4 #include "git-compat-util.h"
12 #include "sha1-lookup.h"
13 #include "commit-graph.h"
14 #include "object-store.h"
17 #include "replace-object.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)
53 static struct commit_graph *alloc_commit_graph(void)
55 struct commit_graph *g = xcalloc(1, sizeof(*g));
61 extern int read_replace_refs;
63 static int commit_graph_compatible(struct repository *r)
68 if (read_replace_refs) {
69 prepare_replace_object(r);
70 if (hashmap_get_size(&r->objects->replace_map->map))
74 prepare_commit_graft(r);
75 if (r->parsed_objects && r->parsed_objects->grafts_nr)
77 if (is_repository_shallow(r))
83 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
85 *fd = git_open(graph_file);
95 struct commit_graph *load_commit_graph_one_fd_st(int fd, struct stat *st)
99 struct commit_graph *ret;
101 graph_size = xsize_t(st->st_size);
103 if (graph_size < GRAPH_MIN_SIZE) {
105 error(_("commit-graph file is too small"));
108 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
109 ret = parse_commit_graph(graph_map, fd, graph_size);
112 munmap(graph_map, graph_size);
119 static int verify_commit_graph_lite(struct commit_graph *g)
122 * Basic validation shared between parse_commit_graph()
123 * which'll be called every time the graph is used, and the
124 * much more expensive verify_commit_graph() used by
125 * "commit-graph verify".
127 * There should only be very basic checks here to ensure that
128 * we don't e.g. segfault in fill_commit_in_graph(), but
129 * because this is a very hot codepath nothing that e.g. loops
130 * over g->num_commits, or runs a checksum on the commit-graph
133 if (!g->chunk_oid_fanout) {
134 error("commit-graph is missing the OID Fanout chunk");
137 if (!g->chunk_oid_lookup) {
138 error("commit-graph is missing the OID Lookup chunk");
141 if (!g->chunk_commit_data) {
142 error("commit-graph is missing the Commit Data chunk");
149 struct commit_graph *parse_commit_graph(void *graph_map, int fd,
152 const unsigned char *data, *chunk_lookup;
154 struct commit_graph *graph;
155 uint64_t last_chunk_offset;
156 uint32_t last_chunk_id;
157 uint32_t graph_signature;
158 unsigned char graph_version, hash_version;
163 if (graph_size < GRAPH_MIN_SIZE)
166 data = (const unsigned char *)graph_map;
168 graph_signature = get_be32(data);
169 if (graph_signature != GRAPH_SIGNATURE) {
170 error(_("commit-graph signature %X does not match signature %X"),
171 graph_signature, GRAPH_SIGNATURE);
175 graph_version = *(unsigned char*)(data + 4);
176 if (graph_version != GRAPH_VERSION) {
177 error(_("commit-graph version %X does not match version %X"),
178 graph_version, GRAPH_VERSION);
182 hash_version = *(unsigned char*)(data + 5);
183 if (hash_version != oid_version()) {
184 error(_("commit-graph hash version %X does not match version %X"),
185 hash_version, oid_version());
189 graph = alloc_commit_graph();
191 graph->hash_len = the_hash_algo->rawsz;
192 graph->num_chunks = *(unsigned char*)(data + 6);
193 graph->graph_fd = fd;
194 graph->data = graph_map;
195 graph->data_len = graph_size;
198 last_chunk_offset = 8;
199 chunk_lookup = data + 8;
200 for (i = 0; i < graph->num_chunks; i++) {
202 uint64_t chunk_offset;
203 int chunk_repeated = 0;
205 if (data + graph_size - chunk_lookup <
206 GRAPH_CHUNKLOOKUP_WIDTH) {
207 error(_("commit-graph chunk lookup table entry missing; file may be incomplete"));
212 chunk_id = get_be32(chunk_lookup + 0);
213 chunk_offset = get_be64(chunk_lookup + 4);
215 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
217 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
218 error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
219 (uint32_t)chunk_offset);
225 case GRAPH_CHUNKID_OIDFANOUT:
226 if (graph->chunk_oid_fanout)
229 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
232 case GRAPH_CHUNKID_OIDLOOKUP:
233 if (graph->chunk_oid_lookup)
236 graph->chunk_oid_lookup = data + chunk_offset;
239 case GRAPH_CHUNKID_DATA:
240 if (graph->chunk_commit_data)
243 graph->chunk_commit_data = data + chunk_offset;
246 case GRAPH_CHUNKID_EXTRAEDGES:
247 if (graph->chunk_extra_edges)
250 graph->chunk_extra_edges = data + chunk_offset;
254 if (chunk_repeated) {
255 error(_("commit-graph chunk id %08x appears multiple times"), chunk_id);
260 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
262 graph->num_commits = (chunk_offset - last_chunk_offset)
266 last_chunk_id = chunk_id;
267 last_chunk_offset = chunk_offset;
270 if (verify_commit_graph_lite(graph))
276 static struct commit_graph *load_commit_graph_one(const char *graph_file)
281 int open_ok = open_commit_graph(graph_file, &fd, &st);
286 return load_commit_graph_one_fd_st(fd, &st);
289 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
293 if (r->objects->commit_graph)
296 graph_name = get_commit_graph_filename(obj_dir);
297 r->objects->commit_graph =
298 load_commit_graph_one(graph_name);
300 FREE_AND_NULL(graph_name);
304 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
306 * On the first invocation, this function attemps to load the commit
307 * graph if the_repository is configured to have one.
