]> git.ipfire.org Git - thirdparty/git.git/blob - midx.c
multi-pack-index: verify missing pack
[thirdparty/git.git] / midx.c
1 #include "cache.h"
2 #include "config.h"
3 #include "csum-file.h"
4 #include "dir.h"
5 #include "lockfile.h"
6 #include "packfile.h"
7 #include "object-store.h"
8 #include "sha1-lookup.h"
9 #include "midx.h"
10
11 #define MIDX_SIGNATURE 0x4d494458 /* "MIDX" */
12 #define MIDX_VERSION 1
13 #define MIDX_BYTE_FILE_VERSION 4
14 #define MIDX_BYTE_HASH_VERSION 5
15 #define MIDX_BYTE_NUM_CHUNKS 6
16 #define MIDX_BYTE_NUM_PACKS 8
17 #define MIDX_HASH_VERSION 1
18 #define MIDX_HEADER_SIZE 12
19 #define MIDX_HASH_LEN 20
20 #define MIDX_MIN_SIZE (MIDX_HEADER_SIZE + MIDX_HASH_LEN)
21
22 #define MIDX_MAX_CHUNKS 5
23 #define MIDX_CHUNK_ALIGNMENT 4
24 #define MIDX_CHUNKID_PACKNAMES 0x504e414d /* "PNAM" */
25 #define MIDX_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
26 #define MIDX_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
27 #define MIDX_CHUNKID_OBJECTOFFSETS 0x4f4f4646 /* "OOFF" */
28 #define MIDX_CHUNKID_LARGEOFFSETS 0x4c4f4646 /* "LOFF" */
29 #define MIDX_CHUNKLOOKUP_WIDTH (sizeof(uint32_t) + sizeof(uint64_t))
30 #define MIDX_CHUNK_FANOUT_SIZE (sizeof(uint32_t) * 256)
31 #define MIDX_CHUNK_OFFSET_WIDTH (2 * sizeof(uint32_t))
32 #define MIDX_CHUNK_LARGE_OFFSET_WIDTH (sizeof(uint64_t))
33 #define MIDX_LARGE_OFFSET_NEEDED 0x80000000
34
35 static char *get_midx_filename(const char *object_dir)
36 {
37 return xstrfmt("%s/pack/multi-pack-index", object_dir);
38 }
39
40 struct multi_pack_index *load_multi_pack_index(const char *object_dir, int local)
41 {
42 struct multi_pack_index *m = NULL;
43 int fd;
44 struct stat st;
45 size_t midx_size;
46 void *midx_map = NULL;
47 uint32_t hash_version;
48 char *midx_name = get_midx_filename(object_dir);
49 uint32_t i;
50 const char *cur_pack_name;
51
52 fd = git_open(midx_name);
53
54 if (fd < 0)
55 goto cleanup_fail;
56 if (fstat(fd, &st)) {
57 error_errno(_("failed to read %s"), midx_name);
58 goto cleanup_fail;
59 }
60
61 midx_size = xsize_t(st.st_size);
62
63 if (midx_size < MIDX_MIN_SIZE) {
64 error(_("multi-pack-index file %s is too small"), midx_name);
65 goto cleanup_fail;
66 }
67
68 FREE_AND_NULL(midx_name);
69
70 midx_map = xmmap(NULL, midx_size, PROT_READ, MAP_PRIVATE, fd, 0);
71
72 FLEX_ALLOC_MEM(m, object_dir, object_dir, strlen(object_dir));
73 m->fd = fd;
74 m->data = midx_map;
75 m->data_len = midx_size;
76 m->local = local;
77
78 m->signature = get_be32(m->data);
79 if (m->signature != MIDX_SIGNATURE)
80 die(_("multi-pack-index signature 0x%08x does not match signature 0x%08x"),
81 m->signature, MIDX_SIGNATURE);
82
83 m->version = m->data[MIDX_BYTE_FILE_VERSION];
84 if (m->version != MIDX_VERSION)
85 die(_("multi-pack-index version %d not recognized"),
86 m->version);
87
88 hash_version = m->data[MIDX_BYTE_HASH_VERSION];
89 if (hash_version != MIDX_HASH_VERSION)
90 die(_("hash version %u does not match"), hash_version);
91 m->hash_len = MIDX_HASH_LEN;
92
93 m->num_chunks = m->data[MIDX_BYTE_NUM_CHUNKS];
94
95 m->num_packs = get_be32(m->data + MIDX_BYTE_NUM_PACKS);
96
97 for (i = 0; i < m->num_chunks; i++) {
98 uint32_t chunk_id = get_be32(m->data + MIDX_HEADER_SIZE +
99 MIDX_CHUNKLOOKUP_WIDTH * i);
100 uint64_t chunk_offset = get_be64(m->data + MIDX_HEADER_SIZE + 4 +
101 MIDX_CHUNKLOOKUP_WIDTH * i);
102
103 if (chunk_offset >= m->data_len)
104 die(_("invalid chunk offset (too large)"));
105
106 switch (chunk_id) {
107 case MIDX_CHUNKID_PACKNAMES:
108 m->chunk_pack_names = m->data + chunk_offset;
109 break;
110
111 case MIDX_CHUNKID_OIDFANOUT:
112 m->chunk_oid_fanout = (uint32_t *)(m->data + chunk_offset);
113 break;
114
115 case MIDX_CHUNKID_OIDLOOKUP:
116 m->chunk_oid_lookup = m->data + chunk_offset;
117 break;
118
119 case MIDX_CHUNKID_OBJECTOFFSETS:
120 m->chunk_object_offsets = m->data + chunk_offset;
121 break;
122
123 case MIDX_CHUNKID_LARGEOFFSETS:
124 m->chunk_large_offsets = m->data + chunk_offset;
125 break;
126
127 case 0:
128 die(_("terminating multi-pack-index chunk id appears earlier than expected"));
129 break;
130
131 default:
132 /*
133 * Do nothing on unrecognized chunks, allowing future
134 * extensions to add optional chunks.
135 */
136 break;
137 }
138 }
139
140 if (!m->chunk_pack_names)
141 die(_("multi-pack-index missing required pack-name chunk"));
142 if (!m->chunk_oid_fanout)
143 die(_("multi-pack-index missing required OID fanout chunk"));
144 if (!m->chunk_oid_lookup)
145 die(_("multi-pack-index missing required OID lookup chunk"));
146 if (!m->chunk_object_offsets)
147 die(_("multi-pack-index missing required object offsets chunk"));
148
149 m->num_objects = ntohl(m->chunk_oid_fanout[255]);
150
151 m->pack_names = xcalloc(m->num_packs, sizeof(*m->pack_names));
152 m->packs = xcalloc(m->num_packs, sizeof(*m->packs));
153
154 cur_pack_name = (const char *)m->chunk_pack_names;
155 for (i = 0; i < m->num_packs; i++) {
156 m->pack_names[i] = cur_pack_name;
157
158 cur_pack_name += strlen(cur_pack_name) + 1;
159
160 if (i && strcmp(m->pack_names[i], m->pack_names[i - 1]) <= 0)
161 die(_("multi-pack-index pack names out of order: '%s' before '%s'"),
162 m->pack_names[i - 1],
163 m->pack_names[i]);
164 }
165
166 return m;
167
168 cleanup_fail:
169 free(m);
170 free(midx_name);
171 if (midx_map)
172 munmap(midx_map, midx_size);
173 if (0 <= fd)
174 close(fd);
175 return NULL;
176 }
177
178 static void close_midx(struct multi_pack_index *m)
179 {
180 uint32_t i;
181 munmap((unsigned char *)m->data, m->data_len);
182 close(m->fd);
183 m->fd = -1;
184
185 for (i = 0; i < m->num_packs; i++) {
186 if (m->packs[i]) {
187 close_pack(m->packs[i]);
188 free(m->packs);
189 }
190 }
191 FREE_AND_NULL(m->packs);
192 FREE_AND_NULL(m->pack_names);
193 }
194
195 int prepare_midx_pack(struct multi_pack_index *m, uint32_t pack_int_id)
196 {
197 struct strbuf pack_name = STRBUF_INIT;
198
199 if (pack_int_id >= m->num_packs)
200 BUG("bad pack-int-id");
201
202 if (m->packs[pack_int_id])
203 return 0;
204
205 strbuf_addf(&pack_name, "%s/pack/%s", m->object_dir,
206 m->pack_names[pack_int_id]);
207
208 m->packs[pack_int_id] = add_packed_git(pack_name.buf, pack_name.len, m->local);
209 strbuf_release(&pack_name);
210 return !m->packs[pack_int_id];
211 }
212
213 int bsearch_midx(const struct object_id *oid, struct multi_pack_index *m, uint32_t *result)
214 {
215 return bsearch_hash(oid->hash, m->chunk_oid_fanout, m->chunk_oid_lookup,
216 MIDX_HASH_LEN, result);
217 }
218
219 struct object_id *nth_midxed_object_oid(struct object_id *oid,
220 struct multi_pack_index *m,
221 uint32_t n)
222 {
223 if (n >= m->num_objects)
224 return NULL;
225
226 hashcpy(oid->hash, m->chunk_oid_lookup + m->hash_len * n);
227 return oid;
228 }
229
230 static off_t nth_midxed_offset(struct multi_pack_index *m, uint32_t pos)
231 {
232 const unsigned char *offset_data;
233 uint32_t offset32;
234
235 offset_data = m->chunk_object_offsets + pos * MIDX_CHUNK_OFFSET_WIDTH;
236 offset32 = get_be32(offset_data + sizeof(uint32_t));
237
238 if (m->chunk_large_offsets && offset32 & MIDX_LARGE_OFFSET_NEEDED) {
239 if (sizeof(offset32) < sizeof(uint64_t))
240 die(_("multi-pack-index stores a 64-bit offset, but off_t is too small"));
