Jonathan Tan | 4f39cd8 | 2017-08-18 15:20:16 -0700 | [diff] [blame] | 1 | #include "cache.h" |
Gargi Sharma | ec2dd32 | 2018-01-23 18:46:51 -0500 | [diff] [blame] | 2 | #include "list.h" |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 3 | #include "pack.h" |
Stefan Beller | 031dc92 | 2018-03-23 18:20:57 +0100 | [diff] [blame] | 4 | #include "repository.h" |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 5 | #include "dir.h" |
| 6 | #include "mergesort.h" |
| 7 | #include "packfile.h" |
Jonathan Tan | 7b3aa75 | 2017-08-18 15:20:28 -0700 | [diff] [blame] | 8 | #include "delta.h" |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 9 | #include "streaming.h" |
Martin Ågren | bc62692 | 2020-12-31 12:56:23 +0100 | [diff] [blame] | 10 | #include "hash-lookup.h" |
Jonathan Tan | 498f1f6 | 2017-12-05 16:58:44 +0000 | [diff] [blame] | 11 | #include "commit.h" |
| 12 | #include "object.h" |
| 13 | #include "tag.h" |
| 14 | #include "tree-walk.h" |
| 15 | #include "tree.h" |
Stefan Beller | 0d4a132 | 2018-03-23 18:20:56 +0100 | [diff] [blame] | 16 | #include "object-store.h" |
Derrick Stolee | c4d2522 | 2018-07-12 15:39:33 -0400 | [diff] [blame] | 17 | #include "midx.h" |
Derrick Stolee | 5472c32 | 2019-05-17 11:41:48 -0700 | [diff] [blame] | 18 | #include "commit-graph.h" |
Christian Couder | b14ed5a | 2019-06-25 15:40:31 +0200 | [diff] [blame] | 19 | #include "promisor-remote.h" |
Jonathan Tan | 4f39cd8 | 2017-08-18 15:20:16 -0700 | [diff] [blame] | 20 | |
| 21 | char *odb_pack_name(struct strbuf *buf, |
brian m. carlson | 3a4d7aa | 2019-08-18 20:04:22 +0000 | [diff] [blame] | 22 | const unsigned char *hash, |
Jonathan Tan | 4f39cd8 | 2017-08-18 15:20:16 -0700 | [diff] [blame] | 23 | const char *ext) |
| 24 | { |
| 25 | strbuf_reset(buf); |
| 26 | strbuf_addf(buf, "%s/pack/pack-%s.%s", get_object_directory(), |
brian m. carlson | 3a4d7aa | 2019-08-18 20:04:22 +0000 | [diff] [blame] | 27 | hash_to_hex(hash), ext); |
Jonathan Tan | 4f39cd8 | 2017-08-18 15:20:16 -0700 | [diff] [blame] | 28 | return buf->buf; |
| 29 | } |
| 30 | |
| 31 | char *sha1_pack_name(const unsigned char *sha1) |
| 32 | { |
| 33 | static struct strbuf buf = STRBUF_INIT; |
| 34 | return odb_pack_name(&buf, sha1, "pack"); |
| 35 | } |
| 36 | |
| 37 | char *sha1_pack_index_name(const unsigned char *sha1) |
| 38 | { |
| 39 | static struct strbuf buf = STRBUF_INIT; |
| 40 | return odb_pack_name(&buf, sha1, "idx"); |
| 41 | } |
Jonathan Tan | 6d6a80e | 2017-08-18 15:20:17 -0700 | [diff] [blame] | 42 | |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 43 | static unsigned int pack_used_ctr; |
| 44 | static unsigned int pack_mmap_calls; |
| 45 | static unsigned int peak_pack_open_windows; |
| 46 | static unsigned int pack_open_windows; |
Jonathan Tan | e65f186 | 2017-08-18 15:20:25 -0700 | [diff] [blame] | 47 | static unsigned int pack_open_fds; |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 48 | static unsigned int pack_max_fds; |
| 49 | static size_t peak_pack_mapped; |
| 50 | static size_t pack_mapped; |
Jonathan Tan | 8e21176 | 2017-08-18 15:20:18 -0700 | [diff] [blame] | 51 | |
| 52 | #define SZ_FMT PRIuMAX |
| 53 | static inline uintmax_t sz_fmt(size_t s) { return s; } |
| 54 | |
| 55 | void pack_report(void) |
| 56 | { |
| 57 | fprintf(stderr, |
| 58 | "pack_report: getpagesize() = %10" SZ_FMT "\n" |
| 59 | "pack_report: core.packedGitWindowSize = %10" SZ_FMT "\n" |
| 60 | "pack_report: core.packedGitLimit = %10" SZ_FMT "\n", |
| 61 | sz_fmt(getpagesize()), |
| 62 | sz_fmt(packed_git_window_size), |
| 63 | sz_fmt(packed_git_limit)); |
| 64 | fprintf(stderr, |
| 65 | "pack_report: pack_used_ctr = %10u\n" |
| 66 | "pack_report: pack_mmap_calls = %10u\n" |
| 67 | "pack_report: pack_open_windows = %10u / %10u\n" |
| 68 | "pack_report: pack_mapped = " |
| 69 | "%10" SZ_FMT " / %10" SZ_FMT "\n", |
| 70 | pack_used_ctr, |
| 71 | pack_mmap_calls, |
| 72 | pack_open_windows, peak_pack_open_windows, |
| 73 | sz_fmt(pack_mapped), sz_fmt(peak_pack_mapped)); |
| 74 | } |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 75 | |
| 76 | /* |
| 77 | * Open and mmap the index file at path, perform a couple of |
| 78 | * consistency checks, then record its information to p. Return 0 on |
| 79 | * success. |
| 80 | */ |
| 81 | static int check_packed_git_idx(const char *path, struct packed_git *p) |
| 82 | { |
| 83 | void *idx_map; |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 84 | size_t idx_size; |
Josh Steadmon | 1127a98 | 2018-10-12 17:58:41 -0700 | [diff] [blame] | 85 | int fd = git_open(path), ret; |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 86 | struct stat st; |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 87 | const unsigned int hashsz = the_hash_algo->rawsz; |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 88 | |
| 89 | if (fd < 0) |
| 90 | return -1; |
| 91 | if (fstat(fd, &st)) { |
| 92 | close(fd); |
| 93 | return -1; |
| 94 | } |
| 95 | idx_size = xsize_t(st.st_size); |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 96 | if (idx_size < 4 * 256 + hashsz + hashsz) { |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 97 | close(fd); |
| 98 | return error("index file %s is too small", path); |
| 99 | } |
| 100 | idx_map = xmmap(NULL, idx_size, PROT_READ, MAP_PRIVATE, fd, 0); |
| 101 | close(fd); |
| 102 | |
Josh Steadmon | 1127a98 | 2018-10-12 17:58:41 -0700 | [diff] [blame] | 103 | ret = load_idx(path, hashsz, idx_map, idx_size, p); |
| 104 | |
| 105 | if (ret) |
| 106 | munmap(idx_map, idx_size); |
| 107 | |
| 108 | return ret; |
| 109 | } |
| 110 | |
| 111 | int load_idx(const char *path, const unsigned int hashsz, void *idx_map, |
| 112 | size_t idx_size, struct packed_git *p) |
| 113 | { |
| 114 | struct pack_idx_header *hdr = idx_map; |
| 115 | uint32_t version, nr, i, *index; |
| 116 | |
| 117 | if (idx_size < 4 * 256 + hashsz + hashsz) |
| 118 | return error("index file %s is too small", path); |
| 119 | if (idx_map == NULL) |
| 120 | return error("empty data"); |
| 121 | |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 122 | if (hdr->idx_signature == htonl(PACK_IDX_SIGNATURE)) { |
| 123 | version = ntohl(hdr->idx_version); |
Josh Steadmon | 1127a98 | 2018-10-12 17:58:41 -0700 | [diff] [blame] | 124 | if (version < 2 || version > 2) |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 125 | return error("index file %s is version %"PRIu32 |
| 126 | " and is not supported by this binary" |
| 127 | " (try upgrading GIT to a newer version)", |
| 128 | path, version); |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 129 | } else |
| 130 | version = 1; |
| 131 | |
| 132 | nr = 0; |
| 133 | index = idx_map; |
| 134 | if (version > 1) |
| 135 | index += 2; /* skip index header */ |
| 136 | for (i = 0; i < 256; i++) { |
| 137 | uint32_t n = ntohl(index[i]); |
Josh Steadmon | 1127a98 | 2018-10-12 17:58:41 -0700 | [diff] [blame] | 138 | if (n < nr) |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 139 | return error("non-monotonic index %s", path); |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 140 | nr = n; |
| 141 | } |
| 142 | |
| 143 | if (version == 1) { |
| 144 | /* |
| 145 | * Total size: |
| 146 | * - 256 index entries 4 bytes each |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 147 | * - 24-byte entries * nr (object ID + 4-byte offset) |
| 148 | * - hash of the packfile |
| 149 | * - file checksum |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 150 | */ |
Jeff King | 81c4c5c | 2020-11-13 00:07:19 -0500 | [diff] [blame] | 151 | if (idx_size != st_add(4 * 256 + hashsz + hashsz, st_mult(nr, hashsz + 4))) |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 152 | return error("wrong index v1 file size in %s", path); |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 153 | } else if (version == 2) { |
| 154 | /* |
| 155 | * Minimum size: |
| 156 | * - 8 bytes of header |
| 157 | * - 256 index entries 4 bytes each |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 158 | * - object ID entry * nr |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 159 | * - 4-byte crc entry * nr |
| 160 | * - 4-byte offset entry * nr |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 161 | * - hash of the packfile |
| 162 | * - file checksum |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 163 | * And after the 4-byte offset table might be a |
| 164 | * variable sized table containing 8-byte entries |
| 165 | * for offsets larger than 2^31. |
| 166 | */ |
Jeff King | 81c4c5c | 2020-11-13 00:07:19 -0500 | [diff] [blame] | 167 | size_t min_size = st_add(8 + 4*256 + hashsz + hashsz, st_mult(nr, hashsz + 4 + 4)); |
Jeff King | a9bc372 | 2020-11-13 00:07:01 -0500 | [diff] [blame] | 168 | size_t max_size = min_size; |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 169 | if (nr) |
Jeff King | 81c4c5c | 2020-11-13 00:07:19 -0500 | [diff] [blame] | 170 | max_size = st_add(max_size, st_mult(nr - 1, 8)); |
Josh Steadmon | 1127a98 | 2018-10-12 17:58:41 -0700 | [diff] [blame] | 171 | if (idx_size < min_size || idx_size > max_size) |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 172 | return error("wrong index v2 file size in %s", path); |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 173 | if (idx_size != min_size && |
| 174 | /* |
| 175 | * make sure we can deal with large pack offsets. |
| 176 | * 31-bit signed offset won't be enough, neither |
| 177 | * 32-bit unsigned one will be. |
| 178 | */ |
Josh Steadmon | 1127a98 | 2018-10-12 17:58:41 -0700 | [diff] [blame] | 179 | (sizeof(off_t) <= 4)) |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 180 | return error("pack too large for current definition of off_t in %s", path); |
brian m. carlson | 629dffc | 2020-05-25 19:59:10 +0000 | [diff] [blame] | 181 | p->crc_offset = 8 + 4 * 256 + nr * hashsz; |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 182 | } |
| 183 | |
| 184 | p->index_version = version; |
| 185 | p->index_data = idx_map; |
| 186 | p->index_size = idx_size; |
| 187 | p->num_objects = nr; |
| 188 | return 0; |
| 189 | } |
| 190 | |
| 191 | int open_pack_index(struct packed_git *p) |
| 192 | { |
| 193 | char *idx_name; |
| 194 | size_t len; |
| 195 | int ret; |
| 196 | |
| 197 | if (p->index_data) |
| 198 | return 0; |
| 199 | |
| 200 | if (!strip_suffix(p->pack_name, ".pack", &len)) |
Johannes Schindelin | 033abf9 | 2018-05-02 11:38:39 +0200 | [diff] [blame] | 201 | BUG("pack_name does not end in .pack"); |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 202 | idx_name = xstrfmt("%.*s.idx", (int)len, p->pack_name); |
| 203 | ret = check_packed_git_idx(idx_name, p); |
| 204 | free(idx_name); |
| 205 | return ret; |
| 206 | } |
| 207 | |
Derrick Stolee | fe1ed56 | 2018-07-12 15:39:29 -0400 | [diff] [blame] | 208 | uint32_t get_pack_fanout(struct packed_git *p, uint32_t value) |
| 209 | { |
| 210 | const uint32_t *level1_ofs = p->index_data; |
| 211 | |
| 212 | if (!level1_ofs) { |
| 213 | if (open_pack_index(p)) |
| 214 | return 0; |
| 215 | level1_ofs = p->index_data; |
| 216 | } |
| 217 | |
| 218 | if (p->index_version > 1) { |
| 219 | level1_ofs += 2; |
| 220 | } |
| 221 | |
| 222 | return ntohl(level1_ofs[value]); |
| 223 | } |
| 224 | |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 225 | static struct packed_git *alloc_packed_git(int extra) |
| 226 | { |
| 227 | struct packed_git *p = xmalloc(st_add(sizeof(*p), extra)); |
| 228 | memset(p, 0, sizeof(*p)); |
| 229 | p->pack_fd = -1; |
| 230 | return p; |
| 231 | } |
| 232 | |
| 233 | struct packed_git *parse_pack_index(unsigned char *sha1, const char *idx_path) |
| 234 | { |
| 235 | const char *path = sha1_pack_name(sha1); |
| 236 | size_t alloc = st_add(strlen(path), 1); |
| 237 | struct packed_git *p = alloc_packed_git(alloc); |
| 238 | |
| 239 | memcpy(p->pack_name, path, alloc); /* includes NUL */ |
brian m. carlson | 538b152 | 2019-02-19 00:05:03 +0000 | [diff] [blame] | 240 | hashcpy(p->hash, sha1); |
Jonathan Tan | 0317f45 | 2017-08-18 15:20:19 -0700 | [diff] [blame] | 241 | if (check_packed_git_idx(idx_path, p)) { |
| 242 | free(p); |
| 243 | return NULL; |
| 244 | } |
| 245 | |
| 246 | return p; |
| 247 | } |
Jonathan Tan | f0e17e8 | 2017-08-18 15:20:20 -0700 | [diff] [blame] | 248 | |
| 249 | static void scan_windows(struct packed_git *p, |
| 250 | struct packed_git **lru_p, |
| 251 | struct pack_window **lru_w, |
| 252 | struct pack_window **lru_l) |
| 253 | { |
| 254 | struct pack_window *w, *w_l; |
| 255 | |
| 256 | for (w_l = NULL, w = p->windows; w; w = w->next) { |
| 257 | if (!w->inuse_cnt) { |
| 258 | if (!*lru_w || w->last_used < (*lru_w)->last_used) { |
| 259 | *lru_p = p; |
| 260 | *lru_w = w; |
| 261 | *lru_l = w_l; |
| 262 | } |
| 263 | } |
| 264 | w_l = w; |
| 265 | } |
| 266 | } |
| 267 | |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 268 | static int unuse_one_window(struct packed_git *current) |
Jonathan Tan | f0e17e8 | 2017-08-18 15:20:20 -0700 | [diff] [blame] | 269 | { |
| 270 | struct packed_git *p, *lru_p = NULL; |
| 271 | struct pack_window *lru_w = NULL, *lru_l = NULL; |
| 272 | |
| 273 | if (current) |
| 274 | scan_windows(current, &lru_p, &lru_w, &lru_l); |
Stefan Beller | a80d72d | 2018-03-23 18:20:59 +0100 | [diff] [blame] | 275 | for (p = the_repository->objects->packed_git; p; p = p->next) |
Jonathan Tan | f0e17e8 | 2017-08-18 15:20:20 -0700 | [diff] [blame] | 276 | scan_windows(p, &lru_p, &lru_w, &lru_l); |
| 277 | if (lru_p) { |
| 278 | munmap(lru_w->base, lru_w->len); |
| 279 | pack_mapped -= lru_w->len; |
| 280 | if (lru_l) |
| 281 | lru_l->next = lru_w->next; |
| 282 | else |
| 283 | lru_p->windows = lru_w->next; |
| 284 | free(lru_w); |
| 285 | pack_open_windows--; |
| 286 | return 1; |
| 287 | } |
| 288 | return 0; |
| 289 | } |
| 290 | |
Jonathan Tan | 3836d88 | 2017-08-18 15:20:21 -0700 | [diff] [blame] | 291 | void close_pack_windows(struct packed_git *p) |
| 292 | { |
| 293 | while (p->windows) { |
| 294 | struct pack_window *w = p->windows; |
