| 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 | |
| 15 | static const char *asmfn; |
| 16 | static int asmln; |
| 17 | |
| 18 | // note: must be in ascending order |
| 19 | enum dx_type { |
| 20 | DXT_UNSPEC, |
| 21 | DXT_BYTE, |
| 22 | DXT_WORD, |
| 23 | DXT_DWORD, |
| 24 | DXT_QUAD, |
| 25 | DXT_TEN, |
| 26 | }; |
| 27 | |
| 28 | #define anote(fmt, ...) \ |
| 29 | printf("%s:%d: note: " fmt, asmfn, asmln, ##__VA_ARGS__) |
| 30 | #define aerr(fmt, ...) do { \ |
| 31 | printf("%s:%d: error: " fmt, asmfn, asmln, ##__VA_ARGS__); \ |
| 32 | fcloseall(); \ |
| 33 | exit(1); \ |
| 34 | } while (0) |
| 35 | |
| 36 | #include "masm_tools.h" |
| 37 | |
| 38 | static char *next_word_s(char *w, size_t wsize, char *s) |
| 39 | { |
| 40 | int quote = 0; |
| 41 | size_t i; |
| 42 | |
| 43 | s = sskip(s); |
| 44 | |
| 45 | for (i = 0; i < wsize - 1; i++) { |
| 46 | if (s[i] == '\'') |
| 47 | quote ^= 1; |
| 48 | if (s[i] == 0 || (!quote && (my_isblank(s[i]) || s[i] == ','))) |
| 49 | break; |
| 50 | w[i] = s[i]; |
| 51 | } |
| 52 | w[i] = 0; |
| 53 | |
| 54 | if (s[i] != 0 && !my_isblank(s[i]) && s[i] != ',') |
| 55 | printf("warning: '%s' truncated\n", w); |
| 56 | |
| 57 | return s + i; |
| 58 | } |
| 59 | |
| 60 | static void next_section(FILE *fasm, char *name) |
| 61 | { |
| 62 | char words[2][256]; |
| 63 | char line[256]; |
| 64 | int wordc; |
| 65 | char *p; |
| 66 | |
| 67 | name[0] = 0; |
| 68 | |
| 69 | while (fgets(line, sizeof(line), fasm)) |
| 70 | { |
| 71 | wordc = 0; |
| 72 | asmln++; |
| 73 | |
| 74 | p = sskip(line); |
| 75 | if (*p == 0) |
| 76 | continue; |
| 77 | |
| 78 | if (*p == ';') { |
| 79 | while (strlen(line) == sizeof(line) - 1) { |
| 80 | // one of those long comment lines.. |
| 81 | if (!fgets(line, sizeof(line), fasm)) |
| 82 | break; |
| 83 | } |
| 84 | continue; |
| 85 | } |
| 86 | |
| 87 | for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) { |
| 88 | p = sskip(next_word(words[wordc], sizeof(words[0]), p)); |
| 89 | if (*p == 0 || *p == ';') { |
| 90 | wordc++; |
| 91 | break; |
| 92 | } |
| 93 | } |
| 94 | |
| 95 | if (wordc < 2) |
| 96 | continue; |
| 97 | |
| 98 | if (!IS(words[1], "segment")) |
| 99 | continue; |
| 100 | |
| 101 | strcpy(name, words[0]); |
| 102 | break; |
| 103 | } |
| 104 | } |
| 105 | |
| 106 | static enum dx_type parse_dx_directive(const char *name) |
| 107 | { |
| 108 | if (IS(name, "dd")) |
| 109 | return DXT_DWORD; |
| 110 | if (IS(name, "dw")) |
| 111 | return DXT_WORD; |
| 112 | if (IS(name, "db")) |
| 113 | return DXT_BYTE; |
| 114 | if (IS(name, "dq")) |
| 115 | return DXT_QUAD; |
| 116 | if (IS(name, "dt")) |
| 117 | return DXT_TEN; |
| 118 | |
| 119 | return DXT_UNSPEC; |
| 120 | } |
| 121 | |
| 122 | static const char *type_name(enum dx_type type) |
| 123 | { |
| 124 | switch (type) { |
| 125 | case DXT_BYTE: |
| 126 | return ".byte"; |
| 127 | case DXT_WORD: |
| 128 | return ".word"; |
| 129 | case DXT_DWORD: |
| 130 | return ".long"; |
