commit-graph: close_commit_graph before shallow walk
[git/git.git] / commit-graph.c
1 #include "cache.h"
2 #include "config.h"
3 #include "dir.h"
4 #include "git-compat-util.h"
5 #include "lockfile.h"
6 #include "pack.h"
7 #include "packfile.h"
8 #include "commit.h"
9 #include "object.h"
10 #include "refs.h"
11 #include "revision.h"
12 #include "sha1-lookup.h"
13 #include "commit-graph.h"
14 #include "object-store.h"
15 #include "alloc.h"
16 #include "hashmap.h"
17 #include "replace-object.h"
18
19 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
20 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
21 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
22 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
23 #define GRAPH_CHUNKID_LARGEEDGES 0x45444745 /* "EDGE" */
24
25 #define GRAPH_DATA_WIDTH 36
26
27 #define GRAPH_VERSION_1 0x1
28 #define GRAPH_VERSION GRAPH_VERSION_1
29
30 #define GRAPH_OID_VERSION_SHA1 1
31 #define GRAPH_OID_LEN_SHA1 GIT_SHA1_RAWSZ
32 #define GRAPH_OID_VERSION GRAPH_OID_VERSION_SHA1
33 #define GRAPH_OID_LEN GRAPH_OID_LEN_SHA1
34
35 #define GRAPH_OCTOPUS_EDGES_NEEDED 0x80000000
36 #define GRAPH_PARENT_MISSING 0x7fffffff
37 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
38 #define GRAPH_PARENT_NONE 0x70000000
39
40 #define GRAPH_LAST_EDGE 0x80000000
41
42 #define GRAPH_HEADER_SIZE 8
43 #define GRAPH_FANOUT_SIZE (4 * 256)
44 #define GRAPH_CHUNKLOOKUP_WIDTH 12
45 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
46 + GRAPH_FANOUT_SIZE + GRAPH_OID_LEN)
47
48 char *get_commit_graph_filename(const char *obj_dir)
49 {
50 return xstrfmt("%s/info/commit-graph", obj_dir);
51 }
52
53 static struct commit_graph *alloc_commit_graph(void)
54 {
55 struct commit_graph *g = xcalloc(1, sizeof(*g));
56 g->graph_fd = -1;
57
58 return g;
59 }
60
61 extern int read_replace_refs;
62
63 static int commit_graph_compatible(struct repository *r)
64 {
65 if (!r->gitdir)
66 return 0;
67
68 if (read_replace_refs) {
69 prepare_replace_object(r);
70 if (hashmap_get_size(&r->objects->replace_map->map))
71 return 0;
72 }
73
74 prepare_commit_graft(r);
75 if (r->parsed_objects && r->parsed_objects->grafts_nr)
76 return 0;
77 if (is_repository_shallow(r))
78 return 0;
79
80 return 1;
81 }
82
83 struct commit_graph *load_commit_graph_one(const char *graph_file)
84 {
85 void *graph_map;
86 const unsigned char *data, *chunk_lookup;
87 size_t graph_size;
88 struct stat st;
89 uint32_t i;
90 struct commit_graph *graph;
91 int fd = git_open(graph_file);
92 uint64_t last_chunk_offset;
93 uint32_t last_chunk_id;
94 uint32_t graph_signature;
95 unsigned char graph_version, hash_version;
96
97 if (fd < 0)
98 return NULL;
99 if (fstat(fd, &st)) {
100 close(fd);
101 return NULL;
102 }
103 graph_size = xsize_t(st.st_size);
104
105 if (graph_size < GRAPH_MIN_SIZE) {
106 close(fd);
107 die("graph file %s is too small", graph_file);
108 }
109 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
110 data = (const unsigned char *)graph_map;
111
112 graph_signature = get_be32(data);
113 if (graph_signature != GRAPH_SIGNATURE) {
114 error("graph signature %X does not match signature %X",
115 graph_signature, GRAPH_SIGNATURE);
116 goto cleanup_fail;
117 }
118
119 graph_version = *(unsigned char*)(data + 4);
120 if (graph_version != GRAPH_VERSION) {