309 static int prepare_commit_graph(struct repository *r)
311 struct object_directory *odb;
314 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
315 die("dying as requested by the '%s' variable on commit-graph load!",
316 GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD);
318 if (r->objects->commit_graph_attempted)
319 return !!r->objects->commit_graph;
320 r->objects->commit_graph_attempted = 1;
322 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
323 (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
326 * This repository is not configured to use commit graphs, so
327 * do not load one. (But report commit_graph_attempted anyway
328 * so that commit graph loading is not attempted again for this
333 if (!commit_graph_compatible(r))
337 for (odb = r->objects->odb;
338 !r->objects->commit_graph && odb;
340 prepare_commit_graph_one(r, odb->path);
341 return !!r->objects->commit_graph;
344 int generation_numbers_enabled(struct repository *r)
346 uint32_t first_generation;
347 struct commit_graph *g;
348 if (!prepare_commit_graph(r))
351 g = r->objects->commit_graph;
356 first_generation = get_be32(g->chunk_commit_data +
357 g->hash_len + 8) >> 2;
359 return !!first_generation;
362 void close_commit_graph(struct raw_object_store *o)
364 free_commit_graph(o->commit_graph);
365 o->commit_graph = NULL;
368 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
370 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
371 g->chunk_oid_lookup, g->hash_len, pos);
374 static struct commit_list **insert_parent_or_die(struct repository *r,
375 struct commit_graph *g,
377 struct commit_list **pptr)
380 struct object_id oid;
382 if (pos >= g->num_commits)
383 die("invalid parent position %"PRIu64, pos);
385 hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
386 c = lookup_commit(r, &oid);
388 die(_("could not find commit %s"), oid_to_hex(&oid));
390 return &commit_list_insert(c, pptr)->next;
393 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
395 const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
396 item->graph_pos = pos;
397 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
400 static int fill_commit_in_graph(struct repository *r,
402 struct commit_graph *g, uint32_t pos)
405 uint32_t *parent_data_ptr;
406 uint64_t date_low, date_high;
407 struct commit_list **pptr;
408 const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
410 item->object.parsed = 1;
411 item->graph_pos = pos;
413 item->maybe_tree = NULL;
415 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
416 date_low = get_be32(commit_data + g->hash_len + 12);
417 item->date = (timestamp_t)((date_high << 32) | date_low);
419 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
421 pptr = &item->parents;
423 edge_value = get_be32(commit_data + g->hash_len);
424 if (edge_value == GRAPH_PARENT_NONE)
426 pptr = insert_parent_or_die(r, g, edge_value, pptr);
428 edge_value = get_be32(commit_data + g->hash_len + 4);
429 if (edge_value == GRAPH_PARENT_NONE)
431 if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
432 pptr = insert_parent_or_die(r, g, edge_value, pptr);
436 parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
437 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
439 edge_value = get_be32(parent_data_ptr);
440 pptr = insert_parent_or_die(r, g,
441 edge_value & GRAPH_EDGE_LAST_MASK,
444 } while (!(edge_value & GRAPH_LAST_EDGE));
449 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
451 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
452 *pos = item->graph_pos;
455 return bsearch_graph(g, &(item->object.oid), pos);
459 static int parse_commit_in_graph_one(struct repository *r,
460 struct commit_graph *g,
465 if (item->object.parsed)
468 if (find_commit_in_graph(item, g, &pos))
469 return fill_commit_in_graph(r, item, g, pos);
474 int parse_commit_in_graph(struct repository *r, struct commit *item)
476 if (!prepare_commit_graph(r))
478 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
481 void load_commit_graph_info(struct repository *r, struct commit *item)
484 if (!prepare_commit_graph(r))
486 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
487 fill_commit_graph_info(item, r->objects->commit_graph, pos);
