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