lib.c

   1 /*
   2  * 'sparse' library helper routines.
   3  *
   4  * Copyright (C) 2003 Transmeta Corp.
   5  *               2003-2004 Linus Torvalds
   6  *
   7  *  Licensed under the Open Software License version 1.1
   8  */
   9 #include <ctype.h>
  10 #include <fcntl.h>
  11 #include <stdarg.h>
  12 #include <stddef.h>
  13 #include <stdio.h>
  14 #include <stdlib.h>
  15 #include <string.h>
  16 #include <unistd.h>
  17 #include <assert.h>
  18
  19 #include <sys/types.h>
  20
  21 #include "lib.h"
  22 #include "allocate.h"
  23 #include "token.h"
  24 #include "parse.h"
  25 #include "symbol.h"
  26 #include "expression.h"
  27 #include "scope.h"
  28 #include "linearize.h"
  29 #include "target.h"
  30
  31 int verbose, optimize, optimize_size, preprocessing;
  32 int die_if_error = 0;
  33
  34 #ifndef __GNUC__
  35 # define __GNUC__ 2
  36 # define __GNUC_MINOR__ 95
  37 # define __GNUC_PATCHLEVEL__ 0
  38 #endif
  39
  40 int gcc_major = __GNUC__;
  41 int gcc_minor = __GNUC_MINOR__;
  42 int gcc_patchlevel = __GNUC_PATCHLEVEL__;
  43
  44 struct token *skip_to(struct token *token, int op)
  45 {
  46         while (!match_op(token, op) && !eof_token(token))
  47                 token = token->next;
  48         return token;
  49 }
  50
  51 struct token *expect(struct token *token, int op, const char *where)
  52 {
  53         if (!match_op(token, op)) {
  54                 static struct token bad_token;
  55                 if (token != &bad_token) {
  56                         bad_token.next = token;
  57                         sparse_error(token->pos, "Expected %s %s", show_special(op), where);
  58                         sparse_error(token->pos, "got %s", show_token(token));
  59                 }
  60                 if (op == ';')
  61                         return skip_to(token, op);
  62                 return &bad_token;
  63         }
  64         return token->next;
  65 }
  66
  67 unsigned int hexval(unsigned int c)
  68 {
  69         int retval = 256;
  70         switch (c) {
  71         case '0'...'9':
  72                 retval = c - '0';
  73                 break;
  74         case 'a'...'f':
  75                 retval = c - 'a' + 10;
  76                 break;
  77         case 'A'...'F':
  78                 retval = c - 'A' + 10;
  79                 break;
  80         }
  81         return retval;
  82 }
  83
  84 static void do_warn(const char *type, struct position pos, const char * fmt, va_list args)
  85 {
  86         static char buffer[512];
  87         const char *name;
  88
  89         vsprintf(buffer, fmt, args);
  90         name = stream_name(pos.stream);
  91
  92         fprintf(stderr, "%s:%d:%d: %s%s\n",
  93                 name, pos.line, pos.pos, type, buffer);
  94 }
  95
  96 static int max_warnings = 100;
  97 static int show_info = 1;
  98
  99 void info(struct position pos, const char * fmt, ...)
 100 {
 101         va_list args;
 102
 103         if (!show_info)
 104                 return;
 105         va_start(args, fmt);
 106         do_warn("", pos, fmt, args);
 107         va_end(args);
 108 }
 109
 110 void warning(struct position pos, const char * fmt, ...)
 111 {
 112         va_list args;
 113
 114         if (!max_warnings) {
 115                 show_info = 0;
 116                 return;
 117         }
 118
 119         if (!--max_warnings) {
 120                 show_info = 0;
 121                 fmt = "too many warnings";
 122         }
 123
 124         va_start(args, fmt);
 125         do_warn("warning: ", pos, fmt, args);
 126         va_end(args);
 127 }
 128
 129 void sparse_error(struct position pos, const char * fmt, ...)
