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 Wone_bit_signed_bitfield = 1;
 172 int Wbitwise = 0;
 173 int Wtypesign = 0;
 174 int Wcontext = 0;
 175 int Wundefined_preprocessor = 0;
 176 int Wptr_subtraction_blows = 0;
 177 int Wcast_to_address_space = 0;
 178 int Wdecl = 0;
 179 int Wtransparent_union = 1;
 180 int preprocess_only;
 181 char *include;
 182 int include_fd = -1;
 183
 184
 185 void add_pre_buffer(const char *fmt, ...)
 186 {
 187         va_list args;
 188         unsigned int size;
 189
 190         va_start(args, fmt);
 191         size = pre_buffer_size;
 192         size += vsnprintf(pre_buffer + size,
 193                 sizeof(pre_buffer) - size,
 194                 fmt, args);
 195         pre_buffer_size = size;
 196         va_end(args);
 197 }
 198
 199 static char **handle_switch_D(char *arg, char **next)
 200 {
 201         const char *name = arg + 1;
 202         const char *value = "1";
 203         for (;;) {
 204                 char c;
 205                 c = *++arg;
 206                 if (!c)
 207                         break;
 208                 if (isspace((unsigned char)c) || c == '=') {
 209                         *arg = '\0';
 210                         value = arg + 1;
 211                         break;
 212                 }
 213         }
 214         add_pre_buffer("#define %s %s\n", name, value);
 215         return next;
 216 }
 217
 218 static char **handle_switch_E(char *arg, char **next)
 219 {
 220         preprocess_only = 1;
 221         return next;
 222 }
 223
 224 static char **handle_switch_v(char *arg, char **next)
 225 {
 226         do {
 227                 verbose++;
 228         } while (*++arg == 'v');
 229         return next;
 230 }
 231
 232 static char **handle_switch_I(char *arg, char **next)
 233 {
 234         char *path = arg+1;
 235
 236         switch (arg[1]) {
 237         case '-':
 238                 add_pre_buffer("#split_include\n");
 239                 break;
 240
 241         case '\0':      /* Plain "-I" */
 242                 path = *++next;
 243                 if (!path)
 244                         die("missing argument for -I option");
 245                 /* Fallthrough */
 246         default:
 247                 add_pre_buffer("#add_include \"%s/\"\n", path);
 248         }
 249         return next;
 250 }
 251
 252 static char **handle_switch_i(char *arg, char **next)
 253 {
 254         if (*next && !strcmp(arg, "include")) {
 255                 char *name = *++next;
 256                 int fd = open(name, O_RDONLY);
 257
 258                 include_fd = fd;
 259                 include = name;
 260                 if (fd < 0)
 261                         perror(name);
 262         }
 263         if (*next && !strcmp(arg, "imacros")) {
 264                 char *name = *++next;
 265                 int fd = open(name, O_RDONLY);
 266
 267                 include_fd = fd;
 268                 include = name;
 269                 if (fd < 0)
 270                         perror(name);
 271         }
 272         else if (*next && !strcmp(arg, "isystem")) {
 273                 char *path = *++next;
 274                 if (!path)
 275                         die("missing argument for -isystem option");
 276                 add_pre_buffer("#add_isystem \"%s/\"\n", path);
 277         }
 278         return next;
 279 }
 280
 281 static char **handle_switch_M(char *arg, char **next)
 282 {
 283         if (!strcmp(arg, "MF") || !strcmp(arg,"MQ") || !strcmp(arg,"MT")) {
 284                 if (!*next)
 285                         die("missing argument for -%s option", arg);
 286                 return next + 1;
 287         }
 288         return next;
 289 }
 290
 291 static char **handle_switch_m(char *arg, char **next)
 292 {
 293         if (!strcmp(arg, "m64")) {
 294                 bits_in_long = 64;
 295                 max_int_alignment = 8;
 296                 bits_in_pointer = 64;
 297                 pointer_alignment = 8;
 298         }
 299         return next;
 300 }
 301
 302 static char **handle_switch_o(char *arg, char **next)
 303 {
 304         if (!strcmp (arg, "o") && *next)
 305                 return next + 1; // "-o foo"
 306         else
 307                 return next;     // "-ofoo" or (bogus) terminal "-o"
 308 }
 309
 310 static const struct warning {
 311         const char *name;
 312         int *flag;
 313 } warnings[] = {
 314         { "cast-to-as", &Wcast_to_address_space },
 315         { "decl", &Wdecl },
 316         { "one-bit-signed-bitfield", &Wone_bit_signed_bitfield },
 317         { "ptr-subtraction-blows", &Wptr_subtraction_blows },
 318         { "default-bitfield-sign", &Wdefault_bitfield_sign },
 319         { "undef", &Wundefined_preprocessor },
 320         { "bitwise", &Wbitwise },
 321         { "typesign", &Wtypesign },
 322         { "context", &Wcontext },
 323         { "transparent-union", &Wtransparent_union },
 324 };
 325
 326
 327 static char **handle_switch_W(char *arg, char **next)
 328 {
 329         int no = 0;
 330         char *p = arg + 1;
 331         unsigned i;
 332
 333         // Prefixes "no" and "no-" mean to turn warning off.
