4432 g_func_pp->argc_stack * 4);
4433 if (strstr(ops[k].operand[1].name, buf)
4434 || strstr(ops[k].operand[1].name, "arglist"))
4435 {
4436 ops[k].flags |= OPF_RMD | OPF_NOREGS | OPF_DONE;
4437 ops[j].flags |= OPF_RMD | OPF_NOREGS | OPF_VAPUSH;
4438 save_args &= ~(1 << arg);
4439 reg = -1;
4440 }
4441 else
4442 ferr(&ops[k], "va_list arg detection failed\n");
4443 }
4444 // check for va_list from g_func_pp arg too
4445 else if (ops[k].op == OP_MOV
4446 && is_stack_access(&ops[k], &ops[k].operand[1]))
4447 {
4448 ret = stack_frame_access(&ops[k], &ops[k].operand[1],
4449 buf, sizeof(buf), ops[k].operand[1].name, "", 1, 0);
4450 if (ret >= 0) {
4451 ops[k].flags |= OPF_RMD | OPF_DONE;
4452 ops[j].flags |= OPF_RMD;
4453 ops[j].p_argpass = ret + 1;
4454 save_args &= ~(1 << arg);
4455 reg = -1;
4456 }
4457 }
4458 }
4459 }
4460
4461 *save_arg_vars |= save_args;
4462
4463 // tracking reg usage
4464 if (reg >= 0)
4465 *regmask |= 1 << reg;
4466
4467 arg++;
4468 if (!pp->is_unresolved) {
4469 // next arg
4470 for (; arg < pp->argc; arg++)
4471 if (pp->arg[arg].reg == NULL)
4472 break;
4473 }
4474 magic = (magic & 0xffffff) | (arg << 24);
4475 }
4476
4477 if (ops[j].p_arggrp > arg_grp_current) {
4478 save_args_seen = 0;
4479 arg_grp_current = ops[j].p_arggrp;
4480 }
4481 if (ops[j].p_argnum > 0)
4482 save_args_seen |= 1 << (ops[j].p_argnum - 1);
4483 }
4484
4485 if (arg < pp->argc) {
4486 ferr(po, "arg collect failed for '%s': %d/%d\n",
4487 pp->name, arg, pp->argc);
4488 return -1;
4489 }
4490
4491 if (arg_grp_current > *arg_grp)
4492 *arg_grp = arg_grp_current;
4493
4494 return arg;
4495}
4496
4497static int collect_call_args(struct parsed_op *po, int i,
4498 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
4499 int magic)
4500{
4501 // arg group is for cases when pushes for
4502 // multiple funcs are going on
4503 struct parsed_op *po_tmp;
4504 int save_arg_vars_current = 0;
4505 int arg_grp = 0;
4506 int ret;
4507 int a;
4508
4509 ret = collect_call_args_r(po, i, pp, regmask,
4510 &save_arg_vars_current, &arg_grp, 0, magic, 0, 0);
4511 if (ret < 0)
4512 return ret;
4513
4514 if (arg_grp != 0) {
4515 // propagate arg_grp
4516 for (a = 0; a < pp->argc; a++) {
4517 if (pp->arg[a].reg != NULL)
4518 continue;
4519
4520 po_tmp = pp->arg[a].datap;
4521 while (po_tmp != NULL) {
4522 po_tmp->p_arggrp = arg_grp;
4523 if (po_tmp->p_argnext > 0)
4524 po_tmp = &ops[po_tmp->p_argnext];
4525 else
4526 po_tmp = NULL;
4527 }
4528 }
4529 }
4530 save_arg_vars[arg_grp] |= save_arg_vars_current;
4531
4532 if (pp->is_unresolved) {
4533 pp->argc += ret;
4534 pp->argc_stack += ret;
4535 for (a = 0; a < pp->argc; a++)
4536 if (pp->arg[a].type.name == NULL)
4537 pp->arg[a].type.name = strdup("int");
4538 }
4539
4540 return ret;
4541}
4542
4543static void reg_use_pass(int i, int opcnt, unsigned char *cbits,
4544 int regmask_now, int *regmask,
4545 int regmask_save_now, int *regmask_save,
4546 int *regmask_init, int regmask_arg)
4547{
4548 struct parsed_op *po;
4549 unsigned int mask;
4550 int already_saved;
4551 int regmask_new;
4552 int regmask_op;
4553 int flags_set;
4554 int ret, reg;
4555 int j;
4556
4557 for (; i < opcnt; i++)
4558 {
4559 po = &ops[i];
4560 if (cbits[i >> 3] & (1 << (i & 7)))
4561 return;
4562 cbits[i >> 3] |= (1 << (i & 7));
4563
4564 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
4565 if (po->flags & (OPF_RMD|OPF_DONE))
4566 continue;
4567 if (po->btj != NULL) {
4568 for (j = 0; j < po->btj->count; j++) {
4569 check_i(po, po->btj->d[j].bt_i);
4570 reg_use_pass(po->btj->d[j].bt_i, opcnt, cbits,
4571 regmask_now, regmask, regmask_save_now, regmask_save,
4572 regmask_init, regmask_arg);
4573 }
4574 return;
4575 }
4576
4577 check_i(po, po->bt_i);
4578 if (po->flags & OPF_CJMP)
4579 reg_use_pass(po->bt_i, opcnt, cbits,
4580 regmask_now, regmask, regmask_save_now, regmask_save,
4581 regmask_init, regmask_arg);
4582 else
4583 i = po->bt_i - 1;
4584 continue;
4585 }
4586
4587 if (po->op == OP_PUSH && !(po->flags & (OPF_FARG|OPF_DONE))
4588 && !g_func_pp->is_userstack
4589 && po->operand[0].type == OPT_REG)
4590 {
4591 reg = po->operand[0].reg;
4592 ferr_assert(po, reg >= 0);
4593
4594 already_saved = 0;
4595 flags_set = OPF_RSAVE | OPF_RMD | OPF_DONE;
4596 if (regmask_now & (1 << reg)) {
4597 already_saved = regmask_save_now & (1 << reg);
4598 flags_set = OPF_RSAVE | OPF_DONE;
4599 }
4600
4601 ret = scan_for_pop(i + 1, opcnt, i + opcnt * 3, reg, 0, 0);
4602 if (ret == 1) {
4603 scan_for_pop(i + 1, opcnt, i + opcnt * 4, reg, 0, flags_set);
4604 }
4605 else {
4606 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].reg, 0);
4607 if (ret == 1) {
4608 scan_for_pop_ret(i + 1, opcnt, po->operand[0].reg,
4609 flags_set);
4610 }
4611 }
4612 if (ret == 1) {
4613 ferr_assert(po, !already_saved);
4614 po->flags |= flags_set;
4615
4616 if (regmask_now & (1 << reg)) {
4617 regmask_save_now |= (1 << reg);
4618 *regmask_save |= regmask_save_now;
4619 }
4620 continue;
4621 }
4622 }
4623 else if (po->op == OP_POP && (po->flags & OPF_RSAVE)) {
4624 reg = po->operand[0].reg;
4625 ferr_assert(po, reg >= 0);
4626
4627 if (regmask_save_now & (1 << reg))
4628 regmask_save_now &= ~(1 << reg);
4629 else
4630 regmask_now &= ~(1 << reg);
4631 continue;
4632 }
4633 else if (po->op == OP_CALL) {
4634 if ((po->regmask_dst & (1 << xAX))
4635 && !(po->regmask_dst & (1 << xDX)))
4636 {
4637 if (po->flags & OPF_TAIL)
4638 // don't need eax, will do "return f();" or "f(); return;"
4639 po->regmask_dst &= ~(1 << xAX);
4640 else {
4641 struct parsed_opr opr = OPR_INIT(OPT_REG, OPLM_DWORD, xAX);
4642 j = -1;
4643 find_next_read(i + 1, opcnt, &opr, i + opcnt * 17, &j);
4644 if (j == -1)
4645 // not used
4646 po->regmask_dst &= ~(1 << xAX);
4647 }
4648 }
4649 }
4650
4651 if (po->flags & OPF_NOREGS)
4652 continue;
4653
4654 if (po->flags & OPF_FPUSH) {
4655 if (regmask_now & mxST1)
4656 ferr(po, "TODO: FPUSH on active ST1\n");
4657 if (regmask_now & mxST0)
4658 po->flags |= OPF_FSHIFT;
4659 mask = mxST0 | mxST1;
4660 regmask_now = (regmask_now & ~mask) | ((regmask_now & mxST0) << 1);
4661 }
4662
4663 // if incomplete register is used, clear it on init to avoid
4664 // later use of uninitialized upper part in some situations
4665 if ((po->flags & OPF_DATA) && po->operand[0].type == OPT_REG
4666 && po->operand[0].lmod != OPLM_DWORD)
4667 {
4668 reg = po->operand[0].reg;
4669 ferr_assert(po, reg >= 0);
4670
4671 if (!(regmask_now & (1 << reg)))
4672 *regmask_init |= 1 << reg;
4673 }
4674
4675 regmask_op = po->regmask_src | po->regmask_dst;
4676
4677 regmask_new = po->regmask_src & ~regmask_now & ~regmask_arg;
4678 regmask_new &= ~(1 << xSP);
4679 if (g_bp_frame && !(po->flags & OPF_EBP_S))
4680 regmask_new &= ~(1 << xBP);
4681
4682 if (po->op == OP_CALL) {
4683 // allow fastcall calls from anywhere, calee may be also sitting
4684 // in some fastcall table even when it's not using reg args
4685 if (regmask_new & po->regmask_src & (1 << xCX)) {
4686 *regmask_init |= (1 << xCX);
4687 regmask_now |= (1 << xCX);
4688 regmask_new &= ~(1 << xCX);
4689 }
4690 if (regmask_new & po->regmask_src & (1 << xDX)) {
4691 *regmask_init |= (1 << xDX);
4692 regmask_now |= (1 << xDX);
4693 regmask_new &= ~(1 << xDX);
4694 }
4695 }
4696
4697 if (regmask_new != 0)
4698 fnote(po, "uninitialized reg mask: %x\n", regmask_new);
4699
4700 if (regmask_op & (1 << xBP)) {
4701 if (g_bp_frame && !(po->flags & OPF_EBP_S)) {
4702 if (po->regmask_dst & (1 << xBP))
4703 // compiler decided to drop bp frame and use ebp as scratch
4704 scan_fwd_set_flags(i + 1, opcnt, i + opcnt * 5, OPF_EBP_S);
4705 else
4706 regmask_op &= ~(1 << xBP);
4707 }
4708 }
4709
4710 regmask_now |= regmask_op;
4711 *regmask |= regmask_now;
4712
4713 // released regs
4714 if (po->flags & OPF_FPOP) {
4715 mask = mxST0 | mxST1;
4716 if (!(regmask_now & mask))
4717 ferr(po, "float pop on empty stack?\n");
4718 if (regmask_now & mxST1)
4719 po->flags |= OPF_FSHIFT;
4720 regmask_now = (regmask_now & ~mask) | ((regmask_now & mxST1) >> 1);
4721 }
4722
4723 if (po->flags & OPF_TAIL) {
4724 if (regmask_now & (mxST0 | mxST1))
4725 ferr(po, "float regs on tail: %x\n", regmask_now);
4726 return;
4727 }
4728 }
4729}
4730
4731static void pp_insert_reg_arg(struct parsed_proto *pp, const char *reg)
4732{
4733 int i;
4734
4735 for (i = 0; i < pp->argc; i++)
4736 if (pp->arg[i].reg == NULL)
4737 break;
4738
4739 if (pp->argc_stack)
4740 memmove(&pp->arg[i + 1], &pp->arg[i],
4741 sizeof(pp->arg[0]) * pp->argc_stack);
4742 memset(&pp->arg[i], 0, sizeof(pp->arg[i]));
4743 pp->arg[i].reg = strdup(reg);
4744 pp->arg[i].type.name = strdup("int");
4745 pp->argc++;
4746 pp->argc_reg++;
4747}
4748
4749static void output_std_flags(FILE *fout, struct parsed_op *po,
4750 int *pfomask, const char *dst_opr_text)
4751{
4752 if (*pfomask & (1 << PFO_Z)) {
4753 fprintf(fout, "\n cond_z = (%s%s == 0);",
4754 lmod_cast_u(po, po->operand[0].lmod), dst_opr_text);
4755 *pfomask &= ~(1 << PFO_Z);
4756 }
4757 if (*pfomask & (1 << PFO_S)) {
4758 fprintf(fout, "\n cond_s = (%s%s < 0);",
4759 lmod_cast_s(po, po->operand[0].lmod), dst_opr_text);
4760 *pfomask &= ~(1 << PFO_S);
4761 }
4762}
4763
4764enum {
4765 OPP_FORCE_NORETURN = (1 << 0),
4766 OPP_SIMPLE_ARGS = (1 << 1),
4767 OPP_ALIGN = (1 << 2),
4768};
4769
4770static void output_pp_attrs(FILE *fout, const struct parsed_proto *pp,
4771 int flags)
4772{
4773 const char *cconv = "";
4774
4775 if (pp->is_fastcall)
4776 cconv = "__fastcall ";
4777 else if (pp->is_stdcall && pp->argc_reg == 0)
4778 cconv = "__stdcall ";
4779
4780 fprintf(fout, (flags & OPP_ALIGN) ? "%-16s" : "%s", cconv);
4781
4782 if (pp->is_noreturn || (flags & OPP_FORCE_NORETURN))
4783 fprintf(fout, "noreturn ");
4784}
4785
4786static void output_pp(FILE *fout, const struct parsed_proto *pp,
4787 int flags)
4788{
4789 int i;
4790
4791 fprintf(fout, (flags & OPP_ALIGN) ? "%-5s" : "%s ",
4792 pp->ret_type.name);
4793 if (pp->is_fptr)
4794 fprintf(fout, "(");
4795 output_pp_attrs(fout, pp, flags);
4796 if (pp->is_fptr)
4797 fprintf(fout, "*");
4798 fprintf(fout, "%s", pp->name);
4799 if (pp->is_fptr)
4800 fprintf(fout, ")");
4801
4802 fprintf(fout, "(");
4803 for (i = 0; i < pp->argc; i++) {
4804 if (i > 0)
4805 fprintf(fout, ", ");
4806 if (pp->arg[i].fptr != NULL && !(flags & OPP_SIMPLE_ARGS)) {
4807 // func pointer
4808 output_pp(fout, pp->arg[i].fptr, 0);
4809 }
4810 else if (pp->arg[i].type.is_retreg) {
4811 fprintf(fout, "u32 *r_%s", pp->arg[i].reg);
4812 }
4813 else {
4814 fprintf(fout, "%s", pp->arg[i].type.name);
4815 if (!pp->is_fptr)
4816 fprintf(fout, " a%d", i + 1);
4817 }
4818 }
4819 if (pp->is_vararg) {
4820 if (i > 0)
4821 fprintf(fout, ", ");
4822 fprintf(fout, "...");
4823 }
4824 fprintf(fout, ")");
4825}
4826
4827static char *saved_arg_name(char *buf, size_t buf_size, int grp, int num)
4828{
4829 char buf1[16];
4830
4831 buf1[0] = 0;
4832 if (grp > 0)
4833 snprintf(buf1, sizeof(buf1), "%d", grp);
4834 snprintf(buf, buf_size, "s%s_a%d", buf1, num);
4835
4836 return buf;
4837}
4838
4839static void gen_x_cleanup(int opcnt);
4840
4841static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
4842{
4843 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
4844 struct parsed_opr *last_arith_dst = NULL;
4845 char buf1[256], buf2[256], buf3[256], cast[64];
4846 struct parsed_proto *pp, *pp_tmp;
4847 struct parsed_data *pd;
4848 unsigned int uval;
4849 int save_arg_vars[MAX_ARG_GRP] = { 0, };
4850 unsigned char cbits[MAX_OPS / 8];
4851 int cond_vars = 0;
4852 int need_tmp_var = 0;
4853 int need_tmp64 = 0;
4854 int had_decl = 0;
4855 int label_pending = 0;
4856 int regmask_save = 0; // regs saved/restored in this func
4857 int regmask_arg; // regs from this function args (fastcall, etc)
4858 int regmask_ret; // regs needed on ret
4859 int regmask_now; // temp
4860 int regmask_init = 0; // regs that need zero initialization
4861 int regmask_pp = 0; // regs used in complex push-pop graph
4862 int regmask = 0; // used regs
4863 int pfomask = 0;
4864 int found = 0;
4865 int no_output;
4866 int i, j, l;
4867 int arg;
4868 int reg;
4869 int ret;
4870
4871 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
4872 g_stack_frame_used = 0;
4873
4874 g_func_pp = proto_parse(fhdr, funcn, 0);
4875 if (g_func_pp == NULL)
4876 ferr(ops, "proto_parse failed for '%s'\n", funcn);
4877
4878 regmask_arg = get_pp_arg_regmask_src(g_func_pp);
4879 regmask_ret = get_pp_arg_regmask_dst(g_func_pp);
4880
4881 if (g_func_pp->has_retreg) {
4882 for (arg = 0; arg < g_func_pp->argc; arg++) {
4883 if (g_func_pp->arg[arg].type.is_retreg) {
4884 reg = char_array_i(regs_r32,
4885 ARRAY_SIZE(regs_r32), g_func_pp->arg[arg].reg);
4886 ferr_assert(ops, reg >= 0);
4887 regmask_ret |= 1 << reg;
4888 }
4889 }
4890 }
4891
4892 // pass1:
4893 // - resolve all branches
4894 // - parse calls with labels
4895 resolve_branches_parse_calls(opcnt);
4896
4897 // pass2:
4898 // - handle ebp/esp frame, remove ops related to it
4899 scan_prologue_epilogue(opcnt);
4900
4901 // pass3:
4902 // - remove dead labels
4903 // - set regs needed at ret
4904 for (i = 0; i < opcnt; i++)
4905 {
4906 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
4907 free(g_labels[i]);
4908 g_labels[i] = NULL;
4909 }
4910
4911 if (ops[i].op == OP_RET)
4912 ops[i].regmask_src |= regmask_ret;
4913 }
4914
4915 // pass4:
4916 // - process trivial calls
4917 for (i = 0; i < opcnt; i++)
4918 {
4919 po = &ops[i];
4920 if (po->flags & (OPF_RMD|OPF_DONE))
4921 continue;
4922
4923 if (po->op == OP_CALL)
4924 {
4925 pp = process_call_early(i, opcnt, &j);
4926 if (pp != NULL) {
4927 if (!(po->flags & OPF_ATAIL))
4928 // since we know the args, try to collect them