121 error("graph version %X does not match version %X",
122 graph_version, GRAPH_VERSION);
123 goto cleanup_fail;
124 }
125
126 hash_version = *(unsigned char*)(data + 5);
127 if (hash_version != GRAPH_OID_VERSION) {
128 error("hash version %X does not match version %X",
129 hash_version, GRAPH_OID_VERSION);
130 goto cleanup_fail;
131 }
132
133 graph = alloc_commit_graph();
134
135 graph->hash_len = GRAPH_OID_LEN;
136 graph->num_chunks = *(unsigned char*)(data + 6);
137 graph->graph_fd = fd;
138 graph->data = graph_map;
139 graph->data_len = graph_size;
140
141 last_chunk_id = 0;
142 last_chunk_offset = 8;
143 chunk_lookup = data + 8;
144 for (i = 0; i < graph->num_chunks; i++) {
145 uint32_t chunk_id = get_be32(chunk_lookup + 0);
146 uint64_t chunk_offset = get_be64(chunk_lookup + 4);
147 int chunk_repeated = 0;
148
149 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
150
151 if (chunk_offset > graph_size - GIT_MAX_RAWSZ) {
152 error("improper chunk offset %08x%08x", (uint32_t)(chunk_offset >> 32),
153 (uint32_t)chunk_offset);
154 goto cleanup_fail;
155 }
156
157 switch (chunk_id) {
158 case GRAPH_CHUNKID_OIDFANOUT:
159 if (graph->chunk_oid_fanout)
160 chunk_repeated = 1;
161 else
162 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
163 break;
164
165 case GRAPH_CHUNKID_OIDLOOKUP:
166 if (graph->chunk_oid_lookup)
167 chunk_repeated = 1;
168 else
169 graph->chunk_oid_lookup = data + chunk_offset;
170 break;
171
172 case GRAPH_CHUNKID_DATA:
173 if (graph->chunk_commit_data)
174 chunk_repeated = 1;
175 else
176 graph->chunk_commit_data = data + chunk_offset;
177 break;
178
179 case GRAPH_CHUNKID_LARGEEDGES:
180 if (graph->chunk_large_edges)
181 chunk_repeated = 1;
182 else
183 graph->chunk_large_edges = data + chunk_offset;
184 break;
185 }
186
187 if (chunk_repeated) {
188 error("chunk id %08x appears multiple times", chunk_id);
189 goto cleanup_fail;
190 }
191
192 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
193 {
194 graph->num_commits = (chunk_offset - last_chunk_offset)
195 / graph->hash_len;
196 }
197
198 last_chunk_id = chunk_id;
199 last_chunk_offset = chunk_offset;
200 }
201
202 return graph;
203
204 cleanup_fail:
205 munmap(graph_map, graph_size);
206 close(fd);
207 exit(1);
208 }
209
210 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
211 {
212 char *graph_name;
213
214 if (r->objects->commit_graph)
215 return;
216
217 graph_name = get_commit_graph_filename(obj_dir);
218 r->objects->commit_graph =
219 load_commit_graph_one(graph_name);
220
221 FREE_AND_NULL(graph_name);
222 }
223
224 /*
225 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
226 *
227 * On the first invocation, this function attemps to load the commit
228 * graph if the_repository is configured to have one.
229 */
230 static int prepare_commit_graph(struct repository *r)
231 {
232 struct alternate_object_database *alt;
233 char *obj_dir;
234 int config_value;
235
236 if (r->objects->commit_graph_attempted)
237 return !!r->objects->commit_graph;
238 r->objects->commit_graph_attempted = 1;
239
240 if (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
241 !config_value)
242 /*
243 * This repository is not configured to use commit graphs, so
244 * do not load one. (But report commit_graph_attempted anyway
245 * so that commit graph loading is not attempted again for this
246 * repository.)
