libcpp/pch.c

   1 /* Part of CPP library.  (Precompiled header reading/writing.)
   2    Copyright (C) 2000, 2001, 2002, 2003, 2004, 2005
   3    Free Software Foundation, Inc.
   4
   5 This program is free software; you can redistribute it and/or modify it
   6 under the terms of the GNU General Public License as published by the
   7 Free Software Foundation; either version 2, or (at your option) any
   8 later version.
   9
  10 This program is distributed in the hope that it will be useful,
  11 but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 GNU General Public License for more details.
  14
  15 You should have received a copy of the GNU General Public License
  16 along with this program; if not, write to the Free Software
  17 Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */
  18
  19 #include "config.h"
  20 #include "system.h"
  21 #include "cpplib.h"
  22 #include "internal.h"
  23 #include "hashtab.h"
  24 #include "mkdeps.h"
  25
  26 static int write_macdef (cpp_reader *, cpp_hashnode *, void *);
  27 static int save_idents (cpp_reader *, cpp_hashnode *, void *);
  28 static hashval_t hashmem (const void *, size_t);
  29 static hashval_t cpp_string_hash (const void *);
  30 static int cpp_string_eq (const void *, const void *);
  31 static int count_defs (cpp_reader *, cpp_hashnode *, void *);
  32 static int comp_hashnodes (const void *, const void *);
  33 static int collect_ht_nodes (cpp_reader *, cpp_hashnode *, void *);
  34 static int write_defs (cpp_reader *, cpp_hashnode *, void *);
  35 static int save_macros (cpp_reader *, cpp_hashnode *, void *);
  36
  37 /* This structure represents a macro definition on disk.  */
  38 struct macrodef_struct
  39 {
  40   unsigned int definition_length;
  41   unsigned short name_length;
  42   unsigned short flags;
  43 };
  44
  45 /* This is how we write out a macro definition.
  46    Suitable for being called by cpp_forall_identifiers.  */
  47
  48 static int
  49 write_macdef (cpp_reader *pfile, cpp_hashnode *hn, void *file_p)
  50 {
  51   FILE *f = (FILE *) file_p;
  52   switch (hn->type)
  53     {
  54     case NT_VOID:
  55       if (! (hn->flags & NODE_POISONED))
  56         return 1;
  57
  58     case NT_MACRO:
  59       if ((hn->flags & NODE_BUILTIN))
  60         return 1;
  61
  62       {
  63         struct macrodef_struct s;
  64         const unsigned char *defn;
  65
  66         s.name_length = NODE_LEN (hn);
  67         s.flags = hn->flags & NODE_POISONED;
  68
  69         if (hn->type == NT_MACRO)
  70           {
  71             defn = cpp_macro_definition (pfile, hn);
  72             s.definition_length = ustrlen (defn);
  73           }
  74         else
  75           {
  76             defn = NODE_NAME (hn);
  77             s.definition_length = s.name_length;
  78           }
  79
  80         if (fwrite (&s, sizeof (s), 1, f) != 1
  81             || fwrite (defn, 1, s.definition_length, f) != s.definition_length)
  82           {
  83             cpp_errno (pfile, CPP_DL_ERROR,
  84                        "while writing precompiled header");
  85             return 0;
  86           }
  87       }
  88       return 1;
  89
  90     case NT_ASSERTION:
  91       /* Not currently implemented.  */
  92       return 1;
  93
  94     default:
  95       abort ();
  96     }
  97 }
  98
  99 /* This structure records the names of the defined macros.
