read.c revision 1.64
1 /* $NetBSD: read.c,v 1.64 2021/09/04 19:16:38 rillig 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 1.64 2021/09/04 19:16:38 rillig Exp $"); 42 #endif 43 44 #include <ctype.h> 45 #include <limits.h> 46 #include <stdarg.h> 47 #include <stdio.h> 48 #include <stdlib.h> 49 #include <string.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 indices 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 unsigned short th_idx; 82 struct thtab *th_next; 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 specified 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 unsigned short inptype(const char *, const char **); 103 static int gettlen(const char *, const char **); 104 static unsigned short findtype(const char *, size_t, int); 105 static unsigned 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 static bool 112 try_parse_int(const char **p, int *num) 113 { 114 char *end; 115 116 *num = (int)strtol(*p, &end, 10); 117 if (end == *p) 118 return false; 119 *p = end; 120 return true; 121 } 122 123 static int 124 parse_int(const char **p) 125 { 126 char *end; 127 int n; 128 129 n = (int)strtol(*p, &end, 10); 130 if (end == *p) 131 inperr("not a number: %s", *p); 132 *p = end; 133 return n; 134 } 135 136 static short 137 parse_short(const char **p) 138 { 139 return (short)parse_int(p); 140 } 141 142 void 143 readfile(const char *name) 144 { 145 FILE *inp; 146 size_t len; 147 const char *cp; 148 char *line, rt = '\0'; 149 int cline, isrc, iline; 150 pos_t pos; 151 152 if (inpfns == NULL) 153 inpfns = xcalloc(ninpfns = 128, sizeof(*inpfns)); 154 if (fnames == NULL) 155 fnames = xcalloc(nfnames = 256, sizeof(*fnames)); 156 if (flines == NULL) 157 flines = xcalloc(nfnames, sizeof(*flines)); 158 if (tlstlen == 0) 159 tlst = xcalloc(tlstlen = 256, sizeof(*tlst)); 160 if (thtab == NULL) 161 thtab = xcalloc(THSHSIZ2, sizeof(*thtab)); 162 163 _inithash(&renametab); 164 165 srcfile = getfnidx(name); 166 167 if ((inp = fopen(name, "r")) == NULL) 168 err(1, "cannot open %s", name); 169 170 while ((line = fgetln(inp, &len)) != NULL) { 171 flines[srcfile]++; 172 173 if (len == 0 || line[len - 1] != '\n') 174 inperr("%s", &line[len - 1]); 175 line[len - 1] = '\0'; 176 cp = line; 177 178 /* line number in csrcfile */ 179 if (!try_parse_int(&cp, &cline)) 180 cline = -1; 181 182 /* record type */ 183 if (*cp == '\0') 184 inperr("missing record type"); 185 rt = *cp++; 186 187 if (rt == 'S') { 188 setsrc(cp); 189 continue; 190 } else if (rt == 's') { 191 setfnid(cline, cp); 192 continue; 193 } 194 195 /* 196 * Index of (included) source file. If this index is 197 * different from csrcfile, it refers to an included 198 * file. 199 */ 200 isrc = parse_int(&cp); 201 isrc = inpfns[isrc]; 202 203 /* line number in isrc */ 204 if (*cp++ != '.') 205 inperr("bad line number"); 206 iline = parse_int(&cp); 207 208 pos.p_src = (unsigned short)csrcfile; 209 pos.p_line = (unsigned short)cline; 210 pos.p_isrc = (unsigned short)isrc; 211 pos.p_iline = (unsigned short)iline; 212 213 /* process rest of this record */ 214 switch (rt) { 215 case 'c': 216 funccall(&pos, cp); 217 break; 218 case 'd': 219 decldef(&pos, cp); 220 break; 221 case 'u': 222 usedsym(&pos, cp); 223 break; 224 default: 225 inperr("bad record type %c", rt); 226 } 227 228 } 229 230 _destroyhash(renametab); 231 232 if (ferror(inp) != 0) 233 err(1, "read error on %s", name); 234 235 (void)fclose(inp); 236 } 237 238 239 static void __attribute__((format(printf, 3, 4))) __attribute__((noreturn)) 240 inperror(const char *file, size_t line, const char *fmt, ...) 