notaz.gp2x.de / ia32rtools.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
translate: many improvements for header gen
[ia32rtools.git] / tools / translate.c
... / ...
CommitLineData
1/*
2 * ia32rtools
3 * (C) notaz, 2013-2015
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 from code generation */
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 */
52 OPF_FARGNR = (1 << 12), /* push collected as func arg (no reuse) */
53 OPF_EBP_S = (1 << 13), /* ebp used as scratch here, not BP */
54 OPF_DF = (1 << 14), /* DF flag set */
55 OPF_ATAIL = (1 << 15), /* tail call with reused arg frame */
56 OPF_32BIT = (1 << 16), /* 32bit division */
57 OPF_LOCK = (1 << 17), /* op has lock prefix */
58 OPF_VAPUSH = (1 << 18), /* vararg ptr push (as call arg) */
59 OPF_DONE = (1 << 19), /* already fully handled by analysis */
60};
61
62enum op_op {
63 OP_INVAL,
64 OP_NOP,
65 OP_PUSH,
66 OP_POP,
67 OP_LEAVE,
68 OP_MOV,
69 OP_LEA,
70 OP_MOVZX,
71 OP_MOVSX,
72 OP_XCHG,
73 OP_NOT,
74 OP_CDQ,
75 OP_LODS,
76 OP_STOS,
77 OP_MOVS,
78 OP_CMPS,
79 OP_SCAS,
80 OP_STD,
81 OP_CLD,
82 OP_RET,
83 OP_ADD,
84 OP_SUB,
85 OP_AND,
86 OP_OR,
87 OP_XOR,
88 OP_SHL,
89 OP_SHR,
90 OP_SAR,
91 OP_SHRD,
92 OP_ROL,
93 OP_ROR,
94 OP_RCL,
95 OP_RCR,
96 OP_ADC,
97 OP_SBB,
98 OP_BSF,
99 OP_INC,
100 OP_DEC,
101 OP_NEG,
102 OP_MUL,
103 OP_IMUL,
104 OP_DIV,
105 OP_IDIV,
106 OP_TEST,
107 OP_CMP,
108 OP_CALL,
109 OP_JMP,
110 OP_JECXZ,
111 OP_JCC,
112 OP_SCC,
113 // x87
114 // mmx
115 OP_EMMS,
116 // mmx
117 OP_UD2,
118};
119
120enum opr_type {
121 OPT_UNSPEC,
122 OPT_REG,
123 OPT_REGMEM,
124 OPT_LABEL,
125 OPT_OFFSET,
126 OPT_CONST,
127};
128
129// must be sorted (larger len must be further in enum)
130enum opr_lenmod {
131 OPLM_UNSPEC,
132 OPLM_BYTE,
133 OPLM_WORD,
134 OPLM_DWORD,
135 OPLM_QWORD,
136};
137
138#define MAX_OPERANDS 3
139#define NAMELEN 112
140
141struct parsed_opr {
142 enum opr_type type;
143 enum opr_lenmod lmod;
144 unsigned int is_ptr:1; // pointer in C
145 unsigned int is_array:1; // array in C
146 unsigned int type_from_var:1; // .. in header, sometimes wrong
147 unsigned int size_mismatch:1; // type override differs from C
148 unsigned int size_lt:1; // type override is larger than C
149 unsigned int had_ds:1; // had ds: prefix
150 const struct parsed_proto *pp; // for OPT_LABEL
151 int reg;
152 unsigned int val;
153 char name[NAMELEN];
154};
155
156struct parsed_op {
157 enum op_op op;
158 struct parsed_opr operand[MAX_OPERANDS];
159 unsigned int flags;
160 unsigned char pfo;
161 unsigned char pfo_inv;
162 unsigned char operand_cnt;
163 unsigned char p_argnum; // arg push: altered before call arg #
164 unsigned char p_arggrp; // arg push: arg group # for above
165 unsigned char p_argpass;// arg push: arg of host func
166 short p_argnext;// arg push: same arg pushed elsewhere or -1
167 int regmask_src; // all referensed regs
168 int regmask_dst;
169 int pfomask; // flagop: parsed_flag_op that can't be delayed
170 int cc_scratch; // scratch storage during analysis
171 int bt_i; // branch target for branches
172 struct parsed_data *btj;// branch targets for jumptables
173 struct parsed_proto *pp;// parsed_proto for OP_CALL
174 void *datap;
175 int asmln;
176};
177
178// datap:
179// OP_CALL - parser proto hint (str)
180// (OPF_CC) - points to one of (OPF_FLAGS) that affects cc op
181// OP_POP - points to OP_PUSH in push/pop pair
182
183struct parsed_equ {
184 char name[64];
185 enum opr_lenmod lmod;
186 int offset;
187};
188
189struct parsed_data {
190 char label[256];
191 enum opr_type type;
192 enum opr_lenmod lmod;
193 int count;
194 int count_alloc;
195 struct {
196 union {
197 char *label;
198 unsigned int val;
199 } u;
200 int bt_i;
201 } *d;
202};
203
204struct label_ref {
205 int i;
206 struct label_ref *next;
207};
208
209enum ida_func_attr {
210 IDAFA_BP_FRAME = (1 << 0),
211 IDAFA_LIB_FUNC = (1 << 1),
212 IDAFA_STATIC = (1 << 2),
213 IDAFA_NORETURN = (1 << 3),
214 IDAFA_THUNK = (1 << 4),
215 IDAFA_FPD = (1 << 5),
216};
217
218// note: limited to 32k due to p_argnext
219#define MAX_OPS 4096
220#define MAX_ARG_GRP 2
221
222static struct parsed_op ops[MAX_OPS];
223static struct parsed_equ *g_eqs;
224static int g_eqcnt;
225static char *g_labels[MAX_OPS];
226static struct label_ref g_label_refs[MAX_OPS];
227static const struct parsed_proto *g_func_pp;
228static struct parsed_data *g_func_pd;
229static int g_func_pd_cnt;
230static char g_func[256];
231static char g_comment[256];
232static int g_bp_frame;
233static int g_sp_frame;
234static int g_stack_frame_used;
235static int g_stack_fsz;
236static int g_ida_func_attr;
237static int g_allow_regfunc;
238static int g_quiet_pp;
239static int g_header_mode;
240
241#define ferr(op_, fmt, ...) do { \
242 printf("%s:%d: error: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
243 dump_op(op_), ##__VA_ARGS__); \
244 fcloseall(); \
245 exit(1); \
246} while (0)
247#define fnote(op_, fmt, ...) \
248 printf("%s:%d: note: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
249 dump_op(op_), ##__VA_ARGS__)
250
251#define ferr_assert(op_, cond) do { \
252 if (!(cond)) ferr(op_, "assertion '%s' failed on ln :%d\n", #cond, \
253 __LINE__); \
254} while (0)
255
256const char *regs_r32[] = {
257 "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp",
258 // not r32, but list here for easy parsing and printing
259 "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7",
260};
261const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" };
262const char *regs_r8l[] = { "al", "bl", "cl", "dl" };
263const char *regs_r8h[] = { "ah", "bh", "ch", "dh" };
264
265enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP };
266
267// possible basic comparison types (without inversion)
268enum parsed_flag_op {
269 PFO_O, // 0 OF=1
270 PFO_C, // 2 CF=1
271 PFO_Z, // 4 ZF=1
272 PFO_BE, // 6 CF=1||ZF=1
273 PFO_S, // 8 SF=1
274 PFO_P, // a PF=1
275 PFO_L, // c SF!=OF
276 PFO_LE, // e ZF=1||SF!=OF
277};
278
279#define PFOB_O (1 << PFO_O)
280#define PFOB_C (1 << PFO_C)
281#define PFOB_Z (1 << PFO_Z)
282#define PFOB_S (1 << PFO_S)
283
284static const char *parsed_flag_op_names[] = {
285 "o", "c", "z", "be", "s", "p", "l", "le"
286};
287
288static int char_array_i(const char *array[], size_t len, const char *s)
289{
290 int i;
291
292 for (i = 0; i < len; i++)
293 if (IS(s, array[i]))
294 return i;
295
296 return -1;
297}
298
299static void printf_number(char *buf, size_t buf_size,
300 unsigned long number)
301{
302 // output in C-friendly form
303 snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
304}
305
306static int check_segment_prefix(const char *s)
307{
308 if (s[0] == 0 || s[1] != 's' || s[2] != ':')
309 return 0;
310
311 switch (s[0]) {
312 case 'c': return 1;
313 case 'd': return 2;
314 case 's': return 3;
315 case 'e': return 4;
316 case 'f': return 5;
317 case 'g': return 6;
318 default: return 0;
319 }
320}
321
322static int parse_reg(enum opr_lenmod *reg_lmod, const char *s)
323{
324 int reg;
325
326 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s);
327 if (reg >= 8) {
328 *reg_lmod = OPLM_QWORD;
329 return reg;
330 }
331 if (reg >= 0) {
332 *reg_lmod = OPLM_DWORD;
333 return reg;
334 }
335 reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s);
336 if (reg >= 0) {
337 *reg_lmod = OPLM_WORD;
338 return reg;
339 }
340 reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s);
341 if (reg >= 0) {
342 *reg_lmod = OPLM_BYTE;
343 return reg;
344 }
345 reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s);
346 if (reg >= 0) {
347 *reg_lmod = OPLM_BYTE;
348 return reg;
349 }
350
351 return -1;
352}
353
354static int parse_indmode(char *name, int *regmask, int need_c_cvt)
355{
356 enum opr_lenmod lmod;
357 char cvtbuf[256];
358 char *d = cvtbuf;
359 char *s = name;
360 char w[64];
361 long number;
362 int reg;
363 int c = 0;
364
365 *d = 0;
366
367 while (*s != 0) {
368 d += strlen(d);
369 while (my_isblank(*s))
370 s++;
371 for (; my_issep(*s); d++, s++)
372 *d = *s;
373 while (my_isblank(*s))
374 s++;
375 *d = 0;
376
377 // skip '?s:' prefixes
378 if (check_segment_prefix(s))
379 s += 3;
380
381 s = next_idt(w, sizeof(w), s);
382 if (w[0] == 0)
383 break;
384 c++;
385
386 reg = parse_reg(&lmod, w);
387 if (reg >= 0) {
388 *regmask |= 1 << reg;
389 goto pass;
390 }
391
392 if ('0' <= w[0] && w[0] <= '9') {
393 number = parse_number(w);
394 printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number);
395 continue;
396 }
397
398 // probably some label/identifier - pass
399
400pass:
401 snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w);
402 }
403
404 if (need_c_cvt)
405 strcpy(name, cvtbuf);
406
407 return c;
408}
409
410static int is_reg_in_str(const char *s)
411{
412 int i;
413
414 if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
415 return 0;
416
417 for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
418 if (!strncmp(s, regs_r32[i], 3))
419 return 1;
420
421 return 0;
422}
423
424static const char *parse_stack_el(const char *name, char *extra_reg,
425 int early_try)
426{
427 const char *p, *p2, *s;
428 char *endp = NULL;
429 char buf[32];
430 long val;
431 int len;
432
433 if (g_bp_frame || early_try)
434 {
435 p = name;
436 if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
437 p += 4;
438 if (extra_reg != NULL) {
439 strncpy(extra_reg, name, 3);
440 extra_reg[4] = 0;
441 }
442 }
443
444 if (IS_START(p, "ebp+")) {
445 p += 4;
446
447 p2 = strchr(p, '+');
448 if (p2 != NULL && is_reg_in_str(p)) {
449 if (extra_reg != NULL) {
450 strncpy(extra_reg, p, p2 - p);
451 extra_reg[p2 - p] = 0;
452 }
453 p = p2 + 1;
454 }
455
456 if (!('0' <= *p && *p <= '9'))
457 return p;
458
459 return NULL;
460 }
461 }
462
463 if (!IS_START(name, "esp+"))
464 return NULL;
465
466 s = name + 4;
467 p = strchr(s, '+');
468 if (p) {
469 if (is_reg_in_str(s)) {
470 if (extra_reg != NULL) {
471 strncpy(extra_reg, s, p - s);
472 extra_reg[p - s] = 0;
473 }
474 s = p + 1;
475 p = strchr(s, '+');
476 if (p == NULL)
477 aerr("%s IDA stackvar not set?\n", __func__);
478 }
479 if (!('0' <= *s && *s <= '9')) {
480 aerr("%s IDA stackvar offset not set?\n", __func__);
481 return NULL;
482 }
483 if (s[0] == '0' && s[1] == 'x')
484 s += 2;
485 len = p - s;
486 if (len < sizeof(buf) - 1) {
487 strncpy(buf, s, len);
488 buf[len] = 0;
489 val = strtol(buf, &endp, 16);
490 if (val == 0 || *endp != 0) {
491 aerr("%s num parse fail for '%s'\n", __func__, buf);
492 return NULL;
493 }
494 }
495 p++;
496 }
497 else
498 p = name + 4;
499
500 if ('0' <= *p && *p <= '9')
501 return NULL;
502
503 return p;
504}
505
506static int guess_lmod_from_name(struct parsed_opr *opr)
507{
508 if (!strncmp(opr->name, "dword_", 6)) {
509 opr->lmod = OPLM_DWORD;
510 return 1;
511 }
512 if (!strncmp(opr->name, "word_", 5)) {
513 opr->lmod = OPLM_WORD;
514 return 1;
515 }
516 if (!strncmp(opr->name, "byte_", 5)) {
517 opr->lmod = OPLM_BYTE;
518 return 1;
519 }
520 if (!strncmp(opr->name, "qword_", 6)) {
521 opr->lmod = OPLM_QWORD;
522 return 1;
523 }
524 return 0;
525}
526
527static int guess_lmod_from_c_type(enum opr_lenmod *lmod,
528 const struct parsed_type *c_type)
529{
530 static const char *dword_types[] = {
531 "int", "_DWORD", "UINT_PTR", "DWORD",
532 "WPARAM", "LPARAM", "UINT", "__int32",
533 "LONG", "HIMC", "BOOL", "size_t",
534 "float",
535 };
536 static const char *word_types[] = {
537 "uint16_t", "int16_t", "_WORD", "WORD",
538 "unsigned __int16", "__int16",
539 };
540 static const char *byte_types[] = {
541 "uint8_t", "int8_t", "char",
542 "unsigned __int8", "__int8", "BYTE", "_BYTE",
543 "CHAR", "_UNKNOWN",
544 // structures.. deal the same as with _UNKNOWN for now
545 "CRITICAL_SECTION",
546 };
547 const char *n;
548 int i;
549
550 if (c_type->is_ptr) {
551 *lmod = OPLM_DWORD;
552 return 1;
553 }
554
555 n = skip_type_mod(c_type->name);
556
557 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
558 if (IS(n, dword_types[i])) {
559 *lmod = OPLM_DWORD;
560 return 1;
561 }
562 }
563
564 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
565 if (IS(n, word_types[i])) {
566 *lmod = OPLM_WORD;
567 return 1;
568 }
569 }
570
571 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
572 if (IS(n, byte_types[i])) {
573 *lmod = OPLM_BYTE;
574 return 1;
575 }
576 }
577
578 return 0;
579}
580
581static char *default_cast_to(char *buf, size_t buf_size,
582 struct parsed_opr *opr)
583{
584 buf[0] = 0;
585
586 if (!opr->is_ptr)
587 return buf;
588 if (opr->pp == NULL || opr->pp->type.name == NULL
589 || opr->pp->is_fptr)
590 {
591 snprintf(buf, buf_size, "%s", "(void *)");
592 return buf;
593 }
594
595 snprintf(buf, buf_size, "(%s)", opr->pp->type.name);
596 return buf;
597}
598
599static enum opr_type lmod_from_directive(const char *d)
600{
601 if (IS(d, "dd"))
602 return OPLM_DWORD;
603 else if (IS(d, "dw"))
604 return OPLM_WORD;
605 else if (IS(d, "db"))
606 return OPLM_BYTE;
607
608 aerr("unhandled directive: '%s'\n", d);
609 return OPLM_UNSPEC;
610}
611
612static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
613 int *regmask)
614{
615 opr->type = OPT_REG;
616 opr->reg = reg;
617 opr->lmod = lmod;
618 *regmask |= 1 << reg;
619}
620
621static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
622 int *extra_offs);
623
624static int parse_operand(struct parsed_opr *opr,
625 int *regmask, int *regmask_indirect,
626 char words[16][256], int wordc, int w, unsigned int op_flags)
627{
628 const struct parsed_proto *pp = NULL;
629 enum opr_lenmod tmplmod;
630 unsigned long number;
631 char buf[256];
632 int ret, len;
633 int wordc_in;
634 char *p;
635 int i;
636
637 if (w >= wordc)
638 aerr("parse_operand w %d, wordc %d\n", w, wordc);
639
640 opr->reg = xUNSPEC;
641
642 for (i = w; i < wordc; i++) {
643 len = strlen(words[i]);
644 if (words[i][len - 1] == ',') {
645 words[i][len - 1] = 0;
646 wordc = i + 1;
647 break;
648 }
649 }
650
651 wordc_in = wordc - w;
652
653 if ((op_flags & OPF_JMP) && wordc_in > 0
654 && !('0' <= words[w][0] && words[w][0] <= '9'))
655 {
656 const char *label = NULL;
657
658 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
659 && IS(words[w + 1], "ptr"))
660 label = words[w + 2];
661 else if (wordc_in == 2 && IS(words[w], "short"))
662 label = words[w + 1];
663 else if (wordc_in == 1
664 && strchr(words[w], '[') == NULL
665 && parse_reg(&tmplmod, words[w]) < 0)
666 label = words[w];
667
668 if (label != NULL) {
669 opr->type = OPT_LABEL;
670 ret = check_segment_prefix(label);
671 if (ret != 0) {
672 if (ret >= 5)
673 aerr("fs/gs used\n");
674 opr->had_ds = 1;
675 label += 3;
676 }
677 strcpy(opr->name, label);
678 return wordc;
679 }
680 }
681
682 if (wordc_in >= 3) {
683 if (IS(words[w + 1], "ptr")) {
684 if (IS(words[w], "dword"))
685 opr->lmod = OPLM_DWORD;
686 else if (IS(words[w], "word"))
687 opr->lmod = OPLM_WORD;
688 else if (IS(words[w], "byte"))
689 opr->lmod = OPLM_BYTE;
690 else if (IS(words[w], "qword"))
691 opr->lmod = OPLM_QWORD;
692 else
693 aerr("type parsing failed\n");
694 w += 2;
695 wordc_in = wordc - w;
696 }
697 }
698
699 if (wordc_in == 2) {
700 if (IS(words[w], "offset")) {
701 opr->type = OPT_OFFSET;
702 opr->lmod = OPLM_DWORD;
703 strcpy(opr->name, words[w + 1]);
704 pp = proto_parse(g_fhdr, opr->name, 1);
705 goto do_label;
706 }
707 if (IS(words[w], "(offset")) {
708 p = strchr(words[w + 1], ')');
709 if (p == NULL)
710 aerr("parse of bracketed offset failed\n");
711 *p = 0;
712 opr->type = OPT_OFFSET;
713 strcpy(opr->name, words[w + 1]);
714 return wordc;
715 }
716 }
717
718 if (wordc_in != 1)
719 aerr("parse_operand 1 word expected\n");
720
721 ret = check_segment_prefix(words[w]);
722 if (ret != 0) {
723 if (ret >= 5)
724 aerr("fs/gs used\n");
725 opr->had_ds = 1;
726 memmove(words[w], words[w] + 3, strlen(words[w]) - 2);
727 }
728 strcpy(opr->name, words[w]);
729
730 if (words[w][0] == '[') {
731 opr->type = OPT_REGMEM;
732 ret = sscanf(words[w], "[%[^]]]", opr->name);
733 if (ret != 1)
734 aerr("[] parse failure\n");
735
736 parse_indmode(opr->name, regmask_indirect, 1);
737 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL, 1))
738 {
739 // might be an equ
740 struct parsed_equ *eq =
741 equ_find(NULL, parse_stack_el(opr->name, NULL, 1), &i);
742 if (eq)
743 opr->lmod = eq->lmod;
744 }
745 return wordc;
746 }
747 else if (strchr(words[w], '[')) {
748 // label[reg] form
749 p = strchr(words[w], '[');
750 opr->type = OPT_REGMEM;
751 parse_indmode(p, regmask_indirect, 0);
752 strncpy(buf, words[w], p - words[w]);
753 buf[p - words[w]] = 0;
754 pp = proto_parse(g_fhdr, buf, 1);
755 goto do_label;
756 }
757 else if (('0' <= words[w][0] && words[w][0] <= '9')
758 || words[w][0] == '-')
759 {
760 number = parse_number(words[w]);
761 opr->type = OPT_CONST;
762 opr->val = number;
763 printf_number(opr->name, sizeof(opr->name), number);
764 return wordc;
765 }
766
767 ret = parse_reg(&tmplmod, opr->name);
768 if (ret >= 0) {
769 setup_reg_opr(opr, ret, tmplmod, regmask);
770 return wordc;
771 }
772
773 // most likely var in data segment
774 opr->type = OPT_LABEL;
775 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
776
777do_label:
778 if (pp != NULL) {
779 if (pp->is_fptr || pp->is_func) {
780 opr->lmod = OPLM_DWORD;
781 opr->is_ptr = 1;
782 }
783 else {
784 tmplmod = OPLM_UNSPEC;
785 if (!guess_lmod_from_c_type(&tmplmod, &pp->type))
786 anote("unhandled C type '%s' for '%s'\n",
787 pp->type.name, opr->name);
788
789 if (opr->lmod == OPLM_UNSPEC) {
790 opr->lmod = tmplmod;
791 opr->type_from_var = 1;
792 }
793 else if (opr->lmod != tmplmod) {
794 opr->size_mismatch = 1;
795 if (tmplmod < opr->lmod)
796 opr->size_lt = 1;
797 }
798 opr->is_ptr = pp->type.is_ptr;
799 }
800 opr->is_array = pp->type.is_array;
801 }
802 opr->pp = pp;
803
804 if (opr->lmod == OPLM_UNSPEC)
805 guess_lmod_from_name(opr);
806 return wordc;
807}
808
809static const struct {
810 const char *name;
811 unsigned int flags;
812} pref_table[] = {
813 { "rep", OPF_REP },
814 { "repe", OPF_REP|OPF_REPZ },
815 { "repz", OPF_REP|OPF_REPZ },
816 { "repne", OPF_REP|OPF_REPNZ },
817 { "repnz", OPF_REP|OPF_REPNZ },
818 { "lock", OPF_LOCK }, // ignored for now..
819};
820
821#define OPF_CJMP_CC (OPF_JMP|OPF_CJMP|OPF_CC)
822
823static const struct {
824 const char *name;
825 enum op_op op;
826 unsigned short minopr;
827 unsigned short maxopr;
828 unsigned int flags;
829 unsigned char pfo;
830 unsigned char pfo_inv;
831} op_table[] = {
832 { "nop", OP_NOP, 0, 0, 0 },
833 { "push", OP_PUSH, 1, 1, 0 },
834 { "pop", OP_POP, 1, 1, OPF_DATA },
835 { "leave",OP_LEAVE, 0, 0, OPF_DATA },
836 { "mov" , OP_MOV, 2, 2, OPF_DATA },
837 { "lea", OP_LEA, 2, 2, OPF_DATA },
838 { "movzx",OP_MOVZX, 2, 2, OPF_DATA },
839 { "movsx",OP_MOVSX, 2, 2, OPF_DATA },
840 { "xchg", OP_XCHG, 2, 2, OPF_DATA },
841 { "not", OP_NOT, 1, 1, OPF_DATA },
842 { "cdq", OP_CDQ, 0, 0, OPF_DATA },
843 { "lodsb",OP_LODS, 0, 0, OPF_DATA },
844 { "lodsw",OP_LODS, 0, 0, OPF_DATA },
845 { "lodsd",OP_LODS, 0, 0, OPF_DATA },
846 { "stosb",OP_STOS, 0, 0, OPF_DATA },
847 { "stosw",OP_STOS, 0, 0, OPF_DATA },
848 { "stosd",OP_STOS, 0, 0, OPF_DATA },
849 { "movsb",OP_MOVS, 0, 0, OPF_DATA },
850 { "movsw",OP_MOVS, 0, 0, OPF_DATA },
851 { "movsd",OP_MOVS, 0, 0, OPF_DATA },
852 { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
853 { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
854 { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
855 { "scasb",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
856 { "scasw",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
857 { "scasd",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
858 { "std", OP_STD, 0, 0, OPF_DATA }, // special flag
859 { "cld", OP_CLD, 0, 0, OPF_DATA },
860 { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
861 { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
862 { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
863 { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS },
864 { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS },
865 { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
866 { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
867 { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
868 { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
869 { "shrd", OP_SHRD, 3, 3, OPF_DATA|OPF_FLAGS },
870 { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
871 { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
872 { "rcl", OP_RCL, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
873 { "rcr", OP_RCR, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
874 { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
875 { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
876 { "bsf", OP_BSF, 2, 2, OPF_DATA|OPF_FLAGS },
877 { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
878 { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS },
879 { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS },
880 { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS },
881 { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS },
882 { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS },
883 { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS },
884 { "test", OP_TEST, 2, 2, OPF_FLAGS },
885 { "cmp", OP_CMP, 2, 2, OPF_FLAGS },
886 { "retn", OP_RET, 0, 1, OPF_TAIL },
887 { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS },
888 { "jmp", OP_JMP, 1, 1, OPF_JMP },
889 { "jecxz",OP_JECXZ, 1, 1, OPF_JMP|OPF_CJMP },
890 { "jo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 0 }, // 70 OF=1
891 { "jno", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 1 }, // 71 OF=0
892 { "jc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72 CF=1
893 { "jb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72
894 { "jnc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73 CF=0
895 { "jnb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
896 { "jae", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
897 { "jz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74 ZF=1
898 { "je", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74
899 { "jnz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75 ZF=0
900 { "jne", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75
901 { "jbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76 CF=1||ZF=1
902 { "jna", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76
903 { "ja", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77 CF=0&&ZF=0
904 { "jnbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77
905 { "js", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 0 }, // 78 SF=1
906 { "jns", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 1 }, // 79 SF=0
907 { "jp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a PF=1
908 { "jpe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a
909 { "jnp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b PF=0
910 { "jpo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b
911 { "jl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c SF!=OF
912 { "jnge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c
913 { "jge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d SF=OF
914 { "jnl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d
915 { "jle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e ZF=1||SF!=OF
916 { "jng", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e
917 { "jg", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f ZF=0&&SF=OF
918 { "jnle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f
919 { "seto", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 0 },
920 { "setno", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 1 },
921 { "setc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
922 { "setb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
923 { "setnc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
924 { "setae", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
925 { "setnb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
926 { "setz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
927 { "sete", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
928 { "setnz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
929 { "setne", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
930 { "setbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
931 { "setna", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
932 { "seta", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
933 { "setnbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
934 { "sets", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 0 },
935 { "setns", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 1 },
936 { "setp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
937 { "setpe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
938 { "setnp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
939 { "setpo", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
940 { "setl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
941 { "setnge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
942 { "setge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
943 { "setnl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
944 { "setle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
945 { "setng", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
946 { "setg", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
947 { "setnle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
948 // x87
949 // mmx
950 { "emms", OP_EMMS, 0, 0, OPF_DATA },
951 { "movq", OP_MOV, 2, 2, OPF_DATA },
952 // must be last
953 { "ud2", OP_UD2 },
954};
955
956static void parse_op(struct parsed_op *op, char words[16][256], int wordc)
957{
958 enum opr_lenmod lmod = OPLM_UNSPEC;
959 int prefix_flags = 0;
960 int regmask_ind;
961 int regmask;
962 int op_w = 0;
963 int opr = 0;
964 int w = 0;
965 int i;
966
967 for (i = 0; i < ARRAY_SIZE(pref_table); i++) {
968 if (IS(words[w], pref_table[i].name)) {
969 prefix_flags = pref_table[i].flags;
970 break;
971 }
972 }
973
974 if (prefix_flags) {
975 if (wordc <= 1)
976 aerr("lone prefix: '%s'\n", words[0]);
977 w++;
978 }
979
980 op_w = w;
981 for (i = 0; i < ARRAY_SIZE(op_table); i++) {
982 if (IS(words[w], op_table[i].name))
983 break;
984 }
985
986 if (i == ARRAY_SIZE(op_table)) {
987 anote("unhandled op: '%s'\n", words[0]);
988 i--; // OP_UD2
989 }
990 w++;
991
992 op->op = op_table[i].op;
993 op->flags = op_table[i].flags | prefix_flags;
994 op->pfo = op_table[i].pfo;
995 op->pfo_inv = op_table[i].pfo_inv;
996 op->regmask_src = op->regmask_dst = 0;
997 op->asmln = asmln;
998
999 if (op->op == OP_UD2)
1000 return;
1001
1002 for (opr = 0; opr < op_table[i].maxopr; opr++) {
1003 if (opr >= op_table[i].minopr && w >= wordc)
1004 break;
1005
1006 regmask = regmask_ind = 0;
1007 w = parse_operand(&op->operand[opr], &regmask, &regmask_ind,
1008 words, wordc, w, op->flags);
1009
1010 if (opr == 0 && (op->flags & OPF_DATA))
1011 op->regmask_dst = regmask;
1012 else
1013 op->regmask_src |= regmask;
1014 op->regmask_src |= regmask_ind;
1015 }
1016
1017 if (w < wordc)
1018 aerr("parse_op %s incomplete: %d/%d\n",
1019 words[0], w, wordc);
1020
1021 // special cases
1022 op->operand_cnt = opr;
1023 if (!strncmp(op_table[i].name, "set", 3))
1024 op->operand[0].lmod = OPLM_BYTE;
1025
1026 switch (op->op) {
1027 // first operand is not dst
1028 case OP_CMP:
1029 case OP_TEST:
1030 op->regmask_src |= op->regmask_dst;
1031 op->regmask_dst = 0;
1032 break;
1033
1034 // first operand is src too
1035 case OP_NOT:
1036 case OP_ADD:
1037 case OP_AND:
1038 case OP_OR:
1039 case OP_RCL:
1040 case OP_RCR:
1041 case OP_ADC:
1042 case OP_INC:
1043 case OP_DEC:
1044 case OP_NEG:
1045 // more below..
