asmproc: add option to ignore missing syms
[ia32rtools.git] / plugin / saveasm.cpp
CommitLineData
d8891fcc 1#define NO_OBSOLETE_FUNCS
2#include <ida.hpp>
3#include <idp.hpp>
4#include <bytes.hpp>
5#include <loader.hpp>
6#include <kernwin.hpp>
7
8#include <name.hpp>
9#include <frame.hpp>
10#include <struct.hpp>
11#include <auto.hpp>
15c7b2a4 12#include <intel.hpp>
d8891fcc 13
14#define IS_START(w, y) !strncmp(w, y, strlen(y))
15#define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
16
15c7b2a4 17// non-local branch targets
18static ea_t *nonlocal_bt;
19static int nonlocal_bt_alloc;
20static int nonlocal_bt_cnt;
21
d8891fcc 22//--------------------------------------------------------------------------
23static int idaapi init(void)
24{
25 return PLUGIN_OK;
26}
27
28//--------------------------------------------------------------------------
29static void idaapi term(void)
30{
15c7b2a4 31 if (nonlocal_bt != NULL) {
32 free(nonlocal_bt);
33 nonlocal_bt = NULL;
34 }
35 nonlocal_bt_alloc = 0;
d8891fcc 36}
37
38//--------------------------------------------------------------------------
39
40static const char *reserved_names[] = {
41 "name",
b587e6ae 42 "type",
d8891fcc 43 "offset",
b587e6ae 44 "aam",
1402b79d 45 "text",
d8891fcc 46};
47
48static int is_name_reserved(const char *name)
49{
50 int i;
51 for (i = 0; i < ARRAY_SIZE(reserved_names); i++)
52 if (strcasecmp(name, reserved_names[i]) == 0)
53 return 1;
54
55 return 0;
56}
57
15c7b2a4 58static int nonlocal_bt_cmp(const void *p1, const void *p2)
59{
60 const ea_t *e1 = (const ea_t *)p1, *e2 = (const ea_t *)p2;
61 return *e1 - *e2;
62}
63
64static void nonlocal_add(ea_t ea)
65{
66 if (nonlocal_bt_cnt >= nonlocal_bt_alloc) {
67 nonlocal_bt_alloc += nonlocal_bt_alloc * 2 + 64;
68 nonlocal_bt = (ea_t *)realloc(nonlocal_bt,
69 nonlocal_bt_alloc * sizeof(nonlocal_bt[0]));
70 if (nonlocal_bt == NULL) {
71 msg("OOM\n");
72 return;
73 }
74 }
75 nonlocal_bt[nonlocal_bt_cnt++] = ea;
76}
77
b587e6ae 78// is instruction a (un)conditional jump (not call)?
79static int is_insn_jmp(uint16 itype)
80{
81 return itype == NN_jmp || (NN_ja <= itype && itype <= NN_jz);
82}
83
d8891fcc 84static void do_def_line(char *buf, size_t buf_size, const char *line)
85{
15c7b2a4 86 char *endp = NULL;
87 ea_t ea, *ea_ret;
d8891fcc 88 int len;
89
90 tag_remove(line, buf, buf_size); // remove color codes
91 len = strlen(buf);
92 if (len < 9) {
93 buf[0] = 0;
94 return;
95 }
96 memmove(buf, buf + 9, len - 9 + 1); // rm address
15c7b2a4 97
98 if (IS_START(buf, "loc_")) {
99 ea = strtoul(buf + 4, &endp, 16);
100 if (ea != 0 && *endp == ':') {
101 ea_ret = (ea_t *)bsearch(&ea, nonlocal_bt, nonlocal_bt_cnt,
102 sizeof(nonlocal_bt[0]), nonlocal_bt_cmp);
103 if (ea_ret != 0) {
104 if (endp[1] != ' ')
105 msg("no trailing blank in '%s'\n", buf);
106 else
107 endp[1] = ':';
108 }
109 }
110 }
d8891fcc 111}
112
113static void idaapi run(int /*arg*/)
114{
15c7b2a4 115 // isEnabled(ea) // address belongs to disassembly
d8891fcc 116 // ea_t ea = get_screen_ea();
15c7b2a4 117 // foo = DecodeInstruction(ScreenEA());
d8891fcc 118 FILE *fout = NULL;
119 int fout_line = 0;
120 char buf[MAXSTR];
1caf86bb 121 char buf2[MAXSTR];
b587e6ae 122 const char *name;
d8891fcc 123 struc_t *frame;
124 func_t *func;
15c7b2a4 125 ea_t ui_ea_block = 0, ea_size;
126 ea_t tmp_ea, target_ea;
d8891fcc 127 ea_t ea;
b587e6ae 128 flags_t ea_flags;
1caf86bb 129 uval_t idx;
d8891fcc 130 int i, o, m, n;
131 int ret;
132 char *p;
133
15c7b2a4 134 nonlocal_bt_cnt = 0;
