notaz.gp2x.de / ia32rtools.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
translate: handle another stackframe setup way
[ia32rtools.git] / tools / translate.c
... / ...
CommitLineData
1/*
2 * ia32rtools
3 * (C) notaz, 2013,2014
4 *
5 * This work is licensed under the terms of 3-clause BSD license.
6 * See COPYING file in the top-level directory.
7 */
8
9#define _GNU_SOURCE
10#include <stdio.h>
11#include <stdlib.h>
12#include <string.h>
13
14#include "my_assert.h"
15#include "my_str.h"
16
17#define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
18#define IS(w, y) !strcmp(w, y)
19#define IS_START(w, y) !strncmp(w, y, strlen(y))
20
21#include "protoparse.h"
22
23static const char *asmfn;
24static int asmln;
25static FILE *g_fhdr;
26
27#define anote(fmt, ...) \
28 printf("%s:%d: note: " fmt, asmfn, asmln, ##__VA_ARGS__)
29#define awarn(fmt, ...) \
30 printf("%s:%d: warning: " fmt, asmfn, asmln, ##__VA_ARGS__)
31#define aerr(fmt, ...) do { \
32 printf("%s:%d: error: " fmt, asmfn, asmln, ##__VA_ARGS__); \
33 fcloseall(); \
34 exit(1); \
35} while (0)
36
37#include "masm_tools.h"
38
39enum op_flags {
40 OPF_RMD = (1 << 0), /* removed or optimized out */
41 OPF_DATA = (1 << 1), /* data processing - writes to dst opr */
42 OPF_FLAGS = (1 << 2), /* sets flags */
43 OPF_JMP = (1 << 3), /* branch, call */
44 OPF_CJMP = (1 << 4), /* cond. branch (cc or jecxz) */
45 OPF_CC = (1 << 5), /* uses flags */
46 OPF_TAIL = (1 << 6), /* ret or tail call */
47 OPF_RSAVE = (1 << 7), /* push/pop is local reg save/load */
48 OPF_REP = (1 << 8), /* prefixed by rep */
49 OPF_REPZ = (1 << 9), /* rep is repe/repz */
50 OPF_REPNZ = (1 << 10), /* rep is repne/repnz */
51 OPF_FARG = (1 << 11), /* push collected as func arg (no reuse) */
52 OPF_EBP_S = (1 << 12), /* ebp used as scratch here, not BP */
53 OPF_DF = (1 << 13), /* DF flag set */
54 OPF_ATAIL = (1 << 14), /* tail call with reused arg frame */
55 OPF_32BIT = (1 << 15), /* 32bit division */
56 OPF_LOCK = (1 << 16), /* op has lock prefix */
57 OPF_VAPUSH = (1 << 17), /* vararg ptr push (as call arg) */
58};
59
60enum op_op {
61 OP_INVAL,
62 OP_NOP,
63 OP_PUSH,
64 OP_POP,
65 OP_LEAVE,
66 OP_MOV,
67 OP_LEA,
68 OP_MOVZX,
69 OP_MOVSX,
70 OP_XCHG,
71 OP_NOT,
72 OP_CDQ,
73 OP_LODS,
74 OP_STOS,
75 OP_MOVS,
76 OP_CMPS,
77 OP_SCAS,
78 OP_STD,
79 OP_CLD,
80 OP_RET,
81 OP_ADD,
82 OP_SUB,
83 OP_AND,
84 OP_OR,
85 OP_XOR,
86 OP_SHL,
87 OP_SHR,
88 OP_SAR,
89 OP_ROL,
90 OP_ROR,
91 OP_RCL,
92 OP_RCR,
93 OP_ADC,
94 OP_SBB,
95 OP_BSF,
96 OP_INC,
97 OP_DEC,
98 OP_NEG,
99 OP_MUL,
100 OP_IMUL,
101 OP_DIV,
102 OP_IDIV,
103 OP_TEST,
104 OP_CMP,
105 OP_CALL,
106 OP_JMP,
107 OP_JECXZ,
108 OP_JCC,
109 OP_SCC,
110};
111
112enum opr_type {
113 OPT_UNSPEC,
114 OPT_REG,
115 OPT_REGMEM,
116 OPT_LABEL,
117 OPT_OFFSET,
118 OPT_CONST,
119};
120
121// must be sorted (larger len must be further in enum)
122enum opr_lenmod {
123 OPLM_UNSPEC,
124 OPLM_BYTE,
125 OPLM_WORD,
126 OPLM_DWORD,
127};
128
129#define MAX_OPERANDS 3
130
131struct parsed_opr {
132 enum opr_type type;
133 enum opr_lenmod lmod;
134 unsigned int is_ptr:1; // pointer in C
135 unsigned int is_array:1; // array in C
136 unsigned int type_from_var:1; // .. in header, sometimes wrong
137 unsigned int size_mismatch:1; // type override differs from C
138 unsigned int size_lt:1; // type override is larger than C
139 unsigned int had_ds:1; // had ds: prefix
140 const struct parsed_proto *pp; // for OPT_LABEL
141 int reg;
142 unsigned int val;
143 char name[112];
144};
145
146struct parsed_op {
147 enum op_op op;
148 struct parsed_opr operand[MAX_OPERANDS];
149 unsigned int flags;
150 unsigned char pfo;
151 unsigned char pfo_inv;
152 unsigned char operand_cnt;
153 unsigned char p_argnum; // push: altered before call arg #
154 unsigned char p_argpass;// push: arg of host func
155 unsigned char pad[3];
156 int regmask_src; // all referensed regs
157 int regmask_dst;
158 int pfomask; // flagop: parsed_flag_op that can't be delayed
159 int cc_scratch; // scratch storage during analysis
160 int bt_i; // branch target for branches
161 struct parsed_data *btj;// branch targets for jumptables
162 struct parsed_proto *pp;// parsed_proto for OP_CALL
163 void *datap;
164 int asmln;
165};
166
167// datap:
168// OP_CALL - parser proto hint (str)
169// (OPF_CC) - point to one of (OPF_FLAGS) that affects cc op
170// OP_POP - point to OP_PUSH in push/pop pair
171
172struct parsed_equ {
173 char name[64];
174 enum opr_lenmod lmod;
175 int offset;
176};
177
178struct parsed_data {
179 char label[256];
180 enum opr_type type;
181 enum opr_lenmod lmod;
182 int count;
183 int count_alloc;
184 struct {
185 union {
186 char *label;
187 unsigned int val;
188 } u;
189 int bt_i;
190 } *d;
191};
192
193struct label_ref {
194 int i;
195 struct label_ref *next;
196};
197
198enum ida_func_attr {
199 IDAFA_BP_FRAME = (1 << 0),
200 IDAFA_LIB_FUNC = (1 << 1),
201 IDAFA_STATIC = (1 << 2),
202 IDAFA_NORETURN = (1 << 3),
203 IDAFA_THUNK = (1 << 4),
204 IDAFA_FPD = (1 << 5),
205};
206
207#define MAX_OPS 4096
208
209static struct parsed_op ops[MAX_OPS];
210static struct parsed_equ *g_eqs;
211static int g_eqcnt;
212static char g_labels[MAX_OPS][48];
213static struct label_ref g_label_refs[MAX_OPS];
214static const struct parsed_proto *g_func_pp;
215static struct parsed_data *g_func_pd;
216static int g_func_pd_cnt;
217static char g_func[256];
218static char g_comment[256];
219static int g_bp_frame;
220static int g_sp_frame;
221static int g_stack_frame_used;
222static int g_stack_fsz;
223static int g_ida_func_attr;
224static int g_allow_regfunc;
225#define ferr(op_, fmt, ...) do { \
226 printf("%s:%d: error: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
227 dump_op(op_), ##__VA_ARGS__); \
228 fcloseall(); \
229 exit(1); \
230} while (0)
231#define fnote(op_, fmt, ...) \
232 printf("%s:%d: note: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
233 dump_op(op_), ##__VA_ARGS__)
234
235#define MAX_REGS 8
236
237const char *regs_r32[] = { "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp" };
238const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" };
239const char *regs_r8l[] = { "al", "bl", "cl", "dl" };
240const char *regs_r8h[] = { "ah", "bh", "ch", "dh" };
241
242enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP };
243
244// possible basic comparison types (without inversion)
245enum parsed_flag_op {
246 PFO_O, // 0 OF=1
247 PFO_C, // 2 CF=1
248 PFO_Z, // 4 ZF=1
249 PFO_BE, // 6 CF=1||ZF=1
250 PFO_S, // 8 SF=1
251 PFO_P, // a PF=1
252 PFO_L, // c SF!=OF
253 PFO_LE, // e ZF=1||SF!=OF
254};
255
256static const char *parsed_flag_op_names[] = {
257 "o", "c", "z", "be", "s", "p", "l", "le"
258};
259
260static int char_array_i(const char *array[], size_t len, const char *s)
261{
262 int i;
263
264 for (i = 0; i < len; i++)
265 if (IS(s, array[i]))
266 return i;
267
268 return -1;
269}
270
271static void printf_number(char *buf, size_t buf_size,
272 unsigned long number)
273{
274 // output in C-friendly form
275 snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
276}
277
278static int parse_reg(enum opr_lenmod *reg_lmod, const char *s)
279{
280 int reg;
281
282 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s);
283 if (reg >= 0) {
284 *reg_lmod = OPLM_DWORD;
285 return reg;
286 }
287 reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s);
288 if (reg >= 0) {
289 *reg_lmod = OPLM_WORD;
290 return reg;
291 }
292 reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s);
293 if (reg >= 0) {
294 *reg_lmod = OPLM_BYTE;
295 return reg;
296 }
297 reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s);
298 if (reg >= 0) {
299 *reg_lmod = OPLM_BYTE;
300 return reg;
301 }
302
303 return -1;
304}
305
306static int parse_indmode(char *name, int *regmask, int need_c_cvt)
307{
308 enum opr_lenmod lmod;
309 char cvtbuf[256];
310 char *d = cvtbuf;
311 char *s = name;
312 char w[64];
313 long number;
314 int reg;
315 int c = 0;
316
317 *d = 0;
318
319 while (*s != 0) {
320 d += strlen(d);
321 while (my_isblank(*s))
322 s++;
323 for (; my_issep(*s); d++, s++)
324 *d = *s;
325 while (my_isblank(*s))
326 s++;
327 *d = 0;
328
329 // skip 'ds:' prefix
330 if (IS_START(s, "ds:"))
331 s += 3;
332
333 s = next_idt(w, sizeof(w), s);
334 if (w[0] == 0)
335 break;
336 c++;
337
338 reg = parse_reg(&lmod, w);
339 if (reg >= 0) {
340 *regmask |= 1 << reg;
341 goto pass;
342 }
343
344 if ('0' <= w[0] && w[0] <= '9') {
345 number = parse_number(w);
346 printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number);
347 continue;
348 }
349
350 // probably some label/identifier - pass
351
352pass:
353 snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w);
354 }
355
356 if (need_c_cvt)
357 strcpy(name, cvtbuf);
358
359 return c;
360}
361
362static int is_reg_in_str(const char *s)
363{
364 int i;
365
366 if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
367 return 0;
368
369 for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
370 if (!strncmp(s, regs_r32[i], 3))
371 return 1;
372
373 return 0;
374}
375
376static const char *parse_stack_el(const char *name, char *extra_reg,
377 int early_try)
378{
379 const char *p, *p2, *s;
380 char *endp = NULL;
381 char buf[32];
382 long val;
383 int len;
384
385 if (g_bp_frame || early_try)
386 {
387 p = name;
388 if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
389 p += 4;
390 if (extra_reg != NULL) {
391 strncpy(extra_reg, name, 3);
392 extra_reg[4] = 0;
393 }
394 }
395
396 if (IS_START(p, "ebp+")) {
397 p += 4;
398
399 p2 = strchr(p, '+');
400 if (p2 != NULL && is_reg_in_str(p)) {
401 if (extra_reg != NULL) {
402 strncpy(extra_reg, p, p2 - p);
403 extra_reg[p2 - p] = 0;
404 }
405 p = p2 + 1;
406 }
407
408 if (!('0' <= *p && *p <= '9'))
409 return p;
410
411 return NULL;
412 }
413 }
414
415 if (!IS_START(name, "esp+"))
416 return NULL;
417
418 s = name + 4;
419 p = strchr(s, '+');
420 if (p) {
421 if (is_reg_in_str(s)) {
422 if (extra_reg != NULL) {
423 strncpy(extra_reg, s, p - s);
424 extra_reg[p - s] = 0;
425 }
426 s = p + 1;
427 p = strchr(s, '+');
428 if (p == NULL)
429 aerr("%s IDA stackvar not set?\n", __func__);
430 }
431 if (!('0' <= *s && *s <= '9')) {
432 aerr("%s IDA stackvar offset not set?\n", __func__);
433 return NULL;
434 }
435 if (s[0] == '0' && s[1] == 'x')
436 s += 2;
437 len = p - s;
438 if (len < sizeof(buf) - 1) {
439 strncpy(buf, s, len);
440 buf[len] = 0;
441 val = strtol(buf, &endp, 16);
442 if (val == 0 || *endp != 0) {
443 aerr("%s num parse fail for '%s'\n", __func__, buf);
444 return NULL;
445 }
446 }
447 p++;
448 }
449 else
450 p = name + 4;
451
452 if ('0' <= *p && *p <= '9')
453 return NULL;
454
455 return p;
456}
457
458static int guess_lmod_from_name(struct parsed_opr *opr)
459{
460 if (!strncmp(opr->name, "dword_", 6)) {
461 opr->lmod = OPLM_DWORD;
462 return 1;
463 }
464 if (!strncmp(opr->name, "word_", 5)) {
465 opr->lmod = OPLM_WORD;
466 return 1;
467 }
468 if (!strncmp(opr->name, "byte_", 5)) {
469 opr->lmod = OPLM_BYTE;
470 return 1;
471 }
472 return 0;
473}
474
475static int guess_lmod_from_c_type(enum opr_lenmod *lmod,
476 const struct parsed_type *c_type)
477{
478 static const char *dword_types[] = {
479 "int", "_DWORD", "UINT_PTR", "DWORD",
480 "WPARAM", "LPARAM", "UINT", "__int32",
481 "LONG", "HIMC", "BOOL", "size_t",
482 };
483 static const char *word_types[] = {
484 "uint16_t", "int16_t", "_WORD", "WORD",
485 "unsigned __int16", "__int16",
486 };
487 static const char *byte_types[] = {
488 "uint8_t", "int8_t", "char",
489 "unsigned __int8", "__int8", "BYTE", "_BYTE",
490 "CHAR", "_UNKNOWN",
491 // structures.. deal the same as with _UNKNOWN for now
492 "CRITICAL_SECTION",
493 };
494 const char *n;
495 int i;
496
497 if (c_type->is_ptr) {
498 *lmod = OPLM_DWORD;
499 return 1;
500 }
501
502 n = skip_type_mod(c_type->name);
503
504 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
505 if (IS(n, dword_types[i])) {
506 *lmod = OPLM_DWORD;
507 return 1;
508 }
509 }
510
511 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
512 if (IS(n, word_types[i])) {
513 *lmod = OPLM_WORD;
514 return 1;
515 }
516 }
517
518 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
519 if (IS(n, byte_types[i])) {
520 *lmod = OPLM_BYTE;
521 return 1;
522 }
523 }
524
525 return 0;
526}
527
528static char *default_cast_to(char *buf, size_t buf_size,
529 struct parsed_opr *opr)
530{
531 buf[0] = 0;
532
533 if (!opr->is_ptr)
534 return buf;
535 if (opr->pp == NULL || opr->pp->type.name == NULL
536 || opr->pp->is_fptr)
537 {
538 snprintf(buf, buf_size, "%s", "(void *)");
539 return buf;
540 }
541
542 snprintf(buf, buf_size, "(%s)", opr->pp->type.name);
543 return buf;
544}
545
546static enum opr_type lmod_from_directive(const char *d)
547{
548 if (IS(d, "dd"))
549 return OPLM_DWORD;
550 else if (IS(d, "dw"))
551 return OPLM_WORD;
552 else if (IS(d, "db"))
553 return OPLM_BYTE;
554
555 aerr("unhandled directive: '%s'\n", d);
556 return OPLM_UNSPEC;
557}
558
559static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
560 int *regmask)
561{
562 opr->type = OPT_REG;
563 opr->reg = reg;
564 opr->lmod = lmod;
565 *regmask |= 1 << reg;
566}
567
568static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
569 int *extra_offs);
570
571static int parse_operand(struct parsed_opr *opr,
572 int *regmask, int *regmask_indirect,
573 char words[16][256], int wordc, int w, unsigned int op_flags)
574{
575 const struct parsed_proto *pp;
576 enum opr_lenmod tmplmod;
577 unsigned long number;
578 char buf[256];
579 int ret, len;
580 int wordc_in;
581 char *p;
582 int i;
583
584 if (w >= wordc)
585 aerr("parse_operand w %d, wordc %d\n", w, wordc);
586
587 opr->reg = xUNSPEC;
588
589 for (i = w; i < wordc; i++) {
590 len = strlen(words[i]);
591 if (words[i][len - 1] == ',') {
592 words[i][len - 1] = 0;
593 wordc = i + 1;
594 break;
595 }
596 }
597
598 wordc_in = wordc - w;
599
600 if ((op_flags & OPF_JMP) && wordc_in > 0
601 && !('0' <= words[w][0] && words[w][0] <= '9'))
602 {
603 const char *label = NULL;
604
605 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
606 && IS(words[w + 1], "ptr"))
607 label = words[w + 2];
608 else if (wordc_in == 2 && IS(words[w], "short"))
609 label = words[w + 1];
610 else if (wordc_in == 1
611 && strchr(words[w], '[') == NULL
612 && parse_reg(&tmplmod, words[w]) < 0)
613 label = words[w];
614
615 if (label != NULL) {
616 opr->type = OPT_LABEL;
617 if (IS_START(label, "ds:")) {
618 opr->had_ds = 1;
619 label += 3;
620 }
621 strcpy(opr->name, label);
622 return wordc;
623 }
624 }
625
626 if (wordc_in >= 3) {
627 if (IS(words[w + 1], "ptr")) {
628 if (IS(words[w], "dword"))
629 opr->lmod = OPLM_DWORD;
630 else if (IS(words[w], "word"))
631 opr->lmod = OPLM_WORD;
632 else if (IS(words[w], "byte"))
633 opr->lmod = OPLM_BYTE;
634 else
635 aerr("type parsing failed\n");
636 w += 2;
637 wordc_in = wordc - w;
638 }
639 }
640
641 if (wordc_in == 2) {
642 if (IS(words[w], "offset")) {
643 opr->type = OPT_OFFSET;
644 opr->lmod = OPLM_DWORD;
645 strcpy(opr->name, words[w + 1]);
646 pp = proto_parse(g_fhdr, opr->name, 1);
647 goto do_label;
648 }
649 if (IS(words[w], "(offset")) {
650 p = strchr(words[w + 1], ')');
651 if (p == NULL)
652 aerr("parse of bracketed offset failed\n");
653 *p = 0;
654 opr->type = OPT_OFFSET;
655 strcpy(opr->name, words[w + 1]);
656 return wordc;
657 }
658 }
659
660 if (wordc_in != 1)
661 aerr("parse_operand 1 word expected\n");
662
663 if (IS_START(words[w], "ds:")) {
664 opr->had_ds = 1;
665 memmove(words[w], words[w] + 3, strlen(words[w]) - 2);
666 }
667 strcpy(opr->name, words[w]);
668
669 if (words[w][0] == '[') {
670 opr->type = OPT_REGMEM;
671 ret = sscanf(words[w], "[%[^]]]", opr->name);
672 if (ret != 1)
673 aerr("[] parse failure\n");
674
675 parse_indmode(opr->name, regmask_indirect, 1);
676 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL, 1))
677 {
678 // might be an equ
679 struct parsed_equ *eq =
680 equ_find(NULL, parse_stack_el(opr->name, NULL, 1), &i);
681 if (eq)
682 opr->lmod = eq->lmod;
683 }
684 return wordc;
685 }
686 else if (strchr(words[w], '[')) {
687 // label[reg] form
688 p = strchr(words[w], '[');
689 opr->type = OPT_REGMEM;
690 parse_indmode(p, regmask_indirect, 0);
691 strncpy(buf, words[w], p - words[w]);
692 buf[p - words[w]] = 0;
693 pp = proto_parse(g_fhdr, buf, 1);
694 goto do_label;
695 }
696 else if (('0' <= words[w][0] && words[w][0] <= '9')
697 || words[w][0] == '-')
698 {
699 number = parse_number(words[w]);
700 opr->type = OPT_CONST;
701 opr->val = number;
702 printf_number(opr->name, sizeof(opr->name), number);
703 return wordc;
704 }
705
706 ret = parse_reg(&tmplmod, opr->name);
707 if (ret >= 0) {
708 setup_reg_opr(opr, ret, tmplmod, regmask);
709 return wordc;
710 }
711
712 // most likely var in data segment
713 opr->type = OPT_LABEL;
714 pp = proto_parse(g_fhdr, opr->name, 0);
715
716do_label:
717 if (pp != NULL) {
718 if (pp->is_fptr || pp->is_func) {
719 opr->lmod = OPLM_DWORD;
720 opr->is_ptr = 1;
721 }
722 else {
723 tmplmod = OPLM_UNSPEC;
724 if (!guess_lmod_from_c_type(&tmplmod, &pp->type))
725 anote("unhandled C type '%s' for '%s'\n",
726 pp->type.name, opr->name);
727
728 if (opr->lmod == OPLM_UNSPEC) {
729 opr->lmod = tmplmod;
730 opr->type_from_var = 1;
731 }
732 else if (opr->lmod != tmplmod) {
733 opr->size_mismatch = 1;
734 if (tmplmod < opr->lmod)
735 opr->size_lt = 1;
736 }
737 opr->is_ptr = pp->type.is_ptr;
738 }
739 opr->is_array = pp->type.is_array;
740 }
741 opr->pp = pp;
742
743 if (opr->lmod == OPLM_UNSPEC)
744 guess_lmod_from_name(opr);
745 return wordc;
746}
747
748static const struct {
749 const char *name;
750 unsigned int flags;
751} pref_table[] = {
752 { "rep", OPF_REP },
753 { "repe", OPF_REP|OPF_REPZ },
754 { "repz", OPF_REP|OPF_REPZ },
755 { "repne", OPF_REP|OPF_REPNZ },
756 { "repnz", OPF_REP|OPF_REPNZ },
757 { "lock", OPF_LOCK }, // ignored for now..
