notaz.gp2x.de / ia32rtools.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
fix tests
[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// on start: function/data type hint (sctproto)
212// after analysis:
213// (OPF_CC) - points to one of (OPF_FLAGS) that affects cc op
214// OP_PUSH - points to OP_POP in complex push/pop graph
215// OP_POP - points to OP_PUSH in simple push/pop pair
216
217struct parsed_equ {
218 char name[64];
219 enum opr_lenmod lmod;
220 int offset;
221};
222
223struct parsed_data {
224 char label[256];
225 enum opr_type type;
226 enum opr_lenmod lmod;
227 int count;
228 int count_alloc;
229 struct {
230 union {
231 char *label;
232 unsigned int val;
233 } u;
234 int bt_i;
235 } *d;
236};
237
238struct label_ref {
239 int i;
240 struct label_ref *next;
241};
242
243enum ida_func_attr {
244 IDAFA_BP_FRAME = (1 << 0),
245 IDAFA_LIB_FUNC = (1 << 1),
246 IDAFA_STATIC = (1 << 2),
247 IDAFA_NORETURN = (1 << 3),
248 IDAFA_THUNK = (1 << 4),
249 IDAFA_FPD = (1 << 5),
250};
251
252enum sct_func_attr {
253 SCTFA_CLEAR_SF = (1 << 0), // clear stack frame
254 SCTFA_CLEAR_REGS = (1 << 1), // clear registers (mask)
255};
256
257enum x87_const {
258 X87_CONST_1 = 1,
259 X87_CONST_2T,
260 X87_CONST_2E,
261 X87_CONST_PI,
262 X87_CONST_LG2,
263 X87_CONST_LN2,
264 X87_CONST_Z,
265};
266
267// note: limited to 32k due to p_argnext
268#define MAX_OPS 4096
269#define MAX_ARG_GRP 2
270
271static struct parsed_op ops[MAX_OPS];
272static struct parsed_equ *g_eqs;
273static int g_eqcnt;
274static char *g_labels[MAX_OPS];
275static struct label_ref g_label_refs[MAX_OPS];
276static const struct parsed_proto *g_func_pp;
277static struct parsed_data *g_func_pd;
278static int g_func_pd_cnt;
279static int g_func_lmods;
280static char g_func[256];
281static char g_comment[256];
282static int g_bp_frame;
283static int g_sp_frame;
284static int g_stack_frame_used;
285static int g_stack_fsz;
286static int g_ida_func_attr;
287static int g_sct_func_attr;
288static int g_stack_clear_start; // in dwords
289static int g_stack_clear_len;
290static int g_regmask_init;
291static int g_skip_func;
292static int g_allow_regfunc;
293static int g_quiet_pp;
294static int g_header_mode;
295
296#define ferr(op_, fmt, ...) do { \
297 printf("%s:%d: error %u: [%s] '%s': " fmt, asmfn, (op_)->asmln, \
298 __LINE__, g_func, dump_op(op_), ##__VA_ARGS__); \
299 fcloseall(); \
300 exit(1); \
301} while (0)
302#define fnote(op_, fmt, ...) \
303 printf("%s:%d: note: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
304 dump_op(op_), ##__VA_ARGS__)
305
306#define ferr_assert(op_, cond) do { \
307 if (!(cond)) ferr(op_, "assertion '%s' failed\n", #cond); \
308} while (0)
309
310const char *regs_r32[] = {
311 "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp",
312 // not r32, but list here for easy parsing and printing
313 "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7",
314 "st", "st(1)", "st(2)", "st(3)", "st(4)", "st(5)", "st(6)", "st(7)"
315};
316const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" };
317const char *regs_r8l[] = { "al", "bl", "cl", "dl" };
318const char *regs_r8h[] = { "ah", "bh", "ch", "dh" };
319
320enum x86_regs {
321 xUNSPEC = -1,
322 xAX, xBX, xCX, xDX,
323 xSI, xDI, xBP, xSP,
324 xMM0, xMM1, xMM2, xMM3, // mmx
325 xMM4, xMM5, xMM6, xMM7,
326 xST0, xST1, xST2, xST3, // x87
327 xST4, xST5, xST6, xST7,
328};
329
330#define mxAX (1 << xAX)
331#define mxCX (1 << xCX)
332#define mxDX (1 << xDX)
333#define mxST0 (1 << xST0)
334#define mxST1 (1 << xST1)
335
336// possible basic comparison types (without inversion)
337enum parsed_flag_op {
338 PFO_O, // 0 OF=1
339 PFO_C, // 2 CF=1
340 PFO_Z, // 4 ZF=1
341 PFO_BE, // 6 CF=1||ZF=1
342 PFO_S, // 8 SF=1
343 PFO_P, // a PF=1
344 PFO_L, // c SF!=OF
345 PFO_LE, // e ZF=1||SF!=OF
346};
347
348#define PFOB_O (1 << PFO_O)
349#define PFOB_C (1 << PFO_C)
350#define PFOB_Z (1 << PFO_Z)
351#define PFOB_S (1 << PFO_S)
352
353static const char *parsed_flag_op_names[] = {
354 "o", "c", "z", "be", "s", "p", "l", "le"
355};
356
357static int char_array_i(const char *array[], size_t len, const char *s)
358{
359 int i;
360
361 for (i = 0; i < len; i++)
362 if (IS(s, array[i]))
363 return i;
364
365 return -1;
366}
367
368static void printf_number(char *buf, size_t buf_size,
369 unsigned long number)
370{
371 // output in C-friendly form
372 snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
373}
374
375static int check_segment_prefix(const char *s)
376{
377 if (s[0] == 0 || s[1] != 's' || s[2] != ':')
378 return 0;
379
380 switch (s[0]) {
381 case 'c': return 1;
382 case 'd': return 2;
383 case 's': return 3;
384 case 'e': return 4;
385 case 'f': return 5;
386 case 'g': return 6;
387 default: return 0;
388 }
389}
390
391static int parse_reg(enum opr_lenmod *reg_lmod, const char *s)
392{
393 int reg;
394
395 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s);
396 if (reg >= 8) {
397 *reg_lmod = OPLM_QWORD;
398 return reg;
399 }
400 if (reg >= 0) {
401 *reg_lmod = OPLM_DWORD;
402 return reg;
403 }
404 reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s);
405 if (reg >= 0) {
406 *reg_lmod = OPLM_WORD;
407 return reg;
408 }
409 reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s);
410 if (reg >= 0) {
411 *reg_lmod = OPLM_BYTE;
412 return reg;
413 }
414 reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s);
415 if (reg >= 0) {
416 *reg_lmod = OPLM_BYTE;
417 return reg;
418 }
419
420 return -1;
421}
422
423static int parse_indmode(char *name, int *regmask, int need_c_cvt)
424{
425 enum opr_lenmod lmod;
426 char cvtbuf[256];
427 char *d = cvtbuf;
428 char *s = name;
429 char w[64];
430 long number;
431 int reg;
432 int c = 0;
433
434 *d = 0;
435
436 while (*s != 0) {
437 d += strlen(d);
438 while (my_isblank(*s))
439 s++;
440 for (; my_issep(*s); d++, s++)
441 *d = *s;
442 while (my_isblank(*s))
443 s++;
444 *d = 0;
445
446 // skip '?s:' prefixes
447 if (check_segment_prefix(s))
448 s += 3;
449
450 s = next_idt(w, sizeof(w), s);
451 if (w[0] == 0)
452 break;
453 c++;
454
455 reg = parse_reg(&lmod, w);
456 if (reg >= 0) {
457 *regmask |= 1 << reg;
458 goto pass;
459 }
460
461 if ('0' <= w[0] && w[0] <= '9') {
462 number = parse_number(w);
463 printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number);
464 continue;
465 }
466
467 // probably some label/identifier - pass
468
469pass:
470 snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w);
471 }
472
473 if (need_c_cvt)
474 strcpy(name, cvtbuf);
475
476 return c;
477}
478
479static int is_reg_in_str(const char *s)
480{
481 int i;
482
483 if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
484 return 0;
485
486 for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
487 if (!strncmp(s, regs_r32[i], 3))
488 return 1;
489
490 return 0;
491}
492
493static const char *parse_stack_el(const char *name, char *extra_reg,
494 int early_try)
495{
496 const char *p, *p2, *s;
497 char *endp = NULL;
498 char buf[32];
499 long val;
500 int len;
501
502 if (g_bp_frame || early_try)
503 {
504 p = name;
505 if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
506 p += 4;
507 if (extra_reg != NULL) {
508 strncpy(extra_reg, name, 3);
509 extra_reg[4] = 0;
510 }
511 }
512
513 if (IS_START(p, "ebp+")) {
514 p += 4;
515
516 p2 = strchr(p, '+');
517 if (p2 != NULL && is_reg_in_str(p)) {
518 if (extra_reg != NULL) {
519 strncpy(extra_reg, p, p2 - p);
520 extra_reg[p2 - p] = 0;
521 }
522 p = p2 + 1;
523 }
524
525 if (!('0' <= *p && *p <= '9'))
526 return p;
527
528 return NULL;
529 }
530 }
531
532 if (!IS_START(name, "esp+"))
533 return NULL;
534
535 s = name + 4;
536 p = strchr(s, '+');
537 if (p) {
538 if (is_reg_in_str(s)) {
539 if (extra_reg != NULL) {
540 strncpy(extra_reg, s, p - s);
541 extra_reg[p - s] = 0;
542 }
543 s = p + 1;
544 p = strchr(s, '+');
545 if (p == NULL)
546 aerr("%s IDA stackvar not set?\n", __func__);
547 }
548 if (!('0' <= *s && *s <= '9')) {
549 aerr("%s IDA stackvar offset not set?\n", __func__);
550 return NULL;
551 }
552 if (s[0] == '0' && s[1] == 'x')
553 s += 2;
554 len = p - s;
555 if (len < sizeof(buf) - 1) {
556 strncpy(buf, s, len);
557 buf[len] = 0;
558 val = strtol(buf, &endp, 16);
559 if (val == 0 || *endp != 0) {
560 aerr("%s num parse fail for '%s'\n", __func__, buf);
561 return NULL;
562 }
563 }
564 p++;
565 }
566 else
567 p = name + 4;
568
569 if ('0' <= *p && *p <= '9')
570 return NULL;
571
572 return p;
573}
574
575static int guess_lmod_from_name(struct parsed_opr *opr)
576{
577 if (IS_START(opr->name, "dword_") || IS_START(opr->name, "off_")) {
578 opr->lmod = OPLM_DWORD;
579 return 1;
580 }
581 if (IS_START(opr->name, "word_")) {
582 opr->lmod = OPLM_WORD;
583 return 1;
584 }
585 if (IS_START(opr->name, "byte_")) {
586 opr->lmod = OPLM_BYTE;
587 return 1;
588 }
589 if (IS_START(opr->name, "qword_")) {
590 opr->lmod = OPLM_QWORD;
591 return 1;
592 }
593 return 0;
594}
595
596static int guess_lmod_from_c_type(enum opr_lenmod *lmod,
597 const struct parsed_type *c_type)
598{
599 static const char *dword_types[] = {
600 "uint32_t", "int", "_DWORD", "UINT_PTR", "DWORD",
601 "WPARAM", "LPARAM", "UINT", "__int32",
602 "LONG", "HIMC", "BOOL", "size_t",
603 "float",
604 };
605 static const char *word_types[] = {
606 "uint16_t", "int16_t", "_WORD", "WORD",
607 "unsigned __int16", "__int16",
608 };
609 static const char *byte_types[] = {
610 "uint8_t", "int8_t", "char",
611 "unsigned __int8", "__int8", "BYTE", "_BYTE",
612 "CHAR", "_UNKNOWN",
613 // structures.. deal the same as with _UNKNOWN for now
614 "CRITICAL_SECTION",
615 };
616 const char *n;
617 int i;
618
619 if (c_type->is_ptr) {
620 *lmod = OPLM_DWORD;
621 return 1;
622 }
623
624 n = skip_type_mod(c_type->name);
625
626 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
627 if (IS(n, dword_types[i])) {
628 *lmod = OPLM_DWORD;
629 return 1;
630 }
631 }
632
633 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
634 if (IS(n, word_types[i])) {
635 *lmod = OPLM_WORD;
636 return 1;
637 }
638 }
639
640 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
641 if (IS(n, byte_types[i])) {
642 *lmod = OPLM_BYTE;
643 return 1;
644 }
645 }
646
647 return 0;
648}
649
650static char *default_cast_to(char *buf, size_t buf_size,
651 struct parsed_opr *opr)
652{
653 buf[0] = 0;
654
655 if (!opr->is_ptr || strchr(opr->name, '['))
656 return buf;
657 if (opr->pp == NULL || opr->pp->type.name == NULL
658 || opr->pp->is_fptr)
659 {
660 snprintf(buf, buf_size, "%s", "(void *)");
661 return buf;
662 }
663
664 snprintf(buf, buf_size, "(%s)", opr->pp->type.name);
665 return buf;
666}
667
668static enum opr_type lmod_from_directive(const char *d)
669{
670 if (IS(d, "dd"))
671 return OPLM_DWORD;
672 else if (IS(d, "dw"))
673 return OPLM_WORD;
674 else if (IS(d, "db"))
675 return OPLM_BYTE;
676
677 aerr("unhandled directive: '%s'\n", d);
678 return OPLM_UNSPEC;
679}
680
681static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
682 int *regmask)
683{
684 opr->type = OPT_REG;
685 opr->reg = reg;
686 opr->lmod = lmod;
687 *regmask |= 1 << reg;
688}
689
690static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
691 int *extra_offs);
692
693static int parse_operand(struct parsed_opr *opr,
694 int *regmask, int *regmask_indirect,
695 char words[16][256], int wordc, int w, unsigned int op_flags)
696{
697 const struct parsed_proto *pp = NULL;
698 enum opr_lenmod tmplmod;
699 unsigned long number;
700 char buf[256];
701 int ret, len;
702 int wordc_in;
703 char *p;
704 int i;
705
706 if (w >= wordc)
707 aerr("parse_operand w %d, wordc %d\n", w, wordc);
708
709 opr->reg = xUNSPEC;
710
711 for (i = w; i < wordc; i++) {
712 len = strlen(words[i]);
713 if (words[i][len - 1] == ',') {
714 words[i][len - 1] = 0;
715 wordc = i + 1;
716 break;
717 }
718 }
719
720 wordc_in = wordc - w;
721
722 if ((op_flags & OPF_JMP) && wordc_in > 0
723 && !('0' <= words[w][0] && words[w][0] <= '9'))
724 {
725 const char *label = NULL;
726
727 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
728 && IS(words[w + 1], "ptr"))
729 label = words[w + 2];
730 else if (wordc_in == 2 && IS(words[w], "short"))
731 label = words[w + 1];
732 else if (wordc_in == 1
733 && strchr(words[w], '[') == NULL
734 && parse_reg(&tmplmod, words[w]) < 0)
735 label = words[w];
736
737 if (label != NULL) {
738 opr->type = OPT_LABEL;
739 ret = check_segment_prefix(label);
740 if (ret != 0) {
741 if (ret >= 5)
742 aerr("fs/gs used\n");
743 opr->had_ds = 1;
744 label += 3;
745 }
746 strcpy(opr->name, label);
747 return wordc;
748 }
749 }
750
751 if (wordc_in >= 3) {
752 if (IS(words[w + 1], "ptr")) {
753 if (IS(words[w], "dword"))
754 opr->lmod = OPLM_DWORD;
755 else if (IS(words[w], "word"))
756 opr->lmod = OPLM_WORD;
757 else if (IS(words[w], "byte"))
758 opr->lmod = OPLM_BYTE;
759 else if (IS(words[w], "qword"))
760 opr->lmod = OPLM_QWORD;
761 else
762 aerr("type parsing failed\n");
763 w += 2;
764 wordc_in = wordc - w;
765 }
766 }
767
768 if (wordc_in == 2) {
769 if (IS(words[w], "offset")) {
770 opr->type = OPT_OFFSET;
771 opr->lmod = OPLM_DWORD;
772 strcpy(opr->name, words[w + 1]);
773 pp = proto_parse(g_fhdr, opr->name, 1);
774 goto do_label;
775 }
776 if (IS(words[w], "(offset")) {
777 p = strchr(words[w + 1], ')');
778 if (p == NULL)
779 aerr("parse of bracketed offset failed\n");
780 *p = 0;
781 opr->type = OPT_OFFSET;
782 strcpy(opr->name, words[w + 1]);
783 return wordc;
784 }
785 }
786
787 if (wordc_in != 1)
788 aerr("parse_operand 1 word expected\n");
789
790 ret = check_segment_prefix(words[w]);
791 if (ret != 0) {
792 if (ret >= 5)
793 aerr("fs/gs used\n");
794 opr->had_ds = 1;
795 memmove(words[w], words[w] + 3, strlen(words[w]) - 2);
796 }
797 strcpy(opr->name, words[w]);
798
799 if (words[w][0] == '[') {
800 opr->type = OPT_REGMEM;
801 ret = sscanf(words[w], "[%[^]]]", opr->name);
802 if (ret != 1)
803 aerr("[] parse failure\n");
804
805 parse_indmode(opr->name, regmask_indirect, 1);
806 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL, 1))
807 {
808 // might be an equ
809 struct parsed_equ *eq =
810 equ_find(NULL, parse_stack_el(opr->name, NULL, 1), &i);
811 if (eq)
812 opr->lmod = eq->lmod;
813
814 // might be unaligned access
815 g_func_lmods |= 1 << OPLM_BYTE;
816 }
817 return wordc;
818 }
819 else if (strchr(words[w], '[')) {
820 // label[reg] form
821 p = strchr(words[w], '[');
822 opr->type = OPT_REGMEM;
823 parse_indmode(p, regmask_indirect, 0);
824 strncpy(buf, words[w], p - words[w]);
825 buf[p - words[w]] = 0;
826 pp = proto_parse(g_fhdr, buf, 1);
827 goto do_label;
828 }
829 else if (('0' <= words[w][0] && words[w][0] <= '9')
830 || words[w][0] == '-')
831 {
832 number = parse_number(words[w]);
833 opr->type = OPT_CONST;
834 opr->val = number;
835 printf_number(opr->name, sizeof(opr->name), number);
836 return wordc;
837 }
838
839 ret = parse_reg(&tmplmod, opr->name);
840 if (ret >= 0) {
841 setup_reg_opr(opr, ret, tmplmod, regmask);
842 return wordc;
843 }
844
845 // most likely var in data segment
846 opr->type = OPT_LABEL;
847 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
848
849do_label:
850 if (pp != NULL) {
851 if (pp->is_fptr || pp->is_func) {
852 opr->lmod = OPLM_DWORD;
853 opr->is_ptr = 1;
854 }
855 else {
856 tmplmod = OPLM_UNSPEC;
857 if (!guess_lmod_from_c_type(&tmplmod, &pp->type))
858 anote("unhandled C type '%s' for '%s'\n",
859 pp->type.name, opr->name);
860
861 if (opr->lmod == OPLM_UNSPEC) {
862 opr->lmod = tmplmod;
863 opr->type_from_var = 1;
864 }
865 else if (opr->lmod != tmplmod) {
866 opr->size_mismatch = 1;
867 if (tmplmod < opr->lmod)
868 opr->size_lt = 1;
869 }
870 opr->is_ptr = pp->type.is_ptr;
871 }
872 opr->is_array = pp->type.is_array;
873 }
874 opr->pp = pp;
875
876 if (opr->lmod == OPLM_UNSPEC)
877 guess_lmod_from_name(opr);
878 return wordc;
879}
880
881static const struct {
882 const char *name;
883 unsigned int flags;
884} pref_table[] = {
885 { "rep", OPF_REP },
886 { "repe", OPF_REP|OPF_REPZ },
887 { "repz", OPF_REP|OPF_REPZ },
888 { "repne", OPF_REP|OPF_REPNZ },
889 { "repnz", OPF_REP|OPF_REPNZ },
890 { "lock", OPF_LOCK }, // ignored for now..
891};
892
893#define OPF_CJMP_CC (OPF_JMP|OPF_CJMP|OPF_CC)
894
895static const struct {
896 const char *name;
897 enum op_op op;
898 unsigned short minopr;
899 unsigned short maxopr;
900 unsigned int flags;
901 unsigned char pfo;
902 unsigned char pfo_inv;
903} op_table[] = {
904 { "nop", OP_NOP, 0, 0, 0 },
905 { "push", OP_PUSH, 1, 1, 0 },
906 { "pop", OP_POP, 1, 1, OPF_DATA },
907 { "leave",OP_LEAVE, 0, 0, OPF_DATA },
908 { "mov" , OP_MOV, 2, 2, OPF_DATA },
909 { "lea", OP_LEA, 2, 2, OPF_DATA },
910 { "movzx",OP_MOVZX, 2, 2, OPF_DATA },
911 { "movsx",OP_MOVSX, 2, 2, OPF_DATA },
912 { "xchg", OP_XCHG, 2, 2, OPF_DATA },
913 { "not", OP_NOT, 1, 1, OPF_DATA },
914 { "xlat", OP_XLAT, 0, 0, OPF_DATA },
915 { "cdq", OP_CDQ, 0, 0, OPF_DATA },
916 { "lodsb",OP_LODS, 0, 0, OPF_DATA },
917 { "lodsw",OP_LODS, 0, 0, OPF_DATA },
918 { "lodsd",OP_LODS, 0, 0, OPF_DATA },
919 { "stosb",OP_STOS, 0, 0, OPF_DATA },
920 { "stosw",OP_STOS, 0, 0, OPF_DATA },
921 { "stosd",OP_STOS, 0, 0, OPF_DATA },
922 { "movsb",OP_MOVS, 0, 0, OPF_DATA },
923 { "movsw",OP_MOVS, 0, 0, OPF_DATA },
924 { "movsd",OP_MOVS, 0, 0, OPF_DATA },
925 { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
926 { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
927 { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
928 { "scasb",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
929 { "scasw",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
930 { "scasd",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
931 { "std", OP_STD, 0, 0, OPF_DATA }, // special flag
932 { "cld", OP_CLD, 0, 0, OPF_DATA },
933 { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
934 { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
935 { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
936 { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS },
937 { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS },
938 { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
939 { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
940 { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
941 { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
942 { "shld", OP_SHLD, 3, 3, OPF_DATA|OPF_FLAGS },
943 { "shrd", OP_SHRD, 3, 3, OPF_DATA|OPF_FLAGS },
944 { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
945 { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
946 { "rcl", OP_RCL, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
947 { "rcr", OP_RCR, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
948 { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
949 { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
950 { "bsf", OP_BSF, 2, 2, OPF_DATA|OPF_FLAGS },
951 { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
952 { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS },
953 { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS },
954 { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS },
955 { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS },
956 { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS },
957 { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS },
958 { "test", OP_TEST, 2, 2, OPF_FLAGS },
959 { "cmp", OP_CMP, 2, 2, OPF_FLAGS },
960 { "retn", OP_RET, 0, 1, OPF_TAIL },
961 { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS },
962 { "jmp", OP_JMP, 1, 1, OPF_JMP },
963 { "jecxz",OP_JECXZ, 1, 1, OPF_JMP|OPF_CJMP },
964 { "loop", OP_LOOP, 1, 1, OPF_JMP|OPF_CJMP|OPF_DATA },
965 { "jo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 0 }, // 70 OF=1
966 { "jno", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 1 }, // 71 OF=0
967 { "jc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72 CF=1
968 { "jb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72
969 { "jnc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73 CF=0
970 { "jnb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
971 { "jae", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
972 { "jz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74 ZF=1
973 { "je", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74
974 { "jnz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75 ZF=0
975 { "jne", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75
976 { "jbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76 CF=1||ZF=1
977 { "jna", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76
978 { "ja", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77 CF=0&&ZF=0
979 { "jnbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77
980 { "js", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 0 }, // 78 SF=1
981 { "jns", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 1 }, // 79 SF=0
982 { "jp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a PF=1
983 { "jpe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a
984 { "jnp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b PF=0
985 { "jpo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b
986 { "jl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c SF!=OF
987 { "jnge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c
988 { "jge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d SF=OF
989 { "jnl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d
990 { "jle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e ZF=1||SF!=OF
991 { "jng", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e
992 { "jg", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f ZF=0&&SF=OF
993 { "jnle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f
994 { "seto", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 0 },
995 { "setno", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 1 },
996 { "setc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
997 { "setb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
998 { "setnc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
999 { "setae", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
1000 { "setnb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
1001 { "setz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
1002 { "sete", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
1003 { "setnz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
1004 { "setne", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
1005 { "setbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
1006 { "setna", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
1007 { "seta", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
1008 { "setnbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
1009 { "sets", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 0 },
1010 { "setns", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 1 },
1011 { "setp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
1012 { "setpe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
1013 { "setnp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
1014 { "setpo", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
1015 { "setl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
1016 { "setnge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
1017 { "setge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
1018 { "setnl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
1019 { "setle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
1020 { "setng", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
1021 { "setg", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
1022 { "setnle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
1023 // x87
1024 { "fld", OP_FLD, 1, 1, OPF_FPUSH },
1025 { "fild", OP_FILD, 1, 1, OPF_FPUSH },
1026 { "fld1", OP_FLDc, 0, 0, OPF_FPUSH },
1027 { "fldz", OP_FLDc, 0, 0, OPF_FPUSH },
1028 { "fstp", OP_FST, 1, 1, OPF_FPOP },
1029 { "fst", OP_FST, 1, 1, 0 },
1030 { "fadd", OP_FADD, 0, 2, 0 },
1031 { "faddp", OP_FADD, 0, 2, OPF_FPOP },
1032 { "fdiv", OP_FDIV, 0, 2, 0 },
1033 { "fdivp", OP_FDIV, 0, 2, OPF_FPOP },
1034 { "fmul", OP_FMUL, 0, 2, 0 },
1035 { "fmulp", OP_FMUL, 0, 2, OPF_FPOP },
1036 { "fsub", OP_FSUB, 0, 2, 0 },
1037 { "fsubp", OP_FSUB, 0, 2, OPF_FPOP },
1038 { "fdivr", OP_FDIVR, 0, 2, 0 },
1039 { "fdivrp", OP_FDIVR, 0, 2, OPF_FPOP },
1040 { "fsubr", OP_FSUBR, 0, 2, 0 },
1041 { "fsubrp", OP_FSUBR, 0, 2, OPF_FPOP },
1042 { "fiadd", OP_FIADD, 1, 1, 0 },
1043 { "fidiv", OP_FIDIV, 1, 1, 0 },
1044 { "fimul", OP_FIMUL, 1, 1, 0 },
1045 { "fisub", OP_FISUB, 1, 1, 0 },
1046 { "fidivr", OP_FIDIVR, 1, 1, 0 },
1047 { "fisubr", OP_FISUBR, 1, 1, 0 },
1048 // mmx
1049 { "emms", OP_EMMS, 0, 0, OPF_DATA },
1050 { "movq", OP_MOV, 2, 2, OPF_DATA },
1051 // pseudo-ops for lib calls
1052 { "_ftol", OPP_FTOL },
1053 // must be last
1054 { "ud2", OP_UD2 },
1055};
1056
1057static void parse_op(struct parsed_op *op, char words[16][256], int wordc)
1058{
1059 enum opr_lenmod lmod = OPLM_UNSPEC;
1060 int prefix_flags = 0;
1061 int regmask_ind;
1062 int regmask;
1063 int op_w = 0;
1064 int opr = 0;
1065 int w = 0;
1066 int i, j;
1067
1068 for (i = 0; i < ARRAY_SIZE(pref_table); i++) {
1069 if (IS(words[w], pref_table[i].name)) {
1070 prefix_flags = pref_table[i].flags;
1071 break;
1072 }
1073 }
1074
1075 if (prefix_flags) {
1076 if (wordc <= 1)
1077 aerr("lone prefix: '%s'\n", words[0]);
1078 w++;
1079 }
1080
1081 op_w = w;
1082 for (i = 0; i < ARRAY_SIZE(op_table); i++) {
1083 if (IS(words[w], op_table[i].name))
1084 break;
1085 }
1086
1087 if (i == ARRAY_SIZE(op_table)) {
1088 if (!g_skip_func)
1089 aerr("unhandled op: '%s'\n", words[0]);
1090 i--; // OP_UD2
1091 }
1092 w++;
1093
1094 op->op = op_table[i].op;
1095 op->flags = op_table[i].flags | prefix_flags;
1096 op->pfo = op_table[i].pfo;
1097 op->pfo_inv = op_table[i].pfo_inv;
1098 op->regmask_src = op->regmask_dst = 0;
1099 op->asmln = asmln;
1100
1101 if (op->op == OP_UD2)
1102 return;
1103
1104 for (opr = 0; opr < op_table[i].maxopr; opr++) {
1105 if (opr >= op_table[i].minopr && w >= wordc)
1106 break;
1107
1108 regmask = regmask_ind = 0;
1109 w = parse_operand(&op->operand[opr], &regmask, &regmask_ind,
1110 words, wordc, w, op->flags);
1111
1112 if (opr == 0 && (op->flags & OPF_DATA))
1113 op->regmask_dst = regmask;
1114 else
1115 op->regmask_src |= regmask;
1116 op->regmask_src |= regmask_ind;
1117
1118 if (op->operand[opr].lmod != OPLM_UNSPEC)
1119 g_func_lmods |= 1 << op->operand[opr].lmod;
1120 }
1121
1122 if (w < wordc)
1123 aerr("parse_op %s incomplete: %d/%d\n",
1124 words[0], w, wordc);
1125
1126 // special cases
1127 op->operand_cnt = opr;
1128 if (!strncmp(op_table[i].name, "set", 3))
1129 op->operand[0].lmod = OPLM_BYTE;
1130
1131 switch (op->op) {
1132 // first operand is not dst
1133 case OP_CMP:
1134 case OP_TEST:
1135 op->regmask_src |= op->regmask_dst;
1136 op->regmask_dst = 0;
1137 break;
1138
1139 // first operand is src too
1140 case OP_NOT:
1141 case OP_ADD:
1142 case OP_AND:
1143 case OP_OR:
1144 case OP_RCL:
1145 case OP_RCR:
1146 case OP_ADC:
1147 case OP_INC:
1148 case OP_DEC:
1149 case OP_NEG:
1150 // more below..