135
1caf86bb 136 // get rid of structs, masm doesn't understand them
137 idx = get_first_struc_idx();
138 while (idx != BADNODE) {
139 tid_t tid = get_struc_by_idx(idx);
140 struc_t *struc = get_struc(tid);
141 get_struc_name(tid, buf, sizeof(buf));
142 msg("removing struct '%s'\n", buf);
143 //del_struc_members(struc, 0, get_max_offset(struc));
144 del_struc(struc);
145
146 idx = get_first_struc_idx();
147 }
148
15c7b2a4 149 // 1st pass: walk through all funcs
150 func = get_func(inf.minEA);
d8891fcc 151 while (func != NULL)
152 {
15c7b2a4 153 func_tail_iterator_t fti(func);
154 if (!fti.main()) {
155 msg("%x: func_tail_iterator_t main failed\n", ea);
156 return;
157 }
158 const area_t &f_area = fti.chunk();
159 ea = f_area.startEA;
160
161 // rename global syms which conflict with frame member names
d8891fcc 162 frame = get_frame(func);
163 if (frame != NULL)
164 {
165 for (m = 0; m < (int)frame->memqty; m++)
166 {
167 ret = get_member_name(frame->members[m].id, buf, sizeof(buf));
168 if (ret <= 0) {
169 msg("%x: member has no name?\n", ea);
170 return;
171 }
172 if (buf[0] == ' ') // what's this?
173 continue;
174 if (IS_START(buf, "arg_") || IS_START(buf, "var_"))
175 continue;
176
1caf86bb 177 // check for dupe names
178 int m1, dupe = 0;
179 for (m1 = 0; m1 < m; m1++) {
180 get_member_name(frame->members[m1].id, buf2, sizeof(buf2));
181 if (stricmp(buf, buf2) == 0)
182 dupe = 1;
183 }
184
185 if (is_name_reserved(buf) || dupe) {
d8891fcc 186 msg("%x: renaming '%s'\n", ea, buf);
187 qstrncat(buf, "_", sizeof(buf));
188 ret = set_member_name(frame, frame->members[m].soff, buf);
189 if (!ret) {
190 msg("%x: renaming failed\n", ea);
191 return;
192 }
193 }
194
195 tmp_ea = get_name_ea(ea, buf);
196 if (tmp_ea == 0 || tmp_ea == ~0)
197 continue;
198
199 msg("%x: from %x: renaming '%s'\n", tmp_ea, ea, buf);
200 qstrncat(buf, "_g", sizeof(buf));
201 set_name(tmp_ea, buf);
202 }
203 }
204
205 func = get_next_func(ea);
15c7b2a4 206 }
207
b587e6ae 208 // 2nd pass over whole .text and .(ro)data segments
15c7b2a4 209 for (ea = inf.minEA; ea != BADADDR; ea = next_head(ea, inf.maxEA))
210 {
211 segment_t *seg = getseg(ea);
b587e6ae 212 if (!seg)
213 break;
214 if (seg->type == SEG_XTRN)
215 continue;
216 if (seg->type != SEG_CODE && seg->type != SEG_DATA)
15c7b2a4 217 break;
218
b587e6ae 219 ea_flags = get_flags_novalue(ea);
15c7b2a4 220 func = get_func(ea);
221 if (isCode(ea_flags))
222 {
223 if (!decode_insn(ea)) {
224 msg("%x: decode_insn() failed\n", ea);
225 continue;
226 }
227
b587e6ae 228 // masm doesn't understand IDA's float/xmm types
229 if (cmd.itype == NN_fld || cmd.itype == NN_fst
230 || cmd.itype == NN_movapd || cmd.itype == NN_movlpd)
231 {
232 for (o = 0; o < UA_MAXOP; o++) {
233 if (cmd.Operands[o].type == o_void)
234 break;
235
236 if (cmd.Operands[o].type == o_mem) {
237 tmp_ea = cmd.Operands[o].addr;
238 flags_t tmp_ea_flags = get_flags_novalue(tmp_ea);
239 if (!isUnknown(tmp_ea_flags)) {
240 buf[0] = 0;
241 get_name(ea, tmp_ea, buf, sizeof(buf));
242 msg("%x: undefining %x '%s'\n", ea, tmp_ea, buf);
243 do_unknown(tmp_ea, DOUNK_EXPAND);
244 }
245 }
246 }
247 }
1caf86bb 248 // detect code alignment
249 else if (cmd.itype == NN_lea) {
250 if (cmd.Operands[0].reg == cmd.Operands[1].reg
251 && cmd.Operands[1].type == o_displ
252 && cmd.Operands[1].addr == 0)
253 {
254 tmp_ea = next_head(ea, inf.maxEA);
255 if ((tmp_ea & 0x03) == 0) {
256 n = calc_max_align(tmp_ea);
257 if (n > 4) // masm doesn't like more..
