Merge branch 'jt/non-blob-lazy-fetch'
[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_PARENT_MISSING 0x7fffffff
38 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
39 #define GRAPH_PARENT_NONE 0x70000000
40
41 #define GRAPH_LAST_EDGE 0x80000000
42
43 #define GRAPH_HEADER_SIZE 8
44 #define GRAPH_FANOUT_SIZE (4 * 256)
45 #define GRAPH_CHUNKLOOKUP_WIDTH 12
46 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
47 + GRAPH_FANOUT_SIZE + GRAPH_OID_LEN)
48
49 char *get_commit_graph_filename(const char *obj_dir)
50 {
51 return xstrfmt("%s/info/commit-graph", obj_dir);
52 }
53
54 static struct commit_graph *alloc_commit_graph(void)
55 {
56 struct commit_graph *g = xcalloc(1, sizeof(*g));
57 g->graph_fd = -1;
58
59 return g;
60 }
61
62 extern int read_replace_refs;
63
64 static int commit_graph_compatible(struct repository *r)
65 {
66 if (!r->gitdir)
67 return 0;
68
69 if (read_replace_refs) {
70 prepare_replace_object(r);
71 if (hashmap_get_size(&r->objects->replace_map->map))
72 return 0;
73 }
74
75 prepare_commit_graft(r);
76 if (r->parsed_objects && r->parsed_objects->grafts_nr)
77 return 0;
78 if (is_repository_shallow(r))
79 return 0;
80
81 return 1;
82 }
83
84 struct commit_graph *load_commit_graph_one(const char *graph_file)
85 {
86 void *graph_map;
87 const unsigned char *data, *chunk_lookup;
88 size_t graph_size;
89 struct stat st;
90 uint32_t i;
91 struct commit_graph *graph;
92 int fd = git_open(graph_file);
93 uint64_t last_chunk_offset;
94 uint32_t last_chunk_id;
95 uint32_t graph_signature;
96 unsigned char graph_version, hash_version;
97
98 if (fd < 0)
99 return NULL;
100 if (fstat(fd, &st)) {
101 close(fd);
102 return NULL;
103 }
104 graph_size = xsize_t(st.st_size);
105
106 if (graph_size < GRAPH_MIN_SIZE) {
107 close(fd);
108 die(_("graph file %s is too small"), graph_file);
109 }
110 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
111 data = (const unsigned char *)graph_map;
112
113 graph_signature = get_be32(data);
114 if (graph_signature != GRAPH_SIGNATURE) {
115 error(_("graph signature %X does not match signature %X"),
116 graph_signature, GRAPH_SIGNATURE);
117 goto cleanup_fail;
118 }
119
120 graph_version = *(unsigned char*)(data + 4);
121 if (graph_version != GRAPH_VERSION) {
122 error(_("graph version %X does not match version %X"),
123 graph_version, GRAPH_VERSION);
124 goto cleanup_fail;
125 }
126
127 hash_version = *(unsigned char*)(data + 5);
128 if (hash_version != GRAPH_OID_VERSION) {
129 error(_("hash version %X does not match version %X"),
130 hash_version, GRAPH_OID_VERSION);
131 goto cleanup_fail;
132 }
133
134 graph = alloc_commit_graph();
135
136 graph->hash_len = GRAPH_OID_LEN;
137 graph->num_chunks = *(unsigned char*)(data + 6);
138 graph->graph_fd = fd;
139 graph->data = graph_map;
140 graph->data_len = graph_size;
141
142 last_chunk_id = 0;
143 last_chunk_offset = 8;
144 chunk_lookup = data + 8;
145 for (i = 0; i < graph->num_chunks; i++) {
146 uint32_t chunk_id = get_be32(chunk_lookup + 0);
147 uint64_t chunk_offset = get_be64(chunk_lookup + 4);
148 int chunk_repeated = 0;
149
150 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
151
152 if (chunk_offset > graph_size - GIT_MAX_RAWSZ) {
153 error(_("improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
154 (uint32_t)chunk_offset);
155 goto cleanup_fail;
156 }
157
158 switch (chunk_id) {
159 case GRAPH_CHUNKID_OIDFANOUT:
160 if (graph->chunk_oid_fanout)
161 chunk_repeated = 1;
162 else
163 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
164 break;
165
166 case GRAPH_CHUNKID_OIDLOOKUP:
167 if (graph->chunk_oid_lookup)
168 chunk_repeated = 1;
169 else
170 graph->chunk_oid_lookup = data + chunk_offset;
171 break;
172
