notaz.gp2x.de / ia32rtools.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
translate: improve esp adjust yet more
[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
17#define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
18#define IS(w, y) !strcmp(w, y)
19#define IS_START(w, y) !strncmp(w, y, strlen(y))
20
21#include "protoparse.h"
22
23static const char *asmfn;
24static int asmln;
25static FILE *g_fhdr;
26
27#define anote(fmt, ...) \
28 printf("%s:%d: note: " fmt, asmfn, asmln, ##__VA_ARGS__)
29#define awarn(fmt, ...) \
30 printf("%s:%d: warning: " fmt, asmfn, asmln, ##__VA_ARGS__)
31#define aerr(fmt, ...) do { \
32 printf("%s:%d: error: " fmt, asmfn, asmln, ##__VA_ARGS__); \
33 fcloseall(); \
34 exit(1); \
35} while (0)
36
37#include "masm_tools.h"
38
39enum op_flags {
40 OPF_RMD = (1 << 0), /* removed from code generation */
41 OPF_DATA = (1 << 1), /* data processing - writes to dst opr */
42 OPF_FLAGS = (1 << 2), /* sets flags */
43 OPF_JMP = (1 << 3), /* branch, call */
44 OPF_CJMP = (1 << 4), /* cond. branch (cc or jecxz) */
45 OPF_CC = (1 << 5), /* uses flags */
46 OPF_TAIL = (1 << 6), /* ret or tail call */
47 OPF_RSAVE = (1 << 7), /* push/pop is local reg save/load */
48 OPF_REP = (1 << 8), /* prefixed by rep */
49 OPF_REPZ = (1 << 9), /* rep is repe/repz */
50 OPF_REPNZ = (1 << 10), /* rep is repne/repnz */
51 OPF_FARG = (1 << 11), /* push collected as func arg */
52 OPF_FARGNR = (1 << 12), /* push collected as func arg (no reuse) */
53 OPF_EBP_S = (1 << 13), /* ebp used as scratch here, not BP */
54 OPF_DF = (1 << 14), /* DF flag set */
55 OPF_ATAIL = (1 << 15), /* tail call with reused arg frame */
56 OPF_32BIT = (1 << 16), /* 32bit division */
57 OPF_LOCK = (1 << 17), /* op has lock prefix */
58 OPF_VAPUSH = (1 << 18), /* vararg ptr push (as call arg) */
59 OPF_DONE = (1 << 19), /* already fully handled by analysis */
60};
61
62enum op_op {
63 OP_INVAL,
64 OP_NOP,
65 OP_PUSH,
66 OP_POP,
67 OP_LEAVE,
68 OP_MOV,
69 OP_LEA,
70 OP_MOVZX,
71 OP_MOVSX,
72 OP_XCHG,
73 OP_NOT,
74 OP_CDQ,
75 OP_LODS,
76 OP_STOS,
77 OP_MOVS,
78 OP_CMPS,
79 OP_SCAS,
80 OP_STD,
81 OP_CLD,
82 OP_RET,
83 OP_ADD,
84 OP_SUB,
85 OP_AND,
86 OP_OR,
87 OP_XOR,
88 OP_SHL,
89 OP_SHR,
90 OP_SAR,
91 OP_SHRD,
92 OP_ROL,
93 OP_ROR,
94 OP_RCL,
95 OP_RCR,
96 OP_ADC,
97 OP_SBB,
98 OP_BSF,
99 OP_INC,
100 OP_DEC,
101 OP_NEG,
102 OP_MUL,
103 OP_IMUL,
104 OP_DIV,
105 OP_IDIV,
106 OP_TEST,
107 OP_CMP,
108 OP_CALL,
109 OP_JMP,
110 OP_JECXZ,
111 OP_JCC,
112 OP_SCC,
113 // x87
114 // mmx
115 OP_EMMS,
116 // mmx
117 OP_UD2,
118};
119
120enum opr_type {
121 OPT_UNSPEC,
122 OPT_REG,
123 OPT_REGMEM,
124 OPT_LABEL,
125 OPT_OFFSET,
126 OPT_CONST,
127};
128
129// must be sorted (larger len must be further in enum)
130enum opr_lenmod {
131 OPLM_UNSPEC,
132 OPLM_BYTE,
133 OPLM_WORD,
134 OPLM_DWORD,
135 OPLM_QWORD,
136};
137
138#define MAX_OPERANDS 3
139#define NAMELEN 112
140
141struct parsed_opr {
142 enum opr_type type;
143 enum opr_lenmod lmod;
144 unsigned int is_ptr:1; // pointer in C
145 unsigned int is_array:1; // array in C
146 unsigned int type_from_var:1; // .. in header, sometimes wrong
147 unsigned int size_mismatch:1; // type override differs from C
148 unsigned int size_lt:1; // type override is larger than C
149 unsigned int had_ds:1; // had ds: prefix
150 const struct parsed_proto *pp; // for OPT_LABEL
151 int reg;
152 unsigned int val;
153 char name[NAMELEN];
154};
155
156struct parsed_op {
157 enum op_op op;
158 struct parsed_opr operand[MAX_OPERANDS];
159 unsigned int flags;
160 unsigned char pfo;
161 unsigned char pfo_inv;
162 unsigned char operand_cnt;
163 unsigned char p_argnum; // arg push: altered before call arg #
164 unsigned char p_arggrp; // arg push: arg group # for above
165 unsigned char p_argpass;// arg push: arg of host func
166 short p_argnext;// arg push: same arg pushed elsewhere or -1
167 int regmask_src; // all referensed regs
168 int regmask_dst;
169 int pfomask; // flagop: parsed_flag_op that can't be delayed
170 int cc_scratch; // scratch storage during analysis
171 int bt_i; // branch target for branches
172 struct parsed_data *btj;// branch targets for jumptables
173 struct parsed_proto *pp;// parsed_proto for OP_CALL
174 void *datap;
175 int asmln;
176};
177
178// datap:
179// OP_CALL - parser proto hint (str)
180// (OPF_CC) - points to one of (OPF_FLAGS) that affects cc op
181// OP_POP - points to OP_PUSH in push/pop pair
182
183struct parsed_equ {
184 char name[64];
185 enum opr_lenmod lmod;
186 int offset;
187};
188
189struct parsed_data {
190 char label[256];
191 enum opr_type type;
192 enum opr_lenmod lmod;
193 int count;
194 int count_alloc;
195 struct {
196 union {
197 char *label;
198 unsigned int val;
199 } u;
200 int bt_i;
201 } *d;
202};
203
204struct label_ref {
205 int i;
206 struct label_ref *next;
207};
208
209enum ida_func_attr {
210 IDAFA_BP_FRAME = (1 << 0),
211 IDAFA_LIB_FUNC = (1 << 1),
212 IDAFA_STATIC = (1 << 2),
213 IDAFA_NORETURN = (1 << 3),
214 IDAFA_THUNK = (1 << 4),
215 IDAFA_FPD = (1 << 5),
216};
217
218// note: limited to 32k due to p_argnext
219#define MAX_OPS 4096
220#define MAX_ARG_GRP 2
221
222static struct parsed_op ops[MAX_OPS];
223static struct parsed_equ *g_eqs;
224static int g_eqcnt;
225static char *g_labels[MAX_OPS];
226static struct label_ref g_label_refs[MAX_OPS];
227static const struct parsed_proto *g_func_pp;
228static struct parsed_data *g_func_pd;
229static int g_func_pd_cnt;
230static char g_func[256];
231static char g_comment[256];
232static int g_bp_frame;
233static int g_sp_frame;
234static int g_stack_frame_used;
235static int g_stack_fsz;
236static int g_ida_func_attr;
237static int g_allow_regfunc;
238static int g_quiet_pp;
239static int g_header_mode;
240
241#define ferr(op_, fmt, ...) do { \
242 printf("%s:%d: error: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
243 dump_op(op_), ##__VA_ARGS__); \
244 fcloseall(); \
245 exit(1); \
246} while (0)
247#define fnote(op_, fmt, ...) \
248 printf("%s:%d: note: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
249 dump_op(op_), ##__VA_ARGS__)
250
251#define ferr_assert(op_, cond) do { \
252 if (!(cond)) ferr(op_, "assertion '%s' failed on ln :%d\n", #cond, \
253 __LINE__); \
254} while (0)
255
256const char *regs_r32[] = {
257 "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp",
258 // not r32, but list here for easy parsing and printing
259 "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7",
260};
261const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" };
262const char *regs_r8l[] = { "al", "bl", "cl", "dl" };
263const char *regs_r8h[] = { "ah", "bh", "ch", "dh" };
264
265enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP };
266
267// possible basic comparison types (without inversion)
268enum parsed_flag_op {
269 PFO_O, // 0 OF=1
270 PFO_C, // 2 CF=1
271 PFO_Z, // 4 ZF=1
272 PFO_BE, // 6 CF=1||ZF=1
273 PFO_S, // 8 SF=1
274 PFO_P, // a PF=1
275 PFO_L, // c SF!=OF
276 PFO_LE, // e ZF=1||SF!=OF
277};
278
279#define PFOB_O (1 << PFO_O)
280#define PFOB_C (1 << PFO_C)
281#define PFOB_Z (1 << PFO_Z)
282#define PFOB_S (1 << PFO_S)
283
284static const char *parsed_flag_op_names[] = {
285 "o", "c", "z", "be", "s", "p", "l", "le"
286};
287
288static int char_array_i(const char *array[], size_t len, const char *s)
289{
290 int i;
291
292 for (i = 0; i < len; i++)
293 if (IS(s, array[i]))
294 return i;
295
296 return -1;
297}
298
299static void printf_number(char *buf, size_t buf_size,
300 unsigned long number)
301{
302 // output in C-friendly form
303 snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
304}
305
306static int check_segment_prefix(const char *s)
307{
308 if (s[0] == 0 || s[1] != 's' || s[2] != ':')
309 return 0;
310
311 switch (s[0]) {
312 case 'c': return 1;
313 case 'd': return 2;
314 case 's': return 3;
315 case 'e': return 4;
316 case 'f': return 5;
317 case 'g': return 6;
318 default: return 0;
319 }
320}
321
322static int parse_reg(enum opr_lenmod *reg_lmod, const char *s)
323{
324 int reg;
325
326 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s);
327 if (reg >= 8) {
328 *reg_lmod = OPLM_QWORD;
329 return reg;
330 }
331 if (reg >= 0) {
332 *reg_lmod = OPLM_DWORD;
333 return reg;
334 }
335 reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s);
336 if (reg >= 0) {
337 *reg_lmod = OPLM_WORD;
338 return reg;
339 }
340 reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s);
341 if (reg >= 0) {
342 *reg_lmod = OPLM_BYTE;
343 return reg;
344 }
345 reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s);
346 if (reg >= 0) {
347 *reg_lmod = OPLM_BYTE;
348 return reg;
349 }
350
351 return -1;
352}
353
354static int parse_indmode(char *name, int *regmask, int need_c_cvt)
355{
356 enum opr_lenmod lmod;
357 char cvtbuf[256];
358 char *d = cvtbuf;
359 char *s = name;
360 char w[64];
361 long number;
362 int reg;
363 int c = 0;
364
365 *d = 0;
366
367 while (*s != 0) {
368 d += strlen(d);
369 while (my_isblank(*s))
370 s++;
371 for (; my_issep(*s); d++, s++)
372 *d = *s;
373 while (my_isblank(*s))
374 s++;
375 *d = 0;
376
377 // skip '?s:' prefixes
378 if (check_segment_prefix(s))
379 s += 3;
380
381 s = next_idt(w, sizeof(w), s);
382 if (w[0] == 0)
383 break;
384 c++;
385
386 reg = parse_reg(&lmod, w);
387 if (reg >= 0) {
388 *regmask |= 1 << reg;
389 goto pass;
390 }
391
392 if ('0' <= w[0] && w[0] <= '9') {
393 number = parse_number(w);
394 printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number);
395 continue;
396 }
397
398 // probably some label/identifier - pass
399
400pass:
401 snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w);
402 }
403
404 if (need_c_cvt)
405 strcpy(name, cvtbuf);
406
407 return c;
408}
409
410static int is_reg_in_str(const char *s)
411{
412 int i;
413
414 if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
415 return 0;
416
417 for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
418 if (!strncmp(s, regs_r32[i], 3))
419 return 1;
420
421 return 0;
422}
423
424static const char *parse_stack_el(const char *name, char *extra_reg,
425 int early_try)
426{
427 const char *p, *p2, *s;
428 char *endp = NULL;
429 char buf[32];
430 long val;
431 int len;
432
433 if (g_bp_frame || early_try)
434 {
435 p = name;
436 if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
437 p += 4;
438 if (extra_reg != NULL) {
439 strncpy(extra_reg, name, 3);
440 extra_reg[4] = 0;
441 }
442 }
443
444 if (IS_START(p, "ebp+")) {
445 p += 4;
446
447 p2 = strchr(p, '+');
448 if (p2 != NULL && is_reg_in_str(p)) {
449 if (extra_reg != NULL) {
450 strncpy(extra_reg, p, p2 - p);
451 extra_reg[p2 - p] = 0;
452 }
453 p = p2 + 1;
454 }
455
456 if (!('0' <= *p && *p <= '9'))
457 return p;
458
459 return NULL;
460 }
461 }
462
463 if (!IS_START(name, "esp+"))
464 return NULL;
465
466 s = name + 4;
467 p = strchr(s, '+');
468 if (p) {
469 if (is_reg_in_str(s)) {
470 if (extra_reg != NULL) {
471 strncpy(extra_reg, s, p - s);
472 extra_reg[p - s] = 0;
473 }
474 s = p + 1;
475 p = strchr(s, '+');
476 if (p == NULL)
477 aerr("%s IDA stackvar not set?\n", __func__);
478 }
479 if (!('0' <= *s && *s <= '9')) {
480 aerr("%s IDA stackvar offset not set?\n", __func__);
481 return NULL;
482 }
483 if (s[0] == '0' && s[1] == 'x')
484 s += 2;
485 len = p - s;
486 if (len < sizeof(buf) - 1) {
487 strncpy(buf, s, len);
488 buf[len] = 0;
489 val = strtol(buf, &endp, 16);
490 if (val == 0 || *endp != 0) {
491 aerr("%s num parse fail for '%s'\n", __func__, buf);
492 return NULL;
493 }
494 }
495 p++;
496 }
497 else
498 p = name + 4;
499
500 if ('0' <= *p && *p <= '9')
501 return NULL;
502
503 return p;
504}
505
506static int guess_lmod_from_name(struct parsed_opr *opr)
507{
508 if (!strncmp(opr->name, "dword_", 6)) {
509 opr->lmod = OPLM_DWORD;
510 return 1;
511 }
512 if (!strncmp(opr->name, "word_", 5)) {
513 opr->lmod = OPLM_WORD;
514 return 1;
515 }
516 if (!strncmp(opr->name, "byte_", 5)) {
517 opr->lmod = OPLM_BYTE;
518 return 1;
519 }
520 if (!strncmp(opr->name, "qword_", 6)) {
521 opr->lmod = OPLM_QWORD;
522 return 1;
523 }
524 return 0;
525}
526
527static int guess_lmod_from_c_type(enum opr_lenmod *lmod,
528 const struct parsed_type *c_type)
529{
530 static const char *dword_types[] = {
531 "int", "_DWORD", "UINT_PTR", "DWORD",
532 "WPARAM", "LPARAM", "UINT", "__int32",
533 "LONG", "HIMC", "BOOL", "size_t",
534 "float",
535 };
536 static const char *word_types[] = {
537 "uint16_t", "int16_t", "_WORD", "WORD",
538 "unsigned __int16", "__int16",
539 };
540 static const char *byte_types[] = {
541 "uint8_t", "int8_t", "char",
542 "unsigned __int8", "__int8", "BYTE", "_BYTE",
543 "CHAR", "_UNKNOWN",
544 // structures.. deal the same as with _UNKNOWN for now
545 "CRITICAL_SECTION",
546 };
547 const char *n;
548 int i;
549
550 if (c_type->is_ptr) {
551 *lmod = OPLM_DWORD;
552 return 1;
553 }
554
555 n = skip_type_mod(c_type->name);
556
557 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
558 if (IS(n, dword_types[i])) {
559 *lmod = OPLM_DWORD;
560 return 1;
561 }
562 }
563
564 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
565 if (IS(n, word_types[i])) {
566 *lmod = OPLM_WORD;
567 return 1;
568 }
569 }
570
571 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
572 if (IS(n, byte_types[i])) {
573 *lmod = OPLM_BYTE;
574 return 1;
575 }
576 }
577
578 return 0;
579}
580
581static char *default_cast_to(char *buf, size_t buf_size,
582 struct parsed_opr *opr)
583{
584 buf[0] = 0;
585
586 if (!opr->is_ptr)
587 return buf;
588 if (opr->pp == NULL || opr->pp->type.name == NULL
589 || opr->pp->is_fptr)
590 {
591 snprintf(buf, buf_size, "%s", "(void *)");
592 return buf;
593 }
594
595 snprintf(buf, buf_size, "(%s)", opr->pp->type.name);
596 return buf;
597}
598
599static enum opr_type lmod_from_directive(const char *d)
600{
601 if (IS(d, "dd"))
602 return OPLM_DWORD;
603 else if (IS(d, "dw"))
604 return OPLM_WORD;
605 else if (IS(d, "db"))
606 return OPLM_BYTE;
607
608 aerr("unhandled directive: '%s'\n", d);
609 return OPLM_UNSPEC;
610}
611
612static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
613 int *regmask)
614{
615 opr->type = OPT_REG;
616 opr->reg = reg;
617 opr->lmod = lmod;
618 *regmask |= 1 << reg;
619}
620
621static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
622 int *extra_offs);
623
624static int parse_operand(struct parsed_opr *opr,
625 int *regmask, int *regmask_indirect,
626 char words[16][256], int wordc, int w, unsigned int op_flags)
627{
628 const struct parsed_proto *pp = NULL;
629 enum opr_lenmod tmplmod;
630 unsigned long number;
631 char buf[256];
632 int ret, len;
633 int wordc_in;
634 char *p;
635 int i;
636
637 if (w >= wordc)
638 aerr("parse_operand w %d, wordc %d\n", w, wordc);
639
640 opr->reg = xUNSPEC;
641
642 for (i = w; i < wordc; i++) {
643 len = strlen(words[i]);
644 if (words[i][len - 1] == ',') {
645 words[i][len - 1] = 0;
646 wordc = i + 1;
647 break;
648 }
649 }
650
651 wordc_in = wordc - w;
652
653 if ((op_flags & OPF_JMP) && wordc_in > 0
654 && !('0' <= words[w][0] && words[w][0] <= '9'))
655 {
656 const char *label = NULL;
657
658 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
659 && IS(words[w + 1], "ptr"))
660 label = words[w + 2];
661 else if (wordc_in == 2 && IS(words[w], "short"))
662 label = words[w + 1];
663 else if (wordc_in == 1
664 && strchr(words[w], '[') == NULL
665 && parse_reg(&tmplmod, words[w]) < 0)
666 label = words[w];
667
668 if (label != NULL) {
669 opr->type = OPT_LABEL;
670 ret = check_segment_prefix(label);
671 if (ret != 0) {
672 if (ret >= 5)
673 aerr("fs/gs used\n");
674 opr->had_ds = 1;
675 label += 3;
676 }
677 strcpy(opr->name, label);
678 return wordc;
679 }
680 }
681
682 if (wordc_in >= 3) {
683 if (IS(words[w + 1], "ptr")) {
684 if (IS(words[w], "dword"))
685 opr->lmod = OPLM_DWORD;
686 else if (IS(words[w], "word"))
687 opr->lmod = OPLM_WORD;
688 else if (IS(words[w], "byte"))
689 opr->lmod = OPLM_BYTE;
690 else if (IS(words[w], "qword"))
691 opr->lmod = OPLM_QWORD;
692 else
693 aerr("type parsing failed\n");
694 w += 2;
695 wordc_in = wordc - w;
696 }
697 }
698
699 if (wordc_in == 2) {
700 if (IS(words[w], "offset")) {
701 opr->type = OPT_OFFSET;
702 opr->lmod = OPLM_DWORD;
703 strcpy(opr->name, words[w + 1]);
704 pp = proto_parse(g_fhdr, opr->name, 1);
705 goto do_label;
706 }
707 if (IS(words[w], "(offset")) {
708 p = strchr(words[w + 1], ')');
709 if (p == NULL)
710 aerr("parse of bracketed offset failed\n");
711 *p = 0;
712 opr->type = OPT_OFFSET;
713 strcpy(opr->name, words[w + 1]);
714 return wordc;
715 }
716 }
717
718 if (wordc_in != 1)
719 aerr("parse_operand 1 word expected\n");
720
721 ret = check_segment_prefix(words[w]);
722 if (ret != 0) {
723 if (ret >= 5)
724 aerr("fs/gs used\n");
725 opr->had_ds = 1;
726 memmove(words[w], words[w] + 3, strlen(words[w]) - 2);
727 }
728 strcpy(opr->name, words[w]);
729
730 if (words[w][0] == '[') {
731 opr->type = OPT_REGMEM;
732 ret = sscanf(words[w], "[%[^]]]", opr->name);
733 if (ret != 1)
734 aerr("[] parse failure\n");
735
736 parse_indmode(opr->name, regmask_indirect, 1);
737 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL, 1))
738 {
739 // might be an equ
740 struct parsed_equ *eq =
741 equ_find(NULL, parse_stack_el(opr->name, NULL, 1), &i);
742 if (eq)
743 opr->lmod = eq->lmod;
744 }
745 return wordc;
746 }
747 else if (strchr(words[w], '[')) {
748 // label[reg] form
749 p = strchr(words[w], '[');
750 opr->type = OPT_REGMEM;
751 parse_indmode(p, regmask_indirect, 0);
752 strncpy(buf, words[w], p - words[w]);
753 buf[p - words[w]] = 0;
754 pp = proto_parse(g_fhdr, buf, 1);
755 goto do_label;
756 }
757 else if (('0' <= words[w][0] && words[w][0] <= '9')
758 || words[w][0] == '-')
759 {
760 number = parse_number(words[w]);
761 opr->type = OPT_CONST;
762 opr->val = number;
763 printf_number(opr->name, sizeof(opr->name), number);
764 return wordc;
765 }
766
767 ret = parse_reg(&tmplmod, opr->name);
768 if (ret >= 0) {
769 setup_reg_opr(opr, ret, tmplmod, regmask);
770 return wordc;
771 }
772
773 // most likely var in data segment
774 opr->type = OPT_LABEL;
775 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
776
777do_label:
778 if (pp != NULL) {
779 if (pp->is_fptr || pp->is_func) {
780 opr->lmod = OPLM_DWORD;
781 opr->is_ptr = 1;
782 }
783 else {
784 tmplmod = OPLM_UNSPEC;
785 if (!guess_lmod_from_c_type(&tmplmod, &pp->type))
786 anote("unhandled C type '%s' for '%s'\n",
787 pp->type.name, opr->name);
788
789 if (opr->lmod == OPLM_UNSPEC) {
790 opr->lmod = tmplmod;
791 opr->type_from_var = 1;
792 }
793 else if (opr->lmod != tmplmod) {
794 opr->size_mismatch = 1;
795 if (tmplmod < opr->lmod)
796 opr->size_lt = 1;
797 }
798 opr->is_ptr = pp->type.is_ptr;
799 }
800 opr->is_array = pp->type.is_array;
801 }
802 opr->pp = pp;
803
804 if (opr->lmod == OPLM_UNSPEC)
805 guess_lmod_from_name(opr);
806 return wordc;
807}
808
809static const struct {
810 const char *name;
811 unsigned int flags;
812} pref_table[] = {
813 { "rep", OPF_REP },
814 { "repe", OPF_REP|OPF_REPZ },
815 { "repz", OPF_REP|OPF_REPZ },
816 { "repne", OPF_REP|OPF_REPNZ },
817 { "repnz", OPF_REP|OPF_REPNZ },
818 { "lock", OPF_LOCK }, // ignored for now..
819};
820
821#define OPF_CJMP_CC (OPF_JMP|OPF_CJMP|OPF_CC)
822
823static const struct {
824 const char *name;
825 enum op_op op;
826 unsigned short minopr;
827 unsigned short maxopr;
828 unsigned int flags;
829 unsigned char pfo;
830 unsigned char pfo_inv;
831} op_table[] = {
832 { "nop", OP_NOP, 0, 0, 0 },
833 { "push", OP_PUSH, 1, 1, 0 },
834 { "pop", OP_POP, 1, 1, OPF_DATA },
835 { "leave",OP_LEAVE, 0, 0, OPF_DATA },
836 { "mov" , OP_MOV, 2, 2, OPF_DATA },
837 { "lea", OP_LEA, 2, 2, OPF_DATA },
838 { "movzx",OP_MOVZX, 2, 2, OPF_DATA },
839 { "movsx",OP_MOVSX, 2, 2, OPF_DATA },
840 { "xchg", OP_XCHG, 2, 2, OPF_DATA },
841 { "not", OP_NOT, 1, 1, OPF_DATA },
842 { "cdq", OP_CDQ, 0, 0, OPF_DATA },
843 { "lodsb",OP_LODS, 0, 0, OPF_DATA },
844 { "lodsw",OP_LODS, 0, 0, OPF_DATA },
845 { "lodsd",OP_LODS, 0, 0, OPF_DATA },
846 { "stosb",OP_STOS, 0, 0, OPF_DATA },
847 { "stosw",OP_STOS, 0, 0, OPF_DATA },
848 { "stosd",OP_STOS, 0, 0, OPF_DATA },
849 { "movsb",OP_MOVS, 0, 0, OPF_DATA },
850 { "movsw",OP_MOVS, 0, 0, OPF_DATA },
851 { "movsd",OP_MOVS, 0, 0, OPF_DATA },
852 { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
853 { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
854 { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
855 { "scasb",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
856 { "scasw",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
857 { "scasd",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
858 { "std", OP_STD, 0, 0, OPF_DATA }, // special flag
859 { "cld", OP_CLD, 0, 0, OPF_DATA },
860 { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
861 { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
862 { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
863 { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS },
864 { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS },
865 { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
866 { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
867 { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
868 { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
869 { "shrd", OP_SHRD, 3, 3, OPF_DATA|OPF_FLAGS },
870 { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
871 { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
872 { "rcl", OP_RCL, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
873 { "rcr", OP_RCR, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
874 { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
875 { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
876 { "bsf", OP_BSF, 2, 2, OPF_DATA|OPF_FLAGS },
877 { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
878 { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS },
879 { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS },
880 { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS },
881 { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS },
882 { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS },
883 { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS },
884 { "test", OP_TEST, 2, 2, OPF_FLAGS },
885 { "cmp", OP_CMP, 2, 2, OPF_FLAGS },
886 { "retn", OP_RET, 0, 1, OPF_TAIL },
887 { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS },
888 { "jmp", OP_JMP, 1, 1, OPF_JMP },
889 { "jecxz",OP_JECXZ, 1, 1, OPF_JMP|OPF_CJMP },
890 { "jo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 0 }, // 70 OF=1
891 { "jno", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 1 }, // 71 OF=0
892 { "jc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72 CF=1
893 { "jb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72
894 { "jnc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73 CF=0
895 { "jnb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
896 { "jae", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
897 { "jz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74 ZF=1
898 { "je", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74
899 { "jnz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75 ZF=0
900 { "jne", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75
901 { "jbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76 CF=1||ZF=1
902 { "jna", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76
903 { "ja", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77 CF=0&&ZF=0
904 { "jnbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77
905 { "js", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 0 }, // 78 SF=1
906 { "jns", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 1 }, // 79 SF=0
907 { "jp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a PF=1
908 { "jpe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a
909 { "jnp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b PF=0
910 { "jpo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b
911 { "jl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c SF!=OF
912 { "jnge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c
913 { "jge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d SF=OF
914 { "jnl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d
915 { "jle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e ZF=1||SF!=OF
916 { "jng", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e
917 { "jg", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f ZF=0&&SF=OF
918 { "jnle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f
919 { "seto", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 0 },
920 { "setno", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 1 },
921 { "setc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
922 { "setb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
923 { "setnc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
924 { "setae", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
925 { "setnb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
926 { "setz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
927 { "sete", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
928 { "setnz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
929 { "setne", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
930 { "setbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
931 { "setna", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
932 { "seta", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
933 { "setnbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
934 { "sets", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 0 },
935 { "setns", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 1 },
936 { "setp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
937 { "setpe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
938 { "setnp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
939 { "setpo", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
940 { "setl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
941 { "setnge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
942 { "setge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
943 { "setnl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
944 { "setle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
945 { "setng", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
946 { "setg", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
947 { "setnle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
948 // x87
949 // mmx
950 { "emms", OP_EMMS, 0, 0, OPF_DATA },
951 { "movq", OP_MOV, 2, 2, OPF_DATA },
952 // must be last
953 { "ud2", OP_UD2 },
954};
955
956static void parse_op(struct parsed_op *op, char words[16][256], int wordc)
957{
958 enum opr_lenmod lmod = OPLM_UNSPEC;
959 int prefix_flags = 0;
960 int regmask_ind;
961 int regmask;
962 int op_w = 0;
963 int opr = 0;
964 int w = 0;
965 int i;
966
967 for (i = 0; i < ARRAY_SIZE(pref_table); i++) {
968 if (IS(words[w], pref_table[i].name)) {
969 prefix_flags = pref_table[i].flags;
970 break;
971 }
972 }
973
974 if (prefix_flags) {
975 if (wordc <= 1)
976 aerr("lone prefix: '%s'\n", words[0]);
977 w++;
978 }
979
980 op_w = w;
981 for (i = 0; i < ARRAY_SIZE(op_table); i++) {
982 if (IS(words[w], op_table[i].name))
983 break;
984 }
985
986 if (i == ARRAY_SIZE(op_table)) {
987 anote("unhandled op: '%s'\n", words[0]);
988 i--; // OP_UD2
989 }
990 w++;
991
992 op->op = op_table[i].op;
993 op->flags = op_table[i].flags | prefix_flags;
994 op->pfo = op_table[i].pfo;
995 op->pfo_inv = op_table[i].pfo_inv;
996 op->regmask_src = op->regmask_dst = 0;
997 op->asmln = asmln;
998
999 if (op->op == OP_UD2)
1000 return;
1001
1002 for (opr = 0; opr < op_table[i].maxopr; opr++) {
1003 if (opr >= op_table[i].minopr && w >= wordc)
1004 break;
1005
1006 regmask = regmask_ind = 0;
1007 w = parse_operand(&op->operand[opr], &regmask, &regmask_ind,
1008 words, wordc, w, op->flags);
1009
1010 if (opr == 0 && (op->flags & OPF_DATA))
1011 op->regmask_dst = regmask;
1012 else
1013 op->regmask_src |= regmask;
1014 op->regmask_src |= regmask_ind;
1015 }
1016
1017 if (w < wordc)
1018 aerr("parse_op %s incomplete: %d/%d\n",
1019 words[0], w, wordc);
1020
1021 // special cases
1022 op->operand_cnt = opr;
1023 if (!strncmp(op_table[i].name, "set", 3))
1024 op->operand[0].lmod = OPLM_BYTE;
1025
1026 switch (op->op) {
1027 // first operand is not dst
1028 case OP_CMP:
1029 case OP_TEST:
1030 op->regmask_src |= op->regmask_dst;
1031 op->regmask_dst = 0;
1032 break;
1033
1034 // first operand is src too
1035 case OP_NOT:
1036 case OP_ADD:
1037 case OP_AND:
1038 case OP_OR:
1039 case OP_RCL:
1040 case OP_RCR:
1041 case OP_ADC:
1042 case OP_INC:
1043 case OP_DEC:
1044 case OP_NEG:
1045 // more below..
