pack-objects.con commit ref-filter.c: find disjoint pattern prefixes (b31e268)
   1#include "cache.h"
   2#include "object.h"
   3#include "pack.h"
   4#include "pack-objects.h"
   5#include "packfile.h"
   6#include "config.h"
   7
   8static uint32_t locate_object_entry_hash(struct packing_data *pdata,
   9                                         const unsigned char *sha1,
  10                                         int *found)
  11{
  12        uint32_t i, mask = (pdata->index_size - 1);
  13
  14        i = sha1hash(sha1) & mask;
  15
  16        while (pdata->index[i] > 0) {
  17                uint32_t pos = pdata->index[i] - 1;
  18
  19                if (hasheq(sha1, pdata->objects[pos].idx.oid.hash)) {
  20                        *found = 1;
  21                        return i;
  22                }
  23
  24                i = (i + 1) & mask;
  25        }
  26
  27        *found = 0;
  28        return i;
  29}
  30
  31static inline uint32_t closest_pow2(uint32_t v)
  32{
  33        v = v - 1;
  34        v |= v >> 1;
  35        v |= v >> 2;
  36        v |= v >> 4;
  37        v |= v >> 8;
  38        v |= v >> 16;
  39        return v + 1;
  40}
  41
  42static void rehash_objects(struct packing_data *pdata)
  43{
  44        uint32_t i;
  45        struct object_entry *entry;
  46
  47        pdata->index_size = closest_pow2(pdata->nr_objects * 3);
  48        if (pdata->index_size < 1024)
  49                pdata->index_size = 1024;
  50
  51        free(pdata->index);
  52        pdata->index = xcalloc(pdata->index_size, sizeof(*pdata->index));
  53
  54        entry = pdata->objects;
  55
  56        for (i = 0; i < pdata->nr_objects; i++) {
  57                int found;
  58                uint32_t ix = locate_object_entry_hash(pdata,
  59                                                       entry->idx.oid.hash,
  60                                                       &found);
  61
  62                if (found)
  63                        BUG("Duplicate object in hash");
  64
  65                pdata->index[ix] = i + 1;
  66                entry++;
  67        }
  68}
  69
  70struct object_entry *packlist_find(struct packing_data *pdata,
  71                                   const unsigned char *sha1,
  72                                   uint32_t *index_pos)
  73{
  74        uint32_t i;
  75        int found;
  76
  77        if (!pdata->index_size)
  78                return NULL;
  79
  80        i = locate_object_entry_hash(pdata, sha1, &found);
  81
  82        if (index_pos)
  83                *index_pos = i;
  84
  85        if (!found)
  86                return NULL;
  87
  88        return &pdata->objects[pdata->index[i] - 1];
  89}
  90
  91static void prepare_in_pack_by_idx(struct packing_data *pdata)
  92{
  93        struct packed_git **mapping, *p;
  94        int cnt = 0, nr = 1U << OE_IN_PACK_BITS;
  95
  96        ALLOC_ARRAY(mapping, nr);
  97        /*
  98         * oe_in_pack() on an all-zero'd object_entry
  99         * (i.e. in_pack_idx also zero) should return NULL.
 100         */
 101        mapping[cnt++] = NULL;
 102        for (p = get_all_packs(pdata->repo); p; p = p->next, cnt++) {
 103                if (cnt == nr) {
 104                        free(mapping);
 105                        return;
 106                }
 107                p->index = cnt;
 108                mapping[cnt] = p;
 109        }
 110        pdata->in_pack_by_idx = mapping;
 111}
 112
 113/*
 114 * A new pack appears after prepare_in_pack_by_idx() has been
 115 * run. This is likely a race.
 116 *
 117 * We could map this new pack to in_pack_by_idx[] array, but then we
 118 * have to deal with full array anyway. And since it's hard to test
 119 * this fall back code, just stay simple and fall back to using
 120 * in_pack[] array.
