commit-graph: writing missing parents is a BUG
[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 #include "progress.h"
19
20 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
21 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
22 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
23 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
24 #define GRAPH_CHUNKID_LARGEEDGES 0x45444745 /* "EDGE" */
25
26 #define GRAPH_DATA_WIDTH 36
27
28 #define GRAPH_VERSION_1 0x1
29 #define GRAPH_VERSION GRAPH_VERSION_1
30
31 #define GRAPH_OID_VERSION_SHA1 1
32 #define GRAPH_OID_LEN_SHA1 GIT_SHA1_RAWSZ
33 #define GRAPH_OID_VERSION GRAPH_OID_VERSION_SHA1
34 #define GRAPH_OID_LEN GRAPH_OID_LEN_SHA1
35
36 #define GRAPH_OCTOPUS_EDGES_NEEDED 0x80000000
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 (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
241 (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
242 !config_value))
243 /*
244 * This repository is not configured to use commit graphs, so
245 * do not load one. (But report commit_graph_attempted anyway
246 * so that commit graph loading is not attempted again for this
247 * repository.)
248 */
249 return 0;
250
251 if (!commit_graph_compatible(r))
252 return 0;
253
254 obj_dir = r->objects->objectdir;
255 prepare_commit_graph_one(r, obj_dir);
256 prepare_alt_odb(r);
257 for (alt = r->objects->alt_odb_list;
258 !r->objects->commit_graph && alt;
259 alt = alt->next)
260 prepare_commit_graph_one(r, alt->path);
261 return !!r->objects->commit_graph;
262 }
263
264 int generation_numbers_enabled(struct repository *r)
265 {
266 uint32_t first_generation;
267 struct commit_graph *g;
268 if (!prepare_commit_graph(r))
269 return 0;
270
271 g = r->objects->commit_graph;
272
273 if (!g->num_commits)
274 return 0;
275
276 first_generation = get_be32(g->chunk_commit_data +
277 g->hash_len + 8) >> 2;
278
279 return !!first_generation;
280 }
281
282 void close_commit_graph(struct repository *r)
283 {
284 free_commit_graph(r->objects->commit_graph);
285 r->objects->commit_graph = NULL;
286 }
287
288 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
289 {
290 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
291 g->chunk_oid_lookup, g->hash_len, pos);
292 }
293
294 static struct commit_list **insert_parent_or_die(struct commit_graph *g,
295 uint64_t pos,
296 struct commit_list **pptr)
297 {
298 struct commit *c;
299 struct object_id oid;
300
301 if (pos >= g->num_commits)
302 die("invalid parent position %"PRIu64, pos);
303
304 hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
305 c = lookup_commit(the_repository, &oid);
306 if (!c)
307 die(_("could not find commit %s"), oid_to_hex(&oid));
308 c->graph_pos = pos;
309 return &commit_list_insert(c, pptr)->next;
310 }
311
312 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
313 {
314 const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
315 item->graph_pos = pos;
316 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
317 }
318
319 static int fill_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t pos)
320 {
321 uint32_t edge_value;
322 uint32_t *parent_data_ptr;
323 uint64_t date_low, date_high;
324 struct commit_list **pptr;
325 const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
326
327 item->object.parsed = 1;
328 item->graph_pos = pos;
329
330 item->maybe_tree = NULL;
331
332 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
333 date_low = get_be32(commit_data + g->hash_len + 12);
334 item->date = (timestamp_t)((date_high << 32) | date_low);
335
336 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
337
338 pptr = &item->parents;
339
340 edge_value = get_be32(commit_data + g->hash_len);
341 if (edge_value == GRAPH_PARENT_NONE)
342 return 1;
343 pptr = insert_parent_or_die(g, edge_value, pptr);
344
345 edge_value = get_be32(commit_data + g->hash_len + 4);
346 if (edge_value == GRAPH_PARENT_NONE)
347 return 1;
348 if (!(edge_value & GRAPH_OCTOPUS_EDGES_NEEDED)) {
349 pptr = insert_parent_or_die(g, edge_value, pptr);
350 return 1;
351 }
352
353 parent_data_ptr = (uint32_t*)(g->chunk_large_edges +