173 case GRAPH_CHUNKID_DATA:
174 if (graph->chunk_commit_data)
175 chunk_repeated = 1;
176 else
177 graph->chunk_commit_data = data + chunk_offset;
178 break;
179
180 case GRAPH_CHUNKID_LARGEEDGES:
181 if (graph->chunk_large_edges)
182 chunk_repeated = 1;
183 else
184 graph->chunk_large_edges = data + chunk_offset;
185 break;
186 }
187
188 if (chunk_repeated) {
189 error(_("chunk id %08x appears multiple times"), chunk_id);
190 goto cleanup_fail;
191 }
192
193 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
194 {
195 graph->num_commits = (chunk_offset - last_chunk_offset)
196 / graph->hash_len;
197 }
198
199 last_chunk_id = chunk_id;
200 last_chunk_offset = chunk_offset;
201 }
202
203 return graph;
204
205 cleanup_fail:
206 munmap(graph_map, graph_size);
207 close(fd);
208 exit(1);
209 }
210
211 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
212 {
213 char *graph_name;
214
215 if (r->objects->commit_graph)
216 return;
217
218 graph_name = get_commit_graph_filename(obj_dir);
219 r->objects->commit_graph =
220 load_commit_graph_one(graph_name);
221
222 FREE_AND_NULL(graph_name);
223 }
224
225 /*
226 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
227 *
228 * On the first invocation, this function attemps to load the commit
229 * graph if the_repository is configured to have one.
230 */
231 static int prepare_commit_graph(struct repository *r)
232 {
233 struct alternate_object_database *alt;
234 char *obj_dir;
235 int config_value;
236
237 if (r->objects->commit_graph_attempted)
238 return !!r->objects->commit_graph;
239 r->objects->commit_graph_attempted = 1;
240
241 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
242 (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
243 !config_value))
244 /*
245 * This repository is not configured to use commit graphs, so
246 * do not load one. (But report commit_graph_attempted anyway
247 * so that commit graph loading is not attempted again for this
248 * repository.)
249 */
250 return 0;
251
252 if (!commit_graph_compatible(r))
253 return 0;
254
255 obj_dir = r->objects->objectdir;
256 prepare_commit_graph_one(r, obj_dir);
257 prepare_alt_odb(r);
258 for (alt = r->objects->alt_odb_list;
259 !r->objects->commit_graph && alt;
260 alt = alt->next)
261 prepare_commit_graph_one(r, alt->path);
262 return !!r->objects->commit_graph;
263 }
264
265 int generation_numbers_enabled(struct repository *r)
266 {
267 uint32_t first_generation;
268 struct commit_graph *g;
269 if (!prepare_commit_graph(r))
270 return 0;
271
272 g = r->objects->commit_graph;
273
274 if (!g->num_commits)
275 return 0;
276
277 first_generation = get_be32(g->chunk_commit_data +
278 g->hash_len + 8) >> 2;
279
280 return !!first_generation;
281 }
282
283 void close_commit_graph(struct repository *r)
284 {
285 free_commit_graph(r->objects->commit_graph);
286 r->objects->commit_graph = NULL;
287 }
288
289 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
290 {
291 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
292 g->chunk_oid_lookup, g->hash_len, pos);
293 }
294
295 static struct commit_list **insert_parent_or_die(struct commit_graph *g,
296 uint64_t pos,
297 struct commit_list **pptr)
298 {
299 struct commit *c;
300 struct object_id oid;
301
302 if (pos >= g->num_commits)
303 die("invalid parent position %"PRIu64, pos);
304
305 hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
306 c = lookup_commit(the_repository, &oid);
307 if (!c)
308 die(_("could not find commit %s"), oid_to_hex(&oid));
309 c->graph_pos = pos;
310 return &commit_list_insert(c, pptr)->next;
311 }
312
313 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
314 {
315 const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
