notaz.gp2x.de / ia32rtools.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
update stdc.list for d1ui
[ia32rtools.git] / tools / translate.c
... / ...
CommitLineData
1/*
2 * ia32rtools
3 * (C) notaz, 2013,2014
4 *
5 * This work is licensed under the terms of 3-clause BSD license.
6 * See COPYING file in the top-level directory.
7 */
8
9#define _GNU_SOURCE
10#include <stdio.h>
11#include <stdlib.h>
12#include <string.h>
13
14#include "my_assert.h"
15#include "my_str.h"
16
17#define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
18#define IS(w, y) !strcmp(w, y)
19#define IS_START(w, y) !strncmp(w, y, strlen(y))
20
21#include "protoparse.h"
22
23static const char *asmfn;
24static int asmln;
25static FILE *g_fhdr;
26
27#define anote(fmt, ...) \
28 printf("%s:%d: note: " fmt, asmfn, asmln, ##__VA_ARGS__)
29#define awarn(fmt, ...) \
30 printf("%s:%d: warning: " fmt, asmfn, asmln, ##__VA_ARGS__)
31#define aerr(fmt, ...) do { \
32 printf("%s:%d: error: " fmt, asmfn, asmln, ##__VA_ARGS__); \
33 fcloseall(); \
34 exit(1); \
35} while (0)
36
37#include "masm_tools.h"
38
39enum op_flags {
40 OPF_RMD = (1 << 0), /* removed or optimized out */
41 OPF_DATA = (1 << 1), /* data processing - writes to dst opr */
42 OPF_FLAGS = (1 << 2), /* sets flags */
43 OPF_JMP = (1 << 3), /* branch, call */
44 OPF_CJMP = (1 << 4), /* cond. branch (cc or jecxz) */
45 OPF_CC = (1 << 5), /* uses flags */
46 OPF_TAIL = (1 << 6), /* ret or tail call */
47 OPF_RSAVE = (1 << 7), /* push/pop is local reg save/load */
48 OPF_REP = (1 << 8), /* prefixed by rep */
49 OPF_REPZ = (1 << 9), /* rep is repe/repz */
50 OPF_REPNZ = (1 << 10), /* rep is repne/repnz */
51 OPF_FARG = (1 << 11), /* push collected as func arg (no reuse) */
52 OPF_EBP_S = (1 << 12), /* ebp used as scratch here, not BP */
53 OPF_DF = (1 << 13), /* DF flag set */
54 OPF_ATAIL = (1 << 14), /* tail call with reused arg frame */
55 OPF_32BIT = (1 << 15), /* 32bit division */
56 OPF_LOCK = (1 << 16), /* op has lock prefix */
57 OPF_VAPUSH = (1 << 17), /* vararg ptr push (as call arg) */
58};
59
60enum op_op {
61 OP_INVAL,
62 OP_NOP,
63 OP_PUSH,
64 OP_POP,
65 OP_LEAVE,
66 OP_MOV,
67 OP_LEA,
68 OP_MOVZX,
69 OP_MOVSX,
70 OP_XCHG,
71 OP_NOT,
72 OP_CDQ,
73 OP_LODS,
74 OP_STOS,
75 OP_MOVS,
76 OP_CMPS,
77 OP_SCAS,
78 OP_STD,
79 OP_CLD,
80 OP_RET,
81 OP_ADD,
82 OP_SUB,
83 OP_AND,
84 OP_OR,
85 OP_XOR,
86 OP_SHL,
87 OP_SHR,
88 OP_SAR,
89 OP_SHRD,
90 OP_ROL,
91 OP_ROR,
92 OP_RCL,
93 OP_RCR,
94 OP_ADC,
95 OP_SBB,
96 OP_BSF,
97 OP_INC,
98 OP_DEC,
99 OP_NEG,
100 OP_MUL,
101 OP_IMUL,
102 OP_DIV,
103 OP_IDIV,
104 OP_TEST,
105 OP_CMP,
106 OP_CALL,
107 OP_JMP,
108 OP_JECXZ,
109 OP_JCC,
110 OP_SCC,
111 // x87
112 // mmx
113 OP_EMMS,
114};
115
116enum opr_type {
117 OPT_UNSPEC,
118 OPT_REG,
119 OPT_REGMEM,
120 OPT_LABEL,
121 OPT_OFFSET,
122 OPT_CONST,
123};
124
125// must be sorted (larger len must be further in enum)
126enum opr_lenmod {
127 OPLM_UNSPEC,
128 OPLM_BYTE,
129 OPLM_WORD,
130 OPLM_DWORD,
131 OPLM_QWORD,
132};
133
134#define MAX_OPERANDS 3
135#define NAMELEN 112
136
137struct parsed_opr {
138 enum opr_type type;
139 enum opr_lenmod lmod;
140 unsigned int is_ptr:1; // pointer in C
141 unsigned int is_array:1; // array in C
142 unsigned int type_from_var:1; // .. in header, sometimes wrong
143 unsigned int size_mismatch:1; // type override differs from C
144 unsigned int size_lt:1; // type override is larger than C
145 unsigned int had_ds:1; // had ds: prefix
146 const struct parsed_proto *pp; // for OPT_LABEL
147 int reg;
148 unsigned int val;
149 char name[NAMELEN];
150};
151
152struct parsed_op {
153 enum op_op op;
154 struct parsed_opr operand[MAX_OPERANDS];
155 unsigned int flags;
156 unsigned char pfo;
157 unsigned char pfo_inv;
158 unsigned char operand_cnt;
159 unsigned char p_argnum; // arg push: altered before call arg #
160 unsigned char p_arggrp; // arg push: arg group # for above
161 unsigned char p_argpass;// arg push: arg of host func
162 short p_argnext;// arg push: same arg pushed elsewhere or -1
163 int regmask_src; // all referensed regs
164 int regmask_dst;
165 int pfomask; // flagop: parsed_flag_op that can't be delayed
166 int cc_scratch; // scratch storage during analysis
167 int bt_i; // branch target for branches
168 struct parsed_data *btj;// branch targets for jumptables
169 struct parsed_proto *pp;// parsed_proto for OP_CALL
170 void *datap;
171 int asmln;
172};
173
174// datap:
175// OP_CALL - parser proto hint (str)
176// (OPF_CC) - point to one of (OPF_FLAGS) that affects cc op
177// OP_POP - point to OP_PUSH in push/pop pair
178
179struct parsed_equ {
180 char name[64];
181 enum opr_lenmod lmod;
182 int offset;
183};
184
185struct parsed_data {
186 char label[256];
187 enum opr_type type;
188 enum opr_lenmod lmod;
189 int count;
190 int count_alloc;
191 struct {
192 union {
193 char *label;
194 unsigned int val;
195 } u;
196 int bt_i;
197 } *d;
198};
199
200struct label_ref {
201 int i;
202 struct label_ref *next;
203};
204
205enum ida_func_attr {
206 IDAFA_BP_FRAME = (1 << 0),
207 IDAFA_LIB_FUNC = (1 << 1),
208 IDAFA_STATIC = (1 << 2),
209 IDAFA_NORETURN = (1 << 3),
210 IDAFA_THUNK = (1 << 4),
211 IDAFA_FPD = (1 << 5),
212};
213
214// note: limited to 32k due to p_argnext
215#define MAX_OPS 4096
216#define MAX_ARG_GRP 2
217
218static struct parsed_op ops[MAX_OPS];
219static struct parsed_equ *g_eqs;
220static int g_eqcnt;
221static char g_labels[MAX_OPS][48];
222static struct label_ref g_label_refs[MAX_OPS];
223static const struct parsed_proto *g_func_pp;
224static struct parsed_data *g_func_pd;
225static int g_func_pd_cnt;
226static char g_func[256];
227static char g_comment[256];
228static int g_bp_frame;
229static int g_sp_frame;
230static int g_stack_frame_used;
231static int g_stack_fsz;
232static int g_ida_func_attr;
233static int g_allow_regfunc;
234static int g_quiet_pp;
235
236#define ferr(op_, fmt, ...) do { \
237 printf("%s:%d: error: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
238 dump_op(op_), ##__VA_ARGS__); \
239 fcloseall(); \
240 exit(1); \
241} while (0)
242#define fnote(op_, fmt, ...) \
243 printf("%s:%d: note: [%s] '%s': " fmt, asmfn, (op_)->asmln, g_func, \
244 dump_op(op_), ##__VA_ARGS__)
245
246const char *regs_r32[] = {
247 "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp",
248 // not r32, but list here for easy parsing and printing
249 "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7",
250};
251const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" };
252const char *regs_r8l[] = { "al", "bl", "cl", "dl" };
253const char *regs_r8h[] = { "ah", "bh", "ch", "dh" };
254
255enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP };
256
257// possible basic comparison types (without inversion)
258enum parsed_flag_op {
259 PFO_O, // 0 OF=1
260 PFO_C, // 2 CF=1
261 PFO_Z, // 4 ZF=1
262 PFO_BE, // 6 CF=1||ZF=1
263 PFO_S, // 8 SF=1
264 PFO_P, // a PF=1
265 PFO_L, // c SF!=OF
266 PFO_LE, // e ZF=1||SF!=OF
267};
268
269#define PFOB_O (1 << PFO_O)
270#define PFOB_C (1 << PFO_C)
271#define PFOB_Z (1 << PFO_Z)
272#define PFOB_S (1 << PFO_S)
273
274static const char *parsed_flag_op_names[] = {
275 "o", "c", "z", "be", "s", "p", "l", "le"
276};
277
278static int char_array_i(const char *array[], size_t len, const char *s)
279{
280 int i;
281
282 for (i = 0; i < len; i++)
283 if (IS(s, array[i]))
284 return i;
285
286 return -1;
287}
288
289static void printf_number(char *buf, size_t buf_size,
290 unsigned long number)
291{
292 // output in C-friendly form
293 snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number);
294}
295
296static int check_segment_prefix(const char *s)
297{
298 if (s[0] == 0 || s[1] != 's' || s[2] != ':')
299 return 0;
300
301 switch (s[0]) {
302 case 'c': return 1;
303 case 'd': return 2;
304 case 's': return 3;
305 case 'e': return 4;
306 case 'f': return 5;
307 case 'g': return 6;
308 default: return 0;
309 }
310}
311
312static int parse_reg(enum opr_lenmod *reg_lmod, const char *s)
313{
314 int reg;
315
316 reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s);
317 if (reg >= 8) {
318 *reg_lmod = OPLM_QWORD;
319 return reg;
320 }
321 if (reg >= 0) {
322 *reg_lmod = OPLM_DWORD;
323 return reg;
324 }
325 reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s);
326 if (reg >= 0) {
327 *reg_lmod = OPLM_WORD;
328 return reg;
329 }
330 reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s);
331 if (reg >= 0) {
332 *reg_lmod = OPLM_BYTE;
333 return reg;
334 }
335 reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s);
336 if (reg >= 0) {
337 *reg_lmod = OPLM_BYTE;
338 return reg;
339 }
340
341 return -1;
342}
343
344static int parse_indmode(char *name, int *regmask, int need_c_cvt)
345{
346 enum opr_lenmod lmod;
347 char cvtbuf[256];
348 char *d = cvtbuf;
349 char *s = name;
350 char w[64];
351 long number;
352 int reg;
353 int c = 0;
354
355 *d = 0;
356
357 while (*s != 0) {
358 d += strlen(d);
359 while (my_isblank(*s))
360 s++;
361 for (; my_issep(*s); d++, s++)
362 *d = *s;
363 while (my_isblank(*s))
364 s++;
365 *d = 0;
366
367 // skip '?s:' prefixes
368 if (check_segment_prefix(s))
369 s += 3;
370
371 s = next_idt(w, sizeof(w), s);
372 if (w[0] == 0)
373 break;
374 c++;
375
376 reg = parse_reg(&lmod, w);
377 if (reg >= 0) {
378 *regmask |= 1 << reg;
379 goto pass;
380 }
381
382 if ('0' <= w[0] && w[0] <= '9') {
383 number = parse_number(w);
384 printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number);
385 continue;
386 }
387
388 // probably some label/identifier - pass
389
390pass:
391 snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w);
392 }
393
394 if (need_c_cvt)
395 strcpy(name, cvtbuf);
396
397 return c;
398}
399
400static int is_reg_in_str(const char *s)
401{
402 int i;
403
404 if (strlen(s) < 3 || (s[3] && !my_issep(s[3]) && !my_isblank(s[3])))
405 return 0;
406
407 for (i = 0; i < ARRAY_SIZE(regs_r32); i++)
408 if (!strncmp(s, regs_r32[i], 3))
409 return 1;
410
411 return 0;
412}
413
414static const char *parse_stack_el(const char *name, char *extra_reg,
415 int early_try)
416{
417 const char *p, *p2, *s;
418 char *endp = NULL;
419 char buf[32];
420 long val;
421 int len;
422
423 if (g_bp_frame || early_try)
424 {
425 p = name;
426 if (IS_START(p + 3, "+ebp+") && is_reg_in_str(p)) {
427 p += 4;
428 if (extra_reg != NULL) {
429 strncpy(extra_reg, name, 3);
430 extra_reg[4] = 0;
431 }
432 }
433
434 if (IS_START(p, "ebp+")) {
435 p += 4;
436
437 p2 = strchr(p, '+');
438 if (p2 != NULL && is_reg_in_str(p)) {
439 if (extra_reg != NULL) {
440 strncpy(extra_reg, p, p2 - p);
441 extra_reg[p2 - p] = 0;
442 }
443 p = p2 + 1;
444 }
445
446 if (!('0' <= *p && *p <= '9'))
447 return p;
448
449 return NULL;
450 }
451 }
452
453 if (!IS_START(name, "esp+"))
454 return NULL;
455
456 s = name + 4;
457 p = strchr(s, '+');
458 if (p) {
459 if (is_reg_in_str(s)) {
460 if (extra_reg != NULL) {
461 strncpy(extra_reg, s, p - s);
462 extra_reg[p - s] = 0;
463 }
464 s = p + 1;
465 p = strchr(s, '+');
466 if (p == NULL)
467 aerr("%s IDA stackvar not set?\n", __func__);
468 }
469 if (!('0' <= *s && *s <= '9')) {
470 aerr("%s IDA stackvar offset not set?\n", __func__);
471 return NULL;
472 }
473 if (s[0] == '0' && s[1] == 'x')
474 s += 2;
475 len = p - s;
476 if (len < sizeof(buf) - 1) {
477 strncpy(buf, s, len);
478 buf[len] = 0;
479 val = strtol(buf, &endp, 16);
480 if (val == 0 || *endp != 0) {
481 aerr("%s num parse fail for '%s'\n", __func__, buf);
482 return NULL;
483 }
484 }
485 p++;
486 }
487 else
488 p = name + 4;
489
490 if ('0' <= *p && *p <= '9')
491 return NULL;
492
493 return p;
494}
495
496static int guess_lmod_from_name(struct parsed_opr *opr)
497{
498 if (!strncmp(opr->name, "dword_", 6)) {
499 opr->lmod = OPLM_DWORD;
500 return 1;
501 }
502 if (!strncmp(opr->name, "word_", 5)) {
503 opr->lmod = OPLM_WORD;
504 return 1;
505 }
506 if (!strncmp(opr->name, "byte_", 5)) {
507 opr->lmod = OPLM_BYTE;
508 return 1;
509 }
510 if (!strncmp(opr->name, "qword_", 6)) {
511 opr->lmod = OPLM_QWORD;
512 return 1;
513 }
514 return 0;
515}
516
517static int guess_lmod_from_c_type(enum opr_lenmod *lmod,
518 const struct parsed_type *c_type)
519{
520 static const char *dword_types[] = {
521 "int", "_DWORD", "UINT_PTR", "DWORD",
522 "WPARAM", "LPARAM", "UINT", "__int32",
523 "LONG", "HIMC", "BOOL", "size_t",
524 "float",
525 };
526 static const char *word_types[] = {
527 "uint16_t", "int16_t", "_WORD", "WORD",
528 "unsigned __int16", "__int16",
529 };
530 static const char *byte_types[] = {
531 "uint8_t", "int8_t", "char",
532 "unsigned __int8", "__int8", "BYTE", "_BYTE",
533 "CHAR", "_UNKNOWN",
534 // structures.. deal the same as with _UNKNOWN for now
535 "CRITICAL_SECTION",
536 };
537 const char *n;
538 int i;
539
540 if (c_type->is_ptr) {
541 *lmod = OPLM_DWORD;
542 return 1;
543 }
544
545 n = skip_type_mod(c_type->name);
546
547 for (i = 0; i < ARRAY_SIZE(dword_types); i++) {
548 if (IS(n, dword_types[i])) {
549 *lmod = OPLM_DWORD;
550 return 1;
551 }
552 }
553
554 for (i = 0; i < ARRAY_SIZE(word_types); i++) {
555 if (IS(n, word_types[i])) {
556 *lmod = OPLM_WORD;
557 return 1;
558 }
559 }
560
561 for (i = 0; i < ARRAY_SIZE(byte_types); i++) {
562 if (IS(n, byte_types[i])) {
563 *lmod = OPLM_BYTE;
564 return 1;
565 }
566 }
567
568 return 0;
569}
570
571static char *default_cast_to(char *buf, size_t buf_size,
572 struct parsed_opr *opr)
573{
574 buf[0] = 0;
575
576 if (!opr->is_ptr)
577 return buf;
578 if (opr->pp == NULL || opr->pp->type.name == NULL
579 || opr->pp->is_fptr)
580 {
581 snprintf(buf, buf_size, "%s", "(void *)");
582 return buf;
583 }
584
585 snprintf(buf, buf_size, "(%s)", opr->pp->type.name);
586 return buf;
587}
588
589static enum opr_type lmod_from_directive(const char *d)
590{
591 if (IS(d, "dd"))
592 return OPLM_DWORD;
593 else if (IS(d, "dw"))
594 return OPLM_WORD;
595 else if (IS(d, "db"))
596 return OPLM_BYTE;
597
598 aerr("unhandled directive: '%s'\n", d);
599 return OPLM_UNSPEC;
600}
601
602static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod,
603 int *regmask)
604{
605 opr->type = OPT_REG;
606 opr->reg = reg;
607 opr->lmod = lmod;
608 *regmask |= 1 << reg;
609}
610
611static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
612 int *extra_offs);
613
614static int parse_operand(struct parsed_opr *opr,
615 int *regmask, int *regmask_indirect,
616 char words[16][256], int wordc, int w, unsigned int op_flags)
617{
618 const struct parsed_proto *pp = NULL;
619 enum opr_lenmod tmplmod;
620 unsigned long number;
621 char buf[256];
622 int ret, len;
623 int wordc_in;
624 char *p;
625 int i;
626
627 if (w >= wordc)
628 aerr("parse_operand w %d, wordc %d\n", w, wordc);
629
630 opr->reg = xUNSPEC;
631
632 for (i = w; i < wordc; i++) {
633 len = strlen(words[i]);
634 if (words[i][len - 1] == ',') {
635 words[i][len - 1] = 0;
636 wordc = i + 1;
637 break;
638 }
639 }
640
641 wordc_in = wordc - w;
642
643 if ((op_flags & OPF_JMP) && wordc_in > 0
644 && !('0' <= words[w][0] && words[w][0] <= '9'))
645 {
646 const char *label = NULL;
647
648 if (wordc_in == 3 && !strncmp(words[w], "near", 4)
649 && IS(words[w + 1], "ptr"))
650 label = words[w + 2];
651 else if (wordc_in == 2 && IS(words[w], "short"))
652 label = words[w + 1];
653 else if (wordc_in == 1
654 && strchr(words[w], '[') == NULL
655 && parse_reg(&tmplmod, words[w]) < 0)
656 label = words[w];
657
658 if (label != NULL) {
659 opr->type = OPT_LABEL;
660 ret = check_segment_prefix(label);
661 if (ret != 0) {
662 if (ret >= 5)
663 aerr("fs/gs used\n");
664 opr->had_ds = 1;
665 label += 3;
666 }
667 strcpy(opr->name, label);
668 return wordc;
669 }
670 }
671
672 if (wordc_in >= 3) {
673 if (IS(words[w + 1], "ptr")) {
674 if (IS(words[w], "dword"))
675 opr->lmod = OPLM_DWORD;
676 else if (IS(words[w], "word"))
677 opr->lmod = OPLM_WORD;
678 else if (IS(words[w], "byte"))
679 opr->lmod = OPLM_BYTE;
680 else if (IS(words[w], "qword"))
681 opr->lmod = OPLM_QWORD;
682 else
683 aerr("type parsing failed\n");
684 w += 2;
685 wordc_in = wordc - w;
686 }
687 }
688
689 if (wordc_in == 2) {
690 if (IS(words[w], "offset")) {
691 opr->type = OPT_OFFSET;
692 opr->lmod = OPLM_DWORD;
693 strcpy(opr->name, words[w + 1]);
694 pp = proto_parse(g_fhdr, opr->name, 1);
695 goto do_label;
696 }
697 if (IS(words[w], "(offset")) {
698 p = strchr(words[w + 1], ')');
699 if (p == NULL)
700 aerr("parse of bracketed offset failed\n");
701 *p = 0;
702 opr->type = OPT_OFFSET;
703 strcpy(opr->name, words[w + 1]);
704 return wordc;
705 }
706 }
707
708 if (wordc_in != 1)
709 aerr("parse_operand 1 word expected\n");
710
711 ret = check_segment_prefix(words[w]);
712 if (ret != 0) {
713 if (ret >= 5)
714 aerr("fs/gs used\n");
715 opr->had_ds = 1;
716 memmove(words[w], words[w] + 3, strlen(words[w]) - 2);
717 }
718 strcpy(opr->name, words[w]);
719
720 if (words[w][0] == '[') {
721 opr->type = OPT_REGMEM;
722 ret = sscanf(words[w], "[%[^]]]", opr->name);
723 if (ret != 1)
724 aerr("[] parse failure\n");
725
726 parse_indmode(opr->name, regmask_indirect, 1);
727 if (opr->lmod == OPLM_UNSPEC && parse_stack_el(opr->name, NULL, 1))
728 {
729 // might be an equ
730 struct parsed_equ *eq =
731 equ_find(NULL, parse_stack_el(opr->name, NULL, 1), &i);
732 if (eq)
733 opr->lmod = eq->lmod;
734 }
735 return wordc;
736 }
737 else if (strchr(words[w], '[')) {
738 // label[reg] form
739 p = strchr(words[w], '[');
740 opr->type = OPT_REGMEM;
741 parse_indmode(p, regmask_indirect, 0);
742 strncpy(buf, words[w], p - words[w]);
743 buf[p - words[w]] = 0;
744 pp = proto_parse(g_fhdr, buf, 1);
745 goto do_label;
746 }
747 else if (('0' <= words[w][0] && words[w][0] <= '9')
748 || words[w][0] == '-')
749 {
750 number = parse_number(words[w]);
751 opr->type = OPT_CONST;
752 opr->val = number;
753 printf_number(opr->name, sizeof(opr->name), number);
754 return wordc;
755 }
756
757 ret = parse_reg(&tmplmod, opr->name);
758 if (ret >= 0) {
759 setup_reg_opr(opr, ret, tmplmod, regmask);
760 return wordc;
761 }
762
763 // most likely var in data segment
764 opr->type = OPT_LABEL;
765 pp = proto_parse(g_fhdr, opr->name, g_quiet_pp);
766
767do_label:
768 if (pp != NULL) {
769 if (pp->is_fptr || pp->is_func) {
770 opr->lmod = OPLM_DWORD;
771 opr->is_ptr = 1;
772 }
773 else {
774 tmplmod = OPLM_UNSPEC;
775 if (!guess_lmod_from_c_type(&tmplmod, &pp->type))
776 anote("unhandled C type '%s' for '%s'\n",
777 pp->type.name, opr->name);
778
779 if (opr->lmod == OPLM_UNSPEC) {
780 opr->lmod = tmplmod;
781 opr->type_from_var = 1;
782 }
783 else if (opr->lmod != tmplmod) {
784 opr->size_mismatch = 1;
785 if (tmplmod < opr->lmod)
786 opr->size_lt = 1;
787 }
788 opr->is_ptr = pp->type.is_ptr;
789 }
790 opr->is_array = pp->type.is_array;
791 }
792 opr->pp = pp;
793
794 if (opr->lmod == OPLM_UNSPEC)
795 guess_lmod_from_name(opr);
796 return wordc;
797}
798
799static const struct {
800 const char *name;
801 unsigned int flags;
802} pref_table[] = {
803 { "rep", OPF_REP },
804 { "repe", OPF_REP|OPF_REPZ },
805 { "repz", OPF_REP|OPF_REPZ },
806 { "repne", OPF_REP|OPF_REPNZ },
807 { "repnz", OPF_REP|OPF_REPNZ },
808 { "lock", OPF_LOCK }, // ignored for now..