1151 op->regmask_src |= op->regmask_dst;
1152 break;
1153
1154 // special
1155 case OP_XCHG:
1156 op->regmask_src |= op->regmask_dst;
1157 op->regmask_dst |= op->regmask_src;
1158 goto check_align;
1159
1160 case OP_SUB:
1161 case OP_SBB:
1162 case OP_XOR:
1163 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1164 && op->operand[0].lmod == op->operand[1].lmod
1165 && op->operand[0].reg == op->operand[1].reg
1166 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
1167 {
1168 op->regmask_src = 0;
1169 }
1170 else
1171 op->regmask_src |= op->regmask_dst;
1172 break;
1173
1174 // ops with implicit argumets
1175 case OP_XLAT:
1176 op->operand_cnt = 2;
1177 setup_reg_opr(&op->operand[0], xAX, OPLM_BYTE, &op->regmask_src);
1178 op->regmask_dst = op->regmask_src;
1179 setup_reg_opr(&op->operand[1], xBX, OPLM_DWORD, &op->regmask_src);
1180 break;
1181
1182 case OP_CDQ:
1183 op->operand_cnt = 2;
1184 setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst);
1185 setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src);
1186 break;
1187
1188 case OP_LODS:
1189 case OP_STOS:
1190 case OP_SCAS:
1191 if (words[op_w][4] == 'b')
1192 lmod = OPLM_BYTE;
1193 else if (words[op_w][4] == 'w')
1194 lmod = OPLM_WORD;
1195 else if (words[op_w][4] == 'd')
1196 lmod = OPLM_DWORD;
1197 j = 0;
1198 op->regmask_src = 0;
1199 setup_reg_opr(&op->operand[j++], op->op == OP_LODS ? xSI : xDI,
1200 OPLM_DWORD, &op->regmask_src);
1201 op->regmask_dst = op->regmask_src;
1202 setup_reg_opr(&op->operand[j++], xAX, lmod,
1203 op->op == OP_LODS ? &op->regmask_dst : &op->regmask_src);
1204 if (op->flags & OPF_REP) {
1205 setup_reg_opr(&op->operand[j++], xCX, OPLM_DWORD, &op->regmask_src);
1206 op->regmask_dst |= 1 << xCX;
1207 }
1208 op->operand_cnt = j;
1209 break;
1210
1211 case OP_MOVS:
1212 case OP_CMPS:
1213 if (words[op_w][4] == 'b')
1214 lmod = OPLM_BYTE;
1215 else if (words[op_w][4] == 'w')
1216 lmod = OPLM_WORD;
1217 else if (words[op_w][4] == 'd')
1218 lmod = OPLM_DWORD;
1219 j = 0;
1220 op->regmask_src = 0;
1221 // note: lmod is not correct, don't have where to place it
1222 setup_reg_opr(&op->operand[j++], xDI, lmod, &op->regmask_src);
1223 setup_reg_opr(&op->operand[j++], xSI, OPLM_DWORD, &op->regmask_src);
1224 if (op->flags & OPF_REP)
1225 setup_reg_opr(&op->operand[j++], xCX, OPLM_DWORD, &op->regmask_src);
1226 op->operand_cnt = j;
1227 op->regmask_dst = op->regmask_src;
1228 break;
1229
1230 case OP_LOOP:
1231 op->regmask_dst = 1 << xCX;
1232 // fallthrough
1233 case OP_JECXZ:
1234 op->operand_cnt = 2;
1235 op->regmask_src = 1 << xCX;
1236 op->operand[1].type = OPT_REG;
1237 op->operand[1].reg = xCX;
1238 op->operand[1].lmod = OPLM_DWORD;
1239 break;
1240
1241 case OP_IMUL:
1242 if (op->operand_cnt == 2) {
1243 if (op->operand[0].type != OPT_REG)
1244 aerr("reg expected\n");
1245 op->regmask_src |= 1 << op->operand[0].reg;
1246 }
1247 if (op->operand_cnt != 1)
1248 break;
1249 // fallthrough
1250 case OP_MUL:
1251 // singleop mul
1252 op->regmask_src |= op->regmask_dst;
1253 op->regmask_dst = (1 << xDX) | (1 << xAX);
1254 if (op->operand[0].lmod == OPLM_UNSPEC)
1255 op->operand[0].lmod = OPLM_DWORD;
1256 break;
1257
1258 case OP_DIV:
1259 case OP_IDIV:
1260 // we could set up operands for edx:eax, but there is no real need to
1261 // (see is_opr_modified())
1262 op->regmask_src |= op->regmask_dst;
1263 op->regmask_dst = (1 << xDX) | (1 << xAX);
1264 if (op->operand[0].lmod == OPLM_UNSPEC)
1265 op->operand[0].lmod = OPLM_DWORD;
1266 break;
1267
1268 case OP_SHL:
1269 case OP_SHR:
1270 case OP_SAR:
1271 case OP_ROL:
1272 case OP_ROR:
1273 op->regmask_src |= op->regmask_dst;
1274 if (op->operand[1].lmod == OPLM_UNSPEC)
1275 op->operand[1].lmod = OPLM_BYTE;
1276 break;
1277
1278 case OP_SHLD:
1279 case OP_SHRD:
1280 op->regmask_src |= op->regmask_dst;
1281 if (op->operand[2].lmod == OPLM_UNSPEC)
1282 op->operand[2].lmod = OPLM_BYTE;
1283 break;
1284
1285 case OP_PUSH:
1286 op->regmask_src |= op->regmask_dst;
1287 op->regmask_dst = 0;
1288 if (op->operand[0].lmod == OPLM_UNSPEC
1289 && (op->operand[0].type == OPT_CONST
1290 || op->operand[0].type == OPT_OFFSET
1291 || op->operand[0].type == OPT_LABEL))
1292 op->operand[0].lmod = OPLM_DWORD;
1293 break;
1294
1295 // alignment
1296 case OP_MOV:
1297 check_align:
1298 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1299 && op->operand[0].lmod == op->operand[1].lmod
1300 && op->operand[0].reg == op->operand[1].reg
1301 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
1302 {
1303 op->flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
1304 op->regmask_src = op->regmask_dst = 0;
1305 }
1306 break;
1307
1308 case OP_LEA:
1309 if (op->operand[0].type == OPT_REG
1310 && op->operand[1].type == OPT_REGMEM)
1311 {
1312 char buf[16];
1313 snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name);
1314 if (IS(buf, op->operand[1].name))
1315 op->flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
1316 }
1317 break;
1318
1319 case OP_CALL:
1320 // trashed regs must be explicitly detected later
1321 op->regmask_dst = 0;
1322 break;
1323
1324 case OP_LEAVE:
1325 op->regmask_dst = (1 << xBP) | (1 << xSP);
1326 op->regmask_src = 1 << xBP;
1327 break;
1328
1329 case OP_FLD:
1330 case OP_FILD:
1331 op->regmask_dst |= mxST0;
1332 break;
1333
1334 case OP_FLDc:
1335 op->regmask_dst |= mxST0;
1336 if (IS(words[op_w] + 3, "1"))
1337 op->operand[0].val = X87_CONST_1;
1338 else if (IS(words[op_w] + 3, "z"))
1339 op->operand[0].val = X87_CONST_Z;
1340 else
1341 aerr("TODO\n");
1342 break;
1343
1344 case OP_FST:
1345 op->regmask_src |= mxST0;
1346 break;
1347
1348 case OP_FADD:
1349 case OP_FDIV:
1350 case OP_FMUL:
1351 case OP_FSUB:
1352 case OP_FDIVR:
1353 case OP_FSUBR:
1354 op->regmask_src |= mxST0;
1355 if (op->operand_cnt == 2)
1356 op->regmask_src |= op->regmask_dst;
1357 else if (op->operand_cnt == 1) {
1358 memcpy(&op->operand[1], &op->operand[0], sizeof(op->operand[1]));
1359 op->operand[0].type = OPT_REG;
1360 op->operand[0].lmod = OPLM_QWORD;
1361 op->operand[0].reg = xST0;
1362 op->regmask_dst |= mxST0;
1363 }
1364 else
1365 // IDA doesn't use this
1366 aerr("no operands?\n");
1367 break;
1368
1369 case OP_FIADD:
1370 case OP_FIDIV:
1371 case OP_FIMUL:
1372 case OP_FISUB:
1373 case OP_FIDIVR:
1374 case OP_FISUBR:
1375 op->regmask_src |= mxST0;
1376 op->regmask_dst |= mxST0;
1377 break;
1378
1379 default:
1380 break;
1381 }
1382
1383 if (op->operand[0].type == OPT_REG
1384 && op->operand[1].type == OPT_CONST)
1385 {
1386 struct parsed_opr *op1 = &op->operand[1];
1387 if ((op->op == OP_AND && op1->val == 0)
1388 || (op->op == OP_OR
1389 && (op1->val == ~0
1390 || (op->operand[0].lmod == OPLM_WORD && op1->val == 0xffff)
1391 || (op->operand[0].lmod == OPLM_BYTE && op1->val == 0xff))))
1392 {
1393 op->regmask_src = 0;
1394 }
1395 }
1396}
1397
1398static const char *op_name(struct parsed_op *po)
1399{
1400 static char buf[16];
1401 char *p;
1402 int i;
1403
1404 if (po->op == OP_JCC || po->op == OP_SCC) {
1405 p = buf;
1406 *p++ = (po->op == OP_JCC) ? 'j' : 's';
1407 if (po->pfo_inv)
1408 *p++ = 'n';
1409 strcpy(p, parsed_flag_op_names[po->pfo]);
1410 return buf;
1411 }
1412
1413 for (i = 0; i < ARRAY_SIZE(op_table); i++)
1414 if (op_table[i].op == po->op)
1415 return op_table[i].name;
1416
1417 return "???";
1418}
1419
1420// debug
1421static const char *dump_op(struct parsed_op *po)
1422{
1423 static char out[128];
1424 char *p = out;
1425 int i;
1426
1427 if (po == NULL)
1428 return "???";
1429
1430 snprintf(out, sizeof(out), "%s", op_name(po));
1431 for (i = 0; i < po->operand_cnt; i++) {
1432 p += strlen(p);
1433 if (i > 0)
1434 *p++ = ',';
1435 snprintf(p, sizeof(out) - (p - out),
1436 po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s",
1437 po->operand[i].name);
1438 }
1439
1440 return out;
1441}
1442
1443static const char *lmod_type_u(struct parsed_op *po,
1444 enum opr_lenmod lmod)
1445{
1446 switch (lmod) {
1447 case OPLM_QWORD:
1448 return "u64";
1449 case OPLM_DWORD:
1450 return "u32";
1451 case OPLM_WORD:
1452 return "u16";
1453 case OPLM_BYTE:
1454 return "u8";
1455 default:
1456 ferr(po, "invalid lmod: %d\n", lmod);
1457 return "(_invalid_)";
1458 }
1459}
1460
1461static const char *lmod_cast_u(struct parsed_op *po,
1462 enum opr_lenmod lmod)
1463{
1464 switch (lmod) {
1465 case OPLM_QWORD:
1466 return "";
1467 case OPLM_DWORD:
1468 return "";
1469 case OPLM_WORD:
1470 return "(u16)";
1471 case OPLM_BYTE:
1472 return "(u8)";
1473 default:
1474 ferr(po, "invalid lmod: %d\n", lmod);
1475 return "(_invalid_)";
1476 }
1477}
1478
1479static const char *lmod_cast_u_ptr(struct parsed_op *po,
1480 enum opr_lenmod lmod)
1481{
1482 switch (lmod) {
1483 case OPLM_QWORD:
1484 return "*(u64 *)";
1485 case OPLM_DWORD:
1486 return "*(u32 *)";
1487 case OPLM_WORD:
1488 return "*(u16 *)";
1489 case OPLM_BYTE:
1490 return "*(u8 *)";
1491 default:
1492 ferr(po, "invalid lmod: %d\n", lmod);
1493 return "(_invalid_)";
1494 }
1495}
1496
1497static const char *lmod_cast_s(struct parsed_op *po,
1498 enum opr_lenmod lmod)
1499{
1500 switch (lmod) {
1501 case OPLM_QWORD:
1502 return "(s64)";
1503 case OPLM_DWORD:
1504 return "(s32)";
1505 case OPLM_WORD:
1506 return "(s16)";
1507 case OPLM_BYTE:
1508 return "(s8)";
1509 default:
1510 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1511 return "(_invalid_)";
1512 }
1513}
1514
1515static const char *lmod_cast(struct parsed_op *po,
1516 enum opr_lenmod lmod, int is_signed)
1517{
1518 return is_signed ?
1519 lmod_cast_s(po, lmod) :
1520 lmod_cast_u(po, lmod);
1521}
1522
1523static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
1524{
1525 switch (lmod) {
1526 case OPLM_QWORD:
1527 return 8;
1528 case OPLM_DWORD:
1529 return 4;
1530 case OPLM_WORD:
1531 return 2;
1532 case OPLM_BYTE:
1533 return 1;
1534 default:
1535 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1536 return 0;
1537 }
1538}
1539
1540static const char *opr_name(struct parsed_op *po, int opr_num)
1541{
1542 if (opr_num >= po->operand_cnt)
1543 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1544 return po->operand[opr_num].name;
1545}
1546
1547static unsigned int opr_const(struct parsed_op *po, int opr_num)
1548{
1549 if (opr_num >= po->operand_cnt)
1550 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1551 if (po->operand[opr_num].type != OPT_CONST)
1552 ferr(po, "opr %d: const expected\n", opr_num);
1553 return po->operand[opr_num].val;
1554}
1555
1556static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr)
1557{
1558 if ((unsigned int)popr->reg >= ARRAY_SIZE(regs_r32))
1559 ferr(po, "invalid reg: %d\n", popr->reg);
1560 return regs_r32[popr->reg];
1561}
1562
1563static int check_simple_cast(const char *cast, int *bits, int *is_signed)
1564{
1565 if (IS_START(cast, "(s8)") || IS_START(cast, "(u8)"))
1566 *bits = 8;
1567 else if (IS_START(cast, "(s16)") || IS_START(cast, "(u16)"))
1568 *bits = 16;
1569 else if (IS_START(cast, "(s32)") || IS_START(cast, "(u32)"))
1570 *bits = 32;
1571 else if (IS_START(cast, "(s64)") || IS_START(cast, "(u64)"))
1572 *bits = 64;
1573 else
1574 return -1;
1575
1576 *is_signed = cast[1] == 's' ? 1 : 0;
1577 return 0;
1578}
1579
1580static int check_deref_cast(const char *cast, int *bits)
1581{
1582 if (IS_START(cast, "*(u8 *)"))
1583 *bits = 8;
1584 else if (IS_START(cast, "*(u16 *)"))
1585 *bits = 16;
1586 else if (IS_START(cast, "*(u32 *)"))
1587 *bits = 32;
1588 else if (IS_START(cast, "*(u64 *)"))
1589 *bits = 64;
1590 else
1591 return -1;
1592
1593 return 0;
1594}
1595
1596// cast1 is the "final" cast
1597static const char *simplify_cast(const char *cast1, const char *cast2)
1598{
1599 static char buf[256];
1600 int bits1, bits2;
1601 int s1, s2;
1602
1603 if (cast1[0] == 0)
1604 return cast2;
1605 if (cast2[0] == 0)
1606 return cast1;
1607 if (IS(cast1, cast2))
1608 return cast1;
1609
1610 if (check_simple_cast(cast1, &bits1, &s1) == 0
1611 && check_simple_cast(cast2, &bits2, &s2) == 0)
1612 {
1613 if (bits1 <= bits2)
1614 return cast1;
1615 }
1616 if (check_simple_cast(cast1, &bits1, &s1) == 0
1617 && check_deref_cast(cast2, &bits2) == 0)
1618 {
1619 if (bits1 == bits2) {
1620 snprintf(buf, sizeof(buf), "*(%c%d *)", s1 ? 's' : 'u', bits1);
1621 return buf;
1622 }
1623 }
1624
1625 if (strchr(cast1, '*') && IS_START(cast2, "(u32)"))
1626 return cast1;
1627
1628 snprintf(buf, sizeof(buf), "%s%s", cast1, cast2);
1629 return buf;
1630}
1631
1632static const char *simplify_cast_num(const char *cast, unsigned int val)
1633{
1634 if (IS(cast, "(u8)") && val < 0x100)
1635 return "";
1636 if (IS(cast, "(s8)") && val < 0x80)
1637 return "";
1638 if (IS(cast, "(u16)") && val < 0x10000)
1639 return "";
1640 if (IS(cast, "(s16)") && val < 0x8000)
1641 return "";
1642 if (IS(cast, "(s32)") && val < 0x80000000)
1643 return "";
1644
1645 return cast;
1646}
1647
1648static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
1649 int *extra_offs)
1650{
1651 const char *p;
1652 char *endp;
1653 int namelen;
1654 int i;
1655
1656 *extra_offs = 0;
1657 namelen = strlen(name);
1658
1659 p = strchr(name, '+');
1660 if (p != NULL) {
1661 namelen = p - name;
1662 if (namelen <= 0)
1663 ferr(po, "equ parse failed for '%s'\n", name);
1664
1665 if (IS_START(p, "0x"))
1666 p += 2;
1667 *extra_offs = strtol(p, &endp, 16);
1668 if (*endp != 0)
1669 ferr(po, "equ parse failed for '%s'\n", name);
1670 }
1671
1672 for (i = 0; i < g_eqcnt; i++)
1673 if (strncmp(g_eqs[i].name, name, namelen) == 0
1674 && g_eqs[i].name[namelen] == 0)
1675 break;
1676 if (i >= g_eqcnt) {
1677 if (po != NULL)
1678 ferr(po, "unresolved equ name: '%s'\n", name);
1679 return NULL;
1680 }
1681
1682 return &g_eqs[i];
1683}
1684
1685static int is_stack_access(struct parsed_op *po,
1686 const struct parsed_opr *popr)
1687{
1688 return (parse_stack_el(popr->name, NULL, 0)
1689 || (g_bp_frame && !(po->flags & OPF_EBP_S)
1690 && IS_START(popr->name, "ebp")));
1691}
1692
1693static void parse_stack_access(struct parsed_op *po,
1694 const char *name, char *ofs_reg, int *offset_out,
1695 int *stack_ra_out, const char **bp_arg_out, int is_lea)
1696{
1697 const char *bp_arg = "";
1698 const char *p = NULL;
1699 struct parsed_equ *eq;
1700 char *endp = NULL;
1701 int stack_ra = 0;
1702 int offset = 0;
1703
1704 ofs_reg[0] = 0;
1705
1706 if (IS_START(name, "ebp-")
1707 || (IS_START(name, "ebp+") && '0' <= name[4] && name[4] <= '9'))
1708 {
1709 p = name + 4;
1710 if (IS_START(p, "0x"))
1711 p += 2;
1712 offset = strtoul(p, &endp, 16);
1713 if (name[3] == '-')
1714 offset = -offset;
1715 if (*endp != 0)
1716 ferr(po, "ebp- parse of '%s' failed\n", name);
1717 }
1718 else {
1719 bp_arg = parse_stack_el(name, ofs_reg, 0);
1720 snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
1721 eq = equ_find(po, bp_arg, &offset);
1722 if (eq == NULL)
1723 ferr(po, "detected but missing eq\n");
1724 offset += eq->offset;
1725 }
1726
1727 if (!strncmp(name, "ebp", 3))
1728 stack_ra = 4;
1729
1730 // yes it sometimes LEAs ra for compares..
1731 if (!is_lea && ofs_reg[0] == 0
1732 && stack_ra <= offset && offset < stack_ra + 4)
1733 {
1734 ferr(po, "reference to ra? %d %d\n", offset, stack_ra);
1735 }
1736
1737 *offset_out = offset;
1738 *stack_ra_out = stack_ra;
1739 if (bp_arg_out)
1740 *bp_arg_out = bp_arg;
1741}
1742
1743static int stack_frame_access(struct parsed_op *po,
1744 struct parsed_opr *popr, char *buf, size_t buf_size,
1745 const char *name, const char *cast, int is_src, int is_lea)
1746{
1747 enum opr_lenmod tmp_lmod = OPLM_UNSPEC;
1748 const char *prefix = "";
1749 const char *bp_arg = NULL;
1750 char ofs_reg[16] = { 0, };
1751 int i, arg_i, arg_s;
1752 int unaligned = 0;
1753 int stack_ra = 0;
1754 int offset = 0;
1755 int retval = -1;
1756 int sf_ofs;
1757 int lim;
1758
1759 if (po->flags & OPF_EBP_S)
1760 ferr(po, "stack_frame_access while ebp is scratch\n");
1761
1762 parse_stack_access(po, name, ofs_reg, &offset,
1763 &stack_ra, &bp_arg, is_lea);
1764
1765 if (offset > stack_ra)
1766 {
1767 arg_i = (offset - stack_ra - 4) / 4;
1768 if (arg_i < 0 || arg_i >= g_func_pp->argc_stack)
1769 {
1770 if (g_func_pp->is_vararg
1771 && arg_i == g_func_pp->argc_stack && is_lea)
1772 {
1773 // should be va_list
1774 if (cast[0] == 0)
1775 cast = "(u32)";
1776 snprintf(buf, buf_size, "%sap", cast);
1777 return -1;
1778 }
1779 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1780 offset, bp_arg, arg_i);
1781 }
1782 if (ofs_reg[0] != 0)
1783 ferr(po, "offset reg on arg access?\n");
1784
1785 for (i = arg_s = 0; i < g_func_pp->argc; i++) {
1786 if (g_func_pp->arg[i].reg != NULL)
1787 continue;
1788 if (arg_s == arg_i)
1789 break;
1790 arg_s++;
1791 }
1792 if (i == g_func_pp->argc)
1793 ferr(po, "arg %d not in prototype?\n", arg_i);
1794
1795 popr->is_ptr = g_func_pp->arg[i].type.is_ptr;
1796 retval = i;
1797
1798 switch (popr->lmod)
1799 {
1800 case OPLM_BYTE:
1801 if (is_lea)
1802 ferr(po, "lea/byte to arg?\n");
1803 if (is_src && (offset & 3) == 0)
1804 snprintf(buf, buf_size, "%sa%d",
1805 simplify_cast(cast, "(u8)"), i + 1);
1806 else
1807 snprintf(buf, buf_size, "%sBYTE%d(a%d)",
1808 cast, offset & 3, i + 1);
1809 break;
1810
1811 case OPLM_WORD:
1812 if (is_lea)
1813 ferr(po, "lea/word to arg?\n");
1814 if (offset & 1) {
1815 unaligned = 1;
1816 if (!is_src) {
1817 if (offset & 2)
1818 ferr(po, "problematic arg store\n");
1819 snprintf(buf, buf_size, "%s((char *)&a%d + 1)",
1820 simplify_cast(cast, "*(u16 *)"), i + 1);
1821 }
1822 else
1823 ferr(po, "unaligned arg word load\n");
1824 }
1825 else if (is_src && (offset & 2) == 0)
1826 snprintf(buf, buf_size, "%sa%d",
1827 simplify_cast(cast, "(u16)"), i + 1);
1828 else
1829 snprintf(buf, buf_size, "%s%sWORD(a%d)",
1830 cast, (offset & 2) ? "HI" : "LO", i + 1);
1831 break;
1832
1833 case OPLM_DWORD:
1834 if (cast[0])
1835 prefix = cast;
1836 else if (is_src)
1837 prefix = "(u32)";
1838
1839 if (offset & 3) {
1840 unaligned = 1;
1841 if (is_lea)
1842 snprintf(buf, buf_size, "(u32)&a%d + %d",
1843 i + 1, offset & 3);
1844 else if (!is_src)
1845 ferr(po, "unaligned arg store\n");
1846 else {
1847 // mov edx, [ebp+arg_4+2]; movsx ecx, dx
1848 snprintf(buf, buf_size, "%s(a%d >> %d)",
1849 prefix, i + 1, (offset & 3) * 8);
1850 }
1851 }
1852 else {
1853 snprintf(buf, buf_size, "%s%sa%d",
1854 prefix, is_lea ? "&" : "", i + 1);
1855 }
1856 break;
1857
1858 default:
1859 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
1860 }
1861
1862 if (unaligned)
1863 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1864
1865 // common problem
1866 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type);
1867 if (tmp_lmod != OPLM_DWORD
1868 && (unaligned || (!is_src && lmod_bytes(po, tmp_lmod)
1869 < lmod_bytes(po, popr->lmod) + (offset & 3))))
1870 {
1871 ferr(po, "bp_arg arg%d/w offset %d and type '%s' is too small\n",
1872 i + 1, offset, g_func_pp->arg[i].type.name);
1873 }
1874 // can't check this because msvc likes to reuse
1875 // arg space for scratch..
1876 //if (popr->is_ptr && popr->lmod != OPLM_DWORD)
1877 // ferr(po, "bp_arg arg%d: non-dword ptr access\n", i + 1);
1878 }
1879 else
1880 {
1881 if (g_stack_fsz == 0)
1882 ferr(po, "stack var access without stackframe\n");
1883 g_stack_frame_used = 1;
1884
1885 sf_ofs = g_stack_fsz + offset;
1886 lim = (ofs_reg[0] != 0) ? -4 : 0;
1887 if (offset > 0 || sf_ofs < lim)
1888 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
1889
1890 if (is_lea)
1891 prefix = "(u32)&";
1892 else
1893 prefix = cast;
1894
1895 switch (popr->lmod)
1896 {
1897 case OPLM_BYTE:
1898 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1899 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1900 break;
1901
1902 case OPLM_WORD:
1903 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1904 // known unaligned or possibly unaligned
1905 strcat(g_comment, " unaligned");
1906 if (prefix[0] == 0)
1907 prefix = "*(u16 *)&";
1908 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1909 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1910 break;
1911 }
1912 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
1913 break;
1914
1915 case OPLM_DWORD:
1916 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1917 // known unaligned or possibly unaligned
1918 strcat(g_comment, " unaligned");
1919 if (prefix[0] == 0)
1920 prefix = "*(u32 *)&";
1921 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1922 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1923 break;
1924 }
1925 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
1926 break;
1927
1928 case OPLM_QWORD:
1929 ferr_assert(po, !(sf_ofs & 7));
1930 ferr_assert(po, ofs_reg[0] == 0);
1931 // float callers set is_lea
1932 ferr_assert(po, is_lea);
1933 snprintf(buf, buf_size, "%ssf.q[%d]", prefix, sf_ofs / 8);
1934 break;
1935
1936 default:
1937 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
1938 }
1939 }
1940
1941 return retval;
1942}
1943
1944static void check_func_pp(struct parsed_op *po,
1945 const struct parsed_proto *pp, const char *pfx)
1946{
1947 enum opr_lenmod tmp_lmod;
1948 char buf[256];
1949 int ret, i;
1950
1951 if (pp->argc_reg != 0) {
1952 if (/*!g_allow_regfunc &&*/ !pp->is_fastcall) {
1953 pp_print(buf, sizeof(buf), pp);
1954 ferr(po, "%s: unexpected reg arg in icall: %s\n", pfx, buf);
1955 }
1956 if (pp->argc_stack > 0 && pp->argc_reg != 2)
1957 ferr(po, "%s: %d reg arg(s) with %d stack arg(s)\n",
1958 pfx, pp->argc_reg, pp->argc_stack);
1959 }
1960
1961 // fptrs must use 32bit args, callsite might have no information and
1962 // lack a cast to smaller types, which results in incorrectly masked
1963 // args passed (callee may assume masked args, it does on ARM)
1964 if (!pp->is_osinc) {
1965 for (i = 0; i < pp->argc; i++) {
1966 ret = guess_lmod_from_c_type(&tmp_lmod, &pp->arg[i].type);
1967 if (ret && tmp_lmod != OPLM_DWORD)
1968 ferr(po, "reference to %s with arg%d '%s'\n", pp->name,
1969 i + 1, pp->arg[i].type.name);
1970 }
1971 }
1972}
1973
1974static const char *check_label_read_ref(struct parsed_op *po,
1975 const char *name)
1976{
1977 const struct parsed_proto *pp;
1978
1979 pp = proto_parse(g_fhdr, name, 0);
1980 if (pp == NULL)
1981 ferr(po, "proto_parse failed for ref '%s'\n", name);
1982
1983 if (pp->is_func)
1984 check_func_pp(po, pp, "ref");
1985
1986 return pp->name;
1987}
1988
1989static char *out_src_opr(char *buf, size_t buf_size,
1990 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
1991 int is_lea)
1992{
1993 char tmp1[256], tmp2[256];
1994 char expr[256];
1995 const char *name;
1996 char *p;
1997 int ret;
1998
1999 if (cast == NULL)
2000 cast = "";
2001
2002 switch (popr->type) {
2003 case OPT_REG:
2004 if (is_lea)
2005 ferr(po, "lea from reg?\n");
2006
2007 switch (popr->lmod) {
2008 case OPLM_QWORD:
2009 snprintf(buf, buf_size, "%s%s.q", cast, opr_reg_p(po, popr));
2010 break;
2011 case OPLM_DWORD:
2012 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
2013 break;
2014 case OPLM_WORD:
2015 snprintf(buf, buf_size, "%s%s",
2016 simplify_cast(cast, "(u16)"), opr_reg_p(po, popr));
2017 break;
2018 case OPLM_BYTE:
2019 if (popr->name[1] == 'h') // XXX..
2020 snprintf(buf, buf_size, "%s(%s >> 8)",
2021 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
2022 else
2023 snprintf(buf, buf_size, "%s%s",
2024 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
2025 break;
2026 default:
2027 ferr(po, "invalid src lmod: %d\n", popr->lmod);
2028 }
2029 break;
2030
2031 case OPT_REGMEM:
2032 if (is_stack_access(po, popr)) {
2033 stack_frame_access(po, popr, buf, buf_size,
2034 popr->name, cast, 1, is_lea);
2035 break;
2036 }
2037
2038 strcpy(expr, popr->name);
2039 if (strchr(expr, '[')) {
2040 // special case: '[' can only be left for label[reg] form
2041 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
2042 if (ret != 2)
2043 ferr(po, "parse failure for '%s'\n", expr);
2044 if (tmp1[0] == '(') {
2045 // (off_4FFF50+3)[eax]
2046 p = strchr(tmp1 + 1, ')');
2047 if (p == NULL || p[1] != 0)
2048 ferr(po, "parse failure (2) for '%s'\n", expr);
2049 *p = 0;
2050 memmove(tmp1, tmp1 + 1, strlen(tmp1));
2051 }
2052 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
2053 }
2054
2055 // XXX: do we need more parsing?
2056 if (is_lea) {
2057 snprintf(buf, buf_size, "%s", expr);
2058 break;
2059 }
2060
2061 snprintf(buf, buf_size, "%s(%s)",
2062 simplify_cast(cast, lmod_cast_u_ptr(po, popr->lmod)), expr);
2063 break;
2064
2065 case OPT_LABEL:
2066 name = check_label_read_ref(po, popr->name);
2067 if (cast[0] == 0 && popr->is_ptr)
2068 cast = "(u32)";
2069
2070 if (is_lea)
2071 snprintf(buf, buf_size, "(u32)&%s", name);
2072 else if (popr->size_lt)
2073 snprintf(buf, buf_size, "%s%s%s%s", cast,
2074 lmod_cast_u_ptr(po, popr->lmod),
2075 popr->is_array ? "" : "&", name);
2076 else
2077 snprintf(buf, buf_size, "%s%s%s", cast, name,
2078 popr->is_array ? "[0]" : "");
2079 break;
2080
2081 case OPT_OFFSET:
2082 name = check_label_read_ref(po, popr->name);
2083 if (cast[0] == 0)
2084 cast = "(u32)";
2085 if (is_lea)
2086 ferr(po, "lea an offset?\n");
2087 snprintf(buf, buf_size, "%s&%s", cast, name);
2088 break;
2089
2090 case OPT_CONST:
2091 if (is_lea)
2092 ferr(po, "lea from const?\n");
2093
2094 printf_number(tmp1, sizeof(tmp1), popr->val);
2095 if (popr->val == 0 && strchr(cast, '*'))
2096 snprintf(buf, buf_size, "NULL");
2097 else
2098 snprintf(buf, buf_size, "%s%s",
2099 simplify_cast_num(cast, popr->val), tmp1);
2100 break;
2101
2102 default:
2103 ferr(po, "invalid src type: %d\n", popr->type);
2104 }
2105
2106 return buf;
2107}
2108
2109// note: may set is_ptr (we find that out late for ebp frame..)
2110static char *out_dst_opr(char *buf, size_t buf_size,
2111 struct parsed_op *po, struct parsed_opr *popr)
2112{
2113 switch (popr->type) {
2114 case OPT_REG:
2115 switch (popr->lmod) {
2116 case OPLM_QWORD:
2117 snprintf(buf, buf_size, "%s.q", opr_reg_p(po, popr));
2118 break;
2119 case OPLM_DWORD:
2120 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
2121 break;
2122 case OPLM_WORD:
2123 // ugh..
2124 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
2125 break;
2126 case OPLM_BYTE:
2127 // ugh..
2128 if (popr->name[1] == 'h') // XXX..
2129 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
2130 else
2131 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
2132 break;
2133 default:
2134 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
2135 }
2136 break;
2137
2138 case OPT_REGMEM:
2139 if (is_stack_access(po, popr)) {
2140 stack_frame_access(po, popr, buf, buf_size,
2141 popr->name, "", 0, 0);
2142 break;
2143 }
2144
2145 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
2146
2147 case OPT_LABEL:
2148 if (popr->size_mismatch)
2149 snprintf(buf, buf_size, "%s%s%s",
2150 lmod_cast_u_ptr(po, popr->lmod),
2151 popr->is_array ? "" : "&", popr->name);
2152 else
2153 snprintf(buf, buf_size, "%s%s", popr->name,
2154 popr->is_array ? "[0]" : "");
2155 break;
2156
2157 default:
2158 ferr(po, "invalid dst type: %d\n", popr->type);
2159 }
2160
2161 return buf;
2162}
2163
2164static char *out_src_opr_u32(char *buf, size_t buf_size,
2165 struct parsed_op *po, struct parsed_opr *popr)
2166{
2167 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
2168}
2169
2170static char *out_src_opr_float(char *buf, size_t buf_size,
2171 struct parsed_op *po, struct parsed_opr *popr)
2172{
2173 const char *cast = NULL;
2174 char tmp[256];
2175
2176 switch (popr->type) {
2177 case OPT_REG:
2178 if (popr->reg < xST0 || popr->reg > xST7)
2179 ferr(po, "bad reg: %d\n", popr->reg);
2180
2181 snprintf(buf, buf_size, "f_st%d", popr->reg - xST0);
2182 break;
2183
2184 case OPT_REGMEM:
2185 case OPT_LABEL:
2186 case OPT_OFFSET:
2187 switch (popr->lmod) {
2188 case OPLM_QWORD:
2189 cast = "double";
2190 break;
2191 case OPLM_DWORD:
2192 cast = "float";
2193 break;
2194 default:
2195 ferr(po, "unhandled lmod: %d\n", popr->lmod);
2196 break;
2197 }
2198 out_src_opr(tmp, sizeof(tmp), po, popr, "", 1);
2199 snprintf(buf, buf_size, "*((%s *)%s)", cast, tmp);
2200 break;
2201
2202 default:
2203 ferr(po, "invalid float type: %d\n", popr->type);
2204 }
2205
2206 return buf;
2207}
2208
2209static char *out_dst_opr_float(char *buf, size_t buf_size,
2210 struct parsed_op *po, struct parsed_opr *popr)
2211{
2212 // same?