316 item->graph_pos = pos;
317 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
318 }
319
320 static int fill_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t pos)
321 {
322 uint32_t edge_value;
323 uint32_t *parent_data_ptr;
324 uint64_t date_low, date_high;
325 struct commit_list **pptr;
326 const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
327
328 item->object.parsed = 1;
329 item->graph_pos = pos;
330
331 item->maybe_tree = NULL;
332
333 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
334 date_low = get_be32(commit_data + g->hash_len + 12);
335 item->date = (timestamp_t)((date_high << 32) | date_low);
336
337 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
338
339 pptr = &item->parents;
340
341 edge_value = get_be32(commit_data + g->hash_len);
342 if (edge_value == GRAPH_PARENT_NONE)
343 return 1;
344 pptr = insert_parent_or_die(g, edge_value, pptr);
345
346 edge_value = get_be32(commit_data + g->hash_len + 4);
347 if (edge_value == GRAPH_PARENT_NONE)
348 return 1;
349 if (!(edge_value & GRAPH_OCTOPUS_EDGES_NEEDED)) {
350 pptr = insert_parent_or_die(g, edge_value, pptr);
351 return 1;
352 }
353
354 parent_data_ptr = (uint32_t*)(g->chunk_large_edges +
355 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
356 do {
357 edge_value = get_be32(parent_data_ptr);
358 pptr = insert_parent_or_die(g,
359 edge_value & GRAPH_EDGE_LAST_MASK,
360 pptr);
361 parent_data_ptr++;
362 } while (!(edge_value & GRAPH_LAST_EDGE));
363
364 return 1;
365 }
366
367 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
368 {
369 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
370 *pos = item->graph_pos;
371 return 1;
372 } else {
373 return bsearch_graph(g, &(item->object.oid), pos);
374 }
375 }
376
377 static int parse_commit_in_graph_one(struct commit_graph *g, struct commit *item)
378 {
379 uint32_t pos;
380
381 if (item->object.parsed)
382 return 1;
383
384 if (find_commit_in_graph(item, g, &pos))
385 return fill_commit_in_graph(item, g, pos);
386
387 return 0;
388 }
389
390 int parse_commit_in_graph(struct repository *r, struct commit *item)
391 {
392 if (!prepare_commit_graph(r))
393 return 0;
394 return parse_commit_in_graph_one(r->objects->commit_graph, item);
395 }
396
397 void load_commit_graph_info(struct repository *r, struct commit *item)
398 {
399 uint32_t pos;
400 if (!prepare_commit_graph(r))
401 return;
402 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
403 fill_commit_graph_info(item, r->objects->commit_graph, pos);
404 }
405
406 static struct tree *load_tree_for_commit(struct commit_graph *g, struct commit *c)
407 {
408 struct object_id oid;
409 const unsigned char *commit_data = g->chunk_commit_data +
410 GRAPH_DATA_WIDTH * (c->graph_pos);
411
412 hashcpy(oid.hash, commit_data);
413 c->maybe_tree = lookup_tree(the_repository, &oid);
414
415 return c->maybe_tree;
416 }
417
418 static struct tree *get_commit_tree_in_graph_one(struct commit_graph *g,
419 const struct commit *c)
420 {
421 if (c->maybe_tree)
422 return c->maybe_tree;
423 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
424 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
425
426 return load_tree_for_commit(g, (struct commit *)c);
427 }
428
429 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
430 {
431 return get_commit_tree_in_graph_one(r->objects->commit_graph, c);
432 }
433
434 static void write_graph_chunk_fanout(struct hashfile *f,
435 struct commit **commits,
436 int nr_commits)
437 {
438 int i, count = 0;
439 struct commit **list = commits;
440
441 /*
442 * Write the first-level table (the list is sorted,
443 * but we use a 256-entry lookup to be able to avoid
444 * having to do eight extra binary search iterations).
