Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 1 | #include "cache.h" |
| 2 | #include "commit.h" |
| 3 | #include "tag.h" |
| 4 | #include "diff.h" |
| 5 | #include "revision.h" |
| 6 | #include "progress.h" |
| 7 | #include "list-objects.h" |
| 8 | #include "pack.h" |
| 9 | #include "pack-bitmap.h" |
| 10 | #include "pack-revindex.h" |
| 11 | #include "pack-objects.h" |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 12 | #include "packfile.h" |
Stefan Beller | a80d72d | 2018-03-23 18:20:59 +0100 | [diff] [blame] | 13 | #include "repository.h" |
| 14 | #include "object-store.h" |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 15 | |
| 16 | /* |
| 17 | * An entry on the bitmap index, representing the bitmap for a given |
| 18 | * commit. |
| 19 | */ |
| 20 | struct stored_bitmap { |
brian m. carlson | 5363653 | 2019-02-19 00:04:55 +0000 | [diff] [blame] | 21 | struct object_id oid; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 22 | struct ewah_bitmap *root; |
| 23 | struct stored_bitmap *xor; |
| 24 | int flags; |
| 25 | }; |
| 26 | |
| 27 | /* |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 28 | * The active bitmap index for a repository. By design, repositories only have |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 29 | * a single bitmap index available (the index for the biggest packfile in |
| 30 | * the repository), since bitmap indexes need full closure. |
| 31 | * |
| 32 | * If there is more than one bitmap index available (e.g. because of alternates), |
| 33 | * the active bitmap index is the largest one. |
| 34 | */ |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 35 | struct bitmap_index { |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 36 | /* Packfile to which this bitmap index belongs to */ |
| 37 | struct packed_git *pack; |
| 38 | |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 39 | /* |
| 40 | * Mark the first `reuse_objects` in the packfile as reused: |
| 41 | * they will be sent as-is without using them for repacking |
| 42 | * calculations |
| 43 | */ |
| 44 | uint32_t reuse_objects; |
| 45 | |
| 46 | /* mmapped buffer of the whole bitmap index */ |
| 47 | unsigned char *map; |
| 48 | size_t map_size; /* size of the mmaped buffer */ |
| 49 | size_t map_pos; /* current position when loading the index */ |
| 50 | |
| 51 | /* |
| 52 | * Type indexes. |
| 53 | * |
| 54 | * Each bitmap marks which objects in the packfile are of the given |
| 55 | * type. This provides type information when yielding the objects from |
| 56 | * the packfile during a walk, which allows for better delta bases. |
| 57 | */ |
| 58 | struct ewah_bitmap *commits; |
| 59 | struct ewah_bitmap *trees; |
| 60 | struct ewah_bitmap *blobs; |
| 61 | struct ewah_bitmap *tags; |
| 62 | |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 63 | /* Map from object ID -> `stored_bitmap` for all the bitmapped commits */ |
| 64 | kh_oid_map_t *bitmaps; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 65 | |
| 66 | /* Number of bitmapped commits */ |
| 67 | uint32_t entry_count; |
| 68 | |
Jonathan Tan | f3c23db | 2018-06-07 12:04:14 -0700 | [diff] [blame] | 69 | /* If not NULL, this is a name-hash cache pointing into map. */ |
Vicent Marti | ae4f07f | 2013-12-21 09:00:45 -0500 | [diff] [blame] | 70 | uint32_t *hashes; |
| 71 | |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 72 | /* |
| 73 | * Extended index. |
| 74 | * |
| 75 | * When trying to perform bitmap operations with objects that are not |
| 76 | * packed in `pack`, these objects are added to this "fake index" and |
| 77 | * are assumed to appear at the end of the packfile for all operations |
| 78 | */ |
| 79 | struct eindex { |
| 80 | struct object **objects; |
| 81 | uint32_t *hashes; |
| 82 | uint32_t count, alloc; |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 83 | kh_oid_pos_t *positions; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 84 | } ext_index; |
| 85 | |
| 86 | /* Bitmap result of the last performed walk */ |
| 87 | struct bitmap *result; |
| 88 | |
Jeff King | 30cdc33 | 2018-08-21 15:07:01 -0400 | [diff] [blame] | 89 | /* "have" bitmap from the last performed walk */ |
| 90 | struct bitmap *haves; |
| 91 | |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 92 | /* Version of the bitmap index */ |
| 93 | unsigned int version; |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 94 | }; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 95 | |
| 96 | static struct ewah_bitmap *lookup_stored_bitmap(struct stored_bitmap *st) |
| 97 | { |
| 98 | struct ewah_bitmap *parent; |
| 99 | struct ewah_bitmap *composed; |
| 100 | |
| 101 | if (st->xor == NULL) |
| 102 | return st->root; |
| 103 | |
| 104 | composed = ewah_pool_new(); |
| 105 | parent = lookup_stored_bitmap(st->xor); |
| 106 | ewah_xor(st->root, parent, composed); |
| 107 | |
| 108 | ewah_pool_free(st->root); |
| 109 | st->root = composed; |
| 110 | st->xor = NULL; |
| 111 | |
| 112 | return composed; |
| 113 | } |
| 114 | |
| 115 | /* |
| 116 | * Read a bitmap from the current read position on the mmaped |
| 117 | * index, and increase the read position accordingly |
| 118 | */ |
| 119 | static struct ewah_bitmap *read_bitmap_1(struct bitmap_index *index) |
| 120 | { |
| 121 | struct ewah_bitmap *b = ewah_pool_new(); |
| 122 | |
Jeff King | 1140bf0 | 2018-06-14 23:44:43 -0400 | [diff] [blame] | 123 | ssize_t bitmap_size = ewah_read_mmap(b, |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 124 | index->map + index->map_pos, |
| 125 | index->map_size - index->map_pos); |
| 126 | |
| 127 | if (bitmap_size < 0) { |
| 128 | error("Failed to load bitmap index (corrupted?)"); |
| 129 | ewah_pool_free(b); |
| 130 | return NULL; |
| 131 | } |
| 132 | |
| 133 | index->map_pos += bitmap_size; |
| 134 | return b; |
| 135 | } |
| 136 | |
| 137 | static int load_bitmap_header(struct bitmap_index *index) |
| 138 | { |
| 139 | struct bitmap_disk_header *header = (void *)index->map; |
| 140 | |
brian m. carlson | 9941e92 | 2019-02-19 00:04:57 +0000 | [diff] [blame] | 141 | if (index->map_size < sizeof(*header) + the_hash_algo->rawsz) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 142 | return error("Corrupted bitmap index (missing header data)"); |
| 143 | |
| 144 | if (memcmp(header->magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE)) != 0) |
| 145 | return error("Corrupted bitmap index file (wrong header)"); |
| 146 | |
| 147 | index->version = ntohs(header->version); |
| 148 | if (index->version != 1) |
| 149 | return error("Unsupported version for bitmap index file (%d)", index->version); |
| 150 | |
| 151 | /* Parse known bitmap format options */ |
| 152 | { |
| 153 | uint32_t flags = ntohs(header->options); |
| 154 | |
| 155 | if ((flags & BITMAP_OPT_FULL_DAG) == 0) |
| 156 | return error("Unsupported options for bitmap index file " |
| 157 | "(Git requires BITMAP_OPT_FULL_DAG)"); |
Vicent Marti | ae4f07f | 2013-12-21 09:00:45 -0500 | [diff] [blame] | 158 | |
| 159 | if (flags & BITMAP_OPT_HASH_CACHE) { |
