The third batch
[git.git] / commit-graph.c
blob47e9be0a3aad883c17221b11c972b8376a70a555
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"
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_EXTRAEDGES 0x45444745 /* "EDGE" */
26 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
28 #define GRAPH_VERSION_1 0x1
29 #define GRAPH_VERSION GRAPH_VERSION_1
31 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
32 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
33 #define GRAPH_PARENT_NONE 0x70000000
35 #define GRAPH_LAST_EDGE 0x80000000
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)
43 char *get_commit_graph_filename(const char *obj_dir)
45 return xstrfmt("%s/info/commit-graph", obj_dir);
48 static uint8_t oid_version(void)
50 return 1;
53 static struct commit_graph *alloc_commit_graph(void)
55 struct commit_graph *g = xcalloc(1, sizeof(*g));
56 g->graph_fd = -1;
58 return g;
61 extern int read_replace_refs;
63 static int commit_graph_compatible(struct repository *r)
65 if (!r->gitdir)
66 return 0;
68 if (read_replace_refs) {
69 prepare_replace_object(r);
70 if (hashmap_get_size(&r->objects->replace_map->map))
71 return 0;
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;
80 return 1;
83 struct commit_graph *load_commit_graph_one(const char *graph_file)
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);
91 if (fd < 0)
92 return NULL;
93 if (fstat(fd, &st)) {
94 close(fd);
95 return NULL;
97 graph_size = xsize_t(st.st_size);
99 if (graph_size < GRAPH_MIN_SIZE) {
100 close(fd);
101 die(_("graph file %s is too small"), graph_file);
103 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
104 ret = parse_commit_graph(graph_map, fd, graph_size);
106 if (!ret) {
107 munmap(graph_map, graph_size);
108 close(fd);
109 exit(1);
112 return ret;
115 struct commit_graph *parse_commit_graph(void *graph_map, int fd,
116 size_t graph_size)
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;
126 if (!graph_map)
127 return NULL;
129 if (graph_size < GRAPH_MIN_SIZE)
130 return NULL;
132 data = (const unsigned char *)graph_map;
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;
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;
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;
155 graph = alloc_commit_graph();
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;
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;
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;
178 chunk_id = get_be32(chunk_lookup + 0);
179 chunk_offset = get_be64(chunk_lookup + 4);
181 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
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;
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;
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;
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;
212 case GRAPH_CHUNKID_EXTRAEDGES:
213 if (graph->chunk_extra_edges)
214 chunk_repeated = 1;
215 else
216 graph->chunk_extra_edges = data + chunk_offset;
217 break;
220 if (chunk_repeated) {
221 error(_("chunk id %08x appears multiple times"), chunk_id);
222 free(graph);
223 return NULL;
226 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
228 graph->num_commits = (chunk_offset - last_chunk_offset)
229 / graph->hash_len;
232 last_chunk_id = chunk_id;
233 last_chunk_offset = chunk_offset;
236 return graph;
239 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
241 char *graph_name;
243 if (r->objects->commit_graph)
244 return;
246 graph_name = get_commit_graph_filename(obj_dir);
247 r->objects->commit_graph =
248 load_commit_graph_one(graph_name);
250 FREE_AND_NULL(graph_name);
254 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
256 * On the first invocation, this function attemps to load the commit
257 * graph if the_repository is configured to have one.
259 static int prepare_commit_graph(struct repository *r)
261 struct object_directory *odb;
262 int config_value;
264 if (r->objects->commit_graph_attempted)
265 return !!r->objects->commit_graph;
266 r->objects->commit_graph_attempted = 1;
268 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
269 (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
270 !config_value))
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.)