1046 op->regmask_src |= op->regmask_dst;
1047 break;
1048
1049 // special
1050 case OP_XCHG:
1051 op->regmask_src |= op->regmask_dst;
1052 op->regmask_dst |= op->regmask_src;
1053 goto check_align;
1054
1055 case OP_SUB:
1056 case OP_SBB:
1057 case OP_XOR:
1058 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1059 && op->operand[0].lmod == op->operand[1].lmod
1060 && op->operand[0].reg == op->operand[1].reg
1061 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
1062 {
1063 op->regmask_src = 0;
1064 }
1065 else
1066 op->regmask_src |= op->regmask_dst;
1067 break;
1068
1069 // ops with implicit argumets
1070 case OP_CDQ:
1071 op->operand_cnt = 2;
1072 setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst);
1073 setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src);
1074 break;
1075
1076 case OP_LODS:
1077 case OP_STOS:
1078 case OP_SCAS:
1079 if (op->operand_cnt != 0)
1080 break;
1081 if (words[op_w][4] == 'b')
1082 lmod = OPLM_BYTE;
1083 else if (words[op_w][4] == 'w')
1084 lmod = OPLM_WORD;
1085 else if (words[op_w][4] == 'd')
1086 lmod = OPLM_DWORD;
1087 op->operand_cnt = 3;
1088 setup_reg_opr(&op->operand[0], op->op == OP_LODS ? xSI : xDI,
1089 lmod, &op->regmask_src);
1090 setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src);
1091 op->regmask_dst = op->regmask_src;
1092 setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD,
1093 op->op == OP_LODS ? &op->regmask_dst : &op->regmask_src);
1094 break;
1095
1096 case OP_MOVS:
1097 case OP_CMPS:
1098 if (op->operand_cnt != 0)
1099 break;
1100 if (words[op_w][4] == 'b')
1101 lmod = OPLM_BYTE;
1102 else if (words[op_w][4] == 'w')
1103 lmod = OPLM_WORD;
1104 else if (words[op_w][4] == 'd')
1105 lmod = OPLM_DWORD;
1106 op->operand_cnt = 3;
1107 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
1108 setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src);
1109 setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src);
1110 op->regmask_dst = op->regmask_src;
1111 break;
1112
1113 case OP_JECXZ:
1114 op->operand_cnt = 1;
1115 op->regmask_src = 1 << xCX;
1116 op->operand[0].type = OPT_REG;
1117 op->operand[0].reg = xCX;
1118 op->operand[0].lmod = OPLM_DWORD;
1119 break;
1120
1121 case OP_IMUL:
1122 if (op->operand_cnt != 1)
1123 break;
1124 // fallthrough
1125 case OP_MUL:
1126 // singleop mul
1127 op->regmask_src |= op->regmask_dst;
1128 op->regmask_dst = (1 << xDX) | (1 << xAX);
1129 if (op->operand[0].lmod == OPLM_UNSPEC)
1130 op->operand[0].lmod = OPLM_DWORD;
1131 break;
1132
1133 case OP_DIV:
1134 case OP_IDIV:
1135 // we could set up operands for edx:eax, but there is no real need to
1136 // (see is_opr_modified())
1137 op->regmask_src |= op->regmask_dst;
1138 op->regmask_dst = (1 << xDX) | (1 << xAX);
1139 if (op->operand[0].lmod == OPLM_UNSPEC)
1140 op->operand[0].lmod = OPLM_DWORD;
1141 break;
1142
1143 case OP_SHL:
1144 case OP_SHR:
1145 case OP_SAR:
1146 case OP_ROL:
1147 case OP_ROR:
1148 op->regmask_src |= op->regmask_dst;
1149 if (op->operand[1].lmod == OPLM_UNSPEC)
1150 op->operand[1].lmod = OPLM_BYTE;
1151 break;
1152
1153 case OP_SHRD:
1154 op->regmask_src |= op->regmask_dst;
1155 if (op->operand[2].lmod == OPLM_UNSPEC)
1156 op->operand[2].lmod = OPLM_BYTE;
1157 break;
1158
1159 case OP_PUSH:
1160 op->regmask_src |= op->regmask_dst;
1161 op->regmask_dst = 0;
1162 if (op->operand[0].lmod == OPLM_UNSPEC
1163 && (op->operand[0].type == OPT_CONST
1164 || op->operand[0].type == OPT_OFFSET
1165 || op->operand[0].type == OPT_LABEL))
1166 op->operand[0].lmod = OPLM_DWORD;
1167 break;
1168
1169 // alignment
1170 case OP_MOV:
1171 check_align:
1172 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1173 && op->operand[0].lmod == op->operand[1].lmod
1174 && op->operand[0].reg == op->operand[1].reg
1175 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
1176 {
1177 op->flags |= OPF_RMD | OPF_DONE;
1178 op->regmask_src = op->regmask_dst = 0;
1179 }
1180 break;
1181
1182 case OP_LEA:
1183 if (op->operand[0].type == OPT_REG
1184 && op->operand[1].type == OPT_REGMEM)
1185 {
1186 char buf[16];
1187 snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name);
1188 if (IS(buf, op->operand[1].name))
1189 op->flags |= OPF_RMD | OPF_DONE;
1190 }
1191 break;
1192
1193 case OP_CALL:
1194 // trashed regs must be explicitly detected later
1195 op->regmask_dst = 0;
1196 break;
1197
1198 case OP_LEAVE:
1199 op->regmask_dst = (1 << xBP) | (1 << xSP);
1200 op->regmask_src = 1 << xBP;
1201 break;
1202
1203 default:
1204 break;
1205 }
1206}
1207
1208static const char *op_name(struct parsed_op *po)
1209{
1210 static char buf[16];
1211 char *p;
1212 int i;
1213
1214 if (po->op == OP_JCC || po->op == OP_SCC) {
1215 p = buf;
1216 *p++ = (po->op == OP_JCC) ? 'j' : 's';
1217 if (po->pfo_inv)
1218 *p++ = 'n';
1219 strcpy(p, parsed_flag_op_names[po->pfo]);
1220 return buf;
1221 }
1222
1223 for (i = 0; i < ARRAY_SIZE(op_table); i++)
1224 if (op_table[i].op == po->op)
1225 return op_table[i].name;
1226
1227 return "???";
1228}
1229
1230// debug
1231static const char *dump_op(struct parsed_op *po)
1232{
1233 static char out[128];
1234 char *p = out;
1235 int i;
1236
1237 if (po == NULL)
1238 return "???";
1239
1240 snprintf(out, sizeof(out), "%s", op_name(po));
1241 for (i = 0; i < po->operand_cnt; i++) {
1242 p += strlen(p);
1243 if (i > 0)
1244 *p++ = ',';
1245 snprintf(p, sizeof(out) - (p - out),
1246 po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s",
1247 po->operand[i].name);
1248 }
1249
1250 return out;
1251}
1252
1253static const char *lmod_type_u(struct parsed_op *po,
1254 enum opr_lenmod lmod)
1255{
1256 switch (lmod) {
1257 case OPLM_QWORD:
1258 return "u64";
1259 case OPLM_DWORD:
1260 return "u32";
1261 case OPLM_WORD:
1262 return "u16";
1263 case OPLM_BYTE:
1264 return "u8";
1265 default:
1266 ferr(po, "invalid lmod: %d\n", lmod);
1267 return "(_invalid_)";
1268 }
1269}
1270
1271static const char *lmod_cast_u(struct parsed_op *po,
1272 enum opr_lenmod lmod)
1273{
1274 switch (lmod) {
1275 case OPLM_QWORD:
1276 return "";
1277 case OPLM_DWORD:
1278 return "";
1279 case OPLM_WORD:
1280 return "(u16)";
1281 case OPLM_BYTE:
1282 return "(u8)";
1283 default:
1284 ferr(po, "invalid lmod: %d\n", lmod);
1285 return "(_invalid_)";
1286 }
1287}
1288
1289static const char *lmod_cast_u_ptr(struct parsed_op *po,
1290 enum opr_lenmod lmod)
1291{
1292 switch (lmod) {
1293 case OPLM_QWORD:
1294 return "*(u64 *)";
1295 case OPLM_DWORD:
1296 return "*(u32 *)";
1297 case OPLM_WORD:
1298 return "*(u16 *)";
1299 case OPLM_BYTE:
1300 return "*(u8 *)";
1301 default:
1302 ferr(po, "invalid lmod: %d\n", lmod);
1303 return "(_invalid_)";
1304 }
1305}
1306
1307static const char *lmod_cast_s(struct parsed_op *po,
1308 enum opr_lenmod lmod)
1309{
1310 switch (lmod) {
1311 case OPLM_QWORD:
1312 return "(s64)";
1313 case OPLM_DWORD:
1314 return "(s32)";
1315 case OPLM_WORD:
1316 return "(s16)";
1317 case OPLM_BYTE:
1318 return "(s8)";
1319 default:
1320 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1321 return "(_invalid_)";
1322 }
1323}
1324
1325static const char *lmod_cast(struct parsed_op *po,
1326 enum opr_lenmod lmod, int is_signed)
1327{
1328 return is_signed ?
1329 lmod_cast_s(po, lmod) :
1330 lmod_cast_u(po, lmod);
1331}
1332
1333static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
1334{
1335 switch (lmod) {
1336 case OPLM_QWORD:
1337 return 8;
1338 case OPLM_DWORD:
1339 return 4;
1340 case OPLM_WORD:
1341 return 2;
1342 case OPLM_BYTE:
1343 return 1;
1344 default:
1345 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1346 return 0;
1347 }
1348}
1349
1350static const char *opr_name(struct parsed_op *po, int opr_num)
1351{
1352 if (opr_num >= po->operand_cnt)
1353 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1354 return po->operand[opr_num].name;
1355}
1356
1357static unsigned int opr_const(struct parsed_op *po, int opr_num)
1358{
1359 if (opr_num >= po->operand_cnt)
1360 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1361 if (po->operand[opr_num].type != OPT_CONST)
1362 ferr(po, "opr %d: const expected\n", opr_num);
1363 return po->operand[opr_num].val;
1364}
1365
1366static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr)
1367{
1368 if ((unsigned int)popr->reg >= ARRAY_SIZE(regs_r32))
1369 ferr(po, "invalid reg: %d\n", popr->reg);
1370 return regs_r32[popr->reg];
1371}
1372
1373// cast1 is the "final" cast
1374static const char *simplify_cast(const char *cast1, const char *cast2)
1375{
1376 static char buf[256];
1377
1378 if (cast1[0] == 0)
1379 return cast2;
1380 if (cast2[0] == 0)
1381 return cast1;
1382 if (IS(cast1, cast2))
1383 return cast1;
1384 if (IS(cast1, "(s8)") && IS(cast2, "(u8)"))
1385 return cast1;
1386 if (IS(cast1, "(s16)") && IS(cast2, "(u16)"))
1387 return cast1;
1388 if (IS(cast1, "(u8)") && IS_START(cast2, "*(u8 *)"))
1389 return cast2;
1390 if (IS(cast1, "(u16)") && IS_START(cast2, "*(u16 *)"))
1391 return cast2;
1392 if (strchr(cast1, '*') && IS_START(cast2, "(u32)"))
1393 return cast1;
1394
1395 snprintf(buf, sizeof(buf), "%s%s", cast1, cast2);
1396 return buf;
1397}
1398
1399static const char *simplify_cast_num(const char *cast, unsigned int val)
1400{
1401 if (IS(cast, "(u8)") && val < 0x100)
1402 return "";
1403 if (IS(cast, "(s8)") && val < 0x80)
1404 return "";
1405 if (IS(cast, "(u16)") && val < 0x10000)
1406 return "";
1407 if (IS(cast, "(s16)") && val < 0x8000)
1408 return "";
1409 if (IS(cast, "(s32)") && val < 0x80000000)
1410 return "";
1411
1412 return cast;
1413}
1414
1415static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
1416 int *extra_offs)
1417{
1418 const char *p;
1419 char *endp;
1420 int namelen;
1421 int i;
1422
1423 *extra_offs = 0;
1424 namelen = strlen(name);
1425
1426 p = strchr(name, '+');
1427 if (p != NULL) {
1428 namelen = p - name;
1429 if (namelen <= 0)
1430 ferr(po, "equ parse failed for '%s'\n", name);
1431
1432 if (IS_START(p, "0x"))
1433 p += 2;
1434 *extra_offs = strtol(p, &endp, 16);
1435 if (*endp != 0)
1436 ferr(po, "equ parse failed for '%s'\n", name);
1437 }
1438
1439 for (i = 0; i < g_eqcnt; i++)
1440 if (strncmp(g_eqs[i].name, name, namelen) == 0
1441 && g_eqs[i].name[namelen] == 0)
1442 break;
1443 if (i >= g_eqcnt) {
1444 if (po != NULL)
1445 ferr(po, "unresolved equ name: '%s'\n", name);
1446 return NULL;
1447 }
1448
1449 return &g_eqs[i];
1450}
1451
1452static int is_stack_access(struct parsed_op *po,
1453 const struct parsed_opr *popr)
1454{
1455 return (parse_stack_el(popr->name, NULL, 0)
1456 || (g_bp_frame && !(po->flags & OPF_EBP_S)
1457 && IS_START(popr->name, "ebp")));
1458}
1459
1460static void parse_stack_access(struct parsed_op *po,
1461 const char *name, char *ofs_reg, int *offset_out,
1462 int *stack_ra_out, const char **bp_arg_out, int is_lea)
1463{
1464 const char *bp_arg = "";
1465 const char *p = NULL;
1466 struct parsed_equ *eq;
1467 char *endp = NULL;
1468 int stack_ra = 0;
1469 int offset = 0;
1470
1471 ofs_reg[0] = 0;
1472
1473 if (IS_START(name, "ebp-")
1474 || (IS_START(name, "ebp+") && '0' <= name[4] && name[4] <= '9'))
1475 {
1476 p = name + 4;
1477 if (IS_START(p, "0x"))
1478 p += 2;
1479 offset = strtoul(p, &endp, 16);
1480 if (name[3] == '-')
1481 offset = -offset;
1482 if (*endp != 0)
1483 ferr(po, "ebp- parse of '%s' failed\n", name);
1484 }
1485 else {
1486 bp_arg = parse_stack_el(name, ofs_reg, 0);
1487 snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
1488 eq = equ_find(po, bp_arg, &offset);
1489 if (eq == NULL)
1490 ferr(po, "detected but missing eq\n");
1491 offset += eq->offset;
1492 }
1493
1494 if (!strncmp(name, "ebp", 3))
1495 stack_ra = 4;
1496
1497 // yes it sometimes LEAs ra for compares..
1498 if (!is_lea && ofs_reg[0] == 0
1499 && stack_ra <= offset && offset < stack_ra + 4)
1500 {
1501 ferr(po, "reference to ra? %d %d\n", offset, stack_ra);
1502 }
1503
1504 *offset_out = offset;
1505 *stack_ra_out = stack_ra;
1506 if (bp_arg_out)
1507 *bp_arg_out = bp_arg;
1508}
1509
1510static int stack_frame_access(struct parsed_op *po,
1511 struct parsed_opr *popr, char *buf, size_t buf_size,
1512 const char *name, const char *cast, int is_src, int is_lea)
1513{
1514 enum opr_lenmod tmp_lmod = OPLM_UNSPEC;
1515 const char *prefix = "";
1516 const char *bp_arg = NULL;
1517 char ofs_reg[16] = { 0, };
1518 int i, arg_i, arg_s;
1519 int unaligned = 0;
1520 int stack_ra = 0;
1521 int offset = 0;
1522 int retval = -1;
1523 int sf_ofs;
1524 int lim;
1525
1526 if (po->flags & OPF_EBP_S)
1527 ferr(po, "stack_frame_access while ebp is scratch\n");
1528
1529 parse_stack_access(po, name, ofs_reg, &offset,
1530 &stack_ra, &bp_arg, is_lea);
1531
1532 if (offset > stack_ra)
1533 {
1534 arg_i = (offset - stack_ra - 4) / 4;
1535 if (arg_i < 0 || arg_i >= g_func_pp->argc_stack)
1536 {
1537 if (g_func_pp->is_vararg
1538 && arg_i == g_func_pp->argc_stack && is_lea)
1539 {
1540 // should be va_list
1541 if (cast[0] == 0)
1542 cast = "(u32)";
1543 snprintf(buf, buf_size, "%sap", cast);
1544 return -1;
1545 }
1546 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1547 offset, bp_arg, arg_i);
1548 }
1549 if (ofs_reg[0] != 0)
1550 ferr(po, "offset reg on arg access?\n");
1551
1552 for (i = arg_s = 0; i < g_func_pp->argc; i++) {
1553 if (g_func_pp->arg[i].reg != NULL)
1554 continue;
1555 if (arg_s == arg_i)
1556 break;
1557 arg_s++;
1558 }
1559 if (i == g_func_pp->argc)
1560 ferr(po, "arg %d not in prototype?\n", arg_i);
1561
1562 popr->is_ptr = g_func_pp->arg[i].type.is_ptr;
1563 retval = i;
1564
1565 switch (popr->lmod)
1566 {
1567 case OPLM_BYTE:
1568 if (is_lea)
1569 ferr(po, "lea/byte to arg?\n");
1570 if (is_src && (offset & 3) == 0)
1571 snprintf(buf, buf_size, "%sa%d",
1572 simplify_cast(cast, "(u8)"), i + 1);
1573 else
1574 snprintf(buf, buf_size, "%sBYTE%d(a%d)",
1575 cast, offset & 3, i + 1);
1576 break;
1577
1578 case OPLM_WORD:
1579 if (is_lea)
1580 ferr(po, "lea/word to arg?\n");
1581 if (offset & 1) {
1582 unaligned = 1;
1583 if (!is_src) {
1584 if (offset & 2)
1585 ferr(po, "problematic arg store\n");
1586 snprintf(buf, buf_size, "%s((char *)&a%d + 1)",
1587 simplify_cast(cast, "*(u16 *)"), i + 1);
1588 }
1589 else
1590 ferr(po, "unaligned arg word load\n");
1591 }
1592 else if (is_src && (offset & 2) == 0)
1593 snprintf(buf, buf_size, "%sa%d",
1594 simplify_cast(cast, "(u16)"), i + 1);
1595 else
1596 snprintf(buf, buf_size, "%s%sWORD(a%d)",
1597 cast, (offset & 2) ? "HI" : "LO", i + 1);
1598 break;
1599
1600 case OPLM_DWORD:
1601 if (cast[0])
1602 prefix = cast;
1603 else if (is_src)
1604 prefix = "(u32)";
1605
1606 if (offset & 3) {
1607 unaligned = 1;
1608 if (is_lea)
1609 snprintf(buf, buf_size, "(u32)&a%d + %d",
1610 i + 1, offset & 3);
1611 else if (!is_src)
1612 ferr(po, "unaligned arg store\n");
1613 else {
1614 // mov edx, [ebp+arg_4+2]; movsx ecx, dx
1615 snprintf(buf, buf_size, "%s(a%d >> %d)",
1616 prefix, i + 1, (offset & 3) * 8);
1617 }
1618 }
1619 else {
1620 snprintf(buf, buf_size, "%s%sa%d",
1621 prefix, is_lea ? "&" : "", i + 1);
1622 }
1623 break;
1624
1625 default:
1626 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
1627 }
1628
1629 if (unaligned)
1630 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1631
1632 // common problem
1633 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type);
1634 if (tmp_lmod != OPLM_DWORD
1635 && (unaligned || (!is_src && lmod_bytes(po, tmp_lmod)
1636 < lmod_bytes(po, popr->lmod) + (offset & 3))))
1637 {
1638 ferr(po, "bp_arg arg%d/w offset %d and type '%s' is too small\n",
1639 i + 1, offset, g_func_pp->arg[i].type.name);
1640 }
1641 // can't check this because msvc likes to reuse
1642 // arg space for scratch..
1643 //if (popr->is_ptr && popr->lmod != OPLM_DWORD)
1644 // ferr(po, "bp_arg arg%d: non-dword ptr access\n", i + 1);
1645 }
1646 else
1647 {
1648 if (g_stack_fsz == 0)
1649 ferr(po, "stack var access without stackframe\n");
1650 g_stack_frame_used = 1;
1651
1652 sf_ofs = g_stack_fsz + offset;
1653 lim = (ofs_reg[0] != 0) ? -4 : 0;
1654 if (offset > 0 || sf_ofs < lim)
1655 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
1656
1657 if (is_lea)
1658 prefix = "(u32)&";
1659 else
1660 prefix = cast;
1661
1662 switch (popr->lmod)
1663 {
1664 case OPLM_BYTE:
1665 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1666 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1667 break;
1668
1669 case OPLM_WORD:
1670 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1671 // known unaligned or possibly unaligned
1672 strcat(g_comment, " unaligned");
1673 if (prefix[0] == 0)
1674 prefix = "*(u16 *)&";
1675 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1676 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1677 break;
1678 }
1679 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
1680 break;
1681
1682 case OPLM_DWORD:
1683 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1684 // known unaligned or possibly unaligned
1685 strcat(g_comment, " unaligned");
1686 if (prefix[0] == 0)
1687 prefix = "*(u32 *)&";
1688 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1689 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1690 break;
1691 }
1692 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
1693 break;
1694
1695 default:
1696 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
1697 }
1698 }
1699
1700 return retval;
1701}
1702
1703static void check_func_pp(struct parsed_op *po,
1704 const struct parsed_proto *pp, const char *pfx)
1705{
1706 enum opr_lenmod tmp_lmod;
1707 char buf[256];
1708 int ret, i;
1709
1710 if (pp->argc_reg != 0) {
1711 if (/*!g_allow_regfunc &&*/ !pp->is_fastcall) {
1712 pp_print(buf, sizeof(buf), pp);
1713 ferr(po, "%s: unexpected reg arg in icall: %s\n", pfx, buf);
1714 }
1715 if (pp->argc_stack > 0 && pp->argc_reg != 2)
1716 ferr(po, "%s: %d reg arg(s) with %d stack arg(s)\n",
1717 pfx, pp->argc_reg, pp->argc_stack);
1718 }
1719
1720 // fptrs must use 32bit args, callsite might have no information and
1721 // lack a cast to smaller types, which results in incorrectly masked
1722 // args passed (callee may assume masked args, it does on ARM)
1723 if (!pp->is_osinc) {
1724 for (i = 0; i < pp->argc; i++) {
1725 ret = guess_lmod_from_c_type(&tmp_lmod, &pp->arg[i].type);
1726 if (ret && tmp_lmod != OPLM_DWORD)
1727 ferr(po, "reference to %s with arg%d '%s'\n", pp->name,
1728 i + 1, pp->arg[i].type.name);
1729 }
1730 }
1731}
1732
1733static const char *check_label_read_ref(struct parsed_op *po,
1734 const char *name)
1735{
1736 const struct parsed_proto *pp;
1737
1738 pp = proto_parse(g_fhdr, name, 0);
1739 if (pp == NULL)
1740 ferr(po, "proto_parse failed for ref '%s'\n", name);
1741
1742 if (pp->is_func)
1743 check_func_pp(po, pp, "ref");
1744
1745 return pp->name;
1746}
1747
1748static char *out_src_opr(char *buf, size_t buf_size,
1749 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
1750 int is_lea)
1751{
1752 char tmp1[256], tmp2[256];
1753 char expr[256];
1754 const char *name;
1755 char *p;
1756 int ret;
1757
1758 if (cast == NULL)
1759 cast = "";
1760
1761 switch (popr->type) {
1762 case OPT_REG:
1763 if (is_lea)
1764 ferr(po, "lea from reg?\n");
1765
1766 switch (popr->lmod) {
1767 case OPLM_QWORD:
1768 snprintf(buf, buf_size, "%s%s.q", cast, opr_reg_p(po, popr));
1769 break;
1770 case OPLM_DWORD:
1771 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
1772 break;
1773 case OPLM_WORD:
1774 snprintf(buf, buf_size, "%s%s",
1775 simplify_cast(cast, "(u16)"), opr_reg_p(po, popr));
1776 break;
1777 case OPLM_BYTE:
1778 if (popr->name[1] == 'h') // XXX..
1779 snprintf(buf, buf_size, "%s(%s >> 8)",
1780 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
1781 else
1782 snprintf(buf, buf_size, "%s%s",
1783 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
1784 break;
1785 default:
1786 ferr(po, "invalid src lmod: %d\n", popr->lmod);
1787 }
1788 break;
1789
1790 case OPT_REGMEM:
1791 if (is_stack_access(po, popr)) {
1792 stack_frame_access(po, popr, buf, buf_size,
1793 popr->name, cast, 1, is_lea);
1794 break;
1795 }
1796
1797 strcpy(expr, popr->name);
1798 if (strchr(expr, '[')) {
1799 // special case: '[' can only be left for label[reg] form
1800 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
1801 if (ret != 2)
1802 ferr(po, "parse failure for '%s'\n", expr);
1803 if (tmp1[0] == '(') {
1804 // (off_4FFF50+3)[eax]
1805 p = strchr(tmp1 + 1, ')');
1806 if (p == NULL || p[1] != 0)
1807 ferr(po, "parse failure (2) for '%s'\n", expr);
1808 *p = 0;
1809 memmove(tmp1, tmp1 + 1, strlen(tmp1));
1810 }
1811 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
1812 }
1813
1814 // XXX: do we need more parsing?
