notaz.gp2x.de / ia32rtools.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
reg return, __userstack, repe fix, more..
[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 OPF_32BIT = (1 << 15), /* 32bit division */
48 OPF_LOCK = (1 << 16), /* op has lock prefix */
49};
50
51enum op_op {
52 OP_INVAL,
53 OP_NOP,
54 OP_PUSH,
55 OP_POP,
56 OP_LEAVE,
57 OP_MOV,
58 OP_LEA,
59 OP_MOVZX,
60 OP_MOVSX,
61 OP_XCHG,
62 OP_NOT,
63 OP_CDQ,
64 OP_STOS,
65 OP_MOVS,
66 OP_CMPS,
67 OP_SCAS,
68 OP_STD,
69 OP_CLD,
70 OP_RET,
71 OP_ADD,
72 OP_SUB,
73 OP_AND,
74 OP_OR,
75 OP_XOR,
76 OP_SHL,
77 OP_SHR,
78 OP_SAR,
79 OP_ROL,
80 OP_ROR,
81 OP_RCL,
82 OP_RCR,
83 OP_ADC,
84 OP_SBB,
85 OP_BSF,
86 OP_INC,
87 OP_DEC,
88 OP_NEG,
89 OP_MUL,
90 OP_IMUL,
91 OP_DIV,
92 OP_IDIV,
93 OP_TEST,
94 OP_CMP,
95 OP_CALL,
96 OP_JMP,
97 OP_JECXZ,
98 OP_JO,
99 OP_JNO,
100 OP_JC,
101 OP_JNC,
102 OP_JZ,
103 OP_JNZ,
104 OP_JBE,
105 OP_JA,
106 OP_JS,
107 OP_JNS,
108 OP_JP,
109 OP_JNP,
110 OP_JL,
111 OP_JGE,
112 OP_JLE,
113 OP_JG,
114};
115
116enum opr_type {
117 OPT_UNSPEC,
118 OPT_REG,
119 OPT_REGMEM,
120 OPT_LABEL,
121 OPT_OFFSET,
122 OPT_CONST,
123};
124
125// must be sorted (larger len must be further in enum)
126enum opr_lenmod {
127 OPLM_UNSPEC,
128 OPLM_BYTE,
129 OPLM_WORD,
130 OPLM_DWORD,
131};
132
133#define MAX_OPERANDS 3
134
135struct parsed_opr {
136 enum opr_type type;
137 enum opr_lenmod lmod;
138 unsigned int is_ptr:1; // pointer in C
139 unsigned int is_array:1; // array in C
140 unsigned int type_from_var:1; // .. in header, sometimes wrong
141 unsigned int size_mismatch:1; // type override differs from C
142 unsigned int size_lt:1; // type override is larger than C
143 unsigned int had_ds:1; // had ds: prefix
144 int reg;
145 unsigned int val;
146 char name[256];
147};
148
149struct parsed_op {
150 enum op_op op;
151 struct parsed_opr operand[MAX_OPERANDS];
152 unsigned int flags;
153 int operand_cnt;
154 int regmask_src; // all referensed regs
155 int regmask_dst;
156 int pfomask; // flagop: parsed_flag_op that can't be delayed
157 int argnum; // push: altered before call arg #
158 int cc_scratch; // scratch storage during analysis
159 int bt_i; // branch target for branches
160 struct parsed_data *btj;// branch targets for jumptables
161 void *datap;
162};
163
164// datap:
165// OP_CALL - parser proto hint (str), ptr to struct parsed_proto
166// (OPF_CC) - point to one of (OPF_FLAGS) that affects cc op
167// OP_POP - point to OP_PUSH in push/pop pair
168
169struct parsed_equ {
170 char name[64];
171 enum opr_lenmod lmod;
172 int offset;
173};
174
175struct parsed_data {
176 char label[256];
177 enum opr_type type;
178 enum opr_lenmod lmod;
179 int count;
180 int count_alloc;
181 struct {
182 union {
183 char *label;
184 unsigned int val;
185 } u;
186 int bt_i;
187 } *d;
188};
189
190struct label_ref {
191 int i;
192 struct label_ref *next;
193};
194
195enum ida_func_attr {
196 IDAFA_BP_FRAME = (1 << 0),
197 IDAFA_LIB_FUNC = (1 << 1),
198 IDAFA_STATIC = (1 << 2),
199 IDAFA_NORETURN = (1 << 3),
200 IDAFA_THUNK = (1 << 4),
201 IDAFA_FPD = (1 << 5),
202};
203
204#define MAX_OPS 4096
205
206static struct parsed_op ops[MAX_OPS];
207static struct parsed_equ *g_eqs;
208static int g_eqcnt;
209static char g_labels[MAX_OPS][48];
210static struct label_ref g_label_refs[MAX_OPS];
211static const struct parsed_proto *g_func_pp;
212static struct parsed_data *g_func_pd;
213static int g_func_pd_cnt;
214static char g_func[256];
215static char g_comment[256];
216static int g_bp_frame;
217static int g_sp_frame;
218static int g_stack_frame_used;
219static int g_stack_fsz;
220static int g_ida_func_attr;
221static int g_allow_regfunc;
222#define ferr(op_, fmt, ...) do { \
223 printf("error:%s:#%zd: '%s': " fmt, g_func, (op_) - ops, \
224 dump_op(op_), ##__VA_ARGS__); \
225 fcloseall(); \
226 exit(1); \
227} while (0)
228#define fnote(op_, fmt, ...) \
229 printf("error:%s:#%zd: '%s': " fmt, g_func, (op_) - ops, \
230 dump_op(op_), ##__VA_ARGS__)
231
232#define MAX_REGS 8
233
234const char *regs_r32[] = { "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp" };
235const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" };
236const char *regs_r8l[] = { "al", "bl", "cl", "dl" };
237const char *regs_r8h[] = { "ah", "bh", "ch", "dh" };
238
239enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP };
240
241// possible basic comparison types (without inversion)
242enum parsed_flag_op {
243 PFO_O, // 0 OF=1
244 PFO_C, // 2 CF=1
245 PFO_Z, // 4 ZF=1
246 PFO_BE, // 6 CF=1||ZF=1
247 PFO_S, // 8 SF=1
248 PFO_P, // a PF=1
249 PFO_L, // c SF!=OF
250 PFO_LE, // e ZF=1||SF!=OF
251};
252
253static const char *parsed_flag_op_names[] = {
254 "o", "c", "z", "be", "s", "p", "l", "le"
255};
256
257static int char_array_i(const char *array[], size_t len, const char *s)
258{
259 int i;
260
261 for (i = 0; i < len; i++)
262 if (IS(s, array[i]))
263 return i;
264
265 return -1;
266}
267
268static void printf_number(char *buf, size_t buf_size,
269 unsigned long number)
270{
271 // output in C-friendly form
272 snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
273}
274
275static int parse_reg(enum opr_lenmod *reg_lmod, const char *s)
276{
277 int reg;
278
279 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s);
280 if (reg >= 0) {
281 *reg_lmod = OPLM_DWORD;
282 return reg;
283 }
284 reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s);
285 if (reg >= 0) {
286 *reg_lmod = OPLM_WORD;
287 return reg;
288 }
289 reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s);
290 if (reg >= 0) {
291 *reg_lmod = OPLM_BYTE;
292 return reg;
293 }
294 reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s);
295 if (reg >= 0) {
296 *reg_lmod = OPLM_BYTE;
297 return reg;
298 }
299
300 return -1;
301}
302
303static int parse_indmode(char *name, int *regmask, int need_c_cvt)
304{
305 enum opr_lenmod lmod;
306 char cvtbuf[256];
307 char *d = cvtbuf;
308 char *s = name;
309 char w[64];
310 long number;
311 int reg;
312 int c = 0;
313
314 *d = 0;
315
316 while (*s != 0) {
317 d += strlen(d);
318 while (my_isblank(*s))
319 s++;
320 for (; my_issep(*s); d++, s++)
321 *d = *s;
322 while (my_isblank(*s))
323 s++;
324 *d = 0;
325
326 // skip 'ds:' prefix
327 if (IS_START(s, "ds:"))
328 s += 3;
329
330 s = next_idt(w, sizeof(w), s);
331 if (w[0] == 0)
332 break;
333 c++;
334
335 reg = parse_reg(&lmod, w);
336 if (reg >= 0) {
337 *regmask |= 1 << reg;
338 goto pass;
339 }
340
341 if ('0' <= w[0] && w[0] <= '9') {
342 number = parse_number(w);
343 printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number);
344 continue;
345 }
346
347 // probably some label/identifier - pass
348
349pass:
350 snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w);
351 }
352
353 if (need_c_cvt)
354 strcpy(name, cvtbuf);
355
356 return c;
357}
358
359static int is_reg_in_str(const char *s)
360{
361 int i;
362
363 if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
364 return 0;
365
366 for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
367 if (!strncmp(s, regs_r32[i], 3))
368 return 1;
369
370 return 0;
371}
372
373static const char *parse_stack_el(const char *name, char *extra_reg)
374{
375 const char *p, *p2, *s;
376 char *endp = NULL;
377 char buf[32];
378 long val;
379 int len;
380
381 p = name;
382 if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
383 p += 4;
384 if (extra_reg != NULL) {
385 strncpy(extra_reg, name, 3);
386 extra_reg[4] = 0;
387 }
388 }
389
390 if (IS_START(p, "ebp+")) {
391 p += 4;
392
393 p2 = strchr(p, '+');
394 if (p2 != NULL && is_reg_in_str(p)) {
395 if (extra_reg != NULL) {
396 strncpy(extra_reg, p, p2 - p);
397 extra_reg[p2 - p] = 0;
398 }
399 p = p2 + 1;
400 }
401
402 if (!('0' <= *p && *p <= '9'))
403 return p;
404
405 return NULL;
406 }
407
408 if (!IS_START(name, "esp+"))
409 return NULL;
410
411 s = name + 4;
412 p = strchr(s, '+');
413 if (p) {
414 if (is_reg_in_str(s)) {
415 if (extra_reg != NULL) {
416 strncpy(extra_reg, s, p - s);
417 extra_reg[p - s] = 0;
418 }
419 s = p + 1;
420 p = strchr(s, '+');
421 if (p == NULL)
422 aerr("%s IDA stackvar not set?\n", __func__);
423 }
424 if (!('0' <= *s && *s <= '9')) {
425 aerr("%s IDA stackvar offset not set?\n", __func__);
426 return NULL;
427 }
428 if (s[0] == '0' && s[1] == 'x')
429 s += 2;
430 len = p - s;
431 if (len < sizeof(buf) - 1) {
432 strncpy(buf, s, len);
433 buf[len] = 0;
434 val = strtol(buf, &endp, 16);
435 if (val == 0 || *endp != 0) {
436 aerr("%s num parse fail for '%s'\n", __func__, buf);
437 return NULL;
438 }
439 }
440 p++;
441 }
442 else
443 p = name + 4;
444
445 if ('0' <= *p && *p <= '9')
446 return NULL;
447
448 return p;
449}
450
451static int guess_lmod_from_name(struct parsed_opr *opr)
452{
453 if (!strncmp(opr->name, "dword_", 6)) {
454 opr->lmod = OPLM_DWORD;
455 return 1;
456 }
457 if (!strncmp(opr->name, "word_", 5)) {
458 opr->lmod = OPLM_WORD;
459 return 1;
460 }
461 if (!strncmp(opr->name, "byte_", 5)) {
462 opr->lmod = OPLM_BYTE;
463 return 1;
464 }
465 return 0;
466}
467
468static int guess_lmod_from_c_type(enum opr_lenmod *lmod,
469 const struct parsed_type *c_type)
470{
471 static const char *dword_types[] = {
472 "int", "_DWORD", "UINT_PTR", "DWORD",
473 "WPARAM", "LPARAM", "UINT", "__int32",
474 "LONG", "HIMC",
475 };
476 static const char *word_types[] = {
477 "uint16_t", "int16_t", "_WORD",
478 "unsigned __int16", "__int16",
479 };
480 static const char *byte_types[] = {
481 "uint8_t", "int8_t", "char",
482 "unsigned __int8", "__int8", "BYTE", "_BYTE",
483 "CHAR", "_UNKNOWN",
484 };
485 const char *n;
486 int i;
487
488 if (c_type->is_ptr) {
489 *lmod = OPLM_DWORD;
490 return 1;
491 }
492
493 n = skip_type_mod(c_type->name);
494
495 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
496 if (IS(n, dword_types[i])) {
497 *lmod = OPLM_DWORD;
498 return 1;
499 }
500 }
501
502 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
503 if (IS(n, word_types[i])) {
504 *lmod = OPLM_WORD;
505 return 1;
506 }
507 }
508
509 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
510 if (IS(n, byte_types[i])) {
511 *lmod = OPLM_BYTE;
512 return 1;
513 }
514 }
515
516 return 0;
517}
518
519static enum opr_type lmod_from_directive(const char *d)
520{
521 if (IS(d, "dd"))
522 return OPLM_DWORD;
523 else if (IS(d, "dw"))
524 return OPLM_WORD;
525 else if (IS(d, "db"))
526 return OPLM_BYTE;
527
528 aerr("unhandled directive: '%s'\n", d);
529 return OPLM_UNSPEC;
530}
531
532static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
533 int *regmask)
534{
535 opr->type = OPT_REG;
536 opr->reg = reg;
537 opr->lmod = lmod;
538 *regmask |= 1 << reg;
539}
540
541static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
542 int *extra_offs);
543
544static int parse_operand(struct parsed_opr *opr,
545 int *regmask, int *regmask_indirect,
546 char words[16][256], int wordc, int w, unsigned int op_flags)
547{
548 const struct parsed_proto *pp;
549 enum opr_lenmod tmplmod;
550 unsigned long number;
551 char buf[256];
552 int ret, len;
553 int wordc_in;
554 char *p;
555 int i;
556
557 if (w >= wordc)
558 aerr("parse_operand w %d, wordc %d\n", w, wordc);
559
560 opr->reg = xUNSPEC;
561
562 for (i = w; i < wordc; i++) {
563 len = strlen(words[i]);
564 if (words[i][len - 1] == ',') {
565 words[i][len - 1] = 0;
566 wordc = i + 1;
567 break;
568 }
569 }
570
571 wordc_in = wordc - w;
572
573 if ((op_flags & OPF_JMP) && wordc_in > 0
574 && !('0' <= words[w][0] && words[w][0] <= '9'))
575 {
576 const char *label = NULL;
577
578 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
579 && IS(words[w + 1], "ptr"))
580 label = words[w + 2];
581 else if (wordc_in == 2 && IS(words[w], "short"))
582 label = words[w + 1];
583 else if (wordc_in == 1
584 && strchr(words[w], '[') == NULL
585 && parse_reg(&tmplmod, words[w]) < 0)
586 label = words[w];
587
588 if (label != NULL) {
589 opr->type = OPT_LABEL;
590 if (IS_START(label, "ds:")) {
591 opr->had_ds = 1;
592 label += 3;
593 }
594 strcpy(opr->name, label);
595 return wordc;
596 }
597 }
598
599 if (wordc_in >= 3) {
600 if (IS(words[w + 1], "ptr")) {
601 if (IS(words[w], "dword"))
602 opr->lmod = OPLM_DWORD;
603 else if (IS(words[w], "word"))
604 opr->lmod = OPLM_WORD;
605 else if (IS(words[w], "byte"))
606 opr->lmod = OPLM_BYTE;
607 else
608 aerr("type parsing failed\n");
609 w += 2;
610 wordc_in = wordc - w;
611 }
612 }
613
614 if (wordc_in == 2) {
615 if (IS(words[w], "offset")) {
616 opr->type = OPT_OFFSET;
617 strcpy(opr->name, words[w + 1]);
618 return wordc;
619 }
620 if (IS(words[w], "(offset")) {
621 p = strchr(words[w + 1], ')');
622 if (p == NULL)
623 aerr("parse of bracketed offset failed\n");
624 *p = 0;
625 opr->type = OPT_OFFSET;
626 strcpy(opr->name, words[w + 1]);
627 return wordc;
628 }
629 }
630
631 if (wordc_in != 1)
632 aerr("parse_operand 1 word expected\n");
633
634 if (IS_START(words[w], "ds:")) {
635 opr->had_ds = 1;
636 memmove(words[w], words[w] + 3, strlen(words[w]) - 2);
637 }
638 strcpy(opr->name, words[w]);
639
640 if (words[w][0] == '[') {
641 opr->type = OPT_REGMEM;
642 ret = sscanf(words[w], "[%[^]]]", opr->name);
643 if (ret != 1)
644 aerr("[] parse failure\n");
645
646 parse_indmode(opr->name, regmask_indirect, 1);
647 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL)) {
648 // might be an equ
649 struct parsed_equ *eq =
650 equ_find(NULL, parse_stack_el(opr->name, NULL), &i);
651 if (eq)
652 opr->lmod = eq->lmod;
653 }
654 return wordc;
655 }
656 else if (strchr(words[w], '[')) {
657 // label[reg] form
658 p = strchr(words[w], '[');
659 opr->type = OPT_REGMEM;
660 parse_indmode(p, regmask_indirect, 0);
661 strncpy(buf, words[w], p - words[w]);
662 buf[p - words[w]] = 0;
663 pp = proto_parse(g_fhdr, buf, 1);
664 goto do_label;
665 }
666 else if (('0' <= words[w][0] && words[w][0] <= '9')
667 || words[w][0] == '-')
668 {
669 number = parse_number(words[w]);
670 opr->type = OPT_CONST;
671 opr->val = number;
672 printf_number(opr->name, sizeof(opr->name), number);
673 return wordc;
674 }
675
676 ret = parse_reg(&tmplmod, opr->name);
677 if (ret >= 0) {
678 setup_reg_opr(opr, ret, tmplmod, regmask);
679 return wordc;
680 }
681
682 // most likely var in data segment
683 opr->type = OPT_LABEL;
684 pp = proto_parse(g_fhdr, opr->name, 0);
685
686do_label:
687 if (pp != NULL) {
688 if (pp->is_fptr || pp->is_func) {
689 opr->lmod = OPLM_DWORD;
690 opr->is_ptr = 1;
691 }
692 else {
693 tmplmod = OPLM_UNSPEC;
694 if (!guess_lmod_from_c_type(&tmplmod, &pp->type))
695 anote("unhandled C type '%s' for '%s'\n",
696 pp->type.name, opr->name);
697
698 if (opr->lmod == OPLM_UNSPEC) {
699 opr->lmod = tmplmod;
700 opr->type_from_var = 1;
701 }
702 else if (opr->lmod != tmplmod) {
703 opr->size_mismatch = 1;
704 if (tmplmod < opr->lmod)
705 opr->size_lt = 1;
706 }
707 opr->is_ptr = pp->type.is_ptr;
708 }
709 opr->is_array = pp->type.is_array;
710 }
711
712 if (opr->lmod == OPLM_UNSPEC)
713 guess_lmod_from_name(opr);
714 return wordc;
715}
716
717static const struct {
718 const char *name;
719 unsigned int flags;
720} pref_table[] = {
721 { "rep", OPF_REP },
722 { "repe", OPF_REP|OPF_REPZ },
723 { "repz", OPF_REP|OPF_REPZ },
724 { "repne", OPF_REP|OPF_REPNZ },
725 { "repnz", OPF_REP|OPF_REPNZ },
726 { "lock", OPF_LOCK }, // ignored for now..