2213 return out_src_opr_float(buf, buf_size, po, popr);
2214}
2215
2216static void out_test_for_cc(char *buf, size_t buf_size,
2217 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
2218 enum opr_lenmod lmod, const char *expr)
2219{
2220 const char *cast, *scast;
2221
2222 cast = lmod_cast_u(po, lmod);
2223 scast = lmod_cast_s(po, lmod);
2224
2225 switch (pfo) {
2226 case PFO_Z:
2227 case PFO_BE: // CF=1||ZF=1; CF=0
2228 snprintf(buf, buf_size, "(%s%s %s 0)",
2229 cast, expr, is_inv ? "!=" : "==");
2230 break;
2231
2232 case PFO_S:
2233 case PFO_L: // SF!=OF; OF=0
2234 snprintf(buf, buf_size, "(%s%s %s 0)",
2235 scast, expr, is_inv ? ">=" : "<");
2236 break;
2237
2238 case PFO_LE: // ZF=1||SF!=OF; OF=0
2239 snprintf(buf, buf_size, "(%s%s %s 0)",
2240 scast, expr, is_inv ? ">" : "<=");
2241 break;
2242
2243 default:
2244 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
2245 }
2246}
2247
2248static void out_cmp_for_cc(char *buf, size_t buf_size,
2249 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
2250{
2251 const char *cast, *scast, *cast_use;
2252 char buf1[256], buf2[256];
2253 enum opr_lenmod lmod;
2254
2255 if (po->op != OP_DEC && po->operand[0].lmod != po->operand[1].lmod)
2256 ferr(po, "%s: lmod mismatch: %d %d\n", __func__,
2257 po->operand[0].lmod, po->operand[1].lmod);
2258 lmod = po->operand[0].lmod;
2259
2260 cast = lmod_cast_u(po, lmod);
2261 scast = lmod_cast_s(po, lmod);
2262
2263 switch (pfo) {
2264 case PFO_C:
2265 case PFO_Z:
2266 case PFO_BE: // !a
2267 cast_use = cast;
2268 break;
2269
2270 case PFO_S:
2271 case PFO_L: // !ge
2272 case PFO_LE:
2273 cast_use = scast;
2274 break;
2275
2276 default:
2277 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
2278 }
2279
2280 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], cast_use, 0);
2281 if (po->op == OP_DEC)
2282 snprintf(buf2, sizeof(buf2), "1");
2283 else
2284 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], cast_use, 0);
2285
2286 switch (pfo) {
2287 case PFO_C:
2288 // note: must be unsigned compare
2289 snprintf(buf, buf_size, "(%s %s %s)",
2290 buf1, is_inv ? ">=" : "<", buf2);
2291 break;
2292
2293 case PFO_Z:
2294 snprintf(buf, buf_size, "(%s %s %s)",
2295 buf1, is_inv ? "!=" : "==", buf2);
2296 break;
2297
2298 case PFO_BE: // !a
2299 // note: must be unsigned compare
2300 snprintf(buf, buf_size, "(%s %s %s)",
2301 buf1, is_inv ? ">" : "<=", buf2);
2302
2303 // annoying case
2304 if (is_inv && lmod == OPLM_BYTE
2305 && po->operand[1].type == OPT_CONST
2306 && po->operand[1].val == 0xff)
2307 {
2308 snprintf(g_comment, sizeof(g_comment), "if %s", buf);
2309 snprintf(buf, buf_size, "(0)");
2310 }
2311 break;
2312
2313 // note: must be signed compare
2314 case PFO_S:
2315 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
2316 scast, buf1, buf2, is_inv ? ">=" : "<");
2317 break;
2318
2319 case PFO_L: // !ge
2320 snprintf(buf, buf_size, "(%s %s %s)",
2321 buf1, is_inv ? ">=" : "<", buf2);
2322 break;
2323
2324 case PFO_LE: // !g
2325 snprintf(buf, buf_size, "(%s %s %s)",
2326 buf1, is_inv ? ">" : "<=", buf2);
2327 break;
2328
2329 default:
2330 break;
2331 }
2332}
2333
2334static void out_cmp_test(char *buf, size_t buf_size,
2335 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
2336{
2337 char buf1[256], buf2[256], buf3[256];
2338
2339 if (po->op == OP_TEST) {
2340 if (IS(opr_name(po, 0), opr_name(po, 1))) {
2341 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
2342 }
2343 else {
2344 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
2345 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
2346 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
2347 }
2348 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
2349 po->operand[0].lmod, buf3);
2350 }
2351 else if (po->op == OP_CMP) {
2352 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv);
2353 }
2354 else
2355 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
2356}
2357
2358static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
2359 struct parsed_opr *popr2)
2360{
2361 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
2362 ferr(po, "missing lmod for both operands\n");
2363
2364 if (popr1->lmod == OPLM_UNSPEC)
2365 popr1->lmod = popr2->lmod;
2366 else if (popr2->lmod == OPLM_UNSPEC)
2367 popr2->lmod = popr1->lmod;
2368 else if (popr1->lmod != popr2->lmod) {
2369 if (popr1->type_from_var) {
2370 popr1->size_mismatch = 1;
2371 if (popr1->lmod < popr2->lmod)
2372 popr1->size_lt = 1;
2373 popr1->lmod = popr2->lmod;
2374 }
2375 else if (popr2->type_from_var) {
2376 popr2->size_mismatch = 1;
2377 if (popr2->lmod < popr1->lmod)
2378 popr2->size_lt = 1;
2379 popr2->lmod = popr1->lmod;
2380 }
2381 else
2382 ferr(po, "conflicting lmods: %d vs %d\n",
2383 popr1->lmod, popr2->lmod);
2384 }
2385}
2386
2387static const char *op_to_c(struct parsed_op *po)
2388{
2389 switch (po->op)
2390 {
2391 case OP_ADD:
2392 case OP_ADC:
2393 return "+";
2394 case OP_SUB:
2395 case OP_SBB:
2396 return "-";
2397 case OP_AND:
2398 return "&";
2399 case OP_OR:
2400 return "|";
2401 case OP_XOR:
2402 return "^";
2403 case OP_SHL:
2404 return "<<";
2405 case OP_SHR:
2406 return ">>";
2407 case OP_MUL:
2408 case OP_IMUL:
2409 return "*";
2410 default:
2411 ferr(po, "op_to_c was supplied with %d\n", po->op);
2412 }
2413}
2414
2415// last op in stream - unconditional branch or ret
2416#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
2417 || ((ops[_i].flags & (OPF_JMP|OPF_CJMP|OPF_RMD)) == OPF_JMP \
2418 && ops[_i].op != OP_CALL))
2419
2420#define check_i(po, i) \
2421 if ((i) < 0) \
2422 ferr(po, "bad " #i ": %d\n", i)
2423
2424// note: this skips over calls and rm'd stuff assuming they're handled
2425// so it's intended to use at one of final passes
2426static int scan_for_pop(int i, int opcnt, int magic, int reg,
2427 int depth, int flags_set)
2428{
2429 struct parsed_op *po;
2430 int relevant;
2431 int ret = 0;
2432 int j;
2433
2434 for (; i < opcnt; i++) {
2435 po = &ops[i];
2436 if (po->cc_scratch == magic)
2437 return ret; // already checked
2438 po->cc_scratch = magic;
2439
2440 if (po->flags & OPF_TAIL) {
2441 if (po->op == OP_CALL) {
2442 if (po->pp != NULL && po->pp->is_noreturn)
2443 // assume no stack cleanup for noreturn
2444 return 1;
2445 }
2446 return -1; // deadend
2447 }
2448
2449 if (po->flags & (OPF_RMD|OPF_DONE|OPF_FARG))
2450 continue;
2451
2452 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2453 if (po->btj != NULL) {
2454 // jumptable
2455 for (j = 0; j < po->btj->count; j++) {
2456 check_i(po, po->btj->d[j].bt_i);
2457 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, magic, reg,
2458 depth, flags_set);
2459 if (ret < 0)
2460 return ret; // dead end
2461 }
2462 return ret;
2463 }
2464
2465 check_i(po, po->bt_i);
2466 if (po->flags & OPF_CJMP) {
2467 ret |= scan_for_pop(po->bt_i, opcnt, magic, reg,
2468 depth, flags_set);
2469 if (ret < 0)
2470 return ret; // dead end
2471 }
2472 else {
2473 i = po->bt_i - 1;
2474 }
2475 continue;
2476 }
2477
2478 relevant = 0;
2479 if ((po->op == OP_POP || po->op == OP_PUSH)
2480 && po->operand[0].type == OPT_REG && po->operand[0].reg == reg)
2481 {
2482 relevant = 1;
2483 }
2484
2485 if (po->op == OP_PUSH) {
2486 depth++;
2487 }
2488 else if (po->op == OP_POP) {
2489 if (relevant && depth == 0) {
2490 po->flags |= flags_set;
2491 return 1;
2492 }
2493 depth--;
2494 }
2495 }
2496
2497 return -1;
2498}
2499
2500// scan for 'reg' pop backwards starting from i
2501// intended to use for register restore search, so other reg
2502// references are considered an error
2503static int scan_for_rsave_pop_reg(int i, int magic, int reg, int set_flags)
2504{
2505 struct parsed_op *po;
2506 struct label_ref *lr;
2507 int ret = 0;
2508
2509 ops[i].cc_scratch = magic;
2510
2511 while (1)
2512 {
2513 if (g_labels[i] != NULL) {
2514 lr = &g_label_refs[i];
2515 for (; lr != NULL; lr = lr->next) {
2516 check_i(&ops[i], lr->i);
2517 ret |= scan_for_rsave_pop_reg(lr->i, magic, reg, set_flags);
2518 if (ret < 0)
2519 return ret;
2520 }
2521 if (i > 0 && LAST_OP(i - 1))
2522 return ret;
2523 }
2524
2525 i--;
2526 if (i < 0)
2527 break;
2528
2529 if (ops[i].cc_scratch == magic)
2530 return ret;
2531 ops[i].cc_scratch = magic;
2532
2533 po = &ops[i];
2534 if (po->op == OP_POP && po->operand[0].reg == reg) {
2535 if (po->flags & (OPF_RMD|OPF_DONE))
2536 return -1;
2537
2538 po->flags |= set_flags;
2539 return 1;
2540 }
2541
2542 // this also covers the case where we reach corresponding push
2543 if ((po->regmask_dst | po->regmask_src) & (1 << reg))
2544 return -1;
2545 }
2546
2547 // nothing interesting on this path
2548 return 0;
2549}
2550
2551static void find_reachable_exits(int i, int opcnt, int magic,
2552 int *exits, int *exit_count)
2553{
2554 struct parsed_op *po;
2555 int j;
2556
2557 for (; i < opcnt; i++)
2558 {
2559 po = &ops[i];
2560 if (po->cc_scratch == magic)
2561 return;
2562 po->cc_scratch = magic;
2563
2564 if (po->flags & OPF_TAIL) {
2565 ferr_assert(po, *exit_count < MAX_EXITS);
2566 exits[*exit_count] = i;
2567 (*exit_count)++;
2568 return;
2569 }
2570
2571 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2572 if (po->flags & OPF_RMD)
2573 continue;
2574
2575 if (po->btj != NULL) {
2576 for (j = 0; j < po->btj->count; j++) {
2577 check_i(po, po->btj->d[j].bt_i);
2578 find_reachable_exits(po->btj->d[j].bt_i, opcnt, magic,
2579 exits, exit_count);
2580 }
2581 return;
2582 }
2583
2584 check_i(po, po->bt_i);
2585 if (po->flags & OPF_CJMP)
2586 find_reachable_exits(po->bt_i, opcnt, magic, exits, exit_count);
2587 else
2588 i = po->bt_i - 1;
2589 continue;
2590 }
2591 }
2592}
2593
2594// scan for 'reg' pop backwards starting from exits (all paths)
2595static int scan_for_pop_ret(int i, int opcnt, int reg, int set_flags)
2596{
2597 static int exits[MAX_EXITS];
2598 static int exit_count;
2599 int j, ret;
2600
2601 if (!set_flags) {
2602 exit_count = 0;
2603 find_reachable_exits(i, opcnt, i + opcnt * 15, exits,
2604 &exit_count);
2605 ferr_assert(&ops[i], exit_count > 0);
2606 }
2607
2608 for (j = 0; j < exit_count; j++) {
2609 ret = scan_for_rsave_pop_reg(exits[j], i + opcnt * 16 + set_flags,
2610 reg, set_flags);
2611 if (ret == -1)
2612 return -1;
2613 }
2614
2615 return 1;
2616}
2617
2618// scan for one or more pop of push <const>
2619static int scan_for_pop_const_r(int i, int opcnt, int magic,
2620 int push_i, int is_probe)
2621{
2622 struct parsed_op *po;
2623 struct label_ref *lr;
2624 int ret = 0;
2625 int j;
2626
2627 for (; i < opcnt; i++)
2628 {
2629 po = &ops[i];
2630 if (po->cc_scratch == magic)
2631 return ret; // already checked
2632 po->cc_scratch = magic;
2633
2634 if (po->flags & OPF_JMP) {
2635 if (po->flags & OPF_RMD)
2636 continue;
2637 if (po->op == OP_CALL)
2638 return -1;
2639
2640 if (po->btj != NULL) {
2641 for (j = 0; j < po->btj->count; j++) {
2642 check_i(po, po->btj->d[j].bt_i);
2643 ret |= scan_for_pop_const_r(po->btj->d[j].bt_i, opcnt, magic,
2644 push_i, is_probe);
2645 if (ret < 0)
2646 return ret;
2647 }
2648 return ret;
2649 }
2650
2651 check_i(po, po->bt_i);
2652 if (po->flags & OPF_CJMP) {
2653 ret |= scan_for_pop_const_r(po->bt_i, opcnt, magic, push_i,
2654 is_probe);
2655 if (ret < 0)
2656 return ret;
2657 }
2658 else {
2659 i = po->bt_i - 1;
2660 }
2661 continue;
2662 }
2663
2664 if ((po->flags & (OPF_TAIL|OPF_RSAVE)) || po->op == OP_PUSH)
2665 return -1;
2666
2667 if (g_labels[i] != NULL) {
2668 // all refs must be visited
2669 lr = &g_label_refs[i];
2670 for (; lr != NULL; lr = lr->next) {
2671 check_i(po, lr->i);
2672 if (ops[lr->i].cc_scratch != magic)
2673 return -1;
2674 }
2675 if (i > 0 && !LAST_OP(i - 1) && ops[i - 1].cc_scratch != magic)
2676 return -1;
2677 }
2678
2679 if (po->op == OP_POP)
2680 {
2681 if (po->flags & (OPF_RMD|OPF_DONE))
2682 return -1;
2683
2684 if (!is_probe) {
2685 po->flags |= OPF_DONE;
2686 po->datap = &ops[push_i];
2687 }
2688 return 1;
2689 }
2690 }
2691
2692 return -1;
2693}
2694
2695static void scan_for_pop_const(int i, int opcnt, int magic)
2696{
2697 int ret;
2698
2699 ret = scan_for_pop_const_r(i + 1, opcnt, magic, i, 1);
2700 if (ret == 1) {
2701 ops[i].flags |= OPF_RMD | OPF_DONE;
2702 scan_for_pop_const_r(i + 1, opcnt, magic + 1, i, 0);
2703 }
2704}
2705
2706// check if all branch targets within a marked path are also marked
2707// note: the path checked must not be empty or end with a branch
2708static int check_path_branches(int opcnt, int magic)
2709{
2710 struct parsed_op *po;
2711 int i, j;
2712
2713 for (i = 0; i < opcnt; i++) {
2714 po = &ops[i];
2715 if (po->cc_scratch != magic)
2716 continue;
2717
2718 if (po->flags & OPF_JMP) {
2719 if ((po->flags & OPF_RMD) || po->op == OP_CALL)
2720 continue;
2721
2722 if (po->btj != NULL) {
2723 for (j = 0; j < po->btj->count; j++) {
2724 check_i(po, po->btj->d[j].bt_i);
2725 if (ops[po->btj->d[j].bt_i].cc_scratch != magic)
2726 return 0;
2727 }
2728 }
2729
2730 check_i(po, po->bt_i);
2731 if (ops[po->bt_i].cc_scratch != magic)
2732 return 0;
2733 if ((po->flags & OPF_CJMP) && ops[i + 1].cc_scratch != magic)
2734 return 0;
2735 }
2736 }
2737
2738 return 1;
2739}
2740
2741// scan for multiple pushes for given pop
2742static int scan_pushes_for_pop_r(int i, int magic, int pop_i,
2743 int is_probe)
2744{
2745 int reg = ops[pop_i].operand[0].reg;
2746 struct parsed_op *po;
2747 struct label_ref *lr;
2748 int ret = 0;
2749
2750 ops[i].cc_scratch = magic;
2751
2752 while (1)
2753 {
2754 if (g_labels[i] != NULL) {
2755 lr = &g_label_refs[i];
2756 for (; lr != NULL; lr = lr->next) {
2757 check_i(&ops[i], lr->i);
2758 ret |= scan_pushes_for_pop_r(lr->i, magic, pop_i, is_probe);
2759 if (ret < 0)
2760 return ret;
2761 }
2762 if (i > 0 && LAST_OP(i - 1))
2763 return ret;
2764 }
2765
2766 i--;
2767 if (i < 0)
2768 break;
2769
2770 if (ops[i].cc_scratch == magic)
2771 return ret;
2772 ops[i].cc_scratch = magic;
2773
2774 po = &ops[i];
2775 if (po->op == OP_CALL)
2776 return -1;
2777 if ((po->flags & (OPF_TAIL|OPF_RSAVE)) || po->op == OP_POP)
2778 return -1;
2779
2780 if (po->op == OP_PUSH)
2781 {
2782 if (po->datap != NULL)
2783 return -1;
2784 if (po->operand[0].type == OPT_REG && po->operand[0].reg == reg)
2785 // leave this case for reg save/restore handlers
2786 return -1;
2787
2788 if (!is_probe) {
2789 po->flags |= OPF_PPUSH | OPF_DONE;
2790 po->datap = &ops[pop_i];
2791 }
2792 return 1;
2793 }
2794 }
2795
2796 return -1;
2797}
2798
2799static void scan_pushes_for_pop(int i, int opcnt, int *regmask_pp)
2800{
2801 int magic = i + opcnt * 14;
2802 int ret;
2803
2804 ret = scan_pushes_for_pop_r(i, magic, i, 1);
2805 if (ret == 1) {
2806 ret = check_path_branches(opcnt, magic);
2807 if (ret == 1) {
2808 ops[i].flags |= OPF_PPUSH | OPF_DONE;
2809 *regmask_pp |= 1 << ops[i].operand[0].reg;
2810 scan_pushes_for_pop_r(i, magic + 1, i, 0);
2811 }
2812 }
2813}
2814
2815static void scan_propagate_df(int i, int opcnt)
2816{
2817 struct parsed_op *po = &ops[i];
2818 int j;
2819
2820 for (; i < opcnt; i++) {
2821 po = &ops[i];
2822 if (po->flags & OPF_DF)
2823 return; // already resolved
2824 po->flags |= OPF_DF;
2825
2826 if (po->op == OP_CALL)
2827 ferr(po, "call with DF set?\n");
2828
2829 if (po->flags & OPF_JMP) {
2830 if (po->btj != NULL) {
2831 // jumptable
2832 for (j = 0; j < po->btj->count; j++) {
2833 check_i(po, po->btj->d[j].bt_i);
2834 scan_propagate_df(po->btj->d[j].bt_i, opcnt);
2835 }
2836 return;
2837 }
2838
2839 if (po->flags & OPF_RMD)
2840 continue;
2841 check_i(po, po->bt_i);
2842 if (po->flags & OPF_CJMP)
2843 scan_propagate_df(po->bt_i, opcnt);
2844 else
2845 i = po->bt_i - 1;
2846 continue;
2847 }
2848
2849 if (po->flags & OPF_TAIL)
2850 break;
2851
2852 if (po->op == OP_CLD) {
2853 po->flags |= OPF_RMD | OPF_DONE;
2854 return;
2855 }
2856 }
2857
2858 ferr(po, "missing DF clear?\n");
2859}
2860
2861// is operand 'opr' referenced by parsed_op 'po'?
2862static int is_opr_referenced(const struct parsed_opr *opr,
2863 const struct parsed_op *po)
2864{
2865 int i, mask;
2866
2867 if (opr->type == OPT_REG) {
2868 mask = po->regmask_dst | po->regmask_src;
2869 if (po->op == OP_CALL)
2870 mask |= (1 << xAX) | (1 << xCX) | (1 << xDX);
2871 if ((1 << opr->reg) & mask)
2872 return 1;
2873 else
2874 return 0;
2875 }
2876
2877 for (i = 0; i < po->operand_cnt; i++)
2878 if (IS(po->operand[0].name, opr->name))
2879 return 1;
2880
2881 return 0;
2882}
2883
2884// is operand 'opr' read by parsed_op 'po'?
2885static int is_opr_read(const struct parsed_opr *opr,
2886 const struct parsed_op *po)
2887{
2888 if (opr->type == OPT_REG) {
2889 if (po->regmask_src & (1 << opr->reg))
2890 return 1;
2891 else
2892 return 0;
2893 }
2894
2895 // yes I'm lazy
2896 return 0;
2897}
2898
2899// is operand 'opr' modified by parsed_op 'po'?
2900static int is_opr_modified(const struct parsed_opr *opr,
2901 const struct parsed_op *po)
2902{
2903 int mask;
2904
2905 if (opr->type == OPT_REG) {
2906 if (po->op == OP_CALL) {
2907 mask = po->regmask_dst;
2908 mask |= (1 << xAX) | (1 << xCX) | (1 << xDX); // ?
2909 if (mask & (1 << opr->reg))
2910 return 1;
2911 else
2912 return 0;
2913 }
2914
2915 if (po->regmask_dst & (1 << opr->reg))
2916 return 1;
2917 else
2918 return 0;
2919 }
2920
2921 return IS(po->operand[0].name, opr->name);
2922}
2923
2924// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
2925static int is_any_opr_modified(const struct parsed_op *po_test,
2926 const struct parsed_op *po, int c_mode)
2927{
2928 int mask;
2929 int i;
2930
2931 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2932 return 0;
2933
2934 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2935 return 0;
2936
2937 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
2938 return 1;
2939
2940 // in reality, it can wreck any register, but in decompiled C
2941 // version it can only overwrite eax or edx:eax
2942 mask = (1 << xAX) | (1 << xDX);
2943 if (!c_mode)
2944 mask |= 1 << xCX;
2945
2946 if (po->op == OP_CALL
2947 && ((po_test->regmask_src | po_test->regmask_dst) & mask))
2948 return 1;
2949
2950 for (i = 0; i < po_test->operand_cnt; i++)
2951 if (IS(po_test->operand[i].name, po->operand[0].name))
2952 return 1;
2953
2954 return 0;
2955}
2956
2957// scan for any po_test operand modification in range given
2958static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt,
2959 int c_mode)
2960{
2961 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2962 return -1;
2963
2964 for (; i < opcnt; i++) {
2965 if (is_any_opr_modified(po_test, &ops[i], c_mode))
2966 return i;
2967 }
2968
2969 return -1;
2970}
2971
2972// scan for po_test operand[0] modification in range given
2973static int scan_for_mod_opr0(struct parsed_op *po_test,
2974 int i, int opcnt)
2975{
2976 for (; i < opcnt; i++) {
2977 if (is_opr_modified(&po_test->operand[0], &ops[i]))
2978 return i;
2979 }
2980
2981 return -1;
2982}
2983
2984static int scan_for_flag_set(int i, int magic, int *branched,
2985 int *setters, int *setter_cnt)
2986{
2987 struct label_ref *lr;
2988 int ret;
2989
2990 while (i >= 0) {
2991 if (ops[i].cc_scratch == magic) {
2992 // is this a problem?
2993 //ferr(&ops[i], "%s looped\n", __func__);
2994 return 0;
2995 }
2996 ops[i].cc_scratch = magic;
2997
2998 if (g_labels[i] != NULL) {
2999 *branched = 1;
3000
3001 lr = &g_label_refs[i];
3002 for (; lr->next; lr = lr->next) {
3003 check_i(&ops[i], lr->i);
3004 ret = scan_for_flag_set(lr->i, magic,
3005 branched, setters, setter_cnt);
3006 if (ret < 0)
3007 return ret;
3008 }
3009
3010 check_i(&ops[i], lr->i);
3011 if (i > 0 && LAST_OP(i - 1)) {
3012 i = lr->i;
3013 continue;
3014 }
3015 ret = scan_for_flag_set(lr->i, magic,
3016 branched, setters, setter_cnt);
3017 if (ret < 0)
3018 return ret;
3019 }
3020 i--;
3021
3022 if (ops[i].flags & OPF_FLAGS) {
3023 setters[*setter_cnt] = i;
3024 (*setter_cnt)++;
3025 return 0;
3026 }
3027
3028 if ((ops[i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP)
3029 return -1;
3030 }
3031
3032 return -1;
3033}
3034
3035// scan back for cdq, if anything modifies edx, fail
3036static int scan_for_cdq_edx(int i)
3037{
3038 while (i >= 0) {
3039 if (g_labels[i] != NULL) {
3040 if (g_label_refs[i].next != NULL)
3041 return -1;
3042 if (i > 0 && LAST_OP(i - 1)) {
3043 i = g_label_refs[i].i;
3044 continue;
3045 }
3046 return -1;
3047 }
3048 i--;
3049
3050 if (ops[i].op == OP_CDQ)
3051 return i;
3052
3053 if (ops[i].regmask_dst & (1 << xDX))
3054 return -1;
3055 }
3056
3057 return -1;
3058}
3059
3060static int scan_for_reg_clear(int i, int reg)
3061{
3062 while (i >= 0) {
3063 if (g_labels[i] != NULL) {
3064 if (g_label_refs[i].next != NULL)
3065 return -1;
3066 if (i > 0 && LAST_OP(i - 1)) {
3067 i = g_label_refs[i].i;
3068 continue;
3069 }
3070 return -1;
3071 }
3072 i--;
3073
3074 if (ops[i].op == OP_XOR
3075 && ops[i].operand[0].lmod == OPLM_DWORD
3076 && ops[i].operand[0].reg == ops[i].operand[1].reg
3077 && ops[i].operand[0].reg == reg)
3078 return i;
3079
3080 if (ops[i].regmask_dst & (1 << reg))
3081 return -1;
3082 }
3083
3084 return -1;
3085}
3086
3087static void patch_esp_adjust(struct parsed_op *po, int adj)
3088{
3089 ferr_assert(po, po->op == OP_ADD);
3090 ferr_assert(po, IS(opr_name(po, 0), "esp"));
3091 ferr_assert(po, po->operand[1].type == OPT_CONST);
3092
3093 // this is a bit of a hack, but deals with use of
3094 // single adj for multiple calls
3095 po->operand[1].val -= adj;
3096 po->flags |= OPF_RMD;
3097 if (po->operand[1].val == 0)
3098 po->flags |= OPF_DONE;
3099 ferr_assert(po, (int)po->operand[1].val >= 0);
3100}
3101
3102// scan for positive, constant esp adjust
3103// multipath case is preliminary
3104static int scan_for_esp_adjust(int i, int opcnt,
3105 int adj_expect, int *adj, int *is_multipath, int do_update)
3106{
3107 int adj_expect_unknown = 0;
3108 struct parsed_op *po;
3109 int first_pop = -1;
3110 int adj_best = 0;
3111
3112 *adj = *is_multipath = 0;
3113 if (adj_expect < 0) {
3114 adj_expect_unknown = 1;
3115 adj_expect = 32 * 4; // enough?
3116 }
3117
3118 for (; i < opcnt && *adj < adj_expect; i++) {
3119 if (g_labels[i] != NULL)
3120 *is_multipath = 1;
3121
3122 po = &ops[i];
3123 if (po->flags & OPF_DONE)
3124 continue;
3125
3126 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
3127 if (po->operand[1].type != OPT_CONST)
3128 ferr(&ops[i], "non-const esp adjust?\n");
3129 *adj += po->operand[1].val;
3130 if (*adj & 3)
3131 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
3132 if (do_update) {
3133 if (!*is_multipath)
3134 patch_esp_adjust(po, adj_expect);
3135 else
3136 po->flags |= OPF_RMD;
3137 }
3138 return i;
3139 }
3140 else if (po->op == OP_PUSH) {
3141 //if (first_pop == -1)
3142 // first_pop = -2; // none
3143 *adj -= lmod_bytes(po, po->operand[0].lmod);
3144 }
3145 else if (po->op == OP_POP) {
3146 if (!(po->flags & OPF_DONE)) {
3147 // seems like msvc only uses 'pop ecx' for stack realignment..
3148 if (po->operand[0].type != OPT_REG || po->operand[0].reg != xCX)
3149 break;
3150 if (first_pop == -1 && *adj >= 0)
3151 first_pop = i;
3152 }
3153 if (do_update && *adj >= 0) {
3154 po->flags |= OPF_RMD;
3155 if (!*is_multipath)
3156 po->flags |= OPF_DONE | OPF_NOREGS;
3157 }
3158
3159 *adj += lmod_bytes(po, po->operand[0].lmod);
3160 if (*adj > adj_best)
3161 adj_best = *adj;
3162 }
3163 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
3164 if (po->op == OP_JMP && po->btj == NULL) {
3165 if (po->bt_i <= i)
3166 break;
3167 i = po->bt_i - 1;
3168 continue;
3169 }
3170 if (po->op != OP_CALL)
3171 break;
3172 if (po->operand[0].type != OPT_LABEL)
3173 break;
3174 if (po->pp != NULL && po->pp->is_stdcall)
3175 break;
3176 if (adj_expect_unknown && first_pop >= 0)
3177 break;
3178 // assume it's another cdecl call
3179 }
3180 }
3181
3182 if (first_pop >= 0) {
3183 // probably only 'pop ecx' was used
3184 *adj = adj_best;
3185 return first_pop;
3186 }
3187
3188 return -1;
3189}
3190
3191static void scan_fwd_set_flags(int i, int opcnt, int magic, int flags)
3192{
3193 struct parsed_op *po;
3194 int j;
3195
3196 if (i < 0)
3197 ferr(ops, "%s: followed bad branch?\n", __func__);
3198
3199 for (; i < opcnt; i++) {
3200 po = &ops[i];
3201 if (po->cc_scratch == magic)
3202 return;
3203 po->cc_scratch = magic;
3204 po->flags |= flags;
3205
3206 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
3207 if (po->btj != NULL) {
3208 // jumptable
3209 for (j = 0; j < po->btj->count; j++)
3210 scan_fwd_set_flags(po->btj->d[j].bt_i, opcnt, magic, flags);
3211 return;
3212 }
3213
3214 scan_fwd_set_flags(po->bt_i, opcnt, magic, flags);
3215 if (!(po->flags & OPF_CJMP))
3216 return;
3217 }
3218 if (po->flags & OPF_TAIL)
3219 return;
3220 }
3221}
3222
3223static const struct parsed_proto *try_recover_pp(
3224 struct parsed_op *po, const struct parsed_opr *opr, int *search_instead)
3225{
3226 const struct parsed_proto *pp = NULL;
3227 char buf[256];
3228 char *p;
3229
3230 // maybe an arg of g_func?
