split-index.c

   1 #include "git-compat-util.h"
   2 #include "alloc.h"
   3 #include "gettext.h"
   4 #include "hash.h"
   5 #include "mem-pool.h"
   6 #include "read-cache-ll.h"
   7 #include "split-index.h"
   8 #include "strbuf.h"
   9 #include "ewah/ewok.h"
  10
  11 struct split_index *init_split_index(struct index_state *istate)
  12 {
  13         if (!istate->split_index) {
  14                 if (istate->sparse_index)
  15                         die(_("cannot use split index with a sparse index"));
  16
  17                 CALLOC_ARRAY(istate->split_index, 1);
  18                 istate->split_index->refcount = 1;
  19         }
  20         return istate->split_index;
  21 }
  22
  23 int read_link_extension(struct index_state *istate,
  24                          const void *data_, unsigned long sz)
  25 {
  26         const unsigned char *data = data_;
  27         struct split_index *si;
  28         int ret;
  29
  30         if (sz < the_hash_algo->rawsz)
  31                 return error("corrupt link extension (too short)");
  32         si = init_split_index(istate);
  33         oidread(&si->base_oid, data);
  34         data += the_hash_algo->rawsz;
  35         sz -= the_hash_algo->rawsz;
  36         if (!sz)
  37                 return 0;
  38         si->delete_bitmap = ewah_new();
  39         ret = ewah_read_mmap(si->delete_bitmap, data, sz);
  40         if (ret < 0)
  41                 return error("corrupt delete bitmap in link extension");
  42         data += ret;
  43         sz -= ret;
  44         si->replace_bitmap = ewah_new();
  45         ret = ewah_read_mmap(si->replace_bitmap, data, sz);
  46         if (ret < 0)
  47                 return error("corrupt replace bitmap in link extension");
  48         if (ret != sz)
  49                 return error("garbage at the end of link extension");
  50         return 0;
  51 }
  52
  53 int write_link_extension(struct strbuf *sb,
  54                          struct index_state *istate)
  55 {
  56         struct split_index *si = istate->split_index;
  57         strbuf_add(sb, si->base_oid.hash, the_hash_algo->rawsz);
  58         if (!si->delete_bitmap && !si->replace_bitmap)
  59                 return 0;
  60         ewah_serialize_strbuf(si->delete_bitmap, sb);
  61         ewah_serialize_strbuf(si->replace_bitmap, sb);
  62         return 0;
  63 }
  64
  65 static void mark_base_index_entries(struct index_state *base)
  66 {
  67         int i;
  68         /*
  69          * To keep track of the shared entries between
  70          * istate->base->cache[] and istate->cache[], base entry
  71          * position is stored in each base entry. All positions start
  72          * from 1 instead of 0, which is reserved to say "this is a new
  73          * entry".
  74          */
  75         for (i = 0; i < base->cache_nr; i++)
  76                 base->cache[i]->index = i + 1;
  77 }
  78
  79 void move_cache_to_base_index(struct index_state *istate)
  80 {
  81         struct split_index *si = istate->split_index;
  82         int i;
  83
  84         /*
  85          * If there was a previous base index, then transfer ownership of allocated
  86          * entries to the parent index.
  87          */
  88         if (si->base &&
  89                 si->base->ce_mem_pool) {
  90
  91                 if (!istate->ce_mem_pool) {
  92                         istate->ce_mem_pool = xmalloc(sizeof(struct mem_pool));
  93                         mem_pool_init(istate->ce_mem_pool, 0);
  94                 }
  95
  96                 mem_pool_combine(istate->ce_mem_pool, istate->split_index->base->ce_mem_pool);
  97         }
  98
  99         ALLOC_ARRAY(si->base, 1);
 100         index_state_init(si->base, istate->repo);
 101         si->base->version = istate->version;
 102         /* zero timestamp disables racy test in ce_write_index() */
 103         si->base->timestamp = istate->timestamp;
 104         ALLOC_GROW(si->base->cache, istate->cache_nr, si->base->cache_alloc);
 105         si->base->cache_nr = istate->cache_nr;
 106
 107         /*
 108          * The mem_pool needs to move with the allocated entries.
