notaz.gp2x.de / ia32rtools.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
translate: some 64bit math stuff, arg-reuse tailcall
[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
29#include "masm_tools.h"
30
31enum op_flags {
32 OPF_RMD = (1 << 0), /* removed or optimized out */
33 OPF_DATA = (1 << 1), /* data processing - writes to dst opr */
34 OPF_FLAGS = (1 << 2), /* sets flags */
35 OPF_JMP = (1 << 3), /* branch, call */
36 OPF_CJMP = (1 << 4), /* cond. branch (cc or jecxz) */
37 OPF_CC = (1 << 5), /* uses flags */
38 OPF_TAIL = (1 << 6), /* ret or tail call */
39 OPF_RSAVE = (1 << 7), /* push/pop is local reg save/load */
40 OPF_REP = (1 << 8), /* prefixed by rep */
41 OPF_REPZ = (1 << 9), /* rep is repe/repz */
42 OPF_REPNZ = (1 << 10), /* rep is repne/repnz */
43 OPF_FARG = (1 << 11), /* push collected as func arg (no reuse) */
44 OPF_EBP_S = (1 << 12), /* ebp used as scratch here, not BP */
45 OPF_DF = (1 << 13), /* DF flag set */
46 OPF_ATAIL = (1 << 14), /* tail call with reused arg frame */
47};
48
49enum op_op {
50 OP_INVAL,
51 OP_NOP,
52 OP_PUSH,
53 OP_POP,
54 OP_LEAVE,
55 OP_MOV,
56 OP_LEA,
57 OP_MOVZX,
58 OP_MOVSX,
59 OP_XCHG,
60 OP_NOT,
61 OP_CDQ,
62 OP_STOS,
63 OP_MOVS,
64 OP_CMPS,
65 OP_SCAS,
66 OP_STD,
67 OP_CLD,
68 OP_RET,
69 OP_ADD,
70 OP_SUB,
71 OP_AND,
72 OP_OR,
73 OP_XOR,
74 OP_SHL,
75 OP_SHR,
76 OP_SAR,
77 OP_ROL,
78 OP_ROR,
79 OP_ADC,
80 OP_SBB,
81 OP_INC,
82 OP_DEC,
83 OP_NEG,
84 OP_MUL,
85 OP_IMUL,
86 OP_DIV,
87 OP_IDIV,
88 OP_TEST,
89 OP_CMP,
90 OP_CALL,
91 OP_JMP,
92 OP_JECXZ,
93 OP_JO,
94 OP_JNO,
95 OP_JC,
96 OP_JNC,
97 OP_JZ,
98 OP_JNZ,
99 OP_JBE,
100 OP_JA,
101 OP_JS,
102 OP_JNS,
103 OP_JP,
104 OP_JNP,
105 OP_JL,
106 OP_JGE,
107 OP_JLE,
108 OP_JG,
109};
110
111enum opr_type {
112 OPT_UNSPEC,
113 OPT_REG,
114 OPT_REGMEM,
115 OPT_LABEL,
116 OPT_OFFSET,
117 OPT_CONST,
118};
119
120// must be sorted (larger len must be further in enum)
121enum opr_lenmod {
122 OPLM_UNSPEC,
123 OPLM_BYTE,
124 OPLM_WORD,
125 OPLM_DWORD,
126};
127
128#define MAX_OPERANDS 3
129
130struct parsed_opr {
131 enum opr_type type;
132 enum opr_lenmod lmod;
133 unsigned int is_ptr:1; // pointer in C
134 unsigned int is_array:1; // array in C
135 unsigned int type_from_var:1; // .. in header, sometimes wrong
136 unsigned int size_mismatch:1; // type override differs from C
137 unsigned int size_lt:1; // type override is larger than C
138 unsigned int had_ds:1; // had ds: prefix
139 int reg;
140 unsigned int val;
141 char name[256];
142};
143
144struct parsed_op {
145 enum op_op op;
146 struct parsed_opr operand[MAX_OPERANDS];
147 unsigned int flags;
148 int operand_cnt;
149 int regmask_src; // all referensed regs
150 int regmask_dst;
151 int pfomask; // flagop: parsed_flag_op that can't be delayed
152 int argnum; // push: altered before call arg #
153 int cc_scratch; // scratch storage during analysis
154 int bt_i; // branch target for branches
155 struct parsed_data *btj;// branch targets for jumptables
156 void *datap;
157};
158
159// datap:
160// OP_CALL - parser proto hint (str), ptr to struct parsed_proto
161// (OPF_CC) - point to one of (OPF_FLAGS) that affects cc op
162// OP_POP - point to OP_PUSH in push/pop pair
163
164struct parsed_equ {
165 char name[64];
166 enum opr_lenmod lmod;
167 int offset;
168};
169
170struct parsed_data {
171 char label[256];
172 enum opr_type type;
173 enum opr_lenmod lmod;
174 int count;
175 int count_alloc;
176 struct {
177 union {
178 char *label;
179 unsigned int val;
180 } u;
181 int bt_i;
182 } *d;
183};
184
185struct label_ref {
186 int i;
187 struct label_ref *next;
188};
189
190enum ida_func_attr {
191 IDAFA_BP_FRAME = (1 << 0),
192 IDAFA_LIB_FUNC = (1 << 1),
193 IDAFA_STATIC = (1 << 2),
194 IDAFA_NORETURN = (1 << 3),
195 IDAFA_THUNK = (1 << 4),
196 IDAFA_FPD = (1 << 5),
197};
198
199#define MAX_OPS 4096
200
201static struct parsed_op ops[MAX_OPS];
202static struct parsed_equ *g_eqs;
203static int g_eqcnt;
204static char g_labels[MAX_OPS][32];
205static struct label_ref g_label_refs[MAX_OPS];
206static const struct parsed_proto *g_func_pp;
207static struct parsed_data *g_func_pd;
208static int g_func_pd_cnt;
209static char g_func[256];
210static char g_comment[256];
211static int g_bp_frame;
212static int g_sp_frame;
213static int g_stack_frame_used;
214static int g_stack_fsz;
215static int g_ida_func_attr;
216static int g_allow_regfunc;
217#define ferr(op_, fmt, ...) do { \
218 printf("error:%s:#%zd: '%s': " fmt, g_func, (op_) - ops, \
219 dump_op(op_), ##__VA_ARGS__); \
220 fcloseall(); \
221 exit(1); \
222} while (0)
223#define fnote(op_, fmt, ...) \
224 printf("error:%s:#%zd: '%s': " fmt, g_func, (op_) - ops, \
225 dump_op(op_), ##__VA_ARGS__)
226
227#define MAX_REGS 8
228
229const char *regs_r32[] = { "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp" };
230const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" };
231const char *regs_r8l[] = { "al", "bl", "cl", "dl" };
232const char *regs_r8h[] = { "ah", "bh", "ch", "dh" };
233
234enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP };
235
236// possible basic comparison types (without inversion)
237enum parsed_flag_op {
238 PFO_O, // 0 OF=1
239 PFO_C, // 2 CF=1
240 PFO_Z, // 4 ZF=1
241 PFO_BE, // 6 CF=1||ZF=1
242 PFO_S, // 8 SF=1
243 PFO_P, // a PF=1
244 PFO_L, // c SF!=OF
245 PFO_LE, // e ZF=1||SF!=OF
246};
247
248static const char *parsed_flag_op_names[] = {
249 "o", "c", "z", "be", "s", "p", "l", "le"
250};
251
252static int char_array_i(const char *array[], size_t len, const char *s)
253{
254 int i;
255
256 for (i = 0; i < len; i++)
257 if (IS(s, array[i]))
258 return i;
259
260 return -1;
261}
262
263static void printf_number(char *buf, size_t buf_size,
264 unsigned long number)
265{
266 // output in C-friendly form
267 snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
268}
269
270static int parse_reg(enum opr_lenmod *reg_lmod, const char *s)
271{
272 int reg;
273
274 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s);
275 if (reg >= 0) {
276 *reg_lmod = OPLM_DWORD;
277 return reg;
278 }
279 reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s);
280 if (reg >= 0) {
281 *reg_lmod = OPLM_WORD;
282 return reg;
283 }
284 reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s);
285 if (reg >= 0) {
286 *reg_lmod = OPLM_BYTE;
287 return reg;
288 }
289 reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s);
290 if (reg >= 0) {
291 *reg_lmod = OPLM_BYTE;
292 return reg;
293 }
294
295 return -1;
296}
297
298static int parse_indmode(char *name, int *regmask, int need_c_cvt)
299{
300 enum opr_lenmod lmod;
301 char cvtbuf[256];
302 char *d = cvtbuf;
303 char *s = name;
304 char w[64];
305 long number;
306 int reg;
307 int c = 0;
308
309 *d = 0;
310
311 while (*s != 0) {
312 d += strlen(d);
313 while (my_isblank(*s))
314 s++;
315 for (; my_issep(*s); d++, s++)
316 *d = *s;
317 while (my_isblank(*s))
318 s++;
319 *d = 0;
320
321 // skip 'ds:' prefix
322 if (IS_START(s, "ds:"))
323 s += 3;
324
325 s = next_idt(w, sizeof(w), s);
326 if (w[0] == 0)
327 break;
328 c++;
329
330 reg = parse_reg(&lmod, w);
331 if (reg >= 0) {
332 *regmask |= 1 << reg;
333 goto pass;
334 }
335
336 if ('0' <= w[0] && w[0] <= '9') {
337 number = parse_number(w);
338 printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number);
339 continue;
340 }
341
342 // probably some label/identifier - pass
343
344pass:
345 snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w);
346 }
347
348 if (need_c_cvt)
349 strcpy(name, cvtbuf);
350
351 return c;
352}
353
354static int is_reg_in_str(const char *s)
355{
356 int i;
357
358 if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
359 return 0;
360
361 for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
362 if (!strncmp(s, regs_r32[i], 3))
363 return 1;
364
365 return 0;
366}
367
368static const char *parse_stack_el(const char *name, char *extra_reg)
369{
370 const char *p, *p2, *s;
371 char *endp = NULL;
372 char buf[32];
373 long val;
374 int len;
375
376 p = name;
377 if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
378 p += 4;
379 if (extra_reg != NULL) {
380 strncpy(extra_reg, name, 3);
381 extra_reg[4] = 0;
382 }
383 }
384
385 if (IS_START(p, "ebp+")) {
386 p += 4;
387
388 p2 = strchr(p, '+');
389 if (p2 != NULL && is_reg_in_str(p)) {
390 if (extra_reg != NULL) {
391 strncpy(extra_reg, p, p2 - p);
392 extra_reg[p2 - p] = 0;
393 }
394 p = p2 + 1;
395 }
396
397 if (!('0' <= *p && *p <= '9'))
398 return p;
399
400 return NULL;
401 }
402
403 if (!IS_START(name, "esp+"))
404 return NULL;
405
406 p = strchr(name + 4, '+');
407 if (p) {
408 // must be a number after esp+, already converted to 0x..
409 s = name + 4;
410 if (!('0' <= *s && *s <= '9')) {
411 aerr("%s nan?\n", __func__);
412 return NULL;
413 }
414 if (s[0] == '0' && s[1] == 'x')
415 s += 2;
416 len = p - s;
417 if (len < sizeof(buf) - 1) {
418 strncpy(buf, s, len);
419 buf[len] = 0;
420 val = strtol(buf, &endp, 16);
421 if (val == 0 || *endp != 0) {
422 aerr("%s num parse fail for '%s'\n", __func__, buf);
423 return NULL;
424 }
425 }
426 p++;
427 }
428 else
429 p = name + 4;
430
431 if ('0' <= *p && *p <= '9')
432 return NULL;
433
434 return p;
435}
436
437static int guess_lmod_from_name(struct parsed_opr *opr)
438{
439 if (!strncmp(opr->name, "dword_", 6)) {
440 opr->lmod = OPLM_DWORD;
441 return 1;
442 }
443 if (!strncmp(opr->name, "word_", 5)) {
444 opr->lmod = OPLM_WORD;
445 return 1;
446 }
447 if (!strncmp(opr->name, "byte_", 5)) {
448 opr->lmod = OPLM_BYTE;
449 return 1;
450 }
451 return 0;
452}
453
454static int guess_lmod_from_c_type(enum opr_lenmod *lmod,
455 const struct parsed_type *c_type)
456{
457 static const char *dword_types[] = {
458 "int", "_DWORD", "UINT_PTR", "DWORD",
459 "WPARAM", "LPARAM", "UINT", "__int32",
460 "LONG", "HIMC",
461 };
462 static const char *word_types[] = {
463 "uint16_t", "int16_t", "_WORD",
464 "unsigned __int16", "__int16",
465 };
466 static const char *byte_types[] = {
467 "uint8_t", "int8_t", "char",
468 "unsigned __int8", "__int8", "BYTE", "_BYTE",
469 "CHAR", "_UNKNOWN",
470 };
471 const char *n;
472 int i;
473
474 if (c_type->is_ptr) {
475 *lmod = OPLM_DWORD;
476 return 1;
477 }
478
479 n = skip_type_mod(c_type->name);
480
481 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
482 if (IS(n, dword_types[i])) {
483 *lmod = OPLM_DWORD;
484 return 1;
485 }
486 }
487
488 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
489 if (IS(n, word_types[i])) {
490 *lmod = OPLM_WORD;
491 return 1;
492 }
493 }
494
495 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
496 if (IS(n, byte_types[i])) {
497 *lmod = OPLM_BYTE;
498 return 1;
499 }
500 }
501
502 return 0;
503}
504
505static enum opr_type lmod_from_directive(const char *d)
506{
507 if (IS(d, "dd"))
508 return OPLM_DWORD;
509 else if (IS(d, "dw"))
510 return OPLM_WORD;
511 else if (IS(d, "db"))
512 return OPLM_BYTE;
513
514 aerr("unhandled directive: '%s'\n", d);
515 return OPLM_UNSPEC;
516}
517
518static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
519 int *regmask)
520{
521 opr->type = OPT_REG;
522 opr->reg = reg;
523 opr->lmod = lmod;
524 *regmask |= 1 << reg;
525}
526
527static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
528 int *extra_offs);
529
530static int parse_operand(struct parsed_opr *opr,
531 int *regmask, int *regmask_indirect,
532 char words[16][256], int wordc, int w, unsigned int op_flags)
533{
534 const struct parsed_proto *pp;
535 enum opr_lenmod tmplmod;
536 unsigned long number;
537 char buf[256];
538 int ret, len;
539 int wordc_in;
540 char *p;
541 int i;
542
543 if (w >= wordc)
544 aerr("parse_operand w %d, wordc %d\n", w, wordc);
545
546 opr->reg = xUNSPEC;
547
548 for (i = w; i < wordc; i++) {
549 len = strlen(words[i]);
550 if (words[i][len - 1] == ',') {
551 words[i][len - 1] = 0;
552 wordc = i + 1;
553 break;
554 }
555 }
556
557 wordc_in = wordc - w;
558
559 if ((op_flags & OPF_JMP) && wordc_in > 0
560 && !('0' <= words[w][0] && words[w][0] <= '9'))
561 {
562 const char *label = NULL;
563
564 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
565 && IS(words[w + 1], "ptr"))
566 label = words[w + 2];
567 else if (wordc_in == 2 && IS(words[w], "short"))
568 label = words[w + 1];
569 else if (wordc_in == 1
570 && strchr(words[w], '[') == NULL
571 && parse_reg(&tmplmod, words[w]) < 0)
572 label = words[w];
573
574 if (label != NULL) {
575 opr->type = OPT_LABEL;
576 if (IS_START(label, "ds:")) {
577 opr->had_ds = 1;
578 label += 3;
579 }
580 strcpy(opr->name, label);
581 return wordc;
582 }
583 }
584
585 if (wordc_in >= 3) {
586 if (IS(words[w + 1], "ptr")) {
587 if (IS(words[w], "dword"))
588 opr->lmod = OPLM_DWORD;
589 else if (IS(words[w], "word"))
590 opr->lmod = OPLM_WORD;
591 else if (IS(words[w], "byte"))
592 opr->lmod = OPLM_BYTE;
593 else
594 aerr("type parsing failed\n");
595 w += 2;
596 wordc_in = wordc - w;
597 }
598 }
599
600 if (wordc_in == 2) {
601 if (IS(words[w], "offset")) {
602 opr->type = OPT_OFFSET;
603 strcpy(opr->name, words[w + 1]);
604 return wordc;
605 }
606 if (IS(words[w], "(offset")) {
607 p = strchr(words[w + 1], ')');
608 if (p == NULL)
609 aerr("parse of bracketed offset failed\n");
610 *p = 0;
611 opr->type = OPT_OFFSET;
612 strcpy(opr->name, words[w + 1]);
613 return wordc;
614 }
615 }
616
617 if (wordc_in != 1)
618 aerr("parse_operand 1 word expected\n");
619
620 if (IS_START(words[w], "ds:")) {
621 opr->had_ds = 1;
622 memmove(words[w], words[w] + 3, strlen(words[w]) - 2);
623 }
624 strcpy(opr->name, words[w]);
625
626 if (words[w][0] == '[') {
627 opr->type = OPT_REGMEM;
628 ret = sscanf(words[w], "[%[^]]]", opr->name);
629 if (ret != 1)
630 aerr("[] parse failure\n");
631
632 parse_indmode(opr->name, regmask_indirect, 1);
633 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL)) {
634 // might be an equ
635 struct parsed_equ *eq =
636 equ_find(NULL, parse_stack_el(opr->name, NULL), &i);
637 if (eq)
638 opr->lmod = eq->lmod;
639 }
640 return wordc;
641 }
642 else if (strchr(words[w], '[')) {
643 // label[reg] form
644 p = strchr(words[w], '[');
645 opr->type = OPT_REGMEM;
646 parse_indmode(p, regmask_indirect, 0);
647 strncpy(buf, words[w], p - words[w]);
648 buf[p - words[w]] = 0;
