git.oblomov.eu Git - git/blob - pack-objects.c

   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
   8 static 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 (!hashcmp(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
  31 static 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
  42 static 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
  70 struct 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
  91 static 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_packed_git(the_repository); 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  */
 122 void oe_map_new_pack(struct packing_data *pack,
 123                      struct packed_git *p)
 124 {
 125         uint32_t i;
 126
 127         REALLOC_ARRAY(pack->in_pack, pack->nr_alloc);
 128
 129         for (i = 0; i < pack->nr_objects; i++)
 130                 pack->in_pack[i] = oe_in_pack(pack, pack->objects + i);
 131
 132         FREE_AND_NULL(pack->in_pack_by_idx);
 133 }
 134
 135 /* assume pdata is already zero'd by caller */
 136 void prepare_packing_data(struct packing_data *pdata)
 137 {
 138         if (git_env_bool("GIT_TEST_FULL_IN_PACK_ARRAY", 0)) {
 139                 /*
 140                  * do not initialize in_pack_by_idx[] to force the
 141                  * slow path in oe_in_pack()
 142                  */
 143         } else {
 144                 prepare_in_pack_by_idx(pdata);
 145         }
 146
 147         pdata->oe_size_limit = git_env_ulong("GIT_TEST_OE_SIZE",
 148                                              1U << OE_SIZE_BITS);
 149         pdata->oe_delta_size_limit = git_env_ulong("GIT_TEST_OE_DELTA_SIZE",
 150                                                    1UL << OE_DELTA_SIZE_BITS);
 151 #ifndef NO_PTHREADS
 152         pthread_mutex_init(&pdata->lock, NULL);
 153 #endif
 154 }
 155
 156 struct object_entry *packlist_alloc(struct packing_data *pdata,
 157                                     const unsigned char *sha1,
 158                                     uint32_t index_pos)
 159 {
 160         struct object_entry *new_entry;
 161
 162         if (pdata->nr_objects >= pdata->nr_alloc) {
 163                 pdata->nr_alloc = (pdata->nr_alloc  + 1024) * 3 / 2;
 164                 REALLOC_ARRAY(pdata->objects, pdata->nr_alloc);
 165
 166                 if (!pdata->in_pack_by_idx)
 167                         REALLOC_ARRAY(pdata->in_pack, pdata->nr_alloc);
 168                 if (pdata->delta_size)
 169                         REALLOC_ARRAY(pdata->delta_size, pdata->nr_alloc);
 170         }
 171
 172         new_entry = pdata->objects + pdata->nr_objects++;
 173
 174         memset(new_entry, 0, sizeof(*new_entry));
 175         hashcpy(new_entry->idx.oid.hash, sha1);
 176
 177         if (pdata->index_size * 3 <= pdata->nr_objects * 4)
 178                 rehash_objects(pdata);
 179         else
 180                 pdata->index[index_pos] = pdata->nr_objects;
 181
 182         if (pdata->in_pack)
 183                 pdata->in_pack[pdata->nr_objects - 1] = NULL;
 184
 185         return new_entry;
 186 }