3231 if (opr->type == OPT_REGMEM && is_stack_access(po, opr))
3232 {
3233 char ofs_reg[16] = { 0, };
3234 int arg, arg_s, arg_i;
3235 int stack_ra = 0;
3236 int offset = 0;
3237
3238 if (g_header_mode)
3239 return NULL;
3240
3241 parse_stack_access(po, opr->name, ofs_reg,
3242 &offset, &stack_ra, NULL, 0);
3243 if (ofs_reg[0] != 0)
3244 ferr(po, "offset reg on arg access?\n");
3245 if (offset <= stack_ra) {
3246 // search who set the stack var instead
3247 if (search_instead != NULL)
3248 *search_instead = 1;
3249 return NULL;
3250 }
3251
3252 arg_i = (offset - stack_ra - 4) / 4;
3253 for (arg = arg_s = 0; arg < g_func_pp->argc; arg++) {
3254 if (g_func_pp->arg[arg].reg != NULL)
3255 continue;
3256 if (arg_s == arg_i)
3257 break;
3258 arg_s++;
3259 }
3260 if (arg == g_func_pp->argc)
3261 ferr(po, "stack arg %d not in prototype?\n", arg_i);
3262
3263 pp = g_func_pp->arg[arg].fptr;
3264 if (pp == NULL)
3265 ferr(po, "icall sa: arg%d is not a fptr?\n", arg + 1);
3266 check_func_pp(po, pp, "icall arg");
3267 }
3268 else if (opr->type == OPT_REGMEM && strchr(opr->name + 1, '[')) {
3269 // label[index]
3270 p = strchr(opr->name + 1, '[');
3271 memcpy(buf, opr->name, p - opr->name);
3272 buf[p - opr->name] = 0;
3273 pp = proto_parse(g_fhdr, buf, g_quiet_pp);
3274 }
3275 else if (opr->type == OPT_OFFSET || opr->type == OPT_LABEL) {
3276 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
3277 if (pp == NULL) {
3278 if (!g_header_mode)
3279 ferr(po, "proto_parse failed for icall to '%s'\n", opr->name);
3280 }
3281 else
3282 check_func_pp(po, pp, "reg-fptr ref");
3283 }
3284
3285 return pp;
3286}
3287
3288static void scan_for_call_type(int i, const struct parsed_opr *opr,
3289 int magic, const struct parsed_proto **pp_found, int *pp_i,
3290 int *multi)
3291{
3292 const struct parsed_proto *pp = NULL;
3293 struct parsed_op *po;
3294 struct label_ref *lr;
3295
3296 ops[i].cc_scratch = magic;
3297
3298 while (1) {
3299 if (g_labels[i] != NULL) {
3300 lr = &g_label_refs[i];
3301 for (; lr != NULL; lr = lr->next) {
3302 check_i(&ops[i], lr->i);
3303 scan_for_call_type(lr->i, opr, magic, pp_found, pp_i, multi);
3304 }
3305 if (i > 0 && LAST_OP(i - 1))
3306 return;
3307 }
3308
3309 i--;
3310 if (i < 0)
3311 break;
3312
3313 if (ops[i].cc_scratch == magic)
3314 return;
3315 ops[i].cc_scratch = magic;
3316
3317 if (!(ops[i].flags & OPF_DATA))
3318 continue;
3319 if (!is_opr_modified(opr, &ops[i]))
3320 continue;
3321 if (ops[i].op != OP_MOV && ops[i].op != OP_LEA) {
3322 // most probably trashed by some processing
3323 *pp_found = NULL;
3324 return;
3325 }
3326
3327 opr = &ops[i].operand[1];
3328 if (opr->type != OPT_REG)
3329 break;
3330 }
3331
3332 po = (i >= 0) ? &ops[i] : ops;
3333
3334 if (i < 0) {
3335 // reached the top - can only be an arg-reg
3336 if (opr->type != OPT_REG || g_func_pp == NULL)
3337 return;
3338
3339 for (i = 0; i < g_func_pp->argc; i++) {
3340 if (g_func_pp->arg[i].reg == NULL)
3341 continue;
3342 if (IS(opr->name, g_func_pp->arg[i].reg))
3343 break;
3344 }
3345 if (i == g_func_pp->argc)
3346 return;
3347 pp = g_func_pp->arg[i].fptr;
3348 if (pp == NULL)
3349 ferr(po, "icall: arg%d (%s) is not a fptr?\n",
3350 i + 1, g_func_pp->arg[i].reg);
3351 check_func_pp(po, pp, "icall reg-arg");
3352 }
3353 else
3354 pp = try_recover_pp(po, opr, NULL);
3355
3356 if (*pp_found != NULL && pp != NULL && *pp_found != pp) {
3357 if (!IS((*pp_found)->ret_type.name, pp->ret_type.name)
3358 || (*pp_found)->is_stdcall != pp->is_stdcall
3359 || (*pp_found)->is_fptr != pp->is_fptr
3360 || (*pp_found)->argc != pp->argc
3361 || (*pp_found)->argc_reg != pp->argc_reg
3362 || (*pp_found)->argc_stack != pp->argc_stack)
3363 {
3364 ferr(po, "icall: parsed_proto mismatch\n");
3365 }
3366 *multi = 1;
3367 }
3368 if (pp != NULL) {
3369 *pp_found = pp;
3370 *pp_i = po - ops;
3371 }
3372}
3373
3374static void add_label_ref(struct label_ref *lr, int op_i)
3375{
3376 struct label_ref *lr_new;
3377
3378 if (lr->i == -1) {
3379 lr->i = op_i;
3380 return;
3381 }
3382
3383 lr_new = calloc(1, sizeof(*lr_new));
3384 lr_new->i = op_i;
3385 lr_new->next = lr->next;
3386 lr->next = lr_new;
3387}
3388
3389static struct parsed_data *try_resolve_jumptab(int i, int opcnt)
3390{
3391 struct parsed_op *po = &ops[i];
3392 struct parsed_data *pd;
3393 char label[NAMELEN], *p;
3394 int len, j, l;
3395
3396 p = strchr(po->operand[0].name, '[');
3397 if (p == NULL)
3398 return NULL;
3399
3400 len = p - po->operand[0].name;
3401 strncpy(label, po->operand[0].name, len);
3402 label[len] = 0;
3403
3404 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
3405 if (IS(g_func_pd[j].label, label)) {
3406 pd = &g_func_pd[j];
3407 break;
3408 }
3409 }
3410 if (pd == NULL)
3411 //ferr(po, "label '%s' not parsed?\n", label);
3412 return NULL;
3413
3414 if (pd->type != OPT_OFFSET)
3415 ferr(po, "label '%s' with non-offset data?\n", label);
3416
3417 // find all labels, link
3418 for (j = 0; j < pd->count; j++) {
3419 for (l = 0; l < opcnt; l++) {
3420 if (g_labels[l] != NULL && IS(g_labels[l], pd->d[j].u.label)) {
3421 add_label_ref(&g_label_refs[l], i);
3422 pd->d[j].bt_i = l;
3423 break;
3424 }
3425 }
3426 }
3427
3428 return pd;
3429}
3430
3431static void clear_labels(int count)
3432{
3433 int i;
3434
3435 for (i = 0; i < count; i++) {
3436 if (g_labels[i] != NULL) {
3437 free(g_labels[i]);
3438 g_labels[i] = NULL;
3439 }
3440 }
3441}
3442
3443static int get_pp_arg_regmask_src(const struct parsed_proto *pp)
3444{
3445 int regmask = 0;
3446 int i, reg;
3447
3448 for (i = 0; i < pp->argc; i++) {
3449 if (pp->arg[i].reg != NULL) {
3450 reg = char_array_i(regs_r32,
3451 ARRAY_SIZE(regs_r32), pp->arg[i].reg);
3452 if (reg < 0)
3453 ferr(ops, "arg '%s' of func '%s' is not a reg?\n",
3454 pp->arg[i].reg, pp->name);
3455 regmask |= 1 << reg;
3456 }
3457 }
3458
3459 return regmask;
3460}
3461
3462static int get_pp_arg_regmask_dst(const struct parsed_proto *pp)
3463{
3464 int regmask = 0;
3465 int i, reg;
3466
3467 if (pp->has_retreg) {
3468 for (i = 0; i < pp->argc; i++) {
3469 if (pp->arg[i].type.is_retreg) {
3470 reg = char_array_i(regs_r32,
3471 ARRAY_SIZE(regs_r32), pp->arg[i].reg);
3472 ferr_assert(ops, reg >= 0);
3473 regmask |= 1 << reg;
3474 }
3475 }
3476 }
3477
3478 if (strstr(pp->ret_type.name, "int64"))
3479 return regmask | (1 << xAX) | (1 << xDX);
3480 if (IS(pp->ret_type.name, "float")
3481 || IS(pp->ret_type.name, "double"))
3482 {
3483 return regmask | mxST0;
3484 }
3485 if (strcasecmp(pp->ret_type.name, "void") == 0)
3486 return regmask;
3487
3488 return regmask | mxAX;
3489}
3490
3491static void resolve_branches_parse_calls(int opcnt)
3492{
3493 static const struct {
3494 const char *name;
3495 enum op_op op;
3496 unsigned int flags;
3497 unsigned int regmask_src;
3498 unsigned int regmask_dst;
3499 } pseudo_ops[] = {
3500 { "__ftol", OPP_FTOL, OPF_FPOP, mxST0, mxAX | mxDX },
3501 };
3502 const struct parsed_proto *pp_c;
3503 struct parsed_proto *pp;
3504 struct parsed_data *pd;
3505 struct parsed_op *po;
3506 const char *tmpname;
3507 int i, l;
3508 int ret;
3509
3510 for (i = 0; i < opcnt; i++)
3511 {
3512 po = &ops[i];
3513 po->bt_i = -1;
3514 po->btj = NULL;
3515
3516 if (po->datap != NULL) {
3517 pp = calloc(1, sizeof(*pp));
3518 my_assert_not(pp, NULL);
3519
3520 ret = parse_protostr(po->datap, pp);
3521 if (ret < 0)
3522 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
3523 free(po->datap);
3524 po->datap = NULL;
3525 po->pp = pp;
3526 }
3527
3528 if (po->op == OP_CALL) {
3529 pp = NULL;
3530
3531 if (po->pp != NULL)
3532 pp = po->pp;
3533 else if (po->operand[0].type == OPT_LABEL)
3534 {
3535 tmpname = opr_name(po, 0);
3536 if (IS_START(tmpname, "loc_"))
3537 ferr(po, "call to loc_*\n");
3538
3539 // convert some calls to pseudo-ops
3540 for (l = 0; l < ARRAY_SIZE(pseudo_ops); l++) {
3541 if (!IS(tmpname, pseudo_ops[l].name))
3542 continue;
3543
3544 po->op = pseudo_ops[l].op;
3545 po->operand_cnt = 0;
3546 po->regmask_src = pseudo_ops[l].regmask_src;
3547 po->regmask_dst = pseudo_ops[l].regmask_dst;
3548 po->flags = pseudo_ops[l].flags;
3549 po->flags |= po->regmask_dst ? OPF_DATA : 0;
3550 break;
3551 }
3552 if (l < ARRAY_SIZE(pseudo_ops))
3553 continue;
3554
3555 pp_c = proto_parse(g_fhdr, tmpname, g_header_mode);
3556 if (!g_header_mode && pp_c == NULL)
3557 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
3558
3559 if (pp_c != NULL) {
3560 pp = proto_clone(pp_c);
3561 my_assert_not(pp, NULL);
3562 }
3563 }
3564
3565 if (pp != NULL) {
3566 if (pp->is_fptr)
3567 check_func_pp(po, pp, "fptr var call");
3568 if (pp->is_noreturn)
3569 po->flags |= OPF_TAIL;
3570 }
3571 po->pp = pp;
3572 continue;
3573 }
3574
3575 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
3576 continue;
3577
3578 if (po->operand[0].type == OPT_REGMEM) {
3579 pd = try_resolve_jumptab(i, opcnt);
3580 if (pd == NULL)
3581 goto tailcall;
3582
3583 po->btj = pd;
3584 continue;
3585 }
3586
3587 for (l = 0; l < opcnt; l++) {
3588 if (g_labels[l] != NULL
3589 && IS(po->operand[0].name, g_labels[l]))
3590 {
3591 if (l == i + 1 && po->op == OP_JMP) {
3592 // yet another alignment type..
3593 po->flags |= OPF_RMD|OPF_DONE;
3594 break;
3595 }
3596 add_label_ref(&g_label_refs[l], i);
3597 po->bt_i = l;
3598 break;
3599 }
3600 }
3601
3602 if (po->bt_i != -1 || (po->flags & OPF_RMD))
3603 continue;
3604
3605 if (po->operand[0].type == OPT_LABEL)
3606 // assume tail call
3607 goto tailcall;
3608
3609 ferr(po, "unhandled branch\n");
3610
3611tailcall:
3612 po->op = OP_CALL;
3613 po->flags |= OPF_TAIL;
3614 if (i > 0 && ops[i - 1].op == OP_POP)
3615 po->flags |= OPF_ATAIL;
3616 i--; // reprocess
3617 }
3618}
3619
3620static void scan_prologue_epilogue(int opcnt)
3621{
3622 int ecx_push = 0, esp_sub = 0;
3623 int found;
3624 int i, j, l;
3625
3626 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
3627 && ops[1].op == OP_MOV
3628 && IS(opr_name(&ops[1], 0), "ebp")
3629 && IS(opr_name(&ops[1], 1), "esp"))
3630 {
3631 g_bp_frame = 1;
3632 ops[0].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3633 ops[1].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3634 i = 2;
3635
3636 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
3637 g_stack_fsz = opr_const(&ops[2], 1);
3638 ops[2].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3639 i++;
3640 }
3641 else {
3642 // another way msvc builds stack frame..
3643 i = 2;
3644 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
3645 g_stack_fsz += 4;
3646 ops[i].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3647 ecx_push++;
3648 i++;
3649 }
3650 // and another way..
3651 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
3652 && ops[i].operand[1].type == OPT_CONST
3653 && ops[i + 1].op == OP_CALL
3654 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
3655 {
3656 g_stack_fsz += ops[i].operand[1].val;
3657 ops[i].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3658 i++;
3659 ops[i].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3660 i++;
3661 }
3662 }
3663
3664 found = 0;
3665 do {
3666 for (; i < opcnt; i++)
3667 if (ops[i].flags & OPF_TAIL)
3668 break;
3669 j = i - 1;
3670 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
3671 if (ops[j].bt_i != -1 || ops[j].btj != NULL)
3672 break;
3673 i--;
3674 j--;
3675 }
3676
3677 if ((ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ebp"))
3678 || ops[j].op == OP_LEAVE)
3679 {
3680 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3681 }
3682 else if (ops[i].op == OP_CALL && ops[i].pp != NULL
3683 && ops[i].pp->is_noreturn)
3684 {
3685 // on noreturn, msvc sometimes cleans stack, sometimes not
3686 i++;
3687 found = 1;
3688 continue;
3689 }
3690 else if (!(g_ida_func_attr & IDAFA_NORETURN))
3691 ferr(&ops[j], "'pop ebp' expected\n");
3692
3693 if (g_stack_fsz != 0) {
3694 if (ops[j].op == OP_LEAVE)
3695 j--;
3696 else if (ops[j].op == OP_POP
3697 && ops[j - 1].op == OP_MOV
3698 && IS(opr_name(&ops[j - 1], 0), "esp")
3699 && IS(opr_name(&ops[j - 1], 1), "ebp"))
3700 {
3701 ops[j - 1].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3702 j -= 2;
3703 }
3704 else if (!(g_ida_func_attr & IDAFA_NORETURN))
3705 {
3706 ferr(&ops[j], "esp restore expected\n");
3707 }
3708
3709 if (ecx_push && j >= 0 && ops[j].op == OP_POP
3710 && IS(opr_name(&ops[j], 0), "ecx"))
3711 {
3712 ferr(&ops[j], "unexpected ecx pop\n");
3713 }
3714 }
3715
3716 found = 1;
3717 i++;
3718 } while (i < opcnt);
3719
3720 if (!found)
3721 ferr(ops, "missing ebp epilogue\n");
3722 return;
3723 }
3724
3725 // non-bp frame
3726 i = 0;
3727 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
3728 ops[i].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3729 g_stack_fsz += 4;
3730 ecx_push++;
3731 i++;
3732 }
3733
3734 for (; i < opcnt; i++) {
3735 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
3736 break;
3737 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
3738 && ops[i].operand[1].type == OPT_CONST)
3739 {
3740 g_stack_fsz = ops[i].operand[1].val;
3741 ops[i].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3742 esp_sub = 1;
3743 break;
3744 }
3745 }
3746
3747 if (ecx_push && !esp_sub) {
3748 // could actually be args for a call..
3749 for (; i < opcnt; i++)
3750 if (ops[i].op != OP_PUSH)
3751 break;
3752
3753 if (ops[i].op == OP_CALL && ops[i].operand[0].type == OPT_LABEL) {
3754 const struct parsed_proto *pp;
3755 pp = proto_parse(g_fhdr, opr_name(&ops[i], 0), 1);
3756 j = pp ? pp->argc_stack : 0;
3757 while (i > 0 && j > 0) {
3758 i--;
3759 if (ops[i].op == OP_PUSH) {
3760 ops[i].flags &= ~(OPF_RMD | OPF_DONE | OPF_NOREGS);
3761 j--;
3762 }
3763 }
3764 if (j != 0)
3765 ferr(&ops[i], "unhandled prologue\n");
3766
3767 // recheck
3768 i = g_stack_fsz = ecx_push = 0;
3769 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
3770 if (!(ops[i].flags & OPF_RMD))
3771 break;
3772 g_stack_fsz += 4;
3773 ecx_push++;
3774 i++;
3775 }
3776 }
3777 }
3778
3779 found = 0;
3780 if (ecx_push || esp_sub)
3781 {
3782 g_sp_frame = 1;
3783
3784 i++;
3785 do {
3786 for (; i < opcnt; i++)
3787 if (ops[i].flags & OPF_TAIL)
3788 break;
3789 j = i - 1;
3790 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
3791 if (ops[j].bt_i != -1 || ops[j].btj != NULL)
3792 break;
3793 i--;
3794 j--;
3795 }
3796
3797 if (ecx_push > 0) {
3798 for (l = 0; l < ecx_push; l++) {
3799 if (ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ecx"))
3800 /* pop ecx */;
3801 else if (ops[j].op == OP_ADD
3802 && IS(opr_name(&ops[j], 0), "esp")
3803 && ops[j].operand[1].type == OPT_CONST)
3804 {
3805 /* add esp, N */
3806 l += ops[j].operand[1].val / 4 - 1;
3807 }
3808 else
3809 ferr(&ops[j], "'pop ecx' expected\n");
3810
3811 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3812 j--;
3813 }
3814 if (l != ecx_push)
3815 ferr(&ops[j], "epilogue scan failed\n");
3816
3817 found = 1;
3818 }
3819
3820 if (esp_sub) {
3821 if (ops[j].op != OP_ADD
3822 || !IS(opr_name(&ops[j], 0), "esp")
3823 || ops[j].operand[1].type != OPT_CONST
3824 || ops[j].operand[1].val != g_stack_fsz)
3825 ferr(&ops[j], "'add esp' expected\n");
3826
3827 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3828 ops[j].operand[1].val = 0; // hack for stack arg scanner
3829 found = 1;
3830 }
3831
3832 i++;
3833 } while (i < opcnt);
3834
3835 if (!found)
3836 ferr(ops, "missing esp epilogue\n");
3837 }
3838}
3839
3840// find an instruction that changed opr before i op
3841// *op_i must be set to -1 by the caller
3842// *is_caller is set to 1 if one source is determined to be g_func arg
3843// returns 1 if found, *op_i is then set to origin
3844// returns -1 if multiple origins are found
3845static int resolve_origin(int i, const struct parsed_opr *opr,
3846 int magic, int *op_i, int *is_caller)
3847{
3848 struct label_ref *lr;
3849 int ret = 0;
3850
3851 if (ops[i].cc_scratch == magic)
3852 return 0;
3853 ops[i].cc_scratch = magic;
3854
3855 while (1) {
3856 if (g_labels[i] != NULL) {
3857 lr = &g_label_refs[i];
3858 for (; lr != NULL; lr = lr->next) {
3859 check_i(&ops[i], lr->i);
3860 ret |= resolve_origin(lr->i, opr, magic, op_i, is_caller);
3861 }
3862 if (i > 0 && LAST_OP(i - 1))
3863 return ret;
3864 }
3865
3866 i--;
3867 if (i < 0) {
3868 if (is_caller != NULL)
3869 *is_caller = 1;
3870 return -1;
3871 }
3872
3873 if (ops[i].cc_scratch == magic)
3874 return ret;
3875 ops[i].cc_scratch = magic;
3876
3877 if (!(ops[i].flags & OPF_DATA))
3878 continue;
3879 if (!is_opr_modified(opr, &ops[i]))
3880 continue;
3881
3882 if (*op_i >= 0) {
3883 if (*op_i == i)
3884 return ret | 1;
3885
3886 // XXX: could check if the other op does the same
3887 return -1;
3888 }
3889
3890 *op_i = i;
3891 return ret | 1;
3892 }
3893}
3894
3895// find an instruction that previously referenced opr
3896// if multiple results are found - fail
3897// *op_i must be set to -1 by the caller
3898// returns 1 if found, *op_i is then set to referencer insn
3899static int resolve_last_ref(int i, const struct parsed_opr *opr,
3900 int magic, int *op_i)
3901{
3902 struct label_ref *lr;
3903 int ret = 0;
3904
3905 if (ops[i].cc_scratch == magic)
3906 return 0;
3907 ops[i].cc_scratch = magic;
3908
3909 while (1) {
3910 if (g_labels[i] != NULL) {
3911 lr = &g_label_refs[i];
3912 for (; lr != NULL; lr = lr->next) {
3913 check_i(&ops[i], lr->i);
3914 ret |= resolve_last_ref(lr->i, opr, magic, op_i);
3915 }
3916 if (i > 0 && LAST_OP(i - 1))
3917 return ret;
3918 }
3919
3920 i--;
3921 if (i < 0)
3922 return -1;
3923
3924 if (ops[i].cc_scratch == magic)
3925 return 0;
3926 ops[i].cc_scratch = magic;
3927
3928 if (!is_opr_referenced(opr, &ops[i]))
3929 continue;
3930
3931 if (*op_i >= 0)
3932 return -1;
3933
3934 *op_i = i;
3935 return 1;
3936 }
3937}
3938
3939// find next instruction that reads opr
3940// *op_i must be set to -1 by the caller
3941// on return, *op_i is set to first referencer insn
3942// returns 1 if exactly 1 referencer is found
3943static int find_next_read(int i, int opcnt,
3944 const struct parsed_opr *opr, int magic, int *op_i)
3945{
3946 struct parsed_op *po;
3947 int j, ret = 0;
3948
3949 for (; i < opcnt; i++)
3950 {
3951 if (ops[i].cc_scratch == magic)
3952 return ret;
3953 ops[i].cc_scratch = magic;
3954
3955 po = &ops[i];
3956 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
3957 if (po->btj != NULL) {
3958 // jumptable
3959 for (j = 0; j < po->btj->count; j++) {
3960 check_i(po, po->btj->d[j].bt_i);
3961 ret |= find_next_read(po->btj->d[j].bt_i, opcnt, opr,
3962 magic, op_i);
3963 }
3964 return ret;
3965 }
3966
3967 if (po->flags & OPF_RMD)
3968 continue;
3969 check_i(po, po->bt_i);
3970 if (po->flags & OPF_CJMP) {
3971 ret |= find_next_read(po->bt_i, opcnt, opr, magic, op_i);
3972 if (ret < 0)
3973 return ret;
3974 }
3975 else
3976 i = po->bt_i - 1;
3977 continue;
3978 }
3979
3980 if (!is_opr_read(opr, po)) {
3981 if (is_opr_modified(opr, po)
3982 && (po->op == OP_CALL
3983 || ((po->flags & OPF_DATA)
3984 && po->operand[0].lmod == OPLM_DWORD)))
3985 {
3986 // it's overwritten
3987 return ret;
3988 }
3989 if (po->flags & OPF_TAIL)
3990 return ret;
3991 continue;
3992 }
3993
3994 if (*op_i >= 0)
3995 return -1;
3996
3997 *op_i = i;
3998 return 1;
3999 }
4000
4001 return 0;
4002}
4003
4004static int try_resolve_const(int i, const struct parsed_opr *opr,
4005 int magic, unsigned int *val)
4006{
4007 int s_i = -1;
4008 int ret;
4009
4010 ret = resolve_origin(i, opr, magic, &s_i, NULL);
4011 if (ret == 1) {
4012 i = s_i;
4013 if (ops[i].op != OP_MOV && ops[i].operand[1].type != OPT_CONST)
4014 return -1;
4015
4016 *val = ops[i].operand[1].val;
4017 return 1;
4018 }
4019
4020 return -1;
4021}
4022
4023static const struct parsed_proto *resolve_icall(int i, int opcnt,
4024 int *pp_i, int *multi_src)
4025{
4026 const struct parsed_proto *pp = NULL;
4027 int search_advice = 0;
4028 int offset = -1;
4029 char name[256];
4030 char s_reg[4];
4031 int reg, len;
4032 int ret;
4033
4034 *multi_src = 0;
4035 *pp_i = -1;
4036
4037 switch (ops[i].operand[0].type) {
4038 case OPT_REGMEM:
4039 // try to resolve struct member calls
4040 ret = sscanf(ops[i].operand[0].name, "%3s+%x%n",
4041 s_reg, &offset, &len);
4042 if (ret == 2 && len == strlen(ops[i].operand[0].name))
4043 {
4044 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s_reg);
4045 if (reg >= 0) {
4046 struct parsed_opr opr = OPR_INIT(OPT_REG, OPLM_DWORD, reg);
4047 int j = -1;
4048 ret = resolve_origin(i, &opr, i + opcnt * 19, &j, NULL);
4049 if (ret != 1)
4050 break;
4051 if (ops[j].op == OP_MOV && ops[j].operand[1].type == OPT_REGMEM
4052 && ops[j].operand[0].lmod == OPLM_DWORD
4053 && ops[j].pp == NULL) // no hint
4054 {
4055 // allow one simple dereference (directx)
4056 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32),
4057 ops[j].operand[1].name);
4058 if (reg < 0)
4059 break;
4060 struct parsed_opr opr2 = OPR_INIT(OPT_REG, OPLM_DWORD, reg);
4061 int k = -1;
4062 ret = resolve_origin(j, &opr2, j + opcnt * 19, &k, NULL);
4063 if (ret != 1)
4064 break;
4065 j = k;
4066 }
4067 if (ops[j].op != OP_MOV)
4068 break;
4069 if (ops[j].operand[0].lmod != OPLM_DWORD)
4070 break;
4071 if (ops[j].pp != NULL) {
4072 // type hint in asm
4073 pp = ops[j].pp;
4074 }
4075 else if (ops[j].operand[1].type == OPT_REGMEM) {
4076 // allow 'hello[ecx]' - assume array of same type items
4077 ret = sscanf(ops[j].operand[1].name, "%[^[][e%2s]",
4078 name, s_reg);
4079 if (ret != 2)
4080 break;
4081 pp = proto_parse(g_fhdr, name, g_quiet_pp);
4082 }
4083 else if (ops[j].operand[1].type == OPT_LABEL)
4084 pp = proto_parse(g_fhdr, ops[j].operand[1].name, g_quiet_pp);
4085 else
4086 break;
4087 if (pp == NULL)
4088 break;
4089 if (pp->is_func || pp->is_fptr || !pp->type.is_struct) {
4090 pp = NULL;
4091 break;
4092 }
4093 pp = proto_lookup_struct(g_fhdr, pp->type.name, offset);
4094 }
4095 break;
4096 }
4097 // fallthrough
4098 case OPT_LABEL:
4099 case OPT_OFFSET:
4100 pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice);
4101 if (!search_advice)
4102 break;
4103 // fallthrough
4104 default:
4105 scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp,
4106 pp_i, multi_src);
4107 break;
4108 }
4109
4110 return pp;
4111}
4112
4113static struct parsed_proto *process_call_early(int i, int opcnt,
4114 int *adj_i)
4115{
4116 struct parsed_op *po = &ops[i];
4117 struct parsed_proto *pp;
4118 int multipath = 0;
4119 int adj = 0;
4120 int j, ret;
4121
4122 pp = po->pp;
4123 if (pp == NULL || pp->is_vararg || pp->argc_reg != 0)
4124 // leave for later
4125 return NULL;
4126
4127 // look for and make use of esp adjust
4128 *adj_i = ret = -1;
4129 if (!pp->is_stdcall && pp->argc_stack > 0)
4130 ret = scan_for_esp_adjust(i + 1, opcnt,
4131 pp->argc_stack * 4, &adj, &multipath, 0);
4132 if (ret >= 0) {
4133 if (pp->argc_stack > adj / 4)
4134 return NULL;
4135 if (multipath)
4136 return NULL;
4137 if (ops[ret].op == OP_POP) {
4138 for (j = 1; j < adj / 4; j++) {
4139 if (ops[ret + j].op != OP_POP
4140 || ops[ret + j].operand[0].reg != xCX)
4141 {
4142 return NULL;
4143 }
4144 }
4145 }
4146 }
4147
4148 *adj_i = ret;
4149 return pp;
4150}
4151
4152static struct parsed_proto *process_call(int i, int opcnt)
4153{
4154 struct parsed_op *po = &ops[i];
4155 const struct parsed_proto *pp_c;
4156 struct parsed_proto *pp;
4157 const char *tmpname;
4158 int call_i = -1, ref_i = -1;
4159 int adj = 0, multipath = 0;
4160 int ret, arg;
4161
4162 tmpname = opr_name(po, 0);
4163 pp = po->pp;
4164 if (pp == NULL)
4165 {
4166 // indirect call
4167 pp_c = resolve_icall(i, opcnt, &call_i, &multipath);
4168 if (pp_c != NULL) {
4169 if (!pp_c->is_func && !pp_c->is_fptr)
4170 ferr(po, "call to non-func: %s\n", pp_c->name);
4171 pp = proto_clone(pp_c);
4172 my_assert_not(pp, NULL);
4173 if (multipath)
4174 // not resolved just to single func
4175 pp->is_fptr = 1;
4176
4177 switch (po->operand[0].type) {
4178 case OPT_REG:
4179 // we resolved this call and no longer need the register
4180 po->regmask_src &= ~(1 << po->operand[0].reg);
4181
4182 if (!multipath && i != call_i && ops[call_i].op == OP_MOV
4183 && ops[call_i].operand[1].type == OPT_LABEL)
4184 {
4185 // no other source users?
4186 ret = resolve_last_ref(i, &po->operand[0], i + opcnt * 10,
4187 &ref_i);
4188 if (ret == 1 && call_i == ref_i) {
4189 // and nothing uses it after us?
4190 ref_i = -1;
4191 find_next_read(i + 1, opcnt, &po->operand[0],
4192 i + opcnt * 11, &ref_i);
4193 if (ref_i == -1)
4194 // then also don't need the source mov
4195 ops[call_i].flags |= OPF_RMD | OPF_NOREGS;
4196 }
4197 }
4198 break;
4199 case OPT_REGMEM:
4200 pp->is_fptr = 1;
4201 break;
4202 default:
4203 break;
4204 }
4205 }
4206 if (pp == NULL) {
4207 pp = calloc(1, sizeof(*pp));
4208 my_assert_not(pp, NULL);
4209
4210 pp->is_fptr = 1;
4211 ret = scan_for_esp_adjust(i + 1, opcnt,
4212 -1, &adj, &multipath, 0);
4213 if (ret < 0 || adj < 0) {
4214 if (!g_allow_regfunc)
4215 ferr(po, "non-__cdecl indirect call unhandled yet\n");
4216 pp->is_unresolved = 1;
4217 adj = 0;
4218 }
4219 adj /= 4;
4220 if (adj > ARRAY_SIZE(pp->arg))
4221 ferr(po, "esp adjust too large: %d\n", adj);
4222 pp->ret_type.name = strdup("int");
4223 pp->argc = pp->argc_stack = adj;
4224 for (arg = 0; arg < pp->argc; arg++)
4225 pp->arg[arg].type.name = strdup("int");
4226 }
4227 po->pp = pp;
4228 }
4229
4230 // look for and make use of esp adjust
4231 multipath = 0;
4232 ret = -1;
4233 if (!pp->is_stdcall && pp->argc_stack > 0) {
4234 int adj_expect = pp->is_vararg ? -1 : pp->argc_stack * 4;
4235 ret = scan_for_esp_adjust(i + 1, opcnt,
4236 adj_expect, &adj, &multipath, 0);
4237 }
4238 if (ret >= 0) {
4239 if (pp->is_vararg) {
4240 if (adj / 4 < pp->argc_stack) {
4241 fnote(po, "(this call)\n");
4242 ferr(&ops[ret], "esp adjust is too small: %x < %x\n",
4243 adj, pp->argc_stack * 4);
4244 }
4245 // modify pp to make it have varargs as normal args
4246 arg = pp->argc;
4247 pp->argc += adj / 4 - pp->argc_stack;
4248 for (; arg < pp->argc; arg++) {
4249 pp->arg[arg].type.name = strdup("int");
4250 pp->argc_stack++;
4251 }
4252 if (pp->argc > ARRAY_SIZE(pp->arg))
4253 ferr(po, "too many args for '%s'\n", tmpname);
4254 }
4255 if (pp->argc_stack > adj / 4) {
4256 fnote(po, "(this call)\n");
4257 ferr(&ops[ret], "stack tracking failed for '%s': %x %x\n",
4258 tmpname, pp->argc_stack * 4, adj);
4259 }
4260
4261 scan_for_esp_adjust(i + 1, opcnt,
4262 pp->argc_stack * 4, &adj, &multipath, 1);
4263 }
4264 else if (pp->is_vararg)
4265 ferr(po, "missing esp_adjust for vararg func '%s'\n",
4266 pp->name);
4267
4268 return pp;
4269}
4270
4271static int collect_call_args_early(struct parsed_op *po, int i,
4272 struct parsed_proto *pp, int *regmask)
4273{
4274 int arg, ret;
4275 int j;
4276
4277 for (arg = 0; arg < pp->argc; arg++)
4278 if (pp->arg[arg].reg == NULL)
4279 break;
4280
4281 // first see if it can be easily done
4282 for (j = i; j > 0 && arg < pp->argc; )
4283 {
4284 if (g_labels[j] != NULL)
4285 return -1;
4286 j--;
4287
4288 if (ops[j].op == OP_CALL)
4289 return -1;
4290 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP)
4291 return -1;
4292 else if (ops[j].op == OP_POP)
4293 return -1;
4294 else if (ops[j].flags & OPF_CJMP)
4295 return -1;
4296 else if (ops[j].op == OP_PUSH) {
4297 if (ops[j].flags & (OPF_FARG|OPF_FARGNR))
4298 return -1;
4299 ret = scan_for_mod(&ops[j], j + 1, i, 1);
4300 if (ret >= 0)
4301 return -1;
4302
4303 if (pp->arg[arg].type.is_va_list)
4304 return -1;
4305
4306 // next arg
4307 for (arg++; arg < pp->argc; arg++)
4308 if (pp->arg[arg].reg == NULL)
4309 break;
4310 }
4311 }
4312
4313 if (arg < pp->argc)
4314 return -1;
4315
4316 // now do it
4317 for (arg = 0; arg < pp->argc; arg++)
4318 if (pp->arg[arg].reg == NULL)
4319 break;
4320
4321 for (j = i; j > 0 && arg < pp->argc; )
4322 {
4323 j--;
4324
4325 if (ops[j].op == OP_PUSH)
4326 {
4327 ops[j].p_argnext = -1;
4328 ferr_assert(&ops[j], pp->arg[arg].datap == NULL);
4329 pp->arg[arg].datap = &ops[j];
4330
4331 if (ops[j].operand[0].type == OPT_REG)
4332 *regmask |= 1 << ops[j].operand[0].reg;
4333
4334 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_FARGNR | OPF_FARG;
4335 ops[j].flags &= ~OPF_RSAVE;
4336
4337 // next arg
4338 for (arg++; arg < pp->argc; arg++)
4339 if (pp->arg[arg].reg == NULL)
4340 break;
4341 }
4342 }
4343
4344 return 0;
4345}
4346
4347static int collect_call_args_r(struct parsed_op *po, int i,
4348 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
4349 int *arg_grp, int arg, int magic, int need_op_saving, int may_reuse)
4350{
4351 struct parsed_proto *pp_tmp;
4352 struct parsed_op *po_tmp;
4353 struct label_ref *lr;
4354 int need_to_save_current;
4355 int arg_grp_current = 0;
4356 int save_args_seen = 0;
4357 int save_args;
4358 int ret = 0;
4359 int reg;
4360 char buf[32];
4361 int j, k;
4362
4363 if (i < 0) {
4364 ferr(po, "dead label encountered\n");
4365 return -1;
4366 }
4367
4368 for (; arg < pp->argc; arg++)
4369 if (pp->arg[arg].reg == NULL)
4370 break;
4371 magic = (magic & 0xffffff) | (arg << 24);
4372
4373 for (j = i; j >= 0 && (arg < pp->argc || pp->is_unresolved); )
4374 {
4375 if (((ops[j].cc_scratch ^ magic) & 0xffffff) == 0) {
4376 if (ops[j].cc_scratch != magic) {
4377 ferr(&ops[j], "arg collect hit same path with diff args for %s\n",
4378 pp->name);
4379 return -1;
4380 }
4381 // ok: have already been here
4382 return 0;
4383 }
4384 ops[j].cc_scratch = magic;
4385
4386 if (g_labels[j] != NULL && g_label_refs[j].i != -1) {
4387 lr = &g_label_refs[j];
4388 if (lr->next != NULL)
4389 need_op_saving = 1;
4390 for (; lr->next; lr = lr->next) {
4391 check_i(&ops[j], lr->i);
4392 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
4393 may_reuse = 1;
4394 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
4395 arg_grp, arg, magic, need_op_saving, may_reuse);
4396 if (ret < 0)
4397 return ret;
4398 }
4399
4400 check_i(&ops[j], lr->i);
4401 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
4402 may_reuse = 1;
4403 if (j > 0 && LAST_OP(j - 1)) {
4404 // follow last branch in reverse
4405 j = lr->i;
4406 continue;
4407 }
4408 need_op_saving = 1;
4409 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
4410 arg_grp, arg, magic, need_op_saving, may_reuse);
4411 if (ret < 0)
4412 return ret;
4413 }
4414 j--;
4415
4416 if (ops[j].op == OP_CALL)
4417 {
4418 if (pp->is_unresolved)
4419 break;
4420
4421 pp_tmp = ops[j].pp;
4422 if (pp_tmp == NULL)
4423 ferr(po, "arg collect hit unparsed call '%s'\n",
4424 ops[j].operand[0].name);
4425 if (may_reuse && pp_tmp->argc_stack > 0)
4426 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
4427 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
4428 }
4429 // esp adjust of 0 means we collected it before
4430 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP
4431 && (ops[j].operand[1].type != OPT_CONST
4432 || ops[j].operand[1].val != 0))
4433 {
4434 if (pp->is_unresolved)
4435 break;
4436
4437 fnote(po, "(this call)\n");
4438 ferr(&ops[j], "arg collect %d/%d hit esp adjust of %d\n",
4439 arg, pp->argc, ops[j].operand[1].val);
4440 }
4441 else if (ops[j].op == OP_POP && !(ops[j].flags & OPF_DONE))
4442 {
4443 if (pp->is_unresolved)
4444 break;
4445
4446 fnote(po, "(this call)\n");
4447 ferr(&ops[j], "arg collect %d/%d hit pop\n", arg, pp->argc);
4448 }
4449 else if (ops[j].flags & OPF_CJMP)
4450 {
4451 if (pp->is_unresolved)
4452 break;
4453
4454 may_reuse = 1;
4455 }
4456 else if (ops[j].op == OP_PUSH
4457 && !(ops[j].flags & (OPF_FARGNR|OPF_DONE)))
4458 {
4459 if (pp->is_unresolved && (ops[j].flags & OPF_RMD))
4460 break;
4461
4462 ops[j].p_argnext = -1;
4463 po_tmp = pp->arg[arg].datap;
4464 if (po_tmp != NULL)
4465 ops[j].p_argnext = po_tmp - ops;
4466 pp->arg[arg].datap = &ops[j];
4467
4468 need_to_save_current = 0;
4469 save_args = 0;
4470 reg = -1;
4471 if (ops[j].operand[0].type == OPT_REG)
4472 reg = ops[j].operand[0].reg;
4473
4474 if (!need_op_saving) {
4475 ret = scan_for_mod(&ops[j], j + 1, i, 1);
4476 need_to_save_current = (ret >= 0);
4477 }
4478 if (need_op_saving || need_to_save_current) {
4479 // mark this push as one that needs operand saving
4480 ops[j].flags &= ~OPF_RMD;
4481 if (ops[j].p_argnum == 0) {
4482 ops[j].p_argnum = arg + 1;
4483 save_args |= 1 << arg;
4484 }
4485 else if (ops[j].p_argnum < arg + 1) {
4486 // XXX: might kill valid var..