 334         if (p[0] == 'n' && p[1] == 'o') {
 335                 p += 2;
 336                 if (p[0] == '-')
 337                         p++;
 338                 no = 1;
 339         }
 340
 341         for (i = 0; i < sizeof(warnings) / sizeof(warnings[0]); i++) {
 342                 if (!strcmp(p,warnings[i].name)) {
 343                         *warnings[i].flag = !no;
 344                         return next;
 345                 }
 346         }
 347
 348         // Unknown.
 349         return next;
 350 }
 351
 352 static char **handle_switch_U(char *arg, char **next)
 353 {
 354         const char *name = arg + 1;
 355         add_pre_buffer ("#undef %s\n", name);
 356         return next;
 357 }
 358
 359 static char **handle_switch_O(char *arg, char **next)
 360 {
 361         int level = 1;
 362         if (arg[1] >= '0' && arg[1] <= '9')
 363                 level = arg[1] - '0';
 364         optimize = level;
 365         optimize_size = arg[1] == 's';
 366         return next;
 367 }
 368
 369 static char **handle_switch_f(char *arg, char **next)
 370 {
 371         int flag = 1;
 372
 373         arg++;
 374         if (!strncmp(arg, "no-", 3)) {
 375                 flag = 0;
 376                 arg += 3;
 377         }
 378         /* handle switch here.. */
 379         return next;
 380 }
 381
 382 static char **handle_switch_G(char *arg, char **next)
 383 {
 384         if (!strcmp (arg, "G") && *next)
 385                 return next + 1; // "-G 0"
 386         else
 387                 return next;     // "-G0" or (bogus) terminal "-G"
 388 }
 389
 390 static char **handle_nostdinc(char *arg, char **next)
 391 {
 392         add_pre_buffer("#nostdinc\n");
 393         return next;
 394 }
 395
 396 static char **handle_dirafter(char *arg, char **next)
 397 {
 398         char *path = *++next;
 399         if (!path)
 400                 die("missing argument for -dirafter option");
 401         add_pre_buffer("#add_dirafter \"%s/\"\n", path);
 402         return next;
 403 }
 404
 405 struct switches {
 406         const char *name;
 407         char **(*fn)(char *, char**);
 408 };
 409
 410 char **handle_switch(char *arg, char **next)
 411 {
 412         static struct switches cmd[] = {
 413                 { "nostdinc", handle_nostdinc },
 414                 { "dirafter", handle_dirafter },
 415                 { NULL, NULL }
 416         };
 417         struct switches *s;
 418
 419         switch (*arg) {
 420         case 'D': return handle_switch_D(arg, next);
 421         case 'E': return handle_switch_E(arg, next);
 422         case 'I': return handle_switch_I(arg, next);
 423         case 'i': return handle_switch_i(arg, next);
 424         case 'M': return handle_switch_M(arg, next);
 425         case 'm': return handle_switch_m(arg, next);
 426         case 'o': return handle_switch_o(arg, next);
 427         case 'U': return handle_switch_U(arg, next);
 428         case 'v': return handle_switch_v(arg, next);
 429         case 'W': return handle_switch_W(arg, next);
 430         case 'O': return handle_switch_O(arg, next);
 431         case 'f': return handle_switch_f(arg, next);
 432         case 'G': return handle_switch_G(arg, next);
 433         default:
 434                 break;
 435         }
 436
 437         s = cmd;
 438         while (s->name) {
 439                 if (!strcmp(s->name, arg))
 440                         return s->fn(arg, next);
 441                 s++;
 442         }
 443
 444         /*
 445          * Ignore unknown command line options:
 446          * they're probably gcc switches
 447          */
 448         return next;
 449 }
 450
 451 void declare_builtin_functions(void)
 452 {
 453         /* Gaah. gcc knows tons of builtin <string.h> functions */
 454         add_pre_buffer("extern void *__builtin_memcpy(void *, const void *, __SIZE_TYPE__);\n");
 455         add_pre_buffer("extern void *__builtin_memset(void *, int, __SIZE_TYPE__);\n");
 456         add_pre_buffer("extern int __builtin_memcmp(const void *, const void *, __SIZE_TYPE__);\n");