 109          */
 110         si->base->ce_mem_pool = istate->ce_mem_pool;
 111         istate->ce_mem_pool = NULL;
 112
 113         COPY_ARRAY(si->base->cache, istate->cache, istate->cache_nr);
 114         mark_base_index_entries(si->base);
 115         for (i = 0; i < si->base->cache_nr; i++)
 116                 si->base->cache[i]->ce_flags &= ~CE_UPDATE_IN_BASE;
 117 }
 118
 119 static void mark_entry_for_delete(size_t pos, void *data)
 120 {
 121         struct index_state *istate = data;
 122         if (pos >= istate->cache_nr)
 123                 die("position for delete %d exceeds base index size %d",
 124                     (int)pos, istate->cache_nr);
 125         istate->cache[pos]->ce_flags |= CE_REMOVE;
 126         istate->split_index->nr_deletions++;
 127 }
 128
 129 static void replace_entry(size_t pos, void *data)
 130 {
 131         struct index_state *istate = data;
 132         struct split_index *si = istate->split_index;
 133         struct cache_entry *dst, *src;
 134
 135         if (pos >= istate->cache_nr)
 136                 die("position for replacement %d exceeds base index size %d",
 137                     (int)pos, istate->cache_nr);
 138         if (si->nr_replacements >= si->saved_cache_nr)
 139                 die("too many replacements (%d vs %d)",
 140                     si->nr_replacements, si->saved_cache_nr);
 141         dst = istate->cache[pos];
 142         if (dst->ce_flags & CE_REMOVE)
 143                 die("entry %d is marked as both replaced and deleted",
 144                     (int)pos);
 145         src = si->saved_cache[si->nr_replacements];
 146         if (ce_namelen(src))
 147                 die("corrupt link extension, entry %d should have "
 148                     "zero length name", (int)pos);
 149         src->index = pos + 1;
 150         src->ce_flags |= CE_UPDATE_IN_BASE;
 151         src->ce_namelen = dst->ce_namelen;
 152         copy_cache_entry(dst, src);
 153         discard_cache_entry(src);
 154         si->nr_replacements++;
 155 }
 156
 157 void merge_base_index(struct index_state *istate)
 158 {
 159         struct split_index *si = istate->split_index;
 160         unsigned int i;
 161
 162         mark_base_index_entries(si->base);
 163
 164         si->saved_cache     = istate->cache;
 165         si->saved_cache_nr  = istate->cache_nr;
 166         istate->cache_nr    = si->base->cache_nr;
 167         istate->cache       = NULL;
 168         istate->cache_alloc = 0;
 169         ALLOC_GROW(istate->cache, istate->cache_nr, istate->cache_alloc);
 170         COPY_ARRAY(istate->cache, si->base->cache, istate->cache_nr);
 171
 172         si->nr_deletions = 0;
 173         si->nr_replacements = 0;
 174         ewah_each_bit(si->replace_bitmap, replace_entry, istate);
 175         ewah_each_bit(si->delete_bitmap, mark_entry_for_delete, istate);
 176         if (si->nr_deletions)
 177                 remove_marked_cache_entries(istate, 0);
 178
 179         for (i = si->nr_replacements; i < si->saved_cache_nr; i++) {
 180                 if (!ce_namelen(si->saved_cache[i]))
 181                         die("corrupt link extension, entry %d should "
 182                             "have non-zero length name", i);
 183                 add_index_entry(istate, si->saved_cache[i],
 184                                 ADD_CACHE_OK_TO_ADD |
 185                                 ADD_CACHE_KEEP_CACHE_TREE |
 186                                 /*
 187                                  * we may have to replay what
 188                                  * merge-recursive.c:update_stages()
 189                                  * does, which has this flag on
 190                                  */
 191                                 ADD_CACHE_SKIP_DFCHECK);
 192                 si->saved_cache[i] = NULL;
 193         }
 194
 195         ewah_free(si->delete_bitmap);
 196         ewah_free(si->replace_bitmap);
 197         FREE_AND_NULL(si->saved_cache);
 198         si->delete_bitmap  = NULL;
 199         si->replace_bitmap = NULL;
 200         si->saved_cache_nr = 0;
 201 }
 202
 203 /*
 204  * Compare most of the fields in two cache entries, i.e. all except the
 205  * hashmap_entry and the name.