 130 {
 131         static int errors = 0;
 132         va_list args;
 133         die_if_error = 1;
 134         show_info = 1;
 135         /* Shut up warnings after an error */
 136         max_warnings = 0;
 137         if (errors > 100) {
 138                 static int once = 0;
 139                 show_info = 0;
 140                 if (once)
 141                         return;
 142                 fmt = "too many errors";
 143                 once = 1;
 144         }
 145
 146         va_start(args, fmt);
 147         do_warn("error: ", pos, fmt, args);
 148         va_end(args);
 149         errors++;
 150 }
 151
 152 void error_die(struct position pos, const char * fmt, ...)
 153 {
 154         va_list args;
 155         va_start(args, fmt);
 156         do_warn("error: ", pos, fmt, args);
 157         va_end(args);
 158         exit(1);
 159 }
 160
 161 void die(const char *fmt, ...)
 162 {
 163         va_list args;
 164         static char buffer[512];
 165
 166         va_start(args, fmt);
 167         vsnprintf(buffer, sizeof(buffer), fmt, args);
 168         va_end(args);
 169
 170         fprintf(stderr, "%s\n", buffer);
 171         exit(1);
 172 }
 173
 174 static unsigned int pre_buffer_size;
 175 static char pre_buffer[8192];
 176
 177 int Wdefault_bitfield_sign = 0;
 178 int Wone_bit_signed_bitfield = 1;
 179 int Wcast_truncate = 1;
 180 int Wbitwise = 0;
 181 int Wtypesign = 0;
 182 int Wcontext = 0;
 183 int Wundefined_preprocessor = 0;
 184 int Wptr_subtraction_blows = 0;
 185 int Wcast_to_address_space = 0;
 186 int Wdecl = 0;
 187 int Wtransparent_union = 1;
 188 int Wshadow = 0;
 189 int Waddress_space = 1;
 190 int Wenum_mismatch = 1;
 191 int preprocess_only;
 192 char *include;
 193 int include_fd = -1;
 194
 195
 196 void add_pre_buffer(const char *fmt, ...)
 197 {
 198         va_list args;
 199         unsigned int size;
 200
 201         va_start(args, fmt);
 202         size = pre_buffer_size;
 203         size += vsnprintf(pre_buffer + size,
 204                 sizeof(pre_buffer) - size,
 205                 fmt, args);
 206         pre_buffer_size = size;
 207         va_end(args);
 208 }
 209
 210 static char **handle_switch_D(char *arg, char **next)
 211 {
 212         const char *name = arg + 1;
 213         const char *value = "1";
 214         for (;;) {
 215                 char c;
 216                 c = *++arg;
 217                 if (!c)
 218                         break;
 219                 if (isspace((unsigned char)c) || c == '=') {
 220                         *arg = '\0';
 221                         value = arg + 1;
 222                         break;
 223                 }
 224         }
 225         add_pre_buffer("#define %s %s\n", name, value);
 226         return next;
 227 }
 228
 229 static char **handle_switch_E(char *arg, char **next)
 230 {
 231         preprocess_only = 1;
 232         return next;
 233 }
 234
 235 static char **handle_switch_v(char *arg, char **next)
 236 {
 237         do {
 238                 verbose++;
 239         } while (*++arg == 'v');
 240         return next;
 241 }
 242
 243 static char **handle_switch_I(char *arg, char **next)
 244 {
 245         char *path = arg+1;
 246
 247         switch (arg[1]) {
 248         case '-':
 249                 add_pre_buffer("#split_include\n");
 250                 break;
 251
 252         case '\0':      /* Plain "-I" */
 253                 path = *++next;
 254                 if (!path)
 255                         die("missing argument for -I option");
 256                 /* Fallthrough */
 257         default:
 258                 add_pre_buffer("#add_include \"%s/\"\n", path);
 259         }
 260         return next;
 261 }
 262
 263 static char **handle_switch_i(char *arg, char **next)
 264 {
 265         if (*next && !strcmp(arg, "include")) {
 266                 char *name = *++next;
 267                 int fd = open(name, O_RDONLY);
 268
 269                 include_fd = fd;
 270                 include = name;
 271                 if (fd < 0)
 272                         perror(name);
 273         }
 274         if (*next && !strcmp(arg, "imacros")) {
 275                 char *name = *++next;
 276                 int fd = open(name, O_RDONLY);
 277
 278                 include_fd = fd;
 279                 include = name;
 280                 if (fd < 0)
 281                         perror(name);
 282         }
 283         else if (*next && !strcmp(arg, "isystem")) {
 284                 char *path = *++next;
 285                 if (!path)
 286                         die("missing argument for -isystem option");
 287                 add_pre_buffer("#add_isystem \"%s/\"\n", path);
 288         }
 289         return next;
 290 }
 291
 292 static char **handle_switch_M(char *arg, char **next)
 293 {
 294         if (!strcmp(arg, "MF") || !strcmp(arg,"MQ") || !strcmp(arg,"MT")) {
 295                 if (!*next)
 296                         die("missing argument for -%s option", arg);
 297                 return next + 1;
 298         }
 299         return next;
 300 }
 301
 302 static char **handle_switch_m(char *arg, char **next)
 303 {
 304         if (!strcmp(arg, "m64")) {
 305                 bits_in_long = 64;
 306                 max_int_alignment = 8;
 307                 bits_in_pointer = 64;
 308                 pointer_alignment = 8;
 309         }
 310         return next;
 311 }
 312
 313 static char **handle_switch_o(char *arg, char **next)
 314 {
 315         if (!strcmp (arg, "o") && *next)
 316                 return next + 1; // "-o foo"
 317         else
 318                 return next;     // "-ofoo" or (bogus) terminal "-o"
 319 }
 320
 321 static const struct warning {
 322         const char *name;
 323         int *flag;
 324 } warnings[] = {
 325         { "cast-to-as", &Wcast_to_address_space },
 326         { "decl", &Wdecl },
 327         { "one-bit-signed-bitfield", &Wone_bit_signed_bitfield },
 328         { "cast-truncate", &Wcast_truncate },
 329         { "ptr-subtraction-blows", &Wptr_subtraction_blows },
 330         { "default-bitfield-sign", &Wdefault_bitfield_sign },
 331         { "undef", &Wundefined_preprocessor },
 332         { "bitwise", &Wbitwise },
 333         { "typesign", &Wtypesign },
 334         { "context", &Wcontext },
 335         { "transparent-union", &Wtransparent_union },
 336         { "shadow", &Wshadow },
 337         { "address-space", &Waddress_space },
 338         { "enum-mismatch", &Wenum_mismatch },
 339 };
 340
 341
 342 static char **handle_switch_W(char *arg, char **next)
 343 {
 344         int no = 0;
 345         char *p = arg + 1;
 346         unsigned i;
 347
 348         // Prefixes "no" and "no-" mean to turn warning off.
 349         if (p[0] == 'n' && p[1] == 'o') {
 350                 p += 2;
 351                 if (p[0] == '-')
 352                         p++;
 353                 no = 1;
 354         }
 355
 356         for (i = 0; i < sizeof(warnings) / sizeof(warnings[0]); i++) {
 357                 if (!strcmp(p,warnings[i].name)) {
 358                         *warnings[i].flag = !no;
 359                         return next;
 360                 }
 361         }
 362
 363         // Unknown.