649 pp = proto_parse(g_fhdr, buf, 1);
650 goto do_label;
651 }
652 else if (('0' <= words[w][0] && words[w][0] <= '9')
653 || words[w][0] == '-')
654 {
655 number = parse_number(words[w]);
656 opr->type = OPT_CONST;
657 opr->val = number;
658 printf_number(opr->name, sizeof(opr->name), number);
659 return wordc;
660 }
661
662 ret = parse_reg(&tmplmod, opr->name);
663 if (ret >= 0) {
664 setup_reg_opr(opr, ret, tmplmod, regmask);
665 return wordc;
666 }
667
668 // most likely var in data segment
669 opr->type = OPT_LABEL;
670 pp = proto_parse(g_fhdr, opr->name, 0);
671
672do_label:
673 if (pp != NULL) {
674 if (pp->is_fptr || pp->is_func) {
675 opr->lmod = OPLM_DWORD;
676 opr->is_ptr = 1;
677 }
678 else {
679 tmplmod = OPLM_UNSPEC;
680 if (!guess_lmod_from_c_type(&tmplmod, &pp->type))
681 anote("unhandled C type '%s' for '%s'\n",
682 pp->type.name, opr->name);
683
684 if (opr->lmod == OPLM_UNSPEC) {
685 opr->lmod = tmplmod;
686 opr->type_from_var = 1;
687 }
688 else if (opr->lmod != tmplmod) {
689 opr->size_mismatch = 1;
690 if (tmplmod < opr->lmod)
691 opr->size_lt = 1;
692 }
693 opr->is_ptr = pp->type.is_ptr;
694 }
695 opr->is_array = pp->type.is_array;
696 }
697
698 if (opr->lmod == OPLM_UNSPEC)
699 guess_lmod_from_name(opr);
700 return wordc;
701}
702
703static const struct {
704 const char *name;
705 unsigned int flags;
706} pref_table[] = {
707 { "rep", OPF_REP },
708 { "repe", OPF_REP|OPF_REPZ },
709 { "repz", OPF_REP|OPF_REPZ },
710 { "repne", OPF_REP|OPF_REPNZ },
711 { "repnz", OPF_REP|OPF_REPNZ },
712};
713
714#define OPF_CJMP_CC (OPF_JMP|OPF_CJMP|OPF_CC)
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 { "leave",OP_LEAVE, 0, 0, OPF_DATA },
727 { "mov" , OP_MOV, 2, 2, OPF_DATA },
728 { "lea", OP_LEA, 2, 2, OPF_DATA },
729 { "movzx",OP_MOVZX, 2, 2, OPF_DATA },
730 { "movsx",OP_MOVSX, 2, 2, OPF_DATA },
731 { "xchg", OP_XCHG, 2, 2, OPF_DATA },
732 { "not", OP_NOT, 1, 1, OPF_DATA },
733 { "cdq", OP_CDQ, 0, 0, OPF_DATA },
734 { "stosb",OP_STOS, 0, 0, OPF_DATA },
735 { "stosw",OP_STOS, 0, 0, OPF_DATA },
736 { "stosd",OP_STOS, 0, 0, OPF_DATA },
737 { "movsb",OP_MOVS, 0, 0, OPF_DATA },
738 { "movsw",OP_MOVS, 0, 0, OPF_DATA },
739 { "movsd",OP_MOVS, 0, 0, OPF_DATA },
740 { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
741 { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
742 { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
743 { "scasb",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
744 { "scasw",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
745 { "scasd",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
746 { "std", OP_STD, 0, 0, OPF_DATA }, // special flag
747 { "cld", OP_CLD, 0, 0, OPF_DATA },
748 { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
749 { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
750 { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
751 { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS },
752 { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS },
753 { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
754 { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
755 { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
756 { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
757 { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
758 { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
759 { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
760 { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
761 { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
762 { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS },
763 { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS },
764 { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS },
765 { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS },
766 { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS },
767 { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS },
768 { "test", OP_TEST, 2, 2, OPF_FLAGS },
769 { "cmp", OP_CMP, 2, 2, OPF_FLAGS },
770 { "retn", OP_RET, 0, 1, OPF_TAIL },
771 { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS },
772 { "jmp", OP_JMP, 1, 1, OPF_JMP },
773 { "jecxz",OP_JECXZ, 1, 1, OPF_JMP|OPF_CJMP },
774 { "jo", OP_JO, 1, 1, OPF_CJMP_CC }, // 70 OF=1
775 { "jno", OP_JNO, 1, 1, OPF_CJMP_CC }, // 71 OF=0
776 { "jc", OP_JC, 1, 1, OPF_CJMP_CC }, // 72 CF=1
777 { "jb", OP_JC, 1, 1, OPF_CJMP_CC }, // 72
778 { "jnc", OP_JNC, 1, 1, OPF_CJMP_CC }, // 73 CF=0
779 { "jnb", OP_JNC, 1, 1, OPF_CJMP_CC }, // 73
780 { "jae", OP_JNC, 1, 1, OPF_CJMP_CC }, // 73
781 { "jz", OP_JZ, 1, 1, OPF_CJMP_CC }, // 74 ZF=1
782 { "je", OP_JZ, 1, 1, OPF_CJMP_CC }, // 74
783 { "jnz", OP_JNZ, 1, 1, OPF_CJMP_CC }, // 75 ZF=0
784 { "jne", OP_JNZ, 1, 1, OPF_CJMP_CC }, // 75
785 { "jbe", OP_JBE, 1, 1, OPF_CJMP_CC }, // 76 CF=1 || ZF=1
786 { "jna", OP_JBE, 1, 1, OPF_CJMP_CC }, // 76
787 { "ja", OP_JA, 1, 1, OPF_CJMP_CC }, // 77 CF=0 && ZF=0
788 { "jnbe", OP_JA, 1, 1, OPF_CJMP_CC }, // 77
789 { "js", OP_JS, 1, 1, OPF_CJMP_CC }, // 78 SF=1
790 { "jns", OP_JNS, 1, 1, OPF_CJMP_CC }, // 79 SF=0
791 { "jp", OP_JP, 1, 1, OPF_CJMP_CC }, // 7a PF=1
792 { "jpe", OP_JP, 1, 1, OPF_CJMP_CC }, // 7a
793 { "jnp", OP_JNP, 1, 1, OPF_CJMP_CC }, // 7b PF=0
794 { "jpo", OP_JNP, 1, 1, OPF_CJMP_CC }, // 7b
795 { "jl", OP_JL, 1, 1, OPF_CJMP_CC }, // 7c SF!=OF
796 { "jnge", OP_JL, 1, 1, OPF_CJMP_CC }, // 7c
797 { "jge", OP_JGE, 1, 1, OPF_CJMP_CC }, // 7d SF=OF
798 { "jnl", OP_JGE, 1, 1, OPF_CJMP_CC }, // 7d
799 { "jle", OP_JLE, 1, 1, OPF_CJMP_CC }, // 7e ZF=1 || SF!=OF
800 { "jng", OP_JLE, 1, 1, OPF_CJMP_CC }, // 7e
801 { "jg", OP_JG, 1, 1, OPF_CJMP_CC }, // 7f ZF=0 && SF=OF
802 { "jnle", OP_JG, 1, 1, OPF_CJMP_CC }, // 7f
803 { "seto", OP_JO, 1, 1, OPF_DATA|OPF_CC },
804 { "setno", OP_JNO, 1, 1, OPF_DATA|OPF_CC },
805 { "setc", OP_JC, 1, 1, OPF_DATA|OPF_CC },
806 { "setb", OP_JC, 1, 1, OPF_DATA|OPF_CC },
807 { "setnc", OP_JNC, 1, 1, OPF_DATA|OPF_CC },
808 { "setae", OP_JNC, 1, 1, OPF_DATA|OPF_CC },
809 { "setz", OP_JZ, 1, 1, OPF_DATA|OPF_CC },
810 { "sete", OP_JZ, 1, 1, OPF_DATA|OPF_CC },
811 { "setnz", OP_JNZ, 1, 1, OPF_DATA|OPF_CC },
812 { "setne", OP_JNZ, 1, 1, OPF_DATA|OPF_CC },
813 { "setbe", OP_JBE, 1, 1, OPF_DATA|OPF_CC },
814 { "setna", OP_JBE, 1, 1, OPF_DATA|OPF_CC },
815 { "seta", OP_JA, 1, 1, OPF_DATA|OPF_CC },
816 { "setnbe", OP_JA, 1, 1, OPF_DATA|OPF_CC },
817 { "sets", OP_JS, 1, 1, OPF_DATA|OPF_CC },
818 { "setns", OP_JNS, 1, 1, OPF_DATA|OPF_CC },
819 { "setp", OP_JP, 1, 1, OPF_DATA|OPF_CC },
820 { "setpe", OP_JP, 1, 1, OPF_DATA|OPF_CC },
821 { "setnp", OP_JNP, 1, 1, OPF_DATA|OPF_CC },
822 { "setpo", OP_JNP, 1, 1, OPF_DATA|OPF_CC },
823 { "setl", OP_JL, 1, 1, OPF_DATA|OPF_CC },
824 { "setnge", OP_JL, 1, 1, OPF_DATA|OPF_CC },
825 { "setge", OP_JGE, 1, 1, OPF_DATA|OPF_CC },
826 { "setnl", OP_JGE, 1, 1, OPF_DATA|OPF_CC },
827 { "setle", OP_JLE, 1, 1, OPF_DATA|OPF_CC },
828 { "setng", OP_JLE, 1, 1, OPF_DATA|OPF_CC },
829 { "setg", OP_JG, 1, 1, OPF_DATA|OPF_CC },
830 { "setnle", OP_JG, 1, 1, OPF_DATA|OPF_CC },
831};
832
833static void parse_op(struct parsed_op *op, char words[16][256], int wordc)
834{
835 enum opr_lenmod lmod = OPLM_UNSPEC;
836 int prefix_flags = 0;
837 int regmask_ind;
838 int regmask;
839 int op_w = 0;
840 int opr = 0;
841 int w = 0;
842 int i;
843
844 for (i = 0; i < ARRAY_SIZE(pref_table); i++) {
845 if (IS(words[w], pref_table[i].name)) {
846 prefix_flags = pref_table[i].flags;
847 break;
848 }
849 }
850
851 if (prefix_flags) {
852 if (wordc <= 1)
853 aerr("lone prefix: '%s'\n", words[0]);
854 w++;
855 }
856
857 op_w = w;
858 for (i = 0; i < ARRAY_SIZE(op_table); i++) {
859 if (IS(words[w], op_table[i].name))
860 break;
861 }
862
863 if (i == ARRAY_SIZE(op_table))
864 aerr("unhandled op: '%s'\n", words[0]);
865 w++;
866
867 op->op = op_table[i].op;
868 op->flags = op_table[i].flags | prefix_flags;
869 op->regmask_src = op->regmask_dst = 0;
870
871 for (opr = 0; opr < op_table[i].minopr; opr++) {
872 regmask = regmask_ind = 0;
873 w = parse_operand(&op->operand[opr], &regmask, &regmask_ind,
874 words, wordc, w, op->flags);
875
876 if (opr == 0 && (op->flags & OPF_DATA))
877 op->regmask_dst = regmask;
878 // for now, mark dst as src too
879 op->regmask_src |= regmask | regmask_ind;
880 }
881
882 for (; w < wordc && opr < op_table[i].maxopr; opr++) {
883 w = parse_operand(&op->operand[opr],
884 &op->regmask_src, &op->regmask_src,
885 words, wordc, w, op->flags);
886 }
887
888 if (w < wordc)
889 aerr("parse_op %s incomplete: %d/%d\n",
890 words[0], w, wordc);
891
892 // special cases
893 op->operand_cnt = opr;
894 if (!strncmp(op_table[i].name, "set", 3))
895 op->operand[0].lmod = OPLM_BYTE;
896
897 // ops with implicit argumets
898 switch (op->op) {
899 case OP_CDQ:
900 op->operand_cnt = 2;
901 setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst);
902 setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src);
903 break;
904
905 case OP_STOS:
906 case OP_SCAS:
907 if (op->operand_cnt != 0)
908 break;
909 if (words[op_w][4] == 'b')
910 lmod = OPLM_BYTE;
911 else if (words[op_w][4] == 'w')
912 lmod = OPLM_WORD;
913 else if (words[op_w][4] == 'd')
914 lmod = OPLM_DWORD;
915 op->operand_cnt = 3;
916 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
917 setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src);
918 op->regmask_dst = op->regmask_src;
919 setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD, &op->regmask_src);
920 break;
921
922 case OP_MOVS:
923 case OP_CMPS:
924 if (op->operand_cnt != 0)
925 break;
926 if (words[op_w][4] == 'b')
927 lmod = OPLM_BYTE;
928 else if (words[op_w][4] == 'w')
929 lmod = OPLM_WORD;
930 else if (words[op_w][4] == 'd')
931 lmod = OPLM_DWORD;
932 op->operand_cnt = 3;
933 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
934 setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src);
935 setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src);
936 op->regmask_dst = op->regmask_src;
937 break;
938
939 case OP_XCHG:
940 op->regmask_src |= op->regmask_dst;
941 op->regmask_dst |= op->regmask_src;
942 break;
943
944 case OP_JECXZ:
945 op->operand_cnt = 1;
946 op->regmask_src = 1 << xCX;
947 op->operand[0].type = OPT_REG;
948 op->operand[0].reg = xCX;
949 op->operand[0].lmod = OPLM_DWORD;
950 break;
951
952 case OP_IMUL:
953 if (op->operand_cnt != 1)
954 break;
955 // fallthrough
956 case OP_MUL:
957 // singleop mul
958 op->regmask_dst = (1 << xDX) | (1 << xAX);
959 op->regmask_src |= (1 << xAX);
960 if (op->operand[0].lmod == OPLM_UNSPEC)
961 op->operand[0].lmod = OPLM_DWORD;
962 break;
963
964 case OP_DIV:
965 case OP_IDIV:
966 // we could set up operands for edx:eax, but there is no real need to
967 // (see is_opr_modified())
968 regmask = (1 << xDX) | (1 << xAX);
969 op->regmask_dst = regmask;
970 op->regmask_src |= regmask;
971 if (op->operand[0].lmod == OPLM_UNSPEC)
972 op->operand[0].lmod = OPLM_DWORD;
973 break;
974
975 case OP_SHL:
976 case OP_SHR:
977 case OP_SAR:
978 case OP_ROL:
979 case OP_ROR:
980 if (op->operand[1].lmod == OPLM_UNSPEC)
981 op->operand[1].lmod = OPLM_BYTE;
982 break;
983
984 case OP_PUSH:
985 if (op->operand[0].lmod == OPLM_UNSPEC
986 && (op->operand[0].type == OPT_CONST
987 || op->operand[0].type == OPT_OFFSET
988 || op->operand[0].type == OPT_LABEL))
989 op->operand[0].lmod = OPLM_DWORD;
990 break;
991
992 // alignment
993 case OP_MOV:
994 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
995 && op->operand[0].reg == xDI && op->operand[1].reg == xDI)
996 {
997 op->flags |= OPF_RMD;
998 }
999 break;
1000
1001 case OP_LEA:
1002 if (op->operand[0].type == OPT_REG
1003 && op->operand[1].type == OPT_REGMEM)
1004 {
1005 char buf[16];
1006 snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name);
1007 if (IS(buf, op->operand[1].name))
1008 op->flags |= OPF_RMD;
1009 }
1010 break;
1011
1012 case OP_CALL:
1013 // trashed regs must be explicitly detected later
1014 op->regmask_dst = 0;
1015 break;
1016
1017 default:
1018 break;
1019 }
1020}
1021
1022static const char *op_name(enum op_op op)
1023{
1024 int i;
1025
1026 for (i = 0; i < ARRAY_SIZE(op_table); i++)
1027 if (op_table[i].op == op)
1028 return op_table[i].name;
1029
1030 return "???";
1031}
1032
1033// debug
1034static const char *dump_op(struct parsed_op *po)
1035{
1036 static char out[128];
1037 char *p = out;
1038 int i;
1039
1040 if (po == NULL)
1041 return "???";
1042
1043 snprintf(out, sizeof(out), "%s", op_name(po->op));
1044 for (i = 0; i < po->operand_cnt; i++) {
1045 p += strlen(p);
1046 if (i > 0)
1047 *p++ = ',';
1048 snprintf(p, sizeof(out) - (p - out),
1049 po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s",
1050 po->operand[i].name);
1051 }
1052
1053 return out;
1054}
1055
1056static const char *lmod_type_u(struct parsed_op *po,
1057 enum opr_lenmod lmod)
1058{
1059 switch (lmod) {
1060 case OPLM_DWORD:
1061 return "u32";
1062 case OPLM_WORD:
1063 return "u16";
1064 case OPLM_BYTE:
1065 return "u8";
1066 default:
1067 ferr(po, "invalid lmod: %d\n", lmod);
1068 return "(_invalid_)";
1069 }
1070}
1071
1072static const char *lmod_cast_u(struct parsed_op *po,
1073 enum opr_lenmod lmod)
1074{
1075 switch (lmod) {
1076 case OPLM_DWORD:
1077 return "";
1078 case OPLM_WORD:
1079 return "(u16)";
1080 case OPLM_BYTE:
1081 return "(u8)";
1082 default:
1083 ferr(po, "invalid lmod: %d\n", lmod);
1084 return "(_invalid_)";
1085 }
1086}
1087
1088static const char *lmod_cast_u_ptr(struct parsed_op *po,
1089 enum opr_lenmod lmod)
1090{
1091 switch (lmod) {
1092 case OPLM_DWORD:
1093 return "*(u32 *)";
1094 case OPLM_WORD:
1095 return "*(u16 *)";
1096 case OPLM_BYTE:
1097 return "*(u8 *)";
1098 default:
1099 ferr(po, "invalid lmod: %d\n", lmod);
1100 return "(_invalid_)";
1101 }
1102}
1103
1104static const char *lmod_cast_s(struct parsed_op *po,
1105 enum opr_lenmod lmod)
1106{
1107 switch (lmod) {
1108 case OPLM_DWORD:
1109 return "(s32)";
1110 case OPLM_WORD:
1111 return "(s16)";
1112 case OPLM_BYTE:
1113 return "(s8)";
1114 default:
1115 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1116 return "(_invalid_)";
1117 }
1118}
1119
1120static const char *lmod_cast(struct parsed_op *po,
1121 enum opr_lenmod lmod, int is_signed)
1122{
1123 return is_signed ?
