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 /* Remember to update object flag allocation in object.h */
45 #define REACHABLE (1u<<15)
47 char *get_commit_graph_filename(const char *obj_dir)
49 char *filename = xstrfmt("%s/info/commit-graph", obj_dir);
50 char *normalized = xmalloc(strlen(filename) + 1);
51 normalize_path_copy(normalized, filename);
56 static char *get_split_graph_filename(const char *obj_dir,
59 char *filename = xstrfmt("%s/info/commit-graphs/graph-%s.graph",
62 char *normalized = xmalloc(strlen(filename) + 1);
63 normalize_path_copy(normalized, filename);
68 static char *get_chain_filename(const char *obj_dir)
70 return xstrfmt("%s/info/commit-graphs/commit-graph-chain", obj_dir);
73 static uint8_t oid_version(void)
78 static struct commit_graph *alloc_commit_graph(void)
80 struct commit_graph *g = xcalloc(1, sizeof(*g));
86 extern int read_replace_refs;
88 static int commit_graph_compatible(struct repository *r)
93 if (read_replace_refs) {
94 prepare_replace_object(r);
95 if (hashmap_get_size(&r->objects->replace_map->map))
99 prepare_commit_graft(r);
100 if (r->parsed_objects && r->parsed_objects->grafts_nr)
102 if (is_repository_shallow(r))
108 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
110 *fd = git_open(graph_file);
113 if (fstat(*fd, st)) {
120 struct commit_graph *load_commit_graph_one_fd_st(int fd, struct stat *st)
124 struct commit_graph *ret;
126 graph_size = xsize_t(st->st_size);
128 if (graph_size < GRAPH_MIN_SIZE) {
130 error(_("commit-graph file is too small"));
133 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
134 ret = parse_commit_graph(graph_map, fd, graph_size);
137 munmap(graph_map, graph_size);
144 static int verify_commit_graph_lite(struct commit_graph *g)
147 * Basic validation shared between parse_commit_graph()
148 * which'll be called every time the graph is used, and the
149 * much more expensive verify_commit_graph() used by
150 * "commit-graph verify".
152 * There should only be very basic checks here to ensure that
153 * we don't e.g. segfault in fill_commit_in_graph(), but
154 * because this is a very hot codepath nothing that e.g. loops
155 * over g->num_commits, or runs a checksum on the commit-graph
158 if (!g->chunk_oid_fanout) {
159 error("commit-graph is missing the OID Fanout chunk");
162 if (!g->chunk_oid_lookup) {
163 error("commit-graph is missing the OID Lookup chunk");
166 if (!g->chunk_commit_data) {
167 error("commit-graph is missing the Commit Data chunk");
174 struct commit_graph *parse_commit_graph(void *graph_map, int fd,
177 const unsigned char *data, *chunk_lookup;
179 struct commit_graph *graph;
180 uint64_t last_chunk_offset;
181 uint32_t last_chunk_id;
182 uint32_t graph_signature;
183 unsigned char graph_version, hash_version;
188 if (graph_size < GRAPH_MIN_SIZE)
191 data = (const unsigned char *)graph_map;
193 graph_signature = get_be32(data);
194 if (graph_signature != GRAPH_SIGNATURE) {
195 error(_("commit-graph signature %X does not match signature %X"),
196 graph_signature, GRAPH_SIGNATURE);
200 graph_version = *(unsigned char*)(data + 4);
201 if (graph_version != GRAPH_VERSION) {
202 error(_("commit-graph version %X does not match version %X"),
203 graph_version, GRAPH_VERSION);
207 hash_version = *(unsigned char*)(data + 5);
208 if (hash_version != oid_version()) {
209 error(_("commit-graph hash version %X does not match version %X"),
210 hash_version, oid_version());
214 graph = alloc_commit_graph();
216 graph->hash_len = the_hash_algo->rawsz;
217 graph->num_chunks = *(unsigned char*)(data + 6);
218 graph->graph_fd = fd;
219 graph->data = graph_map;
220 graph->data_len = graph_size;
223 last_chunk_offset = 8;
224 chunk_lookup = data + 8;
225 for (i = 0; i < graph->num_chunks; i++) {
227 uint64_t chunk_offset;
228 int chunk_repeated = 0;
230 if (data + graph_size - chunk_lookup <
231 GRAPH_CHUNKLOOKUP_WIDTH) {
232 error(_("commit-graph chunk lookup table entry missing; file may be incomplete"));
237 chunk_id = get_be32(chunk_lookup + 0);
238 chunk_offset = get_be64(chunk_lookup + 4);
240 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
242 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
243 error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
244 (uint32_t)chunk_offset);
250 case GRAPH_CHUNKID_OIDFANOUT:
251 if (graph->chunk_oid_fanout)
254 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
257 case GRAPH_CHUNKID_OIDLOOKUP:
258 if (graph->chunk_oid_lookup)
261 graph->chunk_oid_lookup = data + chunk_offset;
264 case GRAPH_CHUNKID_DATA:
265 if (graph->chunk_commit_data)
268 graph->chunk_commit_data = data + chunk_offset;
271 case GRAPH_CHUNKID_EXTRAEDGES:
272 if (graph->chunk_extra_edges)
275 graph->chunk_extra_edges = data + chunk_offset;
278 case GRAPH_CHUNKID_BASE:
279 if (graph->chunk_base_graphs)
282 graph->chunk_base_graphs = data + chunk_offset;
285 if (chunk_repeated) {
286 error(_("commit-graph chunk id %08x appears multiple times"), chunk_id);
291 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
293 graph->num_commits = (chunk_offset - last_chunk_offset)
297 last_chunk_id = chunk_id;
298 last_chunk_offset = chunk_offset;
301 hashcpy(graph->oid.hash, graph->data + graph->data_len - graph->hash_len);
303 if (verify_commit_graph_lite(graph)) {
311 static struct commit_graph *load_commit_graph_one(const char *graph_file)
316 struct commit_graph *g;
317 int open_ok = open_commit_graph(graph_file, &fd, &st);
322 g = load_commit_graph_one_fd_st(fd, &st);
325 g->filename = xstrdup(graph_file);
330 static struct commit_graph *load_commit_graph_v1(struct repository *r, const char *obj_dir)
332 char *graph_name = get_commit_graph_filename(obj_dir);