277 return 0;
279 if (!commit_graph_compatible(r))
280 return 0;
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;
290 int generation_numbers_enabled(struct repository *r)
292 uint32_t first_generation;
293 struct commit_graph *g;
294 if (!prepare_commit_graph(r))
295 return 0;
297 g = r->objects->commit_graph;
299 if (!g->num_commits)
300 return 0;
302 first_generation = get_be32(g->chunk_commit_data +
303 g->hash_len + 8) >> 2;
305 return !!first_generation;
308 void close_commit_graph(struct repository *r)
310 free_commit_graph(r->objects->commit_graph);
311 r->objects->commit_graph = NULL;
314 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
316 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
317 g->chunk_oid_lookup, g->hash_len, pos);
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)
325 struct commit *c;
326 struct object_id oid;
328 if (pos >= g->num_commits)
329 die("invalid parent position %"PRIu64, pos);
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;
339 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
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;
346 static int fill_commit_in_graph(struct repository *r,
347 struct commit *item,
348 struct commit_graph *g, uint32_t pos)
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;
356 item->object.parsed = 1;
357 item->graph_pos = pos;
359 item->maybe_tree = NULL;
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);
365 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
367 pptr = &item->parents;
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);
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_EXTRA_EDGES_NEEDED)) {
378 pptr = insert_parent_or_die(r, g, edge_value, pptr);
379 return 1;
382 parent_data_ptr = (uint32_t*)(g->chunk_extra_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));
392 return 1;
395 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
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);
405 static int parse_commit_in_graph_one(struct repository *r,
406 struct commit_graph *g,
407 struct commit *item)
409 uint32_t pos;
411 if (item->object.parsed)
412 return 1;
414 if (find_commit_in_graph(item, g, &pos))
415 return fill_commit_in_graph(r, item, g, pos);
417 return 0;
420 int parse_commit_in_graph(struct repository *r, struct commit *item)
422 if (!prepare_commit_graph(r))
423 return 0;
424 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
427 void load_commit_graph_info(struct repository *r, struct commit *item)
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);
436 static struct tree *load_tree_for_commit(struct repository *r,
437 struct commit_graph *g,
438 struct commit *c)
440 struct object_id oid;
441 const unsigned char *commit_data = g->chunk_commit_data +
442 GRAPH_DATA_WIDTH * (c->graph_pos);
444 hashcpy(oid.hash, commit_data);
445 c->maybe_tree = lookup_tree(r, &oid);
447 return c->maybe_tree;
450 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
451 struct commit_graph *g,
452 const struct commit *c)
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");
459 return load_tree_for_commit(r, g, (struct commit *)c);
462 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
464 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
467 static void write_graph_chunk_fanout(struct hashfile *f,
468 struct commit **commits,
469 int nr_commits,
470 struct progress *progress,
471 uint64_t *progress_cnt)
473 int i, count = 0;
474 struct commit **list = commits;
477 * Write the first-level table (the list is sorted,
478 * but we use a 256-entry lookup to be able to avoid
479 * having to do eight extra binary search iterations).
481 for (i = 0; i < 256; i++) {
482 while (count < nr_commits) {
483 if ((*list)->object.oid.hash[0] != i)
484 break;
485 display_progress(progress, ++*progress_cnt);
486 count++;
487 list++;
490 hashwrite_be32(f, count);
494 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
495 struct commit **commits, int nr_commits,
496 struct progress *progress,
497 uint64_t *progress_cnt)
499 struct commit **list = commits;
500 int count;
501 for (count = 0; count < nr_commits; count++, list++) {
502 display_progress(progress, ++*progress_cnt);
503 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
507 static const unsigned char *commit_to_sha1(size_t index, void *table)
509 struct commit **commits = table;
510 return commits[index]->object.oid.hash;
513 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
514 struct commit **commits, int nr_commits,
515 struct progress *progress,
516 uint64_t *progress_cnt)
518 struct commit **list = commits;
519 struct commit **last = commits + nr_commits;
520 uint32_t num_extra_edges = 0;
522 while (list < last) {