1815 if (is_lea) {
1816 snprintf(buf, buf_size, "%s", expr);
1817 break;
1818 }
1819
1820 snprintf(buf, buf_size, "%s(%s)",
1821 simplify_cast(cast, lmod_cast_u_ptr(po, popr->lmod)), expr);
1822 break;
1823
1824 case OPT_LABEL:
1825 name = check_label_read_ref(po, popr->name);
1826 if (cast[0] == 0 && popr->is_ptr)
1827 cast = "(u32)";
1828
1829 if (is_lea)
1830 snprintf(buf, buf_size, "(u32)&%s", name);
1831 else if (popr->size_lt)
1832 snprintf(buf, buf_size, "%s%s%s%s", cast,
1833 lmod_cast_u_ptr(po, popr->lmod),
1834 popr->is_array ? "" : "&", name);
1835 else
1836 snprintf(buf, buf_size, "%s%s%s", cast, name,
1837 popr->is_array ? "[0]" : "");
1838 break;
1839
1840 case OPT_OFFSET:
1841 name = check_label_read_ref(po, popr->name);
1842 if (cast[0] == 0)
1843 cast = "(u32)";
1844 if (is_lea)
1845 ferr(po, "lea an offset?\n");
1846 snprintf(buf, buf_size, "%s&%s", cast, name);
1847 break;
1848
1849 case OPT_CONST:
1850 if (is_lea)
1851 ferr(po, "lea from const?\n");
1852
1853 printf_number(tmp1, sizeof(tmp1), popr->val);
1854 if (popr->val == 0 && strchr(cast, '*'))
1855 snprintf(buf, buf_size, "NULL");
1856 else
1857 snprintf(buf, buf_size, "%s%s",
1858 simplify_cast_num(cast, popr->val), tmp1);
1859 break;
1860
1861 default:
1862 ferr(po, "invalid src type: %d\n", popr->type);
1863 }
1864
1865 return buf;
1866}
1867
1868// note: may set is_ptr (we find that out late for ebp frame..)
1869static char *out_dst_opr(char *buf, size_t buf_size,
1870 struct parsed_op *po, struct parsed_opr *popr)
1871{
1872 switch (popr->type) {
1873 case OPT_REG:
1874 switch (popr->lmod) {
1875 case OPLM_QWORD:
1876 snprintf(buf, buf_size, "%s.q", opr_reg_p(po, popr));
1877 break;
1878 case OPLM_DWORD:
1879 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
1880 break;
1881 case OPLM_WORD:
1882 // ugh..
1883 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
1884 break;
1885 case OPLM_BYTE:
1886 // ugh..
1887 if (popr->name[1] == 'h') // XXX..
1888 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
1889 else
1890 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
1891 break;
1892 default:
1893 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
1894 }
1895 break;
1896
1897 case OPT_REGMEM:
1898 if (is_stack_access(po, popr)) {
1899 stack_frame_access(po, popr, buf, buf_size,
1900 popr->name, "", 0, 0);
1901 break;
1902 }
1903
1904 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1905
1906 case OPT_LABEL:
1907 if (popr->size_mismatch)
1908 snprintf(buf, buf_size, "%s%s%s",
1909 lmod_cast_u_ptr(po, popr->lmod),
1910 popr->is_array ? "" : "&", popr->name);
1911 else
1912 snprintf(buf, buf_size, "%s%s", popr->name,
1913 popr->is_array ? "[0]" : "");
1914 break;
1915
1916 default:
1917 ferr(po, "invalid dst type: %d\n", popr->type);
1918 }
1919
1920 return buf;
1921}
1922
1923static char *out_src_opr_u32(char *buf, size_t buf_size,
1924 struct parsed_op *po, struct parsed_opr *popr)
1925{
1926 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1927}
1928
1929static void out_test_for_cc(char *buf, size_t buf_size,
1930 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
1931 enum opr_lenmod lmod, const char *expr)
1932{
1933 const char *cast, *scast;
1934
1935 cast = lmod_cast_u(po, lmod);
1936 scast = lmod_cast_s(po, lmod);
1937
1938 switch (pfo) {
1939 case PFO_Z:
1940 case PFO_BE: // CF=1||ZF=1; CF=0
1941 snprintf(buf, buf_size, "(%s%s %s 0)",
1942 cast, expr, is_inv ? "!=" : "==");
1943 break;
1944
1945 case PFO_S:
1946 case PFO_L: // SF!=OF; OF=0
1947 snprintf(buf, buf_size, "(%s%s %s 0)",
1948 scast, expr, is_inv ? ">=" : "<");
1949 break;
1950
1951 case PFO_LE: // ZF=1||SF!=OF; OF=0
1952 snprintf(buf, buf_size, "(%s%s %s 0)",
1953 scast, expr, is_inv ? ">" : "<=");
1954 break;
1955
1956 default:
1957 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1958 }
1959}
1960
1961static void out_cmp_for_cc(char *buf, size_t buf_size,
1962 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
1963{
1964 const char *cast, *scast, *cast_use;
1965 char buf1[256], buf2[256];
1966 enum opr_lenmod lmod;
1967
1968 if (po->op != OP_DEC && po->operand[0].lmod != po->operand[1].lmod)
1969 ferr(po, "%s: lmod mismatch: %d %d\n", __func__,
1970 po->operand[0].lmod, po->operand[1].lmod);
1971 lmod = po->operand[0].lmod;
1972
1973 cast = lmod_cast_u(po, lmod);
1974 scast = lmod_cast_s(po, lmod);
1975
1976 switch (pfo) {
1977 case PFO_C:
1978 case PFO_Z:
1979 case PFO_BE: // !a
1980 cast_use = cast;
1981 break;
1982
1983 case PFO_S:
1984 case PFO_L: // !ge
1985 case PFO_LE:
1986 cast_use = scast;
1987 break;
1988
1989 default:
1990 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1991 }
1992
1993 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], cast_use, 0);
1994 if (po->op == OP_DEC)
1995 snprintf(buf2, sizeof(buf2), "1");
1996 else
1997 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], cast_use, 0);
1998
1999 switch (pfo) {
2000 case PFO_C:
2001 // note: must be unsigned compare
2002 snprintf(buf, buf_size, "(%s %s %s)",
2003 buf1, is_inv ? ">=" : "<", buf2);
2004 break;
2005
2006 case PFO_Z:
2007 snprintf(buf, buf_size, "(%s %s %s)",
2008 buf1, is_inv ? "!=" : "==", buf2);
2009 break;
2010
2011 case PFO_BE: // !a
2012 // note: must be unsigned compare
2013 snprintf(buf, buf_size, "(%s %s %s)",
2014 buf1, is_inv ? ">" : "<=", buf2);
2015
2016 // annoying case
2017 if (is_inv && lmod == OPLM_BYTE
2018 && po->operand[1].type == OPT_CONST
2019 && po->operand[1].val == 0xff)
2020 {
2021 snprintf(g_comment, sizeof(g_comment), "if %s", buf);
2022 snprintf(buf, buf_size, "(0)");
2023 }
2024 break;
2025
2026 // note: must be signed compare
2027 case PFO_S:
2028 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
2029 scast, buf1, buf2, is_inv ? ">=" : "<");
2030 break;
2031
2032 case PFO_L: // !ge
2033 snprintf(buf, buf_size, "(%s %s %s)",
2034 buf1, is_inv ? ">=" : "<", buf2);
2035 break;
2036
2037 case PFO_LE: // !g
2038 snprintf(buf, buf_size, "(%s %s %s)",
2039 buf1, is_inv ? ">" : "<=", buf2);
2040 break;
2041
2042 default:
2043 break;
2044 }
2045}
2046
2047static void out_cmp_test(char *buf, size_t buf_size,
2048 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
2049{
2050 char buf1[256], buf2[256], buf3[256];
2051
2052 if (po->op == OP_TEST) {
2053 if (IS(opr_name(po, 0), opr_name(po, 1))) {
2054 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
2055 }
2056 else {
2057 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
2058 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
2059 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
2060 }
2061 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
2062 po->operand[0].lmod, buf3);
2063 }
2064 else if (po->op == OP_CMP) {
2065 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv);
2066 }
2067 else
2068 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
2069}
2070
2071static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
2072 struct parsed_opr *popr2)
2073{
2074 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
2075 ferr(po, "missing lmod for both operands\n");
2076
2077 if (popr1->lmod == OPLM_UNSPEC)
2078 popr1->lmod = popr2->lmod;
2079 else if (popr2->lmod == OPLM_UNSPEC)
2080 popr2->lmod = popr1->lmod;
2081 else if (popr1->lmod != popr2->lmod) {
2082 if (popr1->type_from_var) {
2083 popr1->size_mismatch = 1;
2084 if (popr1->lmod < popr2->lmod)
2085 popr1->size_lt = 1;
2086 popr1->lmod = popr2->lmod;
2087 }
2088 else if (popr2->type_from_var) {
2089 popr2->size_mismatch = 1;
2090 if (popr2->lmod < popr1->lmod)
2091 popr2->size_lt = 1;
2092 popr2->lmod = popr1->lmod;
2093 }
2094 else
2095 ferr(po, "conflicting lmods: %d vs %d\n",
2096 popr1->lmod, popr2->lmod);
2097 }
2098}
2099
2100static const char *op_to_c(struct parsed_op *po)
2101{
2102 switch (po->op)
2103 {
2104 case OP_ADD:
2105 case OP_ADC:
2106 return "+";
2107 case OP_SUB:
2108 case OP_SBB:
2109 return "-";
2110 case OP_AND:
2111 return "&";
2112 case OP_OR:
2113 return "|";
2114 case OP_XOR:
2115 return "^";
2116 case OP_SHL:
2117 return "<<";
2118 case OP_SHR:
2119 return ">>";
2120 case OP_MUL:
2121 case OP_IMUL:
2122 return "*";
2123 default:
2124 ferr(po, "op_to_c was supplied with %d\n", po->op);
2125 }
2126}
2127
2128static void op_set_clear_flag(struct parsed_op *po,
2129 enum op_flags flag_set, enum op_flags flag_clear)
2130{
2131 po->flags |= flag_set;
2132 po->flags &= ~flag_clear;
2133}
2134
2135// last op in stream - unconditional branch or ret
2136#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
2137 || ((ops[_i].flags & (OPF_JMP|OPF_CJMP|OPF_RMD)) == OPF_JMP \
2138 && ops[_i].op != OP_CALL))
2139
2140static int scan_for_pop(int i, int opcnt, const char *reg,
2141 int magic, int depth, int *maxdepth, int do_flags)
2142{
2143 const struct parsed_proto *pp;
2144 struct parsed_op *po;
2145 int ret = 0;
2146 int j;
2147
2148 for (; i < opcnt; i++) {
2149 po = &ops[i];
2150 if (po->cc_scratch == magic)
2151 break; // already checked
2152 po->cc_scratch = magic;
2153
2154 if (po->flags & OPF_TAIL) {
2155 if (po->op == OP_CALL) {
2156 pp = proto_parse(g_fhdr, po->operand[0].name, g_quiet_pp);
2157 if (pp != NULL && pp->is_noreturn)
2158 // no stack cleanup for noreturn
2159 return ret;
2160 }
2161 return -1; // deadend
2162 }
2163
2164 if ((po->flags & (OPF_RMD|OPF_DONE))
2165 || (po->op == OP_PUSH && po->p_argnum != 0)) // arg push
2166 continue;
2167
2168 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2169 if (po->btj != NULL) {
2170 // jumptable
2171 for (j = 0; j < po->btj->count; j++) {
2172 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic,
2173 depth, maxdepth, do_flags);
2174 if (ret < 0)
2175 return ret; // dead end
2176 }
2177 return ret;
2178 }
2179
2180 if (po->bt_i < 0) {
2181 ferr(po, "dead branch\n");
2182 return -1;
2183 }
2184
2185 if (po->flags & OPF_CJMP) {
2186 ret |= scan_for_pop(po->bt_i, opcnt, reg, magic,
2187 depth, maxdepth, do_flags);
2188 if (ret < 0)
2189 return ret; // dead end
2190 }
2191 else {
2192 i = po->bt_i - 1;
2193 }
2194 continue;
2195 }
2196
2197 if ((po->op == OP_POP || po->op == OP_PUSH)
2198 && po->operand[0].type == OPT_REG
2199 && IS(po->operand[0].name, reg))
2200 {
2201 if (po->op == OP_PUSH && !(po->flags & OPF_FARGNR)) {
2202 depth++;
2203 if (depth > *maxdepth)
2204 *maxdepth = depth;
2205 if (do_flags)
2206 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2207 }
2208 else if (po->op == OP_POP) {
2209 if (depth == 0) {
2210 if (do_flags)
2211 op_set_clear_flag(po, OPF_RMD, OPF_RSAVE);
2212 return 1;
2213 }
2214 else {
2215 depth--;
2216 if (depth < 0) // should not happen
2217 ferr(po, "fail with depth\n");
2218 if (do_flags)
2219 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2220 }
2221 }
2222 }
2223 }
2224
2225 return ret;
2226}
2227
2228// scan for pop starting from 'ret' op (all paths)
2229static int scan_for_pop_ret(int i, int opcnt, const char *reg,
2230 int flag_set)
2231{
2232 int found = 0;
2233 int j;
2234
2235 for (; i < opcnt; i++) {
2236 if (!(ops[i].flags & OPF_TAIL))
2237 continue;
2238
2239 for (j = i - 1; j >= 0; j--) {
2240 if (ops[j].flags & (OPF_RMD|OPF_DONE))
2241 continue;
2242 if (ops[j].flags & OPF_JMP)
2243 return -1;
2244
2245 if (ops[j].op == OP_POP && ops[j].datap == NULL
2246 && ops[j].operand[0].type == OPT_REG
2247 && IS(ops[j].operand[0].name, reg))
2248 {
2249 found = 1;
2250 ops[j].flags |= flag_set;
2251 break;
2252 }
2253
2254 if (g_labels[j] != NULL)
2255 return -1;
2256 }
2257 }
2258
2259 return found ? 0 : -1;
2260}
2261
2262static void scan_for_pop_const(int i, int opcnt)
2263{
2264 int j;
2265
2266 for (j = i + 1; j < opcnt; j++) {
2267 if ((ops[j].flags & (OPF_JMP|OPF_TAIL|OPF_RSAVE))
2268 || ops[j].op == OP_PUSH || g_labels[i] != NULL)
2269 {
2270 break;
2271 }
2272
2273 if (ops[j].op == OP_POP && !(ops[j].flags & (OPF_RMD|OPF_DONE)))
2274 {
2275 ops[i].flags |= OPF_RMD | OPF_DONE;
2276 ops[j].flags |= OPF_DONE;
2277 ops[j].datap = &ops[i];
2278 break;
2279 }
2280 }
2281}
2282
2283static void scan_propagate_df(int i, int opcnt)
2284{
2285 struct parsed_op *po = &ops[i];
2286 int j;
2287
2288 for (; i < opcnt; i++) {
2289 po = &ops[i];
2290 if (po->flags & OPF_DF)
2291 return; // already resolved
2292 po->flags |= OPF_DF;
2293
2294 if (po->op == OP_CALL)
2295 ferr(po, "call with DF set?\n");
2296
2297 if (po->flags & OPF_JMP) {
2298 if (po->btj != NULL) {
2299 // jumptable
2300 for (j = 0; j < po->btj->count; j++)
2301 scan_propagate_df(po->btj->d[j].bt_i, opcnt);
2302 return;
2303 }
2304
2305 if (po->bt_i < 0) {
2306 ferr(po, "dead branch\n");
2307 return;
2308 }
2309
2310 if (po->flags & OPF_CJMP)
2311 scan_propagate_df(po->bt_i, opcnt);
2312 else
2313 i = po->bt_i - 1;
2314 continue;
2315 }
2316
2317 if (po->flags & OPF_TAIL)
2318 break;
2319
2320 if (po->op == OP_CLD) {
2321 po->flags |= OPF_RMD | OPF_DONE;
2322 return;
2323 }
2324 }
2325
2326 ferr(po, "missing DF clear?\n");
2327}
2328
2329// is operand 'opr' modified by parsed_op 'po'?
2330static int is_opr_modified(const struct parsed_opr *opr,
2331 const struct parsed_op *po)
2332{
2333 int mask;
2334
2335 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2336 return 0;
2337
2338 if (opr->type == OPT_REG) {
2339 if (po->op == OP_CALL) {
2340 mask = (1 << xAX) | (1 << xCX) | (1 << xDX);
2341 if ((1 << opr->reg) & mask)
2342 return 1;
2343 else
2344 return 0;
2345 }
2346
2347 if (po->operand[0].type == OPT_REG) {
2348 if (po->regmask_dst & (1 << opr->reg))
2349 return 1;
2350 else
2351 return 0;
2352 }
2353 }
2354
2355 return IS(po->operand[0].name, opr->name);
2356}
2357
2358// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
2359static int is_any_opr_modified(const struct parsed_op *po_test,
2360 const struct parsed_op *po, int c_mode)
2361{
2362 int mask;
2363 int i;
2364
2365 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2366 return 0;
2367
2368 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2369 return 0;
2370
2371 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
2372 return 1;
2373
2374 // in reality, it can wreck any register, but in decompiled C
2375 // version it can only overwrite eax or edx:eax
2376 mask = (1 << xAX) | (1 << xDX);
2377 if (!c_mode)
2378 mask |= 1 << xCX;
2379
2380 if (po->op == OP_CALL
2381 && ((po_test->regmask_src | po_test->regmask_dst) & mask))
2382 return 1;
2383
2384 for (i = 0; i < po_test->operand_cnt; i++)
2385 if (IS(po_test->operand[i].name, po->operand[0].name))
2386 return 1;
2387
2388 return 0;
2389}
2390
2391// scan for any po_test operand modification in range given
2392static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt,
2393 int c_mode)
2394{
2395 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2396 return -1;
2397
2398 for (; i < opcnt; i++) {
2399 if (is_any_opr_modified(po_test, &ops[i], c_mode))
2400 return i;
2401 }
2402
2403 return -1;
2404}
2405
2406// scan for po_test operand[0] modification in range given
2407static int scan_for_mod_opr0(struct parsed_op *po_test,
2408 int i, int opcnt)
2409{
2410 for (; i < opcnt; i++) {
2411 if (is_opr_modified(&po_test->operand[0], &ops[i]))
2412 return i;
2413 }
2414
2415 return -1;
2416}
2417
2418#define check_i(po, i) \
2419 if ((i) < 0) \
2420 ferr(po, "bad " #i ": %d\n", i)
2421
2422static int scan_for_flag_set(int i, int magic, int *branched,
2423 int *setters, int *setter_cnt)
2424{
2425 struct label_ref *lr;
2426 int ret;
2427
2428 while (i >= 0) {
2429 if (ops[i].cc_scratch == magic) {
2430 ferr(&ops[i], "%s looped\n", __func__);
2431 return -1;
2432 }
2433 ops[i].cc_scratch = magic;
2434
2435 if (g_labels[i] != NULL) {
2436 *branched = 1;
2437
2438 lr = &g_label_refs[i];
2439 for (; lr->next; lr = lr->next) {
2440 check_i(&ops[i], lr->i);
2441 ret = scan_for_flag_set(lr->i, magic,
2442 branched, setters, setter_cnt);
2443 if (ret < 0)
2444 return ret;
2445 }
2446
2447 check_i(&ops[i], lr->i);
2448 if (i > 0 && LAST_OP(i - 1)) {
2449 i = lr->i;
2450 continue;
2451 }
2452 ret = scan_for_flag_set(lr->i, magic,
2453 branched, setters, setter_cnt);
2454 if (ret < 0)
2455 return ret;
2456 }
2457 i--;
2458
2459 if (ops[i].flags & OPF_FLAGS) {
2460 setters[*setter_cnt] = i;
2461 (*setter_cnt)++;
2462 return 0;
2463 }
2464
2465 if ((ops[i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP)
2466 return -1;
2467 }
2468
2469 return -1;
2470}
2471
2472// scan back for cdq, if anything modifies edx, fail
2473static int scan_for_cdq_edx(int i)
2474{
2475 while (i >= 0) {
2476 if (g_labels[i] != NULL) {
2477 if (g_label_refs[i].next != NULL)
2478 return -1;
2479 if (i > 0 && LAST_OP(i - 1)) {
2480 i = g_label_refs[i].i;
2481 continue;
2482 }
2483 return -1;
2484 }
2485 i--;
2486
2487 if (ops[i].op == OP_CDQ)
2488 return i;
2489
2490 if (ops[i].regmask_dst & (1 << xDX))
2491 return -1;
2492 }
2493
2494 return -1;
2495}
2496
2497static int scan_for_reg_clear(int i, int reg)
2498{
2499 while (i >= 0) {
2500 if (g_labels[i] != NULL) {
2501 if (g_label_refs[i].next != NULL)
2502 return -1;
2503 if (i > 0 && LAST_OP(i - 1)) {
2504 i = g_label_refs[i].i;
2505 continue;
2506 }
2507 return -1;
2508 }
2509 i--;
2510
2511 if (ops[i].op == OP_XOR
2512 && ops[i].operand[0].lmod == OPLM_DWORD
2513 && ops[i].operand[0].reg == ops[i].operand[1].reg
2514 && ops[i].operand[0].reg == reg)
2515 return i;
2516
2517 if (ops[i].regmask_dst & (1 << reg))
2518 return -1;
2519 }
2520
2521 return -1;
2522}
2523
2524static void patch_esp_adjust(struct parsed_op *po, int adj)
2525{
2526 ferr_assert(po, po->op == OP_ADD);
2527 ferr_assert(po, IS(opr_name(po, 0), "esp"));
2528 ferr_assert(po, po->operand[1].type == OPT_CONST);
2529
2530 // this is a bit of a hack, but deals with use of
2531 // single adj for multiple calls
2532 po->operand[1].val -= adj;
2533 po->flags |= OPF_RMD;
2534 if (po->operand[1].val == 0)
2535 po->flags |= OPF_DONE;
2536 ferr_assert(po, (int)po->operand[1].val >= 0);
2537}
2538
2539// scan for positive, constant esp adjust
2540// multipath case is preliminary
2541static int scan_for_esp_adjust(int i, int opcnt,
2542 int adj_expect, int *adj, int *is_multipath, int do_update)
2543{
2544 struct parsed_op *po;
2545 int first_pop = -1;
2546
2547 *adj = *is_multipath = 0;
2548
2549 for (; i < opcnt && *adj < adj_expect; i++) {
2550 if (g_labels[i] != NULL)
2551 *is_multipath = 1;
2552
2553 po = &ops[i];
2554 if (po->flags & OPF_DONE)
2555 continue;
2556
2557 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
2558 if (po->operand[1].type != OPT_CONST)
2559 ferr(&ops[i], "non-const esp adjust?\n");
2560 *adj += po->operand[1].val;
2561 if (*adj & 3)
2562 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
2563 if (do_update) {
2564 if (!*is_multipath)
2565 patch_esp_adjust(po, adj_expect);
2566 else
2567 po->flags |= OPF_RMD;
2568 }
2569 return i;
2570 }
2571 else if (po->op == OP_PUSH) {
2572 //if (first_pop == -1)
2573 // first_pop = -2; // none
2574 *adj -= lmod_bytes(po, po->operand[0].lmod);
2575 }
2576 else if (po->op == OP_POP) {
2577 if (!(po->flags & OPF_DONE)) {
2578 // seems like msvc only uses 'pop ecx' for stack realignment..
2579 if (po->operand[0].type != OPT_REG || po->operand[0].reg != xCX)
2580 break;
2581 if (first_pop == -1 && *adj >= 0)
2582 first_pop = i;
2583 }
2584 if (do_update && *adj >= 0) {
2585 po->flags |= OPF_RMD;
2586 if (!*is_multipath)
2587 po->flags |= OPF_DONE;
2588 }
2589
2590 *adj += lmod_bytes(po, po->operand[0].lmod);
2591 }
2592 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
2593 if (po->op == OP_JMP && po->btj == NULL) {
2594 if (po->bt_i <= i)
2595 break;
2596 i = po->bt_i - 1;
2597 continue;
2598 }
2599 if (po->op != OP_CALL)
2600 break;
2601 if (po->operand[0].type != OPT_LABEL)
2602 break;
2603 if (po->pp != NULL && po->pp->is_stdcall)
2604 break;
2605 // assume it's another cdecl call
2606 }
2607 }
2608
2609 if (first_pop >= 0) {
2610 // probably 'pop ecx' was used..
2611 return first_pop;
2612 }
2613
2614 return -1;
2615}
2616
2617static void scan_fwd_set_flags(int i, int opcnt, int magic, int flags)
2618{
2619 struct parsed_op *po;
2620 int j;
2621
2622 if (i < 0)
2623 ferr(ops, "%s: followed bad branch?\n", __func__);
2624
2625 for (; i < opcnt; i++) {
2626 po = &ops[i];
2627 if (po->cc_scratch == magic)
2628 return;
2629 po->cc_scratch = magic;
2630 po->flags |= flags;
2631
2632 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2633 if (po->btj != NULL) {
2634 // jumptable
2635 for (j = 0; j < po->btj->count; j++)
2636 scan_fwd_set_flags(po->btj->d[j].bt_i, opcnt, magic, flags);
2637 return;
2638 }
2639
2640 scan_fwd_set_flags(po->bt_i, opcnt, magic, flags);
2641 if (!(po->flags & OPF_CJMP))
2642 return;
2643 }
2644 if (po->flags & OPF_TAIL)
2645 return;
2646 }
2647}
2648
2649static const struct parsed_proto *try_recover_pp(
2650 struct parsed_op *po, const struct parsed_opr *opr, int *search_instead)
2651{
2652 const struct parsed_proto *pp = NULL;
2653 char buf[256];
2654 char *p;
2655
2656 // maybe an arg of g_func?
2657 if (opr->type == OPT_REGMEM && is_stack_access(po, opr))
2658 {
2659 char ofs_reg[16] = { 0, };
2660 int arg, arg_s, arg_i;
2661 int stack_ra = 0;
2662 int offset = 0;
2663
2664 if (g_header_mode)
2665 return NULL;
2666
2667 parse_stack_access(po, opr->name, ofs_reg,
2668 &offset, &stack_ra, NULL, 0);
2669 if (ofs_reg[0] != 0)
2670 ferr(po, "offset reg on arg access?\n");
2671 if (offset <= stack_ra) {
2672 // search who set the stack var instead
2673 if (search_instead != NULL)
2674 *search_instead = 1;
2675 return NULL;
2676 }
2677
2678 arg_i = (offset - stack_ra - 4) / 4;
2679 for (arg = arg_s = 0; arg < g_func_pp->argc; arg++) {
2680 if (g_func_pp->arg[arg].reg != NULL)
2681 continue;
2682 if (arg_s == arg_i)
2683 break;
2684 arg_s++;
2685 }
2686 if (arg == g_func_pp->argc)
2687 ferr(po, "stack arg %d not in prototype?\n", arg_i);
2688
2689 pp = g_func_pp->arg[arg].fptr;
2690 if (pp == NULL)
2691 ferr(po, "icall sa: arg%d is not a fptr?\n", arg + 1);
2692 check_func_pp(po, pp, "icall arg");
2693 }
2694 else if (opr->type == OPT_REGMEM && strchr(opr->name + 1, '[')) {
2695 // label[index]
2696 p = strchr(opr->name + 1, '[');
2697 memcpy(buf, opr->name, p - opr->name);
2698 buf[p - opr->name] = 0;
2699 pp = proto_parse(g_fhdr, buf, g_quiet_pp);
2700 }
2701 else if (opr->type == OPT_OFFSET || opr->type == OPT_LABEL) {
2702 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
2703 if (pp == NULL) {
2704 if (!g_header_mode)
2705 ferr(po, "proto_parse failed for icall to '%s'\n", opr->name);
2706 }
2707 else
2708 check_func_pp(po, pp, "reg-fptr ref");
2709 }
2710
2711 return pp;
2712}
2713
2714static void scan_for_call_type(int i, const struct parsed_opr *opr,
2715 int magic, const struct parsed_proto **pp_found, int *multi)
2716{
2717 const struct parsed_proto *pp = NULL;
2718 struct parsed_op *po;
2719 struct label_ref *lr;
2720
2721 ops[i].cc_scratch = magic;
2722
2723 while (1) {
2724 if (g_labels[i] != NULL) {
2725 lr = &g_label_refs[i];
2726 for (; lr != NULL; lr = lr->next) {
2727 check_i(&ops[i], lr->i);
2728 scan_for_call_type(lr->i, opr, magic, pp_found, multi);
2729 }
2730 if (i > 0 && LAST_OP(i - 1))
2731 return;
2732 }
2733
2734 i--;
2735 if (i < 0)
2736 break;
2737
2738 if (ops[i].cc_scratch == magic)
2739 return;
2740 ops[i].cc_scratch = magic;
2741
2742 if (!(ops[i].flags & OPF_DATA))
2743 continue;
2744 if (!is_opr_modified(opr, &ops[i]))
2745 continue;
2746 if (ops[i].op != OP_MOV && ops[i].op != OP_LEA) {
2747 // most probably trashed by some processing
2748 *pp_found = NULL;
2749 return;
2750 }
2751
2752 opr = &ops[i].operand[1];
2753 if (opr->type != OPT_REG)
2754 break;
2755 }
2756
2757 po = (i >= 0) ? &ops[i] : ops;
2758
2759 if (i < 0) {
2760 // reached the top - can only be an arg-reg
2761 if (opr->type != OPT_REG)
2762 return;
2763
2764 for (i = 0; i < g_func_pp->argc; i++) {
2765 if (g_func_pp->arg[i].reg == NULL)
2766 continue;
2767 if (IS(opr->name, g_func_pp->arg[i].reg))
2768 break;
2769 }
2770 if (i == g_func_pp->argc)
2771 return;
2772 pp = g_func_pp->arg[i].fptr;
2773 if (pp == NULL)
2774 ferr(po, "icall: arg%d (%s) is not a fptr?\n",
2775 i + 1, g_func_pp->arg[i].reg);
2776 check_func_pp(po, pp, "icall reg-arg");
2777 }
2778 else
2779 pp = try_recover_pp(po, opr, NULL);
2780
2781 if (*pp_found != NULL && pp != NULL && *pp_found != pp) {
2782 if (!IS((*pp_found)->ret_type.name, pp->ret_type.name)
2783 || (*pp_found)->is_stdcall != pp->is_stdcall
2784 || (*pp_found)->is_fptr != pp->is_fptr
2785 || (*pp_found)->argc != pp->argc
2786 || (*pp_found)->argc_reg != pp->argc_reg
2787 || (*pp_found)->argc_stack != pp->argc_stack)
2788 {
2789 ferr(po, "icall: parsed_proto mismatch\n");
2790 }
2791 *multi = 1;
2792 }
2793 if (pp != NULL)
2794 *pp_found = pp;
2795}
2796
2797// early check for tail call or branch back
2798static int is_like_tailjmp(int j)
2799{
2800 if (!(ops[j].flags & OPF_JMP))
2801 return 0;
2802
2803 if (ops[j].op == OP_JMP && !ops[j].operand[0].had_ds)
2804 // probably local branch back..