809};
810
811#define OPF_CJMP_CC (OPF_JMP|OPF_CJMP|OPF_CC)
812
813static const struct {
814 const char *name;
815 enum op_op op;
816 unsigned short minopr;
817 unsigned short maxopr;
818 unsigned int flags;
819 unsigned char pfo;
820 unsigned char pfo_inv;
821} op_table[] = {
822 { "nop", OP_NOP, 0, 0, 0 },
823 { "push", OP_PUSH, 1, 1, 0 },
824 { "pop", OP_POP, 1, 1, OPF_DATA },
825 { "leave",OP_LEAVE, 0, 0, OPF_DATA },
826 { "mov" , OP_MOV, 2, 2, OPF_DATA },
827 { "lea", OP_LEA, 2, 2, OPF_DATA },
828 { "movzx",OP_MOVZX, 2, 2, OPF_DATA },
829 { "movsx",OP_MOVSX, 2, 2, OPF_DATA },
830 { "xchg", OP_XCHG, 2, 2, OPF_DATA },
831 { "not", OP_NOT, 1, 1, OPF_DATA },
832 { "cdq", OP_CDQ, 0, 0, OPF_DATA },
833 { "lodsb",OP_LODS, 0, 0, OPF_DATA },
834 { "lodsw",OP_LODS, 0, 0, OPF_DATA },
835 { "lodsd",OP_LODS, 0, 0, OPF_DATA },
836 { "stosb",OP_STOS, 0, 0, OPF_DATA },
837 { "stosw",OP_STOS, 0, 0, OPF_DATA },
838 { "stosd",OP_STOS, 0, 0, OPF_DATA },
839 { "movsb",OP_MOVS, 0, 0, OPF_DATA },
840 { "movsw",OP_MOVS, 0, 0, OPF_DATA },
841 { "movsd",OP_MOVS, 0, 0, OPF_DATA },
842 { "cmpsb",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
843 { "cmpsw",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
844 { "cmpsd",OP_CMPS, 0, 0, OPF_DATA|OPF_FLAGS },
845 { "scasb",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
846 { "scasw",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
847 { "scasd",OP_SCAS, 0, 0, OPF_DATA|OPF_FLAGS },
848 { "std", OP_STD, 0, 0, OPF_DATA }, // special flag
849 { "cld", OP_CLD, 0, 0, OPF_DATA },
850 { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS },
851 { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS },
852 { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS },
853 { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS },
854 { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS },
855 { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
856 { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS },
857 { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS },
858 { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS },
859 { "shrd", OP_SHRD, 3, 3, OPF_DATA|OPF_FLAGS },
860 { "rol", OP_ROL, 2, 2, OPF_DATA|OPF_FLAGS },
861 { "ror", OP_ROR, 2, 2, OPF_DATA|OPF_FLAGS },
862 { "rcl", OP_RCL, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
863 { "rcr", OP_RCR, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
864 { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
865 { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC, PFO_C },
866 { "bsf", OP_BSF, 2, 2, OPF_DATA|OPF_FLAGS },
867 { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS },
868 { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS },
869 { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS },
870 { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS },
871 { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS },
872 { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS },
873 { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS },
874 { "test", OP_TEST, 2, 2, OPF_FLAGS },
875 { "cmp", OP_CMP, 2, 2, OPF_FLAGS },
876 { "retn", OP_RET, 0, 1, OPF_TAIL },
877 { "call", OP_CALL, 1, 1, OPF_JMP|OPF_DATA|OPF_FLAGS },
878 { "jmp", OP_JMP, 1, 1, OPF_JMP },
879 { "jecxz",OP_JECXZ, 1, 1, OPF_JMP|OPF_CJMP },
880 { "jo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 0 }, // 70 OF=1
881 { "jno", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_O, 1 }, // 71 OF=0
882 { "jc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72 CF=1
883 { "jb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 0 }, // 72
884 { "jnc", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73 CF=0
885 { "jnb", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
886 { "jae", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_C, 1 }, // 73
887 { "jz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74 ZF=1
888 { "je", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 0 }, // 74
889 { "jnz", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75 ZF=0
890 { "jne", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_Z, 1 }, // 75
891 { "jbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76 CF=1||ZF=1
892 { "jna", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 0 }, // 76
893 { "ja", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77 CF=0&&ZF=0
894 { "jnbe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_BE, 1 }, // 77
895 { "js", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 0 }, // 78 SF=1
896 { "jns", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_S, 1 }, // 79 SF=0
897 { "jp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a PF=1
898 { "jpe", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 0 }, // 7a
899 { "jnp", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b PF=0
900 { "jpo", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_P, 1 }, // 7b
901 { "jl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c SF!=OF
902 { "jnge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 0 }, // 7c
903 { "jge", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d SF=OF
904 { "jnl", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_L, 1 }, // 7d
905 { "jle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e ZF=1||SF!=OF
906 { "jng", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 0 }, // 7e
907 { "jg", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f ZF=0&&SF=OF
908 { "jnle", OP_JCC, 1, 1, OPF_CJMP_CC, PFO_LE, 1 }, // 7f
909 { "seto", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 0 },
910 { "setno", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_O, 1 },
911 { "setc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
912 { "setb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 0 },
913 { "setnc", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
914 { "setae", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
915 { "setnb", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_C, 1 },
916 { "setz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
917 { "sete", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 0 },
918 { "setnz", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
919 { "setne", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_Z, 1 },
920 { "setbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
921 { "setna", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 0 },
922 { "seta", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
923 { "setnbe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_BE, 1 },
924 { "sets", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 0 },
925 { "setns", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_S, 1 },
926 { "setp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
927 { "setpe", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 0 },
928 { "setnp", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
929 { "setpo", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_P, 1 },
930 { "setl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
931 { "setnge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 0 },
932 { "setge", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
933 { "setnl", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_L, 1 },
934 { "setle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
935 { "setng", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 0 },
936 { "setg", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
937 { "setnle", OP_SCC, 1, 1, OPF_DATA|OPF_CC, PFO_LE, 1 },
938 // x87
939 // mmx
940 { "emms", OP_EMMS, 0, 0, OPF_DATA },
941 { "movq", OP_MOV, 2, 2, OPF_DATA },
942};
943
944static void parse_op(struct parsed_op *op, char words[16][256], int wordc)
945{
946 enum opr_lenmod lmod = OPLM_UNSPEC;
947 int prefix_flags = 0;
948 int regmask_ind;
949 int regmask;
950 int op_w = 0;
951 int opr = 0;
952 int w = 0;
953 int i;
954
955 for (i = 0; i < ARRAY_SIZE(pref_table); i++) {
956 if (IS(words[w], pref_table[i].name)) {
957 prefix_flags = pref_table[i].flags;
958 break;
959 }
960 }
961
962 if (prefix_flags) {
963 if (wordc <= 1)
964 aerr("lone prefix: '%s'\n", words[0]);
965 w++;
966 }
967
968 op_w = w;
969 for (i = 0; i < ARRAY_SIZE(op_table); i++) {
970 if (IS(words[w], op_table[i].name))
971 break;
972 }
973
974 if (i == ARRAY_SIZE(op_table))
975 aerr("unhandled op: '%s'\n", words[0]);
976 w++;
977
978 op->op = op_table[i].op;
979 op->flags = op_table[i].flags | prefix_flags;
980 op->pfo = op_table[i].pfo;
981 op->pfo_inv = op_table[i].pfo_inv;
982 op->regmask_src = op->regmask_dst = 0;
983 op->asmln = asmln;
984
985 for (opr = 0; opr < op_table[i].maxopr; opr++) {
986 if (opr >= op_table[i].minopr && w >= wordc)
987 break;
988
989 regmask = regmask_ind = 0;
990 w = parse_operand(&op->operand[opr], &regmask, &regmask_ind,
991 words, wordc, w, op->flags);
992
993 if (opr == 0 && (op->flags & OPF_DATA))
994 op->regmask_dst = regmask;
995 else
996 op->regmask_src |= regmask;
997 op->regmask_src |= regmask_ind;
998 }
999
1000 if (w < wordc)
1001 aerr("parse_op %s incomplete: %d/%d\n",
1002 words[0], w, wordc);
1003
1004 // special cases
1005 op->operand_cnt = opr;
1006 if (!strncmp(op_table[i].name, "set", 3))
1007 op->operand[0].lmod = OPLM_BYTE;
1008
1009 switch (op->op) {
1010 // first operand is not dst
1011 case OP_CMP:
1012 case OP_TEST:
1013 op->regmask_src |= op->regmask_dst;
1014 op->regmask_dst = 0;
1015 break;
1016
1017 // first operand is src too
1018 case OP_NOT:
1019 case OP_ADD:
1020 case OP_AND:
1021 case OP_OR:
1022 case OP_RCL:
1023 case OP_RCR:
1024 case OP_ADC:
1025 case OP_INC:
1026 case OP_DEC:
1027 case OP_NEG:
1028 // more below..
1029 op->regmask_src |= op->regmask_dst;
1030 break;
1031
1032 // special
1033 case OP_XCHG:
1034 op->regmask_src |= op->regmask_dst;
1035 op->regmask_dst |= op->regmask_src;
1036 goto check_align;
1037
1038 case OP_SUB:
1039 case OP_SBB:
1040 case OP_XOR:
1041 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1042 && op->operand[0].lmod == op->operand[1].lmod
1043 && op->operand[0].reg == op->operand[1].reg
1044 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
1045 {
1046 op->regmask_src = 0;
1047 }
1048 else
1049 op->regmask_src |= op->regmask_dst;
1050 break;
1051
1052 // ops with implicit argumets
1053 case OP_CDQ:
1054 op->operand_cnt = 2;
1055 setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst);
1056 setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src);
1057 break;
1058
1059 case OP_LODS:
1060 case OP_STOS:
1061 case OP_SCAS:
1062 if (op->operand_cnt != 0)
1063 break;
1064 if (words[op_w][4] == 'b')
1065 lmod = OPLM_BYTE;
1066 else if (words[op_w][4] == 'w')
1067 lmod = OPLM_WORD;
1068 else if (words[op_w][4] == 'd')
1069 lmod = OPLM_DWORD;
1070 op->operand_cnt = 3;
1071 setup_reg_opr(&op->operand[0], op->op == OP_LODS ? xSI : xDI,
1072 lmod, &op->regmask_src);
1073 setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_src);
1074 op->regmask_dst = op->regmask_src;
1075 setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD,
1076 op->op == OP_LODS ? &op->regmask_dst : &op->regmask_src);
1077 break;
1078
1079 case OP_MOVS:
1080 case OP_CMPS:
1081 if (op->operand_cnt != 0)
1082 break;
1083 if (words[op_w][4] == 'b')
1084 lmod = OPLM_BYTE;
1085 else if (words[op_w][4] == 'w')
1086 lmod = OPLM_WORD;
1087 else if (words[op_w][4] == 'd')
1088 lmod = OPLM_DWORD;
1089 op->operand_cnt = 3;
1090 setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_src);
1091 setup_reg_opr(&op->operand[1], xSI, OPLM_DWORD, &op->regmask_src);
1092 setup_reg_opr(&op->operand[2], xCX, OPLM_DWORD, &op->regmask_src);
1093 op->regmask_dst = op->regmask_src;
1094 break;
1095
1096 case OP_JECXZ:
1097 op->operand_cnt = 1;
1098 op->regmask_src = 1 << xCX;
1099 op->operand[0].type = OPT_REG;
1100 op->operand[0].reg = xCX;
1101 op->operand[0].lmod = OPLM_DWORD;
1102 break;
1103
1104 case OP_IMUL:
1105 if (op->operand_cnt != 1)
1106 break;
1107 // fallthrough
1108 case OP_MUL:
1109 // singleop mul
1110 op->regmask_src |= op->regmask_dst;
1111 op->regmask_dst = (1 << xDX) | (1 << xAX);
1112 if (op->operand[0].lmod == OPLM_UNSPEC)
1113 op->operand[0].lmod = OPLM_DWORD;
1114 break;
1115
1116 case OP_DIV:
1117 case OP_IDIV:
1118 // we could set up operands for edx:eax, but there is no real need to
1119 // (see is_opr_modified())
1120 op->regmask_src |= op->regmask_dst;
1121 op->regmask_dst = (1 << xDX) | (1 << xAX);
1122 if (op->operand[0].lmod == OPLM_UNSPEC)
1123 op->operand[0].lmod = OPLM_DWORD;
1124 break;
1125
1126 case OP_SHL:
1127 case OP_SHR:
1128 case OP_SAR:
1129 case OP_ROL:
1130 case OP_ROR:
1131 op->regmask_src |= op->regmask_dst;
1132 if (op->operand[1].lmod == OPLM_UNSPEC)
1133 op->operand[1].lmod = OPLM_BYTE;
1134 break;
1135
1136 case OP_SHRD:
1137 op->regmask_src |= op->regmask_dst;
1138 if (op->operand[2].lmod == OPLM_UNSPEC)
1139 op->operand[2].lmod = OPLM_BYTE;
1140 break;
1141
1142 case OP_PUSH:
1143 op->regmask_src |= op->regmask_dst;
1144 op->regmask_dst = 0;
1145 if (op->operand[0].lmod == OPLM_UNSPEC
1146 && (op->operand[0].type == OPT_CONST
1147 || op->operand[0].type == OPT_OFFSET
1148 || op->operand[0].type == OPT_LABEL))
1149 op->operand[0].lmod = OPLM_DWORD;
1150 break;
1151
1152 // alignment
1153 case OP_MOV:
1154 check_align:
1155 if (op->operand[0].type == OPT_REG && op->operand[1].type == OPT_REG
1156 && op->operand[0].lmod == op->operand[1].lmod
1157 && op->operand[0].reg == op->operand[1].reg
1158 && IS(op->operand[0].name, op->operand[1].name)) // ! ah, al..
1159 {
1160 op->flags |= OPF_RMD;
1161 op->regmask_src = op->regmask_dst = 0;
1162 }
1163 break;
1164
1165 case OP_LEA:
1166 if (op->operand[0].type == OPT_REG
1167 && op->operand[1].type == OPT_REGMEM)
1168 {
1169 char buf[16];
1170 snprintf(buf, sizeof(buf), "%s+0", op->operand[0].name);
1171 if (IS(buf, op->operand[1].name))
1172 op->flags |= OPF_RMD;
1173 }
1174 break;
1175
1176 case OP_CALL:
1177 // trashed regs must be explicitly detected later
1178 op->regmask_dst = 0;
1179 break;
1180
1181 case OP_LEAVE:
1182 op->regmask_dst = (1 << xBP) | (1 << xSP);
1183 op->regmask_src = 1 << xBP;
1184 break;
1185
1186 default:
1187 break;
1188 }
1189}
1190
1191static const char *op_name(struct parsed_op *po)
1192{
1193 static char buf[16];
1194 char *p;
1195 int i;
1196
1197 if (po->op == OP_JCC || po->op == OP_SCC) {
1198 p = buf;
1199 *p++ = (po->op == OP_JCC) ? 'j' : 's';
1200 if (po->pfo_inv)
1201 *p++ = 'n';
1202 strcpy(p, parsed_flag_op_names[po->pfo]);
1203 return buf;
1204 }
1205
1206 for (i = 0; i < ARRAY_SIZE(op_table); i++)
1207 if (op_table[i].op == po->op)
1208 return op_table[i].name;
1209
1210 return "???";
1211}
1212
1213// debug
1214static const char *dump_op(struct parsed_op *po)
1215{
1216 static char out[128];
1217 char *p = out;
1218 int i;
1219
1220 if (po == NULL)
1221 return "???";
1222
1223 snprintf(out, sizeof(out), "%s", op_name(po));
1224 for (i = 0; i < po->operand_cnt; i++) {
1225 p += strlen(p);
1226 if (i > 0)
1227 *p++ = ',';
1228 snprintf(p, sizeof(out) - (p - out),
1229 po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s",
1230 po->operand[i].name);
1231 }
1232
1233 return out;
1234}
1235
1236static const char *lmod_type_u(struct parsed_op *po,
1237 enum opr_lenmod lmod)
1238{
1239 switch (lmod) {
1240 case OPLM_QWORD:
1241 return "u64";
1242 case OPLM_DWORD:
1243 return "u32";
1244 case OPLM_WORD:
1245 return "u16";
1246 case OPLM_BYTE:
1247 return "u8";
1248 default:
1249 ferr(po, "invalid lmod: %d\n", lmod);
1250 return "(_invalid_)";
1251 }
1252}
1253
1254static const char *lmod_cast_u(struct parsed_op *po,
1255 enum opr_lenmod lmod)
1256{
1257 switch (lmod) {
1258 case OPLM_QWORD:
1259 return "";
1260 case OPLM_DWORD:
1261 return "";
1262 case OPLM_WORD:
1263 return "(u16)";
1264 case OPLM_BYTE:
1265 return "(u8)";
1266 default:
1267 ferr(po, "invalid lmod: %d\n", lmod);
1268 return "(_invalid_)";
1269 }
1270}
1271
1272static const char *lmod_cast_u_ptr(struct parsed_op *po,
1273 enum opr_lenmod lmod)
1274{
1275 switch (lmod) {
1276 case OPLM_QWORD:
1277 return "*(u64 *)";
1278 case OPLM_DWORD:
1279 return "*(u32 *)";
1280 case OPLM_WORD:
1281 return "*(u16 *)";
1282 case OPLM_BYTE:
1283 return "*(u8 *)";
1284 default:
1285 ferr(po, "invalid lmod: %d\n", lmod);
1286 return "(_invalid_)";
1287 }
1288}
1289
1290static const char *lmod_cast_s(struct parsed_op *po,
1291 enum opr_lenmod lmod)
1292{
1293 switch (lmod) {
1294 case OPLM_QWORD:
1295 return "(s64)";
1296 case OPLM_DWORD:
1297 return "(s32)";
1298 case OPLM_WORD:
1299 return "(s16)";
1300 case OPLM_BYTE:
1301 return "(s8)";
1302 default:
1303 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1304 return "(_invalid_)";
1305 }
1306}
1307
1308static const char *lmod_cast(struct parsed_op *po,
1309 enum opr_lenmod lmod, int is_signed)
1310{
1311 return is_signed ?
1312 lmod_cast_s(po, lmod) :
1313 lmod_cast_u(po, lmod);
1314}
1315
1316static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod)
1317{
1318 switch (lmod) {
1319 case OPLM_QWORD:
1320 return 8;
1321 case OPLM_DWORD:
1322 return 4;
1323 case OPLM_WORD:
1324 return 2;
1325 case OPLM_BYTE:
1326 return 1;
1327 default:
1328 ferr(po, "%s: invalid lmod: %d\n", __func__, lmod);
1329 return 0;
1330 }
1331}
1332
1333static const char *opr_name(struct parsed_op *po, int opr_num)
1334{
1335 if (opr_num >= po->operand_cnt)
1336 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1337 return po->operand[opr_num].name;
1338}
1339
1340static unsigned int opr_const(struct parsed_op *po, int opr_num)
1341{
1342 if (opr_num >= po->operand_cnt)
1343 ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt);
1344 if (po->operand[opr_num].type != OPT_CONST)
1345 ferr(po, "opr %d: const expected\n", opr_num);
1346 return po->operand[opr_num].val;
1347}
1348
1349static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr)
1350{
1351 if ((unsigned int)popr->reg >= ARRAY_SIZE(regs_r32))
1352 ferr(po, "invalid reg: %d\n", popr->reg);
1353 return regs_r32[popr->reg];
1354}
1355
1356// cast1 is the "final" cast
1357static const char *simplify_cast(const char *cast1, const char *cast2)
1358{
1359 static char buf[256];
1360
1361 if (cast1[0] == 0)
1362 return cast2;
1363 if (cast2[0] == 0)
1364 return cast1;
1365 if (IS(cast1, cast2))
1366 return cast1;
1367 if (IS(cast1, "(s8)") && IS(cast2, "(u8)"))
1368 return cast1;
1369 if (IS(cast1, "(s16)") && IS(cast2, "(u16)"))
1370 return cast1;
1371 if (IS(cast1, "(u8)") && IS_START(cast2, "*(u8 *)"))
1372 return cast2;
1373 if (IS(cast1, "(u16)") && IS_START(cast2, "*(u16 *)"))
1374 return cast2;
1375 if (strchr(cast1, '*') && IS_START(cast2, "(u32)"))
1376 return cast1;
1377
1378 snprintf(buf, sizeof(buf), "%s%s", cast1, cast2);
1379 return buf;
1380}
1381
1382static const char *simplify_cast_num(const char *cast, unsigned int val)
1383{
1384 if (IS(cast, "(u8)") && val < 0x100)
1385 return "";
1386 if (IS(cast, "(s8)") && val < 0x80)
1387 return "";
1388 if (IS(cast, "(u16)") && val < 0x10000)
1389 return "";
1390 if (IS(cast, "(s16)") && val < 0x8000)
1391 return "";
1392 if (IS(cast, "(s32)") && val < 0x80000000)
1393 return "";
1394
1395 return cast;
1396}
1397
1398static struct parsed_equ *equ_find(struct parsed_op *po, const char *name,
1399 int *extra_offs)
1400{
1401 const char *p;
1402 char *endp;
1403 int namelen;
1404 int i;
1405
1406 *extra_offs = 0;
1407 namelen = strlen(name);
1408
1409 p = strchr(name, '+');
1410 if (p != NULL) {
1411 namelen = p - name;
1412 if (namelen <= 0)
1413 ferr(po, "equ parse failed for '%s'\n", name);
1414
1415 if (IS_START(p, "0x"))
1416 p += 2;
1417 *extra_offs = strtol(p, &endp, 16);
1418 if (*endp != 0)
1419 ferr(po, "equ parse failed for '%s'\n", name);
1420 }
1421
1422 for (i = 0; i < g_eqcnt; i++)
1423 if (strncmp(g_eqs[i].name, name, namelen) == 0
1424 && g_eqs[i].name[namelen] == 0)
1425 break;
1426 if (i >= g_eqcnt) {
1427 if (po != NULL)
1428 ferr(po, "unresolved equ name: '%s'\n", name);
1429 return NULL;
1430 }
1431
1432 return &g_eqs[i];
1433}
1434
1435static int is_stack_access(struct parsed_op *po,
1436 const struct parsed_opr *popr)
1437{
1438 return (parse_stack_el(popr->name, NULL, 0)
1439 || (g_bp_frame && !(po->flags & OPF_EBP_S)
1440 && IS_START(popr->name, "ebp")));
1441}
1442
1443static void parse_stack_access(struct parsed_op *po,
1444 const char *name, char *ofs_reg, int *offset_out,
1445 int *stack_ra_out, const char **bp_arg_out, int is_lea)
1446{
1447 const char *bp_arg = "";
1448 const char *p = NULL;
1449 struct parsed_equ *eq;
1450 char *endp = NULL;
1451 int stack_ra = 0;
1452 int offset = 0;
1453
1454 ofs_reg[0] = 0;
1455
1456 if (IS_START(name, "ebp-")
1457 || (IS_START(name, "ebp+") && '0' <= name[4] && name[4] <= '9'))
1458 {
1459 p = name + 4;
1460 if (IS_START(p, "0x"))
1461 p += 2;
1462 offset = strtoul(p, &endp, 16);
1463 if (name[3] == '-')
1464 offset = -offset;
1465 if (*endp != 0)
1466 ferr(po, "ebp- parse of '%s' failed\n", name);
1467 }
1468 else {
1469 bp_arg = parse_stack_el(name, ofs_reg, 0);
1470 snprintf(g_comment, sizeof(g_comment), "%s", bp_arg);
1471 eq = equ_find(po, bp_arg, &offset);
1472 if (eq == NULL)
1473 ferr(po, "detected but missing eq\n");
1474 offset += eq->offset;
1475 }
1476
1477 if (!strncmp(name, "ebp", 3))
1478 stack_ra = 4;
1479
1480 // yes it sometimes LEAs ra for compares..
1481 if (!is_lea && ofs_reg[0] == 0
1482 && stack_ra <= offset && offset < stack_ra + 4)
1483 {
1484 ferr(po, "reference to ra? %d %d\n", offset, stack_ra);
1485 }
1486
1487 *offset_out = offset;
1488 *stack_ra_out = stack_ra;
1489 if (bp_arg_out)
1490 *bp_arg_out = bp_arg;
1491}
1492
1493static int stack_frame_access(struct parsed_op *po,
1494 struct parsed_opr *popr, char *buf, size_t buf_size,
1495 const char *name, const char *cast, int is_src, int is_lea)
1496{
1497 enum opr_lenmod tmp_lmod = OPLM_UNSPEC;
1498 const char *prefix = "";
1499 const char *bp_arg = NULL;
1500 char ofs_reg[16] = { 0, };
1501 int i, arg_i, arg_s;
1502 int unaligned = 0;
1503 int stack_ra = 0;
1504 int offset = 0;
1505 int retval = -1;
1506 int sf_ofs;
1507 int lim;
1508
1509 if (po->flags & OPF_EBP_S)
1510 ferr(po, "stack_frame_access while ebp is scratch\n");
1511
1512 parse_stack_access(po, name, ofs_reg, &offset,
1513 &stack_ra, &bp_arg, is_lea);
1514
1515 if (offset > stack_ra)
1516 {
1517 arg_i = (offset - stack_ra - 4) / 4;
1518 if (arg_i < 0 || arg_i >= g_func_pp->argc_stack)
1519 {
1520 if (g_func_pp->is_vararg
1521 && arg_i == g_func_pp->argc_stack && is_lea)
1522 {
1523 // should be va_list
1524 if (cast[0] == 0)
1525 cast = "(u32)";
1526 snprintf(buf, buf_size, "%sap", cast);
1527 return -1;
1528 }
1529 ferr(po, "offset %d (%s,%d) doesn't map to any arg\n",
1530 offset, bp_arg, arg_i);
1531 }
1532 if (ofs_reg[0] != 0)
1533 ferr(po, "offset reg on arg access?\n");
1534
1535 for (i = arg_s = 0; i < g_func_pp->argc; i++) {
1536 if (g_func_pp->arg[i].reg != NULL)
1537 continue;
1538 if (arg_s == arg_i)
1539 break;
1540 arg_s++;
1541 }
1542 if (i == g_func_pp->argc)
1543 ferr(po, "arg %d not in prototype?\n", arg_i);
1544
1545 popr->is_ptr = g_func_pp->arg[i].type.is_ptr;
1546 retval = i;
1547
1548 switch (popr->lmod)
1549 {
1550 case OPLM_BYTE:
1551 if (is_lea)
1552 ferr(po, "lea/byte to arg?\n");
1553 if (is_src && (offset & 3) == 0)
1554 snprintf(buf, buf_size, "%sa%d",
1555 simplify_cast(cast, "(u8)"), i + 1);
1556 else
1557 snprintf(buf, buf_size, "%sBYTE%d(a%d)",
1558 cast, offset & 3, i + 1);
1559 break;
1560
1561 case OPLM_WORD:
1562 if (is_lea)
1563 ferr(po, "lea/word to arg?\n");
1564 if (offset & 1) {
1565 unaligned = 1;
1566 if (!is_src) {
1567 if (offset & 2)
1568 ferr(po, "problematic arg store\n");
1569 snprintf(buf, buf_size, "%s((char *)&a%d + 1)",
1570 simplify_cast(cast, "*(u16 *)"), i + 1);
1571 }
1572 else
1573 ferr(po, "unaligned arg word load\n");
1574 }
1575 else if (is_src && (offset & 2) == 0)
1576 snprintf(buf, buf_size, "%sa%d",
1577 simplify_cast(cast, "(u16)"), i + 1);
1578 else
1579 snprintf(buf, buf_size, "%s%sWORD(a%d)",
1580 cast, (offset & 2) ? "HI" : "LO", i + 1);
1581 break;
1582
1583 case OPLM_DWORD:
1584 if (cast[0])
1585 prefix = cast;
1586 else if (is_src)
1587 prefix = "(u32)";
1588
1589 if (offset & 3) {
1590 unaligned = 1;
1591 if (is_lea)
1592 snprintf(buf, buf_size, "(u32)&a%d + %d",
1593 i + 1, offset & 3);
1594 else if (!is_src)
1595 ferr(po, "unaligned arg store\n");
1596 else {
1597 // mov edx, [ebp+arg_4+2]; movsx ecx, dx
1598 snprintf(buf, buf_size, "%s(a%d >> %d)",
1599 prefix, i + 1, (offset & 3) * 8);
1600 }
1601 }
1602 else {
1603 snprintf(buf, buf_size, "%s%sa%d",
1604 prefix, is_lea ? "&" : "", i + 1);
1605 }
1606 break;
1607
1608 default:
1609 ferr(po, "bp_arg bad lmod: %d\n", popr->lmod);
1610 }
1611
1612 if (unaligned)
1613 snprintf(g_comment, sizeof(g_comment), "%s unaligned", bp_arg);
1614
1615 // common problem
1616 guess_lmod_from_c_type(&tmp_lmod, &g_func_pp->arg[i].type);
1617 if (tmp_lmod != OPLM_DWORD
1618 && (unaligned || (!is_src && lmod_bytes(po, tmp_lmod)
1619 < lmod_bytes(po, popr->lmod) + (offset & 3))))
1620 {
1621 ferr(po, "bp_arg arg%d/w offset %d and type '%s' is too small\n",
1622 i + 1, offset, g_func_pp->arg[i].type.name);
1623 }
1624 // can't check this because msvc likes to reuse
1625 // arg space for scratch..