 206  */
 207 static int compare_ce_content(struct cache_entry *a, struct cache_entry *b)
 208 {
 209         const unsigned int ondisk_flags = CE_STAGEMASK | CE_VALID |
 210                                           CE_EXTENDED_FLAGS;
 211         unsigned int ce_flags = a->ce_flags;
 212         unsigned int base_flags = b->ce_flags;
 213         int ret;
 214
 215         /* only on-disk flags matter */
 216         a->ce_flags &= ondisk_flags;
 217         b->ce_flags &= ondisk_flags;
 218         ret = memcmp(&a->ce_stat_data, &b->ce_stat_data,
 219                      offsetof(struct cache_entry, name) -
 220                      offsetof(struct cache_entry, oid)) ||
 221                 !oideq(&a->oid, &b->oid);
 222         a->ce_flags = ce_flags;
 223         b->ce_flags = base_flags;
 224
 225         return ret;
 226 }
 227
 228 void prepare_to_write_split_index(struct index_state *istate)
 229 {
 230         struct split_index *si = init_split_index(istate);
 231         struct cache_entry **entries = NULL, *ce;
 232         int i, nr_entries = 0, nr_alloc = 0;
 233
 234         si->delete_bitmap = ewah_new();
 235         si->replace_bitmap = ewah_new();
 236
 237         if (si->base) {
 238                 /* Go through istate->cache[] and mark CE_MATCHED to
 239                  * entry with positive index. We'll go through
 240                  * base->cache[] later to delete all entries in base
 241                  * that are not marked with either CE_MATCHED or
 242                  * CE_UPDATE_IN_BASE. If istate->cache[i] is a
 243                  * duplicate, deduplicate it.
 244                  */
 245                 for (i = 0; i < istate->cache_nr; i++) {
 246                         struct cache_entry *base;
 247                         ce = istate->cache[i];
 248                         if (!ce->index) {
 249                                 /*
 250                                  * During simple update index operations this
 251                                  * is a cache entry that is not present in
 252                                  * the shared index.  It will be added to the
 253                                  * split index.
 254                                  *
 255                                  * However, it might also represent a file
 256                                  * that already has a cache entry in the
 257                                  * shared index, but a new index has just
 258                                  * been constructed by unpack_trees(), and
 259                                  * this entry now refers to different content
 260                                  * than what was recorded in the original
 261                                  * index, e.g. during 'read-tree -m HEAD^' or
 262                                  * 'checkout HEAD^'.  In this case the
 263                                  * original entry in the shared index will be
 264                                  * marked as deleted, and this entry will be
 265                                  * added to the split index.
 266                                  */
 267                                 continue;
 268                         }
 269                         if (ce->index > si->base->cache_nr) {
 270                                 BUG("ce refers to a shared ce at %d, which is beyond the shared index size %d",
 271                                     ce->index, si->base->cache_nr);
 272                         }
 273                         ce->ce_flags |= CE_MATCHED; /* or "shared" */
 274                         base = si->base->cache[ce->index - 1];
 275                         if (ce == base) {
 276                                 /* The entry is present in the shared index. */
 277                                 if (ce->ce_flags & CE_UPDATE_IN_BASE) {
 278                                         /*
 279                                          * Already marked for inclusion in
 280                                          * the split index, either because
 281                                          * the corresponding file was
 282                                          * modified and the cached stat data
 283                                          * was refreshed, or because there
 284                                          * is already a replacement entry in
 285                                          * the split index.
