Merge branch 'ab/commit-graph-write-optim'
[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 (the_hash_algo->rawsz + 16)
27
28 #define GRAPH_VERSION_1 0x1
29 #define GRAPH_VERSION GRAPH_VERSION_1
30
31 #define GRAPH_OCTOPUS_EDGES_NEEDED 0x80000000
32 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
33 #define GRAPH_PARENT_NONE 0x70000000
34
35 #define GRAPH_LAST_EDGE 0x80000000
36
37 #define GRAPH_HEADER_SIZE 8
38 #define GRAPH_FANOUT_SIZE (4 * 256)
39 #define GRAPH_CHUNKLOOKUP_WIDTH 12
40 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
41 + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
42
43 char *get_commit_graph_filename(const char *obj_dir)
44 {
45 return xstrfmt("%s/info/commit-graph", obj_dir);
46 }
47
48 static uint8_t oid_version(void)
49 {
50 return 1;
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 size_t graph_size;
87 struct stat st;
88 struct commit_graph *ret;
89 int fd = git_open(graph_file);
90
91 if (fd < 0)
92 return NULL;
93 if (fstat(fd, &st)) {
94 close(fd);
95 return NULL;
96 }
97 graph_size = xsize_t(st.st_size);
98
99 if (graph_size < GRAPH_MIN_SIZE) {
100 close(fd);
101 die(_("graph file %s is too small"), graph_file);
102 }
103 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
104 ret = parse_commit_graph(graph_map, fd, graph_size);
105
106 if (!ret) {
107 munmap(graph_map, graph_size);
108 close(fd);
109 exit(1);
110 }
111
112 return ret;
113 }
114
115 struct commit_graph *parse_commit_graph(void *graph_map, int fd,
116 size_t graph_size)
117 {
118 const unsigned char *data, *chunk_lookup;
119 uint32_t i;
120 struct commit_graph *graph;
121 uint64_t last_chunk_offset;
122 uint32_t last_chunk_id;
123 uint32_t graph_signature;
124 unsigned char graph_version, hash_version;
125
126 if (!graph_map)
127 return NULL;
128
129 if (graph_size < GRAPH_MIN_SIZE)
130 return NULL;
131
132 data = (const unsigned char *)graph_map;
133
134 graph_signature = get_be32(data);
135 if (graph_signature != GRAPH_SIGNATURE) {
136 error(_("graph signature %X does not match signature %X"),
137 graph_signature, GRAPH_SIGNATURE);
138 return NULL;
139 }
140
141 graph_version = *(unsigned char*)(data + 4);
142 if (graph_version != GRAPH_VERSION) {
143 error(_("graph version %X does not match version %X"),
144 graph_version, GRAPH_VERSION);
145 return NULL;
146 }
147
148 hash_version = *(unsigned char*)(data + 5);
149 if (hash_version != oid_version()) {
150 error(_("hash version %X does not match version %X"),
151 hash_version, oid_version());
152 return NULL;
153 }
154
155 graph = alloc_commit_graph();
156
157 graph->hash_len = the_hash_algo->rawsz;
158 graph->num_chunks = *(unsigned char*)(data + 6);
159 graph->graph_fd = fd;
160 graph->data = graph_map;
161 graph->data_len = graph_size;
162
163 last_chunk_id = 0;
164 last_chunk_offset = 8;
165 chunk_lookup = data + 8;
166 for (i = 0; i < graph->num_chunks; i++) {
167 uint32_t chunk_id;
168 uint64_t chunk_offset;
169 int chunk_repeated = 0;
170
171 if (data + graph_size - chunk_lookup <
172 GRAPH_CHUNKLOOKUP_WIDTH) {
173 error(_("chunk lookup table entry missing; graph file may be incomplete"));
174 free(graph);
175 return NULL;
176 }
177
178 chunk_id = get_be32(chunk_lookup + 0);
179 chunk_offset = get_be64(chunk_lookup + 4);
180
181 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
182
183 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
184 error(_("improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
185 (uint32_t)chunk_offset);
186 free(graph);
187 return NULL;
188 }
189
190 switch (chunk_id) {
191 case GRAPH_CHUNKID_OIDFANOUT:
192 if (graph->chunk_oid_fanout)
193 chunk_repeated = 1;
194 else
195 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
196 break;
197
