notaz.gp2x.de / ia32rtools.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
translate: add custom stack-clear attribute
[ia32rtools.git] / tools / translate.c
... / ...
CommitLineData
1/*
2 * ia32rtools
3 * (C) notaz, 2013-2015
4 *
5 * This work is licensed under the terms of 3-clause BSD license.
6 * See COPYING file in the top-level directory.
7 */
8
9#define _GNU_SOURCE
10#include <stdio.h>
11#include <stdlib.h>
12#include <string.h>
13
14#include "my_assert.h"
15#include "my_str.h"
16#include "common.h"
17
18#define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
19#define IS(w, y) !strcmp(w, y)
20#define IS_START(w, y) !strncmp(w, y, strlen(y))
21
22#include "protoparse.h"
23
24static const char *asmfn;
25static int asmln;
26static FILE *g_fhdr;
27
28#define anote(fmt, ...) \
29 printf("%s:%d: note: " fmt, asmfn, asmln, ##__VA_ARGS__)
30#define awarn(fmt, ...) \
31 printf("%s:%d: warning: " fmt, asmfn, asmln, ##__VA_ARGS__)
32#define aerr(fmt, ...) do { \
33 printf("%s:%d: error: " fmt, asmfn, asmln, ##__VA_ARGS__); \
34 fcloseall(); \
35 exit(1); \
36} while (0)
37
38#include "masm_tools.h"
39
40enum op_flags {
41 OPF_RMD = (1 << 0), /* removed from code generation */
42 OPF_DATA = (1 << 1), /* data processing - writes to dst opr */
43 OPF_FLAGS = (1 << 2), /* sets flags */
44 OPF_JMP = (1 << 3), /* branch, call */
45 OPF_CJMP = (1 << 4), /* cond. branch (cc or jecxz/loop) */
46 OPF_CC = (1 << 5), /* uses flags */
47 OPF_TAIL = (1 << 6), /* ret or tail call */
48 OPF_RSAVE = (1 << 7), /* push/pop is local reg save/load */
49 OPF_REP = (1 << 8), /* prefixed by rep */
50 OPF_REPZ = (1 << 9), /* rep is repe/repz */
51 OPF_REPNZ = (1 << 10), /* rep is repne/repnz */
52 OPF_FARG = (1 << 11), /* push collected as func arg */
53 OPF_FARGNR = (1 << 12), /* push collected as func arg (no reuse) */
54 OPF_EBP_S = (1 << 13), /* ebp used as scratch here, not BP */
55 OPF_DF = (1 << 14), /* DF flag set */
56 OPF_ATAIL = (1 << 15), /* tail call with reused arg frame */
57 OPF_32BIT = (1 << 16), /* 32bit division */
58 OPF_LOCK = (1 << 17), /* op has lock prefix */
59 OPF_VAPUSH = (1 << 18), /* vararg ptr push (as call arg) */
60 OPF_DONE = (1 << 19), /* already fully handled by analysis */
61 OPF_PPUSH = (1 << 20), /* part of complex push-pop graph */
62 OPF_NOREGS = (1 << 21), /* don't track regs of this op */
63 OPF_FPUSH = (1 << 22), /* pushes x87 stack */
64 OPF_FPOP = (1 << 23), /* pops x87 stack */
65 OPF_FSHIFT = (1 << 24), /* x87 stack shift is actually needed */
66};
67
68enum op_op {
69 OP_INVAL,
70 OP_NOP,
71 OP_PUSH,
72 OP_POP,
73 OP_LEAVE,
74 OP_MOV,
75 OP_LEA,
76 OP_MOVZX,
77 OP_MOVSX,
78 OP_XCHG,
79 OP_NOT,
80 OP_XLAT,
81 OP_CDQ,
82 OP_LODS,
83 OP_STOS,
84 OP_MOVS,
85 OP_CMPS,
86 OP_SCAS,
87 OP_STD,
88 OP_CLD,
89 OP_RET,
90 OP_ADD,
91 OP_SUB,
92 OP_AND,
93 OP_OR,
94 OP_XOR,
95 OP_SHL,
96 OP_SHR,
97 OP_SAR,
98 OP_SHLD,
99 OP_SHRD,
100 OP_ROL,
101 OP_ROR,
102 OP_RCL,
103 OP_RCR,
104 OP_ADC,
105 OP_SBB,
106 OP_BSF,
107 OP_INC,
108 OP_DEC,
109 OP_NEG,
110 OP_MUL,
111 OP_IMUL,
112 OP_DIV,
113 OP_IDIV,
114 OP_TEST,
115 OP_CMP,
116 OP_CALL,
117 OP_JMP,
118 OP_JECXZ,
119 OP_LOOP,
120 OP_JCC,
121 OP_SCC,
122 // x87
123 OP_FLD,
124 OP_FILD,
125 OP_FLDc,
126 OP_FST,
127 OP_FADD,
128 OP_FDIV,
129 OP_FMUL,
130 OP_FSUB,
131 OP_FDIVR,
132 OP_FSUBR,
133 OP_FIADD,
134 OP_FIDIV,
135 OP_FIMUL,
136 OP_FISUB,
137 OP_FIDIVR,
138 OP_FISUBR,
139 // mmx
140 OP_EMMS,
141 // pseudo-ops for lib calls
142 OPP_FTOL,
143 // undefined
144 OP_UD2,
145};
146
147enum opr_type {
148 OPT_UNSPEC,
149 OPT_REG,
150 OPT_REGMEM,
151 OPT_LABEL,
152 OPT_OFFSET,
153 OPT_CONST,
154};
155
156// must be sorted (larger len must be further in enum)
157enum opr_lenmod {
158 OPLM_UNSPEC,
159 OPLM_BYTE,
160 OPLM_WORD,
161 OPLM_DWORD,
162 OPLM_QWORD,
163};
164
165#define MAX_EXITS 128
166
167#define MAX_OPERANDS 3
168#define NAMELEN 112
169
170#define OPR_INIT(type_, lmod_, reg_) \
171 { type_, lmod_, reg_, }
172
173struct parsed_opr {
174 enum opr_type type;
175 enum opr_lenmod lmod;
176 int reg;
177 unsigned int is_ptr:1; // pointer in C
178 unsigned int is_array:1; // array in C
179 unsigned int type_from_var:1; // .. in header, sometimes wrong
180 unsigned int size_mismatch:1; // type override differs from C
181 unsigned int size_lt:1; // type override is larger than C
182 unsigned int had_ds:1; // had ds: prefix
183 const struct parsed_proto *pp; // for OPT_LABEL
184 unsigned int val;
185 char name[NAMELEN];
186};
187
188struct parsed_op {
189 enum op_op op;
190 struct parsed_opr operand[MAX_OPERANDS];
191 unsigned int flags;
192 unsigned char pfo;
193 unsigned char pfo_inv;
194 unsigned char operand_cnt;
195 unsigned char p_argnum; // arg push: altered before call arg #
196 unsigned char p_arggrp; // arg push: arg group # for above
197 unsigned char p_argpass;// arg push: arg of host func
198 short p_argnext;// arg push: same arg pushed elsewhere or -1
199 int regmask_src; // all referensed regs
200 int regmask_dst;
201 int pfomask; // flagop: parsed_flag_op that can't be delayed
202 int cc_scratch; // scratch storage during analysis
203 int bt_i; // branch target for branches
204 struct parsed_data *btj;// branch targets for jumptables
205 struct parsed_proto *pp;// parsed_proto for OP_CALL
206 void *datap;
207 int asmln;
208};
209
210// datap:
211// on start: function/data type hint (sctproto)
212// after analysis:
213// (OPF_CC) - points to one of (OPF_FLAGS) that affects cc op
214// OP_PUSH - points to OP_POP in complex push/pop graph
215// OP_POP - points to OP_PUSH in simple push/pop pair
216
217struct parsed_equ {
218 char name[64];
219 enum opr_lenmod lmod;
220 int offset;
221};
222
223struct parsed_data {
224 char label[256];
225 enum opr_type type;
226 enum opr_lenmod lmod;
227 int count;
228 int count_alloc;
229 struct {
230 union {
231 char *label;
232 unsigned int val;
233 } u;
234 int bt_i;
235 } *d;
236};
237
238struct label_ref {
239 int i;
240 struct label_ref *next;
241};
242
243enum ida_func_attr {
244 IDAFA_BP_FRAME = (1 << 0),
245 IDAFA_LIB_FUNC = (1 << 1),
246 IDAFA_STATIC = (1 << 2),
247 IDAFA_NORETURN = (1 << 3),
248 IDAFA_THUNK = (1 << 4),
249 IDAFA_FPD = (1 << 5),
250};
251
252enum sct_func_attr {
253 SCTFA_CLEAR_SF = (1 << 0), // clear stack frame
254};
255
256enum x87_const {
257 X87_CONST_1 = 1,
258 X87_CONST_2T,
259 X87_CONST_2E,
260 X87_CONST_PI,
261 X87_CONST_LG2,
262 X87_CONST_LN2,
263 X87_CONST_Z,
264};
265
266// note: limited to 32k due to p_argnext
267#define MAX_OPS 4096
268#define MAX_ARG_GRP 2
269
270static struct parsed_op ops[MAX_OPS];
271static struct parsed_equ *g_eqs;
272static int g_eqcnt;
273static char *g_labels[MAX_OPS];
274static struct label_ref g_label_refs[MAX_OPS];
275static const struct parsed_proto *g_func_pp;
276static struct parsed_data *g_func_pd;
277static int g_func_pd_cnt;
278static int g_func_lmods;
279static char g_func[256];
280static char g_comment[256];
281static int g_bp_frame;
282static int g_sp_frame;
283static int g_stack_frame_used;
284static int g_stack_fsz;
285static int g_ida_func_attr;
286static int g_sct_func_attr;
287static int g_stack_clear_start; // in dwords
288static int g_stack_clear_len;
289static int g_skip_func;
290static int g_allow_regfunc;
291static int g_quiet_pp;
292static int g_header_mode;
293
294#define ferr(op_, fmt, ...) do { \
295 printf("%s:%d: error %u: [%s] '%s': " fmt, asmfn, (op_)->asmln, \
296 __LINE__, g_func, dump_op(op_), ##__VA_ARGS__); \
297 fcloseall(); \
298 exit(1); \
299} while (0)
300#define fnote(op_, fmt, ...) \
301 printf("%s:%d: note: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
302 dump_op(op_), ##__VA_ARGS__)
303
304#define ferr_assert(op_, cond) do { \
305 if (!(cond)) ferr(op_, "assertion '%s' failed\n", #cond); \
306} while (0)
307
308const char *regs_r32[] = {
309 "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp",
310 // not r32, but list here for easy parsing and printing
311 "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7",
312 "st", "st(1)", "st(2)", "st(3)", "st(4)", "st(5)", "st(6)", "st(7)"
313};
314const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" };
315const char *regs_r8l[] = { "al", "bl", "cl", "dl" };
316const char *regs_r8h[] = { "ah", "bh", "ch", "dh" };
317
318enum x86_regs {
319 xUNSPEC = -1,
320 xAX, xBX, xCX, xDX,
321 xSI, xDI, xBP, xSP,
322 xMM0, xMM1, xMM2, xMM3, // mmx
323 xMM4, xMM5, xMM6, xMM7,
324 xST0, xST1, xST2, xST3, // x87
325 xST4, xST5, xST6, xST7,
326};
327
328#define mxAX (1 << xAX)
329#define mxDX (1 << xDX)
330#define mxST0 (1 << xST0)
331#define mxST1 (1 << xST1)
332
333// possible basic comparison types (without inversion)
334enum parsed_flag_op {
335 PFO_O, // 0 OF=1
336 PFO_C, // 2 CF=1
337 PFO_Z, // 4 ZF=1
338 PFO_BE, // 6 CF=1||ZF=1
339 PFO_S, // 8 SF=1
340 PFO_P, // a PF=1
341 PFO_L, // c SF!=OF
342 PFO_LE, // e ZF=1||SF!=OF
343};
344
345#define PFOB_O (1 << PFO_O)
346#define PFOB_C (1 << PFO_C)
347#define PFOB_Z (1 << PFO_Z)
348#define PFOB_S (1 << PFO_S)
349
350static const char *parsed_flag_op_names[] = {
351 "o", "c", "z", "be", "s", "p", "l", "le"
352};
353
354static int char_array_i(const char *array[], size_t len, const char *s)
355{
356 int i;
357
358 for (i = 0; i < len; i++)
359 if (IS(s, array[i]))
360 return i;
361
362 return -1;
363}
364
365static void printf_number(char *buf, size_t buf_size,
366 unsigned long number)
367{
368 // output in C-friendly form
369 snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
370}
371
372static int check_segment_prefix(const char *s)
373{
374 if (s[0] == 0 || s[1] != 's' || s[2] != ':')
375 return 0;
376
377 switch (s[0]) {
378 case 'c': return 1;
379 case 'd': return 2;
380 case 's': return 3;
381 case 'e': return 4;
382 case 'f': return 5;
383 case 'g': return 6;
384 default: return 0;
385 }
386}
387
388static int parse_reg(enum opr_lenmod *reg_lmod, const char *s)
389{
390 int reg;
391
392 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s);
393 if (reg >= 8) {
394 *reg_lmod = OPLM_QWORD;
395 return reg;
396 }
397 if (reg >= 0) {
398 *reg_lmod = OPLM_DWORD;
399 return reg;
400 }
401 reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s);
402 if (reg >= 0) {
403 *reg_lmod = OPLM_WORD;
404 return reg;
405 }
406 reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s);
407 if (reg >= 0) {
408 *reg_lmod = OPLM_BYTE;
409 return reg;
410 }
411 reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s);
412 if (reg >= 0) {
413 *reg_lmod = OPLM_BYTE;
414 return reg;
415 }
416
417 return -1;
418}
419
420static int parse_indmode(char *name, int *regmask, int need_c_cvt)
421{
422 enum opr_lenmod lmod;
423 char cvtbuf[256];
424 char *d = cvtbuf;
425 char *s = name;
426 char w[64];
427 long number;
428 int reg;
429 int c = 0;
430
431 *d = 0;
432
433 while (*s != 0) {
434 d += strlen(d);
435 while (my_isblank(*s))
436 s++;
437 for (; my_issep(*s); d++, s++)
438 *d = *s;
439 while (my_isblank(*s))
440 s++;
441 *d = 0;
442
443 // skip '?s:' prefixes
444 if (check_segment_prefix(s))
445 s += 3;
446
447 s = next_idt(w, sizeof(w), s);
448 if (w[0] == 0)
449 break;
450 c++;
451
452 reg = parse_reg(&lmod, w);
453 if (reg >= 0) {
454 *regmask |= 1 << reg;
455 goto pass;
456 }
457
458 if ('0' <= w[0] && w[0] <= '9') {
459 number = parse_number(w);
460 printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number);
461 continue;
462 }
463
464 // probably some label/identifier - pass
465
466pass:
467 snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w);
468 }
469
470 if (need_c_cvt)
471 strcpy(name, cvtbuf);
472
473 return c;
474}
475
476static int is_reg_in_str(const char *s)
477{
478 int i;
479
480 if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
481 return 0;
482
483 for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
484 if (!strncmp(s, regs_r32[i], 3))
485 return 1;
486
487 return 0;
488}
489
490static const char *parse_stack_el(const char *name, char *extra_reg,
491 int early_try)
492{
493 const char *p, *p2, *s;
494 char *endp = NULL;
495 char buf[32];
496 long val;
497 int len;
498
499 if (g_bp_frame || early_try)
500 {
501 p = name;
502 if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
503 p += 4;
504 if (extra_reg != NULL) {
505 strncpy(extra_reg, name, 3);
506 extra_reg[4] = 0;
507 }
508 }
509
510 if (IS_START(p, "ebp+")) {
511 p += 4;
512
513 p2 = strchr(p, '+');
514 if (p2 != NULL && is_reg_in_str(p)) {
515 if (extra_reg != NULL) {
516 strncpy(extra_reg, p, p2 - p);
517 extra_reg[p2 - p] = 0;
518 }
519 p = p2 + 1;
520 }
521
522 if (!('0' <= *p && *p <= '9'))
523 return p;
524
525 return NULL;
526 }
527 }
528
529 if (!IS_START(name, "esp+"))
530 return NULL;
531
532 s = name + 4;
533 p = strchr(s, '+');
534 if (p) {
535 if (is_reg_in_str(s)) {
536 if (extra_reg != NULL) {
537 strncpy(extra_reg, s, p - s);
538 extra_reg[p - s] = 0;
539 }
540 s = p + 1;
541 p = strchr(s, '+');
542 if (p == NULL)
543 aerr("%s IDA stackvar not set?\n", __func__);
544 }
545 if (!('0' <= *s && *s <= '9')) {
546 aerr("%s IDA stackvar offset not set?\n", __func__);
547 return NULL;
548 }
549 if (s[0] == '0' && s[1] == 'x')
550 s += 2;
551 len = p - s;
552 if (len < sizeof(buf) - 1) {
553 strncpy(buf, s, len);
554 buf[len] = 0;
555 val = strtol(buf, &endp, 16);
556 if (val == 0 || *endp != 0) {
557 aerr("%s num parse fail for '%s'\n", __func__, buf);
558 return NULL;
559 }
560 }
561 p++;
562 }
563 else
564 p = name + 4;
565
566 if ('0' <= *p && *p <= '9')
567 return NULL;
568
569 return p;
570}
571
572static int guess_lmod_from_name(struct parsed_opr *opr)
573{
574 if (IS_START(opr->name, "dword_") || IS_START(opr->name, "off_")) {
575 opr->lmod = OPLM_DWORD;
576 return 1;
577 }
578 if (IS_START(opr->name, "word_")) {
579 opr->lmod = OPLM_WORD;
580 return 1;
581 }
582 if (IS_START(opr->name, "byte_")) {
583 opr->lmod = OPLM_BYTE;
584 return 1;
585 }
586 if (IS_START(opr->name, "qword_")) {
587 opr->lmod = OPLM_QWORD;
588 return 1;
589 }
590 return 0;
591}
592
593static int guess_lmod_from_c_type(enum opr_lenmod *lmod,
594 const struct parsed_type *c_type)
595{
596 static const char *dword_types[] = {
597 "uint32_t", "int", "_DWORD", "UINT_PTR", "DWORD",
598 "WPARAM", "LPARAM", "UINT", "__int32",
599 "LONG", "HIMC", "BOOL", "size_t",
600 "float",
601 };
602 static const char *word_types[] = {
603 "uint16_t", "int16_t", "_WORD", "WORD",
604 "unsigned __int16", "__int16",
605 };
606 static const char *byte_types[] = {
607 "uint8_t", "int8_t", "char",
608 "unsigned __int8", "__int8", "BYTE", "_BYTE",
609 "CHAR", "_UNKNOWN",
610 // structures.. deal the same as with _UNKNOWN for now
611 "CRITICAL_SECTION",
612 };
613 const char *n;
614 int i;
615
616 if (c_type->is_ptr) {
617 *lmod = OPLM_DWORD;
618 return 1;
619 }
620
621 n = skip_type_mod(c_type->name);
622
623 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
624 if (IS(n, dword_types[i])) {
625 *lmod = OPLM_DWORD;
626 return 1;
627 }
628 }
629
630 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
631 if (IS(n, word_types[i])) {
632 *lmod = OPLM_WORD;
633 return 1;
634 }
635 }
636
637 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
638 if (IS(n, byte_types[i])) {
639 *lmod = OPLM_BYTE;
640 return 1;
641 }
642 }
643
644 return 0;
645}
646
647static char *default_cast_to(char *buf, size_t buf_size,
648 struct parsed_opr *opr)
649{
650 buf[0] = 0;
651
652 if (!opr->is_ptr || strchr(opr->name, '['))
653 return buf;
654 if (opr->pp == NULL || opr->pp->type.name == NULL
655 || opr->pp->is_fptr)
656 {
657 snprintf(buf, buf_size, "%s", "(void *)");
658 return buf;
659 }
660
661 snprintf(buf, buf_size, "(%s)", opr->pp->type.name);
662 return buf;
663}
664
665static enum opr_type lmod_from_directive(const char *d)
666{
667 if (IS(d, "dd"))
668 return OPLM_DWORD;
669 else if (IS(d, "dw"))
670 return OPLM_WORD;
671 else if (IS(d, "db"))
672 return OPLM_BYTE;
673
674 aerr("unhandled directive: '%s'\n", d);
675 return OPLM_UNSPEC;
676}
677
678static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
679 int *regmask)
680{
681 opr->type = OPT_REG;
682 opr->reg = reg;
683 opr->lmod = lmod;
684 *regmask |= 1 << reg;
685}
686
687static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
688 int *extra_offs);
689
690static int parse_operand(struct parsed_opr *opr,
691 int *regmask, int *regmask_indirect,
692 char words[16][256], int wordc, int w, unsigned int op_flags)
693{
694 const struct parsed_proto *pp = NULL;
695 enum opr_lenmod tmplmod;
696 unsigned long number;
697 char buf[256];
698 int ret, len;
699 int wordc_in;
700 char *p;
701 int i;
702
703 if (w >= wordc)
704 aerr("parse_operand w %d, wordc %d\n", w, wordc);
705
706 opr->reg = xUNSPEC;
707
708 for (i = w; i < wordc; i++) {
709 len = strlen(words[i]);
710 if (words[i][len - 1] == ',') {
711 words[i][len - 1] = 0;
712 wordc = i + 1;
713 break;
714 }
715 }
716
717 wordc_in = wordc - w;
718
719 if ((op_flags & OPF_JMP) && wordc_in > 0
720 && !('0' <= words[w][0] && words[w][0] <= '9'))
721 {
722 const char *label = NULL;
723
724 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
725 && IS(words[w + 1], "ptr"))
726 label = words[w + 2];
727 else if (wordc_in == 2 && IS(words[w], "short"))
728 label = words[w + 1];
729 else if (wordc_in == 1
730 && strchr(words[w], '[') == NULL
731 && parse_reg(&tmplmod, words[w]) < 0)
732 label = words[w];
733
734 if (label != NULL) {
735 opr->type = OPT_LABEL;
736 ret = check_segment_prefix(label);
737 if (ret != 0) {
738 if (ret >= 5)
739 aerr("fs/gs used\n");
740 opr->had_ds = 1;
741 label += 3;
742 }
743 strcpy(opr->name, label);
744 return wordc;
745 }
746 }
747
748 if (wordc_in >= 3) {
749 if (IS(words[w + 1], "ptr")) {
750 if (IS(words[w], "dword"))
751 opr->lmod = OPLM_DWORD;
752 else if (IS(words[w], "word"))
753 opr->lmod = OPLM_WORD;
754 else if (IS(words[w], "byte"))
755 opr->lmod = OPLM_BYTE;
756 else if (IS(words[w], "qword"))
757 opr->lmod = OPLM_QWORD;
758 else
759 aerr("type parsing failed\n");
760 w += 2;
761 wordc_in = wordc - w;
762 }
763 }
764
765 if (wordc_in == 2) {
766 if (IS(words[w], "offset")) {
767 opr->type = OPT_OFFSET;
768 opr->lmod = OPLM_DWORD;
769 strcpy(opr->name, words[w + 1]);
770 pp = proto_parse(g_fhdr, opr->name, 1);
771 goto do_label;
772 }
773 if (IS(words[w], "(offset")) {
774 p = strchr(words[w + 1], ')');
775 if (p == NULL)
776 aerr("parse of bracketed offset failed\n");
777 *p = 0;
778 opr->type = OPT_OFFSET;
779 strcpy(opr->name, words[w + 1]);
780 return wordc;
781 }
782 }
783
784 if (wordc_in != 1)
785 aerr("parse_operand 1 word expected\n");
786
787 ret = check_segment_prefix(words[w]);
788 if (ret != 0) {
789 if (ret >= 5)
790 aerr("fs/gs used\n");
791 opr->had_ds = 1;
792 memmove(words[w], words[w] + 3, strlen(words[w]) - 2);
793 }
794 strcpy(opr->name, words[w]);
795
796 if (words[w][0] == '[') {
797 opr->type = OPT_REGMEM;
798 ret = sscanf(words[w], "[%[^]]]", opr->name);
799 if (ret != 1)
800 aerr("[] parse failure\n");
801
802 parse_indmode(opr->name, regmask_indirect, 1);
803 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL, 1))
804 {
805 // might be an equ
806 struct parsed_equ *eq =
807 equ_find(NULL, parse_stack_el(opr->name, NULL, 1), &i);
808 if (eq)
809 opr->lmod = eq->lmod;
810
811 // might be unaligned access
812 g_func_lmods |= 1 << OPLM_BYTE;
813 }
814 return wordc;
815 }
816 else if (strchr(words[w], '[')) {
817 // label[reg] form
818 p = strchr(words[w], '[');
819 opr->type = OPT_REGMEM;
820 parse_indmode(p, regmask_indirect, 0);
821 strncpy(buf, words[w], p - words[w]);
822 buf[p - words[w]] = 0;
823 pp = proto_parse(g_fhdr, buf, 1);
824 goto do_label;
825 }
826 else if (('0' <= words[w][0] && words[w][0] <= '9')
827 || words[w][0] == '-')
828 {
829 number = parse_number(words[w]);
830 opr->type = OPT_CONST;
831 opr->val = number;
832 printf_number(opr->name, sizeof(opr->name), number);
833 return wordc;
834 }
835
836 ret = parse_reg(&tmplmod, opr->name);
837 if (ret >= 0) {
838 setup_reg_opr(opr, ret, tmplmod, regmask);
839 return wordc;
840 }
841
842 // most likely var in data segment
843 opr->type = OPT_LABEL;
844 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
845
846do_label:
847 if (pp != NULL) {
848 if (pp->is_fptr || pp->is_func) {
849 opr->lmod = OPLM_DWORD;
850 opr->is_ptr = 1;
851 }
852 else {
853 tmplmod = OPLM_UNSPEC;
854 if (!guess_lmod_from_c_type(&tmplmod, &pp->type))
855 anote("unhandled C type '%s' for '%s'\n",
856 pp->type.name, opr->name);
857
858 if (opr->lmod == OPLM_UNSPEC) {
859 opr->lmod = tmplmod;
860 opr->type_from_var = 1;
861 }
862 else if (opr->lmod != tmplmod) {
863 opr->size_mismatch = 1;
864 if (tmplmod < opr->lmod)
865 opr->size_lt = 1;
866 }
867 opr->is_ptr = pp->type.is_ptr;
868 }
869 opr->is_array = pp->type.is_array;
870 }
871 opr->pp = pp;
872
873 if (opr->lmod == OPLM_UNSPEC)
874 guess_lmod_from_name(opr);
875 return wordc;
876}
877
878static const struct {
879 const char *name;
880 unsigned int flags;
881} pref_table[] = {
882 { "rep", OPF_REP },
883 { "repe", OPF_REP|OPF_REPZ },
884 { "repz", OPF_REP|OPF_REPZ },
885 { "repne", OPF_REP|OPF_REPNZ },
886 { "repnz", OPF_REP|OPF_REPNZ },
887 { "lock", OPF_LOCK }, // ignored for now..
888};
889
890#define OPF_CJMP_CC (OPF_JMP|OPF_CJMP|OPF_CC)
891
892static const struct {
893 const char *name;
894 enum op_op op;
895 unsigned short minopr;
896 unsigned short maxopr;
897 unsigned int flags;
898 unsigned char pfo;
899 unsigned char pfo_inv;
900} op_table[] = {
901 { "nop", OP_NOP, 0, 0, 0 },
902 { "push", OP_PUSH, 1, 1, 0 },
903 { "pop", OP_POP, 1, 1, OPF_DATA },
904 { "leave",OP_LEAVE, 0, 0, OPF_DATA },
905 { "mov" , OP_MOV, 2, 2, OPF_DATA },
906 { "lea", OP_LEA, 2, 2, OPF_DATA },
907 { "movzx",OP_MOVZX, 2, 2, OPF_DATA },
908 { "movsx",OP_MOVSX, 2, 2, OPF_DATA },
909 { "xchg", OP_XCHG, 2, 2, OPF_DATA },
910 { "not", OP_NOT, 1, 1, OPF_DATA },
911 { "xlat", OP_XLAT, 0, 0, OPF_DATA },
912 { "cdq", OP_CDQ, 0, 0, OPF_DATA },
913 { "lodsb",OP_LODS, 0, 0, OPF_DATA },
914 { "lodsw",OP_LODS, 0, 0, OPF_DATA },
915 { "lodsd",OP_LODS, 0, 0, OPF_DATA },
916 { "stosb",OP_STOS, 0, 0, OPF_DATA },
917 { "stosw",OP_STOS, 0, 0, OPF_DATA },
918 { "stosd",OP_STOS, 0, 0, OPF_DATA },
919 { "movsb",OP_MOVS, 0, 0, OPF_DATA },
920 { "movsw",OP_MOVS, 0, 0, OPF_DATA },
921 { "movsd",OP_MOVS, 0, 0, OPF_DATA },
922 { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
923 { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
924 { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
925 { "scasb",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
926 { "scasw",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
927 { "scasd",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
928 { "std", OP_STD, 0, 0, OPF_DATA }, // special flag
929 { "cld", OP_CLD, 0, 0, OPF_DATA },
930 { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
931 { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
932 { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
933 { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS },
934 { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS },
935 { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
936 { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
937 { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
938 { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
939 { "shld", OP_SHLD, 3, 3, OPF_DATA|OPF_FLAGS },
940 { "shrd", OP_SHRD, 3, 3, OPF_DATA|OPF_FLAGS },
941 { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
942 { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
943 { "rcl", OP_RCL, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
944 { "rcr", OP_RCR, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
945 { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
946 { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
947 { "bsf", OP_BSF, 2, 2, OPF_DATA|OPF_FLAGS },
948 { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
949 { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS },
950 { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS },
951 { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS },
952 { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS },
953 { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS },
954 { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS },
955 { "test", OP_TEST, 2, 2, OPF_FLAGS },
956 { "cmp", OP_CMP, 2, 2, OPF_FLAGS },
957 { "retn", OP_RET, 0, 1, OPF_TAIL },
958 { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS },
959 { "jmp", OP_JMP, 1, 1, OPF_JMP },
960 { "jecxz",OP_JECXZ, 1, 1, OPF_JMP|OPF_CJMP },
961 { "loop", OP_LOOP, 1, 1, OPF_JMP|OPF_CJMP|OPF_DATA },
962 { "jo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 0 }, // 70 OF=1
963 { "jno", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 1 }, // 71 OF=0
964 { "jc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72 CF=1
965 { "jb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72
966 { "jnc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73 CF=0
967 { "jnb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
968 { "jae", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
969 { "jz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74 ZF=1
970 { "je", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74
971 { "jnz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75 ZF=0
972 { "jne", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75
973 { "jbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76 CF=1||ZF=1
974 { "jna", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76
975 { "ja", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77 CF=0&&ZF=0
976 { "jnbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77
977 { "js", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 0 }, // 78 SF=1
978 { "jns", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 1 }, // 79 SF=0
979 { "jp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a PF=1
980 { "jpe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a
981 { "jnp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b PF=0
982 { "jpo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b
983 { "jl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c SF!=OF
984 { "jnge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c
985 { "jge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d SF=OF
986 { "jnl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d
987 { "jle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e ZF=1||SF!=OF
988 { "jng", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e
989 { "jg", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f ZF=0&&SF=OF
990 { "jnle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f
991 { "seto", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 0 },
992 { "setno", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 1 },
993 { "setc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
994 { "setb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
995 { "setnc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
996 { "setae", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
997 { "setnb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
998 { "setz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
999 { "sete", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
1000 { "setnz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
1001 { "setne", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
1002 { "setbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
1003 { "setna", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
1004 { "seta", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
1005 { "setnbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
1006 { "sets", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 0 },
1007 { "setns", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 1 },
1008 { "setp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
1009 { "setpe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
1010 { "setnp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
1011 { "setpo", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
1012 { "setl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
1013 { "setnge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
1014 { "setge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
1015 { "setnl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
1016 { "setle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
1017 { "setng", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
1018 { "setg", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
1019 { "setnle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
1020 // x87
1021 { "fld", OP_FLD, 1, 1, OPF_FPUSH },
1022 { "fild", OP_FILD, 1, 1, OPF_FPUSH },
1023 { "fld1", OP_FLDc, 0, 0, OPF_FPUSH },
1024 { "fldz", OP_FLDc, 0, 0, OPF_FPUSH },
1025 { "fstp", OP_FST, 1, 1, OPF_FPOP },
1026 { "fst", OP_FST, 1, 1, 0 },
1027 { "fadd", OP_FADD, 0, 2, 0 },
1028 { "faddp", OP_FADD, 0, 2, OPF_FPOP },
1029 { "fdiv", OP_FDIV, 0, 2, 0 },
1030 { "fdivp", OP_FDIV, 0, 2, OPF_FPOP },
1031 { "fmul", OP_FMUL, 0, 2, 0 },
1032 { "fmulp", OP_FMUL, 0, 2, OPF_FPOP },
1033 { "fsub", OP_FSUB, 0, 2, 0 },
1034 { "fsubp", OP_FSUB, 0, 2, OPF_FPOP },
1035 { "fdivr", OP_FDIVR, 0, 2, 0 },
1036 { "fdivrp", OP_FDIVR, 0, 2, OPF_FPOP },
1037 { "fsubr", OP_FSUBR, 0, 2, 0 },
1038 { "fsubrp", OP_FSUBR, 0, 2, OPF_FPOP },
1039 { "fiadd", OP_FIADD, 1, 1, 0 },
1040 { "fidiv", OP_FIDIV, 1, 1, 0 },
1041 { "fimul", OP_FIMUL, 1, 1, 0 },
1042 { "fisub", OP_FISUB, 1, 1, 0 },
1043 { "fidivr", OP_FIDIVR, 1, 1, 0 },
1044 { "fisubr", OP_FISUBR, 1, 1, 0 },
1045 // mmx
1046 { "emms", OP_EMMS, 0, 0, OPF_DATA },
1047 { "movq", OP_MOV, 2, 2, OPF_DATA },
1048 // pseudo-ops for lib calls
1049 { "_ftol", OPP_FTOL },
1050 // must be last
1051 { "ud2", OP_UD2 },
1052};
1053
1054static void parse_op(struct parsed_op *op, char words[16][256], int wordc)
1055{
1056 enum opr_lenmod lmod = OPLM_UNSPEC;
1057 int prefix_flags = 0;
1058 int regmask_ind;
1059 int regmask;
1060 int op_w = 0;
1061 int opr = 0;
1062 int w = 0;
1063 int i, j;
1064
1065 for (i = 0; i < ARRAY_SIZE(pref_table); i++) {
1066 if (IS(words[w], pref_table[i].name)) {
1067 prefix_flags = pref_table[i].flags;
1068 break;
1069 }
1070 }
1071
1072 if (prefix_flags) {
1073 if (wordc <= 1)
1074 aerr("lone prefix: '%s'\n", words[0]);
1075 w++;
1076 }
1077
1078 op_w = w;
1079 for (i = 0; i < ARRAY_SIZE(op_table); i++) {
1080 if (IS(words[w], op_table[i].name))
1081 break;
1082 }
1083
1084 if (i == ARRAY_SIZE(op_table)) {
1085 if (!g_skip_func)
1086 aerr("unhandled op: '%s'\n", words[0]);
1087 i--; // OP_UD2
1088 }
1089 w++;
1090
1091 op->op = op_table[i].op;
1092 op->flags = op_table[i].flags | prefix_flags;
1093 op->pfo = op_table[i].pfo;
1094 op->pfo_inv = op_table[i].pfo_inv;
1095 op->regmask_src = op->regmask_dst = 0;
1096 op->asmln = asmln;
1097
1098 if (op->op == OP_UD2)
1099 return;
1100
1101 for (opr = 0; opr < op_table[i].maxopr; opr++) {
1102 if (opr >= op_table[i].minopr && w >= wordc)
1103 break;
1104
1105 regmask = regmask_ind = 0;
1106 w = parse_operand(&op->operand[opr], &regmask, &regmask_ind,
1107 words, wordc, w, op->flags);
1108
1109 if (opr == 0 && (op->flags & OPF_DATA))
1110 op->regmask_dst = regmask;
1111 else
1112 op->regmask_src |= regmask;
1113 op->regmask_src |= regmask_ind;
1114
1115 if (op->operand[opr].lmod != OPLM_UNSPEC)
1116 g_func_lmods |= 1 << op->operand[opr].lmod;
1117 }
1118
1119 if (w < wordc)
1120 aerr("parse_op %s incomplete: %d/%d\n",
1121 words[0], w, wordc);
1122
1123 // special cases
1124 op->operand_cnt = opr;
1125 if (!strncmp(op_table[i].name, "set", 3))
1126 op->operand[0].lmod = OPLM_BYTE;
1127
1128 switch (op->op) {
1129 // first operand is not dst
1130 case OP_CMP:
1131 case OP_TEST:
1132 op->regmask_src |= op->regmask_dst;
1133 op->regmask_dst = 0;
1134 break;
1135
1136 // first operand is src too
1137 case OP_NOT:
1138 case OP_ADD:
1139 case OP_AND:
1140 case OP_OR:
1141 case OP_RCL:
1142 case OP_RCR:
1143 case OP_ADC:
1144 case OP_INC:
1145 case OP_DEC:
1146 case OP_NEG:
1147 // more below..