 121 */
 122void oe_map_new_pack(struct packing_data *pack)
 123{
 124        uint32_t i;
 125
 126        REALLOC_ARRAY(pack->in_pack, pack->nr_alloc);
 127
 128        for (i = 0; i < pack->nr_objects; i++)
 129                pack->in_pack[i] = oe_in_pack(pack, pack->objects + i);
 130
 131        FREE_AND_NULL(pack->in_pack_by_idx);
 132}
 133
 134/* assume pdata is already zero'd by caller */
 135void prepare_packing_data(struct repository *r, struct packing_data *pdata)
 136{
 137        pdata->repo = r;
 138
 139        if (git_env_bool("GIT_TEST_FULL_IN_PACK_ARRAY", 0)) {
 140                /*
 141                 * do not initialize in_pack_by_idx[] to force the
 142                 * slow path in oe_in_pack()
 143                 */
 144        } else {
 145                prepare_in_pack_by_idx(pdata);
 146        }
 147
 148        pdata->oe_size_limit = git_env_ulong("GIT_TEST_OE_SIZE",
 149                                             1U << OE_SIZE_BITS);
 150        pdata->oe_delta_size_limit = git_env_ulong("GIT_TEST_OE_DELTA_SIZE",
 151                                                   1UL << OE_DELTA_SIZE_BITS);
 152        init_recursive_mutex(&pdata->odb_lock);
 153}
 154
 155struct object_entry *packlist_alloc(struct packing_data *pdata,
 156                                    const unsigned char *sha1,
 157                                    uint32_t index_pos)
 158{
 159        struct object_entry *new_entry;
 160
 161        if (pdata->nr_objects >= pdata->nr_alloc) {
 162                pdata->nr_alloc = (pdata->nr_alloc  + 1024) * 3 / 2;
 163                REALLOC_ARRAY(pdata->objects, pdata->nr_alloc);
 164
 165                if (!pdata->in_pack_by_idx)
 166                        REALLOC_ARRAY(pdata->in_pack, pdata->nr_alloc);
 167                if (pdata->delta_size)
 168                        REALLOC_ARRAY(pdata->delta_size, pdata->nr_alloc);
 169
 170                if (pdata->tree_depth)
 171                        REALLOC_ARRAY(pdata->tree_depth, pdata->nr_alloc);
 172
 173                if (pdata->layer)
 174                        REALLOC_ARRAY(pdata->layer, pdata->nr_alloc);
 175        }
 176
 177        new_entry = pdata->objects + pdata->nr_objects++;
 178
 179        memset(new_entry, 0, sizeof(*new_entry));
 180        hashcpy(new_entry->idx.oid.hash, sha1);
 181
 182        if (pdata->index_size * 3 <= pdata->nr_objects * 4)
 183                rehash_objects(pdata);
 184        else
 185                pdata->index[index_pos] = pdata->nr_objects;
 186
 187        if (pdata->in_pack)
 188                pdata->in_pack[pdata->nr_objects - 1] = NULL;
 189
 190        if (pdata->tree_depth)
 191                pdata->tree_depth[pdata->nr_objects - 1] = 0;
 192
 193        if (pdata->layer)
 194                pdata->layer[pdata->nr_objects - 1] = 0;
 195
 196        return new_entry;
 197}
 198
 199void oe_set_delta_ext(struct packing_data *pdata,
 200                      struct object_entry *delta,
 201                      const unsigned char *sha1)
 202{
 203        struct object_entry *base;
 204
 205        ALLOC_GROW(pdata->ext_bases, pdata->nr_ext + 1, pdata->alloc_ext);
 206        base = &pdata->ext_bases[pdata->nr_ext++];
 207        memset(base, 0, sizeof(*base));
 208        hashcpy(base->idx.oid.hash, sha1);
 209
 210        /* These flags mark that we are not part of the actual pack output. */
 211        base->preferred_base = 1;
 212        base->filled = 1;
 213
 214        delta->ext_base = 1;
 215        delta->delta_idx = base - pdata->ext_bases + 1;
 216}