| 131 | case DXT_QUAD: |
| 132 | return ".quad"; |
| 133 | case DXT_TEN: |
| 134 | return ".tfloat"; |
| 135 | case DXT_UNSPEC: |
| 136 | break; |
| 137 | } |
| 138 | return "<bad>"; |
| 139 | } |
| 140 | |
| 141 | static const char *type_name_float(enum dx_type type) |
| 142 | { |
| 143 | switch (type) { |
| 144 | case DXT_DWORD: |
| 145 | return ".float"; |
| 146 | case DXT_QUAD: |
| 147 | return ".double"; |
| 148 | case DXT_TEN: |
| 149 | return ".tfloat"; |
| 150 | default: |
| 151 | break; |
| 152 | } |
| 153 | return "<bad_float>"; |
| 154 | } |
| 155 | |
| 156 | static int type_size(enum dx_type type) |
| 157 | { |
| 158 | switch (type) { |
| 159 | case DXT_BYTE: |
| 160 | return 1; |
| 161 | case DXT_WORD: |
| 162 | return 2; |
| 163 | case DXT_DWORD: |
| 164 | return 4; |
| 165 | case DXT_QUAD: |
| 166 | return 8; |
| 167 | case DXT_TEN: |
| 168 | return 10; |
| 169 | case DXT_UNSPEC: |
| 170 | break; |
| 171 | } |
| 172 | return -1; |
| 173 | } |
| 174 | |
| 175 | static char *escape_string(char *s) |
| 176 | { |
| 177 | char buf[256]; |
| 178 | char *t = buf; |
| 179 | |
| 180 | for (; *s != 0; s++) { |
| 181 | if (*s == '"') { |
| 182 | strcpy(t, "\\22"); |
| 183 | t += strlen(t); |
| 184 | continue; |
| 185 | } |
| 186 | if (*s == '\\') { |
| 187 | strcpy(t, "\\\\"); |
| 188 | t += strlen(t); |
| 189 | continue; |
| 190 | } |
| 191 | *t++ = *s; |
| 192 | } |
| 193 | *t = *s; |
| 194 | return strcpy(s, buf); |
| 195 | } |
| 196 | |
| 197 | static void check_sym(FILE *fhdr, const char *name) |
| 198 | { |
| 199 | const struct parsed_proto *pp; |
| 200 | char buf[256]; |
| 201 | int i, l; |
| 202 | |
| 203 | pp = proto_parse(fhdr, name, 1); |
| 204 | if (pp == NULL) { |
| 205 | if (IS_START(name, "sub_")) |
| 206 | aerr("sub_ sym missing proto: '%s'\n", name); |
| 207 | return; |
| 208 | } |
| 209 | |
| 210 | if (!pp->is_func && !pp->is_fptr) |
| 211 | return; |
| 212 | if (pp->argc_reg == 0) |
| 213 | return; |
| 214 | if (pp->argc_reg == 1 && pp->argc_stack == 0 |
| 215 | && IS(pp->arg[0].reg, "ecx")) |
| 216 | { |
| 217 | return; |
| 218 | } |
| 219 | if (pp->argc_reg == 2 |
| 220 | && IS(pp->arg[0].reg, "ecx") |
| 221 | && IS(pp->arg[1].reg, "edx")) |
| 222 | { |
| 223 | return; |
| 224 | } |
| 225 | snprintf(buf, sizeof(buf), "%s %s(", |
| 226 | pp->ret_type.name, name); |
| 227 | l = strlen(buf); |
| 228 | |
| 229 | for (i = 0; i < pp->argc_reg; i++) { |
| 230 | snprintf(buf + l, sizeof(buf) - l, "%s%s", |
| 231 | i == 0 ? "" : ", ", pp->arg[i].reg); |
| 232 | l = strlen(buf); |
| 233 | } |
| 234 | if (pp->argc_stack > 0) { |
| 235 | snprintf(buf + l, sizeof(buf) - l, ", {%d stack}", pp->argc_stack); |
| 236 | l = strlen(buf); |
| 237 | } |
| 238 | snprintf(buf + l, sizeof(buf) - l, ")"); |
| 239 | |
| 240 | aerr("unhandled reg call: %s\n", buf); |
| 241 | } |
| 242 | |
| 243 | static int cmpstringp(const void *p1, const void *p2) |
| 244 | { |
| 245 | return strcmp(*(char * const *)p1, *(char * const *)p2); |