1148 op->regmask_src |= op->regmask_dst;
1149 break;
1150
1151 // special
1152 case OP_XCHG:
1153 op->regmask_src |= op->regmask_dst;
1154 op->regmask_dst |= op->regmask_src;
1155 goto check_align;
1156
1157 case OP_SUB:
1158 case OP_SBB:
1159 case OP_XOR:
1160 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1161 && op->operand[0].lmod == op->operand[1].lmod
1162 && op->operand[0].reg == op->operand[1].reg
1163 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
1164 {
1165 op->regmask_src = 0;
1166 }
1167 else
1168 op->regmask_src |= op->regmask_dst;
1169 break;
1170
1171 // ops with implicit argumets
1172 case OP_XLAT:
1173 op->operand_cnt = 2;
1174 setup_reg_opr(&op->operand[0], xAX, OPLM_BYTE, &op->regmask_src);
1175 op->regmask_dst = op->regmask_src;
1176 setup_reg_opr(&op->operand[1], xBX, OPLM_DWORD, &op->regmask_src);
1177 break;
1178
1179 case OP_CDQ:
1180 op->operand_cnt = 2;
1181 setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst);
1182 setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src);
1183 break;
1184
1185 case OP_LODS:
1186 case OP_STOS:
1187 case OP_SCAS:
1188 if (words[op_w][4] == 'b')
1189 lmod = OPLM_BYTE;
1190 else if (words[op_w][4] == 'w')
1191 lmod = OPLM_WORD;
1192 else if (words[op_w][4] == 'd')
1193 lmod = OPLM_DWORD;
1194 j = 0;
1195 op->regmask_src = 0;
1196 setup_reg_opr(&op->operand[j++], op->op == OP_LODS ? xSI : xDI,
1197 OPLM_DWORD, &op->regmask_src);
1198 op->regmask_dst = op->regmask_src;
1199 setup_reg_opr(&op->operand[j++], xAX, lmod,
1200 op->op == OP_LODS ? &op->regmask_dst : &op->regmask_src);
1201 if (op->flags & OPF_REP) {
1202 setup_reg_opr(&op->operand[j++], xCX, OPLM_DWORD, &op->regmask_src);
1203 op->regmask_dst |= 1 << xCX;
1204 }
1205 op->operand_cnt = j;
1206 break;
1207
1208 case OP_MOVS:
1209 case OP_CMPS:
1210 if (words[op_w][4] == 'b')
1211 lmod = OPLM_BYTE;
1212 else if (words[op_w][4] == 'w')
1213 lmod = OPLM_WORD;
1214 else if (words[op_w][4] == 'd')
1215 lmod = OPLM_DWORD;
1216 j = 0;
1217 op->regmask_src = 0;
1218 // note: lmod is not correct, don't have where to place it
1219 setup_reg_opr(&op->operand[j++], xDI, lmod, &op->regmask_src);
1220 setup_reg_opr(&op->operand[j++], xSI, OPLM_DWORD, &op->regmask_src);
1221 if (op->flags & OPF_REP)
1222 setup_reg_opr(&op->operand[j++], xCX, OPLM_DWORD, &op->regmask_src);
1223 op->operand_cnt = j;
1224 op->regmask_dst = op->regmask_src;
1225 break;
1226
1227 case OP_LOOP:
1228 op->regmask_dst = 1 << xCX;
1229 // fallthrough
1230 case OP_JECXZ:
1231 op->operand_cnt = 2;
1232 op->regmask_src = 1 << xCX;
1233 op->operand[1].type = OPT_REG;
1234 op->operand[1].reg = xCX;
1235 op->operand[1].lmod = OPLM_DWORD;
1236 break;
1237
1238 case OP_IMUL:
1239 if (op->operand_cnt == 2) {
1240 if (op->operand[0].type != OPT_REG)
1241 aerr("reg expected\n");
1242 op->regmask_src |= 1 << op->operand[0].reg;
1243 }
1244 if (op->operand_cnt != 1)
1245 break;
1246 // fallthrough
1247 case OP_MUL:
1248 // singleop mul
1249 op->regmask_src |= op->regmask_dst;
1250 op->regmask_dst = (1 << xDX) | (1 << xAX);
1251 if (op->operand[0].lmod == OPLM_UNSPEC)
1252 op->operand[0].lmod = OPLM_DWORD;
1253 break;
1254
1255 case OP_DIV:
1256 case OP_IDIV:
1257 // we could set up operands for edx:eax, but there is no real need to
1258 // (see is_opr_modified())
1259 op->regmask_src |= op->regmask_dst;
1260 op->regmask_dst = (1 << xDX) | (1 << xAX);
1261 if (op->operand[0].lmod == OPLM_UNSPEC)
1262 op->operand[0].lmod = OPLM_DWORD;
1263 break;
1264
1265 case OP_SHL:
1266 case OP_SHR:
1267 case OP_SAR:
1268 case OP_ROL:
1269 case OP_ROR:
1270 op->regmask_src |= op->regmask_dst;
1271 if (op->operand[1].lmod == OPLM_UNSPEC)
1272 op->operand[1].lmod = OPLM_BYTE;
1273 break;
1274
1275 case OP_SHLD:
1276 case OP_SHRD:
1277 op->regmask_src |= op->regmask_dst;
1278 if (op->operand[2].lmod == OPLM_UNSPEC)
1279 op->operand[2].lmod = OPLM_BYTE;
1280 break;
1281
1282 case OP_PUSH:
1283 op->regmask_src |= op->regmask_dst;
1284 op->regmask_dst = 0;
1285 if (op->operand[0].lmod == OPLM_UNSPEC
1286 && (op->operand[0].type == OPT_CONST
1287 || op->operand[0].type == OPT_OFFSET
1288 || op->operand[0].type == OPT_LABEL))
1289 op->operand[0].lmod = OPLM_DWORD;
1290 break;
1291
1292 // alignment
1293 case OP_MOV:
1294 check_align:
1295 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1296 && op->operand[0].lmod == op->operand[1].lmod
1297 && op->operand[0].reg == op->operand[1].reg
1298 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
1299 {
1300 op->flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
1301 op->regmask_src = op->regmask_dst = 0;
1302 }
1303 break;
1304
1305 case OP_LEA:
1306 if (op->operand[0].type == OPT_REG
1307 && op->operand[1].type == OPT_REGMEM)
1308 {
1309 char buf[16];
1310 snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name);
1311 if (IS(buf, op->operand[1].name))
1312 op->flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
1313 }
1314 break;
1315
1316 case OP_CALL:
1317 // trashed regs must be explicitly detected later
1318 op->regmask_dst = 0;
1319 break;
1320
1321 case OP_LEAVE:
1322 op->regmask_dst = (1 << xBP) | (1 << xSP);
1323 op->regmask_src = 1 << xBP;
1324 break;
1325
1326 case OP_FLD:
1327 case OP_FILD:
1328 op->regmask_dst |= mxST0;
1329 break;
1330
1331 case OP_FLDc:
1332 op->regmask_dst |= mxST0;
1333 if (IS(words[op_w] + 3, "1"))
1334 op->operand[0].val = X87_CONST_1;
1335 else if (IS(words[op_w] + 3, "z"))
1336 op->operand[0].val = X87_CONST_Z;
1337 else
1338 aerr("TODO\n");
1339 break;
1340
1341 case OP_FST:
1342 op->regmask_src |= mxST0;
1343 break;
1344
1345 case OP_FADD:
1346 case OP_FDIV:
1347 case OP_FMUL:
1348 case OP_FSUB:
1349 case OP_FDIVR:
1350 case OP_FSUBR:
1351 op->regmask_src |= mxST0;
1352 if (op->operand_cnt == 2)
1353 op->regmask_src |= op->regmask_dst;
1354 else if (op->operand_cnt == 1) {
1355 memcpy(&op->operand[1], &op->operand[0], sizeof(op->operand[1]));
1356 op->operand[0].type = OPT_REG;
1357 op->operand[0].lmod = OPLM_QWORD;
1358 op->operand[0].reg = xST0;
1359 op->regmask_dst |= mxST0;
1360 }
1361 else
1362 // IDA doesn't use this
1363 aerr("no operands?\n");
1364 break;
1365
1366 case OP_FIADD:
1367 case OP_FIDIV:
1368 case OP_FIMUL:
1369 case OP_FISUB:
1370 case OP_FIDIVR:
1371 case OP_FISUBR:
1372 op->regmask_src |= mxST0;
1373 op->regmask_dst |= mxST0;
1374 break;
1375
1376 default:
1377 break;
1378 }
1379
1380 if (op->operand[0].type == OPT_REG
1381 && op->operand[1].type == OPT_CONST)
1382 {
1383 struct parsed_opr *op1 = &op->operand[1];
1384 if ((op->op == OP_AND && op1->val == 0)
1385 || (op->op == OP_OR
1386 && (op1->val == ~0
1387 || (op->operand[0].lmod == OPLM_WORD && op1->val == 0xffff)
1388 || (op->operand[0].lmod == OPLM_BYTE && op1->val == 0xff))))
1389 {
1390 op->regmask_src = 0;
1391 }
1392 }
1393}
1394
1395static const char *op_name(struct parsed_op *po)
1396{
1397 static char buf[16];
1398 char *p;
1399 int i;
1400
1401 if (po->op == OP_JCC || po->op == OP_SCC) {
1402 p = buf;
1403 *p++ = (po->op == OP_JCC) ? 'j' : 's';
1404 if (po->pfo_inv)
1405 *p++ = 'n';
1406 strcpy(p, parsed_flag_op_names[po->pfo]);
1407 return buf;
1408 }
1409
1410 for (i = 0; i < ARRAY_SIZE(op_table); i++)
1411 if (op_table[i].op == po->op)
1412 return op_table[i].name;
1413
1414 return "???";
1415}
1416
1417// debug
1418static const char *dump_op(struct parsed_op *po)
1419{
1420 static char out[128];
1421 char *p = out;
1422 int i;
1423
1424 if (po == NULL)
1425 return "???";
1426
1427 snprintf(out, sizeof(out), "%s", op_name(po));
1428 for (i = 0; i < po->operand_cnt; i++) {
1429 p += strlen(p);
1430 if (i > 0)
1431 *p++ = ',';
1432 snprintf(p, sizeof(out) - (p - out),
1433 po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s",
1434 po->operand[i].name);
1435 }
1436
1437 return out;
1438}
1439
1440static const char *lmod_type_u(struct parsed_op *po,
1441 enum opr_lenmod lmod)
1442{
1443 switch (lmod) {
1444 case OPLM_QWORD:
1445 return "u64";
1446 case OPLM_DWORD:
1447 return "u32";
1448 case OPLM_WORD:
1449 return "u16";
1450 case OPLM_BYTE:
1451 return "u8";
1452 default:
1453 ferr(po, "invalid lmod: %d\n", lmod);
1454 return "(_invalid_)";
1455 }
1456}
1457
1458static const char *lmod_cast_u(struct parsed_op *po,
1459 enum opr_lenmod lmod)
1460{
1461 switch (lmod) {
1462 case OPLM_QWORD:
1463 return "";
1464 case OPLM_DWORD:
1465 return "";
1466 case OPLM_WORD:
1467 return "(u16)";
1468 case OPLM_BYTE:
1469 return "(u8)";
1470 default:
1471 ferr(po, "invalid lmod: %d\n", lmod);
1472 return "(_invalid_)";
1473 }
1474}
1475
1476static const char *lmod_cast_u_ptr(struct parsed_op *po,
1477 enum opr_lenmod lmod)
1478{
1479 switch (lmod) {
1480 case OPLM_QWORD:
1481 return "*(u64 *)";
1482 case OPLM_DWORD:
1483 return "*(u32 *)";
1484 case OPLM_WORD:
1485 return "*(u16 *)";
1486 case OPLM_BYTE:
1487 return "*(u8 *)";
1488 default:
1489 ferr(po, "invalid lmod: %d\n", lmod);
1490 return "(_invalid_)";
1491 }
1492}
1493
1494static const char *lmod_cast_s(struct parsed_op *po,
1495 enum opr_lenmod lmod)
1496{
1497 switch (lmod) {
1498 case OPLM_QWORD:
1499 return "(s64)";
1500 case OPLM_DWORD:
1501 return "(s32)";
1502 case OPLM_WORD:
1503 return "(s16)";
1504 case OPLM_BYTE:
1505 return "(s8)";
1506 default:
1507 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1508 return "(_invalid_)";
1509 }
1510}
1511
1512static const char *lmod_cast(struct parsed_op *po,
1513 enum opr_lenmod lmod, int is_signed)
1514{
1515 return is_signed ?
1516 lmod_cast_s(po, lmod) :
1517 lmod_cast_u(po, lmod);
1518}
1519
1520static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
1521{
1522 switch (lmod) {
1523 case OPLM_QWORD:
1524 return 8;
1525 case OPLM_DWORD:
1526 return 4;
1527 case OPLM_WORD:
1528 return 2;
1529 case OPLM_BYTE:
1530 return 1;
1531 default:
1532 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1533 return 0;
1534 }
1535}
1536
1537static const char *opr_name(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 return po->operand[opr_num].name;
1542}
1543
1544static unsigned int opr_const(struct parsed_op *po, int opr_num)
1545{
1546 if (opr_num >= po->operand_cnt)
1547 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1548 if (po->operand[opr_num].type != OPT_CONST)
1549 ferr(po, "opr %d: const expected\n", opr_num);
1550 return po->operand[opr_num].val;
1551}
1552
1553static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr)
1554{
1555 if ((unsigned int)popr->reg >= ARRAY_SIZE(regs_r32))
1556 ferr(po, "invalid reg: %d\n", popr->reg);
1557 return regs_r32[popr->reg];
1558}
1559
1560static int check_simple_cast(const char *cast, int *bits, int *is_signed)
1561{
1562 if (IS_START(cast, "(s8)") || IS_START(cast, "(u8)"))
1563 *bits = 8;
1564 else if (IS_START(cast, "(s16)") || IS_START(cast, "(u16)"))
1565 *bits = 16;
1566 else if (IS_START(cast, "(s32)") || IS_START(cast, "(u32)"))
1567 *bits = 32;
1568 else if (IS_START(cast, "(s64)") || IS_START(cast, "(u64)"))
1569 *bits = 64;
1570 else
1571 return -1;
1572
1573 *is_signed = cast[1] == 's' ? 1 : 0;
1574 return 0;
1575}
1576
1577static int check_deref_cast(const char *cast, int *bits)
1578{
1579 if (IS_START(cast, "*(u8 *)"))
1580 *bits = 8;
1581 else if (IS_START(cast, "*(u16 *)"))
1582 *bits = 16;
1583 else if (IS_START(cast, "*(u32 *)"))
1584 *bits = 32;
1585 else if (IS_START(cast, "*(u64 *)"))
1586 *bits = 64;
1587 else
1588 return -1;
1589
1590 return 0;
1591}
1592
1593// cast1 is the "final" cast
1594static const char *simplify_cast(const char *cast1, const char *cast2)
1595{
1596 static char buf[256];
1597 int bits1, bits2;
1598 int s1, s2;
1599
1600 if (cast1[0] == 0)
1601 return cast2;
1602 if (cast2[0] == 0)
1603 return cast1;
1604 if (IS(cast1, cast2))
1605 return cast1;
1606
1607 if (check_simple_cast(cast1, &bits1, &s1) == 0
1608 && check_simple_cast(cast2, &bits2, &s2) == 0)
1609 {
1610 if (bits1 <= bits2)
1611 return cast1;
1612 }
1613 if (check_simple_cast(cast1, &bits1, &s1) == 0
1614 && check_deref_cast(cast2, &bits2) == 0)
1615 {
1616 if (bits1 == bits2) {
1617 snprintf(buf, sizeof(buf), "*(%c%d *)", s1 ? 's' : 'u', bits1);
1618 return buf;
1619 }
1620 }
1621
1622 if (strchr(cast1, '*') && IS_START(cast2, "(u32)"))
1623 return cast1;
1624
1625 snprintf(buf, sizeof(buf), "%s%s", cast1, cast2);
1626 return buf;
1627}
1628
1629static const char *simplify_cast_num(const char *cast, unsigned int val)
1630{
1631 if (IS(cast, "(u8)") && val < 0x100)
1632 return "";
1633 if (IS(cast, "(s8)") && val < 0x80)
1634 return "";
1635 if (IS(cast, "(u16)") && val < 0x10000)
1636 return "";
1637 if (IS(cast, "(s16)") && val < 0x8000)
1638 return "";
1639 if (IS(cast, "(s32)") && val < 0x80000000)
1640 return "";
1641
1642 return cast;
1643}
1644
1645static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
1646 int *extra_offs)
1647{
1648 const char *p;
1649 char *endp;
1650 int namelen;
1651 int i;
1652
1653 *extra_offs = 0;
1654 namelen = strlen(name);
1655
1656 p = strchr(name, '+');
1657 if (p != NULL) {
1658 namelen = p - name;
1659 if (namelen <= 0)
1660 ferr(po, "equ parse failed for '%s'\n", name);
1661
1662 if (IS_START(p, "0x"))
1663 p += 2;
1664 *extra_offs = strtol(p, &endp, 16);
1665 if (*endp != 0)
1666 ferr(po, "equ parse failed for '%s'\n", name);
1667 }
1668
1669 for (i = 0; i < g_eqcnt; i++)
1670 if (strncmp(g_eqs[i].name, name, namelen) == 0
1671 && g_eqs[i].name[namelen] == 0)
1672 break;
1673 if (i >= g_eqcnt) {
1674 if (po != NULL)
1675 ferr(po, "unresolved equ name: '%s'\n", name);
1676 return NULL;
1677 }
1678
1679 return &g_eqs[i];
1680}
1681
1682static int is_stack_access(struct parsed_op *po,
1683 const struct parsed_opr *popr)
1684{
1685 return (parse_stack_el(popr->name, NULL, 0)
1686 || (g_bp_frame && !(po->flags & OPF_EBP_S)
1687 && IS_START(popr->name, "ebp")));
1688}
1689
1690static void parse_stack_access(struct parsed_op *po,
1691 const char *name, char *ofs_reg, int *offset_out,
1692 int *stack_ra_out, const char **bp_arg_out, int is_lea)
1693{
1694 const char *bp_arg = "";
1695 const char *p = NULL;
1696 struct parsed_equ *eq;
1697 char *endp = NULL;
1698 int stack_ra = 0;
1699 int offset = 0;
1700
1701 ofs_reg[0] = 0;
1702
1703 if (IS_START(name, "ebp-")
1704 || (IS_START(name, "ebp+") && '0' <= name[4] && name[4] <= '9'))
1705 {
1706 p = name + 4;
1707 if (IS_START(p, "0x"))
1708 p += 2;
1709 offset = strtoul(p, &endp, 16);
1710 if (name[3] == '-')
1711 offset = -offset;
1712 if (*endp != 0)
1713 ferr(po, "ebp- parse of '%s' failed\n", name);
1714 }
1715 else {
1716 bp_arg = parse_stack_el(name, ofs_reg, 0);
1717 snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
1718 eq = equ_find(po, bp_arg, &offset);
1719 if (eq == NULL)
1720 ferr(po, "detected but missing eq\n");
1721 offset += eq->offset;
1722 }
1723
1724 if (!strncmp(name, "ebp", 3))
1725 stack_ra = 4;
1726
1727 // yes it sometimes LEAs ra for compares..
1728 if (!is_lea && ofs_reg[0] == 0
1729 && stack_ra <= offset && offset < stack_ra + 4)
1730 {
1731 ferr(po, "reference to ra? %d %d\n", offset, stack_ra);
1732 }
1733
1734 *offset_out = offset;
1735 *stack_ra_out = stack_ra;
1736 if (bp_arg_out)
1737 *bp_arg_out = bp_arg;
1738}
1739
1740static int stack_frame_access(struct parsed_op *po,
1741 struct parsed_opr *popr, char *buf, size_t buf_size,
1742 const char *name, const char *cast, int is_src, int is_lea)
1743{
1744 enum opr_lenmod tmp_lmod = OPLM_UNSPEC;
1745 const char *prefix = "";
1746 const char *bp_arg = NULL;
1747 char ofs_reg[16] = { 0, };
1748 int i, arg_i, arg_s;
1749 int unaligned = 0;
1750 int stack_ra = 0;
1751 int offset = 0;
1752 int retval = -1;
1753 int sf_ofs;
1754 int lim;
1755
1756 if (po->flags & OPF_EBP_S)
1757 ferr(po, "stack_frame_access while ebp is scratch\n");
1758
1759 parse_stack_access(po, name, ofs_reg, &offset,
1760 &stack_ra, &bp_arg, is_lea);
1761
1762 if (offset > stack_ra)
1763 {
1764 arg_i = (offset - stack_ra - 4) / 4;
1765 if (arg_i < 0 || arg_i >= g_func_pp->argc_stack)
1766 {
1767 if (g_func_pp->is_vararg
1768 && arg_i == g_func_pp->argc_stack && is_lea)
1769 {
1770 // should be va_list
1771 if (cast[0] == 0)
1772 cast = "(u32)";
1773 snprintf(buf, buf_size, "%sap", cast);
1774 return -1;
1775 }
1776 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1777 offset, bp_arg, arg_i);
1778 }
1779 if (ofs_reg[0] != 0)
1780 ferr(po, "offset reg on arg access?\n");
1781
1782 for (i = arg_s = 0; i < g_func_pp->argc; i++) {
1783 if (g_func_pp->arg[i].reg != NULL)
1784 continue;
1785 if (arg_s == arg_i)
1786 break;
1787 arg_s++;
1788 }
1789 if (i == g_func_pp->argc)
1790 ferr(po, "arg %d not in prototype?\n", arg_i);
1791
1792 popr->is_ptr = g_func_pp->arg[i].type.is_ptr;
1793 retval = i;
1794
1795 switch (popr->lmod)
1796 {
1797 case OPLM_BYTE:
1798 if (is_lea)
1799 ferr(po, "lea/byte to arg?\n");
1800 if (is_src && (offset & 3) == 0)
1801 snprintf(buf, buf_size, "%sa%d",
1802 simplify_cast(cast, "(u8)"), i + 1);
1803 else
1804 snprintf(buf, buf_size, "%sBYTE%d(a%d)",
1805 cast, offset & 3, i + 1);
1806 break;
1807
1808 case OPLM_WORD:
1809 if (is_lea)
1810 ferr(po, "lea/word to arg?\n");
1811 if (offset & 1) {
1812 unaligned = 1;
1813 if (!is_src) {
1814 if (offset & 2)
1815 ferr(po, "problematic arg store\n");
1816 snprintf(buf, buf_size, "%s((char *)&a%d + 1)",
1817 simplify_cast(cast, "*(u16 *)"), i + 1);
1818 }
1819 else
1820 ferr(po, "unaligned arg word load\n");
1821 }
1822 else if (is_src && (offset & 2) == 0)
1823 snprintf(buf, buf_size, "%sa%d",
1824 simplify_cast(cast, "(u16)"), i + 1);
1825 else
1826 snprintf(buf, buf_size, "%s%sWORD(a%d)",
1827 cast, (offset & 2) ? "HI" : "LO", i + 1);
1828 break;
1829
1830 case OPLM_DWORD:
1831 if (cast[0])
1832 prefix = cast;
1833 else if (is_src)
1834 prefix = "(u32)";
1835
1836 if (offset & 3) {
1837 unaligned = 1;
1838 if (is_lea)
1839 snprintf(buf, buf_size, "(u32)&a%d + %d",
1840 i + 1, offset & 3);
1841 else if (!is_src)
1842 ferr(po, "unaligned arg store\n");
1843 else {
1844 // mov edx, [ebp+arg_4+2]; movsx ecx, dx
1845 snprintf(buf, buf_size, "%s(a%d >> %d)",
1846 prefix, i + 1, (offset & 3) * 8);
1847 }
1848 }
1849 else {
1850 snprintf(buf, buf_size, "%s%sa%d",
1851 prefix, is_lea ? "&" : "", i + 1);
1852 }
1853 break;
1854
1855 default:
1856 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
1857 }
1858
1859 if (unaligned)
1860 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1861
1862 // common problem
1863 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type);
1864 if (tmp_lmod != OPLM_DWORD
1865 && (unaligned || (!is_src && lmod_bytes(po, tmp_lmod)
1866 < lmod_bytes(po, popr->lmod) + (offset & 3))))
1867 {
1868 ferr(po, "bp_arg arg%d/w offset %d and type '%s' is too small\n",
1869 i + 1, offset, g_func_pp->arg[i].type.name);
1870 }
1871 // can't check this because msvc likes to reuse
1872 // arg space for scratch..
1873 //if (popr->is_ptr && popr->lmod != OPLM_DWORD)
1874 // ferr(po, "bp_arg arg%d: non-dword ptr access\n", i + 1);
1875 }
1876 else
1877 {
1878 if (g_stack_fsz == 0)
1879 ferr(po, "stack var access without stackframe\n");
1880 g_stack_frame_used = 1;
1881
1882 sf_ofs = g_stack_fsz + offset;
1883 lim = (ofs_reg[0] != 0) ? -4 : 0;
1884 if (offset > 0 || sf_ofs < lim)
1885 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
1886
1887 if (is_lea)
1888 prefix = "(u32)&";
1889 else
1890 prefix = cast;
1891
1892 switch (popr->lmod)
1893 {
1894 case OPLM_BYTE:
1895 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1896 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1897 break;
1898
1899 case OPLM_WORD:
1900 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1901 // known unaligned or possibly unaligned
1902 strcat(g_comment, " unaligned");
1903 if (prefix[0] == 0)
1904 prefix = "*(u16 *)&";
1905 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1906 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1907 break;
1908 }
1909 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
1910 break;
1911
1912 case OPLM_DWORD:
1913 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1914 // known unaligned or possibly unaligned
1915 strcat(g_comment, " unaligned");
1916 if (prefix[0] == 0)
1917 prefix = "*(u32 *)&";
1918 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1919 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1920 break;
1921 }
1922 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
1923 break;
1924
1925 case OPLM_QWORD:
1926 ferr_assert(po, !(sf_ofs & 7));
1927 ferr_assert(po, ofs_reg[0] == 0);
1928 // float callers set is_lea
1929 ferr_assert(po, is_lea);
1930 snprintf(buf, buf_size, "%ssf.q[%d]", prefix, sf_ofs / 8);
1931 break;
1932
1933 default:
1934 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
1935 }
1936 }
1937
1938 return retval;
1939}
1940
1941static void check_func_pp(struct parsed_op *po,
1942 const struct parsed_proto *pp, const char *pfx)
1943{
1944 enum opr_lenmod tmp_lmod;
1945 char buf[256];
1946 int ret, i;
1947
1948 if (pp->argc_reg != 0) {
1949 if (/*!g_allow_regfunc &&*/ !pp->is_fastcall) {
1950 pp_print(buf, sizeof(buf), pp);
1951 ferr(po, "%s: unexpected reg arg in icall: %s\n", pfx, buf);
1952 }
1953 if (pp->argc_stack > 0 && pp->argc_reg != 2)
1954 ferr(po, "%s: %d reg arg(s) with %d stack arg(s)\n",
1955 pfx, pp->argc_reg, pp->argc_stack);
1956 }
1957
1958 // fptrs must use 32bit args, callsite might have no information and
1959 // lack a cast to smaller types, which results in incorrectly masked
1960 // args passed (callee may assume masked args, it does on ARM)
1961 if (!pp->is_osinc) {
1962 for (i = 0; i < pp->argc; i++) {
1963 ret = guess_lmod_from_c_type(&tmp_lmod, &pp->arg[i].type);
1964 if (ret && tmp_lmod != OPLM_DWORD)
1965 ferr(po, "reference to %s with arg%d '%s'\n", pp->name,
1966 i + 1, pp->arg[i].type.name);
1967 }
1968 }
1969}
1970
1971static const char *check_label_read_ref(struct parsed_op *po,
1972 const char *name)
1973{
1974 const struct parsed_proto *pp;
1975
1976 pp = proto_parse(g_fhdr, name, 0);
1977 if (pp == NULL)
1978 ferr(po, "proto_parse failed for ref '%s'\n", name);
1979
1980 if (pp->is_func)
1981 check_func_pp(po, pp, "ref");
1982
1983 return pp->name;
1984}
1985
1986static char *out_src_opr(char *buf, size_t buf_size,
1987 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
1988 int is_lea)
1989{
1990 char tmp1[256], tmp2[256];
1991 char expr[256];
1992 const char *name;
1993 char *p;
1994 int ret;
1995
1996 if (cast == NULL)
1997 cast = "";
1998
1999 switch (popr->type) {
2000 case OPT_REG:
2001 if (is_lea)
2002 ferr(po, "lea from reg?\n");
2003
2004 switch (popr->lmod) {
2005 case OPLM_QWORD:
2006 snprintf(buf, buf_size, "%s%s.q", cast, opr_reg_p(po, popr));
2007 break;
2008 case OPLM_DWORD:
2009 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
2010 break;
2011 case OPLM_WORD:
2012 snprintf(buf, buf_size, "%s%s",
2013 simplify_cast(cast, "(u16)"), opr_reg_p(po, popr));
2014 break;
2015 case OPLM_BYTE:
2016 if (popr->name[1] == 'h') // XXX..
2017 snprintf(buf, buf_size, "%s(%s >> 8)",
2018 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
2019 else
2020 snprintf(buf, buf_size, "%s%s",
2021 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
2022 break;
2023 default:
2024 ferr(po, "invalid src lmod: %d\n", popr->lmod);
2025 }
2026 break;
2027
2028 case OPT_REGMEM:
2029 if (is_stack_access(po, popr)) {
2030 stack_frame_access(po, popr, buf, buf_size,
2031 popr->name, cast, 1, is_lea);
2032 break;
2033 }
2034
2035 strcpy(expr, popr->name);
2036 if (strchr(expr, '[')) {
2037 // special case: '[' can only be left for label[reg] form
2038 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
2039 if (ret != 2)
2040 ferr(po, "parse failure for '%s'\n", expr);
2041 if (tmp1[0] == '(') {
2042 // (off_4FFF50+3)[eax]
2043 p = strchr(tmp1 + 1, ')');
2044 if (p == NULL || p[1] != 0)
2045 ferr(po, "parse failure (2) for '%s'\n", expr);
2046 *p = 0;
2047 memmove(tmp1, tmp1 + 1, strlen(tmp1));
2048 }
2049 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
2050 }
2051
2052 // XXX: do we need more parsing?
2053 if (is_lea) {
2054 snprintf(buf, buf_size, "%s", expr);
2055 break;
2056 }
2057
2058 snprintf(buf, buf_size, "%s(%s)",
2059 simplify_cast(cast, lmod_cast_u_ptr(po, popr->lmod)), expr);
2060 break;
2061
2062 case OPT_LABEL:
2063 name = check_label_read_ref(po, popr->name);
2064 if (cast[0] == 0 && popr->is_ptr)
2065 cast = "(u32)";
2066
2067 if (is_lea)
2068 snprintf(buf, buf_size, "(u32)&%s", name);
2069 else if (popr->size_lt)
2070 snprintf(buf, buf_size, "%s%s%s%s", cast,
2071 lmod_cast_u_ptr(po, popr->lmod),
2072 popr->is_array ? "" : "&", name);
2073 else
2074 snprintf(buf, buf_size, "%s%s%s", cast, name,
2075 popr->is_array ? "[0]" : "");
2076 break;
2077
2078 case OPT_OFFSET:
2079 name = check_label_read_ref(po, popr->name);
2080 if (cast[0] == 0)
2081 cast = "(u32)";
2082 if (is_lea)
2083 ferr(po, "lea an offset?\n");
2084 snprintf(buf, buf_size, "%s&%s", cast, name);
2085 break;
2086
2087 case OPT_CONST:
2088 if (is_lea)
2089 ferr(po, "lea from const?\n");
2090
2091 printf_number(tmp1, sizeof(tmp1), popr->val);
2092 if (popr->val == 0 && strchr(cast, '*'))
2093 snprintf(buf, buf_size, "NULL");
2094 else
2095 snprintf(buf, buf_size, "%s%s",
2096 simplify_cast_num(cast, popr->val), tmp1);
2097 break;
2098
2099 default:
2100 ferr(po, "invalid src type: %d\n", popr->type);
2101 }
2102
2103 return buf;
2104}
2105
2106// note: may set is_ptr (we find that out late for ebp frame..)
