]>
Commit | Line | Data |
---|---|---|
a3407730 | 1 | #include "cache.h" |
c4d25228 | 2 | #include "config.h" |
fc59e748 | 3 | #include "csum-file.h" |
396f2570 | 4 | #include "dir.h" |
fc59e748 | 5 | #include "lockfile.h" |
396f2570 | 6 | #include "packfile.h" |
4d80560c | 7 | #include "object-store.h" |
3715a633 | 8 | #include "sha1-lookup.h" |
a3407730 | 9 | #include "midx.h" |
144d7033 | 10 | #include "progress.h" |
d829223a | 11 | #include "trace2.h" |
ce1e4a10 | 12 | #include "run-command.h" |
a3407730 | 13 | |
fc59e748 DS |
14 | #define MIDX_SIGNATURE 0x4d494458 /* "MIDX" */ |
15 | #define MIDX_VERSION 1 | |
4d80560c DS |
16 | #define MIDX_BYTE_FILE_VERSION 4 |
17 | #define MIDX_BYTE_HASH_VERSION 5 | |
18 | #define MIDX_BYTE_NUM_CHUNKS 6 | |
19 | #define MIDX_BYTE_NUM_PACKS 8 | |
fc59e748 | 20 | #define MIDX_HEADER_SIZE 12 |
aaa95dfa | 21 | #define MIDX_MIN_SIZE (MIDX_HEADER_SIZE + the_hash_algo->rawsz) |
fc59e748 | 22 | |
662148c4 | 23 | #define MIDX_MAX_CHUNKS 5 |
32f3c541 DS |
24 | #define MIDX_CHUNK_ALIGNMENT 4 |
25 | #define MIDX_CHUNKID_PACKNAMES 0x504e414d /* "PNAM" */ | |
d7cacf29 | 26 | #define MIDX_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */ |
0d5b3a5e | 27 | #define MIDX_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */ |
662148c4 DS |
28 | #define MIDX_CHUNKID_OBJECTOFFSETS 0x4f4f4646 /* "OOFF" */ |
29 | #define MIDX_CHUNKID_LARGEOFFSETS 0x4c4f4646 /* "LOFF" */ | |
32f3c541 | 30 | #define MIDX_CHUNKLOOKUP_WIDTH (sizeof(uint32_t) + sizeof(uint64_t)) |
d7cacf29 | 31 | #define MIDX_CHUNK_FANOUT_SIZE (sizeof(uint32_t) * 256) |
662148c4 DS |
32 | #define MIDX_CHUNK_OFFSET_WIDTH (2 * sizeof(uint32_t)) |
33 | #define MIDX_CHUNK_LARGE_OFFSET_WIDTH (sizeof(uint64_t)) | |
34 | #define MIDX_LARGE_OFFSET_NEEDED 0x80000000 | |
32f3c541 | 35 | |
19575c7c DS |
36 | #define PACK_EXPIRED UINT_MAX |
37 | ||
d9607542 DS |
38 | static uint8_t oid_version(void) |
39 | { | |
40 | switch (hash_algo_by_ptr(the_hash_algo)) { | |
41 | case GIT_HASH_SHA1: | |
42 | return 1; | |
43 | case GIT_HASH_SHA256: | |
44 | return 2; | |
45 | default: | |
46 | die(_("invalid hash version")); | |
47 | } | |
48 | } | |
49 | ||
fc59e748 DS |
50 | static char *get_midx_filename(const char *object_dir) |
51 | { | |
52 | return xstrfmt("%s/pack/multi-pack-index", object_dir); | |
53 | } | |
54 | ||
2cf489a3 | 55 | struct multi_pack_index *load_multi_pack_index(const char *object_dir, int local) |
4d80560c DS |
56 | { |
57 | struct multi_pack_index *m = NULL; | |
58 | int fd; | |
59 | struct stat st; | |
60 | size_t midx_size; | |
61 | void *midx_map = NULL; | |
62 | uint32_t hash_version; | |
63 | char *midx_name = get_midx_filename(object_dir); | |
32f3c541 | 64 | uint32_t i; |
3227565c | 65 | const char *cur_pack_name; |
4d80560c DS |
66 | |
67 | fd = git_open(midx_name); | |
68 | ||
69 | if (fd < 0) | |
70 | goto cleanup_fail; | |
71 | if (fstat(fd, &st)) { | |
72 | error_errno(_("failed to read %s"), midx_name); | |
73 | goto cleanup_fail; | |
74 | } | |
75 | ||
76 | midx_size = xsize_t(st.st_size); | |
77 | ||
78 | if (midx_size < MIDX_MIN_SIZE) { | |
79 | error(_("multi-pack-index file %s is too small"), midx_name); | |
80 | goto cleanup_fail; | |
81 | } | |
82 | ||
83 | FREE_AND_NULL(midx_name); | |
84 | ||
85 | midx_map = xmmap(NULL, midx_size, PROT_READ, MAP_PRIVATE, fd, 0); | |
6c7ff7cf | 86 | close(fd); |
4d80560c | 87 | |
577314ca | 88 | FLEX_ALLOC_STR(m, object_dir, object_dir); |
4d80560c DS |
89 | m->data = midx_map; |
90 | m->data_len = midx_size; | |
2cf489a3 | 91 | m->local = local; |
4d80560c DS |
92 | |
93 | m->signature = get_be32(m->data); | |
53ad0407 DS |
94 | if (m->signature != MIDX_SIGNATURE) |
95 | die(_("multi-pack-index signature 0x%08x does not match signature 0x%08x"), | |
4d80560c | 96 | m->signature, MIDX_SIGNATURE); |
4d80560c DS |
97 | |
98 | m->version = m->data[MIDX_BYTE_FILE_VERSION]; | |
53ad0407 DS |
99 | if (m->version != MIDX_VERSION) |
100 | die(_("multi-pack-index version %d not recognized"), | |
4d80560c | 101 | m->version); |
4d80560c DS |
102 | |
103 | hash_version = m->data[MIDX_BYTE_HASH_VERSION]; | |
d9607542 DS |
104 | if (hash_version != oid_version()) { |
105 | error(_("multi-pack-index hash version %u does not match version %u"), | |
106 | hash_version, oid_version()); | |
107 | goto cleanup_fail; | |
108 | } | |
aaa95dfa | 109 | m->hash_len = the_hash_algo->rawsz; |
4d80560c DS |
110 | |
111 | m->num_chunks = m->data[MIDX_BYTE_NUM_CHUNKS]; | |
112 | ||
113 | m->num_packs = get_be32(m->data + MIDX_BYTE_NUM_PACKS); | |
114 | ||
32f3c541 DS |
115 | for (i = 0; i < m->num_chunks; i++) { |
116 | uint32_t chunk_id = get_be32(m->data + MIDX_HEADER_SIZE + | |
117 | MIDX_CHUNKLOOKUP_WIDTH * i); | |
118 | uint64_t chunk_offset = get_be64(m->data + MIDX_HEADER_SIZE + 4 + | |
119 | MIDX_CHUNKLOOKUP_WIDTH * i); | |
120 | ||
d3f8e211 DS |
121 | if (chunk_offset >= m->data_len) |
122 | die(_("invalid chunk offset (too large)")); | |
123 | ||
32f3c541 DS |
124 | switch (chunk_id) { |
125 | case MIDX_CHUNKID_PACKNAMES: | |
126 | m->chunk_pack_names = m->data + chunk_offset; | |
127 | break; | |
128 | ||
d7cacf29 DS |
129 | case MIDX_CHUNKID_OIDFANOUT: |
130 | m->chunk_oid_fanout = (uint32_t *)(m->data + chunk_offset); | |
131 | break; | |
132 | ||
0d5b3a5e DS |
133 | case MIDX_CHUNKID_OIDLOOKUP: |
134 | m->chunk_oid_lookup = m->data + chunk_offset; | |
135 | break; | |
136 | ||
662148c4 DS |
137 | case MIDX_CHUNKID_OBJECTOFFSETS: |
138 | m->chunk_object_offsets = m->data + chunk_offset; | |
139 | break; | |
140 | ||
141 | case MIDX_CHUNKID_LARGEOFFSETS: | |
142 | m->chunk_large_offsets = m->data + chunk_offset; | |
143 | break; | |
144 | ||
32f3c541 DS |
145 | case 0: |
146 | die(_("terminating multi-pack-index chunk id appears earlier than expected")); | |
147 | break; | |
148 | ||
149 | default: | |
150 | /* | |
151 | * Do nothing on unrecognized chunks, allowing future | |
152 | * extensions to add optional chunks. | |
153 | */ | |
154 | break; | |
155 | } | |
156 | } | |
157 | ||
158 | if (!m->chunk_pack_names) | |
159 | die(_("multi-pack-index missing required pack-name chunk")); | |
d7cacf29 DS |
160 | if (!m->chunk_oid_fanout) |
161 | die(_("multi-pack-index missing required OID fanout chunk")); | |
0d5b3a5e DS |
162 | if (!m->chunk_oid_lookup) |
163 | die(_("multi-pack-index missing required OID lookup chunk")); | |
662148c4 DS |
164 | if (!m->chunk_object_offsets) |
165 | die(_("multi-pack-index missing required object offsets chunk")); | |
32f3c541 | 166 | |
d7cacf29 DS |
167 | m->num_objects = ntohl(m->chunk_oid_fanout[255]); |
168 | ||
3227565c | 169 | m->pack_names = xcalloc(m->num_packs, sizeof(*m->pack_names)); |
3715a633 | 170 | m->packs = xcalloc(m->num_packs, sizeof(*m->packs)); |
3227565c DS |
171 | |
172 | cur_pack_name = (const char *)m->chunk_pack_names; | |
173 | for (i = 0; i < m->num_packs; i++) { | |
174 | m->pack_names[i] = cur_pack_name; | |
175 | ||
