1 /* $NetBSD: read.c,v 1.25 2009/04/15 01:20:57 christos Exp $ */ 2 3 /* 4 * Copyright (c) 1996 Christopher G. Demetriou. All Rights Reserved. 5 * Copyright (c) 1994, 1995 Jochen Pohl 6 * All Rights Reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. All advertising materials mentioning features or use of this software 17 * must display the following acknowledgement: 18 * This product includes software developed by Jochen Pohl for 19 * The NetBSD Project. 20 * 4. The name of the author may not be used to endorse or promote products 21 * derived from this software without specific prior written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 24 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 25 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 26 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 27 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 28 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 29 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 30 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 31 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 32 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 33 */ 34 35 #if HAVE_NBTOOL_CONFIG_H 36 #include "nbtool_config.h" 37 #endif 38 39 #include <sys/cdefs.h> 40 #if defined(__RCSID) && !defined(lint) 41 __RCSID("$NetBSD: read.c,v 1.25 2009/04/15 01:20:57 christos Exp $"); 42 #endif 43 44 #include <ctype.h> 45 #include <limits.h> 46 #include <stdio.h> 47 #include <stdlib.h> 48 #include <string.h> 49 #include <stdarg.h> 50 51 #include "lint2.h" 52 53 54 /* index of current (included) source file */ 55 static int srcfile; 56 57 /* 58 * The array pointed to by inpfns maps the file name indices of input files 59 * to the file name indices used in lint2 60 */ 61 static short *inpfns; 62 static size_t ninpfns; 63 64 /* 65 * The array pointed to by *fnames maps file name indizes to file names. 66 * Indices of type short are used instead of pointers to save memory. 67 */ 68 const char **fnames; 69 static size_t *flines; 70 static size_t nfnames; 71 72 /* 73 * Types are shared (to save memory for the types itself) and accessed 74 * via indices (to save memory for references to types (indices are short)). 75 * To share types, a equal type must be located fast. This is done by a 76 * hash table. Access by indices is done via an array of pointers to the 77 * types. 78 */ 79 typedef struct thtab { 80 const char *th_name; 81 u_short th_idx; 82 struct thtab *th_nxt; 83 } thtab_t; 84 static thtab_t **thtab; /* hash table */ 85 type_t **tlst; /* array for indexed access */ 86 static size_t tlstlen; /* length of tlst */ 87 88 static hte_t **renametab; 89 90 /* index of current C source file (as spezified at the command line) */ 91 static int csrcfile; 92 93 94 #define inperr(fmt, args...) \ 95 inperror(__FILE__, __LINE__, fmt, ##args) 96 static void inperror(const char *, size_t, const char *, ...); 97 static void setsrc(const char *); 98 static void setfnid(int, const char *); 99 static void funccall(pos_t *, const char *); 100 static void decldef(pos_t *, const char *); 101 static void usedsym(pos_t *, const char *); 102 static u_short inptype(const char *, const char **); 103 static int gettlen(const char *, const char **); 104 static u_short findtype(const char *, size_t, int); 105 static u_short storetyp(type_t *, const char *, size_t, int); 106 static int thash(const char *, size_t); 107 static char *inpqstrg(const char *, const char **); 108 static const char *inpname(const char *, const char **); 109 static int getfnidx(const char *); 110 111 void 112 readfile(const char *name) 113 { 114 FILE *inp; 115 size_t len; 116 const