brian m. carlson | 9941e92 | 2019-02-19 00:04:57 +0000 | [diff] [blame] | 160 | unsigned char *end = index->map + index->map_size - the_hash_algo->rawsz; |
Vicent Marti | ae4f07f | 2013-12-21 09:00:45 -0500 | [diff] [blame] | 161 | index->hashes = ((uint32_t *)end) - index->pack->num_objects; |
| 162 | } |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 163 | } |
| 164 | |
| 165 | index->entry_count = ntohl(header->entry_count); |
brian m. carlson | 0f4d6ca | 2019-02-19 00:04:54 +0000 | [diff] [blame] | 166 | index->map_pos += sizeof(*header) - GIT_MAX_RAWSZ + the_hash_algo->rawsz; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 167 | return 0; |
| 168 | } |
| 169 | |
| 170 | static struct stored_bitmap *store_bitmap(struct bitmap_index *index, |
| 171 | struct ewah_bitmap *root, |
brian m. carlson | 0dd4924 | 2019-02-19 00:04:56 +0000 | [diff] [blame] | 172 | const unsigned char *hash, |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 173 | struct stored_bitmap *xor_with, |
| 174 | int flags) |
| 175 | { |
| 176 | struct stored_bitmap *stored; |
| 177 | khiter_t hash_pos; |
| 178 | int ret; |
| 179 | |
| 180 | stored = xmalloc(sizeof(struct stored_bitmap)); |
| 181 | stored->root = root; |
| 182 | stored->xor = xor_with; |
| 183 | stored->flags = flags; |
brian m. carlson | 0dd4924 | 2019-02-19 00:04:56 +0000 | [diff] [blame] | 184 | oidread(&stored->oid, hash); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 185 | |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 186 | hash_pos = kh_put_oid_map(index->bitmaps, stored->oid, &ret); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 187 | |
| 188 | /* a 0 return code means the insertion succeeded with no changes, |
| 189 | * because the SHA1 already existed on the map. this is bad, there |
| 190 | * shouldn't be duplicated commits in the index */ |
| 191 | if (ret == 0) { |
brian m. carlson | 0dd4924 | 2019-02-19 00:04:56 +0000 | [diff] [blame] | 192 | error("Duplicate entry in bitmap index: %s", hash_to_hex(hash)); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 193 | return NULL; |
| 194 | } |
| 195 | |
| 196 | kh_value(index->bitmaps, hash_pos) = stored; |
| 197 | return stored; |
| 198 | } |
| 199 | |
Karsten Blees | b500721 | 2014-11-27 00:24:01 -0500 | [diff] [blame] | 200 | static inline uint32_t read_be32(const unsigned char *buffer, size_t *pos) |
| 201 | { |
| 202 | uint32_t result = get_be32(buffer + *pos); |
| 203 | (*pos) += sizeof(result); |
| 204 | return result; |
| 205 | } |
| 206 | |
| 207 | static inline uint8_t read_u8(const unsigned char *buffer, size_t *pos) |
| 208 | { |
| 209 | return buffer[(*pos)++]; |
| 210 | } |
| 211 | |
René Scharfe | 599dc76 | 2015-05-19 01:24:09 +0200 | [diff] [blame] | 212 | #define MAX_XOR_OFFSET 160 |
| 213 | |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 214 | static int load_bitmap_entries_v1(struct bitmap_index *index) |
| 215 | { |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 216 | uint32_t i; |
René Scharfe | 599dc76 | 2015-05-19 01:24:09 +0200 | [diff] [blame] | 217 | struct stored_bitmap *recent_bitmaps[MAX_XOR_OFFSET] = { NULL }; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 218 | |
| 219 | for (i = 0; i < index->entry_count; ++i) { |
| 220 | int xor_offset, flags; |
| 221 | struct ewah_bitmap *bitmap = NULL; |
| 222 | struct stored_bitmap *xor_bitmap = NULL; |
| 223 | uint32_t commit_idx_pos; |
| 224 | const unsigned char *sha1; |
| 225 | |
Karsten Blees | b500721 | 2014-11-27 00:24:01 -0500 | [diff] [blame] | 226 | commit_idx_pos = read_be32(index->map, &index->map_pos); |
| 227 | xor_offset = read_u8(index->map, &index->map_pos); |
| 228 | flags = read_u8(index->map, &index->map_pos); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 229 | |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 230 | sha1 = nth_packed_object_sha1(index->pack, commit_idx_pos); |
| 231 | |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 232 | bitmap = read_bitmap_1(index); |
| 233 | if (!bitmap) |
| 234 | return -1; |
| 235 | |
| 236 | if (xor_offset > MAX_XOR_OFFSET || xor_offset > i) |
| 237 | return error("Corrupted bitmap pack index"); |
| 238 | |
| 239 | if (xor_offset > 0) { |
| 240 | xor_bitmap = recent_bitmaps[(i - xor_offset) % MAX_XOR_OFFSET]; |
| 241 | |
| 242 | if (xor_bitmap == NULL) |
| 243 | return error("Invalid XOR offset in bitmap pack index"); |
| 244 | } |
| 245 | |
| 246 | recent_bitmaps[i % MAX_XOR_OFFSET] = store_bitmap( |
| 247 | index, bitmap, sha1, xor_bitmap, flags); |
| 248 | } |
| 249 | |
| 250 | return 0; |
| 251 | } |
| 252 | |
Junio C Hamano | cb46805 | 2015-01-14 15:40:53 -0800 | [diff] [blame] | 253 | static char *pack_bitmap_filename(struct packed_git *p) |
| 254 | { |
Jeff King | 9ae9701 | 2015-09-24 17:07:09 -0400 | [diff] [blame] | 255 | size_t len; |
Junio C Hamano | cb46805 | 2015-01-14 15:40:53 -0800 | [diff] [blame] | 256 | |
Jeff King | 9ae9701 | 2015-09-24 17:07:09 -0400 | [diff] [blame] | 257 | if (!strip_suffix(p->pack_name, ".pack", &len)) |
Johannes Schindelin | 033abf9 | 2018-05-02 11:38:39 +0200 | [diff] [blame] | 258 | BUG("pack_name does not end in .pack"); |
Jeff King | 9ae9701 | 2015-09-24 17:07:09 -0400 | [diff] [blame] | 259 | return xstrfmt("%.*s.bitmap", (int)len, p->pack_name); |
Junio C Hamano | cb46805 | 2015-01-14 15:40:53 -0800 | [diff] [blame] | 260 | } |
| 261 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 262 | static int open_pack_bitmap_1(struct bitmap_index *bitmap_git, struct packed_git *packfile) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 263 | { |
| 264 | int fd; |
| 265 | struct stat st; |
| 266 | char *idx_name; |
| 267 | |
| 268 | if (open_pack_index(packfile)) |
| 269 | return -1; |
| 270 | |
| 271 | idx_name = pack_bitmap_filename(packfile); |
Lars Schneider | a5436b5 | 2016-10-24 20:02:59 +0200 | [diff] [blame] | 272 | fd = git_open(idx_name); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 273 | free(idx_name); |
| 274 | |
| 275 | if (fd < 0) |
| 276 | return -1; |
| 277 | |
| 278 | if (fstat(fd, &st)) { |
| 279 | close(fd); |
| 280 | return -1; |
| 281 | } |
| 282 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 283 | if (bitmap_git->pack) { |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 284 | warning("ignoring extra bitmap file: %s", packfile->pack_name); |
| 285 | close(fd); |
| 286 | return -1; |
| 287 | } |
| 288 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 289 | bitmap_git->pack = packfile; |
| 290 | bitmap_git->map_size = xsize_t(st.st_size); |
| 291 | bitmap_git->map = xmmap(NULL, bitmap_git->map_size, PROT_READ, MAP_PRIVATE, fd, 0); |
| 292 | bitmap_git->map_pos = 0; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 293 | close(fd); |
| 294 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 295 | if (load_bitmap_header(bitmap_git) < 0) { |
| 296 | munmap(bitmap_git->map, bitmap_git->map_size); |
| 297 | bitmap_git->map = NULL; |
| 298 | bitmap_git->map_size = 0; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 299 | return -1; |
| 300 | } |
| 301 | |
| 302 | return 0; |
| 303 | } |