2805 return 1;
2806 if (ops[j].op == OP_CALL)
2807 // probably noreturn call..
2808 return 1;
2809
2810 return 0;
2811}
2812
2813static void scan_prologue_epilogue(int opcnt)
2814{
2815 int ecx_push = 0, esp_sub = 0;
2816 int found;
2817 int i, j, l;
2818
2819 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
2820 && ops[1].op == OP_MOV
2821 && IS(opr_name(&ops[1], 0), "ebp")
2822 && IS(opr_name(&ops[1], 1), "esp"))
2823 {
2824 g_bp_frame = 1;
2825 ops[0].flags |= OPF_RMD | OPF_DONE;
2826 ops[1].flags |= OPF_RMD | OPF_DONE;
2827 i = 2;
2828
2829 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
2830 g_stack_fsz = opr_const(&ops[2], 1);
2831 ops[2].flags |= OPF_RMD | OPF_DONE;
2832 i++;
2833 }
2834 else {
2835 // another way msvc builds stack frame..
2836 i = 2;
2837 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
2838 g_stack_fsz += 4;
2839 ops[i].flags |= OPF_RMD | OPF_DONE;
2840 ecx_push++;
2841 i++;
2842 }
2843 // and another way..
2844 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
2845 && ops[i].operand[1].type == OPT_CONST
2846 && ops[i + 1].op == OP_CALL
2847 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
2848 {
2849 g_stack_fsz += ops[i].operand[1].val;
2850 ops[i].flags |= OPF_RMD | OPF_DONE;
2851 i++;
2852 ops[i].flags |= OPF_RMD | OPF_DONE;
2853 i++;
2854 }
2855 }
2856
2857 found = 0;
2858 do {
2859 for (; i < opcnt; i++)
2860 if (ops[i].op == OP_RET)
2861 break;
2862 j = i - 1;
2863 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
2864 if (found && is_like_tailjmp(j))
2865 break;
2866 j--;
2867 }
2868
2869 if ((ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ebp"))
2870 || ops[j].op == OP_LEAVE)
2871 {
2872 ops[j].flags |= OPF_RMD | OPF_DONE;
2873 }
2874 else if (!(g_ida_func_attr & IDAFA_NORETURN))
2875 ferr(&ops[j], "'pop ebp' expected\n");
2876
2877 if (g_stack_fsz != 0) {
2878 if (ops[j - 1].op == OP_MOV
2879 && IS(opr_name(&ops[j - 1], 0), "esp")
2880 && IS(opr_name(&ops[j - 1], 1), "ebp"))
2881 {
2882 ops[j - 1].flags |= OPF_RMD | OPF_DONE;
2883 }
2884 else if (ops[j].op != OP_LEAVE
2885 && !(g_ida_func_attr & IDAFA_NORETURN))
2886 {
2887 ferr(&ops[j - 1], "esp restore expected\n");
2888 }
2889
2890 if (ecx_push && ops[j - 2].op == OP_POP
2891 && IS(opr_name(&ops[j - 2], 0), "ecx"))
2892 {
2893 ferr(&ops[j - 2], "unexpected ecx pop\n");
2894 }
2895 }
2896
2897 found = 1;
2898 i++;
2899 } while (i < opcnt);
2900
2901 return;
2902 }
2903
2904 // non-bp frame
2905 i = 0;
2906 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
2907 ops[i].flags |= OPF_RMD | OPF_DONE;
2908 g_stack_fsz += 4;
2909 ecx_push++;
2910 i++;
2911 }
2912
2913 for (; i < opcnt; i++) {
2914 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
2915 break;
2916 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
2917 && ops[i].operand[1].type == OPT_CONST)
2918 {
2919 g_stack_fsz = ops[i].operand[1].val;
2920 ops[i].flags |= OPF_RMD | OPF_DONE;
2921 esp_sub = 1;
2922 break;
2923 }
2924 }
2925
2926 if (ecx_push && !esp_sub) {
2927 // could actually be args for a call..
2928 for (; i < opcnt; i++)
2929 if (ops[i].op != OP_PUSH)
2930 break;
2931
2932 if (ops[i].op == OP_CALL && ops[i].operand[0].type == OPT_LABEL) {
2933 const struct parsed_proto *pp;
2934 pp = proto_parse(g_fhdr, opr_name(&ops[i], 0), 1);
2935 j = pp ? pp->argc_stack : 0;
2936 while (i > 0 && j > 0) {
2937 i--;
2938 if (ops[i].op == OP_PUSH) {
2939 ops[i].flags &= ~(OPF_RMD | OPF_DONE);
2940 j--;
2941 }
2942 }
2943 if (j != 0)
2944 ferr(&ops[i], "unhandled prologue\n");
2945
2946 // recheck
2947 i = g_stack_fsz = ecx_push = 0;
2948 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
2949 if (!(ops[i].flags & OPF_RMD))
2950 break;
2951 g_stack_fsz += 4;
2952 ecx_push++;
2953 i++;
2954 }
2955 }
2956 }
2957
2958 found = 0;
2959 if (ecx_push || esp_sub)
2960 {
2961 g_sp_frame = 1;
2962
2963 i++;
2964 do {
2965 for (; i < opcnt; i++)
2966 if (ops[i].op == OP_RET)
2967 break;
2968 j = i - 1;
2969 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
2970 if (found && is_like_tailjmp(j))
2971 break;
2972 j--;
2973 }
2974
2975 if (ecx_push > 0) {
2976 for (l = 0; l < ecx_push; l++) {
2977 if (ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ecx"))
2978 /* pop ecx */;
2979 else if (ops[j].op == OP_ADD
2980 && IS(opr_name(&ops[j], 0), "esp")
2981 && ops[j].operand[1].type == OPT_CONST)
2982 {
2983 /* add esp, N */
2984 l += ops[j].operand[1].val / 4 - 1;
2985 }
2986 else
2987 ferr(&ops[j], "'pop ecx' expected\n");
2988
2989 ops[j].flags |= OPF_RMD | OPF_DONE;
2990 j--;
2991 }
2992 if (l != ecx_push)
2993 ferr(&ops[j], "epilogue scan failed\n");
2994
2995 found = 1;
2996 }
2997
2998 if (esp_sub) {
2999 if (ops[j].op != OP_ADD
3000 || !IS(opr_name(&ops[j], 0), "esp")
3001 || ops[j].operand[1].type != OPT_CONST
3002 || ops[j].operand[1].val != g_stack_fsz)
3003 ferr(&ops[j], "'add esp' expected\n");
3004
3005 ops[j].flags |= OPF_RMD | OPF_DONE;
3006 ops[j].operand[1].val = 0; // hack for stack arg scanner
3007 found = 1;
3008 }
3009
3010 i++;
3011 } while (i < opcnt);
3012 }
3013}
3014
3015static const struct parsed_proto *resolve_icall(int i, int opcnt,
3016 int *multi_src)
3017{
3018 const struct parsed_proto *pp = NULL;
3019 int search_advice = 0;
3020
3021 *multi_src = 0;
3022
3023 switch (ops[i].operand[0].type) {
3024 case OPT_REGMEM:
3025 case OPT_LABEL:
3026 case OPT_OFFSET:
3027 pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice);
3028 if (!search_advice)
3029 break;
3030 // fallthrough
3031 default:
3032 scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp,
3033 multi_src);
3034 break;
3035 }
3036
3037 return pp;
3038}
3039
3040// find an instruction that changed opr before i op
3041// *op_i must be set to -1 by caller
3042// *entry is set to 1 if one source is determined to be the caller
3043// returns 1 if found, *op_i is then set to origin
3044static int resolve_origin(int i, const struct parsed_opr *opr,
3045 int magic, int *op_i, int *is_caller)
3046{
3047 struct label_ref *lr;
3048 int ret = 0;
3049
3050 if (ops[i].cc_scratch == magic)
3051 return 0;
3052 ops[i].cc_scratch = magic;
3053
3054 while (1) {
3055 if (g_labels[i] != NULL) {
3056 lr = &g_label_refs[i];
3057 for (; lr != NULL; lr = lr->next) {
3058 check_i(&ops[i], lr->i);
3059 ret |= resolve_origin(lr->i, opr, magic, op_i, is_caller);
3060 }
3061 if (i > 0 && LAST_OP(i - 1))
3062 return ret;
3063 }
3064
3065 i--;
3066 if (i < 0) {
3067 if (is_caller != NULL)
3068 *is_caller = 1;
3069 return -1;
3070 }
3071
3072 if (ops[i].cc_scratch == magic)
3073 return 0;
3074 ops[i].cc_scratch = magic;
3075
3076 if (!(ops[i].flags & OPF_DATA))
3077 continue;
3078 if (!is_opr_modified(opr, &ops[i]))
3079 continue;
3080
3081 if (*op_i >= 0) {
3082 if (*op_i == i)
3083 return 1;
3084 // XXX: could check if the other op does the same
3085 return -1;
3086 }
3087
3088 *op_i = i;
3089 return 1;
3090 }
3091}
3092
3093static int try_resolve_const(int i, const struct parsed_opr *opr,
3094 int magic, unsigned int *val)
3095{
3096 int s_i = -1;
3097 int ret;
3098
3099 ret = resolve_origin(i, opr, magic, &s_i, NULL);
3100 if (ret == 1) {
3101 i = s_i;
3102 if (ops[i].op != OP_MOV && ops[i].operand[1].type != OPT_CONST)
3103 return -1;
3104
3105 *val = ops[i].operand[1].val;
3106 return 1;
3107 }
3108
3109 return -1;
3110}
3111
3112static struct parsed_proto *process_call_early(int i, int opcnt,
3113 int *adj_i)
3114{
3115 struct parsed_op *po = &ops[i];
3116 struct parsed_proto *pp;
3117 int multipath = 0;
3118 int adj = 0;
3119 int ret;
3120
3121 pp = po->pp;
3122 if (pp == NULL || pp->is_vararg || pp->argc_reg != 0)
3123 // leave for later
3124 return NULL;
3125
3126 // look for and make use of esp adjust
3127 *adj_i = ret = -1;
3128 if (!pp->is_stdcall && pp->argc_stack > 0)
3129 ret = scan_for_esp_adjust(i + 1, opcnt,
3130 pp->argc_stack * 4, &adj, &multipath, 0);
3131 if (ret >= 0) {
3132 if (pp->argc_stack > adj / 4)
3133 return NULL;
3134 if (multipath)
3135 return NULL;
3136 if (ops[ret].op == OP_POP && adj != 4)
3137 return NULL;
3138 }
3139
3140 *adj_i = ret;
3141 return pp;
3142}
3143
3144static struct parsed_proto *process_call(int i, int opcnt)
3145{
3146 struct parsed_op *po = &ops[i];
3147 const struct parsed_proto *pp_c;
3148 struct parsed_proto *pp;
3149 const char *tmpname;
3150 int adj = 0, multipath = 0;
3151 int ret, arg;
3152
3153 tmpname = opr_name(po, 0);
3154 pp = po->pp;
3155 if (pp == NULL)
3156 {
3157 // indirect call
3158 pp_c = resolve_icall(i, opcnt, &multipath);
3159 if (pp_c != NULL) {
3160 if (!pp_c->is_func && !pp_c->is_fptr)
3161 ferr(po, "call to non-func: %s\n", pp_c->name);
3162 pp = proto_clone(pp_c);
3163 my_assert_not(pp, NULL);
3164 if (multipath)
3165 // not resolved just to single func
3166 pp->is_fptr = 1;
3167
3168 switch (po->operand[0].type) {
3169 case OPT_REG:
3170 // we resolved this call and no longer need the register
3171 po->regmask_src &= ~(1 << po->operand[0].reg);
3172 break;
3173 case OPT_REGMEM:
3174 pp->is_fptr = 1;
3175 break;
3176 default:
3177 break;
3178 }
3179 }
3180 if (pp == NULL) {
3181 pp = calloc(1, sizeof(*pp));
3182 my_assert_not(pp, NULL);
3183
3184 pp->is_fptr = 1;
3185 ret = scan_for_esp_adjust(i + 1, opcnt,
3186 32*4, &adj, &multipath, 0);
3187 if (ret < 0 || adj < 0) {
3188 if (!g_allow_regfunc)
3189 ferr(po, "non-__cdecl indirect call unhandled yet\n");
3190 pp->is_unresolved = 1;
3191 adj = 0;
3192 }
3193 adj /= 4;
3194 if (adj > ARRAY_SIZE(pp->arg))
3195 ferr(po, "esp adjust too large: %d\n", adj);
3196 pp->ret_type.name = strdup("int");
3197 pp->argc = pp->argc_stack = adj;
3198 for (arg = 0; arg < pp->argc; arg++)
3199 pp->arg[arg].type.name = strdup("int");
3200 }
3201 po->pp = pp;
3202 }
3203
3204 // look for and make use of esp adjust
3205 multipath = 0;
3206 ret = -1;
3207 if (!pp->is_stdcall && pp->argc_stack > 0)
3208 ret = scan_for_esp_adjust(i + 1, opcnt,
3209 pp->argc_stack * 4, &adj, &multipath, 0);
3210 if (ret >= 0) {
3211 if (pp->is_vararg) {
3212 if (adj / 4 < pp->argc_stack) {
3213 fnote(po, "(this call)\n");
3214 ferr(&ops[ret], "esp adjust is too small: %x < %x\n",
3215 adj, pp->argc_stack * 4);
3216 }
3217 // modify pp to make it have varargs as normal args
3218 arg = pp->argc;
3219 pp->argc += adj / 4 - pp->argc_stack;
3220 for (; arg < pp->argc; arg++) {
3221 pp->arg[arg].type.name = strdup("int");
3222 pp->argc_stack++;
3223 }
3224 if (pp->argc > ARRAY_SIZE(pp->arg))
3225 ferr(po, "too many args for '%s'\n", tmpname);
3226 }
3227 if (pp->argc_stack > adj / 4) {
3228 fnote(po, "(this call)\n");
3229 ferr(&ops[ret], "stack tracking failed for '%s': %x %x\n",
3230 tmpname, pp->argc_stack * 4, adj);
3231 }
3232
3233 scan_for_esp_adjust(i + 1, opcnt,
3234 pp->argc_stack * 4, &adj, &multipath, 1);
3235 }
3236 else if (pp->is_vararg)
3237 ferr(po, "missing esp_adjust for vararg func '%s'\n",
3238 pp->name);
3239
3240 return pp;
3241}
3242
3243static int collect_call_args_early(struct parsed_op *po, int i,
3244 struct parsed_proto *pp, int *regmask)
3245{
3246 int arg, ret;
3247 int j;
3248
3249 for (arg = 0; arg < pp->argc; arg++)
3250 if (pp->arg[arg].reg == NULL)
3251 break;
3252
3253 // first see if it can be easily done
3254 for (j = i; j > 0 && arg < pp->argc; )
3255 {
3256 if (g_labels[j] != NULL)
3257 return -1;
3258 j--;
3259
3260 if (ops[j].op == OP_CALL)
3261 return -1;
3262 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP)
3263 return -1;
3264 else if (ops[j].op == OP_POP)
3265 return -1;
3266 else if (ops[j].flags & OPF_CJMP)
3267 return -1;
3268 else if (ops[j].op == OP_PUSH) {
3269 if (ops[j].flags & (OPF_FARG|OPF_FARGNR))
3270 return -1;
3271 ret = scan_for_mod(&ops[j], j + 1, i, 1);
3272 if (ret >= 0)
3273 return -1;
3274
3275 if (pp->arg[arg].type.is_va_list)
3276 return -1;
3277
3278 // next arg
3279 for (arg++; arg < pp->argc; arg++)
3280 if (pp->arg[arg].reg == NULL)
3281 break;
3282 }
3283 }
3284
3285 if (arg < pp->argc)
3286 return -1;
3287
3288 // now do it
3289 for (arg = 0; arg < pp->argc; arg++)
3290 if (pp->arg[arg].reg == NULL)
3291 break;
3292
3293 for (j = i; j > 0 && arg < pp->argc; )
3294 {
3295 j--;
3296
3297 if (ops[j].op == OP_PUSH)
3298 {
3299 ops[j].p_argnext = -1;
3300 ferr_assert(&ops[j], pp->arg[arg].datap == NULL);
3301 pp->arg[arg].datap = &ops[j];
3302
3303 if (ops[j].operand[0].type == OPT_REG)
3304 *regmask |= 1 << ops[j].operand[0].reg;
3305
3306 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_FARGNR | OPF_FARG;
3307 ops[j].flags &= ~OPF_RSAVE;
3308
3309 // next arg
3310 for (arg++; arg < pp->argc; arg++)
3311 if (pp->arg[arg].reg == NULL)
3312 break;
3313 }
3314 }
3315
3316 return 0;
3317}
3318
3319static int collect_call_args_r(struct parsed_op *po, int i,
3320 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
3321 int *arg_grp, int arg, int magic, int need_op_saving, int may_reuse)
3322{
3323 struct parsed_proto *pp_tmp;
3324 struct parsed_op *po_tmp;
3325 struct label_ref *lr;
3326 int need_to_save_current;
3327 int arg_grp_current = 0;
3328 int save_args_seen = 0;
3329 int save_args;
3330 int ret = 0;
3331 int reg;
3332 char buf[32];
3333 int j, k;
3334
3335 if (i < 0) {
3336 ferr(po, "dead label encountered\n");
3337 return -1;
3338 }
3339
3340 for (; arg < pp->argc; arg++)
3341 if (pp->arg[arg].reg == NULL)
3342 break;
3343 magic = (magic & 0xffffff) | (arg << 24);
3344
3345 for (j = i; j >= 0 && (arg < pp->argc || pp->is_unresolved); )
3346 {
3347 if (((ops[j].cc_scratch ^ magic) & 0xffffff) == 0) {
3348 if (ops[j].cc_scratch != magic) {
3349 ferr(&ops[j], "arg collect hit same path with diff args for %s\n",
3350 pp->name);
3351 return -1;
3352 }
3353 // ok: have already been here
3354 return 0;
3355 }
3356 ops[j].cc_scratch = magic;
3357
3358 if (g_labels[j] != NULL && g_label_refs[j].i != -1) {
3359 lr = &g_label_refs[j];
3360 if (lr->next != NULL)
3361 need_op_saving = 1;
3362 for (; lr->next; lr = lr->next) {
3363 check_i(&ops[j], lr->i);
3364 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
3365 may_reuse = 1;
3366 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
3367 arg_grp, arg, magic, need_op_saving, may_reuse);
3368 if (ret < 0)
3369 return ret;
3370 }
3371
3372 check_i(&ops[j], lr->i);
3373 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
3374 may_reuse = 1;
3375 if (j > 0 && LAST_OP(j - 1)) {
3376 // follow last branch in reverse
3377 j = lr->i;
3378 continue;
3379 }
3380 need_op_saving = 1;
3381 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
3382 arg_grp, arg, magic, need_op_saving, may_reuse);
3383 if (ret < 0)
3384 return ret;
3385 }
3386 j--;
3387
3388 if (ops[j].op == OP_CALL)
3389 {
3390 if (pp->is_unresolved)
3391 break;
3392
3393 pp_tmp = ops[j].pp;
3394 if (pp_tmp == NULL)
3395 ferr(po, "arg collect hit unparsed call '%s'\n",
3396 ops[j].operand[0].name);
3397 if (may_reuse && pp_tmp->argc_stack > 0)
3398 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
3399 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
3400 }
3401 // esp adjust of 0 means we collected it before
3402 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP
3403 && (ops[j].operand[1].type != OPT_CONST
3404 || ops[j].operand[1].val != 0))
3405 {
3406 if (pp->is_unresolved)
3407 break;
3408
3409 ferr(po, "arg collect %d/%d hit esp adjust of %d\n",
3410 arg, pp->argc, ops[j].operand[1].val);
3411 }
3412 else if (ops[j].op == OP_POP && !(ops[j].flags & OPF_DONE))
3413 {
3414 if (pp->is_unresolved)
3415 break;
3416
3417 ferr(po, "arg collect %d/%d hit pop\n", arg, pp->argc);
3418 }
3419 else if (ops[j].flags & OPF_CJMP)
3420 {
3421 if (pp->is_unresolved)
3422 break;
3423
3424 may_reuse = 1;
3425 }
3426 else if (ops[j].op == OP_PUSH
3427 && !(ops[j].flags & (OPF_FARGNR|OPF_DONE)))
3428 {
3429 if (pp->is_unresolved && (ops[j].flags & OPF_RMD))
3430 break;
3431
3432 ops[j].p_argnext = -1;
3433 po_tmp = pp->arg[arg].datap;
3434 if (po_tmp != NULL)
3435 ops[j].p_argnext = po_tmp - ops;
3436 pp->arg[arg].datap = &ops[j];
3437
3438 need_to_save_current = 0;
3439 save_args = 0;
3440 reg = -1;
3441 if (ops[j].operand[0].type == OPT_REG)
3442 reg = ops[j].operand[0].reg;
3443
3444 if (!need_op_saving) {
3445 ret = scan_for_mod(&ops[j], j + 1, i, 1);
3446 need_to_save_current = (ret >= 0);
3447 }
3448 if (need_op_saving || need_to_save_current) {
3449 // mark this push as one that needs operand saving
3450 ops[j].flags &= ~OPF_RMD;
3451 if (ops[j].p_argnum == 0) {
3452 ops[j].p_argnum = arg + 1;
3453 save_args |= 1 << arg;
3454 }
3455 else if (ops[j].p_argnum < arg + 1) {
3456 // XXX: might kill valid var..