241
242 offset32 ^= MIDX_LARGE_OFFSET_NEEDED;
243 return get_be64(m->chunk_large_offsets + sizeof(uint64_t) * offset32);
244 }
245
246 return offset32;
247 }
248
249 static uint32_t nth_midxed_pack_int_id(struct multi_pack_index *m, uint32_t pos)
250 {
251 return get_be32(m->chunk_object_offsets + pos * MIDX_CHUNK_OFFSET_WIDTH);
252 }
253
254 static int nth_midxed_pack_entry(struct multi_pack_index *m, struct pack_entry *e, uint32_t pos)
255 {
256 uint32_t pack_int_id;
257 struct packed_git *p;
258
259 if (pos >= m->num_objects)
260 return 0;
261
262 pack_int_id = nth_midxed_pack_int_id(m, pos);
263
264 if (prepare_midx_pack(m, pack_int_id))
265 die(_("error preparing packfile from multi-pack-index"));
266 p = m->packs[pack_int_id];
267
268 /*
269 * We are about to tell the caller where they can locate the
270 * requested object. We better make sure the packfile is
271 * still here and can be accessed before supplying that
272 * answer, as it may have been deleted since the MIDX was
273 * loaded!
274 */
275 if (!is_pack_valid(p))
276 return 0;
277
278 if (p->num_bad_objects) {
279 uint32_t i;
280 struct object_id oid;
281 nth_midxed_object_oid(&oid, m, pos);
282 for (i = 0; i < p->num_bad_objects; i++)
283 if (!hashcmp(oid.hash,
284 p->bad_object_sha1 + the_hash_algo->rawsz * i))
285 return 0;
286 }
287
288 e->offset = nth_midxed_offset(m, pos);
289 e->p = p;
290
291 return 1;
292 }
293
294 int fill_midx_entry(const struct object_id *oid, struct pack_entry *e, struct multi_pack_index *m)
295 {
296 uint32_t pos;
297
298 if (!bsearch_midx(oid, m, &pos))
299 return 0;
300
301 return nth_midxed_pack_entry(m, e, pos);
302 }
303
304 int midx_contains_pack(struct multi_pack_index *m, const char *idx_name)
305 {
306 uint32_t first = 0, last = m->num_packs;
307
308 while (first < last) {
309 uint32_t mid = first + (last - first) / 2;
310 const char *current;
311 int cmp;
312
313 current = m->pack_names[mid];
314 cmp = strcmp(idx_name, current);
315 if (!cmp)
316 return 1;
317 if (cmp > 0) {
318 first = mid + 1;
319 continue;
320 }
321 last = mid;
322 }
323
324 return 0;
325 }
326
327 int prepare_multi_pack_index_one(struct repository *r, const char *object_dir, int local)
328 {
329 struct multi_pack_index *m;
330 struct multi_pack_index *m_search;
331 int config_value;
332
333 if (repo_config_get_bool(r, "core.multipackindex", &config_value) ||
334 !config_value)
335 return 0;
336
337 for (m_search = r->objects->multi_pack_index; m_search; m_search = m_search->next)
338 if (!strcmp(object_dir, m_search->object_dir))
339 return 1;
340
341 m = load_multi_pack_index(object_dir, local);
342
343 if (m) {
344 m->next = r->objects->multi_pack_index;
345 r->objects->multi_pack_index = m;
346 return 1;
347 }
348
349 return 0;
350 }
351
352 static size_t write_midx_header(struct hashfile *f,
353 unsigned char num_chunks,
354 uint32_t num_packs)
355 {
356 unsigned char byte_values[4];
357
358 hashwrite_be32(f, MIDX_SIGNATURE);
359 byte_values[0] = MIDX_VERSION;
360 byte_values[1] = MIDX_HASH_VERSION;
361 byte_values[2] = num_chunks;
362 byte_values[3] = 0; /* unused */
363 hashwrite(f, byte_values, sizeof(byte_values));
364 hashwrite_be32(f, num_packs);
365
366 return MIDX_HEADER_SIZE;
367 }
368
369 struct pack_list {
370 struct packed_git **list;
371 char **names;
372 uint32_t nr;
373 uint32_t alloc_list;
374 uint32_t alloc_names;
375 size_t pack_name_concat_len;
376 struct multi_pack_index *m;
377 };
378
379 static void add_pack_to_midx(const char *full_path, size_t full_path_len,
380 const char *file_name, void *data)
381 {
