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" */
25 #define GRAPH_CHUNKID_BASE 0x42415345 /* "BASE" */
27 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
29 #define GRAPH_VERSION_1 0x1
30 #define GRAPH_VERSION GRAPH_VERSION_1
32 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
33 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
34 #define GRAPH_PARENT_NONE 0x70000000
36 #define GRAPH_LAST_EDGE 0x80000000
38 #define GRAPH_HEADER_SIZE 8
39 #define GRAPH_FANOUT_SIZE (4 * 256)
40 #define GRAPH_CHUNKLOOKUP_WIDTH 12
41 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
42 + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
44 char *get_commit_graph_filename(const char *obj_dir)
46 return xstrfmt("%s/info/commit-graph", obj_dir);
49 static char *get_split_graph_filename(const char *obj_dir,
52 return xstrfmt("%s/info/commit-graphs/graph-%s.graph",
57 static char *get_chain_filename(const char *obj_dir)
59 return xstrfmt("%s/info/commit-graphs/commit-graph-chain", obj_dir);
62 static uint8_t oid_version(void)
67 static struct commit_graph *alloc_commit_graph(void)
69 struct commit_graph *g = xcalloc(1, sizeof(*g));
75 extern int read_replace_refs;
77 static int commit_graph_compatible(struct repository *r)
82 if (read_replace_refs) {
83 prepare_replace_object(r);
84 if (hashmap_get_size(&r->objects->replace_map->map))
88 prepare_commit_graft(r);
89 if (r->parsed_objects && r->parsed_objects->grafts_nr)
91 if (is_repository_shallow(r))
97 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
99 *fd = git_open(graph_file);
102 if (fstat(*fd, st)) {
109 struct commit_graph *load_commit_graph_one_fd_st(int fd, struct stat *st)
113 struct commit_graph *ret;
115 graph_size = xsize_t(st->st_size);
117 if (graph_size < GRAPH_MIN_SIZE) {
119 error(_("commit-graph file is too small"));
122 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
123 ret = parse_commit_graph(graph_map, fd, graph_size);
126 munmap(graph_map, graph_size);
133 static int verify_commit_graph_lite(struct commit_graph *g)
136 * Basic validation shared between parse_commit_graph()
137 * which'll be called every time the graph is used, and the
138 * much more expensive verify_commit_graph() used by
139 * "commit-graph verify".
141 * There should only be very basic checks here to ensure that
142 * we don't e.g. segfault in fill_commit_in_graph(), but
143 * because this is a very hot codepath nothing that e.g. loops
144 * over g->num_commits, or runs a checksum on the commit-graph
147 if (!g->chunk_oid_fanout) {
148 error("commit-graph is missing the OID Fanout chunk");
151 if (!g->chunk_oid_lookup) {
152 error("commit-graph is missing the OID Lookup chunk");
155 if (!g->chunk_commit_data) {
156 error("commit-graph is missing the Commit Data chunk");
163 struct commit_graph *parse_commit_graph(void *graph_map, int fd,
166 const unsigned char *data, *chunk_lookup;
168 struct commit_graph *graph;
169 uint64_t last_chunk_offset;
170 uint32_t last_chunk_id;
171 uint32_t graph_signature;
172 unsigned char graph_version, hash_version;
177 if (graph_size < GRAPH_MIN_SIZE)
180 data = (const unsigned char *)graph_map;
182 graph_signature = get_be32(data);
183 if (graph_signature != GRAPH_SIGNATURE) {
184 error(_("commit-graph signature %X does not match signature %X"),
185 graph_signature, GRAPH_SIGNATURE);
189 graph_version = *(unsigned char*)(data + 4);
190 if (graph_version != GRAPH_VERSION) {
191 error(_("commit-graph version %X does not match version %X"),
192 graph_version, GRAPH_VERSION);
196 hash_version = *(unsigned char*)(data + 5);
197 if (hash_version != oid_version()) {
198 error(_("commit-graph hash version %X does not match version %X"),
199 hash_version, oid_version());
203 graph = alloc_commit_graph();
205 graph->hash_len = the_hash_algo->rawsz;
206 graph->num_chunks = *(unsigned char*)(data + 6);
207 graph->graph_fd = fd;
208 graph->data = graph_map;
209 graph->data_len = graph_size;
212 last_chunk_offset = 8;
213 chunk_lookup = data + 8;
214 for (i = 0; i < graph->num_chunks; i++) {
216 uint64_t chunk_offset;
217 int chunk_repeated = 0;
219 if (data + graph_size - chunk_lookup <
220 GRAPH_CHUNKLOOKUP_WIDTH) {
221 error(_("commit-graph chunk lookup table entry missing; file may be incomplete"));
226 chunk_id = get_be32(chunk_lookup + 0);
227 chunk_offset = get_be64(chunk_lookup + 4);
229 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
231 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
232 error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
233 (uint32_t)chunk_offset);
239 case GRAPH_CHUNKID_OIDFANOUT:
240 if (graph->chunk_oid_fanout)
243 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
246 case GRAPH_CHUNKID_OIDLOOKUP:
247 if (graph->chunk_oid_lookup)
250 graph->chunk_oid_lookup = data + chunk_offset;
253 case GRAPH_CHUNKID_DATA:
254 if (graph->chunk_commit_data)
257 graph->chunk_commit_data = data + chunk_offset;
260 case GRAPH_CHUNKID_EXTRAEDGES:
261 if (graph->chunk_extra_edges)
264 graph->chunk_extra_edges = data + chunk_offset;
267 case GRAPH_CHUNKID_BASE:
268 if (graph->chunk_base_graphs)
271 graph->chunk_base_graphs = data + chunk_offset;
274 if (chunk_repeated) {
275 error(_("commit-graph chunk id %08x appears multiple times"), chunk_id);
280 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
282 graph->num_commits = (chunk_offset - last_chunk_offset)
286 last_chunk_id = chunk_id;
287 last_chunk_offset = chunk_offset;
290 hashcpy(graph->oid.hash, graph->data + graph->data_len - graph->hash_len);
292 if (verify_commit_graph_lite(graph))
298 static struct commit_graph *load_commit_graph_one(const char *graph_file)
303 struct commit_graph *g;
304 int open_ok = open_commit_graph(graph_file, &fd, &st);
309 g = load_commit_graph_one_fd_st(fd, &st);
