notaz.gp2x.de / ia32rtools.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
various fixes
[ia32rtools.git] / tools / translate.c
... / ...
CommitLineData
1#define _GNU_SOURCE
2#include <stdio.h>
3#include <stdlib.h>
4#include <string.h>
5
6#include "my_assert.h"
7#include "my_str.h"
8
9#define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
10#define IS(w, y) !strcmp(w, y)
11#define IS_START(w, y) !strncmp(w, y, strlen(y))
12
13#include "protoparse.h"
14
15static const char *asmfn;
16static int asmln;
17static FILE *g_fhdr;
18
19#define anote(fmt, ...) \
20 printf("%s:%d: note: " fmt, asmfn, asmln, ##__VA_ARGS__)
21#define awarn(fmt, ...) \
22 printf("%s:%d: warning: " fmt, asmfn, asmln, ##__VA_ARGS__)
23#define aerr(fmt, ...) do { \
24 printf("%s:%d: error: " fmt, asmfn, asmln, ##__VA_ARGS__); \
25 fcloseall(); \
26 exit(1); \
27} while (0)
28
29enum op_flags {
30 OPF_RMD = (1 << 0), /* removed or optimized out */
31 OPF_DATA = (1 << 1), /* data processing - writes to dst opr */
32 OPF_FLAGS = (1 << 2), /* sets flags */
33 OPF_JMP = (1 << 3), /* branches, ret and call */
34 OPF_CC = (1 << 4), /* uses flags */
35 OPF_TAIL = (1 << 5), /* ret or tail call */
36 OPF_RSAVE = (1 << 6), /* push/pop is local reg save/load */
37 OPF_REP = (1 << 7), /* prefixed by rep */
38 OPF_REPZ = (1 << 8), /* rep is repe/repz */
39 OPF_REPNZ = (1 << 9), /* rep is repne/repnz */
40 OPF_FARG = (1 << 10), /* push collected as func arg (no reuse) */
41};
42
43enum op_op {
44 OP_INVAL,
45 OP_NOP,
46 OP_PUSH,
47 OP_POP,
48 OP_MOV,
49 OP_LEA,
50 OP_MOVZX,
51 OP_MOVSX,
52 OP_NOT,
53 OP_CDQ,
54 OP_STOS,
55 OP_MOVS,
56 OP_CMPS,
57 OP_RET,
58 OP_ADD,
59 OP_SUB,
60 OP_AND,
61 OP_OR,
62 OP_XOR,
63 OP_SHL,
64 OP_SHR,
65 OP_SAR,
66 OP_ROL,
67 OP_ROR,
68 OP_ADC,
69 OP_SBB,
70 OP_INC,
71 OP_DEC,
72 OP_NEG,
73 OP_MUL,
74 OP_IMUL,
75 OP_DIV,
76 OP_IDIV,
77 OP_TEST,
78 OP_CMP,
79 OP_CALL,
80 OP_JMP,
81 OP_JO,
82 OP_JNO,
83 OP_JC,
84 OP_JNC,
85 OP_JZ,
86 OP_JNZ,
87 OP_JBE,
88 OP_JA,
89 OP_JS,
90 OP_JNS,
91 OP_JP,
92 OP_JNP,
93 OP_JL,
94 OP_JGE,
95 OP_JLE,
96 OP_JG,
97};
98
99enum opr_type {
100 OPT_UNSPEC,
101 OPT_REG,
102 OPT_REGMEM,
103 OPT_LABEL,
104 OPT_OFFSET,
105 OPT_CONST,
106};
107
108// must be sorted (larger len must be further in enum)
109enum opr_lenmod {
110 OPLM_UNSPEC,
111 OPLM_BYTE,
112 OPLM_WORD,
113 OPLM_DWORD,
114};
115
116#define MAX_OPERANDS 3
117
118struct parsed_opr {
119 enum opr_type type;
120 enum opr_lenmod lmod;
121 unsigned int is_ptr:1; // pointer in C
122 unsigned int is_array:1; // array in C
123 unsigned int size_mismatch:1; // type override differs from C
124 unsigned int size_lt:1; // type override is larger than C
125 int reg;
126 unsigned int val;
127 char name[256];
128};
129
130struct parsed_op {
131 enum op_op op;
132 struct parsed_opr operand[MAX_OPERANDS];
133 unsigned int flags;
134 int operand_cnt;
135 int regmask_src; // all referensed regs
136 int regmask_dst;
137 int pfomask; // flagop: parsed_flag_op that can't be delayed
138 int argnum; // push: altered before call arg #
139 int cc_scratch; // scratch storage during analysis
140 int bt_i; // branch target for branches
141 struct parsed_data *btj;// branch targets for jumptables
142 void *datap;
143};
144
145// datap:
146// OP_CALL - ptr to parsed_proto
147// (OPF_CC) - point to one of (OPF_FLAGS) that affects cc op
148
149struct parsed_equ {
150 char name[64];
151 enum opr_lenmod lmod;
152 int offset;
153};
154
155struct parsed_data {
156 char label[256];
157 enum opr_type type;
158 enum opr_lenmod lmod;
159 int count;
160 int count_alloc;
161 struct {
162 union {
163 char *label;
164 unsigned int val;
165 } u;
166 int bt_i;
167 } *d;
168};
169
170struct label_ref {
171 int i;
172 struct label_ref *next;
173};
174
175enum ida_func_attr {
176 IDAFA_BP_FRAME = (1 << 0),
177 IDAFA_LIB_FUNC = (1 << 1),
178 IDAFA_STATIC = (1 << 2),
179 IDAFA_NORETURN = (1 << 3),
180 IDAFA_THUNK = (1 << 4),
181 IDAFA_FPD = (1 << 5),
182};
183
184#define MAX_OPS 4096
185
186static struct parsed_op ops[MAX_OPS];
187static struct parsed_equ *g_eqs;
188static int g_eqcnt;
189static char g_labels[MAX_OPS][32];
190static struct label_ref g_label_refs[MAX_OPS];
191static const struct parsed_proto *g_func_pp;
192static struct parsed_data *g_func_pd;
193static int g_func_pd_cnt;
194static char g_func[256];
195static char g_comment[256];
196static int g_bp_frame;
197static int g_sp_frame;
198static int g_stack_fsz;
199static int g_ida_func_attr;
200#define ferr(op_, fmt, ...) do { \
201 printf("error:%s:#%zd: '%s': " fmt, g_func, (op_) - ops, \
202 dump_op(op_), ##__VA_ARGS__); \
203 fcloseall(); \
204 exit(1); \
205} while (0)
206#define fnote(op_, fmt, ...) \
207 printf("error:%s:#%zd: '%s': " fmt, g_func, (op_) - ops, \
208 dump_op(op_), ##__VA_ARGS__)
209
210#define MAX_REGS 8
211
212const char *regs_r32[] = { "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp" };
213const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" };
214const char *regs_r8l[] = { "al", "bl", "cl", "dl" };
215const char *regs_r8h[] = { "ah", "bh", "ch", "dh" };
216
217enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP };
218
219// possible basic comparison types (without inversion)
220enum parsed_flag_op {
221 PFO_O, // 0 OF=1
222 PFO_C, // 2 CF=1
223 PFO_Z, // 4 ZF=1
224 PFO_BE, // 6 CF=1||ZF=1
225 PFO_S, // 8 SF=1
226 PFO_P, // a PF=1
227 PFO_L, // c SF!=OF
228 PFO_LE, // e ZF=1||SF!=OF
229};
230
231static const char *parsed_flag_op_names[] = {
232 "o", "c", "z", "be", "s", "p", "l", "le"
233};
234
235static int char_array_i(const char *array[], size_t len, const char *s)
236{
237 int i;
238
239 for (i = 0; i < len; i++)
240 if (IS(s, array[i]))
241 return i;
242
243 return -1;
244}
245
246static void printf_number(char *buf, size_t buf_size,
247 unsigned long number)
248{
249 // output in C-friendly form
250 snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
251}
252
253static int parse_reg(enum opr_lenmod *reg_lmod, const char *s)
254{
255 int reg;
256
257 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s);
258 if (reg >= 0) {
259 *reg_lmod = OPLM_DWORD;
260 return reg;
261 }
262 reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s);
263 if (reg >= 0) {
264 *reg_lmod = OPLM_WORD;
265 return reg;
266 }
267 reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s);
268 if (reg >= 0) {
269 *reg_lmod = OPLM_BYTE;
270 return reg;
271 }
272 reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s);
273 if (reg >= 0) {
274 *reg_lmod = OPLM_BYTE;
275 return reg;
276 }
277
278 return -1;
279}
280
281static unsigned long parse_number(const char *number)
282{
283 int len = strlen(number);
284 const char *p = number;
285 char *endp = NULL;
286 unsigned long ret;
287 int neg = 0;
288 int bad;
289
290 if (*p == '-') {
291 neg = 1;
292 p++;
293 }
294 if (len > 1 && *p == '0')
295 p++;
296 if (number[len - 1] == 'h') {
297 ret = strtoul(p, &endp, 16);
298 bad = (*endp != 'h');
299 }
300 else {
301 ret = strtoul(p, &endp, 10);
302 bad = (*endp != 0);
303 }
304 if (bad)
305 aerr("number parsing failed\n");
306 if (neg)
307 ret = -ret;
308 return ret;
309}
310
311static int parse_indmode(char *name, int *regmask, int need_c_cvt)
312{
313 enum opr_lenmod lmod;
314 char cvtbuf[256];
315 char *d = cvtbuf;
316 char *s = name;
317 char w[64];
318 long number;
319 int reg;
320 int c = 0;
321
322 *d = 0;
323
324 while (*s != 0) {
325 d += strlen(d);
326 while (my_isblank(*s))
327 s++;
328 for (; my_issep(*s); d++, s++)
329 *d = *s;
330 while (my_isblank(*s))
331 s++;
332 *d = 0;
333
334 // skip 'ds:' prefix
335 if (IS_START(s, "ds:"))
336 s += 3;
337
338 s = next_idt(w, sizeof(w), s);
339 if (w[0] == 0)
340 break;
341 c++;
342
343 reg = parse_reg(&lmod, w);
344 if (reg >= 0) {
345 *regmask |= 1 << reg;
346 goto pass;
347 }
348
349 if ('0' <= w[0] && w[0] <= '9') {
350 number = parse_number(w);
351 printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number);
352 continue;
353 }
354
355 // probably some label/identifier - pass
356
357pass:
358 snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w);
359 }
360
361 if (need_c_cvt)
362 strcpy(name, cvtbuf);
363
364 return c;
365}
366
367static int is_reg_in_str(const char *s)
368{
369 int i;
370
371 if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
372 return 0;
373
374 for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
375 if (!strncmp(s, regs_r32[i], 3))
376 return 1;
377
378 return 0;
379}
380
381static const char *parse_stack_el(const char *name, char *extra_reg)
382{
383 const char *p, *p2, *s;
384 char *endp = NULL;
385 char buf[32];
386 long val;
387 int len;
388
389 p = name;
390 if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
391 p += 4;
392 if (extra_reg != NULL) {
393 strncpy(extra_reg, name, 3);
394 extra_reg[4] = 0;
395 }
396 }
397
398 if (IS_START(p, "ebp+")) {
399 p += 4;
400
401 p2 = strchr(p, '+');
402 if (p2 != NULL && is_reg_in_str(p)) {
403 if (extra_reg != NULL) {
404 strncpy(extra_reg, p, p2 - p);
405 extra_reg[p2 - p] = 0;
406 }
407 p = p2 + 1;
408 }
409
410 if (!('0' <= *p && *p <= '9'))
411 return p;
412
413 return NULL;
414 }
415
416 if (!IS_START(name, "esp+"))
417 return NULL;
418
419 p = strchr(name + 4, '+');
420 if (p) {
421 // must be a number after esp+, already converted to 0x..