 364         return next;
 365 }
 366
 367 static char **handle_switch_U(char *arg, char **next)
 368 {
 369         const char *name = arg + 1;
 370         add_pre_buffer ("#undef %s\n", name);
 371         return next;
 372 }
 373
 374 static char **handle_switch_O(char *arg, char **next)
 375 {
 376         int level = 1;
 377         if (arg[1] >= '0' && arg[1] <= '9')
 378                 level = arg[1] - '0';
 379         optimize = level;
 380         optimize_size = arg[1] == 's';
 381         return next;
 382 }
 383
 384 static char **handle_switch_f(char *arg, char **next)
 385 {
 386         int flag = 1;
 387
 388         arg++;
 389         if (!strncmp(arg, "no-", 3)) {
 390                 flag = 0;
 391                 arg += 3;
 392         }
 393         /* handle switch here.. */
 394         return next;
 395 }
 396
 397 static char **handle_switch_G(char *arg, char **next)
 398 {
 399         if (!strcmp (arg, "G") && *next)
 400                 return next + 1; // "-G 0"
 401         else
 402                 return next;     // "-G0" or (bogus) terminal "-G"
 403 }
 404
 405 static char **handle_nostdinc(char *arg, char **next)
 406 {
 407         add_pre_buffer("#nostdinc\n");
 408         return next;
 409 }
 410
 411 static char **handle_dirafter(char *arg, char **next)
 412 {
 413         char *path = *++next;
 414         if (!path)
 415                 die("missing argument for -dirafter option");
 416         add_pre_buffer("#add_dirafter \"%s/\"\n", path);
 417         return next;
 418 }
 419
 420 struct switches {
 421         const char *name;
 422         char **(*fn)(char *, char**);
 423 };
 424
 425 char **handle_switch(char *arg, char **next)
 426 {
 427         static struct switches cmd[] = {
 428                 { "nostdinc", handle_nostdinc },
 429                 { "dirafter", handle_dirafter },
 430                 { NULL, NULL }
 431         };
 432         struct switches *s;
 433
 434         switch (*arg) {
 435         case 'D': return handle_switch_D(arg, next);
 436         case 'E': return handle_switch_E(arg, next);
 437         case 'I': return handle_switch_I(arg, next);
 438         case 'i': return handle_switch_i(arg, next);
 439         case 'M': return handle_switch_M(arg, next);
 440         case 'm': return handle_switch_m(arg, next);
 441         case 'o': return handle_switch_o(arg, next);
 442         case 'U': return handle_switch_U(arg, next);
 443         case 'v': return handle_switch_v(arg, next);
 444         case 'W': return handle_switch_W(arg, next);
 445         case 'O': return handle_switch_O(arg, next);
 446         case 'f': return handle_switch_f(arg, next);
 447         case 'G': return handle_switch_G(arg, next);
 448         default:
 449                 break;
 450         }
 451
 452         s = cmd;
 453         while (s->name) {
 454                 if (!strcmp(s->name, arg))
 455                         return s->fn(arg, next);
 456                 s++;
 457         }
 458
 459         /*
 460          * Ignore unknown command line options:
 461          * they're probably gcc switches
 462          */
 463         return next;
 464 }
 465
 466 void declare_builtin_functions(void)
 467 {
 468         /* Gaah. gcc knows tons of builtin <string.h> functions */
 469         add_pre_buffer("extern void *__builtin_memcpy(void *, const void *, __SIZE_TYPE__);\n");
 470         add_pre_buffer("extern void *__builtin_memset(void *, int, __SIZE_TYPE__);\n");
 471         add_pre_buffer("extern int __builtin_memcmp(const void *, const void *, __SIZE_TYPE__);\n");
 472         add_pre_buffer("extern int __builtin_strcmp(const char *, const char *);\n");
 473         add_pre_buffer("extern char *__builtin_strchr(const char *, int);\n");
 474         add_pre_buffer("extern char *__builtin_strcpy(char *, const char *);\n");
 475         add_pre_buffer("extern char *__builtin_strncpy(char *, const char *, __SIZE_TYPE__);\n");
 476         add_pre_buffer("extern __SIZE_TYPE__ __builtin_strspn(const char *, const char *);\n");
 477         add_pre_buffer("extern __SIZE_TYPE__ __builtin_strcspn(const char *, const char *);\n");
 478
 479         /* And some random ones.. */
 480         add_pre_buffer("extern void *__builtin_return_address(unsigned int);\n");
 481         add_pre_buffer("extern void *__builtin_extract_return_addr(void *);\n");
 482         add_pre_buffer("extern void *__builtin_frame_address(unsigned int);\n");