 100    It's also used as a callback structure for size_initial_idents
 101    and save_idents.  */
 102
 103 struct cpp_savedstate
 104 {
 105   /* A hash table of the defined identifiers.  */
 106   htab_t definedhash;
 107   /* The size of the definitions of those identifiers (the size of
 108      'definedstrs').  */
 109   size_t hashsize;
 110   /* Number of definitions */
 111   size_t n_defs;
 112   /* Array of definitions.  In cpp_write_pch_deps it is used for sorting.  */
 113   cpp_hashnode **defs;
 114   /* Space for the next definition.  Definitions are null-terminated
 115      strings.  */
 116   unsigned char *definedstrs;
 117 };
 118
 119 /* Save this identifier into the state: put it in the hash table,
 120    put the definition in 'definedstrs'.  */
 121
 122 static int
 123 save_idents (cpp_reader *pfile ATTRIBUTE_UNUSED, cpp_hashnode *hn, void *ss_p)
 124 {
 125   struct cpp_savedstate *const ss = (struct cpp_savedstate *)ss_p;
 126
 127   if (hn->type != NT_VOID)
 128     {
 129       struct cpp_string news;
 130       void **slot;
 131
 132       news.len = NODE_LEN (hn);
 133       news.text= NODE_NAME (hn);
 134       slot = htab_find_slot (ss->definedhash, &news, INSERT);
 135       if (*slot == NULL)
 136         {
 137           struct cpp_string *sp;
 138           unsigned char *text;
 139
 140           sp = xmalloc (sizeof (struct cpp_string));
 141           *slot = sp;
 142
 143           sp->len = NODE_LEN (hn);
 144           sp->text = text = xmalloc (NODE_LEN (hn));
 145           memcpy (text, NODE_NAME (hn), NODE_LEN (hn));
 146         }
 147     }
 148
 149   return 1;
 150 }
 151
 152 /* Hash some memory in a generic way.  */
 153
 154 static hashval_t
 155 hashmem (const void *p_p, size_t sz)
 156 {
 157   const unsigned char *p = (const unsigned char *)p_p;
 158   size_t i;
 159   hashval_t h;
 160
 161   h = 0;
 162   for (i = 0; i < sz; i++)
 163     h = h * 67 - (*p++ - 113);
 164   return h;
 165 }
 166
 167 /* Hash a cpp string for the hashtable machinery.  */
 168
 169 static hashval_t
 170 cpp_string_hash (const void *a_p)
 171 {
 172   const struct cpp_string *a = (const struct cpp_string *) a_p;
 173   return hashmem (a->text, a->len);
 174 }
 175
 176 /* Compare two cpp strings for the hashtable machinery.  */
 177
 178 static int
 179 cpp_string_eq (const void *a_p, const void *b_p)
 180 {
 181   const struct cpp_string *a = (const struct cpp_string *) a_p;
 182   const struct cpp_string *b = (const struct cpp_string *) b_p;
 183   return (a->len == b->len
 184           && memcmp (a->text, b->text, a->len) == 0);
 185 }
 186
 187 /* Save the current definitions of the cpp_reader for dependency
 188    checking purposes.  When writing a precompiled header, this should
 189    be called at the same point in the compilation as cpp_valid_state
 190    would be called when reading the precompiled header back in.  */
 191
 192 int
 193 cpp_save_state (cpp_reader *r, FILE *f)
 194 {
 195   /* Save the list of non-void identifiers for the dependency checking.  */
 196   r->savedstate = xmalloc (sizeof (struct cpp_savedstate));
 197   r->savedstate->definedhash = htab_create (100, cpp_string_hash,
 198                                             cpp_string_eq, NULL);
 199   cpp_forall_identifiers (r, save_idents, r->savedstate);
 200
 201   /* Write out the list of defined identifiers.  */
 202   cpp_forall_identifiers (r, write_macdef, f);
 203
 204   return 0;
 205 }
 206
 207 /* Calculate the 'hashsize' field of the saved state.  */
 208
 209 static int
 210 count_defs (cpp_reader *pfile ATTRIBUTE_UNUSED, cpp_hashnode *hn, void *ss_p)
 211 {
 212   struct cpp_savedstate *const ss = (struct cpp_savedstate *)ss_p;
 213
 214   switch (hn->type)
 215     {
 216     case NT_MACRO:
 217       if (hn->flags & NODE_BUILTIN)
 218         return 1;
 219
 220       /* else fall through.  */
 221
 222     case NT_VOID:
 223       {
 224         struct cpp_string news;
 225         void **slot;
 226
 227         news.len = NODE_LEN (hn);