727};
728
729#define OPF_CJMP_CC (OPF_JMP|OPF_CJMP|OPF_CC)
730
731static const struct {
732 const char *name;
733 enum op_op op;
734 unsigned int minopr;
735 unsigned int maxopr;
736 unsigned int flags;
737} op_table[] = {
738 { "nop", OP_NOP, 0, 0, 0 },
739 { "push", OP_PUSH, 1, 1, 0 },
740 { "pop", OP_POP, 1, 1, OPF_DATA },
741 { "leave",OP_LEAVE, 0, 0, OPF_DATA },
742 { "mov" , OP_MOV, 2, 2, OPF_DATA },
743 { "lea", OP_LEA, 2, 2, OPF_DATA },
744 { "movzx",OP_MOVZX, 2, 2, OPF_DATA },
745 { "movsx",OP_MOVSX, 2, 2, OPF_DATA },
746 { "xchg", OP_XCHG, 2, 2, OPF_DATA },
747 { "not", OP_NOT, 1, 1, OPF_DATA },
748 { "cdq", OP_CDQ, 0, 0, OPF_DATA },
749 { "stosb",OP_STOS, 0, 0, OPF_DATA },
750 { "stosw",OP_STOS, 0, 0, OPF_DATA },
751 { "stosd",OP_STOS, 0, 0, OPF_DATA },
752 { "movsb",OP_MOVS, 0, 0, OPF_DATA },
753 { "movsw",OP_MOVS, 0, 0, OPF_DATA },
754 { "movsd",OP_MOVS, 0, 0, OPF_DATA },
755 { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
756 { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
757 { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
758 { "scasb",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
759 { "scasw",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
760 { "scasd",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
761 { "std", OP_STD, 0, 0, OPF_DATA }, // special flag
762 { "cld", OP_CLD, 0, 0, OPF_DATA },
763 { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
764 { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
765 { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
766 { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS },
767 { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS },
768 { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
769 { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
770 { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
771 { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
772 { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
773 { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
774 { "rcl", OP_RCL, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
775 { "rcr", OP_RCR, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
776 { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
777 { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC },
778 { "bsf", OP_BSF, 2, 2, OPF_DATA|OPF_FLAGS },
779 { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
780 { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS },
781 { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS },
782 { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS },
783 { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS },
784 { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS },
785 { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS },
786 { "test", OP_TEST, 2, 2, OPF_FLAGS },
787 { "cmp", OP_CMP, 2, 2, OPF_FLAGS },
788 { "retn", OP_RET, 0, 1, OPF_TAIL },
789 { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS },
790 { "jmp", OP_JMP, 1, 1, OPF_JMP },
791 { "jecxz",OP_JECXZ, 1, 1, OPF_JMP|OPF_CJMP },
792 { "jo", OP_JO, 1, 1, OPF_CJMP_CC }, // 70 OF=1
793 { "jno", OP_JNO, 1, 1, OPF_CJMP_CC }, // 71 OF=0
794 { "jc", OP_JC, 1, 1, OPF_CJMP_CC }, // 72 CF=1
795 { "jb", OP_JC, 1, 1, OPF_CJMP_CC }, // 72
796 { "jnc", OP_JNC, 1, 1, OPF_CJMP_CC }, // 73 CF=0
797 { "jnb", OP_JNC, 1, 1, OPF_CJMP_CC }, // 73
798 { "jae", OP_JNC, 1, 1, OPF_CJMP_CC }, // 73
799 { "jz", OP_JZ, 1, 1, OPF_CJMP_CC }, // 74 ZF=1
800 { "je", OP_JZ, 1, 1, OPF_CJMP_CC }, // 74
801 { "jnz", OP_JNZ, 1, 1, OPF_CJMP_CC }, // 75 ZF=0
802 { "jne", OP_JNZ, 1, 1, OPF_CJMP_CC }, // 75
803 { "jbe", OP_JBE, 1, 1, OPF_CJMP_CC }, // 76 CF=1 || ZF=1
804 { "jna", OP_JBE, 1, 1, OPF_CJMP_CC }, // 76
805 { "ja", OP_JA, 1, 1, OPF_CJMP_CC }, // 77 CF=0 && ZF=0
806 { "jnbe", OP_JA, 1, 1, OPF_CJMP_CC }, // 77
807 { "js", OP_JS, 1, 1, OPF_CJMP_CC }, // 78 SF=1
808 { "jns", OP_JNS, 1, 1, OPF_CJMP_CC }, // 79 SF=0
809 { "jp", OP_JP, 1, 1, OPF_CJMP_CC }, // 7a PF=1
810 { "jpe", OP_JP, 1, 1, OPF_CJMP_CC }, // 7a
811 { "jnp", OP_JNP, 1, 1, OPF_CJMP_CC }, // 7b PF=0
812 { "jpo", OP_JNP, 1, 1, OPF_CJMP_CC }, // 7b
813 { "jl", OP_JL, 1, 1, OPF_CJMP_CC }, // 7c SF!=OF
814 { "jnge", OP_JL, 1, 1, OPF_CJMP_CC }, // 7c
815 { "jge", OP_JGE, 1, 1, OPF_CJMP_CC }, // 7d SF=OF
816 { "jnl", OP_JGE, 1, 1, OPF_CJMP_CC }, // 7d
817 { "jle", OP_JLE, 1, 1, OPF_CJMP_CC }, // 7e ZF=1 || SF!=OF
818 { "jng", OP_JLE, 1, 1, OPF_CJMP_CC }, // 7e
819 { "jg", OP_JG, 1, 1, OPF_CJMP_CC }, // 7f ZF=0 && SF=OF
820 { "jnle", OP_JG, 1, 1, OPF_CJMP_CC }, // 7f
821 { "seto", OP_JO, 1, 1, OPF_DATA|OPF_CC },
822 { "setno", OP_JNO, 1, 1, OPF_DATA|OPF_CC },
823 { "setc", OP_JC, 1, 1, OPF_DATA|OPF_CC },
824 { "setb", OP_JC, 1, 1, OPF_DATA|OPF_CC },
825 { "setnc", OP_JNC, 1, 1, OPF_DATA|OPF_CC },
826 { "setae", OP_JNC, 1, 1, OPF_DATA|OPF_CC },
827 { "setz", OP_JZ, 1, 1, OPF_DATA|OPF_CC },
828 { "sete", OP_JZ, 1, 1, OPF_DATA|OPF_CC },
829 { "setnz", OP_JNZ, 1, 1, OPF_DATA|OPF_CC },
830 { "setne", OP_JNZ, 1, 1, OPF_DATA|OPF_CC },
831 { "setbe", OP_JBE, 1, 1, OPF_DATA|OPF_CC },
832 { "setna", OP_JBE, 1, 1, OPF_DATA|OPF_CC },
833 { "seta", OP_JA, 1, 1, OPF_DATA|OPF_CC },
834 { "setnbe", OP_JA, 1, 1, OPF_DATA|OPF_CC },
835 { "sets", OP_JS, 1, 1, OPF_DATA|OPF_CC },
836 { "setns", OP_JNS, 1, 1, OPF_DATA|OPF_CC },
837 { "setp", OP_JP, 1, 1, OPF_DATA|OPF_CC },
838 { "setpe", OP_JP, 1, 1, OPF_DATA|OPF_CC },
839 { "setnp", OP_JNP, 1, 1, OPF_DATA|OPF_CC },
840 { "setpo", OP_JNP, 1, 1, OPF_DATA|OPF_CC },
841 { "setl", OP_JL, 1, 1, OPF_DATA|OPF_CC },
842 { "setnge", OP_JL, 1, 1, OPF_DATA|OPF_CC },
843 { "setge", OP_JGE, 1, 1, OPF_DATA|OPF_CC },
844 { "setnl", OP_JGE, 1, 1, OPF_DATA|OPF_CC },
845 { "setle", OP_JLE, 1, 1, OPF_DATA|OPF_CC },
846 { "setng", OP_JLE, 1, 1, OPF_DATA|OPF_CC },
847 { "setg", OP_JG, 1, 1, OPF_DATA|OPF_CC },
848 { "setnle", OP_JG, 1, 1, OPF_DATA|OPF_CC },
849};
850
851static void parse_op(struct parsed_op *op, char words[16][256], int wordc)
852{
853 enum opr_lenmod lmod = OPLM_UNSPEC;
854 int prefix_flags = 0;
855 int regmask_ind;
856 int regmask;
857 int op_w = 0;
858 int opr = 0;
859 int w = 0;
860 int i;
861
862 for (i = 0; i < ARRAY_SIZE(pref_table); i++) {
863 if (IS(words[w], pref_table[i].name)) {
864 prefix_flags = pref_table[i].flags;
865 break;
866 }
867 }
868
869 if (prefix_flags) {
870 if (wordc <= 1)
871 aerr("lone prefix: '%s'\n", words[0]);
872 w++;
873 }
874
875 op_w = w;
876 for (i = 0; i < ARRAY_SIZE(op_table); i++) {
877 if (IS(words[w], op_table[i].name))
878 break;
879 }
880
881 if (i == ARRAY_SIZE(op_table))
882 aerr("unhandled op: '%s'\n", words[0]);
883 w++;
884
885 op->op = op_table[i].op;
886 op->flags = op_table[i].flags | prefix_flags;
887 op->regmask_src = op->regmask_dst = 0;
888
889 for (opr = 0; opr < op_table[i].minopr; opr++) {
890 regmask = regmask_ind = 0;
891 w = parse_operand(&op->operand[opr], &regmask, &regmask_ind,
892 words, wordc, w, op->flags);
893
894 if (opr == 0 && (op->flags & OPF_DATA))
895 op->regmask_dst = regmask;
896 // for now, mark dst as src too
897 op->regmask_src |= regmask | regmask_ind;
898 }
899
900 for (; w < wordc && opr < op_table[i].maxopr; opr++) {
901 w = parse_operand(&op->operand[opr],
902 &op->regmask_src, &op->regmask_src,
903 words, wordc, w, op->flags);
904 }
905
906 if (w < wordc)
907 aerr("parse_op %s incomplete: %d/%d\n",
908 words[0], w, wordc);
909
910 // special cases
911 op->operand_cnt = opr;
912 if (!strncmp(op_table[i].name, "set", 3))
913 op->operand[0].lmod = OPLM_BYTE;
914
915 // ops with implicit argumets
916 switch (op->op) {
917 case OP_CDQ:
918 op->operand_cnt = 2;
919 setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst);
920 setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src);
921 break;
922
923 case OP_STOS:
924 case OP_SCAS:
925 if (op->operand_cnt != 0)
926 break;
927 if (words[op_w][4] == 'b')
928 lmod = OPLM_BYTE;
929 else if (words[op_w][4] == 'w')
930 lmod = OPLM_WORD;
931 else if (words[op_w][4] == 'd')
932 lmod = OPLM_DWORD;
933 op->operand_cnt = 3;
934 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
935 setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src);
936 op->regmask_dst = op->regmask_src;
937 setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD, &op->regmask_src);
938 break;
939
940 case OP_MOVS:
941 case OP_CMPS:
942 if (op->operand_cnt != 0)
943 break;
944 if (words[op_w][4] == 'b')
945 lmod = OPLM_BYTE;
946 else if (words[op_w][4] == 'w')
947 lmod = OPLM_WORD;
948 else if (words[op_w][4] == 'd')
949 lmod = OPLM_DWORD;
950 op->operand_cnt = 3;
951 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
952 setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src);
953 setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src);
954 op->regmask_dst = op->regmask_src;
955 break;
956
957 case OP_XCHG:
958 op->regmask_src |= op->regmask_dst;
959 op->regmask_dst |= op->regmask_src;
960 break;
961
962 case OP_JECXZ:
963 op->operand_cnt = 1;
964 op->regmask_src = 1 << xCX;
965 op->operand[0].type = OPT_REG;
966 op->operand[0].reg = xCX;
967 op->operand[0].lmod = OPLM_DWORD;
968 break;
969
970 case OP_IMUL:
971 if (op->operand_cnt != 1)
972 break;
973 // fallthrough
974 case OP_MUL:
975 // singleop mul
976 op->regmask_dst = (1 << xDX) | (1 << xAX);
977 op->regmask_src |= (1 << xAX);
978 if (op->operand[0].lmod == OPLM_UNSPEC)
979 op->operand[0].lmod = OPLM_DWORD;
980 break;
981
982 case OP_DIV:
983 case OP_IDIV:
984 // we could set up operands for edx:eax, but there is no real need to
985 // (see is_opr_modified())
986 regmask = (1 << xDX) | (1 << xAX);
987 op->regmask_dst = regmask;
988 op->regmask_src |= regmask;
989 if (op->operand[0].lmod == OPLM_UNSPEC)
990 op->operand[0].lmod = OPLM_DWORD;
991 break;
992
993 case OP_SHL:
994 case OP_SHR:
995 case OP_SAR:
996 case OP_ROL:
997 case OP_ROR:
998 if (op->operand[1].lmod == OPLM_UNSPEC)
999 op->operand[1].lmod = OPLM_BYTE;
1000 break;
1001
1002 case OP_PUSH:
1003 if (op->operand[0].lmod == OPLM_UNSPEC
1004 && (op->operand[0].type == OPT_CONST
1005 || op->operand[0].type == OPT_OFFSET
1006 || op->operand[0].type == OPT_LABEL))
1007 op->operand[0].lmod = OPLM_DWORD;
1008 break;
1009
1010 // alignment
1011 case OP_MOV:
1012 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1013 && op->operand[0].reg == xDI && op->operand[1].reg == xDI)
1014 {
1015 op->flags |= OPF_RMD;
1016 }
1017 break;
1018
1019 case OP_LEA:
1020 if (op->operand[0].type == OPT_REG
1021 && op->operand[1].type == OPT_REGMEM)
1022 {
1023 char buf[16];
1024 snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name);
1025 if (IS(buf, op->operand[1].name))
1026 op->flags |= OPF_RMD;
1027 }
1028 break;
1029
1030 case OP_CALL:
1031 // trashed regs must be explicitly detected later
1032 op->regmask_dst = 0;
1033 break;
1034
1035 default:
1036 break;
1037 }
1038}
1039
1040static const char *op_name(enum op_op op)
1041{
1042 int i;
1043
1044 for (i = 0; i < ARRAY_SIZE(op_table); i++)
1045 if (op_table[i].op == op)
1046 return op_table[i].name;
1047
1048 return "???";
1049}
1050
1051// debug
1052static const char *dump_op(struct parsed_op *po)
1053{
1054 static char out[128];
1055 char *p = out;
1056 int i;
1057
1058 if (po == NULL)
1059 return "???";
1060
1061 snprintf(out, sizeof(out), "%s", op_name(po->op));
1062 for (i = 0; i < po->operand_cnt; i++) {
1063 p += strlen(p);
1064 if (i > 0)
1065 *p++ = ',';
1066 snprintf(p, sizeof(out) - (p - out),
1067 po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s",
1068 po->operand[i].name);
1069 }
1070
1071 return out;
1072}
1073
1074static const char *lmod_type_u(struct parsed_op *po,
1075 enum opr_lenmod lmod)
1076{
1077 switch (lmod) {
1078 case OPLM_DWORD:
1079 return "u32";
1080 case OPLM_WORD:
1081 return "u16";
1082 case OPLM_BYTE:
1083 return "u8";
1084 default:
1085 ferr(po, "invalid lmod: %d\n", lmod);
1086 return "(_invalid_)";
1087 }
1088}
1089
1090static const char *lmod_cast_u(struct parsed_op *po,
1091 enum opr_lenmod lmod)
1092{
1093 switch (lmod) {
1094 case OPLM_DWORD:
1095 return "";
1096 case OPLM_WORD:
1097 return "(u16)";
1098 case OPLM_BYTE:
1099 return "(u8)";
1100 default:
1101 ferr(po, "invalid lmod: %d\n", lmod);
1102 return "(_invalid_)";
1103 }
1104}
1105
1106static const char *lmod_cast_u_ptr(struct parsed_op *po,
1107 enum opr_lenmod lmod)
1108{
1109 switch (lmod) {
1110 case OPLM_DWORD:
1111 return "*(u32 *)";
1112 case OPLM_WORD:
1113 return "*(u16 *)";
1114 case OPLM_BYTE:
1115 return "*(u8 *)";
1116 default:
1117 ferr(po, "invalid lmod: %d\n", lmod);
1118 return "(_invalid_)";
1119 }
1120}
1121
1122static const char *lmod_cast_s(struct parsed_op *po,
1123 enum opr_lenmod lmod)
1124{
1125 switch (lmod) {
1126 case OPLM_DWORD:
1127 return "(s32)";
1128 case OPLM_WORD:
1129 return "(s16)";
1130 case OPLM_BYTE:
1131 return "(s8)";
1132 default:
1133 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1134 return "(_invalid_)";
1135 }
1136}
1137
1138static const char *lmod_cast(struct parsed_op *po,
1139 enum opr_lenmod lmod, int is_signed)
1140{
1141 return is_signed ?
1142 lmod_cast_s(po, lmod) :
1143 lmod_cast_u(po, lmod);
1144}
1145
1146static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
1147{
1148 switch (lmod) {
1149 case OPLM_DWORD:
1150 return 4;
1151 case OPLM_WORD:
1152 return 2;
1153 case OPLM_BYTE:
1154 return 1;
1155 default:
1156 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1157 return 0;
1158 }
1159}
1160
1161static const char *opr_name(struct parsed_op *po, int opr_num)
1162{
1163 if (opr_num >= po->operand_cnt)
1164 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1165 return po->operand[opr_num].name;
1166}
1167
1168static unsigned int opr_const(struct parsed_op *po, int opr_num)
1169{
1170 if (opr_num >= po->operand_cnt)
1171 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1172 if (po->operand[opr_num].type != OPT_CONST)
1173 ferr(po, "opr %d: const expected\n", opr_num);
1174 return po->operand[opr_num].val;
1175}
1176
1177static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr)
1178{
1179 if ((unsigned int)popr->reg >= MAX_REGS)
1180 ferr(po, "invalid reg: %d\n", popr->reg);
1181 return regs_r32[popr->reg];
1182}
1183
1184// cast1 is the "final" cast
1185static const char *simplify_cast(const char *cast1, const char *cast2)
1186{
1187 static char buf[256];
1188
1189 if (cast1[0] == 0)
1190 return cast2;
1191 if (cast2[0] == 0)
1192 return cast1;
1193 if (IS(cast1, cast2))
1194 return cast1;
1195 if (IS(cast1, "(s8)") && IS(cast2, "(u8)"))
1196 return cast1;
1197 if (IS(cast1, "(s16)") && IS(cast2, "(u16)"))
1198 return cast1;
1199 if (IS(cast1, "(u8)") && IS_START(cast2, "*(u8 *)"))
1200 return cast2;
1201 if (IS(cast1, "(u16)") && IS_START(cast2, "*(u16 *)"))
1202 return cast2;
1203 if (strchr(cast1, '*') && IS_START(cast2, "(u32)"))
1204 return cast1;
1205
1206 snprintf(buf, sizeof(buf), "%s%s", cast1, cast2);
1207 return buf;
1208}
1209
1210static const char *simplify_cast_num(const char *cast, unsigned int val)
1211{
1212 if (IS(cast, "(u8)") && val < 0x100)
1213 return "";
1214 if (IS(cast, "(s8)") && val < 0x80)
1215 return "";
1216 if (IS(cast, "(u16)") && val < 0x10000)
1217 return "";
1218 if (IS(cast, "(s16)") && val < 0x8000)
1219 return "";
1220 if (IS(cast, "(s32)") && val < 0x80000000)
1221 return "";
1222
1223 return cast;
1224}
1225
1226static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
1227 int *extra_offs)
1228{
1229 const char *p;
1230 char *endp;
1231 int namelen;
1232 int i;
1233
1234 *extra_offs = 0;
1235 namelen = strlen(name);
1236
1237 p = strchr(name, '+');
1238 if (p != NULL) {
1239 namelen = p - name;
1240 if (namelen <= 0)
1241 ferr(po, "equ parse failed for '%s'\n", name);
1242
1243 if (IS_START(p, "0x"))
1244 p += 2;
1245 *extra_offs = strtol(p, &endp, 16);
1246 if (*endp != 0)
1247 ferr(po, "equ parse failed for '%s'\n", name);
1248 }
1249
1250 for (i = 0; i < g_eqcnt; i++)
1251 if (strncmp(g_eqs[i].name, name, namelen) == 0
1252 && g_eqs[i].name[namelen] == 0)
1253 break;
1254 if (i >= g_eqcnt) {
1255 if (po != NULL)
1256 ferr(po, "unresolved equ name: '%s'\n", name);
1257 return NULL;
1258 }
1259
1260 return &g_eqs[i];
1261}
1262
1263static int is_stack_access(struct parsed_op *po,
1264 const struct parsed_opr *popr)
1265{
1266 return (parse_stack_el(popr->name, NULL)
1267 || (g_bp_frame && !(po->flags & OPF_EBP_S)
1268 && IS_START(popr->name, "ebp")));
1269}
1270
1271static void parse_stack_access(struct parsed_op *po,
1272 const char *name, char *ofs_reg, int *offset_out,
1273 int *stack_ra_out, const char **bp_arg_out, int is_lea)
1274{
1275 const char *bp_arg = "";
1276 const char *p = NULL;
1277 struct parsed_equ *eq;
1278 char *endp = NULL;
1279 int stack_ra = 0;
1280 int offset = 0;
1281
1282 ofs_reg[0] = 0;
1283
1284 if (IS_START(name, "ebp-")
1285 || (IS_START(name, "ebp+") && '0' <= name[4] && name[4] <= '9'))
1286 {
1287 p = name + 4;
1288 if (IS_START(p, "0x"))
1289 p += 2;
1290 offset = strtoul(p, &endp, 16);
1291 if (name[3] == '-')
1292 offset = -offset;
1293 if (*endp != 0)
1294 ferr(po, "ebp- parse of '%s' failed\n", name);
1295 }
1296 else {
1297 bp_arg = parse_stack_el(name, ofs_reg);
1298 snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
1299 eq = equ_find(po, bp_arg, &offset);
1300 if (eq == NULL)
1301 ferr(po, "detected but missing eq\n");
1302 offset += eq->offset;
1303 }
1304
1305 if (!strncmp(name, "ebp", 3))
1306 stack_ra = 4;
1307
1308 // yes it sometimes LEAs ra for compares..