3457 //*save_arg_vars &= ~(1 << (ops[j].p_argnum - 1));
3458 ops[j].p_argnum = arg + 1;
3459 save_args |= 1 << arg;
3460 }
3461
3462 if (save_args_seen & (1 << (ops[j].p_argnum - 1))) {
3463 save_args_seen = 0;
3464 arg_grp_current++;
3465 if (arg_grp_current >= MAX_ARG_GRP)
3466 ferr(&ops[j], "out of arg groups (arg%d), f %s\n",
3467 ops[j].p_argnum, pp->name);
3468 }
3469 }
3470 else if (ops[j].p_argnum == 0)
3471 ops[j].flags |= OPF_RMD;
3472
3473 // some PUSHes are reused by different calls on other branches,
3474 // but that can't happen if we didn't branch, so they
3475 // can be removed from future searches (handles nested calls)
3476 if (!may_reuse)
3477 ops[j].flags |= OPF_FARGNR;
3478
3479 ops[j].flags |= OPF_FARG;
3480 ops[j].flags &= ~OPF_RSAVE;
3481
3482 // check for __VALIST
3483 if (!pp->is_unresolved && pp->arg[arg].type.is_va_list) {
3484 k = -1;
3485 ret = resolve_origin(j, &ops[j].operand[0],
3486 magic + 1, &k, NULL);
3487 if (ret == 1 && k >= 0)
3488 {
3489 if (ops[k].op == OP_LEA) {
3490 snprintf(buf, sizeof(buf), "arg_%X",
3491 g_func_pp->argc_stack * 4);
3492 if (!g_func_pp->is_vararg
3493 || strstr(ops[k].operand[1].name, buf))
3494 {
3495 ops[k].flags |= OPF_RMD | OPF_DONE;
3496 ops[j].flags |= OPF_RMD | OPF_VAPUSH;
3497 save_args &= ~(1 << arg);
3498 reg = -1;
3499 }
3500 else
3501 ferr(&ops[j], "lea va_list used, but no vararg?\n");
3502 }
3503 // check for va_list from g_func_pp arg too
3504 else if (ops[k].op == OP_MOV
3505 && is_stack_access(&ops[k], &ops[k].operand[1]))
3506 {
3507 ret = stack_frame_access(&ops[k], &ops[k].operand[1],
3508 buf, sizeof(buf), ops[k].operand[1].name, "", 1, 0);
3509 if (ret >= 0) {
3510 ops[k].flags |= OPF_RMD | OPF_DONE;
3511 ops[j].flags |= OPF_RMD;
3512 ops[j].p_argpass = ret + 1;
3513 save_args &= ~(1 << arg);
3514 reg = -1;
3515 }
3516 }
3517 }
3518 }
3519
3520 *save_arg_vars |= save_args;
3521
3522 // tracking reg usage
3523 if (reg >= 0)
3524 *regmask |= 1 << reg;
3525
3526 arg++;
3527 if (!pp->is_unresolved) {
3528 // next arg
3529 for (; arg < pp->argc; arg++)
3530 if (pp->arg[arg].reg == NULL)
3531 break;
3532 }
3533 magic = (magic & 0xffffff) | (arg << 24);
3534 }
3535
3536 if (ops[j].p_arggrp > arg_grp_current) {
3537 save_args_seen = 0;
3538 arg_grp_current = ops[j].p_arggrp;
3539 }
3540 if (ops[j].p_argnum > 0)
3541 save_args_seen |= 1 << (ops[j].p_argnum - 1);
3542 }
3543
3544 if (arg < pp->argc) {
3545 ferr(po, "arg collect failed for '%s': %d/%d\n",
3546 pp->name, arg, pp->argc);
3547 return -1;
3548 }
3549
3550 if (arg_grp_current > *arg_grp)
3551 *arg_grp = arg_grp_current;
3552
3553 return arg;
3554}
3555
3556static int collect_call_args(struct parsed_op *po, int i,
3557 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
3558 int magic)
3559{
3560 // arg group is for cases when pushes for
3561 // multiple funcs are going on
3562 struct parsed_op *po_tmp;
3563 int save_arg_vars_current = 0;
3564 int arg_grp = 0;
3565 int ret;
3566 int a;
3567
3568 ret = collect_call_args_r(po, i, pp, regmask,
3569 &save_arg_vars_current, &arg_grp, 0, magic, 0, 0);
3570 if (ret < 0)
3571 return ret;
3572
3573 if (arg_grp != 0) {
3574 // propagate arg_grp
3575 for (a = 0; a < pp->argc; a++) {
3576 if (pp->arg[a].reg != NULL)
3577 continue;
3578
3579 po_tmp = pp->arg[a].datap;
3580 while (po_tmp != NULL) {
3581 po_tmp->p_arggrp = arg_grp;
3582 if (po_tmp->p_argnext > 0)
3583 po_tmp = &ops[po_tmp->p_argnext];
3584 else
3585 po_tmp = NULL;
3586 }
3587 }
3588 }
3589 save_arg_vars[arg_grp] |= save_arg_vars_current;
3590
3591 if (pp->is_unresolved) {
3592 pp->argc += ret;
3593 pp->argc_stack += ret;
3594 for (a = 0; a < pp->argc; a++)
3595 if (pp->arg[a].type.name == NULL)
3596 pp->arg[a].type.name = strdup("int");
3597 }
3598
3599 return ret;
3600}
3601
3602static void pp_insert_reg_arg(struct parsed_proto *pp, const char *reg)
3603{
3604 int i;
3605
3606 for (i = 0; i < pp->argc; i++)
3607 if (pp->arg[i].reg == NULL)
3608 break;
3609
3610 if (pp->argc_stack)
3611 memmove(&pp->arg[i + 1], &pp->arg[i],
3612 sizeof(pp->arg[0]) * pp->argc_stack);
3613 memset(&pp->arg[i], 0, sizeof(pp->arg[i]));
3614 pp->arg[i].reg = strdup(reg);
3615 pp->arg[i].type.name = strdup("int");
3616 pp->argc++;
3617 pp->argc_reg++;
3618}
3619
3620static void add_label_ref(struct label_ref *lr, int op_i)
3621{
3622 struct label_ref *lr_new;
3623
3624 if (lr->i == -1) {
3625 lr->i = op_i;
3626 return;
3627 }
3628
3629 lr_new = calloc(1, sizeof(*lr_new));
3630 lr_new->i = op_i;
3631 lr_new->next = lr->next;
3632 lr->next = lr_new;
3633}
3634
3635static struct parsed_data *try_resolve_jumptab(int i, int opcnt)
3636{
3637 struct parsed_op *po = &ops[i];
3638 struct parsed_data *pd;
3639 char label[NAMELEN], *p;
3640 int len, j, l;
3641
3642 p = strchr(po->operand[0].name, '[');
3643 if (p == NULL)
3644 return NULL;
3645
3646 len = p - po->operand[0].name;
3647 strncpy(label, po->operand[0].name, len);
3648 label[len] = 0;
3649
3650 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
3651 if (IS(g_func_pd[j].label, label)) {
3652 pd = &g_func_pd[j];
3653 break;
3654 }
3655 }
3656 if (pd == NULL)
3657 //ferr(po, "label '%s' not parsed?\n", label);
3658 return NULL;
3659
3660 if (pd->type != OPT_OFFSET)
3661 ferr(po, "label '%s' with non-offset data?\n", label);
3662
3663 // find all labels, link
3664 for (j = 0; j < pd->count; j++) {
3665 for (l = 0; l < opcnt; l++) {
3666 if (g_labels[l] != NULL && IS(g_labels[l], pd->d[j].u.label)) {
3667 add_label_ref(&g_label_refs[l], i);
3668 pd->d[j].bt_i = l;
3669 break;
3670 }
3671 }
3672 }
3673
3674 return pd;
3675}
3676
3677static void clear_labels(int count)
3678{
3679 int i;
3680
3681 for (i = 0; i < count; i++) {
3682 if (g_labels[i] != NULL) {
3683 free(g_labels[i]);
3684 g_labels[i] = NULL;
3685 }
3686 }
3687}
3688
3689static void output_std_flags(FILE *fout, struct parsed_op *po,
3690 int *pfomask, const char *dst_opr_text)
3691{
3692 if (*pfomask & (1 << PFO_Z)) {
3693 fprintf(fout, "\n cond_z = (%s%s == 0);",
3694 lmod_cast_u(po, po->operand[0].lmod), dst_opr_text);
3695 *pfomask &= ~(1 << PFO_Z);
3696 }
3697 if (*pfomask & (1 << PFO_S)) {
3698 fprintf(fout, "\n cond_s = (%s%s < 0);",
3699 lmod_cast_s(po, po->operand[0].lmod), dst_opr_text);
3700 *pfomask &= ~(1 << PFO_S);
3701 }
3702}
3703
3704static void output_pp_attrs(FILE *fout, const struct parsed_proto *pp,
3705 int is_noreturn)
3706{
3707 if (pp->is_fastcall)
3708 fprintf(fout, "__fastcall ");
3709 else if (pp->is_stdcall && pp->argc_reg == 0)
3710 fprintf(fout, "__stdcall ");
3711 if (pp->is_noreturn || is_noreturn)
3712 fprintf(fout, "noreturn ");
3713}
3714
3715static int get_pp_arg_regmask(const struct parsed_proto *pp)
3716{
3717 int regmask = 0;
3718 int i, reg;
3719
3720 for (i = 0; i < pp->argc; i++) {
3721 if (pp->arg[i].reg != NULL) {
3722 reg = char_array_i(regs_r32,
3723 ARRAY_SIZE(regs_r32), pp->arg[i].reg);
3724 if (reg < 0)
3725 ferr(ops, "arg '%s' of func '%s' is not a reg?\n",
3726 pp->arg[i].reg, pp->name);
3727 regmask |= 1 << reg;
3728 }
3729 }
3730
3731 return regmask;
3732}
3733
3734static char *saved_arg_name(char *buf, size_t buf_size, int grp, int num)
3735{
3736 char buf1[16];
3737
3738 buf1[0] = 0;
3739 if (grp > 0)
3740 snprintf(buf1, sizeof(buf1), "%d", grp);
3741 snprintf(buf, buf_size, "s%s_a%d", buf1, num);
3742
3743 return buf;
3744}
3745
3746static void gen_x_cleanup(int opcnt);
3747
3748static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
3749{
3750 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
3751 struct parsed_opr *last_arith_dst = NULL;
3752 char buf1[256], buf2[256], buf3[256], cast[64];
3753 const struct parsed_proto *pp_c;
3754 struct parsed_proto *pp, *pp_tmp;
3755 struct parsed_data *pd;
3756 const char *tmpname;
3757 unsigned int uval;
3758 int save_arg_vars[MAX_ARG_GRP] = { 0, };
3759 int cond_vars = 0;
3760 int need_tmp_var = 0;
3761 int need_tmp64 = 0;
3762 int had_decl = 0;
3763 int label_pending = 0;
3764 int regmask_save = 0;
3765 int regmask_arg = 0;
3766 int regmask_now = 0;
3767 int regmask_init = 0;
3768 int regmask = 0;
3769 int pfomask = 0;
3770 int found = 0;
3771 int depth = 0;
3772 int no_output;
3773 int i, j, l;
3774 int arg;
3775 int reg;
3776 int ret;
3777
3778 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
3779 g_stack_frame_used = 0;
3780
3781 g_func_pp = proto_parse(fhdr, funcn, 0);
3782 if (g_func_pp == NULL)
3783 ferr(ops, "proto_parse failed for '%s'\n", funcn);
3784
3785 regmask_arg = get_pp_arg_regmask(g_func_pp);
3786
3787 // pass1:
3788 // - handle ebp/esp frame, remove ops related to it
3789 scan_prologue_epilogue(opcnt);
3790
3791 // pass2:
3792 // - parse calls with labels
3793 // - resolve all branches
3794 for (i = 0; i < opcnt; i++)
3795 {
3796 po = &ops[i];
3797 po->bt_i = -1;
3798 po->btj = NULL;
3799
3800 if (po->flags & (OPF_RMD|OPF_DONE))
3801 continue;
3802
3803 if (po->op == OP_CALL) {
3804 pp = NULL;
3805
3806 if (po->operand[0].type == OPT_LABEL) {
3807 tmpname = opr_name(po, 0);
3808 if (IS_START(tmpname, "loc_"))
3809 ferr(po, "call to loc_*\n");
3810 pp_c = proto_parse(fhdr, tmpname, 0);
3811 if (pp_c == NULL)
3812 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
3813
3814 pp = proto_clone(pp_c);
3815 my_assert_not(pp, NULL);
3816 }
3817 else if (po->datap != NULL) {
3818 pp = calloc(1, sizeof(*pp));
3819 my_assert_not(pp, NULL);
3820
3821 ret = parse_protostr(po->datap, pp);
3822 if (ret < 0)
3823 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
3824 free(po->datap);
3825 po->datap = NULL;
3826 }
3827
3828 if (pp != NULL) {
3829 if (pp->is_fptr)
3830 check_func_pp(po, pp, "fptr var call");
3831 if (pp->is_noreturn)
3832 po->flags |= OPF_TAIL;
3833 }
3834 po->pp = pp;
3835 continue;
3836 }
3837
3838 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
3839 continue;
3840
3841 if (po->operand[0].type == OPT_REGMEM) {
3842 pd = try_resolve_jumptab(i, opcnt);
3843 if (pd == NULL)
3844 goto tailcall;
3845
3846 po->btj = pd;
3847 continue;
3848 }
3849
3850 for (l = 0; l < opcnt; l++) {
3851 if (g_labels[l] != NULL
3852 && IS(po->operand[0].name, g_labels[l]))
3853 {
3854 if (l == i + 1 && po->op == OP_JMP) {
3855 // yet another alignment type..
3856 po->flags |= OPF_RMD|OPF_DONE;
3857 break;
3858 }
3859 add_label_ref(&g_label_refs[l], i);
3860 po->bt_i = l;
3861 break;
3862 }
3863 }
3864
3865 if (po->bt_i != -1 || (po->flags & OPF_RMD))
3866 continue;
3867
3868 if (po->operand[0].type == OPT_LABEL)
3869 // assume tail call
3870 goto tailcall;
3871
3872 ferr(po, "unhandled branch\n");
3873
3874tailcall:
3875 po->op = OP_CALL;
3876 po->flags |= OPF_TAIL;
3877 if (i > 0 && ops[i - 1].op == OP_POP)
3878 po->flags |= OPF_ATAIL;
3879 i--; // reprocess
3880 }
3881
3882 // pass3:
3883 // - remove dead labels
3884 // - process trivial calls
3885 for (i = 0; i < opcnt; i++)
3886 {
3887 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
3888 free(g_labels[i]);
3889 g_labels[i] = NULL;
3890 }
3891
3892 po = &ops[i];
3893 if (po->flags & (OPF_RMD|OPF_DONE))
3894 continue;
3895
3896 if (po->op == OP_CALL)
3897 {
3898 pp = process_call_early(i, opcnt, &j);
3899 if (pp != NULL) {
3900 if (!(po->flags & OPF_ATAIL))
3901 // since we know the args, try to collect them
3902 if (collect_call_args_early(po, i, pp, &regmask) != 0)
3903 pp = NULL;
3904 }
3905
3906 if (pp != NULL) {
3907 if (j >= 0) {
3908 // commit esp adjust
3909 ops[j].flags |= OPF_RMD;
3910 if (ops[j].op != OP_POP)
3911 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
3912 else
3913 ops[j].flags |= OPF_DONE;
3914 }
3915
3916 if (strstr(pp->ret_type.name, "int64"))
3917 need_tmp64 = 1;
3918
3919 po->flags |= OPF_DONE;
3920 }
3921 }
3922 }
3923
3924 // pass4:
3925 // - process calls
3926 // - handle push <const>/pop pairs
3927 for (i = 0; i < opcnt; i++)
3928 {
3929 po = &ops[i];
3930 if (po->flags & (OPF_RMD|OPF_DONE))
3931 continue;
3932
3933 if (po->op == OP_CALL && !(po->flags & OPF_DONE))
3934 {
3935 pp = process_call(i, opcnt);
3936
3937 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
3938 // since we know the args, collect them
3939 collect_call_args(po, i, pp, &regmask, save_arg_vars,
3940 i + opcnt * 2);
3941 }
3942
3943 if (strstr(pp->ret_type.name, "int64"))
3944 need_tmp64 = 1;
3945 }
3946 else if (po->op == OP_PUSH && !(po->flags & OPF_FARG)
3947 && !(po->flags & OPF_RSAVE) && po->operand[0].type == OPT_CONST)
3948 scan_for_pop_const(i, opcnt);
3949 }
3950
3951 // pass5:
3952 // - find POPs for PUSHes, rm both
3953 // - scan for STD/CLD, propagate DF
3954 // - scan for all used registers
3955 // - find flag set ops for their users
3956 // - do unreselved calls
3957 // - declare indirect functions
3958 for (i = 0; i < opcnt; i++)
3959 {
3960 po = &ops[i];
3961 if (po->flags & (OPF_RMD|OPF_DONE))
3962 continue;
3963
3964 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE)) {
3965 reg = po->operand[0].reg;
3966 if (!(regmask & (1 << reg)))
3967 // not a reg save after all, rerun scan_for_pop
3968 po->flags &= ~OPF_RSAVE;
3969 else
3970 regmask_save |= 1 << reg;
3971 }
3972
3973 if (po->op == OP_PUSH && !(po->flags & OPF_FARG)
3974 && !(po->flags & OPF_RSAVE) && !g_func_pp->is_userstack)
3975 {
3976 if (po->operand[0].type == OPT_REG)
3977 {
3978 reg = po->operand[0].reg;
3979 if (reg < 0)
3980 ferr(po, "reg not set for push?\n");
3981
3982 depth = 0;
3983 ret = scan_for_pop(i + 1, opcnt,
3984 po->operand[0].name, i + opcnt * 3, 0, &depth, 0);
3985 if (ret == 1) {
3986 if (depth > 1)
3987 ferr(po, "too much depth: %d\n", depth);
3988
3989 po->flags |= OPF_RMD;
3990 scan_for_pop(i + 1, opcnt, po->operand[0].name,
3991 i + opcnt * 4, 0, &depth, 1);
3992 continue;
3993 }
3994 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
3995 if (ret == 0) {
3996 arg = OPF_RMD;
3997 if (regmask & (1 << reg)) {
3998 if (regmask_save & (1 << reg))
3999 ferr(po, "%s already saved?\n", po->operand[0].name);
4000 arg = OPF_RSAVE;
4001 }
4002 po->flags |= arg;
4003 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg);
4004 continue;
4005 }
4006 }
4007 }
4008
4009 if (po->op == OP_STD) {
4010 po->flags |= OPF_DF | OPF_RMD | OPF_DONE;
4011 scan_propagate_df(i + 1, opcnt);
4012 }
4013
4014 regmask_now = po->regmask_src | po->regmask_dst;
4015 if (regmask_now & (1 << xBP)) {
4016 if (g_bp_frame && !(po->flags & OPF_EBP_S)) {
4017 if (po->regmask_dst & (1 << xBP))
4018 // compiler decided to drop bp frame and use ebp as scratch
4019 scan_fwd_set_flags(i + 1, opcnt, i + opcnt * 5, OPF_EBP_S);
4020 else
4021 regmask_now &= ~(1 << xBP);
4022 }
4023 }
4024
4025 regmask |= regmask_now;
4026
4027 if (po->flags & OPF_CC)
4028 {
4029 int setters[16], cnt = 0, branched = 0;
4030
4031 ret = scan_for_flag_set(i, i + opcnt * 6,
4032 &branched, setters, &cnt);
4033 if (ret < 0 || cnt <= 0)
4034 ferr(po, "unable to trace flag setter(s)\n");
4035 if (cnt > ARRAY_SIZE(setters))
4036 ferr(po, "too many flag setters\n");
4037
4038 for (j = 0; j < cnt; j++)
4039 {
4040 tmp_op = &ops[setters[j]]; // flag setter
4041 pfomask = 0;
4042
4043 // to get nicer code, we try to delay test and cmp;
4044 // if we can't because of operand modification, or if we
4045 // have arith op, or branch, make it calculate flags explicitly
4046 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP)
4047 {
4048 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i, 0) >= 0)
4049 pfomask = 1 << po->pfo;
4050 }
4051 else if (tmp_op->op == OP_CMPS || tmp_op->op == OP_SCAS) {
4052 pfomask = 1 << po->pfo;
4053 }
4054 else {
4055 // see if we'll be able to handle based on op result
4056 if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR
4057 && po->pfo != PFO_Z && po->pfo != PFO_S
4058 && po->pfo != PFO_P)
4059 || branched
4060 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
4061 {
4062 pfomask = 1 << po->pfo;
4063 }
4064
4065 if (tmp_op->op == OP_ADD && po->pfo == PFO_C) {
4066 propagate_lmod(tmp_op, &tmp_op->operand[0],
4067 &tmp_op->operand[1]);
4068 if (tmp_op->operand[0].lmod == OPLM_DWORD)
4069 need_tmp64 = 1;
4070 }
4071 }
4072 if (pfomask) {
4073 tmp_op->pfomask |= pfomask;
4074 cond_vars |= pfomask;
4075 }
4076 // note: may overwrite, currently not a problem
4077 po->datap = tmp_op;
4078 }
4079
4080 if (po->op == OP_RCL || po->op == OP_RCR
4081 || po->op == OP_ADC || po->op == OP_SBB)
4082 cond_vars |= 1 << PFO_C;
4083 }
4084
4085 if (po->op == OP_CMPS || po->op == OP_SCAS) {
4086 cond_vars |= 1 << PFO_Z;
4087 }
4088 else if (po->op == OP_MUL
4089 || (po->op == OP_IMUL && po->operand_cnt == 1))
4090 {
4091 if (po->operand[0].lmod == OPLM_DWORD)
4092 need_tmp64 = 1;
4093 }
4094 else if (po->op == OP_CALL) {
4095 // note: resolved non-reg calls are OPF_DONE already
4096 pp = po->pp;
4097 if (pp == NULL)
4098 ferr(po, "NULL pp\n");
4099
4100 if (pp->is_unresolved) {
4101 int regmask_stack = 0;
4102 collect_call_args(po, i, pp, &regmask, save_arg_vars,
4103 i + opcnt * 2);
4104
4105 // this is pretty rough guess:
4106 // see ecx and edx were pushed (and not their saved versions)
4107 for (arg = 0; arg < pp->argc; arg++) {
4108 if (pp->arg[arg].reg != NULL)
4109 continue;
4110
4111 tmp_op = pp->arg[arg].datap;
4112 if (tmp_op == NULL)
4113 ferr(po, "parsed_op missing for arg%d\n", arg);
4114 if (tmp_op->p_argnum == 0 && tmp_op->operand[0].type == OPT_REG)
4115 regmask_stack |= 1 << tmp_op->operand[0].reg;
4116 }
4117
4118 if (!((regmask_stack & (1 << xCX))
4119 && (regmask_stack & (1 << xDX))))
4120 {
4121 if (pp->argc_stack != 0
4122 || ((regmask | regmask_arg) & ((1 << xCX)|(1 << xDX))))
4123 {
4124 pp_insert_reg_arg(pp, "ecx");
4125 pp->is_fastcall = 1;
4126 regmask_init |= 1 << xCX;
4127 regmask |= 1 << xCX;
4128 }
4129 if (pp->argc_stack != 0
4130 || ((regmask | regmask_arg) & (1 << xDX)))
4131 {
4132 pp_insert_reg_arg(pp, "edx");
4133 regmask_init |= 1 << xDX;
4134 regmask |= 1 << xDX;
4135 }
4136 }
4137
4138 // note: __cdecl doesn't fall into is_unresolved category
4139 if (pp->argc_stack > 0)
4140 pp->is_stdcall = 1;
4141 }
4142
4143 for (arg = 0; arg < pp->argc; arg++) {
4144 if (pp->arg[arg].reg != NULL) {
4145 reg = char_array_i(regs_r32,
4146 ARRAY_SIZE(regs_r32), pp->arg[arg].reg);
4147 if (reg < 0)
4148 ferr(ops, "arg '%s' is not a reg?\n", pp->arg[arg].reg);
4149 if (!(regmask & (1 << reg))) {
4150 regmask_init |= 1 << reg;
4151 regmask |= 1 << reg;
4152 }
4153 }
4154 }
4155 }
4156 else if (po->op == OP_MOV && po->operand[0].pp != NULL
4157 && po->operand[1].pp != NULL)
4158 {
4159 // <var> = offset <something>
4160 if ((po->operand[1].pp->is_func || po->operand[1].pp->is_fptr)
4161 && !IS_START(po->operand[1].name, "off_"))
4162 {
4163 if (!po->operand[0].pp->is_fptr)
4164 ferr(po, "%s not declared as fptr when it should be\n",
4165 po->operand[0].name);
4166 if (pp_cmp_func(po->operand[0].pp, po->operand[1].pp)) {
4167 pp_print(buf1, sizeof(buf1), po->operand[0].pp);
4168 pp_print(buf2, sizeof(buf2), po->operand[1].pp);
4169 fnote(po, "var: %s\n", buf1);
4170 fnote(po, "func: %s\n", buf2);
4171 ferr(po, "^ mismatch\n");
4172 }
4173 }
4174 }
4175 else if (po->op == OP_RET && !IS(g_func_pp->ret_type.name, "void"))
4176 regmask |= 1 << xAX;
4177 else if (po->op == OP_DIV || po->op == OP_IDIV) {
4178 // 32bit division is common, look for it
4179 if (po->op == OP_DIV)
4180 ret = scan_for_reg_clear(i, xDX);
4181 else
4182 ret = scan_for_cdq_edx(i);
4183 if (ret >= 0)
4184 po->flags |= OPF_32BIT;
4185 else
4186 need_tmp64 = 1;
4187 }
4188 else if (po->op == OP_CLD)
4189 po->flags |= OPF_RMD | OPF_DONE;
4190
4191 if (po->op == OP_RCL || po->op == OP_RCR || po->op == OP_XCHG) {
4192 need_tmp_var = 1;
4193 }
4194 }
4195
4196 // pass6:
4197 // - confirm regmask_save, it might have been reduced
4198 if (regmask_save != 0)
4199 {
4200 regmask_save = 0;
4201 for (i = 0; i < opcnt; i++) {
4202 po = &ops[i];
4203 if (po->flags & OPF_RMD)
4204 continue;
4205
4206 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE))
4207 regmask_save |= 1 << po->operand[0].reg;
4208 }
4209 }
4210
4211 // output starts here
4212
4213 // define userstack size
4214 if (g_func_pp->is_userstack) {
4215 fprintf(fout, "#ifndef US_SZ_%s\n", g_func_pp->name);
4216 fprintf(fout, "#define US_SZ_%s USERSTACK_SIZE\n", g_func_pp->name);
4217 fprintf(fout, "#endif\n");
4218 }
4219
4220 // the function itself
4221 fprintf(fout, "%s ", g_func_pp->ret_type.name);
4222 output_pp_attrs(fout, g_func_pp, g_ida_func_attr & IDAFA_NORETURN);
4223 fprintf(fout, "%s(", g_func_pp->name);
4224
4225 for (i = 0; i < g_func_pp->argc; i++) {
4226 if (i > 0)
4227 fprintf(fout, ", ");
4228 if (g_func_pp->arg[i].fptr != NULL) {
4229 // func pointer..
4230 pp = g_func_pp->arg[i].fptr;
4231 fprintf(fout, "%s (", pp->ret_type.name);
4232 output_pp_attrs(fout, pp, 0);
4233 fprintf(fout, "*a%d)(", i + 1);
4234 for (j = 0; j < pp->argc; j++) {
4235 if (j > 0)
4236 fprintf(fout, ", ");
4237 if (pp->arg[j].fptr)
4238 ferr(ops, "nested fptr\n");
4239 fprintf(fout, "%s", pp->arg[j].type.name);
4240 }
4241 if (pp->is_vararg) {
4242 if (j > 0)
4243 fprintf(fout, ", ");
4244 fprintf(fout, "...");
4245 }
4246 fprintf(fout, ")");
4247 }
4248 else if (g_func_pp->arg[i].type.is_retreg) {
4249 fprintf(fout, "u32 *r_%s", g_func_pp->arg[i].reg);
4250 }
4251 else {
4252 fprintf(fout, "%s a%d", g_func_pp->arg[i].type.name, i + 1);
4253 }
4254 }
4255 if (g_func_pp->is_vararg) {
4256 if (i > 0)
4257 fprintf(fout, ", ");
4258 fprintf(fout, "...");
4259 }
4260
4261 fprintf(fout, ")\n{\n");
4262
4263 // declare indirect functions
4264 for (i = 0; i < opcnt; i++) {
4265 po = &ops[i];
4266 if (po->flags & OPF_RMD)
4267 continue;
4268
4269 if (po->op == OP_CALL) {
4270 pp = po->pp;
4271 if (pp == NULL)
4272 ferr(po, "NULL pp\n");
4273
4274 if (pp->is_fptr && !(pp->name[0] != 0 && pp->is_arg)) {
4275 if (pp->name[0] != 0) {
4276 memmove(pp->name + 2, pp->name, strlen(pp->name) + 1);
4277 memcpy(pp->name, "i_", 2);
4278
4279 // might be declared already
4280 found = 0;
4281 for (j = 0; j < i; j++) {
4282 if (ops[j].op == OP_CALL && (pp_tmp = ops[j].pp)) {
4283 if (pp_tmp->is_fptr && IS(pp->name, pp_tmp->name)) {
4284 found = 1;
4285 break;
4286 }
4287 }
4288 }
4289 if (found)
4290 continue;
4291 }
4292 else
4293 snprintf(pp->name, sizeof(pp->name), "icall%d", i);
4294
4295 fprintf(fout, " %s (", pp->ret_type.name);
4296 output_pp_attrs(fout, pp, 0);
4297 fprintf(fout, "*%s)(", pp->name);
4298 for (j = 0; j < pp->argc; j++) {
4299 if (j > 0)
4300 fprintf(fout, ", ");
4301 fprintf(fout, "%s a%d", pp->arg[j].type.name, j + 1);
4302 }
4303 fprintf(fout, ");\n");
4304 }
4305 }
4306 }
4307
4308 // output LUTs/jumptables
4309 for (i = 0; i < g_func_pd_cnt; i++) {
4310 pd = &g_func_pd[i];
4311 fprintf(fout, " static const ");
4312 if (pd->type == OPT_OFFSET) {
4313 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
4314
4315 for (j = 0; j < pd->count; j++) {
4316 if (j > 0)
4317 fprintf(fout, ", ");
4318 fprintf(fout, "&&%s", pd->d[j].u.label);
4319 }
4320 }
4321 else {
4322 fprintf(fout, "%s %s[] =\n { ",
4323 lmod_type_u(ops, pd->lmod), pd->label);
4324
4325 for (j = 0; j < pd->count; j++) {
4326 if (j > 0)
4327 fprintf(fout, ", ");
4328 fprintf(fout, "%u", pd->d[j].u.val);
4329 }
4330 }
4331 fprintf(fout, " };\n");
4332 had_decl = 1;
4333 }
4334
4335 // declare stack frame, va_arg
4336 if (g_stack_fsz) {
4337 fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
4338 (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz);
4339 had_decl = 1;
4340 }
4341
4342 if (g_func_pp->is_userstack) {
4343 fprintf(fout, " u32 fake_sf[US_SZ_%s / 4];\n", g_func_pp->name);
4344 fprintf(fout, " u32 *esp = &fake_sf[sizeof(fake_sf) / 4];\n");
4345 had_decl = 1;
4346 }
4347
4348 if (g_func_pp->is_vararg) {
4349 fprintf(fout, " va_list ap;\n");
4350 had_decl = 1;
4351 }
4352
4353 // declare arg-registers
4354 for (i = 0; i < g_func_pp->argc; i++) {
4355 if (g_func_pp->arg[i].reg != NULL) {
4356 reg = char_array_i(regs_r32,
4357 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
4358 if (regmask & (1 << reg)) {
4359 if (g_func_pp->arg[i].type.is_retreg)
4360 fprintf(fout, " u32 %s = *r_%s;\n",
4361 g_func_pp->arg[i].reg, g_func_pp->arg[i].reg);
4362 else
4363 fprintf(fout, " u32 %s = (u32)a%d;\n",
4364 g_func_pp->arg[i].reg, i + 1);
4365 }
4366 else {
4367 if (g_func_pp->arg[i].type.is_retreg)
4368 ferr(ops, "retreg '%s' is unused?\n",
4369 g_func_pp->arg[i].reg);
4370 fprintf(fout, " // %s = a%d; // unused\n",
4371 g_func_pp->arg[i].reg, i + 1);
4372 }
4373 had_decl = 1;
4374 }
4375 }
4376
4377 regmask_now = regmask & ~regmask_arg;
4378 regmask_now &= ~(1 << xSP);
4379 if (regmask_now & 0x00ff) {
4380 for (reg = 0; reg < 8; reg++) {
4381 if (regmask_now & (1 << reg)) {
4382 fprintf(fout, " u32 %s", regs_r32[reg]);
4383 if (regmask_init & (1 << reg))
4384 fprintf(fout, " = 0");
4385 fprintf(fout, ";\n");
4386 had_decl = 1;
4387 }
4388 }
4389 }
4390 if (regmask_now & 0xff00) {
4391 for (reg = 8; reg < 16; reg++) {
4392 if (regmask_now & (1 << reg)) {
4393 fprintf(fout, " mmxr %s", regs_r32[reg]);
4394 if (regmask_init & (1 << reg))
4395 fprintf(fout, " = { 0, }");
4396 fprintf(fout, ";\n");
4397 had_decl = 1;
4398 }
4399 }
4400 }
4401
4402 if (regmask_save) {
4403 for (reg = 0; reg < 8; reg++) {
4404 if (regmask_save & (1 << reg)) {
4405 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
4406 had_decl = 1;
4407 }
4408 }
4409 }
4410
4411 for (i = 0; i < ARRAY_SIZE(save_arg_vars); i++) {
4412 if (save_arg_vars[i] == 0)
4413 continue;
4414 for (reg = 0; reg < 32; reg++) {
4415 if (save_arg_vars[i] & (1 << reg)) {
4416 fprintf(fout, " u32 %s;\n",
4417 saved_arg_name(buf1, sizeof(buf1), i, reg + 1));
4418 had_decl = 1;
4419 }
4420 }
4421 }
4422
4423 if (cond_vars) {
4424 for (i = 0; i < 8; i++) {
4425 if (cond_vars & (1 << i)) {
4426 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
4427 had_decl = 1;
4428 }
4429 }
4430 }
4431
4432 if (need_tmp_var) {
4433 fprintf(fout, " u32 tmp;\n");
4434 had_decl = 1;
4435 }
4436
4437 if (need_tmp64) {
4438 fprintf(fout, " u64 tmp64;\n");
4439 had_decl = 1;
4440 }
4441
4442 if (had_decl)
4443 fprintf(fout, "\n");
4444
4445 if (g_func_pp->is_vararg) {
4446 if (g_func_pp->argc_stack == 0)
4447 ferr(ops, "vararg func without stack args?\n");
4448 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
4449 }
4450
4451 // output ops
4452 for (i = 0; i < opcnt; i++)
4453 {
4454 if (g_labels[i] != NULL) {
4455 fprintf(fout, "\n%s:\n", g_labels[i]);
4456 label_pending = 1;
4457
4458 delayed_flag_op = NULL;
4459 last_arith_dst = NULL;
4460 }
4461
4462 po = &ops[i];
4463 if (po->flags & OPF_RMD)
4464 continue;
4465
4466 no_output = 0;
4467
4468 #define assert_operand_cnt(n_) \
4469 if (po->operand_cnt != n_) \
4470 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
4471
4472 // conditional/flag using op?