| 295 | |
| 296 | if (w->inuse_cnt) |
| 297 | die("pack '%s' still has open windows to it", |
| 298 | p->pack_name); |
| 299 | munmap(w->base, w->len); |
| 300 | pack_mapped -= w->len; |
| 301 | pack_open_windows--; |
| 302 | p->windows = w->next; |
| 303 | free(w); |
| 304 | } |
| 305 | } |
| 306 | |
Jeff Hostetler | 5ae18df | 2019-03-21 12:36:15 -0700 | [diff] [blame] | 307 | int close_pack_fd(struct packed_git *p) |
Jonathan Tan | 3836d88 | 2017-08-18 15:20:21 -0700 | [diff] [blame] | 308 | { |
| 309 | if (p->pack_fd < 0) |
| 310 | return 0; |
| 311 | |
| 312 | close(p->pack_fd); |
| 313 | pack_open_fds--; |
| 314 | p->pack_fd = -1; |
| 315 | |
| 316 | return 1; |
| 317 | } |
| 318 | |
| 319 | void close_pack_index(struct packed_git *p) |
| 320 | { |
| 321 | if (p->index_data) { |
| 322 | munmap((void *)p->index_data, p->index_size); |
| 323 | p->index_data = NULL; |
| 324 | } |
| 325 | } |
| 326 | |
Taylor Blau | 2f4ba2a | 2021-01-25 18:37:14 -0500 | [diff] [blame] | 327 | void close_pack_revindex(struct packed_git *p) { |
| 328 | if (!p->revindex_map) |
| 329 | return; |
| 330 | |
| 331 | munmap((void *)p->revindex_map, p->revindex_size); |
| 332 | p->revindex_map = NULL; |
| 333 | p->revindex_data = NULL; |
| 334 | } |
| 335 | |
Derrick Stolee | 049d51a | 2018-04-10 08:56:06 -0400 | [diff] [blame] | 336 | void close_pack(struct packed_git *p) |
Jonathan Tan | 3836d88 | 2017-08-18 15:20:21 -0700 | [diff] [blame] | 337 | { |
| 338 | close_pack_windows(p); |
| 339 | close_pack_fd(p); |
| 340 | close_pack_index(p); |
Taylor Blau | 2f4ba2a | 2021-01-25 18:37:14 -0500 | [diff] [blame] | 341 | close_pack_revindex(p); |
Jonathan Tan | 3836d88 | 2017-08-18 15:20:21 -0700 | [diff] [blame] | 342 | } |
| 343 | |
Derrick Stolee | 2d511cf | 2019-05-17 11:41:49 -0700 | [diff] [blame] | 344 | void close_object_store(struct raw_object_store *o) |
Jonathan Tan | 3836d88 | 2017-08-18 15:20:21 -0700 | [diff] [blame] | 345 | { |
| 346 | struct packed_git *p; |
| 347 | |
Stefan Beller | d0b5986 | 2018-03-23 18:21:00 +0100 | [diff] [blame] | 348 | for (p = o->packed_git; p; p = p->next) |
Jonathan Tan | 3836d88 | 2017-08-18 15:20:21 -0700 | [diff] [blame] | 349 | if (p->do_not_close) |
Johannes Schindelin | 033abf9 | 2018-05-02 11:38:39 +0200 | [diff] [blame] | 350 | BUG("want to close pack marked 'do-not-close'"); |
Jonathan Tan | 3836d88 | 2017-08-18 15:20:21 -0700 | [diff] [blame] | 351 | else |
| 352 | close_pack(p); |
Derrick Stolee | dc7d664 | 2018-10-25 12:54:05 +0000 | [diff] [blame] | 353 | |
| 354 | if (o->multi_pack_index) { |
| 355 | close_midx(o->multi_pack_index); |
| 356 | o->multi_pack_index = NULL; |
| 357 | } |
Derrick Stolee | 5472c32 | 2019-05-17 11:41:48 -0700 | [diff] [blame] | 358 | |
| 359 | close_commit_graph(o); |
Jonathan Tan | 3836d88 | 2017-08-18 15:20:21 -0700 | [diff] [blame] | 360 | } |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 361 | |
Derrick Stolee | 8434e85 | 2019-06-10 16:35:22 -0700 | [diff] [blame] | 362 | void unlink_pack_path(const char *pack_name, int force_delete) |
| 363 | { |
Taylor Blau | 2f4ba2a | 2021-01-25 18:37:14 -0500 | [diff] [blame] | 364 | static const char *exts[] = {".pack", ".idx", ".rev", ".keep", ".bitmap", ".promisor"}; |
Derrick Stolee | 8434e85 | 2019-06-10 16:35:22 -0700 | [diff] [blame] | 365 | int i; |
| 366 | struct strbuf buf = STRBUF_INIT; |
| 367 | size_t plen; |
| 368 | |
| 369 | strbuf_addstr(&buf, pack_name); |
| 370 | strip_suffix_mem(buf.buf, &buf.len, ".pack"); |
| 371 | plen = buf.len; |
| 372 | |
| 373 | if (!force_delete) { |
| 374 | strbuf_addstr(&buf, ".keep"); |
| 375 | if (!access(buf.buf, F_OK)) { |
| 376 | strbuf_release(&buf); |
| 377 | return; |
| 378 | } |
| 379 | } |
| 380 | |
| 381 | for (i = 0; i < ARRAY_SIZE(exts); i++) { |
| 382 | strbuf_setlen(&buf, plen); |
| 383 | strbuf_addstr(&buf, exts[i]); |
| 384 | unlink(buf.buf); |
| 385 | } |
| 386 | |
| 387 | strbuf_release(&buf); |
| 388 | } |
| 389 | |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 390 | /* |
| 391 | * The LRU pack is the one with the oldest MRU window, preferring packs |
| 392 | * with no used windows, or the oldest mtime if it has no windows allocated. |
| 393 | */ |
| 394 | static void find_lru_pack(struct packed_git *p, struct packed_git **lru_p, struct pack_window **mru_w, int *accept_windows_inuse) |
| 395 | { |
| 396 | struct pack_window *w, *this_mru_w; |
| 397 | int has_windows_inuse = 0; |
| 398 | |
| 399 | /* |
| 400 | * Reject this pack if it has windows and the previously selected |
| 401 | * one does not. If this pack does not have windows, reject |
| 402 | * it if the pack file is newer than the previously selected one. |
| 403 | */ |
| 404 | if (*lru_p && !*mru_w && (p->windows || p->mtime > (*lru_p)->mtime)) |
| 405 | return; |
| 406 | |
| 407 | for (w = this_mru_w = p->windows; w; w = w->next) { |
| 408 | /* |
| 409 | * Reject this pack if any of its windows are in use, |
| 410 | * but the previously selected pack did not have any |
| 411 | * inuse windows. Otherwise, record that this pack |
| 412 | * has windows in use. |
| 413 | */ |
| 414 | if (w->inuse_cnt) { |
| 415 | if (*accept_windows_inuse) |
| 416 | has_windows_inuse = 1; |
| 417 | else |
| 418 | return; |
| 419 | } |
| 420 | |
| 421 | if (w->last_used > this_mru_w->last_used) |
| 422 | this_mru_w = w; |
| 423 | |
| 424 | /* |
| 425 | * Reject this pack if it has windows that have been |
| 426 | * used more recently than the previously selected pack. |
| 427 | * If the previously selected pack had windows inuse and |
| 428 | * we have not encountered a window in this pack that is |
| 429 | * inuse, skip this check since we prefer a pack with no |
| 430 | * inuse windows to one that has inuse windows. |
| 431 | */ |
| 432 | if (*mru_w && *accept_windows_inuse == has_windows_inuse && |
| 433 | this_mru_w->last_used > (*mru_w)->last_used) |
| 434 | return; |
| 435 | } |
| 436 | |
| 437 | /* |
| 438 | * Select this pack. |
| 439 | */ |
| 440 | *mru_w = this_mru_w; |
| 441 | *lru_p = p; |
| 442 | *accept_windows_inuse = has_windows_inuse; |
| 443 | } |
| 444 | |
| 445 | static int close_one_pack(void) |
| 446 | { |
| 447 | struct packed_git *p, *lru_p = NULL; |
| 448 | struct pack_window *mru_w = NULL; |
| 449 | int accept_windows_inuse = 1; |
| 450 | |
Stefan Beller | a80d72d | 2018-03-23 18:20:59 +0100 | [diff] [blame] | 451 | for (p = the_repository->objects->packed_git; p; p = p->next) { |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 452 | if (p->pack_fd == -1) |
| 453 | continue; |
| 454 | find_lru_pack(p, &lru_p, &mru_w, &accept_windows_inuse); |
| 455 | } |
| 456 | |
| 457 | if (lru_p) |
| 458 | return close_pack_fd(lru_p); |
| 459 | |
| 460 | return 0; |
| 461 | } |
| 462 | |
| 463 | static unsigned int get_max_fd_limit(void) |
| 464 | { |
| 465 | #ifdef RLIMIT_NOFILE |
| 466 | { |
| 467 | struct rlimit lim; |
| 468 | |
| 469 | if (!getrlimit(RLIMIT_NOFILE, &lim)) |
| 470 | return lim.rlim_cur; |
| 471 | } |
| 472 | #endif |
| 473 | |
| 474 | #ifdef _SC_OPEN_MAX |
| 475 | { |
| 476 | long open_max = sysconf(_SC_OPEN_MAX); |
| 477 | if (0 < open_max) |
| 478 | return open_max; |
| 479 | /* |
| 480 | * Otherwise, we got -1 for one of the two |
| 481 | * reasons: |
| 482 | * |
| 483 | * (1) sysconf() did not understand _SC_OPEN_MAX |
| 484 | * and signaled an error with -1; or |
| 485 | * (2) sysconf() said there is no limit. |
| 486 | * |
| 487 | * We _could_ clear errno before calling sysconf() to |
| 488 | * tell these two cases apart and return a huge number |
| 489 | * in the latter case to let the caller cap it to a |
| 490 | * value that is not so selfish, but letting the |
| 491 | * fallback OPEN_MAX codepath take care of these cases |
| 492 | * is a lot simpler. |
| 493 | */ |
| 494 | } |
| 495 | #endif |
| 496 | |
| 497 | #ifdef OPEN_MAX |
| 498 | return OPEN_MAX; |
| 499 | #else |
| 500 | return 1; /* see the caller ;-) */ |
| 501 | #endif |
| 502 | } |
| 503 | |
Jeff King | fc78915 | 2019-04-05 14:06:22 -0400 | [diff] [blame] | 504 | const char *pack_basename(struct packed_git *p) |
| 505 | { |
| 506 | const char *ret = strrchr(p->pack_name, '/'); |
| 507 | if (ret) |
| 508 | ret = ret + 1; /* skip past slash */ |
| 509 | else |
| 510 | ret = p->pack_name; /* we only have a base */ |
| 511 | return ret; |
| 512 | } |
| 513 | |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 514 | /* |
| 515 | * Do not call this directly as this leaks p->pack_fd on error return; |
| 516 | * call open_packed_git() instead. |
| 517 | */ |
| 518 | static int open_packed_git_1(struct packed_git *p) |
| 519 | { |
| 520 | struct stat st; |
| 521 | struct pack_header hdr; |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 522 | unsigned char hash[GIT_MAX_RAWSZ]; |
| 523 | unsigned char *idx_hash; |
Jeff King | 41dcc4d | 2017-09-27 02:02:11 -0400 | [diff] [blame] | 524 | ssize_t read_result; |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 525 | const unsigned hashsz = the_hash_algo->rawsz; |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 526 | |
Taylor Blau | c8a45eb | 2020-11-25 12:17:28 -0500 | [diff] [blame] | 527 | if (open_pack_index(p)) |
| 528 | return error("packfile %s index unavailable", p->pack_name); |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 529 | |
| 530 | if (!pack_max_fds) { |
| 531 | unsigned int max_fds = get_max_fd_limit(); |
| 532 | |
| 533 | /* Save 3 for stdin/stdout/stderr, 22 for work */ |
| 534 | if (25 < max_fds) |
| 535 | pack_max_fds = max_fds - 25; |
| 536 | else |
| 537 | pack_max_fds = 1; |
| 538 | } |
| 539 | |
| 540 | while (pack_max_fds <= pack_open_fds && close_one_pack()) |
| 541 | ; /* nothing */ |
| 542 | |
| 543 | p->pack_fd = git_open(p->pack_name); |
| 544 | if (p->pack_fd < 0 || fstat(p->pack_fd, &st)) |
| 545 | return -1; |
| 546 | pack_open_fds++; |
| 547 | |
| 548 | /* If we created the struct before we had the pack we lack size. */ |
| 549 | if (!p->pack_size) { |
| 550 | if (!S_ISREG(st.st_mode)) |
| 551 | return error("packfile %s not a regular file", p->pack_name); |
| 552 | p->pack_size = st.st_size; |
| 553 | } else if (p->pack_size != st.st_size) |
| 554 | return error("packfile %s size changed", p->pack_name); |
| 555 | |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 556 | /* Verify we recognize this pack file format. */ |
Jeff King | 41dcc4d | 2017-09-27 02:02:11 -0400 | [diff] [blame] | 557 | read_result = read_in_full(p->pack_fd, &hdr, sizeof(hdr)); |
| 558 | if (read_result < 0) |
| 559 | return error_errno("error reading from %s", p->pack_name); |
| 560 | if (read_result != sizeof(hdr)) |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 561 | return error("file %s is far too short to be a packfile", p->pack_name); |
| 562 | if (hdr.hdr_signature != htonl(PACK_SIGNATURE)) |
| 563 | return error("file %s is not a GIT packfile", p->pack_name); |
| 564 | if (!pack_version_ok(hdr.hdr_version)) |
| 565 | return error("packfile %s is version %"PRIu32" and not" |
| 566 | " supported (try upgrading GIT to a newer version)", |
| 567 | p->pack_name, ntohl(hdr.hdr_version)); |
| 568 | |
| 569 | /* Verify the pack matches its index. */ |
| 570 | if (p->num_objects != ntohl(hdr.hdr_entries)) |
| 571 | return error("packfile %s claims to have %"PRIu32" objects" |
| 572 | " while index indicates %"PRIu32" objects", |
| 573 | p->pack_name, ntohl(hdr.hdr_entries), |
| 574 | p->num_objects); |
Eric Wong | 4e61b22 | 2019-12-26 10:42:20 +0000 | [diff] [blame] | 575 | read_result = pread_in_full(p->pack_fd, hash, hashsz, |
| 576 | p->pack_size - hashsz); |
Jeff King | 41dcc4d | 2017-09-27 02:02:11 -0400 | [diff] [blame] | 577 | if (read_result < 0) |
| 578 | return error_errno("error reading from %s", p->pack_name); |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 579 | if (read_result != hashsz) |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 580 | return error("packfile %s signature is unavailable", p->pack_name); |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 581 | idx_hash = ((unsigned char *)p->index_data) + p->index_size - hashsz * 2; |
Jeff King | 67947c3 | 2018-08-28 17:22:52 -0400 | [diff] [blame] | 582 | if (!hasheq(hash, idx_hash)) |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 583 | return error("packfile %s does not match index", p->pack_name); |
| 584 | return 0; |
| 585 | } |
| 586 | |
Jonathan Tan | a255195 | 2017-08-18 15:20:33 -0700 | [diff] [blame] | 587 | static int open_packed_git(struct packed_git *p) |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 588 | { |
| 589 | if (!open_packed_git_1(p)) |
| 590 | return 0; |
| 591 | close_pack_fd(p); |
| 592 | return -1; |
| 593 | } |
| 594 | |
| 595 | static int in_window(struct pack_window *win, off_t offset) |
| 596 | { |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 597 | /* We must promise at least one full hash after the |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 598 | * offset is available from this window, otherwise the offset |
| 599 | * is not actually in this window and a different window (which |
| 600 | * has that one hash excess) must be used. This is to support |
| 601 | * the object header and delta base parsing routines below. |
| 602 | */ |
| 603 | off_t win_off = win->offset; |
| 604 | return win_off <= offset |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 605 | && (offset + the_hash_algo->rawsz) <= (win_off + win->len); |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 606 | } |
| 607 | |
| 608 | unsigned char *use_pack(struct packed_git *p, |
| 609 | struct pack_window **w_cursor, |
| 610 | off_t offset, |
| 611 | unsigned long *left) |
| 612 | { |
| 613 | struct pack_window *win = *w_cursor; |
| 614 | |
| 615 | /* Since packfiles end in a hash of their content and it's |
| 616 | * pointless to ask for an offset into the middle of that |
| 617 | * hash, and the in_window function above wouldn't match |
| 618 | * don't allow an offset too close to the end of the file. |