312 g->filename = xstrdup(graph_file);
317 static struct commit_graph *load_commit_graph_v1(struct repository *r, const char *obj_dir)
319 char *graph_name = get_commit_graph_filename(obj_dir);
320 struct commit_graph *g = load_commit_graph_one(graph_name);
324 g->obj_dir = obj_dir;
329 static int add_graph_to_chain(struct commit_graph *g,
330 struct commit_graph *chain,
331 struct object_id *oids,
334 struct commit_graph *cur_g = chain;
336 if (n && !g->chunk_base_graphs) {
337 warning(_("commit-graph has no base graphs chunk"));
345 !oideq(&oids[n], &cur_g->oid) ||
346 !hasheq(oids[n].hash, g->chunk_base_graphs + g->hash_len * n)) {
347 warning(_("commit-graph chain does not match"));
351 cur_g = cur_g->base_graph;
354 g->base_graph = chain;
357 g->num_commits_in_base = chain->num_commits + chain->num_commits_in_base;
362 static struct commit_graph *load_commit_graph_chain(struct repository *r, const char *obj_dir)
364 struct commit_graph *graph_chain = NULL;
365 struct strbuf line = STRBUF_INIT;
367 struct object_id *oids;
368 int i = 0, valid = 1, count;
369 char *chain_name = get_chain_filename(obj_dir);
373 fp = fopen(chain_name, "r");
374 stat_res = stat(chain_name, &st);
379 st.st_size <= the_hash_algo->hexsz)
382 count = st.st_size / (the_hash_algo->hexsz + 1);
383 oids = xcalloc(count, sizeof(struct object_id));
387 for (i = 0; i < count; i++) {
388 struct object_directory *odb;
390 if (strbuf_getline_lf(&line, fp) == EOF)
393 if (get_oid_hex(line.buf, &oids[i])) {
394 warning(_("invalid commit-graph chain: line '%s' not a hash"),
401 for (odb = r->objects->odb; odb; odb = odb->next) {
402 char *graph_name = get_split_graph_filename(odb->path, line.buf);
403 struct commit_graph *g = load_commit_graph_one(graph_name);
408 g->obj_dir = odb->path;
410 if (add_graph_to_chain(g, graph_chain, oids, i)) {
420 warning(_("unable to find all commit-graph files"));
431 static struct commit_graph *read_commit_graph_one(struct repository *r, const char *obj_dir)
433 struct commit_graph *g = load_commit_graph_v1(r, obj_dir);
436 g = load_commit_graph_chain(r, obj_dir);
441 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
444 if (r->objects->commit_graph)
447 r->objects->commit_graph = read_commit_graph_one(r, obj_dir);
451 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
453 * On the first invocation, this function attemps to load the commit
454 * graph if the_repository is configured to have one.
456 static int prepare_commit_graph(struct repository *r)
458 struct object_directory *odb;
461 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
462 die("dying as requested by the '%s' variable on commit-graph load!",
463 GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD);
465 if (r->objects->commit_graph_attempted)
466 return !!r->objects->commit_graph;
467 r->objects->commit_graph_attempted = 1;
469 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
470 (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
473 * This repository is not configured to use commit graphs, so
474 * do not load one. (But report commit_graph_attempted anyway
475 * so that commit graph loading is not attempted again for this
480 if (!commit_graph_compatible(r))
484 for (odb = r->objects->odb;
485 !r->objects->commit_graph && odb;
487 prepare_commit_graph_one(r, odb->path);
488 return !!r->objects->commit_graph;
491 int generation_numbers_enabled(struct repository *r)
493 uint32_t first_generation;
494 struct commit_graph *g;
495 if (!prepare_commit_graph(r))
498 g = r->objects->commit_graph;
503 first_generation = get_be32(g->chunk_commit_data +
504 g->hash_len + 8) >> 2;
506 return !!first_generation;
509 static void close_commit_graph_one(struct commit_graph *g)
514 close_commit_graph_one(g->base_graph);
515 free_commit_graph(g);
518 void close_commit_graph(struct raw_object_store *o)
520 close_commit_graph_one(o->commit_graph);
521 o->commit_graph = NULL;
524 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
526 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
527 g->chunk_oid_lookup, g->hash_len, pos);
530 static void load_oid_from_graph(struct commit_graph *g,
532 struct object_id *oid)
536 while (g && pos < g->num_commits_in_base)
540 BUG("NULL commit-graph");
542 if (pos >= g->num_commits + g->num_commits_in_base)
543 die(_("invalid commit position. commit-graph is likely corrupt"));
545 lex_index = pos - g->num_commits_in_base;
547 hashcpy(oid->hash, g->chunk_oid_lookup + g->hash_len * lex_index);
550 static struct commit_list **insert_parent_or_die(struct repository *r,
551 struct commit_graph *g,
553 struct commit_list **pptr)
556 struct object_id oid;
558 if (pos >= g->num_commits + g->num_commits_in_base)
559 die("invalid parent position %"PRIu32, pos);
561 load_oid_from_graph(g, pos, &oid);
562 c = lookup_commit(r, &oid);
564 die(_("could not find commit %s"), oid_to_hex(&oid));
566 return &commit_list_insert(c, pptr)->next;
569 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
571 const unsigned char *commit_data;
574 while (pos < g->num_commits_in_base)
577 lex_index = pos - g->num_commits_in_base;
578 commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * lex_index;
579 item->graph_pos = pos;
580 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
583 static int fill_commit_in_graph(struct repository *r,
585 struct commit_graph *g, uint32_t pos)
588 uint32_t *parent_data_ptr;
589 uint64_t date_low, date_high;
590 struct commit_list **pptr;
591 const unsigned char *commit_data;
594 while (pos < g->num_commits_in_base)
597 if (pos >= g->num_commits + g->num_commits_in_base)
598 die(_("invalid commit position. commit-graph is likely corrupt"));
601 * Store the "full" position, but then use the
602 * "local" position for the rest of the calculation.