176 | cur_pack_name += strlen(cur_pack_name) + 1; | |
177 | ||
8e72a3c3 DS |
178 | if (i && strcmp(m->pack_names[i], m->pack_names[i - 1]) <= 0) |
179 | die(_("multi-pack-index pack names out of order: '%s' before '%s'"), | |
3227565c DS |
180 | m->pack_names[i - 1], |
181 | m->pack_names[i]); | |
3227565c DS |
182 | } |
183 | ||
d829223a JH |
184 | trace2_data_intmax("midx", the_repository, "load/num_packs", m->num_packs); |
185 | trace2_data_intmax("midx", the_repository, "load/num_objects", m->num_objects); | |
186 | ||
4d80560c DS |
187 | return m; |
188 | ||
189 | cleanup_fail: | |
190 | free(m); | |
191 | free(midx_name); | |
192 | if (midx_map) | |
193 | munmap(midx_map, midx_size); | |
194 | if (0 <= fd) | |
195 | close(fd); | |
196 | return NULL; | |
197 | } | |
198 | ||
1dcd9f20 | 199 | void close_midx(struct multi_pack_index *m) |
a40498a1 DS |
200 | { |
201 | uint32_t i; | |
1dcd9f20 DS |
202 | |
203 | if (!m) | |
204 | return; | |
205 | ||
a40498a1 | 206 | munmap((unsigned char *)m->data, m->data_len); |
a40498a1 DS |
207 | |
208 | for (i = 0; i < m->num_packs; i++) { | |
af96fe33 DS |
209 | if (m->packs[i]) |
210 | m->packs[i]->multi_pack_index = 0; | |
a40498a1 DS |
211 | } |
212 | FREE_AND_NULL(m->packs); | |
213 | FREE_AND_NULL(m->pack_names); | |
214 | } | |
215 | ||
64404a24 | 216 | int prepare_midx_pack(struct repository *r, struct multi_pack_index *m, uint32_t pack_int_id) |
3715a633 DS |
217 | { |
218 | struct strbuf pack_name = STRBUF_INIT; | |
af96fe33 | 219 | struct packed_git *p; |
3715a633 DS |
220 | |
221 | if (pack_int_id >= m->num_packs) | |
d355e46a | 222 | die(_("bad pack-int-id: %u (%u total packs)"), |
cc6af73c | 223 | pack_int_id, m->num_packs); |
3715a633 DS |
224 | |
225 | if (m->packs[pack_int_id]) | |
226 | return 0; | |
227 | ||
228 | strbuf_addf(&pack_name, "%s/pack/%s", m->object_dir, | |
229 | m->pack_names[pack_int_id]); | |
230 | ||
af96fe33 | 231 | p = add_packed_git(pack_name.buf, pack_name.len, m->local); |
3715a633 | 232 | strbuf_release(&pack_name); |
af96fe33 DS |
233 | |
234 | if (!p) | |
235 | return 1; | |
236 | ||
237 | p->multi_pack_index = 1; | |
238 | m->packs[pack_int_id] = p; | |
239 | install_packed_git(r, p); | |
240 | list_add_tail(&p->mru, &r->objects->packed_git_mru); | |
241 | ||
242 | return 0; | |
3715a633 DS |
243 | } |
244 | ||
245 | int bsearch_midx(const struct object_id *oid, struct multi_pack_index *m, uint32_t *result) | |
246 | { | |
247 | return bsearch_hash(oid->hash, m->chunk_oid_fanout, m->chunk_oid_lookup, | |
aaa95dfa | 248 | the_hash_algo->rawsz, result); |
3715a633 DS |
249 | } |
250 | ||
8aac67a1 DS |
251 | struct object_id *nth_midxed_object_oid(struct object_id *oid, |
252 | struct multi_pack_index *m, | |
253 | uint32_t n) | |
254 | { | |
255 | if (n >= m->num_objects) | |
256 | return NULL; | |
257 | ||
258 | hashcpy(oid->hash, m->chunk_oid_lookup + m->hash_len * n); | |
259 | return oid; | |
260 | } | |
261 | ||
3715a633 DS |
262 | static off_t nth_midxed_offset(struct multi_pack_index *m, uint32_t pos) |
263 | { | |
264 | const unsigned char *offset_data; | |
265 | uint32_t offset32; | |
266 | ||
267 | offset_data = m->chunk_object_offsets + pos * MIDX_CHUNK_OFFSET_WIDTH; | |
268 | offset32 = get_be32(offset_data + sizeof(uint32_t)); | |
269 | ||
270 | if (m->chunk_large_offsets && offset32 & MIDX_LARGE_OFFSET_NEEDED) { | |
d8ac9ee1 | 271 | if (sizeof(off_t) < sizeof(uint64_t)) |
3715a633 DS |
272 | die(_("multi-pack-index stores a 64-bit offset, but off_t is too small")); |
273 | ||
274 | offset32 ^= MIDX_LARGE_OFFSET_NEEDED; | |
275 | return get_be64(m->chunk_large_offsets + sizeof(uint64_t) * offset32); | |
276 | } | |
277 | ||
278 | return offset32; | |
279 | } | |
280 | ||
281 | static uint32_t nth_midxed_pack_int_id(struct multi_pack_index *m, uint32_t pos) | |
282 | { | |
283 | return get_be32(m->chunk_object_offsets + pos * MIDX_CHUNK_OFFSET_WIDTH); | |
284 | } | |
285 | ||
64404a24 DS |
286 | static int nth_midxed_pack_entry(struct repository *r, |
287 | struct multi_pack_index *m, | |
288 | struct pack_entry *e, | |
289 | uint32_t pos) | |
3715a633 DS |
290 | { |
291 | uint32_t pack_int_id; | |
292 | struct packed_git *p; | |
293 | ||
294 | if (pos >= m->num_objects) | |
295 | return 0; | |
296 | ||
297 | pack_int_id = nth_midxed_pack_int_id(m, pos); | |
298 | ||
64404a24 | 299 | if (prepare_midx_pack(r, m, pack_int_id)) |
3715a633 DS |
300 | die(_("error preparing packfile from multi-pack-index")); |
301 | p = m->packs[pack_int_id]; | |
302 | ||
303 | /* | |
304 | * We are about to tell the caller where they can locate the | |
305 | * requested object. We better make sure the packfile is | |
306 | * still here and can be accessed before supplying that | |
307 | * answer, as it may have been deleted since the MIDX was | |
308 | * loaded! | |
309 | */ | |
310 | if (!is_pack_valid(p)) | |
311 | return 0; | |
312 | ||
c39b02ae DS |
313 | if (p->num_bad_objects) { |
314 | uint32_t i; | |
315 | struct object_id oid; | |
316 | nth_midxed_object_oid(&oid, m, pos); | |
317 | for (i = 0; i < p->num_bad_objects; i++) | |
e43d2dcc JK |
318 | if (hasheq(oid.hash, |
319 | p->bad_object_sha1 + the_hash_algo->rawsz * i)) | |
c39b02ae DS |
320 | return 0; |
321 | } | |
322 | ||
3715a633 DS |
323 | e->offset = nth_midxed_offset(m, pos); |
324 | e->p = p; | |
325 | ||
326 | return 1; | |
327 | } | |
328 | ||
64404a24 DS |
329 | int fill_midx_entry(struct repository * r, |
330 | const struct object_id *oid, | |
331 | struct pack_entry *e, | |
332 | struct multi_pack_index *m) | |
3715a633 DS |
333 | { |
334 | uint32_t pos; | |
335 | ||
336 | if (!bsearch_midx(oid, m, &pos)) | |
337 | return 0; | |
338 | ||
64404a24 | 339 | return nth_midxed_pack_entry(r, m, e, pos); |
3715a633 DS |
340 | } |
341 | ||
013fd7ad JK |
342 | /* Match "foo.idx" against either "foo.pack" _or_ "foo.idx". */ |
343 | static int cmp_idx_or_pack_name(const char *idx_or_pack_name, | |
344 | const char *idx_name) | |
345 | { | |
346 | /* Skip past any initial matching prefix. */ | |
347 | while (*idx_name && *idx_name == *idx_or_pack_name) { | |
348 | idx_name++; | |
349 | idx_or_pack_name++; | |
350 | } | |
351 | ||
352 | /* | |
353 | * If we didn't match completely, we may have matched "pack-1234." and | |
354 | * be left with "idx" and "pack" respectively, which is also OK. We do | |
355 | * not have to check for "idx" and "idx", because that would have been | |
356 | * a complete match (and in that case these strcmps will be false, but | |
357 | * we'll correctly return 0 from the final strcmp() below. | |
358 | * | |
359 | * Technically this matches "fooidx" and "foopack", but we'd never have | |
360 | * such names in the first place. | |
361 | */ | |
362 | if (!strcmp(idx_name, "idx") && !strcmp(idx_or_pack_name, "pack")) | |
363 | return 0; | |
364 | ||
365 | /* | |
366 | * This not only checks for a complete match, but also orders based on | |
367 | * the first non-identical character, which means our ordering will | |