4487 //*save_arg_vars &= ~(1 << (ops[j].p_argnum - 1));
4488 ops[j].p_argnum = arg + 1;
4489 save_args |= 1 << arg;
4490 }
4491
4492 if (save_args_seen & (1 << (ops[j].p_argnum - 1))) {
4493 save_args_seen = 0;
4494 arg_grp_current++;
4495 if (arg_grp_current >= MAX_ARG_GRP)
4496 ferr(&ops[j], "out of arg groups (arg%d), f %s\n",
4497 ops[j].p_argnum, pp->name);
4498 }
4499 }
4500 else if (ops[j].p_argnum == 0)
4501 ops[j].flags |= OPF_RMD;
4502
4503 // some PUSHes are reused by different calls on other branches,
4504 // but that can't happen if we didn't branch, so they
4505 // can be removed from future searches (handles nested calls)
4506 if (!may_reuse)
4507 ops[j].flags |= OPF_FARGNR;
4508
4509 ops[j].flags |= OPF_FARG;
4510 ops[j].flags &= ~OPF_RSAVE;
4511
4512 // check for __VALIST
4513 if (!pp->is_unresolved && g_func_pp != NULL
4514 && pp->arg[arg].type.is_va_list)
4515 {
4516 k = -1;
4517 ret = resolve_origin(j, &ops[j].operand[0],
4518 magic + 1, &k, NULL);
4519 if (ret == 1 && k >= 0)
4520 {
4521 if (ops[k].op == OP_LEA) {
4522 if (!g_func_pp->is_vararg)
4523 ferr(&ops[k], "lea <arg> used, but %s is not vararg?\n",
4524 g_func_pp->name);
4525
4526 snprintf(buf, sizeof(buf), "arg_%X",
4527 g_func_pp->argc_stack * 4);
4528 if (strstr(ops[k].operand[1].name, buf)
4529 || strstr(ops[k].operand[1].name, "arglist"))
4530 {
4531 ops[k].flags |= OPF_RMD | OPF_NOREGS | OPF_DONE;
4532 ops[j].flags |= OPF_RMD | OPF_NOREGS | OPF_VAPUSH;
4533 save_args &= ~(1 << arg);
4534 reg = -1;
4535 }
4536 else
4537 ferr(&ops[k], "va_list arg detection failed\n");
4538 }
4539 // check for va_list from g_func_pp arg too
4540 else if (ops[k].op == OP_MOV
4541 && is_stack_access(&ops[k], &ops[k].operand[1]))
4542 {
4543 ret = stack_frame_access(&ops[k], &ops[k].operand[1],
4544 buf, sizeof(buf), ops[k].operand[1].name, "", 1, 0);
4545 if (ret >= 0) {
4546 ops[k].flags |= OPF_RMD | OPF_DONE;
4547 ops[j].flags |= OPF_RMD;
4548 ops[j].p_argpass = ret + 1;
4549 save_args &= ~(1 << arg);
4550 reg = -1;
4551 }
4552 }
4553 }
4554 }
4555
4556 *save_arg_vars |= save_args;
4557
4558 // tracking reg usage
4559 if (reg >= 0)
4560 *regmask |= 1 << reg;
4561
4562 arg++;
4563 if (!pp->is_unresolved) {
4564 // next arg
4565 for (; arg < pp->argc; arg++)
4566 if (pp->arg[arg].reg == NULL)
4567 break;
4568 }
4569 magic = (magic & 0xffffff) | (arg << 24);
4570 }
4571
4572 if (ops[j].p_arggrp > arg_grp_current) {
4573 save_args_seen = 0;
4574 arg_grp_current = ops[j].p_arggrp;
4575 }
4576 if (ops[j].p_argnum > 0)
4577 save_args_seen |= 1 << (ops[j].p_argnum - 1);
4578 }
4579
4580 if (arg < pp->argc) {
4581 ferr(po, "arg collect failed for '%s': %d/%d\n",
4582 pp->name, arg, pp->argc);
4583 return -1;
4584 }
4585
4586 if (arg_grp_current > *arg_grp)
4587 *arg_grp = arg_grp_current;
4588
4589 return arg;
4590}
4591
4592static int collect_call_args(struct parsed_op *po, int i,
4593 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
4594 int magic)
4595{
4596 // arg group is for cases when pushes for
4597 // multiple funcs are going on
4598 struct parsed_op *po_tmp;
4599 int save_arg_vars_current = 0;
4600 int arg_grp = 0;
4601 int ret;
4602 int a;
4603
4604 ret = collect_call_args_r(po, i, pp, regmask,
4605 &save_arg_vars_current, &arg_grp, 0, magic, 0, 0);
4606 if (ret < 0)
4607 return ret;
4608
4609 if (arg_grp != 0) {
4610 // propagate arg_grp
4611 for (a = 0; a < pp->argc; a++) {
4612 if (pp->arg[a].reg != NULL)
4613 continue;
4614
4615 po_tmp = pp->arg[a].datap;
4616 while (po_tmp != NULL) {
4617 po_tmp->p_arggrp = arg_grp;
4618 if (po_tmp->p_argnext > 0)
4619 po_tmp = &ops[po_tmp->p_argnext];
4620 else
4621 po_tmp = NULL;
4622 }
4623 }
4624 }
4625 save_arg_vars[arg_grp] |= save_arg_vars_current;
4626
4627 if (pp->is_unresolved) {
4628 pp->argc += ret;
4629 pp->argc_stack += ret;
4630 for (a = 0; a < pp->argc; a++)
4631 if (pp->arg[a].type.name == NULL)
4632 pp->arg[a].type.name = strdup("int");
4633 }
4634
4635 return ret;
4636}
4637
4638static void reg_use_pass(int i, int opcnt, unsigned char *cbits,
4639 int regmask_now, int *regmask,
4640 int regmask_save_now, int *regmask_save,
4641 int *regmask_init, int regmask_arg)
4642{
4643 struct parsed_op *po;
4644 unsigned int mask;
4645 int already_saved;
4646 int regmask_new;
4647 int regmask_op;
4648 int flags_set;
4649 int ret, reg;
4650 int j;
4651
4652 for (; i < opcnt; i++)
4653 {
4654 po = &ops[i];
4655 if (cbits[i >> 3] & (1 << (i & 7)))
4656 return;
4657 cbits[i >> 3] |= (1 << (i & 7));
4658
4659 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
4660 if (po->flags & (OPF_RMD|OPF_DONE))
4661 continue;
4662 if (po->btj != NULL) {
4663 for (j = 0; j < po->btj->count; j++) {
4664 check_i(po, po->btj->d[j].bt_i);
4665 reg_use_pass(po->btj->d[j].bt_i, opcnt, cbits,
4666 regmask_now, regmask, regmask_save_now, regmask_save,
4667 regmask_init, regmask_arg);
4668 }
4669 return;
4670 }
4671
4672 check_i(po, po->bt_i);
4673 if (po->flags & OPF_CJMP)
4674 reg_use_pass(po->bt_i, opcnt, cbits,
4675 regmask_now, regmask, regmask_save_now, regmask_save,
4676 regmask_init, regmask_arg);
4677 else
4678 i = po->bt_i - 1;
4679 continue;
4680 }
4681
4682 if (po->op == OP_PUSH && !(po->flags & (OPF_FARG|OPF_DONE))
4683 && !g_func_pp->is_userstack
4684 && po->operand[0].type == OPT_REG)
4685 {
4686 reg = po->operand[0].reg;
4687 ferr_assert(po, reg >= 0);
4688
4689 already_saved = 0;
4690 flags_set = OPF_RSAVE | OPF_RMD | OPF_DONE;
4691 if (regmask_now & (1 << reg)) {
4692 already_saved = regmask_save_now & (1 << reg);
4693 flags_set = OPF_RSAVE | OPF_DONE;
4694 }
4695
4696 ret = scan_for_pop(i + 1, opcnt, i + opcnt * 3, reg, 0, 0);
4697 if (ret == 1) {
4698 scan_for_pop(i + 1, opcnt, i + opcnt * 4, reg, 0, flags_set);
4699 }
4700 else {
4701 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].reg, 0);
4702 if (ret == 1) {
4703 scan_for_pop_ret(i + 1, opcnt, po->operand[0].reg,
4704 flags_set);
4705 }
4706 }
4707 if (ret == 1) {
4708 ferr_assert(po, !already_saved);
4709 po->flags |= flags_set;
4710
4711 if (regmask_now & (1 << reg)) {
4712 regmask_save_now |= (1 << reg);
4713 *regmask_save |= regmask_save_now;
4714 }
4715 continue;
4716 }
4717 }
4718 else if (po->op == OP_POP && (po->flags & OPF_RSAVE)) {
4719 reg = po->operand[0].reg;
4720 ferr_assert(po, reg >= 0);
4721
4722 if (regmask_save_now & (1 << reg))
4723 regmask_save_now &= ~(1 << reg);
4724 else
4725 regmask_now &= ~(1 << reg);
4726 continue;
4727 }
4728 else if (po->op == OP_CALL) {
4729 if ((po->regmask_dst & (1 << xAX))
4730 && !(po->regmask_dst & (1 << xDX)))
4731 {
4732 if (po->flags & OPF_TAIL)
4733 // don't need eax, will do "return f();" or "f(); return;"
4734 po->regmask_dst &= ~(1 << xAX);
4735 else {
4736 struct parsed_opr opr = OPR_INIT(OPT_REG, OPLM_DWORD, xAX);
4737 j = -1;
4738 find_next_read(i + 1, opcnt, &opr, i + opcnt * 17, &j);
4739 if (j == -1)
4740 // not used
4741 po->regmask_dst &= ~(1 << xAX);
4742 }
4743 }
4744 }
4745
4746 if (po->flags & OPF_NOREGS)
4747 continue;
4748
4749 if (po->flags & OPF_FPUSH) {
4750 if (regmask_now & mxST1)
4751 ferr(po, "TODO: FPUSH on active ST1\n");
4752 if (regmask_now & mxST0)
4753 po->flags |= OPF_FSHIFT;
4754 mask = mxST0 | mxST1;
4755 regmask_now = (regmask_now & ~mask) | ((regmask_now & mxST0) << 1);
4756 }
4757
4758 // if incomplete register is used, clear it on init to avoid
4759 // later use of uninitialized upper part in some situations
4760 if ((po->flags & OPF_DATA) && po->operand[0].type == OPT_REG
4761 && po->operand[0].lmod != OPLM_DWORD)
4762 {
4763 reg = po->operand[0].reg;
4764 ferr_assert(po, reg >= 0);
4765
4766 if (!(regmask_now & (1 << reg)))
4767 *regmask_init |= 1 << reg;
4768 }
4769
4770 regmask_op = po->regmask_src | po->regmask_dst;
4771
4772 regmask_new = po->regmask_src & ~regmask_now & ~regmask_arg;
4773 regmask_new &= ~(1 << xSP);
4774 if (g_bp_frame && !(po->flags & OPF_EBP_S))
4775 regmask_new &= ~(1 << xBP);
4776
4777 if (regmask_new != 0)
4778 fnote(po, "uninitialized reg mask: %x\n", regmask_new);
4779
4780 if (regmask_op & (1 << xBP)) {
4781 if (g_bp_frame && !(po->flags & OPF_EBP_S)) {
4782 if (po->regmask_dst & (1 << xBP))
4783 // compiler decided to drop bp frame and use ebp as scratch
4784 scan_fwd_set_flags(i + 1, opcnt, i + opcnt * 5, OPF_EBP_S);
4785 else
4786 regmask_op &= ~(1 << xBP);
4787 }
4788 }
4789
4790 regmask_now |= regmask_op;
4791 *regmask |= regmask_now;
4792
4793 // released regs
4794 if (po->flags & OPF_FPOP) {
4795 mask = mxST0 | mxST1;
4796 if (!(regmask_now & mask))
4797 ferr(po, "float pop on empty stack?\n");
4798 if (regmask_now & mxST1)
4799 po->flags |= OPF_FSHIFT;
4800 regmask_now = (regmask_now & ~mask) | ((regmask_now & mxST1) >> 1);
4801 }
4802
4803 if (po->flags & OPF_TAIL) {
4804 if (regmask_now & (mxST0 | mxST1))
4805 ferr(po, "float regs on tail: %x\n", regmask_now);
4806
4807 // there is support for "conditional tailcall", sort of
4808 if (!(po->flags & OPF_CC))
4809 return;
4810 }
4811 }
4812}
4813
4814static void pp_insert_reg_arg(struct parsed_proto *pp, const char *reg)
4815{
4816 int i;
4817
4818 for (i = 0; i < pp->argc; i++)
4819 if (pp->arg[i].reg == NULL)
4820 break;
4821
4822 if (pp->argc_stack)
4823 memmove(&pp->arg[i + 1], &pp->arg[i],
4824 sizeof(pp->arg[0]) * pp->argc_stack);
4825 memset(&pp->arg[i], 0, sizeof(pp->arg[i]));
4826 pp->arg[i].reg = strdup(reg);
4827 pp->arg[i].type.name = strdup("int");
4828 pp->argc++;
4829 pp->argc_reg++;
4830}
4831
4832static void output_std_flags(FILE *fout, struct parsed_op *po,
4833 int *pfomask, const char *dst_opr_text)
4834{
4835 if (*pfomask & (1 << PFO_Z)) {
4836 fprintf(fout, "\n cond_z = (%s%s == 0);",
4837 lmod_cast_u(po, po->operand[0].lmod), dst_opr_text);
4838 *pfomask &= ~(1 << PFO_Z);
4839 }
4840 if (*pfomask & (1 << PFO_S)) {
4841 fprintf(fout, "\n cond_s = (%s%s < 0);",
4842 lmod_cast_s(po, po->operand[0].lmod), dst_opr_text);
4843 *pfomask &= ~(1 << PFO_S);
4844 }
4845}
4846
4847enum {
4848 OPP_FORCE_NORETURN = (1 << 0),
4849 OPP_SIMPLE_ARGS = (1 << 1),
4850 OPP_ALIGN = (1 << 2),
4851};
4852
4853static void output_pp_attrs(FILE *fout, const struct parsed_proto *pp,
4854 int flags)
4855{
4856 const char *cconv = "";
4857
4858 if (pp->is_fastcall)
4859 cconv = "__fastcall ";
4860 else if (pp->is_stdcall && pp->argc_reg == 0)
4861 cconv = "__stdcall ";
4862
4863 fprintf(fout, (flags & OPP_ALIGN) ? "%-16s" : "%s", cconv);
4864
4865 if (pp->is_noreturn || (flags & OPP_FORCE_NORETURN))
4866 fprintf(fout, "noreturn ");
4867}
4868
4869static void output_pp(FILE *fout, const struct parsed_proto *pp,
4870 int flags)
4871{
4872 int i;
4873
4874 fprintf(fout, (flags & OPP_ALIGN) ? "%-5s" : "%s ",
4875 pp->ret_type.name);
4876 if (pp->is_fptr)
4877 fprintf(fout, "(");
4878 output_pp_attrs(fout, pp, flags);
4879 if (pp->is_fptr)
4880 fprintf(fout, "*");
4881 fprintf(fout, "%s", pp->name);
4882 if (pp->is_fptr)
4883 fprintf(fout, ")");
4884
4885 fprintf(fout, "(");
4886 for (i = 0; i < pp->argc; i++) {
4887 if (i > 0)
4888 fprintf(fout, ", ");
4889 if (pp->arg[i].fptr != NULL && !(flags & OPP_SIMPLE_ARGS)) {
4890 // func pointer
4891 output_pp(fout, pp->arg[i].fptr, 0);
4892 }
4893 else if (pp->arg[i].type.is_retreg) {
4894 fprintf(fout, "u32 *r_%s", pp->arg[i].reg);
4895 }
4896 else {
4897 fprintf(fout, "%s", pp->arg[i].type.name);
4898 if (!pp->is_fptr)
4899 fprintf(fout, " a%d", i + 1);
4900 }
4901 }
4902 if (pp->is_vararg) {
4903 if (i > 0)
4904 fprintf(fout, ", ");
4905 fprintf(fout, "...");
4906 }
4907 fprintf(fout, ")");
4908}
4909
4910static char *saved_arg_name(char *buf, size_t buf_size, int grp, int num)
4911{
4912 char buf1[16];
4913
4914 buf1[0] = 0;
4915 if (grp > 0)
4916 snprintf(buf1, sizeof(buf1), "%d", grp);
4917 snprintf(buf, buf_size, "s%s_a%d", buf1, num);
4918
4919 return buf;
4920}
4921
4922static void gen_x_cleanup(int opcnt);
4923
4924static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
4925{
4926 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
4927 struct parsed_opr *last_arith_dst = NULL;
4928 char buf1[256], buf2[256], buf3[256], cast[64];
4929 struct parsed_proto *pp, *pp_tmp;
4930 struct parsed_data *pd;
4931 unsigned int uval;
4932 int save_arg_vars[MAX_ARG_GRP] = { 0, };
4933 unsigned char cbits[MAX_OPS / 8];
4934 int cond_vars = 0;
4935 int need_tmp_var = 0;
4936 int need_tmp64 = 0;
4937 int had_decl = 0;
4938 int label_pending = 0;
4939 int regmask_save = 0; // regs saved/restored in this func
4940 int regmask_arg; // regs from this function args (fastcall, etc)
4941 int regmask_ret; // regs needed on ret
4942 int regmask_now; // temp
4943 int regmask_init = 0; // regs that need zero initialization
4944 int regmask_pp = 0; // regs used in complex push-pop graph
4945 int regmask = 0; // used regs
4946 int pfomask = 0;
4947 int found = 0;
4948 int no_output;
4949 int i, j, l;
4950 int arg;
4951 int reg;
4952 int ret;
4953
4954 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
4955 g_stack_frame_used = 0;
4956 if (g_sct_func_attr & SCTFA_CLEAR_REGS)
4957 regmask_init = g_regmask_init;
4958
4959 g_func_pp = proto_parse(fhdr, funcn, 0);
4960 if (g_func_pp == NULL)
4961 ferr(ops, "proto_parse failed for '%s'\n", funcn);
4962
4963 regmask_arg = get_pp_arg_regmask_src(g_func_pp);
4964 regmask_ret = get_pp_arg_regmask_dst(g_func_pp);
4965
4966 // pass1:
4967 // - resolve all branches
4968 // - parse calls with labels
4969 resolve_branches_parse_calls(opcnt);
4970
4971 // pass2:
4972 // - handle ebp/esp frame, remove ops related to it
4973 scan_prologue_epilogue(opcnt);
4974
4975 // pass3:
4976 // - remove dead labels
4977 // - set regs needed at ret
4978 for (i = 0; i < opcnt; i++)
4979 {
4980 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
4981 free(g_labels[i]);
4982 g_labels[i] = NULL;
4983 }
4984
4985 if (ops[i].op == OP_RET)
4986 ops[i].regmask_src |= regmask_ret;
4987 }
4988
4989 // pass4:
4990 // - process trivial calls
4991 for (i = 0; i < opcnt; i++)
4992 {
4993 po = &ops[i];
4994 if (po->flags & (OPF_RMD|OPF_DONE))
4995 continue;
4996
4997 if (po->op == OP_CALL)
4998 {
4999 pp = process_call_early(i, opcnt, &j);
5000 if (pp != NULL) {
5001 if (!(po->flags & OPF_ATAIL))
5002 // since we know the args, try to collect them
5003 if (collect_call_args_early(po, i, pp, &regmask) != 0)
5004 pp = NULL;
5005 }
5006
5007 if (pp != NULL) {
5008 if (j >= 0) {
5009 // commit esp adjust
5010 if (ops[j].op != OP_POP)
5011 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
5012 else {
5013 for (l = 0; l < pp->argc_stack; l++)
5014 ops[j + l].flags |= OPF_DONE | OPF_RMD | OPF_NOREGS;
5015 }
5016 }
5017
5018 if (strstr(pp->ret_type.name, "int64"))
5019 need_tmp64 = 1;
5020
5021 po->flags |= OPF_DONE;
5022 }
5023 }
5024 }
5025
5026 // pass5:
5027 // - process calls, stage 2
5028 // - handle some push/pop pairs
5029 // - scan for STD/CLD, propagate DF
5030 for (i = 0; i < opcnt; i++)
5031 {
5032 po = &ops[i];
5033 if (po->flags & OPF_RMD)
5034 continue;
5035
5036 if (po->op == OP_CALL)
5037 {
5038 if (!(po->flags & OPF_DONE)) {
5039 pp = process_call(i, opcnt);
5040
5041 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
5042 // since we know the args, collect them
5043 collect_call_args(po, i, pp, &regmask, save_arg_vars,
5044 i + opcnt * 2);
5045 }
5046 // for unresolved, collect after other passes
5047 }
5048
5049 pp = po->pp;
5050 ferr_assert(po, pp != NULL);
5051
5052 po->regmask_src |= get_pp_arg_regmask_src(pp);
5053 po->regmask_dst |= get_pp_arg_regmask_dst(pp);
5054
5055 if (po->regmask_dst & mxST0)
5056 po->flags |= OPF_FPUSH;
5057
5058 if (strstr(pp->ret_type.name, "int64"))
5059 need_tmp64 = 1;
5060
5061 continue;
5062 }
5063
5064 if (po->flags & OPF_DONE)
5065 continue;
5066
5067 if (po->op == OP_PUSH && !(po->flags & OPF_FARG)
5068 && !(po->flags & OPF_RSAVE) && po->operand[0].type == OPT_CONST)
5069 {
5070 scan_for_pop_const(i, opcnt, i + opcnt * 12);
5071 }
5072 else if (po->op == OP_POP)
5073 scan_pushes_for_pop(i, opcnt, &regmask_pp);
5074 else if (po->op == OP_STD) {
5075 po->flags |= OPF_DF | OPF_RMD | OPF_DONE;
5076 scan_propagate_df(i + 1, opcnt);
5077 }
5078 }
5079
5080 // pass6:
5081 // - find POPs for PUSHes, rm both
5082 // - scan for all used registers
5083 memset(cbits, 0, sizeof(cbits));
5084 reg_use_pass(0, opcnt, cbits, regmask_init, &regmask,
5085 0, &regmask_save, &regmask_init, regmask_arg);
5086
5087 // pass7:
5088 // - find flag set ops for their users
5089 // - do unresolved calls
5090 // - declare indirect functions
5091 for (i = 0; i < opcnt; i++)
5092 {
5093 po = &ops[i];
5094 if (po->flags & (OPF_RMD|OPF_DONE))
5095 continue;
5096
5097 if (po->flags & OPF_CC)
5098 {
5099 int setters[16], cnt = 0, branched = 0;
5100
5101 ret = scan_for_flag_set(i, i + opcnt * 6,
5102 &branched, setters, &cnt);
5103 if (ret < 0 || cnt <= 0)
5104 ferr(po, "unable to trace flag setter(s)\n");
5105 if (cnt > ARRAY_SIZE(setters))
5106 ferr(po, "too many flag setters\n");
5107
5108 for (j = 0; j < cnt; j++)
5109 {
5110 tmp_op = &ops[setters[j]]; // flag setter
5111 pfomask = 0;
5112
5113 // to get nicer code, we try to delay test and cmp;
5114 // if we can't because of operand modification, or if we
5115 // have arith op, or branch, make it calculate flags explicitly
5116 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP)
5117 {
5118 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i, 0) >= 0)
5119 pfomask = 1 << po->pfo;
5120 }
5121 else if (tmp_op->op == OP_CMPS || tmp_op->op == OP_SCAS) {
5122 pfomask = 1 << po->pfo;
5123 }
5124 else {
5125 // see if we'll be able to handle based on op result
5126 if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR
5127 && po->pfo != PFO_Z && po->pfo != PFO_S
5128 && po->pfo != PFO_P)
5129 || branched
5130 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
5131 {
5132 pfomask = 1 << po->pfo;
5133 }
5134
5135 if (tmp_op->op == OP_ADD && po->pfo == PFO_C) {
5136 propagate_lmod(tmp_op, &tmp_op->operand[0],
5137 &tmp_op->operand[1]);
5138 if (tmp_op->operand[0].lmod == OPLM_DWORD)
5139 need_tmp64 = 1;
5140 }
5141 }
5142 if (pfomask) {
5143 tmp_op->pfomask |= pfomask;
5144 cond_vars |= pfomask;
5145 }
5146 // note: may overwrite, currently not a problem
5147 po->datap = tmp_op;
5148 }
5149
5150 if (po->op == OP_RCL || po->op == OP_RCR
5151 || po->op == OP_ADC || po->op == OP_SBB)
5152 cond_vars |= 1 << PFO_C;
5153 }
5154
5155 if (po->op == OP_CMPS || po->op == OP_SCAS) {
5156 cond_vars |= 1 << PFO_Z;
5157 }
5158 else if (po->op == OP_MUL
5159 || (po->op == OP_IMUL && po->operand_cnt == 1))
5160 {
5161 if (po->operand[0].lmod == OPLM_DWORD)
5162 need_tmp64 = 1;
5163 }
5164 else if (po->op == OP_CALL) {
5165 // note: resolved non-reg calls are OPF_DONE already
5166 pp = po->pp;
5167 ferr_assert(po, pp != NULL);
5168
5169 if (pp->is_unresolved) {
5170 int regmask_stack = 0;
5171 collect_call_args(po, i, pp, &regmask, save_arg_vars,
5172 i + opcnt * 2);
5173
5174 // this is pretty rough guess:
5175 // see ecx and edx were pushed (and not their saved versions)
5176 for (arg = 0; arg < pp->argc; arg++) {
5177 if (pp->arg[arg].reg != NULL)
5178 continue;
5179
5180 tmp_op = pp->arg[arg].datap;
5181 if (tmp_op == NULL)
5182 ferr(po, "parsed_op missing for arg%d\n", arg);
5183 if (tmp_op->p_argnum == 0 && tmp_op->operand[0].type == OPT_REG)
5184 regmask_stack |= 1 << tmp_op->operand[0].reg;
5185 }
5186
5187 if (!((regmask_stack & (1 << xCX))
5188 && (regmask_stack & (1 << xDX))))
5189 {
5190 if (pp->argc_stack != 0
5191 || ((regmask | regmask_arg) & ((1 << xCX)|(1 << xDX))))
5192 {
5193 pp_insert_reg_arg(pp, "ecx");
5194 pp->is_fastcall = 1;
5195 regmask_init |= 1 << xCX;
5196 regmask |= 1 << xCX;
5197 }
5198 if (pp->argc_stack != 0
5199 || ((regmask | regmask_arg) & (1 << xDX)))
5200 {
5201 pp_insert_reg_arg(pp, "edx");
5202 regmask_init |= 1 << xDX;
5203 regmask |= 1 << xDX;
5204 }
5205 }
5206
5207 // note: __cdecl doesn't fall into is_unresolved category
5208 if (pp->argc_stack > 0)
5209 pp->is_stdcall = 1;
5210 }
5211 }
5212 else if (po->op == OP_MOV && po->operand[0].pp != NULL
5213 && po->operand[1].pp != NULL)
5214 {
5215 // <var> = offset <something>
5216 if ((po->operand[1].pp->is_func || po->operand[1].pp->is_fptr)
5217 && !IS_START(po->operand[1].name, "off_"))
5218 {
5219 if (!po->operand[0].pp->is_fptr)
5220 ferr(po, "%s not declared as fptr when it should be\n",
5221 po->operand[0].name);
5222 if (pp_cmp_func(po->operand[0].pp, po->operand[1].pp)) {
5223 pp_print(buf1, sizeof(buf1), po->operand[0].pp);
5224 pp_print(buf2, sizeof(buf2), po->operand[1].pp);
5225 fnote(po, "var: %s\n", buf1);
5226 fnote(po, "func: %s\n", buf2);
5227 ferr(po, "^ mismatch\n");
5228 }
5229 }
5230 }
5231 else if (po->op == OP_DIV || po->op == OP_IDIV) {
5232 if (po->operand[0].lmod == OPLM_DWORD) {
5233 // 32bit division is common, look for it
5234 if (po->op == OP_DIV)
5235 ret = scan_for_reg_clear(i, xDX);
5236 else
5237 ret = scan_for_cdq_edx(i);
5238 if (ret >= 0)
5239 po->flags |= OPF_32BIT;
5240 else
5241 need_tmp64 = 1;
5242 }
5243 else
5244 need_tmp_var = 1;
5245 }
5246 else if (po->op == OP_CLD)
5247 po->flags |= OPF_RMD | OPF_DONE;
5248 else if (po->op == OPP_FTOL) {
5249 struct parsed_opr opr = OPR_INIT(OPT_REG, OPLM_DWORD, xDX);
5250 j = -1;
5251 find_next_read(i + 1, opcnt, &opr, i + opcnt * 18, &j);
5252 if (j == -1)
5253 po->flags |= OPF_32BIT;
5254 }
5255
5256 if (po->op == OP_RCL || po->op == OP_RCR || po->op == OP_XCHG)
5257 need_tmp_var = 1;
5258 }
5259
5260 // output starts here
5261
5262 // define userstack size
5263 if (g_func_pp->is_userstack) {
5264 fprintf(fout, "#ifndef US_SZ_%s\n", g_func_pp->name);
5265 fprintf(fout, "#define US_SZ_%s USERSTACK_SIZE\n", g_func_pp->name);
5266 fprintf(fout, "#endif\n");
5267 }
5268
5269 // the function itself
5270 ferr_assert(ops, !g_func_pp->is_fptr);
5271 output_pp(fout, g_func_pp,
5272 (g_ida_func_attr & IDAFA_NORETURN) ? OPP_FORCE_NORETURN : 0);
5273 fprintf(fout, "\n{\n");
5274
5275 // declare indirect functions
5276 for (i = 0; i < opcnt; i++) {
5277 po = &ops[i];
5278 if (po->flags & OPF_RMD)
5279 continue;
5280
5281 if (po->op == OP_CALL) {
5282 pp = po->pp;
5283 if (pp == NULL)
5284 ferr(po, "NULL pp\n");
5285
5286 if (pp->is_fptr && !(pp->name[0] != 0 && pp->is_arg)) {
5287 if (pp->name[0] != 0) {
5288 memmove(pp->name + 2, pp->name, strlen(pp->name) + 1);
5289 memcpy(pp->name, "i_", 2);
5290
5291 // might be declared already
5292 found = 0;
5293 for (j = 0; j < i; j++) {
5294 if (ops[j].op == OP_CALL && (pp_tmp = ops[j].pp)) {
5295 if (pp_tmp->is_fptr && IS(pp->name, pp_tmp->name)) {
5296 found = 1;
5297 break;
5298 }
5299 }
5300 }
5301 if (found)
5302 continue;
5303 }
5304 else
5305 snprintf(pp->name, sizeof(pp->name), "icall%d", i);
5306
5307 fprintf(fout, " ");
5308 output_pp(fout, pp, OPP_SIMPLE_ARGS);
5309 fprintf(fout, ";\n");
5310 }
5311 }
5312 }
5313
5314 // output LUTs/jumptables
5315 for (i = 0; i < g_func_pd_cnt; i++) {
5316 pd = &g_func_pd[i];
5317 fprintf(fout, " static const ");
5318 if (pd->type == OPT_OFFSET) {
5319 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
5320
5321 for (j = 0; j < pd->count; j++) {
5322 if (j > 0)
5323 fprintf(fout, ", ");
5324 fprintf(fout, "&&%s", pd->d[j].u.label);
5325 }
5326 }
5327 else {
5328 fprintf(fout, "%s %s[] =\n { ",
5329 lmod_type_u(ops, pd->lmod), pd->label);
5330
5331 for (j = 0; j < pd->count; j++) {
5332 if (j > 0)
5333 fprintf(fout, ", ");
5334 fprintf(fout, "%u", pd->d[j].u.val);
5335 }
5336 }
5337 fprintf(fout, " };\n");
5338 had_decl = 1;
5339 }
5340
5341 // declare stack frame, va_arg
5342 if (g_stack_fsz) {
5343 fprintf(fout, " union { u32 d[%d];", (g_stack_fsz + 3) / 4);
5344 if (g_func_lmods & (1 << OPLM_WORD))
5345 fprintf(fout, " u16 w[%d];", (g_stack_fsz + 1) / 2);
5346 if (g_func_lmods & (1 << OPLM_BYTE))
5347 fprintf(fout, " u8 b[%d];", g_stack_fsz);
5348 if (g_func_lmods & (1 << OPLM_QWORD))
5349 fprintf(fout, " double q[%d];", (g_stack_fsz + 7) / 8);
5350 fprintf(fout, " } sf;\n");
5351 had_decl = 1;
5352 }
5353
5354 if (g_func_pp->is_userstack) {
5355 fprintf(fout, " u32 fake_sf[US_SZ_%s / 4];\n", g_func_pp->name);
5356 fprintf(fout, " u32 *esp = &fake_sf[sizeof(fake_sf) / 4];\n");
5357 had_decl = 1;
5358 }
5359
5360 if (g_func_pp->is_vararg) {
5361 fprintf(fout, " va_list ap;\n");
5362 had_decl = 1;
5363 }
5364
5365 // declare arg-registers
5366 for (i = 0; i < g_func_pp->argc; i++) {
5367 if (g_func_pp->arg[i].reg != NULL) {
5368 reg = char_array_i(regs_r32,
5369 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
5370 if (regmask & (1 << reg)) {
5371 if (g_func_pp->arg[i].type.is_retreg)
5372 fprintf(fout, " u32 %s = *r_%s;\n",
5373 g_func_pp->arg[i].reg, g_func_pp->arg[i].reg);
5374 else
5375 fprintf(fout, " u32 %s = (u32)a%d;\n",
5376 g_func_pp->arg[i].reg, i + 1);
5377 }
5378 else {
5379 if (g_func_pp->arg[i].type.is_retreg)
5380 ferr(ops, "retreg '%s' is unused?\n",
5381 g_func_pp->arg[i].reg);
5382 fprintf(fout, " // %s = a%d; // unused\n",
5383 g_func_pp->arg[i].reg, i + 1);
5384 }
5385 had_decl = 1;
5386 }
5387 }
5388
5389 // declare normal registers
5390 regmask_now = regmask & ~regmask_arg;
5391 regmask_now &= ~(1 << xSP);
5392 if (regmask_now & 0x00ff) {
5393 for (reg = 0; reg < 8; reg++) {
5394 if (regmask_now & (1 << reg)) {
5395 fprintf(fout, " u32 %s", regs_r32[reg]);
5396 if (regmask_init & (1 << reg))
5397 fprintf(fout, " = 0");
5398 fprintf(fout, ";\n");
5399 had_decl = 1;
5400 }
5401 }
5402 }
5403 // ... mmx
5404 if (regmask_now & 0xff00) {
5405 for (reg = 8; reg < 16; reg++) {
5406 if (regmask_now & (1 << reg)) {
5407 fprintf(fout, " mmxr %s", regs_r32[reg]);
5408 if (regmask_init & (1 << reg))
5409 fprintf(fout, " = { 0, }");
5410 fprintf(fout, ";\n");
5411 had_decl = 1;
5412 }
5413 }
5414 }
5415 // ... x87
5416 if (regmask_now & 0xff0000) {
5417 for (reg = 16; reg < 24; reg++) {
5418 if (regmask_now & (1 << reg)) {
5419 fprintf(fout, " double f_st%d", reg - 16);
5420 if (regmask_init & (1 << reg))
5421 fprintf(fout, " = 0");
5422 fprintf(fout, ";\n");
5423 had_decl = 1;
5424 }
5425 }
5426 }
5427
5428 if (regmask_save) {
5429 for (reg = 0; reg < 8; reg++) {
5430 if (regmask_save & (1 << reg)) {
5431 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
5432 had_decl = 1;
5433 }
5434 }
5435 }
5436
5437 for (i = 0; i < ARRAY_SIZE(save_arg_vars); i++) {
5438 if (save_arg_vars[i] == 0)
5439 continue;
5440 for (reg = 0; reg < 32; reg++) {
5441 if (save_arg_vars[i] & (1 << reg)) {
5442 fprintf(fout, " u32 %s;\n",
5443 saved_arg_name(buf1, sizeof(buf1), i, reg + 1));
5444 had_decl = 1;
5445 }
5446 }
5447 }
5448
5449 // declare push-pop temporaries
5450 if (regmask_pp) {
5451 for (reg = 0; reg < 8; reg++) {
5452 if (regmask_pp & (1 << reg)) {
5453 fprintf(fout, " u32 pp_%s;\n", regs_r32[reg]);
5454 had_decl = 1;
5455 }
5456 }
5457 }
5458
5459 if (cond_vars) {
5460 for (i = 0; i < 8; i++) {
5461 if (cond_vars & (1 << i)) {
5462 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
5463 had_decl = 1;
5464 }
5465 }
5466 }
5467
5468 if (need_tmp_var) {
5469 fprintf(fout, " u32 tmp;\n");
5470 had_decl = 1;
5471 }
5472
5473 if (need_tmp64) {
5474 fprintf(fout, " u64 tmp64;\n");
5475 had_decl = 1;
5476 }
5477
5478 if (had_decl)
5479 fprintf(fout, "\n");
5480
5481 // do stack clear, if needed
5482 if (g_sct_func_attr & SCTFA_CLEAR_SF) {
5483 fprintf(fout, " ");
5484 if (g_stack_clear_len != 0) {
5485 if (g_stack_clear_len <= 4) {
5486 for (i = 0; i < g_stack_clear_len; i++)
5487 fprintf(fout, "sf.d[%d] = ", g_stack_clear_start + i);
5488 fprintf(fout, "0;\n");
5489 }
5490 else {
5491 fprintf(fout, "memset(&sf[%d], 0, %d);\n",
5492 g_stack_clear_start, g_stack_clear_len * 4);
5493 }
5494 }
5495 else
5496 fprintf(fout, "memset(&sf, 0, sizeof(sf));\n");
5497 }
5498
5499 if (g_func_pp->is_vararg) {
5500 if (g_func_pp->argc_stack == 0)
5501 ferr(ops, "vararg func without stack args?\n");
5502 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
5503 }
5504
5505 // output ops
5506 for (i = 0; i < opcnt; i++)
5507 {
5508 if (g_labels[i] != NULL) {
5509 fprintf(fout, "\n%s:\n", g_labels[i]);
5510 label_pending = 1;
5511
5512 delayed_flag_op = NULL;
5513 last_arith_dst = NULL;
5514 }
5515
5516 po = &ops[i];
5517 if (po->flags & OPF_RMD)
5518 continue;
5519
5520 no_output = 0;
5521
5522 #define assert_operand_cnt(n_) \
5523 if (po->operand_cnt != n_) \
5524 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
5525
5526 // conditional/flag using op?