247 */
248 return 0;
249
250 if (!commit_graph_compatible(r))
251 return 0;
252
253 obj_dir = r->objects->objectdir;
254 prepare_commit_graph_one(r, obj_dir);
255 prepare_alt_odb(r);
256 for (alt = r->objects->alt_odb_list;
257 !r->objects->commit_graph && alt;
258 alt = alt->next)
259 prepare_commit_graph_one(r, alt->path);
260 return !!r->objects->commit_graph;
261 }
262
263 void close_commit_graph(struct repository *r)
264 {
265 free_commit_graph(r->objects->commit_graph);
266 r->objects->commit_graph = NULL;
267 }
268
269 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
270 {
271 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
272 g->chunk_oid_lookup, g->hash_len, pos);
273 }
274
275 static struct commit_list **insert_parent_or_die(struct commit_graph *g,
276 uint64_t pos,
277 struct commit_list **pptr)
278 {
279 struct commit *c;
280 struct object_id oid;
281
282 if (pos >= g->num_commits)
283 die("invalid parent position %"PRIu64, pos);
284
285 hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
286 c = lookup_commit(the_repository, &oid);
287 if (!c)
288 die("could not find commit %s", oid_to_hex(&oid));
289 c->graph_pos = pos;
290 return &commit_list_insert(c, pptr)->next;
291 }
292
293 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
294 {
295 const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
296 item->graph_pos = pos;
297 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
298 }
299
300 static int fill_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t pos)
301 {
302 uint32_t edge_value;
303 uint32_t *parent_data_ptr;
304 uint64_t date_low, date_high;
305 struct commit_list **pptr;
306 const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
307
308 item->object.parsed = 1;
309 item->graph_pos = pos;
310
311 item->maybe_tree = NULL;
312
313 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
314 date_low = get_be32(commit_data + g->hash_len + 12);
315 item->date = (timestamp_t)((date_high << 32) | date_low);
316
317 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
318
319 pptr = &item->parents;
320
321 edge_value = get_be32(commit_data + g->hash_len);
322 if (edge_value == GRAPH_PARENT_NONE)
323 return 1;
324 pptr = insert_parent_or_die(g, edge_value, pptr);
325
326 edge_value = get_be32(commit_data + g->hash_len + 4);
327 if (edge_value == GRAPH_PARENT_NONE)
328 return 1;
329 if (!(edge_value & GRAPH_OCTOPUS_EDGES_NEEDED)) {
330 pptr = insert_parent_or_die(g, edge_value, pptr);
331 return 1;
332 }
333
334 parent_data_ptr = (uint32_t*)(g->chunk_large_edges +
335 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
336 do {
337 edge_value = get_be32(parent_data_ptr);
338 pptr = insert_parent_or_die(g,
339 edge_value & GRAPH_EDGE_LAST_MASK,
340 pptr);
341 parent_data_ptr++;
342 } while (!(edge_value & GRAPH_LAST_EDGE));
343
344 return 1;
345 }
346
347 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
348 {
349 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
350 *pos = item->graph_pos;
351 return 1;
352 } else {
353 return bsearch_graph(g, &(item->object.oid), pos);
354 }
355 }
356
357 static int parse_commit_in_graph_one(struct commit_graph *g, struct commit *item)
358 {
359 uint32_t pos;
360
361 if (item->object.parsed)
362 return 1;
363
364 if (find_commit_in_graph(item, g, &pos))
365 return fill_commit_in_graph(item, g, pos);
366
367 return 0;
368 }
369
370 int parse_commit_in_graph(struct repository *r, struct commit *item)
371 {
372 if (!prepare_commit_graph(r))
373 return 0;
374 return parse_commit_in_graph_one(r->objects->commit_graph, item);
375 }
376
377 void load_commit_graph_info(struct repository *r, struct commit *item)
378 {
379 uint32_t pos;
380 if (!prepare_commit_graph(r))
381 return;
382 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
383 fill_commit_graph_info(item, r->objects->commit_graph, pos);
384 }
385
386 static struct tree *load_tree_for_commit(struct commit_graph *g, struct commit *c)
387 {
388 struct object_id oid;
389 const unsigned char *commit_data = g->chunk_commit_data +
390 GRAPH_DATA_WIDTH * (c->graph_pos);
391
392 hashcpy(oid.hash, commit_data);
393 c->maybe_tree = lookup_tree(the_repository, &oid);
394
395 return c->maybe_tree;
396 }
397
398 static struct tree *get_commit_tree_in_graph_one(struct commit_graph *g,
399 const struct commit *c)
400 {
401 if (c->maybe_tree)
402 return c->maybe_tree;
403 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
404 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
405
406 return load_tree_for_commit(g, (struct commit *)c);
407 }
408
409 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
410 {
411 return get_commit_tree_in_graph_one(r->objects->commit_graph, c);
412 }
413
414 static void write_graph_chunk_fanout(struct hashfile *f,
415 struct commit **commits,
416 int nr_commits)
417 {
418 int i, count = 0;
419 struct commit **list = commits;
420
421 /*
422 * Write the first-level table (the list is sorted,
423 * but we use a 256-entry lookup to be able to avoid
424 * having to do eight extra binary search iterations).
