notaz.gp2x.de / ia32rtools.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
flag/cond handling improvements
[ia32rtools.git] / tools / translate.c
... / ...
CommitLineData
1#define _GNU_SOURCE
2#include <stdio.h>
3#include <stdlib.h>
4#include <string.h>
5
6#include "my_assert.h"
7#include "my_str.h"
8
9#define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
10#define IS(w, y) !strcmp(w, y)
11#define IS_START(w, y) !strncmp(w, y, strlen(y))
12
13#include "protoparse.h"
14
15static const char *asmfn;
16static int asmln;
17static FILE *g_fhdr;
18
19#define anote(fmt, ...) \
20 printf("%s:%d: note: " fmt, asmfn, asmln, ##__VA_ARGS__)
21#define awarn(fmt, ...) \
22 printf("%s:%d: warning: " fmt, asmfn, asmln, ##__VA_ARGS__)
23#define aerr(fmt, ...) do { \
24 printf("%s:%d: error: " fmt, asmfn, asmln, ##__VA_ARGS__); \
25 fcloseall(); \
26 exit(1); \
27} while (0)
28
29enum op_flags {
30 OPF_RMD = (1 << 0), /* removed or optimized out */
31 OPF_DATA = (1 << 1), /* data processing - writes to dst opr */
32 OPF_FLAGS = (1 << 2), /* sets flags */
33 OPF_JMP = (1 << 3), /* branches, ret and call */
34 OPF_CC = (1 << 4), /* uses flags */
35 OPF_TAIL = (1 << 5), /* ret or tail call */
36 OPF_RSAVE = (1 << 6), /* push/pop is local reg save/load */
37 OPF_REP = (1 << 7), /* prefixed by rep */
38 OPF_REPZ = (1 << 8), /* rep is repe/repz */
39 OPF_REPNZ = (1 << 9), /* rep is repne/repnz */
40 OPF_FARG = (1 << 10), /* push collected as func arg (no reuse) */
41};
42
43enum op_op {
44 OP_INVAL,
45 OP_NOP,
46 OP_PUSH,
47 OP_POP,
48 OP_MOV,
49 OP_LEA,
50 OP_MOVZX,
51 OP_MOVSX,
52 OP_NOT,
53 OP_CDQ,
54 OP_STOS,
55 OP_MOVS,
56 OP_CMPS,
57 OP_RET,
58 OP_ADD,
59 OP_SUB,
60 OP_AND,
61 OP_OR,
62 OP_XOR,
63 OP_SHL,
64 OP_SHR,
65 OP_SAR,
66 OP_ROL,
67 OP_ROR,
68 OP_ADC,
69 OP_SBB,
70 OP_INC,
71 OP_DEC,
72 OP_NEG,
73 OP_MUL,
74 OP_IMUL,
75 OP_DIV,
76 OP_IDIV,
77 OP_TEST,
78 OP_CMP,
79 OP_CALL,
80 OP_JMP,
81 OP_JO,
82 OP_JNO,
83 OP_JC,
84 OP_JNC,
85 OP_JZ,
86 OP_JNZ,
87 OP_JBE,
88 OP_JA,
89 OP_JS,
90 OP_JNS,
91 OP_JP,
92 OP_JNP,
93 OP_JL,
94 OP_JGE,
95 OP_JLE,
96 OP_JG,
97};
98
99enum opr_type {
100 OPT_UNSPEC,
101 OPT_REG,
102 OPT_REGMEM,
103 OPT_LABEL,
104 OPT_OFFSET,
105 OPT_CONST,
106};
107
108// must be sorted (larger len must be further in enum)
109enum opr_lenmod {
110 OPLM_UNSPEC,
111 OPLM_BYTE,
112 OPLM_WORD,
113 OPLM_DWORD,
114};
115
116#define MAX_OPERANDS 3
117
118struct parsed_opr {
119 enum opr_type type;
120 enum opr_lenmod lmod;
121 unsigned int is_ptr:1; // pointer in C
122 unsigned int is_array:1; // array in C
123 unsigned int size_mismatch:1; // type override differs from C
124 unsigned int size_lt:1; // type override is larger than C
125 int reg;
126 unsigned int val;
127 char name[256];
128};
129
130struct parsed_op {
131 enum op_op op;
132 struct parsed_opr operand[MAX_OPERANDS];
133 unsigned int flags;
134 int operand_cnt;
135 int regmask_src; // all referensed regs
136 int regmask_dst;
137 int pfomask; // flagop: parsed_flag_op that can't be delayed
138 int argnum; // push: altered before call arg #
139 int cc_scratch; // scratch storage during analysis
140 int bt_i; // branch target for branches
141 struct parsed_data *btj;// branch targets for jumptables
142 void *datap;
143};
144
145// datap:
146// OP_CALL - ptr to parsed_proto
147// (OPF_CC) - point to one of (OPF_FLAGS) that affects cc op
148
149struct parsed_equ {
150 char name[64];
151 enum opr_lenmod lmod;
152 int offset;
153};
154
155struct parsed_data {
156 char label[256];
157 enum opr_type type;
158 enum opr_lenmod lmod;
159 int count;
160 int count_alloc;
161 struct {
162 union {
163 char *label;
164 unsigned int val;
165 } u;
166 int bt_i;
167 } *d;
168};
169
170struct label_ref {
171 int i;
172 struct label_ref *next;
173};
174
175enum ida_func_attr {
176 IDAFA_BP_FRAME = (1 << 0),
177 IDAFA_LIB_FUNC = (1 << 1),
178 IDAFA_STATIC = (1 << 2),
179 IDAFA_NORETURN = (1 << 3),
180 IDAFA_THUNK = (1 << 4),
181 IDAFA_FPD = (1 << 5),
182};
183
184#define MAX_OPS 4096
185
186static struct parsed_op ops[MAX_OPS];
187static struct parsed_equ *g_eqs;
188static int g_eqcnt;
189static char g_labels[MAX_OPS][32];
190static struct label_ref g_label_refs[MAX_OPS];
191static const struct parsed_proto *g_func_pp;
192static struct parsed_data *g_func_pd;
193static int g_func_pd_cnt;
194static char g_func[256];
195static char g_comment[256];
196static int g_bp_frame;
197static int g_sp_frame;
198static int g_stack_fsz;
199static int g_ida_func_attr;
200#define ferr(op_, fmt, ...) do { \
201 printf("error:%s:#%zd: '%s': " fmt, g_func, (op_) - ops, \
202 dump_op(op_), ##__VA_ARGS__); \
203 fcloseall(); \
204 exit(1); \
205} while (0)
206#define fnote(op_, fmt, ...) \
207 printf("error:%s:#%zd: '%s': " fmt, g_func, (op_) - ops, \
208 dump_op(op_), ##__VA_ARGS__)
209
210#define MAX_REGS 8
211
212const char *regs_r32[] = { "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp" };
213const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" };
214const char *regs_r8l[] = { "al", "bl", "cl", "dl" };
215const char *regs_r8h[] = { "ah", "bh", "ch", "dh" };
216
217enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP };
218
219// possible basic comparison types (without inversion)
220enum parsed_flag_op {
221 PFO_O, // 0 OF=1
222 PFO_C, // 2 CF=1
223 PFO_Z, // 4 ZF=1
224 PFO_BE, // 6 CF=1||ZF=1
225 PFO_S, // 8 SF=1
226 PFO_P, // a PF=1
227 PFO_L, // c SF!=OF
228 PFO_LE, // e ZF=1||SF!=OF
229};
230
231static const char *parsed_flag_op_names[] = {
232 "o", "c", "z", "be", "s", "p", "l", "le"
233};
234
235static int char_array_i(const char *array[], size_t len, const char *s)
236{
237 int i;
238
239 for (i = 0; i < len; i++)
240 if (IS(s, array[i]))
241 return i;
242
243 return -1;
244}
245
246static void printf_number(char *buf, size_t buf_size,
247 unsigned long number)
248{
249 // output in C-friendly form
250 snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
251}
252
253static int parse_reg(enum opr_lenmod *reg_lmod, const char *s)
254{
255 int reg;
256
257 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s);
258 if (reg >= 0) {
259 *reg_lmod = OPLM_DWORD;
260 return reg;
261 }
262 reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s);
263 if (reg >= 0) {
264 *reg_lmod = OPLM_WORD;
265 return reg;
266 }
267 reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s);
268 if (reg >= 0) {
269 *reg_lmod = OPLM_BYTE;
270 return reg;
271 }
272 reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s);
273 if (reg >= 0) {
274 *reg_lmod = OPLM_BYTE;
275 return reg;
276 }
277
278 return -1;
279}
280
281static unsigned long parse_number(const char *number)
282{
283 int len = strlen(number);
284 const char *p = number;
285 char *endp = NULL;
286 unsigned long ret;
287 int neg = 0;
288 int bad;
289
290 if (*p == '-') {
291 neg = 1;
292 p++;
293 }
294 if (len > 1 && *p == '0')
295 p++;
296 if (number[len - 1] == 'h') {
297 ret = strtoul(p, &endp, 16);
298 bad = (*endp != 'h');
299 }
300 else {
301 ret = strtoul(p, &endp, 10);
302 bad = (*endp != 0);
303 }
304 if (bad)
305 aerr("number parsing failed\n");
306 if (neg)
307 ret = -ret;
308 return ret;
309}
310
311static int parse_indmode(char *name, int *regmask, int need_c_cvt)
312{
313 enum opr_lenmod lmod;
314 char cvtbuf[256];
315 char *d = cvtbuf;
316 char *s = name;
317 char w[64];
318 long number;
319 int reg;
320 int c = 0;
321
322 *d = 0;
323
324 while (*s != 0) {
325 d += strlen(d);
326 while (my_isblank(*s))
327 s++;
328 for (; my_issep(*s); d++, s++)
329 *d = *s;
330 while (my_isblank(*s))
331 s++;
332 *d = 0;
333
334 // skip 'ds:' prefix
335 if (IS_START(s, "ds:"))
336 s += 3;
337
338 s = next_idt(w, sizeof(w), s);
339 if (w[0] == 0)
340 break;
341 c++;
342
343 reg = parse_reg(&lmod, w);
344 if (reg >= 0) {
345 *regmask |= 1 << reg;
346 goto pass;
347 }
348
349 if ('0' <= w[0] && w[0] <= '9') {
350 number = parse_number(w);
351 printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number);
352 continue;
353 }
354
355 // probably some label/identifier - pass
356
357pass:
358 snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w);
359 }
360
361 if (need_c_cvt)
362 strcpy(name, cvtbuf);
363
364 return c;
365}
366
367static int is_reg_in_str(const char *s)
368{
369 int i;
370
371 if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
372 return 0;
373
374 for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
375 if (!strncmp(s, regs_r32[i], 3))
376 return 1;
377
378 return 0;
379}
380
381static const char *parse_stack_el(const char *name, char *extra_reg)
382{
383 const char *p, *p2, *s;
384 char *endp = NULL;
385 char buf[32];
386 long val;
387 int len;
388
389 p = name;
390 if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
391 p += 4;
392 if (extra_reg != NULL) {
393 strncpy(extra_reg, name, 3);
394 extra_reg[4] = 0;
395 }
396 }
397
398 if (IS_START(p, "ebp+")) {
399 p += 4;
400
401 p2 = strchr(p, '+');
402 if (p2 != NULL && is_reg_in_str(p)) {
403 if (extra_reg != NULL) {
404 strncpy(extra_reg, p, p2 - p);
405 extra_reg[p2 - p] = 0;
406 }
407 p = p2 + 1;
408 }
409
410 if (!('0' <= *p && *p <= '9'))
411 return p;
412
413 return NULL;
414 }
415
416 if (!IS_START(name, "esp+"))
417 return NULL;
418
419 p = strchr(name + 4, '+');
420 if (p) {
421 // must be a number after esp+, already converted to 0x..