490 static struct tree *load_tree_for_commit(struct repository *r,
491 struct commit_graph *g,
494 struct object_id oid;
495 const unsigned char *commit_data = g->chunk_commit_data +
496 GRAPH_DATA_WIDTH * (c->graph_pos);
498 hashcpy(oid.hash, commit_data);
499 c->maybe_tree = lookup_tree(r, &oid);
501 return c->maybe_tree;
504 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
505 struct commit_graph *g,
506 const struct commit *c)
509 return c->maybe_tree;
510 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
511 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
513 return load_tree_for_commit(r, g, (struct commit *)c);
516 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
518 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
521 struct packed_commit_list {
522 struct commit **list;
527 struct packed_oid_list {
528 struct object_id *list;
533 struct write_commit_graph_context {
534 struct repository *r;
537 struct packed_oid_list oids;
538 struct packed_commit_list commits;
540 unsigned long approx_nr_objects;
541 struct progress *progress;
543 uint64_t progress_cnt;
548 static void write_graph_chunk_fanout(struct hashfile *f,
549 struct write_commit_graph_context *ctx)
552 struct commit **list = ctx->commits.list;
555 * Write the first-level table (the list is sorted,
556 * but we use a 256-entry lookup to be able to avoid
557 * having to do eight extra binary search iterations).
559 for (i = 0; i < 256; i++) {
560 while (count < ctx->commits.nr) {
561 if ((*list)->object.oid.hash[0] != i)
563 display_progress(ctx->progress, ++ctx->progress_cnt);
568 hashwrite_be32(f, count);
572 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
573 struct write_commit_graph_context *ctx)
575 struct commit **list = ctx->commits.list;
577 for (count = 0; count < ctx->commits.nr; count++, list++) {
578 display_progress(ctx->progress, ++ctx->progress_cnt);
579 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
583 static const unsigned char *commit_to_sha1(size_t index, void *table)
585 struct commit **commits = table;
586 return commits[index]->object.oid.hash;
589 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
590 struct write_commit_graph_context *ctx)
592 struct commit **list = ctx->commits.list;
593 struct commit **last = ctx->commits.list + ctx->commits.nr;
594 uint32_t num_extra_edges = 0;
596 while (list < last) {
597 struct commit_list *parent;
599 uint32_t packedDate[2];
600 display_progress(ctx->progress, ++ctx->progress_cnt);
602 parse_commit_no_graph(*list);
603 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
605 parent = (*list)->parents;
608 edge_value = GRAPH_PARENT_NONE;
610 edge_value = sha1_pos(parent->item->object.oid.hash,
616 BUG("missing parent %s for commit %s",
617 oid_to_hex(&parent->item->object.oid),
618 oid_to_hex(&(*list)->object.oid));
621 hashwrite_be32(f, edge_value);
624 parent = parent->next;
627 edge_value = GRAPH_PARENT_NONE;
628 else if (parent->next)
629 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
631 edge_value = sha1_pos(parent->item->object.oid.hash,
636 BUG("missing parent %s for commit %s",
637 oid_to_hex(&parent->item->object.oid),
638 oid_to_hex(&(*list)->object.oid));
641 hashwrite_be32(f, edge_value);
643 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
646 parent = parent->next;
650 if (sizeof((*list)->date) > 4)
651 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
655 packedDate[0] |= htonl((*list)->generation << 2);
657 packedDate[1] = htonl((*list)->date);
658 hashwrite(f, packedDate, 8);
664 static void write_graph_chunk_extra_edges(struct hashfile *f,
665 struct write_commit_graph_context *ctx)
667 struct commit **list = ctx->commits.list;
668 struct commit **last = ctx->commits.list + ctx->commits.nr;
669 struct commit_list *parent;
671 while (list < last) {
674 display_progress(ctx->progress, ++ctx->progress_cnt);
676 for (parent = (*list)->parents; num_parents < 3 && parent;
677 parent = parent->next)
680 if (num_parents <= 2) {
685 /* Since num_parents > 2, this initializer is safe. */
686 for (parent = (*list)->parents->next; parent; parent = parent->next) {