333 struct commit_graph *g = load_commit_graph_one(graph_name);
337 g->obj_dir = obj_dir;
342 static int add_graph_to_chain(struct commit_graph *g,
343 struct commit_graph *chain,
344 struct object_id *oids,
347 struct commit_graph *cur_g = chain;
349 if (n && !g->chunk_base_graphs) {
350 warning(_("commit-graph has no base graphs chunk"));
358 !oideq(&oids[n], &cur_g->oid) ||
359 !hasheq(oids[n].hash, g->chunk_base_graphs + g->hash_len * n)) {
360 warning(_("commit-graph chain does not match"));
364 cur_g = cur_g->base_graph;
367 g->base_graph = chain;
370 g->num_commits_in_base = chain->num_commits + chain->num_commits_in_base;
375 static struct commit_graph *load_commit_graph_chain(struct repository *r, const char *obj_dir)
377 struct commit_graph *graph_chain = NULL;
378 struct strbuf line = STRBUF_INIT;
380 struct object_id *oids;
381 int i = 0, valid = 1, count;
382 char *chain_name = get_chain_filename(obj_dir);
386 fp = fopen(chain_name, "r");
387 stat_res = stat(chain_name, &st);
392 st.st_size <= the_hash_algo->hexsz)
395 count = st.st_size / (the_hash_algo->hexsz + 1);
396 oids = xcalloc(count, sizeof(struct object_id));
400 for (i = 0; i < count; i++) {
401 struct object_directory *odb;
403 if (strbuf_getline_lf(&line, fp) == EOF)
406 if (get_oid_hex(line.buf, &oids[i])) {
407 warning(_("invalid commit-graph chain: line '%s' not a hash"),
414 for (odb = r->objects->odb; odb; odb = odb->next) {
415 char *graph_name = get_split_graph_filename(odb->path, line.buf);
416 struct commit_graph *g = load_commit_graph_one(graph_name);
421 g->obj_dir = odb->path;
423 if (add_graph_to_chain(g, graph_chain, oids, i)) {
433 warning(_("unable to find all commit-graph files"));
444 struct commit_graph *read_commit_graph_one(struct repository *r, const char *obj_dir)
446 struct commit_graph *g = load_commit_graph_v1(r, obj_dir);
449 g = load_commit_graph_chain(r, obj_dir);
454 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
457 if (r->objects->commit_graph)
460 r->objects->commit_graph = read_commit_graph_one(r, obj_dir);
464 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
466 * On the first invocation, this function attemps to load the commit
467 * graph if the_repository is configured to have one.
469 static int prepare_commit_graph(struct repository *r)
471 struct object_directory *odb;
473 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
474 die("dying as requested by the '%s' variable on commit-graph load!",
475 GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD);
477 if (r->objects->commit_graph_attempted)
478 return !!r->objects->commit_graph;
479 r->objects->commit_graph_attempted = 1;
481 prepare_repo_settings(r);
483 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
484 r->settings.core_commit_graph != 1)
486 * This repository is not configured to use commit graphs, so
487 * do not load one. (But report commit_graph_attempted anyway
488 * so that commit graph loading is not attempted again for this
493 if (!commit_graph_compatible(r))
497 for (odb = r->objects->odb;
498 !r->objects->commit_graph && odb;
500 prepare_commit_graph_one(r, odb->path);
501 return !!r->objects->commit_graph;
504 int generation_numbers_enabled(struct repository *r)
506 uint32_t first_generation;
507 struct commit_graph *g;
508 if (!prepare_commit_graph(r))
511 g = r->objects->commit_graph;
516 first_generation = get_be32(g->chunk_commit_data +
517 g->hash_len + 8) >> 2;
519 return !!first_generation;
522 static void close_commit_graph_one(struct commit_graph *g)
527 close_commit_graph_one(g->base_graph);
528 free_commit_graph(g);
531 void close_commit_graph(struct raw_object_store *o)
533 close_commit_graph_one(o->commit_graph);
534 o->commit_graph = NULL;
537 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
539 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
540 g->chunk_oid_lookup, g->hash_len, pos);
543 static void load_oid_from_graph(struct commit_graph *g,
545 struct object_id *oid)
549 while (g && pos < g->num_commits_in_base)
553 BUG("NULL commit-graph");
555 if (pos >= g->num_commits + g->num_commits_in_base)
556 die(_("invalid commit position. commit-graph is likely corrupt"));
558 lex_index = pos - g->num_commits_in_base;
560 hashcpy(oid->hash, g->chunk_oid_lookup + g->hash_len * lex_index);
563 static struct commit_list **insert_parent_or_die(struct repository *r,
564 struct commit_graph *g,
566 struct commit_list **pptr)
569 struct object_id oid;
571 if (pos >= g->num_commits + g->num_commits_in_base)
572 die("invalid parent position %"PRIu32, pos);
574 load_oid_from_graph(g, pos, &oid);
575 c = lookup_commit(r, &oid);
577 die(_("could not find commit %s"), oid_to_hex(&oid));
579 return &commit_list_insert(c, pptr)->next;
582 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
584 const unsigned char *commit_data;
587 while (pos < g->num_commits_in_base)
590 lex_index = pos - g->num_commits_in_base;
591 commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * lex_index;
592 item->graph_pos = pos;
593 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
596 static inline void set_commit_tree(struct commit *c, struct tree *t)
601 static int fill_commit_in_graph(struct repository *r,
603 struct commit_graph *g, uint32_t pos)
606 uint32_t *parent_data_ptr;
607 uint64_t date_low, date_high;
608 struct commit_list **pptr;
609 const unsigned char *commit_data;
612 while (pos < g->num_commits_in_base)
615 if (pos >= g->num_commits + g->num_commits_in_base)
616 die(_("invalid commit position. commit-graph is likely corrupt"));
619 * Store the "full" position, but then use the
620 * "local" position for the rest of the calculation.