198 case GRAPH_CHUNKID_OIDLOOKUP:
199 if (graph->chunk_oid_lookup)
200 chunk_repeated = 1;
201 else
202 graph->chunk_oid_lookup = data + chunk_offset;
203 break;
204
205 case GRAPH_CHUNKID_DATA:
206 if (graph->chunk_commit_data)
207 chunk_repeated = 1;
208 else
209 graph->chunk_commit_data = data + chunk_offset;
210 break;
211
212 case GRAPH_CHUNKID_LARGEEDGES:
213 if (graph->chunk_large_edges)
214 chunk_repeated = 1;
215 else
216 graph->chunk_large_edges = data + chunk_offset;
217 break;
218 }
219
220 if (chunk_repeated) {
221 error(_("chunk id %08x appears multiple times"), chunk_id);
222 free(graph);
223 return NULL;
224 }
225
226 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
227 {
228 graph->num_commits = (chunk_offset - last_chunk_offset)
229 / graph->hash_len;
230 }
231
232 last_chunk_id = chunk_id;
233 last_chunk_offset = chunk_offset;
234 }
235
236 return graph;
237 }
238
239 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
240 {
241 char *graph_name;
242
243 if (r->objects->commit_graph)
244 return;
245
246 graph_name = get_commit_graph_filename(obj_dir);
247 r->objects->commit_graph =
248 load_commit_graph_one(graph_name);
249
250 FREE_AND_NULL(graph_name);
251 }
252
253 /*
254 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
255 *
256 * On the first invocation, this function attemps to load the commit
257 * graph if the_repository is configured to have one.
258 */
259 static int prepare_commit_graph(struct repository *r)
260 {
261 struct object_directory *odb;
262 int config_value;
263
264 if (r->objects->commit_graph_attempted)
265 return !!r->objects->commit_graph;
266 r->objects->commit_graph_attempted = 1;
267
268 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
269 (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
270 !config_value))
271 /*
272 * This repository is not configured to use commit graphs, so
273 * do not load one. (But report commit_graph_attempted anyway
274 * so that commit graph loading is not attempted again for this
275 * repository.)
276 */
277 return 0;
278
279 if (!commit_graph_compatible(r))
280 return 0;
281
282 prepare_alt_odb(r);
283 for (odb = r->objects->odb;
284 !r->objects->commit_graph && odb;
285 odb = odb->next)
286 prepare_commit_graph_one(r, odb->path);
287 return !!r->objects->commit_graph;
288 }
289
290 int generation_numbers_enabled(struct repository *r)
291 {
292 uint32_t first_generation;
293 struct commit_graph *g;
294 if (!prepare_commit_graph(r))
295 return 0;
296
297 g = r->objects->commit_graph;
298
299 if (!g->num_commits)
300 return 0;
301
302 first_generation = get_be32(g->chunk_commit_data +
303 g->hash_len + 8) >> 2;
304
305 return !!first_generation;
306 }
307
308 void close_commit_graph(struct repository *r)
309 {
310 free_commit_graph(r->objects->commit_graph);
311 r->objects->commit_graph = NULL;
312 }
313
314 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
315 {
316 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
317 g->chunk_oid_lookup, g->hash_len, pos);
318 }
319
320 static struct commit_list **insert_parent_or_die(struct repository *r,
321 struct commit_graph *g,
322 uint64_t pos,
323 struct commit_list **pptr)
324 {
325 struct commit *c;
326 struct object_id oid;
327
328 if (pos >= g->num_commits)
329 die("invalid parent position %"PRIu64, pos);
330
331 hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
332 c = lookup_commit(r, &oid);
333 if (!c)
334 die(_("could not find commit %s"), oid_to_hex(&oid));
335 c->graph_pos = pos;
336 return &commit_list_insert(c, pptr)->next;
337 }
338
339 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
340 {
341 const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
342 item->graph_pos = pos;
343 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
