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