622 item->graph_pos = pos;
623 lex_index = pos - g->num_commits_in_base;
625 commit_data = g->chunk_commit_data + (g->hash_len + 16) * lex_index;
627 item->object.parsed = 1;
629 set_commit_tree(item, NULL);
631 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
632 date_low = get_be32(commit_data + g->hash_len + 12);
633 item->date = (timestamp_t)((date_high << 32) | date_low);
635 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
637 pptr = &item->parents;
639 edge_value = get_be32(commit_data + g->hash_len);
640 if (edge_value == GRAPH_PARENT_NONE)
642 pptr = insert_parent_or_die(r, g, edge_value, pptr);
644 edge_value = get_be32(commit_data + g->hash_len + 4);
645 if (edge_value == GRAPH_PARENT_NONE)
647 if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
648 pptr = insert_parent_or_die(r, g, edge_value, pptr);
652 parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
653 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
655 edge_value = get_be32(parent_data_ptr);
656 pptr = insert_parent_or_die(r, g,
657 edge_value & GRAPH_EDGE_LAST_MASK,
660 } while (!(edge_value & GRAPH_LAST_EDGE));
665 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
667 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
668 *pos = item->graph_pos;
671 struct commit_graph *cur_g = g;
674 while (cur_g && !bsearch_graph(cur_g, &(item->object.oid), &lex_index))
675 cur_g = cur_g->base_graph;
678 *pos = lex_index + cur_g->num_commits_in_base;
686 static int parse_commit_in_graph_one(struct repository *r,
687 struct commit_graph *g,
692 if (item->object.parsed)
695 if (find_commit_in_graph(item, g, &pos))
696 return fill_commit_in_graph(r, item, g, pos);
701 int parse_commit_in_graph(struct repository *r, struct commit *item)
703 if (!prepare_commit_graph(r))
705 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
708 void load_commit_graph_info(struct repository *r, struct commit *item)
711 if (!prepare_commit_graph(r))
713 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
714 fill_commit_graph_info(item, r->objects->commit_graph, pos);
717 static struct tree *load_tree_for_commit(struct repository *r,
718 struct commit_graph *g,
721 struct object_id oid;
722 const unsigned char *commit_data;
724 while (c->graph_pos < g->num_commits_in_base)
727 commit_data = g->chunk_commit_data +
728 GRAPH_DATA_WIDTH * (c->graph_pos - g->num_commits_in_base);
730 hashcpy(oid.hash, commit_data);
731 set_commit_tree(c, lookup_tree(r, &oid));
733 return c->maybe_tree;
736 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
737 struct commit_graph *g,
738 const struct commit *c)
741 return c->maybe_tree;
742 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
743 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
745 return load_tree_for_commit(r, g, (struct commit *)c);
748 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
750 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
753 struct packed_commit_list {
754 struct commit **list;
759 struct packed_oid_list {
760 struct object_id *list;
765 struct write_commit_graph_context {
766 struct repository *r;
769 struct packed_oid_list oids;
770 struct packed_commit_list commits;
772 unsigned long approx_nr_objects;
773 struct progress *progress;
775 uint64_t progress_cnt;
777 char *base_graph_name;
778 int num_commit_graphs_before;
779 int num_commit_graphs_after;
780 char **commit_graph_filenames_before;
781 char **commit_graph_filenames_after;
782 char **commit_graph_hash_after;
783 uint32_t new_num_commits_in_base;
784 struct commit_graph *new_base_graph;
790 const struct split_commit_graph_opts *split_opts;
793 static void write_graph_chunk_fanout(struct hashfile *f,
794 struct write_commit_graph_context *ctx)
797 struct commit **list = ctx->commits.list;
800 * Write the first-level table (the list is sorted,
801 * but we use a 256-entry lookup to be able to avoid
802 * having to do eight extra binary search iterations).
804 for (i = 0; i < 256; i++) {
805 while (count < ctx->commits.nr) {
806 if ((*list)->object.oid.hash[0] != i)
808 display_progress(ctx->progress, ++ctx->progress_cnt);
813 hashwrite_be32(f, count);
817 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
818 struct write_commit_graph_context *ctx)
820 struct commit **list = ctx->commits.list;
822 for (count = 0; count < ctx->commits.nr; count++, list++) {
823 display_progress(ctx->progress, ++ctx->progress_cnt);
824 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
828 static const unsigned char *commit_to_sha1(size_t index, void *table)
830 struct commit **commits = table;
831 return commits[index]->object.oid.hash;
834 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
835 struct write_commit_graph_context *ctx)
837 struct commit **list = ctx->commits.list;
838 struct commit **last = ctx->commits.list + ctx->commits.nr;
839 uint32_t num_extra_edges = 0;
841 while (list < last) {
842 struct commit_list *parent;
844 uint32_t packedDate[2];
845 display_progress(ctx->progress, ++ctx->progress_cnt);
847 parse_commit_no_graph(*list);
848 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
850 parent = (*list)->parents;
853 edge_value = GRAPH_PARENT_NONE;
855 edge_value = sha1_pos(parent->item->object.oid.hash,
861 edge_value += ctx->new_num_commits_in_base;
864 if (find_commit_in_graph(parent->item,
871 BUG("missing parent %s for commit %s",
872 oid_to_hex(&parent->item->object.oid),
873 oid_to_hex(&(*list)->object.oid));
876 hashwrite_be32(f, edge_value);
879 parent = parent->next;
882 edge_value = GRAPH_PARENT_NONE;
883 else if (parent->next)
