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 static void close_commit_graph_one(struct commit_graph *g)
367 close_commit_graph_one(g->base_graph);
368 free_commit_graph(g);
371 void close_commit_graph(struct raw_object_store *o)
373 close_commit_graph_one(o->commit_graph);
374 o->commit_graph = NULL;
377 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
379 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
380 g->chunk_oid_lookup, g->hash_len, pos);
383 static void load_oid_from_graph(struct commit_graph *g,
385 struct object_id *oid)
389 while (g && pos < g->num_commits_in_base)
393 BUG("NULL commit-graph");
395 if (pos >= g->num_commits + g->num_commits_in_base)
396 die(_("invalid commit position. commit-graph is likely corrupt"));
398 lex_index = pos - g->num_commits_in_base;
400 hashcpy(oid->hash, g->chunk_oid_lookup + g->hash_len * lex_index);
403 static struct commit_list **insert_parent_or_die(struct repository *r,
404 struct commit_graph *g,
406 struct commit_list **pptr)
409 struct object_id oid;
411 if (pos >= g->num_commits + g->num_commits_in_base)
412 die("invalid parent position %"PRIu32, pos);
414 load_oid_from_graph(g, pos, &oid);
415 c = lookup_commit(r, &oid);
417 die(_("could not find commit %s"), oid_to_hex(&oid));
419 return &commit_list_insert(c, pptr)->next;
422 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
424 const unsigned char *commit_data;
427 while (pos < g->num_commits_in_base)
430 lex_index = pos - g->num_commits_in_base;
431 commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * lex_index;
432 item->graph_pos = pos;
433 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
436 static int fill_commit_in_graph(struct repository *r,
438 struct commit_graph *g, uint32_t pos)
441 uint32_t *parent_data_ptr;
442 uint64_t date_low, date_high;
443 struct commit_list **pptr;
444 const unsigned char *commit_data;
447 while (pos < g->num_commits_in_base)
450 if (pos >= g->num_commits + g->num_commits_in_base)
451 die(_("invalid commit position. commit-graph is likely corrupt"));
454 * Store the "full" position, but then use the
455 * "local" position for the rest of the calculation.
457 item->graph_pos = pos;
458 lex_index = pos - g->num_commits_in_base;
460 commit_data = g->chunk_commit_data + (g->hash_len + 16) * lex_index;
462 item->object.parsed = 1;
464 item->maybe_tree = NULL;
466 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
467 date_low = get_be32(commit_data + g->hash_len + 12);
468 item->date = (timestamp_t)((date_high << 32) | date_low);
470 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
472 pptr = &item->parents;
474 edge_value = get_be32(commit_data + g->hash_len);
475 if (edge_value == GRAPH_PARENT_NONE)
477 pptr = insert_parent_or_die(r, g, edge_value, pptr);
479 edge_value = get_be32(commit_data + g->hash_len + 4);
480 if (edge_value == GRAPH_PARENT_NONE)
482 if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
483 pptr = insert_parent_or_die(r, g, edge_value, pptr);
487 parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
488 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
490 edge_value = get_be32(parent_data_ptr);
491 pptr = insert_parent_or_die(r, g,
492 edge_value & GRAPH_EDGE_LAST_MASK,
495 } while (!(edge_value & GRAPH_LAST_EDGE));
500 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
502 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
503 *pos = item->graph_pos;
506 struct commit_graph *cur_g = g;
509 while (cur_g && !bsearch_graph(cur_g, &(item->object.oid), &lex_index))
510 cur_g = cur_g->base_graph;
513 *pos = lex_index + cur_g->num_commits_in_base;
521 static int parse_commit_in_graph_one(struct repository *r,
522 struct commit_graph *g,
527 if (item->object.parsed)
530 if (find_commit_in_graph(item, g, &pos))
531 return fill_commit_in_graph(r, item, g, pos);
536 int parse_commit_in_graph(struct repository *r, struct commit *item)
538 if (!prepare_commit_graph(r))
540 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
543 void load_commit_graph_info(struct repository *r, struct commit *item)
546 if (!prepare_commit_graph(r))
548 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
549 fill_commit_graph_info(item, r->objects->commit_graph, pos);
552 static struct tree *load_tree_for_commit(struct repository *r,
553 struct commit_graph *g,
556 struct object_id oid;
557 const unsigned char *commit_data;
559 while (c->graph_pos < g->num_commits_in_base)
562 commit_data = g->chunk_commit_data +
563 GRAPH_DATA_WIDTH * (c->graph_pos - g->num_commits_in_base);
565 hashcpy(oid.hash, commit_data);
566 c->maybe_tree = lookup_tree(r, &oid);
568 return c->maybe_tree;
571 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
572 struct commit_graph *g,
573 const struct commit *c)
576 return c->maybe_tree;
577 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
578 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
580 return load_tree_for_commit(r, g, (struct commit *)c);
583 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
585 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
588 struct packed_commit_list {
589 struct commit **list;
594 struct packed_oid_list {
595 struct object_id *list;
600 struct write_commit_graph_context {
601 struct repository *r;
604 struct packed_oid_list oids;
605 struct packed_commit_list commits;
607 unsigned long approx_nr_objects;
608 struct progress *progress;
610 uint64_t progress_cnt;
615 static void write_graph_chunk_fanout(struct hashfile *f,
616 struct write_commit_graph_context *ctx)
619 struct commit **list = ctx->commits.list;
622 * Write the first-level table (the list is sorted,
623 * but we use a 256-entry lookup to be able to avoid
624 * having to do eight extra binary search iterations).