| 619 | */ |
| 620 | if (!p->pack_size && p->pack_fd == -1 && open_packed_git(p)) |
| 621 | die("packfile %s cannot be accessed", p->pack_name); |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 622 | if (offset > (p->pack_size - the_hash_algo->rawsz)) |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 623 | die("offset beyond end of packfile (truncated pack?)"); |
| 624 | if (offset < 0) |
| 625 | die(_("offset before end of packfile (broken .idx?)")); |
| 626 | |
| 627 | if (!win || !in_window(win, offset)) { |
| 628 | if (win) |
| 629 | win->inuse_cnt--; |
| 630 | for (win = p->windows; win; win = win->next) { |
| 631 | if (in_window(win, offset)) |
| 632 | break; |
| 633 | } |
| 634 | if (!win) { |
| 635 | size_t window_align = packed_git_window_size / 2; |
| 636 | off_t len; |
| 637 | |
| 638 | if (p->pack_fd == -1 && open_packed_git(p)) |
| 639 | die("packfile %s cannot be accessed", p->pack_name); |
| 640 | |
René Scharfe | ca56dad | 2021-03-13 17:17:22 +0100 | [diff] [blame] | 641 | CALLOC_ARRAY(win, 1); |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 642 | win->offset = (offset / window_align) * window_align; |
| 643 | len = p->pack_size - win->offset; |
| 644 | if (len > packed_git_window_size) |
| 645 | len = packed_git_window_size; |
| 646 | win->len = (size_t)len; |
| 647 | pack_mapped += win->len; |
| 648 | while (packed_git_limit < pack_mapped |
| 649 | && unuse_one_window(p)) |
| 650 | ; /* nothing */ |
Mike Hommey | 3203566 | 2019-05-16 09:37:36 +0900 | [diff] [blame] | 651 | win->base = xmmap_gently(NULL, win->len, |
Jonathan Tan | 84f80ad | 2017-08-18 15:20:22 -0700 | [diff] [blame] | 652 | PROT_READ, MAP_PRIVATE, |
| 653 | p->pack_fd, win->offset); |
| 654 | if (win->base == MAP_FAILED) |
| 655 | die_errno("packfile %s cannot be mapped", |
| 656 | p->pack_name); |
| 657 | if (!win->offset && win->len == p->pack_size |
| 658 | && !p->do_not_close) |
| 659 | close_pack_fd(p); |
| 660 | pack_mmap_calls++; |
| 661 | pack_open_windows++; |
| 662 | if (pack_mapped > peak_pack_mapped) |
| 663 | peak_pack_mapped = pack_mapped; |
| 664 | if (pack_open_windows > peak_pack_open_windows) |
| 665 | peak_pack_open_windows = pack_open_windows; |
| 666 | win->next = p->windows; |
| 667 | p->windows = win; |
| 668 | } |
| 669 | } |
| 670 | if (win != *w_cursor) { |
| 671 | win->last_used = pack_used_ctr++; |
| 672 | win->inuse_cnt++; |
| 673 | *w_cursor = win; |
| 674 | } |
| 675 | offset -= win->offset; |
| 676 | if (left) |
| 677 | *left = win->len - xsize_t(offset); |
| 678 | return win->base + offset; |
| 679 | } |
Jonathan Tan | 97de180 | 2017-08-18 15:20:23 -0700 | [diff] [blame] | 680 | |
| 681 | void unuse_pack(struct pack_window **w_cursor) |
| 682 | { |
| 683 | struct pack_window *w = *w_cursor; |
| 684 | if (w) { |
| 685 | w->inuse_cnt--; |
| 686 | *w_cursor = NULL; |
| 687 | } |
| 688 | } |
Jonathan Tan | 9a42865 | 2017-08-18 15:20:24 -0700 | [diff] [blame] | 689 | |
Jonathan Tan | 9a42865 | 2017-08-18 15:20:24 -0700 | [diff] [blame] | 690 | struct packed_git *add_packed_git(const char *path, size_t path_len, int local) |
| 691 | { |
Jonathan Tan | 9a42865 | 2017-08-18 15:20:24 -0700 | [diff] [blame] | 692 | struct stat st; |
| 693 | size_t alloc; |
| 694 | struct packed_git *p; |
| 695 | |
Jonathan Tan | 9a42865 | 2017-08-18 15:20:24 -0700 | [diff] [blame] | 696 | /* |
| 697 | * Make sure a corresponding .pack file exists and that |
| 698 | * the index looks sane. |
| 699 | */ |
| 700 | if (!strip_suffix_mem(path, &path_len, ".idx")) |
| 701 | return NULL; |
| 702 | |
| 703 | /* |
Jonathan Tan | 498f1f6 | 2017-12-05 16:58:44 +0000 | [diff] [blame] | 704 | * ".promisor" is long enough to hold any suffix we're adding (and |
Jonathan Tan | 9a42865 | 2017-08-18 15:20:24 -0700 | [diff] [blame] | 705 | * the use xsnprintf double-checks that) |
| 706 | */ |
Jonathan Tan | 498f1f6 | 2017-12-05 16:58:44 +0000 | [diff] [blame] | 707 | alloc = st_add3(path_len, strlen(".promisor"), 1); |
Jonathan Tan | 9a42865 | 2017-08-18 15:20:24 -0700 | [diff] [blame] | 708 | p = alloc_packed_git(alloc); |
| 709 | memcpy(p->pack_name, path, path_len); |
| 710 | |
| 711 | xsnprintf(p->pack_name + path_len, alloc - path_len, ".keep"); |
| 712 | if (!access(p->pack_name, F_OK)) |
| 713 | p->pack_keep = 1; |
| 714 | |
Jonathan Tan | 498f1f6 | 2017-12-05 16:58:44 +0000 | [diff] [blame] | 715 | xsnprintf(p->pack_name + path_len, alloc - path_len, ".promisor"); |
| 716 | if (!access(p->pack_name, F_OK)) |
| 717 | p->pack_promisor = 1; |
| 718 | |
Jonathan Tan | 9a42865 | 2017-08-18 15:20:24 -0700 | [diff] [blame] | 719 | xsnprintf(p->pack_name + path_len, alloc - path_len, ".pack"); |
| 720 | if (stat(p->pack_name, &st) || !S_ISREG(st.st_mode)) { |
| 721 | free(p); |
| 722 | return NULL; |
| 723 | } |
| 724 | |
| 725 | /* ok, it looks sane as far as we can check without |
| 726 | * actually mapping the pack file. |
| 727 | */ |
| 728 | p->pack_size = st.st_size; |
| 729 | p->pack_local = local; |
| 730 | p->mtime = st.st_mtime; |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 731 | if (path_len < the_hash_algo->hexsz || |
brian m. carlson | 538b152 | 2019-02-19 00:05:03 +0000 | [diff] [blame] | 732 | get_sha1_hex(path + path_len - the_hash_algo->hexsz, p->hash)) |
| 733 | hashclr(p->hash); |
Jonathan Tan | 9a42865 | 2017-08-18 15:20:24 -0700 | [diff] [blame] | 734 | return p; |
| 735 | } |
Jonathan Tan | e65f186 | 2017-08-18 15:20:25 -0700 | [diff] [blame] | 736 | |
Stefan Beller | 5babff1 | 2018-03-23 18:45:18 +0100 | [diff] [blame] | 737 | void install_packed_git(struct repository *r, struct packed_git *pack) |
Jonathan Tan | e65f186 | 2017-08-18 15:20:25 -0700 | [diff] [blame] | 738 | { |
| 739 | if (pack->pack_fd != -1) |
| 740 | pack_open_fds++; |
| 741 | |
Stefan Beller | 5babff1 | 2018-03-23 18:45:18 +0100 | [diff] [blame] | 742 | pack->next = r->objects->packed_git; |
| 743 | r->objects->packed_git = pack; |
Colin Stolley | ec48540 | 2019-11-27 16:24:53 -0600 | [diff] [blame] | 744 | |
| 745 | hashmap_entry_init(&pack->packmap_ent, strhash(pack->pack_name)); |
| 746 | hashmap_add(&r->objects->pack_map, &pack->packmap_ent); |
Jonathan Tan | e65f186 | 2017-08-18 15:20:25 -0700 | [diff] [blame] | 747 | } |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 748 | |
| 749 | void (*report_garbage)(unsigned seen_bits, const char *path); |
| 750 | |
| 751 | static void report_helper(const struct string_list *list, |
| 752 | int seen_bits, int first, int last) |
| 753 | { |
| 754 | if (seen_bits == (PACKDIR_FILE_PACK|PACKDIR_FILE_IDX)) |
| 755 | return; |
| 756 | |
| 757 | for (; first < last; first++) |
| 758 | report_garbage(seen_bits, list->items[first].string); |
| 759 | } |
| 760 | |
| 761 | static void report_pack_garbage(struct string_list *list) |
| 762 | { |
| 763 | int i, baselen = -1, first = 0, seen_bits = 0; |
| 764 | |
| 765 | if (!report_garbage) |
| 766 | return; |
| 767 | |
| 768 | string_list_sort(list); |
| 769 | |
| 770 | for (i = 0; i < list->nr; i++) { |
| 771 | const char *path = list->items[i].string; |
| 772 | if (baselen != -1 && |
| 773 | strncmp(path, list->items[first].string, baselen)) { |
| 774 | report_helper(list, seen_bits, first, i); |
| 775 | baselen = -1; |
| 776 | seen_bits = 0; |
| 777 | } |
| 778 | if (baselen == -1) { |
| 779 | const char *dot = strrchr(path, '.'); |
| 780 | if (!dot) { |
| 781 | report_garbage(PACKDIR_FILE_GARBAGE, path); |
| 782 | continue; |
| 783 | } |
| 784 | baselen = dot - path + 1; |
| 785 | first = i; |
| 786 | } |
| 787 | if (!strcmp(path + baselen, "pack")) |
| 788 | seen_bits |= 1; |
| 789 | else if (!strcmp(path + baselen, "idx")) |
| 790 | seen_bits |= 2; |
| 791 | } |
| 792 | report_helper(list, seen_bits, first, list->nr); |
| 793 | } |
| 794 | |
Derrick Stolee | 9208e31 | 2018-07-12 15:39:25 -0400 | [diff] [blame] | 795 | void for_each_file_in_pack_dir(const char *objdir, |
| 796 | each_file_in_pack_dir_fn fn, |
| 797 | void *data) |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 798 | { |
| 799 | struct strbuf path = STRBUF_INIT; |
| 800 | size_t dirnamelen; |
| 801 | DIR *dir; |
| 802 | struct dirent *de; |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 803 | |
| 804 | strbuf_addstr(&path, objdir); |
| 805 | strbuf_addstr(&path, "/pack"); |
| 806 | dir = opendir(path.buf); |
| 807 | if (!dir) { |
| 808 | if (errno != ENOENT) |
| 809 | error_errno("unable to open object pack directory: %s", |
| 810 | path.buf); |
| 811 | strbuf_release(&path); |
| 812 | return; |
| 813 | } |
| 814 | strbuf_addch(&path, '/'); |
| 815 | dirnamelen = path.len; |
| 816 | while ((de = readdir(dir)) != NULL) { |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 817 | if (is_dot_or_dotdot(de->d_name)) |
| 818 | continue; |
| 819 | |
| 820 | strbuf_setlen(&path, dirnamelen); |
| 821 | strbuf_addstr(&path, de->d_name); |
| 822 | |
Derrick Stolee | 9208e31 | 2018-07-12 15:39:25 -0400 | [diff] [blame] | 823 | fn(path.buf, path.len, de->d_name, data); |
| 824 | } |
| 825 | |
| 826 | closedir(dir); |
| 827 | strbuf_release(&path); |
| 828 | } |
| 829 | |
| 830 | struct prepare_pack_data { |
| 831 | struct repository *r; |
| 832 | struct string_list *garbage; |
| 833 | int local; |
Derrick Stolee | f3a002b | 2018-07-12 15:39:38 -0400 | [diff] [blame] | 834 | struct multi_pack_index *m; |
Derrick Stolee | 9208e31 | 2018-07-12 15:39:25 -0400 | [diff] [blame] | 835 | }; |
| 836 | |
| 837 | static void prepare_pack(const char *full_name, size_t full_name_len, |
| 838 | const char *file_name, void *_data) |
| 839 | { |
| 840 | struct prepare_pack_data *data = (struct prepare_pack_data *)_data; |
| 841 | struct packed_git *p; |
| 842 | size_t base_len = full_name_len; |
| 843 | |
Derrick Stolee | fe86c3b | 2018-08-20 16:51:59 +0000 | [diff] [blame] | 844 | if (strip_suffix_mem(full_name, &base_len, ".idx") && |
| 845 | !(data->m && midx_contains_pack(data->m, file_name))) { |
Colin Stolley | ec48540 | 2019-11-27 16:24:53 -0600 | [diff] [blame] | 846 | struct hashmap_entry hent; |
| 847 | char *pack_name = xstrfmt("%.*s.pack", (int)base_len, full_name); |
| 848 | unsigned int hash = strhash(pack_name); |
| 849 | hashmap_entry_init(&hent, hash); |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 850 | |
Colin Stolley | ec48540 | 2019-11-27 16:24:53 -0600 | [diff] [blame] | 851 | /* Don't reopen a pack we already have. */ |
| 852 | if (!hashmap_get(&data->r->objects->pack_map, &hent, pack_name)) { |
Derrick Stolee | 9208e31 | 2018-07-12 15:39:25 -0400 | [diff] [blame] | 853 | p = add_packed_git(full_name, full_name_len, data->local); |
| 854 | if (p) |
| 855 | install_packed_git(data->r, p); |
| 856 | } |
Colin Stolley | ec48540 | 2019-11-27 16:24:53 -0600 | [diff] [blame] | 857 | free(pack_name); |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 858 | } |
Derrick Stolee | 9208e31 | 2018-07-12 15:39:25 -0400 | [diff] [blame] | 859 | |
| 860 | if (!report_garbage) |
| 861 | return; |
| 862 | |
Derrick Stolee | fe86c3b | 2018-08-20 16:51:59 +0000 | [diff] [blame] | 863 | if (!strcmp(file_name, "multi-pack-index")) |
| 864 | return; |
Taylor Blau | f894081 | 2021-03-30 11:04:26 -0400 | [diff] [blame] | 865 | if (starts_with(file_name, "multi-pack-index") && |
| 866 | ends_with(file_name, ".rev")) |
| 867 | return; |
Derrick Stolee | 9208e31 | 2018-07-12 15:39:25 -0400 | [diff] [blame] | 868 | if (ends_with(file_name, ".idx") || |
Taylor Blau | 2f4ba2a | 2021-01-25 18:37:14 -0500 | [diff] [blame] | 869 | ends_with(file_name, ".rev") || |
Derrick Stolee | 9208e31 | 2018-07-12 15:39:25 -0400 | [diff] [blame] | 870 | ends_with(file_name, ".pack") || |
| 871 | ends_with(file_name, ".bitmap") || |
| 872 | ends_with(file_name, ".keep") || |
| 873 | ends_with(file_name, ".promisor")) |
| 874 | string_list_append(data->garbage, full_name); |
| 875 | else |
| 876 | report_garbage(PACKDIR_FILE_GARBAGE, full_name); |
| 877 | } |
| 878 | |
| 879 | static void prepare_packed_git_one(struct repository *r, char *objdir, int local) |
| 880 | { |
| 881 | struct prepare_pack_data data; |
| 882 | struct string_list garbage = STRING_LIST_INIT_DUP; |
| 883 | |
Derrick Stolee | f3a002b | 2018-07-12 15:39:38 -0400 | [diff] [blame] | 884 | data.m = r->objects->multi_pack_index; |
| 885 | |
| 886 | /* look for the multi-pack-index for this object directory */ |
| 887 | while (data.m && strcmp(data.m->object_dir, objdir)) |
| 888 | data.m = data.m->next; |
| 889 | |
Derrick Stolee | 9208e31 | 2018-07-12 15:39:25 -0400 | [diff] [blame] | 890 | data.r = r; |
| 891 | data.garbage = &garbage; |
| 892 | data.local = local; |
| 893 | |
| 894 | for_each_file_in_pack_dir(objdir, prepare_pack, &data); |
| 895 | |
| 896 | report_pack_garbage(data.garbage); |
| 897 | string_list_clear(data.garbage, 0); |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 898 | } |
| 899 | |
Nguyễn Thái Ngọc Duy | 464416a | 2018-03-23 18:45:27 +0100 | [diff] [blame] | 900 | static void prepare_packed_git(struct repository *r); |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 901 | /* |
| 902 | * Give a fast, rough count of the number of objects in the repository. This |
| 903 | * ignores loose objects completely. If you have a lot of them, then either |
| 904 | * you should repack because your performance will be awful, or they are |
| 905 | * all unreachable objects about to be pruned, in which case they're not really |
| 906 | * interesting as a measure of repo size in the first place. |
| 907 | */ |
Nguyễn Thái Ngọc Duy | 5038de1 | 2019-04-06 18:34:23 +0700 | [diff] [blame] | 908 | unsigned long repo_approximate_object_count(struct repository *r) |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 909 | { |
Nguyễn Thái Ngọc Duy | 5038de1 | 2019-04-06 18:34:23 +0700 | [diff] [blame] | 910 | if (!r->objects->approximate_object_count_valid) { |
Stefan Beller | 9a00580 | 2018-03-23 18:21:02 +0100 | [diff] [blame] | 911 | unsigned long count; |