1626 //if (popr->is_ptr && popr->lmod != OPLM_DWORD)
1627 // ferr(po, "bp_arg arg%d: non-dword ptr access\n", i + 1);
1628 }
1629 else
1630 {
1631 if (g_stack_fsz == 0)
1632 ferr(po, "stack var access without stackframe\n");
1633 g_stack_frame_used = 1;
1634
1635 sf_ofs = g_stack_fsz + offset;
1636 lim = (ofs_reg[0] != 0) ? -4 : 0;
1637 if (offset > 0 || sf_ofs < lim)
1638 ferr(po, "bp_stack offset %d/%d\n", offset, g_stack_fsz);
1639
1640 if (is_lea)
1641 prefix = "(u32)&";
1642 else
1643 prefix = cast;
1644
1645 switch (popr->lmod)
1646 {
1647 case OPLM_BYTE:
1648 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1649 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1650 break;
1651
1652 case OPLM_WORD:
1653 if ((sf_ofs & 1) || ofs_reg[0] != 0) {
1654 // known unaligned or possibly unaligned
1655 strcat(g_comment, " unaligned");
1656 if (prefix[0] == 0)
1657 prefix = "*(u16 *)&";
1658 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1659 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1660 break;
1661 }
1662 snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2);
1663 break;
1664
1665 case OPLM_DWORD:
1666 if ((sf_ofs & 3) || ofs_reg[0] != 0) {
1667 // known unaligned or possibly unaligned
1668 strcat(g_comment, " unaligned");
1669 if (prefix[0] == 0)
1670 prefix = "*(u32 *)&";
1671 snprintf(buf, buf_size, "%ssf.b[%d%s%s]",
1672 prefix, sf_ofs, ofs_reg[0] ? "+" : "", ofs_reg);
1673 break;
1674 }
1675 snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4);
1676 break;
1677
1678 default:
1679 ferr(po, "bp_stack bad lmod: %d\n", popr->lmod);
1680 }
1681 }
1682
1683 return retval;
1684}
1685
1686static void check_func_pp(struct parsed_op *po,
1687 const struct parsed_proto *pp, const char *pfx)
1688{
1689 enum opr_lenmod tmp_lmod;
1690 char buf[256];
1691 int ret, i;
1692
1693 if (pp->argc_reg != 0) {
1694 if (/*!g_allow_regfunc &&*/ !pp->is_fastcall) {
1695 pp_print(buf, sizeof(buf), pp);
1696 ferr(po, "%s: unexpected reg arg in icall: %s\n", pfx, buf);
1697 }
1698 if (pp->argc_stack > 0 && pp->argc_reg != 2)
1699 ferr(po, "%s: %d reg arg(s) with %d stack arg(s)\n",
1700 pfx, pp->argc_reg, pp->argc_stack);
1701 }
1702
1703 // fptrs must use 32bit args, callsite might have no information and
1704 // lack a cast to smaller types, which results in incorrectly masked
1705 // args passed (callee may assume masked args, it does on ARM)
1706 if (!pp->is_oslib) {
1707 for (i = 0; i < pp->argc; i++) {
1708 ret = guess_lmod_from_c_type(&tmp_lmod, &pp->arg[i].type);
1709 if (ret && tmp_lmod != OPLM_DWORD)
1710 ferr(po, "reference to %s with arg%d '%s'\n", pp->name,
1711 i + 1, pp->arg[i].type.name);
1712 }
1713 }
1714}
1715
1716static const char *check_label_read_ref(struct parsed_op *po,
1717 const char *name)
1718{
1719 const struct parsed_proto *pp;
1720
1721 pp = proto_parse(g_fhdr, name, 0);
1722 if (pp == NULL)
1723 ferr(po, "proto_parse failed for ref '%s'\n", name);
1724
1725 if (pp->is_func)
1726 check_func_pp(po, pp, "ref");
1727
1728 return pp->name;
1729}
1730
1731static char *out_src_opr(char *buf, size_t buf_size,
1732 struct parsed_op *po, struct parsed_opr *popr, const char *cast,
1733 int is_lea)
1734{
1735 char tmp1[256], tmp2[256];
1736 char expr[256];
1737 const char *name;
1738 char *p;
1739 int ret;
1740
1741 if (cast == NULL)
1742 cast = "";
1743
1744 switch (popr->type) {
1745 case OPT_REG:
1746 if (is_lea)
1747 ferr(po, "lea from reg?\n");
1748
1749 switch (popr->lmod) {
1750 case OPLM_QWORD:
1751 snprintf(buf, buf_size, "%s%s.q", cast, opr_reg_p(po, popr));
1752 break;
1753 case OPLM_DWORD:
1754 snprintf(buf, buf_size, "%s%s", cast, opr_reg_p(po, popr));
1755 break;
1756 case OPLM_WORD:
1757 snprintf(buf, buf_size, "%s%s",
1758 simplify_cast(cast, "(u16)"), opr_reg_p(po, popr));
1759 break;
1760 case OPLM_BYTE:
1761 if (popr->name[1] == 'h') // XXX..
1762 snprintf(buf, buf_size, "%s(%s >> 8)",
1763 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
1764 else
1765 snprintf(buf, buf_size, "%s%s",
1766 simplify_cast(cast, "(u8)"), opr_reg_p(po, popr));
1767 break;
1768 default:
1769 ferr(po, "invalid src lmod: %d\n", popr->lmod);
1770 }
1771 break;
1772
1773 case OPT_REGMEM:
1774 if (is_stack_access(po, popr)) {
1775 stack_frame_access(po, popr, buf, buf_size,
1776 popr->name, cast, 1, is_lea);
1777 break;
1778 }
1779
1780 strcpy(expr, popr->name);
1781 if (strchr(expr, '[')) {
1782 // special case: '[' can only be left for label[reg] form
1783 ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2);
1784 if (ret != 2)
1785 ferr(po, "parse failure for '%s'\n", expr);
1786 if (tmp1[0] == '(') {
1787 // (off_4FFF50+3)[eax]
1788 p = strchr(tmp1 + 1, ')');
1789 if (p == NULL || p[1] != 0)
1790 ferr(po, "parse failure (2) for '%s'\n", expr);
1791 *p = 0;
1792 memmove(tmp1, tmp1 + 1, strlen(tmp1));
1793 }
1794 snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2);
1795 }
1796
1797 // XXX: do we need more parsing?
1798 if (is_lea) {
1799 snprintf(buf, buf_size, "%s", expr);
1800 break;
1801 }
1802
1803 snprintf(buf, buf_size, "%s(%s)",
1804 simplify_cast(cast, lmod_cast_u_ptr(po, popr->lmod)), expr);
1805 break;
1806
1807 case OPT_LABEL:
1808 name = check_label_read_ref(po, popr->name);
1809 if (cast[0] == 0 && popr->is_ptr)
1810 cast = "(u32)";
1811
1812 if (is_lea)
1813 snprintf(buf, buf_size, "(u32)&%s", name);
1814 else if (popr->size_lt)
1815 snprintf(buf, buf_size, "%s%s%s%s", cast,
1816 lmod_cast_u_ptr(po, popr->lmod),
1817 popr->is_array ? "" : "&", name);
1818 else
1819 snprintf(buf, buf_size, "%s%s%s", cast, name,
1820 popr->is_array ? "[0]" : "");
1821 break;
1822
1823 case OPT_OFFSET:
1824 name = check_label_read_ref(po, popr->name);
1825 if (cast[0] == 0)
1826 cast = "(u32)";
1827 if (is_lea)
1828 ferr(po, "lea an offset?\n");
1829 snprintf(buf, buf_size, "%s&%s", cast, name);
1830 break;
1831
1832 case OPT_CONST:
1833 if (is_lea)
1834 ferr(po, "lea from const?\n");
1835
1836 printf_number(tmp1, sizeof(tmp1), popr->val);
1837 if (popr->val == 0 && strchr(cast, '*'))
1838 snprintf(buf, buf_size, "NULL");
1839 else
1840 snprintf(buf, buf_size, "%s%s",
1841 simplify_cast_num(cast, popr->val), tmp1);
1842 break;
1843
1844 default:
1845 ferr(po, "invalid src type: %d\n", popr->type);
1846 }
1847
1848 return buf;
1849}
1850
1851// note: may set is_ptr (we find that out late for ebp frame..)
1852static char *out_dst_opr(char *buf, size_t buf_size,
1853 struct parsed_op *po, struct parsed_opr *popr)
1854{
1855 switch (popr->type) {
1856 case OPT_REG:
1857 switch (popr->lmod) {
1858 case OPLM_QWORD:
1859 snprintf(buf, buf_size, "%s.q", opr_reg_p(po, popr));
1860 break;
1861 case OPLM_DWORD:
1862 snprintf(buf, buf_size, "%s", opr_reg_p(po, popr));
1863 break;
1864 case OPLM_WORD:
1865 // ugh..
1866 snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr));
1867 break;
1868 case OPLM_BYTE:
1869 // ugh..
1870 if (popr->name[1] == 'h') // XXX..
1871 snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr));
1872 else
1873 snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr));
1874 break;
1875 default:
1876 ferr(po, "invalid dst lmod: %d\n", popr->lmod);
1877 }
1878 break;
1879
1880 case OPT_REGMEM:
1881 if (is_stack_access(po, popr)) {
1882 stack_frame_access(po, popr, buf, buf_size,
1883 popr->name, "", 0, 0);
1884 break;
1885 }
1886
1887 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1888
1889 case OPT_LABEL:
1890 if (popr->size_mismatch)
1891 snprintf(buf, buf_size, "%s%s%s",
1892 lmod_cast_u_ptr(po, popr->lmod),
1893 popr->is_array ? "" : "&", popr->name);
1894 else
1895 snprintf(buf, buf_size, "%s%s", popr->name,
1896 popr->is_array ? "[0]" : "");
1897 break;
1898
1899 default:
1900 ferr(po, "invalid dst type: %d\n", popr->type);
1901 }
1902
1903 return buf;
1904}
1905
1906static char *out_src_opr_u32(char *buf, size_t buf_size,
1907 struct parsed_op *po, struct parsed_opr *popr)
1908{
1909 return out_src_opr(buf, buf_size, po, popr, NULL, 0);
1910}
1911
1912static void out_test_for_cc(char *buf, size_t buf_size,
1913 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv,
1914 enum opr_lenmod lmod, const char *expr)
1915{
1916 const char *cast, *scast;
1917
1918 cast = lmod_cast_u(po, lmod);
1919 scast = lmod_cast_s(po, lmod);
1920
1921 switch (pfo) {
1922 case PFO_Z:
1923 case PFO_BE: // CF=1||ZF=1; CF=0
1924 snprintf(buf, buf_size, "(%s%s %s 0)",
1925 cast, expr, is_inv ? "!=" : "==");
1926 break;
1927
1928 case PFO_S:
1929 case PFO_L: // SF!=OF; OF=0
1930 snprintf(buf, buf_size, "(%s%s %s 0)",
1931 scast, expr, is_inv ? ">=" : "<");
1932 break;
1933
1934 case PFO_LE: // ZF=1||SF!=OF; OF=0
1935 snprintf(buf, buf_size, "(%s%s %s 0)",
1936 scast, expr, is_inv ? ">" : "<=");
1937 break;
1938
1939 default:
1940 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1941 }
1942}
1943
1944static void out_cmp_for_cc(char *buf, size_t buf_size,
1945 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
1946{
1947 const char *cast, *scast, *cast_use;
1948 char buf1[256], buf2[256];
1949 enum opr_lenmod lmod;
1950
1951 if (po->op != OP_DEC && po->operand[0].lmod != po->operand[1].lmod)
1952 ferr(po, "%s: lmod mismatch: %d %d\n", __func__,
1953 po->operand[0].lmod, po->operand[1].lmod);
1954 lmod = po->operand[0].lmod;
1955
1956 cast = lmod_cast_u(po, lmod);
1957 scast = lmod_cast_s(po, lmod);
1958
1959 switch (pfo) {
1960 case PFO_C:
1961 case PFO_Z:
1962 case PFO_BE: // !a
1963 cast_use = cast;
1964 break;
1965
1966 case PFO_S:
1967 case PFO_L: // !ge
1968 case PFO_LE:
1969 cast_use = scast;
1970 break;
1971
1972 default:
1973 ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo);
1974 }
1975
1976 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], cast_use, 0);
1977 if (po->op == OP_DEC)
1978 snprintf(buf2, sizeof(buf2), "1");
1979 else
1980 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], cast_use, 0);
1981
1982 switch (pfo) {
1983 case PFO_C:
1984 // note: must be unsigned compare
1985 snprintf(buf, buf_size, "(%s %s %s)",
1986 buf1, is_inv ? ">=" : "<", buf2);
1987 break;
1988
1989 case PFO_Z:
1990 snprintf(buf, buf_size, "(%s %s %s)",
1991 buf1, is_inv ? "!=" : "==", buf2);
1992 break;
1993
1994 case PFO_BE: // !a
1995 // note: must be unsigned compare
1996 snprintf(buf, buf_size, "(%s %s %s)",
1997 buf1, is_inv ? ">" : "<=", buf2);
1998
1999 // annoying case
2000 if (is_inv && lmod == OPLM_BYTE
2001 && po->operand[1].type == OPT_CONST
2002 && po->operand[1].val == 0xff)
2003 {
2004 snprintf(g_comment, sizeof(g_comment), "if %s", buf);
2005 snprintf(buf, buf_size, "(0)");
2006 }
2007 break;
2008
2009 // note: must be signed compare
2010 case PFO_S:
2011 snprintf(buf, buf_size, "(%s(%s - %s) %s 0)",
2012 scast, buf1, buf2, is_inv ? ">=" : "<");
2013 break;
2014
2015 case PFO_L: // !ge
2016 snprintf(buf, buf_size, "(%s %s %s)",
2017 buf1, is_inv ? ">=" : "<", buf2);
2018 break;
2019
2020 case PFO_LE: // !g
2021 snprintf(buf, buf_size, "(%s %s %s)",
2022 buf1, is_inv ? ">" : "<=", buf2);
2023 break;
2024
2025 default:
2026 break;
2027 }
2028}
2029
2030static void out_cmp_test(char *buf, size_t buf_size,
2031 struct parsed_op *po, enum parsed_flag_op pfo, int is_inv)
2032{
2033 char buf1[256], buf2[256], buf3[256];
2034
2035 if (po->op == OP_TEST) {
2036 if (IS(opr_name(po, 0), opr_name(po, 1))) {
2037 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[0]);
2038 }
2039 else {
2040 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
2041 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
2042 snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2);
2043 }
2044 out_test_for_cc(buf, buf_size, po, pfo, is_inv,
2045 po->operand[0].lmod, buf3);
2046 }
2047 else if (po->op == OP_CMP) {
2048 out_cmp_for_cc(buf, buf_size, po, pfo, is_inv);
2049 }
2050 else
2051 ferr(po, "%s: unhandled op: %d\n", __func__, po->op);
2052}
2053
2054static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1,
2055 struct parsed_opr *popr2)
2056{
2057 if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC)
2058 ferr(po, "missing lmod for both operands\n");
2059
2060 if (popr1->lmod == OPLM_UNSPEC)
2061 popr1->lmod = popr2->lmod;
2062 else if (popr2->lmod == OPLM_UNSPEC)
2063 popr2->lmod = popr1->lmod;
2064 else if (popr1->lmod != popr2->lmod) {
2065 if (popr1->type_from_var) {
2066 popr1->size_mismatch = 1;
2067 if (popr1->lmod < popr2->lmod)
2068 popr1->size_lt = 1;
2069 popr1->lmod = popr2->lmod;
2070 }
2071 else if (popr2->type_from_var) {
2072 popr2->size_mismatch = 1;
2073 if (popr2->lmod < popr1->lmod)
2074 popr2->size_lt = 1;
2075 popr2->lmod = popr1->lmod;
2076 }
2077 else
2078 ferr(po, "conflicting lmods: %d vs %d\n",
2079 popr1->lmod, popr2->lmod);
2080 }
2081}
2082
2083static const char *op_to_c(struct parsed_op *po)
2084{
2085 switch (po->op)
2086 {
2087 case OP_ADD:
2088 case OP_ADC:
2089 return "+";
2090 case OP_SUB:
2091 case OP_SBB:
2092 return "-";
2093 case OP_AND:
2094 return "&";
2095 case OP_OR:
2096 return "|";
2097 case OP_XOR:
2098 return "^";
2099 case OP_SHL:
2100 return "<<";
2101 case OP_SHR:
2102 return ">>";
2103 case OP_MUL:
2104 case OP_IMUL:
2105 return "*";
2106 default:
2107 ferr(po, "op_to_c was supplied with %d\n", po->op);
2108 }
2109}
2110
2111static void op_set_clear_flag(struct parsed_op *po,
2112 enum op_flags flag_set, enum op_flags flag_clear)
2113{
2114 po->flags |= flag_set;
2115 po->flags &= ~flag_clear;
2116}
2117
2118// last op in stream - unconditional branch or ret
2119#define LAST_OP(_i) ((ops[_i].flags & OPF_TAIL) \
2120 || ((ops[_i].flags & (OPF_JMP|OPF_CJMP|OPF_RMD)) == OPF_JMP \
2121 && ops[_i].op != OP_CALL))
2122
2123static int scan_for_pop(int i, int opcnt, const char *reg,
2124 int magic, int depth, int *maxdepth, int do_flags)
2125{
2126 const struct parsed_proto *pp;
2127 struct parsed_op *po;
2128 int ret = 0;
2129 int j;
2130
2131 for (; i < opcnt; i++) {
2132 po = &ops[i];
2133 if (po->cc_scratch == magic)
2134 break; // already checked
2135 po->cc_scratch = magic;
2136
2137 if (po->flags & OPF_TAIL) {
2138 if (po->op == OP_CALL) {
2139 pp = proto_parse(g_fhdr, po->operand[0].name, g_quiet_pp);
2140 if (pp != NULL && pp->is_noreturn)
2141 // no stack cleanup for noreturn
2142 return ret;
2143 }
2144 return -1; // deadend
2145 }
2146
2147 if ((po->flags & OPF_RMD)
2148 || (po->op == OP_PUSH && po->p_argnum != 0)) // arg push
2149 continue;
2150
2151 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2152 if (po->btj != NULL) {
2153 // jumptable
2154 for (j = 0; j < po->btj->count; j++) {
2155 ret |= scan_for_pop(po->btj->d[j].bt_i, opcnt, reg, magic,
2156 depth, maxdepth, do_flags);
2157 if (ret < 0)
2158 return ret; // dead end
2159 }
2160 return ret;
2161 }
2162
2163 if (po->bt_i < 0) {
2164 ferr(po, "dead branch\n");
2165 return -1;
2166 }
2167
2168 if (po->flags & OPF_CJMP) {
2169 ret |= scan_for_pop(po->bt_i, opcnt, reg, magic,
2170 depth, maxdepth, do_flags);
2171 if (ret < 0)
2172 return ret; // dead end
2173 }
2174 else {
2175 i = po->bt_i - 1;
2176 }
2177 continue;
2178 }
2179
2180 if ((po->op == OP_POP || po->op == OP_PUSH)
2181 && po->operand[0].type == OPT_REG
2182 && IS(po->operand[0].name, reg))
2183 {
2184 if (po->op == OP_PUSH && !(po->flags & OPF_FARG)) {
2185 depth++;
2186 if (depth > *maxdepth)
2187 *maxdepth = depth;
2188 if (do_flags)
2189 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2190 }
2191 else if (po->op == OP_POP) {
2192 if (depth == 0) {
2193 if (do_flags)
2194 op_set_clear_flag(po, OPF_RMD, OPF_RSAVE);
2195 return 1;
2196 }
2197 else {
2198 depth--;
2199 if (depth < 0) // should not happen
2200 ferr(po, "fail with depth\n");
2201 if (do_flags)
2202 op_set_clear_flag(po, OPF_RSAVE, OPF_RMD);
2203 }
2204 }
2205 }
2206 }
2207
2208 return ret;
2209}
2210
2211// scan for pop starting from 'ret' op (all paths)
2212static int scan_for_pop_ret(int i, int opcnt, const char *reg,
2213 int flag_set)
2214{
2215 int found = 0;
2216 int j;
2217
2218 for (; i < opcnt; i++) {
2219 if (!(ops[i].flags & OPF_TAIL))
2220 continue;
2221
2222 for (j = i - 1; j >= 0; j--) {
2223 if (ops[j].flags & OPF_RMD)
2224 continue;
2225 if (ops[j].flags & OPF_JMP)
2226 return -1;
2227
2228 if (ops[j].op == OP_POP && ops[j].operand[0].type == OPT_REG
2229 && IS(ops[j].operand[0].name, reg))
2230 {
2231 found = 1;
2232 ops[j].flags |= flag_set;
2233 break;
2234 }
2235
2236 if (g_labels[j][0] != 0)
2237 return -1;
2238 }
2239 }
2240
2241 return found ? 0 : -1;
2242}
2243
2244static void scan_propagate_df(int i, int opcnt)
2245{
2246 struct parsed_op *po = &ops[i];
2247 int j;
2248
2249 for (; i < opcnt; i++) {
2250 po = &ops[i];
2251 if (po->flags & OPF_DF)
2252 return; // already resolved
2253 po->flags |= OPF_DF;
2254
2255 if (po->op == OP_CALL)
2256 ferr(po, "call with DF set?\n");
2257
2258 if (po->flags & OPF_JMP) {
2259 if (po->btj != NULL) {
2260 // jumptable
2261 for (j = 0; j < po->btj->count; j++)
2262 scan_propagate_df(po->btj->d[j].bt_i, opcnt);
2263 return;
2264 }
2265
2266 if (po->bt_i < 0) {
2267 ferr(po, "dead branch\n");
2268 return;
2269 }
2270
2271 if (po->flags & OPF_CJMP)
2272 scan_propagate_df(po->bt_i, opcnt);
2273 else
2274 i = po->bt_i - 1;
2275 continue;
2276 }
2277
2278 if (po->flags & OPF_TAIL)
2279 break;
2280
2281 if (po->op == OP_CLD) {
2282 po->flags |= OPF_RMD;
2283 return;
2284 }
2285 }
2286
2287 ferr(po, "missing DF clear?\n");
2288}
2289
2290// is operand 'opr' modified by parsed_op 'po'?
2291static int is_opr_modified(const struct parsed_opr *opr,
2292 const struct parsed_op *po)
2293{
2294 int mask;
2295
2296 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2297 return 0;
2298
2299 if (opr->type == OPT_REG) {
2300 if (po->op == OP_CALL) {
2301 mask = (1 << xAX) | (1 << xCX) | (1 << xDX);
2302 if ((1 << opr->reg) & mask)
2303 return 1;
2304 else
2305 return 0;
2306 }
2307
2308 if (po->operand[0].type == OPT_REG) {
2309 if (po->regmask_dst & (1 << opr->reg))
2310 return 1;
2311 else
2312 return 0;
2313 }
2314 }
2315
2316 return IS(po->operand[0].name, opr->name);
2317}
2318
2319// is any operand of parsed_op 'po_test' modified by parsed_op 'po'?
2320static int is_any_opr_modified(const struct parsed_op *po_test,
2321 const struct parsed_op *po, int c_mode)
2322{
2323 int mask;
2324 int i;
2325
2326 if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA))
2327 return 0;
2328
2329 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2330 return 0;
2331
2332 if ((po_test->regmask_src | po_test->regmask_dst) & po->regmask_dst)
2333 return 1;
2334
2335 // in reality, it can wreck any register, but in decompiled C
2336 // version it can only overwrite eax or edx:eax
2337 mask = (1 << xAX) | (1 << xDX);
2338 if (!c_mode)
2339 mask |= 1 << xCX;
2340
2341 if (po->op == OP_CALL
2342 && ((po_test->regmask_src | po_test->regmask_dst) & mask))
2343 return 1;
2344
2345 for (i = 0; i < po_test->operand_cnt; i++)
2346 if (IS(po_test->operand[i].name, po->operand[0].name))
2347 return 1;
2348
2349 return 0;
2350}
2351
2352// scan for any po_test operand modification in range given
2353static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt,
2354 int c_mode)
2355{
2356 if (po_test->operand_cnt == 1 && po_test->operand[0].type == OPT_CONST)
2357 return -1;
2358
2359 for (; i < opcnt; i++) {
2360 if (is_any_opr_modified(po_test, &ops[i], c_mode))
2361 return i;
2362 }
2363
2364 return -1;
2365}
2366
2367// scan for po_test operand[0] modification in range given
2368static int scan_for_mod_opr0(struct parsed_op *po_test,
2369 int i, int opcnt)
2370{
2371 for (; i < opcnt; i++) {
2372 if (is_opr_modified(&po_test->operand[0], &ops[i]))
2373 return i;
2374 }
2375
2376 return -1;
2377}
2378
2379#define check_i(po, i) \
2380 if ((i) < 0) \
2381 ferr(po, "bad " #i ": %d\n", i)
2382
2383static int scan_for_flag_set(int i, int magic, int *branched,
2384 int *setters, int *setter_cnt)
2385{
2386 struct label_ref *lr;
2387 int ret;
2388
2389 while (i >= 0) {
2390 if (ops[i].cc_scratch == magic) {
2391 ferr(&ops[i], "%s looped\n", __func__);
2392 return -1;
2393 }
2394 ops[i].cc_scratch = magic;
2395
2396 if (g_labels[i][0] != 0) {
2397 *branched = 1;
2398
2399 lr = &g_label_refs[i];
2400 for (; lr->next; lr = lr->next) {
2401 check_i(&ops[i], lr->i);
2402 ret = scan_for_flag_set(lr->i, magic,
2403 branched, setters, setter_cnt);
2404 if (ret < 0)
2405 return ret;
2406 }
2407
2408 check_i(&ops[i], lr->i);
2409 if (i > 0 && LAST_OP(i - 1)) {
2410 i = lr->i;
2411 continue;
2412 }
2413 ret = scan_for_flag_set(lr->i, magic,
2414 branched, setters, setter_cnt);
2415 if (ret < 0)
2416 return ret;
2417 }
2418 i--;
2419
2420 if (ops[i].flags & OPF_FLAGS) {
2421 setters[*setter_cnt] = i;
2422 (*setter_cnt)++;
2423 return 0;
2424 }
2425
2426 if ((ops[i].flags & (OPF_JMP|OPF_CJMP)) == OPF_JMP)
2427 return -1;
2428 }
2429
2430 return -1;
2431}
2432
2433// scan back for cdq, if anything modifies edx, fail
2434static int scan_for_cdq_edx(int i)
2435{
2436 while (i >= 0) {
2437 if (g_labels[i][0] != 0) {
2438 if (g_label_refs[i].next != NULL)
2439 return -1;
2440 if (i > 0 && LAST_OP(i - 1)) {
2441 i = g_label_refs[i].i;
2442 continue;
2443 }
2444 return -1;
2445 }
2446 i--;
2447
2448 if (ops[i].op == OP_CDQ)
2449 return i;
2450
2451 if (ops[i].regmask_dst & (1 << xDX))
2452 return -1;
2453 }
2454
2455 return -1;
2456}
2457
2458static int scan_for_reg_clear(int i, int reg)
2459{
2460 while (i >= 0) {
2461 if (g_labels[i][0] != 0) {
2462 if (g_label_refs[i].next != NULL)
2463 return -1;
2464 if (i > 0 && LAST_OP(i - 1)) {
2465 i = g_label_refs[i].i;
2466 continue;
2467 }
2468 return -1;
2469 }
2470 i--;
2471
2472 if (ops[i].op == OP_XOR
2473 && ops[i].operand[0].lmod == OPLM_DWORD
2474 && ops[i].operand[0].reg == ops[i].operand[1].reg
2475 && ops[i].operand[0].reg == reg)
2476 return i;
2477
2478 if (ops[i].regmask_dst & (1 << reg))
2479 return -1;
2480 }
2481
2482 return -1;
2483}
2484
2485// scan for positive, constant esp adjust
2486static int scan_for_esp_adjust(int i, int opcnt, int *adj,
2487 int *multipath)
2488{
2489 struct parsed_op *po;
2490 int first_pop = -1;
2491
2492 *adj = *multipath = 0;
2493
2494 for (; i < opcnt; i++) {
2495 po = &ops[i];
2496
2497 if (g_labels[i][0] != 0)
2498 *multipath = 1;
2499
2500 if (po->op == OP_ADD && po->operand[0].reg == xSP) {
2501 if (po->operand[1].type != OPT_CONST)
2502 ferr(&ops[i], "non-const esp adjust?\n");
2503 *adj += po->operand[1].val;
2504 if (*adj & 3)
2505 ferr(&ops[i], "unaligned esp adjust: %x\n", *adj);
2506 return i;
2507 }
2508 else if (po->op == OP_PUSH && !(po->flags & OPF_RMD)) {
2509 //if (first_pop == -1)
2510 // first_pop = -2; // none
2511 *adj -= lmod_bytes(po, po->operand[0].lmod);
2512 }
2513 else if (po->op == OP_POP && !(po->flags & OPF_RMD)) {
2514 // seems like msvc only uses 'pop ecx' for stack realignment..