354 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
355 do {
356 edge_value = get_be32(parent_data_ptr);
357 pptr = insert_parent_or_die(g,
358 edge_value & GRAPH_EDGE_LAST_MASK,
359 pptr);
360 parent_data_ptr++;
361 } while (!(edge_value & GRAPH_LAST_EDGE));
362
363 return 1;
364 }
365
366 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
367 {
368 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
369 *pos = item->graph_pos;
370 return 1;
371 } else {
372 return bsearch_graph(g, &(item->object.oid), pos);
373 }
374 }
375
376 static int parse_commit_in_graph_one(struct commit_graph *g, struct commit *item)
377 {
378 uint32_t pos;
379
380 if (item->object.parsed)
381 return 1;
382
383 if (find_commit_in_graph(item, g, &pos))
384 return fill_commit_in_graph(item, g, pos);
385
386 return 0;
387 }
388
389 int parse_commit_in_graph(struct repository *r, struct commit *item)
390 {
391 if (!prepare_commit_graph(r))
392 return 0;
393 return parse_commit_in_graph_one(r->objects->commit_graph, item);
394 }
395
396 void load_commit_graph_info(struct repository *r, struct commit *item)
397 {
398 uint32_t pos;
399 if (!prepare_commit_graph(r))
400 return;
401 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
402 fill_commit_graph_info(item, r->objects->commit_graph, pos);
403 }
404
405 static struct tree *load_tree_for_commit(struct commit_graph *g, struct commit *c)
406 {
407 struct object_id oid;
408 const unsigned char *commit_data = g->chunk_commit_data +
409 GRAPH_DATA_WIDTH * (c->graph_pos);
410
411 hashcpy(oid.hash, commit_data);
412 c->maybe_tree = lookup_tree(the_repository, &oid);
413
414 return c->maybe_tree;
415 }
416
417 static struct tree *get_commit_tree_in_graph_one(struct commit_graph *g,
418 const struct commit *c)
419 {
420 if (c->maybe_tree)
421 return c->maybe_tree;
422 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
423 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
424
425 return load_tree_for_commit(g, (struct commit *)c);
426 }
427
428 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
429 {
430 return get_commit_tree_in_graph_one(r->objects->commit_graph, c);
431 }
432
433 static void write_graph_chunk_fanout(struct hashfile *f,
434 struct commit **commits,
435 int nr_commits)
436 {
437 int i, count = 0;
438 struct commit **list = commits;
439
440 /*
441 * Write the first-level table (the list is sorted,
442 * but we use a 256-entry lookup to be able to avoid
443 * having to do eight extra binary search iterations).
444 */
445 for (i = 0; i < 256; i++) {
446 while (count < nr_commits) {
447 if ((*list)->object.oid.hash[0] != i)
448 break;
449 count++;
450 list++;
451 }
452
453 hashwrite_be32(f, count);
454 }
455 }
456
457 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
458 struct commit **commits, int nr_commits)
459 {
460 struct commit **list = commits;
461 int count;
462 for (count = 0; count < nr_commits; count++, list++)
463 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
464 }
465
466 static const unsigned char *commit_to_sha1(size_t index, void *table)
467 {
468 struct commit **commits = table;
469 return commits[index]->object.oid.hash;
470 }
471
472 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
473 struct commit **commits, int nr_commits)
474 {
475 struct commit **list = commits;
476 struct commit **last = commits + nr_commits;
477 uint32_t num_extra_edges = 0;
478
479 while (list < last) {
480 struct commit_list *parent;
481 int edge_value;
482 uint32_t packedDate[2];
483
484 parse_commit(*list);
485 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
486
487 parent = (*list)->parents;
488
489 if (!parent)
490 edge_value = GRAPH_PARENT_NONE;
491 else {
492 edge_value = sha1_pos(parent->item->object.oid.hash,
493 commits,
494 nr_commits,
495 commit_to_sha1);
496
497 if (edge_value < 0)
498 BUG("missing parent %s for commit %s",
499 oid_to_hex(&parent->item->object.oid),
500 oid_to_hex(&(*list)->object.oid));
501 }
502
503 hashwrite_be32(f, edge_value);
504
505 if (parent)
506 parent = parent->next;
507
508 if (!parent)
509 edge_value = GRAPH_PARENT_NONE;
510 else if (parent->next)