 483         add_pre_buffer("extern void __builtin_trap(void);\n");
 484         add_pre_buffer("extern int __builtin_ffs(int);\n");
 485         add_pre_buffer("extern void *__builtin_alloca(__SIZE_TYPE__);\n");
 486 }
 487
 488 void create_builtin_stream(void)
 489 {
 490         add_pre_buffer("#weak_define __GNUC__ %d\n", gcc_major);
 491         add_pre_buffer("#weak_define __GNUC_MINOR__ %d\n", gcc_minor);
 492         add_pre_buffer("#weak_define __GNUC_PATCHLEVEL__ %d\n", gcc_patchlevel);
 493         add_pre_buffer("#define __extension__\n");
 494         add_pre_buffer("#define __pragma__\n");
 495
 496         // gcc defines __SIZE_TYPE__ to be size_t.  For linux/i86 and
 497         // solaris/sparc that is really "unsigned int" and for linux/x86_64
 498         // it is "long unsigned int".  In either case we can probably
 499         // get away with this.  We need the #ifndef as cgcc will define
 500         // the right __SIZE_TYPE__.
 501         add_pre_buffer("#weak_define __SIZE_TYPE__ long unsigned int\n");
 502         add_pre_buffer("#weak_define __STDC__ 1\n");
 503
 504         add_pre_buffer("#define __builtin_stdarg_start(a,b) ((a) = (__builtin_va_list)(&(b)))\n");
 505         add_pre_buffer("#define __builtin_va_start(a,b) ((a) = (__builtin_va_list)(&(b)))\n");
 506         add_pre_buffer("#define __builtin_va_arg(arg,type)  ({ type __va_arg_ret = *(type *)(arg); arg += sizeof(type); __va_arg_ret; })\n");
 507         add_pre_buffer("#define __builtin_va_alist (*(void *)0)\n");
 508         add_pre_buffer("#define __builtin_va_arg_incr(x) ((x) + 1)\n");
 509         add_pre_buffer("#define __builtin_va_copy(dest, src) ({ dest = src; (void)0; })\n");
 510         add_pre_buffer("#define __builtin_va_end(arg)\n");
 511         add_pre_buffer("#define __builtin_offsetof(type, name) ((__SIZE_TYPE__)&((type *)(0ul))->name)\n");
 512
 513         /* FIXME! We need to do these as special magic macros at expansion time! */
 514         add_pre_buffer("#define __BASE_FILE__ \"base_file.c\"\n");
 515         add_pre_buffer("#define __DATE__ \"??? ?? ????\"\n");
 516         add_pre_buffer("#define __TIME__ \"??:??:??\"\n");
 517
 518         if (optimize)
 519                 add_pre_buffer("#define __OPTIMIZE__ 1\n");
 520         if (optimize_size)
 521                 add_pre_buffer("#define __OPTIMIZE_SIZE__ 1\n");
 522 }
 523
 524 static struct symbol_list *sparse_tokenstream(struct token *token)
 525 {
 526         // Pre-process the stream
 527         token = preprocess(token);
 528
 529         if (preprocess_only) {
 530                 while (!eof_token(token)) {
 531                         int prec = 1;
 532                         struct token *next = token->next;
 533                         const char *separator = "";
 534                         if (next->pos.whitespace)
 535                                 separator = " ";
 536                         if (next->pos.newline) {
 537                                 separator = "\n\t\t\t\t\t";
 538                                 prec = next->pos.pos;
 539                                 if (prec > 4)
 540                                         prec = 4;
 541                         }
 542                         printf("%s%.*s", show_token(token), prec, separator);
 543                         token = next;
 544                 }
 545                 putchar('\n');
 546
 547                 return NULL;
 548         }
 549
 550         // Parse the resulting C code
 551         while (!eof_token(token))
 552                 token = external_declaration(token, &translation_unit_used_list);
 553         return translation_unit_used_list;
 554 }
 555
 556 static struct symbol_list *sparse_file(const char *filename)
 557 {
 558         int fd;
 559         struct token *token;
 560
 561         if (strcmp (filename, "-") == 0) {
 562                 fd = 0;
 563         } else {
 564                 fd = open(filename, O_RDONLY);
 565                 if (fd < 0)
 566                         die("No such file: %s", filename);
 567         }
 568
 569         // Tokenize the input stream
 570         token = tokenize(filename, fd, NULL, includepath);
 571         close(fd);
 572
 573         return sparse_tokenstream(token);
 574 }
 575
 576 /*
 577  * This handles the "-include" directive etc: we're in global
 578  * scope, and all types/macros etc will affect all the following
 579  * files.