422 s = name + 4;
423 if (!('0' <= *s && *s <= '9')) {
424 aerr("%s nan?\n", __func__);
425 return NULL;
426 }
427 if (s[0] == '0' && s[1] == 'x')
428 s += 2;
429 len = p - s;
430 if (len < sizeof(buf) - 1) {
431 strncpy(buf, s, len);
432 buf[len] = 0;
433 val = strtol(buf, &endp, 16);
434 if (val == 0 || *endp != 0) {
435 aerr("%s num parse fail for '%s'\n", __func__, buf);
436 return NULL;
437 }
438 }
439 p++;
440 }
441 else
442 p = name + 4;
443
444 if ('0' <= *p && *p <= '9')
445 return NULL;
446
447 return p;
448}
449
450static int guess_lmod_from_name(struct parsed_opr *opr)
451{
452 if (!strncmp(opr->name, "dword_", 6)) {
453 opr->lmod = OPLM_DWORD;
454 return 1;
455 }
456 if (!strncmp(opr->name, "word_", 5)) {
457 opr->lmod = OPLM_WORD;
458 return 1;
459 }
460 if (!strncmp(opr->name, "byte_", 5)) {
461 opr->lmod = OPLM_BYTE;
462 return 1;
463 }
464 return 0;
465}
466
467static int guess_lmod_from_c_type(enum opr_lenmod *lmod,
468 const struct parsed_type *c_type)
469{
470 static const char *dword_types[] = {
471 "int", "_DWORD", "DWORD", "HANDLE", "HWND", "HMODULE",
472 "WPARAM", "LPARAM", "UINT",
473 };
474 static const char *word_types[] = {
475 "uint16_t", "int16_t",
476 "unsigned __int16", "__int16",
477 };
478 static const char *byte_types[] = {
479 "uint8_t", "int8_t", "char",
480 "unsigned __int8", "__int8", "BYTE",
481 };
482 const char *n;
483 int i;
484
485 if (c_type->is_ptr) {
486 *lmod = OPLM_DWORD;
487 return 1;
488 }
489
490 n = skip_type_mod(c_type->name);
491
492 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
493 if (IS(n, dword_types[i])) {
494 *lmod = OPLM_DWORD;
495 return 1;
496 }
497 }
498
499 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
500 if (IS(n, word_types[i])) {
501 *lmod = OPLM_WORD;
502 return 1;
503 }
504 }
505
506 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
507 if (IS(n, byte_types[i])) {
508 *lmod = OPLM_BYTE;
509 return 1;
510 }
511 }
512
513 return 0;
514}
515
516static enum opr_type lmod_from_directive(const char *d)
517{
518 if (IS(d, "dd"))
519 return OPLM_DWORD;
520 else if (IS(d, "dw"))
521 return OPLM_WORD;
522 else if (IS(d, "db"))
523 return OPLM_BYTE;
524
525 aerr("unhandled directive: '%s'\n", d);
526 return OPLM_UNSPEC;
527}
528
529static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
530 int *regmask)
531{
532 opr->type = OPT_REG;
533 opr->reg = reg;
534 opr->lmod = lmod;
535 *regmask |= 1 << reg;
536}
537
538static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
539 int *extra_offs);
540
541static int parse_operand(struct parsed_opr *opr,
542 int *regmask, int *regmask_indirect,
543 char words[16][256], int wordc, int w, unsigned int op_flags)
544{
545 const struct parsed_proto *pp;
546 enum opr_lenmod tmplmod;
547 unsigned long number;
548 int ret, len;
549 int wordc_in;
550 char *tmp;
551 int i;
552
553 if (w >= wordc)
554 aerr("parse_operand w %d, wordc %d\n", w, wordc);
555
556 opr->reg = xUNSPEC;
557
558 for (i = w; i < wordc; i++) {
559 len = strlen(words[i]);
560 if (words[i][len - 1] == ',') {
561 words[i][len - 1] = 0;
562 wordc = i + 1;
563 break;
564 }
565 }
566
567 wordc_in = wordc - w;
568
569 if ((op_flags & OPF_JMP) && wordc_in > 0
570 && !('0' <= words[w][0] && words[w][0] <= '9'))
571 {
572 const char *label = NULL;
573
574 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
575 && IS(words[w + 1], "ptr"))
576 label = words[w + 2];
577 else if (wordc_in == 2 && IS(words[w], "short"))
578 label = words[w + 1];
579 else if (wordc_in == 1
580 && strchr(words[w], '[') == NULL
581 && parse_reg(&tmplmod, words[w]) < 0)
582 label = words[w];
583
584 if (label != NULL) {
585 opr->type = OPT_LABEL;
586 if (IS_START(label, "ds:"))
587 label += 3;
588 strcpy(opr->name, label);
589 return wordc;
590 }
591 }
592
593 if (wordc_in >= 3) {
594 if (IS(words[w + 1], "ptr")) {
595 if (IS(words[w], "dword"))
596 opr->lmod = OPLM_DWORD;
597 else if (IS(words[w], "word"))
598 opr->lmod = OPLM_WORD;
599 else if (IS(words[w], "byte"))
600 opr->lmod = OPLM_BYTE;
601 else
602 aerr("type parsing failed\n");
603 w += 2;
604 wordc_in = wordc - w;
605 }
606 }
607
608 if (wordc_in == 2) {
609 if (IS(words[w], "offset")) {
610 opr->type = OPT_OFFSET;
611 strcpy(opr->name, words[w + 1]);
612 return wordc;
613 }
614 if (IS(words[w], "(offset")) {
615 char *p = strchr(words[w + 1], ')');
616 if (p == NULL)
617 aerr("parse of bracketed offset failed\n");
618 *p = 0;
619 opr->type = OPT_OFFSET;
620 strcpy(opr->name, words[w + 1]);
621 return wordc;
622 }
623 }
624
625 if (wordc_in != 1)
626 aerr("parse_operand 1 word expected\n");
627
628 tmp = words[w];
629 if (IS_START(tmp, "ds:"))
630 tmp += 3;
631 strcpy(opr->name, tmp);
632
633 if (words[w][0] == '[') {
634 opr->type = OPT_REGMEM;
635 ret = sscanf(words[w], "[%[^]]]", opr->name);
636 if (ret != 1)
637 aerr("[] parse failure\n");
638
639 parse_indmode(opr->name, regmask_indirect, 1);
640 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL)) {
641 // might be an equ
642 struct parsed_equ *eq =
643 equ_find(NULL, parse_stack_el(opr->name, NULL), &i);
644 if (eq)
645 opr->lmod = eq->lmod;
646 }
647 return wordc;
648 }
649 else if (strchr(words[w], '[')) {
650 // label[reg] form
651 opr->type = OPT_REGMEM;
652 if (opr->lmod == OPLM_UNSPEC)
653 guess_lmod_from_name(opr);
654 parse_indmode(strchr(words[w], '['), regmask_indirect, 0);
655 return wordc;
656 }
657 else if (('0' <= words[w][0] && words[w][0] <= '9')
658 || words[w][0] == '-')
659 {
660 number = parse_number(words[w]);
661 opr->type = OPT_CONST;
662 opr->val = number;
663 printf_number(opr->name, sizeof(opr->name), number);
664 return wordc;
665 }
666
667 ret = parse_reg(&tmplmod, opr->name);
668 if (ret >= 0) {
669 setup_reg_opr(opr, ret, tmplmod, regmask);
670 return wordc;
671 }
672
673 // most likely var in data segment
674 opr->type = OPT_LABEL;
675
676 pp = proto_parse(g_fhdr, opr->name);
677 if (pp != NULL) {
678 if (pp->is_fptr) {
679 opr->lmod = OPLM_DWORD;
680 opr->is_ptr = 1;
681 }
682 else {
683 tmplmod = OPLM_UNSPEC;
684 if (!guess_lmod_from_c_type(&tmplmod, &pp->type))
685 anote("unhandled C type '%s' for '%s'\n",
686 pp->type.name, opr->name);
687
688 if (opr->lmod == OPLM_UNSPEC)
689 opr->lmod = tmplmod;
690 else if (opr->lmod != tmplmod) {
691 opr->size_mismatch = 1;
692 if (tmplmod < opr->lmod)
693 opr->size_lt = 1;
694 }
695 opr->is_ptr = pp->type.is_ptr;
696 }
697 opr->is_array = pp->type.is_array;
698 }
699
700 if (opr->lmod == OPLM_UNSPEC)
701 guess_lmod_from_name(opr);
702 return wordc;
703}
704
705static const struct {
706 const char *name;
707 unsigned int flags;
708} pref_table[] = {
709 { "rep", OPF_REP },
710 { "repe", OPF_REP|OPF_REPZ },
711 { "repz", OPF_REP|OPF_REPZ },
712 { "repne", OPF_REP|OPF_REPNZ },
713 { "repnz", OPF_REP|OPF_REPNZ },
714};
715
716static const struct {
717 const char *name;
718 enum op_op op;
719 unsigned int minopr;
720 unsigned int maxopr;
721 unsigned int flags;
722} op_table[] = {
723 { "nop", OP_NOP, 0, 0, 0 },
724 { "push", OP_PUSH, 1, 1, 0 },
725 { "pop", OP_POP, 1, 1, OPF_DATA },
726 { "mov" , OP_MOV, 2, 2, OPF_DATA },
727 { "lea", OP_LEA, 2, 2, OPF_DATA },
728 { "movzx",OP_MOVZX, 2, 2, OPF_DATA },
729 { "movsx",OP_MOVSX, 2, 2, OPF_DATA },
730 { "not", OP_NOT, 1, 1, OPF_DATA },
731 { "cdq", OP_CDQ, 0, 0, OPF_DATA },
732 { "stosb",OP_STOS, 0, 0, OPF_DATA },
733 { "stosw",OP_STOS, 0, 0, OPF_DATA },
734 { "stosd",OP_STOS, 0, 0, OPF_DATA },
735 { "movsb",OP_MOVS, 0, 0, OPF_DATA },
736 { "movsw",OP_MOVS, 0, 0, OPF_DATA },
737 { "movsd",OP_MOVS, 0, 0, OPF_DATA },
738 { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
739 { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
740 { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
741 { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
742 { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
743 { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
744 { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS },
745 { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS },
746 { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
747 { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
748 { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
749 { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
750 { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
751 { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
752 { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
753 { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
754 { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
755 { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS },
756 { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS },
757 { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS },
758 { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS },
759 { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS },
760 { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS },
761 { "test", OP_TEST, 2, 2, OPF_FLAGS },
762 { "cmp", OP_CMP, 2, 2, OPF_FLAGS },
763 { "retn", OP_RET, 0, 1, OPF_JMP|OPF_TAIL },
764 { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS },
765 { "jmp", OP_JMP, 1, 1, OPF_JMP },
766 { "jo", OP_JO, 1, 1, OPF_JMP|OPF_CC }, // 70 OF=1
767 { "jno", OP_JNO, 1, 1, OPF_JMP|OPF_CC }, // 71 OF=0
768 { "jc", OP_JC, 1, 1, OPF_JMP|OPF_CC }, // 72 CF=1
769 { "jb", OP_JC, 1, 1, OPF_JMP|OPF_CC }, // 72
770 { "jnc", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 CF=0
771 { "jnb", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73
772 { "jae", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73
773 { "jz", OP_JZ, 1, 1, OPF_JMP|OPF_CC }, // 74 ZF=1
774 { "je", OP_JZ, 1, 1, OPF_JMP|OPF_CC }, // 74
775 { "jnz", OP_JNZ, 1, 1, OPF_JMP|OPF_CC }, // 75 ZF=0
776 { "jne", OP_JNZ, 1, 1, OPF_JMP|OPF_CC }, // 75
777 { "jbe", OP_JBE, 1, 1, OPF_JMP|OPF_CC }, // 76 CF=1 || ZF=1
778 { "jna", OP_JBE, 1, 1, OPF_JMP|OPF_CC }, // 76
779 { "ja", OP_JA, 1, 1, OPF_JMP|OPF_CC }, // 77 CF=0 && ZF=0
780 { "jnbe", OP_JA, 1, 1, OPF_JMP|OPF_CC }, // 77
781 { "js", OP_JS, 1, 1, OPF_JMP|OPF_CC }, // 78 SF=1
782 { "jns", OP_JNS, 1, 1, OPF_JMP|OPF_CC }, // 79 SF=0
783 { "jp", OP_JP, 1, 1, OPF_JMP|OPF_CC }, // 7a PF=1
784 { "jpe", OP_JP, 1, 1, OPF_JMP|OPF_CC }, // 7a
785 { "jnp", OP_JNP, 1, 1, OPF_JMP|OPF_CC }, // 7b PF=0
786 { "jpo", OP_JNP, 1, 1, OPF_JMP|OPF_CC }, // 7b
787 { "jl", OP_JL, 1, 1, OPF_JMP|OPF_CC }, // 7c SF!=OF
788 { "jnge", OP_JL, 1, 1, OPF_JMP|OPF_CC }, // 7c
789 { "jge", OP_JGE, 1, 1, OPF_JMP|OPF_CC }, // 7d SF=OF
790 { "jnl", OP_JGE, 1, 1, OPF_JMP|OPF_CC }, // 7d
791 { "jle", OP_JLE, 1, 1, OPF_JMP|OPF_CC }, // 7e ZF=1 || SF!=OF
792 { "jng", OP_JLE, 1, 1, OPF_JMP|OPF_CC }, // 7e
793 { "jg", OP_JG, 1, 1, OPF_JMP|OPF_CC }, // 7f ZF=0 && SF=OF
794 { "jnle", OP_JG, 1, 1, OPF_JMP|OPF_CC }, // 7f
795 { "seto", OP_JO, 1, 1, OPF_DATA|OPF_CC },
796 { "setno", OP_JNO, 1, 1, OPF_DATA|OPF_CC },
797 { "setc", OP_JC, 1, 1, OPF_DATA|OPF_CC },
798 { "setb", OP_JC, 1, 1, OPF_DATA|OPF_CC },
799 { "setnc", OP_JNC, 1, 1, OPF_DATA|OPF_CC },
800 { "setae", OP_JNC, 1, 1, OPF_DATA|OPF_CC },
801 { "setz", OP_JZ, 1, 1, OPF_DATA|OPF_CC },
802 { "sete", OP_JZ, 1, 1, OPF_DATA|OPF_CC },
803 { "setnz", OP_JNZ, 1, 1, OPF_DATA|OPF_CC },
804 { "setne", OP_JNZ, 1, 1, OPF_DATA|OPF_CC },
805 { "setbe", OP_JBE, 1, 1, OPF_DATA|OPF_CC },
806 { "setna", OP_JBE, 1, 1, OPF_DATA|OPF_CC },
807 { "seta", OP_JA, 1, 1, OPF_DATA|OPF_CC },
808 { "setnbe", OP_JA, 1, 1, OPF_DATA|OPF_CC },
809 { "sets", OP_JS, 1, 1, OPF_DATA|OPF_CC },
810 { "setns", OP_JNS, 1, 1, OPF_DATA|OPF_CC },
811 { "setp", OP_JP, 1, 1, OPF_DATA|OPF_CC },
812 { "setpe", OP_JP, 1, 1, OPF_DATA|OPF_CC },
813 { "setnp", OP_JNP, 1, 1, OPF_DATA|OPF_CC },
814 { "setpo", OP_JNP, 1, 1, OPF_DATA|OPF_CC },
815 { "setl", OP_JL, 1, 1, OPF_DATA|OPF_CC },
816 { "setnge", OP_JL, 1, 1, OPF_DATA|OPF_CC },
817 { "setge", OP_JGE, 1, 1, OPF_DATA|OPF_CC },
818 { "setnl", OP_JGE, 1, 1, OPF_DATA|OPF_CC },
819 { "setle", OP_JLE, 1, 1, OPF_DATA|OPF_CC },
820 { "setng", OP_JLE, 1, 1, OPF_DATA|OPF_CC },
821 { "setg", OP_JG, 1, 1, OPF_DATA|OPF_CC },
822 { "setnle", OP_JG, 1, 1, OPF_DATA|OPF_CC },
823};
824
825static void parse_op(struct parsed_op *op, char words[16][256], int wordc)
826{
827 enum opr_lenmod lmod = OPLM_UNSPEC;
828 int prefix_flags = 0;
829 int regmask_ind;
830 int regmask;
831 int op_w = 0;
832 int opr = 0;
833 int w = 0;
834 int i;
835
836 for (i = 0; i < ARRAY_SIZE(pref_table); i++) {
837 if (IS(words[w], pref_table[i].name)) {
838 prefix_flags = pref_table[i].flags;
839 break;
840 }
841 }
842
843 if (prefix_flags) {
844 if (wordc <= 1)
845 aerr("lone prefix: '%s'\n", words[0]);
846 w++;
847 }
848
849 op_w = w;
850 for (i = 0; i < ARRAY_SIZE(op_table); i++) {
851 if (IS(words[w], op_table[i].name))
852 break;
853 }
854
855 if (i == ARRAY_SIZE(op_table))
856 aerr("unhandled op: '%s'\n", words[0]);
857 w++;
858
859 op->op = op_table[i].op;
860 op->flags = op_table[i].flags | prefix_flags;
861 op->regmask_src = op->regmask_dst = 0;
862
863 for (opr = 0; opr < op_table[i].minopr; opr++) {
864 regmask = regmask_ind = 0;
865 w = parse_operand(&op->operand[opr], &regmask, &regmask_ind,
866 words, wordc, w, op->flags);
867
868 if (opr == 0 && (op->flags & OPF_DATA))
869 op->regmask_dst = regmask;
870 // for now, mark dst as src too
871 op->regmask_src |= regmask | regmask_ind;
872 }
873
874 for (; w < wordc && opr < op_table[i].maxopr; opr++) {
875 w = parse_operand(&op->operand[opr],
876 &op->regmask_src, &op->regmask_src,
877 words, wordc, w, op->flags);
878 }
879
880 if (w < wordc)
881 aerr("parse_op %s incomplete: %d/%d\n",
882 words[0], w, wordc);
883
884 // special cases
885 op->operand_cnt = opr;
886 if (!strncmp(op_table[i].name, "set", 3))
887 op->operand[0].lmod = OPLM_BYTE;
888
889 // ops with implicit argumets
890 switch (op->op) {
891 case OP_CDQ:
892 op->operand_cnt = 2;
893 setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst);
894 setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src);
895 break;
896
897 case OP_STOS:
898 if (op->operand_cnt != 0)
899 break;
900 if (IS(words[op_w], "stosb"))
901 lmod = OPLM_BYTE;
902 else if (IS(words[op_w], "stosw"))
903 lmod = OPLM_WORD;
904 else if (IS(words[op_w], "stosd"))
905 lmod = OPLM_DWORD;
906 op->operand_cnt = 3;
907 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
908 setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src);
909 op->regmask_dst = op->regmask_src;
910 setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD, &op->regmask_src);
911 break;
912
913 case OP_MOVS:
914 case OP_CMPS:
915 if (op->operand_cnt != 0)
916 break;
917 if (words[op_w][4] == 'b')
918 lmod = OPLM_BYTE;
919 else if (words[op_w][4] == 'w')
920 lmod = OPLM_WORD;
921 else if (words[op_w][4] == 'd')
922 lmod = OPLM_DWORD;
923 op->operand_cnt = 3;
924 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
925 setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src);
926 setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src);
927 op->regmask_dst = op->regmask_src;
928 break;
929
930 case OP_IMUL:
931 if (op->operand_cnt != 1)
932 break;
933 // fallthrough
934 case OP_MUL:
935 // singleop mul
936 op->regmask_dst = (1 << xDX) | (1 << xAX);
937 op->regmask_src |= (1 << xAX);
938 if (op->operand[0].lmod == OPLM_UNSPEC)
939 op->operand[0].lmod = OPLM_DWORD;
940 break;
941
942 case OP_DIV:
943 case OP_IDIV:
944 // we could set up operands for edx:eax, but there is no real need to
945 // (see is_opr_modified())
946 regmask = (1 << xDX) | (1 << xAX);
947 op->regmask_dst = regmask;
948 op->regmask_src |= regmask;
949 if (op->operand[0].lmod == OPLM_UNSPEC)
950 op->operand[0].lmod = OPLM_DWORD;
951 break;
952
953 case OP_SHL:
954 case OP_SHR:
955 case OP_SAR:
956 case OP_ROL:
957 case OP_ROR:
958 if (op->operand[1].lmod == OPLM_UNSPEC)
959 op->operand[1].lmod = OPLM_BYTE;
960 break;
961
962 case OP_PUSH:
963 if (op->operand[0].lmod == OPLM_UNSPEC
964 && (op->operand[0].type == OPT_CONST
965 || op->operand[0].type == OPT_OFFSET
966 || op->operand[0].type == OPT_LABEL))
967 op->operand[0].lmod = OPLM_DWORD;
968 break;
969
970 // alignment
971 case OP_MOV:
972 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
973 && op->operand[0].reg == xDI && op->operand[1].reg == xDI)
974 {
975 op->flags |= OPF_RMD;
976 }
977 break;
978
979 case OP_LEA:
980 if (op->operand[0].type == OPT_REG
981 && op->operand[1].type == OPT_REGMEM)
982 {
983 char buf[16];
984 snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name);
985 if (IS(buf, op->operand[1].name))
986 op->flags |= OPF_RMD;
987 }
988 break;
989
990 default:
991 break;
992 }
993}
994
995static const char *op_name(enum op_op op)
996{
997 int i;
998
999 for (i = 0; i < ARRAY_SIZE(op_table); i++)
1000 if (op_table[i].op == op)
1001 return op_table[i].name;
1002
1003 return "???";
1004}
1005
1006// debug
1007static const char *dump_op(struct parsed_op *po)
1008{
1009 static char out[128];
1010 char *p = out;
1011 int i;
1012
1013 if (po == NULL)
1014 return "???";
1015
1016 snprintf(out, sizeof(out), "%s", op_name(po->op));
1017 for (i = 0; i < po->operand_cnt; i++) {
1018 p += strlen(p);
1019 if (i > 0)
1020 *p++ = ',';
1021 snprintf(p, sizeof(out) - (p - out),
1022 po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s",
1023 po->operand[i].name);
1024 }
1025
1026 return out;
1027}
1028
1029static const char *lmod_type_u(struct parsed_op *po,
1030 enum opr_lenmod lmod)
1031{
1032 switch (lmod) {
1033 case OPLM_DWORD:
1034 return "u32";
1035 case OPLM_WORD:
1036 return "u16";
1037 case OPLM_BYTE:
1038 return "u8";
1039 default:
1040 ferr(po, "invalid lmod: %d\n", lmod);
1041 return "(_invalid_)";
1042 }
1043}
1044
1045static const char *lmod_cast_u(struct parsed_op *po,
1046 enum opr_lenmod lmod)
1047{
1048 switch (lmod) {
1049 case OPLM_DWORD:
1050 return "";
1051 case OPLM_WORD:
1052 return "(u16)";
1053 case OPLM_BYTE:
1054 return "(u8)";
1055 default:
1056 ferr(po, "invalid lmod: %d\n", lmod);
1057 return "(_invalid_)";
1058 }
1059}
1060
1061static const char *lmod_cast_u_ptr(struct parsed_op *po,
1062 enum opr_lenmod lmod)
1063{
1064 switch (lmod) {
1065 case OPLM_DWORD:
1066 return "*(u32 *)";
1067 case OPLM_WORD:
1068 return "*(u16 *)";
1069 case OPLM_BYTE:
1070 return "*(u8 *)";
1071 default:
1072 ferr(po, "invalid lmod: %d\n", lmod);
1073 return "(_invalid_)";
1074 }
1075}
1076
1077static const char *lmod_cast_s(struct parsed_op *po,
1078 enum opr_lenmod lmod)
1079{
1080 switch (lmod) {
1081 case OPLM_DWORD:
1082 return "(s32)";
1083 case OPLM_WORD:
1084 return "(s16)";
1085 case OPLM_BYTE:
1086 return "(s8)";
1087 default:
1088 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1089 return "(_invalid_)";
1090 }
1091}
1092
1093static const char *lmod_cast(struct parsed_op *po,
1094 enum opr_lenmod lmod, int is_signed)
1095{
1096 return is_signed ?