1309 if (!is_lea && ofs_reg[0] == 0
1310 && stack_ra <= offset && offset < stack_ra + 4)
1311 {
1312 ferr(po, "reference to ra? %d %d\n", offset, stack_ra);
1313 }
1314
1315 *offset_out = offset;
1316 *stack_ra_out = stack_ra;
1317 if (bp_arg_out)
1318 *bp_arg_out = bp_arg;
1319}
1320
1321static void stack_frame_access(struct parsed_op *po,
1322 struct parsed_opr *popr, char *buf, size_t buf_size,
1323 const char *name, const char *cast, int is_src, int is_lea)
1324{
1325 enum opr_lenmod tmp_lmod = OPLM_UNSPEC;
1326 const char *prefix = "";
1327 const char *bp_arg = NULL;
1328 char ofs_reg[16] = { 0, };
1329 int i, arg_i, arg_s;
1330 int unaligned = 0;
1331 int stack_ra = 0;
1332 int offset = 0;
1333 int sf_ofs;
1334 int lim;
1335
1336 if (po->flags & OPF_EBP_S)
1337 ferr(po, "stack_frame_access while ebp is scratch\n");
1338
1339 parse_stack_access(po, name, ofs_reg, &offset,
1340 &stack_ra, &bp_arg, is_lea);
1341
1342 if (offset > stack_ra)
1343 {
1344 arg_i = (offset - stack_ra - 4) / 4;
1345 if (arg_i < 0 || arg_i >= g_func_pp->argc_stack)
1346 {
1347 if (g_func_pp->is_vararg
1348 && arg_i == g_func_pp->argc_stack && is_lea)
1349 {
1350 // should be va_list
1351 if (cast[0] == 0)
1352 cast = "(u32)";
1353 snprintf(buf, buf_size, "%sap", cast);
1354 return;
1355 }
1356 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1357 offset, bp_arg, arg_i);
1358 }
1359 if (ofs_reg[0] != 0)
1360 ferr(po, "offset reg on arg access?\n");
1361
1362 for (i = arg_s = 0; i < g_func_pp->argc; i++) {
1363 if (g_func_pp->arg[i].reg != NULL)
1364 continue;
1365 if (arg_s == arg_i)
1366 break;
1367 arg_s++;
1368 }
1369 if (i == g_func_pp->argc)
1370 ferr(po, "arg %d not in prototype?\n", arg_i);
1371
1372 popr->is_ptr = g_func_pp->arg[i].type.is_ptr;
1373
1374 switch (popr->lmod)
1375 {
1376 case OPLM_BYTE:
1377 if (is_lea)
1378 ferr(po, "lea/byte to arg?\n");
1379 if (is_src && (offset & 3) == 0)
1380 snprintf(buf, buf_size, "%sa%d",
1381 simplify_cast(cast, "(u8)"), i + 1);
1382 else
1383 snprintf(buf, buf_size, "%sBYTE%d(a%d)",
1384 cast, offset & 3, i + 1);
1385 break;
1386
1387 case OPLM_WORD:
1388 if (is_lea)
1389 ferr(po, "lea/word to arg?\n");
1390 if (offset & 1) {
1391 unaligned = 1;
1392 if (!is_src) {
1393 if (offset & 2)
1394 ferr(po, "problematic arg store\n");
1395 snprintf(buf, buf_size, "%s((char *)&a%d + 1)",
1396 simplify_cast(cast, "*(u16 *)"), i + 1);
1397 }
1398 else
1399 ferr(po, "unaligned arg word load\n");
1400 }
1401 else if (is_src && (offset & 2) == 0)
1402 snprintf(buf, buf_size, "%sa%d",
1403 simplify_cast(cast, "(u16)"), i + 1);
1404 else
1405 snprintf(buf, buf_size, "%s%sWORD(a%d)",
1406 cast, (offset & 2) ? "HI" : "LO", i + 1);
1407 break;
1408
1409 case OPLM_DWORD:
1410 if (cast[0])
1411 prefix = cast;
1412 else if (is_src)
1413 prefix = "(u32)";
1414
1415 if (offset & 3) {
1416 unaligned = 1;
1417 if (is_lea)
1418 snprintf(buf, buf_size, "(u32)&a%d + %d",
1419 i + 1, offset & 3);
1420 else if (!is_src)
1421 ferr(po, "unaligned arg store\n");
1422 else {
1423 // mov edx, [ebp+arg_4+2]; movsx ecx, dx
1424 snprintf(buf, buf_size, "%s(a%d >> %d)",
1425 prefix, i + 1, (offset & 3) * 8);
1426 }
1427 }
1428 else {
1429 snprintf(buf, buf_size, "%s%sa%d",
1430 prefix, is_lea ? "&" : "", i + 1);
1431 }
1432 break;
1433
1434 default:
1435 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
1436 }
1437
1438 if (unaligned)
1439 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1440
1441 // common problem
1442 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type);
1443 if (tmp_lmod != OPLM_DWORD
1444 && (unaligned || (!is_src && tmp_lmod < popr->lmod)))
1445 {
1446 ferr(po, "bp_arg arg%d/w offset %d and type '%s' is too small\n",
1447 i + 1, offset, g_func_pp->arg[i].type.name);
1448 }
1449 // can't check this because msvc likes to reuse
1450 // arg space for scratch..
1451 //if (popr->is_ptr && popr->lmod != OPLM_DWORD)
1452 // ferr(po, "bp_arg arg%d: non-dword ptr access\n", i + 1);
1453 }
1454 else
1455 {
1456 if (g_stack_fsz == 0)
1457 ferr(po, "stack var access without stackframe\n");
1458 g_stack_frame_used = 1;
1459
1460 sf_ofs = g_stack_fsz + offset;
1461 lim = (ofs_reg[0] != 0) ? -4 : 0;
1462 if (offset > 0 || sf_ofs < lim)
1463 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
1464
1465 if (is_lea)
1466 prefix = "(u32)&";
1467 else
1468 prefix = cast;
1469
1470 switch (popr->lmod)
1471 {
1472 case OPLM_BYTE:
1473 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1474 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1475 break;
1476
1477 case OPLM_WORD:
1478 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1479 // known unaligned or possibly unaligned
1480 strcat(g_comment, " unaligned");
1481 if (prefix[0] == 0)
1482 prefix = "*(u16 *)&";
1483 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1484 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1485 break;
1486 }
1487 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
1488 break;
1489
1490 case OPLM_DWORD:
1491 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1492 // known unaligned or possibly unaligned
1493 strcat(g_comment, " unaligned");
1494 if (prefix[0] == 0)
1495 prefix = "*(u32 *)&";
1496 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1497 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1498 break;
1499 }
1500 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
1501 break;
1502
1503 default:
1504 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
1505 }
1506 }
1507}
1508
1509static void check_func_pp(struct parsed_op *po,
1510 const struct parsed_proto *pp, const char *pfx)
1511{
1512 char buf[256];
1513
1514 if (pp->argc_reg != 0) {
1515 if (/*!g_allow_regfunc &&*/ !pp->is_fastcall) {
1516 pp_print(buf, sizeof(buf), pp);
1517 ferr(po, "%s: unexpected reg arg in icall: %s\n", pfx, buf);
1518 }
1519 if (pp->argc_stack > 0 && pp->argc_reg != 2)
1520 ferr(po, "%s: %d reg arg(s) with %d stack arg(s)\n",
1521 pfx, pp->argc_reg, pp->argc_stack);
1522 }
1523}
1524
1525static const char *check_label_read_ref(struct parsed_op *po,
1526 const char *name)
1527{
1528 const struct parsed_proto *pp;
1529
1530 pp = proto_parse(g_fhdr, name, 0);
1531 if (pp == NULL)
1532 ferr(po, "proto_parse failed for ref '%s'\n", name);
1533
1534 if (pp->is_func)
1535 check_func_pp(po, pp, "ref");
1536
1537 return pp->name;
1538}
1539
1540static char *out_src_opr(char *buf, size_t buf_size,
1541 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
1542 int is_lea)
1543{
1544 char tmp1[256], tmp2[256];
1545 char expr[256];
1546 const char *name;
1547 char *p;
1548 int ret;
1549
1550 if (cast == NULL)
1551 cast = "";
1552
1553 switch (popr->type) {
1554 case OPT_REG:
1555 if (is_lea)
1556 ferr(po, "lea from reg?\n");
1557
1558 switch (popr->lmod) {
1559 case OPLM_DWORD:
1560 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
1561 break;
1562 case OPLM_WORD:
1563 snprintf(buf, buf_size, "%s%s",
1564 simplify_cast(cast, "(u16)"), opr_reg_p(po, popr));
1565 break;
1566 case OPLM_BYTE:
1567 if (popr->name[1] == 'h') // XXX..
1568 snprintf(buf, buf_size, "%s(%s >> 8)",
1569 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
1570 else
1571 snprintf(buf, buf_size, "%s%s",
1572 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
1573 break;
1574 default:
1575 ferr(po, "invalid src lmod: %d\n", popr->lmod);
1576 }
1577 break;
1578
1579 case OPT_REGMEM:
1580 if (is_stack_access(po, popr)) {
1581 stack_frame_access(po, popr, buf, buf_size,
1582 popr->name, cast, 1, is_lea);
1583 break;
1584 }
1585
1586 strcpy(expr, popr->name);
1587 if (strchr(expr, '[')) {
1588 // special case: '[' can only be left for label[reg] form
1589 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
1590 if (ret != 2)
1591 ferr(po, "parse failure for '%s'\n", expr);
1592 if (tmp1[0] == '(') {
1593 // (off_4FFF50+3)[eax]
1594 p = strchr(tmp1 + 1, ')');
1595 if (p == NULL || p[1] != 0)
1596 ferr(po, "parse failure (2) for '%s'\n", expr);
1597 *p = 0;
1598 memmove(tmp1, tmp1 + 1, strlen(tmp1));
1599 }
1600 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
1601 }
1602
1603 // XXX: do we need more parsing?
1604 if (is_lea) {
1605 snprintf(buf, buf_size, "%s", expr);
1606 break;
1607 }
1608
1609 snprintf(buf, buf_size, "%s(%s)",
1610 simplify_cast(cast, lmod_cast_u_ptr(po, popr->lmod)), expr);
1611 break;
1612
1613 case OPT_LABEL:
1614 name = check_label_read_ref(po, popr->name);
1615 if (cast[0] == 0 && popr->is_ptr)
1616 cast = "(u32)";
1617
1618 if (is_lea)
1619 snprintf(buf, buf_size, "(u32)&%s", name);
1620 else if (popr->size_lt)
1621 snprintf(buf, buf_size, "%s%s%s%s", cast,
1622 lmod_cast_u_ptr(po, popr->lmod),
1623 popr->is_array ? "" : "&", name);
1624 else
1625 snprintf(buf, buf_size, "%s%s%s", cast, name,
1626 popr->is_array ? "[0]" : "");
1627 break;
1628
1629 case OPT_OFFSET:
1630 name = check_label_read_ref(po, popr->name);
1631 if (cast[0] == 0)
1632 cast = "(u32)";
1633 if (is_lea)
1634 ferr(po, "lea an offset?\n");
1635 snprintf(buf, buf_size, "%s&%s", cast, name);
1636 break;
1637
1638 case OPT_CONST:
1639 if (is_lea)
1640 ferr(po, "lea from const?\n");
1641
1642 printf_number(tmp1, sizeof(tmp1), popr->val);
1643 if (popr->val == 0 && strchr(cast, '*'))
1644 snprintf(buf, buf_size, "NULL");
1645 else
1646 snprintf(buf, buf_size, "%s%s",
1647 simplify_cast_num(cast, popr->val), tmp1);
1648 break;
1649
1650 default:
1651 ferr(po, "invalid src type: %d\n", popr->type);
1652 }
1653
1654 return buf;
1655}
1656
1657// note: may set is_ptr (we find that out late for ebp frame..)
1658static char *out_dst_opr(char *buf, size_t buf_size,
1659 struct parsed_op *po, struct parsed_opr *popr)
1660{
1661 switch (popr->type) {
1662 case OPT_REG:
1663 switch (popr->lmod) {
1664 case OPLM_DWORD:
1665 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
1666 break;
1667 case OPLM_WORD:
1668 // ugh..
1669 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
1670 break;
1671 case OPLM_BYTE:
1672 // ugh..
1673 if (popr->name[1] == 'h') // XXX..
1674 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
1675 else
1676 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
1677 break;
1678 default:
1679 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
1680 }
1681 break;
1682
1683 case OPT_REGMEM:
1684 if (is_stack_access(po, popr)) {
1685 stack_frame_access(po, popr, buf, buf_size,
1686 popr->name, "", 0, 0);
1687 break;
1688 }
1689
1690 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1691
1692 case OPT_LABEL:
1693 if (popr->size_mismatch)
1694 snprintf(buf, buf_size, "%s%s%s",
1695 lmod_cast_u_ptr(po, popr->lmod),
1696 popr->is_array ? "" : "&", popr->name);
1697 else
1698 snprintf(buf, buf_size, "%s%s", popr->name,
1699 popr->is_array ? "[0]" : "");
1700 break;
1701
1702 default:
1703 ferr(po, "invalid dst type: %d\n", popr->type);
1704 }
1705
1706 return buf;
1707}
1708
1709static char *out_src_opr_u32(char *buf, size_t buf_size,
1710 struct parsed_op *po, struct parsed_opr *popr)
1711{
1712 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1713}
1714
1715static enum parsed_flag_op split_cond(struct parsed_op *po,
1716 enum op_op op, int *is_inv)
1717{
1718 *is_inv = 0;
1719
1720 switch (op) {
1721 case OP_JO:
1722 return PFO_O;
1723 case OP_JC:
1724 return PFO_C;
1725 case OP_JZ:
1726 return PFO_Z;
1727 case OP_JBE:
1728 return PFO_BE;
1729 case OP_JS:
1730 return PFO_S;
1731 case OP_JP:
1732 return PFO_P;
1733 case OP_JL:
1734 return PFO_L;
1735 case OP_JLE:
1736 return PFO_LE;
1737
1738 case OP_JNO:
1739 *is_inv = 1;
1740 return PFO_O;
1741 case OP_JNC:
1742 *is_inv = 1;
1743 return PFO_C;
1744 case OP_JNZ:
1745 *is_inv = 1;
1746 return PFO_Z;
1747 case OP_JA:
1748 *is_inv = 1;
1749 return PFO_BE;
1750 case OP_JNS:
1751 *is_inv = 1;
1752 return PFO_S;
1753 case OP_JNP:
1754 *is_inv = 1;
1755 return PFO_P;
1756 case OP_JGE:
1757 *is_inv = 1;
1758 return PFO_L;
1759 case OP_JG:
1760 *is_inv = 1;
1761 return PFO_LE;
1762
1763 case OP_RCL:
1764 case OP_RCR:
1765 case OP_ADC:
1766 case OP_SBB:
1767 return PFO_C;
1768
1769 default:
1770 ferr(po, "split_cond: bad op %d\n", op);
1771 return -1;
1772 }
1773}
1774
1775static void out_test_for_cc(char *buf, size_t buf_size,
1776 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
1777 enum opr_lenmod lmod, const char *expr)
1778{
1779 const char *cast, *scast;
1780
1781 cast = lmod_cast_u(po, lmod);
1782 scast = lmod_cast_s(po, lmod);
1783
1784 switch (pfo) {
1785 case PFO_Z:
1786 case PFO_BE: // CF=1||ZF=1; CF=0
1787 snprintf(buf, buf_size, "(%s%s %s 0)",
1788 cast, expr, is_inv ? "!=" : "==");
1789 break;
1790
1791 case PFO_S:
1792 case PFO_L: // SF!=OF; OF=0
1793 snprintf(buf, buf_size, "(%s%s %s 0)",
1794 scast, expr, is_inv ? ">=" : "<");
1795 break;
1796
1797 case PFO_LE: // ZF=1||SF!=OF; OF=0
1798 snprintf(buf, buf_size, "(%s%s %s 0)",
1799 scast, expr, is_inv ? ">" : "<=");
1800 break;
1801
1802 default:
1803 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1804 }
1805}
1806
1807static void out_cmp_for_cc(char *buf, size_t buf_size,
1808 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
1809{
1810 const char *cast, *scast, *cast_use;
1811 char buf1[256], buf2[256];
1812 enum opr_lenmod lmod;
1813
1814 if (po->operand[0].lmod != po->operand[1].lmod)
1815 ferr(po, "%s: lmod mismatch: %d %d\n", __func__,
1816 po->operand[0].lmod, po->operand[1].lmod);
1817 lmod = po->operand[0].lmod;
1818
1819 cast = lmod_cast_u(po, lmod);
1820 scast = lmod_cast_s(po, lmod);
1821
1822 switch (pfo) {
1823 case PFO_C:
1824 case PFO_Z:
1825 case PFO_BE: // !a
1826 cast_use = cast;
1827 break;
1828
1829 case PFO_S:
1830 case PFO_L: // !ge
1831 case PFO_LE:
1832 cast_use = scast;
1833 break;
1834
1835 default:
1836 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1837 }
1838
1839 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], cast_use, 0);
1840 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], cast_use, 0);
1841
1842 switch (pfo) {
1843 case PFO_C:
1844 // note: must be unsigned compare
1845 snprintf(buf, buf_size, "(%s %s %s)",
1846 buf1, is_inv ? ">=" : "<", buf2);
1847 break;
1848
1849 case PFO_Z:
1850 snprintf(buf, buf_size, "(%s %s %s)",
1851 buf1, is_inv ? "!=" : "==", buf2);
1852 break;
1853
1854 case PFO_BE: // !a
1855 // note: must be unsigned compare
1856 snprintf(buf, buf_size, "(%s %s %s)",
1857 buf1, is_inv ? ">" : "<=", buf2);
1858
1859 // annoying case
1860 if (is_inv && lmod == OPLM_BYTE
1861 && po->operand[1].type == OPT_CONST
1862 && po->operand[1].val == 0xff)
1863 {
1864 snprintf(g_comment, sizeof(g_comment), "if %s", buf);
1865 snprintf(buf, buf_size, "(0)");
1866 }
1867 break;
1868
1869 // note: must be signed compare
1870 case PFO_S:
1871 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
1872 scast, buf1, buf2, is_inv ? ">=" : "<");
1873 break;
1874
1875 case PFO_L: // !ge
1876 snprintf(buf, buf_size, "(%s %s %s)",
1877 buf1, is_inv ? ">=" : "<", buf2);
1878 break;
1879
1880 case PFO_LE:
1881 snprintf(buf, buf_size, "(%s %s %s)",
1882 buf1, is_inv ? ">" : "<=", buf2);
1883 break;
1884
1885 default:
1886 break;
1887 }
1888}
1889
1890static void out_cmp_test(char *buf, size_t buf_size,
1891 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
1892{
1893 char buf1[256], buf2[256], buf3[256];
1894
1895 if (po->op == OP_TEST) {
1896 if (IS(opr_name(po, 0), opr_name(po, 1))) {
1897 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
1898 }
1899 else {
1900 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
1901 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
1902 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
1903 }
1904 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
1905 po->operand[0].lmod, buf3);
1906 }
1907 else if (po->op == OP_CMP) {
1908 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv);
1909 }
1910 else
1911 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
1912}
1913
1914static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
1915 struct parsed_opr *popr2)
1916{
1917 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
1918 ferr(po, "missing lmod for both operands\n");
1919
1920 if (popr1->lmod == OPLM_UNSPEC)
1921 popr1->lmod = popr2->lmod;
1922 else if (popr2->lmod == OPLM_UNSPEC)
1923 popr2->lmod = popr1->lmod;
1924 else if (popr1->lmod != popr2->lmod) {
1925 if (popr1->type_from_var) {
1926 popr1->size_mismatch = 1;
1927 if (popr1->lmod < popr2->lmod)
1928 popr1->size_lt = 1;
1929 popr1->lmod = popr2->lmod;
1930 }
1931 else if (popr2->type_from_var) {
1932 popr2->size_mismatch = 1;
1933 if (popr2->lmod < popr1->lmod)
1934 popr2->size_lt = 1;
1935 popr2->lmod = popr1->lmod;
1936 }
1937 else
1938 ferr(po, "conflicting lmods: %d vs %d\n",
1939 popr1->lmod, popr2->lmod);
1940 }
1941}
1942
1943static const char *op_to_c(struct parsed_op *po)
1944{
1945 switch (po->op)
1946 {
1947 case OP_ADD:
1948 case OP_ADC:
1949 return "+";
1950 case OP_SUB:
1951 case OP_SBB:
1952 return "-";
1953 case OP_AND:
1954 return "&";
1955 case OP_OR:
1956 return "|";
1957 case OP_XOR:
1958 return "^";
1959 case OP_SHL:
1960 return "<<";
1961 case OP_SHR:
1962 return ">>";
1963 case OP_MUL:
1964 case OP_IMUL:
1965 return "*";
1966 default:
1967 ferr(po, "op_to_c was supplied with %d\n", po->op);
1968 }
1969}
1970
1971static void op_set_clear_flag(struct parsed_op *po,
1972 enum op_flags flag_set, enum op_flags flag_clear)
1973{
1974 po->flags |= flag_set;
1975 po->flags &= ~flag_clear;
1976}
1977
1978// last op in stream - unconditional branch or ret
1979#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
1980 || ((ops[_i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP \
1981 && ops[_i].op != OP_CALL))
1982
1983static int scan_for_pop(int i, int opcnt, const char *reg,
1984 int magic, int depth, int *maxdepth, int do_flags)
1985{
1986 const struct parsed_proto *pp;
1987 struct parsed_op *po;
1988 int ret = 0;
1989 int j;
1990
1991 for (; i < opcnt; i++) {
1992 po = &ops[i];
1993 if (po->cc_scratch == magic)
1994 break; // already checked
1995 po->cc_scratch = magic;
1996
1997 if (po->flags & OPF_TAIL) {
1998 if (po->op == OP_CALL) {
1999 pp = proto_parse(g_fhdr, po->operand[0].name, 0);
2000 if (pp != NULL && pp->is_noreturn)
2001 // no stack cleanup for noreturn
2002 return ret;
2003 }
2004 return -1; // deadend
2005 }
2006
2007 if ((po->flags & OPF_RMD)
2008 || (po->op == OP_PUSH && po->argnum != 0)) // arg push
2009 continue;
2010
2011 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2012 if (po->btj != NULL) {
2013 // jumptable
2014 for (j = 0; j < po->btj->count; j++) {
2015 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic,
2016 depth, maxdepth, do_flags);
2017 if (ret < 0)
2018 return ret; // dead end
2019 }
2020 return ret;
2021 }
2022
2023 if (po->bt_i < 0) {
2024 ferr(po, "dead branch\n");
2025 return -1;
2026 }
2027
2028 if (po->flags & OPF_CJMP) {
2029 ret |= scan_for_pop(po->bt_i, opcnt, reg, magic,
2030 depth, maxdepth, do_flags);
2031 if (ret < 0)
2032 return ret; // dead end
2033 }
2034 else {
2035 i = po->bt_i - 1;
2036 }
2037 continue;
2038 }
2039
2040 if ((po->op == OP_POP || po->op == OP_PUSH)
2041 && po->operand[0].type == OPT_REG
2042 && IS(po->operand[0].name, reg))
2043 {
2044 if (po->op == OP_PUSH && !(po->flags & OPF_FARG)) {
2045 depth++;
2046 if (depth > *maxdepth)
2047 *maxdepth = depth;
2048 if (do_flags)
2049 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2050 }
2051 else if (po->op == OP_POP) {
2052 if (depth == 0) {
2053 if (do_flags)
2054 op_set_clear_flag(po, OPF_RMD, OPF_RSAVE);
2055 return 1;
2056 }
2057 else {
2058 depth--;
2059 if (depth < 0) // should not happen
2060 ferr(po, "fail with depth\n");
2061 if (do_flags)
2062 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2063 }
2064 }
2065 }
2066 }
2067
2068 return ret;
2069}
2070
2071// scan for pop starting from 'ret' op (all paths)
2072static int scan_for_pop_ret(int i, int opcnt, const char *reg,
2073 int flag_set)
2074{
2075 int found = 0;
2076 int j;
2077
2078 for (; i < opcnt; i++) {
2079 if (!(ops[i].flags & OPF_TAIL))
2080 continue;
2081
2082 for (j = i - 1; j >= 0; j--) {
2083 if (ops[j].flags & OPF_RMD)
2084 continue;
2085 if (ops[j].flags & OPF_JMP)
2086 return -1;
2087
2088 if (ops[j].op == OP_POP && ops[j].operand[0].type == OPT_REG
2089 && IS(ops[j].operand[0].name, reg))
2090 {
2091 found = 1;
2092 ops[j].flags |= flag_set;
2093 break;
2094 }
2095
2096 if (g_labels[j][0] != 0)
2097 return -1;
2098 }
2099 }
2100
2101 return found ? 0 : -1;
2102}
2103
2104static void scan_propagate_df(int i, int opcnt)
2105{
2106 struct parsed_op *po = &ops[i];
2107 int j;
2108
2109 for (; i < opcnt; i++) {
2110 po = &ops[i];
2111 if (po->flags & OPF_DF)
2112 return; // already resolved
2113 po->flags |= OPF_DF;
2114
2115 if (po->op == OP_CALL)
2116 ferr(po, "call with DF set?\n");
2117
2118 if (po->flags & OPF_JMP) {
2119 if (po->btj != NULL) {
2120 // jumptable
2121 for (j = 0; j < po->btj->count; j++)
2122 scan_propagate_df(po->btj->d[j].bt_i, opcnt);
2123 return;
2124 }
2125
2126 if (po->bt_i < 0) {
2127 ferr(po, "dead branch\n");
2128 return;
2129 }
2130
2131 if (po->flags & OPF_CJMP)
2132 scan_propagate_df(po->bt_i, opcnt);
2133 else
2134 i = po->bt_i - 1;
2135 continue;
2136 }
2137
2138 if (po->flags & OPF_TAIL)
2139 break;
2140
2141 if (po->op == OP_CLD) {
2142 po->flags |= OPF_RMD;
2143 return;
2144 }
2145 }
2146
2147 ferr(po, "missing DF clear?\n");
2148}
2149
2150// is operand 'opr' modified by parsed_op 'po'?