241 { 242 va_list ap; 243 char buf[1024]; 244 245 va_start(ap, fmt); 246 (void)vsnprintf(buf, sizeof(buf), fmt, ap); 247 va_end(ap); 248 249 errx(1, "%s,%zu: input file error: %s,%zu (%s)", file, line, 250 fnames[srcfile], flines[srcfile], buf); 251 } 252 253 /* 254 * Set the name of the C source file of the .ln file which is 255 * currently read. 256 */ 257 static void 258 setsrc(const char *cp) 259 { 260 261 csrcfile = getfnidx(cp); 262 } 263 264 /* 265 * setfnid() gets as input an index as used in an input file and the 266 * associated file name. If necessary, it creates a new lint2 file 267 * name index for this file name and creates the mapping of the index 268 * as used in the input file to the index used in lint2. 269 */ 270 static void 271 setfnid(int fid, const char *cp) 272 { 273 274 if (fid < 0) 275 inperr("bad fid"); 276 277 if ((size_t)fid >= ninpfns) { 278 inpfns = xrealloc(inpfns, (ninpfns * 2) * sizeof(*inpfns)); 279 (void)memset(inpfns + ninpfns, 0, ninpfns * sizeof(*inpfns)); 280 ninpfns *= 2; 281 } 282 /* 283 * Should always be true because indices written in the output 284 * file by lint1 are always the previous index + 1. 285 */ 286 if ((size_t)fid >= ninpfns) 287 errx(1, "internal error: setfnid()"); 288 inpfns[fid] = (unsigned short)getfnidx(cp); 289 } 290 291 /* 292 * Process a function call record (c-record). 293 */ 294 static void 295 funccall(pos_t *posp, const char *cp) 296 { 297 arginf_t *ai, **lai; 298 char c; 299 bool rused, rdisc; 300 hte_t *hte; 301 fcall_t *fcall; 302 const char *name; 303 304 fcall = xalloc(sizeof(*fcall)); 305 fcall->f_pos = *posp; 306 307 /* read flags */ 308 rused = rdisc = false; 309 lai = &fcall->f_args; 310 311 again: 312 c = *cp++; 313 switch (c) { 314 case 'u': 315 if (rused || rdisc) 316 inperr("used or discovered: %c", c); 317 rused = true; 318 goto again; 319 case 'i': 320 if (rused || rdisc) 321 inperr("used or discovered: %c", c); 322 goto again; 323 case 'd': 324 if (rused || rdisc) 325 inperr("used or discovered: %c", c); 326 rdisc = true; 327 goto again; 328 case 'z': 329 case 'p': 330 case 'n': 331 case 's': 332 ai = xalloc(sizeof(*ai)); 333 ai->a_num = parse_int(&cp); 334 if (c == 'z') { 335 ai->a_pcon = ai->a_zero = true; 336 } else if (c == 'p') { 337 ai->a_pcon = true; 338 } else if (c == 'n') { 339 ai->a_ncon = true; 340 } else { 341 ai->a_fmt = true; 342 ai->a_fstrg = inpqstrg(cp, &cp); 343 } 344 *lai = ai; 345 lai = &ai->a_next; 346 goto again; 347 default: 348 cp--; 349 } 350 351 fcall->f_rused = rused; 352 fcall->f_rdisc = rdisc; 353 354 /* read name of function */ 355 name = inpname(cp, &cp); 356 357 /* first look it up in the renaming table, then in the normal table */ 358 hte = _hsearch(renametab, name, false); 359 if (hte != NULL) 360 hte = hte->h_hte; 361 else 362 hte = hsearch(name, true); 363 hte->h_used = true; 364 365 fcall->f_type = inptype(cp, &cp); 366 367 *hte->h_lcall = fcall; 368 hte->h_lcall = &fcall->f_next; 369 370 if (*cp != '\0') 371 inperr("trailing line data: %s", cp); 372 } 373 374 static bool 375 parse_function_attribute(const char **pp, sym_t *sym, bool *used) 376 { 377 378 switch (*(*pp)++) { 379 case 'd': 380 if (sym->s_def != NODECL) 381 inperr("def"); 382 sym->s_def = DEF; 383 break; 384 case 'e': 385 if (sym->s_def != NODECL) 386 inperr("decl"); 387 sym->s_def = DECL; 388 break; 389 case 'i': 390 if (sym->s_inline) 391 inperr("inline"); 392 sym->s_inline = true; 393 break; 394 case 'o': 395 if (sym->s_old_style_function) 396 inperr("osdef"); 397 sym->s_old_style_function = true; 398 break; 399 case 'r': 400 if (sym->s_function_has_return_value) 401 inperr("r"); 402 sym->s_function_has_return_value = true; 403 break; 404 case 's': 405 if (sym->s_static) 406 inperr("static"); 407 sym->s_static = true; 408 break; 409 case 't': 410 if (sym->s_def != NODECL) 411 inperr("tdef"); 412 sym->s_def = TDEF; 413 break; 414 case 'u': 415 if (*used) 416 inperr("used"); 417 *used = true; 418 break; 419 case 'v': 420 if (sym->s_check_only_first_args) 421 inperr("v"); 422 sym->s_check_only_first_args = true; 423 sym->s_check_num_args = parse_short(pp); 424 break; 425 case 'P': 426 if (sym->s_printflike) 427 inperr("P"); 428 sym->s_printflike = true; 429 sym->s_printflike_arg = parse_short(pp); 430 break; 431 case 'S': 432 if (sym->s_scanflike) 433 inperr("S"); 434 sym->s_scanflike = true; 435 sym->s_scanflike_arg = parse_short(pp); 436 break; 437 default: 438 (*pp)--; 439 return false; 440 } 441 return true; 442 } 443 444 /* 445 * Process a declaration or definition (d-record). 446 */ 447 static void 448 decldef(pos_t *posp, const char *cp) 449 { 450 sym_t *symp, sym; 451 char *pos1, *tname; 452 bool used, renamed; 453 hte_t *hte, *renamehte = NULL; 454 const char *name, *newname; 455 456 (void)memset(&sym, 0, sizeof(sym)); 457 sym.s_pos = *posp; 458 sym.s_def = NODECL; 459 460 used = false; 461 462 while (parse_function_attribute(&cp, &sym, &used)) 463 continue; 464 465 /* read symbol name, doing renaming if necessary */ 466 name = inpname(cp, &cp); 467 renamed = false; 468 if (*cp == 'r') { 469 cp++; 470 tname = xstrdup(name); 471 newname = inpname(cp, &cp); 472 473 /* enter it and see if it's already been renamed */ 474 renamehte = _hsearch(renametab, tname, true); 475 if (renamehte->h_hte == NULL) { 476 hte = hsearch(newname, true); 477 renamehte->h_hte = hte; 478 renamed = true; 479 } else if (hte = renamehte->h_hte, 480 strcmp(hte->h_name, newname) != 0) { 481 pos1 = xstrdup(mkpos(&renamehte->h_syms->s_pos)); 482 /* %s renamed multiple times \t%s :: %s */ 483 msg(18, tname, pos1, mkpos(&sym.s_pos)); 484 free(pos1); 485 } 486 free(tname); 487 } else { 488 /* it might be a previously-done rename */ 489 hte = _hsearch(renametab, name, false); 490 if (hte != NULL) 491 hte = hte->h_hte; 492 else 493 hte = hsearch(name, true); 494 } 495 hte->h_used |= used; 496 if (sym.s_def == DEF || sym.s_def == TDEF) 497 hte->h_def = true; 498 499 sym.s_type = inptype(cp, &cp); 500 501 /* 502 * Allocate memory for this symbol only if it was not already 503 * declared or tentatively defined at the same location with 504 * the same type. Works only for symbols with external linkage, 505 * because static symbols, tentatively defined at the same location 506 * but in different translation units are really different symbols. 