| 304 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 305 | static int load_pack_bitmap(struct bitmap_index *bitmap_git) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 306 | { |
Jeff King | 199c86b | 2018-09-01 03:50:57 -0400 | [diff] [blame] | 307 | assert(bitmap_git->map); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 308 | |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 309 | bitmap_git->bitmaps = kh_init_oid_map(); |
| 310 | bitmap_git->ext_index.positions = kh_init_oid_pos(); |
Jeff King | 4828ce9 | 2019-04-05 14:04:24 -0400 | [diff] [blame] | 311 | if (load_pack_revindex(bitmap_git->pack)) |
| 312 | goto failed; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 313 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 314 | if (!(bitmap_git->commits = read_bitmap_1(bitmap_git)) || |
| 315 | !(bitmap_git->trees = read_bitmap_1(bitmap_git)) || |
| 316 | !(bitmap_git->blobs = read_bitmap_1(bitmap_git)) || |
| 317 | !(bitmap_git->tags = read_bitmap_1(bitmap_git))) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 318 | goto failed; |
| 319 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 320 | if (load_bitmap_entries_v1(bitmap_git) < 0) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 321 | goto failed; |
| 322 | |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 323 | return 0; |
| 324 | |
| 325 | failed: |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 326 | munmap(bitmap_git->map, bitmap_git->map_size); |
| 327 | bitmap_git->map = NULL; |
| 328 | bitmap_git->map_size = 0; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 329 | return -1; |
| 330 | } |
| 331 | |
Nguyễn Thái Ngọc Duy | 7c14112 | 2018-11-10 06:49:08 +0100 | [diff] [blame] | 332 | static int open_pack_bitmap(struct repository *r, |
| 333 | struct bitmap_index *bitmap_git) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 334 | { |
| 335 | struct packed_git *p; |
| 336 | int ret = -1; |
| 337 | |
Jeff King | 199c86b | 2018-09-01 03:50:57 -0400 | [diff] [blame] | 338 | assert(!bitmap_git->map); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 339 | |
Nguyễn Thái Ngọc Duy | 7c14112 | 2018-11-10 06:49:08 +0100 | [diff] [blame] | 340 | for (p = get_all_packs(r); p; p = p->next) { |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 341 | if (open_pack_bitmap_1(bitmap_git, p) == 0) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 342 | ret = 0; |
| 343 | } |
| 344 | |
| 345 | return ret; |
| 346 | } |
| 347 | |
Nguyễn Thái Ngọc Duy | 7c14112 | 2018-11-10 06:49:08 +0100 | [diff] [blame] | 348 | struct bitmap_index *prepare_bitmap_git(struct repository *r) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 349 | { |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 350 | struct bitmap_index *bitmap_git = xcalloc(1, sizeof(*bitmap_git)); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 351 | |
Nguyễn Thái Ngọc Duy | 7c14112 | 2018-11-10 06:49:08 +0100 | [diff] [blame] | 352 | if (!open_pack_bitmap(r, bitmap_git) && !load_pack_bitmap(bitmap_git)) |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 353 | return bitmap_git; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 354 | |
Jonathan Tan | f3c23db | 2018-06-07 12:04:14 -0700 | [diff] [blame] | 355 | free_bitmap_index(bitmap_git); |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 356 | return NULL; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 357 | } |
| 358 | |
| 359 | struct include_data { |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 360 | struct bitmap_index *bitmap_git; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 361 | struct bitmap *base; |
| 362 | struct bitmap *seen; |
| 363 | }; |
| 364 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 365 | static inline int bitmap_position_extended(struct bitmap_index *bitmap_git, |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 366 | const struct object_id *oid) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 367 | { |
Jeff King | 4ed43d1 | 2019-06-20 03:41:25 -0400 | [diff] [blame] | 368 | kh_oid_pos_t *positions = bitmap_git->ext_index.positions; |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 369 | khiter_t pos = kh_get_oid_pos(positions, *oid); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 370 | |
| 371 | if (pos < kh_end(positions)) { |
| 372 | int bitmap_pos = kh_value(positions, pos); |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 373 | return bitmap_pos + bitmap_git->pack->num_objects; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 374 | } |
| 375 | |
| 376 | return -1; |
| 377 | } |
| 378 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 379 | static inline int bitmap_position_packfile(struct bitmap_index *bitmap_git, |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 380 | const struct object_id *oid) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 381 | { |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 382 | off_t offset = find_pack_entry_one(oid->hash, bitmap_git->pack); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 383 | if (!offset) |
| 384 | return -1; |
| 385 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 386 | return find_revindex_position(bitmap_git->pack, offset); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 387 | } |
| 388 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 389 | static int bitmap_position(struct bitmap_index *bitmap_git, |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 390 | const struct object_id *oid) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 391 | { |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 392 | int pos = bitmap_position_packfile(bitmap_git, oid); |
| 393 | return (pos >= 0) ? pos : bitmap_position_extended(bitmap_git, oid); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 394 | } |
| 395 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 396 | static int ext_index_add_object(struct bitmap_index *bitmap_git, |
| 397 | struct object *object, const char *name) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 398 | { |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 399 | struct eindex *eindex = &bitmap_git->ext_index; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 400 | |
| 401 | khiter_t hash_pos; |
| 402 | int hash_ret; |
| 403 | int bitmap_pos; |
| 404 | |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 405 | hash_pos = kh_put_oid_pos(eindex->positions, object->oid, &hash_ret); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 406 | if (hash_ret > 0) { |
| 407 | if (eindex->count >= eindex->alloc) { |
| 408 | eindex->alloc = (eindex->alloc + 16) * 3 / 2; |
René Scharfe | 2756ca4 | 2014-09-16 20:56:57 +0200 | [diff] [blame] | 409 | REALLOC_ARRAY(eindex->objects, eindex->alloc); |
| 410 | REALLOC_ARRAY(eindex->hashes, eindex->alloc); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 411 | } |
| 412 | |
| 413 | bitmap_pos = eindex->count; |
| 414 | eindex->objects[eindex->count] = object; |
| 415 | eindex->hashes[eindex->count] = pack_name_hash(name); |
| 416 | kh_value(eindex->positions, hash_pos) = bitmap_pos; |
| 417 | eindex->count++; |
| 418 | } else { |
| 419 | bitmap_pos = kh_value(eindex->positions, hash_pos); |