4929 if (collect_call_args_early(po, i, pp, &regmask) != 0)
4930 pp = NULL;
4931 }
4932
4933 if (pp != NULL) {
4934 if (j >= 0) {
4935 // commit esp adjust
4936 if (ops[j].op != OP_POP)
4937 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
4938 else {
4939 for (l = 0; l < pp->argc_stack; l++)
4940 ops[j + l].flags |= OPF_DONE | OPF_RMD | OPF_NOREGS;
4941 }
4942 }
4943
4944 if (strstr(pp->ret_type.name, "int64"))
4945 need_tmp64 = 1;
4946
4947 po->flags |= OPF_DONE;
4948 }
4949 }
4950 }
4951
4952 // pass5:
4953 // - process calls, stage 2
4954 // - handle some push/pop pairs
4955 // - scan for STD/CLD, propagate DF
4956 for (i = 0; i < opcnt; i++)
4957 {
4958 po = &ops[i];
4959 if (po->flags & OPF_RMD)
4960 continue;
4961
4962 if (po->op == OP_CALL)
4963 {
4964 if (!(po->flags & OPF_DONE)) {
4965 pp = process_call(i, opcnt);
4966
4967 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
4968 // since we know the args, collect them
4969 collect_call_args(po, i, pp, &regmask, save_arg_vars,
4970 i + opcnt * 2);
4971 }
4972 // for unresolved, collect after other passes
4973 }
4974
4975 pp = po->pp;
4976 ferr_assert(po, pp != NULL);
4977
4978 po->regmask_src |= get_pp_arg_regmask_src(pp);
4979 po->regmask_dst |= get_pp_arg_regmask_dst(pp);
4980
4981 if (po->regmask_dst & mxST0)
4982 po->flags |= OPF_FPUSH;
4983
4984 if (strstr(pp->ret_type.name, "int64"))
4985 need_tmp64 = 1;
4986
4987 continue;
4988 }
4989
4990 if (po->flags & OPF_DONE)
4991 continue;
4992
4993 if (po->op == OP_PUSH && !(po->flags & OPF_FARG)
4994 && !(po->flags & OPF_RSAVE) && po->operand[0].type == OPT_CONST)
4995 {
4996 scan_for_pop_const(i, opcnt, i + opcnt * 12);
4997 }
4998 else if (po->op == OP_POP)
4999 scan_pushes_for_pop(i, opcnt, &regmask_pp);
5000 else if (po->op == OP_STD) {
5001 po->flags |= OPF_DF | OPF_RMD | OPF_DONE;
5002 scan_propagate_df(i + 1, opcnt);
5003 }
5004 }
5005
5006 // pass6:
5007 // - find POPs for PUSHes, rm both
5008 // - scan for all used registers
5009 memset(cbits, 0, sizeof(cbits));
5010 reg_use_pass(0, opcnt, cbits, 0, &regmask,
5011 0, &regmask_save, &regmask_init, regmask_arg);
5012
5013 // pass7:
5014 // - find flag set ops for their users
5015 // - do unresolved calls
5016 // - declare indirect functions
5017 for (i = 0; i < opcnt; i++)
5018 {
5019 po = &ops[i];
5020 if (po->flags & (OPF_RMD|OPF_DONE))
5021 continue;
5022
5023 if (po->flags & OPF_CC)
5024 {
5025 int setters[16], cnt = 0, branched = 0;
5026
5027 ret = scan_for_flag_set(i, i + opcnt * 6,
5028 &branched, setters, &cnt);
5029 if (ret < 0 || cnt <= 0)
5030 ferr(po, "unable to trace flag setter(s)\n");
5031 if (cnt > ARRAY_SIZE(setters))
5032 ferr(po, "too many flag setters\n");
5033
5034 for (j = 0; j < cnt; j++)
5035 {
5036 tmp_op = &ops[setters[j]]; // flag setter
5037 pfomask = 0;
5038
5039 // to get nicer code, we try to delay test and cmp;
5040 // if we can't because of operand modification, or if we
5041 // have arith op, or branch, make it calculate flags explicitly
5042 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP)
5043 {
5044 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i, 0) >= 0)
5045 pfomask = 1 << po->pfo;
5046 }
5047 else if (tmp_op->op == OP_CMPS || tmp_op->op == OP_SCAS) {
5048 pfomask = 1 << po->pfo;
5049 }
5050 else {
5051 // see if we'll be able to handle based on op result
5052 if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR
5053 && po->pfo != PFO_Z && po->pfo != PFO_S
5054 && po->pfo != PFO_P)
5055 || branched
5056 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
5057 {
5058 pfomask = 1 << po->pfo;
5059 }
5060
5061 if (tmp_op->op == OP_ADD && po->pfo == PFO_C) {
5062 propagate_lmod(tmp_op, &tmp_op->operand[0],
5063 &tmp_op->operand[1]);
5064 if (tmp_op->operand[0].lmod == OPLM_DWORD)
5065 need_tmp64 = 1;
5066 }
5067 }
5068 if (pfomask) {
5069 tmp_op->pfomask |= pfomask;
5070 cond_vars |= pfomask;
5071 }
5072 // note: may overwrite, currently not a problem
5073 po->datap = tmp_op;
5074 }
5075
5076 if (po->op == OP_RCL || po->op == OP_RCR
5077 || po->op == OP_ADC || po->op == OP_SBB)
5078 cond_vars |= 1 << PFO_C;
5079 }
5080
5081 if (po->op == OP_CMPS || po->op == OP_SCAS) {
5082 cond_vars |= 1 << PFO_Z;
5083 }
5084 else if (po->op == OP_MUL
5085 || (po->op == OP_IMUL && po->operand_cnt == 1))
5086 {
5087 if (po->operand[0].lmod == OPLM_DWORD)
5088 need_tmp64 = 1;
5089 }
5090 else if (po->op == OP_CALL) {
5091 // note: resolved non-reg calls are OPF_DONE already
5092 pp = po->pp;
5093 ferr_assert(po, pp != NULL);
5094
5095 if (pp->is_unresolved) {
5096 int regmask_stack = 0;
5097 collect_call_args(po, i, pp, &regmask, save_arg_vars,
5098 i + opcnt * 2);
5099
5100 // this is pretty rough guess:
5101 // see ecx and edx were pushed (and not their saved versions)
5102 for (arg = 0; arg < pp->argc; arg++) {
5103 if (pp->arg[arg].reg != NULL)
5104 continue;
5105
5106 tmp_op = pp->arg[arg].datap;
5107 if (tmp_op == NULL)
5108 ferr(po, "parsed_op missing for arg%d\n", arg);
5109 if (tmp_op->p_argnum == 0 && tmp_op->operand[0].type == OPT_REG)
5110 regmask_stack |= 1 << tmp_op->operand[0].reg;
5111 }
5112
5113 if (!((regmask_stack & (1 << xCX))
5114 && (regmask_stack & (1 << xDX))))
5115 {
5116 if (pp->argc_stack != 0
5117 || ((regmask | regmask_arg) & ((1 << xCX)|(1 << xDX))))
5118 {
5119 pp_insert_reg_arg(pp, "ecx");
5120 pp->is_fastcall = 1;
5121 regmask_init |= 1 << xCX;
5122 regmask |= 1 << xCX;
5123 }
5124 if (pp->argc_stack != 0
5125 || ((regmask | regmask_arg) & (1 << xDX)))
5126 {
5127 pp_insert_reg_arg(pp, "edx");
5128 regmask_init |= 1 << xDX;
5129 regmask |= 1 << xDX;
5130 }
5131 }
5132
5133 // note: __cdecl doesn't fall into is_unresolved category
5134 if (pp->argc_stack > 0)
5135 pp->is_stdcall = 1;
5136 }
5137 }
5138 else if (po->op == OP_MOV && po->operand[0].pp != NULL
5139 && po->operand[1].pp != NULL)
5140 {
5141 // <var> = offset <something>
5142 if ((po->operand[1].pp->is_func || po->operand[1].pp->is_fptr)
5143 && !IS_START(po->operand[1].name, "off_"))
5144 {
5145 if (!po->operand[0].pp->is_fptr)
5146 ferr(po, "%s not declared as fptr when it should be\n",
5147 po->operand[0].name);
5148 if (pp_cmp_func(po->operand[0].pp, po->operand[1].pp)) {
5149 pp_print(buf1, sizeof(buf1), po->operand[0].pp);
5150 pp_print(buf2, sizeof(buf2), po->operand[1].pp);
5151 fnote(po, "var: %s\n", buf1);
5152 fnote(po, "func: %s\n", buf2);
5153 ferr(po, "^ mismatch\n");
5154 }
5155 }
5156 }
5157 else if (po->op == OP_DIV || po->op == OP_IDIV) {
5158 if (po->operand[0].lmod == OPLM_DWORD) {
5159 // 32bit division is common, look for it
5160 if (po->op == OP_DIV)
5161 ret = scan_for_reg_clear(i, xDX);
5162 else
5163 ret = scan_for_cdq_edx(i);
5164 if (ret >= 0)
5165 po->flags |= OPF_32BIT;
5166 else
5167 need_tmp64 = 1;
5168 }
5169 else
5170 need_tmp_var = 1;
5171 }
5172 else if (po->op == OP_CLD)
5173 po->flags |= OPF_RMD | OPF_DONE;
5174 else if (po->op == OPP_FTOL) {
5175 struct parsed_opr opr = OPR_INIT(OPT_REG, OPLM_DWORD, xDX);
5176 j = -1;
5177 find_next_read(i + 1, opcnt, &opr, i + opcnt * 18, &j);
5178 if (j == -1)
5179 po->flags |= OPF_32BIT;
5180 }
5181
5182 if (po->op == OP_RCL || po->op == OP_RCR || po->op == OP_XCHG)
5183 need_tmp_var = 1;
5184 }
5185
5186 // output starts here
5187
5188 // define userstack size
5189 if (g_func_pp->is_userstack) {
5190 fprintf(fout, "#ifndef US_SZ_%s\n", g_func_pp->name);
5191 fprintf(fout, "#define US_SZ_%s USERSTACK_SIZE\n", g_func_pp->name);
5192 fprintf(fout, "#endif\n");
5193 }
5194
5195 // the function itself
5196 ferr_assert(ops, !g_func_pp->is_fptr);
5197 output_pp(fout, g_func_pp,
5198 (g_ida_func_attr & IDAFA_NORETURN) ? OPP_FORCE_NORETURN : 0);
5199 fprintf(fout, "\n{\n");
5200
5201 // declare indirect functions
5202 for (i = 0; i < opcnt; i++) {
5203 po = &ops[i];
5204 if (po->flags & OPF_RMD)
5205 continue;
5206
5207 if (po->op == OP_CALL) {
5208 pp = po->pp;
5209 if (pp == NULL)
5210 ferr(po, "NULL pp\n");
5211
5212 if (pp->is_fptr && !(pp->name[0] != 0 && pp->is_arg)) {
5213 if (pp->name[0] != 0) {
5214 memmove(pp->name + 2, pp->name, strlen(pp->name) + 1);
5215 memcpy(pp->name, "i_", 2);
5216
5217 // might be declared already
5218 found = 0;
5219 for (j = 0; j < i; j++) {
5220 if (ops[j].op == OP_CALL && (pp_tmp = ops[j].pp)) {
5221 if (pp_tmp->is_fptr && IS(pp->name, pp_tmp->name)) {
5222 found = 1;
5223 break;
5224 }
5225 }
5226 }
5227 if (found)
5228 continue;
5229 }
5230 else
5231 snprintf(pp->name, sizeof(pp->name), "icall%d", i);
5232
5233 fprintf(fout, " ");
5234 output_pp(fout, pp, OPP_SIMPLE_ARGS);
5235 fprintf(fout, ";\n");
5236 }
5237 }
5238 }
5239
5240 // output LUTs/jumptables
5241 for (i = 0; i < g_func_pd_cnt; i++) {
5242 pd = &g_func_pd[i];
5243 fprintf(fout, " static const ");
5244 if (pd->type == OPT_OFFSET) {
5245 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
5246
5247 for (j = 0; j < pd->count; j++) {
5248 if (j > 0)
5249 fprintf(fout, ", ");
5250 fprintf(fout, "&&%s", pd->d[j].u.label);
5251 }
5252 }
5253 else {
5254 fprintf(fout, "%s %s[] =\n { ",
5255 lmod_type_u(ops, pd->lmod), pd->label);
5256
5257 for (j = 0; j < pd->count; j++) {
5258 if (j > 0)
5259 fprintf(fout, ", ");
5260 fprintf(fout, "%u", pd->d[j].u.val);
5261 }
5262 }
5263 fprintf(fout, " };\n");
5264 had_decl = 1;
5265 }
5266
5267 // declare stack frame, va_arg
5268 if (g_stack_fsz) {
5269 fprintf(fout, " union { u32 d[%d];", (g_stack_fsz + 3) / 4);
5270 if (g_func_lmods & (1 << OPLM_WORD))
5271 fprintf(fout, " u16 w[%d];", (g_stack_fsz + 1) / 2);
5272 if (g_func_lmods & (1 << OPLM_BYTE))
5273 fprintf(fout, " u8 b[%d];", g_stack_fsz);
5274 if (g_func_lmods & (1 << OPLM_QWORD))
5275 fprintf(fout, " double q[%d];", (g_stack_fsz + 7) / 8);
5276 fprintf(fout, " } sf;\n");
5277 had_decl = 1;
5278 }
5279
5280 if (g_func_pp->is_userstack) {
5281 fprintf(fout, " u32 fake_sf[US_SZ_%s / 4];\n", g_func_pp->name);
5282 fprintf(fout, " u32 *esp = &fake_sf[sizeof(fake_sf) / 4];\n");
5283 had_decl = 1;
5284 }
5285
5286 if (g_func_pp->is_vararg) {
5287 fprintf(fout, " va_list ap;\n");
5288 had_decl = 1;
5289 }
5290
5291 // declare arg-registers
5292 for (i = 0; i < g_func_pp->argc; i++) {
5293 if (g_func_pp->arg[i].reg != NULL) {
5294 reg = char_array_i(regs_r32,
5295 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
5296 if (regmask & (1 << reg)) {
5297 if (g_func_pp->arg[i].type.is_retreg)
5298 fprintf(fout, " u32 %s = *r_%s;\n",
5299 g_func_pp->arg[i].reg, g_func_pp->arg[i].reg);
5300 else
5301 fprintf(fout, " u32 %s = (u32)a%d;\n",
5302 g_func_pp->arg[i].reg, i + 1);
5303 }
5304 else {
5305 if (g_func_pp->arg[i].type.is_retreg)
5306 ferr(ops, "retreg '%s' is unused?\n",
5307 g_func_pp->arg[i].reg);
5308 fprintf(fout, " // %s = a%d; // unused\n",
5309 g_func_pp->arg[i].reg, i + 1);
5310 }
5311 had_decl = 1;
5312 }
5313 }
5314
5315 // declare normal registers
5316 regmask_now = regmask & ~regmask_arg;
5317 regmask_now &= ~(1 << xSP);
5318 if (regmask_now & 0x00ff) {
5319 for (reg = 0; reg < 8; reg++) {
5320 if (regmask_now & (1 << reg)) {
5321 fprintf(fout, " u32 %s", regs_r32[reg]);
5322 if (regmask_init & (1 << reg))
5323 fprintf(fout, " = 0");
5324 fprintf(fout, ";\n");
5325 had_decl = 1;
5326 }
5327 }
5328 }
5329 // ... mmx
5330 if (regmask_now & 0xff00) {
5331 for (reg = 8; reg < 16; reg++) {
5332 if (regmask_now & (1 << reg)) {
5333 fprintf(fout, " mmxr %s", regs_r32[reg]);
5334 if (regmask_init & (1 << reg))
5335 fprintf(fout, " = { 0, }");
5336 fprintf(fout, ";\n");
5337 had_decl = 1;
5338 }
5339 }
5340 }
5341 // ... x87
5342 if (regmask_now & 0xff0000) {
5343 for (reg = 16; reg < 24; reg++) {
5344 if (regmask_now & (1 << reg)) {
5345 fprintf(fout, " double f_st%d", reg - 16);
5346 if (regmask_init & (1 << reg))
5347 fprintf(fout, " = 0");
5348 fprintf(fout, ";\n");
5349 had_decl = 1;
5350 }
5351 }
5352 }
5353
5354 if (regmask_save) {
5355 for (reg = 0; reg < 8; reg++) {
5356 if (regmask_save & (1 << reg)) {
5357 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
5358 had_decl = 1;
5359 }
5360 }
5361 }
5362
5363 for (i = 0; i < ARRAY_SIZE(save_arg_vars); i++) {
5364 if (save_arg_vars[i] == 0)
5365 continue;
5366 for (reg = 0; reg < 32; reg++) {
5367 if (save_arg_vars[i] & (1 << reg)) {
5368 fprintf(fout, " u32 %s;\n",
5369 saved_arg_name(buf1, sizeof(buf1), i, reg + 1));
5370 had_decl = 1;
5371 }
5372 }
5373 }
5374
5375 // declare push-pop temporaries
5376 if (regmask_pp) {
5377 for (reg = 0; reg < 8; reg++) {
5378 if (regmask_pp & (1 << reg)) {
5379 fprintf(fout, " u32 pp_%s;\n", regs_r32[reg]);
5380 had_decl = 1;
5381 }
5382 }
5383 }
5384
5385 if (cond_vars) {
5386 for (i = 0; i < 8; i++) {
5387 if (cond_vars & (1 << i)) {
5388 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
5389 had_decl = 1;
5390 }
5391 }
5392 }
5393
5394 if (need_tmp_var) {
5395 fprintf(fout, " u32 tmp;\n");
5396 had_decl = 1;
5397 }
5398
5399 if (need_tmp64) {
5400 fprintf(fout, " u64 tmp64;\n");
5401 had_decl = 1;
5402 }
5403
5404 if (had_decl)
5405 fprintf(fout, "\n");
5406
5407 if (g_func_pp->is_vararg) {
5408 if (g_func_pp->argc_stack == 0)
5409 ferr(ops, "vararg func without stack args?\n");
5410 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
5411 }
5412
5413 // output ops
5414 for (i = 0; i < opcnt; i++)
5415 {
5416 if (g_labels[i] != NULL) {
5417 fprintf(fout, "\n%s:\n", g_labels[i]);
5418 label_pending = 1;
5419
5420 delayed_flag_op = NULL;
5421 last_arith_dst = NULL;
5422 }
5423
5424 po = &ops[i];
5425 if (po->flags & OPF_RMD)
5426 continue;
5427
5428 no_output = 0;
5429
5430 #define assert_operand_cnt(n_) \
5431 if (po->operand_cnt != n_) \
5432 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
5433
5434 // conditional/flag using op?
