notaz.gp2x.de / ia32rtools.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
use unsigned types, cc+nested branch fix, var type cast
[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 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 };
473 static const char *word_types[] = {
474 "uint16_t", "int16_t",
475 "unsigned __int16", "__int16",
476 };
477 static const char *byte_types[] = {
478 "uint8_t", "int8_t", "char",
479 "unsigned __int8", "__int8", "BYTE",
480 };
481 const char *n;
482 int i;
483
484 if (c_type->is_ptr) {
485 *lmod = OPLM_DWORD;
486 return 1;
487 }
488
489 n = skip_type_mod(c_type->name);
490
491 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
492 if (IS(n, dword_types[i])) {
493 *lmod = OPLM_DWORD;
494 return 1;
495 }
496 }
497
498 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
499 if (IS(n, word_types[i])) {
500 *lmod = OPLM_WORD;
501 return 1;
502 }
503 }
504
505 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
506 if (IS(n, byte_types[i])) {
507 *lmod = OPLM_BYTE;
508 return 1;
509 }
510 }
511
512 return 0;
513}
514
515static enum opr_type lmod_from_directive(const char *d)
516{
517 if (IS(d, "dd"))
518 return OPLM_DWORD;
519 else if (IS(d, "dw"))
520 return OPLM_WORD;
521 else if (IS(d, "db"))
522 return OPLM_BYTE;
523
524 aerr("unhandled directive: '%s'\n", d);
525 return OPLM_UNSPEC;
526}
527
528static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
529 int *regmask)
530{
531 opr->type = OPT_REG;
532 opr->reg = reg;
533 opr->lmod = lmod;
534 *regmask |= 1 << reg;
535}
536
537static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
538 int *extra_offs);
539
540static int parse_operand(struct parsed_opr *opr,
541 int *regmask, int *regmask_indirect,
542 char words[16][256], int wordc, int w, unsigned int op_flags)
543{
544 struct parsed_proto pp;
545 enum opr_lenmod tmplmod;
546 unsigned long number;
547 int ret, len;
548 int wordc_in;
549 char *tmp;
550 int i;
551
552 if (w >= wordc)
553 aerr("parse_operand w %d, wordc %d\n", w, wordc);
554
555 opr->reg = xUNSPEC;
556
557 for (i = w; i < wordc; i++) {
558 len = strlen(words[i]);
559 if (words[i][len - 1] == ',') {
560 words[i][len - 1] = 0;
561 wordc = i + 1;
562 break;
563 }
564 }
565
566 wordc_in = wordc - w;
567
568 if ((op_flags & OPF_JMP) && wordc_in > 0
569 && !('0' <= words[w][0] && words[w][0] <= '9'))
570 {
571 const char *label = NULL;
572
573 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
574 && IS(words[w + 1], "ptr"))
575 label = words[w + 2];
576 else if (wordc_in == 2 && IS(words[w], "short"))
577 label = words[w + 1];
578 else if (wordc_in == 1
579 && strchr(words[w], '[') == NULL
580 && parse_reg(&tmplmod, words[w]) < 0)
581 label = words[w];
582
583 if (label != NULL) {
584 opr->type = OPT_LABEL;
585 if (IS_START(label, "ds:"))
586 label += 3;
587 strcpy(opr->name, label);
588 return wordc;
589 }
590 }
591
592 if (wordc_in >= 3) {
593 if (IS(words[w + 1], "ptr")) {
594 if (IS(words[w], "dword"))
595 opr->lmod = OPLM_DWORD;
596 else if (IS(words[w], "word"))
597 opr->lmod = OPLM_WORD;
598 else if (IS(words[w], "byte"))
599 opr->lmod = OPLM_BYTE;
600 else
601 aerr("type parsing failed\n");
602 w += 2;
603 wordc_in = wordc - w;
604 }
605 }
606
607 if (wordc_in == 2) {
608 if (IS(words[w], "offset")) {
609 opr->type = OPT_OFFSET;
610 strcpy(opr->name, words[w + 1]);
611 return wordc;
612 }
613 if (IS(words[w], "(offset")) {
614 char *p = strchr(words[w + 1], ')');
615 if (p == NULL)
616 aerr("parse of bracketed offset failed\n");
617 *p = 0;
618 opr->type = OPT_OFFSET;
619 strcpy(opr->name, words[w + 1]);
620 return wordc;
621 }
622 }
623
624 if (wordc_in != 1)
625 aerr("parse_operand 1 word expected\n");
626
627 tmp = words[w];
628 if (IS_START(tmp, "ds:"))
629 tmp += 3;
630 strcpy(opr->name, tmp);
631
632 if (words[w][0] == '[') {
633 opr->type = OPT_REGMEM;
634 ret = sscanf(words[w], "[%[^]]]", opr->name);
635 if (ret != 1)
636 aerr("[] parse failure\n");
637
638 parse_indmode(opr->name, regmask_indirect, 1);
639 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL)) {
640 // might be an equ
641 struct parsed_equ *eq =
642 equ_find(NULL, parse_stack_el(opr->name, NULL), &i);
643 if (eq)
644 opr->lmod = eq->lmod;
645 }
646 return wordc;
647 }
648 else if (strchr(words[w], '[')) {
649 // label[reg] form
650 opr->type = OPT_REGMEM;
651 if (opr->lmod == OPLM_UNSPEC)
652 guess_lmod_from_name(opr);
653 parse_indmode(strchr(words[w], '['), regmask_indirect, 0);
654 return wordc;
655 }
656 else if (('0' <= words[w][0] && words[w][0] <= '9')
657 || words[w][0] == '-')
658 {
659 number = parse_number(words[w]);
660 opr->type = OPT_CONST;
661 opr->val = number;
662 printf_number(opr->name, sizeof(opr->name), number);
663 return wordc;
664 }
665
666 ret = parse_reg(&tmplmod, opr->name);
667 if (ret >= 0) {
668 setup_reg_opr(opr, ret, tmplmod, regmask);
669 return wordc;
670 }
671
672 // most likely var in data segment
673 opr->type = OPT_LABEL;
674
675 ret = proto_parse(g_fhdr, opr->name, &pp);
676 if (ret == 0) {
677 if (pp.is_fptr) {
678 opr->lmod = OPLM_DWORD;
679 opr->is_ptr = 1;
680 }
681 else {
682 if (!guess_lmod_from_c_type(&tmplmod, &pp.type))
683 anote("unhandled C type '%s' for '%s'\n",
684 pp.type.name, opr->name);
685
686 if (opr->lmod == OPLM_UNSPEC)
687 opr->lmod = tmplmod;
688 else if (opr->lmod != tmplmod) {
689 opr->size_mismatch = 1;
690 if (tmplmod < opr->lmod)
691 opr->size_lt = 1;
692 }
693 }
694 opr->is_ptr = pp.type.is_ptr;
695 opr->is_array = pp.type.is_array;
696 }
697 proto_release(&pp);
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 && arg_i == g_func_pp.argc_stack && is_lea) {
1221 // should be va_list
1222 if (cast[0] == 0)
1223 cast = "(u32)";
1224 snprintf(buf, buf_size, "%sap", cast);
1225 return;
1226 }
1227 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1228 offset, bp_arg, arg_i);
1229 }
1230 if (ofs_reg[0] != 0)
1231 ferr(po, "offset reg on arg access?\n");
1232
1233 for (i = arg_s = 0; i < g_func_pp.argc; i++) {
1234 if (g_func_pp.arg[i].reg != NULL)
1235 continue;
1236 if (arg_s == arg_i)
1237 break;
1238 arg_s++;
1239 }
1240 if (i == g_func_pp.argc)
1241 ferr(po, "arg %d not in prototype?\n", arg_i);
1242
1243 popr->is_ptr = g_func_pp.arg[i].type.is_ptr;
1244
1245 switch (popr->lmod)
1246 {
1247 case OPLM_BYTE:
1248 if (is_lea)
1249 ferr(po, "lea/byte to arg?\n");
1250 if (is_src && (offset & 3) == 0)
1251 snprintf(buf, buf_size, "(u8)a%d", i + 1);
1252 else
1253 snprintf(buf, buf_size, "BYTE%d(a%d)", offset & 3, i + 1);
1254 break;
1255
1256 case OPLM_WORD:
1257 if (is_lea)
1258 ferr(po, "lea/word to arg?\n");
1259 if (offset & 1)
1260 ferr(po, "unaligned arg access\n");
1261 if (is_src && (offset & 2) == 0)
1262 snprintf(buf, buf_size, "(u16)a%d", i + 1);
1263 else
1264 snprintf(buf, buf_size, "%sWORD(a%d)",
1265 (offset & 2) ? "HI" : "LO", i + 1);
1266 break;
1267
1268 case OPLM_DWORD:
1269 if (cast[0])
1270 prefix = cast;
1271 else if (is_src)
1272 prefix = "(u32)";
1273 if (offset & 3) {
1274 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1275 if (is_lea)
1276 snprintf(buf, buf_size, "(u32)&a%d + %d",
1277 i + 1, offset & 3);
1278 else
1279 snprintf(buf, buf_size, "%s(a%d >> %d)",
1280 prefix, i + 1, (offset & 3) * 8);
1281 }
1282 else {
1283 snprintf(buf, buf_size, "%s%sa%d",
1284 prefix, is_lea ? "&" : "", i + 1);
1285 }
1286 break;
1287
1288 default:
1289 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
1290 }
1291
1292 // common problem
1293 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp.arg[i].type);
1294 if ((offset & 3) && tmp_lmod != OPLM_DWORD)
1295 ferr(po, "bp_arg arg/w offset %d and type '%s'\n",
1296 offset, g_func_pp.arg[i].type.name);
1297 }
1298 else
1299 {
1300 if (g_stack_fsz == 0)
1301 ferr(po, "stack var access without stackframe\n");
1302
1303 sf_ofs = g_stack_fsz + offset;
1304 lim = (ofs_reg[0] != 0) ? -4 : 0;
1305 if (offset > 0 || sf_ofs < lim)
1306 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
1307
1308 if (is_lea)
1309 prefix = "(u32)&";
1310 else
1311 prefix = cast;
1312
1313 switch (popr->lmod)
1314 {
1315 case OPLM_BYTE:
1316 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1317 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1318 break;
1319
1320 case OPLM_WORD:
1321 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1322 // known unaligned or possibly unaligned
1323 strcat(g_comment, " unaligned");
1324 if (prefix[0] == 0)
1325 prefix = "*(u16 *)&";
1326 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1327 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1328 break;
1329 }
1330 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
1331 break;
1332
1333 case OPLM_DWORD:
1334 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1335 // known unaligned or possibly unaligned
1336 strcat(g_comment, " unaligned");
1337 if (prefix[0] == 0)
1338 prefix = "*(u32 *)&";
1339 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1340 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1341 break;
1342 }
1343 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
1344 break;
1345
1346 default:
1347 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
1348 }
1349 }
1350}
1351
1352static void check_label_read_ref(struct parsed_op *po, const char *name)
1353{
1354 if (IS_START(name, "sub_"))
1355 ferr(po, "func reference?\n");
1356}
1357
1358static char *out_src_opr(char *buf, size_t buf_size,
1359 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
1360 int is_lea)
1361{
1362 char tmp1[256], tmp2[256];
1363 char expr[256];
1364 int ret;
1365
1366 if (cast == NULL)
1367 cast = "";
1368
1369 switch (popr->type) {
1370 case OPT_REG:
1371 if (is_lea)
1372 ferr(po, "lea from reg?\n");
1373
1374 switch (popr->lmod) {
1375 case OPLM_DWORD:
1376 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
1377 break;
1378 case OPLM_WORD:
1379 snprintf(buf, buf_size, "(u16)%s", opr_reg_p(po, popr));
1380 break;
1381 case OPLM_BYTE:
1382 if (popr->name[1] == 'h') // XXX..