| 246 | } |
| 247 | |
| 248 | int main(int argc, char *argv[]) |
| 249 | { |
| 250 | FILE *fout, *fasm, *fhdr, *frlist; |
| 251 | char words[20][256]; |
| 252 | char word[256]; |
| 253 | char line[256]; |
| 254 | char comment[256]; |
| 255 | unsigned long val; |
| 256 | unsigned long cnt; |
| 257 | const char *sym; |
| 258 | enum dx_type type; |
| 259 | char **pub_syms; |
| 260 | int pub_sym_cnt = 0; |
| 261 | int pub_sym_alloc; |
| 262 | char **rlist; |
| 263 | int rlist_cnt = 0; |
| 264 | int rlist_alloc; |
| 265 | int is_label; |
| 266 | int is_bss; |
| 267 | int wordc; |
| 268 | int first; |
| 269 | int arg_out; |
| 270 | int arg = 1; |
| 271 | int len; |
| 272 | int w, i; |
| 273 | char *p; |
| 274 | char *p2; |
| 275 | |
| 276 | if (argc < 4) { |
| 277 | printf("usage:\n%s <.s> <.asm> <hdrf> [rlist]*\n", |
| 278 | argv[0]); |
| 279 | return 1; |
| 280 | } |
| 281 | |
| 282 | arg_out = arg++; |
| 283 | |
| 284 | asmfn = argv[arg++]; |
| 285 | fasm = fopen(asmfn, "r"); |
| 286 | my_assert_not(fasm, NULL); |
| 287 | |
| 288 | hdrfn = argv[arg++]; |
| 289 | fhdr = fopen(hdrfn, "r"); |
| 290 | my_assert_not(fhdr, NULL); |
| 291 | |
| 292 | fout = fopen(argv[arg_out], "w"); |
| 293 | my_assert_not(fout, NULL); |
| 294 | |
| 295 | comment[0] = 0; |
| 296 | |
| 297 | pub_sym_alloc = 64; |
| 298 | pub_syms = malloc(pub_sym_alloc * sizeof(pub_syms[0])); |
| 299 | my_assert_not(pub_syms, NULL); |
| 300 | |
| 301 | rlist_alloc = 64; |
| 302 | rlist = malloc(rlist_alloc * sizeof(rlist[0])); |
| 303 | my_assert_not(rlist, NULL); |
| 304 | |
| 305 | for (; arg < argc; arg++) { |
| 306 | frlist = fopen(argv[arg], "r"); |
| 307 | my_assert_not(frlist, NULL); |
| 308 | |
| 309 | while (fgets(line, sizeof(line), frlist)) { |
| 310 | p = sskip(line); |
| 311 | if (*p == 0 || *p == ';') |
| 312 | continue; |
| 313 | |
| 314 | p = next_word(words[0], sizeof(words[0]), p); |
| 315 | if (words[0][0] == 0) |
| 316 | continue; |
| 317 | |
| 318 | if (rlist_cnt >= rlist_alloc) { |
| 319 | rlist_alloc = rlist_alloc * 2 + 64; |
| 320 | rlist = realloc(rlist, rlist_alloc * sizeof(rlist[0])); |
| 321 | my_assert_not(rlist, NULL); |
| 322 | } |
| 323 | rlist[rlist_cnt++] = strdup(words[0]); |
| 324 | } |
| 325 | |
| 326 | fclose(frlist); |
| 327 | frlist = NULL; |
| 328 | } |
| 329 | |
| 330 | if (rlist_cnt > 0) |
| 331 | qsort(rlist, rlist_cnt, sizeof(rlist[0]), cmpstringp); |
| 332 | |
| 333 | while (1) { |
| 334 | next_section(fasm, line); |
| 335 | if (feof(fasm)) |
| 336 | break; |
| 337 | if (IS(line + 1, "text")) |
| 338 | continue; |
| 339 | |
| 340 | if (IS(line + 1, "rdata")) |
| 341 | fprintf(fout, "\n.section .rodata\n"); |
| 342 | else if (IS(line + 1, "data")) |
| 343 | fprintf(fout, "\n.data\n"); |
| 344 | else |
| 345 | aerr("unhandled section: '%s'\n", line); |
| 346 | |
| 347 | fprintf(fout, ".align 4\n"); |
| 348 | |
| 349 | while (fgets(line, sizeof(line), fasm)) |
| 350 | { |
| 351 | sym = NULL; |