422 s = name + 4;
423 if (!('0' <= *s && *s <= '9')) {
424 aerr("%s nan?\n", __func__);
425 return NULL;
426 }
427 if (s[0] == '0' && s[1] == 'x')
428 s += 2;
429 len = p - s;
430 if (len < sizeof(buf) - 1) {
431 strncpy(buf, s, len);
432 buf[len] = 0;
433 val = strtol(buf, &endp, 16);
434 if (val == 0 || *endp != 0) {
435 aerr("%s num parse fail for '%s'\n", __func__, buf);
436 return NULL;
437 }
438 }
439 p++;
440 }
441 else
442 p = name + 4;
443
444 if ('0' <= *p && *p <= '9')
445 return NULL;
446
447 return p;
448}
449
450static int guess_lmod_from_name(struct parsed_opr *opr)
451{
452 if (!strncmp(opr->name, "dword_", 6)) {
453 opr->lmod = OPLM_DWORD;
454 return 1;
455 }
456 if (!strncmp(opr->name, "word_", 5)) {
457 opr->lmod = OPLM_WORD;
458 return 1;
459 }
460 if (!strncmp(opr->name, "byte_", 5)) {
461 opr->lmod = OPLM_BYTE;
462 return 1;
463 }
464 return 0;
465}
466
467static int guess_lmod_from_c_type(enum opr_lenmod *lmod,
468 const struct parsed_type *c_type)
469{
470 static const char *dword_types[] = {
471 "int", "_DWORD", "DWORD", "HANDLE", "HWND", "HMODULE",
472 "WPARAM", "LPARAM", "UINT",
473 };
474 static const char *word_types[] = {
475 "uint16_t", "int16_t",
476 "unsigned __int16", "__int16",
477 };
478 static const char *byte_types[] = {
479 "uint8_t", "int8_t", "char",
480 "unsigned __int8", "__int8", "BYTE",
481 };
482 const char *n;
483 int i;
484
485 if (c_type->is_ptr) {
486 *lmod = OPLM_DWORD;
487 return 1;
488 }
489
490 n = skip_type_mod(c_type->name);
491
492 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
493 if (IS(n, dword_types[i])) {
494 *lmod = OPLM_DWORD;
495 return 1;
496 }
497 }
498
499 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
500 if (IS(n, word_types[i])) {
501 *lmod = OPLM_WORD;
502 return 1;
503 }
504 }
505
506 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
507 if (IS(n, byte_types[i])) {
508 *lmod = OPLM_BYTE;
509 return 1;
510 }
511 }
512
513 return 0;
514}
515
516static enum opr_type lmod_from_directive(const char *d)
517{
518 if (IS(d, "dd"))
519 return OPLM_DWORD;
520 else if (IS(d, "dw"))
521 return OPLM_WORD;
522 else if (IS(d, "db"))
523 return OPLM_BYTE;
524
525 aerr("unhandled directive: '%s'\n", d);
526 return OPLM_UNSPEC;
527}
528
529static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
530 int *regmask)
531{
532 opr->type = OPT_REG;
533 opr->reg = reg;
534 opr->lmod = lmod;
535 *regmask |= 1 << reg;
536}
537
538static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
539 int *extra_offs);
540
541static int parse_operand(struct parsed_opr *opr,
542 int *regmask, int *regmask_indirect,
543 char words[16][256], int wordc, int w, unsigned int op_flags)
544{
545 const struct parsed_proto *pp;
546 enum opr_lenmod tmplmod;
547 unsigned long number;
548 int ret, len;
549 int wordc_in;
550 char *tmp;
551 int i;
552
553 if (w >= wordc)
554 aerr("parse_operand w %d, wordc %d\n", w, wordc);
555
556 opr->reg = xUNSPEC;
557
558 for (i = w; i < wordc; i++) {
559 len = strlen(words[i]);
560 if (words[i][len - 1] == ',') {
561 words[i][len - 1] = 0;
562 wordc = i + 1;
563 break;
564 }
565 }
566
567 wordc_in = wordc - w;
568
569 if ((op_flags & OPF_JMP) && wordc_in > 0
570 && !('0' <= words[w][0] && words[w][0] <= '9'))
571 {
572 const char *label = NULL;
573
574 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
575 && IS(words[w + 1], "ptr"))
576 label = words[w + 2];
577 else if (wordc_in == 2 && IS(words[w], "short"))
578 label = words[w + 1];
579 else if (wordc_in == 1
580 && strchr(words[w], '[') == NULL
581 && parse_reg(&tmplmod, words[w]) < 0)
582 label = words[w];
583
584 if (label != NULL) {
585 opr->type = OPT_LABEL;
586 if (IS_START(label, "ds:"))
587 label += 3;
588 strcpy(opr->name, label);
589 return wordc;
590 }
591 }
592
593 if (wordc_in >= 3) {
594 if (IS(words[w + 1], "ptr")) {
595 if (IS(words[w], "dword"))
596 opr->lmod = OPLM_DWORD;
597 else if (IS(words[w], "word"))
598 opr->lmod = OPLM_WORD;
599 else if (IS(words[w], "byte"))
600 opr->lmod = OPLM_BYTE;
601 else
602 aerr("type parsing failed\n");
603 w += 2;
604 wordc_in = wordc - w;
605 }
606 }
607
608 if (wordc_in == 2) {
609 if (IS(words[w], "offset")) {
610 opr->type = OPT_OFFSET;
611 strcpy(opr->name, words[w + 1]);
612 return wordc;
613 }
614 if (IS(words[w], "(offset")) {
615 char *p = strchr(words[w + 1], ')');
616 if (p == NULL)
617 aerr("parse of bracketed offset failed\n");
618 *p = 0;
619 opr->type = OPT_OFFSET;
620 strcpy(opr->name, words[w + 1]);
621 return wordc;
622 }
623 }
624
625 if (wordc_in != 1)
626 aerr("parse_operand 1 word expected\n");
627
628 tmp = words[w];
629 if (IS_START(tmp, "ds:"))
630 tmp += 3;
631 strcpy(opr->name, tmp);
632
633 if (words[w][0] == '[') {
634 opr->type = OPT_REGMEM;
635 ret = sscanf(words[w], "[%[^]]]", opr->name);
636 if (ret != 1)
637 aerr("[] parse failure\n");
638
639 parse_indmode(opr->name, regmask_indirect, 1);
640 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL)) {
641 // might be an equ
642 struct parsed_equ *eq =
643 equ_find(NULL, parse_stack_el(opr->name, NULL), &i);
644 if (eq)
645 opr->lmod = eq->lmod;
646 }
647 return wordc;
648 }
649 else if (strchr(words[w], '[')) {
650 // label[reg] form
651 opr->type = OPT_REGMEM;
652 if (opr->lmod == OPLM_UNSPEC)
653 guess_lmod_from_name(opr);
654 parse_indmode(strchr(words[w], '['), regmask_indirect, 0);
655 return wordc;
656 }
657 else if (('0' <= words[w][0] && words[w][0] <= '9')
658 || words[w][0] == '-')
659 {
660 number = parse_number(words[w]);
661 opr->type = OPT_CONST;
662 opr->val = number;
663 printf_number(opr->name, sizeof(opr->name), number);
664 return wordc;
665 }
666
667 ret = parse_reg(&tmplmod, opr->name);
668 if (ret >= 0) {
669 setup_reg_opr(opr, ret, tmplmod, regmask);
670 return wordc;
671 }
672
673 // most likely var in data segment
674 opr->type = OPT_LABEL;
675
676 pp = proto_parse(g_fhdr, opr->name);
677 if (pp != NULL) {
678 if (pp->is_fptr) {
679 opr->lmod = OPLM_DWORD;
680 opr->is_ptr = 1;
681 }
682 else {
683 tmplmod = OPLM_UNSPEC;
684 if (!guess_lmod_from_c_type(&tmplmod, &pp->type))
685 anote("unhandled C type '%s' for '%s'\n",
686 pp->type.name, opr->name);
687
688 if (opr->lmod == OPLM_UNSPEC)
689 opr->lmod = tmplmod;
690 else if (opr->lmod != tmplmod) {
691 opr->size_mismatch = 1;
692 if (tmplmod < opr->lmod)
693 opr->size_lt = 1;
694 }
695 opr->is_ptr = pp->type.is_ptr;
696 }
697 opr->is_array = pp->type.is_array;
698 }
699
700 if (opr->lmod == OPLM_UNSPEC)
701 guess_lmod_from_name(opr);
702 return wordc;
703}
704
705static const struct {
706 const char *name;
707 unsigned int flags;
708} pref_table[] = {
709 { "rep", OPF_REP },
710 { "repe", OPF_REP|OPF_REPZ },
711 { "repz", OPF_REP|OPF_REPZ },
712 { "repne", OPF_REP|OPF_REPNZ },
713 { "repnz", OPF_REP|OPF_REPNZ },
714};
715
716static const struct {
717 const char *name;
718 enum op_op op;
719 unsigned int minopr;
720 unsigned int maxopr;
721 unsigned int flags;
722} op_table[] = {
723 { "nop", OP_NOP, 0, 0, 0 },
724 { "push", OP_PUSH, 1, 1, 0 },
725 { "pop", OP_POP, 1, 1, OPF_DATA },
726 { "mov" , OP_MOV, 2, 2, OPF_DATA },
727 { "lea", OP_LEA, 2, 2, OPF_DATA },
728 { "movzx",OP_MOVZX, 2, 2, OPF_DATA },
729 { "movsx",OP_MOVSX, 2, 2, OPF_DATA },
730 { "not", OP_NOT, 1, 1, OPF_DATA },
731 { "cdq", OP_CDQ, 0, 0, OPF_DATA },
732 { "stosb",OP_STOS, 0, 0, OPF_DATA },
733 { "stosw",OP_STOS, 0, 0, OPF_DATA },
734 { "stosd",OP_STOS, 0, 0, OPF_DATA },
735 { "movsb",OP_MOVS, 0, 0, OPF_DATA },
736 { "movsw",OP_MOVS, 0, 0, OPF_DATA },
737 { "movsd",OP_MOVS, 0, 0, OPF_DATA },
738 { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
739 { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
740 { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
741 { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
742 { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
743 { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
744 { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS },
745 { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS },
746 { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
747 { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
748 { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
749 { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
750 { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
751 { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
752 { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
753 { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
754 { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
755 { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS },
756 { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS },
757 { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS },
758 { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS },
759 { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS },
760 { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS },
761 { "test", OP_TEST, 2, 2, OPF_FLAGS },
762 { "cmp", OP_CMP, 2, 2, OPF_FLAGS },
763 { "retn", OP_RET, 0, 1, OPF_JMP|OPF_TAIL },
764 { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS },
765 { "jmp", OP_JMP, 1, 1, OPF_JMP },
766 { "jo", OP_JO, 1, 1, OPF_JMP|OPF_CC }, // 70 OF=1
767 { "jno", OP_JNO, 1, 1, OPF_JMP|OPF_CC }, // 71 OF=0
768 { "jc", OP_JC, 1, 1, OPF_JMP|OPF_CC }, // 72 CF=1
769 { "jb", OP_JC, 1, 1, OPF_JMP|OPF_CC }, // 72
770 { "jnc", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 CF=0
771 { "jnb", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73
772 { "jae", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73
773 { "jz", OP_JZ, 1, 1, OPF_JMP|OPF_CC }, // 74 ZF=1
774 { "je", OP_JZ, 1, 1, OPF_JMP|OPF_CC }, // 74
775 { "jnz", OP_JNZ, 1, 1, OPF_JMP|OPF_CC }, // 75 ZF=0
776 { "jne", OP_JNZ, 1, 1, OPF_JMP|OPF_CC }, // 75
777 { "jbe", OP_JBE, 1, 1, OPF_JMP|OPF_CC }, // 76 CF=1 || ZF=1
778 { "jna", OP_JBE, 1, 1, OPF_JMP|OPF_CC }, // 76
779 { "ja", OP_JA, 1, 1, OPF_JMP|OPF_CC }, // 77 CF=0 && ZF=0
780 { "jnbe", OP_JA, 1, 1, OPF_JMP|OPF_CC }, // 77
781 { "js", OP_JS, 1, 1, OPF_JMP|OPF_CC }, // 78 SF=1
782 { "jns", OP_JNS, 1, 1, OPF_JMP|OPF_CC }, // 79 SF=0
783 { "jp", OP_JP, 1, 1, OPF_JMP|OPF_CC }, // 7a PF=1
784 { "jpe", OP_JP, 1, 1, OPF_JMP|OPF_CC }, // 7a
785 { "jnp", OP_JNP, 1, 1, OPF_JMP|OPF_CC }, // 7b PF=0
786 { "jpo", OP_JNP, 1, 1, OPF_JMP|OPF_CC }, // 7b
787 { "jl", OP_JL, 1, 1, OPF_JMP|OPF_CC }, // 7c SF!=OF
788 { "jnge", OP_JL, 1, 1, OPF_JMP|OPF_CC }, // 7c
789 { "jge", OP_JGE, 1, 1, OPF_JMP|OPF_CC }, // 7d SF=OF
790 { "jnl", OP_JGE, 1, 1, OPF_JMP|OPF_CC }, // 7d
791 { "jle", OP_JLE, 1, 1, OPF_JMP|OPF_CC }, // 7e ZF=1 || SF!=OF
792 { "jng", OP_JLE, 1, 1, OPF_JMP|OPF_CC }, // 7e
793 { "jg", OP_JG, 1, 1, OPF_JMP|OPF_CC }, // 7f ZF=0 && SF=OF
794 { "jnle", OP_JG, 1, 1, OPF_JMP|OPF_CC }, // 7f
795 { "seto", OP_JO, 1, 1, OPF_DATA|OPF_CC },
796 { "setno", OP_JNO, 1, 1, OPF_DATA|OPF_CC },
797 { "setc", OP_JC, 1, 1, OPF_DATA|OPF_CC },
798 { "setb", OP_JC, 1, 1, OPF_DATA|OPF_CC },
799 { "setnc", OP_JNC, 1, 1, OPF_DATA|OPF_CC },
800 { "setae", OP_JNC, 1, 1, OPF_DATA|OPF_CC },
801 { "setz", OP_JZ, 1, 1, OPF_DATA|OPF_CC },
802 { "sete", OP_JZ, 1, 1, OPF_DATA|OPF_CC },
803 { "setnz", OP_JNZ, 1, 1, OPF_DATA|OPF_CC },
804 { "setne", OP_JNZ, 1, 1, OPF_DATA|OPF_CC },
805 { "setbe", OP_JBE, 1, 1, OPF_DATA|OPF_CC },
806 { "setna", OP_JBE, 1, 1, OPF_DATA|OPF_CC },
807 { "seta", OP_JA, 1, 1, OPF_DATA|OPF_CC },
808 { "setnbe", OP_JA, 1, 1, OPF_DATA|OPF_CC },
809 { "sets", OP_JS, 1, 1, OPF_DATA|OPF_CC },
810 { "setns", OP_JNS, 1, 1, OPF_DATA|OPF_CC },
811 { "setp", OP_JP, 1, 1, OPF_DATA|OPF_CC },
812 { "setpe", OP_JP, 1, 1, OPF_DATA|OPF_CC },
813 { "setnp", OP_JNP, 1, 1, OPF_DATA|OPF_CC },
814 { "setpo", OP_JNP, 1, 1, OPF_DATA|OPF_CC },
815 { "setl", OP_JL, 1, 1, OPF_DATA|OPF_CC },
816 { "setnge", OP_JL, 1, 1, OPF_DATA|OPF_CC },
817 { "setge", OP_JGE, 1, 1, OPF_DATA|OPF_CC },
818 { "setnl", OP_JGE, 1, 1, OPF_DATA|OPF_CC },
819 { "setle", OP_JLE, 1, 1, OPF_DATA|OPF_CC },
820 { "setng", OP_JLE, 1, 1, OPF_DATA|OPF_CC },
821 { "setg", OP_JG, 1, 1, OPF_DATA|OPF_CC },
822 { "setnle", OP_JG, 1, 1, OPF_DATA|OPF_CC },
823};
824
825static void parse_op(struct parsed_op *op, char words[16][256], int wordc)
826{
827 enum opr_lenmod lmod = OPLM_UNSPEC;
828 int prefix_flags = 0;
829 int regmask_ind;
830 int regmask;
831 int op_w = 0;
832 int opr = 0;
833 int w = 0;
834 int i;
835
836 for (i = 0; i < ARRAY_SIZE(pref_table); i++) {
837 if (IS(words[w], pref_table[i].name)) {
838 prefix_flags = pref_table[i].flags;
839 break;
840 }
841 }
842
843 if (prefix_flags) {
844 if (wordc <= 1)
845 aerr("lone prefix: '%s'\n", words[0]);
846 w++;
847 }
848
849 op_w = w;
850 for (i = 0; i < ARRAY_SIZE(op_table); i++) {
851 if (IS(words[w], op_table[i].name))
852 break;
853 }
854
855 if (i == ARRAY_SIZE(op_table))
856 aerr("unhandled op: '%s'\n", words[0]);
857 w++;
858
859 op->op = op_table[i].op;
860 op->flags = op_table[i].flags | prefix_flags;
861 op->regmask_src = op->regmask_dst = 0;
862
863 for (opr = 0; opr < op_table[i].minopr; opr++) {
864 regmask = regmask_ind = 0;
865 w = parse_operand(&op->operand[opr], &regmask, &regmask_ind,
866 words, wordc, w, op->flags);
867
868 if (opr == 0 && (op->flags & OPF_DATA))
869 op->regmask_dst = regmask;
870 // for now, mark dst as src too
871 op->regmask_src |= regmask | regmask_ind;
872 }
873
874 for (; w < wordc && opr < op_table[i].maxopr; opr++) {
875 w = parse_operand(&op->operand[opr],
876 &op->regmask_src, &op->regmask_src,
877 words, wordc, w, op->flags);
878 }
879
880 if (w < wordc)
881 aerr("parse_op %s incomplete: %d/%d\n",
882 words[0], w, wordc);
883
884 // special cases
885 op->operand_cnt = opr;
886 if (!strncmp(op_table[i].name, "set", 3))
887 op->operand[0].lmod = OPLM_BYTE;
888
889 // ops with implicit argumets
890 switch (op->op) {
891 case OP_CDQ:
892 op->operand_cnt = 2;
893 setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst);
894 setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src);
895 break;
896
897 case OP_STOS:
898 if (op->operand_cnt != 0)
899 break;
900 if (IS(words[op_w], "stosb"))
901 lmod = OPLM_BYTE;
902 else if (IS(words[op_w], "stosw"))
903 lmod = OPLM_WORD;
904 else if (IS(words[op_w], "stosd"))
905 lmod = OPLM_DWORD;
906 op->operand_cnt = 3;
907 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
908 setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src);
909 op->regmask_dst = op->regmask_src;
910 setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD, &op->regmask_src);
911 break;
912
913 case OP_MOVS:
914 case OP_CMPS:
915 if (op->operand_cnt != 0)
916 break;
917 if (words[op_w][4] == 'b')
918 lmod = OPLM_BYTE;
919 else if (words[op_w][4] == 'w')
920 lmod = OPLM_WORD;
921 else if (words[op_w][4] == 'd')
922 lmod = OPLM_DWORD;
923 op->operand_cnt = 3;
924 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
925 setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src);
926 setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src);
927 op->regmask_dst = op->regmask_src;
928 break;
929
930 case OP_IMUL:
931 if (op->operand_cnt != 1)
932 break;
933 // fallthrough
934 case OP_MUL:
935 // singleop mul
936 op->regmask_dst = (1 << xDX) | (1 << xAX);
937 op->regmask_src |= (1 << xAX);
938 if (op->operand[0].lmod == OPLM_UNSPEC)
939 op->operand[0].lmod = OPLM_DWORD;
940 break;
941
942 case OP_DIV:
943 case OP_IDIV:
944 // we could set up operands for edx:eax, but there is no real need to
945 // (see is_opr_modified())
946 regmask = (1 << xDX) | (1 << xAX);
947 op->regmask_dst = regmask;
948 op->regmask_src |= regmask;
949 if (op->operand[0].lmod == OPLM_UNSPEC)
950 op->operand[0].lmod = OPLM_DWORD;
951 break;
952
953 case OP_SHL:
954 case OP_SHR:
955 case OP_SAR:
956 case OP_ROL:
957 case OP_ROR:
958 if (op->operand[1].lmod == OPLM_UNSPEC)
959 op->operand[1].lmod = OPLM_BYTE;
960 break;
961
962 case OP_PUSH:
963 if (op->operand[0].lmod == OPLM_UNSPEC
964 && (op->operand[0].type == OPT_CONST
965 || op->operand[0].type == OPT_OFFSET
966 || op->operand[0].type == OPT_LABEL))
967 op->operand[0].lmod = OPLM_DWORD;
968 break;
969
970 // alignment
971 case OP_MOV:
972 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
973 && op->operand[0].reg == xDI && op->operand[1].reg == xDI)
974 {
975 op->flags |= OPF_RMD;
976 }
977 break;
978
979 case OP_LEA:
980 if (op->operand[0].type == OPT_REG
981 && op->operand[1].type == OPT_REGMEM)
982 {
983 char buf[16];
984 snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name);
985 if (IS(buf, op->operand[1].name))
986 op->flags |= OPF_RMD;
987 }
988 break;
989
990 default:
991 break;
992 }
993}
994
995static const char *op_name(enum op_op op)
996{
997 int i;
998
999 for (i = 0; i < ARRAY_SIZE(op_table); i++)
1000 if (op_table[i].op == op)
1001 return op_table[i].name;
1002
1003 return "???";
1004}
1005
1006// debug
1007static const char *dump_op(struct parsed_op *po)
1008{
1009 static char out[128];
1010 char *p = out;
1011 int i;
1012
1013 if (po == NULL)
1014 return "???";
1015
1016 snprintf(out, sizeof(out), "%s", op_name(po->op));
1017 for (i = 0; i < po->operand_cnt; i++) {
1018 p += strlen(p);
1019 if (i > 0)
1020 *p++ = ',';
1021 snprintf(p, sizeof(out) - (p - out),
1022 po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s",
1023 po->operand[i].name);
1024 }
1025
1026 return out;
1027}
1028
1029static const char *lmod_type_u(struct parsed_op *po,
1030 enum opr_lenmod lmod)
1031{
1032 switch (lmod) {
1033 case OPLM_DWORD:
1034 return "u32";
1035 case OPLM_WORD:
1036 return "u16";
1037 case OPLM_BYTE:
1038 return "u8";
1039 default:
1040 ferr(po, "invalid lmod: %d\n", lmod);
1041 return "(_invalid_)";
1042 }
1043}
1044
1045static const char *lmod_cast_u(struct parsed_op *po,
1046 enum opr_lenmod lmod)
1047{
1048 switch (lmod) {
1049 case OPLM_DWORD:
1050 return "";
1051 case OPLM_WORD:
1052 return "(u16)";
1053 case OPLM_BYTE:
1054 return "(u8)";
1055 default:
1056 ferr(po, "invalid lmod: %d\n", lmod);
1057 return "(_invalid_)";
1058 }
1059}
1060
1061static const char *lmod_cast_u_ptr(struct parsed_op *po,
1062 enum opr_lenmod lmod)
1063{
1064 switch (lmod) {
1065 case OPLM_DWORD:
1066 return "*(u32 *)";
1067 case OPLM_WORD:
1068 return "*(u16 *)";
1069 case OPLM_BYTE:
1070 return "*(u8 *)";
1071 default:
1072 ferr(po, "invalid lmod: %d\n", lmod);
1073 return "(_invalid_)";
1074 }
1075}
1076
1077static const char *lmod_cast_s(struct parsed_op *po,
1078 enum opr_lenmod lmod)
1079{
1080 switch (lmod) {
1081 case OPLM_DWORD:
1082 return "(s32)";
1083 case OPLM_WORD:
1084 return "(s16)";
1085 case OPLM_BYTE:
1086 return "(s8)";
1087 default:
1088 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1089 return "(_invalid_)";
1090 }
1091}
1092
1093static const char *lmod_cast(struct parsed_op *po,
1094 enum opr_lenmod lmod, int is_signed)
1095{
1096 return is_signed ?