Derrick Stolee | b8990fb | 2018-07-12 15:39:37 -0400 | [diff] [blame] | 912 | struct multi_pack_index *m; |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 913 | struct packed_git *p; |
| 914 | |
Nguyễn Thái Ngọc Duy | 5038de1 | 2019-04-06 18:34:23 +0700 | [diff] [blame] | 915 | prepare_packed_git(r); |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 916 | count = 0; |
Nguyễn Thái Ngọc Duy | 5038de1 | 2019-04-06 18:34:23 +0700 | [diff] [blame] | 917 | for (m = get_multi_pack_index(r); m; m = m->next) |
Derrick Stolee | b8990fb | 2018-07-12 15:39:37 -0400 | [diff] [blame] | 918 | count += m->num_objects; |
Nguyễn Thái Ngọc Duy | 5038de1 | 2019-04-06 18:34:23 +0700 | [diff] [blame] | 919 | for (p = r->objects->packed_git; p; p = p->next) { |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 920 | if (open_pack_index(p)) |
| 921 | continue; |
| 922 | count += p->num_objects; |
| 923 | } |
Nguyễn Thái Ngọc Duy | 5038de1 | 2019-04-06 18:34:23 +0700 | [diff] [blame] | 924 | r->objects->approximate_object_count = count; |
Jeff King | 67bb65d | 2020-09-17 12:47:43 -0400 | [diff] [blame] | 925 | r->objects->approximate_object_count_valid = 1; |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 926 | } |
Nguyễn Thái Ngọc Duy | 5038de1 | 2019-04-06 18:34:23 +0700 | [diff] [blame] | 927 | return r->objects->approximate_object_count; |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 928 | } |
| 929 | |
| 930 | static void *get_next_packed_git(const void *p) |
| 931 | { |
| 932 | return ((const struct packed_git *)p)->next; |
| 933 | } |
| 934 | |
| 935 | static void set_next_packed_git(void *p, void *next) |
| 936 | { |
| 937 | ((struct packed_git *)p)->next = next; |
| 938 | } |
| 939 | |
| 940 | static int sort_pack(const void *a_, const void *b_) |
| 941 | { |
| 942 | const struct packed_git *a = a_; |
| 943 | const struct packed_git *b = b_; |
| 944 | int st; |
| 945 | |
| 946 | /* |
| 947 | * Local packs tend to contain objects specific to our |
| 948 | * variant of the project than remote ones. In addition, |
| 949 | * remote ones could be on a network mounted filesystem. |
| 950 | * Favor local ones for these reasons. |
| 951 | */ |
| 952 | st = a->pack_local - b->pack_local; |
| 953 | if (st) |
| 954 | return -st; |
| 955 | |
| 956 | /* |
| 957 | * Younger packs tend to contain more recent objects, |
| 958 | * and more recent objects tend to get accessed more |
| 959 | * often. |
| 960 | */ |
| 961 | if (a->mtime < b->mtime) |
| 962 | return 1; |
| 963 | else if (a->mtime == b->mtime) |
| 964 | return 0; |
| 965 | return -1; |
| 966 | } |
| 967 | |
Stefan Beller | c235bea | 2018-03-23 18:45:17 +0100 | [diff] [blame] | 968 | static void rearrange_packed_git(struct repository *r) |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 969 | { |
Stefan Beller | c235bea | 2018-03-23 18:45:17 +0100 | [diff] [blame] | 970 | r->objects->packed_git = llist_mergesort( |
| 971 | r->objects->packed_git, get_next_packed_git, |
Stefan Beller | a80d72d | 2018-03-23 18:20:59 +0100 | [diff] [blame] | 972 | set_next_packed_git, sort_pack); |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 973 | } |
| 974 | |
Stefan Beller | 804be79 | 2018-03-23 18:45:16 +0100 | [diff] [blame] | 975 | static void prepare_packed_git_mru(struct repository *r) |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 976 | { |
| 977 | struct packed_git *p; |
| 978 | |
Stefan Beller | 804be79 | 2018-03-23 18:45:16 +0100 | [diff] [blame] | 979 | INIT_LIST_HEAD(&r->objects->packed_git_mru); |
Gargi Sharma | ec2dd32 | 2018-01-23 18:46:51 -0500 | [diff] [blame] | 980 | |
Stefan Beller | 804be79 | 2018-03-23 18:45:16 +0100 | [diff] [blame] | 981 | for (p = r->objects->packed_git; p; p = p->next) |
| 982 | list_add_tail(&p->mru, &r->objects->packed_git_mru); |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 983 | } |
| 984 | |
Nguyễn Thái Ngọc Duy | 464416a | 2018-03-23 18:45:27 +0100 | [diff] [blame] | 985 | static void prepare_packed_git(struct repository *r) |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 986 | { |
Jeff King | 263db40 | 2018-11-12 09:48:47 -0500 | [diff] [blame] | 987 | struct object_directory *odb; |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 988 | |
Stefan Beller | 0f90a9f | 2018-03-23 18:45:23 +0100 | [diff] [blame] | 989 | if (r->objects->packed_git_initialized) |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 990 | return; |
Jeff King | f0eaf63 | 2018-11-12 09:50:39 -0500 | [diff] [blame] | 991 | |
Stefan Beller | 0f90a9f | 2018-03-23 18:45:23 +0100 | [diff] [blame] | 992 | prepare_alt_odb(r); |
Jeff King | f0eaf63 | 2018-11-12 09:50:39 -0500 | [diff] [blame] | 993 | for (odb = r->objects->odb; odb; odb = odb->next) { |
| 994 | int local = (odb == r->objects->odb); |
| 995 | prepare_multi_pack_index_one(r, odb->path, local); |
| 996 | prepare_packed_git_one(r, odb->path, local); |
Derrick Stolee | c4d2522 | 2018-07-12 15:39:33 -0400 | [diff] [blame] | 997 | } |
Stefan Beller | 0f90a9f | 2018-03-23 18:45:23 +0100 | [diff] [blame] | 998 | rearrange_packed_git(r); |
Derrick Stolee | 0bff526 | 2018-08-20 16:52:02 +0000 | [diff] [blame] | 999 | |
Stefan Beller | 0f90a9f | 2018-03-23 18:45:23 +0100 | [diff] [blame] | 1000 | prepare_packed_git_mru(r); |
| 1001 | r->objects->packed_git_initialized = 1; |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 1002 | } |
| 1003 | |
Stefan Beller | 4c2a13b | 2018-03-23 18:45:24 +0100 | [diff] [blame] | 1004 | void reprepare_packed_git(struct repository *r) |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 1005 | { |
Jeff King | 3a2e082 | 2018-11-12 09:50:56 -0500 | [diff] [blame] | 1006 | struct object_directory *odb; |
| 1007 | |
Matheus Tavares | 6c30762 | 2020-01-15 23:39:57 -0300 | [diff] [blame] | 1008 | obj_read_lock(); |
René Scharfe | d4e19e5 | 2019-01-06 17:45:39 +0100 | [diff] [blame] | 1009 | for (odb = r->objects->odb; odb; odb = odb->next) |
| 1010 | odb_clear_loose_cache(odb); |
Jeff King | 3a2e082 | 2018-11-12 09:50:56 -0500 | [diff] [blame] | 1011 | |
Stefan Beller | 4c2a13b | 2018-03-23 18:45:24 +0100 | [diff] [blame] | 1012 | r->objects->approximate_object_count_valid = 0; |
| 1013 | r->objects->packed_git_initialized = 0; |
| 1014 | prepare_packed_git(r); |
Matheus Tavares | 6c30762 | 2020-01-15 23:39:57 -0300 | [diff] [blame] | 1015 | obj_read_unlock(); |
Jonathan Tan | 0abe14f | 2017-08-18 15:20:26 -0700 | [diff] [blame] | 1016 | } |
Jonathan Tan | 32b42e1 | 2017-08-18 15:20:27 -0700 | [diff] [blame] | 1017 | |
Stefan Beller | a80d72d | 2018-03-23 18:20:59 +0100 | [diff] [blame] | 1018 | struct packed_git *get_packed_git(struct repository *r) |
| 1019 | { |
Nguyễn Thái Ngọc Duy | 464416a | 2018-03-23 18:45:27 +0100 | [diff] [blame] | 1020 | prepare_packed_git(r); |
Stefan Beller | a80d72d | 2018-03-23 18:20:59 +0100 | [diff] [blame] | 1021 | return r->objects->packed_git; |
| 1022 | } |
| 1023 | |
Derrick Stolee | 8aac67a | 2018-07-12 15:39:35 -0400 | [diff] [blame] | 1024 | struct multi_pack_index *get_multi_pack_index(struct repository *r) |
| 1025 | { |
| 1026 | prepare_packed_git(r); |
| 1027 | return r->objects->multi_pack_index; |
| 1028 | } |
| 1029 | |
Taylor Blau | 59552fb | 2020-08-28 16:22:13 -0400 | [diff] [blame] | 1030 | struct multi_pack_index *get_local_multi_pack_index(struct repository *r) |
| 1031 | { |
| 1032 | struct multi_pack_index *m = get_multi_pack_index(r); |
| 1033 | |
| 1034 | /* no need to iterate; we always put the local one first (if any) */ |
| 1035 | if (m && m->local) |
| 1036 | return m; |
| 1037 | |
| 1038 | return NULL; |
| 1039 | } |
| 1040 | |
Derrick Stolee | 0bff526 | 2018-08-20 16:52:02 +0000 | [diff] [blame] | 1041 | struct packed_git *get_all_packs(struct repository *r) |
| 1042 | { |
Derrick Stolee | af96fe3 | 2019-04-29 09:18:56 -0700 | [diff] [blame] | 1043 | struct multi_pack_index *m; |
| 1044 | |
Derrick Stolee | 0bff526 | 2018-08-20 16:52:02 +0000 | [diff] [blame] | 1045 | prepare_packed_git(r); |
Derrick Stolee | af96fe3 | 2019-04-29 09:18:56 -0700 | [diff] [blame] | 1046 | for (m = r->objects->multi_pack_index; m; m = m->next) { |
| 1047 | uint32_t i; |
| 1048 | for (i = 0; i < m->num_packs; i++) |
| 1049 | prepare_midx_pack(r, m, i); |
Derrick Stolee | 0bff526 | 2018-08-20 16:52:02 +0000 | [diff] [blame] | 1050 | } |
| 1051 | |
Derrick Stolee | af96fe3 | 2019-04-29 09:18:56 -0700 | [diff] [blame] | 1052 | return r->objects->packed_git; |
Derrick Stolee | 0bff526 | 2018-08-20 16:52:02 +0000 | [diff] [blame] | 1053 | } |
| 1054 | |
Stefan Beller | a80d72d | 2018-03-23 18:20:59 +0100 | [diff] [blame] | 1055 | struct list_head *get_packed_git_mru(struct repository *r) |
| 1056 | { |
Nguyễn Thái Ngọc Duy | 464416a | 2018-03-23 18:45:27 +0100 | [diff] [blame] | 1057 | prepare_packed_git(r); |
Stefan Beller | a80d72d | 2018-03-23 18:20:59 +0100 | [diff] [blame] | 1058 | return &r->objects->packed_git_mru; |
| 1059 | } |
| 1060 | |
Jonathan Tan | 32b42e1 | 2017-08-18 15:20:27 -0700 | [diff] [blame] | 1061 | unsigned long unpack_object_header_buffer(const unsigned char *buf, |
| 1062 | unsigned long len, enum object_type *type, unsigned long *sizep) |
| 1063 | { |
| 1064 | unsigned shift; |
| 1065 | unsigned long size, c; |
| 1066 | unsigned long used = 0; |
| 1067 | |
| 1068 | c = buf[used++]; |
| 1069 | *type = (c >> 4) & 7; |
| 1070 | size = c & 15; |
| 1071 | shift = 4; |
| 1072 | while (c & 0x80) { |
| 1073 | if (len <= used || bitsizeof(long) <= shift) { |
| 1074 | error("bad object header"); |
| 1075 | size = used = 0; |
| 1076 | break; |
| 1077 | } |
| 1078 | c = buf[used++]; |
| 1079 | size += (c & 0x7f) << shift; |
| 1080 | shift += 7; |
| 1081 | } |
| 1082 | *sizep = size; |
| 1083 | return used; |
| 1084 | } |
Jonathan Tan | 7b3aa75 | 2017-08-18 15:20:28 -0700 | [diff] [blame] | 1085 | |
| 1086 | unsigned long get_size_from_delta(struct packed_git *p, |
| 1087 | struct pack_window **w_curs, |
| 1088 | off_t curpos) |
| 1089 | { |
| 1090 | const unsigned char *data; |
| 1091 | unsigned char delta_head[20], *in; |
| 1092 | git_zstream stream; |
| 1093 | int st; |
| 1094 | |
| 1095 | memset(&stream, 0, sizeof(stream)); |
| 1096 | stream.next_out = delta_head; |
| 1097 | stream.avail_out = sizeof(delta_head); |
| 1098 | |
| 1099 | git_inflate_init(&stream); |
| 1100 | do { |
| 1101 | in = use_pack(p, w_curs, curpos, &stream.avail_in); |
| 1102 | stream.next_in = in; |
Matheus Tavares | 31877c9 | 2020-01-15 23:39:53 -0300 | [diff] [blame] | 1103 | /* |
| 1104 | * Note: the window section returned by use_pack() must be |
| 1105 | * available throughout git_inflate()'s unlocked execution. To |
| 1106 | * ensure no other thread will modify the window in the |
| 1107 | * meantime, we rely on the packed_window.inuse_cnt. This |
| 1108 | * counter is incremented before window reading and checked |
| 1109 | * before window disposal. |
| 1110 | * |
| 1111 | * Other worrying sections could be the call to close_pack_fd(), |
| 1112 | * which can close packs even with in-use windows, and to |
| 1113 | * reprepare_packed_git(). Regarding the former, mmap doc says: |
| 1114 | * "closing the file descriptor does not unmap the region". And |
| 1115 | * for the latter, it won't re-open already available packs. |
| 1116 | */ |
| 1117 | obj_read_unlock(); |
Jonathan Tan | 7b3aa75 | 2017-08-18 15:20:28 -0700 | [diff] [blame] | 1118 | st = git_inflate(&stream, Z_FINISH); |
Matheus Tavares | 31877c9 | 2020-01-15 23:39:53 -0300 | [diff] [blame] | 1119 | obj_read_lock(); |
Jonathan Tan | 7b3aa75 | 2017-08-18 15:20:28 -0700 | [diff] [blame] | 1120 | curpos += stream.next_in - in; |
| 1121 | } while ((st == Z_OK || st == Z_BUF_ERROR) && |
| 1122 | stream.total_out < sizeof(delta_head)); |
| 1123 | git_inflate_end(&stream); |
| 1124 | if ((st != Z_STREAM_END) && stream.total_out != sizeof(delta_head)) { |
| 1125 | error("delta data unpack-initial failed"); |
| 1126 | return 0; |
| 1127 | } |
| 1128 | |
| 1129 | /* Examine the initial part of the delta to figure out |
| 1130 | * the result size. |
| 1131 | */ |
| 1132 | data = delta_head; |
| 1133 | |
| 1134 | /* ignore base size */ |
| 1135 | get_delta_hdr_size(&data, delta_head+sizeof(delta_head)); |
| 1136 | |
| 1137 | /* Read the result size */ |
| 1138 | return get_delta_hdr_size(&data, delta_head+sizeof(delta_head)); |
| 1139 | } |
Jonathan Tan | 3588dd6 | 2017-08-18 15:20:29 -0700 | [diff] [blame] | 1140 | |
| 1141 | int unpack_object_header(struct packed_git *p, |
| 1142 | struct pack_window **w_curs, |
| 1143 | off_t *curpos, |
| 1144 | unsigned long *sizep) |
| 1145 | { |
| 1146 | unsigned char *base; |
| 1147 | unsigned long left; |
| 1148 | unsigned long used; |
| 1149 | enum object_type type; |
| 1150 | |
| 1151 | /* use_pack() assures us we have [base, base + 20) available |
| 1152 | * as a range that we can look at. (Its actually the hash |
| 1153 | * size that is assured.) With our object header encoding |
| 1154 | * the maximum deflated object size is 2^137, which is just |
| 1155 | * insane, so we know won't exceed what we have been given. |
| 1156 | */ |
| 1157 | base = use_pack(p, w_curs, *curpos, &left); |
| 1158 | used = unpack_object_header_buffer(base, left, &type, sizep); |
| 1159 | if (!used) { |
| 1160 | type = OBJ_BAD; |
| 1161 | } else |
| 1162 | *curpos += used; |
| 1163 | |
| 1164 | return type; |
| 1165 | } |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1166 | |
| 1167 | void mark_bad_packed_object(struct packed_git *p, const unsigned char *sha1) |
| 1168 | { |
| 1169 | unsigned i; |
brian m. carlson | 268babd | 2018-10-15 00:01:54 +0000 | [diff] [blame] | 1170 | const unsigned hashsz = the_hash_algo->rawsz; |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1171 | for (i = 0; i < p->num_bad_objects; i++) |
brian m. carlson | 268babd | 2018-10-15 00:01:54 +0000 | [diff] [blame] | 1172 | if (hasheq(sha1, p->bad_object_sha1 + hashsz * i)) |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1173 | return; |
