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