258 n = 4;
259 msg("%x: align %d\n", ea, 1 << n);
260 do_unknown(ea, DOUNK_SIMPLE);
261 doAlign(ea, tmp_ea - ea, n);
262 }
263 }
264 }
b587e6ae 265
15c7b2a4 266 // find non-local branches
b587e6ae 267 if (is_insn_jmp(cmd.itype) && cmd.Operands[0].type == o_near)
15c7b2a4 268 {
269 target_ea = cmd.Operands[0].addr;
270 if (func == NULL)
271 nonlocal_add(target_ea);
272 else {
273 ret = get_func_chunknum(func, target_ea);
274 if (ret != 0) {
275 // a jump to another func or chunk
276 // check if it lands on func start
277 if (!isFunc(get_flags_novalue(target_ea)))
278 nonlocal_add(target_ea);
279 }
280 }
281 }
282 }
283 else { // not code
1caf86bb 284 int do_undef = 0;
285 ea_size = get_item_size(ea);
286
15c7b2a4 287 if (func == NULL && isOff0(ea_flags)) {
15c7b2a4 288 for (tmp_ea = 0; tmp_ea < ea_size; tmp_ea += 4)
289 nonlocal_add(get_long(ea + tmp_ea));
290 }
b587e6ae 291
292 // IDA vs masm float/mmx/xmm type incompatibility
293 if (isDouble(ea_flags) || isTbyt(ea_flags)
294 || isPackReal(ea_flags))
295 {
1caf86bb 296 do_undef = 1;
297 }
298 else if (isOwrd(ea_flags)) {
b587e6ae 299 buf[0] = 0;
300 get_name(BADADDR, ea, buf, sizeof(buf));
1caf86bb 301 if (IS_START(buf, "xmm"))
302 do_undef = 1;
303 }
304 // masm doesn't understand IDA's unicode
305 else if (isASCII(ea_flags) && ea_size >= 4
306 && (get_long(ea) & 0xff00ff00) == 0) // lame..
307 {
308 do_undef = 1;
309 }
310 // masm doesn't understand large aligns
311 else if (isAlign(ea_flags) && ea_size > 0x10) {
312 msg("%x: undefining align %d\n", ea, ea_size);
b587e6ae 313 do_unknown(ea, DOUNK_EXPAND);
314 }
1caf86bb 315
316 if (do_undef) {
b587e6ae 317 buf[0] = 0;
318 get_name(BADADDR, ea, buf, sizeof(buf));
1caf86bb 319 msg("%x: undefining '%s'\n", ea, buf);
320 do_unknown(ea, DOUNK_EXPAND);
b587e6ae 321 }
322 }
323 }
324
325 // check namelist for reserved names
326 n = get_nlist_size();
327 for (i = 0; i < n; i++) {
328 ea = get_nlist_ea(i);
329 name = get_nlist_name(i);
330 if (name == NULL) {
331 msg("%x: null name?\n", ea);
332 continue;
333 }
334
335 if (is_name_reserved(name)) {
336 msg("%x: renaming name '%s'\n", ea, name);
337 qsnprintf(buf, sizeof(buf), "%s_g", name);
338 set_name(ea, buf);
15c7b2a4 339 }
340 }
341
342 if (nonlocal_bt_cnt > 1) {
343 qsort(nonlocal_bt, nonlocal_bt_cnt,
344 sizeof(nonlocal_bt[0]), nonlocal_bt_cmp);
d8891fcc 345 }
346
347 char *fname = askfile_c(1, NULL, "Save asm file");
348 if (fname == NULL)
349 return;
350 fout = qfopen(fname, "w");
351 if (fout == NULL) {
352 msg("couldn't open '%s'\n", fname);
353 return;
354 }
355
356 show_wait_box("Saving..");
357
358 // deal with the beginning
359 ea = inf.minEA;
360 int flags = 0; // calc_default_idaplace_flags();
361 linearray_t ln(&flags);
362 idaplace_t pl;
363 pl.ea = ea;
364 pl.lnnum = 0;