| 1174 | p->bad_object_sha1 = xrealloc(p->bad_object_sha1, |
| 1175 | st_mult(GIT_MAX_RAWSZ, |
| 1176 | st_add(p->num_bad_objects, 1))); |
brian m. carlson | 268babd | 2018-10-15 00:01:54 +0000 | [diff] [blame] | 1177 | hashcpy(p->bad_object_sha1 + hashsz * p->num_bad_objects, sha1); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1178 | p->num_bad_objects++; |
| 1179 | } |
| 1180 | |
Stefan Beller | 33b9406 | 2018-10-16 16:35:33 -0700 | [diff] [blame] | 1181 | const struct packed_git *has_packed_and_bad(struct repository *r, |
| 1182 | const unsigned char *sha1) |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1183 | { |
| 1184 | struct packed_git *p; |
| 1185 | unsigned i; |
| 1186 | |
Stefan Beller | 33b9406 | 2018-10-16 16:35:33 -0700 | [diff] [blame] | 1187 | for (p = r->objects->packed_git; p; p = p->next) |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1188 | for (i = 0; i < p->num_bad_objects; i++) |
Jeff King | e3ff068 | 2018-08-28 17:22:44 -0400 | [diff] [blame] | 1189 | if (hasheq(sha1, |
| 1190 | p->bad_object_sha1 + the_hash_algo->rawsz * i)) |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1191 | return p; |
| 1192 | return NULL; |
| 1193 | } |
| 1194 | |
Jeff King | 56d9cbe | 2019-09-13 15:02:18 +0200 | [diff] [blame] | 1195 | off_t get_delta_base(struct packed_git *p, |
| 1196 | struct pack_window **w_curs, |
| 1197 | off_t *curpos, |
| 1198 | enum object_type type, |
| 1199 | off_t delta_obj_offset) |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1200 | { |
| 1201 | unsigned char *base_info = use_pack(p, w_curs, *curpos, NULL); |
| 1202 | off_t base_offset; |
| 1203 | |
| 1204 | /* use_pack() assured us we have [base_info, base_info + 20) |
| 1205 | * as a range that we can look at without walking off the |
| 1206 | * end of the mapped window. Its actually the hash size |
| 1207 | * that is assured. An OFS_DELTA longer than the hash size |
| 1208 | * is stupid, as then a REF_DELTA would be smaller to store. |
| 1209 | */ |
| 1210 | if (type == OBJ_OFS_DELTA) { |
| 1211 | unsigned used = 0; |
| 1212 | unsigned char c = base_info[used++]; |
| 1213 | base_offset = c & 127; |
| 1214 | while (c & 128) { |
| 1215 | base_offset += 1; |
| 1216 | if (!base_offset || MSB(base_offset, 7)) |
| 1217 | return 0; /* overflow */ |
| 1218 | c = base_info[used++]; |
| 1219 | base_offset = (base_offset << 7) + (c & 127); |
| 1220 | } |
| 1221 | base_offset = delta_obj_offset - base_offset; |
| 1222 | if (base_offset <= 0 || base_offset >= delta_obj_offset) |
| 1223 | return 0; /* out of bound */ |
| 1224 | *curpos += used; |
| 1225 | } else if (type == OBJ_REF_DELTA) { |
| 1226 | /* The base entry _must_ be in the same pack */ |
| 1227 | base_offset = find_pack_entry_one(base_info, p); |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 1228 | *curpos += the_hash_algo->rawsz; |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1229 | } else |
| 1230 | die("I am totally screwed"); |
| 1231 | return base_offset; |
| 1232 | } |
| 1233 | |
| 1234 | /* |
| 1235 | * Like get_delta_base above, but we return the sha1 instead of the pack |
| 1236 | * offset. This means it is cheaper for REF deltas (we do not have to do |
| 1237 | * the final object lookup), but more expensive for OFS deltas (we |
| 1238 | * have to load the revidx to convert the offset back into a sha1). |
| 1239 | */ |
Jeff King | 6ac9760 | 2020-02-23 23:37:31 -0500 | [diff] [blame] | 1240 | static int get_delta_base_oid(struct packed_git *p, |
| 1241 | struct pack_window **w_curs, |
| 1242 | off_t curpos, |
| 1243 | struct object_id *oid, |
| 1244 | enum object_type type, |
| 1245 | off_t delta_obj_offset) |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1246 | { |
| 1247 | if (type == OBJ_REF_DELTA) { |
| 1248 | unsigned char *base = use_pack(p, w_curs, curpos, NULL); |
Jeff King | 6ac9760 | 2020-02-23 23:37:31 -0500 | [diff] [blame] | 1249 | oidread(oid, base); |
| 1250 | return 0; |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1251 | } else if (type == OBJ_OFS_DELTA) { |
Taylor Blau | 45bef5c | 2021-01-13 17:24:32 -0500 | [diff] [blame] | 1252 | uint32_t base_pos; |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1253 | off_t base_offset = get_delta_base(p, w_curs, &curpos, |
| 1254 | type, delta_obj_offset); |
| 1255 | |
| 1256 | if (!base_offset) |
Jeff King | 6ac9760 | 2020-02-23 23:37:31 -0500 | [diff] [blame] | 1257 | return -1; |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1258 | |
Taylor Blau | 45bef5c | 2021-01-13 17:24:32 -0500 | [diff] [blame] | 1259 | if (offset_to_pack_pos(p, base_offset, &base_pos) < 0) |
Jeff King | 6ac9760 | 2020-02-23 23:37:31 -0500 | [diff] [blame] | 1260 | return -1; |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1261 | |
Taylor Blau | 45bef5c | 2021-01-13 17:24:32 -0500 | [diff] [blame] | 1262 | return nth_packed_object_id(oid, p, |
| 1263 | pack_pos_to_index(p, base_pos)); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1264 | } else |
Jeff King | 6ac9760 | 2020-02-23 23:37:31 -0500 | [diff] [blame] | 1265 | return -1; |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1266 | } |
| 1267 | |
Stefan Beller | 9d98354 | 2018-04-25 11:21:06 -0700 | [diff] [blame] | 1268 | static int retry_bad_packed_offset(struct repository *r, |
| 1269 | struct packed_git *p, |
| 1270 | off_t obj_offset) |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1271 | { |
| 1272 | int type; |
Taylor Blau | 3a3f54d | 2021-01-13 17:24:36 -0500 | [diff] [blame] | 1273 | uint32_t pos; |
brian m. carlson | d169d66 | 2018-03-12 02:27:43 +0000 | [diff] [blame] | 1274 | struct object_id oid; |
Taylor Blau | 3a3f54d | 2021-01-13 17:24:36 -0500 | [diff] [blame] | 1275 | if (offset_to_pack_pos(p, obj_offset, &pos) < 0) |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1276 | return OBJ_BAD; |
Taylor Blau | 3a3f54d | 2021-01-13 17:24:36 -0500 | [diff] [blame] | 1277 | nth_packed_object_id(&oid, p, pack_pos_to_index(p, pos)); |
brian m. carlson | d169d66 | 2018-03-12 02:27:43 +0000 | [diff] [blame] | 1278 | mark_bad_packed_object(p, oid.hash); |
Stefan Beller | 9d98354 | 2018-04-25 11:21:06 -0700 | [diff] [blame] | 1279 | type = oid_object_info(r, &oid, NULL); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1280 | if (type <= OBJ_NONE) |
| 1281 | return OBJ_BAD; |
| 1282 | return type; |
| 1283 | } |
| 1284 | |
| 1285 | #define POI_STACK_PREALLOC 64 |
| 1286 | |
Stefan Beller | 9d98354 | 2018-04-25 11:21:06 -0700 | [diff] [blame] | 1287 | static enum object_type packed_to_object_type(struct repository *r, |
| 1288 | struct packed_git *p, |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1289 | off_t obj_offset, |
| 1290 | enum object_type type, |
| 1291 | struct pack_window **w_curs, |
| 1292 | off_t curpos) |
| 1293 | { |
| 1294 | off_t small_poi_stack[POI_STACK_PREALLOC]; |
| 1295 | off_t *poi_stack = small_poi_stack; |
| 1296 | int poi_stack_nr = 0, poi_stack_alloc = POI_STACK_PREALLOC; |
| 1297 | |
| 1298 | while (type == OBJ_OFS_DELTA || type == OBJ_REF_DELTA) { |
| 1299 | off_t base_offset; |
| 1300 | unsigned long size; |
| 1301 | /* Push the object we're going to leave behind */ |
| 1302 | if (poi_stack_nr >= poi_stack_alloc && poi_stack == small_poi_stack) { |
| 1303 | poi_stack_alloc = alloc_nr(poi_stack_nr); |
| 1304 | ALLOC_ARRAY(poi_stack, poi_stack_alloc); |
René Scharfe | 921d49b | 2019-06-15 20:36:35 +0200 | [diff] [blame] | 1305 | COPY_ARRAY(poi_stack, small_poi_stack, poi_stack_nr); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1306 | } else { |
| 1307 | ALLOC_GROW(poi_stack, poi_stack_nr+1, poi_stack_alloc); |
| 1308 | } |
| 1309 | poi_stack[poi_stack_nr++] = obj_offset; |
| 1310 | /* If parsing the base offset fails, just unwind */ |
| 1311 | base_offset = get_delta_base(p, w_curs, &curpos, type, obj_offset); |
| 1312 | if (!base_offset) |
| 1313 | goto unwind; |
| 1314 | curpos = obj_offset = base_offset; |
| 1315 | type = unpack_object_header(p, w_curs, &curpos, &size); |
| 1316 | if (type <= OBJ_NONE) { |
| 1317 | /* If getting the base itself fails, we first |
| 1318 | * retry the base, otherwise unwind */ |
Stefan Beller | 9d98354 | 2018-04-25 11:21:06 -0700 | [diff] [blame] | 1319 | type = retry_bad_packed_offset(r, p, base_offset); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1320 | if (type > OBJ_NONE) |
| 1321 | goto out; |
| 1322 | goto unwind; |
| 1323 | } |
| 1324 | } |
| 1325 | |
| 1326 | switch (type) { |
| 1327 | case OBJ_BAD: |
| 1328 | case OBJ_COMMIT: |
| 1329 | case OBJ_TREE: |
| 1330 | case OBJ_BLOB: |
| 1331 | case OBJ_TAG: |
| 1332 | break; |
| 1333 | default: |
| 1334 | error("unknown object type %i at offset %"PRIuMAX" in %s", |
| 1335 | type, (uintmax_t)obj_offset, p->pack_name); |
| 1336 | type = OBJ_BAD; |
| 1337 | } |
| 1338 | |
| 1339 | out: |
| 1340 | if (poi_stack != small_poi_stack) |
| 1341 | free(poi_stack); |
| 1342 | return type; |
| 1343 | |
| 1344 | unwind: |
| 1345 | while (poi_stack_nr) { |
| 1346 | obj_offset = poi_stack[--poi_stack_nr]; |
Stefan Beller | 9d98354 | 2018-04-25 11:21:06 -0700 | [diff] [blame] | 1347 | type = retry_bad_packed_offset(r, p, obj_offset); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1348 | if (type > OBJ_NONE) |
| 1349 | goto out; |
| 1350 | } |
| 1351 | type = OBJ_BAD; |
| 1352 | goto out; |
| 1353 | } |
| 1354 | |
| 1355 | static struct hashmap delta_base_cache; |
| 1356 | static size_t delta_base_cached; |
| 1357 | |
| 1358 | static LIST_HEAD(delta_base_cache_lru); |
| 1359 | |
| 1360 | struct delta_base_cache_key { |
| 1361 | struct packed_git *p; |
| 1362 | off_t base_offset; |
| 1363 | }; |
| 1364 | |
| 1365 | struct delta_base_cache_entry { |
Eric Wong | d0a48a0 | 2019-10-06 23:30:26 +0000 | [diff] [blame] | 1366 | struct hashmap_entry ent; |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1367 | struct delta_base_cache_key key; |
| 1368 | struct list_head lru; |
| 1369 | void *data; |
| 1370 | unsigned long size; |
| 1371 | enum object_type type; |
| 1372 | }; |
| 1373 | |
| 1374 | static unsigned int pack_entry_hash(struct packed_git *p, off_t base_offset) |
| 1375 | { |
| 1376 | unsigned int hash; |
| 1377 | |
| 1378 | hash = (unsigned int)(intptr_t)p + (unsigned int)base_offset; |
| 1379 | hash += (hash >> 8) + (hash >> 16); |
| 1380 | return hash; |
| 1381 | } |
| 1382 | |
| 1383 | static struct delta_base_cache_entry * |
| 1384 | get_delta_base_cache_entry(struct packed_git *p, off_t base_offset) |
| 1385 | { |
Eric Wong | f23a465 | 2019-10-06 23:30:36 +0000 | [diff] [blame] | 1386 | struct hashmap_entry entry, *e; |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1387 | struct delta_base_cache_key key; |
| 1388 | |
| 1389 | if (!delta_base_cache.cmpfn) |
| 1390 | return NULL; |
| 1391 | |
| 1392 | hashmap_entry_init(&entry, pack_entry_hash(p, base_offset)); |
| 1393 | key.p = p; |
| 1394 | key.base_offset = base_offset; |
Eric Wong | f23a465 | 2019-10-06 23:30:36 +0000 | [diff] [blame] | 1395 | e = hashmap_get(&delta_base_cache, &entry, &key); |
| 1396 | return e ? container_of(e, struct delta_base_cache_entry, ent) : NULL; |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1397 | } |
| 1398 | |
| 1399 | static int delta_base_cache_key_eq(const struct delta_base_cache_key *a, |
| 1400 | const struct delta_base_cache_key *b) |
| 1401 | { |
| 1402 | return a->p == b->p && a->base_offset == b->base_offset; |
| 1403 | } |
| 1404 | |
| 1405 | static int delta_base_cache_hash_cmp(const void *unused_cmp_data, |
Eric Wong | 939af16 | 2019-10-06 23:30:37 +0000 | [diff] [blame] | 1406 | const struct hashmap_entry *va, |
| 1407 | const struct hashmap_entry *vb, |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1408 | const void *vkey) |
| 1409 | { |
Eric Wong | 939af16 | 2019-10-06 23:30:37 +0000 | [diff] [blame] | 1410 | const struct delta_base_cache_entry *a, *b; |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1411 | const struct delta_base_cache_key *key = vkey; |
Eric Wong | 939af16 | 2019-10-06 23:30:37 +0000 | [diff] [blame] | 1412 | |
| 1413 | a = container_of(va, const struct delta_base_cache_entry, ent); |
| 1414 | b = container_of(vb, const struct delta_base_cache_entry, ent); |
| 1415 | |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1416 | if (key) |
| 1417 | return !delta_base_cache_key_eq(&a->key, key); |
| 1418 | else |
| 1419 | return !delta_base_cache_key_eq(&a->key, &b->key); |
| 1420 | } |
| 1421 | |
| 1422 | static int in_delta_base_cache(struct packed_git *p, off_t base_offset) |
| 1423 | { |
| 1424 | return !!get_delta_base_cache_entry(p, base_offset); |
| 1425 | } |
| 1426 | |
| 1427 | /* |
| 1428 | * Remove the entry from the cache, but do _not_ free the associated |
| 1429 | * entry data. The caller takes ownership of the "data" buffer, and |
| 1430 | * should copy out any fields it wants before detaching. |
| 1431 | */ |
| 1432 | static void detach_delta_base_cache_entry(struct delta_base_cache_entry *ent) |
| 1433 | { |
Eric Wong | 28ee794 | 2019-10-06 23:30:31 +0000 | [diff] [blame] | 1434 | hashmap_remove(&delta_base_cache, &ent->ent, &ent->key); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1435 | list_del(&ent->lru); |
| 1436 | delta_base_cached -= ent->size; |
| 1437 | free(ent); |
| 1438 | } |
| 1439 | |
Stefan Beller | 9d98354 | 2018-04-25 11:21:06 -0700 | [diff] [blame] | 1440 | static void *cache_or_unpack_entry(struct repository *r, struct packed_git *p, |
| 1441 | off_t base_offset, unsigned long *base_size, |
| 1442 | enum object_type *type) |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1443 | { |
| 1444 | struct delta_base_cache_entry *ent; |
| 1445 | |
| 1446 | ent = get_delta_base_cache_entry(p, base_offset); |
| 1447 | if (!ent) |
Stefan Beller | 9d98354 | 2018-04-25 11:21:06 -0700 | [diff] [blame] | 1448 | return unpack_entry(r, p, base_offset, type, base_size); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1449 | |
| 1450 | if (type) |
| 1451 | *type = ent->type; |
| 1452 | if (base_size) |