507 */ 508 for (symp = hte->h_syms; symp != NULL; symp = symp->s_next) { 509 if (symp->s_pos.p_isrc == sym.s_pos.p_isrc && 510 symp->s_pos.p_iline == sym.s_pos.p_iline && 511 symp->s_type == sym.s_type && 512 ((symp->s_def == DECL && sym.s_def == DECL) || 513 (!sflag && symp->s_def == TDEF && sym.s_def == TDEF)) && 514 !symp->s_static && !sym.s_static) { 515 break; 516 } 517 } 518 519 if (symp == NULL) { 520 if (sym.s_check_only_first_args || 521 sym.s_printflike || sym.s_scanflike) { 522 symp = xalloc(sizeof(*symp)); 523 *symp = sym; 524 } else { 525 /* no need to allocate memory for unused members */ 526 symp = xalloc(sizeof(symp->s_s)); 527 symp->s_s = sym.s_s; 528 } 529 *hte->h_lsym = symp; 530 hte->h_lsym = &symp->s_next; 531 532 /* XXX hack so we can remember where a symbol was renamed */ 533 if (renamed) 534 renamehte->h_syms = symp; 535 } 536 537 if (*cp != '\0') 538 inperr("trailing line: %s", cp); 539 } 540 541 /* 542 * Read an u-record (emitted by lint1 if a symbol was used). 543 */ 544 static void 545 usedsym(pos_t *posp, const char *cp) 546 { 547 usym_t *usym; 548 hte_t *hte; 549 const char *name; 550 551 usym = xalloc(sizeof(*usym)); 552 usym->u_pos = *posp; 553 554 /* needed as delimiter between two numbers */ 555 if (*cp++ != 'x') 556 inperr("bad delim %c", cp[-1]); 557 558 name = inpname(cp, &cp); 559 hte = _hsearch(renametab, name, false); 560 if (hte != NULL) 561 hte = hte->h_hte; 562 else 563 hte = hsearch(name, true); 564 hte->h_used = true; 565 566 *hte->h_lusym = usym; 567 hte->h_lusym = &usym->u_next; 568 } 569 570 static tspec_t 571 parse_tspec(const char **pp, char c, bool *osdef) 572 { 573 char s; 574 575 switch (c) { 576 case 's': /* 'signed' or 'struct' or 'float' */ 577 case 'u': /* 'unsigned' or 'union' */ 578 case 'l': /* 'long double' */ 579 case 'e': /* 'enum' */ 580 s = c; 581 c = *(*pp)++; 582 break; 583 default: 584 s = '\0'; 585 break; 586 } 587 588 switch (c) { 589 case 'B': 590 return BOOL; 591 case 'C': 592 return s == 's' ? SCHAR : (s == 'u' ? UCHAR : CHAR); 593 case 'S': 594 return s == 'u' ? USHORT : SHORT; 595 case 'I': 596 return s == 'u' ? UINT : INT; 597 case 'L': 598 return s == 'u' ? ULONG : LONG; 599 case 'Q': 600 return s == 'u' ? UQUAD : QUAD; 601 #ifdef INT128_SIZE 602 case 'J': 603 return s == 'u' ? UINT128 : INT128; 604 #endif 605 case 'D': 606 return s == 's' ? FLOAT : (s == 'l' ? LDOUBLE : DOUBLE); 607 case 'V': 608 return VOID; 609 case 'P': 610 return PTR; 611 case 'A': 612 return ARRAY; 613 case 'F': 614 case 'f': 615 *osdef = c == 'f'; 616 return FUNC; 617 case 'T': 618 return s == 'e' ? ENUM : (s == 's' ? STRUCT : UNION); 619 case 'X': 620 return s == 's' ? FCOMPLEX 621 : (s == 'l' ? LCOMPLEX : DCOMPLEX); 622 default: 623 inperr("tspec '%c'", c); 624 } 625 } 626 627 /* 628 * Read a type and return the index of this type. 629 */ 630 static unsigned short 631 inptype(const char *cp, const char **epp) 632 { 633 char c; 634 const char *ep; 635 type_t *tp; 636 int narg, i; 637 bool osdef = false; 