884 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
886 edge_value = sha1_pos(parent->item->object.oid.hash,
892 edge_value += ctx->new_num_commits_in_base;
895 if (find_commit_in_graph(parent->item,
902 BUG("missing parent %s for commit %s",
903 oid_to_hex(&parent->item->object.oid),
904 oid_to_hex(&(*list)->object.oid));
907 hashwrite_be32(f, edge_value);
909 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
912 parent = parent->next;
916 if (sizeof((*list)->date) > 4)
917 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
921 packedDate[0] |= htonl((*list)->generation << 2);
923 packedDate[1] = htonl((*list)->date);
924 hashwrite(f, packedDate, 8);
930 static void write_graph_chunk_extra_edges(struct hashfile *f,
931 struct write_commit_graph_context *ctx)
933 struct commit **list = ctx->commits.list;
934 struct commit **last = ctx->commits.list + ctx->commits.nr;
935 struct commit_list *parent;
937 while (list < last) {
940 display_progress(ctx->progress, ++ctx->progress_cnt);
942 for (parent = (*list)->parents; num_parents < 3 && parent;
943 parent = parent->next)
946 if (num_parents <= 2) {
951 /* Since num_parents > 2, this initializer is safe. */
952 for (parent = (*list)->parents->next; parent; parent = parent->next) {
953 int edge_value = sha1_pos(parent->item->object.oid.hash,
959 edge_value += ctx->new_num_commits_in_base;
962 if (find_commit_in_graph(parent->item,
969 BUG("missing parent %s for commit %s",
970 oid_to_hex(&parent->item->object.oid),
971 oid_to_hex(&(*list)->object.oid));
972 else if (!parent->next)
973 edge_value |= GRAPH_LAST_EDGE;
975 hashwrite_be32(f, edge_value);
982 static int oid_compare(const void *_a, const void *_b)
984 const struct object_id *a = (const struct object_id *)_a;
985 const struct object_id *b = (const struct object_id *)_b;
989 static int add_packed_commits(const struct object_id *oid,
990 struct packed_git *pack,
994 struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
995 enum object_type type;
996 off_t offset = nth_packed_object_offset(pack, pos);
997 struct object_info oi = OBJECT_INFO_INIT;
1000 display_progress(ctx->progress, ++ctx->progress_done);
1003 if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
1004 die(_("unable to get type of object %s"), oid_to_hex(oid));
1006 if (type != OBJ_COMMIT)
1009 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1010 oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
1016 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
1018 struct commit_list *parent;
1019 for (parent = commit->parents; parent; parent = parent->next) {
1020 if (!(parent->item->object.flags & REACHABLE)) {
1021 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1022 oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
1024 parent->item->object.flags |= REACHABLE;
1029 static void close_reachable(struct write_commit_graph_context *ctx)
1032 struct commit *commit;
1034 if (ctx->report_progress)
1035 ctx->progress = start_delayed_progress(
1036 _("Loading known commits in commit graph"),
1038 for (i = 0; i < ctx->oids.nr; i++) {
1039 display_progress(ctx->progress, i + 1);
1040 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1042 commit->object.flags |= REACHABLE;
1044 stop_progress(&ctx->progress);
1047 * As this loop runs, ctx->oids.nr may grow, but not more
1048 * than the number of missing commits in the reachable
1051 if (ctx->report_progress)
1052 ctx->progress = start_delayed_progress(
1053 _("Expanding reachable commits in commit graph"),
1055 for (i = 0; i < ctx->oids.nr; i++) {
1056 display_progress(ctx->progress, i + 1);
1057 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1062 if (!parse_commit(commit) &&
1063 commit->graph_pos == COMMIT_NOT_FROM_GRAPH)
1064 add_missing_parents(ctx, commit);
1065 } else if (!parse_commit_no_graph(commit))
1066 add_missing_parents(ctx, commit);
1068 stop_progress(&ctx->progress);
1070 if (ctx->report_progress)
1071 ctx->progress = start_delayed_progress(
1072 _("Clearing commit marks in commit graph"),
1074 for (i = 0; i < ctx->oids.nr; i++) {
1075 display_progress(ctx->progress, i + 1);
1076 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1079 commit->object.flags &= ~REACHABLE;
1081 stop_progress(&ctx->progress);
1084 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1087 struct commit_list *list = NULL;
1089 if (ctx->report_progress)
1090 ctx->progress = start_progress(
1091 _("Computing commit graph generation numbers"),
1093 for (i = 0; i < ctx->commits.nr; i++) {
1094 display_progress(ctx->progress, i + 1);
1095 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
1096 ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
1099 commit_list_insert(ctx->commits.list[i], &list);
1101 struct commit *current = list->item;
1102 struct commit_list *parent;
1103 int all_parents_computed = 1;
1104 uint32_t max_generation = 0;
1106 for (parent = current->parents; parent; parent = parent->next) {
1107 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
1108 parent->item->generation == GENERATION_NUMBER_ZERO) {
1109 all_parents_computed = 0;
1110 commit_list_insert(parent->item, &list);
1112 } else if (parent->item->generation > max_generation) {
1113 max_generation = parent->item->generation;
1117 if (all_parents_computed) {
1118 current->generation = max_generation + 1;
1121 if (current->generation > GENERATION_NUMBER_MAX)
1122 current->generation = GENERATION_NUMBER_MAX;
1126 stop_progress(&ctx->progress);
1129 static int add_ref_to_list(const char *refname,
1130 const struct object_id *oid,