 580  *
 581  * NOTE NOTE NOTE! "#undef" of anything in this stage will
 582  * affect all subsequent files too, ie we can have non-local
 583  * behaviour between files!
 584  */
 585 static struct symbol_list *sparse_initial(void)
 586 {
 587         struct token *token;
 588
 589         // Prepend any "include" file to the stream.
 590         // We're in global scope, it will affect all files!
 591         token = NULL;
 592         if (include_fd >= 0)
 593                 token = tokenize(include, include_fd, NULL, includepath);
 594
 595         // Prepend the initial built-in stream
 596         token = tokenize_buffer(pre_buffer, pre_buffer_size, token);
 597         return sparse_tokenstream(token);
 598 }
 599
 600 struct symbol_list *sparse_initialize(int argc, char **argv)
 601 {
 602         char **args;
 603         int files = 0;
 604         struct symbol_list *list;
 605
 606         // Initialize symbol stream first, so that we can add defines etc
 607         init_symbols();
 608
 609         args = argv;
 610         for (;;) {
 611                 char *arg = *++args;
 612                 if (!arg)
 613                         break;
 614
 615                 if (arg[0] == '-' && arg[1]) {
 616                         args = handle_switch(arg+1, args);
 617                         continue;
 618                 }
 619
 620                 /*
 621                  * Hacky hacky hacky: we re-use the argument space
 622                  * to save the filenames.
 623                  */
 624                 argv[files++] = arg;
 625         }
 626
 627         list = NULL;
 628         argv[files] = NULL;
 629         if (files) {
 630                 // Initialize type system
 631                 init_ctype();
 632
 633                 create_builtin_stream();
 634                 add_pre_buffer("#define __CHECKER__ 1\n");
 635                 if (!preprocess_only)
 636                         declare_builtin_functions();
 637
 638                 list = sparse_initial();
 639
 640                 /*
 641                  * Protect the initial token allocations, since
 642                  * they need to survive all the others
 643                  */
 644                 protect_token_alloc();
 645         }
 646         return list;
 647 }
 648
 649 struct symbol_list * __sparse(char **argv)
 650 {
 651         struct symbol_list *res;
 652         char *filename, *next;
 653
 654         /* Clear previous symbol list */
 655         translation_unit_used_list = NULL;
 656
 657         filename = *argv;
 658         if (!filename)
 659                 return NULL;
 660         do {
 661                 next = argv[1];
 662                 *argv++ = next;
 663         } while (next);
 664
 665         start_file_scope();
 666         res = sparse_file(filename);
 667         end_file_scope();
 668
 669         /* Drop the tokens for this file after parsing */
 670         clear_token_alloc();
 671
 672         /* And return it */
 673         return res;
 674 }
 675
 676 struct symbol_list * sparse(char **argv)
 677 {
 678         struct symbol_list *res = __sparse(argv);
 679
 680         /* Evaluate the complete symbol list */
 681         evaluate_symbol_list(res);
 682
 683         return res;
 684 }