1124 lmod_cast_s(po, lmod) :
1125 lmod_cast_u(po, lmod);
1126}
1127
1128static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
1129{
1130 switch (lmod) {
1131 case OPLM_DWORD:
1132 return 4;
1133 case OPLM_WORD:
1134 return 2;
1135 case OPLM_BYTE:
1136 return 1;
1137 default:
1138 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1139 return 0;
1140 }
1141}
1142
1143static const char *opr_name(struct parsed_op *po, int opr_num)
1144{
1145 if (opr_num >= po->operand_cnt)
1146 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1147 return po->operand[opr_num].name;
1148}
1149
1150static unsigned int opr_const(struct parsed_op *po, int opr_num)
1151{
1152 if (opr_num >= po->operand_cnt)
1153 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1154 if (po->operand[opr_num].type != OPT_CONST)
1155 ferr(po, "opr %d: const expected\n", opr_num);
1156 return po->operand[opr_num].val;
1157}
1158
1159static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr)
1160{
1161 if ((unsigned int)popr->reg >= MAX_REGS)
1162 ferr(po, "invalid reg: %d\n", popr->reg);
1163 return regs_r32[popr->reg];
1164}
1165
1166// cast1 is the "final" cast
1167static const char *simplify_cast(const char *cast1, const char *cast2)
1168{
1169 static char buf[256];
1170
1171 if (cast1[0] == 0)
1172 return cast2;
1173 if (cast2[0] == 0)
1174 return cast1;
1175 if (IS(cast1, cast2))
1176 return cast1;
1177 if (IS(cast1, "(s8)") && IS(cast2, "(u8)"))
1178 return cast1;
1179 if (IS(cast1, "(s16)") && IS(cast2, "(u16)"))
1180 return cast1;
1181 if (IS(cast1, "(u8)") && IS_START(cast2, "*(u8 *)"))
1182 return cast2;
1183 if (IS(cast1, "(u16)") && IS_START(cast2, "*(u16 *)"))
1184 return cast2;
1185 if (strchr(cast1, '*') && IS_START(cast2, "(u32)"))
1186 return cast1;
1187
1188 snprintf(buf, sizeof(buf), "%s%s", cast1, cast2);
1189 return buf;
1190}
1191
1192static const char *simplify_cast_num(const char *cast, unsigned int val)
1193{
1194 if (IS(cast, "(u8)") && val < 0x100)
1195 return "";
1196 if (IS(cast, "(s8)") && val < 0x80)
1197 return "";
1198 if (IS(cast, "(u16)") && val < 0x10000)
1199 return "";
1200 if (IS(cast, "(s16)") && val < 0x8000)
1201 return "";
1202 if (IS(cast, "(s32)") && val < 0x80000000)
1203 return "";
1204
1205 return cast;
1206}
1207
1208static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
1209 int *extra_offs)
1210{
1211 const char *p;
1212 char *endp;
1213 int namelen;
1214 int i;
1215
1216 *extra_offs = 0;
1217 namelen = strlen(name);
1218
1219 p = strchr(name, '+');
1220 if (p != NULL) {
1221 namelen = p - name;
1222 if (namelen <= 0)
1223 ferr(po, "equ parse failed for '%s'\n", name);
1224
1225 if (IS_START(p, "0x"))
1226 p += 2;
1227 *extra_offs = strtol(p, &endp, 16);
1228 if (*endp != 0)
1229 ferr(po, "equ parse failed for '%s'\n", name);
1230 }
1231
1232 for (i = 0; i < g_eqcnt; i++)
1233 if (strncmp(g_eqs[i].name, name, namelen) == 0
1234 && g_eqs[i].name[namelen] == 0)
1235 break;
1236 if (i >= g_eqcnt) {
1237 if (po != NULL)
1238 ferr(po, "unresolved equ name: '%s'\n", name);
1239 return NULL;
1240 }
1241
1242 return &g_eqs[i];
1243}
1244
1245static int is_stack_access(struct parsed_op *po,
1246 const struct parsed_opr *popr)
1247{
1248 return (parse_stack_el(popr->name, NULL)
1249 || (g_bp_frame && !(po->flags & OPF_EBP_S)
1250 && IS_START(popr->name, "ebp")));
1251}
1252
1253static void parse_stack_access(struct parsed_op *po,
1254 const char *name, char *ofs_reg, int *offset_out,
1255 int *stack_ra_out, const char **bp_arg_out)
1256{
1257 const char *bp_arg = "";
1258 const char *p = NULL;
1259 struct parsed_equ *eq;
1260 char *endp = NULL;
1261 int stack_ra = 0;
1262 int offset = 0;
1263
1264 ofs_reg[0] = 0;
1265
1266 if (IS_START(name, "ebp-")
1267 || (IS_START(name, "ebp+") && '0' <= name[4] && name[4] <= '9'))
1268 {
1269 p = name + 4;
1270 if (IS_START(p, "0x"))
1271 p += 2;
1272 offset = strtoul(p, &endp, 16);
1273 if (name[3] == '-')
1274 offset = -offset;
1275 if (*endp != 0)
1276 ferr(po, "ebp- parse of '%s' failed\n", name);
1277 }
1278 else {
1279 bp_arg = parse_stack_el(name, ofs_reg);
1280 snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
1281 eq = equ_find(po, bp_arg, &offset);
1282 if (eq == NULL)
1283 ferr(po, "detected but missing eq\n");
1284 offset += eq->offset;
1285 }
1286
1287 if (!strncmp(name, "ebp", 3))
1288 stack_ra = 4;
1289
1290 if (ofs_reg[0] == 0 && stack_ra <= offset && offset < stack_ra + 4)
1291 ferr(po, "reference to ra? %d %d\n", offset, stack_ra);
1292
1293 *offset_out = offset;
1294 *stack_ra_out = stack_ra;
1295 if (bp_arg_out)
1296 *bp_arg_out = bp_arg;
1297}
1298
1299static void stack_frame_access(struct parsed_op *po,
1300 struct parsed_opr *popr, char *buf, size_t buf_size,
1301 const char *name, const char *cast, int is_src, int is_lea)
1302{
1303 enum opr_lenmod tmp_lmod = OPLM_UNSPEC;
1304 const char *prefix = "";
1305 const char *bp_arg = NULL;
1306 char ofs_reg[16] = { 0, };
1307 int i, arg_i, arg_s;
1308 int unaligned = 0;
1309 int stack_ra = 0;
1310 int offset = 0;
1311 int sf_ofs;
1312 int lim;
1313
1314 if (po->flags & OPF_EBP_S)
1315 ferr(po, "stack_frame_access while ebp is scratch\n");
1316
1317 parse_stack_access(po, name, ofs_reg, &offset, &stack_ra, &bp_arg);
1318
1319 if (offset > stack_ra)
1320 {
1321 arg_i = (offset - stack_ra - 4) / 4;
1322 if (arg_i < 0 || arg_i >= g_func_pp->argc_stack)
1323 {
1324 if (g_func_pp->is_vararg
1325 && arg_i == g_func_pp->argc_stack && is_lea)
1326 {
1327 // should be va_list
1328 if (cast[0] == 0)
1329 cast = "(u32)";
1330 snprintf(buf, buf_size, "%sap", cast);
1331 return;
1332 }
1333 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1334 offset, bp_arg, arg_i);
1335 }
1336 if (ofs_reg[0] != 0)
1337 ferr(po, "offset reg on arg access?\n");
1338
1339 for (i = arg_s = 0; i < g_func_pp->argc; i++) {
1340 if (g_func_pp->arg[i].reg != NULL)
1341 continue;
1342 if (arg_s == arg_i)
1343 break;
1344 arg_s++;
1345 }
1346 if (i == g_func_pp->argc)
1347 ferr(po, "arg %d not in prototype?\n", arg_i);
1348
1349 popr->is_ptr = g_func_pp->arg[i].type.is_ptr;
1350
1351 switch (popr->lmod)
1352 {
1353 case OPLM_BYTE:
1354 if (is_lea)
1355 ferr(po, "lea/byte to arg?\n");
1356 if (is_src && (offset & 3) == 0)
1357 snprintf(buf, buf_size, "%sa%d",
1358 simplify_cast(cast, "(u8)"), i + 1);
1359 else
1360 snprintf(buf, buf_size, "%sBYTE%d(a%d)",
1361 cast, offset & 3, i + 1);
1362 break;
1363
1364 case OPLM_WORD:
1365 if (is_lea)
1366 ferr(po, "lea/word to arg?\n");
1367 if (offset & 1) {
1368 unaligned = 1;
1369 if (!is_src) {
1370 if (offset & 2)
1371 ferr(po, "problematic arg store\n");
1372 snprintf(buf, buf_size, "%s((char *)&a%d + 1)",
1373 simplify_cast(cast, "*(u16 *)"), i + 1);
1374 }
1375 else
1376 ferr(po, "unaligned arg word load\n");
1377 }
1378 else if (is_src && (offset & 2) == 0)
1379 snprintf(buf, buf_size, "%sa%d",
1380 simplify_cast(cast, "(u16)"), i + 1);
1381 else
1382 snprintf(buf, buf_size, "%s%sWORD(a%d)",
1383 cast, (offset & 2) ? "HI" : "LO", i + 1);
1384 break;
1385
1386 case OPLM_DWORD:
1387 if (cast[0])
1388 prefix = cast;
1389 else if (is_src)
1390 prefix = "(u32)";
1391
1392 if (offset & 3) {
1393 unaligned = 1;
1394 if (is_lea)
1395 snprintf(buf, buf_size, "(u32)&a%d + %d",
1396 i + 1, offset & 3);
1397 else if (!is_src)
1398 ferr(po, "unaligned arg store\n");
1399 else {
1400 // mov edx, [ebp+arg_4+2]; movsx ecx, dx
1401 snprintf(buf, buf_size, "%s(a%d >> %d)",
1402 prefix, i + 1, (offset & 3) * 8);
1403 }
1404 }
1405 else {
1406 snprintf(buf, buf_size, "%s%sa%d",
1407 prefix, is_lea ? "&" : "", i + 1);
1408 }
1409 break;
1410
1411 default:
1412 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
1413 }
1414
1415 if (unaligned)
1416 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1417
1418 // common problem
1419 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type);
1420 if (tmp_lmod != OPLM_DWORD
1421 && (unaligned || (!is_src && tmp_lmod < popr->lmod)))
1422 {
1423 ferr(po, "bp_arg arg%d/w offset %d and type '%s' is too small\n",
1424 i + 1, offset, g_func_pp->arg[i].type.name);
1425 }
1426 // can't check this because msvc likes to reuse
1427 // arg space for scratch..
1428 //if (popr->is_ptr && popr->lmod != OPLM_DWORD)
1429 // ferr(po, "bp_arg arg%d: non-dword ptr access\n", i + 1);
1430 }
1431 else
1432 {
1433 if (g_stack_fsz == 0)
1434 ferr(po, "stack var access without stackframe\n");
1435 g_stack_frame_used = 1;
1436
1437 sf_ofs = g_stack_fsz + offset;
1438 lim = (ofs_reg[0] != 0) ? -4 : 0;
1439 if (offset > 0 || sf_ofs < lim)
1440 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
1441
1442 if (is_lea)
1443 prefix = "(u32)&";
1444 else
1445 prefix = cast;
1446
1447 switch (popr->lmod)
1448 {
1449 case OPLM_BYTE:
1450 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1451 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1452 break;
1453
1454 case OPLM_WORD:
1455 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1456 // known unaligned or possibly unaligned
1457 strcat(g_comment, " unaligned");
1458 if (prefix[0] == 0)
1459 prefix = "*(u16 *)&";
1460 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1461 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1462 break;
1463 }
1464 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
1465 break;
1466
1467 case OPLM_DWORD:
1468 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1469 // known unaligned or possibly unaligned
1470 strcat(g_comment, " unaligned");
1471 if (prefix[0] == 0)
1472 prefix = "*(u32 *)&";
1473 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1474 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1475 break;
1476 }
1477 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
1478 break;
1479
1480 default:
1481 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
1482 }
1483 }
1484}
1485
1486static void check_func_pp(struct parsed_op *po,
1487 const struct parsed_proto *pp, const char *pfx)
1488{
1489 char buf[256];
1490
1491 if (pp->argc_reg != 0) {
1492 if (/*!g_allow_regfunc &&*/ !pp->is_fastcall) {
1493 pp_print(buf, sizeof(buf), pp);
1494 ferr(po, "%s: unexpected reg arg in icall: %s\n", pfx, buf);
1495 }
1496 if (pp->argc_stack > 0 && pp->argc_reg != 2)
1497 ferr(po, "%s: %d reg arg(s) with %d stack arg(s)\n",
1498 pfx, pp->argc_reg, pp->argc_stack);
1499 }
1500}
1501
1502static void check_label_read_ref(struct parsed_op *po, const char *name)
1503{
1504 const struct parsed_proto *pp;
1505
1506 pp = proto_parse(g_fhdr, name, 0);
1507 if (pp == NULL)
1508 ferr(po, "proto_parse failed for ref '%s'\n", name);
1509
1510 if (pp->is_func)
1511 check_func_pp(po, pp, "ref");
1512}
1513
1514static char *out_src_opr(char *buf, size_t buf_size,
1515 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
1516 int is_lea)
1517{
1518 char tmp1[256], tmp2[256];
1519 char expr[256];
1520 char *p;
1521 int ret;
1522
1523 if (cast == NULL)
1524 cast = "";
1525
1526 switch (popr->type) {
1527 case OPT_REG:
1528 if (is_lea)
1529 ferr(po, "lea from reg?\n");
1530
1531 switch (popr->lmod) {
1532 case OPLM_DWORD:
1533 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
1534 break;
1535 case OPLM_WORD:
1536 snprintf(buf, buf_size, "%s%s",
1537 simplify_cast(cast, "(u16)"), opr_reg_p(po, popr));
1538 break;
1539 case OPLM_BYTE:
1540 if (popr->name[1] == 'h') // XXX..
1541 snprintf(buf, buf_size, "%s(%s >> 8)",
1542 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
1543 else
1544 snprintf(buf, buf_size, "%s%s",
1545 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
1546 break;
1547 default:
1548 ferr(po, "invalid src lmod: %d\n", popr->lmod);
1549 }
1550 break;
1551
1552 case OPT_REGMEM:
1553 if (is_stack_access(po, popr)) {
1554 stack_frame_access(po, popr, buf, buf_size,
1555 popr->name, cast, 1, is_lea);
1556 break;
1557 }
1558
1559 strcpy(expr, popr->name);
1560 if (strchr(expr, '[')) {
1561 // special case: '[' can only be left for label[reg] form
1562 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
1563 if (ret != 2)
1564 ferr(po, "parse failure for '%s'\n", expr);
1565 if (tmp1[0] == '(') {
1566 // (off_4FFF50+3)[eax]
1567 p = strchr(tmp1 + 1, ')');
1568 if (p == NULL || p[1] != 0)
1569 ferr(po, "parse failure (2) for '%s'\n", expr);
1570 *p = 0;
1571 memmove(tmp1, tmp1 + 1, strlen(tmp1));
1572 }
1573 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
1574 }
1575
1576 // XXX: do we need more parsing?
1577 if (is_lea) {
1578 snprintf(buf, buf_size, "%s", expr);
1579 break;
1580 }
1581
1582 snprintf(buf, buf_size, "%s(%s)",
1583 simplify_cast(cast, lmod_cast_u_ptr(po, popr->lmod)), expr);
1584 break;
1585
1586 case OPT_LABEL:
1587 check_label_read_ref(po, popr->name);
1588 if (cast[0] == 0 && popr->is_ptr)
1589 cast = "(u32)";
1590
1591 if (is_lea)
1592 snprintf(buf, buf_size, "(u32)&%s", popr->name);
1593 else if (popr->size_lt)
1594 snprintf(buf, buf_size, "%s%s%s%s", cast,
1595 lmod_cast_u_ptr(po, popr->lmod),
1596 popr->is_array ? "" : "&",
1597 popr->name);
1598 else
1599 snprintf(buf, buf_size, "%s%s%s", cast, popr->name,
1600 popr->is_array ? "[0]" : "");
1601 break;
1602
1603 case OPT_OFFSET:
1604 check_label_read_ref(po, popr->name);
1605 if (cast[0] == 0)
1606 cast = "(u32)";
1607 if (is_lea)
1608 ferr(po, "lea an offset?\n");
1609 snprintf(buf, buf_size, "%s&%s", cast, popr->name);
1610 break;
1611
1612 case OPT_CONST:
1613 if (is_lea)
1614 ferr(po, "lea from const?\n");
1615
1616 printf_number(tmp1, sizeof(tmp1), popr->val);
1617 if (popr->val == 0 && strchr(cast, '*'))
1618 snprintf(buf, buf_size, "NULL");
1619 else
1620 snprintf(buf, buf_size, "%s%s",
1621 simplify_cast_num(cast, popr->val), tmp1);
1622 break;
1623
1624 default:
1625 ferr(po, "invalid src type: %d\n", popr->type);
1626 }
1627
1628 return buf;
1629}
1630
1631// note: may set is_ptr (we find that out late for ebp frame..)
1632static char *out_dst_opr(char *buf, size_t buf_size,
1633 struct parsed_op *po, struct parsed_opr *popr)
1634{
1635 switch (popr->type) {
1636 case OPT_REG:
1637 switch (popr->lmod) {
1638 case OPLM_DWORD:
1639 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
1640 break;
1641 case OPLM_WORD:
1642 // ugh..
1643 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
1644 break;
1645 case OPLM_BYTE:
1646 // ugh..
1647 if (popr->name[1] == 'h') // XXX..
1648 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
1649 else
1650 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
1651 break;
1652 default:
1653 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
1654 }
1655 break;
1656
1657 case OPT_REGMEM:
1658 if (is_stack_access(po, popr)) {
1659 stack_frame_access(po, popr, buf, buf_size,
1660 popr->name, "", 0, 0);
1661 break;
1662 }
1663
1664 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1665
1666 case OPT_LABEL:
1667 if (popr->size_mismatch)
1668 snprintf(buf, buf_size, "%s%s%s",
1669 lmod_cast_u_ptr(po, popr->lmod),
1670 popr->is_array ? "" : "&", popr->name);
1671 else
1672 snprintf(buf, buf_size, "%s%s", popr->name,
1673 popr->is_array ? "[0]" : "");
1674 break;
1675
1676 default:
1677 ferr(po, "invalid dst type: %d\n", popr->type);
1678 }
1679
1680 return buf;
1681}
1682
1683static char *out_src_opr_u32(char *buf, size_t buf_size,
1684 struct parsed_op *po, struct parsed_opr *popr)
1685{
1686 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1687}
1688
1689static enum parsed_flag_op split_cond(struct parsed_op *po,
1690 enum op_op op, int *is_inv)
1691{
1692 *is_inv = 0;
1693
1694 switch (op) {
1695 case OP_JO:
1696 return PFO_O;
1697 case OP_JC:
1698 return PFO_C;
1699 case OP_JZ:
1700 return PFO_Z;
1701 case OP_JBE:
1702 return PFO_BE;
1703 case OP_JS:
1704 return PFO_S;
1705 case OP_JP:
1706 return PFO_P;
1707 case OP_JL:
1708 return PFO_L;
1709 case OP_JLE:
1710 return PFO_LE;
1711
1712 case OP_JNO:
1713 *is_inv = 1;
1714 return PFO_O;
1715 case OP_JNC:
1716 *is_inv = 1;
1717 return PFO_C;
1718 case OP_JNZ:
1719 *is_inv = 1;
1720 return PFO_Z;
1721 case OP_JA:
1722 *is_inv = 1;
1723 return PFO_BE;
1724 case OP_JNS:
1725 *is_inv = 1;
1726 return PFO_S;
1727 case OP_JNP:
1728 *is_inv = 1;
1729 return PFO_P;
1730 case OP_JGE:
1731 *is_inv = 1;
1732 return PFO_L;
1733 case OP_JG:
1734 *is_inv = 1;
1735 return PFO_LE;
1736
1737 case OP_ADC:
1738 case OP_SBB:
1739 return PFO_C;
1740
1741 default:
1742 ferr(po, "split_cond: bad op %d\n", op);
1743 return -1;
1744 }
1745}
1746
1747static void out_test_for_cc(char *buf, size_t buf_size,
1748 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
1749 enum opr_lenmod lmod, const char *expr)
1750{
1751 const char *cast, *scast;
1752
1753 cast = lmod_cast_u(po, lmod);
1754 scast = lmod_cast_s(po, lmod);
1755
1756 switch (pfo) {
1757 case PFO_Z:
1758 case PFO_BE: // CF=1||ZF=1; CF=0
1759 snprintf(buf, buf_size, "(%s%s %s 0)",
1760 cast, expr, is_inv ? "!=" : "==");
1761 break;
1762
1763 case PFO_S:
1764 case PFO_L: // SF!=OF; OF=0
1765 snprintf(buf, buf_size, "(%s%s %s 0)",
1766 scast, expr, is_inv ? ">=" : "<");
1767 break;
1768
1769 case PFO_LE: // ZF=1||SF!=OF; OF=0
1770 snprintf(buf, buf_size, "(%s%s %s 0)",
1771 scast, expr, is_inv ? ">" : "<=");
1772 break;
1773
1774 default:
1775 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1776 }
1777}
1778
1779static void out_cmp_for_cc(char *buf, size_t buf_size,
1780 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
1781{
1782 const char *cast, *scast, *cast_use;
1783 char buf1[256], buf2[256];
1784 enum opr_lenmod lmod;
1785
1786 if (po->operand[0].lmod != po->operand[1].lmod)
1787 ferr(po, "%s: lmod mismatch: %d %d\n", __func__,
1788 po->operand[0].lmod, po->operand[1].lmod);
1789 lmod = po->operand[0].lmod;
1790
1791 cast = lmod_cast_u(po, lmod);
1792 scast = lmod_cast_s(po, lmod);
1793
1794 switch (pfo) {
1795 case PFO_C:
1796 case PFO_Z:
1797 case PFO_BE: // !a
1798 cast_use = cast;
1799 break;
1800
1801 case PFO_S:
1802 case PFO_L: // !ge
1803 case PFO_LE:
1804 cast_use = scast;
1805 break;
1806
1807 default:
1808 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1809 }
1810
1811 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], cast_use, 0);
1812 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], cast_use, 0);
1813
1814 switch (pfo) {
1815 case PFO_C:
1816 // note: must be unsigned compare
1817 snprintf(buf, buf_size, "(%s %s %s)",
1818 buf1, is_inv ? ">=" : "<", buf2);
1819 break;
1820
1821 case PFO_Z:
1822 snprintf(buf, buf_size, "(%s %s %s)",
1823 buf1, is_inv ? "!=" : "==", buf2);
1824 break;
1825
1826 case PFO_BE: // !a
1827 // note: must be unsigned compare
1828 snprintf(buf, buf_size, "(%s %s %s)",
1829 buf1, is_inv ? ">" : "<=", buf2);
1830
1831 // annoying case
1832 if (is_inv && lmod == OPLM_BYTE
1833 && po->operand[1].type == OPT_CONST
1834 && po->operand[1].val == 0xff)
1835 {
1836 snprintf(g_comment, sizeof(g_comment), "if %s", buf);
1837 snprintf(buf, buf_size, "(0)");
1838 }
1839 break;
1840
1841 // note: must be signed compare
1842 case PFO_S:
1843 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
1844 scast, buf1, buf2, is_inv ? ">=" : "<");
1845 break;
1846
1847 case PFO_L: // !ge
1848 snprintf(buf, buf_size, "(%s %s %s)",
1849 buf1, is_inv ? ">=" : "<", buf2);
1850 break;
1851
1852 case PFO_LE:
1853 snprintf(buf, buf_size, "(%s %s %s)",
1854 buf1, is_inv ? ">" : "<=", buf2);
1855 break;
1856
1857 default:
1858 break;
1859 }
1860}
1861
1862static void out_cmp_test(char *buf, size_t buf_size,
1863 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
1864{
1865 char buf1[256], buf2[256], buf3[256];
1866
1867 if (po->op == OP_TEST) {
1868 if (IS(opr_name(po, 0), opr_name(po, 1))) {
1869 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
1870 }
1871 else {
1872 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
1873 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
1874 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
1875 }
1876 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
1877 po->operand[0].lmod, buf3);
1878 }
1879 else if (po->op == OP_CMP) {
1880 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv);
1881 }
1882 else
1883 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
1884}
1885
1886static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
1887 struct parsed_opr *popr2)
1888{
1889 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
1890 ferr(po, "missing lmod for both operands\n");
1891
1892 if (popr1->lmod == OPLM_UNSPEC)
1893 popr1->lmod = popr2->lmod;
1894 else if (popr2->lmod == OPLM_UNSPEC)
1895 popr2->lmod = popr1->lmod;
1896 else if (popr1->lmod != popr2->lmod) {
1897 if (popr1->type_from_var) {
1898 popr1->size_mismatch = 1;
1899 if (popr1->lmod < popr2->lmod)
1900 popr1->size_lt = 1;
1901 popr1->lmod = popr2->lmod;
1902 }
1903 else if (popr2->type_from_var) {
1904 popr2->size_mismatch = 1;
1905 if (popr2->lmod < popr1->lmod)
1906 popr2->size_lt = 1;
1907 popr2->lmod = popr1->lmod;
1908 }
1909 else
1910 ferr(po, "conflicting lmods: %d vs %d\n",
1911 popr1->lmod, popr2->lmod);
1912 }
1913}
1914
1915static const char *op_to_c(struct parsed_op *po)
1916{
1917 switch (po->op)
1918 {
1919 case OP_ADD:
1920 case OP_ADC:
1921 return "+";
1922 case OP_SUB:
1923 case OP_SBB:
1924 return "-";
1925 case OP_AND:
1926 return "&";
1927 case OP_OR:
1928 return "|";
1929 case OP_XOR:
1930 return "^";
1931 case OP_SHL:
1932 return "<<";
1933 case OP_SHR:
1934 return ">>";
1935 case OP_MUL:
1936 case OP_IMUL:
1937 return "*";
1938 default:
1939 ferr(po, "op_to_c was supplied with %d\n", po->op);
1940 }
1941}
1942
1943static void op_set_clear_flag(struct parsed_op *po,
1944 enum op_flags flag_set, enum op_flags flag_clear)
1945{
1946 po->flags |= flag_set;
1947 po->flags &= ~flag_clear;
1948}
1949
1950// last op in stream - unconditional branch or ret
1951#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
1952 || (ops[_i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP)
1953
1954static int scan_for_pop(int i, int opcnt, const char *reg,
1955 int magic, int depth, int *maxdepth, int do_flags)
1956{
1957 const struct parsed_proto *pp;
1958 struct parsed_op *po;
1959 int ret = 0;
1960 int j;
1961
1962 for (; i < opcnt; i++) {
1963 po = &ops[i];
1964 if (po->cc_scratch == magic)
1965 break; // already checked
1966 po->cc_scratch = magic;
1967
1968 if (po->flags & OPF_TAIL) {
1969 if (po->op == OP_CALL) {
1970 pp = proto_parse(g_fhdr, po->operand[0].name, 0);
1971 if (pp != NULL && pp->is_noreturn)
1972 // no stack cleanup for noreturn
1973 return ret;
1974 }
1975 return -1; // deadend
1976 }
1977
1978 if ((po->flags & OPF_RMD)
1979 || (po->op == OP_PUSH && po->argnum != 0)) // arg push
1980 continue;
1981
1982 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
1983 if (po->btj != NULL) {
1984 // jumptable
1985 for (j = 0; j < po->btj->count; j++) {
1986 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic,
1987 depth, maxdepth, do_flags);
1988 if (ret < 0)
1989 return ret; // dead end
1990 }
1991 return ret;
1992 }
1993
1994 if (po->bt_i < 0) {
1995 ferr(po, "dead branch\n");
1996 return -1;
1997 }
1998
1999 if (po->flags & OPF_CJMP) {
2000 ret |= scan_for_pop(po->bt_i, opcnt, reg, magic,
2001 depth, maxdepth, do_flags);
2002 if (ret < 0)
2003 return ret; // dead end
2004 }
2005 else {
2006 i = po->bt_i - 1;
2007 }
2008 continue;
2009 }
2010
2011 if ((po->op == OP_POP || po->op == OP_PUSH)
2012 && po->operand[0].type == OPT_REG
2013 && IS(po->operand[0].name, reg))
2014 {
2015 if (po->op == OP_PUSH && !(po->flags & OPF_FARG)) {
2016 depth++;
2017 if (depth > *maxdepth)
2018 *maxdepth = depth;
2019 if (do_flags)
2020 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2021 }
2022 else if (po->op == OP_POP) {
2023 if (depth == 0) {
2024 if (do_flags)
2025 op_set_clear_flag(po, OPF_RMD, OPF_RSAVE);
2026 return 1;
2027 }
2028 else {
2029 depth--;
2030 if (depth < 0) // should not happen
2031 ferr(po, "fail with depth\n");
2032 if (do_flags)
2033 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2034 }
2035 }
2036 }
2037 }
2038
2039 return ret;
2040}
2041
2042// scan for pop starting from 'ret' op (all paths)
2043static int scan_for_pop_ret(int i, int opcnt, const char *reg,
2044 int flag_set)
2045{
2046 int found = 0;
2047 int j;
2048
2049 for (; i < opcnt; i++) {
2050 if (!(ops[i].flags & OPF_TAIL))
2051 continue;
2052
2053 for (j = i - 1; j >= 0; j--) {
2054 if (ops[j].flags & OPF_RMD)
2055 continue;
2056 if (ops[j].flags & OPF_JMP)
2057 return -1;
2058
2059 if (ops[j].op == OP_POP && ops[j].operand[0].type == OPT_REG
2060 && IS(ops[j].operand[0].name, reg))
2061 {
2062 found = 1;
2063 ops[j].flags |= flag_set;
2064 break;
2065 }
2066
2067 if (g_labels[j][0] != 0)
2068 return -1;
2069 }
2070 }
2071
2072 return found ? 0 : -1;
2073}
2074
2075static void scan_propagate_df(int i, int opcnt)
2076{
2077 struct parsed_op *po = &ops[i];
2078 int j;
2079
2080 for (; i < opcnt; i++) {
2081 po = &ops[i];
2082 if (po->flags & OPF_DF)
2083 return; // already resolved
2084 po->flags |= OPF_DF;
2085
2086 if (po->op == OP_CALL)
2087 ferr(po, "call with DF set?\n");
2088
2089 if (po->flags & OPF_JMP) {
2090 if (po->btj != NULL) {
2091 // jumptable
2092 for (j = 0; j < po->btj->count; j++)
2093 scan_propagate_df(po->btj->d[j].bt_i, opcnt);
2094 return;
2095 }
2096
2097 if (po->bt_i < 0) {
2098 ferr(po, "dead branch\n");
2099 return;
2100 }
2101
2102 if (po->flags & OPF_CJMP)
2103 scan_propagate_df(po->bt_i, opcnt);
2104 else
2105 i = po->bt_i - 1;
2106 continue;
2107 }
2108
2109 if (po->flags & OPF_TAIL)
2110 break;
2111
2112 if (po->op == OP_CLD) {
2113 po->flags |= OPF_RMD;
2114 return;
2115 }
2116 }
2117
2118 ferr(po, "missing DF clear?\n");
2119}
2120
2121// is operand 'opr' modified by parsed_op 'po'?