2107static char *out_dst_opr(char *buf, size_t buf_size,
2108 struct parsed_op *po, struct parsed_opr *popr)
2109{
2110 switch (popr->type) {
2111 case OPT_REG:
2112 switch (popr->lmod) {
2113 case OPLM_QWORD:
2114 snprintf(buf, buf_size, "%s.q", opr_reg_p(po, popr));
2115 break;
2116 case OPLM_DWORD:
2117 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
2118 break;
2119 case OPLM_WORD:
2120 // ugh..
2121 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
2122 break;
2123 case OPLM_BYTE:
2124 // ugh..
2125 if (popr->name[1] == 'h') // XXX..
2126 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
2127 else
2128 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
2129 break;
2130 default:
2131 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
2132 }
2133 break;
2134
2135 case OPT_REGMEM:
2136 if (is_stack_access(po, popr)) {
2137 stack_frame_access(po, popr, buf, buf_size,
2138 popr->name, "", 0, 0);
2139 break;
2140 }
2141
2142 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
2143
2144 case OPT_LABEL:
2145 if (popr->size_mismatch)
2146 snprintf(buf, buf_size, "%s%s%s",
2147 lmod_cast_u_ptr(po, popr->lmod),
2148 popr->is_array ? "" : "&", popr->name);
2149 else
2150 snprintf(buf, buf_size, "%s%s", popr->name,
2151 popr->is_array ? "[0]" : "");
2152 break;
2153
2154 default:
2155 ferr(po, "invalid dst type: %d\n", popr->type);
2156 }
2157
2158 return buf;
2159}
2160
2161static char *out_src_opr_u32(char *buf, size_t buf_size,
2162 struct parsed_op *po, struct parsed_opr *popr)
2163{
2164 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
2165}
2166
2167static char *out_src_opr_float(char *buf, size_t buf_size,
2168 struct parsed_op *po, struct parsed_opr *popr)
2169{
2170 const char *cast = NULL;
2171 char tmp[256];
2172
2173 switch (popr->type) {
2174 case OPT_REG:
2175 if (popr->reg < xST0 || popr->reg > xST7)
2176 ferr(po, "bad reg: %d\n", popr->reg);
2177
2178 snprintf(buf, buf_size, "f_st%d", popr->reg - xST0);
2179 break;
2180
2181 case OPT_REGMEM:
2182 case OPT_LABEL:
2183 case OPT_OFFSET:
2184 switch (popr->lmod) {
2185 case OPLM_QWORD:
2186 cast = "double";
2187 break;
2188 case OPLM_DWORD:
2189 cast = "float";
2190 break;
2191 default:
2192 ferr(po, "unhandled lmod: %d\n", popr->lmod);
2193 break;
2194 }
2195 out_src_opr(tmp, sizeof(tmp), po, popr, "", 1);
2196 snprintf(buf, buf_size, "*((%s *)%s)", cast, tmp);
2197 break;
2198
2199 default:
2200 ferr(po, "invalid float type: %d\n", popr->type);
2201 }
2202
2203 return buf;
2204}
2205
2206static char *out_dst_opr_float(char *buf, size_t buf_size,
2207 struct parsed_op *po, struct parsed_opr *popr)
2208{
2209 // same?
2210 return out_src_opr_float(buf, buf_size, po, popr);
2211}
2212
2213static void out_test_for_cc(char *buf, size_t buf_size,
2214 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
2215 enum opr_lenmod lmod, const char *expr)
2216{
2217 const char *cast, *scast;
2218
2219 cast = lmod_cast_u(po, lmod);
2220 scast = lmod_cast_s(po, lmod);
2221
2222 switch (pfo) {
2223 case PFO_Z:
2224 case PFO_BE: // CF=1||ZF=1; CF=0
2225 snprintf(buf, buf_size, "(%s%s %s 0)",
2226 cast, expr, is_inv ? "!=" : "==");
2227 break;
2228
2229 case PFO_S:
2230 case PFO_L: // SF!=OF; OF=0
2231 snprintf(buf, buf_size, "(%s%s %s 0)",
2232 scast, expr, is_inv ? ">=" : "<");
2233 break;
2234
2235 case PFO_LE: // ZF=1||SF!=OF; OF=0
2236 snprintf(buf, buf_size, "(%s%s %s 0)",
2237 scast, expr, is_inv ? ">" : "<=");
2238 break;
2239
2240 default:
2241 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
2242 }
2243}
2244
2245static void out_cmp_for_cc(char *buf, size_t buf_size,
2246 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
2247{
2248 const char *cast, *scast, *cast_use;
2249 char buf1[256], buf2[256];
2250 enum opr_lenmod lmod;
2251
2252 if (po->op != OP_DEC && po->operand[0].lmod != po->operand[1].lmod)
2253 ferr(po, "%s: lmod mismatch: %d %d\n", __func__,
2254 po->operand[0].lmod, po->operand[1].lmod);
2255 lmod = po->operand[0].lmod;
2256
2257 cast = lmod_cast_u(po, lmod);
2258 scast = lmod_cast_s(po, lmod);
2259
2260 switch (pfo) {
2261 case PFO_C:
2262 case PFO_Z:
2263 case PFO_BE: // !a
2264 cast_use = cast;
2265 break;
2266
2267 case PFO_S:
2268 case PFO_L: // !ge
2269 case PFO_LE:
2270 cast_use = scast;
2271 break;
2272
2273 default:
2274 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
2275 }
2276
2277 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], cast_use, 0);
2278 if (po->op == OP_DEC)
2279 snprintf(buf2, sizeof(buf2), "1");
2280 else
2281 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], cast_use, 0);
2282
2283 switch (pfo) {
2284 case PFO_C:
2285 // note: must be unsigned compare
2286 snprintf(buf, buf_size, "(%s %s %s)",
2287 buf1, is_inv ? ">=" : "<", buf2);
2288 break;
2289
2290 case PFO_Z:
2291 snprintf(buf, buf_size, "(%s %s %s)",
2292 buf1, is_inv ? "!=" : "==", buf2);
2293 break;
2294
2295 case PFO_BE: // !a
2296 // note: must be unsigned compare
2297 snprintf(buf, buf_size, "(%s %s %s)",
2298 buf1, is_inv ? ">" : "<=", buf2);
2299
2300 // annoying case
2301 if (is_inv && lmod == OPLM_BYTE
2302 && po->operand[1].type == OPT_CONST
2303 && po->operand[1].val == 0xff)
2304 {
2305 snprintf(g_comment, sizeof(g_comment), "if %s", buf);
2306 snprintf(buf, buf_size, "(0)");
2307 }
2308 break;
2309
2310 // note: must be signed compare
2311 case PFO_S:
2312 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
2313 scast, buf1, buf2, is_inv ? ">=" : "<");
2314 break;
2315
2316 case PFO_L: // !ge
2317 snprintf(buf, buf_size, "(%s %s %s)",
2318 buf1, is_inv ? ">=" : "<", buf2);
2319 break;
2320
2321 case PFO_LE: // !g
2322 snprintf(buf, buf_size, "(%s %s %s)",
2323 buf1, is_inv ? ">" : "<=", buf2);
2324 break;
2325
2326 default:
2327 break;
2328 }
2329}
2330
2331static void out_cmp_test(char *buf, size_t buf_size,
2332 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
2333{
2334 char buf1[256], buf2[256], buf3[256];
2335
2336 if (po->op == OP_TEST) {
2337 if (IS(opr_name(po, 0), opr_name(po, 1))) {
2338 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
2339 }
2340 else {
2341 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
2342 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
2343 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
2344 }
2345 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
2346 po->operand[0].lmod, buf3);
2347 }
2348 else if (po->op == OP_CMP) {
2349 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv);
2350 }
2351 else
2352 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
2353}
2354
2355static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
2356 struct parsed_opr *popr2)
2357{
2358 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
2359 ferr(po, "missing lmod for both operands\n");
2360
2361 if (popr1->lmod == OPLM_UNSPEC)
2362 popr1->lmod = popr2->lmod;
2363 else if (popr2->lmod == OPLM_UNSPEC)
2364 popr2->lmod = popr1->lmod;
2365 else if (popr1->lmod != popr2->lmod) {
2366 if (popr1->type_from_var) {
2367 popr1->size_mismatch = 1;
2368 if (popr1->lmod < popr2->lmod)
2369 popr1->size_lt = 1;
2370 popr1->lmod = popr2->lmod;
2371 }
2372 else if (popr2->type_from_var) {
2373 popr2->size_mismatch = 1;
2374 if (popr2->lmod < popr1->lmod)
2375 popr2->size_lt = 1;
2376 popr2->lmod = popr1->lmod;
2377 }
2378 else
2379 ferr(po, "conflicting lmods: %d vs %d\n",
2380 popr1->lmod, popr2->lmod);
2381 }
2382}
2383
2384static const char *op_to_c(struct parsed_op *po)
2385{
2386 switch (po->op)
2387 {
2388 case OP_ADD:
2389 case OP_ADC:
2390 return "+";
2391 case OP_SUB:
2392 case OP_SBB:
2393 return "-";
2394 case OP_AND:
2395 return "&";
2396 case OP_OR:
2397 return "|";
2398 case OP_XOR:
2399 return "^";
2400 case OP_SHL:
2401 return "<<";
2402 case OP_SHR:
2403 return ">>";
2404 case OP_MUL:
2405 case OP_IMUL:
2406 return "*";
2407 default:
2408 ferr(po, "op_to_c was supplied with %d\n", po->op);
2409 }
2410}
2411
2412// last op in stream - unconditional branch or ret
2413#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
2414 || ((ops[_i].flags & (OPF_JMP|OPF_CJMP|OPF_RMD)) == OPF_JMP \
2415 && ops[_i].op != OP_CALL))
2416
2417#define check_i(po, i) \
2418 if ((i) < 0) \
2419 ferr(po, "bad " #i ": %d\n", i)
2420
2421// note: this skips over calls and rm'd stuff assuming they're handled
2422// so it's intended to use at one of final passes
2423static int scan_for_pop(int i, int opcnt, int magic, int reg,
2424 int depth, int flags_set)
2425{
2426 struct parsed_op *po;
2427 int relevant;
2428 int ret = 0;
2429 int j;
2430
2431 for (; i < opcnt; i++) {
2432 po = &ops[i];
2433 if (po->cc_scratch == magic)
2434 return ret; // already checked
2435 po->cc_scratch = magic;
2436
2437 if (po->flags & OPF_TAIL) {
2438 if (po->op == OP_CALL) {
2439 if (po->pp != NULL && po->pp->is_noreturn)
2440 // assume no stack cleanup for noreturn
2441 return 1;
2442 }
2443 return -1; // deadend
2444 }
2445
2446 if (po->flags & (OPF_RMD|OPF_DONE|OPF_FARG))
2447 continue;
2448
2449 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2450 if (po->btj != NULL) {
2451 // jumptable
2452 for (j = 0; j < po->btj->count; j++) {
2453 check_i(po, po->btj->d[j].bt_i);
2454 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, magic, reg,
2455 depth, flags_set);
2456 if (ret < 0)
2457 return ret; // dead end
2458 }
2459 return ret;
2460 }
2461
2462 check_i(po, po->bt_i);
2463 if (po->flags & OPF_CJMP) {
2464 ret |= scan_for_pop(po->bt_i, opcnt, magic, reg,
2465 depth, flags_set);
2466 if (ret < 0)
2467 return ret; // dead end
2468 }
2469 else {
2470 i = po->bt_i - 1;
2471 }
2472 continue;
2473 }
2474
2475 relevant = 0;
2476 if ((po->op == OP_POP || po->op == OP_PUSH)
2477 && po->operand[0].type == OPT_REG && po->operand[0].reg == reg)
2478 {
2479 relevant = 1;
2480 }
2481
2482 if (po->op == OP_PUSH) {
2483 depth++;
2484 }
2485 else if (po->op == OP_POP) {
2486 if (relevant && depth == 0) {
2487 po->flags |= flags_set;
2488 return 1;
2489 }
2490 depth--;
2491 }
2492 }
2493
2494 return -1;
2495}
2496
2497// scan for 'reg' pop backwards starting from i
2498// intended to use for register restore search, so other reg
2499// references are considered an error
2500static int scan_for_rsave_pop_reg(int i, int magic, int reg, int set_flags)
2501{
2502 struct parsed_op *po;
2503 struct label_ref *lr;
2504 int ret = 0;
2505
2506 ops[i].cc_scratch = magic;
2507
2508 while (1)
2509 {
2510 if (g_labels[i] != NULL) {
2511 lr = &g_label_refs[i];
2512 for (; lr != NULL; lr = lr->next) {
2513 check_i(&ops[i], lr->i);
2514 ret |= scan_for_rsave_pop_reg(lr->i, magic, reg, set_flags);
2515 if (ret < 0)
2516 return ret;
2517 }
2518 if (i > 0 && LAST_OP(i - 1))
2519 return ret;
2520 }
2521
2522 i--;
2523 if (i < 0)
2524 break;
2525
2526 if (ops[i].cc_scratch == magic)
2527 return ret;
2528 ops[i].cc_scratch = magic;
2529
2530 po = &ops[i];
2531 if (po->op == OP_POP && po->operand[0].reg == reg) {
2532 if (po->flags & (OPF_RMD|OPF_DONE))
2533 return -1;
2534
2535 po->flags |= set_flags;
2536 return 1;
2537 }
2538
2539 // this also covers the case where we reach corresponding push
2540 if ((po->regmask_dst | po->regmask_src) & (1 << reg))
2541 return -1;
2542 }
2543
2544 // nothing interesting on this path
2545 return 0;
2546}
2547
2548static void find_reachable_exits(int i, int opcnt, int magic,
2549 int *exits, int *exit_count)
2550{
2551 struct parsed_op *po;
2552 int j;
2553
2554 for (; i < opcnt; i++)
2555 {
2556 po = &ops[i];
2557 if (po->cc_scratch == magic)
2558 return;
2559 po->cc_scratch = magic;
2560
2561 if (po->flags & OPF_TAIL) {
2562 ferr_assert(po, *exit_count < MAX_EXITS);
2563 exits[*exit_count] = i;
2564 (*exit_count)++;
2565 return;
2566 }
2567
2568 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2569 if (po->flags & OPF_RMD)
2570 continue;
2571
2572 if (po->btj != NULL) {
2573 for (j = 0; j < po->btj->count; j++) {
2574 check_i(po, po->btj->d[j].bt_i);
2575 find_reachable_exits(po->btj->d[j].bt_i, opcnt, magic,
2576 exits, exit_count);
2577 }
2578 return;
2579 }
2580
2581 check_i(po, po->bt_i);
2582 if (po->flags & OPF_CJMP)
2583 find_reachable_exits(po->bt_i, opcnt, magic, exits, exit_count);
2584 else
2585 i = po->bt_i - 1;
2586 continue;
2587 }
2588 }
2589}
2590
2591// scan for 'reg' pop backwards starting from exits (all paths)
2592static int scan_for_pop_ret(int i, int opcnt, int reg, int set_flags)
2593{
2594 static int exits[MAX_EXITS];
2595 static int exit_count;
2596 int j, ret;
2597
2598 if (!set_flags) {
2599 exit_count = 0;
2600 find_reachable_exits(i, opcnt, i + opcnt * 15, exits,
2601 &exit_count);
2602 ferr_assert(&ops[i], exit_count > 0);
2603 }
2604
2605 for (j = 0; j < exit_count; j++) {
2606 ret = scan_for_rsave_pop_reg(exits[j], i + opcnt * 16 + set_flags,
2607 reg, set_flags);
2608 if (ret == -1)
2609 return -1;
2610 }
2611
2612 return 1;
2613}
2614
2615// scan for one or more pop of push <const>
2616static int scan_for_pop_const_r(int i, int opcnt, int magic,
2617 int push_i, int is_probe)
2618{
2619 struct parsed_op *po;
2620 struct label_ref *lr;
2621 int ret = 0;
2622 int j;
2623
2624 for (; i < opcnt; i++)
2625 {
2626 po = &ops[i];
2627 if (po->cc_scratch == magic)
2628 return ret; // already checked
2629 po->cc_scratch = magic;
2630
2631 if (po->flags & OPF_JMP) {
2632 if (po->flags & OPF_RMD)
2633 continue;
2634 if (po->op == OP_CALL)
2635 return -1;
2636
2637 if (po->btj != NULL) {
2638 for (j = 0; j < po->btj->count; j++) {
2639 check_i(po, po->btj->d[j].bt_i);
2640 ret |= scan_for_pop_const_r(po->btj->d[j].bt_i, opcnt, magic,
2641 push_i, is_probe);
2642 if (ret < 0)
2643 return ret;
2644 }
2645 return ret;
2646 }
2647
2648 check_i(po, po->bt_i);
2649 if (po->flags & OPF_CJMP) {
2650 ret |= scan_for_pop_const_r(po->bt_i, opcnt, magic, push_i,
2651 is_probe);
2652 if (ret < 0)
2653 return ret;
2654 }
2655 else {
2656 i = po->bt_i - 1;
2657 }
2658 continue;
2659 }
2660
2661 if ((po->flags & (OPF_TAIL|OPF_RSAVE)) || po->op == OP_PUSH)
2662 return -1;
2663
2664 if (g_labels[i] != NULL) {
2665 // all refs must be visited
2666 lr = &g_label_refs[i];
2667 for (; lr != NULL; lr = lr->next) {
2668 check_i(po, lr->i);
2669 if (ops[lr->i].cc_scratch != magic)
2670 return -1;
2671 }
2672 if (i > 0 && !LAST_OP(i - 1) && ops[i - 1].cc_scratch != magic)
2673 return -1;
2674 }
2675
2676 if (po->op == OP_POP)
2677 {
2678 if (po->flags & (OPF_RMD|OPF_DONE))
2679 return -1;
2680
2681 if (!is_probe) {
2682 po->flags |= OPF_DONE;
2683 po->datap = &ops[push_i];
2684 }
2685 return 1;
2686 }
2687 }
2688
2689 return -1;
2690}
2691
2692static void scan_for_pop_const(int i, int opcnt, int magic)
2693{
2694 int ret;
2695
2696 ret = scan_for_pop_const_r(i + 1, opcnt, magic, i, 1);
2697 if (ret == 1) {
2698 ops[i].flags |= OPF_RMD | OPF_DONE;
2699 scan_for_pop_const_r(i + 1, opcnt, magic + 1, i, 0);
2700 }
2701}
2702
2703// check if all branch targets within a marked path are also marked
2704// note: the path checked must not be empty or end with a branch
2705static int check_path_branches(int opcnt, int magic)
2706{
2707 struct parsed_op *po;
2708 int i, j;
2709
2710 for (i = 0; i < opcnt; i++) {
2711 po = &ops[i];
2712 if (po->cc_scratch != magic)
2713 continue;
2714
2715 if (po->flags & OPF_JMP) {
2716 if ((po->flags & OPF_RMD) || po->op == OP_CALL)
2717 continue;
2718
2719 if (po->btj != NULL) {
2720 for (j = 0; j < po->btj->count; j++) {
2721 check_i(po, po->btj->d[j].bt_i);
2722 if (ops[po->btj->d[j].bt_i].cc_scratch != magic)
2723 return 0;
2724 }
2725 }
2726
2727 check_i(po, po->bt_i);
2728 if (ops[po->bt_i].cc_scratch != magic)
2729 return 0;
2730 if ((po->flags & OPF_CJMP) && ops[i + 1].cc_scratch != magic)
2731 return 0;
2732 }
2733 }
2734
2735 return 1;
2736}
2737
2738// scan for multiple pushes for given pop
2739static int scan_pushes_for_pop_r(int i, int magic, int pop_i,
2740 int is_probe)
2741{
2742 int reg = ops[pop_i].operand[0].reg;
2743 struct parsed_op *po;
2744 struct label_ref *lr;
2745 int ret = 0;
2746
2747 ops[i].cc_scratch = magic;
2748
2749 while (1)
2750 {
2751 if (g_labels[i] != NULL) {
2752 lr = &g_label_refs[i];
2753 for (; lr != NULL; lr = lr->next) {
2754 check_i(&ops[i], lr->i);
2755 ret |= scan_pushes_for_pop_r(lr->i, magic, pop_i, is_probe);
2756 if (ret < 0)
2757 return ret;
2758 }
2759 if (i > 0 && LAST_OP(i - 1))
2760 return ret;
2761 }
2762
2763 i--;
2764 if (i < 0)
2765 break;
2766
2767 if (ops[i].cc_scratch == magic)
2768 return ret;
2769 ops[i].cc_scratch = magic;
2770
2771 po = &ops[i];
2772 if (po->op == OP_CALL)
2773 return -1;
2774 if ((po->flags & (OPF_TAIL|OPF_RSAVE)) || po->op == OP_POP)
2775 return -1;
2776
2777 if (po->op == OP_PUSH)
2778 {
2779 if (po->datap != NULL)
2780 return -1;
2781 if (po->operand[0].type == OPT_REG && po->operand[0].reg == reg)
2782 // leave this case for reg save/restore handlers
2783 return -1;
2784
2785 if (!is_probe) {
2786 po->flags |= OPF_PPUSH | OPF_DONE;
2787 po->datap = &ops[pop_i];
2788 }
2789 return 1;
2790 }
2791 }
2792
2793 return -1;
2794}
2795
2796static void scan_pushes_for_pop(int i, int opcnt, int *regmask_pp)
2797{
2798 int magic = i + opcnt * 14;
2799 int ret;
2800
2801 ret = scan_pushes_for_pop_r(i, magic, i, 1);
2802 if (ret == 1) {
2803 ret = check_path_branches(opcnt, magic);
2804 if (ret == 1) {
2805 ops[i].flags |= OPF_PPUSH | OPF_DONE;
2806 *regmask_pp |= 1 << ops[i].operand[0].reg;
2807 scan_pushes_for_pop_r(i, magic + 1, i, 0);
2808 }
2809 }
2810}
2811
2812static void scan_propagate_df(int i, int opcnt)
2813{
2814 struct parsed_op *po = &ops[i];
2815 int j;
2816
2817 for (; i < opcnt; i++) {
2818 po = &ops[i];
2819 if (po->flags & OPF_DF)
2820 return; // already resolved
2821 po->flags |= OPF_DF;
2822
2823 if (po->op == OP_CALL)
2824 ferr(po, "call with DF set?\n");
2825
2826 if (po->flags & OPF_JMP) {
2827 if (po->btj != NULL) {
2828 // jumptable
2829 for (j = 0; j < po->btj->count; j++) {
2830 check_i(po, po->btj->d[j].bt_i);
2831 scan_propagate_df(po->btj->d[j].bt_i, opcnt);
2832 }
2833 return;
2834 }
2835
2836 if (po->flags & OPF_RMD)
2837 continue;
2838 check_i(po, po->bt_i);
2839 if (po->flags & OPF_CJMP)
2840 scan_propagate_df(po->bt_i, opcnt);
2841 else
2842 i = po->bt_i - 1;
2843 continue;
2844 }
2845
2846 if (po->flags & OPF_TAIL)
2847 break;
2848
2849 if (po->op == OP_CLD) {
2850 po->flags |= OPF_RMD | OPF_DONE;
2851 return;
2852 }
2853 }
2854
2855 ferr(po, "missing DF clear?\n");
2856}
2857
2858// is operand 'opr' referenced by parsed_op 'po'?
2859static int is_opr_referenced(const struct parsed_opr *opr,
2860 const struct parsed_op *po)
2861{
2862 int i, mask;
2863
2864 if (opr->type == OPT_REG) {
2865 mask = po->regmask_dst | po->regmask_src;
2866 if (po->op == OP_CALL)
2867 mask |= (1 << xAX) | (1 << xCX) | (1 << xDX);
2868 if ((1 << opr->reg) & mask)
2869 return 1;
2870 else
2871 return 0;
2872 }
2873
2874 for (i = 0; i < po->operand_cnt; i++)
2875 if (IS(po->operand[0].name, opr->name))
2876 return 1;
2877
2878 return 0;
2879}
2880
2881// is operand 'opr' read by parsed_op 'po'?
2882static int is_opr_read(const struct parsed_opr *opr,
2883 const struct parsed_op *po)
2884{
2885 if (opr->type == OPT_REG) {
2886 if (po->regmask_src & (1 << opr->reg))
2887 return 1;
2888 else
2889 return 0;
2890 }
2891
2892 // yes I'm lazy
2893 return 0;
2894}
2895
2896// is operand 'opr' modified by parsed_op 'po'?
2897static int is_opr_modified(const struct parsed_opr *opr,
2898 const struct parsed_op *po)
2899{
2900 int mask;
2901
2902 if (opr->type == OPT_REG) {
2903 if (po->op == OP_CALL) {
2904 mask = po->regmask_dst;
2905 mask |= (1 << xAX) | (1 << xCX) | (1 << xDX); // ?
2906 if (mask & (1 << opr->reg))
2907 return 1;
2908 else
2909 return 0;
2910 }
2911
2912 if (po->regmask_dst & (1 << opr->reg))
2913 return 1;
2914 else
2915 return 0;
2916 }
2917
2918 return IS(po->operand[0].name, opr->name);
2919}
2920
2921// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
2922static int is_any_opr_modified(const struct parsed_op *po_test,
2923 const struct parsed_op *po, int c_mode)
2924{
2925 int mask;
2926 int i;
2927
2928 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2929 return 0;
2930
2931 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2932 return 0;
2933
2934 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
2935 return 1;
2936
2937 // in reality, it can wreck any register, but in decompiled C
2938 // version it can only overwrite eax or edx:eax
2939 mask = (1 << xAX) | (1 << xDX);
2940 if (!c_mode)
2941 mask |= 1 << xCX;
2942
2943 if (po->op == OP_CALL
2944 && ((po_test->regmask_src | po_test->regmask_dst) & mask))
2945 return 1;
2946
2947 for (i = 0; i < po_test->operand_cnt; i++)
2948 if (IS(po_test->operand[i].name, po->operand[0].name))
2949 return 1;
2950
2951 return 0;
2952}
2953
2954// scan for any po_test operand modification in range given
2955static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt,
2956 int c_mode)
2957{
2958 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2959 return -1;
2960
2961 for (; i < opcnt; i++) {
2962 if (is_any_opr_modified(po_test, &ops[i], c_mode))
2963 return i;
2964 }
2965
2966 return -1;
2967}
2968
2969// scan for po_test operand[0] modification in range given
2970static int scan_for_mod_opr0(struct parsed_op *po_test,
2971 int i, int opcnt)
2972{
2973 for (; i < opcnt; i++) {
2974 if (is_opr_modified(&po_test->operand[0], &ops[i]))
2975 return i;
2976 }
2977
2978 return -1;
2979}
2980
2981static int scan_for_flag_set(int i, int magic, int *branched,
2982 int *setters, int *setter_cnt)
2983{
2984 struct label_ref *lr;
2985 int ret;
2986
2987 while (i >= 0) {
2988 if (ops[i].cc_scratch == magic) {
2989 // is this a problem?
2990 //ferr(&ops[i], "%s looped\n", __func__);
2991 return 0;
2992 }
2993 ops[i].cc_scratch = magic;
2994
2995 if (g_labels[i] != NULL) {
2996 *branched = 1;
2997
2998 lr = &g_label_refs[i];
2999 for (; lr->next; lr = lr->next) {
3000 check_i(&ops[i], lr->i);
3001 ret = scan_for_flag_set(lr->i, magic,
3002 branched, setters, setter_cnt);
3003 if (ret < 0)
3004 return ret;
3005 }
3006
3007 check_i(&ops[i], lr->i);
3008 if (i > 0 && LAST_OP(i - 1)) {
3009 i = lr->i;
3010 continue;
3011 }
3012 ret = scan_for_flag_set(lr->i, magic,
3013 branched, setters, setter_cnt);
3014 if (ret < 0)
3015 return ret;
3016 }
3017 i--;
3018
3019 if (ops[i].flags & OPF_FLAGS) {
3020 setters[*setter_cnt] = i;
3021 (*setter_cnt)++;
3022 return 0;
3023 }
3024
3025 if ((ops[i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP)
3026 return -1;
3027 }
3028
3029 return -1;
3030}
3031
3032// scan back for cdq, if anything modifies edx, fail
3033static int scan_for_cdq_edx(int i)
3034{
3035 while (i >= 0) {
3036 if (g_labels[i] != NULL) {
3037 if (g_label_refs[i].next != NULL)
3038 return -1;
3039 if (i > 0 && LAST_OP(i - 1)) {
3040 i = g_label_refs[i].i;
3041 continue;
3042 }
3043 return -1;
3044 }
3045 i--;
3046
3047 if (ops[i].op == OP_CDQ)
3048 return i;
3049
3050 if (ops[i].regmask_dst & (1 << xDX))
3051 return -1;
3052 }
3053
3054 return -1;
3055}
3056
3057static int scan_for_reg_clear(int i, int reg)
3058{
3059 while (i >= 0) {
3060 if (g_labels[i] != NULL) {
3061 if (g_label_refs[i].next != NULL)
3062 return -1;
3063 if (i > 0 && LAST_OP(i - 1)) {
3064 i = g_label_refs[i].i;
3065 continue;
3066 }
3067 return -1;
3068 }
3069 i--;
3070
3071 if (ops[i].op == OP_XOR
3072 && ops[i].operand[0].lmod == OPLM_DWORD
3073 && ops[i].operand[0].reg == ops[i].operand[1].reg
3074 && ops[i].operand[0].reg == reg)
3075 return i;
3076
3077 if (ops[i].regmask_dst & (1 << reg))
3078 return -1;
3079 }
3080
3081 return -1;
3082}
3083
3084static void patch_esp_adjust(struct parsed_op *po, int adj)
3085{
3086 ferr_assert(po, po->op == OP_ADD);
3087 ferr_assert(po, IS(opr_name(po, 0), "esp"));
3088 ferr_assert(po, po->operand[1].type == OPT_CONST);
3089
3090 // this is a bit of a hack, but deals with use of
3091 // single adj for multiple calls
3092 po->operand[1].val -= adj;
3093 po->flags |= OPF_RMD;
3094 if (po->operand[1].val == 0)
3095 po->flags |= OPF_DONE;
3096 ferr_assert(po, (int)po->operand[1].val >= 0);
3097}
3098
3099// scan for positive, constant esp adjust
3100// multipath case is preliminary
3101static int scan_for_esp_adjust(int i, int opcnt,
3102 int adj_expect, int *adj, int *is_multipath, int do_update)
3103{
3104 int adj_expect_unknown = 0;
3105 struct parsed_op *po;
3106 int first_pop = -1;
3107 int adj_best = 0;
3108
3109 *adj = *is_multipath = 0;
3110 if (adj_expect < 0) {
3111 adj_expect_unknown = 1;
3112 adj_expect = 32 * 4; // enough?
3113 }
3114
3115 for (; i < opcnt && *adj < adj_expect; i++) {
3116 if (g_labels[i] != NULL)
3117 *is_multipath = 1;
3118
3119 po = &ops[i];
3120 if (po->flags & OPF_DONE)
3121 continue;
3122
3123 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
3124 if (po->operand[1].type != OPT_CONST)
3125 ferr(&ops[i], "non-const esp adjust?\n");
3126 *adj += po->operand[1].val;
3127 if (*adj & 3)
3128 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
3129 if (do_update) {
3130 if (!*is_multipath)
3131 patch_esp_adjust(po, adj_expect);
3132 else
3133 po->flags |= OPF_RMD;
3134 }
3135 return i;
3136 }
3137 else if (po->op == OP_PUSH) {
3138 //if (first_pop == -1)
3139 // first_pop = -2; // none
3140 *adj -= lmod_bytes(po, po->operand[0].lmod);
3141 }
3142 else if (po->op == OP_POP) {
3143 if (!(po->flags & OPF_DONE)) {
3144 // seems like msvc only uses 'pop ecx' for stack realignment..
3145 if (po->operand[0].type != OPT_REG || po->operand[0].reg != xCX)
3146 break;
3147 if (first_pop == -1 && *adj >= 0)
3148 first_pop = i;
3149 }
3150 if (do_update && *adj >= 0) {
3151 po->flags |= OPF_RMD;
3152 if (!*is_multipath)
3153 po->flags |= OPF_DONE | OPF_NOREGS;
3154 }
3155
3156 *adj += lmod_bytes(po, po->operand[0].lmod);
3157 if (*adj > adj_best)
3158 adj_best = *adj;
3159 }
3160 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
3161 if (po->op == OP_JMP && po->btj == NULL) {
3162 if (po->bt_i <= i)
3163 break;
3164 i = po->bt_i - 1;
3165 continue;
3166 }
3167 if (po->op != OP_CALL)
3168 break;
3169 if (po->operand[0].type != OPT_LABEL)
3170 break;
3171 if (po->pp != NULL && po->pp->is_stdcall)
3172 break;
3173 if (adj_expect_unknown && first_pop >= 0)
3174 break;
3175 // assume it's another cdecl call
3176 }
3177 }
3178
3179 if (first_pop >= 0) {
3180 // probably only 'pop ecx' was used
3181 *adj = adj_best;
3182 return first_pop;
3183 }
3184
3185 return -1;
3186}
3187
3188static void scan_fwd_set_flags(int i, int opcnt, int magic, int flags)
3189{
3190 struct parsed_op *po;
3191 int j;
3192
3193 if (i < 0)
3194 ferr(ops, "%s: followed bad branch?\n", __func__);
3195
3196 for (; i < opcnt; i++) {
3197 po = &ops[i];
3198 if (po->cc_scratch == magic)
3199 return;
3200 po->cc_scratch = magic;
3201 po->flags |= flags;
3202
3203 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
3204 if (po->btj != NULL) {
3205 // jumptable
3206 for (j = 0; j < po->btj->count; j++)
3207 scan_fwd_set_flags(po->btj->d[j].bt_i, opcnt, magic, flags);
3208 return;
3209 }
3210
3211 scan_fwd_set_flags(po->bt_i, opcnt, magic, flags);
3212 if (!(po->flags & OPF_CJMP))
3213 return;
3214 }
3215 if (po->flags & OPF_TAIL)
3216 return;
3217 }
3218}
3219
3220static const struct parsed_proto *try_recover_pp(
3221 struct parsed_op *po, const struct parsed_opr *opr, int *search_instead)
3222{
3223 const struct parsed_proto *pp = NULL;
3224 char buf[256];
3225 char *p;
3226
3227 // maybe an arg of g_func?