425 */
426 for (i = 0; i < 256; i++) {
427 while (count < nr_commits) {
428 if ((*list)->object.oid.hash[0] != i)
429 break;
430 count++;
431 list++;
432 }
433
434 hashwrite_be32(f, count);
435 }
436 }
437
438 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
439 struct commit **commits, int nr_commits)
440 {
441 struct commit **list = commits;
442 int count;
443 for (count = 0; count < nr_commits; count++, list++)
444 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
445 }
446
447 static const unsigned char *commit_to_sha1(size_t index, void *table)
448 {
449 struct commit **commits = table;
450 return commits[index]->object.oid.hash;
451 }
452
453 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
454 struct commit **commits, int nr_commits)
455 {
456 struct commit **list = commits;
457 struct commit **last = commits + nr_commits;
458 uint32_t num_extra_edges = 0;
459
460 while (list < last) {
461 struct commit_list *parent;
462 int edge_value;
463 uint32_t packedDate[2];
464
465 parse_commit(*list);
466 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
467
468 parent = (*list)->parents;
469
470 if (!parent)
471 edge_value = GRAPH_PARENT_NONE;
472 else {
473 edge_value = sha1_pos(parent->item->object.oid.hash,
474 commits,
475 nr_commits,
476 commit_to_sha1);
477
478 if (edge_value < 0)
479 edge_value = GRAPH_PARENT_MISSING;
480 }
481
482 hashwrite_be32(f, edge_value);
483
484 if (parent)
485 parent = parent->next;
486
487 if (!parent)
488 edge_value = GRAPH_PARENT_NONE;
489 else if (parent->next)
490 edge_value = GRAPH_OCTOPUS_EDGES_NEEDED | num_extra_edges;
491 else {
492 edge_value = sha1_pos(parent->item->object.oid.hash,
493 commits,
494 nr_commits,
495 commit_to_sha1);
496 if (edge_value < 0)
497 edge_value = GRAPH_PARENT_MISSING;
498 }
499
500 hashwrite_be32(f, edge_value);
501
502 if (edge_value & GRAPH_OCTOPUS_EDGES_NEEDED) {
503 do {
504 num_extra_edges++;
505 parent = parent->next;
506 } while (parent);
507 }
508
509 if (sizeof((*list)->date) > 4)
510 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
511 else
512 packedDate[0] = 0;
513
514 packedDate[0] |= htonl((*list)->generation << 2);
515
516 packedDate[1] = htonl((*list)->date);
517 hashwrite(f, packedDate, 8);
518
519 list++;
520 }
521 }
522
523 static void write_graph_chunk_large_edges(struct hashfile *f,
524 struct commit **commits,
525 int nr_commits)
526 {
527 struct commit **list = commits;
528 struct commit **last = commits + nr_commits;
529 struct commit_list *parent;
530
531 while (list < last) {
532 int num_parents = 0;
533 for (parent = (*list)->parents; num_parents < 3 && parent;
534 parent = parent->next)
535 num_parents++;
536
537 if (num_parents <= 2) {
538 list++;
539 continue;
540 }
541
542 /* Since num_parents > 2, this initializer is safe. */
543 for (parent = (*list)->parents->next; parent; parent = parent->next) {
544 int edge_value = sha1_pos(parent->item->object.oid.hash,
545 commits,
546 nr_commits,
547 commit_to_sha1);
548
549 if (edge_value < 0)
550 edge_value = GRAPH_PARENT_MISSING;
551 else if (!parent->next)
552 edge_value |= GRAPH_LAST_EDGE;
553
554 hashwrite_be32(f, edge_value);
555 }
556
557 list++;
558 }
559 }
560
561 static int commit_compare(const void *_a, const void *_b)
562 {
563 const struct object_id *a = (const struct object_id *)_a;
564 const struct object_id *b = (const struct object_id *)_b;
565 return oidcmp(a, b);
566 }
567
568 struct packed_commit_list {
569 struct commit **list;
570 int nr;
571 int alloc;
572 };
573
574 struct packed_oid_list {
575 struct object_id *list;
576 int nr;
577 int alloc;
578 };
579
580 static int add_packed_commits(const struct object_id *oid,
581 struct packed_git *pack,