2122static int is_opr_modified(const struct parsed_opr *opr,
2123 const struct parsed_op *po)
2124{
2125 int mask;
2126
2127 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2128 return 0;
2129
2130 if (opr->type == OPT_REG) {
2131 if (po->op == OP_CALL) {
2132 mask = (1 << xAX) | (1 << xCX) | (1 << xDX);
2133 if ((1 << opr->reg) & mask)
2134 return 1;
2135 else
2136 return 0;
2137 }
2138
2139 if (po->operand[0].type == OPT_REG) {
2140 if (po->regmask_dst & (1 << opr->reg))
2141 return 1;
2142 else
2143 return 0;
2144 }
2145 }
2146
2147 return IS(po->operand[0].name, opr->name);
2148}
2149
2150// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
2151static int is_any_opr_modified(const struct parsed_op *po_test,
2152 const struct parsed_op *po, int c_mode)
2153{
2154 int mask;
2155 int i;
2156
2157 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2158 return 0;
2159
2160 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2161 return 0;
2162
2163 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
2164 return 1;
2165
2166 // in reality, it can wreck any register, but in decompiled C
2167 // version it can only overwrite eax or edx:eax
2168 mask = (1 << xAX) | (1 << xDX);
2169 if (!c_mode)
2170 mask |= 1 << xCX;
2171
2172 if (po->op == OP_CALL
2173 && ((po_test->regmask_src | po_test->regmask_dst) & mask))
2174 return 1;
2175
2176 for (i = 0; i < po_test->operand_cnt; i++)
2177 if (IS(po_test->operand[i].name, po->operand[0].name))
2178 return 1;
2179
2180 return 0;
2181}
2182
2183// scan for any po_test operand modification in range given
2184static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt,
2185 int c_mode)
2186{
2187 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2188 return -1;
2189
2190 for (; i < opcnt; i++) {
2191 if (is_any_opr_modified(po_test, &ops[i], c_mode))
2192 return i;
2193 }
2194
2195 return -1;
2196}
2197
2198// scan for po_test operand[0] modification in range given
2199static int scan_for_mod_opr0(struct parsed_op *po_test,
2200 int i, int opcnt)
2201{
2202 for (; i < opcnt; i++) {
2203 if (is_opr_modified(&po_test->operand[0], &ops[i]))
2204 return i;
2205 }
2206
2207 return -1;
2208}
2209
2210static int scan_for_flag_set(int i, int magic, int *branched,
2211 int *setters, int *setter_cnt)
2212{
2213 struct label_ref *lr;
2214 int ret;
2215
2216 while (i >= 0) {
2217 if (ops[i].cc_scratch == magic) {
2218 ferr(&ops[i], "%s looped\n", __func__);
2219 return -1;
2220 }
2221 ops[i].cc_scratch = magic;
2222
2223 if (g_labels[i][0] != 0) {
2224 *branched = 1;
2225
2226 lr = &g_label_refs[i];
2227 for (; lr->next; lr = lr->next) {
2228 ret = scan_for_flag_set(lr->i, magic,
2229 branched, setters, setter_cnt);
2230 if (ret < 0)
2231 return ret;
2232 }
2233
2234 if (i > 0 && LAST_OP(i - 1)) {
2235 i = lr->i;
2236 continue;
2237 }
2238 ret = scan_for_flag_set(lr->i, magic,
2239 branched, setters, setter_cnt);
2240 if (ret < 0)
2241 return ret;
2242 }
2243 i--;
2244
2245 if (ops[i].flags & OPF_FLAGS) {
2246 setters[*setter_cnt] = i;
2247 (*setter_cnt)++;
2248 return 0;
2249 }
2250
2251 if ((ops[i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP)
2252 return -1;
2253 }
2254
2255 return -1;
2256}
2257
2258// scan back for cdq, if anything modifies edx, fail
2259static int scan_for_cdq_edx(int i)
2260{
2261 while (i >= 0) {
2262 if (g_labels[i][0] != 0) {
2263 if (g_label_refs[i].next != NULL)
2264 return -1;
2265 if (i > 0 && LAST_OP(i - 1)) {
2266 i = g_label_refs[i].i;
2267 continue;
2268 }
2269 return -1;
2270 }
2271 i--;
2272
2273 if (ops[i].op == OP_CDQ)
2274 return i;
2275
2276 if (ops[i].regmask_dst & (1 << xDX))
2277 return -1;
2278 }
2279
2280 return -1;
2281}
2282
2283static int scan_for_reg_clear(int i, int reg)
2284{
2285 while (i >= 0) {
2286 if (g_labels[i][0] != 0) {
2287 if (g_label_refs[i].next != NULL)
2288 return -1;
2289 if (i > 0 && LAST_OP(i - 1)) {
2290 i = g_label_refs[i].i;
2291 continue;
2292 }
2293 return -1;
2294 }
2295 i--;
2296
2297 if (ops[i].op == OP_XOR
2298 && ops[i].operand[0].lmod == OPLM_DWORD
2299 && ops[i].operand[0].reg == ops[i].operand[1].reg
2300 && ops[i].operand[0].reg == reg)
2301 return i;
2302
2303 if (ops[i].regmask_dst & (1 << reg))
2304 return -1;
2305 }
2306
2307 return -1;
2308}
2309
2310// scan for positive, constant esp adjust
2311static int scan_for_esp_adjust(int i, int opcnt, int *adj,
2312 int *multipath)
2313{
2314 const struct parsed_proto *pp;
2315 struct parsed_op *po;
2316 int first_pop = -1;
2317
2318 *adj = *multipath = 0;
2319
2320 for (; i < opcnt; i++) {
2321 po = &ops[i];
2322
2323 if (g_labels[i][0] != 0)
2324 *multipath = 1;
2325
2326 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
2327 if (po->operand[1].type != OPT_CONST)
2328 ferr(&ops[i], "non-const esp adjust?\n");
2329 *adj += po->operand[1].val;
2330 if (*adj & 3)
2331 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
2332 return i;
2333 }
2334 else if (po->op == OP_PUSH && !(po->flags & OPF_RMD)) {
2335 //if (first_pop == -1)
2336 // first_pop = -2; // none
2337 *adj -= lmod_bytes(po, po->operand[0].lmod);
2338 }
2339 else if (po->op == OP_POP && !(po->flags & OPF_RMD)) {
2340 // seems like msvc only uses 'pop ecx' for stack realignment..
2341 if (po->operand[0].type != OPT_REG || po->operand[0].reg != xCX)
2342 break;
2343 if (first_pop == -1 && *adj >= 0)
2344 first_pop = i;
2345 *adj += lmod_bytes(po, po->operand[0].lmod);
2346 }
2347 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
2348 if (po->op == OP_JMP && po->btj == NULL) {
2349 i = po->bt_i - 1;
2350 continue;
2351 }
2352 if (po->op != OP_CALL)
2353 break;
2354 if (po->operand[0].type != OPT_LABEL)
2355 break;
2356 pp = po->datap;
2357 if (pp != NULL && pp->is_stdcall)
2358 break;
2359 }
2360 }
2361
2362 if (first_pop >= 0) {
2363 // probably 'pop ecx' was used..
2364 return first_pop;
2365 }
2366
2367 return -1;
2368}
2369
2370static void scan_fwd_set_flags(int i, int opcnt, int magic, int flags)
2371{
2372 struct parsed_op *po;
2373 int j;
2374
2375 if (i < 0)
2376 ferr(ops, "%s: followed bad branch?\n", __func__);
2377
2378 for (; i < opcnt; i++) {
2379 po = &ops[i];
2380 if (po->cc_scratch == magic)
2381 return;
2382 po->cc_scratch = magic;
2383 po->flags |= flags;
2384
2385 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2386 if (po->btj != NULL) {
2387 // jumptable
2388 for (j = 0; j < po->btj->count; j++)
2389 scan_fwd_set_flags(po->btj->d[j].bt_i, opcnt, magic, flags);
2390 return;
2391 }
2392
2393 scan_fwd_set_flags(po->bt_i, opcnt, magic, flags);
2394 if (!(po->flags & OPF_CJMP))
2395 return;
2396 }
2397 if (po->flags & OPF_TAIL)
2398 return;
2399 }
2400}
2401
2402static const struct parsed_proto *try_recover_pp(
2403 struct parsed_op *po, const struct parsed_opr *opr, int *search_instead)
2404{
2405 const struct parsed_proto *pp = NULL;
2406 char buf[256];
2407 char *p;
2408
2409 // maybe an arg of g_func?
2410 if (opr->type == OPT_REGMEM && is_stack_access(po, opr))
2411 {
2412 char ofs_reg[16] = { 0, };
2413 int arg, arg_s, arg_i;
2414 int stack_ra = 0;
2415 int offset = 0;
2416
2417 parse_stack_access(po, opr->name, ofs_reg,
2418 &offset, &stack_ra, NULL);
2419 if (ofs_reg[0] != 0)
2420 ferr(po, "offset reg on arg access?\n");
2421 if (offset <= stack_ra) {
2422 // search who set the stack var instead
2423 if (search_instead != NULL)
2424 *search_instead = 1;
2425 return NULL;
2426 }
2427
2428 arg_i = (offset - stack_ra - 4) / 4;
2429 for (arg = arg_s = 0; arg < g_func_pp->argc; arg++) {
2430 if (g_func_pp->arg[arg].reg != NULL)
2431 continue;
2432 if (arg_s == arg_i)
2433 break;
2434 arg_s++;
2435 }
2436 if (arg == g_func_pp->argc)
2437 ferr(po, "stack arg %d not in prototype?\n", arg_i);
2438
2439 pp = g_func_pp->arg[arg].fptr;
2440 if (pp == NULL)
2441 ferr(po, "icall sa: arg%d is not a fptr?\n", arg + 1);
2442 check_func_pp(po, pp, "icall arg");
2443 }
2444 else if (opr->type == OPT_REGMEM && strchr(opr->name + 1, '[')) {
2445 // label[index]
2446 p = strchr(opr->name + 1, '[');
2447 memcpy(buf, opr->name, p - opr->name);
2448 buf[p - opr->name] = 0;
2449 pp = proto_parse(g_fhdr, buf, 0);
2450 }
2451 else if (opr->type == OPT_OFFSET || opr->type == OPT_LABEL) {
2452 pp = proto_parse(g_fhdr, opr->name, 0);
2453 if (pp == NULL)
2454 ferr(po, "proto_parse failed for icall from '%s'\n", opr->name);
2455 check_func_pp(po, pp, "reg-fptr ref");
2456 }
2457
2458 return pp;
2459}
2460
2461static void scan_for_call_type(int i, const struct parsed_opr *opr,
2462 int magic, const struct parsed_proto **pp_found, int *multi)
2463{
2464 const struct parsed_proto *pp = NULL;
2465 struct parsed_op *po;
2466 struct label_ref *lr;
2467
2468 while (i >= 0) {
2469 if (ops[i].cc_scratch == magic)
2470 return;
2471 ops[i].cc_scratch = magic;
2472
2473 if (g_labels[i][0] != 0) {
2474 lr = &g_label_refs[i];
2475 for (; lr != NULL; lr = lr->next)
2476 scan_for_call_type(lr->i, opr, magic, pp_found, multi);
2477 if (i > 0 && LAST_OP(i - 1))
2478 return;
2479 }
2480 i--;
2481
2482 if (!(ops[i].flags & OPF_DATA))
2483 continue;
2484 if (!is_opr_modified(opr, &ops[i]))
2485 continue;
2486 if (ops[i].op != OP_MOV && ops[i].op != OP_LEA) {
2487 // most probably trashed by some processing
2488 *pp_found = NULL;
2489 return;
2490 }
2491
2492 opr = &ops[i].operand[1];
2493 if (opr->type != OPT_REG)
2494 break;
2495 }
2496
2497 po = (i >= 0) ? &ops[i] : ops;
2498
2499 if (i < 0) {
2500 // reached the top - can only be an arg-reg
2501 if (opr->type != OPT_REG)
2502 return;
2503
2504 for (i = 0; i < g_func_pp->argc; i++) {
2505 if (g_func_pp->arg[i].reg == NULL)
2506 continue;
2507 if (IS(opr->name, g_func_pp->arg[i].reg))
2508 break;
2509 }
2510 if (i == g_func_pp->argc)
2511 return;
2512 pp = g_func_pp->arg[i].fptr;
2513 if (pp == NULL)
2514 ferr(po, "icall: arg%d (%s) is not a fptr?\n",
2515 i + 1, g_func_pp->arg[i].reg);
2516 check_func_pp(po, pp, "icall reg-arg");
2517 }
2518 else
2519 pp = try_recover_pp(po, opr, NULL);
2520
2521 if (*pp_found != NULL && pp != NULL && *pp_found != pp) {
2522 if (!IS((*pp_found)->ret_type.name, pp->ret_type.name)
2523 || (*pp_found)->is_stdcall != pp->is_stdcall
2524 || (*pp_found)->is_fptr != pp->is_fptr
2525 || (*pp_found)->argc != pp->argc
2526 || (*pp_found)->argc_reg != pp->argc_reg
2527 || (*pp_found)->argc_stack != pp->argc_stack)
2528 {
2529 ferr(po, "icall: parsed_proto mismatch\n");
2530 }
2531 *multi = 1;
2532 }
2533 if (pp != NULL)
2534 *pp_found = pp;
2535}
2536
2537static const struct parsed_proto *resolve_icall(int i, int opcnt,
2538 int *multi_src)
2539{
2540 const struct parsed_proto *pp = NULL;
2541 int search_advice = 0;
2542
2543 *multi_src = 0;
2544
2545 switch (ops[i].operand[0].type) {
2546 case OPT_REGMEM:
2547 case OPT_LABEL:
2548 case OPT_OFFSET:
2549 pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice);
2550 if (!search_advice)
2551 break;
2552 // fallthrough
2553 default:
2554 scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp,
2555 multi_src);
2556 break;
2557 }
2558
2559 return pp;
2560}
2561
2562static int collect_call_args_r(struct parsed_op *po, int i,
2563 struct parsed_proto *pp, int *regmask, int *save_arg_vars, int arg,
2564 int magic, int need_op_saving, int may_reuse)
2565{
2566 struct parsed_proto *pp_tmp;
2567 struct label_ref *lr;
2568 int need_to_save_current;
2569 int ret = 0;
2570 int j;
2571
2572 if (i < 0) {
2573 ferr(po, "dead label encountered\n");
2574 return -1;
2575 }
2576
2577 for (; arg < pp->argc; arg++)
2578 if (pp->arg[arg].reg == NULL)
2579 break;
2580 magic = (magic & 0xffffff) | (arg << 24);
2581
2582 for (j = i; j >= 0 && (arg < pp->argc || pp->is_unresolved); )
2583 {
2584 if (((ops[j].cc_scratch ^ magic) & 0xffffff) == 0) {
2585 if (ops[j].cc_scratch != magic) {
2586 ferr(&ops[j], "arg collect hit same path with diff args for %s\n",
2587 pp->name);
2588 return -1;
2589 }
2590 // ok: have already been here
2591 return 0;
2592 }
2593 ops[j].cc_scratch = magic;
2594
2595 if (g_labels[j][0] != 0 && g_label_refs[j].i != -1) {
2596 lr = &g_label_refs[j];
2597 if (lr->next != NULL)
2598 need_op_saving = 1;
2599 for (; lr->next; lr = lr->next) {
2600 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
2601 may_reuse = 1;
2602 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
2603 arg, magic, need_op_saving, may_reuse);
2604 if (ret < 0)
2605 return ret;
2606 }
2607
2608 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
2609 may_reuse = 1;
2610 if (j > 0 && LAST_OP(j - 1)) {
2611 // follow last branch in reverse
2612 j = lr->i;
2613 continue;
2614 }
2615 need_op_saving = 1;
2616 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
2617 arg, magic, need_op_saving, may_reuse);
2618 if (ret < 0)
2619 return ret;
2620 }
2621 j--;
2622
2623 if (ops[j].op == OP_CALL)
2624 {
2625 if (pp->is_unresolved)
2626 break;
2627
2628 pp_tmp = ops[j].datap;
2629 if (pp_tmp == NULL)
2630 ferr(po, "arg collect hit unparsed call '%s'\n",
2631 ops[j].operand[0].name);
2632 if (may_reuse && pp_tmp->argc_stack > 0)
2633 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
2634 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
2635 }
2636 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP) {
2637 if (pp->is_unresolved)
2638 break;
2639
2640 ferr(po, "arg collect %d/%d hit esp adjust\n",
2641 arg, pp->argc);
2642 }
2643 else if (ops[j].op == OP_POP) {
2644 if (pp->is_unresolved)
2645 break;
2646
2647 ferr(po, "arg collect %d/%d hit pop\n", arg, pp->argc);
2648 }
2649 else if (ops[j].flags & OPF_CJMP)
2650 {
2651 if (pp->is_unresolved)
2652 break;
2653
2654 may_reuse = 1;
2655 }
2656 else if (ops[j].op == OP_PUSH && !(ops[j].flags & OPF_FARG))
2657 {
2658 if (pp->is_unresolved && (ops[j].flags & OPF_RMD))
2659 break;
2660
2661 pp->arg[arg].datap = &ops[j];
2662 need_to_save_current = 0;
2663 if (!need_op_saving) {
2664 ret = scan_for_mod(&ops[j], j + 1, i, 1);
2665 need_to_save_current = (ret >= 0);
2666 }
2667 if (need_op_saving || need_to_save_current) {
2668 // mark this push as one that needs operand saving
2669 ops[j].flags &= ~OPF_RMD;
2670 if (ops[j].argnum == 0) {
2671 ops[j].argnum = arg + 1;
2672 *save_arg_vars |= 1 << arg;
2673 }
2674 else if (ops[j].argnum < arg + 1)
2675 ferr(&ops[j], "argnum conflict (%d<%d) for '%s'\n",
2676 ops[j].argnum, arg + 1, pp->name);
2677 }
2678 else if (ops[j].argnum == 0)
2679 ops[j].flags |= OPF_RMD;
2680
2681 // some PUSHes are reused by different calls on other branches,
2682 // but that can't happen if we didn't branch, so they
2683 // can be removed from future searches (handles nested calls)
2684 if (!may_reuse)
2685 ops[j].flags |= OPF_FARG;
2686
2687 ops[j].flags &= ~OPF_RSAVE;
2688
2689 arg++;
2690 if (!pp->is_unresolved) {
2691 // next arg
2692 for (; arg < pp->argc; arg++)
2693 if (pp->arg[arg].reg == NULL)
2694 break;
2695 }
2696 magic = (magic & 0xffffff) | (arg << 24);
2697
2698 // tracking reg usage
2699 if (ops[j].operand[0].type == OPT_REG)
2700 *regmask |= 1 << ops[j].operand[0].reg;
2701 }