1097 lmod_cast_s(po, lmod) :
1098 lmod_cast_u(po, lmod);
1099}
1100
1101static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
1102{
1103 switch (lmod) {
1104 case OPLM_DWORD:
1105 return 4;
1106 case OPLM_WORD:
1107 return 2;
1108 case OPLM_BYTE:
1109 return 1;
1110 default:
1111 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1112 return 0;
1113 }
1114}
1115
1116static const char *opr_name(struct parsed_op *po, int opr_num)
1117{
1118 if (opr_num >= po->operand_cnt)
1119 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1120 return po->operand[opr_num].name;
1121}
1122
1123static unsigned int opr_const(struct parsed_op *po, int opr_num)
1124{
1125 if (opr_num >= po->operand_cnt)
1126 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1127 if (po->operand[opr_num].type != OPT_CONST)
1128 ferr(po, "opr %d: const expected\n", opr_num);
1129 return po->operand[opr_num].val;
1130}
1131
1132static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr)
1133{
1134 if ((unsigned int)popr->reg >= MAX_REGS)
1135 ferr(po, "invalid reg: %d\n", popr->reg);
1136 return regs_r32[popr->reg];
1137}
1138
1139static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
1140 int *extra_offs)
1141{
1142 const char *p;
1143 char *endp;
1144 int namelen;
1145 int i;
1146
1147 *extra_offs = 0;
1148 namelen = strlen(name);
1149
1150 p = strchr(name, '+');
1151 if (p != NULL) {
1152 namelen = p - name;
1153 if (namelen <= 0)
1154 ferr(po, "equ parse failed for '%s'\n", name);
1155
1156 if (IS_START(p, "0x"))
1157 p += 2;
1158 *extra_offs = strtol(p, &endp, 16);
1159 if (*endp != 0)
1160 ferr(po, "equ parse failed for '%s'\n", name);
1161 }
1162
1163 for (i = 0; i < g_eqcnt; i++)
1164 if (strncmp(g_eqs[i].name, name, namelen) == 0
1165 && g_eqs[i].name[namelen] == 0)
1166 break;
1167 if (i >= g_eqcnt) {
1168 if (po != NULL)
1169 ferr(po, "unresolved equ name: '%s'\n", name);
1170 return NULL;
1171 }
1172
1173 return &g_eqs[i];
1174}
1175
1176static void stack_frame_access(struct parsed_op *po,
1177 struct parsed_opr *popr, char *buf, size_t buf_size,
1178 const char *name, const char *cast, int is_src, int is_lea)
1179{
1180 enum opr_lenmod tmp_lmod = OPLM_UNSPEC;
1181 const char *prefix = "";
1182 const char *bp_arg = NULL;
1183 char ofs_reg[16] = { 0, };
1184 struct parsed_equ *eq;
1185 const char *p;
1186 char *endp = NULL;
1187 int i, arg_i, arg_s;
1188 int stack_ra = 0;
1189 int offset = 0;
1190 int sf_ofs;
1191 int lim;
1192
1193 if (!IS_START(name, "ebp-")) {
1194 bp_arg = parse_stack_el(name, ofs_reg);
1195 snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
1196 eq = equ_find(po, bp_arg, &offset);
1197 if (eq == NULL)
1198 ferr(po, "detected but missing eq\n");
1199 offset += eq->offset;
1200 }
1201 else {
1202 p = name + 4;
1203 if (IS_START(p, "0x"))
1204 p += 2;
1205 offset = -strtoul(p, &endp, 16);
1206 if (*endp != 0)
1207 ferr(po, "ebp- parse of '%s' failed\n", name);
1208 }
1209
1210 if (!strncmp(name, "ebp", 3))
1211 stack_ra = 4;
1212
1213 if (stack_ra <= offset && offset < stack_ra + 4)
1214 ferr(po, "reference to ra? %d %d\n", offset, stack_ra);
1215
1216 if (offset > stack_ra)
1217 {
1218 arg_i = (offset - stack_ra - 4) / 4;
1219 if (arg_i < 0 || arg_i >= g_func_pp->argc_stack)
1220 {
1221 if (g_func_pp->is_vararg
1222 && arg_i == g_func_pp->argc_stack && is_lea)
1223 {
1224 // should be va_list
1225 if (cast[0] == 0)
1226 cast = "(u32)";
1227 snprintf(buf, buf_size, "%sap", cast);
1228 return;
1229 }
1230 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1231 offset, bp_arg, arg_i);
1232 }
1233 if (ofs_reg[0] != 0)
1234 ferr(po, "offset reg on arg access?\n");
1235
1236 for (i = arg_s = 0; i < g_func_pp->argc; i++) {
1237 if (g_func_pp->arg[i].reg != NULL)
1238 continue;
1239 if (arg_s == arg_i)
1240 break;
1241 arg_s++;
1242 }
1243 if (i == g_func_pp->argc)
1244 ferr(po, "arg %d not in prototype?\n", arg_i);
1245
1246 popr->is_ptr = g_func_pp->arg[i].type.is_ptr;
1247
1248 switch (popr->lmod)
1249 {
1250 case OPLM_BYTE:
1251 if (is_lea)
1252 ferr(po, "lea/byte to arg?\n");
1253 if (is_src && (offset & 3) == 0)
1254 snprintf(buf, buf_size, "(u8)a%d", i + 1);
1255 else
1256 snprintf(buf, buf_size, "BYTE%d(a%d)", offset & 3, i + 1);
1257 break;
1258
1259 case OPLM_WORD:
1260 if (is_lea)
1261 ferr(po, "lea/word to arg?\n");
1262 if (offset & 1)
1263 ferr(po, "unaligned arg access\n");
1264 if (is_src && (offset & 2) == 0)
1265 snprintf(buf, buf_size, "(u16)a%d", i + 1);
1266 else
1267 snprintf(buf, buf_size, "%sWORD(a%d)",
1268 (offset & 2) ? "HI" : "LO", i + 1);
1269 break;
1270
1271 case OPLM_DWORD:
1272 if (cast[0])
1273 prefix = cast;
1274 else if (is_src)
1275 prefix = "(u32)";
1276 if (offset & 3) {
1277 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1278 if (is_lea)
1279 snprintf(buf, buf_size, "(u32)&a%d + %d",
1280 i + 1, offset & 3);
1281 else
1282 snprintf(buf, buf_size, "%s(a%d >> %d)",
1283 prefix, i + 1, (offset & 3) * 8);
1284 }
1285 else {
1286 snprintf(buf, buf_size, "%s%sa%d",
1287 prefix, is_lea ? "&" : "", i + 1);
1288 }
1289 break;
1290
1291 default:
1292 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
1293 }
1294
1295 // common problem
1296 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type);
1297 if ((offset & 3) && tmp_lmod != OPLM_DWORD)
1298 ferr(po, "bp_arg arg/w offset %d and type '%s'\n",
1299 offset, g_func_pp->arg[i].type.name);
1300 }
1301 else
1302 {
1303 if (g_stack_fsz == 0)
1304 ferr(po, "stack var access without stackframe\n");
1305
1306 sf_ofs = g_stack_fsz + offset;
1307 lim = (ofs_reg[0] != 0) ? -4 : 0;
1308 if (offset > 0 || sf_ofs < lim)
1309 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
1310
1311 if (is_lea)
1312 prefix = "(u32)&";
1313 else
1314 prefix = cast;
1315
1316 switch (popr->lmod)
1317 {
1318 case OPLM_BYTE:
1319 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1320 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1321 break;
1322
1323 case OPLM_WORD:
1324 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1325 // known unaligned or possibly unaligned
1326 strcat(g_comment, " unaligned");
1327 if (prefix[0] == 0)
1328 prefix = "*(u16 *)&";
1329 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1330 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1331 break;
1332 }
1333 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
1334 break;
1335
1336 case OPLM_DWORD:
1337 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1338 // known unaligned or possibly unaligned
1339 strcat(g_comment, " unaligned");
1340 if (prefix[0] == 0)
1341 prefix = "*(u32 *)&";
1342 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1343 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1344 break;
1345 }
1346 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
1347 break;
1348
1349 default:
1350 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
1351 }
1352 }
1353}
1354
1355static void check_label_read_ref(struct parsed_op *po, const char *name)
1356{
1357 const struct parsed_proto *pp;
1358
1359 pp = proto_parse(g_fhdr, name);
1360 if (pp == NULL)
1361 ferr(po, "proto_parse failed for ref '%s'\n", name);
1362
1363 // currently we can take __cdecl and __stdcall
1364 if (pp->is_func && pp->argc_reg != 0)
1365 ferr(po, "reg-arg func reference?\n");
1366}
1367
1368static char *out_src_opr(char *buf, size_t buf_size,
1369 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
1370 int is_lea)
1371{
1372 char tmp1[256], tmp2[256];
1373 char expr[256];
1374 int ret;
1375
1376 if (cast == NULL)
1377 cast = "";
1378
1379 switch (popr->type) {
1380 case OPT_REG:
1381 if (is_lea)
1382 ferr(po, "lea from reg?\n");
1383
1384 switch (popr->lmod) {
1385 case OPLM_DWORD:
1386 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
1387 break;
1388 case OPLM_WORD:
1389 snprintf(buf, buf_size, "(u16)%s", opr_reg_p(po, popr));
1390 break;
1391 case OPLM_BYTE:
1392 if (popr->name[1] == 'h') // XXX..
1393 snprintf(buf, buf_size, "(u8)(%s >> 8)", opr_reg_p(po, popr));
1394 else
1395 snprintf(buf, buf_size, "(u8)%s", opr_reg_p(po, popr));
1396 break;
1397 default:
1398 ferr(po, "invalid src lmod: %d\n", popr->lmod);
1399 }
1400 break;
1401
1402 case OPT_REGMEM:
1403 if (parse_stack_el(popr->name, NULL)
1404 || (g_bp_frame && IS_START(popr->name, "ebp-")))
1405 {
1406 stack_frame_access(po, popr, buf, buf_size,
1407 popr->name, cast, 1, is_lea);
1408 break;
1409 }
1410
1411 strcpy(expr, popr->name);
1412 if (strchr(expr, '[')) {
1413 // special case: '[' can only be left for label[reg] form
1414 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
1415 if (ret != 2)
1416 ferr(po, "parse failure for '%s'\n", expr);
1417 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
1418 }
1419
1420 // XXX: do we need more parsing?
1421 if (is_lea) {
1422 snprintf(buf, buf_size, "%s", expr);
1423 break;
1424 }
1425
1426 if (cast[0] == 0)
1427 cast = lmod_cast_u_ptr(po, popr->lmod);
1428 snprintf(buf, buf_size, "%s(%s)", cast, expr);
1429 break;
1430
1431 case OPT_LABEL:
1432 check_label_read_ref(po, popr->name);
1433 if (cast[0] == 0 && popr->is_ptr)
1434 cast = "(u32)";
1435
1436 if (is_lea)
1437 snprintf(buf, buf_size, "(u32)&%s", popr->name);
1438 else if (popr->size_lt)
1439 snprintf(buf, buf_size, "%s%s%s%s", cast,
1440 lmod_cast_u_ptr(po, popr->lmod),
1441 popr->is_array ? "" : "&",
1442 popr->name);
1443 else
1444 snprintf(buf, buf_size, "%s%s%s", cast, popr->name,
1445 popr->is_array ? "[0]" : "");
1446 break;
1447
1448 case OPT_OFFSET:
1449 check_label_read_ref(po, popr->name);
1450 if (cast[0] == 0)
1451 cast = "(u32)";
1452 if (is_lea)
1453 ferr(po, "lea an offset?\n");
1454 snprintf(buf, buf_size, "%s&%s", cast, popr->name);
1455 break;
1456
1457 case OPT_CONST:
1458 if (is_lea)
1459 ferr(po, "lea from const?\n");
1460
1461 snprintf(buf, buf_size, "%s", cast);
1462 ret = strlen(buf);
1463 printf_number(buf + ret, buf_size - ret, popr->val);
1464 break;
1465
1466 default:
1467 ferr(po, "invalid src type: %d\n", popr->type);
1468 }
1469
1470 return buf;
1471}
1472
1473// note: may set is_ptr (we find that out late for ebp frame..)