| 420 | } |
| 421 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 422 | return bitmap_pos + bitmap_git->pack->num_objects; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 423 | } |
| 424 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 425 | struct bitmap_show_data { |
| 426 | struct bitmap_index *bitmap_git; |
| 427 | struct bitmap *base; |
| 428 | }; |
| 429 | |
| 430 | static void show_object(struct object *object, const char *name, void *data_) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 431 | { |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 432 | struct bitmap_show_data *data = data_; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 433 | int bitmap_pos; |
| 434 | |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 435 | bitmap_pos = bitmap_position(data->bitmap_git, &object->oid); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 436 | |
Jeff King | de1e67d | 2016-02-11 17:28:36 -0500 | [diff] [blame] | 437 | if (bitmap_pos < 0) |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 438 | bitmap_pos = ext_index_add_object(data->bitmap_git, object, |
| 439 | name); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 440 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 441 | bitmap_set(data->base, bitmap_pos); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 442 | } |
| 443 | |
| 444 | static void show_commit(struct commit *commit, void *data) |
| 445 | { |
| 446 | } |
| 447 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 448 | static int add_to_include_set(struct bitmap_index *bitmap_git, |
| 449 | struct include_data *data, |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 450 | const struct object_id *oid, |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 451 | int bitmap_pos) |
| 452 | { |
| 453 | khiter_t hash_pos; |
| 454 | |
| 455 | if (data->seen && bitmap_get(data->seen, bitmap_pos)) |
| 456 | return 0; |
| 457 | |
| 458 | if (bitmap_get(data->base, bitmap_pos)) |
| 459 | return 0; |
| 460 | |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 461 | hash_pos = kh_get_oid_map(bitmap_git->bitmaps, *oid); |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 462 | if (hash_pos < kh_end(bitmap_git->bitmaps)) { |
| 463 | struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, hash_pos); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 464 | bitmap_or_ewah(data->base, lookup_stored_bitmap(st)); |
| 465 | return 0; |
| 466 | } |
| 467 | |
| 468 | bitmap_set(data->base, bitmap_pos); |
| 469 | return 1; |
| 470 | } |
| 471 | |
| 472 | static int should_include(struct commit *commit, void *_data) |
| 473 | { |
| 474 | struct include_data *data = _data; |
| 475 | int bitmap_pos; |
| 476 | |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 477 | bitmap_pos = bitmap_position(data->bitmap_git, &commit->object.oid); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 478 | if (bitmap_pos < 0) |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 479 | bitmap_pos = ext_index_add_object(data->bitmap_git, |
| 480 | (struct object *)commit, |
| 481 | NULL); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 482 | |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 483 | if (!add_to_include_set(data->bitmap_git, data, &commit->object.oid, |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 484 | bitmap_pos)) { |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 485 | struct commit_list *parent = commit->parents; |
| 486 | |
| 487 | while (parent) { |
| 488 | parent->item->object.flags |= SEEN; |
| 489 | parent = parent->next; |
| 490 | } |
| 491 | |
| 492 | return 0; |
| 493 | } |
| 494 | |
| 495 | return 1; |
| 496 | } |
| 497 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 498 | static struct bitmap *find_objects(struct bitmap_index *bitmap_git, |
| 499 | struct rev_info *revs, |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 500 | struct object_list *roots, |
| 501 | struct bitmap *seen) |
| 502 | { |
| 503 | struct bitmap *base = NULL; |
| 504 | int needs_walk = 0; |
| 505 | |
| 506 | struct object_list *not_mapped = NULL; |
| 507 | |
| 508 | /* |
| 509 | * Go through all the roots for the walk. The ones that have bitmaps |
| 510 | * on the bitmap index will be `or`ed together to form an initial |
| 511 | * global reachability analysis. |
| 512 | * |
| 513 | * The ones without bitmaps in the index will be stored in the |
| 514 | * `not_mapped_list` for further processing. |
| 515 | */ |
| 516 | while (roots) { |
| 517 | struct object *object = roots->item; |
| 518 | roots = roots->next; |
| 519 | |
| 520 | if (object->type == OBJ_COMMIT) { |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 521 | khiter_t pos = kh_get_oid_map(bitmap_git->bitmaps, object->oid); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 522 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 523 | if (pos < kh_end(bitmap_git->bitmaps)) { |
| 524 | struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, pos); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 525 | struct ewah_bitmap *or_with = lookup_stored_bitmap(st); |
| 526 | |
| 527 | if (base == NULL) |
| 528 | base = ewah_to_bitmap(or_with); |
| 529 | else |
| 530 | bitmap_or_ewah(base, or_with); |
| 531 | |
| 532 | object->flags |= SEEN; |
| 533 | continue; |
| 534 | } |
| 535 | } |
| 536 | |
| 537 | object_list_insert(object, ¬_mapped); |
| 538 | } |
| 539 | |
| 540 | /* |
| 541 | * Best case scenario: We found bitmaps for all the roots, |
| 542 | * so the resulting `or` bitmap has the full reachability analysis |
| 543 | */ |
| 544 | if (not_mapped == NULL) |
| 545 | return base; |
| 546 | |
| 547 | roots = not_mapped; |
| 548 | |
| 549 | /* |
| 550 | * Let's iterate through all the roots that don't have bitmaps to |
| 551 | * check if we can determine them to be reachable from the existing |
| 552 | * global bitmap. |
| 553 | * |
| 554 | * If we cannot find them in the existing global bitmap, we'll need |
| 555 | * to push them to an actual walk and run it until we can confirm |
| 556 | * they are reachable |
| 557 | */ |
| 558 | while (roots) { |
| 559 | struct object *object = roots->item; |
| 560 | int pos; |
| 561 | |
| 562 | roots = roots->next; |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 563 | pos = bitmap_position(bitmap_git, &object->oid); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 564 | |
| 565 | if (pos < 0 || base == NULL || !bitmap_get(base, pos)) { |
| 566 | object->flags &= ~UNINTERESTING; |
| 567 | add_pending_object(revs, object, ""); |
| 568 | needs_walk = 1; |
| 569 | } else { |
| 570 | object->flags |= SEEN; |
| 571 | } |
| 572 | } |
| 573 | |
| 574 | if (needs_walk) { |
| 575 | struct include_data incdata; |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 576 | struct bitmap_show_data show_data; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 577 | |
| 578 | if (base == NULL) |
| 579 | base = bitmap_new(); |
| 580 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 581 | incdata.bitmap_git = bitmap_git; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 582 | incdata.base = base; |