2702 }
2703
2704 if (arg < pp->argc) {
2705 ferr(po, "arg collect failed for '%s': %d/%d\n",
2706 pp->name, arg, pp->argc);
2707 return -1;
2708 }
2709
2710 return arg;
2711}
2712
2713static int collect_call_args(struct parsed_op *po, int i,
2714 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
2715 int magic)
2716{
2717 int ret;
2718 int a;
2719
2720 ret = collect_call_args_r(po, i, pp, regmask, save_arg_vars,
2721 0, magic, 0, 0);
2722 if (ret < 0)
2723 return ret;
2724
2725 if (pp->is_unresolved) {
2726 pp->argc += ret;
2727 pp->argc_stack += ret;
2728 for (a = 0; a < pp->argc; a++)
2729 if (pp->arg[a].type.name == NULL)
2730 pp->arg[a].type.name = strdup("int");
2731 }
2732
2733 return ret;
2734}
2735
2736static void pp_insert_reg_arg(struct parsed_proto *pp, const char *reg)
2737{
2738 int i;
2739
2740 for (i = 0; i < pp->argc; i++)
2741 if (pp->arg[i].reg == NULL)
2742 break;
2743
2744 if (pp->argc_stack)
2745 memmove(&pp->arg[i + 1], &pp->arg[i],
2746 sizeof(pp->arg[0]) * pp->argc_stack);
2747 memset(&pp->arg[i], 0, sizeof(pp->arg[i]));
2748 pp->arg[i].reg = strdup(reg);
2749 pp->arg[i].type.name = strdup("int");
2750 pp->argc++;
2751 pp->argc_reg++;
2752}
2753
2754static void add_label_ref(struct label_ref *lr, int op_i)
2755{
2756 struct label_ref *lr_new;
2757
2758 if (lr->i == -1) {
2759 lr->i = op_i;
2760 return;
2761 }
2762
2763 lr_new = calloc(1, sizeof(*lr_new));
2764 lr_new->i = op_i;
2765 lr_new->next = lr->next;
2766 lr->next = lr_new;
2767}
2768
2769static void output_std_flags(FILE *fout, struct parsed_op *po,
2770 int *pfomask, const char *dst_opr_text)
2771{
2772 if (*pfomask & (1 << PFO_Z)) {
2773 fprintf(fout, "\n cond_z = (%s%s == 0);",
2774 lmod_cast_u(po, po->operand[0].lmod), dst_opr_text);
2775 *pfomask &= ~(1 << PFO_Z);
2776 }
2777 if (*pfomask & (1 << PFO_S)) {
2778 fprintf(fout, "\n cond_s = (%s%s < 0);",
2779 lmod_cast_s(po, po->operand[0].lmod), dst_opr_text);
2780 *pfomask &= ~(1 << PFO_S);
2781 }
2782}
2783
2784static void output_pp_attrs(FILE *fout, const struct parsed_proto *pp,
2785 int is_noreturn)
2786{
2787 if (pp->is_fastcall)
2788 fprintf(fout, "__fastcall ");
2789 else if (pp->is_stdcall && pp->argc_reg == 0)
2790 fprintf(fout, "__stdcall ");
2791 if (pp->is_noreturn || is_noreturn)
2792 fprintf(fout, "noreturn ");
2793}
2794
2795static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
2796{
2797 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
2798 struct parsed_opr *last_arith_dst = NULL;
2799 char buf1[256], buf2[256], buf3[256], cast[64];
2800 const struct parsed_proto *pp_c;
2801 struct parsed_proto *pp, *pp_tmp;
2802 struct parsed_data *pd;
2803 const char *tmpname;
2804 enum parsed_flag_op pfo;
2805 int save_arg_vars = 0;
2806 int cmp_result_vars = 0;
2807 int need_tmp_var = 0;
2808 int need_tmp64 = 0;
2809 int had_decl = 0;
2810 int label_pending = 0;
2811 int regmask_save = 0;
2812 int regmask_arg = 0;
2813 int regmask_now = 0;
2814 int regmask_init = 0;
2815 int regmask = 0;
2816 int pfomask = 0;
2817 int found = 0;
2818 int depth = 0;
2819 int no_output;
2820 int i, j, l;
2821 int dummy;
2822 int arg;
2823 int reg;
2824 int ret;
2825
2826 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
2827 g_stack_frame_used = 0;
2828
2829 g_func_pp = proto_parse(fhdr, funcn, 0);
2830 if (g_func_pp == NULL)
2831 ferr(ops, "proto_parse failed for '%s'\n", funcn);
2832
2833 fprintf(fout, "%s ", g_func_pp->ret_type.name);
2834 output_pp_attrs(fout, g_func_pp, g_ida_func_attr & IDAFA_NORETURN);
2835 fprintf(fout, "%s(", funcn);
2836
2837 for (i = 0; i < g_func_pp->argc; i++) {
2838 if (i > 0)
2839 fprintf(fout, ", ");
2840 if (g_func_pp->arg[i].fptr != NULL) {
2841 // func pointer..
2842 pp = g_func_pp->arg[i].fptr;
2843 fprintf(fout, "%s (", pp->ret_type.name);
2844 output_pp_attrs(fout, pp, 0);
2845 fprintf(fout, "*a%d)(", i + 1);
2846 for (j = 0; j < pp->argc; j++) {
2847 if (j > 0)
2848 fprintf(fout, ", ");
2849 if (pp->arg[j].fptr)
2850 ferr(ops, "nested fptr\n");
2851 fprintf(fout, "%s", pp->arg[j].type.name);
2852 }
2853 fprintf(fout, ")");
2854 }
2855 else {
2856 fprintf(fout, "%s a%d", g_func_pp->arg[i].type.name, i + 1);
2857 }
2858 if (g_func_pp->arg[i].reg != NULL) {
2859 reg = char_array_i(regs_r32,
2860 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
2861 if (reg < 0)
2862 ferr(ops, "arg '%s' is not a reg?\n", g_func_pp->arg[i].reg);
2863 regmask_arg |= 1 << reg;
2864 }
2865 }
2866 if (g_func_pp->is_vararg) {
2867 if (i > 0)
2868 fprintf(fout, ", ");
2869 fprintf(fout, "...");
2870 }
2871
2872 fprintf(fout, ")\n{\n");
2873
2874 // pass1:
2875 // - handle ebp/esp frame, remove ops related to it
2876 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
2877 && ops[1].op == OP_MOV
2878 && IS(opr_name(&ops[1], 0), "ebp")
2879 && IS(opr_name(&ops[1], 1), "esp"))
2880 {
2881 int ecx_push = 0;
2882
2883 g_bp_frame = 1;
2884 ops[0].flags |= OPF_RMD;
2885 ops[1].flags |= OPF_RMD;
2886 i = 2;
2887
2888 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
2889 g_stack_fsz = opr_const(&ops[2], 1);
2890 ops[2].flags |= OPF_RMD;
2891 i++;
2892 }
2893 else {
2894 // another way msvc builds stack frame..
2895 i = 2;
2896 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
2897 g_stack_fsz += 4;
2898 ops[i].flags |= OPF_RMD;
2899 ecx_push++;
2900 i++;
2901 }
2902 // and another way..
2903 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
2904 && ops[i].operand[1].type == OPT_CONST
2905 && ops[i + 1].op == OP_CALL
2906 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
2907 {
2908 g_stack_fsz += ops[i].operand[1].val;
2909 ops[i].flags |= OPF_RMD;
2910 i++;
2911 ops[i].flags |= OPF_RMD;
2912 i++;
2913 }
2914 }
2915
2916 found = 0;
2917 do {
2918 for (; i < opcnt; i++)
2919 if (ops[i].op == OP_RET)
2920 break;
2921 j = i - 1;
2922 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
2923 if (found) {
2924 if (ops[j].op == OP_JMP && !ops[j].operand[0].had_ds)
2925 // probably local branch back..
2926 break;
2927 if (ops[j].op == OP_CALL)
2928 // probably noreturn call..
2929 break;
2930 }
2931 j--;
2932 }
2933
2934 if ((ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ebp"))
2935 || ops[j].op == OP_LEAVE)
2936 {
2937 ops[j].flags |= OPF_RMD;
2938 }
2939 else if (!(g_ida_func_attr & IDAFA_NORETURN))
2940 ferr(&ops[j], "'pop ebp' expected\n");
2941
2942 if (g_stack_fsz != 0) {
2943 if (ops[j - 1].op == OP_MOV
2944 && IS(opr_name(&ops[j - 1], 0), "esp")
2945 && IS(opr_name(&ops[j - 1], 1), "ebp"))
2946 {
2947 ops[j - 1].flags |= OPF_RMD;
2948 }
2949 else if (ops[j].op != OP_LEAVE
2950 && !(g_ida_func_attr & IDAFA_NORETURN))
2951 {
2952 ferr(&ops[j - 1], "esp restore expected\n");
2953 }
2954
2955 if (ecx_push && ops[j - 2].op == OP_POP
2956 && IS(opr_name(&ops[j - 2], 0), "ecx"))
2957 {
2958 ferr(&ops[j - 2], "unexpected ecx pop\n");
2959 }
2960 }
2961
2962 found = 1;
2963 i++;
2964 } while (i < opcnt);
2965 }
2966 else {
2967 for (i = 0; i < opcnt; i++) {
2968 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
2969 break;
2970 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
2971 && ops[i].operand[1].type == OPT_CONST)
2972 {
2973 g_sp_frame = 1;
2974 break;
2975 }
2976 }
2977
2978 if (g_sp_frame)
2979 {
2980 g_stack_fsz = ops[i].operand[1].val;
2981 ops[i].flags |= OPF_RMD;
2982
2983 i++;
2984 do {
2985 for (; i < opcnt; i++)
2986 if (ops[i].op == OP_RET)
2987 break;
2988 if (ops[i - 1].op != OP_ADD
2989 || !IS(opr_name(&ops[i - 1], 0), "esp")
2990 || ops[i - 1].operand[1].type != OPT_CONST
2991 || ops[i - 1].operand[1].val != g_stack_fsz)
2992 ferr(&ops[i - 1], "'add esp' expected\n");
2993 ops[i - 1].flags |= OPF_RMD;
2994
2995 i++;
2996 } while (i < opcnt);
2997 }
2998 }
2999
3000 // pass2:
3001 // - parse calls with labels
3002 // - resolve all branches
3003 for (i = 0; i < opcnt; i++)
3004 {
3005 po = &ops[i];
3006 po->bt_i = -1;
3007 po->btj = NULL;
3008
3009 if (po->flags & OPF_RMD)
3010 continue;
3011
3012 if (po->op == OP_CALL) {
3013 pp = NULL;
3014
3015 if (po->operand[0].type == OPT_LABEL) {
3016 tmpname = opr_name(po, 0);
3017 if (IS_START(tmpname, "loc_"))
3018 ferr(po, "call to loc_*\n");
3019 pp_c = proto_parse(fhdr, tmpname, 0);
3020 if (pp_c == NULL)
3021 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
3022
3023 pp = proto_clone(pp_c);
3024 my_assert_not(pp, NULL);
3025 }
3026 else if (po->datap != NULL) {
3027 pp = calloc(1, sizeof(*pp));
3028 my_assert_not(pp, NULL);
3029
3030 ret = parse_protostr(po->datap, pp);
3031 if (ret < 0)
3032 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
3033 free(po->datap);
3034 po->datap = NULL;
3035 }
3036
3037 if (pp != NULL) {
3038 if (pp->is_fptr)
3039 check_func_pp(po, pp, "fptr var call");
3040 if (pp->is_noreturn)
3041 po->flags |= OPF_TAIL;
3042 po->datap = pp;
3043 }
3044 continue;
3045 }
3046
3047 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
3048 continue;
3049
3050 if (po->operand[0].type == OPT_REGMEM) {
3051 char *p = strchr(po->operand[0].name, '[');
3052 if (p == NULL)
3053 goto tailcall;
3054 ret = p - po->operand[0].name;
3055 strncpy(buf1, po->operand[0].name, ret);
3056 buf1[ret] = 0;
3057
3058 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
3059 if (IS(g_func_pd[j].label, buf1)) {
3060 pd = &g_func_pd[j];
3061 break;
3062 }
3063 }
3064 if (pd == NULL)
3065 //ferr(po, "label '%s' not parsed?\n", buf1);
3066 goto tailcall;
3067 if (pd->type != OPT_OFFSET)
3068 ferr(po, "label '%s' with non-offset data?\n", buf1);
3069
3070 // find all labels, link
3071 for (j = 0; j < pd->count; j++) {
3072 for (l = 0; l < opcnt; l++) {
3073 if (g_labels[l][0] && IS(g_labels[l], pd->d[j].u.label)) {
3074 add_label_ref(&g_label_refs[l], i);
3075 pd->d[j].bt_i = l;
3076 break;
3077 }
3078 }
3079 }
3080
3081 po->btj = pd;
3082 continue;
3083 }
3084
3085 for (l = 0; l < opcnt; l++) {
3086 if (g_labels[l][0] && IS(po->operand[0].name, g_labels[l])) {
3087 add_label_ref(&g_label_refs[l], i);
3088 po->bt_i = l;
3089 break;
3090 }
3091 }
3092
3093 if (po->bt_i != -1)
3094 continue;
3095
3096 if (po->operand[0].type == OPT_LABEL)
3097 // assume tail call
3098 goto tailcall;
3099
3100 ferr(po, "unhandled branch\n");
3101
3102tailcall:
3103 po->op = OP_CALL;
3104 po->flags |= OPF_TAIL;
3105 if (i > 0 && ops[i - 1].op == OP_POP)
3106 po->flags |= OPF_ATAIL;
3107 i--; // reprocess
3108 }
3109
3110 // pass3:
3111 // - remove dead labels
3112 // - process calls
3113 for (i = 0; i < opcnt; i++)
3114 {
3115 if (g_labels[i][0] != 0 && g_label_refs[i].i == -1)
3116 g_labels[i][0] = 0;
3117
3118 po = &ops[i];
3119 if (po->flags & OPF_RMD)
3120 continue;
3121
3122 if (po->op == OP_CALL)
3123 {
3124 tmpname = opr_name(po, 0);
3125 pp = po->datap;
3126 if (pp == NULL)
3127 {
3128 // indirect call
3129 pp_c = resolve_icall(i, opcnt, &l);
3130 if (pp_c != NULL) {
3131 pp = proto_clone(pp_c);
3132 my_assert_not(pp, NULL);
3133 if (l)
3134 // not resolved just to single func
3135 pp->is_fptr = 1;
3136 }
3137 if (pp == NULL) {
3138 pp = calloc(1, sizeof(*pp));
3139 my_assert_not(pp, NULL);
3140 pp->is_fptr = 1;
3141 ret = scan_for_esp_adjust(i + 1, opcnt, &j, &l);
3142 if (ret < 0) {
3143 if (!g_allow_regfunc)
3144 ferr(po, "non-__cdecl indirect call unhandled yet\n");
3145 pp->is_unresolved = 1;
3146 j = 0;
3147 }
3148 j /= 4;
3149 if (j > ARRAY_SIZE(pp->arg))
3150 ferr(po, "esp adjust too large: %d\n", j);
3151 pp->ret_type.name = strdup("int");
3152 pp->argc = pp->argc_stack = j;
3153 for (arg = 0; arg < pp->argc; arg++)
3154 pp->arg[arg].type.name = strdup("int");
3155 }
3156 po->datap = pp;
3157 }
3158
3159 // look for and make use of esp adjust
3160 ret = -1;
3161 if (!pp->is_stdcall && pp->argc_stack > 0)
3162 ret = scan_for_esp_adjust(i + 1, opcnt, &j, &l);
3163 if (ret >= 0) {
3164 if (pp->is_vararg) {
3165 if (j / 4 < pp->argc_stack)
3166 ferr(po, "esp adjust is too small: %x < %x\n",
3167 j, pp->argc_stack * 4);
3168 // modify pp to make it have varargs as normal args
3169 arg = pp->argc;
3170 pp->argc += j / 4 - pp->argc_stack;
3171 for (; arg < pp->argc; arg++) {
3172 pp->arg[arg].type.name = strdup("int");
3173 pp->argc_stack++;
3174 }
3175 if (pp->argc > ARRAY_SIZE(pp->arg))
3176 ferr(po, "too many args for '%s'\n", tmpname);
3177 }
3178 if (pp->argc_stack != j / 4)
3179 ferr(po, "stack tracking failed for '%s': %x %x\n",
3180 tmpname, pp->argc_stack * 4, j);
3181
3182 ops[ret].flags |= OPF_RMD;
3183 if (ops[ret].op == OP_POP && j > 4) {
3184 // deal with multi-pop stack adjust
3185 j = pp->argc_stack;
3186 while (ops[ret].op == OP_POP && j > 0 && ret < opcnt) {
3187 ops[ret].flags |= OPF_RMD;
3188 j--;
3189 ret++;
3190 }
3191 }
3192 else if (!l) {
3193 // a bit of a hack, but deals with use of
3194 // single adj for multiple calls
3195 ops[ret].operand[1].val -= j;
3196 }
3197 }
3198 else if (pp->is_vararg)
3199 ferr(po, "missing esp_adjust for vararg func '%s'\n",
3200 pp->name);
3201
3202 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
3203 // since we know the args, collect them
3204 collect_call_args(po, i, pp, &regmask, &save_arg_vars,
3205 i + opcnt * 2);
3206 }
3207
3208 if (strstr(pp->ret_type.name, "int64"))
3209 need_tmp64 = 1;
3210 }
3211 }
3212
3213 // pass4:
3214 // - find POPs for PUSHes, rm both
3215 // - scan for STD/CLD, propagate DF
3216 // - scan for all used registers
3217 // - find flag set ops for their users
3218 // - do unreselved calls
3219 // - declare indirect functions
3220 for (i = 0; i < opcnt; i++) {
3221 po = &ops[i];
3222 if (po->flags & OPF_RMD)
3223 continue;
3224
3225 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE)) {
3226 reg = po->operand[0].reg;
3227 if (!(regmask & (1 << reg)))
3228 // not a reg save after all, rerun scan_for_pop
3229 po->flags &= ~OPF_RSAVE;
3230 else
3231 regmask_save |= 1 << reg;
3232 }
3233
3234 if (po->op == OP_PUSH && po->argnum == 0
3235 && !(po->flags & OPF_RSAVE))
3236 {
3237 if (po->operand[0].type == OPT_REG)
3238 {
3239 reg = po->operand[0].reg;
3240 if (reg < 0)
3241 ferr(po, "reg not set for push?\n");
3242
3243 depth = 0;
3244 ret = scan_for_pop(i + 1, opcnt,
3245 po->operand[0].name, i + opcnt * 3, 0, &depth, 0);
3246 if (ret == 1) {
3247 if (depth > 1)
3248 ferr(po, "too much depth: %d\n", depth);
3249
3250 po->flags |= OPF_RMD;
3251 scan_for_pop(i + 1, opcnt, po->operand[0].name,
3252 i + opcnt * 4, 0, &depth, 1);
3253 continue;
3254 }
3255 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
3256 if (ret == 0) {
3257 arg = OPF_RMD;
3258 if (regmask & (1 << reg)) {
3259 if (regmask_save & (1 << reg))
3260 ferr(po, "%s already saved?\n", po->operand[0].name);
3261 arg = OPF_RSAVE;
3262 }
3263 po->flags |= arg;
3264 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg);
3265 continue;
3266 }
3267 }
3268 else if (po->operand[0].type == OPT_CONST) {
3269 for (j = i + 1; j < opcnt; j++) {
3270 if ((ops[j].flags & (OPF_JMP|OPF_TAIL|OPF_RSAVE))
3271 || ops[j].op == OP_PUSH || g_labels[i][0] != 0)
3272 {
3273 break;
3274 }
3275