5527 if (po->flags & OPF_CC)
5528 {
5529 int is_delayed = 0;
5530
5531 tmp_op = po->datap;
5532
5533 // we go through all this trouble to avoid using parsed_flag_op,
5534 // which makes generated code much nicer
5535 if (delayed_flag_op != NULL)
5536 {
5537 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op,
5538 po->pfo, po->pfo_inv);
5539 is_delayed = 1;
5540 }
5541 else if (last_arith_dst != NULL
5542 && (po->pfo == PFO_Z || po->pfo == PFO_S || po->pfo == PFO_P
5543 || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR))
5544 ))
5545 {
5546 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
5547 out_test_for_cc(buf1, sizeof(buf1), po, po->pfo, po->pfo_inv,
5548 last_arith_dst->lmod, buf3);
5549 is_delayed = 1;
5550 }
5551 else if (tmp_op != NULL) {
5552 // use preprocessed flag calc results
5553 if (!(tmp_op->pfomask & (1 << po->pfo)))
5554 ferr(po, "not prepared for pfo %d\n", po->pfo);
5555
5556 // note: pfo_inv was not yet applied
5557 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
5558 po->pfo_inv ? "!" : "", parsed_flag_op_names[po->pfo]);
5559 }
5560 else {
5561 ferr(po, "all methods of finding comparison failed\n");
5562 }
5563
5564 if (po->flags & OPF_JMP) {
5565 fprintf(fout, " if %s", buf1);
5566 }
5567 else if (po->op == OP_RCL || po->op == OP_RCR
5568 || po->op == OP_ADC || po->op == OP_SBB)
5569 {
5570 if (is_delayed)
5571 fprintf(fout, " cond_%s = %s;\n",
5572 parsed_flag_op_names[po->pfo], buf1);
5573 }
5574 else if (po->flags & OPF_DATA) { // SETcc
5575 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
5576 fprintf(fout, " %s = %s;", buf2, buf1);
5577 }
5578 else {
5579 ferr(po, "unhandled conditional op\n");
5580 }
5581 }
5582
5583 pfomask = po->pfomask;
5584
5585 if (po->flags & (OPF_REPZ|OPF_REPNZ)) {
5586 struct parsed_opr opr = OPR_INIT(OPT_REG, OPLM_DWORD, xCX);
5587 ret = try_resolve_const(i, &opr, opcnt * 7 + i, &uval);
5588
5589 if (ret != 1 || uval == 0) {
5590 // we need initial flags for ecx=0 case..
5591 if (i > 0 && ops[i - 1].op == OP_XOR
5592 && IS(ops[i - 1].operand[0].name,
5593 ops[i - 1].operand[1].name))
5594 {
5595 fprintf(fout, " cond_z = ");
5596 if (pfomask & (1 << PFO_C))
5597 fprintf(fout, "cond_c = ");
5598 fprintf(fout, "0;\n");
5599 }
5600 else if (last_arith_dst != NULL) {
5601 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
5602 out_test_for_cc(buf1, sizeof(buf1), po, PFO_Z, 0,
5603 last_arith_dst->lmod, buf3);
5604 fprintf(fout, " cond_z = %s;\n", buf1);
5605 }
5606 else
5607 ferr(po, "missing initial ZF\n");
5608 }
5609 }
5610
5611 switch (po->op)
5612 {
5613 case OP_MOV:
5614 assert_operand_cnt(2);
5615 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5616 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5617 default_cast_to(buf3, sizeof(buf3), &po->operand[0]);
5618 fprintf(fout, " %s = %s;", buf1,
5619 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
5620 buf3, 0));
5621 break;
5622
5623 case OP_LEA:
5624 assert_operand_cnt(2);
5625 po->operand[1].lmod = OPLM_DWORD; // always
5626 fprintf(fout, " %s = %s;",
5627 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5628 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
5629 NULL, 1));
5630 break;
5631
5632 case OP_MOVZX:
5633 assert_operand_cnt(2);
5634 fprintf(fout, " %s = %s;",
5635 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5636 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5637 break;
5638
5639 case OP_MOVSX:
5640 assert_operand_cnt(2);
5641 switch (po->operand[1].lmod) {
5642 case OPLM_BYTE:
5643 strcpy(buf3, "(s8)");
5644 break;
5645 case OPLM_WORD:
5646 strcpy(buf3, "(s16)");
5647 break;
5648 default:
5649 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
5650 }
5651 fprintf(fout, " %s = %s;",
5652 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5653 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
5654 buf3, 0));
5655 break;
5656
5657 case OP_XCHG:
5658 assert_operand_cnt(2);
5659 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5660 fprintf(fout, " tmp = %s;",
5661 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], "", 0));
5662 fprintf(fout, " %s = %s;",
5663 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5664 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
5665 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
5666 fprintf(fout, " %s = %stmp;",
5667 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]),
5668 default_cast_to(buf3, sizeof(buf3), &po->operand[1]));
5669 snprintf(g_comment, sizeof(g_comment), "xchg");
5670 break;
5671
5672 case OP_NOT:
5673 assert_operand_cnt(1);
5674 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5675 fprintf(fout, " %s = ~%s;", buf1, buf1);
5676 break;
5677
5678 case OP_XLAT:
5679 assert_operand_cnt(2);
5680 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5681 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
5682 fprintf(fout, " %s = *(u8 *)(%s + %s);", buf1, buf2, buf1);
5683 strcpy(g_comment, "xlat");
5684 break;
5685
5686 case OP_CDQ:
5687 assert_operand_cnt(2);
5688 fprintf(fout, " %s = (s32)%s >> 31;",
5689 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5690 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5691 strcpy(g_comment, "cdq");
5692 break;
5693
5694 case OP_LODS:
5695 if (po->flags & OPF_REP) {
5696 assert_operand_cnt(3);
5697 // hmh..
5698 ferr(po, "TODO\n");
5699 }
5700 else {
5701 assert_operand_cnt(2);
5702 fprintf(fout, " %s = %sesi; esi %c= %d;",
5703 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]),
5704 lmod_cast_u_ptr(po, po->operand[1].lmod),
5705 (po->flags & OPF_DF) ? '-' : '+',
5706 lmod_bytes(po, po->operand[1].lmod));
5707 strcpy(g_comment, "lods");
5708 }
5709 break;
5710
5711 case OP_STOS:
5712 if (po->flags & OPF_REP) {
5713 assert_operand_cnt(3);
5714 fprintf(fout, " for (; ecx != 0; ecx--, edi %c= %d)\n",
5715 (po->flags & OPF_DF) ? '-' : '+',
5716 lmod_bytes(po, po->operand[1].lmod));
5717 fprintf(fout, " %sedi = eax;",
5718 lmod_cast_u_ptr(po, po->operand[1].lmod));
5719 strcpy(g_comment, "rep stos");
5720 }
5721 else {
5722 assert_operand_cnt(2);
5723 fprintf(fout, " %sedi = eax; edi %c= %d;",
5724 lmod_cast_u_ptr(po, po->operand[1].lmod),
5725 (po->flags & OPF_DF) ? '-' : '+',
5726 lmod_bytes(po, po->operand[1].lmod));
5727 strcpy(g_comment, "stos");
5728 }
5729 break;
5730
5731 case OP_MOVS:
5732 j = lmod_bytes(po, po->operand[0].lmod);
5733 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
5734 l = (po->flags & OPF_DF) ? '-' : '+';
5735 if (po->flags & OPF_REP) {
5736 assert_operand_cnt(3);
5737 fprintf(fout,
5738 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n",
5739 l, j, l, j);
5740 fprintf(fout,
5741 " %sedi = %sesi;", buf1, buf1);
5742 strcpy(g_comment, "rep movs");
5743 }
5744 else {
5745 assert_operand_cnt(2);
5746 fprintf(fout, " %sedi = %sesi; edi %c= %d; esi %c= %d;",
5747 buf1, buf1, l, j, l, j);
5748 strcpy(g_comment, "movs");
5749 }
5750 break;
5751
5752 case OP_CMPS:
5753 // repe ~ repeat while ZF=1
5754 j = lmod_bytes(po, po->operand[0].lmod);
5755 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
5756 l = (po->flags & OPF_DF) ? '-' : '+';
5757 if (po->flags & OPF_REP) {
5758 assert_operand_cnt(3);
5759 fprintf(fout,
5760 " for (; ecx != 0; ecx--) {\n");
5761 if (pfomask & (1 << PFO_C)) {
5762 // ugh..
5763 fprintf(fout,
5764 " cond_c = %sesi < %sedi;\n", buf1, buf1);
5765 pfomask &= ~(1 << PFO_C);
5766 }
5767 fprintf(fout,
5768 " cond_z = (%sesi == %sedi); esi %c= %d, edi %c= %d;\n",
5769 buf1, buf1, l, j, l, j);
5770 fprintf(fout,
5771 " if (cond_z %s 0) break;\n",
5772 (po->flags & OPF_REPZ) ? "==" : "!=");
5773 fprintf(fout,
5774 " }");
5775 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
5776 (po->flags & OPF_REPZ) ? "e" : "ne");
5777 }
5778 else {
5779 assert_operand_cnt(2);
5780 fprintf(fout,
5781 " cond_z = (%sesi == %sedi); esi %c= %d; edi %c= %d;",
5782 buf1, buf1, l, j, l, j);
5783 strcpy(g_comment, "cmps");
5784 }
5785 pfomask &= ~(1 << PFO_Z);
5786 last_arith_dst = NULL;
5787 delayed_flag_op = NULL;
5788 break;
5789
5790 case OP_SCAS:
5791 // only does ZF (for now)
5792 // repe ~ repeat while ZF=1
5793 j = lmod_bytes(po, po->operand[1].lmod);
5794 l = (po->flags & OPF_DF) ? '-' : '+';
5795 if (po->flags & OPF_REP) {
5796 assert_operand_cnt(3);
5797 fprintf(fout,
5798 " for (; ecx != 0; ecx--) {\n");
5799 fprintf(fout,
5800 " cond_z = (%seax == %sedi); edi %c= %d;\n",
5801 lmod_cast_u(po, po->operand[1].lmod),
5802 lmod_cast_u_ptr(po, po->operand[1].lmod), l, j);
5803 fprintf(fout,
5804 " if (cond_z %s 0) break;\n",
5805 (po->flags & OPF_REPZ) ? "==" : "!=");
5806 fprintf(fout,
5807 " }");
5808 snprintf(g_comment, sizeof(g_comment), "rep%s scas",
5809 (po->flags & OPF_REPZ) ? "e" : "ne");
5810 }
5811 else {
5812 assert_operand_cnt(2);
5813 fprintf(fout, " cond_z = (%seax == %sedi); edi %c= %d;",
5814 lmod_cast_u(po, po->operand[1].lmod),
5815 lmod_cast_u_ptr(po, po->operand[1].lmod), l, j);
5816 strcpy(g_comment, "scas");
5817 }
5818 pfomask &= ~(1 << PFO_Z);
5819 last_arith_dst = NULL;
5820 delayed_flag_op = NULL;
5821 break;
5822
5823 // arithmetic w/flags
5824 case OP_AND:
5825 if (po->operand[1].type == OPT_CONST && !po->operand[1].val)
5826 goto dualop_arith_const;
5827 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5828 goto dualop_arith;
5829
5830 case OP_OR:
5831 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5832 if (po->operand[1].type == OPT_CONST) {
5833 j = lmod_bytes(po, po->operand[0].lmod);
5834 if (((1ull << j * 8) - 1) == po->operand[1].val)
5835 goto dualop_arith_const;
5836 }
5837 goto dualop_arith;
5838
5839 dualop_arith:
5840 assert_operand_cnt(2);
5841 fprintf(fout, " %s %s= %s;",
5842 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5843 op_to_c(po),
5844 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5845 output_std_flags(fout, po, &pfomask, buf1);
5846 last_arith_dst = &po->operand[0];
5847 delayed_flag_op = NULL;
5848 break;
5849
5850 dualop_arith_const:
5851 // and 0, or ~0 used instead mov
5852 assert_operand_cnt(2);
5853 fprintf(fout, " %s = %s;",
5854 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5855 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
5856 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
5857 output_std_flags(fout, po, &pfomask, buf1);
5858 last_arith_dst = &po->operand[0];
5859 delayed_flag_op = NULL;
5860 break;
5861
5862 case OP_SHL:
5863 case OP_SHR:
5864 assert_operand_cnt(2);
5865 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5866 if (pfomask & (1 << PFO_C)) {
5867 if (po->operand[1].type == OPT_CONST) {
5868 l = lmod_bytes(po, po->operand[0].lmod) * 8;
5869 j = po->operand[1].val;
5870 j %= l;
5871 if (j != 0) {
5872 if (po->op == OP_SHL)
5873 j = l - j;
5874 else
5875 j -= 1;
5876 fprintf(fout, " cond_c = (%s >> %d) & 1;\n",
5877 buf1, j);
5878 }
5879 else
5880 ferr(po, "zero shift?\n");
5881 }
5882 else
5883 ferr(po, "TODO\n");
5884 pfomask &= ~(1 << PFO_C);
5885 }
5886 fprintf(fout, " %s %s= %s", buf1, op_to_c(po),
5887 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5888 if (po->operand[1].type != OPT_CONST)
5889 fprintf(fout, " & 0x1f");
5890 fprintf(fout, ";");
5891 output_std_flags(fout, po, &pfomask, buf1);
5892 last_arith_dst = &po->operand[0];
5893 delayed_flag_op = NULL;
5894 break;
5895
5896 case OP_SAR:
5897 assert_operand_cnt(2);
5898 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5899 fprintf(fout, " %s = %s%s >> %s;", buf1,
5900 lmod_cast_s(po, po->operand[0].lmod), buf1,
5901 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5902 output_std_flags(fout, po, &pfomask, buf1);
5903 last_arith_dst = &po->operand[0];
5904 delayed_flag_op = NULL;
5905 break;
5906
5907 case OP_SHLD:
5908 case OP_SHRD:
5909 assert_operand_cnt(3);
5910 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5911 l = lmod_bytes(po, po->operand[0].lmod) * 8;
5912 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[2]);
5913 if (po->operand[2].type != OPT_CONST) {
5914 // no handling for "undefined" case, hopefully not needed
5915 snprintf(buf2, sizeof(buf2), "(%s & 0x1f)", buf3);
5916 strcpy(buf3, buf2);
5917 }
5918 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
5919 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5920 if (po->op == OP_SHLD) {
5921 fprintf(fout, " %s <<= %s; %s |= %s >> (%d - %s);",
5922 buf1, buf3, buf1, buf2, l, buf3);
5923 strcpy(g_comment, "shld");
5924 }
5925 else {
5926 fprintf(fout, " %s >>= %s; %s |= %s << (%d - %s);",
5927 buf1, buf3, buf1, buf2, l, buf3);
5928 strcpy(g_comment, "shrd");
5929 }
5930 output_std_flags(fout, po, &pfomask, buf1);
5931 last_arith_dst = &po->operand[0];
5932 delayed_flag_op = NULL;
5933 break;
5934
5935 case OP_ROL:
5936 case OP_ROR:
5937 assert_operand_cnt(2);
5938 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5939 if (po->operand[1].type == OPT_CONST) {
5940 j = po->operand[1].val;
5941 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
5942 fprintf(fout, po->op == OP_ROL ?
5943 " %s = (%s << %d) | (%s >> %d);" :
5944 " %s = (%s >> %d) | (%s << %d);",
5945 buf1, buf1, j, buf1,
5946 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
5947 }
5948 else
5949 ferr(po, "TODO\n");
5950 output_std_flags(fout, po, &pfomask, buf1);
5951 last_arith_dst = &po->operand[0];
5952 delayed_flag_op = NULL;
5953 break;
5954
5955 case OP_RCL:
5956 case OP_RCR:
5957 assert_operand_cnt(2);
5958 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5959 l = lmod_bytes(po, po->operand[0].lmod) * 8;
5960 if (po->operand[1].type == OPT_CONST) {
5961 j = po->operand[1].val % l;
5962 if (j == 0)
5963 ferr(po, "zero rotate\n");
5964 fprintf(fout, " tmp = (%s >> %d) & 1;\n",
5965 buf1, (po->op == OP_RCL) ? (l - j) : (j - 1));
5966 if (po->op == OP_RCL) {
5967 fprintf(fout,
5968 " %s = (%s << %d) | (cond_c << %d)",
5969 buf1, buf1, j, j - 1);
5970 if (j != 1)
5971 fprintf(fout, " | (%s >> %d)", buf1, l + 1 - j);
5972 }
5973 else {
5974 fprintf(fout,
5975 " %s = (%s >> %d) | (cond_c << %d)",
5976 buf1, buf1, j, l - j);
5977 if (j != 1)
5978 fprintf(fout, " | (%s << %d)", buf1, l + 1 - j);
5979 }
5980 fprintf(fout, ";\n");
5981 fprintf(fout, " cond_c = tmp;");
5982 }
5983 else
5984 ferr(po, "TODO\n");
5985 strcpy(g_comment, (po->op == OP_RCL) ? "rcl" : "rcr");
5986 output_std_flags(fout, po, &pfomask, buf1);
5987 last_arith_dst = &po->operand[0];
5988 delayed_flag_op = NULL;
5989 break;
5990
5991 case OP_XOR:
5992 assert_operand_cnt(2);
5993 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5994 if (IS(opr_name(po, 0), opr_name(po, 1))) {
5995 // special case for XOR
5996 if (pfomask & (1 << PFO_BE)) { // weird, but it happens..
5997 fprintf(fout, " cond_be = 1;\n");
5998 pfomask &= ~(1 << PFO_BE);
5999 }
6000 fprintf(fout, " %s = 0;",
6001 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
6002 last_arith_dst = &po->operand[0];
6003 delayed_flag_op = NULL;
6004 break;
6005 }
6006 goto dualop_arith;
6007
6008 case OP_ADD:
6009 assert_operand_cnt(2);
6010 propagate_lmod(po, &po->operand[0], &po->operand[1]);
6011 if (pfomask & (1 << PFO_C)) {
6012 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
6013 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
6014 if (po->operand[0].lmod == OPLM_DWORD) {
6015 fprintf(fout, " tmp64 = (u64)%s + %s;\n", buf1, buf2);
6016 fprintf(fout, " cond_c = tmp64 >> 32;\n");
6017 fprintf(fout, " %s = (u32)tmp64;",
6018 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
6019 strcat(g_comment, " add64");
6020 }
6021 else {
6022 fprintf(fout, " cond_c = ((u32)%s + %s) >> %d;\n",
6023 buf1, buf2, lmod_bytes(po, po->operand[0].lmod) * 8);
6024 fprintf(fout, " %s += %s;",
6025 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
6026 buf2);
6027 }
6028 pfomask &= ~(1 << PFO_C);
6029 output_std_flags(fout, po, &pfomask, buf1);
6030 last_arith_dst = &po->operand[0];
6031 delayed_flag_op = NULL;
6032 break;
6033 }
6034 goto dualop_arith;
6035
6036 case OP_SUB:
6037 assert_operand_cnt(2);
6038 propagate_lmod(po, &po->operand[0], &po->operand[1]);
6039 if (pfomask & ~((1 << PFO_Z) | (1 << PFO_S))) {
6040 for (j = 0; j <= PFO_LE; j++) {
6041 if (!(pfomask & (1 << j)))
6042 continue;
6043 if (j == PFO_Z || j == PFO_S)
6044 continue;
6045
6046 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
6047 fprintf(fout, " cond_%s = %s;\n",
6048 parsed_flag_op_names[j], buf1);
6049 pfomask &= ~(1 << j);
6050 }
6051 }
6052 goto dualop_arith;
6053
6054 case OP_ADC:
6055 case OP_SBB:
6056 assert_operand_cnt(2);
6057 propagate_lmod(po, &po->operand[0], &po->operand[1]);
6058 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
6059 if (po->op == OP_SBB
6060 && IS(po->operand[0].name, po->operand[1].name))
6061 {
6062 // avoid use of unitialized var
6063 fprintf(fout, " %s = -cond_c;", buf1);
6064 // carry remains what it was
6065 pfomask &= ~(1 << PFO_C);
6066 }
6067 else {
6068 fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po),
6069 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
6070 }
6071 output_std_flags(fout, po, &pfomask, buf1);
6072 last_arith_dst = &po->operand[0];
6073 delayed_flag_op = NULL;
6074 break;
6075
6076 case OP_BSF:
6077 assert_operand_cnt(2);
6078 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
6079 fprintf(fout, " %s = %s ? __builtin_ffs(%s) - 1 : 0;",
6080 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
6081 buf2, buf2);
6082 output_std_flags(fout, po, &pfomask, buf1);
6083 last_arith_dst = &po->operand[0];
6084 delayed_flag_op = NULL;
6085 strcat(g_comment, " bsf");
6086 break;
6087
6088 case OP_DEC:
6089 if (pfomask & ~(PFOB_S | PFOB_S | PFOB_C)) {
6090 for (j = 0; j <= PFO_LE; j++) {
6091 if (!(pfomask & (1 << j)))
6092 continue;
6093 if (j == PFO_Z || j == PFO_S || j == PFO_C)
6094 continue;
6095
6096 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
6097 fprintf(fout, " cond_%s = %s;\n",
6098 parsed_flag_op_names[j], buf1);
6099 pfomask &= ~(1 << j);
6100 }
6101 }
6102 // fallthrough
6103
6104 case OP_INC:
6105 if (pfomask & (1 << PFO_C))
6106 // carry is unaffected by inc/dec.. wtf?