2151static int is_opr_modified(const struct parsed_opr *opr,
2152 const struct parsed_op *po)
2153{
2154 int mask;
2155
2156 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2157 return 0;
2158
2159 if (opr->type == OPT_REG) {
2160 if (po->op == OP_CALL) {
2161 mask = (1 << xAX) | (1 << xCX) | (1 << xDX);
2162 if ((1 << opr->reg) & mask)
2163 return 1;
2164 else
2165 return 0;
2166 }
2167
2168 if (po->operand[0].type == OPT_REG) {
2169 if (po->regmask_dst & (1 << opr->reg))
2170 return 1;
2171 else
2172 return 0;
2173 }
2174 }
2175
2176 return IS(po->operand[0].name, opr->name);
2177}
2178
2179// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
2180static int is_any_opr_modified(const struct parsed_op *po_test,
2181 const struct parsed_op *po, int c_mode)
2182{
2183 int mask;
2184 int i;
2185
2186 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2187 return 0;
2188
2189 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2190 return 0;
2191
2192 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
2193 return 1;
2194
2195 // in reality, it can wreck any register, but in decompiled C
2196 // version it can only overwrite eax or edx:eax
2197 mask = (1 << xAX) | (1 << xDX);
2198 if (!c_mode)
2199 mask |= 1 << xCX;
2200
2201 if (po->op == OP_CALL
2202 && ((po_test->regmask_src | po_test->regmask_dst) & mask))
2203 return 1;
2204
2205 for (i = 0; i < po_test->operand_cnt; i++)
2206 if (IS(po_test->operand[i].name, po->operand[0].name))
2207 return 1;
2208
2209 return 0;
2210}
2211
2212// scan for any po_test operand modification in range given
2213static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt,
2214 int c_mode)
2215{
2216 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2217 return -1;
2218
2219 for (; i < opcnt; i++) {
2220 if (is_any_opr_modified(po_test, &ops[i], c_mode))
2221 return i;
2222 }
2223
2224 return -1;
2225}
2226
2227// scan for po_test operand[0] modification in range given
2228static int scan_for_mod_opr0(struct parsed_op *po_test,
2229 int i, int opcnt)
2230{
2231 for (; i < opcnt; i++) {
2232 if (is_opr_modified(&po_test->operand[0], &ops[i]))
2233 return i;
2234 }
2235
2236 return -1;
2237}
2238
2239static int scan_for_flag_set(int i, int magic, int *branched,
2240 int *setters, int *setter_cnt)
2241{
2242 struct label_ref *lr;
2243 int ret;
2244
2245 while (i >= 0) {
2246 if (ops[i].cc_scratch == magic) {
2247 ferr(&ops[i], "%s looped\n", __func__);
2248 return -1;
2249 }
2250 ops[i].cc_scratch = magic;
2251
2252 if (g_labels[i][0] != 0) {
2253 *branched = 1;
2254
2255 lr = &g_label_refs[i];
2256 for (; lr->next; lr = lr->next) {
2257 ret = scan_for_flag_set(lr->i, magic,
2258 branched, setters, setter_cnt);
2259 if (ret < 0)
2260 return ret;
2261 }
2262
2263 if (i > 0 && LAST_OP(i - 1)) {
2264 i = lr->i;
2265 continue;
2266 }
2267 ret = scan_for_flag_set(lr->i, magic,
2268 branched, setters, setter_cnt);
2269 if (ret < 0)
2270 return ret;
2271 }
2272 i--;
2273
2274 if (ops[i].flags & OPF_FLAGS) {
2275 setters[*setter_cnt] = i;
2276 (*setter_cnt)++;
2277 return 0;
2278 }
2279
2280 if ((ops[i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP)
2281 return -1;
2282 }
2283
2284 return -1;
2285}
2286
2287// scan back for cdq, if anything modifies edx, fail
2288static int scan_for_cdq_edx(int i)
2289{
2290 while (i >= 0) {
2291 if (g_labels[i][0] != 0) {
2292 if (g_label_refs[i].next != NULL)
2293 return -1;
2294 if (i > 0 && LAST_OP(i - 1)) {
2295 i = g_label_refs[i].i;
2296 continue;
2297 }
2298 return -1;
2299 }
2300 i--;
2301
2302 if (ops[i].op == OP_CDQ)
2303 return i;
2304
2305 if (ops[i].regmask_dst & (1 << xDX))
2306 return -1;
2307 }
2308
2309 return -1;
2310}
2311
2312static int scan_for_reg_clear(int i, int reg)
2313{
2314 while (i >= 0) {
2315 if (g_labels[i][0] != 0) {
2316 if (g_label_refs[i].next != NULL)
2317 return -1;
2318 if (i > 0 && LAST_OP(i - 1)) {
2319 i = g_label_refs[i].i;
2320 continue;
2321 }
2322 return -1;
2323 }
2324 i--;
2325
2326 if (ops[i].op == OP_XOR
2327 && ops[i].operand[0].lmod == OPLM_DWORD
2328 && ops[i].operand[0].reg == ops[i].operand[1].reg
2329 && ops[i].operand[0].reg == reg)
2330 return i;
2331
2332 if (ops[i].regmask_dst & (1 << reg))
2333 return -1;
2334 }
2335
2336 return -1;
2337}
2338
2339// scan for positive, constant esp adjust
2340static int scan_for_esp_adjust(int i, int opcnt, int *adj,
2341 int *multipath)
2342{
2343 const struct parsed_proto *pp;
2344 struct parsed_op *po;
2345 int first_pop = -1;
2346
2347 *adj = *multipath = 0;
2348
2349 for (; i < opcnt; i++) {
2350 po = &ops[i];
2351
2352 if (g_labels[i][0] != 0)
2353 *multipath = 1;
2354
2355 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
2356 if (po->operand[1].type != OPT_CONST)
2357 ferr(&ops[i], "non-const esp adjust?\n");
2358 *adj += po->operand[1].val;
2359 if (*adj & 3)
2360 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
2361 return i;
2362 }
2363 else if (po->op == OP_PUSH && !(po->flags & OPF_RMD)) {
2364 //if (first_pop == -1)
2365 // first_pop = -2; // none
2366 *adj -= lmod_bytes(po, po->operand[0].lmod);
2367 }
2368 else if (po->op == OP_POP && !(po->flags & OPF_RMD)) {
2369 // seems like msvc only uses 'pop ecx' for stack realignment..
2370 if (po->operand[0].type != OPT_REG || po->operand[0].reg != xCX)
2371 break;
2372 if (first_pop == -1 && *adj >= 0)
2373 first_pop = i;
2374 *adj += lmod_bytes(po, po->operand[0].lmod);
2375 }
2376 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
2377 if (po->op == OP_JMP && po->btj == NULL) {
2378 i = po->bt_i - 1;
2379 continue;
2380 }
2381 if (po->op != OP_CALL)
2382 break;
2383 if (po->operand[0].type != OPT_LABEL)
2384 break;
2385 pp = po->datap;
2386 if (pp != NULL && pp->is_stdcall)
2387 break;
2388 }
2389 }
2390
2391 if (first_pop >= 0) {
2392 // probably 'pop ecx' was used..
2393 return first_pop;
2394 }
2395
2396 return -1;
2397}
2398
2399static void scan_fwd_set_flags(int i, int opcnt, int magic, int flags)
2400{
2401 struct parsed_op *po;
2402 int j;
2403
2404 if (i < 0)
2405 ferr(ops, "%s: followed bad branch?\n", __func__);
2406
2407 for (; i < opcnt; i++) {
2408 po = &ops[i];
2409 if (po->cc_scratch == magic)
2410 return;
2411 po->cc_scratch = magic;
2412 po->flags |= flags;
2413
2414 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2415 if (po->btj != NULL) {
2416 // jumptable
2417 for (j = 0; j < po->btj->count; j++)
2418 scan_fwd_set_flags(po->btj->d[j].bt_i, opcnt, magic, flags);
2419 return;
2420 }
2421
2422 scan_fwd_set_flags(po->bt_i, opcnt, magic, flags);
2423 if (!(po->flags & OPF_CJMP))
2424 return;
2425 }
2426 if (po->flags & OPF_TAIL)
2427 return;
2428 }
2429}
2430
2431static const struct parsed_proto *try_recover_pp(
2432 struct parsed_op *po, const struct parsed_opr *opr, int *search_instead)
2433{
2434 const struct parsed_proto *pp = NULL;
2435 char buf[256];
2436 char *p;
2437
2438 // maybe an arg of g_func?
2439 if (opr->type == OPT_REGMEM && is_stack_access(po, opr))
2440 {
2441 char ofs_reg[16] = { 0, };
2442 int arg, arg_s, arg_i;
2443 int stack_ra = 0;
2444 int offset = 0;
2445
2446 parse_stack_access(po, opr->name, ofs_reg,
2447 &offset, &stack_ra, NULL, 0);
2448 if (ofs_reg[0] != 0)
2449 ferr(po, "offset reg on arg access?\n");
2450 if (offset <= stack_ra) {
2451 // search who set the stack var instead
2452 if (search_instead != NULL)
2453 *search_instead = 1;
2454 return NULL;
2455 }
2456
2457 arg_i = (offset - stack_ra - 4) / 4;
2458 for (arg = arg_s = 0; arg < g_func_pp->argc; arg++) {
2459 if (g_func_pp->arg[arg].reg != NULL)
2460 continue;
2461 if (arg_s == arg_i)
2462 break;
2463 arg_s++;
2464 }
2465 if (arg == g_func_pp->argc)
2466 ferr(po, "stack arg %d not in prototype?\n", arg_i);
2467
2468 pp = g_func_pp->arg[arg].fptr;
2469 if (pp == NULL)
2470 ferr(po, "icall sa: arg%d is not a fptr?\n", arg + 1);
2471 check_func_pp(po, pp, "icall arg");
2472 }
2473 else if (opr->type == OPT_REGMEM && strchr(opr->name + 1, '[')) {
2474 // label[index]
2475 p = strchr(opr->name + 1, '[');
2476 memcpy(buf, opr->name, p - opr->name);
2477 buf[p - opr->name] = 0;
2478 pp = proto_parse(g_fhdr, buf, 0);
2479 }
2480 else if (opr->type == OPT_OFFSET || opr->type == OPT_LABEL) {
2481 pp = proto_parse(g_fhdr, opr->name, 0);
2482 if (pp == NULL)
2483 ferr(po, "proto_parse failed for icall from '%s'\n", opr->name);
2484 check_func_pp(po, pp, "reg-fptr ref");
2485 }
2486
2487 return pp;
2488}
2489
2490static void scan_for_call_type(int i, const struct parsed_opr *opr,
2491 int magic, const struct parsed_proto **pp_found, int *multi)
2492{
2493 const struct parsed_proto *pp = NULL;
2494 struct parsed_op *po;
2495 struct label_ref *lr;
2496
2497 ops[i].cc_scratch = magic;
2498
2499 while (1) {
2500 if (g_labels[i][0] != 0) {
2501 lr = &g_label_refs[i];
2502 for (; lr != NULL; lr = lr->next)
2503 scan_for_call_type(lr->i, opr, magic, pp_found, multi);
2504 if (i > 0 && LAST_OP(i - 1))
2505 return;
2506 }
2507
2508 i--;
2509 if (i < 0)
2510 break;
2511
2512 if (ops[i].cc_scratch == magic)
2513 return;
2514 ops[i].cc_scratch = magic;
2515
2516 if (!(ops[i].flags & OPF_DATA))
2517 continue;
2518 if (!is_opr_modified(opr, &ops[i]))
2519 continue;
2520 if (ops[i].op != OP_MOV && ops[i].op != OP_LEA) {
2521 // most probably trashed by some processing
2522 *pp_found = NULL;
2523 return;
2524 }
2525
2526 opr = &ops[i].operand[1];
2527 if (opr->type != OPT_REG)
2528 break;
2529 }
2530
2531 po = (i >= 0) ? &ops[i] : ops;
2532
2533 if (i < 0) {
2534 // reached the top - can only be an arg-reg
2535 if (opr->type != OPT_REG)
2536 return;
2537
2538 for (i = 0; i < g_func_pp->argc; i++) {
2539 if (g_func_pp->arg[i].reg == NULL)
2540 continue;
2541 if (IS(opr->name, g_func_pp->arg[i].reg))
2542 break;
2543 }
2544 if (i == g_func_pp->argc)
2545 return;
2546 pp = g_func_pp->arg[i].fptr;
2547 if (pp == NULL)
2548 ferr(po, "icall: arg%d (%s) is not a fptr?\n",
2549 i + 1, g_func_pp->arg[i].reg);
2550 check_func_pp(po, pp, "icall reg-arg");
2551 }
2552 else
2553 pp = try_recover_pp(po, opr, NULL);
2554
2555 if (*pp_found != NULL && pp != NULL && *pp_found != pp) {
2556 if (!IS((*pp_found)->ret_type.name, pp->ret_type.name)
2557 || (*pp_found)->is_stdcall != pp->is_stdcall
2558 || (*pp_found)->is_fptr != pp->is_fptr
2559 || (*pp_found)->argc != pp->argc
2560 || (*pp_found)->argc_reg != pp->argc_reg
2561 || (*pp_found)->argc_stack != pp->argc_stack)
2562 {
2563 ferr(po, "icall: parsed_proto mismatch\n");
2564 }
2565 *multi = 1;
2566 }
2567 if (pp != NULL)
2568 *pp_found = pp;
2569}
2570
2571static const struct parsed_proto *resolve_icall(int i, int opcnt,
2572 int *multi_src)
2573{
2574 const struct parsed_proto *pp = NULL;
2575 int search_advice = 0;
2576
2577 *multi_src = 0;
2578
2579 switch (ops[i].operand[0].type) {
2580 case OPT_REGMEM:
2581 case OPT_LABEL:
2582 case OPT_OFFSET:
2583 pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice);
2584 if (!search_advice)
2585 break;
2586 // fallthrough
2587 default:
2588 scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp,
2589 multi_src);
2590 break;
2591 }
2592
2593 return pp;
2594}
2595
2596static int try_resolve_const(int i, const struct parsed_opr *opr,
2597 int magic, unsigned int *val)
2598{
2599 struct label_ref *lr;
2600 int ret = 0;
2601
2602 ops[i].cc_scratch = magic;
2603
2604 while (1) {
2605 if (g_labels[i][0] != 0) {
2606 lr = &g_label_refs[i];
2607 for (; lr != NULL; lr = lr->next)
2608 ret |= try_resolve_const(lr->i, opr, magic, val);
2609 if (i > 0 && LAST_OP(i - 1))
2610 return ret;
2611 }
2612
2613 i--;
2614 if (i < 0)
2615 return -1;
2616
2617 if (ops[i].cc_scratch == magic)
2618 return 0;
2619 ops[i].cc_scratch = magic;
2620
2621 if (!(ops[i].flags & OPF_DATA))
2622 continue;
2623 if (!is_opr_modified(opr, &ops[i]))
2624 continue;
2625 if (ops[i].op != OP_MOV && ops[i].operand[1].type != OPT_CONST)
2626 return -1;
2627
2628 *val = ops[i].operand[1].val;
2629 return 1;
2630 }
2631}
2632
2633static int collect_call_args_r(struct parsed_op *po, int i,
2634 struct parsed_proto *pp, int *regmask, int *save_arg_vars, int arg,
2635 int magic, int need_op_saving, int may_reuse)
2636{
2637 struct parsed_proto *pp_tmp;
2638 struct label_ref *lr;
2639 int need_to_save_current;
2640 int ret = 0;
2641 int j;
2642
2643 if (i < 0) {
2644 ferr(po, "dead label encountered\n");
2645 return -1;
2646 }
2647
2648 for (; arg < pp->argc; arg++)
2649 if (pp->arg[arg].reg == NULL)
2650 break;
2651 magic = (magic & 0xffffff) | (arg << 24);
2652
2653 for (j = i; j >= 0 && (arg < pp->argc || pp->is_unresolved); )
2654 {
2655 if (((ops[j].cc_scratch ^ magic) & 0xffffff) == 0) {
2656 if (ops[j].cc_scratch != magic) {
2657 ferr(&ops[j], "arg collect hit same path with diff args for %s\n",
2658 pp->name);
2659 return -1;
2660 }
2661 // ok: have already been here
2662 return 0;
2663 }
2664 ops[j].cc_scratch = magic;
2665
2666 if (g_labels[j][0] != 0 && g_label_refs[j].i != -1) {
2667 lr = &g_label_refs[j];
2668 if (lr->next != NULL)
2669 need_op_saving = 1;
2670 for (; lr->next; lr = lr->next) {
2671 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
2672 may_reuse = 1;
2673 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
2674 arg, magic, need_op_saving, may_reuse);
2675 if (ret < 0)
2676 return ret;
2677 }
2678
2679 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
2680 may_reuse = 1;
2681 if (j > 0 && LAST_OP(j - 1)) {
2682 // follow last branch in reverse
2683 j = lr->i;
2684 continue;
2685 }
2686 need_op_saving = 1;
2687 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
2688 arg, magic, need_op_saving, may_reuse);
2689 if (ret < 0)
2690 return ret;
2691 }
2692 j--;
2693
2694 if (ops[j].op == OP_CALL)
2695 {
2696 if (pp->is_unresolved)
2697 break;
2698
2699 pp_tmp = ops[j].datap;
2700 if (pp_tmp == NULL)
2701 ferr(po, "arg collect hit unparsed call '%s'\n",
2702 ops[j].operand[0].name);
2703 if (may_reuse && pp_tmp->argc_stack > 0)
2704 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
2705 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
2706 }
2707 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP) {
2708 if (pp->is_unresolved)
2709 break;
2710
2711 ferr(po, "arg collect %d/%d hit esp adjust\n",
2712 arg, pp->argc);
2713 }
2714 else if (ops[j].op == OP_POP) {
2715 if (pp->is_unresolved)
2716 break;
2717
2718 ferr(po, "arg collect %d/%d hit pop\n", arg, pp->argc);
2719 }
2720 else if (ops[j].flags & OPF_CJMP)
2721 {
2722 if (pp->is_unresolved)
2723 break;
2724
2725 may_reuse = 1;
2726 }
2727 else if (ops[j].op == OP_PUSH && !(ops[j].flags & OPF_FARG))
2728 {
2729 if (pp->is_unresolved && (ops[j].flags & OPF_RMD))
2730 break;
2731
2732 pp->arg[arg].datap = &ops[j];
2733 need_to_save_current = 0;
2734 if (!need_op_saving) {
2735 ret = scan_for_mod(&ops[j], j + 1, i, 1);
2736 need_to_save_current = (ret >= 0);
2737 }
2738 if (need_op_saving || need_to_save_current) {
2739 // mark this push as one that needs operand saving
2740 ops[j].flags &= ~OPF_RMD;
2741 if (ops[j].argnum == 0) {
2742 ops[j].argnum = arg + 1;
2743 *save_arg_vars |= 1 << arg;
2744 }
2745 else if (ops[j].argnum < arg + 1)
2746 ferr(&ops[j], "argnum conflict (%d<%d) for '%s'\n",
2747 ops[j].argnum, arg + 1, pp->name);
2748 }
2749 else if (ops[j].argnum == 0)
2750 ops[j].flags |= OPF_RMD;
2751
2752 // some PUSHes are reused by different calls on other branches,
2753 // but that can't happen if we didn't branch, so they
2754 // can be removed from future searches (handles nested calls)
2755 if (!may_reuse)
2756 ops[j].flags |= OPF_FARG;
2757
2758 ops[j].flags &= ~OPF_RSAVE;
2759
2760 arg++;
2761 if (!pp->is_unresolved) {
2762 // next arg
2763 for (; arg < pp->argc; arg++)
2764 if (pp->arg[arg].reg == NULL)
2765 break;
2766 }
2767 magic = (magic & 0xffffff) | (arg << 24);
2768
2769 // tracking reg usage
2770 if (ops[j].operand[0].type == OPT_REG)
2771 *regmask |= 1 << ops[j].operand[0].reg;
2772 }
2773 }
2774
2775 if (arg < pp->argc) {
2776 ferr(po, "arg collect failed for '%s': %d/%d\n",
2777 pp->name, arg, pp->argc);
2778 return -1;
2779 }
2780
2781 return arg;
2782}
2783
2784static int collect_call_args(struct parsed_op *po, int i,
2785 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
2786 int magic)
2787{
2788 int ret;
2789 int a;
2790
2791 ret = collect_call_args_r(po, i, pp, regmask, save_arg_vars,
2792 0, magic, 0, 0);
2793 if (ret < 0)
2794 return ret;
2795
2796 if (pp->is_unresolved) {
2797 pp->argc += ret;
2798 pp->argc_stack += ret;
2799 for (a = 0; a < pp->argc; a++)
2800 if (pp->arg[a].type.name == NULL)
2801 pp->arg[a].type.name = strdup("int");
2802 }
2803
2804 return ret;
2805}
2806
2807// early check for tail call or branch back
2808static int is_like_tailjmp(int j)
2809{
2810 if (!(ops[j].flags & OPF_JMP))
2811 return 0;
2812
2813 if (ops[j].op == OP_JMP && !ops[j].operand[0].had_ds)
2814 // probably local branch back..