1474static char *out_dst_opr(char *buf, size_t buf_size,
1475 struct parsed_op *po, struct parsed_opr *popr)
1476{
1477 switch (popr->type) {
1478 case OPT_REG:
1479 switch (popr->lmod) {
1480 case OPLM_DWORD:
1481 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
1482 break;
1483 case OPLM_WORD:
1484 // ugh..
1485 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
1486 break;
1487 case OPLM_BYTE:
1488 // ugh..
1489 if (popr->name[1] == 'h') // XXX..
1490 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
1491 else
1492 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
1493 break;
1494 default:
1495 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
1496 }
1497 break;
1498
1499 case OPT_REGMEM:
1500 if (parse_stack_el(popr->name, NULL)
1501 || (g_bp_frame && IS_START(popr->name, "ebp-")))
1502 {
1503 stack_frame_access(po, popr, buf, buf_size,
1504 popr->name, "", 0, 0);
1505 break;
1506 }
1507
1508 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1509
1510 case OPT_LABEL:
1511 if (popr->size_mismatch)
1512 snprintf(buf, buf_size, "%s%s%s",
1513 lmod_cast_u_ptr(po, popr->lmod),
1514 popr->is_array ? "" : "&", popr->name);
1515 else
1516 snprintf(buf, buf_size, "%s%s", popr->name,
1517 popr->is_array ? "[0]" : "");
1518 break;
1519
1520 default:
1521 ferr(po, "invalid dst type: %d\n", popr->type);
1522 }
1523
1524 return buf;
1525}
1526
1527static char *out_src_opr_u32(char *buf, size_t buf_size,
1528 struct parsed_op *po, struct parsed_opr *popr)
1529{
1530 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1531}
1532
1533static enum parsed_flag_op split_cond(struct parsed_op *po,
1534 enum op_op op, int *is_inv)
1535{
1536 *is_inv = 0;
1537
1538 switch (op) {
1539 case OP_JO:
1540 return PFO_O;
1541 case OP_JC:
1542 return PFO_C;
1543 case OP_JZ:
1544 return PFO_Z;
1545 case OP_JBE:
1546 return PFO_BE;
1547 case OP_JS:
1548 return PFO_S;
1549 case OP_JP:
1550 return PFO_P;
1551 case OP_JL:
1552 return PFO_L;
1553 case OP_JLE:
1554 return PFO_LE;
1555
1556 case OP_JNO:
1557 *is_inv = 1;
1558 return PFO_O;
1559 case OP_JNC:
1560 *is_inv = 1;
1561 return PFO_C;
1562 case OP_JNZ:
1563 *is_inv = 1;
1564 return PFO_Z;
1565 case OP_JA:
1566 *is_inv = 1;
1567 return PFO_BE;
1568 case OP_JNS:
1569 *is_inv = 1;
1570 return PFO_S;
1571 case OP_JNP:
1572 *is_inv = 1;
1573 return PFO_P;
1574 case OP_JGE:
1575 *is_inv = 1;
1576 return PFO_L;
1577 case OP_JG:
1578 *is_inv = 1;
1579 return PFO_LE;
1580
1581 case OP_ADC:
1582 case OP_SBB:
1583 return PFO_C;
1584
1585 default:
1586 ferr(po, "split_cond: bad op %d\n", op);
1587 return -1;
1588 }
1589}
1590
1591static void out_test_for_cc(char *buf, size_t buf_size,
1592 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
1593 enum opr_lenmod lmod, const char *expr)
1594{
1595 const char *cast, *scast;
1596
1597 cast = lmod_cast_u(po, lmod);
1598 scast = lmod_cast_s(po, lmod);
1599
1600 switch (pfo) {
1601 case PFO_Z:
1602 case PFO_BE: // CF=1||ZF=1; CF=0
1603 snprintf(buf, buf_size, "(%s%s %s 0)",
1604 cast, expr, is_inv ? "!=" : "==");
1605 break;
1606
1607 case PFO_S:
1608 case PFO_L: // SF!=OF; OF=0
1609 snprintf(buf, buf_size, "(%s%s %s 0)",
1610 scast, expr, is_inv ? ">=" : "<");
1611 break;
1612
1613 case PFO_LE: // ZF=1||SF!=OF; OF=0
1614 snprintf(buf, buf_size, "(%s%s %s 0)",
1615 scast, expr, is_inv ? ">" : "<=");
1616 break;
1617
1618 default:
1619 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1620 }
1621}
1622
1623static void out_cmp_for_cc(char *buf, size_t buf_size,
1624 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
1625 enum opr_lenmod lmod, const char *expr1, const char *expr2)
1626{
1627 const char *cast, *scast;
1628
1629 cast = lmod_cast_u(po, lmod);
1630 scast = lmod_cast_s(po, lmod);
1631
1632 switch (pfo) {
1633 case PFO_C:
1634 // note: must be unsigned compare
1635 snprintf(buf, buf_size, "(%s%s %s %s%s)",
1636 cast, expr1, is_inv ? ">=" : "<", cast, expr2);
1637 break;
1638
1639 case PFO_Z:
1640 snprintf(buf, buf_size, "(%s%s %s %s%s)",
1641 cast, expr1, is_inv ? "!=" : "==", cast, expr2);
1642 break;
1643
1644 case PFO_BE: // !a
1645 // note: must be unsigned compare
1646 snprintf(buf, buf_size, "(%s%s %s %s%s)",
1647 cast, expr1, is_inv ? ">" : "<=", cast, expr2);
1648 break;
1649
1650 // note: must be signed compare
1651 case PFO_S:
1652 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
1653 scast, expr1, expr2, is_inv ? ">=" : "<");
1654 break;
1655
1656 case PFO_L: // !ge
1657 snprintf(buf, buf_size, "(%s%s %s %s%s)",
1658 scast, expr1, is_inv ? ">=" : "<", scast, expr2);
1659 break;
1660
1661 case PFO_LE:
1662 snprintf(buf, buf_size, "(%s%s %s %s%s)",
1663 scast, expr1, is_inv ? ">" : "<=", scast, expr2);
1664 break;
1665
1666 default:
1667 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1668 }
1669}
1670
1671static void out_cmp_test(char *buf, size_t buf_size,
1672 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
1673{
1674 char buf1[256], buf2[256], buf3[256];
1675
1676 if (po->op == OP_TEST) {
1677 if (IS(opr_name(po, 0), opr_name(po, 1))) {
1678 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
1679 }
1680 else {
1681 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
1682 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
1683 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
1684 }
1685 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
1686 po->operand[0].lmod, buf3);
1687 }
1688 else if (po->op == OP_CMP) {
1689 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
1690 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[1]);
1691 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv,
1692 po->operand[0].lmod, buf2, buf3);
1693 }
1694 else
1695 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
1696}
1697
1698static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
1699 struct parsed_opr *popr2)
1700{
1701 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
1702 ferr(po, "missing lmod for both operands\n");
1703
1704 if (popr1->lmod == OPLM_UNSPEC)
1705 popr1->lmod = popr2->lmod;
1706 else if (popr2->lmod == OPLM_UNSPEC)
1707 popr2->lmod = popr1->lmod;
1708 else if (popr1->lmod != popr2->lmod)
1709 ferr(po, "conflicting lmods: %d vs %d\n", popr1->lmod, popr2->lmod);
1710}
1711
1712static const char *op_to_c(struct parsed_op *po)
1713{
1714 switch (po->op)
1715 {
1716 case OP_ADD:
1717 case OP_ADC:
1718 return "+";
1719 case OP_SUB:
1720 case OP_SBB:
1721 return "-";
1722 case OP_AND:
1723 return "&";
1724 case OP_OR:
1725 return "|";
1726 case OP_XOR:
1727 return "^";
1728 case OP_SHL:
1729 return "<<";
1730 case OP_SHR:
1731 return ">>";
1732 case OP_MUL:
1733 case OP_IMUL:
1734 return "*";
1735 default:
1736 ferr(po, "op_to_c was supplied with %d\n", po->op);
1737 }
1738}
1739
1740static void set_flag_no_dup(struct parsed_op *po, enum op_flags flag,
1741 enum op_flags flag_check)
1742{
1743 if (po->flags & flag)
1744 ferr(po, "flag %x already set\n", flag);
1745 if (po->flags & flag_check)
1746 ferr(po, "flag_check %x already set\n", flag_check);
1747
1748 po->flags |= flag;
1749}
1750
1751// last op in stream - unconditional branch or ret
1752#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
1753 || (ops[_i].flags & (OPF_JMP|OPF_CC)) == OPF_JMP)
1754
1755static int scan_for_pop(int i, int opcnt, const char *reg,
1756 int magic, int depth, int *maxdepth, int do_flags)
1757{
1758 struct parsed_op *po;
1759 int ret = 0;
1760 int j;
1761
1762 for (; i < opcnt; i++) {
1763 po = &ops[i];
1764 if (po->cc_scratch == magic)
1765 break; // already checked
1766 po->cc_scratch = magic;
1767
1768 if (po->flags & OPF_TAIL)
1769 return -1; // deadend
1770
1771 if ((po->flags & OPF_RMD)
1772 || (po->op == OP_PUSH && po->argnum != 0)) // arg push
1773 continue;
1774
1775 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
1776 if (po->btj != NULL) {
1777 // jumptable
1778 for (j = 0; j < po->btj->count - 1; j++) {
1779 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic,
1780 depth, maxdepth, do_flags);
1781 if (ret < 0)
1782 return ret; // dead end
1783 }
1784 // follow last jumptable entry
1785 i = po->btj->d[j].bt_i - 1;
1786 continue;
1787 }
1788
1789 if (po->bt_i < 0) {
1790 ferr(po, "dead branch\n");
1791 return -1;
1792 }
1793
1794 if (po->flags & OPF_CC) {
1795 ret |= scan_for_pop(po->bt_i, opcnt, reg, magic,
1796 depth, maxdepth, do_flags);
1797 if (ret < 0)
1798 return ret; // dead end
1799 }
1800 else {
1801 i = po->bt_i - 1;
1802 }
1803 continue;
1804 }
1805
1806 if ((po->op == OP_POP || po->op == OP_PUSH)
1807 && po->operand[0].type == OPT_REG
1808 && IS(po->operand[0].name, reg))
1809 {
1810 if (po->op == OP_PUSH) {
1811 depth++;
1812 if (depth > *maxdepth)
1813 *maxdepth = depth;
1814 if (do_flags)
1815 set_flag_no_dup(po, OPF_RSAVE, OPF_RMD);
1816 }
1817 else if (depth == 0) {
1818 if (do_flags)
1819 set_flag_no_dup(po, OPF_RMD, OPF_RSAVE);
1820 return 1;
1821 }
1822 else {
1823 depth--;
1824 if (depth < 0) // should not happen
1825 ferr(po, "fail with depth\n");
1826 if (do_flags)
1827 set_flag_no_dup(po, OPF_RSAVE, OPF_RMD);
1828 }
1829 }
1830 }
1831
1832 return ret;
1833}
1834
1835// scan for pop starting from 'ret' op (all paths)
1836static int scan_for_pop_ret(int i, int opcnt, const char *reg,
1837 int flag_set)
1838{
1839 int found = 0;
1840 int j;
1841
1842 for (; i < opcnt; i++) {
1843 if (!(ops[i].flags & OPF_TAIL))
1844 continue;
1845
1846 for (j = i - 1; j >= 0; j--) {
1847 if (ops[j].flags & OPF_RMD)
1848 continue;
1849 if (ops[j].flags & OPF_JMP)
1850 return -1;
1851
1852 if (ops[j].op == OP_POP && ops[j].operand[0].type == OPT_REG
1853 && IS(ops[j].operand[0].name, reg))
1854 {
1855 found = 1;
1856 ops[j].flags |= flag_set;
1857 break;
1858 }
1859
1860 if (g_labels[j][0] != 0)
1861 return -1;
1862 }
1863 }
1864
1865 return found ? 0 : -1;
1866}
1867
1868// is operand 'opr modified' by parsed_op 'po'?
1869static int is_opr_modified(const struct parsed_opr *opr,
1870 const struct parsed_op *po)
1871{
1872 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
1873 return 0;
1874
1875 if (opr->type == OPT_REG && po->operand[0].type == OPT_REG) {
1876 if (po->regmask_dst & (1 << opr->reg))
1877 return 1;
1878 else
1879 return 0;
1880 }
1881
1882 return IS(po->operand[0].name, opr->name);
1883}
1884
1885// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
1886static int is_any_opr_modified(const struct parsed_op *po_test,
1887 const struct parsed_op *po)
1888{
1889 int i;
1890
1891 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
1892 return 0;
1893
1894 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
1895 return 0;
1896
1897 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
1898 return 1;
1899
1900 // in reality, it can wreck any register, but in decompiled C
1901 // version it can only overwrite eax or edx:eax
1902 if (po->op == OP_CALL
1903 && ((po_test->regmask_src | po_test->regmask_dst)
1904 & ((1 << xAX)|(1 << xDX))))
1905 return 1;
1906
1907 for (i = 0; i < po_test->operand_cnt; i++)
1908 if (IS(po_test->operand[i].name, po->operand[0].name))
1909 return 1;
1910
1911 return 0;
1912}
1913
1914// scan for any po_test operand modification in range given
1915static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt)
1916{
1917 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
1918 return -1;
1919
1920 for (; i < opcnt; i++) {
1921 if (is_any_opr_modified(po_test, &ops[i]))
1922 return i;
1923 }
1924
1925 return -1;
1926}
1927
1928// scan for po_test operand[0] modification in range given
1929static int scan_for_mod_opr0(struct parsed_op *po_test,
1930 int i, int opcnt)
1931{
1932 for (; i < opcnt; i++) {
1933 if (is_opr_modified(&po_test->operand[0], &ops[i]))
1934 return i;
1935 }
1936
1937 return -1;
1938}
1939
1940static int scan_for_flag_set(int i, int magic, int *branched,
1941 int *setters, int *setter_cnt)
1942{
1943 struct label_ref *lr;
1944 int ret;
1945
1946 while (i >= 0) {
1947 if (ops[i].cc_scratch == magic) {
1948 ferr(&ops[i], "%s looped\n", __func__);
1949 return -1;
1950 }
1951 ops[i].cc_scratch = magic;
1952
1953 if (g_labels[i][0] != 0) {
1954 *branched = 1;
1955
1956 lr = &g_label_refs[i];
1957 for (; lr->next; lr = lr->next) {
1958 ret = scan_for_flag_set(lr->i, magic,
1959 branched, setters, setter_cnt);
1960 if (ret < 0)
1961 return ret;
1962 }
1963
1964 if (i > 0 && LAST_OP(i - 1)) {
1965 i = g_label_refs[i].i;
1966 continue;
1967 }
1968 ret = scan_for_flag_set(lr->i, magic,
1969 branched, setters, setter_cnt);
1970 if (ret < 0)
1971 return ret;
1972 }
1973 i--;
1974
1975 if (ops[i].flags & OPF_FLAGS) {
1976 setters[*setter_cnt] = i;
1977 (*setter_cnt)++;
1978 return 0;
1979 }
1980
1981 if ((ops[i].flags & OPF_JMP) && !(ops[i].flags & OPF_CC))
1982 return -1;
1983 }
1984
1985 return -1;
1986}
1987
1988// scan back for cdq, if anything modifies edx, fail
1989static int scan_for_cdq_edx(int i)
1990{
1991 for (; i >= 0; i--) {
1992 if (ops[i].op == OP_CDQ)
1993 return i;
1994
1995 if (ops[i].regmask_dst & (1 << xDX))
1996 return -1;
1997 if (g_labels[i][0] != 0)
1998 return -1;
1999 }
2000
2001 return -1;
2002}
2003
2004static int scan_for_reg_clear(int i, int reg)
2005{
2006 for (; i >= 0; i--) {
2007 if (ops[i].op == OP_XOR
2008 && ops[i].operand[0].lmod == OPLM_DWORD
2009 && ops[i].operand[0].reg == ops[i].operand[1].reg
2010 && ops[i].operand[0].reg == reg)
2011 return i;
2012
2013 if (ops[i].regmask_dst & (1 << reg))
2014 return -1;
2015 if (g_labels[i][0] != 0)
2016 return -1;
2017 }
2018
2019 return -1;
2020}
2021
2022// scan for positive, constant esp adjust
2023static int scan_for_esp_adjust(int i, int opcnt, int *adj)
2024{
2025 struct parsed_op *po;
2026 *adj = 0;
2027
2028 for (; i < opcnt; i++) {
2029 po = &ops[i];
2030
2031 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
2032 if (po->operand[1].type != OPT_CONST)
2033 ferr(&ops[i], "non-const esp adjust?\n");
2034 *adj += po->operand[1].val;
2035 if (*adj & 3)
2036 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
2037 return i;
2038 }
2039 else if (po->op == OP_PUSH)
2040 *adj -= lmod_bytes(po, po->operand[0].lmod);
2041 else if (po->op == OP_POP)
2042 *adj += lmod_bytes(po, po->operand[0].lmod);
2043 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
2044 if (po->op != OP_CALL)
2045 return -1;
2046 if (po->operand[0].type != OPT_LABEL)
2047 return -1;
2048 // TODO: should only allow combining __cdecl calls..