1046 op->regmask_src |= op->regmask_dst;
1047 break;
1048
1049 // special
1050 case OP_XCHG:
1051 op->regmask_src |= op->regmask_dst;
1052 op->regmask_dst |= op->regmask_src;
1053 goto check_align;
1054
1055 case OP_SUB:
1056 case OP_SBB:
1057 case OP_XOR:
1058 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1059 && op->operand[0].lmod == op->operand[1].lmod
1060 && op->operand[0].reg == op->operand[1].reg
1061 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
1062 {
1063 op->regmask_src = 0;
1064 }
1065 else
1066 op->regmask_src |= op->regmask_dst;
1067 break;
1068
1069 // ops with implicit argumets
1070 case OP_CDQ:
1071 op->operand_cnt = 2;
1072 setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst);
1073 setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src);
1074 break;
1075
1076 case OP_LODS:
1077 case OP_STOS:
1078 case OP_SCAS:
1079 if (op->operand_cnt != 0)
1080 break;
1081 if (words[op_w][4] == 'b')
1082 lmod = OPLM_BYTE;
1083 else if (words[op_w][4] == 'w')
1084 lmod = OPLM_WORD;
1085 else if (words[op_w][4] == 'd')
1086 lmod = OPLM_DWORD;
1087 op->operand_cnt = 3;
1088 setup_reg_opr(&op->operand[0], op->op == OP_LODS ? xSI : xDI,
1089 lmod, &op->regmask_src);
1090 setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src);
1091 op->regmask_dst = op->regmask_src;
1092 setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD,
1093 op->op == OP_LODS ? &op->regmask_dst : &op->regmask_src);
1094 break;
1095
1096 case OP_MOVS:
1097 case OP_CMPS:
1098 if (op->operand_cnt != 0)
1099 break;
1100 if (words[op_w][4] == 'b')
1101 lmod = OPLM_BYTE;
1102 else if (words[op_w][4] == 'w')
1103 lmod = OPLM_WORD;
1104 else if (words[op_w][4] == 'd')
1105 lmod = OPLM_DWORD;
1106 op->operand_cnt = 3;
1107 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
1108 setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src);
1109 setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src);
1110 op->regmask_dst = op->regmask_src;
1111 break;
1112
1113 case OP_JECXZ:
1114 op->operand_cnt = 1;
1115 op->regmask_src = 1 << xCX;
1116 op->operand[0].type = OPT_REG;
1117 op->operand[0].reg = xCX;
1118 op->operand[0].lmod = OPLM_DWORD;
1119 break;
1120
1121 case OP_IMUL:
1122 if (op->operand_cnt != 1)
1123 break;
1124 // fallthrough
1125 case OP_MUL:
1126 // singleop mul
1127 op->regmask_src |= op->regmask_dst;
1128 op->regmask_dst = (1 << xDX) | (1 << xAX);
1129 if (op->operand[0].lmod == OPLM_UNSPEC)
1130 op->operand[0].lmod = OPLM_DWORD;
1131 break;
1132
1133 case OP_DIV:
1134 case OP_IDIV:
1135 // we could set up operands for edx:eax, but there is no real need to
1136 // (see is_opr_modified())
1137 op->regmask_src |= op->regmask_dst;
1138 op->regmask_dst = (1 << xDX) | (1 << xAX);
1139 if (op->operand[0].lmod == OPLM_UNSPEC)
1140 op->operand[0].lmod = OPLM_DWORD;
1141 break;
1142
1143 case OP_SHL:
1144 case OP_SHR:
1145 case OP_SAR:
1146 case OP_ROL:
1147 case OP_ROR:
1148 op->regmask_src |= op->regmask_dst;
1149 if (op->operand[1].lmod == OPLM_UNSPEC)
1150 op->operand[1].lmod = OPLM_BYTE;
1151 break;
1152
1153 case OP_SHRD:
1154 op->regmask_src |= op->regmask_dst;
1155 if (op->operand[2].lmod == OPLM_UNSPEC)
1156 op->operand[2].lmod = OPLM_BYTE;
1157 break;
1158
1159 case OP_PUSH:
1160 op->regmask_src |= op->regmask_dst;
1161 op->regmask_dst = 0;
1162 if (op->operand[0].lmod == OPLM_UNSPEC
1163 && (op->operand[0].type == OPT_CONST
1164 || op->operand[0].type == OPT_OFFSET
1165 || op->operand[0].type == OPT_LABEL))
1166 op->operand[0].lmod = OPLM_DWORD;
1167 break;
1168
1169 // alignment
1170 case OP_MOV:
1171 check_align:
1172 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1173 && op->operand[0].lmod == op->operand[1].lmod
1174 && op->operand[0].reg == op->operand[1].reg
1175 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
1176 {
1177 op->flags |= OPF_RMD | OPF_DONE;
1178 op->regmask_src = op->regmask_dst = 0;
1179 }
1180 break;
1181
1182 case OP_LEA:
1183 if (op->operand[0].type == OPT_REG
1184 && op->operand[1].type == OPT_REGMEM)
1185 {
1186 char buf[16];
1187 snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name);
1188 if (IS(buf, op->operand[1].name))
1189 op->flags |= OPF_RMD | OPF_DONE;
1190 }
1191 break;
1192
1193 case OP_CALL:
1194 // trashed regs must be explicitly detected later
1195 op->regmask_dst = 0;
1196 break;
1197
1198 case OP_LEAVE:
1199 op->regmask_dst = (1 << xBP) | (1 << xSP);
1200 op->regmask_src = 1 << xBP;
1201 break;
1202
1203 default:
1204 break;
1205 }
1206}
1207
1208static const char *op_name(struct parsed_op *po)
1209{
1210 static char buf[16];
1211 char *p;
1212 int i;
1213
1214 if (po->op == OP_JCC || po->op == OP_SCC) {
1215 p = buf;
1216 *p++ = (po->op == OP_JCC) ? 'j' : 's';
1217 if (po->pfo_inv)
1218 *p++ = 'n';
1219 strcpy(p, parsed_flag_op_names[po->pfo]);
1220 return buf;
1221 }
1222
1223 for (i = 0; i < ARRAY_SIZE(op_table); i++)
1224 if (op_table[i].op == po->op)
1225 return op_table[i].name;
1226
1227 return "???";
1228}
1229
1230// debug
1231static const char *dump_op(struct parsed_op *po)
1232{
1233 static char out[128];
1234 char *p = out;
1235 int i;
1236
1237 if (po == NULL)
1238 return "???";
1239
1240 snprintf(out, sizeof(out), "%s", op_name(po));
1241 for (i = 0; i < po->operand_cnt; i++) {
1242 p += strlen(p);
1243 if (i > 0)
1244 *p++ = ',';
1245 snprintf(p, sizeof(out) - (p - out),
1246 po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s",
1247 po->operand[i].name);
1248 }
1249
1250 return out;
1251}
1252
1253static const char *lmod_type_u(struct parsed_op *po,
1254 enum opr_lenmod lmod)
1255{
1256 switch (lmod) {
1257 case OPLM_QWORD:
1258 return "u64";
1259 case OPLM_DWORD:
1260 return "u32";
1261 case OPLM_WORD:
1262 return "u16";
1263 case OPLM_BYTE:
1264 return "u8";
1265 default:
1266 ferr(po, "invalid lmod: %d\n", lmod);
1267 return "(_invalid_)";
1268 }
1269}
1270
1271static const char *lmod_cast_u(struct parsed_op *po,
1272 enum opr_lenmod lmod)
1273{
1274 switch (lmod) {
1275 case OPLM_QWORD:
1276 return "";
1277 case OPLM_DWORD:
1278 return "";
1279 case OPLM_WORD:
1280 return "(u16)";
1281 case OPLM_BYTE:
1282 return "(u8)";
1283 default:
1284 ferr(po, "invalid lmod: %d\n", lmod);
1285 return "(_invalid_)";
1286 }
1287}
1288
1289static const char *lmod_cast_u_ptr(struct parsed_op *po,
1290 enum opr_lenmod lmod)
1291{
1292 switch (lmod) {
1293 case OPLM_QWORD:
1294 return "*(u64 *)";
1295 case OPLM_DWORD:
1296 return "*(u32 *)";
1297 case OPLM_WORD:
1298 return "*(u16 *)";
1299 case OPLM_BYTE:
1300 return "*(u8 *)";
1301 default:
1302 ferr(po, "invalid lmod: %d\n", lmod);
1303 return "(_invalid_)";
1304 }
1305}
1306
1307static const char *lmod_cast_s(struct parsed_op *po,
1308 enum opr_lenmod lmod)
1309{
1310 switch (lmod) {
1311 case OPLM_QWORD:
1312 return "(s64)";
1313 case OPLM_DWORD:
1314 return "(s32)";
1315 case OPLM_WORD:
1316 return "(s16)";
1317 case OPLM_BYTE:
1318 return "(s8)";
1319 default:
1320 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1321 return "(_invalid_)";
1322 }
1323}
1324
1325static const char *lmod_cast(struct parsed_op *po,
1326 enum opr_lenmod lmod, int is_signed)
1327{
1328 return is_signed ?
1329 lmod_cast_s(po, lmod) :
1330 lmod_cast_u(po, lmod);
1331}
1332
1333static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
1334{
1335 switch (lmod) {
1336 case OPLM_QWORD:
1337 return 8;
1338 case OPLM_DWORD:
1339 return 4;
1340 case OPLM_WORD:
1341 return 2;
1342 case OPLM_BYTE:
1343 return 1;
1344 default:
1345 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1346 return 0;
1347 }
1348}
1349
1350static const char *opr_name(struct parsed_op *po, int opr_num)
1351{
1352 if (opr_num >= po->operand_cnt)
1353 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1354 return po->operand[opr_num].name;
1355}
1356
1357static unsigned int opr_const(struct parsed_op *po, int opr_num)
1358{
1359 if (opr_num >= po->operand_cnt)
1360 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1361 if (po->operand[opr_num].type != OPT_CONST)
1362 ferr(po, "opr %d: const expected\n", opr_num);
1363 return po->operand[opr_num].val;
1364}
1365
1366static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr)
1367{
1368 if ((unsigned int)popr->reg >= ARRAY_SIZE(regs_r32))
1369 ferr(po, "invalid reg: %d\n", popr->reg);
1370 return regs_r32[popr->reg];
1371}
1372
1373// cast1 is the "final" cast
1374static const char *simplify_cast(const char *cast1, const char *cast2)
1375{
1376 static char buf[256];
1377
1378 if (cast1[0] == 0)
1379 return cast2;
1380 if (cast2[0] == 0)
1381 return cast1;
1382 if (IS(cast1, cast2))
1383 return cast1;
1384 if (IS(cast1, "(s8)") && IS(cast2, "(u8)"))
1385 return cast1;
1386 if (IS(cast1, "(s16)") && IS(cast2, "(u16)"))
1387 return cast1;
1388 if (IS(cast1, "(u8)") && IS_START(cast2, "*(u8 *)"))
1389 return cast2;
1390 if (IS(cast1, "(u16)") && IS_START(cast2, "*(u16 *)"))
1391 return cast2;
1392 if (strchr(cast1, '*') && IS_START(cast2, "(u32)"))
1393 return cast1;
1394
1395 snprintf(buf, sizeof(buf), "%s%s", cast1, cast2);
1396 return buf;
1397}
1398
1399static const char *simplify_cast_num(const char *cast, unsigned int val)
1400{
1401 if (IS(cast, "(u8)") && val < 0x100)
1402 return "";
1403 if (IS(cast, "(s8)") && val < 0x80)
1404 return "";
1405 if (IS(cast, "(u16)") && val < 0x10000)
1406 return "";
1407 if (IS(cast, "(s16)") && val < 0x8000)
1408 return "";
1409 if (IS(cast, "(s32)") && val < 0x80000000)
1410 return "";
1411
1412 return cast;
1413}
1414
1415static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
1416 int *extra_offs)
1417{
1418 const char *p;
1419 char *endp;
1420 int namelen;
1421 int i;
1422
1423 *extra_offs = 0;
1424 namelen = strlen(name);
1425
1426 p = strchr(name, '+');
1427 if (p != NULL) {
1428 namelen = p - name;
1429 if (namelen <= 0)
1430 ferr(po, "equ parse failed for '%s'\n", name);
1431
1432 if (IS_START(p, "0x"))
1433 p += 2;
1434 *extra_offs = strtol(p, &endp, 16);
1435 if (*endp != 0)
1436 ferr(po, "equ parse failed for '%s'\n", name);
1437 }
1438
1439 for (i = 0; i < g_eqcnt; i++)
1440 if (strncmp(g_eqs[i].name, name, namelen) == 0
1441 && g_eqs[i].name[namelen] == 0)
1442 break;
1443 if (i >= g_eqcnt) {
1444 if (po != NULL)
1445 ferr(po, "unresolved equ name: '%s'\n", name);
1446 return NULL;
1447 }
1448
1449 return &g_eqs[i];
1450}
1451
1452static int is_stack_access(struct parsed_op *po,
1453 const struct parsed_opr *popr)
1454{
1455 return (parse_stack_el(popr->name, NULL, 0)
1456 || (g_bp_frame && !(po->flags & OPF_EBP_S)
1457 && IS_START(popr->name, "ebp")));
1458}
1459
1460static void parse_stack_access(struct parsed_op *po,
1461 const char *name, char *ofs_reg, int *offset_out,
1462 int *stack_ra_out, const char **bp_arg_out, int is_lea)
1463{
1464 const char *bp_arg = "";
1465 const char *p = NULL;
1466 struct parsed_equ *eq;
1467 char *endp = NULL;
1468 int stack_ra = 0;
1469 int offset = 0;
1470
1471 ofs_reg[0] = 0;
1472
1473 if (IS_START(name, "ebp-")
1474 || (IS_START(name, "ebp+") && '0' <= name[4] && name[4] <= '9'))
1475 {
1476 p = name + 4;
1477 if (IS_START(p, "0x"))
1478 p += 2;
1479 offset = strtoul(p, &endp, 16);
1480 if (name[3] == '-')
1481 offset = -offset;
1482 if (*endp != 0)
1483 ferr(po, "ebp- parse of '%s' failed\n", name);
1484 }
1485 else {
1486 bp_arg = parse_stack_el(name, ofs_reg, 0);
1487 snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
1488 eq = equ_find(po, bp_arg, &offset);
1489 if (eq == NULL)
1490 ferr(po, "detected but missing eq\n");
1491 offset += eq->offset;
1492 }
1493
1494 if (!strncmp(name, "ebp", 3))
1495 stack_ra = 4;
1496
1497 // yes it sometimes LEAs ra for compares..
1498 if (!is_lea && ofs_reg[0] == 0
1499 && stack_ra <= offset && offset < stack_ra + 4)
1500 {
1501 ferr(po, "reference to ra? %d %d\n", offset, stack_ra);
1502 }
1503
1504 *offset_out = offset;
1505 *stack_ra_out = stack_ra;
1506 if (bp_arg_out)
1507 *bp_arg_out = bp_arg;
1508}
1509
1510static int stack_frame_access(struct parsed_op *po,
1511 struct parsed_opr *popr, char *buf, size_t buf_size,
1512 const char *name, const char *cast, int is_src, int is_lea)
1513{
1514 enum opr_lenmod tmp_lmod = OPLM_UNSPEC;
1515 const char *prefix = "";
1516 const char *bp_arg = NULL;
1517 char ofs_reg[16] = { 0, };
1518 int i, arg_i, arg_s;
1519 int unaligned = 0;
1520 int stack_ra = 0;
1521 int offset = 0;
1522 int retval = -1;
1523 int sf_ofs;
1524 int lim;
1525
1526 if (po->flags & OPF_EBP_S)
1527 ferr(po, "stack_frame_access while ebp is scratch\n");
1528
1529 parse_stack_access(po, name, ofs_reg, &offset,
1530 &stack_ra, &bp_arg, is_lea);
1531
1532 if (offset > stack_ra)
1533 {
1534 arg_i = (offset - stack_ra - 4) / 4;
1535 if (arg_i < 0 || arg_i >= g_func_pp->argc_stack)
1536 {
1537 if (g_func_pp->is_vararg
1538 && arg_i == g_func_pp->argc_stack && is_lea)
1539 {
1540 // should be va_list
1541 if (cast[0] == 0)
1542 cast = "(u32)";
1543 snprintf(buf, buf_size, "%sap", cast);
1544 return -1;
1545 }
1546 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1547 offset, bp_arg, arg_i);
1548 }
1549 if (ofs_reg[0] != 0)
1550 ferr(po, "offset reg on arg access?\n");
1551
1552 for (i = arg_s = 0; i < g_func_pp->argc; i++) {
1553 if (g_func_pp->arg[i].reg != NULL)
1554 continue;
1555 if (arg_s == arg_i)
1556 break;
1557 arg_s++;
1558 }
1559 if (i == g_func_pp->argc)
1560 ferr(po, "arg %d not in prototype?\n", arg_i);
1561
1562 popr->is_ptr = g_func_pp->arg[i].type.is_ptr;
1563 retval = i;
1564
1565 switch (popr->lmod)
1566 {
1567 case OPLM_BYTE:
1568 if (is_lea)
1569 ferr(po, "lea/byte to arg?\n");
1570 if (is_src && (offset & 3) == 0)
1571 snprintf(buf, buf_size, "%sa%d",
1572 simplify_cast(cast, "(u8)"), i + 1);
1573 else
1574 snprintf(buf, buf_size, "%sBYTE%d(a%d)",
1575 cast, offset & 3, i + 1);
1576 break;
1577
1578 case OPLM_WORD:
1579 if (is_lea)
1580 ferr(po, "lea/word to arg?\n");
1581 if (offset & 1) {
1582 unaligned = 1;
1583 if (!is_src) {
1584 if (offset & 2)
1585 ferr(po, "problematic arg store\n");
1586 snprintf(buf, buf_size, "%s((char *)&a%d + 1)",
1587 simplify_cast(cast, "*(u16 *)"), i + 1);
1588 }
1589 else
1590 ferr(po, "unaligned arg word load\n");
1591 }
1592 else if (is_src && (offset & 2) == 0)
1593 snprintf(buf, buf_size, "%sa%d",
1594 simplify_cast(cast, "(u16)"), i + 1);
1595 else
1596 snprintf(buf, buf_size, "%s%sWORD(a%d)",
1597 cast, (offset & 2) ? "HI" : "LO", i + 1);
1598 break;
1599
1600 case OPLM_DWORD:
1601 if (cast[0])
1602 prefix = cast;
1603 else if (is_src)
1604 prefix = "(u32)";
1605
1606 if (offset & 3) {
1607 unaligned = 1;
1608 if (is_lea)
1609 snprintf(buf, buf_size, "(u32)&a%d + %d",
1610 i + 1, offset & 3);
1611 else if (!is_src)
1612 ferr(po, "unaligned arg store\n");
1613 else {
1614 // mov edx, [ebp+arg_4+2]; movsx ecx, dx
1615 snprintf(buf, buf_size, "%s(a%d >> %d)",
1616 prefix, i + 1, (offset & 3) * 8);
1617 }
1618 }
1619 else {
1620 snprintf(buf, buf_size, "%s%sa%d",
1621 prefix, is_lea ? "&" : "", i + 1);
1622 }
1623 break;
1624
1625 default:
1626 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
1627 }
1628
1629 if (unaligned)
1630 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1631
1632 // common problem
1633 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type);
1634 if (tmp_lmod != OPLM_DWORD
1635 && (unaligned || (!is_src && lmod_bytes(po, tmp_lmod)
1636 < lmod_bytes(po, popr->lmod) + (offset & 3))))
1637 {
1638 ferr(po, "bp_arg arg%d/w offset %d and type '%s' is too small\n",
1639 i + 1, offset, g_func_pp->arg[i].type.name);
1640 }
1641 // can't check this because msvc likes to reuse
1642 // arg space for scratch..
1643 //if (popr->is_ptr && popr->lmod != OPLM_DWORD)
1644 // ferr(po, "bp_arg arg%d: non-dword ptr access\n", i + 1);
1645 }
1646 else
1647 {
1648 if (g_stack_fsz == 0)
1649 ferr(po, "stack var access without stackframe\n");
1650 g_stack_frame_used = 1;
1651
1652 sf_ofs = g_stack_fsz + offset;
1653 lim = (ofs_reg[0] != 0) ? -4 : 0;
1654 if (offset > 0 || sf_ofs < lim)
1655 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
1656
1657 if (is_lea)
1658 prefix = "(u32)&";
1659 else
1660 prefix = cast;
1661
1662 switch (popr->lmod)
1663 {
1664 case OPLM_BYTE:
1665 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1666 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1667 break;
1668
1669 case OPLM_WORD:
1670 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1671 // known unaligned or possibly unaligned
1672 strcat(g_comment, " unaligned");
1673 if (prefix[0] == 0)
1674 prefix = "*(u16 *)&";
1675 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1676 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1677 break;
1678 }
1679 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
1680 break;
1681
1682 case OPLM_DWORD:
1683 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1684 // known unaligned or possibly unaligned
1685 strcat(g_comment, " unaligned");
1686 if (prefix[0] == 0)
1687 prefix = "*(u32 *)&";
1688 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1689 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1690 break;
1691 }
1692 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
1693 break;
1694
1695 default:
1696 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
1697 }
1698 }
1699
1700 return retval;
1701}
1702
1703static void check_func_pp(struct parsed_op *po,
1704 const struct parsed_proto *pp, const char *pfx)
1705{
1706 enum opr_lenmod tmp_lmod;
1707 char buf[256];
1708 int ret, i;
1709
1710 if (pp->argc_reg != 0) {
1711 if (/*!g_allow_regfunc &&*/ !pp->is_fastcall) {
1712 pp_print(buf, sizeof(buf), pp);
1713 ferr(po, "%s: unexpected reg arg in icall: %s\n", pfx, buf);
1714 }
1715 if (pp->argc_stack > 0 && pp->argc_reg != 2)
1716 ferr(po, "%s: %d reg arg(s) with %d stack arg(s)\n",
1717 pfx, pp->argc_reg, pp->argc_stack);
1718 }
1719
1720 // fptrs must use 32bit args, callsite might have no information and
1721 // lack a cast to smaller types, which results in incorrectly masked
1722 // args passed (callee may assume masked args, it does on ARM)
1723 if (!pp->is_osinc) {
1724 for (i = 0; i < pp->argc; i++) {
1725 ret = guess_lmod_from_c_type(&tmp_lmod, &pp->arg[i].type);
1726 if (ret && tmp_lmod != OPLM_DWORD)
1727 ferr(po, "reference to %s with arg%d '%s'\n", pp->name,
1728 i + 1, pp->arg[i].type.name);
1729 }
1730 }
1731}
1732
1733static const char *check_label_read_ref(struct parsed_op *po,
1734 const char *name)
1735{
1736 const struct parsed_proto *pp;
1737
1738 pp = proto_parse(g_fhdr, name, 0);
1739 if (pp == NULL)
1740 ferr(po, "proto_parse failed for ref '%s'\n", name);
1741
1742 if (pp->is_func)
1743 check_func_pp(po, pp, "ref");
1744
1745 return pp->name;
1746}
1747
1748static char *out_src_opr(char *buf, size_t buf_size,
1749 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
1750 int is_lea)
1751{
1752 char tmp1[256], tmp2[256];
1753 char expr[256];
1754 const char *name;
1755 char *p;
1756 int ret;
1757
1758 if (cast == NULL)
1759 cast = "";
1760
1761 switch (popr->type) {
1762 case OPT_REG:
1763 if (is_lea)
1764 ferr(po, "lea from reg?\n");
1765
1766 switch (popr->lmod) {
1767 case OPLM_QWORD:
1768 snprintf(buf, buf_size, "%s%s.q", cast, opr_reg_p(po, popr));
1769 break;
1770 case OPLM_DWORD:
1771 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
1772 break;
1773 case OPLM_WORD:
1774 snprintf(buf, buf_size, "%s%s",
1775 simplify_cast(cast, "(u16)"), opr_reg_p(po, popr));
1776 break;
1777 case OPLM_BYTE:
1778 if (popr->name[1] == 'h') // XXX..
1779 snprintf(buf, buf_size, "%s(%s >> 8)",
1780 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
1781 else
1782 snprintf(buf, buf_size, "%s%s",
1783 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
1784 break;
1785 default:
1786 ferr(po, "invalid src lmod: %d\n", popr->lmod);
1787 }
1788 break;
1789
1790 case OPT_REGMEM:
1791 if (is_stack_access(po, popr)) {
1792 stack_frame_access(po, popr, buf, buf_size,
1793 popr->name, cast, 1, is_lea);
1794 break;
1795 }
1796
1797 strcpy(expr, popr->name);
1798 if (strchr(expr, '[')) {
1799 // special case: '[' can only be left for label[reg] form
1800 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
1801 if (ret != 2)
1802 ferr(po, "parse failure for '%s'\n", expr);
1803 if (tmp1[0] == '(') {
1804 // (off_4FFF50+3)[eax]
1805 p = strchr(tmp1 + 1, ')');
1806 if (p == NULL || p[1] != 0)
1807 ferr(po, "parse failure (2) for '%s'\n", expr);
1808 *p = 0;
1809 memmove(tmp1, tmp1 + 1, strlen(tmp1));
1810 }
1811 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
1812 }
1813
1814 // XXX: do we need more parsing?
1815 if (is_lea) {
1816 snprintf(buf, buf_size, "%s", expr);
1817 break;
1818 }
1819
1820 snprintf(buf, buf_size, "%s(%s)",
1821 simplify_cast(cast, lmod_cast_u_ptr(po, popr->lmod)), expr);
1822 break;
1823
1824 case OPT_LABEL:
1825 name = check_label_read_ref(po, popr->name);
1826 if (cast[0] == 0 && popr->is_ptr)
1827 cast = "(u32)";
1828
1829 if (is_lea)
1830 snprintf(buf, buf_size, "(u32)&%s", name);
1831 else if (popr->size_lt)
1832 snprintf(buf, buf_size, "%s%s%s%s", cast,
1833 lmod_cast_u_ptr(po, popr->lmod),
1834 popr->is_array ? "" : "&", name);
1835 else
1836 snprintf(buf, buf_size, "%s%s%s", cast, name,
1837 popr->is_array ? "[0]" : "");
1838 break;
1839
1840 case OPT_OFFSET:
1841 name = check_label_read_ref(po, popr->name);
1842 if (cast[0] == 0)
1843 cast = "(u32)";
1844 if (is_lea)
1845 ferr(po, "lea an offset?\n");
1846 snprintf(buf, buf_size, "%s&%s", cast, name);
1847 break;
1848
1849 case OPT_CONST:
1850 if (is_lea)
1851 ferr(po, "lea from const?\n");
1852
1853 printf_number(tmp1, sizeof(tmp1), popr->val);
1854 if (popr->val == 0 && strchr(cast, '*'))
1855 snprintf(buf, buf_size, "NULL");
1856 else
1857 snprintf(buf, buf_size, "%s%s",
1858 simplify_cast_num(cast, popr->val), tmp1);
1859 break;
1860
1861 default:
1862 ferr(po, "invalid src type: %d\n", popr->type);
1863 }
1864
1865 return buf;
1866}
1867
1868// note: may set is_ptr (we find that out late for ebp frame..)