344 }
345
346 static int fill_commit_in_graph(struct repository *r,
347 struct commit *item,
348 struct commit_graph *g, uint32_t pos)
349 {
350 uint32_t edge_value;
351 uint32_t *parent_data_ptr;
352 uint64_t date_low, date_high;
353 struct commit_list **pptr;
354 const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
355
356 item->object.parsed = 1;
357 item->graph_pos = pos;
358
359 item->maybe_tree = NULL;
360
361 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
362 date_low = get_be32(commit_data + g->hash_len + 12);
363 item->date = (timestamp_t)((date_high << 32) | date_low);
364
365 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
366
367 pptr = &item->parents;
368
369 edge_value = get_be32(commit_data + g->hash_len);
370 if (edge_value == GRAPH_PARENT_NONE)
371 return 1;
372 pptr = insert_parent_or_die(r, g, edge_value, pptr);
373
374 edge_value = get_be32(commit_data + g->hash_len + 4);
375 if (edge_value == GRAPH_PARENT_NONE)
376 return 1;
377 if (!(edge_value & GRAPH_OCTOPUS_EDGES_NEEDED)) {
378 pptr = insert_parent_or_die(r, g, edge_value, pptr);
379 return 1;
380 }
381
382 parent_data_ptr = (uint32_t*)(g->chunk_large_edges +
383 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
384 do {
385 edge_value = get_be32(parent_data_ptr);
386 pptr = insert_parent_or_die(r, g,
387 edge_value & GRAPH_EDGE_LAST_MASK,
388 pptr);
389 parent_data_ptr++;
390 } while (!(edge_value & GRAPH_LAST_EDGE));
391
392 return 1;
393 }
394
395 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
396 {
397 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
398 *pos = item->graph_pos;
399 return 1;
400 } else {
401 return bsearch_graph(g, &(item->object.oid), pos);
402 }
403 }
404
405 static int parse_commit_in_graph_one(struct repository *r,
406 struct commit_graph *g,
407 struct commit *item)
408 {
409 uint32_t pos;
410
411 if (item->object.parsed)
412 return 1;
413
414 if (find_commit_in_graph(item, g, &pos))
415 return fill_commit_in_graph(r, item, g, pos);
416
417 return 0;
418 }
419
420 int parse_commit_in_graph(struct repository *r, struct commit *item)
421 {
422 if (!prepare_commit_graph(r))
423 return 0;
424 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
425 }
426
427 void load_commit_graph_info(struct repository *r, struct commit *item)
428 {
429 uint32_t pos;
430 if (!prepare_commit_graph(r))
431 return;
432 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
433 fill_commit_graph_info(item, r->objects->commit_graph, pos);
434 }
435
436 static struct tree *load_tree_for_commit(struct repository *r,
437 struct commit_graph *g,
438 struct commit *c)
439 {
440 struct object_id oid;
441 const unsigned char *commit_data = g->chunk_commit_data +
442 GRAPH_DATA_WIDTH * (c->graph_pos);
443
444 hashcpy(oid.hash, commit_data);
445 c->maybe_tree = lookup_tree(r, &oid);
446
447 return c->maybe_tree;
448 }
449
450 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
451 struct commit_graph *g,
452 const struct commit *c)
453 {
454 if (c->maybe_tree)
455 return c->maybe_tree;
456 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
457 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
458
459 return load_tree_for_commit(r, g, (struct commit *)c);
460 }
461
462 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
463 {
464 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
465 }
466
467 static void write_graph_chunk_fanout(struct hashfile *f,
468 struct commit **commits,
469 int nr_commits)
470 {
471 int i, count = 0;
472 struct commit **list = commits;
473
474 /*
475 * Write the first-level table (the list is sorted,
476 * but we use a 256-entry lookup to be able to avoid
477 * having to do eight extra binary search iterations).