 286                                          * Nothing more to do here.
 287                                          */
 288                                 } else if (!ce_uptodate(ce) &&
 289                                            is_racy_timestamp(istate, ce)) {
 290                                         /*
 291                                          * A racily clean cache entry stored
 292                                          * only in the shared index: it must
 293                                          * be added to the split index, so
 294                                          * the subsequent do_write_index()
 295                                          * can smudge its stat data.
 296                                          */
 297                                         ce->ce_flags |= CE_UPDATE_IN_BASE;
 298                                 } else {
 299                                         /*
 300                                          * The entry is only present in the
 301                                          * shared index and it was not
 302                                          * refreshed.
 303                                          * Just leave it there.
 304                                          */
 305                                 }
 306                                 continue;
 307                         }
 308                         if (ce->ce_namelen != base->ce_namelen ||
 309                             strcmp(ce->name, base->name)) {
 310                                 ce->index = 0;
 311                                 continue;
 312                         }
 313                         /*
 314                          * This is the copy of a cache entry that is present
 315                          * in the shared index, created by unpack_trees()
 316                          * while it constructed a new index.
 317                          */
 318                         if (ce->ce_flags & CE_UPDATE_IN_BASE) {
 319                                 /*
 320                                  * Already marked for inclusion in the split
 321                                  * index, either because the corresponding
 322                                  * file was modified and the cached stat data
 323                                  * was refreshed, or because the original
 324                                  * entry already had a replacement entry in
 325                                  * the split index.
 326                                  * Nothing to do.
 327                                  */
 328                         } else if (!ce_uptodate(ce) &&
 329                                    is_racy_timestamp(istate, ce)) {
 330                                 /*
 331                                  * A copy of a racily clean cache entry from
 332                                  * the shared index.  It must be added to
 333                                  * the split index, so the subsequent
 334                                  * do_write_index() can smudge its stat data.
 335                                  */
 336                                 ce->ce_flags |= CE_UPDATE_IN_BASE;
 337                         } else {
 338                                 /*
 339                                  * Thoroughly compare the cached data to see
 340                                  * whether it should be marked for inclusion
 341                                  * in the split index.
 342                                  *
 343                                  * This comparison might be unnecessary, as
 344                                  * code paths modifying the cached data do
 345                                  * set CE_UPDATE_IN_BASE as well.
 346                                  */
 347                                 if (compare_ce_content(ce, base))
 348                                         ce->ce_flags |= CE_UPDATE_IN_BASE;
 349                         }
 350                         discard_cache_entry(base);
 351                         si->base->cache[ce->index - 1] = ce;
 352                 }
 353                 for (i = 0; i < si->base->cache_nr; i++) {
 354                         ce = si->base->cache[i];
 355                         if ((ce->ce_flags & CE_REMOVE) ||
 356                             !(ce->ce_flags & CE_MATCHED))
 357                                 ewah_set(si->delete_bitmap, i);
 358                         else if (ce->ce_flags & CE_UPDATE_IN_BASE) {
 359                                 ewah_set(si->replace_bitmap, i);
 360                                 ce->ce_flags |= CE_STRIP_NAME;
 361                                 ALLOC_GROW(entries, nr_entries+1, nr_alloc);
 362                                 entries[nr_entries++] = ce;
 363                         }
 364                         if (is_null_oid(&ce->oid))
 365                                 istate->drop_cache_tree = 1;
