split-index.con commit p5551: add a script to test fetch pack-dir rescans (7893bf1)
   1#include "cache.h"
   2#include "split-index.h"
   3#include "ewah/ewok.h"
   4
   5struct 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
  14int 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 < 20)
  22                return error("corrupt link extension (too short)");
  23        si = init_split_index(istate);
  24        hashcpy(si->base_sha1, data);
  25        data += 20;
  26        sz -= 20;
  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
  44int 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_sha1, 20);
  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
  56static 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
  70void move_cache_to_base_index(struct index_state *istate)
  71{
  72        struct split_index *si = istate->split_index;
  73        int i;
  74
  75        /*
  76         * do not delete old si->base, its index entries may be shared
  77         * with istate->cache[]. Accept a bit of leaking here because
  78         * this code is only used by short-lived update-index.
  79         */
  80        si->base = xcalloc(1, sizeof(*si->base));
  81        si->base->version = istate->version;
  82        /* zero timestamp disables racy test in ce_write_index() */
  83        si->base->timestamp = istate->timestamp;
  84        ALLOC_GROW(si->base->cache, istate->cache_nr, si->base->cache_alloc);
  85        si->base->cache_nr = istate->cache_nr;
  86        COPY_ARRAY(si->base->cache, istate->cache, istate->cache_nr);
  87        mark_base_index_entries(si->base);
  88        for (i = 0; i < si->base->cache_nr; i++)
  89                si->base->cache[i]->ce_flags &= ~CE_UPDATE_IN_BASE;
  90}
  91
  92static void mark_entry_for_delete(size_t pos, void *data)
  93{
  94        struct index_state *istate = data;
  95        if (pos >= istate->cache_nr)
  96                die("position for delete %d exceeds base index size %d",
  97                    (int)pos, istate->cache_nr);
  98        istate->cache[pos]->ce_flags |= CE_REMOVE;
  99        istate->split_index->nr_deletions = 1;
 100}
 101
 102static void replace_entry(size_t pos, void *data)
 103{
 104        struct index_state *istate = data;
 105        struct split_index *si = istate->split_index;
 106        struct cache_entry *dst, *src;
 107
 108        if (pos >= istate->cache_nr)
 109                die("position for replacement %d exceeds base index size %d",
 110                    (int)pos, istate->cache_nr);
 111        if (si->nr_replacements >= si->saved_cache_nr)
 112                die("too many replacements (%d vs %d)",
 113                    si->nr_replacements, si->saved_cache_nr);
 114        dst = istate->cache[pos];
 115        if (dst->ce_flags & CE_REMOVE)
 116                die("entry %d is marked as both replaced and deleted",
 117                    (int)pos);
 118        src = si->saved_cache[si->nr_replacements];
 119        if (ce_namelen(src))
 120                die("corrupt link extension, entry %d should have "
 121                    "zero length name", (int)pos);
 122        src->index = pos + 1;
 123        src->ce_flags |= CE_UPDATE_IN_BASE;
 124        src->ce_namelen = dst->ce_namelen;
 125        copy_cache_entry(dst, src);
 126        free(src);
 127        si->nr_replacements++;
 128}
 129
 130void merge_base_index(struct index_state *istate)
 131{
 132        struct split_index *si = istate->split_index;
 133        unsigned int i;
 134
 135        mark_base_index_entries(si->base);
 136
 137        si->saved_cache     = istate->cache;
 138        si->saved_cache_nr  = istate->cache_nr;
 139        istate->cache_nr    = si->base->cache_nr;
 140        istate->cache       = NULL;
 141        istate->cache_alloc = 0;
 142        ALLOC_GROW(istate->cache, istate->cache_nr, istate->cache_alloc);
 143        COPY_ARRAY(istate->cache, si->base->cache, istate->cache_nr);
 144
 145        si->nr_deletions = 0;
 146        si->nr_replacements = 0;
 147        ewah_each_bit(si->replace_bitmap, replace_entry, istate);
 148        ewah_each_bit(si->delete_bitmap, mark_entry_for_delete, istate);