511 edge_value = GRAPH_OCTOPUS_EDGES_NEEDED | num_extra_edges;
512 else {
513 edge_value = sha1_pos(parent->item->object.oid.hash,
514 commits,
515 nr_commits,
516 commit_to_sha1);
517 if (edge_value < 0)
518 BUG("missing parent %s for commit %s",
519 oid_to_hex(&parent->item->object.oid),
520 oid_to_hex(&(*list)->object.oid));
521 }
522
523 hashwrite_be32(f, edge_value);
524
525 if (edge_value & GRAPH_OCTOPUS_EDGES_NEEDED) {
526 do {
527 num_extra_edges++;
528 parent = parent->next;
529 } while (parent);
530 }
531
532 if (sizeof((*list)->date) > 4)
533 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
534 else
535 packedDate[0] = 0;
536
537 packedDate[0] |= htonl((*list)->generation << 2);
538
539 packedDate[1] = htonl((*list)->date);
540 hashwrite(f, packedDate, 8);
541
542 list++;
543 }
544 }
545
546 static void write_graph_chunk_large_edges(struct hashfile *f,
547 struct commit **commits,
548 int nr_commits)
549 {
550 struct commit **list = commits;
551 struct commit **last = commits + nr_commits;
552 struct commit_list *parent;
553
554 while (list < last) {
555 int num_parents = 0;
556 for (parent = (*list)->parents; num_parents < 3 && parent;
557 parent = parent->next)
558 num_parents++;
559
560 if (num_parents <= 2) {
561 list++;
562 continue;
563 }
564
565 /* Since num_parents > 2, this initializer is safe. */
566 for (parent = (*list)->parents->next; parent; parent = parent->next) {
567 int edge_value = sha1_pos(parent->item->object.oid.hash,
568 commits,
569 nr_commits,
570 commit_to_sha1);
571
572 if (edge_value < 0)
573 BUG("missing parent %s for commit %s",
574 oid_to_hex(&parent->item->object.oid),
575 oid_to_hex(&(*list)->object.oid));
576 else if (!parent->next)
577 edge_value |= GRAPH_LAST_EDGE;
578
579 hashwrite_be32(f, edge_value);
580 }
581
582 list++;
583 }
584 }
585
586 static int commit_compare(const void *_a, const void *_b)
587 {
588 const struct object_id *a = (const struct object_id *)_a;
589 const struct object_id *b = (const struct object_id *)_b;
590 return oidcmp(a, b);
591 }
592
593 struct packed_commit_list {
594 struct commit **list;
595 int nr;
596 int alloc;
597 };
598
599 struct packed_oid_list {
600 struct object_id *list;
601 int nr;
602 int alloc;
603 struct progress *progress;
604 int progress_done;
605 };
606
607 static int add_packed_commits(const struct object_id *oid,
608 struct packed_git *pack,
609 uint32_t pos,
610 void *data)
611 {
612 struct packed_oid_list *list = (struct packed_oid_list*)data;
613 enum object_type type;
614 off_t offset = nth_packed_object_offset(pack, pos);
615 struct object_info oi = OBJECT_INFO_INIT;
616
617 if (list->progress)
618 display_progress(list->progress, ++list->progress_done);
619
620 oi.typep = &type;
621 if (packed_object_info(the_repository, pack, offset, &oi) < 0)
622 die(_("unable to get type of object %s"), oid_to_hex(oid));
623
624 if (type != OBJ_COMMIT)
625 return 0;
626
627 ALLOC_GROW(list->list, list->nr + 1, list->alloc);
628 oidcpy(&(list->list[list->nr]), oid);
629 list->nr++;
630
631 return 0;
632 }
633
634 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
635 {
636 struct commit_list *parent;
637 for (parent = commit->parents; parent; parent = parent->next) {
638 if (!(parent->item->object.flags & UNINTERESTING)) {
639 ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
640 oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
641 oids->nr++;
642 parent->item->object.flags |= UNINTERESTING;
643 }
644 }
645 }
646
647 static void close_reachable(struct packed_oid_list *oids, int report_progress)
648 {
649 int i;
650 struct commit *commit;
651 struct progress *progress = NULL;
652 int j = 0;
653
654 if (report_progress)
655 progress = start_delayed_progress(
656 _("Annotating commits in commit graph"), 0);
657 for (i = 0; i < oids->nr; i++) {
658 display_progress(progress, ++j);
659 commit = lookup_commit(the_repository, &oids->list[i]);
660 if (commit)
661 commit->object.flags |= UNINTERESTING;
662 }
663
664 /*
665 * As this loop runs, oids->nr may grow, but not more
666 * than the number of missing commits in the reachable
667 * closure.