char *cp; 117 char *line, *eptr, rt = '\0'; 118 int cline, isrc, iline; 119 pos_t pos; 120 121 if (inpfns == NULL) 122 inpfns = xcalloc(ninpfns = 128, sizeof (short)); 123 if (fnames == NULL) 124 fnames = xcalloc(nfnames = 256, sizeof (char *)); 125 if (flines == NULL) 126 flines = xcalloc(nfnames, sizeof (size_t)); 127 if (tlstlen == 0) 128 tlst = xcalloc(tlstlen = 256, sizeof (type_t *)); 129 if (thtab == NULL) 130 thtab = xcalloc(THSHSIZ2, sizeof (thtab_t)); 131 132 _inithash(&renametab); 133 134 srcfile = getfnidx(name); 135 136 if ((inp = fopen(name, "r")) == NULL) 137 err(1, "cannot open %s", name); 138 139 while ((line = fgetln(inp, &len)) != NULL) { 140 flines[srcfile]++; 141 142 if (len == 0 || line[len - 1] != '\n') 143 inperr("%s", &line[len - 1]); 144 line[len - 1] = '\0'; 145 cp = line; 146 147 /* line number in csrcfile */ 148 cline = (int)strtol(cp, &eptr, 10); 149 if (cp == eptr) { 150 cline = -1; 151 } else { 152 cp = eptr; 153 } 154 155 /* record type */ 156 if (*cp != '\0') { 157 rt = *cp++; 158 } else { 159 inperr("null cp"); 160 } 161 162 if (rt == 'S') { 163 setsrc(cp); 164 continue; 165 } else if (rt == 's') { 166 setfnid(cline, cp); 167 continue; 168 } 169 170 /* 171 * Index of (included) source file. If this index is 172 * different from csrcfile, it refers to an included 173 * file. 174 */ 175 isrc = (int)strtol(cp, &eptr, 10); 176 if (cp == eptr) 177 inperr("not a number: %s", cp); 178 cp = eptr; 179 isrc = inpfns[isrc]; 180 181 /* line number in isrc */ 182 if (*cp++ != '.') 183 inperr("bad line number"); 184 iline = (int)strtol(cp, &eptr, 10); 185 if (cp == eptr) 186 inperr("not a number: %s", cp); 187 cp = eptr; 188 189 pos.p_src = (u_short)csrcfile; 190 pos.p_line = (u_short)cline; 191 pos.p_isrc = (u_short)isrc; 192 pos.p_iline = (u_short)iline; 193 194 /* process rest of this record */ 195 switch (rt) { 196 case 'c': 197 funccall(&pos, cp); 198 break; 199 case 'd': 200 decldef(&pos, cp); 201 break; 202 case 'u': 203 usedsym(&pos, cp); 204 break; 205 default: 206 inperr("bad record type %c", rt); 207 } 208 209 } 210 211 _destroyhash(renametab); 212 213 if (ferror(inp)) 214 err(1, "read error on %s", name); 215 216 (void)fclose(inp); 217 } 218 219 220 static void 221 inperror(const char *file, size_t line, const char *fmt, ...) 222 { 223 va_list ap; 224 char buf[1024]; 225 226 va_start(ap, fmt); 227 (void)vsnprintf(buf, sizeof(buf), fmt, ap); 228 va_end(ap); 229 230 errx(1, "%s,%zu: input file error: %s,%zu (%s)", file, line, 231 fnames[srcfile], flines[srcfile], buf); 232 } 233 234 /* 235 * Set the name of the C source file of the .ln file which is 236 * currently read. 237 */ 238 static void 239 setsrc(const char *cp) 240 { 241 242 csrcfile = getfnidx(cp); 243 } 244 245 /* 246 * setfnid() gets as input an index as used in an input file and the 247 * associated file name. If necessary, it creates a new lint2 file 248 * name index for this file name and creates the mapping of the index 249 * as used in the input file to the index used in lint2. 250 */ 251 static void 252 setfnid(int fid, const char *cp) 253 { 254 255 if (fid < 0) 256 inperr("bad fid"); 257 258 if ((size_t)fid >= ninpfns) { 259 inpfns = xrealloc(inpfns, (ninpfns * 2) * sizeof (short)); 260 (void)memset(inpfns + ninpfns, 0, ninpfns * sizeof (short)); 261 ninpfns *= 2; 262 } 263 /* 264 * Should always be true because indices written in the output 265 * file by lint1 are always the previous index + 1. 266 */ 267 if ((size_t)fid >= ninpfns) 268 errx(1, "internal error: setfnid()"); 269 inpfns[fid] = (u_short)getfnidx(cp); 270 } 271 272 /* 273 * Process a function call record (c-record). 