2049 }
2050
2051 if (g_labels[i][0] != 0)
2052 return -1;
2053 }
2054
2055 return -1;
2056}
2057
2058static int collect_call_args(struct parsed_op *po, int i,
2059 struct parsed_proto *pp, int *save_arg_vars, int arg,
2060 int need_op_saving, int may_reuse)
2061{
2062 struct parsed_proto *pp_tmp;
2063 struct label_ref *lr;
2064 int need_to_save_current;
2065 int ret = 0;
2066 int j;
2067
2068 if (i < 0)
2069 ferr(po, "no refs for '%s'?\n", g_labels[i]);
2070
2071 for (; arg < pp->argc; arg++)
2072 if (pp->arg[arg].reg == NULL)
2073 break;
2074
2075 for (j = i; j >= 0 && arg < pp->argc; )
2076 {
2077 if (g_labels[j][0] != 0) {
2078 lr = &g_label_refs[j];
2079 if (lr->next != NULL)
2080 need_op_saving = 1;
2081 for (; lr->next; lr = lr->next) {
2082 if ((ops[lr->i].flags & (OPF_JMP|OPF_CC)) != OPF_JMP)
2083 may_reuse = 1;
2084 ret |= collect_call_args(po, lr->i, pp, save_arg_vars,
2085 arg, need_op_saving, may_reuse);
2086 }
2087
2088 if ((ops[lr->i].flags & (OPF_JMP|OPF_CC)) != OPF_JMP)
2089 may_reuse = 1;
2090 if (j > 0 && LAST_OP(j - 1)) {
2091 // follow last branch in reverse
2092 j = lr->i;
2093 continue;
2094 }
2095 need_op_saving = 1;
2096 ret |= collect_call_args(po, lr->i, pp, save_arg_vars,
2097 arg, need_op_saving, may_reuse);
2098 }
2099 j--;
2100
2101 if (ops[j].op == OP_CALL)
2102 {
2103 pp_tmp = ops[j].datap;
2104 if (pp_tmp == NULL)
2105 ferr(po, "arg collect hit unparsed call\n");
2106 if (may_reuse && pp_tmp->argc_stack > 0)
2107 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
2108 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
2109 }
2110 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP) {
2111 ferr(po, "arg collect %d/%d hit esp adjust\n",
2112 arg, pp->argc);
2113 }
2114 else if (ops[j].op == OP_POP) {
2115 ferr(po, "arg collect %d/%d hit pop\n", arg, pp->argc);
2116 }
2117 else if ((ops[j].flags & (OPF_JMP|OPF_CC)) == (OPF_JMP|OPF_CC))
2118 {
2119 may_reuse = 1;
2120 }
2121 else if (ops[j].op == OP_PUSH && !(ops[j].flags & OPF_FARG))
2122 {
2123 pp->arg[arg].datap = &ops[j];
2124 need_to_save_current = 0;
2125 if (!need_op_saving) {
2126 ret = scan_for_mod(&ops[j], j + 1, i);
2127 need_to_save_current = (ret >= 0);
2128 }
2129 if (need_op_saving || need_to_save_current) {
2130 // mark this push as one that needs operand saving
2131 ops[j].flags &= ~OPF_RMD;
2132 if (ops[j].argnum == 0) {
2133 ops[j].argnum = arg + 1;
2134 *save_arg_vars |= 1 << arg;
2135 }
2136 else if (ops[j].argnum < arg + 1)
2137 ferr(&ops[j], "argnum conflict (%d<%d) for '%s'\n",
2138 ops[j].argnum, arg + 1, pp->name);
2139 }
2140 else if (ops[j].argnum == 0)
2141 ops[j].flags |= OPF_RMD;
2142
2143 // some PUSHes are reused by different calls on other branches,
2144 // but that can't happen if we didn't branch, so they
2145 // can be removed from future searches (handles nested calls)
2146 if (!may_reuse)
2147 ops[j].flags |= OPF_FARG;
2148
2149 // next arg
2150 for (arg++; arg < pp->argc; arg++)
2151 if (pp->arg[arg].reg == NULL)
2152 break;
2153 }
2154 }
2155
2156 if (arg < pp->argc) {
2157 ferr(po, "arg collect failed for '%s': %d/%d\n",
2158 pp->name, arg, pp->argc);
2159 ret = -1;
2160 }
2161 return ret;
2162}
2163
2164static void add_label_ref(struct label_ref *lr, int op_i)
2165{
2166 struct label_ref *lr_new;
2167
2168 if (lr->i == -1) {
2169 lr->i = op_i;
2170 return;
2171 }
2172
2173 lr_new = calloc(1, sizeof(*lr_new));
2174 lr_new->i = op_i;
2175 lr_new->next = lr->next;
2176 lr->next = lr_new;
2177}
2178
2179static void output_std_flags(FILE *fout, struct parsed_op *po,
2180 int *pfomask, const char *dst_opr_text)
2181{
2182 if (*pfomask & (1 << PFO_Z)) {
2183 fprintf(fout, "\n cond_z = (%s%s == 0);",
2184 lmod_cast_u(po, po->operand[0].lmod), dst_opr_text);
2185 *pfomask &= ~(1 << PFO_Z);
2186 }
2187 if (*pfomask & (1 << PFO_S)) {
2188 fprintf(fout, "\n cond_s = (%s%s < 0);",
2189 lmod_cast_s(po, po->operand[0].lmod), dst_opr_text);
2190 *pfomask &= ~(1 << PFO_S);
2191 }
2192}
2193
2194static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
2195{
2196 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
2197 struct parsed_opr *last_arith_dst = NULL;
2198 char buf1[256], buf2[256], buf3[256], cast[64];
2199 const struct parsed_proto *pp_c;
2200 struct parsed_proto *pp, *pp_tmp;
2201 struct parsed_data *pd;
2202 const char *tmpname;
2203 enum parsed_flag_op pfo;
2204 int save_arg_vars = 0;
2205 int cmp_result_vars = 0;
2206 int need_mul_var = 0;
2207 int had_decl = 0;
2208 int label_pending = 0;
2209 int regmask_save = 0;
2210 int regmask_arg = 0;
2211 int regmask = 0;
2212 int pfomask = 0;
2213 int found = 0;
2214 int depth = 0;
2215 int no_output;
2216 int i, j, l;
2217 int dummy;
2218 int arg;
2219 int reg;
2220 int ret;
2221
2222 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
2223
2224 g_func_pp = proto_parse(fhdr, funcn);
2225 if (g_func_pp == NULL)
2226 ferr(ops, "proto_parse failed for '%s'\n", funcn);
2227
2228 fprintf(fout, "%s ", g_func_pp->ret_type.name);
2229 if (g_func_pp->is_stdcall && g_func_pp->argc_reg == 0)
2230 fprintf(fout, "__stdcall ");
2231 if (g_ida_func_attr & IDAFA_NORETURN)
2232 fprintf(fout, "noreturn ");
2233 fprintf(fout, "%s(", funcn);
2234
2235 for (i = 0; i < g_func_pp->argc; i++) {
2236 if (i > 0)
2237 fprintf(fout, ", ");
2238 if (g_func_pp->arg[i].fptr != NULL) {
2239 // func pointer..
2240 pp = g_func_pp->arg[i].fptr;
2241 fprintf(fout, "%s (", pp->ret_type.name);
2242 if (pp->is_stdcall && pp->argc_reg == 0)
2243 fprintf(fout, "__stdcall ");
2244 fprintf(fout, "*a%d)(", i + 1);
2245 for (j = 0; j < pp->argc; j++) {
2246 if (j > 0)
2247 fprintf(fout, ", ");
2248 if (pp->arg[j].fptr)
2249 ferr(ops, "nested fptr\n");
2250 fprintf(fout, "%s", pp->arg[j].type.name);
2251 }
2252 fprintf(fout, ")");
2253 }
2254 else {
2255 fprintf(fout, "%s a%d", g_func_pp->arg[i].type.name, i + 1);
2256 }
2257 }
2258 if (g_func_pp->is_vararg) {
2259 if (i > 0)
2260 fprintf(fout, ", ");
2261 fprintf(fout, "...");
2262 }
2263
2264 fprintf(fout, ")\n{\n");
2265
2266 // pass1:
2267 // - handle ebp/esp frame, remove ops related to it
2268 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
2269 && ops[1].op == OP_MOV
2270 && IS(opr_name(&ops[1], 0), "ebp")
2271 && IS(opr_name(&ops[1], 1), "esp"))
2272 {
2273 int ecx_push = 0;
2274
2275 g_bp_frame = 1;
2276 ops[0].flags |= OPF_RMD;
2277 ops[1].flags |= OPF_RMD;
2278 i = 2;
2279
2280 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
2281 g_stack_fsz = opr_const(&ops[2], 1);
2282 ops[2].flags |= OPF_RMD;
2283 i++;
2284 }
2285 else {
2286 // another way msvc builds stack frame..
2287 i = 2;
2288 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
2289 g_stack_fsz += 4;
2290 ops[i].flags |= OPF_RMD;
2291 ecx_push++;
2292 i++;
2293 }
2294 // and another way..
2295 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
2296 && ops[i].operand[1].type == OPT_CONST
2297 && ops[i + 1].op == OP_CALL
2298 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
2299 {
2300 g_stack_fsz += ops[i].operand[1].val;
2301 ops[i].flags |= OPF_RMD;
2302 i++;
2303 ops[i].flags |= OPF_RMD;
2304 i++;
2305 }
2306 }
2307
2308 found = 0;
2309 do {
2310 for (; i < opcnt; i++)
2311 if (ops[i].op == OP_RET)
2312 break;
2313 if (i == opcnt && (ops[i - 1].flags & OPF_JMP) && found)
2314 break;
2315
2316 if (ops[i - 1].op == OP_POP && IS(opr_name(&ops[i - 1], 0), "ebp"))
2317 ops[i - 1].flags |= OPF_RMD;
2318 else if (!(g_ida_func_attr & IDAFA_NORETURN))
2319 ferr(&ops[i - 1], "'pop ebp' expected\n");
2320
2321 if (g_stack_fsz != 0) {
2322 if (ops[i - 2].op == OP_MOV
2323 && IS(opr_name(&ops[i - 2], 0), "esp")
2324 && IS(opr_name(&ops[i - 2], 1), "ebp"))
2325 {
2326 ops[i - 2].flags |= OPF_RMD;
2327 }
2328 else if (!(g_ida_func_attr & IDAFA_NORETURN))
2329 ferr(&ops[i - 2], "esp restore expected\n");
2330
2331 if (ecx_push && ops[i - 3].op == OP_POP
2332 && IS(opr_name(&ops[i - 3], 0), "ecx"))
2333 {
2334 ferr(&ops[i - 3], "unexpected ecx pop\n");
2335 }
2336 }
2337
2338 found = 1;
2339 i++;
2340 } while (i < opcnt);
2341 }
2342 else {
2343 for (i = 0; i < opcnt; i++) {
2344 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
2345 break;
2346 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
2347 && ops[i].operand[1].type == OPT_CONST)
2348 {
2349 g_sp_frame = 1;
2350 break;
2351 }
2352 }
2353
2354 if (g_sp_frame)
2355 {
2356 g_stack_fsz = ops[i].operand[1].val;
2357 ops[i].flags |= OPF_RMD;
2358
2359 i++;
2360 do {
2361 for (; i < opcnt; i++)
2362 if (ops[i].op == OP_RET)
2363 break;
2364 if (ops[i - 1].op != OP_ADD
2365 || !IS(opr_name(&ops[i - 1], 0), "esp")
2366 || ops[i - 1].operand[1].type != OPT_CONST
2367 || ops[i - 1].operand[1].val != g_stack_fsz)
2368 ferr(&ops[i - 1], "'add esp' expected\n");
2369 ops[i - 1].flags |= OPF_RMD;
2370
2371 i++;
2372 } while (i < opcnt);
2373 }
2374 }
2375
2376 // pass2:
2377 // - resolve all branches
2378 for (i = 0; i < opcnt; i++)
2379 {
2380 po = &ops[i];
2381 po->bt_i = -1;
2382 po->btj = NULL;
2383
2384 if ((po->flags & OPF_RMD) || !(po->flags & OPF_JMP)
2385 || po->op == OP_CALL || po->op == OP_RET)
2386 continue;
2387
2388 if (po->operand[0].type == OPT_REGMEM) {
2389 char *p = strchr(po->operand[0].name, '[');
2390 if (p == NULL)
2391 ferr(po, "unhandled indirect branch\n");
2392 ret = p - po->operand[0].name;
2393 strncpy(buf1, po->operand[0].name, ret);
2394 buf1[ret] = 0;
2395
2396 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
2397 if (IS(g_func_pd[j].label, buf1)) {
2398 pd = &g_func_pd[j];
2399 break;
2400 }
2401 }
2402 if (pd == NULL)
2403 //ferr(po, "label '%s' not parsed?\n", buf1);
2404 goto tailcall;
2405 if (pd->type != OPT_OFFSET)
2406 ferr(po, "label '%s' with non-offset data?\n", buf1);
2407
2408 // find all labels, link
2409 for (j = 0; j < pd->count; j++) {
2410 for (l = 0; l < opcnt; l++) {
2411 if (g_labels[l][0] && IS(g_labels[l], pd->d[j].u.label)) {
2412 add_label_ref(&g_label_refs[l], i);
2413 pd->d[j].bt_i = l;
2414 break;
2415 }
2416 }
2417 }
2418
2419 po->btj = pd;
2420 continue;
2421 }
2422
2423 for (l = 0; l < opcnt; l++) {
2424 if (g_labels[l][0] && IS(po->operand[0].name, g_labels[l])) {
2425 add_label_ref(&g_label_refs[l], i);
2426 po->bt_i = l;
2427 break;
2428 }
2429 }
2430
2431 if (po->bt_i != -1)
2432 continue;
2433
2434 if (po->operand[0].type == OPT_LABEL)
2435 // assume tail call
2436 goto tailcall;
2437
2438 ferr(po, "unhandled branch\n");
2439
2440tailcall:
2441 po->op = OP_CALL;
2442 po->flags |= OPF_TAIL;
2443 }
2444
2445 // pass3:
2446 // - process calls
2447 for (i = 0; i < opcnt; i++)
2448 {
2449 po = &ops[i];
2450 if (po->flags & OPF_RMD)
2451 continue;
2452
2453 if (po->op == OP_CALL)
2454 {
2455 pp = calloc(1, sizeof(*pp));
2456 my_assert_not(pp, NULL);
2457 tmpname = opr_name(po, 0);
2458 if (po->operand[0].type != OPT_LABEL)
2459 {
2460 ret = scan_for_esp_adjust(i + 1, opcnt, &j);
2461 if (ret < 0)
2462 ferr(po, "non-__cdecl indirect call unhandled yet\n");
2463 j /= 4;
2464 if (j > ARRAY_SIZE(pp->arg))
2465 ferr(po, "esp adjust too large?\n");
2466 pp->ret_type.name = strdup("int");
2467 pp->argc = pp->argc_stack = j;
2468 for (arg = 0; arg < pp->argc; arg++)
2469 pp->arg[arg].type.name = strdup("int");
2470 }
2471 else {
2472 pp_c = proto_parse(fhdr, tmpname);
2473 if (pp_c == NULL)
2474 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
2475 pp = proto_clone(pp_c);
2476 }
2477
2478 // look for and make use of esp adjust
2479 ret = -1;
2480 if (!pp->is_stdcall && pp->argc_stack > 0)
2481 ret = scan_for_esp_adjust(i + 1, opcnt, &j);
2482 if (ret >= 0) {
2483 if (pp->is_vararg) {
2484 if (j / 4 < pp->argc_stack)
2485 ferr(po, "esp adjust is too small: %x < %x\n",
2486 j, pp->argc_stack * 4);
2487 // modify pp to make it have varargs as normal args
2488 arg = pp->argc;
2489 pp->argc += j / 4 - pp->argc_stack;
2490 for (; arg < pp->argc; arg++) {
2491 pp->arg[arg].type.name = strdup("int");
2492 pp->argc_stack++;
2493 }
2494 if (pp->argc > ARRAY_SIZE(pp->arg))
2495 ferr(po, "too many args for '%s'\n", tmpname);
2496 }
2497 if (pp->argc_stack != j / 4)
2498 ferr(po, "stack tracking failed for '%s': %x %x\n",
2499 tmpname, pp->argc_stack * 4, j);
2500
2501 ops[ret].flags |= OPF_RMD;
2502 // a bit of a hack, but deals with use of
2503 // single adj for multiple calls
2504 ops[ret].operand[1].val -= j;
2505 }
2506 else if (pp->is_vararg)
2507 ferr(po, "missing esp_adjust for vararg func '%s'\n",
2508 pp->name);
2509
2510 // can't call functions with non-__cdecl callbacks yet
2511 for (arg = 0; arg < pp->argc; arg++) {
2512 if (pp->arg[arg].fptr != NULL) {
2513 pp_tmp = pp->arg[arg].fptr;
2514 if (pp_tmp->is_stdcall || pp_tmp->argc != pp_tmp->argc_stack)
2515 ferr(po, "'%s' has a non-__cdecl callback\n", tmpname);
2516 }
2517 }
2518
2519 collect_call_args(po, i, pp, &save_arg_vars, 0, 0, 0);
2520
2521 if (strstr(pp->ret_type.name, "int64"))
2522 need_mul_var = 1;
2523 po->datap = pp;
2524 }
2525 }
2526
2527 // pass4:
2528 // - find POPs for PUSHes, rm both
2529 // - scan for all used registers
2530 // - find flag set ops for their users
2531 // - declare indirect functions
2532 for (i = 0; i < opcnt; i++) {
2533 po = &ops[i];
2534 if (po->flags & OPF_RMD)
2535 continue;
2536
2537 if (po->op == OP_PUSH
2538 && po->argnum == 0 && !(po->flags & OPF_RSAVE)
2539 && po->operand[0].type == OPT_REG)
2540 {
2541 reg = po->operand[0].reg;
2542 if (reg < 0)
2543 ferr(po, "reg not set for push?\n");
2544
2545 depth = 0;
2546 ret = scan_for_pop(i + 1, opcnt,