1097 lmod_cast_s(po, lmod) :
1098 lmod_cast_u(po, lmod);
1099}
1100
1101static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
1102{
1103 switch (lmod) {
1104 case OPLM_DWORD:
1105 return 4;
1106 case OPLM_WORD:
1107 return 2;
1108 case OPLM_BYTE:
1109 return 1;
1110 default:
1111 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1112 return 0;
1113 }
1114}
1115
1116static const char *opr_name(struct parsed_op *po, int opr_num)
1117{
1118 if (opr_num >= po->operand_cnt)
1119 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1120 return po->operand[opr_num].name;
1121}
1122
1123static unsigned int opr_const(struct parsed_op *po, int opr_num)
1124{
1125 if (opr_num >= po->operand_cnt)
1126 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1127 if (po->operand[opr_num].type != OPT_CONST)
1128 ferr(po, "opr %d: const expected\n", opr_num);
1129 return po->operand[opr_num].val;
1130}
1131
1132static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr)
1133{
1134 if ((unsigned int)popr->reg >= MAX_REGS)
1135 ferr(po, "invalid reg: %d\n", popr->reg);
1136 return regs_r32[popr->reg];
1137}
1138
1139static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
1140 int *extra_offs)
1141{
1142 const char *p;
1143 char *endp;
1144 int namelen;
1145 int i;
1146
1147 *extra_offs = 0;
1148 namelen = strlen(name);
1149
1150 p = strchr(name, '+');
1151 if (p != NULL) {
1152 namelen = p - name;
1153 if (namelen <= 0)
1154 ferr(po, "equ parse failed for '%s'\n", name);
1155
1156 if (IS_START(p, "0x"))
1157 p += 2;
1158 *extra_offs = strtol(p, &endp, 16);
1159 if (*endp != 0)
1160 ferr(po, "equ parse failed for '%s'\n", name);
1161 }
1162
1163 for (i = 0; i < g_eqcnt; i++)
1164 if (strncmp(g_eqs[i].name, name, namelen) == 0
1165 && g_eqs[i].name[namelen] == 0)
1166 break;
1167 if (i >= g_eqcnt) {
1168 if (po != NULL)
1169 ferr(po, "unresolved equ name: '%s'\n", name);
1170 return NULL;
1171 }
1172
1173 return &g_eqs[i];
1174}
1175
1176static void stack_frame_access(struct parsed_op *po,
1177 struct parsed_opr *popr, char *buf, size_t buf_size,
1178 const char *name, const char *cast, int is_src, int is_lea)
1179{
1180 enum opr_lenmod tmp_lmod = OPLM_UNSPEC;
1181 const char *prefix = "";
1182 const char *bp_arg = NULL;
1183 char ofs_reg[16] = { 0, };
1184 struct parsed_equ *eq;
1185 const char *p;
1186 char *endp = NULL;
1187 int i, arg_i, arg_s;
1188 int stack_ra = 0;
1189 int offset = 0;
1190 int sf_ofs;
1191 int lim;
1192
1193 if (!IS_START(name, "ebp-")) {
1194 bp_arg = parse_stack_el(name, ofs_reg);
1195 snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
1196 eq = equ_find(po, bp_arg, &offset);
1197 if (eq == NULL)
1198 ferr(po, "detected but missing eq\n");
1199 offset += eq->offset;
1200 }
1201 else {
1202 p = name + 4;
1203 if (IS_START(p, "0x"))
1204 p += 2;
1205 offset = -strtoul(p, &endp, 16);
1206 if (*endp != 0)
1207 ferr(po, "ebp- parse of '%s' failed\n", name);
1208 }
1209
1210 if (!strncmp(name, "ebp", 3))
1211 stack_ra = 4;
1212
1213 if (stack_ra <= offset && offset < stack_ra + 4)
1214 ferr(po, "reference to ra? %d %d\n", offset, stack_ra);
1215
1216 if (offset > stack_ra)
1217 {
1218 arg_i = (offset - stack_ra - 4) / 4;
1219 if (arg_i < 0 || arg_i >= g_func_pp->argc_stack)
1220 {
1221 if (g_func_pp->is_vararg
1222 && arg_i == g_func_pp->argc_stack && is_lea)
1223 {
1224 // should be va_list
1225 if (cast[0] == 0)
1226 cast = "(u32)";
1227 snprintf(buf, buf_size, "%sap", cast);
1228 return;
1229 }
1230 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1231 offset, bp_arg, arg_i);
1232 }
1233 if (ofs_reg[0] != 0)
1234 ferr(po, "offset reg on arg access?\n");
1235
1236 for (i = arg_s = 0; i < g_func_pp->argc; i++) {
1237 if (g_func_pp->arg[i].reg != NULL)
1238 continue;
1239 if (arg_s == arg_i)
1240 break;
1241 arg_s++;
1242 }
1243 if (i == g_func_pp->argc)
1244 ferr(po, "arg %d not in prototype?\n", arg_i);
1245
1246 popr->is_ptr = g_func_pp->arg[i].type.is_ptr;
1247
1248 switch (popr->lmod)
1249 {
1250 case OPLM_BYTE:
1251 if (is_lea)
1252 ferr(po, "lea/byte to arg?\n");
1253 if (is_src && (offset & 3) == 0)
1254 snprintf(buf, buf_size, "(u8)a%d", i + 1);
1255 else
1256 snprintf(buf, buf_size, "BYTE%d(a%d)", offset & 3, i + 1);
1257 break;
1258
1259 case OPLM_WORD:
1260 if (is_lea)
1261 ferr(po, "lea/word to arg?\n");
1262 if (offset & 1)
1263 ferr(po, "unaligned arg access\n");
1264 if (is_src && (offset & 2) == 0)
1265 snprintf(buf, buf_size, "(u16)a%d", i + 1);
1266 else
1267 snprintf(buf, buf_size, "%sWORD(a%d)",
1268 (offset & 2) ? "HI" : "LO", i + 1);
1269 break;
1270
1271 case OPLM_DWORD:
1272 if (cast[0])
1273 prefix = cast;
1274 else if (is_src)
1275 prefix = "(u32)";
1276 if (offset & 3) {
1277 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1278 if (is_lea)
1279 snprintf(buf, buf_size, "(u32)&a%d + %d",
1280 i + 1, offset & 3);
1281 else
1282 snprintf(buf, buf_size, "%s(a%d >> %d)",
1283 prefix, i + 1, (offset & 3) * 8);
1284 }
1285 else {
1286 snprintf(buf, buf_size, "%s%sa%d",
1287 prefix, is_lea ? "&" : "", i + 1);
1288 }
1289 break;
1290
1291 default:
1292 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
1293 }
1294
1295 // common problem
1296 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type);
1297 if ((offset & 3) && tmp_lmod != OPLM_DWORD)
1298 ferr(po, "bp_arg arg/w offset %d and type '%s'\n",
1299 offset, g_func_pp->arg[i].type.name);
1300 }
1301 else
1302 {
1303 if (g_stack_fsz == 0)
1304 ferr(po, "stack var access without stackframe\n");
1305
1306 sf_ofs = g_stack_fsz + offset;
1307 lim = (ofs_reg[0] != 0) ? -4 : 0;
1308 if (offset > 0 || sf_ofs < lim)
1309 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
1310
1311 if (is_lea)
1312 prefix = "(u32)&";
1313 else
1314 prefix = cast;
1315
1316 switch (popr->lmod)
1317 {
1318 case OPLM_BYTE:
1319 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1320 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1321 break;
1322
1323 case OPLM_WORD:
1324 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1325 // known unaligned or possibly unaligned
1326 strcat(g_comment, " unaligned");
1327 if (prefix[0] == 0)
1328 prefix = "*(u16 *)&";
1329 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1330 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1331 break;
1332 }
1333 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
1334 break;
1335
1336 case OPLM_DWORD:
1337 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1338 // known unaligned or possibly unaligned
1339 strcat(g_comment, " unaligned");
1340 if (prefix[0] == 0)
1341 prefix = "*(u32 *)&";
1342 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1343 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1344 break;
1345 }
1346 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
1347 break;
1348
1349 default:
1350 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
1351 }
1352 }
1353}
1354
1355static void check_label_read_ref(struct parsed_op *po, const char *name)
1356{
1357 const struct parsed_proto *pp;
1358
1359 pp = proto_parse(g_fhdr, name);
1360 if (pp == NULL)
1361 ferr(po, "proto_parse failed for ref '%s'\n", name);
1362
1363 // currently we can take __cdecl and __stdcall
1364 if (pp->is_func && pp->argc_reg != 0)
1365 ferr(po, "reg-arg func reference?\n");
1366}
1367
1368static char *out_src_opr(char *buf, size_t buf_size,
1369 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
1370 int is_lea)
1371{
1372 char tmp1[256], tmp2[256];
1373 char expr[256];
1374 int ret;
1375
1376 if (cast == NULL)
1377 cast = "";
1378
1379 switch (popr->type) {
1380 case OPT_REG:
1381 if (is_lea)
1382 ferr(po, "lea from reg?\n");
1383
1384 switch (popr->lmod) {
1385 case OPLM_DWORD:
1386 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
1387 break;
1388 case OPLM_WORD:
1389 snprintf(buf, buf_size, "(u16)%s", opr_reg_p(po, popr));
1390 break;
1391 case OPLM_BYTE:
1392 if (popr->name[1] == 'h') // XXX..
1393 snprintf(buf, buf_size, "(u8)(%s >> 8)", opr_reg_p(po, popr));
1394 else
1395 snprintf(buf, buf_size, "(u8)%s", opr_reg_p(po, popr));
1396 break;
1397 default:
1398 ferr(po, "invalid src lmod: %d\n", popr->lmod);
1399 }
1400 break;
1401
1402 case OPT_REGMEM:
1403 if (parse_stack_el(popr->name, NULL)
1404 || (g_bp_frame && IS_START(popr->name, "ebp-")))
1405 {
1406 stack_frame_access(po, popr, buf, buf_size,
1407 popr->name, cast, 1, is_lea);
1408 break;
1409 }
1410
1411 strcpy(expr, popr->name);
1412 if (strchr(expr, '[')) {
1413 // special case: '[' can only be left for label[reg] form
1414 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
1415 if (ret != 2)
1416 ferr(po, "parse failure for '%s'\n", expr);
1417 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
1418 }
1419
1420 // XXX: do we need more parsing?
1421 if (is_lea) {
1422 snprintf(buf, buf_size, "%s", expr);
1423 break;
1424 }
1425
1426 if (cast[0] == 0)
1427 cast = lmod_cast_u_ptr(po, popr->lmod);
1428 snprintf(buf, buf_size, "%s(%s)", cast, expr);
1429 break;
1430
1431 case OPT_LABEL:
1432 check_label_read_ref(po, popr->name);
1433 if (cast[0] == 0 && popr->is_ptr)
1434 cast = "(u32)";
1435
1436 if (is_lea)
1437 snprintf(buf, buf_size, "(u32)&%s", popr->name);
1438 else if (popr->size_lt)
1439 snprintf(buf, buf_size, "%s%s%s%s", cast,
1440 lmod_cast_u_ptr(po, popr->lmod),
1441 popr->is_array ? "" : "&",
1442 popr->name);
1443 else
1444 snprintf(buf, buf_size, "%s%s%s", cast, popr->name,
1445 popr->is_array ? "[0]" : "");
1446 break;
1447
1448 case OPT_OFFSET:
1449 check_label_read_ref(po, popr->name);
1450 if (cast[0] == 0)
1451 cast = "(u32)";
1452 if (is_lea)
1453 ferr(po, "lea an offset?\n");
1454 snprintf(buf, buf_size, "%s&%s", cast, popr->name);
1455 break;
1456
1457 case OPT_CONST:
1458 if (is_lea)
1459 ferr(po, "lea from const?\n");
1460
1461 snprintf(buf, buf_size, "%s", cast);
1462 ret = strlen(buf);
1463 printf_number(buf + ret, buf_size - ret, popr->val);
1464 break;
1465
1466 default:
1467 ferr(po, "invalid src type: %d\n", popr->type);
1468 }
1469
1470 return buf;
1471}
1472
1473// note: may set is_ptr (we find that out late for ebp frame..)