 149        if (si->nr_deletions)
 150                remove_marked_cache_entries(istate);
 151
 152        for (i = si->nr_replacements; i < si->saved_cache_nr; i++) {
 153                if (!ce_namelen(si->saved_cache[i]))
 154                        die("corrupt link extension, entry %d should "
 155                            "have non-zero length name", i);
 156                add_index_entry(istate, si->saved_cache[i],
 157                                ADD_CACHE_OK_TO_ADD |
 158                                ADD_CACHE_KEEP_CACHE_TREE |
 159                                /*
 160                                 * we may have to replay what
 161                                 * merge-recursive.c:update_stages()
 162                                 * does, which has this flag on
 163                                 */
 164                                ADD_CACHE_SKIP_DFCHECK);
 165                si->saved_cache[i] = NULL;
 166        }
 167
 168        ewah_free(si->delete_bitmap);
 169        ewah_free(si->replace_bitmap);
 170        FREE_AND_NULL(si->saved_cache);
 171        si->delete_bitmap  = NULL;
 172        si->replace_bitmap = NULL;
 173        si->saved_cache_nr = 0;
 174}
 175
 176void prepare_to_write_split_index(struct index_state *istate)
 177{
 178        struct split_index *si = init_split_index(istate);
 179        struct cache_entry **entries = NULL, *ce;
 180        int i, nr_entries = 0, nr_alloc = 0;
 181
 182        si->delete_bitmap = ewah_new();
 183        si->replace_bitmap = ewah_new();
 184
 185        if (si->base) {
 186                /* Go through istate->cache[] and mark CE_MATCHED to
 187                 * entry with positive index. We'll go through
 188                 * base->cache[] later to delete all entries in base
 189                 * that are not marked with either CE_MATCHED or
 190                 * CE_UPDATE_IN_BASE. If istate->cache[i] is a
 191                 * duplicate, deduplicate it.
 192                 */
 193                for (i = 0; i < istate->cache_nr; i++) {
 194                        struct cache_entry *base;
 195                        /* namelen is checked separately */
 196                        const unsigned int ondisk_flags =
 197                                CE_STAGEMASK | CE_VALID | CE_EXTENDED_FLAGS;
 198                        unsigned int ce_flags, base_flags, ret;
 199                        ce = istate->cache[i];
 200                        if (!ce->index)
 201                                continue;
 202                        if (ce->index > si->base->cache_nr) {
 203                                ce->index = 0;
 204                                continue;
 205                        }
 206                        ce->ce_flags |= CE_MATCHED; /* or "shared" */
 207                        base = si->base->cache[ce->index - 1];
 208                        if (ce == base)
 209                                continue;
 210                        if (ce->ce_namelen != base->ce_namelen ||
 211                            strcmp(ce->name, base->name)) {
 212                                ce->index = 0;
 213                                continue;
 214                        }
 215                        ce_flags = ce->ce_flags;
 216                        base_flags = base->ce_flags;
 217                        /* only on-disk flags matter */
 218                        ce->ce_flags   &= ondisk_flags;
 219                        base->ce_flags &= ondisk_flags;
 220                        ret = memcmp(&ce->ce_stat_data, &base->ce_stat_data,
 221                                     offsetof(struct cache_entry, name) -
 222                                     offsetof(struct cache_entry, ce_stat_data));
 223                        ce->ce_flags = ce_flags;
 224                        base->ce_flags = base_flags;
 225                        if (ret)
 226                                ce->ce_flags |= CE_UPDATE_IN_BASE;
 227                        free(base);
 228                        si->base->cache[ce->index - 1] = ce;
 229                }
 230                for (i = 0; i < si->base->cache_nr; i++) {
 231                        ce = si->base->cache[i];