 228         news.text = NODE_NAME (hn);
 229         slot = htab_find (ss->definedhash, &news);
 230         if (slot == NULL)
 231           {
 232             ss->hashsize += NODE_LEN (hn) + 1;
 233             ss->n_defs += 1;
 234           }
 235       }
 236       return 1;
 237
 238     case NT_ASSERTION:
 239       /* Not currently implemented.  */
 240       return 1;
 241
 242     default:
 243       abort ();
 244     }
 245 }
 246
 247 /* Collect the identifiers into the state's string table.  */
 248 static int
 249 write_defs (cpp_reader *pfile ATTRIBUTE_UNUSED, cpp_hashnode *hn, void *ss_p)
 250 {
 251   struct cpp_savedstate *const ss = (struct cpp_savedstate *)ss_p;
 252
 253   switch (hn->type)
 254     {
 255     case NT_MACRO:
 256       if (hn->flags & NODE_BUILTIN)
 257         return 1;
 258
 259       /* else fall through.  */
 260
 261     case NT_VOID:
 262       {
 263         struct cpp_string news;
 264         void **slot;
 265
 266         news.len = NODE_LEN (hn);
 267         news.text = NODE_NAME (hn);
 268         slot = htab_find (ss->definedhash, &news);
 269         if (slot == NULL)
 270           {
 271             ss->defs[ss->n_defs] = hn;
 272             ss->n_defs += 1;
 273           }
 274       }
 275       return 1;
 276
 277     case NT_ASSERTION:
 278       /* Not currently implemented.  */
 279       return 1;
 280
 281     default:
 282       abort ();
 283     }
 284 }
 285
 286 /* Comparison function for qsort.  The arguments point to pointers of
 287    type ht_hashnode *.  */
 288 static int
 289 comp_hashnodes (const void *px, const void *py)
 290 {
 291   cpp_hashnode *x = *(cpp_hashnode **) px;
 292   cpp_hashnode *y = *(cpp_hashnode **) py;
 293   return ustrcmp (NODE_NAME (x), NODE_NAME (y));
 294 }
 295
 296 /* Write out the remainder of the dependency information.  This should be
 297    called after the PCH is ready to be saved.  */
 298
 299 int
 300 cpp_write_pch_deps (cpp_reader *r, FILE *f)
 301 {
 302   struct macrodef_struct z;
 303   struct cpp_savedstate *const ss = r->savedstate;
 304   unsigned char *definedstrs;
 305   size_t i;
 306
 307   /* Collect the list of identifiers which have been seen and
 308      weren't defined to anything previously.  */
 309   ss->hashsize = 0;
 310   ss->n_defs = 0;
 311   cpp_forall_identifiers (r, count_defs, ss);
 312
 313   ss->defs = xmalloc (ss->n_defs * sizeof (cpp_hashnode *));
 314   ss->n_defs = 0;
 315   cpp_forall_identifiers (r, write_defs, ss);
 316
 317   /* Sort the list, copy it into a buffer, and write it out.  */
 318   qsort (ss->defs, ss->n_defs, sizeof (cpp_hashnode *), &comp_hashnodes);
 319   definedstrs = ss->definedstrs = xmalloc (ss->hashsize);
 320   for (i = 0; i < ss->n_defs; ++i)
 321     {
 322       size_t len = NODE_LEN (ss->defs[i]);
 323       memcpy (definedstrs, NODE_NAME (ss->defs[i]), len + 1);
 324       definedstrs += len + 1;
 325     }
 326
 327   memset (&z, 0, sizeof (z));
 328   z.definition_length = ss->hashsize;
 329   if (fwrite (&z, sizeof (z), 1, f) != 1
 330       || fwrite (ss->definedstrs, ss->hashsize, 1, f) != 1)
 331     {
 332       cpp_errno (r, CPP_DL_ERROR, "while writing precompiled header");
 333       return -1;
 334     }
 335   free (ss->definedstrs);
 336
 337   /* Free the saved state.  */
 338   free (ss);
 339   r->savedstate = NULL;
 340   return 0;
 341 }
 342
 343 /* Write out the definitions of the preprocessor, in a form suitable for
 344    cpp_read_state.  */
 345
 346 int
 347 cpp_write_pch_state (cpp_reader *r, FILE *f)
 348 {
 349   if (!r->deps)
 350     r->deps = deps_init ();
 351
 352   if (deps_save (r->deps, f) != 0)
 353     {
 354       cpp_errno (r, CPP_DL_ERROR, "while writing precompiled header");
 355       return -1;
 356     }
 357
 358   if (! _cpp_save_file_entries (r, f))
 359     {
 360       cpp_errno (r, CPP_DL_ERROR, "while writing precompiled header");
 361       return -1;
 362     }
 363
 364   return 0;
 365 }
 366
 367