582 uint32_t pos,
583 void *data)
584 {
585 struct packed_oid_list *list = (struct packed_oid_list*)data;
586 enum object_type type;
587 off_t offset = nth_packed_object_offset(pack, pos);
588 struct object_info oi = OBJECT_INFO_INIT;
589
590 oi.typep = &type;
591 if (packed_object_info(the_repository, pack, offset, &oi) < 0)
592 die("unable to get type of object %s", oid_to_hex(oid));
593
594 if (type != OBJ_COMMIT)
595 return 0;
596
597 ALLOC_GROW(list->list, list->nr + 1, list->alloc);
598 oidcpy(&(list->list[list->nr]), oid);
599 list->nr++;
600
601 return 0;
602 }
603
604 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
605 {
606 struct commit_list *parent;
607 for (parent = commit->parents; parent; parent = parent->next) {
608 if (!(parent->item->object.flags & UNINTERESTING)) {
609 ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
610 oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
611 oids->nr++;
612 parent->item->object.flags |= UNINTERESTING;
613 }
614 }
615 }
616
617 static void close_reachable(struct packed_oid_list *oids)
618 {
619 int i;
620 struct commit *commit;
621
622 for (i = 0; i < oids->nr; i++) {
623 commit = lookup_commit(the_repository, &oids->list[i]);
624 if (commit)
625 commit->object.flags |= UNINTERESTING;
626 }
627
628 /*
629 * As this loop runs, oids->nr may grow, but not more
630 * than the number of missing commits in the reachable
631 * closure.
632 */
633 for (i = 0; i < oids->nr; i++) {
634 commit = lookup_commit(the_repository, &oids->list[i]);
635
636 if (commit && !parse_commit(commit))
637 add_missing_parents(oids, commit);
638 }
639
640 for (i = 0; i < oids->nr; i++) {
641 commit = lookup_commit(the_repository, &oids->list[i]);
642
643 if (commit)
644 commit->object.flags &= ~UNINTERESTING;
645 }
646 }
647
648 static void compute_generation_numbers(struct packed_commit_list* commits)
649 {
650 int i;
651 struct commit_list *list = NULL;
652
653 for (i = 0; i < commits->nr; i++) {
654 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
655 commits->list[i]->generation != GENERATION_NUMBER_ZERO)
656 continue;
657
658 commit_list_insert(commits->list[i], &list);
659 while (list) {
660 struct commit *current = list->item;
661 struct commit_list *parent;
662 int all_parents_computed = 1;
663 uint32_t max_generation = 0;
664
665 for (parent = current->parents; parent; parent = parent->next) {
666 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
667 parent->item->generation == GENERATION_NUMBER_ZERO) {
668 all_parents_computed = 0;
669 commit_list_insert(parent->item, &list);
670 break;
671 } else if (parent->item->generation > max_generation) {
672 max_generation = parent->item->generation;
673 }
674 }
675
676 if (all_parents_computed) {
677 current->generation = max_generation + 1;
678 pop_commit(&list);
679
680 if (current->generation > GENERATION_NUMBER_MAX)
681 current->generation = GENERATION_NUMBER_MAX;
682 }
683 }
684 }
685 }
686
687 static int add_ref_to_list(const char *refname,
688 const struct object_id *oid,
689 int flags, void *cb_data)
690 {
691 struct string_list *list = (struct string_list *)cb_data;
692
693 string_list_append(list, oid_to_hex(oid));
694 return 0;
695 }
696
697 void write_commit_graph_reachable(const char *obj_dir, int append)
698 {
699 struct string_list list;
700
701 string_list_init(&list, 1);
702 for_each_ref(add_ref_to_list, &list);
703 write_commit_graph(obj_dir, NULL, &list, append);
704 }
705
706 void write_commit_graph(const char *obj_dir,
707 struct string_list *pack_indexes,
708 struct string_list *commit_hex,
709 int append)
710 {
711 struct packed_oid_list oids;
712 struct packed_commit_list commits;
713 struct hashfile *f;