| 583 | incdata.seen = seen; |
| 584 | |
| 585 | revs->include_check = should_include; |
| 586 | revs->include_check_data = &incdata; |
| 587 | |
| 588 | if (prepare_revision_walk(revs)) |
| 589 | die("revision walk setup failed"); |
| 590 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 591 | show_data.bitmap_git = bitmap_git; |
| 592 | show_data.base = base; |
| 593 | |
| 594 | traverse_commit_list(revs, show_commit, show_object, |
| 595 | &show_data); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 596 | } |
| 597 | |
| 598 | return base; |
| 599 | } |
| 600 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 601 | static void show_extended_objects(struct bitmap_index *bitmap_git, |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 602 | show_reachable_fn show_reach) |
| 603 | { |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 604 | struct bitmap *objects = bitmap_git->result; |
| 605 | struct eindex *eindex = &bitmap_git->ext_index; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 606 | uint32_t i; |
| 607 | |
| 608 | for (i = 0; i < eindex->count; ++i) { |
| 609 | struct object *obj; |
| 610 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 611 | if (!bitmap_get(objects, bitmap_git->pack->num_objects + i)) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 612 | continue; |
| 613 | |
| 614 | obj = eindex->objects[i]; |
brian m. carlson | 2066496 | 2017-10-15 22:07:00 +0000 | [diff] [blame] | 615 | show_reach(&obj->oid, obj->type, 0, eindex->hashes[i], NULL, 0); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 616 | } |
| 617 | } |
| 618 | |
| 619 | static void show_objects_for_type( |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 620 | struct bitmap_index *bitmap_git, |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 621 | struct ewah_bitmap *type_filter, |
| 622 | enum object_type object_type, |
| 623 | show_reachable_fn show_reach) |
| 624 | { |
| 625 | size_t pos = 0, i = 0; |
| 626 | uint32_t offset; |
| 627 | |
| 628 | struct ewah_iterator it; |
| 629 | eword_t filter; |
| 630 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 631 | struct bitmap *objects = bitmap_git->result; |
| 632 | |
| 633 | if (bitmap_git->reuse_objects == bitmap_git->pack->num_objects) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 634 | return; |
| 635 | |
| 636 | ewah_iterator_init(&it, type_filter); |
| 637 | |
| 638 | while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) { |
| 639 | eword_t word = objects->words[i] & filter; |
| 640 | |
Jeff King | 34b935c | 2015-06-03 02:39:37 -0400 | [diff] [blame] | 641 | for (offset = 0; offset < BITS_IN_EWORD; ++offset) { |
brian m. carlson | 2066496 | 2017-10-15 22:07:00 +0000 | [diff] [blame] | 642 | struct object_id oid; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 643 | struct revindex_entry *entry; |
| 644 | uint32_t hash = 0; |
| 645 | |
| 646 | if ((word >> offset) == 0) |
| 647 | break; |
| 648 | |
| 649 | offset += ewah_bit_ctz64(word >> offset); |
| 650 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 651 | if (pos + offset < bitmap_git->reuse_objects) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 652 | continue; |
| 653 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 654 | entry = &bitmap_git->pack->revindex[pos + offset]; |
| 655 | nth_packed_object_oid(&oid, bitmap_git->pack, entry->nr); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 656 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 657 | if (bitmap_git->hashes) |
| 658 | hash = get_be32(bitmap_git->hashes + entry->nr); |
Vicent Marti | ae4f07f | 2013-12-21 09:00:45 -0500 | [diff] [blame] | 659 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 660 | show_reach(&oid, object_type, 0, hash, bitmap_git->pack, entry->offset); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 661 | } |
| 662 | |
Jeff King | 34b935c | 2015-06-03 02:39:37 -0400 | [diff] [blame] | 663 | pos += BITS_IN_EWORD; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 664 | i++; |
| 665 | } |
| 666 | } |
| 667 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 668 | static int in_bitmapped_pack(struct bitmap_index *bitmap_git, |
| 669 | struct object_list *roots) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 670 | { |
| 671 | while (roots) { |
| 672 | struct object *object = roots->item; |
| 673 | roots = roots->next; |
| 674 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 675 | if (find_pack_entry_one(object->oid.hash, bitmap_git->pack) > 0) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 676 | return 1; |
| 677 | } |
| 678 | |
| 679 | return 0; |
| 680 | } |
| 681 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 682 | struct bitmap_index *prepare_bitmap_walk(struct rev_info *revs) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 683 | { |
| 684 | unsigned int i; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 685 | |
| 686 | struct object_list *wants = NULL; |
| 687 | struct object_list *haves = NULL; |
| 688 | |
| 689 | struct bitmap *wants_bitmap = NULL; |
| 690 | struct bitmap *haves_bitmap = NULL; |
| 691 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 692 | struct bitmap_index *bitmap_git = xcalloc(1, sizeof(*bitmap_git)); |
| 693 | /* try to open a bitmapped pack, but don't parse it yet |
| 694 | * because we may not need to use it */ |
Nguyễn Thái Ngọc Duy | 7c14112 | 2018-11-10 06:49:08 +0100 | [diff] [blame] | 695 | if (open_pack_bitmap(revs->repo, bitmap_git) < 0) |
Jonathan Tan | f3c23db | 2018-06-07 12:04:14 -0700 | [diff] [blame] | 696 | goto cleanup; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 697 | |
Martin Ågren | 4d01a7f | 2017-09-23 01:34:54 +0200 | [diff] [blame] | 698 | for (i = 0; i < revs->pending.nr; ++i) { |
| 699 | struct object *object = revs->pending.objects[i].item; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 700 | |
| 701 | if (object->type == OBJ_NONE) |
brian m. carlson | c251c83 | 2017-05-06 22:10:38 +0000 | [diff] [blame] | 702 | parse_object_or_die(&object->oid, NULL); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 703 | |
| 704 | while (object->type == OBJ_TAG) { |
| 705 | struct tag *tag = (struct tag *) object; |
| 706 | |
| 707 | if (object->flags & UNINTERESTING) |
| 708 | object_list_insert(object, &haves); |
| 709 | else |
| 710 | object_list_insert(object, &wants); |
| 711 | |
| 712 | if (!tag->tagged) |
| 713 | die("bad tag"); |
brian m. carlson | c251c83 | 2017-05-06 22:10:38 +0000 | [diff] [blame] | 714 | object = parse_object_or_die(&tag->tagged->oid, NULL); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 715 | } |
| 716 | |
| 717 | if (object->flags & UNINTERESTING) |
| 718 | object_list_insert(object, &haves); |
| 719 | else |
| 720 | object_list_insert(object, &wants); |
| 721 | } |
| 722 | |
| 723 | /* |
| 724 | * if we have a HAVES list, but none of those haves is contained |
| 725 | * in the packfile that has a bitmap, we don't have anything to |
| 726 | * optimize here |
| 727 | */ |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 728 | if (haves && !in_bitmapped_pack(bitmap_git, haves)) |