 368 /* Data structure to transform hash table nodes into a sorted list */
 369
 370 struct ht_node_list
 371 {
 372   /* Array of nodes */
 373   cpp_hashnode **defs;
 374   /* Number of nodes in the array */
 375   size_t n_defs;
 376   /* Size of the allocated array */
 377   size_t asize;
 378 };
 379
 380 /* Callback for collecting identifiers from hash table */
 381
 382 static int
 383 collect_ht_nodes (cpp_reader *pfile ATTRIBUTE_UNUSED, cpp_hashnode *hn,
 384                   void *nl_p)
 385 {
 386   struct ht_node_list *const nl = (struct ht_node_list *)nl_p;
 387
 388   if (hn->type != NT_VOID || hn->flags & NODE_POISONED)
 389     {
 390       if (nl->n_defs == nl->asize)
 391         {
 392           nl->asize *= 2;
 393           nl->defs = xrealloc (nl->defs, nl->asize * sizeof (cpp_hashnode *));
 394         }
 395
 396       nl->defs[nl->n_defs] = hn;
 397       ++nl->n_defs;
 398     }
 399   return 1;
 400 }
 401
 402
 403 /* Return nonzero if FD is a precompiled header which is consistent
 404    with the preprocessor's current definitions.  It will be consistent
 405    when:
 406
 407    - anything that was defined just before the PCH was generated
 408      is defined the same way now; and
 409    - anything that was not defined then, but is defined now, was not
 410      used by the PCH.
 411
 412    NAME is used to print warnings if `warn_invalid_pch' is set in the
 413    reader's flags.
 414 */
 415
 416 int
 417 cpp_valid_state (cpp_reader *r, const char *name, int fd)
 418 {
 419   struct macrodef_struct m;
 420   size_t namebufsz = 256;
 421   unsigned char *namebuf = xmalloc (namebufsz);
 422   unsigned char *undeftab = NULL;
 423   struct ht_node_list nl = { 0, 0, 0 };
 424   unsigned char *first, *last;
 425   unsigned int i;
 426
 427   /* Read in the list of identifiers that must be defined
 428      Check that they are defined in the same way.  */
 429   for (;;)
 430     {
 431       cpp_hashnode *h;
 432       const unsigned char *newdefn;
 433
 434       if (read (fd, &m, sizeof (m)) != sizeof (m))
 435         goto error;
 436
 437       if (m.name_length == 0)
 438         break;
 439
 440       /* If this file is already preprocessed, there won't be any
 441          macros defined, and that's OK.  */
 442       if (CPP_OPTION (r, preprocessed))
 443         {
 444           if (lseek (fd, m.definition_length, SEEK_CUR) == -1)
 445             goto error;
 446           continue;
 447         }
 448
 449       if (m.definition_length > namebufsz)
 450         {
 451           free (namebuf);
 452           namebufsz = m.definition_length + 256;
 453           namebuf = xmalloc (namebufsz);
 454         }
 455
 456       if ((size_t)read (fd, namebuf, m.definition_length)
 457           != m.definition_length)
 458         goto error;
 459
 460       h = cpp_lookup (r, namebuf, m.name_length);
 461       if (m.flags & NODE_POISONED
 462           || h->type != NT_MACRO
 463           || h->flags & NODE_POISONED)
 464         {
 465           if (CPP_OPTION (r, warn_invalid_pch))
 466             cpp_error (r, CPP_DL_WARNING_SYSHDR,
 467                        "%s: not used because `%.*s' not defined",
 468                        name, m.name_length, namebuf);
 469           goto fail;
 470         }
 471
 472       newdefn = cpp_macro_definition (r, h);
 473
 474       if (m.definition_length != ustrlen (newdefn)
 475           || memcmp (namebuf, newdefn, m.definition_length) != 0)
 476         {
 477           if (CPP_OPTION (r, warn_invalid_pch))
 478             cpp_error (r, CPP_DL_WARNING_SYSHDR,
 479                "%s: not used because `%.*s' defined as `%s' not `%.*s'",
 480                        name, m.name_length, namebuf, newdefn + m.name_length,
 481                        m.definition_length - m.name_length,
 482                        namebuf +  m.name_length);
 483           goto fail;
 484         }
 485     }
 486   free (namebuf);
 487   namebuf = NULL;
 488
 489   /* Read in the list of identifiers that must not be defined.