5435 if (po->flags & OPF_CC)
5436 {
5437 int is_delayed = 0;
5438
5439 tmp_op = po->datap;
5440
5441 // we go through all this trouble to avoid using parsed_flag_op,
5442 // which makes generated code much nicer
5443 if (delayed_flag_op != NULL)
5444 {
5445 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op,
5446 po->pfo, po->pfo_inv);
5447 is_delayed = 1;
5448 }
5449 else if (last_arith_dst != NULL
5450 && (po->pfo == PFO_Z || po->pfo == PFO_S || po->pfo == PFO_P
5451 || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR))
5452 ))
5453 {
5454 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
5455 out_test_for_cc(buf1, sizeof(buf1), po, po->pfo, po->pfo_inv,
5456 last_arith_dst->lmod, buf3);
5457 is_delayed = 1;
5458 }
5459 else if (tmp_op != NULL) {
5460 // use preprocessed flag calc results
5461 if (!(tmp_op->pfomask & (1 << po->pfo)))
5462 ferr(po, "not prepared for pfo %d\n", po->pfo);
5463
5464 // note: pfo_inv was not yet applied
5465 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
5466 po->pfo_inv ? "!" : "", parsed_flag_op_names[po->pfo]);
5467 }
5468 else {
5469 ferr(po, "all methods of finding comparison failed\n");
5470 }
5471
5472 if (po->flags & OPF_JMP) {
5473 fprintf(fout, " if %s", buf1);
5474 }
5475 else if (po->op == OP_RCL || po->op == OP_RCR
5476 || po->op == OP_ADC || po->op == OP_SBB)
5477 {
5478 if (is_delayed)
5479 fprintf(fout, " cond_%s = %s;\n",
5480 parsed_flag_op_names[po->pfo], buf1);
5481 }
5482 else if (po->flags & OPF_DATA) { // SETcc
5483 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
5484 fprintf(fout, " %s = %s;", buf2, buf1);
5485 }
5486 else {
5487 ferr(po, "unhandled conditional op\n");
5488 }
5489 }
5490
5491 pfomask = po->pfomask;
5492
5493 if (po->flags & (OPF_REPZ|OPF_REPNZ)) {
5494 struct parsed_opr opr = OPR_INIT(OPT_REG, OPLM_DWORD, xCX);
5495 ret = try_resolve_const(i, &opr, opcnt * 7 + i, &uval);
5496
5497 if (ret != 1 || uval == 0) {
5498 // we need initial flags for ecx=0 case..
5499 if (i > 0 && ops[i - 1].op == OP_XOR
5500 && IS(ops[i - 1].operand[0].name,
5501 ops[i - 1].operand[1].name))
5502 {
5503 fprintf(fout, " cond_z = ");
5504 if (pfomask & (1 << PFO_C))
5505 fprintf(fout, "cond_c = ");
5506 fprintf(fout, "0;\n");
5507 }
5508 else if (last_arith_dst != NULL) {
5509 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
5510 out_test_for_cc(buf1, sizeof(buf1), po, PFO_Z, 0,
5511 last_arith_dst->lmod, buf3);
5512 fprintf(fout, " cond_z = %s;\n", buf1);
5513 }
5514 else
5515 ferr(po, "missing initial ZF\n");
5516 }
5517 }
5518
5519 switch (po->op)
5520 {
5521 case OP_MOV:
5522 assert_operand_cnt(2);
5523 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5524 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5525 default_cast_to(buf3, sizeof(buf3), &po->operand[0]);
5526 fprintf(fout, " %s = %s;", buf1,
5527 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
5528 buf3, 0));
5529 break;
5530
5531 case OP_LEA:
5532 assert_operand_cnt(2);
5533 po->operand[1].lmod = OPLM_DWORD; // always
5534 fprintf(fout, " %s = %s;",
5535 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5536 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
5537 NULL, 1));
5538 break;
5539
5540 case OP_MOVZX:
5541 assert_operand_cnt(2);
5542 fprintf(fout, " %s = %s;",
5543 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5544 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5545 break;
5546
5547 case OP_MOVSX:
5548 assert_operand_cnt(2);
5549 switch (po->operand[1].lmod) {
5550 case OPLM_BYTE:
5551 strcpy(buf3, "(s8)");
5552 break;
5553 case OPLM_WORD:
5554 strcpy(buf3, "(s16)");
5555 break;
5556 default:
5557 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
5558 }
5559 fprintf(fout, " %s = %s;",
5560 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5561 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
5562 buf3, 0));
5563 break;
5564
5565 case OP_XCHG:
5566 assert_operand_cnt(2);
5567 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5568 fprintf(fout, " tmp = %s;",
5569 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], "", 0));
5570 fprintf(fout, " %s = %s;",
5571 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5572 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
5573 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
5574 fprintf(fout, " %s = %stmp;",
5575 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]),
5576 default_cast_to(buf3, sizeof(buf3), &po->operand[1]));
5577 snprintf(g_comment, sizeof(g_comment), "xchg");
5578 break;
5579
5580 case OP_NOT:
5581 assert_operand_cnt(1);
5582 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5583 fprintf(fout, " %s = ~%s;", buf1, buf1);
5584 break;
5585
5586 case OP_XLAT:
5587 assert_operand_cnt(2);
5588 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5589 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
5590 fprintf(fout, " %s = *(u8 *)(%s + %s);", buf1, buf2, buf1);
5591 strcpy(g_comment, "xlat");
5592 break;
5593
5594 case OP_CDQ:
5595 assert_operand_cnt(2);
5596 fprintf(fout, " %s = (s32)%s >> 31;",
5597 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5598 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5599 strcpy(g_comment, "cdq");
5600 break;
5601
5602 case OP_LODS:
5603 if (po->flags & OPF_REP) {
5604 assert_operand_cnt(3);
5605 // hmh..
5606 ferr(po, "TODO\n");
5607 }
5608 else {
5609 assert_operand_cnt(2);
5610 fprintf(fout, " %s = %sesi; esi %c= %d;",
5611 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]),
5612 lmod_cast_u_ptr(po, po->operand[1].lmod),
5613 (po->flags & OPF_DF) ? '-' : '+',
5614 lmod_bytes(po, po->operand[1].lmod));
5615 strcpy(g_comment, "lods");
5616 }
5617 break;
5618
5619 case OP_STOS:
5620 if (po->flags & OPF_REP) {
5621 assert_operand_cnt(3);
5622 fprintf(fout, " for (; ecx != 0; ecx--, edi %c= %d)\n",
5623 (po->flags & OPF_DF) ? '-' : '+',
5624 lmod_bytes(po, po->operand[1].lmod));
5625 fprintf(fout, " %sedi = eax;",
5626 lmod_cast_u_ptr(po, po->operand[1].lmod));
5627 strcpy(g_comment, "rep stos");
5628 }
5629 else {
5630 assert_operand_cnt(2);
5631 fprintf(fout, " %sedi = eax; edi %c= %d;",
5632 lmod_cast_u_ptr(po, po->operand[1].lmod),
5633 (po->flags & OPF_DF) ? '-' : '+',
5634 lmod_bytes(po, po->operand[1].lmod));
5635 strcpy(g_comment, "stos");
5636 }
5637 break;
5638
5639 case OP_MOVS:
5640 j = lmod_bytes(po, po->operand[0].lmod);
5641 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
5642 l = (po->flags & OPF_DF) ? '-' : '+';
5643 if (po->flags & OPF_REP) {
5644 assert_operand_cnt(3);
5645 fprintf(fout,
5646 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n",
5647 l, j, l, j);
5648 fprintf(fout,
5649 " %sedi = %sesi;", buf1, buf1);
5650 strcpy(g_comment, "rep movs");
5651 }
5652 else {
5653 assert_operand_cnt(2);
5654 fprintf(fout, " %sedi = %sesi; edi %c= %d; esi %c= %d;",
5655 buf1, buf1, l, j, l, j);
5656 strcpy(g_comment, "movs");
5657 }
5658 break;
5659
5660 case OP_CMPS:
5661 // repe ~ repeat while ZF=1
5662 j = lmod_bytes(po, po->operand[0].lmod);
5663 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
5664 l = (po->flags & OPF_DF) ? '-' : '+';
5665 if (po->flags & OPF_REP) {
5666 assert_operand_cnt(3);
5667 fprintf(fout,
5668 " for (; ecx != 0; ecx--) {\n");
5669 if (pfomask & (1 << PFO_C)) {
5670 // ugh..