382 struct pack_list *packs = (struct pack_list *)data;
383
384 if (ends_with(file_name, ".idx")) {
385 if (packs->m && midx_contains_pack(packs->m, file_name))
386 return;
387
388 ALLOC_GROW(packs->list, packs->nr + 1, packs->alloc_list);
389 ALLOC_GROW(packs->names, packs->nr + 1, packs->alloc_names);
390
391 packs->list[packs->nr] = add_packed_git(full_path,
392 full_path_len,
393 0);
394
395 if (!packs->list[packs->nr]) {
396 warning(_("failed to add packfile '%s'"),
397 full_path);
398 return;
399 }
400
401 if (open_pack_index(packs->list[packs->nr])) {
402 warning(_("failed to open pack-index '%s'"),
403 full_path);
404 close_pack(packs->list[packs->nr]);
405 FREE_AND_NULL(packs->list[packs->nr]);
406 return;
407 }
408
409 packs->names[packs->nr] = xstrdup(file_name);
410 packs->pack_name_concat_len += strlen(file_name) + 1;
411 packs->nr++;
412 }
413 }
414
415 struct pack_pair {
416 uint32_t pack_int_id;
417 char *pack_name;
418 };
419
420 static int pack_pair_compare(const void *_a, const void *_b)
421 {
422 struct pack_pair *a = (struct pack_pair *)_a;
423 struct pack_pair *b = (struct pack_pair *)_b;
424 return strcmp(a->pack_name, b->pack_name);
425 }
426
427 static void sort_packs_by_name(char **pack_names, uint32_t nr_packs, uint32_t *perm)
428 {
429 uint32_t i;
430 struct pack_pair *pairs;
431
432 ALLOC_ARRAY(pairs, nr_packs);
433
434 for (i = 0; i < nr_packs; i++) {
435 pairs[i].pack_int_id = i;
436 pairs[i].pack_name = pack_names[i];
437 }
438
439 QSORT(pairs, nr_packs, pack_pair_compare);
440
441 for (i = 0; i < nr_packs; i++) {
442 pack_names[i] = pairs[i].pack_name;
443 perm[pairs[i].pack_int_id] = i;
444 }
445
446 free(pairs);
447 }
448
449 struct pack_midx_entry {
450 struct object_id oid;
451 uint32_t pack_int_id;
452 time_t pack_mtime;
453 uint64_t offset;
454 };
455
456 static int midx_oid_compare(const void *_a, const void *_b)
457 {
458 const struct pack_midx_entry *a = (const struct pack_midx_entry *)_a;
459 const struct pack_midx_entry *b = (const struct pack_midx_entry *)_b;
460 int cmp = oidcmp(&a->oid, &b->oid);
461
462 if (cmp)
463 return cmp;
464
465 if (a->pack_mtime > b->pack_mtime)
466 return -1;
467 else if (a->pack_mtime < b->pack_mtime)
468 return 1;
469
470 return a->pack_int_id - b->pack_int_id;
471 }
472
473 static int nth_midxed_pack_midx_entry(struct multi_pack_index *m,
474 uint32_t *pack_perm,
475 struct pack_midx_entry *e,
476 uint32_t pos)
477 {
478 if (pos >= m->num_objects)
479 return 1;
480
481 nth_midxed_object_oid(&e->oid, m, pos);
482 e->pack_int_id = pack_perm[nth_midxed_pack_int_id(m, pos)];
483 e->offset = nth_midxed_offset(m, pos);
484
485 /* consider objects in midx to be from "old" packs */
486 e->pack_mtime = 0;
487 return 0;
488 }
489
490 static void fill_pack_entry(uint32_t pack_int_id,
491 struct packed_git *p,
492 uint32_t cur_object,
493 struct pack_midx_entry *entry)
494 {
495 if (!nth_packed_object_oid(&entry->oid, p, cur_object))
496 die(_("failed to locate object %d in packfile"), cur_object);
497
498 entry->pack_int_id = pack_int_id;
499 entry->pack_mtime = p->mtime;
500
501 entry->offset = nth_packed_object_offset(p, cur_object);
502 }
503
504 /*
505 * It is possible to artificially get into a state where there are many
506 * duplicate copies of objects. That can create high memory pressure if
507 * we are to create a list of all objects before de-duplication. To reduce
508 * this memory pressure without a significant performance drop, automatically
509 * group objects by the first byte of their object id. Use the IDX fanout
510 * tables to group the data, copy to a local array, then sort.