523 struct commit_list *parent;
524 int edge_value;
525 uint32_t packedDate[2];
526 display_progress(progress, ++*progress_cnt);
528 parse_commit(*list);
529 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
531 parent = (*list)->parents;
533 if (!parent)
534 edge_value = GRAPH_PARENT_NONE;
535 else {
536 edge_value = sha1_pos(parent->item->object.oid.hash,
537 commits,
538 nr_commits,
539 commit_to_sha1);
541 if (edge_value < 0)
542 BUG("missing parent %s for commit %s",
543 oid_to_hex(&parent->item->object.oid),
544 oid_to_hex(&(*list)->object.oid));
547 hashwrite_be32(f, edge_value);
549 if (parent)
550 parent = parent->next;
552 if (!parent)
553 edge_value = GRAPH_PARENT_NONE;
554 else if (parent->next)
555 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
556 else {
557 edge_value = sha1_pos(parent->item->object.oid.hash,
558 commits,
559 nr_commits,
560 commit_to_sha1);
561 if (edge_value < 0)
562 BUG("missing parent %s for commit %s",
563 oid_to_hex(&parent->item->object.oid),
564 oid_to_hex(&(*list)->object.oid));
567 hashwrite_be32(f, edge_value);
569 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
570 do {
571 num_extra_edges++;
572 parent = parent->next;
573 } while (parent);
576 if (sizeof((*list)->date) > 4)
577 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
578 else
579 packedDate[0] = 0;
581 packedDate[0] |= htonl((*list)->generation << 2);
583 packedDate[1] = htonl((*list)->date);
584 hashwrite(f, packedDate, 8);
586 list++;
590 static void write_graph_chunk_extra_edges(struct hashfile *f,
591 struct commit **commits,
592 int nr_commits,
593 struct progress *progress,
594 uint64_t *progress_cnt)
596 struct commit **list = commits;
597 struct commit **last = commits + nr_commits;
598 struct commit_list *parent;
600 while (list < last) {
601 int num_parents = 0;
603 display_progress(progress, ++*progress_cnt);
605 for (parent = (*list)->parents; num_parents < 3 && parent;
606 parent = parent->next)
607 num_parents++;
609 if (num_parents <= 2) {
610 list++;
611 continue;
614 /* Since num_parents > 2, this initializer is safe. */
615 for (parent = (*list)->parents->next; parent; parent = parent->next) {
616 int edge_value = sha1_pos(parent->item->object.oid.hash,
617 commits,
618 nr_commits,
619 commit_to_sha1);
621 if (edge_value < 0)
622 BUG("missing parent %s for commit %s",
623 oid_to_hex(&parent->item->object.oid),
624 oid_to_hex(&(*list)->object.oid));
625 else if (!parent->next)
626 edge_value |= GRAPH_LAST_EDGE;
628 hashwrite_be32(f, edge_value);
631 list++;
635 static int commit_compare(const void *_a, const void *_b)
637 const struct object_id *a = (const struct object_id *)_a;
638 const struct object_id *b = (const struct object_id *)_b;
639 return oidcmp(a, b);
642 struct packed_commit_list {
643 struct commit **list;
644 int nr;
645 int alloc;
648 struct packed_oid_list {
649 struct object_id *list;
650 int nr;
651 int alloc;
652 struct progress *progress;
653 int progress_done;
656 static int add_packed_commits(const struct object_id *oid,
657 struct packed_git *pack,
658 uint32_t pos,
659 void *data)
661 struct packed_oid_list *list = (struct packed_oid_list*)data;
662 enum object_type type;
663 off_t offset = nth_packed_object_offset(pack, pos);
664 struct object_info oi = OBJECT_INFO_INIT;
666 if (list->progress)
667 display_progress(list->progress, ++list->progress_done);
669 oi.typep = &type;
670 if (packed_object_info(the_repository, pack, offset, &oi) < 0)
671 die(_("unable to get type of object %s"), oid_to_hex(oid));
673 if (type != OBJ_COMMIT)
674 return 0;
676 ALLOC_GROW(list->list, list->nr + 1, list->alloc);
677 oidcpy(&(list->list[list->nr]), oid);
678 list->nr++;
680 return 0;
683 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
685 struct commit_list *parent;
686 for (parent = commit->parents; parent; parent = parent->next) {
687 if (!(parent->item->object.flags & UNINTERESTING)) {
688 ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
689 oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
690 oids->nr++;
691 parent->item->object.flags |= UNINTERESTING;
696 static void close_reachable(struct packed_oid_list *oids, int report_progress)
698 int i;
699 struct commit *commit;
700 struct progress *progress = NULL;
702 if (report_progress)
703 progress = start_delayed_progress(
704 _("Loading known commits in commit graph"), oids->nr);
705 for (i = 0; i < oids->nr; i++) {
706 display_progress(progress, i + 1);
707 commit = lookup_commit(the_repository, &oids->list[i]);
708 if (commit)
709 commit->object.flags |= UNINTERESTING;
711 stop_progress(&progress);
714 * As this loop runs, oids->nr may grow, but not more
715 * than the number of missing commits in the reachable
716 * closure.