3228 if (opr->type == OPT_REGMEM && is_stack_access(po, opr))
3229 {
3230 char ofs_reg[16] = { 0, };
3231 int arg, arg_s, arg_i;
3232 int stack_ra = 0;
3233 int offset = 0;
3234
3235 if (g_header_mode)
3236 return NULL;
3237
3238 parse_stack_access(po, opr->name, ofs_reg,
3239 &offset, &stack_ra, NULL, 0);
3240 if (ofs_reg[0] != 0)
3241 ferr(po, "offset reg on arg access?\n");
3242 if (offset <= stack_ra) {
3243 // search who set the stack var instead
3244 if (search_instead != NULL)
3245 *search_instead = 1;
3246 return NULL;
3247 }
3248
3249 arg_i = (offset - stack_ra - 4) / 4;
3250 for (arg = arg_s = 0; arg < g_func_pp->argc; arg++) {
3251 if (g_func_pp->arg[arg].reg != NULL)
3252 continue;
3253 if (arg_s == arg_i)
3254 break;
3255 arg_s++;
3256 }
3257 if (arg == g_func_pp->argc)
3258 ferr(po, "stack arg %d not in prototype?\n", arg_i);
3259
3260 pp = g_func_pp->arg[arg].fptr;
3261 if (pp == NULL)
3262 ferr(po, "icall sa: arg%d is not a fptr?\n", arg + 1);
3263 check_func_pp(po, pp, "icall arg");
3264 }
3265 else if (opr->type == OPT_REGMEM && strchr(opr->name + 1, '[')) {
3266 // label[index]
3267 p = strchr(opr->name + 1, '[');
3268 memcpy(buf, opr->name, p - opr->name);
3269 buf[p - opr->name] = 0;
3270 pp = proto_parse(g_fhdr, buf, g_quiet_pp);
3271 }
3272 else if (opr->type == OPT_OFFSET || opr->type == OPT_LABEL) {
3273 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
3274 if (pp == NULL) {
3275 if (!g_header_mode)
3276 ferr(po, "proto_parse failed for icall to '%s'\n", opr->name);
3277 }
3278 else
3279 check_func_pp(po, pp, "reg-fptr ref");
3280 }
3281
3282 return pp;
3283}
3284
3285static void scan_for_call_type(int i, const struct parsed_opr *opr,
3286 int magic, const struct parsed_proto **pp_found, int *pp_i,
3287 int *multi)
3288{
3289 const struct parsed_proto *pp = NULL;
3290 struct parsed_op *po;
3291 struct label_ref *lr;
3292
3293 ops[i].cc_scratch = magic;
3294
3295 while (1) {
3296 if (g_labels[i] != NULL) {
3297 lr = &g_label_refs[i];
3298 for (; lr != NULL; lr = lr->next) {
3299 check_i(&ops[i], lr->i);
3300 scan_for_call_type(lr->i, opr, magic, pp_found, pp_i, multi);
3301 }
3302 if (i > 0 && LAST_OP(i - 1))
3303 return;
3304 }
3305
3306 i--;
3307 if (i < 0)
3308 break;
3309
3310 if (ops[i].cc_scratch == magic)
3311 return;
3312 ops[i].cc_scratch = magic;
3313
3314 if (!(ops[i].flags & OPF_DATA))
3315 continue;
3316 if (!is_opr_modified(opr, &ops[i]))
3317 continue;
3318 if (ops[i].op != OP_MOV && ops[i].op != OP_LEA) {
3319 // most probably trashed by some processing
3320 *pp_found = NULL;
3321 return;
3322 }
3323
3324 opr = &ops[i].operand[1];
3325 if (opr->type != OPT_REG)
3326 break;
3327 }
3328
3329 po = (i >= 0) ? &ops[i] : ops;
3330
3331 if (i < 0) {
3332 // reached the top - can only be an arg-reg
3333 if (opr->type != OPT_REG || g_func_pp == NULL)
3334 return;
3335
3336 for (i = 0; i < g_func_pp->argc; i++) {
3337 if (g_func_pp->arg[i].reg == NULL)
3338 continue;
3339 if (IS(opr->name, g_func_pp->arg[i].reg))
3340 break;
3341 }
3342 if (i == g_func_pp->argc)
3343 return;
3344 pp = g_func_pp->arg[i].fptr;
3345 if (pp == NULL)
3346 ferr(po, "icall: arg%d (%s) is not a fptr?\n",
3347 i + 1, g_func_pp->arg[i].reg);
3348 check_func_pp(po, pp, "icall reg-arg");
3349 }
3350 else
3351 pp = try_recover_pp(po, opr, NULL);
3352
3353 if (*pp_found != NULL && pp != NULL && *pp_found != pp) {
3354 if (!IS((*pp_found)->ret_type.name, pp->ret_type.name)
3355 || (*pp_found)->is_stdcall != pp->is_stdcall
3356 || (*pp_found)->is_fptr != pp->is_fptr
3357 || (*pp_found)->argc != pp->argc
3358 || (*pp_found)->argc_reg != pp->argc_reg
3359 || (*pp_found)->argc_stack != pp->argc_stack)
3360 {
3361 ferr(po, "icall: parsed_proto mismatch\n");
3362 }
3363 *multi = 1;
3364 }
3365 if (pp != NULL) {
3366 *pp_found = pp;
3367 *pp_i = po - ops;
3368 }
3369}
3370
3371static void add_label_ref(struct label_ref *lr, int op_i)
3372{
3373 struct label_ref *lr_new;
3374
3375 if (lr->i == -1) {
3376 lr->i = op_i;
3377 return;
3378 }
3379
3380 lr_new = calloc(1, sizeof(*lr_new));
3381 lr_new->i = op_i;
3382 lr_new->next = lr->next;
3383 lr->next = lr_new;
3384}
3385
3386static struct parsed_data *try_resolve_jumptab(int i, int opcnt)
3387{
3388 struct parsed_op *po = &ops[i];
3389 struct parsed_data *pd;
3390 char label[NAMELEN], *p;
3391 int len, j, l;
3392
3393 p = strchr(po->operand[0].name, '[');
3394 if (p == NULL)
3395 return NULL;
3396
3397 len = p - po->operand[0].name;
3398 strncpy(label, po->operand[0].name, len);
3399 label[len] = 0;
3400
3401 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
3402 if (IS(g_func_pd[j].label, label)) {
3403 pd = &g_func_pd[j];
3404 break;
3405 }
3406 }
3407 if (pd == NULL)
3408 //ferr(po, "label '%s' not parsed?\n", label);
3409 return NULL;
3410
3411 if (pd->type != OPT_OFFSET)
3412 ferr(po, "label '%s' with non-offset data?\n", label);
3413
3414 // find all labels, link
3415 for (j = 0; j < pd->count; j++) {
3416 for (l = 0; l < opcnt; l++) {
3417 if (g_labels[l] != NULL && IS(g_labels[l], pd->d[j].u.label)) {
3418 add_label_ref(&g_label_refs[l], i);
3419 pd->d[j].bt_i = l;
3420 break;
3421 }
3422 }
3423 }
3424
3425 return pd;
3426}
3427
3428static void clear_labels(int count)
3429{
3430 int i;
3431
3432 for (i = 0; i < count; i++) {
3433 if (g_labels[i] != NULL) {
3434 free(g_labels[i]);
3435 g_labels[i] = NULL;
3436 }
3437 }
3438}
3439
3440static int get_pp_arg_regmask_src(const struct parsed_proto *pp)
3441{
3442 int regmask = 0;
3443 int i, reg;
3444
3445 for (i = 0; i < pp->argc; i++) {
3446 if (pp->arg[i].reg != NULL) {
3447 reg = char_array_i(regs_r32,
3448 ARRAY_SIZE(regs_r32), pp->arg[i].reg);
3449 if (reg < 0)
3450 ferr(ops, "arg '%s' of func '%s' is not a reg?\n",
3451 pp->arg[i].reg, pp->name);
3452 regmask |= 1 << reg;
3453 }
3454 }
3455
3456 return regmask;
3457}
3458
3459static int get_pp_arg_regmask_dst(const struct parsed_proto *pp)
3460{
3461 if (strstr(pp->ret_type.name, "int64"))
3462 return (1 << xAX) | (1 << xDX);
3463 if (IS(pp->ret_type.name, "float")
3464 || IS(pp->ret_type.name, "double"))
3465 {
3466 return mxST0;
3467 }
3468 if (strcasecmp(pp->ret_type.name, "void") == 0)
3469 return 0;
3470
3471 return mxAX;
3472}
3473
3474static void resolve_branches_parse_calls(int opcnt)
3475{
3476 static const struct {
3477 const char *name;
3478 enum op_op op;
3479 unsigned int flags;
3480 unsigned int regmask_src;
3481 unsigned int regmask_dst;
3482 } pseudo_ops[] = {
3483 { "__ftol", OPP_FTOL, OPF_FPOP, mxST0, mxAX | mxDX },
3484 };
3485 const struct parsed_proto *pp_c;
3486 struct parsed_proto *pp;
3487 struct parsed_data *pd;
3488 struct parsed_op *po;
3489 const char *tmpname;
3490 int i, l;
3491 int ret;
3492
3493 for (i = 0; i < opcnt; i++)
3494 {
3495 po = &ops[i];
3496 po->bt_i = -1;
3497 po->btj = NULL;
3498
3499 if (po->datap != NULL) {
3500 pp = calloc(1, sizeof(*pp));
3501 my_assert_not(pp, NULL);
3502
3503 ret = parse_protostr(po->datap, pp);
3504 if (ret < 0)
3505 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
3506 free(po->datap);
3507 po->datap = NULL;
3508 po->pp = pp;
3509 }
3510
3511 if (po->op == OP_CALL) {
3512 pp = NULL;
3513
3514 if (po->pp != NULL)
3515 pp = po->pp;
3516 else if (po->operand[0].type == OPT_LABEL)
3517 {
3518 tmpname = opr_name(po, 0);
3519 if (IS_START(tmpname, "loc_"))
3520 ferr(po, "call to loc_*\n");
3521
3522 // convert some calls to pseudo-ops
3523 for (l = 0; l < ARRAY_SIZE(pseudo_ops); l++) {
3524 if (!IS(tmpname, pseudo_ops[l].name))
3525 continue;
3526
3527 po->op = pseudo_ops[l].op;
3528 po->operand_cnt = 0;
3529 po->regmask_src = pseudo_ops[l].regmask_src;
3530 po->regmask_dst = pseudo_ops[l].regmask_dst;
3531 po->flags = pseudo_ops[l].flags;
3532 po->flags |= po->regmask_dst ? OPF_DATA : 0;
3533 break;
3534 }
3535 if (l < ARRAY_SIZE(pseudo_ops))
3536 continue;
3537
3538 pp_c = proto_parse(g_fhdr, tmpname, g_header_mode);
3539 if (!g_header_mode && pp_c == NULL)
3540 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
3541
3542 if (pp_c != NULL) {
3543 pp = proto_clone(pp_c);
3544 my_assert_not(pp, NULL);
3545 }
3546 }
3547
3548 if (pp != NULL) {
3549 if (pp->is_fptr)
3550 check_func_pp(po, pp, "fptr var call");
3551 if (pp->is_noreturn)
3552 po->flags |= OPF_TAIL;
3553 }
3554 po->pp = pp;
3555 continue;
3556 }
3557
3558 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
3559 continue;
3560
3561 if (po->operand[0].type == OPT_REGMEM) {
3562 pd = try_resolve_jumptab(i, opcnt);
3563 if (pd == NULL)
3564 goto tailcall;
3565
3566 po->btj = pd;
3567 continue;
3568 }
3569
3570 for (l = 0; l < opcnt; l++) {
3571 if (g_labels[l] != NULL
3572 && IS(po->operand[0].name, g_labels[l]))
3573 {
3574 if (l == i + 1 && po->op == OP_JMP) {
3575 // yet another alignment type..
3576 po->flags |= OPF_RMD|OPF_DONE;
3577 break;
3578 }
3579 add_label_ref(&g_label_refs[l], i);
3580 po->bt_i = l;
3581 break;
3582 }
3583 }
3584
3585 if (po->bt_i != -1 || (po->flags & OPF_RMD))
3586 continue;
3587
3588 if (po->operand[0].type == OPT_LABEL)
3589 // assume tail call
3590 goto tailcall;
3591
3592 ferr(po, "unhandled branch\n");
3593
3594tailcall:
3595 po->op = OP_CALL;
3596 po->flags |= OPF_TAIL;
3597 if (i > 0 && ops[i - 1].op == OP_POP)
3598 po->flags |= OPF_ATAIL;
3599 i--; // reprocess
3600 }
3601}
3602
3603static void scan_prologue_epilogue(int opcnt)
3604{
3605 int ecx_push = 0, esp_sub = 0;
3606 int found;
3607 int i, j, l;
3608
3609 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
3610 && ops[1].op == OP_MOV
3611 && IS(opr_name(&ops[1], 0), "ebp")
3612 && IS(opr_name(&ops[1], 1), "esp"))
3613 {
3614 g_bp_frame = 1;
3615 ops[0].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3616 ops[1].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3617 i = 2;
3618
3619 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
3620 g_stack_fsz = opr_const(&ops[2], 1);
3621 ops[2].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3622 i++;
3623 }
3624 else {
3625 // another way msvc builds stack frame..
3626 i = 2;
3627 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
3628 g_stack_fsz += 4;
3629 ops[i].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3630 ecx_push++;
3631 i++;
3632 }
3633 // and another way..
3634 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
3635 && ops[i].operand[1].type == OPT_CONST
3636 && ops[i + 1].op == OP_CALL
3637 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
3638 {
3639 g_stack_fsz += ops[i].operand[1].val;
3640 ops[i].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3641 i++;
3642 ops[i].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3643 i++;
3644 }
3645 }
3646
3647 found = 0;
3648 do {
3649 for (; i < opcnt; i++)
3650 if (ops[i].flags & OPF_TAIL)
3651 break;
3652 j = i - 1;
3653 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
3654 if (ops[j].bt_i != -1 || ops[j].btj != NULL)
3655 break;
3656 i--;
3657 j--;
3658 }
3659
3660 if ((ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ebp"))
3661 || ops[j].op == OP_LEAVE)
3662 {
3663 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3664 }
3665 else if (ops[i].op == OP_CALL && ops[i].pp != NULL
3666 && ops[i].pp->is_noreturn)
3667 {
3668 // on noreturn, msvc sometimes cleans stack, sometimes not
3669 i++;
3670 found = 1;
3671 continue;
3672 }
3673 else if (!(g_ida_func_attr & IDAFA_NORETURN))
3674 ferr(&ops[j], "'pop ebp' expected\n");
3675
3676 if (g_stack_fsz != 0) {
3677 if (ops[j].op == OP_LEAVE)
3678 j--;
3679 else if (ops[j].op == OP_POP
3680 && ops[j - 1].op == OP_MOV
3681 && IS(opr_name(&ops[j - 1], 0), "esp")
3682 && IS(opr_name(&ops[j - 1], 1), "ebp"))
3683 {
3684 ops[j - 1].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3685 j -= 2;
3686 }
3687 else if (!(g_ida_func_attr & IDAFA_NORETURN))
3688 {
3689 ferr(&ops[j], "esp restore expected\n");
3690 }
3691
3692 if (ecx_push && j >= 0 && ops[j].op == OP_POP
3693 && IS(opr_name(&ops[j], 0), "ecx"))
3694 {
3695 ferr(&ops[j], "unexpected ecx pop\n");
3696 }
3697 }
3698
3699 found = 1;
3700 i++;
3701 } while (i < opcnt);
3702
3703 if (!found)
3704 ferr(ops, "missing ebp epilogue\n");
3705 return;
3706 }
3707
3708 // non-bp frame
3709 i = 0;
3710 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
3711 ops[i].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3712 g_stack_fsz += 4;
3713 ecx_push++;
3714 i++;
3715 }
3716
3717 for (; i < opcnt; i++) {
3718 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
3719 break;
3720 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
3721 && ops[i].operand[1].type == OPT_CONST)
3722 {
3723 g_stack_fsz = ops[i].operand[1].val;
3724 ops[i].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3725 esp_sub = 1;
3726 break;
3727 }
3728 }
3729
3730 if (ecx_push && !esp_sub) {
3731 // could actually be args for a call..
3732 for (; i < opcnt; i++)
3733 if (ops[i].op != OP_PUSH)
3734 break;
3735
3736 if (ops[i].op == OP_CALL && ops[i].operand[0].type == OPT_LABEL) {
3737 const struct parsed_proto *pp;
3738 pp = proto_parse(g_fhdr, opr_name(&ops[i], 0), 1);
3739 j = pp ? pp->argc_stack : 0;
3740 while (i > 0 && j > 0) {
3741 i--;
3742 if (ops[i].op == OP_PUSH) {
3743 ops[i].flags &= ~(OPF_RMD | OPF_DONE | OPF_NOREGS);
3744 j--;
3745 }
3746 }
3747 if (j != 0)
3748 ferr(&ops[i], "unhandled prologue\n");
3749
3750 // recheck
3751 i = g_stack_fsz = ecx_push = 0;
3752 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
3753 if (!(ops[i].flags & OPF_RMD))
3754 break;
3755 g_stack_fsz += 4;
3756 ecx_push++;
3757 i++;
3758 }
3759 }
3760 }
3761
3762 found = 0;
3763 if (ecx_push || esp_sub)
3764 {
3765 g_sp_frame = 1;
3766
3767 i++;
3768 do {
3769 for (; i < opcnt; i++)
3770 if (ops[i].flags & OPF_TAIL)
3771 break;
3772 j = i - 1;
3773 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
3774 if (ops[j].bt_i != -1 || ops[j].btj != NULL)
3775 break;
3776 i--;
3777 j--;
3778 }
3779
3780 if (ecx_push > 0) {
3781 for (l = 0; l < ecx_push; l++) {
3782 if (ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ecx"))
3783 /* pop ecx */;
3784 else if (ops[j].op == OP_ADD
3785 && IS(opr_name(&ops[j], 0), "esp")
3786 && ops[j].operand[1].type == OPT_CONST)
3787 {
3788 /* add esp, N */
3789 l += ops[j].operand[1].val / 4 - 1;
3790 }
3791 else
3792 ferr(&ops[j], "'pop ecx' expected\n");
3793
3794 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3795 j--;
3796 }
3797 if (l != ecx_push)
3798 ferr(&ops[j], "epilogue scan failed\n");
3799
3800 found = 1;
3801 }
3802
3803 if (esp_sub) {
3804 if (ops[j].op != OP_ADD
3805 || !IS(opr_name(&ops[j], 0), "esp")
3806 || ops[j].operand[1].type != OPT_CONST
3807 || ops[j].operand[1].val != g_stack_fsz)
3808 ferr(&ops[j], "'add esp' expected\n");
3809
3810 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_NOREGS;
3811 ops[j].operand[1].val = 0; // hack for stack arg scanner
3812 found = 1;
3813 }
3814
3815 i++;
3816 } while (i < opcnt);
3817
3818 if (!found)
3819 ferr(ops, "missing esp epilogue\n");
3820 }
3821}
3822
3823// find an instruction that changed opr before i op
3824// *op_i must be set to -1 by the caller
3825// *is_caller is set to 1 if one source is determined to be g_func arg
3826// returns 1 if found, *op_i is then set to origin
3827// returns -1 if multiple origins are found
3828static int resolve_origin(int i, const struct parsed_opr *opr,
3829 int magic, int *op_i, int *is_caller)
3830{
3831 struct label_ref *lr;
3832 int ret = 0;
3833
3834 if (ops[i].cc_scratch == magic)
3835 return 0;
3836 ops[i].cc_scratch = magic;
3837
3838 while (1) {
3839 if (g_labels[i] != NULL) {
3840 lr = &g_label_refs[i];
3841 for (; lr != NULL; lr = lr->next) {
3842 check_i(&ops[i], lr->i);
3843 ret |= resolve_origin(lr->i, opr, magic, op_i, is_caller);
3844 }
3845 if (i > 0 && LAST_OP(i - 1))
3846 return ret;
3847 }
3848
3849 i--;
3850 if (i < 0) {
3851 if (is_caller != NULL)
3852 *is_caller = 1;
3853 return -1;
3854 }
3855
3856 if (ops[i].cc_scratch == magic)
3857 return ret;
3858 ops[i].cc_scratch = magic;
3859
3860 if (!(ops[i].flags & OPF_DATA))
3861 continue;
3862 if (!is_opr_modified(opr, &ops[i]))
3863 continue;
3864
3865 if (*op_i >= 0) {
3866 if (*op_i == i)
3867 return ret | 1;
3868
3869 // XXX: could check if the other op does the same
3870 return -1;
3871 }
3872
3873 *op_i = i;
3874 return ret | 1;
3875 }
3876}
3877
3878// find an instruction that previously referenced opr
3879// if multiple results are found - fail
3880// *op_i must be set to -1 by the caller
3881// returns 1 if found, *op_i is then set to referencer insn
3882static int resolve_last_ref(int i, const struct parsed_opr *opr,
3883 int magic, int *op_i)
3884{
3885 struct label_ref *lr;
3886 int ret = 0;
3887
3888 if (ops[i].cc_scratch == magic)
3889 return 0;
3890 ops[i].cc_scratch = magic;
3891
3892 while (1) {
3893 if (g_labels[i] != NULL) {
3894 lr = &g_label_refs[i];
3895 for (; lr != NULL; lr = lr->next) {
3896 check_i(&ops[i], lr->i);
3897 ret |= resolve_last_ref(lr->i, opr, magic, op_i);
3898 }
3899 if (i > 0 && LAST_OP(i - 1))
3900 return ret;
3901 }
3902
3903 i--;
3904 if (i < 0)
3905 return -1;
3906
3907 if (ops[i].cc_scratch == magic)
3908 return 0;
3909 ops[i].cc_scratch = magic;
3910
3911 if (!is_opr_referenced(opr, &ops[i]))
3912 continue;
3913
3914 if (*op_i >= 0)
3915 return -1;
3916
3917 *op_i = i;
3918 return 1;
3919 }
3920}
3921
3922// find next instruction that reads opr
3923// *op_i must be set to -1 by the caller
3924// on return, *op_i is set to first referencer insn
3925// returns 1 if exactly 1 referencer is found
3926static int find_next_read(int i, int opcnt,
3927 const struct parsed_opr *opr, int magic, int *op_i)
3928{
3929 struct parsed_op *po;
3930 int j, ret = 0;
3931
3932 for (; i < opcnt; i++)
3933 {
3934 if (ops[i].cc_scratch == magic)
3935 return ret;
3936 ops[i].cc_scratch = magic;
3937
3938 po = &ops[i];
3939 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
3940 if (po->btj != NULL) {
3941 // jumptable
3942 for (j = 0; j < po->btj->count; j++) {
3943 check_i(po, po->btj->d[j].bt_i);
3944 ret |= find_next_read(po->btj->d[j].bt_i, opcnt, opr,
3945 magic, op_i);
3946 }
3947 return ret;
3948 }
3949
3950 if (po->flags & OPF_RMD)
3951 continue;
3952 check_i(po, po->bt_i);
3953 if (po->flags & OPF_CJMP) {
3954 ret |= find_next_read(po->bt_i, opcnt, opr, magic, op_i);
3955 if (ret < 0)
3956 return ret;
3957 }
3958 else
3959 i = po->bt_i - 1;
3960 continue;
3961 }
3962
3963 if (!is_opr_read(opr, po)) {
3964 if (is_opr_modified(opr, po)
3965 && (po->op == OP_CALL
3966 || ((po->flags & OPF_DATA)
3967 && po->operand[0].lmod == OPLM_DWORD)))
3968 {
3969 // it's overwritten
3970 return ret;
3971 }
3972 if (po->flags & OPF_TAIL)
3973 return ret;
3974 continue;
3975 }
3976
3977 if (*op_i >= 0)
3978 return -1;
3979
3980 *op_i = i;
3981 return 1;
3982 }
3983
3984 return 0;
3985}
3986
3987static int try_resolve_const(int i, const struct parsed_opr *opr,
3988 int magic, unsigned int *val)
3989{
3990 int s_i = -1;
3991 int ret;
3992
3993 ret = resolve_origin(i, opr, magic, &s_i, NULL);
3994 if (ret == 1) {
3995 i = s_i;
3996 if (ops[i].op != OP_MOV && ops[i].operand[1].type != OPT_CONST)
3997 return -1;
3998
3999 *val = ops[i].operand[1].val;
4000 return 1;
4001 }
4002
4003 return -1;
4004}
4005
4006static const struct parsed_proto *resolve_icall(int i, int opcnt,
4007 int *pp_i, int *multi_src)
4008{
4009 const struct parsed_proto *pp = NULL;
4010 int search_advice = 0;
4011 int offset = -1;
4012 char name[256];
4013 char s_reg[4];
4014 int reg, len;
4015 int ret;
4016
4017 *multi_src = 0;
4018 *pp_i = -1;
4019
4020 switch (ops[i].operand[0].type) {
4021 case OPT_REGMEM:
4022 // try to resolve struct member calls
4023 ret = sscanf(ops[i].operand[0].name, "%3s+%x%n",
4024 s_reg, &offset, &len);
4025 if (ret == 2 && len == strlen(ops[i].operand[0].name))
4026 {
4027 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s_reg);
4028 if (reg >= 0) {
4029 struct parsed_opr opr = OPR_INIT(OPT_REG, OPLM_DWORD, reg);
4030 int j = -1;
4031 ret = resolve_origin(i, &opr, i + opcnt * 19, &j, NULL);
4032 if (ret != 1)
4033 break;
4034 if (ops[j].op == OP_MOV && ops[j].operand[1].type == OPT_REGMEM
4035 && ops[j].operand[0].lmod == OPLM_DWORD
4036 && ops[j].pp == NULL) // no hint
4037 {
4038 // allow one simple dereference (directx)
4039 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32),
4040 ops[j].operand[1].name);
4041 if (reg < 0)
4042 break;
4043 struct parsed_opr opr2 = OPR_INIT(OPT_REG, OPLM_DWORD, reg);
4044 int k = -1;
4045 ret = resolve_origin(j, &opr2, j + opcnt * 19, &k, NULL);
4046 if (ret != 1)
4047 break;
4048 j = k;
4049 }
4050 if (ops[j].op != OP_MOV)
4051 break;
4052 if (ops[j].operand[0].lmod != OPLM_DWORD)
4053 break;
4054 if (ops[j].pp != NULL) {
4055 // type hint in asm
4056 pp = ops[j].pp;
4057 }
4058 else if (ops[j].operand[1].type == OPT_REGMEM) {
4059 // allow 'hello[ecx]' - assume array of same type items
4060 ret = sscanf(ops[j].operand[1].name, "%[^[][e%2s]",
4061 name, s_reg);
4062 if (ret != 2)
4063 break;
4064 pp = proto_parse(g_fhdr, name, g_quiet_pp);
4065 }
4066 else if (ops[j].operand[1].type == OPT_LABEL)
4067 pp = proto_parse(g_fhdr, ops[j].operand[1].name, g_quiet_pp);
4068 else
4069 break;
4070 if (pp == NULL)
4071 break;
4072 if (pp->is_func || pp->is_fptr || !pp->type.is_struct) {
4073 pp = NULL;
4074 break;
4075 }
4076 pp = proto_lookup_struct(g_fhdr, pp->type.name, offset);
4077 }
4078 break;
4079 }
4080 // fallthrough
4081 case OPT_LABEL:
4082 case OPT_OFFSET:
4083 pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice);
4084 if (!search_advice)
4085 break;
4086 // fallthrough
4087 default:
4088 scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp,
4089 pp_i, multi_src);
4090 break;
4091 }
4092
4093 return pp;
4094}
4095
4096static struct parsed_proto *process_call_early(int i, int opcnt,
4097 int *adj_i)
4098{
4099 struct parsed_op *po = &ops[i];
4100 struct parsed_proto *pp;
4101 int multipath = 0;
4102 int adj = 0;
4103 int j, ret;
4104
4105 pp = po->pp;
4106 if (pp == NULL || pp->is_vararg || pp->argc_reg != 0)
4107 // leave for later
4108 return NULL;
4109
4110 // look for and make use of esp adjust
4111 *adj_i = ret = -1;
4112 if (!pp->is_stdcall && pp->argc_stack > 0)
4113 ret = scan_for_esp_adjust(i + 1, opcnt,
4114 pp->argc_stack * 4, &adj, &multipath, 0);
4115 if (ret >= 0) {
4116 if (pp->argc_stack > adj / 4)
4117 return NULL;
4118 if (multipath)
4119 return NULL;
4120 if (ops[ret].op == OP_POP) {
4121 for (j = 1; j < adj / 4; j++) {
4122 if (ops[ret + j].op != OP_POP
4123 || ops[ret + j].operand[0].reg != xCX)
4124 {
4125 return NULL;
4126 }
4127 }
4128 }
4129 }
4130
4131 *adj_i = ret;
4132 return pp;
4133}
4134
4135static struct parsed_proto *process_call(int i, int opcnt)
4136{
4137 struct parsed_op *po = &ops[i];
4138 const struct parsed_proto *pp_c;
4139 struct parsed_proto *pp;
4140 const char *tmpname;
4141 int call_i = -1, ref_i = -1;
4142 int adj = 0, multipath = 0;
4143 int ret, arg;
4144
4145 tmpname = opr_name(po, 0);
4146 pp = po->pp;
4147 if (pp == NULL)
4148 {
4149 // indirect call
4150 pp_c = resolve_icall(i, opcnt, &call_i, &multipath);
4151 if (pp_c != NULL) {
4152 if (!pp_c->is_func && !pp_c->is_fptr)
4153 ferr(po, "call to non-func: %s\n", pp_c->name);
4154 pp = proto_clone(pp_c);
4155 my_assert_not(pp, NULL);
4156 if (multipath)
4157 // not resolved just to single func
4158 pp->is_fptr = 1;
4159
4160 switch (po->operand[0].type) {
4161 case OPT_REG:
4162 // we resolved this call and no longer need the register
4163 po->regmask_src &= ~(1 << po->operand[0].reg);
4164
4165 if (!multipath && i != call_i && ops[call_i].op == OP_MOV
4166 && ops[call_i].operand[1].type == OPT_LABEL)
4167 {
4168 // no other source users?
4169 ret = resolve_last_ref(i, &po->operand[0], i + opcnt * 10,
4170 &ref_i);
4171 if (ret == 1 && call_i == ref_i) {
4172 // and nothing uses it after us?
4173 ref_i = -1;
4174 find_next_read(i + 1, opcnt, &po->operand[0],
4175 i + opcnt * 11, &ref_i);
4176 if (ref_i == -1)
4177 // then also don't need the source mov
4178 ops[call_i].flags |= OPF_RMD | OPF_NOREGS;
4179 }
4180 }
4181 break;
4182 case OPT_REGMEM:
4183 pp->is_fptr = 1;
4184 break;
4185 default:
4186 break;
4187 }
4188 }
4189 if (pp == NULL) {
4190 pp = calloc(1, sizeof(*pp));
4191 my_assert_not(pp, NULL);
4192
4193 pp->is_fptr = 1;
4194 ret = scan_for_esp_adjust(i + 1, opcnt,
4195 -1, &adj, &multipath, 0);
4196 if (ret < 0 || adj < 0) {
4197 if (!g_allow_regfunc)
4198 ferr(po, "non-__cdecl indirect call unhandled yet\n");
4199 pp->is_unresolved = 1;
4200 adj = 0;
4201 }
4202 adj /= 4;
4203 if (adj > ARRAY_SIZE(pp->arg))
4204 ferr(po, "esp adjust too large: %d\n", adj);
4205 pp->ret_type.name = strdup("int");
4206 pp->argc = pp->argc_stack = adj;
4207 for (arg = 0; arg < pp->argc; arg++)
4208 pp->arg[arg].type.name = strdup("int");
4209 }
4210 po->pp = pp;
4211 }
4212
4213 // look for and make use of esp adjust
4214 multipath = 0;
4215 ret = -1;
4216 if (!pp->is_stdcall && pp->argc_stack > 0) {
4217 int adj_expect = pp->is_vararg ? -1 : pp->argc_stack * 4;
4218 ret = scan_for_esp_adjust(i + 1, opcnt,
4219 adj_expect, &adj, &multipath, 0);
4220 }
4221 if (ret >= 0) {
4222 if (pp->is_vararg) {
4223 if (adj / 4 < pp->argc_stack) {
4224 fnote(po, "(this call)\n");
4225 ferr(&ops[ret], "esp adjust is too small: %x < %x\n",
4226 adj, pp->argc_stack * 4);
4227 }
4228 // modify pp to make it have varargs as normal args
4229 arg = pp->argc;
4230 pp->argc += adj / 4 - pp->argc_stack;
4231 for (; arg < pp->argc; arg++) {
4232 pp->arg[arg].type.name = strdup("int");
4233 pp->argc_stack++;
4234 }
4235 if (pp->argc > ARRAY_SIZE(pp->arg))
4236 ferr(po, "too many args for '%s'\n", tmpname);
4237 }
4238 if (pp->argc_stack > adj / 4) {
4239 fnote(po, "(this call)\n");
4240 ferr(&ops[ret], "stack tracking failed for '%s': %x %x\n",
4241 tmpname, pp->argc_stack * 4, adj);
4242 }
4243
4244 scan_for_esp_adjust(i + 1, opcnt,
4245 pp->argc_stack * 4, &adj, &multipath, 1);
4246 }
4247 else if (pp->is_vararg)
4248 ferr(po, "missing esp_adjust for vararg func '%s'\n",
4249 pp->name);
4250
4251 return pp;
4252}
4253
4254static int collect_call_args_early(struct parsed_op *po, int i,
4255 struct parsed_proto *pp, int *regmask)
4256{
4257 int arg, ret;
4258 int j;
4259
4260 for (arg = 0; arg < pp->argc; arg++)
4261 if (pp->arg[arg].reg == NULL)
4262 break;
4263
4264 // first see if it can be easily done
4265 for (j = i; j > 0 && arg < pp->argc; )
4266 {
4267 if (g_labels[j] != NULL)
4268 return -1;
4269 j--;
4270
4271 if (ops[j].op == OP_CALL)
4272 return -1;
4273 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP)
4274 return -1;
4275 else if (ops[j].op == OP_POP)
4276 return -1;
4277 else if (ops[j].flags & OPF_CJMP)
4278 return -1;
4279 else if (ops[j].op == OP_PUSH) {
4280 if (ops[j].flags & (OPF_FARG|OPF_FARGNR))
4281 return -1;
4282 ret = scan_for_mod(&ops[j], j + 1, i, 1);
4283 if (ret >= 0)
4284 return -1;
4285
4286 if (pp->arg[arg].type.is_va_list)
4287 return -1;
4288
4289 // next arg
4290 for (arg++; arg < pp->argc; arg++)
4291 if (pp->arg[arg].reg == NULL)
4292 break;
4293 }
4294 }
4295
4296 if (arg < pp->argc)
4297 return -1;
4298
4299 // now do it
4300 for (arg = 0; arg < pp->argc; arg++)
4301 if (pp->arg[arg].reg == NULL)
4302 break;
4303
4304 for (j = i; j > 0 && arg < pp->argc; )
4305 {
4306 j--;
4307
4308 if (ops[j].op == OP_PUSH)
4309 {
4310 ops[j].p_argnext = -1;
4311 ferr_assert(&ops[j], pp->arg[arg].datap == NULL);
4312 pp->arg[arg].datap = &ops[j];
4313
4314 if (ops[j].operand[0].type == OPT_REG)
4315 *regmask |= 1 << ops[j].operand[0].reg;
4316
4317 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_FARGNR | OPF_FARG;
4318 ops[j].flags &= ~OPF_RSAVE;
4319
4320 // next arg
4321 for (arg++; arg < pp->argc; arg++)
4322 if (pp->arg[arg].reg == NULL)
4323 break;
4324 }
4325 }
4326
4327 return 0;
4328}
4329
4330static int collect_call_args_r(struct parsed_op *po, int i,
4331 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
4332 int *arg_grp, int arg, int magic, int need_op_saving, int may_reuse)
4333{
4334 struct parsed_proto *pp_tmp;
4335 struct parsed_op *po_tmp;
4336 struct label_ref *lr;
4337 int need_to_save_current;
4338 int arg_grp_current = 0;
4339 int save_args_seen = 0;
4340 int save_args;
4341 int ret = 0;
4342 int reg;
4343 char buf[32];
4344 int j, k;
4345
4346 if (i < 0) {
4347 ferr(po, "dead label encountered\n");
4348 return -1;
4349 }
4350
4351 for (; arg < pp->argc; arg++)
4352 if (pp->arg[arg].reg == NULL)
4353 break;
4354 magic = (magic & 0xffffff) | (arg << 24);
4355
4356 for (j = i; j >= 0 && (arg < pp->argc || pp->is_unresolved); )
4357 {
4358 if (((ops[j].cc_scratch ^ magic) & 0xffffff) == 0) {
4359 if (ops[j].cc_scratch != magic) {
4360 ferr(&ops[j], "arg collect hit same path with diff args for %s\n",
4361 pp->name);
4362 return -1;
4363 }
4364 // ok: have already been here
4365 return 0;
4366 }
4367 ops[j].cc_scratch = magic;
4368
4369 if (g_labels[j] != NULL && g_label_refs[j].i != -1) {
4370 lr = &g_label_refs[j];
4371 if (lr->next != NULL)
4372 need_op_saving = 1;
4373 for (; lr->next; lr = lr->next) {
4374 check_i(&ops[j], lr->i);
4375 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
4376 may_reuse = 1;
4377 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
4378 arg_grp, arg, magic, need_op_saving, may_reuse);
4379 if (ret < 0)
4380 return ret;
4381 }
4382
4383 check_i(&ops[j], lr->i);
4384 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
4385 may_reuse = 1;
4386 if (j > 0 && LAST_OP(j - 1)) {
4387 // follow last branch in reverse
4388 j = lr->i;
4389 continue;
4390 }
4391 need_op_saving = 1;
4392 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
4393 arg_grp, arg, magic, need_op_saving, may_reuse);
4394 if (ret < 0)
4395 return ret;
4396 }
4397 j--;
4398
4399 if (ops[j].op == OP_CALL)
4400 {
4401 if (pp->is_unresolved)
4402 break;
4403
4404 pp_tmp = ops[j].pp;
4405 if (pp_tmp == NULL)
4406 ferr(po, "arg collect hit unparsed call '%s'\n",
4407 ops[j].operand[0].name);
4408 if (may_reuse && pp_tmp->argc_stack > 0)
4409 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
4410 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
4411 }
4412 // esp adjust of 0 means we collected it before
4413 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP
4414 && (ops[j].operand[1].type != OPT_CONST
4415 || ops[j].operand[1].val != 0))
4416 {
4417 if (pp->is_unresolved)
4418 break;
4419
4420 fnote(po, "(this call)\n");
4421 ferr(&ops[j], "arg collect %d/%d hit esp adjust of %d\n",
4422 arg, pp->argc, ops[j].operand[1].val);
4423 }
4424 else if (ops[j].op == OP_POP && !(ops[j].flags & OPF_DONE))
4425 {
4426 if (pp->is_unresolved)
4427 break;
4428
4429 fnote(po, "(this call)\n");
4430 ferr(&ops[j], "arg collect %d/%d hit pop\n", arg, pp->argc);
4431 }
4432 else if (ops[j].flags & OPF_CJMP)
4433 {
4434 if (pp->is_unresolved)
4435 break;
4436
4437 may_reuse = 1;
4438 }
4439 else if (ops[j].op == OP_PUSH
4440 && !(ops[j].flags & (OPF_FARGNR|OPF_DONE)))
4441 {
4442 if (pp->is_unresolved && (ops[j].flags & OPF_RMD))
4443 break;
4444
4445 ops[j].p_argnext = -1;
4446 po_tmp = pp->arg[arg].datap;
4447 if (po_tmp != NULL)
4448 ops[j].p_argnext = po_tmp - ops;
4449 pp->arg[arg].datap = &ops[j];
4450
4451 need_to_save_current = 0;
4452 save_args = 0;
4453 reg = -1;
4454 if (ops[j].operand[0].type == OPT_REG)
4455 reg = ops[j].operand[0].reg;
4456
4457 if (!need_op_saving) {
4458 ret = scan_for_mod(&ops[j], j + 1, i, 1);
4459 need_to_save_current = (ret >= 0);
4460 }
4461 if (need_op_saving || need_to_save_current) {
4462 // mark this push as one that needs operand saving
4463 ops[j].flags &= ~OPF_RMD;
4464 if (ops[j].p_argnum == 0) {
4465 ops[j].p_argnum = arg + 1;
4466 save_args |= 1 << arg;
4467 }
4468 else if (ops[j].p_argnum < arg + 1) {
4469 // XXX: might kill valid var..