 490      Check that they really aren't.  */
 491   undeftab = xmalloc (m.definition_length);
 492   if ((size_t) read (fd, undeftab, m.definition_length) != m.definition_length)
 493     goto error;
 494
 495   /* Collect identifiers from the current hash table.  */
 496   nl.n_defs = 0;
 497   nl.asize = 10;
 498   nl.defs = xmalloc (nl.asize * sizeof (cpp_hashnode *));
 499   cpp_forall_identifiers (r, &collect_ht_nodes, &nl);
 500   qsort (nl.defs, nl.n_defs, sizeof (cpp_hashnode *), &comp_hashnodes);
 501
 502   /* Loop through nl.defs and undeftab, both of which are sorted lists.
 503      There should be no matches.  */
 504   first = undeftab;
 505   last = undeftab + m.definition_length;
 506   i = 0;
 507
 508   while (first < last && i < nl.n_defs)
 509     {
 510       int cmp = ustrcmp (first, NODE_NAME (nl.defs[i]));
 511
 512       if (cmp < 0)
 513         first += ustrlen (first) + 1;
 514       else if (cmp > 0)
 515         ++i;
 516       else
 517         {
 518           if (CPP_OPTION (r, warn_invalid_pch))
 519             cpp_error (r, CPP_DL_WARNING_SYSHDR,
 520                        "%s: not used because `%s' is defined",
 521                        name, first);
 522           goto fail;
 523         }
 524     }
 525
 526   free(nl.defs);
 527   free (undeftab);
 528
 529   /* We win!  */
 530   return 0;
 531
 532  error:
 533   cpp_errno (r, CPP_DL_ERROR, "while reading precompiled header");
 534   return -1;
 535
 536  fail:
 537   if (namebuf != NULL)
 538     free (namebuf);
 539   if (undeftab != NULL)
 540     free (undeftab);
 541   if (nl.defs != NULL)
 542     free (nl.defs);
 543   return 1;
 544 }
 545
 546 /* Save all the existing macros.  */
 547
 548 struct save_macro_data
 549 {
 550   uchar **defns;
 551   size_t count;
 552   size_t array_size;
 553   char **saved_pragmas;
 554 };
 555
 556 /* Save the definition of a single macro, so that it will persist
 557    across a PCH restore.  Because macro data is in GCed memory, which
 558    will be blown away by PCH, it must be temporarily copied to
 559    malloced memory.  (The macros will refer to identifier nodes which
 560    are also GCed and so on, so the copying is done by turning them
 561    into self-contained strings.)  The assumption is that most macro
 562    definitions will come from the PCH file, not from the compilation
 563    before the PCH file is loaded, so it doesn't matter that this is
 564    a little expensive.