478 */
479 for (i = 0; i < 256; i++) {
480 while (count < nr_commits) {
481 if ((*list)->object.oid.hash[0] != i)
482 break;
483 count++;
484 list++;
485 }
486
487 hashwrite_be32(f, count);
488 }
489 }
490
491 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
492 struct commit **commits, int nr_commits)
493 {
494 struct commit **list = commits;
495 int count;
496 for (count = 0; count < nr_commits; count++, list++)
497 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
498 }
499
500 static const unsigned char *commit_to_sha1(size_t index, void *table)
501 {
502 struct commit **commits = table;
503 return commits[index]->object.oid.hash;
504 }
505
506 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
507 struct commit **commits, int nr_commits)
508 {
509 struct commit **list = commits;
510 struct commit **last = commits + nr_commits;
511 uint32_t num_extra_edges = 0;
512
513 while (list < last) {
514 struct commit_list *parent;
515 int edge_value;
516 uint32_t packedDate[2];
517
518 parse_commit(*list);
519 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
520
521 parent = (*list)->parents;
522
523 if (!parent)
524 edge_value = GRAPH_PARENT_NONE;
525 else {
526 edge_value = sha1_pos(parent->item->object.oid.hash,
527 commits,
528 nr_commits,
529 commit_to_sha1);
530
531 if (edge_value < 0)
532 BUG("missing parent %s for commit %s",
533 oid_to_hex(&parent->item->object.oid),
534 oid_to_hex(&(*list)->object.oid));
535 }
536
537 hashwrite_be32(f, edge_value);
538
539 if (parent)
540 parent = parent->next;
541
542 if (!parent)
543 edge_value = GRAPH_PARENT_NONE;
544 else if (parent->next)
545 edge_value = GRAPH_OCTOPUS_EDGES_NEEDED | num_extra_edges;
546 else {
547 edge_value = sha1_pos(parent->item->object.oid.hash,
548 commits,
549 nr_commits,
550 commit_to_sha1);
551 if (edge_value < 0)
552 BUG("missing parent %s for commit %s",
553 oid_to_hex(&parent->item->object.oid),
554 oid_to_hex(&(*list)->object.oid));
555 }
556
557 hashwrite_be32(f, edge_value);
558
559 if (edge_value & GRAPH_OCTOPUS_EDGES_NEEDED) {
560 do {
561 num_extra_edges++;
562 parent = parent->next;
563 } while (parent);
564 }
565
566 if (sizeof((*list)->date) > 4)
567 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
568 else
569 packedDate[0] = 0;
570
571 packedDate[0] |= htonl((*list)->generation << 2);
572
573 packedDate[1] = htonl((*list)->date);
574 hashwrite(f, packedDate, 8);
575
576 list++;
577 }
578 }
579
580 static void write_graph_chunk_large_edges(struct hashfile *f,
581 struct commit **commits,
582 int nr_commits)
583 {
584 struct commit **list = commits;
585 struct commit **last = commits + nr_commits;
586 struct commit_list *parent;
587
588 while (list < last) {
589 int num_parents = 0;
590 for (parent = (*list)->parents; num_parents < 3 && parent;
591 parent = parent->next)
592 num_parents++;
593
594 if (num_parents <= 2) {
595 list++;
596 continue;
597 }
598
599 /* Since num_parents > 2, this initializer is safe. */
600 for (parent = (*list)->parents->next; parent; parent = parent->next) {
601 int edge_value = sha1_pos(parent->item->object.oid.hash,
602 commits,
603 nr_commits,
604 commit_to_sha1);
605
606 if (edge_value < 0)
607 BUG("missing parent %s for commit %s",
608 oid_to_hex(&parent->item->object.oid),
609 oid_to_hex(&(*list)->object.oid));
610 else if (!parent->next)
611 edge_value |= GRAPH_LAST_EDGE;
612
613 hashwrite_be32(f, edge_value);
614 }
615
616 list++;
617 }
618 }
619
620 static int commit_compare(const void *_a, const void *_b)
621 {
622 const struct object_id *a = (const struct object_id *)_a;
623 const struct object_id *b = (const struct object_id *)_b;
624 return oidcmp(a, b);
625 }
626
627 struct packed_commit_list {
628 struct commit **list;
629 int nr;
630 int alloc;
631 };
632
633 struct packed_oid_list {
634 struct object_id *list;
635 int nr;
636 int alloc;
637 struct progress *progress;
638 int progress_done;
639 };
640
641 static int add_packed_commits(const struct object_id *oid,
642 struct packed_git *pack,
643 uint32_t pos,
644 void *data)
645 {
646 struct packed_oid_list *list = (struct packed_oid_list*)data;
647 enum object_type type;
648 off_t offset = nth_packed_object_offset(pack, pos);
649 struct object_info oi = OBJECT_INFO_INIT;
650
651 if (list->progress)
652 display_progress(list->progress, ++list->progress_done);
653
654 oi.typep = &type;
655 if (packed_object_info(the_repository, pack, offset, &oi) < 0)
656 die(_("unable to get type of object %s"), oid_to_hex(oid));
657
658 if (type != OBJ_COMMIT)
659 return 0;
660
661 ALLOC_GROW(list->list, list->nr + 1, list->alloc);
662 oidcpy(&(list->list[list->nr]), oid);
663 list->nr++;
664
665 return 0;
666 }
667
668 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
669 {
670 struct commit_list *parent;
671 for (parent = commit->parents; parent; parent = parent->next) {
672 if (!(parent->item->object.flags & UNINTERESTING)) {
673 ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
674 oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
675 oids->nr++;
676 parent->item->object.flags |= UNINTERESTING;
677 }
678 }
679 }
680
681 static void close_reachable(struct packed_oid_list *oids, int report_progress)
682 {
683 int i, j;
684 struct commit *commit;
685 struct progress *progress = NULL;
686
687 if (report_progress)
688 progress = start_delayed_progress(
689 _("Loading known commits in commit graph"), j = 0);
690 for (i = 0; i < oids->nr; i++) {
691 display_progress(progress, ++j);
692 commit = lookup_commit(the_repository, &oids->list[i]);
693 if (commit)
694 commit->object.flags |= UNINTERESTING;
695 }
696 stop_progress(&progress);
697
698 /*
699 * As this loop runs, oids->nr may grow, but not more
700 * than the number of missing commits in the reachable
701 * closure.