626 for (i = 0; i < 256; i++) {
627 while (count < ctx->commits.nr) {
628 if ((*list)->object.oid.hash[0] != i)
630 display_progress(ctx->progress, ++ctx->progress_cnt);
635 hashwrite_be32(f, count);
639 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
640 struct write_commit_graph_context *ctx)
642 struct commit **list = ctx->commits.list;
644 for (count = 0; count < ctx->commits.nr; count++, list++) {
645 display_progress(ctx->progress, ++ctx->progress_cnt);
646 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
650 static const unsigned char *commit_to_sha1(size_t index, void *table)
652 struct commit **commits = table;
653 return commits[index]->object.oid.hash;
656 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
657 struct write_commit_graph_context *ctx)
659 struct commit **list = ctx->commits.list;
660 struct commit **last = ctx->commits.list + ctx->commits.nr;
661 uint32_t num_extra_edges = 0;
663 while (list < last) {
664 struct commit_list *parent;
666 uint32_t packedDate[2];
667 display_progress(ctx->progress, ++ctx->progress_cnt);
669 parse_commit_no_graph(*list);
670 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
672 parent = (*list)->parents;
675 edge_value = GRAPH_PARENT_NONE;
677 edge_value = sha1_pos(parent->item->object.oid.hash,
683 BUG("missing parent %s for commit %s",
684 oid_to_hex(&parent->item->object.oid),
685 oid_to_hex(&(*list)->object.oid));
688 hashwrite_be32(f, edge_value);
691 parent = parent->next;
694 edge_value = GRAPH_PARENT_NONE;
695 else if (parent->next)
696 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
698 edge_value = sha1_pos(parent->item->object.oid.hash,
703 BUG("missing parent %s for commit %s",
704 oid_to_hex(&parent->item->object.oid),
705 oid_to_hex(&(*list)->object.oid));
708 hashwrite_be32(f, edge_value);
710 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
713 parent = parent->next;
717 if (sizeof((*list)->date) > 4)
718 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
722 packedDate[0] |= htonl((*list)->generation << 2);
724 packedDate[1] = htonl((*list)->date);
725 hashwrite(f, packedDate, 8);
731 static void write_graph_chunk_extra_edges(struct hashfile *f,
732 struct write_commit_graph_context *ctx)
734 struct commit **list = ctx->commits.list;
735 struct commit **last = ctx->commits.list + ctx->commits.nr;
736 struct commit_list *parent;
738 while (list < last) {
741 display_progress(ctx->progress, ++ctx->progress_cnt);
743 for (parent = (*list)->parents; num_parents < 3 && parent;
744 parent = parent->next)
747 if (num_parents <= 2) {
752 /* Since num_parents > 2, this initializer is safe. */
753 for (parent = (*list)->parents->next; parent; parent = parent->next) {
754 int edge_value = sha1_pos(parent->item->object.oid.hash,
760 BUG("missing parent %s for commit %s",
761 oid_to_hex(&parent->item->object.oid),
762 oid_to_hex(&(*list)->object.oid));
763 else if (!parent->next)
764 edge_value |= GRAPH_LAST_EDGE;
766 hashwrite_be32(f, edge_value);
773 static int oid_compare(const void *_a, const void *_b)
775 const struct object_id *a = (const struct object_id *)_a;
776 const struct object_id *b = (const struct object_id *)_b;
780 static int add_packed_commits(const struct object_id *oid,
781 struct packed_git *pack,
785 struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
786 enum object_type type;
787 off_t offset = nth_packed_object_offset(pack, pos);
788 struct object_info oi = OBJECT_INFO_INIT;
791 display_progress(ctx->progress, ++ctx->progress_done);
794 if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
795 die(_("unable to get type of object %s"), oid_to_hex(oid));
797 if (type != OBJ_COMMIT)
800 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
801 oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
807 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