1474static char *out_dst_opr(char *buf, size_t buf_size,
1475 struct parsed_op *po, struct parsed_opr *popr)
1476{
1477 switch (popr->type) {
1478 case OPT_REG:
1479 switch (popr->lmod) {
1480 case OPLM_DWORD:
1481 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
1482 break;
1483 case OPLM_WORD:
1484 // ugh..
1485 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
1486 break;
1487 case OPLM_BYTE:
1488 // ugh..
1489 if (popr->name[1] == 'h') // XXX..
1490 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
1491 else
1492 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
1493 break;
1494 default:
1495 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
1496 }
1497 break;
1498
1499 case OPT_REGMEM:
1500 if (parse_stack_el(popr->name, NULL)
1501 || (g_bp_frame && IS_START(popr->name, "ebp-")))
1502 {
1503 stack_frame_access(po, popr, buf, buf_size,
1504 popr->name, "", 0, 0);
1505 break;
1506 }
1507
1508 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1509
1510 case OPT_LABEL:
1511 if (popr->size_mismatch)
1512 snprintf(buf, buf_size, "%s%s%s",
1513 lmod_cast_u_ptr(po, popr->lmod),
1514 popr->is_array ? "" : "&", popr->name);
1515 else
1516 snprintf(buf, buf_size, "%s%s", popr->name,
1517 popr->is_array ? "[0]" : "");
1518 break;
1519
1520 default:
1521 ferr(po, "invalid dst type: %d\n", popr->type);
1522 }
1523
1524 return buf;
1525}
1526
1527static char *out_src_opr_u32(char *buf, size_t buf_size,
1528 struct parsed_op *po, struct parsed_opr *popr)
1529{
1530 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1531}
1532
1533static enum parsed_flag_op split_cond(struct parsed_op *po,
1534 enum op_op op, int *is_inv)
1535{
1536 *is_inv = 0;
1537
1538 switch (op) {
1539 case OP_JO:
1540 return PFO_O;
1541 case OP_JC:
1542 return PFO_C;
1543 case OP_JZ:
1544 return PFO_Z;
1545 case OP_JBE:
1546 return PFO_BE;
1547 case OP_JS:
1548 return PFO_S;
1549 case OP_JP:
1550 return PFO_P;
1551 case OP_JL:
1552 return PFO_L;
1553 case OP_JLE:
1554 return PFO_LE;
1555
1556 case OP_JNO:
1557 *is_inv = 1;
1558 return PFO_O;
1559 case OP_JNC:
1560 *is_inv = 1;
1561 return PFO_C;
1562 case OP_JNZ:
1563 *is_inv = 1;
1564 return PFO_Z;
1565 case OP_JA:
1566 *is_inv = 1;
1567 return PFO_BE;
1568 case OP_JNS:
1569 *is_inv = 1;
1570 return PFO_S;
1571 case OP_JNP:
1572 *is_inv = 1;
1573 return PFO_P;
1574 case OP_JGE:
1575 *is_inv = 1;
1576 return PFO_L;
1577 case OP_JG:
1578 *is_inv = 1;
1579 return PFO_LE;
1580
1581 case OP_ADC:
1582 case OP_SBB:
1583 return PFO_C;
1584
1585 default:
1586 ferr(po, "split_cond: bad op %d\n", op);
1587 return -1;
1588 }
1589}
1590
1591static void out_test_for_cc(char *buf, size_t buf_size,
1592 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
1593 enum opr_lenmod lmod, const char *expr)
1594{
1595 const char *cast, *scast;
1596
1597 cast = lmod_cast_u(po, lmod);
1598 scast = lmod_cast_s(po, lmod);
1599
1600 switch (pfo) {
1601 case PFO_Z:
1602 case PFO_BE: // CF=1||ZF=1; CF=0
1603 snprintf(buf, buf_size, "(%s%s %s 0)",
1604 cast, expr, is_inv ? "!=" : "==");
1605 break;
1606
1607 case PFO_S:
1608 case PFO_L: // SF!=OF; OF=0
1609 snprintf(buf, buf_size, "(%s%s %s 0)",
1610 scast, expr, is_inv ? ">=" : "<");
1611 break;
1612
1613 case PFO_LE: // ZF=1||SF!=OF; OF=0
1614 snprintf(buf, buf_size, "(%s%s %s 0)",
1615 scast, expr, is_inv ? ">" : "<=");
1616 break;
1617
1618 default:
1619 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1620 }
1621}
1622
1623static void out_cmp_for_cc(char *buf, size_t buf_size,
1624 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
1625 enum opr_lenmod lmod, const char *expr1, const char *expr2)
1626{
1627 const char *cast, *scast;
1628
1629 cast = lmod_cast_u(po, lmod);
1630 scast = lmod_cast_s(po, lmod);
1631
1632 switch (pfo) {
1633 case PFO_C:
1634 // note: must be unsigned compare
1635 snprintf(buf, buf_size, "(%s%s %s %s%s)",
1636 cast, expr1, is_inv ? ">=" : "<", cast, expr2);
1637 break;
1638
1639 case PFO_Z:
1640 snprintf(buf, buf_size, "(%s%s %s %s%s)",
1641 cast, expr1, is_inv ? "!=" : "==", cast, expr2);
1642 break;
1643
1644 case PFO_BE: // !a
1645 // note: must be unsigned compare
1646 snprintf(buf, buf_size, "(%s%s %s %s%s)",
1647 cast, expr1, is_inv ? ">" : "<=", cast, expr2);
1648 break;
1649
1650 // note: must be signed compare
1651 case PFO_S:
1652 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
1653 scast, expr1, expr2, is_inv ? ">=" : "<");
1654 break;
1655
1656 case PFO_L: // !ge
1657 snprintf(buf, buf_size, "(%s%s %s %s%s)",
1658 scast, expr1, is_inv ? ">=" : "<", scast, expr2);
1659 break;
1660
1661 case PFO_LE:
1662 snprintf(buf, buf_size, "(%s%s %s %s%s)",
1663 scast, expr1, is_inv ? ">" : "<=", scast, expr2);
1664 break;
1665
1666 default:
1667 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1668 }
1669}
1670
1671static void out_cmp_test(char *buf, size_t buf_size,
1672 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
1673{
1674 char buf1[256], buf2[256], buf3[256];
1675
1676 if (po->op == OP_TEST) {
1677 if (IS(opr_name(po, 0), opr_name(po, 1))) {
1678 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
1679 }
1680 else {
1681 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
1682 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
1683 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
1684 }
1685 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
1686 po->operand[0].lmod, buf3);
1687 }
1688 else if (po->op == OP_CMP) {
1689 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
1690 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[1]);
1691 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv,
1692 po->operand[0].lmod, buf2, buf3);
1693 }
1694 else
1695 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
1696}
1697
1698static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
1699 struct parsed_opr *popr2)
1700{
1701 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
1702 ferr(po, "missing lmod for both operands\n");
1703
1704 if (popr1->lmod == OPLM_UNSPEC)
1705 popr1->lmod = popr2->lmod;
1706 else if (popr2->lmod == OPLM_UNSPEC)
1707 popr2->lmod = popr1->lmod;
1708 else if (popr1->lmod != popr2->lmod)
1709 ferr(po, "conflicting lmods: %d vs %d\n", popr1->lmod, popr2->lmod);
1710}
1711
1712static const char *op_to_c(struct parsed_op *po)
1713{
1714 switch (po->op)
1715 {
1716 case OP_ADD:
1717 case OP_ADC:
1718 return "+";
1719 case OP_SUB:
1720 case OP_SBB:
1721 return "-";
1722 case OP_AND:
1723 return "&";
1724 case OP_OR:
1725 return "|";
1726 case OP_XOR:
1727 return "^";
1728 case OP_SHL:
1729 return "<<";
1730 case OP_SHR:
1731 return ">>";
1732 case OP_MUL:
1733 case OP_IMUL:
1734 return "*";
1735 default:
1736 ferr(po, "op_to_c was supplied with %d\n", po->op);
1737 }
1738}
1739
1740static void set_flag_no_dup(struct parsed_op *po, enum op_flags flag,
1741 enum op_flags flag_check)
1742{
1743 if (po->flags & flag)
1744 ferr(po, "flag %x already set\n", flag);
1745 if (po->flags & flag_check)
1746 ferr(po, "flag_check %x already set\n", flag_check);
1747
1748 po->flags |= flag;
1749}
1750
1751// last op in stream - unconditional branch or ret
1752#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
1753 || (ops[_i].flags & (OPF_JMP|OPF_CC)) == OPF_JMP)
1754
1755static int scan_for_pop(int i, int opcnt, const char *reg,
1756 int magic, int depth, int *maxdepth, int do_flags)
1757{
1758 struct parsed_op *po;
1759 int ret = 0;
1760 int j;
1761
1762 for (; i < opcnt; i++) {
1763 po = &ops[i];
1764 if (po->cc_scratch == magic)
1765 break; // already checked
1766 po->cc_scratch = magic;
1767
1768 if (po->flags & OPF_TAIL)
1769 return -1; // deadend
1770
1771 if ((po->flags & OPF_RMD)
1772 || (po->op == OP_PUSH && po->argnum != 0)) // arg push
1773 continue;
1774
1775 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
1776 if (po->btj != NULL) {
1777 // jumptable
1778 for (j = 0; j < po->btj->count - 1; j++) {
1779 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic,
1780 depth, maxdepth, do_flags);
1781 if (ret < 0)
1782 return ret; // dead end
1783 }
1784 // follow last jumptable entry
1785 i = po->btj->d[j].bt_i - 1;
1786 continue;
1787 }
1788
1789 if (po->bt_i < 0) {
1790 ferr(po, "dead branch\n");
1791 return -1;
1792 }
1793
1794 if (po->flags & OPF_CC) {
1795 ret |= scan_for_pop(po->bt_i, opcnt, reg, magic,
1796 depth, maxdepth, do_flags);
1797 if (ret < 0)
1798 return ret; // dead end
1799 }
1800 else {
1801 i = po->bt_i - 1;
1802 }
1803 continue;
1804 }
1805
1806 if ((po->op == OP_POP || po->op == OP_PUSH)
1807 && po->operand[0].type == OPT_REG
1808 && IS(po->operand[0].name, reg))
1809 {
1810 if (po->op == OP_PUSH) {
1811 depth++;
1812 if (depth > *maxdepth)
1813 *maxdepth = depth;
1814 if (do_flags)
1815 set_flag_no_dup(po, OPF_RSAVE, OPF_RMD);
1816 }
1817 else if (depth == 0) {
1818 if (do_flags)
1819 set_flag_no_dup(po, OPF_RMD, OPF_RSAVE);
1820 return 1;
1821 }
1822 else {
1823 depth--;
1824 if (depth < 0) // should not happen
1825 ferr(po, "fail with depth\n");
1826 if (do_flags)
1827 set_flag_no_dup(po, OPF_RSAVE, OPF_RMD);
1828 }
1829 }
1830 }
1831
1832 return ret;
1833}
1834
1835// scan for pop starting from 'ret' op (all paths)
1836static int scan_for_pop_ret(int i, int opcnt, const char *reg,
1837 int flag_set)
1838{
1839 int found = 0;
1840 int j;
1841
1842 for (; i < opcnt; i++) {
1843 if (!(ops[i].flags & OPF_TAIL))
1844 continue;
1845
1846 for (j = i - 1; j >= 0; j--) {
1847 if (ops[j].flags & OPF_RMD)
1848 continue;
1849 if (ops[j].flags & OPF_JMP)
1850 return -1;
1851
1852 if (ops[j].op == OP_POP && ops[j].operand[0].type == OPT_REG
1853 && IS(ops[j].operand[0].name, reg))
1854 {
1855 found = 1;
1856 ops[j].flags |= flag_set;
1857 break;
1858 }
1859
1860 if (g_labels[j][0] != 0)
1861 return -1;
1862 }
1863 }
1864
1865 return found ? 0 : -1;
1866}
1867
1868// is operand 'opr modified' by parsed_op 'po'?
1869static int is_opr_modified(const struct parsed_opr *opr,
1870 const struct parsed_op *po)
1871{
1872 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
1873 return 0;
1874
1875 if (opr->type == OPT_REG && po->operand[0].type == OPT_REG) {
1876 if (po->regmask_dst & (1 << opr->reg))
1877 return 1;
1878 else
1879 return 0;
1880 }
1881
1882 return IS(po->operand[0].name, opr->name);
1883}
1884
1885// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
1886static int is_any_opr_modified(const struct parsed_op *po_test,
1887 const struct parsed_op *po)
1888{
1889 int i;
1890
1891 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
1892 return 0;
1893
1894 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
1895 return 0;
1896
1897 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
1898 return 1;
1899
1900 // in reality, it can wreck any register, but in decompiled C
1901 // version it can only overwrite eax or edx:eax
1902 if (po->op == OP_CALL
1903 && ((po_test->regmask_src | po_test->regmask_dst)
1904 & ((1 << xAX)|(1 << xDX))))
1905 return 1;
1906
1907 for (i = 0; i < po_test->operand_cnt; i++)
1908 if (IS(po_test->operand[i].name, po->operand[0].name))
1909 return 1;
1910
1911 return 0;
1912}
1913
1914// scan for any po_test operand modification in range given
1915static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt)
1916{
1917 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
1918 return -1;
1919
1920 for (; i < opcnt; i++) {
1921 if (is_any_opr_modified(po_test, &ops[i]))
1922 return i;
1923 }
1924
1925 return -1;
1926}
1927
1928// scan for po_test operand[0] modification in range given
1929static int scan_for_mod_opr0(struct parsed_op *po_test,
1930 int i, int opcnt)
1931{
1932 for (; i < opcnt; i++) {
1933 if (is_opr_modified(&po_test->operand[0], &ops[i]))
1934 return i;
1935 }
1936
1937 return -1;
1938}
1939
1940static int scan_for_flag_set(int i, int *branched, int *setters,
1941 int *setter_cnt)
1942{
1943 int ret;
1944
1945 while (i >= 0) {
1946 if (g_labels[i][0] != 0) {
1947 *branched = 1;
1948 if (g_label_refs[i].next != NULL)
1949 return -1;
1950 if (i > 0 && LAST_OP(i - 1)) {
1951 i = g_label_refs[i].i;
1952 continue;
1953 }
1954 ret = scan_for_flag_set(g_label_refs[i].i, branched,
1955 setters, setter_cnt);
1956 if (ret < 0)
1957 return ret;
1958 }
1959 i--;
1960
1961 if (ops[i].flags & OPF_FLAGS) {
1962 setters[*setter_cnt] = i;
1963 (*setter_cnt)++;
1964 return 0;
1965 }
1966
1967 if ((ops[i].flags & OPF_JMP) && !(ops[i].flags & OPF_CC))
1968 return -1;
1969 }
1970
1971 return -1;
1972}
1973
1974// scan back for cdq, if anything modifies edx, fail
1975static int scan_for_cdq_edx(int i)
1976{
1977 for (; i >= 0; i--) {
1978 if (ops[i].op == OP_CDQ)
1979 return i;
1980
1981 if (ops[i].regmask_dst & (1 << xDX))
1982 return -1;
1983 if (g_labels[i][0] != 0)
1984 return -1;
1985 }
1986
1987 return -1;
1988}
1989
1990static int scan_for_reg_clear(int i, int reg)
1991{
1992 for (; i >= 0; i--) {
1993 if (ops[i].op == OP_XOR
1994 && ops[i].operand[0].lmod == OPLM_DWORD
1995 && ops[i].operand[0].reg == ops[i].operand[1].reg
1996 && ops[i].operand[0].reg == reg)
1997 return i;
1998
1999 if (ops[i].regmask_dst & (1 << reg))
2000 return -1;
2001 if (g_labels[i][0] != 0)
2002 return -1;
2003 }
2004
2005 return -1;
2006}
2007
2008// scan for positive, constant esp adjust
2009static int scan_for_esp_adjust(int i, int opcnt, int *adj)
2010{
2011 struct parsed_op *po;
2012 *adj = 0;
2013
2014 for (; i < opcnt; i++) {
2015 po = &ops[i];
2016
2017 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
2018 if (po->operand[1].type != OPT_CONST)
2019 ferr(&ops[i], "non-const esp adjust?\n");
2020 *adj += po->operand[1].val;
2021 if (*adj & 3)
2022 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
2023 return i;
2024 }
2025 else if (po->op == OP_PUSH)
2026 *adj -= lmod_bytes(po, po->operand[0].lmod);
2027 else if (po->op == OP_POP)
2028 *adj += lmod_bytes(po, po->operand[0].lmod);
2029 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
2030 if (po->op != OP_CALL)
2031 return -1;
2032 if (po->operand[0].type != OPT_LABEL)
2033 return -1;
2034 // TODO: should only allow combining __cdecl calls..