368 | * match a raw strcmp(). That makes it OK to use this to binary search | |
369 | * a naively-sorted list. | |
370 | */ | |
371 | return strcmp(idx_or_pack_name, idx_name); | |
372 | } | |
373 | ||
374 | int midx_contains_pack(struct multi_pack_index *m, const char *idx_or_pack_name) | |
a40498a1 DS |
375 | { |
376 | uint32_t first = 0, last = m->num_packs; | |
377 | ||
378 | while (first < last) { | |
379 | uint32_t mid = first + (last - first) / 2; | |
380 | const char *current; | |
381 | int cmp; | |
382 | ||
383 | current = m->pack_names[mid]; | |
013fd7ad | 384 | cmp = cmp_idx_or_pack_name(idx_or_pack_name, current); |
a40498a1 DS |
385 | if (!cmp) |
386 | return 1; | |
387 | if (cmp > 0) { | |
388 | first = mid + 1; | |
389 | continue; | |
390 | } | |
391 | last = mid; | |
392 | } | |
393 | ||
394 | return 0; | |
395 | } | |
396 | ||
2cf489a3 | 397 | int prepare_multi_pack_index_one(struct repository *r, const char *object_dir, int local) |
c4d25228 | 398 | { |
29e2016b | 399 | struct multi_pack_index *m; |
c4d25228 DS |
400 | struct multi_pack_index *m_search; |
401 | int config_value; | |
0465a505 | 402 | static int env_value = -1; |
c4d25228 | 403 | |
0465a505 DS |
404 | if (env_value < 0) |
405 | env_value = git_env_bool(GIT_TEST_MULTI_PACK_INDEX, 0); | |
c4d25228 | 406 | |
0465a505 DS |
407 | if (!env_value && |
408 | (repo_config_get_bool(r, "core.multipackindex", &config_value) || | |
409 | !config_value)) | |
c4d25228 DS |
410 | return 0; |
411 | ||
29e2016b | 412 | for (m_search = r->objects->multi_pack_index; m_search; m_search = m_search->next) |
c4d25228 DS |
413 | if (!strcmp(object_dir, m_search->object_dir)) |
414 | return 1; | |
415 | ||
29e2016b | 416 | m = load_multi_pack_index(object_dir, local); |
c4d25228 | 417 | |
29e2016b | 418 | if (m) { |
59552fb3 TB |
419 | struct multi_pack_index *mp = r->objects->multi_pack_index; |
420 | if (mp) { | |
421 | m->next = mp->next; | |
422 | mp->next = m; | |
423 | } else | |
424 | r->objects->multi_pack_index = m; | |
c4d25228 DS |
425 | return 1; |
426 | } | |
427 | ||
428 | return 0; | |
429 | } | |
430 | ||
fc59e748 DS |
431 | static size_t write_midx_header(struct hashfile *f, |
432 | unsigned char num_chunks, | |
433 | uint32_t num_packs) | |
434 | { | |
fc59e748 | 435 | hashwrite_be32(f, MIDX_SIGNATURE); |
014f1447 RS |
436 | hashwrite_u8(f, MIDX_VERSION); |
437 | hashwrite_u8(f, oid_version()); | |
438 | hashwrite_u8(f, num_chunks); | |
439 | hashwrite_u8(f, 0); /* unused */ | |
fc59e748 DS |
440 | hashwrite_be32(f, num_packs); |
441 | ||
442 | return MIDX_HEADER_SIZE; | |
443 | } | |
444 | ||
d01bf2e6 DS |
445 | struct pack_info { |
446 | uint32_t orig_pack_int_id; | |
447 | char *pack_name; | |
448 | struct packed_git *p; | |
19575c7c | 449 | unsigned expired : 1; |
d01bf2e6 DS |
450 | }; |
451 | ||
452 | static int pack_info_compare(const void *_a, const void *_b) | |
453 | { | |
454 | struct pack_info *a = (struct pack_info *)_a; | |
455 | struct pack_info *b = (struct pack_info *)_b; | |
456 | return strcmp(a->pack_name, b->pack_name); | |
457 | } | |
458 | ||
396f2570 | 459 | struct pack_list { |
d01bf2e6 | 460 | struct pack_info *info; |
396f2570 | 461 | uint32_t nr; |
d01bf2e6 | 462 | uint32_t alloc; |
a40498a1 | 463 | struct multi_pack_index *m; |
840cef0c WB |
464 | struct progress *progress; |
465 | unsigned pack_paths_checked; | |
396f2570 DS |
466 | }; |
467 | ||
468 | static void add_pack_to_midx(const char *full_path, size_t full_path_len, | |
469 | const char *file_name, void *data) | |
470 | { | |
471 | struct pack_list *packs = (struct pack_list *)data; | |
472 | ||
473 | if (ends_with(file_name, ".idx")) { | |
840cef0c | 474 | display_progress(packs->progress, ++packs->pack_paths_checked); |
a40498a1 DS |
475 | if (packs->m && midx_contains_pack(packs->m, file_name)) |
476 | return; | |
477 | ||
d01bf2e6 | 478 | ALLOC_GROW(packs->info, packs->nr + 1, packs->alloc); |
396f2570 | 479 | |
d01bf2e6 DS |
480 | packs->info[packs->nr].p = add_packed_git(full_path, |
481 | full_path_len, | |
482 | 0); | |
fe1ed56f | 483 | |
d01bf2e6 | 484 | if (!packs->info[packs->nr].p) { |
396f2570 DS |
485 | warning(_("failed to add packfile '%s'"), |
486 | full_path); | |
487 | return; | |
488 | } | |
489 | ||
d01bf2e6 | 490 | if (open_pack_index(packs->info[packs->nr].p)) { |
fe1ed56f DS |
491 | warning(_("failed to open pack-index '%s'"), |
492 | full_path); | |
d01bf2e6 DS |
493 | close_pack(packs->info[packs->nr].p); |
494 | FREE_AND_NULL(packs->info[packs->nr].p); | |
fe1ed56f DS |
495 | return; |
496 | } | |
497 | ||
d01bf2e6 DS |
498 | packs->info[packs->nr].pack_name = xstrdup(file_name); |
499 | packs->info[packs->nr].orig_pack_int_id = packs->nr; | |
19575c7c | 500 | packs->info[packs->nr].expired = 0; |
396f2570 DS |
501 | packs->nr++; |
502 | } | |
503 | } | |
504 | ||
fe1ed56f DS |
505 | struct pack_midx_entry { |
506 | struct object_id oid; | |
507 | uint32_t pack_int_id; | |
508 | time_t pack_mtime; | |
509 | uint64_t offset; | |
510 | }; | |
511 | ||
512 | static int midx_oid_compare(const void *_a, const void *_b) | |
513 | { | |
514 | const struct pack_midx_entry *a = (const struct pack_midx_entry *)_a; | |
515 | const struct pack_midx_entry *b = (const struct pack_midx_entry *)_b; | |
516 | int cmp = oidcmp(&a->oid, &b->oid); | |
517 | ||
518 | if (cmp) | |
519 | return cmp; | |
520 | ||
521 | if (a->pack_mtime > b->pack_mtime) | |
522 | return -1; | |
523 | else if (a->pack_mtime < b->pack_mtime) | |
524 | return 1; | |
525 | ||
526 | return a->pack_int_id - b->pack_int_id; | |
527 | } | |
528 | ||
a40498a1 | 529 | static int nth_midxed_pack_midx_entry(struct multi_pack_index *m, |
a40498a1 DS |
530 | struct pack_midx_entry *e, |
531 | uint32_t pos) | |
532 | { | |
533 | if (pos >= m->num_objects) | |
534 | return 1; | |
535 | ||
536 | nth_midxed_object_oid(&e->oid, m, pos); | |
d01bf2e6 | 537 | e->pack_int_id = nth_midxed_pack_int_id(m, pos); |
a40498a1 DS |
538 | e->offset = nth_midxed_offset(m, pos); |
539 | ||
540 | /* consider objects in midx to be from "old" packs */ | |
541 | e->pack_mtime = 0; | |
542 | return 0; | |
543 | } | |
544 | ||
fe1ed56f DS |
545 | static void fill_pack_entry(uint32_t pack_int_id, |
546 | struct packed_git *p, | |
547 | uint32_t cur_object, | |
548 | struct pack_midx_entry *entry) | |
549 | { | |
0763671b | 550 | if (nth_packed_object_id(&entry->oid, p, cur_object) < 0) |
fe1ed56f DS |
551 | die(_("failed to locate object %d in packfile"), cur_object); |
552 | ||
553 | entry->pack_int_id = pack_int_id; | |
554 | entry->pack_mtime = p->mtime; | |
555 | ||
556 | entry->offset = nth_packed_object_offset(p, cur_object); | |
557 | } | |
558 | ||
559 | /* | |
560 | * It is possible to artificially get into a state where there are many | |
561 | * duplicate copies of objects. That can create high memory pressure if | |
562 | * we are to create a list of all objects before de-duplication. To reduce | |