809 struct commit_list *parent;
810 for (parent = commit->parents; parent; parent = parent->next) {
811 if (!(parent->item->object.flags & UNINTERESTING)) {
812 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
813 oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
815 parent->item->object.flags |= UNINTERESTING;
820 static void close_reachable(struct write_commit_graph_context *ctx)
823 struct commit *commit;
825 if (ctx->report_progress)
826 ctx->progress = start_delayed_progress(
827 _("Loading known commits in commit graph"),
829 for (i = 0; i < ctx->oids.nr; i++) {
830 display_progress(ctx->progress, i + 1);
831 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
833 commit->object.flags |= UNINTERESTING;
835 stop_progress(&ctx->progress);
838 * As this loop runs, ctx->oids.nr may grow, but not more
839 * than the number of missing commits in the reachable
842 if (ctx->report_progress)
843 ctx->progress = start_delayed_progress(
844 _("Expanding reachable commits in commit graph"),
846 for (i = 0; i < ctx->oids.nr; i++) {
847 display_progress(ctx->progress, i + 1);
848 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
850 if (commit && !parse_commit_no_graph(commit))
851 add_missing_parents(ctx, commit);
853 stop_progress(&ctx->progress);
855 if (ctx->report_progress)
856 ctx->progress = start_delayed_progress(
857 _("Clearing commit marks in commit graph"),
859 for (i = 0; i < ctx->oids.nr; i++) {
860 display_progress(ctx->progress, i + 1);
861 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
864 commit->object.flags &= ~UNINTERESTING;
866 stop_progress(&ctx->progress);
869 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
872 struct commit_list *list = NULL;
874 if (ctx->report_progress)
875 ctx->progress = start_progress(
876 _("Computing commit graph generation numbers"),
878 for (i = 0; i < ctx->commits.nr; i++) {
879 display_progress(ctx->progress, i + 1);
880 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
881 ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
884 commit_list_insert(ctx->commits.list[i], &list);
886 struct commit *current = list->item;
887 struct commit_list *parent;
888 int all_parents_computed = 1;
889 uint32_t max_generation = 0;
891 for (parent = current->parents; parent; parent = parent->next) {
892 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
893 parent->item->generation == GENERATION_NUMBER_ZERO) {
894 all_parents_computed = 0;
895 commit_list_insert(parent->item, &list);
897 } else if (parent->item->generation > max_generation) {
898 max_generation = parent->item->generation;
902 if (all_parents_computed) {
903 current->generation = max_generation + 1;
906 if (current->generation > GENERATION_NUMBER_MAX)
907 current->generation = GENERATION_NUMBER_MAX;
911 stop_progress(&ctx->progress);
914 static int add_ref_to_list(const char *refname,
915 const struct object_id *oid,
916 int flags, void *cb_data)
918 struct string_list *list = (struct string_list *)cb_data;
920 string_list_append(list, oid_to_hex(oid));
924 int write_commit_graph_reachable(const char *obj_dir, unsigned int flags)
926 struct string_list list = STRING_LIST_INIT_DUP;
929 for_each_ref(add_ref_to_list, &list);
930 result = write_commit_graph(obj_dir, NULL, &list,
933 string_list_clear(&list, 0);
937 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
938 struct string_list *pack_indexes)
941 struct strbuf progress_title = STRBUF_INIT;
942 struct strbuf packname = STRBUF_INIT;
945 strbuf_addf(&packname, "%s/pack/", ctx->obj_dir);
946 dirlen = packname.len;
947 if (ctx->report_progress) {
948 strbuf_addf(&progress_title,
949 Q_("Finding commits for commit graph in %d pack",
950 "Finding commits for commit graph in %d packs",
953 ctx->progress = start_delayed_progress(progress_title.buf, 0);
954 ctx->progress_done = 0;