604 item->graph_pos = pos;
605 lex_index = pos - g->num_commits_in_base;
607 commit_data = g->chunk_commit_data + (g->hash_len + 16) * lex_index;
609 item->object.parsed = 1;
611 item->maybe_tree = NULL;
613 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
614 date_low = get_be32(commit_data + g->hash_len + 12);
615 item->date = (timestamp_t)((date_high << 32) | date_low);
617 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
619 pptr = &item->parents;
621 edge_value = get_be32(commit_data + g->hash_len);
622 if (edge_value == GRAPH_PARENT_NONE)
624 pptr = insert_parent_or_die(r, g, edge_value, pptr);
626 edge_value = get_be32(commit_data + g->hash_len + 4);
627 if (edge_value == GRAPH_PARENT_NONE)
629 if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
630 pptr = insert_parent_or_die(r, g, edge_value, pptr);
634 parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
635 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
637 edge_value = get_be32(parent_data_ptr);
638 pptr = insert_parent_or_die(r, g,
639 edge_value & GRAPH_EDGE_LAST_MASK,
642 } while (!(edge_value & GRAPH_LAST_EDGE));
647 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
649 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
650 *pos = item->graph_pos;
653 struct commit_graph *cur_g = g;
656 while (cur_g && !bsearch_graph(cur_g, &(item->object.oid), &lex_index))
657 cur_g = cur_g->base_graph;
660 *pos = lex_index + cur_g->num_commits_in_base;
668 static int parse_commit_in_graph_one(struct repository *r,
669 struct commit_graph *g,
674 if (item->object.parsed)
677 if (find_commit_in_graph(item, g, &pos))
678 return fill_commit_in_graph(r, item, g, pos);
683 int parse_commit_in_graph(struct repository *r, struct commit *item)
685 if (!prepare_commit_graph(r))
687 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
690 void load_commit_graph_info(struct repository *r, struct commit *item)
693 if (!prepare_commit_graph(r))
695 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
696 fill_commit_graph_info(item, r->objects->commit_graph, pos);
699 static struct tree *load_tree_for_commit(struct repository *r,
700 struct commit_graph *g,
703 struct object_id oid;
704 const unsigned char *commit_data;
706 while (c->graph_pos < g->num_commits_in_base)
709 commit_data = g->chunk_commit_data +
710 GRAPH_DATA_WIDTH * (c->graph_pos - g->num_commits_in_base);
712 hashcpy(oid.hash, commit_data);
713 c->maybe_tree = lookup_tree(r, &oid);
715 return c->maybe_tree;
718 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
719 struct commit_graph *g,
720 const struct commit *c)
723 return c->maybe_tree;
724 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
725 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
727 return load_tree_for_commit(r, g, (struct commit *)c);
730 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
732 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
735 struct packed_commit_list {
736 struct commit **list;
741 struct packed_oid_list {
742 struct object_id *list;
747 struct write_commit_graph_context {
748 struct repository *r;
751 struct packed_oid_list oids;
752 struct packed_commit_list commits;
754 unsigned long approx_nr_objects;
755 struct progress *progress;
757 uint64_t progress_cnt;
759 char *base_graph_name;
760 int num_commit_graphs_before;
761 int num_commit_graphs_after;
762 char **commit_graph_filenames_before;
763 char **commit_graph_filenames_after;
764 char **commit_graph_hash_after;
765 uint32_t new_num_commits_in_base;
766 struct commit_graph *new_base_graph;
772 const struct split_commit_graph_opts *split_opts;
775 static void write_graph_chunk_fanout(struct hashfile *f,
776 struct write_commit_graph_context *ctx)
779 struct commit **list = ctx->commits.list;
782 * Write the first-level table (the list is sorted,
783 * but we use a 256-entry lookup to be able to avoid
784 * having to do eight extra binary search iterations).
786 for (i = 0; i < 256; i++) {
787 while (count < ctx->commits.nr) {
788 if ((*list)->object.oid.hash[0] != i)
790 display_progress(ctx->progress, ++ctx->progress_cnt);
795 hashwrite_be32(f, count);
799 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
800 struct write_commit_graph_context *ctx)
802 struct commit **list = ctx->commits.list;
804 for (count = 0; count < ctx->commits.nr; count++, list++) {
805 display_progress(ctx->progress, ++ctx->progress_cnt);
806 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
810 static const unsigned char *commit_to_sha1(size_t index, void *table)
812 struct commit **commits = table;
813 return commits[index]->object.oid.hash;
816 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
817 struct write_commit_graph_context *ctx)
819 struct commit **list = ctx->commits.list;
820 struct commit **last = ctx->commits.list + ctx->commits.nr;
821 uint32_t num_extra_edges = 0;
823 while (list < last) {
824 struct commit_list *parent;
826 uint32_t packedDate[2];
827 display_progress(ctx->progress, ++ctx->progress_cnt);
829 parse_commit_no_graph(*list);
830 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
832 parent = (*list)->parents;
835 edge_value = GRAPH_PARENT_NONE;
837 edge_value = sha1_pos(parent->item->object.oid.hash,
843 edge_value += ctx->new_num_commits_in_base;
846 if (find_commit_in_graph(parent->item,
853 BUG("missing parent %s for commit %s",
854 oid_to_hex(&parent->item->object.oid),
855 oid_to_hex(&(*list)->object.oid));
858 hashwrite_be32(f, edge_value);
861 parent = parent->next;
864 edge_value = GRAPH_PARENT_NONE;
865 else if (parent->next)
866 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
868 edge_value = sha1_pos(parent->item->object.oid.hash,
874 edge_value += ctx->new_num_commits_in_base;
877 if (find_commit_in_graph(parent->item,
884 BUG("missing parent %s for commit %s",
885 oid_to_hex(&parent->item->object.oid),
886 oid_to_hex(&(*list)->object.oid));
889 hashwrite_be32(f, edge_value);
