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