2815 return 1;
2816 if (ops[j].op == OP_CALL)
2817 // probably noreturn call..
2818 return 1;
2819
2820 return 0;
2821}
2822
2823static void pp_insert_reg_arg(struct parsed_proto *pp, const char *reg)
2824{
2825 int i;
2826
2827 for (i = 0; i < pp->argc; i++)
2828 if (pp->arg[i].reg == NULL)
2829 break;
2830
2831 if (pp->argc_stack)
2832 memmove(&pp->arg[i + 1], &pp->arg[i],
2833 sizeof(pp->arg[0]) * pp->argc_stack);
2834 memset(&pp->arg[i], 0, sizeof(pp->arg[i]));
2835 pp->arg[i].reg = strdup(reg);
2836 pp->arg[i].type.name = strdup("int");
2837 pp->argc++;
2838 pp->argc_reg++;
2839}
2840
2841static void add_label_ref(struct label_ref *lr, int op_i)
2842{
2843 struct label_ref *lr_new;
2844
2845 if (lr->i == -1) {
2846 lr->i = op_i;
2847 return;
2848 }
2849
2850 lr_new = calloc(1, sizeof(*lr_new));
2851 lr_new->i = op_i;
2852 lr_new->next = lr->next;
2853 lr->next = lr_new;
2854}
2855
2856static void output_std_flags(FILE *fout, struct parsed_op *po,
2857 int *pfomask, const char *dst_opr_text)
2858{
2859 if (*pfomask & (1 << PFO_Z)) {
2860 fprintf(fout, "\n cond_z = (%s%s == 0);",
2861 lmod_cast_u(po, po->operand[0].lmod), dst_opr_text);
2862 *pfomask &= ~(1 << PFO_Z);
2863 }
2864 if (*pfomask & (1 << PFO_S)) {
2865 fprintf(fout, "\n cond_s = (%s%s < 0);",
2866 lmod_cast_s(po, po->operand[0].lmod), dst_opr_text);
2867 *pfomask &= ~(1 << PFO_S);
2868 }
2869}
2870
2871static void output_pp_attrs(FILE *fout, const struct parsed_proto *pp,
2872 int is_noreturn)
2873{
2874 if (pp->is_fastcall)
2875 fprintf(fout, "__fastcall ");
2876 else if (pp->is_stdcall && pp->argc_reg == 0)
2877 fprintf(fout, "__stdcall ");
2878 if (pp->is_noreturn || is_noreturn)
2879 fprintf(fout, "noreturn ");
2880}
2881
2882static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
2883{
2884 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
2885 struct parsed_opr *last_arith_dst = NULL;
2886 char buf1[256], buf2[256], buf3[256], cast[64];
2887 const struct parsed_proto *pp_c;
2888 struct parsed_proto *pp, *pp_tmp;
2889 struct parsed_data *pd;
2890 const char *tmpname;
2891 enum parsed_flag_op pfo;
2892 unsigned int uval;
2893 int save_arg_vars = 0;
2894 int cond_vars = 0;
2895 int need_tmp_var = 0;
2896 int need_tmp64 = 0;
2897 int had_decl = 0;
2898 int label_pending = 0;
2899 int regmask_save = 0;
2900 int regmask_arg = 0;
2901 int regmask_now = 0;
2902 int regmask_init = 0;
2903 int regmask = 0;
2904 int pfomask = 0;
2905 int found = 0;
2906 int depth = 0;
2907 int no_output;
2908 int i, j, l;
2909 int dummy;
2910 int arg;
2911 int reg;
2912 int ret;
2913
2914 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
2915 g_stack_frame_used = 0;
2916
2917 g_func_pp = proto_parse(fhdr, funcn, 0);
2918 if (g_func_pp == NULL)
2919 ferr(ops, "proto_parse failed for '%s'\n", funcn);
2920
2921 fprintf(fout, "%s ", g_func_pp->ret_type.name);
2922 output_pp_attrs(fout, g_func_pp, g_ida_func_attr & IDAFA_NORETURN);
2923 fprintf(fout, "%s(", g_func_pp->name);
2924
2925 for (i = 0; i < g_func_pp->argc; i++) {
2926 if (i > 0)
2927 fprintf(fout, ", ");
2928 if (g_func_pp->arg[i].fptr != NULL) {
2929 // func pointer..
2930 pp = g_func_pp->arg[i].fptr;
2931 fprintf(fout, "%s (", pp->ret_type.name);
2932 output_pp_attrs(fout, pp, 0);
2933 fprintf(fout, "*a%d)(", i + 1);
2934 for (j = 0; j < pp->argc; j++) {
2935 if (j > 0)
2936 fprintf(fout, ", ");
2937 if (pp->arg[j].fptr)
2938 ferr(ops, "nested fptr\n");
2939 fprintf(fout, "%s", pp->arg[j].type.name);
2940 }
2941 if (pp->is_vararg) {
2942 if (j > 0)
2943 fprintf(fout, ", ");
2944 fprintf(fout, "...");
2945 }
2946 fprintf(fout, ")");
2947 }
2948 else if (g_func_pp->arg[i].type.is_retreg) {
2949 fprintf(fout, "u32 *r_%s", g_func_pp->arg[i].reg);
2950 }
2951 else {
2952 fprintf(fout, "%s a%d", g_func_pp->arg[i].type.name, i + 1);
2953 }
2954 if (g_func_pp->arg[i].reg != NULL) {
2955 reg = char_array_i(regs_r32,
2956 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
2957 if (reg < 0)
2958 ferr(ops, "arg '%s' is not a reg?\n", g_func_pp->arg[i].reg);
2959 regmask_arg |= 1 << reg;
2960 }
2961 }
2962 if (g_func_pp->is_vararg) {
2963 if (i > 0)
2964 fprintf(fout, ", ");
2965 fprintf(fout, "...");
2966 }
2967
2968 fprintf(fout, ")\n{\n");
2969
2970 // pass1:
2971 // - handle ebp/esp frame, remove ops related to it
2972 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
2973 && ops[1].op == OP_MOV
2974 && IS(opr_name(&ops[1], 0), "ebp")
2975 && IS(opr_name(&ops[1], 1), "esp"))
2976 {
2977 int ecx_push = 0;
2978
2979 g_bp_frame = 1;
2980 ops[0].flags |= OPF_RMD;
2981 ops[1].flags |= OPF_RMD;
2982 i = 2;
2983
2984 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
2985 g_stack_fsz = opr_const(&ops[2], 1);
2986 ops[2].flags |= OPF_RMD;
2987 i++;
2988 }
2989 else {
2990 // another way msvc builds stack frame..
2991 i = 2;
2992 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
2993 g_stack_fsz += 4;
2994 ops[i].flags |= OPF_RMD;
2995 ecx_push++;
2996 i++;
2997 }
2998 // and another way..
2999 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
3000 && ops[i].operand[1].type == OPT_CONST
3001 && ops[i + 1].op == OP_CALL
3002 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
3003 {
3004 g_stack_fsz += ops[i].operand[1].val;
3005 ops[i].flags |= OPF_RMD;
3006 i++;
3007 ops[i].flags |= OPF_RMD;
3008 i++;
3009 }
3010 }
3011
3012 found = 0;
3013 do {
3014 for (; i < opcnt; i++)
3015 if (ops[i].op == OP_RET)
3016 break;
3017 j = i - 1;
3018 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
3019 if (found && is_like_tailjmp(j))
3020 break;
3021 j--;
3022 }
3023
3024 if ((ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ebp"))
3025 || ops[j].op == OP_LEAVE)
3026 {
3027 ops[j].flags |= OPF_RMD;
3028 }
3029 else if (!(g_ida_func_attr & IDAFA_NORETURN))
3030 ferr(&ops[j], "'pop ebp' expected\n");
3031
3032 if (g_stack_fsz != 0) {
3033 if (ops[j - 1].op == OP_MOV
3034 && IS(opr_name(&ops[j - 1], 0), "esp")
3035 && IS(opr_name(&ops[j - 1], 1), "ebp"))
3036 {
3037 ops[j - 1].flags |= OPF_RMD;
3038 }
3039 else if (ops[j].op != OP_LEAVE
3040 && !(g_ida_func_attr & IDAFA_NORETURN))
3041 {
3042 ferr(&ops[j - 1], "esp restore expected\n");
3043 }
3044
3045 if (ecx_push && ops[j - 2].op == OP_POP
3046 && IS(opr_name(&ops[j - 2], 0), "ecx"))
3047 {
3048 ferr(&ops[j - 2], "unexpected ecx pop\n");
3049 }
3050 }
3051
3052 found = 1;
3053 i++;
3054 } while (i < opcnt);
3055 }
3056 else {
3057 for (i = 0; i < opcnt; i++) {
3058 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
3059 break;
3060 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
3061 && ops[i].operand[1].type == OPT_CONST)
3062 {
3063 g_sp_frame = 1;
3064 break;
3065 }
3066 }
3067
3068 found = 0;
3069 if (g_sp_frame)
3070 {
3071 g_stack_fsz = ops[i].operand[1].val;
3072 ops[i].flags |= OPF_RMD;
3073
3074 i++;
3075 do {
3076 for (; i < opcnt; i++)
3077 if (ops[i].op == OP_RET)
3078 break;
3079 j = i - 1;
3080 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
3081 if (found && is_like_tailjmp(j))
3082 break;
3083 j--;
3084 }
3085
3086 if (ops[j].op != OP_ADD
3087 || !IS(opr_name(&ops[j], 0), "esp")
3088 || ops[j].operand[1].type != OPT_CONST
3089 || ops[j].operand[1].val != g_stack_fsz)
3090 ferr(&ops[j], "'add esp' expected\n");
3091 ops[j].flags |= OPF_RMD;
3092
3093 found = 1;
3094 i++;
3095 } while (i < opcnt);
3096 }
3097 }
3098
3099 // pass2:
3100 // - parse calls with labels
3101 // - resolve all branches
3102 for (i = 0; i < opcnt; i++)
3103 {
3104 po = &ops[i];
3105 po->bt_i = -1;
3106 po->btj = NULL;
3107
3108 if (po->flags & OPF_RMD)
3109 continue;
3110
3111 if (po->op == OP_CALL) {
3112 pp = NULL;
3113
3114 if (po->operand[0].type == OPT_LABEL) {
3115 tmpname = opr_name(po, 0);
3116 if (IS_START(tmpname, "loc_"))
3117 ferr(po, "call to loc_*\n");
3118 pp_c = proto_parse(fhdr, tmpname, 0);
3119 if (pp_c == NULL)
3120 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
3121
3122 pp = proto_clone(pp_c);
3123 my_assert_not(pp, NULL);
3124 }
3125 else if (po->datap != NULL) {
3126 pp = calloc(1, sizeof(*pp));
3127 my_assert_not(pp, NULL);
3128
3129 ret = parse_protostr(po->datap, pp);
3130 if (ret < 0)
3131 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
3132 free(po->datap);
3133 po->datap = NULL;
3134 }
3135
3136 if (pp != NULL) {
3137 if (pp->is_fptr)
3138 check_func_pp(po, pp, "fptr var call");
3139 if (pp->is_noreturn)
3140 po->flags |= OPF_TAIL;
3141 po->datap = pp;
3142 }
3143 continue;
3144 }
3145
3146 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
3147 continue;
3148
3149 if (po->operand[0].type == OPT_REGMEM) {
3150 char *p = strchr(po->operand[0].name, '[');
3151 if (p == NULL)
3152 goto tailcall;
3153 ret = p - po->operand[0].name;
3154 strncpy(buf1, po->operand[0].name, ret);
3155 buf1[ret] = 0;
3156
3157 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
3158 if (IS(g_func_pd[j].label, buf1)) {
3159 pd = &g_func_pd[j];
3160 break;
3161 }
3162 }
3163 if (pd == NULL)
3164 //ferr(po, "label '%s' not parsed?\n", buf1);
3165 goto tailcall;
3166 if (pd->type != OPT_OFFSET)
3167 ferr(po, "label '%s' with non-offset data?\n", buf1);
3168
3169 // find all labels, link
3170 for (j = 0; j < pd->count; j++) {
3171 for (l = 0; l < opcnt; l++) {
3172 if (g_labels[l][0] && IS(g_labels[l], pd->d[j].u.label)) {
3173 add_label_ref(&g_label_refs[l], i);
3174 pd->d[j].bt_i = l;
3175 break;
3176 }
3177 }
3178 }
3179
3180 po->btj = pd;
3181 continue;
3182 }
3183
3184 for (l = 0; l < opcnt; l++) {
3185 if (g_labels[l][0] && IS(po->operand[0].name, g_labels[l])) {
3186 add_label_ref(&g_label_refs[l], i);
3187 po->bt_i = l;
3188 break;
3189 }
3190 }
3191
3192 if (po->bt_i != -1)
3193 continue;
3194
3195 if (po->operand[0].type == OPT_LABEL)
3196 // assume tail call
3197 goto tailcall;
3198
3199 ferr(po, "unhandled branch\n");
3200
3201tailcall:
3202 po->op = OP_CALL;
3203 po->flags |= OPF_TAIL;
3204 if (i > 0 && ops[i - 1].op == OP_POP)
3205 po->flags |= OPF_ATAIL;
3206 i--; // reprocess
3207 }
3208
3209 // pass3:
3210 // - remove dead labels
3211 // - process calls
3212 for (i = 0; i < opcnt; i++)
3213 {
3214 if (g_labels[i][0] != 0 && g_label_refs[i].i == -1)
3215 g_labels[i][0] = 0;
3216
3217 po = &ops[i];
3218 if (po->flags & OPF_RMD)
3219 continue;
3220
3221 if (po->op == OP_CALL)
3222 {
3223 tmpname = opr_name(po, 0);
3224 pp = po->datap;
3225 if (pp == NULL)
3226 {
3227 // indirect call
3228 pp_c = resolve_icall(i, opcnt, &l);
3229 if (pp_c != NULL) {
3230 pp = proto_clone(pp_c);
3231 my_assert_not(pp, NULL);
3232 if (l)
3233 // not resolved just to single func
3234 pp->is_fptr = 1;
3235
3236 switch (po->operand[0].type) {
3237 case OPT_REG:
3238 // we resolved this call and no longer need the register
3239 po->regmask_src &= ~(1 << po->operand[0].reg);
3240 break;
3241 case OPT_REGMEM:
3242 pp->is_fptr = 1;
3243 break;
3244 default:
3245 break;
3246 }
3247 }
3248 if (pp == NULL) {
3249 pp = calloc(1, sizeof(*pp));
3250 my_assert_not(pp, NULL);
3251 pp->is_fptr = 1;
3252 ret = scan_for_esp_adjust(i + 1, opcnt, &j, &l);
3253 if (ret < 0) {
3254 if (!g_allow_regfunc)
3255 ferr(po, "non-__cdecl indirect call unhandled yet\n");
3256 pp->is_unresolved = 1;
3257 j = 0;
3258 }
3259 j /= 4;
3260 if (j > ARRAY_SIZE(pp->arg))
3261 ferr(po, "esp adjust too large: %d\n", j);
3262 pp->ret_type.name = strdup("int");
3263 pp->argc = pp->argc_stack = j;
3264 for (arg = 0; arg < pp->argc; arg++)
3265 pp->arg[arg].type.name = strdup("int");
3266 }
3267 po->datap = pp;
3268 }
3269
3270 // look for and make use of esp adjust
3271 ret = -1;
3272 if (!pp->is_stdcall && pp->argc_stack > 0)
3273 ret = scan_for_esp_adjust(i + 1, opcnt, &j, &l);
3274 if (ret >= 0) {
3275 if (pp->is_vararg) {
3276 if (j / 4 < pp->argc_stack)
3277 ferr(po, "esp adjust is too small: %x < %x\n",
3278 j, pp->argc_stack * 4);
3279 // modify pp to make it have varargs as normal args
3280 arg = pp->argc;
3281 pp->argc += j / 4 - pp->argc_stack;
3282 for (; arg < pp->argc; arg++) {
3283 pp->arg[arg].type.name = strdup("int");
3284 pp->argc_stack++;
3285 }
3286 if (pp->argc > ARRAY_SIZE(pp->arg))
3287 ferr(po, "too many args for '%s'\n", tmpname);
3288 }
3289 if (pp->argc_stack != j / 4)
3290 ferr(po, "stack tracking failed for '%s': %x %x\n",
3291 tmpname, pp->argc_stack * 4, j);
3292
3293 ops[ret].flags |= OPF_RMD;
3294 if (ops[ret].op == OP_POP && j > 4) {
3295 // deal with multi-pop stack adjust
3296 j = pp->argc_stack;
3297 while (ops[ret].op == OP_POP && j > 0 && ret < opcnt) {
3298 ops[ret].flags |= OPF_RMD;
3299 j--;
3300 ret++;
3301 }
3302 }
3303 else if (!l) {
3304 // a bit of a hack, but deals with use of
3305 // single adj for multiple calls
3306 ops[ret].operand[1].val -= j;
3307 }
3308 }
3309 else if (pp->is_vararg)
3310 ferr(po, "missing esp_adjust for vararg func '%s'\n",
3311 pp->name);
3312
3313 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
3314 // since we know the args, collect them
3315 collect_call_args(po, i, pp, &regmask, &save_arg_vars,
3316 i + opcnt * 2);
3317 }
3318
3319 if (strstr(pp->ret_type.name, "int64"))
3320 need_tmp64 = 1;
3321 }
3322 }
3323
3324 // pass4:
3325 // - find POPs for PUSHes, rm both
3326 // - scan for STD/CLD, propagate DF
3327 // - scan for all used registers
3328 // - find flag set ops for their users
3329 // - do unreselved calls
3330 // - declare indirect functions
3331 for (i = 0; i < opcnt; i++) {
3332 po = &ops[i];
3333 if (po->flags & OPF_RMD)
3334 continue;
3335
3336 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE)) {
3337 reg = po->operand[0].reg;
3338 if (!(regmask & (1 << reg)))
3339 // not a reg save after all, rerun scan_for_pop
3340 po->flags &= ~OPF_RSAVE;
3341 else
3342 regmask_save |= 1 << reg;
3343 }
3344
3345 if (po->op == OP_PUSH && po->argnum == 0
3346 && !(po->flags & OPF_RSAVE) && !g_func_pp->is_userstack)
3347 {
3348 if (po->operand[0].type == OPT_REG)
3349 {
3350 reg = po->operand[0].reg;
3351 if (reg < 0)
3352 ferr(po, "reg not set for push?\n");
3353
3354 depth = 0;
3355 ret = scan_for_pop(i + 1, opcnt,
3356 po->operand[0].name, i + opcnt * 3, 0, &depth, 0);
3357 if (ret == 1) {
3358 if (depth > 1)
3359 ferr(po, "too much depth: %d\n", depth);
3360
3361 po->flags |= OPF_RMD;
3362 scan_for_pop(i + 1, opcnt, po->operand[0].name,
3363 i + opcnt * 4, 0, &depth, 1);
3364 continue;
3365 }
3366 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
3367 if (ret == 0) {
3368 arg = OPF_RMD;
3369 if (regmask & (1 << reg)) {
3370 if (regmask_save & (1 << reg))
3371 ferr(po, "%s already saved?\n", po->operand[0].name);
3372 arg = OPF_RSAVE;
3373 }
3374 po->flags |= arg;
3375 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg);
3376 continue;
3377 }
3378 }
3379 else if (po->operand[0].type == OPT_CONST) {
3380 for (j = i + 1; j < opcnt; j++) {
3381 if ((ops[j].flags & (OPF_JMP|OPF_TAIL|OPF_RSAVE))
3382 || ops[j].op == OP_PUSH || g_labels[i][0] != 0)
3383 {
3384 break;
3385 }
3386
3387 if (!(ops[j].flags & OPF_RMD) && ops[j].op == OP_POP)
3388 {
3389 po->flags |= OPF_RMD;
3390 ops[j].datap = po;
3391 break;
3392 }
3393 }
3394 }
3395 }
3396
3397 if (po->op == OP_STD) {
3398 po->flags |= OPF_DF | OPF_RMD;
3399 scan_propagate_df(i + 1, opcnt);
3400 }
3401
3402 regmask_now = po->regmask_src | po->regmask_dst;
3403 if (regmask_now & (1 << xBP)) {
3404 if (g_bp_frame && !(po->flags & OPF_EBP_S)) {
3405 if (po->regmask_dst & (1 << xBP))
3406 // compiler decided to drop bp frame and use ebp as scratch
3407 scan_fwd_set_flags(i + 1, opcnt, i + opcnt * 5, OPF_EBP_S);
3408 else
3409 regmask_now &= ~(1 << xBP);
3410 }
3411 }
3412