758};
759
760#define OPF_CJMP_CC (OPF_JMP|OPF_CJMP|OPF_CC)
761
762static const struct {
763 const char *name;
764 enum op_op op;
765 unsigned short minopr;
766 unsigned short maxopr;
767 unsigned int flags;
768 unsigned char pfo;
769 unsigned char pfo_inv;
770} op_table[] = {
771 { "nop", OP_NOP, 0, 0, 0 },
772 { "push", OP_PUSH, 1, 1, 0 },
773 { "pop", OP_POP, 1, 1, OPF_DATA },
774 { "leave",OP_LEAVE, 0, 0, OPF_DATA },
775 { "mov" , OP_MOV, 2, 2, OPF_DATA },
776 { "lea", OP_LEA, 2, 2, OPF_DATA },
777 { "movzx",OP_MOVZX, 2, 2, OPF_DATA },
778 { "movsx",OP_MOVSX, 2, 2, OPF_DATA },
779 { "xchg", OP_XCHG, 2, 2, OPF_DATA },
780 { "not", OP_NOT, 1, 1, OPF_DATA },
781 { "cdq", OP_CDQ, 0, 0, OPF_DATA },
782 { "lodsb",OP_LODS, 0, 0, OPF_DATA },
783 { "lodsw",OP_LODS, 0, 0, OPF_DATA },
784 { "lodsd",OP_LODS, 0, 0, OPF_DATA },
785 { "stosb",OP_STOS, 0, 0, OPF_DATA },
786 { "stosw",OP_STOS, 0, 0, OPF_DATA },
787 { "stosd",OP_STOS, 0, 0, OPF_DATA },
788 { "movsb",OP_MOVS, 0, 0, OPF_DATA },
789 { "movsw",OP_MOVS, 0, 0, OPF_DATA },
790 { "movsd",OP_MOVS, 0, 0, OPF_DATA },
791 { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
792 { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
793 { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
794 { "scasb",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
795 { "scasw",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
796 { "scasd",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
797 { "std", OP_STD, 0, 0, OPF_DATA }, // special flag
798 { "cld", OP_CLD, 0, 0, OPF_DATA },
799 { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
800 { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
801 { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
802 { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS },
803 { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS },
804 { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
805 { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
806 { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
807 { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
808 { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
809 { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
810 { "rcl", OP_RCL, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
811 { "rcr", OP_RCR, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
812 { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
813 { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
814 { "bsf", OP_BSF, 2, 2, OPF_DATA|OPF_FLAGS },
815 { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
816 { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS },
817 { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS },
818 { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS },
819 { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS },
820 { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS },
821 { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS },
822 { "test", OP_TEST, 2, 2, OPF_FLAGS },
823 { "cmp", OP_CMP, 2, 2, OPF_FLAGS },
824 { "retn", OP_RET, 0, 1, OPF_TAIL },
825 { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS },
826 { "jmp", OP_JMP, 1, 1, OPF_JMP },
827 { "jecxz",OP_JECXZ, 1, 1, OPF_JMP|OPF_CJMP },
828 { "jo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 0 }, // 70 OF=1
829 { "jno", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 1 }, // 71 OF=0
830 { "jc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72 CF=1
831 { "jb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72
832 { "jnc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73 CF=0
833 { "jnb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
834 { "jae", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
835 { "jz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74 ZF=1
836 { "je", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74
837 { "jnz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75 ZF=0
838 { "jne", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75
839 { "jbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76 CF=1||ZF=1
840 { "jna", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76
841 { "ja", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77 CF=0&&ZF=0
842 { "jnbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77
843 { "js", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 0 }, // 78 SF=1
844 { "jns", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 1 }, // 79 SF=0
845 { "jp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a PF=1
846 { "jpe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a
847 { "jnp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b PF=0
848 { "jpo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b
849 { "jl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c SF!=OF
850 { "jnge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c
851 { "jge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d SF=OF
852 { "jnl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d
853 { "jle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e ZF=1||SF!=OF
854 { "jng", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e
855 { "jg", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f ZF=0&&SF=OF
856 { "jnle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f
857 { "seto", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 0 },
858 { "setno", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 1 },
859 { "setc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
860 { "setb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
861 { "setnc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
862 { "setae", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
863 { "setnb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
864 { "setz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
865 { "sete", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
866 { "setnz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
867 { "setne", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
868 { "setbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
869 { "setna", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
870 { "seta", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
871 { "setnbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
872 { "sets", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 0 },
873 { "setns", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 1 },
874 { "setp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
875 { "setpe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
876 { "setnp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
877 { "setpo", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
878 { "setl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
879 { "setnge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
880 { "setge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
881 { "setnl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
882 { "setle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
883 { "setng", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
884 { "setg", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
885 { "setnle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
886};
887
888static void parse_op(struct parsed_op *op, char words[16][256], int wordc)
889{
890 enum opr_lenmod lmod = OPLM_UNSPEC;
891 int prefix_flags = 0;
892 int regmask_ind;
893 int regmask;
894 int op_w = 0;
895 int opr = 0;
896 int w = 0;
897 int i;
898
899 for (i = 0; i < ARRAY_SIZE(pref_table); i++) {
900 if (IS(words[w], pref_table[i].name)) {
901 prefix_flags = pref_table[i].flags;
902 break;
903 }
904 }
905
906 if (prefix_flags) {
907 if (wordc <= 1)
908 aerr("lone prefix: '%s'\n", words[0]);
909 w++;
910 }
911
912 op_w = w;
913 for (i = 0; i < ARRAY_SIZE(op_table); i++) {
914 if (IS(words[w], op_table[i].name))
915 break;
916 }
917
918 if (i == ARRAY_SIZE(op_table))
919 aerr("unhandled op: '%s'\n", words[0]);
920 w++;
921
922 op->op = op_table[i].op;
923 op->flags = op_table[i].flags | prefix_flags;
924 op->pfo = op_table[i].pfo;
925 op->pfo_inv = op_table[i].pfo_inv;
926 op->regmask_src = op->regmask_dst = 0;
927 op->asmln = asmln;
928
929 for (opr = 0; opr < op_table[i].minopr; opr++) {
930 regmask = regmask_ind = 0;
931 w = parse_operand(&op->operand[opr], &regmask, &regmask_ind,
932 words, wordc, w, op->flags);
933
934 if (opr == 0 && (op->flags & OPF_DATA))
935 op->regmask_dst = regmask;
936 // for now, mark dst as src too
937 op->regmask_src |= regmask | regmask_ind;
938 }
939
940 for (; w < wordc && opr < op_table[i].maxopr; opr++) {
941 w = parse_operand(&op->operand[opr],
942 &op->regmask_src, &op->regmask_src,
943 words, wordc, w, op->flags);
944 }
945
946 if (w < wordc)
947 aerr("parse_op %s incomplete: %d/%d\n",
948 words[0], w, wordc);
949
950 // special cases
951 op->operand_cnt = opr;
952 if (!strncmp(op_table[i].name, "set", 3))
953 op->operand[0].lmod = OPLM_BYTE;
954
955 // ops with implicit argumets
956 switch (op->op) {
957 case OP_CDQ:
958 op->operand_cnt = 2;
959 setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst);
960 setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src);
961 break;
962
963 case OP_LODS:
964 case OP_STOS:
965 case OP_SCAS:
966 if (op->operand_cnt != 0)
967 break;
968 if (words[op_w][4] == 'b')
969 lmod = OPLM_BYTE;
970 else if (words[op_w][4] == 'w')
971 lmod = OPLM_WORD;
972 else if (words[op_w][4] == 'd')
973 lmod = OPLM_DWORD;
974 op->operand_cnt = 3;
975 setup_reg_opr(&op->operand[0], op->op == OP_LODS ? xSI : xDI,
976 lmod, &op->regmask_src);
977 setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src);
978 op->regmask_dst = op->regmask_src;
979 setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD,
980 op->op == OP_LODS ? &op->regmask_dst : &op->regmask_src);
981 break;
982
983 case OP_MOVS:
984 case OP_CMPS:
985 if (op->operand_cnt != 0)
986 break;
987 if (words[op_w][4] == 'b')
988 lmod = OPLM_BYTE;
989 else if (words[op_w][4] == 'w')
990 lmod = OPLM_WORD;
991 else if (words[op_w][4] == 'd')
992 lmod = OPLM_DWORD;
993 op->operand_cnt = 3;
994 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
995 setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src);
996 setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src);
997 op->regmask_dst = op->regmask_src;
998 break;
999
1000 case OP_XCHG:
1001 op->regmask_src |= op->regmask_dst;
1002 op->regmask_dst |= op->regmask_src;
1003 goto check_align;
1004
1005 case OP_JECXZ:
1006 op->operand_cnt = 1;
1007 op->regmask_src = 1 << xCX;
1008 op->operand[0].type = OPT_REG;
1009 op->operand[0].reg = xCX;
1010 op->operand[0].lmod = OPLM_DWORD;
1011 break;
1012
1013 case OP_IMUL:
1014 if (op->operand_cnt != 1)
1015 break;
1016 // fallthrough
1017 case OP_MUL:
1018 // singleop mul
1019 op->regmask_dst = (1 << xDX) | (1 << xAX);
1020 op->regmask_src |= (1 << xAX);
1021 if (op->operand[0].lmod == OPLM_UNSPEC)
1022 op->operand[0].lmod = OPLM_DWORD;
1023 break;
1024
1025 case OP_DIV:
1026 case OP_IDIV:
1027 // we could set up operands for edx:eax, but there is no real need to
1028 // (see is_opr_modified())
1029 regmask = (1 << xDX) | (1 << xAX);
1030 op->regmask_dst = regmask;
1031 op->regmask_src |= regmask;
1032 if (op->operand[0].lmod == OPLM_UNSPEC)
1033 op->operand[0].lmod = OPLM_DWORD;
1034 break;
1035
1036 case OP_SHL:
1037 case OP_SHR:
1038 case OP_SAR:
1039 case OP_ROL:
1040 case OP_ROR:
1041 if (op->operand[1].lmod == OPLM_UNSPEC)
1042 op->operand[1].lmod = OPLM_BYTE;
1043 break;
1044
1045 case OP_PUSH:
1046 if (op->operand[0].lmod == OPLM_UNSPEC
1047 && (op->operand[0].type == OPT_CONST
1048 || op->operand[0].type == OPT_OFFSET
1049 || op->operand[0].type == OPT_LABEL))
1050 op->operand[0].lmod = OPLM_DWORD;
1051 break;
1052
1053 // alignment
1054 case OP_MOV:
1055 check_align:
1056 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1057 && op->operand[0].lmod == op->operand[1].lmod
1058 && op->operand[0].reg == op->operand[1].reg
1059 && IS(op->operand[0].name, op->operand[1].name)) // ah, al..
1060 {
1061 op->flags |= OPF_RMD;
1062 op->regmask_src = op->regmask_dst = 0;
1063 }
1064 break;
1065
1066 case OP_LEA:
1067 if (op->operand[0].type == OPT_REG
1068 && op->operand[1].type == OPT_REGMEM)
1069 {
1070 char buf[16];
1071 snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name);
1072 if (IS(buf, op->operand[1].name))
1073 op->flags |= OPF_RMD;
1074 }
1075 break;
1076
1077 case OP_CALL:
1078 // trashed regs must be explicitly detected later
1079 op->regmask_dst = 0;
1080 break;
1081
1082 default:
1083 break;
1084 }
1085}
1086
1087static const char *op_name(struct parsed_op *po)
1088{
1089 static char buf[16];
1090 char *p;
1091 int i;
1092
1093 if (po->op == OP_JCC || po->op == OP_SCC) {
1094 p = buf;
1095 *p++ = (po->op == OP_JCC) ? 'j' : 's';
1096 if (po->pfo_inv)
1097 *p++ = 'n';
1098 strcpy(p, parsed_flag_op_names[po->pfo]);
1099 return buf;
1100 }
1101
1102 for (i = 0; i < ARRAY_SIZE(op_table); i++)
1103 if (op_table[i].op == po->op)
1104 return op_table[i].name;
1105
1106 return "???";
1107}
1108
1109// debug
1110static const char *dump_op(struct parsed_op *po)
1111{
1112 static char out[128];
1113 char *p = out;
1114 int i;
1115
1116 if (po == NULL)
1117 return "???";
1118
1119 snprintf(out, sizeof(out), "%s", op_name(po));
1120 for (i = 0; i < po->operand_cnt; i++) {
1121 p += strlen(p);
1122 if (i > 0)
1123 *p++ = ',';
1124 snprintf(p, sizeof(out) - (p - out),
1125 po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s",
1126 po->operand[i].name);
1127 }
1128
1129 return out;
1130}
1131
1132static const char *lmod_type_u(struct parsed_op *po,
1133 enum opr_lenmod lmod)
1134{
1135 switch (lmod) {
1136 case OPLM_DWORD:
1137 return "u32";
1138 case OPLM_WORD:
1139 return "u16";
1140 case OPLM_BYTE:
1141 return "u8";
1142 default:
1143 ferr(po, "invalid lmod: %d\n", lmod);
1144 return "(_invalid_)";
1145 }
1146}
1147
1148static const char *lmod_cast_u(struct parsed_op *po,
1149 enum opr_lenmod lmod)
1150{
1151 switch (lmod) {
1152 case OPLM_DWORD:
1153 return "";
1154 case OPLM_WORD:
1155 return "(u16)";
1156 case OPLM_BYTE:
1157 return "(u8)";
1158 default:
1159 ferr(po, "invalid lmod: %d\n", lmod);
1160 return "(_invalid_)";
1161 }
1162}
1163
1164static const char *lmod_cast_u_ptr(struct parsed_op *po,
1165 enum opr_lenmod lmod)
1166{
1167 switch (lmod) {
1168 case OPLM_DWORD:
1169 return "*(u32 *)";
1170 case OPLM_WORD:
1171 return "*(u16 *)";
1172 case OPLM_BYTE:
1173 return "*(u8 *)";
1174 default:
1175 ferr(po, "invalid lmod: %d\n", lmod);
1176 return "(_invalid_)";
1177 }
1178}
1179
1180static const char *lmod_cast_s(struct parsed_op *po,
1181 enum opr_lenmod lmod)
1182{
1183 switch (lmod) {
1184 case OPLM_DWORD:
1185 return "(s32)";
1186 case OPLM_WORD:
1187 return "(s16)";
1188 case OPLM_BYTE:
1189 return "(s8)";
1190 default:
1191 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1192 return "(_invalid_)";
1193 }
1194}
1195
1196static const char *lmod_cast(struct parsed_op *po,
1197 enum opr_lenmod lmod, int is_signed)
1198{
1199 return is_signed ?
1200 lmod_cast_s(po, lmod) :
1201 lmod_cast_u(po, lmod);
1202}
1203
1204static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
1205{
1206 switch (lmod) {
1207 case OPLM_DWORD:
1208 return 4;
1209 case OPLM_WORD:
1210 return 2;
1211 case OPLM_BYTE:
1212 return 1;
1213 default:
1214 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1215 return 0;
1216 }
1217}
1218
1219static const char *opr_name(struct parsed_op *po, int opr_num)
1220{
1221 if (opr_num >= po->operand_cnt)
1222 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1223 return po->operand[opr_num].name;
1224}
1225
1226static unsigned int opr_const(struct parsed_op *po, int opr_num)
1227{
1228 if (opr_num >= po->operand_cnt)
1229 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1230 if (po->operand[opr_num].type != OPT_CONST)
1231 ferr(po, "opr %d: const expected\n", opr_num);
1232 return po->operand[opr_num].val;
1233}
1234
1235static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr)
1236{
1237 if ((unsigned int)popr->reg >= MAX_REGS)
1238 ferr(po, "invalid reg: %d\n", popr->reg);
1239 return regs_r32[popr->reg];
1240}
1241
1242// cast1 is the "final" cast
1243static const char *simplify_cast(const char *cast1, const char *cast2)
1244{
1245 static char buf[256];
1246
1247 if (cast1[0] == 0)
1248 return cast2;
1249 if (cast2[0] == 0)
1250 return cast1;
1251 if (IS(cast1, cast2))
1252 return cast1;
1253 if (IS(cast1, "(s8)") && IS(cast2, "(u8)"))
1254 return cast1;
1255 if (IS(cast1, "(s16)") && IS(cast2, "(u16)"))
1256 return cast1;
1257 if (IS(cast1, "(u8)") && IS_START(cast2, "*(u8 *)"))
1258 return cast2;
1259 if (IS(cast1, "(u16)") && IS_START(cast2, "*(u16 *)"))
1260 return cast2;
1261 if (strchr(cast1, '*') && IS_START(cast2, "(u32)"))
1262 return cast1;
1263
1264 snprintf(buf, sizeof(buf), "%s%s", cast1, cast2);
1265 return buf;
1266}
1267
1268static const char *simplify_cast_num(const char *cast, unsigned int val)
1269{
1270 if (IS(cast, "(u8)") && val < 0x100)
1271 return "";
1272 if (IS(cast, "(s8)") && val < 0x80)
1273 return "";
1274 if (IS(cast, "(u16)") && val < 0x10000)
1275 return "";
1276 if (IS(cast, "(s16)") && val < 0x8000)
1277 return "";
1278 if (IS(cast, "(s32)") && val < 0x80000000)
1279 return "";
1280
1281 return cast;
1282}
1283
1284static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
1285 int *extra_offs)
1286{
1287 const char *p;
1288 char *endp;
1289 int namelen;
1290 int i;
1291
1292 *extra_offs = 0;
1293 namelen = strlen(name);
1294
1295 p = strchr(name, '+');
1296 if (p != NULL) {
1297 namelen = p - name;
1298 if (namelen <= 0)
1299 ferr(po, "equ parse failed for '%s'\n", name);
1300
1301 if (IS_START(p, "0x"))
1302 p += 2;
1303 *extra_offs = strtol(p, &endp, 16);
1304 if (*endp != 0)
1305 ferr(po, "equ parse failed for '%s'\n", name);
1306 }
1307
1308 for (i = 0; i < g_eqcnt; i++)
1309 if (strncmp(g_eqs[i].name, name, namelen) == 0
1310 && g_eqs[i].name[namelen] == 0)
1311 break;
1312 if (i >= g_eqcnt) {
1313 if (po != NULL)
1314 ferr(po, "unresolved equ name: '%s'\n", name);
1315 return NULL;
1316 }
1317
1318 return &g_eqs[i];
1319}
1320
1321static int is_stack_access(struct parsed_op *po,
1322 const struct parsed_opr *popr)
1323{
1324 return (parse_stack_el(popr->name, NULL, 0)
1325 || (g_bp_frame && !(po->flags & OPF_EBP_S)
1326 && IS_START(popr->name, "ebp")));
1327}
1328
1329static void parse_stack_access(struct parsed_op *po,
1330 const char *name, char *ofs_reg, int *offset_out,
1331 int *stack_ra_out, const char **bp_arg_out, int is_lea)
1332{
1333 const char *bp_arg = "";
1334 const char *p = NULL;
1335 struct parsed_equ *eq;
1336 char *endp = NULL;
1337 int stack_ra = 0;
1338 int offset = 0;
1339
1340 ofs_reg[0] = 0;
1341
1342 if (IS_START(name, "ebp-")
1343 || (IS_START(name, "ebp+") && '0' <= name[4] && name[4] <= '9'))
1344 {
1345 p = name + 4;
1346 if (IS_START(p, "0x"))
1347 p += 2;
1348 offset = strtoul(p, &endp, 16);
1349 if (name[3] == '-')
1350 offset = -offset;
1351 if (*endp != 0)
1352 ferr(po, "ebp- parse of '%s' failed\n", name);
1353 }
1354 else {
1355 bp_arg = parse_stack_el(name, ofs_reg, 0);
1356 snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
1357 eq = equ_find(po, bp_arg, &offset);
1358 if (eq == NULL)
1359 ferr(po, "detected but missing eq\n");
1360 offset += eq->offset;
1361 }
1362
1363 if (!strncmp(name, "ebp", 3))
1364 stack_ra = 4;
1365
1366 // yes it sometimes LEAs ra for compares..
1367 if (!is_lea && ofs_reg[0] == 0
1368 && stack_ra <= offset && offset < stack_ra + 4)
1369 {
1370 ferr(po, "reference to ra? %d %d\n", offset, stack_ra);
1371 }
1372
1373 *offset_out = offset;
1374 *stack_ra_out = stack_ra;
1375 if (bp_arg_out)
1376 *bp_arg_out = bp_arg;
1377}
1378
1379static int stack_frame_access(struct parsed_op *po,
1380 struct parsed_opr *popr, char *buf, size_t buf_size,
1381 const char *name, const char *cast, int is_src, int is_lea)
1382{
1383 enum opr_lenmod tmp_lmod = OPLM_UNSPEC;
1384 const char *prefix = "";
1385 const char *bp_arg = NULL;
1386 char ofs_reg[16] = { 0, };
1387 int i, arg_i, arg_s;
1388 int unaligned = 0;
1389 int stack_ra = 0;
1390 int offset = 0;
1391 int retval = -1;
1392 int sf_ofs;
1393 int lim;
1394
1395 if (po->flags & OPF_EBP_S)
1396 ferr(po, "stack_frame_access while ebp is scratch\n");
1397
1398 parse_stack_access(po, name, ofs_reg, &offset,
1399 &stack_ra, &bp_arg, is_lea);
1400
1401 if (offset > stack_ra)
1402 {
1403 arg_i = (offset - stack_ra - 4) / 4;
1404 if (arg_i < 0 || arg_i >= g_func_pp->argc_stack)
1405 {
1406 if (g_func_pp->is_vararg
1407 && arg_i == g_func_pp->argc_stack && is_lea)
1408 {
1409 // should be va_list
1410 if (cast[0] == 0)
1411 cast = "(u32)";
1412 snprintf(buf, buf_size, "%sap", cast);
1413 return -1;
1414 }
1415 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1416 offset, bp_arg, arg_i);
1417 }
1418 if (ofs_reg[0] != 0)
1419 ferr(po, "offset reg on arg access?\n");
1420
1421 for (i = arg_s = 0; i < g_func_pp->argc; i++) {
1422 if (g_func_pp->arg[i].reg != NULL)
1423 continue;
1424 if (arg_s == arg_i)
1425 break;
1426 arg_s++;
1427 }
1428 if (i == g_func_pp->argc)
1429 ferr(po, "arg %d not in prototype?\n", arg_i);
1430
1431 popr->is_ptr = g_func_pp->arg[i].type.is_ptr;
1432 retval = i;
1433
1434 switch (popr->lmod)
1435 {
1436 case OPLM_BYTE:
1437 if (is_lea)
1438 ferr(po, "lea/byte to arg?\n");
1439 if (is_src && (offset & 3) == 0)
1440 snprintf(buf, buf_size, "%sa%d",
1441 simplify_cast(cast, "(u8)"), i + 1);
1442 else
1443 snprintf(buf, buf_size, "%sBYTE%d(a%d)",
1444 cast, offset & 3, i + 1);
1445 break;
1446
1447 case OPLM_WORD:
1448 if (is_lea)
1449 ferr(po, "lea/word to arg?\n");
1450 if (offset & 1) {
1451 unaligned = 1;
1452 if (!is_src) {
1453 if (offset & 2)
1454 ferr(po, "problematic arg store\n");
1455 snprintf(buf, buf_size, "%s((char *)&a%d + 1)",
1456 simplify_cast(cast, "*(u16 *)"), i + 1);
1457 }
1458 else
1459 ferr(po, "unaligned arg word load\n");
1460 }
1461 else if (is_src && (offset & 2) == 0)
1462 snprintf(buf, buf_size, "%sa%d",
1463 simplify_cast(cast, "(u16)"), i + 1);
1464 else
1465 snprintf(buf, buf_size, "%s%sWORD(a%d)",
1466 cast, (offset & 2) ? "HI" : "LO", i + 1);
1467 break;
1468
1469 case OPLM_DWORD:
1470 if (cast[0])
1471 prefix = cast;
1472 else if (is_src)
1473 prefix = "(u32)";
1474
1475 if (offset & 3) {
1476 unaligned = 1;
1477 if (is_lea)
1478 snprintf(buf, buf_size, "(u32)&a%d + %d",
1479 i + 1, offset & 3);
1480 else if (!is_src)
1481 ferr(po, "unaligned arg store\n");
1482 else {
1483 // mov edx, [ebp+arg_4+2]; movsx ecx, dx
1484 snprintf(buf, buf_size, "%s(a%d >> %d)",
1485 prefix, i + 1, (offset & 3) * 8);
1486 }
1487 }
1488 else {
1489 snprintf(buf, buf_size, "%s%sa%d",
1490 prefix, is_lea ? "&" : "", i + 1);
1491 }
1492 break;
1493
1494 default:
1495 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
1496 }
1497
1498 if (unaligned)
1499 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1500
1501 // common problem
1502 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type);
1503 if (tmp_lmod != OPLM_DWORD
1504 && (unaligned || (!is_src && lmod_bytes(po, tmp_lmod)
1505 < lmod_bytes(po, popr->lmod) + (offset & 3))))
1506 {
1507 ferr(po, "bp_arg arg%d/w offset %d and type '%s' is too small\n",
1508 i + 1, offset, g_func_pp->arg[i].type.name);
1509 }
1510 // can't check this because msvc likes to reuse
1511 // arg space for scratch..