4470 //*save_arg_vars &= ~(1 << (ops[j].p_argnum - 1));
4471 ops[j].p_argnum = arg + 1;
4472 save_args |= 1 << arg;
4473 }
4474
4475 if (save_args_seen & (1 << (ops[j].p_argnum - 1))) {
4476 save_args_seen = 0;
4477 arg_grp_current++;
4478 if (arg_grp_current >= MAX_ARG_GRP)
4479 ferr(&ops[j], "out of arg groups (arg%d), f %s\n",
4480 ops[j].p_argnum, pp->name);
4481 }
4482 }
4483 else if (ops[j].p_argnum == 0)
4484 ops[j].flags |= OPF_RMD;
4485
4486 // some PUSHes are reused by different calls on other branches,
4487 // but that can't happen if we didn't branch, so they
4488 // can be removed from future searches (handles nested calls)
4489 if (!may_reuse)
4490 ops[j].flags |= OPF_FARGNR;
4491
4492 ops[j].flags |= OPF_FARG;
4493 ops[j].flags &= ~OPF_RSAVE;
4494
4495 // check for __VALIST
4496 if (!pp->is_unresolved && g_func_pp != NULL
4497 && pp->arg[arg].type.is_va_list)
4498 {
4499 k = -1;
4500 ret = resolve_origin(j, &ops[j].operand[0],
4501 magic + 1, &k, NULL);
4502 if (ret == 1 && k >= 0)
4503 {
4504 if (ops[k].op == OP_LEA) {
4505 if (!g_func_pp->is_vararg)
4506 ferr(&ops[k], "lea <arg> used, but %s is not vararg?\n",
4507 g_func_pp->name);
4508
4509 snprintf(buf, sizeof(buf), "arg_%X",
4510 g_func_pp->argc_stack * 4);
4511 if (strstr(ops[k].operand[1].name, buf)
4512 || strstr(ops[k].operand[1].name, "arglist"))
4513 {
4514 ops[k].flags |= OPF_RMD | OPF_NOREGS | OPF_DONE;
4515 ops[j].flags |= OPF_RMD | OPF_NOREGS | OPF_VAPUSH;
4516 save_args &= ~(1 << arg);
4517 reg = -1;
4518 }
4519 else
4520 ferr(&ops[k], "va_list arg detection failed\n");
4521 }
4522 // check for va_list from g_func_pp arg too
4523 else if (ops[k].op == OP_MOV
4524 && is_stack_access(&ops[k], &ops[k].operand[1]))
4525 {
4526 ret = stack_frame_access(&ops[k], &ops[k].operand[1],
4527 buf, sizeof(buf), ops[k].operand[1].name, "", 1, 0);
4528 if (ret >= 0) {
4529 ops[k].flags |= OPF_RMD | OPF_DONE;
4530 ops[j].flags |= OPF_RMD;
4531 ops[j].p_argpass = ret + 1;
4532 save_args &= ~(1 << arg);
4533 reg = -1;
4534 }
4535 }
4536 }
4537 }
4538
4539 *save_arg_vars |= save_args;
4540
4541 // tracking reg usage
4542 if (reg >= 0)
4543 *regmask |= 1 << reg;
4544
4545 arg++;
4546 if (!pp->is_unresolved) {
4547 // next arg
4548 for (; arg < pp->argc; arg++)
4549 if (pp->arg[arg].reg == NULL)
4550 break;
4551 }
4552 magic = (magic & 0xffffff) | (arg << 24);
4553 }
4554
4555 if (ops[j].p_arggrp > arg_grp_current) {
4556 save_args_seen = 0;
4557 arg_grp_current = ops[j].p_arggrp;
4558 }
4559 if (ops[j].p_argnum > 0)
4560 save_args_seen |= 1 << (ops[j].p_argnum - 1);
4561 }
4562
4563 if (arg < pp->argc) {
4564 ferr(po, "arg collect failed for '%s': %d/%d\n",
4565 pp->name, arg, pp->argc);
4566 return -1;
4567 }
4568
4569 if (arg_grp_current > *arg_grp)
4570 *arg_grp = arg_grp_current;
4571
4572 return arg;
4573}
4574
4575static int collect_call_args(struct parsed_op *po, int i,
4576 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
4577 int magic)
4578{
4579 // arg group is for cases when pushes for
4580 // multiple funcs are going on
4581 struct parsed_op *po_tmp;
4582 int save_arg_vars_current = 0;
4583 int arg_grp = 0;
4584 int ret;
4585 int a;
4586
4587 ret = collect_call_args_r(po, i, pp, regmask,
4588 &save_arg_vars_current, &arg_grp, 0, magic, 0, 0);
4589 if (ret < 0)
4590 return ret;
4591
4592 if (arg_grp != 0) {
4593 // propagate arg_grp
4594 for (a = 0; a < pp->argc; a++) {
4595 if (pp->arg[a].reg != NULL)
4596 continue;
4597
4598 po_tmp = pp->arg[a].datap;
4599 while (po_tmp != NULL) {
4600 po_tmp->p_arggrp = arg_grp;
4601 if (po_tmp->p_argnext > 0)
4602 po_tmp = &ops[po_tmp->p_argnext];
4603 else
4604 po_tmp = NULL;
4605 }
4606 }
4607 }
4608 save_arg_vars[arg_grp] |= save_arg_vars_current;
4609
4610 if (pp->is_unresolved) {
4611 pp->argc += ret;
4612 pp->argc_stack += ret;
4613 for (a = 0; a < pp->argc; a++)
4614 if (pp->arg[a].type.name == NULL)
4615 pp->arg[a].type.name = strdup("int");
4616 }
4617
4618 return ret;
4619}
4620
4621static void reg_use_pass(int i, int opcnt, unsigned char *cbits,
4622 int regmask_now, int *regmask,
4623 int regmask_save_now, int *regmask_save,
4624 int *regmask_init, int regmask_arg)
4625{
4626 struct parsed_op *po;
4627 unsigned int mask;
4628 int already_saved;
4629 int regmask_new;
4630 int regmask_op;
4631 int flags_set;
4632 int ret, reg;
4633 int j;
4634
4635 for (; i < opcnt; i++)
4636 {
4637 po = &ops[i];
4638 if (cbits[i >> 3] & (1 << (i & 7)))
4639 return;
4640 cbits[i >> 3] |= (1 << (i & 7));
4641
4642 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
4643 if (po->flags & (OPF_RMD|OPF_DONE))
4644 continue;
4645 if (po->btj != NULL) {
4646 for (j = 0; j < po->btj->count; j++) {
4647 check_i(po, po->btj->d[j].bt_i);
4648 reg_use_pass(po->btj->d[j].bt_i, opcnt, cbits,
4649 regmask_now, regmask, regmask_save_now, regmask_save,
4650 regmask_init, regmask_arg);
4651 }
4652 return;
4653 }
4654
4655 check_i(po, po->bt_i);
4656 if (po->flags & OPF_CJMP)
4657 reg_use_pass(po->bt_i, opcnt, cbits,
4658 regmask_now, regmask, regmask_save_now, regmask_save,
4659 regmask_init, regmask_arg);
4660 else
4661 i = po->bt_i - 1;
4662 continue;
4663 }
4664
4665 if (po->op == OP_PUSH && !(po->flags & (OPF_FARG|OPF_DONE))
4666 && !g_func_pp->is_userstack
4667 && po->operand[0].type == OPT_REG)
4668 {
4669 reg = po->operand[0].reg;
4670 ferr_assert(po, reg >= 0);
4671
4672 already_saved = 0;
4673 flags_set = OPF_RSAVE | OPF_RMD | OPF_DONE;
4674 if (regmask_now & (1 << reg)) {
4675 already_saved = regmask_save_now & (1 << reg);
4676 flags_set = OPF_RSAVE | OPF_DONE;
4677 }
4678
4679 ret = scan_for_pop(i + 1, opcnt, i + opcnt * 3, reg, 0, 0);
4680 if (ret == 1) {
4681 scan_for_pop(i + 1, opcnt, i + opcnt * 4, reg, 0, flags_set);
4682 }
4683 else {
4684 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].reg, 0);
4685 if (ret == 1) {
4686 scan_for_pop_ret(i + 1, opcnt, po->operand[0].reg,
4687 flags_set);
4688 }
4689 }
4690 if (ret == 1) {
4691 ferr_assert(po, !already_saved);
4692 po->flags |= flags_set;
4693
4694 if (regmask_now & (1 << reg)) {
4695 regmask_save_now |= (1 << reg);
4696 *regmask_save |= regmask_save_now;
4697 }
4698 continue;
4699 }
4700 }
4701 else if (po->op == OP_POP && (po->flags & OPF_RSAVE)) {
4702 reg = po->operand[0].reg;
4703 ferr_assert(po, reg >= 0);
4704
4705 if (regmask_save_now & (1 << reg))
4706 regmask_save_now &= ~(1 << reg);
4707 else
4708 regmask_now &= ~(1 << reg);
4709 continue;
4710 }
4711 else if (po->op == OP_CALL) {
4712 if ((po->regmask_dst & (1 << xAX))
4713 && !(po->regmask_dst & (1 << xDX)))
4714 {
4715 if (po->flags & OPF_TAIL)
4716 // don't need eax, will do "return f();" or "f(); return;"
4717 po->regmask_dst &= ~(1 << xAX);
4718 else {
4719 struct parsed_opr opr = OPR_INIT(OPT_REG, OPLM_DWORD, xAX);
4720 j = -1;
4721 find_next_read(i + 1, opcnt, &opr, i + opcnt * 17, &j);
4722 if (j == -1)
4723 // not used
4724 po->regmask_dst &= ~(1 << xAX);
4725 }
4726 }
4727 }
4728
4729 if (po->flags & OPF_NOREGS)
4730 continue;
4731
4732 if (po->flags & OPF_FPUSH) {
4733 if (regmask_now & mxST1)
4734 ferr(po, "TODO: FPUSH on active ST1\n");
4735 if (regmask_now & mxST0)
4736 po->flags |= OPF_FSHIFT;
4737 mask = mxST0 | mxST1;
4738 regmask_now = (regmask_now & ~mask) | ((regmask_now & mxST0) << 1);
4739 }
4740
4741 // if incomplete register is used, clear it on init to avoid
4742 // later use of uninitialized upper part in some situations
4743 if ((po->flags & OPF_DATA) && po->operand[0].type == OPT_REG
4744 && po->operand[0].lmod != OPLM_DWORD)
4745 {
4746 reg = po->operand[0].reg;
4747 ferr_assert(po, reg >= 0);
4748
4749 if (!(regmask_now & (1 << reg)))
4750 *regmask_init |= 1 << reg;
4751 }
4752
4753 regmask_op = po->regmask_src | po->regmask_dst;
4754
4755 regmask_new = po->regmask_src & ~regmask_now & ~regmask_arg;
4756 regmask_new &= ~(1 << xSP);
4757 if (g_bp_frame && !(po->flags & OPF_EBP_S))
4758 regmask_new &= ~(1 << xBP);
4759
4760 if (po->op == OP_CALL) {
4761 // allow fastcall calls from anywhere, calee may be also sitting
4762 // in some fastcall table even when it's not using reg args
4763 if (regmask_new & po->regmask_src & (1 << xCX)) {
4764 *regmask_init |= (1 << xCX);
4765 regmask_now |= (1 << xCX);
4766 regmask_new &= ~(1 << xCX);
4767 }
4768 if (regmask_new & po->regmask_src & (1 << xDX)) {
4769 *regmask_init |= (1 << xDX);
4770 regmask_now |= (1 << xDX);
4771 regmask_new &= ~(1 << xDX);
4772 }
4773 }
4774
4775 if (regmask_new != 0)
4776 fnote(po, "uninitialized reg mask: %x\n", regmask_new);
4777
4778 if (regmask_op & (1 << xBP)) {
4779 if (g_bp_frame && !(po->flags & OPF_EBP_S)) {
4780 if (po->regmask_dst & (1 << xBP))
4781 // compiler decided to drop bp frame and use ebp as scratch
4782 scan_fwd_set_flags(i + 1, opcnt, i + opcnt * 5, OPF_EBP_S);
4783 else
4784 regmask_op &= ~(1 << xBP);
4785 }
4786 }
4787
4788 regmask_now |= regmask_op;
4789 *regmask |= regmask_now;
4790
4791 // released regs
4792 if (po->flags & OPF_FPOP) {
4793 mask = mxST0 | mxST1;
4794 if (!(regmask_now & mask))
4795 ferr(po, "float pop on empty stack?\n");
4796 if (regmask_now & mxST1)
4797 po->flags |= OPF_FSHIFT;
4798 regmask_now = (regmask_now & ~mask) | ((regmask_now & mxST1) >> 1);
4799 }
4800
4801 if (po->flags & OPF_TAIL) {
4802 if (regmask_now & (mxST0 | mxST1))
4803 ferr(po, "float regs on tail: %x\n", regmask_now);
4804 return;
4805 }
4806 }
4807}
4808
4809static void pp_insert_reg_arg(struct parsed_proto *pp, const char *reg)
4810{
4811 int i;
4812
4813 for (i = 0; i < pp->argc; i++)
4814 if (pp->arg[i].reg == NULL)
4815 break;
4816
4817 if (pp->argc_stack)
4818 memmove(&pp->arg[i + 1], &pp->arg[i],
4819 sizeof(pp->arg[0]) * pp->argc_stack);
4820 memset(&pp->arg[i], 0, sizeof(pp->arg[i]));
4821 pp->arg[i].reg = strdup(reg);
4822 pp->arg[i].type.name = strdup("int");
4823 pp->argc++;
4824 pp->argc_reg++;
4825}
4826
4827static void output_std_flags(FILE *fout, struct parsed_op *po,
4828 int *pfomask, const char *dst_opr_text)
4829{
4830 if (*pfomask & (1 << PFO_Z)) {
4831 fprintf(fout, "\n cond_z = (%s%s == 0);",
4832 lmod_cast_u(po, po->operand[0].lmod), dst_opr_text);
4833 *pfomask &= ~(1 << PFO_Z);
4834 }
4835 if (*pfomask & (1 << PFO_S)) {
4836 fprintf(fout, "\n cond_s = (%s%s < 0);",
4837 lmod_cast_s(po, po->operand[0].lmod), dst_opr_text);
4838 *pfomask &= ~(1 << PFO_S);
4839 }
4840}
4841
4842enum {
4843 OPP_FORCE_NORETURN = (1 << 0),
4844 OPP_SIMPLE_ARGS = (1 << 1),
4845 OPP_ALIGN = (1 << 2),
4846};
4847
4848static void output_pp_attrs(FILE *fout, const struct parsed_proto *pp,
4849 int flags)
4850{
4851 const char *cconv = "";
4852
4853 if (pp->is_fastcall)
4854 cconv = "__fastcall ";
4855 else if (pp->is_stdcall && pp->argc_reg == 0)
4856 cconv = "__stdcall ";
4857
4858 fprintf(fout, (flags & OPP_ALIGN) ? "%-16s" : "%s", cconv);
4859
4860 if (pp->is_noreturn || (flags & OPP_FORCE_NORETURN))
4861 fprintf(fout, "noreturn ");
4862}
4863
4864static void output_pp(FILE *fout, const struct parsed_proto *pp,
4865 int flags)
4866{
4867 int i;
4868
4869 fprintf(fout, (flags & OPP_ALIGN) ? "%-5s" : "%s ",
4870 pp->ret_type.name);
4871 if (pp->is_fptr)
4872 fprintf(fout, "(");
4873 output_pp_attrs(fout, pp, flags);
4874 if (pp->is_fptr)
4875 fprintf(fout, "*");
4876 fprintf(fout, "%s", pp->name);
4877 if (pp->is_fptr)
4878 fprintf(fout, ")");
4879
4880 fprintf(fout, "(");
4881 for (i = 0; i < pp->argc; i++) {
4882 if (i > 0)
4883 fprintf(fout, ", ");
4884 if (pp->arg[i].fptr != NULL && !(flags & OPP_SIMPLE_ARGS)) {
4885 // func pointer
4886 output_pp(fout, pp->arg[i].fptr, 0);
4887 }
4888 else if (pp->arg[i].type.is_retreg) {
4889 fprintf(fout, "u32 *r_%s", pp->arg[i].reg);
4890 }
4891 else {
4892 fprintf(fout, "%s", pp->arg[i].type.name);
4893 if (!pp->is_fptr)
4894 fprintf(fout, " a%d", i + 1);
4895 }
4896 }
4897 if (pp->is_vararg) {
4898 if (i > 0)
4899 fprintf(fout, ", ");
4900 fprintf(fout, "...");
4901 }
4902 fprintf(fout, ")");
4903}
4904
4905static char *saved_arg_name(char *buf, size_t buf_size, int grp, int num)
4906{
4907 char buf1[16];
4908
4909 buf1[0] = 0;
4910 if (grp > 0)
4911 snprintf(buf1, sizeof(buf1), "%d", grp);
4912 snprintf(buf, buf_size, "s%s_a%d", buf1, num);
4913
4914 return buf;
4915}
4916
4917static void gen_x_cleanup(int opcnt);
4918
4919static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
4920{
4921 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
4922 struct parsed_opr *last_arith_dst = NULL;
4923 char buf1[256], buf2[256], buf3[256], cast[64];
4924 struct parsed_proto *pp, *pp_tmp;
4925 struct parsed_data *pd;
4926 unsigned int uval;
4927 int save_arg_vars[MAX_ARG_GRP] = { 0, };
4928 unsigned char cbits[MAX_OPS / 8];
4929 int cond_vars = 0;
4930 int need_tmp_var = 0;
4931 int need_tmp64 = 0;
4932 int had_decl = 0;
4933 int label_pending = 0;
4934 int regmask_save = 0; // regs saved/restored in this func
4935 int regmask_arg; // regs from this function args (fastcall, etc)
4936 int regmask_ret; // regs needed on ret
4937 int regmask_now; // temp
4938 int regmask_init = 0; // regs that need zero initialization
4939 int regmask_pp = 0; // regs used in complex push-pop graph
4940 int regmask = 0; // used regs
4941 int pfomask = 0;
4942 int found = 0;
4943 int no_output;
4944 int i, j, l;
4945 int arg;
4946 int reg;
4947 int ret;
4948
4949 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
4950 g_stack_frame_used = 0;
4951
4952 g_func_pp = proto_parse(fhdr, funcn, 0);
4953 if (g_func_pp == NULL)
4954 ferr(ops, "proto_parse failed for '%s'\n", funcn);
4955
4956 regmask_arg = get_pp_arg_regmask_src(g_func_pp);
4957 regmask_ret = get_pp_arg_regmask_dst(g_func_pp);
4958
4959 if (g_func_pp->has_retreg) {
4960 for (arg = 0; arg < g_func_pp->argc; arg++) {
4961 if (g_func_pp->arg[arg].type.is_retreg) {
4962 reg = char_array_i(regs_r32,
4963 ARRAY_SIZE(regs_r32), g_func_pp->arg[arg].reg);
4964 ferr_assert(ops, reg >= 0);
4965 regmask_ret |= 1 << reg;
4966 }
4967 }
4968 }
4969
4970 // pass1:
4971 // - resolve all branches
4972 // - parse calls with labels
4973 resolve_branches_parse_calls(opcnt);
4974
4975 // pass2:
4976 // - handle ebp/esp frame, remove ops related to it
4977 scan_prologue_epilogue(opcnt);
4978
4979 // pass3:
4980 // - remove dead labels
4981 // - set regs needed at ret
4982 for (i = 0; i < opcnt; i++)
4983 {
4984 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
4985 free(g_labels[i]);
4986 g_labels[i] = NULL;
4987 }
4988
4989 if (ops[i].op == OP_RET)
4990 ops[i].regmask_src |= regmask_ret;
4991 }
4992
4993 // pass4:
4994 // - process trivial calls
4995 for (i = 0; i < opcnt; i++)
4996 {
4997 po = &ops[i];
4998 if (po->flags & (OPF_RMD|OPF_DONE))
4999 continue;
5000
5001 if (po->op == OP_CALL)
5002 {
5003 pp = process_call_early(i, opcnt, &j);
5004 if (pp != NULL) {
5005 if (!(po->flags & OPF_ATAIL))
5006 // since we know the args, try to collect them
5007 if (collect_call_args_early(po, i, pp, &regmask) != 0)
5008 pp = NULL;
5009 }
5010
5011 if (pp != NULL) {
5012 if (j >= 0) {
5013 // commit esp adjust
5014 if (ops[j].op != OP_POP)
5015 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
5016 else {
5017 for (l = 0; l < pp->argc_stack; l++)
5018 ops[j + l].flags |= OPF_DONE | OPF_RMD | OPF_NOREGS;
5019 }
5020 }
5021
5022 if (strstr(pp->ret_type.name, "int64"))
5023 need_tmp64 = 1;
5024
5025 po->flags |= OPF_DONE;
5026 }
5027 }
5028 }
5029
5030 // pass5:
5031 // - process calls, stage 2
5032 // - handle some push/pop pairs
5033 // - scan for STD/CLD, propagate DF
5034 for (i = 0; i < opcnt; i++)
5035 {
5036 po = &ops[i];
5037 if (po->flags & OPF_RMD)
5038 continue;
5039
5040 if (po->op == OP_CALL)
5041 {
5042 if (!(po->flags & OPF_DONE)) {
5043 pp = process_call(i, opcnt);
5044
5045 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
5046 // since we know the args, collect them
5047 collect_call_args(po, i, pp, &regmask, save_arg_vars,
5048 i + opcnt * 2);
5049 }
5050 // for unresolved, collect after other passes
5051 }
5052
5053 pp = po->pp;
5054 ferr_assert(po, pp != NULL);
5055
5056 po->regmask_src |= get_pp_arg_regmask_src(pp);
5057 po->regmask_dst |= get_pp_arg_regmask_dst(pp);
5058
5059 if (po->regmask_dst & mxST0)
5060 po->flags |= OPF_FPUSH;
5061
5062 if (strstr(pp->ret_type.name, "int64"))
5063 need_tmp64 = 1;
5064
5065 continue;
5066 }
5067
5068 if (po->flags & OPF_DONE)
5069 continue;
5070
5071 if (po->op == OP_PUSH && !(po->flags & OPF_FARG)
5072 && !(po->flags & OPF_RSAVE) && po->operand[0].type == OPT_CONST)
5073 {
5074 scan_for_pop_const(i, opcnt, i + opcnt * 12);
5075 }
5076 else if (po->op == OP_POP)
5077 scan_pushes_for_pop(i, opcnt, &regmask_pp);
5078 else if (po->op == OP_STD) {
5079 po->flags |= OPF_DF | OPF_RMD | OPF_DONE;
5080 scan_propagate_df(i + 1, opcnt);
5081 }
5082 }
5083
5084 // pass6:
5085 // - find POPs for PUSHes, rm both
5086 // - scan for all used registers
5087 memset(cbits, 0, sizeof(cbits));
5088 reg_use_pass(0, opcnt, cbits, 0, &regmask,
5089 0, &regmask_save, &regmask_init, regmask_arg);
5090
5091 // pass7:
5092 // - find flag set ops for their users
5093 // - do unresolved calls
5094 // - declare indirect functions
5095 for (i = 0; i < opcnt; i++)
5096 {
5097 po = &ops[i];
5098 if (po->flags & (OPF_RMD|OPF_DONE))
5099 continue;
5100
5101 if (po->flags & OPF_CC)
5102 {
5103 int setters[16], cnt = 0, branched = 0;
5104
5105 ret = scan_for_flag_set(i, i + opcnt * 6,
5106 &branched, setters, &cnt);
5107 if (ret < 0 || cnt <= 0)
5108 ferr(po, "unable to trace flag setter(s)\n");
5109 if (cnt > ARRAY_SIZE(setters))
5110 ferr(po, "too many flag setters\n");
5111
5112 for (j = 0; j < cnt; j++)
5113 {
5114 tmp_op = &ops[setters[j]]; // flag setter
5115 pfomask = 0;
5116
5117 // to get nicer code, we try to delay test and cmp;
5118 // if we can't because of operand modification, or if we
5119 // have arith op, or branch, make it calculate flags explicitly
5120 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP)
5121 {
5122 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i, 0) >= 0)
5123 pfomask = 1 << po->pfo;
5124 }
5125 else if (tmp_op->op == OP_CMPS || tmp_op->op == OP_SCAS) {
5126 pfomask = 1 << po->pfo;
5127 }
5128 else {
5129 // see if we'll be able to handle based on op result
5130 if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR
5131 && po->pfo != PFO_Z && po->pfo != PFO_S
5132 && po->pfo != PFO_P)
5133 || branched
5134 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
5135 {
5136 pfomask = 1 << po->pfo;
5137 }
5138
5139 if (tmp_op->op == OP_ADD && po->pfo == PFO_C) {
5140 propagate_lmod(tmp_op, &tmp_op->operand[0],
5141 &tmp_op->operand[1]);
5142 if (tmp_op->operand[0].lmod == OPLM_DWORD)
5143 need_tmp64 = 1;
5144 }
5145 }
5146 if (pfomask) {
5147 tmp_op->pfomask |= pfomask;
5148 cond_vars |= pfomask;
5149 }
5150 // note: may overwrite, currently not a problem
5151 po->datap = tmp_op;
5152 }
5153
5154 if (po->op == OP_RCL || po->op == OP_RCR
5155 || po->op == OP_ADC || po->op == OP_SBB)
5156 cond_vars |= 1 << PFO_C;
5157 }
5158
5159 if (po->op == OP_CMPS || po->op == OP_SCAS) {
5160 cond_vars |= 1 << PFO_Z;
5161 }
5162 else if (po->op == OP_MUL
5163 || (po->op == OP_IMUL && po->operand_cnt == 1))
5164 {
5165 if (po->operand[0].lmod == OPLM_DWORD)
5166 need_tmp64 = 1;
5167 }
5168 else if (po->op == OP_CALL) {
5169 // note: resolved non-reg calls are OPF_DONE already
5170 pp = po->pp;
5171 ferr_assert(po, pp != NULL);
5172
5173 if (pp->is_unresolved) {
5174 int regmask_stack = 0;
5175 collect_call_args(po, i, pp, &regmask, save_arg_vars,
5176 i + opcnt * 2);
5177
5178 // this is pretty rough guess:
5179 // see ecx and edx were pushed (and not their saved versions)
5180 for (arg = 0; arg < pp->argc; arg++) {
5181 if (pp->arg[arg].reg != NULL)
5182 continue;
5183
5184 tmp_op = pp->arg[arg].datap;
5185 if (tmp_op == NULL)
5186 ferr(po, "parsed_op missing for arg%d\n", arg);
5187 if (tmp_op->p_argnum == 0 && tmp_op->operand[0].type == OPT_REG)
5188 regmask_stack |= 1 << tmp_op->operand[0].reg;
5189 }
5190
5191 if (!((regmask_stack & (1 << xCX))
5192 && (regmask_stack & (1 << xDX))))
5193 {
5194 if (pp->argc_stack != 0
5195 || ((regmask | regmask_arg) & ((1 << xCX)|(1 << xDX))))
5196 {
5197 pp_insert_reg_arg(pp, "ecx");
5198 pp->is_fastcall = 1;
5199 regmask_init |= 1 << xCX;
5200 regmask |= 1 << xCX;
5201 }
5202 if (pp->argc_stack != 0
5203 || ((regmask | regmask_arg) & (1 << xDX)))
5204 {
5205 pp_insert_reg_arg(pp, "edx");
5206 regmask_init |= 1 << xDX;
5207 regmask |= 1 << xDX;
5208 }
5209 }
5210
5211 // note: __cdecl doesn't fall into is_unresolved category
5212 if (pp->argc_stack > 0)
5213 pp->is_stdcall = 1;
5214 }
5215 }
5216 else if (po->op == OP_MOV && po->operand[0].pp != NULL
5217 && po->operand[1].pp != NULL)
5218 {
5219 // <var> = offset <something>
5220 if ((po->operand[1].pp->is_func || po->operand[1].pp->is_fptr)
5221 && !IS_START(po->operand[1].name, "off_"))
5222 {
5223 if (!po->operand[0].pp->is_fptr)
5224 ferr(po, "%s not declared as fptr when it should be\n",
5225 po->operand[0].name);
5226 if (pp_cmp_func(po->operand[0].pp, po->operand[1].pp)) {
5227 pp_print(buf1, sizeof(buf1), po->operand[0].pp);
5228 pp_print(buf2, sizeof(buf2), po->operand[1].pp);
5229 fnote(po, "var: %s\n", buf1);
5230 fnote(po, "func: %s\n", buf2);
5231 ferr(po, "^ mismatch\n");
5232 }
5233 }
5234 }
5235 else if (po->op == OP_DIV || po->op == OP_IDIV) {
5236 if (po->operand[0].lmod == OPLM_DWORD) {
5237 // 32bit division is common, look for it
5238 if (po->op == OP_DIV)
5239 ret = scan_for_reg_clear(i, xDX);
5240 else
5241 ret = scan_for_cdq_edx(i);
5242 if (ret >= 0)
5243 po->flags |= OPF_32BIT;
5244 else
5245 need_tmp64 = 1;
5246 }
5247 else
5248 need_tmp_var = 1;
5249 }
5250 else if (po->op == OP_CLD)
5251 po->flags |= OPF_RMD | OPF_DONE;
5252 else if (po->op == OPP_FTOL) {
5253 struct parsed_opr opr = OPR_INIT(OPT_REG, OPLM_DWORD, xDX);
5254 j = -1;
5255 find_next_read(i + 1, opcnt, &opr, i + opcnt * 18, &j);
5256 if (j == -1)
5257 po->flags |= OPF_32BIT;
5258 }
5259
5260 if (po->op == OP_RCL || po->op == OP_RCR || po->op == OP_XCHG)
5261 need_tmp_var = 1;
5262 }
5263
5264 // output starts here
5265
5266 // define userstack size
5267 if (g_func_pp->is_userstack) {
5268 fprintf(fout, "#ifndef US_SZ_%s\n", g_func_pp->name);
5269 fprintf(fout, "#define US_SZ_%s USERSTACK_SIZE\n", g_func_pp->name);
5270 fprintf(fout, "#endif\n");
5271 }
5272
5273 // the function itself
5274 ferr_assert(ops, !g_func_pp->is_fptr);
5275 output_pp(fout, g_func_pp,
5276 (g_ida_func_attr & IDAFA_NORETURN) ? OPP_FORCE_NORETURN : 0);
5277 fprintf(fout, "\n{\n");
5278
5279 // declare indirect functions
5280 for (i = 0; i < opcnt; i++) {
5281 po = &ops[i];
5282 if (po->flags & OPF_RMD)
5283 continue;
5284
5285 if (po->op == OP_CALL) {
5286 pp = po->pp;
5287 if (pp == NULL)
5288 ferr(po, "NULL pp\n");
5289
5290 if (pp->is_fptr && !(pp->name[0] != 0 && pp->is_arg)) {
5291 if (pp->name[0] != 0) {
5292 memmove(pp->name + 2, pp->name, strlen(pp->name) + 1);
5293 memcpy(pp->name, "i_", 2);
5294
5295 // might be declared already
5296 found = 0;
5297 for (j = 0; j < i; j++) {
5298 if (ops[j].op == OP_CALL && (pp_tmp = ops[j].pp)) {
5299 if (pp_tmp->is_fptr && IS(pp->name, pp_tmp->name)) {
5300 found = 1;
5301 break;
5302 }
5303 }
5304 }
5305 if (found)
5306 continue;
5307 }
5308 else
5309 snprintf(pp->name, sizeof(pp->name), "icall%d", i);
5310
5311 fprintf(fout, " ");
5312 output_pp(fout, pp, OPP_SIMPLE_ARGS);
5313 fprintf(fout, ";\n");
5314 }
5315 }
5316 }
5317
5318 // output LUTs/jumptables
5319 for (i = 0; i < g_func_pd_cnt; i++) {
5320 pd = &g_func_pd[i];
5321 fprintf(fout, " static const ");
5322 if (pd->type == OPT_OFFSET) {
5323 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
5324
5325 for (j = 0; j < pd->count; j++) {
5326 if (j > 0)
5327 fprintf(fout, ", ");
5328 fprintf(fout, "&&%s", pd->d[j].u.label);
5329 }
5330 }
5331 else {
5332 fprintf(fout, "%s %s[] =\n { ",
5333 lmod_type_u(ops, pd->lmod), pd->label);
5334
5335 for (j = 0; j < pd->count; j++) {
5336 if (j > 0)
5337 fprintf(fout, ", ");
5338 fprintf(fout, "%u", pd->d[j].u.val);
5339 }
5340 }
5341 fprintf(fout, " };\n");
5342 had_decl = 1;
5343 }
5344
5345 // declare stack frame, va_arg
5346 if (g_stack_fsz) {
5347 fprintf(fout, " union { u32 d[%d];", (g_stack_fsz + 3) / 4);
5348 if (g_func_lmods & (1 << OPLM_WORD))
5349 fprintf(fout, " u16 w[%d];", (g_stack_fsz + 1) / 2);
5350 if (g_func_lmods & (1 << OPLM_BYTE))
5351 fprintf(fout, " u8 b[%d];", g_stack_fsz);
5352 if (g_func_lmods & (1 << OPLM_QWORD))
5353 fprintf(fout, " double q[%d];", (g_stack_fsz + 7) / 8);
5354 fprintf(fout, " } sf;\n");
5355 had_decl = 1;
5356 }
5357
5358 if (g_func_pp->is_userstack) {
5359 fprintf(fout, " u32 fake_sf[US_SZ_%s / 4];\n", g_func_pp->name);
5360 fprintf(fout, " u32 *esp = &fake_sf[sizeof(fake_sf) / 4];\n");
5361 had_decl = 1;
5362 }
5363
5364 if (g_func_pp->is_vararg) {
5365 fprintf(fout, " va_list ap;\n");
5366 had_decl = 1;
5367 }
5368
5369 // declare arg-registers
5370 for (i = 0; i < g_func_pp->argc; i++) {
5371 if (g_func_pp->arg[i].reg != NULL) {
5372 reg = char_array_i(regs_r32,
5373 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
5374 if (regmask & (1 << reg)) {
5375 if (g_func_pp->arg[i].type.is_retreg)
5376 fprintf(fout, " u32 %s = *r_%s;\n",
5377 g_func_pp->arg[i].reg, g_func_pp->arg[i].reg);
5378 else
5379 fprintf(fout, " u32 %s = (u32)a%d;\n",
5380 g_func_pp->arg[i].reg, i + 1);
5381 }
5382 else {
5383 if (g_func_pp->arg[i].type.is_retreg)
5384 ferr(ops, "retreg '%s' is unused?\n",
5385 g_func_pp->arg[i].reg);
5386 fprintf(fout, " // %s = a%d; // unused\n",
5387 g_func_pp->arg[i].reg, i + 1);
5388 }
5389 had_decl = 1;
5390 }
5391 }
5392
5393 // declare normal registers
5394 regmask_now = regmask & ~regmask_arg;
5395 regmask_now &= ~(1 << xSP);
5396 if (regmask_now & 0x00ff) {
5397 for (reg = 0; reg < 8; reg++) {
5398 if (regmask_now & (1 << reg)) {
5399 fprintf(fout, " u32 %s", regs_r32[reg]);
5400 if (regmask_init & (1 << reg))
5401 fprintf(fout, " = 0");
5402 fprintf(fout, ";\n");
5403 had_decl = 1;
5404 }
5405 }
5406 }
5407 // ... mmx
5408 if (regmask_now & 0xff00) {
5409 for (reg = 8; reg < 16; reg++) {
5410 if (regmask_now & (1 << reg)) {
5411 fprintf(fout, " mmxr %s", regs_r32[reg]);
5412 if (regmask_init & (1 << reg))
5413 fprintf(fout, " = { 0, }");
5414 fprintf(fout, ";\n");
5415 had_decl = 1;
5416 }
5417 }
5418 }
5419 // ... x87
5420 if (regmask_now & 0xff0000) {
5421 for (reg = 16; reg < 24; reg++) {
5422 if (regmask_now & (1 << reg)) {
5423 fprintf(fout, " double f_st%d", reg - 16);
5424 if (regmask_init & (1 << reg))
5425 fprintf(fout, " = 0");
5426 fprintf(fout, ";\n");
5427 had_decl = 1;
5428 }
5429 }
5430 }
5431
5432 if (regmask_save) {
5433 for (reg = 0; reg < 8; reg++) {
5434 if (regmask_save & (1 << reg)) {
5435 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
5436 had_decl = 1;
5437 }
5438 }
5439 }
5440
5441 for (i = 0; i < ARRAY_SIZE(save_arg_vars); i++) {
5442 if (save_arg_vars[i] == 0)
5443 continue;
5444 for (reg = 0; reg < 32; reg++) {
5445 if (save_arg_vars[i] & (1 << reg)) {
5446 fprintf(fout, " u32 %s;\n",
5447 saved_arg_name(buf1, sizeof(buf1), i, reg + 1));
5448 had_decl = 1;
5449 }
5450 }
5451 }
5452
5453 // declare push-pop temporaries
5454 if (regmask_pp) {
5455 for (reg = 0; reg < 8; reg++) {
5456 if (regmask_pp & (1 << reg)) {
5457 fprintf(fout, " u32 pp_%s;\n", regs_r32[reg]);
5458 had_decl = 1;
5459 }
5460 }
5461 }
5462
5463 if (cond_vars) {
5464 for (i = 0; i < 8; i++) {
5465 if (cond_vars & (1 << i)) {
5466 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
5467 had_decl = 1;
5468 }
5469 }
5470 }
5471
5472 if (need_tmp_var) {
5473 fprintf(fout, " u32 tmp;\n");
5474 had_decl = 1;
5475 }
5476
5477 if (need_tmp64) {
5478 fprintf(fout, " u64 tmp64;\n");
5479 had_decl = 1;
5480 }
5481
5482 if (had_decl)
5483 fprintf(fout, "\n");
5484
5485 // do stack clear, if needed
5486 if (g_sct_func_attr & SCTFA_CLEAR_SF) {
5487 fprintf(fout, " ");
5488 if (g_stack_clear_len != 0) {
5489 if (g_stack_clear_len <= 4) {
5490 for (i = 0; i < g_stack_clear_len; i++)
5491 fprintf(fout, "sf.d[%d] = ", g_stack_clear_start + i);
5492 fprintf(fout, "0;\n");
5493 }
5494 else {
5495 fprintf(fout, "memset(&sf[%d], 0, %d);\n",
5496 g_stack_clear_start, g_stack_clear_len * 4);
5497 }
5498 }
5499 else
5500 fprintf(fout, "memset(&sf, 0, sizeof(sf));\n");
5501 }
5502
5503 if (g_func_pp->is_vararg) {
5504 if (g_func_pp->argc_stack == 0)
5505 ferr(ops, "vararg func without stack args?\n");
5506 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
5507 }
5508
5509 // output ops
5510 for (i = 0; i < opcnt; i++)
5511 {
5512 if (g_labels[i] != NULL) {
5513 fprintf(fout, "\n%s:\n", g_labels[i]);
5514 label_pending = 1;
5515
5516 delayed_flag_op = NULL;
5517 last_arith_dst = NULL;
5518 }
5519
5520 po = &ops[i];
5521 if (po->flags & OPF_RMD)
5522 continue;
5523
5524 no_output = 0;
5525
5526 #define assert_operand_cnt(n_) \
5527 if (po->operand_cnt != n_) \
5528 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
5529
5530 // conditional/flag using op?