718 if (report_progress)
719 progress = start_delayed_progress(
720 _("Expanding reachable commits in commit graph"), oids->nr);
721 for (i = 0; i < oids->nr; i++) {
722 display_progress(progress, i + 1);
723 commit = lookup_commit(the_repository, &oids->list[i]);
725 if (commit && !parse_commit(commit))
726 add_missing_parents(oids, commit);
728 stop_progress(&progress);
730 if (report_progress)
731 progress = start_delayed_progress(
732 _("Clearing commit marks in commit graph"), oids->nr);
733 for (i = 0; i < oids->nr; i++) {
734 display_progress(progress, i + 1);
735 commit = lookup_commit(the_repository, &oids->list[i]);
737 if (commit)
738 commit->object.flags &= ~UNINTERESTING;
740 stop_progress(&progress);
743 static void compute_generation_numbers(struct packed_commit_list* commits,
744 int report_progress)
746 int i;
747 struct commit_list *list = NULL;
748 struct progress *progress = NULL;
750 if (report_progress)
751 progress = start_progress(
752 _("Computing commit graph generation numbers"),
753 commits->nr);
754 for (i = 0; i < commits->nr; i++) {
755 display_progress(progress, i + 1);
756 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
757 commits->list[i]->generation != GENERATION_NUMBER_ZERO)
758 continue;
760 commit_list_insert(commits->list[i], &list);
761 while (list) {
762 struct commit *current = list->item;
763 struct commit_list *parent;
764 int all_parents_computed = 1;
765 uint32_t max_generation = 0;
767 for (parent = current->parents; parent; parent = parent->next) {
768 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
769 parent->item->generation == GENERATION_NUMBER_ZERO) {
770 all_parents_computed = 0;
771 commit_list_insert(parent->item, &list);
772 break;
773 } else if (parent->item->generation > max_generation) {
774 max_generation = parent->item->generation;
778 if (all_parents_computed) {
779 current->generation = max_generation + 1;
780 pop_commit(&list);
782 if (current->generation > GENERATION_NUMBER_MAX)
783 current->generation = GENERATION_NUMBER_MAX;
787 stop_progress(&progress);
790 static int add_ref_to_list(const char *refname,
791 const struct object_id *oid,
792 int flags, void *cb_data)
794 struct string_list *list = (struct string_list *)cb_data;
796 string_list_append(list, oid_to_hex(oid));
797 return 0;
800 void write_commit_graph_reachable(const char *obj_dir, int append,
801 int report_progress)
803 struct string_list list = STRING_LIST_INIT_DUP;
805 for_each_ref(add_ref_to_list, &list);
806 write_commit_graph(obj_dir, NULL, &list, append, report_progress);
808 string_list_clear(&list, 0);
811 void write_commit_graph(const char *obj_dir,
812 struct string_list *pack_indexes,
813 struct string_list *commit_hex,
814 int append, int report_progress)
816 struct packed_oid_list oids;
817 struct packed_commit_list commits;
818 struct hashfile *f;
819 uint32_t i, count_distinct = 0;
820 char *graph_name;
821 struct lock_file lk = LOCK_INIT;
822 uint32_t chunk_ids[5];
823 uint64_t chunk_offsets[5];
824 int num_chunks;
825 int num_extra_edges;
826 struct commit_list *parent;
827 struct progress *progress = NULL;
828 const unsigned hashsz = the_hash_algo->rawsz;
829 uint64_t progress_cnt = 0;
830 struct strbuf progress_title = STRBUF_INIT;
831 unsigned long approx_nr_objects;
833 if (!commit_graph_compatible(the_repository))
834 return;
836 oids.nr = 0;
837 approx_nr_objects = approximate_object_count();
838 oids.alloc = approx_nr_objects / 32;
839 oids.progress = NULL;
840 oids.progress_done = 0;
842 if (append) {
843 prepare_commit_graph_one(the_repository, obj_dir);
844 if (the_repository->objects->commit_graph)
845 oids.alloc += the_repository->objects->commit_graph->num_commits;
848 if (oids.alloc < 1024)
849 oids.alloc = 1024;
850 ALLOC_ARRAY(oids.list, oids.alloc);
852 if (append && the_repository->objects->commit_graph) {
853 struct commit_graph *commit_graph =
854 the_repository->objects->commit_graph;
855 for (i = 0; i < commit_graph->num_commits; i++) {
856 const unsigned char *hash = commit_graph->chunk_oid_lookup +