2515 if (po->operand[0].type != OPT_REG || po->operand[0].reg != xCX)
2516 break;
2517 if (first_pop == -1 && *adj >= 0)
2518 first_pop = i;
2519 *adj += lmod_bytes(po, po->operand[0].lmod);
2520 }
2521 else if (po->flags & (OPF_JMP|OPF_TAIL)) {
2522 if (po->op == OP_JMP && po->btj == NULL) {
2523 i = po->bt_i - 1;
2524 continue;
2525 }
2526 if (po->op != OP_CALL)
2527 break;
2528 if (po->operand[0].type != OPT_LABEL)
2529 break;
2530 if (po->pp != NULL && po->pp->is_stdcall)
2531 break;
2532 }
2533 }
2534
2535 if (first_pop >= 0) {
2536 // probably 'pop ecx' was used..
2537 return first_pop;
2538 }
2539
2540 return -1;
2541}
2542
2543static void scan_fwd_set_flags(int i, int opcnt, int magic, int flags)
2544{
2545 struct parsed_op *po;
2546 int j;
2547
2548 if (i < 0)
2549 ferr(ops, "%s: followed bad branch?\n", __func__);
2550
2551 for (; i < opcnt; i++) {
2552 po = &ops[i];
2553 if (po->cc_scratch == magic)
2554 return;
2555 po->cc_scratch = magic;
2556 po->flags |= flags;
2557
2558 if ((po->flags & OPF_JMP) && po->op != OP_CALL) {
2559 if (po->btj != NULL) {
2560 // jumptable
2561 for (j = 0; j < po->btj->count; j++)
2562 scan_fwd_set_flags(po->btj->d[j].bt_i, opcnt, magic, flags);
2563 return;
2564 }
2565
2566 scan_fwd_set_flags(po->bt_i, opcnt, magic, flags);
2567 if (!(po->flags & OPF_CJMP))
2568 return;
2569 }
2570 if (po->flags & OPF_TAIL)
2571 return;
2572 }
2573}
2574
2575static const struct parsed_proto *try_recover_pp(
2576 struct parsed_op *po, const struct parsed_opr *opr, int *search_instead)
2577{
2578 const struct parsed_proto *pp = NULL;
2579 char buf[256];
2580 char *p;
2581
2582 // maybe an arg of g_func?
2583 if (opr->type == OPT_REGMEM && is_stack_access(po, opr))
2584 {
2585 char ofs_reg[16] = { 0, };
2586 int arg, arg_s, arg_i;
2587 int stack_ra = 0;
2588 int offset = 0;
2589
2590 parse_stack_access(po, opr->name, ofs_reg,
2591 &offset, &stack_ra, NULL, 0);
2592 if (ofs_reg[0] != 0)
2593 ferr(po, "offset reg on arg access?\n");
2594 if (offset <= stack_ra) {
2595 // search who set the stack var instead
2596 if (search_instead != NULL)
2597 *search_instead = 1;
2598 return NULL;
2599 }
2600
2601 arg_i = (offset - stack_ra - 4) / 4;
2602 for (arg = arg_s = 0; arg < g_func_pp->argc; arg++) {
2603 if (g_func_pp->arg[arg].reg != NULL)
2604 continue;
2605 if (arg_s == arg_i)
2606 break;
2607 arg_s++;
2608 }
2609 if (arg == g_func_pp->argc)
2610 ferr(po, "stack arg %d not in prototype?\n", arg_i);
2611
2612 pp = g_func_pp->arg[arg].fptr;
2613 if (pp == NULL)
2614 ferr(po, "icall sa: arg%d is not a fptr?\n", arg + 1);
2615 check_func_pp(po, pp, "icall arg");
2616 }
2617 else if (opr->type == OPT_REGMEM && strchr(opr->name + 1, '[')) {
2618 // label[index]
2619 p = strchr(opr->name + 1, '[');
2620 memcpy(buf, opr->name, p - opr->name);
2621 buf[p - opr->name] = 0;
2622 pp = proto_parse(g_fhdr, buf, g_quiet_pp);
2623 }
2624 else if (opr->type == OPT_OFFSET || opr->type == OPT_LABEL) {
2625 pp = proto_parse(g_fhdr, opr->name, 0);
2626 if (pp == NULL)
2627 ferr(po, "proto_parse failed for icall from '%s'\n", opr->name);
2628 check_func_pp(po, pp, "reg-fptr ref");
2629 }
2630
2631 return pp;
2632}
2633
2634static void scan_for_call_type(int i, const struct parsed_opr *opr,
2635 int magic, const struct parsed_proto **pp_found, int *multi)
2636{
2637 const struct parsed_proto *pp = NULL;
2638 struct parsed_op *po;
2639 struct label_ref *lr;
2640
2641 ops[i].cc_scratch = magic;
2642
2643 while (1) {
2644 if (g_labels[i][0] != 0) {
2645 lr = &g_label_refs[i];
2646 for (; lr != NULL; lr = lr->next) {
2647 check_i(&ops[i], lr->i);
2648 scan_for_call_type(lr->i, opr, magic, pp_found, multi);
2649 }
2650 if (i > 0 && LAST_OP(i - 1))
2651 return;
2652 }
2653
2654 i--;
2655 if (i < 0)
2656 break;
2657
2658 if (ops[i].cc_scratch == magic)
2659 return;
2660 ops[i].cc_scratch = magic;
2661
2662 if (!(ops[i].flags & OPF_DATA))
2663 continue;
2664 if (!is_opr_modified(opr, &ops[i]))
2665 continue;
2666 if (ops[i].op != OP_MOV && ops[i].op != OP_LEA) {
2667 // most probably trashed by some processing
2668 *pp_found = NULL;
2669 return;
2670 }
2671
2672 opr = &ops[i].operand[1];
2673 if (opr->type != OPT_REG)
2674 break;
2675 }
2676
2677 po = (i >= 0) ? &ops[i] : ops;
2678
2679 if (i < 0) {
2680 // reached the top - can only be an arg-reg
2681 if (opr->type != OPT_REG)
2682 return;
2683
2684 for (i = 0; i < g_func_pp->argc; i++) {
2685 if (g_func_pp->arg[i].reg == NULL)
2686 continue;
2687 if (IS(opr->name, g_func_pp->arg[i].reg))
2688 break;
2689 }
2690 if (i == g_func_pp->argc)
2691 return;
2692 pp = g_func_pp->arg[i].fptr;
2693 if (pp == NULL)
2694 ferr(po, "icall: arg%d (%s) is not a fptr?\n",
2695 i + 1, g_func_pp->arg[i].reg);
2696 check_func_pp(po, pp, "icall reg-arg");
2697 }
2698 else
2699 pp = try_recover_pp(po, opr, NULL);
2700
2701 if (*pp_found != NULL && pp != NULL && *pp_found != pp) {
2702 if (!IS((*pp_found)->ret_type.name, pp->ret_type.name)
2703 || (*pp_found)->is_stdcall != pp->is_stdcall
2704 || (*pp_found)->is_fptr != pp->is_fptr
2705 || (*pp_found)->argc != pp->argc
2706 || (*pp_found)->argc_reg != pp->argc_reg
2707 || (*pp_found)->argc_stack != pp->argc_stack)
2708 {
2709 ferr(po, "icall: parsed_proto mismatch\n");
2710 }
2711 *multi = 1;
2712 }
2713 if (pp != NULL)
2714 *pp_found = pp;
2715}
2716
2717static const struct parsed_proto *resolve_icall(int i, int opcnt,
2718 int *multi_src)
2719{
2720 const struct parsed_proto *pp = NULL;
2721 int search_advice = 0;
2722
2723 *multi_src = 0;
2724
2725 switch (ops[i].operand[0].type) {
2726 case OPT_REGMEM:
2727 case OPT_LABEL:
2728 case OPT_OFFSET:
2729 pp = try_recover_pp(&ops[i], &ops[i].operand[0], &search_advice);
2730 if (!search_advice)
2731 break;
2732 // fallthrough
2733 default:
2734 scan_for_call_type(i, &ops[i].operand[0], i + opcnt * 9, &pp,
2735 multi_src);
2736 break;
2737 }
2738
2739 return pp;
2740}
2741
2742// find an instruction that changed opr before i op
2743// *op_i must be set to -1 by caller
2744// *entry is set to 1 if one source is determined to be the caller
2745// returns 1 if found, *op_i is then set to origin
2746static int resolve_origin(int i, const struct parsed_opr *opr,
2747 int magic, int *op_i, int *is_caller)
2748{
2749 struct label_ref *lr;
2750 int ret = 0;
2751
2752 if (ops[i].cc_scratch == magic)
2753 return 0;
2754 ops[i].cc_scratch = magic;
2755
2756 while (1) {
2757 if (g_labels[i][0] != 0) {
2758 lr = &g_label_refs[i];
2759 for (; lr != NULL; lr = lr->next) {
2760 check_i(&ops[i], lr->i);
2761 ret |= resolve_origin(lr->i, opr, magic, op_i, is_caller);
2762 }
2763 if (i > 0 && LAST_OP(i - 1))
2764 return ret;
2765 }
2766
2767 i--;
2768 if (i < 0) {
2769 if (is_caller != NULL)
2770 *is_caller = 1;
2771 return -1;
2772 }
2773
2774 if (ops[i].cc_scratch == magic)
2775 return 0;
2776 ops[i].cc_scratch = magic;
2777
2778 if (!(ops[i].flags & OPF_DATA))
2779 continue;
2780 if (!is_opr_modified(opr, &ops[i]))
2781 continue;
2782
2783 if (*op_i >= 0) {
2784 if (*op_i == i)
2785 return 1;
2786 // XXX: could check if the other op does the same
2787 return -1;
2788 }
2789
2790 *op_i = i;
2791 return 1;
2792 }
2793}
2794
2795static int try_resolve_const(int i, const struct parsed_opr *opr,
2796 int magic, unsigned int *val)
2797{
2798 int s_i = -1;
2799 int ret;
2800
2801 ret = resolve_origin(i, opr, magic, &s_i, NULL);
2802 if (ret == 1) {
2803 i = s_i;
2804 if (ops[i].op != OP_MOV && ops[i].operand[1].type != OPT_CONST)
2805 return -1;
2806
2807 *val = ops[i].operand[1].val;
2808 return 1;
2809 }
2810
2811 return -1;
2812}
2813
2814static int collect_call_args_r(struct parsed_op *po, int i,
2815 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
2816 int *arg_grp, int arg, int magic, int need_op_saving, int may_reuse)
2817{
2818 struct parsed_proto *pp_tmp;
2819 struct parsed_op *po_tmp;
2820 struct label_ref *lr;
2821 int need_to_save_current;
2822 int arg_grp_current = 0;
2823 int save_args_seen = 0;
2824 int save_args;
2825 int ret = 0;
2826 int reg;
2827 char buf[32];
2828 int j, k;
2829
2830 if (i < 0) {
2831 ferr(po, "dead label encountered\n");
2832 return -1;
2833 }
2834
2835 for (; arg < pp->argc; arg++)
2836 if (pp->arg[arg].reg == NULL)
2837 break;
2838 magic = (magic & 0xffffff) | (arg << 24);
2839
2840 for (j = i; j >= 0 && (arg < pp->argc || pp->is_unresolved); )
2841 {
2842 if (((ops[j].cc_scratch ^ magic) & 0xffffff) == 0) {
2843 if (ops[j].cc_scratch != magic) {
2844 ferr(&ops[j], "arg collect hit same path with diff args for %s\n",
2845 pp->name);
2846 return -1;
2847 }
2848 // ok: have already been here
2849 return 0;
2850 }
2851 ops[j].cc_scratch = magic;
2852
2853 if (g_labels[j][0] != 0 && g_label_refs[j].i != -1) {
2854 lr = &g_label_refs[j];
2855 if (lr->next != NULL)
2856 need_op_saving = 1;
2857 for (; lr->next; lr = lr->next) {
2858 check_i(&ops[j], lr->i);
2859 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
2860 may_reuse = 1;
2861 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
2862 arg_grp, arg, magic, need_op_saving, may_reuse);
2863 if (ret < 0)
2864 return ret;
2865 }
2866
2867 check_i(&ops[j], lr->i);
2868 if ((ops[lr->i].flags & (OPF_JMP|OPF_CJMP)) != OPF_JMP)
2869 may_reuse = 1;
2870 if (j > 0 && LAST_OP(j - 1)) {
2871 // follow last branch in reverse
2872 j = lr->i;
2873 continue;
2874 }
2875 need_op_saving = 1;
2876 ret = collect_call_args_r(po, lr->i, pp, regmask, save_arg_vars,
2877 arg_grp, arg, magic, need_op_saving, may_reuse);
2878 if (ret < 0)
2879 return ret;
2880 }
2881 j--;
2882
2883 if (ops[j].op == OP_CALL)
2884 {
2885 if (pp->is_unresolved)
2886 break;
2887
2888 pp_tmp = ops[j].pp;
2889 if (pp_tmp == NULL)
2890 ferr(po, "arg collect hit unparsed call '%s'\n",
2891 ops[j].operand[0].name);
2892 if (may_reuse && pp_tmp->argc_stack > 0)
2893 ferr(po, "arg collect %d/%d hit '%s' with %d stack args\n",
2894 arg, pp->argc, opr_name(&ops[j], 0), pp_tmp->argc_stack);
2895 }
2896 // esp adjust of 0 means we collected it before
2897 else if (ops[j].op == OP_ADD && ops[j].operand[0].reg == xSP
2898 && (ops[j].operand[1].type != OPT_CONST
2899 || ops[j].operand[1].val != 0))
2900 {
2901 if (pp->is_unresolved)
2902 break;
2903
2904 ferr(po, "arg collect %d/%d hit esp adjust of %d\n",
2905 arg, pp->argc, ops[j].operand[1].val);
2906 }
2907 else if (ops[j].op == OP_POP) {
2908 if (pp->is_unresolved)
2909 break;
2910
2911 ferr(po, "arg collect %d/%d hit pop\n", arg, pp->argc);
2912 }
2913 else if (ops[j].flags & OPF_CJMP)
2914 {
2915 if (pp->is_unresolved)
2916 break;
2917
2918 may_reuse = 1;
2919 }
2920 else if (ops[j].op == OP_PUSH && !(ops[j].flags & OPF_FARG))
2921 {
2922 if (pp->is_unresolved && (ops[j].flags & OPF_RMD))
2923 break;
2924
2925 ops[j].p_argnext = -1;
2926 po_tmp = pp->arg[arg].datap;
2927 if (po_tmp != NULL)
2928 ops[j].p_argnext = po_tmp - ops;
2929 pp->arg[arg].datap = &ops[j];
2930
2931 need_to_save_current = 0;
2932 save_args = 0;
2933 reg = -1;
2934 if (ops[j].operand[0].type == OPT_REG)
2935 reg = ops[j].operand[0].reg;
2936
2937 if (!need_op_saving) {
2938 ret = scan_for_mod(&ops[j], j + 1, i, 1);
2939 need_to_save_current = (ret >= 0);
2940 }
2941 if (need_op_saving || need_to_save_current) {
2942 // mark this push as one that needs operand saving
2943 ops[j].flags &= ~OPF_RMD;
2944 if (ops[j].p_argnum == 0) {
2945 ops[j].p_argnum = arg + 1;
2946 save_args |= 1 << arg;
2947 }
2948 else if (ops[j].p_argnum < arg + 1) {
2949 // XXX: might kill valid var..
2950 //*save_arg_vars &= ~(1 << (ops[j].p_argnum - 1));
2951 ops[j].p_argnum = arg + 1;
2952 save_args |= 1 << arg;
2953 }
2954
2955 if (save_args_seen & (1 << (ops[j].p_argnum - 1))) {
2956 save_args_seen = 0;
2957 arg_grp_current++;
2958 if (arg_grp_current >= MAX_ARG_GRP)
2959 ferr(&ops[j], "out of arg groups (arg%d), f %s\n",
2960 ops[j].p_argnum, pp->name);
2961 }
2962 }
2963 else if (ops[j].p_argnum == 0)
2964 ops[j].flags |= OPF_RMD;
2965
2966 // some PUSHes are reused by different calls on other branches,
2967 // but that can't happen if we didn't branch, so they
2968 // can be removed from future searches (handles nested calls)
2969 if (!may_reuse)
2970 ops[j].flags |= OPF_FARG;
2971
2972 ops[j].flags &= ~OPF_RSAVE;
2973
2974 // check for __VALIST
2975 if (!pp->is_unresolved && pp->arg[arg].type.is_va_list) {
2976 k = -1;
2977 ret = resolve_origin(j, &ops[j].operand[0],
2978 magic + 1, &k, NULL);
2979 if (ret == 1 && k >= 0)
2980 {
2981 if (ops[k].op == OP_LEA) {
2982 snprintf(buf, sizeof(buf), "arg_%X",
2983 g_func_pp->argc_stack * 4);
2984 if (!g_func_pp->is_vararg
2985 || strstr(ops[k].operand[1].name, buf))
2986 {
2987 ops[k].flags |= OPF_RMD;
2988 ops[j].flags |= OPF_RMD | OPF_VAPUSH;
2989 save_args &= ~(1 << arg);
2990 reg = -1;
2991 }
2992 else
2993 ferr(&ops[j], "lea va_list used, but no vararg?\n");
2994 }
2995 // check for va_list from g_func_pp arg too
2996 else if (ops[k].op == OP_MOV
2997 && is_stack_access(&ops[k], &ops[k].operand[1]))
2998 {
2999 ret = stack_frame_access(&ops[k], &ops[k].operand[1],
3000 buf, sizeof(buf), ops[k].operand[1].name, "", 1, 0);
3001 if (ret >= 0) {
3002 ops[k].flags |= OPF_RMD;
3003 ops[j].flags |= OPF_RMD;
3004 ops[j].p_argpass = ret + 1;
3005 save_args &= ~(1 << arg);
3006 reg = -1;
3007 }
3008 }
3009 }
3010 }
3011
3012 *save_arg_vars |= save_args;
3013
3014 // tracking reg usage
3015 if (reg >= 0)
3016 *regmask |= 1 << reg;
3017
3018 arg++;
3019 if (!pp->is_unresolved) {
3020 // next arg
3021 for (; arg < pp->argc; arg++)
3022 if (pp->arg[arg].reg == NULL)
3023 break;
3024 }
3025 magic = (magic & 0xffffff) | (arg << 24);
3026 }
3027
3028 if (ops[j].p_arggrp > arg_grp_current) {
3029 save_args_seen = 0;
3030 arg_grp_current = ops[j].p_arggrp;
3031 }
3032 if (ops[j].p_argnum > 0)
3033 save_args_seen |= 1 << (ops[j].p_argnum - 1);
3034 }
3035
3036 if (arg < pp->argc) {
3037 ferr(po, "arg collect failed for '%s': %d/%d\n",
3038 pp->name, arg, pp->argc);
3039 return -1;
3040 }
3041
3042 if (arg_grp_current > *arg_grp)
3043 *arg_grp = arg_grp_current;
3044
3045 return arg;
3046}
3047
3048static int collect_call_args(struct parsed_op *po, int i,
3049 struct parsed_proto *pp, int *regmask, int *save_arg_vars,
3050 int magic)
3051{
3052 // arg group is for cases when pushes for
3053 // multiple funcs are going on
3054 struct parsed_op *po_tmp;
3055 int save_arg_vars_current = 0;
3056 int arg_grp = 0;
3057 int ret;
3058 int a;
3059
3060 ret = collect_call_args_r(po, i, pp, regmask,
3061 &save_arg_vars_current, &arg_grp, 0, magic, 0, 0);
3062 if (ret < 0)
3063 return ret;
3064
3065 if (arg_grp != 0) {
3066 // propagate arg_grp
3067 for (a = 0; a < pp->argc; a++) {
3068 if (pp->arg[a].reg != NULL)
3069 continue;
3070
3071 po_tmp = pp->arg[a].datap;
3072 while (po_tmp != NULL) {
3073 po_tmp->p_arggrp = arg_grp;
3074 if (po_tmp->p_argnext > 0)
3075 po_tmp = &ops[po_tmp->p_argnext];
3076 else
3077 po_tmp = NULL;
3078 }
3079 }
3080 }
3081 save_arg_vars[arg_grp] |= save_arg_vars_current;
3082
3083 if (pp->is_unresolved) {
3084 pp->argc += ret;
3085 pp->argc_stack += ret;
3086 for (a = 0; a < pp->argc; a++)
3087 if (pp->arg[a].type.name == NULL)
3088 pp->arg[a].type.name = strdup("int");
3089 }
3090
3091 return ret;
3092}
3093
3094// early check for tail call or branch back
3095static int is_like_tailjmp(int j)
3096{
3097 if (!(ops[j].flags & OPF_JMP))
3098 return 0;
3099
3100 if (ops[j].op == OP_JMP && !ops[j].operand[0].had_ds)
3101 // probably local branch back..
3102 return 1;
3103 if (ops[j].op == OP_CALL)
3104 // probably noreturn call..
3105 return 1;
3106
3107 return 0;
3108}
3109
3110static void pp_insert_reg_arg(struct parsed_proto *pp, const char *reg)
3111{
3112 int i;
3113
3114 for (i = 0; i < pp->argc; i++)
3115 if (pp->arg[i].reg == NULL)
3116 break;
3117
3118 if (pp->argc_stack)
3119 memmove(&pp->arg[i + 1], &pp->arg[i],
3120 sizeof(pp->arg[0]) * pp->argc_stack);
3121 memset(&pp->arg[i], 0, sizeof(pp->arg[i]));
3122 pp->arg[i].reg = strdup(reg);
3123 pp->arg[i].type.name = strdup("int");
3124 pp->argc++;
3125 pp->argc_reg++;
3126}
3127
3128static void add_label_ref(struct label_ref *lr, int op_i)
3129{
3130 struct label_ref *lr_new;
3131
3132 if (lr->i == -1) {
3133 lr->i = op_i;
3134 return;
3135 }
3136
3137 lr_new = calloc(1, sizeof(*lr_new));
3138 lr_new->i = op_i;
3139 lr_new->next = lr->next;
3140 lr->next = lr_new;
3141}
3142
3143static struct parsed_data *try_resolve_jumptab(int i, int opcnt)
3144{
3145 struct parsed_op *po = &ops[i];
3146 struct parsed_data *pd;
3147 char label[NAMELEN], *p;
3148 int len, j, l;
3149
3150 p = strchr(po->operand[0].name, '[');
3151 if (p == NULL)
3152 return NULL;
3153
3154 len = p - po->operand[0].name;
3155 strncpy(label, po->operand[0].name, len);
3156 label[len] = 0;
3157
3158 for (j = 0, pd = NULL; j < g_func_pd_cnt; j++) {
3159 if (IS(g_func_pd[j].label, label)) {
3160 pd = &g_func_pd[j];
3161 break;
3162 }
3163 }
3164 if (pd == NULL)
3165 //ferr(po, "label '%s' not parsed?\n", label);
3166 return NULL;
3167
3168 if (pd->type != OPT_OFFSET)
3169 ferr(po, "label '%s' with non-offset data?\n", label);
3170
3171 // find all labels, link
3172 for (j = 0; j < pd->count; j++) {
3173 for (l = 0; l < opcnt; l++) {
3174 if (g_labels[l][0] && IS(g_labels[l], pd->d[j].u.label)) {
3175 add_label_ref(&g_label_refs[l], i);
3176 pd->d[j].bt_i = l;
3177 break;
3178 }
3179 }
3180 }
3181
3182 return pd;
3183}
3184
3185static void output_std_flags(FILE *fout, struct parsed_op *po,
3186 int *pfomask, const char *dst_opr_text)
3187{
3188 if (*pfomask & (1 << PFO_Z)) {
3189 fprintf(fout, "\n cond_z = (%s%s == 0);",
3190 lmod_cast_u(po, po->operand[0].lmod), dst_opr_text);
3191 *pfomask &= ~(1 << PFO_Z);
3192 }
3193 if (*pfomask & (1 << PFO_S)) {
3194 fprintf(fout, "\n cond_s = (%s%s < 0);",
3195 lmod_cast_s(po, po->operand[0].lmod), dst_opr_text);
3196 *pfomask &= ~(1 << PFO_S);
3197 }
3198}
3199
3200static void output_pp_attrs(FILE *fout, const struct parsed_proto *pp,
3201 int is_noreturn)
3202{
3203 if (pp->is_fastcall)
3204 fprintf(fout, "__fastcall ");
3205 else if (pp->is_stdcall && pp->argc_reg == 0)
3206 fprintf(fout, "__stdcall ");
3207 if (pp->is_noreturn || is_noreturn)
3208 fprintf(fout, "noreturn ");
3209}
3210
3211static char *saved_arg_name(char *buf, size_t buf_size, int grp, int num)
3212{
3213 char buf1[16];
3214
3215 buf1[0] = 0;
3216 if (grp > 0)
3217 snprintf(buf1, sizeof(buf1), "%d", grp);
3218 snprintf(buf, buf_size, "s%s_a%d", buf1, num);
3219
3220 return buf;
3221}
3222
3223static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt)
3224{
3225 struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op;
3226 struct parsed_opr *last_arith_dst = NULL;
3227 char buf1[256], buf2[256], buf3[256], cast[64];
3228 const struct parsed_proto *pp_c;
3229 struct parsed_proto *pp, *pp_tmp;
3230 struct parsed_data *pd;
3231 const char *tmpname;
3232 unsigned int uval;
3233 int save_arg_vars[MAX_ARG_GRP] = { 0, };
3234 int cond_vars = 0;
3235 int need_tmp_var = 0;
3236 int need_tmp64 = 0;
3237 int had_decl = 0;
3238 int label_pending = 0;
3239 int regmask_save = 0;
3240 int regmask_arg = 0;
3241 int regmask_now = 0;
3242 int regmask_init = 0;
3243 int regmask = 0;
3244 int pfomask = 0;
3245 int found = 0;
3246 int depth = 0;
3247 int no_output;
3248 int i, j, l;
3249 int arg;
3250 int reg;
3251 int ret;
3252
3253 g_bp_frame = g_sp_frame = g_stack_fsz = 0;
3254 g_stack_frame_used = 0;
3255
3256 g_func_pp = proto_parse(fhdr, funcn, 0);
3257 if (g_func_pp == NULL)
3258 ferr(ops, "proto_parse failed for '%s'\n", funcn);
3259
3260 for (i = 0; i < g_func_pp->argc; i++) {
3261 if (g_func_pp->arg[i].reg != NULL) {
3262 reg = char_array_i(regs_r32,
3263 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
3264 if (reg < 0)
3265 ferr(ops, "arg '%s' is not a reg?\n", g_func_pp->arg[i].reg);
3266 regmask_arg |= 1 << reg;
3267 }
3268 }
3269
3270 // pass1:
3271 // - handle ebp/esp frame, remove ops related to it
3272 if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp")
3273 && ops[1].op == OP_MOV
3274 && IS(opr_name(&ops[1], 0), "ebp")
3275 && IS(opr_name(&ops[1], 1), "esp"))
3276 {
3277 int ecx_push = 0;
3278
3279 g_bp_frame = 1;
3280 ops[0].flags |= OPF_RMD;
3281 ops[1].flags |= OPF_RMD;
3282 i = 2;
3283
3284 if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) {
3285 g_stack_fsz = opr_const(&ops[2], 1);
3286 ops[2].flags |= OPF_RMD;
3287 i++;
3288 }
3289 else {
3290 // another way msvc builds stack frame..