3276 if (!(ops[j].flags & OPF_RMD) && ops[j].op == OP_POP)
3277 {
3278 po->flags |= OPF_RMD;
3279 ops[j].datap = po;
3280 break;
3281 }
3282 }
3283 }
3284 }
3285
3286 if (po->op == OP_STD) {
3287 po->flags |= OPF_DF | OPF_RMD;
3288 scan_propagate_df(i + 1, opcnt);
3289 }
3290
3291 regmask_now = po->regmask_src | po->regmask_dst;
3292 if (regmask_now & (1 << xBP)) {
3293 if (g_bp_frame && !(po->flags & OPF_EBP_S)) {
3294 if (po->regmask_dst & (1 << xBP))
3295 // compiler decided to drop bp frame and use ebp as scratch
3296 scan_fwd_set_flags(i, opcnt, i + opcnt * 5, OPF_EBP_S);
3297 else
3298 regmask_now &= ~(1 << xBP);
3299 }
3300 }
3301
3302 regmask |= regmask_now;
3303
3304 if (po->flags & OPF_CC)
3305 {
3306 int setters[16], cnt = 0, branched = 0;
3307
3308 ret = scan_for_flag_set(i, i + opcnt * 6,
3309 &branched, setters, &cnt);
3310 if (ret < 0 || cnt <= 0)
3311 ferr(po, "unable to trace flag setter(s)\n");
3312 if (cnt > ARRAY_SIZE(setters))
3313 ferr(po, "too many flag setters\n");
3314
3315 pfo = split_cond(po, po->op, &dummy);
3316 for (j = 0; j < cnt; j++)
3317 {
3318 tmp_op = &ops[setters[j]]; // flag setter
3319 pfomask = 0;
3320
3321 // to get nicer code, we try to delay test and cmp;
3322 // if we can't because of operand modification, or if we
3323 // have arith op, or branch, make it calculate flags explicitly
3324 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP)
3325 {
3326 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i, 0) >= 0)
3327 pfomask = 1 << pfo;
3328 }
3329 else if (tmp_op->op == OP_CMPS || tmp_op->op == OP_SCAS) {
3330 pfomask = 1 << pfo;
3331 }
3332 else {
3333 // see if we'll be able to handle based on op result
3334 if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR
3335 && pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P)
3336 || branched
3337 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
3338 pfomask = 1 << pfo;
3339
3340 if (tmp_op->op == OP_ADD && pfo == PFO_C)
3341 need_tmp64 = 1;
3342 }
3343 if (pfomask) {
3344 tmp_op->pfomask |= pfomask;
3345 cmp_result_vars |= pfomask;
3346 }
3347 // note: may overwrite, currently not a problem
3348 po->datap = tmp_op;
3349 }
3350
3351 if (po->op == OP_ADC || po->op == OP_SBB)
3352 cmp_result_vars |= 1 << PFO_C;
3353 }
3354 else if (po->op == OP_CMPS || po->op == OP_SCAS) {
3355 cmp_result_vars |= 1 << PFO_Z;
3356 }
3357 else if (po->op == OP_MUL
3358 || (po->op == OP_IMUL && po->operand_cnt == 1))
3359 {
3360 need_tmp64 = 1;
3361 }
3362 else if (po->op == OP_XCHG) {
3363 need_tmp_var = 1;
3364 }
3365 else if (po->op == OP_CALL) {
3366 pp = po->datap;
3367 if (pp == NULL)
3368 ferr(po, "NULL pp\n");
3369
3370 if (pp->is_unresolved) {
3371 int regmask_stack = 0;
3372 collect_call_args(po, i, pp, &regmask, &save_arg_vars,
3373 i + opcnt * 2);
3374
3375 // this is pretty rough guess:
3376 // see ecx and edx were pushed (and not their saved versions)
3377 for (arg = 0; arg < pp->argc; arg++) {
3378 if (pp->arg[arg].reg != NULL)
3379 continue;
3380
3381 tmp_op = pp->arg[arg].datap;
3382 if (tmp_op == NULL)
3383 ferr(po, "parsed_op missing for arg%d\n", arg);
3384 if (tmp_op->argnum == 0 && tmp_op->operand[0].type == OPT_REG)
3385 regmask_stack |= 1 << tmp_op->operand[0].reg;
3386 }
3387
3388 if (!((regmask_stack & (1 << xCX))
3389 && (regmask_stack & (1 << xDX))))
3390 {
3391 if (pp->argc_stack != 0
3392 || ((regmask | regmask_arg) & ((1 << xCX)|(1 << xDX))))
3393 {
3394 pp_insert_reg_arg(pp, "ecx");
3395 pp->is_fastcall = 1;
3396 regmask_init |= 1 << xCX;
3397 regmask |= 1 << xCX;
3398 }
3399 if (pp->argc_stack != 0
3400 || ((regmask | regmask_arg) & (1 << xDX)))
3401 {
3402 pp_insert_reg_arg(pp, "edx");
3403 regmask_init |= 1 << xDX;
3404 regmask |= 1 << xDX;
3405 }
3406 }
3407
3408 // note: __cdecl doesn't fall into is_unresolved category
3409 if (pp->argc_stack > 0)
3410 pp->is_stdcall = 1;
3411 }
3412
3413 for (arg = 0; arg < pp->argc; arg++) {
3414 if (pp->arg[arg].reg != NULL) {
3415 reg = char_array_i(regs_r32,
3416 ARRAY_SIZE(regs_r32), pp->arg[arg].reg);
3417 if (reg < 0)
3418 ferr(ops, "arg '%s' is not a reg?\n", pp->arg[arg].reg);
3419 if (!(regmask & (1 << reg))) {
3420 regmask_init |= 1 << reg;
3421 regmask |= 1 << reg;
3422 }
3423 }
3424 }
3425
3426 if (pp->is_fptr) {
3427 if (pp->name[0] != 0) {
3428 memmove(pp->name + 2, pp->name, strlen(pp->name) + 1);
3429 memcpy(pp->name, "i_", 2);
3430
3431 // might be declared already
3432 found = 0;
3433 for (j = 0; j < i; j++) {
3434 if (ops[j].op == OP_CALL && (pp_tmp = ops[j].datap)) {
3435 if (pp_tmp->is_fptr && IS(pp->name, pp_tmp->name)) {
3436 found = 1;
3437 break;
3438 }
3439 }
3440 }
3441 if (found)
3442 continue;
3443 }
3444 else
3445 snprintf(pp->name, sizeof(pp->name), "icall%d", i);
3446
3447 fprintf(fout, " %s (", pp->ret_type.name);
3448 output_pp_attrs(fout, pp, 0);
3449 fprintf(fout, "*%s)(", pp->name);
3450 for (j = 0; j < pp->argc; j++) {
3451 if (j > 0)
3452 fprintf(fout, ", ");
3453 fprintf(fout, "%s a%d", pp->arg[j].type.name, j + 1);
3454 }
3455 fprintf(fout, ");\n");
3456 }
3457 }
3458 else if (po->op == OP_RET && !IS(g_func_pp->ret_type.name, "void"))
3459 regmask |= 1 << xAX;
3460 }
3461
3462 // pass4:
3463 // - confirm regmask_save, it might have been reduced
3464 if (regmask_save != 0)
3465 {
3466 regmask_save = 0;
3467 for (i = 0; i < opcnt; i++) {
3468 po = &ops[i];
3469 if (po->flags & OPF_RMD)
3470 continue;
3471
3472 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE))
3473 regmask_save |= 1 << po->operand[0].reg;
3474 }
3475 }
3476
3477
3478 // output LUTs/jumptables
3479 for (i = 0; i < g_func_pd_cnt; i++) {
3480 pd = &g_func_pd[i];
3481 fprintf(fout, " static const ");
3482 if (pd->type == OPT_OFFSET) {
3483 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
3484
3485 for (j = 0; j < pd->count; j++) {
3486 if (j > 0)
3487 fprintf(fout, ", ");
3488 fprintf(fout, "&&%s", pd->d[j].u.label);
3489 }
3490 }
3491 else {
3492 fprintf(fout, "%s %s[] =\n { ",
3493 lmod_type_u(ops, pd->lmod), pd->label);
3494
3495 for (j = 0; j < pd->count; j++) {
3496 if (j > 0)
3497 fprintf(fout, ", ");
3498 fprintf(fout, "%u", pd->d[j].u.val);
3499 }
3500 }
3501 fprintf(fout, " };\n");
3502 }
3503
3504 // declare stack frame, va_arg
3505 if (g_stack_fsz)
3506 fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
3507 (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz);
3508
3509 if (g_func_pp->is_vararg)
3510 fprintf(fout, " va_list ap;\n");
3511
3512 // declare arg-registers
3513 for (i = 0; i < g_func_pp->argc; i++) {
3514 if (g_func_pp->arg[i].reg != NULL) {
3515 reg = char_array_i(regs_r32,
3516 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
3517 if (regmask & (1 << reg)) {
3518 fprintf(fout, " u32 %s = (u32)a%d;\n",
3519 g_func_pp->arg[i].reg, i + 1);
3520 }
3521 else
3522 fprintf(fout, " // %s = a%d; // unused\n",
3523 g_func_pp->arg[i].reg, i + 1);
3524 had_decl = 1;
3525 }
3526 }
3527
3528 regmask_now = regmask & ~regmask_arg;
3529 regmask_now &= ~(1 << xSP);
3530 if (regmask_now) {
3531 for (reg = 0; reg < 8; reg++) {
3532 if (regmask_now & (1 << reg)) {
3533 fprintf(fout, " u32 %s", regs_r32[reg]);
3534 if (regmask_init & (1 << reg))
3535 fprintf(fout, " = 0");
3536 fprintf(fout, ";\n");
3537 had_decl = 1;
3538 }
3539 }
3540 }
3541
3542 if (regmask_save) {
3543 for (reg = 0; reg < 8; reg++) {
3544 if (regmask_save & (1 << reg)) {
3545 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
3546 had_decl = 1;
3547 }
3548 }
3549 }
3550
3551 if (save_arg_vars) {
3552 for (reg = 0; reg < 32; reg++) {
3553 if (save_arg_vars & (1 << reg)) {
3554 fprintf(fout, " u32 s_a%d;\n", reg + 1);
3555 had_decl = 1;
3556 }
3557 }
3558 }
3559
3560 if (cmp_result_vars) {
3561 for (i = 0; i < 8; i++) {
3562 if (cmp_result_vars & (1 << i)) {
3563 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
3564 had_decl = 1;
3565 }
3566 }
3567 }
3568
3569 if (need_tmp_var) {
3570 fprintf(fout, " u32 tmp;\n");
3571 had_decl = 1;
3572 }
3573
3574 if (need_tmp64) {
3575 fprintf(fout, " u64 tmp64;\n");
3576 had_decl = 1;
3577 }
3578
3579 if (had_decl)
3580 fprintf(fout, "\n");
3581
3582 if (g_func_pp->is_vararg) {
3583 if (g_func_pp->argc_stack == 0)
3584 ferr(ops, "vararg func without stack args?\n");
3585 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
3586 }
3587
3588 // output ops
3589 for (i = 0; i < opcnt; i++)
3590 {
3591 if (g_labels[i][0] != 0) {
3592 fprintf(fout, "\n%s:\n", g_labels[i]);
3593 label_pending = 1;
3594
3595 delayed_flag_op = NULL;
3596 last_arith_dst = NULL;
3597 }
3598
3599 po = &ops[i];
3600 if (po->flags & OPF_RMD)
3601 continue;
3602
3603 no_output = 0;
3604
3605 #define assert_operand_cnt(n_) \
3606 if (po->operand_cnt != n_) \
3607 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
3608
3609 // conditional/flag using op?
3610 if (po->flags & OPF_CC)
3611 {
3612 int is_delayed = 0;
3613 int is_inv = 0;
3614
3615 pfo = split_cond(po, po->op, &is_inv);
3616 tmp_op = po->datap;
3617
3618 // we go through all this trouble to avoid using parsed_flag_op,
3619 // which makes generated code much nicer
3620 if (delayed_flag_op != NULL)
3621 {
3622 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op, pfo, is_inv);
3623 is_delayed = 1;
3624 }
3625 else if (last_arith_dst != NULL
3626 && (pfo == PFO_Z || pfo == PFO_S || pfo == PFO_P
3627 || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR))
3628 ))
3629 {
3630 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
3631 out_test_for_cc(buf1, sizeof(buf1), po, pfo, is_inv,
3632 last_arith_dst->lmod, buf3);
3633 is_delayed = 1;
3634 }
3635 else if (tmp_op != NULL) {
3636 // use preprocessed flag calc results
3637 if (!(tmp_op->pfomask & (1 << pfo)))
3638 ferr(po, "not prepared for pfo %d\n", pfo);
3639
3640 // note: is_inv was not yet applied
3641 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
3642 is_inv ? "!" : "", parsed_flag_op_names[pfo]);
3643 }
3644 else {
3645 ferr(po, "all methods of finding comparison failed\n");
3646 }
3647
3648 if (po->flags & OPF_JMP) {
3649 fprintf(fout, " if %s\n", buf1);
3650 }
3651 else if (po->op == OP_ADC || po->op == OP_SBB) {
3652 if (is_delayed)
3653 fprintf(fout, " cond_%s = %s;\n",
3654 parsed_flag_op_names[pfo], buf1);
3655 }
3656 else if (po->flags & OPF_DATA) { // SETcc
3657 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
3658 fprintf(fout, " %s = %s;", buf2, buf1);
3659 }
3660 else {
3661 ferr(po, "unhandled conditional op\n");
3662 }
3663 }
3664
3665 pfomask = po->pfomask;
3666
3667 if (po->flags & (OPF_REPZ|OPF_REPNZ)) {
3668 // we need initial flags for ecx=0 case..
3669 if (i > 0 && ops[i - 1].op == OP_XOR
3670 && IS(ops[i - 1].operand[0].name,
3671 ops[i - 1].operand[1].name))
3672 {
3673 fprintf(fout, " cond_z = ");
3674 if (pfomask & (1 << PFO_C))
3675 fprintf(fout, "cond_c = ");
3676 fprintf(fout, "0;\n");
3677 }
3678 else if (last_arith_dst != NULL) {
3679 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
3680 out_test_for_cc(buf1, sizeof(buf1), po, PFO_Z, 0,
3681 last_arith_dst->lmod, buf3);
3682 fprintf(fout, " cond_z = %s;\n", buf1);
3683 }
3684 else
3685 ferr(po, "missing initial ZF\n");
3686 }
3687
3688 switch (po->op)
3689 {
3690 case OP_MOV:
3691 assert_operand_cnt(2);
3692 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3693 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3694 fprintf(fout, " %s = %s;", buf1,
3695 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
3696 po->operand[0].is_ptr ? "(void *)" : "", 0));
3697 break;
3698
3699 case OP_LEA:
3700 assert_operand_cnt(2);
3701 po->operand[1].lmod = OPLM_DWORD; // always
3702 fprintf(fout, " %s = %s;",
3703 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3704 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
3705 NULL, 1));
3706 break;
3707
3708 case OP_MOVZX:
3709 assert_operand_cnt(2);
3710 fprintf(fout, " %s = %s;",
3711 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3712 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
3713 break;
3714
3715 case OP_MOVSX:
3716 assert_operand_cnt(2);
3717 switch (po->operand[1].lmod) {
3718 case OPLM_BYTE:
3719 strcpy(buf3, "(s8)");
3720 break;
3721 case OPLM_WORD:
3722 strcpy(buf3, "(s16)");
3723 break;
3724 default:
3725 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
3726 }
3727 fprintf(fout, " %s = %s;",
3728 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3729 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
3730 buf3, 0));
3731 break;
3732
3733 case OP_XCHG:
3734 assert_operand_cnt(2);
3735 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3736 fprintf(fout, " tmp = %s;",
3737 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], "", 0));
3738 fprintf(fout, " %s = %s;",
3739 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3740 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], "", 0));
3741 fprintf(fout, " %s = tmp;",
3742 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]));
3743 snprintf(g_comment, sizeof(g_comment), "xchg");
3744 break;
3745
3746 case OP_NOT:
3747 assert_operand_cnt(1);
3748 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3749 fprintf(fout, " %s = ~%s;", buf1, buf1);
3750 break;
3751
3752 case OP_CDQ:
3753 assert_operand_cnt(2);
3754 fprintf(fout, " %s = (s32)%s >> 31;",
3755 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3756 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
3757 strcpy(g_comment, "cdq");
3758 break;
3759
3760 case OP_STOS:
3761 assert_operand_cnt(3);
3762 if (po->flags & OPF_REP) {
3763 fprintf(fout, " for (; ecx != 0; ecx--, edi %c= %d)\n",
3764 (po->flags & OPF_DF) ? '-' : '+',
3765 lmod_bytes(po, po->operand[0].lmod));
3766 fprintf(fout, " %sedi = eax;",
3767 lmod_cast_u_ptr(po, po->operand[0].lmod));
3768 strcpy(g_comment, "rep stos");
3769 }
3770 else {
3771 fprintf(fout, " %sedi = eax; edi %c= %d;",
3772 lmod_cast_u_ptr(po, po->operand[0].lmod),
3773 (po->flags & OPF_DF) ? '-' : '+',
3774 lmod_bytes(po, po->operand[0].lmod));
3775 strcpy(g_comment, "stos");
3776 }
3777 break;
3778
3779 case OP_MOVS:
3780 assert_operand_cnt(3);
3781 j = lmod_bytes(po, po->operand[0].lmod);
3782 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
3783 l = (po->flags & OPF_DF) ? '-' : '+';
3784 if (po->flags & OPF_REP) {
3785 fprintf(fout,
3786 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n",
3787 l, j, l, j);
3788 fprintf(fout,
3789 " %sedi = %sesi;", buf1, buf1);
3790 strcpy(g_comment, "rep movs");
3791 }
3792 else {
3793 fprintf(fout, " %sedi = %sesi; edi %c= %d; esi %c= %d;",
3794 buf1, buf1, l, j, l, j);
3795 strcpy(g_comment, "movs");
3796 }
3797 break;
3798
3799 case OP_CMPS:
3800 // repe ~ repeat while ZF=1
3801 assert_operand_cnt(3);
3802 j = lmod_bytes(po, po->operand[0].lmod);
3803 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
3804 l = (po->flags & OPF_DF) ? '-' : '+';
3805 if (po->flags & OPF_REP) {
3806 fprintf(fout,
3807 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d) {\n",
3808 l, j, l, j);
3809 if (pfomask & (1 << PFO_C)) {
3810 // ugh..