 232                        if ((ce->ce_flags & CE_REMOVE) ||
 233                            !(ce->ce_flags & CE_MATCHED))
 234                                ewah_set(si->delete_bitmap, i);
 235                        else if (ce->ce_flags & CE_UPDATE_IN_BASE) {
 236                                ewah_set(si->replace_bitmap, i);
 237                                ce->ce_flags |= CE_STRIP_NAME;
 238                                ALLOC_GROW(entries, nr_entries+1, nr_alloc);
 239                                entries[nr_entries++] = ce;
 240                        }
 241                }
 242        }
 243
 244        for (i = 0; i < istate->cache_nr; i++) {
 245                ce = istate->cache[i];
 246                if ((!si->base || !ce->index) && !(ce->ce_flags & CE_REMOVE)) {
 247                        assert(!(ce->ce_flags & CE_STRIP_NAME));
 248                        ALLOC_GROW(entries, nr_entries+1, nr_alloc);
 249                        entries[nr_entries++] = ce;
 250                }
 251                ce->ce_flags &= ~CE_MATCHED;
 252        }
 253
 254        /*
 255         * take cache[] out temporarily, put entries[] in its place
 256         * for writing
 257         */
 258        si->saved_cache = istate->cache;
 259        si->saved_cache_nr = istate->cache_nr;
 260        istate->cache = entries;
 261        istate->cache_nr = nr_entries;
 262}
 263
 264void finish_writing_split_index(struct index_state *istate)
 265{
 266        struct split_index *si = init_split_index(istate);
 267
 268        ewah_free(si->delete_bitmap);
 269        ewah_free(si->replace_bitmap);
 270        si->delete_bitmap = NULL;
 271        si->replace_bitmap = NULL;
 272        free(istate->cache);
 273        istate->cache = si->saved_cache;
 274        istate->cache_nr = si->saved_cache_nr;
 275}
 276
 277void discard_split_index(struct index_state *istate)
 278{
 279        struct split_index *si = istate->split_index;
 280        if (!si)
 281                return;
 282        istate->split_index = NULL;
 283        si->refcount--;
 284        if (si->refcount)
 285                return;
 286        if (si->base) {
 287                discard_index(si->base);
 288                free(si->base);
 289        }
 290        free(si);
 291}
 292
 293void save_or_free_index_entry(struct index_state *istate, struct cache_entry *ce)
 294{
 295        if (ce->index &&
 296            istate->split_index &&
 297            istate->split_index->base &&
 298            ce->index <= istate->split_index->base->cache_nr &&
 299            ce == istate->split_index->base->cache[ce->index - 1])
 300                ce->ce_flags |= CE_REMOVE;
 301        else
 302                free(ce);
 303}
 304
 305void replace_index_entry_in_base(struct index_state *istate,
 306                                 struct cache_entry *old,
 307                                 struct cache_entry *new)
 308{
 309        if (old->index &&
 310            istate->split_index &&
 311            istate->split_index->base &&
 312            old->index <= istate->split_index->base->cache_nr) {
 313                new->index = old->index;
 314                if (old != istate->split_index->base->cache[new->index - 1])
 315                        free(istate->split_index->base->cache[new->index - 1]);
 316                istate->split_index->base->cache[new->index - 1] = new;
 317        }
 318}
 319
 320void add_split_index(struct index_state *istate)
 321{
 322        if (!istate->split_index) {
 323                init_split_index(istate);
 324                istate->cache_changed |= SPLIT_INDEX_ORDERED;
 325        }
 326}
 327
 328void remove_split_index(struct index_state *istate)
 329{
 330        if (istate->split_index) {
 331                /*
 332                 * can't discard_split_index(&the_index); because that
 333                 * will destroy split_index->base->cache[], which may
 334                 * be shared with the_index.cache[]. So yeah we're
 335                 * leaking a bit here.
 336                 */
 337                istate->split_index = NULL;
 338                istate->cache_changed |= SOMETHING_CHANGED;
 339        }
 340}