668 */
669 for (i = 0; i < oids->nr; i++) {
670 display_progress(progress, ++j);
671 commit = lookup_commit(the_repository, &oids->list[i]);
672
673 if (commit && !parse_commit(commit))
674 add_missing_parents(oids, commit);
675 }
676
677 for (i = 0; i < oids->nr; i++) {
678 display_progress(progress, ++j);
679 commit = lookup_commit(the_repository, &oids->list[i]);
680
681 if (commit)
682 commit->object.flags &= ~UNINTERESTING;
683 }
684 stop_progress(&progress);
685 }
686
687 static void compute_generation_numbers(struct packed_commit_list* commits,
688 int report_progress)
689 {
690 int i;
691 struct commit_list *list = NULL;
692 struct progress *progress = NULL;
693
694 if (report_progress)
695 progress = start_progress(
696 _("Computing commit graph generation numbers"),
697 commits->nr);
698 for (i = 0; i < commits->nr; i++) {
699 display_progress(progress, i + 1);
700 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
701 commits->list[i]->generation != GENERATION_NUMBER_ZERO)
702 continue;
703
704 commit_list_insert(commits->list[i], &list);
705 while (list) {
706 struct commit *current = list->item;
707 struct commit_list *parent;
708 int all_parents_computed = 1;
709 uint32_t max_generation = 0;
710
711 for (parent = current->parents; parent; parent = parent->next) {
712 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
713 parent->item->generation == GENERATION_NUMBER_ZERO) {
714 all_parents_computed = 0;
715 commit_list_insert(parent->item, &list);
716 break;
717 } else if (parent->item->generation > max_generation) {
718 max_generation = parent->item->generation;
719 }
720 }
721
722 if (all_parents_computed) {
723 current->generation = max_generation + 1;
724 pop_commit(&list);
725
726 if (current->generation > GENERATION_NUMBER_MAX)
727 current->generation = GENERATION_NUMBER_MAX;
728 }
729 }
730 }
731 stop_progress(&progress);
732 }
733
734 static int add_ref_to_list(const char *refname,
735 const struct object_id *oid,
736 int flags, void *cb_data)
737 {
738 struct string_list *list = (struct string_list *)cb_data;
739
740 string_list_append(list, oid_to_hex(oid));
741 return 0;
742 }
743
744 void write_commit_graph_reachable(const char *obj_dir, int append,
745 int report_progress)
746 {
747 struct string_list list = STRING_LIST_INIT_DUP;
748
749 for_each_ref(add_ref_to_list, &list);
750 write_commit_graph(obj_dir, NULL, &list, append, report_progress);
751
752 string_list_clear(&list, 0);
753 }
754
755 void write_commit_graph(const char *obj_dir,
756 struct string_list *pack_indexes,
757 struct string_list *commit_hex,
758 int append, int report_progress)
759 {
760 struct packed_oid_list oids;
761 struct packed_commit_list commits;
762 struct hashfile *f;
763 uint32_t i, count_distinct = 0;
764 char *graph_name;
765 struct lock_file lk = LOCK_INIT;
766 uint32_t chunk_ids[5];
767 uint64_t chunk_offsets[5];
768 int num_chunks;
769 int num_extra_edges;
770 struct commit_list *parent;
771 struct progress *progress = NULL;
772
773 if (!commit_graph_compatible(the_repository))
774 return;
775
776 oids.nr = 0;
777 oids.alloc = approximate_object_count() / 32;
778 oids.progress = NULL;
779 oids.progress_done = 0;
780
781 if (append) {
782 prepare_commit_graph_one(the_repository, obj_dir);
783 if (the_repository->objects->commit_graph)
784 oids.alloc += the_repository->objects->commit_graph->num_commits;
785 }
786
787 if (oids.alloc < 1024)
788 oids.alloc = 1024;
789 ALLOC_ARRAY(oids.list, oids.alloc);
790
791 if (append && the_repository->objects->commit_graph) {
792 struct commit_graph *commit_graph =
793 the_repository->objects->commit_graph;
794 for (i = 0; i < commit_graph->num_commits; i++) {