| 1453 | *base_size = ent->size; |
| 1454 | return xmemdupz(ent->data, ent->size); |
| 1455 | } |
| 1456 | |
| 1457 | static inline void release_delta_base_cache(struct delta_base_cache_entry *ent) |
| 1458 | { |
| 1459 | free(ent->data); |
| 1460 | detach_delta_base_cache_entry(ent); |
| 1461 | } |
| 1462 | |
| 1463 | void clear_delta_base_cache(void) |
| 1464 | { |
| 1465 | struct list_head *lru, *tmp; |
| 1466 | list_for_each_safe(lru, tmp, &delta_base_cache_lru) { |
| 1467 | struct delta_base_cache_entry *entry = |
| 1468 | list_entry(lru, struct delta_base_cache_entry, lru); |
| 1469 | release_delta_base_cache(entry); |
| 1470 | } |
| 1471 | } |
| 1472 | |
| 1473 | static void add_delta_base_cache(struct packed_git *p, off_t base_offset, |
| 1474 | void *base, unsigned long base_size, enum object_type type) |
| 1475 | { |
Matheus Tavares | bda959c | 2020-09-28 21:01:53 -0300 | [diff] [blame] | 1476 | struct delta_base_cache_entry *ent; |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1477 | struct list_head *lru, *tmp; |
| 1478 | |
Matheus Tavares | 31877c9 | 2020-01-15 23:39:53 -0300 | [diff] [blame] | 1479 | /* |
| 1480 | * Check required to avoid redundant entries when more than one thread |
| 1481 | * is unpacking the same object, in unpack_entry() (since its phases I |
| 1482 | * and III might run concurrently across multiple threads). |
| 1483 | */ |
Matheus Tavares | bda959c | 2020-09-28 21:01:53 -0300 | [diff] [blame] | 1484 | if (in_delta_base_cache(p, base_offset)) { |
| 1485 | free(base); |
Matheus Tavares | 31877c9 | 2020-01-15 23:39:53 -0300 | [diff] [blame] | 1486 | return; |
Matheus Tavares | bda959c | 2020-09-28 21:01:53 -0300 | [diff] [blame] | 1487 | } |
Matheus Tavares | 31877c9 | 2020-01-15 23:39:53 -0300 | [diff] [blame] | 1488 | |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1489 | delta_base_cached += base_size; |
| 1490 | |
| 1491 | list_for_each_safe(lru, tmp, &delta_base_cache_lru) { |
| 1492 | struct delta_base_cache_entry *f = |
| 1493 | list_entry(lru, struct delta_base_cache_entry, lru); |
| 1494 | if (delta_base_cached <= delta_base_cache_limit) |
| 1495 | break; |
| 1496 | release_delta_base_cache(f); |
| 1497 | } |
| 1498 | |
Matheus Tavares | bda959c | 2020-09-28 21:01:53 -0300 | [diff] [blame] | 1499 | ent = xmalloc(sizeof(*ent)); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1500 | ent->key.p = p; |
| 1501 | ent->key.base_offset = base_offset; |
| 1502 | ent->type = type; |
| 1503 | ent->data = base; |
| 1504 | ent->size = base_size; |
| 1505 | list_add_tail(&ent->lru, &delta_base_cache_lru); |
| 1506 | |
| 1507 | if (!delta_base_cache.cmpfn) |
| 1508 | hashmap_init(&delta_base_cache, delta_base_cache_hash_cmp, NULL, 0); |
Eric Wong | d22245a | 2019-10-06 23:30:27 +0000 | [diff] [blame] | 1509 | hashmap_entry_init(&ent->ent, pack_entry_hash(p, base_offset)); |
Eric Wong | b94e5c1 | 2019-10-06 23:30:29 +0000 | [diff] [blame] | 1510 | hashmap_add(&delta_base_cache, &ent->ent); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1511 | } |
| 1512 | |
Stefan Beller | 9d98354 | 2018-04-25 11:21:06 -0700 | [diff] [blame] | 1513 | int packed_object_info(struct repository *r, struct packed_git *p, |
| 1514 | off_t obj_offset, struct object_info *oi) |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1515 | { |
| 1516 | struct pack_window *w_curs = NULL; |
| 1517 | unsigned long size; |
| 1518 | off_t curpos = obj_offset; |
| 1519 | enum object_type type; |
| 1520 | |
| 1521 | /* |
| 1522 | * We always get the representation type, but only convert it to |
| 1523 | * a "real" type later if the caller is interested. |
| 1524 | */ |
| 1525 | if (oi->contentp) { |
Stefan Beller | 9d98354 | 2018-04-25 11:21:06 -0700 | [diff] [blame] | 1526 | *oi->contentp = cache_or_unpack_entry(r, p, obj_offset, oi->sizep, |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1527 | &type); |
| 1528 | if (!*oi->contentp) |
| 1529 | type = OBJ_BAD; |
| 1530 | } else { |
| 1531 | type = unpack_object_header(p, &w_curs, &curpos, &size); |
| 1532 | } |
| 1533 | |
| 1534 | if (!oi->contentp && oi->sizep) { |
| 1535 | if (type == OBJ_OFS_DELTA || type == OBJ_REF_DELTA) { |
| 1536 | off_t tmp_pos = curpos; |
| 1537 | off_t base_offset = get_delta_base(p, &w_curs, &tmp_pos, |
| 1538 | type, obj_offset); |
| 1539 | if (!base_offset) { |
| 1540 | type = OBJ_BAD; |
| 1541 | goto out; |
| 1542 | } |
| 1543 | *oi->sizep = get_size_from_delta(p, &w_curs, tmp_pos); |
| 1544 | if (*oi->sizep == 0) { |
| 1545 | type = OBJ_BAD; |
| 1546 | goto out; |
| 1547 | } |
| 1548 | } else { |
| 1549 | *oi->sizep = size; |
| 1550 | } |
| 1551 | } |
| 1552 | |
| 1553 | if (oi->disk_sizep) { |
Taylor Blau | fc150ca | 2021-01-13 17:24:41 -0500 | [diff] [blame] | 1554 | uint32_t pos; |
| 1555 | if (offset_to_pack_pos(p, obj_offset, &pos) < 0) { |
| 1556 | error("could not find object at offset %"PRIuMAX" " |
| 1557 | "in pack %s", (uintmax_t)obj_offset, p->pack_name); |
| 1558 | type = OBJ_BAD; |
| 1559 | goto out; |
| 1560 | } |
| 1561 | |
| 1562 | *oi->disk_sizep = pack_pos_to_offset(p, pos + 1) - obj_offset; |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1563 | } |
| 1564 | |
Brandon Williams | 6ca32f4 | 2018-02-14 10:59:23 -0800 | [diff] [blame] | 1565 | if (oi->typep || oi->type_name) { |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1566 | enum object_type ptot; |
Stefan Beller | 9d98354 | 2018-04-25 11:21:06 -0700 | [diff] [blame] | 1567 | ptot = packed_to_object_type(r, p, obj_offset, |
Stefan Beller | 144f494 | 2018-04-25 11:21:01 -0700 | [diff] [blame] | 1568 | type, &w_curs, curpos); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1569 | if (oi->typep) |
| 1570 | *oi->typep = ptot; |
Brandon Williams | 6ca32f4 | 2018-02-14 10:59:23 -0800 | [diff] [blame] | 1571 | if (oi->type_name) { |
Brandon Williams | debca9d | 2018-02-14 10:59:24 -0800 | [diff] [blame] | 1572 | const char *tn = type_name(ptot); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1573 | if (tn) |
Brandon Williams | 6ca32f4 | 2018-02-14 10:59:23 -0800 | [diff] [blame] | 1574 | strbuf_addstr(oi->type_name, tn); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1575 | } |
| 1576 | if (ptot < 0) { |
| 1577 | type = OBJ_BAD; |
| 1578 | goto out; |
| 1579 | } |
| 1580 | } |
| 1581 | |
Jeff King | b99b6bc | 2020-02-23 23:36:56 -0500 | [diff] [blame] | 1582 | if (oi->delta_base_oid) { |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1583 | if (type == OBJ_OFS_DELTA || type == OBJ_REF_DELTA) { |
Jeff King | 6ac9760 | 2020-02-23 23:37:31 -0500 | [diff] [blame] | 1584 | if (get_delta_base_oid(p, &w_curs, curpos, |
| 1585 | oi->delta_base_oid, |
| 1586 | type, obj_offset) < 0) { |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1587 | type = OBJ_BAD; |
| 1588 | goto out; |
| 1589 | } |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1590 | } else |
Jeff King | b99b6bc | 2020-02-23 23:36:56 -0500 | [diff] [blame] | 1591 | oidclr(oi->delta_base_oid); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1592 | } |
| 1593 | |
| 1594 | oi->whence = in_delta_base_cache(p, obj_offset) ? OI_DBCACHED : |
| 1595 | OI_PACKED; |
| 1596 | |
| 1597 | out: |
| 1598 | unuse_pack(&w_curs); |
| 1599 | return type; |
| 1600 | } |
| 1601 | |
| 1602 | static void *unpack_compressed_entry(struct packed_git *p, |
| 1603 | struct pack_window **w_curs, |
| 1604 | off_t curpos, |
| 1605 | unsigned long size) |
| 1606 | { |
| 1607 | int st; |
| 1608 | git_zstream stream; |
| 1609 | unsigned char *buffer, *in; |
| 1610 | |
| 1611 | buffer = xmallocz_gently(size); |
| 1612 | if (!buffer) |
| 1613 | return NULL; |
| 1614 | memset(&stream, 0, sizeof(stream)); |
| 1615 | stream.next_out = buffer; |
| 1616 | stream.avail_out = size + 1; |
| 1617 | |
| 1618 | git_inflate_init(&stream); |
| 1619 | do { |
| 1620 | in = use_pack(p, w_curs, curpos, &stream.avail_in); |
| 1621 | stream.next_in = in; |
Matheus Tavares | 31877c9 | 2020-01-15 23:39:53 -0300 | [diff] [blame] | 1622 | /* |
| 1623 | * Note: we must ensure the window section returned by |
| 1624 | * use_pack() will be available throughout git_inflate()'s |
| 1625 | * unlocked execution. Please refer to the comment at |
| 1626 | * get_size_from_delta() to see how this is done. |
| 1627 | */ |
| 1628 | obj_read_unlock(); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1629 | st = git_inflate(&stream, Z_FINISH); |
Matheus Tavares | 31877c9 | 2020-01-15 23:39:53 -0300 | [diff] [blame] | 1630 | obj_read_lock(); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1631 | if (!stream.avail_out) |
| 1632 | break; /* the payload is larger than it should be */ |
| 1633 | curpos += stream.next_in - in; |
| 1634 | } while (st == Z_OK || st == Z_BUF_ERROR); |
| 1635 | git_inflate_end(&stream); |
| 1636 | if ((st != Z_STREAM_END) || stream.total_out != size) { |
| 1637 | free(buffer); |
| 1638 | return NULL; |
| 1639 | } |
| 1640 | |
Jeremy Linton | b611396 | 2018-06-13 09:22:07 -0500 | [diff] [blame] | 1641 | /* versions of zlib can clobber unconsumed portion of outbuf */ |
| 1642 | buffer[size] = '\0'; |
| 1643 | |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1644 | return buffer; |
| 1645 | } |
| 1646 | |
| 1647 | static void write_pack_access_log(struct packed_git *p, off_t obj_offset) |
| 1648 | { |
| 1649 | static struct trace_key pack_access = TRACE_KEY_INIT(PACK_ACCESS); |
| 1650 | trace_printf_key(&pack_access, "%s %"PRIuMAX"\n", |
| 1651 | p->pack_name, (uintmax_t)obj_offset); |
| 1652 | } |
| 1653 | |
| 1654 | int do_check_packed_object_crc; |
| 1655 | |
| 1656 | #define UNPACK_ENTRY_STACK_PREALLOC 64 |
| 1657 | struct unpack_entry_stack_ent { |
| 1658 | off_t obj_offset; |
| 1659 | off_t curpos; |
| 1660 | unsigned long size; |
| 1661 | }; |
| 1662 | |
Stefan Beller | 9d98354 | 2018-04-25 11:21:06 -0700 | [diff] [blame] | 1663 | static void *read_object(struct repository *r, |
| 1664 | const struct object_id *oid, |
| 1665 | enum object_type *type, |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1666 | unsigned long *size) |
| 1667 | { |
| 1668 | struct object_info oi = OBJECT_INFO_INIT; |
| 1669 | void *content; |
| 1670 | oi.typep = type; |
| 1671 | oi.sizep = size; |
| 1672 | oi.contentp = &content; |
| 1673 | |
Stefan Beller | 9d98354 | 2018-04-25 11:21:06 -0700 | [diff] [blame] | 1674 | if (oid_object_info_extended(r, oid, &oi, 0) < 0) |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1675 | return NULL; |
| 1676 | return content; |
| 1677 | } |
| 1678 | |
Stefan Beller | 9d98354 | 2018-04-25 11:21:06 -0700 | [diff] [blame] | 1679 | void *unpack_entry(struct repository *r, struct packed_git *p, off_t obj_offset, |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1680 | enum object_type *final_type, unsigned long *final_size) |
| 1681 | { |
| 1682 | struct pack_window *w_curs = NULL; |
| 1683 | off_t curpos = obj_offset; |
| 1684 | void *data = NULL; |
| 1685 | unsigned long size; |
| 1686 | enum object_type type; |
| 1687 | struct unpack_entry_stack_ent small_delta_stack[UNPACK_ENTRY_STACK_PREALLOC]; |
| 1688 | struct unpack_entry_stack_ent *delta_stack = small_delta_stack; |
| 1689 | int delta_stack_nr = 0, delta_stack_alloc = UNPACK_ENTRY_STACK_PREALLOC; |
| 1690 | int base_from_cache = 0; |
| 1691 | |
| 1692 | write_pack_access_log(p, obj_offset); |
| 1693 | |
| 1694 | /* PHASE 1: drill down to the innermost base object */ |
| 1695 | for (;;) { |
| 1696 | off_t base_offset; |
| 1697 | int i; |
| 1698 | struct delta_base_cache_entry *ent; |
| 1699 | |
| 1700 | ent = get_delta_base_cache_entry(p, curpos); |
| 1701 | if (ent) { |
| 1702 | type = ent->type; |
| 1703 | data = ent->data; |
| 1704 | size = ent->size; |
| 1705 | detach_delta_base_cache_entry(ent); |
| 1706 | base_from_cache = 1; |
| 1707 | break; |
| 1708 | } |
| 1709 | |
| 1710 | if (do_check_packed_object_crc && p->index_version > 1) { |
Taylor Blau | 0a7e364 | 2021-01-13 17:24:45 -0500 | [diff] [blame] | 1711 | uint32_t pack_pos, index_pos; |
| 1712 | off_t len; |
| 1713 | |
| 1714 | if (offset_to_pack_pos(p, obj_offset, &pack_pos) < 0) { |
| 1715 | error("could not find object at offset %"PRIuMAX" in pack %s", |
| 1716 | (uintmax_t)obj_offset, p->pack_name); |
| 1717 | data = NULL; |
| 1718 | goto out; |
| 1719 | } |
| 1720 | |
| 1721 | len = pack_pos_to_offset(p, pack_pos + 1) - obj_offset; |
| 1722 | index_pos = pack_pos_to_index(p, pack_pos); |
| 1723 | if (check_pack_crc(p, &w_curs, obj_offset, len, index_pos)) { |
brian m. carlson | 4310b0c | 2018-03-12 02:27:44 +0000 | [diff] [blame] | 1724 | struct object_id oid; |
Taylor Blau | 0a7e364 | 2021-01-13 17:24:45 -0500 | [diff] [blame] | 1725 | nth_packed_object_id(&oid, p, index_pos); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1726 | error("bad packed object CRC for %s", |
brian m. carlson | 4310b0c | 2018-03-12 02:27:44 +0000 | [diff] [blame] | 1727 | oid_to_hex(&oid)); |
| 1728 | mark_bad_packed_object(p, oid.hash); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1729 | data = NULL; |
| 1730 | goto out; |
| 1731 | } |
| 1732 | } |
| 1733 | |
| 1734 | type = unpack_object_header(p, &w_curs, &curpos, &size); |
| 1735 | if (type != OBJ_OFS_DELTA && type != OBJ_REF_DELTA) |
| 1736 | break; |
| 1737 | |
| 1738 | base_offset = get_delta_base(p, &w_curs, &curpos, type, obj_offset); |
| 1739 | if (!base_offset) { |
| 1740 | error("failed to validate delta base reference " |
| 1741 | "at offset %"PRIuMAX" from %s", |
| 1742 | (uintmax_t)curpos, p->pack_name); |
| 1743 | /* bail to phase 2, in hopes of recovery */ |
| 1744 | data = NULL; |
| 1745 | break; |
| 1746 | } |
| 1747 | |
| 1748 | /* push object, proceed to base */ |
| 1749 | if (delta_stack_nr >= delta_stack_alloc |
| 1750 | && delta_stack == small_delta_stack) { |
| 1751 | delta_stack_alloc = alloc_nr(delta_stack_nr); |
| 1752 | ALLOC_ARRAY(delta_stack, delta_stack_alloc); |
René Scharfe | 921d49b | 2019-06-15 20:36:35 +0200 | [diff] [blame] | 1753 | COPY_ARRAY(delta_stack, small_delta_stack, |