714 uint32_t i, count_distinct = 0;
715 char *graph_name;
716 struct lock_file lk = LOCK_INIT;
717 uint32_t chunk_ids[5];
718 uint64_t chunk_offsets[5];
719 int num_chunks;
720 int num_extra_edges;
721 struct commit_list *parent;
722
723 if (!commit_graph_compatible(the_repository))
724 return;
725
726 oids.nr = 0;
727 oids.alloc = approximate_object_count() / 4;
728
729 if (append) {
730 prepare_commit_graph_one(the_repository, obj_dir);
731 if (the_repository->objects->commit_graph)
732 oids.alloc += the_repository->objects->commit_graph->num_commits;
733 }
734
735 if (oids.alloc < 1024)
736 oids.alloc = 1024;
737 ALLOC_ARRAY(oids.list, oids.alloc);
738
739 if (append && the_repository->objects->commit_graph) {
740 struct commit_graph *commit_graph =
741 the_repository->objects->commit_graph;
742 for (i = 0; i < commit_graph->num_commits; i++) {
743 const unsigned char *hash = commit_graph->chunk_oid_lookup +
744 commit_graph->hash_len * i;
745 hashcpy(oids.list[oids.nr++].hash, hash);
746 }
747 }
748
749 if (pack_indexes) {
750 struct strbuf packname = STRBUF_INIT;
751 int dirlen;
752 strbuf_addf(&packname, "%s/pack/", obj_dir);
753 dirlen = packname.len;
754 for (i = 0; i < pack_indexes->nr; i++) {
755 struct packed_git *p;
756 strbuf_setlen(&packname, dirlen);
757 strbuf_addstr(&packname, pack_indexes->items[i].string);
758 p = add_packed_git(packname.buf, packname.len, 1);
759 if (!p)
760 die("error adding pack %s", packname.buf);
761 if (open_pack_index(p))
762 die("error opening index for %s", packname.buf);
763 for_each_object_in_pack(p, add_packed_commits, &oids);
764 close_pack(p);
765 }
766 strbuf_release(&packname);
767 }
768
769 if (commit_hex) {
770 for (i = 0; i < commit_hex->nr; i++) {
771 const char *end;
772 struct object_id oid;
773 struct commit *result;
774
775 if (commit_hex->items[i].string &&
776 parse_oid_hex(commit_hex->items[i].string, &oid, &end))
777 continue;
778
779 result = lookup_commit_reference_gently(the_repository, &oid, 1);
780
781 if (result) {
782 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
783 oidcpy(&oids.list[oids.nr], &(result->object.oid));
784 oids.nr++;
785 }
786 }
787 }
788
789 if (!pack_indexes && !commit_hex)
790 for_each_packed_object(add_packed_commits, &oids, 0);
791
792 close_reachable(&oids);
793
794 QSORT(oids.list, oids.nr, commit_compare);
795
796 count_distinct = 1;
797 for (i = 1; i < oids.nr; i++) {
798 if (oidcmp(&oids.list[i-1], &oids.list[i]))
799 count_distinct++;
800 }
801
802 if (count_distinct >= GRAPH_PARENT_MISSING)
803 die(_("the commit graph format cannot write %d commits"), count_distinct);
804
805 commits.nr = 0;
806 commits.alloc = count_distinct;
807 ALLOC_ARRAY(commits.list, commits.alloc);
808
809 num_extra_edges = 0;
810 for (i = 0; i < oids.nr; i++) {
811 int num_parents = 0;
812 if (i > 0 && !oidcmp(&oids.list[i-1], &oids.list[i]))
813 continue;
814
815 commits.list[commits.nr] = lookup_commit(the_repository, &oids.list[i]);
816 parse_commit(commits.list[commits.nr]);
817
818 for (parent = commits.list[commits.nr]->parents;
819 parent; parent = parent->next)
820 num_parents++;
821
822 if (num_parents > 2)
823 num_extra_edges += num_parents - 1;
824
825 commits.nr++;
826 }
827 num_chunks = num_extra_edges ? 4 : 3;
828
829 if (commits.nr >= GRAPH_PARENT_MISSING)
830 die(_("too many commits to write graph"));
831
832 compute_generation_numbers(&commits);
833
834 graph_name = get_commit_graph_filename(obj_dir);
835 if (safe_create_leading_directories(graph_name))
836 die_errno(_("unable to create leading directories of %s"),
837 graph_name);
838
839 hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
840 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
841