795 const unsigned char *hash = commit_graph->chunk_oid_lookup +
796 commit_graph->hash_len * i;
797 hashcpy(oids.list[oids.nr++].hash, hash);
798 }
799 }
800
801 if (pack_indexes) {
802 struct strbuf packname = STRBUF_INIT;
803 int dirlen;
804 strbuf_addf(&packname, "%s/pack/", obj_dir);
805 dirlen = packname.len;
806 if (report_progress) {
807 oids.progress = start_delayed_progress(
808 _("Finding commits for commit graph"), 0);
809 oids.progress_done = 0;
810 }
811 for (i = 0; i < pack_indexes->nr; i++) {
812 struct packed_git *p;
813 strbuf_setlen(&packname, dirlen);
814 strbuf_addstr(&packname, pack_indexes->items[i].string);
815 p = add_packed_git(packname.buf, packname.len, 1);
816 if (!p)
817 die(_("error adding pack %s"), packname.buf);
818 if (open_pack_index(p))
819 die(_("error opening index for %s"), packname.buf);
820 for_each_object_in_pack(p, add_packed_commits, &oids, 0);
821 close_pack(p);
822 free(p);
823 }
824 stop_progress(&oids.progress);
825 strbuf_release(&packname);
826 }
827
828 if (commit_hex) {
829 if (report_progress)
830 progress = start_delayed_progress(
831 _("Finding commits for commit graph"),
832 commit_hex->nr);
833 for (i = 0; i < commit_hex->nr; i++) {
834 const char *end;
835 struct object_id oid;
836 struct commit *result;
837
838 display_progress(progress, i + 1);
839 if (commit_hex->items[i].string &&
840 parse_oid_hex(commit_hex->items[i].string, &oid, &end))
841 continue;
842
843 result = lookup_commit_reference_gently(the_repository, &oid, 1);
844
845 if (result) {
846 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
847 oidcpy(&oids.list[oids.nr], &(result->object.oid));
848 oids.nr++;
849 }
850 }
851 stop_progress(&progress);
852 }
853
854 if (!pack_indexes && !commit_hex) {
855 if (report_progress)
856 oids.progress = start_delayed_progress(
857 _("Finding commits for commit graph"), 0);
858 for_each_packed_object(add_packed_commits, &oids, 0);
859 stop_progress(&oids.progress);
860 }
861
862 close_reachable(&oids, report_progress);
863
864 QSORT(oids.list, oids.nr, commit_compare);
865
866 count_distinct = 1;
867 for (i = 1; i < oids.nr; i++) {
868 if (!oideq(&oids.list[i - 1], &oids.list[i]))
869 count_distinct++;
870 }
871
872 if (count_distinct >= GRAPH_EDGE_LAST_MASK)
873 die(_("the commit graph format cannot write %d commits"), count_distinct);
874
875 commits.nr = 0;
876 commits.alloc = count_distinct;
877 ALLOC_ARRAY(commits.list, commits.alloc);
878
879 num_extra_edges = 0;
880 for (i = 0; i < oids.nr; i++) {
881 int num_parents = 0;
882 if (i > 0 && oideq(&oids.list[i - 1], &oids.list[i]))
883 continue;
884
885 commits.list[commits.nr] = lookup_commit(the_repository, &oids.list[i]);
886 parse_commit(commits.list[commits.nr]);
887
888 for (parent = commits.list[commits.nr]->parents;
889 parent; parent = parent->next)
890 num_parents++;
891
892 if (num_parents > 2)
893 num_extra_edges += num_parents - 1;
894
895 commits.nr++;
896 }
897 num_chunks = num_extra_edges ? 4 : 3;
898
899 if (commits.nr >= GRAPH_EDGE_LAST_MASK)
900 die(_("too many commits to write graph"));
901
902 compute_generation_numbers(&commits, report_progress);
903
904 graph_name = get_commit_graph_filename(obj_dir);
905 if (safe_create_leading_directories(graph_name)) {
906 UNLEAK(graph_name);
907 die_errno(_("unable to create leading directories of %s"),
908 graph_name);
909 }
910
911 hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
912 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
913
914 hashwrite_be32(f, GRAPH_SIGNATURE);
915
916 hashwrite_u8(f, GRAPH_VERSION);
917 hashwrite_u8(f, GRAPH_OID_VERSION);
918 hashwrite_u8(f, num_chunks);
919 hashwrite_u8(f, 0); /* unused padding byte */
920
921 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