Jonathan Tan | f3c23db | 2018-06-07 12:04:14 -0700 | [diff] [blame] | 729 | goto cleanup; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 730 | |
| 731 | /* if we don't want anything, we're done here */ |
| 732 | if (!wants) |
Jonathan Tan | f3c23db | 2018-06-07 12:04:14 -0700 | [diff] [blame] | 733 | goto cleanup; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 734 | |
| 735 | /* |
| 736 | * now we're going to use bitmaps, so load the actual bitmap entries |
| 737 | * from disk. this is the point of no return; after this the rev_list |
| 738 | * becomes invalidated and we must perform the revwalk through bitmaps |
| 739 | */ |
Jeff King | 199c86b | 2018-09-01 03:50:57 -0400 | [diff] [blame] | 740 | if (load_pack_bitmap(bitmap_git) < 0) |
Jonathan Tan | f3c23db | 2018-06-07 12:04:14 -0700 | [diff] [blame] | 741 | goto cleanup; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 742 | |
Martin Ågren | 4d01a7f | 2017-09-23 01:34:54 +0200 | [diff] [blame] | 743 | object_array_clear(&revs->pending); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 744 | |
| 745 | if (haves) { |
Vicent Marti | 2db1a43 | 2014-03-28 06:00:43 -0400 | [diff] [blame] | 746 | revs->ignore_missing_links = 1; |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 747 | haves_bitmap = find_objects(bitmap_git, revs, haves, NULL); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 748 | reset_revision_walk(); |
Vicent Marti | 2db1a43 | 2014-03-28 06:00:43 -0400 | [diff] [blame] | 749 | revs->ignore_missing_links = 0; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 750 | |
| 751 | if (haves_bitmap == NULL) |
Johannes Schindelin | 033abf9 | 2018-05-02 11:38:39 +0200 | [diff] [blame] | 752 | BUG("failed to perform bitmap walk"); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 753 | } |
| 754 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 755 | wants_bitmap = find_objects(bitmap_git, revs, wants, haves_bitmap); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 756 | |
| 757 | if (!wants_bitmap) |
Johannes Schindelin | 033abf9 | 2018-05-02 11:38:39 +0200 | [diff] [blame] | 758 | BUG("failed to perform bitmap walk"); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 759 | |
| 760 | if (haves_bitmap) |
| 761 | bitmap_and_not(wants_bitmap, haves_bitmap); |
| 762 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 763 | bitmap_git->result = wants_bitmap; |
Jeff King | 30cdc33 | 2018-08-21 15:07:01 -0400 | [diff] [blame] | 764 | bitmap_git->haves = haves_bitmap; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 765 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 766 | return bitmap_git; |
Jonathan Tan | f3c23db | 2018-06-07 12:04:14 -0700 | [diff] [blame] | 767 | |
| 768 | cleanup: |
| 769 | free_bitmap_index(bitmap_git); |
| 770 | return NULL; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 771 | } |
| 772 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 773 | int reuse_partial_packfile_from_bitmap(struct bitmap_index *bitmap_git, |
| 774 | struct packed_git **packfile, |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 775 | uint32_t *entries, |
| 776 | off_t *up_to) |
| 777 | { |
| 778 | /* |
| 779 | * Reuse the packfile content if we need more than |
| 780 | * 90% of its objects |
| 781 | */ |
| 782 | static const double REUSE_PERCENT = 0.9; |
| 783 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 784 | struct bitmap *result = bitmap_git->result; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 785 | uint32_t reuse_threshold; |
| 786 | uint32_t i, reuse_objects = 0; |
| 787 | |
| 788 | assert(result); |
| 789 | |
| 790 | for (i = 0; i < result->word_alloc; ++i) { |
| 791 | if (result->words[i] != (eword_t)~0) { |
| 792 | reuse_objects += ewah_bit_ctz64(~result->words[i]); |
| 793 | break; |
| 794 | } |
| 795 | |
Jeff King | 34b935c | 2015-06-03 02:39:37 -0400 | [diff] [blame] | 796 | reuse_objects += BITS_IN_EWORD; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 797 | } |
| 798 | |
| 799 | #ifdef GIT_BITMAP_DEBUG |
| 800 | { |
| 801 | const unsigned char *sha1; |
| 802 | struct revindex_entry *entry; |
| 803 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 804 | entry = &bitmap_git->reverse_index->revindex[reuse_objects]; |
| 805 | sha1 = nth_packed_object_sha1(bitmap_git->pack, entry->nr); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 806 | |
| 807 | fprintf(stderr, "Failed to reuse at %d (%016llx)\n", |
| 808 | reuse_objects, result->words[i]); |
brian m. carlson | 0dd4924 | 2019-02-19 00:04:56 +0000 | [diff] [blame] | 809 | fprintf(stderr, " %s\n", hash_to_hex(sha1)); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 810 | } |
| 811 | #endif |
| 812 | |
| 813 | if (!reuse_objects) |
| 814 | return -1; |
| 815 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 816 | if (reuse_objects >= bitmap_git->pack->num_objects) { |
| 817 | bitmap_git->reuse_objects = *entries = bitmap_git->pack->num_objects; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 818 | *up_to = -1; /* reuse the full pack */ |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 819 | *packfile = bitmap_git->pack; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 820 | return 0; |
| 821 | } |
| 822 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 823 | reuse_threshold = bitmap_popcount(bitmap_git->result) * REUSE_PERCENT; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 824 | |
| 825 | if (reuse_objects < reuse_threshold) |
| 826 | return -1; |
| 827 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 828 | bitmap_git->reuse_objects = *entries = reuse_objects; |
| 829 | *up_to = bitmap_git->pack->revindex[reuse_objects].offset; |
| 830 | *packfile = bitmap_git->pack; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 831 | |
| 832 | return 0; |
| 833 | } |
| 834 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 835 | void traverse_bitmap_commit_list(struct bitmap_index *bitmap_git, |
| 836 | show_reachable_fn show_reachable) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 837 | { |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 838 | assert(bitmap_git->result); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 839 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 840 | show_objects_for_type(bitmap_git, bitmap_git->commits, |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 841 | OBJ_COMMIT, show_reachable); |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 842 | show_objects_for_type(bitmap_git, bitmap_git->trees, |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 843 | OBJ_TREE, show_reachable); |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 844 | show_objects_for_type(bitmap_git, bitmap_git->blobs, |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 845 | OBJ_BLOB, show_reachable); |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 846 | show_objects_for_type(bitmap_git, bitmap_git->tags, |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 847 | OBJ_TAG, show_reachable); |