3413 regmask |= regmask_now;
3414
3415 if (po->flags & OPF_CC)
3416 {
3417 int setters[16], cnt = 0, branched = 0;
3418
3419 ret = scan_for_flag_set(i, i + opcnt * 6,
3420 &branched, setters, &cnt);
3421 if (ret < 0 || cnt <= 0)
3422 ferr(po, "unable to trace flag setter(s)\n");
3423 if (cnt > ARRAY_SIZE(setters))
3424 ferr(po, "too many flag setters\n");
3425
3426 pfo = split_cond(po, po->op, &dummy);
3427 for (j = 0; j < cnt; j++)
3428 {
3429 tmp_op = &ops[setters[j]]; // flag setter
3430 pfomask = 0;
3431
3432 // to get nicer code, we try to delay test and cmp;
3433 // if we can't because of operand modification, or if we
3434 // have arith op, or branch, make it calculate flags explicitly
3435 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP)
3436 {
3437 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i, 0) >= 0)
3438 pfomask = 1 << pfo;
3439 }
3440 else if (tmp_op->op == OP_CMPS || tmp_op->op == OP_SCAS) {
3441 pfomask = 1 << pfo;
3442 }
3443 else {
3444 // see if we'll be able to handle based on op result
3445 if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR
3446 && pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P)
3447 || branched
3448 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
3449 pfomask = 1 << pfo;
3450
3451 if (tmp_op->op == OP_ADD && pfo == PFO_C)
3452 need_tmp64 = 1;
3453 }
3454 if (pfomask) {
3455 tmp_op->pfomask |= pfomask;
3456 cond_vars |= pfomask;
3457 }
3458 // note: may overwrite, currently not a problem
3459 po->datap = tmp_op;
3460 }
3461
3462 if (po->op == OP_RCL || po->op == OP_RCR
3463 || po->op == OP_ADC || po->op == OP_SBB)
3464 cond_vars |= 1 << PFO_C;
3465 }
3466 else if (po->op == OP_CMPS || po->op == OP_SCAS) {
3467 cond_vars |= 1 << PFO_Z;
3468 }
3469 else if (po->op == OP_MUL
3470 || (po->op == OP_IMUL && po->operand_cnt == 1))
3471 {
3472 need_tmp64 = 1;
3473 }
3474 else if (po->op == OP_CALL) {
3475 pp = po->datap;
3476 if (pp == NULL)
3477 ferr(po, "NULL pp\n");
3478
3479 if (pp->is_unresolved) {
3480 int regmask_stack = 0;
3481 collect_call_args(po, i, pp, &regmask, &save_arg_vars,
3482 i + opcnt * 2);
3483
3484 // this is pretty rough guess:
3485 // see ecx and edx were pushed (and not their saved versions)
3486 for (arg = 0; arg < pp->argc; arg++) {
3487 if (pp->arg[arg].reg != NULL)
3488 continue;
3489
3490 tmp_op = pp->arg[arg].datap;
3491 if (tmp_op == NULL)
3492 ferr(po, "parsed_op missing for arg%d\n", arg);
3493 if (tmp_op->argnum == 0 && tmp_op->operand[0].type == OPT_REG)
3494 regmask_stack |= 1 << tmp_op->operand[0].reg;
3495 }
3496
3497 if (!((regmask_stack & (1 << xCX))
3498 && (regmask_stack & (1 << xDX))))
3499 {
3500 if (pp->argc_stack != 0
3501 || ((regmask | regmask_arg) & ((1 << xCX)|(1 << xDX))))
3502 {
3503 pp_insert_reg_arg(pp, "ecx");
3504 pp->is_fastcall = 1;
3505 regmask_init |= 1 << xCX;
3506 regmask |= 1 << xCX;
3507 }
3508 if (pp->argc_stack != 0
3509 || ((regmask | regmask_arg) & (1 << xDX)))
3510 {
3511 pp_insert_reg_arg(pp, "edx");
3512 regmask_init |= 1 << xDX;
3513 regmask |= 1 << xDX;
3514 }
3515 }
3516
3517 // note: __cdecl doesn't fall into is_unresolved category
3518 if (pp->argc_stack > 0)
3519 pp->is_stdcall = 1;
3520 }
3521
3522 for (arg = 0; arg < pp->argc; arg++) {
3523 if (pp->arg[arg].reg != NULL) {
3524 reg = char_array_i(regs_r32,
3525 ARRAY_SIZE(regs_r32), pp->arg[arg].reg);
3526 if (reg < 0)
3527 ferr(ops, "arg '%s' is not a reg?\n", pp->arg[arg].reg);
3528 if (!(regmask & (1 << reg))) {
3529 regmask_init |= 1 << reg;
3530 regmask |= 1 << reg;
3531 }
3532 }
3533 }
3534
3535 // declare indirect funcs
3536 if (pp->is_fptr && !(pp->name[0] != 0 && pp->is_arg)) {
3537 if (pp->name[0] != 0) {
3538 memmove(pp->name + 2, pp->name, strlen(pp->name) + 1);
3539 memcpy(pp->name, "i_", 2);
3540
3541 // might be declared already
3542 found = 0;
3543 for (j = 0; j < i; j++) {
3544 if (ops[j].op == OP_CALL && (pp_tmp = ops[j].datap)) {
3545 if (pp_tmp->is_fptr && IS(pp->name, pp_tmp->name)) {
3546 found = 1;
3547 break;
3548 }
3549 }
3550 }
3551 if (found)
3552 continue;
3553 }
3554 else
3555 snprintf(pp->name, sizeof(pp->name), "icall%d", i);
3556
3557 fprintf(fout, " %s (", pp->ret_type.name);
3558 output_pp_attrs(fout, pp, 0);
3559 fprintf(fout, "*%s)(", pp->name);
3560 for (j = 0; j < pp->argc; j++) {
3561 if (j > 0)
3562 fprintf(fout, ", ");
3563 fprintf(fout, "%s a%d", pp->arg[j].type.name, j + 1);
3564 }
3565 fprintf(fout, ");\n");
3566 }
3567 }
3568 else if (po->op == OP_RET && !IS(g_func_pp->ret_type.name, "void"))
3569 regmask |= 1 << xAX;
3570 else if (po->op == OP_DIV || po->op == OP_IDIV) {
3571 // 32bit division is common, look for it
3572 if (po->op == OP_DIV)
3573 ret = scan_for_reg_clear(i, xDX);
3574 else
3575 ret = scan_for_cdq_edx(i);
3576 if (ret >= 0)
3577 po->flags |= OPF_32BIT;
3578 else
3579 need_tmp64 = 1;
3580 }
3581 else if (po->op == OP_CLD)
3582 po->flags |= OPF_RMD;
3583
3584 if (po->op == OP_RCL || po->op == OP_RCR || po->op == OP_XCHG) {
3585 need_tmp_var = 1;
3586 }
3587 }
3588
3589 // pass4:
3590 // - confirm regmask_save, it might have been reduced
3591 if (regmask_save != 0)
3592 {
3593 regmask_save = 0;
3594 for (i = 0; i < opcnt; i++) {
3595 po = &ops[i];
3596 if (po->flags & OPF_RMD)
3597 continue;
3598
3599 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE))
3600 regmask_save |= 1 << po->operand[0].reg;
3601 }
3602 }
3603
3604
3605 // output LUTs/jumptables
3606 for (i = 0; i < g_func_pd_cnt; i++) {
3607 pd = &g_func_pd[i];
3608 fprintf(fout, " static const ");
3609 if (pd->type == OPT_OFFSET) {
3610 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
3611
3612 for (j = 0; j < pd->count; j++) {
3613 if (j > 0)
3614 fprintf(fout, ", ");
3615 fprintf(fout, "&&%s", pd->d[j].u.label);
3616 }
3617 }
3618 else {
3619 fprintf(fout, "%s %s[] =\n { ",
3620 lmod_type_u(ops, pd->lmod), pd->label);
3621
3622 for (j = 0; j < pd->count; j++) {
3623 if (j > 0)
3624 fprintf(fout, ", ");
3625 fprintf(fout, "%u", pd->d[j].u.val);
3626 }
3627 }
3628 fprintf(fout, " };\n");
3629 had_decl = 1;
3630 }
3631
3632 // declare stack frame, va_arg
3633 if (g_stack_fsz) {
3634 fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
3635 (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz);
3636 had_decl = 1;
3637 }
3638
3639 if (g_func_pp->is_userstack) {
3640 fprintf(fout, " u32 fake_sf[1024];\n");
3641 fprintf(fout, " u32 *esp = &fake_sf[1024];\n");
3642 had_decl = 1;
3643 }
3644
3645 if (g_func_pp->is_vararg) {
3646 fprintf(fout, " va_list ap;\n");
3647 had_decl = 1;
3648 }
3649
3650 // declare arg-registers
3651 for (i = 0; i < g_func_pp->argc; i++) {
3652 if (g_func_pp->arg[i].reg != NULL) {
3653 reg = char_array_i(regs_r32,
3654 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
3655 if (regmask & (1 << reg)) {
3656 if (g_func_pp->arg[i].type.is_retreg)
3657 fprintf(fout, " u32 %s = *r_%s;\n",
3658 g_func_pp->arg[i].reg, g_func_pp->arg[i].reg);
3659 else
3660 fprintf(fout, " u32 %s = (u32)a%d;\n",
3661 g_func_pp->arg[i].reg, i + 1);
3662 }
3663 else {
3664 if (g_func_pp->arg[i].type.is_retreg)
3665 ferr(ops, "retreg '%s' is unused?\n",
3666 g_func_pp->arg[i].reg);
3667 fprintf(fout, " // %s = a%d; // unused\n",
3668 g_func_pp->arg[i].reg, i + 1);
3669 }
3670 had_decl = 1;
3671 }
3672 }
3673
3674 regmask_now = regmask & ~regmask_arg;
3675 regmask_now &= ~(1 << xSP);
3676 if (regmask_now) {
3677 for (reg = 0; reg < 8; reg++) {
3678 if (regmask_now & (1 << reg)) {
3679 fprintf(fout, " u32 %s", regs_r32[reg]);
3680 if (regmask_init & (1 << reg))
3681 fprintf(fout, " = 0");
3682 fprintf(fout, ";\n");
3683 had_decl = 1;
3684 }
3685 }
3686 }
3687
3688 if (regmask_save) {
3689 for (reg = 0; reg < 8; reg++) {
3690 if (regmask_save & (1 << reg)) {
3691 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
3692 had_decl = 1;
3693 }
3694 }
3695 }
3696
3697 if (save_arg_vars) {
3698 for (reg = 0; reg < 32; reg++) {
3699 if (save_arg_vars & (1 << reg)) {
3700 fprintf(fout, " u32 s_a%d;\n", reg + 1);
3701 had_decl = 1;
3702 }
3703 }
3704 }
3705
3706 if (cond_vars) {
3707 for (i = 0; i < 8; i++) {
3708 if (cond_vars & (1 << i)) {
3709 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
3710 had_decl = 1;
3711 }
3712 }
3713 }
3714
3715 if (need_tmp_var) {
3716 fprintf(fout, " u32 tmp;\n");
3717 had_decl = 1;
3718 }
3719
3720 if (need_tmp64) {
3721 fprintf(fout, " u64 tmp64;\n");
3722 had_decl = 1;
3723 }
3724
3725 if (had_decl)
3726 fprintf(fout, "\n");
3727
3728 if (g_func_pp->is_vararg) {
3729 if (g_func_pp->argc_stack == 0)
3730 ferr(ops, "vararg func without stack args?\n");
3731 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
3732 }
3733
3734 // output ops
3735 for (i = 0; i < opcnt; i++)
3736 {
3737 if (g_labels[i][0] != 0) {
3738 fprintf(fout, "\n%s:\n", g_labels[i]);
3739 label_pending = 1;
3740
3741 delayed_flag_op = NULL;
3742 last_arith_dst = NULL;
3743 }
3744
3745 po = &ops[i];
3746 if (po->flags & OPF_RMD)
3747 continue;
3748
3749 no_output = 0;
3750
3751 #define assert_operand_cnt(n_) \
3752 if (po->operand_cnt != n_) \
3753 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
3754
3755 // conditional/flag using op?
3756 if (po->flags & OPF_CC)
3757 {
3758 int is_delayed = 0;
3759 int is_inv = 0;
3760
3761 pfo = split_cond(po, po->op, &is_inv);
3762 tmp_op = po->datap;
3763
3764 // we go through all this trouble to avoid using parsed_flag_op,
3765 // which makes generated code much nicer
3766 if (delayed_flag_op != NULL)
3767 {
3768 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op, pfo, is_inv);
3769 is_delayed = 1;
3770 }
3771 else if (last_arith_dst != NULL
3772 && (pfo == PFO_Z || pfo == PFO_S || pfo == PFO_P
3773 || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR))
3774 ))
3775 {
3776 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
3777 out_test_for_cc(buf1, sizeof(buf1), po, pfo, is_inv,
3778 last_arith_dst->lmod, buf3);
3779 is_delayed = 1;
3780 }
3781 else if (tmp_op != NULL) {
3782 // use preprocessed flag calc results
3783 if (!(tmp_op->pfomask & (1 << pfo)))
3784 ferr(po, "not prepared for pfo %d\n", pfo);
3785
3786 // note: is_inv was not yet applied
3787 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
3788 is_inv ? "!" : "", parsed_flag_op_names[pfo]);
3789 }
3790 else {
3791 ferr(po, "all methods of finding comparison failed\n");
3792 }
3793
3794 if (po->flags & OPF_JMP) {
3795 fprintf(fout, " if %s\n", buf1);
3796 }
3797 else if (po->op == OP_RCL || po->op == OP_RCR
3798 || po->op == OP_ADC || po->op == OP_SBB)
3799 {
3800 if (is_delayed)
3801 fprintf(fout, " cond_%s = %s;\n",
3802 parsed_flag_op_names[pfo], buf1);
3803 }
3804 else if (po->flags & OPF_DATA) { // SETcc
3805 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
3806 fprintf(fout, " %s = %s;", buf2, buf1);
3807 }
3808 else {
3809 ferr(po, "unhandled conditional op\n");
3810 }
3811 }
3812
3813 pfomask = po->pfomask;
3814
3815 if (po->flags & (OPF_REPZ|OPF_REPNZ)) {
3816 struct parsed_opr opr = {0,};
3817 opr.type = OPT_REG;
3818 opr.reg = xCX;
3819 opr.lmod = OPLM_DWORD;
3820 ret = try_resolve_const(i, &opr, opcnt * 7 + i, &uval);
3821
3822 if (ret != 1 || uval == 0) {
3823 // we need initial flags for ecx=0 case..
3824 if (i > 0 && ops[i - 1].op == OP_XOR
3825 && IS(ops[i - 1].operand[0].name,
3826 ops[i - 1].operand[1].name))
3827 {
3828 fprintf(fout, " cond_z = ");
3829 if (pfomask & (1 << PFO_C))
3830 fprintf(fout, "cond_c = ");
3831 fprintf(fout, "0;\n");
3832 }
3833 else if (last_arith_dst != NULL) {
3834 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
3835 out_test_for_cc(buf1, sizeof(buf1), po, PFO_Z, 0,
3836 last_arith_dst->lmod, buf3);
3837 fprintf(fout, " cond_z = %s;\n", buf1);
3838 }
3839 else
3840 ferr(po, "missing initial ZF\n");
3841 }
3842 }
3843
3844 switch (po->op)
3845 {
3846 case OP_MOV:
3847 assert_operand_cnt(2);
3848 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3849 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3850 fprintf(fout, " %s = %s;", buf1,
3851 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
3852 po->operand[0].is_ptr ? "(void *)" : "", 0));
3853 break;
3854
3855 case OP_LEA:
3856 assert_operand_cnt(2);
3857 po->operand[1].lmod = OPLM_DWORD; // always
3858 fprintf(fout, " %s = %s;",
3859 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3860 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
3861 NULL, 1));
3862 break;
3863
3864 case OP_MOVZX:
3865 assert_operand_cnt(2);
3866 fprintf(fout, " %s = %s;",
3867 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3868 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
3869 break;
3870
3871 case OP_MOVSX:
3872 assert_operand_cnt(2);
3873 switch (po->operand[1].lmod) {
3874 case OPLM_BYTE:
3875 strcpy(buf3, "(s8)");
3876 break;
3877 case OPLM_WORD:
3878 strcpy(buf3, "(s16)");
3879 break;
3880 default:
3881 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
3882 }
3883 fprintf(fout, " %s = %s;",
3884 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3885 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
3886 buf3, 0));
3887 break;
3888
3889 case OP_XCHG:
3890 assert_operand_cnt(2);
3891 propagate_lmod(po, &po->operand[0], &po->operand[1]);
3892 fprintf(fout, " tmp = %s;",
3893 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], "", 0));
3894 fprintf(fout, " %s = %s;",
3895 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3896 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], "", 0));
3897 fprintf(fout, " %s = tmp;",
3898 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]));
3899 snprintf(g_comment, sizeof(g_comment), "xchg");
3900 break;
3901
3902 case OP_NOT:
3903 assert_operand_cnt(1);
3904 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
3905 fprintf(fout, " %s = ~%s;", buf1, buf1);
3906 break;
3907
3908 case OP_CDQ:
3909 assert_operand_cnt(2);
3910 fprintf(fout, " %s = (s32)%s >> 31;",
3911 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
3912 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
3913 strcpy(g_comment, "cdq");
3914 break;
3915
3916 case OP_STOS:
3917 assert_operand_cnt(3);
3918 if (po->flags & OPF_REP) {
3919 fprintf(fout, " for (; ecx != 0; ecx--, edi %c= %d)\n",
3920 (po->flags & OPF_DF) ? '-' : '+',
3921 lmod_bytes(po, po->operand[0].lmod));
3922 fprintf(fout, " %sedi = eax;",
3923 lmod_cast_u_ptr(po, po->operand[0].lmod));
3924 strcpy(g_comment, "rep stos");
3925 }
3926 else {
3927 fprintf(fout, " %sedi = eax; edi %c= %d;",
3928 lmod_cast_u_ptr(po, po->operand[0].lmod),
3929 (po->flags & OPF_DF) ? '-' : '+',
3930 lmod_bytes(po, po->operand[0].lmod));
3931 strcpy(g_comment, "stos");
3932 }
3933 break;
3934
3935 case OP_MOVS:
3936 assert_operand_cnt(3);
3937 j = lmod_bytes(po, po->operand[0].lmod);
3938 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
3939 l = (po->flags & OPF_DF) ? '-' : '+';
3940 if (po->flags & OPF_REP) {
3941 fprintf(fout,
3942 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n",
3943 l, j, l, j);
3944 fprintf(fout,
3945 " %sedi = %sesi;", buf1, buf1);
3946 strcpy(g_comment, "rep movs");
3947 }
3948 else {
3949 fprintf(fout, " %sedi = %sesi; edi %c= %d; esi %c= %d;",
3950 buf1, buf1, l, j, l, j);
3951 strcpy(g_comment, "movs");
3952 }
3953 break;
3954
3955 case OP_CMPS:
3956 // repe ~ repeat while ZF=1
3957 assert_operand_cnt(3);
3958 j = lmod_bytes(po, po->operand[0].lmod);
3959 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
3960 l = (po->flags & OPF_DF) ? '-' : '+';
3961 if (po->flags & OPF_REP) {
3962 fprintf(fout,
3963 " for (; ecx != 0; ecx--) {\n");
3964 if (pfomask & (1 << PFO_C)) {
3965 // ugh..