 457         add_pre_buffer("extern int __builtin_strcmp(const char *, const char *);\n");
 458         add_pre_buffer("extern char *__builtin_strchr(const char *, int);\n");
 459         add_pre_buffer("extern char *__builtin_strcpy(char *, const char *);\n");
 460         add_pre_buffer("extern char *__builtin_strncpy(char *, const char *, __SIZE_TYPE__);\n");
 461         add_pre_buffer("extern __SIZE_TYPE__ __builtin_strspn(const char *, const char *);\n");
 462         add_pre_buffer("extern __SIZE_TYPE__ __builtin_strcspn(const char *, const char *);\n");
 463
 464         /* And some random ones.. */
 465         add_pre_buffer("extern void *__builtin_return_address(unsigned int);\n");
 466         add_pre_buffer("extern void *__builtin_frame_address(unsigned int);\n");
 467         add_pre_buffer("extern void __builtin_trap(void);\n");
 468         add_pre_buffer("extern int __builtin_ffs(int);\n");
 469         add_pre_buffer("extern void *__builtin_alloca(__SIZE_TYPE__);\n");
 470 }
 471
 472 void create_builtin_stream(void)
 473 {
 474         add_pre_buffer("#define __GNUC__ %d\n", gcc_major);
 475         add_pre_buffer("#define __GNUC_MINOR__ %d\n", gcc_minor);
 476         add_pre_buffer("#define __GNUC_PATCHLEVEL__ %d\n", gcc_patchlevel);
 477         add_pre_buffer("#define __extension__\n");
 478         add_pre_buffer("#define __pragma__\n");
 479
 480         // gcc defines __SIZE_TYPE__ to be size_t.  For linux/i86 and
 481         // solaris/sparc that is really "unsigned int" and for linux/x86_64
 482         // it is "long unsigned int".  In either case we can probably
 483         // get away with this.  We need the #ifndef as cgcc will define
 484         // the right __SIZE_TYPE__.
 485         add_pre_buffer("#weak_define __SIZE_TYPE__ long unsigned int\n");
 486         add_pre_buffer("#weak_define __STDC__ 1\n");
 487
 488         add_pre_buffer("#define __builtin_stdarg_start(a,b) ((a) = (__builtin_va_list)(&(b)))\n");
 489         add_pre_buffer("#define __builtin_va_start(a,b) ((a) = (__builtin_va_list)(&(b)))\n");
 490         add_pre_buffer("#define __builtin_va_arg(arg,type)  ({ type __va_arg_ret = *(type *)(arg); arg += sizeof(type); __va_arg_ret; })\n");
 491         add_pre_buffer("#define __builtin_va_alist (*(void *)0)\n");
 492         add_pre_buffer("#define __builtin_va_arg_incr(x) ((x) + 1)\n");
 493         add_pre_buffer("#define __builtin_va_end(arg)\n");
 494         add_pre_buffer("#define __builtin_offsetof(type, name) ((__SIZE_TYPE__)&((type *)(0ul))->name)\n");
 495
 496         /* FIXME! We need to do these as special magic macros at expansion time! */
 497         add_pre_buffer("#define __BASE_FILE__ \"base_file.c\"\n");
 498         add_pre_buffer("#define __DATE__ \"??? ?? ????\"\n");
 499         add_pre_buffer("#define __TIME__ \"??:??:??\"\n");
 500
 501         if (optimize)
 502                 add_pre_buffer("#define __OPTIMIZE__ 1\n");
 503         if (optimize_size)
 504                 add_pre_buffer("#define __OPTIMIZE_SIZE__ 1\n");
 505 }
 506
 507 static struct symbol_list *sparse_tokenstream(struct token *token)
 508 {
 509         // Pre-process the stream
 510         token = preprocess(token);
 511
 512         if (preprocess_only) {
 513                 while (!eof_token(token)) {
 514                         int prec = 1;
 515                         struct token *next = token->next;
 516                         const char *separator = "";
 517                         if (next->pos.whitespace)
 518                                 separator = " ";
 519                         if (next->pos.newline) {
 520                                 separator = "\n\t\t\t\t\t";
 521                                 prec = next->pos.pos;
 522                                 if (prec > 4)
 523                                         prec = 4;
 524                         }
 525                         printf("%s%.*s", show_token(token), prec, separator);
 526                         token = next;
 527                 }
 528                 putchar('\n');
 529
 530                 return NULL;
 531         }