274 */ 275 static void 276 funccall(pos_t *posp, const char *cp) 277 { 278 arginf_t *ai, **lai; 279 char c, *eptr; 280 int rused, rdisc; 281 hte_t *hte; 282 fcall_t *fcall; 283 const char *name; 284 285 fcall = xalloc(sizeof (fcall_t)); 286 STRUCT_ASSIGN(fcall->f_pos, *posp); 287 288 /* read flags */ 289 rused = rdisc = 0; 290 lai = &fcall->f_args; 291 while ((c = *cp) == 'u' || c == 'i' || c == 'd' || 292 c == 'z' || c == 'p' || c == 'n' || c == 's') { 293 cp++; 294 switch (c) { 295 case 'u': 296 if (rused || rdisc) 297 inperr("used or discovered: %c", c); 298 rused = 1; 299 break; 300 case 'i': 301 if (rused || rdisc) 302 inperr("used or discovered: %c", c); 303 break; 304 case 'd': 305 if (rused || rdisc) 306 inperr("used or discovered: %c", c); 307 rdisc = 1; 308 break; 309 case 'z': 310 case 'p': 311 case 'n': 312 case 's': 313 ai = xalloc(sizeof (arginf_t)); 314 ai->a_num = (int)strtol(cp, &eptr, 10); 315 if (cp == eptr) 316 inperr("bad number: %s", cp); 317 cp = eptr; 318 if (c == 'z') { 319 ai->a_pcon = ai->a_zero = 1; 320 } else if (c == 'p') { 321 ai->a_pcon = 1; 322 } else if (c == 'n') { 323 ai->a_ncon = 1; 324 } else { 325 ai->a_fmt = 1; 326 ai->a_fstrg = inpqstrg(cp, &cp); 327 } 328 *lai = ai; 329 lai = &ai->a_nxt; 330 break; 331 } 332 } 333 fcall->f_rused = rused; 334 fcall->f_rdisc = rdisc; 335 336 /* read name of function */ 337 name = inpname(cp, &cp); 338 339 /* first look it up in the renaming table, then in the normal table */ 340 hte = _hsearch(renametab, name, 0); 341 if (hte != NULL) 342 hte = hte->h_hte; 343 else 344 hte = hsearch(name, 1); 345 hte->h_used = 1; 346 347 fcall->f_type = inptype(cp, &cp); 348 349 *hte->h_lcall = fcall; 350 hte->h_lcall = &fcall->f_nxt; 351 352 if (*cp != '\0') 353 inperr("trailing line data: %s", cp); 354 } 355 356 /* 357 * Process a declaration or definition (d-record). 358 */ 359 static void 360 decldef(pos_t *posp, const char *cp) 361 { 362 sym_t *symp, sym; 363 char c, *ep, *pos1, *tname; 364 int used, renamed; 365 hte_t *hte, *renamehte = NULL; 366 const char *name, *newname; 367 368 (void)memset(&sym, 0, sizeof (sym)); 369 STRUCT_ASSIGN(sym.s_pos, *posp); 370 sym.s_def = NODECL; 371 372 used = 0; 373 374 while (strchr("tdeurosvPS", (c = *cp)) != NULL) { 375 cp++; 376 switch (c) { 377 case 't': 378 if (sym.s_def != NODECL) 379 inperr("nodecl %c", c); 380 sym.s_def = TDEF; 381 break; 382 case 'd': 383 if (sym.s_def != NODECL) 384 inperr("nodecl %c", c); 385 sym.s_def = DEF; 386 break; 387 case 'e': 388 if (sym.s_def != NODECL) 389 inperr("nodecl %c", c); 390 sym.s_def = DECL; 391 break; 392 case 'u': 393 if (used) 394 inperr("used %c", c); 395 used = 1; 396 break; 397 case 'r': 398 if (sym.s_rval) 399 inperr("rval"); 400 sym.s_rval = 1; 401 break; 402 case 'o': 403 if (sym.s_osdef) 404 inperr("osdef"); 405 sym.s_osdef = 1; 406 break; 407 case 's': 408 if (sym.s_static) 409 inperr("static"); 410 sym.s_static = 1; 411 break; 412 case 'v': 413 if (sym.s_va) 414 inperr("va"); 415 sym.s_va = 1; 416 sym.s_nva = (short)strtol(cp, &ep, 10); 417 if (cp == ep) 418 inperr("bad number: %s", cp); 419 cp = ep; 420 break; 421 case 'P': 422 if (sym.s_prfl) 423 inperr("prfl"); 424 sym.s_prfl = 1; 425 sym.s_nprfl = (short)strtol(cp, &ep, 10); 426 if (cp == ep) 427 inperr("bad number: %s", cp); 428 cp = ep; 429 break; 430 case 'S': 431 if (sym.s_scfl) 432 inperr("scfl"); 433 sym.s_scfl = 1; 434 sym.s_nscfl = (short)strtol(cp, &ep, 10); 435 if (cp == ep) 436 inperr("bad number: %s", cp); 437 cp = ep; 438 break; 439 } 440 } 441 442 /* read symbol name, doing renaming if necessary */ 443 name = inpname(cp, &cp); 444 renamed = 0; 445 if (*cp == 'r') { 446 cp++; 447 tname = xstrdup(name); 448 newname = inpname(cp, &cp); 