687 int edge_value = sha1_pos(parent->item->object.oid.hash,
693 BUG("missing parent %s for commit %s",
694 oid_to_hex(&parent->item->object.oid),
695 oid_to_hex(&(*list)->object.oid));
696 else if (!parent->next)
697 edge_value |= GRAPH_LAST_EDGE;
699 hashwrite_be32(f, edge_value);
706 static int commit_compare(const void *_a, const void *_b)
708 const struct object_id *a = (const struct object_id *)_a;
709 const struct object_id *b = (const struct object_id *)_b;
713 static int add_packed_commits(const struct object_id *oid,
714 struct packed_git *pack,
718 struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
719 enum object_type type;
720 off_t offset = nth_packed_object_offset(pack, pos);
721 struct object_info oi = OBJECT_INFO_INIT;
724 display_progress(ctx->progress, ++ctx->progress_done);
727 if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
728 die(_("unable to get type of object %s"), oid_to_hex(oid));
730 if (type != OBJ_COMMIT)
733 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
734 oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
740 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
742 struct commit_list *parent;
743 for (parent = commit->parents; parent; parent = parent->next) {
744 if (!(parent->item->object.flags & UNINTERESTING)) {
745 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
746 oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
748 parent->item->object.flags |= UNINTERESTING;
753 static void close_reachable(struct write_commit_graph_context *ctx)
756 struct commit *commit;
758 if (ctx->report_progress)
759 ctx->progress = start_delayed_progress(
760 _("Loading known commits in commit graph"),
762 for (i = 0; i < ctx->oids.nr; i++) {
763 display_progress(ctx->progress, i + 1);
764 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
766 commit->object.flags |= UNINTERESTING;
768 stop_progress(&ctx->progress);
771 * As this loop runs, ctx->oids.nr may grow, but not more
772 * than the number of missing commits in the reachable
775 if (ctx->report_progress)
776 ctx->progress = start_delayed_progress(
777 _("Expanding reachable commits in commit graph"),
779 for (i = 0; i < ctx->oids.nr; i++) {
780 display_progress(ctx->progress, i + 1);
781 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
783 if (commit && !parse_commit_no_graph(commit))
784 add_missing_parents(ctx, commit);
786 stop_progress(&ctx->progress);
788 if (ctx->report_progress)
789 ctx->progress = start_delayed_progress(
790 _("Clearing commit marks in commit graph"),
792 for (i = 0; i < ctx->oids.nr; i++) {
793 display_progress(ctx->progress, i + 1);
794 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
797 commit->object.flags &= ~UNINTERESTING;
799 stop_progress(&ctx->progress);
802 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
805 struct commit_list *list = NULL;
807 if (ctx->report_progress)
808 ctx->progress = start_progress(
809 _("Computing commit graph generation numbers"),
811 for (i = 0; i < ctx->commits.nr; i++) {
812 display_progress(ctx->progress, i + 1);
813 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
814 ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
817 commit_list_insert(ctx->commits.list[i], &list);
819 struct commit *current = list->item;
820 struct commit_list *parent;
821 int all_parents_computed = 1;
822 uint32_t max_generation = 0;
824 for (parent = current->parents; parent; parent = parent->next) {
825 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
826 parent->item->generation == GENERATION_NUMBER_ZERO) {
827 all_parents_computed = 0;
828 commit_list_insert(parent->item, &list);
830 } else if (parent->item->generation > max_generation) {
831 max_generation = parent->item->generation;
835 if (all_parents_computed) {
836 current->generation = max_generation + 1;
839 if (current->generation > GENERATION_NUMBER_MAX)
840 current->generation = GENERATION_NUMBER_MAX;
844 stop_progress(&ctx->progress);
847 static int add_ref_to_list(const char *refname,
848 const struct object_id *oid,
849 int flags, void *cb_data)
851 struct string_list *list = (struct string_list *)cb_data;
853 string_list_append(list, oid_to_hex(oid));