891 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
894 parent = parent->next;
898 if (sizeof((*list)->date) > 4)
899 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
903 packedDate[0] |= htonl((*list)->generation << 2);
905 packedDate[1] = htonl((*list)->date);
906 hashwrite(f, packedDate, 8);
912 static void write_graph_chunk_extra_edges(struct hashfile *f,
913 struct write_commit_graph_context *ctx)
915 struct commit **list = ctx->commits.list;
916 struct commit **last = ctx->commits.list + ctx->commits.nr;
917 struct commit_list *parent;
919 while (list < last) {
922 display_progress(ctx->progress, ++ctx->progress_cnt);
924 for (parent = (*list)->parents; num_parents < 3 && parent;
925 parent = parent->next)
928 if (num_parents <= 2) {
933 /* Since num_parents > 2, this initializer is safe. */
934 for (parent = (*list)->parents->next; parent; parent = parent->next) {
935 int edge_value = sha1_pos(parent->item->object.oid.hash,
941 edge_value += ctx->new_num_commits_in_base;
944 if (find_commit_in_graph(parent->item,
951 BUG("missing parent %s for commit %s",
952 oid_to_hex(&parent->item->object.oid),
953 oid_to_hex(&(*list)->object.oid));
954 else if (!parent->next)
955 edge_value |= GRAPH_LAST_EDGE;
957 hashwrite_be32(f, edge_value);
964 static int oid_compare(const void *_a, const void *_b)
966 const struct object_id *a = (const struct object_id *)_a;
967 const struct object_id *b = (const struct object_id *)_b;
971 static int add_packed_commits(const struct object_id *oid,
972 struct packed_git *pack,
976 struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
977 enum object_type type;
978 off_t offset = nth_packed_object_offset(pack, pos);
979 struct object_info oi = OBJECT_INFO_INIT;
982 display_progress(ctx->progress, ++ctx->progress_done);
985 if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
986 die(_("unable to get type of object %s"), oid_to_hex(oid));
988 if (type != OBJ_COMMIT)
991 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
992 oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
998 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
1000 struct commit_list *parent;
1001 for (parent = commit->parents; parent; parent = parent->next) {
1002 if (!(parent->item->object.flags & UNINTERESTING)) {
1003 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1004 oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
1006 parent->item->object.flags |= UNINTERESTING;
1011 static void close_reachable(struct write_commit_graph_context *ctx)
1014 struct commit *commit;
1016 if (ctx->report_progress)
1017 ctx->progress = start_delayed_progress(
1018 _("Loading known commits in commit graph"),
1020 for (i = 0; i < ctx->oids.nr; i++) {
1021 display_progress(ctx->progress, i + 1);
1022 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1024 commit->object.flags |= UNINTERESTING;
1026 stop_progress(&ctx->progress);
1029 * As this loop runs, ctx->oids.nr may grow, but not more
1030 * than the number of missing commits in the reachable
1033 if (ctx->report_progress)
1034 ctx->progress = start_delayed_progress(
1035 _("Expanding reachable commits in commit graph"),
1037 for (i = 0; i < ctx->oids.nr; i++) {
1038 display_progress(ctx->progress, i + 1);
1039 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1044 if (!parse_commit(commit) &&
1045 commit->graph_pos == COMMIT_NOT_FROM_GRAPH)
1046 add_missing_parents(ctx, commit);
1047 } else if (!parse_commit_no_graph(commit))
1048 add_missing_parents(ctx, commit);
1050 stop_progress(&ctx->progress);
1052 if (ctx->report_progress)
1053 ctx->progress = start_delayed_progress(
1054 _("Clearing commit marks in commit graph"),
1056 for (i = 0; i < ctx->oids.nr; i++) {
1057 display_progress(ctx->progress, i + 1);
1058 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1061 commit->object.flags &= ~UNINTERESTING;
1063 stop_progress(&ctx->progress);
1066 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1069 struct commit_list *list = NULL;
1071 if (ctx->report_progress)
1072 ctx->progress = start_progress(
1073 _("Computing commit graph generation numbers"),
1075 for (i = 0; i < ctx->commits.nr; i++) {
1076 display_progress(ctx->progress, i + 1);
1077 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
1078 ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
1081 commit_list_insert(ctx->commits.list[i], &list);
1083 struct commit *current = list->item;
1084 struct commit_list *parent;
1085 int all_parents_computed = 1;
1086 uint32_t max_generation = 0;
1088 for (parent = current->parents; parent; parent = parent->next) {
1089 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
1090 parent->item->generation == GENERATION_NUMBER_ZERO) {
1091 all_parents_computed = 0;
1092 commit_list_insert(parent->item, &list);
1094 } else if (parent->item->generation > max_generation) {
1095 max_generation = parent->item->generation;
1099 if (all_parents_computed) {
1100 current->generation = max_generation + 1;
1103 if (current->generation > GENERATION_NUMBER_MAX)
1104 current->generation = GENERATION_NUMBER_MAX;
1108 stop_progress(&ctx->progress);
1111 static int add_ref_to_list(const char *refname,
1112 const struct object_id *oid,
1113 int flags, void *cb_data)
1115 struct string_list *list = (struct string_list *)cb_data;
1117 string_list_append(list, oid_to_hex(oid));
1121 int write_commit_graph_reachable(const char *obj_dir, unsigned int flags,
1122 const struct split_commit_graph_opts *split_opts)
1124 struct string_list list = STRING_LIST_INIT_DUP;
1127 for_each_ref(add_ref_to_list, &list);
1128 result = write_commit_graph(obj_dir, NULL, &list,
1131 string_list_clear(&list, 0);
1135 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
1136 struct string_list *pack_indexes)
1139 struct strbuf progress_title = STRBUF_INIT;
1140 struct strbuf packname = STRBUF_INIT;
1143 strbuf_addf(&packname, "%s/pack/", ctx->obj_dir);
1144 dirlen = packname.len;
1145 if (ctx->report_progress) {
1146 strbuf_addf(&progress_title,