449 450 /* enter it and see if it's already been renamed */ 451 renamehte = _hsearch(renametab, tname, 1); 452 if (renamehte->h_hte == NULL) { 453 hte = hsearch(newname, 1); 454 renamehte->h_hte = hte; 455 renamed = 1; 456 } else if (strcmp((hte = renamehte->h_hte)->h_name, newname)) { 457 pos1 = xstrdup(mkpos(&renamehte->h_syms->s_pos)); 458 /* %s renamed multiple times\t%s :: %s */ 459 msg(18, tname, pos1, mkpos(&sym.s_pos)); 460 free(pos1); 461 } 462 free(tname); 463 } else { 464 /* it might be a previously-done rename */ 465 hte = _hsearch(renametab, name, 0); 466 if (hte != NULL) 467 hte = hte->h_hte; 468 else 469 hte = hsearch(name, 1); 470 } 471 hte->h_used |= used; 472 if (sym.s_def == DEF || sym.s_def == TDEF) 473 hte->h_def = 1; 474 475 sym.s_type = inptype(cp, &cp); 476 477 /* 478 * Allocate memory for this symbol only if it was not already 479 * declared or tentatively defined at the same location with 480 * the same type. Works only for symbols with external linkage, 481 * because static symbols, tentatively defined at the same location 482 * but in different translation units are really different symbols. 483 */ 484 for (symp = hte->h_syms; symp != NULL; symp = symp->s_nxt) { 485 if (symp->s_pos.p_isrc == sym.s_pos.p_isrc && 486 symp->s_pos.p_iline == sym.s_pos.p_iline && 487 symp->s_type == sym.s_type && 488 ((symp->s_def == DECL && sym.s_def == DECL) || 489 (!sflag && symp->s_def == TDEF && sym.s_def == TDEF)) && 490 !symp->s_static && !sym.s_static) { 491 break; 492 } 493 } 494 495 if (symp == NULL) { 496 /* allocsym reserviert keinen Platz fuer s_nva */ 497 if (sym.s_va || sym.s_prfl || sym.s_scfl) { 498 symp = xalloc(sizeof (sym_t)); 499 STRUCT_ASSIGN(*symp, sym); 500 } else { 501 symp = xalloc(sizeof (symp->s_s)); 502 STRUCT_ASSIGN(symp->s_s, sym.s_s); 503 } 504 *hte->h_lsym = symp; 505 hte->h_lsym = &symp->s_nxt; 506 507 /* XXX hack so we can remember where a symbol was renamed */ 508 if (renamed) 509 renamehte->h_syms = symp; 510 } 511 512 if (*cp != '\0') 513 inperr("trailing line: %s", cp); 514 } 515 516 /* 517 * Read an u-record (emited by lint1 if a symbol was used). 518 */ 519 static void 520 usedsym(pos_t *posp, const char *cp) 521 { 522 usym_t *usym; 523 hte_t *hte; 524 const char *name; 525 526 usym = xalloc(sizeof (usym_t)); 527 STRUCT_ASSIGN(usym->u_pos, *posp); 528 529 /* needed as delimiter between two numbers */ 530 if (*cp++ != 'x') 531 inperr("bad delim %c", cp[-1]); 532 533 name = inpname(cp, &cp); 534 hte = _hsearch(renametab, name, 0); 535 if (hte != NULL) 536 hte = hte->h_hte; 537 else 538 hte = hsearch(name, 1); 539 hte->h_used = 1; 540 541 *hte->h_lusym = usym; 542 hte->h_lusym = &usym->u_nxt; 543 } 544 545 /* 546 * Read a type and return the index of this type. 547 */ 548 static u_short 549 inptype(const char *cp, const char **epp) 550 { 551 char c, s, *eptr; 552 const char *ep; 553 type_t *tp; 554 int narg, i, osdef = 0; 555 size_t tlen; 556 u_short tidx, sidx; 557 int h; 558 559 /* If we have this type already, return it's index. */ 560 tlen = gettlen(cp, &ep); 561 h = thash(cp, tlen); 562 if ((tidx = findtype(cp, tlen, h)) != 0) { 563 *epp = ep; 564 return (tidx); 565 } 566 567 /* No, we must create a new type. */ 568 tp = xalloc(sizeof (type_t)); 569 570 tidx = storetyp(tp, cp, tlen, h); 571 572 c = *cp++; 573 574 while (c == 'c' || c == 'v') { 575 if (c == 'c') { 576 tp->t_const = 1; 577 } else { 578 tp->t_volatile = 1; 579 } 580 c = *cp++; 581 } 582 583 switch (c) { 584 case 's': 585 case 'u': 586 case 'l': 587 case 'e': 588 s = c; 589 c = *cp++; 590 break; 591 default: 592 s = '\0'; 593 break; 594 } 595 596 switch (c) { 597 case 'B': 598 tp->t_tspec = BOOL; 