split-index.c

   1 #include "cache.h"
   2 #include "split-index.h"
   3 #include "ewah/ewok.h"
   4
   5 struct split_index *init_split_index(struct index_state *istate)
   6 {
   7         if (!istate->split_index) {
   8                 istate->split_index = xcalloc(1, sizeof(*istate->split_index));
   9                 istate->split_index->refcount = 1;
  10         }
  11         return istate->split_index;
  12 }
  13
  14 int read_link_extension(struct index_state *istate,
  15                          const void *data_, unsigned long sz)
  16 {
  17         const unsigned char *data = data_;
  18         struct split_index *si;
  19         int ret;
  20
  21         if (sz < the_hash_algo->rawsz)
  22                 return error("corrupt link extension (too short)");
  23         si = init_split_index(istate);
  24         hashcpy(si->base_oid.hash, data);
  25         data += the_hash_algo->rawsz;
  26         sz -= the_hash_algo->rawsz;
  27         if (!sz)
  28                 return 0;
  29         si->delete_bitmap = ewah_new();
  30         ret = ewah_read_mmap(si->delete_bitmap, data, sz);
  31         if (ret < 0)
  32                 return error("corrupt delete bitmap in link extension");
  33         data += ret;
  34         sz -= ret;
  35         si->replace_bitmap = ewah_new();
  36         ret = ewah_read_mmap(si->replace_bitmap, data, sz);
  37         if (ret < 0)
  38                 return error("corrupt replace bitmap in link extension");
  39         if (ret != sz)
  40                 return error("garbage at the end of link extension");
  41         return 0;
  42 }
  43
  44 int write_link_extension(struct strbuf *sb,
  45                          struct index_state *istate)
  46 {
  47         struct split_index *si = istate->split_index;
  48         strbuf_add(sb, si->base_oid.hash, the_hash_algo->rawsz);
  49         if (!si->delete_bitmap && !si->replace_bitmap)
  50                 return 0;
  51         ewah_serialize_strbuf(si->delete_bitmap, sb);
  52         ewah_serialize_strbuf(si->replace_bitmap, sb);
  53         return 0;
  54 }
  55
  56 static void mark_base_index_entries(struct index_state *base)
  57 {
  58         int i;
  59         /*
  60          * To keep track of the shared entries between
  61          * istate->base->cache[] and istate->cache[], base entry
  62          * position is stored in each base entry. All positions start
  63          * from 1 instead of 0, which is reserved to say "this is a new
  64          * entry".
  65          */
  66         for (i = 0; i < base->cache_nr; i++)
  67                 base->cache[i]->index = i + 1;
  68 }
  69
  70 void move_cache_to_base_index(struct index_state *istate)
  71 {
  72         struct split_index *si = istate->split_index;
  73         int i;
  74
  75         /*
  76          * If there was a previous base index, then transfer ownership of allocated
  77          * entries to the parent index.
  78          */
  79         if (si->base &&
  80                 si->base->ce_mem_pool) {
  81
  82                 if (!istate->ce_mem_pool)
  83                         mem_pool_init(&istate->ce_mem_pool, 0);
  84
  85                 mem_pool_combine(istate->ce_mem_pool, istate->split_index->base->ce_mem_pool);
  86         }
  87
  88         si->base = xcalloc(1, sizeof(*si->base));
  89         si->base->version = istate->version;
  90         /* zero timestamp disables racy test in ce_write_index() */
  91         si->base->timestamp = istate->timestamp;
  92         ALLOC_GROW(si->base->cache, istate->cache_nr, si->base->cache_alloc);
  93         si->base->cache_nr = istate->cache_nr;
  94
  95         /*
  96          * The mem_pool needs to move with the allocated entries.