445 */
446 for (i = 0; i < 256; i++) {
447 while (count < nr_commits) {
448 if ((*list)->object.oid.hash[0] != i)
449 break;
450 count++;
451 list++;
452 }
453
454 hashwrite_be32(f, count);
455 }
456 }
457
458 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
459 struct commit **commits, int nr_commits)
460 {
461 struct commit **list = commits;
462 int count;
463 for (count = 0; count < nr_commits; count++, list++)
464 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
465 }
466
467 static const unsigned char *commit_to_sha1(size_t index, void *table)
468 {
469 struct commit **commits = table;
470 return commits[index]->object.oid.hash;
471 }
472
473 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
474 struct commit **commits, int nr_commits)
475 {
476 struct commit **list = commits;
477 struct commit **last = commits + nr_commits;
478 uint32_t num_extra_edges = 0;
479
480 while (list < last) {
481 struct commit_list *parent;
482 int edge_value;
483 uint32_t packedDate[2];
484
485 parse_commit(*list);
486 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
487
488 parent = (*list)->parents;
489
490 if (!parent)
491 edge_value = GRAPH_PARENT_NONE;
492 else {
493 edge_value = sha1_pos(parent->item->object.oid.hash,
494 commits,
495 nr_commits,
496 commit_to_sha1);
497
498 if (edge_value < 0)
499 edge_value = GRAPH_PARENT_MISSING;
500 }
501
502 hashwrite_be32(f, edge_value);
503
504 if (parent)
505 parent = parent->next;
506
507 if (!parent)
508 edge_value = GRAPH_PARENT_NONE;
509 else if (parent->next)
510 edge_value = GRAPH_OCTOPUS_EDGES_NEEDED | num_extra_edges;
511 else {
512 edge_value = sha1_pos(parent->item->object.oid.hash,
513 commits,
514 nr_commits,
515 commit_to_sha1);
516 if (edge_value < 0)
517 edge_value = GRAPH_PARENT_MISSING;
518 }
519
520 hashwrite_be32(f, edge_value);
521
522 if (edge_value & GRAPH_OCTOPUS_EDGES_NEEDED) {
523 do {
524 num_extra_edges++;
525 parent = parent->next;
526 } while (parent);
527 }
528
529 if (sizeof((*list)->date) > 4)
530 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
531 else
532 packedDate[0] = 0;
533
534 packedDate[0] |= htonl((*list)->generation << 2);
535
536 packedDate[1] = htonl((*list)->date);
537 hashwrite(f, packedDate, 8);
538
539 list++;
540 }
541 }
542
543 static void write_graph_chunk_large_edges(struct hashfile *f,
544 struct commit **commits,
545 int nr_commits)
546 {
547 struct commit **list = commits;
548 struct commit **last = commits + nr_commits;
549 struct commit_list *parent;
550
551 while (list < last) {
552 int num_parents = 0;
553 for (parent = (*list)->parents; num_parents < 3 && parent;
554 parent = parent->next)
555 num_parents++;
556
557 if (num_parents <= 2) {
558 list++;
559 continue;
560 }
561
562 /* Since num_parents > 2, this initializer is safe. */
563 for (parent = (*list)->parents->next; parent; parent = parent->next) {
564 int edge_value = sha1_pos(parent->item->object.oid.hash,
565 commits,
566 nr_commits,
567 commit_to_sha1);
568
569 if (edge_value < 0)
570 edge_value = GRAPH_PARENT_MISSING;
571 else if (!parent->next)
572 edge_value |= GRAPH_LAST_EDGE;
573
574 hashwrite_be32(f, edge_value);
575 }
576
577 list++;
578 }
579 }
580
581 static int commit_compare(const void *_a, const void *_b)
582 {
583 const struct object_id *a = (const struct object_id *)_a;
584 const struct object_id *b = (const struct object_id *)_b;
585 return oidcmp(a, b);
586 }
587
588 struct packed_commit_list {
589 struct commit **list;
590 int nr;
591 int alloc;
592 };
593
594 struct packed_oid_list {
595 struct object_id *list;
596 int nr;
597 int alloc;
598 struct progress *progress;
599 int progress_done;
600 };
601
602 static int add_packed_commits(const struct object_id *oid,
603 struct packed_git *pack,
604 uint32_t pos,
605 void *data)
606 {
607 struct packed_oid_list *list = (struct packed_oid_list*)data;
608 enum object_type type;
609 off_t offset = nth_packed_object_offset(pack, pos);
610 struct object_info oi = OBJECT_INFO_INIT;
611
612 if (list->progress)
613 display_progress(list->progress, ++list->progress_done);
614
615 oi.typep = &type;
616 if (packed_object_info(the_repository, pack, offset, &oi) < 0)
617 die(_("unable to get type of object %s"), oid_to_hex(oid));
618
619 if (type != OBJ_COMMIT)
620 return 0;
621
622 ALLOC_GROW(list->list, list->nr + 1, list->alloc);
623 oidcpy(&(list->list[list->nr]), oid);
624 list->nr++;
625
626 return 0;
627 }
628
629 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
630 {
631 struct commit_list *parent;
632 for (parent = commit->parents; parent; parent = parent->next) {
633 if (!(parent->item->object.flags & UNINTERESTING)) {
634 ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
635 oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
636 oids->nr++;
637 parent->item->object.flags |= UNINTERESTING;
638 }
639 }
640 }
641
642 static void close_reachable(struct packed_oid_list *oids, int report_progress)
643 {
644 int i;
645 struct commit *commit;
646 struct progress *progress = NULL;
647 int j = 0;
648
649 if (report_progress)
650 progress = start_delayed_progress(
651 _("Annotating commits in commit graph"), 0);
652 for (i = 0; i < oids->nr; i++) {
653 display_progress(progress, ++j);
654 commit = lookup_commit(the_repository, &oids->list[i]);
655 if (commit)
656 commit->object.flags |= UNINTERESTING;
657 }
658
659 /*
660 * As this loop runs, oids->nr may grow, but not more
661 * than the number of missing commits in the reachable
662 * closure.