| 848 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 849 | show_extended_objects(bitmap_git, show_reachable); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 850 | } |
| 851 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 852 | static uint32_t count_object_type(struct bitmap_index *bitmap_git, |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 853 | enum object_type type) |
| 854 | { |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 855 | struct bitmap *objects = bitmap_git->result; |
| 856 | struct eindex *eindex = &bitmap_git->ext_index; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 857 | |
| 858 | uint32_t i = 0, count = 0; |
| 859 | struct ewah_iterator it; |
| 860 | eword_t filter; |
| 861 | |
| 862 | switch (type) { |
| 863 | case OBJ_COMMIT: |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 864 | ewah_iterator_init(&it, bitmap_git->commits); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 865 | break; |
| 866 | |
| 867 | case OBJ_TREE: |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 868 | ewah_iterator_init(&it, bitmap_git->trees); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 869 | break; |
| 870 | |
| 871 | case OBJ_BLOB: |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 872 | ewah_iterator_init(&it, bitmap_git->blobs); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 873 | break; |
| 874 | |
| 875 | case OBJ_TAG: |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 876 | ewah_iterator_init(&it, bitmap_git->tags); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 877 | break; |
| 878 | |
| 879 | default: |
| 880 | return 0; |
| 881 | } |
| 882 | |
| 883 | while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) { |
| 884 | eword_t word = objects->words[i++] & filter; |
| 885 | count += ewah_bit_popcount64(word); |
| 886 | } |
| 887 | |
| 888 | for (i = 0; i < eindex->count; ++i) { |
| 889 | if (eindex->objects[i]->type == type && |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 890 | bitmap_get(objects, bitmap_git->pack->num_objects + i)) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 891 | count++; |
| 892 | } |
| 893 | |
| 894 | return count; |
| 895 | } |
| 896 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 897 | void count_bitmap_commit_list(struct bitmap_index *bitmap_git, |
| 898 | uint32_t *commits, uint32_t *trees, |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 899 | uint32_t *blobs, uint32_t *tags) |
| 900 | { |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 901 | assert(bitmap_git->result); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 902 | |
| 903 | if (commits) |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 904 | *commits = count_object_type(bitmap_git, OBJ_COMMIT); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 905 | |
| 906 | if (trees) |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 907 | *trees = count_object_type(bitmap_git, OBJ_TREE); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 908 | |
| 909 | if (blobs) |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 910 | *blobs = count_object_type(bitmap_git, OBJ_BLOB); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 911 | |
| 912 | if (tags) |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 913 | *tags = count_object_type(bitmap_git, OBJ_TAG); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 914 | } |
| 915 | |
| 916 | struct bitmap_test_data { |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 917 | struct bitmap_index *bitmap_git; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 918 | struct bitmap *base; |
| 919 | struct progress *prg; |
| 920 | size_t seen; |
| 921 | }; |
| 922 | |
Jeff King | de1e67d | 2016-02-11 17:28:36 -0500 | [diff] [blame] | 923 | static void test_show_object(struct object *object, const char *name, |
| 924 | void *data) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 925 | { |
| 926 | struct bitmap_test_data *tdata = data; |
| 927 | int bitmap_pos; |
| 928 | |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 929 | bitmap_pos = bitmap_position(tdata->bitmap_git, &object->oid); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 930 | if (bitmap_pos < 0) |
brian m. carlson | f2fd076 | 2015-11-10 02:22:28 +0000 | [diff] [blame] | 931 | die("Object not in bitmap: %s\n", oid_to_hex(&object->oid)); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 932 | |
| 933 | bitmap_set(tdata->base, bitmap_pos); |
| 934 | display_progress(tdata->prg, ++tdata->seen); |
| 935 | } |
| 936 | |
| 937 | static void test_show_commit(struct commit *commit, void *data) |
| 938 | { |
| 939 | struct bitmap_test_data *tdata = data; |
| 940 | int bitmap_pos; |
| 941 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 942 | bitmap_pos = bitmap_position(tdata->bitmap_git, |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 943 | &commit->object.oid); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 944 | if (bitmap_pos < 0) |
brian m. carlson | f2fd076 | 2015-11-10 02:22:28 +0000 | [diff] [blame] | 945 | die("Object not in bitmap: %s\n", oid_to_hex(&commit->object.oid)); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 946 | |
| 947 | bitmap_set(tdata->base, bitmap_pos); |
| 948 | display_progress(tdata->prg, ++tdata->seen); |
| 949 | } |
| 950 | |
| 951 | void test_bitmap_walk(struct rev_info *revs) |
| 952 | { |
| 953 | struct object *root; |
| 954 | struct bitmap *result = NULL; |
| 955 | khiter_t pos; |
| 956 | size_t result_popcnt; |
| 957 | struct bitmap_test_data tdata; |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 958 | struct bitmap_index *bitmap_git; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 959 | |
Nguyễn Thái Ngọc Duy | 7c14112 | 2018-11-10 06:49:08 +0100 | [diff] [blame] | 960 | if (!(bitmap_git = prepare_bitmap_git(revs->repo))) |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 961 | die("failed to load bitmap indexes"); |
| 962 | |
| 963 | if (revs->pending.nr != 1) |
| 964 | die("you must specify exactly one commit to test"); |
| 965 | |
| 966 | fprintf(stderr, "Bitmap v%d test (%d entries loaded)\n", |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 967 | bitmap_git->version, bitmap_git->entry_count); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 968 | |
| 969 | root = revs->pending.objects[0].item; |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 970 | pos = kh_get_oid_map(bitmap_git->bitmaps, root->oid); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 971 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 972 | if (pos < kh_end(bitmap_git->bitmaps)) { |
| 973 | struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, pos); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 974 | struct ewah_bitmap *bm = lookup_stored_bitmap(st); |
| 975 | |
| 976 | fprintf(stderr, "Found bitmap for %s. %d bits / %08x checksum\n", |
brian m. carlson | f2fd076 | 2015-11-10 02:22:28 +0000 | [diff] [blame] | 977 | oid_to_hex(&root->oid), (int)bm->bit_size, ewah_checksum(bm)); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 978 | |
| 979 | result = ewah_to_bitmap(bm); |
| 980 | } |
| 981 | |
| 982 | if (result == NULL) |