  97          */
  98         si->base->ce_mem_pool = istate->ce_mem_pool;
  99         istate->ce_mem_pool = NULL;
 100
 101         COPY_ARRAY(si->base->cache, istate->cache, istate->cache_nr);
 102         mark_base_index_entries(si->base);
 103         for (i = 0; i < si->base->cache_nr; i++)
 104                 si->base->cache[i]->ce_flags &= ~CE_UPDATE_IN_BASE;
 105 }
 106
 107 static void mark_entry_for_delete(size_t pos, void *data)
 108 {
 109         struct index_state *istate = data;
 110         if (pos >= istate->cache_nr)
 111                 die("position for delete %d exceeds base index size %d",
 112                     (int)pos, istate->cache_nr);
 113         istate->cache[pos]->ce_flags |= CE_REMOVE;
 114         istate->split_index->nr_deletions = 1;
 115 }
 116
 117 static void replace_entry(size_t pos, void *data)
 118 {
 119         struct index_state *istate = data;
 120         struct split_index *si = istate->split_index;
 121         struct cache_entry *dst, *src;
 122
 123         if (pos >= istate->cache_nr)
 124                 die("position for replacement %d exceeds base index size %d",
 125                     (int)pos, istate->cache_nr);
 126         if (si->nr_replacements >= si->saved_cache_nr)
 127                 die("too many replacements (%d vs %d)",
 128                     si->nr_replacements, si->saved_cache_nr);
 129         dst = istate->cache[pos];
 130         if (dst->ce_flags & CE_REMOVE)
 131                 die("entry %d is marked as both replaced and deleted",
 132                     (int)pos);
 133         src = si->saved_cache[si->nr_replacements];
 134         if (ce_namelen(src))
 135                 die("corrupt link extension, entry %d should have "
 136                     "zero length name", (int)pos);
 137         src->index = pos + 1;
 138         src->ce_flags |= CE_UPDATE_IN_BASE;
 139         src->ce_namelen = dst->ce_namelen;
 140         copy_cache_entry(dst, src);
 141         discard_cache_entry(src);
 142         si->nr_replacements++;
 143 }
 144
 145 void merge_base_index(struct index_state *istate)
 146 {
 147         struct split_index *si = istate->split_index;
 148         unsigned int i;
 149
 150         mark_base_index_entries(si->base);
 151
 152         si->saved_cache     = istate->cache;
 153         si->saved_cache_nr  = istate->cache_nr;
 154         istate->cache_nr    = si->base->cache_nr;
 155         istate->cache       = NULL;
 156         istate->cache_alloc = 0;
 157         ALLOC_GROW(istate->cache, istate->cache_nr, istate->cache_alloc);
 158         COPY_ARRAY(istate->cache, si->base->cache, istate->cache_nr);
 159
 160         si->nr_deletions = 0;
 161         si->nr_replacements = 0;
 162         ewah_each_bit(si->replace_bitmap, replace_entry, istate);
 163         ewah_each_bit(si->delete_bitmap, mark_entry_for_delete, istate);
 164         if (si->nr_deletions)
 165                 remove_marked_cache_entries(istate);
 166
 167         for (i = si->nr_replacements; i < si->saved_cache_nr; i++) {
 168                 if (!ce_namelen(si->saved_cache[i]))
 169                         die("corrupt link extension, entry %d should "
 170                             "have non-zero length name", i);
 171                 add_index_entry(istate, si->saved_cache[i],
 172                                 ADD_CACHE_OK_TO_ADD |
 173                                 ADD_CACHE_KEEP_CACHE_TREE |
 174                                 /*
 175                                  * we may have to replay what
 176                                  * merge-recursive.c:update_stages()
 177                                  * does, which has this flag on
 178                                  */
 179                                 ADD_CACHE_SKIP_DFCHECK);
 180                 si->saved_cache[i] = NULL;
 181         }
 182
 183         ewah_free(si->delete_bitmap);
 184         ewah_free(si->replace_bitmap);
 185         FREE_AND_NULL(si->saved_cache);
 186         si->delete_bitmap  = NULL;
 187         si->replace_bitmap = NULL;
 188         si->saved_cache_nr = 0;
 189 }
 190
 191 void prepare_to_write_split_index(struct index_state *istate)
 192 {
 193         struct split_index *si = init_split_index(istate);
 194         struct cache_entry **entries = NULL, *ce;
 195         int i, nr_entries = 0, nr_alloc = 0;
 196
 197         si->delete_bitmap = ewah_new();
 198         si->replace_bitmap = ewah_new();
 199
 200         if (si->base) {
 201                 /* Go through istate->cache[] and mark CE_MATCHED to
 202                  * entry with positive index. We'll go through
 203                  * base->cache[] later to delete all entries in base
 204                  * that are not marked with either CE_MATCHED or
 205                  * CE_UPDATE_IN_BASE. If istate->cache[i] is a
 206                  * duplicate, deduplicate it.