956 for (i = 0; i < pack_indexes->nr; i++) {
957 struct packed_git *p;
958 strbuf_setlen(&packname, dirlen);
959 strbuf_addstr(&packname, pack_indexes->items[i].string);
960 p = add_packed_git(packname.buf, packname.len, 1);
962 error(_("error adding pack %s"), packname.buf);
965 if (open_pack_index(p)) {
966 error(_("error opening index for %s"), packname.buf);
969 for_each_object_in_pack(p, add_packed_commits, ctx,
970 FOR_EACH_OBJECT_PACK_ORDER);
975 stop_progress(&ctx->progress);
976 strbuf_reset(&progress_title);
977 strbuf_release(&packname);
982 static void fill_oids_from_commit_hex(struct write_commit_graph_context *ctx,
983 struct string_list *commit_hex)
986 struct strbuf progress_title = STRBUF_INIT;
988 if (ctx->report_progress) {
989 strbuf_addf(&progress_title,
990 Q_("Finding commits for commit graph from %d ref",
991 "Finding commits for commit graph from %d refs",
994 ctx->progress = start_delayed_progress(
998 for (i = 0; i < commit_hex->nr; i++) {
1000 struct object_id oid;
1001 struct commit *result;
1003 display_progress(ctx->progress, i + 1);
1004 if (commit_hex->items[i].string &&
1005 parse_oid_hex(commit_hex->items[i].string, &oid, &end))
1008 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1011 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1012 oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
1016 stop_progress(&ctx->progress);
1017 strbuf_release(&progress_title);
1020 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1022 if (ctx->report_progress)
1023 ctx->progress = start_delayed_progress(
1024 _("Finding commits for commit graph among packed objects"),
1025 ctx->approx_nr_objects);
1026 for_each_packed_object(add_packed_commits, ctx,
1027 FOR_EACH_OBJECT_PACK_ORDER);
1028 if (ctx->progress_done < ctx->approx_nr_objects)
1029 display_progress(ctx->progress, ctx->approx_nr_objects);
1030 stop_progress(&ctx->progress);
1033 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
1035 uint32_t i, count_distinct = 1;
1037 if (ctx->report_progress)
1038 ctx->progress = start_delayed_progress(
1039 _("Counting distinct commits in commit graph"),
1041 display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1042 QSORT(ctx->oids.list, ctx->oids.nr, oid_compare);
1044 for (i = 1; i < ctx->oids.nr; i++) {
1045 display_progress(ctx->progress, i + 1);
1046 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1049 stop_progress(&ctx->progress);
1051 return count_distinct;
1054 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1057 struct commit_list *parent;
1059 ctx->num_extra_edges = 0;
1060 if (ctx->report_progress)
1061 ctx->progress = start_delayed_progress(
1062 _("Finding extra edges in commit graph"),
1064 for (i = 0; i < ctx->oids.nr; i++) {
1065 int num_parents = 0;
1066 display_progress(ctx->progress, i + 1);
1067 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1070 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1071 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1073 for (parent = ctx->commits.list[ctx->commits.nr]->parents;
1074 parent; parent = parent->next)
1077 if (num_parents > 2)
1078 ctx->num_extra_edges += num_parents - 1;
1082 stop_progress(&ctx->progress);
1085 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1089 struct lock_file lk = LOCK_INIT;
1090 uint32_t chunk_ids[5];
1091 uint64_t chunk_offsets[5];
1092 const unsigned hashsz = the_hash_algo->rawsz;
1093 struct strbuf progress_title = STRBUF_INIT;
1094 int num_chunks = ctx->num_extra_edges ? 4 : 3;
1096 ctx->graph_name = get_commit_graph_filename(ctx->obj_dir);
1097 if (safe_create_leading_directories(ctx->graph_name)) {
1098 UNLEAK(ctx->graph_name);
1099 error(_("unable to create leading directories of %s"),
1104 hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