1147 Q_("Finding commits for commit graph in %d pack",
1148 "Finding commits for commit graph in %d packs",
1151 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1152 ctx->progress_done = 0;
1154 for (i = 0; i < pack_indexes->nr; i++) {
1155 struct packed_git *p;
1156 strbuf_setlen(&packname, dirlen);
1157 strbuf_addstr(&packname, pack_indexes->items[i].string);
1158 p = add_packed_git(packname.buf, packname.len, 1);
1160 error(_("error adding pack %s"), packname.buf);
1163 if (open_pack_index(p)) {
1164 error(_("error opening index for %s"), packname.buf);
1167 for_each_object_in_pack(p, add_packed_commits, ctx,
1168 FOR_EACH_OBJECT_PACK_ORDER);
1173 stop_progress(&ctx->progress);
1174 strbuf_reset(&progress_title);
1175 strbuf_release(&packname);
1180 static void fill_oids_from_commit_hex(struct write_commit_graph_context *ctx,
1181 struct string_list *commit_hex)
1184 struct strbuf progress_title = STRBUF_INIT;
1186 if (ctx->report_progress) {
1187 strbuf_addf(&progress_title,
1188 Q_("Finding commits for commit graph from %d ref",
1189 "Finding commits for commit graph from %d refs",
1192 ctx->progress = start_delayed_progress(
1196 for (i = 0; i < commit_hex->nr; i++) {
1198 struct object_id oid;
1199 struct commit *result;
1201 display_progress(ctx->progress, i + 1);
1202 if (commit_hex->items[i].string &&
1203 parse_oid_hex(commit_hex->items[i].string, &oid, &end))
1206 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1209 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1210 oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
1214 stop_progress(&ctx->progress);
1215 strbuf_release(&progress_title);
1218 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1220 if (ctx->report_progress)
1221 ctx->progress = start_delayed_progress(
1222 _("Finding commits for commit graph among packed objects"),
1223 ctx->approx_nr_objects);
1224 for_each_packed_object(add_packed_commits, ctx,
1225 FOR_EACH_OBJECT_PACK_ORDER);
1226 if (ctx->progress_done < ctx->approx_nr_objects)
1227 display_progress(ctx->progress, ctx->approx_nr_objects);
1228 stop_progress(&ctx->progress);
1231 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
1233 uint32_t i, count_distinct = 1;
1235 if (ctx->report_progress)
1236 ctx->progress = start_delayed_progress(
1237 _("Counting distinct commits in commit graph"),
1239 display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1240 QSORT(ctx->oids.list, ctx->oids.nr, oid_compare);
1242 for (i = 1; i < ctx->oids.nr; i++) {
1243 display_progress(ctx->progress, i + 1);
1244 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i])) {
1246 struct commit *c = lookup_commit(ctx->r, &ctx->oids.list[i]);
1248 if (!c || c->graph_pos != COMMIT_NOT_FROM_GRAPH)
1255 stop_progress(&ctx->progress);
1257 return count_distinct;
1260 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1263 struct commit_list *parent;
1265 ctx->num_extra_edges = 0;
1266 if (ctx->report_progress)
1267 ctx->progress = start_delayed_progress(
1268 _("Finding extra edges in commit graph"),
1270 for (i = 0; i < ctx->oids.nr; i++) {
1271 int num_parents = 0;
1272 display_progress(ctx->progress, i + 1);
1273 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1276 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + 1, ctx->commits.alloc);
1277 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1280 ctx->commits.list[ctx->commits.nr]->graph_pos != COMMIT_NOT_FROM_GRAPH)
1283 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1285 for (parent = ctx->commits.list[ctx->commits.nr]->parents;
1286 parent; parent = parent->next)
1289 if (num_parents > 2)
1290 ctx->num_extra_edges += num_parents - 1;
1294 stop_progress(&ctx->progress);
1297 static int write_graph_chunk_base_1(struct hashfile *f,
1298 struct commit_graph *g)
1305 num = write_graph_chunk_base_1(f, g->base_graph);
1306 hashwrite(f, g->oid.hash, the_hash_algo->rawsz);
1310 static int write_graph_chunk_base(struct hashfile *f,
1311 struct write_commit_graph_context *ctx)
1313 int num = write_graph_chunk_base_1(f, ctx->new_base_graph);
1315 if (num != ctx->num_commit_graphs_after - 1) {
1316 error(_("failed to write correct number of base graph ids"));
1323 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1328 struct lock_file lk = LOCK_INIT;
1329 uint32_t chunk_ids[6];
1330 uint64_t chunk_offsets[6];
1331 const unsigned hashsz = the_hash_algo->rawsz;
1332 struct strbuf progress_title = STRBUF_INIT;
1334 struct object_id file_hash;
1337 struct strbuf tmp_file = STRBUF_INIT;
1339 strbuf_addf(&tmp_file,
1340 "%s/info/commit-graphs/tmp_graph_XXXXXX",
1342 ctx->graph_name = strbuf_detach(&tmp_file, NULL);
1344 ctx->graph_name = get_commit_graph_filename(ctx->obj_dir);
1347 if (safe_create_leading_directories(ctx->graph_name)) {
1348 UNLEAK(ctx->graph_name);
1349 error(_("unable to create leading directories of %s"),
1355 char *lock_name = get_chain_filename(ctx->obj_dir);
1357 hold_lock_file_for_update(&lk, lock_name, LOCK_DIE_ON_ERROR);
1359 fd = git_mkstemp_mode(ctx->graph_name, 0444);
1361 error(_("unable to create '%s'"), ctx->graph_name);
1365 f = hashfd(fd, ctx->graph_name);
1367 hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
1368 fd = lk.tempfile->fd;
1369 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1372 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1373 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1374 chunk_ids[2] = GRAPH_CHUNKID_DATA;
1375 if (ctx->num_extra_edges) {
1376 chunk_ids[num_chunks] = GRAPH_CHUNKID_EXTRAEDGES;
1379 if (ctx->num_commit_graphs_after > 1) {
1380 chunk_ids[num_chunks] = GRAPH_CHUNKID_BASE;
1384 chunk_ids[num_chunks] = 0;
1386 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1387 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1388 chunk_offsets[2] = chunk_offsets[1] + hashsz * ctx->commits.nr;
1389 chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * ctx->commits.nr;
1392 if (ctx->num_extra_edges) {