2547 po->operand[0].name, i + opcnt, 0, &depth, 0);
2548 if (ret == 1) {
2549 if (depth > 1)
2550 ferr(po, "too much depth: %d\n", depth);
2551 if (depth > 0)
2552 regmask_save |= 1 << reg;
2553
2554 po->flags |= OPF_RMD;
2555 scan_for_pop(i + 1, opcnt, po->operand[0].name,
2556 i + opcnt * 2, 0, &depth, 1);
2557 continue;
2558 }
2559 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
2560 if (ret == 0) {
2561 arg = OPF_RMD;
2562 if (regmask & (1 << reg)) {
2563 if (regmask_save & (1 << reg))
2564 ferr(po, "%s already saved?\n", po->operand[0].name);
2565 arg = OPF_RSAVE;
2566 }
2567 po->flags |= arg;
2568 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg);
2569 continue;
2570 }
2571 }
2572
2573 regmask |= po->regmask_src | po->regmask_dst;
2574
2575 if (po->flags & OPF_CC)
2576 {
2577 int setters[16], cnt = 0, branched = 0;
2578
2579 ret = scan_for_flag_set(i, i + opcnt * 6,
2580 &branched, setters, &cnt);
2581 if (ret < 0 || cnt <= 0)
2582 ferr(po, "unable to trace flag setter(s)\n");
2583 if (cnt > ARRAY_SIZE(setters))
2584 ferr(po, "too many flag setters\n");
2585
2586 pfo = split_cond(po, po->op, &dummy);
2587 for (j = 0; j < cnt; j++)
2588 {
2589 tmp_op = &ops[setters[j]]; // flag setter
2590 pfomask = 0;
2591
2592 // to get nicer code, we try to delay test and cmp;
2593 // if we can't because of operand modification, or if we
2594 // have math op, or branch, make it calculate flags explicitly
2595 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP) {
2596 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i) >= 0)
2597 pfomask = 1 << pfo;
2598 }
2599 else if (tmp_op->op == OP_CMPS) {
2600 pfomask = 1 << PFO_Z;
2601 }
2602 else {
2603 // see if we'll be able to handle based on op result
2604 if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR
2605 && pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P)
2606 || branched
2607 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
2608 pfomask = 1 << pfo;
2609 }
2610 if (pfomask) {
2611 tmp_op->pfomask |= pfomask;
2612 cmp_result_vars |= pfomask;
2613 }
2614 // note: may overwrite, currently not a problem
2615 po->datap = tmp_op;
2616 }
2617
2618 if (po->op == OP_ADC || po->op == OP_SBB)
2619 cmp_result_vars |= 1 << PFO_C;
2620 }
2621 else if (po->op == OP_MUL
2622 || (po->op == OP_IMUL && po->operand_cnt == 1))
2623 {
2624 need_mul_var = 1;
2625 }
2626 else if (po->op == OP_CALL && po->operand[0].type != OPT_LABEL) {
2627 pp = po->datap;
2628 my_assert_not(pp, NULL);
2629 fprintf(fout, " %s (*icall%d)(", pp->ret_type.name, i);
2630 for (j = 0; j < pp->argc; j++) {
2631 if (j > 0)
2632 fprintf(fout, ", ");
2633 fprintf(fout, "%s a%d", pp->arg[j].type.name, j + 1);
2634 }
2635 fprintf(fout, ");\n");
2636 }
2637 }
2638
2639 // output LUTs/jumptables
2640 for (i = 0; i < g_func_pd_cnt; i++) {
2641 pd = &g_func_pd[i];
2642 fprintf(fout, " static const ");
2643 if (pd->type == OPT_OFFSET) {
2644 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
2645
2646 for (j = 0; j < pd->count; j++) {
2647 if (j > 0)
2648 fprintf(fout, ", ");
2649 fprintf(fout, "&&%s", pd->d[j].u.label);
2650 }
2651 }
2652 else {
2653 fprintf(fout, "%s %s[] =\n { ",
2654 lmod_type_u(ops, pd->lmod), pd->label);
2655
2656 for (j = 0; j < pd->count; j++) {
2657 if (j > 0)
2658 fprintf(fout, ", ");
2659 fprintf(fout, "%u", pd->d[j].u.val);
2660 }
2661 }
2662 fprintf(fout, " };\n");
2663 }
2664
2665 // declare stack frame, va_arg
2666 if (g_stack_fsz)
2667 fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
2668 (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz);
2669
2670 if (g_func_pp->is_vararg)
2671 fprintf(fout, " va_list ap;\n");
2672
2673 // declare arg-registers
2674 for (i = 0; i < g_func_pp->argc; i++) {
2675 if (g_func_pp->arg[i].reg != NULL) {
2676 reg = char_array_i(regs_r32,
2677 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
2678 if (reg < 0)
2679 ferr(ops, "arg '%s' is not a reg?\n", g_func_pp->arg[i].reg);
2680
2681 regmask_arg |= 1 << reg;
2682 fprintf(fout, " u32 %s = (u32)a%d;\n",
2683 g_func_pp->arg[i].reg, i + 1);
2684 had_decl = 1;
2685 }
2686 }
2687
2688 // declare other regs - special case for eax
2689 if (!((regmask | regmask_arg) & 1)
2690 && !IS(g_func_pp->ret_type.name, "void"))
2691 {
2692 fprintf(fout, " u32 eax = 0;\n");
2693 had_decl = 1;
2694 }
2695
2696 regmask &= ~regmask_arg;
2697 regmask &= ~(1 << xSP);
2698 if (g_bp_frame)
2699 regmask &= ~(1 << xBP);
2700 if (regmask) {
2701 for (reg = 0; reg < 8; reg++) {
2702 if (regmask & (1 << reg)) {
2703 fprintf(fout, " u32 %s;\n", regs_r32[reg]);
2704 had_decl = 1;
2705 }
2706 }
2707 }
2708
2709 if (regmask_save) {
2710 for (reg = 0; reg < 8; reg++) {
2711 if (regmask_save & (1 << reg)) {
2712 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
2713 had_decl = 1;
2714 }
2715 }
2716 }
2717
2718 if (save_arg_vars) {
2719 for (reg = 0; reg < 32; reg++) {
2720 if (save_arg_vars & (1 << reg)) {
2721 fprintf(fout, " u32 s_a%d;\n", reg + 1);
2722 had_decl = 1;
2723 }
2724 }
2725 }
2726
2727 if (cmp_result_vars) {
2728 for (i = 0; i < 8; i++) {
2729 if (cmp_result_vars & (1 << i)) {
2730 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
2731 had_decl = 1;
2732 }
2733 }
2734 }
2735
2736 if (need_mul_var) {
2737 fprintf(fout, " u64 mul_tmp;\n");
2738 had_decl = 1;
2739 }
2740
2741 if (had_decl)
2742 fprintf(fout, "\n");
2743
2744 if (g_func_pp->is_vararg) {
2745 if (g_func_pp->argc_stack == 0)
2746 ferr(ops, "vararg func without stack args?\n");
2747 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
2748 }
2749
2750 // output ops
2751 for (i = 0; i < opcnt; i++)
2752 {
2753 if (g_labels[i][0] != 0 && g_label_refs[i].i != -1) {
2754 fprintf(fout, "\n%s:\n", g_labels[i]);
2755 label_pending = 1;
2756
2757 delayed_flag_op = NULL;
2758 last_arith_dst = NULL;
2759 }
2760
2761 po = &ops[i];
2762 if (po->flags & OPF_RMD)
2763 continue;
2764
2765 no_output = 0;
2766
2767 #define assert_operand_cnt(n_) \
2768 if (po->operand_cnt != n_) \
2769 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
2770
2771 // conditional/flag using op?
2772 if (po->flags & OPF_CC)
2773 {
2774 int is_delayed = 0;
2775 int is_inv = 0;
2776
2777 pfo = split_cond(po, po->op, &is_inv);
2778 tmp_op = po->datap;
2779
2780 // we go through all this trouble to avoid using parsed_flag_op,
2781 // which makes generated code much nicer
2782 if (delayed_flag_op != NULL)
2783 {
2784 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op, pfo, is_inv);
2785 is_delayed = 1;
2786 }
2787 else if (last_arith_dst != NULL
2788 && (pfo == PFO_Z || pfo == PFO_S || pfo == PFO_P
2789 || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR))
2790 ))
2791 {
2792 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
2793 out_test_for_cc(buf1, sizeof(buf1), po, pfo, is_inv,
2794 last_arith_dst->lmod, buf3);
2795 is_delayed = 1;
2796 }
2797 else if (tmp_op != NULL) {
2798 // use preprocessed flag calc results
2799 if (!(tmp_op->pfomask & (1 << pfo)))
2800 ferr(po, "not prepared for pfo %d\n", pfo);
2801
2802 // note: is_inv was not yet applied
2803 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
2804 is_inv ? "!" : "", parsed_flag_op_names[pfo]);
2805 }
2806 else {
2807 ferr(po, "all methods of finding comparison failed\n");
2808 }
2809
2810 if (po->flags & OPF_JMP) {
2811 fprintf(fout, " if %s\n", buf1);
2812 }
2813 else if (po->op == OP_ADC || po->op == OP_SBB) {
2814 if (is_delayed)
2815 fprintf(fout, " cond_%s = %s;\n",
2816 parsed_flag_op_names[pfo], buf1);
2817 }
2818 else if (po->flags & OPF_DATA) { // SETcc
2819 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
2820 fprintf(fout, " %s = %s;", buf2, buf1);
2821 }
2822 else {
2823 ferr(po, "unhandled conditional op\n");
2824 }
2825 }
2826
2827 pfomask = po->pfomask;
2828
2829 switch (po->op)
2830 {
2831 case OP_MOV:
2832 assert_operand_cnt(2);
2833 propagate_lmod(po, &po->operand[0], &po->operand[1]);
2834 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
2835 fprintf(fout, " %s = %s;", buf1,
2836 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
2837 po->operand[0].is_ptr ? "(void *)" : "", 0));
2838 break;
2839
2840 case OP_LEA:
2841 assert_operand_cnt(2);
2842 po->operand[1].lmod = OPLM_DWORD; // always
2843 fprintf(fout, " %s = %s;",
2844 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2845 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
2846 NULL, 1));
2847 break;
2848
2849 case OP_MOVZX:
2850 assert_operand_cnt(2);
2851 fprintf(fout, " %s = %s;",
2852 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2853 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2854 break;
2855
2856 case OP_MOVSX:
2857 assert_operand_cnt(2);
2858 switch (po->operand[1].lmod) {
2859 case OPLM_BYTE:
2860 strcpy(buf3, "(s8)");
2861 break;
2862 case OPLM_WORD:
2863 strcpy(buf3, "(s16)");
2864 break;
2865 default:
2866 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
2867 }
2868 fprintf(fout, " %s = %s%s;",
2869 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2870 buf3,
2871 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2872 break;
2873
2874 case OP_NOT:
2875 assert_operand_cnt(1);
2876 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
2877 fprintf(fout, " %s = ~%s;", buf1, buf1);
2878 break;
2879
2880 case OP_CDQ:
2881 assert_operand_cnt(2);
2882 fprintf(fout, " %s = (s32)%s >> 31;",
2883 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2884 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2885 strcpy(g_comment, "cdq");
2886 break;
2887
2888 case OP_STOS:
2889 // assumes DF=0
2890 assert_operand_cnt(3);
2891 if (po->flags & OPF_REP) {
2892 fprintf(fout, " for (; ecx != 0; ecx--, edi += %d)\n",
2893 lmod_bytes(po, po->operand[0].lmod));
2894 fprintf(fout, " %sedi = eax;",
2895 lmod_cast_u_ptr(po, po->operand[0].lmod));
2896 strcpy(g_comment, "rep stos");
2897 }
2898 else {
2899 fprintf(fout, " %sedi = eax; edi += %d;",
2900 lmod_cast_u_ptr(po, po->operand[0].lmod),
2901 lmod_bytes(po, po->operand[0].lmod));
2902 strcpy(g_comment, "stos");
2903 }
2904 break;
2905
2906 case OP_MOVS:
2907 // assumes DF=0
2908 assert_operand_cnt(3);
2909 j = lmod_bytes(po, po->operand[0].lmod);
2910 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
2911 if (po->flags & OPF_REP) {
2912 fprintf(fout,
2913 " for (; ecx != 0; ecx--, edi += %d, esi += %d)\n",
2914 j, j);
2915 fprintf(fout,
2916 " %sedi = %sesi;", buf1, buf1);
2917 strcpy(g_comment, "rep movs");
2918 }
2919 else {
2920 fprintf(fout, " %sedi = %sesi; edi += %d; esi += %d;",
2921 buf1, buf1, j, j);
2922 strcpy(g_comment, "movs");
2923 }
2924 break;
2925
2926 case OP_CMPS:
2927 // assumes DF=0
2928 // repe ~ repeat while ZF=1
2929 assert_operand_cnt(3);
2930 j = lmod_bytes(po, po->operand[0].lmod);
2931 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
2932 if (po->flags & OPF_REP) {
2933 fprintf(fout,
2934 " for (; ecx != 0; ecx--, edi += %d, esi += %d)\n",
2935 j, j);
2936 fprintf(fout,
2937 " if ((cond_z = (%sedi == %sesi)) %s 0)\n",
2938 buf1, buf1, (po->flags & OPF_REPZ) ? "==" : "!=");
2939 fprintf(fout,
2940 " break;");
2941 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
2942 (po->flags & OPF_REPZ) ? "e" : "ne");
2943 }
2944 else {
2945 fprintf(fout,
2946 " cond_z = (%sedi = %sesi); edi += %d; esi += %d;",
2947 buf1, buf1, j, j);
2948 strcpy(g_comment, "cmps");
2949 }
2950 pfomask &= ~(1 << PFO_Z);
2951 last_arith_dst = NULL;
2952 delayed_flag_op = NULL;
2953 break;
2954
2955 // arithmetic w/flags
2956 case OP_ADD:
2957 case OP_SUB:
2958 case OP_AND:
2959 case OP_OR:
2960 propagate_lmod(po, &po->operand[0], &po->operand[1]);
2961 // fallthrough
2962 dualop_arith:
2963 assert_operand_cnt(2);
2964 fprintf(fout, " %s %s= %s;",
2965 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
2966 op_to_c(po),
2967 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2968 output_std_flags(fout, po, &pfomask, buf1);
2969 last_arith_dst = &po->operand[0];
2970 delayed_flag_op = NULL;
2971 break;
2972
2973 case OP_SHL:
2974 case OP_SHR:
2975 assert_operand_cnt(2);
2976 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
2977 if (pfomask & (1 << PFO_C)) {
2978 if (po->operand[1].type == OPT_CONST) {
2979 l = lmod_bytes(po, po->operand[0].lmod) * 8;
2980 j = po->operand[1].val;
2981 j %= l;
2982 if (j != 0) {
2983 if (po->op == OP_SHL)
2984 j = l - j;
2985 else
2986 j -= 1;
2987 fprintf(fout, " cond_c = (%s & 0x%02x) ? 1 : 0;\n",
2988 buf1, 1 << j);
2989 }
2990 else
2991 ferr(po, "zero shift?\n");
2992 }
2993 else
2994 ferr(po, "TODO\n");
2995 pfomask &= ~(1 << PFO_C);
2996 }
2997 fprintf(fout, " %s %s= %s;", buf1, op_to_c(po),
2998 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
2999 output_std_flags(fout, po, &pfomask, buf1);
3000 last_arith_dst = &po->operand[0];
3001 delayed_flag_op = NULL;
3002 break;
3003
3004 case OP_SAR:
3005 assert_operand_cnt(2);
3006 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3007 fprintf(fout, " %s = %s%s >> %s;", buf1,
3008 lmod_cast_s(po, po->operand[0].lmod), buf1,
3009 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
3010 output_std_flags(fout, po, &pfomask, buf1);
3011 last_arith_dst = &po->operand[0];
3012 delayed_flag_op = NULL;
3013 break;
3014
3015 case OP_ROL:
3016 case OP_ROR:
3017 assert_operand_cnt(2);
3018 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3019 if (po->operand[1].type == OPT_CONST) {
3020 j = po->operand[1].val;
3021 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
3022 fprintf(fout, po->op == OP_ROL ?
