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