1512 //if (popr->is_ptr && popr->lmod != OPLM_DWORD)
1513 // ferr(po, "bp_arg arg%d: non-dword ptr access\n", i + 1);
1514 }
1515 else
1516 {
1517 if (g_stack_fsz == 0)
1518 ferr(po, "stack var access without stackframe\n");
1519 g_stack_frame_used = 1;
1520
1521 sf_ofs = g_stack_fsz + offset;
1522 lim = (ofs_reg[0] != 0) ? -4 : 0;
1523 if (offset > 0 || sf_ofs < lim)
1524 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
1525
1526 if (is_lea)
1527 prefix = "(u32)&";
1528 else
1529 prefix = cast;
1530
1531 switch (popr->lmod)
1532 {
1533 case OPLM_BYTE:
1534 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1535 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1536 break;
1537
1538 case OPLM_WORD:
1539 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1540 // known unaligned or possibly unaligned
1541 strcat(g_comment, " unaligned");
1542 if (prefix[0] == 0)
1543 prefix = "*(u16 *)&";
1544 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1545 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1546 break;
1547 }
1548 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
1549 break;
1550
1551 case OPLM_DWORD:
1552 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1553 // known unaligned or possibly unaligned
1554 strcat(g_comment, " unaligned");
1555 if (prefix[0] == 0)
1556 prefix = "*(u32 *)&";
1557 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1558 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1559 break;
1560 }
1561 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
1562 break;
1563
1564 default:
1565 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
1566 }
1567 }
1568
1569 return retval;
1570}
1571
1572static void check_func_pp(struct parsed_op *po,
1573 const struct parsed_proto *pp, const char *pfx)
1574{
1575 enum opr_lenmod tmp_lmod;
1576 char buf[256];
1577 int ret, i;
1578
1579 if (pp->argc_reg != 0) {
1580 if (/*!g_allow_regfunc &&*/ !pp->is_fastcall) {
1581 pp_print(buf, sizeof(buf), pp);
1582 ferr(po, "%s: unexpected reg arg in icall: %s\n", pfx, buf);
1583 }
1584 if (pp->argc_stack > 0 && pp->argc_reg != 2)
1585 ferr(po, "%s: %d reg arg(s) with %d stack arg(s)\n",
1586 pfx, pp->argc_reg, pp->argc_stack);
1587 }
1588
1589 // fptrs must use 32bit args, callsite might have no information and
1590 // lack a cast to smaller types, which results in incorrectly masked
1591 // args passed (callee may assume masked args, it does on ARM)
1592 if (!pp->is_oslib) {
1593 for (i = 0; i < pp->argc; i++) {
1594 ret = guess_lmod_from_c_type(&tmp_lmod, &pp->arg[i].type);
1595 if (ret && tmp_lmod != OPLM_DWORD)
1596 ferr(po, "reference to %s with arg%d '%s'\n", pp->name,
1597 i + 1, pp->arg[i].type.name);
1598 }
1599 }
1600}
1601
1602static const char *check_label_read_ref(struct parsed_op *po,
1603 const char *name)
1604{
1605 const struct parsed_proto *pp;
1606
1607 pp = proto_parse(g_fhdr, name, 0);
1608 if (pp == NULL)
1609 ferr(po, "proto_parse failed for ref '%s'\n", name);
1610
1611 if (pp->is_func)
1612 check_func_pp(po, pp, "ref");
1613
1614 return pp->name;
1615}
1616
1617static char *out_src_opr(char *buf, size_t buf_size,
1618 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
1619 int is_lea)
1620{
1621 char tmp1[256], tmp2[256];
1622 char expr[256];
1623 const char *name;
1624 char *p;
1625 int ret;
1626
1627 if (cast == NULL)
1628 cast = "";
1629
1630 switch (popr->type) {
1631 case OPT_REG:
1632 if (is_lea)
1633 ferr(po, "lea from reg?\n");
1634
1635 switch (popr->lmod) {
1636 case OPLM_DWORD:
1637 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
1638 break;
1639 case OPLM_WORD:
1640 snprintf(buf, buf_size, "%s%s",
1641 simplify_cast(cast, "(u16)"), opr_reg_p(po, popr));
1642 break;
1643 case OPLM_BYTE:
1644 if (popr->name[1] == 'h') // XXX..
1645 snprintf(buf, buf_size, "%s(%s >> 8)",
1646 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
1647 else
1648 snprintf(buf, buf_size, "%s%s",
1649 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
1650 break;
1651 default:
1652 ferr(po, "invalid src lmod: %d\n", popr->lmod);
1653 }
1654 break;
1655
1656 case OPT_REGMEM:
1657 if (is_stack_access(po, popr)) {
1658 stack_frame_access(po, popr, buf, buf_size,
1659 popr->name, cast, 1, is_lea);
1660 break;
1661 }
1662
1663 strcpy(expr, popr->name);
1664 if (strchr(expr, '[')) {
1665 // special case: '[' can only be left for label[reg] form
1666 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
1667 if (ret != 2)
1668 ferr(po, "parse failure for '%s'\n", expr);
1669 if (tmp1[0] == '(') {
1670 // (off_4FFF50+3)[eax]
1671 p = strchr(tmp1 + 1, ')');
1672 if (p == NULL || p[1] != 0)
1673 ferr(po, "parse failure (2) for '%s'\n", expr);
1674 *p = 0;
1675 memmove(tmp1, tmp1 + 1, strlen(tmp1));
1676 }
1677 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
1678 }
1679
1680 // XXX: do we need more parsing?
1681 if (is_lea) {
1682 snprintf(buf, buf_size, "%s", expr);
1683 break;
1684 }
1685
1686 snprintf(buf, buf_size, "%s(%s)",
1687 simplify_cast(cast, lmod_cast_u_ptr(po, popr->lmod)), expr);
1688 break;
1689
1690 case OPT_LABEL:
1691 name = check_label_read_ref(po, popr->name);
1692 if (cast[0] == 0 && popr->is_ptr)
1693 cast = "(u32)";
1694
1695 if (is_lea)
1696 snprintf(buf, buf_size, "(u32)&%s", name);
1697 else if (popr->size_lt)
1698 snprintf(buf, buf_size, "%s%s%s%s", cast,
1699 lmod_cast_u_ptr(po, popr->lmod),
1700 popr->is_array ? "" : "&", name);
1701 else
1702 snprintf(buf, buf_size, "%s%s%s", cast, name,
1703 popr->is_array ? "[0]" : "");
1704 break;
1705
1706 case OPT_OFFSET:
1707 name = check_label_read_ref(po, popr->name);
1708 if (cast[0] == 0)
1709 cast = "(u32)";
1710 if (is_lea)
1711 ferr(po, "lea an offset?\n");
1712 snprintf(buf, buf_size, "%s&%s", cast, name);
1713 break;
1714
1715 case OPT_CONST:
1716 if (is_lea)
1717 ferr(po, "lea from const?\n");
1718
1719 printf_number(tmp1, sizeof(tmp1), popr->val);
1720 if (popr->val == 0 && strchr(cast, '*'))
1721 snprintf(buf, buf_size, "NULL");
1722 else
1723 snprintf(buf, buf_size, "%s%s",
1724 simplify_cast_num(cast, popr->val), tmp1);
1725 break;
1726
1727 default:
1728 ferr(po, "invalid src type: %d\n", popr->type);
1729 }
1730
1731 return buf;
1732}
1733
1734// note: may set is_ptr (we find that out late for ebp frame..)
1735static char *out_dst_opr(char *buf, size_t buf_size,
1736 struct parsed_op *po, struct parsed_opr *popr)
1737{
1738 switch (popr->type) {
1739 case OPT_REG:
1740 switch (popr->lmod) {
1741 case OPLM_DWORD:
1742 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
1743 break;
1744 case OPLM_WORD:
1745 // ugh..
1746 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
1747 break;
1748 case OPLM_BYTE:
1749 // ugh..
1750 if (popr->name[1] == 'h') // XXX..
1751 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
1752 else
1753 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
1754 break;
1755 default:
1756 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
1757 }
1758 break;
1759
1760 case OPT_REGMEM:
1761 if (is_stack_access(po, popr)) {
1762 stack_frame_access(po, popr, buf, buf_size,
1763 popr->name, "", 0, 0);
1764 break;
1765 }
1766
1767 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1768
1769 case OPT_LABEL:
1770 if (popr->size_mismatch)
1771 snprintf(buf, buf_size, "%s%s%s",
1772 lmod_cast_u_ptr(po, popr->lmod),
1773 popr->is_array ? "" : "&", popr->name);
1774 else
1775 snprintf(buf, buf_size, "%s%s", popr->name,
1776 popr->is_array ? "[0]" : "");
1777 break;
1778
1779 default:
1780 ferr(po, "invalid dst type: %d\n", popr->type);
1781 }
1782
1783 return buf;
1784}
1785
1786static char *out_src_opr_u32(char *buf, size_t buf_size,
1787 struct parsed_op *po, struct parsed_opr *popr)
1788{
1789 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1790}
1791
1792static void out_test_for_cc(char *buf, size_t buf_size,
1793 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
1794 enum opr_lenmod lmod, const char *expr)
1795{
1796 const char *cast, *scast;
1797
1798 cast = lmod_cast_u(po, lmod);
1799 scast = lmod_cast_s(po, lmod);
1800
1801 switch (pfo) {
1802 case PFO_Z:
1803 case PFO_BE: // CF=1||ZF=1; CF=0
1804 snprintf(buf, buf_size, "(%s%s %s 0)",
1805 cast, expr, is_inv ? "!=" : "==");
1806 break;
1807
1808 case PFO_S:
1809 case PFO_L: // SF!=OF; OF=0
1810 snprintf(buf, buf_size, "(%s%s %s 0)",
1811 scast, expr, is_inv ? ">=" : "<");
1812 break;
1813
1814 case PFO_LE: // ZF=1||SF!=OF; OF=0
1815 snprintf(buf, buf_size, "(%s%s %s 0)",
1816 scast, expr, is_inv ? ">" : "<=");
1817 break;
1818
1819 default:
1820 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1821 }
1822}
1823
1824static void out_cmp_for_cc(char *buf, size_t buf_size,
1825 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
1826{
1827 const char *cast, *scast, *cast_use;
1828 char buf1[256], buf2[256];
1829 enum opr_lenmod lmod;
1830
1831 if (po->operand[0].lmod != po->operand[1].lmod)
1832 ferr(po, "%s: lmod mismatch: %d %d\n", __func__,
1833 po->operand[0].lmod, po->operand[1].lmod);
1834 lmod = po->operand[0].lmod;
1835
1836 cast = lmod_cast_u(po, lmod);
1837 scast = lmod_cast_s(po, lmod);
1838
1839 switch (pfo) {
1840 case PFO_C:
1841 case PFO_Z:
1842 case PFO_BE: // !a
1843 cast_use = cast;
1844 break;
1845
1846 case PFO_S:
1847 case PFO_L: // !ge
1848 case PFO_LE:
1849 cast_use = scast;
1850 break;
1851
1852 default:
1853 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1854 }
1855
1856 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], cast_use, 0);
1857 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], cast_use, 0);
1858
1859 switch (pfo) {
1860 case PFO_C:
1861 // note: must be unsigned compare
1862 snprintf(buf, buf_size, "(%s %s %s)",
1863 buf1, is_inv ? ">=" : "<", buf2);
1864 break;
1865
1866 case PFO_Z:
1867 snprintf(buf, buf_size, "(%s %s %s)",
1868 buf1, is_inv ? "!=" : "==", buf2);
1869 break;
1870
1871 case PFO_BE: // !a
1872 // note: must be unsigned compare
1873 snprintf(buf, buf_size, "(%s %s %s)",
1874 buf1, is_inv ? ">" : "<=", buf2);
1875
1876 // annoying case
1877 if (is_inv && lmod == OPLM_BYTE
1878 && po->operand[1].type == OPT_CONST
1879 && po->operand[1].val == 0xff)
1880 {
1881 snprintf(g_comment, sizeof(g_comment), "if %s", buf);
1882 snprintf(buf, buf_size, "(0)");
1883 }
1884 break;
1885
1886 // note: must be signed compare
1887 case PFO_S:
1888 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
1889 scast, buf1, buf2, is_inv ? ">=" : "<");
1890 break;
1891
1892 case PFO_L: // !ge
1893 snprintf(buf, buf_size, "(%s %s %s)",
1894 buf1, is_inv ? ">=" : "<", buf2);
1895 break;
1896
1897 case PFO_LE:
1898 snprintf(buf, buf_size, "(%s %s %s)",
1899 buf1, is_inv ? ">" : "<=", buf2);
1900 break;
1901
1902 default:
1903 break;
1904 }
1905}
1906
1907static void out_cmp_test(char *buf, size_t buf_size,
1908 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
1909{
1910 char buf1[256], buf2[256], buf3[256];
1911
1912 if (po->op == OP_TEST) {
1913 if (IS(opr_name(po, 0), opr_name(po, 1))) {
1914 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
1915 }
1916 else {
1917 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
1918 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
1919 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
1920 }
1921 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
1922 po->operand[0].lmod, buf3);
1923 }
1924 else if (po->op == OP_CMP) {
1925 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv);
1926 }
1927 else
1928 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
1929}
1930
1931static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
1932 struct parsed_opr *popr2)
1933{
1934 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
1935 ferr(po, "missing lmod for both operands\n");
1936
1937 if (popr1->lmod == OPLM_UNSPEC)
1938 popr1->lmod = popr2->lmod;
1939 else if (popr2->lmod == OPLM_UNSPEC)
1940 popr2->lmod = popr1->lmod;
1941 else if (popr1->lmod != popr2->lmod) {
1942 if (popr1->type_from_var) {
1943 popr1->size_mismatch = 1;
1944 if (popr1->lmod < popr2->lmod)
1945 popr1->size_lt = 1;
1946 popr1->lmod = popr2->lmod;
1947 }
1948 else if (popr2->type_from_var) {
1949 popr2->size_mismatch = 1;
1950 if (popr2->lmod < popr1->lmod)
1951 popr2->size_lt = 1;
1952 popr2->lmod = popr1->lmod;
1953 }
1954 else
1955 ferr(po, "conflicting lmods: %d vs %d\n",
1956 popr1->lmod, popr2->lmod);
1957 }
1958}
1959
1960static const char *op_to_c(struct parsed_op *po)
1961{
1962 switch (po->op)
1963 {
1964 case OP_ADD:
1965 case OP_ADC:
1966 return "+";
1967 case OP_SUB:
1968 case OP_SBB:
1969 return "-";
1970 case OP_AND:
1971 return "&";
1972 case OP_OR:
1973 return "|";
1974 case OP_XOR:
1975 return "^";
1976 case OP_SHL:
1977 return "<<";
1978 case OP_SHR:
1979 return ">>";
1980 case OP_MUL:
1981 case OP_IMUL:
1982 return "*";
1983 default:
1984 ferr(po, "op_to_c was supplied with %d\n", po->op);
1985 }
1986}
1987
1988static void op_set_clear_flag(struct parsed_op *po,
1989 enum op_flags flag_set, enum op_flags flag_clear)
1990{
1991 po->flags |= flag_set;
1992 po->flags &= ~flag_clear;
1993}
1994
1995// last op in stream - unconditional branch or ret
1996#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
1997 || ((ops[_i].flags & (OPF_JMP|OPF_CJMP|OPF_RMD)) == OPF_JMP \
1998 && ops[_i].op != OP_CALL))
1999
2000static int scan_for_pop(int i, int opcnt, const char *reg,
2001 int magic, int depth, int *maxdepth, int do_flags)
2002{
2003 const struct parsed_proto *pp;
2004 struct parsed_op *po;
2005 int ret = 0;
2006 int j;
2007
2008 for (; i < opcnt; i++) {
2009 po = &ops[i];
2010 if (po->cc_scratch == magic)
2011 break; // already checked
2012 po->cc_scratch = magic;
2013
2014 if (po->flags & OPF_TAIL) {
2015 if (po->op == OP_CALL) {
2016 pp = proto_parse(g_fhdr, po->operand[0].name, 0);
2017 if (pp != NULL && pp->is_noreturn)
2018 // no stack cleanup for noreturn
2019 return ret;
2020 }
2021 return -1; // deadend
2022 }
2023
2024 if ((po->flags & OPF_RMD)
2025 || (po->op == OP_PUSH && po->p_argnum != 0)) // arg push
2026 continue;
2027
2028 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2029 if (po->btj != NULL) {
2030 // jumptable
2031 for (j = 0; j < po->btj->count; j++) {
2032 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic,
2033 depth, maxdepth, do_flags);
2034 if (ret < 0)
2035 return ret; // dead end
2036 }
2037 return ret;
2038 }
2039
2040 if (po->bt_i < 0) {
2041 ferr(po, "dead branch\n");
2042 return -1;
2043 }
2044
2045 if (po->flags & OPF_CJMP) {
2046 ret |= scan_for_pop(po->bt_i, opcnt, reg, magic,
2047 depth, maxdepth, do_flags);
2048 if (ret < 0)
2049 return ret; // dead end
2050 }
2051 else {
2052 i = po->bt_i - 1;
2053 }
2054 continue;
2055 }
2056
2057 if ((po->op == OP_POP || po->op == OP_PUSH)
2058 && po->operand[0].type == OPT_REG
2059 && IS(po->operand[0].name, reg))
2060 {
2061 if (po->op == OP_PUSH && !(po->flags & OPF_FARG)) {
2062 depth++;
2063 if (depth > *maxdepth)
2064 *maxdepth = depth;
2065 if (do_flags)
2066 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2067 }
2068 else if (po->op == OP_POP) {
2069 if (depth == 0) {
2070 if (do_flags)
2071 op_set_clear_flag(po, OPF_RMD, OPF_RSAVE);
2072 return 1;
2073 }
2074 else {
2075 depth--;
2076 if (depth < 0) // should not happen
2077 ferr(po, "fail with depth\n");
2078 if (do_flags)
2079 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2080 }
2081 }
2082 }
2083 }
2084
2085 return ret;
2086}
2087
2088// scan for pop starting from 'ret' op (all paths)
2089static int scan_for_pop_ret(int i, int opcnt, const char *reg,
2090 int flag_set)
2091{
2092 int found = 0;
2093 int j;
2094
2095 for (; i < opcnt; i++) {
2096 if (!(ops[i].flags & OPF_TAIL))
2097 continue;
2098
2099 for (j = i - 1; j >= 0; j--) {
2100 if (ops[j].flags & OPF_RMD)
2101 continue;
2102 if (ops[j].flags & OPF_JMP)
2103 return -1;
2104
2105 if (ops[j].op == OP_POP && ops[j].operand[0].type == OPT_REG
2106 && IS(ops[j].operand[0].name, reg))
2107 {
2108 found = 1;
2109 ops[j].flags |= flag_set;
2110 break;
2111 }
2112
2113 if (g_labels[j][0] != 0)
2114 return -1;
2115 }
2116 }
2117
2118 return found ? 0 : -1;
2119}
2120
2121static void scan_propagate_df(int i, int opcnt)
2122{
2123 struct parsed_op *po = &ops[i];
2124 int j;
2125
2126 for (; i < opcnt; i++) {
2127 po = &ops[i];
2128 if (po->flags & OPF_DF)
2129 return; // already resolved
2130 po->flags |= OPF_DF;
2131
2132 if (po->op == OP_CALL)
2133 ferr(po, "call with DF set?\n");
2134
2135 if (po->flags & OPF_JMP) {
2136 if (po->btj != NULL) {
2137 // jumptable
2138 for (j = 0; j < po->btj->count; j++)
2139 scan_propagate_df(po->btj->d[j].bt_i, opcnt);
2140 return;
2141 }
2142
2143 if (po->bt_i < 0) {
2144 ferr(po, "dead branch\n");
2145 return;
2146 }
2147
2148 if (po->flags & OPF_CJMP)
2149 scan_propagate_df(po->bt_i, opcnt);
2150 else
2151 i = po->bt_i - 1;
2152 continue;
2153 }
2154
2155 if (po->flags & OPF_TAIL)
2156 break;
2157
2158 if (po->op == OP_CLD) {
2159 po->flags |= OPF_RMD;
2160 return;
2161 }
2162 }
2163
2164 ferr(po, "missing DF clear?\n");
2165}
2166
2167// is operand 'opr' modified by parsed_op 'po'?