3811 fprintf(fout,
3812 " cond_c = %sedi < %sesi;\n", buf1, buf1);
3813 pfomask &= ~(1 << PFO_C);
3814 }
3815 fprintf(fout,
3816 " if ((cond_z = (%sedi == %sesi)) %s 0)\n",
3817 buf1, buf1, (po->flags & OPF_REPZ) ? "==" : "!=");
3818 fprintf(fout,
3819 " break;\n"
3820 " }");
3821 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
3822 (po->flags & OPF_REPZ) ? "e" : "ne");
3823 }
3824 else {
3825 fprintf(fout,
3826 " cond_z = (%sedi = %sesi); edi %c= %d; esi %c= %d;",
3827 buf1, buf1, l, j, l, j);
3828 strcpy(g_comment, "cmps");
3829 }
3830 pfomask &= ~(1 << PFO_Z);
3831 last_arith_dst = NULL;
3832 delayed_flag_op = NULL;
3833 break;
3834
3835 case OP_SCAS:
3836 // only does ZF (for now)
3837 // repe ~ repeat while ZF=1
3838 assert_operand_cnt(3);
3839 j = lmod_bytes(po, po->operand[0].lmod);
3840 l = (po->flags & OPF_DF) ? '-' : '+';
3841 if (po->flags & OPF_REP) {
3842 fprintf(fout,
3843 " for (; ecx != 0; ecx--, edi %c= %d)\n", l, j);
3844 fprintf(fout,
3845 " if ((cond_z = (%seax == %sedi)) %s 0)\n",
3846 lmod_cast_u(po, po->operand[0].lmod),
3847 lmod_cast_u_ptr(po, po->operand[0].lmod),
3848 (po->flags & OPF_REPZ) ? "==" : "!=");
3849 fprintf(fout,
3850 " break;");
3851 snprintf(g_comment, sizeof(g_comment), "rep%s scas",
3852 (po->flags & OPF_REPZ) ? "e" : "ne");
3853 }
3854 else {
3855 fprintf(fout, " cond_z = (%seax = %sedi); edi %c= %d;",
3856 lmod_cast_u(po, po->operand[0].lmod),
3857 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
3858 strcpy(g_comment, "scas");
3859 }
3860 pfomask &= ~(1 << PFO_Z);
3861 last_arith_dst = NULL;
3862 delayed_flag_op = NULL;
3863 break;
3864
3865 // arithmetic w/flags
3866 case OP_AND:
3867 case OP_OR:
3868 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3869 // fallthrough
3870 dualop_arith:
3871 assert_operand_cnt(2);
3872 fprintf(fout, " %s %s= %s;",
3873 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3874 op_to_c(po),
3875 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
3876 output_std_flags(fout, po, &pfomask, buf1);
3877 last_arith_dst = &po->operand[0];
3878 delayed_flag_op = NULL;
3879 break;
3880
3881 case OP_SHL:
3882 case OP_SHR:
3883 assert_operand_cnt(2);
3884 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3885 if (pfomask & (1 << PFO_C)) {
3886 if (po->operand[1].type == OPT_CONST) {
3887 l = lmod_bytes(po, po->operand[0].lmod) * 8;
3888 j = po->operand[1].val;
3889 j %= l;
3890 if (j != 0) {
3891 if (po->op == OP_SHL)
3892 j = l - j;
3893 else
3894 j -= 1;
3895 fprintf(fout, " cond_c = (%s & 0x%02x) ? 1 : 0;\n",
3896 buf1, 1 << j);
3897 }
3898 else
3899 ferr(po, "zero shift?\n");
3900 }
3901 else
3902 ferr(po, "TODO\n");
3903 pfomask &= ~(1 << PFO_C);
3904 }
3905 fprintf(fout, " %s %s= %s;", buf1, op_to_c(po),
3906 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
3907 output_std_flags(fout, po, &pfomask, buf1);
3908 last_arith_dst = &po->operand[0];
3909 delayed_flag_op = NULL;
3910 break;
3911
3912 case OP_SAR:
3913 assert_operand_cnt(2);
3914 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3915 fprintf(fout, " %s = %s%s >> %s;", buf1,
3916 lmod_cast_s(po, po->operand[0].lmod), buf1,
3917 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
3918 output_std_flags(fout, po, &pfomask, buf1);
3919 last_arith_dst = &po->operand[0];
3920 delayed_flag_op = NULL;
3921 break;
3922
3923 case OP_ROL:
3924 case OP_ROR:
3925 assert_operand_cnt(2);
3926 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3927 if (po->operand[1].type == OPT_CONST) {
3928 j = po->operand[1].val;
3929 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
3930 fprintf(fout, po->op == OP_ROL ?
3931 " %s = (%s << %d) | (%s >> %d);" :
3932 " %s = (%s >> %d) | (%s << %d);",
3933 buf1, buf1, j, buf1,
3934 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
3935 }
3936 else
3937 ferr(po, "TODO\n");
3938 output_std_flags(fout, po, &pfomask, buf1);
3939 last_arith_dst = &po->operand[0];
3940 delayed_flag_op = NULL;
3941 break;
3942
3943 case OP_XOR:
3944 assert_operand_cnt(2);
3945 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3946 if (IS(opr_name(po, 0), opr_name(po, 1))) {
3947 // special case for XOR
3948 if (pfomask & (1 << PFO_BE)) { // weird, but it happens..
3949 fprintf(fout, " cond_be = 1;\n");
3950 pfomask &= ~(1 << PFO_BE);
3951 }
3952 fprintf(fout, " %s = 0;",
3953 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
3954 last_arith_dst = &po->operand[0];
3955 delayed_flag_op = NULL;
3956 break;
3957 }
3958 goto dualop_arith;
3959
3960 case OP_ADD:
3961 assert_operand_cnt(2);
3962 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3963 if (pfomask & (1 << PFO_C)) {
3964 fprintf(fout, " tmp64 = (u64)%s + %s;\n",
3965 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]),
3966 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
3967 fprintf(fout, " cond_c = tmp64 >> 32;\n");
3968 fprintf(fout, " %s = (u32)tmp64;",
3969 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
3970 strcat(g_comment, "add64");
3971 pfomask &= ~(1 << PFO_C);
3972 output_std_flags(fout, po, &pfomask, buf1);
3973 last_arith_dst = &po->operand[0];
3974 delayed_flag_op = NULL;
3975 break;
3976 }
3977 goto dualop_arith;
3978
3979 case OP_SUB:
3980 assert_operand_cnt(2);
3981 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3982 if (pfomask & (1 << PFO_C)) {
3983 fprintf(fout, " cond_c = %s < %s;\n",
3984 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]),
3985 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
3986 pfomask &= ~(1 << PFO_C);
3987 }
3988 goto dualop_arith;
3989
3990 case OP_ADC:
3991 case OP_SBB:
3992 assert_operand_cnt(2);
3993 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3994 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3995 if (po->op == OP_SBB
3996 && IS(po->operand[0].name, po->operand[1].name))
3997 {
3998 // avoid use of unitialized var
3999 fprintf(fout, " %s = -cond_c;", buf1);
4000 // carry remains what it was
4001 pfomask &= ~(1 << PFO_C);
4002 }
4003 else {
4004 fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po),
4005 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4006 }
4007 output_std_flags(fout, po, &pfomask, buf1);
4008 last_arith_dst = &po->operand[0];
4009 delayed_flag_op = NULL;
4010 break;
4011
4012 case OP_INC:
4013 case OP_DEC:
4014 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4015 if (po->operand[0].type == OPT_REG) {
4016 strcpy(buf2, po->op == OP_INC ? "++" : "--");
4017 fprintf(fout, " %s%s;", buf1, buf2);
4018 }
4019 else {
4020 strcpy(buf2, po->op == OP_INC ? "+" : "-");
4021 fprintf(fout, " %s %s= 1;", buf1, buf2);
4022 }
4023 output_std_flags(fout, po, &pfomask, buf1);
4024 last_arith_dst = &po->operand[0];
4025 delayed_flag_op = NULL;
4026 break;
4027
4028 case OP_NEG:
4029 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4030 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
4031 fprintf(fout, " %s = -%s%s;", buf1,
4032 lmod_cast_s(po, po->operand[0].lmod), buf2);
4033 last_arith_dst = &po->operand[0];
4034 delayed_flag_op = NULL;
4035 if (pfomask & (1 << PFO_C)) {
4036 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
4037 pfomask &= ~(1 << PFO_C);
4038 }
4039 break;
4040
4041 case OP_IMUL:
4042 if (po->operand_cnt == 2) {
4043 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4044 goto dualop_arith;
4045 }
4046 if (po->operand_cnt == 3)
4047 ferr(po, "TODO imul3\n");
4048 // fallthrough
4049 case OP_MUL:
4050 assert_operand_cnt(1);
4051 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
4052 fprintf(fout, " tmp64 = %seax * %s%s;\n", buf1, buf1,
4053 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
4054 fprintf(fout, " edx = tmp64 >> 32;\n");
4055 fprintf(fout, " eax = tmp64;");
4056 last_arith_dst = NULL;
4057 delayed_flag_op = NULL;
4058 break;
4059
4060 case OP_DIV:
4061 case OP_IDIV:
4062 assert_operand_cnt(1);
4063 if (po->operand[0].lmod != OPLM_DWORD)
4064 ferr(po, "unhandled lmod %d\n", po->operand[0].lmod);
4065
4066 // 32bit division is common, look for it
4067 if (po->op == OP_DIV)
4068 ret = scan_for_reg_clear(i, xDX);
4069 else
4070 ret = scan_for_cdq_edx(i);
4071 if (ret >= 0) {
4072 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4073 strcpy(buf2, lmod_cast(po, po->operand[0].lmod,
4074 po->op == OP_IDIV));
4075 fprintf(fout, " edx = %seax %% %s%s;\n", buf2, buf2, buf1);
4076 fprintf(fout, " eax = %seax / %s%s;", buf2, buf2, buf1);
4077 }
4078 else
4079 ferr(po, "TODO 64bit divident\n");
4080 last_arith_dst = NULL;
4081 delayed_flag_op = NULL;
4082 break;
4083
4084 case OP_TEST:
4085 case OP_CMP:
4086 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4087 if (pfomask != 0) {
4088 for (j = 0; j < 8; j++) {
4089 if (pfomask & (1 << j)) {
4090 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
4091 fprintf(fout, " cond_%s = %s;",
4092 parsed_flag_op_names[j], buf1);
4093 }
4094 }
4095 pfomask = 0;
4096 }
4097 else
4098 no_output = 1;
4099 last_arith_dst = NULL;
4100 delayed_flag_op = po;
4101 break;
4102
4103 // note: we reuse OP_Jcc for SETcc, only flags differ
4104 case OP_JO ... OP_JG:
4105 if (po->flags & OPF_JMP)
4106 fprintf(fout, " goto %s;", po->operand[0].name);
4107 // else SETcc - should already be handled
4108 break;
4109
4110 case OP_JECXZ:
4111 fprintf(fout, " if (ecx == 0)\n");
4112 fprintf(fout, " goto %s;", po->operand[0].name);
4113 strcat(g_comment, "jecxz");
4114 break;
4115
4116 case OP_JMP:
4117 assert_operand_cnt(1);
4118 last_arith_dst = NULL;
4119 delayed_flag_op = NULL;
4120
4121 if (po->operand[0].type == OPT_REGMEM) {
4122 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
4123 buf1, buf2);
4124 if (ret != 2)
4125 ferr(po, "parse failure for jmp '%s'\n",
4126 po->operand[0].name);
4127 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
4128 break;
4129 }
4130 else if (po->operand[0].type != OPT_LABEL)
4131 ferr(po, "unhandled jmp type\n");
4132
4133 fprintf(fout, " goto %s;", po->operand[0].name);
4134 break;
4135
4136 case OP_CALL:
4137 assert_operand_cnt(1);
4138 pp = po->datap;
4139 my_assert_not(pp, NULL);
4140
4141 if (pp->is_fptr)
4142 fprintf(fout, " %s = %s;\n", pp->name,
4143 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
4144 "(void *)", 0));
4145
4146 fprintf(fout, " ");
4147 if (strstr(pp->ret_type.name, "int64")) {
4148 if (po->flags & OPF_TAIL)
4149 ferr(po, "int64 and tail?\n");
4150 fprintf(fout, "tmp64 = ");
4151 }
4152 else if (!IS(pp->ret_type.name, "void")) {
4153 if (po->flags & OPF_TAIL) {
4154 if (!IS(g_func_pp->ret_type.name, "void")) {
4155 fprintf(fout, "return ");
4156 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
4157 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
4158 }
4159 }
4160 else if (regmask & (1 << xAX)) {
4161 fprintf(fout, "eax = ");
4162 if (pp->ret_type.is_ptr)
4163 fprintf(fout, "(u32)");
4164 }
4165 }
4166
4167 if (pp->name[0] == 0)
4168 ferr(po, "missing pp->name\n");
4169 fprintf(fout, "%s%s(", pp->name,
4170 pp->has_structarg ? "_sa" : "");
4171
4172 if (po->flags & OPF_ATAIL) {
4173 if (pp->argc_stack != g_func_pp->argc_stack
4174 || (pp->argc_stack > 0
4175 && pp->is_stdcall != g_func_pp->is_stdcall))
4176 ferr(po, "incompatible tailcall\n");
4177
4178 for (arg = j = 0; arg < pp->argc; arg++) {
4179 if (arg > 0)
4180 fprintf(fout, ", ");
4181
4182 cast[0] = 0;
4183 if (pp->arg[arg].type.is_ptr)
4184 snprintf(cast, sizeof(cast), "(%s)",
4185 pp->arg[arg].type.name);
4186
4187 if (pp->arg[arg].reg != NULL) {
4188 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
4189 continue;
4190 }
4191 // stack arg
4192 for (; j < g_func_pp->argc; j++)
4193 if (g_func_pp->arg[j].reg == NULL)
4194 break;
4195 fprintf(fout, "%sa%d", cast, j + 1);
4196 j++;
4197 }
4198 }
4199 else {
4200 for (arg = 0; arg < pp->argc; arg++) {
4201 if (arg > 0)
4202 fprintf(fout, ", ");
4203
4204 cast[0] = 0;
4205 if (pp->arg[arg].type.is_ptr)
4206 snprintf(cast, sizeof(cast), "(%s)",
4207 pp->arg[arg].type.name);
4208
4209 if (pp->arg[arg].reg != NULL) {
4210 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
4211 continue;
4212 }
4213
4214 // stack arg
4215 tmp_op = pp->arg[arg].datap;
4216 if (tmp_op == NULL)
4217 ferr(po, "parsed_op missing for arg%d\n", arg);
4218 if (tmp_op->argnum != 0) {
4219 fprintf(fout, "%ss_a%d", cast, tmp_op->argnum);
4220 }
4221 else {
4222 fprintf(fout, "%s",
4223 out_src_opr(buf1, sizeof(buf1),
4224 tmp_op, &tmp_op->operand[0], cast, 0));
4225 }
4226 }
4227 }
4228 fprintf(fout, ");");
4229
4230 if (strstr(pp->ret_type.name, "int64")) {
4231 fprintf(fout, "\n");
4232 fprintf(fout, " edx = tmp64 >> 32;\n");
4233 fprintf(fout, " eax = tmp64;");
4234 }
4235
4236 if (pp->is_unresolved) {
4237 snprintf(buf3, sizeof(buf3), " unresoved %dreg",
4238 pp->argc_reg);
4239 strcat(g_comment, buf3);
4240 }
4241
4242 if (po->flags & OPF_TAIL) {
4243 ret = 0;
4244 if (i == opcnt - 1 || pp->is_noreturn)
4245 ret = 0;
4246 else if (IS(pp->ret_type.name, "void"))
4247 ret = 1;
4248 else if (IS(g_func_pp->ret_type.name, "void"))
4249 ret = 1;
4250 // else already handled as 'return f()'
4251
4252 if (ret) {
4253 if (!IS(g_func_pp->ret_type.name, "void")) {
4254 ferr(po, "int func -> void func tailcall?\n");
4255 }
4256 else {
4257 fprintf(fout, "\n return;");
4258 strcat(g_comment, " ^ tailcall");
4259 }
4260 }
4261 else
4262 strcat(g_comment, " tailcall");
4263 }
4264 if (pp->is_noreturn)
4265 strcat(g_comment, " noreturn");
4266 if ((po->flags & OPF_ATAIL) && pp->argc_stack > 0)
4267 strcat(g_comment, " argframe");
4268 delayed_flag_op = NULL;
4269 last_arith_dst = NULL;
4270 break;
4271
4272 case OP_RET:
4273 if (g_func_pp->is_vararg)
4274 fprintf(fout, " va_end(ap);\n");
4275
4276 if (IS(g_func_pp->ret_type.name, "void")) {
4277 if (i != opcnt - 1 || label_pending)
4278 fprintf(fout, " return;");
4279 }
4280 else if (g_func_pp->ret_type.is_ptr) {
4281 fprintf(fout, " return (%s)eax;",
4282 g_func_pp->ret_type.name);
4283 }
4284 else if (IS(g_func_pp->ret_type.name, "__int64"))
4285 fprintf(fout, " return ((u64)edx << 32) | eax;");
4286 else
4287 fprintf(fout, " return eax;");
4288
4289 last_arith_dst = NULL;
4290 delayed_flag_op = NULL;
4291 break;
4292
4293 case OP_PUSH:
4294 if (po->argnum != 0) {
4295 // special case - saved func arg
4296 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4297 fprintf(fout, " s_a%d = %s;", po->argnum, buf1);
4298 break;
4299 }
4300 else if (po->flags & OPF_RSAVE) {
4301 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4302 fprintf(fout, " s_%s = %s;", buf1, buf1);
4303 break;
4304 }
4305 if (!(g_ida_func_attr & IDAFA_NORETURN))
4306 ferr(po, "stray push encountered\n");
4307 no_output = 1;
4308 break;
4309
4310 case OP_POP:
4311 if (po->flags & OPF_RSAVE) {
4312 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4313 fprintf(fout, " %s = s_%s;", buf1, buf1);
4314 break;
4315 }
4316 else if (po->datap != NULL) {
4317 // push/pop pair
4318 tmp_op = po->datap;
4319 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4320 fprintf(fout, " %s = %s;", buf1,
4321 out_src_opr(buf2, sizeof(buf2),
4322 tmp_op, &tmp_op->operand[0],
4323 po->operand[0].is_ptr ? "(void *)" : "", 0));
4324 break;
4325 }
4326 ferr(po, "stray pop encountered\n");
4327 break;
4328
4329 case OP_NOP:
4330 no_output = 1;
4331 break;
4332
4333 default:
4334 no_output = 1;
4335 ferr(po, "unhandled op type %d, flags %x\n",
4336 po->op, po->flags);
4337 break;
4338 }
4339
4340 if (g_comment[0] != 0) {
4341 char *p = g_comment;
4342 while (my_isblank(*p))
4343 p++;
4344 fprintf(fout, " // %s", p);
4345 g_comment[0] = 0;
4346 no_output = 0;
4347 }
4348 if (!no_output)
4349 fprintf(fout, "\n");
4350
4351 // some sanity checking
4352 if (po->flags & OPF_REP) {
4353 if (po->op != OP_STOS && po->op != OP_MOVS
4354 && po->op != OP_CMPS && po->op != OP_SCAS)
4355 ferr(po, "unexpected rep\n");
4356 if (!(po->flags & (OPF_REPZ|OPF_REPNZ))
4357 && (po->op == OP_CMPS || po->op == OP_SCAS))
4358 ferr(po, "cmps/scas with plain rep\n");
4359 }
4360 if ((po->flags & (OPF_REPZ|OPF_REPNZ))
4361 && po->op != OP_CMPS && po->op != OP_SCAS)
4362 ferr(po, "unexpected repz/repnz\n");
4363
4364 if (pfomask != 0)
4365 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
4366
4367 // see is delayed flag stuff is still valid
4368 if (delayed_flag_op != NULL && delayed_flag_op != po) {
4369 if (is_any_opr_modified(delayed_flag_op, po, 0))
4370 delayed_flag_op = NULL;
4371 }
4372
4373 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
4374 if (is_opr_modified(last_arith_dst, po))
4375 last_arith_dst = NULL;
4376 }
4377
4378 label_pending = 0;
4379 }
4380
4381 if (g_stack_fsz && !g_stack_frame_used)
4382 fprintf(fout, " (void)sf;\n");
4383
4384 fprintf(fout, "}\n\n");
4385
4386 // cleanup
4387 for (i = 0; i < opcnt; i++) {
4388 struct label_ref *lr, *lr_del;
4389
4390 lr = g_label_refs[i].next;
4391 while (lr != NULL) {
4392 lr_del = lr;
4393 lr = lr->next;
4394 free(lr_del);
4395 }
4396 g_label_refs[i].i = -1;
4397 g_label_refs[i].next = NULL;
4398
4399 if (ops[i].op == OP_CALL) {
4400 pp = ops[i].datap;
4401 if (pp)
4402 proto_release(pp);
4403 }
4404 }
4405 g_func_pp = NULL;
4406}
4407
4408static void set_label(int i, const char *name)
4409{
4410 const char *p;
4411 int len;
4412
4413 len = strlen(name);
4414 p = strchr(name, ':');
4415 if (p != NULL)
4416 len = p - name;
4417
4418 if (len > sizeof(g_labels[0]) - 1)
4419 aerr("label '%s' too long: %d\n", name, len);
4420 if (g_labels[i][0] != 0 && !IS_START(g_labels[i], "algn_"))
4421 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
4422 memcpy(g_labels[i], name, len);
4423 g_labels[i][len] = 0;
4424}
4425
4426// '=' needs special treatment..