4473 if (po->flags & OPF_CC)
4474 {
4475 int is_delayed = 0;
4476
4477 tmp_op = po->datap;
4478
4479 // we go through all this trouble to avoid using parsed_flag_op,
4480 // which makes generated code much nicer
4481 if (delayed_flag_op != NULL)
4482 {
4483 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op,
4484 po->pfo, po->pfo_inv);
4485 is_delayed = 1;
4486 }
4487 else if (last_arith_dst != NULL
4488 && (po->pfo == PFO_Z || po->pfo == PFO_S || po->pfo == PFO_P
4489 || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR))
4490 ))
4491 {
4492 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
4493 out_test_for_cc(buf1, sizeof(buf1), po, po->pfo, po->pfo_inv,
4494 last_arith_dst->lmod, buf3);
4495 is_delayed = 1;
4496 }
4497 else if (tmp_op != NULL) {
4498 // use preprocessed flag calc results
4499 if (!(tmp_op->pfomask & (1 << po->pfo)))
4500 ferr(po, "not prepared for pfo %d\n", po->pfo);
4501
4502 // note: pfo_inv was not yet applied
4503 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
4504 po->pfo_inv ? "!" : "", parsed_flag_op_names[po->pfo]);
4505 }
4506 else {
4507 ferr(po, "all methods of finding comparison failed\n");
4508 }
4509
4510 if (po->flags & OPF_JMP) {
4511 fprintf(fout, " if %s", buf1);
4512 }
4513 else if (po->op == OP_RCL || po->op == OP_RCR
4514 || po->op == OP_ADC || po->op == OP_SBB)
4515 {
4516 if (is_delayed)
4517 fprintf(fout, " cond_%s = %s;\n",
4518 parsed_flag_op_names[po->pfo], buf1);
4519 }
4520 else if (po->flags & OPF_DATA) { // SETcc
4521 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
4522 fprintf(fout, " %s = %s;", buf2, buf1);
4523 }
4524 else {
4525 ferr(po, "unhandled conditional op\n");
4526 }
4527 }
4528
4529 pfomask = po->pfomask;
4530
4531 if (po->flags & (OPF_REPZ|OPF_REPNZ)) {
4532 struct parsed_opr opr = {0,};
4533 opr.type = OPT_REG;
4534 opr.reg = xCX;
4535 opr.lmod = OPLM_DWORD;
4536 ret = try_resolve_const(i, &opr, opcnt * 7 + i, &uval);
4537
4538 if (ret != 1 || uval == 0) {
4539 // we need initial flags for ecx=0 case..
4540 if (i > 0 && ops[i - 1].op == OP_XOR
4541 && IS(ops[i - 1].operand[0].name,
4542 ops[i - 1].operand[1].name))
4543 {
4544 fprintf(fout, " cond_z = ");
4545 if (pfomask & (1 << PFO_C))
4546 fprintf(fout, "cond_c = ");
4547 fprintf(fout, "0;\n");
4548 }
4549 else if (last_arith_dst != NULL) {
4550 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
4551 out_test_for_cc(buf1, sizeof(buf1), po, PFO_Z, 0,
4552 last_arith_dst->lmod, buf3);
4553 fprintf(fout, " cond_z = %s;\n", buf1);
4554 }
4555 else
4556 ferr(po, "missing initial ZF\n");
4557 }
4558 }
4559
4560 switch (po->op)
4561 {
4562 case OP_MOV:
4563 assert_operand_cnt(2);
4564 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4565 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4566 default_cast_to(buf3, sizeof(buf3), &po->operand[0]);
4567 fprintf(fout, " %s = %s;", buf1,
4568 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4569 buf3, 0));
4570 break;
4571
4572 case OP_LEA:
4573 assert_operand_cnt(2);
4574 po->operand[1].lmod = OPLM_DWORD; // always
4575 fprintf(fout, " %s = %s;",
4576 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4577 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4578 NULL, 1));
4579 break;
4580
4581 case OP_MOVZX:
4582 assert_operand_cnt(2);
4583 fprintf(fout, " %s = %s;",
4584 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4585 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4586 break;
4587
4588 case OP_MOVSX:
4589 assert_operand_cnt(2);
4590 switch (po->operand[1].lmod) {
4591 case OPLM_BYTE:
4592 strcpy(buf3, "(s8)");
4593 break;
4594 case OPLM_WORD:
4595 strcpy(buf3, "(s16)");
4596 break;
4597 default:
4598 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
4599 }
4600 fprintf(fout, " %s = %s;",
4601 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4602 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4603 buf3, 0));
4604 break;
4605
4606 case OP_XCHG:
4607 assert_operand_cnt(2);
4608 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4609 fprintf(fout, " tmp = %s;",
4610 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], "", 0));
4611 fprintf(fout, " %s = %s;",
4612 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4613 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4614 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
4615 fprintf(fout, " %s = %stmp;",
4616 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]),
4617 default_cast_to(buf3, sizeof(buf3), &po->operand[1]));
4618 snprintf(g_comment, sizeof(g_comment), "xchg");
4619 break;
4620
4621 case OP_NOT:
4622 assert_operand_cnt(1);
4623 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4624 fprintf(fout, " %s = ~%s;", buf1, buf1);
4625 break;
4626
4627 case OP_CDQ:
4628 assert_operand_cnt(2);
4629 fprintf(fout, " %s = (s32)%s >> 31;",
4630 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4631 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4632 strcpy(g_comment, "cdq");
4633 break;
4634
4635 case OP_LODS:
4636 assert_operand_cnt(3);
4637 if (po->flags & OPF_REP) {
4638 // hmh..
4639 ferr(po, "TODO\n");
4640 }
4641 else {
4642 fprintf(fout, " eax = %sesi; esi %c= %d;",
4643 lmod_cast_u_ptr(po, po->operand[0].lmod),
4644 (po->flags & OPF_DF) ? '-' : '+',
4645 lmod_bytes(po, po->operand[0].lmod));
4646 strcpy(g_comment, "lods");
4647 }
4648 break;
4649
4650 case OP_STOS:
4651 assert_operand_cnt(3);
4652 if (po->flags & OPF_REP) {
4653 fprintf(fout, " for (; ecx != 0; ecx--, edi %c= %d)\n",
4654 (po->flags & OPF_DF) ? '-' : '+',
4655 lmod_bytes(po, po->operand[0].lmod));
4656 fprintf(fout, " %sedi = eax;",
4657 lmod_cast_u_ptr(po, po->operand[0].lmod));
4658 strcpy(g_comment, "rep stos");
4659 }
4660 else {
4661 fprintf(fout, " %sedi = eax; edi %c= %d;",
4662 lmod_cast_u_ptr(po, po->operand[0].lmod),
4663 (po->flags & OPF_DF) ? '-' : '+',
4664 lmod_bytes(po, po->operand[0].lmod));
4665 strcpy(g_comment, "stos");
4666 }
4667 break;
4668
4669 case OP_MOVS:
4670 assert_operand_cnt(3);
4671 j = lmod_bytes(po, po->operand[0].lmod);
4672 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
4673 l = (po->flags & OPF_DF) ? '-' : '+';
4674 if (po->flags & OPF_REP) {
4675 fprintf(fout,
4676 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n",
4677 l, j, l, j);
4678 fprintf(fout,
4679 " %sedi = %sesi;", buf1, buf1);
4680 strcpy(g_comment, "rep movs");
4681 }
4682 else {
4683 fprintf(fout, " %sedi = %sesi; edi %c= %d; esi %c= %d;",
4684 buf1, buf1, l, j, l, j);
4685 strcpy(g_comment, "movs");
4686 }
4687 break;
4688
4689 case OP_CMPS:
4690 // repe ~ repeat while ZF=1
4691 assert_operand_cnt(3);
4692 j = lmod_bytes(po, po->operand[0].lmod);
4693 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
4694 l = (po->flags & OPF_DF) ? '-' : '+';
4695 if (po->flags & OPF_REP) {
4696 fprintf(fout,
4697 " for (; ecx != 0; ecx--) {\n");
4698 if (pfomask & (1 << PFO_C)) {
4699 // ugh..
4700 fprintf(fout,
4701 " cond_c = %sesi < %sedi;\n", buf1, buf1);
4702 pfomask &= ~(1 << PFO_C);
4703 }
4704 fprintf(fout,
4705 " cond_z = (%sesi == %sedi); esi %c= %d, edi %c= %d;\n",
4706 buf1, buf1, l, j, l, j);
4707 fprintf(fout,
4708 " if (cond_z %s 0) break;\n",
4709 (po->flags & OPF_REPZ) ? "==" : "!=");
4710 fprintf(fout,
4711 " }");
4712 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
4713 (po->flags & OPF_REPZ) ? "e" : "ne");
4714 }
4715 else {
4716 fprintf(fout,
4717 " cond_z = (%sesi == %sedi); esi %c= %d; edi %c= %d;",
4718 buf1, buf1, l, j, l, j);
4719 strcpy(g_comment, "cmps");
4720 }
4721 pfomask &= ~(1 << PFO_Z);
4722 last_arith_dst = NULL;
4723 delayed_flag_op = NULL;
4724 break;
4725
4726 case OP_SCAS:
4727 // only does ZF (for now)
4728 // repe ~ repeat while ZF=1
4729 assert_operand_cnt(3);
4730 j = lmod_bytes(po, po->operand[0].lmod);
4731 l = (po->flags & OPF_DF) ? '-' : '+';
4732 if (po->flags & OPF_REP) {
4733 fprintf(fout,
4734 " for (; ecx != 0; ecx--) {\n");
4735 fprintf(fout,
4736 " cond_z = (%seax == %sedi); edi %c= %d;\n",
4737 lmod_cast_u(po, po->operand[0].lmod),
4738 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
4739 fprintf(fout,
4740 " if (cond_z %s 0) break;\n",
4741 (po->flags & OPF_REPZ) ? "==" : "!=");
4742 fprintf(fout,
4743 " }");
4744 snprintf(g_comment, sizeof(g_comment), "rep%s scas",
4745 (po->flags & OPF_REPZ) ? "e" : "ne");
4746 }
4747 else {
4748 fprintf(fout, " cond_z = (%seax == %sedi); edi %c= %d;",
4749 lmod_cast_u(po, po->operand[0].lmod),
4750 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
4751 strcpy(g_comment, "scas");
4752 }
4753 pfomask &= ~(1 << PFO_Z);
4754 last_arith_dst = NULL;
4755 delayed_flag_op = NULL;
4756 break;
4757
4758 // arithmetic w/flags
4759 case OP_AND:
4760 case OP_OR:
4761 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4762 // fallthrough
4763 dualop_arith:
4764 assert_operand_cnt(2);
4765 fprintf(fout, " %s %s= %s;",
4766 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4767 op_to_c(po),
4768 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4769 output_std_flags(fout, po, &pfomask, buf1);
4770 last_arith_dst = &po->operand[0];
4771 delayed_flag_op = NULL;
4772 break;
4773
4774 case OP_SHL:
4775 case OP_SHR:
4776 assert_operand_cnt(2);
4777 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4778 if (pfomask & (1 << PFO_C)) {
4779 if (po->operand[1].type == OPT_CONST) {
4780 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4781 j = po->operand[1].val;
4782 j %= l;
4783 if (j != 0) {
4784 if (po->op == OP_SHL)
4785 j = l - j;
4786 else
4787 j -= 1;
4788 fprintf(fout, " cond_c = (%s >> %d) & 1;\n",
4789 buf1, j);
4790 }
4791 else
4792 ferr(po, "zero shift?\n");
4793 }
4794 else
4795 ferr(po, "TODO\n");
4796 pfomask &= ~(1 << PFO_C);
4797 }
4798 fprintf(fout, " %s %s= %s;", buf1, op_to_c(po),
4799 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4800 output_std_flags(fout, po, &pfomask, buf1);
4801 last_arith_dst = &po->operand[0];
4802 delayed_flag_op = NULL;
4803 break;
4804
4805 case OP_SAR:
4806 assert_operand_cnt(2);
4807 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4808 fprintf(fout, " %s = %s%s >> %s;", buf1,
4809 lmod_cast_s(po, po->operand[0].lmod), buf1,
4810 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4811 output_std_flags(fout, po, &pfomask, buf1);
4812 last_arith_dst = &po->operand[0];
4813 delayed_flag_op = NULL;
4814 break;
4815
4816 case OP_SHRD:
4817 assert_operand_cnt(3);
4818 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4819 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4820 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4821 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4822 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[2]);
4823 fprintf(fout, " %s >>= %s; %s |= %s << (%d - %s);",
4824 buf1, buf3, buf1, buf2, l, buf3);
4825 strcpy(g_comment, "shrd");
4826 output_std_flags(fout, po, &pfomask, buf1);
4827 last_arith_dst = &po->operand[0];
4828 delayed_flag_op = NULL;
4829 break;
4830
4831 case OP_ROL:
4832 case OP_ROR:
4833 assert_operand_cnt(2);
4834 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4835 if (po->operand[1].type == OPT_CONST) {
4836 j = po->operand[1].val;
4837 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
4838 fprintf(fout, po->op == OP_ROL ?
4839 " %s = (%s << %d) | (%s >> %d);" :
4840 " %s = (%s >> %d) | (%s << %d);",
4841 buf1, buf1, j, buf1,
4842 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
4843 }
4844 else
4845 ferr(po, "TODO\n");
4846 output_std_flags(fout, po, &pfomask, buf1);
4847 last_arith_dst = &po->operand[0];
4848 delayed_flag_op = NULL;
4849 break;
4850
4851 case OP_RCL:
4852 case OP_RCR:
4853 assert_operand_cnt(2);
4854 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4855 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4856 if (po->operand[1].type == OPT_CONST) {
4857 j = po->operand[1].val % l;
4858 if (j == 0)
4859 ferr(po, "zero rotate\n");
4860 fprintf(fout, " tmp = (%s >> %d) & 1;\n",
4861 buf1, (po->op == OP_RCL) ? (l - j) : (j - 1));
4862 if (po->op == OP_RCL) {
4863 fprintf(fout,
4864 " %s = (%s << %d) | (cond_c << %d)",
4865 buf1, buf1, j, j - 1);
4866 if (j != 1)
4867 fprintf(fout, " | (%s >> %d)", buf1, l + 1 - j);
4868 }
4869 else {
4870 fprintf(fout,
4871 " %s = (%s >> %d) | (cond_c << %d)",
4872 buf1, buf1, j, l - j);
4873 if (j != 1)
4874 fprintf(fout, " | (%s << %d)", buf1, l + 1 - j);
4875 }
4876 fprintf(fout, ";\n");
4877 fprintf(fout, " cond_c = tmp;");
4878 }
4879 else
4880 ferr(po, "TODO\n");
4881 strcpy(g_comment, (po->op == OP_RCL) ? "rcl" : "rcr");
4882 output_std_flags(fout, po, &pfomask, buf1);
4883 last_arith_dst = &po->operand[0];
4884 delayed_flag_op = NULL;
4885 break;
4886
4887 case OP_XOR:
4888 assert_operand_cnt(2);
4889 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4890 if (IS(opr_name(po, 0), opr_name(po, 1))) {
4891 // special case for XOR
4892 if (pfomask & (1 << PFO_BE)) { // weird, but it happens..
4893 fprintf(fout, " cond_be = 1;\n");
4894 pfomask &= ~(1 << PFO_BE);
4895 }
4896 fprintf(fout, " %s = 0;",
4897 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4898 last_arith_dst = &po->operand[0];
4899 delayed_flag_op = NULL;
4900 break;
4901 }
4902 goto dualop_arith;
4903
4904 case OP_ADD:
4905 assert_operand_cnt(2);
4906 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4907 if (pfomask & (1 << PFO_C)) {
4908 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4909 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4910 if (po->operand[0].lmod == OPLM_DWORD) {
4911 fprintf(fout, " tmp64 = (u64)%s + %s;\n", buf1, buf2);
4912 fprintf(fout, " cond_c = tmp64 >> 32;\n");
4913 fprintf(fout, " %s = (u32)tmp64;",
4914 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4915 strcat(g_comment, "add64");
4916 }
4917 else {
4918 fprintf(fout, " cond_c = ((u32)%s + %s) >> %d;\n",
4919 buf1, buf2, lmod_bytes(po, po->operand[0].lmod) * 8);
4920 fprintf(fout, " %s += %s;",
4921 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4922 buf2);
4923 }
4924 pfomask &= ~(1 << PFO_C);
4925 output_std_flags(fout, po, &pfomask, buf1);
4926 last_arith_dst = &po->operand[0];
4927 delayed_flag_op = NULL;
4928 break;
4929 }
4930 goto dualop_arith;
4931
4932 case OP_SUB:
4933 assert_operand_cnt(2);
4934 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4935 if (pfomask & ~((1 << PFO_Z) | (1 << PFO_S))) {
4936 for (j = 0; j <= PFO_LE; j++) {
4937 if (!(pfomask & (1 << j)))
4938 continue;
4939 if (j == PFO_Z || j == PFO_S)
4940 continue;
4941
4942 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
4943 fprintf(fout, " cond_%s = %s;\n",
4944 parsed_flag_op_names[j], buf1);
4945 pfomask &= ~(1 << j);
4946 }
4947 }
4948 goto dualop_arith;
4949
4950 case OP_ADC:
4951 case OP_SBB:
4952 assert_operand_cnt(2);
4953 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4954 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4955 if (po->op == OP_SBB
4956 && IS(po->operand[0].name, po->operand[1].name))
4957 {
4958 // avoid use of unitialized var
4959 fprintf(fout, " %s = -cond_c;", buf1);
4960 // carry remains what it was
4961 pfomask &= ~(1 << PFO_C);
4962 }
4963 else {
4964 fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po),
4965 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4966 }
4967 output_std_flags(fout, po, &pfomask, buf1);
4968 last_arith_dst = &po->operand[0];
4969 delayed_flag_op = NULL;
4970 break;
4971
4972 case OP_BSF:
4973 assert_operand_cnt(2);
4974 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4975 fprintf(fout, " %s = %s ? __builtin_ffs(%s) - 1 : 0;",
4976 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4977 buf2, buf2);
4978 output_std_flags(fout, po, &pfomask, buf1);
4979 last_arith_dst = &po->operand[0];
4980 delayed_flag_op = NULL;
4981 strcat(g_comment, "bsf");
4982 break;
4983
4984 case OP_DEC:
4985 if (pfomask & ~(PFOB_S | PFOB_S | PFOB_C)) {
4986 for (j = 0; j <= PFO_LE; j++) {
4987 if (!(pfomask & (1 << j)))
4988 continue;
4989 if (j == PFO_Z || j == PFO_S || j == PFO_C)
4990 continue;
4991
4992 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
4993 fprintf(fout, " cond_%s = %s;\n",
4994 parsed_flag_op_names[j], buf1);
4995 pfomask &= ~(1 << j);
4996 }
4997 }
4998 // fallthrough
4999
5000 case OP_INC:
5001 if (pfomask & (1 << PFO_C))
5002 // carry is unaffected by inc/dec.. wtf?