1105 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1107 hashwrite_be32(f, GRAPH_SIGNATURE);
1109 hashwrite_u8(f, GRAPH_VERSION);
1110 hashwrite_u8(f, oid_version());
1111 hashwrite_u8(f, num_chunks);
1112 hashwrite_u8(f, 0); /* unused padding byte */
1114 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1115 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1116 chunk_ids[2] = GRAPH_CHUNKID_DATA;
1117 if (ctx->num_extra_edges)
1118 chunk_ids[3] = GRAPH_CHUNKID_EXTRAEDGES;
1123 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1124 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1125 chunk_offsets[2] = chunk_offsets[1] + hashsz * ctx->commits.nr;
1126 chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * ctx->commits.nr;
1127 chunk_offsets[4] = chunk_offsets[3] + 4 * ctx->num_extra_edges;
1129 for (i = 0; i <= num_chunks; i++) {
1130 uint32_t chunk_write[3];
1132 chunk_write[0] = htonl(chunk_ids[i]);
1133 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1134 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1135 hashwrite(f, chunk_write, 12);
1138 if (ctx->report_progress) {
1139 strbuf_addf(&progress_title,
1140 Q_("Writing out commit graph in %d pass",
1141 "Writing out commit graph in %d passes",
1144 ctx->progress = start_delayed_progress(
1146 num_chunks * ctx->commits.nr);
1148 write_graph_chunk_fanout(f, ctx);
1149 write_graph_chunk_oids(f, hashsz, ctx);
1150 write_graph_chunk_data(f, hashsz, ctx);
1151 if (ctx->num_extra_edges)
1152 write_graph_chunk_extra_edges(f, ctx);
1153 stop_progress(&ctx->progress);
1154 strbuf_release(&progress_title);
1156 close_commit_graph(ctx->r->objects);
1157 finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1158 commit_lock_file(&lk);
1163 int write_commit_graph(const char *obj_dir,
1164 struct string_list *pack_indexes,
1165 struct string_list *commit_hex,
1168 struct write_commit_graph_context *ctx;
1169 uint32_t i, count_distinct = 0;
1172 if (!commit_graph_compatible(the_repository))
1175 ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
1176 ctx->r = the_repository;
1177 ctx->obj_dir = obj_dir;
1178 ctx->append = flags & COMMIT_GRAPH_APPEND ? 1 : 0;
1179 ctx->report_progress = flags & COMMIT_GRAPH_PROGRESS ? 1 : 0;
1181 ctx->approx_nr_objects = approximate_object_count();
1182 ctx->oids.alloc = ctx->approx_nr_objects / 32;
1185 prepare_commit_graph_one(ctx->r, ctx->obj_dir);
1186 if (ctx->r->objects->commit_graph)
1187 ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
1190 if (ctx->oids.alloc < 1024)
1191 ctx->oids.alloc = 1024;
1192 ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
1194 if (ctx->append && ctx->r->objects->commit_graph) {
1195 struct commit_graph *g = ctx->r->objects->commit_graph;
1196 for (i = 0; i < g->num_commits; i++) {
1197 const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
1198 hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
1203 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
1208 fill_oids_from_commit_hex(ctx, commit_hex);
1210 if (!pack_indexes && !commit_hex)
1211 fill_oids_from_all_packs(ctx);
1213 close_reachable(ctx);
1215 count_distinct = count_distinct_commits(ctx);
1217 if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
1218 error(_("the commit graph format cannot write %d commits"), count_distinct);
1223 ctx->commits.alloc = count_distinct;
1224 ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
1226 copy_oids_to_commits(ctx);
1228 if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
1229 error(_("too many commits to write graph"));
1234 compute_generation_numbers(ctx);
1236 res = write_commit_graph_file(ctx);
1239 free(ctx->graph_name);
1240 free(ctx->commits.list);
1241 free(ctx->oids.list);
1247 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1248 static int verify_commit_graph_error;
1250 static void graph_report(const char *fmt, ...)
