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