4427static char *next_word_s(char *w, size_t wsize, char *s)
4428{
4429 size_t i;
4430
4431 s = sskip(s);
4432
4433 for (i = 0; i < wsize - 1; i++) {
4434 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
4435 break;
4436 w[i] = s[i];
4437 }
4438 w[i] = 0;
4439
4440 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
4441 printf("warning: '%s' truncated\n", w);
4442
4443 return s + i;
4444}
4445
4446struct chunk_item {
4447 char *name;
4448 long fptr;
4449 int asmln;
4450};
4451
4452static struct chunk_item *func_chunks;
4453static int func_chunk_cnt;
4454static int func_chunk_alloc;
4455
4456static void add_func_chunk(FILE *fasm, const char *name, int line)
4457{
4458 if (func_chunk_cnt >= func_chunk_alloc) {
4459 func_chunk_alloc *= 2;
4460 func_chunks = realloc(func_chunks,
4461 func_chunk_alloc * sizeof(func_chunks[0]));
4462 my_assert_not(func_chunks, NULL);
4463 }
4464 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
4465 func_chunks[func_chunk_cnt].name = strdup(name);
4466 func_chunks[func_chunk_cnt].asmln = line;
4467 func_chunk_cnt++;
4468}
4469
4470static int cmp_chunks(const void *p1, const void *p2)
4471{
4472 const struct chunk_item *c1 = p1, *c2 = p2;
4473 return strcmp(c1->name, c2->name);
4474}
4475
4476static int cmpstringp(const void *p1, const void *p2)
4477{
4478 return strcmp(*(char * const *)p1, *(char * const *)p2);
4479}
4480
4481static void scan_ahead(FILE *fasm)
4482{
4483 char words[2][256];
4484 char line[256];
4485 long oldpos;
4486 int oldasmln;
4487 int wordc;
4488 char *p;
4489 int i;
4490
4491 oldpos = ftell(fasm);
4492 oldasmln = asmln;
4493
4494 while (fgets(line, sizeof(line), fasm))
4495 {
4496 wordc = 0;
4497 asmln++;
4498
4499 p = sskip(line);
4500 if (*p == 0)
4501 continue;
4502
4503 if (*p == ';')
4504 {
4505 // get rid of random tabs
4506 for (i = 0; line[i] != 0; i++)
4507 if (line[i] == '\t')
4508 line[i] = ' ';
4509
4510 if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
4511 {
4512 p += 30;
4513 next_word(words[0], sizeof(words[0]), p);
4514 if (words[0][0] == 0)
4515 aerr("missing name for func chunk?\n");
4516
4517 add_func_chunk(fasm, words[0], asmln);
4518 }
4519 continue;
4520 } // *p == ';'
4521
4522 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
4523 words[wordc][0] = 0;
4524 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
4525 if (*p == 0 || *p == ';') {
4526 wordc++;
4527 break;
4528 }
4529 }
4530
4531 if (wordc == 2 && IS(words[1], "ends"))
4532 break;
4533 }
4534
4535 fseek(fasm, oldpos, SEEK_SET);
4536 asmln = oldasmln;
4537}
4538
4539int main(int argc, char *argv[])
4540{
4541 FILE *fout, *fasm, *frlist;
4542 struct parsed_data *pd = NULL;
4543 int pd_alloc = 0;
4544 char **rlist = NULL;
4545 int rlist_len = 0;
4546 int rlist_alloc = 0;
4547 int func_chunks_used = 0;
4548 int func_chunks_sorted = 0;
4549 int func_chunk_i = -1;
4550 long func_chunk_ret = 0;
4551 int func_chunk_ret_ln = 0;
4552 int scanned_ahead = 0;
4553 char line[256];
4554 char words[20][256];
4555 enum opr_lenmod lmod;
4556 char *sctproto = NULL;
4557 int in_func = 0;
4558 int pending_endp = 0;
4559 int skip_func = 0;
4560 int skip_warned = 0;
4561 int eq_alloc;
4562 int verbose = 0;
4563 int arg_out;
4564 int arg;
4565 int pi = 0;
4566 int i, j;
4567 int ret, len;
4568 char *p;
4569 int wordc;
4570
4571 for (arg = 1; arg < argc; arg++) {
4572 if (IS(argv[arg], "-v"))
4573 verbose = 1;
4574 else if (IS(argv[arg], "-rf"))
4575 g_allow_regfunc = 1;
4576 else
4577 break;
4578 }
4579
4580 if (argc < arg + 3) {
4581 printf("usage:\n%s [-v] [-rf] <.c> <.asm> <hdrf> [rlist]*\n",
4582 argv[0]);
4583 return 1;
4584 }
4585
4586 arg_out = arg++;
4587
4588 asmfn = argv[arg++];
4589 fasm = fopen(asmfn, "r");
4590 my_assert_not(fasm, NULL);
4591
4592 hdrfn = argv[arg++];
4593 g_fhdr = fopen(hdrfn, "r");
4594 my_assert_not(g_fhdr, NULL);
4595
4596 rlist_alloc = 64;
4597 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
4598 my_assert_not(rlist, NULL);
4599 // needs special handling..
4600 rlist[rlist_len++] = "__alloca_probe";
4601
4602 func_chunk_alloc = 32;
4603 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
4604 my_assert_not(func_chunks, NULL);
4605
4606 memset(words, 0, sizeof(words));
4607
4608 for (; arg < argc; arg++) {
4609 frlist = fopen(argv[arg], "r");
4610 my_assert_not(frlist, NULL);
4611
4612 while (fgets(line, sizeof(line), frlist)) {
4613 p = sskip(line);
4614 if (*p == 0 || *p == ';')
4615 continue;
4616 if (*p == '#') {
4617 if (IS_START(p, "#if 0")
4618 || (g_allow_regfunc && IS_START(p, "#if NO_REGFUNC")))
4619 {
4620 skip_func = 1;
4621 }
4622 else if (IS_START(p, "#endif"))
4623 skip_func = 0;
4624 continue;
4625 }
4626 if (skip_func)
4627 continue;
4628
4629 p = next_word(words[0], sizeof(words[0]), p);
4630 if (words[0][0] == 0)
4631 continue;
4632
4633 if (rlist_len >= rlist_alloc) {
4634 rlist_alloc = rlist_alloc * 2 + 64;
4635 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
4636 my_assert_not(rlist, NULL);
4637 }
4638 rlist[rlist_len++] = strdup(words[0]);
4639 }
4640 skip_func = 0;
4641
4642 fclose(frlist);
4643 frlist = NULL;
4644 }
4645
4646 if (rlist_len > 0)
4647 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
4648
4649 fout = fopen(argv[arg_out], "w");
4650 my_assert_not(fout, NULL);
4651
4652 eq_alloc = 128;
4653 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
4654 my_assert_not(g_eqs, NULL);
4655
4656 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
4657 g_label_refs[i].i = -1;
4658 g_label_refs[i].next = NULL;
4659 }
4660
4661 while (fgets(line, sizeof(line), fasm))
4662 {
4663 wordc = 0;
4664 asmln++;
4665
4666 p = sskip(line);
4667 if (*p == 0)
4668 continue;
4669
4670 // get rid of random tabs
4671 for (i = 0; line[i] != 0; i++)
4672 if (line[i] == '\t')
4673 line[i] = ' ';
4674
4675 if (*p == ';')
4676 {
4677 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
4678 goto do_pending_endp; // eww..
4679
4680 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
4681 {
4682 static const char *attrs[] = {
4683 "bp-based frame",
4684 "library function",
4685 "static",
4686 "noreturn",
4687 "thunk",
4688 "fpd=",
4689 };
4690
4691 // parse IDA's attribute-list comment
4692 g_ida_func_attr = 0;
4693 p = sskip(p + 13);
4694
4695 for (; *p != 0; p = sskip(p)) {
4696 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
4697 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
4698 g_ida_func_attr |= 1 << i;
4699 p += strlen(attrs[i]);
4700 break;
4701 }
4702 }
4703 if (i == ARRAY_SIZE(attrs)) {
4704 anote("unparsed IDA attr: %s\n", p);
4705 break;
4706 }
4707 if (IS(attrs[i], "fpd=")) {
4708 p = next_word(words[0], sizeof(words[0]), p);
4709 // ignore for now..
4710 }
4711 }
4712 }
4713 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
4714 {
4715 p += 30;
4716 next_word(words[0], sizeof(words[0]), p);
4717 if (words[0][0] == 0)
4718 aerr("missing name for func chunk?\n");
4719
4720 if (!scanned_ahead) {
4721 add_func_chunk(fasm, words[0], asmln);
4722 func_chunks_sorted = 0;
4723 }
4724 }
4725 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
4726 {
4727 if (func_chunk_i >= 0) {
4728 if (func_chunk_i < func_chunk_cnt
4729 && IS(func_chunks[func_chunk_i].name, g_func))
4730 {
4731 // move on to next chunk
4732 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
4733 if (ret)
4734 aerr("seek failed for '%s' chunk #%d\n",
4735 g_func, func_chunk_i);
4736 asmln = func_chunks[func_chunk_i].asmln;
4737 func_chunk_i++;
4738 }
4739 else {
4740 if (func_chunk_ret == 0)
4741 aerr("no return from chunk?\n");
4742 fseek(fasm, func_chunk_ret, SEEK_SET);
4743 asmln = func_chunk_ret_ln;
4744 func_chunk_ret = 0;
4745 pending_endp = 1;
4746 }
4747 }
4748 }
4749 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
4750 func_chunks_used = 1;
4751 p += 20;
4752 if (IS_START(g_func, "sub_")) {
4753 unsigned long addr = strtoul(p, NULL, 16);
4754 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
4755 if (addr > f_addr && !scanned_ahead) {
4756 anote("scan_ahead caused by '%s', addr %lx\n",
4757 g_func, addr);
4758 scan_ahead(fasm);
4759 scanned_ahead = 1;
4760 func_chunks_sorted = 0;
4761 }
4762 }
4763 }
4764 continue;
4765 } // *p == ';'
4766
4767parse_words:
4768 for (i = wordc; i < ARRAY_SIZE(words); i++)
4769 words[i][0] = 0;
4770 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
4771 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
4772 if (*p == 0 || *p == ';') {
4773 wordc++;
4774 break;
4775 }
4776 }
4777 if (*p != 0 && *p != ';')
4778 aerr("too many words\n");
4779
4780 // alow asm patches in comments
4781 if (*p == ';') {
4782 if (IS_START(p, "; sctpatch:")) {
4783 p = sskip(p + 11);
4784 if (*p == 0 || *p == ';')
4785 continue;
4786 goto parse_words; // lame
4787 }
4788 if (IS_START(p, "; sctproto:")) {
4789 sctproto = strdup(p + 11);
4790 }
4791 }
4792
4793 if (wordc == 0) {
4794 // shouldn't happen
4795 awarn("wordc == 0?\n");
4796 continue;
4797 }
4798
4799 // don't care about this:
4800 if (words[0][0] == '.'
4801 || IS(words[0], "include")
4802 || IS(words[0], "assume") || IS(words[1], "segment")
4803 || IS(words[0], "align"))
4804 {
4805 continue;
4806 }
4807
4808do_pending_endp:
4809 // do delayed endp processing to collect switch jumptables
4810 if (pending_endp) {
4811 if (in_func && !skip_func && wordc >= 2
4812 && ((words[0][0] == 'd' && words[0][2] == 0)
4813 || (words[1][0] == 'd' && words[1][2] == 0)))
4814 {
4815 i = 1;
4816 if (words[1][0] == 'd' && words[1][2] == 0) {
4817 // label
4818 if (g_func_pd_cnt >= pd_alloc) {
4819 pd_alloc = pd_alloc * 2 + 16;
4820 g_func_pd = realloc(g_func_pd,
4821 sizeof(g_func_pd[0]) * pd_alloc);
4822 my_assert_not(g_func_pd, NULL);
4823 }
4824 pd = &g_func_pd[g_func_pd_cnt];
4825 g_func_pd_cnt++;
4826 memset(pd, 0, sizeof(*pd));
4827 strcpy(pd->label, words[0]);
4828 pd->type = OPT_CONST;
4829 pd->lmod = lmod_from_directive(words[1]);
4830 i = 2;
4831 }
4832 else {
4833 if (pd == NULL) {
4834 if (verbose)
4835 anote("skipping alignment byte?\n");
4836 continue;
4837 }
4838 lmod = lmod_from_directive(words[0]);
4839 if (lmod != pd->lmod)
4840 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
4841 }
4842
4843 if (pd->count_alloc < pd->count + wordc) {
4844 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
4845 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
4846 my_assert_not(pd->d, NULL);
4847 }
4848 for (; i < wordc; i++) {
4849 if (IS(words[i], "offset")) {
4850 pd->type = OPT_OFFSET;
4851 i++;
4852 }
4853 p = strchr(words[i], ',');
4854 if (p != NULL)
4855 *p = 0;
4856 if (pd->type == OPT_OFFSET)
4857 pd->d[pd->count].u.label = strdup(words[i]);
4858 else
4859 pd->d[pd->count].u.val = parse_number(words[i]);
4860 pd->d[pd->count].bt_i = -1;
4861 pd->count++;
4862 }
4863 continue;
4864 }
4865
4866 if (in_func && !skip_func)
4867 gen_func(fout, g_fhdr, g_func, pi);
4868
4869 pending_endp = 0;
4870 in_func = 0;
4871 g_ida_func_attr = 0;
4872 skip_warned = 0;
4873 skip_func = 0;
4874 g_func[0] = 0;
4875 func_chunks_used = 0;
4876 func_chunk_i = -1;
4877 if (pi != 0) {
4878 memset(&ops, 0, pi * sizeof(ops[0]));
4879 memset(g_labels, 0, pi * sizeof(g_labels[0]));
4880 pi = 0;
4881 }
4882 g_eqcnt = 0;
4883 for (i = 0; i < g_func_pd_cnt; i++) {
4884 pd = &g_func_pd[i];
4885 if (pd->type == OPT_OFFSET) {
4886 for (j = 0; j < pd->count; j++)
4887 free(pd->d[j].u.label);
4888 }
4889 free(pd->d);
4890 pd->d = NULL;
4891 }
4892 g_func_pd_cnt = 0;
4893 pd = NULL;
4894 if (wordc == 0)
4895 continue;
4896 }
4897
4898 if (IS(words[1], "proc")) {
4899 if (in_func)
4900 aerr("proc '%s' while in_func '%s'?\n",
4901 words[0], g_func);
4902 p = words[0];
4903 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
4904 skip_func = 1;
4905 strcpy(g_func, words[0]);
4906 set_label(0, words[0]);
4907 in_func = 1;
4908 continue;
4909 }
4910
4911 if (IS(words[1], "endp"))
4912 {
4913 if (!in_func)
4914 aerr("endp '%s' while not in_func?\n", words[0]);
4915 if (!IS(g_func, words[0]))
4916 aerr("endp '%s' while in_func '%s'?\n",
4917 words[0], g_func);
4918
4919 if ((g_ida_func_attr & IDAFA_THUNK) && pi == 1
4920 && ops[0].op == OP_JMP && ops[0].operand[0].had_ds)
4921 {
4922 // import jump
4923 skip_func = 1;
4924 }
4925
4926 if (!skip_func && func_chunks_used) {
4927 // start processing chunks
4928 struct chunk_item *ci, key = { g_func, 0 };
4929
4930 func_chunk_ret = ftell(fasm);
4931 func_chunk_ret_ln = asmln;
4932 if (!func_chunks_sorted) {
4933 qsort(func_chunks, func_chunk_cnt,
4934 sizeof(func_chunks[0]), cmp_chunks);
4935 func_chunks_sorted = 1;
4936 }
4937 ci = bsearch(&key, func_chunks, func_chunk_cnt,
4938 sizeof(func_chunks[0]), cmp_chunks);
4939 if (ci == NULL)
4940 aerr("'%s' needs chunks, but none found\n", g_func);
4941 func_chunk_i = ci - func_chunks;
4942 for (; func_chunk_i > 0; func_chunk_i--)
4943 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
4944 break;
4945
4946 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
4947 if (ret)
4948 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
4949 asmln = func_chunks[func_chunk_i].asmln;
4950 func_chunk_i++;
4951 continue;
4952 }
4953 pending_endp = 1;
4954 continue;
4955 }
4956
4957 if (wordc == 2 && IS(words[1], "ends"))
4958 break;
4959
4960 p = strchr(words[0], ':');
4961 if (p != NULL) {
4962 set_label(pi, words[0]);
4963 continue;
4964 }
4965
4966 if (!in_func || skip_func) {
4967 if (!skip_warned && !skip_func && g_labels[pi][0] != 0) {
4968 if (verbose)
4969 anote("skipping from '%s'\n", g_labels[pi]);
4970 skip_warned = 1;
4971 }
4972 g_labels[pi][0] = 0;
4973 continue;
4974 }
4975
4976 if (wordc > 1 && IS(words[1], "="))
4977 {
4978 if (wordc != 5)
4979 aerr("unhandled equ, wc=%d\n", wordc);
4980 if (g_eqcnt >= eq_alloc) {
4981 eq_alloc *= 2;
4982 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
4983 my_assert_not(g_eqs, NULL);
4984 }
4985
4986 len = strlen(words[0]);
4987 if (len > sizeof(g_eqs[0].name) - 1)
4988 aerr("equ name too long: %d\n", len);
4989 strcpy(g_eqs[g_eqcnt].name, words[0]);
4990
4991 if (!IS(words[3], "ptr"))
4992 aerr("unhandled equ\n");
4993 if (IS(words[2], "dword"))
4994 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
4995 else if (IS(words[2], "word"))
4996 g_eqs[g_eqcnt].lmod = OPLM_WORD;
4997 else if (IS(words[2], "byte"))
4998 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
4999 else
5000 aerr("bad lmod: '%s'\n", words[2]);
5001
5002 g_eqs[g_eqcnt].offset = parse_number(words[4]);
5003 g_eqcnt++;
5004 continue;
5005 }
5006
5007 if (pi >= ARRAY_SIZE(ops))
5008 aerr("too many ops\n");
5009
5010 parse_op(&ops[pi], words, wordc);
5011
5012 if (sctproto != NULL) {
5013 if (ops[pi].op == OP_CALL)
5014 ops[pi].datap = sctproto;
5015 sctproto = NULL;
5016 }
5017 pi++;
5018 }
5019
5020 fclose(fout);
5021 fclose(fasm);
5022 fclose(g_fhdr);
5023
5024 return 0;
5025}
5026
5027// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.