6107 ferr(po, "carry propagation needed\n");
6108
6109 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
6110 if (po->operand[0].type == OPT_REG) {
6111 strcpy(buf2, po->op == OP_INC ? "++" : "--");
6112 fprintf(fout, " %s%s;", buf1, buf2);
6113 }
6114 else {
6115 strcpy(buf2, po->op == OP_INC ? "+" : "-");
6116 fprintf(fout, " %s %s= 1;", buf1, buf2);
6117 }
6118 output_std_flags(fout, po, &pfomask, buf1);
6119 last_arith_dst = &po->operand[0];
6120 delayed_flag_op = NULL;
6121 break;
6122
6123 case OP_NEG:
6124 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
6125 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
6126 fprintf(fout, " %s = -%s%s;", buf1,
6127 lmod_cast_s(po, po->operand[0].lmod), buf2);
6128 last_arith_dst = &po->operand[0];
6129 delayed_flag_op = NULL;
6130 if (pfomask & (1 << PFO_C)) {
6131 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
6132 pfomask &= ~(1 << PFO_C);
6133 }
6134 break;
6135
6136 case OP_IMUL:
6137 if (po->operand_cnt == 2) {
6138 propagate_lmod(po, &po->operand[0], &po->operand[1]);
6139 goto dualop_arith;
6140 }
6141 if (po->operand_cnt == 3)
6142 ferr(po, "TODO imul3\n");
6143 // fallthrough
6144 case OP_MUL:
6145 assert_operand_cnt(1);
6146 switch (po->operand[0].lmod) {
6147 case OPLM_DWORD:
6148 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
6149 fprintf(fout, " tmp64 = %seax * %s%s;\n", buf1, buf1,
6150 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
6151 fprintf(fout, " edx = tmp64 >> 32;\n");
6152 fprintf(fout, " eax = tmp64;");
6153 break;
6154 case OPLM_BYTE:
6155 strcpy(buf1, po->op == OP_IMUL ? "(s16)(s8)" : "(u16)(u8)");
6156 fprintf(fout, " LOWORD(eax) = %seax * %s;", buf1,
6157 out_src_opr(buf2, sizeof(buf2), po, &po->operand[0],
6158 buf1, 0));
6159 break;
6160 default:
6161 ferr(po, "TODO: unhandled mul type\n");
6162 break;
6163 }
6164 last_arith_dst = NULL;
6165 delayed_flag_op = NULL;
6166 break;
6167
6168 case OP_DIV:
6169 case OP_IDIV:
6170 assert_operand_cnt(1);
6171 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
6172 strcpy(cast, lmod_cast(po, po->operand[0].lmod,
6173 po->op == OP_IDIV));
6174 switch (po->operand[0].lmod) {
6175 case OPLM_DWORD:
6176 if (po->flags & OPF_32BIT)
6177 snprintf(buf2, sizeof(buf2), "%seax", cast);
6178 else {
6179 fprintf(fout, " tmp64 = ((u64)edx << 32) | eax;\n");
6180 snprintf(buf2, sizeof(buf2), "%stmp64",
6181 (po->op == OP_IDIV) ? "(s64)" : "");
6182 }
6183 if (po->operand[0].type == OPT_REG
6184 && po->operand[0].reg == xDX)
6185 {
6186 fprintf(fout, " eax = %s / %s%s;\n", buf2, cast, buf1);
6187 fprintf(fout, " edx = %s %% %s%s;", buf2, cast, buf1);
6188 }
6189 else {
6190 fprintf(fout, " edx = %s %% %s%s;\n", buf2, cast, buf1);
6191 fprintf(fout, " eax = %s / %s%s;", buf2, cast, buf1);
6192 }
6193 break;
6194 case OPLM_WORD:
6195 fprintf(fout, " tmp = (edx << 16) | (eax & 0xffff);\n");
6196 snprintf(buf2, sizeof(buf2), "%stmp",
6197 (po->op == OP_IDIV) ? "(s32)" : "");
6198 if (po->operand[0].type == OPT_REG
6199 && po->operand[0].reg == xDX)
6200 {
6201 fprintf(fout, " LOWORD(eax) = %s / %s%s;\n",
6202 buf2, cast, buf1);
6203 fprintf(fout, " LOWORD(edx) = %s %% %s%s;",
6204 buf2, cast, buf1);
6205 }
6206 else {
6207 fprintf(fout, " LOWORD(edx) = %s %% %s%s;\n",
6208 buf2, cast, buf1);
6209 fprintf(fout, " LOWORD(eax) = %s / %s%s;",
6210 buf2, cast, buf1);
6211 }
6212 strcat(g_comment, " div16");
6213 break;
6214 default:
6215 ferr(po, "unhandled div lmod %d\n", po->operand[0].lmod);
6216 }
6217 last_arith_dst = NULL;
6218 delayed_flag_op = NULL;
6219 break;
6220
6221 case OP_TEST:
6222 case OP_CMP:
6223 propagate_lmod(po, &po->operand[0], &po->operand[1]);
6224 if (pfomask != 0) {
6225 for (j = 0; j < 8; j++) {
6226 if (pfomask & (1 << j)) {
6227 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
6228 fprintf(fout, " cond_%s = %s;",
6229 parsed_flag_op_names[j], buf1);
6230 }
6231 }
6232 pfomask = 0;
6233 }
6234 else
6235 no_output = 1;
6236 last_arith_dst = NULL;
6237 delayed_flag_op = po;
6238 break;
6239
6240 case OP_SCC:
6241 // SETcc - should already be handled
6242 break;
6243
6244 // note: we reuse OP_Jcc for SETcc, only flags differ
6245 case OP_JCC:
6246 fprintf(fout, "\n goto %s;", po->operand[0].name);
6247 break;
6248
6249 case OP_JECXZ:
6250 fprintf(fout, " if (ecx == 0)\n");
6251 fprintf(fout, " goto %s;", po->operand[0].name);
6252 strcat(g_comment, " jecxz");
6253 break;
6254
6255 case OP_LOOP:
6256 fprintf(fout, " if (--ecx != 0)\n");
6257 fprintf(fout, " goto %s;", po->operand[0].name);
6258 strcat(g_comment, " loop");
6259 break;
6260
6261 case OP_JMP:
6262 assert_operand_cnt(1);
6263 last_arith_dst = NULL;
6264 delayed_flag_op = NULL;
6265
6266 if (po->operand[0].type == OPT_REGMEM) {
6267 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
6268 buf1, buf2);
6269 if (ret != 2)
6270 ferr(po, "parse failure for jmp '%s'\n",
6271 po->operand[0].name);
6272 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
6273 break;
6274 }
6275 else if (po->operand[0].type != OPT_LABEL)
6276 ferr(po, "unhandled jmp type\n");
6277
6278 fprintf(fout, " goto %s;", po->operand[0].name);
6279 break;
6280
6281 case OP_CALL:
6282 assert_operand_cnt(1);
6283 pp = po->pp;
6284 my_assert_not(pp, NULL);
6285
6286 strcpy(buf3, " ");
6287 if (po->flags & OPF_CC) {
6288 // we treat conditional branch to another func
6289 // (yes such code exists..) as conditional tailcall
6290 strcat(buf3, " ");
6291 fprintf(fout, " {\n");
6292 }
6293
6294 if (pp->is_fptr && !pp->is_arg) {
6295 fprintf(fout, "%s%s = %s;\n", buf3, pp->name,
6296 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
6297 "(void *)", 0));
6298 if (pp->is_unresolved)
6299 fprintf(fout, "%sunresolved_call(\"%s:%d\", %s);\n",
6300 buf3, asmfn, po->asmln, pp->name);
6301 }
6302
6303 fprintf(fout, "%s", buf3);
6304 if (strstr(pp->ret_type.name, "int64")) {
6305 if (po->flags & OPF_TAIL)
6306 ferr(po, "int64 and tail?\n");
6307 fprintf(fout, "tmp64 = ");
6308 }
6309 else if (!IS(pp->ret_type.name, "void")) {
6310 if (po->flags & OPF_TAIL) {
6311 if (regmask_ret & mxAX) {
6312 fprintf(fout, "return ");
6313 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
6314 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
6315 }
6316 else if (regmask_ret & mxST0)
6317 ferr(po, "float tailcall\n");
6318 }
6319 else if (po->regmask_dst & mxAX) {
6320 fprintf(fout, "eax = ");
6321 if (pp->ret_type.is_ptr)
6322 fprintf(fout, "(u32)");
6323 }
6324 else if (po->regmask_dst & mxST0) {
6325 fprintf(fout, "f_st0 = ");
6326 }
6327 }
6328
6329 if (pp->name[0] == 0)
6330 ferr(po, "missing pp->name\n");
6331 fprintf(fout, "%s%s(", pp->name,
6332 pp->has_structarg ? "_sa" : "");
6333
6334 if (po->flags & OPF_ATAIL) {
6335 if (pp->argc_stack != g_func_pp->argc_stack
6336 || (pp->argc_stack > 0
6337 && pp->is_stdcall != g_func_pp->is_stdcall))
6338 ferr(po, "incompatible tailcall\n");
6339 if (g_func_pp->has_retreg)
6340 ferr(po, "TODO: retreg+tailcall\n");
6341
6342 for (arg = j = 0; arg < pp->argc; arg++) {
6343 if (arg > 0)
6344 fprintf(fout, ", ");
6345
6346 cast[0] = 0;
6347 if (pp->arg[arg].type.is_ptr)
6348 snprintf(cast, sizeof(cast), "(%s)",
6349 pp->arg[arg].type.name);
6350
6351 if (pp->arg[arg].reg != NULL) {
6352 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
6353 continue;
6354 }
6355 // stack arg
6356 for (; j < g_func_pp->argc; j++)
6357 if (g_func_pp->arg[j].reg == NULL)
6358 break;
6359 fprintf(fout, "%sa%d", cast, j + 1);
6360 j++;
6361 }
6362 }
6363 else {
6364 for (arg = 0; arg < pp->argc; arg++) {
6365 if (arg > 0)
6366 fprintf(fout, ", ");
6367
6368 cast[0] = 0;
6369 if (pp->arg[arg].type.is_ptr)
6370 snprintf(cast, sizeof(cast), "(%s)",
6371 pp->arg[arg].type.name);
6372
6373 if (pp->arg[arg].reg != NULL) {
6374 if (pp->arg[arg].type.is_retreg)
6375 fprintf(fout, "&%s", pp->arg[arg].reg);
6376 else
6377 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
6378 continue;
6379 }
6380
6381 // stack arg
6382 tmp_op = pp->arg[arg].datap;
6383 if (tmp_op == NULL)
6384 ferr(po, "parsed_op missing for arg%d\n", arg);
6385
6386 if (tmp_op->flags & OPF_VAPUSH) {
6387 fprintf(fout, "ap");
6388 }
6389 else if (tmp_op->p_argpass != 0) {
6390 fprintf(fout, "a%d", tmp_op->p_argpass);
6391 }
6392 else if (tmp_op->p_argnum != 0) {
6393 fprintf(fout, "%s%s", cast,
6394 saved_arg_name(buf1, sizeof(buf1),
6395 tmp_op->p_arggrp, tmp_op->p_argnum));
6396 }
6397 else {
6398 fprintf(fout, "%s",
6399 out_src_opr(buf1, sizeof(buf1),
6400 tmp_op, &tmp_op->operand[0], cast, 0));
6401 }
6402 }
6403 }
6404 fprintf(fout, ");");
6405
6406 if (strstr(pp->ret_type.name, "int64")) {
6407 fprintf(fout, "\n");
6408 fprintf(fout, "%sedx = tmp64 >> 32;\n", buf3);
6409 fprintf(fout, "%seax = tmp64;", buf3);
6410 }
6411
6412 if (pp->is_unresolved) {
6413 snprintf(buf2, sizeof(buf2), " unresolved %dreg",
6414 pp->argc_reg);
6415 strcat(g_comment, buf2);
6416 }
6417
6418 if (po->flags & OPF_TAIL) {
6419 ret = 0;
6420 if (i == opcnt - 1 || pp->is_noreturn)
6421 ret = 0;
6422 else if (IS(pp->ret_type.name, "void"))
6423 ret = 1;
6424 else if (!(regmask_ret & (1 << xAX)))
6425 ret = 1;
6426 // else already handled as 'return f()'
6427
6428 if (ret) {
6429 fprintf(fout, "\n%sreturn;", buf3);
6430 strcat(g_comment, " ^ tailcall");
6431 }
6432 else
6433 strcat(g_comment, " tailcall");
6434
6435 if ((regmask_ret & (1 << xAX))
6436 && IS(pp->ret_type.name, "void") && !pp->is_noreturn)
6437 {
6438 ferr(po, "int func -> void func tailcall?\n");
6439 }
6440 }
6441 if (pp->is_noreturn)
6442 strcat(g_comment, " noreturn");
6443 if ((po->flags & OPF_ATAIL) && pp->argc_stack > 0)
6444 strcat(g_comment, " argframe");
6445 if (po->flags & OPF_CC)
6446 strcat(g_comment, " cond");
6447
6448 if (po->flags & OPF_CC)
6449 fprintf(fout, "\n }");
6450
6451 delayed_flag_op = NULL;
6452 last_arith_dst = NULL;
6453 break;
6454
6455 case OP_RET:
6456 if (g_func_pp->is_vararg)
6457 fprintf(fout, " va_end(ap);\n");
6458 if (g_func_pp->has_retreg) {
6459 for (arg = 0; arg < g_func_pp->argc; arg++)
6460 if (g_func_pp->arg[arg].type.is_retreg)
6461 fprintf(fout, " *r_%s = %s;\n",
6462 g_func_pp->arg[arg].reg, g_func_pp->arg[arg].reg);
6463 }
6464
6465 if (!(regmask_ret & (1 << xAX))) {
6466 if (i != opcnt - 1 || label_pending)
6467 fprintf(fout, " return;");
6468 }
6469 else if (g_func_pp->ret_type.is_ptr) {
6470 fprintf(fout, " return (%s)eax;",
6471 g_func_pp->ret_type.name);
6472 }
6473 else if (IS(g_func_pp->ret_type.name, "__int64"))
6474 fprintf(fout, " return ((u64)edx << 32) | eax;");
6475 else
6476 fprintf(fout, " return eax;");
6477
6478 last_arith_dst = NULL;
6479 delayed_flag_op = NULL;
6480 break;
6481
6482 case OP_PUSH:
6483 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
6484 if (po->p_argnum != 0) {
6485 // special case - saved func arg
6486 fprintf(fout, " %s = %s;",
6487 saved_arg_name(buf2, sizeof(buf2),
6488 po->p_arggrp, po->p_argnum), buf1);
6489 break;
6490 }
6491 else if (po->flags & OPF_RSAVE) {
6492 fprintf(fout, " s_%s = %s;", buf1, buf1);
6493 break;
6494 }
6495 else if (po->flags & OPF_PPUSH) {
6496 tmp_op = po->datap;
6497 ferr_assert(po, tmp_op != NULL);
6498 out_dst_opr(buf2, sizeof(buf2), po, &tmp_op->operand[0]);
6499 fprintf(fout, " pp_%s = %s;", buf2, buf1);
6500 break;
6501 }
6502 else if (g_func_pp->is_userstack) {
6503 fprintf(fout, " *(--esp) = %s;", buf1);
6504 break;
6505 }
6506 if (!(g_ida_func_attr & IDAFA_NORETURN))
6507 ferr(po, "stray push encountered\n");
6508 no_output = 1;
6509 break;
6510
6511 case OP_POP:
6512 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
6513 if (po->flags & OPF_RSAVE) {
6514 fprintf(fout, " %s = s_%s;", buf1, buf1);
6515 break;
6516 }
6517 else if (po->flags & OPF_PPUSH) {
6518 // push/pop graph / non-const
6519 ferr_assert(po, po->datap == NULL);
6520 fprintf(fout, " %s = pp_%s;", buf1, buf1);
6521 break;
6522 }
6523 else if (po->datap != NULL) {
6524 // push/pop pair
6525 tmp_op = po->datap;
6526 fprintf(fout, " %s = %s;", buf1,
6527 out_src_opr(buf2, sizeof(buf2),
6528 tmp_op, &tmp_op->operand[0],
6529 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
6530 break;
6531 }
6532 else if (g_func_pp->is_userstack) {
6533 fprintf(fout, " %s = *esp++;", buf1);
6534 break;
6535 }
6536 else
6537 ferr(po, "stray pop encountered\n");
6538 break;
6539
6540 case OP_NOP:
6541 no_output = 1;
6542 break;
6543
6544 // x87
6545 case OP_FLD:
6546 if (po->flags & OPF_FSHIFT)
6547 fprintf(fout, " f_st1 = f_st0;\n");
6548 if (po->operand[0].type == OPT_REG
6549 && po->operand[0].reg == xST0)
6550 {
6551 strcat(g_comment, " fld st");
6552 break;
6553 }
6554 fprintf(fout, " f_st0 = %s;",
6555 out_src_opr_float(buf1, sizeof(buf1), po, &po->operand[0]));
6556 strcat(g_comment, " fld");
6557 break;
6558
6559 case OP_FILD:
6560 if (po->flags & OPF_FSHIFT)
6561 fprintf(fout, " f_st1 = f_st0;\n");
6562 fprintf(fout, " f_st0 = (double)%s;",
6563 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
6564 lmod_cast(po, po->operand[0].lmod, 1), 0));
6565 strcat(g_comment, " fild");
6566 break;
6567
6568 case OP_FLDc:
6569 if (po->flags & OPF_FSHIFT)
6570 fprintf(fout, " f_st1 = f_st0;\n");
6571 fprintf(fout, " f_st0 = ");
6572 switch (po->operand[0].val) {
6573 case X87_CONST_1: fprintf(fout, "1.0;"); break;
6574 case X87_CONST_Z: fprintf(fout, "0.0;"); break;
6575 default: ferr(po, "TODO\n"); break;
6576 }
6577 break;
6578
6579 case OP_FST:
6580 if ((po->flags & OPF_FPOP) && po->operand[0].type == OPT_REG
6581 && po->operand[0].reg == xST0)
6582 {
6583 no_output = 1;
6584 break;
6585 }
6586 fprintf(fout, " %s = f_st0;",
6587 out_dst_opr_float(buf1, sizeof(buf1), po, &po->operand[0]));
6588 if (po->flags & OPF_FSHIFT)
6589 fprintf(fout, "\n f_st0 = f_st1;");
6590 strcat(g_comment, " fst");
6591 break;
6592
6593 case OP_FADD:
6594 case OP_FDIV:
6595 case OP_FMUL:
6596 case OP_FSUB:
6597 switch (po->op) {
6598 case OP_FADD: j = '+'; break;
6599 case OP_FDIV: j = '/'; break;
6600 case OP_FMUL: j = '*'; break;
6601 case OP_FSUB: j = '-'; break;
6602 default: j = 'x'; break;
6603 }
6604 if (po->flags & OPF_FSHIFT) {
6605 fprintf(fout, " f_st0 = f_st1 %c f_st0;", j);
6606 }
6607 else {
6608 fprintf(fout, " %s %c= %s;",
6609 out_dst_opr_float(buf1, sizeof(buf1), po, &po->operand[0]),
6610 j,
6611 out_src_opr_float(buf2, sizeof(buf2), po, &po->operand[1]));
6612 }
6613 break;
6614
6615 case OP_FDIVR:
6616 case OP_FSUBR:
6617 if (po->flags & OPF_FSHIFT)
6618 snprintf(buf1, sizeof(buf1), "f_st0");
6619 else
6620 out_dst_opr_float(buf1, sizeof(buf1), po, &po->operand[0]);
6621 fprintf(fout, " %s = %s %c %s;", buf1,
6622 out_src_opr_float(buf2, sizeof(buf2), po, &po->operand[1]),
6623 po->op == OP_FDIVR ? '/' : '-',
6624 out_src_opr_float(buf3, sizeof(buf3), po, &po->operand[0]));
6625 break;
6626
6627 case OP_FIADD:
6628 case OP_FIDIV:
6629 case OP_FIMUL:
6630 case OP_FISUB:
6631 switch (po->op) {
6632 case OP_FIADD: j = '+'; break;
6633 case OP_FIDIV: j = '/'; break;
6634 case OP_FIMUL: j = '*'; break;
6635 case OP_FISUB: j = '-'; break;
6636 default: j = 'x'; break;
6637 }
6638 fprintf(fout, " f_st0 %c= (double)%s;", j,
6639 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
6640 lmod_cast(po, po->operand[0].lmod, 1), 0));
6641 break;
6642
6643 case OP_FIDIVR:
6644 case OP_FISUBR:
6645 fprintf(fout, " f_st0 = %s %c f_st0;",
6646 out_src_opr_float(buf2, sizeof(buf2), po, &po->operand[1]),
6647 po->op == OP_FIDIVR ? '/' : '-');
6648 break;
6649
6650 case OPP_FTOL:
6651 ferr_assert(po, po->flags & OPF_32BIT);
6652 fprintf(fout, " eax = (s32)f_st0;");
6653 if (po->flags & OPF_FSHIFT)
6654 fprintf(fout, "\n f_st0 = f_st1;");
6655 strcat(g_comment, " ftol");
6656 break;
6657
6658 // mmx
6659 case OP_EMMS:
6660 strcpy(g_comment, " (emms)");
6661 break;
6662
6663 default:
6664 no_output = 1;
6665 ferr(po, "unhandled op type %d, flags %x\n",
6666 po->op, po->flags);
6667 break;
6668 }
6669
6670 if (g_comment[0] != 0) {
6671 char *p = g_comment;
6672 while (my_isblank(*p))
6673 p++;
6674 fprintf(fout, " // %s", p);
6675 g_comment[0] = 0;
6676 no_output = 0;
6677 }
6678 if (!no_output)
6679 fprintf(fout, "\n");
6680
6681 // some sanity checking
6682 if (po->flags & OPF_REP) {
6683 if (po->op != OP_STOS && po->op != OP_MOVS
6684 && po->op != OP_CMPS && po->op != OP_SCAS)
6685 ferr(po, "unexpected rep\n");
6686 if (!(po->flags & (OPF_REPZ|OPF_REPNZ))
6687 && (po->op == OP_CMPS || po->op == OP_SCAS))
6688 ferr(po, "cmps/scas with plain rep\n");
6689 }
6690 if ((po->flags & (OPF_REPZ|OPF_REPNZ))
6691 && po->op != OP_CMPS && po->op != OP_SCAS)
6692 ferr(po, "unexpected repz/repnz\n");
6693
6694 if (pfomask != 0)
6695 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
6696
6697 // see is delayed flag stuff is still valid
6698 if (delayed_flag_op != NULL && delayed_flag_op != po) {
6699 if (is_any_opr_modified(delayed_flag_op, po, 0))
6700 delayed_flag_op = NULL;
6701 }
6702
6703 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
6704 if (is_opr_modified(last_arith_dst, po))
6705 last_arith_dst = NULL;
6706 }
6707
6708 label_pending = 0;
6709 }
6710
6711 if (g_stack_fsz && !g_stack_frame_used)
6712 fprintf(fout, " (void)sf;\n");
6713
6714 fprintf(fout, "}\n\n");
6715
6716 gen_x_cleanup(opcnt);
6717}
6718
6719static void gen_x_cleanup(int opcnt)
6720{
6721 int i;
6722
6723 for (i = 0; i < opcnt; i++) {
6724 struct label_ref *lr, *lr_del;
6725
6726 lr = g_label_refs[i].next;
6727 while (lr != NULL) {
6728 lr_del = lr;
6729 lr = lr->next;
6730 free(lr_del);
6731 }
6732 g_label_refs[i].i = -1;
6733 g_label_refs[i].next = NULL;
6734
6735 if (ops[i].op == OP_CALL) {
6736 if (ops[i].pp)
6737 proto_release(ops[i].pp);
6738 }
6739 }
6740 g_func_pp = NULL;
6741}
6742
6743struct func_proto_dep;
6744
6745struct func_prototype {
6746 char name[NAMELEN];
6747 int id;
6748 int argc_stack;
6749 int regmask_dep;
6750 int has_ret:3; // -1, 0, 1: unresolved, no, yes
6751 unsigned int dep_resolved:1;
6752 unsigned int is_stdcall:1;
6753 struct func_proto_dep *dep_func;
6754 int dep_func_cnt;
6755 const struct parsed_proto *pp; // seed pp, if any
6756};
6757
6758struct func_proto_dep {
6759 char *name;
6760 struct func_prototype *proto;
6761 int regmask_live; // .. at the time of call
6762 unsigned int ret_dep:1; // return from this is caller's return
6763};
6764
6765static struct func_prototype *hg_fp;
6766static int hg_fp_cnt;
6767
6768static struct scanned_var {
6769 char name[NAMELEN];
6770 enum opr_lenmod lmod;
6771 unsigned int is_seeded:1;
6772 unsigned int is_c_str:1;
6773 const struct parsed_proto *pp; // seed pp, if any
6774} *hg_vars;
6775static int hg_var_cnt;
6776
6777static char **hg_refs;
6778static int hg_ref_cnt;
6779
6780static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
6781 int count);
6782
6783static struct func_prototype *hg_fp_add(const char *funcn)
6784{
6785 struct func_prototype *fp;
6786
6787 if ((hg_fp_cnt & 0xff) == 0) {
6788 hg_fp = realloc(hg_fp, sizeof(hg_fp[0]) * (hg_fp_cnt + 0x100));
6789 my_assert_not(hg_fp, NULL);
6790 memset(hg_fp + hg_fp_cnt, 0, sizeof(hg_fp[0]) * 0x100);
6791 }
6792
6793 fp = &hg_fp[hg_fp_cnt];
6794 snprintf(fp->name, sizeof(fp->name), "%s", funcn);
6795 fp->id = hg_fp_cnt;
6796 fp->argc_stack = -1;
6797 hg_fp_cnt++;
6798
6799 return fp;
6800}
6801
6802static struct func_proto_dep *hg_fp_find_dep(struct func_prototype *fp,
6803 const char *name)
6804{
6805 int i;
6806
6807 for (i = 0; i < fp->dep_func_cnt; i++)
6808 if (IS(fp->dep_func[i].name, name))
6809 return &fp->dep_func[i];
6810
6811 return NULL;
6812}
6813
6814static void hg_fp_add_dep(struct func_prototype *fp, const char *name)
6815{
6816 // is it a dupe?
6817 if (hg_fp_find_dep(fp, name))
6818 return;
6819
6820 if ((fp->dep_func_cnt & 0xff) == 0) {
6821 fp->dep_func = realloc(fp->dep_func,
6822 sizeof(fp->dep_func[0]) * (fp->dep_func_cnt + 0x100));
6823 my_assert_not(fp->dep_func, NULL);
6824 memset(&fp->dep_func[fp->dep_func_cnt], 0,
6825 sizeof(fp->dep_func[0]) * 0x100);
6826 }
6827 fp->dep_func[fp->dep_func_cnt].name = strdup(name);
6828 fp->dep_func_cnt++;
6829}
6830
6831static int hg_fp_cmp_name(const void *p1_, const void *p2_)
6832{
6833 const struct func_prototype *p1 = p1_, *p2 = p2_;
6834 return strcmp(p1->name, p2->name);
6835}
6836
6837#if 0
6838static int hg_fp_cmp_id(const void *p1_, const void *p2_)
6839{
6840 const struct func_prototype *p1 = p1_, *p2 = p2_;
6841 return p1->id - p2->id;
6842}
6843#endif
6844
6845static void hg_ref_add(const char *name)
6846{
6847 if ((hg_ref_cnt & 0xff) == 0) {
6848 hg_refs = realloc(hg_refs, sizeof(hg_refs[0]) * (hg_ref_cnt + 0x100));
6849 my_assert_not(hg_refs, NULL);
6850 memset(hg_refs + hg_ref_cnt, 0, sizeof(hg_refs[0]) * 0x100);
6851 }
6852
6853 hg_refs[hg_ref_cnt] = strdup(name);
6854 my_assert_not(hg_refs[hg_ref_cnt], NULL);
6855 hg_ref_cnt++;
6856}
6857
6858// recursive register dep pass
6859// - track saved regs (part 2)
6860// - try to figure out arg-regs
6861// - calculate reg deps
6862static void gen_hdr_dep_pass(int i, int opcnt, unsigned char *cbits,
6863 struct func_prototype *fp, int regmask_save, int regmask_dst,
6864 int *regmask_dep, int *has_ret)
6865{
6866 struct func_proto_dep *dep;
6867 struct parsed_op *po;
6868 int from_caller = 0;
6869 int j, l;
6870 int reg;
6871 int ret;
6872
6873 for (; i < opcnt; i++)
6874 {
6875 if (cbits[i >> 3] & (1 << (i & 7)))
6876 return;
6877 cbits[i >> 3] |= (1 << (i & 7));
6878
6879 po = &ops[i];
6880
6881 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
6882 if (po->flags & OPF_RMD)
6883 continue;
6884
6885 if (po->btj != NULL) {
6886 // jumptable
6887 for (j = 0; j < po->btj->count; j++) {
6888 check_i(po, po->btj->d[j].bt_i);
6889 gen_hdr_dep_pass(po->btj->d[j].bt_i, opcnt, cbits, fp,
6890 regmask_save, regmask_dst, regmask_dep, has_ret);
6891 }
6892 return;
6893 }
6894
6895 check_i(po, po->bt_i);
6896 if (po->flags & OPF_CJMP) {
6897 gen_hdr_dep_pass(po->bt_i, opcnt, cbits, fp,
6898 regmask_save, regmask_dst, regmask_dep, has_ret);
6899 }
6900 else {
6901 i = po->bt_i - 1;
6902 }
6903 continue;
6904 }
6905
6906 if (po->flags & OPF_FARG)
6907 /* (just calculate register deps) */;
6908 else if (po->op == OP_PUSH && po->operand[0].type == OPT_REG)
6909 {
6910 reg = po->operand[0].reg;
6911 ferr_assert(po, reg >= 0);
6912
6913 if (po->flags & OPF_RSAVE) {
6914 regmask_save |= 1 << reg;
6915 continue;
6916 }
6917 if (po->flags & OPF_DONE)
6918 continue;
6919
6920 ret = scan_for_pop(i + 1, opcnt, i + opcnt * 2, reg, 0, 0);
6921 if (ret == 1) {
6922 regmask_save |= 1 << reg;
6923 po->flags |= OPF_RMD;
6924 scan_for_pop(i + 1, opcnt, i + opcnt * 3, reg, 0, OPF_RMD);
6925 continue;
6926 }
6927 }
6928 else if (po->flags & OPF_RMD)
6929 continue;
6930 else if (po->op == OP_CALL) {
6931 po->regmask_dst |= 1 << xAX;
6932
6933 dep = hg_fp_find_dep(fp, po->operand[0].name);
6934 if (dep != NULL)
6935 dep->regmask_live = regmask_save | regmask_dst;
6936 }
6937 else if (po->op == OP_RET) {
6938 if (po->operand_cnt > 0) {
6939 fp->is_stdcall = 1;
6940 if (fp->argc_stack >= 0
6941 && fp->argc_stack != po->operand[0].val / 4)
6942 ferr(po, "ret mismatch? (%d)\n", fp->argc_stack * 4);
6943 fp->argc_stack = po->operand[0].val / 4;
6944 }
6945 }
6946
6947 // if has_ret is 0, there is uninitialized eax path,
6948 // which means it's most likely void func
6949 if (*has_ret != 0 && (po->flags & OPF_TAIL)) {
6950 if (po->op == OP_CALL) {
6951 j = i;
6952 ret = 1;
6953 }
6954 else {
6955 struct parsed_opr opr = OPR_INIT(OPT_REG, OPLM_DWORD, xAX);
6956 j = -1;
6957 from_caller = 0;
6958 ret = resolve_origin(i, &opr, i + opcnt * 4, &j, &from_caller);
6959 }
6960
6961 if (ret != 1 && from_caller) {
6962 // unresolved eax - probably void func
6963 *has_ret = 0;
6964 }
6965 else {
6966 if (j >= 0 && ops[j].op == OP_CALL) {
6967 dep = hg_fp_find_dep(fp, ops[j].operand[0].name);
6968 if (dep != NULL)
6969 dep->ret_dep = 1;
6970 else
6971 *has_ret = 1;
6972 }
6973 else
6974 *has_ret = 1;
6975 }
6976 }
6977
6978 l = regmask_save | regmask_dst;
6979 if (g_bp_frame && !(po->flags & OPF_EBP_S))
6980 l |= 1 << xBP;
6981
6982 l = po->regmask_src & ~l;
6983#if 0
6984 if (l)
6985 fnote(po, "dep |= %04x, dst %04x, save %04x (f %x)\n",
6986 l, regmask_dst, regmask_save, po->flags);
6987#endif
6988 *regmask_dep |= l;
6989 regmask_dst |= po->regmask_dst;
6990
6991 if (po->flags & OPF_TAIL)
6992 return;
6993 }
6994}
6995
6996static void gen_hdr(const char *funcn, int opcnt)
6997{
6998 int save_arg_vars[MAX_ARG_GRP] = { 0, };
6999 unsigned char cbits[MAX_OPS / 8];
7000 const struct parsed_proto *pp_c;
7001 struct parsed_proto *pp;
7002 struct func_prototype *fp;
7003 struct parsed_op *po;
7004 int regmask_dummy = 0;
7005 int regmask_dep;
7006 int max_bp_offset = 0;
7007 int has_ret;
7008 int i, j, l;
7009 int ret;
7010
7011 pp_c = proto_parse(g_fhdr, funcn, 1);
7012 if (pp_c != NULL)
7013 // already in seed, will add to hg_fp later
7014 return;
7015
7016 fp = hg_fp_add(funcn);
7017
7018 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
7019 g_stack_frame_used = 0;
7020
7021 // pass1:
7022 // - resolve all branches
7023 // - parse calls with labels
7024 resolve_branches_parse_calls(opcnt);
7025
7026 // pass2:
7027 // - handle ebp/esp frame, remove ops related to it
7028 scan_prologue_epilogue(opcnt);
7029
7030 // pass3:
7031 // - remove dead labels
7032 // - collect calls
7033 for (i = 0; i < opcnt; i++)
7034 {
7035 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
7036 free(g_labels[i]);
7037 g_labels[i] = NULL;
7038 }
7039
7040 po = &ops[i];
7041 if (po->flags & (OPF_RMD|OPF_DONE))
7042 continue;
7043
7044 if (po->op == OP_CALL) {
7045 if (po->operand[0].type == OPT_LABEL)
7046 hg_fp_add_dep(fp, opr_name(po, 0));
7047 else if (po->pp != NULL)
7048 hg_fp_add_dep(fp, po->pp->name);
7049 }
7050 }
7051
7052 // pass4:
7053 // - remove dead labels
7054 // - handle push <const>/pop pairs
7055 for (i = 0; i < opcnt; i++)
7056 {
7057 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
7058 free(g_labels[i]);
7059 g_labels[i] = NULL;
7060 }
7061
7062 po = &ops[i];
7063 if (po->flags & (OPF_RMD|OPF_DONE))
7064 continue;
7065
7066 if (po->op == OP_PUSH && po->operand[0].type == OPT_CONST)
7067 scan_for_pop_const(i, opcnt, i + opcnt * 13);
7068 }
7069
7070 // pass5:
7071 // - process trivial calls
7072 for (i = 0; i < opcnt; i++)
7073 {
7074 po = &ops[i];
7075 if (po->flags & (OPF_RMD|OPF_DONE))
7076 continue;
7077
7078 if (po->op == OP_CALL)
7079 {
7080 pp = process_call_early(i, opcnt, &j);
7081 if (pp != NULL) {
7082 if (!(po->flags & OPF_ATAIL))
7083 // since we know the args, try to collect them
7084 if (collect_call_args_early(po, i, pp, &regmask_dummy) != 0)
7085 pp = NULL;
7086 }
7087
7088 if (pp != NULL) {
7089 if (j >= 0) {
7090 // commit esp adjust
7091 if (ops[j].op != OP_POP)
7092 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
7093 else {
7094 for (l = 0; l < pp->argc_stack; l++)
7095 ops[j + l].flags |= OPF_DONE | OPF_RMD | OPF_NOREGS;
7096 }
7097 }
7098
7099 po->flags |= OPF_DONE;
7100 }
7101 }
7102 }
7103
7104 // pass6:
7105 // - track saved regs (simple)
7106 // - process calls
7107 for (i = 0; i < opcnt; i++)
7108 {
7109 po = &ops[i];
7110 if (po->flags & (OPF_RMD|OPF_DONE))
7111 continue;
7112
7113 if (po->op == OP_PUSH && po->operand[0].type == OPT_REG
7114 && po->operand[0].reg != xCX)
7115 {
7116 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].reg, 0);
7117 if (ret == 1) {
7118 // regmask_save |= 1 << po->operand[0].reg; // do it later
7119 po->flags |= OPF_RSAVE | OPF_RMD | OPF_DONE;
7120 scan_for_pop_ret(i + 1, opcnt, po->operand[0].reg, OPF_RMD);
7121 }
7122 }
7123 else if (po->op == OP_CALL)
7124 {
7125 pp = process_call(i, opcnt);
7126
7127 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
7128 // since we know the args, collect them
7129 ret = collect_call_args(po, i, pp, &regmask_dummy, save_arg_vars,
7130 i + opcnt * 1);
7131 }
7132 }
7133 }
7134
7135 // pass7
7136 memset(cbits, 0, sizeof(cbits));
7137 regmask_dep = 0;
7138 has_ret = -1;
7139
7140 gen_hdr_dep_pass(0, opcnt, cbits, fp, 0, 0, &regmask_dep, &has_ret);
7141
7142 // find unreachable code - must be fixed in IDA
7143 for (i = 0; i < opcnt; i++)
7144 {
7145 if (cbits[i >> 3] & (1 << (i & 7)))
7146 continue;
7147
7148 if (g_labels[i] == NULL && i > 0 && ops[i - 1].op == OP_CALL
7149 && ops[i - 1].pp != NULL && ops[i - 1].pp->is_osinc)
7150 {
7151 // the compiler sometimes still generates code after
7152 // noreturn OS functions
7153 break;
7154 }
7155 if (ops[i].op != OP_NOP)
7156 ferr(&ops[i], "unreachable code\n");
7157 }
7158
7159 for (i = 0; i < g_eqcnt; i++) {
7160 if (g_eqs[i].offset > max_bp_offset && g_eqs[i].offset < 4*32)
7161 max_bp_offset = g_eqs[i].offset;
7162 }
7163
7164 if (fp->argc_stack < 0) {
7165 max_bp_offset = (max_bp_offset + 3) & ~3;
7166 fp->argc_stack = max_bp_offset / 4;
7167 if ((g_ida_func_attr & IDAFA_BP_FRAME) && fp->argc_stack > 0)
7168 fp->argc_stack--;
7169 }
7170
7171 fp->regmask_dep = regmask_dep & ~(1 << xSP);
7172 fp->has_ret = has_ret;
7173#if 0
7174 printf("// has_ret %d, regmask_dep %x\n",
7175 fp->has_ret, fp->regmask_dep);
7176 output_hdr_fp(stdout, fp, 1);
7177 if (IS(funcn, "sub_10007F72")) exit(1);
7178#endif
7179
7180 gen_x_cleanup(opcnt);
7181}
7182
7183static void hg_fp_resolve_deps(struct func_prototype *fp)
7184{
7185 struct func_prototype fp_s;
7186 int dep;
7187 int i;
7188
7189 // this thing is recursive, so mark first..