702 */
703 if (report_progress)
704 progress = start_delayed_progress(
705 _("Expanding reachable commits in commit graph"), j = 0);
706 for (i = 0; i < oids->nr; i++) {
707 display_progress(progress, ++j);
708 commit = lookup_commit(the_repository, &oids->list[i]);
709
710 if (commit && !parse_commit(commit))
711 add_missing_parents(oids, commit);
712 }
713 stop_progress(&progress);
714
715 if (report_progress)
716 progress = start_delayed_progress(
717 _("Clearing commit marks in commit graph"), j = 0);
718 for (i = 0; i < oids->nr; i++) {
719 display_progress(progress, ++j);
720 commit = lookup_commit(the_repository, &oids->list[i]);
721
722 if (commit)
723 commit->object.flags &= ~UNINTERESTING;
724 }
725 stop_progress(&progress);
726 }
727
728 static void compute_generation_numbers(struct packed_commit_list* commits,
729 int report_progress)
730 {
731 int i;
732 struct commit_list *list = NULL;
733 struct progress *progress = NULL;
734
735 if (report_progress)
736 progress = start_progress(
737 _("Computing commit graph generation numbers"),
738 commits->nr);
739 for (i = 0; i < commits->nr; i++) {
740 display_progress(progress, i + 1);
741 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
742 commits->list[i]->generation != GENERATION_NUMBER_ZERO)
743 continue;
744
745 commit_list_insert(commits->list[i], &list);
746 while (list) {
747 struct commit *current = list->item;
748 struct commit_list *parent;
749 int all_parents_computed = 1;
750 uint32_t max_generation = 0;
751
752 for (parent = current->parents; parent; parent = parent->next) {
753 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
754 parent->item->generation == GENERATION_NUMBER_ZERO) {
755 all_parents_computed = 0;
756 commit_list_insert(parent->item, &list);
757 break;
758 } else if (parent->item->generation > max_generation) {
759 max_generation = parent->item->generation;
760 }
761 }
762
763 if (all_parents_computed) {
764 current->generation = max_generation + 1;
765 pop_commit(&list);
766
767 if (current->generation > GENERATION_NUMBER_MAX)
768 current->generation = GENERATION_NUMBER_MAX;
769 }
770 }
771 }
772 stop_progress(&progress);
773 }
774
775 static int add_ref_to_list(const char *refname,
776 const struct object_id *oid,
777 int flags, void *cb_data)
778 {
779 struct string_list *list = (struct string_list *)cb_data;
780
781 string_list_append(list, oid_to_hex(oid));
782 return 0;
783 }
784
785 void write_commit_graph_reachable(const char *obj_dir, int append,
786 int report_progress)
787 {
788 struct string_list list = STRING_LIST_INIT_DUP;
789
790 for_each_ref(add_ref_to_list, &list);
791 write_commit_graph(obj_dir, NULL, &list, append, report_progress);
792
793 string_list_clear(&list, 0);
794 }
795
796 void write_commit_graph(const char *obj_dir,
797 struct string_list *pack_indexes,
798 struct string_list *commit_hex,
799 int append, int report_progress)
800 {
801 struct packed_oid_list oids;
802 struct packed_commit_list commits;
803 struct hashfile *f;
804 uint32_t i, count_distinct = 0;
805 char *graph_name;
806 struct lock_file lk = LOCK_INIT;
807 uint32_t chunk_ids[5];
808 uint64_t chunk_offsets[5];
809 int num_chunks;
810 int num_extra_edges;
811 struct commit_list *parent;
812 struct progress *progress = NULL;
813 const unsigned hashsz = the_hash_algo->rawsz;
814
815 if (!commit_graph_compatible(the_repository))
816 return;
817
818 oids.nr = 0;
819 oids.alloc = approximate_object_count() / 32;
820 oids.progress = NULL;
821 oids.progress_done = 0;
822
823 if (append) {
824 prepare_commit_graph_one(the_repository, obj_dir);
825 if (the_repository->objects->commit_graph)
826 oids.alloc += the_repository->objects->commit_graph->num_commits;
827 }
828
829 if (oids.alloc < 1024)
830 oids.alloc = 1024;
831 ALLOC_ARRAY(oids.list, oids.alloc);