 207                  */
 208                 for (i = 0; i < istate->cache_nr; i++) {
 209                         struct cache_entry *base;
 210                         /* namelen is checked separately */
 211                         const unsigned int ondisk_flags =
 212                                 CE_STAGEMASK | CE_VALID | CE_EXTENDED_FLAGS;
 213                         unsigned int ce_flags, base_flags, ret;
 214                         ce = istate->cache[i];
 215                         if (!ce->index)
 216                                 continue;
 217                         if (ce->index > si->base->cache_nr) {
 218                                 ce->index = 0;
 219                                 continue;
 220                         }
 221                         ce->ce_flags |= CE_MATCHED; /* or "shared" */
 222                         base = si->base->cache[ce->index - 1];
 223                         if (ce == base)
 224                                 continue;
 225                         if (ce->ce_namelen != base->ce_namelen ||
 226                             strcmp(ce->name, base->name)) {
 227                                 ce->index = 0;
 228                                 continue;
 229                         }
 230                         ce_flags = ce->ce_flags;
 231                         base_flags = base->ce_flags;
 232                         /* only on-disk flags matter */
 233                         ce->ce_flags   &= ondisk_flags;
 234                         base->ce_flags &= ondisk_flags;
 235                         ret = memcmp(&ce->ce_stat_data, &base->ce_stat_data,
 236                                      offsetof(struct cache_entry, name) -
 237                                      offsetof(struct cache_entry, ce_stat_data));
 238                         ce->ce_flags = ce_flags;
 239                         base->ce_flags = base_flags;
 240                         if (ret)
 241                                 ce->ce_flags |= CE_UPDATE_IN_BASE;
 242                         discard_cache_entry(base);
 243                         si->base->cache[ce->index - 1] = ce;
 244                 }
 245                 for (i = 0; i < si->base->cache_nr; i++) {
 246                         ce = si->base->cache[i];
 247                         if ((ce->ce_flags & CE_REMOVE) ||
 248                             !(ce->ce_flags & CE_MATCHED))
 249                                 ewah_set(si->delete_bitmap, i);
 250                         else if (ce->ce_flags & CE_UPDATE_IN_BASE) {
 251                                 ewah_set(si->replace_bitmap, i);
 252                                 ce->ce_flags |= CE_STRIP_NAME;
 253                                 ALLOC_GROW(entries, nr_entries+1, nr_alloc);
 254                                 entries[nr_entries++] = ce;
 255                         }
 256                         if (is_null_oid(&ce->oid))
 257                                 istate->drop_cache_tree = 1;
 258                 }
 259         }
 260
 261         for (i = 0; i < istate->cache_nr; i++) {
 262                 ce = istate->cache[i];
 263                 if ((!si->base || !ce->index) && !(ce->ce_flags & CE_REMOVE)) {
 264                         assert(!(ce->ce_flags & CE_STRIP_NAME));
 265                         ALLOC_GROW(entries, nr_entries+1, nr_alloc);
 266                         entries[nr_entries++] = ce;
 267                 }
 268                 ce->ce_flags &= ~CE_MATCHED;
 269         }
 270
 271         /*
 272          * take cache[] out temporarily, put entries[] in its place
 273          * for writing
 274          */
 275         si->saved_cache = istate->cache;
 276         si->saved_cache_nr = istate->cache_nr;
 277         istate->cache = entries;
 278         istate->cache_nr = nr_entries;
 279 }
 280
 281 void finish_writing_split_index(struct index_state *istate)