1393 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1394 4 * ctx->num_extra_edges;
1397 if (ctx->num_commit_graphs_after > 1) {
1398 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1399 hashsz * (ctx->num_commit_graphs_after - 1);
1403 hashwrite_be32(f, GRAPH_SIGNATURE);
1405 hashwrite_u8(f, GRAPH_VERSION);
1406 hashwrite_u8(f, oid_version());
1407 hashwrite_u8(f, num_chunks);
1408 hashwrite_u8(f, ctx->num_commit_graphs_after - 1);
1410 for (i = 0; i <= num_chunks; i++) {
1411 uint32_t chunk_write[3];
1413 chunk_write[0] = htonl(chunk_ids[i]);
1414 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1415 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1416 hashwrite(f, chunk_write, 12);
1419 if (ctx->report_progress) {
1420 strbuf_addf(&progress_title,
1421 Q_("Writing out commit graph in %d pass",
1422 "Writing out commit graph in %d passes",
1425 ctx->progress = start_delayed_progress(
1427 num_chunks * ctx->commits.nr);
1429 write_graph_chunk_fanout(f, ctx);
1430 write_graph_chunk_oids(f, hashsz, ctx);
1431 write_graph_chunk_data(f, hashsz, ctx);
1432 if (ctx->num_extra_edges)
1433 write_graph_chunk_extra_edges(f, ctx);
1434 if (ctx->num_commit_graphs_after > 1 &&
1435 write_graph_chunk_base(f, ctx)) {
1438 stop_progress(&ctx->progress);
1439 strbuf_release(&progress_title);
1441 if (ctx->split && ctx->base_graph_name && ctx->num_commit_graphs_after > 1) {
1442 char *new_base_hash = xstrdup(oid_to_hex(&ctx->new_base_graph->oid));
1443 char *new_base_name = get_split_graph_filename(ctx->new_base_graph->obj_dir, new_base_hash);
1445 free(ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2]);
1446 free(ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2]);
1447 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2] = new_base_name;
1448 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2] = new_base_hash;
1451 close_commit_graph(ctx->r->objects);
1452 finalize_hashfile(f, file_hash.hash, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1455 FILE *chainf = fdopen_lock_file(&lk, "w");
1456 char *final_graph_name;
1462 error(_("unable to open commit-graph chain file"));
1466 if (ctx->base_graph_name) {
1467 const char *dest = ctx->commit_graph_filenames_after[
1468 ctx->num_commit_graphs_after - 2];
1470 if (strcmp(ctx->base_graph_name, dest)) {
1471 result = rename(ctx->base_graph_name, dest);
1474 error(_("failed to rename base commit-graph file"));
1479 char *graph_name = get_commit_graph_filename(ctx->obj_dir);
1483 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1] = xstrdup(oid_to_hex(&file_hash));
1484 final_graph_name = get_split_graph_filename(ctx->obj_dir,
1485 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1]);
1486 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 1] = final_graph_name;
1488 result = rename(ctx->graph_name, final_graph_name);
1490 for (i = 0; i < ctx->num_commit_graphs_after; i++)
1491 fprintf(lk.tempfile->fp, "%s\n", ctx->commit_graph_hash_after[i]);
1494 error(_("failed to rename temporary commit-graph file"));
1499 commit_lock_file(&lk);
1504 static void split_graph_merge_strategy(struct write_commit_graph_context *ctx)
1506 struct commit_graph *g = ctx->r->objects->commit_graph;
1507 uint32_t num_commits = ctx->commits.nr;
1510 int max_commits = 0;
1513 if (ctx->split_opts) {
1514 max_commits = ctx->split_opts->max_commits;
1515 size_mult = ctx->split_opts->size_multiple;
1518 g = ctx->r->objects->commit_graph;
1519 ctx->num_commit_graphs_after = ctx->num_commit_graphs_before + 1;
1521 while (g && (g->num_commits <= size_mult * num_commits ||
1522 (max_commits && num_commits > max_commits))) {
1523 if (strcmp(g->obj_dir, ctx->obj_dir))
1526 num_commits += g->num_commits;
1529 ctx->num_commit_graphs_after--;
1532 ctx->new_base_graph = g;
1534 if (ctx->num_commit_graphs_after == 2) {
1535 char *old_graph_name = get_commit_graph_filename(g->obj_dir);
1537 if (!strcmp(g->filename, old_graph_name) &&
1538 strcmp(g->obj_dir, ctx->obj_dir)) {
1539 ctx->num_commit_graphs_after = 1;
1540 ctx->new_base_graph = NULL;
1543 free(old_graph_name);
1546 ALLOC_ARRAY(ctx->commit_graph_filenames_after, ctx->num_commit_graphs_after);
1547 ALLOC_ARRAY(ctx->commit_graph_hash_after, ctx->num_commit_graphs_after);
1549 for (i = 0; i < ctx->num_commit_graphs_after &&
1550 i < ctx->num_commit_graphs_before; i++)
1551 ctx->commit_graph_filenames_after[i] = xstrdup(ctx->commit_graph_filenames_before[i]);
1553 i = ctx->num_commit_graphs_before - 1;
1554 g = ctx->r->objects->commit_graph;
1557 if (i < ctx->num_commit_graphs_after)
1558 ctx->commit_graph_hash_after[i] = xstrdup(oid_to_hex(&g->oid));
1565 static void merge_commit_graph(struct write_commit_graph_context *ctx,
1566 struct commit_graph *g)
1569 uint32_t offset = g->num_commits_in_base;
1571 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + g->num_commits, ctx->commits.alloc);
1573 for (i = 0; i < g->num_commits; i++) {
1574 struct object_id oid;
1575 struct commit *result;
1577 display_progress(ctx->progress, i + 1);
1579 load_oid_from_graph(g, i + offset, &oid);
1581 /* only add commits if they still exist in the repo */
1582 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1585 ctx->commits.list[ctx->commits.nr] = result;
1591 static int commit_compare(const void *_a, const void *_b)
1593 const struct commit *a = *(const struct commit **)_a;
1594 const struct commit *b = *(const struct commit **)_b;
1595 return oidcmp(&a->object.oid, &b->object.oid);
1598 static void sort_and_scan_merged_commits(struct write_commit_graph_context *ctx)
1600 uint32_t i, num_parents;
1601 struct commit_list *parent;
1603 if (ctx->report_progress)
1604 ctx->progress = start_delayed_progress(
1605 _("Scanning merged commits"),
1608 QSORT(ctx->commits.list, ctx->commits.nr, commit_compare);
1610 ctx->num_extra_edges = 0;
1611 for (i = 0; i < ctx->commits.nr; i++) {
1612 display_progress(ctx->progress, i);
1614 if (i && oideq(&ctx->commits.list[i - 1]->object.oid,