663 */
664 for (i = 0; i < oids->nr; i++) {
665 display_progress(progress, ++j);
666 commit = lookup_commit(the_repository, &oids->list[i]);
667
668 if (commit && !parse_commit(commit))
669 add_missing_parents(oids, commit);
670 }
671
672 for (i = 0; i < oids->nr; i++) {
673 display_progress(progress, ++j);
674 commit = lookup_commit(the_repository, &oids->list[i]);
675
676 if (commit)
677 commit->object.flags &= ~UNINTERESTING;
678 }
679 stop_progress(&progress);
680 }
681
682 static void compute_generation_numbers(struct packed_commit_list* commits,
683 int report_progress)
684 {
685 int i;
686 struct commit_list *list = NULL;
687 struct progress *progress = NULL;
688
689 if (report_progress)
690 progress = start_progress(
691 _("Computing commit graph generation numbers"),
692 commits->nr);
693 for (i = 0; i < commits->nr; i++) {
694 display_progress(progress, i + 1);
695 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
696 commits->list[i]->generation != GENERATION_NUMBER_ZERO)
697 continue;
698
699 commit_list_insert(commits->list[i], &list);
700 while (list) {
701 struct commit *current = list->item;
702 struct commit_list *parent;
703 int all_parents_computed = 1;
704 uint32_t max_generation = 0;
705
706 for (parent = current->parents; parent; parent = parent->next) {
707 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
708 parent->item->generation == GENERATION_NUMBER_ZERO) {
709 all_parents_computed = 0;
710 commit_list_insert(parent->item, &list);
711 break;
712 } else if (parent->item->generation > max_generation) {
713 max_generation = parent->item->generation;
714 }
715 }
716
717 if (all_parents_computed) {
718 current->generation = max_generation + 1;
719 pop_commit(&list);
720
721 if (current->generation > GENERATION_NUMBER_MAX)
722 current->generation = GENERATION_NUMBER_MAX;
723 }
724 }
725 }
726 stop_progress(&progress);
727 }
728
729 static int add_ref_to_list(const char *refname,
730 const struct object_id *oid,
731 int flags, void *cb_data)
732 {
733 struct string_list *list = (struct string_list *)cb_data;
734
735 string_list_append(list, oid_to_hex(oid));
736 return 0;
737 }
738
739 void write_commit_graph_reachable(const char *obj_dir, int append,
740 int report_progress)
741 {
742 struct string_list list;
743
744 string_list_init(&list, 1);
745 for_each_ref(add_ref_to_list, &list);
746 write_commit_graph(obj_dir, NULL, &list, append, report_progress);
747 }
748
749 void write_commit_graph(const char *obj_dir,
750 struct string_list *pack_indexes,
751 struct string_list *commit_hex,
752 int append, int report_progress)
753 {
754 struct packed_oid_list oids;
755 struct packed_commit_list commits;
756 struct hashfile *f;
757 uint32_t i, count_distinct = 0;
758 char *graph_name;
759 struct lock_file lk = LOCK_INIT;
760 uint32_t chunk_ids[5];
761 uint64_t chunk_offsets[5];
762 int num_chunks;
763 int num_extra_edges;
764 struct commit_list *parent;
765 struct progress *progress = NULL;
766
767 if (!commit_graph_compatible(the_repository))
768 return;
769
770 oids.nr = 0;
771 oids.alloc = approximate_object_count() / 4;
772 oids.progress = NULL;
773 oids.progress_done = 0;
774
775 if (append) {
776 prepare_commit_graph_one(the_repository, obj_dir);
777 if (the_repository->objects->commit_graph)
778 oids.alloc += the_repository->objects->commit_graph->num_commits;
779 }
780
781 if (oids.alloc < 1024)
782 oids.alloc = 1024;
783 ALLOC_ARRAY(oids.list, oids.alloc);
784
785 if (append && the_repository->objects->commit_graph) {
786 struct commit_graph *commit_graph =
787 the_repository->objects->commit_graph;
788 for (i = 0; i < commit_graph->num_commits; i++) {
789 const unsigned char *hash = commit_graph->chunk_oid_lookup +