5671 fprintf(fout,
5672 " cond_c = %sesi < %sedi;\n", buf1, buf1);
5673 pfomask &= ~(1 << PFO_C);
5674 }
5675 fprintf(fout,
5676 " cond_z = (%sesi == %sedi); esi %c= %d, edi %c= %d;\n",
5677 buf1, buf1, l, j, l, j);
5678 fprintf(fout,
5679 " if (cond_z %s 0) break;\n",
5680 (po->flags & OPF_REPZ) ? "==" : "!=");
5681 fprintf(fout,
5682 " }");
5683 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
5684 (po->flags & OPF_REPZ) ? "e" : "ne");
5685 }
5686 else {
5687 assert_operand_cnt(2);
5688 fprintf(fout,
5689 " cond_z = (%sesi == %sedi); esi %c= %d; edi %c= %d;",
5690 buf1, buf1, l, j, l, j);
5691 strcpy(g_comment, "cmps");
5692 }
5693 pfomask &= ~(1 << PFO_Z);
5694 last_arith_dst = NULL;
5695 delayed_flag_op = NULL;
5696 break;
5697
5698 case OP_SCAS:
5699 // only does ZF (for now)
5700 // repe ~ repeat while ZF=1
5701 j = lmod_bytes(po, po->operand[1].lmod);
5702 l = (po->flags & OPF_DF) ? '-' : '+';
5703 if (po->flags & OPF_REP) {
5704 assert_operand_cnt(3);
5705 fprintf(fout,
5706 " for (; ecx != 0; ecx--) {\n");
5707 fprintf(fout,
5708 " cond_z = (%seax == %sedi); edi %c= %d;\n",
5709 lmod_cast_u(po, po->operand[1].lmod),
5710 lmod_cast_u_ptr(po, po->operand[1].lmod), l, j);
5711 fprintf(fout,
5712 " if (cond_z %s 0) break;\n",
5713 (po->flags & OPF_REPZ) ? "==" : "!=");
5714 fprintf(fout,
5715 " }");
5716 snprintf(g_comment, sizeof(g_comment), "rep%s scas",
5717 (po->flags & OPF_REPZ) ? "e" : "ne");
5718 }
5719 else {
5720 assert_operand_cnt(2);
5721 fprintf(fout, " cond_z = (%seax == %sedi); edi %c= %d;",
5722 lmod_cast_u(po, po->operand[1].lmod),
5723 lmod_cast_u_ptr(po, po->operand[1].lmod), l, j);
5724 strcpy(g_comment, "scas");
5725 }
5726 pfomask &= ~(1 << PFO_Z);
5727 last_arith_dst = NULL;
5728 delayed_flag_op = NULL;
5729 break;
5730
5731 // arithmetic w/flags
5732 case OP_AND:
5733 if (po->operand[1].type == OPT_CONST && !po->operand[1].val)
5734 goto dualop_arith_const;
5735 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5736 goto dualop_arith;
5737
5738 case OP_OR:
5739 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5740 if (po->operand[1].type == OPT_CONST) {
5741 j = lmod_bytes(po, po->operand[0].lmod);
5742 if (((1ull << j * 8) - 1) == po->operand[1].val)
5743 goto dualop_arith_const;
5744 }
5745 goto dualop_arith;
5746
5747 dualop_arith:
5748 assert_operand_cnt(2);
5749 fprintf(fout, " %s %s= %s;",
5750 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5751 op_to_c(po),
5752 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5753 output_std_flags(fout, po, &pfomask, buf1);
5754 last_arith_dst = &po->operand[0];
5755 delayed_flag_op = NULL;
5756 break;
5757
5758 dualop_arith_const:
5759 // and 0, or ~0 used instead mov
5760 assert_operand_cnt(2);
5761 fprintf(fout, " %s = %s;",
5762 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5763 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
5764 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
5765 output_std_flags(fout, po, &pfomask, buf1);
5766 last_arith_dst = &po->operand[0];
5767 delayed_flag_op = NULL;
5768 break;
5769
5770 case OP_SHL:
5771 case OP_SHR:
5772 assert_operand_cnt(2);
5773 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5774 if (pfomask & (1 << PFO_C)) {
5775 if (po->operand[1].type == OPT_CONST) {
5776 l = lmod_bytes(po, po->operand[0].lmod) * 8;
5777 j = po->operand[1].val;
5778 j %= l;
5779 if (j != 0) {
5780 if (po->op == OP_SHL)
5781 j = l - j;
5782 else
5783 j -= 1;
5784 fprintf(fout, " cond_c = (%s >> %d) & 1;\n",
5785 buf1, j);
5786 }
5787 else
5788 ferr(po, "zero shift?\n");
5789 }
5790 else
5791 ferr(po, "TODO\n");
5792 pfomask &= ~(1 << PFO_C);
5793 }
5794 fprintf(fout, " %s %s= %s", buf1, op_to_c(po),
5795 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5796 if (po->operand[1].type != OPT_CONST)
5797 fprintf(fout, " & 0x1f");
5798 fprintf(fout, ";");
5799 output_std_flags(fout, po, &pfomask, buf1);
5800 last_arith_dst = &po->operand[0];
5801 delayed_flag_op = NULL;
5802 break;
5803
5804 case OP_SAR:
5805 assert_operand_cnt(2);
5806 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5807 fprintf(fout, " %s = %s%s >> %s;", buf1,
5808 lmod_cast_s(po, po->operand[0].lmod), buf1,
5809 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5810 output_std_flags(fout, po, &pfomask, buf1);
5811 last_arith_dst = &po->operand[0];
5812 delayed_flag_op = NULL;
5813 break;
5814
5815 case OP_SHLD:
5816 case OP_SHRD:
5817 assert_operand_cnt(3);
5818 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5819 l = lmod_bytes(po, po->operand[0].lmod) * 8;
5820 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[2]);
5821 if (po->operand[2].type != OPT_CONST) {
5822 // no handling for "undefined" case, hopefully not needed
5823 snprintf(buf2, sizeof(buf2), "(%s & 0x1f)", buf3);
5824 strcpy(buf3, buf2);
5825 }
5826 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
5827 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5828 if (po->op == OP_SHLD) {
5829 fprintf(fout, " %s <<= %s; %s |= %s >> (%d - %s);",
5830 buf1, buf3, buf1, buf2, l, buf3);
5831 strcpy(g_comment, "shld");
5832 }
5833 else {
5834 fprintf(fout, " %s >>= %s; %s |= %s << (%d - %s);",
5835 buf1, buf3, buf1, buf2, l, buf3);
5836 strcpy(g_comment, "shrd");
5837 }
5838 output_std_flags(fout, po, &pfomask, buf1);
5839 last_arith_dst = &po->operand[0];
5840 delayed_flag_op = NULL;
5841 break;
5842
5843 case OP_ROL:
5844 case OP_ROR:
5845 assert_operand_cnt(2);
5846 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5847 if (po->operand[1].type == OPT_CONST) {
5848 j = po->operand[1].val;
5849 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
5850 fprintf(fout, po->op == OP_ROL ?
5851 " %s = (%s << %d) | (%s >> %d);" :
5852 " %s = (%s >> %d) | (%s << %d);",
5853 buf1, buf1, j, buf1,
5854 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
5855 }
5856 else
5857 ferr(po, "TODO\n");
5858 output_std_flags(fout, po, &pfomask, buf1);
5859 last_arith_dst = &po->operand[0];
5860 delayed_flag_op = NULL;
5861 break;
5862
5863 case OP_RCL:
5864 case OP_RCR:
5865 assert_operand_cnt(2);
5866 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5867 l = lmod_bytes(po, po->operand[0].lmod) * 8;
5868 if (po->operand[1].type == OPT_CONST) {
5869 j = po->operand[1].val % l;
5870 if (j == 0)
5871 ferr(po, "zero rotate\n");
5872 fprintf(fout, " tmp = (%s >> %d) & 1;\n",
5873 buf1, (po->op == OP_RCL) ? (l - j) : (j - 1));
5874 if (po->op == OP_RCL) {
5875 fprintf(fout,
5876 " %s = (%s << %d) | (cond_c << %d)",
5877 buf1, buf1, j, j - 1);
5878 if (j != 1)
5879 fprintf(fout, " | (%s >> %d)", buf1, l + 1 - j);
5880 }
5881 else {
5882 fprintf(fout,
5883 " %s = (%s >> %d) | (cond_c << %d)",
5884 buf1, buf1, j, l - j);
5885 if (j != 1)
5886 fprintf(fout, " | (%s << %d)", buf1, l + 1 - j);
5887 }
5888 fprintf(fout, ";\n");
5889 fprintf(fout, " cond_c = tmp;");
5890 }
5891 else
5892 ferr(po, "TODO\n");
5893 strcpy(g_comment, (po->op == OP_RCL) ? "rcl" : "rcr");
5894 output_std_flags(fout, po, &pfomask, buf1);
5895 last_arith_dst = &po->operand[0];
5896 delayed_flag_op = NULL;
5897 break;
5898
5899 case OP_XOR:
5900 assert_operand_cnt(2);
5901 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5902 if (IS(opr_name(po, 0), opr_name(po, 1))) {
5903 // special case for XOR
5904 if (pfomask & (1 << PFO_BE)) { // weird, but it happens..
5905 fprintf(fout, " cond_be = 1;\n");
5906 pfomask &= ~(1 << PFO_BE);
5907 }
5908 fprintf(fout, " %s = 0;",
5909 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
5910 last_arith_dst = &po->operand[0];
5911 delayed_flag_op = NULL;
5912 break;
5913 }
5914 goto dualop_arith;
5915
5916 case OP_ADD:
5917 assert_operand_cnt(2);
5918 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5919 if (pfomask & (1 << PFO_C)) {
5920 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
5921 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
5922 if (po->operand[0].lmod == OPLM_DWORD) {
5923 fprintf(fout, " tmp64 = (u64)%s + %s;\n", buf1, buf2);
5924 fprintf(fout, " cond_c = tmp64 >> 32;\n");
5925 fprintf(fout, " %s = (u32)tmp64;",
5926 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
5927 strcat(g_comment, " add64");
5928 }
5929 else {
5930 fprintf(fout, " cond_c = ((u32)%s + %s) >> %d;\n",
5931 buf1, buf2, lmod_bytes(po, po->operand[0].lmod) * 8);
5932 fprintf(fout, " %s += %s;",
5933 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5934 buf2);
5935 }
5936 pfomask &= ~(1 << PFO_C);
5937 output_std_flags(fout, po, &pfomask, buf1);
5938 last_arith_dst = &po->operand[0];
5939 delayed_flag_op = NULL;
5940 break;
5941 }
5942 goto dualop_arith;
5943
5944 case OP_SUB:
5945 assert_operand_cnt(2);
5946 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5947 if (pfomask & ~((1 << PFO_Z) | (1 << PFO_S))) {
5948 for (j = 0; j <= PFO_LE; j++) {
5949 if (!(pfomask & (1 << j)))
5950 continue;
5951 if (j == PFO_Z || j == PFO_S)
5952 continue;
5953
5954 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
5955 fprintf(fout, " cond_%s = %s;\n",
5956 parsed_flag_op_names[j], buf1);
5957 pfomask &= ~(1 << j);
5958 }
5959 }
5960 goto dualop_arith;
5961
5962 case OP_ADC:
5963 case OP_SBB:
5964 assert_operand_cnt(2);
5965 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5966 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5967 if (po->op == OP_SBB
5968 && IS(po->operand[0].name, po->operand[1].name))
5969 {
5970 // avoid use of unitialized var
5971 fprintf(fout, " %s = -cond_c;", buf1);
5972 // carry remains what it was
5973 pfomask &= ~(1 << PFO_C);
5974 }
5975 else {
5976 fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po),
5977 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5978 }
5979 output_std_flags(fout, po, &pfomask, buf1);
5980 last_arith_dst = &po->operand[0];
5981 delayed_flag_op = NULL;
5982 break;
5983
5984 case OP_BSF:
5985 assert_operand_cnt(2);
5986 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
5987 fprintf(fout, " %s = %s ? __builtin_ffs(%s) - 1 : 0;",
5988 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5989 buf2, buf2);
5990 output_std_flags(fout, po, &pfomask, buf1);
5991 last_arith_dst = &po->operand[0];
5992 delayed_flag_op = NULL;
5993 strcat(g_comment, " bsf");
5994 break;
5995
5996 case OP_DEC:
5997 if (pfomask & ~(PFOB_S | PFOB_S | PFOB_C)) {
5998 for (j = 0; j <= PFO_LE; j++) {
5999 if (!(pfomask & (1 << j)))
6000 continue;
6001 if (j == PFO_Z || j == PFO_S || j == PFO_C)
6002 continue;
6003
6004 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
6005 fprintf(fout, " cond_%s = %s;\n",
6006 parsed_flag_op_names[j], buf1);
6007 pfomask &= ~(1 << j);
6008 }
6009 }
6010 // fallthrough
6011
6012 case OP_INC:
6013 if (pfomask & (1 << PFO_C))
6014 // carry is unaffected by inc/dec.. wtf?
6015 ferr(po, "carry propagation needed\n");
6016
6017 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
6018 if (po->operand[0].type == OPT_REG) {
6019 strcpy(buf2, po->op == OP_INC ? "++" : "--");
6020 fprintf(fout, " %s%s;", buf1, buf2);
6021 }
6022 else {
6023 strcpy(buf2, po->op == OP_INC ? "+" : "-");
6024 fprintf(fout, " %s %s= 1;", buf1, buf2);
6025 }
6026 output_std_flags(fout, po, &pfomask, buf1);
6027 last_arith_dst = &po->operand[0];
6028 delayed_flag_op = NULL;
6029 break;
6030
6031 case OP_NEG:
6032 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
6033 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
6034 fprintf(fout, " %s = -%s%s;", buf1,
6035 lmod_cast_s(po, po->operand[0].lmod), buf2);
6036 last_arith_dst = &po->operand[0];
6037 delayed_flag_op = NULL;
6038 if (pfomask & (1 << PFO_C)) {
6039 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
6040 pfomask &= ~(1 << PFO_C);
6041 }
6042 break;
6043
6044 case OP_IMUL:
6045 if (po->operand_cnt == 2) {
6046 propagate_lmod(po, &po->operand[0], &po->operand[1]);
6047 goto dualop_arith;
6048 }
6049 if (po->operand_cnt == 3)
6050 ferr(po, "TODO imul3\n");
6051 // fallthrough
6052 case OP_MUL:
6053 assert_operand_cnt(1);
6054 switch (po->operand[0].lmod) {
6055 case OPLM_DWORD:
6056 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
6057 fprintf(fout, " tmp64 = %seax * %s%s;\n", buf1, buf1,
6058 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
6059 fprintf(fout, " edx = tmp64 >> 32;\n");
6060 fprintf(fout, " eax = tmp64;");
6061 break;
6062 case OPLM_BYTE:
6063 strcpy(buf1, po->op == OP_IMUL ? "(s16)(s8)" : "(u16)(u8)");
6064 fprintf(fout, " LOWORD(eax) = %seax * %s;", buf1,
6065 out_src_opr(buf2, sizeof(buf2), po, &po->operand[0],
6066 buf1, 0));
6067 break;
6068 default:
6069 ferr(po, "TODO: unhandled mul type\n");
6070 break;
6071 }
6072 last_arith_dst = NULL;
6073 delayed_flag_op = NULL;
6074 break;
6075
6076 case OP_DIV:
6077 case OP_IDIV:
6078 assert_operand_cnt(1);
6079 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
6080 strcpy(cast, lmod_cast(po, po->operand[0].lmod,
6081 po->op == OP_IDIV));
6082 switch (po->operand[0].lmod) {
6083 case OPLM_DWORD:
6084 if (po->flags & OPF_32BIT)
6085 snprintf(buf2, sizeof(buf2), "%seax", cast);
6086 else {
6087 fprintf(fout, " tmp64 = ((u64)edx << 32) | eax;\n");
6088 snprintf(buf2, sizeof(buf2), "%stmp64",
6089 (po->op == OP_IDIV) ? "(s64)" : "");
6090 }
6091 if (po->operand[0].type == OPT_REG
6092 && po->operand[0].reg == xDX)
6093 {
6094 fprintf(fout, " eax = %s / %s%s;\n", buf2, cast, buf1);
6095 fprintf(fout, " edx = %s %% %s%s;", buf2, cast, buf1);
6096 }
6097 else {
6098 fprintf(fout, " edx = %s %% %s%s;\n", buf2, cast, buf1);
6099 fprintf(fout, " eax = %s / %s%s;", buf2, cast, buf1);
6100 }
6101 break;
6102 case OPLM_WORD:
6103 fprintf(fout, " tmp = (edx << 16) | (eax & 0xffff);\n");
6104 snprintf(buf2, sizeof(buf2), "%stmp",
6105 (po->op == OP_IDIV) ? "(s32)" : "");
6106 if (po->operand[0].type == OPT_REG
6107 && po->operand[0].reg == xDX)
6108 {
6109 fprintf(fout, " LOWORD(eax) = %s / %s%s;\n",
6110 buf2, cast, buf1);
6111 fprintf(fout, " LOWORD(edx) = %s %% %s%s;",
6112 buf2, cast, buf1);
6113 }
6114 else {
6115 fprintf(fout, " LOWORD(edx) = %s %% %s%s;\n",
6116 buf2, cast, buf1);
6117 fprintf(fout, " LOWORD(eax) = %s / %s%s;",
6118 buf2, cast, buf1);
6119 }
6120 strcat(g_comment, " div16");
6121 break;
6122 default:
6123 ferr(po, "unhandled div lmod %d\n", po->operand[0].lmod);
6124 }
6125 last_arith_dst = NULL;
6126 delayed_flag_op = NULL;
6127 break;
6128
6129 case OP_TEST:
6130 case OP_CMP:
6131 propagate_lmod(po, &po->operand[0], &po->operand[1]);
6132 if (pfomask != 0) {
6133 for (j = 0; j < 8; j++) {
6134 if (pfomask & (1 << j)) {
6135 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
6136 fprintf(fout, " cond_%s = %s;",
6137 parsed_flag_op_names[j], buf1);
6138 }
6139 }
6140 pfomask = 0;
6141 }
6142 else
6143 no_output = 1;
6144 last_arith_dst = NULL;
6145 delayed_flag_op = po;
6146 break;
6147
6148 case OP_SCC:
6149 // SETcc - should already be handled
6150 break;
6151
6152 // note: we reuse OP_Jcc for SETcc, only flags differ
6153 case OP_JCC:
6154 fprintf(fout, "\n goto %s;", po->operand[0].name);
6155 break;
6156
6157 case OP_JECXZ:
6158 fprintf(fout, " if (ecx == 0)\n");
6159 fprintf(fout, " goto %s;", po->operand[0].name);
6160 strcat(g_comment, " jecxz");
6161 break;