2168static int is_opr_modified(const struct parsed_opr *opr,
2169 const struct parsed_op *po)
2170{
2171 int mask;
2172
2173 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2174 return 0;
2175
2176 if (opr->type == OPT_REG) {
2177 if (po->op == OP_CALL) {
2178 mask = (1 << xAX) | (1 << xCX) | (1 << xDX);
2179 if ((1 << opr->reg) & mask)
2180 return 1;
2181 else
2182 return 0;
2183 }
2184
2185 if (po->operand[0].type == OPT_REG) {
2186 if (po->regmask_dst & (1 << opr->reg))
2187 return 1;
2188 else
2189 return 0;
2190 }
2191 }
2192
2193 return IS(po->operand[0].name, opr->name);
2194}
2195
2196// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
2197static int is_any_opr_modified(const struct parsed_op *po_test,
2198 const struct parsed_op *po, int c_mode)
2199{
2200 int mask;
2201 int i;
2202
2203 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2204 return 0;
2205
2206 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2207 return 0;
2208
2209 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
2210 return 1;
2211
2212 // in reality, it can wreck any register, but in decompiled C
2213 // version it can only overwrite eax or edx:eax
2214 mask = (1 << xAX) | (1 << xDX);
2215 if (!c_mode)
2216 mask |= 1 << xCX;
2217
2218 if (po->op == OP_CALL
2219 && ((po_test->regmask_src | po_test->regmask_dst) & mask))
2220 return 1;
2221
2222 for (i = 0; i < po_test->operand_cnt; i++)
2223 if (IS(po_test->operand[i].name, po->operand[0].name))
2224 return 1;
2225
2226 return 0;
2227}
2228
2229// scan for any po_test operand modification in range given
2230static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt,
2231 int c_mode)
2232{
2233 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2234 return -1;
2235
2236 for (; i < opcnt; i++) {
2237 if (is_any_opr_modified(po_test, &ops[i], c_mode))
2238 return i;
2239 }
2240
2241 return -1;
2242}
2243
2244// scan for po_test operand[0] modification in range given
2245static int scan_for_mod_opr0(struct parsed_op *po_test,
2246 int i, int opcnt)
2247{
2248 for (; i < opcnt; i++) {
2249 if (is_opr_modified(&po_test->operand[0], &ops[i]))
2250 return i;
2251 }
2252
2253 return -1;
2254}
2255
2256static int scan_for_flag_set(int i, int magic, int *branched,
2257 int *setters, int *setter_cnt)
2258{
2259 struct label_ref *lr;
2260 int ret;
2261
2262 while (i >= 0) {
2263 if (ops[i].cc_scratch == magic) {
2264 ferr(&ops[i], "%s looped\n", __func__);
2265 return -1;
2266 }
2267 ops[i].cc_scratch = magic;
2268
2269 if (g_labels[i][0] != 0) {
2270 *branched = 1;
2271
2272 lr = &g_label_refs[i];
2273 for (; lr->next; lr = lr->next) {
2274 ret = scan_for_flag_set(lr->i, magic,
2275 branched, setters, setter_cnt);
2276 if (ret < 0)
2277 return ret;
2278 }
2279
2280 if (i > 0 && LAST_OP(i - 1)) {
2281 i = lr->i;
2282 continue;
2283 }
2284 ret = scan_for_flag_set(lr->i, magic,
2285 branched, setters, setter_cnt);
2286 if (ret < 0)
2287 return ret;
2288 }
2289 i--;
2290
2291 if (ops[i].flags & OPF_FLAGS) {
2292 setters[*setter_cnt] = i;
2293 (*setter_cnt)++;
2294 return 0;
2295 }
2296
2297 if ((ops[i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP)
2298 return -1;
2299 }
2300
2301 return -1;
2302}
2303
2304// scan back for cdq, if anything modifies edx, fail
2305static int scan_for_cdq_edx(int i)
2306{
2307 while (i >= 0) {
2308 if (g_labels[i][0] != 0) {
2309 if (g_label_refs[i].next != NULL)
2310 return -1;
2311 if (i > 0 && LAST_OP(i - 1)) {
2312 i = g_label_refs[i].i;
2313 continue;
2314 }
2315 return -1;
2316 }
2317 i--;
2318
2319 if (ops[i].op == OP_CDQ)
2320 return i;
2321
2322 if (ops[i].regmask_dst & (1 << xDX))
2323 return -1;
2324 }
2325
2326 return -1;
2327}
2328
2329static int scan_for_reg_clear(int i, int reg)
2330{
2331 while (i >= 0) {
2332 if (g_labels[i][0] != 0) {
2333 if (g_label_refs[i].next != NULL)
2334 return -1;
2335 if (i > 0 && LAST_OP(i - 1)) {
2336 i = g_label_refs[i].i;
2337 continue;
2338 }
2339 return -1;
2340 }
2341 i--;
2342
2343 if (ops[i].op == OP_XOR
2344 && ops[i].operand[0].lmod == OPLM_DWORD
2345 && ops[i].operand[0].reg == ops[i].operand[1].reg
2346 && ops[i].operand[0].reg == reg)
2347 return i;
2348
2349 if (ops[i].regmask_dst & (1 << reg))
2350 return -1;
2351 }
2352
2353 return -1;
2354}
2355
2356// scan for positive, constant esp adjust
2357static int scan_for_esp_adjust(int i, int opcnt, int *adj,
2358 int *multipath)
2359{
2360 struct parsed_op *po;
2361 int first_pop = -1;
2362
2363 *adj = *multipath = 0;
2364
2365 for (; i < opcnt; i++) {
2366 po = &ops[i];
2367
2368 if (g_labels[i][0] != 0)
2369 *multipath = 1;
2370
2371 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
2372 if (po->operand[1].type != OPT_CONST)
2373 ferr(&ops[i], "non-const esp adjust?\n");
2374 *adj += po->operand[1].val;
2375 if (*adj & 3)
2376 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
2377 return i;
2378 }
2379 else if (po->op == OP_PUSH && !(po->flags & OPF_RMD)) {
2380 //if (first_pop == -1)
2381 // first_pop = -2; // none
2382 *adj -= lmod_bytes(po, po->operand[0].lmod);
2383 }
2384 else if (po->op == OP_POP && !(po->flags & OPF_RMD)) {
2385 // seems like msvc only uses 'pop ecx' for stack realignment..
2386 if (po->operand[0].type != OPT_REG || po->operand[0].reg != xCX)
2387 break;
2388 if (first_pop == -1 && *adj >= 0)
2389 first_pop = i;
2390 *adj += lmod_bytes(po, po->operand[0].lmod);
2391 }
2392 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
2393 if (po->op == OP_JMP && po->btj == NULL) {
2394 i = po->bt_i - 1;
2395 continue;
2396 }
2397 if (po->op != OP_CALL)
2398 break;
2399 if (po->operand[0].type != OPT_LABEL)
2400 break;
2401 if (po->pp != NULL && po->pp->is_stdcall)
2402 break;
2403 }
2404 }
2405
2406 if (first_pop >= 0) {
2407 // probably 'pop ecx' was used..
2408 return first_pop;
2409 }
2410
2411 return -1;
2412}
2413
2414static void scan_fwd_set_flags(int i, int opcnt, int magic, int flags)
2415{
2416 struct parsed_op *po;
2417 int j;
2418
2419 if (i < 0)
2420 ferr(ops, "%s: followed bad branch?\n", __func__);
2421
2422 for (; i < opcnt; i++) {
2423 po = &ops[i];
2424 if (po->cc_scratch == magic)
2425 return;
2426 po->cc_scratch = magic;
2427 po->flags |= flags;
2428
2429 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2430 if (po->btj != NULL) {
2431 // jumptable
2432 for (j = 0; j < po->btj->count; j++)
2433 scan_fwd_set_flags(po->btj->d[j].bt_i, opcnt, magic, flags);
2434 return;
2435 }
2436
2437 scan_fwd_set_flags(po->bt_i, opcnt, magic, flags);
2438 if (!(po->flags & OPF_CJMP))
2439 return;
2440 }
2441 if (po->flags & OPF_TAIL)
2442 return;
2443 }
2444}
2445
2446static const struct parsed_proto *try_recover_pp(
2447 struct parsed_op *po, const struct parsed_opr *opr, int *search_instead)
2448{
2449 const struct parsed_proto *pp = NULL;
2450 char buf[256];
2451 char *p;
2452
2453 // maybe an arg of g_func?
2454 if (opr->type == OPT_REGMEM && is_stack_access(po, opr))
2455 {
2456 char ofs_reg[16] = { 0, };
2457 int arg, arg_s, arg_i;
2458 int stack_ra = 0;
2459 int offset = 0;
2460
2461 parse_stack_access(po, opr->name, ofs_reg,
2462 &offset, &stack_ra, NULL, 0);
2463 if (ofs_reg[0] != 0)
2464 ferr(po, "offset reg on arg access?\n");
2465 if (offset <= stack_ra) {
2466 // search who set the stack var instead
2467 if (search_instead != NULL)
2468 *search_instead = 1;
2469 return NULL;
2470 }
2471
2472 arg_i = (offset - stack_ra - 4) / 4;
2473 for (arg = arg_s = 0; arg < g_func_pp->argc; arg++) {
2474 if (g_func_pp->arg[arg].reg != NULL)
2475 continue;
2476 if (arg_s == arg_i)
2477 break;
2478 arg_s++;
2479 }
2480 if (arg == g_func_pp->argc)
2481 ferr(po, "stack arg %d not in prototype?\n", arg_i);
2482
2483 pp = g_func_pp->arg[arg].fptr;
2484 if (pp == NULL)
2485 ferr(po, "icall sa: arg%d is not a fptr?\n", arg + 1);
2486 check_func_pp(po, pp, "icall arg");
2487 }
2488 else if (opr->type == OPT_REGMEM && strchr(opr->name + 1, '[')) {
2489 // label[index]
2490 p = strchr(opr->name + 1, '[');
2491 memcpy(buf, opr->name, p - opr->name);
2492 buf[p - opr->name] = 0;
2493 pp = proto_parse(g_fhdr, buf, 0);
2494 }
2495 else if (opr->type == OPT_OFFSET || opr->type == OPT_LABEL) {
2496 pp = proto_parse(g_fhdr, opr->name, 0);
2497 if (pp == NULL)
2498 ferr(po, "proto_parse failed for icall from '%s'\n", opr->name);
2499 check_func_pp(po, pp, "reg-fptr ref");
2500 }
2501
2502 return pp;
2503}
2504
2505static void scan_for_call_type(int i, const struct parsed_opr *opr,
2506 int magic, const struct parsed_proto **pp_found, int *multi)
2507{
2508 const struct parsed_proto *pp = NULL;
2509 struct parsed_op *po;
2510 struct label_ref *lr;
2511
2512 ops[i].cc_scratch = magic;
2513
2514 while (1) {
2515 if (g_labels[i][0] != 0) {
2516 lr = &g_label_refs[i];
2517 for (; lr != NULL; lr = lr->next)
2518 scan_for_call_type(lr->i, opr, magic, pp_found, multi);
2519 if (i > 0 && LAST_OP(i - 1))
2520 return;
2521 }
2522
2523 i--;
2524 if (i < 0)
2525 break;
2526
2527 if (ops[i].cc_scratch == magic)
2528 return;
2529 ops[i].cc_scratch = magic;
2530
2531 if (!(ops[i].flags & OPF_DATA))
2532 continue;
2533 if (!is_opr_modified(opr, &ops[i]))
2534 continue;
2535 if (ops[i].op != OP_MOV && ops[i].op != OP_LEA) {
2536 // most probably trashed by some processing
2537 *pp_found = NULL;
2538 return;
2539 }
2540
2541 opr = &ops[i].operand[1];
2542 if (opr->type != OPT_REG)
2543 break;
2544 }
2545
2546 po = (i >= 0) ? &ops[i] : ops;
2547
2548 if (i < 0) {
2549 // reached the top - can only be an arg-reg
2550 if (opr->type != OPT_REG)
2551 return;
2552
2553 for (i = 0; i < g_func_pp->argc; i++) {
2554 if (g_func_pp->arg[i].reg == NULL)
2555 continue;
2556 if (IS(opr->name, g_func_pp->arg[i].reg))
2557 break;
2558 }
2559 if (i == g_func_pp->argc)
2560 return;
2561 pp = g_func_pp->arg[i].fptr;
2562 if (pp == NULL)
2563 ferr(po, "icall: arg%d (%s) is not a fptr?\n",
2564 i + 1, g_func_pp->arg[i].reg);
2565 check_func_pp(po, pp, "icall reg-arg");
2566 }
2567 else
2568 pp = try_recover_pp(po, opr, NULL);
2569
2570 if (*pp_found != NULL && pp != NULL && *pp_found != pp) {
2571 if (!IS((*pp_found)->ret_type.name, pp->ret_type.name)
2572 || (*pp_found)->is_stdcall != pp->is_stdcall
2573 || (*pp_found)->is_fptr != pp->is_fptr
2574 || (*pp_found)->argc != pp->argc
2575 || (*pp_found)->argc_reg != pp->argc_reg
2576 || (*pp_found)->argc_stack != pp->argc_stack)
2577 {
2578 ferr(po, "icall: parsed_proto mismatch\n");
2579 }
2580 *multi = 1;
2581 }
2582 if (pp != NULL)
2583 *pp_found = pp;
2584}
2585
2586static const struct parsed_proto *resolve_icall(int i, int opcnt,
2587 int *multi_src)
2588{
2589 const struct parsed_proto *pp = NULL;
2590 int search_advice = 0;
2591
2592 *multi_src = 0;
2593
2594 switch (ops[i].operand[0].type) {
2595 case OPT_REGMEM:
2596 case OPT_LABEL:
2597 case OPT_OFFSET:
2598 pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice);
2599 if (!search_advice)
2600 break;
2601 // fallthrough
2602 default:
2603 scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp,
2604 multi_src);
2605 break;
2606 }
2607
2608 return pp;
2609}
2610
2611// find an instruction that changed opr before i op
2612// *op_i must be set to -1
2613static int resolve_origin(int i, const struct parsed_opr *opr,
2614 int magic, int *op_i)
2615{
2616 struct label_ref *lr;
2617 int ret = 0;
2618
2619 ops[i].cc_scratch = magic;
2620
2621 while (1) {
2622 if (g_labels[i][0] != 0) {
2623 lr = &g_label_refs[i];
2624 for (; lr != NULL; lr = lr->next)
2625 ret |= resolve_origin(lr->i, opr, magic, op_i);
2626 if (i > 0 && LAST_OP(i - 1))
2627 return ret;
2628 }
2629
2630 i--;
2631 if (i < 0)
2632 return -1;
2633
2634 if (ops[i].cc_scratch == magic)
2635 return 0;
2636 ops[i].cc_scratch = magic;
2637
2638 if (!(ops[i].flags & OPF_DATA))
2639 continue;
2640 if (!is_opr_modified(opr, &ops[i]))
2641 continue;
2642
2643 if (*op_i >= 0) {
2644 if (*op_i == i)
2645 return 1;
2646 // XXX: could check if the other op does the same
2647 return -1;
2648 }
2649
2650 *op_i = i;
2651 return 1;
2652 }
2653}
2654
2655static int try_resolve_const(int i, const struct parsed_opr *opr,
2656 int magic, unsigned int *val)
2657{
2658 int s_i = -1;
2659 int ret = 0;
2660
2661 ret = resolve_origin(i, opr, magic, &s_i);
2662 if (ret == 1) {
2663 i = s_i;
2664 if (ops[i].op != OP_MOV && ops[i].operand[1].type != OPT_CONST)
2665 return -1;
2666
2667 *val = ops[i].operand[1].val;
2668 return 1;
2669 }
2670
2671 return -1;
2672}
2673
2674static int collect_call_args_r(struct parsed_op *po, int i,
2675 struct parsed_proto *pp, int *regmask, int *save_arg_vars, int arg,
2676 int magic, int need_op_saving, int may_reuse)
2677{
2678 struct parsed_proto *pp_tmp;
2679 struct label_ref *lr;
2680 int need_to_save_current;
2681 int save_args;
2682 int ret = 0;
2683 int reg;
2684 char buf[32];
2685 int j, k;
2686
2687 if (i < 0) {
2688 ferr(po, "dead label encountered\n");
2689 return -1;
2690 }
2691
2692 for (; arg < pp->argc; arg++)
2693 if (pp->arg[arg].reg == NULL)
2694 break;
2695 magic = (magic & 0xffffff) | (arg << 24);
2696
2697 for (j = i; j >= 0 && (arg < pp->argc || pp->is_unresolved); )
2698 {
2699 if (((ops[j].cc_scratch ^ magic) & 0xffffff) == 0) {
2700 if (ops[j].cc_scratch != magic) {
2701 ferr(&ops[j], "arg collect hit same path with diff args for %s\n",
2702 pp->name);
2703 return -1;
2704 }
2705 // ok: have already been here
2706 return 0;
2707 }
2708 ops[j].cc_scratch = magic;
2709
2710 if (g_labels[j][0] != 0 && g_label_refs[j].i != -1) {
2711 lr = &g_label_refs[j];
2712 if (lr->next != NULL)
2713 need_op_saving = 1;
2714 for (; lr->next; lr = lr->next) {
2715 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
2716 may_reuse = 1;
2717 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
2718 arg, magic, need_op_saving, may_reuse);
2719 if (ret < 0)
2720 return ret;
2721 }
2722
2723 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
2724 may_reuse = 1;
2725 if (j > 0 && LAST_OP(j - 1)) {
2726 // follow last branch in reverse
2727 j = lr->i;
2728 continue;
2729 }
2730 need_op_saving = 1;
2731 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
2732 arg, magic, need_op_saving, may_reuse);
2733 if (ret < 0)
2734 return ret;
2735 }
2736 j--;
2737
2738 if (ops[j].op == OP_CALL)
2739 {
2740 if (pp->is_unresolved)
2741 break;
2742
2743 pp_tmp = ops[j].pp;
2744 if (pp_tmp == NULL)
2745 ferr(po, "arg collect hit unparsed call '%s'\n",
2746 ops[j].operand[0].name);
2747 if (may_reuse && pp_tmp->argc_stack > 0)
2748 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
2749 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
2750 }
2751 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP) {
2752 if (pp->is_unresolved)
2753 break;
2754
2755 ferr(po, "arg collect %d/%d hit esp adjust\n",
2756 arg, pp->argc);
2757 }
2758 else if (ops[j].op == OP_POP) {
2759 if (pp->is_unresolved)
2760 break;
2761
2762 ferr(po, "arg collect %d/%d hit pop\n", arg, pp->argc);
2763 }
2764 else if (ops[j].flags & OPF_CJMP)
2765 {
2766 if (pp->is_unresolved)
2767 break;
2768
2769 may_reuse = 1;
2770 }
2771 else if (ops[j].op == OP_PUSH && !(ops[j].flags & OPF_FARG))
2772 {
2773 if (pp->is_unresolved && (ops[j].flags & OPF_RMD))
2774 break;
2775
2776 pp->arg[arg].datap = &ops[j];
2777 need_to_save_current = 0;
2778 save_args = 0;
2779 reg = -1;
2780 if (ops[j].operand[0].type == OPT_REG)
2781 reg = ops[j].operand[0].reg;
2782
2783 if (!need_op_saving) {
2784 ret = scan_for_mod(&ops[j], j + 1, i, 1);
2785 need_to_save_current = (ret >= 0);
2786 }
2787 if (need_op_saving || need_to_save_current) {
2788 // mark this push as one that needs operand saving
2789 ops[j].flags &= ~OPF_RMD;
2790 if (ops[j].p_argnum == 0) {
2791 ops[j].p_argnum = arg + 1;
2792 save_args |= 1 << arg;
2793 }
2794 else if (ops[j].p_argnum < arg + 1)
2795 ferr(&ops[j], "p_argnum conflict (%d<%d) for '%s'\n",
2796 ops[j].p_argnum, arg + 1, pp->name);
2797 }
2798 else if (ops[j].p_argnum == 0)
2799 ops[j].flags |= OPF_RMD;
2800
2801 // some PUSHes are reused by different calls on other branches,
2802 // but that can't happen if we didn't branch, so they
2803 // can be removed from future searches (handles nested calls)
2804 if (!may_reuse)
2805 ops[j].flags |= OPF_FARG;
2806
2807 ops[j].flags &= ~OPF_RSAVE;
2808
2809 // check for __VALIST
2810 if (!pp->is_unresolved && pp->arg[arg].type.is_va_list) {
2811 k = -1;
2812 ret = resolve_origin(j, &ops[j].operand[0], magic + 1, &k);
2813 if (ret == 1 && k >= 0)
2814 {
2815 if (ops[k].op == OP_LEA) {
2816 snprintf(buf, sizeof(buf), "arg_%X",
2817 g_func_pp->argc_stack * 4);
2818 if (!g_func_pp->is_vararg
2819 || strstr(ops[k].operand[1].name, buf))
2820 {
2821 ops[k].flags |= OPF_RMD;
2822 ops[j].flags |= OPF_RMD | OPF_VAPUSH;
2823 save_args &= ~(1 << arg);
2824 reg = -1;
2825 }
2826 else
2827 ferr(&ops[j], "lea va_list used, but no vararg?\n");
2828 }
2829 // check for va_list from g_func_pp arg too
2830 else if (ops[k].op == OP_MOV
2831 && is_stack_access(&ops[k], &ops[k].operand[1]))
2832 {
2833 ret = stack_frame_access(&ops[k], &ops[k].operand[1],
2834 buf, sizeof(buf), ops[k].operand[1].name, "", 1, 0);
2835 if (ret >= 0) {
2836 ops[k].flags |= OPF_RMD;
2837 ops[j].flags |= OPF_RMD;
2838 ops[j].p_argpass = ret + 1;
2839 save_args &= ~(1 << arg);
2840 reg = -1;
2841 }
2842 }
2843 }
2844 }
2845
2846 *save_arg_vars |= save_args;
2847
2848 // tracking reg usage
2849 if (reg >= 0)
2850 *regmask |= 1 << reg;
2851
2852 arg++;
2853 if (!pp->is_unresolved) {
2854 // next arg
2855 for (; arg < pp->argc; arg++)
2856 if (pp->arg[arg].reg == NULL)
2857 break;
2858 }
2859 magic = (magic & 0xffffff) | (arg << 24);
2860 }
2861 }
2862
2863 if (arg < pp->argc) {
2864 ferr(po, "arg collect failed for '%s': %d/%d\n",
2865 pp->name, arg, pp->argc);
2866 return -1;
2867 }
2868
2869 return arg;
2870}
2871
2872static int collect_call_args(struct parsed_op *po, int i,
2873 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
2874 int magic)
2875{
2876 int ret;
2877 int a;
2878
2879 ret = collect_call_args_r(po, i, pp, regmask, save_arg_vars,
2880 0, magic, 0, 0);
2881 if (ret < 0)
2882 return ret;
2883
2884 if (pp->is_unresolved) {
2885 pp->argc += ret;
2886 pp->argc_stack += ret;
2887 for (a = 0; a < pp->argc; a++)
2888 if (pp->arg[a].type.name == NULL)
2889 pp->arg[a].type.name = strdup("int");
2890 }
2891
2892 return ret;
2893}
2894
2895// early check for tail call or branch back
2896static int is_like_tailjmp(int j)
2897{
2898 if (!(ops[j].flags & OPF_JMP))
2899 return 0;
2900
2901 if (ops[j].op == OP_JMP && !ops[j].operand[0].had_ds)
2902 // probably local branch back..