563 | * this memory pressure without a significant performance drop, automatically | |
564 | * group objects by the first byte of their object id. Use the IDX fanout | |
565 | * tables to group the data, copy to a local array, then sort. | |
566 | * | |
567 | * Copy only the de-duplicated entries (selected by most-recent modified time | |
568 | * of a packfile containing the object). | |
569 | */ | |
a40498a1 | 570 | static struct pack_midx_entry *get_sorted_entries(struct multi_pack_index *m, |
d01bf2e6 | 571 | struct pack_info *info, |
fe1ed56f DS |
572 | uint32_t nr_packs, |
573 | uint32_t *nr_objects) | |
574 | { | |
575 | uint32_t cur_fanout, cur_pack, cur_object; | |
576 | uint32_t alloc_fanout, alloc_objects, total_objects = 0; | |
577 | struct pack_midx_entry *entries_by_fanout = NULL; | |
578 | struct pack_midx_entry *deduplicated_entries = NULL; | |
a40498a1 | 579 | uint32_t start_pack = m ? m->num_packs : 0; |
fe1ed56f | 580 | |
a40498a1 | 581 | for (cur_pack = start_pack; cur_pack < nr_packs; cur_pack++) |
d01bf2e6 | 582 | total_objects += info[cur_pack].p->num_objects; |
fe1ed56f DS |
583 | |
584 | /* | |
585 | * As we de-duplicate by fanout value, we expect the fanout | |
586 | * slices to be evenly distributed, with some noise. Hence, | |
587 | * allocate slightly more than one 256th. | |
588 | */ | |
589 | alloc_objects = alloc_fanout = total_objects > 3200 ? total_objects / 200 : 16; | |
590 | ||
591 | ALLOC_ARRAY(entries_by_fanout, alloc_fanout); | |
592 | ALLOC_ARRAY(deduplicated_entries, alloc_objects); | |
593 | *nr_objects = 0; | |
594 | ||
595 | for (cur_fanout = 0; cur_fanout < 256; cur_fanout++) { | |
596 | uint32_t nr_fanout = 0; | |
597 | ||
a40498a1 DS |
598 | if (m) { |
599 | uint32_t start = 0, end; | |
600 | ||
601 | if (cur_fanout) | |
602 | start = ntohl(m->chunk_oid_fanout[cur_fanout - 1]); | |
603 | end = ntohl(m->chunk_oid_fanout[cur_fanout]); | |
604 | ||
605 | for (cur_object = start; cur_object < end; cur_object++) { | |
606 | ALLOC_GROW(entries_by_fanout, nr_fanout + 1, alloc_fanout); | |
d01bf2e6 | 607 | nth_midxed_pack_midx_entry(m, |
a40498a1 DS |
608 | &entries_by_fanout[nr_fanout], |
609 | cur_object); | |
610 | nr_fanout++; | |
611 | } | |
612 | } | |
613 | ||
614 | for (cur_pack = start_pack; cur_pack < nr_packs; cur_pack++) { | |
fe1ed56f DS |
615 | uint32_t start = 0, end; |
616 | ||
617 | if (cur_fanout) | |
d01bf2e6 DS |
618 | start = get_pack_fanout(info[cur_pack].p, cur_fanout - 1); |
619 | end = get_pack_fanout(info[cur_pack].p, cur_fanout); | |
fe1ed56f DS |
620 | |
621 | for (cur_object = start; cur_object < end; cur_object++) { | |
622 | ALLOC_GROW(entries_by_fanout, nr_fanout + 1, alloc_fanout); | |
d01bf2e6 | 623 | fill_pack_entry(cur_pack, info[cur_pack].p, cur_object, &entries_by_fanout[nr_fanout]); |
fe1ed56f DS |
624 | nr_fanout++; |
625 | } | |
626 | } | |
627 | ||
628 | QSORT(entries_by_fanout, nr_fanout, midx_oid_compare); | |
629 | ||
630 | /* | |
631 | * The batch is now sorted by OID and then mtime (descending). | |
632 | * Take only the first duplicate. | |
633 | */ | |
634 | for (cur_object = 0; cur_object < nr_fanout; cur_object++) { | |
e43d2dcc JK |
635 | if (cur_object && oideq(&entries_by_fanout[cur_object - 1].oid, |
636 | &entries_by_fanout[cur_object].oid)) | |
fe1ed56f DS |
637 | continue; |
638 | ||
639 | ALLOC_GROW(deduplicated_entries, *nr_objects + 1, alloc_objects); | |
640 | memcpy(&deduplicated_entries[*nr_objects], | |
641 | &entries_by_fanout[cur_object], | |
642 | sizeof(struct pack_midx_entry)); | |
643 | (*nr_objects)++; | |
644 | } | |
645 | } | |
646 | ||
647 | free(entries_by_fanout); | |
648 | return deduplicated_entries; | |
649 | } | |
650 | ||
32f3c541 | 651 | static size_t write_midx_pack_names(struct hashfile *f, |
d01bf2e6 | 652 | struct pack_info *info, |
32f3c541 DS |
653 | uint32_t num_packs) |
654 | { | |
655 | uint32_t i; | |
656 | unsigned char padding[MIDX_CHUNK_ALIGNMENT]; | |
657 | size_t written = 0; | |
658 | ||
659 | for (i = 0; i < num_packs; i++) { | |
19575c7c DS |
660 | size_t writelen; |
661 | ||
662 | if (info[i].expired) | |
663 | continue; | |
32f3c541 | 664 | |
d01bf2e6 | 665 | if (i && strcmp(info[i].pack_name, info[i - 1].pack_name) <= 0) |
32f3c541 | 666 | BUG("incorrect pack-file order: %s before %s", |
d01bf2e6 DS |
667 | info[i - 1].pack_name, |
668 | info[i].pack_name); | |
32f3c541 | 669 | |
19575c7c | 670 | writelen = strlen(info[i].pack_name) + 1; |
d01bf2e6 | 671 | hashwrite(f, info[i].pack_name, writelen); |
32f3c541 DS |
672 | written += writelen; |
673 | } | |
674 | ||
675 | /* add padding to be aligned */ | |
676 | i = MIDX_CHUNK_ALIGNMENT - (written % MIDX_CHUNK_ALIGNMENT); | |
677 | if (i < MIDX_CHUNK_ALIGNMENT) { | |
678 | memset(padding, 0, sizeof(padding)); | |
679 | hashwrite(f, padding, i); | |
680 | written += i; | |
681 | } | |
682 | ||
683 | return written; | |
684 | } | |
685 | ||
d7cacf29 DS |
686 | static size_t write_midx_oid_fanout(struct hashfile *f, |
687 | struct pack_midx_entry *objects, | |
688 | uint32_t nr_objects) | |
689 | { | |
690 | struct pack_midx_entry *list = objects; | |
691 | struct pack_midx_entry *last = objects + nr_objects; | |
692 | uint32_t count = 0; | |
693 | uint32_t i; | |
694 | ||
695 | /* | |
696 | * Write the first-level table (the list is sorted, | |
697 | * but we use a 256-entry lookup to be able to avoid | |
698 | * having to do eight extra binary search iterations). | |
699 | */ | |
700 | for (i = 0; i < 256; i++) { | |
701 | struct pack_midx_entry *next = list; | |
702 | ||
703 | while (next < last && next->oid.hash[0] == i) { | |
704 | count++; | |
705 | next++; | |
706 | } | |
707 | ||
708 | hashwrite_be32(f, count); | |
709 | list = next; | |
710 | } | |
711 | ||
712 | return MIDX_CHUNK_FANOUT_SIZE; | |
713 | } | |
714 | ||
0d5b3a5e DS |
715 | static size_t write_midx_oid_lookup(struct hashfile *f, unsigned char hash_len, |
716 | struct pack_midx_entry *objects, | |
717 | uint32_t nr_objects) | |
718 | { | |
719 | struct pack_midx_entry *list = objects; | |
720 | uint32_t i; | |
721 | size_t written = 0; | |
722 | ||
723 | for (i = 0; i < nr_objects; i++) { | |
724 | struct pack_midx_entry *obj = list++; | |
725 | ||
726 | if (i < nr_objects - 1) { | |
727 | struct pack_midx_entry *next = list; | |
728 | if (oidcmp(&obj->oid, &next->oid) >= 0) | |
729 | BUG("OIDs not in order: %s >= %s", | |
730 | oid_to_hex(&obj->oid), | |
731 | oid_to_hex(&next->oid)); | |
732 | } | |
733 | ||
734 | hashwrite(f, obj->oid.hash, (int)hash_len); | |
735 | written += hash_len; | |
736 | } | |
737 | ||
738 | return written; | |
739 | } | |
740 | ||
662148c4 | 741 | static size_t write_midx_object_offsets(struct hashfile *f, int large_offset_needed, |
d01bf2e6 | 742 | uint32_t *perm, |
662148c4 DS |
743 | struct pack_midx_entry *objects, uint32_t nr_objects) |
744 | { | |
745 | struct pack_midx_entry *list = objects; | |
746 | uint32_t i, nr_large_offset = 0; | |