599 break; 600 case 'C': 601 tp->t_tspec = s == 's' ? SCHAR : (s == 'u' ? UCHAR : CHAR); 602 break; 603 case 'S': 604 tp->t_tspec = s == 'u' ? USHORT : SHORT; 605 break; 606 case 'I': 607 tp->t_tspec = s == 'u' ? UINT : INT; 608 break; 609 case 'L': 610 tp->t_tspec = s == 'u' ? ULONG : LONG; 611 break; 612 case 'Q': 613 tp->t_tspec = s == 'u' ? UQUAD : QUAD; 614 break; 615 case 'D': 616 tp->t_tspec = s == 's' ? FLOAT : (s == 'l' ? LDOUBLE : DOUBLE); 617 break; 618 case 'V': 619 tp->t_tspec = VOID; 620 break; 621 case 'P': 622 tp->t_tspec = PTR; 623 break; 624 case 'A': 625 tp->t_tspec = ARRAY; 626 break; 627 case 'F': 628 case 'f': 629 osdef = c == 'f'; 630 tp->t_tspec = FUNC; 631 break; 632 case 'T': 633 tp->t_tspec = s == 'e' ? ENUM : (s == 's' ? STRUCT : UNION); 634 break; 635 case 'X': 636 tp->t_tspec = s == 's' ? FCOMPLEX 637 : (s == 'l' ? LCOMPLEX : DCOMPLEX); 638 break; 639 } 640 641 switch (tp->t_tspec) { 642 case ARRAY: 643 tp->t_dim = (int)strtol(cp, &eptr, 10); 644 cp = eptr; 645 sidx = inptype(cp, &cp); /* force seq. point! (ditto below) */ 646 tp->t_subt = TP(sidx); 647 break; 648 case PTR: 649 sidx = inptype(cp, &cp); 650 tp->t_subt = TP(sidx); 651 break; 652 case FUNC: 653 c = *cp; 654 if (isdigit((u_char)c)) { 655 if (!osdef) 656 tp->t_proto = 1; 657 narg = (int)strtol(cp, &eptr, 10); 658 cp = eptr; 659 tp->t_args = xcalloc((size_t)(narg + 1), 660 sizeof (type_t *)); 661 for (i = 0; i < narg; i++) { 662 if (i == narg - 1 && *cp == 'E') { 663 tp->t_vararg = 1; 664 cp++; 665 } else { 666 sidx = inptype(cp, &cp); 667 tp->t_args[i] = TP(sidx); 668 } 669 } 670 } 671 sidx = inptype(cp, &cp); 672 tp->t_subt = TP(sidx); 673 break; 674 case ENUM: 675 tp->t_tspec = INT; 676 tp->t_isenum = 1; 677 /* FALLTHROUGH */ 678 case STRUCT: 679 case UNION: 680 switch (*cp++) { 681 case '1': 682 tp->t_istag = 1; 683 tp->t_tag = hsearch(inpname(cp, &cp), 1); 684 break; 685 case '2': 686 tp->t_istynam = 1; 687 tp->t_tynam = hsearch(inpname(cp, &cp), 1); 688 break; 689 case '3': 690 tp->t_isuniqpos = 1; 691 tp->t_uniqpos.p_line = strtol(cp, &eptr, 10); 692 cp = eptr; 693 cp++; 694 /* xlate to 'global' file name. */ 695 tp->t_uniqpos.p_file = 696 addoutfile(inpfns[strtol(cp, &eptr, 10)]); 697 cp = eptr; 698 cp++; 699 tp->t_uniqpos.p_uniq = strtol(cp, &eptr, 10); 700 cp = eptr; 701 break; 702 } 703 break; 704 case LONG: 705 case VOID: 706 case LDOUBLE: 707 case DOUBLE: 708 case FLOAT: 709 case UQUAD: 710 case QUAD: 711 case ULONG: 712 case UINT: 713 case INT: 714 case USHORT: 715 case SHORT: 716 case UCHAR: 717 case SCHAR: 718 case CHAR: 719 case BOOL: 720 case UNSIGN: 721 case SIGNED: 722 case NOTSPEC: 723 case FCOMPLEX: 724 case DCOMPLEX: 725 case LCOMPLEX: 726 case COMPLEX: 727 break; 728 case NTSPEC: 729 abort(); 730 } 731 732 *epp = cp; 733 return (tidx); 734 } 735 736 /* 737 * Get the length of a type string. 738 */ 739 static int 740 gettlen(const char *cp, const char **epp) 741 { 742 const char *cp1; 743 char c, s, *eptr; 744 tspec_t t; 745 int narg, i, cm, vm; 746 747 cp1 = cp; 748 749 c = *cp++; 750 751 cm = vm = 0; 752 753 while (c == 'c' || c == 'v') { 754 if (c == 'c') { 755 if (cm) 756 inperr("cm: %c", c); 757 cm = 1; 758 } else { 759 if (vm) 760 inperr("vm: %c", c); 761 vm = 1; 762 } 763 c = *cp++; 764 } 765 766 switch (c) { 767 case 's': 768 case 'u': 769 case 'l': 770 case 'e': 771 s = c; 772 c = *cp++; 773 break; 774 default: 775 s = '\0'; 776 break; 777 } 778 779 t = NOTSPEC; 780 781 switch (c) { 782 case 'B': 783 if (s == '\0') 784 t = BOOL; 785 break; 786 case 'C': 787 if (s == 's') { 788 t = SCHAR; 789 } else if (s == 'u') { 790 t = UCHAR; 791 } else if (s == '\0') { 792 t = CHAR; 793 } 794 break; 795 case 