1383 snprintf(buf, buf_size, "(u8)(%s >> 8)", opr_reg_p(po, popr));
1384 else
1385 snprintf(buf, buf_size, "(u8)%s", opr_reg_p(po, popr));
1386 break;
1387 default:
1388 ferr(po, "invalid src lmod: %d\n", popr->lmod);
1389 }
1390 break;
1391
1392 case OPT_REGMEM:
1393 if (parse_stack_el(popr->name, NULL)
1394 || (g_bp_frame && IS_START(popr->name, "ebp-")))
1395 {
1396 stack_frame_access(po, popr, buf, buf_size,
1397 popr->name, cast, 1, is_lea);
1398 break;
1399 }
1400
1401 strcpy(expr, popr->name);
1402 if (strchr(expr, '[')) {
1403 // special case: '[' can only be left for label[reg] form
1404 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
1405 if (ret != 2)
1406 ferr(po, "parse failure for '%s'\n", expr);
1407 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
1408 }
1409
1410 // XXX: do we need more parsing?
1411 if (is_lea) {
1412 snprintf(buf, buf_size, "%s", expr);
1413 break;
1414 }
1415
1416 if (cast[0] == 0)
1417 cast = lmod_cast_u_ptr(po, popr->lmod);
1418 snprintf(buf, buf_size, "%s(%s)", cast, expr);
1419 break;
1420
1421 case OPT_LABEL:
1422 check_label_read_ref(po, popr->name);
1423 if (cast[0] == 0 && popr->is_ptr)
1424 cast = "(u32)";
1425
1426 if (is_lea)
1427 snprintf(buf, buf_size, "(u32)&%s", popr->name);
1428 else if (popr->size_lt)
1429 snprintf(buf, buf_size, "%s%s%s%s", cast,
1430 lmod_cast_u_ptr(po, popr->lmod),
1431 popr->is_array ? "" : "&",
1432 popr->name);
1433 else
1434 snprintf(buf, buf_size, "%s%s%s", cast, popr->name,
1435 popr->is_array ? "[0]" : "");
1436 break;
1437
1438 case OPT_OFFSET:
1439 check_label_read_ref(po, popr->name);
1440 if (cast[0] == 0)
1441 cast = "(u32)";
1442 if (is_lea)
1443 ferr(po, "lea an offset?\n");
1444 snprintf(buf, buf_size, "%s&%s", cast, popr->name);
1445 break;
1446
1447 case OPT_CONST:
1448 if (is_lea)
1449 ferr(po, "lea from const?\n");
1450
1451 snprintf(buf, buf_size, "%s", cast);
1452 ret = strlen(buf);
1453 printf_number(buf + ret, buf_size - ret, popr->val);
1454 break;
1455
1456 default:
1457 ferr(po, "invalid src type: %d\n", popr->type);
1458 }
1459
1460 return buf;
1461}
1462
1463// note: may set is_ptr (we find that out late for ebp frame..)
1464static char *out_dst_opr(char *buf, size_t buf_size,
1465 struct parsed_op *po, struct parsed_opr *popr)
1466{
1467 switch (popr->type) {
1468 case OPT_REG:
1469 switch (popr->lmod) {
1470 case OPLM_DWORD:
1471 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
1472 break;
1473 case OPLM_WORD:
1474 // ugh..
1475 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
1476 break;
1477 case OPLM_BYTE:
1478 // ugh..
1479 if (popr->name[1] == 'h') // XXX..
1480 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
1481 else
1482 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
1483 break;
1484 default:
1485 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
1486 }
1487 break;
1488
1489 case OPT_REGMEM:
1490 if (parse_stack_el(popr->name, NULL)
1491 || (g_bp_frame && IS_START(popr->name, "ebp-")))
1492 {
1493 stack_frame_access(po, popr, buf, buf_size,
1494 popr->name, "", 0, 0);
1495 break;
1496 }
1497
1498 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1499
1500 case OPT_LABEL:
1501 if (popr->size_mismatch)
1502 snprintf(buf, buf_size, "%s%s%s",
1503 lmod_cast_u_ptr(po, popr->lmod),
1504 popr->is_array ? "" : "&", popr->name);
1505 else
1506 snprintf(buf, buf_size, "%s%s", popr->name,
1507 popr->is_array ? "[0]" : "");
1508 break;
1509
1510 default:
1511 ferr(po, "invalid dst type: %d\n", popr->type);
1512 }
1513
1514 return buf;
1515}
1516
1517static char *out_src_opr_u32(char *buf, size_t buf_size,
1518 struct parsed_op *po, struct parsed_opr *popr)
1519{
1520 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1521}
1522
1523static enum parsed_flag_op split_cond(struct parsed_op *po,
1524 enum op_op op, int *is_inv)
1525{
1526 *is_inv = 0;
1527
1528 switch (op) {
1529 case OP_JO:
1530 return PFO_O;
1531 case OP_JC:
1532 return PFO_C;
1533 case OP_JZ:
1534 return PFO_Z;
1535 case OP_JBE:
1536 return PFO_BE;
1537 case OP_JS:
1538 return PFO_S;
1539 case OP_JP:
1540 return PFO_P;
1541 case OP_JL:
1542 return PFO_L;
1543 case OP_JLE:
1544 return PFO_LE;
1545
1546 case OP_JNO:
1547 *is_inv = 1;
1548 return PFO_O;
1549 case OP_JNC:
1550 *is_inv = 1;
1551 return PFO_C;
1552 case OP_JNZ:
1553 *is_inv = 1;
1554 return PFO_Z;
1555 case OP_JA:
1556 *is_inv = 1;
1557 return PFO_BE;
1558 case OP_JNS:
1559 *is_inv = 1;
1560 return PFO_S;
1561 case OP_JNP:
1562 *is_inv = 1;
1563 return PFO_P;
1564 case OP_JGE:
1565 *is_inv = 1;
1566 return PFO_L;
1567 case OP_JG:
1568 *is_inv = 1;
1569 return PFO_LE;
1570
1571 case OP_ADC:
1572 case OP_SBB:
1573 return PFO_C;
1574
1575 default:
1576 ferr(po, "split_cond: bad op %d\n", op);
1577 return -1;
1578 }
1579}
1580
1581static void out_test_for_cc(char *buf, size_t buf_size,
1582 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
1583 enum opr_lenmod lmod, const char *expr)
1584{
1585 const char *cast, *scast;
1586
1587 cast = lmod_cast_u(po, lmod);
1588 scast = lmod_cast_s(po, lmod);
1589
1590 switch (pfo) {
1591 case PFO_Z:
1592 case PFO_BE: // CF=1||ZF=1; CF=0
1593 snprintf(buf, buf_size, "(%s%s %s 0)",
1594 cast, expr, is_inv ? "!=" : "==");
1595 break;
1596
1597 case PFO_S:
1598 case PFO_L: // SF!=OF; OF=0
1599 snprintf(buf, buf_size, "(%s%s %s 0)",
1600 scast, expr, is_inv ? ">=" : "<");
1601 break;
1602
1603 case PFO_LE: // ZF=1||SF!=OF; OF=0
1604 snprintf(buf, buf_size, "(%s%s %s 0)",
1605 scast, expr, is_inv ? ">" : "<=");
1606 break;
1607
1608 default:
1609 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1610 }
1611}
1612
1613static void out_cmp_for_cc(char *buf, size_t buf_size,
1614 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
1615 enum opr_lenmod lmod, const char *expr1, const char *expr2)
1616{
1617 const char *cast, *scast;
1618
1619 cast = lmod_cast_u(po, lmod);
1620 scast = lmod_cast_s(po, lmod);
1621
1622 switch (pfo) {
1623 case PFO_C:
1624 // note: must be unsigned compare
1625 snprintf(buf, buf_size, "(%s%s %s %s%s)",
1626 cast, expr1, is_inv ? ">=" : "<", cast, expr2);
1627 break;
1628
1629 case PFO_Z:
1630 snprintf(buf, buf_size, "(%s%s %s %s%s)",
1631 cast, expr1, is_inv ? "!=" : "==", cast, expr2);
1632 break;
1633
1634 case PFO_BE: // !a
1635 // note: must be unsigned compare
1636 snprintf(buf, buf_size, "(%s%s %s %s%s)",
1637 cast, expr1, is_inv ? ">" : "<=", cast, expr2);
1638 break;
1639
1640 // note: must be signed compare
1641 case PFO_S:
1642 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
1643 scast, expr1, expr2, is_inv ? ">=" : "<");
1644 break;
1645
1646 case PFO_L: // !ge
1647 snprintf(buf, buf_size, "(%s%s %s %s%s)",
1648 scast, expr1, is_inv ? ">=" : "<", scast, expr2);
1649 break;
1650
1651 case PFO_LE:
1652 snprintf(buf, buf_size, "(%s%s %s %s%s)",
1653 scast, expr1, is_inv ? ">" : "<=", scast, expr2);
1654 break;
1655
1656 default:
1657 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1658 }
1659}
1660
1661static void out_cmp_test(char *buf, size_t buf_size,
1662 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
1663{
1664 char buf1[256], buf2[256], buf3[256];
1665
1666 if (po->op == OP_TEST) {
1667 if (IS(opr_name(po, 0), opr_name(po, 1))) {
1668 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
1669 }
1670 else {
1671 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
1672 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
1673 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
1674 }
1675 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
1676 po->operand[0].lmod, buf3);
1677 }
1678 else if (po->op == OP_CMP) {
1679 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
1680 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[1]);
1681 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv,
1682 po->operand[0].lmod, buf2, buf3);
1683 }
1684 else
1685 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
1686}
1687
1688static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
1689 struct parsed_opr *popr2)
1690{
1691 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
1692 ferr(po, "missing lmod for both operands\n");