| 1754 | delta_stack_nr); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1755 | } else { |
| 1756 | ALLOC_GROW(delta_stack, delta_stack_nr+1, delta_stack_alloc); |
| 1757 | } |
| 1758 | i = delta_stack_nr++; |
| 1759 | delta_stack[i].obj_offset = obj_offset; |
| 1760 | delta_stack[i].curpos = curpos; |
| 1761 | delta_stack[i].size = size; |
| 1762 | |
| 1763 | curpos = obj_offset = base_offset; |
| 1764 | } |
| 1765 | |
| 1766 | /* PHASE 2: handle the base */ |
| 1767 | switch (type) { |
| 1768 | case OBJ_OFS_DELTA: |
| 1769 | case OBJ_REF_DELTA: |
| 1770 | if (data) |
Johannes Schindelin | 033abf9 | 2018-05-02 11:38:39 +0200 | [diff] [blame] | 1771 | BUG("unpack_entry: left loop at a valid delta"); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1772 | break; |
| 1773 | case OBJ_COMMIT: |
| 1774 | case OBJ_TREE: |
| 1775 | case OBJ_BLOB: |
| 1776 | case OBJ_TAG: |
| 1777 | if (!base_from_cache) |
| 1778 | data = unpack_compressed_entry(p, &w_curs, curpos, size); |
| 1779 | break; |
| 1780 | default: |
| 1781 | data = NULL; |
| 1782 | error("unknown object type %i at offset %"PRIuMAX" in %s", |
| 1783 | type, (uintmax_t)obj_offset, p->pack_name); |
| 1784 | } |
| 1785 | |
| 1786 | /* PHASE 3: apply deltas in order */ |
| 1787 | |
| 1788 | /* invariants: |
| 1789 | * 'data' holds the base data, or NULL if there was corruption |
| 1790 | */ |
| 1791 | while (delta_stack_nr) { |
| 1792 | void *delta_data; |
| 1793 | void *base = data; |
| 1794 | void *external_base = NULL; |
| 1795 | unsigned long delta_size, base_size = size; |
| 1796 | int i; |
Matheus Tavares | 74b052f | 2020-09-28 21:01:52 -0300 | [diff] [blame] | 1797 | off_t base_obj_offset = obj_offset; |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1798 | |
| 1799 | data = NULL; |
| 1800 | |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1801 | if (!base) { |
| 1802 | /* |
| 1803 | * We're probably in deep shit, but let's try to fetch |
| 1804 | * the required base anyway from another pack or loose. |
| 1805 | * This is costly but should happen only in the presence |
| 1806 | * of a corrupted pack, and is better than failing outright. |
| 1807 | */ |
Taylor Blau | 0a7e364 | 2021-01-13 17:24:45 -0500 | [diff] [blame] | 1808 | uint32_t pos; |
brian m. carlson | 4310b0c | 2018-03-12 02:27:44 +0000 | [diff] [blame] | 1809 | struct object_id base_oid; |
Taylor Blau | 0a7e364 | 2021-01-13 17:24:45 -0500 | [diff] [blame] | 1810 | if (!(offset_to_pack_pos(p, obj_offset, &pos))) { |
| 1811 | nth_packed_object_id(&base_oid, p, |
| 1812 | pack_pos_to_index(p, pos)); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1813 | error("failed to read delta base object %s" |
| 1814 | " at offset %"PRIuMAX" from %s", |
brian m. carlson | 4310b0c | 2018-03-12 02:27:44 +0000 | [diff] [blame] | 1815 | oid_to_hex(&base_oid), (uintmax_t)obj_offset, |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1816 | p->pack_name); |
brian m. carlson | 4310b0c | 2018-03-12 02:27:44 +0000 | [diff] [blame] | 1817 | mark_bad_packed_object(p, base_oid.hash); |
Stefan Beller | 9d98354 | 2018-04-25 11:21:06 -0700 | [diff] [blame] | 1818 | base = read_object(r, &base_oid, &type, &base_size); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1819 | external_base = base; |
| 1820 | } |
| 1821 | } |
| 1822 | |
| 1823 | i = --delta_stack_nr; |
| 1824 | obj_offset = delta_stack[i].obj_offset; |
| 1825 | curpos = delta_stack[i].curpos; |
| 1826 | delta_size = delta_stack[i].size; |
| 1827 | |
| 1828 | if (!base) |
| 1829 | continue; |
| 1830 | |
| 1831 | delta_data = unpack_compressed_entry(p, &w_curs, curpos, delta_size); |
| 1832 | |
| 1833 | if (!delta_data) { |
| 1834 | error("failed to unpack compressed delta " |
| 1835 | "at offset %"PRIuMAX" from %s", |
| 1836 | (uintmax_t)curpos, p->pack_name); |
| 1837 | data = NULL; |
Matheus Tavares | 74b052f | 2020-09-28 21:01:52 -0300 | [diff] [blame] | 1838 | } else { |
| 1839 | data = patch_delta(base, base_size, delta_data, |
| 1840 | delta_size, &size); |
| 1841 | |
| 1842 | /* |
| 1843 | * We could not apply the delta; warn the user, but |
| 1844 | * keep going. Our failure will be noticed either in |
| 1845 | * the next iteration of the loop, or if this is the |
| 1846 | * final delta, in the caller when we return NULL. |
| 1847 | * Those code paths will take care of making a more |
| 1848 | * explicit warning and retrying with another copy of |
| 1849 | * the object. |
| 1850 | */ |
| 1851 | if (!data) |
| 1852 | error("failed to apply delta"); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1853 | } |
| 1854 | |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1855 | /* |
Matheus Tavares | 74b052f | 2020-09-28 21:01:52 -0300 | [diff] [blame] | 1856 | * We delay adding `base` to the cache until the end of the loop |
| 1857 | * because unpack_compressed_entry() momentarily releases the |
| 1858 | * obj_read_mutex, giving another thread the chance to access |
| 1859 | * the cache. Therefore, if `base` was already there, this other |
| 1860 | * thread could free() it (e.g. to make space for another entry) |
| 1861 | * before we are done using it. |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1862 | */ |
Matheus Tavares | 74b052f | 2020-09-28 21:01:52 -0300 | [diff] [blame] | 1863 | if (!external_base) |
| 1864 | add_delta_base_cache(p, base_obj_offset, base, base_size, type); |
Jonathan Tan | f1d8130 | 2017-08-18 15:20:30 -0700 | [diff] [blame] | 1865 | |
| 1866 | free(delta_data); |
| 1867 | free(external_base); |
| 1868 | } |
| 1869 | |
| 1870 | if (final_type) |
| 1871 | *final_type = type; |
| 1872 | if (final_size) |
| 1873 | *final_size = size; |
| 1874 | |
| 1875 | out: |
| 1876 | unuse_pack(&w_curs); |
| 1877 | |
| 1878 | if (delta_stack != small_delta_stack) |
| 1879 | free(delta_stack); |
| 1880 | |
| 1881 | return data; |
| 1882 | } |
Jonathan Tan | d5a1676 | 2017-08-18 15:20:31 -0700 | [diff] [blame] | 1883 | |
Derrick Stolee | 3d475f4 | 2018-03-22 13:40:09 -0400 | [diff] [blame] | 1884 | int bsearch_pack(const struct object_id *oid, const struct packed_git *p, uint32_t *result) |
| 1885 | { |
| 1886 | const unsigned char *index_fanout = p->index_data; |
| 1887 | const unsigned char *index_lookup; |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 1888 | const unsigned int hashsz = the_hash_algo->rawsz; |
Derrick Stolee | 3d475f4 | 2018-03-22 13:40:09 -0400 | [diff] [blame] | 1889 | int index_lookup_width; |
| 1890 | |
| 1891 | if (!index_fanout) |
| 1892 | BUG("bsearch_pack called without a valid pack-index"); |
| 1893 | |
| 1894 | index_lookup = index_fanout + 4 * 256; |
| 1895 | if (p->index_version == 1) { |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 1896 | index_lookup_width = hashsz + 4; |
Derrick Stolee | 3d475f4 | 2018-03-22 13:40:09 -0400 | [diff] [blame] | 1897 | index_lookup += 4; |
| 1898 | } else { |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 1899 | index_lookup_width = hashsz; |
Derrick Stolee | 3d475f4 | 2018-03-22 13:40:09 -0400 | [diff] [blame] | 1900 | index_fanout += 8; |
| 1901 | index_lookup += 8; |
| 1902 | } |
| 1903 | |
| 1904 | return bsearch_hash(oid->hash, (const uint32_t*)index_fanout, |
| 1905 | index_lookup, index_lookup_width, result); |
| 1906 | } |
| 1907 | |
Jeff King | 2fecc48 | 2020-02-23 23:37:54 -0500 | [diff] [blame] | 1908 | int nth_packed_object_id(struct object_id *oid, |
| 1909 | struct packed_git *p, |
| 1910 | uint32_t n) |
Jonathan Tan | d5a1676 | 2017-08-18 15:20:31 -0700 | [diff] [blame] | 1911 | { |
| 1912 | const unsigned char *index = p->index_data; |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 1913 | const unsigned int hashsz = the_hash_algo->rawsz; |
Jonathan Tan | d5a1676 | 2017-08-18 15:20:31 -0700 | [diff] [blame] | 1914 | if (!index) { |
| 1915 | if (open_pack_index(p)) |
Jeff King | 2fecc48 | 2020-02-23 23:37:54 -0500 | [diff] [blame] | 1916 | return -1; |
Jonathan Tan | d5a1676 | 2017-08-18 15:20:31 -0700 | [diff] [blame] | 1917 | index = p->index_data; |
| 1918 | } |
| 1919 | if (n >= p->num_objects) |
Jeff King | 2fecc48 | 2020-02-23 23:37:54 -0500 | [diff] [blame] | 1920 | return -1; |
Jonathan Tan | d5a1676 | 2017-08-18 15:20:31 -0700 | [diff] [blame] | 1921 | index += 4 * 256; |
| 1922 | if (p->index_version == 1) { |
Jeff King | 2fecc48 | 2020-02-23 23:37:54 -0500 | [diff] [blame] | 1923 | oidread(oid, index + (hashsz + 4) * n + 4); |
Jonathan Tan | d5a1676 | 2017-08-18 15:20:31 -0700 | [diff] [blame] | 1924 | } else { |
| 1925 | index += 8; |
Jeff King | 2fecc48 | 2020-02-23 23:37:54 -0500 | [diff] [blame] | 1926 | oidread(oid, index + hashsz * n); |
Jonathan Tan | d5a1676 | 2017-08-18 15:20:31 -0700 | [diff] [blame] | 1927 | } |
Jeff King | 0763671 | 2020-02-23 23:27:36 -0500 | [diff] [blame] | 1928 | return 0; |
Jonathan Tan | d5a1676 | 2017-08-18 15:20:31 -0700 | [diff] [blame] | 1929 | } |
Jonathan Tan | 9e0f45f | 2017-08-18 15:20:32 -0700 | [diff] [blame] | 1930 | |
| 1931 | void check_pack_index_ptr(const struct packed_git *p, const void *vptr) |
| 1932 | { |
| 1933 | const unsigned char *ptr = vptr; |
| 1934 | const unsigned char *start = p->index_data; |
| 1935 | const unsigned char *end = start + p->index_size; |
| 1936 | if (ptr < start) |
| 1937 | die(_("offset before start of pack index for %s (corrupt index?)"), |
| 1938 | p->pack_name); |
| 1939 | /* No need to check for underflow; .idx files must be at least 8 bytes */ |
| 1940 | if (ptr >= end - 8) |
| 1941 | die(_("offset beyond end of pack index for %s (truncated index?)"), |
| 1942 | p->pack_name); |
| 1943 | } |
| 1944 | |
| 1945 | off_t nth_packed_object_offset(const struct packed_git *p, uint32_t n) |
| 1946 | { |
| 1947 | const unsigned char *index = p->index_data; |
brian m. carlson | 37fec86 | 2018-05-02 00:25:36 +0000 | [diff] [blame] | 1948 | const unsigned int hashsz = the_hash_algo->rawsz; |
Jonathan Tan | 9e0f45f | 2017-08-18 15:20:32 -0700 | [diff] [blame] | 1949 | index += 4 * 256; |
| 1950 | if (p->index_version == 1) { |
Jeff King | f86f769 | 2020-11-13 00:06:48 -0500 | [diff] [blame] | 1951 | return ntohl(*((uint32_t *)(index + (hashsz + 4) * (size_t)n))); |
Jonathan Tan | 9e0f45f | 2017-08-18 15:20:32 -0700 | [diff] [blame] | 1952 | } else { |
| 1953 | uint32_t off; |
Jeff King | f86f769 | 2020-11-13 00:06:48 -0500 | [diff] [blame] | 1954 | index += 8 + (size_t)p->num_objects * (hashsz + 4); |
Jonathan Tan | 9e0f45f | 2017-08-18 15:20:32 -0700 | [diff] [blame] | 1955 | off = ntohl(*((uint32_t *)(index + 4 * n))); |
| 1956 | if (!(off & 0x80000000)) |
| 1957 | return off; |
Jeff King | f86f769 | 2020-11-13 00:06:48 -0500 | [diff] [blame] | 1958 | index += (size_t)p->num_objects * 4 + (off & 0x7fffffff) * 8; |
Jonathan Tan | 9e0f45f | 2017-08-18 15:20:32 -0700 | [diff] [blame] | 1959 | check_pack_index_ptr(p, index); |
Derrick Stolee | ad622a2 | 2018-01-17 14:08:23 -0500 | [diff] [blame] | 1960 | return get_be64(index); |
Jonathan Tan | 9e0f45f | 2017-08-18 15:20:32 -0700 | [diff] [blame] | 1961 | } |
| 1962 | } |
Jonathan Tan | a255195 | 2017-08-18 15:20:33 -0700 | [diff] [blame] | 1963 | |
| 1964 | off_t find_pack_entry_one(const unsigned char *sha1, |
| 1965 | struct packed_git *p) |
| 1966 | { |
Jonathan Tan | a255195 | 2017-08-18 15:20:33 -0700 | [diff] [blame] | 1967 | const unsigned char *index = p->index_data; |
Derrick Stolee | 3d475f4 | 2018-03-22 13:40:09 -0400 | [diff] [blame] | 1968 | struct object_id oid; |
Jonathan Tan | b4e00f7 | 2018-02-13 10:39:39 -0800 | [diff] [blame] | 1969 | uint32_t result; |
Jonathan Tan | a255195 | 2017-08-18 15:20:33 -0700 | [diff] [blame] | 1970 | |
| 1971 | if (!index) { |
| 1972 | if (open_pack_index(p)) |
| 1973 | return 0; |
Jonathan Tan | a255195 | 2017-08-18 15:20:33 -0700 | [diff] [blame] | 1974 | } |
| 1975 | |
Derrick Stolee | 3d475f4 | 2018-03-22 13:40:09 -0400 | [diff] [blame] | 1976 | hashcpy(oid.hash, sha1); |
| 1977 | if (bsearch_pack(&oid, p, &result)) |
Jonathan Tan | b4e00f7 | 2018-02-13 10:39:39 -0800 | [diff] [blame] | 1978 | return nth_packed_object_offset(p, result); |
Jonathan Tan | a255195 | 2017-08-18 15:20:33 -0700 | [diff] [blame] | 1979 | return 0; |
| 1980 | } |
| 1981 | |
| 1982 | int is_pack_valid(struct packed_git *p) |
| 1983 | { |
| 1984 | /* An already open pack is known to be valid. */ |
| 1985 | if (p->pack_fd != -1) |
| 1986 | return 1; |
| 1987 | |
| 1988 | /* If the pack has one window completely covering the |
| 1989 | * file size, the pack is known to be valid even if |
| 1990 | * the descriptor is not currently open. |
| 1991 | */ |
| 1992 | if (p->windows) { |
| 1993 | struct pack_window *w = p->windows; |
| 1994 | |
| 1995 | if (!w->offset && w->len == p->pack_size) |
| 1996 | return 1; |
| 1997 | } |
| 1998 | |
| 1999 | /* Force the pack to open to prove its valid. */ |
| 2000 | return !open_packed_git(p); |
| 2001 | } |
Jonathan Tan | d6fe003 | 2017-08-18 15:20:34 -0700 | [diff] [blame] | 2002 | |
| 2003 | struct packed_git *find_sha1_pack(const unsigned char *sha1, |
| 2004 | struct packed_git *packs) |
| 2005 | { |
| 2006 | struct packed_git *p; |
| 2007 | |
| 2008 | for (p = packs; p; p = p->next) { |
| 2009 | if (find_pack_entry_one(sha1, p)) |
| 2010 | return p; |
| 2011 | } |
| 2012 | return NULL; |
| 2013 | |
| 2014 | } |
Jonathan Tan | 1a1e5d4 | 2017-08-18 15:20:35 -0700 | [diff] [blame] | 2015 | |
brian m. carlson | 544443c | 2018-05-02 00:25:35 +0000 | [diff] [blame] | 2016 | static int fill_pack_entry(const struct object_id *oid, |
Jonathan Tan | 1a1e5d4 | 2017-08-18 15:20:35 -0700 | [diff] [blame] | 2017 | struct pack_entry *e, |
| 2018 | struct packed_git *p) |
| 2019 | { |
| 2020 | off_t offset; |
| 2021 | |
| 2022 | if (p->num_bad_objects) { |
| 2023 | unsigned i; |
| 2024 | for (i = 0; i < p->num_bad_objects; i++) |
Jeff King | e3ff068 | 2018-08-28 17:22:44 -0400 | [diff] [blame] | 2025 | if (hasheq(oid->hash, |
| 2026 | p->bad_object_sha1 + the_hash_algo->rawsz * i)) |