1869static char *out_dst_opr(char *buf, size_t buf_size,
1870 struct parsed_op *po, struct parsed_opr *popr)
1871{
1872 switch (popr->type) {
1873 case OPT_REG:
1874 switch (popr->lmod) {
1875 case OPLM_QWORD:
1876 snprintf(buf, buf_size, "%s.q", opr_reg_p(po, popr));
1877 break;
1878 case OPLM_DWORD:
1879 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
1880 break;
1881 case OPLM_WORD:
1882 // ugh..
1883 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
1884 break;
1885 case OPLM_BYTE:
1886 // ugh..
1887 if (popr->name[1] == 'h') // XXX..
1888 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
1889 else
1890 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
1891 break;
1892 default:
1893 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
1894 }
1895 break;
1896
1897 case OPT_REGMEM:
1898 if (is_stack_access(po, popr)) {
1899 stack_frame_access(po, popr, buf, buf_size,
1900 popr->name, "", 0, 0);
1901 break;
1902 }
1903
1904 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1905
1906 case OPT_LABEL:
1907 if (popr->size_mismatch)
1908 snprintf(buf, buf_size, "%s%s%s",
1909 lmod_cast_u_ptr(po, popr->lmod),
1910 popr->is_array ? "" : "&", popr->name);
1911 else
1912 snprintf(buf, buf_size, "%s%s", popr->name,
1913 popr->is_array ? "[0]" : "");
1914 break;
1915
1916 default:
1917 ferr(po, "invalid dst type: %d\n", popr->type);
1918 }
1919
1920 return buf;
1921}
1922
1923static char *out_src_opr_u32(char *buf, size_t buf_size,
1924 struct parsed_op *po, struct parsed_opr *popr)
1925{
1926 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1927}
1928
1929static void out_test_for_cc(char *buf, size_t buf_size,
1930 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
1931 enum opr_lenmod lmod, const char *expr)
1932{
1933 const char *cast, *scast;
1934
1935 cast = lmod_cast_u(po, lmod);
1936 scast = lmod_cast_s(po, lmod);
1937
1938 switch (pfo) {
1939 case PFO_Z:
1940 case PFO_BE: // CF=1||ZF=1; CF=0
1941 snprintf(buf, buf_size, "(%s%s %s 0)",
1942 cast, expr, is_inv ? "!=" : "==");
1943 break;
1944
1945 case PFO_S:
1946 case PFO_L: // SF!=OF; OF=0
1947 snprintf(buf, buf_size, "(%s%s %s 0)",
1948 scast, expr, is_inv ? ">=" : "<");
1949 break;
1950
1951 case PFO_LE: // ZF=1||SF!=OF; OF=0
1952 snprintf(buf, buf_size, "(%s%s %s 0)",
1953 scast, expr, is_inv ? ">" : "<=");
1954 break;
1955
1956 default:
1957 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1958 }
1959}
1960
1961static void out_cmp_for_cc(char *buf, size_t buf_size,
1962 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
1963{
1964 const char *cast, *scast, *cast_use;
1965 char buf1[256], buf2[256];
1966 enum opr_lenmod lmod;
1967
1968 if (po->op != OP_DEC && po->operand[0].lmod != po->operand[1].lmod)
1969 ferr(po, "%s: lmod mismatch: %d %d\n", __func__,
1970 po->operand[0].lmod, po->operand[1].lmod);
1971 lmod = po->operand[0].lmod;
1972
1973 cast = lmod_cast_u(po, lmod);
1974 scast = lmod_cast_s(po, lmod);
1975
1976 switch (pfo) {
1977 case PFO_C:
1978 case PFO_Z:
1979 case PFO_BE: // !a
1980 cast_use = cast;
1981 break;
1982
1983 case PFO_S:
1984 case PFO_L: // !ge
1985 case PFO_LE:
1986 cast_use = scast;
1987 break;
1988
1989 default:
1990 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1991 }
1992
1993 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], cast_use, 0);
1994 if (po->op == OP_DEC)
1995 snprintf(buf2, sizeof(buf2), "1");
1996 else
1997 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], cast_use, 0);
1998
1999 switch (pfo) {
2000 case PFO_C:
2001 // note: must be unsigned compare
2002 snprintf(buf, buf_size, "(%s %s %s)",
2003 buf1, is_inv ? ">=" : "<", buf2);
2004 break;
2005
2006 case PFO_Z:
2007 snprintf(buf, buf_size, "(%s %s %s)",
2008 buf1, is_inv ? "!=" : "==", buf2);
2009 break;
2010
2011 case PFO_BE: // !a
2012 // note: must be unsigned compare
2013 snprintf(buf, buf_size, "(%s %s %s)",
2014 buf1, is_inv ? ">" : "<=", buf2);
2015
2016 // annoying case
2017 if (is_inv && lmod == OPLM_BYTE
2018 && po->operand[1].type == OPT_CONST
2019 && po->operand[1].val == 0xff)
2020 {
2021 snprintf(g_comment, sizeof(g_comment), "if %s", buf);
2022 snprintf(buf, buf_size, "(0)");
2023 }
2024 break;
2025
2026 // note: must be signed compare
2027 case PFO_S:
2028 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
2029 scast, buf1, buf2, is_inv ? ">=" : "<");
2030 break;
2031
2032 case PFO_L: // !ge
2033 snprintf(buf, buf_size, "(%s %s %s)",
2034 buf1, is_inv ? ">=" : "<", buf2);
2035 break;
2036
2037 case PFO_LE: // !g
2038 snprintf(buf, buf_size, "(%s %s %s)",
2039 buf1, is_inv ? ">" : "<=", buf2);
2040 break;
2041
2042 default:
2043 break;
2044 }
2045}
2046
2047static void out_cmp_test(char *buf, size_t buf_size,
2048 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
2049{
2050 char buf1[256], buf2[256], buf3[256];
2051
2052 if (po->op == OP_TEST) {
2053 if (IS(opr_name(po, 0), opr_name(po, 1))) {
2054 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
2055 }
2056 else {
2057 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
2058 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
2059 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
2060 }
2061 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
2062 po->operand[0].lmod, buf3);
2063 }
2064 else if (po->op == OP_CMP) {
2065 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv);
2066 }
2067 else
2068 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
2069}
2070
2071static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
2072 struct parsed_opr *popr2)
2073{
2074 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
2075 ferr(po, "missing lmod for both operands\n");
2076
2077 if (popr1->lmod == OPLM_UNSPEC)
2078 popr1->lmod = popr2->lmod;
2079 else if (popr2->lmod == OPLM_UNSPEC)
2080 popr2->lmod = popr1->lmod;
2081 else if (popr1->lmod != popr2->lmod) {
2082 if (popr1->type_from_var) {
2083 popr1->size_mismatch = 1;
2084 if (popr1->lmod < popr2->lmod)
2085 popr1->size_lt = 1;
2086 popr1->lmod = popr2->lmod;
2087 }
2088 else if (popr2->type_from_var) {
2089 popr2->size_mismatch = 1;
2090 if (popr2->lmod < popr1->lmod)
2091 popr2->size_lt = 1;
2092 popr2->lmod = popr1->lmod;
2093 }
2094 else
2095 ferr(po, "conflicting lmods: %d vs %d\n",
2096 popr1->lmod, popr2->lmod);
2097 }
2098}
2099
2100static const char *op_to_c(struct parsed_op *po)
2101{
2102 switch (po->op)
2103 {
2104 case OP_ADD:
2105 case OP_ADC:
2106 return "+";
2107 case OP_SUB:
2108 case OP_SBB:
2109 return "-";
2110 case OP_AND:
2111 return "&";
2112 case OP_OR:
2113 return "|";
2114 case OP_XOR:
2115 return "^";
2116 case OP_SHL:
2117 return "<<";
2118 case OP_SHR:
2119 return ">>";
2120 case OP_MUL:
2121 case OP_IMUL:
2122 return "*";
2123 default:
2124 ferr(po, "op_to_c was supplied with %d\n", po->op);
2125 }
2126}
2127
2128static void op_set_clear_flag(struct parsed_op *po,
2129 enum op_flags flag_set, enum op_flags flag_clear)
2130{
2131 po->flags |= flag_set;
2132 po->flags &= ~flag_clear;
2133}
2134
2135// last op in stream - unconditional branch or ret
2136#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
2137 || ((ops[_i].flags & (OPF_JMP|OPF_CJMP|OPF_RMD)) == OPF_JMP \
2138 && ops[_i].op != OP_CALL))
2139
2140static int scan_for_pop(int i, int opcnt, const char *reg,
2141 int magic, int depth, int *maxdepth, int do_flags)
2142{
2143 const struct parsed_proto *pp;
2144 struct parsed_op *po;
2145 int ret = 0;
2146 int j;
2147
2148 for (; i < opcnt; i++) {
2149 po = &ops[i];
2150 if (po->cc_scratch == magic)
2151 break; // already checked
2152 po->cc_scratch = magic;
2153
2154 if (po->flags & OPF_TAIL) {
2155 if (po->op == OP_CALL) {
2156 pp = proto_parse(g_fhdr, po->operand[0].name, g_quiet_pp);
2157 if (pp != NULL && pp->is_noreturn)
2158 // no stack cleanup for noreturn
2159 return ret;
2160 }
2161 return -1; // deadend
2162 }
2163
2164 if ((po->flags & (OPF_RMD|OPF_DONE))
2165 || (po->op == OP_PUSH && po->p_argnum != 0)) // arg push
2166 continue;
2167
2168 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2169 if (po->btj != NULL) {
2170 // jumptable
2171 for (j = 0; j < po->btj->count; j++) {
2172 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic,
2173 depth, maxdepth, do_flags);
2174 if (ret < 0)
2175 return ret; // dead end
2176 }
2177 return ret;
2178 }
2179
2180 if (po->bt_i < 0) {
2181 ferr(po, "dead branch\n");
2182 return -1;
2183 }
2184
2185 if (po->flags & OPF_CJMP) {
2186 ret |= scan_for_pop(po->bt_i, opcnt, reg, magic,
2187 depth, maxdepth, do_flags);
2188 if (ret < 0)
2189 return ret; // dead end
2190 }
2191 else {
2192 i = po->bt_i - 1;
2193 }
2194 continue;
2195 }
2196
2197 if ((po->op == OP_POP || po->op == OP_PUSH)
2198 && po->operand[0].type == OPT_REG
2199 && IS(po->operand[0].name, reg))
2200 {
2201 if (po->op == OP_PUSH && !(po->flags & OPF_FARGNR)) {
2202 depth++;
2203 if (depth > *maxdepth)
2204 *maxdepth = depth;
2205 if (do_flags)
2206 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2207 }
2208 else if (po->op == OP_POP) {
2209 if (depth == 0) {
2210 if (do_flags)
2211 op_set_clear_flag(po, OPF_RMD, OPF_RSAVE);
2212 return 1;
2213 }
2214 else {
2215 depth--;
2216 if (depth < 0) // should not happen
2217 ferr(po, "fail with depth\n");
2218 if (do_flags)
2219 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2220 }
2221 }
2222 }
2223 }
2224
2225 return ret;
2226}
2227
2228// scan for pop starting from 'ret' op (all paths)
2229static int scan_for_pop_ret(int i, int opcnt, const char *reg,
2230 int flag_set)
2231{
2232 int found = 0;
2233 int j;
2234
2235 for (; i < opcnt; i++) {
2236 if (!(ops[i].flags & OPF_TAIL))
2237 continue;
2238
2239 for (j = i - 1; j >= 0; j--) {
2240 if (ops[j].flags & (OPF_RMD|OPF_DONE))
2241 continue;
2242 if (ops[j].flags & OPF_JMP)
2243 return -1;
2244
2245 if (ops[j].op == OP_POP && ops[j].datap == NULL
2246 && ops[j].operand[0].type == OPT_REG
2247 && IS(ops[j].operand[0].name, reg))
2248 {
2249 found = 1;
2250 ops[j].flags |= flag_set;
2251 break;
2252 }
2253
2254 if (g_labels[j] != NULL)
2255 return -1;
2256 }
2257 }
2258
2259 return found ? 0 : -1;
2260}
2261
2262static void scan_for_pop_const(int i, int opcnt)
2263{
2264 int j;
2265
2266 for (j = i + 1; j < opcnt; j++) {
2267 if ((ops[j].flags & (OPF_JMP|OPF_TAIL|OPF_RSAVE))
2268 || ops[j].op == OP_PUSH || g_labels[i] != NULL)
2269 {
2270 break;
2271 }
2272
2273 if (ops[j].op == OP_POP && !(ops[j].flags & (OPF_RMD|OPF_DONE)))
2274 {
2275 ops[i].flags |= OPF_RMD | OPF_DONE;
2276 ops[j].flags |= OPF_DONE;
2277 ops[j].datap = &ops[i];
2278 break;
2279 }
2280 }
2281}
2282
2283static void scan_propagate_df(int i, int opcnt)
2284{
2285 struct parsed_op *po = &ops[i];
2286 int j;
2287
2288 for (; i < opcnt; i++) {
2289 po = &ops[i];
2290 if (po->flags & OPF_DF)
2291 return; // already resolved
2292 po->flags |= OPF_DF;
2293
2294 if (po->op == OP_CALL)
2295 ferr(po, "call with DF set?\n");
2296
2297 if (po->flags & OPF_JMP) {
2298 if (po->btj != NULL) {
2299 // jumptable
2300 for (j = 0; j < po->btj->count; j++)
2301 scan_propagate_df(po->btj->d[j].bt_i, opcnt);
2302 return;
2303 }
2304
2305 if (po->bt_i < 0) {
2306 ferr(po, "dead branch\n");
2307 return;
2308 }
2309
2310 if (po->flags & OPF_CJMP)
2311 scan_propagate_df(po->bt_i, opcnt);
2312 else
2313 i = po->bt_i - 1;
2314 continue;
2315 }
2316
2317 if (po->flags & OPF_TAIL)
2318 break;
2319
2320 if (po->op == OP_CLD) {
2321 po->flags |= OPF_RMD | OPF_DONE;
2322 return;
2323 }
2324 }
2325
2326 ferr(po, "missing DF clear?\n");
2327}
2328
2329// is operand 'opr' modified by parsed_op 'po'?
2330static int is_opr_modified(const struct parsed_opr *opr,
2331 const struct parsed_op *po)
2332{
2333 int mask;
2334
2335 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2336 return 0;
2337
2338 if (opr->type == OPT_REG) {
2339 if (po->op == OP_CALL) {
2340 mask = (1 << xAX) | (1 << xCX) | (1 << xDX);
2341 if ((1 << opr->reg) & mask)
2342 return 1;
2343 else
2344 return 0;
2345 }
2346
2347 if (po->operand[0].type == OPT_REG) {
2348 if (po->regmask_dst & (1 << opr->reg))
2349 return 1;
2350 else
2351 return 0;
2352 }
2353 }
2354
2355 return IS(po->operand[0].name, opr->name);
2356}
2357
2358// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
2359static int is_any_opr_modified(const struct parsed_op *po_test,
2360 const struct parsed_op *po, int c_mode)
2361{
2362 int mask;
2363 int i;
2364
2365 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2366 return 0;
2367
2368 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2369 return 0;
2370
2371 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
2372 return 1;
2373
2374 // in reality, it can wreck any register, but in decompiled C
2375 // version it can only overwrite eax or edx:eax
2376 mask = (1 << xAX) | (1 << xDX);
2377 if (!c_mode)
2378 mask |= 1 << xCX;
2379
2380 if (po->op == OP_CALL
2381 && ((po_test->regmask_src | po_test->regmask_dst) & mask))
2382 return 1;
2383
2384 for (i = 0; i < po_test->operand_cnt; i++)
2385 if (IS(po_test->operand[i].name, po->operand[0].name))
2386 return 1;
2387
2388 return 0;
2389}
2390
2391// scan for any po_test operand modification in range given
2392static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt,
2393 int c_mode)
2394{
2395 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2396 return -1;
2397
2398 for (; i < opcnt; i++) {
2399 if (is_any_opr_modified(po_test, &ops[i], c_mode))
2400 return i;
2401 }
2402
2403 return -1;
2404}
2405
2406// scan for po_test operand[0] modification in range given
2407static int scan_for_mod_opr0(struct parsed_op *po_test,
2408 int i, int opcnt)
2409{
2410 for (; i < opcnt; i++) {
2411 if (is_opr_modified(&po_test->operand[0], &ops[i]))
2412 return i;
2413 }
2414
2415 return -1;
2416}
2417
2418#define check_i(po, i) \
2419 if ((i) < 0) \
2420 ferr(po, "bad " #i ": %d\n", i)
2421
2422static int scan_for_flag_set(int i, int magic, int *branched,
2423 int *setters, int *setter_cnt)
2424{
2425 struct label_ref *lr;
2426 int ret;
2427
2428 while (i >= 0) {
2429 if (ops[i].cc_scratch == magic) {
2430 ferr(&ops[i], "%s looped\n", __func__);
2431 return -1;
2432 }
2433 ops[i].cc_scratch = magic;
2434
2435 if (g_labels[i] != NULL) {
2436 *branched = 1;
2437
2438 lr = &g_label_refs[i];
2439 for (; lr->next; lr = lr->next) {
2440 check_i(&ops[i], lr->i);
2441 ret = scan_for_flag_set(lr->i, magic,
2442 branched, setters, setter_cnt);
2443 if (ret < 0)
2444 return ret;
2445 }
2446
2447 check_i(&ops[i], lr->i);
2448 if (i > 0 && LAST_OP(i - 1)) {
2449 i = lr->i;
2450 continue;
2451 }
2452 ret = scan_for_flag_set(lr->i, magic,
2453 branched, setters, setter_cnt);
2454 if (ret < 0)
2455 return ret;
2456 }
2457 i--;
2458
2459 if (ops[i].flags & OPF_FLAGS) {
2460 setters[*setter_cnt] = i;
2461 (*setter_cnt)++;
2462 return 0;
2463 }
2464
2465 if ((ops[i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP)
2466 return -1;
2467 }
2468
2469 return -1;
2470}
2471
2472// scan back for cdq, if anything modifies edx, fail
2473static int scan_for_cdq_edx(int i)
2474{
2475 while (i >= 0) {
2476 if (g_labels[i] != NULL) {
2477 if (g_label_refs[i].next != NULL)
2478 return -1;
2479 if (i > 0 && LAST_OP(i - 1)) {
2480 i = g_label_refs[i].i;
2481 continue;
2482 }
2483 return -1;
2484 }
2485 i--;
2486
2487 if (ops[i].op == OP_CDQ)
2488 return i;
2489
2490 if (ops[i].regmask_dst & (1 << xDX))
2491 return -1;
2492 }
2493
2494 return -1;
2495}
2496
2497static int scan_for_reg_clear(int i, int reg)
2498{
2499 while (i >= 0) {
2500 if (g_labels[i] != NULL) {
2501 if (g_label_refs[i].next != NULL)
2502 return -1;
2503 if (i > 0 && LAST_OP(i - 1)) {
2504 i = g_label_refs[i].i;
2505 continue;
2506 }
2507 return -1;
2508 }
2509 i--;
2510
2511 if (ops[i].op == OP_XOR
2512 && ops[i].operand[0].lmod == OPLM_DWORD
2513 && ops[i].operand[0].reg == ops[i].operand[1].reg
2514 && ops[i].operand[0].reg == reg)
2515 return i;
2516
2517 if (ops[i].regmask_dst & (1 << reg))
2518 return -1;
2519 }
2520
2521 return -1;
2522}
2523
2524// scan for positive, constant esp adjust
2525static int scan_for_esp_adjust(int i, int opcnt,
2526 int adj_expect, int *adj, int *multipath)
2527{
2528 struct parsed_op *po;
2529 int first_pop = -1;
2530
2531 *adj = *multipath = 0;
2532
2533 for (; i < opcnt && *adj < adj_expect; i++) {
2534 if (g_labels[i] != NULL)
2535 *multipath = 1;
2536
2537 po = &ops[i];
2538 if (po->flags & OPF_DONE)
2539 continue;
2540
2541 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
2542 if (po->operand[1].type != OPT_CONST)
2543 ferr(&ops[i], "non-const esp adjust?\n");
2544 *adj += po->operand[1].val;
2545 if (*adj & 3)
2546 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
2547 return i;
2548 }
2549 else if (po->op == OP_PUSH) {
2550 //if (first_pop == -1)
2551 // first_pop = -2; // none
2552 *adj -= lmod_bytes(po, po->operand[0].lmod);
2553 }
2554 else if (po->op == OP_POP) {
2555 if (!(po->flags & OPF_DONE)) {
2556 // seems like msvc only uses 'pop ecx' for stack realignment..
2557 if (po->operand[0].type != OPT_REG || po->operand[0].reg != xCX)
2558 break;
2559 if (first_pop == -1 && *adj >= 0)
2560 first_pop = i;
2561 }
2562 *adj += lmod_bytes(po, po->operand[0].lmod);
2563 }
2564 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
2565 if (po->op == OP_JMP && po->btj == NULL) {
2566 if (po->bt_i <= i)
2567 break;
2568 i = po->bt_i - 1;
2569 continue;
2570 }
2571 if (po->op != OP_CALL)
2572 break;
2573 if (po->operand[0].type != OPT_LABEL)
2574 break;
2575 if (po->pp != NULL && po->pp->is_stdcall)
2576 break;
2577 // assume it's another cdecl call
2578 }
2579 }
2580
2581 if (first_pop >= 0) {
2582 // probably 'pop ecx' was used..
2583 return first_pop;
2584 }
2585
2586 return -1;
2587}
2588
2589static void scan_fwd_set_flags(int i, int opcnt, int magic, int flags)
2590{
2591 struct parsed_op *po;
2592 int j;
2593
2594 if (i < 0)
2595 ferr(ops, "%s: followed bad branch?\n", __func__);
2596
2597 for (; i < opcnt; i++) {
2598 po = &ops[i];
2599 if (po->cc_scratch == magic)
2600 return;
2601 po->cc_scratch = magic;
2602 po->flags |= flags;
2603
2604 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2605 if (po->btj != NULL) {
2606 // jumptable
2607 for (j = 0; j < po->btj->count; j++)
2608 scan_fwd_set_flags(po->btj->d[j].bt_i, opcnt, magic, flags);
2609 return;
2610 }
2611
2612 scan_fwd_set_flags(po->bt_i, opcnt, magic, flags);
2613 if (!(po->flags & OPF_CJMP))
2614 return;
2615 }
2616 if (po->flags & OPF_TAIL)
2617 return;
2618 }
2619}
2620
2621static const struct parsed_proto *try_recover_pp(
2622 struct parsed_op *po, const struct parsed_opr *opr, int *search_instead)
2623{
2624 const struct parsed_proto *pp = NULL;
2625 char buf[256];
2626 char *p;
2627
2628 // maybe an arg of g_func?
2629 if (opr->type == OPT_REGMEM && is_stack_access(po, opr))
2630 {
2631 char ofs_reg[16] = { 0, };
2632 int arg, arg_s, arg_i;
2633 int stack_ra = 0;
2634 int offset = 0;
2635
2636 if (g_header_mode)
2637 return NULL;
2638
2639 parse_stack_access(po, opr->name, ofs_reg,
2640 &offset, &stack_ra, NULL, 0);
2641 if (ofs_reg[0] != 0)
2642 ferr(po, "offset reg on arg access?\n");
2643 if (offset <= stack_ra) {
2644 // search who set the stack var instead
2645 if (search_instead != NULL)
2646 *search_instead = 1;
2647 return NULL;
2648 }
2649
2650 arg_i = (offset - stack_ra - 4) / 4;
2651 for (arg = arg_s = 0; arg < g_func_pp->argc; arg++) {
2652 if (g_func_pp->arg[arg].reg != NULL)
2653 continue;
2654 if (arg_s == arg_i)
2655 break;
2656 arg_s++;
2657 }
2658 if (arg == g_func_pp->argc)
2659 ferr(po, "stack arg %d not in prototype?\n", arg_i);
2660
2661 pp = g_func_pp->arg[arg].fptr;
2662 if (pp == NULL)
2663 ferr(po, "icall sa: arg%d is not a fptr?\n", arg + 1);
2664 check_func_pp(po, pp, "icall arg");
2665 }
2666 else if (opr->type == OPT_REGMEM && strchr(opr->name + 1, '[')) {
2667 // label[index]
2668 p = strchr(opr->name + 1, '[');
2669 memcpy(buf, opr->name, p - opr->name);
2670 buf[p - opr->name] = 0;
2671 pp = proto_parse(g_fhdr, buf, g_quiet_pp);
2672 }
2673 else if (opr->type == OPT_OFFSET || opr->type == OPT_LABEL) {
2674 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
2675 if (pp == NULL) {
2676 if (!g_header_mode)
2677 ferr(po, "proto_parse failed for icall to '%s'\n", opr->name);
2678 }
2679 else
2680 check_func_pp(po, pp, "reg-fptr ref");
2681 }
2682
2683 return pp;
2684}
2685
2686static void scan_for_call_type(int i, const struct parsed_opr *opr,
2687 int magic, const struct parsed_proto **pp_found, int *multi)
2688{
2689 const struct parsed_proto *pp = NULL;
2690 struct parsed_op *po;
2691 struct label_ref *lr;
2692
2693 ops[i].cc_scratch = magic;
2694
2695 while (1) {
2696 if (g_labels[i] != NULL) {
2697 lr = &g_label_refs[i];
2698 for (; lr != NULL; lr = lr->next) {
2699 check_i(&ops[i], lr->i);
2700 scan_for_call_type(lr->i, opr, magic, pp_found, multi);
2701 }
2702 if (i > 0 && LAST_OP(i - 1))
2703 return;
2704 }
2705
2706 i--;
2707 if (i < 0)
2708 break;
2709
2710 if (ops[i].cc_scratch == magic)
2711 return;
2712 ops[i].cc_scratch = magic;
2713
2714 if (!(ops[i].flags & OPF_DATA))
2715 continue;
2716 if (!is_opr_modified(opr, &ops[i]))
2717 continue;
2718 if (ops[i].op != OP_MOV && ops[i].op != OP_LEA) {
2719 // most probably trashed by some processing
2720 *pp_found = NULL;
2721 return;
2722 }
2723
2724 opr = &ops[i].operand[1];
2725 if (opr->type != OPT_REG)
2726 break;
2727 }
2728
2729 po = (i >= 0) ? &ops[i] : ops;
2730
2731 if (i < 0) {
2732 // reached the top - can only be an arg-reg
2733 if (opr->type != OPT_REG)
2734 return;
2735
2736 for (i = 0; i < g_func_pp->argc; i++) {
2737 if (g_func_pp->arg[i].reg == NULL)
2738 continue;
2739 if (IS(opr->name, g_func_pp->arg[i].reg))
2740 break;
2741 }
2742 if (i == g_func_pp->argc)
2743 return;
2744 pp = g_func_pp->arg[i].fptr;
2745 if (pp == NULL)
2746 ferr(po, "icall: arg%d (%s) is not a fptr?\n",
2747 i + 1, g_func_pp->arg[i].reg);
2748 check_func_pp(po, pp, "icall reg-arg");
2749 }
2750 else
2751 pp = try_recover_pp(po, opr, NULL);
2752
2753 if (*pp_found != NULL && pp != NULL && *pp_found != pp) {
2754 if (!IS((*pp_found)->ret_type.name, pp->ret_type.name)
2755 || (*pp_found)->is_stdcall != pp->is_stdcall
2756 || (*pp_found)->is_fptr != pp->is_fptr
2757 || (*pp_found)->argc != pp->argc
2758 || (*pp_found)->argc_reg != pp->argc_reg
2759 || (*pp_found)->argc_stack != pp->argc_stack)
2760 {
2761 ferr(po, "icall: parsed_proto mismatch\n");
2762 }
2763 *multi = 1;
2764 }
2765 if (pp != NULL)
2766 *pp_found = pp;
2767}
2768
2769// early check for tail call or branch back
2770static int is_like_tailjmp(int j)
2771{
2772 if (!(ops[j].flags & OPF_JMP))
2773 return 0;
2774
2775 if (ops[j].op == OP_JMP && !ops[j].operand[0].had_ds)
2776 // probably local branch back..