790 commit_graph->hash_len * i;
791 hashcpy(oids.list[oids.nr++].hash, hash);
792 }
793 }
794
795 if (pack_indexes) {
796 struct strbuf packname = STRBUF_INIT;
797 int dirlen;
798 strbuf_addf(&packname, "%s/pack/", obj_dir);
799 dirlen = packname.len;
800 if (report_progress) {
801 oids.progress = start_delayed_progress(
802 _("Finding commits for commit graph"), 0);
803 oids.progress_done = 0;
804 }
805 for (i = 0; i < pack_indexes->nr; i++) {
806 struct packed_git *p;
807 strbuf_setlen(&packname, dirlen);
808 strbuf_addstr(&packname, pack_indexes->items[i].string);
809 p = add_packed_git(packname.buf, packname.len, 1);
810 if (!p)
811 die(_("error adding pack %s"), packname.buf);
812 if (open_pack_index(p))
813 die(_("error opening index for %s"), packname.buf);
814 for_each_object_in_pack(p, add_packed_commits, &oids, 0);
815 close_pack(p);
816 }
817 stop_progress(&oids.progress);
818 strbuf_release(&packname);
819 }
820
821 if (commit_hex) {
822 if (report_progress)
823 progress = start_delayed_progress(
824 _("Finding commits for commit graph"),
825 commit_hex->nr);
826 for (i = 0; i < commit_hex->nr; i++) {
827 const char *end;
828 struct object_id oid;
829 struct commit *result;
830
831 display_progress(progress, i + 1);
832 if (commit_hex->items[i].string &&
833 parse_oid_hex(commit_hex->items[i].string, &oid, &end))
834 continue;
835
836 result = lookup_commit_reference_gently(the_repository, &oid, 1);
837
838 if (result) {
839 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
840 oidcpy(&oids.list[oids.nr], &(result->object.oid));
841 oids.nr++;
842 }
843 }
844 stop_progress(&progress);
845 }
846
847 if (!pack_indexes && !commit_hex) {
848 if (report_progress)
849 oids.progress = start_delayed_progress(
850 _("Finding commits for commit graph"), 0);
851 for_each_packed_object(add_packed_commits, &oids, 0);
852 stop_progress(&oids.progress);
853 }
854
855 close_reachable(&oids, report_progress);
856
857 QSORT(oids.list, oids.nr, commit_compare);
858
859 count_distinct = 1;
860 for (i = 1; i < oids.nr; i++) {
861 if (!oideq(&oids.list[i - 1], &oids.list[i]))
862 count_distinct++;
863 }
864
865 if (count_distinct >= GRAPH_PARENT_MISSING)
866 die(_("the commit graph format cannot write %d commits"), count_distinct);
867
868 commits.nr = 0;
869 commits.alloc = count_distinct;
870 ALLOC_ARRAY(commits.list, commits.alloc);
871
872 num_extra_edges = 0;
873 for (i = 0; i < oids.nr; i++) {
874 int num_parents = 0;
875 if (i > 0 && oideq(&oids.list[i - 1], &oids.list[i]))
876 continue;
877
878 commits.list[commits.nr] = lookup_commit(the_repository, &oids.list[i]);
879 parse_commit(commits.list[commits.nr]);
880
881 for (parent = commits.list[commits.nr]->parents;
882 parent; parent = parent->next)
883 num_parents++;
884
885 if (num_parents > 2)
886 num_extra_edges += num_parents - 1;
887
888 commits.nr++;
889 }
890 num_chunks = num_extra_edges ? 4 : 3;
891
892 if (commits.nr >= GRAPH_PARENT_MISSING)
893 die(_("too many commits to write graph"));
894
895 compute_generation_numbers(&commits, report_progress);
896
897 graph_name = get_commit_graph_filename(obj_dir);
898 if (safe_create_leading_directories(graph_name))
899 die_errno(_("unable to create leading directories of %s"),
900 graph_name);
901
902 hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
903 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
904
905 hashwrite_be32(f, GRAPH_SIGNATURE);
906
907 hashwrite_u8(f, GRAPH_VERSION);
908 hashwrite_u8(f, GRAPH_OID_VERSION);
909 hashwrite_u8(f, num_chunks);
910 hashwrite_u8(f, 0); /* unused padding byte */
911
912 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