747 | size_t written = 0; | |
748 | ||
749 | for (i = 0; i < nr_objects; i++) { | |
750 | struct pack_midx_entry *obj = list++; | |
751 | ||
19575c7c DS |
752 | if (perm[obj->pack_int_id] == PACK_EXPIRED) |
753 | BUG("object %s is in an expired pack with int-id %d", | |
754 | oid_to_hex(&obj->oid), | |
755 | obj->pack_int_id); | |
756 | ||
d01bf2e6 | 757 | hashwrite_be32(f, perm[obj->pack_int_id]); |
662148c4 DS |
758 | |
759 | if (large_offset_needed && obj->offset >> 31) | |
760 | hashwrite_be32(f, MIDX_LARGE_OFFSET_NEEDED | nr_large_offset++); | |
761 | else if (!large_offset_needed && obj->offset >> 32) | |
762 | BUG("object %s requires a large offset (%"PRIx64") but the MIDX is not writing large offsets!", | |
763 | oid_to_hex(&obj->oid), | |
764 | obj->offset); | |
765 | else | |
766 | hashwrite_be32(f, (uint32_t)obj->offset); | |
767 | ||
768 | written += MIDX_CHUNK_OFFSET_WIDTH; | |
769 | } | |
770 | ||
771 | return written; | |
772 | } | |
773 | ||
774 | static size_t write_midx_large_offsets(struct hashfile *f, uint32_t nr_large_offset, | |
775 | struct pack_midx_entry *objects, uint32_t nr_objects) | |
776 | { | |
61b0fcbb | 777 | struct pack_midx_entry *list = objects, *end = objects + nr_objects; |
662148c4 DS |
778 | size_t written = 0; |
779 | ||
780 | while (nr_large_offset) { | |
61b0fcbb JK |
781 | struct pack_midx_entry *obj; |
782 | uint64_t offset; | |
783 | ||
784 | if (list >= end) | |
785 | BUG("too many large-offset objects"); | |
786 | ||
787 | obj = list++; | |
788 | offset = obj->offset; | |
662148c4 DS |
789 | |
790 | if (!(offset >> 31)) | |
791 | continue; | |
792 | ||
793 | hashwrite_be32(f, offset >> 32); | |
794 | hashwrite_be32(f, offset & 0xffffffffUL); | |
795 | written += 2 * sizeof(uint32_t); | |
796 | ||
797 | nr_large_offset--; | |
798 | } | |
799 | ||
800 | return written; | |
801 | } | |
802 | ||
19575c7c | 803 | static int write_midx_internal(const char *object_dir, struct multi_pack_index *m, |
840cef0c | 804 | struct string_list *packs_to_drop, unsigned flags) |
a3407730 | 805 | { |
32f3c541 | 806 | unsigned char cur_chunk, num_chunks = 0; |
fc59e748 | 807 | char *midx_name; |
396f2570 | 808 | uint32_t i; |
fc59e748 DS |
809 | struct hashfile *f = NULL; |
810 | struct lock_file lk; | |
396f2570 | 811 | struct pack_list packs; |
32f3c541 DS |
812 | uint32_t *pack_perm = NULL; |
813 | uint64_t written = 0; | |
814 | uint32_t chunk_ids[MIDX_MAX_CHUNKS + 1]; | |
815 | uint64_t chunk_offsets[MIDX_MAX_CHUNKS + 1]; | |
662148c4 | 816 | uint32_t nr_entries, num_large_offsets = 0; |
fe1ed56f | 817 | struct pack_midx_entry *entries = NULL; |
840cef0c | 818 | struct progress *progress = NULL; |
662148c4 | 819 | int large_offsets_needed = 0; |
dba6175c | 820 | int pack_name_concat_len = 0; |
19575c7c DS |
821 | int dropped_packs = 0; |
822 | int result = 0; | |
fc59e748 DS |
823 | |
824 | midx_name = get_midx_filename(object_dir); | |
d5e1961c | 825 | if (safe_create_leading_directories(midx_name)) |
fc59e748 DS |
826 | die_errno(_("unable to create leading directories of %s"), |
827 | midx_name); | |
fc59e748 | 828 | |
19575c7c DS |
829 | if (m) |
830 | packs.m = m; | |
831 | else | |
832 | packs.m = load_multi_pack_index(object_dir, 1); | |
a40498a1 | 833 | |
396f2570 | 834 | packs.nr = 0; |
d01bf2e6 DS |
835 | packs.alloc = packs.m ? packs.m->num_packs : 16; |
836 | packs.info = NULL; | |
837 | ALLOC_ARRAY(packs.info, packs.alloc); | |
396f2570 | 838 | |
a40498a1 DS |
839 | if (packs.m) { |
840 | for (i = 0; i < packs.m->num_packs; i++) { | |
d01bf2e6 | 841 | ALLOC_GROW(packs.info, packs.nr + 1, packs.alloc); |
a40498a1 | 842 | |
d01bf2e6 DS |
843 | packs.info[packs.nr].orig_pack_int_id = i; |
844 | packs.info[packs.nr].pack_name = xstrdup(packs.m->pack_names[i]); | |
845 | packs.info[packs.nr].p = NULL; | |
19575c7c | 846 | packs.info[packs.nr].expired = 0; |
a40498a1 DS |
847 | packs.nr++; |
848 | } | |
849 | } | |
850 | ||
840cef0c WB |
851 | packs.pack_paths_checked = 0; |
852 | if (flags & MIDX_PROGRESS) | |
853 | packs.progress = start_progress(_("Adding packfiles to multi-pack-index"), 0); | |
854 | else | |
855 | packs.progress = NULL; | |
856 | ||
396f2570 | 857 | for_each_file_in_pack_dir(object_dir, add_pack_to_midx, &packs); |
840cef0c | 858 | stop_progress(&packs.progress); |
396f2570 | 859 | |
19575c7c | 860 | if (packs.m && packs.nr == packs.m->num_packs && !packs_to_drop) |
a40498a1 DS |
861 | goto cleanup; |
862 | ||
d01bf2e6 | 863 | entries = get_sorted_entries(packs.m, packs.info, packs.nr, &nr_entries); |
a40498a1 | 864 | |
662148c4 DS |
865 | for (i = 0; i < nr_entries; i++) { |
866 | if (entries[i].offset > 0x7fffffff) | |
867 | num_large_offsets++; | |
868 | if (entries[i].offset > 0xffffffff) | |
869 | large_offsets_needed = 1; | |
870 | } | |
fe1ed56f | 871 | |
d01bf2e6 DS |
872 | QSORT(packs.info, packs.nr, pack_info_compare); |
873 | ||
19575c7c DS |
874 | if (packs_to_drop && packs_to_drop->nr) { |
875 | int drop_index = 0; | |
876 | int missing_drops = 0; | |
877 | ||
878 | for (i = 0; i < packs.nr && drop_index < packs_to_drop->nr; i++) { | |
879 | int cmp = strcmp(packs.info[i].pack_name, | |
880 | packs_to_drop->items[drop_index].string); | |
881 | ||
882 | if (!cmp) { | |
883 | drop_index++; | |
884 | packs.info[i].expired = 1; | |
885 | } else if (cmp > 0) { | |
886 | error(_("did not see pack-file %s to drop"), | |
887 | packs_to_drop->items[drop_index].string); | |
888 | drop_index++; | |
889 | missing_drops++; | |
890 | i--; | |
891 | } else { | |
892 | packs.info[i].expired = 0; | |
893 | } | |
894 | } | |
895 | ||
896 | if (missing_drops) { | |
897 | result = 1; | |
898 | goto cleanup; | |
899 | } | |
900 | } | |
901 | ||
d01bf2e6 DS |
902 | /* |
903 | * pack_perm stores a permutation between pack-int-ids from the | |
904 | * previous multi-pack-index to the new one we are writing: | |
905 | * | |
906 | * pack_perm[old_id] = new_id | |
907 | */ | |
908 | ALLOC_ARRAY(pack_perm, packs.nr); | |
909 | for (i = 0; i < packs.nr; i++) { | |
19575c7c DS |
910 | if (packs.info[i].expired) { |
911 | dropped_packs++; | |
912 | pack_perm[packs.info[i].orig_pack_int_id] = PACK_EXPIRED; | |
913 | } else { | |
914 | pack_perm[packs.info[i].orig_pack_int_id] = i - dropped_packs; | |
915 | } | |
d01bf2e6 DS |
916 | } |
917 | ||
19575c7c DS |
918 | for (i = 0; i < packs.nr; i++) { |
919 | if (!packs.info[i].expired) | |
920 | pack_name_concat_len += strlen(packs.info[i].pack_name) + 1; | |
921 | } | |
dba6175c DS |
922 | |
923 | if (pack_name_concat_len % MIDX_CHUNK_ALIGNMENT) | |
924 | pack_name_concat_len += MIDX_CHUNK_ALIGNMENT - | |
925 | (pack_name_concat_len % MIDX_CHUNK_ALIGNMENT); | |
926 | ||
fc59e748 DS |
927 | hold_lock_file_for_update(&lk, midx_name, LOCK_DIE_ON_ERROR); |
928 | f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf); | |
929 | FREE_AND_NULL(midx_name); | |
930 | ||
a40498a1 DS |
931 | if (packs.m) |
932 | close_midx(packs.m); | |
933 | ||