6162
6163 case OP_LOOP:
6164 fprintf(fout, " if (--ecx != 0)\n");
6165 fprintf(fout, " goto %s;", po->operand[0].name);
6166 strcat(g_comment, " loop");
6167 break;
6168
6169 case OP_JMP:
6170 assert_operand_cnt(1);
6171 last_arith_dst = NULL;
6172 delayed_flag_op = NULL;
6173
6174 if (po->operand[0].type == OPT_REGMEM) {
6175 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
6176 buf1, buf2);
6177 if (ret != 2)
6178 ferr(po, "parse failure for jmp '%s'\n",
6179 po->operand[0].name);
6180 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
6181 break;
6182 }
6183 else if (po->operand[0].type != OPT_LABEL)
6184 ferr(po, "unhandled jmp type\n");
6185
6186 fprintf(fout, " goto %s;", po->operand[0].name);
6187 break;
6188
6189 case OP_CALL:
6190 assert_operand_cnt(1);
6191 pp = po->pp;
6192 my_assert_not(pp, NULL);
6193
6194 strcpy(buf3, " ");
6195 if (po->flags & OPF_CC) {
6196 // we treat conditional branch to another func
6197 // (yes such code exists..) as conditional tailcall
6198 strcat(buf3, " ");
6199 fprintf(fout, " {\n");
6200 }
6201
6202 if (pp->is_fptr && !pp->is_arg) {
6203 fprintf(fout, "%s%s = %s;\n", buf3, pp->name,
6204 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
6205 "(void *)", 0));
6206 if (pp->is_unresolved)
6207 fprintf(fout, "%sunresolved_call(\"%s:%d\", %s);\n",
6208 buf3, asmfn, po->asmln, pp->name);
6209 }
6210
6211 fprintf(fout, "%s", buf3);
6212 if (strstr(pp->ret_type.name, "int64")) {
6213 if (po->flags & OPF_TAIL)
6214 ferr(po, "int64 and tail?\n");
6215 fprintf(fout, "tmp64 = ");
6216 }
6217 else if (!IS(pp->ret_type.name, "void")) {
6218 if (po->flags & OPF_TAIL) {
6219 if (regmask_ret & mxAX) {
6220 fprintf(fout, "return ");
6221 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
6222 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
6223 }
6224 else if (regmask_ret & mxST0)
6225 ferr(po, "float tailcall\n");
6226 }
6227 else if (po->regmask_dst & mxAX) {
6228 fprintf(fout, "eax = ");
6229 if (pp->ret_type.is_ptr)
6230 fprintf(fout, "(u32)");
6231 }
6232 else if (po->regmask_dst & mxST0) {
6233 fprintf(fout, "f_st0 = ");
6234 }
6235 }
6236
6237 if (pp->name[0] == 0)
6238 ferr(po, "missing pp->name\n");
6239 fprintf(fout, "%s%s(", pp->name,
6240 pp->has_structarg ? "_sa" : "");
6241
6242 if (po->flags & OPF_ATAIL) {
6243 if (pp->argc_stack != g_func_pp->argc_stack
6244 || (pp->argc_stack > 0
6245 && pp->is_stdcall != g_func_pp->is_stdcall))
6246 ferr(po, "incompatible tailcall\n");
6247 if (g_func_pp->has_retreg)
6248 ferr(po, "TODO: retreg+tailcall\n");
6249
6250 for (arg = j = 0; arg < pp->argc; arg++) {
6251 if (arg > 0)
6252 fprintf(fout, ", ");
6253
6254 cast[0] = 0;
6255 if (pp->arg[arg].type.is_ptr)
6256 snprintf(cast, sizeof(cast), "(%s)",
6257 pp->arg[arg].type.name);
6258
6259 if (pp->arg[arg].reg != NULL) {
6260 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
6261 continue;
6262 }
6263 // stack arg
6264 for (; j < g_func_pp->argc; j++)
6265 if (g_func_pp->arg[j].reg == NULL)
6266 break;
6267 fprintf(fout, "%sa%d", cast, j + 1);
6268 j++;
6269 }
6270 }
6271 else {
6272 for (arg = 0; arg < pp->argc; arg++) {
6273 if (arg > 0)
6274 fprintf(fout, ", ");
6275
6276 cast[0] = 0;
6277 if (pp->arg[arg].type.is_ptr)
6278 snprintf(cast, sizeof(cast), "(%s)",
6279 pp->arg[arg].type.name);
6280
6281 if (pp->arg[arg].reg != NULL) {
6282 if (pp->arg[arg].type.is_retreg)
6283 fprintf(fout, "&%s", pp->arg[arg].reg);
6284 else
6285 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
6286 continue;
6287 }
6288
6289 // stack arg
6290 tmp_op = pp->arg[arg].datap;
6291 if (tmp_op == NULL)
6292 ferr(po, "parsed_op missing for arg%d\n", arg);
6293
6294 if (tmp_op->flags & OPF_VAPUSH) {
6295 fprintf(fout, "ap");
6296 }
6297 else if (tmp_op->p_argpass != 0) {
6298 fprintf(fout, "a%d", tmp_op->p_argpass);
6299 }
6300 else if (tmp_op->p_argnum != 0) {
6301 fprintf(fout, "%s%s", cast,
6302 saved_arg_name(buf1, sizeof(buf1),
6303 tmp_op->p_arggrp, tmp_op->p_argnum));
6304 }
6305 else {
6306 fprintf(fout, "%s",
6307 out_src_opr(buf1, sizeof(buf1),
6308 tmp_op, &tmp_op->operand[0], cast, 0));
6309 }
6310 }
6311 }
6312 fprintf(fout, ");");
6313
6314 if (strstr(pp->ret_type.name, "int64")) {
6315 fprintf(fout, "\n");
6316 fprintf(fout, "%sedx = tmp64 >> 32;\n", buf3);
6317 fprintf(fout, "%seax = tmp64;", buf3);
6318 }
6319
6320 if (pp->is_unresolved) {
6321 snprintf(buf2, sizeof(buf2), " unresolved %dreg",
6322 pp->argc_reg);
6323 strcat(g_comment, buf2);
6324 }
6325
6326 if (po->flags & OPF_TAIL) {
6327 ret = 0;
6328 if (i == opcnt - 1 || pp->is_noreturn)
6329 ret = 0;
6330 else if (IS(pp->ret_type.name, "void"))
6331 ret = 1;
6332 else if (!(regmask_ret & (1 << xAX)))
6333 ret = 1;
6334 // else already handled as 'return f()'
6335
6336 if (ret) {
6337 fprintf(fout, "\n%sreturn;", buf3);
6338 strcat(g_comment, " ^ tailcall");
6339 }
6340 else
6341 strcat(g_comment, " tailcall");
6342
6343 if ((regmask_ret & (1 << xAX))
6344 && IS(pp->ret_type.name, "void") && !pp->is_noreturn)
6345 {
6346 ferr(po, "int func -> void func tailcall?\n");
6347 }
6348 }
6349 if (pp->is_noreturn)
6350 strcat(g_comment, " noreturn");
6351 if ((po->flags & OPF_ATAIL) && pp->argc_stack > 0)
6352 strcat(g_comment, " argframe");
6353 if (po->flags & OPF_CC)
6354 strcat(g_comment, " cond");
6355
6356 if (po->flags & OPF_CC)
6357 fprintf(fout, "\n }");
6358
6359 delayed_flag_op = NULL;
6360 last_arith_dst = NULL;
6361 break;
6362
6363 case OP_RET:
6364 if (g_func_pp->is_vararg)
6365 fprintf(fout, " va_end(ap);\n");
6366 if (g_func_pp->has_retreg) {
6367 for (arg = 0; arg < g_func_pp->argc; arg++)
6368 if (g_func_pp->arg[arg].type.is_retreg)
6369 fprintf(fout, " *r_%s = %s;\n",
6370 g_func_pp->arg[arg].reg, g_func_pp->arg[arg].reg);
6371 }
6372
6373 if (!(regmask_ret & (1 << xAX))) {
6374 if (i != opcnt - 1 || label_pending)
6375 fprintf(fout, " return;");
6376 }
6377 else if (g_func_pp->ret_type.is_ptr) {
6378 fprintf(fout, " return (%s)eax;",
6379 g_func_pp->ret_type.name);
6380 }
6381 else if (IS(g_func_pp->ret_type.name, "__int64"))
6382 fprintf(fout, " return ((u64)edx << 32) | eax;");
6383 else
6384 fprintf(fout, " return eax;");
6385
6386 last_arith_dst = NULL;
6387 delayed_flag_op = NULL;
6388 break;
6389
6390 case OP_PUSH:
6391 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
6392 if (po->p_argnum != 0) {
6393 // special case - saved func arg
6394 fprintf(fout, " %s = %s;",
6395 saved_arg_name(buf2, sizeof(buf2),
6396 po->p_arggrp, po->p_argnum), buf1);
6397 break;
6398 }
6399 else if (po->flags & OPF_RSAVE) {
6400 fprintf(fout, " s_%s = %s;", buf1, buf1);
6401 break;
6402 }
6403 else if (po->flags & OPF_PPUSH) {
6404 tmp_op = po->datap;
6405 ferr_assert(po, tmp_op != NULL);
6406 out_dst_opr(buf2, sizeof(buf2), po, &tmp_op->operand[0]);
6407 fprintf(fout, " pp_%s = %s;", buf2, buf1);
6408 break;
6409 }
6410 else if (g_func_pp->is_userstack) {
6411 fprintf(fout, " *(--esp) = %s;", buf1);
6412 break;
6413 }
6414 if (!(g_ida_func_attr & IDAFA_NORETURN))
6415 ferr(po, "stray push encountered\n");
6416 no_output = 1;
6417 break;
6418
6419 case OP_POP:
6420 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
6421 if (po->flags & OPF_RSAVE) {
6422 fprintf(fout, " %s = s_%s;", buf1, buf1);
6423 break;
6424 }
6425 else if (po->flags & OPF_PPUSH) {
6426 // push/pop graph / non-const
6427 ferr_assert(po, po->datap == NULL);
6428 fprintf(fout, " %s = pp_%s;", buf1, buf1);
6429 break;
6430 }
6431 else if (po->datap != NULL) {
6432 // push/pop pair
6433 tmp_op = po->datap;
6434 fprintf(fout, " %s = %s;", buf1,
6435 out_src_opr(buf2, sizeof(buf2),
6436 tmp_op, &tmp_op->operand[0],
6437 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
6438 break;
6439 }
6440 else if (g_func_pp->is_userstack) {
6441 fprintf(fout, " %s = *esp++;", buf1);
6442 break;
6443 }
6444 else
6445 ferr(po, "stray pop encountered\n");
6446 break;
6447
6448 case OP_NOP:
6449 no_output = 1;
6450 break;
6451
6452 // x87
6453 case OP_FLD:
6454 if (po->flags & OPF_FSHIFT)
6455 fprintf(fout, " f_st1 = f_st0;\n");
6456 if (po->operand[0].type == OPT_REG
6457 && po->operand[0].reg == xST0)
6458 {
6459 strcat(g_comment, " fld st");
6460 break;
6461 }
6462 fprintf(fout, " f_st0 = %s;",
6463 out_src_opr_float(buf1, sizeof(buf1), po, &po->operand[0]));
6464 strcat(g_comment, " fld");
6465 break;
6466
6467 case OP_FILD:
6468 if (po->flags & OPF_FSHIFT)
6469 fprintf(fout, " f_st1 = f_st0;\n");
6470 fprintf(fout, " f_st0 = (double)%s;",
6471 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
6472 lmod_cast(po, po->operand[0].lmod, 1), 0));
6473 strcat(g_comment, " fild");
6474 break;
6475
6476 case OP_FLDc:
6477 if (po->flags & OPF_FSHIFT)
6478 fprintf(fout, " f_st1 = f_st0;\n");
6479 fprintf(fout, " f_st0 = ");
6480 switch (po->operand[0].val) {
6481 case X87_CONST_1: fprintf(fout, "1.0;"); break;
6482 case X87_CONST_Z: fprintf(fout, "0.0;"); break;
6483 default: ferr(po, "TODO\n"); break;
6484 }
6485 break;
6486
6487 case OP_FST:
6488 if ((po->flags & OPF_FPOP) && po->operand[0].type == OPT_REG
6489 && po->operand[0].reg == xST0)
6490 {
6491 no_output = 1;
6492 break;
6493 }
6494 fprintf(fout, " %s = f_st0;",
6495 out_dst_opr_float(buf1, sizeof(buf1), po, &po->operand[0]));
6496 if (po->flags & OPF_FSHIFT)
6497 fprintf(fout, "\n f_st0 = f_st1;");
6498 strcat(g_comment, " fst");
6499 break;
6500
6501 case OP_FADD:
6502 case OP_FDIV:
6503 case OP_FMUL:
6504 case OP_FSUB:
6505 switch (po->op) {
6506 case OP_FADD: j = '+'; break;
6507 case OP_FDIV: j = '/'; break;
6508 case OP_FMUL: j = '*'; break;
6509 case OP_FSUB: j = '-'; break;
6510 default: j = 'x'; break;
6511 }
6512 if (po->flags & OPF_FSHIFT) {
6513 fprintf(fout, " f_st0 = f_st1 %c f_st0;", j);
6514 }
6515 else {
6516 fprintf(fout, " %s %c= %s;",
6517 out_dst_opr_float(buf1, sizeof(buf1), po, &po->operand[0]),
6518 j,
6519 out_src_opr_float(buf2, sizeof(buf2), po, &po->operand[1]));
6520 }
6521 break;
6522
6523 case OP_FDIVR:
6524 case OP_FSUBR:
6525 if (po->flags & OPF_FSHIFT)
6526 snprintf(buf1, sizeof(buf1), "f_st0");
6527 else
6528 out_dst_opr_float(buf1, sizeof(buf1), po, &po->operand[0]);
6529 fprintf(fout, " %s = %s %c %s;", buf1,
6530 out_src_opr_float(buf2, sizeof(buf2), po, &po->operand[1]),
6531 po->op == OP_FDIVR ? '/' : '-',
6532 out_src_opr_float(buf3, sizeof(buf3), po, &po->operand[0]));
6533 break;
6534
6535 case OP_FIADD:
6536 case OP_FIDIV:
6537 case OP_FIMUL:
6538 case OP_FISUB:
6539 switch (po->op) {
6540 case OP_FIADD: j = '+'; break;
6541 case OP_FIDIV: j = '/'; break;
6542 case OP_FIMUL: j = '*'; break;
6543 case OP_FISUB: j = '-'; break;
6544 default: j = 'x'; break;
6545 }
6546 fprintf(fout, " f_st0 %c= (double)%s;", j,
6547 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
6548 lmod_cast(po, po->operand[0].lmod, 1), 0));
6549 break;
6550
6551 case OP_FIDIVR:
6552 case OP_FISUBR:
6553 fprintf(fout, " f_st0 = %s %c f_st0;",
6554 out_src_opr_float(buf2, sizeof(buf2), po, &po->operand[1]),
6555 po->op == OP_FIDIVR ? '/' : '-');
6556 break;
6557
6558 case OPP_FTOL:
6559 ferr_assert(po, po->flags & OPF_32BIT);
6560 fprintf(fout, " eax = (s32)f_st0;");
6561 if (po->flags & OPF_FSHIFT)
6562 fprintf(fout, "\n f_st0 = f_st1;");
6563 strcat(g_comment, " ftol");
6564 break;
6565
6566 // mmx
6567 case OP_EMMS:
6568 strcpy(g_comment, " (emms)");
6569 break;
6570
6571 default:
6572 no_output = 1;
6573 ferr(po, "unhandled op type %d, flags %x\n",
6574 po->op, po->flags);
6575 break;
6576 }
6577
6578 if (g_comment[0] != 0) {
6579 char *p = g_comment;
6580 while (my_isblank(*p))
6581 p++;
6582 fprintf(fout, " // %s", p);
6583 g_comment[0] = 0;
6584 no_output = 0;
6585 }
6586 if (!no_output)
6587 fprintf(fout, "\n");
6588
6589 // some sanity checking
6590 if (po->flags & OPF_REP) {
6591 if (po->op != OP_STOS && po->op != OP_MOVS
6592 && po->op != OP_CMPS && po->op != OP_SCAS)
6593 ferr(po, "unexpected rep\n");
6594 if (!(po->flags & (OPF_REPZ|OPF_REPNZ))
6595 && (po->op == OP_CMPS || po->op == OP_SCAS))
6596 ferr(po, "cmps/scas with plain rep\n");
6597 }
6598 if ((po->flags & (OPF_REPZ|OPF_REPNZ))
6599 && po->op != OP_CMPS && po->op != OP_SCAS)
6600 ferr(po, "unexpected repz/repnz\n");
6601
6602 if (pfomask != 0)
6603 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
6604
6605 // see is delayed flag stuff is still valid
6606 if (delayed_flag_op != NULL && delayed_flag_op != po) {
6607 if (is_any_opr_modified(delayed_flag_op, po, 0))
6608 delayed_flag_op = NULL;
6609 }
6610
6611 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
6612 if (is_opr_modified(last_arith_dst, po))
6613 last_arith_dst = NULL;
6614 }
6615
6616 label_pending = 0;
6617 }
6618
6619 if (g_stack_fsz && !g_stack_frame_used)
6620 fprintf(fout, " (void)sf;\n");
6621
6622 fprintf(fout, "}\n\n");
6623
6624 gen_x_cleanup(opcnt);
6625}
6626
6627static void gen_x_cleanup(int opcnt)
6628{
6629 int i;
6630
6631 for (i = 0; i < opcnt; i++) {
6632 struct label_ref *lr, *lr_del;
6633
6634 lr = g_label_refs[i].next;
6635 while (lr != NULL) {
6636 lr_del = lr;
6637 lr = lr->next;
6638 free(lr_del);
6639 }
6640 g_label_refs[i].i = -1;
6641 g_label_refs[i].next = NULL;
6642
6643 if (ops[i].op == OP_CALL) {
6644 if (ops[i].pp)
6645 proto_release(ops[i].pp);
6646 }
6647 }
6648 g_func_pp = NULL;
6649}
6650
6651struct func_proto_dep;
6652
6653struct func_prototype {
6654 char name[NAMELEN];
6655 int id;
6656 int argc_stack;
6657 int regmask_dep;
6658 int has_ret:3; // -1, 0, 1: unresolved, no, yes
6659 unsigned int dep_resolved:1;
6660 unsigned int is_stdcall:1;
6661 struct func_proto_dep *dep_func;
6662 int dep_func_cnt;
6663 const struct parsed_proto *pp; // seed pp, if any
6664};
6665
6666struct func_proto_dep {
6667 char *name;
6668 struct func_prototype *proto;
6669 int regmask_live; // .. at the time of call
6670 unsigned int ret_dep:1; // return from this is caller's return
6671};
6672
6673static struct func_prototype *hg_fp;
6674static int hg_fp_cnt;
6675
6676static struct scanned_var {
6677 char name[NAMELEN];
6678 enum opr_lenmod lmod;
6679 unsigned int is_seeded:1;
6680 unsigned int is_c_str:1;
6681 const struct parsed_proto *pp; // seed pp, if any
6682} *hg_vars;
6683static int hg_var_cnt;
6684
6685static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
6686 int count);
6687
6688struct func_prototype *hg_fp_add(const char *funcn)
6689{
6690 struct func_prototype *fp;
6691
6692 if ((hg_fp_cnt & 0xff) == 0) {
6693 hg_fp = realloc(hg_fp, sizeof(hg_fp[0]) * (hg_fp_cnt + 0x100));
6694 my_assert_not(hg_fp, NULL);
6695 memset(hg_fp + hg_fp_cnt, 0, sizeof(hg_fp[0]) * 0x100);
6696 }
6697
6698 fp = &hg_fp[hg_fp_cnt];
6699 snprintf(fp->name, sizeof(fp->name), "%s", funcn);
6700 fp->id = hg_fp_cnt;
6701 fp->argc_stack = -1;
6702 hg_fp_cnt++;
6703
6704 return fp;
6705}
6706
6707static struct func_proto_dep *hg_fp_find_dep(struct func_prototype *fp,
6708 const char *name)
6709{
6710 int i;
6711
6712 for (i = 0; i < fp->dep_func_cnt; i++)
6713 if (IS(fp->dep_func[i].name, name))
6714 return &fp->dep_func[i];
6715
6716 return NULL;
6717}
6718
6719static void hg_fp_add_dep(struct func_prototype *fp, const char *name)
6720{
6721 // is it a dupe?