638 size_t tlen; 639 unsigned short tidx; 640 int h; 641 642 /* If we have this type already, return its index. */ 643 tlen = gettlen(cp, &ep); 644 h = thash(cp, tlen); 645 if ((tidx = findtype(cp, tlen, h)) != 0) { 646 *epp = ep; 647 return tidx; 648 } 649 650 /* No, we must create a new type. */ 651 tp = xalloc(sizeof(*tp)); 652 653 tidx = storetyp(tp, cp, tlen, h); 654 655 c = *cp++; 656 657 if (c == 'c') { 658 tp->t_const = true; 659 c = *cp++; 660 } 661 if (c == 'v') { 662 tp->t_volatile = true; 663 c = *cp++; 664 } 665 666 tp->t_tspec = parse_tspec(&cp, c, &osdef); 667 668 switch (tp->t_tspec) { 669 case ARRAY: 670 tp->t_dim = parse_int(&cp); 671 tp->t_subt = TP(inptype(cp, &cp)); 672 break; 673 case PTR: 674 tp->t_subt = TP(inptype(cp, &cp)); 675 break; 676 case FUNC: 677 c = *cp; 678 if (ch_isdigit(c)) { 679 if (!osdef) 680 tp->t_proto = true; 681 narg = parse_int(&cp); 682 tp->t_args = xcalloc((size_t)narg + 1, 683 sizeof(*tp->t_args)); 684 for (i = 0; i < narg; i++) { 685 if (i == narg - 1 && *cp == 'E') { 686 tp->t_vararg = true; 687 cp++; 688 } else { 689 tp->t_args[i] = TP(inptype(cp, &cp)); 690 } 691 } 692 } 693 tp->t_subt = TP(inptype(cp, &cp)); 694 break; 695 case ENUM: 696 tp->t_tspec = INT; 697 tp->t_is_enum = true; 698 /* FALLTHROUGH */ 699 case STRUCT: 700 case UNION: 701 switch (*cp++) { 702 case '1': 703 tp->t_istag = true; 704 tp->t_tag = hsearch(inpname(cp, &cp), true); 705 break; 706 case '2': 707 tp->t_istynam = true; 708 tp->t_tynam = hsearch(inpname(cp, &cp), true); 709 break; 710 case '3': 711 tp->t_isuniqpos = true; 712 tp->t_uniqpos.p_line = parse_int(&cp); 713 cp++; 714 /* xlate to 'global' file name. */ 715 tp->t_uniqpos.p_file = 716 addoutfile(inpfns[parse_int(&cp)]); 717 cp++; 718 tp->t_uniqpos.p_uniq = parse_int(&cp); 719 break; 720 } 721 break; 722 default: 723 break; 724 } 725 726 *epp = cp; 727 return tidx; 728 } 729 730 /* 731 * Get the length of a type string. 732 */ 733 static int 734 gettlen(const char *cp, const char **epp) 735 { 736 const char *cp1; 737 char c, s; 738 tspec_t t; 739 int narg, i; 740 741 cp1 = cp; 742 743 c = *cp++; 744 745 if (c == 'c') 746 c = *cp++; 747 if (c == 'v') 748 c = *cp++; 749 750 switch (c) { 751 case 's': 752 case 'u': 753 case 'l': 754 case 'e': 755 s = c; 756 c = *cp++; 757 break; 758 default: 759 s = '\0'; 760 break; 761 } 762 763 t = NOTSPEC; 764 765 switch (c) { 766 case 'B': 767 if (s == '\0') 768 t = BOOL; 769 break; 770 case 'C': 771 if (s == 's') { 772 t = SCHAR; 773 } else if (s == 'u') { 774 t = UCHAR; 775 } else if (s == '\0') { 776 t = CHAR; 777 } 778 break; 779 case 'S': 780 if (s == 'u') { 781 t = USHORT; 782 } else if (s == '\0') { 783 t = SHORT; 784 } 785 break; 786 case 'I': 787 if (s == 'u') { 788 t = UINT; 789 } else if (s == '\0') { 790 t = INT; 791 } 792 break; 793 case 'L': 794 if (s == 'u') { 795 t = ULONG; 796 } else if (s == '\0') { 797 t = LONG; 798 } 799 break; 800 case 'Q': 801 if (s == 'u') { 802 t = UQUAD; 803 } else if (s == '\0') { 804 t = QUAD; 805 } 806 break; 807 #ifdef INT128_SIZE 808 case 'J': 809 if (s == 'u') { 810 t = UINT128; 811 } else if (s == '\0') { 812 t = INT128; 813 } 814 break; 815 #endif 816 case 'D': 817 if (s == 's') { 818 t = FLOAT; 819 } else