1693
1694 if (popr1->lmod == OPLM_UNSPEC)
1695 popr1->lmod = popr2->lmod;
1696 else if (popr2->lmod == OPLM_UNSPEC)
1697 popr2->lmod = popr1->lmod;
1698 else if (popr1->lmod != popr2->lmod)
1699 ferr(po, "conflicting lmods: %d vs %d\n", popr1->lmod, popr2->lmod);
1700}
1701
1702static const char *op_to_c(struct parsed_op *po)
1703{
1704 switch (po->op)
1705 {
1706 case OP_ADD:
1707 case OP_ADC:
1708 return "+";
1709 case OP_SUB:
1710 case OP_SBB:
1711 return "-";
1712 case OP_AND:
1713 return "&";
1714 case OP_OR:
1715 return "|";
1716 case OP_XOR:
1717 return "^";
1718 case OP_SHL:
1719 return "<<";
1720 case OP_SHR:
1721 return ">>";
1722 case OP_MUL:
1723 case OP_IMUL:
1724 return "*";
1725 default:
1726 ferr(po, "op_to_c was supplied with %d\n", po->op);
1727 }
1728}
1729
1730static void set_flag_no_dup(struct parsed_op *po, enum op_flags flag,
1731 enum op_flags flag_check)
1732{
1733 if (po->flags & flag)
1734 ferr(po, "flag %x already set\n", flag);
1735 if (po->flags & flag_check)
1736 ferr(po, "flag_check %x already set\n", flag_check);
1737
1738 po->flags |= flag;
1739}
1740
1741// last op in stream - unconditional branch or ret
1742#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
1743 || (ops[_i].flags & (OPF_JMP|OPF_CC)) == OPF_JMP)
1744
1745static int scan_for_pop(int i, int opcnt, const char *reg,
1746 int magic, int depth, int *maxdepth, int do_flags)
1747{
1748 struct parsed_op *po;
1749 int ret = 0;
1750 int j;
1751
1752 for (; i < opcnt; i++) {
1753 po = &ops[i];
1754 if (po->cc_scratch == magic)
1755 break; // already checked
1756 po->cc_scratch = magic;
1757
1758 if (po->flags & OPF_TAIL)
1759 return -1; // deadend
1760
1761 if ((po->flags & OPF_RMD)
1762 || (po->op == OP_PUSH && po->argnum != 0)) // arg push
1763 continue;
1764
1765 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
1766 if (po->btj != NULL) {
1767 // jumptable
1768 for (j = 0; j < po->btj->count - 1; j++) {
1769 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic,
1770 depth, maxdepth, do_flags);
1771 if (ret < 0)
1772 return ret; // dead end
1773 }
1774 // follow last jumptable entry
1775 i = po->btj->d[j].bt_i - 1;
1776 continue;
1777 }
1778
1779 if (po->bt_i < 0) {
1780 ferr(po, "dead branch\n");
1781 return -1;
1782 }
1783
1784 if (po->flags & OPF_CC) {
1785 ret |= scan_for_pop(po->bt_i, opcnt, reg, magic,
1786 depth, maxdepth, do_flags);
1787 if (ret < 0)
1788 return ret; // dead end
1789 }
1790 else {
1791 i = po->bt_i - 1;
1792 }
1793 continue;
1794 }
1795
1796 if ((po->op == OP_POP || po->op == OP_PUSH)
1797 && po->operand[0].type == OPT_REG
1798 && IS(po->operand[0].name, reg))
1799 {
1800 if (po->op == OP_PUSH) {
1801 depth++;
1802 if (depth > *maxdepth)
1803 *maxdepth = depth;
1804 if (do_flags)
1805 set_flag_no_dup(po, OPF_RSAVE, OPF_RMD);
1806 }
1807 else if (depth == 0) {
1808 if (do_flags)
1809 set_flag_no_dup(po, OPF_RMD, OPF_RSAVE);
1810 return 1;
1811 }
1812 else {
1813 depth--;
1814 if (depth < 0) // should not happen
1815 ferr(po, "fail with depth\n");
1816 if (do_flags)
1817 set_flag_no_dup(po, OPF_RSAVE, OPF_RMD);
1818 }
1819 }
1820 }
1821
1822 return ret;
1823}
1824
1825// scan for pop starting from 'ret' op (all paths)
1826static int scan_for_pop_ret(int i, int opcnt, const char *reg,
1827 int flag_set)
1828{
1829 int found = 0;
1830 int j;
1831
1832 for (; i < opcnt; i++) {
1833 if (!(ops[i].flags & OPF_TAIL))
1834 continue;
1835
1836 for (j = i - 1; j >= 0; j--) {
1837 if (ops[j].flags & OPF_RMD)
1838 continue;
1839 if (ops[j].flags & OPF_JMP)
1840 return -1;
1841
1842 if (ops[j].op == OP_POP && ops[j].operand[0].type == OPT_REG
1843 && IS(ops[j].operand[0].name, reg))
1844 {
1845 found = 1;
1846 ops[j].flags |= flag_set;
1847 break;
1848 }
1849
1850 if (g_labels[j][0] != 0)
1851 return -1;
1852 }
1853 }
1854
1855 return found ? 0 : -1;
1856}
1857
1858// is operand 'opr modified' by parsed_op 'po'?
1859static int is_opr_modified(const struct parsed_opr *opr,
1860 const struct parsed_op *po)
1861{
1862 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
1863 return 0;
1864
1865 if (opr->type == OPT_REG && po->operand[0].type == OPT_REG) {
1866 if (po->regmask_dst & (1 << opr->reg))
1867 return 1;
1868 else
1869 return 0;
1870 }
1871
1872 return IS(po->operand[0].name, opr->name);
1873}
1874
1875// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
1876static int is_any_opr_modified(const struct parsed_op *po_test,
1877 const struct parsed_op *po)
1878{
1879 int i;
1880
1881 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
1882 return 0;
1883
1884 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
1885 return 0;
1886
1887 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
1888 return 1;
1889
1890 // in reality, it can wreck any register, but in decompiled C
1891 // version it can only overwrite eax or edx:eax
1892 if (po->op == OP_CALL
1893 && ((po_test->regmask_src | po_test->regmask_dst)
1894 & ((1 << xAX)|(1 << xDX))))
1895 return 1;
1896
1897 for (i = 0; i < po_test->operand_cnt; i++)
1898 if (IS(po_test->operand[i].name, po->operand[0].name))
1899 return 1;
1900
1901 return 0;
1902}
1903
1904// scan for any po_test operand modification in range given
1905static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt)
1906{
1907 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
1908 return -1;
1909
1910 for (; i < opcnt; i++) {
1911 if (is_any_opr_modified(po_test, &ops[i]))
1912 return i;
1913 }
1914
1915 return -1;
1916}
1917
1918// scan for po_test operand[0] modification in range given
1919static int scan_for_mod_opr0(struct parsed_op *po_test,
1920 int i, int opcnt)
1921{
1922 for (; i < opcnt; i++) {
1923 if (is_opr_modified(&po_test->operand[0], &ops[i]))
1924 return i;
1925 }
1926
1927 return -1;
1928}
1929
1930static int scan_for_flag_set(int i, int *branched, int *setters,
1931 int *setter_cnt)
1932{
1933 int ret;
1934
1935 while (i >= 0) {
1936 if (g_labels[i][0] != 0) {
1937 *branched = 1;
1938 if (g_label_refs[i].next != NULL)
1939 return -1;
1940 if (i > 0 && LAST_OP(i - 1)) {
1941 i = g_label_refs[i].i;
1942 continue;
1943 }
1944 ret = scan_for_flag_set(g_label_refs[i].i, branched,
1945 setters, setter_cnt);
1946 if (ret < 0)
1947 return ret;
1948 }
1949 i--;
1950
1951 if (ops[i].flags & OPF_FLAGS) {
1952 setters[*setter_cnt] = i;
1953 (*setter_cnt)++;
1954 return 0;
1955 }
1956
1957 if ((ops[i].flags & OPF_JMP) && !(ops[i].flags & OPF_CC))
1958 return -1;
1959 }
1960
1961 return -1;
1962}
1963
1964// scan back for cdq, if anything modifies edx, fail
1965static int scan_for_cdq_edx(int i)
1966{
1967 for (; i >= 0; i--) {
1968 if (ops[i].op == OP_CDQ)
1969 return i;
1970
1971 if (ops[i].regmask_dst & (1 << xDX))
1972 return -1;
1973 if (g_labels[i][0] != 0)
1974 return -1;
1975 }
1976
1977 return -1;
1978}
1979
1980static int scan_for_reg_clear(int i, int reg)
1981{
1982 for (; i >= 0; i--) {
1983 if (ops[i].op == OP_XOR
1984 && ops[i].operand[0].lmod == OPLM_DWORD
1985 && ops[i].operand[0].reg == ops[i].operand[1].reg
1986 && ops[i].operand[0].reg == reg)
1987 return i;
1988
1989 if (ops[i].regmask_dst & (1 << reg))
1990 return -1;
1991 if (g_labels[i][0] != 0)
1992 return -1;
1993 }
1994
1995 return -1;
1996}
1997
1998// scan for positive, constant esp adjust
1999static int scan_for_esp_adjust(int i, int opcnt, int *adj)
2000{
2001 struct parsed_op *po;
2002 *adj = 0;
2003
2004 for (; i < opcnt; i++) {
2005 po = &ops[i];
2006
2007 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
2008 if (po->operand[1].type != OPT_CONST)
2009 ferr(&ops[i], "non-const esp adjust?\n");
2010 *adj += po->operand[1].val;
2011 if (*adj & 3)
2012 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
2013 return i;
2014 }
2015 else if (po->op == OP_PUSH)
2016 *adj -= lmod_bytes(po, po->operand[0].lmod);
2017 else if (po->op == OP_POP)
2018 *adj += lmod_bytes(po, po->operand[0].lmod);
2019 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
2020 if (po->op != OP_CALL)
2021 return -1;
2022 if (po->operand[0].type != OPT_LABEL)
2023 return -1;
2024 // TODO: should only allow combining __cdecl calls..