2903 return 1;
2904 if (ops[j].op == OP_CALL)
2905 // probably noreturn call..
2906 return 1;
2907
2908 return 0;
2909}
2910
2911static void pp_insert_reg_arg(struct parsed_proto *pp, const char *reg)
2912{
2913 int i;
2914
2915 for (i = 0; i < pp->argc; i++)
2916 if (pp->arg[i].reg == NULL)
2917 break;
2918
2919 if (pp->argc_stack)
2920 memmove(&pp->arg[i + 1], &pp->arg[i],
2921 sizeof(pp->arg[0]) * pp->argc_stack);
2922 memset(&pp->arg[i], 0, sizeof(pp->arg[i]));
2923 pp->arg[i].reg = strdup(reg);
2924 pp->arg[i].type.name = strdup("int");
2925 pp->argc++;
2926 pp->argc_reg++;
2927}
2928
2929static void add_label_ref(struct label_ref *lr, int op_i)
2930{
2931 struct label_ref *lr_new;
2932
2933 if (lr->i == -1) {
2934 lr->i = op_i;
2935 return;
2936 }
2937
2938 lr_new = calloc(1, sizeof(*lr_new));
2939 lr_new->i = op_i;
2940 lr_new->next = lr->next;
2941 lr->next = lr_new;
2942}
2943
2944static void output_std_flags(FILE *fout, struct parsed_op *po,
2945 int *pfomask, const char *dst_opr_text)
2946{
2947 if (*pfomask & (1 << PFO_Z)) {
2948 fprintf(fout, "\n cond_z = (%s%s == 0);",
2949 lmod_cast_u(po, po->operand[0].lmod), dst_opr_text);
2950 *pfomask &= ~(1 << PFO_Z);
2951 }
2952 if (*pfomask & (1 << PFO_S)) {
2953 fprintf(fout, "\n cond_s = (%s%s < 0);",
2954 lmod_cast_s(po, po->operand[0].lmod), dst_opr_text);
2955 *pfomask &= ~(1 << PFO_S);
2956 }
2957}
2958
2959static void output_pp_attrs(FILE *fout, const struct parsed_proto *pp,
2960 int is_noreturn)
2961{
2962 if (pp->is_fastcall)
2963 fprintf(fout, "__fastcall ");
2964 else if (pp->is_stdcall && pp->argc_reg == 0)
2965 fprintf(fout, "__stdcall ");
2966 if (pp->is_noreturn || is_noreturn)
2967 fprintf(fout, "noreturn ");
2968}
2969
2970static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
2971{
2972 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
2973 struct parsed_opr *last_arith_dst = NULL;
2974 char buf1[256], buf2[256], buf3[256], cast[64];
2975 const struct parsed_proto *pp_c;
2976 struct parsed_proto *pp, *pp_tmp;
2977 struct parsed_data *pd;
2978 const char *tmpname;
2979 unsigned int uval;
2980 int save_arg_vars = 0;
2981 int cond_vars = 0;
2982 int need_tmp_var = 0;
2983 int need_tmp64 = 0;
2984 int had_decl = 0;
2985 int label_pending = 0;
2986 int regmask_save = 0;
2987 int regmask_arg = 0;
2988 int regmask_now = 0;
2989 int regmask_init = 0;
2990 int regmask = 0;
2991 int pfomask = 0;
2992 int found = 0;
2993 int depth = 0;
2994 int no_output;
2995 int i, j, l;
2996 int arg;
2997 int reg;
2998 int ret;
2999
3000 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
3001 g_stack_frame_used = 0;
3002
3003 g_func_pp = proto_parse(fhdr, funcn, 0);
3004 if (g_func_pp == NULL)
3005 ferr(ops, "proto_parse failed for '%s'\n", funcn);
3006
3007 for (i = 0; i < g_func_pp->argc; i++) {
3008 if (g_func_pp->arg[i].reg != NULL) {
3009 reg = char_array_i(regs_r32,
3010 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
3011 if (reg < 0)
3012 ferr(ops, "arg '%s' is not a reg?\n", g_func_pp->arg[i].reg);
3013 regmask_arg |= 1 << reg;
3014 }
3015 }
3016
3017 // pass1:
3018 // - handle ebp/esp frame, remove ops related to it
3019 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
3020 && ops[1].op == OP_MOV
3021 && IS(opr_name(&ops[1], 0), "ebp")
3022 && IS(opr_name(&ops[1], 1), "esp"))
3023 {
3024 int ecx_push = 0;
3025
3026 g_bp_frame = 1;
3027 ops[0].flags |= OPF_RMD;
3028 ops[1].flags |= OPF_RMD;
3029 i = 2;
3030
3031 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
3032 g_stack_fsz = opr_const(&ops[2], 1);
3033 ops[2].flags |= OPF_RMD;
3034 i++;
3035 }
3036 else {
3037 // another way msvc builds stack frame..
3038 i = 2;
3039 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
3040 g_stack_fsz += 4;
3041 ops[i].flags |= OPF_RMD;
3042 ecx_push++;
3043 i++;
3044 }
3045 // and another way..
3046 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
3047 && ops[i].operand[1].type == OPT_CONST
3048 && ops[i + 1].op == OP_CALL
3049 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
3050 {
3051 g_stack_fsz += ops[i].operand[1].val;
3052 ops[i].flags |= OPF_RMD;
3053 i++;
3054 ops[i].flags |= OPF_RMD;
3055 i++;
3056 }
3057 }
3058
3059 found = 0;
3060 do {
3061 for (; i < opcnt; i++)
3062 if (ops[i].op == OP_RET)
3063 break;
3064 j = i - 1;
3065 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
3066 if (found && is_like_tailjmp(j))
3067 break;
3068 j--;
3069 }
3070
3071 if ((ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ebp"))
3072 || ops[j].op == OP_LEAVE)
3073 {
3074 ops[j].flags |= OPF_RMD;
3075 }
3076 else if (!(g_ida_func_attr & IDAFA_NORETURN))
3077 ferr(&ops[j], "'pop ebp' expected\n");
3078
3079 if (g_stack_fsz != 0) {
3080 if (ops[j - 1].op == OP_MOV
3081 && IS(opr_name(&ops[j - 1], 0), "esp")
3082 && IS(opr_name(&ops[j - 1], 1), "ebp"))
3083 {
3084 ops[j - 1].flags |= OPF_RMD;
3085 }
3086 else if (ops[j].op != OP_LEAVE
3087 && !(g_ida_func_attr & IDAFA_NORETURN))
3088 {
3089 ferr(&ops[j - 1], "esp restore expected\n");
3090 }
3091
3092 if (ecx_push && ops[j - 2].op == OP_POP
3093 && IS(opr_name(&ops[j - 2], 0), "ecx"))
3094 {
3095 ferr(&ops[j - 2], "unexpected ecx pop\n");
3096 }
3097 }
3098
3099 found = 1;
3100 i++;
3101 } while (i < opcnt);
3102 }
3103 else {
3104 int ecx_push = 0, esp_sub = 0;
3105
3106 i = 0;
3107 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
3108 ops[i].flags |= OPF_RMD;
3109 g_stack_fsz += 4;
3110 ecx_push++;
3111 i++;
3112 }
3113
3114 for (; i < opcnt; i++) {
3115 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
3116 break;
3117 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
3118 && ops[i].operand[1].type == OPT_CONST)
3119 {
3120 g_stack_fsz = ops[i].operand[1].val;
3121 ops[i].flags |= OPF_RMD;
3122 esp_sub = 1;
3123 break;
3124 }
3125 }
3126
3127 found = 0;
3128 if (ecx_push || esp_sub)
3129 {
3130 g_sp_frame = 1;
3131
3132 i++;
3133 do {
3134 for (; i < opcnt; i++)
3135 if (ops[i].op == OP_RET)
3136 break;
3137 j = i - 1;
3138 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
3139 if (found && is_like_tailjmp(j))
3140 break;
3141 j--;
3142 }
3143
3144 if (ecx_push > 0) {
3145 for (l = 0; l < ecx_push; l++) {
3146 if (ops[j].op != OP_POP
3147 || !IS(opr_name(&ops[j], 0), "ecx"))
3148 {
3149 ferr(&ops[j], "'pop ecx' expected\n");
3150 }
3151 ops[j].flags |= OPF_RMD;
3152 j--;
3153 }
3154
3155 found = 1;
3156 }
3157
3158 if (esp_sub) {
3159 if (ops[j].op != OP_ADD
3160 || !IS(opr_name(&ops[j], 0), "esp")
3161 || ops[j].operand[1].type != OPT_CONST
3162 || ops[j].operand[1].val != g_stack_fsz)
3163 ferr(&ops[j], "'add esp' expected\n");
3164 ops[j].flags |= OPF_RMD;
3165
3166 found = 1;
3167 }
3168
3169 i++;
3170 } while (i < opcnt);
3171 }
3172 }
3173
3174 // pass2:
3175 // - parse calls with labels
3176 // - resolve all branches
3177 for (i = 0; i < opcnt; i++)
3178 {
3179 po = &ops[i];
3180 po->bt_i = -1;
3181 po->btj = NULL;
3182
3183 if (po->flags & OPF_RMD)
3184 continue;
3185
3186 if (po->op == OP_CALL) {
3187 pp = NULL;
3188
3189 if (po->operand[0].type == OPT_LABEL) {
3190 tmpname = opr_name(po, 0);
3191 if (IS_START(tmpname, "loc_"))
3192 ferr(po, "call to loc_*\n");
3193 pp_c = proto_parse(fhdr, tmpname, 0);
3194 if (pp_c == NULL)
3195 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
3196
3197 pp = proto_clone(pp_c);
3198 my_assert_not(pp, NULL);
3199 }
3200 else if (po->datap != NULL) {
3201 pp = calloc(1, sizeof(*pp));
3202 my_assert_not(pp, NULL);
3203
3204 ret = parse_protostr(po->datap, pp);
3205 if (ret < 0)
3206 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
3207 free(po->datap);
3208 po->datap = NULL;
3209 }
3210
3211 if (pp != NULL) {
3212 if (pp->is_fptr)
3213 check_func_pp(po, pp, "fptr var call");
3214 if (pp->is_noreturn)
3215 po->flags |= OPF_TAIL;
3216 }
3217 po->pp = pp;
3218 continue;
3219 }
3220
3221 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
3222 continue;
3223
3224 if (po->operand[0].type == OPT_REGMEM) {
3225 char *p = strchr(po->operand[0].name, '[');
3226 if (p == NULL)
3227 goto tailcall;
3228 ret = p - po->operand[0].name;
3229 strncpy(buf1, po->operand[0].name, ret);
3230 buf1[ret] = 0;
3231
3232 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
3233 if (IS(g_func_pd[j].label, buf1)) {
3234 pd = &g_func_pd[j];
3235 break;
3236 }
3237 }
3238 if (pd == NULL)
3239 //ferr(po, "label '%s' not parsed?\n", buf1);
3240 goto tailcall;
3241 if (pd->type != OPT_OFFSET)
3242 ferr(po, "label '%s' with non-offset data?\n", buf1);
3243
3244 // find all labels, link
3245 for (j = 0; j < pd->count; j++) {
3246 for (l = 0; l < opcnt; l++) {
3247 if (g_labels[l][0] && IS(g_labels[l], pd->d[j].u.label)) {
3248 add_label_ref(&g_label_refs[l], i);
3249 pd->d[j].bt_i = l;
3250 break;
3251 }
3252 }
3253 }
3254
3255 po->btj = pd;
3256 continue;
3257 }
3258
3259 for (l = 0; l < opcnt; l++) {
3260 if (g_labels[l][0] && IS(po->operand[0].name, g_labels[l])) {
3261 if (l == i + 1 && po->op == OP_JMP) {
3262 // yet another alignment type..
3263 po->flags |= OPF_RMD;
3264 break;
3265 }
3266 add_label_ref(&g_label_refs[l], i);
3267 po->bt_i = l;
3268 break;
3269 }
3270 }
3271
3272 if (po->bt_i != -1 || (po->flags & OPF_RMD))
3273 continue;
3274
3275 if (po->operand[0].type == OPT_LABEL)
3276 // assume tail call
3277 goto tailcall;
3278
3279 ferr(po, "unhandled branch\n");
3280
3281tailcall:
3282 po->op = OP_CALL;
3283 po->flags |= OPF_TAIL;
3284 if (i > 0 && ops[i - 1].op == OP_POP)
3285 po->flags |= OPF_ATAIL;
3286 i--; // reprocess
3287 }
3288
3289 // pass3:
3290 // - remove dead labels
3291 // - process calls
3292 for (i = 0; i < opcnt; i++)
3293 {
3294 if (g_labels[i][0] != 0 && g_label_refs[i].i == -1)
3295 g_labels[i][0] = 0;
3296
3297 po = &ops[i];
3298 if (po->flags & OPF_RMD)
3299 continue;
3300
3301 if (po->op == OP_CALL)
3302 {
3303 tmpname = opr_name(po, 0);
3304 pp = po->pp;
3305 if (pp == NULL)
3306 {
3307 // indirect call
3308 pp_c = resolve_icall(i, opcnt, &l);
3309 if (pp_c != NULL) {
3310 if (!pp_c->is_func && !pp_c->is_fptr)
3311 ferr(po, "call to non-func: %s\n", pp_c->name);
3312 pp = proto_clone(pp_c);
3313 my_assert_not(pp, NULL);
3314 if (l)
3315 // not resolved just to single func
3316 pp->is_fptr = 1;
3317
3318 switch (po->operand[0].type) {
3319 case OPT_REG:
3320 // we resolved this call and no longer need the register
3321 po->regmask_src &= ~(1 << po->operand[0].reg);
3322 break;
3323 case OPT_REGMEM:
3324 pp->is_fptr = 1;
3325 break;
3326 default:
3327 break;
3328 }
3329 }
3330 if (pp == NULL) {
3331 pp = calloc(1, sizeof(*pp));
3332 my_assert_not(pp, NULL);
3333 pp->is_fptr = 1;
3334 ret = scan_for_esp_adjust(i + 1, opcnt, &j, &l);
3335 if (ret < 0) {
3336 if (!g_allow_regfunc)
3337 ferr(po, "non-__cdecl indirect call unhandled yet\n");
3338 pp->is_unresolved = 1;
3339 j = 0;
3340 }
3341 j /= 4;
3342 if (j > ARRAY_SIZE(pp->arg))
3343 ferr(po, "esp adjust too large: %d\n", j);
3344 pp->ret_type.name = strdup("int");
3345 pp->argc = pp->argc_stack = j;
3346 for (arg = 0; arg < pp->argc; arg++)
3347 pp->arg[arg].type.name = strdup("int");
3348 }
3349 po->pp = pp;
3350 }
3351
3352 // look for and make use of esp adjust
3353 ret = -1;
3354 if (!pp->is_stdcall && pp->argc_stack > 0)
3355 ret = scan_for_esp_adjust(i + 1, opcnt, &j, &l);
3356 if (ret >= 0) {
3357 if (pp->is_vararg) {
3358 if (j / 4 < pp->argc_stack)
3359 ferr(po, "esp adjust is too small: %x < %x\n",
3360 j, pp->argc_stack * 4);
3361 // modify pp to make it have varargs as normal args
3362 arg = pp->argc;
3363 pp->argc += j / 4 - pp->argc_stack;
3364 for (; arg < pp->argc; arg++) {
3365 pp->arg[arg].type.name = strdup("int");
3366 pp->argc_stack++;
3367 }
3368 if (pp->argc > ARRAY_SIZE(pp->arg))
3369 ferr(po, "too many args for '%s'\n", tmpname);
3370 }
3371 if (pp->argc_stack != j / 4)
3372 ferr(po, "stack tracking failed for '%s': %x %x\n",
3373 tmpname, pp->argc_stack * 4, j);
3374
3375 ops[ret].flags |= OPF_RMD;
3376 if (ops[ret].op == OP_POP && j > 4) {
3377 // deal with multi-pop stack adjust
3378 j = pp->argc_stack;
3379 while (ops[ret].op == OP_POP && j > 0 && ret < opcnt) {
3380 ops[ret].flags |= OPF_RMD;
3381 j--;
3382 ret++;
3383 }
3384 }
3385 else if (!l) {
3386 // a bit of a hack, but deals with use of
3387 // single adj for multiple calls
3388 ops[ret].operand[1].val -= j;
3389 }
3390 }
3391 else if (pp->is_vararg)
3392 ferr(po, "missing esp_adjust for vararg func '%s'\n",
3393 pp->name);
3394
3395 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
3396 // since we know the args, collect them
3397 collect_call_args(po, i, pp, &regmask, &save_arg_vars,
3398 i + opcnt * 2);
3399 }
3400
3401 if (strstr(pp->ret_type.name, "int64"))
3402 need_tmp64 = 1;
3403 }
3404 }
3405
3406 // pass4:
3407 // - find POPs for PUSHes, rm both
3408 // - scan for STD/CLD, propagate DF
3409 // - scan for all used registers
3410 // - find flag set ops for their users
3411 // - do unreselved calls
3412 // - declare indirect functions
3413 for (i = 0; i < opcnt; i++) {
3414 po = &ops[i];
3415 if (po->flags & OPF_RMD)
3416 continue;
3417
3418 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE)) {
3419 reg = po->operand[0].reg;
3420 if (!(regmask & (1 << reg)))
3421 // not a reg save after all, rerun scan_for_pop
3422 po->flags &= ~OPF_RSAVE;
3423 else
3424 regmask_save |= 1 << reg;
3425 }
3426
3427 if (po->op == OP_PUSH && po->p_argnum == 0
3428 && !(po->flags & OPF_RSAVE) && !g_func_pp->is_userstack)
3429 {
3430 if (po->operand[0].type == OPT_REG)
3431 {
3432 reg = po->operand[0].reg;
3433 if (reg < 0)
3434 ferr(po, "reg not set for push?\n");
3435
3436 depth = 0;
3437 ret = scan_for_pop(i + 1, opcnt,
3438 po->operand[0].name, i + opcnt * 3, 0, &depth, 0);
3439 if (ret == 1) {
3440 if (depth > 1)
3441 ferr(po, "too much depth: %d\n", depth);
3442
3443 po->flags |= OPF_RMD;
3444 scan_for_pop(i + 1, opcnt, po->operand[0].name,
3445 i + opcnt * 4, 0, &depth, 1);
3446 continue;
3447 }
3448 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
3449 if (ret == 0) {
3450 arg = OPF_RMD;
3451 if (regmask & (1 << reg)) {
3452 if (regmask_save & (1 << reg))
3453 ferr(po, "%s already saved?\n", po->operand[0].name);
3454 arg = OPF_RSAVE;
3455 }
3456 po->flags |= arg;
3457 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg);
3458 continue;
3459 }
3460 }
3461 else if (po->operand[0].type == OPT_CONST) {
3462 for (j = i + 1; j < opcnt; j++) {
3463 if ((ops[j].flags & (OPF_JMP|OPF_TAIL|OPF_RSAVE))
3464 || ops[j].op == OP_PUSH || g_labels[i][0] != 0)
3465 {
3466 break;
3467 }
3468
3469 if (!(ops[j].flags & OPF_RMD) && ops[j].op == OP_POP)
3470 {
3471 po->flags |= OPF_RMD;
3472 ops[j].datap = po;
3473 break;
3474 }
3475 }
3476 }
3477 }
3478
3479 if (po->op == OP_STD) {
3480 po->flags |= OPF_DF | OPF_RMD;
3481 scan_propagate_df(i + 1, opcnt);
3482 }
3483
3484 regmask_now = po->regmask_src | po->regmask_dst;
3485 if (regmask_now & (1 << xBP)) {
3486 if (g_bp_frame && !(po->flags & OPF_EBP_S)) {
3487 if (po->regmask_dst & (1 << xBP))
3488 // compiler decided to drop bp frame and use ebp as scratch
3489 scan_fwd_set_flags(i + 1, opcnt, i + opcnt * 5, OPF_EBP_S);
3490 else
3491 regmask_now &= ~(1 << xBP);
3492 }
3493 }
3494
3495 regmask |= regmask_now;
3496
3497 if (po->flags & OPF_CC)
3498 {
3499 int setters[16], cnt = 0, branched = 0;
3500
3501 ret = scan_for_flag_set(i, i + opcnt * 6,
3502 &branched, setters, &cnt);
3503 if (ret < 0 || cnt <= 0)
3504 ferr(po, "unable to trace flag setter(s)\n");
3505 if (cnt > ARRAY_SIZE(setters))
3506 ferr(po, "too many flag setters\n");
3507
3508 for (j = 0; j < cnt; j++)
3509 {
3510 tmp_op = &ops[setters[j]]; // flag setter
3511 pfomask = 0;
3512
3513 // to get nicer code, we try to delay test and cmp;
3514 // if we can't because of operand modification, or if we
3515 // have arith op, or branch, make it calculate flags explicitly
3516 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP)
3517 {
3518 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i, 0) >= 0)
3519 pfomask = 1 << po->pfo;
3520 }
3521 else if (tmp_op->op == OP_CMPS || tmp_op->op == OP_SCAS) {
3522 pfomask = 1 << po->pfo;
3523 }
3524 else {
3525 // see if we'll be able to handle based on op result
3526 if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR
3527 && po->pfo != PFO_Z && po->pfo != PFO_S
3528 && po->pfo != PFO_P)
3529 || branched
3530 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
3531 {
3532 pfomask = 1 << po->pfo;
3533 }
3534
3535 if (tmp_op->op == OP_ADD && po->pfo == PFO_C) {
3536 propagate_lmod(tmp_op, &tmp_op->operand[0],
3537 &tmp_op->operand[1]);
3538 if (tmp_op->operand[0].lmod == OPLM_DWORD)
3539 need_tmp64 = 1;
3540 }
3541 }
3542 if (pfomask) {
3543 tmp_op->pfomask |= pfomask;
3544 cond_vars |= pfomask;
3545 }
3546 // note: may overwrite, currently not a problem
3547 po->datap = tmp_op;
3548 }
3549
3550 if (po->op == OP_RCL || po->op == OP_RCR
3551 || po->op == OP_ADC || po->op == OP_SBB)
3552 cond_vars |= 1 << PFO_C;
3553 }
3554
3555 if (po->op == OP_CMPS || po->op == OP_SCAS) {
3556 cond_vars |= 1 << PFO_Z;
3557 }
3558 else if (po->op == OP_MUL
3559 || (po->op == OP_IMUL && po->operand_cnt == 1))
3560 {
3561 if (po->operand[0].lmod == OPLM_DWORD)
3562 need_tmp64 = 1;
3563 }
3564 else if (po->op == OP_CALL) {
3565 pp = po->pp;
3566 if (pp == NULL)
3567 ferr(po, "NULL pp\n");
3568
3569 if (pp->is_unresolved) {
3570 int regmask_stack = 0;
3571 collect_call_args(po, i, pp, &regmask, &save_arg_vars,
3572 i + opcnt * 2);
3573
3574 // this is pretty rough guess:
3575 // see ecx and edx were pushed (and not their saved versions)
3576 for (arg = 0; arg < pp->argc; arg++) {
3577 if (pp->arg[arg].reg != NULL)
3578 continue;
3579
3580 tmp_op = pp->arg[arg].datap;
3581 if (tmp_op == NULL)
3582 ferr(po, "parsed_op missing for arg%d\n", arg);
3583 if (tmp_op->p_argnum == 0 && tmp_op->operand[0].type == OPT_REG)
3584 regmask_stack |= 1 << tmp_op->operand[0].reg;
3585 }
3586
3587 if (!((regmask_stack & (1 << xCX))
3588 && (regmask_stack & (1 << xDX))))
3589 {
3590 if (pp->argc_stack != 0
3591 || ((regmask | regmask_arg) & ((1 << xCX)|(1 << xDX))))
3592 {
3593 pp_insert_reg_arg(pp, "ecx");
3594 pp->is_fastcall = 1;
3595 regmask_init |= 1 << xCX;
3596 regmask |= 1 << xCX;
3597 }
3598 if (pp->argc_stack != 0
3599 || ((regmask | regmask_arg) & (1 << xDX)))
3600 {
3601 pp_insert_reg_arg(pp, "edx");
3602 regmask_init |= 1 << xDX;
3603 regmask |= 1 << xDX;
3604 }
3605 }
3606
3607 // note: __cdecl doesn't fall into is_unresolved category
3608 if (pp->argc_stack > 0)
3609 pp->is_stdcall = 1;
3610 }
3611
3612 for (arg = 0; arg < pp->argc; arg++) {
3613 if (pp->arg[arg].reg != NULL) {
3614 reg = char_array_i(regs_r32,
3615 ARRAY_SIZE(regs_r32), pp->arg[arg].reg);
3616 if (reg < 0)
3617 ferr(ops, "arg '%s' is not a reg?\n", pp->arg[arg].reg);
3618 if (!(regmask & (1 << reg))) {
3619 regmask_init |= 1 << reg;
3620 regmask |= 1 << reg;
3621 }
3622 }
3623 }
3624 }
3625 else if (po->op == OP_MOV && po->operand[0].pp != NULL
3626 && po->operand[1].pp != NULL)
3627 {
3628 // <var> = offset <something>
3629 if ((po->operand[1].pp->is_func || po->operand[1].pp->is_fptr)
3630 && !IS_START(po->operand[1].name, "off_"))
3631 {
3632 if (!po->operand[0].pp->is_fptr)
3633 ferr(po, "%s not declared as fptr when it should be\n",
3634 po->operand[0].name);
3635 if (pp_cmp_func(po->operand[0].pp, po->operand[1].pp)) {
3636 pp_print(buf1, sizeof(buf1), po->operand[0].pp);
3637 pp_print(buf2, sizeof(buf2), po->operand[1].pp);
3638 fnote(po, "var: %s\n", buf1);
3639 fnote(po, "func: %s\n", buf2);
3640 ferr(po, "^ mismatch\n");
3641 }
3642 }
3643 }
3644 else if (po->op == OP_RET && !IS(g_func_pp->ret_type.name, "void"))
3645 regmask |= 1 << xAX;
3646 else if (po->op == OP_DIV || po->op == OP_IDIV) {
3647 // 32bit division is common, look for it
3648 if (po->op == OP_DIV)
3649 ret = scan_for_reg_clear(i, xDX);
3650 else
3651 ret = scan_for_cdq_edx(i);
3652 if (ret >= 0)
3653 po->flags |= OPF_32BIT;
3654 else
3655 need_tmp64 = 1;
3656 }
3657 else if (po->op == OP_CLD)
3658 po->flags |= OPF_RMD;
3659
3660 if (po->op == OP_RCL || po->op == OP_RCR || po->op == OP_XCHG) {
3661 need_tmp_var = 1;
3662 }
3663 }
3664
3665 // pass4:
3666 // - confirm regmask_save, it might have been reduced
3667 if (regmask_save != 0)
3668 {
3669 regmask_save = 0;
3670 for (i = 0; i < opcnt; i++) {
3671 po = &ops[i];
3672 if (po->flags & OPF_RMD)
3673 continue;
3674
3675 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE))
3676 regmask_save |= 1 << po->operand[0].reg;
3677 }
3678 }
3679
3680 // output starts here
3681
3682 // define userstack size
3683 if (g_func_pp->is_userstack) {
3684 fprintf(fout, "#ifndef US_SZ_%s\n", g_func_pp->name);
3685 fprintf(fout, "#define US_SZ_%s USERSTACK_SIZE\n", g_func_pp->name);
3686 fprintf(fout, "#endif\n");
3687 }
3688
3689 // the function itself
3690 fprintf(fout, "%s ", g_func_pp->ret_type.name);
3691 output_pp_attrs(fout, g_func_pp, g_ida_func_attr & IDAFA_NORETURN);
3692 fprintf(fout, "%s(", g_func_pp->name);
3693
3694 for (i = 0; i < g_func_pp->argc; i++) {
3695 if (i > 0)
3696 fprintf(fout, ", ");
3697 if (g_func_pp->arg[i].fptr != NULL) {
3698 // func pointer..