913 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
914 chunk_ids[2] = GRAPH_CHUNKID_DATA;
915 if (num_extra_edges)
916 chunk_ids[3] = GRAPH_CHUNKID_LARGEEDGES;
917 else
918 chunk_ids[3] = 0;
919 chunk_ids[4] = 0;
920
921 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
922 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
923 chunk_offsets[2] = chunk_offsets[1] + GRAPH_OID_LEN * commits.nr;
924 chunk_offsets[3] = chunk_offsets[2] + (GRAPH_OID_LEN + 16) * commits.nr;
925 chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
926
927 for (i = 0; i <= num_chunks; i++) {
928 uint32_t chunk_write[3];
929
930 chunk_write[0] = htonl(chunk_ids[i]);
931 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
932 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
933 hashwrite(f, chunk_write, 12);
934 }
935
936 write_graph_chunk_fanout(f, commits.list, commits.nr);
937 write_graph_chunk_oids(f, GRAPH_OID_LEN, commits.list, commits.nr);
938 write_graph_chunk_data(f, GRAPH_OID_LEN, commits.list, commits.nr);
939 write_graph_chunk_large_edges(f, commits.list, commits.nr);
940
941 close_commit_graph(the_repository);
942 finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
943 commit_lock_file(&lk);
944
945 free(oids.list);
946 oids.alloc = 0;
947 oids.nr = 0;
948 }
949
950 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
951 static int verify_commit_graph_error;
952
953 static void graph_report(const char *fmt, ...)
954 {
955 va_list ap;
956
957 verify_commit_graph_error = 1;
958 va_start(ap, fmt);
959 vfprintf(stderr, fmt, ap);
960 fprintf(stderr, "\n");
961 va_end(ap);
962 }
963
964 #define GENERATION_ZERO_EXISTS 1
965 #define GENERATION_NUMBER_EXISTS 2
966
967 int verify_commit_graph(struct repository *r, struct commit_graph *g)
968 {
969 uint32_t i, cur_fanout_pos = 0;
970 struct object_id prev_oid, cur_oid, checksum;
971 int generation_zero = 0;
972 struct hashfile *f;
973 int devnull;
974 struct progress *progress = NULL;
975
976 if (!g) {
977 graph_report("no commit-graph file loaded");
978 return 1;
979 }
980
981 verify_commit_graph_error = 0;
982
983 if (!g->chunk_oid_fanout)
984 graph_report("commit-graph is missing the OID Fanout chunk");
985 if (!g->chunk_oid_lookup)
986 graph_report("commit-graph is missing the OID Lookup chunk");
987 if (!g->chunk_commit_data)
988 graph_report("commit-graph is missing the Commit Data chunk");
989
990 if (verify_commit_graph_error)
991 return verify_commit_graph_error;
992
993 devnull = open("/dev/null", O_WRONLY);
994 f = hashfd(devnull, NULL);
995 hashwrite(f, g->data, g->data_len - g->hash_len);
996 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
997 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
998 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
999 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1000 }
1001
1002 for (i = 0; i < g->num_commits; i++) {
1003 struct commit *graph_commit;
1004
1005 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1006
1007 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1008 graph_report("commit-graph has incorrect OID order: %s then %s",
1009 oid_to_hex(&prev_oid),
1010 oid_to_hex(&cur_oid));
1011
1012 oidcpy(&prev_oid, &cur_oid);
1013
1014 while (cur_oid.hash[0] > cur_fanout_pos) {
1015 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1016
1017 if (i != fanout_value)
1018 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1019 cur_fanout_pos, fanout_value, i);
1020 cur_fanout_pos++;
1021 }
1022
1023 graph_commit = lookup_commit(r, &cur_oid);
1024 if (!parse_commit_in_graph_one(g, graph_commit))
1025 graph_report("failed to parse %s from commit-graph",
1026 oid_to_hex(&cur_oid));
1027 }
1028