 565
 566    It would reduce the cost even further if macros defined in the PCH
 567    file were not saved in this way, but this is not done (yet), except
 568    for builtins, and for #assert by default.  */
 569
 570 static int
 571 save_macros (cpp_reader *r, cpp_hashnode *h, void *data_p)
 572 {
 573   struct save_macro_data *data = (struct save_macro_data *)data_p;
 574   if (h->type != NT_VOID
 575       && (h->flags & NODE_BUILTIN) == 0)
 576     {
 577       if (data->count == data->array_size)
 578         {
 579           data->array_size *= 2;
 580           data->defns = xrealloc (data->defns, (data->array_size
 581                                                 * sizeof (uchar *)));
 582         }
 583
 584       switch (h->type)
 585         {
 586         case NT_ASSERTION:
 587           /* Not currently implemented.  */
 588           return 1;
 589
 590         case NT_MACRO:
 591           {
 592             const uchar * defn = cpp_macro_definition (r, h);
 593             size_t defnlen = ustrlen (defn);
 594
 595             data->defns[data->count] = xmemdup (defn, defnlen, defnlen + 2);
 596             data->defns[data->count][defnlen] = '\n';
 597           }
 598           break;
 599
 600         default:
 601           abort ();
 602         }
 603       data->count++;
 604     }
 605   return 1;
 606 }
 607
 608 /* Prepare to restore the state, by saving the currently-defined
 609    macros in 'data'.  */
 610
 611 void
 612 cpp_prepare_state (cpp_reader *r, struct save_macro_data **data)
 613 {
 614   struct save_macro_data *d = xmalloc (sizeof (struct save_macro_data));
 615
 616   d->array_size = 512;
 617   d->defns = xmalloc (d->array_size * sizeof (d->defns[0]));
 618   d->count = 0;
 619   cpp_forall_identifiers (r, save_macros, d);
 620   d->saved_pragmas = _cpp_save_pragma_names (r);
 621   *data = d;
 622 }
 623
 624 /* Given a precompiled header that was previously determined to be valid,
 625    apply all its definitions (and undefinitions) to the current state.
 626    DEPNAME is passed to deps_restore.  */
 627
 628 int
 629 cpp_read_state (cpp_reader *r, const char *name, FILE *f,
 630                 struct save_macro_data *data)
 631 {
 632   size_t i;
 633   struct lexer_state old_state;
 634
 635   /* Restore spec_nodes, which will be full of references to the old
 636      hashtable entries and so will now be invalid.  */
 637   {
 638     struct spec_nodes *s = &r->spec_nodes;
 639     s->n_defined        = cpp_lookup (r, DSC("defined"));
 640     s->n_true           = cpp_lookup (r, DSC("true"));
 641     s->n_false          = cpp_lookup (r, DSC("false"));
 642     s->n__VA_ARGS__     = cpp_lookup (r, DSC("__VA_ARGS__"));
 643   }
 644
 645   old_state = r->state;
 646   r->state.in_directive = 1;
 647   r->state.prevent_expansion = 1;
 648   r->state.angled_headers = 0;
 649
 650   /* Run through the carefully-saved macros, insert them.  */
 651   for (i = 0; i < data->count; i++)
 652     {
 653       cpp_hashnode *h;
 654       size_t namelen;
 655       uchar *defn;
 656
 657       namelen = ustrcspn (data->defns[i], "( \n");
 658       h = cpp_lookup (r, data->defns[i], namelen);
 659       defn = data->defns[i] + namelen;
 660
 661       /* The PCH file is valid, so we know that if there is a definition
 662          from the PCH file it must be the same as the one we had
 663          originally, and so do not need to restore it.  */
 664       if (h->type == NT_VOID)
 665         {
 666           if (cpp_push_buffer (r, defn, ustrchr (defn, '\n') - defn, true)
 667               != NULL)
 668             {
 669               _cpp_clean_line (r);
 670               if (!_cpp_create_definition (r, h))
 671                 abort ();
 672               _cpp_pop_buffer (r);
 673             }
 674           else
 675             abort ();
 676         }
 677
 678       free (data->defns[i]);
 679     }
 680   r->state = old_state;
 681
 682   _cpp_restore_pragma_names (r, data->saved_pragmas);
 683
 684   free (data);
 685
 686   if (deps_restore (r->deps, f, CPP_OPTION (r, restore_pch_deps) ? name : NULL)
 687       != 0)
 688     goto error;
 689
 690   if (! _cpp_read_file_entries (r, f))
 691     goto error;
 692
 693   return 0;
 694
 695  error:
 696   cpp_errno (r, CPP_DL_ERROR, "while reading precompiled header");
 697   return -1;
 698 }