 366                 }
 367         }
 368
 369         for (i = 0; i < istate->cache_nr; i++) {
 370                 ce = istate->cache[i];
 371                 if ((!si->base || !ce->index) && !(ce->ce_flags & CE_REMOVE)) {
 372                         assert(!(ce->ce_flags & CE_STRIP_NAME));
 373                         ALLOC_GROW(entries, nr_entries+1, nr_alloc);
 374                         entries[nr_entries++] = ce;
 375                 }
 376                 ce->ce_flags &= ~CE_MATCHED;
 377         }
 378
 379         /*
 380          * take cache[] out temporarily, put entries[] in its place
 381          * for writing
 382          */
 383         si->saved_cache = istate->cache;
 384         si->saved_cache_nr = istate->cache_nr;
 385         istate->cache = entries;
 386         istate->cache_nr = nr_entries;
 387 }
 388
 389 void finish_writing_split_index(struct index_state *istate)
 390 {
 391         struct split_index *si = init_split_index(istate);
 392
 393         ewah_free(si->delete_bitmap);
 394         ewah_free(si->replace_bitmap);
 395         si->delete_bitmap = NULL;
 396         si->replace_bitmap = NULL;
 397         free(istate->cache);
 398         istate->cache = si->saved_cache;
 399         istate->cache_nr = si->saved_cache_nr;
 400 }
 401
 402 void discard_split_index(struct index_state *istate)
 403 {
 404         struct split_index *si = istate->split_index;
 405         if (!si)
 406                 return;
 407         istate->split_index = NULL;
 408         si->refcount--;
 409         if (si->refcount)
 410                 return;
 411         if (si->base) {
 412                 discard_index(si->base);
 413                 free(si->base);
 414         }
 415         free(si);
 416 }
 417
 418 void save_or_free_index_entry(struct index_state *istate, struct cache_entry *ce)
 419 {
 420         if (ce->index &&
 421             istate->split_index &&
 422             istate->split_index->base &&
 423             ce->index <= istate->split_index->base->cache_nr &&
 424             ce == istate->split_index->base->cache[ce->index - 1])
 425                 ce->ce_flags |= CE_REMOVE;
 426         else
 427                 discard_cache_entry(ce);
 428 }
 429
 430 void replace_index_entry_in_base(struct index_state *istate,
 431                                  struct cache_entry *old_entry,
 432                                  struct cache_entry *new_entry)
 433 {
 434         if (old_entry->index &&
 435             istate->split_index &&
 436             istate->split_index->base &&
 437             old_entry->index <= istate->split_index->base->cache_nr) {
 438                 new_entry->index = old_entry->index;
 439                 if (old_entry != istate->split_index->base->cache[new_entry->index - 1])
 440                         discard_cache_entry(istate->split_index->base->cache[new_entry->index - 1]);
 441                 istate->split_index->base->cache[new_entry->index - 1] = new_entry;
 442         }
 443 }
 444
 445 void add_split_index(struct index_state *istate)
 446 {
 447         if (!istate->split_index) {
 448                 init_split_index(istate);
 449                 istate->cache_changed |= SPLIT_INDEX_ORDERED;
 450         }
 451 }
 452
 453 void remove_split_index(struct index_state *istate)
 454 {
 455         if (istate->split_index) {
 456                 if (istate->split_index->base) {
 457                         /*
 458                          * When removing the split index, we need to move
 459                          * ownership of the mem_pool associated with the
 460                          * base index to the main index. There may be cache entries
 461                          * allocated from the base's memory pool that are shared with
 462                          * the_index.cache[].
 463                          */
 464                         mem_pool_combine(istate->ce_mem_pool,
 465                                          istate->split_index->base->ce_mem_pool);
 466
 467                         /*
 468                          * The split index no longer owns the mem_pool backing
 469                          * its cache array. As we are discarding this index,
 470                          * mark the index as having no cache entries, so it
 471                          * will not attempt to clean up the cache entries or
 472                          * validate them.
 473                          */
 474                         istate->split_index->base->cache_nr = 0;
 475                 }
 476
 477                 /*
 478                  * We can discard the split index because its
 479                  * memory pool has been incorporated into the
 480                  * memory pool associated with the the_index.
 481                  */
 482                 discard_split_index(istate);
 483
 484                 istate->cache_changed |= SOMETHING_CHANGED;
 485         }
 486 }