2035 }
2036
2037 if (g_labels[i][0] != 0)
2038 return -1;
2039 }
2040
2041 return -1;
2042}
2043
2044static int collect_call_args(struct parsed_op *po, int i,
2045 struct parsed_proto *pp, int *save_arg_vars, int arg,
2046 int need_op_saving, int may_reuse)
2047{
2048 struct parsed_proto *pp_tmp;
2049 struct label_ref *lr;
2050 int need_to_save_current;
2051 int ret = 0;
2052 int j;
2053
2054 if (i < 0)
2055 ferr(po, "no refs for '%s'?\n", g_labels[i]);
2056
2057 for (; arg < pp->argc; arg++)
2058 if (pp->arg[arg].reg == NULL)
2059 break;
2060
2061 for (j = i; j >= 0 && arg < pp->argc; )
2062 {
2063 if (g_labels[j][0] != 0) {
2064 lr = &g_label_refs[j];
2065 if (lr->next != NULL)
2066 need_op_saving = 1;
2067 for (; lr->next; lr = lr->next) {
2068 if ((ops[lr->i].flags & (OPF_JMP|OPF_CC)) != OPF_JMP)
2069 may_reuse = 1;
2070 ret |= collect_call_args(po, lr->i, pp, save_arg_vars,
2071 arg, need_op_saving, may_reuse);
2072 }
2073
2074 if ((ops[lr->i].flags & (OPF_JMP|OPF_CC)) != OPF_JMP)
2075 may_reuse = 1;
2076 if (j > 0 && LAST_OP(j - 1)) {
2077 // follow last branch in reverse
2078 j = lr->i;
2079 continue;
2080 }
2081 need_op_saving = 1;
2082 ret |= collect_call_args(po, lr->i, pp, save_arg_vars,
2083 arg, need_op_saving, may_reuse);
2084 }
2085 j--;
2086
2087 if (ops[j].op == OP_CALL)
2088 {
2089 pp_tmp = ops[j].datap;
2090 if (pp_tmp == NULL)
2091 ferr(po, "arg collect hit unparsed call\n");
2092 if (may_reuse && pp_tmp->argc_stack > 0)
2093 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
2094 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
2095 }
2096 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP) {
2097 ferr(po, "arg collect %d/%d hit esp adjust\n",
2098 arg, pp->argc);
2099 }
2100 else if (ops[j].op == OP_POP) {
2101 ferr(po, "arg collect %d/%d hit pop\n", arg, pp->argc);
2102 }
2103 else if ((ops[j].flags & (OPF_JMP|OPF_CC)) == (OPF_JMP|OPF_CC))
2104 {
2105 may_reuse = 1;
2106 }
2107 else if (ops[j].op == OP_PUSH && !(ops[j].flags & OPF_FARG))
2108 {
2109 pp->arg[arg].datap = &ops[j];
2110 need_to_save_current = 0;
2111 if (!need_op_saving) {
2112 ret = scan_for_mod(&ops[j], j + 1, i);
2113 need_to_save_current = (ret >= 0);
2114 }
2115 if (need_op_saving || need_to_save_current) {
2116 // mark this push as one that needs operand saving
2117 ops[j].flags &= ~OPF_RMD;
2118 if (ops[j].argnum == 0) {
2119 ops[j].argnum = arg + 1;
2120 *save_arg_vars |= 1 << arg;
2121 }
2122 else if (ops[j].argnum < arg + 1)
2123 ferr(&ops[j], "argnum conflict (%d<%d) for '%s'\n",
2124 ops[j].argnum, arg + 1, pp->name);
2125 }
2126 else if (ops[j].argnum == 0)
2127 ops[j].flags |= OPF_RMD;
2128
2129 // some PUSHes are reused by different calls on other branches,
2130 // but that can't happen if we didn't branch, so they
2131 // can be removed from future searches (handles nested calls)
2132 if (!may_reuse)
2133 ops[j].flags |= OPF_FARG;
2134
2135 // next arg
2136 for (arg++; arg < pp->argc; arg++)
2137 if (pp->arg[arg].reg == NULL)
2138 break;
2139 }
2140 }
2141
2142 if (arg < pp->argc) {
2143 ferr(po, "arg collect failed for '%s': %d/%d\n",
2144 pp->name, arg, pp->argc);
2145 ret = -1;
2146 }
2147 return ret;
2148}
2149
2150static void add_label_ref(struct label_ref *lr, int op_i)
2151{
2152 struct label_ref *lr_new;
2153
2154 if (lr->i == -1) {
2155 lr->i = op_i;
2156 return;
2157 }
2158
2159 lr_new = calloc(1, sizeof(*lr_new));
2160 lr_new->i = op_i;
2161 lr_new->next = lr->next;
2162 lr->next = lr_new;
2163}
2164
2165static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
2166{
2167 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
2168 struct parsed_opr *last_arith_dst = NULL;
2169 char buf1[256], buf2[256], buf3[256], cast[64];
2170 const struct parsed_proto *pp_c;
2171 struct parsed_proto *pp, *pp_tmp;
2172 struct parsed_data *pd;
2173 const char *tmpname;
2174 enum parsed_flag_op pfo;
2175 int save_arg_vars = 0;
2176 int cmp_result_vars = 0;
2177 int need_mul_var = 0;
2178 int had_decl = 0;
2179 int label_pending = 0;
2180 int regmask_save = 0;
2181 int regmask_arg = 0;
2182 int regmask = 0;
2183 int pfomask = 0;
2184 int found = 0;
2185 int depth = 0;
2186 int no_output;
2187 int i, j, l;
2188 int dummy;
2189 int arg;
2190 int reg;
2191 int ret;
2192
2193 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
2194
2195 g_func_pp = proto_parse(fhdr, funcn);
2196 if (g_func_pp == NULL)
2197 ferr(ops, "proto_parse failed for '%s'\n", funcn);
2198
2199 fprintf(fout, "%s ", g_func_pp->ret_type.name);
2200 if (g_func_pp->is_stdcall && g_func_pp->argc_reg == 0)
2201 fprintf(fout, "__stdcall ");
2202 if (g_ida_func_attr & IDAFA_NORETURN)
2203 fprintf(fout, "noreturn ");
2204 fprintf(fout, "%s(", funcn);
2205
2206 for (i = 0; i < g_func_pp->argc; i++) {
2207 if (i > 0)
2208 fprintf(fout, ", ");
2209 if (g_func_pp->arg[i].fptr != NULL) {
2210 // func pointer..
2211 pp = g_func_pp->arg[i].fptr;
2212 fprintf(fout, "%s (", pp->ret_type.name);
2213 if (pp->is_stdcall && pp->argc_reg == 0)
2214 fprintf(fout, "__stdcall ");
2215 fprintf(fout, "*a%d)(", i + 1);
2216 for (j = 0; j < pp->argc; j++) {
2217 if (j > 0)
2218 fprintf(fout, ", ");
2219 if (pp->arg[j].fptr)
2220 ferr(ops, "nested fptr\n");
2221 fprintf(fout, "%s", pp->arg[j].type.name);
2222 }
2223 fprintf(fout, ")");
2224 }
2225 else {
2226 fprintf(fout, "%s a%d", g_func_pp->arg[i].type.name, i + 1);
2227 }
2228 }
2229 if (g_func_pp->is_vararg) {
2230 if (i > 0)
2231 fprintf(fout, ", ");
2232 fprintf(fout, "...");
2233 }
2234
2235 fprintf(fout, ")\n{\n");
2236
2237 // pass1:
2238 // - handle ebp/esp frame, remove ops related to it
2239 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
2240 && ops[1].op == OP_MOV
2241 && IS(opr_name(&ops[1], 0), "ebp")
2242 && IS(opr_name(&ops[1], 1), "esp"))
2243 {
2244 int ecx_push = 0;
2245
2246 g_bp_frame = 1;
2247 ops[0].flags |= OPF_RMD;
2248 ops[1].flags |= OPF_RMD;
2249 i = 2;
2250
2251 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
2252 g_stack_fsz = opr_const(&ops[2], 1);
2253 ops[2].flags |= OPF_RMD;
2254 i++;
2255 }
2256 else {
2257 // another way msvc builds stack frame..
2258 i = 2;
2259 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
2260 g_stack_fsz += 4;
2261 ops[i].flags |= OPF_RMD;
2262 ecx_push++;
2263 i++;
2264 }
2265 // and another way..
2266 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
2267 && ops[i].operand[1].type == OPT_CONST
2268 && ops[i + 1].op == OP_CALL
2269 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
2270 {
2271 g_stack_fsz += ops[i].operand[1].val;
2272 ops[i].flags |= OPF_RMD;
2273 i++;
2274 ops[i].flags |= OPF_RMD;
2275 i++;
2276 }
2277 }
2278
2279 found = 0;
2280 do {
2281 for (; i < opcnt; i++)
2282 if (ops[i].op == OP_RET)
2283 break;
2284 if (i == opcnt && (ops[i - 1].flags & OPF_JMP) && found)
2285 break;
2286
2287 if (ops[i - 1].op == OP_POP && IS(opr_name(&ops[i - 1], 0), "ebp"))
2288 ops[i - 1].flags |= OPF_RMD;
2289 else if (!(g_ida_func_attr & IDAFA_NORETURN))
2290 ferr(&ops[i - 1], "'pop ebp' expected\n");
2291
2292 if (g_stack_fsz != 0) {
2293 if (ops[i - 2].op == OP_MOV
2294 && IS(opr_name(&ops[i - 2], 0), "esp")
2295 && IS(opr_name(&ops[i - 2], 1), "ebp"))
2296 {
2297 ops[i - 2].flags |= OPF_RMD;
2298 }
2299 else if (!(g_ida_func_attr & IDAFA_NORETURN))
2300 ferr(&ops[i - 2], "esp restore expected\n");
2301
2302 if (ecx_push && ops[i - 3].op == OP_POP
2303 && IS(opr_name(&ops[i - 3], 0), "ecx"))
2304 {
2305 ferr(&ops[i - 3], "unexpected ecx pop\n");
2306 }
2307 }
2308
2309 found = 1;
2310 i++;
2311 } while (i < opcnt);
2312 }
2313 else {
2314 for (i = 0; i < opcnt; i++) {
2315 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
2316 break;
2317 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
2318 && ops[i].operand[1].type == OPT_CONST)
2319 {
2320 g_sp_frame = 1;
2321 break;
2322 }
2323 }
2324
2325 if (g_sp_frame)
2326 {
2327 g_stack_fsz = ops[i].operand[1].val;
2328 ops[i].flags |= OPF_RMD;
2329
2330 i++;
2331 do {
2332 for (; i < opcnt; i++)
2333 if (ops[i].op == OP_RET)
2334 break;
2335 if (ops[i - 1].op != OP_ADD
2336 || !IS(opr_name(&ops[i - 1], 0), "esp")
2337 || ops[i - 1].operand[1].type != OPT_CONST
2338 || ops[i - 1].operand[1].val != g_stack_fsz)
2339 ferr(&ops[i - 1], "'add esp' expected\n");
2340 ops[i - 1].flags |= OPF_RMD;
2341
2342 i++;
2343 } while (i < opcnt);
2344 }
2345 }
2346
2347 // pass2:
2348 // - resolve all branches
2349 for (i = 0; i < opcnt; i++)
2350 {
2351 po = &ops[i];
2352 po->bt_i = -1;
2353 po->btj = NULL;
2354
2355 if ((po->flags & OPF_RMD) || !(po->flags & OPF_JMP)
2356 || po->op == OP_CALL || po->op == OP_RET)
2357 continue;
2358
2359 if (po->operand[0].type == OPT_REGMEM) {
2360 char *p = strchr(po->operand[0].name, '[');
2361 if (p == NULL)
2362 ferr(po, "unhandled indirect branch\n");
2363 ret = p - po->operand[0].name;
2364 strncpy(buf1, po->operand[0].name, ret);
2365 buf1[ret] = 0;
2366
2367 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
2368 if (IS(g_func_pd[j].label, buf1)) {
2369 pd = &g_func_pd[j];
2370 break;
2371 }
2372 }
2373 if (pd == NULL)
2374 //ferr(po, "label '%s' not parsed?\n", buf1);
2375 goto tailcall;
2376 if (pd->type != OPT_OFFSET)
2377 ferr(po, "label '%s' with non-offset data?\n", buf1);
2378
2379 // find all labels, link
2380 for (j = 0; j < pd->count; j++) {
2381 for (l = 0; l < opcnt; l++) {
2382 if (g_labels[l][0] && IS(g_labels[l], pd->d[j].u.label)) {
2383 add_label_ref(&g_label_refs[l], i);
2384 pd->d[j].bt_i = l;
2385 break;
2386 }
2387 }
2388 }
2389
2390 po->btj = pd;
2391 continue;
2392 }
2393
2394 for (l = 0; l < opcnt; l++) {
2395 if (g_labels[l][0] && IS(po->operand[0].name, g_labels[l])) {
2396 add_label_ref(&g_label_refs[l], i);
2397 po->bt_i = l;
2398 break;
2399 }
2400 }
2401
2402 if (po->bt_i != -1)
2403 continue;
2404
2405 if (po->operand[0].type == OPT_LABEL)
2406 // assume tail call
2407 goto tailcall;
2408
2409 ferr(po, "unhandled branch\n");
2410
2411tailcall:
2412 po->op = OP_CALL;
2413 po->flags |= OPF_TAIL;
2414 }
2415
2416 // pass3:
2417 // - process calls
2418 for (i = 0; i < opcnt; i++)
2419 {
2420 po = &ops[i];
2421 if (po->flags & OPF_RMD)
2422 continue;
2423
2424 if (po->op == OP_CALL)
2425 {
2426 pp = calloc(1, sizeof(*pp));
2427 my_assert_not(pp, NULL);
2428 tmpname = opr_name(po, 0);
2429 if (po->operand[0].type != OPT_LABEL)
2430 {
2431 ret = scan_for_esp_adjust(i + 1, opcnt, &j);
2432 if (ret < 0)
2433 ferr(po, "non-__cdecl indirect call unhandled yet\n");
2434 j /= 4;
2435 if (j > ARRAY_SIZE(pp->arg))
2436 ferr(po, "esp adjust too large?\n");
2437 pp->ret_type.name = strdup("int");
2438 pp->argc = pp->argc_stack = j;
2439 for (arg = 0; arg < pp->argc; arg++)
2440 pp->arg[arg].type.name = strdup("int");
2441 }
2442 else {
2443 pp_c = proto_parse(fhdr, tmpname);
2444 if (pp_c == NULL)
2445 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
2446 pp = proto_clone(pp_c);
2447 }
2448
2449 // look for and make use of esp adjust
2450 ret = -1;
2451 if (!pp->is_stdcall && pp->argc_stack > 0)
2452 ret = scan_for_esp_adjust(i + 1, opcnt, &j);
2453 if (ret >= 0) {
2454 if (pp->is_vararg) {
2455 if (j / 4 < pp->argc_stack)
2456 ferr(po, "esp adjust is too small: %x < %x\n",
2457 j, pp->argc_stack * 4);
2458 // modify pp to make it have varargs as normal args
2459 arg = pp->argc;
2460 pp->argc += j / 4 - pp->argc_stack;
2461 for (; arg < pp->argc; arg++) {
2462 pp->arg[arg].type.name = strdup("int");
2463 pp->argc_stack++;
2464 }
2465 if (pp->argc > ARRAY_SIZE(pp->arg))
2466 ferr(po, "too many args for '%s'\n", tmpname);
2467 }
2468 if (pp->argc_stack != j / 4)
2469 ferr(po, "stack tracking failed for '%s': %x %x\n",
2470 tmpname, pp->argc_stack * 4, j);
2471
2472 ops[ret].flags |= OPF_RMD;
2473 // a bit of a hack, but deals with use of
2474 // single adj for multiple calls
2475 ops[ret].operand[1].val -= j;
2476 }
2477 else if (pp->is_vararg)
2478 ferr(po, "missing esp_adjust for vararg func '%s'\n",
2479 pp->name);
2480
2481 // can't call functions with non-__cdecl callbacks yet
2482 for (arg = 0; arg < pp->argc; arg++) {
2483 if (pp->arg[arg].fptr != NULL) {
2484 pp_tmp = pp->arg[arg].fptr;
2485 if (pp_tmp->is_stdcall || pp_tmp->argc != pp_tmp->argc_stack)
2486 ferr(po, "'%s' has a non-__cdecl callback\n", tmpname);
2487 }
2488 }
2489
2490 collect_call_args(po, i, pp, &save_arg_vars, 0, 0, 0);
2491
2492 if (strstr(pp->ret_type.name, "int64"))
2493 need_mul_var = 1;
2494 po->datap = pp;
2495 }
2496 }
2497
2498 // pass4:
2499 // - find POPs for PUSHes, rm both
2500 // - scan for all used registers
2501 // - find flag set ops for their users
2502 // - declare indirect functions
2503 for (i = 0; i < opcnt; i++) {
2504 po = &ops[i];
2505 if (po->flags & OPF_RMD)
2506 continue;
2507
2508 if (po->op == OP_PUSH
2509 && po->argnum == 0 && !(po->flags & OPF_RSAVE)