3291 i = 2;
3292 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
3293 g_stack_fsz += 4;
3294 ops[i].flags |= OPF_RMD;
3295 ecx_push++;
3296 i++;
3297 }
3298 // and another way..
3299 if (i == 2 && ops[i].op == OP_MOV && ops[i].operand[0].reg == xAX
3300 && ops[i].operand[1].type == OPT_CONST
3301 && ops[i + 1].op == OP_CALL
3302 && IS(opr_name(&ops[i + 1], 0), "__alloca_probe"))
3303 {
3304 g_stack_fsz += ops[i].operand[1].val;
3305 ops[i].flags |= OPF_RMD;
3306 i++;
3307 ops[i].flags |= OPF_RMD;
3308 i++;
3309 }
3310 }
3311
3312 found = 0;
3313 do {
3314 for (; i < opcnt; i++)
3315 if (ops[i].op == OP_RET)
3316 break;
3317 j = i - 1;
3318 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
3319 if (found && is_like_tailjmp(j))
3320 break;
3321 j--;
3322 }
3323
3324 if ((ops[j].op == OP_POP && IS(opr_name(&ops[j], 0), "ebp"))
3325 || ops[j].op == OP_LEAVE)
3326 {
3327 ops[j].flags |= OPF_RMD;
3328 }
3329 else if (!(g_ida_func_attr & IDAFA_NORETURN))
3330 ferr(&ops[j], "'pop ebp' expected\n");
3331
3332 if (g_stack_fsz != 0) {
3333 if (ops[j - 1].op == OP_MOV
3334 && IS(opr_name(&ops[j - 1], 0), "esp")
3335 && IS(opr_name(&ops[j - 1], 1), "ebp"))
3336 {
3337 ops[j - 1].flags |= OPF_RMD;
3338 }
3339 else if (ops[j].op != OP_LEAVE
3340 && !(g_ida_func_attr & IDAFA_NORETURN))
3341 {
3342 ferr(&ops[j - 1], "esp restore expected\n");
3343 }
3344
3345 if (ecx_push && ops[j - 2].op == OP_POP
3346 && IS(opr_name(&ops[j - 2], 0), "ecx"))
3347 {
3348 ferr(&ops[j - 2], "unexpected ecx pop\n");
3349 }
3350 }
3351
3352 found = 1;
3353 i++;
3354 } while (i < opcnt);
3355 }
3356 else {
3357 int ecx_push = 0, esp_sub = 0;
3358
3359 i = 0;
3360 while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) {
3361 ops[i].flags |= OPF_RMD;
3362 g_stack_fsz += 4;
3363 ecx_push++;
3364 i++;
3365 }
3366
3367 for (; i < opcnt; i++) {
3368 if (ops[i].op == OP_PUSH || (ops[i].flags & (OPF_JMP|OPF_TAIL)))
3369 break;
3370 if (ops[i].op == OP_SUB && ops[i].operand[0].reg == xSP
3371 && ops[i].operand[1].type == OPT_CONST)
3372 {
3373 g_stack_fsz = ops[i].operand[1].val;
3374 ops[i].flags |= OPF_RMD;
3375 esp_sub = 1;
3376 break;
3377 }
3378 }
3379
3380 found = 0;
3381 if (ecx_push || esp_sub)
3382 {
3383 g_sp_frame = 1;
3384
3385 i++;
3386 do {
3387 for (; i < opcnt; i++)
3388 if (ops[i].op == OP_RET)
3389 break;
3390 j = i - 1;
3391 if (i == opcnt && (ops[j].flags & OPF_JMP)) {
3392 if (found && is_like_tailjmp(j))
3393 break;
3394 j--;
3395 }
3396
3397 if (ecx_push > 0) {
3398 for (l = 0; l < ecx_push; l++) {
3399 if (ops[j].op != OP_POP
3400 || !IS(opr_name(&ops[j], 0), "ecx"))
3401 {
3402 ferr(&ops[j], "'pop ecx' expected\n");
3403 }
3404 ops[j].flags |= OPF_RMD;
3405 j--;
3406 }
3407
3408 found = 1;
3409 }
3410
3411 if (esp_sub) {
3412 if (ops[j].op != OP_ADD
3413 || !IS(opr_name(&ops[j], 0), "esp")
3414 || ops[j].operand[1].type != OPT_CONST
3415 || ops[j].operand[1].val != g_stack_fsz)
3416 ferr(&ops[j], "'add esp' expected\n");
3417 ops[j].flags |= OPF_RMD;
3418
3419 found = 1;
3420 }
3421
3422 i++;
3423 } while (i < opcnt);
3424 }
3425 }
3426
3427 // pass2:
3428 // - parse calls with labels
3429 // - resolve all branches
3430 for (i = 0; i < opcnt; i++)
3431 {
3432 po = &ops[i];
3433 po->bt_i = -1;
3434 po->btj = NULL;
3435
3436 if (po->flags & OPF_RMD)
3437 continue;
3438
3439 if (po->op == OP_CALL) {
3440 pp = NULL;
3441
3442 if (po->operand[0].type == OPT_LABEL) {
3443 tmpname = opr_name(po, 0);
3444 if (IS_START(tmpname, "loc_"))
3445 ferr(po, "call to loc_*\n");
3446 pp_c = proto_parse(fhdr, tmpname, 0);
3447 if (pp_c == NULL)
3448 ferr(po, "proto_parse failed for call '%s'\n", tmpname);
3449
3450 pp = proto_clone(pp_c);
3451 my_assert_not(pp, NULL);
3452 }
3453 else if (po->datap != NULL) {
3454 pp = calloc(1, sizeof(*pp));
3455 my_assert_not(pp, NULL);
3456
3457 ret = parse_protostr(po->datap, pp);
3458 if (ret < 0)
3459 ferr(po, "bad protostr supplied: %s\n", (char *)po->datap);
3460 free(po->datap);
3461 po->datap = NULL;
3462 }
3463
3464 if (pp != NULL) {
3465 if (pp->is_fptr)
3466 check_func_pp(po, pp, "fptr var call");
3467 if (pp->is_noreturn)
3468 po->flags |= OPF_TAIL;
3469 }
3470 po->pp = pp;
3471 continue;
3472 }
3473
3474 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
3475 continue;
3476
3477 if (po->operand[0].type == OPT_REGMEM) {
3478 pd = try_resolve_jumptab(i, opcnt);
3479 if (pd == NULL)
3480 goto tailcall;
3481
3482 po->btj = pd;
3483 continue;
3484 }
3485
3486 for (l = 0; l < opcnt; l++) {
3487 if (g_labels[l][0] && IS(po->operand[0].name, g_labels[l])) {
3488 if (l == i + 1 && po->op == OP_JMP) {
3489 // yet another alignment type..
3490 po->flags |= OPF_RMD;
3491 break;
3492 }
3493 add_label_ref(&g_label_refs[l], i);
3494 po->bt_i = l;
3495 break;
3496 }
3497 }
3498
3499 if (po->bt_i != -1 || (po->flags & OPF_RMD))
3500 continue;
3501
3502 if (po->operand[0].type == OPT_LABEL)
3503 // assume tail call
3504 goto tailcall;
3505
3506 ferr(po, "unhandled branch\n");
3507
3508tailcall:
3509 po->op = OP_CALL;
3510 po->flags |= OPF_TAIL;
3511 if (i > 0 && ops[i - 1].op == OP_POP)
3512 po->flags |= OPF_ATAIL;
3513 i--; // reprocess
3514 }
3515
3516 // pass3:
3517 // - remove dead labels
3518 // - process calls
3519 for (i = 0; i < opcnt; i++)
3520 {
3521 if (g_labels[i][0] != 0 && g_label_refs[i].i == -1)
3522 g_labels[i][0] = 0;
3523
3524 po = &ops[i];
3525 if (po->flags & OPF_RMD)
3526 continue;
3527
3528 if (po->op == OP_CALL)
3529 {
3530 tmpname = opr_name(po, 0);
3531 pp = po->pp;
3532 if (pp == NULL)
3533 {
3534 // indirect call
3535 pp_c = resolve_icall(i, opcnt, &l);
3536 if (pp_c != NULL) {
3537 if (!pp_c->is_func && !pp_c->is_fptr)
3538 ferr(po, "call to non-func: %s\n", pp_c->name);
3539 pp = proto_clone(pp_c);
3540 my_assert_not(pp, NULL);
3541 if (l)
3542 // not resolved just to single func
3543 pp->is_fptr = 1;
3544
3545 switch (po->operand[0].type) {
3546 case OPT_REG:
3547 // we resolved this call and no longer need the register
3548 po->regmask_src &= ~(1 << po->operand[0].reg);
3549 break;
3550 case OPT_REGMEM:
3551 pp->is_fptr = 1;
3552 break;
3553 default:
3554 break;
3555 }
3556 }
3557 if (pp == NULL) {
3558 pp = calloc(1, sizeof(*pp));
3559 my_assert_not(pp, NULL);
3560 pp->is_fptr = 1;
3561 ret = scan_for_esp_adjust(i + 1, opcnt, &j, &l);
3562 if (ret < 0) {
3563 if (!g_allow_regfunc)
3564 ferr(po, "non-__cdecl indirect call unhandled yet\n");
3565 pp->is_unresolved = 1;
3566 j = 0;
3567 }
3568 j /= 4;
3569 if (j > ARRAY_SIZE(pp->arg))
3570 ferr(po, "esp adjust too large: %d\n", j);
3571 pp->ret_type.name = strdup("int");
3572 pp->argc = pp->argc_stack = j;
3573 for (arg = 0; arg < pp->argc; arg++)
3574 pp->arg[arg].type.name = strdup("int");
3575 }
3576 po->pp = pp;
3577 }
3578
3579 // look for and make use of esp adjust
3580 ret = -1;
3581 if (!pp->is_stdcall && pp->argc_stack > 0)
3582 ret = scan_for_esp_adjust(i + 1, opcnt, &j, &l);
3583 if (ret >= 0) {
3584 if (pp->is_vararg) {
3585 if (j / 4 < pp->argc_stack)
3586 ferr(po, "esp adjust is too small: %x < %x\n",
3587 j, pp->argc_stack * 4);
3588 // modify pp to make it have varargs as normal args
3589 arg = pp->argc;
3590 pp->argc += j / 4 - pp->argc_stack;
3591 for (; arg < pp->argc; arg++) {
3592 pp->arg[arg].type.name = strdup("int");
3593 pp->argc_stack++;
3594 }
3595 if (pp->argc > ARRAY_SIZE(pp->arg))
3596 ferr(po, "too many args for '%s'\n", tmpname);
3597 }
3598 if (pp->argc_stack != j / 4)
3599 ferr(po, "stack tracking failed for '%s': %x %x\n",
3600 tmpname, pp->argc_stack * 4, j);
3601
3602 ops[ret].flags |= OPF_RMD;
3603 if (ops[ret].op == OP_POP && j > 4) {
3604 // deal with multi-pop stack adjust
3605 j = pp->argc_stack;
3606 while (ops[ret].op == OP_POP && j > 0 && ret < opcnt) {
3607 ops[ret].flags |= OPF_RMD;
3608 j--;
3609 ret++;
3610 }
3611 }
3612 else if (!l) {
3613 // a bit of a hack, but deals with use of
3614 // single adj for multiple calls
3615 ops[ret].operand[1].val -= j;
3616 }
3617 }
3618 else if (pp->is_vararg)
3619 ferr(po, "missing esp_adjust for vararg func '%s'\n",
3620 pp->name);
3621
3622 if (!pp->is_unresolved && !(po->flags & OPF_ATAIL)) {
3623 // since we know the args, collect them
3624 collect_call_args(po, i, pp, &regmask, save_arg_vars,
3625 i + opcnt * 2);
3626 }
3627
3628 if (strstr(pp->ret_type.name, "int64"))
3629 need_tmp64 = 1;
3630 }
3631 }
3632
3633 // pass4:
3634 // - find POPs for PUSHes, rm both
3635 // - scan for STD/CLD, propagate DF
3636 // - scan for all used registers
3637 // - find flag set ops for their users
3638 // - do unreselved calls
3639 // - declare indirect functions
3640 for (i = 0; i < opcnt; i++) {
3641 po = &ops[i];
3642 if (po->flags & OPF_RMD)
3643 continue;
3644
3645 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE)) {
3646 reg = po->operand[0].reg;
3647 if (!(regmask & (1 << reg)))
3648 // not a reg save after all, rerun scan_for_pop
3649 po->flags &= ~OPF_RSAVE;
3650 else
3651 regmask_save |= 1 << reg;
3652 }
3653
3654 if (po->op == OP_PUSH && po->p_argnum == 0
3655 && !(po->flags & OPF_RSAVE) && !g_func_pp->is_userstack)
3656 {
3657 if (po->operand[0].type == OPT_REG)
3658 {
3659 reg = po->operand[0].reg;
3660 if (reg < 0)
3661 ferr(po, "reg not set for push?\n");
3662
3663 depth = 0;
3664 ret = scan_for_pop(i + 1, opcnt,
3665 po->operand[0].name, i + opcnt * 3, 0, &depth, 0);
3666 if (ret == 1) {
3667 if (depth > 1)
3668 ferr(po, "too much depth: %d\n", depth);
3669
3670 po->flags |= OPF_RMD;
3671 scan_for_pop(i + 1, opcnt, po->operand[0].name,
3672 i + opcnt * 4, 0, &depth, 1);
3673 continue;
3674 }
3675 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
3676 if (ret == 0) {
3677 arg = OPF_RMD;
3678 if (regmask & (1 << reg)) {
3679 if (regmask_save & (1 << reg))
3680 ferr(po, "%s already saved?\n", po->operand[0].name);
3681 arg = OPF_RSAVE;
3682 }
3683 po->flags |= arg;
3684 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, arg);
3685 continue;
3686 }
3687 }
3688 else if (po->operand[0].type == OPT_CONST) {
3689 for (j = i + 1; j < opcnt; j++) {
3690 if ((ops[j].flags & (OPF_JMP|OPF_TAIL|OPF_RSAVE))
3691 || ops[j].op == OP_PUSH || g_labels[i][0] != 0)
3692 {
3693 break;
3694 }
3695
3696 if (!(ops[j].flags & OPF_RMD) && ops[j].op == OP_POP)
3697 {
3698 po->flags |= OPF_RMD;
3699 ops[j].datap = po;
3700 break;
3701 }
3702 }
3703 }
3704 }
3705
3706 if (po->op == OP_STD) {
3707 po->flags |= OPF_DF | OPF_RMD;
3708 scan_propagate_df(i + 1, opcnt);
3709 }
3710
3711 regmask_now = po->regmask_src | po->regmask_dst;
3712 if (regmask_now & (1 << xBP)) {
3713 if (g_bp_frame && !(po->flags & OPF_EBP_S)) {
3714 if (po->regmask_dst & (1 << xBP))
3715 // compiler decided to drop bp frame and use ebp as scratch
3716 scan_fwd_set_flags(i + 1, opcnt, i + opcnt * 5, OPF_EBP_S);
3717 else
3718 regmask_now &= ~(1 << xBP);
3719 }
3720 }
3721
3722 regmask |= regmask_now;
3723
3724 if (po->flags & OPF_CC)
3725 {
3726 int setters[16], cnt = 0, branched = 0;
3727
3728 ret = scan_for_flag_set(i, i + opcnt * 6,
3729 &branched, setters, &cnt);
3730 if (ret < 0 || cnt <= 0)
3731 ferr(po, "unable to trace flag setter(s)\n");
3732 if (cnt > ARRAY_SIZE(setters))
3733 ferr(po, "too many flag setters\n");
3734
3735 for (j = 0; j < cnt; j++)
3736 {
3737 tmp_op = &ops[setters[j]]; // flag setter
3738 pfomask = 0;
3739
3740 // to get nicer code, we try to delay test and cmp;
3741 // if we can't because of operand modification, or if we
3742 // have arith op, or branch, make it calculate flags explicitly
3743 if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP)
3744 {
3745 if (branched || scan_for_mod(tmp_op, setters[j] + 1, i, 0) >= 0)
3746 pfomask = 1 << po->pfo;
3747 }
3748 else if (tmp_op->op == OP_CMPS || tmp_op->op == OP_SCAS) {
3749 pfomask = 1 << po->pfo;
3750 }
3751 else {
3752 // see if we'll be able to handle based on op result
3753 if ((tmp_op->op != OP_AND && tmp_op->op != OP_OR
3754 && po->pfo != PFO_Z && po->pfo != PFO_S
3755 && po->pfo != PFO_P)
3756 || branched
3757 || scan_for_mod_opr0(tmp_op, setters[j] + 1, i) >= 0)
3758 {
3759 pfomask = 1 << po->pfo;
3760 }
3761
3762 if (tmp_op->op == OP_ADD && po->pfo == PFO_C) {
3763 propagate_lmod(tmp_op, &tmp_op->operand[0],
3764 &tmp_op->operand[1]);
3765 if (tmp_op->operand[0].lmod == OPLM_DWORD)
3766 need_tmp64 = 1;
3767 }
3768 }
3769 if (pfomask) {
3770 tmp_op->pfomask |= pfomask;
3771 cond_vars |= pfomask;
3772 }
3773 // note: may overwrite, currently not a problem
3774 po->datap = tmp_op;
3775 }
3776
3777 if (po->op == OP_RCL || po->op == OP_RCR
3778 || po->op == OP_ADC || po->op == OP_SBB)
3779 cond_vars |= 1 << PFO_C;
3780 }
3781
3782 if (po->op == OP_CMPS || po->op == OP_SCAS) {
3783 cond_vars |= 1 << PFO_Z;
3784 }
3785 else if (po->op == OP_MUL
3786 || (po->op == OP_IMUL && po->operand_cnt == 1))
3787 {
3788 if (po->operand[0].lmod == OPLM_DWORD)
3789 need_tmp64 = 1;
3790 }
3791 else if (po->op == OP_CALL) {
3792 pp = po->pp;
3793 if (pp == NULL)
3794 ferr(po, "NULL pp\n");
3795
3796 if (pp->is_unresolved) {
3797 int regmask_stack = 0;
3798 collect_call_args(po, i, pp, &regmask, save_arg_vars,
3799 i + opcnt * 2);
3800
3801 // this is pretty rough guess:
3802 // see ecx and edx were pushed (and not their saved versions)
3803 for (arg = 0; arg < pp->argc; arg++) {
3804 if (pp->arg[arg].reg != NULL)
3805 continue;
3806
3807 tmp_op = pp->arg[arg].datap;
3808 if (tmp_op == NULL)
3809 ferr(po, "parsed_op missing for arg%d\n", arg);
3810 if (tmp_op->p_argnum == 0 && tmp_op->operand[0].type == OPT_REG)
3811 regmask_stack |= 1 << tmp_op->operand[0].reg;
3812 }
3813
3814 if (!((regmask_stack & (1 << xCX))
3815 && (regmask_stack & (1 << xDX))))
3816 {
3817 if (pp->argc_stack != 0
3818 || ((regmask | regmask_arg) & ((1 << xCX)|(1 << xDX))))
3819 {
3820 pp_insert_reg_arg(pp, "ecx");
3821 pp->is_fastcall = 1;
3822 regmask_init |= 1 << xCX;
3823 regmask |= 1 << xCX;
3824 }
3825 if (pp->argc_stack != 0
3826 || ((regmask | regmask_arg) & (1 << xDX)))
3827 {
3828 pp_insert_reg_arg(pp, "edx");
3829 regmask_init |= 1 << xDX;
3830 regmask |= 1 << xDX;
3831 }
3832 }
3833
3834 // note: __cdecl doesn't fall into is_unresolved category
3835 if (pp->argc_stack > 0)
3836 pp->is_stdcall = 1;
3837 }
3838
3839 for (arg = 0; arg < pp->argc; arg++) {
3840 if (pp->arg[arg].reg != NULL) {
3841 reg = char_array_i(regs_r32,
3842 ARRAY_SIZE(regs_r32), pp->arg[arg].reg);
3843 if (reg < 0)
3844 ferr(ops, "arg '%s' is not a reg?\n", pp->arg[arg].reg);
3845 if (!(regmask & (1 << reg))) {
3846 regmask_init |= 1 << reg;
3847 regmask |= 1 << reg;
3848 }
3849 }
3850 }
3851 }
3852 else if (po->op == OP_MOV && po->operand[0].pp != NULL
3853 && po->operand[1].pp != NULL)
3854 {
3855 // <var> = offset <something>
3856 if ((po->operand[1].pp->is_func || po->operand[1].pp->is_fptr)
3857 && !IS_START(po->operand[1].name, "off_"))
3858 {
3859 if (!po->operand[0].pp->is_fptr)
3860 ferr(po, "%s not declared as fptr when it should be\n",
3861 po->operand[0].name);
3862 if (pp_cmp_func(po->operand[0].pp, po->operand[1].pp)) {
3863 pp_print(buf1, sizeof(buf1), po->operand[0].pp);
3864 pp_print(buf2, sizeof(buf2), po->operand[1].pp);
3865 fnote(po, "var: %s\n", buf1);
3866 fnote(po, "func: %s\n", buf2);
3867 ferr(po, "^ mismatch\n");
3868 }
3869 }
3870 }
3871 else if (po->op == OP_RET && !IS(g_func_pp->ret_type.name, "void"))
3872 regmask |= 1 << xAX;
3873 else if (po->op == OP_DIV || po->op == OP_IDIV) {
3874 // 32bit division is common, look for it
3875 if (po->op == OP_DIV)
3876 ret = scan_for_reg_clear(i, xDX);
3877 else
3878 ret = scan_for_cdq_edx(i);
3879 if (ret >= 0)
3880 po->flags |= OPF_32BIT;
3881 else
3882 need_tmp64 = 1;
3883 }
3884 else if (po->op == OP_CLD)
3885 po->flags |= OPF_RMD;
3886
3887 if (po->op == OP_RCL || po->op == OP_RCR || po->op == OP_XCHG) {
3888 need_tmp_var = 1;
3889 }
3890 }
3891
3892 // pass4:
3893 // - confirm regmask_save, it might have been reduced
3894 if (regmask_save != 0)
3895 {
3896 regmask_save = 0;
3897 for (i = 0; i < opcnt; i++) {
3898 po = &ops[i];
3899 if (po->flags & OPF_RMD)
3900 continue;
3901
3902 if (po->op == OP_PUSH && (po->flags & OPF_RSAVE))
3903 regmask_save |= 1 << po->operand[0].reg;
3904 }
3905 }
3906
3907 // output starts here
3908
3909 // define userstack size
3910 if (g_func_pp->is_userstack) {
3911 fprintf(fout, "#ifndef US_SZ_%s\n", g_func_pp->name);
3912 fprintf(fout, "#define US_SZ_%s USERSTACK_SIZE\n", g_func_pp->name);
3913 fprintf(fout, "#endif\n");
3914 }
3915
3916 // the function itself
3917 fprintf(fout, "%s ", g_func_pp->ret_type.name);
3918 output_pp_attrs(fout, g_func_pp, g_ida_func_attr & IDAFA_NORETURN);
3919 fprintf(fout, "%s(", g_func_pp->name);
3920
3921 for (i = 0; i < g_func_pp->argc; i++) {
3922 if (i > 0)
3923 fprintf(fout, ", ");
3924 if (g_func_pp->arg[i].fptr != NULL) {
3925 // func pointer..