32f3c541 | 934 | cur_chunk = 0; |
662148c4 | 935 | num_chunks = large_offsets_needed ? 5 : 4; |
32f3c541 | 936 | |
796d61cd DR |
937 | if (packs.nr - dropped_packs == 0) { |
938 | error(_("no pack files to index.")); | |
939 | result = 1; | |
940 | goto cleanup; | |
941 | } | |
942 | ||
19575c7c | 943 | written = write_midx_header(f, num_chunks, packs.nr - dropped_packs); |
32f3c541 DS |
944 | |
945 | chunk_ids[cur_chunk] = MIDX_CHUNKID_PACKNAMES; | |
946 | chunk_offsets[cur_chunk] = written + (num_chunks + 1) * MIDX_CHUNKLOOKUP_WIDTH; | |
947 | ||
948 | cur_chunk++; | |
d7cacf29 | 949 | chunk_ids[cur_chunk] = MIDX_CHUNKID_OIDFANOUT; |
dba6175c | 950 | chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + pack_name_concat_len; |
32f3c541 | 951 | |
d7cacf29 DS |
952 | cur_chunk++; |
953 | chunk_ids[cur_chunk] = MIDX_CHUNKID_OIDLOOKUP; | |
954 | chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + MIDX_CHUNK_FANOUT_SIZE; | |
955 | ||
0d5b3a5e | 956 | cur_chunk++; |
662148c4 | 957 | chunk_ids[cur_chunk] = MIDX_CHUNKID_OBJECTOFFSETS; |
aaa95dfa | 958 | chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + nr_entries * the_hash_algo->rawsz; |
0d5b3a5e | 959 | |
662148c4 DS |
960 | cur_chunk++; |
961 | chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + nr_entries * MIDX_CHUNK_OFFSET_WIDTH; | |
962 | if (large_offsets_needed) { | |
963 | chunk_ids[cur_chunk] = MIDX_CHUNKID_LARGEOFFSETS; | |
964 | ||
965 | cur_chunk++; | |
966 | chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + | |
967 | num_large_offsets * MIDX_CHUNK_LARGE_OFFSET_WIDTH; | |
968 | } | |
969 | ||
970 | chunk_ids[cur_chunk] = 0; | |
971 | ||
32f3c541 DS |
972 | for (i = 0; i <= num_chunks; i++) { |
973 | if (i && chunk_offsets[i] < chunk_offsets[i - 1]) | |
974 | BUG("incorrect chunk offsets: %"PRIu64" before %"PRIu64, | |
975 | chunk_offsets[i - 1], | |
976 | chunk_offsets[i]); | |
977 | ||
978 | if (chunk_offsets[i] % MIDX_CHUNK_ALIGNMENT) | |
979 | BUG("chunk offset %"PRIu64" is not properly aligned", | |
980 | chunk_offsets[i]); | |
981 | ||
982 | hashwrite_be32(f, chunk_ids[i]); | |
983 | hashwrite_be32(f, chunk_offsets[i] >> 32); | |
984 | hashwrite_be32(f, chunk_offsets[i]); | |
985 | ||
986 | written += MIDX_CHUNKLOOKUP_WIDTH; | |
987 | } | |
988 | ||
840cef0c WB |
989 | if (flags & MIDX_PROGRESS) |
990 | progress = start_progress(_("Writing chunks to multi-pack-index"), | |
991 | num_chunks); | |
32f3c541 DS |
992 | for (i = 0; i < num_chunks; i++) { |
993 | if (written != chunk_offsets[i]) | |
994 | BUG("incorrect chunk offset (%"PRIu64" != %"PRIu64") for chunk id %"PRIx32, | |
995 | chunk_offsets[i], | |
996 | written, | |
997 | chunk_ids[i]); | |
998 | ||
999 | switch (chunk_ids[i]) { | |
1000 | case MIDX_CHUNKID_PACKNAMES: | |
d01bf2e6 | 1001 | written += write_midx_pack_names(f, packs.info, packs.nr); |
32f3c541 DS |
1002 | break; |
1003 | ||
d7cacf29 DS |
1004 | case MIDX_CHUNKID_OIDFANOUT: |
1005 | written += write_midx_oid_fanout(f, entries, nr_entries); | |
1006 | break; | |
1007 | ||
0d5b3a5e | 1008 | case MIDX_CHUNKID_OIDLOOKUP: |
aaa95dfa | 1009 | written += write_midx_oid_lookup(f, the_hash_algo->rawsz, entries, nr_entries); |
0d5b3a5e DS |
1010 | break; |
1011 | ||
662148c4 | 1012 | case MIDX_CHUNKID_OBJECTOFFSETS: |
d01bf2e6 | 1013 | written += write_midx_object_offsets(f, large_offsets_needed, pack_perm, entries, nr_entries); |
662148c4 DS |
1014 | break; |
1015 | ||
1016 | case MIDX_CHUNKID_LARGEOFFSETS: | |
1017 | written += write_midx_large_offsets(f, num_large_offsets, entries, nr_entries); | |
1018 | break; | |
1019 | ||
32f3c541 DS |
1020 | default: |
1021 | BUG("trying to write unknown chunk id %"PRIx32, | |
1022 | chunk_ids[i]); | |
1023 | } | |
840cef0c WB |
1024 | |
1025 | display_progress(progress, i + 1); | |
32f3c541 | 1026 | } |
840cef0c | 1027 | stop_progress(&progress); |
32f3c541 DS |
1028 | |
1029 | if (written != chunk_offsets[num_chunks]) | |
1030 | BUG("incorrect final offset %"PRIu64" != %"PRIu64, | |
1031 | written, | |
1032 | chunk_offsets[num_chunks]); | |
fc59e748 DS |
1033 | |
1034 | finalize_hashfile(f, NULL, CSUM_FSYNC | CSUM_HASH_IN_STREAM); | |
1035 | commit_lock_file(&lk); | |
1036 | ||
a40498a1 | 1037 | cleanup: |
396f2570 | 1038 | for (i = 0; i < packs.nr; i++) { |
d01bf2e6 DS |
1039 | if (packs.info[i].p) { |
1040 | close_pack(packs.info[i].p); | |
1041 | free(packs.info[i].p); | |
396f2570 | 1042 | } |
d01bf2e6 | 1043 | free(packs.info[i].pack_name); |
396f2570 DS |
1044 | } |
1045 | ||
d01bf2e6 | 1046 | free(packs.info); |
fe1ed56f | 1047 | free(entries); |
a40498a1 DS |
1048 | free(pack_perm); |
1049 | free(midx_name); | |
19575c7c DS |
1050 | return result; |
1051 | } | |
1052 | ||
efbc3aee | 1053 | int write_midx_file(const char *object_dir, unsigned flags) |
19575c7c | 1054 | { |
840cef0c | 1055 | return write_midx_internal(object_dir, NULL, NULL, flags); |
a3407730 | 1056 | } |
525e18c0 | 1057 | |
1dcd9f20 | 1058 | void clear_midx_file(struct repository *r) |
525e18c0 | 1059 | { |
3b2f8a02 | 1060 | char *midx = get_midx_filename(r->objects->odb->path); |
1dcd9f20 DS |
1061 | |
1062 | if (r->objects && r->objects->multi_pack_index) { | |
1063 | close_midx(r->objects->multi_pack_index); | |
1064 | r->objects->multi_pack_index = NULL; | |
1065 | } | |
525e18c0 | 1066 | |
d5e1961c | 1067 | if (remove_path(midx)) |
525e18c0 | 1068 | die(_("failed to clear multi-pack-index at %s"), midx); |
525e18c0 DS |
1069 | |
1070 | free(midx); | |
1071 | } | |
56ee7ff1 DS |
1072 | |
1073 | static int verify_midx_error; | |
1074 | ||
d4bf1d88 DS |
1075 | static void midx_report(const char *fmt, ...) |
1076 | { | |
1077 | va_list ap; | |
1078 | verify_midx_error = 1; | |
1079 | va_start(ap, fmt); | |
1080 | vfprintf(stderr, fmt, ap); | |
1081 | fprintf(stderr, "\n"); | |
1082 | va_end(ap); | |
1083 | } | |
1084 | ||
5ae18df9 JH |
1085 | struct pair_pos_vs_id |
1086 | { | |
1087 | uint32_t pos; | |
1088 | uint32_t pack_int_id; | |
1089 | }; | |
1090 | ||
1091 | static int compare_pair_pos_vs_id(const void *_a, const void *_b) | |
1092 | { | |
1093 | struct pair_pos_vs_id *a = (struct pair_pos_vs_id *)_a; | |
1094 | struct pair_pos_vs_id *b = (struct pair_pos_vs_id *)_b; | |
1095 | ||
1096 | return b->pack_int_id - a->pack_int_id; | |
1097 | } | |
1098 | ||
430efb8a JH |
1099 | /* |
1100 | * Limit calls to display_progress() for performance reasons. | |
1101 | * The interval here was arbitrarily chosen. | |
1102 | */ | |
1103 | #define SPARSE_PROGRESS_INTERVAL (1 << 12) | |
1104 | #define midx_display_sparse_progress(progress, n) \ | |
1105 | do { \ | |
1106 | uint64_t _n = (n); \ | |
1107 | if ((_n & (SPARSE_PROGRESS_INTERVAL - 1)) == 0) \ | |
1108 | display_progress(progress, _n); \ | |
1109 | } while (0) | |
1110 | ||
efbc3aee | 1111 | int verify_midx_file(struct repository *r, const char *object_dir, unsigned flags) |
56ee7ff1 | 1112 | { |
5ae18df9 | 1113 | struct pair_pos_vs_id *pairs = NULL; |
