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