2025 }
2026
2027 if (g_labels[i][0] != 0)
2028 return -1;
2029 }
2030
2031 return -1;
2032}
2033
2034static int collect_call_args(struct parsed_op *po, int i,
2035 struct parsed_proto *pp, int *save_arg_vars, int arg,
2036 int need_op_saving, int branched)
2037{
2038 struct parsed_proto *pp_tmp;
2039 struct label_ref *lr;
2040 int need_to_save_current;
2041 int ret = 0;
2042 int j;
2043
2044 if (i < 0)
2045 ferr(po, "no refs for '%s'?\n", g_labels[i]);
2046
2047 for (; arg < pp->argc; arg++)
2048 if (pp->arg[arg].reg == NULL)
2049 break;
2050
2051 for (j = i; j >= 0 && arg < pp->argc; )
2052 {
2053 if (g_labels[j][0] != 0) {
2054 branched = 1;
2055 lr = &g_label_refs[j];
2056 if (lr->next != NULL)
2057 need_op_saving = 1;
2058 for (; lr->next; lr = lr->next)
2059 ret |= collect_call_args(po, lr->i, pp, save_arg_vars,
2060 arg, need_op_saving, branched);
2061
2062 if (j > 0 && LAST_OP(j - 1)) {
2063 // follow last branch in reverse
2064 j = lr->i;
2065 continue;
2066 }
2067 need_op_saving = 1;
2068 ret |= collect_call_args(po, lr->i, pp, save_arg_vars,
2069 arg, need_op_saving, branched);
2070 }
2071 j--;
2072
2073 if (ops[j].op == OP_CALL)
2074 {
2075 pp_tmp = ops[j].datap;
2076 if (pp_tmp == NULL)
2077 ferr(po, "arg collect hit unparsed call\n");
2078 if (branched && pp_tmp->argc_stack > 0)
2079 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
2080 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
2081 }
2082 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP) {
2083 ferr(po, "arg collect %d/%d hit esp adjust\n",
2084 arg, pp->argc);
2085 }
2086 else if (ops[j].op == OP_POP) {
2087 ferr(po, "arg collect %d/%d hit pop\n", arg, pp->argc);
2088 }
2089 else if (LAST_OP(j)) {
2090 break;
2091 }
2092 else if ((ops[j].flags & (OPF_JMP|OPF_CC)) == (OPF_JMP|OPF_CC))
2093 {
2094 branched = 1;
2095 }
2096 else if (ops[j].op == OP_PUSH && !(ops[j].flags & OPF_FARG))
2097 {
2098 pp->arg[arg].datap = &ops[j];
2099 need_to_save_current = 0;
2100 if (!need_op_saving) {
2101 ret = scan_for_mod(&ops[j], j + 1, i);
2102 need_to_save_current = (ret >= 0);
2103 }
2104 if (need_op_saving || need_to_save_current) {
2105 // mark this push as one that needs operand saving
2106 ops[j].flags &= ~OPF_RMD;
2107 if (ops[j].argnum == 0) {
2108 ops[j].argnum = arg + 1;
2109 *save_arg_vars |= 1 << arg;
2110 }
2111 else if (ops[j].argnum < arg + 1)
2112 ferr(&ops[j], "argnum conflict (%d<%d) for '%s'\n",
2113 ops[j].argnum, arg + 1, pp->name);
2114 }
2115 else if (ops[j].argnum == 0)
2116 ops[j].flags |= OPF_RMD;
2117
2118 // some PUSHes are reused by calls on multiple branches,
2119 // but that can't happen if we didn't branch, so they
2120 // can be removed from future searches (handles nested calls)
2121 if (!branched)
2122 ops[j].flags |= OPF_FARG;
2123
2124 // next arg
2125 for (arg++; arg < pp->argc; arg++)
2126 if (pp->arg[arg].reg == NULL)
2127 break;
2128 }
2129 }
2130
2131 if (arg < pp->argc) {
2132 ferr(po, "arg collect failed for '%s': %d/%d\n",
2133 pp->name, arg, pp->argc);
2134 ret = -1;
2135 }
2136 return ret;
2137}
2138
2139static void add_label_ref(struct label_ref *lr, int op_i)
2140{
2141 struct label_ref *lr_new;
2142
2143 if (lr->i == -1) {
2144 lr->i = op_i;
2145 return;
2146 }
2147
2148 lr_new = calloc(1, sizeof(*lr_new));
2149 lr_new->i = op_i;
2150 lr->next = lr_new;
2151}
2152
2153static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
2154{
2155 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
2156 struct parsed_opr *last_arith_dst = NULL;
2157 char buf1[256], buf2[256], buf3[256], cast[64];
2158 struct parsed_proto *pp, *pp_tmp;
2159 struct parsed_data *pd;
2160 const char *tmpname;
2161 enum parsed_flag_op pfo;
2162 int save_arg_vars = 0;
2163 int cmp_result_vars = 0;
2164 int need_mul_var = 0;
2165 int had_decl = 0;
2166 int label_pending = 0;
2167 int regmask_save = 0;
2168 int regmask_arg = 0;
2169 int regmask = 0;
2170 int pfomask = 0;
2171 int found = 0;
2172 int depth = 0;
2173 int no_output;
2174 int i, j, l;
2175 int dummy;
2176 int arg;
2177 int reg;
2178 int ret;
2179
2180 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
2181
2182 ret = proto_parse(fhdr, funcn, &g_func_pp);
2183 if (ret)
2184 ferr(ops, "proto_parse failed for '%s'\n", funcn);
2185
2186 fprintf(fout, "%s ", g_func_pp.ret_type.name);
2187 if (g_ida_func_attr & IDAFA_NORETURN)
2188 fprintf(fout, "noreturn ");
2189 fprintf(fout, "%s(", funcn);
2190 for (i = 0; i < g_func_pp.argc; i++) {
2191 if (i > 0)
2192 fprintf(fout, ", ");
2193 fprintf(fout, "%s a%d", g_func_pp.arg[i].type.name, i + 1);
2194 }
2195 if (g_func_pp.is_vararg) {
2196 if (i > 0)
2197 fprintf(fout, ", ");
2198 fprintf(fout, "...");
2199 }
2200 fprintf(fout, ")\n{\n");
2201
2202 // pass1:
2203 // - handle ebp/esp frame, remove ops related to it
2204 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
2205 && ops[1].op == OP_MOV
2206 && IS(opr_name(&ops[1], 0), "ebp")
2207 && IS(opr_name(&ops[1], 1), "esp"))
2208 {
2209 int ecx_push = 0;
2210
2211 g_bp_frame = 1;
2212 ops[0].flags |= OPF_RMD;
2213 ops[1].flags |= OPF_RMD;
2214 i = 2;
2215
2216 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
2217 g_stack_fsz = opr_const(&ops[2], 1);
2218 ops[2].flags |= OPF_RMD;
2219 i++;
2220 }
2221 else {
2222 // another way msvc builds stack frame..
2223 i = 2;
2224 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
2225 g_stack_fsz += 4;
2226 ops[i].flags |= OPF_RMD;
2227 ecx_push++;
2228 i++;
2229 }
2230 // and another way..
2231 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
2232 && ops[i].operand[1].type == OPT_CONST
2233 && ops[i + 1].op == OP_CALL
2234 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
2235 {
2236 g_stack_fsz += ops[i].operand[1].val;
2237 ops[i].flags |= OPF_RMD;
2238 i++;
2239 ops[i].flags |= OPF_RMD;
2240 i++;
2241 }
2242 }
2243
2244 found = 0;
2245 do {
2246 for (; i < opcnt; i++)
2247 if (ops[i].op == OP_RET)
2248 break;
2249 if (i == opcnt && (ops[i - 1].flags & OPF_JMP) && found)
2250 break;
2251
2252 if (ops[i - 1].op == OP_POP && IS(opr_name(&ops[i - 1], 0), "ebp"))
2253 ops[i - 1].flags |= OPF_RMD;
2254 else if (!(g_ida_func_attr & IDAFA_NORETURN))
2255 ferr(&ops[i - 1], "'pop ebp' expected\n");
2256
2257 if (g_stack_fsz != 0) {
2258 if (ops[i - 2].op == OP_MOV
2259 && IS(opr_name(&ops[i - 2], 0), "esp")
2260 && IS(opr_name(&ops[i - 2], 1), "ebp"))
2261 {
2262 ops[i - 2].flags |= OPF_RMD;
2263 }
2264 else if (!(g_ida_func_attr & IDAFA_NORETURN))
2265 ferr(&ops[i - 2], "esp restore expected\n");
2266
2267 if (ecx_push && ops[i - 3].op == OP_POP
2268 && IS(opr_name(&ops[i - 3], 0), "ecx"))
2269 {
2270 ferr(&ops[i - 3], "unexpected ecx pop\n");
2271 }
2272 }
2273
2274 found = 1;
2275 i++;
2276 } while (i < opcnt);
2277 }
2278 else {
2279 for (i = 0; i < opcnt; i++) {
2280 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
2281 break;
2282 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
2283 && ops[i].operand[1].type == OPT_CONST)
2284 {
2285 g_sp_frame = 1;
2286 break;
2287 }
2288 }
2289
2290 if (g_sp_frame)
2291 {
2292 g_stack_fsz = ops[i].operand[1].val;
2293 ops[i].flags |= OPF_RMD;
2294
2295 i++;
2296 do {
2297 for (; i < opcnt; i++)
2298 if (ops[i].op == OP_RET)
2299 break;
2300 if (ops[i - 1].op != OP_ADD
2301 || !IS(opr_name(&ops[i - 1], 0), "esp")
2302 || ops[i - 1].operand[1].type != OPT_CONST
2303 || ops[i - 1].operand[1].val != g_stack_fsz)
2304 ferr(&ops[i - 1], "'add esp' expected\n");
2305 ops[i - 1].flags |= OPF_RMD;
2306
2307 i++;
2308 } while (i < opcnt);
2309 }
2310 }
2311
2312 // pass2:
2313 // - resolve all branches
2314 for (i = 0; i < opcnt; i++)
2315 {
2316 po = &ops[i];
2317 po->bt_i = -1;
2318 po->btj = NULL;
2319
2320 if ((po->flags & OPF_RMD) || !(po->flags & OPF_JMP)
2321 || po->op == OP_CALL || po->op == OP_RET)
2322 continue;
2323
2324 if (po->operand[0].type == OPT_REGMEM) {
2325 char *p = strchr(po->operand[0].name, '[');
2326 if (p == NULL)
2327 ferr(po, "unhandled indirect branch\n");
2328 ret = p - po->operand[0].name;
2329 strncpy(buf1, po->operand[0].name, ret);
2330 buf1[ret] = 0;
2331
2332 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
2333 if (IS(g_func_pd[j].label, buf1)) {
2334 pd = &g_func_pd[j];
2335 break;
2336 }
2337 }
2338 if (pd == NULL)
2339 ferr(po, "label '%s' not parsed?\n", buf1);
2340 if (pd->type != OPT_OFFSET)
2341 ferr(po, "label '%s' with non-offset data?\n", buf1);
2342
2343 // find all labels, link
2344 for (j = 0; j < pd->count; j++) {
2345 for (l = 0; l < opcnt; l++) {
2346 if (g_labels[l][0] && IS(g_labels[l], pd->d[j].u.label)) {
2347 add_label_ref(&g_label_refs[l], i);
2348 pd->d[j].bt_i = l;
2349 break;
2350 }
2351 }
2352 }
2353
2354 po->btj = pd;
2355 continue;
2356 }
2357
2358 for (l = 0; l < opcnt; l++) {
2359 if (g_labels[l][0] && IS(po->operand[0].name, g_labels[l])) {
2360 add_label_ref(&g_label_refs[l], i);
2361 po->bt_i = l;
2362 break;
2363 }
2364 }
2365
2366 if (po->bt_i != -1)
2367 continue;
2368
2369 if (po->operand[0].type == OPT_LABEL) {