922 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
923 chunk_ids[2] = GRAPH_CHUNKID_DATA;
924 if (num_extra_edges)
925 chunk_ids[3] = GRAPH_CHUNKID_LARGEEDGES;
926 else
927 chunk_ids[3] = 0;
928 chunk_ids[4] = 0;
929
930 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
931 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
932 chunk_offsets[2] = chunk_offsets[1] + GRAPH_OID_LEN * commits.nr;
933 chunk_offsets[3] = chunk_offsets[2] + (GRAPH_OID_LEN + 16) * commits.nr;
934 chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
935
936 for (i = 0; i <= num_chunks; i++) {
937 uint32_t chunk_write[3];
938
939 chunk_write[0] = htonl(chunk_ids[i]);
940 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
941 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
942 hashwrite(f, chunk_write, 12);
943 }
944
945 write_graph_chunk_fanout(f, commits.list, commits.nr);
946 write_graph_chunk_oids(f, GRAPH_OID_LEN, commits.list, commits.nr);
947 write_graph_chunk_data(f, GRAPH_OID_LEN, commits.list, commits.nr);
948 write_graph_chunk_large_edges(f, commits.list, commits.nr);
949
950 close_commit_graph(the_repository);
951 finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
952 commit_lock_file(&lk);
953
954 free(graph_name);
955 free(commits.list);
956 free(oids.list);
957 }
958
959 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
960 static int verify_commit_graph_error;
961
962 static void graph_report(const char *fmt, ...)
963 {
964 va_list ap;
965
966 verify_commit_graph_error = 1;
967 va_start(ap, fmt);
968 vfprintf(stderr, fmt, ap);
969 fprintf(stderr, "\n");
970 va_end(ap);
971 }
972
973 #define GENERATION_ZERO_EXISTS 1
974 #define GENERATION_NUMBER_EXISTS 2
975
976 int verify_commit_graph(struct repository *r, struct commit_graph *g)
977 {
978 uint32_t i, cur_fanout_pos = 0;
979 struct object_id prev_oid, cur_oid, checksum;
980 int generation_zero = 0;
981 struct hashfile *f;
982 int devnull;
983 struct progress *progress = NULL;
984
985 if (!g) {
986 graph_report("no commit-graph file loaded");
987 return 1;
988 }
989
990 verify_commit_graph_error = 0;
991
992 if (!g->chunk_oid_fanout)
993 graph_report("commit-graph is missing the OID Fanout chunk");
994 if (!g->chunk_oid_lookup)
995 graph_report("commit-graph is missing the OID Lookup chunk");
996 if (!g->chunk_commit_data)
997 graph_report("commit-graph is missing the Commit Data chunk");
998
999 if (verify_commit_graph_error)
1000 return verify_commit_graph_error;
1001
1002 devnull = open("/dev/null", O_WRONLY);
1003 f = hashfd(devnull, NULL);
1004 hashwrite(f, g->data, g->data_len - g->hash_len);
1005 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1006 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1007 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1008 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1009 }
1010
1011 for (i = 0; i < g->num_commits; i++) {
1012 struct commit *graph_commit;
1013
1014 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1015
1016 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1017 graph_report("commit-graph has incorrect OID order: %s then %s",
1018 oid_to_hex(&prev_oid),
1019 oid_to_hex(&cur_oid));
1020
1021 oidcpy(&prev_oid, &cur_oid);
1022
1023 while (cur_oid.hash[0] > cur_fanout_pos) {
1024 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1025
1026 if (i != fanout_value)
1027 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1028 cur_fanout_pos, fanout_value, i);
1029 cur_fanout_pos++;
1030 }
1031
1032 graph_commit = lookup_commit(r, &cur_oid);
1033 if (!parse_commit_in_graph_one(g, graph_commit))
1034 graph_report("failed to parse %s from commit-graph",
1035 oid_to_hex(&cur_oid));