1131 int flags, void *cb_data)
1133 struct string_list *list = (struct string_list *)cb_data;
1135 string_list_append(list, oid_to_hex(oid));
1139 int write_commit_graph_reachable(const char *obj_dir, unsigned int flags,
1140 const struct split_commit_graph_opts *split_opts)
1142 struct string_list list = STRING_LIST_INIT_DUP;
1145 for_each_ref(add_ref_to_list, &list);
1146 result = write_commit_graph(obj_dir, NULL, &list,
1149 string_list_clear(&list, 0);
1153 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
1154 struct string_list *pack_indexes)
1157 struct strbuf progress_title = STRBUF_INIT;
1158 struct strbuf packname = STRBUF_INIT;
1161 strbuf_addf(&packname, "%s/pack/", ctx->obj_dir);
1162 dirlen = packname.len;
1163 if (ctx->report_progress) {
1164 strbuf_addf(&progress_title,
1165 Q_("Finding commits for commit graph in %d pack",
1166 "Finding commits for commit graph in %d packs",
1169 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1170 ctx->progress_done = 0;
1172 for (i = 0; i < pack_indexes->nr; i++) {
1173 struct packed_git *p;
1174 strbuf_setlen(&packname, dirlen);
1175 strbuf_addstr(&packname, pack_indexes->items[i].string);
1176 p = add_packed_git(packname.buf, packname.len, 1);
1178 error(_("error adding pack %s"), packname.buf);
1181 if (open_pack_index(p)) {
1182 error(_("error opening index for %s"), packname.buf);
1185 for_each_object_in_pack(p, add_packed_commits, ctx,
1186 FOR_EACH_OBJECT_PACK_ORDER);
1191 stop_progress(&ctx->progress);
1192 strbuf_reset(&progress_title);
1193 strbuf_release(&packname);
1198 static void fill_oids_from_commit_hex(struct write_commit_graph_context *ctx,
1199 struct string_list *commit_hex)
1202 struct strbuf progress_title = STRBUF_INIT;
1204 if (ctx->report_progress) {
1205 strbuf_addf(&progress_title,
1206 Q_("Finding commits for commit graph from %d ref",
1207 "Finding commits for commit graph from %d refs",
1210 ctx->progress = start_delayed_progress(
1214 for (i = 0; i < commit_hex->nr; i++) {
1216 struct object_id oid;
1217 struct commit *result;
1219 display_progress(ctx->progress, i + 1);
1220 if (commit_hex->items[i].string &&
1221 parse_oid_hex(commit_hex->items[i].string, &oid, &end))
1224 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1227 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1228 oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
1232 stop_progress(&ctx->progress);
1233 strbuf_release(&progress_title);
1236 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1238 if (ctx->report_progress)
1239 ctx->progress = start_delayed_progress(
1240 _("Finding commits for commit graph among packed objects"),
1241 ctx->approx_nr_objects);
1242 for_each_packed_object(add_packed_commits, ctx,
1243 FOR_EACH_OBJECT_PACK_ORDER);
1244 if (ctx->progress_done < ctx->approx_nr_objects)
1245 display_progress(ctx->progress, ctx->approx_nr_objects);
1246 stop_progress(&ctx->progress);
1249 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
1251 uint32_t i, count_distinct = 1;
1253 if (ctx->report_progress)
1254 ctx->progress = start_delayed_progress(
1255 _("Counting distinct commits in commit graph"),
1257 display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1258 QSORT(ctx->oids.list, ctx->oids.nr, oid_compare);
1260 for (i = 1; i < ctx->oids.nr; i++) {
1261 display_progress(ctx->progress, i + 1);
1262 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i])) {
1264 struct commit *c = lookup_commit(ctx->r, &ctx->oids.list[i]);
1266 if (!c || c->graph_pos != COMMIT_NOT_FROM_GRAPH)
1273 stop_progress(&ctx->progress);
1275 return count_distinct;
1278 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1281 struct commit_list *parent;
1283 ctx->num_extra_edges = 0;
1284 if (ctx->report_progress)
1285 ctx->progress = start_delayed_progress(
1286 _("Finding extra edges in commit graph"),
1288 for (i = 0; i < ctx->oids.nr; i++) {
1289 int num_parents = 0;
1290 display_progress(ctx->progress, i + 1);
1291 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1294 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + 1, ctx->commits.alloc);
1295 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1298 ctx->commits.list[ctx->commits.nr]->graph_pos != COMMIT_NOT_FROM_GRAPH)
1301 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1303 for (parent = ctx->commits.list[ctx->commits.nr]->parents;
1304 parent; parent = parent->next)
1307 if (num_parents > 2)
1308 ctx->num_extra_edges += num_parents - 1;
1312 stop_progress(&ctx->progress);
1315 static int write_graph_chunk_base_1(struct hashfile *f,
1316 struct commit_graph *g)
1323 num = write_graph_chunk_base_1(f, g->base_graph);
1324 hashwrite(f, g->oid.hash, the_hash_algo->rawsz);
1328 static int write_graph_chunk_base(struct hashfile *f,
1329 struct write_commit_graph_context *ctx)
1331 int num = write_graph_chunk_base_1(f, ctx->new_base_graph);
1333 if (num != ctx->num_commit_graphs_after - 1) {
1334 error(_("failed to write correct number of base graph ids"));
1341 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1346 struct lock_file lk = LOCK_INIT;
1347 uint32_t chunk_ids[6];
1348 uint64_t chunk_offsets[6];
1349 const unsigned hashsz = the_hash_algo->rawsz;
1350 struct strbuf progress_title = STRBUF_INIT;
1352 struct object_id file_hash;
1355 struct strbuf tmp_file = STRBUF_INIT;
1357 strbuf_addf(&tmp_file,
1358 "%s/info/commit-graphs/tmp_graph_XXXXXX",
1360 ctx->graph_name = strbuf_detach(&tmp_file, NULL);
1362 ctx->graph_name = get_commit_graph_filename(ctx->obj_dir);
1365 if (safe_create_leading_directories(ctx->graph_name)) {