5531 if (po->flags & OPF_CC)
5532 {
5533 int is_delayed = 0;
5534
5535 tmp_op = po->datap;
5536
5537 // we go through all this trouble to avoid using parsed_flag_op,
5538 // which makes generated code much nicer
5539 if (delayed_flag_op != NULL)
5540 {
5541 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op,
5542 po->pfo, po->pfo_inv);
5543 is_delayed = 1;
5544 }
5545 else if (last_arith_dst != NULL
5546 && (po->pfo == PFO_Z || po->pfo == PFO_S || po->pfo == PFO_P
5547 || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR))
5548 ))
5549 {
5550 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
5551 out_test_for_cc(buf1, sizeof(buf1), po, po->pfo, po->pfo_inv,
5552 last_arith_dst->lmod, buf3);
5553 is_delayed = 1;
5554 }
5555 else if (tmp_op != NULL) {
5556 // use preprocessed flag calc results
5557 if (!(tmp_op->pfomask & (1 << po->pfo)))
5558 ferr(po, "not prepared for pfo %d\n", po->pfo);
5559
5560 // note: pfo_inv was not yet applied
5561 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
5562 po->pfo_inv ? "!" : "", parsed_flag_op_names[po->pfo]);
5563 }
5564 else {
5565 ferr(po, "all methods of finding comparison failed\n");
5566 }
5567
5568 if (po->flags & OPF_JMP) {
5569 fprintf(fout, " if %s", buf1);
5570 }
5571 else if (po->op == OP_RCL || po->op == OP_RCR
5572 || po->op == OP_ADC || po->op == OP_SBB)
5573 {
5574 if (is_delayed)
5575 fprintf(fout, " cond_%s = %s;\n",
5576 parsed_flag_op_names[po->pfo], buf1);
5577 }
5578 else if (po->flags & OPF_DATA) { // SETcc
5579 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
5580 fprintf(fout, " %s = %s;", buf2, buf1);
5581 }
5582 else {
5583 ferr(po, "unhandled conditional op\n");
5584 }
5585 }
5586
5587 pfomask = po->pfomask;
5588
5589 if (po->flags & (OPF_REPZ|OPF_REPNZ)) {
5590 struct parsed_opr opr = OPR_INIT(OPT_REG, OPLM_DWORD, xCX);
5591 ret = try_resolve_const(i, &opr, opcnt * 7 + i, &uval);
5592
5593 if (ret != 1 || uval == 0) {
5594 // we need initial flags for ecx=0 case..
5595 if (i > 0 && ops[i - 1].op == OP_XOR
5596 && IS(ops[i - 1].operand[0].name,
5597 ops[i - 1].operand[1].name))
5598 {
5599 fprintf(fout, " cond_z = ");
5600 if (pfomask & (1 << PFO_C))
5601 fprintf(fout, "cond_c = ");
5602 fprintf(fout, "0;\n");
5603 }
5604 else if (last_arith_dst != NULL) {
5605 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
5606 out_test_for_cc(buf1, sizeof(buf1), po, PFO_Z, 0,
5607 last_arith_dst->lmod, buf3);
5608 fprintf(fout, " cond_z = %s;\n", buf1);
5609 }
5610 else
5611 ferr(po, "missing initial ZF\n");
5612 }
5613 }
5614
5615 switch (po->op)
5616 {
5617 case OP_MOV:
5618 assert_operand_cnt(2);
5619 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5620 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5621 default_cast_to(buf3, sizeof(buf3), &po->operand[0]);
5622 fprintf(fout, " %s = %s;", buf1,
5623 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
5624 buf3, 0));
5625 break;
5626
5627 case OP_LEA:
5628 assert_operand_cnt(2);
5629 po->operand[1].lmod = OPLM_DWORD; // always
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 NULL, 1));
5634 break;
5635
5636 case OP_MOVZX:
5637 assert_operand_cnt(2);
5638 fprintf(fout, " %s = %s;",
5639 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5640 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5641 break;
5642
5643 case OP_MOVSX:
5644 assert_operand_cnt(2);
5645 switch (po->operand[1].lmod) {
5646 case OPLM_BYTE:
5647 strcpy(buf3, "(s8)");
5648 break;
5649 case OPLM_WORD:
5650 strcpy(buf3, "(s16)");
5651 break;
5652 default:
5653 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
5654 }
5655 fprintf(fout, " %s = %s;",
5656 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5657 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
5658 buf3, 0));
5659 break;
5660
5661 case OP_XCHG:
5662 assert_operand_cnt(2);
5663 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5664 fprintf(fout, " tmp = %s;",
5665 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], "", 0));
5666 fprintf(fout, " %s = %s;",
5667 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5668 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
5669 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
5670 fprintf(fout, " %s = %stmp;",
5671 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]),
5672 default_cast_to(buf3, sizeof(buf3), &po->operand[1]));
5673 snprintf(g_comment, sizeof(g_comment), "xchg");
5674 break;
5675
5676 case OP_NOT:
5677 assert_operand_cnt(1);
5678 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5679 fprintf(fout, " %s = ~%s;", buf1, buf1);
5680 break;
5681
5682 case OP_XLAT:
5683 assert_operand_cnt(2);
5684 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5685 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
5686 fprintf(fout, " %s = *(u8 *)(%s + %s);", buf1, buf2, buf1);
5687 strcpy(g_comment, "xlat");
5688 break;
5689
5690 case OP_CDQ:
5691 assert_operand_cnt(2);
5692 fprintf(fout, " %s = (s32)%s >> 31;",
5693 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5694 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5695 strcpy(g_comment, "cdq");
5696 break;
5697
5698 case OP_LODS:
5699 if (po->flags & OPF_REP) {
5700 assert_operand_cnt(3);
5701 // hmh..
5702 ferr(po, "TODO\n");
5703 }
5704 else {
5705 assert_operand_cnt(2);
5706 fprintf(fout, " %s = %sesi; esi %c= %d;",
5707 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]),
5708 lmod_cast_u_ptr(po, po->operand[1].lmod),
5709 (po->flags & OPF_DF) ? '-' : '+',
5710 lmod_bytes(po, po->operand[1].lmod));
5711 strcpy(g_comment, "lods");
5712 }
5713 break;
5714
5715 case OP_STOS:
5716 if (po->flags & OPF_REP) {
5717 assert_operand_cnt(3);
5718 fprintf(fout, " for (; ecx != 0; ecx--, edi %c= %d)\n",
5719 (po->flags & OPF_DF) ? '-' : '+',
5720 lmod_bytes(po, po->operand[1].lmod));
5721 fprintf(fout, " %sedi = eax;",
5722 lmod_cast_u_ptr(po, po->operand[1].lmod));
5723 strcpy(g_comment, "rep stos");
5724 }
5725 else {
5726 assert_operand_cnt(2);
5727 fprintf(fout, " %sedi = eax; edi %c= %d;",
5728 lmod_cast_u_ptr(po, po->operand[1].lmod),
5729 (po->flags & OPF_DF) ? '-' : '+',
5730 lmod_bytes(po, po->operand[1].lmod));
5731 strcpy(g_comment, "stos");
5732 }
5733 break;
5734
5735 case OP_MOVS:
5736 j = lmod_bytes(po, po->operand[0].lmod);
5737 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
5738 l = (po->flags & OPF_DF) ? '-' : '+';
5739 if (po->flags & OPF_REP) {
5740 assert_operand_cnt(3);
5741 fprintf(fout,
5742 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n",
5743 l, j, l, j);
5744 fprintf(fout,
5745 " %sedi = %sesi;", buf1, buf1);
5746 strcpy(g_comment, "rep movs");
5747 }
5748 else {
5749 assert_operand_cnt(2);
5750 fprintf(fout, " %sedi = %sesi; edi %c= %d; esi %c= %d;",
5751 buf1, buf1, l, j, l, j);
5752 strcpy(g_comment, "movs");
5753 }
5754 break;
5755
5756 case OP_CMPS:
5757 // repe ~ repeat while ZF=1
5758 j = lmod_bytes(po, po->operand[0].lmod);
5759 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
5760 l = (po->flags & OPF_DF) ? '-' : '+';
5761 if (po->flags & OPF_REP) {
5762 assert_operand_cnt(3);
5763 fprintf(fout,
5764 " for (; ecx != 0; ecx--) {\n");
5765 if (pfomask & (1 << PFO_C)) {
5766 // ugh..
5767 fprintf(fout,
5768 " cond_c = %sesi < %sedi;\n", buf1, buf1);
5769 pfomask &= ~(1 << PFO_C);
5770 }
5771 fprintf(fout,
5772 " cond_z = (%sesi == %sedi); esi %c= %d, edi %c= %d;\n",
5773 buf1, buf1, l, j, l, j);
5774 fprintf(fout,
5775 " if (cond_z %s 0) break;\n",
5776 (po->flags & OPF_REPZ) ? "==" : "!=");
5777 fprintf(fout,
5778 " }");
5779 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
5780 (po->flags & OPF_REPZ) ? "e" : "ne");
5781 }
5782 else {
5783 assert_operand_cnt(2);
5784 fprintf(fout,
5785 " cond_z = (%sesi == %sedi); esi %c= %d; edi %c= %d;",
5786 buf1, buf1, l, j, l, j);
5787 strcpy(g_comment, "cmps");
5788 }
5789 pfomask &= ~(1 << PFO_Z);
5790 last_arith_dst = NULL;
5791 delayed_flag_op = NULL;
5792 break;
5793
5794 case OP_SCAS:
5795 // only does ZF (for now)
5796 // repe ~ repeat while ZF=1
5797 j = lmod_bytes(po, po->operand[1].lmod);
5798 l = (po->flags & OPF_DF) ? '-' : '+';
5799 if (po->flags & OPF_REP) {
5800 assert_operand_cnt(3);
5801 fprintf(fout,
5802 " for (; ecx != 0; ecx--) {\n");
5803 fprintf(fout,
5804 " cond_z = (%seax == %sedi); edi %c= %d;\n",
5805 lmod_cast_u(po, po->operand[1].lmod),
5806 lmod_cast_u_ptr(po, po->operand[1].lmod), l, j);
5807 fprintf(fout,
5808 " if (cond_z %s 0) break;\n",
5809 (po->flags & OPF_REPZ) ? "==" : "!=");
5810 fprintf(fout,
5811 " }");
5812 snprintf(g_comment, sizeof(g_comment), "rep%s scas",
5813 (po->flags & OPF_REPZ) ? "e" : "ne");
5814 }
5815 else {
5816 assert_operand_cnt(2);
5817 fprintf(fout, " cond_z = (%seax == %sedi); edi %c= %d;",
5818 lmod_cast_u(po, po->operand[1].lmod),
5819 lmod_cast_u_ptr(po, po->operand[1].lmod), l, j);
5820 strcpy(g_comment, "scas");
5821 }
5822 pfomask &= ~(1 << PFO_Z);
5823 last_arith_dst = NULL;
5824 delayed_flag_op = NULL;
5825 break;
5826
5827 // arithmetic w/flags
5828 case OP_AND:
5829 if (po->operand[1].type == OPT_CONST && !po->operand[1].val)
5830 goto dualop_arith_const;
5831 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5832 goto dualop_arith;
5833
5834 case OP_OR:
5835 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5836 if (po->operand[1].type == OPT_CONST) {
5837 j = lmod_bytes(po, po->operand[0].lmod);
5838 if (((1ull << j * 8) - 1) == po->operand[1].val)
5839 goto dualop_arith_const;
5840 }
5841 goto dualop_arith;
5842
5843 dualop_arith:
5844 assert_operand_cnt(2);
5845 fprintf(fout, " %s %s= %s;",
5846 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5847 op_to_c(po),
5848 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5849 output_std_flags(fout, po, &pfomask, buf1);
5850 last_arith_dst = &po->operand[0];
5851 delayed_flag_op = NULL;
5852 break;
5853
5854 dualop_arith_const:
5855 // and 0, or ~0 used instead mov
5856 assert_operand_cnt(2);
5857 fprintf(fout, " %s = %s;",
5858 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
5859 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
5860 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
5861 output_std_flags(fout, po, &pfomask, buf1);
5862 last_arith_dst = &po->operand[0];
5863 delayed_flag_op = NULL;
5864 break;
5865
5866 case OP_SHL:
5867 case OP_SHR:
5868 assert_operand_cnt(2);
5869 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5870 if (pfomask & (1 << PFO_C)) {
5871 if (po->operand[1].type == OPT_CONST) {
5872 l = lmod_bytes(po, po->operand[0].lmod) * 8;
5873 j = po->operand[1].val;
5874 j %= l;
5875 if (j != 0) {
5876 if (po->op == OP_SHL)
5877 j = l - j;
5878 else
5879 j -= 1;
5880 fprintf(fout, " cond_c = (%s >> %d) & 1;\n",
5881 buf1, j);
5882 }
5883 else
5884 ferr(po, "zero shift?\n");
5885 }
5886 else
5887 ferr(po, "TODO\n");
5888 pfomask &= ~(1 << PFO_C);
5889 }
5890 fprintf(fout, " %s %s= %s", buf1, op_to_c(po),
5891 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5892 if (po->operand[1].type != OPT_CONST)
5893 fprintf(fout, " & 0x1f");
5894 fprintf(fout, ";");
5895 output_std_flags(fout, po, &pfomask, buf1);
5896 last_arith_dst = &po->operand[0];
5897 delayed_flag_op = NULL;
5898 break;
5899
5900 case OP_SAR:
5901 assert_operand_cnt(2);
5902 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5903 fprintf(fout, " %s = %s%s >> %s;", buf1,
5904 lmod_cast_s(po, po->operand[0].lmod), buf1,
5905 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
5906 output_std_flags(fout, po, &pfomask, buf1);
5907 last_arith_dst = &po->operand[0];
5908 delayed_flag_op = NULL;
5909 break;
5910
5911 case OP_SHLD:
5912 case OP_SHRD:
5913 assert_operand_cnt(3);
5914 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5915 l = lmod_bytes(po, po->operand[0].lmod) * 8;
5916 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[2]);
5917 if (po->operand[2].type != OPT_CONST) {
5918 // no handling for "undefined" case, hopefully not needed
5919 snprintf(buf2, sizeof(buf2), "(%s & 0x1f)", buf3);
5920 strcpy(buf3, buf2);
5921 }
5922 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
5923 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5924 if (po->op == OP_SHLD) {
5925 fprintf(fout, " %s <<= %s; %s |= %s >> (%d - %s);",
5926 buf1, buf3, buf1, buf2, l, buf3);
5927 strcpy(g_comment, "shld");
5928 }
5929 else {
5930 fprintf(fout, " %s >>= %s; %s |= %s << (%d - %s);",
5931 buf1, buf3, buf1, buf2, l, buf3);
5932 strcpy(g_comment, "shrd");
5933 }
5934 output_std_flags(fout, po, &pfomask, buf1);
5935 last_arith_dst = &po->operand[0];
5936 delayed_flag_op = NULL;
5937 break;
5938
5939 case OP_ROL:
5940 case OP_ROR:
5941 assert_operand_cnt(2);
5942 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5943 if (po->operand[1].type == OPT_CONST) {
5944 j = po->operand[1].val;
5945 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
5946 fprintf(fout, po->op == OP_ROL ?
5947 " %s = (%s << %d) | (%s >> %d);" :
5948 " %s = (%s >> %d) | (%s << %d);",
5949 buf1, buf1, j, buf1,
5950 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
5951 }
5952 else
5953 ferr(po, "TODO\n");
5954 output_std_flags(fout, po, &pfomask, buf1);
5955 last_arith_dst = &po->operand[0];
5956 delayed_flag_op = NULL;
5957 break;
5958
5959 case OP_RCL:
5960 case OP_RCR:
5961 assert_operand_cnt(2);
5962 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5963 l = lmod_bytes(po, po->operand[0].lmod) * 8;
5964 if (po->operand[1].type == OPT_CONST) {
5965 j = po->operand[1].val % l;
5966 if (j == 0)
5967 ferr(po, "zero rotate\n");
5968 fprintf(fout, " tmp = (%s >> %d) & 1;\n",
5969 buf1, (po->op == OP_RCL) ? (l - j) : (j - 1));
5970 if (po->op == OP_RCL) {
5971 fprintf(fout,
5972 " %s = (%s << %d) | (cond_c << %d)",
5973 buf1, buf1, j, j - 1);
5974 if (j != 1)
5975 fprintf(fout, " | (%s >> %d)", buf1, l + 1 - j);
5976 }
5977 else {
5978 fprintf(fout,
5979 " %s = (%s >> %d) | (cond_c << %d)",
5980 buf1, buf1, j, l - j);
5981 if (j != 1)
5982 fprintf(fout, " | (%s << %d)", buf1, l + 1 - j);
5983 }
5984 fprintf(fout, ";\n");
5985 fprintf(fout, " cond_c = tmp;");
5986 }
5987 else
5988 ferr(po, "TODO\n");
5989 strcpy(g_comment, (po->op == OP_RCL) ? "rcl" : "rcr");
5990 output_std_flags(fout, po, &pfomask, buf1);
5991 last_arith_dst = &po->operand[0];
5992 delayed_flag_op = NULL;
5993 break;
5994
5995 case OP_XOR:
5996 assert_operand_cnt(2);
5997 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5998 if (IS(opr_name(po, 0), opr_name(po, 1))) {
5999 // special case for XOR
6000 if (pfomask & (1 << PFO_BE)) { // weird, but it happens..
6001 fprintf(fout, " cond_be = 1;\n");
6002 pfomask &= ~(1 << PFO_BE);
6003 }
6004 fprintf(fout, " %s = 0;",
6005 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
6006 last_arith_dst = &po->operand[0];
6007 delayed_flag_op = NULL;
6008 break;
6009 }
6010 goto dualop_arith;
6011
6012 case OP_ADD:
6013 assert_operand_cnt(2);
6014 propagate_lmod(po, &po->operand[0], &po->operand[1]);
6015 if (pfomask & (1 << PFO_C)) {
6016 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
6017 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
6018 if (po->operand[0].lmod == OPLM_DWORD) {
6019 fprintf(fout, " tmp64 = (u64)%s + %s;\n", buf1, buf2);
6020 fprintf(fout, " cond_c = tmp64 >> 32;\n");
6021 fprintf(fout, " %s = (u32)tmp64;",
6022 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
6023 strcat(g_comment, " add64");
6024 }
6025 else {
6026 fprintf(fout, " cond_c = ((u32)%s + %s) >> %d;\n",
6027 buf1, buf2, lmod_bytes(po, po->operand[0].lmod) * 8);
6028 fprintf(fout, " %s += %s;",
6029 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
6030 buf2);
6031 }
6032 pfomask &= ~(1 << PFO_C);
6033 output_std_flags(fout, po, &pfomask, buf1);
6034 last_arith_dst = &po->operand[0];
6035 delayed_flag_op = NULL;
6036 break;
6037 }
6038 goto dualop_arith;
6039
6040 case OP_SUB:
6041 assert_operand_cnt(2);
6042 propagate_lmod(po, &po->operand[0], &po->operand[1]);
6043 if (pfomask & ~((1 << PFO_Z) | (1 << PFO_S))) {
6044 for (j = 0; j <= PFO_LE; j++) {
6045 if (!(pfomask & (1 << j)))
6046 continue;
6047 if (j == PFO_Z || j == PFO_S)
6048 continue;
6049
6050 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
6051 fprintf(fout, " cond_%s = %s;\n",
6052 parsed_flag_op_names[j], buf1);
6053 pfomask &= ~(1 << j);
6054 }
6055 }
6056 goto dualop_arith;
6057
6058 case OP_ADC:
6059 case OP_SBB:
6060 assert_operand_cnt(2);
6061 propagate_lmod(po, &po->operand[0], &po->operand[1]);
6062 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
6063 if (po->op == OP_SBB
6064 && IS(po->operand[0].name, po->operand[1].name))
6065 {
6066 // avoid use of unitialized var
6067 fprintf(fout, " %s = -cond_c;", buf1);
6068 // carry remains what it was
6069 pfomask &= ~(1 << PFO_C);
6070 }
6071 else {
6072 fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po),
6073 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
6074 }
6075 output_std_flags(fout, po, &pfomask, buf1);
6076 last_arith_dst = &po->operand[0];
6077 delayed_flag_op = NULL;
6078 break;
6079
6080 case OP_BSF:
6081 assert_operand_cnt(2);
6082 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
6083 fprintf(fout, " %s = %s ? __builtin_ffs(%s) - 1 : 0;",
6084 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
6085 buf2, buf2);
6086 output_std_flags(fout, po, &pfomask, buf1);
6087 last_arith_dst = &po->operand[0];
6088 delayed_flag_op = NULL;
6089 strcat(g_comment, " bsf");
6090 break;
6091
6092 case OP_DEC:
6093 if (pfomask & ~(PFOB_S | PFOB_S | PFOB_C)) {
6094 for (j = 0; j <= PFO_LE; j++) {
6095 if (!(pfomask & (1 << j)))
6096 continue;
6097 if (j == PFO_Z || j == PFO_S || j == PFO_C)
6098 continue;
6099
6100 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
6101 fprintf(fout, " cond_%s = %s;\n",
6102 parsed_flag_op_names[j], buf1);
6103 pfomask &= ~(1 << j);
6104 }
6105 }
6106 // fallthrough
6107
6108 case OP_INC:
6109 if (pfomask & (1 << PFO_C))
6110 // carry is unaffected by inc/dec.. wtf?