Jonathan Tan | 1a1e5d4 | 2017-08-18 15:20:35 -0700 | [diff] [blame] | 2027 | return 0; |
| 2028 | } |
| 2029 | |
brian m. carlson | 544443c | 2018-05-02 00:25:35 +0000 | [diff] [blame] | 2030 | offset = find_pack_entry_one(oid->hash, p); |
Jonathan Tan | 1a1e5d4 | 2017-08-18 15:20:35 -0700 | [diff] [blame] | 2031 | if (!offset) |
| 2032 | return 0; |
| 2033 | |
| 2034 | /* |
| 2035 | * We are about to tell the caller where they can locate the |
| 2036 | * requested object. We better make sure the packfile is |
| 2037 | * still here and can be accessed before supplying that |
| 2038 | * answer, as it may have been deleted since the index was |
| 2039 | * loaded! |
| 2040 | */ |
| 2041 | if (!is_pack_valid(p)) |
| 2042 | return 0; |
| 2043 | e->offset = offset; |
| 2044 | e->p = p; |
Jonathan Tan | 1a1e5d4 | 2017-08-18 15:20:35 -0700 | [diff] [blame] | 2045 | return 1; |
| 2046 | } |
| 2047 | |
brian m. carlson | 544443c | 2018-05-02 00:25:35 +0000 | [diff] [blame] | 2048 | int find_pack_entry(struct repository *r, const struct object_id *oid, struct pack_entry *e) |
Jonathan Tan | 1a1e5d4 | 2017-08-18 15:20:35 -0700 | [diff] [blame] | 2049 | { |
Olga Telezhnaya | 8865859 | 2017-09-30 17:51:01 +0000 | [diff] [blame] | 2050 | struct list_head *pos; |
Derrick Stolee | 3715a63 | 2018-07-12 15:39:34 -0400 | [diff] [blame] | 2051 | struct multi_pack_index *m; |
Jonathan Tan | 1a1e5d4 | 2017-08-18 15:20:35 -0700 | [diff] [blame] | 2052 | |
Stefan Beller | 0a0dd63 | 2018-03-23 18:45:26 +0100 | [diff] [blame] | 2053 | prepare_packed_git(r); |
Derrick Stolee | 3715a63 | 2018-07-12 15:39:34 -0400 | [diff] [blame] | 2054 | if (!r->objects->packed_git && !r->objects->multi_pack_index) |
Jonathan Tan | 1a1e5d4 | 2017-08-18 15:20:35 -0700 | [diff] [blame] | 2055 | return 0; |
| 2056 | |
Derrick Stolee | 3715a63 | 2018-07-12 15:39:34 -0400 | [diff] [blame] | 2057 | for (m = r->objects->multi_pack_index; m; m = m->next) { |
Derrick Stolee | 64404a2 | 2019-04-29 09:18:55 -0700 | [diff] [blame] | 2058 | if (fill_midx_entry(r, oid, e, m)) |
Derrick Stolee | 3715a63 | 2018-07-12 15:39:34 -0400 | [diff] [blame] | 2059 | return 1; |
| 2060 | } |
| 2061 | |
Stefan Beller | 0a0dd63 | 2018-03-23 18:45:26 +0100 | [diff] [blame] | 2062 | list_for_each(pos, &r->objects->packed_git_mru) { |
Gargi Sharma | ec2dd32 | 2018-01-23 18:46:51 -0500 | [diff] [blame] | 2063 | struct packed_git *p = list_entry(pos, struct packed_git, mru); |
Derrick Stolee | af96fe3 | 2019-04-29 09:18:56 -0700 | [diff] [blame] | 2064 | if (!p->multi_pack_index && fill_pack_entry(oid, e, p)) { |
Stefan Beller | 0a0dd63 | 2018-03-23 18:45:26 +0100 | [diff] [blame] | 2065 | list_move(&p->mru, &r->objects->packed_git_mru); |
Jonathan Tan | 1a1e5d4 | 2017-08-18 15:20:35 -0700 | [diff] [blame] | 2066 | return 1; |
| 2067 | } |
| 2068 | } |
| 2069 | return 0; |
| 2070 | } |
Jonathan Tan | 150e300 | 2017-08-18 15:20:36 -0700 | [diff] [blame] | 2071 | |
Jeff King | 20b031f | 2021-02-22 21:25:23 -0500 | [diff] [blame] | 2072 | static void maybe_invalidate_kept_pack_cache(struct repository *r, |
| 2073 | unsigned flags) |
Taylor Blau | f62312e | 2021-02-22 21:25:03 -0500 | [diff] [blame] | 2074 | { |
Jeff King | 20b031f | 2021-02-22 21:25:23 -0500 | [diff] [blame] | 2075 | if (!r->objects->kept_pack_cache.packs) |
| 2076 | return; |
| 2077 | if (r->objects->kept_pack_cache.flags == flags) |
| 2078 | return; |
| 2079 | FREE_AND_NULL(r->objects->kept_pack_cache.packs); |
| 2080 | r->objects->kept_pack_cache.flags = 0; |
| 2081 | } |
| 2082 | |
| 2083 | static struct packed_git **kept_pack_cache(struct repository *r, unsigned flags) |
| 2084 | { |
| 2085 | maybe_invalidate_kept_pack_cache(r, flags); |
| 2086 | |
| 2087 | if (!r->objects->kept_pack_cache.packs) { |
| 2088 | struct packed_git **packs = NULL; |
| 2089 | size_t nr = 0, alloc = 0; |
| 2090 | struct packed_git *p; |
| 2091 | |
| 2092 | /* |
| 2093 | * We want "all" packs here, because we need to cover ones that |
| 2094 | * are used by a midx, as well. We need to look in every one of |
| 2095 | * them (instead of the midx itself) to cover duplicates. It's |
| 2096 | * possible that an object is found in two packs that the midx |
| 2097 | * covers, one kept and one not kept, but the midx returns only |
| 2098 | * the non-kept version. |
| 2099 | */ |
| 2100 | for (p = get_all_packs(r); p; p = p->next) { |
| 2101 | if ((p->pack_keep && (flags & ON_DISK_KEEP_PACKS)) || |
| 2102 | (p->pack_keep_in_core && (flags & IN_CORE_KEEP_PACKS))) { |
| 2103 | ALLOC_GROW(packs, nr + 1, alloc); |
| 2104 | packs[nr++] = p; |
| 2105 | } |
| 2106 | } |
| 2107 | ALLOC_GROW(packs, nr + 1, alloc); |
| 2108 | packs[nr] = NULL; |
| 2109 | |
| 2110 | r->objects->kept_pack_cache.packs = packs; |
| 2111 | r->objects->kept_pack_cache.flags = flags; |
| 2112 | } |
| 2113 | |
| 2114 | return r->objects->kept_pack_cache.packs; |
Taylor Blau | f62312e | 2021-02-22 21:25:03 -0500 | [diff] [blame] | 2115 | } |
| 2116 | |
| 2117 | int find_kept_pack_entry(struct repository *r, |
| 2118 | const struct object_id *oid, |
| 2119 | unsigned flags, |
| 2120 | struct pack_entry *e) |
| 2121 | { |
Jeff King | 20b031f | 2021-02-22 21:25:23 -0500 | [diff] [blame] | 2122 | struct packed_git **cache; |
| 2123 | |
| 2124 | for (cache = kept_pack_cache(r, flags); *cache; cache++) { |
| 2125 | struct packed_git *p = *cache; |
| 2126 | if (fill_pack_entry(oid, e, p)) |
| 2127 | return 1; |
| 2128 | } |
| 2129 | |
| 2130 | return 0; |
Taylor Blau | f62312e | 2021-02-22 21:25:03 -0500 | [diff] [blame] | 2131 | } |
| 2132 | |
brian m. carlson | 14c3c80 | 2018-05-02 00:25:33 +0000 | [diff] [blame] | 2133 | int has_object_pack(const struct object_id *oid) |
Jonathan Tan | 150e300 | 2017-08-18 15:20:36 -0700 | [diff] [blame] | 2134 | { |
| 2135 | struct pack_entry e; |
brian m. carlson | 544443c | 2018-05-02 00:25:35 +0000 | [diff] [blame] | 2136 | return find_pack_entry(the_repository, oid, &e); |
Jonathan Tan | 150e300 | 2017-08-18 15:20:36 -0700 | [diff] [blame] | 2137 | } |
Jonathan Tan | f9a8672 | 2017-08-18 15:20:37 -0700 | [diff] [blame] | 2138 | |
Taylor Blau | f62312e | 2021-02-22 21:25:03 -0500 | [diff] [blame] | 2139 | int has_object_kept_pack(const struct object_id *oid, unsigned flags) |
| 2140 | { |
| 2141 | struct pack_entry e; |
| 2142 | return find_kept_pack_entry(the_repository, oid, flags, &e); |
| 2143 | } |
| 2144 | |
Jonathan Tan | f9a8672 | 2017-08-18 15:20:37 -0700 | [diff] [blame] | 2145 | int has_pack_index(const unsigned char *sha1) |
| 2146 | { |
| 2147 | struct stat st; |
| 2148 | if (stat(sha1_pack_index_name(sha1), &st)) |
| 2149 | return 0; |
| 2150 | return 1; |
| 2151 | } |
Jonathan Tan | 7709f46 | 2017-08-18 15:20:38 -0700 | [diff] [blame] | 2152 | |
Jeff King | 736eb88 | 2018-08-10 19:15:49 -0400 | [diff] [blame] | 2153 | int for_each_object_in_pack(struct packed_git *p, |
| 2154 | each_packed_object_fn cb, void *data, |
| 2155 | enum for_each_object_flags flags) |
Jonathan Tan | 7709f46 | 2017-08-18 15:20:38 -0700 | [diff] [blame] | 2156 | { |
| 2157 | uint32_t i; |
| 2158 | int r = 0; |
| 2159 | |
Jeff King | 4828ce9 | 2019-04-05 14:04:24 -0400 | [diff] [blame] | 2160 | if (flags & FOR_EACH_OBJECT_PACK_ORDER) { |
| 2161 | if (load_pack_revindex(p)) |
| 2162 | return -1; |
| 2163 | } |
Jeff King | 736eb88 | 2018-08-10 19:15:49 -0400 | [diff] [blame] | 2164 | |
Jonathan Tan | 7709f46 | 2017-08-18 15:20:38 -0700 | [diff] [blame] | 2165 | for (i = 0; i < p->num_objects; i++) { |
Jeff King | 779412b | 2021-01-14 15:11:10 -0500 | [diff] [blame] | 2166 | uint32_t index_pos; |
Jonathan Tan | 7709f46 | 2017-08-18 15:20:38 -0700 | [diff] [blame] | 2167 | struct object_id oid; |
| 2168 | |
Jeff King | 779412b | 2021-01-14 15:11:10 -0500 | [diff] [blame] | 2169 | /* |
| 2170 | * We are iterating "i" from 0 up to num_objects, but its |
| 2171 | * meaning may be different, depending on the requested output |
| 2172 | * order: |
| 2173 | * |
| 2174 | * - in object-name order, it is the same as the index order |
| 2175 | * used by nth_packed_object_id(), so we can pass it |
| 2176 | * directly |
| 2177 | * |
| 2178 | * - in pack-order, it is pack position, which we must |
| 2179 | * convert to an index position in order to get the oid. |
| 2180 | */ |
Jeff King | 736eb88 | 2018-08-10 19:15:49 -0400 | [diff] [blame] | 2181 | if (flags & FOR_EACH_OBJECT_PACK_ORDER) |
Jeff King | 779412b | 2021-01-14 15:11:10 -0500 | [diff] [blame] | 2182 | index_pos = pack_pos_to_index(p, i); |
Jeff King | 736eb88 | 2018-08-10 19:15:49 -0400 | [diff] [blame] | 2183 | else |
Jeff King | 779412b | 2021-01-14 15:11:10 -0500 | [diff] [blame] | 2184 | index_pos = i; |
Jonathan Tan | 7709f46 | 2017-08-18 15:20:38 -0700 | [diff] [blame] | 2185 | |
Jeff King | 779412b | 2021-01-14 15:11:10 -0500 | [diff] [blame] | 2186 | if (nth_packed_object_id(&oid, p, index_pos) < 0) |
Jeff King | 736eb88 | 2018-08-10 19:15:49 -0400 | [diff] [blame] | 2187 | return error("unable to get sha1 of object %u in %s", |
Jeff King | 779412b | 2021-01-14 15:11:10 -0500 | [diff] [blame] | 2188 | index_pos, p->pack_name); |
Jeff King | 736eb88 | 2018-08-10 19:15:49 -0400 | [diff] [blame] | 2189 | |
Jeff King | 779412b | 2021-01-14 15:11:10 -0500 | [diff] [blame] | 2190 | r = cb(&oid, p, index_pos, data); |
Jonathan Tan | 7709f46 | 2017-08-18 15:20:38 -0700 | [diff] [blame] | 2191 | if (r) |
| 2192 | break; |
| 2193 | } |
| 2194 | return r; |
| 2195 | } |
| 2196 | |
Jeff King | a7ff6f5 | 2018-08-10 19:09:44 -0400 | [diff] [blame] | 2197 | int for_each_packed_object(each_packed_object_fn cb, void *data, |
| 2198 | enum for_each_object_flags flags) |
Jonathan Tan | 7709f46 | 2017-08-18 15:20:38 -0700 | [diff] [blame] | 2199 | { |
| 2200 | struct packed_git *p; |
| 2201 | int r = 0; |
| 2202 | int pack_errors = 0; |
| 2203 | |
Stefan Beller | 6fdb4e9 | 2018-03-23 18:45:20 +0100 | [diff] [blame] | 2204 | prepare_packed_git(the_repository); |
Derrick Stolee | 454ea2e | 2018-08-20 16:52:04 +0000 | [diff] [blame] | 2205 | for (p = get_all_packs(the_repository); p; p = p->next) { |
Jonathan Tan | 7709f46 | 2017-08-18 15:20:38 -0700 | [diff] [blame] | 2206 | if ((flags & FOR_EACH_OBJECT_LOCAL_ONLY) && !p->pack_local) |
| 2207 | continue; |
Jonathan Tan | 498f1f6 | 2017-12-05 16:58:44 +0000 | [diff] [blame] | 2208 | if ((flags & FOR_EACH_OBJECT_PROMISOR_ONLY) && |
| 2209 | !p->pack_promisor) |
| 2210 | continue; |
Jonathan Tan | 7709f46 | 2017-08-18 15:20:38 -0700 | [diff] [blame] | 2211 | if (open_pack_index(p)) { |
| 2212 | pack_errors = 1; |
| 2213 | continue; |
| 2214 | } |
Jeff King | 736eb88 | 2018-08-10 19:15:49 -0400 | [diff] [blame] | 2215 | r = for_each_object_in_pack(p, cb, data, flags); |
Jonathan Tan | 7709f46 | 2017-08-18 15:20:38 -0700 | [diff] [blame] | 2216 | if (r) |
| 2217 | break; |
| 2218 | } |
| 2219 | return r ? r : pack_errors; |
| 2220 | } |
Jonathan Tan | 498f1f6 | 2017-12-05 16:58:44 +0000 | [diff] [blame] | 2221 | |
| 2222 | static int add_promisor_object(const struct object_id *oid, |
| 2223 | struct packed_git *pack, |
| 2224 | uint32_t pos, |
| 2225 | void *set_) |
| 2226 | { |
| 2227 | struct oidset *set = set_; |
Stefan Beller | 109cd76 | 2018-06-28 18:21:51 -0700 | [diff] [blame] | 2228 | struct object *obj = parse_object(the_repository, oid); |
Jonathan Tan | 498f1f6 | 2017-12-05 16:58:44 +0000 | [diff] [blame] | 2229 | if (!obj) |
| 2230 | return 1; |
| 2231 | |
| 2232 | oidset_insert(set, oid); |
| 2233 | |
| 2234 | /* |
| 2235 | * If this is a tree, commit, or tag, the objects it refers |
Stefan Beller | a80d72d | 2018-03-23 18:20:59 +0100 | [diff] [blame] | 2236 | * to are also promisor objects. (Blobs refer to no objects->) |
Jonathan Tan | 498f1f6 | 2017-12-05 16:58:44 +0000 | [diff] [blame] | 2237 | */ |
| 2238 | if (obj->type == OBJ_TREE) { |
| 2239 | struct tree *tree = (struct tree *)obj; |
| 2240 | struct tree_desc desc; |
| 2241 | struct name_entry entry; |
| 2242 | if (init_tree_desc_gently(&desc, tree->buffer, tree->size)) |
| 2243 | /* |
| 2244 | * Error messages are given when packs are |
| 2245 | * verified, so do not print any here. |
| 2246 | */ |
| 2247 | return 0; |
| 2248 | while (tree_entry_gently(&desc, &entry)) |
brian m. carlson | ea82b2a | 2019-01-15 00:39:44 +0000 | [diff] [blame] | 2249 | oidset_insert(set, &entry.oid); |
Jeff King | fcc07e9 | 2021-04-13 03:15:54 -0400 | [diff] [blame] | 2250 | free_tree_buffer(tree); |
Jonathan Tan | 498f1f6 | 2017-12-05 16:58:44 +0000 | [diff] [blame] | 2251 | } else if (obj->type == OBJ_COMMIT) { |
| 2252 | struct commit *commit = (struct commit *) obj; |
| 2253 | struct commit_list *parents = commit->parents; |
| 2254 | |
Derrick Stolee | 2e27bd7 | 2018-04-06 19:09:38 +0000 | [diff] [blame] | 2255 | oidset_insert(set, get_commit_tree_oid(commit)); |
Jonathan Tan | 498f1f6 | 2017-12-05 16:58:44 +0000 | [diff] [blame] | 2256 | for (; parents; parents = parents->next) |
| 2257 | oidset_insert(set, &parents->item->object.oid); |
| 2258 | } else if (obj->type == OBJ_TAG) { |
| 2259 | struct tag *tag = (struct tag *) obj; |
René Scharfe | c77722b | 2019-09-05 21:59:42 +0200 | [diff] [blame] | 2260 | oidset_insert(set, get_tagged_oid(tag)); |
Jonathan Tan | 498f1f6 | 2017-12-05 16:58:44 +0000 | [diff] [blame] | 2261 | } |
| 2262 | return 0; |
| 2263 | } |
| 2264 | |
| 2265 | int is_promisor_object(const struct object_id *oid) |
| 2266 | { |
| 2267 | static struct oidset promisor_objects; |
| 2268 | static int promisor_objects_prepared; |
| 2269 | |
| 2270 | if (!promisor_objects_prepared) { |
Christian Couder | b14ed5a | 2019-06-25 15:40:31 +0200 | [diff] [blame] | 2271 | if (has_promisor_remote()) { |
Jonathan Tan | 498f1f6 | 2017-12-05 16:58:44 +0000 | [diff] [blame] | 2272 | for_each_packed_object(add_promisor_object, |
| 2273 | &promisor_objects, |
| 2274 | FOR_EACH_OBJECT_PROMISOR_ONLY); |
| 2275 | } |
| 2276 | promisor_objects_prepared = 1; |
| 2277 | } |
| 2278 | return oidset_contains(&promisor_objects, oid); |
| 2279 | } |