1366 UNLEAK(ctx->graph_name);
1367 error(_("unable to create leading directories of %s"),
1373 char *lock_name = get_chain_filename(ctx->obj_dir);
1375 hold_lock_file_for_update(&lk, lock_name, LOCK_DIE_ON_ERROR);
1377 fd = git_mkstemp_mode(ctx->graph_name, 0444);
1379 error(_("unable to create '%s'"), ctx->graph_name);
1383 f = hashfd(fd, ctx->graph_name);
1385 hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
1386 fd = lk.tempfile->fd;
1387 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1390 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1391 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1392 chunk_ids[2] = GRAPH_CHUNKID_DATA;
1393 if (ctx->num_extra_edges) {
1394 chunk_ids[num_chunks] = GRAPH_CHUNKID_EXTRAEDGES;
1397 if (ctx->num_commit_graphs_after > 1) {
1398 chunk_ids[num_chunks] = GRAPH_CHUNKID_BASE;
1402 chunk_ids[num_chunks] = 0;
1404 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1405 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1406 chunk_offsets[2] = chunk_offsets[1] + hashsz * ctx->commits.nr;
1407 chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * ctx->commits.nr;
1410 if (ctx->num_extra_edges) {
1411 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1412 4 * ctx->num_extra_edges;
1415 if (ctx->num_commit_graphs_after > 1) {
1416 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1417 hashsz * (ctx->num_commit_graphs_after - 1);
1421 hashwrite_be32(f, GRAPH_SIGNATURE);
1423 hashwrite_u8(f, GRAPH_VERSION);
1424 hashwrite_u8(f, oid_version());
1425 hashwrite_u8(f, num_chunks);
1426 hashwrite_u8(f, ctx->num_commit_graphs_after - 1);
1428 for (i = 0; i <= num_chunks; i++) {
1429 uint32_t chunk_write[3];
1431 chunk_write[0] = htonl(chunk_ids[i]);
1432 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1433 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1434 hashwrite(f, chunk_write, 12);
1437 if (ctx->report_progress) {
1438 strbuf_addf(&progress_title,
1439 Q_("Writing out commit graph in %d pass",
1440 "Writing out commit graph in %d passes",
1443 ctx->progress = start_delayed_progress(
1445 num_chunks * ctx->commits.nr);
1447 write_graph_chunk_fanout(f, ctx);
1448 write_graph_chunk_oids(f, hashsz, ctx);
1449 write_graph_chunk_data(f, hashsz, ctx);
1450 if (ctx->num_extra_edges)
1451 write_graph_chunk_extra_edges(f, ctx);
1452 if (ctx->num_commit_graphs_after > 1 &&
1453 write_graph_chunk_base(f, ctx)) {
1456 stop_progress(&ctx->progress);
1457 strbuf_release(&progress_title);
1459 if (ctx->split && ctx->base_graph_name && ctx->num_commit_graphs_after > 1) {
1460 char *new_base_hash = xstrdup(oid_to_hex(&ctx->new_base_graph->oid));
1461 char *new_base_name = get_split_graph_filename(ctx->new_base_graph->obj_dir, new_base_hash);
1463 free(ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2]);
1464 free(ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2]);
1465 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2] = new_base_name;
1466 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2] = new_base_hash;
1469 close_commit_graph(ctx->r->objects);
1470 finalize_hashfile(f, file_hash.hash, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1473 FILE *chainf = fdopen_lock_file(&lk, "w");
1474 char *final_graph_name;
1480 error(_("unable to open commit-graph chain file"));
1484 if (ctx->base_graph_name) {
1485 const char *dest = ctx->commit_graph_filenames_after[
1486 ctx->num_commit_graphs_after - 2];
1488 if (strcmp(ctx->base_graph_name, dest)) {
1489 result = rename(ctx->base_graph_name, dest);
1492 error(_("failed to rename base commit-graph file"));
1497 char *graph_name = get_commit_graph_filename(ctx->obj_dir);
1501 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1] = xstrdup(oid_to_hex(&file_hash));
1502 final_graph_name = get_split_graph_filename(ctx->obj_dir,
1503 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1]);
1504 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 1] = final_graph_name;
1506 result = rename(ctx->graph_name, final_graph_name);
1508 for (i = 0; i < ctx->num_commit_graphs_after; i++)
1509 fprintf(lk.tempfile->fp, "%s\n", ctx->commit_graph_hash_after[i]);
1512 error(_("failed to rename temporary commit-graph file"));
1517 commit_lock_file(&lk);
1522 static void split_graph_merge_strategy(struct write_commit_graph_context *ctx)
1524 struct commit_graph *g = ctx->r->objects->commit_graph;
1525 uint32_t num_commits = ctx->commits.nr;
1528 int max_commits = 0;
1531 if (ctx->split_opts) {
1532 max_commits = ctx->split_opts->max_commits;
1533 size_mult = ctx->split_opts->size_multiple;
1536 g = ctx->r->objects->commit_graph;
1537 ctx->num_commit_graphs_after = ctx->num_commit_graphs_before + 1;
1539 while (g && (g->num_commits <= size_mult * num_commits ||
1540 (max_commits && num_commits > max_commits))) {
1541 if (strcmp(g->obj_dir, ctx->obj_dir))
1544 num_commits += g->num_commits;
1547 ctx->num_commit_graphs_after--;
1550 ctx->new_base_graph = g;
1552 if (ctx->num_commit_graphs_after == 2) {
1553 char *old_graph_name = get_commit_graph_filename(g->obj_dir);
1555 if (!strcmp(g->filename, old_graph_name) &&
1556 strcmp(g->obj_dir, ctx->obj_dir)) {
1557 ctx->num_commit_graphs_after = 1;
1558 ctx->new_base_graph = NULL;
1561 free(old_graph_name);
1564 ALLOC_ARRAY(ctx->commit_graph_filenames_after, ctx->num_commit_graphs_after);
1565 ALLOC_ARRAY(ctx->commit_graph_hash_after, ctx->num_commit_graphs_after);
1567 for (i = 0; i < ctx->num_commit_graphs_after &&
1568 i < ctx->num_commit_graphs_before; i++)
1569 ctx->commit_graph_filenames_after[i] = xstrdup(ctx->commit_graph_filenames_before[i]);