3023 " %s = (%s << %d) | (%s >> %d);" :
3024 " %s = (%s >> %d) | (%s << %d);",
3025 buf1, buf1, j, buf1,
3026 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
3027 }
3028 else
3029 ferr(po, "TODO\n");
3030 output_std_flags(fout, po, &pfomask, buf1);
3031 last_arith_dst = &po->operand[0];
3032 delayed_flag_op = NULL;
3033 break;
3034
3035 case OP_XOR:
3036 assert_operand_cnt(2);
3037 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3038 if (IS(opr_name(po, 0), opr_name(po, 1))) {
3039 // special case for XOR
3040 fprintf(fout, " %s = 0;",
3041 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
3042 last_arith_dst = &po->operand[0];
3043 delayed_flag_op = NULL;
3044 break;
3045 }
3046 goto dualop_arith;
3047
3048 case OP_ADC:
3049 case OP_SBB:
3050 assert_operand_cnt(2);
3051 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3052 if (po->op == OP_SBB
3053 && IS(po->operand[0].name, po->operand[1].name))
3054 {
3055 // avoid use of unitialized var
3056 fprintf(fout, " %s = -cond_c;",
3057 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
3058 }
3059 else {
3060 fprintf(fout, " %s %s= %s + cond_c;",
3061 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3062 op_to_c(po),
3063 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
3064 }
3065 last_arith_dst = &po->operand[0];
3066 delayed_flag_op = NULL;
3067 break;
3068
3069 case OP_INC:
3070 case OP_DEC:
3071 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3072 if (po->operand[0].type == OPT_REG) {
3073 strcpy(buf2, po->op == OP_INC ? "++" : "--");
3074 fprintf(fout, " %s%s;", buf1, buf2);
3075 }
3076 else {
3077 strcpy(buf2, po->op == OP_INC ? "+" : "-");
3078 fprintf(fout, " %s %s= 1;", buf1, buf2);
3079 }
3080 last_arith_dst = &po->operand[0];
3081 delayed_flag_op = NULL;
3082 break;
3083
3084 case OP_NEG:
3085 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3086 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
3087 fprintf(fout, " %s = -%s%s;", buf1,
3088 lmod_cast_s(po, po->operand[0].lmod), buf2);
3089 last_arith_dst = &po->operand[0];
3090 delayed_flag_op = NULL;
3091 if (pfomask & (1 << PFO_C)) {
3092 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
3093 pfomask &= ~(1 << PFO_C);
3094 }
3095 break;
3096
3097 case OP_IMUL:
3098 if (po->operand_cnt == 2) {
3099 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3100 goto dualop_arith;
3101 }
3102 if (po->operand_cnt == 3)
3103 ferr(po, "TODO imul3\n");
3104 // fallthrough
3105 case OP_MUL:
3106 assert_operand_cnt(1);
3107 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
3108 fprintf(fout, " mul_tmp = %seax * %s%s;\n", buf1, buf1,
3109 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
3110 fprintf(fout, " edx = mul_tmp >> 32;\n");
3111 fprintf(fout, " eax = mul_tmp;");
3112 last_arith_dst = NULL;
3113 delayed_flag_op = NULL;
3114 break;
3115
3116 case OP_DIV:
3117 case OP_IDIV:
3118 assert_operand_cnt(1);
3119 if (po->operand[0].lmod != OPLM_DWORD)
3120 ferr(po, "unhandled lmod %d\n", po->operand[0].lmod);
3121
3122 // 32bit division is common, look for it
3123 if (po->op == OP_DIV)
3124 ret = scan_for_reg_clear(i - 1, xDX);
3125 else
3126 ret = scan_for_cdq_edx(i - 1);
3127 if (ret >= 0) {
3128 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
3129 strcpy(buf2, lmod_cast(po, po->operand[0].lmod,
3130 po->op == OP_IDIV));
3131 fprintf(fout, " edx = %seax %% %s%s;\n", buf2, buf2, buf1);
3132 fprintf(fout, " eax = %seax / %s%s;", buf2, buf2, buf1);
3133 }
3134 else
3135 ferr(po, "TODO 64bit divident\n");
3136 last_arith_dst = NULL;
3137 delayed_flag_op = NULL;
3138 break;
3139
3140 case OP_TEST:
3141 case OP_CMP:
3142 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3143 if (pfomask != 0) {
3144 for (j = 0; j < 8; j++) {
3145 if (pfomask & (1 << j)) {
3146 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
3147 fprintf(fout, " cond_%s = %s;",
3148 parsed_flag_op_names[j], buf1);
3149 }
3150 }
3151 pfomask = 0;
3152 }
3153 else
3154 no_output = 1;
3155 last_arith_dst = NULL;
3156 delayed_flag_op = po;
3157 break;
3158
3159 // note: we reuse OP_Jcc for SETcc, only flags differ
3160 case OP_JO ... OP_JG:
3161 if (po->flags & OPF_JMP)
3162 fprintf(fout, " goto %s;", po->operand[0].name);
3163 // else SETcc - should already be handled
3164 break;
3165
3166 case OP_JMP:
3167 assert_operand_cnt(1);
3168 last_arith_dst = NULL;
3169 delayed_flag_op = NULL;
3170
3171 if (po->operand[0].type == OPT_REGMEM) {
3172 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
3173 buf1, buf2);
3174 if (ret != 2)
3175 ferr(po, "parse failure for jmp '%s'\n",
3176 po->operand[0].name);
3177 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
3178 break;
3179 }
3180 else if (po->operand[0].type != OPT_LABEL)
3181 ferr(po, "unhandled jmp type\n");
3182
3183 fprintf(fout, " goto %s;", po->operand[0].name);
3184 break;
3185
3186 case OP_CALL:
3187 assert_operand_cnt(1);
3188 pp = po->datap;
3189 if (pp == NULL)
3190 ferr(po, "NULL pp\n");
3191
3192 if (po->operand[0].type != OPT_LABEL)
3193 fprintf(fout, " icall%d = (void *)%s;\n", i,
3194 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]));
3195
3196 fprintf(fout, " ");
3197 if (strstr(pp->ret_type.name, "int64")) {
3198 if (po->flags & OPF_TAIL)
3199 ferr(po, "int64 and tail?\n");
3200 fprintf(fout, "mul_tmp = ");
3201 }
3202 else if (!IS(pp->ret_type.name, "void")) {
3203 if (po->flags & OPF_TAIL) {
3204 if (!IS(g_func_pp->ret_type.name, "void")) {
3205 fprintf(fout, "return ");
3206 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
3207 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
3208 }
3209 }
3210 else {
3211 fprintf(fout, "eax = ");
3212 if (pp->ret_type.is_ptr)
3213 fprintf(fout, "(u32)");
3214 }
3215 }
3216
3217 if (po->operand[0].type != OPT_LABEL) {
3218 fprintf(fout, "icall%d(", i);
3219 }
3220 else {
3221 if (pp->name[0] == 0)
3222 ferr(po, "missing pp->name\n");
3223 fprintf(fout, "%s%s(", pp->name,
3224 pp->has_structarg ? "_sa" : "");
3225 }
3226
3227 for (arg = 0; arg < pp->argc; arg++) {
3228 if (arg > 0)
3229 fprintf(fout, ", ");
3230
3231 cast[0] = 0;
3232 if (pp->arg[arg].type.is_ptr)
3233 snprintf(cast, sizeof(cast), "(%s)", pp->arg[arg].type.name);
3234
3235 if (pp->arg[arg].reg != NULL) {
3236 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
3237 continue;
3238 }
3239
3240 // stack arg
3241 tmp_op = pp->arg[arg].datap;
3242 if (tmp_op == NULL)
3243 ferr(po, "parsed_op missing for arg%d\n", arg);
3244 if (tmp_op->argnum != 0) {
3245 fprintf(fout, "%ss_a%d", cast, tmp_op->argnum);
3246 }
3247 else {
3248 fprintf(fout, "%s",
3249 out_src_opr(buf1, sizeof(buf1),
3250 tmp_op, &tmp_op->operand[0], cast, 0));
3251 }
3252 }
3253 fprintf(fout, ");");
3254
3255 if (strstr(pp->ret_type.name, "int64")) {
3256 fprintf(fout, "\n");
3257 fprintf(fout, " edx = mul_tmp >> 32;\n");
3258 fprintf(fout, " eax = mul_tmp;");
3259 }
3260
3261 if (po->flags & OPF_TAIL) {
3262 strcpy(g_comment, "tailcall");
3263 ret = 0;
3264 if (i == opcnt - 1)
3265 ret = 0;
3266 else if (IS(pp->ret_type.name, "void"))
3267 ret = 1;
3268 else if (IS(g_func_pp->ret_type.name, "void"))
3269 ret = 1;
3270 // else already handled as 'return f()'
3271
3272 if (ret) {
3273 fprintf(fout, "\n return;");
3274 strcpy(g_comment, "^ tailcall");
3275 }
3276 }
3277 delayed_flag_op = NULL;
3278 last_arith_dst = NULL;
3279 break;
3280
3281 case OP_RET:
3282 if (g_func_pp->is_vararg)
3283 fprintf(fout, " va_end(ap);\n");
3284
3285 if (IS(g_func_pp->ret_type.name, "void")) {
3286 if (i != opcnt - 1 || label_pending)
3287 fprintf(fout, " return;");
3288 }
3289 else if (g_func_pp->ret_type.is_ptr) {
3290 fprintf(fout, " return (%s)eax;",
3291 g_func_pp->ret_type.name);
3292 }
3293 else
3294 fprintf(fout, " return eax;");
3295
3296 last_arith_dst = NULL;
3297 delayed_flag_op = NULL;
3298 break;
3299
3300 case OP_PUSH:
3301 if (po->argnum != 0) {
3302 // special case - saved func arg
3303 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
3304 fprintf(fout, " s_a%d = %s;", po->argnum, buf1);
3305 break;
3306 }
3307 else if (po->flags & OPF_RSAVE) {
3308 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
3309 fprintf(fout, " s_%s = %s;", buf1, buf1);
3310 break;
3311 }
3312 if (!(g_ida_func_attr & IDAFA_NORETURN))
3313 ferr(po, "stray push encountered\n");
3314 no_output = 1;
3315 break;
3316
3317 case OP_POP:
3318 if (po->flags & OPF_RSAVE) {
3319 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3320 fprintf(fout, " %s = s_%s;", buf1, buf1);
3321 break;
3322 }
3323 ferr(po, "stray pop encountered\n");
3324 break;
3325
3326 case OP_NOP:
3327 no_output = 1;
3328 break;
3329
3330 default:
3331 no_output = 1;
3332 ferr(po, "unhandled op type %d, flags %x\n",
3333 po->op, po->flags);
3334 break;
3335 }
3336
3337 if (g_comment[0] != 0) {
3338 fprintf(fout, " // %s", g_comment);
3339 g_comment[0] = 0;
3340 no_output = 0;
3341 }
3342 if (!no_output)
3343 fprintf(fout, "\n");
3344
3345 // some sanity checking
3346 if ((po->flags & OPF_REP) && po->op != OP_STOS
3347 && po->op != OP_MOVS && po->op != OP_CMPS)
3348 ferr(po, "unexpected rep\n");
3349 if ((po->flags & (OPF_REPZ|OPF_REPNZ)) && po->op != OP_CMPS)
3350 ferr(po, "unexpected repz/repnz\n");
3351
3352 if (pfomask != 0)
3353 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
3354
3355 // see is delayed flag stuff is still valid
3356 if (delayed_flag_op != NULL && delayed_flag_op != po) {
3357 if (is_any_opr_modified(delayed_flag_op, po))
3358 delayed_flag_op = NULL;
3359 }
3360
3361 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
3362 if (is_opr_modified(last_arith_dst, po))
3363 last_arith_dst = NULL;
3364 }
3365
3366 label_pending = 0;
3367 }
3368
3369 fprintf(fout, "}\n\n");
3370
3371 // cleanup
3372 for (i = 0; i < opcnt; i++) {
3373 struct label_ref *lr, *lr_del;
3374
3375 lr = g_label_refs[i].next;
3376 while (lr != NULL) {
3377 lr_del = lr;
3378 lr = lr->next;
3379 free(lr_del);
3380 }
3381 g_label_refs[i].i = -1;
3382 g_label_refs[i].next = NULL;
3383
3384 if (ops[i].op == OP_CALL) {
3385 pp = ops[i].datap;
3386 if (pp)
3387 proto_release(pp);
3388 }
3389 }
3390 g_func_pp = NULL;
3391}
3392
3393static void set_label(int i, const char *name)
3394{
3395 const char *p;
3396 int len;
3397
3398 len = strlen(name);
3399 p = strchr(name, ':');
3400 if (p != NULL)
3401 len = p - name;
3402
3403 if (len > sizeof(g_labels[0]) - 1)
3404 aerr("label '%s' too long: %d\n", name, len);
3405 if (g_labels[i][0] != 0 && !IS_START(g_labels[i], "algn_"))
3406 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
3407 memcpy(g_labels[i], name, len);
3408 g_labels[i][len] = 0;
3409}
3410
3411// '=' needs special treatment..