365 ln.set_place(&pl);
366 n = ln.get_linecnt();
367 for (i = 0; i < n - 1; i++) {
368 do_def_line(buf, sizeof(buf), ln.down());
369 if (strstr(buf, "include"))
370 continue;
a6e7d4e6 371 p = strstr(buf, "assume cs");
372 if (p != NULL) {
373 memmove(p + 1, p, strlen(p) + 1);
374 *p = ';';
375 }
d8891fcc 376
377 fout_line++;
378 qfprintf(fout, "%s\n", buf);
379 p = strstr(buf, ".mmx");
380 if (p != NULL) {
381 memcpy(p, ".xmm", 4);
382 fout_line++;
383 qfprintf(fout, "%s\n", buf);
1402b79d 384 continue;
385 }
386 p = strstr(buf, ".model");
387 if (p != NULL) {
388 qstrncpy(p, "include imports.inc", sizeof(buf) - (p - buf));
389 fout_line++;
390 qfprintf(fout, "\n%s\n", buf);
391 continue;
d8891fcc 392 }
393 }
b587e6ae 394 pl.lnnum = i;
d8891fcc 395
396 for (;;)
397 {
1caf86bb 398 int drop_large = 0, drop_rva = 0, set_scale = 0, jmp_near = 0;
399 int word_imm = 0, dword_imm = 0, do_pushf = 0;
15c7b2a4 400
d8891fcc 401 if ((ea >> 14) != ui_ea_block) {
402 ui_ea_block = ea >> 14;
403 showAddr(ea);
404 if (wasBreak())
405 break;
406 }
407
408 segment_t *seg = getseg(ea);
b587e6ae 409 if (!seg || (seg->type != SEG_CODE && seg->type != SEG_DATA))
d8891fcc 410 goto pass;
411
b587e6ae 412 ea_flags = get_flags_novalue(ea);
413 if (isCode(ea_flags))
414 {
415 if (!decode_insn(ea))
416 goto pass;
d8891fcc 417
1caf86bb 418 if (is_insn_jmp(cmd.itype) && cmd.Operands[0].type == o_near
419 && cmd.Operands[0].dtyp == dt_dword)
420 {
421 jmp_near = 1;
422 }
423 else if ((cmd.itype == NN_pushf || cmd.itype == NN_popf)
424 && natop())
425 {
426 do_pushf = 1;
427 }
428
b587e6ae 429 for (o = 0; o < UA_MAXOP; o++) {
1caf86bb 430 const op_t &opr = cmd.Operands[o];
431 if (opr.type == o_void)
b587e6ae 432 break;
d8891fcc 433
1caf86bb 434 // correct?
435 if (opr.type == o_mem && opr.specval_shorts.high == 0x21)
b587e6ae 436 drop_large = 1;
1caf86bb 437 if (opr.hasSIB && x86_scale(opr) == 0
438 && x86_index(opr) != INDEX_NONE)
439 {
440 set_scale = 1;
441 }
442 // annoying alignment variant..
443 if (opr.type == o_imm && opr.dtyp == dt_dword
444 && (opr.value < 0x80 || opr.value > 0xffffff80)
445 && cmd.size >= opr.offb + 4)
446 {
447 if (get_long(ea + opr.offb) == opr.value)
448 dword_imm = 1;
449 }
450 else if (opr.type == o_imm && opr.dtyp == dt_word
451 && (opr.value < 0x80 || opr.value > 0xff80)
452 && cmd.size >= opr.offb + 2)
453 {
454 if (get_word(ea + opr.offb) == (ushort)opr.value)
455 word_imm = 1;
15c7b2a4 456 }
15c7b2a4 457 }
b587e6ae 458 }
459 else { // not code
460 if (isOff0(ea_flags))
461 drop_rva = 1;
d8891fcc 462 }
463
464pass:
b587e6ae 465 n = ln.get_linecnt();
466 for (i = pl.lnnum; i < n; i++) {
467 do_def_line(buf, sizeof(buf), ln.down());
15c7b2a4 468
1402b79d 469 char *fw;
470 for (fw = buf; *fw != 0 && *fw == ' '; )
471 fw++;
472
1caf86bb 473 // patches..