3926 pp = g_func_pp->arg[i].fptr;
3927 fprintf(fout, "%s (", pp->ret_type.name);
3928 output_pp_attrs(fout, pp, 0);
3929 fprintf(fout, "*a%d)(", i + 1);
3930 for (j = 0; j < pp->argc; j++) {
3931 if (j > 0)
3932 fprintf(fout, ", ");
3933 if (pp->arg[j].fptr)
3934 ferr(ops, "nested fptr\n");
3935 fprintf(fout, "%s", pp->arg[j].type.name);
3936 }
3937 if (pp->is_vararg) {
3938 if (j > 0)
3939 fprintf(fout, ", ");
3940 fprintf(fout, "...");
3941 }
3942 fprintf(fout, ")");
3943 }
3944 else if (g_func_pp->arg[i].type.is_retreg) {
3945 fprintf(fout, "u32 *r_%s", g_func_pp->arg[i].reg);
3946 }
3947 else {
3948 fprintf(fout, "%s a%d", g_func_pp->arg[i].type.name, i + 1);
3949 }
3950 }
3951 if (g_func_pp->is_vararg) {
3952 if (i > 0)
3953 fprintf(fout, ", ");
3954 fprintf(fout, "...");
3955 }
3956
3957 fprintf(fout, ")\n{\n");
3958
3959 // declare indirect functions
3960 for (i = 0; i < opcnt; i++) {
3961 po = &ops[i];
3962 if (po->flags & OPF_RMD)
3963 continue;
3964
3965 if (po->op == OP_CALL) {
3966 pp = po->pp;
3967 if (pp == NULL)
3968 ferr(po, "NULL pp\n");
3969
3970 if (pp->is_fptr && !(pp->name[0] != 0 && pp->is_arg)) {
3971 if (pp->name[0] != 0) {
3972 memmove(pp->name + 2, pp->name, strlen(pp->name) + 1);
3973 memcpy(pp->name, "i_", 2);
3974
3975 // might be declared already
3976 found = 0;
3977 for (j = 0; j < i; j++) {
3978 if (ops[j].op == OP_CALL && (pp_tmp = ops[j].pp)) {
3979 if (pp_tmp->is_fptr && IS(pp->name, pp_tmp->name)) {
3980 found = 1;
3981 break;
3982 }
3983 }
3984 }
3985 if (found)
3986 continue;
3987 }
3988 else
3989 snprintf(pp->name, sizeof(pp->name), "icall%d", i);
3990
3991 fprintf(fout, " %s (", pp->ret_type.name);
3992 output_pp_attrs(fout, pp, 0);
3993 fprintf(fout, "*%s)(", pp->name);
3994 for (j = 0; j < pp->argc; j++) {
3995 if (j > 0)
3996 fprintf(fout, ", ");
3997 fprintf(fout, "%s a%d", pp->arg[j].type.name, j + 1);
3998 }
3999 fprintf(fout, ");\n");
4000 }
4001 }
4002 }
4003
4004 // output LUTs/jumptables
4005 for (i = 0; i < g_func_pd_cnt; i++) {
4006 pd = &g_func_pd[i];
4007 fprintf(fout, " static const ");
4008 if (pd->type == OPT_OFFSET) {
4009 fprintf(fout, "void *jt_%s[] =\n { ", pd->label);
4010
4011 for (j = 0; j < pd->count; j++) {
4012 if (j > 0)
4013 fprintf(fout, ", ");
4014 fprintf(fout, "&&%s", pd->d[j].u.label);
4015 }
4016 }
4017 else {
4018 fprintf(fout, "%s %s[] =\n { ",
4019 lmod_type_u(ops, pd->lmod), pd->label);
4020
4021 for (j = 0; j < pd->count; j++) {
4022 if (j > 0)
4023 fprintf(fout, ", ");
4024 fprintf(fout, "%u", pd->d[j].u.val);
4025 }
4026 }
4027 fprintf(fout, " };\n");
4028 had_decl = 1;
4029 }
4030
4031 // declare stack frame, va_arg
4032 if (g_stack_fsz) {
4033 fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n",
4034 (g_stack_fsz + 3) / 4, (g_stack_fsz + 1) / 2, g_stack_fsz);
4035 had_decl = 1;
4036 }
4037
4038 if (g_func_pp->is_userstack) {
4039 fprintf(fout, " u32 fake_sf[US_SZ_%s / 4];\n", g_func_pp->name);
4040 fprintf(fout, " u32 *esp = &fake_sf[sizeof(fake_sf) / 4];\n");
4041 had_decl = 1;
4042 }
4043
4044 if (g_func_pp->is_vararg) {
4045 fprintf(fout, " va_list ap;\n");
4046 had_decl = 1;
4047 }
4048
4049 // declare arg-registers
4050 for (i = 0; i < g_func_pp->argc; i++) {
4051 if (g_func_pp->arg[i].reg != NULL) {
4052 reg = char_array_i(regs_r32,
4053 ARRAY_SIZE(regs_r32), g_func_pp->arg[i].reg);
4054 if (regmask & (1 << reg)) {
4055 if (g_func_pp->arg[i].type.is_retreg)
4056 fprintf(fout, " u32 %s = *r_%s;\n",
4057 g_func_pp->arg[i].reg, g_func_pp->arg[i].reg);
4058 else
4059 fprintf(fout, " u32 %s = (u32)a%d;\n",
4060 g_func_pp->arg[i].reg, i + 1);
4061 }
4062 else {
4063 if (g_func_pp->arg[i].type.is_retreg)
4064 ferr(ops, "retreg '%s' is unused?\n",
4065 g_func_pp->arg[i].reg);
4066 fprintf(fout, " // %s = a%d; // unused\n",
4067 g_func_pp->arg[i].reg, i + 1);
4068 }
4069 had_decl = 1;
4070 }
4071 }
4072
4073 regmask_now = regmask & ~regmask_arg;
4074 regmask_now &= ~(1 << xSP);
4075 if (regmask_now & 0x00ff) {
4076 for (reg = 0; reg < 8; reg++) {
4077 if (regmask_now & (1 << reg)) {
4078 fprintf(fout, " u32 %s", regs_r32[reg]);
4079 if (regmask_init & (1 << reg))
4080 fprintf(fout, " = 0");
4081 fprintf(fout, ";\n");
4082 had_decl = 1;
4083 }
4084 }
4085 }
4086 if (regmask_now & 0xff00) {
4087 for (reg = 8; reg < 16; reg++) {
4088 if (regmask_now & (1 << reg)) {
4089 fprintf(fout, " mmxr %s", regs_r32[reg]);
4090 if (regmask_init & (1 << reg))
4091 fprintf(fout, " = { 0, }");
4092 fprintf(fout, ";\n");
4093 had_decl = 1;
4094 }
4095 }
4096 }
4097
4098 if (regmask_save) {
4099 for (reg = 0; reg < 8; reg++) {
4100 if (regmask_save & (1 << reg)) {
4101 fprintf(fout, " u32 s_%s;\n", regs_r32[reg]);
4102 had_decl = 1;
4103 }
4104 }
4105 }
4106
4107 for (i = 0; i < ARRAY_SIZE(save_arg_vars); i++) {
4108 if (save_arg_vars[i] == 0)
4109 continue;
4110 for (reg = 0; reg < 32; reg++) {
4111 if (save_arg_vars[i] & (1 << reg)) {
4112 fprintf(fout, " u32 %s;\n",
4113 saved_arg_name(buf1, sizeof(buf1), i, reg + 1));
4114 had_decl = 1;
4115 }
4116 }
4117 }
4118
4119 if (cond_vars) {
4120 for (i = 0; i < 8; i++) {
4121 if (cond_vars & (1 << i)) {
4122 fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]);
4123 had_decl = 1;
4124 }
4125 }
4126 }
4127
4128 if (need_tmp_var) {
4129 fprintf(fout, " u32 tmp;\n");
4130 had_decl = 1;
4131 }
4132
4133 if (need_tmp64) {
4134 fprintf(fout, " u64 tmp64;\n");
4135 had_decl = 1;
4136 }
4137
4138 if (had_decl)
4139 fprintf(fout, "\n");
4140
4141 if (g_func_pp->is_vararg) {
4142 if (g_func_pp->argc_stack == 0)
4143 ferr(ops, "vararg func without stack args?\n");
4144 fprintf(fout, " va_start(ap, a%d);\n", g_func_pp->argc);
4145 }
4146
4147 // output ops
4148 for (i = 0; i < opcnt; i++)
4149 {
4150 if (g_labels[i][0] != 0) {
4151 fprintf(fout, "\n%s:\n", g_labels[i]);
4152 label_pending = 1;
4153
4154 delayed_flag_op = NULL;
4155 last_arith_dst = NULL;
4156 }
4157
4158 po = &ops[i];
4159 if (po->flags & OPF_RMD)
4160 continue;
4161
4162 no_output = 0;
4163
4164 #define assert_operand_cnt(n_) \
4165 if (po->operand_cnt != n_) \
4166 ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_)
4167
4168 // conditional/flag using op?
4169 if (po->flags & OPF_CC)
4170 {
4171 int is_delayed = 0;
4172
4173 tmp_op = po->datap;
4174
4175 // we go through all this trouble to avoid using parsed_flag_op,
4176 // which makes generated code much nicer
4177 if (delayed_flag_op != NULL)
4178 {
4179 out_cmp_test(buf1, sizeof(buf1), delayed_flag_op,
4180 po->pfo, po->pfo_inv);
4181 is_delayed = 1;
4182 }
4183 else if (last_arith_dst != NULL
4184 && (po->pfo == PFO_Z || po->pfo == PFO_S || po->pfo == PFO_P
4185 || (tmp_op && (tmp_op->op == OP_AND || tmp_op->op == OP_OR))
4186 ))
4187 {
4188 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
4189 out_test_for_cc(buf1, sizeof(buf1), po, po->pfo, po->pfo_inv,
4190 last_arith_dst->lmod, buf3);
4191 is_delayed = 1;
4192 }
4193 else if (tmp_op != NULL) {
4194 // use preprocessed flag calc results
4195 if (!(tmp_op->pfomask & (1 << po->pfo)))
4196 ferr(po, "not prepared for pfo %d\n", po->pfo);
4197
4198 // note: pfo_inv was not yet applied
4199 snprintf(buf1, sizeof(buf1), "(%scond_%s)",
4200 po->pfo_inv ? "!" : "", parsed_flag_op_names[po->pfo]);
4201 }
4202 else {
4203 ferr(po, "all methods of finding comparison failed\n");
4204 }
4205
4206 if (po->flags & OPF_JMP) {
4207 fprintf(fout, " if %s", buf1);
4208 }
4209 else if (po->op == OP_RCL || po->op == OP_RCR
4210 || po->op == OP_ADC || po->op == OP_SBB)
4211 {
4212 if (is_delayed)
4213 fprintf(fout, " cond_%s = %s;\n",
4214 parsed_flag_op_names[po->pfo], buf1);
4215 }
4216 else if (po->flags & OPF_DATA) { // SETcc
4217 out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]);
4218 fprintf(fout, " %s = %s;", buf2, buf1);
4219 }
4220 else {
4221 ferr(po, "unhandled conditional op\n");
4222 }
4223 }
4224
4225 pfomask = po->pfomask;
4226
4227 if (po->flags & (OPF_REPZ|OPF_REPNZ)) {
4228 struct parsed_opr opr = {0,};
4229 opr.type = OPT_REG;
4230 opr.reg = xCX;
4231 opr.lmod = OPLM_DWORD;
4232 ret = try_resolve_const(i, &opr, opcnt * 7 + i, &uval);
4233
4234 if (ret != 1 || uval == 0) {
4235 // we need initial flags for ecx=0 case..
4236 if (i > 0 && ops[i - 1].op == OP_XOR
4237 && IS(ops[i - 1].operand[0].name,
4238 ops[i - 1].operand[1].name))
4239 {
4240 fprintf(fout, " cond_z = ");
4241 if (pfomask & (1 << PFO_C))
4242 fprintf(fout, "cond_c = ");
4243 fprintf(fout, "0;\n");
4244 }
4245 else if (last_arith_dst != NULL) {
4246 out_src_opr_u32(buf3, sizeof(buf3), po, last_arith_dst);
4247 out_test_for_cc(buf1, sizeof(buf1), po, PFO_Z, 0,
4248 last_arith_dst->lmod, buf3);
4249 fprintf(fout, " cond_z = %s;\n", buf1);
4250 }
4251 else
4252 ferr(po, "missing initial ZF\n");
4253 }
4254 }
4255
4256 switch (po->op)
4257 {
4258 case OP_MOV:
4259 assert_operand_cnt(2);
4260 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4261 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4262 default_cast_to(buf3, sizeof(buf3), &po->operand[0]);
4263 fprintf(fout, " %s = %s;", buf1,
4264 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4265 buf3, 0));
4266 break;
4267
4268 case OP_LEA:
4269 assert_operand_cnt(2);
4270 po->operand[1].lmod = OPLM_DWORD; // always
4271 fprintf(fout, " %s = %s;",
4272 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4273 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4274 NULL, 1));
4275 break;
4276
4277 case OP_MOVZX:
4278 assert_operand_cnt(2);
4279 fprintf(fout, " %s = %s;",
4280 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4281 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4282 break;
4283
4284 case OP_MOVSX:
4285 assert_operand_cnt(2);
4286 switch (po->operand[1].lmod) {
4287 case OPLM_BYTE:
4288 strcpy(buf3, "(s8)");
4289 break;
4290 case OPLM_WORD:
4291 strcpy(buf3, "(s16)");
4292 break;
4293 default:
4294 ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod);
4295 }
4296 fprintf(fout, " %s = %s;",
4297 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4298 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4299 buf3, 0));
4300 break;
4301
4302 case OP_XCHG:
4303 assert_operand_cnt(2);
4304 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4305 fprintf(fout, " tmp = %s;",
4306 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], "", 0));
4307 fprintf(fout, " %s = %s;",
4308 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4309 out_src_opr(buf2, sizeof(buf2), po, &po->operand[1],
4310 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
4311 fprintf(fout, " %s = %stmp;",
4312 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[1]),
4313 default_cast_to(buf3, sizeof(buf3), &po->operand[1]));
4314 snprintf(g_comment, sizeof(g_comment), "xchg");
4315 break;
4316
4317 case OP_NOT:
4318 assert_operand_cnt(1);
4319 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4320 fprintf(fout, " %s = ~%s;", buf1, buf1);
4321 break;
4322
4323 case OP_CDQ:
4324 assert_operand_cnt(2);
4325 fprintf(fout, " %s = (s32)%s >> 31;",
4326 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4327 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4328 strcpy(g_comment, "cdq");
4329 break;
4330
4331 case OP_LODS:
4332 assert_operand_cnt(3);
4333 if (po->flags & OPF_REP) {
4334 // hmh..
4335 ferr(po, "TODO\n");
4336 }
4337 else {
4338 fprintf(fout, " eax = %sesi; esi %c= %d;",
4339 lmod_cast_u_ptr(po, po->operand[0].lmod),
4340 (po->flags & OPF_DF) ? '-' : '+',
4341 lmod_bytes(po, po->operand[0].lmod));
4342 strcpy(g_comment, "lods");
4343 }
4344 break;
4345
4346 case OP_STOS:
4347 assert_operand_cnt(3);
4348 if (po->flags & OPF_REP) {
4349 fprintf(fout, " for (; ecx != 0; ecx--, edi %c= %d)\n",
4350 (po->flags & OPF_DF) ? '-' : '+',
4351 lmod_bytes(po, po->operand[0].lmod));
4352 fprintf(fout, " %sedi = eax;",
4353 lmod_cast_u_ptr(po, po->operand[0].lmod));
4354 strcpy(g_comment, "rep stos");
4355 }
4356 else {
4357 fprintf(fout, " %sedi = eax; edi %c= %d;",
4358 lmod_cast_u_ptr(po, po->operand[0].lmod),
4359 (po->flags & OPF_DF) ? '-' : '+',
4360 lmod_bytes(po, po->operand[0].lmod));
4361 strcpy(g_comment, "stos");
4362 }
4363 break;
4364
4365 case OP_MOVS:
4366 assert_operand_cnt(3);
4367 j = lmod_bytes(po, po->operand[0].lmod);
4368 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
4369 l = (po->flags & OPF_DF) ? '-' : '+';
4370 if (po->flags & OPF_REP) {
4371 fprintf(fout,
4372 " for (; ecx != 0; ecx--, edi %c= %d, esi %c= %d)\n",
4373 l, j, l, j);
4374 fprintf(fout,
4375 " %sedi = %sesi;", buf1, buf1);
4376 strcpy(g_comment, "rep movs");
4377 }
4378 else {
4379 fprintf(fout, " %sedi = %sesi; edi %c= %d; esi %c= %d;",
4380 buf1, buf1, l, j, l, j);
4381 strcpy(g_comment, "movs");
4382 }
4383 break;
4384
4385 case OP_CMPS:
4386 // repe ~ repeat while ZF=1
4387 assert_operand_cnt(3);
4388 j = lmod_bytes(po, po->operand[0].lmod);
4389 strcpy(buf1, lmod_cast_u_ptr(po, po->operand[0].lmod));
4390 l = (po->flags & OPF_DF) ? '-' : '+';
4391 if (po->flags & OPF_REP) {
4392 fprintf(fout,
4393 " for (; ecx != 0; ecx--) {\n");
4394 if (pfomask & (1 << PFO_C)) {
4395 // ugh..
4396 fprintf(fout,
4397 " cond_c = %sesi < %sedi;\n", buf1, buf1);
4398 pfomask &= ~(1 << PFO_C);
4399 }
4400 fprintf(fout,
4401 " cond_z = (%sesi == %sedi); esi %c= %d, edi %c= %d;\n",
4402 buf1, buf1, l, j, l, j);
4403 fprintf(fout,
4404 " if (cond_z %s 0) break;\n",
4405 (po->flags & OPF_REPZ) ? "==" : "!=");
4406 fprintf(fout,
4407 " }");
4408 snprintf(g_comment, sizeof(g_comment), "rep%s cmps",
4409 (po->flags & OPF_REPZ) ? "e" : "ne");
4410 }
4411 else {
4412 fprintf(fout,
4413 " cond_z = (%sesi == %sedi); esi %c= %d; edi %c= %d;",
4414 buf1, buf1, l, j, l, j);
4415 strcpy(g_comment, "cmps");
4416 }
4417 pfomask &= ~(1 << PFO_Z);
4418 last_arith_dst = NULL;
4419 delayed_flag_op = NULL;
4420 break;
4421
4422 case OP_SCAS:
4423 // only does ZF (for now)
4424 // repe ~ repeat while ZF=1
4425 assert_operand_cnt(3);
4426 j = lmod_bytes(po, po->operand[0].lmod);
4427 l = (po->flags & OPF_DF) ? '-' : '+';
4428 if (po->flags & OPF_REP) {
4429 fprintf(fout,
4430 " for (; ecx != 0; ecx--) {\n");
4431 fprintf(fout,
4432 " cond_z = (%seax == %sedi); edi %c= %d;\n",
4433 lmod_cast_u(po, po->operand[0].lmod),
4434 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
4435 fprintf(fout,
4436 " if (cond_z %s 0) break;\n",
4437 (po->flags & OPF_REPZ) ? "==" : "!=");
4438 fprintf(fout,
4439 " }");
4440 snprintf(g_comment, sizeof(g_comment), "rep%s scas",
4441 (po->flags & OPF_REPZ) ? "e" : "ne");
4442 }
4443 else {
4444 fprintf(fout, " cond_z = (%seax == %sedi); edi %c= %d;",
4445 lmod_cast_u(po, po->operand[0].lmod),
4446 lmod_cast_u_ptr(po, po->operand[0].lmod), l, j);
4447 strcpy(g_comment, "scas");
4448 }
4449 pfomask &= ~(1 << PFO_Z);
4450 last_arith_dst = NULL;
4451 delayed_flag_op = NULL;
4452 break;
4453
4454 // arithmetic w/flags
4455 case OP_AND:
4456 case OP_OR:
4457 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4458 // fallthrough
4459 dualop_arith:
4460 assert_operand_cnt(2);
4461 fprintf(fout, " %s %s= %s;",
4462 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4463 op_to_c(po),
4464 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4465 output_std_flags(fout, po, &pfomask, buf1);
4466 last_arith_dst = &po->operand[0];
4467 delayed_flag_op = NULL;
4468 break;
4469
4470 case OP_SHL:
4471 case OP_SHR:
4472 assert_operand_cnt(2);
4473 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4474 if (pfomask & (1 << PFO_C)) {
4475 if (po->operand[1].type == OPT_CONST) {
4476 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4477 j = po->operand[1].val;
4478 j %= l;
4479 if (j != 0) {
4480 if (po->op == OP_SHL)
4481 j = l - j;
4482 else
4483 j -= 1;
4484 fprintf(fout, " cond_c = (%s >> %d) & 1;\n",
4485 buf1, j);
4486 }
4487 else
4488 ferr(po, "zero shift?\n");
4489 }
4490 else
4491 ferr(po, "TODO\n");
4492 pfomask &= ~(1 << PFO_C);
4493 }
4494 fprintf(fout, " %s %s= %s;", buf1, op_to_c(po),
4495 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4496 output_std_flags(fout, po, &pfomask, buf1);
4497 last_arith_dst = &po->operand[0];
4498 delayed_flag_op = NULL;
4499 break;
4500
4501 case OP_SAR:
4502 assert_operand_cnt(2);
4503 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4504 fprintf(fout, " %s = %s%s >> %s;", buf1,
4505 lmod_cast_s(po, po->operand[0].lmod), buf1,
4506 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4507 output_std_flags(fout, po, &pfomask, buf1);
4508 last_arith_dst = &po->operand[0];
4509 delayed_flag_op = NULL;
4510 break;
4511
4512 case OP_SHRD:
4513 assert_operand_cnt(3);
4514 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4515 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4516 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4517 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4518 out_src_opr_u32(buf3, sizeof(buf3), po, &po->operand[2]);
4519 fprintf(fout, " %s >>= %s; %s |= %s << (%d - %s);",
4520 buf1, buf3, buf1, buf2, l, buf3);
4521 strcpy(g_comment, "shrd");
4522 output_std_flags(fout, po, &pfomask, buf1);
4523 last_arith_dst = &po->operand[0];
4524 delayed_flag_op = NULL;
4525 break;
4526
4527 case OP_ROL:
4528 case OP_ROR:
4529 assert_operand_cnt(2);
4530 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4531 if (po->operand[1].type == OPT_CONST) {
4532 j = po->operand[1].val;
4533 j %= lmod_bytes(po, po->operand[0].lmod) * 8;
4534 fprintf(fout, po->op == OP_ROL ?
4535 " %s = (%s << %d) | (%s >> %d);" :
4536 " %s = (%s >> %d) | (%s << %d);",
4537 buf1, buf1, j, buf1,
4538 lmod_bytes(po, po->operand[0].lmod) * 8 - j);
4539 }
4540 else
4541 ferr(po, "TODO\n");
4542 output_std_flags(fout, po, &pfomask, buf1);
4543 last_arith_dst = &po->operand[0];
4544 delayed_flag_op = NULL;
4545 break;
4546
4547 case OP_RCL:
4548 case OP_RCR:
4549 assert_operand_cnt(2);
4550 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4551 l = lmod_bytes(po, po->operand[0].lmod) * 8;
4552 if (po->operand[1].type == OPT_CONST) {
4553 j = po->operand[1].val % l;
4554 if (j == 0)
4555 ferr(po, "zero rotate\n");
4556 fprintf(fout, " tmp = (%s >> %d) & 1;\n",
4557 buf1, (po->op == OP_RCL) ? (l - j) : (j - 1));
4558 if (po->op == OP_RCL) {
4559 fprintf(fout,
4560 " %s = (%s << %d) | (cond_c << %d)",
4561 buf1, buf1, j, j - 1);
4562 if (j != 1)
4563 fprintf(fout, " | (%s >> %d)", buf1, l + 1 - j);
4564 }
4565 else {
4566 fprintf(fout,
4567 " %s = (%s >> %d) | (cond_c << %d)",
4568 buf1, buf1, j, l - j);
4569 if (j != 1)
4570 fprintf(fout, " | (%s << %d)", buf1, l + 1 - j);
4571 }
4572 fprintf(fout, ";\n");
4573 fprintf(fout, " cond_c = tmp;");
4574 }
4575 else
4576 ferr(po, "TODO\n");
4577 strcpy(g_comment, (po->op == OP_RCL) ? "rcl" : "rcr");
4578 output_std_flags(fout, po, &pfomask, buf1);
4579 last_arith_dst = &po->operand[0];
4580 delayed_flag_op = NULL;
4581 break;
4582
4583 case OP_XOR:
4584 assert_operand_cnt(2);
4585 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4586 if (IS(opr_name(po, 0), opr_name(po, 1))) {
4587 // special case for XOR
4588 if (pfomask & (1 << PFO_BE)) { // weird, but it happens..
4589 fprintf(fout, " cond_be = 1;\n");
4590 pfomask &= ~(1 << PFO_BE);
4591 }
4592 fprintf(fout, " %s = 0;",
4593 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4594 last_arith_dst = &po->operand[0];
4595 delayed_flag_op = NULL;
4596 break;
4597 }
4598 goto dualop_arith;
4599
4600 case OP_ADD:
4601 assert_operand_cnt(2);
4602 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4603 if (pfomask & (1 << PFO_C)) {
4604 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4605 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4606 if (po->operand[0].lmod == OPLM_DWORD) {
4607 fprintf(fout, " tmp64 = (u64)%s + %s;\n", buf1, buf2);
4608 fprintf(fout, " cond_c = tmp64 >> 32;\n");
4609 fprintf(fout, " %s = (u32)tmp64;",
4610 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
4611 strcat(g_comment, "add64");
4612 }
4613 else {
4614 fprintf(fout, " cond_c = ((u32)%s + %s) >> %d;\n",
4615 buf1, buf2, lmod_bytes(po, po->operand[0].lmod) * 8);
4616 fprintf(fout, " %s += %s;",
4617 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4618 buf2);
4619 }
4620 pfomask &= ~(1 << PFO_C);
4621 output_std_flags(fout, po, &pfomask, buf1);
4622 last_arith_dst = &po->operand[0];
4623 delayed_flag_op = NULL;
4624 break;
4625 }
4626 goto dualop_arith;
4627
4628 case OP_SUB:
4629 assert_operand_cnt(2);
4630 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4631 if (pfomask & ~((1 << PFO_Z) | (1 << PFO_S))) {
4632 for (j = 0; j <= PFO_LE; j++) {
4633 if (!(pfomask & (1 << j)))
4634 continue;
4635 if (j == PFO_Z || j == PFO_S)
4636 continue;
4637
4638 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
4639 fprintf(fout, " cond_%s = %s;\n",
4640 parsed_flag_op_names[j], buf1);
4641 pfomask &= ~(1 << j);
4642 }
4643 }
4644 goto dualop_arith;
4645
4646 case OP_ADC:
4647 case OP_SBB:
4648 assert_operand_cnt(2);
4649 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4650 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4651 if (po->op == OP_SBB
4652 && IS(po->operand[0].name, po->operand[1].name))
4653 {
4654 // avoid use of unitialized var
4655 fprintf(fout, " %s = -cond_c;", buf1);
4656 // carry remains what it was
4657 pfomask &= ~(1 << PFO_C);
4658 }
4659 else {
4660 fprintf(fout, " %s %s= %s + cond_c;", buf1, op_to_c(po),
4661 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]));
4662 }
4663 output_std_flags(fout, po, &pfomask, buf1);
4664 last_arith_dst = &po->operand[0];
4665 delayed_flag_op = NULL;
4666 break;
4667
4668 case OP_BSF:
4669 assert_operand_cnt(2);
4670 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[1]);
4671 fprintf(fout, " %s = %s ? __builtin_ffs(%s) - 1 : 0;",
4672 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]),
4673 buf2, buf2);
4674 output_std_flags(fout, po, &pfomask, buf1);
4675 last_arith_dst = &po->operand[0];
4676 delayed_flag_op = NULL;
4677 strcat(g_comment, "bsf");
4678 break;
4679
4680 case OP_DEC:
4681 if (pfomask & ~(PFOB_S | PFOB_S | PFOB_C)) {
4682 for (j = 0; j <= PFO_LE; j++) {
4683 if (!(pfomask & (1 << j)))
4684 continue;
4685 if (j == PFO_Z || j == PFO_S || j == PFO_C)
4686 continue;
4687
4688 out_cmp_for_cc(buf1, sizeof(buf1), po, j, 0);
4689 fprintf(fout, " cond_%s = %s;\n",
4690 parsed_flag_op_names[j], buf1);
4691 pfomask &= ~(1 << j);
4692 }
4693 }
4694 // fallthrough
4695
4696 case OP_INC:
4697 if (pfomask & (1 << PFO_C))
4698 // carry is unaffected by inc/dec.. wtf?