832
833 if (append && the_repository->objects->commit_graph) {
834 struct commit_graph *commit_graph =
835 the_repository->objects->commit_graph;
836 for (i = 0; i < commit_graph->num_commits; i++) {
837 const unsigned char *hash = commit_graph->chunk_oid_lookup +
838 commit_graph->hash_len * i;
839 hashcpy(oids.list[oids.nr++].hash, hash);
840 }
841 }
842
843 if (pack_indexes) {
844 struct strbuf packname = STRBUF_INIT;
845 int dirlen;
846 strbuf_addf(&packname, "%s/pack/", obj_dir);
847 dirlen = packname.len;
848 if (report_progress) {
849 oids.progress = start_delayed_progress(
850 _("Finding commits for commit graph"), 0);
851 oids.progress_done = 0;
852 }
853 for (i = 0; i < pack_indexes->nr; i++) {
854 struct packed_git *p;
855 strbuf_setlen(&packname, dirlen);
856 strbuf_addstr(&packname, pack_indexes->items[i].string);
857 p = add_packed_git(packname.buf, packname.len, 1);
858 if (!p)
859 die(_("error adding pack %s"), packname.buf);
860 if (open_pack_index(p))
861 die(_("error opening index for %s"), packname.buf);
862 for_each_object_in_pack(p, add_packed_commits, &oids,
863 FOR_EACH_OBJECT_PACK_ORDER);
864 close_pack(p);
865 free(p);
866 }
867 stop_progress(&oids.progress);
868 strbuf_release(&packname);
869 }
870
871 if (commit_hex) {
872 if (report_progress)
873 progress = start_delayed_progress(
874 _("Finding commits for commit graph"),
875 commit_hex->nr);
876 for (i = 0; i < commit_hex->nr; i++) {
877 const char *end;
878 struct object_id oid;
879 struct commit *result;
880
881 display_progress(progress, i + 1);
882 if (commit_hex->items[i].string &&
883 parse_oid_hex(commit_hex->items[i].string, &oid, &end))
884 continue;
885
886 result = lookup_commit_reference_gently(the_repository, &oid, 1);
887
888 if (result) {
889 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
890 oidcpy(&oids.list[oids.nr], &(result->object.oid));
891 oids.nr++;
892 }
893 }
894 stop_progress(&progress);
895 }
896
897 if (!pack_indexes && !commit_hex) {
898 if (report_progress)
899 oids.progress = start_delayed_progress(
900 _("Finding commits for commit graph"), 0);
901 for_each_packed_object(add_packed_commits, &oids,
902 FOR_EACH_OBJECT_PACK_ORDER);
903 stop_progress(&oids.progress);
904 }
905
906 close_reachable(&oids, report_progress);
907
908 QSORT(oids.list, oids.nr, commit_compare);
909
910 count_distinct = 1;
911 for (i = 1; i < oids.nr; i++) {
912 if (!oideq(&oids.list[i - 1], &oids.list[i]))
913 count_distinct++;
914 }
915
916 if (count_distinct >= GRAPH_EDGE_LAST_MASK)
917 die(_("the commit graph format cannot write %d commits"), count_distinct);
918
919 commits.nr = 0;
920 commits.alloc = count_distinct;
921 ALLOC_ARRAY(commits.list, commits.alloc);
922
923 num_extra_edges = 0;
924 for (i = 0; i < oids.nr; i++) {
925 int num_parents = 0;
926 if (i > 0 && oideq(&oids.list[i - 1], &oids.list[i]))
927 continue;
928
929 commits.list[commits.nr] = lookup_commit(the_repository, &oids.list[i]);
930 parse_commit(commits.list[commits.nr]);
931
932 for (parent = commits.list[commits.nr]->parents;
933 parent; parent = parent->next)
934 num_parents++;
935
936 if (num_parents > 2)
937 num_extra_edges += num_parents - 1;
938
939 commits.nr++;
940 }
941 num_chunks = num_extra_edges ? 4 : 3;
942
943 if (commits.nr >= GRAPH_EDGE_LAST_MASK)
944 die(_("too many commits to write graph"));
945
946 compute_generation_numbers(&commits, report_progress);
947
948 graph_name = get_commit_graph_filename(obj_dir);
949 if (safe_create_leading_directories(graph_name)) {
950 UNLEAK(graph_name);
951 die_errno(_("unable to create leading directories of %s"),
952 graph_name);
953 }
954
955 hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
956 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
957
958 hashwrite_be32(f, GRAPH_SIGNATURE);
959