3966 fprintf(fout,
3967 " cond_c = %sedi < %sesi;\n", buf1, buf1);
3968 pfomask &= ~(1 << PFO_C);
3969 }
3970 fprintf(fout,
3971 " cond_z = (%sedi == %sesi); edi %c= %d, esi %c= %d;\n",
3972 buf1, buf1, l, j, l, j);
3973 fprintf(fout,
3974 " if (cond_z %s 0) break;\n",
3975 (po->flags & OPF_REPZ) ? "==" : "!=");
3976 fprintf(fout,
3977 " }");
3978 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
3979 (po->flags & OPF_REPZ) ? "e" : "ne");
3980 }
3981 else {
3982 fprintf(fout,
3983 " cond_z = (%sedi = %sesi); edi %c= %d; esi %c= %d;",
3984 buf1, buf1, l, j, l, j);
3985 strcpy(g_comment, "cmps");
3986 }
3987 pfomask &= ~(1 << PFO_Z);
3988 last_arith_dst = NULL;
3989 delayed_flag_op = NULL;
3990 break;
3991
3992 case OP_SCAS:
3993 // only does ZF (for now)
3994 // repe ~ repeat while ZF=1
3995 assert_operand_cnt(3);
3996 j = lmod_bytes(po, po->operand[0].lmod);
3997 l = (po->flags & OPF_DF) ? '-' : '+';
3998 if (po->flags & OPF_REP) {
3999 fprintf(fout,
4000 " for (; ecx != 0; ecx--) {\n");
4001 fprintf(fout,
4002 " cond_z = (%seax == %sedi); edi %c= %d;\n",
4003 lmod_cast_u(po, po->operand[0].lmod),
4004 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
4005 fprintf(fout,
4006 " if (cond_z %s 0) break;\n",
4007 (po->flags & OPF_REPZ) ? "==" : "!=");
4008 fprintf(fout,
4009 " }");
4010 snprintf(g_comment, sizeof(g_comment), "rep%s scas",
4011 (po->flags & OPF_REPZ) ? "e" : "ne");
4012 }
4013 else {
4014 fprintf(fout, " cond_z = (%seax = %sedi); edi %c= %d;",
4015 lmod_cast_u(po, po->operand[0].lmod),
4016 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
4017 strcpy(g_comment, "scas");
4018 }
4019 pfomask &= ~(1 << PFO_Z);
4020 last_arith_dst = NULL;
4021 delayed_flag_op = NULL;
4022 break;
4023
4024 // arithmetic w/flags
4025 case OP_AND:
4026 case OP_OR:
4027 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4028 // fallthrough
4029 dualop_arith:
4030 assert_operand_cnt(2);
4031 fprintf(fout, " %s %s= %s;",
4032 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4033 op_to_c(po),
4034 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4035 output_std_flags(fout, po, &pfomask, buf1);
4036 last_arith_dst = &po->operand[0];
4037 delayed_flag_op = NULL;
4038 break;
4039
4040 case OP_SHL:
4041 case OP_SHR:
4042 assert_operand_cnt(2);
4043 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4044 if (pfomask & (1 << PFO_C)) {
4045 if (po->operand[1].type == OPT_CONST) {
4046 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4047 j = po->operand[1].val;
4048 j %= l;
4049 if (j != 0) {
4050 if (po->op == OP_SHL)
4051 j = l - j;
4052 else
4053 j -= 1;
4054 fprintf(fout, " cond_c = (%s >> %d) & 1;\n",
4055 buf1, j);
4056 }
4057 else
4058 ferr(po, "zero shift?\n");
4059 }
4060 else
4061 ferr(po, "TODO\n");
4062 pfomask &= ~(1 << PFO_C);
4063 }
4064 fprintf(fout, " %s %s= %s;", buf1, op_to_c(po),
4065 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4066 output_std_flags(fout, po, &pfomask, buf1);
4067 last_arith_dst = &po->operand[0];
4068 delayed_flag_op = NULL;
4069 break;
4070
4071 case OP_SAR:
4072 assert_operand_cnt(2);
4073 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4074 fprintf(fout, " %s = %s%s >> %s;", buf1,
4075 lmod_cast_s(po, po->operand[0].lmod), buf1,
4076 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4077 output_std_flags(fout, po, &pfomask, buf1);
4078 last_arith_dst = &po->operand[0];
4079 delayed_flag_op = NULL;
4080 break;
4081
4082 case OP_ROL:
4083 case OP_ROR:
4084 assert_operand_cnt(2);
4085 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4086 if (po->operand[1].type == OPT_CONST) {
4087 j = po->operand[1].val;
4088 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
4089 fprintf(fout, po->op == OP_ROL ?
4090 " %s = (%s << %d) | (%s >> %d);" :
4091 " %s = (%s >> %d) | (%s << %d);",
4092 buf1, buf1, j, buf1,
4093 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
4094 }
4095 else
4096 ferr(po, "TODO\n");
4097 output_std_flags(fout, po, &pfomask, buf1);
4098 last_arith_dst = &po->operand[0];
4099 delayed_flag_op = NULL;
4100 break;
4101
4102 case OP_RCL:
4103 case OP_RCR:
4104 assert_operand_cnt(2);
4105 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4106 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4107 if (po->operand[1].type == OPT_CONST) {
4108 j = po->operand[1].val % l;
4109 if (j == 0)
4110 ferr(po, "zero rotate\n");
4111 fprintf(fout, " tmp = (%s >> %d) & 1;\n",
4112 buf1, (po->op == OP_RCL) ? (l - j) : (j - 1));
4113 if (po->op == OP_RCL) {
4114 fprintf(fout,
4115 " %s = (%s << %d) | (cond_c << %d)",
4116 buf1, buf1, j, j - 1);
4117 if (j != 1)
4118 fprintf(fout, " | (%s >> %d)", buf1, l + 1 - j);
4119 }
4120 else {
4121 fprintf(fout,
4122 " %s = (%s >> %d) | (cond_c << %d)",
4123 buf1, buf1, j, l - j);
4124 if (j != 1)
4125 fprintf(fout, " | (%s << %d)", buf1, l + 1 - j);
4126 }
4127 fprintf(fout, ";\n");
4128 fprintf(fout, " cond_c = tmp;");
4129 }
4130 else
4131 ferr(po, "TODO\n");
4132 strcpy(g_comment, (po->op == OP_RCL) ? "rcl" : "rcr");
4133 output_std_flags(fout, po, &pfomask, buf1);
4134 last_arith_dst = &po->operand[0];
4135 delayed_flag_op = NULL;
4136 break;
4137
4138 case OP_XOR:
4139 assert_operand_cnt(2);
4140 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4141 if (IS(opr_name(po, 0), opr_name(po, 1))) {
4142 // special case for XOR
4143 if (pfomask & (1 << PFO_BE)) { // weird, but it happens..
4144 fprintf(fout, " cond_be = 1;\n");
4145 pfomask &= ~(1 << PFO_BE);
4146 }
4147 fprintf(fout, " %s = 0;",
4148 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4149 last_arith_dst = &po->operand[0];
4150 delayed_flag_op = NULL;
4151 break;
4152 }
4153 goto dualop_arith;
4154
4155 case OP_ADD:
4156 assert_operand_cnt(2);
4157 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4158 if (pfomask & (1 << PFO_C)) {
4159 fprintf(fout, " tmp64 = (u64)%s + %s;\n",
4160 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]),
4161 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4162 fprintf(fout, " cond_c = tmp64 >> 32;\n");
4163 fprintf(fout, " %s = (u32)tmp64;",
4164 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4165 strcat(g_comment, "add64");
4166 pfomask &= ~(1 << PFO_C);
4167 output_std_flags(fout, po, &pfomask, buf1);
4168 last_arith_dst = &po->operand[0];
4169 delayed_flag_op = NULL;
4170 break;
4171 }
4172 goto dualop_arith;
4173
4174 case OP_SUB:
4175 assert_operand_cnt(2);
4176 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4177 if (pfomask & (1 << PFO_C)) {
4178 fprintf(fout, " cond_c = %s < %s;\n",
4179 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]),
4180 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4181 pfomask &= ~(1 << PFO_C);
4182 }
4183 goto dualop_arith;
4184
4185 case OP_ADC:
4186 case OP_SBB:
4187 assert_operand_cnt(2);
4188 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4189 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4190 if (po->op == OP_SBB
4191 && IS(po->operand[0].name, po->operand[1].name))
4192 {
4193 // avoid use of unitialized var
4194 fprintf(fout, " %s = -cond_c;", buf1);
4195 // carry remains what it was
4196 pfomask &= ~(1 << PFO_C);
4197 }
4198 else {
4199 fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po),
4200 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4201 }
4202 output_std_flags(fout, po, &pfomask, buf1);
4203 last_arith_dst = &po->operand[0];
4204 delayed_flag_op = NULL;
4205 break;
4206
4207 case OP_BSF:
4208 assert_operand_cnt(2);
4209 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4210 fprintf(fout, " %s = %s ? __builtin_ffs(%s) - 1 : 0;",
4211 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4212 buf2, buf2);
4213 output_std_flags(fout, po, &pfomask, buf1);
4214 last_arith_dst = &po->operand[0];
4215 delayed_flag_op = NULL;
4216 strcat(g_comment, "bsf");
4217 break;
4218
4219 case OP_INC:
4220 case OP_DEC:
4221 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4222 if (po->operand[0].type == OPT_REG) {
4223 strcpy(buf2, po->op == OP_INC ? "++" : "--");
4224 fprintf(fout, " %s%s;", buf1, buf2);
4225 }
4226 else {
4227 strcpy(buf2, po->op == OP_INC ? "+" : "-");
4228 fprintf(fout, " %s %s= 1;", buf1, buf2);
4229 }
4230 output_std_flags(fout, po, &pfomask, buf1);
4231 last_arith_dst = &po->operand[0];
4232 delayed_flag_op = NULL;
4233 break;
4234
4235 case OP_NEG:
4236 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4237 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
4238 fprintf(fout, " %s = -%s%s;", buf1,
4239 lmod_cast_s(po, po->operand[0].lmod), buf2);
4240 last_arith_dst = &po->operand[0];
4241 delayed_flag_op = NULL;
4242 if (pfomask & (1 << PFO_C)) {
4243 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
4244 pfomask &= ~(1 << PFO_C);
4245 }
4246 break;
4247
4248 case OP_IMUL:
4249 if (po->operand_cnt == 2) {
4250 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4251 goto dualop_arith;
4252 }
4253 if (po->operand_cnt == 3)
4254 ferr(po, "TODO imul3\n");
4255 // fallthrough
4256 case OP_MUL:
4257 assert_operand_cnt(1);
4258 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
4259 fprintf(fout, " tmp64 = %seax * %s%s;\n", buf1, buf1,
4260 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
4261 fprintf(fout, " edx = tmp64 >> 32;\n");
4262 fprintf(fout, " eax = tmp64;");
4263 last_arith_dst = NULL;
4264 delayed_flag_op = NULL;
4265 break;
4266
4267 case OP_DIV:
4268 case OP_IDIV:
4269 assert_operand_cnt(1);
4270 if (po->operand[0].lmod != OPLM_DWORD)
4271 ferr(po, "unhandled lmod %d\n", po->operand[0].lmod);
4272
4273 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4274 strcpy(buf2, lmod_cast(po, po->operand[0].lmod,
4275 po->op == OP_IDIV));
4276 switch (po->operand[0].lmod) {
4277 case OPLM_DWORD:
4278 if (po->flags & OPF_32BIT)
4279 snprintf(buf3, sizeof(buf3), "%seax", buf2);
4280 else {
4281 fprintf(fout, " tmp64 = ((u64)edx << 32) | eax;\n");
4282 snprintf(buf3, sizeof(buf3), "%stmp64",
4283 (po->op == OP_IDIV) ? "(s64)" : "");
4284 }
4285 if (po->operand[0].type == OPT_REG
4286 && po->operand[0].reg == xDX)
4287 {
4288 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
4289 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
4290 }
4291 else {
4292 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
4293 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
4294 }
4295 break;
4296 default:
4297 ferr(po, "unhandled division type\n");
4298 }
4299 last_arith_dst = NULL;
4300 delayed_flag_op = NULL;
4301 break;
4302
4303 case OP_TEST:
4304 case OP_CMP:
4305 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4306 if (pfomask != 0) {
4307 for (j = 0; j < 8; j++) {
4308 if (pfomask & (1 << j)) {
4309 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
4310 fprintf(fout, " cond_%s = %s;",
4311 parsed_flag_op_names[j], buf1);
4312 }
4313 }
4314 pfomask = 0;
4315 }
4316 else
4317 no_output = 1;
4318 last_arith_dst = NULL;
4319 delayed_flag_op = po;
4320 break;
4321
4322 // note: we reuse OP_Jcc for SETcc, only flags differ
4323 case OP_JO ... OP_JG:
4324 if (po->flags & OPF_JMP)
4325 fprintf(fout, " goto %s;", po->operand[0].name);
4326 // else SETcc - should already be handled
4327 break;
4328
4329 case OP_JECXZ:
4330 fprintf(fout, " if (ecx == 0)\n");
4331 fprintf(fout, " goto %s;", po->operand[0].name);
4332 strcat(g_comment, "jecxz");
4333 break;
4334
4335 case OP_JMP:
4336 assert_operand_cnt(1);
4337 last_arith_dst = NULL;
4338 delayed_flag_op = NULL;
4339
4340 if (po->operand[0].type == OPT_REGMEM) {
4341 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
4342 buf1, buf2);
4343 if (ret != 2)
4344 ferr(po, "parse failure for jmp '%s'\n",
4345 po->operand[0].name);
4346 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
4347 break;
4348 }
4349 else if (po->operand[0].type != OPT_LABEL)
4350 ferr(po, "unhandled jmp type\n");
4351
4352 fprintf(fout, " goto %s;", po->operand[0].name);
4353 break;
4354
4355 case OP_CALL:
4356 assert_operand_cnt(1);
4357 pp = po->datap;
4358 my_assert_not(pp, NULL);
4359
4360 if (pp->is_fptr && !pp->is_arg)
4361 fprintf(fout, " %s = %s;\n", pp->name,
4362 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
4363 "(void *)", 0));
4364
4365 fprintf(fout, " ");
4366 if (strstr(pp->ret_type.name, "int64")) {
4367 if (po->flags & OPF_TAIL)
4368 ferr(po, "int64 and tail?\n");
4369 fprintf(fout, "tmp64 = ");
4370 }
4371 else if (!IS(pp->ret_type.name, "void")) {
4372 if (po->flags & OPF_TAIL) {
4373 if (!IS(g_func_pp->ret_type.name, "void")) {
4374 fprintf(fout, "return ");
4375 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
4376 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
4377 }
4378 }
4379 else if (regmask & (1 << xAX)) {
4380 fprintf(fout, "eax = ");
4381 if (pp->ret_type.is_ptr)
4382 fprintf(fout, "(u32)");
4383 }
4384 }
4385
4386 if (pp->name[0] == 0)
4387 ferr(po, "missing pp->name\n");
4388 fprintf(fout, "%s%s(", pp->name,
4389 pp->has_structarg ? "_sa" : "");
4390
4391 if (po->flags & OPF_ATAIL) {
4392 if (pp->argc_stack != g_func_pp->argc_stack
4393 || (pp->argc_stack > 0
4394 && pp->is_stdcall != g_func_pp->is_stdcall))
4395 ferr(po, "incompatible tailcall\n");
4396 if (g_func_pp->has_retreg)
4397 ferr(po, "TODO: retreg+tailcall\n");
4398
4399 for (arg = j = 0; arg < pp->argc; arg++) {
4400 if (arg > 0)
4401 fprintf(fout, ", ");
4402
4403 cast[0] = 0;
4404 if (pp->arg[arg].type.is_ptr)
4405 snprintf(cast, sizeof(cast), "(%s)",
4406 pp->arg[arg].type.name);
4407
4408 if (pp->arg[arg].reg != NULL) {
4409 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
4410 continue;
4411 }
4412 // stack arg
4413 for (; j < g_func_pp->argc; j++)
4414 if (g_func_pp->arg[j].reg == NULL)
4415 break;
4416 fprintf(fout, "%sa%d", cast, j + 1);
4417 j++;
4418 }
4419 }
4420 else {
4421 for (arg = 0; arg < pp->argc; arg++) {
4422 if (arg > 0)
4423 fprintf(fout, ", ");
4424
4425 cast[0] = 0;
4426 if (pp->arg[arg].type.is_ptr)
4427 snprintf(cast, sizeof(cast), "(%s)",
4428 pp->arg[arg].type.name);
4429
4430 if (pp->arg[arg].reg != NULL) {
4431 if (pp->arg[arg].type.is_retreg)
4432 fprintf(fout, "&%s", pp->arg[arg].reg);
4433 else
4434 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
4435 continue;
4436 }
4437
4438 // stack arg
4439 tmp_op = pp->arg[arg].datap;
4440 if (tmp_op == NULL)
4441 ferr(po, "parsed_op missing for arg%d\n", arg);
4442 if (tmp_op->argnum != 0) {
4443 fprintf(fout, "%ss_a%d", cast, tmp_op->argnum);
4444 }
4445 else {
4446 fprintf(fout, "%s",
4447 out_src_opr(buf1, sizeof(buf1),
4448 tmp_op, &tmp_op->operand[0], cast, 0));
4449 }
4450 }
4451 }
4452 fprintf(fout, ");");
4453
4454 if (strstr(pp->ret_type.name, "int64")) {
4455 fprintf(fout, "\n");
4456 fprintf(fout, " edx = tmp64 >> 32;\n");
4457 fprintf(fout, " eax = tmp64;");
4458 }
4459
4460 if (pp->is_unresolved) {
4461 snprintf(buf3, sizeof(buf3), " unresoved %dreg",
4462 pp->argc_reg);
4463 strcat(g_comment, buf3);
4464 }
4465
4466 if (po->flags & OPF_TAIL) {
4467 ret = 0;
4468 if (i == opcnt - 1 || pp->is_noreturn)
4469 ret = 0;
4470 else if (IS(pp->ret_type.name, "void"))
4471 ret = 1;
4472 else if (IS(g_func_pp->ret_type.name, "void"))
4473 ret = 1;
4474 // else already handled as 'return f()'
4475
4476 if (ret) {
4477 if (!IS(g_func_pp->ret_type.name, "void")) {
4478 ferr(po, "int func -> void func tailcall?\n");
4479 }
4480 else {
4481 fprintf(fout, "\n return;");
4482 strcat(g_comment, " ^ tailcall");
4483 }
4484 }
4485 else
4486 strcat(g_comment, " tailcall");
4487 }
4488 if (pp->is_noreturn)
4489 strcat(g_comment, " noreturn");
4490 if ((po->flags & OPF_ATAIL) && pp->argc_stack > 0)
4491 strcat(g_comment, " argframe");
4492 delayed_flag_op = NULL;
4493 last_arith_dst = NULL;
4494 break;
4495
4496 case OP_RET:
4497 if (g_func_pp->is_vararg)
4498 fprintf(fout, " va_end(ap);\n");
4499 if (g_func_pp->has_retreg) {
4500 for (arg = 0; arg < g_func_pp->argc; arg++)
4501 if (g_func_pp->arg[arg].type.is_retreg)
4502 fprintf(fout, " *r_%s = %s;\n",
4503 g_func_pp->arg[arg].reg, g_func_pp->arg[arg].reg);
4504 }
4505
4506 if (IS(g_func_pp->ret_type.name, "void")) {
4507 if (i != opcnt - 1 || label_pending)
4508 fprintf(fout, " return;");
4509 }
4510 else if (g_func_pp->ret_type.is_ptr) {
4511 fprintf(fout, " return (%s)eax;",
4512 g_func_pp->ret_type.name);
4513 }
4514 else if (IS(g_func_pp->ret_type.name, "__int64"))
4515 fprintf(fout, " return ((u64)edx << 32) | eax;");
4516 else
4517 fprintf(fout, " return eax;");
4518
4519 last_arith_dst = NULL;
4520 delayed_flag_op = NULL;
4521 break;
4522
4523 case OP_PUSH:
4524 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4525 if (po->argnum != 0) {
4526 // special case - saved func arg
4527 fprintf(fout, " s_a%d = %s;", po->argnum, buf1);
4528 break;
4529 }
4530 else if (po->flags & OPF_RSAVE) {
4531 fprintf(fout, " s_%s = %s;", buf1, buf1);
4532 break;
4533 }
4534 else if (g_func_pp->is_userstack) {
4535 fprintf(fout, " *(--esp) = %s;", buf1);
4536 break;
4537 }
4538 if (!(g_ida_func_attr & IDAFA_NORETURN))
4539 ferr(po, "stray push encountered\n");
4540 no_output = 1;
4541 break;
4542
4543 case OP_POP:
4544 if (po->flags & OPF_RSAVE) {
4545 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4546 fprintf(fout, " %s = s_%s;", buf1, buf1);
4547 break;
4548 }
4549 else if (po->datap != NULL) {
4550 // push/pop pair
4551 tmp_op = po->datap;
4552 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4553 fprintf(fout, " %s = %s;", buf1,
4554 out_src_opr(buf2, sizeof(buf2),
4555 tmp_op, &tmp_op->operand[0],
4556 po->operand[0].is_ptr ? "(void *)" : "", 0));
4557 break;
4558 }
4559 else if (g_func_pp->is_userstack) {
4560 fprintf(fout, " %s = *esp++;",
4561 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4562 break;
4563 }
4564 else
4565 ferr(po, "stray pop encountered\n");
4566 break;
4567
4568 case OP_NOP:
4569 no_output = 1;
4570 break;
4571
4572 default:
4573 no_output = 1;
4574 ferr(po, "unhandled op type %d, flags %x\n",
4575 po->op, po->flags);
4576 break;
4577 }
4578
4579 if (g_comment[0] != 0) {
4580 char *p = g_comment;
4581 while (my_isblank(*p))
4582 p++;
4583 fprintf(fout, " // %s", p);
4584 g_comment[0] = 0;
4585 no_output = 0;
4586 }
4587 if (!no_output)
4588 fprintf(fout, "\n");
4589
4590 // some sanity checking
4591 if (po->flags & OPF_REP) {
4592 if (po->op != OP_STOS && po->op != OP_MOVS
4593 && po->op != OP_CMPS && po->op != OP_SCAS)
4594 ferr(po, "unexpected rep\n");
4595 if (!(po->flags & (OPF_REPZ|OPF_REPNZ))
4596 && (po->op == OP_CMPS || po->op == OP_SCAS))
4597 ferr(po, "cmps/scas with plain rep\n");
4598 }
4599 if ((po->flags & (OPF_REPZ|OPF_REPNZ))
4600 && po->op != OP_CMPS && po->op != OP_SCAS)
4601 ferr(po, "unexpected repz/repnz\n");
4602
4603 if (pfomask != 0)
4604 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
4605
4606 // see is delayed flag stuff is still valid
4607 if (delayed_flag_op != NULL && delayed_flag_op != po) {
4608 if (is_any_opr_modified(delayed_flag_op, po, 0))
4609 delayed_flag_op = NULL;
4610 }
4611
4612 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
4613 if (is_opr_modified(last_arith_dst, po))
4614 last_arith_dst = NULL;
4615 }
4616
4617 label_pending = 0;
4618 }
4619
4620 if (g_stack_fsz && !g_stack_frame_used)
4621 fprintf(fout, " (void)sf;\n");
4622
4623 fprintf(fout, "}\n\n");
4624
4625 // cleanup
4626 for (i = 0; i < opcnt; i++) {
4627 struct label_ref *lr, *lr_del;
4628
4629 lr = g_label_refs[i].next;
4630 while (lr != NULL) {
4631 lr_del = lr;
4632 lr = lr->next;
4633 free(lr_del);
4634 }
4635 g_label_refs[i].i = -1;
4636 g_label_refs[i].next = NULL;
4637
4638 if (ops[i].op == OP_CALL) {
4639 pp = ops[i].datap;
4640 if (pp)
4641 proto_release(pp);
4642 }
4643 }
4644 g_func_pp = NULL;
4645}
4646
4647static void set_label(int i, const char *name)
4648{
4649 const char *p;
4650 int len;
4651
4652 len = strlen(name);
4653 p = strchr(name, ':');
4654 if (p != NULL)
4655 len = p - name;
4656
4657 if (len > sizeof(g_labels[0]) - 1)
4658 aerr("label '%s' too long: %d\n", name, len);
4659 if (g_labels[i][0] != 0 && !IS_START(g_labels[i], "algn_"))
4660 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
4661 memcpy(g_labels[i], name, len);
4662 g_labels[i][len] = 0;
4663}
4664
4665// '=' needs special treatment..