842 hashwrite_be32(f, GRAPH_SIGNATURE);
843
844 hashwrite_u8(f, GRAPH_VERSION);
845 hashwrite_u8(f, GRAPH_OID_VERSION);
846 hashwrite_u8(f, num_chunks);
847 hashwrite_u8(f, 0); /* unused padding byte */
848
849 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
850 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
851 chunk_ids[2] = GRAPH_CHUNKID_DATA;
852 if (num_extra_edges)
853 chunk_ids[3] = GRAPH_CHUNKID_LARGEEDGES;
854 else
855 chunk_ids[3] = 0;
856 chunk_ids[4] = 0;
857
858 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
859 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
860 chunk_offsets[2] = chunk_offsets[1] + GRAPH_OID_LEN * commits.nr;
861 chunk_offsets[3] = chunk_offsets[2] + (GRAPH_OID_LEN + 16) * commits.nr;
862 chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
863
864 for (i = 0; i <= num_chunks; i++) {
865 uint32_t chunk_write[3];
866
867 chunk_write[0] = htonl(chunk_ids[i]);
868 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
869 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
870 hashwrite(f, chunk_write, 12);
871 }
872
873 write_graph_chunk_fanout(f, commits.list, commits.nr);
874 write_graph_chunk_oids(f, GRAPH_OID_LEN, commits.list, commits.nr);
875 write_graph_chunk_data(f, GRAPH_OID_LEN, commits.list, commits.nr);
876 write_graph_chunk_large_edges(f, commits.list, commits.nr);
877
878 close_commit_graph(the_repository);
879 finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
880 commit_lock_file(&lk);
881
882 free(oids.list);
883 oids.alloc = 0;
884 oids.nr = 0;
885 }
886
887 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
888 static int verify_commit_graph_error;
889
890 static void graph_report(const char *fmt, ...)
891 {
892 va_list ap;
893
894 verify_commit_graph_error = 1;
895 va_start(ap, fmt);
896 vfprintf(stderr, fmt, ap);
897 fprintf(stderr, "\n");
898 va_end(ap);
899 }
900
901 #define GENERATION_ZERO_EXISTS 1
902 #define GENERATION_NUMBER_EXISTS 2
903
904 int verify_commit_graph(struct repository *r, struct commit_graph *g)
905 {
906 uint32_t i, cur_fanout_pos = 0;
907 struct object_id prev_oid, cur_oid, checksum;
908 int generation_zero = 0;
909 struct hashfile *f;
910 int devnull;
911
912 if (!g) {
913 graph_report("no commit-graph file loaded");
914 return 1;
915 }
916
917 verify_commit_graph_error = 0;
918
919 if (!g->chunk_oid_fanout)
920 graph_report("commit-graph is missing the OID Fanout chunk");
921 if (!g->chunk_oid_lookup)
922 graph_report("commit-graph is missing the OID Lookup chunk");
923 if (!g->chunk_commit_data)
924 graph_report("commit-graph is missing the Commit Data chunk");
925
926 if (verify_commit_graph_error)
927 return verify_commit_graph_error;
928
929 devnull = open("/dev/null", O_WRONLY);
930 f = hashfd(devnull, NULL);
931 hashwrite(f, g->data, g->data_len - g->hash_len);
932 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
933 if (hashcmp(checksum.hash, g->data + g->data_len - g->hash_len)) {
934 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
935 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
936 }
937
938 for (i = 0; i < g->num_commits; i++) {
939 struct commit *graph_commit;
940
941 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
942
943 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
944 graph_report("commit-graph has incorrect OID order: %s then %s",
945 oid_to_hex(&prev_oid),
946 oid_to_hex(&cur_oid));
947
948 oidcpy(&prev_oid, &cur_oid);
949
950 while (cur_oid.hash[0] > cur_fanout_pos) {
951 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
952
953 if (i != fanout_value)
954 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
955 cur_fanout_pos, fanout_value, i);
956 cur_fanout_pos++;
957 }