6722 if (hg_fp_find_dep(fp, name))
6723 return;
6724
6725 if ((fp->dep_func_cnt & 0xff) == 0) {
6726 fp->dep_func = realloc(fp->dep_func,
6727 sizeof(fp->dep_func[0]) * (fp->dep_func_cnt + 0x100));
6728 my_assert_not(fp->dep_func, NULL);
6729 memset(&fp->dep_func[fp->dep_func_cnt], 0,
6730 sizeof(fp->dep_func[0]) * 0x100);
6731 }
6732 fp->dep_func[fp->dep_func_cnt].name = strdup(name);
6733 fp->dep_func_cnt++;
6734}
6735
6736static int hg_fp_cmp_name(const void *p1_, const void *p2_)
6737{
6738 const struct func_prototype *p1 = p1_, *p2 = p2_;
6739 return strcmp(p1->name, p2->name);
6740}
6741
6742#if 0
6743static int hg_fp_cmp_id(const void *p1_, const void *p2_)
6744{
6745 const struct func_prototype *p1 = p1_, *p2 = p2_;
6746 return p1->id - p2->id;
6747}
6748#endif
6749
6750// recursive register dep pass
6751// - track saved regs (part 2)
6752// - try to figure out arg-regs
6753// - calculate reg deps
6754static void gen_hdr_dep_pass(int i, int opcnt, unsigned char *cbits,
6755 struct func_prototype *fp, int regmask_save, int regmask_dst,
6756 int *regmask_dep, int *has_ret)
6757{
6758 struct func_proto_dep *dep;
6759 struct parsed_op *po;
6760 int from_caller = 0;
6761 int j, l;
6762 int reg;
6763 int ret;
6764
6765 for (; i < opcnt; i++)
6766 {
6767 if (cbits[i >> 3] & (1 << (i & 7)))
6768 return;
6769 cbits[i >> 3] |= (1 << (i & 7));
6770
6771 po = &ops[i];
6772
6773 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
6774 if (po->flags & OPF_RMD)
6775 continue;
6776
6777 if (po->btj != NULL) {
6778 // jumptable
6779 for (j = 0; j < po->btj->count; j++) {
6780 check_i(po, po->btj->d[j].bt_i);
6781 gen_hdr_dep_pass(po->btj->d[j].bt_i, opcnt, cbits, fp,
6782 regmask_save, regmask_dst, regmask_dep, has_ret);
6783 }
6784 return;
6785 }
6786
6787 check_i(po, po->bt_i);
6788 if (po->flags & OPF_CJMP) {
6789 gen_hdr_dep_pass(po->bt_i, opcnt, cbits, fp,
6790 regmask_save, regmask_dst, regmask_dep, has_ret);
6791 }
6792 else {
6793 i = po->bt_i - 1;
6794 }
6795 continue;
6796 }
6797
6798 if (po->flags & OPF_FARG)
6799 /* (just calculate register deps) */;
6800 else if (po->op == OP_PUSH && po->operand[0].type == OPT_REG)
6801 {
6802 reg = po->operand[0].reg;
6803 ferr_assert(po, reg >= 0);
6804
6805 if (po->flags & OPF_RSAVE) {
6806 regmask_save |= 1 << reg;
6807 continue;
6808 }
6809 if (po->flags & OPF_DONE)
6810 continue;
6811
6812 ret = scan_for_pop(i + 1, opcnt, i + opcnt * 2, reg, 0, 0);
6813 if (ret == 1) {
6814 regmask_save |= 1 << reg;
6815 po->flags |= OPF_RMD;
6816 scan_for_pop(i + 1, opcnt, i + opcnt * 3, reg, 0, OPF_RMD);
6817 continue;
6818 }
6819 }
6820 else if (po->flags & OPF_RMD)
6821 continue;
6822 else if (po->op == OP_CALL) {
6823 po->regmask_dst |= 1 << xAX;
6824
6825 dep = hg_fp_find_dep(fp, po->operand[0].name);
6826 if (dep != NULL)
6827 dep->regmask_live = regmask_save | regmask_dst;
6828 }
6829 else if (po->op == OP_RET) {
6830 if (po->operand_cnt > 0) {
6831 fp->is_stdcall = 1;
6832 if (fp->argc_stack >= 0
6833 && fp->argc_stack != po->operand[0].val / 4)
6834 ferr(po, "ret mismatch? (%d)\n", fp->argc_stack * 4);
6835 fp->argc_stack = po->operand[0].val / 4;
6836 }
6837 }
6838
6839 // if has_ret is 0, there is uninitialized eax path,
6840 // which means it's most likely void func
6841 if (*has_ret != 0 && (po->flags & OPF_TAIL)) {
6842 if (po->op == OP_CALL) {
6843 j = i;
6844 ret = 1;
6845 }
6846 else {
6847 struct parsed_opr opr = OPR_INIT(OPT_REG, OPLM_DWORD, xAX);
6848 j = -1;
6849 from_caller = 0;
6850 ret = resolve_origin(i, &opr, i + opcnt * 4, &j, &from_caller);
6851 }
6852
6853 if (ret != 1 && from_caller) {
6854 // unresolved eax - probably void func
6855 *has_ret = 0;
6856 }
6857 else {
6858 if (j >= 0 && ops[j].op == OP_CALL) {
6859 dep = hg_fp_find_dep(fp, ops[j].operand[0].name);
6860 if (dep != NULL)
6861 dep->ret_dep = 1;
6862 else
6863 *has_ret = 1;
6864 }
6865 else
6866 *has_ret = 1;
6867 }
6868 }
6869
6870 l = regmask_save | regmask_dst;
6871 if (g_bp_frame && !(po->flags & OPF_EBP_S))
6872 l |= 1 << xBP;
6873
6874 l = po->regmask_src & ~l;
6875#if 0
6876 if (l)
6877 fnote(po, "dep |= %04x, dst %04x, save %04x (f %x)\n",
6878 l, regmask_dst, regmask_save, po->flags);
6879#endif
6880 *regmask_dep |= l;
6881 regmask_dst |= po->regmask_dst;
6882
6883 if (po->flags & OPF_TAIL)
6884 return;
6885 }
6886}
6887
6888static void gen_hdr(const char *funcn, int opcnt)
6889{
6890 int save_arg_vars[MAX_ARG_GRP] = { 0, };
6891 unsigned char cbits[MAX_OPS / 8];
6892 const struct parsed_proto *pp_c;
6893 struct parsed_proto *pp;
6894 struct func_prototype *fp;
6895 struct parsed_op *po;
6896 int regmask_dummy = 0;
6897 int regmask_dep;
6898 int max_bp_offset = 0;
6899 int has_ret;
6900 int i, j, l;
6901 int ret;
6902
6903 pp_c = proto_parse(g_fhdr, funcn, 1);
6904 if (pp_c != NULL)
6905 // already in seed, will add to hg_fp later
6906 return;
6907
6908 fp = hg_fp_add(funcn);
6909
6910 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
6911 g_stack_frame_used = 0;
6912
6913 // pass1:
6914 // - resolve all branches
6915 // - parse calls with labels
6916 resolve_branches_parse_calls(opcnt);
6917
6918 // pass2:
6919 // - handle ebp/esp frame, remove ops related to it
6920 scan_prologue_epilogue(opcnt);
6921
6922 // pass3:
6923 // - remove dead labels
6924 // - collect calls
6925 for (i = 0; i < opcnt; i++)
6926 {
6927 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
6928 free(g_labels[i]);
6929 g_labels[i] = NULL;
6930 }
6931
6932 po = &ops[i];
6933 if (po->flags & (OPF_RMD|OPF_DONE))
6934 continue;
6935
6936 if (po->op == OP_CALL) {
6937 if (po->operand[0].type == OPT_LABEL)
6938 hg_fp_add_dep(fp, opr_name(po, 0));
6939 else if (po->pp != NULL)
6940 hg_fp_add_dep(fp, po->pp->name);
6941 }
6942 }
6943
6944 // pass4:
6945 // - remove dead labels
6946 // - handle push <const>/pop pairs
6947 for (i = 0; i < opcnt; i++)
6948 {
6949 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
6950 free(g_labels[i]);
6951 g_labels[i] = NULL;
6952 }
6953
6954 po = &ops[i];
6955 if (po->flags & (OPF_RMD|OPF_DONE))
6956 continue;
6957
6958 if (po->op == OP_PUSH && po->operand[0].type == OPT_CONST)
6959 scan_for_pop_const(i, opcnt, i + opcnt * 13);
6960 }
6961
6962 // pass5:
6963 // - process trivial calls
6964 for (i = 0; i < opcnt; i++)
6965 {
6966 po = &ops[i];
6967 if (po->flags & (OPF_RMD|OPF_DONE))
6968 continue;
6969
6970 if (po->op == OP_CALL)
6971 {
6972 pp = process_call_early(i, opcnt, &j);
6973 if (pp != NULL) {
6974 if (!(po->flags & OPF_ATAIL))
6975 // since we know the args, try to collect them
6976 if (collect_call_args_early(po, i, pp, &regmask_dummy) != 0)
6977 pp = NULL;
6978 }
6979
6980 if (pp != NULL) {
6981 if (j >= 0) {
6982 // commit esp adjust
6983 if (ops[j].op != OP_POP)
6984 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
6985 else {
6986 for (l = 0; l < pp->argc_stack; l++)
6987 ops[j + l].flags |= OPF_DONE | OPF_RMD | OPF_NOREGS;
6988 }
6989 }
6990
6991 po->flags |= OPF_DONE;
6992 }
6993 }
6994 }
6995
6996 // pass6:
6997 // - track saved regs (simple)
6998 // - process calls
6999 for (i = 0; i < opcnt; i++)
7000 {
7001 po = &ops[i];
7002 if (po->flags & (OPF_RMD|OPF_DONE))
7003 continue;
7004
7005 if (po->op == OP_PUSH && po->operand[0].type == OPT_REG
7006 && po->operand[0].reg != xCX)
7007 {
7008 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].reg, 0);
7009 if (ret == 1) {
7010 // regmask_save |= 1 << po->operand[0].reg; // do it later
7011 po->flags |= OPF_RSAVE | OPF_RMD | OPF_DONE;
7012 scan_for_pop_ret(i + 1, opcnt, po->operand[0].reg, OPF_RMD);
7013 }
7014 }
7015 else if (po->op == OP_CALL)
7016 {
7017 pp = process_call(i, opcnt);
7018
7019 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
7020 // since we know the args, collect them
7021 ret = collect_call_args(po, i, pp, &regmask_dummy, save_arg_vars,
7022 i + opcnt * 1);
7023 }
7024 }
7025 }
7026
7027 // pass7
7028 memset(cbits, 0, sizeof(cbits));
7029 regmask_dep = 0;
7030 has_ret = -1;
7031
7032 gen_hdr_dep_pass(0, opcnt, cbits, fp, 0, 0, &regmask_dep, &has_ret);
7033
7034 // find unreachable code - must be fixed in IDA
7035 for (i = 0; i < opcnt; i++)
7036 {
7037 if (cbits[i >> 3] & (1 << (i & 7)))
7038 continue;
7039
7040 if (g_labels[i] == NULL && i > 0 && ops[i - 1].op == OP_CALL
7041 && ops[i - 1].pp != NULL && ops[i - 1].pp->is_osinc)
7042 {
7043 // the compiler sometimes still generates code after
7044 // noreturn OS functions
7045 break;
7046 }
7047 if (ops[i].op != OP_NOP)
7048 ferr(&ops[i], "unreachable code\n");
7049 }
7050
7051 for (i = 0; i < g_eqcnt; i++) {
7052 if (g_eqs[i].offset > max_bp_offset && g_eqs[i].offset < 4*32)
7053 max_bp_offset = g_eqs[i].offset;
7054 }
7055
7056 if (fp->argc_stack < 0) {
7057 max_bp_offset = (max_bp_offset + 3) & ~3;
7058 fp->argc_stack = max_bp_offset / 4;
7059 if ((g_ida_func_attr & IDAFA_BP_FRAME) && fp->argc_stack > 0)
7060 fp->argc_stack--;
7061 }
7062
7063 fp->regmask_dep = regmask_dep & ~(1 << xSP);
7064 fp->has_ret = has_ret;
7065#if 0
7066 printf("// has_ret %d, regmask_dep %x\n",
7067 fp->has_ret, fp->regmask_dep);
7068 output_hdr_fp(stdout, fp, 1);
7069 if (IS(funcn, "sub_10007F72")) exit(1);
7070#endif
7071
7072 gen_x_cleanup(opcnt);
7073}
7074
7075static void hg_fp_resolve_deps(struct func_prototype *fp)
7076{
7077 struct func_prototype fp_s;
7078 int dep;
7079 int i;
7080
7081 // this thing is recursive, so mark first..