857 int write_commit_graph_reachable(const char *obj_dir, unsigned int flags)
859 struct string_list list = STRING_LIST_INIT_DUP;
862 for_each_ref(add_ref_to_list, &list);
863 result = write_commit_graph(obj_dir, NULL, &list,
866 string_list_clear(&list, 0);
870 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
871 struct string_list *pack_indexes)
874 struct strbuf progress_title = STRBUF_INIT;
875 struct strbuf packname = STRBUF_INIT;
878 strbuf_addf(&packname, "%s/pack/", ctx->obj_dir);
879 dirlen = packname.len;
880 if (ctx->report_progress) {
881 strbuf_addf(&progress_title,
882 Q_("Finding commits for commit graph in %d pack",
883 "Finding commits for commit graph in %d packs",
886 ctx->progress = start_delayed_progress(progress_title.buf, 0);
887 ctx->progress_done = 0;
889 for (i = 0; i < pack_indexes->nr; i++) {
890 struct packed_git *p;
891 strbuf_setlen(&packname, dirlen);
892 strbuf_addstr(&packname, pack_indexes->items[i].string);
893 p = add_packed_git(packname.buf, packname.len, 1);
895 error(_("error adding pack %s"), packname.buf);
898 if (open_pack_index(p)) {
899 error(_("error opening index for %s"), packname.buf);
902 for_each_object_in_pack(p, add_packed_commits, ctx,
903 FOR_EACH_OBJECT_PACK_ORDER);
908 stop_progress(&ctx->progress);
909 strbuf_reset(&progress_title);
910 strbuf_release(&packname);
915 static void fill_oids_from_commit_hex(struct write_commit_graph_context *ctx,
916 struct string_list *commit_hex)
919 struct strbuf progress_title = STRBUF_INIT;
921 if (ctx->report_progress) {
922 strbuf_addf(&progress_title,
923 Q_("Finding commits for commit graph from %d ref",
924 "Finding commits for commit graph from %d refs",
927 ctx->progress = start_delayed_progress(
931 for (i = 0; i < commit_hex->nr; i++) {
933 struct object_id oid;
934 struct commit *result;
936 display_progress(ctx->progress, i + 1);
937 if (commit_hex->items[i].string &&
938 parse_oid_hex(commit_hex->items[i].string, &oid, &end))
941 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
944 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
945 oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
949 stop_progress(&ctx->progress);
950 strbuf_release(&progress_title);
953 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
955 if (ctx->report_progress)
956 ctx->progress = start_delayed_progress(
957 _("Finding commits for commit graph among packed objects"),
958 ctx->approx_nr_objects);
959 for_each_packed_object(add_packed_commits, ctx,
960 FOR_EACH_OBJECT_PACK_ORDER);
961 if (ctx->progress_done < ctx->approx_nr_objects)
962 display_progress(ctx->progress, ctx->approx_nr_objects);
963 stop_progress(&ctx->progress);
966 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
968 uint32_t i, count_distinct = 1;
970 if (ctx->report_progress)
971 ctx->progress = start_delayed_progress(
972 _("Counting distinct commits in commit graph"),
974 display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
975 QSORT(ctx->oids.list, ctx->oids.nr, commit_compare);
977 for (i = 1; i < ctx->oids.nr; i++) {
978 display_progress(ctx->progress, i + 1);
979 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
982 stop_progress(&ctx->progress);
984 return count_distinct;
987 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
990 struct commit_list *parent;
992 ctx->num_extra_edges = 0;
993 if (ctx->report_progress)
994 ctx->progress = start_delayed_progress(
995 _("Finding extra edges in commit graph"),
997 for (i = 0; i < ctx->oids.nr; i++) {
999 display_progress(ctx->progress, i + 1);
1000 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1003 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1004 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1006 for (parent = ctx->commits.list[ctx->commits.nr]->parents;
1007 parent; parent = parent->next)
1010 if (num_parents > 2)
1011 ctx->num_extra_edges += num_parents - 1;
1015 stop_progress(&ctx->progress);
1018 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1022 struct lock_file lk = LOCK_INIT;
1023 uint32_t chunk_ids[5];
1024 uint64_t chunk_offsets[5];
1025 const unsigned hashsz = the_hash_algo->rawsz;
1026 struct strbuf progress_title = STRBUF_INIT;