2370 // assume tail call
2371 po->op = OP_CALL;
2372 po->flags |= OPF_TAIL;
2373 continue;
2374 }
2375
2376 ferr(po, "unhandled branch\n");
2377 }
2378
2379 // pass3:
2380 // - process calls
2381 for (i = 0; i < opcnt; i++)
2382 {
2383 po = &ops[i];
2384 if (po->flags & OPF_RMD)
2385 continue;
2386
2387 if (po->op == OP_CALL)
2388 {
2389 pp = calloc(1, sizeof(*pp));
2390 my_assert_not(pp, NULL);
2391 tmpname = opr_name(po, 0);
2392 if (po->operand[0].type != OPT_LABEL)
2393 {
2394 ret = scan_for_esp_adjust(i + 1, opcnt, &j);
2395 if (ret < 0)
2396 ferr(po, "non-__cdecl indirect call unhandled yet\n");
2397 j /= 4;
2398 if (j > ARRAY_SIZE(pp->arg))
2399 ferr(po, "esp adjust too large?\n");
2400 pp->ret_type.name = strdup("int");
2401 pp->argc = pp->argc_stack = j;
2402 for (arg = 0; arg < pp->argc; arg++)
2403 pp->arg[arg].type.name = strdup("int");
2404 }
2405 else {
2406 ret = proto_parse(fhdr, tmpname, pp);
2407 if (ret)
2408 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
2409 }
2410
2411 // look for and make use of esp adjust
2412 ret = -1;
2413 if (!pp->is_stdcall && pp->argc_stack > 0)
2414 ret = scan_for_esp_adjust(i + 1, opcnt, &j);
2415 if (ret >= 0) {
2416 if (pp->is_vararg) {
2417 if (j / 4 < pp->argc_stack)
2418 ferr(po, "esp adjust is too small: %x < %x\n",
2419 j, pp->argc_stack * 4);
2420 // modify pp to make it have varargs as normal args
2421 arg = pp->argc;
2422 pp->argc += j / 4 - pp->argc_stack;
2423 for (; arg < pp->argc; arg++) {
2424 pp->arg[arg].type.name = strdup("int");
2425 pp->argc_stack++;
2426 }
2427 if (pp->argc > ARRAY_SIZE(pp->arg))
2428 ferr(po, "too many args for '%s'\n", tmpname);
2429 }
2430 if (pp->argc_stack != j / 4)
2431 ferr(po, "stack tracking failed for '%s': %x %x\n",
2432 tmpname, pp->argc_stack * 4, j);
2433
2434 ops[ret].flags |= OPF_RMD;
2435 // a bit of a hack, but deals with use of
2436 // single adj for multiple calls
2437 ops[ret].operand[1].val -= j;
2438 }
2439 else if (pp->is_vararg)
2440 ferr(po, "missing esp_adjust for vararg func '%s'\n",
2441 pp->name);
2442
2443 // can't call functions with non-__cdecl callbacks yet
2444 for (arg = 0; arg < pp->argc; arg++) {
2445 if (pp->arg[arg].fptr != NULL) {
2446 pp_tmp = pp->arg[arg].fptr;
2447 if (pp_tmp->is_stdcall || pp_tmp->argc != pp_tmp->argc_stack)
2448 ferr(po, "'%s' has a non-__cdecl callback\n", tmpname);
2449 }
2450 }
2451
2452 collect_call_args(po, i, pp, &save_arg_vars, 0, 0, 0);
2453
2454 if (strstr(pp->ret_type.name, "int64"))
2455 need_mul_var = 1;
2456 po->datap = pp;
2457 }
2458 }
2459
2460 // pass4:
2461 // - find POPs for PUSHes, rm both
2462 // - scan for all used registers
2463 // - find flag set ops for their users
2464 // - declare indirect functions
2465 for (i = 0; i < opcnt; i++) {
2466 po = &ops[i];
2467 if (po->flags & OPF_RMD)
2468 continue;
2469
2470 if (po->op == OP_PUSH
2471 && po->argnum == 0 && !(po->flags & OPF_RSAVE)
2472 && po->operand[0].type == OPT_REG)
2473 {
2474 reg = po->operand[0].reg;
2475 if (reg < 0)
2476 ferr(po, "reg not set for push?\n");
2477
2478 depth = 0;
2479 ret = scan_for_pop(i + 1, opcnt,
2480 po->operand[0].name, i + opcnt, 0, &depth, 0);
2481 if (ret == 1) {
2482 if (depth > 1)
2483 ferr(po, "too much depth: %d\n", depth);
2484 if (depth > 0)
2485 regmask_save |= 1 << reg;
2486
2487 po->flags |= OPF_RMD;
2488 scan_for_pop(i + 1, opcnt, po->operand[0].name,
2489 i + opcnt * 2, 0, &depth, 1);
2490 continue;
2491 }
2492 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
2493 if (ret == 0) {
2494 arg = OPF_RMD;
2495 if (regmask & (1 << reg)) {
2496 if (regmask_save & (1 << reg))
2497 ferr(po, "%s already saved?\n", po->operand[0].name);
2498 arg = OPF_RSAVE;
2499 }
2500 po->flags |= arg;
2501 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg);
2502 continue;
2503 }
2504 }
2505
2506 regmask |= po->regmask_src | po->regmask_dst;
2507
2508 if (po->flags & OPF_CC)
2509 {
2510 int setters[16], cnt = 0, branched = 0;
2511
2512 ret = scan_for_flag_set(i, &branched, setters, &cnt);
2513 if (ret < 0 || cnt <= 0)
2514 ferr(po, "unable to trace flag setter(s)\n");
2515 if (cnt > ARRAY_SIZE(setters))
2516 ferr(po, "too many flag setters\n");
2517
2518 pfo = split_cond(po, po->op, &dummy);
2519 for (j = 0; j < cnt; j++)
2520 {
2521 tmp_op = &ops[setters[j]]; // flag setter
2522 pfomask = 0;
2523
2524 // to get nicer code, we try to delay test and cmp;
2525 // if we can't because of operand modification, or if we
2526 // have math op, or branch, make it calculate flags explicitly
2527 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP) {
2528 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i) >= 0)
2529 pfomask = 1 << pfo;
2530 }
2531 else if (tmp_op->op == OP_CMPS) {
2532 pfomask = 1 << PFO_Z;
2533 }
2534 else {
2535 if ((pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P)
2536 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
2537 pfomask = 1 << pfo;
2538 }
2539 if (pfomask) {
2540 tmp_op->pfomask |= pfomask;
2541 cmp_result_vars |= pfomask;
2542 // note: may overwrite, currently not a problem
2543 po->datap = tmp_op;
2544 }
2545 }
2546
2547 if (po->op == OP_ADC || po->op == OP_SBB)
2548 cmp_result_vars |= 1 << PFO_C;
2549 }
2550 else if (po->op == OP_MUL
2551 || (po->op == OP_IMUL && po->operand_cnt == 1))
2552 {
2553 need_mul_var = 1;
2554 }
2555 else if (po->op == OP_CALL && po->operand[0].type != OPT_LABEL) {
2556 pp = po->datap;
2557 my_assert_not(pp, NULL);
2558 fprintf(fout, " %s (*icall%d)(", pp->ret_type.name, i);
2559 for (j = 0; j < pp->argc; j++) {
2560 if (j > 0)
2561 fprintf(fout, ", ");
2562 fprintf(fout, "%s a%d", pp->arg[j].type.name, j + 1);
2563 }
2564 fprintf(fout, ");\n");
2565 }
2566 }
2567
2568 // output LUTs/jumptables
2569 for (i = 0; i < g_func_pd_cnt; i++) {
2570 pd = &g_func_pd[i];
2571 fprintf(fout, " static const ");
2572 if (pd->type == OPT_OFFSET) {
2573 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
2574
2575 for (j = 0; j < pd->count; j++) {
2576 if (j > 0)
2577 fprintf(fout, ", ");
2578 fprintf(fout, "&&%s", pd->d[j].u.label);
2579 }
2580 }
2581 else {
2582 fprintf(fout, "%s %s[] =\n { ",
2583 lmod_type_u(ops, pd->lmod), pd->label);
2584
2585 for (j = 0; j < pd->count; j++) {
2586 if (j > 0)
2587 fprintf(fout, ", ");
2588 fprintf(fout, "%u", pd->d[j].u.val);
2589 }
2590 }
2591 fprintf(fout, " };\n");
2592 }
2593
2594 // declare stack frame, va_arg
2595 if (g_stack_fsz)
2596 fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
2597 (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz);
2598
2599 if (g_func_pp.is_vararg)
2600 fprintf(fout, " va_list ap;\n");
2601
2602 // declare arg-registers
2603 for (i = 0; i < g_func_pp.argc; i++) {
2604 if (g_func_pp.arg[i].reg != NULL) {
2605 reg = char_array_i(regs_r32,
2606 ARRAY_SIZE(regs_r32), g_func_pp.arg[i].reg);
2607 if (reg < 0)
2608 ferr(ops, "arg '%s' is not a reg?\n", g_func_pp.arg[i].reg);
2609
2610 regmask_arg |= 1 << reg;
2611 fprintf(fout, " u32 %s = (u32)a%d;\n",
2612 g_func_pp.arg[i].reg, i + 1);
2613 had_decl = 1;
2614 }
2615 }
2616
2617 // declare other regs - special case for eax
2618 if (!((regmask | regmask_arg) & 1)
2619 && !IS(g_func_pp.ret_type.name, "void"))
2620 {
2621 fprintf(fout, " u32 eax = 0;\n");
2622 had_decl = 1;
2623 }
2624
2625 regmask &= ~regmask_arg;
2626 regmask &= ~(1 << xSP);
2627 if (g_bp_frame)
2628 regmask &= ~(1 << xBP);
2629 if (regmask) {
2630 for (reg = 0; reg < 8; reg++) {
2631 if (regmask & (1 << reg)) {
2632 fprintf(fout, " u32 %s;\n", regs_r32[reg]);
2633 had_decl = 1;
2634 }
2635 }
2636 }
2637
2638 if (regmask_save) {
2639 for (reg = 0; reg < 8; reg++) {
2640 if (regmask_save & (1 << reg)) {
2641 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
2642 had_decl = 1;
2643 }
2644 }
2645 }
2646
2647 if (save_arg_vars) {
2648 for (reg = 0; reg < 32; reg++) {
2649 if (save_arg_vars & (1 << reg)) {
2650 fprintf(fout, " u32 s_a%d;\n", reg + 1);
2651 had_decl = 1;
2652 }
2653 }
2654 }
2655
2656 if (cmp_result_vars) {
2657 for (i = 0; i < 8; i++) {
2658 if (cmp_result_vars & (1 << i)) {
2659 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
2660 had_decl = 1;
2661 }
2662 }
2663 }
2664
2665 if (need_mul_var) {
2666 fprintf(fout, " u64 mul_tmp;\n");
2667 had_decl = 1;
2668 }
2669
2670 if (had_decl)
2671 fprintf(fout, "\n");
2672
2673 if (g_func_pp.is_vararg) {
2674 if (g_func_pp.argc_stack == 0)
2675 ferr(ops, "vararg func without stack args?\n");
2676 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp.argc);
2677 }
2678
2679 // output ops
2680 for (i = 0; i < opcnt; i++)
2681 {
2682 if (g_labels[i][0] != 0 && g_label_refs[i].i != -1) {
2683 fprintf(fout, "\n%s:\n", g_labels[i]);
2684 label_pending = 1;
2685
2686 delayed_flag_op = NULL;
2687 last_arith_dst = NULL;
2688 }
2689
2690 po = &ops[i];
2691 if (po->flags & OPF_RMD)
2692 continue;
2693
2694 no_output = 0;
2695
2696 #define assert_operand_cnt(n_) \
2697 if (po->operand_cnt != n_) \
2698 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
2699
2700 // conditional/flag using op?