d4bf1d88 | 1114 | uint32_t i; |
ad60096d | 1115 | struct progress *progress = NULL; |
56ee7ff1 DS |
1116 | struct multi_pack_index *m = load_multi_pack_index(object_dir, 1); |
1117 | verify_midx_error = 0; | |
1118 | ||
d9607542 DS |
1119 | if (!m) { |
1120 | int result = 0; | |
1121 | struct stat sb; | |
1122 | char *filename = get_midx_filename(object_dir); | |
1123 | if (!stat(filename, &sb)) { | |
1124 | error(_("multi-pack-index file exists, but failed to parse")); | |
1125 | result = 1; | |
1126 | } | |
1127 | free(filename); | |
1128 | return result; | |
1129 | } | |
56ee7ff1 | 1130 | |
ad60096d WB |
1131 | if (flags & MIDX_PROGRESS) |
1132 | progress = start_progress(_("Looking for referenced packfiles"), | |
1133 | m->num_packs); | |
d4bf1d88 | 1134 | for (i = 0; i < m->num_packs; i++) { |
64404a24 | 1135 | if (prepare_midx_pack(r, m, i)) |
d4bf1d88 | 1136 | midx_report("failed to load pack in position %d", i); |
430efb8a JH |
1137 | |
1138 | display_progress(progress, i + 1); | |
d4bf1d88 | 1139 | } |
430efb8a | 1140 | stop_progress(&progress); |
d4bf1d88 | 1141 | |
2f23d3f3 DS |
1142 | for (i = 0; i < 255; i++) { |
1143 | uint32_t oid_fanout1 = ntohl(m->chunk_oid_fanout[i]); | |
1144 | uint32_t oid_fanout2 = ntohl(m->chunk_oid_fanout[i + 1]); | |
1145 | ||
1146 | if (oid_fanout1 > oid_fanout2) | |
1147 | midx_report(_("oid fanout out of order: fanout[%d] = %"PRIx32" > %"PRIx32" = fanout[%d]"), | |
1148 | i, oid_fanout1, oid_fanout2, i + 1); | |
1149 | } | |
1150 | ||
796d61cd DR |
1151 | if (m->num_objects == 0) { |
1152 | midx_report(_("the midx contains no oid")); | |
1153 | /* | |
1154 | * Remaining tests assume that we have objects, so we can | |
1155 | * return here. | |
1156 | */ | |
1157 | return verify_midx_error; | |
1158 | } | |
1159 | ||
ad60096d WB |
1160 | if (flags & MIDX_PROGRESS) |
1161 | progress = start_sparse_progress(_("Verifying OID order in multi-pack-index"), | |
1162 | m->num_objects - 1); | |
55c5648d DS |
1163 | for (i = 0; i < m->num_objects - 1; i++) { |
1164 | struct object_id oid1, oid2; | |
1165 | ||
1166 | nth_midxed_object_oid(&oid1, m, i); | |
1167 | nth_midxed_object_oid(&oid2, m, i + 1); | |
1168 | ||
1169 | if (oidcmp(&oid1, &oid2) >= 0) | |
1170 | midx_report(_("oid lookup out of order: oid[%d] = %s >= %s = oid[%d]"), | |
1171 | i, oid_to_hex(&oid1), oid_to_hex(&oid2), i + 1); | |
430efb8a JH |
1172 | |
1173 | midx_display_sparse_progress(progress, i + 1); | |
55c5648d | 1174 | } |
430efb8a | 1175 | stop_progress(&progress); |
55c5648d | 1176 | |
5ae18df9 JH |
1177 | /* |
1178 | * Create an array mapping each object to its packfile id. Sort it | |
1179 | * to group the objects by packfile. Use this permutation to visit | |
1180 | * each of the objects and only require 1 packfile to be open at a | |
1181 | * time. | |
1182 | */ | |
1183 | ALLOC_ARRAY(pairs, m->num_objects); | |
1184 | for (i = 0; i < m->num_objects; i++) { | |
1185 | pairs[i].pos = i; | |
1186 | pairs[i].pack_int_id = nth_midxed_pack_int_id(m, i); | |
1187 | } | |
1188 | ||
ad60096d WB |
1189 | if (flags & MIDX_PROGRESS) |
1190 | progress = start_sparse_progress(_("Sorting objects by packfile"), | |
1191 | m->num_objects); | |
5ae18df9 JH |
1192 | display_progress(progress, 0); /* TODO: Measure QSORT() progress */ |
1193 | QSORT(pairs, m->num_objects, compare_pair_pos_vs_id); | |
1194 | stop_progress(&progress); | |
1195 | ||
ad60096d WB |
1196 | if (flags & MIDX_PROGRESS) |
1197 | progress = start_sparse_progress(_("Verifying object offsets"), m->num_objects); | |
cc6af73c DS |
1198 | for (i = 0; i < m->num_objects; i++) { |
1199 | struct object_id oid; | |
1200 | struct pack_entry e; | |
1201 | off_t m_offset, p_offset; | |
1202 | ||
5ae18df9 JH |
1203 | if (i > 0 && pairs[i-1].pack_int_id != pairs[i].pack_int_id && |
1204 | m->packs[pairs[i-1].pack_int_id]) | |
1205 | { | |
1206 | close_pack_fd(m->packs[pairs[i-1].pack_int_id]); | |
1207 | close_pack_index(m->packs[pairs[i-1].pack_int_id]); | |
1208 | } | |
1209 | ||
1210 | nth_midxed_object_oid(&oid, m, pairs[i].pos); | |
1211 | ||
64404a24 | 1212 | if (!fill_midx_entry(r, &oid, &e, m)) { |
cc6af73c | 1213 | midx_report(_("failed to load pack entry for oid[%d] = %s"), |
5ae18df9 | 1214 | pairs[i].pos, oid_to_hex(&oid)); |
cc6af73c DS |
1215 | continue; |
1216 | } | |
1217 | ||
1218 | if (open_pack_index(e.p)) { | |
1219 | midx_report(_("failed to load pack-index for packfile %s"), | |
1220 | e.p->pack_name); | |
1221 | break; | |
1222 | } | |
1223 | ||
1224 | m_offset = e.offset; | |
1225 | p_offset = find_pack_entry_one(oid.hash, e.p); | |
1226 | ||
1227 | if (m_offset != p_offset) | |
1228 | midx_report(_("incorrect object offset for oid[%d] = %s: %"PRIx64" != %"PRIx64), | |
5ae18df9 | 1229 | pairs[i].pos, oid_to_hex(&oid), m_offset, p_offset); |
144d7033 | 1230 | |
430efb8a | 1231 | midx_display_sparse_progress(progress, i + 1); |
cc6af73c | 1232 | } |
144d7033 | 1233 | stop_progress(&progress); |
cc6af73c | 1234 | |
5ae18df9 JH |
1235 | free(pairs); |
1236 | ||
56ee7ff1 DS |
1237 | return verify_midx_error; |
1238 | } | |
cff97116 | 1239 | |
efbc3aee | 1240 | int expire_midx_packs(struct repository *r, const char *object_dir, unsigned flags) |
cff97116 | 1241 | { |
19575c7c DS |
1242 | uint32_t i, *count, result = 0; |
1243 | struct string_list packs_to_drop = STRING_LIST_INIT_DUP; | |
1244 | struct multi_pack_index *m = load_multi_pack_index(object_dir, 1); | |
8dc18f89 | 1245 | struct progress *progress = NULL; |
19575c7c DS |
1246 | |
1247 | if (!m) | |
1248 | return 0; | |
1249 | ||
1250 | count = xcalloc(m->num_packs, sizeof(uint32_t)); | |
8dc18f89 WB |
1251 | |
1252 | if (flags & MIDX_PROGRESS) | |
1253 | progress = start_progress(_("Counting referenced objects"), | |
1254 | m->num_objects); | |
19575c7c DS |
1255 | for (i = 0; i < m->num_objects; i++) { |
1256 | int pack_int_id = nth_midxed_pack_int_id(m, i); | |
1257 | count[pack_int_id]++; | |
8dc18f89 | 1258 | display_progress(progress, i + 1); |
19575c7c | 1259 | } |
8dc18f89 | 1260 | stop_progress(&progress); |
19575c7c | 1261 | |
8dc18f89 WB |
1262 | if (flags & MIDX_PROGRESS) |
1263 | progress = start_progress(_("Finding and deleting unreferenced packfiles"), | |
1264 | m->num_packs); | |
19575c7c DS |
1265 | for (i = 0; i < m->num_packs; i++) { |
1266 | char *pack_name; | |
8dc18f89 | 1267 | display_progress(progress, i + 1); |
19575c7c DS |
1268 | |
1269 | if (count[i]) | |
1270 | continue; | |
1271 | ||
1272 | if (prepare_midx_pack(r, m, i)) | |
1273 | continue; | |
1274 | ||
1275 | if (m->packs[i]->pack_keep) | |
1276 | continue; | |
1277 | ||
1278 | pack_name = xstrdup(m->packs[i]->pack_name); | |
1279 | close_pack(m->packs[i]); | |
1280 | ||
1281 | string_list_insert(&packs_to_drop, m->pack_names[i]); | |
1282 | unlink_pack_path(pack_name, 0); | |
1283 | free(pack_name); | |
1284 | } | |
8dc18f89 | 1285 | stop_progress(&progress); |
19575c7c DS |
1286 | |
1287 | free(count); | |
1288 | ||