2510 && po->operand[0].type == OPT_REG)
2511 {
2512 reg = po->operand[0].reg;
2513 if (reg < 0)
2514 ferr(po, "reg not set for push?\n");
2515
2516 depth = 0;
2517 ret = scan_for_pop(i + 1, opcnt,
2518 po->operand[0].name, i + opcnt, 0, &depth, 0);
2519 if (ret == 1) {
2520 if (depth > 1)
2521 ferr(po, "too much depth: %d\n", depth);
2522 if (depth > 0)
2523 regmask_save |= 1 << reg;
2524
2525 po->flags |= OPF_RMD;
2526 scan_for_pop(i + 1, opcnt, po->operand[0].name,
2527 i + opcnt * 2, 0, &depth, 1);
2528 continue;
2529 }
2530 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
2531 if (ret == 0) {
2532 arg = OPF_RMD;
2533 if (regmask & (1 << reg)) {
2534 if (regmask_save & (1 << reg))
2535 ferr(po, "%s already saved?\n", po->operand[0].name);
2536 arg = OPF_RSAVE;
2537 }
2538 po->flags |= arg;
2539 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg);
2540 continue;
2541 }
2542 }
2543
2544 regmask |= po->regmask_src | po->regmask_dst;
2545
2546 if (po->flags & OPF_CC)
2547 {
2548 int setters[16], cnt = 0, branched = 0;
2549
2550 ret = scan_for_flag_set(i, &branched, setters, &cnt);
2551 if (ret < 0 || cnt <= 0)
2552 ferr(po, "unable to trace flag setter(s)\n");
2553 if (cnt > ARRAY_SIZE(setters))
2554 ferr(po, "too many flag setters\n");
2555
2556 pfo = split_cond(po, po->op, &dummy);
2557 for (j = 0; j < cnt; j++)
2558 {
2559 tmp_op = &ops[setters[j]]; // flag setter
2560 pfomask = 0;
2561
2562 // to get nicer code, we try to delay test and cmp;
2563 // if we can't because of operand modification, or if we
2564 // have math op, or branch, make it calculate flags explicitly
2565 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP) {
2566 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i) >= 0)
2567 pfomask = 1 << pfo;
2568 }
2569 else if (tmp_op->op == OP_CMPS) {
2570 pfomask = 1 << PFO_Z;
2571 }
2572 else {
2573 if ((pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P)
2574 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
2575 pfomask = 1 << pfo;
2576 }
2577 if (pfomask) {
2578 tmp_op->pfomask |= pfomask;
2579 cmp_result_vars |= pfomask;
2580 // note: may overwrite, currently not a problem
2581 po->datap = tmp_op;
2582 }
2583 }
2584
2585 if (po->op == OP_ADC || po->op == OP_SBB)
2586 cmp_result_vars |= 1 << PFO_C;
2587 }
2588 else if (po->op == OP_MUL
2589 || (po->op == OP_IMUL && po->operand_cnt == 1))
2590 {
2591 need_mul_var = 1;
2592 }
2593 else if (po->op == OP_CALL && po->operand[0].type != OPT_LABEL) {
2594 pp = po->datap;
2595 my_assert_not(pp, NULL);
2596 fprintf(fout, " %s (*icall%d)(", pp->ret_type.name, i);
2597 for (j = 0; j < pp->argc; j++) {
2598 if (j > 0)
2599 fprintf(fout, ", ");
2600 fprintf(fout, "%s a%d", pp->arg[j].type.name, j + 1);
2601 }
2602 fprintf(fout, ");\n");
2603 }
2604 }
2605
2606 // output LUTs/jumptables
2607 for (i = 0; i < g_func_pd_cnt; i++) {
2608 pd = &g_func_pd[i];
2609 fprintf(fout, " static const ");
2610 if (pd->type == OPT_OFFSET) {
2611 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
2612
2613 for (j = 0; j < pd->count; j++) {
2614 if (j > 0)
2615 fprintf(fout, ", ");
2616 fprintf(fout, "&&%s", pd->d[j].u.label);
2617 }
2618 }
2619 else {
2620 fprintf(fout, "%s %s[] =\n { ",
2621 lmod_type_u(ops, pd->lmod), pd->label);
2622
2623 for (j = 0; j < pd->count; j++) {
2624 if (j > 0)
2625 fprintf(fout, ", ");
2626 fprintf(fout, "%u", pd->d[j].u.val);
2627 }
2628 }
2629 fprintf(fout, " };\n");
2630 }
2631
2632 // declare stack frame, va_arg
2633 if (g_stack_fsz)
2634 fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
2635 (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz);
2636
2637 if (g_func_pp->is_vararg)
2638 fprintf(fout, " va_list ap;\n");
2639
2640 // declare arg-registers
2641 for (i = 0; i < g_func_pp->argc; i++) {
2642 if (g_func_pp->arg[i].reg != NULL) {
2643 reg = char_array_i(regs_r32,
2644 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
2645 if (reg < 0)
2646 ferr(ops, "arg '%s' is not a reg?\n", g_func_pp->arg[i].reg);
2647
2648 regmask_arg |= 1 << reg;
2649 fprintf(fout, " u32 %s = (u32)a%d;\n",
2650 g_func_pp->arg[i].reg, i + 1);
2651 had_decl = 1;
2652 }
2653 }
2654
2655 // declare other regs - special case for eax
2656 if (!((regmask | regmask_arg) & 1)
2657 && !IS(g_func_pp->ret_type.name, "void"))
2658 {
2659 fprintf(fout, " u32 eax = 0;\n");
2660 had_decl = 1;
2661 }
2662
2663 regmask &= ~regmask_arg;
2664 regmask &= ~(1 << xSP);
2665 if (g_bp_frame)
2666 regmask &= ~(1 << xBP);
2667 if (regmask) {
2668 for (reg = 0; reg < 8; reg++) {
2669 if (regmask & (1 << reg)) {
2670 fprintf(fout, " u32 %s;\n", regs_r32[reg]);
2671 had_decl = 1;
2672 }
2673 }
2674 }
2675
2676 if (regmask_save) {
2677 for (reg = 0; reg < 8; reg++) {
2678 if (regmask_save & (1 << reg)) {
2679 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
2680 had_decl = 1;
2681 }
2682 }
2683 }
2684
2685 if (save_arg_vars) {
2686 for (reg = 0; reg < 32; reg++) {
2687 if (save_arg_vars & (1 << reg)) {
2688 fprintf(fout, " u32 s_a%d;\n", reg + 1);
2689 had_decl = 1;
2690 }
2691 }
2692 }
2693
2694 if (cmp_result_vars) {
2695 for (i = 0; i < 8; i++) {
2696 if (cmp_result_vars & (1 << i)) {
2697 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
2698 had_decl = 1;
2699 }
2700 }
2701 }
2702
2703 if (need_mul_var) {
2704 fprintf(fout, " u64 mul_tmp;\n");
2705 had_decl = 1;
2706 }
2707
2708 if (had_decl)
2709 fprintf(fout, "\n");
2710
2711 if (g_func_pp->is_vararg) {
2712 if (g_func_pp->argc_stack == 0)
2713 ferr(ops, "vararg func without stack args?\n");
2714 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
2715 }
2716
2717 // output ops
2718 for (i = 0; i < opcnt; i++)
2719 {
2720 if (g_labels[i][0] != 0 && g_label_refs[i].i != -1) {
2721 fprintf(fout, "\n%s:\n", g_labels[i]);
2722 label_pending = 1;
2723
2724 delayed_flag_op = NULL;
2725 last_arith_dst = NULL;
2726 }
2727
2728 po = &ops[i];
2729 if (po->flags & OPF_RMD)
2730 continue;
2731
2732 no_output = 0;
2733
2734 #define assert_operand_cnt(n_) \
2735 if (po->operand_cnt != n_) \
2736 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
2737
2738 // conditional/flag using op?
2739 if (po->flags & OPF_CC)
2740 {
2741 int is_delayed = 0;
2742 int is_inv = 0;
2743
2744 pfo = split_cond(po, po->op, &is_inv);
2745
2746 // we go through all this trouble to avoid using parsed_flag_op,
2747 // which makes generated code much nicer
2748 if (delayed_flag_op != NULL)
2749 {
2750 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op, pfo, is_inv);
2751 is_delayed = 1;
2752 }
2753 else if (last_arith_dst != NULL
2754 && (pfo == PFO_Z || pfo == PFO_S || pfo == PFO_P))
2755 {
2756 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
2757 out_test_for_cc(buf1, sizeof(buf1), po, pfo, is_inv,
2758 last_arith_dst->lmod, buf3);
2759 is_delayed = 1;
2760 }
2761 else if (po->datap != NULL) {
2762 // use preprocessed flag calc results
2763 tmp_op = po->datap;
2764 if (!tmp_op || !(tmp_op->pfomask & (1 << pfo)))
2765 ferr(po, "not prepared for pfo %d\n", pfo);
2766
2767 // note: is_inv was not yet applied
2768 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
2769 is_inv ? "!" : "", parsed_flag_op_names[pfo]);
2770 }
2771 else {
2772 ferr(po, "all methods of finding comparison failed\n");
2773 }
2774
2775 if (po->flags & OPF_JMP) {
2776 fprintf(fout, " if %s\n", buf1);
2777 }
2778 else if (po->op == OP_ADC || po->op == OP_SBB) {
2779 if (is_delayed)
2780 fprintf(fout, " cond_%s = %s;\n",
2781 parsed_flag_op_names[pfo], buf1);
2782 }
2783 else if (po->flags & OPF_DATA) { // SETcc
2784 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
2785 fprintf(fout, " %s = %s;", buf2, buf1);
2786 }
2787 else {
2788 ferr(po, "unhandled conditional op\n");
2789 }
2790 }
2791
2792 pfomask = po->pfomask;
2793
2794 switch (po->op)
2795 {
2796 case OP_MOV:
2797 assert_operand_cnt(2);
2798 propagate_lmod(po, &po->operand[0], &po->operand[1]);
2799 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
2800 fprintf(fout, " %s = %s;", buf1,
2801 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
2802 po->operand[0].is_ptr ? "(void *)" : "", 0));
2803 break;
2804
2805 case OP_LEA:
2806 assert_operand_cnt(2);
2807 po->operand[1].lmod = OPLM_DWORD; // always
2808 fprintf(fout, " %s = %s;",
2809 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2810 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
2811 NULL, 1));
2812 break;
2813
2814 case OP_MOVZX:
2815 assert_operand_cnt(2);
2816 fprintf(fout, " %s = %s;",
2817 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2818 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2819 break;
2820
2821 case OP_MOVSX:
2822 assert_operand_cnt(2);
2823 switch (po->operand[1].lmod) {
2824 case OPLM_BYTE:
2825 strcpy(buf3, "(s8)");
2826 break;
2827 case OPLM_WORD:
2828 strcpy(buf3, "(s16)");
2829 break;
2830 default:
2831 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
2832 }
2833 fprintf(fout, " %s = %s%s;",
2834 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2835 buf3,
2836 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2837 break;
2838
2839 case OP_NOT:
2840 assert_operand_cnt(1);
2841 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
2842 fprintf(fout, " %s = ~%s;", buf1, buf1);
2843 break;
2844
2845 case OP_CDQ:
2846 assert_operand_cnt(2);
2847 fprintf(fout, " %s = (s32)%s >> 31;",
2848 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2849 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2850 strcpy(g_comment, "cdq");
2851 break;
2852
2853 case OP_STOS:
2854 // assumes DF=0
2855 assert_operand_cnt(3);
2856 if (po->flags & OPF_REP) {
2857 fprintf(fout, " for (; ecx != 0; ecx--, edi += %d)\n",
2858 lmod_bytes(po, po->operand[0].lmod));
2859 fprintf(fout, " %sedi = eax;",
2860 lmod_cast_u_ptr(po, po->operand[0].lmod));
2861 strcpy(g_comment, "rep stos");
2862 }
2863 else {
2864 fprintf(fout, " %sedi = eax; edi += %d;",
2865 lmod_cast_u_ptr(po, po->operand[0].lmod),
2866 lmod_bytes(po, po->operand[0].lmod));
2867 strcpy(g_comment, "stos");
2868 }
2869 break;
2870
2871 case OP_MOVS:
2872 // assumes DF=0
2873 assert_operand_cnt(3);
2874 j = lmod_bytes(po, po->operand[0].lmod);
2875 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
2876 if (po->flags & OPF_REP) {
2877 fprintf(fout,
2878 " for (; ecx != 0; ecx--, edi += %d, esi += %d)\n",
2879 j, j);
2880 fprintf(fout,
2881 " %sedi = %sesi;", buf1, buf1);
2882 strcpy(g_comment, "rep movs");
2883 }
2884 else {
2885 fprintf(fout, " %sedi = %sesi; edi += %d; esi += %d;",
2886 buf1, buf1, j, j);
2887 strcpy(g_comment, "movs");
2888 }
2889 break;
2890
2891 case OP_CMPS:
2892 // assumes DF=0
2893 // repe ~ repeat while ZF=1
2894 assert_operand_cnt(3);
2895 j = lmod_bytes(po, po->operand[0].lmod);
2896 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
2897 if (po->flags & OPF_REP) {
2898 fprintf(fout,
2899 " for (; ecx != 0; ecx--, edi += %d, esi += %d)\n",
2900 j, j);
2901 fprintf(fout,
2902 " if ((cond_z = (%sedi == %sesi)) %s 0)\n",
2903 buf1, buf1, (po->flags & OPF_REPZ) ? "==" : "!=");
2904 fprintf(fout,
2905 " break;");
2906 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
2907 (po->flags & OPF_REPZ) ? "e" : "ne");
2908 }
2909 else {
2910 fprintf(fout,
2911 " cond_z = (%sedi = %sesi); edi += %d; esi += %d;",
2912 buf1, buf1, j, j);
2913 strcpy(g_comment, "cmps");
2914 }
2915 pfomask &= ~(1 << PFO_Z);
2916 last_arith_dst = NULL;
2917 delayed_flag_op = NULL;
2918 break;
2919
2920 // arithmetic w/flags
2921 case OP_ADD:
2922 case OP_SUB:
2923 case OP_AND:
2924 case OP_OR:
2925 propagate_lmod(po, &po->operand[0], &po->operand[1]);
2926 // fallthrough
2927 case OP_SHL:
2928 case OP_SHR:
2929 dualop_arith:
2930 assert_operand_cnt(2);
2931 fprintf(fout, " %s %s= %s;",
2932 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2933 op_to_c(po),
2934 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2935 if (pfomask & (1 << PFO_Z)) {
2936 fprintf(fout, "\n cond_z = (%s == 0);", buf1);
2937 pfomask &= ~(1 << PFO_Z);
2938 }
2939 if (pfomask & (1 << PFO_S)) {
2940 fprintf(fout, "\n cond_s = ((s32)%s < 0);", buf1);
2941 pfomask &= ~(1 << PFO_S);
2942 }
2943 last_arith_dst = &po->operand[0];
2944 delayed_flag_op = NULL;
2945 break;
2946
2947 case OP_SAR:
2948 assert_operand_cnt(2);
2949 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
2950 fprintf(fout, " %s = %s%s >> %s;", buf1,
2951 lmod_cast_s(po, po->operand[0].lmod), buf1,
2952 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2953 last_arith_dst = &po->operand[0];
2954 delayed_flag_op = NULL;
2955 break;
2956
2957 case OP_ROL:
2958 case OP_ROR:
2959 assert_operand_cnt(2);
2960 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
2961 if (po->operand[1].type == OPT_CONST) {
2962 j = po->operand[1].val;
2963 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
2964 fprintf(fout, po->op == OP_ROL ?