3699 pp = g_func_pp->arg[i].fptr;
3700 fprintf(fout, "%s (", pp->ret_type.name);
3701 output_pp_attrs(fout, pp, 0);
3702 fprintf(fout, "*a%d)(", i + 1);
3703 for (j = 0; j < pp->argc; j++) {
3704 if (j > 0)
3705 fprintf(fout, ", ");
3706 if (pp->arg[j].fptr)
3707 ferr(ops, "nested fptr\n");
3708 fprintf(fout, "%s", pp->arg[j].type.name);
3709 }
3710 if (pp->is_vararg) {
3711 if (j > 0)
3712 fprintf(fout, ", ");
3713 fprintf(fout, "...");
3714 }
3715 fprintf(fout, ")");
3716 }
3717 else if (g_func_pp->arg[i].type.is_retreg) {
3718 fprintf(fout, "u32 *r_%s", g_func_pp->arg[i].reg);
3719 }
3720 else {
3721 fprintf(fout, "%s a%d", g_func_pp->arg[i].type.name, i + 1);
3722 }
3723 }
3724 if (g_func_pp->is_vararg) {
3725 if (i > 0)
3726 fprintf(fout, ", ");
3727 fprintf(fout, "...");
3728 }
3729
3730 fprintf(fout, ")\n{\n");
3731
3732 // declare indirect functions
3733 for (i = 0; i < opcnt; i++) {
3734 po = &ops[i];
3735 if (po->flags & OPF_RMD)
3736 continue;
3737
3738 if (po->op == OP_CALL) {
3739 pp = po->pp;
3740 if (pp == NULL)
3741 ferr(po, "NULL pp\n");
3742
3743 if (pp->is_fptr && !(pp->name[0] != 0 && pp->is_arg)) {
3744 if (pp->name[0] != 0) {
3745 memmove(pp->name + 2, pp->name, strlen(pp->name) + 1);
3746 memcpy(pp->name, "i_", 2);
3747
3748 // might be declared already
3749 found = 0;
3750 for (j = 0; j < i; j++) {
3751 if (ops[j].op == OP_CALL && (pp_tmp = ops[j].pp)) {
3752 if (pp_tmp->is_fptr && IS(pp->name, pp_tmp->name)) {
3753 found = 1;
3754 break;
3755 }
3756 }
3757 }
3758 if (found)
3759 continue;
3760 }
3761 else
3762 snprintf(pp->name, sizeof(pp->name), "icall%d", i);
3763
3764 fprintf(fout, " %s (", pp->ret_type.name);
3765 output_pp_attrs(fout, pp, 0);
3766 fprintf(fout, "*%s)(", pp->name);
3767 for (j = 0; j < pp->argc; j++) {
3768 if (j > 0)
3769 fprintf(fout, ", ");
3770 fprintf(fout, "%s a%d", pp->arg[j].type.name, j + 1);
3771 }
3772 fprintf(fout, ");\n");
3773 }
3774 }
3775 }
3776
3777 // output LUTs/jumptables
3778 for (i = 0; i < g_func_pd_cnt; i++) {
3779 pd = &g_func_pd[i];
3780 fprintf(fout, " static const ");
3781 if (pd->type == OPT_OFFSET) {
3782 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
3783
3784 for (j = 0; j < pd->count; j++) {
3785 if (j > 0)
3786 fprintf(fout, ", ");
3787 fprintf(fout, "&&%s", pd->d[j].u.label);
3788 }
3789 }
3790 else {
3791 fprintf(fout, "%s %s[] =\n { ",
3792 lmod_type_u(ops, pd->lmod), pd->label);
3793
3794 for (j = 0; j < pd->count; j++) {
3795 if (j > 0)
3796 fprintf(fout, ", ");
3797 fprintf(fout, "%u", pd->d[j].u.val);
3798 }
3799 }
3800 fprintf(fout, " };\n");
3801 had_decl = 1;
3802 }
3803
3804 // declare stack frame, va_arg
3805 if (g_stack_fsz) {
3806 fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
3807 (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz);
3808 had_decl = 1;
3809 }
3810
3811 if (g_func_pp->is_userstack) {
3812 fprintf(fout, " u32 fake_sf[US_SZ_%s / 4];\n", g_func_pp->name);
3813 fprintf(fout, " u32 *esp = &fake_sf[sizeof(fake_sf) / 4];\n");
3814 had_decl = 1;
3815 }
3816
3817 if (g_func_pp->is_vararg) {
3818 fprintf(fout, " va_list ap;\n");
3819 had_decl = 1;
3820 }
3821
3822 // declare arg-registers
3823 for (i = 0; i < g_func_pp->argc; i++) {
3824 if (g_func_pp->arg[i].reg != NULL) {
3825 reg = char_array_i(regs_r32,
3826 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
3827 if (regmask & (1 << reg)) {
3828 if (g_func_pp->arg[i].type.is_retreg)
3829 fprintf(fout, " u32 %s = *r_%s;\n",
3830 g_func_pp->arg[i].reg, g_func_pp->arg[i].reg);
3831 else
3832 fprintf(fout, " u32 %s = (u32)a%d;\n",
3833 g_func_pp->arg[i].reg, i + 1);
3834 }
3835 else {
3836 if (g_func_pp->arg[i].type.is_retreg)
3837 ferr(ops, "retreg '%s' is unused?\n",
3838 g_func_pp->arg[i].reg);
3839 fprintf(fout, " // %s = a%d; // unused\n",
3840 g_func_pp->arg[i].reg, i + 1);
3841 }
3842 had_decl = 1;
3843 }
3844 }
3845
3846 regmask_now = regmask & ~regmask_arg;
3847 regmask_now &= ~(1 << xSP);
3848 if (regmask_now) {
3849 for (reg = 0; reg < 8; reg++) {
3850 if (regmask_now & (1 << reg)) {
3851 fprintf(fout, " u32 %s", regs_r32[reg]);
3852 if (regmask_init & (1 << reg))
3853 fprintf(fout, " = 0");
3854 fprintf(fout, ";\n");
3855 had_decl = 1;
3856 }
3857 }
3858 }
3859
3860 if (regmask_save) {
3861 for (reg = 0; reg < 8; reg++) {
3862 if (regmask_save & (1 << reg)) {
3863 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
3864 had_decl = 1;
3865 }
3866 }
3867 }
3868
3869 if (save_arg_vars) {
3870 for (reg = 0; reg < 32; reg++) {
3871 if (save_arg_vars & (1 << reg)) {
3872 fprintf(fout, " u32 s_a%d;\n", reg + 1);
3873 had_decl = 1;
3874 }
3875 }
3876 }
3877
3878 if (cond_vars) {
3879 for (i = 0; i < 8; i++) {
3880 if (cond_vars & (1 << i)) {
3881 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
3882 had_decl = 1;
3883 }
3884 }
3885 }
3886
3887 if (need_tmp_var) {
3888 fprintf(fout, " u32 tmp;\n");
3889 had_decl = 1;
3890 }
3891
3892 if (need_tmp64) {
3893 fprintf(fout, " u64 tmp64;\n");
3894 had_decl = 1;
3895 }
3896
3897 if (had_decl)
3898 fprintf(fout, "\n");
3899
3900 if (g_func_pp->is_vararg) {
3901 if (g_func_pp->argc_stack == 0)
3902 ferr(ops, "vararg func without stack args?\n");
3903 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
3904 }
3905
3906 // output ops
3907 for (i = 0; i < opcnt; i++)
3908 {
3909 if (g_labels[i][0] != 0) {
3910 fprintf(fout, "\n%s:\n", g_labels[i]);
3911 label_pending = 1;
3912
3913 delayed_flag_op = NULL;
3914 last_arith_dst = NULL;
3915 }
3916
3917 po = &ops[i];
3918 if (po->flags & OPF_RMD)
3919 continue;
3920
3921 no_output = 0;
3922
3923 #define assert_operand_cnt(n_) \
3924 if (po->operand_cnt != n_) \
3925 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
3926
3927 // conditional/flag using op?
3928 if (po->flags & OPF_CC)
3929 {
3930 int is_delayed = 0;
3931
3932 tmp_op = po->datap;
3933
3934 // we go through all this trouble to avoid using parsed_flag_op,
3935 // which makes generated code much nicer
3936 if (delayed_flag_op != NULL)
3937 {
3938 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op,
3939 po->pfo, po->pfo_inv);
3940 is_delayed = 1;
3941 }
3942 else if (last_arith_dst != NULL
3943 && (po->pfo == PFO_Z || po->pfo == PFO_S || po->pfo == PFO_P
3944 || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR))
3945 ))
3946 {
3947 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
3948 out_test_for_cc(buf1, sizeof(buf1), po, po->pfo, po->pfo_inv,
3949 last_arith_dst->lmod, buf3);
3950 is_delayed = 1;
3951 }
3952 else if (tmp_op != NULL) {
3953 // use preprocessed flag calc results
3954 if (!(tmp_op->pfomask & (1 << po->pfo)))
3955 ferr(po, "not prepared for pfo %d\n", po->pfo);
3956
3957 // note: pfo_inv was not yet applied
3958 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
3959 po->pfo_inv ? "!" : "", parsed_flag_op_names[po->pfo]);
3960 }
3961 else {
3962 ferr(po, "all methods of finding comparison failed\n");
3963 }
3964
3965 if (po->flags & OPF_JMP) {
3966 fprintf(fout, " if %s", buf1);
3967 }
3968 else if (po->op == OP_RCL || po->op == OP_RCR
3969 || po->op == OP_ADC || po->op == OP_SBB)
3970 {
3971 if (is_delayed)
3972 fprintf(fout, " cond_%s = %s;\n",
3973 parsed_flag_op_names[po->pfo], buf1);
3974 }
3975 else if (po->flags & OPF_DATA) { // SETcc
3976 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
3977 fprintf(fout, " %s = %s;", buf2, buf1);
3978 }
3979 else {
3980 ferr(po, "unhandled conditional op\n");
3981 }
3982 }
3983
3984 pfomask = po->pfomask;
3985
3986 if (po->flags & (OPF_REPZ|OPF_REPNZ)) {
3987 struct parsed_opr opr = {0,};
3988 opr.type = OPT_REG;
3989 opr.reg = xCX;
3990 opr.lmod = OPLM_DWORD;
3991 ret = try_resolve_const(i, &opr, opcnt * 7 + i, &uval);
3992
3993 if (ret != 1 || uval == 0) {
3994 // we need initial flags for ecx=0 case..
3995 if (i > 0 && ops[i - 1].op == OP_XOR
3996 && IS(ops[i - 1].operand[0].name,
3997 ops[i - 1].operand[1].name))
3998 {
3999 fprintf(fout, " cond_z = ");
4000 if (pfomask & (1 << PFO_C))
4001 fprintf(fout, "cond_c = ");
4002 fprintf(fout, "0;\n");
4003 }
4004 else if (last_arith_dst != NULL) {
4005 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
4006 out_test_for_cc(buf1, sizeof(buf1), po, PFO_Z, 0,
4007 last_arith_dst->lmod, buf3);
4008 fprintf(fout, " cond_z = %s;\n", buf1);
4009 }
4010 else
4011 ferr(po, "missing initial ZF\n");
4012 }
4013 }
4014
4015 switch (po->op)
4016 {
4017 case OP_MOV:
4018 assert_operand_cnt(2);
4019 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4020 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4021 default_cast_to(buf3, sizeof(buf3), &po->operand[0]);
4022 fprintf(fout, " %s = %s;", buf1,
4023 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4024 buf3, 0));
4025 break;
4026
4027 case OP_LEA:
4028 assert_operand_cnt(2);
4029 po->operand[1].lmod = OPLM_DWORD; // always
4030 fprintf(fout, " %s = %s;",
4031 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4032 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4033 NULL, 1));
4034 break;
4035
4036 case OP_MOVZX:
4037 assert_operand_cnt(2);
4038 fprintf(fout, " %s = %s;",
4039 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4040 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4041 break;
4042
4043 case OP_MOVSX:
4044 assert_operand_cnt(2);
4045 switch (po->operand[1].lmod) {
4046 case OPLM_BYTE:
4047 strcpy(buf3, "(s8)");
4048 break;
4049 case OPLM_WORD:
4050 strcpy(buf3, "(s16)");
4051 break;
4052 default:
4053 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
4054 }
4055 fprintf(fout, " %s = %s;",
4056 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4057 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4058 buf3, 0));
4059 break;
4060
4061 case OP_XCHG:
4062 assert_operand_cnt(2);
4063 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4064 fprintf(fout, " tmp = %s;",
4065 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], "", 0));
4066 fprintf(fout, " %s = %s;",
4067 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4068 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4069 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
4070 fprintf(fout, " %s = %stmp;",
4071 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]),
4072 default_cast_to(buf3, sizeof(buf3), &po->operand[1]));
4073 snprintf(g_comment, sizeof(g_comment), "xchg");
4074 break;
4075
4076 case OP_NOT:
4077 assert_operand_cnt(1);
4078 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4079 fprintf(fout, " %s = ~%s;", buf1, buf1);
4080 break;
4081
4082 case OP_CDQ:
4083 assert_operand_cnt(2);
4084 fprintf(fout, " %s = (s32)%s >> 31;",
4085 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4086 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4087 strcpy(g_comment, "cdq");
4088 break;
4089
4090 case OP_LODS:
4091 assert_operand_cnt(3);
4092 if (po->flags & OPF_REP) {
4093 // hmh..
4094 ferr(po, "TODO\n");
4095 }
4096 else {
4097 fprintf(fout, " eax = %sesi; esi %c= %d;",
4098 lmod_cast_u_ptr(po, po->operand[0].lmod),
4099 (po->flags & OPF_DF) ? '-' : '+',
4100 lmod_bytes(po, po->operand[0].lmod));
4101 strcpy(g_comment, "lods");
4102 }
4103 break;
4104
4105 case OP_STOS:
4106 assert_operand_cnt(3);
4107 if (po->flags & OPF_REP) {
4108 fprintf(fout, " for (; ecx != 0; ecx--, edi %c= %d)\n",
4109 (po->flags & OPF_DF) ? '-' : '+',
4110 lmod_bytes(po, po->operand[0].lmod));
4111 fprintf(fout, " %sedi = eax;",
4112 lmod_cast_u_ptr(po, po->operand[0].lmod));
4113 strcpy(g_comment, "rep stos");
4114 }
4115 else {
4116 fprintf(fout, " %sedi = eax; edi %c= %d;",
4117 lmod_cast_u_ptr(po, po->operand[0].lmod),
4118 (po->flags & OPF_DF) ? '-' : '+',
4119 lmod_bytes(po, po->operand[0].lmod));
4120 strcpy(g_comment, "stos");
4121 }
4122 break;
4123
4124 case OP_MOVS:
4125 assert_operand_cnt(3);
4126 j = lmod_bytes(po, po->operand[0].lmod);
4127 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
4128 l = (po->flags & OPF_DF) ? '-' : '+';
4129 if (po->flags & OPF_REP) {
4130 fprintf(fout,
4131 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n",
4132 l, j, l, j);
4133 fprintf(fout,
4134 " %sedi = %sesi;", buf1, buf1);
4135 strcpy(g_comment, "rep movs");
4136 }
4137 else {
4138 fprintf(fout, " %sedi = %sesi; edi %c= %d; esi %c= %d;",
4139 buf1, buf1, l, j, l, j);
4140 strcpy(g_comment, "movs");
4141 }
4142 break;
4143
4144 case OP_CMPS:
4145 // repe ~ repeat while ZF=1
4146 assert_operand_cnt(3);
4147 j = lmod_bytes(po, po->operand[0].lmod);
4148 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
4149 l = (po->flags & OPF_DF) ? '-' : '+';
4150 if (po->flags & OPF_REP) {
4151 fprintf(fout,
4152 " for (; ecx != 0; ecx--) {\n");
4153 if (pfomask & (1 << PFO_C)) {
4154 // ugh..