6111 ferr(po, "carry propagation needed\n");
6112
6113 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
6114 if (po->operand[0].type == OPT_REG) {
6115 strcpy(buf2, po->op == OP_INC ? "++" : "--");
6116 fprintf(fout, " %s%s;", buf1, buf2);
6117 }
6118 else {
6119 strcpy(buf2, po->op == OP_INC ? "+" : "-");
6120 fprintf(fout, " %s %s= 1;", buf1, buf2);
6121 }
6122 output_std_flags(fout, po, &pfomask, buf1);
6123 last_arith_dst = &po->operand[0];
6124 delayed_flag_op = NULL;
6125 break;
6126
6127 case OP_NEG:
6128 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
6129 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
6130 fprintf(fout, " %s = -%s%s;", buf1,
6131 lmod_cast_s(po, po->operand[0].lmod), buf2);
6132 last_arith_dst = &po->operand[0];
6133 delayed_flag_op = NULL;
6134 if (pfomask & (1 << PFO_C)) {
6135 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
6136 pfomask &= ~(1 << PFO_C);
6137 }
6138 break;
6139
6140 case OP_IMUL:
6141 if (po->operand_cnt == 2) {
6142 propagate_lmod(po, &po->operand[0], &po->operand[1]);
6143 goto dualop_arith;
6144 }
6145 if (po->operand_cnt == 3)
6146 ferr(po, "TODO imul3\n");
6147 // fallthrough
6148 case OP_MUL:
6149 assert_operand_cnt(1);
6150 switch (po->operand[0].lmod) {
6151 case OPLM_DWORD:
6152 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
6153 fprintf(fout, " tmp64 = %seax * %s%s;\n", buf1, buf1,
6154 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
6155 fprintf(fout, " edx = tmp64 >> 32;\n");
6156 fprintf(fout, " eax = tmp64;");
6157 break;
6158 case OPLM_BYTE:
6159 strcpy(buf1, po->op == OP_IMUL ? "(s16)(s8)" : "(u16)(u8)");
6160 fprintf(fout, " LOWORD(eax) = %seax * %s;", buf1,
6161 out_src_opr(buf2, sizeof(buf2), po, &po->operand[0],
6162 buf1, 0));
6163 break;
6164 default:
6165 ferr(po, "TODO: unhandled mul type\n");
6166 break;
6167 }
6168 last_arith_dst = NULL;
6169 delayed_flag_op = NULL;
6170 break;
6171
6172 case OP_DIV:
6173 case OP_IDIV:
6174 assert_operand_cnt(1);
6175 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
6176 strcpy(cast, lmod_cast(po, po->operand[0].lmod,
6177 po->op == OP_IDIV));
6178 switch (po->operand[0].lmod) {
6179 case OPLM_DWORD:
6180 if (po->flags & OPF_32BIT)
6181 snprintf(buf2, sizeof(buf2), "%seax", cast);
6182 else {
6183 fprintf(fout, " tmp64 = ((u64)edx << 32) | eax;\n");
6184 snprintf(buf2, sizeof(buf2), "%stmp64",
6185 (po->op == OP_IDIV) ? "(s64)" : "");
6186 }
6187 if (po->operand[0].type == OPT_REG
6188 && po->operand[0].reg == xDX)
6189 {
6190 fprintf(fout, " eax = %s / %s%s;\n", buf2, cast, buf1);
6191 fprintf(fout, " edx = %s %% %s%s;", buf2, cast, buf1);
6192 }
6193 else {
6194 fprintf(fout, " edx = %s %% %s%s;\n", buf2, cast, buf1);
6195 fprintf(fout, " eax = %s / %s%s;", buf2, cast, buf1);
6196 }
6197 break;
6198 case OPLM_WORD:
6199 fprintf(fout, " tmp = (edx << 16) | (eax & 0xffff);\n");
6200 snprintf(buf2, sizeof(buf2), "%stmp",
6201 (po->op == OP_IDIV) ? "(s32)" : "");
6202 if (po->operand[0].type == OPT_REG
6203 && po->operand[0].reg == xDX)
6204 {
6205 fprintf(fout, " LOWORD(eax) = %s / %s%s;\n",
6206 buf2, cast, buf1);
6207 fprintf(fout, " LOWORD(edx) = %s %% %s%s;",
6208 buf2, cast, buf1);
6209 }
6210 else {
6211 fprintf(fout, " LOWORD(edx) = %s %% %s%s;\n",
6212 buf2, cast, buf1);
6213 fprintf(fout, " LOWORD(eax) = %s / %s%s;",
6214 buf2, cast, buf1);
6215 }
6216 strcat(g_comment, " div16");
6217 break;
6218 default:
6219 ferr(po, "unhandled div lmod %d\n", po->operand[0].lmod);
6220 }
6221 last_arith_dst = NULL;
6222 delayed_flag_op = NULL;
6223 break;
6224
6225 case OP_TEST:
6226 case OP_CMP:
6227 propagate_lmod(po, &po->operand[0], &po->operand[1]);
6228 if (pfomask != 0) {
6229 for (j = 0; j < 8; j++) {
6230 if (pfomask & (1 << j)) {
6231 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
6232 fprintf(fout, " cond_%s = %s;",
6233 parsed_flag_op_names[j], buf1);
6234 }
6235 }
6236 pfomask = 0;
6237 }
6238 else
6239 no_output = 1;
6240 last_arith_dst = NULL;
6241 delayed_flag_op = po;
6242 break;
6243
6244 case OP_SCC:
6245 // SETcc - should already be handled
6246 break;
6247
6248 // note: we reuse OP_Jcc for SETcc, only flags differ
6249 case OP_JCC:
6250 fprintf(fout, "\n goto %s;", po->operand[0].name);
6251 break;
6252
6253 case OP_JECXZ:
6254 fprintf(fout, " if (ecx == 0)\n");
6255 fprintf(fout, " goto %s;", po->operand[0].name);
6256 strcat(g_comment, " jecxz");
6257 break;
6258
6259 case OP_LOOP:
6260 fprintf(fout, " if (--ecx != 0)\n");
6261 fprintf(fout, " goto %s;", po->operand[0].name);
6262 strcat(g_comment, " loop");
6263 break;
6264
6265 case OP_JMP:
6266 assert_operand_cnt(1);
6267 last_arith_dst = NULL;
6268 delayed_flag_op = NULL;
6269
6270 if (po->operand[0].type == OPT_REGMEM) {
6271 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
6272 buf1, buf2);
6273 if (ret != 2)
6274 ferr(po, "parse failure for jmp '%s'\n",
6275 po->operand[0].name);
6276 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
6277 break;
6278 }
6279 else if (po->operand[0].type != OPT_LABEL)
6280 ferr(po, "unhandled jmp type\n");
6281
6282 fprintf(fout, " goto %s;", po->operand[0].name);
6283 break;
6284
6285 case OP_CALL:
6286 assert_operand_cnt(1);
6287 pp = po->pp;
6288 my_assert_not(pp, NULL);
6289
6290 strcpy(buf3, " ");
6291 if (po->flags & OPF_CC) {
6292 // we treat conditional branch to another func
6293 // (yes such code exists..) as conditional tailcall
6294 strcat(buf3, " ");
6295 fprintf(fout, " {\n");
6296 }
6297
6298 if (pp->is_fptr && !pp->is_arg) {
6299 fprintf(fout, "%s%s = %s;\n", buf3, pp->name,
6300 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
6301 "(void *)", 0));
6302 if (pp->is_unresolved)
6303 fprintf(fout, "%sunresolved_call(\"%s:%d\", %s);\n",
6304 buf3, asmfn, po->asmln, pp->name);
6305 }
6306
6307 fprintf(fout, "%s", buf3);
6308 if (strstr(pp->ret_type.name, "int64")) {
6309 if (po->flags & OPF_TAIL)
6310 ferr(po, "int64 and tail?\n");
6311 fprintf(fout, "tmp64 = ");
6312 }
6313 else if (!IS(pp->ret_type.name, "void")) {
6314 if (po->flags & OPF_TAIL) {
6315 if (regmask_ret & mxAX) {
6316 fprintf(fout, "return ");
6317 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
6318 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
6319 }
6320 else if (regmask_ret & mxST0)
6321 ferr(po, "float tailcall\n");
6322 }
6323 else if (po->regmask_dst & mxAX) {
6324 fprintf(fout, "eax = ");
6325 if (pp->ret_type.is_ptr)
6326 fprintf(fout, "(u32)");
6327 }
6328 else if (po->regmask_dst & mxST0) {
6329 fprintf(fout, "f_st0 = ");
6330 }
6331 }
6332
6333 if (pp->name[0] == 0)
6334 ferr(po, "missing pp->name\n");
6335 fprintf(fout, "%s%s(", pp->name,
6336 pp->has_structarg ? "_sa" : "");
6337
6338 if (po->flags & OPF_ATAIL) {
6339 if (pp->argc_stack != g_func_pp->argc_stack
6340 || (pp->argc_stack > 0
6341 && pp->is_stdcall != g_func_pp->is_stdcall))
6342 ferr(po, "incompatible tailcall\n");
6343 if (g_func_pp->has_retreg)
6344 ferr(po, "TODO: retreg+tailcall\n");
6345
6346 for (arg = j = 0; arg < pp->argc; arg++) {
6347 if (arg > 0)
6348 fprintf(fout, ", ");
6349
6350 cast[0] = 0;
6351 if (pp->arg[arg].type.is_ptr)
6352 snprintf(cast, sizeof(cast), "(%s)",
6353 pp->arg[arg].type.name);
6354
6355 if (pp->arg[arg].reg != NULL) {
6356 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
6357 continue;
6358 }
6359 // stack arg
6360 for (; j < g_func_pp->argc; j++)
6361 if (g_func_pp->arg[j].reg == NULL)
6362 break;
6363 fprintf(fout, "%sa%d", cast, j + 1);
6364 j++;
6365 }
6366 }
6367 else {
6368 for (arg = 0; arg < pp->argc; arg++) {
6369 if (arg > 0)
6370 fprintf(fout, ", ");
6371
6372 cast[0] = 0;
6373 if (pp->arg[arg].type.is_ptr)
6374 snprintf(cast, sizeof(cast), "(%s)",
6375 pp->arg[arg].type.name);
6376
6377 if (pp->arg[arg].reg != NULL) {
6378 if (pp->arg[arg].type.is_retreg)
6379 fprintf(fout, "&%s", pp->arg[arg].reg);
6380 else
6381 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
6382 continue;
6383 }
6384
6385 // stack arg
6386 tmp_op = pp->arg[arg].datap;
6387 if (tmp_op == NULL)
6388 ferr(po, "parsed_op missing for arg%d\n", arg);
6389
6390 if (tmp_op->flags & OPF_VAPUSH) {
6391 fprintf(fout, "ap");
6392 }
6393 else if (tmp_op->p_argpass != 0) {
6394 fprintf(fout, "a%d", tmp_op->p_argpass);
6395 }
6396 else if (tmp_op->p_argnum != 0) {
6397 fprintf(fout, "%s%s", cast,
6398 saved_arg_name(buf1, sizeof(buf1),
6399 tmp_op->p_arggrp, tmp_op->p_argnum));
6400 }
6401 else {
6402 fprintf(fout, "%s",
6403 out_src_opr(buf1, sizeof(buf1),
6404 tmp_op, &tmp_op->operand[0], cast, 0));
6405 }
6406 }
6407 }
6408 fprintf(fout, ");");
6409
6410 if (strstr(pp->ret_type.name, "int64")) {
6411 fprintf(fout, "\n");
6412 fprintf(fout, "%sedx = tmp64 >> 32;\n", buf3);
6413 fprintf(fout, "%seax = tmp64;", buf3);
6414 }
6415
6416 if (pp->is_unresolved) {
6417 snprintf(buf2, sizeof(buf2), " unresolved %dreg",
6418 pp->argc_reg);
6419 strcat(g_comment, buf2);
6420 }
6421
6422 if (po->flags & OPF_TAIL) {
6423 ret = 0;
6424 if (i == opcnt - 1 || pp->is_noreturn)
6425 ret = 0;
6426 else if (IS(pp->ret_type.name, "void"))
6427 ret = 1;
6428 else if (!(regmask_ret & (1 << xAX)))
6429 ret = 1;
6430 // else already handled as 'return f()'
6431
6432 if (ret) {
6433 fprintf(fout, "\n%sreturn;", buf3);
6434 strcat(g_comment, " ^ tailcall");
6435 }
6436 else
6437 strcat(g_comment, " tailcall");
6438
6439 if ((regmask_ret & (1 << xAX))
6440 && IS(pp->ret_type.name, "void") && !pp->is_noreturn)
6441 {
6442 ferr(po, "int func -> void func tailcall?\n");
6443 }
6444 }
6445 if (pp->is_noreturn)
6446 strcat(g_comment, " noreturn");
6447 if ((po->flags & OPF_ATAIL) && pp->argc_stack > 0)
6448 strcat(g_comment, " argframe");
6449 if (po->flags & OPF_CC)
6450 strcat(g_comment, " cond");
6451
6452 if (po->flags & OPF_CC)
6453 fprintf(fout, "\n }");
6454
6455 delayed_flag_op = NULL;
6456 last_arith_dst = NULL;
6457 break;
6458
6459 case OP_RET:
6460 if (g_func_pp->is_vararg)
6461 fprintf(fout, " va_end(ap);\n");
6462 if (g_func_pp->has_retreg) {
6463 for (arg = 0; arg < g_func_pp->argc; arg++)
6464 if (g_func_pp->arg[arg].type.is_retreg)
6465 fprintf(fout, " *r_%s = %s;\n",
6466 g_func_pp->arg[arg].reg, g_func_pp->arg[arg].reg);
6467 }
6468
6469 if (!(regmask_ret & (1 << xAX))) {
6470 if (i != opcnt - 1 || label_pending)
6471 fprintf(fout, " return;");
6472 }
6473 else if (g_func_pp->ret_type.is_ptr) {
6474 fprintf(fout, " return (%s)eax;",
6475 g_func_pp->ret_type.name);
6476 }
6477 else if (IS(g_func_pp->ret_type.name, "__int64"))
6478 fprintf(fout, " return ((u64)edx << 32) | eax;");
6479 else
6480 fprintf(fout, " return eax;");
6481
6482 last_arith_dst = NULL;
6483 delayed_flag_op = NULL;
6484 break;
6485
6486 case OP_PUSH:
6487 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
6488 if (po->p_argnum != 0) {
6489 // special case - saved func arg
6490 fprintf(fout, " %s = %s;",
6491 saved_arg_name(buf2, sizeof(buf2),
6492 po->p_arggrp, po->p_argnum), buf1);
6493 break;
6494 }
6495 else if (po->flags & OPF_RSAVE) {
6496 fprintf(fout, " s_%s = %s;", buf1, buf1);
6497 break;
6498 }
6499 else if (po->flags & OPF_PPUSH) {
6500 tmp_op = po->datap;
6501 ferr_assert(po, tmp_op != NULL);
6502 out_dst_opr(buf2, sizeof(buf2), po, &tmp_op->operand[0]);
6503 fprintf(fout, " pp_%s = %s;", buf2, buf1);
6504 break;
6505 }
6506 else if (g_func_pp->is_userstack) {
6507 fprintf(fout, " *(--esp) = %s;", buf1);
6508 break;
6509 }
6510 if (!(g_ida_func_attr & IDAFA_NORETURN))
6511 ferr(po, "stray push encountered\n");
6512 no_output = 1;
6513 break;
6514
6515 case OP_POP:
6516 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
6517 if (po->flags & OPF_RSAVE) {
6518 fprintf(fout, " %s = s_%s;", buf1, buf1);
6519 break;
6520 }
6521 else if (po->flags & OPF_PPUSH) {
6522 // push/pop graph / non-const
6523 ferr_assert(po, po->datap == NULL);
6524 fprintf(fout, " %s = pp_%s;", buf1, buf1);
6525 break;
6526 }
6527 else if (po->datap != NULL) {
6528 // push/pop pair
6529 tmp_op = po->datap;
6530 fprintf(fout, " %s = %s;", buf1,
6531 out_src_opr(buf2, sizeof(buf2),
6532 tmp_op, &tmp_op->operand[0],
6533 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
6534 break;
6535 }
6536 else if (g_func_pp->is_userstack) {
6537 fprintf(fout, " %s = *esp++;", buf1);
6538 break;
6539 }
6540 else
6541 ferr(po, "stray pop encountered\n");
6542 break;
6543
6544 case OP_NOP:
6545 no_output = 1;
6546 break;
6547
6548 // x87
6549 case OP_FLD:
6550 if (po->flags & OPF_FSHIFT)
6551 fprintf(fout, " f_st1 = f_st0;\n");
6552 if (po->operand[0].type == OPT_REG
6553 && po->operand[0].reg == xST0)
6554 {
6555 strcat(g_comment, " fld st");
6556 break;
6557 }
6558 fprintf(fout, " f_st0 = %s;",
6559 out_src_opr_float(buf1, sizeof(buf1), po, &po->operand[0]));
6560 strcat(g_comment, " fld");
6561 break;
6562
6563 case OP_FILD:
6564 if (po->flags & OPF_FSHIFT)
6565 fprintf(fout, " f_st1 = f_st0;\n");
6566 fprintf(fout, " f_st0 = (double)%s;",
6567 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
6568 lmod_cast(po, po->operand[0].lmod, 1), 0));
6569 strcat(g_comment, " fild");
6570 break;
6571
6572 case OP_FLDc:
6573 if (po->flags & OPF_FSHIFT)
6574 fprintf(fout, " f_st1 = f_st0;\n");
6575 fprintf(fout, " f_st0 = ");
6576 switch (po->operand[0].val) {
6577 case X87_CONST_1: fprintf(fout, "1.0;"); break;
6578 case X87_CONST_Z: fprintf(fout, "0.0;"); break;
6579 default: ferr(po, "TODO\n"); break;
6580 }
6581 break;
6582
6583 case OP_FST:
6584 if ((po->flags & OPF_FPOP) && po->operand[0].type == OPT_REG
6585 && po->operand[0].reg == xST0)
6586 {
6587 no_output = 1;
6588 break;
6589 }
6590 fprintf(fout, " %s = f_st0;",
6591 out_dst_opr_float(buf1, sizeof(buf1), po, &po->operand[0]));
6592 if (po->flags & OPF_FSHIFT)
6593 fprintf(fout, "\n f_st0 = f_st1;");
6594 strcat(g_comment, " fst");
6595 break;
6596
6597 case OP_FADD:
6598 case OP_FDIV:
6599 case OP_FMUL:
6600 case OP_FSUB:
6601 switch (po->op) {
6602 case OP_FADD: j = '+'; break;
6603 case OP_FDIV: j = '/'; break;
6604 case OP_FMUL: j = '*'; break;
6605 case OP_FSUB: j = '-'; break;
6606 default: j = 'x'; break;
6607 }
6608 if (po->flags & OPF_FSHIFT) {
6609 fprintf(fout, " f_st0 = f_st1 %c f_st0;", j);
6610 }
6611 else {
6612 fprintf(fout, " %s %c= %s;",
6613 out_dst_opr_float(buf1, sizeof(buf1), po, &po->operand[0]),
6614 j,
6615 out_src_opr_float(buf2, sizeof(buf2), po, &po->operand[1]));
6616 }
6617 break;
6618
6619 case OP_FDIVR:
6620 case OP_FSUBR:
6621 if (po->flags & OPF_FSHIFT)
6622 snprintf(buf1, sizeof(buf1), "f_st0");
6623 else
6624 out_dst_opr_float(buf1, sizeof(buf1), po, &po->operand[0]);
6625 fprintf(fout, " %s = %s %c %s;", buf1,
6626 out_src_opr_float(buf2, sizeof(buf2), po, &po->operand[1]),
6627 po->op == OP_FDIVR ? '/' : '-',
6628 out_src_opr_float(buf3, sizeof(buf3), po, &po->operand[0]));
6629 break;
6630
6631 case OP_FIADD:
6632 case OP_FIDIV:
6633 case OP_FIMUL:
6634 case OP_FISUB:
6635 switch (po->op) {
6636 case OP_FIADD: j = '+'; break;
6637 case OP_FIDIV: j = '/'; break;
6638 case OP_FIMUL: j = '*'; break;
6639 case OP_FISUB: j = '-'; break;
6640 default: j = 'x'; break;
6641 }
6642 fprintf(fout, " f_st0 %c= (double)%s;", j,
6643 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
6644 lmod_cast(po, po->operand[0].lmod, 1), 0));
6645 break;
6646
6647 case OP_FIDIVR:
6648 case OP_FISUBR:
6649 fprintf(fout, " f_st0 = %s %c f_st0;",
6650 out_src_opr_float(buf2, sizeof(buf2), po, &po->operand[1]),
6651 po->op == OP_FIDIVR ? '/' : '-');
6652 break;
6653
6654 case OPP_FTOL:
6655 ferr_assert(po, po->flags & OPF_32BIT);
6656 fprintf(fout, " eax = (s32)f_st0;");
6657 if (po->flags & OPF_FSHIFT)
6658 fprintf(fout, "\n f_st0 = f_st1;");
6659 strcat(g_comment, " ftol");
6660 break;
6661
6662 // mmx
6663 case OP_EMMS:
6664 strcpy(g_comment, " (emms)");
6665 break;
6666
6667 default:
6668 no_output = 1;
6669 ferr(po, "unhandled op type %d, flags %x\n",
6670 po->op, po->flags);
6671 break;
6672 }
6673
6674 if (g_comment[0] != 0) {
6675 char *p = g_comment;
6676 while (my_isblank(*p))
6677 p++;
6678 fprintf(fout, " // %s", p);
6679 g_comment[0] = 0;
6680 no_output = 0;
6681 }
6682 if (!no_output)
6683 fprintf(fout, "\n");
6684
6685 // some sanity checking
6686 if (po->flags & OPF_REP) {
6687 if (po->op != OP_STOS && po->op != OP_MOVS
6688 && po->op != OP_CMPS && po->op != OP_SCAS)
6689 ferr(po, "unexpected rep\n");
6690 if (!(po->flags & (OPF_REPZ|OPF_REPNZ))
6691 && (po->op == OP_CMPS || po->op == OP_SCAS))
6692 ferr(po, "cmps/scas with plain rep\n");
6693 }
6694 if ((po->flags & (OPF_REPZ|OPF_REPNZ))
6695 && po->op != OP_CMPS && po->op != OP_SCAS)
6696 ferr(po, "unexpected repz/repnz\n");
6697
6698 if (pfomask != 0)
6699 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
6700
6701 // see is delayed flag stuff is still valid
6702 if (delayed_flag_op != NULL && delayed_flag_op != po) {
6703 if (is_any_opr_modified(delayed_flag_op, po, 0))
6704 delayed_flag_op = NULL;
6705 }
6706
6707 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
6708 if (is_opr_modified(last_arith_dst, po))
6709 last_arith_dst = NULL;
6710 }
6711
6712 label_pending = 0;
6713 }
6714
6715 if (g_stack_fsz && !g_stack_frame_used)
6716 fprintf(fout, " (void)sf;\n");
6717
6718 fprintf(fout, "}\n\n");
6719
6720 gen_x_cleanup(opcnt);
6721}
6722
6723static void gen_x_cleanup(int opcnt)
6724{
6725 int i;
6726
6727 for (i = 0; i < opcnt; i++) {
6728 struct label_ref *lr, *lr_del;
6729
6730 lr = g_label_refs[i].next;
6731 while (lr != NULL) {
6732 lr_del = lr;
6733 lr = lr->next;
6734 free(lr_del);
6735 }
6736 g_label_refs[i].i = -1;
6737 g_label_refs[i].next = NULL;
6738
6739 if (ops[i].op == OP_CALL) {
6740 if (ops[i].pp)
6741 proto_release(ops[i].pp);
6742 }
6743 }
6744 g_func_pp = NULL;
6745}
6746
6747struct func_proto_dep;
6748
6749struct func_prototype {
6750 char name[NAMELEN];
6751 int id;
6752 int argc_stack;
6753 int regmask_dep;
6754 int has_ret:3; // -1, 0, 1: unresolved, no, yes
6755 unsigned int dep_resolved:1;
6756 unsigned int is_stdcall:1;
6757 struct func_proto_dep *dep_func;
6758 int dep_func_cnt;
6759 const struct parsed_proto *pp; // seed pp, if any
6760};
6761
6762struct func_proto_dep {
6763 char *name;
6764 struct func_prototype *proto;
6765 int regmask_live; // .. at the time of call
6766 unsigned int ret_dep:1; // return from this is caller's return
6767};
6768
6769static struct func_prototype *hg_fp;
6770static int hg_fp_cnt;
6771
6772static struct scanned_var {
6773 char name[NAMELEN];
6774 enum opr_lenmod lmod;
6775 unsigned int is_seeded:1;
6776 unsigned int is_c_str:1;
6777 const struct parsed_proto *pp; // seed pp, if any
6778} *hg_vars;
6779static int hg_var_cnt;
6780
6781static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
6782 int count);
6783
6784struct func_prototype *hg_fp_add(const char *funcn)
6785{
6786 struct func_prototype *fp;
6787
6788 if ((hg_fp_cnt & 0xff) == 0) {
6789 hg_fp = realloc(hg_fp, sizeof(hg_fp[0]) * (hg_fp_cnt + 0x100));
6790 my_assert_not(hg_fp, NULL);
6791 memset(hg_fp + hg_fp_cnt, 0, sizeof(hg_fp[0]) * 0x100);
6792 }
6793
6794 fp = &hg_fp[hg_fp_cnt];
6795 snprintf(fp->name, sizeof(fp->name), "%s", funcn);
6796 fp->id = hg_fp_cnt;
6797 fp->argc_stack = -1;
6798 hg_fp_cnt++;
6799
6800 return fp;
6801}
6802
6803static struct func_proto_dep *hg_fp_find_dep(struct func_prototype *fp,
6804 const char *name)
6805{
6806 int i;
6807
6808 for (i = 0; i < fp->dep_func_cnt; i++)
6809 if (IS(fp->dep_func[i].name, name))
6810 return &fp->dep_func[i];
6811
6812 return NULL;
6813}
6814
6815static void hg_fp_add_dep(struct func_prototype *fp, const char *name)
6816{
6817 // is it a dupe?
6818 if (hg_fp_find_dep(fp, name))
6819 return;
6820
6821 if ((fp->dep_func_cnt & 0xff) == 0) {
6822 fp->dep_func = realloc(fp->dep_func,
6823 sizeof(fp->dep_func[0]) * (fp->dep_func_cnt + 0x100));
6824 my_assert_not(fp->dep_func, NULL);
6825 memset(&fp->dep_func[fp->dep_func_cnt], 0,
6826 sizeof(fp->dep_func[0]) * 0x100);
6827 }
6828 fp->dep_func[fp->dep_func_cnt].name = strdup(name);
6829 fp->dep_func_cnt++;
6830}
6831
6832static int hg_fp_cmp_name(const void *p1_, const void *p2_)
6833{
6834 const struct func_prototype *p1 = p1_, *p2 = p2_;
6835 return strcmp(p1->name, p2->name);
6836}
6837
6838#if 0
6839static int hg_fp_cmp_id(const void *p1_, const void *p2_)
6840{
6841 const struct func_prototype *p1 = p1_, *p2 = p2_;
6842 return p1->id - p2->id;
6843}
6844#endif
6845
6846// recursive register dep pass
6847// - track saved regs (part 2)
6848// - try to figure out arg-regs
6849// - calculate reg deps
6850static void gen_hdr_dep_pass(int i, int opcnt, unsigned char *cbits,
6851 struct func_prototype *fp, int regmask_save, int regmask_dst,
6852 int *regmask_dep, int *has_ret)
6853{
6854 struct func_proto_dep *dep;
6855 struct parsed_op *po;
6856 int from_caller = 0;
6857 int j, l;
6858 int reg;
6859 int ret;
6860
6861 for (; i < opcnt; i++)
6862 {
6863 if (cbits[i >> 3] & (1 << (i & 7)))
6864 return;
6865 cbits[i >> 3] |= (1 << (i & 7));
6866
6867 po = &ops[i];
6868
6869 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
6870 if (po->flags & OPF_RMD)
6871 continue;
6872
6873 if (po->btj != NULL) {
6874 // jumptable
6875 for (j = 0; j < po->btj->count; j++) {
6876 check_i(po, po->btj->d[j].bt_i);
6877 gen_hdr_dep_pass(po->btj->d[j].bt_i, opcnt, cbits, fp,
6878 regmask_save, regmask_dst, regmask_dep, has_ret);
6879 }
6880 return;
6881 }
6882
6883 check_i(po, po->bt_i);
6884 if (po->flags & OPF_CJMP) {
6885 gen_hdr_dep_pass(po->bt_i, opcnt, cbits, fp,
6886 regmask_save, regmask_dst, regmask_dep, has_ret);
6887 }
6888 else {
6889 i = po->bt_i - 1;
6890 }
6891 continue;
6892 }
6893
6894 if (po->flags & OPF_FARG)
6895 /* (just calculate register deps) */;
6896 else if (po->op == OP_PUSH && po->operand[0].type == OPT_REG)
6897 {
6898 reg = po->operand[0].reg;
6899 ferr_assert(po, reg >= 0);
6900
6901 if (po->flags & OPF_RSAVE) {
6902 regmask_save |= 1 << reg;
6903 continue;
6904 }
6905 if (po->flags & OPF_DONE)
6906 continue;
6907
6908 ret = scan_for_pop(i + 1, opcnt, i + opcnt * 2, reg, 0, 0);
6909 if (ret == 1) {
6910 regmask_save |= 1 << reg;
6911 po->flags |= OPF_RMD;
6912 scan_for_pop(i + 1, opcnt, i + opcnt * 3, reg, 0, OPF_RMD);
6913 continue;
6914 }
6915 }
6916 else if (po->flags & OPF_RMD)
6917 continue;
6918 else if (po->op == OP_CALL) {
6919 po->regmask_dst |= 1 << xAX;
6920
6921 dep = hg_fp_find_dep(fp, po->operand[0].name);
6922 if (dep != NULL)
6923 dep->regmask_live = regmask_save | regmask_dst;
6924 }
6925 else if (po->op == OP_RET) {
6926 if (po->operand_cnt > 0) {
6927 fp->is_stdcall = 1;
6928 if (fp->argc_stack >= 0
6929 && fp->argc_stack != po->operand[0].val / 4)
6930 ferr(po, "ret mismatch? (%d)\n", fp->argc_stack * 4);
6931 fp->argc_stack = po->operand[0].val / 4;
6932 }
6933 }
6934
6935 // if has_ret is 0, there is uninitialized eax path,
6936 // which means it's most likely void func
6937 if (*has_ret != 0 && (po->flags & OPF_TAIL)) {
6938 if (po->op == OP_CALL) {
6939 j = i;
6940 ret = 1;
6941 }
6942 else {
6943 struct parsed_opr opr = OPR_INIT(OPT_REG, OPLM_DWORD, xAX);
6944 j = -1;
6945 from_caller = 0;
6946 ret = resolve_origin(i, &opr, i + opcnt * 4, &j, &from_caller);
6947 }
6948
6949 if (ret != 1 && from_caller) {
6950 // unresolved eax - probably void func
6951 *has_ret = 0;
6952 }
6953 else {
6954 if (j >= 0 && ops[j].op == OP_CALL) {
6955 dep = hg_fp_find_dep(fp, ops[j].operand[0].name);
6956 if (dep != NULL)
6957 dep->ret_dep = 1;
6958 else
6959 *has_ret = 1;
6960 }
6961 else
6962 *has_ret = 1;
6963 }
6964 }
6965
6966 l = regmask_save | regmask_dst;
6967 if (g_bp_frame && !(po->flags & OPF_EBP_S))
6968 l |= 1 << xBP;
6969
6970 l = po->regmask_src & ~l;
6971#if 0
6972 if (l)
6973 fnote(po, "dep |= %04x, dst %04x, save %04x (f %x)\n",
6974 l, regmask_dst, regmask_save, po->flags);
6975#endif
6976 *regmask_dep |= l;
6977 regmask_dst |= po->regmask_dst;
6978
6979 if (po->flags & OPF_TAIL)
6980 return;
6981 }
6982}
6983
6984static void gen_hdr(const char *funcn, int opcnt)
6985{
6986 int save_arg_vars[MAX_ARG_GRP] = { 0, };
6987 unsigned char cbits[MAX_OPS / 8];
6988 const struct parsed_proto *pp_c;
6989 struct parsed_proto *pp;
6990 struct func_prototype *fp;
6991 struct parsed_op *po;
6992 int regmask_dummy = 0;
6993 int regmask_dep;
6994 int max_bp_offset = 0;
6995 int has_ret;
6996 int i, j, l;
6997 int ret;
6998
6999 pp_c = proto_parse(g_fhdr, funcn, 1);
7000 if (pp_c != NULL)
7001 // already in seed, will add to hg_fp later
7002 return;
7003
7004 fp = hg_fp_add(funcn);
7005
7006 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
7007 g_stack_frame_used = 0;
7008
7009 // pass1:
7010 // - resolve all branches
7011 // - parse calls with labels
7012 resolve_branches_parse_calls(opcnt);
7013
7014 // pass2:
7015 // - handle ebp/esp frame, remove ops related to it
7016 scan_prologue_epilogue(opcnt);
7017
7018 // pass3:
7019 // - remove dead labels
7020 // - collect calls
7021 for (i = 0; i < opcnt; i++)
7022 {
7023 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
7024 free(g_labels[i]);
7025 g_labels[i] = NULL;
7026 }
7027
7028 po = &ops[i];
7029 if (po->flags & (OPF_RMD|OPF_DONE))
7030 continue;
7031
7032 if (po->op == OP_CALL) {
7033 if (po->operand[0].type == OPT_LABEL)
7034 hg_fp_add_dep(fp, opr_name(po, 0));
7035 else if (po->pp != NULL)
7036 hg_fp_add_dep(fp, po->pp->name);
7037 }
7038 }
7039
7040 // pass4:
7041 // - remove dead labels
7042 // - handle push <const>/pop pairs
7043 for (i = 0; i < opcnt; i++)
7044 {
7045 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
7046 free(g_labels[i]);
7047 g_labels[i] = NULL;
7048 }
7049
7050 po = &ops[i];
7051 if (po->flags & (OPF_RMD|OPF_DONE))
7052 continue;
7053
7054 if (po->op == OP_PUSH && po->operand[0].type == OPT_CONST)
7055 scan_for_pop_const(i, opcnt, i + opcnt * 13);
7056 }
7057
7058 // pass5:
7059 // - process trivial calls
7060 for (i = 0; i < opcnt; i++)
7061 {
7062 po = &ops[i];
7063 if (po->flags & (OPF_RMD|OPF_DONE))
7064 continue;
7065
7066 if (po->op == OP_CALL)
7067 {
7068 pp = process_call_early(i, opcnt, &j);
7069 if (pp != NULL) {
7070 if (!(po->flags & OPF_ATAIL))
7071 // since we know the args, try to collect them
7072 if (collect_call_args_early(po, i, pp, &regmask_dummy) != 0)
7073 pp = NULL;
7074 }
7075
7076 if (pp != NULL) {
7077 if (j >= 0) {
7078 // commit esp adjust
7079 if (ops[j].op != OP_POP)
7080 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
7081 else {
7082 for (l = 0; l < pp->argc_stack; l++)
7083 ops[j + l].flags |= OPF_DONE | OPF_RMD | OPF_NOREGS;
7084 }
7085 }
7086
7087 po->flags |= OPF_DONE;
7088 }
7089 }
7090 }
7091
7092 // pass6:
7093 // - track saved regs (simple)
7094 // - process calls
7095 for (i = 0; i < opcnt; i++)
7096 {
7097 po = &ops[i];
7098 if (po->flags & (OPF_RMD|OPF_DONE))
7099 continue;
7100
7101 if (po->op == OP_PUSH && po->operand[0].type == OPT_REG
7102 && po->operand[0].reg != xCX)
7103 {
7104 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].reg, 0);
7105 if (ret == 1) {
7106 // regmask_save |= 1 << po->operand[0].reg; // do it later
7107 po->flags |= OPF_RSAVE | OPF_RMD | OPF_DONE;
7108 scan_for_pop_ret(i + 1, opcnt, po->operand[0].reg, OPF_RMD);
7109 }
7110 }
7111 else if (po->op == OP_CALL)
7112 {
7113 pp = process_call(i, opcnt);
7114
7115 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
7116 // since we know the args, collect them
7117 ret = collect_call_args(po, i, pp, &regmask_dummy, save_arg_vars,
7118 i + opcnt * 1);
7119 }
7120 }
7121 }
7122
7123 // pass7
7124 memset(cbits, 0, sizeof(cbits));
7125 regmask_dep = 0;
7126 has_ret = -1;
7127
7128 gen_hdr_dep_pass(0, opcnt, cbits, fp, 0, 0, &regmask_dep, &has_ret);
7129
7130 // find unreachable code - must be fixed in IDA
7131 for (i = 0; i < opcnt; i++)
7132 {
7133 if (cbits[i >> 3] & (1 << (i & 7)))
7134 continue;
7135
7136 if (g_labels[i] == NULL && i > 0 && ops[i - 1].op == OP_CALL
7137 && ops[i - 1].pp != NULL && ops[i - 1].pp->is_osinc)
7138 {
7139 // the compiler sometimes still generates code after
7140 // noreturn OS functions
7141 break;
7142 }
7143 if (ops[i].op != OP_NOP)
7144 ferr(&ops[i], "unreachable code\n");
7145 }
7146
7147 for (i = 0; i < g_eqcnt; i++) {
7148 if (g_eqs[i].offset > max_bp_offset && g_eqs[i].offset < 4*32)
7149 max_bp_offset = g_eqs[i].offset;
7150 }
7151
7152 if (fp->argc_stack < 0) {
7153 max_bp_offset = (max_bp_offset + 3) & ~3;
7154 fp->argc_stack = max_bp_offset / 4;
7155 if ((g_ida_func_attr & IDAFA_BP_FRAME) && fp->argc_stack > 0)
7156 fp->argc_stack--;
7157 }
7158
7159 fp->regmask_dep = regmask_dep & ~(1 << xSP);
7160 fp->has_ret = has_ret;
7161#if 0
7162 printf("// has_ret %d, regmask_dep %x\n",
7163 fp->has_ret, fp->regmask_dep);
7164 output_hdr_fp(stdout, fp, 1);
7165 if (IS(funcn, "sub_10007F72")) exit(1);
7166#endif
7167
7168 gen_x_cleanup(opcnt);
7169}
7170
7171static void hg_fp_resolve_deps(struct func_prototype *fp)
7172{
7173 struct func_prototype fp_s;
7174 int dep;
7175 int i;
7176
7177 // this thing is recursive, so mark first..