2701 if (po->flags & OPF_CC)
2702 {
2703 int is_delayed = 0;
2704 int is_inv = 0;
2705
2706 pfo = split_cond(po, po->op, &is_inv);
2707
2708 // we go through all this trouble to avoid using parsed_flag_op,
2709 // which makes generated code much nicer
2710 if (delayed_flag_op != NULL)
2711 {
2712 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op, pfo, is_inv);
2713 is_delayed = 1;
2714 }
2715 else if (last_arith_dst != NULL
2716 && (pfo == PFO_Z || pfo == PFO_S || pfo == PFO_P))
2717 {
2718 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
2719 out_test_for_cc(buf1, sizeof(buf1), po, pfo, is_inv,
2720 last_arith_dst->lmod, buf3);
2721 is_delayed = 1;
2722 }
2723 else if (po->datap != NULL) {
2724 // use preprocessed flag calc results
2725 tmp_op = po->datap;
2726 if (!tmp_op || !(tmp_op->pfomask & (1 << pfo)))
2727 ferr(po, "not prepared for pfo %d\n", pfo);
2728
2729 // note: is_inv was not yet applied
2730 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
2731 is_inv ? "!" : "", parsed_flag_op_names[pfo]);
2732 }
2733 else {
2734 ferr(po, "all methods of finding comparison failed\n");
2735 }
2736
2737 if (po->flags & OPF_JMP) {
2738 fprintf(fout, " if %s\n", buf1);
2739 }
2740 else if (po->op == OP_ADC || po->op == OP_SBB) {
2741 if (is_delayed)
2742 fprintf(fout, " cond_%s = %s;\n",
2743 parsed_flag_op_names[pfo], buf1);
2744 }
2745 else if (po->flags & OPF_DATA) { // SETcc
2746 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
2747 fprintf(fout, " %s = %s;", buf2, buf1);
2748 }
2749 else {
2750 ferr(po, "unhandled conditional op\n");
2751 }
2752 }
2753
2754 pfomask = po->pfomask;
2755
2756 switch (po->op)
2757 {
2758 case OP_MOV:
2759 assert_operand_cnt(2);
2760 propagate_lmod(po, &po->operand[0], &po->operand[1]);
2761 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
2762 fprintf(fout, " %s = %s;", buf1,
2763 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
2764 po->operand[0].is_ptr ? "(void *)" : "", 0));
2765 break;
2766
2767 case OP_LEA:
2768 assert_operand_cnt(2);
2769 po->operand[1].lmod = OPLM_DWORD; // always
2770 fprintf(fout, " %s = %s;",
2771 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2772 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
2773 NULL, 1));
2774 break;
2775
2776 case OP_MOVZX:
2777 assert_operand_cnt(2);
2778 fprintf(fout, " %s = %s;",
2779 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2780 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2781 break;
2782
2783 case OP_MOVSX:
2784 assert_operand_cnt(2);
2785 switch (po->operand[1].lmod) {
2786 case OPLM_BYTE:
2787 strcpy(buf3, "(s8)");
2788 break;
2789 case OPLM_WORD:
2790 strcpy(buf3, "(s16)");
2791 break;
2792 default:
2793 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
2794 }
2795 fprintf(fout, " %s = %s%s;",
2796 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2797 buf3,
2798 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2799 break;
2800
2801 case OP_NOT:
2802 assert_operand_cnt(1);
2803 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
2804 fprintf(fout, " %s = ~%s;", buf1, buf1);
2805 break;
2806
2807 case OP_CDQ:
2808 assert_operand_cnt(2);
2809 fprintf(fout, " %s = (s32)%s >> 31;",
2810 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2811 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2812 strcpy(g_comment, "cdq");
2813 break;
2814
2815 case OP_STOS:
2816 // assumes DF=0
2817 assert_operand_cnt(3);
2818 if (po->flags & OPF_REP) {
2819 fprintf(fout, " for (; ecx != 0; ecx--, edi += %d)\n",
2820 lmod_bytes(po, po->operand[0].lmod));
2821 fprintf(fout, " %sedi = eax;",
2822 lmod_cast_u_ptr(po, po->operand[0].lmod));
2823 strcpy(g_comment, "rep stos");
2824 }
2825 else {
2826 fprintf(fout, " %sedi = eax; edi += %d;",
2827 lmod_cast_u_ptr(po, po->operand[0].lmod),
2828 lmod_bytes(po, po->operand[0].lmod));
2829 strcpy(g_comment, "stos");
2830 }
2831 break;
2832
2833 case OP_MOVS:
2834 // assumes DF=0
2835 assert_operand_cnt(3);
2836 j = lmod_bytes(po, po->operand[0].lmod);
2837 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
2838 if (po->flags & OPF_REP) {
2839 fprintf(fout,
2840 " for (; ecx != 0; ecx--, edi += %d, esi += %d)\n",
2841 j, j);
2842 fprintf(fout,
2843 " %sedi = %sesi;", buf1, buf1);
2844 strcpy(g_comment, "rep movs");
2845 }
2846 else {
2847 fprintf(fout, " %sedi = %sesi; edi += %d; esi += %d;",
2848 buf1, buf1, j, j);
2849 strcpy(g_comment, "movs");
2850 }
2851 break;
2852
2853 case OP_CMPS:
2854 // assumes DF=0
2855 // repe ~ repeat while ZF=1
2856 assert_operand_cnt(3);
2857 j = lmod_bytes(po, po->operand[0].lmod);
2858 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
2859 if (po->flags & OPF_REP) {
2860 fprintf(fout,
2861 " for (; ecx != 0; ecx--, edi += %d, esi += %d)\n",
2862 j, j);
2863 fprintf(fout,
2864 " if ((cond_z = (%sedi == %sesi)) %s 0)\n",
2865 buf1, buf1, (po->flags & OPF_REPZ) ? "==" : "!=");
2866 fprintf(fout,
2867 " break;");
2868 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
2869 (po->flags & OPF_REPZ) ? "e" : "ne");
2870 }
2871 else {
2872 fprintf(fout,
2873 " cond_z = (%sedi = %sesi); edi += %d; esi += %d;",
2874 buf1, buf1, j, j);
2875 strcpy(g_comment, "cmps");
2876 }
2877 pfomask &= ~(1 << PFO_Z);
2878 last_arith_dst = NULL;
2879 delayed_flag_op = NULL;
2880 break;
2881
2882 // arithmetic w/flags
2883 case OP_ADD:
2884 case OP_SUB:
2885 case OP_AND:
2886 case OP_OR:
2887 propagate_lmod(po, &po->operand[0], &po->operand[1]);
2888 // fallthrough
2889 case OP_SHL:
2890 case OP_SHR:
2891 dualop_arith:
2892 assert_operand_cnt(2);
2893 fprintf(fout, " %s %s= %s;",
2894 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2895 op_to_c(po),
2896 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2897 last_arith_dst = &po->operand[0];
2898 delayed_flag_op = NULL;
2899 break;
2900
2901 case OP_SAR:
2902 assert_operand_cnt(2);
2903 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
2904 fprintf(fout, " %s = %s%s >> %s;", buf1,
2905 lmod_cast_s(po, po->operand[0].lmod), buf1,
2906 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2907 last_arith_dst = &po->operand[0];
2908 delayed_flag_op = NULL;
2909 break;
2910
2911 case OP_ROL:
2912 case OP_ROR:
2913 assert_operand_cnt(2);
2914 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
2915 if (po->operand[1].type == OPT_CONST) {
2916 j = po->operand[1].val;
2917 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
2918 fprintf(fout, po->op == OP_ROL ?
2919 " %s = (%s << %d) | (%s >> %d);" :
2920 " %s = (%s >> %d) | (%s << %d);",
2921 buf1, buf1, j, buf1,
2922 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
2923 }
2924 else
2925 ferr(po, "TODO\n");
2926 last_arith_dst = &po->operand[0];
2927 delayed_flag_op = NULL;
2928 break;
2929
2930 case OP_XOR:
2931 assert_operand_cnt(2);
2932 propagate_lmod(po, &po->operand[0], &po->operand[1]);
2933 if (IS(opr_name(po, 0), opr_name(po, 1))) {
2934 // special case for XOR
2935 fprintf(fout, " %s = 0;",
2936 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
2937 last_arith_dst = &po->operand[0];
2938 delayed_flag_op = NULL;
2939 break;
2940 }
2941 goto dualop_arith;
2942
2943 case OP_ADC:
2944 case OP_SBB:
2945 assert_operand_cnt(2);
2946 propagate_lmod(po, &po->operand[0], &po->operand[1]);
2947 fprintf(fout, " %s %s= %s + cond_c;",
2948 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2949 op_to_c(po),
2950 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2951 last_arith_dst = &po->operand[0];
2952 delayed_flag_op = NULL;
2953 break;
2954
2955 case OP_INC:
2956 case OP_DEC:
2957 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
2958 if (po->operand[0].type == OPT_REG) {
2959 strcpy(buf2, po->op == OP_INC ? "++" : "--");
2960 fprintf(fout, " %s%s;", buf1, buf2);
2961 }
2962 else {
2963 strcpy(buf2, po->op == OP_INC ? "+" : "-");
2964 fprintf(fout, " %s %s= 1;", buf1, buf2);
2965 }
2966 last_arith_dst = &po->operand[0];
2967 delayed_flag_op = NULL;
2968 break;
2969
2970 case OP_NEG:
2971 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
2972 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
2973 fprintf(fout, " %s = -%s%s;", buf1,
2974 lmod_cast_s(po, po->operand[0].lmod), buf2);
2975 last_arith_dst = &po->operand[0];
2976 delayed_flag_op = NULL;
2977 if (pfomask & (1 << PFO_C)) {
2978 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
2979 pfomask &= ~(1 << PFO_C);
2980 }
2981 break;
2982
2983 case OP_IMUL:
2984 if (po->operand_cnt == 2) {
2985 propagate_lmod(po, &po->operand[0], &po->operand[1]);
2986 goto dualop_arith;
2987 }
2988 if (po->operand_cnt == 3)
2989 ferr(po, "TODO imul3\n");
2990 // fallthrough
2991 case OP_MUL:
2992 assert_operand_cnt(1);
2993 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
2994 fprintf(fout, " mul_tmp = %seax * %s%s;\n", buf1, buf1,
2995 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
2996 fprintf(fout, " edx = mul_tmp >> 32;\n");
2997 fprintf(fout, " eax = mul_tmp;");
2998 last_arith_dst = NULL;
2999 delayed_flag_op = NULL;
3000 break;
3001
3002 case OP_DIV:
3003 case OP_IDIV:
3004 assert_operand_cnt(1);
3005 if (po->operand[0].lmod != OPLM_DWORD)
3006 ferr(po, "unhandled lmod %d\n", po->operand[0].lmod);
3007
3008 // 32bit division is common, look for it
3009 if (po->op == OP_DIV)
3010 ret = scan_for_reg_clear(i - 1, xDX);
3011 else
3012 ret = scan_for_cdq_edx(i - 1);
3013 if (ret >= 0) {
3014 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
3015 strcpy(buf2, lmod_cast(po, po->operand[0].lmod,
3016 po->op == OP_IDIV));
3017 fprintf(fout, " edx = %seax %% %s%s;\n", buf2, buf2, buf1);
3018 fprintf(fout, " eax = %seax / %s%s;", buf2, buf2, buf1);
3019 }
3020 else
3021 ferr(po, "TODO 64bit divident\n");
3022 last_arith_dst = NULL;
3023 delayed_flag_op = NULL;
3024 break;
3025
3026 case OP_TEST:
3027 case OP_CMP:
3028 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3029 if (pfomask != 0) {
3030 for (j = 0; j < 8; j++) {
3031 if (pfomask & (1 << j)) {
3032 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
3033 fprintf(fout, " cond_%s = %s;",
3034 parsed_flag_op_names[j], buf1);
3035 }
3036 }
3037 pfomask = 0;
3038 }
3039 else
3040 no_output = 1;
3041 last_arith_dst = NULL;