1615 &ctx->commits.list[i]->object.oid)) {
1616 die(_("unexpected duplicate commit id %s"),
1617 oid_to_hex(&ctx->commits.list[i]->object.oid));
1620 for (parent = ctx->commits.list[i]->parents; parent; parent = parent->next)
1623 if (num_parents > 2)
1624 ctx->num_extra_edges += num_parents - 2;
1628 stop_progress(&ctx->progress);
1631 static void merge_commit_graphs(struct write_commit_graph_context *ctx)
1633 struct commit_graph *g = ctx->r->objects->commit_graph;
1634 uint32_t current_graph_number = ctx->num_commit_graphs_before;
1635 struct strbuf progress_title = STRBUF_INIT;
1637 while (g && current_graph_number >= ctx->num_commit_graphs_after) {
1638 current_graph_number--;
1640 if (ctx->report_progress) {
1641 strbuf_addstr(&progress_title, _("Merging commit-graph"));
1642 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1645 merge_commit_graph(ctx, g);
1646 stop_progress(&ctx->progress);
1647 strbuf_release(&progress_title);
1653 ctx->new_base_graph = g;
1654 ctx->new_num_commits_in_base = g->num_commits + g->num_commits_in_base;
1657 if (ctx->new_base_graph)
1658 ctx->base_graph_name = xstrdup(ctx->new_base_graph->filename);
1660 sort_and_scan_merged_commits(ctx);
1663 static void mark_commit_graphs(struct write_commit_graph_context *ctx)
1666 time_t now = time(NULL);
1668 for (i = ctx->num_commit_graphs_after - 1; i < ctx->num_commit_graphs_before; i++) {
1670 struct utimbuf updated_time;
1672 stat(ctx->commit_graph_filenames_before[i], &st);
1674 updated_time.actime = st.st_atime;
1675 updated_time.modtime = now;
1676 utime(ctx->commit_graph_filenames_before[i], &updated_time);
1680 static void expire_commit_graphs(struct write_commit_graph_context *ctx)
1682 struct strbuf path = STRBUF_INIT;
1686 timestamp_t expire_time = time(NULL);
1688 if (ctx->split_opts && ctx->split_opts->expire_time)
1689 expire_time -= ctx->split_opts->expire_time;
1691 strbuf_addstr(&path, ctx->obj_dir);
1692 strbuf_addstr(&path, "/info/commit-graphs");
1693 dir = opendir(path.buf);
1696 strbuf_release(&path);
1700 strbuf_addch(&path, '/');
1701 dirnamelen = path.len;
1702 while ((de = readdir(dir)) != NULL) {
1704 uint32_t i, found = 0;
1706 strbuf_setlen(&path, dirnamelen);
1707 strbuf_addstr(&path, de->d_name);
1709 stat(path.buf, &st);
1711 if (st.st_mtime > expire_time)
1713 if (path.len < 6 || strcmp(path.buf + path.len - 6, ".graph"))
1716 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
1717 if (!strcmp(ctx->commit_graph_filenames_after[i],
1730 int write_commit_graph(const char *obj_dir,
1731 struct string_list *pack_indexes,
1732 struct string_list *commit_hex,
1734 const struct split_commit_graph_opts *split_opts)
1736 struct write_commit_graph_context *ctx;
1737 uint32_t i, count_distinct = 0;
1740 if (!commit_graph_compatible(the_repository))
1743 ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
1744 ctx->r = the_repository;
1745 ctx->obj_dir = obj_dir;
1746 ctx->append = flags & COMMIT_GRAPH_APPEND ? 1 : 0;
1747 ctx->report_progress = flags & COMMIT_GRAPH_PROGRESS ? 1 : 0;
1748 ctx->split = flags & COMMIT_GRAPH_SPLIT ? 1 : 0;
1749 ctx->split_opts = split_opts;
1752 struct commit_graph *g;
1753 prepare_commit_graph(ctx->r);
1755 g = ctx->r->objects->commit_graph;
1758 ctx->num_commit_graphs_before++;
1762 if (ctx->num_commit_graphs_before) {
1763 ALLOC_ARRAY(ctx->commit_graph_filenames_before, ctx->num_commit_graphs_before);
1764 i = ctx->num_commit_graphs_before;
1765 g = ctx->r->objects->commit_graph;
1768 ctx->commit_graph_filenames_before[--i] = xstrdup(g->filename);
1774 ctx->approx_nr_objects = approximate_object_count();
1775 ctx->oids.alloc = ctx->approx_nr_objects / 32;
1777 if (ctx->split && split_opts && ctx->oids.alloc > split_opts->max_commits)
1778 ctx->oids.alloc = split_opts->max_commits;
1781 prepare_commit_graph_one(ctx->r, ctx->obj_dir);
1782 if (ctx->r->objects->commit_graph)
1783 ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
1786 if (ctx->oids.alloc < 1024)
1787 ctx->oids.alloc = 1024;
1788 ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
1790 if (ctx->append && ctx->r->objects->commit_graph) {
1791 struct commit_graph *g = ctx->r->objects->commit_graph;
1792 for (i = 0; i < g->num_commits; i++) {
1793 const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
1794 hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
1799 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
1804 fill_oids_from_commit_hex(ctx, commit_hex);
1806 if (!pack_indexes && !commit_hex)
1807 fill_oids_from_all_packs(ctx);
1809 close_reachable(ctx);
1811 count_distinct = count_distinct_commits(ctx);
1813 if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
1814 error(_("the commit graph format cannot write %d commits"), count_distinct);
1819 ctx->commits.alloc = count_distinct;
1820 ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
1822 copy_oids_to_commits(ctx);
1824 if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
1825 error(_("too many commits to write graph"));
1830 if (!ctx->commits.nr)
1834 split_graph_merge_strategy(ctx);
1836 merge_commit_graphs(ctx);
1838 ctx->num_commit_graphs_after = 1;
1840 compute_generation_numbers(ctx);
1842 res = write_commit_graph_file(ctx);
1845 mark_commit_graphs(ctx);
1846 expire_commit_graphs(ctx);
1850 free(ctx->graph_name);
1851 free(ctx->commits.list);
1852 free(ctx->oids.list);
1854 if (ctx->commit_graph_filenames_after) {
1855 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
1856 free(ctx->commit_graph_filenames_after[i]);
1857 free(ctx->commit_graph_hash_after[i]);
1860 for (i = 0; i < ctx->num_commit_graphs_before; i++)
1861 free(ctx->commit_graph_filenames_before[i]);
1863 free(ctx->commit_graph_filenames_after);
1864 free(ctx->commit_graph_filenames_before);
1865 free(ctx->commit_graph_hash_after);
1873 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1874 static int verify_commit_graph_error;
1876 static void graph_report(const char *fmt, ...)