2965 " %s = (%s << %d) | (%s >> %d);" :
2966 " %s = (%s >> %d) | (%s << %d);",
2967 buf1, buf1, j, buf1,
2968 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
2969 }
2970 else
2971 ferr(po, "TODO\n");
2972 last_arith_dst = &po->operand[0];
2973 delayed_flag_op = NULL;
2974 break;
2975
2976 case OP_XOR:
2977 assert_operand_cnt(2);
2978 propagate_lmod(po, &po->operand[0], &po->operand[1]);
2979 if (IS(opr_name(po, 0), opr_name(po, 1))) {
2980 // special case for XOR
2981 fprintf(fout, " %s = 0;",
2982 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
2983 last_arith_dst = &po->operand[0];
2984 delayed_flag_op = NULL;
2985 break;
2986 }
2987 goto dualop_arith;
2988
2989 case OP_ADC:
2990 case OP_SBB:
2991 assert_operand_cnt(2);
2992 propagate_lmod(po, &po->operand[0], &po->operand[1]);
2993 if (po->op == OP_SBB
2994 && IS(po->operand[0].name, po->operand[1].name))
2995 {
2996 // avoid use of unitialized var
2997 fprintf(fout, " %s = -cond_c;",
2998 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
2999 }
3000 else {
3001 fprintf(fout, " %s %s= %s + cond_c;",
3002 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3003 op_to_c(po),
3004 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
3005 }
3006 last_arith_dst = &po->operand[0];
3007 delayed_flag_op = NULL;
3008 break;
3009
3010 case OP_INC:
3011 case OP_DEC:
3012 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3013 if (po->operand[0].type == OPT_REG) {
3014 strcpy(buf2, po->op == OP_INC ? "++" : "--");
3015 fprintf(fout, " %s%s;", buf1, buf2);
3016 }
3017 else {
3018 strcpy(buf2, po->op == OP_INC ? "+" : "-");
3019 fprintf(fout, " %s %s= 1;", buf1, buf2);
3020 }
3021 last_arith_dst = &po->operand[0];
3022 delayed_flag_op = NULL;
3023 break;
3024
3025 case OP_NEG:
3026 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3027 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
3028 fprintf(fout, " %s = -%s%s;", buf1,
3029 lmod_cast_s(po, po->operand[0].lmod), buf2);
3030 last_arith_dst = &po->operand[0];
3031 delayed_flag_op = NULL;
3032 if (pfomask & (1 << PFO_C)) {
3033 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
3034 pfomask &= ~(1 << PFO_C);
3035 }
3036 break;
3037
3038 case OP_IMUL:
3039 if (po->operand_cnt == 2) {
3040 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3041 goto dualop_arith;
3042 }
3043 if (po->operand_cnt == 3)
3044 ferr(po, "TODO imul3\n");
3045 // fallthrough
3046 case OP_MUL:
3047 assert_operand_cnt(1);
3048 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
3049 fprintf(fout, " mul_tmp = %seax * %s%s;\n", buf1, buf1,
3050 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
3051 fprintf(fout, " edx = mul_tmp >> 32;\n");
3052 fprintf(fout, " eax = mul_tmp;");
3053 last_arith_dst = NULL;
3054 delayed_flag_op = NULL;
3055 break;
3056
3057 case OP_DIV:
3058 case OP_IDIV:
3059 assert_operand_cnt(1);
3060 if (po->operand[0].lmod != OPLM_DWORD)
3061 ferr(po, "unhandled lmod %d\n", po->operand[0].lmod);
3062
3063 // 32bit division is common, look for it
3064 if (po->op == OP_DIV)
3065 ret = scan_for_reg_clear(i - 1, xDX);
3066 else
3067 ret = scan_for_cdq_edx(i - 1);
3068 if (ret >= 0) {
3069 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
3070 strcpy(buf2, lmod_cast(po, po->operand[0].lmod,
3071 po->op == OP_IDIV));
3072 fprintf(fout, " edx = %seax %% %s%s;\n", buf2, buf2, buf1);
3073 fprintf(fout, " eax = %seax / %s%s;", buf2, buf2, buf1);
3074 }
3075 else
3076 ferr(po, "TODO 64bit divident\n");
3077 last_arith_dst = NULL;
3078 delayed_flag_op = NULL;
3079 break;
3080
3081 case OP_TEST:
3082 case OP_CMP:
3083 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3084 if (pfomask != 0) {
3085 for (j = 0; j < 8; j++) {
3086 if (pfomask & (1 << j)) {
3087 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
3088 fprintf(fout, " cond_%s = %s;",
3089 parsed_flag_op_names[j], buf1);
3090 }
3091 }
3092 pfomask = 0;
3093 }
3094 else
3095 no_output = 1;
3096 last_arith_dst = NULL;
3097 delayed_flag_op = po;
3098 break;
3099
3100 // note: we reuse OP_Jcc for SETcc, only flags differ
3101 case OP_JO ... OP_JG:
3102 if (po->flags & OPF_JMP)
3103 fprintf(fout, " goto %s;", po->operand[0].name);
3104 // else SETcc - should already be handled
3105 break;
3106
3107 case OP_JMP:
3108 assert_operand_cnt(1);
3109 last_arith_dst = NULL;
3110 delayed_flag_op = NULL;
3111
3112 if (po->operand[0].type == OPT_REGMEM) {
3113 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
3114 buf1, buf2);
3115 if (ret != 2)
3116 ferr(po, "parse failure for jmp '%s'\n",
3117 po->operand[0].name);
3118 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
3119 break;
3120 }
3121 else if (po->operand[0].type != OPT_LABEL)
3122 ferr(po, "unhandled jmp type\n");
3123
3124 fprintf(fout, " goto %s;", po->operand[0].name);
3125 break;
3126
3127 case OP_CALL:
3128 assert_operand_cnt(1);
3129 pp = po->datap;
3130 if (pp == NULL)
3131 ferr(po, "NULL pp\n");
3132
3133 if (po->operand[0].type != OPT_LABEL)
3134 fprintf(fout, " icall%d = (void *)%s;\n", i,
3135 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]));
3136
3137 fprintf(fout, " ");
3138 if (strstr(pp->ret_type.name, "int64")) {
3139 if (po->flags & OPF_TAIL)
3140 ferr(po, "int64 and tail?\n");
3141 fprintf(fout, "mul_tmp = ");
3142 }
3143 else if (!IS(pp->ret_type.name, "void")) {
3144 if (po->flags & OPF_TAIL) {
3145 if (!IS(g_func_pp->ret_type.name, "void")) {
3146 fprintf(fout, "return ");
3147 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
3148 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
3149 }
3150 }
3151 else {
3152 fprintf(fout, "eax = ");
3153 if (pp->ret_type.is_ptr)
3154 fprintf(fout, "(u32)");
3155 }
3156 }
3157
3158 if (po->operand[0].type != OPT_LABEL) {
3159 fprintf(fout, "icall%d(", i);
3160 }
3161 else {
3162 if (pp->name[0] == 0)
3163 ferr(po, "missing pp->name\n");
3164 fprintf(fout, "%s%s(", pp->name,
3165 pp->has_structarg ? "_sa" : "");
3166 }
3167
3168 for (arg = 0; arg < pp->argc; arg++) {
3169 if (arg > 0)
3170 fprintf(fout, ", ");
3171
3172 cast[0] = 0;
3173 if (pp->arg[arg].type.is_ptr)
3174 snprintf(cast, sizeof(cast), "(%s)", pp->arg[arg].type.name);
3175
3176 if (pp->arg[arg].reg != NULL) {
3177 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
3178 continue;
3179 }
3180
3181 // stack arg
3182 tmp_op = pp->arg[arg].datap;
3183 if (tmp_op == NULL)
3184 ferr(po, "parsed_op missing for arg%d\n", arg);
3185 if (tmp_op->argnum != 0) {
3186 fprintf(fout, "%ss_a%d", cast, tmp_op->argnum);
3187 }
3188 else {
3189 fprintf(fout, "%s",
3190 out_src_opr(buf1, sizeof(buf1),
3191 tmp_op, &tmp_op->operand[0], cast, 0));
3192 }
3193 }
3194 fprintf(fout, ");");
3195
3196 if (strstr(pp->ret_type.name, "int64")) {
3197 fprintf(fout, "\n");
3198 fprintf(fout, " edx = mul_tmp >> 32;\n");
3199 fprintf(fout, " eax = mul_tmp;");
3200 }
3201
3202 if (po->flags & OPF_TAIL) {
3203 strcpy(g_comment, "tailcall");
3204 ret = 0;
3205 if (i == opcnt - 1)
3206 ret = 0;
3207 else if (IS(pp->ret_type.name, "void"))
3208 ret = 1;
3209 else if (IS(g_func_pp->ret_type.name, "void"))
3210 ret = 1;
3211 // else already handled as 'return f()'
3212
3213 if (ret) {
3214 fprintf(fout, "\n return;");
3215 strcpy(g_comment, "^ tailcall");
3216 }
3217 }
3218 delayed_flag_op = NULL;
3219 last_arith_dst = NULL;
3220 break;
3221
3222 case OP_RET:
3223 if (g_func_pp->is_vararg)
3224 fprintf(fout, " va_end(ap);\n");
3225
3226 if (IS(g_func_pp->ret_type.name, "void")) {
3227 if (i != opcnt - 1 || label_pending)
3228 fprintf(fout, " return;");
3229 }
3230 else if (g_func_pp->ret_type.is_ptr) {
3231 fprintf(fout, " return (%s)eax;",
3232 g_func_pp->ret_type.name);
3233 }
3234 else
3235 fprintf(fout, " return eax;");
3236
3237 last_arith_dst = NULL;
3238 delayed_flag_op = NULL;
3239 break;
3240
3241 case OP_PUSH:
3242 if (po->argnum != 0) {
3243 // special case - saved func arg
3244 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
3245 fprintf(fout, " s_a%d = %s;", po->argnum, buf1);
3246 break;
3247 }
3248 else if (po->flags & OPF_RSAVE) {
3249 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
3250 fprintf(fout, " s_%s = %s;", buf1, buf1);
3251 break;
3252 }
3253 if (!(g_ida_func_attr & IDAFA_NORETURN))
3254 ferr(po, "stray push encountered\n");
3255 no_output = 1;
3256 break;
3257
3258 case OP_POP:
3259 if (po->flags & OPF_RSAVE) {
3260 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3261 fprintf(fout, " %s = s_%s;", buf1, buf1);
3262 break;
3263 }
3264 ferr(po, "stray pop encountered\n");
3265 break;
3266
3267 case OP_NOP:
3268 no_output = 1;
3269 break;
3270
3271 default:
3272 no_output = 1;
3273 ferr(po, "unhandled op type %d, flags %x\n",
3274 po->op, po->flags);
3275 break;
3276 }
3277
3278 if (g_comment[0] != 0) {
3279 fprintf(fout, " // %s", g_comment);
3280 g_comment[0] = 0;
3281 no_output = 0;
3282 }
3283 if (!no_output)
3284 fprintf(fout, "\n");
3285
3286 // some sanity checking
3287 if ((po->flags & OPF_REP) && po->op != OP_STOS
3288 && po->op != OP_MOVS && po->op != OP_CMPS)
3289 ferr(po, "unexpected rep\n");
3290 if ((po->flags & (OPF_REPZ|OPF_REPNZ)) && po->op != OP_CMPS)
3291 ferr(po, "unexpected repz/repnz\n");
3292
3293 if (pfomask != 0)
3294 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
3295
3296 // see is delayed flag stuff is still valid
3297 if (delayed_flag_op != NULL && delayed_flag_op != po) {
3298 if (is_any_opr_modified(delayed_flag_op, po))
3299 delayed_flag_op = NULL;
3300 }
3301
3302 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
3303 if (is_opr_modified(last_arith_dst, po))
3304 last_arith_dst = NULL;
3305 }
3306
3307 label_pending = 0;
3308 }
3309
3310 fprintf(fout, "}\n\n");
3311
3312 // cleanup
3313 for (i = 0; i < opcnt; i++) {
3314 struct label_ref *lr, *lr_del;
3315
3316 lr = g_label_refs[i].next;
3317 while (lr != NULL) {
3318 lr_del = lr;
3319 lr = lr->next;
3320 free(lr_del);
3321 }
3322 g_label_refs[i].i = -1;
3323 g_label_refs[i].next = NULL;
3324
3325 if (ops[i].op == OP_CALL) {
3326 pp = ops[i].datap;
3327 if (pp)
3328 proto_release(pp);
3329 }
3330 }
3331 g_func_pp = NULL;
3332}
3333
3334static void set_label(int i, const char *name)
3335{
3336 const char *p;
3337 int len;
3338
3339 len = strlen(name);
3340 p = strchr(name, ':');
3341 if (p != NULL)
3342 len = p - name;
3343
3344 if (len > sizeof(g_labels[0]) - 1)
3345 aerr("label '%s' too long: %d\n", name, len);
3346 if (g_labels[i][0] != 0 && !IS_START(g_labels[i], "algn_"))
3347 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
3348 memcpy(g_labels[i], name, len);
3349 g_labels[i][len] = 0;
3350}
3351
3352// '=' needs special treatment..