| 352 | asmln++; |
| 353 | |
| 354 | p = sskip(line); |
| 355 | if (*p == 0 || *p == ';') |
| 356 | continue; |
| 357 | |
| 358 | for (wordc = 0; wordc < ARRAY_SIZE(words); wordc++) { |
| 359 | p = sskip(next_word_s(words[wordc], sizeof(words[0]), p)); |
| 360 | if (*p == 0 || *p == ';') { |
| 361 | wordc++; |
| 362 | break; |
| 363 | } |
| 364 | if (*p == ',') { |
| 365 | p = sskip(p + 1); |
| 366 | } |
| 367 | } |
| 368 | |
| 369 | if (wordc == 2 && IS(words[1], "ends")) |
| 370 | break; |
| 371 | if (wordc <= 2 && IS(words[0], "end")) |
| 372 | break; |
| 373 | if (wordc < 2) |
| 374 | aerr("unhandled: '%s'\n", words[0]); |
| 375 | |
| 376 | // don't cares |
| 377 | if (IS(words[0], "assume")) |
| 378 | continue; |
| 379 | |
| 380 | if (IS(words[0], "align")) { |
| 381 | val = parse_number(words[1]); |
| 382 | fprintf(fout, "\t\t .align %ld", val); |
| 383 | goto fin; |
| 384 | } |
| 385 | |
| 386 | w = 1; |
| 387 | type = parse_dx_directive(words[0]); |
| 388 | if (type == DXT_UNSPEC) { |
| 389 | type = parse_dx_directive(words[1]); |
| 390 | sym = words[0]; |
| 391 | w = 2; |
| 392 | } |
| 393 | if (type == DXT_UNSPEC) |
| 394 | aerr("unhandled decl: '%s %s'\n", words[0], words[1]); |
| 395 | |
| 396 | if (sym != NULL) { |
| 397 | // public/global name |
| 398 | if (pub_sym_cnt >= pub_sym_alloc) { |
| 399 | pub_sym_alloc *= 2; |
| 400 | pub_syms = realloc(pub_syms, pub_sym_alloc * sizeof(pub_syms[0])); |
| 401 | my_assert_not(pub_syms, NULL); |
| 402 | } |
| 403 | pub_syms[pub_sym_cnt++] = strdup(sym); |
| 404 | |
| 405 | len = strlen(sym); |
| 406 | fprintf(fout, "_%s:", sym); |
| 407 | |
| 408 | len += 2; |
| 409 | if (len < 8) |
| 410 | fprintf(fout, "\t"); |
| 411 | if (len < 16) |
| 412 | fprintf(fout, "\t"); |
| 413 | if (len <= 16) |
| 414 | fprintf(fout, " "); |
| 415 | else |
| 416 | fprintf(fout, " "); |
| 417 | } |
| 418 | else { |
| 419 | fprintf(fout, "\t\t "); |
| 420 | } |
| 421 | |
| 422 | if (type == DXT_BYTE && words[w][0] == '\'') { |
| 423 | // string; use asciz for most common case |
| 424 | if (w == wordc - 2 && IS(words[w + 1], "0")) { |
| 425 | fprintf(fout, ".asciz \""); |
| 426 | wordc--; |
| 427 | } |
| 428 | else |
| 429 | fprintf(fout, ".ascii \""); |
| 430 | |
| 431 | for (; w < wordc; w++) { |
| 432 | if (words[w][0] == '\'') { |
| 433 | p = words[w] + 1; |
| 434 | p2 = strchr(p, '\''); |
| 435 | if (p2 == NULL) |
| 436 | aerr("unterminated string? '%s'\n", p); |
| 437 | memcpy(word, p, p2 - p); |
| 438 | word[p2 - p] = 0; |
| 439 | fprintf(fout, "%s", escape_string(word)); |
| 440 | } |
| 441 | else { |
| 442 | val = parse_number(words[w]); |
| 443 | if (val & ~0xff) |
| 444 | aerr("bad string trailing byte?\n"); |
| 445 | fprintf(fout, "\\x%02lx", val); |
| 446 | } |
| 447 | } |
| 448 | fprintf(fout, "\""); |
| 449 | goto fin; |
| 450 | } |
| 451 | |
| 452 | if (w == wordc - 2) { |
| 453 | if (IS_START(words[w + 1], "dup(")) { |
| 454 | cnt = parse_number(words[w]); |