b587e6ae 474 if (drop_large) {
1402b79d 475 p = strstr(fw, "large ");
b587e6ae 476 if (p != NULL)
477 memmove(p, p + 6, strlen(p + 6) + 1);
478 }
479 while (drop_rva) {
1402b79d 480 p = strstr(fw, " rva ");
b587e6ae 481 if (p == NULL)
482 break;
483 memmove(p, p + 4, strlen(p + 4) + 1);
484 }
1caf86bb 485 if (set_scale) {
1402b79d 486 p = strchr(fw, '[');
1caf86bb 487 if (p != NULL)
488 p = strchr(p, '+');
489 if (p != NULL && p[1] == 'e') {
490 p += 4;
491 // scale is 1, must specify it explicitly so that
492 // masm chooses the right scaled reg
493 memmove(p + 2, p, strlen(p) + 1);
494 memcpy(p, "*1", 2);
495 }
496 }
497 else if (jmp_near) {
1402b79d 498 p = strchr(fw, 'j');
1caf86bb 499 while (p && *p != ' ')
500 p++;
501 while (p && *p == ' ')
502 p++;
503 if (p != NULL) {
504 memmove(p + 9, p, strlen(p) + 1);
505 memcpy(p, "near ptr ", 9);
506 }
507 }
508 if (word_imm) {
1402b79d 509 p = strstr(fw, ", ");
1caf86bb 510 if (p != NULL && '0' <= p[2] && p[2] <= '9') {
511 p += 2;
512 memmove(p + 9, p, strlen(p) + 1);
513 memcpy(p, "word ptr ", 9);
514 }
515 }
516 else if (dword_imm) {
1402b79d 517 p = strstr(fw, ", ");
1caf86bb 518 if (p != NULL && '0' <= p[2] && p[2] <= '9') {
519 p += 2;
520 memmove(p + 10, p, strlen(p) + 1);
521 memcpy(p, "dword ptr ", 10);
522 }
523 }
524 else if (do_pushf) {
1402b79d 525 p = strstr(fw, "pushf");
1caf86bb 526 if (p == NULL)
1402b79d 527 p = strstr(fw, "popf");
1caf86bb 528 if (p != NULL) {
529 p = strchr(p, 'f') + 1;
530 memmove(p + 1, p, strlen(p) + 1);
531 *p = 'd';
532 }
533 }
b587e6ae 534
1402b79d 535 if (fw[0] == 'e' && IS_START(fw, "end") && fw[3] == ' ') {
536 fout_line++;
537 qfprintf(fout, "include public.inc\n\n");
538
539 // kill entry point
540 fw[3] = 0;
541 }
542
b587e6ae 543 fout_line++;
544 qfprintf(fout, "%s\n", buf);
545 }
d8891fcc 546
15c7b2a4 547 // note: next_head skips some undefined stuff
d8891fcc 548 ea = next_not_tail(ea); // correct?
15c7b2a4 549 if (ea == BADADDR)
d8891fcc 550 break;
551
552 pl.ea = ea;
553 pl.lnnum = 0;
554 ln.set_place(&pl);
d8891fcc 555 }
556
557 if (fout != NULL)
558 qfclose(fout);
15c7b2a4 559 if (fname != NULL)
560 qfree(fname);
d8891fcc 561
562 hide_wait_box();
563 msg("%d lines saved.\n", fout_line);
564}
565
566//--------------------------------------------------------------------------
567
568static const char comment[] = "Generate disassembly lines for one address";
569static const char help[] = "Generate asm file\n";
570static const char wanted_name[] = "Save asm";
571static const char wanted_hotkey[] = "Ctrl-F6";
572
573//--------------------------------------------------------------------------
574//
575// PLUGIN DESCRIPTION BLOCK
576//
577//--------------------------------------------------------------------------
578plugin_t PLUGIN =
579{
580 IDP_INTERFACE_VERSION,
581 0, // plugin flags
582 init, // initialize
583 term, // terminate. this pointer may be NULL.
584 run, // invoke plugin
585 comment, // long comment about the plugin
586 // it could appear in the status line
587 // or as a hint
588 help, // multiline help about the plugin
589 wanted_name, // the preferred short name of the plugin
590 wanted_hotkey // the preferred hotkey to run the plugin
591};
592
593// vim:ts=2:shiftwidth=2:expandtab