857 commit_graph->hash_len * i;
858 hashcpy(oids.list[oids.nr++].hash, hash);
862 if (pack_indexes) {
863 struct strbuf packname = STRBUF_INIT;
864 int dirlen;
865 strbuf_addf(&packname, "%s/pack/", obj_dir);
866 dirlen = packname.len;
867 if (report_progress) {
868 strbuf_addf(&progress_title,
869 Q_("Finding commits for commit graph in %d pack",
870 "Finding commits for commit graph in %d packs",
871 pack_indexes->nr),
872 pack_indexes->nr);
873 oids.progress = start_delayed_progress(progress_title.buf, 0);
874 oids.progress_done = 0;
876 for (i = 0; i < pack_indexes->nr; i++) {
877 struct packed_git *p;
878 strbuf_setlen(&packname, dirlen);
879 strbuf_addstr(&packname, pack_indexes->items[i].string);
880 p = add_packed_git(packname.buf, packname.len, 1);
881 if (!p)
882 die(_("error adding pack %s"), packname.buf);
883 if (open_pack_index(p))
884 die(_("error opening index for %s"), packname.buf);
885 for_each_object_in_pack(p, add_packed_commits, &oids,
886 FOR_EACH_OBJECT_PACK_ORDER);
887 close_pack(p);
888 free(p);
890 stop_progress(&oids.progress);
891 strbuf_reset(&progress_title);
892 strbuf_release(&packname);
895 if (commit_hex) {
896 if (report_progress) {
897 strbuf_addf(&progress_title,
898 Q_("Finding commits for commit graph from %d ref",
899 "Finding commits for commit graph from %d refs",
900 commit_hex->nr),
901 commit_hex->nr);
902 progress = start_delayed_progress(progress_title.buf,
903 commit_hex->nr);
905 for (i = 0; i < commit_hex->nr; i++) {
906 const char *end;
907 struct object_id oid;
908 struct commit *result;
910 display_progress(progress, i + 1);
911 if (commit_hex->items[i].string &&
912 parse_oid_hex(commit_hex->items[i].string, &oid, &end))
913 continue;
915 result = lookup_commit_reference_gently(the_repository, &oid, 1);
917 if (result) {
918 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
919 oidcpy(&oids.list[oids.nr], &(result->object.oid));
920 oids.nr++;
923 stop_progress(&progress);
924 strbuf_reset(&progress_title);
927 if (!pack_indexes && !commit_hex) {
928 if (report_progress)
929 oids.progress = start_delayed_progress(
930 _("Finding commits for commit graph among packed objects"),
931 approx_nr_objects);
932 for_each_packed_object(add_packed_commits, &oids,
933 FOR_EACH_OBJECT_PACK_ORDER);
934 if (oids.progress_done < approx_nr_objects)
935 display_progress(oids.progress, approx_nr_objects);
936 stop_progress(&oids.progress);
939 close_reachable(&oids, report_progress);
941 if (report_progress)
942 progress = start_delayed_progress(
943 _("Counting distinct commits in commit graph"),
944 oids.nr);
945 display_progress(progress, 0); /* TODO: Measure QSORT() progress */
946 QSORT(oids.list, oids.nr, commit_compare);
947 count_distinct = 1;
948 for (i = 1; i < oids.nr; i++) {
949 display_progress(progress, i + 1);
950 if (!oideq(&oids.list[i - 1], &oids.list[i]))
951 count_distinct++;
953 stop_progress(&progress);
955 if (count_distinct >= GRAPH_EDGE_LAST_MASK)
956 die(_("the commit graph format cannot write %d commits"), count_distinct);
958 commits.nr = 0;
959 commits.alloc = count_distinct;
960 ALLOC_ARRAY(commits.list, commits.alloc);
962 num_extra_edges = 0;
963 if (report_progress)
964 progress = start_delayed_progress(
965 _("Finding extra edges in commit graph"),
966 oids.nr);
967 for (i = 0; i < oids.nr; i++) {
968 int num_parents = 0;
969 display_progress(progress, i + 1);
970 if (i > 0 && oideq(&oids.list[i - 1], &oids.list[i]))
971 continue;
973 commits.list[commits.nr] = lookup_commit(the_repository, &oids.list[i]);
974 parse_commit(commits.list[commits.nr]);
976 for (parent = commits.list[commits.nr]->parents;
977 parent; parent = parent->next)
978 num_parents++;
980 if (num_parents > 2)
981 num_extra_edges += num_parents - 1;
983 commits.nr++;
985 num_chunks = num_extra_edges ? 4 : 3;
986 stop_progress(&progress);
988 if (commits.nr >= GRAPH_EDGE_LAST_MASK)
989 die(_("too many commits to write graph"));
991 compute_generation_numbers(&commits, report_progress);
993 graph_name = get_commit_graph_filename(obj_dir);