3412static char *next_word_s(char *w, size_t wsize, char *s)
3413{
3414 size_t i;
3415
3416 s = sskip(s);
3417
3418 for (i = 0; i < wsize - 1; i++) {
3419 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
3420 break;
3421 w[i] = s[i];
3422 }
3423 w[i] = 0;
3424
3425 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
3426 printf("warning: '%s' truncated\n", w);
3427
3428 return s + i;
3429}
3430
3431struct chunk_item {
3432 char *name;
3433 long fptr;
3434 int asmln;
3435};
3436
3437static int cmp_chunks(const void *p1, const void *p2)
3438{
3439 const struct chunk_item *c1 = p1, *c2 = p2;
3440 return strcmp(c1->name, c2->name);
3441}
3442
3443static int cmpstringp(const void *p1, const void *p2)
3444{
3445 return strcmp(*(char * const *)p1, *(char * const *)p2);
3446}
3447
3448int main(int argc, char *argv[])
3449{
3450 FILE *fout, *fasm, *frlist;
3451 struct parsed_data *pd = NULL;
3452 int pd_alloc = 0;
3453 char **rlist = NULL;
3454 int rlist_len = 0;
3455 int rlist_alloc = 0;
3456 struct chunk_item *func_chunks;
3457 int func_chunks_used = 0;
3458 int func_chunks_sorted = 0;
3459 int func_chunk_cnt = 0;
3460 int func_chunk_alloc;
3461 int func_chunk_i = -1;
3462 long func_chunk_ret = 0;
3463 int func_chunk_ret_ln = 0;
3464 char line[256];
3465 char words[16][256];
3466 enum opr_lenmod lmod;
3467 int in_func = 0;
3468 int pending_endp = 0;
3469 int skip_func = 0;
3470 int skip_warned = 0;
3471 int eq_alloc;
3472 int verbose = 0;
3473 int arg_out;
3474 int arg = 1;
3475 int pi = 0;
3476 int i, j;
3477 int ret, len;
3478 char *p;
3479 int wordc;
3480
3481 if (argv[1] && IS(argv[1], "-v")) {
3482 verbose = 1;
3483 arg++;
3484 }
3485
3486 if (argc < arg + 3) {
3487 printf("usage:\n%s [-v] <.c> <.asm> <hdrf> [rlist]*\n",
3488 argv[0]);
3489 return 1;
3490 }
3491
3492 arg_out = arg++;
3493
3494 asmfn = argv[arg++];
3495 fasm = fopen(asmfn, "r");
3496 my_assert_not(fasm, NULL);
3497
3498 hdrfn = argv[arg++];
3499 g_fhdr = fopen(hdrfn, "r");
3500 my_assert_not(g_fhdr, NULL);
3501
3502 rlist_alloc = 64;
3503 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
3504 my_assert_not(rlist, NULL);
3505 // needs special handling..
3506 rlist[rlist_len++] = "__alloca_probe";
3507
3508 func_chunk_alloc = 32;
3509 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
3510 my_assert_not(func_chunks, NULL);
3511
3512 for (; arg < argc; arg++) {
3513 frlist = fopen(argv[arg], "r");
3514 my_assert_not(frlist, NULL);
3515
3516 while (fgets(line, sizeof(line), frlist)) {
3517 p = sskip(line);
3518 if (*p == 0 || *p == ';' || *p == '#')
3519 continue;
3520
3521 p = next_word(words[0], sizeof(words[0]), p);
3522 if (words[0][0] == 0)
3523 continue;
3524
3525 if (rlist_len >= rlist_alloc) {
3526 rlist_alloc = rlist_alloc * 2 + 64;
3527 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
3528 my_assert_not(rlist, NULL);
3529 }
3530 rlist[rlist_len++] = strdup(words[0]);
3531 }
3532
3533 fclose(frlist);
3534 frlist = NULL;
3535 }
3536
3537 if (rlist_len > 0)
3538 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
3539
3540 fout = fopen(argv[arg_out], "w");
3541 my_assert_not(fout, NULL);
3542
3543 eq_alloc = 128;
3544 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
3545 my_assert_not(g_eqs, NULL);
3546
3547 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
3548 g_label_refs[i].i = -1;
3549 g_label_refs[i].next = NULL;
3550 }
3551
3552 while (fgets(line, sizeof(line), fasm))
3553 {
3554 wordc = 0;
3555 asmln++;
3556
3557 p = sskip(line);
3558 if (*p == 0)
3559 continue;
3560
3561 // get rid of random tabs
3562 for (i = 0; line[i] != 0; i++)
3563 if (line[i] == '\t')
3564 line[i] = ' ';
3565
3566 if (*p == ';')
3567 {
3568 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
3569 goto do_pending_endp; // eww..
3570
3571 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
3572 {
3573 static const char *attrs[] = {
3574 "bp-based frame",
3575 "library function",
3576 "static",
3577 "noreturn",
3578 "thunk",
3579 "fpd=",
3580 };
3581
3582 // parse IDA's attribute-list comment
3583 g_ida_func_attr = 0;
3584 p = sskip(p + 13);
3585
3586 for (; *p != 0; p = sskip(p)) {
3587 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
3588 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
3589 g_ida_func_attr |= 1 << i;
3590 p += strlen(attrs[i]);
3591 break;
3592 }
3593 }
3594 if (i == ARRAY_SIZE(attrs)) {
3595 anote("unparsed IDA attr: %s\n", p);
3596 break;
3597 }
3598 if (IS(attrs[i], "fpd=")) {
3599 p = next_word(words[0], sizeof(words[0]), p);
3600 // ignore for now..
3601 }
3602 }
3603 }
3604 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
3605 {
3606 p += 30;
3607 next_word(words[0], sizeof(words[0]), p);
3608 if (words[0][0] == 0)
3609 aerr("missing nam for func chunk?\n");
3610 if (func_chunk_cnt >= func_chunk_alloc) {
3611 func_chunk_alloc *= 2;
3612 func_chunks = realloc(func_chunks,
3613 func_chunk_alloc * sizeof(func_chunks[0]));
3614 my_assert_not(func_chunks, NULL);
3615 }
3616 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
3617 func_chunks[func_chunk_cnt].name = strdup(words[0]);
3618 func_chunks[func_chunk_cnt].asmln = asmln;
3619 func_chunk_cnt++;
3620 func_chunks_sorted = 0;
3621 }
3622 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
3623 {
3624 if (func_chunk_i >= 0) {
3625 if (func_chunk_i < func_chunk_cnt
3626 && IS(func_chunks[func_chunk_i].name, g_func))
3627 {
3628 // move on to next chunk
3629 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
3630 if (ret)
3631 aerr("seek failed for '%s' chunk #%d\n",
3632 g_func, func_chunk_i);
3633 asmln = func_chunks[func_chunk_i].asmln;
3634 func_chunk_i++;
3635 }
3636 else {
3637 if (func_chunk_ret == 0)
3638 aerr("no return from chunk?\n");
3639 fseek(fasm, func_chunk_ret, SEEK_SET);
3640 asmln = func_chunk_ret_ln;
3641 func_chunk_ret = 0;
3642 pending_endp = 1;
3643 }
3644 }
3645 }
3646 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
3647 func_chunks_used = 1;
3648 p += 20;
3649 if (IS_START(g_func, "sub_")) {
3650 unsigned long addr = strtoul(p, NULL, 16);
3651 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
3652 if (addr > f_addr)
3653 aerr("need a chunk %lX that is after %s\n", addr, g_func);
3654 }
3655 }
3656 continue;
3657 } // *p == ';'
3658
3659parse_words:
3660 memset(words, 0, sizeof(words));
3661 for (wordc = 0; wordc < 16; wordc++) {
3662 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
3663 if (*p == 0 || *p == ';') {
3664 wordc++;
3665 break;
3666 }
3667 }
3668
3669 // alow asm patches in comments
3670 if (*p == ';' && IS_START(p, "; sctpatch:")) {
3671 p = sskip(p + 11);
3672 if (*p == 0 || *p == ';')
3673 continue;
3674 goto parse_words; // lame
3675 }
3676
3677 if (wordc == 0) {
3678 // shouldn't happen
3679 awarn("wordc == 0?\n");
3680 continue;
3681 }
3682
3683 // don't care about this:
3684 if (words[0][0] == '.'
3685 || IS(words[0], "include")
3686 || IS(words[0], "assume") || IS(words[1], "segment")
3687 || IS(words[0], "align"))
3688 {
3689 continue;
3690 }
3691
3692do_pending_endp:
3693 // do delayed endp processing to collect switch jumptables
3694 if (pending_endp) {
3695 if (in_func && !skip_func && wordc >= 2
3696 && ((words[0][0] == 'd' && words[0][2] == 0)
3697 || (words[1][0] == 'd' && words[1][2] == 0)))
3698 {
3699 i = 1;
3700 if (words[1][0] == 'd' && words[1][2] == 0) {
3701 // label
3702 if (g_func_pd_cnt >= pd_alloc) {
3703 pd_alloc = pd_alloc * 2 + 16;
3704 g_func_pd = realloc(g_func_pd,
3705 sizeof(g_func_pd[0]) * pd_alloc);
3706 my_assert_not(g_func_pd, NULL);
3707 }
3708 pd = &g_func_pd[g_func_pd_cnt];
3709 g_func_pd_cnt++;
3710 memset(pd, 0, sizeof(*pd));
3711 strcpy(pd->label, words[0]);
3712 pd->type = OPT_CONST;
3713 pd->lmod = lmod_from_directive(words[1]);
3714 i = 2;
3715 }
3716 else {
3717 lmod = lmod_from_directive(words[0]);
3718 if (lmod != pd->lmod)
3719 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
3720 }
3721
3722 if (pd->count_alloc < pd->count + wordc) {
3723 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
3724 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
3725 my_assert_not(pd->d, NULL);
3726 }
3727 for (; i < wordc; i++) {
3728 if (IS(words[i], "offset")) {
3729 pd->type = OPT_OFFSET;
3730 i++;
3731 }
3732 p = strchr(words[i], ',');
3733 if (p != NULL)
3734 *p = 0;
3735 if (pd->type == OPT_OFFSET)
3736 pd->d[pd->count].u.label = strdup(words[i]);
3737 else
3738 pd->d[pd->count].u.val = parse_number(words[i]);
3739 pd->d[pd->count].bt_i = -1;
3740 pd->count++;
3741 }
3742 continue;
3743 }
3744
3745 if (in_func && !skip_func)
3746 gen_func(fout, g_fhdr, g_func, pi);
3747
3748 pending_endp = 0;
3749 in_func = 0;
3750 g_ida_func_attr = 0;
3751 skip_warned = 0;
3752 skip_func = 0;
3753 g_func[0] = 0;
3754 func_chunks_used = 0;
3755 func_chunk_i = -1;
3756 if (pi != 0) {
3757 memset(&ops, 0, pi * sizeof(ops[0]));
3758 memset(g_labels, 0, pi * sizeof(g_labels[0]));
3759 pi = 0;
3760 }
3761 g_eqcnt = 0;
3762 for (i = 0; i < g_func_pd_cnt; i++) {
3763 pd = &g_func_pd[i];
3764 if (pd->type == OPT_OFFSET) {
3765 for (j = 0; j < pd->count; j++)
3766 free(pd->d[j].u.label);
3767 }
3768 free(pd->d);
3769 pd->d = NULL;
3770 }
3771 g_func_pd_cnt = 0;
3772 pd = NULL;
3773 if (wordc == 0)
3774 continue;
3775 }
3776
3777 if (IS(words[1], "proc")) {
3778 if (in_func)
3779 aerr("proc '%s' while in_func '%s'?\n",
3780 words[0], g_func);
3781 p = words[0];
3782 if ((g_ida_func_attr & IDAFA_THUNK)
3783 || bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
3784 skip_func = 1;
3785 strcpy(g_func, words[0]);
3786 set_label(0, words[0]);
3787 in_func = 1;
3788 continue;
3789 }
3790
3791 if (IS(words[1], "endp"))
3792 {
3793 if (!in_func)
3794 aerr("endp '%s' while not in_func?\n", words[0]);
3795 if (!IS(g_func, words[0]))
3796 aerr("endp '%s' while in_func '%s'?\n",
3797 words[0], g_func);
3798
3799 if (!skip_func && func_chunks_used) {
3800 // start processing chunks
3801 struct chunk_item *ci, key = { g_func, 0 };
3802
3803 func_chunk_ret = ftell(fasm);
3804 func_chunk_ret_ln = asmln;
3805 if (!func_chunks_sorted) {
3806 qsort(func_chunks, func_chunk_cnt,
3807 sizeof(func_chunks[0]), cmp_chunks);
3808 func_chunks_sorted = 1;
3809 }
3810 ci = bsearch(&key, func_chunks, func_chunk_cnt,
3811 sizeof(func_chunks[0]), cmp_chunks);
3812 if (ci == NULL)
3813 aerr("'%s' needs chunks, but none found\n", g_func);
3814 func_chunk_i = ci - func_chunks;
3815 for (; func_chunk_i > 0; func_chunk_i--)
3816 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
3817 break;
3818
3819 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
3820 if (ret)
3821 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
3822 asmln = func_chunks[func_chunk_i].asmln;
3823 func_chunk_i++;
3824 continue;
3825 }
3826 pending_endp = 1;
3827 continue;
3828 }
3829
3830 p = strchr(words[0], ':');
3831 if (p != NULL) {
3832 set_label(pi, words[0]);
3833 continue;
3834 }
3835
3836 if (!in_func || skip_func) {
3837 if (!skip_warned && !skip_func && g_labels[pi][0] != 0) {
3838 if (verbose)
3839 anote("skipping from '%s'\n", g_labels[pi]);
3840 skip_warned = 1;
3841 }
3842 g_labels[pi][0] = 0;
3843 continue;
3844 }
3845
3846 if (wordc > 1 && IS(words[1], "=")) {
3847 if (wordc != 5)
3848 aerr("unhandled equ, wc=%d\n", wordc);
3849 if (g_eqcnt >= eq_alloc) {
3850 eq_alloc *= 2;
3851 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
3852 my_assert_not(g_eqs, NULL);
3853 }
3854
3855 len = strlen(words[0]);
3856 if (len > sizeof(g_eqs[0].name) - 1)
3857 aerr("equ name too long: %d\n", len);
3858 strcpy(g_eqs[g_eqcnt].name, words[0]);
3859
3860 if (!IS(words[3], "ptr"))
3861 aerr("unhandled equ\n");
3862 if (IS(words[2], "dword"))
3863 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
3864 else if (IS(words[2], "word"))
3865 g_eqs[g_eqcnt].lmod = OPLM_WORD;
3866 else if (IS(words[2], "byte"))
3867 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
3868 else
3869 aerr("bad lmod: '%s'\n", words[2]);
3870
3871 g_eqs[g_eqcnt].offset = parse_number(words[4]);
3872 g_eqcnt++;
3873 continue;
3874 }
3875
3876 if (pi >= ARRAY_SIZE(ops))
3877 aerr("too many ops\n");
3878
3879 parse_op(&ops[pi], words, wordc);
3880 pi++;
3881 }
3882
3883 fclose(fout);
3884 fclose(fasm);
3885 fclose(g_fhdr);
3886
3887 return 0;
3888}
3889
3890// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.