'S': 796 if (s == 'u') { 797 t = USHORT; 798 } else if (s == '\0') { 799 t = SHORT; 800 } 801 break; 802 case 'I': 803 if (s == 'u') { 804 t = UINT; 805 } else if (s == '\0') { 806 t = INT; 807 } 808 break; 809 case 'L': 810 if (s == 'u') { 811 t = ULONG; 812 } else if (s == '\0') { 813 t = LONG; 814 } 815 break; 816 case 'Q': 817 if (s == 'u') { 818 t = UQUAD; 819 } else if (s == '\0') { 820 t = QUAD; 821 } 822 break; 823 case 'D': 824 if (s == 's') { 825 t = FLOAT; 826 } else if (s == 'l') { 827 t = LDOUBLE; 828 } else if (s == '\0') { 829 t = DOUBLE; 830 } 831 break; 832 case 'V': 833 if (s == '\0') 834 t = VOID; 835 break; 836 case 'P': 837 if (s == '\0') 838 t = PTR; 839 break; 840 case 'A': 841 if (s == '\0') 842 t = ARRAY; 843 break; 844 case 'F': 845 case 'f': 846 if (s == '\0') 847 t = FUNC; 848 break; 849 case 'T': 850 if (s == 'e') { 851 t = ENUM; 852 } else if (s == 's') { 853 t = STRUCT; 854 } else if (s == 'u') { 855 t = UNION; 856 } 857 break; 858 case 'X': 859 if (s == 's') { 860 t = FCOMPLEX; 861 } else if (s == 'l') { 862 t = LCOMPLEX; 863 } else if (s == '\0') { 864 t = DCOMPLEX; 865 } 866 break; 867 default: 868 inperr("bad type: %c %c", c, s); 869 } 870 871 if (t == NOTSPEC) { 872 inperr("undefined type: %c %c", c, s); 873 } 874 875 switch (t) { 876 case ARRAY: 877 (void)strtol(cp, &eptr, 10); 878 if (cp == eptr) 879 inperr("bad number: %s", cp); 880 cp = eptr; 881 (void)gettlen(cp, &cp); 882 break; 883 case PTR: 884 (void)gettlen(cp, &cp); 885 break; 886 case FUNC: 887 c = *cp; 888 if (isdigit((u_char)c)) { 889 narg = (int)strtol(cp, &eptr, 10); 890 cp = eptr; 891 for (i = 0; i < narg; i++) { 892 if (i == narg - 1 && *cp == 'E') { 893 cp++; 894 } else { 895 (void)gettlen(cp, &cp); 896 } 897 } 898 } 899 (void)gettlen(cp, &cp); 900 break; 901 case ENUM: 902 case STRUCT: 903 case UNION: 904 switch (*cp++) { 905 case '1': 906 (void)inpname(cp, &cp); 907 break; 908 case '2': 909 (void)inpname(cp, &cp); 910 break; 911 case '3': 912 /* unique position: line.file.uniquifier */ 913 (void)strtol(cp, &eptr, 10); 914 if (cp == eptr) 915 inperr("bad number: %s", cp); 916 cp = eptr; 917 if (*cp++ != '.') 918 inperr("not dot: %c", cp[-1]); 919 (void)strtol(cp, &eptr, 10); 920 if (cp == eptr) 921 inperr("bad number: %s", cp); 922 cp = eptr; 923 if (*cp++ != '.') 924 inperr("not dot: %c", cp[-1]); 925 (void)strtol(cp, &eptr, 10); 926 if (cp == eptr) 927 inperr("bad number: %s", cp); 928 cp = eptr; 929 break; 930 default: 931 inperr("bad value: %c\n", cp[-1]); 932 } 933 break; 934 case FLOAT: 935 case USHORT: 936 case SHORT: 937 case UCHAR: 938 case SCHAR: 939 case CHAR: 940 case BOOL: 941 case UNSIGN: 942 case SIGNED: 943 case NOTSPEC: 944 case INT: 945 case UINT: 946 case DOUBLE: 947 case LDOUBLE: 948 case VOID: 949 case ULONG: 950 case QUAD: 951 case UQUAD: 952 case LONG: 953 case FCOMPLEX: 954 case DCOMPLEX: 955 case LCOMPLEX: 956 case COMPLEX: 957 break; 958 #ifndef __COVERITY__ 959 case NTSPEC: 960 abort(); 961 #endif 962 } 963 964 *epp = cp; 965 return (cp - cp1); 966 } 967 968 /* 969 * Search a type by it's type string. 970 */ 971 static u_short 972 findtype(const char *cp, size_t len, int h) 973 { 974 thtab_t *thte; 975 976 for (thte = thtab[h]; thte != NULL; thte = thte->th_nxt) { 977 if (strncmp(thte->th_name, cp, len) != 0) 978 continue; 979 if (thte->th_name[len] == '\0') 980 return (thte->th_idx); 981 } 982 983 return (0); 984 } 985 986 /* 987 * Store a type and it's type string so we can later share this type 988 * if we read the same type string from the input file. 989 */ 990 static u_short 991 storetyp(type_t *tp, const char *cp, size_t len, int h) 992 { 993 static u_int tidx = 1; /* 0 is