3042 delayed_flag_op = po;
3043 break;
3044
3045 // note: we reuse OP_Jcc for SETcc, only flags differ
3046 case OP_JO ... OP_JG:
3047 if (po->flags & OPF_JMP)
3048 fprintf(fout, " goto %s;", po->operand[0].name);
3049 // else SETcc - should already be handled
3050 break;
3051
3052 case OP_JMP:
3053 assert_operand_cnt(1);
3054 last_arith_dst = NULL;
3055 delayed_flag_op = NULL;
3056
3057 if (po->operand[0].type == OPT_REGMEM) {
3058 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
3059 buf1, buf2);
3060 if (ret != 2)
3061 ferr(po, "parse failure for jmp '%s'\n",
3062 po->operand[0].name);
3063 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
3064 break;
3065 }
3066 else if (po->operand[0].type != OPT_LABEL)
3067 ferr(po, "unhandled jmp type\n");
3068
3069 fprintf(fout, " goto %s;", po->operand[0].name);
3070 break;
3071
3072 case OP_CALL:
3073 assert_operand_cnt(1);
3074 pp = po->datap;
3075 if (pp == NULL)
3076 ferr(po, "NULL pp\n");
3077
3078 if (po->operand[0].type != OPT_LABEL)
3079 fprintf(fout, " icall%d = (void *)%s;\n", i,
3080 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]));
3081
3082 fprintf(fout, " ");
3083 if (strstr(pp->ret_type.name, "int64")) {
3084 if (po->flags & OPF_TAIL)
3085 ferr(po, "int64 and tail?\n");
3086 fprintf(fout, "mul_tmp = ");
3087 }
3088 else if (!IS(pp->ret_type.name, "void")) {
3089 if (po->flags & OPF_TAIL) {
3090 fprintf(fout, "return ");
3091 if (g_func_pp.ret_type.is_ptr != pp->ret_type.is_ptr)
3092 fprintf(fout, "(%s)", g_func_pp.ret_type.name);
3093 }
3094 else {
3095 fprintf(fout, "eax = ");
3096 if (pp->ret_type.is_ptr)
3097 fprintf(fout, "(u32)");
3098 }
3099 }
3100
3101 if (po->operand[0].type != OPT_LABEL) {
3102 fprintf(fout, "icall%d(", i);
3103 }
3104 else {
3105 if (pp->name[0] == 0)
3106 ferr(po, "missing pp->name\n");
3107 fprintf(fout, "%s(", pp->name);
3108 }
3109
3110 for (arg = 0; arg < pp->argc; arg++) {
3111 if (arg > 0)
3112 fprintf(fout, ", ");
3113
3114 cast[0] = 0;
3115 if (pp->arg[arg].type.is_ptr)
3116 snprintf(cast, sizeof(cast), "(%s)", pp->arg[arg].type.name);
3117
3118 if (pp->arg[arg].reg != NULL) {
3119 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
3120 continue;
3121 }
3122
3123 // stack arg
3124 tmp_op = pp->arg[arg].datap;
3125 if (tmp_op == NULL)
3126 ferr(po, "parsed_op missing for arg%d\n", arg);
3127 if (tmp_op->argnum != 0) {
3128 fprintf(fout, "%ss_a%d", cast, tmp_op->argnum);
3129 }
3130 else {
3131 fprintf(fout, "%s",
3132 out_src_opr(buf1, sizeof(buf1),
3133 tmp_op, &tmp_op->operand[0], cast, 0));
3134 }
3135 }
3136 fprintf(fout, ");");
3137
3138 if (strstr(pp->ret_type.name, "int64")) {
3139 fprintf(fout, "\n");
3140 fprintf(fout, " edx = mul_tmp >> 32;\n");
3141 fprintf(fout, " eax = mul_tmp;");
3142 }
3143
3144 if (po->flags & OPF_TAIL) {
3145 strcpy(g_comment, "tailcall");
3146 if (IS(pp->ret_type.name, "void")
3147 && !(g_ida_func_attr & IDAFA_NORETURN))
3148 {
3149 fprintf(fout, "\n return;");
3150 strcpy(g_comment, "^ tailcall");
3151 }
3152 }
3153 delayed_flag_op = NULL;
3154 last_arith_dst = NULL;
3155 break;
3156
3157 case OP_RET:
3158 if (g_func_pp.is_vararg)
3159 fprintf(fout, " va_end(ap);\n");
3160
3161 if (IS(g_func_pp.ret_type.name, "void")) {
3162 if (i != opcnt - 1 || label_pending)
3163 fprintf(fout, " return;");
3164 }
3165 else if (g_func_pp.ret_type.is_ptr) {
3166 fprintf(fout, " return (%s)eax;",
3167 g_func_pp.ret_type.name);
3168 }
3169 else
3170 fprintf(fout, " return eax;");
3171
3172 last_arith_dst = NULL;
3173 delayed_flag_op = NULL;
3174 break;
3175
3176 case OP_PUSH:
3177 if (po->argnum != 0) {
3178 // special case - saved func arg
3179 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
3180 fprintf(fout, " s_a%d = %s;", po->argnum, buf1);
3181 break;
3182 }
3183 else if (po->flags & OPF_RSAVE) {
3184 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
3185 fprintf(fout, " s_%s = %s;", buf1, buf1);
3186 break;
3187 }
3188 if (!(g_ida_func_attr & IDAFA_NORETURN))
3189 ferr(po, "stray push encountered\n");
3190 no_output = 1;
3191 break;
3192
3193 case OP_POP:
3194 if (po->flags & OPF_RSAVE) {
3195 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3196 fprintf(fout, " %s = s_%s;", buf1, buf1);
3197 break;
3198 }
3199 ferr(po, "stray pop encountered\n");
3200 break;
3201
3202 case OP_NOP:
3203 no_output = 1;
3204 break;
3205
3206 default:
3207 no_output = 1;
3208 ferr(po, "unhandled op type %d, flags %x\n",
3209 po->op, po->flags);
3210 break;
3211 }
3212
3213 if (g_comment[0] != 0) {
3214 fprintf(fout, " // %s", g_comment);
3215 g_comment[0] = 0;
3216 no_output = 0;
3217 }
3218 if (!no_output)
3219 fprintf(fout, "\n");
3220
3221 // some sanity checking
3222 if ((po->flags & OPF_REP) && po->op != OP_STOS
3223 && po->op != OP_MOVS && po->op != OP_CMPS)
3224 ferr(po, "unexpected rep\n");
3225 if ((po->flags & (OPF_REPZ|OPF_REPNZ)) && po->op != OP_CMPS)
3226 ferr(po, "unexpected repz/repnz\n");
3227
3228 if (pfomask != 0)
3229 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
3230
3231 // see is delayed flag stuff is still valid
3232 if (delayed_flag_op != NULL && delayed_flag_op != po) {
3233 if (is_any_opr_modified(delayed_flag_op, po))
3234 delayed_flag_op = NULL;
3235 }
3236
3237 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
3238 if (is_opr_modified(last_arith_dst, po))
3239 last_arith_dst = NULL;
3240 }
3241
3242 label_pending = 0;
3243 }
3244
3245 fprintf(fout, "}\n\n");
3246
3247 // cleanup
3248 for (i = 0; i < opcnt; i++) {
3249 struct label_ref *lr, *lr_del;
3250
3251 lr = g_label_refs[i].next;
3252 while (lr != NULL) {
3253 lr_del = lr;
3254 lr = lr->next;
3255 free(lr_del);
3256 }
3257 g_label_refs[i].i = -1;
3258 g_label_refs[i].next = NULL;
3259
3260 if (ops[i].op == OP_CALL) {
3261 pp = ops[i].datap;
3262 if (pp) {
3263 proto_release(pp);
3264 free(pp);
3265 }
3266 }
3267 }
3268 proto_release(&g_func_pp);
3269}
3270
3271static void set_label(int i, const char *name)
3272{
3273 const char *p;
3274 int len;
3275
3276 len = strlen(name);
3277 p = strchr(name, ':');
3278 if (p != NULL)
3279 len = p - name;
3280
3281 if (len > sizeof(g_labels[0]) - 1)
3282 aerr("label '%s' too long: %d\n", name, len);
3283 if (g_labels[i][0] != 0 && !IS_START(g_labels[i], "algn_"))
3284 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
3285 memcpy(g_labels[i], name, len);
3286 g_labels[i][len] = 0;
3287}
3288
3289// '=' needs special treatment..
3290static char *next_word_s(char *w, size_t wsize, char *s)
3291{
3292 size_t i;
3293
3294 s = sskip(s);
3295
3296 for (i = 0; i < wsize - 1; i++) {
3297 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
3298 break;
3299 w[i] = s[i];
3300 }
3301 w[i] = 0;
3302
3303 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
3304 printf("warning: '%s' truncated\n", w);
3305
3306 return s + i;
3307}
3308
3309struct chunk_item {
3310 char *name;
3311 long fptr;
3312 int asmln;
3313};
3314
3315static int cmp_chunks(const void *p1, const void *p2)
3316{
3317 const struct chunk_item *c1 = p1, *c2 = p2;
3318 return strcmp(c1->name, c2->name);
3319}
3320
3321static int cmpstringp(const void *p1, const void *p2)
3322{
3323 return strcmp(*(char * const *)p1, *(char * const *)p2);
3324}
3325
3326int main(int argc, char *argv[])
3327{
3328 FILE *fout, *fasm, *frlist;
3329 struct parsed_data *pd = NULL;
3330 int pd_alloc = 0;
3331 char **rlist = NULL;
3332 int rlist_len = 0;
3333 int rlist_alloc = 0;
3334 struct chunk_item *func_chunks;
3335 int func_chunks_used = 0;
3336 int func_chunks_sorted = 0;
3337 int func_chunk_cnt = 0;
3338 int func_chunk_alloc;
3339 int func_chunk_i = -1;
3340 long func_chunk_ret = 0;
3341 int func_chunk_ret_ln = 0;
3342 char line[256];
3343 char words[16][256];
3344 enum opr_lenmod lmod;
3345 int in_func = 0;
3346 int pending_endp = 0;
3347 int skip_func = 0;
3348 int skip_warned = 0;
3349 int eq_alloc;
3350 int verbose = 0;
3351 int arg_out;
3352 int arg = 1;
3353 int pi = 0;
3354 int i, j;
3355 int ret, len;
3356 char *p;
3357 int wordc;
3358
3359 if (argv[1] && IS(argv[1], "-v")) {
3360 verbose = 1;
3361 arg++;
3362 }
3363
3364 if (argc < arg + 3) {
3365 printf("usage:\n%s [-v] <.c> <.asm> <hdrf> [rlist]*\n",
3366 argv[0]);
3367 return 1;
3368 }
3369
3370 arg_out = arg++;
3371
3372 asmfn = argv[arg++];
3373 fasm = fopen(asmfn, "r");
3374 my_assert_not(fasm, NULL);
3375
3376 hdrfn = argv[arg++];
3377 g_fhdr = fopen(hdrfn, "r");
3378 my_assert_not(g_fhdr, NULL);
3379
3380 rlist_alloc = 64;
3381 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
3382 my_assert_not(rlist, NULL);
3383 // needs special handling..
3384 rlist[rlist_len++] = "__alloca_probe";
3385
3386 func_chunk_alloc = 32;
3387 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
3388 my_assert_not(func_chunks, NULL);
3389
3390 for (; arg < argc; arg++) {
3391 frlist = fopen(argv[arg], "r");
3392 my_assert_not(frlist, NULL);
3393
3394 while (fgets(line, sizeof(line), frlist)) {
3395 p = sskip(line);
3396 if (*p == 0 || *p == ';' || *p == '#')
3397 continue;
3398
3399 p = next_word(words[0], sizeof(words[0]), p);
3400 if (words[0][0] == 0)
3401 continue;
3402
3403 if (rlist_len >= rlist_alloc) {
3404 rlist_alloc = rlist_alloc * 2 + 64;
3405 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
3406 my_assert_not(rlist, NULL);
3407 }
3408 rlist[rlist_len++] = strdup(words[0]);
3409 }
3410
3411 fclose(frlist);
3412 frlist = NULL;
3413 }
3414
3415 if (rlist_len > 0)
3416 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
3417
3418 fout = fopen(argv[arg_out], "w");
3419 my_assert_not(fout, NULL);
3420
3421 eq_alloc = 128;
3422 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
3423 my_assert_not(g_eqs, NULL);
3424
3425 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
3426 g_label_refs[i].i = -1;
3427 g_label_refs[i].next = NULL;
3428 }
3429
3430 while (fgets(line, sizeof(line), fasm))
3431 {
3432 wordc = 0;
3433 asmln++;
3434
3435 p = sskip(line);
3436 if (*p == 0)
3437 continue;
3438
3439 // get rid of random tabs
3440 for (i = 0; line[i] != 0; i++)
3441 if (line[i] == '\t')
3442 line[i] = ' ';
3443
3444 if (*p == ';')
3445 {
3446 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
3447 goto do_pending_endp; // eww..