7178 fp->dep_resolved = 1;
7179
7180 for (i = 0; i < fp->dep_func_cnt; i++) {
7181 strcpy(fp_s.name, fp->dep_func[i].name);
7182 fp->dep_func[i].proto = bsearch(&fp_s, hg_fp, hg_fp_cnt,
7183 sizeof(hg_fp[0]), hg_fp_cmp_name);
7184 if (fp->dep_func[i].proto != NULL) {
7185 if (!fp->dep_func[i].proto->dep_resolved)
7186 hg_fp_resolve_deps(fp->dep_func[i].proto);
7187
7188 dep = ~fp->dep_func[i].regmask_live
7189 & fp->dep_func[i].proto->regmask_dep;
7190 fp->regmask_dep |= dep;
7191 // printf("dep %s %s |= %x\n", fp->name,
7192 // fp->dep_func[i].name, dep);
7193
7194 if (fp->has_ret == -1 && fp->dep_func[i].ret_dep)
7195 fp->has_ret = fp->dep_func[i].proto->has_ret;
7196 }
7197 }
7198}
7199
7200static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
7201 int count)
7202{
7203 const struct parsed_proto *pp;
7204 char *p, namebuf[NAMELEN];
7205 const char *name;
7206 int regmask_dep;
7207 int argc_stack;
7208 int j, arg;
7209
7210 for (; count > 0; count--, fp++) {
7211 if (fp->has_ret == -1)
7212 fprintf(fout, "// ret unresolved\n");
7213#if 0
7214 fprintf(fout, "// dep:");
7215 for (j = 0; j < fp->dep_func_cnt; j++) {
7216 fprintf(fout, " %s/", fp->dep_func[j].name);
7217 if (fp->dep_func[j].proto != NULL)
7218 fprintf(fout, "%04x/%d", fp->dep_func[j].proto->regmask_dep,
7219 fp->dep_func[j].proto->has_ret);
7220 }
7221 fprintf(fout, "\n");
7222#endif
7223
7224 p = strchr(fp->name, '@');
7225 if (p != NULL) {
7226 memcpy(namebuf, fp->name, p - fp->name);
7227 namebuf[p - fp->name] = 0;
7228 name = namebuf;
7229 }
7230 else
7231 name = fp->name;
7232 if (name[0] == '_')
7233 name++;
7234
7235 pp = proto_parse(g_fhdr, name, 1);
7236 if (pp != NULL && pp->is_include)
7237 continue;
7238
7239 if (fp->pp != NULL) {
7240 // part of seed, output later
7241 continue;
7242 }
7243
7244 regmask_dep = fp->regmask_dep;
7245 argc_stack = fp->argc_stack;
7246
7247 fprintf(fout, "%-5s", fp->pp ? fp->pp->ret_type.name :
7248 (fp->has_ret ? "int" : "void"));
7249 if (regmask_dep && (fp->is_stdcall || argc_stack == 0)
7250 && (regmask_dep & ~((1 << xCX) | (1 << xDX))) == 0)
7251 {
7252 fprintf(fout, " __fastcall ");
7253 if (!(regmask_dep & (1 << xDX)) && argc_stack == 0)
7254 argc_stack = 1;
7255 else
7256 argc_stack += 2;
7257 regmask_dep = 0;
7258 }
7259 else if (regmask_dep && !fp->is_stdcall) {
7260 fprintf(fout, "/*__usercall*/ ");
7261 }
7262 else if (regmask_dep) {
7263 fprintf(fout, "/*__userpurge*/ ");
7264 }
7265 else if (fp->is_stdcall)
7266 fprintf(fout, " __stdcall ");
7267 else
7268 fprintf(fout, " __cdecl ");
7269
7270 fprintf(fout, "%s(", name);
7271
7272 arg = 0;
7273 for (j = 0; j < xSP; j++) {
7274 if (regmask_dep & (1 << j)) {
7275 arg++;
7276 if (arg != 1)
7277 fprintf(fout, ", ");
7278 if (fp->pp != NULL)
7279 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
7280 else
7281 fprintf(fout, "int");
7282 fprintf(fout, " a%d/*<%s>*/", arg, regs_r32[j]);
7283 }
7284 }
7285
7286 for (j = 0; j < argc_stack; j++) {
7287 arg++;
7288 if (arg != 1)
7289 fprintf(fout, ", ");
7290 if (fp->pp != NULL) {
7291 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
7292 if (!fp->pp->arg[arg - 1].type.is_ptr)
7293 fprintf(fout, " ");
7294 }
7295 else
7296 fprintf(fout, "int ");
7297 fprintf(fout, "a%d", arg);
7298 }
7299
7300 fprintf(fout, ");\n");
7301 }
7302}
7303
7304static void output_hdr(FILE *fout)
7305{
7306 static const char *lmod_c_names[] = {
7307 [OPLM_UNSPEC] = "???",
7308 [OPLM_BYTE] = "uint8_t",
7309 [OPLM_WORD] = "uint16_t",
7310 [OPLM_DWORD] = "uint32_t",
7311 [OPLM_QWORD] = "uint64_t",
7312 };
7313 const struct scanned_var *var;
7314 struct func_prototype *fp;
7315 char line[256] = { 0, };
7316 char name[256];
7317 int i;
7318
7319 // add stuff from headers
7320 for (i = 0; i < pp_cache_size; i++) {
7321 if (pp_cache[i].is_cinc && !pp_cache[i].is_stdcall)
7322 snprintf(name, sizeof(name), "_%s", pp_cache[i].name);
7323 else
7324 snprintf(name, sizeof(name), "%s", pp_cache[i].name);
7325 fp = hg_fp_add(name);
7326 fp->pp = &pp_cache[i];
7327 fp->argc_stack = fp->pp->argc_stack;
7328 fp->is_stdcall = fp->pp->is_stdcall;
7329 fp->regmask_dep = get_pp_arg_regmask_src(fp->pp);
7330 fp->has_ret = !IS(fp->pp->ret_type.name, "void");
7331 }
7332
7333 // resolve deps
7334 qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_name);
7335 for (i = 0; i < hg_fp_cnt; i++)
7336 hg_fp_resolve_deps(&hg_fp[i]);
7337
7338 // note: messes up .proto ptr, don't use
7339 //qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_id);
7340
7341 // output variables
7342 for (i = 0; i < hg_var_cnt; i++) {
7343 var = &hg_vars[i];
7344
7345 if (var->pp != NULL)
7346 // part of seed
7347 continue;
7348 else if (var->is_c_str)
7349 fprintf(fout, "extern %-8s %s[];", "char", var->name);
7350 else
7351 fprintf(fout, "extern %-8s %s;",
7352 lmod_c_names[var->lmod], var->name);
7353
7354 if (var->is_seeded)
7355 fprintf(fout, " // seeded");
7356 fprintf(fout, "\n");
7357 }
7358
7359 fprintf(fout, "\n");
7360
7361 // output function prototypes
7362 output_hdr_fp(fout, hg_fp, hg_fp_cnt);
7363
7364 // seed passthrough
7365 fprintf(fout, "\n// - seed -\n");
7366
7367 rewind(g_fhdr);
7368 while (fgets(line, sizeof(line), g_fhdr))
7369 fwrite(line, 1, strlen(line), fout);
7370}
7371
7372// '=' needs special treatment
7373// also ' quote
7374static char *next_word_s(char *w, size_t wsize, char *s)
7375{
7376 size_t i;
7377
7378 s = sskip(s);
7379
7380 i = 0;
7381 if (*s == '\'') {
7382 w[0] = s[0];
7383 for (i = 1; i < wsize - 1; i++) {
7384 if (s[i] == 0) {
7385 printf("warning: missing closing quote: \"%s\"\n", s);
7386 break;
7387 }
7388 if (s[i] == '\'')
7389 break;
7390 w[i] = s[i];
7391 }
7392 }
7393
7394 for (; i < wsize - 1; i++) {
7395 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
7396 break;
7397 w[i] = s[i];
7398 }
7399 w[i] = 0;
7400
7401 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
7402 printf("warning: '%s' truncated\n", w);
7403
7404 return s + i;
7405}
7406
7407static int cmpstringp(const void *p1, const void *p2)
7408{
7409 return strcmp(*(char * const *)p1, *(char * const *)p2);
7410}
7411
7412static int is_xref_needed(char *p, char **rlist, int rlist_len)
7413{
7414 char *p2;
7415
7416 p = sskip(p);
7417 if (strstr(p, "..."))
7418 // unable to determine, assume needed
7419 return 1;
7420
7421 if (*p == '.') // .text, .data, ...
7422 // ref from other data or non-function -> no
7423 return 0;
7424
7425 p2 = strpbrk(p, "+:\r\n\x18");
7426 if (p2 != NULL)
7427 *p2 = 0;
7428 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
7429 // referenced from removed code
7430 return 0;
7431
7432 return 1;
7433}
7434
7435static int xrefs_show_need(FILE *fasm, char *p,
7436 char **rlist, int rlist_len)
7437{
7438 int found_need = 0;
7439 char line[256];
7440 long pos;
7441
7442 p = strrchr(p, ';');
7443 if (p != NULL && *p == ';' && IS_START(p + 2, "DATA XREF: ")) {
7444 p += 13;
7445 if (is_xref_needed(p, rlist, rlist_len))
7446 return 1;
7447 }
7448
7449 pos = ftell(fasm);
7450 while (1)
7451 {
7452 if (!my_fgets(line, sizeof(line), fasm))
7453 break;
7454 // non-first line is always indented
7455 if (!my_isblank(line[0]))
7456 break;
7457
7458 // should be no content, just comment
7459 p = sskip(line);
7460 if (*p != ';')
7461 break;
7462
7463 p = strrchr(p, ';');
7464 p += 2;
7465 // it's printed once, but no harm to check again
7466 if (IS_START(p, "DATA XREF: "))
7467 p += 11;
7468
7469 if (is_xref_needed(p, rlist, rlist_len)) {
7470 found_need = 1;
7471 break;
7472 }
7473 }
7474 fseek(fasm, pos, SEEK_SET);
7475 return found_need;
7476}
7477
7478static void scan_variables(FILE *fasm, char **rlist, int rlist_len)
7479{
7480 struct scanned_var *var;
7481 char line[256] = { 0, };
7482 char words[3][256];
7483 char *p = NULL;
7484 int wordc;
7485 int l;
7486
7487 while (!feof(fasm))
7488 {
7489 // skip to next data section
7490 while (my_fgets(line, sizeof(line), fasm))
7491 {
7492 asmln++;
7493
7494 p = sskip(line);
7495 if (*p == 0 || *p == ';')
7496 continue;
7497
7498 p = sskip(next_word_s(words[0], sizeof(words[0]), p));
7499 if (*p == 0 || *p == ';')
7500 continue;
7501
7502 if (*p != 's' || !IS_START(p, "segment para public"))
7503 continue;
7504
7505 break;
7506 }
7507
7508 if (p == NULL || !IS_START(p, "segment para public"))
7509 break;
7510 p = sskip(p + 19);
7511
7512 if (!IS_START(p, "'DATA'"))
7513 continue;
7514
7515 // now process it
7516 while (my_fgets(line, sizeof(line), fasm))
7517 {
7518 asmln++;
7519
7520 p = line;
7521 if (my_isblank(*p))
7522 continue;
7523
7524 p = sskip(p);
7525 if (*p == 0 || *p == ';')
7526 continue;
7527
7528 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
7529 words[wordc][0] = 0;
7530 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
7531 if (*p == 0 || *p == ';') {
7532 wordc++;
7533 break;
7534 }
7535 }
7536
7537 if (wordc == 2 && IS(words[1], "ends"))
7538 break;
7539 if (wordc < 2)
7540 continue;
7541
7542 if (IS_START(words[0], "__IMPORT_DESCRIPTOR_")) {
7543 // when this starts, we don't need anything from this section
7544 break;
7545 }
7546
7547 // check refs comment(s)
7548 if (!xrefs_show_need(fasm, p, rlist, rlist_len))
7549 continue;
7550
7551 if ((hg_var_cnt & 0xff) == 0) {
7552 hg_vars = realloc(hg_vars, sizeof(hg_vars[0])
7553 * (hg_var_cnt + 0x100));
7554 my_assert_not(hg_vars, NULL);
7555 memset(hg_vars + hg_var_cnt, 0, sizeof(hg_vars[0]) * 0x100);
7556 }
7557
7558 var = &hg_vars[hg_var_cnt++];
7559 snprintf(var->name, sizeof(var->name), "%s", words[0]);
7560
7561 // maybe already in seed header?
7562 var->pp = proto_parse(g_fhdr, var->name, 1);
7563 if (var->pp != NULL) {
7564 if (var->pp->is_fptr) {
7565 var->lmod = OPLM_DWORD;
7566 //var->is_ptr = 1;
7567 }
7568 else if (var->pp->is_func)
7569 aerr("func?\n");
7570 else if (!guess_lmod_from_c_type(&var->lmod, &var->pp->type))
7571 aerr("unhandled C type '%s' for '%s'\n",
7572 var->pp->type.name, var->name);
7573
7574 var->is_seeded = 1;
7575 continue;
7576 }
7577
7578 if (IS(words[1], "dd"))
7579 var->lmod = OPLM_DWORD;
7580 else if (IS(words[1], "dw"))
7581 var->lmod = OPLM_WORD;
7582 else if (IS(words[1], "db")) {
7583 var->lmod = OPLM_BYTE;
7584 if (wordc >= 3 && (l = strlen(words[2])) > 4) {
7585 if (words[2][0] == '\'' && IS(words[2] + l - 2, ",0"))
7586 var->is_c_str = 1;
7587 }
7588 }
7589 else if (IS(words[1], "dq"))
7590 var->lmod = OPLM_QWORD;
7591 //else if (IS(words[1], "dt"))
7592 else
7593 aerr("type '%s' not known\n", words[1]);
7594 }
7595 }
7596
7597 rewind(fasm);
7598 asmln = 0;
7599}
7600
7601static void set_label(int i, const char *name)
7602{
7603 const char *p;
7604 int len;
7605
7606 len = strlen(name);
7607 p = strchr(name, ':');
7608 if (p != NULL)
7609 len = p - name;
7610
7611 if (g_labels[i] != NULL && !IS_START(g_labels[i], "algn_"))
7612 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
7613 g_labels[i] = realloc(g_labels[i], len + 1);
7614 my_assert_not(g_labels[i], NULL);
7615 memcpy(g_labels[i], name, len);
7616 g_labels[i][len] = 0;
7617}
7618
7619struct chunk_item {
7620 char *name;
7621 long fptr;
7622 int asmln;
7623};
7624
7625static struct chunk_item *func_chunks;
7626static int func_chunk_cnt;
7627static int func_chunk_alloc;
7628
7629static void add_func_chunk(FILE *fasm, const char *name, int line)
7630{
7631 if (func_chunk_cnt >= func_chunk_alloc) {
7632 func_chunk_alloc *= 2;
7633 func_chunks = realloc(func_chunks,
7634 func_chunk_alloc * sizeof(func_chunks[0]));
7635 my_assert_not(func_chunks, NULL);
7636 }
7637 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
7638 func_chunks[func_chunk_cnt].name = strdup(name);
7639 func_chunks[func_chunk_cnt].asmln = line;
7640 func_chunk_cnt++;
7641}
7642
7643static int cmp_chunks(const void *p1, const void *p2)
7644{
7645 const struct chunk_item *c1 = p1, *c2 = p2;
7646 return strcmp(c1->name, c2->name);
7647}
7648
7649static void scan_ahead(FILE *fasm)
7650{
7651 char words[2][256];
7652 char line[256];
7653 long oldpos;
7654 int oldasmln;
7655 int wordc;
7656 char *p;
7657 int i;
7658
7659 oldpos = ftell(fasm);
7660 oldasmln = asmln;
7661
7662 while (my_fgets(line, sizeof(line), fasm))
7663 {
7664 wordc = 0;
7665 asmln++;
7666
7667 p = sskip(line);
7668 if (*p == 0)
7669 continue;
7670
7671 if (*p == ';')
7672 {
7673 // get rid of random tabs
7674 for (i = 0; line[i] != 0; i++)
7675 if (line[i] == '\t')
7676 line[i] = ' ';
7677
7678 if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
7679 {
7680 p += 30;
7681 next_word(words[0], sizeof(words[0]), p);
7682 if (words[0][0] == 0)
7683 aerr("missing name for func chunk?\n");
7684
7685 add_func_chunk(fasm, words[0], asmln);
7686 }
7687 else if (IS_START(p, "; sctend"))
7688 break;
7689
7690 continue;
7691 } // *p == ';'
7692
7693 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
7694 words[wordc][0] = 0;
7695 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
7696 if (*p == 0 || *p == ';') {
7697 wordc++;
7698 break;
7699 }
7700 }
7701
7702 if (wordc == 2 && IS(words[1], "ends"))
7703 break;
7704 }
7705
7706 fseek(fasm, oldpos, SEEK_SET);
7707 asmln = oldasmln;
7708}
7709
7710int main(int argc, char *argv[])
7711{
7712 FILE *fout, *fasm, *frlist;
7713 struct parsed_data *pd = NULL;
7714 int pd_alloc = 0;
7715 char **rlist = NULL;
7716 int rlist_len = 0;
7717 int rlist_alloc = 0;
7718 int func_chunks_used = 0;
7719 int func_chunks_sorted = 0;
7720 int func_chunk_i = -1;
7721 long func_chunk_ret = 0;
7722 int func_chunk_ret_ln = 0;
7723 int scanned_ahead = 0;
7724 char line[256];
7725 char words[20][256];
7726 enum opr_lenmod lmod;
7727 char *sctproto = NULL;
7728 int in_func = 0;
7729 int pending_endp = 0;
7730 int skip_func = 0;
7731 int skip_warned = 0;
7732 int eq_alloc;
7733 int verbose = 0;
7734 int multi_seg = 0;
7735 int end = 0;
7736 int arg_out;
7737 int arg;
7738 int pi = 0;
7739 int i, j;
7740 int ret, len;
7741 char *p;
7742 int wordc;
7743
7744 for (arg = 1; arg < argc; arg++) {
7745 if (IS(argv[arg], "-v"))
7746 verbose = 1;
7747 else if (IS(argv[arg], "-rf"))
7748 g_allow_regfunc = 1;
7749 else if (IS(argv[arg], "-m"))
7750 multi_seg = 1;
7751 else if (IS(argv[arg], "-hdr"))
7752 g_header_mode = g_quiet_pp = g_allow_regfunc = 1;
7753 else
7754 break;
7755 }
7756
7757 if (argc < arg + 3) {
7758 printf("usage:\n%s [-v] [-rf] [-m] <.c> <.asm> <hdr.h> [rlist]*\n"
7759 "%s -hdr <out.h> <.asm> <seed.h> [rlist]*\n"
7760 "options:\n"
7761 " -hdr - header generation mode\n"
7762 " -rf - allow unannotated indirect calls\n"
7763 " -m - allow multiple .text sections\n"
7764 "[rlist] is a file with function names to skip,"
7765 " one per line\n",
7766 argv[0], argv[0]);
7767 return 1;
7768 }
7769
7770 arg_out = arg++;
7771
7772 asmfn = argv[arg++];
7773 fasm = fopen(asmfn, "r");
7774 my_assert_not(fasm, NULL);
7775
7776 hdrfn = argv[arg++];
7777 g_fhdr = fopen(hdrfn, "r");
7778 my_assert_not(g_fhdr, NULL);
7779
7780 rlist_alloc = 64;
7781 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
7782 my_assert_not(rlist, NULL);
7783 // needs special handling..
7784 rlist[rlist_len++] = "__alloca_probe";
7785
7786 func_chunk_alloc = 32;
7787 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
7788 my_assert_not(func_chunks, NULL);
7789
7790 memset(words, 0, sizeof(words));
7791
7792 for (; arg < argc; arg++) {
7793 frlist = fopen(argv[arg], "r");
7794 my_assert_not(frlist, NULL);
7795
7796 while (my_fgets(line, sizeof(line), frlist)) {
7797 p = sskip(line);
7798 if (*p == 0 || *p == ';')
7799 continue;
7800 if (*p == '#') {
7801 if (IS_START(p, "#if 0")
7802 || (g_allow_regfunc && IS_START(p, "#if NO_REGFUNC")))
7803 {
7804 skip_func = 1;
7805 }
7806 else if (IS_START(p, "#endif"))
7807 skip_func = 0;
7808 continue;
7809 }
7810 if (skip_func)
7811 continue;
7812
7813 p = next_word(words[0], sizeof(words[0]), p);
7814 if (words[0][0] == 0)
7815 continue;
7816
7817 if (rlist_len >= rlist_alloc) {
7818 rlist_alloc = rlist_alloc * 2 + 64;
7819 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
7820 my_assert_not(rlist, NULL);
7821 }
7822 rlist[rlist_len++] = strdup(words[0]);
7823 }
7824 skip_func = 0;
7825
7826 fclose(frlist);
7827 frlist = NULL;
7828 }
7829
7830 if (rlist_len > 0)
7831 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
7832
7833 fout = fopen(argv[arg_out], "w");
7834 my_assert_not(fout, NULL);
7835
7836 eq_alloc = 128;
7837 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
7838 my_assert_not(g_eqs, NULL);
7839
7840 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
7841 g_label_refs[i].i = -1;
7842 g_label_refs[i].next = NULL;
7843 }
7844
7845 if (g_header_mode)
7846 scan_variables(fasm, rlist, rlist_len);
7847
7848 while (my_fgets(line, sizeof(line), fasm))
7849 {
7850 wordc = 0;
7851 asmln++;
7852
7853 p = sskip(line);
7854 if (*p == 0)
7855 continue;
7856
7857 // get rid of random tabs
7858 for (i = 0; line[i] != 0; i++)
7859 if (line[i] == '\t')
7860 line[i] = ' ';
7861
7862 if (*p == ';')
7863 {
7864 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
7865 goto do_pending_endp; // eww..
7866
7867 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
7868 {
7869 static const char *attrs[] = {
7870 "bp-based frame",
7871 "library function",
7872 "static",
7873 "noreturn",
7874 "thunk",
7875 "fpd=",
7876 };
7877
7878 // parse IDA's attribute-list comment
7879 g_ida_func_attr = 0;
7880 p = sskip(p + 13);
7881
7882 for (; *p != 0; p = sskip(p)) {
7883 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
7884 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
7885 g_ida_func_attr |= 1 << i;
7886 p += strlen(attrs[i]);
7887 break;
7888 }
7889 }
7890 if (i == ARRAY_SIZE(attrs)) {
7891 anote("unparsed IDA attr: %s\n", p);
7892 break;
7893 }
7894 if (IS(attrs[i], "fpd=")) {
7895 p = next_word(words[0], sizeof(words[0]), p);
7896 // ignore for now..
7897 }
7898 }
7899 }
7900 else if (p[2] == 's' && IS_START(p, "; sctattr:"))
7901 {
7902 static const char *attrs[] = {
7903 "clear_sf",
7904 };
7905
7906 // parse manual attribute-list comment
7907 g_sct_func_attr = 0;
7908 p = sskip(p + 10);
7909
7910 for (; *p != 0; p = sskip(p)) {
7911 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
7912 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
7913 g_sct_func_attr |= 1 << i;
7914 p += strlen(attrs[i]);
7915 break;
7916 }
7917 }
7918 if (i == 0 && *p == '=') {
7919 // clear_sf=start,len (in dwords)
7920 ret = sscanf(p, "=%d,%d%n", &g_stack_clear_start,
7921 &g_stack_clear_len, &j);
7922 if (ret < 2) {
7923 anote("unparsed clear_sf attr value: %s\n", p);
7924 break;
7925 }
7926 p += j;
7927 }
7928 else if (i == ARRAY_SIZE(attrs)) {
7929 anote("unparsed sct attr: %s\n", p);
7930 break;
7931 }
7932 }
7933 }
7934 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
7935 {
7936 p += 30;
7937 next_word(words[0], sizeof(words[0]), p);
7938 if (words[0][0] == 0)
7939 aerr("missing name for func chunk?\n");
7940
7941 if (!scanned_ahead) {
7942 add_func_chunk(fasm, words[0], asmln);
7943 func_chunks_sorted = 0;
7944 }
7945 }
7946 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
7947 {
7948 if (func_chunk_i >= 0) {
7949 if (func_chunk_i < func_chunk_cnt
7950 && IS(func_chunks[func_chunk_i].name, g_func))
7951 {
7952 // move on to next chunk
7953 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
7954 if (ret)
7955 aerr("seek failed for '%s' chunk #%d\n",
7956 g_func, func_chunk_i);
7957 asmln = func_chunks[func_chunk_i].asmln;
7958 func_chunk_i++;
7959 }
7960 else {
7961 if (func_chunk_ret == 0)
7962 aerr("no return from chunk?\n");
7963 fseek(fasm, func_chunk_ret, SEEK_SET);
7964 asmln = func_chunk_ret_ln;
7965 func_chunk_ret = 0;
7966 pending_endp = 1;
7967 }
7968 }
7969 }
7970 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
7971 func_chunks_used = 1;
7972 p += 20;
7973 if (IS_START(g_func, "sub_")) {
7974 unsigned long addr = strtoul(p, NULL, 16);
7975 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
7976 if (addr > f_addr && !scanned_ahead) {
7977 //anote("scan_ahead caused by '%s', addr %lx\n",
7978 // g_func, addr);
7979 scan_ahead(fasm);
7980 scanned_ahead = 1;
7981 func_chunks_sorted = 0;
7982 }
7983 }
7984 }
7985 continue;
7986 } // *p == ';'
7987
7988parse_words:
7989 for (i = wordc; i < ARRAY_SIZE(words); i++)
7990 words[i][0] = 0;
7991 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
7992 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
7993 if (*p == 0 || *p == ';') {
7994 wordc++;
7995 break;
7996 }
7997 }
7998 if (*p != 0 && *p != ';')
7999 aerr("too many words\n");
8000
8001 // alow asm patches in comments
8002 if (*p == ';') {
8003 if (IS_START(p, "; sctpatch:")) {
8004 p = sskip(p + 11);
8005 if (*p == 0 || *p == ';')
8006 continue;
8007 goto parse_words; // lame
8008 }
8009 if (IS_START(p, "; sctproto:")) {
8010 sctproto = strdup(p + 11);
8011 }
8012 else if (IS_START(p, "; sctend")) {
8013 end = 1;
8014 if (!pending_endp)
8015 break;
8016 }
8017 }
8018
8019 if (wordc == 0) {
8020 // shouldn't happen
8021 awarn("wordc == 0?\n");
8022 continue;
8023 }
8024
8025 // don't care about this:
8026 if (words[0][0] == '.'
8027 || IS(words[0], "include")
8028 || IS(words[0], "assume") || IS(words[1], "segment")
8029 || IS(words[0], "align"))
8030 {
8031 continue;
8032 }
8033
8034do_pending_endp:
8035 // do delayed endp processing to collect switch jumptables
8036 if (pending_endp) {
8037 if (in_func && !g_skip_func && !end && wordc >= 2
8038 && ((words[0][0] == 'd' && words[0][2] == 0)
8039 || (words[1][0] == 'd' && words[1][2] == 0)))
8040 {
8041 i = 1;
8042 if (words[1][0] == 'd' && words[1][2] == 0) {
8043 // label
8044 if (g_func_pd_cnt >= pd_alloc) {
8045 pd_alloc = pd_alloc * 2 + 16;
8046 g_func_pd = realloc(g_func_pd,
8047 sizeof(g_func_pd[0]) * pd_alloc);
8048 my_assert_not(g_func_pd, NULL);
8049 }
8050 pd = &g_func_pd[g_func_pd_cnt];
8051 g_func_pd_cnt++;
8052 memset(pd, 0, sizeof(*pd));
8053 strcpy(pd->label, words[0]);
8054 pd->type = OPT_CONST;
8055 pd->lmod = lmod_from_directive(words[1]);
8056 i = 2;
8057 }
8058 else {
8059 if (pd == NULL) {
8060 if (verbose)
8061 anote("skipping alignment byte?\n");
8062 continue;
8063 }
8064 lmod = lmod_from_directive(words[0]);
8065 if (lmod != pd->lmod)
8066 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
8067 }
8068
8069 if (pd->count_alloc < pd->count + wordc) {
8070 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
8071 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
8072 my_assert_not(pd->d, NULL);
8073 }
8074 for (; i < wordc; i++) {
8075 if (IS(words[i], "offset")) {
8076 pd->type = OPT_OFFSET;
8077 i++;
8078 }
8079 p = strchr(words[i], ',');
8080 if (p != NULL)
8081 *p = 0;
8082 if (pd->type == OPT_OFFSET)
8083 pd->d[pd->count].u.label = strdup(words[i]);
8084 else
8085 pd->d[pd->count].u.val = parse_number(words[i]);
8086 pd->d[pd->count].bt_i = -1;
8087 pd->count++;
8088 }
8089 continue;
8090 }
8091
8092 if (in_func && !g_skip_func) {
8093 if (g_header_mode)
8094 gen_hdr(g_func, pi);
8095 else
8096 gen_func(fout, g_fhdr, g_func, pi);
8097 }
8098
8099 pending_endp = 0;
8100 in_func = 0;
8101 g_ida_func_attr = 0;
8102 g_sct_func_attr = 0;
8103 g_stack_clear_start = 0;
8104 g_stack_clear_len = 0;
8105 skip_warned = 0;
8106 g_skip_func = 0;
8107 g_func[0] = 0;
8108 func_chunks_used = 0;
8109 func_chunk_i = -1;
8110 if (pi != 0) {
8111 memset(&ops, 0, pi * sizeof(ops[0]));
8112 clear_labels(pi);
8113 pi = 0;
8114 }
8115 g_eqcnt = 0;
8116 for (i = 0; i < g_func_pd_cnt; i++) {
8117 pd = &g_func_pd[i];
8118 if (pd->type == OPT_OFFSET) {
8119 for (j = 0; j < pd->count; j++)
8120 free(pd->d[j].u.label);
8121 }
8122 free(pd->d);
8123 pd->d = NULL;
8124 }
8125 g_func_pd_cnt = 0;
8126 g_func_lmods = 0;
8127 pd = NULL;
8128
8129 if (end)
8130 break;
8131 if (wordc == 0)
8132 continue;
8133 }
8134
8135 if (IS(words[1], "proc")) {
8136 if (in_func)
8137 aerr("proc '%s' while in_func '%s'?\n",
8138 words[0], g_func);
8139 p = words[0];
8140 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
8141 g_skip_func = 1;
8142 strcpy(g_func, words[0]);
8143 set_label(0, words[0]);
8144 in_func = 1;
8145 continue;
8146 }
8147
8148 if (IS(words[1], "endp"))
8149 {
8150 if (!in_func)
8151 aerr("endp '%s' while not in_func?\n", words[0]);
8152 if (!IS(g_func, words[0]))
8153 aerr("endp '%s' while in_func '%s'?\n",
8154 words[0], g_func);
8155
8156 if ((g_ida_func_attr & IDAFA_THUNK) && pi == 1
8157 && ops[0].op == OP_JMP && ops[0].operand[0].had_ds)
8158 {
8159 // import jump
8160 g_skip_func = 1;
8161 }
8162
8163 if (!g_skip_func && func_chunks_used) {
8164 // start processing chunks
8165 struct chunk_item *ci, key = { g_func, 0 };
8166
8167 func_chunk_ret = ftell(fasm);
8168 func_chunk_ret_ln = asmln;
8169 if (!func_chunks_sorted) {
8170 qsort(func_chunks, func_chunk_cnt,
8171 sizeof(func_chunks[0]), cmp_chunks);
8172 func_chunks_sorted = 1;
8173 }
8174 ci = bsearch(&key, func_chunks, func_chunk_cnt,
8175 sizeof(func_chunks[0]), cmp_chunks);
8176 if (ci == NULL)
8177 aerr("'%s' needs chunks, but none found\n", g_func);
8178 func_chunk_i = ci - func_chunks;
8179 for (; func_chunk_i > 0; func_chunk_i--)
8180 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
8181 break;
8182
8183 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
8184 if (ret)
8185 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
8186 asmln = func_chunks[func_chunk_i].asmln;
8187 func_chunk_i++;
8188 continue;
8189 }
8190 pending_endp = 1;
8191 continue;
8192 }
8193
8194 if (wordc == 2 && IS(words[1], "ends")) {
8195 if (!multi_seg) {
8196 end = 1;
8197 if (pending_endp)
8198 goto do_pending_endp;
8199 break;
8200 }
8201
8202 // scan for next text segment
8203 while (my_fgets(line, sizeof(line), fasm)) {
8204 asmln++;
8205 p = sskip(line);
8206 if (*p == 0 || *p == ';')
8207 continue;
8208
8209 if (strstr(p, "segment para public 'CODE' use32"))
8210 break;
8211 }
8212
8213 continue;
8214 }
8215
8216 p = strchr(words[0], ':');
8217 if (p != NULL) {
8218 set_label(pi, words[0]);
8219 continue;
8220 }
8221
8222 if (!in_func || g_skip_func) {
8223 if (!skip_warned && !g_skip_func && g_labels[pi] != NULL) {
8224 if (verbose)
8225 anote("skipping from '%s'\n", g_labels[pi]);
8226 skip_warned = 1;
8227 }
8228 free(g_labels[pi]);
8229 g_labels[pi] = NULL;
8230 continue;
8231 }
8232
8233 if (wordc > 1 && IS(words[1], "="))
8234 {
8235 if (wordc != 5)
8236 aerr("unhandled equ, wc=%d\n", wordc);
8237 if (g_eqcnt >= eq_alloc) {
8238 eq_alloc *= 2;
8239 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
8240 my_assert_not(g_eqs, NULL);
8241 }
8242
8243 len = strlen(words[0]);
8244 if (len > sizeof(g_eqs[0].name) - 1)
8245 aerr("equ name too long: %d\n", len);
8246 strcpy(g_eqs[g_eqcnt].name, words[0]);
8247
8248 if (!IS(words[3], "ptr"))
8249 aerr("unhandled equ\n");
8250 if (IS(words[2], "dword"))
8251 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
8252 else if (IS(words[2], "word"))
8253 g_eqs[g_eqcnt].lmod = OPLM_WORD;
8254 else if (IS(words[2], "byte"))
8255 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
8256 else if (IS(words[2], "qword"))
8257 g_eqs[g_eqcnt].lmod = OPLM_QWORD;
8258 else
8259 aerr("bad lmod: '%s'\n", words[2]);
8260
8261 g_eqs[g_eqcnt].offset = parse_number(words[4]);
8262 g_eqcnt++;
8263 continue;
8264 }
8265
8266 if (pi >= ARRAY_SIZE(ops))
8267 aerr("too many ops\n");
8268
8269 parse_op(&ops[pi], words, wordc);
8270
8271 ops[pi].datap = sctproto;
8272 sctproto = NULL;
8273 pi++;
8274 }
8275
8276 if (g_header_mode)
8277 output_hdr(fout);
8278
8279 fclose(fout);
8280 fclose(fasm);
8281 fclose(g_fhdr);
8282
8283 return 0;
8284}
8285
8286// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.