2777 return 1;
2778 if (ops[j].op == OP_CALL)
2779 // probably noreturn call..
2780 return 1;
2781
2782 return 0;
2783}
2784
2785static void scan_prologue_epilogue(int opcnt)
2786{
2787 int ecx_push = 0, esp_sub = 0;
2788 int found;
2789 int i, j, l;
2790
2791 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
2792 && ops[1].op == OP_MOV
2793 && IS(opr_name(&ops[1], 0), "ebp")
2794 && IS(opr_name(&ops[1], 1), "esp"))
2795 {
2796 g_bp_frame = 1;
2797 ops[0].flags |= OPF_RMD | OPF_DONE;
2798 ops[1].flags |= OPF_RMD | OPF_DONE;
2799 i = 2;
2800
2801 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
2802 g_stack_fsz = opr_const(&ops[2], 1);
2803 ops[2].flags |= OPF_RMD | OPF_DONE;
2804 i++;
2805 }
2806 else {
2807 // another way msvc builds stack frame..
2808 i = 2;
2809 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
2810 g_stack_fsz += 4;
2811 ops[i].flags |= OPF_RMD | OPF_DONE;
2812 ecx_push++;
2813 i++;
2814 }
2815 // and another way..
2816 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
2817 && ops[i].operand[1].type == OPT_CONST
2818 && ops[i + 1].op == OP_CALL
2819 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
2820 {
2821 g_stack_fsz += ops[i].operand[1].val;
2822 ops[i].flags |= OPF_RMD | OPF_DONE;
2823 i++;
2824 ops[i].flags |= OPF_RMD | OPF_DONE;
2825 i++;
2826 }
2827 }
2828
2829 found = 0;
2830 do {
2831 for (; i < opcnt; i++)
2832 if (ops[i].op == OP_RET)
2833 break;
2834 j = i - 1;
2835 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
2836 if (found && is_like_tailjmp(j))
2837 break;
2838 j--;
2839 }
2840
2841 if ((ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ebp"))
2842 || ops[j].op == OP_LEAVE)
2843 {
2844 ops[j].flags |= OPF_RMD | OPF_DONE;
2845 }
2846 else if (!(g_ida_func_attr & IDAFA_NORETURN))
2847 ferr(&ops[j], "'pop ebp' expected\n");
2848
2849 if (g_stack_fsz != 0) {
2850 if (ops[j - 1].op == OP_MOV
2851 && IS(opr_name(&ops[j - 1], 0), "esp")
2852 && IS(opr_name(&ops[j - 1], 1), "ebp"))
2853 {
2854 ops[j - 1].flags |= OPF_RMD | OPF_DONE;
2855 }
2856 else if (ops[j].op != OP_LEAVE
2857 && !(g_ida_func_attr & IDAFA_NORETURN))
2858 {
2859 ferr(&ops[j - 1], "esp restore expected\n");
2860 }
2861
2862 if (ecx_push && ops[j - 2].op == OP_POP
2863 && IS(opr_name(&ops[j - 2], 0), "ecx"))
2864 {
2865 ferr(&ops[j - 2], "unexpected ecx pop\n");
2866 }
2867 }
2868
2869 found = 1;
2870 i++;
2871 } while (i < opcnt);
2872
2873 return;
2874 }
2875
2876 // non-bp frame
2877 i = 0;
2878 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
2879 ops[i].flags |= OPF_RMD | OPF_DONE;
2880 g_stack_fsz += 4;
2881 ecx_push++;
2882 i++;
2883 }
2884
2885 for (; i < opcnt; i++) {
2886 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
2887 break;
2888 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
2889 && ops[i].operand[1].type == OPT_CONST)
2890 {
2891 g_stack_fsz = ops[i].operand[1].val;
2892 ops[i].flags |= OPF_RMD | OPF_DONE;
2893 esp_sub = 1;
2894 break;
2895 }
2896 }
2897
2898 if (ecx_push && !esp_sub) {
2899 // could actually be args for a call..
2900 for (; i < opcnt; i++)
2901 if (ops[i].op != OP_PUSH)
2902 break;
2903
2904 if (ops[i].op == OP_CALL && ops[i].operand[0].type == OPT_LABEL) {
2905 const struct parsed_proto *pp;
2906 pp = proto_parse(g_fhdr, opr_name(&ops[i], 0), 1);
2907 j = pp ? pp->argc_stack : 0;
2908 while (i > 0 && j > 0) {
2909 i--;
2910 if (ops[i].op == OP_PUSH) {
2911 ops[i].flags &= ~(OPF_RMD | OPF_DONE);
2912 j--;
2913 }
2914 }
2915 if (j != 0)
2916 ferr(&ops[i], "unhandled prologue\n");
2917
2918 // recheck
2919 i = g_stack_fsz = ecx_push = 0;
2920 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
2921 if (!(ops[i].flags & OPF_RMD))
2922 break;
2923 g_stack_fsz += 4;
2924 ecx_push++;
2925 i++;
2926 }
2927 }
2928 }
2929
2930 found = 0;
2931 if (ecx_push || esp_sub)
2932 {
2933 g_sp_frame = 1;
2934
2935 i++;
2936 do {
2937 for (; i < opcnt; i++)
2938 if (ops[i].op == OP_RET)
2939 break;
2940 j = i - 1;
2941 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
2942 if (found && is_like_tailjmp(j))
2943 break;
2944 j--;
2945 }
2946
2947 if (ecx_push > 0) {
2948 for (l = 0; l < ecx_push; l++) {
2949 if (ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ecx"))
2950 /* pop ecx */;
2951 else if (ops[j].op == OP_ADD
2952 && IS(opr_name(&ops[j], 0), "esp")
2953 && ops[j].operand[1].type == OPT_CONST)
2954 {
2955 /* add esp, N */
2956 l += ops[j].operand[1].val / 4 - 1;
2957 }
2958 else
2959 ferr(&ops[j], "'pop ecx' expected\n");
2960
2961 ops[j].flags |= OPF_RMD | OPF_DONE;
2962 j--;
2963 }
2964 if (l != ecx_push)
2965 ferr(&ops[j], "epilogue scan failed\n");
2966
2967 found = 1;
2968 }
2969
2970 if (esp_sub) {
2971 if (ops[j].op != OP_ADD
2972 || !IS(opr_name(&ops[j], 0), "esp")
2973 || ops[j].operand[1].type != OPT_CONST
2974 || ops[j].operand[1].val != g_stack_fsz)
2975 ferr(&ops[j], "'add esp' expected\n");
2976
2977 ops[j].flags |= OPF_RMD | OPF_DONE;
2978 ops[j].operand[1].val = 0; // hack for stack arg scanner
2979 found = 1;
2980 }
2981
2982 i++;
2983 } while (i < opcnt);
2984 }
2985}
2986
2987static const struct parsed_proto *resolve_icall(int i, int opcnt,
2988 int *multi_src)
2989{
2990 const struct parsed_proto *pp = NULL;
2991 int search_advice = 0;
2992
2993 *multi_src = 0;
2994
2995 switch (ops[i].operand[0].type) {
2996 case OPT_REGMEM:
2997 case OPT_LABEL:
2998 case OPT_OFFSET:
2999 pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice);
3000 if (!search_advice)
3001 break;
3002 // fallthrough
3003 default:
3004 scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp,
3005 multi_src);
3006 break;
3007 }
3008
3009 return pp;
3010}
3011
3012// find an instruction that changed opr before i op
3013// *op_i must be set to -1 by caller
3014// *entry is set to 1 if one source is determined to be the caller
3015// returns 1 if found, *op_i is then set to origin
3016static int resolve_origin(int i, const struct parsed_opr *opr,
3017 int magic, int *op_i, int *is_caller)
3018{
3019 struct label_ref *lr;
3020 int ret = 0;
3021
3022 if (ops[i].cc_scratch == magic)
3023 return 0;
3024 ops[i].cc_scratch = magic;
3025
3026 while (1) {
3027 if (g_labels[i] != NULL) {
3028 lr = &g_label_refs[i];
3029 for (; lr != NULL; lr = lr->next) {
3030 check_i(&ops[i], lr->i);
3031 ret |= resolve_origin(lr->i, opr, magic, op_i, is_caller);
3032 }
3033 if (i > 0 && LAST_OP(i - 1))
3034 return ret;
3035 }
3036
3037 i--;
3038 if (i < 0) {
3039 if (is_caller != NULL)
3040 *is_caller = 1;
3041 return -1;
3042 }
3043
3044 if (ops[i].cc_scratch == magic)
3045 return 0;
3046 ops[i].cc_scratch = magic;
3047
3048 if (!(ops[i].flags & OPF_DATA))
3049 continue;
3050 if (!is_opr_modified(opr, &ops[i]))
3051 continue;
3052
3053 if (*op_i >= 0) {
3054 if (*op_i == i)
3055 return 1;
3056 // XXX: could check if the other op does the same
3057 return -1;
3058 }
3059
3060 *op_i = i;
3061 return 1;
3062 }
3063}
3064
3065static int try_resolve_const(int i, const struct parsed_opr *opr,
3066 int magic, unsigned int *val)
3067{
3068 int s_i = -1;
3069 int ret;
3070
3071 ret = resolve_origin(i, opr, magic, &s_i, NULL);
3072 if (ret == 1) {
3073 i = s_i;
3074 if (ops[i].op != OP_MOV && ops[i].operand[1].type != OPT_CONST)
3075 return -1;
3076
3077 *val = ops[i].operand[1].val;
3078 return 1;
3079 }
3080
3081 return -1;
3082}
3083
3084static struct parsed_proto *process_call_early(int i, int opcnt,
3085 int *adj_i)
3086{
3087 struct parsed_op *po = &ops[i];
3088 struct parsed_proto *pp;
3089 int multipath = 0;
3090 int adj = 0;
3091 int ret;
3092
3093 pp = po->pp;
3094 if (pp == NULL || pp->is_vararg || pp->argc_reg != 0)
3095 // leave for later
3096 return NULL;
3097
3098 // look for and make use of esp adjust
3099 *adj_i = ret = -1;
3100 if (!pp->is_stdcall && pp->argc_stack > 0)
3101 ret = scan_for_esp_adjust(i + 1, opcnt,
3102 pp->argc_stack * 4, &adj, &multipath);
3103 if (ret >= 0) {
3104 if (pp->argc_stack > adj / 4)
3105 return NULL;
3106 if (multipath)
3107 return NULL;
3108 if (ops[ret].op == OP_POP && adj != 4)
3109 return NULL;
3110 }
3111
3112 *adj_i = ret;
3113 return pp;
3114}
3115
3116static void patch_esp_adjust(struct parsed_op *po, int adj)
3117{
3118 ferr_assert(po, po->op == OP_ADD);
3119 ferr_assert(po, IS(opr_name(po, 0), "esp"));
3120 ferr_assert(po, po->operand[1].type == OPT_CONST);
3121
3122 // this is a bit of a hack, but deals with use of
3123 // single adj for multiple calls
3124 po->operand[1].val -= adj;
3125 po->flags |= OPF_RMD;
3126 if (po->operand[1].val == 0)
3127 po->flags |= OPF_DONE;
3128 ferr_assert(po, (int)po->operand[1].val >= 0);
3129}
3130
3131static struct parsed_proto *process_call(int i, int opcnt)
3132{
3133 struct parsed_op *po = &ops[i];
3134 const struct parsed_proto *pp_c;
3135 struct parsed_proto *pp;
3136 const char *tmpname;
3137 int adj = 0, multipath = 0;
3138 int ret, arg;
3139
3140 tmpname = opr_name(po, 0);
3141 pp = po->pp;
3142 if (pp == NULL)
3143 {
3144 // indirect call
3145 pp_c = resolve_icall(i, opcnt, &multipath);
3146 if (pp_c != NULL) {
3147 if (!pp_c->is_func && !pp_c->is_fptr)
3148 ferr(po, "call to non-func: %s\n", pp_c->name);
3149 pp = proto_clone(pp_c);
3150 my_assert_not(pp, NULL);
3151 if (multipath)
3152 // not resolved just to single func
3153 pp->is_fptr = 1;
3154
3155 switch (po->operand[0].type) {
3156 case OPT_REG:
3157 // we resolved this call and no longer need the register
3158 po->regmask_src &= ~(1 << po->operand[0].reg);
3159 break;
3160 case OPT_REGMEM:
3161 pp->is_fptr = 1;
3162 break;
3163 default:
3164 break;
3165 }
3166 }
3167 if (pp == NULL) {
3168 pp = calloc(1, sizeof(*pp));
3169 my_assert_not(pp, NULL);
3170
3171 pp->is_fptr = 1;
3172 ret = scan_for_esp_adjust(i + 1, opcnt, 32*4, &adj, &multipath);
3173 if (ret < 0 || adj < 0) {
3174 if (!g_allow_regfunc)
3175 ferr(po, "non-__cdecl indirect call unhandled yet\n");
3176 pp->is_unresolved = 1;
3177 adj = 0;
3178 }
3179 adj /= 4;
3180 if (adj > ARRAY_SIZE(pp->arg))
3181 ferr(po, "esp adjust too large: %d\n", adj);
3182 pp->ret_type.name = strdup("int");
3183 pp->argc = pp->argc_stack = adj;
3184 for (arg = 0; arg < pp->argc; arg++)
3185 pp->arg[arg].type.name = strdup("int");
3186 }
3187 po->pp = pp;
3188 }
3189
3190 // look for and make use of esp adjust
3191 multipath = 0;
3192 ret = -1;
3193 if (!pp->is_stdcall && pp->argc_stack > 0)
3194 ret = scan_for_esp_adjust(i + 1, opcnt,
3195 pp->argc_stack * 4, &adj, &multipath);
3196 if (ret >= 0) {
3197 if (pp->is_vararg) {
3198 if (adj / 4 < pp->argc_stack) {
3199 fnote(po, "(this call)\n");
3200 ferr(&ops[ret], "esp adjust is too small: %x < %x\n",
3201 adj, pp->argc_stack * 4);
3202 }
3203 // modify pp to make it have varargs as normal args
3204 arg = pp->argc;
3205 pp->argc += adj / 4 - pp->argc_stack;
3206 for (; arg < pp->argc; arg++) {
3207 pp->arg[arg].type.name = strdup("int");
3208 pp->argc_stack++;
3209 }
3210 if (pp->argc > ARRAY_SIZE(pp->arg))
3211 ferr(po, "too many args for '%s'\n", tmpname);
3212 }
3213 if (pp->argc_stack > adj / 4) {
3214 fnote(po, "(this call)\n");
3215 ferr(&ops[ret], "stack tracking failed for '%s': %x %x\n",
3216 tmpname, pp->argc_stack * 4, adj);
3217 }
3218
3219 ops[ret].flags |= OPF_RMD;
3220 if (ops[ret].op == OP_POP) {
3221 if (adj > 4) {
3222 // deal with multi-pop stack adjust
3223 adj = pp->argc_stack;
3224 while (ops[ret].op == OP_POP && adj > 0 && ret < opcnt) {
3225 ops[ret].flags |= OPF_RMD | OPF_DONE;
3226 adj--;
3227 ret++;
3228 }
3229 }
3230 }
3231 else if (!multipath)
3232 patch_esp_adjust(&ops[ret], pp->argc_stack * 4);
3233 }
3234 else if (pp->is_vararg)
3235 ferr(po, "missing esp_adjust for vararg func '%s'\n",
3236 pp->name);
3237
3238 return pp;
3239}
3240
3241static int collect_call_args_early(struct parsed_op *po, int i,
3242 struct parsed_proto *pp, int *regmask)
3243{
3244 int arg, ret;
3245 int j;
3246
3247 for (arg = 0; arg < pp->argc; arg++)
3248 if (pp->arg[arg].reg == NULL)
3249 break;
3250
3251 // first see if it can be easily done
3252 for (j = i; j > 0 && arg < pp->argc; )
3253 {
3254 if (g_labels[j] != NULL)
3255 return -1;
3256 j--;
3257
3258 if (ops[j].op == OP_CALL)
3259 return -1;
3260 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP)
3261 return -1;
3262 else if (ops[j].op == OP_POP)
3263 return -1;
3264 else if (ops[j].flags & OPF_CJMP)
3265 return -1;
3266 else if (ops[j].op == OP_PUSH) {
3267 if (ops[j].flags & (OPF_FARG|OPF_FARGNR))
3268 return -1;
3269 ret = scan_for_mod(&ops[j], j + 1, i, 1);
3270 if (ret >= 0)
3271 return -1;
3272
3273 if (pp->arg[arg].type.is_va_list)
3274 return -1;
3275
3276 // next arg
3277 for (arg++; arg < pp->argc; arg++)
3278 if (pp->arg[arg].reg == NULL)
3279 break;
3280 }
3281 }
3282
3283 if (arg < pp->argc)
3284 return -1;
3285
3286 // now do it
3287 for (arg = 0; arg < pp->argc; arg++)
3288 if (pp->arg[arg].reg == NULL)
3289 break;
3290
3291 for (j = i; j > 0 && arg < pp->argc; )
3292 {
3293 j--;
3294
3295 if (ops[j].op == OP_PUSH)
3296 {
3297 ops[j].p_argnext = -1;
3298 ferr_assert(&ops[j], pp->arg[arg].datap == NULL);
3299 pp->arg[arg].datap = &ops[j];
3300
3301 if (ops[j].operand[0].type == OPT_REG)
3302 *regmask |= 1 << ops[j].operand[0].reg;
3303
3304 ops[j].flags |= OPF_RMD | OPF_DONE | OPF_FARGNR | OPF_FARG;
3305 ops[j].flags &= ~OPF_RSAVE;
3306
3307 // next arg
3308 for (arg++; arg < pp->argc; arg++)
3309 if (pp->arg[arg].reg == NULL)
3310 break;
3311 }
3312 }
3313
3314 return 0;
3315}
3316
3317static int collect_call_args_r(struct parsed_op *po, int i,
3318 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
3319 int *arg_grp, int arg, int magic, int need_op_saving, int may_reuse)
3320{
3321 struct parsed_proto *pp_tmp;
3322 struct parsed_op *po_tmp;
3323 struct label_ref *lr;
3324 int need_to_save_current;
3325 int arg_grp_current = 0;
3326 int save_args_seen = 0;
3327 int save_args;
3328 int ret = 0;
3329 int reg;
3330 char buf[32];
3331 int j, k;
3332
3333 if (i < 0) {
3334 ferr(po, "dead label encountered\n");
3335 return -1;
3336 }
3337
3338 for (; arg < pp->argc; arg++)
3339 if (pp->arg[arg].reg == NULL)
3340 break;
3341 magic = (magic & 0xffffff) | (arg << 24);
3342
3343 for (j = i; j >= 0 && (arg < pp->argc || pp->is_unresolved); )
3344 {
3345 if (((ops[j].cc_scratch ^ magic) & 0xffffff) == 0) {
3346 if (ops[j].cc_scratch != magic) {
3347 ferr(&ops[j], "arg collect hit same path with diff args for %s\n",
3348 pp->name);
3349 return -1;
3350 }
3351 // ok: have already been here
3352 return 0;
3353 }
3354 ops[j].cc_scratch = magic;
3355
3356 if (g_labels[j] != NULL && g_label_refs[j].i != -1) {
3357 lr = &g_label_refs[j];
3358 if (lr->next != NULL)
3359 need_op_saving = 1;
3360 for (; lr->next; lr = lr->next) {
3361 check_i(&ops[j], lr->i);
3362 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
3363 may_reuse = 1;
3364 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
3365 arg_grp, arg, magic, need_op_saving, may_reuse);
3366 if (ret < 0)
3367 return ret;
3368 }
3369
3370 check_i(&ops[j], lr->i);
3371 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
3372 may_reuse = 1;
3373 if (j > 0 && LAST_OP(j - 1)) {
3374 // follow last branch in reverse
3375 j = lr->i;
3376 continue;
3377 }
3378 need_op_saving = 1;
3379 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
3380 arg_grp, arg, magic, need_op_saving, may_reuse);
3381 if (ret < 0)
3382 return ret;
3383 }
3384 j--;
3385
3386 if (ops[j].op == OP_CALL)
3387 {
3388 if (pp->is_unresolved)
3389 break;
3390
3391 pp_tmp = ops[j].pp;
3392 if (pp_tmp == NULL)
3393 ferr(po, "arg collect hit unparsed call '%s'\n",
3394 ops[j].operand[0].name);
3395 if (may_reuse && pp_tmp->argc_stack > 0)
3396 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
3397 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
3398 }
3399 // esp adjust of 0 means we collected it before
3400 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP
3401 && (ops[j].operand[1].type != OPT_CONST
3402 || ops[j].operand[1].val != 0))
3403 {
3404 if (pp->is_unresolved)
3405 break;
3406
3407 ferr(po, "arg collect %d/%d hit esp adjust of %d\n",
3408 arg, pp->argc, ops[j].operand[1].val);
3409 }
3410 else if (ops[j].op == OP_POP && !(ops[j].flags & OPF_DONE))
3411 {
3412 if (pp->is_unresolved)
3413 break;
3414
3415 ferr(po, "arg collect %d/%d hit pop\n", arg, pp->argc);
3416 }
3417 else if (ops[j].flags & OPF_CJMP)
3418 {
3419 if (pp->is_unresolved)
3420 break;
3421
3422 may_reuse = 1;
3423 }
3424 else if (ops[j].op == OP_PUSH
3425 && !(ops[j].flags & (OPF_FARGNR|OPF_DONE)))
3426 {
3427 if (pp->is_unresolved && (ops[j].flags & OPF_RMD))
3428 break;
3429
3430 ops[j].p_argnext = -1;
3431 po_tmp = pp->arg[arg].datap;
3432 if (po_tmp != NULL)
3433 ops[j].p_argnext = po_tmp - ops;
3434 pp->arg[arg].datap = &ops[j];
3435
3436 need_to_save_current = 0;
3437 save_args = 0;
3438 reg = -1;
3439 if (ops[j].operand[0].type == OPT_REG)
3440 reg = ops[j].operand[0].reg;
3441
3442 if (!need_op_saving) {
3443 ret = scan_for_mod(&ops[j], j + 1, i, 1);
3444 need_to_save_current = (ret >= 0);
3445 }
3446 if (need_op_saving || need_to_save_current) {
3447 // mark this push as one that needs operand saving
3448 ops[j].flags &= ~OPF_RMD;
3449 if (ops[j].p_argnum == 0) {
3450 ops[j].p_argnum = arg + 1;
3451 save_args |= 1 << arg;
3452 }
3453 else if (ops[j].p_argnum < arg + 1) {
3454 // XXX: might kill valid var..