994 if (safe_create_leading_directories(graph_name)) {
995 UNLEAK(graph_name);
996 die_errno(_("unable to create leading directories of %s"),
997 graph_name);
1000 hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
1001 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1003 hashwrite_be32(f, GRAPH_SIGNATURE);
1005 hashwrite_u8(f, GRAPH_VERSION);
1006 hashwrite_u8(f, oid_version());
1007 hashwrite_u8(f, num_chunks);
1008 hashwrite_u8(f, 0); /* unused padding byte */
1010 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1011 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1012 chunk_ids[2] = GRAPH_CHUNKID_DATA;
1013 if (num_extra_edges)
1014 chunk_ids[3] = GRAPH_CHUNKID_EXTRAEDGES;
1015 else
1016 chunk_ids[3] = 0;
1017 chunk_ids[4] = 0;
1019 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1020 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1021 chunk_offsets[2] = chunk_offsets[1] + hashsz * commits.nr;
1022 chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * commits.nr;
1023 chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
1025 for (i = 0; i <= num_chunks; i++) {
1026 uint32_t chunk_write[3];
1028 chunk_write[0] = htonl(chunk_ids[i]);
1029 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1030 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1031 hashwrite(f, chunk_write, 12);
1034 if (report_progress) {
1035 strbuf_addf(&progress_title,
1036 Q_("Writing out commit graph in %d pass",
1037 "Writing out commit graph in %d passes",
1038 num_chunks),
1039 num_chunks);
1040 progress = start_delayed_progress(
1041 progress_title.buf,
1042 num_chunks * commits.nr);
1044 write_graph_chunk_fanout(f, commits.list, commits.nr, progress, &progress_cnt);
1045 write_graph_chunk_oids(f, hashsz, commits.list, commits.nr, progress, &progress_cnt);
1046 write_graph_chunk_data(f, hashsz, commits.list, commits.nr, progress, &progress_cnt);
1047 if (num_extra_edges)
1048 write_graph_chunk_extra_edges(f, commits.list, commits.nr, progress, &progress_cnt);
1049 stop_progress(&progress);
1050 strbuf_release(&progress_title);
1052 close_commit_graph(the_repository);
1053 finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1054 commit_lock_file(&lk);
1056 free(graph_name);
1057 free(commits.list);
1058 free(oids.list);
1061 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1062 static int verify_commit_graph_error;
1064 static void graph_report(const char *fmt, ...)
1066 va_list ap;
1068 verify_commit_graph_error = 1;
1069 va_start(ap, fmt);
1070 vfprintf(stderr, fmt, ap);
1071 fprintf(stderr, "\n");
1072 va_end(ap);
1075 #define GENERATION_ZERO_EXISTS 1
1076 #define GENERATION_NUMBER_EXISTS 2
1078 int verify_commit_graph(struct repository *r, struct commit_graph *g)
1080 uint32_t i, cur_fanout_pos = 0;
1081 struct object_id prev_oid, cur_oid, checksum;
1082 int generation_zero = 0;
1083 struct hashfile *f;
1084 int devnull;
1085 struct progress *progress = NULL;
1087 if (!g) {
1088 graph_report("no commit-graph file loaded");
1089 return 1;
1092 verify_commit_graph_error = 0;
1094 if (!g->chunk_oid_fanout)
1095 graph_report("commit-graph is missing the OID Fanout chunk");
1096 if (!g->chunk_oid_lookup)
1097 graph_report("commit-graph is missing the OID Lookup chunk");
1098 if (!g->chunk_commit_data)
1099 graph_report("commit-graph is missing the Commit Data chunk");
1101 if (verify_commit_graph_error)
1102 return verify_commit_graph_error;
1104 devnull = open("/dev/null", O_WRONLY);
1105 f = hashfd(devnull, NULL);
1106 hashwrite(f, g->data, g->data_len - g->hash_len);
1107 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1108 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1109 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1110 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1113 for (i = 0; i < g->num_commits; i++) {
1114 struct commit *graph_commit;
1116 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1118 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1119 graph_report("commit-graph has incorrect OID order: %s then %s",