960 hashwrite_u8(f, GRAPH_VERSION);
961 hashwrite_u8(f, oid_version());
962 hashwrite_u8(f, num_chunks);
963 hashwrite_u8(f, 0); /* unused padding byte */
964
965 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
966 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
967 chunk_ids[2] = GRAPH_CHUNKID_DATA;
968 if (num_extra_edges)
969 chunk_ids[3] = GRAPH_CHUNKID_LARGEEDGES;
970 else
971 chunk_ids[3] = 0;
972 chunk_ids[4] = 0;
973
974 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
975 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
976 chunk_offsets[2] = chunk_offsets[1] + hashsz * commits.nr;
977 chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * commits.nr;
978 chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
979
980 for (i = 0; i <= num_chunks; i++) {
981 uint32_t chunk_write[3];
982
983 chunk_write[0] = htonl(chunk_ids[i]);
984 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
985 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
986 hashwrite(f, chunk_write, 12);
987 }
988
989 write_graph_chunk_fanout(f, commits.list, commits.nr);
990 write_graph_chunk_oids(f, hashsz, commits.list, commits.nr);
991 write_graph_chunk_data(f, hashsz, commits.list, commits.nr);
992 write_graph_chunk_large_edges(f, commits.list, commits.nr);
993
994 close_commit_graph(the_repository);
995 finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
996 commit_lock_file(&lk);
997
998 free(graph_name);
999 free(commits.list);
1000 free(oids.list);
1001 }
1002
1003 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1004 static int verify_commit_graph_error;
1005
1006 static void graph_report(const char *fmt, ...)
1007 {
1008 va_list ap;
1009
1010 verify_commit_graph_error = 1;
1011 va_start(ap, fmt);
1012 vfprintf(stderr, fmt, ap);
1013 fprintf(stderr, "\n");
1014 va_end(ap);
1015 }
1016
1017 #define GENERATION_ZERO_EXISTS 1
1018 #define GENERATION_NUMBER_EXISTS 2
1019
1020 int verify_commit_graph(struct repository *r, struct commit_graph *g)
1021 {
1022 uint32_t i, cur_fanout_pos = 0;
1023 struct object_id prev_oid, cur_oid, checksum;
1024 int generation_zero = 0;
1025 struct hashfile *f;
1026 int devnull;
1027 struct progress *progress = NULL;
1028
1029 if (!g) {
1030 graph_report("no commit-graph file loaded");
1031 return 1;
1032 }
1033
1034 verify_commit_graph_error = 0;
1035
1036 if (!g->chunk_oid_fanout)
1037 graph_report("commit-graph is missing the OID Fanout chunk");
1038 if (!g->chunk_oid_lookup)
1039 graph_report("commit-graph is missing the OID Lookup chunk");
1040 if (!g->chunk_commit_data)
1041 graph_report("commit-graph is missing the Commit Data chunk");
1042
1043 if (verify_commit_graph_error)
1044 return verify_commit_graph_error;
1045
1046 devnull = open("/dev/null", O_WRONLY);
1047 f = hashfd(devnull, NULL);
1048 hashwrite(f, g->data, g->data_len - g->hash_len);
1049 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1050 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1051 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1052 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1053 }
1054
1055 for (i = 0; i < g->num_commits; i++) {
1056 struct commit *graph_commit;
1057
1058 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1059
1060 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1061 graph_report("commit-graph has incorrect OID order: %s then %s",
1062 oid_to_hex(&prev_oid),
1063 oid_to_hex(&cur_oid));
1064
1065 oidcpy(&prev_oid, &cur_oid);
1066
1067 while (cur_oid.hash[0] > cur_fanout_pos) {
1068 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1069
1070 if (i != fanout_value)
1071 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1072 cur_fanout_pos, fanout_value, i);
1073 cur_fanout_pos++;
1074 }
1075
1076 graph_commit = lookup_commit(r, &cur_oid);