4666static char *next_word_s(char *w, size_t wsize, char *s)
4667{
4668 size_t i;
4669
4670 s = sskip(s);
4671
4672 for (i = 0; i < wsize - 1; i++) {
4673 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
4674 break;
4675 w[i] = s[i];
4676 }
4677 w[i] = 0;
4678
4679 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
4680 printf("warning: '%s' truncated\n", w);
4681
4682 return s + i;
4683}
4684
4685struct chunk_item {
4686 char *name;
4687 long fptr;
4688 int asmln;
4689};
4690
4691static struct chunk_item *func_chunks;
4692static int func_chunk_cnt;
4693static int func_chunk_alloc;
4694
4695static void add_func_chunk(FILE *fasm, const char *name, int line)
4696{
4697 if (func_chunk_cnt >= func_chunk_alloc) {
4698 func_chunk_alloc *= 2;
4699 func_chunks = realloc(func_chunks,
4700 func_chunk_alloc * sizeof(func_chunks[0]));
4701 my_assert_not(func_chunks, NULL);
4702 }
4703 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
4704 func_chunks[func_chunk_cnt].name = strdup(name);
4705 func_chunks[func_chunk_cnt].asmln = line;
4706 func_chunk_cnt++;
4707}
4708
4709static int cmp_chunks(const void *p1, const void *p2)
4710{
4711 const struct chunk_item *c1 = p1, *c2 = p2;
4712 return strcmp(c1->name, c2->name);
4713}
4714
4715static int cmpstringp(const void *p1, const void *p2)
4716{
4717 return strcmp(*(char * const *)p1, *(char * const *)p2);
4718}
4719
4720static void scan_ahead(FILE *fasm)
4721{
4722 char words[2][256];
4723 char line[256];
4724 long oldpos;
4725 int oldasmln;
4726 int wordc;
4727 char *p;
4728 int i;
4729
4730 oldpos = ftell(fasm);
4731 oldasmln = asmln;
4732
4733 while (fgets(line, sizeof(line), fasm))
4734 {
4735 wordc = 0;
4736 asmln++;
4737
4738 p = sskip(line);
4739 if (*p == 0)
4740 continue;
4741
4742 if (*p == ';')
4743 {
4744 // get rid of random tabs
4745 for (i = 0; line[i] != 0; i++)
4746 if (line[i] == '\t')
4747 line[i] = ' ';
4748
4749 if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
4750 {
4751 p += 30;
4752 next_word(words[0], sizeof(words[0]), p);
4753 if (words[0][0] == 0)
4754 aerr("missing name for func chunk?\n");
4755
4756 add_func_chunk(fasm, words[0], asmln);
4757 }
4758 continue;
4759 } // *p == ';'
4760
4761 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
4762 words[wordc][0] = 0;
4763 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
4764 if (*p == 0 || *p == ';') {
4765 wordc++;
4766 break;
4767 }
4768 }
4769
4770 if (wordc == 2 && IS(words[1], "ends"))
4771 break;
4772 }
4773
4774 fseek(fasm, oldpos, SEEK_SET);
4775 asmln = oldasmln;
4776}
4777
4778int main(int argc, char *argv[])
4779{
4780 FILE *fout, *fasm, *frlist;
4781 struct parsed_data *pd = NULL;
4782 int pd_alloc = 0;
4783 char **rlist = NULL;
4784 int rlist_len = 0;
4785 int rlist_alloc = 0;
4786 int func_chunks_used = 0;
4787 int func_chunks_sorted = 0;
4788 int func_chunk_i = -1;
4789 long func_chunk_ret = 0;
4790 int func_chunk_ret_ln = 0;
4791 int scanned_ahead = 0;
4792 char line[256];
4793 char words[20][256];
4794 enum opr_lenmod lmod;
4795 char *sctproto = NULL;
4796 int in_func = 0;
4797 int pending_endp = 0;
4798 int skip_func = 0;
4799 int skip_warned = 0;
4800 int eq_alloc;
4801 int verbose = 0;
4802 int multi_seg = 0;
4803 int arg_out;
4804 int arg;
4805 int pi = 0;
4806 int i, j;
4807 int ret, len;
4808 char *p;
4809 int wordc;
4810
4811 for (arg = 1; arg < argc; arg++) {
4812 if (IS(argv[arg], "-v"))
4813 verbose = 1;
4814 else if (IS(argv[arg], "-rf"))
4815 g_allow_regfunc = 1;
4816 else if (IS(argv[arg], "-m"))
4817 multi_seg = 1;
4818 else
4819 break;
4820 }
4821
4822 if (argc < arg + 3) {
4823 printf("usage:\n%s [-v] [-rf] [-m] <.c> <.asm> <hdrf> [rlist]*\n",
4824 argv[0]);
4825 return 1;
4826 }
4827
4828 arg_out = arg++;
4829
4830 asmfn = argv[arg++];
4831 fasm = fopen(asmfn, "r");
4832 my_assert_not(fasm, NULL);
4833
4834 hdrfn = argv[arg++];
4835 g_fhdr = fopen(hdrfn, "r");
4836 my_assert_not(g_fhdr, NULL);
4837
4838 rlist_alloc = 64;
4839 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
4840 my_assert_not(rlist, NULL);
4841 // needs special handling..
4842 rlist[rlist_len++] = "__alloca_probe";
4843
4844 func_chunk_alloc = 32;
4845 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
4846 my_assert_not(func_chunks, NULL);
4847
4848 memset(words, 0, sizeof(words));
4849
4850 for (; arg < argc; arg++) {
4851 frlist = fopen(argv[arg], "r");
4852 my_assert_not(frlist, NULL);
4853
4854 while (fgets(line, sizeof(line), frlist)) {
4855 p = sskip(line);
4856 if (*p == 0 || *p == ';')
4857 continue;
4858 if (*p == '#') {
4859 if (IS_START(p, "#if 0")
4860 || (g_allow_regfunc && IS_START(p, "#if NO_REGFUNC")))
4861 {
4862 skip_func = 1;
4863 }
4864 else if (IS_START(p, "#endif"))
4865 skip_func = 0;
4866 continue;
4867 }
4868 if (skip_func)
4869 continue;
4870
4871 p = next_word(words[0], sizeof(words[0]), p);
4872 if (words[0][0] == 0)
4873 continue;
4874
4875 if (rlist_len >= rlist_alloc) {
4876 rlist_alloc = rlist_alloc * 2 + 64;
4877 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
4878 my_assert_not(rlist, NULL);
4879 }
4880 rlist[rlist_len++] = strdup(words[0]);
4881 }
4882 skip_func = 0;
4883
4884 fclose(frlist);
4885 frlist = NULL;
4886 }
4887
4888 if (rlist_len > 0)
4889 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
4890
4891 fout = fopen(argv[arg_out], "w");
4892 my_assert_not(fout, NULL);
4893
4894 eq_alloc = 128;
4895 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
4896 my_assert_not(g_eqs, NULL);
4897
4898 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
4899 g_label_refs[i].i = -1;
4900 g_label_refs[i].next = NULL;
4901 }
4902
4903 while (fgets(line, sizeof(line), fasm))
4904 {
4905 wordc = 0;
4906 asmln++;
4907
4908 p = sskip(line);
4909 if (*p == 0)
4910 continue;
4911
4912 // get rid of random tabs
4913 for (i = 0; line[i] != 0; i++)
4914 if (line[i] == '\t')
4915 line[i] = ' ';
4916
4917 if (*p == ';')
4918 {
4919 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
4920 goto do_pending_endp; // eww..
4921
4922 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
4923 {
4924 static const char *attrs[] = {
4925 "bp-based frame",
4926 "library function",
4927 "static",
4928 "noreturn",
4929 "thunk",
4930 "fpd=",
4931 };
4932
4933 // parse IDA's attribute-list comment
4934 g_ida_func_attr = 0;
4935 p = sskip(p + 13);
4936
4937 for (; *p != 0; p = sskip(p)) {
4938 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
4939 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
4940 g_ida_func_attr |= 1 << i;
4941 p += strlen(attrs[i]);
4942 break;
4943 }
4944 }
4945 if (i == ARRAY_SIZE(attrs)) {
4946 anote("unparsed IDA attr: %s\n", p);
4947 break;
4948 }
4949 if (IS(attrs[i], "fpd=")) {
4950 p = next_word(words[0], sizeof(words[0]), p);
4951 // ignore for now..
4952 }
4953 }
4954 }
4955 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
4956 {
4957 p += 30;
4958 next_word(words[0], sizeof(words[0]), p);
4959 if (words[0][0] == 0)
4960 aerr("missing name for func chunk?\n");
4961
4962 if (!scanned_ahead) {
4963 add_func_chunk(fasm, words[0], asmln);
4964 func_chunks_sorted = 0;
4965 }
4966 }
4967 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
4968 {
4969 if (func_chunk_i >= 0) {
4970 if (func_chunk_i < func_chunk_cnt
4971 && IS(func_chunks[func_chunk_i].name, g_func))
4972 {
4973 // move on to next chunk
4974 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
4975 if (ret)
4976 aerr("seek failed for '%s' chunk #%d\n",
4977 g_func, func_chunk_i);
4978 asmln = func_chunks[func_chunk_i].asmln;
4979 func_chunk_i++;
4980 }
4981 else {
4982 if (func_chunk_ret == 0)
4983 aerr("no return from chunk?\n");
4984 fseek(fasm, func_chunk_ret, SEEK_SET);
4985 asmln = func_chunk_ret_ln;
4986 func_chunk_ret = 0;
4987 pending_endp = 1;
4988 }
4989 }
4990 }
4991 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
4992 func_chunks_used = 1;
4993 p += 20;
4994 if (IS_START(g_func, "sub_")) {
4995 unsigned long addr = strtoul(p, NULL, 16);
4996 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
4997 if (addr > f_addr && !scanned_ahead) {
4998 anote("scan_ahead caused by '%s', addr %lx\n",
4999 g_func, addr);
5000 scan_ahead(fasm);
5001 scanned_ahead = 1;
5002 func_chunks_sorted = 0;
5003 }
5004 }
5005 }
5006 continue;
5007 } // *p == ';'
5008
5009parse_words:
5010 for (i = wordc; i < ARRAY_SIZE(words); i++)
5011 words[i][0] = 0;
5012 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
5013 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
5014 if (*p == 0 || *p == ';') {
5015 wordc++;
5016 break;
5017 }
5018 }
5019 if (*p != 0 && *p != ';')
5020 aerr("too many words\n");
5021
5022 // alow asm patches in comments
5023 if (*p == ';') {
5024 if (IS_START(p, "; sctpatch:")) {
5025 p = sskip(p + 11);
5026 if (*p == 0 || *p == ';')
5027 continue;
5028 goto parse_words; // lame
5029 }
5030 if (IS_START(p, "; sctproto:")) {
5031 sctproto = strdup(p + 11);
5032 }
5033 }
5034
5035 if (wordc == 0) {
5036 // shouldn't happen
5037 awarn("wordc == 0?\n");
5038 continue;
5039 }
5040
5041 // don't care about this:
5042 if (words[0][0] == '.'
5043 || IS(words[0], "include")
5044 || IS(words[0], "assume") || IS(words[1], "segment")
5045 || IS(words[0], "align"))
5046 {
5047 continue;
5048 }
5049
5050do_pending_endp:
5051 // do delayed endp processing to collect switch jumptables
5052 if (pending_endp) {
5053 if (in_func && !skip_func && wordc >= 2
5054 && ((words[0][0] == 'd' && words[0][2] == 0)
5055 || (words[1][0] == 'd' && words[1][2] == 0)))
5056 {
5057 i = 1;
5058 if (words[1][0] == 'd' && words[1][2] == 0) {
5059 // label
5060 if (g_func_pd_cnt >= pd_alloc) {
5061 pd_alloc = pd_alloc * 2 + 16;
5062 g_func_pd = realloc(g_func_pd,
5063 sizeof(g_func_pd[0]) * pd_alloc);
5064 my_assert_not(g_func_pd, NULL);
5065 }
5066 pd = &g_func_pd[g_func_pd_cnt];
5067 g_func_pd_cnt++;
5068 memset(pd, 0, sizeof(*pd));
5069 strcpy(pd->label, words[0]);
5070 pd->type = OPT_CONST;
5071 pd->lmod = lmod_from_directive(words[1]);
5072 i = 2;
5073 }
5074 else {
5075 if (pd == NULL) {
5076 if (verbose)
5077 anote("skipping alignment byte?\n");
5078 continue;
5079 }
5080 lmod = lmod_from_directive(words[0]);
5081 if (lmod != pd->lmod)
5082 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
5083 }
5084
5085 if (pd->count_alloc < pd->count + wordc) {
5086 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
5087 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
5088 my_assert_not(pd->d, NULL);
5089 }
5090 for (; i < wordc; i++) {
5091 if (IS(words[i], "offset")) {
5092 pd->type = OPT_OFFSET;
5093 i++;
5094 }
5095 p = strchr(words[i], ',');
5096 if (p != NULL)
5097 *p = 0;
5098 if (pd->type == OPT_OFFSET)
5099 pd->d[pd->count].u.label = strdup(words[i]);
5100 else
5101 pd->d[pd->count].u.val = parse_number(words[i]);
5102 pd->d[pd->count].bt_i = -1;
5103 pd->count++;
5104 }
5105 continue;
5106 }
5107
5108 if (in_func && !skip_func)
5109 gen_func(fout, g_fhdr, g_func, pi);
5110
5111 pending_endp = 0;
5112 in_func = 0;
5113 g_ida_func_attr = 0;
5114 skip_warned = 0;
5115 skip_func = 0;
5116 g_func[0] = 0;
5117 func_chunks_used = 0;
5118 func_chunk_i = -1;
5119 if (pi != 0) {
5120 memset(&ops, 0, pi * sizeof(ops[0]));
5121 memset(g_labels, 0, pi * sizeof(g_labels[0]));
5122 pi = 0;
5123 }
5124 g_eqcnt = 0;
5125 for (i = 0; i < g_func_pd_cnt; i++) {
5126 pd = &g_func_pd[i];
5127 if (pd->type == OPT_OFFSET) {
5128 for (j = 0; j < pd->count; j++)
5129 free(pd->d[j].u.label);
5130 }
5131 free(pd->d);
5132 pd->d = NULL;
5133 }
5134 g_func_pd_cnt = 0;
5135 pd = NULL;
5136 if (wordc == 0)
5137 continue;
5138 }
5139
5140 if (IS(words[1], "proc")) {
5141 if (in_func)
5142 aerr("proc '%s' while in_func '%s'?\n",
5143 words[0], g_func);
5144 p = words[0];
5145 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
5146 skip_func = 1;
5147 strcpy(g_func, words[0]);
5148 set_label(0, words[0]);
5149 in_func = 1;
5150 continue;
5151 }
5152
5153 if (IS(words[1], "endp"))
5154 {
5155 if (!in_func)
5156 aerr("endp '%s' while not in_func?\n", words[0]);
5157 if (!IS(g_func, words[0]))
5158 aerr("endp '%s' while in_func '%s'?\n",
5159 words[0], g_func);
5160
5161 if ((g_ida_func_attr & IDAFA_THUNK) && pi == 1
5162 && ops[0].op == OP_JMP && ops[0].operand[0].had_ds)
5163 {
5164 // import jump
5165 skip_func = 1;
5166 }
5167
5168 if (!skip_func && func_chunks_used) {
5169 // start processing chunks
5170 struct chunk_item *ci, key = { g_func, 0 };
5171
5172 func_chunk_ret = ftell(fasm);
5173 func_chunk_ret_ln = asmln;
5174 if (!func_chunks_sorted) {
5175 qsort(func_chunks, func_chunk_cnt,
5176 sizeof(func_chunks[0]), cmp_chunks);
5177 func_chunks_sorted = 1;
5178 }
5179 ci = bsearch(&key, func_chunks, func_chunk_cnt,
5180 sizeof(func_chunks[0]), cmp_chunks);
5181 if (ci == NULL)
5182 aerr("'%s' needs chunks, but none found\n", g_func);
5183 func_chunk_i = ci - func_chunks;
5184 for (; func_chunk_i > 0; func_chunk_i--)
5185 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
5186 break;
5187
5188 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
5189 if (ret)
5190 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
5191 asmln = func_chunks[func_chunk_i].asmln;
5192 func_chunk_i++;
5193 continue;
5194 }
5195 pending_endp = 1;
5196 continue;
5197 }
5198
5199 if (wordc == 2 && IS(words[1], "ends")) {
5200 if (!multi_seg)
5201 break;
5202
5203 // scan for next text segment
5204 while (fgets(line, sizeof(line), fasm)) {
5205 asmln++;
5206 p = sskip(line);
5207 if (*p == 0 || *p == ';')
5208 continue;
5209
5210 if (strstr(p, "segment para public 'CODE' use32"))
5211 break;
5212 }
5213
5214 continue;
5215 }
5216
5217 p = strchr(words[0], ':');
5218 if (p != NULL) {
5219 set_label(pi, words[0]);
5220 continue;
5221 }
5222
5223 if (!in_func || skip_func) {
5224 if (!skip_warned && !skip_func && g_labels[pi][0] != 0) {
5225 if (verbose)
5226 anote("skipping from '%s'\n", g_labels[pi]);
5227 skip_warned = 1;
5228 }
5229 g_labels[pi][0] = 0;
5230 continue;
5231 }
5232
5233 if (wordc > 1 && IS(words[1], "="))
5234 {
5235 if (wordc != 5)
5236 aerr("unhandled equ, wc=%d\n", wordc);
5237 if (g_eqcnt >= eq_alloc) {
5238 eq_alloc *= 2;
5239 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
5240 my_assert_not(g_eqs, NULL);
5241 }
5242
5243 len = strlen(words[0]);
5244 if (len > sizeof(g_eqs[0].name) - 1)
5245 aerr("equ name too long: %d\n", len);
5246 strcpy(g_eqs[g_eqcnt].name, words[0]);
5247
5248 if (!IS(words[3], "ptr"))
5249 aerr("unhandled equ\n");
5250 if (IS(words[2], "dword"))
5251 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
5252 else if (IS(words[2], "word"))
5253 g_eqs[g_eqcnt].lmod = OPLM_WORD;
5254 else if (IS(words[2], "byte"))
5255 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
5256 else
5257 aerr("bad lmod: '%s'\n", words[2]);
5258
5259 g_eqs[g_eqcnt].offset = parse_number(words[4]);
5260 g_eqcnt++;
5261 continue;
5262 }
5263
5264 if (pi >= ARRAY_SIZE(ops))
5265 aerr("too many ops\n");
5266
5267 parse_op(&ops[pi], words, wordc);
5268
5269 if (sctproto != NULL) {
5270 if (ops[pi].op == OP_CALL)
5271 ops[pi].datap = sctproto;
5272 sctproto = NULL;
5273 }
5274 pi++;
5275 }
5276
5277 fclose(fout);
5278 fclose(fasm);
5279 fclose(g_fhdr);
5280
5281 return 0;
5282}
5283
5284// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.