1027 int num_chunks = ctx->num_extra_edges ? 4 : 3;
1029 ctx->graph_name = get_commit_graph_filename(ctx->obj_dir);
1030 if (safe_create_leading_directories(ctx->graph_name)) {
1031 UNLEAK(ctx->graph_name);
1032 error(_("unable to create leading directories of %s"),
1037 hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
1038 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1040 hashwrite_be32(f, GRAPH_SIGNATURE);
1042 hashwrite_u8(f, GRAPH_VERSION);
1043 hashwrite_u8(f, oid_version());
1044 hashwrite_u8(f, num_chunks);
1045 hashwrite_u8(f, 0); /* unused padding byte */
1047 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1048 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1049 chunk_ids[2] = GRAPH_CHUNKID_DATA;
1050 if (ctx->num_extra_edges)
1051 chunk_ids[3] = GRAPH_CHUNKID_EXTRAEDGES;
1056 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1057 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1058 chunk_offsets[2] = chunk_offsets[1] + hashsz * ctx->commits.nr;
1059 chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * ctx->commits.nr;
1060 chunk_offsets[4] = chunk_offsets[3] + 4 * ctx->num_extra_edges;
1062 for (i = 0; i <= num_chunks; i++) {
1063 uint32_t chunk_write[3];
1065 chunk_write[0] = htonl(chunk_ids[i]);
1066 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1067 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1068 hashwrite(f, chunk_write, 12);
1071 if (ctx->report_progress) {
1072 strbuf_addf(&progress_title,
1073 Q_("Writing out commit graph in %d pass",
1074 "Writing out commit graph in %d passes",
1077 ctx->progress = start_delayed_progress(
1079 num_chunks * ctx->commits.nr);
1081 write_graph_chunk_fanout(f, ctx);
1082 write_graph_chunk_oids(f, hashsz, ctx);
1083 write_graph_chunk_data(f, hashsz, ctx);
1084 if (ctx->num_extra_edges)
1085 write_graph_chunk_extra_edges(f, ctx);
1086 stop_progress(&ctx->progress);
1087 strbuf_release(&progress_title);
1089 close_commit_graph(ctx->r->objects);
1090 finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1091 commit_lock_file(&lk);
1096 int write_commit_graph(const char *obj_dir,
1097 struct string_list *pack_indexes,
1098 struct string_list *commit_hex,
1101 struct write_commit_graph_context *ctx;
1102 uint32_t i, count_distinct = 0;
1105 if (!commit_graph_compatible(the_repository))
1108 ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
1109 ctx->r = the_repository;
1110 ctx->obj_dir = obj_dir;
1111 ctx->append = flags & COMMIT_GRAPH_APPEND ? 1 : 0;
1112 ctx->report_progress = flags & COMMIT_GRAPH_PROGRESS ? 1 : 0;
1114 ctx->approx_nr_objects = approximate_object_count();
1115 ctx->oids.alloc = ctx->approx_nr_objects / 32;
1118 prepare_commit_graph_one(ctx->r, ctx->obj_dir);
1119 if (ctx->r->objects->commit_graph)
1120 ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
1123 if (ctx->oids.alloc < 1024)
1124 ctx->oids.alloc = 1024;
1125 ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
1127 if (ctx->append && ctx->r->objects->commit_graph) {
1128 struct commit_graph *g = ctx->r->objects->commit_graph;
1129 for (i = 0; i < g->num_commits; i++) {
1130 const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
1131 hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
1136 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
1141 fill_oids_from_commit_hex(ctx, commit_hex);
1143 if (!pack_indexes && !commit_hex)
1144 fill_oids_from_all_packs(ctx);
1146 close_reachable(ctx);
1148 count_distinct = count_distinct_commits(ctx);
1150 if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
1151 error(_("the commit graph format cannot write %d commits"), count_distinct);
1156 ctx->commits.alloc = count_distinct;
1157 ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
1159 copy_oids_to_commits(ctx);
1161 if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
1162 error(_("too many commits to write graph"));
1167 compute_generation_numbers(ctx);
1169 res = write_commit_graph_file(ctx);
1172 free(ctx->graph_name);
1173 free(ctx->commits.list);
1174 free(ctx->oids.list);
1180 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1181 static int verify_commit_graph_error;
1183 static void graph_report(const char *fmt, ...)