5003 ferr(po, "carry propagation needed\n");
5004
5005 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5006 if (po->operand[0].type == OPT_REG) {
5007 strcpy(buf2, po->op == OP_INC ? "++" : "--");
5008 fprintf(fout, " %s%s;", buf1, buf2);
5009 }
5010 else {
5011 strcpy(buf2, po->op == OP_INC ? "+" : "-");
5012 fprintf(fout, " %s %s= 1;", buf1, buf2);
5013 }
5014 output_std_flags(fout, po, &pfomask, buf1);
5015 last_arith_dst = &po->operand[0];
5016 delayed_flag_op = NULL;
5017 break;
5018
5019 case OP_NEG:
5020 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5021 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
5022 fprintf(fout, " %s = -%s%s;", buf1,
5023 lmod_cast_s(po, po->operand[0].lmod), buf2);
5024 last_arith_dst = &po->operand[0];
5025 delayed_flag_op = NULL;
5026 if (pfomask & (1 << PFO_C)) {
5027 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
5028 pfomask &= ~(1 << PFO_C);
5029 }
5030 break;
5031
5032 case OP_IMUL:
5033 if (po->operand_cnt == 2) {
5034 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5035 goto dualop_arith;
5036 }
5037 if (po->operand_cnt == 3)
5038 ferr(po, "TODO imul3\n");
5039 // fallthrough
5040 case OP_MUL:
5041 assert_operand_cnt(1);
5042 switch (po->operand[0].lmod) {
5043 case OPLM_DWORD:
5044 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
5045 fprintf(fout, " tmp64 = %seax * %s%s;\n", buf1, buf1,
5046 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
5047 fprintf(fout, " edx = tmp64 >> 32;\n");
5048 fprintf(fout, " eax = tmp64;");
5049 break;
5050 case OPLM_BYTE:
5051 strcpy(buf1, po->op == OP_IMUL ? "(s16)(s8)" : "(u16)(u8)");
5052 fprintf(fout, " LOWORD(eax) = %seax * %s;", buf1,
5053 out_src_opr(buf2, sizeof(buf2), po, &po->operand[0],
5054 buf1, 0));
5055 break;
5056 default:
5057 ferr(po, "TODO: unhandled mul type\n");
5058 break;
5059 }
5060 last_arith_dst = NULL;
5061 delayed_flag_op = NULL;
5062 break;
5063
5064 case OP_DIV:
5065 case OP_IDIV:
5066 assert_operand_cnt(1);
5067 if (po->operand[0].lmod != OPLM_DWORD)
5068 ferr(po, "unhandled lmod %d\n", po->operand[0].lmod);
5069
5070 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
5071 strcpy(buf2, lmod_cast(po, po->operand[0].lmod,
5072 po->op == OP_IDIV));
5073 switch (po->operand[0].lmod) {
5074 case OPLM_DWORD:
5075 if (po->flags & OPF_32BIT)
5076 snprintf(buf3, sizeof(buf3), "%seax", buf2);
5077 else {
5078 fprintf(fout, " tmp64 = ((u64)edx << 32) | eax;\n");
5079 snprintf(buf3, sizeof(buf3), "%stmp64",
5080 (po->op == OP_IDIV) ? "(s64)" : "");
5081 }
5082 if (po->operand[0].type == OPT_REG
5083 && po->operand[0].reg == xDX)
5084 {
5085 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
5086 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
5087 }
5088 else {
5089 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
5090 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
5091 }
5092 break;
5093 default:
5094 ferr(po, "unhandled division type\n");
5095 }
5096 last_arith_dst = NULL;
5097 delayed_flag_op = NULL;
5098 break;
5099
5100 case OP_TEST:
5101 case OP_CMP:
5102 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5103 if (pfomask != 0) {
5104 for (j = 0; j < 8; j++) {
5105 if (pfomask & (1 << j)) {
5106 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
5107 fprintf(fout, " cond_%s = %s;",
5108 parsed_flag_op_names[j], buf1);
5109 }
5110 }
5111 pfomask = 0;
5112 }
5113 else
5114 no_output = 1;
5115 last_arith_dst = NULL;
5116 delayed_flag_op = po;
5117 break;
5118
5119 case OP_SCC:
5120 // SETcc - should already be handled
5121 break;
5122
5123 // note: we reuse OP_Jcc for SETcc, only flags differ
5124 case OP_JCC:
5125 fprintf(fout, "\n goto %s;", po->operand[0].name);
5126 break;
5127
5128 case OP_JECXZ:
5129 fprintf(fout, " if (ecx == 0)\n");
5130 fprintf(fout, " goto %s;", po->operand[0].name);
5131 strcat(g_comment, "jecxz");
5132 break;
5133
5134 case OP_JMP:
5135 assert_operand_cnt(1);
5136 last_arith_dst = NULL;
5137 delayed_flag_op = NULL;
5138
5139 if (po->operand[0].type == OPT_REGMEM) {
5140 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
5141 buf1, buf2);
5142 if (ret != 2)
5143 ferr(po, "parse failure for jmp '%s'\n",
5144 po->operand[0].name);
5145 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
5146 break;
5147 }
5148 else if (po->operand[0].type != OPT_LABEL)
5149 ferr(po, "unhandled jmp type\n");
5150
5151 fprintf(fout, " goto %s;", po->operand[0].name);
5152 break;
5153
5154 case OP_CALL:
5155 assert_operand_cnt(1);
5156 pp = po->pp;
5157 my_assert_not(pp, NULL);
5158
5159 strcpy(buf3, " ");
5160 if (po->flags & OPF_CC) {
5161 // we treat conditional branch to another func
5162 // (yes such code exists..) as conditional tailcall
5163 strcat(buf3, " ");
5164 fprintf(fout, " {\n");
5165 }
5166
5167 if (pp->is_fptr && !pp->is_arg) {
5168 fprintf(fout, "%s%s = %s;\n", buf3, pp->name,
5169 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
5170 "(void *)", 0));
5171 if (pp->is_unresolved)
5172 fprintf(fout, "%sunresolved_call(\"%s:%d\", %s);\n",
5173 buf3, asmfn, po->asmln, pp->name);
5174 }
5175
5176 fprintf(fout, "%s", buf3);
5177 if (strstr(pp->ret_type.name, "int64")) {
5178 if (po->flags & OPF_TAIL)
5179 ferr(po, "int64 and tail?\n");
5180 fprintf(fout, "tmp64 = ");
5181 }
5182 else if (!IS(pp->ret_type.name, "void")) {
5183 if (po->flags & OPF_TAIL) {
5184 if (!IS(g_func_pp->ret_type.name, "void")) {
5185 fprintf(fout, "return ");
5186 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
5187 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
5188 }
5189 }
5190 else if (regmask & (1 << xAX)) {
5191 fprintf(fout, "eax = ");
5192 if (pp->ret_type.is_ptr)
5193 fprintf(fout, "(u32)");
5194 }
5195 }
5196
5197 if (pp->name[0] == 0)
5198 ferr(po, "missing pp->name\n");
5199 fprintf(fout, "%s%s(", pp->name,
5200 pp->has_structarg ? "_sa" : "");
5201
5202 if (po->flags & OPF_ATAIL) {
5203 if (pp->argc_stack != g_func_pp->argc_stack
5204 || (pp->argc_stack > 0
5205 && pp->is_stdcall != g_func_pp->is_stdcall))
5206 ferr(po, "incompatible tailcall\n");
5207 if (g_func_pp->has_retreg)
5208 ferr(po, "TODO: retreg+tailcall\n");
5209
5210 for (arg = j = 0; arg < pp->argc; arg++) {
5211 if (arg > 0)
5212 fprintf(fout, ", ");
5213
5214 cast[0] = 0;
5215 if (pp->arg[arg].type.is_ptr)
5216 snprintf(cast, sizeof(cast), "(%s)",
5217 pp->arg[arg].type.name);
5218
5219 if (pp->arg[arg].reg != NULL) {
5220 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
5221 continue;
5222 }
5223 // stack arg
5224 for (; j < g_func_pp->argc; j++)
5225 if (g_func_pp->arg[j].reg == NULL)
5226 break;
5227 fprintf(fout, "%sa%d", cast, j + 1);
5228 j++;
5229 }
5230 }
5231 else {
5232 for (arg = 0; arg < pp->argc; arg++) {
5233 if (arg > 0)
5234 fprintf(fout, ", ");
5235
5236 cast[0] = 0;
5237 if (pp->arg[arg].type.is_ptr)
5238 snprintf(cast, sizeof(cast), "(%s)",
5239 pp->arg[arg].type.name);
5240
5241 if (pp->arg[arg].reg != NULL) {
5242 if (pp->arg[arg].type.is_retreg)
5243 fprintf(fout, "&%s", pp->arg[arg].reg);
5244 else
5245 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
5246 continue;
5247 }
5248
5249 // stack arg
5250 tmp_op = pp->arg[arg].datap;
5251 if (tmp_op == NULL)
5252 ferr(po, "parsed_op missing for arg%d\n", arg);
5253
5254 if (tmp_op->flags & OPF_VAPUSH) {
5255 fprintf(fout, "ap");
5256 }
5257 else if (tmp_op->p_argpass != 0) {
5258 fprintf(fout, "a%d", tmp_op->p_argpass);
5259 }
5260 else if (tmp_op->p_argnum != 0) {
5261 fprintf(fout, "%s%s", cast,
5262 saved_arg_name(buf1, sizeof(buf1),
5263 tmp_op->p_arggrp, tmp_op->p_argnum));
5264 }
5265 else {
5266 fprintf(fout, "%s",
5267 out_src_opr(buf1, sizeof(buf1),
5268 tmp_op, &tmp_op->operand[0], cast, 0));
5269 }
5270 }
5271 }
5272 fprintf(fout, ");");
5273
5274 if (strstr(pp->ret_type.name, "int64")) {
5275 fprintf(fout, "\n");
5276 fprintf(fout, "%sedx = tmp64 >> 32;\n", buf3);
5277 fprintf(fout, "%seax = tmp64;", buf3);
5278 }
5279
5280 if (pp->is_unresolved) {
5281 snprintf(buf2, sizeof(buf2), " unresolved %dreg",
5282 pp->argc_reg);
5283 strcat(g_comment, buf2);
5284 }
5285
5286 if (po->flags & OPF_TAIL) {
5287 ret = 0;
5288 if (i == opcnt - 1 || pp->is_noreturn)
5289 ret = 0;
5290 else if (IS(pp->ret_type.name, "void"))
5291 ret = 1;
5292 else if (IS(g_func_pp->ret_type.name, "void"))
5293 ret = 1;
5294 // else already handled as 'return f()'
5295
5296 if (ret) {
5297 if (!IS(g_func_pp->ret_type.name, "void")) {
5298 ferr(po, "int func -> void func tailcall?\n");
5299 }
5300 else {
5301 fprintf(fout, "\n%sreturn;", buf3);
5302 strcat(g_comment, " ^ tailcall");
5303 }
5304 }
5305 else
5306 strcat(g_comment, " tailcall");
5307 }
5308 if (pp->is_noreturn)
5309 strcat(g_comment, " noreturn");
5310 if ((po->flags & OPF_ATAIL) && pp->argc_stack > 0)
5311 strcat(g_comment, " argframe");
5312 if (po->flags & OPF_CC)
5313 strcat(g_comment, " cond");
5314
5315 if (po->flags & OPF_CC)
5316 fprintf(fout, "\n }");
5317
5318 delayed_flag_op = NULL;
5319 last_arith_dst = NULL;
5320 break;
5321
5322 case OP_RET:
5323 if (g_func_pp->is_vararg)
5324 fprintf(fout, " va_end(ap);\n");
5325 if (g_func_pp->has_retreg) {
5326 for (arg = 0; arg < g_func_pp->argc; arg++)
5327 if (g_func_pp->arg[arg].type.is_retreg)
5328 fprintf(fout, " *r_%s = %s;\n",
5329 g_func_pp->arg[arg].reg, g_func_pp->arg[arg].reg);
5330 }
5331
5332 if (IS(g_func_pp->ret_type.name, "void")) {
5333 if (i != opcnt - 1 || label_pending)
5334 fprintf(fout, " return;");
5335 }
5336 else if (g_func_pp->ret_type.is_ptr) {
5337 fprintf(fout, " return (%s)eax;",
5338 g_func_pp->ret_type.name);
5339 }
5340 else if (IS(g_func_pp->ret_type.name, "__int64"))
5341 fprintf(fout, " return ((u64)edx << 32) | eax;");
5342 else
5343 fprintf(fout, " return eax;");
5344
5345 last_arith_dst = NULL;
5346 delayed_flag_op = NULL;
5347 break;
5348
5349 case OP_PUSH:
5350 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
5351 if (po->p_argnum != 0) {
5352 // special case - saved func arg
5353 fprintf(fout, " %s = %s;",
5354 saved_arg_name(buf2, sizeof(buf2),
5355 po->p_arggrp, po->p_argnum), buf1);
5356 break;
5357 }
5358 else if (po->flags & OPF_RSAVE) {
5359 fprintf(fout, " s_%s = %s;", buf1, buf1);
5360 break;
5361 }
5362 else if (g_func_pp->is_userstack) {
5363 fprintf(fout, " *(--esp) = %s;", buf1);
5364 break;
5365 }
5366 if (!(g_ida_func_attr & IDAFA_NORETURN))
5367 ferr(po, "stray push encountered\n");
5368 no_output = 1;
5369 break;
5370
5371 case OP_POP:
5372 if (po->flags & OPF_RSAVE) {
5373 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5374 fprintf(fout, " %s = s_%s;", buf1, buf1);
5375 break;
5376 }
5377 else if (po->datap != NULL) {
5378 // push/pop pair
5379 tmp_op = po->datap;
5380 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5381 fprintf(fout, " %s = %s;", buf1,
5382 out_src_opr(buf2, sizeof(buf2),
5383 tmp_op, &tmp_op->operand[0],
5384 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
5385 break;
5386 }
5387 else if (g_func_pp->is_userstack) {
5388 fprintf(fout, " %s = *esp++;",
5389 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
5390 break;
5391 }
5392 else
5393 ferr(po, "stray pop encountered\n");
5394 break;
5395
5396 case OP_NOP:
5397 no_output = 1;
5398 break;
5399
5400 // mmx
5401 case OP_EMMS:
5402 strcpy(g_comment, "(emms)");
5403 break;
5404
5405 default:
5406 no_output = 1;
5407 ferr(po, "unhandled op type %d, flags %x\n",
5408 po->op, po->flags);
5409 break;
5410 }
5411
5412 if (g_comment[0] != 0) {
5413 char *p = g_comment;
5414 while (my_isblank(*p))
5415 p++;
5416 fprintf(fout, " // %s", p);
5417 g_comment[0] = 0;
5418 no_output = 0;
5419 }
5420 if (!no_output)
5421 fprintf(fout, "\n");
5422
5423 // some sanity checking
5424 if (po->flags & OPF_REP) {
5425 if (po->op != OP_STOS && po->op != OP_MOVS
5426 && po->op != OP_CMPS && po->op != OP_SCAS)
5427 ferr(po, "unexpected rep\n");
5428 if (!(po->flags & (OPF_REPZ|OPF_REPNZ))
5429 && (po->op == OP_CMPS || po->op == OP_SCAS))
5430 ferr(po, "cmps/scas with plain rep\n");
5431 }
5432 if ((po->flags & (OPF_REPZ|OPF_REPNZ))
5433 && po->op != OP_CMPS && po->op != OP_SCAS)
5434 ferr(po, "unexpected repz/repnz\n");
5435
5436 if (pfomask != 0)
5437 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
5438
5439 // see is delayed flag stuff is still valid
5440 if (delayed_flag_op != NULL && delayed_flag_op != po) {
5441 if (is_any_opr_modified(delayed_flag_op, po, 0))
5442 delayed_flag_op = NULL;
5443 }
5444
5445 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
5446 if (is_opr_modified(last_arith_dst, po))
5447 last_arith_dst = NULL;
5448 }
5449
5450 label_pending = 0;
5451 }
5452
5453 if (g_stack_fsz && !g_stack_frame_used)
5454 fprintf(fout, " (void)sf;\n");
5455
5456 fprintf(fout, "}\n\n");
5457
5458 gen_x_cleanup(opcnt);
5459}
5460
5461static void gen_x_cleanup(int opcnt)
5462{
5463 int i;
5464
5465 for (i = 0; i < opcnt; i++) {
5466 struct label_ref *lr, *lr_del;
5467
5468 lr = g_label_refs[i].next;
5469 while (lr != NULL) {
5470 lr_del = lr;
5471 lr = lr->next;
5472 free(lr_del);
5473 }
5474 g_label_refs[i].i = -1;
5475 g_label_refs[i].next = NULL;
5476
5477 if (ops[i].op == OP_CALL) {
5478 if (ops[i].pp)
5479 proto_release(ops[i].pp);
5480 }
5481 }
5482 g_func_pp = NULL;
5483}
5484
5485struct func_proto_dep;
5486
5487struct func_prototype {
5488 char name[NAMELEN];
5489 int id;
5490 int argc_stack;
5491 int regmask_dep;
5492 int has_ret:3; // -1, 0, 1: unresolved, no, yes
5493 unsigned int dep_resolved:1;
5494 unsigned int is_stdcall:1;
5495 struct func_proto_dep *dep_func;
5496 int dep_func_cnt;
5497 const struct parsed_proto *pp; // seed pp, if any
5498};
5499
5500struct func_proto_dep {
5501 char *name;
5502 struct func_prototype *proto;
5503 int regmask_live; // .. at the time of call
5504 unsigned int ret_dep:1; // return from this is caller's return
5505};
5506
5507static struct func_prototype *hg_fp;
5508static int hg_fp_cnt;
5509
5510static struct scanned_var {
5511 char name[NAMELEN];
5512 enum opr_lenmod lmod;
5513 unsigned int is_seeded:1;
5514 unsigned int is_c_str:1;
5515} *hg_vars;
5516static int hg_var_cnt;
5517
5518static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
5519 int count);
5520
5521static struct func_proto_dep *hg_fp_find_dep(struct func_prototype *fp,
5522 const char *name)
5523{
5524 int i;
5525
5526 for (i = 0; i < fp->dep_func_cnt; i++)
5527 if (IS(fp->dep_func[i].name, name))
5528 return &fp->dep_func[i];
5529
5530 return NULL;
5531}
5532
5533static void hg_fp_add_dep(struct func_prototype *fp, const char *name)
5534{
5535 // is it a dupe?
5536 if (hg_fp_find_dep(fp, name))
5537 return;
5538
5539 if ((fp->dep_func_cnt & 0xff) == 0) {
5540 fp->dep_func = realloc(fp->dep_func,
5541 sizeof(fp->dep_func[0]) * (fp->dep_func_cnt + 0x100));
5542 my_assert_not(fp->dep_func, NULL);
5543 memset(&fp->dep_func[fp->dep_func_cnt], 0,
5544 sizeof(fp->dep_func[0]) * 0x100);
5545 }
5546 fp->dep_func[fp->dep_func_cnt].name = strdup(name);
5547 fp->dep_func_cnt++;
5548}
5549
5550static int hg_fp_cmp_name(const void *p1_, const void *p2_)
5551{
5552 const struct func_prototype *p1 = p1_, *p2 = p2_;
5553 return strcmp(p1->name, p2->name);
5554}
5555
5556#if 0
5557static int hg_fp_cmp_id(const void *p1_, const void *p2_)
5558{
5559 const struct func_prototype *p1 = p1_, *p2 = p2_;
5560 return p1->id - p2->id;
5561}
5562#endif
5563
5564// recursive register dep pass
5565// - track saved regs (part 2)
5566// - try to figure out arg-regs
5567// - calculate reg deps
5568static void gen_hdr_dep_pass(int i, int opcnt, unsigned char *cbits,
5569 struct func_prototype *fp, int regmask_save, int regmask_dst,
5570 int *regmask_dep, int *has_ret)
5571{
5572 struct func_proto_dep *dep;
5573 struct parsed_op *po;
5574 int from_caller = 0;
5575 int depth;
5576 int j, l;
5577 int reg;
5578 int ret;
5579
5580 for (; i < opcnt; i++)
5581 {
5582 if (cbits[i >> 3] & (1 << (i & 7)))
5583 return;
5584 cbits[i >> 3] |= (1 << (i & 7));
5585
5586 po = &ops[i];
5587
5588 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
5589 if (po->btj != NULL) {
5590 // jumptable
5591 for (j = 0; j < po->btj->count; j++) {
5592 gen_hdr_dep_pass(po->btj->d[j].bt_i, opcnt, cbits, fp,
5593 regmask_save, regmask_dst, regmask_dep, has_ret);
5594 }
5595 return;
5596 }
5597
5598 if (po->bt_i < 0) {
5599 ferr(po, "dead branch\n");
5600 return;
5601 }
5602
5603 if (po->flags & OPF_CJMP) {
5604 gen_hdr_dep_pass(po->bt_i, opcnt, cbits, fp,
5605 regmask_save, regmask_dst, regmask_dep, has_ret);
5606 }
5607 else {
5608 i = po->bt_i - 1;
5609 }
5610 continue;
5611 }
5612
5613 if (po->flags & OPF_FARG)
5614 /* (just calculate register deps) */;
5615 else if (po->op == OP_PUSH && po->operand[0].type == OPT_REG)
5616 {
5617 reg = po->operand[0].reg;
5618 if (reg < 0)
5619 ferr(po, "reg not set for push?\n");
5620
5621 if (po->flags & OPF_RSAVE) {
5622 regmask_save |= 1 << reg;
5623 continue;
5624 }
5625 if (po->flags & OPF_DONE)
5626 continue;
5627
5628 depth = 0;
5629 ret = scan_for_pop(i + 1, opcnt,
5630 po->operand[0].name, i + opcnt * 2, 0, &depth, 0);
5631 if (ret == 1) {
5632 regmask_save |= 1 << reg;
5633 po->flags |= OPF_RMD;
5634 scan_for_pop(i + 1, opcnt,
5635 po->operand[0].name, i + opcnt * 3, 0, &depth, 1);
5636 continue;
5637 }
5638 }
5639 else if (po->flags & OPF_RMD)
5640 continue;
5641 else if (po->op == OP_CALL) {
5642 po->regmask_dst |= 1 << xAX;
5643
5644 dep = hg_fp_find_dep(fp, po->operand[0].name);
5645 if (dep != NULL)
5646 dep->regmask_live = regmask_save | regmask_dst;
5647 }
5648 else if (po->op == OP_RET) {
5649 if (po->operand_cnt > 0) {
5650 fp->is_stdcall = 1;
5651 if (fp->argc_stack >= 0
5652 && fp->argc_stack != po->operand[0].val / 4)
5653 ferr(po, "ret mismatch? (%d)\n", fp->argc_stack * 4);
5654 fp->argc_stack = po->operand[0].val / 4;
5655 }
5656 }
5657
5658 if (*has_ret != 0 && (po->flags & OPF_TAIL)) {
5659 if (po->op == OP_CALL) {
5660 j = i;
5661 ret = 1;
5662 }
5663 else {
5664 struct parsed_opr opr = { 0, };
5665 opr.type = OPT_REG;
5666 opr.reg = xAX;
5667 j = -1;
5668 from_caller = 0;
5669 ret = resolve_origin(i, &opr, i + opcnt * 4, &j, &from_caller);
5670 }
5671
5672 if (ret == -1 && from_caller) {
5673 // unresolved eax - probably void func
5674 *has_ret = 0;
5675 }
5676 else {
5677 if (ops[j].op == OP_CALL) {
5678 dep = hg_fp_find_dep(fp, po->operand[0].name);
5679 if (dep != NULL)
5680 dep->ret_dep = 1;
5681 else
5682 *has_ret = 1;
5683 }
5684 else
5685 *has_ret = 1;
5686 }
5687 }
5688
5689 l = regmask_save | regmask_dst;
5690 if (g_bp_frame && !(po->flags & OPF_EBP_S))
5691 l |= 1 << xBP;
5692
5693 l = po->regmask_src & ~l;
5694#if 0
5695 if (l)
5696 fnote(po, "dep |= %04x, dst %04x, save %04x (f %x)\n",
5697 l, regmask_dst, regmask_save, po->flags);
5698#endif
5699 *regmask_dep |= l;
5700 regmask_dst |= po->regmask_dst;
5701
5702 if (po->flags & OPF_TAIL)
5703 return;
5704 }
5705}
5706
5707static void gen_hdr(const char *funcn, int opcnt)
5708{
5709 int save_arg_vars[MAX_ARG_GRP] = { 0, };
5710 unsigned char cbits[MAX_OPS / 8];
5711 const struct parsed_proto *pp_c;
5712 struct parsed_proto *pp;
5713 struct func_prototype *fp;
5714 struct parsed_data *pd;
5715 struct parsed_op *po;
5716 const char *tmpname;
5717 int regmask_dummy = 0;
5718 int regmask_dep;
5719 int max_bp_offset = 0;
5720 int has_ret;
5721 int i, j, l, ret;
5722
5723 if ((hg_fp_cnt & 0xff) == 0) {
5724 hg_fp = realloc(hg_fp, sizeof(hg_fp[0]) * (hg_fp_cnt + 0x100));
5725 my_assert_not(hg_fp, NULL);
5726 memset(hg_fp + hg_fp_cnt, 0, sizeof(hg_fp[0]) * 0x100);
5727 }
5728
5729 fp = &hg_fp[hg_fp_cnt];
5730 snprintf(fp->name, sizeof(fp->name), "%s", funcn);
5731 fp->id = hg_fp_cnt;
5732 fp->argc_stack = -1;
5733 hg_fp_cnt++;
5734
5735 // perhaps already in seed header?
5736 fp->pp = proto_parse(g_fhdr, funcn, 1);
5737 if (fp->pp != NULL) {
5738 fp->argc_stack = fp->pp->argc_stack;
5739 fp->is_stdcall = fp->pp->is_stdcall;
5740 fp->regmask_dep = get_pp_arg_regmask(fp->pp);
5741 fp->has_ret = !IS(fp->pp->ret_type.name, "void");
5742 return;
5743 }
5744
5745 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
5746 g_stack_frame_used = 0;
5747
5748 // pass1:
5749 // - handle ebp/esp frame, remove ops related to it
5750 scan_prologue_epilogue(opcnt);
5751
5752 // pass2:
5753 // - collect calls
5754 // - resolve all branches
5755 for (i = 0; i < opcnt; i++)
5756 {
5757 po = &ops[i];
5758 po->bt_i = -1;
5759 po->btj = NULL;
5760
5761 if (po->flags & (OPF_RMD|OPF_DONE))
5762 continue;
5763
5764 if (po->op == OP_CALL) {
5765 tmpname = opr_name(po, 0);
5766 pp = NULL;
5767 if (po->operand[0].type == OPT_LABEL) {
5768 hg_fp_add_dep(fp, tmpname);
5769
5770 // perhaps a call to already known func?
5771 pp_c = proto_parse(g_fhdr, tmpname, 1);
5772 if (pp_c != NULL)
5773 pp = proto_clone(pp_c);
5774 }
5775 else if (po->datap != NULL) {
5776 pp = calloc(1, sizeof(*pp));
5777 my_assert_not(pp, NULL);
5778
5779 ret = parse_protostr(po->datap, pp);
5780 if (ret < 0)
5781 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
5782 free(po->datap);
5783 po->datap = NULL;
5784 }
5785 if (pp != NULL && pp->is_noreturn)
5786 po->flags |= OPF_TAIL;
5787
5788 po->pp = pp;
5789 continue;
5790 }
5791
5792 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
5793 continue;
5794
5795 if (po->operand[0].type == OPT_REGMEM) {
5796 pd = try_resolve_jumptab(i, opcnt);
5797 if (pd == NULL)
5798 goto tailcall;
5799
5800 po->btj = pd;
5801 continue;
5802 }
5803
5804 for (l = 0; l < opcnt; l++) {
5805 if (g_labels[l] != NULL
5806 && IS(po->operand[0].name, g_labels[l]))
5807 {
5808 add_label_ref(&g_label_refs[l], i);
5809 po->bt_i = l;
5810 break;
5811 }
5812 }
5813
5814 if (po->bt_i != -1 || (po->flags & OPF_RMD))
5815 continue;
5816
5817 if (po->operand[0].type == OPT_LABEL)
5818 // assume tail call
5819 goto tailcall;
5820
5821 ferr(po, "unhandled branch\n");
5822
5823tailcall:
5824 po->op = OP_CALL;
5825 po->flags |= OPF_TAIL;
5826 if (i > 0 && ops[i - 1].op == OP_POP)
5827 po->flags |= OPF_ATAIL;
5828 i--; // reprocess
5829 }
5830
5831 // pass3:
5832 // - remove dead labels
5833 // - handle push <const>/pop pairs
5834 for (i = 0; i < opcnt; i++)
5835 {
5836 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
5837 free(g_labels[i]);
5838 g_labels[i] = NULL;
5839 }
5840
5841 po = &ops[i];
5842 if (po->flags & (OPF_RMD|OPF_DONE))
5843 continue;
5844
5845 if (po->op == OP_PUSH && po->operand[0].type == OPT_CONST)
5846 scan_for_pop_const(i, opcnt);
5847 }
5848
5849 // pass4:
5850 // - process trivial calls
5851 for (i = 0; i < opcnt; i++)
5852 {
5853 po = &ops[i];
5854 if (po->flags & (OPF_RMD|OPF_DONE))
5855 continue;
5856
5857 if (po->op == OP_CALL)
5858 {
5859 pp = process_call_early(i, opcnt, &j);
5860 if (pp != NULL) {
5861 if (!(po->flags & OPF_ATAIL))
5862 // since we know the args, try to collect them
5863 if (collect_call_args_early(po, i, pp, &regmask_dummy) != 0)
5864 pp = NULL;
5865 }
5866
5867 if (pp != NULL) {
5868 if (j >= 0) {
5869 // commit esp adjust
5870 ops[j].flags |= OPF_RMD;
5871 if (ops[j].op != OP_POP)
5872 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
5873 else
5874 ops[j].flags |= OPF_DONE;
5875 }
5876
5877 po->flags |= OPF_DONE;
5878 }
5879 }
5880 }
5881
5882 // pass5:
5883 // - track saved regs (simple)
5884 // - process calls
5885 for (i = 0; i < opcnt; i++)
5886 {
5887 po = &ops[i];
5888 if (po->flags & (OPF_RMD|OPF_DONE))
5889 continue;
5890
5891 if (po->op == OP_PUSH && po->operand[0].type == OPT_REG)
5892 {
5893 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
5894 if (ret == 0) {
5895 // regmask_save |= 1 << po->operand[0].reg; // do it later
5896 po->flags |= OPF_RSAVE | OPF_RMD | OPF_DONE;
5897 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, OPF_RMD);
5898 }
5899 }
5900 else if (po->op == OP_CALL && !(po->flags & OPF_DONE))
5901 {
5902 pp = process_call(i, opcnt);
5903
5904 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
5905 // since we know the args, collect them
5906 ret = collect_call_args(po, i, pp, &regmask_dummy, save_arg_vars,
5907 i + opcnt * 1);
5908 }
5909 }
5910 }
5911
5912 // pass6
5913 memset(cbits, 0, sizeof(cbits));
5914 regmask_dep = 0;
5915 has_ret = -1;
5916
5917 gen_hdr_dep_pass(0, opcnt, cbits, fp, 0, 0, &regmask_dep, &has_ret);
5918
5919 // find unreachable code - must be fixed in IDA
5920 for (i = 0; i < opcnt; i++)
5921 {
5922 if (cbits[i >> 3] & (1 << (i & 7)))
5923 continue;
5924
5925 if (ops[i].op != OP_NOP)
5926 ferr(&ops[i], "unreachable code\n");
5927 }
5928
5929 if (has_ret == -1 && (regmask_dep & (1 << xAX)))
5930 has_ret = 1;
5931
5932 for (i = 0; i < g_eqcnt; i++) {
5933 if (g_eqs[i].offset > max_bp_offset && g_eqs[i].offset < 4*32)
5934 max_bp_offset = g_eqs[i].offset;
5935 }
5936
5937 if (fp->argc_stack < 0) {
5938 max_bp_offset = (max_bp_offset + 3) & ~3;
5939 fp->argc_stack = max_bp_offset / 4;
5940 if ((g_ida_func_attr & IDAFA_BP_FRAME) && fp->argc_stack > 0)
5941 fp->argc_stack--;
5942 }
5943
5944 fp->regmask_dep = regmask_dep & ~(1 << xSP);
5945 fp->has_ret = has_ret;
5946#if 0
5947 printf("// has_ret %d, regmask_dep %x\n",
5948 fp->has_ret, fp->regmask_dep);
5949 output_hdr_fp(stdout, fp, 1);
5950 if (IS(funcn, "sub_100073FD")) exit(1);
5951#endif
5952
5953 gen_x_cleanup(opcnt);
5954}
5955
5956static void hg_fp_resolve_deps(struct func_prototype *fp)
5957{
5958 struct func_prototype fp_s;
5959 int dep;
5960 int i;
5961
5962 // this thing is recursive, so mark first..