511 *
512 * Copy only the de-duplicated entries (selected by most-recent modified time
513 * of a packfile containing the object).
514 */
515 static struct pack_midx_entry *get_sorted_entries(struct multi_pack_index *m,
516 struct packed_git **p,
517 uint32_t *perm,
518 uint32_t nr_packs,
519 uint32_t *nr_objects)
520 {
521 uint32_t cur_fanout, cur_pack, cur_object;
522 uint32_t alloc_fanout, alloc_objects, total_objects = 0;
523 struct pack_midx_entry *entries_by_fanout = NULL;
524 struct pack_midx_entry *deduplicated_entries = NULL;
525 uint32_t start_pack = m ? m->num_packs : 0;
526
527 for (cur_pack = start_pack; cur_pack < nr_packs; cur_pack++)
528 total_objects += p[cur_pack]->num_objects;
529
530 /*
531 * As we de-duplicate by fanout value, we expect the fanout
532 * slices to be evenly distributed, with some noise. Hence,
533 * allocate slightly more than one 256th.
534 */
535 alloc_objects = alloc_fanout = total_objects > 3200 ? total_objects / 200 : 16;
536
537 ALLOC_ARRAY(entries_by_fanout, alloc_fanout);
538 ALLOC_ARRAY(deduplicated_entries, alloc_objects);
539 *nr_objects = 0;
540
541 for (cur_fanout = 0; cur_fanout < 256; cur_fanout++) {
542 uint32_t nr_fanout = 0;
543
544 if (m) {
545 uint32_t start = 0, end;
546
547 if (cur_fanout)
548 start = ntohl(m->chunk_oid_fanout[cur_fanout - 1]);
549 end = ntohl(m->chunk_oid_fanout[cur_fanout]);
550
551 for (cur_object = start; cur_object < end; cur_object++) {
552 ALLOC_GROW(entries_by_fanout, nr_fanout + 1, alloc_fanout);
553 nth_midxed_pack_midx_entry(m, perm,
554 &entries_by_fanout[nr_fanout],
555 cur_object);
556 nr_fanout++;
557 }
558 }
559
560 for (cur_pack = start_pack; cur_pack < nr_packs; cur_pack++) {
561 uint32_t start = 0, end;
562
563 if (cur_fanout)
564 start = get_pack_fanout(p[cur_pack], cur_fanout - 1);
565 end = get_pack_fanout(p[cur_pack], cur_fanout);
566
567 for (cur_object = start; cur_object < end; cur_object++) {
568 ALLOC_GROW(entries_by_fanout, nr_fanout + 1, alloc_fanout);
569 fill_pack_entry(perm[cur_pack], p[cur_pack], cur_object, &entries_by_fanout[nr_fanout]);
570 nr_fanout++;
571 }
572 }
573
574 QSORT(entries_by_fanout, nr_fanout, midx_oid_compare);
575
576 /*
577 * The batch is now sorted by OID and then mtime (descending).
578 * Take only the first duplicate.