1571 i = ctx->num_commit_graphs_before - 1;
1572 g = ctx->r->objects->commit_graph;
1575 if (i < ctx->num_commit_graphs_after)
1576 ctx->commit_graph_hash_after[i] = xstrdup(oid_to_hex(&g->oid));
1583 static void merge_commit_graph(struct write_commit_graph_context *ctx,
1584 struct commit_graph *g)
1587 uint32_t offset = g->num_commits_in_base;
1589 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + g->num_commits, ctx->commits.alloc);
1591 for (i = 0; i < g->num_commits; i++) {
1592 struct object_id oid;
1593 struct commit *result;
1595 display_progress(ctx->progress, i + 1);
1597 load_oid_from_graph(g, i + offset, &oid);
1599 /* only add commits if they still exist in the repo */
1600 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1603 ctx->commits.list[ctx->commits.nr] = result;
1609 static int commit_compare(const void *_a, const void *_b)
1611 const struct commit *a = *(const struct commit **)_a;
1612 const struct commit *b = *(const struct commit **)_b;
1613 return oidcmp(&a->object.oid, &b->object.oid);
1616 static void sort_and_scan_merged_commits(struct write_commit_graph_context *ctx)
1618 uint32_t i, num_parents;
1619 struct commit_list *parent;
1621 if (ctx->report_progress)
1622 ctx->progress = start_delayed_progress(
1623 _("Scanning merged commits"),
1626 QSORT(ctx->commits.list, ctx->commits.nr, commit_compare);
1628 ctx->num_extra_edges = 0;
1629 for (i = 0; i < ctx->commits.nr; i++) {
1630 display_progress(ctx->progress, i);
1632 if (i && oideq(&ctx->commits.list[i - 1]->object.oid,
1633 &ctx->commits.list[i]->object.oid)) {
1634 die(_("unexpected duplicate commit id %s"),
1635 oid_to_hex(&ctx->commits.list[i]->object.oid));
1638 for (parent = ctx->commits.list[i]->parents; parent; parent = parent->next)
1641 if (num_parents > 2)
1642 ctx->num_extra_edges += num_parents - 2;
1646 stop_progress(&ctx->progress);
1649 static void merge_commit_graphs(struct write_commit_graph_context *ctx)
1651 struct commit_graph *g = ctx->r->objects->commit_graph;
1652 uint32_t current_graph_number = ctx->num_commit_graphs_before;
1653 struct strbuf progress_title = STRBUF_INIT;
1655 while (g && current_graph_number >= ctx->num_commit_graphs_after) {
1656 current_graph_number--;
1658 if (ctx->report_progress) {
1659 strbuf_addstr(&progress_title, _("Merging commit-graph"));
1660 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1663 merge_commit_graph(ctx, g);
1664 stop_progress(&ctx->progress);
1665 strbuf_release(&progress_title);
1671 ctx->new_base_graph = g;
1672 ctx->new_num_commits_in_base = g->num_commits + g->num_commits_in_base;
1675 if (ctx->new_base_graph)
1676 ctx->base_graph_name = xstrdup(ctx->new_base_graph->filename);
1678 sort_and_scan_merged_commits(ctx);
1681 static void mark_commit_graphs(struct write_commit_graph_context *ctx)
1684 time_t now = time(NULL);
1686 for (i = ctx->num_commit_graphs_after - 1; i < ctx->num_commit_graphs_before; i++) {
1688 struct utimbuf updated_time;
1690 stat(ctx->commit_graph_filenames_before[i], &st);
1692 updated_time.actime = st.st_atime;
1693 updated_time.modtime = now;
1694 utime(ctx->commit_graph_filenames_before[i], &updated_time);
1698 static void expire_commit_graphs(struct write_commit_graph_context *ctx)
1700 struct strbuf path = STRBUF_INIT;
1704 timestamp_t expire_time = time(NULL);
1706 if (ctx->split_opts && ctx->split_opts->expire_time)
1707 expire_time -= ctx->split_opts->expire_time;
1709 char *chain_file_name = get_chain_filename(ctx->obj_dir);
1710 unlink(chain_file_name);
1711 free(chain_file_name);
1712 ctx->num_commit_graphs_after = 0;
1715 strbuf_addstr(&path, ctx->obj_dir);
1716 strbuf_addstr(&path, "/info/commit-graphs");
1717 dir = opendir(path.buf);
1720 strbuf_release(&path);
1724 strbuf_addch(&path, '/');
1725 dirnamelen = path.len;
1726 while ((de = readdir(dir)) != NULL) {
1728 uint32_t i, found = 0;
1730 strbuf_setlen(&path, dirnamelen);
1731 strbuf_addstr(&path, de->d_name);
1733 stat(path.buf, &st);
1735 if (st.st_mtime > expire_time)
1737 if (path.len < 6 || strcmp(path.buf + path.len - 6, ".graph"))
1740 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
1741 if (!strcmp(ctx->commit_graph_filenames_after[i],
1753 int write_commit_graph(const char *obj_dir,
1754 struct string_list *pack_indexes,
1755 struct string_list *commit_hex,
1757 const struct split_commit_graph_opts *split_opts)
1759 struct write_commit_graph_context *ctx;
1760 uint32_t i, count_distinct = 0;
1764 if (!commit_graph_compatible(the_repository))
1767 ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
1768 ctx->r = the_repository;
1770 /* normalize object dir with no trailing slash */
1771 ctx->obj_dir = xmallocz(strlen(obj_dir) + 1);
1772 normalize_path_copy(ctx->obj_dir, obj_dir);
1773 len = strlen(ctx->obj_dir);
1774 if (len && ctx->obj_dir[len - 1] == '/')
1775 ctx->obj_dir[len - 1] = 0;
1777 ctx->append = flags & COMMIT_GRAPH_APPEND ? 1 : 0;
1778 ctx->report_progress = flags & COMMIT_GRAPH_PROGRESS ? 1 : 0;
1779 ctx->split = flags & COMMIT_GRAPH_SPLIT ? 1 : 0;
1780 ctx->split_opts = split_opts;
1783 struct commit_graph *g;
1784 prepare_commit_graph(ctx->r);
1786 g = ctx->r->objects->commit_graph;
1789 ctx->num_commit_graphs_before++;
1793 if (ctx->num_commit_graphs_before) {
1794 ALLOC_ARRAY(ctx->commit_graph_filenames_before, ctx->num_commit_graphs_before);
1795 i = ctx->num_commit_graphs_before;
1796 g = ctx->r->objects->commit_graph;
1799 ctx->commit_graph_filenames_before[--i] = xstrdup(g->filename);
1805 ctx->approx_nr_objects = approximate_object_count();