4699 ferr(po, "carry propagation needed\n");
4700
4701 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4702 if (po->operand[0].type == OPT_REG) {
4703 strcpy(buf2, po->op == OP_INC ? "++" : "--");
4704 fprintf(fout, " %s%s;", buf1, buf2);
4705 }
4706 else {
4707 strcpy(buf2, po->op == OP_INC ? "+" : "-");
4708 fprintf(fout, " %s %s= 1;", buf1, buf2);
4709 }
4710 output_std_flags(fout, po, &pfomask, buf1);
4711 last_arith_dst = &po->operand[0];
4712 delayed_flag_op = NULL;
4713 break;
4714
4715 case OP_NEG:
4716 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
4717 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]);
4718 fprintf(fout, " %s = -%s%s;", buf1,
4719 lmod_cast_s(po, po->operand[0].lmod), buf2);
4720 last_arith_dst = &po->operand[0];
4721 delayed_flag_op = NULL;
4722 if (pfomask & (1 << PFO_C)) {
4723 fprintf(fout, "\n cond_c = (%s != 0);", buf1);
4724 pfomask &= ~(1 << PFO_C);
4725 }
4726 break;
4727
4728 case OP_IMUL:
4729 if (po->operand_cnt == 2) {
4730 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4731 goto dualop_arith;
4732 }
4733 if (po->operand_cnt == 3)
4734 ferr(po, "TODO imul3\n");
4735 // fallthrough
4736 case OP_MUL:
4737 assert_operand_cnt(1);
4738 switch (po->operand[0].lmod) {
4739 case OPLM_DWORD:
4740 strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)");
4741 fprintf(fout, " tmp64 = %seax * %s%s;\n", buf1, buf1,
4742 out_src_opr_u32(buf2, sizeof(buf2), po, &po->operand[0]));
4743 fprintf(fout, " edx = tmp64 >> 32;\n");
4744 fprintf(fout, " eax = tmp64;");
4745 break;
4746 case OPLM_BYTE:
4747 strcpy(buf1, po->op == OP_IMUL ? "(s16)(s8)" : "(u16)(u8)");
4748 fprintf(fout, " LOWORD(eax) = %seax * %s;", buf1,
4749 out_src_opr(buf2, sizeof(buf2), po, &po->operand[0],
4750 buf1, 0));
4751 break;
4752 default:
4753 ferr(po, "TODO: unhandled mul type\n");
4754 break;
4755 }
4756 last_arith_dst = NULL;
4757 delayed_flag_op = NULL;
4758 break;
4759
4760 case OP_DIV:
4761 case OP_IDIV:
4762 assert_operand_cnt(1);
4763 if (po->operand[0].lmod != OPLM_DWORD)
4764 ferr(po, "unhandled lmod %d\n", po->operand[0].lmod);
4765
4766 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
4767 strcpy(buf2, lmod_cast(po, po->operand[0].lmod,
4768 po->op == OP_IDIV));
4769 switch (po->operand[0].lmod) {
4770 case OPLM_DWORD:
4771 if (po->flags & OPF_32BIT)
4772 snprintf(buf3, sizeof(buf3), "%seax", buf2);
4773 else {
4774 fprintf(fout, " tmp64 = ((u64)edx << 32) | eax;\n");
4775 snprintf(buf3, sizeof(buf3), "%stmp64",
4776 (po->op == OP_IDIV) ? "(s64)" : "");
4777 }
4778 if (po->operand[0].type == OPT_REG
4779 && po->operand[0].reg == xDX)
4780 {
4781 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
4782 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
4783 }
4784 else {
4785 fprintf(fout, " edx = %s %% %s%s;\n", buf3, buf2, buf1);
4786 fprintf(fout, " eax = %s / %s%s;", buf3, buf2, buf1);
4787 }
4788 break;
4789 default:
4790 ferr(po, "unhandled division type\n");
4791 }
4792 last_arith_dst = NULL;
4793 delayed_flag_op = NULL;
4794 break;
4795
4796 case OP_TEST:
4797 case OP_CMP:
4798 propagate_lmod(po, &po->operand[0], &po->operand[1]);
4799 if (pfomask != 0) {
4800 for (j = 0; j < 8; j++) {
4801 if (pfomask & (1 << j)) {
4802 out_cmp_test(buf1, sizeof(buf1), po, j, 0);
4803 fprintf(fout, " cond_%s = %s;",
4804 parsed_flag_op_names[j], buf1);
4805 }
4806 }
4807 pfomask = 0;
4808 }
4809 else
4810 no_output = 1;
4811 last_arith_dst = NULL;
4812 delayed_flag_op = po;
4813 break;
4814
4815 case OP_SCC:
4816 // SETcc - should already be handled
4817 break;
4818
4819 // note: we reuse OP_Jcc for SETcc, only flags differ
4820 case OP_JCC:
4821 fprintf(fout, "\n goto %s;", po->operand[0].name);
4822 break;
4823
4824 case OP_JECXZ:
4825 fprintf(fout, " if (ecx == 0)\n");
4826 fprintf(fout, " goto %s;", po->operand[0].name);
4827 strcat(g_comment, "jecxz");
4828 break;
4829
4830 case OP_JMP:
4831 assert_operand_cnt(1);
4832 last_arith_dst = NULL;
4833 delayed_flag_op = NULL;
4834
4835 if (po->operand[0].type == OPT_REGMEM) {
4836 ret = sscanf(po->operand[0].name, "%[^[][%[^*]*4]",
4837 buf1, buf2);
4838 if (ret != 2)
4839 ferr(po, "parse failure for jmp '%s'\n",
4840 po->operand[0].name);
4841 fprintf(fout, " goto *jt_%s[%s];", buf1, buf2);
4842 break;
4843 }
4844 else if (po->operand[0].type != OPT_LABEL)
4845 ferr(po, "unhandled jmp type\n");
4846
4847 fprintf(fout, " goto %s;", po->operand[0].name);
4848 break;
4849
4850 case OP_CALL:
4851 assert_operand_cnt(1);
4852 pp = po->pp;
4853 my_assert_not(pp, NULL);
4854
4855 strcpy(buf3, " ");
4856 if (po->flags & OPF_CC) {
4857 // we treat conditional branch to another func
4858 // (yes such code exists..) as conditional tailcall
4859 strcat(buf3, " ");
4860 fprintf(fout, " {\n");
4861 }
4862
4863 if (pp->is_fptr && !pp->is_arg) {
4864 fprintf(fout, "%s%s = %s;\n", buf3, pp->name,
4865 out_src_opr(buf1, sizeof(buf1), po, &po->operand[0],
4866 "(void *)", 0));
4867 if (pp->is_unresolved)
4868 fprintf(fout, "%sunresolved_call(\"%s:%d\", %s);\n",
4869 buf3, asmfn, po->asmln, pp->name);
4870 }
4871
4872 fprintf(fout, "%s", buf3);
4873 if (strstr(pp->ret_type.name, "int64")) {
4874 if (po->flags & OPF_TAIL)
4875 ferr(po, "int64 and tail?\n");
4876 fprintf(fout, "tmp64 = ");
4877 }
4878 else if (!IS(pp->ret_type.name, "void")) {
4879 if (po->flags & OPF_TAIL) {
4880 if (!IS(g_func_pp->ret_type.name, "void")) {
4881 fprintf(fout, "return ");
4882 if (g_func_pp->ret_type.is_ptr != pp->ret_type.is_ptr)
4883 fprintf(fout, "(%s)", g_func_pp->ret_type.name);
4884 }
4885 }
4886 else if (regmask & (1 << xAX)) {
4887 fprintf(fout, "eax = ");
4888 if (pp->ret_type.is_ptr)
4889 fprintf(fout, "(u32)");
4890 }
4891 }
4892
4893 if (pp->name[0] == 0)
4894 ferr(po, "missing pp->name\n");
4895 fprintf(fout, "%s%s(", pp->name,
4896 pp->has_structarg ? "_sa" : "");
4897
4898 if (po->flags & OPF_ATAIL) {
4899 if (pp->argc_stack != g_func_pp->argc_stack
4900 || (pp->argc_stack > 0
4901 && pp->is_stdcall != g_func_pp->is_stdcall))
4902 ferr(po, "incompatible tailcall\n");
4903 if (g_func_pp->has_retreg)
4904 ferr(po, "TODO: retreg+tailcall\n");
4905
4906 for (arg = j = 0; arg < pp->argc; arg++) {
4907 if (arg > 0)
4908 fprintf(fout, ", ");
4909
4910 cast[0] = 0;
4911 if (pp->arg[arg].type.is_ptr)
4912 snprintf(cast, sizeof(cast), "(%s)",
4913 pp->arg[arg].type.name);
4914
4915 if (pp->arg[arg].reg != NULL) {
4916 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
4917 continue;
4918 }
4919 // stack arg
4920 for (; j < g_func_pp->argc; j++)
4921 if (g_func_pp->arg[j].reg == NULL)
4922 break;
4923 fprintf(fout, "%sa%d", cast, j + 1);
4924 j++;
4925 }
4926 }
4927 else {
4928 for (arg = 0; arg < pp->argc; arg++) {
4929 if (arg > 0)
4930 fprintf(fout, ", ");
4931
4932 cast[0] = 0;
4933 if (pp->arg[arg].type.is_ptr)
4934 snprintf(cast, sizeof(cast), "(%s)",
4935 pp->arg[arg].type.name);
4936
4937 if (pp->arg[arg].reg != NULL) {
4938 if (pp->arg[arg].type.is_retreg)
4939 fprintf(fout, "&%s", pp->arg[arg].reg);
4940 else
4941 fprintf(fout, "%s%s", cast, pp->arg[arg].reg);
4942 continue;
4943 }
4944
4945 // stack arg
4946 tmp_op = pp->arg[arg].datap;
4947 if (tmp_op == NULL)
4948 ferr(po, "parsed_op missing for arg%d\n", arg);
4949
4950 if (tmp_op->flags & OPF_VAPUSH) {
4951 fprintf(fout, "ap");
4952 }
4953 else if (tmp_op->p_argpass != 0) {
4954 fprintf(fout, "a%d", tmp_op->p_argpass);
4955 }
4956 else if (tmp_op->p_argnum != 0) {
4957 fprintf(fout, "%s%s", cast,
4958 saved_arg_name(buf1, sizeof(buf1),
4959 tmp_op->p_arggrp, tmp_op->p_argnum));
4960 }
4961 else {
4962 fprintf(fout, "%s",
4963 out_src_opr(buf1, sizeof(buf1),
4964 tmp_op, &tmp_op->operand[0], cast, 0));
4965 }
4966 }
4967 }
4968 fprintf(fout, ");");
4969
4970 if (strstr(pp->ret_type.name, "int64")) {
4971 fprintf(fout, "\n");
4972 fprintf(fout, "%sedx = tmp64 >> 32;\n", buf3);
4973 fprintf(fout, "%seax = tmp64;", buf3);
4974 }
4975
4976 if (pp->is_unresolved) {
4977 snprintf(buf2, sizeof(buf2), " unresolved %dreg",
4978 pp->argc_reg);
4979 strcat(g_comment, buf2);
4980 }
4981
4982 if (po->flags & OPF_TAIL) {
4983 ret = 0;
4984 if (i == opcnt - 1 || pp->is_noreturn)
4985 ret = 0;
4986 else if (IS(pp->ret_type.name, "void"))
4987 ret = 1;
4988 else if (IS(g_func_pp->ret_type.name, "void"))
4989 ret = 1;
4990 // else already handled as 'return f()'
4991
4992 if (ret) {
4993 if (!IS(g_func_pp->ret_type.name, "void")) {
4994 ferr(po, "int func -> void func tailcall?\n");
4995 }
4996 else {
4997 fprintf(fout, "\n%sreturn;", buf3);
4998 strcat(g_comment, " ^ tailcall");
4999 }
5000 }
5001 else
5002 strcat(g_comment, " tailcall");
5003 }
5004 if (pp->is_noreturn)
5005 strcat(g_comment, " noreturn");
5006 if ((po->flags & OPF_ATAIL) && pp->argc_stack > 0)
5007 strcat(g_comment, " argframe");
5008 if (po->flags & OPF_CC)
5009 strcat(g_comment, " cond");
5010
5011 if (po->flags & OPF_CC)
5012 fprintf(fout, "\n }");
5013
5014 delayed_flag_op = NULL;
5015 last_arith_dst = NULL;
5016 break;
5017
5018 case OP_RET:
5019 if (g_func_pp->is_vararg)
5020 fprintf(fout, " va_end(ap);\n");
5021 if (g_func_pp->has_retreg) {
5022 for (arg = 0; arg < g_func_pp->argc; arg++)
5023 if (g_func_pp->arg[arg].type.is_retreg)
5024 fprintf(fout, " *r_%s = %s;\n",
5025 g_func_pp->arg[arg].reg, g_func_pp->arg[arg].reg);
5026 }
5027
5028 if (IS(g_func_pp->ret_type.name, "void")) {
5029 if (i != opcnt - 1 || label_pending)
5030 fprintf(fout, " return;");
5031 }
5032 else if (g_func_pp->ret_type.is_ptr) {
5033 fprintf(fout, " return (%s)eax;",
5034 g_func_pp->ret_type.name);
5035 }
5036 else if (IS(g_func_pp->ret_type.name, "__int64"))
5037 fprintf(fout, " return ((u64)edx << 32) | eax;");
5038 else
5039 fprintf(fout, " return eax;");
5040
5041 last_arith_dst = NULL;
5042 delayed_flag_op = NULL;
5043 break;
5044
5045 case OP_PUSH:
5046 out_src_opr_u32(buf1, sizeof(buf1), po, &po->operand[0]);
5047 if (po->p_argnum != 0) {
5048 // special case - saved func arg
5049 fprintf(fout, " %s = %s;",
5050 saved_arg_name(buf2, sizeof(buf2),
5051 po->p_arggrp, po->p_argnum), buf1);
5052 break;
5053 }
5054 else if (po->flags & OPF_RSAVE) {
5055 fprintf(fout, " s_%s = %s;", buf1, buf1);
5056 break;
5057 }
5058 else if (g_func_pp->is_userstack) {
5059 fprintf(fout, " *(--esp) = %s;", buf1);
5060 break;
5061 }
5062 if (!(g_ida_func_attr & IDAFA_NORETURN))
5063 ferr(po, "stray push encountered\n");
5064 no_output = 1;
5065 break;
5066
5067 case OP_POP:
5068 if (po->flags & OPF_RSAVE) {
5069 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5070 fprintf(fout, " %s = s_%s;", buf1, buf1);
5071 break;
5072 }
5073 else if (po->datap != NULL) {
5074 // push/pop pair
5075 tmp_op = po->datap;
5076 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]);
5077 fprintf(fout, " %s = %s;", buf1,
5078 out_src_opr(buf2, sizeof(buf2),
5079 tmp_op, &tmp_op->operand[0],
5080 default_cast_to(buf3, sizeof(buf3), &po->operand[0]), 0));
5081 break;
5082 }
5083 else if (g_func_pp->is_userstack) {
5084 fprintf(fout, " %s = *esp++;",
5085 out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]));
5086 break;
5087 }
5088 else
5089 ferr(po, "stray pop encountered\n");
5090 break;
5091
5092 case OP_NOP:
5093 no_output = 1;
5094 break;
5095
5096 // mmx
5097 case OP_EMMS:
5098 strcpy(g_comment, "(emms)");
5099 break;
5100
5101 default:
5102 no_output = 1;
5103 ferr(po, "unhandled op type %d, flags %x\n",
5104 po->op, po->flags);
5105 break;
5106 }
5107
5108 if (g_comment[0] != 0) {
5109 char *p = g_comment;
5110 while (my_isblank(*p))
5111 p++;
5112 fprintf(fout, " // %s", p);
5113 g_comment[0] = 0;
5114 no_output = 0;
5115 }
5116 if (!no_output)
5117 fprintf(fout, "\n");
5118
5119 // some sanity checking
5120 if (po->flags & OPF_REP) {
5121 if (po->op != OP_STOS && po->op != OP_MOVS
5122 && po->op != OP_CMPS && po->op != OP_SCAS)
5123 ferr(po, "unexpected rep\n");
5124 if (!(po->flags & (OPF_REPZ|OPF_REPNZ))
5125 && (po->op == OP_CMPS || po->op == OP_SCAS))
5126 ferr(po, "cmps/scas with plain rep\n");
5127 }
5128 if ((po->flags & (OPF_REPZ|OPF_REPNZ))
5129 && po->op != OP_CMPS && po->op != OP_SCAS)
5130 ferr(po, "unexpected repz/repnz\n");
5131
5132 if (pfomask != 0)
5133 ferr(po, "missed flag calc, pfomask=%x\n", pfomask);
5134
5135 // see is delayed flag stuff is still valid
5136 if (delayed_flag_op != NULL && delayed_flag_op != po) {
5137 if (is_any_opr_modified(delayed_flag_op, po, 0))
5138 delayed_flag_op = NULL;
5139 }
5140
5141 if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) {
5142 if (is_opr_modified(last_arith_dst, po))
5143 last_arith_dst = NULL;
5144 }
5145
5146 label_pending = 0;
5147 }
5148
5149 if (g_stack_fsz && !g_stack_frame_used)
5150 fprintf(fout, " (void)sf;\n");
5151
5152 fprintf(fout, "}\n\n");
5153
5154 // cleanup
5155 for (i = 0; i < opcnt; i++) {
5156 struct label_ref *lr, *lr_del;
5157
5158 lr = g_label_refs[i].next;
5159 while (lr != NULL) {
5160 lr_del = lr;
5161 lr = lr->next;
5162 free(lr_del);
5163 }
5164 g_label_refs[i].i = -1;
5165 g_label_refs[i].next = NULL;
5166
5167 if (ops[i].op == OP_CALL) {
5168 if (ops[i].pp)
5169 proto_release(ops[i].pp);
5170 }
5171 }
5172 g_func_pp = NULL;
5173}
5174
5175struct func_proto_dep;
5176
5177struct func_prototype {
5178 char name[NAMELEN];
5179 int id;
5180 int argc_stack;
5181 int regmask_dep;
5182 int has_ret:3; // -1, 0, 1: unresolved, no, yes
5183 unsigned int dep_resolved:1;
5184 unsigned int is_stdcall:1;
5185 struct func_proto_dep *dep_func;
5186 int dep_func_cnt;
5187};
5188
5189struct func_proto_dep {
5190 char *name;
5191 struct func_prototype *proto;
5192 int regmask_live; // .. at the time of call
5193 unsigned int ret_dep:1; // return from this is caller's return
5194};
5195
5196static struct func_prototype *hg_fp;
5197static int hg_fp_cnt;
5198
5199static struct func_proto_dep *hg_fp_find_dep(struct func_prototype *fp,
5200 const char *name)
5201{
5202 int i;
5203
5204 for (i = 0; i < fp->dep_func_cnt; i++)
5205 if (IS(fp->dep_func[i].name, name))
5206 return &fp->dep_func[i];
5207
5208 return NULL;
5209}
5210
5211static void hg_fp_add_dep(struct func_prototype *fp, const char *name)
5212{
5213 // is it a dupe?
5214 if (hg_fp_find_dep(fp, name))
5215 return;
5216
5217 if ((fp->dep_func_cnt & 0xff) == 0) {
5218 fp->dep_func = realloc(fp->dep_func,
5219 sizeof(fp->dep_func[0]) * (fp->dep_func_cnt + 0x100));
5220 my_assert_not(fp->dep_func, NULL);
5221 memset(&fp->dep_func[fp->dep_func_cnt], 0,
5222 sizeof(fp->dep_func[0]) * 0x100);
5223 }
5224 fp->dep_func[fp->dep_func_cnt].name = strdup(name);
5225 fp->dep_func_cnt++;
5226}
5227
5228static int hg_fp_cmp_name(const void *p1_, const void *p2_)
5229{
5230 const struct func_prototype *p1 = p1_, *p2 = p2_;
5231 return strcmp(p1->name, p2->name);
5232}
5233
5234#if 0
5235static int hg_fp_cmp_id(const void *p1_, const void *p2_)
5236{
5237 const struct func_prototype *p1 = p1_, *p2 = p2_;
5238 return p1->id - p2->id;
5239}
5240#endif
5241
5242static void gen_hdr(const char *funcn, int opcnt)
5243{
5244 struct func_prototype *fp;
5245 struct func_proto_dep *dep;
5246 struct parsed_data *pd;
5247 struct parsed_op *po;
5248 int regmask_save = 0;
5249 int regmask_dst = 0;
5250 int regmask_dep = 0;
5251 int max_bp_offset = 0;
5252 int has_ret = -1;
5253 int from_caller = 0;
5254 int i, j, l, ret;
5255 int depth, reg;
5256
5257 if ((hg_fp_cnt & 0xff) == 0) {
5258 hg_fp = realloc(hg_fp, sizeof(hg_fp[0]) * (hg_fp_cnt + 0x100));
5259 my_assert_not(hg_fp, NULL);
5260 memset(hg_fp + hg_fp_cnt, 0, sizeof(hg_fp[0]) * 0x100);
5261 }
5262
5263 fp = &hg_fp[hg_fp_cnt];
5264 snprintf(fp->name, sizeof(fp->name), "%s", funcn);
5265 fp->id = hg_fp_cnt;
5266 hg_fp_cnt++;
5267
5268 // pass1:
5269 // - collect calls
5270 // - resolve all branches
5271 for (i = 0; i < opcnt; i++)
5272 {
5273 po = &ops[i];
5274 po->bt_i = -1;
5275 po->btj = NULL;
5276
5277 if (po->flags & OPF_RMD)
5278 continue;
5279
5280 if (po->op == OP_CALL) {
5281 if (po->operand[0].type == OPT_LABEL)
5282 hg_fp_add_dep(fp, opr_name(po, 0));
5283
5284 continue;
5285 }
5286
5287 if (!(po->flags & OPF_JMP) || po->op == OP_RET)
5288 continue;
5289
5290 if (po->operand[0].type == OPT_REGMEM) {
5291 pd = try_resolve_jumptab(i, opcnt);
5292 if (pd == NULL)
5293 goto tailcall;
5294
5295 po->btj = pd;
5296 continue;
5297 }
5298
5299 for (l = 0; l < opcnt; l++) {
5300 if (g_labels[l][0] && IS(po->operand[0].name, g_labels[l])) {
5301 add_label_ref(&g_label_refs[l], i);
5302 po->bt_i = l;
5303 break;
5304 }
5305 }
5306
5307 if (po->bt_i != -1 || (po->flags & OPF_RMD))
5308 continue;
5309
5310 if (po->operand[0].type == OPT_LABEL)
5311 // assume tail call
5312 goto tailcall;
5313
5314 ferr(po, "unhandled branch\n");
5315
5316tailcall:
5317 po->op = OP_CALL;
5318 po->flags |= OPF_TAIL;
5319 if (i > 0 && ops[i - 1].op == OP_POP)
5320 po->flags |= OPF_ATAIL;
5321 i--; // reprocess
5322 }
5323
5324 // pass2:
5325 // - remove dead labels
5326 for (i = 0; i < opcnt; i++)
5327 {
5328 if (g_labels[i][0] != 0 && g_label_refs[i].i == -1)
5329 g_labels[i][0] = 0;
5330 }
5331
5332 // pass3:
5333 // - track saved regs
5334 // - try to figure out arg-regs
5335 for (i = 0; i < opcnt; i++)
5336 {
5337 po = &ops[i];
5338 if (po->flags & OPF_RMD)
5339 continue;
5340
5341 if (po->op == OP_PUSH && po->operand[0].type == OPT_REG)
5342 {
5343 reg = po->operand[0].reg;
5344 if (reg < 0)
5345 ferr(po, "reg not set for push?\n");
5346
5347 depth = 0;
5348 ret = scan_for_pop(i + 1, opcnt,
5349 po->operand[0].name, i + opcnt * 1, 0, &depth, 0);
5350 if (ret == 1) {
5351 regmask_save |= 1 << reg;
5352 po->flags |= OPF_RMD;
5353 scan_for_pop(i + 1, opcnt,
5354 po->operand[0].name, i + opcnt * 2, 0, &depth, 1);
5355 continue;
5356 }
5357 ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0);
5358 if (ret == 0) {
5359 regmask_save |= 1 << reg;
5360 po->flags |= OPF_RMD;
5361 scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, OPF_RMD);
5362 continue;
5363 }
5364 }
5365 else if (po->op == OP_PUSH && po->operand[0].type == OPT_CONST) {
5366 for (j = i + 1; j < opcnt; j++) {
5367 if ((ops[j].flags & (OPF_JMP|OPF_TAIL|OPF_RSAVE))
5368 || ops[j].op == OP_PUSH || g_labels[i][0] != 0)
5369 {
5370 break;
5371 }
5372
5373 if (!(ops[j].flags & OPF_RMD) && ops[j].op == OP_POP)
5374 {
5375 po->flags |= OPF_RMD;
5376 ops[j].datap = po;
5377 break;
5378 }
5379 }
5380 continue;
5381 }
5382 else if (po->op == OP_CALL) {
5383 po->regmask_dst |= 1 << xAX;
5384
5385 dep = hg_fp_find_dep(fp, po->operand[0].name);
5386 if (dep != NULL)
5387 dep->regmask_live = regmask_save | regmask_dst;
5388 }
5389 else if (po->op == OP_RET) {
5390 if (po->operand_cnt > 0)
5391 fp->is_stdcall = 1;
5392 }
5393
5394 if (has_ret != 0 && (po->flags & OPF_TAIL)) {
5395 if (po->op == OP_CALL) {
5396 j = i;
5397 ret = 1;
5398 }
5399 else {
5400 struct parsed_opr opr = { 0, };
5401 opr.type = OPT_REG;
5402 opr.reg = xAX;
5403 j = -1;
5404 from_caller = 0;
5405 ret = resolve_origin(i, &opr, i + opcnt * 3, &j, &from_caller);
5406 }
5407
5408 if (ret == -1 && from_caller) {
5409 // unresolved eax - probably void func
5410 has_ret = 0;
5411 }
5412 else {
5413 if (ops[j].op == OP_CALL) {
5414 dep = hg_fp_find_dep(fp, po->operand[0].name);
5415 if (dep != NULL)
5416 dep->ret_dep = 1;
5417 else
5418 has_ret = 1;
5419 }
5420 else
5421 has_ret = 1;
5422 }
5423 }
5424
5425 l = po->regmask_src & ~(regmask_save | regmask_dst);
5426#if 0
5427 if (l)
5428 fnote(po, "dep |= %04x, dst %04x, save %04x\n", l,
5429 regmask_dst, regmask_save);
5430#endif
5431 regmask_dep |= l;
5432 regmask_dst |= po->regmask_dst;
5433 }
5434
5435 if (has_ret == -1 && (regmask_dep & (1 << xAX)))
5436 has_ret = 1;
5437
5438 for (i = 0; i < g_eqcnt; i++)
5439 if (g_eqs[i].offset > max_bp_offset && g_eqs[i].offset < 4*32)
5440 max_bp_offset = g_eqs[i].offset;
5441
5442 if (max_bp_offset > 0) {
5443 max_bp_offset = (max_bp_offset + 3) & ~3;
5444 fp->argc_stack = max_bp_offset / 4 - 1;
5445 if (!(g_ida_func_attr & IDAFA_BP_FRAME))
5446 fp->argc_stack--;
5447 }
5448
5449 fp->regmask_dep = regmask_dep & ~(1 << xSP);
5450 fp->has_ret = has_ret;
5451}
5452
5453static void hg_fp_resolve_deps(struct func_prototype *fp)
5454{
5455 struct func_prototype fp_s;
5456 int i;
5457
5458 // this thing is recursive, so mark first..