1029 while (cur_fanout_pos < 256) {
1030 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1031
1032 if (g->num_commits != fanout_value)
1033 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1034 cur_fanout_pos, fanout_value, i);
1035
1036 cur_fanout_pos++;
1037 }
1038
1039 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1040 return verify_commit_graph_error;
1041
1042 progress = start_progress(_("Verifying commits in commit graph"),
1043 g->num_commits);
1044 for (i = 0; i < g->num_commits; i++) {
1045 struct commit *graph_commit, *odb_commit;
1046 struct commit_list *graph_parents, *odb_parents;
1047 uint32_t max_generation = 0;
1048
1049 display_progress(progress, i + 1);
1050 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1051
1052 graph_commit = lookup_commit(r, &cur_oid);
1053 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1054 if (parse_commit_internal(odb_commit, 0, 0)) {
1055 graph_report("failed to parse %s from object database",
1056 oid_to_hex(&cur_oid));
1057 continue;
1058 }
1059
1060 if (!oideq(&get_commit_tree_in_graph_one(g, graph_commit)->object.oid,
1061 get_commit_tree_oid(odb_commit)))
1062 graph_report("root tree OID for commit %s in commit-graph is %s != %s",
1063 oid_to_hex(&cur_oid),
1064 oid_to_hex(get_commit_tree_oid(graph_commit)),
1065 oid_to_hex(get_commit_tree_oid(odb_commit)));
1066
1067 graph_parents = graph_commit->parents;
1068 odb_parents = odb_commit->parents;
1069
1070 while (graph_parents) {
1071 if (odb_parents == NULL) {
1072 graph_report("commit-graph parent list for commit %s is too long",
1073 oid_to_hex(&cur_oid));
1074 break;
1075 }
1076
1077 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1078 graph_report("commit-graph parent for %s is %s != %s",
1079 oid_to_hex(&cur_oid),
1080 oid_to_hex(&graph_parents->item->object.oid),
1081 oid_to_hex(&odb_parents->item->object.oid));
1082
1083 if (graph_parents->item->generation > max_generation)
1084 max_generation = graph_parents->item->generation;
1085
1086 graph_parents = graph_parents->next;
1087 odb_parents = odb_parents->next;
1088 }
1089
1090 if (odb_parents != NULL)
1091 graph_report("commit-graph parent list for commit %s terminates early",
1092 oid_to_hex(&cur_oid));
1093
1094 if (!graph_commit->generation) {
1095 if (generation_zero == GENERATION_NUMBER_EXISTS)
1096 graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
1097 oid_to_hex(&cur_oid));
1098 generation_zero = GENERATION_ZERO_EXISTS;
1099 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1100 graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
1101 oid_to_hex(&cur_oid));
1102
1103 if (generation_zero == GENERATION_ZERO_EXISTS)
1104 continue;
1105
1106 /*
1107 * If one of our parents has generation GENERATION_NUMBER_MAX, then
1108 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1109 * extra logic in the following condition.
1110 */
1111 if (max_generation == GENERATION_NUMBER_MAX)
1112 max_generation--;
1113
1114 if (graph_commit->generation != max_generation + 1)
1115 graph_report("commit-graph generation for commit %s is %u != %u",
1116 oid_to_hex(&cur_oid),
1117 graph_commit->generation,
1118 max_generation + 1);
1119
1120 if (graph_commit->date != odb_commit->date)
1121 graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
1122 oid_to_hex(&cur_oid),
1123 graph_commit->date,
1124 odb_commit->date);
1125 }
1126 stop_progress(&progress);
1127
1128 return verify_commit_graph_error;
1129 }
1130
1131 void free_commit_graph(struct commit_graph *g)
1132 {
1133 if (!g)
1134 return;
1135 if (g->graph_fd >= 0) {
1136 munmap((void *)g->data, g->data_len);
1137 g->data = NULL;
1138 close(g->graph_fd);
1139 }
1140 free(g);
1141 }