reserved */ 994 thtab_t *thte; 995 char *name; 996 997 if (tidx >= USHRT_MAX) 998 errx(1, "sorry, too many types"); 999 1000 if (tidx == tlstlen - 1) { 1001 tlst = xrealloc(tlst, (tlstlen * 2) * sizeof (type_t *)); 1002 (void)memset(tlst + tlstlen, 0, tlstlen * sizeof (type_t *)); 1003 tlstlen *= 2; 1004 } 1005 1006 tlst[tidx] = tp; 1007 1008 /* create a hash table entry */ 1009 name = xalloc(len + 1); 1010 (void)memcpy(name, cp, len); 1011 name[len] = '\0'; 1012 1013 thte = xalloc(sizeof (thtab_t)); 1014 thte->th_name = name; 1015 thte->th_idx = tidx; 1016 thte->th_nxt = thtab[h]; 1017 thtab[h] = thte; 1018 1019 return ((u_short)tidx++); 1020 } 1021 1022 /* 1023 * Hash function for types 1024 */ 1025 static int 1026 thash(const char *s, size_t len) 1027 { 1028 u_int v; 1029 1030 v = 0; 1031 while (len-- != 0) { 1032 v = (v << sizeof (v)) + (u_char)*s++; 1033 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); 1034 } 1035 return (v % THSHSIZ2); 1036 } 1037 1038 /* 1039 * Read a string enclosed by "". This string may contain quoted chars. 1040 */ 1041 static char * 1042 inpqstrg(const char *src, const char **epp) 1043 { 1044 char *strg, *dst; 1045 size_t slen; 1046 int c; 1047 int v; 1048 1049 dst = strg = xmalloc(slen = 32); 1050 1051 if ((c = *src++) != '"') 1052 inperr("not quote: %c", c); 1053 if ((c = *src++) == '\0') 1054 inperr("trailing data: %c", c); 1055 1056 while (c != '"') { 1057 if (c == '\\') { 1058 if ((c = *src++) == '\0') 1059 inperr("missing after \\"); 1060 switch (c) { 1061 case 'n': 1062 c = '\n'; 1063 break; 1064 case 't': 1065 c = '\t'; 1066 break; 1067 case 'v': 1068 c = '\v'; 1069 break; 1070 case 'b': 1071 c = '\b'; 1072 break; 1073 case 'r': 1074 c = '\r'; 1075 break; 1076 case 'f': 1077 c = '\f'; 1078 break; 1079 case 'a': 1080 c = '\a'; 1081 break; 1082 case '\\': 1083 c = '\\'; 1084 break; 1085 case '"': 1086 c = '"'; 1087 break; 1088 case '\'': 1089 c = '\''; 1090 break; 1091 case '0': case '1': case '2': case '3': 1092 v = (c - '0') << 6; 1093 if ((c = *src++) < '0' || c > '7') 1094 inperr("not octal: %c", c); 1095 v |= (c - '0') << 3; 1096 if ((c = *src++) < '0' || c > '7') 1097 inperr("not octal: %c", c); 1098 v |= c - '0'; 1099 c = (u_char)v; 1100 break; 1101 default: 1102 inperr("bad \\ escape: %c", c); 1103 } 1104 } 1105 /* keep space for trailing '\0' */ 1106 if ((size_t)(dst - strg) == slen - 1) { 1107 strg = xrealloc(strg, slen * 2); 1108 dst = strg + (slen - 1); 1109 slen *= 2; 1110 } 1111 *dst++ = (char)c; 1112 if ((c = *src++) == '\0') 1113 inperr("missing closing quote"); 1114 } 1115 *dst = '\0'; 1116 1117 *epp = src; 1118 return (strg); 1119 } 1120 1121 /* 1122 * Read the name of a symbol in static memory. 1123 */ 1124 static const char * 1125 inpname(const char *cp, const char **epp) 1126 { 1127 static char *buf; 1128 static size_t blen = 0; 1129 size_t len, i; 1130 char *eptr, c; 1131 1132 len = (int)strtol(cp, &eptr, 10); 1133 if (cp == eptr) 1134 inperr("bad number: %s", cp); 1135 cp = eptr; 1136 if (len + 1 > blen) 1137 buf = xrealloc(buf, blen = len + 1); 1138 for (i = 0; i < len; i++) { 1139 c = *cp++; 1140 if (!isalnum((unsigned char)c) && c != '_') 1141 inperr("not alnum or _: %c", c); 1142 buf[i] = c; 1143 } 1144 buf[i] = '\0'; 1145 1146 *epp = cp; 1147 return (buf); 1148 } 1149 1150 /* 1151 * Return the index of a file name. If the name cannot be found, create 1152 * a new entry and return the index of the newly created entry. 