579 */
580 for (cur_object = 0; cur_object < nr_fanout; cur_object++) {
581 if (cur_object && !oidcmp(&entries_by_fanout[cur_object - 1].oid,
582 &entries_by_fanout[cur_object].oid))
583 continue;
584
585 ALLOC_GROW(deduplicated_entries, *nr_objects + 1, alloc_objects);
586 memcpy(&deduplicated_entries[*nr_objects],
587 &entries_by_fanout[cur_object],
588 sizeof(struct pack_midx_entry));
589 (*nr_objects)++;
590 }
591 }
592
593 free(entries_by_fanout);
594 return deduplicated_entries;
595 }
596
597 static size_t write_midx_pack_names(struct hashfile *f,
598 char **pack_names,
599 uint32_t num_packs)
600 {
601 uint32_t i;
602 unsigned char padding[MIDX_CHUNK_ALIGNMENT];
603 size_t written = 0;
604
605 for (i = 0; i < num_packs; i++) {
606 size_t writelen = strlen(pack_names[i]) + 1;
607
608 if (i && strcmp(pack_names[i], pack_names[i - 1]) <= 0)
609 BUG("incorrect pack-file order: %s before %s",
610 pack_names[i - 1],
611 pack_names[i]);
612
613 hashwrite(f, pack_names[i], writelen);
614 written += writelen;
615 }
616
617 /* add padding to be aligned */
618 i = MIDX_CHUNK_ALIGNMENT - (written % MIDX_CHUNK_ALIGNMENT);
619 if (i < MIDX_CHUNK_ALIGNMENT) {
620 memset(padding, 0, sizeof(padding));
621 hashwrite(f, padding, i);
622 written += i;
623 }
624
625 return written;
626 }
627
628 static size_t write_midx_oid_fanout(struct hashfile *f,
629 struct pack_midx_entry *objects,
630 uint32_t nr_objects)
631 {
632 struct pack_midx_entry *list = objects;
633 struct pack_midx_entry *last = objects + nr_objects;
634 uint32_t count = 0;
635 uint32_t i;
636
637 /*
638 * Write the first-level table (the list is sorted,
639 * but we use a 256-entry lookup to be able to avoid
640 * having to do eight extra binary search iterations).
641 */
642 for (i = 0; i < 256; i++) {
643 struct pack_midx_entry *next = list;
644
645 while (next < last && next->oid.hash[0] == i) {
646 count++;
647 next++;
648 }
649
650 hashwrite_be32(f, count);
651 list = next;
652 }
653
654 return MIDX_CHUNK_FANOUT_SIZE;
655 }
656
657 static size_t write_midx_oid_lookup(struct hashfile *f, unsigned char hash_len,
658 struct pack_midx_entry *objects,
659 uint32_t nr_objects)
660 {
661 struct pack_midx_entry *list = objects;
662 uint32_t i;
663 size_t written = 0;
664
665 for (i = 0; i < nr_objects; i++) {
666 struct pack_midx_entry *obj = list++;
667
668 if (i < nr_objects - 1) {
669 struct pack_midx_entry *next = list;
670 if (oidcmp(&obj->oid, &next->oid) >= 0)
671 BUG("OIDs not in order: %s >= %s",
672 oid_to_hex(&obj->oid),
673 oid_to_hex(&next->oid));
674 }
675
676 hashwrite(f, obj->oid.hash, (int)hash_len);
677 written += hash_len;
678 }
679
680 return written;
681 }
682
683 static size_t write_midx_object_offsets(struct hashfile *f, int large_offset_needed,
684 struct pack_midx_entry *objects, uint32_t nr_objects)
685 {
686 struct pack_midx_entry *list = objects;
687 uint32_t i, nr_large_offset = 0;
688 size_t written = 0;
689
690 for (i = 0; i < nr_objects; i++) {
691 struct pack_midx_entry *obj = list++;
692
693 hashwrite_be32(f, obj->pack_int_id);
694
695 if (large_offset_needed && obj->offset >> 31)
696 hashwrite_be32(f, MIDX_LARGE_OFFSET_NEEDED | nr_large_offset++);
697 else if (!large_offset_needed && obj->offset >> 32)
698 BUG("object %s requires a large offset (%"PRIx64") but the MIDX is not writing large offsets!",
699 oid_to_hex(&obj->oid),
700 obj->offset);
701 else
702 hashwrite_be32(f, (uint32_t)obj->offset);
703
704 written += MIDX_CHUNK_OFFSET_WIDTH;
705 }
706