3448
3449 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
3450 {
3451 static const char *attrs[] = {
3452 "bp-based frame",
3453 "library function",
3454 "static",
3455 "noreturn",
3456 "thunk",
3457 "fpd=",
3458 };
3459
3460 // parse IDA's attribute-list comment
3461 g_ida_func_attr = 0;
3462 p = sskip(p + 13);
3463
3464 for (; *p != 0; p = sskip(p)) {
3465 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
3466 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
3467 g_ida_func_attr |= 1 << i;
3468 p += strlen(attrs[i]);
3469 break;
3470 }
3471 }
3472 if (i == ARRAY_SIZE(attrs)) {
3473 anote("unparsed IDA attr: %s\n", p);
3474 break;
3475 }
3476 if (IS(attrs[i], "fpd=")) {
3477 p = next_word(words[0], sizeof(words[0]), p);
3478 // ignore for now..
3479 }
3480 }
3481 }
3482 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
3483 {
3484 p += 30;
3485 next_word(words[0], sizeof(words[0]), p);
3486 if (words[0][0] == 0)
3487 aerr("missing nam for func chunk?\n");
3488 if (func_chunk_cnt >= func_chunk_alloc) {
3489 func_chunk_alloc *= 2;
3490 func_chunks = realloc(func_chunks,
3491 func_chunk_alloc * sizeof(func_chunks[0]));
3492 my_assert_not(func_chunks, NULL);
3493 }
3494 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
3495 func_chunks[func_chunk_cnt].name = strdup(words[0]);
3496 func_chunks[func_chunk_cnt].asmln = asmln;
3497 func_chunk_cnt++;
3498 func_chunks_sorted = 0;
3499 }
3500 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
3501 {
3502 if (func_chunk_i >= 0) {
3503 if (func_chunk_i < func_chunk_cnt
3504 && IS(func_chunks[func_chunk_i].name, g_func))
3505 {
3506 // move on to next chunk
3507 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
3508 if (ret)
3509 aerr("seek failed for '%s' chunk #%d\n",
3510 g_func, func_chunk_i);
3511 asmln = func_chunks[func_chunk_i].asmln;
3512 func_chunk_i++;
3513 }
3514 else {
3515 if (func_chunk_ret == 0)
3516 aerr("no return from chunk?\n");
3517 fseek(fasm, func_chunk_ret, SEEK_SET);
3518 asmln = func_chunk_ret_ln;
3519 func_chunk_ret = 0;
3520 pending_endp = 1;
3521 }
3522 }
3523 }
3524 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
3525 func_chunks_used = 1;
3526 p += 20;
3527 if (IS_START(g_func, "sub_")) {
3528 unsigned long addr = strtoul(p, NULL, 16);
3529 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
3530 if (addr > f_addr)
3531 aerr("need a chunk %lX that is after %s\n", addr, g_func);
3532 }
3533 }
3534 continue;
3535 } // *p == ';'
3536
3537parse_words:
3538 memset(words, 0, sizeof(words));
3539 for (wordc = 0; wordc < 16; wordc++) {
3540 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
3541 if (*p == 0 || *p == ';') {
3542 wordc++;
3543 break;
3544 }
3545 }
3546
3547 // alow asm patches in comments
3548 if (*p == ';' && IS_START(p, "; sctpatch:")) {
3549 p = sskip(p + 11);
3550 if (*p == 0 || *p == ';')
3551 continue;
3552 goto parse_words; // lame
3553 }
3554
3555 if (wordc == 0) {
3556 // shouldn't happen
3557 awarn("wordc == 0?\n");
3558 continue;
3559 }
3560
3561 // don't care about this:
3562 if (words[0][0] == '.'
3563 || IS(words[0], "include")
3564 || IS(words[0], "assume") || IS(words[1], "segment")
3565 || IS(words[0], "align"))
3566 {
3567 continue;
3568 }
3569
3570do_pending_endp:
3571 // do delayed endp processing to collect switch jumptables
3572 if (pending_endp) {
3573 if (in_func && !skip_func && wordc >= 2
3574 && ((words[0][0] == 'd' && words[0][2] == 0)
3575 || (words[1][0] == 'd' && words[1][2] == 0)))
3576 {
3577 i = 1;
3578 if (words[1][0] == 'd' && words[1][2] == 0) {
3579 // label
3580 if (g_func_pd_cnt >= pd_alloc) {
3581 pd_alloc = pd_alloc * 2 + 16;
3582 g_func_pd = realloc(g_func_pd,
3583 sizeof(g_func_pd[0]) * pd_alloc);
3584 my_assert_not(g_func_pd, NULL);
3585 }
3586 pd = &g_func_pd[g_func_pd_cnt];
3587 g_func_pd_cnt++;
3588 memset(pd, 0, sizeof(*pd));
3589 strcpy(pd->label, words[0]);
3590 pd->type = OPT_CONST;
3591 pd->lmod = lmod_from_directive(words[1]);
3592 i = 2;
3593 }
3594 else {
3595 lmod = lmod_from_directive(words[0]);
3596 if (lmod != pd->lmod)
3597 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
3598 }
3599
3600 if (pd->count_alloc < pd->count + wordc) {
3601 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
3602 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
3603 my_assert_not(pd->d, NULL);
3604 }
3605 for (; i < wordc; i++) {
3606 if (IS(words[i], "offset")) {
3607 pd->type = OPT_OFFSET;
3608 i++;
3609 }
3610 p = strchr(words[i], ',');
3611 if (p != NULL)
3612 *p = 0;
3613 if (pd->type == OPT_OFFSET)
3614 pd->d[pd->count].u.label = strdup(words[i]);
3615 else
3616 pd->d[pd->count].u.val = parse_number(words[i]);
3617 pd->d[pd->count].bt_i = -1;
3618 pd->count++;
3619 }
3620 continue;
3621 }
3622
3623 if (in_func && !skip_func)
3624 gen_func(fout, g_fhdr, g_func, pi);
3625
3626 pending_endp = 0;
3627 in_func = 0;
3628 g_ida_func_attr = 0;
3629 skip_warned = 0;
3630 skip_func = 0;
3631 g_func[0] = 0;
3632 func_chunks_used = 0;
3633 func_chunk_i = -1;
3634 if (pi != 0) {
3635 memset(&ops, 0, pi * sizeof(ops[0]));
3636 memset(g_labels, 0, pi * sizeof(g_labels[0]));
3637 pi = 0;
3638 }
3639 g_eqcnt = 0;
3640 for (i = 0; i < g_func_pd_cnt; i++) {
3641 pd = &g_func_pd[i];
3642 if (pd->type == OPT_OFFSET) {
3643 for (j = 0; j < pd->count; j++)
3644 free(pd->d[j].u.label);
3645 }
3646 free(pd->d);
3647 pd->d = NULL;
3648 }
3649 g_func_pd_cnt = 0;
3650 pd = NULL;
3651 if (wordc == 0)
3652 continue;
3653 }
3654
3655 if (IS(words[1], "proc")) {
3656 if (in_func)
3657 aerr("proc '%s' while in_func '%s'?\n",
3658 words[0], g_func);
3659 p = words[0];
3660 if ((g_ida_func_attr & IDAFA_THUNK)
3661 || bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
3662 skip_func = 1;
3663 strcpy(g_func, words[0]);
3664 set_label(0, words[0]);
3665 in_func = 1;
3666 continue;
3667 }
3668
3669 if (IS(words[1], "endp"))
3670 {
3671 if (!in_func)
3672 aerr("endp '%s' while not in_func?\n", words[0]);
3673 if (!IS(g_func, words[0]))
3674 aerr("endp '%s' while in_func '%s'?\n",
3675 words[0], g_func);
3676
3677 if (!skip_func && func_chunks_used) {
3678 // start processing chunks
3679 struct chunk_item *ci, key = { g_func, 0 };
3680
3681 func_chunk_ret = ftell(fasm);
3682 func_chunk_ret_ln = asmln;
3683 if (!func_chunks_sorted) {
3684 qsort(func_chunks, func_chunk_cnt,
3685 sizeof(func_chunks[0]), cmp_chunks);
3686 func_chunks_sorted = 1;
3687 }
3688 ci = bsearch(&key, func_chunks, func_chunk_cnt,
3689 sizeof(func_chunks[0]), cmp_chunks);
3690 if (ci == NULL)
3691 aerr("'%s' needs chunks, but none found\n", g_func);
3692 func_chunk_i = ci - func_chunks;
3693 for (; func_chunk_i > 0; func_chunk_i--)
3694 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
3695 break;
3696
3697 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
3698 if (ret)
3699 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
3700 asmln = func_chunks[func_chunk_i].asmln;
3701 func_chunk_i++;
3702 continue;
3703 }
3704 pending_endp = 1;
3705 continue;
3706 }
3707
3708 p = strchr(words[0], ':');
3709 if (p != NULL) {
3710 set_label(pi, words[0]);
3711 continue;
3712 }
3713
3714 if (!in_func || skip_func) {
3715 if (!skip_warned && !skip_func && g_labels[pi][0] != 0) {
3716 if (verbose)
3717 anote("skipping from '%s'\n", g_labels[pi]);
3718 skip_warned = 1;
3719 }
3720 g_labels[pi][0] = 0;
3721 continue;
3722 }
3723
3724 if (wordc > 1 && IS(words[1], "=")) {
3725 if (wordc != 5)
3726 aerr("unhandled equ, wc=%d\n", wordc);
3727 if (g_eqcnt >= eq_alloc) {
3728 eq_alloc *= 2;
3729 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
3730 my_assert_not(g_eqs, NULL);
3731 }
3732
3733 len = strlen(words[0]);
3734 if (len > sizeof(g_eqs[0].name) - 1)
3735 aerr("equ name too long: %d\n", len);
3736 strcpy(g_eqs[g_eqcnt].name, words[0]);
3737
3738 if (!IS(words[3], "ptr"))
3739 aerr("unhandled equ\n");
3740 if (IS(words[2], "dword"))
3741 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
3742 else if (IS(words[2], "word"))
3743 g_eqs[g_eqcnt].lmod = OPLM_WORD;
3744 else if (IS(words[2], "byte"))
3745 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
3746 else
3747 aerr("bad lmod: '%s'\n", words[2]);
3748
3749 g_eqs[g_eqcnt].offset = parse_number(words[4]);
3750 g_eqcnt++;
3751 continue;
3752 }
3753
3754 if (pi >= ARRAY_SIZE(ops))
3755 aerr("too many ops\n");
3756
3757 parse_op(&ops[pi], words, wordc);
3758 pi++;
3759 }
3760
3761 fclose(fout);
3762 fclose(fasm);
3763 fclose(g_fhdr);
3764
3765 return 0;
3766}
3767
3768// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.