4155 fprintf(fout,
4156 " cond_c = %sesi < %sedi;\n", buf1, buf1);
4157 pfomask &= ~(1 << PFO_C);
4158 }
4159 fprintf(fout,
4160 " cond_z = (%sesi == %sedi); esi %c= %d, edi %c= %d;\n",
4161 buf1, buf1, l, j, l, j);
4162 fprintf(fout,
4163 " if (cond_z %s 0) break;\n",
4164 (po->flags & OPF_REPZ) ? "==" : "!=");
4165 fprintf(fout,
4166 " }");
4167 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
4168 (po->flags & OPF_REPZ) ? "e" : "ne");
4169 }
4170 else {
4171 fprintf(fout,
4172 " cond_z = (%sesi == %sedi); esi %c= %d; edi %c= %d;",
4173 buf1, buf1, l, j, l, j);
4174 strcpy(g_comment, "cmps");
4175 }
4176 pfomask &= ~(1 << PFO_Z);
4177 last_arith_dst = NULL;
4178 delayed_flag_op = NULL;
4179 break;
4180
4181 case OP_SCAS:
4182 // only does ZF (for now)
4183 // repe ~ repeat while ZF=1
4184 assert_operand_cnt(3);
4185 j = lmod_bytes(po, po->operand[0].lmod);
4186 l = (po->flags & OPF_DF) ? '-' : '+';
4187 if (po->flags & OPF_REP) {
4188 fprintf(fout,
4189 " for (; ecx != 0; ecx--) {\n");
4190 fprintf(fout,
4191 " cond_z = (%seax == %sedi); edi %c= %d;\n",
4192 lmod_cast_u(po, po->operand[0].lmod),
4193 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
4194 fprintf(fout,
4195 " if (cond_z %s 0) break;\n",
4196 (po->flags & OPF_REPZ) ? "==" : "!=");
4197 fprintf(fout,
4198 " }");
4199 snprintf(g_comment, sizeof(g_comment), "rep%s scas",
4200 (po->flags & OPF_REPZ) ? "e" : "ne");
4201 }
4202 else {
4203 fprintf(fout, " cond_z = (%seax == %sedi); edi %c= %d;",
4204 lmod_cast_u(po, po->operand[0].lmod),
4205 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
4206 strcpy(g_comment, "scas");
4207 }
4208 pfomask &= ~(1 << PFO_Z);
4209 last_arith_dst = NULL;
4210 delayed_flag_op = NULL;
4211 break;
4212
4213 // arithmetic w/flags
4214 case OP_AND:
4215 case OP_OR:
4216 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4217 // fallthrough
4218 dualop_arith:
4219 assert_operand_cnt(2);
4220 fprintf(fout, " %s %s= %s;",
4221 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4222 op_to_c(po),
4223 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4224 output_std_flags(fout, po, &pfomask, buf1);
4225 last_arith_dst = &po->operand[0];
4226 delayed_flag_op = NULL;
4227 break;
4228
4229 case OP_SHL:
4230 case OP_SHR:
4231 assert_operand_cnt(2);
4232 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4233 if (pfomask & (1 << PFO_C)) {
4234 if (po->operand[1].type == OPT_CONST) {
4235 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4236 j = po->operand[1].val;
4237 j %= l;
4238 if (j != 0) {
4239 if (po->op == OP_SHL)
4240 j = l - j;
4241 else
4242 j -= 1;
4243 fprintf(fout, " cond_c = (%s >> %d) & 1;\n",
4244 buf1, j);
4245 }
4246 else
4247 ferr(po, "zero shift?\n");
4248 }
4249 else
4250 ferr(po, "TODO\n");
4251 pfomask &= ~(1 << PFO_C);
4252 }
4253 fprintf(fout, " %s %s= %s;", buf1, op_to_c(po),
4254 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4255 output_std_flags(fout, po, &pfomask, buf1);
4256 last_arith_dst = &po->operand[0];
4257 delayed_flag_op = NULL;
4258 break;
4259
4260 case OP_SAR:
4261 assert_operand_cnt(2);
4262 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4263 fprintf(fout, " %s = %s%s >> %s;", buf1,
4264 lmod_cast_s(po, po->operand[0].lmod), buf1,
4265 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4266 output_std_flags(fout, po, &pfomask, buf1);
4267 last_arith_dst = &po->operand[0];
4268 delayed_flag_op = NULL;
4269 break;
4270
4271 case OP_ROL:
4272 case OP_ROR:
4273 assert_operand_cnt(2);
4274 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4275 if (po->operand[1].type == OPT_CONST) {
4276 j = po->operand[1].val;
4277 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
4278 fprintf(fout, po->op == OP_ROL ?
4279 " %s = (%s << %d) | (%s >> %d);" :
4280 " %s = (%s >> %d) | (%s << %d);",
4281 buf1, buf1, j, buf1,
4282 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
4283 }
4284 else
4285 ferr(po, "TODO\n");
4286 output_std_flags(fout, po, &pfomask, buf1);
4287 last_arith_dst = &po->operand[0];
4288 delayed_flag_op = NULL;
4289 break;
4290
4291 case OP_RCL:
4292 case OP_RCR:
4293 assert_operand_cnt(2);
4294 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4295 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4296 if (po->operand[1].type == OPT_CONST) {
4297 j = po->operand[1].val % l;
4298 if (j == 0)
4299 ferr(po, "zero rotate\n");
4300 fprintf(fout, " tmp = (%s >> %d) & 1;\n",
4301 buf1, (po->op == OP_RCL) ? (l - j) : (j - 1));
4302 if (po->op == OP_RCL) {
4303 fprintf(fout,
4304 " %s = (%s << %d) | (cond_c << %d)",
4305 buf1, buf1, j, j - 1);
4306 if (j != 1)
4307 fprintf(fout, " | (%s >> %d)", buf1, l + 1 - j);
4308 }
4309 else {
4310 fprintf(fout,
4311 " %s = (%s >> %d) | (cond_c << %d)",
4312 buf1, buf1, j, l - j);
4313 if (j != 1)
4314 fprintf(fout, " | (%s << %d)", buf1, l + 1 - j);
4315 }
4316 fprintf(fout, ";\n");
4317 fprintf(fout, " cond_c = tmp;");
4318 }
4319 else
4320 ferr(po, "TODO\n");
4321 strcpy(g_comment, (po->op == OP_RCL) ? "rcl" : "rcr");
4322 output_std_flags(fout, po, &pfomask, buf1);
4323 last_arith_dst = &po->operand[0];
4324 delayed_flag_op = NULL;
4325 break;
4326
4327 case OP_XOR:
4328 assert_operand_cnt(2);
4329 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4330 if (IS(opr_name(po, 0), opr_name(po, 1))) {
4331 // special case for XOR
4332 if (pfomask & (1 << PFO_BE)) { // weird, but it happens..
4333 fprintf(fout, " cond_be = 1;\n");
4334 pfomask &= ~(1 << PFO_BE);
4335 }
4336 fprintf(fout, " %s = 0;",
4337 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4338 last_arith_dst = &po->operand[0];
4339 delayed_flag_op = NULL;
4340 break;
4341 }
4342 goto dualop_arith;
4343
4344 case OP_ADD:
4345 assert_operand_cnt(2);
4346 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4347 if (pfomask & (1 << PFO_C)) {
4348 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4349 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4350 if (po->operand[0].lmod == OPLM_DWORD) {
4351 fprintf(fout, " tmp64 = (u64)%s + %s;\n", buf1, buf2);
4352 fprintf(fout, " cond_c = tmp64 >> 32;\n");
4353 fprintf(fout, " %s = (u32)tmp64;",
4354 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4355 strcat(g_comment, "add64");
4356 }
4357 else {
4358 fprintf(fout, " cond_c = ((u32)%s + %s) >> %d;\n",
4359 buf1, buf2, lmod_bytes(po, po->operand[0].lmod) * 8);
4360 fprintf(fout, " %s += %s;",
4361 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4362 buf2);
4363 }
4364 pfomask &= ~(1 << PFO_C);
4365 output_std_flags(fout, po, &pfomask, buf1);
4366 last_arith_dst = &po->operand[0];
4367 delayed_flag_op = NULL;
4368 break;
4369 }
4370 goto dualop_arith;
4371
4372 case OP_SUB:
4373 assert_operand_cnt(2);
4374 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4375 if (pfomask & (1 << PFO_C)) {
4376 fprintf(fout, " cond_c = %s < %s;\n",
4377 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]),
4378 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4379 pfomask &= ~(1 << PFO_C);
4380 }
4381 goto dualop_arith;
4382
4383 case OP_ADC:
4384 case OP_SBB:
4385 assert_operand_cnt(2);
4386 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4387 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4388 if (po->op == OP_SBB
4389 && IS(po->operand[0].name, po->operand[1].name))
4390 {
4391 // avoid use of unitialized var
4392 fprintf(fout, " %s = -cond_c;", buf1);
4393 // carry remains what it was
4394 pfomask &= ~(1 << PFO_C);
4395 }
4396 else {
4397 fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po),
4398 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4399 }
4400 output_std_flags(fout, po, &pfomask, buf1);
4401 last_arith_dst = &po->operand[0];
4402 delayed_flag_op = NULL;
4403 break;
4404
4405 case OP_BSF:
4406 assert_operand_cnt(2);
4407 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4408 fprintf(fout, " %s = %s ? __builtin_ffs(%s) - 1 : 0;",
4409 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4410 buf2, buf2);
4411 output_std_flags(fout, po, &pfomask, buf1);
4412 last_arith_dst = &po->operand[0];
4413 delayed_flag_op = NULL;
4414 strcat(g_comment, "bsf");
4415 break;
4416
4417 case OP_INC:
4418 case OP_DEC:
4419 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4420 if (po->operand[0].type == OPT_REG) {
4421 strcpy(buf2, po->op == OP_INC ? "++" : "--");
4422 fprintf(fout, " %s%s;", buf1, buf2);
4423 }
4424 else {
4425 strcpy(buf2, po->op == OP_INC ? "+" : "-");
4426 fprintf(fout, " %s %s= 1;", buf1, buf2);
4427 }
4428 output_std_flags(fout, po, &pfomask, buf1);
4429 last_arith_dst = &po->operand[0];
4430 delayed_flag_op = NULL;
4431 break;
4432
4433 case OP_NEG:
4434 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4435 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
4436 fprintf(fout, " %s = -%s%s;", buf1,
4437 lmod_cast_s(po, po->operand[0].lmod), buf2);
4438 last_arith_dst = &po->operand[0];
4439 delayed_flag_op = NULL;
4440 if (pfomask & (1 << PFO_C)) {
4441 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
4442 pfomask &= ~(1 << PFO_C);
4443 }
4444 break;
4445
4446 case OP_IMUL:
4447 if (po->operand_cnt == 2) {
4448 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4449 goto dualop_arith;
4450 }
4451 if (po->operand_cnt == 3)
4452 ferr(po, "TODO imul3\n");
4453 // fallthrough
4454 case OP_MUL:
4455 assert_operand_cnt(1);
4456 switch (po->operand[0].lmod) {
4457 case OPLM_DWORD:
4458 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
4459 fprintf(fout, " tmp64 = %seax * %s%s;\n", buf1, buf1,
4460 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
4461 fprintf(fout, " edx = tmp64 >> 32;\n");
4462 fprintf(fout, " eax = tmp64;");
4463 break;
4464 case OPLM_BYTE:
4465 strcpy(buf1, po->op == OP_IMUL ? "(s16)(s8)" : "(u16)(u8)");
4466 fprintf(fout, " LOWORD(eax) = %seax * %s;", buf1,
4467 out_src_opr(buf2, sizeof(buf2), po, &po->operand[0],
4468 buf1, 0));
4469 break;
4470 default:
4471 ferr(po, "TODO: unhandled mul type\n");
4472 break;
4473 }
4474 last_arith_dst = NULL;
4475 delayed_flag_op = NULL;
4476 break;
4477
4478 case OP_DIV:
4479 case OP_IDIV:
4480 assert_operand_cnt(1);
4481 if (po->operand[0].lmod != OPLM_DWORD)
4482 ferr(po, "unhandled lmod %d\n", po->operand[0].lmod);
4483
4484 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4485 strcpy(buf2, lmod_cast(po, po->operand[0].lmod,
4486 po->op == OP_IDIV));
4487 switch (po->operand[0].lmod) {
4488 case OPLM_DWORD:
4489 if (po->flags & OPF_32BIT)
4490 snprintf(buf3, sizeof(buf3), "%seax", buf2);
4491 else {
4492 fprintf(fout, " tmp64 = ((u64)edx << 32) | eax;\n");
4493 snprintf(buf3, sizeof(buf3), "%stmp64",
4494 (po->op == OP_IDIV) ? "(s64)" : "");
4495 }
4496 if (po->operand[0].type == OPT_REG
4497 && po->operand[0].reg == xDX)
4498 {
4499 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
4500 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
4501 }
4502 else {
4503 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
4504 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
4505 }
4506 break;
4507 default:
4508 ferr(po, "unhandled division type\n");
4509 }
4510 last_arith_dst = NULL;
4511 delayed_flag_op = NULL;
4512 break;
4513
4514 case OP_TEST:
4515 case OP_CMP:
4516 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4517 if (pfomask != 0) {
4518 for (j = 0; j < 8; j++) {
4519 if (pfomask & (1 << j)) {
4520 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
4521 fprintf(fout, " cond_%s = %s;",
4522 parsed_flag_op_names[j], buf1);
4523 }
4524 }
4525 pfomask = 0;
4526 }
4527 else
4528 no_output = 1;
4529 last_arith_dst = NULL;
4530 delayed_flag_op = po;
4531 break;
4532
4533 case OP_SCC:
4534 // SETcc - should already be handled
4535 break;
4536
4537 // note: we reuse OP_Jcc for SETcc, only flags differ
4538 case OP_JCC:
4539 fprintf(fout, "\n goto %s;", po->operand[0].name);
4540 break;
4541
4542 case OP_JECXZ:
4543 fprintf(fout, " if (ecx == 0)\n");
4544 fprintf(fout, " goto %s;", po->operand[0].name);
4545 strcat(g_comment, "jecxz");
4546 break;
4547
4548 case OP_JMP:
4549 assert_operand_cnt(1);
4550 last_arith_dst = NULL;
4551 delayed_flag_op = NULL;
4552
4553 if (po->operand[0].type == OPT_REGMEM) {
4554 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
4555 buf1, buf2);
4556 if (ret != 2)
4557 ferr(po, "parse failure for jmp '%s'\n",
4558 po->operand[0].name);
4559 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
4560 break;
4561 }
4562 else if (po->operand[0].type != OPT_LABEL)
4563 ferr(po, "unhandled jmp type\n");
4564
4565 fprintf(fout, " goto %s;", po->operand[0].name);
4566 break;
4567
4568 case OP_CALL:
4569 assert_operand_cnt(1);
4570 pp = po->pp;
4571 my_assert_not(pp, NULL);
4572
4573 strcpy(buf3, " ");
4574 if (po->flags & OPF_CC) {
4575 // we treat conditional branch to another func
4576 // (yes such code exists..) as conditional tailcall
4577 strcat(buf3, " ");
4578 fprintf(fout, " {\n");
4579 }
4580
4581 if (pp->is_fptr && !pp->is_arg) {
4582 fprintf(fout, "%s%s = %s;\n", buf3, pp->name,
4583 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
4584 "(void *)", 0));
4585 if (pp->is_unresolved)
4586 fprintf(fout, "%sunresolved_call(\"%s:%d\", %s);\n",
4587 buf3, asmfn, po->asmln, pp->name);
4588 }
4589
4590 fprintf(fout, "%s", buf3);
4591 if (strstr(pp->ret_type.name, "int64")) {
4592 if (po->flags & OPF_TAIL)
4593 ferr(po, "int64 and tail?\n");
4594 fprintf(fout, "tmp64 = ");
4595 }
4596 else if (!IS(pp->ret_type.name, "void")) {
4597 if (po->flags & OPF_TAIL) {
4598 if (!IS(g_func_pp->ret_type.name, "void")) {
4599 fprintf(fout, "return ");
4600 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
4601 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
4602 }
4603 }
4604 else if (regmask & (1 << xAX)) {
4605 fprintf(fout, "eax = ");
4606 if (pp->ret_type.is_ptr)
4607 fprintf(fout, "(u32)");
4608 }
4609 }
4610
4611 if (pp->name[0] == 0)
4612 ferr(po, "missing pp->name\n");
4613 fprintf(fout, "%s%s(", pp->name,
4614 pp->has_structarg ? "_sa" : "");
4615
4616 if (po->flags & OPF_ATAIL) {
4617 if (pp->argc_stack != g_func_pp->argc_stack
4618 || (pp->argc_stack > 0
4619 && pp->is_stdcall != g_func_pp->is_stdcall))
4620 ferr(po, "incompatible tailcall\n");
4621 if (g_func_pp->has_retreg)
4622 ferr(po, "TODO: retreg+tailcall\n");
4623
4624 for (arg = j = 0; arg < pp->argc; arg++) {
4625 if (arg > 0)
4626 fprintf(fout, ", ");
4627
4628 cast[0] = 0;
4629 if (pp->arg[arg].type.is_ptr)
4630 snprintf(cast, sizeof(cast), "(%s)",
4631 pp->arg[arg].type.name);
4632
4633 if (pp->arg[arg].reg != NULL) {
4634 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
4635 continue;
4636 }
4637 // stack arg
4638 for (; j < g_func_pp->argc; j++)
4639 if (g_func_pp->arg[j].reg == NULL)
4640 break;
4641 fprintf(fout, "%sa%d", cast, j + 1);
4642 j++;
4643 }
4644 }
4645 else {
4646 for (arg = 0; arg < pp->argc; arg++) {
4647 if (arg > 0)
4648 fprintf(fout, ", ");
4649
4650 cast[0] = 0;
4651 if (pp->arg[arg].type.is_ptr)
4652 snprintf(cast, sizeof(cast), "(%s)",
4653 pp->arg[arg].type.name);
4654
4655 if (pp->arg[arg].reg != NULL) {
4656 if (pp->arg[arg].type.is_retreg)
4657 fprintf(fout, "&%s", pp->arg[arg].reg);
4658 else
4659 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
4660 continue;
4661 }
4662
4663 // stack arg
4664 tmp_op = pp->arg[arg].datap;
4665 if (tmp_op == NULL)
4666 ferr(po, "parsed_op missing for arg%d\n", arg);
4667
4668 if (tmp_op->flags & OPF_VAPUSH) {
4669 fprintf(fout, "ap");
4670 }
4671 else if (tmp_op->p_argpass != 0) {
4672 fprintf(fout, "a%d", tmp_op->p_argpass);
4673 }
4674 else if (tmp_op->p_argnum != 0) {
4675 fprintf(fout, "%ss_a%d", cast, tmp_op->p_argnum);
4676 }
4677 else {
4678 fprintf(fout, "%s",
4679 out_src_opr(buf1, sizeof(buf1),
4680 tmp_op, &tmp_op->operand[0], cast, 0));
4681 }
4682 }
4683 }
4684 fprintf(fout, ");");
4685
4686 if (strstr(pp->ret_type.name, "int64")) {
4687 fprintf(fout, "\n");
4688 fprintf(fout, "%sedx = tmp64 >> 32;\n", buf3);
4689 fprintf(fout, "%seax = tmp64;", buf3);
4690 }
4691
4692 if (pp->is_unresolved) {
4693 snprintf(buf2, sizeof(buf2), " unresolved %dreg",
4694 pp->argc_reg);
4695 strcat(g_comment, buf2);
4696 }
4697
4698 if (po->flags & OPF_TAIL) {
4699 ret = 0;
4700 if (i == opcnt - 1 || pp->is_noreturn)
4701 ret = 0;
4702 else if (IS(pp->ret_type.name, "void"))
4703 ret = 1;
4704 else if (IS(g_func_pp->ret_type.name, "void"))
4705 ret = 1;
4706 // else already handled as 'return f()'
4707
4708 if (ret) {
4709 if (!IS(g_func_pp->ret_type.name, "void")) {
4710 ferr(po, "int func -> void func tailcall?\n");
4711 }
4712 else {
4713 fprintf(fout, "\n%sreturn;", buf3);
4714 strcat(g_comment, " ^ tailcall");
4715 }
4716 }
4717 else
4718 strcat(g_comment, " tailcall");
4719 }
4720 if (pp->is_noreturn)
4721 strcat(g_comment, " noreturn");
4722 if ((po->flags & OPF_ATAIL) && pp->argc_stack > 0)
4723 strcat(g_comment, " argframe");
4724 if (po->flags & OPF_CC)
4725 strcat(g_comment, " cond");
4726
4727 if (po->flags & OPF_CC)
4728 fprintf(fout, "\n }");
4729
4730 delayed_flag_op = NULL;
4731 last_arith_dst = NULL;
4732 break;
4733
4734 case OP_RET:
4735 if (g_func_pp->is_vararg)
4736 fprintf(fout, " va_end(ap);\n");
4737 if (g_func_pp->has_retreg) {
4738 for (arg = 0; arg < g_func_pp->argc; arg++)
4739 if (g_func_pp->arg[arg].type.is_retreg)
4740 fprintf(fout, " *r_%s = %s;\n",
4741 g_func_pp->arg[arg].reg, g_func_pp->arg[arg].reg);
4742 }
4743
4744 if (IS(g_func_pp->ret_type.name, "void")) {
4745 if (i != opcnt - 1 || label_pending)
4746 fprintf(fout, " return;");
4747 }
4748 else if (g_func_pp->ret_type.is_ptr) {
4749 fprintf(fout, " return (%s)eax;",
4750 g_func_pp->ret_type.name);
4751 }
4752 else if (IS(g_func_pp->ret_type.name, "__int64"))
4753 fprintf(fout, " return ((u64)edx << 32) | eax;");
4754 else
4755 fprintf(fout, " return eax;");
4756
4757 last_arith_dst = NULL;
4758 delayed_flag_op = NULL;
4759 break;
4760
4761 case OP_PUSH:
4762 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4763 if (po->p_argnum != 0) {
4764 // special case - saved func arg
4765 fprintf(fout, " s_a%d = %s;", po->p_argnum, buf1);
4766 break;
4767 }
4768 else if (po->flags & OPF_RSAVE) {
4769 fprintf(fout, " s_%s = %s;", buf1, buf1);
4770 break;
4771 }
4772 else if (g_func_pp->is_userstack) {
4773 fprintf(fout, " *(--esp) = %s;", buf1);
4774 break;
4775 }
4776 if (!(g_ida_func_attr & IDAFA_NORETURN))
4777 ferr(po, "stray push encountered\n");
4778 no_output = 1;
4779 break;
4780
4781 case OP_POP:
4782 if (po->flags & OPF_RSAVE) {
4783 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4784 fprintf(fout, " %s = s_%s;", buf1, buf1);
4785 break;
4786 }
4787 else if (po->datap != NULL) {
4788 // push/pop pair
4789 tmp_op = po->datap;
4790 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4791 fprintf(fout, " %s = %s;", buf1,
4792 out_src_opr(buf2, sizeof(buf2),
4793 tmp_op, &tmp_op->operand[0],
4794 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
4795 break;
4796 }
4797 else if (g_func_pp->is_userstack) {
4798 fprintf(fout, " %s = *esp++;",
4799 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4800 break;
4801 }
4802 else
4803 ferr(po, "stray pop encountered\n");
4804 break;
4805
4806 case OP_NOP:
4807 no_output = 1;
4808 break;
4809
4810 default:
4811 no_output = 1;
4812 ferr(po, "unhandled op type %d, flags %x\n",
4813 po->op, po->flags);
4814 break;
4815 }
4816
4817 if (g_comment[0] != 0) {
4818 char *p = g_comment;
4819 while (my_isblank(*p))
4820 p++;
4821 fprintf(fout, " // %s", p);
4822 g_comment[0] = 0;
4823 no_output = 0;
4824 }
4825 if (!no_output)
4826 fprintf(fout, "\n");
4827
4828 // some sanity checking
4829 if (po->flags & OPF_REP) {
4830 if (po->op != OP_STOS && po->op != OP_MOVS
4831 && po->op != OP_CMPS && po->op != OP_SCAS)
4832 ferr(po, "unexpected rep\n");
4833 if (!(po->flags & (OPF_REPZ|OPF_REPNZ))
4834 && (po->op == OP_CMPS || po->op == OP_SCAS))
4835 ferr(po, "cmps/scas with plain rep\n");
4836 }
4837 if ((po->flags & (OPF_REPZ|OPF_REPNZ))
4838 && po->op != OP_CMPS && po->op != OP_SCAS)
4839 ferr(po, "unexpected repz/repnz\n");
4840
4841 if (pfomask != 0)
4842 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
4843
4844 // see is delayed flag stuff is still valid
4845 if (delayed_flag_op != NULL && delayed_flag_op != po) {
4846 if (is_any_opr_modified(delayed_flag_op, po, 0))
4847 delayed_flag_op = NULL;
4848 }
4849
4850 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
4851 if (is_opr_modified(last_arith_dst, po))
4852 last_arith_dst = NULL;
4853 }
4854
4855 label_pending = 0;
4856 }
4857
4858 if (g_stack_fsz && !g_stack_frame_used)
4859 fprintf(fout, " (void)sf;\n");
4860
4861 fprintf(fout, "}\n\n");
4862
4863 // cleanup
4864 for (i = 0; i < opcnt; i++) {
4865 struct label_ref *lr, *lr_del;
4866
4867 lr = g_label_refs[i].next;
4868 while (lr != NULL) {
4869 lr_del = lr;
4870 lr = lr->next;
4871 free(lr_del);
4872 }
4873 g_label_refs[i].i = -1;
4874 g_label_refs[i].next = NULL;
4875
4876 if (ops[i].op == OP_CALL) {
4877 if (ops[i].pp)
4878 proto_release(ops[i].pp);
4879 }
4880 }
4881 g_func_pp = NULL;
4882}
4883
4884static void set_label(int i, const char *name)
4885{
4886 const char *p;
4887 int len;
4888
4889 len = strlen(name);
4890 p = strchr(name, ':');
4891 if (p != NULL)
4892 len = p - name;
4893
4894 if (len > sizeof(g_labels[0]) - 1)
4895 aerr("label '%s' too long: %d\n", name, len);
4896 if (g_labels[i][0] != 0 && !IS_START(g_labels[i], "algn_"))
4897 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
4898 memcpy(g_labels[i], name, len);
4899 g_labels[i][len] = 0;
4900}
4901
4902// '=' needs special treatment..