958
959 graph_commit = lookup_commit(r, &cur_oid);
960 if (!parse_commit_in_graph_one(g, graph_commit))
961 graph_report("failed to parse %s from commit-graph",
962 oid_to_hex(&cur_oid));
963 }
964
965 while (cur_fanout_pos < 256) {
966 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
967
968 if (g->num_commits != fanout_value)
969 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
970 cur_fanout_pos, fanout_value, i);
971
972 cur_fanout_pos++;
973 }
974
975 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
976 return verify_commit_graph_error;
977
978 for (i = 0; i < g->num_commits; i++) {
979 struct commit *graph_commit, *odb_commit;
980 struct commit_list *graph_parents, *odb_parents;
981 uint32_t max_generation = 0;
982
983 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
984
985 graph_commit = lookup_commit(r, &cur_oid);
986 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
987 if (parse_commit_internal(odb_commit, 0, 0)) {
988 graph_report("failed to parse %s from object database",
989 oid_to_hex(&cur_oid));
990 continue;
991 }
992
993 if (oidcmp(&get_commit_tree_in_graph_one(g, graph_commit)->object.oid,
994 get_commit_tree_oid(odb_commit)))
995 graph_report("root tree OID for commit %s in commit-graph is %s != %s",
996 oid_to_hex(&cur_oid),
997 oid_to_hex(get_commit_tree_oid(graph_commit)),
998 oid_to_hex(get_commit_tree_oid(odb_commit)));
999
1000 graph_parents = graph_commit->parents;
1001 odb_parents = odb_commit->parents;
1002
1003 while (graph_parents) {
1004 if (odb_parents == NULL) {
1005 graph_report("commit-graph parent list for commit %s is too long",
1006 oid_to_hex(&cur_oid));
1007 break;
1008 }
1009
1010 if (oidcmp(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1011 graph_report("commit-graph parent for %s is %s != %s",
1012 oid_to_hex(&cur_oid),
1013 oid_to_hex(&graph_parents->item->object.oid),
1014 oid_to_hex(&odb_parents->item->object.oid));
1015
1016 if (graph_parents->item->generation > max_generation)
1017 max_generation = graph_parents->item->generation;
1018
1019 graph_parents = graph_parents->next;
1020 odb_parents = odb_parents->next;
1021 }
1022
1023 if (odb_parents != NULL)
1024 graph_report("commit-graph parent list for commit %s terminates early",
1025 oid_to_hex(&cur_oid));
1026
1027 if (!graph_commit->generation) {
1028 if (generation_zero == GENERATION_NUMBER_EXISTS)
1029 graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
1030 oid_to_hex(&cur_oid));
1031 generation_zero = GENERATION_ZERO_EXISTS;
1032 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1033 graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
1034 oid_to_hex(&cur_oid));
1035
1036 if (generation_zero == GENERATION_ZERO_EXISTS)
1037 continue;
1038
1039 /*
1040 * If one of our parents has generation GENERATION_NUMBER_MAX, then
1041 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1042 * extra logic in the following condition.
1043 */
1044 if (max_generation == GENERATION_NUMBER_MAX)
1045 max_generation--;
1046
1047 if (graph_commit->generation != max_generation + 1)
1048 graph_report("commit-graph generation for commit %s is %u != %u",
1049 oid_to_hex(&cur_oid),
1050 graph_commit->generation,
1051 max_generation + 1);
1052
1053 if (graph_commit->date != odb_commit->date)
1054 graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
1055 oid_to_hex(&cur_oid),
1056 graph_commit->date,
1057 odb_commit->date);
1058 }
1059
1060 return verify_commit_graph_error;
1061 }
1062
1063 void free_commit_graph(struct commit_graph *g)
1064 {
1065 if (!g)
1066 return;
1067 if (g->graph_fd >= 0) {
1068 munmap((void *)g->data, g->data_len);
1069 g->data = NULL;
1070 close(g->graph_fd);
1071 }
1072 free(g);
1073 }