if (s == 'l') { 820 t = LDOUBLE; 821 } else if (s == '\0') { 822 t = DOUBLE; 823 } 824 break; 825 case 'V': 826 if (s == '\0') 827 t = VOID; 828 break; 829 case 'P': 830 if (s == '\0') 831 t = PTR; 832 break; 833 case 'A': 834 if (s == '\0') 835 t = ARRAY; 836 break; 837 case 'F': 838 case 'f': 839 if (s == '\0') 840 t = FUNC; 841 break; 842 case 'T': 843 if (s == 'e') { 844 t = ENUM; 845 } else if (s == 's') { 846 t = STRUCT; 847 } else if (s == 'u') { 848 t = UNION; 849 } 850 break; 851 case 'X': 852 if (s == 's') { 853 t = FCOMPLEX; 854 } else if (s == 'l') { 855 t = LCOMPLEX; 856 } else if (s == '\0') { 857 t = DCOMPLEX; 858 } 859 break; 860 default: 861 break; 862 } 863 864 if (t == NOTSPEC) 865 inperr("bad type: %c %c", c, s); 866 867 switch (t) { 868 case ARRAY: 869 (void)parse_int(&cp); 870 (void)gettlen(cp, &cp); 871 break; 872 case PTR: 873 (void)gettlen(cp, &cp); 874 break; 875 case FUNC: 876 c = *cp; 877 if (ch_isdigit(c)) { 878 narg = parse_int(&cp); 879 for (i = 0; i < narg; i++) { 880 if (i == narg - 1 && *cp == 'E') { 881 cp++; 882 } else { 883 (void)gettlen(cp, &cp); 884 } 885 } 886 } 887 (void)gettlen(cp, &cp); 888 break; 889 case ENUM: 890 case STRUCT: 891 case UNION: 892 switch (*cp++) { 893 case '1': 894 case '2': 895 (void)inpname(cp, &cp); 896 break; 897 case '3': 898 /* unique position: line.file.uniquifier */ 899 (void)parse_int(&cp); 900 if (*cp++ != '.') 901 inperr("not dot: %c", cp[-1]); 902 (void)parse_int(&cp); 903 if (*cp++ != '.') 904 inperr("not dot: %c", cp[-1]); 905 (void)parse_int(&cp); 906 break; 907 default: 908 inperr("bad value: %c", cp[-1]); 909 } 910 break; 911 default: 912 break; 913 } 914 915 *epp = cp; 916 return cp - cp1; 917 } 918 919 /* 920 * Search a type by its type string. 921 */ 922 static unsigned short 923 findtype(const char *cp, size_t len, int h) 924 { 925 thtab_t *thte; 926 927 for (thte = thtab[h]; thte != NULL; thte = thte->th_next) { 928 if (strncmp(thte->th_name, cp, len) != 0) 929 continue; 930 if (thte->th_name[len] == '\0') 931 return thte->th_idx; 932 } 933 934 return 0; 935 } 936 937 /* 938 * Store a type and its type string, so we can later share this type 939 * if we read the same type string from the input file. 940 */ 941 static unsigned short 942 storetyp(type_t *tp, const char *cp, size_t len, int h) 943 { 944 static unsigned int tidx = 1; /* 0 is reserved */ 945 thtab_t *thte; 946 char *name; 947 948 if (tidx >= USHRT_MAX) 949 errx(1, "sorry, too many types"); 950 951 if (tidx == tlstlen - 1) { 952 tlst = xrealloc(tlst, (tlstlen * 2) * sizeof(*tlst)); 953 (void)memset(tlst + tlstlen, 0, tlstlen * sizeof(*tlst)); 954 tlstlen *= 2; 955 } 956 957 tlst[tidx] = tp; 958 959 /* create a hash table entry */ 960 name = xalloc(len + 1); 961 (void)memcpy(name, cp, len); 962 name[len] = '\0'; 963 964 thte = xalloc(sizeof(*thte)); 965 thte->th_name = name; 966 thte->th_idx = tidx; 967 thte->th_next = thtab[h]; 968 thtab[h] = thte; 969 970 return (unsigned short)tidx++; 971 } 972 973 /* 974 * Hash function for types 975 */ 976 static int 977 thash(const char *s, size_t len) 978 { 979 unsigned int v; 980 981 v = 0; 982 while (len-- != 0) { 983 v = (v << sizeof(v)) + (unsigned char)*s++; 984 v ^= v >> (sizeof(v) * CHAR_BIT - sizeof(v)); 985 } 986 return v % THSHSIZ2; 987 } 988 989 /* 990 * Read a string enclosed by "". This string may contain quoted chars. 