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