3455 //*save_arg_vars &= ~(1 << (ops[j].p_argnum - 1));
3456 ops[j].p_argnum = arg + 1;
3457 save_args |= 1 << arg;
3458 }
3459
3460 if (save_args_seen & (1 << (ops[j].p_argnum - 1))) {
3461 save_args_seen = 0;
3462 arg_grp_current++;
3463 if (arg_grp_current >= MAX_ARG_GRP)
3464 ferr(&ops[j], "out of arg groups (arg%d), f %s\n",
3465 ops[j].p_argnum, pp->name);
3466 }
3467 }
3468 else if (ops[j].p_argnum == 0)
3469 ops[j].flags |= OPF_RMD;
3470
3471 // some PUSHes are reused by different calls on other branches,
3472 // but that can't happen if we didn't branch, so they
3473 // can be removed from future searches (handles nested calls)
3474 if (!may_reuse)
3475 ops[j].flags |= OPF_FARGNR;
3476
3477 ops[j].flags |= OPF_FARG;
3478 ops[j].flags &= ~OPF_RSAVE;
3479
3480 // check for __VALIST
3481 if (!pp->is_unresolved && pp->arg[arg].type.is_va_list) {
3482 k = -1;
3483 ret = resolve_origin(j, &ops[j].operand[0],
3484 magic + 1, &k, NULL);
3485 if (ret == 1 && k >= 0)
3486 {
3487 if (ops[k].op == OP_LEA) {
3488 snprintf(buf, sizeof(buf), "arg_%X",
3489 g_func_pp->argc_stack * 4);
3490 if (!g_func_pp->is_vararg
3491 || strstr(ops[k].operand[1].name, buf))
3492 {
3493 ops[k].flags |= OPF_RMD | OPF_DONE;
3494 ops[j].flags |= OPF_RMD | OPF_VAPUSH;
3495 save_args &= ~(1 << arg);
3496 reg = -1;
3497 }
3498 else
3499 ferr(&ops[j], "lea va_list used, but no vararg?\n");
3500 }
3501 // check for va_list from g_func_pp arg too
3502 else if (ops[k].op == OP_MOV
3503 && is_stack_access(&ops[k], &ops[k].operand[1]))
3504 {
3505 ret = stack_frame_access(&ops[k], &ops[k].operand[1],
3506 buf, sizeof(buf), ops[k].operand[1].name, "", 1, 0);
3507 if (ret >= 0) {
3508 ops[k].flags |= OPF_RMD | OPF_DONE;
3509 ops[j].flags |= OPF_RMD;
3510 ops[j].p_argpass = ret + 1;
3511 save_args &= ~(1 << arg);
3512 reg = -1;
3513 }
3514 }
3515 }
3516 }
3517
3518 *save_arg_vars |= save_args;
3519
3520 // tracking reg usage
3521 if (reg >= 0)
3522 *regmask |= 1 << reg;
3523
3524 arg++;
3525 if (!pp->is_unresolved) {
3526 // next arg
3527 for (; arg < pp->argc; arg++)
3528 if (pp->arg[arg].reg == NULL)
3529 break;
3530 }
3531 magic = (magic & 0xffffff) | (arg << 24);
3532 }
3533
3534 if (ops[j].p_arggrp > arg_grp_current) {
3535 save_args_seen = 0;
3536 arg_grp_current = ops[j].p_arggrp;
3537 }
3538 if (ops[j].p_argnum > 0)
3539 save_args_seen |= 1 << (ops[j].p_argnum - 1);
3540 }
3541
3542 if (arg < pp->argc) {
3543 ferr(po, "arg collect failed for '%s': %d/%d\n",
3544 pp->name, arg, pp->argc);
3545 return -1;
3546 }
3547
3548 if (arg_grp_current > *arg_grp)
3549 *arg_grp = arg_grp_current;
3550
3551 return arg;
3552}
3553
3554static int collect_call_args(struct parsed_op *po, int i,
3555 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
3556 int magic)
3557{
3558 // arg group is for cases when pushes for
3559 // multiple funcs are going on
3560 struct parsed_op *po_tmp;
3561 int save_arg_vars_current = 0;
3562 int arg_grp = 0;
3563 int ret;
3564 int a;
3565
3566 ret = collect_call_args_r(po, i, pp, regmask,
3567 &save_arg_vars_current, &arg_grp, 0, magic, 0, 0);
3568 if (ret < 0)
3569 return ret;
3570
3571 if (arg_grp != 0) {
3572 // propagate arg_grp
3573 for (a = 0; a < pp->argc; a++) {
3574 if (pp->arg[a].reg != NULL)
3575 continue;
3576
3577 po_tmp = pp->arg[a].datap;
3578 while (po_tmp != NULL) {
3579 po_tmp->p_arggrp = arg_grp;
3580 if (po_tmp->p_argnext > 0)
3581 po_tmp = &ops[po_tmp->p_argnext];
3582 else
3583 po_tmp = NULL;
3584 }
3585 }
3586 }
3587 save_arg_vars[arg_grp] |= save_arg_vars_current;
3588
3589 if (pp->is_unresolved) {
3590 pp->argc += ret;
3591 pp->argc_stack += ret;
3592 for (a = 0; a < pp->argc; a++)
3593 if (pp->arg[a].type.name == NULL)
3594 pp->arg[a].type.name = strdup("int");
3595 }
3596
3597 return ret;
3598}
3599
3600static void pp_insert_reg_arg(struct parsed_proto *pp, const char *reg)
3601{
3602 int i;
3603
3604 for (i = 0; i < pp->argc; i++)
3605 if (pp->arg[i].reg == NULL)
3606 break;
3607
3608 if (pp->argc_stack)
3609 memmove(&pp->arg[i + 1], &pp->arg[i],
3610 sizeof(pp->arg[0]) * pp->argc_stack);
3611 memset(&pp->arg[i], 0, sizeof(pp->arg[i]));
3612 pp->arg[i].reg = strdup(reg);
3613 pp->arg[i].type.name = strdup("int");
3614 pp->argc++;
3615 pp->argc_reg++;
3616}
3617
3618static void add_label_ref(struct label_ref *lr, int op_i)
3619{
3620 struct label_ref *lr_new;
3621
3622 if (lr->i == -1) {
3623 lr->i = op_i;
3624 return;
3625 }
3626
3627 lr_new = calloc(1, sizeof(*lr_new));
3628 lr_new->i = op_i;
3629 lr_new->next = lr->next;
3630 lr->next = lr_new;
3631}
3632
3633static struct parsed_data *try_resolve_jumptab(int i, int opcnt)
3634{
3635 struct parsed_op *po = &ops[i];
3636 struct parsed_data *pd;
3637 char label[NAMELEN], *p;
3638 int len, j, l;
3639
3640 p = strchr(po->operand[0].name, '[');
3641 if (p == NULL)
3642 return NULL;
3643
3644 len = p - po->operand[0].name;
3645 strncpy(label, po->operand[0].name, len);
3646 label[len] = 0;
3647
3648 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
3649 if (IS(g_func_pd[j].label, label)) {
3650 pd = &g_func_pd[j];
3651 break;
3652 }
3653 }
3654 if (pd == NULL)
3655 //ferr(po, "label '%s' not parsed?\n", label);
3656 return NULL;
3657
3658 if (pd->type != OPT_OFFSET)
3659 ferr(po, "label '%s' with non-offset data?\n", label);
3660
3661 // find all labels, link
3662 for (j = 0; j < pd->count; j++) {
3663 for (l = 0; l < opcnt; l++) {
3664 if (g_labels[l] != NULL && IS(g_labels[l], pd->d[j].u.label)) {
3665 add_label_ref(&g_label_refs[l], i);
3666 pd->d[j].bt_i = l;
3667 break;
3668 }
3669 }
3670 }
3671
3672 return pd;
3673}
3674
3675static void clear_labels(int count)
3676{
3677 int i;
3678
3679 for (i = 0; i < count; i++) {
3680 if (g_labels[i] != NULL) {
3681 free(g_labels[i]);
3682 g_labels[i] = NULL;
3683 }
3684 }
3685}
3686
3687static void output_std_flags(FILE *fout, struct parsed_op *po,
3688 int *pfomask, const char *dst_opr_text)
3689{
3690 if (*pfomask & (1 << PFO_Z)) {
3691 fprintf(fout, "\n cond_z = (%s%s == 0);",
3692 lmod_cast_u(po, po->operand[0].lmod), dst_opr_text);
3693 *pfomask &= ~(1 << PFO_Z);
3694 }
3695 if (*pfomask & (1 << PFO_S)) {
3696 fprintf(fout, "\n cond_s = (%s%s < 0);",
3697 lmod_cast_s(po, po->operand[0].lmod), dst_opr_text);
3698 *pfomask &= ~(1 << PFO_S);
3699 }
3700}
3701
3702static void output_pp_attrs(FILE *fout, const struct parsed_proto *pp,
3703 int is_noreturn)
3704{
3705 if (pp->is_fastcall)
3706 fprintf(fout, "__fastcall ");
3707 else if (pp->is_stdcall && pp->argc_reg == 0)
3708 fprintf(fout, "__stdcall ");
3709 if (pp->is_noreturn || is_noreturn)
3710 fprintf(fout, "noreturn ");
3711}
3712
3713static int get_pp_arg_regmask(const struct parsed_proto *pp)
3714{
3715 int regmask = 0;
3716 int i, reg;
3717
3718 for (i = 0; i < pp->argc; i++) {
3719 if (pp->arg[i].reg != NULL) {
3720 reg = char_array_i(regs_r32,
3721 ARRAY_SIZE(regs_r32), pp->arg[i].reg);
3722 if (reg < 0)
3723 ferr(ops, "arg '%s' of func '%s' is not a reg?\n",
3724 pp->arg[i].reg, pp->name);
3725 regmask |= 1 << reg;
3726 }
3727 }
3728
3729 return regmask;
3730}
3731
3732static char *saved_arg_name(char *buf, size_t buf_size, int grp, int num)
3733{
3734 char buf1[16];
3735
3736 buf1[0] = 0;
3737 if (grp > 0)
3738 snprintf(buf1, sizeof(buf1), "%d", grp);
3739 snprintf(buf, buf_size, "s%s_a%d", buf1, num);
3740
3741 return buf;
3742}
3743
3744static void gen_x_cleanup(int opcnt);
3745
3746static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
3747{
3748 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
3749 struct parsed_opr *last_arith_dst = NULL;
3750 char buf1[256], buf2[256], buf3[256], cast[64];
3751 const struct parsed_proto *pp_c;
3752 struct parsed_proto *pp, *pp_tmp;
3753 struct parsed_data *pd;
3754 const char *tmpname;
3755 unsigned int uval;
3756 int save_arg_vars[MAX_ARG_GRP] = { 0, };
3757 int cond_vars = 0;
3758 int need_tmp_var = 0;
3759 int need_tmp64 = 0;
3760 int had_decl = 0;
3761 int label_pending = 0;
3762 int regmask_save = 0;
3763 int regmask_arg = 0;
3764 int regmask_now = 0;
3765 int regmask_init = 0;
3766 int regmask = 0;
3767 int pfomask = 0;
3768 int found = 0;
3769 int depth = 0;
3770 int no_output;
3771 int i, j, l;
3772 int arg;
3773 int reg;
3774 int ret;
3775
3776 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
3777 g_stack_frame_used = 0;
3778
3779 g_func_pp = proto_parse(fhdr, funcn, 0);
3780 if (g_func_pp == NULL)
3781 ferr(ops, "proto_parse failed for '%s'\n", funcn);
3782
3783 regmask_arg = get_pp_arg_regmask(g_func_pp);
3784
3785 // pass1:
3786 // - handle ebp/esp frame, remove ops related to it
3787 scan_prologue_epilogue(opcnt);
3788
3789 // pass2:
3790 // - parse calls with labels
3791 // - resolve all branches
3792 for (i = 0; i < opcnt; i++)
3793 {
3794 po = &ops[i];
3795 po->bt_i = -1;
3796 po->btj = NULL;
3797
3798 if (po->flags & (OPF_RMD|OPF_DONE))
3799 continue;
3800
3801 if (po->op == OP_CALL) {
3802 pp = NULL;
3803
3804 if (po->operand[0].type == OPT_LABEL) {
3805 tmpname = opr_name(po, 0);
3806 if (IS_START(tmpname, "loc_"))
3807 ferr(po, "call to loc_*\n");
3808 pp_c = proto_parse(fhdr, tmpname, 0);
3809 if (pp_c == NULL)
3810 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
3811
3812 pp = proto_clone(pp_c);
3813 my_assert_not(pp, NULL);
3814 }
3815 else if (po->datap != NULL) {
3816 pp = calloc(1, sizeof(*pp));
3817 my_assert_not(pp, NULL);
3818
3819 ret = parse_protostr(po->datap, pp);
3820 if (ret < 0)
3821 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
3822 free(po->datap);
3823 po->datap = NULL;
3824 }
3825
3826 if (pp != NULL) {
3827 if (pp->is_fptr)
3828 check_func_pp(po, pp, "fptr var call");
3829 if (pp->is_noreturn)
3830 po->flags |= OPF_TAIL;
3831 }
3832 po->pp = pp;
3833 continue;
3834 }
3835
3836 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
3837 continue;
3838
3839 if (po->operand[0].type == OPT_REGMEM) {
3840 pd = try_resolve_jumptab(i, opcnt);
3841 if (pd == NULL)
3842 goto tailcall;
3843
3844 po->btj = pd;
3845 continue;
3846 }
3847
3848 for (l = 0; l < opcnt; l++) {
3849 if (g_labels[l] != NULL
3850 && IS(po->operand[0].name, g_labels[l]))
3851 {
3852 if (l == i + 1 && po->op == OP_JMP) {
3853 // yet another alignment type..
3854 po->flags |= OPF_RMD|OPF_DONE;
3855 break;
3856 }
3857 add_label_ref(&g_label_refs[l], i);
3858 po->bt_i = l;
3859 break;
3860 }
3861 }
3862
3863 if (po->bt_i != -1 || (po->flags & OPF_RMD))
3864 continue;
3865
3866 if (po->operand[0].type == OPT_LABEL)
3867 // assume tail call
3868 goto tailcall;
3869
3870 ferr(po, "unhandled branch\n");
3871
3872tailcall:
3873 po->op = OP_CALL;
3874 po->flags |= OPF_TAIL;
3875 if (i > 0 && ops[i - 1].op == OP_POP)
3876 po->flags |= OPF_ATAIL;
3877 i--; // reprocess
3878 }
3879
3880 // pass3:
3881 // - remove dead labels
3882 // - process trivial calls
3883 for (i = 0; i < opcnt; i++)
3884 {
3885 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
3886 free(g_labels[i]);
3887 g_labels[i] = NULL;
3888 }
3889
3890 po = &ops[i];
3891 if (po->flags & (OPF_RMD|OPF_DONE))
3892 continue;
3893
3894 if (po->op == OP_CALL)
3895 {
3896 pp = process_call_early(i, opcnt, &j);
3897 if (pp != NULL) {
3898 if (!(po->flags & OPF_ATAIL))
3899 // since we know the args, try to collect them
3900 if (collect_call_args_early(po, i, pp, &regmask) != 0)
3901 pp = NULL;
3902 }
3903
3904 if (pp != NULL) {
3905 if (j >= 0) {
3906 // commit esp adjust
3907 ops[j].flags |= OPF_RMD;
3908 if (ops[j].op != OP_POP)
3909 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
3910 else
3911 ops[j].flags |= OPF_DONE;
3912 }
3913
3914 if (strstr(pp->ret_type.name, "int64"))
3915 need_tmp64 = 1;
3916
3917 po->flags |= OPF_DONE;
3918 }
3919 }
3920 }
3921
3922 // pass4:
3923 // - process calls
3924 for (i = 0; i < opcnt; i++)
3925 {
3926 po = &ops[i];
3927 if (po->flags & (OPF_RMD|OPF_DONE))
3928 continue;
3929
3930 if (po->op == OP_CALL && !(po->flags & OPF_DONE))
3931 {
3932 pp = process_call(i, opcnt);
3933
3934 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
3935 // since we know the args, collect them
3936 collect_call_args(po, i, pp, &regmask, save_arg_vars,
3937 i + opcnt * 2);
3938 }
3939
3940 if (strstr(pp->ret_type.name, "int64"))
3941 need_tmp64 = 1;
3942 }
3943 }
3944
3945 // pass5:
3946 // - find POPs for PUSHes, rm both
3947 // - scan for STD/CLD, propagate DF
3948 // - scan for all used registers
3949 // - find flag set ops for their users
3950 // - do unreselved calls
3951 // - declare indirect functions
3952 for (i = 0; i < opcnt; i++)
3953 {
3954 po = &ops[i];
3955 if (po->flags & (OPF_RMD|OPF_DONE))
3956 continue;
3957
3958 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE)) {
3959 reg = po->operand[0].reg;
3960 if (!(regmask & (1 << reg)))
3961 // not a reg save after all, rerun scan_for_pop
3962 po->flags &= ~OPF_RSAVE;
3963 else
3964 regmask_save |= 1 << reg;
3965 }
3966
3967 if (po->op == OP_PUSH && !(po->flags & OPF_FARG)
3968 && !(po->flags & OPF_RSAVE) && !g_func_pp->is_userstack)
3969 {
3970 if (po->operand[0].type == OPT_REG)
3971 {
3972 reg = po->operand[0].reg;
3973 if (reg < 0)
3974 ferr(po, "reg not set for push?\n");
3975
3976 depth = 0;
3977 ret = scan_for_pop(i + 1, opcnt,
3978 po->operand[0].name, i + opcnt * 3, 0, &depth, 0);
3979 if (ret == 1) {
3980 if (depth > 1)
3981 ferr(po, "too much depth: %d\n", depth);
3982
3983 po->flags |= OPF_RMD;
3984 scan_for_pop(i + 1, opcnt, po->operand[0].name,
3985 i + opcnt * 4, 0, &depth, 1);
3986 continue;
3987 }
3988 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
3989 if (ret == 0) {
3990 arg = OPF_RMD;
3991 if (regmask & (1 << reg)) {
3992 if (regmask_save & (1 << reg))
3993 ferr(po, "%s already saved?\n", po->operand[0].name);
3994 arg = OPF_RSAVE;
3995 }
3996 po->flags |= arg;
3997 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg);
3998 continue;
3999 }
4000 }
4001 else if (po->operand[0].type == OPT_CONST) {
4002 scan_for_pop_const(i, opcnt);
4003 }
4004 }
4005
4006 if (po->op == OP_STD) {
4007 po->flags |= OPF_DF | OPF_RMD | OPF_DONE;
4008 scan_propagate_df(i + 1, opcnt);
4009 }
4010
4011 regmask_now = po->regmask_src | po->regmask_dst;
4012 if (regmask_now & (1 << xBP)) {
4013 if (g_bp_frame && !(po->flags & OPF_EBP_S)) {
4014 if (po->regmask_dst & (1 << xBP))
4015 // compiler decided to drop bp frame and use ebp as scratch
4016 scan_fwd_set_flags(i + 1, opcnt, i + opcnt * 5, OPF_EBP_S);
4017 else
4018 regmask_now &= ~(1 << xBP);
4019 }
4020 }
4021
4022 regmask |= regmask_now;
4023
4024 if (po->flags & OPF_CC)
4025 {
4026 int setters[16], cnt = 0, branched = 0;
4027
4028 ret = scan_for_flag_set(i, i + opcnt * 6,
4029 &branched, setters, &cnt);
4030 if (ret < 0 || cnt <= 0)
4031 ferr(po, "unable to trace flag setter(s)\n");
4032 if (cnt > ARRAY_SIZE(setters))
4033 ferr(po, "too many flag setters\n");
4034
4035 for (j = 0; j < cnt; j++)
4036 {
4037 tmp_op = &ops[setters[j]]; // flag setter
4038 pfomask = 0;
4039
4040 // to get nicer code, we try to delay test and cmp;
4041 // if we can't because of operand modification, or if we
4042 // have arith op, or branch, make it calculate flags explicitly
4043 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP)
4044 {
4045 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i, 0) >= 0)
4046 pfomask = 1 << po->pfo;
4047 }
4048 else if (tmp_op->op == OP_CMPS || tmp_op->op == OP_SCAS) {
4049 pfomask = 1 << po->pfo;
4050 }
4051 else {
4052 // see if we'll be able to handle based on op result
4053 if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR
4054 && po->pfo != PFO_Z && po->pfo != PFO_S
4055 && po->pfo != PFO_P)
4056 || branched
4057 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
4058 {
4059 pfomask = 1 << po->pfo;
4060 }
4061
4062 if (tmp_op->op == OP_ADD && po->pfo == PFO_C) {
4063 propagate_lmod(tmp_op, &tmp_op->operand[0],
4064 &tmp_op->operand[1]);
4065 if (tmp_op->operand[0].lmod == OPLM_DWORD)
4066 need_tmp64 = 1;
4067 }
4068 }
4069 if (pfomask) {
4070 tmp_op->pfomask |= pfomask;
4071 cond_vars |= pfomask;
4072 }
4073 // note: may overwrite, currently not a problem
4074 po->datap = tmp_op;
4075 }
4076
4077 if (po->op == OP_RCL || po->op == OP_RCR
4078 || po->op == OP_ADC || po->op == OP_SBB)
4079 cond_vars |= 1 << PFO_C;
4080 }
4081
4082 if (po->op == OP_CMPS || po->op == OP_SCAS) {
4083 cond_vars |= 1 << PFO_Z;
4084 }
4085 else if (po->op == OP_MUL
4086 || (po->op == OP_IMUL && po->operand_cnt == 1))
4087 {
4088 if (po->operand[0].lmod == OPLM_DWORD)
4089 need_tmp64 = 1;
4090 }
4091 else if (po->op == OP_CALL) {
4092 // note: resolved non-reg calls are OPF_DONE already
4093 pp = po->pp;
4094 if (pp == NULL)
4095 ferr(po, "NULL pp\n");
4096
4097 if (pp->is_unresolved) {
4098 int regmask_stack = 0;
4099 collect_call_args(po, i, pp, &regmask, save_arg_vars,
4100 i + opcnt * 2);
4101
4102 // this is pretty rough guess:
4103 // see ecx and edx were pushed (and not their saved versions)
4104 for (arg = 0; arg < pp->argc; arg++) {
4105 if (pp->arg[arg].reg != NULL)
4106 continue;
4107
4108 tmp_op = pp->arg[arg].datap;
4109 if (tmp_op == NULL)
4110 ferr(po, "parsed_op missing for arg%d\n", arg);
4111 if (tmp_op->p_argnum == 0 && tmp_op->operand[0].type == OPT_REG)
4112 regmask_stack |= 1 << tmp_op->operand[0].reg;
4113 }
4114
4115 if (!((regmask_stack & (1 << xCX))
4116 && (regmask_stack & (1 << xDX))))
4117 {
4118 if (pp->argc_stack != 0
4119 || ((regmask | regmask_arg) & ((1 << xCX)|(1 << xDX))))
4120 {
4121 pp_insert_reg_arg(pp, "ecx");
4122 pp->is_fastcall = 1;
4123 regmask_init |= 1 << xCX;
4124 regmask |= 1 << xCX;
4125 }
4126 if (pp->argc_stack != 0
4127 || ((regmask | regmask_arg) & (1 << xDX)))
4128 {
4129 pp_insert_reg_arg(pp, "edx");
4130 regmask_init |= 1 << xDX;
4131 regmask |= 1 << xDX;
4132 }
4133 }
4134
4135 // note: __cdecl doesn't fall into is_unresolved category
4136 if (pp->argc_stack > 0)
4137 pp->is_stdcall = 1;
4138 }
4139
4140 for (arg = 0; arg < pp->argc; arg++) {
4141 if (pp->arg[arg].reg != NULL) {
4142 reg = char_array_i(regs_r32,
4143 ARRAY_SIZE(regs_r32), pp->arg[arg].reg);
4144 if (reg < 0)
4145 ferr(ops, "arg '%s' is not a reg?\n", pp->arg[arg].reg);
4146 if (!(regmask & (1 << reg))) {
4147 regmask_init |= 1 << reg;
4148 regmask |= 1 << reg;
4149 }
4150 }
4151 }
4152 }
4153 else if (po->op == OP_MOV && po->operand[0].pp != NULL
4154 && po->operand[1].pp != NULL)
4155 {
4156 // <var> = offset <something>
4157 if ((po->operand[1].pp->is_func || po->operand[1].pp->is_fptr)
4158 && !IS_START(po->operand[1].name, "off_"))
4159 {
4160 if (!po->operand[0].pp->is_fptr)
4161 ferr(po, "%s not declared as fptr when it should be\n",
4162 po->operand[0].name);
4163 if (pp_cmp_func(po->operand[0].pp, po->operand[1].pp)) {
4164 pp_print(buf1, sizeof(buf1), po->operand[0].pp);
4165 pp_print(buf2, sizeof(buf2), po->operand[1].pp);
4166 fnote(po, "var: %s\n", buf1);
4167 fnote(po, "func: %s\n", buf2);
4168 ferr(po, "^ mismatch\n");
4169 }
4170 }
4171 }
4172 else if (po->op == OP_RET && !IS(g_func_pp->ret_type.name, "void"))
4173 regmask |= 1 << xAX;
4174 else if (po->op == OP_DIV || po->op == OP_IDIV) {
4175 // 32bit division is common, look for it
4176 if (po->op == OP_DIV)
4177 ret = scan_for_reg_clear(i, xDX);
4178 else
4179 ret = scan_for_cdq_edx(i);
4180 if (ret >= 0)
4181 po->flags |= OPF_32BIT;
4182 else
4183 need_tmp64 = 1;
4184 }
4185 else if (po->op == OP_CLD)
4186 po->flags |= OPF_RMD | OPF_DONE;
4187
4188 if (po->op == OP_RCL || po->op == OP_RCR || po->op == OP_XCHG) {
4189 need_tmp_var = 1;
4190 }
4191 }
4192
4193 // pass6:
4194 // - confirm regmask_save, it might have been reduced
4195 if (regmask_save != 0)
4196 {
4197 regmask_save = 0;
4198 for (i = 0; i < opcnt; i++) {
4199 po = &ops[i];
4200 if (po->flags & OPF_RMD)
4201 continue;
4202
4203 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE))
4204 regmask_save |= 1 << po->operand[0].reg;
4205 }
4206 }
4207
4208 // output starts here
4209
4210 // define userstack size
4211 if (g_func_pp->is_userstack) {
4212 fprintf(fout, "#ifndef US_SZ_%s\n", g_func_pp->name);
4213 fprintf(fout, "#define US_SZ_%s USERSTACK_SIZE\n", g_func_pp->name);
4214 fprintf(fout, "#endif\n");
4215 }
4216
4217 // the function itself
4218 fprintf(fout, "%s ", g_func_pp->ret_type.name);
4219 output_pp_attrs(fout, g_func_pp, g_ida_func_attr & IDAFA_NORETURN);
4220 fprintf(fout, "%s(", g_func_pp->name);
4221
4222 for (i = 0; i < g_func_pp->argc; i++) {
4223 if (i > 0)
4224 fprintf(fout, ", ");
4225 if (g_func_pp->arg[i].fptr != NULL) {
4226 // func pointer..
4227 pp = g_func_pp->arg[i].fptr;
4228 fprintf(fout, "%s (", pp->ret_type.name);
4229 output_pp_attrs(fout, pp, 0);
4230 fprintf(fout, "*a%d)(", i + 1);
4231 for (j = 0; j < pp->argc; j++) {
4232 if (j > 0)
4233 fprintf(fout, ", ");
4234 if (pp->arg[j].fptr)
4235 ferr(ops, "nested fptr\n");
4236 fprintf(fout, "%s", pp->arg[j].type.name);
4237 }
4238 if (pp->is_vararg) {
4239 if (j > 0)
4240 fprintf(fout, ", ");
4241 fprintf(fout, "...");
4242 }
4243 fprintf(fout, ")");
4244 }
4245 else if (g_func_pp->arg[i].type.is_retreg) {
4246 fprintf(fout, "u32 *r_%s", g_func_pp->arg[i].reg);
4247 }
4248 else {
4249 fprintf(fout, "%s a%d", g_func_pp->arg[i].type.name, i + 1);
4250 }
4251 }
4252 if (g_func_pp->is_vararg) {
4253 if (i > 0)
4254 fprintf(fout, ", ");
4255 fprintf(fout, "...");
4256 }
4257
4258 fprintf(fout, ")\n{\n");
4259
4260 // declare indirect functions
4261 for (i = 0; i < opcnt; i++) {
4262 po = &ops[i];
4263 if (po->flags & OPF_RMD)
4264 continue;
4265
4266 if (po->op == OP_CALL) {
4267 pp = po->pp;
4268 if (pp == NULL)
4269 ferr(po, "NULL pp\n");
4270
4271 if (pp->is_fptr && !(pp->name[0] != 0 && pp->is_arg)) {
4272 if (pp->name[0] != 0) {
4273 memmove(pp->name + 2, pp->name, strlen(pp->name) + 1);
4274 memcpy(pp->name, "i_", 2);
4275
4276 // might be declared already
4277 found = 0;
4278 for (j = 0; j < i; j++) {
4279 if (ops[j].op == OP_CALL && (pp_tmp = ops[j].pp)) {
4280 if (pp_tmp->is_fptr && IS(pp->name, pp_tmp->name)) {
4281 found = 1;
4282 break;
4283 }
4284 }
4285 }
4286 if (found)
4287 continue;
4288 }
4289 else
4290 snprintf(pp->name, sizeof(pp->name), "icall%d", i);
4291
4292 fprintf(fout, " %s (", pp->ret_type.name);
4293 output_pp_attrs(fout, pp, 0);
4294 fprintf(fout, "*%s)(", pp->name);
4295 for (j = 0; j < pp->argc; j++) {
4296 if (j > 0)
4297 fprintf(fout, ", ");
4298 fprintf(fout, "%s a%d", pp->arg[j].type.name, j + 1);
4299 }
4300 fprintf(fout, ");\n");
4301 }
4302 }
4303 }
4304
4305 // output LUTs/jumptables
4306 for (i = 0; i < g_func_pd_cnt; i++) {
4307 pd = &g_func_pd[i];
4308 fprintf(fout, " static const ");
4309 if (pd->type == OPT_OFFSET) {
4310 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
4311
4312 for (j = 0; j < pd->count; j++) {
4313 if (j > 0)
4314 fprintf(fout, ", ");
4315 fprintf(fout, "&&%s", pd->d[j].u.label);
4316 }
4317 }
4318 else {
4319 fprintf(fout, "%s %s[] =\n { ",
4320 lmod_type_u(ops, pd->lmod), pd->label);
4321
4322 for (j = 0; j < pd->count; j++) {
4323 if (j > 0)
4324 fprintf(fout, ", ");
4325 fprintf(fout, "%u", pd->d[j].u.val);
4326 }
4327 }
4328 fprintf(fout, " };\n");
4329 had_decl = 1;
4330 }
4331
4332 // declare stack frame, va_arg
4333 if (g_stack_fsz) {
4334 fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
4335 (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz);
4336 had_decl = 1;
4337 }
4338
4339 if (g_func_pp->is_userstack) {
4340 fprintf(fout, " u32 fake_sf[US_SZ_%s / 4];\n", g_func_pp->name);
4341 fprintf(fout, " u32 *esp = &fake_sf[sizeof(fake_sf) / 4];\n");
4342 had_decl = 1;
4343 }
4344
4345 if (g_func_pp->is_vararg) {
4346 fprintf(fout, " va_list ap;\n");
4347 had_decl = 1;
4348 }
4349
4350 // declare arg-registers
4351 for (i = 0; i < g_func_pp->argc; i++) {
4352 if (g_func_pp->arg[i].reg != NULL) {
4353 reg = char_array_i(regs_r32,
4354 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
4355 if (regmask & (1 << reg)) {
4356 if (g_func_pp->arg[i].type.is_retreg)
4357 fprintf(fout, " u32 %s = *r_%s;\n",
4358 g_func_pp->arg[i].reg, g_func_pp->arg[i].reg);
4359 else
4360 fprintf(fout, " u32 %s = (u32)a%d;\n",
4361 g_func_pp->arg[i].reg, i + 1);
4362 }
4363 else {
4364 if (g_func_pp->arg[i].type.is_retreg)
4365 ferr(ops, "retreg '%s' is unused?\n",
4366 g_func_pp->arg[i].reg);
4367 fprintf(fout, " // %s = a%d; // unused\n",
4368 g_func_pp->arg[i].reg, i + 1);
4369 }
4370 had_decl = 1;
4371 }
4372 }
4373
4374 regmask_now = regmask & ~regmask_arg;
4375 regmask_now &= ~(1 << xSP);
4376 if (regmask_now & 0x00ff) {
4377 for (reg = 0; reg < 8; reg++) {
4378 if (regmask_now & (1 << reg)) {
4379 fprintf(fout, " u32 %s", regs_r32[reg]);
4380 if (regmask_init & (1 << reg))
4381 fprintf(fout, " = 0");
4382 fprintf(fout, ";\n");
4383 had_decl = 1;
4384 }
4385 }
4386 }
4387 if (regmask_now & 0xff00) {
4388 for (reg = 8; reg < 16; reg++) {
4389 if (regmask_now & (1 << reg)) {
4390 fprintf(fout, " mmxr %s", regs_r32[reg]);
4391 if (regmask_init & (1 << reg))
4392 fprintf(fout, " = { 0, }");
4393 fprintf(fout, ";\n");
4394 had_decl = 1;
4395 }
4396 }
4397 }
4398
4399 if (regmask_save) {
4400 for (reg = 0; reg < 8; reg++) {
4401 if (regmask_save & (1 << reg)) {
4402 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
4403 had_decl = 1;
4404 }
4405 }
4406 }
4407
4408 for (i = 0; i < ARRAY_SIZE(save_arg_vars); i++) {
4409 if (save_arg_vars[i] == 0)
4410 continue;
4411 for (reg = 0; reg < 32; reg++) {
4412 if (save_arg_vars[i] & (1 << reg)) {
4413 fprintf(fout, " u32 %s;\n",
4414 saved_arg_name(buf1, sizeof(buf1), i, reg + 1));
4415 had_decl = 1;
4416 }
4417 }
4418 }
4419
4420 if (cond_vars) {
4421 for (i = 0; i < 8; i++) {
4422 if (cond_vars & (1 << i)) {
4423 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
4424 had_decl = 1;
4425 }
4426 }
4427 }
4428
4429 if (need_tmp_var) {
4430 fprintf(fout, " u32 tmp;\n");
4431 had_decl = 1;
4432 }
4433
4434 if (need_tmp64) {
4435 fprintf(fout, " u64 tmp64;\n");
4436 had_decl = 1;
4437 }
4438
4439 if (had_decl)
4440 fprintf(fout, "\n");
4441
4442 if (g_func_pp->is_vararg) {
4443 if (g_func_pp->argc_stack == 0)
4444 ferr(ops, "vararg func without stack args?\n");
4445 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
4446 }
4447
4448 // output ops
4449 for (i = 0; i < opcnt; i++)
4450 {
4451 if (g_labels[i] != NULL) {
4452 fprintf(fout, "\n%s:\n", g_labels[i]);
4453 label_pending = 1;
4454
4455 delayed_flag_op = NULL;
4456 last_arith_dst = NULL;
4457 }
4458
4459 po = &ops[i];
4460 if (po->flags & OPF_RMD)
4461 continue;
4462
4463 no_output = 0;
4464
4465 #define assert_operand_cnt(n_) \
4466 if (po->operand_cnt != n_) \
4467 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
4468
4469 // conditional/flag using op?