| 455 | p = words[w + 1] + 4; |
| 456 | p2 = strchr(p, ')'); |
| 457 | if (p2 == NULL) |
| 458 | aerr("bad dup?\n"); |
| 459 | memmove(word, p, p2 - p); |
| 460 | word[p2 - p] = 0; |
| 461 | |
| 462 | val = 0; |
| 463 | if (!IS(word, "?")) |
| 464 | val = parse_number(word); |
| 465 | |
| 466 | fprintf(fout, ".fill 0x%02lx,%d,0x%02lx", |
| 467 | cnt, type_size(type), val); |
| 468 | goto fin; |
| 469 | } |
| 470 | } |
| 471 | |
| 472 | if (type == DXT_DWORD && words[w][0] == '\'' |
| 473 | && words[w][5] == '\'' && strlen(words[w]) == 6) |
| 474 | { |
| 475 | if (w != wordc - 1) |
| 476 | aerr("TODO\n"); |
| 477 | |
| 478 | p = words[w]; |
| 479 | val = (p[1] << 24) | (p[2] << 16) | (p[3] << 8) | p[4]; |
| 480 | fprintf(fout, ".long 0x%lx", val); |
| 481 | snprintf(comment, sizeof(comment), "%s", words[w]); |
| 482 | goto fin; |
| 483 | } |
| 484 | |
| 485 | if (type >= DXT_DWORD && strchr(words[w], '.')) |
| 486 | { |
| 487 | if (w != wordc - 1) |
| 488 | aerr("TODO\n"); |
| 489 | |
| 490 | fprintf(fout, "%s %s", type_name_float(type), words[w]); |
| 491 | goto fin; |
| 492 | } |
| 493 | |
| 494 | first = 1; |
| 495 | fprintf(fout, "%s ", type_name(type)); |
| 496 | for (; w < wordc; w++) |
| 497 | { |
| 498 | if (!first) |
| 499 | fprintf(fout, ", "); |
| 500 | |
| 501 | is_label = is_bss = 0; |
| 502 | if (w <= wordc - 2 && IS(words[w], "offset")) { |
| 503 | is_label = 1; |
| 504 | w++; |
| 505 | } |
| 506 | else if (IS(words[w], "?")) { |
| 507 | is_bss = 1; |
| 508 | } |
| 509 | else if (type == DXT_DWORD |
| 510 | && !('0' <= words[w][0] && words[w][0] <= '9')) |
| 511 | { |
| 512 | // assume label |
| 513 | is_label = 1; |
| 514 | } |
| 515 | |
| 516 | if (is_bss) { |
| 517 | fprintf(fout, "0"); |
| 518 | } |
| 519 | else if (is_label) { |
| 520 | p = words[w]; |
| 521 | if (IS_START(p, "loc_") || strchr(p, '?') || strchr(p, '@') |
| 522 | || bsearch(&p, rlist, rlist_cnt, sizeof(rlist[0]), |
| 523 | cmpstringp)) |
| 524 | { |
| 525 | fprintf(fout, "0"); |
| 526 | snprintf(comment, sizeof(comment), "%s", p); |
| 527 | } |
| 528 | else { |
| 529 | check_sym(fhdr, p); |
| 530 | fprintf(fout, "_%s", p); |
| 531 | } |
| 532 | } |
| 533 | else { |
| 534 | val = parse_number(words[w]); |
| 535 | if (val < 10) |
| 536 | fprintf(fout, "%ld", val); |
| 537 | else |
| 538 | fprintf(fout, "0x%lx", val); |
| 539 | } |
| 540 | |
| 541 | first = 0; |
| 542 | } |
| 543 | |
| 544 | fin: |
| 545 | if (comment[0] != 0) { |
| 546 | fprintf(fout, "\t\t# %s", comment); |
| 547 | comment[0] = 0; |
| 548 | } |
| 549 | fprintf(fout, "\n"); |
| 550 | (void)proto_parse; |
| 551 | } |
| 552 | } |
| 553 | |
| 554 | fprintf(fout, "\n"); |
| 555 | |
| 556 | // dump public syms |
| 557 | for (i = 0; i < pub_sym_cnt; i++) |
| 558 | fprintf(fout, ".global _%s\n", pub_syms[i]); |
| 559 | |
| 560 | fclose(fout); |
| 561 | fclose(fasm); |
| 562 | fclose(fhdr); |
| 563 | |
| 564 | return 0; |
| 565 | } |
| 566 | |
| 567 | // vim:ts=2:shiftwidth=2:expandtab |