1077 if (!parse_commit_in_graph_one(r, g, graph_commit))
1078 graph_report("failed to parse %s from commit-graph",
1079 oid_to_hex(&cur_oid));
1080 }
1081
1082 while (cur_fanout_pos < 256) {
1083 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1084
1085 if (g->num_commits != fanout_value)
1086 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1087 cur_fanout_pos, fanout_value, i);
1088
1089 cur_fanout_pos++;
1090 }
1091
1092 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1093 return verify_commit_graph_error;
1094
1095 progress = start_progress(_("Verifying commits in commit graph"),
1096 g->num_commits);
1097 for (i = 0; i < g->num_commits; i++) {
1098 struct commit *graph_commit, *odb_commit;
1099 struct commit_list *graph_parents, *odb_parents;
1100 uint32_t max_generation = 0;
1101
1102 display_progress(progress, i + 1);
1103 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1104
1105 graph_commit = lookup_commit(r, &cur_oid);
1106 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1107 if (parse_commit_internal(odb_commit, 0, 0)) {
1108 graph_report("failed to parse %s from object database",
1109 oid_to_hex(&cur_oid));
1110 continue;
1111 }
1112
1113 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
1114 get_commit_tree_oid(odb_commit)))
1115 graph_report("root tree OID for commit %s in commit-graph is %s != %s",
1116 oid_to_hex(&cur_oid),
1117 oid_to_hex(get_commit_tree_oid(graph_commit)),
1118 oid_to_hex(get_commit_tree_oid(odb_commit)));
1119
1120 graph_parents = graph_commit->parents;
1121 odb_parents = odb_commit->parents;
1122
1123 while (graph_parents) {
1124 if (odb_parents == NULL) {
1125 graph_report("commit-graph parent list for commit %s is too long",
1126 oid_to_hex(&cur_oid));
1127 break;
1128 }
1129
1130 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1131 graph_report("commit-graph parent for %s is %s != %s",
1132 oid_to_hex(&cur_oid),
1133 oid_to_hex(&graph_parents->item->object.oid),
1134 oid_to_hex(&odb_parents->item->object.oid));
1135
1136 if (graph_parents->item->generation > max_generation)
1137 max_generation = graph_parents->item->generation;
1138
1139 graph_parents = graph_parents->next;
1140 odb_parents = odb_parents->next;
1141 }
1142
1143 if (odb_parents != NULL)
1144 graph_report("commit-graph parent list for commit %s terminates early",
1145 oid_to_hex(&cur_oid));
1146
1147 if (!graph_commit->generation) {
1148 if (generation_zero == GENERATION_NUMBER_EXISTS)
1149 graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
1150 oid_to_hex(&cur_oid));
1151 generation_zero = GENERATION_ZERO_EXISTS;
1152 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1153 graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
1154 oid_to_hex(&cur_oid));
1155
1156 if (generation_zero == GENERATION_ZERO_EXISTS)
1157 continue;
1158
1159 /*
1160 * If one of our parents has generation GENERATION_NUMBER_MAX, then
1161 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1162 * extra logic in the following condition.
1163 */
1164 if (max_generation == GENERATION_NUMBER_MAX)
1165 max_generation--;
1166
1167 if (graph_commit->generation != max_generation + 1)
1168 graph_report("commit-graph generation for commit %s is %u != %u",
1169 oid_to_hex(&cur_oid),
1170 graph_commit->generation,
1171 max_generation + 1);
1172
1173 if (graph_commit->date != odb_commit->date)
1174 graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
1175 oid_to_hex(&cur_oid),
1176 graph_commit->date,
1177 odb_commit->date);
1178 }
1179 stop_progress(&progress);
1180
1181 return verify_commit_graph_error;
1182 }
1183
1184 void free_commit_graph(struct commit_graph *g)
1185 {
1186 if (!g)
1187 return;
1188 if (g->graph_fd >= 0) {
1189 munmap((void *)g->data, g->data_len);
1190 g->data = NULL;
1191 close(g->graph_fd);
1192 }
1193 free(g);
1194 }