 532
 533         // Parse the resulting C code
 534         while (!eof_token(token))
 535                 token = external_declaration(token, &translation_unit_used_list);
 536         return translation_unit_used_list;
 537 }
 538
 539 static struct symbol_list *sparse_file(const char *filename)
 540 {
 541         int fd;
 542         struct token *token;
 543
 544         if (strcmp (filename, "-") == 0) {
 545                 fd = 0;
 546         } else {
 547                 fd = open(filename, O_RDONLY);
 548                 if (fd < 0)
 549                         die("No such file: %s", filename);
 550         }
 551
 552         // Tokenize the input stream
 553         token = tokenize(filename, fd, NULL, includepath);
 554         close(fd);
 555
 556         return sparse_tokenstream(token);
 557 }
 558
 559 /*
 560  * This handles the "-include" directive etc: we're in global
 561  * scope, and all types/macros etc will affect all the following
 562  * files.
 563  *
 564  * NOTE NOTE NOTE! "#undef" of anything in this stage will
 565  * affect all subsequent files too, ie we can have non-local
 566  * behaviour between files!
 567  */
 568 static struct symbol_list *sparse_initial(void)
 569 {
 570         struct token *token;
 571
 572         // Prepend any "include" file to the stream.
 573         // We're in global scope, it will affect all files!
 574         token = NULL;
 575         if (include_fd >= 0)
 576                 token = tokenize(include, include_fd, NULL, includepath);
 577
 578         // Prepend the initial built-in stream
 579         token = tokenize_buffer(pre_buffer, pre_buffer_size, token);
 580         return sparse_tokenstream(token);
 581 }
 582
 583 struct symbol_list *sparse_initialize(int argc, char **argv)
 584 {
 585         char **args;
 586         int files = 0;
 587         struct symbol_list *list;
 588
 589         // Initialize symbol stream first, so that we can add defines etc
 590         init_symbols();
 591
 592         args = argv;
 593         for (;;) {
 594                 char *arg = *++args;
 595                 if (!arg)
 596                         break;
 597
 598                 if (arg[0] == '-' && arg[1]) {
 599                         args = handle_switch(arg+1, args);
 600                         continue;
 601                 }
 602
 603                 /*
 604                  * Hacky hacky hacky: we re-use the argument space
 605                  * to save the filenames.
 606                  */
 607                 argv[files++] = arg;
 608         }
 609
 610         list = NULL;
 611         argv[files] = NULL;
 612         if (files) {
 613                 // Initialize type system
 614                 init_ctype();
 615
 616                 create_builtin_stream();
 617                 add_pre_buffer("#define __CHECKER__ 1\n");
 618                 if (!preprocess_only)
 619                         declare_builtin_functions();
 620
 621                 list = sparse_initial();
 622
 623                 /*
 624                  * Protect the initial token allocations, since
 625                  * they need to survive all the others
 626                  */
 627                 protect_token_alloc();
 628         }
 629         return list;
 630 }
 631
 632 struct symbol_list * sparse(char **argv)
 633 {
 634         struct symbol_list *res;
 635         char *filename, *next;
 636
 637         /* Clear previous symbol list */
 638         translation_unit_used_list = NULL;
 639
 640         filename = *argv;
 641         if (!filename)
 642                 return NULL;
 643         do {
 644                 next = argv[1];
 645                 *argv++ = next;
 646         } while (next);
 647
 648         start_file_scope();
 649         res = sparse_file(filename);
 650         end_file_scope();
 651
 652         /* Drop the tokens for this file after parsing */
 653         clear_token_alloc();
 654
 655         /* Evaluate the complete symbol list */
 656         evaluate_symbol_list(res);
 657
 658         /* And return it */
 659         return res;
 660 }