5459 fp->dep_resolved = 1;
5460
5461 for (i = 0; i < fp->dep_func_cnt; i++) {
5462 strcpy(fp_s.name, fp->dep_func[i].name);
5463 fp->dep_func[i].proto = bsearch(&fp_s, hg_fp, hg_fp_cnt,
5464 sizeof(hg_fp[0]), hg_fp_cmp_name);
5465 if (fp->dep_func[i].proto != NULL) {
5466 if (!fp->dep_func[i].proto->dep_resolved)
5467 hg_fp_resolve_deps(fp->dep_func[i].proto);
5468
5469 fp->regmask_dep |= ~fp->dep_func[i].regmask_live
5470 & fp->dep_func[i].proto->regmask_dep;
5471
5472 if (fp->has_ret == -1)
5473 fp->has_ret = fp->dep_func[i].proto->has_ret;
5474 }
5475 }
5476}
5477
5478static void output_hdr(FILE *fout)
5479{
5480 struct func_prototype *fp;
5481 int had_usercall = 0;
5482 int regmask_dep;
5483 int argc_stack;
5484 int i, j, arg;
5485
5486 // resolve deps
5487 qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_name);
5488 for (i = 0; i < hg_fp_cnt; i++)
5489 hg_fp_resolve_deps(&hg_fp[i]);
5490
5491 // note: messes up .proto ptr, don't use
5492 //qsort(hg_fp, hg_fp_cnt, sizeof(hg_fp[0]), hg_fp_cmp_id);
5493
5494 for (i = 0; i < hg_fp_cnt; i++) {
5495 fp = &hg_fp[i];
5496 if (fp->has_ret == -1)
5497 fprintf(fout, "// ret unresolved\n");
5498#if 0
5499 fprintf(fout, "// dep:");
5500 for (j = 0; j < fp->dep_func_cnt; j++) {
5501 fprintf(fout, " %s/", fp->dep_func[j].name);
5502 if (fp->dep_func[j].proto != NULL)
5503 fprintf(fout, "%04x/%d", fp->dep_func[j].proto->regmask_dep,
5504 fp->dep_func[j].proto->has_ret);
5505 }
5506 fprintf(fout, "\n");
5507#endif
5508
5509 regmask_dep = fp->regmask_dep;
5510 argc_stack = fp->argc_stack;
5511
5512 fprintf(fout, fp->has_ret ? "int " : "void ");
5513 if (regmask_dep && (fp->is_stdcall || argc_stack == 0)
5514 && (regmask_dep & ~((1 << xCX) | (1 << xDX))) == 0)
5515 {
5516 fprintf(fout, "__fastcall ");
5517 if (had_usercall)
5518 fprintf(fout, " "); // align
5519 if (!(regmask_dep & (1 << xDX)) && argc_stack == 0)
5520 argc_stack = 1;
5521 else
5522 argc_stack += 2;
5523 regmask_dep = 0;
5524 }
5525 else if (regmask_dep && !fp->is_stdcall) {
5526 fprintf(fout, "/*__usercall*/ ");
5527 had_usercall = 1;
5528 }
5529 else if (regmask_dep) {
5530 fprintf(fout, "/*__userpurge*/ ");
5531 had_usercall = 1;
5532 }
5533 else if (fp->is_stdcall)
5534 fprintf(fout, "__stdcall ");
5535 else
5536 fprintf(fout, "__cdecl ");
5537
5538 fprintf(fout, "%s(", fp->name);
5539
5540 arg = 0;
5541 for (j = 0; j < xSP; j++) {
5542 if (regmask_dep & (1 << j)) {
5543 arg++;
5544 if (arg != 1)
5545 fprintf(fout, ", ");
5546 fprintf(fout, "int a%d/*<%s>*/", arg, regs_r32[j]);
5547 }
5548 }
5549
5550 for (j = 0; j < argc_stack; j++) {
5551 arg++;
5552 if (arg != 1)
5553 fprintf(fout, ", ");
5554 fprintf(fout, "int a%d", arg);
5555 }
5556
5557 fprintf(fout, ");\n");
5558 }
5559}
5560
5561static void set_label(int i, const char *name)
5562{
5563 const char *p;
5564 int len;
5565
5566 len = strlen(name);
5567 p = strchr(name, ':');
5568 if (p != NULL)
5569 len = p - name;
5570
5571 if (len > sizeof(g_labels[0]) - 1)
5572 aerr("label '%s' too long: %d\n", name, len);
5573 if (g_labels[i][0] != 0 && !IS_START(g_labels[i], "algn_"))
5574 aerr("dupe label '%s' vs '%s'?\n", name, g_labels[i]);
5575 memcpy(g_labels[i], name, len);
5576 g_labels[i][len] = 0;
5577}
5578
5579// '=' needs special treatment..
5580static char *next_word_s(char *w, size_t wsize, char *s)
5581{
5582 size_t i;
5583
5584 s = sskip(s);
5585
5586 for (i = 0; i < wsize - 1; i++) {
5587 if (s[i] == 0 || my_isblank(s[i]) || (s[i] == '=' && i > 0))
5588 break;
5589 w[i] = s[i];
5590 }
5591 w[i] = 0;
5592
5593 if (s[i] != 0 && !my_isblank(s[i]) && s[i] != '=')
5594 printf("warning: '%s' truncated\n", w);
5595
5596 return s + i;
5597}
5598
5599struct chunk_item {
5600 char *name;
5601 long fptr;
5602 int asmln;
5603};
5604
5605static struct chunk_item *func_chunks;
5606static int func_chunk_cnt;
5607static int func_chunk_alloc;
5608
5609static void add_func_chunk(FILE *fasm, const char *name, int line)
5610{
5611 if (func_chunk_cnt >= func_chunk_alloc) {
5612 func_chunk_alloc *= 2;
5613 func_chunks = realloc(func_chunks,
5614 func_chunk_alloc * sizeof(func_chunks[0]));
5615 my_assert_not(func_chunks, NULL);
5616 }
5617 func_chunks[func_chunk_cnt].fptr = ftell(fasm);
5618 func_chunks[func_chunk_cnt].name = strdup(name);
5619 func_chunks[func_chunk_cnt].asmln = line;
5620 func_chunk_cnt++;
5621}
5622
5623static int cmp_chunks(const void *p1, const void *p2)
5624{
5625 const struct chunk_item *c1 = p1, *c2 = p2;
5626 return strcmp(c1->name, c2->name);
5627}
5628
5629static int cmpstringp(const void *p1, const void *p2)
5630{
5631 return strcmp(*(char * const *)p1, *(char * const *)p2);
5632}
5633
5634static void scan_ahead(FILE *fasm)
5635{
5636 char words[2][256];
5637 char line[256];
5638 long oldpos;
5639 int oldasmln;
5640 int wordc;
5641 char *p;
5642 int i;
5643
5644 oldpos = ftell(fasm);
5645 oldasmln = asmln;
5646
5647 while (fgets(line, sizeof(line), fasm))
5648 {
5649 wordc = 0;
5650 asmln++;
5651
5652 p = sskip(line);
5653 if (*p == 0)
5654 continue;
5655
5656 if (*p == ';')
5657 {
5658 // get rid of random tabs
5659 for (i = 0; line[i] != 0; i++)
5660 if (line[i] == '\t')
5661 line[i] = ' ';
5662
5663 if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
5664 {
5665 p += 30;
5666 next_word(words[0], sizeof(words[0]), p);
5667 if (words[0][0] == 0)
5668 aerr("missing name for func chunk?\n");
5669
5670 add_func_chunk(fasm, words[0], asmln);
5671 }
5672 else if (IS_START(p, "; sctend"))
5673 break;
5674
5675 continue;
5676 } // *p == ';'
5677
5678 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
5679 words[wordc][0] = 0;
5680 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
5681 if (*p == 0 || *p == ';') {
5682 wordc++;
5683 break;
5684 }
5685 }
5686
5687 if (wordc == 2 && IS(words[1], "ends"))
5688 break;
5689 }
5690
5691 fseek(fasm, oldpos, SEEK_SET);
5692 asmln = oldasmln;
5693}
5694
5695int main(int argc, char *argv[])
5696{
5697 FILE *fout, *fasm, *frlist;
5698 struct parsed_data *pd = NULL;
5699 int pd_alloc = 0;
5700 char **rlist = NULL;
5701 int rlist_len = 0;
5702 int rlist_alloc = 0;
5703 int func_chunks_used = 0;
5704 int func_chunks_sorted = 0;
5705 int func_chunk_i = -1;
5706 long func_chunk_ret = 0;
5707 int func_chunk_ret_ln = 0;
5708 int scanned_ahead = 0;
5709 char line[256];
5710 char words[20][256];
5711 enum opr_lenmod lmod;
5712 char *sctproto = NULL;
5713 int in_func = 0;
5714 int pending_endp = 0;
5715 int skip_func = 0;
5716 int skip_warned = 0;
5717 int header_mode = 0;
5718 int eq_alloc;
5719 int verbose = 0;
5720 int multi_seg = 0;
5721 int end = 0;
5722 int arg_out;
5723 int arg;
5724 int pi = 0;
5725 int i, j;
5726 int ret, len;
5727 char *p;
5728 int wordc;
5729
5730 for (arg = 1; arg < argc; arg++) {
5731 if (IS(argv[arg], "-v"))
5732 verbose = 1;
5733 else if (IS(argv[arg], "-rf"))
5734 g_allow_regfunc = 1;
5735 else if (IS(argv[arg], "-m"))
5736 multi_seg = 1;
5737 else if (IS(argv[arg], "-hdr"))
5738 header_mode = g_quiet_pp = 1;
5739 else
5740 break;
5741 }
5742
5743 if (argc < arg + 3) {
5744 printf("usage:\n%s [-v] [-rf] [-m] <.c> <.asm> <hdrf> [rlist]*\n"
5745 "%s -hdr <.h> <.asm> <seed_hdrf> [rlist]*\n",
5746 argv[0], argv[0]);
5747 return 1;
5748 }
5749
5750 arg_out = arg++;
5751
5752 asmfn = argv[arg++];
5753 fasm = fopen(asmfn, "r");
5754 my_assert_not(fasm, NULL);
5755
5756 hdrfn = argv[arg++];
5757 g_fhdr = fopen(hdrfn, "r");
5758 my_assert_not(g_fhdr, NULL);
5759
5760 rlist_alloc = 64;
5761 rlist = malloc(rlist_alloc * sizeof(rlist[0]));
5762 my_assert_not(rlist, NULL);
5763 // needs special handling..
5764 rlist[rlist_len++] = "__alloca_probe";
5765
5766 func_chunk_alloc = 32;
5767 func_chunks = malloc(func_chunk_alloc * sizeof(func_chunks[0]));
5768 my_assert_not(func_chunks, NULL);
5769
5770 memset(words, 0, sizeof(words));
5771
5772 for (; arg < argc; arg++) {
5773 frlist = fopen(argv[arg], "r");
5774 my_assert_not(frlist, NULL);
5775
5776 while (fgets(line, sizeof(line), frlist)) {
5777 p = sskip(line);
5778 if (*p == 0 || *p == ';')
5779 continue;
5780 if (*p == '#') {
5781 if (IS_START(p, "#if 0")
5782 || (g_allow_regfunc && IS_START(p, "#if NO_REGFUNC")))
5783 {
5784 skip_func = 1;
5785 }
5786 else if (IS_START(p, "#endif"))
5787 skip_func = 0;
5788 continue;
5789 }
5790 if (skip_func)
5791 continue;
5792
5793 p = next_word(words[0], sizeof(words[0]), p);
5794 if (words[0][0] == 0)
5795 continue;
5796
5797 if (rlist_len >= rlist_alloc) {
5798 rlist_alloc = rlist_alloc * 2 + 64;
5799 rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0]));
5800 my_assert_not(rlist, NULL);
5801 }
5802 rlist[rlist_len++] = strdup(words[0]);
5803 }
5804 skip_func = 0;
5805
5806 fclose(frlist);
5807 frlist = NULL;
5808 }
5809
5810 if (rlist_len > 0)
5811 qsort(rlist, rlist_len, sizeof(rlist[0]), cmpstringp);
5812
5813 fout = fopen(argv[arg_out], "w");
5814 my_assert_not(fout, NULL);
5815
5816 eq_alloc = 128;
5817 g_eqs = malloc(eq_alloc * sizeof(g_eqs[0]));
5818 my_assert_not(g_eqs, NULL);
5819
5820 for (i = 0; i < ARRAY_SIZE(g_label_refs); i++) {
5821 g_label_refs[i].i = -1;
5822 g_label_refs[i].next = NULL;
5823 }
5824
5825 while (fgets(line, sizeof(line), fasm))
5826 {
5827 wordc = 0;
5828 asmln++;
5829
5830 p = sskip(line);
5831 if (*p == 0)
5832 continue;
5833
5834 // get rid of random tabs
5835 for (i = 0; line[i] != 0; i++)
5836 if (line[i] == '\t')
5837 line[i] = ' ';
5838
5839 if (*p == ';')
5840 {
5841 if (p[2] == '=' && IS_START(p, "; =============== S U B"))
5842 goto do_pending_endp; // eww..
5843
5844 if (p[2] == 'A' && IS_START(p, "; Attributes:"))
5845 {
5846 static const char *attrs[] = {
5847 "bp-based frame",
5848 "library function",
5849 "static",
5850 "noreturn",
5851 "thunk",
5852 "fpd=",
5853 };
5854
5855 // parse IDA's attribute-list comment
5856 g_ida_func_attr = 0;
5857 p = sskip(p + 13);
5858
5859 for (; *p != 0; p = sskip(p)) {
5860 for (i = 0; i < ARRAY_SIZE(attrs); i++) {
5861 if (!strncmp(p, attrs[i], strlen(attrs[i]))) {
5862 g_ida_func_attr |= 1 << i;
5863 p += strlen(attrs[i]);
5864 break;
5865 }
5866 }
5867 if (i == ARRAY_SIZE(attrs)) {
5868 anote("unparsed IDA attr: %s\n", p);
5869 break;
5870 }
5871 if (IS(attrs[i], "fpd=")) {
5872 p = next_word(words[0], sizeof(words[0]), p);
5873 // ignore for now..
5874 }
5875 }
5876 }
5877 else if (p[2] == 'S' && IS_START(p, "; START OF FUNCTION CHUNK FOR "))
5878 {
5879 p += 30;
5880 next_word(words[0], sizeof(words[0]), p);
5881 if (words[0][0] == 0)
5882 aerr("missing name for func chunk?\n");
5883
5884 if (!scanned_ahead) {
5885 add_func_chunk(fasm, words[0], asmln);
5886 func_chunks_sorted = 0;
5887 }
5888 }
5889 else if (p[2] == 'E' && IS_START(p, "; END OF FUNCTION CHUNK"))
5890 {
5891 if (func_chunk_i >= 0) {
5892 if (func_chunk_i < func_chunk_cnt
5893 && IS(func_chunks[func_chunk_i].name, g_func))
5894 {
5895 // move on to next chunk
5896 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
5897 if (ret)
5898 aerr("seek failed for '%s' chunk #%d\n",
5899 g_func, func_chunk_i);
5900 asmln = func_chunks[func_chunk_i].asmln;
5901 func_chunk_i++;
5902 }
5903 else {
5904 if (func_chunk_ret == 0)
5905 aerr("no return from chunk?\n");
5906 fseek(fasm, func_chunk_ret, SEEK_SET);
5907 asmln = func_chunk_ret_ln;
5908 func_chunk_ret = 0;
5909 pending_endp = 1;
5910 }
5911 }
5912 }
5913 else if (p[2] == 'F' && IS_START(p, "; FUNCTION CHUNK AT ")) {
5914 func_chunks_used = 1;
5915 p += 20;
5916 if (IS_START(g_func, "sub_")) {
5917 unsigned long addr = strtoul(p, NULL, 16);
5918 unsigned long f_addr = strtoul(g_func + 4, NULL, 16);
5919 if (addr > f_addr && !scanned_ahead) {
5920 anote("scan_ahead caused by '%s', addr %lx\n",
5921 g_func, addr);
5922 scan_ahead(fasm);
5923 scanned_ahead = 1;
5924 func_chunks_sorted = 0;
5925 }
5926 }
5927 }
5928 continue;
5929 } // *p == ';'
5930
5931parse_words:
5932 for (i = wordc; i < ARRAY_SIZE(words); i++)
5933 words[i][0] = 0;
5934 for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) {
5935 p = sskip(next_word_s(words[wordc], sizeof(words[0]), p));
5936 if (*p == 0 || *p == ';') {
5937 wordc++;
5938 break;
5939 }
5940 }
5941 if (*p != 0 && *p != ';')
5942 aerr("too many words\n");
5943
5944 // alow asm patches in comments
5945 if (*p == ';') {
5946 if (IS_START(p, "; sctpatch:")) {
5947 p = sskip(p + 11);
5948 if (*p == 0 || *p == ';')
5949 continue;
5950 goto parse_words; // lame
5951 }
5952 if (IS_START(p, "; sctproto:")) {
5953 sctproto = strdup(p + 11);
5954 }
5955 else if (IS_START(p, "; sctend")) {
5956 end = 1;
5957 if (!pending_endp)
5958 break;
5959 }
5960 }
5961
5962 if (wordc == 0) {
5963 // shouldn't happen
5964 awarn("wordc == 0?\n");
5965 continue;
5966 }
5967
5968 // don't care about this:
5969 if (words[0][0] == '.'
5970 || IS(words[0], "include")
5971 || IS(words[0], "assume") || IS(words[1], "segment")
5972 || IS(words[0], "align"))
5973 {
5974 continue;
5975 }
5976
5977do_pending_endp:
5978 // do delayed endp processing to collect switch jumptables
5979 if (pending_endp) {
5980 if (in_func && !skip_func && !end && wordc >= 2
5981 && ((words[0][0] == 'd' && words[0][2] == 0)
5982 || (words[1][0] == 'd' && words[1][2] == 0)))
5983 {
5984 i = 1;
5985 if (words[1][0] == 'd' && words[1][2] == 0) {
5986 // label
5987 if (g_func_pd_cnt >= pd_alloc) {
5988 pd_alloc = pd_alloc * 2 + 16;
5989 g_func_pd = realloc(g_func_pd,
5990 sizeof(g_func_pd[0]) * pd_alloc);
5991 my_assert_not(g_func_pd, NULL);
5992 }
5993 pd = &g_func_pd[g_func_pd_cnt];
5994 g_func_pd_cnt++;
5995 memset(pd, 0, sizeof(*pd));
5996 strcpy(pd->label, words[0]);
5997 pd->type = OPT_CONST;
5998 pd->lmod = lmod_from_directive(words[1]);
5999 i = 2;
6000 }
6001 else {
6002 if (pd == NULL) {
6003 if (verbose)
6004 anote("skipping alignment byte?\n");
6005 continue;
6006 }
6007 lmod = lmod_from_directive(words[0]);
6008 if (lmod != pd->lmod)
6009 aerr("lmod change? %d->%d\n", pd->lmod, lmod);
6010 }
6011
6012 if (pd->count_alloc < pd->count + wordc) {
6013 pd->count_alloc = pd->count_alloc * 2 + 14 + wordc;
6014 pd->d = realloc(pd->d, sizeof(pd->d[0]) * pd->count_alloc);
6015 my_assert_not(pd->d, NULL);
6016 }
6017 for (; i < wordc; i++) {
6018 if (IS(words[i], "offset")) {
6019 pd->type = OPT_OFFSET;
6020 i++;
6021 }
6022 p = strchr(words[i], ',');
6023 if (p != NULL)
6024 *p = 0;
6025 if (pd->type == OPT_OFFSET)
6026 pd->d[pd->count].u.label = strdup(words[i]);
6027 else
6028 pd->d[pd->count].u.val = parse_number(words[i]);
6029 pd->d[pd->count].bt_i = -1;
6030 pd->count++;
6031 }
6032 continue;
6033 }
6034
6035 if (in_func && !skip_func) {
6036 if (header_mode)
6037 gen_hdr(g_func, pi);
6038 else
6039 gen_func(fout, g_fhdr, g_func, pi);
6040 }
6041
6042 pending_endp = 0;
6043 in_func = 0;
6044 g_ida_func_attr = 0;
6045 skip_warned = 0;
6046 skip_func = 0;
6047 g_func[0] = 0;
6048 func_chunks_used = 0;
6049 func_chunk_i = -1;
6050 if (pi != 0) {
6051 memset(&ops, 0, pi * sizeof(ops[0]));
6052 memset(g_labels, 0, pi * sizeof(g_labels[0]));
6053 pi = 0;
6054 }
6055 g_eqcnt = 0;
6056 for (i = 0; i < g_func_pd_cnt; i++) {
6057 pd = &g_func_pd[i];
6058 if (pd->type == OPT_OFFSET) {
6059 for (j = 0; j < pd->count; j++)
6060 free(pd->d[j].u.label);
6061 }
6062 free(pd->d);
6063 pd->d = NULL;
6064 }
6065 g_func_pd_cnt = 0;
6066 pd = NULL;
6067
6068 if (end)
6069 break;
6070 if (wordc == 0)
6071 continue;
6072 }
6073
6074 if (IS(words[1], "proc")) {
6075 if (in_func)
6076 aerr("proc '%s' while in_func '%s'?\n",
6077 words[0], g_func);
6078 p = words[0];
6079 if (bsearch(&p, rlist, rlist_len, sizeof(rlist[0]), cmpstringp))
6080 skip_func = 1;
6081 strcpy(g_func, words[0]);
6082 set_label(0, words[0]);
6083 in_func = 1;
6084 continue;
6085 }
6086
6087 if (IS(words[1], "endp"))
6088 {
6089 if (!in_func)
6090 aerr("endp '%s' while not in_func?\n", words[0]);
6091 if (!IS(g_func, words[0]))
6092 aerr("endp '%s' while in_func '%s'?\n",
6093 words[0], g_func);
6094
6095 if ((g_ida_func_attr & IDAFA_THUNK) && pi == 1
6096 && ops[0].op == OP_JMP && ops[0].operand[0].had_ds)
6097 {
6098 // import jump
6099 skip_func = 1;
6100 }
6101
6102 if (!skip_func && func_chunks_used) {
6103 // start processing chunks
6104 struct chunk_item *ci, key = { g_func, 0 };
6105
6106 func_chunk_ret = ftell(fasm);
6107 func_chunk_ret_ln = asmln;
6108 if (!func_chunks_sorted) {
6109 qsort(func_chunks, func_chunk_cnt,
6110 sizeof(func_chunks[0]), cmp_chunks);
6111 func_chunks_sorted = 1;
6112 }
6113 ci = bsearch(&key, func_chunks, func_chunk_cnt,
6114 sizeof(func_chunks[0]), cmp_chunks);
6115 if (ci == NULL)
6116 aerr("'%s' needs chunks, but none found\n", g_func);
6117 func_chunk_i = ci - func_chunks;
6118 for (; func_chunk_i > 0; func_chunk_i--)
6119 if (!IS(func_chunks[func_chunk_i - 1].name, g_func))
6120 break;
6121
6122 ret = fseek(fasm, func_chunks[func_chunk_i].fptr, SEEK_SET);
6123 if (ret)
6124 aerr("seek failed for '%s' chunk #%d\n", g_func, func_chunk_i);
6125 asmln = func_chunks[func_chunk_i].asmln;
6126 func_chunk_i++;
6127 continue;
6128 }
6129 pending_endp = 1;
6130 continue;
6131 }
6132
6133 if (wordc == 2 && IS(words[1], "ends")) {
6134 if (!multi_seg) {
6135 end = 1;
6136 if (pending_endp)
6137 goto do_pending_endp;
6138 break;
6139 }
6140
6141 // scan for next text segment
6142 while (fgets(line, sizeof(line), fasm)) {
6143 asmln++;
6144 p = sskip(line);
6145 if (*p == 0 || *p == ';')
6146 continue;
6147
6148 if (strstr(p, "segment para public 'CODE' use32"))
6149 break;
6150 }
6151
6152 continue;
6153 }
6154
6155 p = strchr(words[0], ':');
6156 if (p != NULL) {
6157 set_label(pi, words[0]);
6158 continue;
6159 }
6160
6161 if (!in_func || skip_func) {
6162 if (!skip_warned && !skip_func && g_labels[pi][0] != 0) {
6163 if (verbose)
6164 anote("skipping from '%s'\n", g_labels[pi]);
6165 skip_warned = 1;
6166 }
6167 g_labels[pi][0] = 0;
6168 continue;
6169 }
6170
6171 if (wordc > 1 && IS(words[1], "="))
6172 {
6173 if (wordc != 5)
6174 aerr("unhandled equ, wc=%d\n", wordc);
6175 if (g_eqcnt >= eq_alloc) {
6176 eq_alloc *= 2;
6177 g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0]));
6178 my_assert_not(g_eqs, NULL);
6179 }
6180
6181 len = strlen(words[0]);
6182 if (len > sizeof(g_eqs[0].name) - 1)
6183 aerr("equ name too long: %d\n", len);
6184 strcpy(g_eqs[g_eqcnt].name, words[0]);
6185
6186 if (!IS(words[3], "ptr"))
6187 aerr("unhandled equ\n");
6188 if (IS(words[2], "dword"))
6189 g_eqs[g_eqcnt].lmod = OPLM_DWORD;
6190 else if (IS(words[2], "word"))
6191 g_eqs[g_eqcnt].lmod = OPLM_WORD;
6192 else if (IS(words[2], "byte"))
6193 g_eqs[g_eqcnt].lmod = OPLM_BYTE;
6194 else if (IS(words[2], "qword"))
6195 g_eqs[g_eqcnt].lmod = OPLM_QWORD;
6196 else
6197 aerr("bad lmod: '%s'\n", words[2]);
6198
6199 g_eqs[g_eqcnt].offset = parse_number(words[4]);
6200 g_eqcnt++;
6201 continue;
6202 }
6203
6204 if (pi >= ARRAY_SIZE(ops))
6205 aerr("too many ops\n");
6206
6207 parse_op(&ops[pi], words, wordc);
6208
6209 if (sctproto != NULL) {
6210 if (ops[pi].op == OP_CALL || ops[pi].op == OP_JMP)
6211 ops[pi].datap = sctproto;
6212 sctproto = NULL;
6213 }
6214 pi++;
6215 }
6216
6217 if (header_mode)
6218 output_hdr(fout);
6219
6220 fclose(fout);
6221 fclose(fasm);
6222 fclose(g_fhdr);
6223
6224 return 0;
6225}
6226
6227// vim:ts=2:shiftwidth=2:expandtab
Unnamed repository; edit this file 'description' to name the repository.