7082 fp->dep_resolved = 1;
7083
7084 for (i = 0; i < fp->dep_func_cnt; i++) {
7085 strcpy(fp_s.name, fp->dep_func[i].name);
7086 fp->dep_func[i].proto = bsearch(&fp_s, hg_fp, hg_fp_cnt,
7087 sizeof(hg_fp[0]), hg_fp_cmp_name);
7088 if (fp->dep_func[i].proto != NULL) {
7089 if (!fp->dep_func[i].proto->dep_resolved)
7090 hg_fp_resolve_deps(fp->dep_func[i].proto);
7091
7092 dep = ~fp->dep_func[i].regmask_live
7093 & fp->dep_func[i].proto->regmask_dep;
7094 fp->regmask_dep |= dep;
7095 // printf("dep %s %s |= %x\n", fp->name,
7096 // fp->dep_func[i].name, dep);
7097
7098 if (fp->has_ret == -1 && fp->dep_func[i].ret_dep)
7099 fp->has_ret = fp->dep_func[i].proto->has_ret;
7100 }
7101 }
7102}
7103
7104static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
7105 int count)
7106{
7107 const struct parsed_proto *pp;
7108 char *p, namebuf[NAMELEN];
7109 const char *name;
7110 int regmask_dep;
7111 int argc_stack;
7112 int j, arg;
7113
7114 for (; count > 0; count--, fp++) {
7115 if (fp->has_ret == -1)
7116 fprintf(fout, "// ret unresolved\n");
7117#if 0
7118 fprintf(fout, "// dep:");
7119 for (j = 0; j < fp->dep_func_cnt; j++) {
7120 fprintf(fout, " %s/", fp->dep_func[j].name);
7121 if (fp->dep_func[j].proto != NULL)
7122 fprintf(fout, "%04x/%d", fp->dep_func[j].proto->regmask_dep,
7123 fp->dep_func[j].proto->has_ret);
7124 }
7125 fprintf(fout, "\n");
7126#endif
7127
7128 p = strchr(fp->name, '@');
7129 if (p != NULL) {
7130 memcpy(namebuf, fp->name, p - fp->name);
7131 namebuf[p - fp->name] = 0;
7132 name = namebuf;
7133 }
7134 else
7135 name = fp->name;
7136 if (name[0] == '_')
7137 name++;
7138
7139 pp = proto_parse(g_fhdr, name, 1);
7140 if (pp != NULL && pp->is_include)
7141 continue;
7142
7143 if (fp->pp != NULL) {
7144 // part of seed, output later
7145 continue;
7146 }
7147
7148 regmask_dep = fp->regmask_dep;
7149 argc_stack = fp->argc_stack;
7150
7151 fprintf(fout, "%-5s", fp->pp ? fp->pp->ret_type.name :
7152 (fp->has_ret ? "int" : "void"));
7153 if (regmask_dep && (fp->is_stdcall || argc_stack == 0)
7154 && (regmask_dep & ~((1 << xCX) | (1 << xDX))) == 0)
7155 {
7156 fprintf(fout, " __fastcall ");
7157 if (!(regmask_dep & (1 << xDX)) && argc_stack == 0)
7158 argc_stack = 1;
7159 else
7160 argc_stack += 2;
7161 regmask_dep = 0;
7162 }
7163 else if (regmask_dep && !fp->is_stdcall) {
7164 fprintf(fout, "/*__usercall*/ ");
7165 }
7166 else if (regmask_dep) {
7167 fprintf(fout, "/*__userpurge*/ ");
7168 }
7169 else if (fp->is_stdcall)
7170 fprintf(fout, " __stdcall ");
7171 else
7172 fprintf(fout, " __cdecl ");
7173
7174 fprintf(fout, "%s(", name);
7175
7176 arg = 0;
7177 for (j = 0; j < xSP; j++) {
7178 if (regmask_dep & (1 << j)) {
7179 arg++;
7180 if (arg != 1)
7181 fprintf(fout, ", ");
7182 if (fp->pp != NULL)
7183 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
7184 else
7185 fprintf(fout, "int");
7186 fprintf(fout, " a%d/*<%s>*/", arg, regs_r32[j]);
7187 }
7188 }
7189
7190 for (j = 0; j < argc_stack; j++) {
7191 arg++;
7192 if (arg != 1)
7193 fprintf(fout, ", ");
7194 if (fp->pp != NULL) {
7195 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
7196 if (!fp->pp->arg[arg - 1].type.is_ptr)
7197 fprintf(fout, " ");
7198 }
7199 else
7200 fprintf(fout, "int ");
7201 fprintf(fout, "a%d", arg);
7202 }
7203
7204 fprintf(fout, ");\n");
7205 }
7206}
7207
7208static void output_hdr(FILE *fout)
7209{
7210 static const char *lmod_c_names[] = {
7211 [OPLM_UNSPEC] = "???",
7212 [OPLM_BYTE] = "uint8_t",
7213 [OPLM_WORD] = "uint16_t",
7214 [OPLM_DWORD] = "uint32_t",
7215 [OPLM_QWORD] = "uint64_t",
7216 };
7217 const struct scanned_var *var;
7218 struct func_prototype *fp;
7219 char line[256] = { 0, };
7220 char name[256];
7221 int i;
7222
7223 // add stuff from headers
7224 for (i = 0; i < pp_cache_size; i++) {
7225 if (pp_cache[i].is_cinc && !pp_cache[i].is_stdcall)
7226 snprintf(name, sizeof(name), "_%s", pp_cache[i].name);
7227 else
7228 snprintf(name, sizeof(name), "%s", pp_cache[i].name);
7229 fp = hg_fp_add(name);
7230 fp->pp = &pp_cache[i];
7231 fp->argc_stack = fp->pp->argc_stack;
7232 fp->is_stdcall = fp->pp->is_stdcall;
7233 fp->regmask_dep = get_pp_arg_regmask_src(fp->pp);
7234 fp->has_ret = !IS(fp->pp->ret_type.name, "void");
7235 }
7236
7237 // resolve deps
7238 qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_name);
7239 for (i = 0; i < hg_fp_cnt; i++)
7240 hg_fp_resolve_deps(&hg_fp[i]);
7241
7242 // note: messes up .proto ptr, don't use
7243 //qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_id);
7244
7245 // output variables
7246 for (i = 0; i < hg_var_cnt; i++) {
7247 var = &hg_vars[i];
7248
7249 if (var->pp != NULL)
7250 // part of seed
7251 continue;
7252 else if (var->is_c_str)
7253 fprintf(fout, "extern %-8s %s[];", "char", var->name);
7254 else
7255 fprintf(fout, "extern %-8s %s;",
7256 lmod_c_names[var->lmod], var->name);
7257
7258 if (var->is_seeded)
7259 fprintf(fout, " // seeded");
7260 fprintf(fout, "\n");
7261 }
7262
7263 fprintf(fout, "\n");
7264
7265 // output function prototypes
7266 output_hdr_fp(fout, hg_fp, hg_fp_cnt);
7267
7268 // seed passthrough
7269 fprintf(fout, "\n// - seed -\n");
7270
7271 rewind(g_fhdr);
7272 while (fgets(line, sizeof(line), g_fhdr))
7273 fwrite(line, 1, strlen(line), fout);
7274}
7275
7276// '=' needs special treatment
7277// also ' quote
7278static char *next_word_s(char *w, size_t wsize, char *s)
7279{
7280 size_t i;
7281
7282 s = sskip(s);
7283
7284 i = 0;
7285 if (*s == '\'') {
7286 w[0] = s[0];
7287 for (i = 1; i < wsize - 1; i++) {
7288 if (s[i] == 0) {
7289 printf("warning: missing closing quote: \"%s\"\n", s);
7290 break;
7291 }
7292 if (s[i] == '\'')
7293 break;
7294 w[i] = s[i];
7295 }
7296 }
7297
7298 for (; i < wsize - 1; i++) {
7299 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
7300 break;
7301 w[i] = s[i];
7302 }
7303 w[i] = 0;
7304
7305 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
7306 printf("warning: '%s' truncated\n", w);
7307
7308 return s + i;
7309}
7310
7311static void scan_variables(FILE *fasm)
7312{
7313 struct scanned_var *var;
7314 char line[256] = { 0, };
7315 char words[3][256];
7316 char *p = NULL;
7317 int wordc;
7318 int l;
7319
7320 while (!feof(fasm))
7321 {
7322 // skip to next data section
7323 while (my_fgets(line, sizeof(line), fasm))
7324 {
7325 asmln++;
7326
7327 p = sskip(line);
7328 if (*p == 0 || *p == ';')
7329 continue;
7330
7331 p = sskip(next_word_s(words[0], sizeof(words[0]), p));
7332 if (*p == 0 || *p == ';')
7333 continue;
7334
7335 if (*p != 's' || !IS_START(p, "segment para public"))
7336 continue;
7337
7338 break;
7339 }
7340
7341 if (p == NULL || !IS_START(p, "segment para public"))
7342 break;
7343 p = sskip(p + 19);
7344
7345 if (!IS_START(p, "'DATA'"))
7346 continue;
7347
7348 // now process it
7349 while (my_fgets(line, sizeof(line), fasm))
7350 {
7351 asmln++;
7352
7353 p = line;
7354 if (my_isblank(*p))
7355 continue;
7356
7357 p = sskip(p);
7358 if (*p == 0 || *p == ';')
7359 continue;
7360
7361 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
7362 words[wordc][0] = 0;
7363 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
7364 if (*p == 0 || *p == ';') {
7365 wordc++;
7366 break;
7367 }
7368 }
7369
7370 if (wordc == 2 && IS(words[1], "ends"))
7371 break;
7372 if (wordc < 2)
7373 continue;
7374
7375 if (IS_START(words[0], "__IMPORT_DESCRIPTOR_")) {
7376 // when this starts, we don't need anything from this section
7377 break;
7378 }
7379
7380 if ((hg_var_cnt & 0xff) == 0) {
7381 hg_vars = realloc(hg_vars, sizeof(hg_vars[0])
7382 * (hg_var_cnt + 0x100));
7383 my_assert_not(hg_vars, NULL);
7384 memset(hg_vars + hg_var_cnt, 0, sizeof(hg_vars[0]) * 0x100);
7385 }
7386
7387 var = &hg_vars[hg_var_cnt++];
7388 snprintf(var->name, sizeof(var->name), "%s", words[0]);
7389
7390 // maybe already in seed header?
7391 var->pp = proto_parse(g_fhdr, var->name, 1);
7392 if (var->pp != NULL) {
7393 if (var->pp->is_fptr) {
7394 var->lmod = OPLM_DWORD;
7395 //var->is_ptr = 1;
7396 }
7397 else if (var->pp->is_func)
7398 aerr("func?\n");
7399 else if (!guess_lmod_from_c_type(&var->lmod, &var->pp->type))
7400 aerr("unhandled C type '%s' for '%s'\n",
7401 var->pp->type.name, var->name);
7402
7403 var->is_seeded = 1;
7404 continue;
7405 }
7406
7407 if (IS(words[1], "dd"))
7408 var->lmod = OPLM_DWORD;
7409 else if (IS(words[1], "dw"))
7410 var->lmod = OPLM_WORD;
7411 else if (IS(words[1], "db")) {
7412 var->lmod = OPLM_BYTE;
7413 if (wordc >= 3 && (l = strlen(words[2])) > 4) {
7414 if (words[2][0] == '\'' && IS(words[2] + l - 2, ",0"))
7415 var->is_c_str = 1;
7416 }
7417 }
7418 else if (IS(words[1], "dq"))
7419 var->lmod = OPLM_QWORD;
7420 //else if (IS(words[1], "dt"))
7421 else
7422 aerr("type '%s' not known\n", words[1]);
7423 }
7424 }
7425
7426 rewind(fasm);
7427 asmln = 0;
7428}
7429
7430static void set_label(int i, const char *name)
7431{
7432 const char *p;
7433 int len;
7434
7435 len = strlen(name);
7436 p = strchr(name, ':');
7437 if (p != NULL)
7438 len = p - name;
7439
7440 if (g_labels[i] != NULL && !IS_START(g_labels[i], "algn_"))
7441 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
7442 g_labels[i] = realloc(g_labels[i], len + 1);
7443 my_assert_not(g_labels[i], NULL);
7444 memcpy(g_labels[i], name, len);
7445 g_labels[i][len] = 0;
7446}
7447
7448struct chunk_item {
7449 char *name;
7450 long fptr;
7451 int asmln;
7452};
7453
7454static struct chunk_item *func_chunks;
7455static int func_chunk_cnt;
7456static int func_chunk_alloc;
7457
7458static void add_func_chunk(FILE *fasm, const char *name, int line)
7459{
7460 if (func_chunk_cnt >= func_chunk_alloc) {
7461 func_chunk_alloc *= 2;
7462 func_chunks = realloc(func_chunks,
7463 func_chunk_alloc * sizeof(func_chunks[0]));
7464 my_assert_not(func_chunks, NULL);
7465 }
7466 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
7467 func_chunks[func_chunk_cnt].name = strdup(name);
7468 func_chunks[func_chunk_cnt].asmln = line;
7469 func_chunk_cnt++;
7470}
7471
7472static int cmp_chunks(const void *p1, const void *p2)
7473{
7474 const struct chunk_item *c1 = p1, *c2 = p2;
7475 return strcmp(c1->name, c2->name);
7476}
7477
7478static int cmpstringp(const void *p1, const void *p2)
7479{
7480 return strcmp(*(char * const *)p1, *(char * const *)p2);
7481}
7482
7483static void scan_ahead(FILE *fasm)
7484{
7485 char words[2][256];
7486 char line[256];
7487 long oldpos;
7488 int oldasmln;
7489 int wordc;
7490 char *p;
7491 int i;
7492
7493 oldpos = ftell(fasm);
7494 oldasmln = asmln;
7495
7496 while (my_fgets(line, sizeof(line), fasm))
7497 {
7498 wordc = 0;
7499 asmln++;
7500
7501 p = sskip(line);
7502 if (*p == 0)
7503 continue;
7504
7505 if (*p == ';')
7506 {
7507 // get rid of random tabs
7508 for (i = 0; line[i] != 0; i++)
7509 if (line[i] == '\t')
7510 line[i] = ' ';
7511
7512 if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
7513 {
7514 p += 30;
7515 next_word(words[0], sizeof(words[0]), p);
7516 if (words[0][0] == 0)
7517 aerr("missing name for func chunk?\n");
7518
7519 add_func_chunk(fasm, words[0], asmln);
7520 }
7521 else if (IS_START(p, "; sctend"))
7522 break;
7523
7524 continue;
7525 } // *p == ';'
7526
7527 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
7528 words[wordc][0] = 0;
7529 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
7530 if (*p == 0 || *p == ';') {
7531 wordc++;
7532 break;
7533 }
7534 }
7535
7536 if (wordc == 2 && IS(words[1], "ends"))
7537 break;
7538 }
7539
7540 fseek(fasm, oldpos, SEEK_SET);
7541 asmln = oldasmln;
7542}
7543
7544int main(int argc, char *argv[])
7545{
7546 FILE *fout, *fasm, *frlist;
7547 struct parsed_data *pd = NULL;
7548 int pd_alloc = 0;
7549 char **rlist = NULL;
7550 int rlist_len = 0;
7551 int rlist_alloc = 0;
7552 int func_chunks_used = 0;
7553 int func_chunks_sorted = 0;
7554 int func_chunk_i = -1;
7555 long func_chunk_ret = 0;
7556 int func_chunk_ret_ln = 0;
7557 int scanned_ahead = 0;
7558 char line[256];
7559 char words[20][256];
7560 enum opr_lenmod lmod;
7561 char *sctproto = NULL;
7562 int in_func = 0;
7563 int pending_endp = 0;
7564 int skip_func = 0;
7565 int skip_warned = 0;
7566 int eq_alloc;
7567 int verbose = 0;
7568 int multi_seg = 0;
7569 int end = 0;
7570 int arg_out;
7571 int arg;
7572 int pi = 0;
7573 int i, j;
7574 int ret, len;
7575 char *p;
7576 int wordc;
7577
7578 for (arg = 1; arg < argc; arg++) {
7579 if (IS(argv[arg], "-v"))
7580 verbose = 1;
7581 else if (IS(argv[arg], "-rf"))
7582 g_allow_regfunc = 1;
7583 else if (IS(argv[arg], "-m"))
7584 multi_seg = 1;
7585 else if (IS(argv[arg], "-hdr"))
7586 g_header_mode = g_quiet_pp = g_allow_regfunc = 1;
7587 else
7588 break;
7589 }
7590
7591 if (argc < arg + 3) {
7592 printf("usage:\n%s [-v] [-rf] [-m] <.c> <.asm> <hdr.h> [rlist]*\n"
7593 "%s -hdr <out.h> <.asm> <seed.h> [rlist]*\n"
7594 "options:\n"
7595 " -hdr - header generation mode\n"
7596 " -rf - allow unannotated indirect calls\n"
7597 " -m - allow multiple .text sections\n"
7598 "[rlist] is a file with function names to skip,"
7599 " one per line\n",
7600 argv[0], argv[0]);
7601 return 1;
7602 }
7603
7604 arg_out = arg++;
7605
7606 asmfn = argv[arg++];
7607 fasm = fopen(asmfn, "r");
7608 my_assert_not(fasm, NULL);
7609
7610 hdrfn = argv[arg++];
7611 g_fhdr = fopen(hdrfn, "r");
7612 my_assert_not(g_fhdr, NULL);
7613
7614 rlist_alloc = 64;
7615 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
7616 my_assert_not(rlist, NULL);
7617 // needs special handling..