3353static char *next_word_s(char *w, size_t wsize, char *s)
3354{
3355 size_t i;
3356
3357 s = sskip(s);
3358
3359 for (i = 0; i < wsize - 1; i++) {
3360 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
3361 break;
3362 w[i] = s[i];
3363 }
3364 w[i] = 0;
3365
3366 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
3367 printf("warning: '%s' truncated\n", w);
3368
3369 return s + i;
3370}
3371
3372struct chunk_item {
3373 char *name;
3374 long fptr;
3375 int asmln;
3376};
3377
3378static int cmp_chunks(const void *p1, const void *p2)
3379{
3380 const struct chunk_item *c1 = p1, *c2 = p2;
3381 return strcmp(c1->name, c2->name);
3382}
3383
3384static int cmpstringp(const void *p1, const void *p2)
3385{
3386 return strcmp(*(char * const *)p1, *(char * const *)p2);
3387}
3388
3389int main(int argc, char *argv[])
3390{
3391 FILE *fout, *fasm, *frlist;
3392 struct parsed_data *pd = NULL;
3393 int pd_alloc = 0;
3394 char **rlist = NULL;
3395 int rlist_len = 0;
3396 int rlist_alloc = 0;
3397 struct chunk_item *func_chunks;
3398 int func_chunks_used = 0;
3399 int func_chunks_sorted = 0;
3400 int func_chunk_cnt = 0;
3401 int func_chunk_alloc;
3402 int func_chunk_i = -1;
3403 long func_chunk_ret = 0;
3404 int func_chunk_ret_ln = 0;
3405 char line[256];
3406 char words[16][256];
3407 enum opr_lenmod lmod;
3408 int in_func = 0;
3409 int pending_endp = 0;
3410 int skip_func = 0;
3411 int skip_warned = 0;
3412 int eq_alloc;
3413 int verbose = 0;
3414 int arg_out;
3415 int arg = 1;
3416 int pi = 0;
3417 int i, j;
3418 int ret, len;
3419 char *p;
3420 int wordc;
3421
3422 if (argv[1] && IS(argv[1], "-v")) {
3423 verbose = 1;
3424 arg++;
3425 }
3426
3427 if (argc < arg + 3) {
3428 printf("usage:\n%s [-v] <.c> <.asm> <hdrf> [rlist]*\n",
3429 argv[0]);
3430 return 1;
3431 }
3432
3433 arg_out = arg++;
3434
3435 asmfn = argv[arg++];
3436 fasm = fopen(asmfn, "r");
3437 my_assert_not(fasm, NULL);
3438
3439 hdrfn = argv[arg++];
3440 g_fhdr = fopen(hdrfn, "r");
3441 my_assert_not(g_fhdr, NULL);
3442
3443 rlist_alloc = 64;
3444 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
3445 my_assert_not(rlist, NULL);
3446 // needs special handling..
3447 rlist[rlist_len++] = "__alloca_probe";
3448
3449 func_chunk_alloc = 32;
3450 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
3451 my_assert_not(func_chunks, NULL);
3452
3453 for (; arg < argc; arg++) {
3454 frlist = fopen(argv[arg], "r");
3455 my_assert_not(frlist, NULL);
3456
3457 while (fgets(line, sizeof(line), frlist)) {
3458 p = sskip(line);
3459 if (*p == 0 || *p == ';' || *p == '#')
3460 continue;
3461
3462 p = next_word(words[0], sizeof(words[0]), p);
3463 if (words[0][0] == 0)
3464 continue;
3465
3466 if (rlist_len >= rlist_alloc) {
3467 rlist_alloc = rlist_alloc * 2 + 64;
3468 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
3469 my_assert_not(rlist, NULL);
3470 }
3471 rlist[rlist_len++] = strdup(words[0]);
3472 }
3473
3474 fclose(frlist);
3475 frlist = NULL;
3476 }
3477
3478 if (rlist_len > 0)
3479 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
3480
3481 fout = fopen(argv[arg_out], "w");
3482 my_assert_not(fout, NULL);
3483
3484 eq_alloc = 128;
3485 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
3486 my_assert_not(g_eqs, NULL);
3487
3488 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
3489 g_label_refs[i].i = -1;
3490 g_label_refs[i].next = NULL;
3491 }
3492
3493 while (fgets(line, sizeof(line), fasm))
3494 {
3495 wordc = 0;
3496 asmln++;
3497
3498 p = sskip(line);
3499 if (*p == 0)
3500 continue;
3501
3502 // get rid of random tabs
3503 for (i = 0; line[i] != 0; i++)
3504 if (line[i] == '\t')
3505 line[i] = ' ';
3506
3507 if (*p == ';')
3508 {
3509 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
3510 goto do_pending_endp; // eww..
3511
3512 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
3513 {
3514 static const char *attrs[] = {
3515 "bp-based frame",
3516 "library function",
3517 "static",
3518 "noreturn",
3519 "thunk",
3520 "fpd=",
3521 };
3522
3523 // parse IDA's attribute-list comment
3524 g_ida_func_attr = 0;
3525 p = sskip(p + 13);
3526
3527 for (; *p != 0; p = sskip(p)) {
3528 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
3529 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
3530 g_ida_func_attr |= 1 << i;
3531 p += strlen(attrs[i]);
3532 break;
3533 }
3534 }
3535 if (i == ARRAY_SIZE(attrs)) {
3536 anote("unparsed IDA attr: %s\n", p);
3537 break;
3538 }
3539 if (IS(attrs[i], "fpd=")) {
3540 p = next_word(words[0], sizeof(words[0]), p);
3541 // ignore for now..
3542 }
3543 }
3544 }
3545 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
3546 {
3547 p += 30;
3548 next_word(words[0], sizeof(words[0]), p);
3549 if (words[0][0] == 0)
3550 aerr("missing nam for func chunk?\n");
3551 if (func_chunk_cnt >= func_chunk_alloc) {
3552 func_chunk_alloc *= 2;
3553 func_chunks = realloc(func_chunks,
3554 func_chunk_alloc * sizeof(func_chunks[0]));
3555 my_assert_not(func_chunks, NULL);
3556 }
3557 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
3558 func_chunks[func_chunk_cnt].name = strdup(words[0]);
3559 func_chunks[func_chunk_cnt].asmln = asmln;
3560 func_chunk_cnt++;
3561 func_chunks_sorted = 0;
3562 }
3563 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
3564 {
3565 if (func_chunk_i >= 0) {
3566 if (func_chunk_i < func_chunk_cnt
3567 && IS(func_chunks[func_chunk_i].name, g_func))
3568 {
3569 // move on to next chunk
3570 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
3571 if (ret)
3572 aerr("seek failed for '%s' chunk #%d\n",
3573 g_func, func_chunk_i);
3574 asmln = func_chunks[func_chunk_i].asmln;
3575 func_chunk_i++;
3576 }
3577 else {
3578 if (func_chunk_ret == 0)
3579 aerr("no return from chunk?\n");
3580 fseek(fasm, func_chunk_ret, SEEK_SET);
3581 asmln = func_chunk_ret_ln;
3582 func_chunk_ret = 0;
3583 pending_endp = 1;
3584 }
3585 }
3586 }
3587 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
3588 func_chunks_used = 1;
3589 p += 20;
3590 if (IS_START(g_func, "sub_")) {
3591 unsigned long addr = strtoul(p, NULL, 16);
3592 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
3593 if (addr > f_addr)
3594 aerr("need a chunk %lX that is after %s\n", addr, g_func);
3595 }
3596 }
3597 continue;
3598 } // *p == ';'
3599
3600parse_words:
3601 memset(words, 0, sizeof(words));
3602 for (wordc = 0; wordc < 16; wordc++) {
3603 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
3604 if (*p == 0 || *p == ';') {
3605 wordc++;
3606 break;
3607 }
3608 }
3609
3610 // alow asm patches in comments
3611 if (*p == ';' && IS_START(p, "; sctpatch:")) {
3612 p = sskip(p + 11);
3613 if (*p == 0 || *p == ';')
3614 continue;
3615 goto parse_words; // lame
3616 }
3617
3618 if (wordc == 0) {
3619 // shouldn't happen
3620 awarn("wordc == 0?\n");
3621 continue;
3622 }
3623
3624 // don't care about this:
3625 if (words[0][0] == '.'
3626 || IS(words[0], "include")
3627 || IS(words[0], "assume") || IS(words[1], "segment")
3628 || IS(words[0], "align"))
3629 {
3630 continue;
3631 }
3632
3633do_pending_endp:
3634 // do delayed endp processing to collect switch jumptables
3635 if (pending_endp) {
3636 if (in_func && !skip_func && wordc >= 2
3637 && ((words[0][0] == 'd' && words[0][2] == 0)
3638 || (words[1][0] == 'd' && words[1][2] == 0)))
3639 {
3640 i = 1;
3641 if (words[1][0] == 'd' && words[1][2] == 0) {
3642 // label
3643 if (g_func_pd_cnt >= pd_alloc) {
3644 pd_alloc = pd_alloc * 2 + 16;
3645 g_func_pd = realloc(g_func_pd,
3646 sizeof(g_func_pd[0]) * pd_alloc);
3647 my_assert_not(g_func_pd, NULL);
3648 }
3649 pd = &g_func_pd[g_func_pd_cnt];
3650 g_func_pd_cnt++;
3651 memset(pd, 0, sizeof(*pd));
3652 strcpy(pd->label, words[0]);
3653 pd->type = OPT_CONST;
3654 pd->lmod = lmod_from_directive(words[1]);
3655 i = 2;
3656 }
3657 else {
3658 lmod = lmod_from_directive(words[0]);
3659 if (lmod != pd->lmod)
3660 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
3661 }
3662
3663 if (pd->count_alloc < pd->count + wordc) {
3664 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
3665 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
3666 my_assert_not(pd->d, NULL);
3667 }
3668 for (; i < wordc; i++) {
3669 if (IS(words[i], "offset")) {
3670 pd->type = OPT_OFFSET;
3671 i++;
3672 }
3673 p = strchr(words[i], ',');
3674 if (p != NULL)
3675 *p = 0;
3676 if (pd->type == OPT_OFFSET)
3677 pd->d[pd->count].u.label = strdup(words[i]);
3678 else
3679 pd->d[pd->count].u.val = parse_number(words[i]);
3680 pd->d[pd->count].bt_i = -1;
3681 pd->count++;
3682 }
3683 continue;
3684 }
3685
3686 if (in_func && !skip_func)
3687 gen_func(fout, g_fhdr, g_func, pi);
3688
3689 pending_endp = 0;
3690 in_func = 0;
3691 g_ida_func_attr = 0;
3692 skip_warned = 0;
3693 skip_func = 0;
3694 g_func[0] = 0;
3695 func_chunks_used = 0;
3696 func_chunk_i = -1;
3697 if (pi != 0) {
3698 memset(&ops, 0, pi * sizeof(ops[0]));
3699 memset(g_labels, 0, pi * sizeof(g_labels[0]));
3700 pi = 0;
3701 }
3702 g_eqcnt = 0;
3703 for (i = 0; i < g_func_pd_cnt; i++) {
3704 pd = &g_func_pd[i];
3705 if (pd->type == OPT_OFFSET) {
3706 for (j = 0; j < pd->count; j++)
3707 free(pd->d[j].u.label);
3708 }
3709 free(pd->d);
3710 pd->d = NULL;
3711 }
3712 g_func_pd_cnt = 0;
3713 pd = NULL;
3714 if (wordc == 0)
3715 continue;
3716 }
3717
3718 if (IS(words[1], "proc")) {
3719 if (in_func)
3720 aerr("proc '%s' while in_func '%s'?\n",
3721 words[0], g_func);
3722 p = words[0];
3723 if ((g_ida_func_attr & IDAFA_THUNK)
3724 || bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
3725 skip_func = 1;
3726 strcpy(g_func, words[0]);
3727 set_label(0, words[0]);
3728 in_func = 1;
3729 continue;
3730 }
3731
3732 if (IS(words[1], "endp"))
3733 {
3734 if (!in_func)
3735 aerr("endp '%s' while not in_func?\n", words[0]);
3736 if (!IS(g_func, words[0]))
3737 aerr("endp '%s' while in_func '%s'?\n",
3738 words[0], g_func);
3739
3740 if (!skip_func && func_chunks_used) {
3741 // start processing chunks
3742 struct chunk_item *ci, key = { g_func, 0 };
3743
3744 func_chunk_ret = ftell(fasm);
3745 func_chunk_ret_ln = asmln;
3746 if (!func_chunks_sorted) {
3747 qsort(func_chunks, func_chunk_cnt,
3748 sizeof(func_chunks[0]), cmp_chunks);
3749 func_chunks_sorted = 1;
3750 }
3751 ci = bsearch(&key, func_chunks, func_chunk_cnt,
3752 sizeof(func_chunks[0]), cmp_chunks);
3753 if (ci == NULL)
3754 aerr("'%s' needs chunks, but none found\n", g_func);
3755 func_chunk_i = ci - func_chunks;
3756 for (; func_chunk_i > 0; func_chunk_i--)
3757 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
3758 break;
3759
3760 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
3761 if (ret)
3762 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
3763 asmln = func_chunks[func_chunk_i].asmln;
3764 func_chunk_i++;
3765 continue;
3766 }
3767 pending_endp = 1;
3768 continue;
3769 }
3770
3771 p = strchr(words[0], ':');
3772 if (p != NULL) {
3773 set_label(pi, words[0]);
3774 continue;
3775 }
3776
3777 if (!in_func || skip_func) {
3778 if (!skip_warned && !skip_func && g_labels[pi][0] != 0) {
3779 if (verbose)
3780 anote("skipping from '%s'\n", g_labels[pi]);
3781 skip_warned = 1;
3782 }
3783 g_labels[pi][0] = 0;
3784 continue;
3785 }
3786
3787 if (wordc > 1 && IS(words[1], "=")) {
3788 if (wordc != 5)
3789 aerr("unhandled equ, wc=%d\n", wordc);
3790 if (g_eqcnt >= eq_alloc) {
3791 eq_alloc *= 2;
3792 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
3793 my_assert_not(g_eqs, NULL);
3794 }
3795
3796 len = strlen(words[0]);
3797 if (len > sizeof(g_eqs[0].name) - 1)
3798 aerr("equ name too long: %d\n", len);
3799 strcpy(g_eqs[g_eqcnt].name, words[0]);
3800
3801 if (!IS(words[3], "ptr"))
3802 aerr("unhandled equ\n");
3803 if (IS(words[2], "dword"))
3804 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
3805 else if (IS(words[2], "word"))
3806 g_eqs[g_eqcnt].lmod = OPLM_WORD;
3807 else if (IS(words[2], "byte"))
3808 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
3809 else
3810 aerr("bad lmod: '%s'\n", words[2]);
3811
3812 g_eqs[g_eqcnt].offset = parse_number(words[4]);
3813 g_eqcnt++;
3814 continue;
3815 }
3816
3817 if (pi >= ARRAY_SIZE(ops))
3818 aerr("too many ops\n");
3819
3820 parse_op(&ops[pi], words, wordc);
3821 pi++;
3822 }
3823
3824 fclose(fout);
3825 fclose(fasm);
3826 fclose(g_fhdr);
3827
3828 return 0;
3829}
3830
3831// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.