4903static char *next_word_s(char *w, size_t wsize, char *s)
4904{
4905 size_t i;
4906
4907 s = sskip(s);
4908
4909 for (i = 0; i < wsize - 1; i++) {
4910 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
4911 break;
4912 w[i] = s[i];
4913 }
4914 w[i] = 0;
4915
4916 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
4917 printf("warning: '%s' truncated\n", w);
4918
4919 return s + i;
4920}
4921
4922struct chunk_item {
4923 char *name;
4924 long fptr;
4925 int asmln;
4926};
4927
4928static struct chunk_item *func_chunks;
4929static int func_chunk_cnt;
4930static int func_chunk_alloc;
4931
4932static void add_func_chunk(FILE *fasm, const char *name, int line)
4933{
4934 if (func_chunk_cnt >= func_chunk_alloc) {
4935 func_chunk_alloc *= 2;
4936 func_chunks = realloc(func_chunks,
4937 func_chunk_alloc * sizeof(func_chunks[0]));
4938 my_assert_not(func_chunks, NULL);
4939 }
4940 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
4941 func_chunks[func_chunk_cnt].name = strdup(name);
4942 func_chunks[func_chunk_cnt].asmln = line;
4943 func_chunk_cnt++;
4944}
4945
4946static int cmp_chunks(const void *p1, const void *p2)
4947{
4948 const struct chunk_item *c1 = p1, *c2 = p2;
4949 return strcmp(c1->name, c2->name);
4950}
4951
4952static int cmpstringp(const void *p1, const void *p2)
4953{
4954 return strcmp(*(char * const *)p1, *(char * const *)p2);
4955}
4956
4957static void scan_ahead(FILE *fasm)
4958{
4959 char words[2][256];
4960 char line[256];
4961 long oldpos;
4962 int oldasmln;
4963 int wordc;
4964 char *p;
4965 int i;
4966
4967 oldpos = ftell(fasm);
4968 oldasmln = asmln;
4969
4970 while (fgets(line, sizeof(line), fasm))
4971 {
4972 wordc = 0;
4973 asmln++;
4974
4975 p = sskip(line);
4976 if (*p == 0)
4977 continue;
4978
4979 if (*p == ';')
4980 {
4981 // get rid of random tabs
4982 for (i = 0; line[i] != 0; i++)
4983 if (line[i] == '\t')
4984 line[i] = ' ';
4985
4986 if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
4987 {
4988 p += 30;
4989 next_word(words[0], sizeof(words[0]), p);
4990 if (words[0][0] == 0)
4991 aerr("missing name for func chunk?\n");
4992
4993 add_func_chunk(fasm, words[0], asmln);
4994 }
4995 else if (IS_START(p, "; sctend"))
4996 break;
4997
4998 continue;
4999 } // *p == ';'
5000
5001 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
5002 words[wordc][0] = 0;
5003 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
5004 if (*p == 0 || *p == ';') {
5005 wordc++;
5006 break;
5007 }
5008 }
5009
5010 if (wordc == 2 && IS(words[1], "ends"))
5011 break;
5012 }
5013
5014 fseek(fasm, oldpos, SEEK_SET);
5015 asmln = oldasmln;
5016}
5017
5018int main(int argc, char *argv[])
5019{
5020 FILE *fout, *fasm, *frlist;
5021 struct parsed_data *pd = NULL;
5022 int pd_alloc = 0;
5023 char **rlist = NULL;
5024 int rlist_len = 0;
5025 int rlist_alloc = 0;
5026 int func_chunks_used = 0;
5027 int func_chunks_sorted = 0;
5028 int func_chunk_i = -1;
5029 long func_chunk_ret = 0;
5030 int func_chunk_ret_ln = 0;
5031 int scanned_ahead = 0;
5032 char line[256];
5033 char words[20][256];
5034 enum opr_lenmod lmod;
5035 char *sctproto = NULL;
5036 int in_func = 0;
5037 int pending_endp = 0;
5038 int skip_func = 0;
5039 int skip_warned = 0;
5040 int eq_alloc;
5041 int verbose = 0;
5042 int multi_seg = 0;
5043 int end = 0;
5044 int arg_out;
5045 int arg;
5046 int pi = 0;
5047 int i, j;
5048 int ret, len;
5049 char *p;
5050 int wordc;
5051
5052 for (arg = 1; arg < argc; arg++) {
5053 if (IS(argv[arg], "-v"))
5054 verbose = 1;
5055 else if (IS(argv[arg], "-rf"))
5056 g_allow_regfunc = 1;
5057 else if (IS(argv[arg], "-m"))
5058 multi_seg = 1;
5059 else
5060 break;
5061 }
5062
5063 if (argc < arg + 3) {
5064 printf("usage:\n%s [-v] [-rf] [-m] <.c> <.asm> <hdrf> [rlist]*\n",
5065 argv[0]);
5066 return 1;
5067 }
5068
5069 arg_out = arg++;
5070
5071 asmfn = argv[arg++];
5072 fasm = fopen(asmfn, "r");
5073 my_assert_not(fasm, NULL);
5074
5075 hdrfn = argv[arg++];
5076 g_fhdr = fopen(hdrfn, "r");
5077 my_assert_not(g_fhdr, NULL);
5078
5079 rlist_alloc = 64;
5080 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
5081 my_assert_not(rlist, NULL);
5082 // needs special handling..
5083 rlist[rlist_len++] = "__alloca_probe";
5084
5085 func_chunk_alloc = 32;
5086 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
5087 my_assert_not(func_chunks, NULL);
5088
5089 memset(words, 0, sizeof(words));
5090
5091 for (; arg < argc; arg++) {
5092 frlist = fopen(argv[arg], "r");
5093 my_assert_not(frlist, NULL);
5094
5095 while (fgets(line, sizeof(line), frlist)) {
5096 p = sskip(line);
5097 if (*p == 0 || *p == ';')
5098 continue;
5099 if (*p == '#') {
5100 if (IS_START(p, "#if 0")
5101 || (g_allow_regfunc && IS_START(p, "#if NO_REGFUNC")))
5102 {
5103 skip_func = 1;
5104 }
5105 else if (IS_START(p, "#endif"))
5106 skip_func = 0;
5107 continue;
5108 }
5109 if (skip_func)
5110 continue;
5111
5112 p = next_word(words[0], sizeof(words[0]), p);
5113 if (words[0][0] == 0)
5114 continue;
5115
5116 if (rlist_len >= rlist_alloc) {
5117 rlist_alloc = rlist_alloc * 2 + 64;
5118 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
5119 my_assert_not(rlist, NULL);
5120 }
5121 rlist[rlist_len++] = strdup(words[0]);
5122 }
5123 skip_func = 0;
5124
5125 fclose(frlist);
5126 frlist = NULL;
5127 }
5128
5129 if (rlist_len > 0)
5130 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
5131
5132 fout = fopen(argv[arg_out], "w");
5133 my_assert_not(fout, NULL);
5134
5135 eq_alloc = 128;
5136 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
5137 my_assert_not(g_eqs, NULL);
5138
5139 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
5140 g_label_refs[i].i = -1;
5141 g_label_refs[i].next = NULL;
5142 }
5143
5144 while (fgets(line, sizeof(line), fasm))
5145 {
5146 wordc = 0;
5147 asmln++;
5148
5149 p = sskip(line);
5150 if (*p == 0)
5151 continue;
5152
5153 // get rid of random tabs
5154 for (i = 0; line[i] != 0; i++)
5155 if (line[i] == '\t')
5156 line[i] = ' ';
5157
5158 if (*p == ';')
5159 {
5160 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
5161 goto do_pending_endp; // eww..
5162
5163 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
5164 {
5165 static const char *attrs[] = {
5166 "bp-based frame",
5167 "library function",
5168 "static",
5169 "noreturn",
5170 "thunk",
5171 "fpd=",
5172 };
5173
5174 // parse IDA's attribute-list comment
5175 g_ida_func_attr = 0;
5176 p = sskip(p + 13);
5177
5178 for (; *p != 0; p = sskip(p)) {
5179 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
5180 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
5181 g_ida_func_attr |= 1 << i;
5182 p += strlen(attrs[i]);
5183 break;
5184 }
5185 }
5186 if (i == ARRAY_SIZE(attrs)) {
5187 anote("unparsed IDA attr: %s\n", p);
5188 break;
5189 }
5190 if (IS(attrs[i], "fpd=")) {
5191 p = next_word(words[0], sizeof(words[0]), p);
5192 // ignore for now..
5193 }
5194 }
5195 }
5196 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
5197 {
5198 p += 30;
5199 next_word(words[0], sizeof(words[0]), p);
5200 if (words[0][0] == 0)
5201 aerr("missing name for func chunk?\n");
5202
5203 if (!scanned_ahead) {
5204 add_func_chunk(fasm, words[0], asmln);
5205 func_chunks_sorted = 0;
5206 }
5207 }
5208 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
5209 {
5210 if (func_chunk_i >= 0) {
5211 if (func_chunk_i < func_chunk_cnt
5212 && IS(func_chunks[func_chunk_i].name, g_func))
5213 {
5214 // move on to next chunk
5215 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
5216 if (ret)
5217 aerr("seek failed for '%s' chunk #%d\n",
5218 g_func, func_chunk_i);
5219 asmln = func_chunks[func_chunk_i].asmln;
5220 func_chunk_i++;
5221 }
5222 else {
5223 if (func_chunk_ret == 0)
5224 aerr("no return from chunk?\n");
5225 fseek(fasm, func_chunk_ret, SEEK_SET);
5226 asmln = func_chunk_ret_ln;
5227 func_chunk_ret = 0;
5228 pending_endp = 1;
5229 }
5230 }
5231 }
5232 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
5233 func_chunks_used = 1;
5234 p += 20;
5235 if (IS_START(g_func, "sub_")) {
5236 unsigned long addr = strtoul(p, NULL, 16);
5237 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
5238 if (addr > f_addr && !scanned_ahead) {
5239 anote("scan_ahead caused by '%s', addr %lx\n",
5240 g_func, addr);
5241 scan_ahead(fasm);
5242 scanned_ahead = 1;
5243 func_chunks_sorted = 0;
5244 }
5245 }
5246 }
5247 continue;
5248 } // *p == ';'
5249
5250parse_words:
5251 for (i = wordc; i < ARRAY_SIZE(words); i++)
5252 words[i][0] = 0;
5253 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
5254 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
5255 if (*p == 0 || *p == ';') {
5256 wordc++;
5257 break;
5258 }
5259 }
5260 if (*p != 0 && *p != ';')
5261 aerr("too many words\n");
5262
5263 // alow asm patches in comments
5264 if (*p == ';') {
5265 if (IS_START(p, "; sctpatch:")) {
5266 p = sskip(p + 11);
5267 if (*p == 0 || *p == ';')
5268 continue;
5269 goto parse_words; // lame
5270 }
5271 if (IS_START(p, "; sctproto:")) {
5272 sctproto = strdup(p + 11);
5273 }
5274 else if (IS_START(p, "; sctend")) {
5275 end = 1;
5276 if (!pending_endp)
5277 break;
5278 }
5279 }
5280
5281 if (wordc == 0) {
5282 // shouldn't happen
5283 awarn("wordc == 0?\n");
5284 continue;
5285 }
5286
5287 // don't care about this:
5288 if (words[0][0] == '.'
5289 || IS(words[0], "include")
5290 || IS(words[0], "assume") || IS(words[1], "segment")
5291 || IS(words[0], "align"))
5292 {
5293 continue;
5294 }
5295
5296do_pending_endp:
5297 // do delayed endp processing to collect switch jumptables
5298 if (pending_endp) {
5299 if (in_func && !skip_func && !end && wordc >= 2
5300 && ((words[0][0] == 'd' && words[0][2] == 0)
5301 || (words[1][0] == 'd' && words[1][2] == 0)))
5302 {
5303 i = 1;
5304 if (words[1][0] == 'd' && words[1][2] == 0) {
5305 // label
5306 if (g_func_pd_cnt >= pd_alloc) {
5307 pd_alloc = pd_alloc * 2 + 16;
5308 g_func_pd = realloc(g_func_pd,
5309 sizeof(g_func_pd[0]) * pd_alloc);
5310 my_assert_not(g_func_pd, NULL);
5311 }
5312 pd = &g_func_pd[g_func_pd_cnt];
5313 g_func_pd_cnt++;
5314 memset(pd, 0, sizeof(*pd));
5315 strcpy(pd->label, words[0]);
5316 pd->type = OPT_CONST;
5317 pd->lmod = lmod_from_directive(words[1]);
5318 i = 2;
5319 }
5320 else {
5321 if (pd == NULL) {
5322 if (verbose)
5323 anote("skipping alignment byte?\n");
5324 continue;
5325 }
5326 lmod = lmod_from_directive(words[0]);
5327 if (lmod != pd->lmod)
5328 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
5329 }
5330
5331 if (pd->count_alloc < pd->count + wordc) {
5332 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
5333 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
5334 my_assert_not(pd->d, NULL);
5335 }
5336 for (; i < wordc; i++) {
5337 if (IS(words[i], "offset")) {
5338 pd->type = OPT_OFFSET;
5339 i++;
5340 }
5341 p = strchr(words[i], ',');
5342 if (p != NULL)
5343 *p = 0;
5344 if (pd->type == OPT_OFFSET)
5345 pd->d[pd->count].u.label = strdup(words[i]);
5346 else
5347 pd->d[pd->count].u.val = parse_number(words[i]);
5348 pd->d[pd->count].bt_i = -1;
5349 pd->count++;
5350 }
5351 continue;
5352 }
5353
5354 if (in_func && !skip_func)
5355 gen_func(fout, g_fhdr, g_func, pi);
5356
5357 pending_endp = 0;
5358 in_func = 0;
5359 g_ida_func_attr = 0;
5360 skip_warned = 0;
5361 skip_func = 0;
5362 g_func[0] = 0;
5363 func_chunks_used = 0;
5364 func_chunk_i = -1;
5365 if (pi != 0) {
5366 memset(&ops, 0, pi * sizeof(ops[0]));
5367 memset(g_labels, 0, pi * sizeof(g_labels[0]));
5368 pi = 0;
5369 }
5370 g_eqcnt = 0;
5371 for (i = 0; i < g_func_pd_cnt; i++) {
5372 pd = &g_func_pd[i];
5373 if (pd->type == OPT_OFFSET) {
5374 for (j = 0; j < pd->count; j++)
5375 free(pd->d[j].u.label);
5376 }
5377 free(pd->d);
5378 pd->d = NULL;
5379 }
5380 g_func_pd_cnt = 0;
5381 pd = NULL;
5382
5383 if (end)
5384 break;
5385 if (wordc == 0)
5386 continue;
5387 }
5388
5389 if (IS(words[1], "proc")) {
5390 if (in_func)
5391 aerr("proc '%s' while in_func '%s'?\n",
5392 words[0], g_func);
5393 p = words[0];
5394 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
5395 skip_func = 1;
5396 strcpy(g_func, words[0]);
5397 set_label(0, words[0]);
5398 in_func = 1;
5399 continue;
5400 }
5401
5402 if (IS(words[1], "endp"))
5403 {
5404 if (!in_func)
5405 aerr("endp '%s' while not in_func?\n", words[0]);
5406 if (!IS(g_func, words[0]))
5407 aerr("endp '%s' while in_func '%s'?\n",
5408 words[0], g_func);
5409
5410 if ((g_ida_func_attr & IDAFA_THUNK) && pi == 1
5411 && ops[0].op == OP_JMP && ops[0].operand[0].had_ds)
5412 {
5413 // import jump
5414 skip_func = 1;
5415 }
5416
5417 if (!skip_func && func_chunks_used) {
5418 // start processing chunks
5419 struct chunk_item *ci, key = { g_func, 0 };
5420
5421 func_chunk_ret = ftell(fasm);
5422 func_chunk_ret_ln = asmln;
5423 if (!func_chunks_sorted) {
5424 qsort(func_chunks, func_chunk_cnt,
5425 sizeof(func_chunks[0]), cmp_chunks);
5426 func_chunks_sorted = 1;
5427 }
5428 ci = bsearch(&key, func_chunks, func_chunk_cnt,
5429 sizeof(func_chunks[0]), cmp_chunks);
5430 if (ci == NULL)
5431 aerr("'%s' needs chunks, but none found\n", g_func);
5432 func_chunk_i = ci - func_chunks;
5433 for (; func_chunk_i > 0; func_chunk_i--)
5434 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
5435 break;
5436
5437 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
5438 if (ret)
5439 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
5440 asmln = func_chunks[func_chunk_i].asmln;
5441 func_chunk_i++;
5442 continue;
5443 }
5444 pending_endp = 1;
5445 continue;
5446 }
5447
5448 if (wordc == 2 && IS(words[1], "ends")) {
5449 if (!multi_seg) {
5450 end = 1;
5451 if (pending_endp)
5452 goto do_pending_endp;
5453 break;
5454 }
5455
5456 // scan for next text segment
5457 while (fgets(line, sizeof(line), fasm)) {
5458 asmln++;
5459 p = sskip(line);
5460 if (*p == 0 || *p == ';')
5461 continue;
5462
5463 if (strstr(p, "segment para public 'CODE' use32"))
5464 break;
5465 }
5466
5467 continue;
5468 }
5469
5470 p = strchr(words[0], ':');
5471 if (p != NULL) {
5472 set_label(pi, words[0]);
5473 continue;
5474 }
5475
5476 if (!in_func || skip_func) {
5477 if (!skip_warned && !skip_func && g_labels[pi][0] != 0) {
5478 if (verbose)
5479 anote("skipping from '%s'\n", g_labels[pi]);
5480 skip_warned = 1;
5481 }
5482 g_labels[pi][0] = 0;
5483 continue;
5484 }
5485
5486 if (wordc > 1 && IS(words[1], "="))
5487 {
5488 if (wordc != 5)
5489 aerr("unhandled equ, wc=%d\n", wordc);
5490 if (g_eqcnt >= eq_alloc) {
5491 eq_alloc *= 2;
5492 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
5493 my_assert_not(g_eqs, NULL);
5494 }
5495
5496 len = strlen(words[0]);
5497 if (len > sizeof(g_eqs[0].name) - 1)
5498 aerr("equ name too long: %d\n", len);
5499 strcpy(g_eqs[g_eqcnt].name, words[0]);
5500
5501 if (!IS(words[3], "ptr"))
5502 aerr("unhandled equ\n");
5503 if (IS(words[2], "dword"))
5504 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
5505 else if (IS(words[2], "word"))
5506 g_eqs[g_eqcnt].lmod = OPLM_WORD;
5507 else if (IS(words[2], "byte"))
5508 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
5509 else
5510 aerr("bad lmod: '%s'\n", words[2]);
5511
5512 g_eqs[g_eqcnt].offset = parse_number(words[4]);
5513 g_eqcnt++;
5514 continue;
5515 }
5516
5517 if (pi >= ARRAY_SIZE(ops))
5518 aerr("too many ops\n");
5519
5520 parse_op(&ops[pi], words, wordc);
5521
5522 if (sctproto != NULL) {
5523 if (ops[pi].op == OP_CALL || ops[pi].op == OP_JMP)
5524 ops[pi].datap = sctproto;
5525 sctproto = NULL;
5526 }
5527 pi++;
5528 }
5529
5530 fclose(fout);
5531 fclose(fasm);
5532 fclose(g_fhdr);
5533
5534 return 0;
5535}
5536
5537// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.