7618 rlist[rlist_len++] = "__alloca_probe";
7619
7620 func_chunk_alloc = 32;
7621 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
7622 my_assert_not(func_chunks, NULL);
7623
7624 memset(words, 0, sizeof(words));
7625
7626 for (; arg < argc; arg++) {
7627 frlist = fopen(argv[arg], "r");
7628 my_assert_not(frlist, NULL);
7629
7630 while (my_fgets(line, sizeof(line), frlist)) {
7631 p = sskip(line);
7632 if (*p == 0 || *p == ';')
7633 continue;
7634 if (*p == '#') {
7635 if (IS_START(p, "#if 0")
7636 || (g_allow_regfunc && IS_START(p, "#if NO_REGFUNC")))
7637 {
7638 skip_func = 1;
7639 }
7640 else if (IS_START(p, "#endif"))
7641 skip_func = 0;
7642 continue;
7643 }
7644 if (skip_func)
7645 continue;
7646
7647 p = next_word(words[0], sizeof(words[0]), p);
7648 if (words[0][0] == 0)
7649 continue;
7650
7651 if (rlist_len >= rlist_alloc) {
7652 rlist_alloc = rlist_alloc * 2 + 64;
7653 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
7654 my_assert_not(rlist, NULL);
7655 }
7656 rlist[rlist_len++] = strdup(words[0]);
7657 }
7658 skip_func = 0;
7659
7660 fclose(frlist);
7661 frlist = NULL;
7662 }
7663
7664 if (rlist_len > 0)
7665 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
7666
7667 fout = fopen(argv[arg_out], "w");
7668 my_assert_not(fout, NULL);
7669
7670 eq_alloc = 128;
7671 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
7672 my_assert_not(g_eqs, NULL);
7673
7674 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
7675 g_label_refs[i].i = -1;
7676 g_label_refs[i].next = NULL;
7677 }
7678
7679 if (g_header_mode)
7680 scan_variables(fasm);
7681
7682 while (my_fgets(line, sizeof(line), fasm))
7683 {
7684 wordc = 0;
7685 asmln++;
7686
7687 p = sskip(line);
7688 if (*p == 0)
7689 continue;
7690
7691 // get rid of random tabs
7692 for (i = 0; line[i] != 0; i++)
7693 if (line[i] == '\t')
7694 line[i] = ' ';
7695
7696 if (*p == ';')
7697 {
7698 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
7699 goto do_pending_endp; // eww..
7700
7701 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
7702 {
7703 static const char *attrs[] = {
7704 "bp-based frame",
7705 "library function",
7706 "static",
7707 "noreturn",
7708 "thunk",
7709 "fpd=",
7710 };
7711
7712 // parse IDA's attribute-list comment
7713 g_ida_func_attr = 0;
7714 p = sskip(p + 13);
7715
7716 for (; *p != 0; p = sskip(p)) {
7717 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
7718 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
7719 g_ida_func_attr |= 1 << i;
7720 p += strlen(attrs[i]);
7721 break;
7722 }
7723 }
7724 if (i == ARRAY_SIZE(attrs)) {
7725 anote("unparsed IDA attr: %s\n", p);
7726 break;
7727 }
7728 if (IS(attrs[i], "fpd=")) {
7729 p = next_word(words[0], sizeof(words[0]), p);
7730 // ignore for now..
7731 }
7732 }
7733 }
7734 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
7735 {
7736 p += 30;
7737 next_word(words[0], sizeof(words[0]), p);
7738 if (words[0][0] == 0)
7739 aerr("missing name for func chunk?\n");
7740
7741 if (!scanned_ahead) {
7742 add_func_chunk(fasm, words[0], asmln);
7743 func_chunks_sorted = 0;
7744 }
7745 }
7746 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
7747 {
7748 if (func_chunk_i >= 0) {
7749 if (func_chunk_i < func_chunk_cnt
7750 && IS(func_chunks[func_chunk_i].name, g_func))
7751 {
7752 // move on to next chunk
7753 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
7754 if (ret)
7755 aerr("seek failed for '%s' chunk #%d\n",
7756 g_func, func_chunk_i);
7757 asmln = func_chunks[func_chunk_i].asmln;
7758 func_chunk_i++;
7759 }
7760 else {
7761 if (func_chunk_ret == 0)
7762 aerr("no return from chunk?\n");
7763 fseek(fasm, func_chunk_ret, SEEK_SET);
7764 asmln = func_chunk_ret_ln;
7765 func_chunk_ret = 0;
7766 pending_endp = 1;
7767 }
7768 }
7769 }
7770 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
7771 func_chunks_used = 1;
7772 p += 20;
7773 if (IS_START(g_func, "sub_")) {
7774 unsigned long addr = strtoul(p, NULL, 16);
7775 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
7776 if (addr > f_addr && !scanned_ahead) {
7777 //anote("scan_ahead caused by '%s', addr %lx\n",
7778 // g_func, addr);
7779 scan_ahead(fasm);
7780 scanned_ahead = 1;
7781 func_chunks_sorted = 0;
7782 }
7783 }
7784 }
7785 continue;
7786 } // *p == ';'
7787
7788parse_words:
7789 for (i = wordc; i < ARRAY_SIZE(words); i++)
7790 words[i][0] = 0;
7791 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
7792 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
7793 if (*p == 0 || *p == ';') {
7794 wordc++;
7795 break;
7796 }
7797 }
7798 if (*p != 0 && *p != ';')
7799 aerr("too many words\n");
7800
7801 // alow asm patches in comments
7802 if (*p == ';') {
7803 if (IS_START(p, "; sctpatch:")) {
7804 p = sskip(p + 11);
7805 if (*p == 0 || *p == ';')
7806 continue;
7807 goto parse_words; // lame
7808 }
7809 if (IS_START(p, "; sctproto:")) {
7810 sctproto = strdup(p + 11);
7811 }
7812 else if (IS_START(p, "; sctend")) {
7813 end = 1;
7814 if (!pending_endp)
7815 break;
7816 }
7817 }
7818
7819 if (wordc == 0) {
7820 // shouldn't happen
7821 awarn("wordc == 0?\n");
7822 continue;
7823 }
7824
7825 // don't care about this:
7826 if (words[0][0] == '.'
7827 || IS(words[0], "include")
7828 || IS(words[0], "assume") || IS(words[1], "segment")
7829 || IS(words[0], "align"))
7830 {
7831 continue;
7832 }
7833
7834do_pending_endp:
7835 // do delayed endp processing to collect switch jumptables
7836 if (pending_endp) {
7837 if (in_func && !g_skip_func && !end && wordc >= 2
7838 && ((words[0][0] == 'd' && words[0][2] == 0)
7839 || (words[1][0] == 'd' && words[1][2] == 0)))
7840 {
7841 i = 1;
7842 if (words[1][0] == 'd' && words[1][2] == 0) {
7843 // label
7844 if (g_func_pd_cnt >= pd_alloc) {
7845 pd_alloc = pd_alloc * 2 + 16;
7846 g_func_pd = realloc(g_func_pd,
7847 sizeof(g_func_pd[0]) * pd_alloc);
7848 my_assert_not(g_func_pd, NULL);
7849 }
7850 pd = &g_func_pd[g_func_pd_cnt];
7851 g_func_pd_cnt++;
7852 memset(pd, 0, sizeof(*pd));
7853 strcpy(pd->label, words[0]);
7854 pd->type = OPT_CONST;
7855 pd->lmod = lmod_from_directive(words[1]);
7856 i = 2;
7857 }
7858 else {
7859 if (pd == NULL) {
7860 if (verbose)
7861 anote("skipping alignment byte?\n");
7862 continue;
7863 }
7864 lmod = lmod_from_directive(words[0]);
7865 if (lmod != pd->lmod)
7866 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
7867 }
7868
7869 if (pd->count_alloc < pd->count + wordc) {
7870 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
7871 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
7872 my_assert_not(pd->d, NULL);
7873 }
7874 for (; i < wordc; i++) {
7875 if (IS(words[i], "offset")) {
7876 pd->type = OPT_OFFSET;
7877 i++;
7878 }
7879 p = strchr(words[i], ',');
7880 if (p != NULL)
7881 *p = 0;
7882 if (pd->type == OPT_OFFSET)
7883 pd->d[pd->count].u.label = strdup(words[i]);
7884 else
7885 pd->d[pd->count].u.val = parse_number(words[i]);
7886 pd->d[pd->count].bt_i = -1;
7887 pd->count++;
7888 }
7889 continue;
7890 }
7891
7892 if (in_func && !g_skip_func) {
7893 if (g_header_mode)
7894 gen_hdr(g_func, pi);
7895 else
7896 gen_func(fout, g_fhdr, g_func, pi);
7897 }
7898
7899 pending_endp = 0;
7900 in_func = 0;
7901 g_ida_func_attr = 0;
7902 skip_warned = 0;
7903 g_skip_func = 0;
7904 g_func[0] = 0;
7905 func_chunks_used = 0;
7906 func_chunk_i = -1;
7907 if (pi != 0) {
7908 memset(&ops, 0, pi * sizeof(ops[0]));
7909 clear_labels(pi);
7910 pi = 0;
7911 }
7912 g_eqcnt = 0;
7913 for (i = 0; i < g_func_pd_cnt; i++) {
7914 pd = &g_func_pd[i];
7915 if (pd->type == OPT_OFFSET) {
7916 for (j = 0; j < pd->count; j++)
7917 free(pd->d[j].u.label);
7918 }
7919 free(pd->d);
7920 pd->d = NULL;
7921 }
7922 g_func_pd_cnt = 0;
7923 g_func_lmods = 0;
7924 pd = NULL;
7925
7926 if (end)
7927 break;
7928 if (wordc == 0)
7929 continue;
7930 }
7931
7932 if (IS(words[1], "proc")) {
7933 if (in_func)
7934 aerr("proc '%s' while in_func '%s'?\n",
7935 words[0], g_func);
7936 p = words[0];
7937 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
7938 g_skip_func = 1;
7939 strcpy(g_func, words[0]);
7940 set_label(0, words[0]);
7941 in_func = 1;
7942 continue;
7943 }
7944
7945 if (IS(words[1], "endp"))
7946 {
7947 if (!in_func)
7948 aerr("endp '%s' while not in_func?\n", words[0]);
7949 if (!IS(g_func, words[0]))
7950 aerr("endp '%s' while in_func '%s'?\n",
7951 words[0], g_func);
7952
7953 if ((g_ida_func_attr & IDAFA_THUNK) && pi == 1
7954 && ops[0].op == OP_JMP && ops[0].operand[0].had_ds)
7955 {
7956 // import jump
7957 g_skip_func = 1;
7958 }
7959
7960 if (!g_skip_func && func_chunks_used) {
7961 // start processing chunks
7962 struct chunk_item *ci, key = { g_func, 0 };
7963
7964 func_chunk_ret = ftell(fasm);
7965 func_chunk_ret_ln = asmln;
7966 if (!func_chunks_sorted) {
7967 qsort(func_chunks, func_chunk_cnt,
7968 sizeof(func_chunks[0]), cmp_chunks);
7969 func_chunks_sorted = 1;
7970 }
7971 ci = bsearch(&key, func_chunks, func_chunk_cnt,
7972 sizeof(func_chunks[0]), cmp_chunks);
7973 if (ci == NULL)
7974 aerr("'%s' needs chunks, but none found\n", g_func);
7975 func_chunk_i = ci - func_chunks;
7976 for (; func_chunk_i > 0; func_chunk_i--)
7977 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
7978 break;
7979
7980 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
7981 if (ret)
7982 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
7983 asmln = func_chunks[func_chunk_i].asmln;
7984 func_chunk_i++;
7985 continue;
7986 }
7987 pending_endp = 1;
7988 continue;
7989 }
7990
7991 if (wordc == 2 && IS(words[1], "ends")) {
7992 if (!multi_seg) {
7993 end = 1;
7994 if (pending_endp)
7995 goto do_pending_endp;
7996 break;
7997 }
7998
7999 // scan for next text segment
8000 while (my_fgets(line, sizeof(line), fasm)) {
8001 asmln++;
8002 p = sskip(line);
8003 if (*p == 0 || *p == ';')
8004 continue;
8005
8006 if (strstr(p, "segment para public 'CODE' use32"))
8007 break;
8008 }
8009
8010 continue;
8011 }
8012
8013 p = strchr(words[0], ':');
8014 if (p != NULL) {
8015 set_label(pi, words[0]);
8016 continue;
8017 }
8018
8019 if (!in_func || g_skip_func) {
8020 if (!skip_warned && !g_skip_func && g_labels[pi] != NULL) {
8021 if (verbose)
8022 anote("skipping from '%s'\n", g_labels[pi]);
8023 skip_warned = 1;
8024 }
8025 free(g_labels[pi]);
8026 g_labels[pi] = NULL;
8027 continue;
8028 }
8029
8030 if (wordc > 1 && IS(words[1], "="))
8031 {
8032 if (wordc != 5)
8033 aerr("unhandled equ, wc=%d\n", wordc);
8034 if (g_eqcnt >= eq_alloc) {
8035 eq_alloc *= 2;
8036 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
8037 my_assert_not(g_eqs, NULL);
8038 }
8039
8040 len = strlen(words[0]);
8041 if (len > sizeof(g_eqs[0].name) - 1)
8042 aerr("equ name too long: %d\n", len);
8043 strcpy(g_eqs[g_eqcnt].name, words[0]);
8044
8045 if (!IS(words[3], "ptr"))
8046 aerr("unhandled equ\n");
8047 if (IS(words[2], "dword"))
8048 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
8049 else if (IS(words[2], "word"))
8050 g_eqs[g_eqcnt].lmod = OPLM_WORD;
8051 else if (IS(words[2], "byte"))
8052 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
8053 else if (IS(words[2], "qword"))
8054 g_eqs[g_eqcnt].lmod = OPLM_QWORD;
8055 else
8056 aerr("bad lmod: '%s'\n", words[2]);
8057
8058 g_eqs[g_eqcnt].offset = parse_number(words[4]);
8059 g_eqcnt++;
8060 continue;
8061 }
8062
8063 if (pi >= ARRAY_SIZE(ops))
8064 aerr("too many ops\n");
8065
8066 parse_op(&ops[pi], words, wordc);
8067
8068 if (sctproto != NULL) {
8069 if (ops[pi].op == OP_CALL || ops[pi].op == OP_JMP)
8070 ops[pi].datap = sctproto;
8071 sctproto = NULL;
8072 }
8073 pi++;
8074 }
8075
8076 if (g_header_mode)
8077 output_hdr(fout);
8078
8079 fclose(fout);
8080 fclose(fasm);
8081 fclose(g_fhdr);
8082
8083 return 0;
8084}
8085
8086// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.