7190 fp->dep_resolved = 1;
7191
7192 for (i = 0; i < fp->dep_func_cnt; i++) {
7193 strcpy(fp_s.name, fp->dep_func[i].name);
7194 fp->dep_func[i].proto = bsearch(&fp_s, hg_fp, hg_fp_cnt,
7195 sizeof(hg_fp[0]), hg_fp_cmp_name);
7196 if (fp->dep_func[i].proto != NULL) {
7197 if (!fp->dep_func[i].proto->dep_resolved)
7198 hg_fp_resolve_deps(fp->dep_func[i].proto);
7199
7200 dep = ~fp->dep_func[i].regmask_live
7201 & fp->dep_func[i].proto->regmask_dep;
7202 fp->regmask_dep |= dep;
7203 // printf("dep %s %s |= %x\n", fp->name,
7204 // fp->dep_func[i].name, dep);
7205
7206 if (fp->has_ret == -1 && fp->dep_func[i].ret_dep)
7207 fp->has_ret = fp->dep_func[i].proto->has_ret;
7208 }
7209 }
7210}
7211
7212// make all thiscall/edx arg functions referenced from .data fastcall
7213static void do_func_refs_from_data(void)
7214{
7215 struct func_prototype *fp, fp_s;
7216 int i;
7217
7218 for (i = 0; i < hg_ref_cnt; i++) {
7219 strcpy(fp_s.name, hg_refs[i]);
7220 fp = bsearch(&fp_s, hg_fp, hg_fp_cnt,
7221 sizeof(hg_fp[0]), hg_fp_cmp_name);
7222 if (fp == NULL)
7223 continue;
7224
7225 if (fp->argc_stack != 0 && (fp->regmask_dep & (mxCX | mxDX)))
7226 fp->regmask_dep |= mxCX | mxDX;
7227 }
7228}
7229
7230static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
7231 int count)
7232{
7233 const struct parsed_proto *pp;
7234 char *p, namebuf[NAMELEN];
7235 const char *name;
7236 int regmask_dep;
7237 int argc_normal;
7238 int j, arg;
7239
7240 for (; count > 0; count--, fp++) {
7241 if (fp->has_ret == -1)
7242 fprintf(fout, "// ret unresolved\n");
7243#if 0
7244 fprintf(fout, "// dep:");
7245 for (j = 0; j < fp->dep_func_cnt; j++) {
7246 fprintf(fout, " %s/", fp->dep_func[j].name);
7247 if (fp->dep_func[j].proto != NULL)
7248 fprintf(fout, "%04x/%d", fp->dep_func[j].proto->regmask_dep,
7249 fp->dep_func[j].proto->has_ret);
7250 }
7251 fprintf(fout, "\n");
7252#endif
7253
7254 p = strchr(fp->name, '@');
7255 if (p != NULL) {
7256 memcpy(namebuf, fp->name, p - fp->name);
7257 namebuf[p - fp->name] = 0;
7258 name = namebuf;
7259 }
7260 else
7261 name = fp->name;
7262 if (name[0] == '_')
7263 name++;
7264
7265 pp = proto_parse(g_fhdr, name, 1);
7266 if (pp != NULL && pp->is_include)
7267 continue;
7268
7269 if (fp->pp != NULL) {
7270 // part of seed, output later
7271 continue;
7272 }
7273
7274 regmask_dep = fp->regmask_dep;
7275 argc_normal = fp->argc_stack;
7276
7277 fprintf(fout, "%-5s", fp->pp ? fp->pp->ret_type.name :
7278 (fp->has_ret ? "int" : "void"));
7279 if (regmask_dep && (fp->is_stdcall || fp->argc_stack > 0)
7280 && (regmask_dep & ~mxCX) == 0)
7281 {
7282 fprintf(fout, "/*__thiscall*/ ");
7283 argc_normal++;
7284 regmask_dep = 0;
7285 }
7286 else if (regmask_dep && (fp->is_stdcall || fp->argc_stack == 0)
7287 && (regmask_dep & ~(mxCX | mxDX)) == 0)
7288 {
7289 fprintf(fout, " __fastcall ");
7290 if (!(regmask_dep & (1 << xDX)) && fp->argc_stack == 0)
7291 argc_normal = 1;
7292 else
7293 argc_normal += 2;
7294 regmask_dep = 0;
7295 }
7296 else if (regmask_dep && !fp->is_stdcall) {
7297 fprintf(fout, "/*__usercall*/ ");
7298 }
7299 else if (regmask_dep) {
7300 fprintf(fout, "/*__userpurge*/ ");
7301 }
7302 else if (fp->is_stdcall)
7303 fprintf(fout, " __stdcall ");
7304 else
7305 fprintf(fout, " __cdecl ");
7306
7307 fprintf(fout, "%s(", name);
7308
7309 arg = 0;
7310 for (j = 0; j < xSP; j++) {
7311 if (regmask_dep & (1 << j)) {
7312 arg++;
7313 if (arg != 1)
7314 fprintf(fout, ", ");
7315 if (fp->pp != NULL)
7316 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
7317 else
7318 fprintf(fout, "int");
7319 fprintf(fout, " a%d/*<%s>*/", arg, regs_r32[j]);
7320 }
7321 }
7322
7323 for (j = 0; j < argc_normal; j++) {
7324 arg++;
7325 if (arg != 1)
7326 fprintf(fout, ", ");
7327 if (fp->pp != NULL) {
7328 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
7329 if (!fp->pp->arg[arg - 1].type.is_ptr)
7330 fprintf(fout, " ");
7331 }
7332 else
7333 fprintf(fout, "int ");
7334 fprintf(fout, "a%d", arg);
7335 }
7336
7337 fprintf(fout, ");\n");
7338 }
7339}
7340
7341static void output_hdr(FILE *fout)
7342{
7343 static const char *lmod_c_names[] = {
7344 [OPLM_UNSPEC] = "???",
7345 [OPLM_BYTE] = "uint8_t",
7346 [OPLM_WORD] = "uint16_t",
7347 [OPLM_DWORD] = "uint32_t",
7348 [OPLM_QWORD] = "uint64_t",
7349 };
7350 const struct scanned_var *var;
7351 struct func_prototype *fp;
7352 char line[256] = { 0, };
7353 char name[256];
7354 int i;
7355
7356 // add stuff from headers
7357 for (i = 0; i < pp_cache_size; i++) {
7358 if (pp_cache[i].is_cinc && !pp_cache[i].is_stdcall)
7359 snprintf(name, sizeof(name), "_%s", pp_cache[i].name);
7360 else
7361 snprintf(name, sizeof(name), "%s", pp_cache[i].name);
7362 fp = hg_fp_add(name);
7363 fp->pp = &pp_cache[i];
7364 fp->argc_stack = fp->pp->argc_stack;
7365 fp->is_stdcall = fp->pp->is_stdcall;
7366 fp->regmask_dep = get_pp_arg_regmask_src(fp->pp);
7367 fp->has_ret = !IS(fp->pp->ret_type.name, "void");
7368 }
7369
7370 // resolve deps
7371 qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_name);
7372 for (i = 0; i < hg_fp_cnt; i++)
7373 hg_fp_resolve_deps(&hg_fp[i]);
7374
7375 // adjust functions referenced from data segment
7376 do_func_refs_from_data();
7377
7378 // note: messes up .proto ptr, don't use
7379 //qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_id);
7380
7381 // output variables
7382 for (i = 0; i < hg_var_cnt; i++) {
7383 var = &hg_vars[i];
7384
7385 if (var->pp != NULL)
7386 // part of seed
7387 continue;
7388 else if (var->is_c_str)
7389 fprintf(fout, "extern %-8s %s[];", "char", var->name);
7390 else
7391 fprintf(fout, "extern %-8s %s;",
7392 lmod_c_names[var->lmod], var->name);
7393
7394 if (var->is_seeded)
7395 fprintf(fout, " // seeded");
7396 fprintf(fout, "\n");
7397 }
7398
7399 fprintf(fout, "\n");
7400
7401 // output function prototypes
7402 output_hdr_fp(fout, hg_fp, hg_fp_cnt);
7403
7404 // seed passthrough
7405 fprintf(fout, "\n// - seed -\n");
7406
7407 rewind(g_fhdr);
7408 while (fgets(line, sizeof(line), g_fhdr))
7409 fwrite(line, 1, strlen(line), fout);
7410}
7411
7412// '=' needs special treatment
7413// also ' quote
7414static char *next_word_s(char *w, size_t wsize, char *s)
7415{
7416 size_t i;
7417
7418 s = sskip(s);
7419
7420 i = 0;
7421 if (*s == '\'' && s[1] != '\r' && s[1] != '\n') {
7422 w[0] = s[0];
7423 for (i = 1; i < wsize - 1; i++) {
7424 if (s[i] == 0) {
7425 printf("warning: missing closing quote: \"%s\"\n", s);
7426 break;
7427 }
7428 if (s[i] == '\'')
7429 break;
7430 w[i] = s[i];
7431 }
7432 }
7433
7434 for (; i < wsize - 1; i++) {
7435 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
7436 break;
7437 w[i] = s[i];
7438 }
7439 w[i] = 0;
7440
7441 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
7442 printf("warning: '%s' truncated\n", w);
7443
7444 return s + i;
7445}
7446
7447static int cmpstringp(const void *p1, const void *p2)
7448{
7449 return strcmp(*(char * const *)p1, *(char * const *)p2);
7450}
7451
7452static int is_xref_needed(char *p, char **rlist, int rlist_len)
7453{
7454 char *p2;
7455
7456 p = sskip(p);
7457 if (strstr(p, "..."))
7458 // unable to determine, assume needed
7459 return 1;
7460
7461 if (*p == '.') // .text, .data, ...
7462 // ref from other data or non-function -> no
7463 return 0;
7464
7465 p2 = strpbrk(p, "+:\r\n\x18");
7466 if (p2 != NULL)
7467 *p2 = 0;
7468 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
7469 // referenced from removed code
7470 return 0;
7471
7472 return 1;
7473}
7474
7475static int ida_xrefs_show_need(FILE *fasm, char *p,
7476 char **rlist, int rlist_len)
7477{
7478 int found_need = 0;
7479 char line[256];
7480 long pos;
7481
7482 p = strrchr(p, ';');
7483 if (p != NULL && *p == ';' && IS_START(p + 2, "DATA XREF: ")) {
7484 p += 13;
7485 if (is_xref_needed(p, rlist, rlist_len))
7486 return 1;
7487 }
7488
7489 pos = ftell(fasm);
7490 while (1)
7491 {
7492 if (!my_fgets(line, sizeof(line), fasm))
7493 break;
7494 // non-first line is always indented
7495 if (!my_isblank(line[0]))
7496 break;
7497
7498 // should be no content, just comment
7499 p = sskip(line);
7500 if (*p != ';')
7501 break;
7502
7503 p = strrchr(p, ';');
7504 p += 2;
7505 // it's printed once, but no harm to check again
7506 if (IS_START(p, "DATA XREF: "))
7507 p += 11;
7508
7509 if (is_xref_needed(p, rlist, rlist_len)) {
7510 found_need = 1;
7511 break;
7512 }
7513 }
7514 fseek(fasm, pos, SEEK_SET);
7515 return found_need;
7516}
7517
7518static void scan_variables(FILE *fasm, char **rlist, int rlist_len)
7519{
7520 struct scanned_var *var;
7521 char line[256] = { 0, };
7522 char words[4][256];
7523 int no_identifier;
7524 char *p = NULL;
7525 int wordc;
7526 int l;
7527
7528 while (!feof(fasm))
7529 {
7530 // skip to next data section
7531 while (my_fgets(line, sizeof(line), fasm))
7532 {
7533 asmln++;
7534
7535 p = sskip(line);
7536 if (*p == 0 || *p == ';')
7537 continue;
7538
7539 p = sskip(next_word_s(words[0], sizeof(words[0]), p));
7540 if (*p == 0 || *p == ';')
7541 continue;
7542
7543 if (*p != 's' || !IS_START(p, "segment para public"))
7544 continue;
7545
7546 break;
7547 }
7548
7549 if (p == NULL || !IS_START(p, "segment para public"))
7550 break;
7551 p = sskip(p + 19);
7552
7553 if (!IS_START(p, "'DATA'"))
7554 continue;
7555
7556 // now process it
7557 while (my_fgets(line, sizeof(line), fasm))
7558 {
7559 asmln++;
7560
7561 p = line;
7562 no_identifier = my_isblank(*p);
7563
7564 p = sskip(p);
7565 if (*p == 0 || *p == ';')
7566 continue;
7567
7568 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
7569 words[wordc][0] = 0;
7570 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
7571 if (*p == 0 || *p == ';') {
7572 wordc++;
7573 break;
7574 }
7575 }
7576
7577 if (wordc == 2 && IS(words[1], "ends"))
7578 break;
7579 if (wordc < 2)
7580 continue;
7581
7582 if (no_identifier) {
7583 if (wordc >= 3 && IS(words[0], "dd") && IS(words[1], "offset"))
7584 hg_ref_add(words[2]);
7585 continue;
7586 }
7587
7588 if (IS_START(words[0], "__IMPORT_DESCRIPTOR_")) {
7589 // when this starts, we don't need anything from this section
7590 break;
7591 }
7592
7593 // check refs comment(s)
7594 if (!ida_xrefs_show_need(fasm, p, rlist, rlist_len))
7595 continue;
7596
7597 if ((hg_var_cnt & 0xff) == 0) {
7598 hg_vars = realloc(hg_vars, sizeof(hg_vars[0])
7599 * (hg_var_cnt + 0x100));
7600 my_assert_not(hg_vars, NULL);
7601 memset(hg_vars + hg_var_cnt, 0, sizeof(hg_vars[0]) * 0x100);
7602 }
7603
7604 var = &hg_vars[hg_var_cnt++];
7605 snprintf(var->name, sizeof(var->name), "%s", words[0]);
7606
7607 // maybe already in seed header?
7608 var->pp = proto_parse(g_fhdr, var->name, 1);
7609 if (var->pp != NULL) {
7610 if (var->pp->is_fptr) {
7611 var->lmod = OPLM_DWORD;
7612 //var->is_ptr = 1;
7613 }
7614 else if (var->pp->is_func)
7615 aerr("func?\n");
7616 else if (!guess_lmod_from_c_type(&var->lmod, &var->pp->type))
7617 aerr("unhandled C type '%s' for '%s'\n",
7618 var->pp->type.name, var->name);
7619
7620 var->is_seeded = 1;
7621 continue;
7622 }
7623
7624 if (IS(words[1], "dd")) {
7625 var->lmod = OPLM_DWORD;
7626 if (wordc >= 4 && IS(words[2], "offset"))
7627 hg_ref_add(words[3]);
7628 }
7629 else if (IS(words[1], "dw"))
7630 var->lmod = OPLM_WORD;
7631 else if (IS(words[1], "db")) {
7632 var->lmod = OPLM_BYTE;
7633 if (wordc >= 3 && (l = strlen(words[2])) > 4) {
7634 if (words[2][0] == '\'' && IS(words[2] + l - 2, ",0"))
7635 var->is_c_str = 1;
7636 }
7637 }
7638 else if (IS(words[1], "dq"))
7639 var->lmod = OPLM_QWORD;
7640 //else if (IS(words[1], "dt"))
7641 else
7642 aerr("type '%s' not known\n", words[1]);
7643 }
7644 }
7645
7646 rewind(fasm);
7647 asmln = 0;
7648}
7649
7650static void set_label(int i, const char *name)
7651{
7652 const char *p;
7653 int len;
7654
7655 len = strlen(name);
7656 p = strchr(name, ':');
7657 if (p != NULL)
7658 len = p - name;
7659
7660 if (g_labels[i] != NULL && !IS_START(g_labels[i], "algn_"))
7661 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
7662 g_labels[i] = realloc(g_labels[i], len + 1);
7663 my_assert_not(g_labels[i], NULL);
7664 memcpy(g_labels[i], name, len);
7665 g_labels[i][len] = 0;
7666}
7667
7668struct chunk_item {
7669 char *name;
7670 long fptr;
7671 int asmln;
7672};
7673
7674static struct chunk_item *func_chunks;
7675static int func_chunk_cnt;
7676static int func_chunk_alloc;
7677
7678static void add_func_chunk(FILE *fasm, const char *name, int line)
7679{
7680 if (func_chunk_cnt >= func_chunk_alloc) {
7681 func_chunk_alloc *= 2;
7682 func_chunks = realloc(func_chunks,
7683 func_chunk_alloc * sizeof(func_chunks[0]));
7684 my_assert_not(func_chunks, NULL);
7685 }
7686 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
7687 func_chunks[func_chunk_cnt].name = strdup(name);
7688 func_chunks[func_chunk_cnt].asmln = line;
7689 func_chunk_cnt++;
7690}
7691
7692static int cmp_chunks(const void *p1, const void *p2)
7693{
7694 const struct chunk_item *c1 = p1, *c2 = p2;
7695 return strcmp(c1->name, c2->name);
7696}
7697
7698static void scan_ahead(FILE *fasm)
7699{
7700 char words[2][256];
7701 char line[256];
7702 long oldpos;
7703 int oldasmln;
7704 int wordc;
7705 char *p;
7706 int i;
7707
7708 oldpos = ftell(fasm);
7709 oldasmln = asmln;
7710
7711 while (my_fgets(line, sizeof(line), fasm))
7712 {
7713 wordc = 0;
7714 asmln++;
7715
7716 p = sskip(line);
7717 if (*p == 0)
7718 continue;
7719
7720 if (*p == ';')
7721 {
7722 // get rid of random tabs
7723 for (i = 0; line[i] != 0; i++)
7724 if (line[i] == '\t')
7725 line[i] = ' ';
7726
7727 if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
7728 {
7729 p += 30;
7730 next_word(words[0], sizeof(words[0]), p);
7731 if (words[0][0] == 0)
7732 aerr("missing name for func chunk?\n");
7733
7734 add_func_chunk(fasm, words[0], asmln);
7735 }
7736 else if (IS_START(p, "; sctend"))
7737 break;
7738
7739 continue;
7740 } // *p == ';'
7741
7742 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
7743 words[wordc][0] = 0;
7744 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
7745 if (*p == 0 || *p == ';') {
7746 wordc++;
7747 break;
7748 }
7749 }
7750
7751 if (wordc == 2 && IS(words[1], "ends"))
7752 break;
7753 }
7754
7755 fseek(fasm, oldpos, SEEK_SET);
7756 asmln = oldasmln;
7757}
7758
7759int main(int argc, char *argv[])
7760{
7761 FILE *fout, *fasm, *frlist;
7762 struct parsed_data *pd = NULL;
7763 int pd_alloc = 0;
7764 char **rlist = NULL;
7765 int rlist_len = 0;
7766 int rlist_alloc = 0;
7767 int func_chunks_used = 0;
7768 int func_chunks_sorted = 0;
7769 int func_chunk_i = -1;
7770 long func_chunk_ret = 0;
7771 int func_chunk_ret_ln = 0;
7772 int scanned_ahead = 0;
7773 char line[256];
7774 char words[20][256];
7775 enum opr_lenmod lmod;
7776 char *sctproto = NULL;
7777 int in_func = 0;
7778 int pending_endp = 0;
7779 int skip_func = 0;
7780 int skip_warned = 0;
7781 int eq_alloc;
7782 int verbose = 0;
7783 int multi_seg = 0;
7784 int end = 0;
7785 int arg_out;
7786 int arg;
7787 int pi = 0;
7788 int i, j;
7789 int ret, len;
7790 char *p;
7791 int wordc;
7792
7793 for (arg = 1; arg < argc; arg++) {
7794 if (IS(argv[arg], "-v"))
7795 verbose = 1;
7796 else if (IS(argv[arg], "-rf"))
7797 g_allow_regfunc = 1;
7798 else if (IS(argv[arg], "-m"))
7799 multi_seg = 1;
7800 else if (IS(argv[arg], "-hdr"))
7801 g_header_mode = g_quiet_pp = g_allow_regfunc = 1;
7802 else
7803 break;
7804 }
7805
7806 if (argc < arg + 3) {
7807 printf("usage:\n%s [-v] [-rf] [-m] <.c> <.asm> <hdr.h> [rlist]*\n"
7808 "%s -hdr <out.h> <.asm> <seed.h> [rlist]*\n"
7809 "options:\n"
7810 " -hdr - header generation mode\n"
7811 " -rf - allow unannotated indirect calls\n"
7812 " -m - allow multiple .text sections\n"
7813 "[rlist] is a file with function names to skip,"
7814 " one per line\n",
7815 argv[0], argv[0]);
7816 return 1;
7817 }
7818
7819 arg_out = arg++;
7820
7821 asmfn = argv[arg++];
7822 fasm = fopen(asmfn, "r");
7823 my_assert_not(fasm, NULL);
7824
7825 hdrfn = argv[arg++];
7826 g_fhdr = fopen(hdrfn, "r");
7827 my_assert_not(g_fhdr, NULL);
7828
7829 rlist_alloc = 64;
7830 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
7831 my_assert_not(rlist, NULL);
7832 // needs special handling..
7833 rlist[rlist_len++] = "__alloca_probe";
7834
7835 func_chunk_alloc = 32;
7836 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
7837 my_assert_not(func_chunks, NULL);
7838
7839 memset(words, 0, sizeof(words));
7840
7841 for (; arg < argc; arg++) {
7842 frlist = fopen(argv[arg], "r");
7843 my_assert_not(frlist, NULL);
7844
7845 while (my_fgets(line, sizeof(line), frlist)) {
7846 p = sskip(line);
7847 if (*p == 0 || *p == ';')
7848 continue;
7849 if (*p == '#') {
7850 if (IS_START(p, "#if 0")
7851 || (g_allow_regfunc && IS_START(p, "#if NO_REGFUNC")))
7852 {
7853 skip_func = 1;
7854 }
7855 else if (IS_START(p, "#endif"))
7856 skip_func = 0;
7857 continue;
7858 }
7859 if (skip_func)
7860 continue;
7861
7862 p = next_word(words[0], sizeof(words[0]), p);
7863 if (words[0][0] == 0)
7864 continue;
7865
7866 if (rlist_len >= rlist_alloc) {
7867 rlist_alloc = rlist_alloc * 2 + 64;
7868 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
7869 my_assert_not(rlist, NULL);
7870 }
7871 rlist[rlist_len++] = strdup(words[0]);
7872 }
7873 skip_func = 0;
7874
7875 fclose(frlist);
7876 frlist = NULL;
7877 }
7878
7879 if (rlist_len > 0)
7880 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
7881
7882 fout = fopen(argv[arg_out], "w");
7883 my_assert_not(fout, NULL);
7884
7885 eq_alloc = 128;
7886 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
7887 my_assert_not(g_eqs, NULL);
7888
7889 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
7890 g_label_refs[i].i = -1;
7891 g_label_refs[i].next = NULL;
7892 }
7893
7894 if (g_header_mode)
7895 scan_variables(fasm, rlist, rlist_len);
7896
7897 while (my_fgets(line, sizeof(line), fasm))
7898 {
7899 wordc = 0;
7900 asmln++;
7901
7902 p = sskip(line);
7903 if (*p == 0)
7904 continue;
7905
7906 // get rid of random tabs
7907 for (i = 0; line[i] != 0; i++)
7908 if (line[i] == '\t')
7909 line[i] = ' ';
7910
7911 if (*p == ';')
7912 {
7913 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
7914 goto do_pending_endp; // eww..
7915
7916 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
7917 {
7918 static const char *attrs[] = {
7919 "bp-based frame",
7920 "library function",
7921 "static",
7922 "noreturn",
7923 "thunk",
7924 "fpd=",
7925 };
7926
7927 // parse IDA's attribute-list comment
7928 g_ida_func_attr = 0;
7929 p = sskip(p + 13);
7930
7931 for (; *p != 0; p = sskip(p)) {
7932 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
7933 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
7934 g_ida_func_attr |= 1 << i;
7935 p += strlen(attrs[i]);
7936 break;
7937 }
7938 }
7939 if (i == ARRAY_SIZE(attrs)) {
7940 anote("unparsed IDA attr: %s\n", p);
7941 break;
7942 }
7943 if (IS(attrs[i], "fpd=")) {
7944 p = next_word(words[0], sizeof(words[0]), p);
7945 // ignore for now..
7946 }
7947 }
7948 }
7949 else if (p[2] == 's' && IS_START(p, "; sctattr:"))
7950 {
7951 static const char *attrs[] = {
7952 "clear_sf",
7953 "clear_regmask",
7954 };
7955
7956 // parse manual attribute-list comment
7957 g_sct_func_attr = 0;
7958 p = sskip(p + 10);
7959
7960 for (; *p != 0; p = sskip(p)) {
7961 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
7962 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
7963 g_sct_func_attr |= 1 << i;
7964 p += strlen(attrs[i]);
7965 break;
7966 }
7967 }
7968 if (*p == '=') {
7969 j = ret = 0;
7970 if (i == 0)
7971 // clear_sf=start,len (in dwords)
7972 ret = sscanf(p, "=%d,%d%n", &g_stack_clear_start,
7973 &g_stack_clear_len, &j);
7974 else if (i == 1)
7975 // clear_regmask=<mask>
7976 ret = sscanf(p, "=%x%n", &g_regmask_init, &j) + 1;
7977 if (ret < 2) {
7978 anote("unparsed attr value: %s\n", p);
7979 break;
7980 }
7981 p += j;
7982 }
7983 else if (i == ARRAY_SIZE(attrs)) {
7984 anote("unparsed sct attr: %s\n", p);
7985 break;
7986 }
7987 }
7988 }
7989 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
7990 {
7991 p += 30;
7992 next_word(words[0], sizeof(words[0]), p);
7993 if (words[0][0] == 0)
7994 aerr("missing name for func chunk?\n");
7995
7996 if (!scanned_ahead) {
7997 add_func_chunk(fasm, words[0], asmln);
7998 func_chunks_sorted = 0;
7999 }
8000 }
8001 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
8002 {
8003 if (func_chunk_i >= 0) {
8004 if (func_chunk_i < func_chunk_cnt
8005 && IS(func_chunks[func_chunk_i].name, g_func))
8006 {
8007 // move on to next chunk
8008 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
8009 if (ret)
8010 aerr("seek failed for '%s' chunk #%d\n",
8011 g_func, func_chunk_i);
8012 asmln = func_chunks[func_chunk_i].asmln;
8013 func_chunk_i++;
8014 }
8015 else {
8016 if (func_chunk_ret == 0)
8017 aerr("no return from chunk?\n");
8018 fseek(fasm, func_chunk_ret, SEEK_SET);
8019 asmln = func_chunk_ret_ln;
8020 func_chunk_ret = 0;
8021 pending_endp = 1;
8022 }
8023 }
8024 }
8025 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
8026 func_chunks_used = 1;
8027 p += 20;
8028 if (IS_START(g_func, "sub_")) {
8029 unsigned long addr = strtoul(p, NULL, 16);
8030 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
8031 if (addr > f_addr && !scanned_ahead) {
8032 //anote("scan_ahead caused by '%s', addr %lx\n",
8033 // g_func, addr);
8034 scan_ahead(fasm);
8035 scanned_ahead = 1;
8036 func_chunks_sorted = 0;
8037 }
8038 }
8039 }
8040 continue;
8041 } // *p == ';'
8042
8043parse_words:
8044 for (i = wordc; i < ARRAY_SIZE(words); i++)
8045 words[i][0] = 0;
8046 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
8047 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
8048 if (*p == 0 || *p == ';') {
8049 wordc++;
8050 break;
8051 }
8052 }
8053 if (*p != 0 && *p != ';')
8054 aerr("too many words\n");
8055
8056 // alow asm patches in comments
8057 if (*p == ';') {
8058 if (IS_START(p, "; sctpatch:")) {
8059 p = sskip(p + 11);
8060 if (*p == 0 || *p == ';')
8061 continue;
8062 goto parse_words; // lame
8063 }
8064 if (IS_START(p, "; sctproto:")) {
8065 sctproto = strdup(p + 11);
8066 }
8067 else if (IS_START(p, "; sctend")) {
8068 end = 1;
8069 if (!pending_endp)
8070 break;
8071 }
8072 }
8073
8074 if (wordc == 0) {
8075 // shouldn't happen
8076 awarn("wordc == 0?\n");
8077 continue;
8078 }
8079
8080 // don't care about this:
8081 if (words[0][0] == '.'
8082 || IS(words[0], "include")
8083 || IS(words[0], "assume") || IS(words[1], "segment")
8084 || IS(words[0], "align"))
8085 {
8086 continue;
8087 }
8088
8089do_pending_endp:
8090 // do delayed endp processing to collect switch jumptables
8091 if (pending_endp) {
8092 if (in_func && !g_skip_func && !end && wordc >= 2
8093 && ((words[0][0] == 'd' && words[0][2] == 0)
8094 || (words[1][0] == 'd' && words[1][2] == 0)))
8095 {
8096 i = 1;
8097 if (words[1][0] == 'd' && words[1][2] == 0) {
8098 // label
8099 if (g_func_pd_cnt >= pd_alloc) {
8100 pd_alloc = pd_alloc * 2 + 16;
8101 g_func_pd = realloc(g_func_pd,
8102 sizeof(g_func_pd[0]) * pd_alloc);
8103 my_assert_not(g_func_pd, NULL);
8104 }
8105 pd = &g_func_pd[g_func_pd_cnt];
8106 g_func_pd_cnt++;
8107 memset(pd, 0, sizeof(*pd));
8108 strcpy(pd->label, words[0]);
8109 pd->type = OPT_CONST;
8110 pd->lmod = lmod_from_directive(words[1]);
8111 i = 2;
8112 }
8113 else {
8114 if (pd == NULL) {
8115 if (verbose)
8116 anote("skipping alignment byte?\n");
8117 continue;
8118 }
8119 lmod = lmod_from_directive(words[0]);
8120 if (lmod != pd->lmod)
8121 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
8122 }
8123
8124 if (pd->count_alloc < pd->count + wordc) {
8125 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
8126 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
8127 my_assert_not(pd->d, NULL);
8128 }
8129 for (; i < wordc; i++) {
8130 if (IS(words[i], "offset")) {
8131 pd->type = OPT_OFFSET;
8132 i++;
8133 }
8134 p = strchr(words[i], ',');
8135 if (p != NULL)
8136 *p = 0;
8137 if (pd->type == OPT_OFFSET)
8138 pd->d[pd->count].u.label = strdup(words[i]);
8139 else
8140 pd->d[pd->count].u.val = parse_number(words[i]);
8141 pd->d[pd->count].bt_i = -1;
8142 pd->count++;
8143 }
8144 continue;
8145 }
8146
8147 if (in_func && !g_skip_func) {
8148 if (g_header_mode)
8149 gen_hdr(g_func, pi);
8150 else
8151 gen_func(fout, g_fhdr, g_func, pi);
8152 }
8153
8154 pending_endp = 0;
8155 in_func = 0;
8156 g_ida_func_attr = 0;
8157 g_sct_func_attr = 0;
8158 g_stack_clear_start = 0;
8159 g_stack_clear_len = 0;
8160 g_regmask_init = 0;
8161 skip_warned = 0;
8162 g_skip_func = 0;
8163 g_func[0] = 0;
8164 func_chunks_used = 0;
8165 func_chunk_i = -1;
8166 if (pi != 0) {
8167 memset(&ops, 0, pi * sizeof(ops[0]));
8168 clear_labels(pi);
8169 pi = 0;
8170 }
8171 g_eqcnt = 0;
8172 for (i = 0; i < g_func_pd_cnt; i++) {
8173 pd = &g_func_pd[i];
8174 if (pd->type == OPT_OFFSET) {
8175 for (j = 0; j < pd->count; j++)
8176 free(pd->d[j].u.label);
8177 }
8178 free(pd->d);
8179 pd->d = NULL;
8180 }
8181 g_func_pd_cnt = 0;
8182 g_func_lmods = 0;
8183 pd = NULL;
8184
8185 if (end)
8186 break;
8187 if (wordc == 0)
8188 continue;
8189 }
8190
8191 if (IS(words[1], "proc")) {
8192 if (in_func)
8193 aerr("proc '%s' while in_func '%s'?\n",
8194 words[0], g_func);
8195 p = words[0];
8196 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
8197 g_skip_func = 1;
8198 strcpy(g_func, words[0]);
8199 set_label(0, words[0]);
8200 in_func = 1;
8201 continue;
8202 }
8203
8204 if (IS(words[1], "endp"))
8205 {
8206 if (!in_func)
8207 aerr("endp '%s' while not in_func?\n", words[0]);
8208 if (!IS(g_func, words[0]))
8209 aerr("endp '%s' while in_func '%s'?\n",
8210 words[0], g_func);
8211
8212 if ((g_ida_func_attr & IDAFA_THUNK) && pi == 1
8213 && ops[0].op == OP_JMP && ops[0].operand[0].had_ds)
8214 {
8215 // import jump
8216 g_skip_func = 1;
8217 }
8218
8219 if (!g_skip_func && func_chunks_used) {
8220 // start processing chunks
8221 struct chunk_item *ci, key = { g_func, 0 };
8222
8223 func_chunk_ret = ftell(fasm);
8224 func_chunk_ret_ln = asmln;
8225 if (!func_chunks_sorted) {
8226 qsort(func_chunks, func_chunk_cnt,
8227 sizeof(func_chunks[0]), cmp_chunks);
8228 func_chunks_sorted = 1;
8229 }
8230 ci = bsearch(&key, func_chunks, func_chunk_cnt,
8231 sizeof(func_chunks[0]), cmp_chunks);
8232 if (ci == NULL)
8233 aerr("'%s' needs chunks, but none found\n", g_func);
8234 func_chunk_i = ci - func_chunks;
8235 for (; func_chunk_i > 0; func_chunk_i--)
8236 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
8237 break;
8238
8239 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
8240 if (ret)
8241 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
8242 asmln = func_chunks[func_chunk_i].asmln;
8243 func_chunk_i++;
8244 continue;
8245 }
8246 pending_endp = 1;
8247 continue;
8248 }
8249
8250 if (wordc == 2 && IS(words[1], "ends")) {
8251 if (!multi_seg) {
8252 end = 1;
8253 if (pending_endp)
8254 goto do_pending_endp;
8255 break;
8256 }
8257
8258 // scan for next text segment
8259 while (my_fgets(line, sizeof(line), fasm)) {
8260 asmln++;
8261 p = sskip(line);
8262 if (*p == 0 || *p == ';')
8263 continue;
8264
8265 if (strstr(p, "segment para public 'CODE' use32"))
8266 break;
8267 }
8268
8269 continue;
8270 }
8271
8272 p = strchr(words[0], ':');
8273 if (p != NULL) {
8274 set_label(pi, words[0]);
8275 continue;
8276 }
8277
8278 if (!in_func || g_skip_func) {
8279 if (!skip_warned && !g_skip_func && g_labels[pi] != NULL) {
8280 if (verbose)
8281 anote("skipping from '%s'\n", g_labels[pi]);
8282 skip_warned = 1;
8283 }
8284 free(g_labels[pi]);
8285 g_labels[pi] = NULL;
8286 continue;
8287 }
8288
8289 if (wordc > 1 && IS(words[1], "="))
8290 {
8291 if (wordc != 5)
8292 aerr("unhandled equ, wc=%d\n", wordc);
8293 if (g_eqcnt >= eq_alloc) {
8294 eq_alloc *= 2;
8295 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
8296 my_assert_not(g_eqs, NULL);
8297 }
8298
8299 len = strlen(words[0]);
8300 if (len > sizeof(g_eqs[0].name) - 1)
8301 aerr("equ name too long: %d\n", len);
8302 strcpy(g_eqs[g_eqcnt].name, words[0]);
8303
8304 if (!IS(words[3], "ptr"))
8305 aerr("unhandled equ\n");
8306 if (IS(words[2], "dword"))
8307 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
8308 else if (IS(words[2], "word"))
8309 g_eqs[g_eqcnt].lmod = OPLM_WORD;
8310 else if (IS(words[2], "byte"))
8311 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
8312 else if (IS(words[2], "qword"))
8313 g_eqs[g_eqcnt].lmod = OPLM_QWORD;
8314 else
8315 aerr("bad lmod: '%s'\n", words[2]);
8316
8317 g_eqs[g_eqcnt].offset = parse_number(words[4]);
8318 g_eqcnt++;
8319 continue;
8320 }
8321
8322 if (pi >= ARRAY_SIZE(ops))
8323 aerr("too many ops\n");
8324
8325 parse_op(&ops[pi], words, wordc);
8326
8327 ops[pi].datap = sctproto;
8328 sctproto = NULL;
8329 pi++;
8330 }
8331
8332 if (g_header_mode)
8333 output_hdr(fout);
8334
8335 fclose(fout);
8336 fclose(fasm);
8337 fclose(g_fhdr);
8338
8339 return 0;
8340}
8341
8342// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.