5963 fp->dep_resolved = 1;
5964
5965 for (i = 0; i < fp->dep_func_cnt; i++) {
5966 strcpy(fp_s.name, fp->dep_func[i].name);
5967 fp->dep_func[i].proto = bsearch(&fp_s, hg_fp, hg_fp_cnt,
5968 sizeof(hg_fp[0]), hg_fp_cmp_name);
5969 if (fp->dep_func[i].proto != NULL) {
5970 if (!fp->dep_func[i].proto->dep_resolved)
5971 hg_fp_resolve_deps(fp->dep_func[i].proto);
5972
5973 dep = ~fp->dep_func[i].regmask_live
5974 & fp->dep_func[i].proto->regmask_dep;
5975 fp->regmask_dep |= dep;
5976 // printf("dep %s %s |= %x\n", fp->name,
5977 // fp->dep_func[i].name, dep);
5978
5979 if (fp->has_ret == -1)
5980 fp->has_ret = fp->dep_func[i].proto->has_ret;
5981 }
5982 }
5983}
5984
5985static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
5986 int count)
5987{
5988 const struct parsed_proto *pp;
5989 char *p, namebuf[NAMELEN];
5990 const char *name;
5991 int regmask_dep;
5992 int argc_stack;
5993 int j, arg;
5994
5995 for (; count > 0; count--, fp++) {
5996 if (fp->has_ret == -1)
5997 fprintf(fout, "// ret unresolved\n");
5998#if 0
5999 fprintf(fout, "// dep:");
6000 for (j = 0; j < fp->dep_func_cnt; j++) {
6001 fprintf(fout, " %s/", fp->dep_func[j].name);
6002 if (fp->dep_func[j].proto != NULL)
6003 fprintf(fout, "%04x/%d", fp->dep_func[j].proto->regmask_dep,
6004 fp->dep_func[j].proto->has_ret);
6005 }
6006 fprintf(fout, "\n");
6007#endif
6008
6009 p = strchr(fp->name, '@');
6010 if (p != NULL) {
6011 memcpy(namebuf, fp->name, p - fp->name);
6012 namebuf[p - fp->name] = 0;
6013 name = namebuf;
6014 }
6015 else
6016 name = fp->name;
6017 if (name[0] == '_')
6018 name++;
6019
6020 pp = proto_parse(g_fhdr, name, 1);
6021 if (pp != NULL && pp->is_include)
6022 continue;
6023
6024 regmask_dep = fp->regmask_dep;
6025 argc_stack = fp->argc_stack;
6026
6027 fprintf(fout, "%-5s", fp->pp ? fp->pp->ret_type.name :
6028 (fp->has_ret ? "int" : "void"));
6029 if (regmask_dep && (fp->is_stdcall || argc_stack == 0)
6030 && (regmask_dep & ~((1 << xCX) | (1 << xDX))) == 0)
6031 {
6032 fprintf(fout, " __fastcall ");
6033 if (!(regmask_dep & (1 << xDX)) && argc_stack == 0)
6034 argc_stack = 1;
6035 else
6036 argc_stack += 2;
6037 regmask_dep = 0;
6038 }
6039 else if (regmask_dep && !fp->is_stdcall) {
6040 fprintf(fout, "/*__usercall*/ ");
6041 }
6042 else if (regmask_dep) {
6043 fprintf(fout, "/*__userpurge*/ ");
6044 }
6045 else if (fp->is_stdcall)
6046 fprintf(fout, " __stdcall ");
6047 else
6048 fprintf(fout, " __cdecl ");
6049
6050 fprintf(fout, "%s(", name);
6051
6052 arg = 0;
6053 for (j = 0; j < xSP; j++) {
6054 if (regmask_dep & (1 << j)) {
6055 arg++;
6056 if (arg != 1)
6057 fprintf(fout, ", ");
6058 if (fp->pp != NULL)
6059 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
6060 else
6061 fprintf(fout, "int");
6062 fprintf(fout, " a%d/*<%s>*/", arg, regs_r32[j]);
6063 }
6064 }
6065
6066 for (j = 0; j < argc_stack; j++) {
6067 arg++;
6068 if (arg != 1)
6069 fprintf(fout, ", ");
6070 if (fp->pp != NULL) {
6071 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
6072 if (!fp->pp->arg[arg - 1].type.is_ptr)
6073 fprintf(fout, " ");
6074 }
6075 else
6076 fprintf(fout, "int ");
6077 fprintf(fout, "a%d", arg);
6078 }
6079
6080 fprintf(fout, ");\n");
6081 }
6082}
6083
6084static void output_hdr(FILE *fout)
6085{
6086 static const char *lmod_c_names[] = {
6087 [OPLM_UNSPEC] = "???",
6088 [OPLM_BYTE] = "uint8_t",
6089 [OPLM_WORD] = "uint16_t",
6090 [OPLM_DWORD] = "uint32_t",
6091 [OPLM_QWORD] = "uint64_t",
6092 };
6093 const struct scanned_var *var;
6094 int i;
6095
6096 // resolve deps
6097 qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_name);
6098 for (i = 0; i < hg_fp_cnt; i++)
6099 hg_fp_resolve_deps(&hg_fp[i]);
6100
6101 // note: messes up .proto ptr, don't use
6102 //qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_id);
6103
6104 // output variables
6105 for (i = 0; i < hg_var_cnt; i++) {
6106 var = &hg_vars[i];
6107
6108 if (var->is_c_str)
6109 fprintf(fout, "extern %-8s %s[];", "char", var->name);
6110 else
6111 fprintf(fout, "extern %-8s %s;",
6112 lmod_c_names[var->lmod], var->name);
6113
6114 if (var->is_seeded)
6115 fprintf(fout, " // seeded");
6116 fprintf(fout, "\n");
6117 }
6118
6119 fprintf(fout, "\n");
6120
6121 // output function prototypes
6122 output_hdr_fp(fout, hg_fp, hg_fp_cnt);
6123}
6124
6125// read a line, truncating it if it doesn't fit
6126static char *my_fgets(char *s, size_t size, FILE *stream)
6127{
6128 char *ret, *ret2;
6129 char buf[64];
6130 int p;
6131
6132 p = size - 2;
6133 if (p >= 0)
6134 s[p] = 0;
6135
6136 ret = fgets(s, size, stream);
6137 if (ret != NULL && p >= 0 && s[p] != 0 && s[p] != '\n') {
6138 p = sizeof(buf) - 2;
6139 do {
6140 buf[p] = 0;
6141 ret2 = fgets(buf, sizeof(buf), stream);
6142 }
6143 while (ret2 != NULL && buf[p] != 0 && buf[p] != '\n');
6144 }
6145
6146 return ret;
6147}
6148
6149// '=' needs special treatment
6150// also ' quote
6151static char *next_word_s(char *w, size_t wsize, char *s)
6152{
6153 size_t i;
6154
6155 s = sskip(s);
6156
6157 i = 0;
6158 if (*s == '\'') {
6159 w[0] = s[0];
6160 for (i = 1; i < wsize - 1; i++) {
6161 if (s[i] == 0) {
6162 printf("warning: missing closing quote: \"%s\"\n", s);
6163 break;
6164 }
6165 if (s[i] == '\'')
6166 break;
6167 w[i] = s[i];
6168 }
6169 }
6170
6171 for (; i < wsize - 1; i++) {
6172 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
6173 break;
6174 w[i] = s[i];
6175 }
6176 w[i] = 0;
6177
6178 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
6179 printf("warning: '%s' truncated\n", w);
6180
6181 return s + i;
6182}
6183
6184static void scan_variables(FILE *fasm)
6185{
6186 const struct parsed_proto *pp_c;
6187 struct scanned_var *var;
6188 char line[256] = { 0, };
6189 char words[3][256];
6190 char *p = NULL;
6191 int wordc;
6192 int l;
6193
6194 while (!feof(fasm))
6195 {
6196 // skip to next data section
6197 while (my_fgets(line, sizeof(line), fasm))
6198 {
6199 asmln++;
6200
6201 p = sskip(line);
6202 if (*p == 0 || *p == ';')
6203 continue;
6204
6205 p = sskip(next_word_s(words[0], sizeof(words[0]), p));
6206 if (*p == 0 || *p == ';')
6207 continue;
6208
6209 if (*p != 's' || !IS_START(p, "segment para public"))
6210 continue;
6211
6212 break;
6213 }
6214
6215 if (p == NULL || !IS_START(p, "segment para public"))
6216 break;
6217 p = sskip(p + 19);
6218
6219 if (!IS_START(p, "'DATA'"))
6220 continue;
6221
6222 // now process it
6223 while (my_fgets(line, sizeof(line), fasm))
6224 {
6225 asmln++;
6226
6227 p = line;
6228 if (my_isblank(*p))
6229 continue;
6230
6231 p = sskip(p);
6232 if (*p == 0 || *p == ';')
6233 continue;
6234
6235 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
6236 words[wordc][0] = 0;
6237 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
6238 if (*p == 0 || *p == ';') {
6239 wordc++;
6240 break;
6241 }
6242 }
6243
6244 if (wordc == 2 && IS(words[1], "ends"))
6245 break;
6246 if (wordc < 2)
6247 continue;
6248
6249 if ((hg_var_cnt & 0xff) == 0) {
6250 hg_vars = realloc(hg_vars, sizeof(hg_vars[0])
6251 * (hg_var_cnt + 0x100));
6252 my_assert_not(hg_vars, NULL);
6253 memset(hg_vars + hg_var_cnt, 0, sizeof(hg_vars[0]) * 0x100);
6254 }
6255
6256 var = &hg_vars[hg_var_cnt++];
6257 snprintf(var->name, sizeof(var->name), "%s", words[0]);
6258
6259 // maybe already in seed header?
6260 pp_c = proto_parse(g_fhdr, var->name, 1);
6261 if (pp_c != NULL) {
6262 if (pp_c->is_func)
6263 aerr("func?\n");
6264 else if (pp_c->is_fptr) {
6265 var->lmod = OPLM_DWORD;
6266 //var->is_ptr = 1;
6267 }
6268 else if (!guess_lmod_from_c_type(&var->lmod, &pp_c->type))
6269 aerr("unhandled C type '%s' for '%s'\n",
6270 pp_c->type.name, var->name);
6271
6272 var->is_seeded = 1;
6273 continue;
6274 }
6275
6276 if (IS(words[1], "dd"))
6277 var->lmod = OPLM_DWORD;
6278 else if (IS(words[1], "dw"))
6279 var->lmod = OPLM_WORD;
6280 else if (IS(words[1], "db")) {
6281 var->lmod = OPLM_BYTE;
6282 if (wordc >= 3 && (l = strlen(words[2])) > 4) {
6283 if (words[2][0] == '\'' && IS(words[2] + l - 2, ",0"))
6284 var->is_c_str = 1;
6285 }
6286 }
6287 else if (IS(words[1], "dq"))
6288 var->lmod = OPLM_QWORD;
6289 //else if (IS(words[1], "dt"))
6290 else
6291 aerr("type '%s' not known\n", words[1]);
6292 }
6293 }
6294
6295 rewind(fasm);
6296 asmln = 0;
6297}
6298
6299static void set_label(int i, const char *name)
6300{
6301 const char *p;
6302 int len;
6303
6304 len = strlen(name);
6305 p = strchr(name, ':');
6306 if (p != NULL)
6307 len = p - name;
6308
6309 if (g_labels[i] != NULL && !IS_START(g_labels[i], "algn_"))
6310 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
6311 g_labels[i] = realloc(g_labels[i], len + 1);
6312 my_assert_not(g_labels[i], NULL);
6313 memcpy(g_labels[i], name, len);
6314 g_labels[i][len] = 0;
6315}
6316
6317struct chunk_item {
6318 char *name;
6319 long fptr;
6320 int asmln;
6321};
6322
6323static struct chunk_item *func_chunks;
6324static int func_chunk_cnt;
6325static int func_chunk_alloc;
6326
6327static void add_func_chunk(FILE *fasm, const char *name, int line)
6328{
6329 if (func_chunk_cnt >= func_chunk_alloc) {
6330 func_chunk_alloc *= 2;
6331 func_chunks = realloc(func_chunks,
6332 func_chunk_alloc * sizeof(func_chunks[0]));
6333 my_assert_not(func_chunks, NULL);
6334 }
6335 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
6336 func_chunks[func_chunk_cnt].name = strdup(name);
6337 func_chunks[func_chunk_cnt].asmln = line;
6338 func_chunk_cnt++;
6339}
6340
6341static int cmp_chunks(const void *p1, const void *p2)
6342{
6343 const struct chunk_item *c1 = p1, *c2 = p2;
6344 return strcmp(c1->name, c2->name);
6345}
6346
6347static int cmpstringp(const void *p1, const void *p2)
6348{
6349 return strcmp(*(char * const *)p1, *(char * const *)p2);
6350}
6351
6352static void scan_ahead(FILE *fasm)
6353{
6354 char words[2][256];
6355 char line[256];
6356 long oldpos;
6357 int oldasmln;
6358 int wordc;
6359 char *p;
6360 int i;
6361
6362 oldpos = ftell(fasm);
6363 oldasmln = asmln;
6364
6365 while (my_fgets(line, sizeof(line), fasm))
6366 {
6367 wordc = 0;
6368 asmln++;
6369
6370 p = sskip(line);
6371 if (*p == 0)
6372 continue;
6373
6374 if (*p == ';')
6375 {
6376 // get rid of random tabs
6377 for (i = 0; line[i] != 0; i++)
6378 if (line[i] == '\t')
6379 line[i] = ' ';
6380
6381 if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
6382 {
6383 p += 30;
6384 next_word(words[0], sizeof(words[0]), p);
6385 if (words[0][0] == 0)
6386 aerr("missing name for func chunk?\n");
6387
6388 add_func_chunk(fasm, words[0], asmln);
6389 }
6390 else if (IS_START(p, "; sctend"))
6391 break;
6392
6393 continue;
6394 } // *p == ';'
6395
6396 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
6397 words[wordc][0] = 0;
6398 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
6399 if (*p == 0 || *p == ';') {
6400 wordc++;
6401 break;
6402 }
6403 }
6404
6405 if (wordc == 2 && IS(words[1], "ends"))
6406 break;
6407 }
6408
6409 fseek(fasm, oldpos, SEEK_SET);
6410 asmln = oldasmln;
6411}
6412
6413int main(int argc, char *argv[])
6414{
6415 FILE *fout, *fasm, *frlist;
6416 struct parsed_data *pd = NULL;
6417 int pd_alloc = 0;
6418 char **rlist = NULL;
6419 int rlist_len = 0;
6420 int rlist_alloc = 0;
6421 int func_chunks_used = 0;
6422 int func_chunks_sorted = 0;
6423 int func_chunk_i = -1;
6424 long func_chunk_ret = 0;
6425 int func_chunk_ret_ln = 0;
6426 int scanned_ahead = 0;
6427 char line[256];
6428 char words[20][256];
6429 enum opr_lenmod lmod;
6430 char *sctproto = NULL;
6431 int in_func = 0;
6432 int pending_endp = 0;
6433 int skip_func = 0;
6434 int skip_warned = 0;
6435 int eq_alloc;
6436 int verbose = 0;
6437 int multi_seg = 0;
6438 int end = 0;
6439 int arg_out;
6440 int arg;
6441 int pi = 0;
6442 int i, j;
6443 int ret, len;
6444 char *p;
6445 int wordc;
6446
6447 for (arg = 1; arg < argc; arg++) {
6448 if (IS(argv[arg], "-v"))
6449 verbose = 1;
6450 else if (IS(argv[arg], "-rf"))
6451 g_allow_regfunc = 1;
6452 else if (IS(argv[arg], "-m"))
6453 multi_seg = 1;
6454 else if (IS(argv[arg], "-hdr"))
6455 g_header_mode = g_quiet_pp = g_allow_regfunc = 1;
6456 else
6457 break;
6458 }
6459
6460 if (argc < arg + 3) {
6461 printf("usage:\n%s [-v] [-rf] [-m] <.c> <.asm> <hdr.h> [rlist]*\n"
6462 "%s -hdr <out.h> <.asm> <seed.h> [rlist]*\n"
6463 "options:\n"
6464 " -hdr - header generation mode\n"
6465 " -rf - allow unannotated indirect calls\n"
6466 " -m - allow multiple .text sections\n"
6467 "[rlist] is a file with function names to skip,"
6468 " one per line\n",
6469 argv[0], argv[0]);
6470 return 1;
6471 }
6472
6473 arg_out = arg++;
6474
6475 asmfn = argv[arg++];
6476 fasm = fopen(asmfn, "r");
6477 my_assert_not(fasm, NULL);
6478
6479 hdrfn = argv[arg++];
6480 g_fhdr = fopen(hdrfn, "r");
6481 my_assert_not(g_fhdr, NULL);
6482
6483 rlist_alloc = 64;
6484 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
6485 my_assert_not(rlist, NULL);
6486 // needs special handling..
6487 rlist[rlist_len++] = "__alloca_probe";
6488
6489 func_chunk_alloc = 32;
6490 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
6491 my_assert_not(func_chunks, NULL);
6492
6493 memset(words, 0, sizeof(words));
6494
6495 for (; arg < argc; arg++) {
6496 frlist = fopen(argv[arg], "r");
6497 my_assert_not(frlist, NULL);
6498
6499 while (my_fgets(line, sizeof(line), frlist)) {
6500 p = sskip(line);
6501 if (*p == 0 || *p == ';')
6502 continue;
6503 if (*p == '#') {
6504 if (IS_START(p, "#if 0")
6505 || (g_allow_regfunc && IS_START(p, "#if NO_REGFUNC")))
6506 {
6507 skip_func = 1;
6508 }
6509 else if (IS_START(p, "#endif"))
6510 skip_func = 0;
6511 continue;
6512 }
6513 if (skip_func)
6514 continue;
6515
6516 p = next_word(words[0], sizeof(words[0]), p);
6517 if (words[0][0] == 0)
6518 continue;
6519
6520 if (rlist_len >= rlist_alloc) {
6521 rlist_alloc = rlist_alloc * 2 + 64;
6522 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
6523 my_assert_not(rlist, NULL);
6524 }
6525 rlist[rlist_len++] = strdup(words[0]);
6526 }
6527 skip_func = 0;
6528
6529 fclose(frlist);
6530 frlist = NULL;
6531 }
6532
6533 if (rlist_len > 0)
6534 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
6535
6536 fout = fopen(argv[arg_out], "w");
6537 my_assert_not(fout, NULL);
6538
6539 eq_alloc = 128;
6540 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
6541 my_assert_not(g_eqs, NULL);
6542
6543 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
6544 g_label_refs[i].i = -1;
6545 g_label_refs[i].next = NULL;
6546 }
6547
6548 if (g_header_mode)
6549 scan_variables(fasm);
6550
6551 while (my_fgets(line, sizeof(line), fasm))
6552 {
6553 wordc = 0;
6554 asmln++;
6555
6556 p = sskip(line);
6557 if (*p == 0)
6558 continue;
6559
6560 // get rid of random tabs
6561 for (i = 0; line[i] != 0; i++)
6562 if (line[i] == '\t')
6563 line[i] = ' ';
6564
6565 if (*p == ';')
6566 {
6567 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
6568 goto do_pending_endp; // eww..
6569
6570 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
6571 {
6572 static const char *attrs[] = {
6573 "bp-based frame",
6574 "library function",
6575 "static",
6576 "noreturn",
6577 "thunk",
6578 "fpd=",
6579 };
6580
6581 // parse IDA's attribute-list comment
6582 g_ida_func_attr = 0;
6583 p = sskip(p + 13);
6584
6585 for (; *p != 0; p = sskip(p)) {
6586 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
6587 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
6588 g_ida_func_attr |= 1 << i;
6589 p += strlen(attrs[i]);
6590 break;
6591 }
6592 }
6593 if (i == ARRAY_SIZE(attrs)) {
6594 anote("unparsed IDA attr: %s\n", p);
6595 break;
6596 }
6597 if (IS(attrs[i], "fpd=")) {
6598 p = next_word(words[0], sizeof(words[0]), p);
6599 // ignore for now..
6600 }
6601 }
6602 }
6603 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
6604 {
6605 p += 30;
6606 next_word(words[0], sizeof(words[0]), p);
6607 if (words[0][0] == 0)
6608 aerr("missing name for func chunk?\n");
6609
6610 if (!scanned_ahead) {
6611 add_func_chunk(fasm, words[0], asmln);
6612 func_chunks_sorted = 0;
6613 }
6614 }
6615 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
6616 {
6617 if (func_chunk_i >= 0) {
6618 if (func_chunk_i < func_chunk_cnt
6619 && IS(func_chunks[func_chunk_i].name, g_func))
6620 {
6621 // move on to next chunk
6622 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
6623 if (ret)
6624 aerr("seek failed for '%s' chunk #%d\n",
6625 g_func, func_chunk_i);
6626 asmln = func_chunks[func_chunk_i].asmln;
6627 func_chunk_i++;
6628 }
6629 else {
6630 if (func_chunk_ret == 0)
6631 aerr("no return from chunk?\n");
6632 fseek(fasm, func_chunk_ret, SEEK_SET);
6633 asmln = func_chunk_ret_ln;
6634 func_chunk_ret = 0;
6635 pending_endp = 1;
6636 }
6637 }
6638 }
6639 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
6640 func_chunks_used = 1;
6641 p += 20;
6642 if (IS_START(g_func, "sub_")) {
6643 unsigned long addr = strtoul(p, NULL, 16);
6644 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
6645 if (addr > f_addr && !scanned_ahead) {
6646 anote("scan_ahead caused by '%s', addr %lx\n",
6647 g_func, addr);
6648 scan_ahead(fasm);
6649 scanned_ahead = 1;
6650 func_chunks_sorted = 0;
6651 }
6652 }
6653 }
6654 continue;
6655 } // *p == ';'
6656
6657parse_words:
6658 for (i = wordc; i < ARRAY_SIZE(words); i++)
6659 words[i][0] = 0;
6660 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
6661 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
6662 if (*p == 0 || *p == ';') {
6663 wordc++;
6664 break;
6665 }
6666 }
6667 if (*p != 0 && *p != ';')
6668 aerr("too many words\n");
6669
6670 // alow asm patches in comments
6671 if (*p == ';') {
6672 if (IS_START(p, "; sctpatch:")) {
6673 p = sskip(p + 11);
6674 if (*p == 0 || *p == ';')
6675 continue;
6676 goto parse_words; // lame
6677 }
6678 if (IS_START(p, "; sctproto:")) {
6679 sctproto = strdup(p + 11);
6680 }
6681 else if (IS_START(p, "; sctend")) {
6682 end = 1;
6683 if (!pending_endp)
6684 break;
6685 }
6686 }
6687
6688 if (wordc == 0) {
6689 // shouldn't happen
6690 awarn("wordc == 0?\n");
6691 continue;
6692 }
6693
6694 // don't care about this:
6695 if (words[0][0] == '.'
6696 || IS(words[0], "include")
6697 || IS(words[0], "assume") || IS(words[1], "segment")
6698 || IS(words[0], "align"))
6699 {
6700 continue;
6701 }
6702
6703do_pending_endp:
6704 // do delayed endp processing to collect switch jumptables
6705 if (pending_endp) {
6706 if (in_func && !skip_func && !end && wordc >= 2
6707 && ((words[0][0] == 'd' && words[0][2] == 0)
6708 || (words[1][0] == 'd' && words[1][2] == 0)))
6709 {
6710 i = 1;
6711 if (words[1][0] == 'd' && words[1][2] == 0) {
6712 // label
6713 if (g_func_pd_cnt >= pd_alloc) {
6714 pd_alloc = pd_alloc * 2 + 16;
6715 g_func_pd = realloc(g_func_pd,
6716 sizeof(g_func_pd[0]) * pd_alloc);
6717 my_assert_not(g_func_pd, NULL);
6718 }
6719 pd = &g_func_pd[g_func_pd_cnt];
6720 g_func_pd_cnt++;
6721 memset(pd, 0, sizeof(*pd));
6722 strcpy(pd->label, words[0]);
6723 pd->type = OPT_CONST;
6724 pd->lmod = lmod_from_directive(words[1]);
6725 i = 2;
6726 }
6727 else {
6728 if (pd == NULL) {
6729 if (verbose)
6730 anote("skipping alignment byte?\n");
6731 continue;
6732 }
6733 lmod = lmod_from_directive(words[0]);
6734 if (lmod != pd->lmod)
6735 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
6736 }
6737
6738 if (pd->count_alloc < pd->count + wordc) {
6739 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
6740 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
6741 my_assert_not(pd->d, NULL);
6742 }
6743 for (; i < wordc; i++) {
6744 if (IS(words[i], "offset")) {
6745 pd->type = OPT_OFFSET;
6746 i++;
6747 }
6748 p = strchr(words[i], ',');
6749 if (p != NULL)
6750 *p = 0;
6751 if (pd->type == OPT_OFFSET)
6752 pd->d[pd->count].u.label = strdup(words[i]);
6753 else
6754 pd->d[pd->count].u.val = parse_number(words[i]);
6755 pd->d[pd->count].bt_i = -1;
6756 pd->count++;
6757 }
6758 continue;
6759 }
6760
6761 if (in_func && !skip_func) {
6762 if (g_header_mode)
6763 gen_hdr(g_func, pi);
6764 else
6765 gen_func(fout, g_fhdr, g_func, pi);
6766 }
6767
6768 pending_endp = 0;
6769 in_func = 0;
6770 g_ida_func_attr = 0;
6771 skip_warned = 0;
6772 skip_func = 0;
6773 g_func[0] = 0;
6774 func_chunks_used = 0;
6775 func_chunk_i = -1;
6776 if (pi != 0) {
6777 memset(&ops, 0, pi * sizeof(ops[0]));
6778 clear_labels(pi);
6779 pi = 0;
6780 }
6781 g_eqcnt = 0;
6782 for (i = 0; i < g_func_pd_cnt; i++) {
6783 pd = &g_func_pd[i];
6784 if (pd->type == OPT_OFFSET) {
6785 for (j = 0; j < pd->count; j++)
6786 free(pd->d[j].u.label);
6787 }
6788 free(pd->d);
6789 pd->d = NULL;
6790 }
6791 g_func_pd_cnt = 0;
6792 pd = NULL;
6793
6794 if (end)
6795 break;
6796 if (wordc == 0)
6797 continue;
6798 }
6799
6800 if (IS(words[1], "proc")) {
6801 if (in_func)
6802 aerr("proc '%s' while in_func '%s'?\n",
6803 words[0], g_func);
6804 p = words[0];
6805 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
6806 skip_func = 1;
6807 strcpy(g_func, words[0]);
6808 set_label(0, words[0]);
6809 in_func = 1;
6810 continue;
6811 }
6812
6813 if (IS(words[1], "endp"))
6814 {
6815 if (!in_func)
6816 aerr("endp '%s' while not in_func?\n", words[0]);
6817 if (!IS(g_func, words[0]))
6818 aerr("endp '%s' while in_func '%s'?\n",
6819 words[0], g_func);
6820
6821 if ((g_ida_func_attr & IDAFA_THUNK) && pi == 1
6822 && ops[0].op == OP_JMP && ops[0].operand[0].had_ds)
6823 {
6824 // import jump
6825 skip_func = 1;
6826 }
6827
6828 if (!skip_func && func_chunks_used) {
6829 // start processing chunks
6830 struct chunk_item *ci, key = { g_func, 0 };
6831
6832 func_chunk_ret = ftell(fasm);
6833 func_chunk_ret_ln = asmln;
6834 if (!func_chunks_sorted) {
6835 qsort(func_chunks, func_chunk_cnt,
6836 sizeof(func_chunks[0]), cmp_chunks);
6837 func_chunks_sorted = 1;
6838 }
6839 ci = bsearch(&key, func_chunks, func_chunk_cnt,
6840 sizeof(func_chunks[0]), cmp_chunks);
6841 if (ci == NULL)
6842 aerr("'%s' needs chunks, but none found\n", g_func);
6843 func_chunk_i = ci - func_chunks;
6844 for (; func_chunk_i > 0; func_chunk_i--)
6845 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
6846 break;
6847
6848 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
6849 if (ret)
6850 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
6851 asmln = func_chunks[func_chunk_i].asmln;
6852 func_chunk_i++;
6853 continue;
6854 }
6855 pending_endp = 1;
6856 continue;
6857 }
6858
6859 if (wordc == 2 && IS(words[1], "ends")) {
6860 if (!multi_seg) {
6861 end = 1;
6862 if (pending_endp)
6863 goto do_pending_endp;
6864 break;
6865 }
6866
6867 // scan for next text segment
6868 while (my_fgets(line, sizeof(line), fasm)) {
6869 asmln++;
6870 p = sskip(line);
6871 if (*p == 0 || *p == ';')
6872 continue;
6873
6874 if (strstr(p, "segment para public 'CODE' use32"))
6875 break;
6876 }
6877
6878 continue;
6879 }
6880
6881 p = strchr(words[0], ':');
6882 if (p != NULL) {
6883 set_label(pi, words[0]);
6884 continue;
6885 }
6886
6887 if (!in_func || skip_func) {
6888 if (!skip_warned && !skip_func && g_labels[pi] != NULL) {
6889 if (verbose)
6890 anote("skipping from '%s'\n", g_labels[pi]);
6891 skip_warned = 1;
6892 }
6893 free(g_labels[pi]);
6894 g_labels[pi] = NULL;
6895 continue;
6896 }
6897
6898 if (wordc > 1 && IS(words[1], "="))
6899 {
6900 if (wordc != 5)
6901 aerr("unhandled equ, wc=%d\n", wordc);
6902 if (g_eqcnt >= eq_alloc) {
6903 eq_alloc *= 2;
6904 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
6905 my_assert_not(g_eqs, NULL);
6906 }
6907
6908 len = strlen(words[0]);
6909 if (len > sizeof(g_eqs[0].name) - 1)
6910 aerr("equ name too long: %d\n", len);
6911 strcpy(g_eqs[g_eqcnt].name, words[0]);
6912
6913 if (!IS(words[3], "ptr"))
6914 aerr("unhandled equ\n");
6915 if (IS(words[2], "dword"))
6916 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
6917 else if (IS(words[2], "word"))
6918 g_eqs[g_eqcnt].lmod = OPLM_WORD;
6919 else if (IS(words[2], "byte"))
6920 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
6921 else if (IS(words[2], "qword"))
6922 g_eqs[g_eqcnt].lmod = OPLM_QWORD;
6923 else
6924 aerr("bad lmod: '%s'\n", words[2]);
6925
6926 g_eqs[g_eqcnt].offset = parse_number(words[4]);
6927 g_eqcnt++;
6928 continue;
6929 }
6930
6931 if (pi >= ARRAY_SIZE(ops))
6932 aerr("too many ops\n");
6933
6934 parse_op(&ops[pi], words, wordc);
6935
6936 if (sctproto != NULL) {
6937 if (ops[pi].op == OP_CALL || ops[pi].op == OP_JMP)
6938 ops[pi].datap = sctproto;
6939 sctproto = NULL;
6940 }
6941 pi++;
6942 }
6943
6944 if (g_header_mode)
6945 output_hdr(fout);
6946
6947 fclose(fout);
6948 fclose(fasm);
6949 fclose(g_fhdr);
6950
6951 return 0;
6952}
6953
6954// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.