brian m. carlson | f2fd076 | 2015-11-10 02:22:28 +0000 | [diff] [blame] | 983 | die("Commit %s doesn't have an indexed bitmap", oid_to_hex(&root->oid)); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 984 | |
| 985 | revs->tag_objects = 1; |
| 986 | revs->tree_objects = 1; |
| 987 | revs->blob_objects = 1; |
| 988 | |
| 989 | result_popcnt = bitmap_popcount(result); |
| 990 | |
| 991 | if (prepare_revision_walk(revs)) |
| 992 | die("revision walk setup failed"); |
| 993 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 994 | tdata.bitmap_git = bitmap_git; |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 995 | tdata.base = bitmap_new(); |
| 996 | tdata.prg = start_progress("Verifying bitmap entries", result_popcnt); |
| 997 | tdata.seen = 0; |
| 998 | |
| 999 | traverse_commit_list(revs, &test_show_commit, &test_show_object, &tdata); |
| 1000 | |
| 1001 | stop_progress(&tdata.prg); |
| 1002 | |
| 1003 | if (bitmap_equals(result, tdata.base)) |
| 1004 | fprintf(stderr, "OK!\n"); |
| 1005 | else |
| 1006 | fprintf(stderr, "Mismatch!\n"); |
Stefan Beller | f86a374 | 2015-03-30 18:22:10 -0700 | [diff] [blame] | 1007 | |
Jonathan Tan | f3c23db | 2018-06-07 12:04:14 -0700 | [diff] [blame] | 1008 | free_bitmap_index(bitmap_git); |
Vicent Marti | fff4275 | 2013-12-21 09:00:01 -0500 | [diff] [blame] | 1009 | } |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 1010 | |
| 1011 | static int rebuild_bitmap(uint32_t *reposition, |
| 1012 | struct ewah_bitmap *source, |
| 1013 | struct bitmap *dest) |
| 1014 | { |
| 1015 | uint32_t pos = 0; |
| 1016 | struct ewah_iterator it; |
| 1017 | eword_t word; |
| 1018 | |
| 1019 | ewah_iterator_init(&it, source); |
| 1020 | |
| 1021 | while (ewah_iterator_next(&word, &it)) { |
| 1022 | uint32_t offset, bit_pos; |
| 1023 | |
Jeff King | 34b935c | 2015-06-03 02:39:37 -0400 | [diff] [blame] | 1024 | for (offset = 0; offset < BITS_IN_EWORD; ++offset) { |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 1025 | if ((word >> offset) == 0) |
| 1026 | break; |
| 1027 | |
| 1028 | offset += ewah_bit_ctz64(word >> offset); |
| 1029 | |
| 1030 | bit_pos = reposition[pos + offset]; |
| 1031 | if (bit_pos > 0) |
| 1032 | bitmap_set(dest, bit_pos - 1); |
| 1033 | else /* can't reuse, we don't have the object */ |
| 1034 | return -1; |
| 1035 | } |
| 1036 | |
Jeff King | 34b935c | 2015-06-03 02:39:37 -0400 | [diff] [blame] | 1037 | pos += BITS_IN_EWORD; |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 1038 | } |
| 1039 | return 0; |
| 1040 | } |
| 1041 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 1042 | int rebuild_existing_bitmaps(struct bitmap_index *bitmap_git, |
| 1043 | struct packing_data *mapping, |
Jeff King | d2bc62b | 2019-06-20 03:41:35 -0400 | [diff] [blame] | 1044 | kh_oid_map_t *reused_bitmaps, |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 1045 | int show_progress) |
| 1046 | { |
| 1047 | uint32_t i, num_objects; |
| 1048 | uint32_t *reposition; |
| 1049 | struct bitmap *rebuild; |
| 1050 | struct stored_bitmap *stored; |
| 1051 | struct progress *progress = NULL; |
| 1052 | |
| 1053 | khiter_t hash_pos; |
| 1054 | int hash_ret; |
| 1055 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 1056 | num_objects = bitmap_git->pack->num_objects; |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 1057 | reposition = xcalloc(num_objects, sizeof(uint32_t)); |
| 1058 | |
| 1059 | for (i = 0; i < num_objects; ++i) { |
Jeff King | 3df28ca | 2019-06-20 03:41:03 -0400 | [diff] [blame] | 1060 | struct object_id oid; |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 1061 | struct revindex_entry *entry; |
| 1062 | struct object_entry *oe; |
| 1063 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 1064 | entry = &bitmap_git->pack->revindex[i]; |
Jeff King | 3df28ca | 2019-06-20 03:41:03 -0400 | [diff] [blame] | 1065 | nth_packed_object_oid(&oid, bitmap_git->pack, entry->nr); |
| 1066 | oe = packlist_find(mapping, &oid, NULL); |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 1067 | |
| 1068 | if (oe) |
Nguyễn Thái Ngọc Duy | 06af3bb | 2018-04-14 17:35:04 +0200 | [diff] [blame] | 1069 | reposition[i] = oe_in_pack_pos(mapping, oe) + 1; |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 1070 | } |
| 1071 | |
| 1072 | rebuild = bitmap_new(); |
| 1073 | i = 0; |
| 1074 | |
| 1075 | if (show_progress) |
| 1076 | progress = start_progress("Reusing bitmaps", 0); |
| 1077 | |
Jonathan Tan | 3ae5fa0 | 2018-06-07 12:04:13 -0700 | [diff] [blame] | 1078 | kh_foreach_value(bitmap_git->bitmaps, stored, { |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 1079 | if (stored->flags & BITMAP_FLAG_REUSE) { |
| 1080 | if (!rebuild_bitmap(reposition, |
| 1081 | lookup_stored_bitmap(stored), |
| 1082 | rebuild)) { |
Jeff King | d2bc62b | 2019-06-20 03:41:35 -0400 | [diff] [blame] | 1083 | hash_pos = kh_put_oid_map(reused_bitmaps, |
| 1084 | stored->oid, |
| 1085 | &hash_ret); |
Vicent Marti | 7cc8f97 | 2013-12-21 09:00:16 -0500 | [diff] [blame] | 1086 | kh_value(reused_bitmaps, hash_pos) = |
| 1087 | bitmap_to_ewah(rebuild); |
| 1088 | } |
| 1089 | bitmap_reset(rebuild); |
| 1090 | display_progress(progress, ++i); |
| 1091 | } |
| 1092 | }); |
| 1093 | |
| 1094 | stop_progress(&progress); |
| 1095 | |
| 1096 | free(reposition); |
| 1097 | bitmap_free(rebuild); |
| 1098 | return 0; |
| 1099 | } |
Jonathan Tan | f3c23db | 2018-06-07 12:04:14 -0700 | [diff] [blame] | 1100 | |
| 1101 | void free_bitmap_index(struct bitmap_index *b) |
| 1102 | { |
| 1103 | if (!b) |
| 1104 | return; |
| 1105 | |
| 1106 | if (b->map) |
| 1107 | munmap(b->map, b->map_size); |
| 1108 | ewah_pool_free(b->commits); |
| 1109 | ewah_pool_free(b->trees); |
| 1110 | ewah_pool_free(b->blobs); |
| 1111 | ewah_pool_free(b->tags); |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 1112 | kh_destroy_oid_map(b->bitmaps); |
Jonathan Tan | f3c23db | 2018-06-07 12:04:14 -0700 | [diff] [blame] | 1113 | free(b->ext_index.objects); |
| 1114 | free(b->ext_index.hashes); |
| 1115 | bitmap_free(b->result); |
Jeff King | 30cdc33 | 2018-08-21 15:07:01 -0400 | [diff] [blame] | 1116 | bitmap_free(b->haves); |
Jonathan Tan | f3c23db | 2018-06-07 12:04:14 -0700 | [diff] [blame] | 1117 | free(b); |
| 1118 | } |
Jeff King | 30cdc33 | 2018-08-21 15:07:01 -0400 | [diff] [blame] | 1119 | |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 1120 | int bitmap_has_oid_in_uninteresting(struct bitmap_index *bitmap_git, |
| 1121 | const struct object_id *oid) |
Jeff King | 30cdc33 | 2018-08-21 15:07:01 -0400 | [diff] [blame] | 1122 | { |
| 1123 | int pos; |
| 1124 | |
| 1125 | if (!bitmap_git) |
| 1126 | return 0; /* no bitmap loaded */ |
Jeff King | 30cdc33 | 2018-08-21 15:07:01 -0400 | [diff] [blame] | 1127 | if (!bitmap_git->haves) |
| 1128 | return 0; /* walk had no "haves" */ |
| 1129 | |
brian m. carlson | 3c77144 | 2019-02-19 00:04:58 +0000 | [diff] [blame] | 1130 | pos = bitmap_position_packfile(bitmap_git, oid); |
Jeff King | 30cdc33 | 2018-08-21 15:07:01 -0400 | [diff] [blame] | 1131 | if (pos < 0) |
| 1132 | return 0; |
| 1133 | |
| 1134 | return bitmap_get(bitmap_git->haves, pos); |
| 1135 | } |