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