4470 if (po->flags & OPF_CC)
4471 {
4472 int is_delayed = 0;
4473
4474 tmp_op = po->datap;
4475
4476 // we go through all this trouble to avoid using parsed_flag_op,
4477 // which makes generated code much nicer
4478 if (delayed_flag_op != NULL)
4479 {
4480 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op,
4481 po->pfo, po->pfo_inv);
4482 is_delayed = 1;
4483 }
4484 else if (last_arith_dst != NULL
4485 && (po->pfo == PFO_Z || po->pfo == PFO_S || po->pfo == PFO_P
4486 || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR))
4487 ))
4488 {
4489 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
4490 out_test_for_cc(buf1, sizeof(buf1), po, po->pfo, po->pfo_inv,
4491 last_arith_dst->lmod, buf3);
4492 is_delayed = 1;
4493 }
4494 else if (tmp_op != NULL) {
4495 // use preprocessed flag calc results
4496 if (!(tmp_op->pfomask & (1 << po->pfo)))
4497 ferr(po, "not prepared for pfo %d\n", po->pfo);
4498
4499 // note: pfo_inv was not yet applied
4500 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
4501 po->pfo_inv ? "!" : "", parsed_flag_op_names[po->pfo]);
4502 }
4503 else {
4504 ferr(po, "all methods of finding comparison failed\n");
4505 }
4506
4507 if (po->flags & OPF_JMP) {
4508 fprintf(fout, " if %s", buf1);
4509 }
4510 else if (po->op == OP_RCL || po->op == OP_RCR
4511 || po->op == OP_ADC || po->op == OP_SBB)
4512 {
4513 if (is_delayed)
4514 fprintf(fout, " cond_%s = %s;\n",
4515 parsed_flag_op_names[po->pfo], buf1);
4516 }
4517 else if (po->flags & OPF_DATA) { // SETcc
4518 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
4519 fprintf(fout, " %s = %s;", buf2, buf1);
4520 }
4521 else {
4522 ferr(po, "unhandled conditional op\n");
4523 }
4524 }
4525
4526 pfomask = po->pfomask;
4527
4528 if (po->flags & (OPF_REPZ|OPF_REPNZ)) {
4529 struct parsed_opr opr = {0,};
4530 opr.type = OPT_REG;
4531 opr.reg = xCX;
4532 opr.lmod = OPLM_DWORD;
4533 ret = try_resolve_const(i, &opr, opcnt * 7 + i, &uval);
4534
4535 if (ret != 1 || uval == 0) {
4536 // we need initial flags for ecx=0 case..
4537 if (i > 0 && ops[i - 1].op == OP_XOR
4538 && IS(ops[i - 1].operand[0].name,
4539 ops[i - 1].operand[1].name))
4540 {
4541 fprintf(fout, " cond_z = ");
4542 if (pfomask & (1 << PFO_C))
4543 fprintf(fout, "cond_c = ");
4544 fprintf(fout, "0;\n");
4545 }
4546 else if (last_arith_dst != NULL) {
4547 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
4548 out_test_for_cc(buf1, sizeof(buf1), po, PFO_Z, 0,
4549 last_arith_dst->lmod, buf3);
4550 fprintf(fout, " cond_z = %s;\n", buf1);
4551 }
4552 else
4553 ferr(po, "missing initial ZF\n");
4554 }
4555 }
4556
4557 switch (po->op)
4558 {
4559 case OP_MOV:
4560 assert_operand_cnt(2);
4561 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4562 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4563 default_cast_to(buf3, sizeof(buf3), &po->operand[0]);
4564 fprintf(fout, " %s = %s;", buf1,
4565 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4566 buf3, 0));
4567 break;
4568
4569 case OP_LEA:
4570 assert_operand_cnt(2);
4571 po->operand[1].lmod = OPLM_DWORD; // always
4572 fprintf(fout, " %s = %s;",
4573 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4574 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4575 NULL, 1));
4576 break;
4577
4578 case OP_MOVZX:
4579 assert_operand_cnt(2);
4580 fprintf(fout, " %s = %s;",
4581 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4582 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4583 break;
4584
4585 case OP_MOVSX:
4586 assert_operand_cnt(2);
4587 switch (po->operand[1].lmod) {
4588 case OPLM_BYTE:
4589 strcpy(buf3, "(s8)");
4590 break;
4591 case OPLM_WORD:
4592 strcpy(buf3, "(s16)");
4593 break;
4594 default:
4595 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
4596 }
4597 fprintf(fout, " %s = %s;",
4598 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4599 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4600 buf3, 0));
4601 break;
4602
4603 case OP_XCHG:
4604 assert_operand_cnt(2);
4605 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4606 fprintf(fout, " tmp = %s;",
4607 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], "", 0));
4608 fprintf(fout, " %s = %s;",
4609 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4610 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4611 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
4612 fprintf(fout, " %s = %stmp;",
4613 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]),
4614 default_cast_to(buf3, sizeof(buf3), &po->operand[1]));
4615 snprintf(g_comment, sizeof(g_comment), "xchg");
4616 break;
4617
4618 case OP_NOT:
4619 assert_operand_cnt(1);
4620 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4621 fprintf(fout, " %s = ~%s;", buf1, buf1);
4622 break;
4623
4624 case OP_CDQ:
4625 assert_operand_cnt(2);
4626 fprintf(fout, " %s = (s32)%s >> 31;",
4627 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4628 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4629 strcpy(g_comment, "cdq");
4630 break;
4631
4632 case OP_LODS:
4633 assert_operand_cnt(3);
4634 if (po->flags & OPF_REP) {
4635 // hmh..
4636 ferr(po, "TODO\n");
4637 }
4638 else {
4639 fprintf(fout, " eax = %sesi; esi %c= %d;",
4640 lmod_cast_u_ptr(po, po->operand[0].lmod),
4641 (po->flags & OPF_DF) ? '-' : '+',
4642 lmod_bytes(po, po->operand[0].lmod));
4643 strcpy(g_comment, "lods");
4644 }
4645 break;
4646
4647 case OP_STOS:
4648 assert_operand_cnt(3);
4649 if (po->flags & OPF_REP) {
4650 fprintf(fout, " for (; ecx != 0; ecx--, edi %c= %d)\n",
4651 (po->flags & OPF_DF) ? '-' : '+',
4652 lmod_bytes(po, po->operand[0].lmod));
4653 fprintf(fout, " %sedi = eax;",
4654 lmod_cast_u_ptr(po, po->operand[0].lmod));
4655 strcpy(g_comment, "rep stos");
4656 }
4657 else {
4658 fprintf(fout, " %sedi = eax; edi %c= %d;",
4659 lmod_cast_u_ptr(po, po->operand[0].lmod),
4660 (po->flags & OPF_DF) ? '-' : '+',
4661 lmod_bytes(po, po->operand[0].lmod));
4662 strcpy(g_comment, "stos");
4663 }
4664 break;
4665
4666 case OP_MOVS:
4667 assert_operand_cnt(3);
4668 j = lmod_bytes(po, po->operand[0].lmod);
4669 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
4670 l = (po->flags & OPF_DF) ? '-' : '+';
4671 if (po->flags & OPF_REP) {
4672 fprintf(fout,
4673 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n",
4674 l, j, l, j);
4675 fprintf(fout,
4676 " %sedi = %sesi;", buf1, buf1);
4677 strcpy(g_comment, "rep movs");
4678 }
4679 else {
4680 fprintf(fout, " %sedi = %sesi; edi %c= %d; esi %c= %d;",
4681 buf1, buf1, l, j, l, j);
4682 strcpy(g_comment, "movs");
4683 }
4684 break;
4685
4686 case OP_CMPS:
4687 // repe ~ repeat while ZF=1
4688 assert_operand_cnt(3);
4689 j = lmod_bytes(po, po->operand[0].lmod);
4690 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
4691 l = (po->flags & OPF_DF) ? '-' : '+';
4692 if (po->flags & OPF_REP) {
4693 fprintf(fout,
4694 " for (; ecx != 0; ecx--) {\n");
4695 if (pfomask & (1 << PFO_C)) {
4696 // ugh..
4697 fprintf(fout,
4698 " cond_c = %sesi < %sedi;\n", buf1, buf1);
4699 pfomask &= ~(1 << PFO_C);
4700 }
4701 fprintf(fout,
4702 " cond_z = (%sesi == %sedi); esi %c= %d, edi %c= %d;\n",
4703 buf1, buf1, l, j, l, j);
4704 fprintf(fout,
4705 " if (cond_z %s 0) break;\n",
4706 (po->flags & OPF_REPZ) ? "==" : "!=");
4707 fprintf(fout,
4708 " }");
4709 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
4710 (po->flags & OPF_REPZ) ? "e" : "ne");
4711 }
4712 else {
4713 fprintf(fout,
4714 " cond_z = (%sesi == %sedi); esi %c= %d; edi %c= %d;",
4715 buf1, buf1, l, j, l, j);
4716 strcpy(g_comment, "cmps");
4717 }
4718 pfomask &= ~(1 << PFO_Z);
4719 last_arith_dst = NULL;
4720 delayed_flag_op = NULL;
4721 break;
4722
4723 case OP_SCAS:
4724 // only does ZF (for now)
4725 // repe ~ repeat while ZF=1
4726 assert_operand_cnt(3);
4727 j = lmod_bytes(po, po->operand[0].lmod);
4728 l = (po->flags & OPF_DF) ? '-' : '+';
4729 if (po->flags & OPF_REP) {
4730 fprintf(fout,
4731 " for (; ecx != 0; ecx--) {\n");
4732 fprintf(fout,
4733 " cond_z = (%seax == %sedi); edi %c= %d;\n",
4734 lmod_cast_u(po, po->operand[0].lmod),
4735 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
4736 fprintf(fout,
4737 " if (cond_z %s 0) break;\n",
4738 (po->flags & OPF_REPZ) ? "==" : "!=");
4739 fprintf(fout,
4740 " }");
4741 snprintf(g_comment, sizeof(g_comment), "rep%s scas",
4742 (po->flags & OPF_REPZ) ? "e" : "ne");
4743 }
4744 else {
4745 fprintf(fout, " cond_z = (%seax == %sedi); edi %c= %d;",
4746 lmod_cast_u(po, po->operand[0].lmod),
4747 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
4748 strcpy(g_comment, "scas");
4749 }
4750 pfomask &= ~(1 << PFO_Z);
4751 last_arith_dst = NULL;
4752 delayed_flag_op = NULL;
4753 break;
4754
4755 // arithmetic w/flags
4756 case OP_AND:
4757 case OP_OR:
4758 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4759 // fallthrough
4760 dualop_arith:
4761 assert_operand_cnt(2);
4762 fprintf(fout, " %s %s= %s;",
4763 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4764 op_to_c(po),
4765 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4766 output_std_flags(fout, po, &pfomask, buf1);
4767 last_arith_dst = &po->operand[0];
4768 delayed_flag_op = NULL;
4769 break;
4770
4771 case OP_SHL:
4772 case OP_SHR:
4773 assert_operand_cnt(2);
4774 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4775 if (pfomask & (1 << PFO_C)) {
4776 if (po->operand[1].type == OPT_CONST) {
4777 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4778 j = po->operand[1].val;
4779 j %= l;
4780 if (j != 0) {
4781 if (po->op == OP_SHL)
4782 j = l - j;
4783 else
4784 j -= 1;
4785 fprintf(fout, " cond_c = (%s >> %d) & 1;\n",
4786 buf1, j);
4787 }
4788 else
4789 ferr(po, "zero shift?\n");
4790 }
4791 else
4792 ferr(po, "TODO\n");
4793 pfomask &= ~(1 << PFO_C);
4794 }
4795 fprintf(fout, " %s %s= %s;", buf1, op_to_c(po),
4796 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4797 output_std_flags(fout, po, &pfomask, buf1);
4798 last_arith_dst = &po->operand[0];
4799 delayed_flag_op = NULL;
4800 break;
4801
4802 case OP_SAR:
4803 assert_operand_cnt(2);
4804 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4805 fprintf(fout, " %s = %s%s >> %s;", buf1,
4806 lmod_cast_s(po, po->operand[0].lmod), buf1,
4807 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4808 output_std_flags(fout, po, &pfomask, buf1);
4809 last_arith_dst = &po->operand[0];
4810 delayed_flag_op = NULL;
4811 break;
4812
4813 case OP_SHRD:
4814 assert_operand_cnt(3);
4815 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4816 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4817 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4818 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4819 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[2]);
4820 fprintf(fout, " %s >>= %s; %s |= %s << (%d - %s);",
4821 buf1, buf3, buf1, buf2, l, buf3);
4822 strcpy(g_comment, "shrd");
4823 output_std_flags(fout, po, &pfomask, buf1);
4824 last_arith_dst = &po->operand[0];
4825 delayed_flag_op = NULL;
4826 break;
4827
4828 case OP_ROL:
4829 case OP_ROR:
4830 assert_operand_cnt(2);
4831 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4832 if (po->operand[1].type == OPT_CONST) {
4833 j = po->operand[1].val;
4834 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
4835 fprintf(fout, po->op == OP_ROL ?
4836 " %s = (%s << %d) | (%s >> %d);" :
4837 " %s = (%s >> %d) | (%s << %d);",
4838 buf1, buf1, j, buf1,
4839 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
4840 }
4841 else
4842 ferr(po, "TODO\n");
4843 output_std_flags(fout, po, &pfomask, buf1);
4844 last_arith_dst = &po->operand[0];
4845 delayed_flag_op = NULL;
4846 break;
4847
4848 case OP_RCL:
4849 case OP_RCR:
4850 assert_operand_cnt(2);
4851 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4852 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4853 if (po->operand[1].type == OPT_CONST) {
4854 j = po->operand[1].val % l;
4855 if (j == 0)
4856 ferr(po, "zero rotate\n");
4857 fprintf(fout, " tmp = (%s >> %d) & 1;\n",
4858 buf1, (po->op == OP_RCL) ? (l - j) : (j - 1));
4859 if (po->op == OP_RCL) {
4860 fprintf(fout,
4861 " %s = (%s << %d) | (cond_c << %d)",
4862 buf1, buf1, j, j - 1);
4863 if (j != 1)
4864 fprintf(fout, " | (%s >> %d)", buf1, l + 1 - j);
4865 }
4866 else {
4867 fprintf(fout,
4868 " %s = (%s >> %d) | (cond_c << %d)",
4869 buf1, buf1, j, l - j);
4870 if (j != 1)
4871 fprintf(fout, " | (%s << %d)", buf1, l + 1 - j);
4872 }
4873 fprintf(fout, ";\n");
4874 fprintf(fout, " cond_c = tmp;");
4875 }
4876 else
4877 ferr(po, "TODO\n");
4878 strcpy(g_comment, (po->op == OP_RCL) ? "rcl" : "rcr");
4879 output_std_flags(fout, po, &pfomask, buf1);
4880 last_arith_dst = &po->operand[0];
4881 delayed_flag_op = NULL;
4882 break;
4883
4884 case OP_XOR:
4885 assert_operand_cnt(2);
4886 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4887 if (IS(opr_name(po, 0), opr_name(po, 1))) {
4888 // special case for XOR
4889 if (pfomask & (1 << PFO_BE)) { // weird, but it happens..
4890 fprintf(fout, " cond_be = 1;\n");
4891 pfomask &= ~(1 << PFO_BE);
4892 }
4893 fprintf(fout, " %s = 0;",
4894 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4895 last_arith_dst = &po->operand[0];
4896 delayed_flag_op = NULL;
4897 break;
4898 }
4899 goto dualop_arith;
4900
4901 case OP_ADD:
4902 assert_operand_cnt(2);
4903 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4904 if (pfomask & (1 << PFO_C)) {
4905 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4906 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4907 if (po->operand[0].lmod == OPLM_DWORD) {
4908 fprintf(fout, " tmp64 = (u64)%s + %s;\n", buf1, buf2);
4909 fprintf(fout, " cond_c = tmp64 >> 32;\n");
4910 fprintf(fout, " %s = (u32)tmp64;",
4911 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4912 strcat(g_comment, "add64");
4913 }
4914 else {
4915 fprintf(fout, " cond_c = ((u32)%s + %s) >> %d;\n",
4916 buf1, buf2, lmod_bytes(po, po->operand[0].lmod) * 8);
4917 fprintf(fout, " %s += %s;",
4918 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4919 buf2);
4920 }
4921 pfomask &= ~(1 << PFO_C);
4922 output_std_flags(fout, po, &pfomask, buf1);
4923 last_arith_dst = &po->operand[0];
4924 delayed_flag_op = NULL;
4925 break;
4926 }
4927 goto dualop_arith;
4928
4929 case OP_SUB:
4930 assert_operand_cnt(2);
4931 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4932 if (pfomask & ~((1 << PFO_Z) | (1 << PFO_S))) {
4933 for (j = 0; j <= PFO_LE; j++) {
4934 if (!(pfomask & (1 << j)))
4935 continue;
4936 if (j == PFO_Z || j == PFO_S)
4937 continue;
4938
4939 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
4940 fprintf(fout, " cond_%s = %s;\n",
4941 parsed_flag_op_names[j], buf1);
4942 pfomask &= ~(1 << j);
4943 }
4944 }
4945 goto dualop_arith;
4946
4947 case OP_ADC:
4948 case OP_SBB:
4949 assert_operand_cnt(2);
4950 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4951 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4952 if (po->op == OP_SBB
4953 && IS(po->operand[0].name, po->operand[1].name))
4954 {
4955 // avoid use of unitialized var
4956 fprintf(fout, " %s = -cond_c;", buf1);
4957 // carry remains what it was
4958 pfomask &= ~(1 << PFO_C);
4959 }
4960 else {
4961 fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po),
4962 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4963 }
4964 output_std_flags(fout, po, &pfomask, buf1);
4965 last_arith_dst = &po->operand[0];
4966 delayed_flag_op = NULL;
4967 break;
4968
4969 case OP_BSF:
4970 assert_operand_cnt(2);
4971 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4972 fprintf(fout, " %s = %s ? __builtin_ffs(%s) - 1 : 0;",
4973 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4974 buf2, buf2);
4975 output_std_flags(fout, po, &pfomask, buf1);
4976 last_arith_dst = &po->operand[0];
4977 delayed_flag_op = NULL;
4978 strcat(g_comment, "bsf");
4979 break;
4980
4981 case OP_DEC:
4982 if (pfomask & ~(PFOB_S | PFOB_S | PFOB_C)) {
4983 for (j = 0; j <= PFO_LE; j++) {
4984 if (!(pfomask & (1 << j)))
4985 continue;
4986 if (j == PFO_Z || j == PFO_S || j == PFO_C)
4987 continue;
4988
4989 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
4990 fprintf(fout, " cond_%s = %s;\n",
4991 parsed_flag_op_names[j], buf1);
4992 pfomask &= ~(1 << j);
4993 }
4994 }
4995 // fallthrough
4996
4997 case OP_INC:
4998 if (pfomask & (1 << PFO_C))
4999 // carry is unaffected by inc/dec.. wtf?