707 return written;
708 }
709
710 static size_t write_midx_large_offsets(struct hashfile *f, uint32_t nr_large_offset,
711 struct pack_midx_entry *objects, uint32_t nr_objects)
712 {
713 struct pack_midx_entry *list = objects;
714 size_t written = 0;
715
716 while (nr_large_offset) {
717 struct pack_midx_entry *obj = list++;
718 uint64_t offset = obj->offset;
719
720 if (!(offset >> 31))
721 continue;
722
723 hashwrite_be32(f, offset >> 32);
724 hashwrite_be32(f, offset & 0xffffffffUL);
725 written += 2 * sizeof(uint32_t);
726
727 nr_large_offset--;
728 }
729
730 return written;
731 }
732
733 int write_midx_file(const char *object_dir)
734 {
735 unsigned char cur_chunk, num_chunks = 0;
736 char *midx_name;
737 uint32_t i;
738 struct hashfile *f = NULL;
739 struct lock_file lk;
740 struct pack_list packs;
741 uint32_t *pack_perm = NULL;
742 uint64_t written = 0;
743 uint32_t chunk_ids[MIDX_MAX_CHUNKS + 1];
744 uint64_t chunk_offsets[MIDX_MAX_CHUNKS + 1];
745 uint32_t nr_entries, num_large_offsets = 0;
746 struct pack_midx_entry *entries = NULL;
747 int large_offsets_needed = 0;
748
749 midx_name = get_midx_filename(object_dir);
750 if (safe_create_leading_directories(midx_name)) {
751 UNLEAK(midx_name);
752 die_errno(_("unable to create leading directories of %s"),
753 midx_name);
754 }
755
756 packs.m = load_multi_pack_index(object_dir, 1);
757
758 packs.nr = 0;
759 packs.alloc_list = packs.m ? packs.m->num_packs : 16;
760 packs.alloc_names = packs.alloc_list;
761 packs.list = NULL;
762 packs.names = NULL;
763 packs.pack_name_concat_len = 0;
764 ALLOC_ARRAY(packs.list, packs.alloc_list);
765 ALLOC_ARRAY(packs.names, packs.alloc_names);
766
767 if (packs.m) {
768 for (i = 0; i < packs.m->num_packs; i++) {
769 ALLOC_GROW(packs.list, packs.nr + 1, packs.alloc_list);
770 ALLOC_GROW(packs.names, packs.nr + 1, packs.alloc_names);
771
772 packs.list[packs.nr] = NULL;
773 packs.names[packs.nr] = xstrdup(packs.m->pack_names[i]);
774 packs.pack_name_concat_len += strlen(packs.names[packs.nr]) + 1;
775 packs.nr++;
776 }
777 }
778
779 for_each_file_in_pack_dir(object_dir, add_pack_to_midx, &packs);
780
781 if (packs.m && packs.nr == packs.m->num_packs)
782 goto cleanup;
783
784 if (packs.pack_name_concat_len % MIDX_CHUNK_ALIGNMENT)
785 packs.pack_name_concat_len += MIDX_CHUNK_ALIGNMENT -
786 (packs.pack_name_concat_len % MIDX_CHUNK_ALIGNMENT);
787
788 ALLOC_ARRAY(pack_perm, packs.nr);
789 sort_packs_by_name(packs.names, packs.nr, pack_perm);
790
791 entries = get_sorted_entries(packs.m, packs.list, pack_perm, packs.nr, &nr_entries);
792
793 for (i = 0; i < nr_entries; i++) {
794 if (entries[i].offset > 0x7fffffff)
795 num_large_offsets++;
796 if (entries[i].offset > 0xffffffff)
797 large_offsets_needed = 1;
798 }
799
800 hold_lock_file_for_update(&lk, midx_name, LOCK_DIE_ON_ERROR);
801 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
802 FREE_AND_NULL(midx_name);
803
804 if (packs.m)
805 close_midx(packs.m);
806
807 cur_chunk = 0;
808 num_chunks = large_offsets_needed ? 5 : 4;
809
810 written = write_midx_header(f, num_chunks, packs.nr);
811
812 chunk_ids[cur_chunk] = MIDX_CHUNKID_PACKNAMES;
813 chunk_offsets[cur_chunk] = written + (num_chunks + 1) * MIDX_CHUNKLOOKUP_WIDTH;
814
815 cur_chunk++;
816 chunk_ids[cur_chunk] = MIDX_CHUNKID_OIDFANOUT;
817 chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + packs.pack_name_concat_len;
818
819 cur_chunk++;
820 chunk_ids[cur_chunk] = MIDX_CHUNKID_OIDLOOKUP;
821 chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + MIDX_CHUNK_FANOUT_SIZE;