991 */ 992 static char * 993 inpqstrg(const char *src, const char **epp) 994 { 995 char *strg, *dst; 996 size_t slen; 997 char c; 998 int v; 999 1000 dst = strg = xmalloc(slen = 32); 1001 1002 if ((c = *src++) != '"') 1003 inperr("not quote: %c", c); 1004 if ((c = *src++) == '\0') 1005 inperr("trailing data: %c", c); 1006 1007 while (c != '"') { 1008 if (c == '\\') { 1009 if ((c = *src++) == '\0') 1010 inperr("missing after \\"); 1011 switch (c) { 1012 case 'n': 1013 c = '\n'; 1014 break; 1015 case 't': 1016 c = '\t'; 1017 break; 1018 case 'v': 1019 c = '\v'; 1020 break; 1021 case 'b': 1022 c = '\b'; 1023 break; 1024 case 'r': 1025 c = '\r'; 1026 break; 1027 case 'f': 1028 c = '\f'; 1029 break; 1030 case 'a': 1031 c = '\a'; 1032 break; 1033 case '\\': 1034 c = '\\'; 1035 break; 1036 case '"': 1037 c = '"'; 1038 break; 1039 case '\'': 1040 c = '\''; 1041 break; 1042 case '0': case '1': case '2': case '3': 1043 v = (c - '0') << 6; 1044 if ((c = *src++) < '0' || c > '7') 1045 inperr("not octal: %c", c); 1046 v |= (c - '0') << 3; 1047 if ((c = *src++) < '0' || c > '7') 1048 inperr("not octal: %c", c); 1049 v |= c - '0'; 1050 c = (char)v; 1051 break; 1052 default: 1053 inperr("bad \\ escape: %c", c); 1054 } 1055 } 1056 /* keep space for trailing '\0' */ 1057 if ((size_t)(dst - strg) == slen - 1) { 1058 strg = xrealloc(strg, slen * 2); 1059 dst = strg + (slen - 1); 1060 slen *= 2; 1061 } 1062 *dst++ = c; 1063 if ((c = *src++) == '\0') 1064 inperr("missing closing quote"); 1065 } 1066 *dst = '\0'; 1067 1068 *epp = src; 1069 return strg; 1070 } 1071 1072 /* 1073 * Read the name of a symbol in static memory. 1074 */ 1075 static const char * 1076 inpname(const char *cp, const char **epp) 1077 { 1078 static char *buf; 1079 static size_t blen = 0; 1080 size_t len, i; 1081 char c; 1082 1083 len = parse_int(&cp); 1084 if (len + 1 > blen) 1085 buf = xrealloc(buf, blen = len + 1); 1086 for (i = 0; i < len; i++) { 1087 c = *cp++; 1088 if (!ch_isalnum(c) && c != '_') 1089 inperr("not alnum or _: %c", c); 1090 buf[i] = c; 1091 } 1092 buf[i] = '\0'; 1093 1094 *epp = cp; 1095 return buf; 1096 } 1097 1098 /* 1099 * Return the index of a file name. If the name cannot be found, create 1100 * a new entry and return the index of the newly created entry. 1101 */ 1102 static int 1103 getfnidx(const char *fn) 1104 { 1105 size_t i; 1106 1107 /* 0 is reserved */ 1108 for (i = 1; fnames[i] != NULL; i++) { 1109 if (strcmp(fnames[i], fn) == 0) 1110 return i; 1111 } 1112 1113 if (i == nfnames - 1) { 1114 size_t nlen = nfnames * 2; 1115 fnames = xrealloc(fnames, nlen * sizeof(*fnames)); 1116 (void)memset(fnames + nfnames, 0, nfnames * sizeof(*fnames)); 1117 flines = xrealloc(flines, nlen * sizeof(*flines)); 1118 (void)memset(flines + nfnames, 0, nfnames * sizeof(*flines)); 1119 nfnames = nlen; 1120 } 1121 1122 fnames[i] = xstrdup(fn); 1123 flines[i] = 0; 1124 return i; 1125 } 1126 1127 /* 1128 * Separate symbols with static and external linkage. 