1153 */ 1154 static int 1155 getfnidx(const char *fn) 1156 { 1157 size_t i; 1158 1159 /* 0 is reserved */ 1160 for (i = 1; fnames[i] != NULL; i++) { 1161 if (strcmp(fnames[i], fn) == 0) 1162 return i; 1163 } 1164 1165 if (i == nfnames - 1) { 1166 size_t nlen = nfnames * 2; 1167 fnames = xrealloc(fnames, nlen * sizeof(char *)); 1168 (void)memset(fnames + nfnames, 0, nfnames * sizeof(char *)); 1169 flines = xrealloc(flines, nlen * sizeof(size_t)); 1170 (void)memset(flines + nfnames, 0, nfnames * sizeof(size_t)); 1171 nfnames = nlen; 1172 } 1173 1174 fnames[i] = xstrdup(fn); 1175 flines[i] = 0; 1176 return (i); 1177 } 1178 1179 /* 1180 * Separate symbols with static and external linkage. 1181 */ 1182 void 1183 mkstatic(hte_t *hte) 1184 { 1185 sym_t *sym1, **symp, *sym; 1186 fcall_t **callp, *call; 1187 usym_t **usymp, *usym; 1188 hte_t *nhte; 1189 int ofnd; 1190 1191 /* Look for first static definition */ 1192 for (sym1 = hte->h_syms; sym1 != NULL; sym1 = sym1->s_nxt) { 1193 if (sym1->s_static) 1194 break; 1195 } 1196 if (sym1 == NULL) 1197 return; 1198 1199 /* Do nothing if this name is used only in one translation unit. */ 1200 ofnd = 0; 1201 for (sym = hte->h_syms; sym != NULL && !ofnd; sym = sym->s_nxt) { 1202 if (sym->s_pos.p_src != sym1->s_pos.p_src) 1203 ofnd = 1; 1204 } 1205 for (call = hte->h_calls; call != NULL && !ofnd; call = call->f_nxt) { 1206 if (call->f_pos.p_src != sym1->s_pos.p_src) 1207 ofnd = 1; 1208 } 1209 for (usym = hte->h_usyms; usym != NULL && !ofnd; usym = usym->u_nxt) { 1210 if (usym->u_pos.p_src != sym1->s_pos.p_src) 1211 ofnd = 1; 1212 } 1213 if (!ofnd) { 1214 hte->h_used = 1; 1215 /* errors about undef. static symbols are printed in lint1 */ 1216 hte->h_def = 1; 1217 hte->h_static = 1; 1218 return; 1219 } 1220 1221 /* 1222 * Create a new hash table entry 1223 * 1224 * XXX this entry should be put at the beginning of the list to 1225 * avoid to process the same symbol twice. 1226 */ 1227 for (nhte = hte; nhte->h_link != NULL; nhte = nhte->h_link) 1228 continue; 1229 nhte->h_link = xmalloc(sizeof (hte_t)); 1230 nhte = nhte->h_link; 1231 nhte->h_name = hte->h_name; 1232 nhte->h_used = 1; 1233 nhte->h_def = 1; /* error in lint1 */ 1234 nhte->h_static = 1; 1235 nhte->h_syms = NULL; 1236 nhte->h_lsym = &nhte->h_syms; 1237 nhte->h_calls = NULL; 1238 nhte->h_lcall = &nhte->h_calls; 1239 nhte->h_usyms = NULL; 1240 nhte->h_lusym = &nhte->h_usyms; 1241 nhte->h_link = NULL; 1242 nhte->h_hte = NULL; 1243 1244 /* 1245 * move all symbols used in this translation unit into the new 1246 * hash table entry. 1247 */ 1248 for (symp = &hte->h_syms; (sym = *symp) != NULL; ) { 1249 if (sym->s_pos.p_src == sym1->s_pos.p_src) { 1250 sym->s_static = 1; 1251 (*symp) = sym->s_nxt; 1252 if (hte->h_lsym == &sym->s_nxt) 1253 hte->h_lsym = symp; 1254 sym->s_nxt = NULL; 1255 *nhte->h_lsym = sym; 1256 nhte->h_lsym = &sym->s_nxt; 1257 } else { 1258 symp = &sym->s_nxt; 1259 } 1260 } 1261 for (callp = &hte->h_calls; (call = *callp) != NULL; ) { 1262 if (call->f_pos.p_src == sym1->s_pos.p_src) { 1263 (*callp) = call->f_nxt; 1264 if (hte->h_lcall == &call->f_nxt) 1265 hte->h_lcall = callp; 1266 call->f_nxt = NULL; 1267 *nhte->h_lcall = call; 1268 nhte->h_lcall = &call->f_nxt; 1269 } else { 1270 callp = &call->f_nxt; 1271 } 1272 } 1273 for (usymp = &hte->h_usyms; (usym = *usymp) != NULL; ) { 1274 if (usym->u_pos.p_src == sym1->s_pos.p_src) { 1275 (*usymp) = usym->u_nxt; 1276 if (hte->h_lusym == &usym->u_nxt) 1277 hte->h_lusym = usymp; 1278 usym->u_nxt = NULL; 1279 *nhte->h_lusym = usym; 1280 nhte->h_lusym = &usym->u_nxt; 1281 } else { 1282 usymp = &usym->u_nxt; 1283 } 1284 } 1285 1286 /* h_def must be recalculated for old hte */ 1287 hte->h_def = nhte->h_def = 0; 1288 for (sym = hte->h_syms; sym != NULL; sym = sym->s_nxt) { 1289 if (sym->s_def == DEF || sym->s_def == TDEF) { 1290 hte->h_def = 1; 1291 break; 1292 } 1293 } 1294 1295 mkstatic(hte); 1296 } 1297