5000 ferr(po, "carry propagation needed\n");
5001
5002 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5003 if (po->operand[0].type == OPT_REG) {
5004 strcpy(buf2, po->op == OP_INC ? "++" : "--");
5005 fprintf(fout, " %s%s;", buf1, buf2);
5006 }
5007 else {
5008 strcpy(buf2, po->op == OP_INC ? "+" : "-");
5009 fprintf(fout, " %s %s= 1;", buf1, buf2);
5010 }
5011 output_std_flags(fout, po, &pfomask, buf1);
5012 last_arith_dst = &po->operand[0];
5013 delayed_flag_op = NULL;
5014 break;
5015
5016 case OP_NEG:
5017 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5018 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
5019 fprintf(fout, " %s = -%s%s;", buf1,
5020 lmod_cast_s(po, po->operand[0].lmod), buf2);
5021 last_arith_dst = &po->operand[0];
5022 delayed_flag_op = NULL;
5023 if (pfomask & (1 << PFO_C)) {
5024 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
5025 pfomask &= ~(1 << PFO_C);
5026 }
5027 break;
5028
5029 case OP_IMUL:
5030 if (po->operand_cnt == 2) {
5031 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5032 goto dualop_arith;
5033 }
5034 if (po->operand_cnt == 3)
5035 ferr(po, "TODO imul3\n");
5036 // fallthrough
5037 case OP_MUL:
5038 assert_operand_cnt(1);
5039 switch (po->operand[0].lmod) {
5040 case OPLM_DWORD:
5041 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
5042 fprintf(fout, " tmp64 = %seax * %s%s;\n", buf1, buf1,
5043 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
5044 fprintf(fout, " edx = tmp64 >> 32;\n");
5045 fprintf(fout, " eax = tmp64;");
5046 break;
5047 case OPLM_BYTE:
5048 strcpy(buf1, po->op == OP_IMUL ? "(s16)(s8)" : "(u16)(u8)");
5049 fprintf(fout, " LOWORD(eax) = %seax * %s;", buf1,
5050 out_src_opr(buf2, sizeof(buf2), po, &po->operand[0],
5051 buf1, 0));
5052 break;
5053 default:
5054 ferr(po, "TODO: unhandled mul type\n");
5055 break;
5056 }
5057 last_arith_dst = NULL;
5058 delayed_flag_op = NULL;
5059 break;
5060
5061 case OP_DIV:
5062 case OP_IDIV:
5063 assert_operand_cnt(1);
5064 if (po->operand[0].lmod != OPLM_DWORD)
5065 ferr(po, "unhandled lmod %d\n", po->operand[0].lmod);
5066
5067 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
5068 strcpy(buf2, lmod_cast(po, po->operand[0].lmod,
5069 po->op == OP_IDIV));
5070 switch (po->operand[0].lmod) {
5071 case OPLM_DWORD:
5072 if (po->flags & OPF_32BIT)
5073 snprintf(buf3, sizeof(buf3), "%seax", buf2);
5074 else {
5075 fprintf(fout, " tmp64 = ((u64)edx << 32) | eax;\n");
5076 snprintf(buf3, sizeof(buf3), "%stmp64",
5077 (po->op == OP_IDIV) ? "(s64)" : "");
5078 }
5079 if (po->operand[0].type == OPT_REG
5080 && po->operand[0].reg == xDX)
5081 {
5082 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
5083 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
5084 }
5085 else {
5086 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
5087 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
5088 }
5089 break;
5090 default:
5091 ferr(po, "unhandled division type\n");
5092 }
5093 last_arith_dst = NULL;
5094 delayed_flag_op = NULL;
5095 break;
5096
5097 case OP_TEST:
5098 case OP_CMP:
5099 propagate_lmod(po, &po->operand[0], &po->operand[1]);
5100 if (pfomask != 0) {
5101 for (j = 0; j < 8; j++) {
5102 if (pfomask & (1 << j)) {
5103 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
5104 fprintf(fout, " cond_%s = %s;",
5105 parsed_flag_op_names[j], buf1);
5106 }
5107 }
5108 pfomask = 0;
5109 }
5110 else
5111 no_output = 1;
5112 last_arith_dst = NULL;
5113 delayed_flag_op = po;
5114 break;
5115
5116 case OP_SCC:
5117 // SETcc - should already be handled
5118 break;
5119
5120 // note: we reuse OP_Jcc for SETcc, only flags differ
5121 case OP_JCC:
5122 fprintf(fout, "\n goto %s;", po->operand[0].name);
5123 break;
5124
5125 case OP_JECXZ:
5126 fprintf(fout, " if (ecx == 0)\n");
5127 fprintf(fout, " goto %s;", po->operand[0].name);
5128 strcat(g_comment, "jecxz");
5129 break;
5130
5131 case OP_JMP:
5132 assert_operand_cnt(1);
5133 last_arith_dst = NULL;
5134 delayed_flag_op = NULL;
5135
5136 if (po->operand[0].type == OPT_REGMEM) {
5137 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
5138 buf1, buf2);
5139 if (ret != 2)
5140 ferr(po, "parse failure for jmp '%s'\n",
5141 po->operand[0].name);
5142 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
5143 break;
5144 }
5145 else if (po->operand[0].type != OPT_LABEL)
5146 ferr(po, "unhandled jmp type\n");
5147
5148 fprintf(fout, " goto %s;", po->operand[0].name);
5149 break;
5150
5151 case OP_CALL:
5152 assert_operand_cnt(1);
5153 pp = po->pp;
5154 my_assert_not(pp, NULL);
5155
5156 strcpy(buf3, " ");
5157 if (po->flags & OPF_CC) {
5158 // we treat conditional branch to another func
5159 // (yes such code exists..) as conditional tailcall
5160 strcat(buf3, " ");
5161 fprintf(fout, " {\n");
5162 }
5163
5164 if (pp->is_fptr && !pp->is_arg) {
5165 fprintf(fout, "%s%s = %s;\n", buf3, pp->name,
5166 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
5167 "(void *)", 0));
5168 if (pp->is_unresolved)
5169 fprintf(fout, "%sunresolved_call(\"%s:%d\", %s);\n",
5170 buf3, asmfn, po->asmln, pp->name);
5171 }
5172
5173 fprintf(fout, "%s", buf3);
5174 if (strstr(pp->ret_type.name, "int64")) {
5175 if (po->flags & OPF_TAIL)
5176 ferr(po, "int64 and tail?\n");
5177 fprintf(fout, "tmp64 = ");
5178 }
5179 else if (!IS(pp->ret_type.name, "void")) {
5180 if (po->flags & OPF_TAIL) {
5181 if (!IS(g_func_pp->ret_type.name, "void")) {
5182 fprintf(fout, "return ");
5183 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
5184 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
5185 }
5186 }
5187 else if (regmask & (1 << xAX)) {
5188 fprintf(fout, "eax = ");
5189 if (pp->ret_type.is_ptr)
5190 fprintf(fout, "(u32)");
5191 }
5192 }
5193
5194 if (pp->name[0] == 0)
5195 ferr(po, "missing pp->name\n");
5196 fprintf(fout, "%s%s(", pp->name,
5197 pp->has_structarg ? "_sa" : "");
5198
5199 if (po->flags & OPF_ATAIL) {
5200 if (pp->argc_stack != g_func_pp->argc_stack
5201 || (pp->argc_stack > 0
5202 && pp->is_stdcall != g_func_pp->is_stdcall))
5203 ferr(po, "incompatible tailcall\n");
5204 if (g_func_pp->has_retreg)
5205 ferr(po, "TODO: retreg+tailcall\n");
5206
5207 for (arg = j = 0; arg < pp->argc; arg++) {
5208 if (arg > 0)
5209 fprintf(fout, ", ");
5210
5211 cast[0] = 0;
5212 if (pp->arg[arg].type.is_ptr)
5213 snprintf(cast, sizeof(cast), "(%s)",
5214 pp->arg[arg].type.name);
5215
5216 if (pp->arg[arg].reg != NULL) {
5217 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
5218 continue;
5219 }
5220 // stack arg
5221 for (; j < g_func_pp->argc; j++)
5222 if (g_func_pp->arg[j].reg == NULL)
5223 break;
5224 fprintf(fout, "%sa%d", cast, j + 1);
5225 j++;
5226 }
5227 }
5228 else {
5229 for (arg = 0; arg < pp->argc; arg++) {
5230 if (arg > 0)
5231 fprintf(fout, ", ");
5232
5233 cast[0] = 0;
5234 if (pp->arg[arg].type.is_ptr)
5235 snprintf(cast, sizeof(cast), "(%s)",
5236 pp->arg[arg].type.name);
5237
5238 if (pp->arg[arg].reg != NULL) {
5239 if (pp->arg[arg].type.is_retreg)
5240 fprintf(fout, "&%s", pp->arg[arg].reg);
5241 else
5242 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
5243 continue;
5244 }
5245
5246 // stack arg
5247 tmp_op = pp->arg[arg].datap;
5248 if (tmp_op == NULL)
5249 ferr(po, "parsed_op missing for arg%d\n", arg);
5250
5251 if (tmp_op->flags & OPF_VAPUSH) {
5252 fprintf(fout, "ap");
5253 }
5254 else if (tmp_op->p_argpass != 0) {
5255 fprintf(fout, "a%d", tmp_op->p_argpass);
5256 }
5257 else if (tmp_op->p_argnum != 0) {
5258 fprintf(fout, "%s%s", cast,
5259 saved_arg_name(buf1, sizeof(buf1),
5260 tmp_op->p_arggrp, tmp_op->p_argnum));
5261 }
5262 else {
5263 fprintf(fout, "%s",
5264 out_src_opr(buf1, sizeof(buf1),
5265 tmp_op, &tmp_op->operand[0], cast, 0));
5266 }
5267 }
5268 }
5269 fprintf(fout, ");");
5270
5271 if (strstr(pp->ret_type.name, "int64")) {
5272 fprintf(fout, "\n");
5273 fprintf(fout, "%sedx = tmp64 >> 32;\n", buf3);
5274 fprintf(fout, "%seax = tmp64;", buf3);
5275 }
5276
5277 if (pp->is_unresolved) {
5278 snprintf(buf2, sizeof(buf2), " unresolved %dreg",
5279 pp->argc_reg);
5280 strcat(g_comment, buf2);
5281 }
5282
5283 if (po->flags & OPF_TAIL) {
5284 ret = 0;
5285 if (i == opcnt - 1 || pp->is_noreturn)
5286 ret = 0;
5287 else if (IS(pp->ret_type.name, "void"))
5288 ret = 1;
5289 else if (IS(g_func_pp->ret_type.name, "void"))
5290 ret = 1;
5291 // else already handled as 'return f()'
5292
5293 if (ret) {
5294 if (!IS(g_func_pp->ret_type.name, "void")) {
5295 ferr(po, "int func -> void func tailcall?\n");
5296 }
5297 else {
5298 fprintf(fout, "\n%sreturn;", buf3);
5299 strcat(g_comment, " ^ tailcall");
5300 }
5301 }
5302 else
5303 strcat(g_comment, " tailcall");
5304 }
5305 if (pp->is_noreturn)
5306 strcat(g_comment, " noreturn");
5307 if ((po->flags & OPF_ATAIL) && pp->argc_stack > 0)
5308 strcat(g_comment, " argframe");
5309 if (po->flags & OPF_CC)
5310 strcat(g_comment, " cond");
5311
5312 if (po->flags & OPF_CC)
5313 fprintf(fout, "\n }");
5314
5315 delayed_flag_op = NULL;
5316 last_arith_dst = NULL;
5317 break;
5318
5319 case OP_RET:
5320 if (g_func_pp->is_vararg)
5321 fprintf(fout, " va_end(ap);\n");
5322 if (g_func_pp->has_retreg) {
5323 for (arg = 0; arg < g_func_pp->argc; arg++)
5324 if (g_func_pp->arg[arg].type.is_retreg)
5325 fprintf(fout, " *r_%s = %s;\n",
5326 g_func_pp->arg[arg].reg, g_func_pp->arg[arg].reg);
5327 }
5328
5329 if (IS(g_func_pp->ret_type.name, "void")) {
5330 if (i != opcnt - 1 || label_pending)
5331 fprintf(fout, " return;");
5332 }
5333 else if (g_func_pp->ret_type.is_ptr) {
5334 fprintf(fout, " return (%s)eax;",
5335 g_func_pp->ret_type.name);
5336 }
5337 else if (IS(g_func_pp->ret_type.name, "__int64"))
5338 fprintf(fout, " return ((u64)edx << 32) | eax;");
5339 else
5340 fprintf(fout, " return eax;");
5341
5342 last_arith_dst = NULL;
5343 delayed_flag_op = NULL;
5344 break;
5345
5346 case OP_PUSH:
5347 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
5348 if (po->p_argnum != 0) {
5349 // special case - saved func arg
5350 fprintf(fout, " %s = %s;",
5351 saved_arg_name(buf2, sizeof(buf2),
5352 po->p_arggrp, po->p_argnum), buf1);
5353 break;
5354 }
5355 else if (po->flags & OPF_RSAVE) {
5356 fprintf(fout, " s_%s = %s;", buf1, buf1);
5357 break;
5358 }
5359 else if (g_func_pp->is_userstack) {
5360 fprintf(fout, " *(--esp) = %s;", buf1);
5361 break;
5362 }
5363 if (!(g_ida_func_attr & IDAFA_NORETURN))
5364 ferr(po, "stray push encountered\n");
5365 no_output = 1;
5366 break;
5367
5368 case OP_POP:
5369 if (po->flags & OPF_RSAVE) {
5370 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5371 fprintf(fout, " %s = s_%s;", buf1, buf1);
5372 break;
5373 }
5374 else if (po->datap != NULL) {
5375 // push/pop pair
5376 tmp_op = po->datap;
5377 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5378 fprintf(fout, " %s = %s;", buf1,
5379 out_src_opr(buf2, sizeof(buf2),
5380 tmp_op, &tmp_op->operand[0],
5381 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
5382 break;
5383 }
5384 else if (g_func_pp->is_userstack) {
5385 fprintf(fout, " %s = *esp++;",
5386 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
5387 break;
5388 }
5389 else
5390 ferr(po, "stray pop encountered\n");
5391 break;
5392
5393 case OP_NOP:
5394 no_output = 1;
5395 break;
5396
5397 // mmx
5398 case OP_EMMS:
5399 strcpy(g_comment, "(emms)");
5400 break;
5401
5402 default:
5403 no_output = 1;
5404 ferr(po, "unhandled op type %d, flags %x\n",
5405 po->op, po->flags);
5406 break;
5407 }
5408
5409 if (g_comment[0] != 0) {
5410 char *p = g_comment;
5411 while (my_isblank(*p))
5412 p++;
5413 fprintf(fout, " // %s", p);
5414 g_comment[0] = 0;
5415 no_output = 0;
5416 }
5417 if (!no_output)
5418 fprintf(fout, "\n");
5419
5420 // some sanity checking
5421 if (po->flags & OPF_REP) {
5422 if (po->op != OP_STOS && po->op != OP_MOVS
5423 && po->op != OP_CMPS && po->op != OP_SCAS)
5424 ferr(po, "unexpected rep\n");
5425 if (!(po->flags & (OPF_REPZ|OPF_REPNZ))
5426 && (po->op == OP_CMPS || po->op == OP_SCAS))
5427 ferr(po, "cmps/scas with plain rep\n");
5428 }
5429 if ((po->flags & (OPF_REPZ|OPF_REPNZ))
5430 && po->op != OP_CMPS && po->op != OP_SCAS)
5431 ferr(po, "unexpected repz/repnz\n");
5432
5433 if (pfomask != 0)
5434 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
5435
5436 // see is delayed flag stuff is still valid
5437 if (delayed_flag_op != NULL && delayed_flag_op != po) {
5438 if (is_any_opr_modified(delayed_flag_op, po, 0))
5439 delayed_flag_op = NULL;
5440 }
5441
5442 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
5443 if (is_opr_modified(last_arith_dst, po))
5444 last_arith_dst = NULL;
5445 }
5446
5447 label_pending = 0;
5448 }
5449
5450 if (g_stack_fsz && !g_stack_frame_used)
5451 fprintf(fout, " (void)sf;\n");
5452
5453 fprintf(fout, "}\n\n");
5454
5455 gen_x_cleanup(opcnt);
5456}
5457
5458static void gen_x_cleanup(int opcnt)
5459{
5460 int i;
5461
5462 for (i = 0; i < opcnt; i++) {
5463 struct label_ref *lr, *lr_del;
5464
5465 lr = g_label_refs[i].next;
5466 while (lr != NULL) {
5467 lr_del = lr;
5468 lr = lr->next;
5469 free(lr_del);
5470 }
5471 g_label_refs[i].i = -1;
5472 g_label_refs[i].next = NULL;
5473
5474 if (ops[i].op == OP_CALL) {
5475 if (ops[i].pp)
5476 proto_release(ops[i].pp);
5477 }
5478 }
5479 g_func_pp = NULL;
5480}
5481
5482struct func_proto_dep;
5483
5484struct func_prototype {
5485 char name[NAMELEN];
5486 int id;
5487 int argc_stack;
5488 int regmask_dep;
5489 int has_ret:3; // -1, 0, 1: unresolved, no, yes
5490 unsigned int dep_resolved:1;
5491 unsigned int is_stdcall:1;
5492 struct func_proto_dep *dep_func;
5493 int dep_func_cnt;
5494 const struct parsed_proto *pp; // seed pp, if any
5495};
5496
5497struct func_proto_dep {
5498 char *name;
5499 struct func_prototype *proto;
5500 int regmask_live; // .. at the time of call
5501 unsigned int ret_dep:1; // return from this is caller's return
5502};
5503
5504static struct func_prototype *hg_fp;
5505static int hg_fp_cnt;
5506
5507static struct scanned_var {
5508 char name[NAMELEN];
5509 enum opr_lenmod lmod;
5510 unsigned int is_seeded:1;
5511 unsigned int is_c_str:1;
5512} *hg_vars;
5513static int hg_var_cnt;
5514
5515static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
5516 int count);
5517
5518static struct func_proto_dep *hg_fp_find_dep(struct func_prototype *fp,
5519 const char *name)
5520{
5521 int i;
5522
5523 for (i = 0; i < fp->dep_func_cnt; i++)
5524 if (IS(fp->dep_func[i].name, name))
5525 return &fp->dep_func[i];
5526
5527 return NULL;
5528}
5529
5530static void hg_fp_add_dep(struct func_prototype *fp, const char *name)
5531{
5532 // is it a dupe?
5533 if (hg_fp_find_dep(fp, name))
5534 return;
5535
5536 if ((fp->dep_func_cnt & 0xff) == 0) {
5537 fp->dep_func = realloc(fp->dep_func,
5538 sizeof(fp->dep_func[0]) * (fp->dep_func_cnt + 0x100));
5539 my_assert_not(fp->dep_func, NULL);
5540 memset(&fp->dep_func[fp->dep_func_cnt], 0,
5541 sizeof(fp->dep_func[0]) * 0x100);
5542 }
5543 fp->dep_func[fp->dep_func_cnt].name = strdup(name);
5544 fp->dep_func_cnt++;
5545}
5546
5547static int hg_fp_cmp_name(const void *p1_, const void *p2_)
5548{
5549 const struct func_prototype *p1 = p1_, *p2 = p2_;
5550 return strcmp(p1->name, p2->name);
5551}
5552
5553#if 0
5554static int hg_fp_cmp_id(const void *p1_, const void *p2_)
5555{
5556 const struct func_prototype *p1 = p1_, *p2 = p2_;
5557 return p1->id - p2->id;
5558}
5559#endif
5560
5561// recursive register dep pass
5562// - track saved regs (part 2)
5563// - try to figure out arg-regs
5564// - calculate reg deps
5565static void gen_hdr_dep_pass(int i, int opcnt, unsigned char *cbits,
5566 struct func_prototype *fp, int regmask_save, int regmask_dst,
5567 int *regmask_dep, int *has_ret)
5568{
5569 struct func_proto_dep *dep;
5570 struct parsed_op *po;
5571 int from_caller = 0;
5572 int depth;
5573 int j, l;
5574 int reg;
5575 int ret;
5576
5577 for (; i < opcnt; i++)
5578 {
5579 if (cbits[i >> 3] & (1 << (i & 7)))
5580 return;
5581 cbits[i >> 3] |= (1 << (i & 7));
5582
5583 po = &ops[i];
5584
5585 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
5586 if (po->btj != NULL) {
5587 // jumptable
5588 for (j = 0; j < po->btj->count; j++) {
5589 gen_hdr_dep_pass(po->btj->d[j].bt_i, opcnt, cbits, fp,
5590 regmask_save, regmask_dst, regmask_dep, has_ret);
5591 }
5592 return;
5593 }
5594
5595 if (po->bt_i < 0) {
5596 ferr(po, "dead branch\n");
5597 return;
5598 }
5599
5600 if (po->flags & OPF_CJMP) {
5601 gen_hdr_dep_pass(po->bt_i, opcnt, cbits, fp,
5602 regmask_save, regmask_dst, regmask_dep, has_ret);
5603 }
5604 else {
5605 i = po->bt_i - 1;
5606 }
5607 continue;
5608 }
5609
5610 if (po->flags & OPF_FARG)
5611 /* (just calculate register deps) */;
5612 else if (po->op == OP_PUSH && po->operand[0].type == OPT_REG)
5613 {
5614 reg = po->operand[0].reg;
5615 if (reg < 0)
5616 ferr(po, "reg not set for push?\n");
5617
5618 if (po->flags & OPF_RSAVE) {
5619 regmask_save |= 1 << reg;
5620 continue;
5621 }
5622 if (po->flags & OPF_DONE)
5623 continue;
5624
5625 depth = 0;
5626 ret = scan_for_pop(i + 1, opcnt,
5627 po->operand[0].name, i + opcnt * 2, 0, &depth, 0);
5628 if (ret == 1) {
5629 regmask_save |= 1 << reg;
5630 po->flags |= OPF_RMD;
5631 scan_for_pop(i + 1, opcnt,
5632 po->operand[0].name, i + opcnt * 3, 0, &depth, 1);
5633 continue;
5634 }
5635 }
5636 else if (po->flags & OPF_RMD)
5637 continue;
5638 else if (po->op == OP_CALL) {
5639 po->regmask_dst |= 1 << xAX;
5640
5641 dep = hg_fp_find_dep(fp, po->operand[0].name);
5642 if (dep != NULL)
5643 dep->regmask_live = regmask_save | regmask_dst;
5644 }
5645 else if (po->op == OP_RET) {
5646 if (po->operand_cnt > 0) {
5647 fp->is_stdcall = 1;
5648 if (fp->argc_stack >= 0
5649 && fp->argc_stack != po->operand[0].val / 4)
5650 ferr(po, "ret mismatch? (%d)\n", fp->argc_stack * 4);
5651 fp->argc_stack = po->operand[0].val / 4;
5652 }
5653 }
5654
5655 if (*has_ret != 0 && (po->flags & OPF_TAIL)) {
5656 if (po->op == OP_CALL) {
5657 j = i;
5658 ret = 1;
5659 }
5660 else {
5661 struct parsed_opr opr = { 0, };
5662 opr.type = OPT_REG;
5663 opr.reg = xAX;
5664 j = -1;
5665 from_caller = 0;
5666 ret = resolve_origin(i, &opr, i + opcnt * 4, &j, &from_caller);
5667 }
5668
5669 if (ret == -1 && from_caller) {
5670 // unresolved eax - probably void func
5671 *has_ret = 0;
5672 }
5673 else {
5674 if (ops[j].op == OP_CALL) {
5675 dep = hg_fp_find_dep(fp, po->operand[0].name);
5676 if (dep != NULL)
5677 dep->ret_dep = 1;
5678 else
5679 *has_ret = 1;
5680 }
5681 else
5682 *has_ret = 1;
5683 }
5684 }
5685
5686 l = regmask_save | regmask_dst;
5687 if (g_bp_frame && !(po->flags & OPF_EBP_S))
5688 l |= 1 << xBP;
5689
5690 l = po->regmask_src & ~l;
5691#if 0
5692 if (l)
5693 fnote(po, "dep |= %04x, dst %04x, save %04x (f %x)\n",
5694 l, regmask_dst, regmask_save, po->flags);
5695#endif
5696 *regmask_dep |= l;
5697 regmask_dst |= po->regmask_dst;
5698
5699 if (po->flags & OPF_TAIL)
5700 return;
5701 }
5702}
5703
5704static void gen_hdr(const char *funcn, int opcnt)
5705{
5706 int save_arg_vars[MAX_ARG_GRP] = { 0, };
5707 unsigned char cbits[MAX_OPS / 8];
5708 const struct parsed_proto *pp_c;
5709 struct parsed_proto *pp;
5710 struct func_prototype *fp;
5711 struct parsed_data *pd;
5712 struct parsed_op *po;
5713 const char *tmpname;
5714 int regmask_dummy = 0;
5715 int regmask_dep;
5716 int max_bp_offset = 0;
5717 int has_ret;
5718 int i, j, l, ret;
5719
5720 if ((hg_fp_cnt & 0xff) == 0) {
5721 hg_fp = realloc(hg_fp, sizeof(hg_fp[0]) * (hg_fp_cnt + 0x100));
5722 my_assert_not(hg_fp, NULL);
5723 memset(hg_fp + hg_fp_cnt, 0, sizeof(hg_fp[0]) * 0x100);
5724 }
5725
5726 fp = &hg_fp[hg_fp_cnt];
5727 snprintf(fp->name, sizeof(fp->name), "%s", funcn);
5728 fp->id = hg_fp_cnt;
5729 fp->argc_stack = -1;
5730 hg_fp_cnt++;
5731
5732 // perhaps already in seed header?
5733 fp->pp = proto_parse(g_fhdr, funcn, 1);
5734 if (fp->pp != NULL) {
5735 fp->argc_stack = fp->pp->argc_stack;
5736 fp->is_stdcall = fp->pp->is_stdcall;
5737 fp->regmask_dep = get_pp_arg_regmask(fp->pp);
5738 fp->has_ret = !IS(fp->pp->ret_type.name, "void");
5739 return;
5740 }
5741
5742 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
5743 g_stack_frame_used = 0;
5744
5745 // pass1:
5746 // - handle ebp/esp frame, remove ops related to it
5747 scan_prologue_epilogue(opcnt);
5748
5749 // pass2:
5750 // - collect calls
5751 // - resolve all branches
5752 for (i = 0; i < opcnt; i++)
5753 {
5754 po = &ops[i];
5755 po->bt_i = -1;
5756 po->btj = NULL;
5757
5758 if (po->flags & (OPF_RMD|OPF_DONE))
5759 continue;
5760
5761 if (po->op == OP_CALL) {
5762 tmpname = opr_name(po, 0);
5763 pp = NULL;
5764 if (po->operand[0].type == OPT_LABEL) {
5765 hg_fp_add_dep(fp, tmpname);
5766
5767 // perhaps a call to already known func?
5768 pp_c = proto_parse(g_fhdr, tmpname, 1);
5769 if (pp_c != NULL)
5770 pp = proto_clone(pp_c);
5771 }
5772 else if (po->datap != NULL) {
5773 pp = calloc(1, sizeof(*pp));
5774 my_assert_not(pp, NULL);
5775
5776 ret = parse_protostr(po->datap, pp);
5777 if (ret < 0)
5778 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
5779 free(po->datap);
5780 po->datap = NULL;
5781 }
5782 if (pp != NULL && pp->is_noreturn)
5783 po->flags |= OPF_TAIL;
5784
5785 po->pp = pp;
5786 continue;
5787 }
5788
5789 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
5790 continue;
5791
5792 if (po->operand[0].type == OPT_REGMEM) {
5793 pd = try_resolve_jumptab(i, opcnt);
5794 if (pd == NULL)
5795 goto tailcall;
5796
5797 po->btj = pd;
5798 continue;
5799 }
5800
5801 for (l = 0; l < opcnt; l++) {
5802 if (g_labels[l] != NULL
5803 && IS(po->operand[0].name, g_labels[l]))
5804 {
5805 add_label_ref(&g_label_refs[l], i);
5806 po->bt_i = l;
5807 break;
5808 }
5809 }
5810
5811 if (po->bt_i != -1 || (po->flags & OPF_RMD))
5812 continue;
5813
5814 if (po->operand[0].type == OPT_LABEL)
5815 // assume tail call
5816 goto tailcall;
5817
5818 ferr(po, "unhandled branch\n");
5819
5820tailcall:
5821 po->op = OP_CALL;
5822 po->flags |= OPF_TAIL;
5823 if (i > 0 && ops[i - 1].op == OP_POP)
5824 po->flags |= OPF_ATAIL;
5825 i--; // reprocess
5826 }
5827
5828 // pass3:
5829 // - remove dead labels
5830 // - handle push <const>/pop pairs
5831 for (i = 0; i < opcnt; i++)
5832 {
5833 if (g_labels[i] != NULL && g_label_refs[i].i == -1) {
5834 free(g_labels[i]);
5835 g_labels[i] = NULL;
5836 }
5837
5838 po = &ops[i];
5839 if (po->flags & (OPF_RMD|OPF_DONE))
5840 continue;
5841
5842 if (po->op == OP_PUSH && po->operand[0].type == OPT_CONST)
5843 scan_for_pop_const(i, opcnt);
5844 }
5845
5846 // pass4:
5847 // - process trivial calls
5848 for (i = 0; i < opcnt; i++)
5849 {
5850 po = &ops[i];
5851 if (po->flags & (OPF_RMD|OPF_DONE))
5852 continue;
5853
5854 if (po->op == OP_CALL)
5855 {
5856 pp = process_call_early(i, opcnt, &j);
5857 if (pp != NULL) {
5858 if (!(po->flags & OPF_ATAIL))
5859 // since we know the args, try to collect them
5860 if (collect_call_args_early(po, i, pp, &regmask_dummy) != 0)
5861 pp = NULL;
5862 }
5863
5864 if (pp != NULL) {
5865 if (j >= 0) {
5866 // commit esp adjust
5867 ops[j].flags |= OPF_RMD;
5868 if (ops[j].op != OP_POP)
5869 patch_esp_adjust(&ops[j], pp->argc_stack * 4);
5870 else
5871 ops[j].flags |= OPF_DONE;
5872 }
5873
5874 po->flags |= OPF_DONE;
5875 }
5876 }
5877 }
5878
5879 // pass5:
5880 // - track saved regs (simple)
5881 // - process calls
5882 for (i = 0; i < opcnt; i++)
5883 {
5884 po = &ops[i];
5885 if (po->flags & (OPF_RMD|OPF_DONE))
5886 continue;
5887
5888 if (po->op == OP_PUSH && po->operand[0].type == OPT_REG)
5889 {
5890 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
5891 if (ret == 0) {
5892 // regmask_save |= 1 << po->operand[0].reg; // do it later
5893 po->flags |= OPF_RSAVE | OPF_RMD | OPF_DONE;
5894 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, OPF_RMD);
5895 }
5896 }
5897 else if (po->op == OP_CALL && !(po->flags & OPF_DONE))
5898 {
5899 pp = process_call(i, opcnt);
5900
5901 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
5902 // since we know the args, collect them
5903 ret = collect_call_args(po, i, pp, &regmask_dummy, save_arg_vars,
5904 i + opcnt * 1);
5905 }
5906 }
5907 }
5908
5909 // pass6
5910 memset(cbits, 0, sizeof(cbits));
5911 regmask_dep = 0;
5912 has_ret = -1;
5913
5914 gen_hdr_dep_pass(0, opcnt, cbits, fp, 0, 0, &regmask_dep, &has_ret);
5915
5916 // find unreachable code - must be fixed in IDA
5917 for (i = 0; i < opcnt; i++)
5918 {
5919 if (cbits[i >> 3] & (1 << (i & 7)))
5920 continue;
5921
5922 if (ops[i].op != OP_NOP)
5923 ferr(&ops[i], "unreachable code\n");
5924 }
5925
5926 if (has_ret == -1 && (regmask_dep & (1 << xAX)))
5927 has_ret = 1;
5928
5929 for (i = 0; i < g_eqcnt; i++) {
5930 if (g_eqs[i].offset > max_bp_offset && g_eqs[i].offset < 4*32)
5931 max_bp_offset = g_eqs[i].offset;
5932 }
5933
5934 if (fp->argc_stack < 0) {
5935 max_bp_offset = (max_bp_offset + 3) & ~3;
5936 fp->argc_stack = max_bp_offset / 4;
5937 if ((g_ida_func_attr & IDAFA_BP_FRAME) && fp->argc_stack > 0)
5938 fp->argc_stack--;
5939 }
5940
5941 fp->regmask_dep = regmask_dep & ~(1 << xSP);
5942 fp->has_ret = has_ret;
5943#if 0
5944 printf("// has_ret %d, regmask_dep %x\n",
5945 fp->has_ret, fp->regmask_dep);
5946 output_hdr_fp(stdout, fp, 1);
5947 if (IS(funcn, "sub_100073FD")) exit(1);
5948#endif
5949
5950 gen_x_cleanup(opcnt);
5951}
5952
5953static void hg_fp_resolve_deps(struct func_prototype *fp)
5954{
5955 struct func_prototype fp_s;
5956 int dep;
5957 int i;
5958
5959 // this thing is recursive, so mark first..