 282 {
 283         struct split_index *si = init_split_index(istate);
 284
 285         ewah_free(si->delete_bitmap);
 286         ewah_free(si->replace_bitmap);
 287         si->delete_bitmap = NULL;
 288         si->replace_bitmap = NULL;
 289         free(istate->cache);
 290         istate->cache = si->saved_cache;
 291         istate->cache_nr = si->saved_cache_nr;
 292 }
 293
 294 void discard_split_index(struct index_state *istate)
 295 {
 296         struct split_index *si = istate->split_index;
 297         if (!si)
 298                 return;
 299         istate->split_index = NULL;
 300         si->refcount--;
 301         if (si->refcount)
 302                 return;
 303         if (si->base) {
 304                 discard_index(si->base);
 305                 free(si->base);
 306         }
 307         free(si);
 308 }
 309
 310 void save_or_free_index_entry(struct index_state *istate, struct cache_entry *ce)
 311 {
 312         if (ce->index &&
 313             istate->split_index &&
 314             istate->split_index->base &&
 315             ce->index <= istate->split_index->base->cache_nr &&
 316             ce == istate->split_index->base->cache[ce->index - 1])
 317                 ce->ce_flags |= CE_REMOVE;
 318         else
 319                 discard_cache_entry(ce);
 320 }
 321
 322 void replace_index_entry_in_base(struct index_state *istate,
 323                                  struct cache_entry *old_entry,
 324                                  struct cache_entry *new_entry)
 325 {
 326         if (old_entry->index &&
 327             istate->split_index &&
 328             istate->split_index->base &&
 329             old_entry->index <= istate->split_index->base->cache_nr) {
 330                 new_entry->index = old_entry->index;
 331                 if (old_entry != istate->split_index->base->cache[new_entry->index - 1])
 332                         discard_cache_entry(istate->split_index->base->cache[new_entry->index - 1]);
 333                 istate->split_index->base->cache[new_entry->index - 1] = new_entry;
 334         }
 335 }
 336
 337 void add_split_index(struct index_state *istate)
 338 {
 339         if (!istate->split_index) {
 340                 init_split_index(istate);
 341                 istate->cache_changed |= SPLIT_INDEX_ORDERED;
 342         }
 343 }
 344
 345 void remove_split_index(struct index_state *istate)
 346 {
 347         if (istate->split_index) {
 348                 /*
 349                  * When removing the split index, we need to move
 350                  * ownership of the mem_pool associated with the
 351                  * base index to the main index. There may be cache entries
 352                  * allocated from the base's memory pool that are shared with
 353                  * the_index.cache[].
 354                  */
 355                 mem_pool_combine(istate->ce_mem_pool, istate->split_index->base->ce_mem_pool);
 356
 357                 /*
 358                  * The split index no longer owns the mem_pool backing
 359                  * its cache array. As we are discarding this index,
 360                  * mark the index as having no cache entries, so it
 361                  * will not attempt to clean up the cache entries or
 362                  * validate them.
 363                  */
 364                 if (istate->split_index->base)
 365                         istate->split_index->base->cache_nr = 0;
 366
 367                 /*
 368                  * We can discard the split index because its
 369                  * memory pool has been incorporated into the
 370                  * memory pool associated with the the_index.
 371                  */
 372                 discard_split_index(istate);
 373
 374                 istate->cache_changed |= SOMETHING_CHANGED;
 375         }
 376 }