1289 | if (packs_to_drop.nr) | |
840cef0c | 1290 | result = write_midx_internal(object_dir, m, &packs_to_drop, flags); |
19575c7c DS |
1291 | |
1292 | string_list_clear(&packs_to_drop, 0); | |
1293 | return result; | |
cff97116 | 1294 | } |
2af890bb | 1295 | |
ce1e4a10 DS |
1296 | struct repack_info { |
1297 | timestamp_t mtime; | |
1298 | uint32_t referenced_objects; | |
1299 | uint32_t pack_int_id; | |
1300 | }; | |
1301 | ||
1302 | static int compare_by_mtime(const void *a_, const void *b_) | |
2af890bb | 1303 | { |
ce1e4a10 DS |
1304 | const struct repack_info *a, *b; |
1305 | ||
1306 | a = (const struct repack_info *)a_; | |
1307 | b = (const struct repack_info *)b_; | |
1308 | ||
1309 | if (a->mtime < b->mtime) | |
1310 | return -1; | |
1311 | if (a->mtime > b->mtime) | |
1312 | return 1; | |
1313 | return 0; | |
1314 | } | |
1315 | ||
3ce4ca0a DS |
1316 | static int fill_included_packs_all(struct repository *r, |
1317 | struct multi_pack_index *m, | |
ce1e4a10 DS |
1318 | unsigned char *include_pack) |
1319 | { | |
3ce4ca0a DS |
1320 | uint32_t i, count = 0; |
1321 | int pack_kept_objects = 0; | |
1322 | ||
1323 | repo_config_get_bool(r, "repack.packkeptobjects", &pack_kept_objects); | |
1324 | ||
1325 | for (i = 0; i < m->num_packs; i++) { | |
1326 | if (prepare_midx_pack(r, m, i)) | |
1327 | continue; | |
1328 | if (!pack_kept_objects && m->packs[i]->pack_keep) | |
1329 | continue; | |
ce1e4a10 | 1330 | |
ce1e4a10 | 1331 | include_pack[i] = 1; |
3ce4ca0a DS |
1332 | count++; |
1333 | } | |
ce1e4a10 | 1334 | |
3ce4ca0a | 1335 | return count < 2; |
ce1e4a10 DS |
1336 | } |
1337 | ||
1338 | static int fill_included_packs_batch(struct repository *r, | |
1339 | struct multi_pack_index *m, | |
1340 | unsigned char *include_pack, | |
1341 | size_t batch_size) | |
1342 | { | |
1343 | uint32_t i, packs_to_repack; | |
1344 | size_t total_size; | |
1345 | struct repack_info *pack_info = xcalloc(m->num_packs, sizeof(struct repack_info)); | |
3ce4ca0a DS |
1346 | int pack_kept_objects = 0; |
1347 | ||
1348 | repo_config_get_bool(r, "repack.packkeptobjects", &pack_kept_objects); | |
ce1e4a10 DS |
1349 | |
1350 | for (i = 0; i < m->num_packs; i++) { | |
1351 | pack_info[i].pack_int_id = i; | |
1352 | ||
1353 | if (prepare_midx_pack(r, m, i)) | |
1354 | continue; | |
1355 | ||
1356 | pack_info[i].mtime = m->packs[i]->mtime; | |
1357 | } | |
1358 | ||
1359 | for (i = 0; batch_size && i < m->num_objects; i++) { | |
1360 | uint32_t pack_int_id = nth_midxed_pack_int_id(m, i); | |
1361 | pack_info[pack_int_id].referenced_objects++; | |
1362 | } | |
1363 | ||
1364 | QSORT(pack_info, m->num_packs, compare_by_mtime); | |
1365 | ||
1366 | total_size = 0; | |
1367 | packs_to_repack = 0; | |
1368 | for (i = 0; total_size < batch_size && i < m->num_packs; i++) { | |
1369 | int pack_int_id = pack_info[i].pack_int_id; | |
1370 | struct packed_git *p = m->packs[pack_int_id]; | |
1371 | size_t expected_size; | |
1372 | ||
1373 | if (!p) | |
1374 | continue; | |
3ce4ca0a DS |
1375 | if (!pack_kept_objects && p->pack_keep) |
1376 | continue; | |
ce1e4a10 DS |
1377 | if (open_pack_index(p) || !p->num_objects) |
1378 | continue; | |
1379 | ||
1380 | expected_size = (size_t)(p->pack_size | |
1381 | * pack_info[i].referenced_objects); | |
1382 | expected_size /= p->num_objects; | |
1383 | ||
1384 | if (expected_size >= batch_size) | |
1385 | continue; | |
1386 | ||
1387 | packs_to_repack++; | |
1388 | total_size += expected_size; | |
1389 | include_pack[pack_int_id] = 1; | |
1390 | } | |
1391 | ||
1392 | free(pack_info); | |
1393 | ||
1eb22c7d | 1394 | if (packs_to_repack < 2) |
ce1e4a10 DS |
1395 | return 1; |
1396 | ||
2af890bb DS |
1397 | return 0; |
1398 | } | |
ce1e4a10 | 1399 | |
efbc3aee | 1400 | int midx_repack(struct repository *r, const char *object_dir, size_t batch_size, unsigned flags) |
ce1e4a10 DS |
1401 | { |
1402 | int result = 0; | |
1403 | uint32_t i; | |
1404 | unsigned char *include_pack; | |
1405 | struct child_process cmd = CHILD_PROCESS_INIT; | |
6af3b00a | 1406 | FILE *cmd_in; |
ce1e4a10 DS |
1407 | struct strbuf base_name = STRBUF_INIT; |
1408 | struct multi_pack_index *m = load_multi_pack_index(object_dir, 1); | |
1409 | ||
e11d86de SLN |
1410 | /* |
1411 | * When updating the default for these configuration | |
1412 | * variables in builtin/repack.c, these must be adjusted | |
1413 | * to match. | |
1414 | */ | |
1415 | int delta_base_offset = 1; | |
1416 | int use_delta_islands = 0; | |
1417 | ||
ce1e4a10 DS |
1418 | if (!m) |
1419 | return 0; | |
1420 | ||
1421 | include_pack = xcalloc(m->num_packs, sizeof(unsigned char)); | |
1422 | ||
1423 | if (batch_size) { | |
1424 | if (fill_included_packs_batch(r, m, include_pack, batch_size)) | |
1425 | goto cleanup; | |
3ce4ca0a | 1426 | } else if (fill_included_packs_all(r, m, include_pack)) |
ce1e4a10 DS |
1427 | goto cleanup; |
1428 | ||
e11d86de SLN |
1429 | repo_config_get_bool(r, "repack.usedeltabaseoffset", &delta_base_offset); |
1430 | repo_config_get_bool(r, "repack.usedeltaislands", &use_delta_islands); | |
1431 | ||
c972bf4c | 1432 | strvec_push(&cmd.args, "pack-objects"); |
ce1e4a10 DS |
1433 | |
1434 | strbuf_addstr(&base_name, object_dir); | |
1435 | strbuf_addstr(&base_name, "/pack/pack"); | |
c972bf4c | 1436 | strvec_push(&cmd.args, base_name.buf); |
64d80e7d | 1437 | |
e11d86de | 1438 | if (delta_base_offset) |
c972bf4c | 1439 | strvec_push(&cmd.args, "--delta-base-offset"); |
e11d86de | 1440 | if (use_delta_islands) |
c972bf4c | 1441 | strvec_push(&cmd.args, "--delta-islands"); |
e11d86de | 1442 | |
64d80e7d | 1443 | if (flags & MIDX_PROGRESS) |
c972bf4c | 1444 | strvec_push(&cmd.args, "--progress"); |
64d80e7d | 1445 | else |
c972bf4c | 1446 | strvec_push(&cmd.args, "-q"); |
64d80e7d | 1447 | |
ce1e4a10 DS |
1448 | strbuf_release(&base_name); |
1449 | ||
1450 | cmd.git_cmd = 1; | |
1451 | cmd.in = cmd.out = -1; | |
1452 | ||
1453 | if (start_command(&cmd)) { | |
1454 | error(_("could not start pack-objects")); | |
1455 | result = 1; | |
1456 | goto cleanup; | |
1457 | } | |
1458 | ||
6af3b00a RS |
1459 | cmd_in = xfdopen(cmd.in, "w"); |
1460 | ||
ce1e4a10 DS |
1461 | for (i = 0; i < m->num_objects; i++) { |
1462 | struct object_id oid; | |
1463 | uint32_t pack_int_id = nth_midxed_pack_int_id(m, i); | |
1464 | ||
1465 | if (!include_pack[pack_int_id]) | |
1466 | continue; | |
1467 | ||
1468 | nth_midxed_object_oid(&oid, m, i); | |
6af3b00a | 1469 | fprintf(cmd_in, "%s\n", oid_to_hex(&oid)); |
ce1e4a10 | 1470 | } |
6af3b00a | 1471 | fclose(cmd_in); |
ce1e4a10 DS |
1472 | |
1473 | if (finish_command(&cmd)) { | |
1474 | error(_("could not finish pack-objects")); | |
1475 | result = 1; | |
1476 | goto cleanup; | |
1477 | } | |
1478 | ||
840cef0c | 1479 | result = write_midx_internal(object_dir, m, NULL, flags); |
ce1e4a10 DS |
1480 | m = NULL; |
1481 | ||
1482 | cleanup: | |
1483 | if (m) | |
1484 | close_midx(m); | |
1485 | free(include_pack); | |
1486 | return result; | |
1487 | } |