1129 */ 1130 void 1131 mkstatic(hte_t *hte) 1132 { 1133 sym_t *sym1, **symp, *sym; 1134 fcall_t **callp, *call; 1135 usym_t **usymp, *usym; 1136 hte_t *nhte; 1137 bool ofnd; 1138 1139 /* Look for first static definition */ 1140 for (sym1 = hte->h_syms; sym1 != NULL; sym1 = sym1->s_next) { 1141 if (sym1->s_static) 1142 break; 1143 } 1144 if (sym1 == NULL) 1145 return; 1146 1147 /* Do nothing if this name is used only in one translation unit. */ 1148 ofnd = false; 1149 for (sym = hte->h_syms; sym != NULL && !ofnd; sym = sym->s_next) { 1150 if (sym->s_pos.p_src != sym1->s_pos.p_src) 1151 ofnd = true; 1152 } 1153 for (call = hte->h_calls; call != NULL && !ofnd; call = call->f_next) { 1154 if (call->f_pos.p_src != sym1->s_pos.p_src) 1155 ofnd = true; 1156 } 1157 for (usym = hte->h_usyms; usym != NULL && !ofnd; usym = usym->u_next) { 1158 if (usym->u_pos.p_src != sym1->s_pos.p_src) 1159 ofnd = true; 1160 } 1161 if (!ofnd) { 1162 hte->h_used = true; 1163 /* errors about undef. static symbols are printed in lint1 */ 1164 hte->h_def = true; 1165 hte->h_static = true; 1166 return; 1167 } 1168 1169 /* 1170 * Create a new hash table entry 1171 * 1172 * XXX this entry should be put at the beginning of the list to 1173 * avoid processing the same symbol twice. 1174 */ 1175 for (nhte = hte; nhte->h_link != NULL; nhte = nhte->h_link) 1176 continue; 1177 nhte->h_link = xmalloc(sizeof(*nhte->h_link)); 1178 nhte = nhte->h_link; 1179 nhte->h_name = hte->h_name; 1180 nhte->h_used = true; 1181 nhte->h_def = true; /* error in lint1 */ 1182 nhte->h_static = true; 1183 nhte->h_syms = NULL; 1184 nhte->h_lsym = &nhte->h_syms; 1185 nhte->h_calls = NULL; 1186 nhte->h_lcall = &nhte->h_calls; 1187 nhte->h_usyms = NULL; 1188 nhte->h_lusym = &nhte->h_usyms; 1189 nhte->h_link = NULL; 1190 nhte->h_hte = NULL; 1191 1192 /* 1193 * move all symbols used in this translation unit into the new 1194 * hash table entry. 1195 */ 1196 for (symp = &hte->h_syms; (sym = *symp) != NULL; ) { 1197 if (sym->s_pos.p_src == sym1->s_pos.p_src) { 1198 sym->s_static = true; 1199 (*symp) = sym->s_next; 1200 if (hte->h_lsym == &sym->s_next) 1201 hte->h_lsym = symp; 1202 sym->s_next = NULL; 1203 *nhte->h_lsym = sym; 1204 nhte->h_lsym = &sym->s_next; 1205 } else { 1206 symp = &sym->s_next; 1207 } 1208 } 1209 for (callp = &hte->h_calls; (call = *callp) != NULL; ) { 1210 if (call->f_pos.p_src == sym1->s_pos.p_src) { 1211 (*callp) = call->f_next; 1212 if (hte->h_lcall == &call->f_next) 1213 hte->h_lcall = callp; 1214 call->f_next = NULL; 1215 *nhte->h_lcall = call; 1216 nhte->h_lcall = &call->f_next; 1217 } else { 1218 callp = &call->f_next; 1219 } 1220 } 1221 for (usymp = &hte->h_usyms; (usym = *usymp) != NULL; ) { 1222 if (usym->u_pos.p_src == sym1->s_pos.p_src) { 1223 (*usymp) = usym->u_next; 1224 if (hte->h_lusym == &usym->u_next) 1225 hte->h_lusym = usymp; 1226 usym->u_next = NULL; 1227 *nhte->h_lusym = usym; 1228 nhte->h_lusym = &usym->u_next; 1229 } else { 1230 usymp = &usym->u_next; 1231 } 1232 } 1233 1234 /* h_def must be recalculated for old hte */ 1235 hte->h_def = nhte->h_def = false; 1236 for (sym = hte->h_syms; sym != NULL; sym = sym->s_next) { 1237 if (sym->s_def == DEF || sym->s_def == TDEF) { 1238 hte->h_def = true; 1239 break; 1240 } 1241 } 1242 1243 mkstatic(hte); 1244 } 1245
Indexes created Fri Sep 26 20:09:58 GMT 2025