1187 verify_commit_graph_error = 1;
1189 vfprintf(stderr, fmt, ap);
1190 fprintf(stderr, "\n");
1194 #define GENERATION_ZERO_EXISTS 1
1195 #define GENERATION_NUMBER_EXISTS 2
1197 int verify_commit_graph(struct repository *r, struct commit_graph *g)
1199 uint32_t i, cur_fanout_pos = 0;
1200 struct object_id prev_oid, cur_oid, checksum;
1201 int generation_zero = 0;
1204 struct progress *progress = NULL;
1207 graph_report("no commit-graph file loaded");
1211 verify_commit_graph_error = verify_commit_graph_lite(g);
1212 if (verify_commit_graph_error)
1213 return verify_commit_graph_error;
1215 devnull = open("/dev/null", O_WRONLY);
1216 f = hashfd(devnull, NULL);
1217 hashwrite(f, g->data, g->data_len - g->hash_len);
1218 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1219 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1220 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1221 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1224 for (i = 0; i < g->num_commits; i++) {
1225 struct commit *graph_commit;
1227 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1229 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1230 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
1231 oid_to_hex(&prev_oid),
1232 oid_to_hex(&cur_oid));
1234 oidcpy(&prev_oid, &cur_oid);
1236 while (cur_oid.hash[0] > cur_fanout_pos) {
1237 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1239 if (i != fanout_value)
1240 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1241 cur_fanout_pos, fanout_value, i);
1245 graph_commit = lookup_commit(r, &cur_oid);
1246 if (!parse_commit_in_graph_one(r, g, graph_commit))
1247 graph_report(_("failed to parse commit %s from commit-graph"),
1248 oid_to_hex(&cur_oid));
1251 while (cur_fanout_pos < 256) {
1252 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1254 if (g->num_commits != fanout_value)
1255 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1256 cur_fanout_pos, fanout_value, i);
1261 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1262 return verify_commit_graph_error;
1264 progress = start_progress(_("Verifying commits in commit graph"),
1266 for (i = 0; i < g->num_commits; i++) {
1267 struct commit *graph_commit, *odb_commit;
1268 struct commit_list *graph_parents, *odb_parents;
1269 uint32_t max_generation = 0;
1271 display_progress(progress, i + 1);
1272 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1274 graph_commit = lookup_commit(r, &cur_oid);
1275 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1276 if (parse_commit_internal(odb_commit, 0, 0)) {
1277 graph_report(_("failed to parse commit %s from object database for commit-graph"),
1278 oid_to_hex(&cur_oid));
1282 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
1283 get_commit_tree_oid(odb_commit)))
1284 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
1285 oid_to_hex(&cur_oid),
1286 oid_to_hex(get_commit_tree_oid(graph_commit)),
1287 oid_to_hex(get_commit_tree_oid(odb_commit)));
1289 graph_parents = graph_commit->parents;
1290 odb_parents = odb_commit->parents;
1292 while (graph_parents) {
1293 if (odb_parents == NULL) {
1294 graph_report(_("commit-graph parent list for commit %s is too long"),
1295 oid_to_hex(&cur_oid));
1299 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1300 graph_report(_("commit-graph parent for %s is %s != %s"),
1301 oid_to_hex(&cur_oid),
1302 oid_to_hex(&graph_parents->item->object.oid),
1303 oid_to_hex(&odb_parents->item->object.oid));
1305 if (graph_parents->item->generation > max_generation)
1306 max_generation = graph_parents->item->generation;
1308 graph_parents = graph_parents->next;
1309 odb_parents = odb_parents->next;
1312 if (odb_parents != NULL)
1313 graph_report(_("commit-graph parent list for commit %s terminates early"),
1314 oid_to_hex(&cur_oid));
1316 if (!graph_commit->generation) {
1317 if (generation_zero == GENERATION_NUMBER_EXISTS)
1318 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
1319 oid_to_hex(&cur_oid));
1320 generation_zero = GENERATION_ZERO_EXISTS;
1321 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1322 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
1323 oid_to_hex(&cur_oid));
1325 if (generation_zero == GENERATION_ZERO_EXISTS)
1329 * If one of our parents has generation GENERATION_NUMBER_MAX, then
1330 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1331 * extra logic in the following condition.
1333 if (max_generation == GENERATION_NUMBER_MAX)
1336 if (graph_commit->generation != max_generation + 1)
1337 graph_report(_("commit-graph generation for commit %s is %u != %u"),
1338 oid_to_hex(&cur_oid),
1339 graph_commit->generation,
1340 max_generation + 1);
1342 if (graph_commit->date != odb_commit->date)
1343 graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
1344 oid_to_hex(&cur_oid),
1348 stop_progress(&progress);
1350 return verify_commit_graph_error;
1353 void free_commit_graph(struct commit_graph *g)
1357 if (g->graph_fd >= 0) {
1358 munmap((void *)g->data, g->data_len);