822
823 cur_chunk++;
824 chunk_ids[cur_chunk] = MIDX_CHUNKID_OBJECTOFFSETS;
825 chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + nr_entries * MIDX_HASH_LEN;
826
827 cur_chunk++;
828 chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + nr_entries * MIDX_CHUNK_OFFSET_WIDTH;
829 if (large_offsets_needed) {
830 chunk_ids[cur_chunk] = MIDX_CHUNKID_LARGEOFFSETS;
831
832 cur_chunk++;
833 chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] +
834 num_large_offsets * MIDX_CHUNK_LARGE_OFFSET_WIDTH;
835 }
836
837 chunk_ids[cur_chunk] = 0;
838
839 for (i = 0; i <= num_chunks; i++) {
840 if (i && chunk_offsets[i] < chunk_offsets[i - 1])
841 BUG("incorrect chunk offsets: %"PRIu64" before %"PRIu64,
842 chunk_offsets[i - 1],
843 chunk_offsets[i]);
844
845 if (chunk_offsets[i] % MIDX_CHUNK_ALIGNMENT)
846 BUG("chunk offset %"PRIu64" is not properly aligned",
847 chunk_offsets[i]);
848
849 hashwrite_be32(f, chunk_ids[i]);
850 hashwrite_be32(f, chunk_offsets[i] >> 32);
851 hashwrite_be32(f, chunk_offsets[i]);
852
853 written += MIDX_CHUNKLOOKUP_WIDTH;
854 }
855
856 for (i = 0; i < num_chunks; i++) {
857 if (written != chunk_offsets[i])
858 BUG("incorrect chunk offset (%"PRIu64" != %"PRIu64") for chunk id %"PRIx32,
859 chunk_offsets[i],
860 written,
861 chunk_ids[i]);
862
863 switch (chunk_ids[i]) {
864 case MIDX_CHUNKID_PACKNAMES:
865 written += write_midx_pack_names(f, packs.names, packs.nr);
866 break;
867
868 case MIDX_CHUNKID_OIDFANOUT:
869 written += write_midx_oid_fanout(f, entries, nr_entries);
870 break;
871
872 case MIDX_CHUNKID_OIDLOOKUP:
873 written += write_midx_oid_lookup(f, MIDX_HASH_LEN, entries, nr_entries);
874 break;
875
876 case MIDX_CHUNKID_OBJECTOFFSETS:
877 written += write_midx_object_offsets(f, large_offsets_needed, entries, nr_entries);
878 break;
879
880 case MIDX_CHUNKID_LARGEOFFSETS:
881 written += write_midx_large_offsets(f, num_large_offsets, entries, nr_entries);
882 break;
883
884 default:
885 BUG("trying to write unknown chunk id %"PRIx32,
886 chunk_ids[i]);
887 }
888 }
889
890 if (written != chunk_offsets[num_chunks])
891 BUG("incorrect final offset %"PRIu64" != %"PRIu64,
892 written,
893 chunk_offsets[num_chunks]);
894
895 finalize_hashfile(f, NULL, CSUM_FSYNC | CSUM_HASH_IN_STREAM);
896 commit_lock_file(&lk);
897
898 cleanup:
899 for (i = 0; i < packs.nr; i++) {
900 if (packs.list[i]) {
901 close_pack(packs.list[i]);
902 free(packs.list[i]);
903 }
904 free(packs.names[i]);
905 }
906
907 free(packs.list);
908 free(packs.names);
909 free(entries);
910 free(pack_perm);
911 free(midx_name);
912 return 0;
913 }
914
915 void clear_midx_file(const char *object_dir)
916 {
917 char *midx = get_midx_filename(object_dir);
918
919 if (remove_path(midx)) {
920 UNLEAK(midx);
921 die(_("failed to clear multi-pack-index at %s"), midx);
922 }
923
924 free(midx);
925 }
926
927 static int verify_midx_error;
928
929 static void midx_report(const char *fmt, ...)
930 {
931 va_list ap;
932 verify_midx_error = 1;
933 va_start(ap, fmt);
934 vfprintf(stderr, fmt, ap);
935 fprintf(stderr, "\n");
936 va_end(ap);
937 }
938
939 int verify_midx_file(const char *object_dir)
940 {
941 uint32_t i;
942 struct multi_pack_index *m = load_multi_pack_index(object_dir, 1);
943 verify_midx_error = 0;
944
945 if (!m)
946 return 0;
947
948 for (i = 0; i < m->num_packs; i++) {
949 if (prepare_midx_pack(m, i))
950 midx_report("failed to load pack in position %d", i);
951 }
952
953 return verify_midx_error;
954 }