1880 verify_commit_graph_error = 1;
1882 vfprintf(stderr, fmt, ap);
1883 fprintf(stderr, "\n");
1887 #define GENERATION_ZERO_EXISTS 1
1888 #define GENERATION_NUMBER_EXISTS 2
1890 int verify_commit_graph(struct repository *r, struct commit_graph *g)
1892 uint32_t i, cur_fanout_pos = 0;
1893 struct object_id prev_oid, cur_oid, checksum;
1894 int generation_zero = 0;
1897 struct progress *progress = NULL;
1900 graph_report("no commit-graph file loaded");
1904 verify_commit_graph_error = verify_commit_graph_lite(g);
1905 if (verify_commit_graph_error)
1906 return verify_commit_graph_error;
1908 devnull = open("/dev/null", O_WRONLY);
1909 f = hashfd(devnull, NULL);
1910 hashwrite(f, g->data, g->data_len - g->hash_len);
1911 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1912 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1913 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1914 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1917 for (i = 0; i < g->num_commits; i++) {
1918 struct commit *graph_commit;
1920 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1922 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1923 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
1924 oid_to_hex(&prev_oid),
1925 oid_to_hex(&cur_oid));
1927 oidcpy(&prev_oid, &cur_oid);
1929 while (cur_oid.hash[0] > cur_fanout_pos) {
1930 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1932 if (i != fanout_value)
1933 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1934 cur_fanout_pos, fanout_value, i);
1938 graph_commit = lookup_commit(r, &cur_oid);
1939 if (!parse_commit_in_graph_one(r, g, graph_commit))
1940 graph_report(_("failed to parse commit %s from commit-graph"),
1941 oid_to_hex(&cur_oid));
1944 while (cur_fanout_pos < 256) {
1945 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1947 if (g->num_commits != fanout_value)
1948 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1949 cur_fanout_pos, fanout_value, i);
1954 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1955 return verify_commit_graph_error;
1957 progress = start_progress(_("Verifying commits in commit graph"),
1959 for (i = 0; i < g->num_commits; i++) {
1960 struct commit *graph_commit, *odb_commit;
1961 struct commit_list *graph_parents, *odb_parents;
1962 uint32_t max_generation = 0;
1964 display_progress(progress, i + 1);
1965 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1967 graph_commit = lookup_commit(r, &cur_oid);
1968 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1969 if (parse_commit_internal(odb_commit, 0, 0)) {
1970 graph_report(_("failed to parse commit %s from object database for commit-graph"),
1971 oid_to_hex(&cur_oid));
1975 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
1976 get_commit_tree_oid(odb_commit)))
1977 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
1978 oid_to_hex(&cur_oid),
1979 oid_to_hex(get_commit_tree_oid(graph_commit)),
1980 oid_to_hex(get_commit_tree_oid(odb_commit)));
1982 graph_parents = graph_commit->parents;
1983 odb_parents = odb_commit->parents;
1985 while (graph_parents) {
1986 if (odb_parents == NULL) {
1987 graph_report(_("commit-graph parent list for commit %s is too long"),
1988 oid_to_hex(&cur_oid));
1992 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1993 graph_report(_("commit-graph parent for %s is %s != %s"),
1994 oid_to_hex(&cur_oid),
1995 oid_to_hex(&graph_parents->item->object.oid),
1996 oid_to_hex(&odb_parents->item->object.oid));
1998 if (graph_parents->item->generation > max_generation)
1999 max_generation = graph_parents->item->generation;
2001 graph_parents = graph_parents->next;
2002 odb_parents = odb_parents->next;
2005 if (odb_parents != NULL)
2006 graph_report(_("commit-graph parent list for commit %s terminates early"),
2007 oid_to_hex(&cur_oid));
2009 if (!graph_commit->generation) {
2010 if (generation_zero == GENERATION_NUMBER_EXISTS)
2011 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2012 oid_to_hex(&cur_oid));
2013 generation_zero = GENERATION_ZERO_EXISTS;
2014 } else if (generation_zero == GENERATION_ZERO_EXISTS)
2015 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2016 oid_to_hex(&cur_oid));
2018 if (generation_zero == GENERATION_ZERO_EXISTS)
2022 * If one of our parents has generation GENERATION_NUMBER_MAX, then
2023 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
2024 * extra logic in the following condition.
2026 if (max_generation == GENERATION_NUMBER_MAX)
2029 if (graph_commit->generation != max_generation + 1)
2030 graph_report(_("commit-graph generation for commit %s is %u != %u"),
2031 oid_to_hex(&cur_oid),
2032 graph_commit->generation,
2033 max_generation + 1);
2035 if (graph_commit->date != odb_commit->date)
2036 graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
2037 oid_to_hex(&cur_oid),
2041 stop_progress(&progress);
2043 return verify_commit_graph_error;
2046 void free_commit_graph(struct commit_graph *g)
2050 if (g->graph_fd >= 0) {
2051 munmap((void *)g->data, g->data_len);