1036 }
1037
1038 while (cur_fanout_pos < 256) {
1039 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1040
1041 if (g->num_commits != fanout_value)
1042 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1043 cur_fanout_pos, fanout_value, i);
1044
1045 cur_fanout_pos++;
1046 }
1047
1048 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1049 return verify_commit_graph_error;
1050
1051 progress = start_progress(_("Verifying commits in commit graph"),
1052 g->num_commits);
1053 for (i = 0; i < g->num_commits; i++) {
1054 struct commit *graph_commit, *odb_commit;
1055 struct commit_list *graph_parents, *odb_parents;
1056 uint32_t max_generation = 0;
1057
1058 display_progress(progress, i + 1);
1059 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1060
1061 graph_commit = lookup_commit(r, &cur_oid);
1062 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1063 if (parse_commit_internal(odb_commit, 0, 0)) {
1064 graph_report("failed to parse %s from object database",
1065 oid_to_hex(&cur_oid));
1066 continue;
1067 }
1068
1069 if (!oideq(&get_commit_tree_in_graph_one(g, graph_commit)->object.oid,
1070 get_commit_tree_oid(odb_commit)))
1071 graph_report("root tree OID for commit %s in commit-graph is %s != %s",
1072 oid_to_hex(&cur_oid),
1073 oid_to_hex(get_commit_tree_oid(graph_commit)),
1074 oid_to_hex(get_commit_tree_oid(odb_commit)));
1075
1076 graph_parents = graph_commit->parents;
1077 odb_parents = odb_commit->parents;
1078
1079 while (graph_parents) {
1080 if (odb_parents == NULL) {
1081 graph_report("commit-graph parent list for commit %s is too long",
1082 oid_to_hex(&cur_oid));
1083 break;
1084 }
1085
1086 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1087 graph_report("commit-graph parent for %s is %s != %s",
1088 oid_to_hex(&cur_oid),
1089 oid_to_hex(&graph_parents->item->object.oid),
1090 oid_to_hex(&odb_parents->item->object.oid));
1091
1092 if (graph_parents->item->generation > max_generation)
1093 max_generation = graph_parents->item->generation;
1094
1095 graph_parents = graph_parents->next;
1096 odb_parents = odb_parents->next;
1097 }
1098
1099 if (odb_parents != NULL)
1100 graph_report("commit-graph parent list for commit %s terminates early",
1101 oid_to_hex(&cur_oid));
1102
1103 if (!graph_commit->generation) {
1104 if (generation_zero == GENERATION_NUMBER_EXISTS)
1105 graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
1106 oid_to_hex(&cur_oid));
1107 generation_zero = GENERATION_ZERO_EXISTS;
1108 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1109 graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
1110 oid_to_hex(&cur_oid));
1111
1112 if (generation_zero == GENERATION_ZERO_EXISTS)
1113 continue;
1114
1115 /*
1116 * If one of our parents has generation GENERATION_NUMBER_MAX, then
1117 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1118 * extra logic in the following condition.
1119 */
1120 if (max_generation == GENERATION_NUMBER_MAX)
1121 max_generation--;
1122
1123 if (graph_commit->generation != max_generation + 1)
1124 graph_report("commit-graph generation for commit %s is %u != %u",
1125 oid_to_hex(&cur_oid),
1126 graph_commit->generation,
1127 max_generation + 1);
1128
1129 if (graph_commit->date != odb_commit->date)
1130 graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
1131 oid_to_hex(&cur_oid),
1132 graph_commit->date,
1133 odb_commit->date);
1134 }
1135 stop_progress(&progress);
1136
1137 return verify_commit_graph_error;
1138 }
1139
1140 void free_commit_graph(struct commit_graph *g)
1141 {
1142 if (!g)
1143 return;
1144 if (g->graph_fd >= 0) {
1145 munmap((void *)g->data, g->data_len);
1146 g->data = NULL;
1147 close(g->graph_fd);
1148 }
1149 free(g);
1150 }