5960 fp->dep_resolved = 1;
5961
5962 for (i = 0; i < fp->dep_func_cnt; i++) {
5963 strcpy(fp_s.name, fp->dep_func[i].name);
5964 fp->dep_func[i].proto = bsearch(&fp_s, hg_fp, hg_fp_cnt,
5965 sizeof(hg_fp[0]), hg_fp_cmp_name);
5966 if (fp->dep_func[i].proto != NULL) {
5967 if (!fp->dep_func[i].proto->dep_resolved)
5968 hg_fp_resolve_deps(fp->dep_func[i].proto);
5969
5970 dep = ~fp->dep_func[i].regmask_live
5971 & fp->dep_func[i].proto->regmask_dep;
5972 fp->regmask_dep |= dep;
5973 // printf("dep %s %s |= %x\n", fp->name,
5974 // fp->dep_func[i].name, dep);
5975
5976 if (fp->has_ret == -1)
5977 fp->has_ret = fp->dep_func[i].proto->has_ret;
5978 }
5979 }
5980}
5981
5982static void output_hdr_fp(FILE *fout, const struct func_prototype *fp,
5983 int count)
5984{
5985 const struct parsed_proto *pp;
5986 char *p, namebuf[NAMELEN];
5987 const char *name;
5988 int regmask_dep;
5989 int argc_stack;
5990 int j, arg;
5991
5992 for (; count > 0; count--, fp++) {
5993 if (fp->has_ret == -1)
5994 fprintf(fout, "// ret unresolved\n");
5995#if 0
5996 fprintf(fout, "// dep:");
5997 for (j = 0; j < fp->dep_func_cnt; j++) {
5998 fprintf(fout, " %s/", fp->dep_func[j].name);
5999 if (fp->dep_func[j].proto != NULL)
6000 fprintf(fout, "%04x/%d", fp->dep_func[j].proto->regmask_dep,
6001 fp->dep_func[j].proto->has_ret);
6002 }
6003 fprintf(fout, "\n");
6004#endif
6005
6006 p = strchr(fp->name, '@');
6007 if (p != NULL) {
6008 memcpy(namebuf, fp->name, p - fp->name);
6009 namebuf[p - fp->name] = 0;
6010 name = namebuf;
6011 }
6012 else
6013 name = fp->name;
6014 if (name[0] == '_')
6015 name++;
6016
6017 pp = proto_parse(g_fhdr, name, 1);
6018 if (pp != NULL && pp->is_include)
6019 continue;
6020
6021 regmask_dep = fp->regmask_dep;
6022 argc_stack = fp->argc_stack;
6023
6024 fprintf(fout, "%-5s", fp->pp ? fp->pp->ret_type.name :
6025 (fp->has_ret ? "int" : "void"));
6026 if (regmask_dep && (fp->is_stdcall || argc_stack == 0)
6027 && (regmask_dep & ~((1 << xCX) | (1 << xDX))) == 0)
6028 {
6029 fprintf(fout, " __fastcall ");
6030 if (!(regmask_dep & (1 << xDX)) && argc_stack == 0)
6031 argc_stack = 1;
6032 else
6033 argc_stack += 2;
6034 regmask_dep = 0;
6035 }
6036 else if (regmask_dep && !fp->is_stdcall) {
6037 fprintf(fout, "/*__usercall*/ ");
6038 }
6039 else if (regmask_dep) {
6040 fprintf(fout, "/*__userpurge*/ ");
6041 }
6042 else if (fp->is_stdcall)
6043 fprintf(fout, " __stdcall ");
6044 else
6045 fprintf(fout, " __cdecl ");
6046
6047 fprintf(fout, "%s(", name);
6048
6049 arg = 0;
6050 for (j = 0; j < xSP; j++) {
6051 if (regmask_dep & (1 << j)) {
6052 arg++;
6053 if (arg != 1)
6054 fprintf(fout, ", ");
6055 if (fp->pp != NULL)
6056 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
6057 else
6058 fprintf(fout, "int");
6059 fprintf(fout, " a%d/*<%s>*/", arg, regs_r32[j]);
6060 }
6061 }
6062
6063 for (j = 0; j < argc_stack; j++) {
6064 arg++;
6065 if (arg != 1)
6066 fprintf(fout, ", ");
6067 if (fp->pp != NULL) {
6068 fprintf(fout, "%s", fp->pp->arg[arg - 1].type.name);
6069 if (!fp->pp->arg[arg - 1].type.is_ptr)
6070 fprintf(fout, " ");
6071 }
6072 else
6073 fprintf(fout, "int ");
6074 fprintf(fout, "a%d", arg);
6075 }
6076
6077 fprintf(fout, ");\n");
6078 }
6079}
6080
6081static void output_hdr(FILE *fout)
6082{
6083 static const char *lmod_c_names[] = {
6084 [OPLM_UNSPEC] = "???",
6085 [OPLM_BYTE] = "uint8_t",
6086 [OPLM_WORD] = "uint16_t",
6087 [OPLM_DWORD] = "uint32_t",
6088 [OPLM_QWORD] = "uint64_t",
6089 };
6090 const struct scanned_var *var;
6091 int i;
6092
6093 // resolve deps
6094 qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_name);
6095 for (i = 0; i < hg_fp_cnt; i++)
6096 hg_fp_resolve_deps(&hg_fp[i]);
6097
6098 // note: messes up .proto ptr, don't use
6099 //qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_id);
6100
6101 // output variables
6102 for (i = 0; i < hg_var_cnt; i++) {
6103 var = &hg_vars[i];
6104
6105 if (var->is_c_str)
6106 fprintf(fout, "extern %-8s %s[];", "char", var->name);
6107 else
6108 fprintf(fout, "extern %-8s %s;",
6109 lmod_c_names[var->lmod], var->name);
6110
6111 if (var->is_seeded)
6112 fprintf(fout, " // seeded");
6113 fprintf(fout, "\n");
6114 }
6115
6116 fprintf(fout, "\n");
6117
6118 // output function prototypes
6119 output_hdr_fp(fout, hg_fp, hg_fp_cnt);
6120}
6121
6122// read a line, truncating it if it doesn't fit
6123static char *my_fgets(char *s, size_t size, FILE *stream)
6124{
6125 char *ret, *ret2;
6126 char buf[64];
6127 int p;
6128
6129 p = size - 2;
6130 if (p >= 0)
6131 s[p] = 0;
6132
6133 ret = fgets(s, size, stream);
6134 if (ret != NULL && p >= 0 && s[p] != 0 && s[p] != '\n') {
6135 p = sizeof(buf) - 2;
6136 do {
6137 buf[p] = 0;
6138 ret2 = fgets(buf, sizeof(buf), stream);
6139 }
6140 while (ret2 != NULL && buf[p] != 0 && buf[p] != '\n');
6141 }
6142
6143 return ret;
6144}
6145
6146// '=' needs special treatment
6147// also ' quote
6148static char *next_word_s(char *w, size_t wsize, char *s)
6149{
6150 size_t i;
6151
6152 s = sskip(s);
6153
6154 i = 0;
6155 if (*s == '\'') {
6156 w[0] = s[0];
6157 for (i = 1; i < wsize - 1; i++) {
6158 if (s[i] == 0) {
6159 printf("warning: missing closing quote: \"%s\"\n", s);
6160 break;
6161 }
6162 if (s[i] == '\'')
6163 break;
6164 w[i] = s[i];
6165 }
6166 }
6167
6168 for (; i < wsize - 1; i++) {
6169 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
6170 break;
6171 w[i] = s[i];
6172 }
6173 w[i] = 0;
6174
6175 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
6176 printf("warning: '%s' truncated\n", w);
6177
6178 return s + i;
6179}
6180
6181static void scan_variables(FILE *fasm)
6182{
6183 const struct parsed_proto *pp_c;
6184 struct scanned_var *var;
6185 char line[256] = { 0, };
6186 char words[3][256];
6187 char *p = NULL;
6188 int wordc;
6189 int l;
6190
6191 while (!feof(fasm))
6192 {
6193 // skip to next data section
6194 while (my_fgets(line, sizeof(line), fasm))
6195 {
6196 asmln++;
6197
6198 p = sskip(line);
6199 if (*p == 0 || *p == ';')
6200 continue;
6201
6202 p = sskip(next_word_s(words[0], sizeof(words[0]), p));
6203 if (*p == 0 || *p == ';')
6204 continue;
6205
6206 if (*p != 's' || !IS_START(p, "segment para public"))
6207 continue;
6208
6209 break;
6210 }
6211
6212 if (p == NULL || !IS_START(p, "segment para public"))
6213 break;
6214 p = sskip(p + 19);
6215
6216 if (!IS_START(p, "'DATA'"))
6217 continue;
6218
6219 // now process it
6220 while (my_fgets(line, sizeof(line), fasm))
6221 {
6222 asmln++;
6223
6224 p = line;
6225 if (my_isblank(*p))
6226 continue;
6227
6228 p = sskip(p);
6229 if (*p == 0 || *p == ';')
6230 continue;
6231
6232 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
6233 words[wordc][0] = 0;
6234 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
6235 if (*p == 0 || *p == ';') {
6236 wordc++;
6237 break;
6238 }
6239 }
6240
6241 if (wordc == 2 && IS(words[1], "ends"))
6242 break;
6243 if (wordc < 2)
6244 continue;
6245
6246 if ((hg_var_cnt & 0xff) == 0) {
6247 hg_vars = realloc(hg_vars, sizeof(hg_vars[0])
6248 * (hg_var_cnt + 0x100));
6249 my_assert_not(hg_vars, NULL);
6250 memset(hg_vars + hg_var_cnt, 0, sizeof(hg_vars[0]) * 0x100);
6251 }
6252
6253 var = &hg_vars[hg_var_cnt++];
6254 snprintf(var->name, sizeof(var->name), "%s", words[0]);
6255
6256 // maybe already in seed header?
6257 pp_c = proto_parse(g_fhdr, var->name, 1);
6258 if (pp_c != NULL) {
6259 if (pp_c->is_func)
6260 aerr("func?\n");
6261 else if (pp_c->is_fptr) {
6262 var->lmod = OPLM_DWORD;
6263 //var->is_ptr = 1;
6264 }
6265 else if (!guess_lmod_from_c_type(&var->lmod, &pp_c->type))
6266 aerr("unhandled C type '%s' for '%s'\n",
6267 pp_c->type.name, var->name);
6268
6269 var->is_seeded = 1;
6270 continue;
6271 }
6272
6273 if (IS(words[1], "dd"))
6274 var->lmod = OPLM_DWORD;
6275 else if (IS(words[1], "dw"))
6276 var->lmod = OPLM_WORD;
6277 else if (IS(words[1], "db")) {
6278 var->lmod = OPLM_BYTE;
6279 if (wordc >= 3 && (l = strlen(words[2])) > 4) {
6280 if (words[2][0] == '\'' && IS(words[2] + l - 2, ",0"))
6281 var->is_c_str = 1;
6282 }
6283 }
6284 else if (IS(words[1], "dq"))
6285 var->lmod = OPLM_QWORD;
6286 //else if (IS(words[1], "dt"))
6287 else
6288 aerr("type '%s' not known\n", words[1]);
6289 }
6290 }
6291
6292 rewind(fasm);
6293 asmln = 0;
6294}
6295
6296static void set_label(int i, const char *name)
6297{
6298 const char *p;
6299 int len;
6300
6301 len = strlen(name);
6302 p = strchr(name, ':');
6303 if (p != NULL)
6304 len = p - name;
6305
6306 if (g_labels[i] != NULL && !IS_START(g_labels[i], "algn_"))
6307 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
6308 g_labels[i] = realloc(g_labels[i], len + 1);
6309 my_assert_not(g_labels[i], NULL);
6310 memcpy(g_labels[i], name, len);
6311 g_labels[i][len] = 0;
6312}
6313
6314struct chunk_item {
6315 char *name;
6316 long fptr;
6317 int asmln;
6318};
6319
6320static struct chunk_item *func_chunks;
6321static int func_chunk_cnt;
6322static int func_chunk_alloc;
6323
6324static void add_func_chunk(FILE *fasm, const char *name, int line)
6325{
6326 if (func_chunk_cnt >= func_chunk_alloc) {
6327 func_chunk_alloc *= 2;
6328 func_chunks = realloc(func_chunks,
6329 func_chunk_alloc * sizeof(func_chunks[0]));
6330 my_assert_not(func_chunks, NULL);
6331 }
6332 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
6333 func_chunks[func_chunk_cnt].name = strdup(name);
6334 func_chunks[func_chunk_cnt].asmln = line;
6335 func_chunk_cnt++;
6336}
6337
6338static int cmp_chunks(const void *p1, const void *p2)
6339{
6340 const struct chunk_item *c1 = p1, *c2 = p2;
6341 return strcmp(c1->name, c2->name);
6342}
6343
6344static int cmpstringp(const void *p1, const void *p2)
6345{
6346 return strcmp(*(char * const *)p1, *(char * const *)p2);
6347}
6348
6349static void scan_ahead(FILE *fasm)
6350{
6351 char words[2][256];
6352 char line[256];
6353 long oldpos;
6354 int oldasmln;
6355 int wordc;
6356 char *p;
6357 int i;
6358
6359 oldpos = ftell(fasm);
6360 oldasmln = asmln;
6361
6362 while (my_fgets(line, sizeof(line), fasm))
6363 {
6364 wordc = 0;
6365 asmln++;
6366
6367 p = sskip(line);
6368 if (*p == 0)
6369 continue;
6370
6371 if (*p == ';')
6372 {
6373 // get rid of random tabs
6374 for (i = 0; line[i] != 0; i++)
6375 if (line[i] == '\t')
6376 line[i] = ' ';
6377
6378 if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
6379 {
6380 p += 30;
6381 next_word(words[0], sizeof(words[0]), p);
6382 if (words[0][0] == 0)
6383 aerr("missing name for func chunk?\n");
6384
6385 add_func_chunk(fasm, words[0], asmln);
6386 }
6387 else if (IS_START(p, "; sctend"))
6388 break;
6389
6390 continue;
6391 } // *p == ';'
6392
6393 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
6394 words[wordc][0] = 0;
6395 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
6396 if (*p == 0 || *p == ';') {
6397 wordc++;
6398 break;
6399 }
6400 }
6401
6402 if (wordc == 2 && IS(words[1], "ends"))
6403 break;
6404 }
6405
6406 fseek(fasm, oldpos, SEEK_SET);
6407 asmln = oldasmln;
6408}
6409
6410int main(int argc, char *argv[])
6411{
6412 FILE *fout, *fasm, *frlist;
6413 struct parsed_data *pd = NULL;
6414 int pd_alloc = 0;
6415 char **rlist = NULL;
6416 int rlist_len = 0;
6417 int rlist_alloc = 0;
6418 int func_chunks_used = 0;
6419 int func_chunks_sorted = 0;
6420 int func_chunk_i = -1;
6421 long func_chunk_ret = 0;
6422 int func_chunk_ret_ln = 0;
6423 int scanned_ahead = 0;
6424 char line[256];
6425 char words[20][256];
6426 enum opr_lenmod lmod;
6427 char *sctproto = NULL;
6428 int in_func = 0;
6429 int pending_endp = 0;
6430 int skip_func = 0;
6431 int skip_warned = 0;
6432 int eq_alloc;
6433 int verbose = 0;
6434 int multi_seg = 0;
6435 int end = 0;
6436 int arg_out;
6437 int arg;
6438 int pi = 0;
6439 int i, j;
6440 int ret, len;
6441 char *p;
6442 int wordc;
6443
6444 for (arg = 1; arg < argc; arg++) {
6445 if (IS(argv[arg], "-v"))
6446 verbose = 1;
6447 else if (IS(argv[arg], "-rf"))
6448 g_allow_regfunc = 1;
6449 else if (IS(argv[arg], "-m"))
6450 multi_seg = 1;
6451 else if (IS(argv[arg], "-hdr"))
6452 g_header_mode = g_quiet_pp = g_allow_regfunc = 1;
6453 else
6454 break;
6455 }
6456
6457 if (argc < arg + 3) {
6458 printf("usage:\n%s [-v] [-rf] [-m] <.c> <.asm> <hdrf> [rlist]*\n"
6459 "%s -hdr <out.h> <.asm> <seed.h> [rlist]*\n",
6460 argv[0], argv[0]);
6461 return 1;
6462 }
6463
6464 arg_out = arg++;
6465
6466 asmfn = argv[arg++];
6467 fasm = fopen(asmfn, "r");
6468 my_assert_not(fasm, NULL);
6469
6470 hdrfn = argv[arg++];
6471 g_fhdr = fopen(hdrfn, "r");
6472 my_assert_not(g_fhdr, NULL);
6473
6474 rlist_alloc = 64;
6475 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
6476 my_assert_not(rlist, NULL);
6477 // needs special handling..
6478 rlist[rlist_len++] = "__alloca_probe";
6479
6480 func_chunk_alloc = 32;
6481 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
6482 my_assert_not(func_chunks, NULL);
6483
6484 memset(words, 0, sizeof(words));
6485
6486 for (; arg < argc; arg++) {
6487 frlist = fopen(argv[arg], "r");
6488 my_assert_not(frlist, NULL);
6489
6490 while (my_fgets(line, sizeof(line), frlist)) {
6491 p = sskip(line);
6492 if (*p == 0 || *p == ';')
6493 continue;
6494 if (*p == '#') {
6495 if (IS_START(p, "#if 0")
6496 || (g_allow_regfunc && IS_START(p, "#if NO_REGFUNC")))
6497 {
6498 skip_func = 1;
6499 }
6500 else if (IS_START(p, "#endif"))
6501 skip_func = 0;
6502 continue;
6503 }
6504 if (skip_func)
6505 continue;
6506
6507 p = next_word(words[0], sizeof(words[0]), p);
6508 if (words[0][0] == 0)
6509 continue;
6510
6511 if (rlist_len >= rlist_alloc) {
6512 rlist_alloc = rlist_alloc * 2 + 64;
6513 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
6514 my_assert_not(rlist, NULL);
6515 }
6516 rlist[rlist_len++] = strdup(words[0]);
6517 }
6518 skip_func = 0;
6519
6520 fclose(frlist);
6521 frlist = NULL;
6522 }
6523
6524 if (rlist_len > 0)
6525 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
6526
6527 fout = fopen(argv[arg_out], "w");
6528 my_assert_not(fout, NULL);
6529
6530 eq_alloc = 128;
6531 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
6532 my_assert_not(g_eqs, NULL);
6533
6534 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
6535 g_label_refs[i].i = -1;
6536 g_label_refs[i].next = NULL;
6537 }
6538
6539 if (g_header_mode)
6540 scan_variables(fasm);
6541
6542 while (my_fgets(line, sizeof(line), fasm))
6543 {
6544 wordc = 0;
6545 asmln++;
6546
6547 p = sskip(line);
6548 if (*p == 0)
6549 continue;
6550
6551 // get rid of random tabs
6552 for (i = 0; line[i] != 0; i++)
6553 if (line[i] == '\t')
6554 line[i] = ' ';
6555
6556 if (*p == ';')
6557 {
6558 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
6559 goto do_pending_endp; // eww..
6560
6561 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
6562 {
6563 static const char *attrs[] = {
6564 "bp-based frame",
6565 "library function",
6566 "static",
6567 "noreturn",
6568 "thunk",
6569 "fpd=",
6570 };
6571
6572 // parse IDA's attribute-list comment
6573 g_ida_func_attr = 0;
6574 p = sskip(p + 13);
6575
6576 for (; *p != 0; p = sskip(p)) {
6577 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
6578 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
6579 g_ida_func_attr |= 1 << i;
6580 p += strlen(attrs[i]);
6581 break;
6582 }
6583 }
6584 if (i == ARRAY_SIZE(attrs)) {
6585 anote("unparsed IDA attr: %s\n", p);
6586 break;
6587 }
6588 if (IS(attrs[i], "fpd=")) {
6589 p = next_word(words[0], sizeof(words[0]), p);
6590 // ignore for now..
6591 }
6592 }
6593 }
6594 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
6595 {
6596 p += 30;
6597 next_word(words[0], sizeof(words[0]), p);
6598 if (words[0][0] == 0)
6599 aerr("missing name for func chunk?\n");
6600
6601 if (!scanned_ahead) {
6602 add_func_chunk(fasm, words[0], asmln);
6603 func_chunks_sorted = 0;
6604 }
6605 }
6606 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
6607 {
6608 if (func_chunk_i >= 0) {
6609 if (func_chunk_i < func_chunk_cnt
6610 && IS(func_chunks[func_chunk_i].name, g_func))
6611 {
6612 // move on to next chunk
6613 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
6614 if (ret)
6615 aerr("seek failed for '%s' chunk #%d\n",
6616 g_func, func_chunk_i);
6617 asmln = func_chunks[func_chunk_i].asmln;
6618 func_chunk_i++;
6619 }
6620 else {
6621 if (func_chunk_ret == 0)
6622 aerr("no return from chunk?\n");
6623 fseek(fasm, func_chunk_ret, SEEK_SET);
6624 asmln = func_chunk_ret_ln;
6625 func_chunk_ret = 0;
6626 pending_endp = 1;
6627 }
6628 }
6629 }
6630 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
6631 func_chunks_used = 1;
6632 p += 20;
6633 if (IS_START(g_func, "sub_")) {
6634 unsigned long addr = strtoul(p, NULL, 16);
6635 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
6636 if (addr > f_addr && !scanned_ahead) {
6637 anote("scan_ahead caused by '%s', addr %lx\n",
6638 g_func, addr);
6639 scan_ahead(fasm);
6640 scanned_ahead = 1;
6641 func_chunks_sorted = 0;
6642 }
6643 }
6644 }
6645 continue;
6646 } // *p == ';'
6647
6648parse_words:
6649 for (i = wordc; i < ARRAY_SIZE(words); i++)
6650 words[i][0] = 0;
6651 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
6652 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
6653 if (*p == 0 || *p == ';') {
6654 wordc++;
6655 break;
6656 }
6657 }
6658 if (*p != 0 && *p != ';')
6659 aerr("too many words\n");
6660
6661 // alow asm patches in comments
6662 if (*p == ';') {
6663 if (IS_START(p, "; sctpatch:")) {
6664 p = sskip(p + 11);
6665 if (*p == 0 || *p == ';')
6666 continue;
6667 goto parse_words; // lame
6668 }
6669 if (IS_START(p, "; sctproto:")) {
6670 sctproto = strdup(p + 11);
6671 }
6672 else if (IS_START(p, "; sctend")) {
6673 end = 1;
6674 if (!pending_endp)
6675 break;
6676 }
6677 }
6678
6679 if (wordc == 0) {
6680 // shouldn't happen
6681 awarn("wordc == 0?\n");
6682 continue;
6683 }
6684
6685 // don't care about this:
6686 if (words[0][0] == '.'
6687 || IS(words[0], "include")
6688 || IS(words[0], "assume") || IS(words[1], "segment")
6689 || IS(words[0], "align"))
6690 {
6691 continue;
6692 }
6693
6694do_pending_endp:
6695 // do delayed endp processing to collect switch jumptables
6696 if (pending_endp) {
6697 if (in_func && !skip_func && !end && wordc >= 2
6698 && ((words[0][0] == 'd' && words[0][2] == 0)
6699 || (words[1][0] == 'd' && words[1][2] == 0)))
6700 {
6701 i = 1;
6702 if (words[1][0] == 'd' && words[1][2] == 0) {
6703 // label
6704 if (g_func_pd_cnt >= pd_alloc) {
6705 pd_alloc = pd_alloc * 2 + 16;
6706 g_func_pd = realloc(g_func_pd,
6707 sizeof(g_func_pd[0]) * pd_alloc);
6708 my_assert_not(g_func_pd, NULL);
6709 }
6710 pd = &g_func_pd[g_func_pd_cnt];
6711 g_func_pd_cnt++;
6712 memset(pd, 0, sizeof(*pd));
6713 strcpy(pd->label, words[0]);
6714 pd->type = OPT_CONST;
6715 pd->lmod = lmod_from_directive(words[1]);
6716 i = 2;
6717 }
6718 else {
6719 if (pd == NULL) {
6720 if (verbose)
6721 anote("skipping alignment byte?\n");
6722 continue;
6723 }
6724 lmod = lmod_from_directive(words[0]);
6725 if (lmod != pd->lmod)
6726 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
6727 }
6728
6729 if (pd->count_alloc < pd->count + wordc) {
6730 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
6731 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
6732 my_assert_not(pd->d, NULL);
6733 }
6734 for (; i < wordc; i++) {
6735 if (IS(words[i], "offset")) {
6736 pd->type = OPT_OFFSET;
6737 i++;
6738 }
6739 p = strchr(words[i], ',');
6740 if (p != NULL)
6741 *p = 0;
6742 if (pd->type == OPT_OFFSET)
6743 pd->d[pd->count].u.label = strdup(words[i]);
6744 else
6745 pd->d[pd->count].u.val = parse_number(words[i]);
6746 pd->d[pd->count].bt_i = -1;
6747 pd->count++;
6748 }
6749 continue;
6750 }
6751
6752 if (in_func && !skip_func) {
6753 if (g_header_mode)
6754 gen_hdr(g_func, pi);
6755 else
6756 gen_func(fout, g_fhdr, g_func, pi);
6757 }
6758
6759 pending_endp = 0;
6760 in_func = 0;
6761 g_ida_func_attr = 0;
6762 skip_warned = 0;
6763 skip_func = 0;
6764 g_func[0] = 0;
6765 func_chunks_used = 0;
6766 func_chunk_i = -1;
6767 if (pi != 0) {
6768 memset(&ops, 0, pi * sizeof(ops[0]));
6769 clear_labels(pi);
6770 pi = 0;
6771 }
6772 g_eqcnt = 0;
6773 for (i = 0; i < g_func_pd_cnt; i++) {
6774 pd = &g_func_pd[i];
6775 if (pd->type == OPT_OFFSET) {
6776 for (j = 0; j < pd->count; j++)
6777 free(pd->d[j].u.label);
6778 }
6779 free(pd->d);
6780 pd->d = NULL;
6781 }
6782 g_func_pd_cnt = 0;
6783 pd = NULL;
6784
6785 if (end)
6786 break;
6787 if (wordc == 0)
6788 continue;
6789 }
6790
6791 if (IS(words[1], "proc")) {
6792 if (in_func)
6793 aerr("proc '%s' while in_func '%s'?\n",
6794 words[0], g_func);
6795 p = words[0];
6796 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
6797 skip_func = 1;
6798 strcpy(g_func, words[0]);
6799 set_label(0, words[0]);
6800 in_func = 1;
6801 continue;
6802 }
6803
6804 if (IS(words[1], "endp"))
6805 {
6806 if (!in_func)
6807 aerr("endp '%s' while not in_func?\n", words[0]);
6808 if (!IS(g_func, words[0]))
6809 aerr("endp '%s' while in_func '%s'?\n",
6810 words[0], g_func);
6811
6812 if ((g_ida_func_attr & IDAFA_THUNK) && pi == 1
6813 && ops[0].op == OP_JMP && ops[0].operand[0].had_ds)
6814 {
6815 // import jump
6816 skip_func = 1;
6817 }
6818
6819 if (!skip_func && func_chunks_used) {
6820 // start processing chunks
6821 struct chunk_item *ci, key = { g_func, 0 };
6822
6823 func_chunk_ret = ftell(fasm);
6824 func_chunk_ret_ln = asmln;
6825 if (!func_chunks_sorted) {
6826 qsort(func_chunks, func_chunk_cnt,
6827 sizeof(func_chunks[0]), cmp_chunks);
6828 func_chunks_sorted = 1;
6829 }
6830 ci = bsearch(&key, func_chunks, func_chunk_cnt,
6831 sizeof(func_chunks[0]), cmp_chunks);
6832 if (ci == NULL)
6833 aerr("'%s' needs chunks, but none found\n", g_func);
6834 func_chunk_i = ci - func_chunks;
6835 for (; func_chunk_i > 0; func_chunk_i--)
6836 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
6837 break;
6838
6839 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
6840 if (ret)
6841 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
6842 asmln = func_chunks[func_chunk_i].asmln;
6843 func_chunk_i++;
6844 continue;
6845 }
6846 pending_endp = 1;
6847 continue;
6848 }
6849
6850 if (wordc == 2 && IS(words[1], "ends")) {
6851 if (!multi_seg) {
6852 end = 1;
6853 if (pending_endp)
6854 goto do_pending_endp;
6855 break;
6856 }
6857
6858 // scan for next text segment
6859 while (my_fgets(line, sizeof(line), fasm)) {
6860 asmln++;
6861 p = sskip(line);
6862 if (*p == 0 || *p == ';')
6863 continue;
6864
6865 if (strstr(p, "segment para public 'CODE' use32"))
6866 break;
6867 }
6868
6869 continue;
6870 }
6871
6872 p = strchr(words[0], ':');
6873 if (p != NULL) {
6874 set_label(pi, words[0]);
6875 continue;
6876 }
6877
6878 if (!in_func || skip_func) {
6879 if (!skip_warned && !skip_func && g_labels[pi] != NULL) {
6880 if (verbose)
6881 anote("skipping from '%s'\n", g_labels[pi]);
6882 skip_warned = 1;
6883 }
6884 free(g_labels[pi]);
6885 g_labels[pi] = NULL;
6886 continue;
6887 }
6888
6889 if (wordc > 1 && IS(words[1], "="))
6890 {
6891 if (wordc != 5)
6892 aerr("unhandled equ, wc=%d\n", wordc);
6893 if (g_eqcnt >= eq_alloc) {
6894 eq_alloc *= 2;
6895 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
6896 my_assert_not(g_eqs, NULL);
6897 }
6898
6899 len = strlen(words[0]);
6900 if (len > sizeof(g_eqs[0].name) - 1)
6901 aerr("equ name too long: %d\n", len);
6902 strcpy(g_eqs[g_eqcnt].name, words[0]);
6903
6904 if (!IS(words[3], "ptr"))
6905 aerr("unhandled equ\n");
6906 if (IS(words[2], "dword"))
6907 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
6908 else if (IS(words[2], "word"))
6909 g_eqs[g_eqcnt].lmod = OPLM_WORD;
6910 else if (IS(words[2], "byte"))
6911 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
6912 else if (IS(words[2], "qword"))
6913 g_eqs[g_eqcnt].lmod = OPLM_QWORD;
6914 else
6915 aerr("bad lmod: '%s'\n", words[2]);
6916
6917 g_eqs[g_eqcnt].offset = parse_number(words[4]);
6918 g_eqcnt++;
6919 continue;
6920 }
6921
6922 if (pi >= ARRAY_SIZE(ops))
6923 aerr("too many ops\n");
6924
6925 parse_op(&ops[pi], words, wordc);
6926
6927 if (sctproto != NULL) {
6928 if (ops[pi].op == OP_CALL || ops[pi].op == OP_JMP)
6929 ops[pi].datap = sctproto;
6930 sctproto = NULL;
6931 }
6932 pi++;
6933 }
6934
6935 if (g_header_mode)
6936 output_hdr(fout);
6937
6938 fclose(fout);
6939 fclose(fasm);
6940 fclose(g_fhdr);
6941
6942 return 0;
6943}
6944
6945// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.