1120 oid_to_hex(&prev_oid),
1121 oid_to_hex(&cur_oid));
1123 oidcpy(&prev_oid, &cur_oid);
1125 while (cur_oid.hash[0] > cur_fanout_pos) {
1126 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1128 if (i != fanout_value)
1129 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1130 cur_fanout_pos, fanout_value, i);
1131 cur_fanout_pos++;
1134 graph_commit = lookup_commit(r, &cur_oid);
1135 if (!parse_commit_in_graph_one(r, g, graph_commit))
1136 graph_report("failed to parse %s from commit-graph",
1137 oid_to_hex(&cur_oid));
1140 while (cur_fanout_pos < 256) {
1141 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1143 if (g->num_commits != fanout_value)
1144 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1145 cur_fanout_pos, fanout_value, i);
1147 cur_fanout_pos++;
1150 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1151 return verify_commit_graph_error;
1153 progress = start_progress(_("Verifying commits in commit graph"),
1154 g->num_commits);
1155 for (i = 0; i < g->num_commits; i++) {
1156 struct commit *graph_commit, *odb_commit;
1157 struct commit_list *graph_parents, *odb_parents;
1158 uint32_t max_generation = 0;
1160 display_progress(progress, i + 1);
1161 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1163 graph_commit = lookup_commit(r, &cur_oid);
1164 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1165 if (parse_commit_internal(odb_commit, 0, 0)) {
1166 graph_report("failed to parse %s from object database",
1167 oid_to_hex(&cur_oid));
1168 continue;
1171 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
1172 get_commit_tree_oid(odb_commit)))
1173 graph_report("root tree OID for commit %s in commit-graph is %s != %s",
1174 oid_to_hex(&cur_oid),
1175 oid_to_hex(get_commit_tree_oid(graph_commit)),
1176 oid_to_hex(get_commit_tree_oid(odb_commit)));
1178 graph_parents = graph_commit->parents;
1179 odb_parents = odb_commit->parents;
1181 while (graph_parents) {
1182 if (odb_parents == NULL) {
1183 graph_report("commit-graph parent list for commit %s is too long",
1184 oid_to_hex(&cur_oid));
1185 break;
1188 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1189 graph_report("commit-graph parent for %s is %s != %s",
1190 oid_to_hex(&cur_oid),
1191 oid_to_hex(&graph_parents->item->object.oid),
1192 oid_to_hex(&odb_parents->item->object.oid));
1194 if (graph_parents->item->generation > max_generation)
1195 max_generation = graph_parents->item->generation;
1197 graph_parents = graph_parents->next;
1198 odb_parents = odb_parents->next;
1201 if (odb_parents != NULL)
1202 graph_report("commit-graph parent list for commit %s terminates early",
1203 oid_to_hex(&cur_oid));
1205 if (!graph_commit->generation) {
1206 if (generation_zero == GENERATION_NUMBER_EXISTS)
1207 graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
1208 oid_to_hex(&cur_oid));
1209 generation_zero = GENERATION_ZERO_EXISTS;
1210 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1211 graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
1212 oid_to_hex(&cur_oid));
1214 if (generation_zero == GENERATION_ZERO_EXISTS)
1215 continue;
1218 * If one of our parents has generation GENERATION_NUMBER_MAX, then
1219 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1220 * extra logic in the following condition.
1222 if (max_generation == GENERATION_NUMBER_MAX)
1223 max_generation--;
1225 if (graph_commit->generation != max_generation + 1)
1226 graph_report("commit-graph generation for commit %s is %u != %u",
1227 oid_to_hex(&cur_oid),
1228 graph_commit->generation,
1229 max_generation + 1);
1231 if (graph_commit->date != odb_commit->date)
1232 graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
1233 oid_to_hex(&cur_oid),
1234 graph_commit->date,
1235 odb_commit->date);
1237 stop_progress(&progress);
1239 return verify_commit_graph_error;
1242 void free_commit_graph(struct commit_graph *g)
1244 if (!g)
1245 return;
1246 if (g->graph_fd >= 0) {
1247 munmap((void *)g->data, g->data_len);
1248 g->data = NULL;
1249 close(g->graph_fd);
1251 free(g);