1254 verify_commit_graph_error = 1;
1256 vfprintf(stderr, fmt, ap);
1257 fprintf(stderr, "\n");
1261 #define GENERATION_ZERO_EXISTS 1
1262 #define GENERATION_NUMBER_EXISTS 2
1264 int verify_commit_graph(struct repository *r, struct commit_graph *g)
1266 uint32_t i, cur_fanout_pos = 0;
1267 struct object_id prev_oid, cur_oid, checksum;
1268 int generation_zero = 0;
1271 struct progress *progress = NULL;
1274 graph_report("no commit-graph file loaded");
1278 verify_commit_graph_error = verify_commit_graph_lite(g);
1279 if (verify_commit_graph_error)
1280 return verify_commit_graph_error;
1282 devnull = open("/dev/null", O_WRONLY);
1283 f = hashfd(devnull, NULL);
1284 hashwrite(f, g->data, g->data_len - g->hash_len);
1285 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1286 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1287 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1288 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1291 for (i = 0; i < g->num_commits; i++) {
1292 struct commit *graph_commit;
1294 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1296 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1297 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
1298 oid_to_hex(&prev_oid),
1299 oid_to_hex(&cur_oid));
1301 oidcpy(&prev_oid, &cur_oid);
1303 while (cur_oid.hash[0] > cur_fanout_pos) {
1304 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1306 if (i != fanout_value)
1307 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1308 cur_fanout_pos, fanout_value, i);
1312 graph_commit = lookup_commit(r, &cur_oid);
1313 if (!parse_commit_in_graph_one(r, g, graph_commit))
1314 graph_report(_("failed to parse commit %s from commit-graph"),
1315 oid_to_hex(&cur_oid));
1318 while (cur_fanout_pos < 256) {
1319 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1321 if (g->num_commits != fanout_value)
1322 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1323 cur_fanout_pos, fanout_value, i);
1328 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1329 return verify_commit_graph_error;
1331 progress = start_progress(_("Verifying commits in commit graph"),
1333 for (i = 0; i < g->num_commits; i++) {
1334 struct commit *graph_commit, *odb_commit;
1335 struct commit_list *graph_parents, *odb_parents;
1336 uint32_t max_generation = 0;
1338 display_progress(progress, i + 1);
1339 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1341 graph_commit = lookup_commit(r, &cur_oid);
1342 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1343 if (parse_commit_internal(odb_commit, 0, 0)) {
1344 graph_report(_("failed to parse commit %s from object database for commit-graph"),
1345 oid_to_hex(&cur_oid));
1349 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
1350 get_commit_tree_oid(odb_commit)))
1351 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
1352 oid_to_hex(&cur_oid),
1353 oid_to_hex(get_commit_tree_oid(graph_commit)),
1354 oid_to_hex(get_commit_tree_oid(odb_commit)));
1356 graph_parents = graph_commit->parents;
1357 odb_parents = odb_commit->parents;
1359 while (graph_parents) {
1360 if (odb_parents == NULL) {
1361 graph_report(_("commit-graph parent list for commit %s is too long"),
1362 oid_to_hex(&cur_oid));
1366 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1367 graph_report(_("commit-graph parent for %s is %s != %s"),
1368 oid_to_hex(&cur_oid),
1369 oid_to_hex(&graph_parents->item->object.oid),
1370 oid_to_hex(&odb_parents->item->object.oid));
1372 if (graph_parents->item->generation > max_generation)
1373 max_generation = graph_parents->item->generation;
1375 graph_parents = graph_parents->next;
1376 odb_parents = odb_parents->next;
1379 if (odb_parents != NULL)
1380 graph_report(_("commit-graph parent list for commit %s terminates early"),
1381 oid_to_hex(&cur_oid));
1383 if (!graph_commit->generation) {
1384 if (generation_zero == GENERATION_NUMBER_EXISTS)
1385 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
1386 oid_to_hex(&cur_oid));
1387 generation_zero = GENERATION_ZERO_EXISTS;
1388 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1389 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
1390 oid_to_hex(&cur_oid));
1392 if (generation_zero == GENERATION_ZERO_EXISTS)
1396 * If one of our parents has generation GENERATION_NUMBER_MAX, then
1397 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1398 * extra logic in the following condition.
1400 if (max_generation == GENERATION_NUMBER_MAX)
1403 if (graph_commit->generation != max_generation + 1)
1404 graph_report(_("commit-graph generation for commit %s is %u != %u"),
1405 oid_to_hex(&cur_oid),
1406 graph_commit->generation,
1407 max_generation + 1);
1409 if (graph_commit->date != odb_commit->date)
1410 graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
1411 oid_to_hex(&cur_oid),
1415 stop_progress(&progress);
1417 return verify_commit_graph_error;
1420 void free_commit_graph(struct commit_graph *g)
1424 if (g->graph_fd >= 0) {
1425 munmap((void *)g->data, g->data_len);