1806 ctx->oids.alloc = ctx->approx_nr_objects / 32;
1808 if (ctx->split && split_opts && ctx->oids.alloc > split_opts->max_commits)
1809 ctx->oids.alloc = split_opts->max_commits;
1812 prepare_commit_graph_one(ctx->r, ctx->obj_dir);
1813 if (ctx->r->objects->commit_graph)
1814 ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
1817 if (ctx->oids.alloc < 1024)
1818 ctx->oids.alloc = 1024;
1819 ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
1821 if (ctx->append && ctx->r->objects->commit_graph) {
1822 struct commit_graph *g = ctx->r->objects->commit_graph;
1823 for (i = 0; i < g->num_commits; i++) {
1824 const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
1825 hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
1830 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
1835 fill_oids_from_commit_hex(ctx, commit_hex);
1837 if (!pack_indexes && !commit_hex)
1838 fill_oids_from_all_packs(ctx);
1840 close_reachable(ctx);
1842 count_distinct = count_distinct_commits(ctx);
1844 if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
1845 error(_("the commit graph format cannot write %d commits"), count_distinct);
1850 ctx->commits.alloc = count_distinct;
1851 ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
1853 copy_oids_to_commits(ctx);
1855 if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
1856 error(_("too many commits to write graph"));
1861 if (!ctx->commits.nr)
1865 split_graph_merge_strategy(ctx);
1867 merge_commit_graphs(ctx);
1869 ctx->num_commit_graphs_after = 1;
1871 compute_generation_numbers(ctx);
1873 res = write_commit_graph_file(ctx);
1876 mark_commit_graphs(ctx);
1878 expire_commit_graphs(ctx);
1881 free(ctx->graph_name);
1882 free(ctx->commits.list);
1883 free(ctx->oids.list);
1886 if (ctx->commit_graph_filenames_after) {
1887 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
1888 free(ctx->commit_graph_filenames_after[i]);
1889 free(ctx->commit_graph_hash_after[i]);
1892 for (i = 0; i < ctx->num_commit_graphs_before; i++)
1893 free(ctx->commit_graph_filenames_before[i]);
1895 free(ctx->commit_graph_filenames_after);
1896 free(ctx->commit_graph_filenames_before);
1897 free(ctx->commit_graph_hash_after);
1905 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1906 static int verify_commit_graph_error;
1908 static void graph_report(const char *fmt, ...)
1912 verify_commit_graph_error = 1;
1914 vfprintf(stderr, fmt, ap);
1915 fprintf(stderr, "\n");
1919 #define GENERATION_ZERO_EXISTS 1
1920 #define GENERATION_NUMBER_EXISTS 2
1922 int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
1924 uint32_t i, cur_fanout_pos = 0;
1925 struct object_id prev_oid, cur_oid, checksum;
1926 int generation_zero = 0;
1929 struct progress *progress = NULL;
1930 int local_error = 0;
1933 graph_report("no commit-graph file loaded");
1937 verify_commit_graph_error = verify_commit_graph_lite(g);
1938 if (verify_commit_graph_error)
1939 return verify_commit_graph_error;
1941 devnull = open("/dev/null", O_WRONLY);
1942 f = hashfd(devnull, NULL);
1943 hashwrite(f, g->data, g->data_len - g->hash_len);
1944 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1945 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1946 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1947 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1950 for (i = 0; i < g->num_commits; i++) {
1951 struct commit *graph_commit;
1953 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1955 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1956 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
1957 oid_to_hex(&prev_oid),
1958 oid_to_hex(&cur_oid));
1960 oidcpy(&prev_oid, &cur_oid);
1962 while (cur_oid.hash[0] > cur_fanout_pos) {
1963 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1965 if (i != fanout_value)
1966 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1967 cur_fanout_pos, fanout_value, i);
1971 graph_commit = lookup_commit(r, &cur_oid);
1972 if (!parse_commit_in_graph_one(r, g, graph_commit))
1973 graph_report(_("failed to parse commit %s from commit-graph"),
1974 oid_to_hex(&cur_oid));
1977 while (cur_fanout_pos < 256) {
1978 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1980 if (g->num_commits != fanout_value)
1981 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1982 cur_fanout_pos, fanout_value, i);
1987 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1988 return verify_commit_graph_error;
1990 progress = start_progress(_("Verifying commits in commit graph"),
1992 for (i = 0; i < g->num_commits; i++) {
1993 struct commit *graph_commit, *odb_commit;
1994 struct commit_list *graph_parents, *odb_parents;
1995 uint32_t max_generation = 0;
1997 display_progress(progress, i + 1);
1998 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2000 graph_commit = lookup_commit(r, &cur_oid);
2001 odb_commit = (struct commit *)create_object(r, &cur_oid, alloc_commit_node(r));
2002 if (parse_commit_internal(odb_commit, 0, 0)) {
2003 graph_report(_("failed to parse commit %s from object database for commit-graph"),
2004 oid_to_hex(&cur_oid));
2008 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
2009 get_commit_tree_oid(odb_commit)))
2010 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2011 oid_to_hex(&cur_oid),
2012 oid_to_hex(get_commit_tree_oid(graph_commit)),
2013 oid_to_hex(get_commit_tree_oid(odb_commit)));
2015 graph_parents = graph_commit->parents;
2016 odb_parents = odb_commit->parents;
2018 while (graph_parents) {
2019 if (odb_parents == NULL) {
2020 graph_report(_("commit-graph parent list for commit %s is too long"),
2021 oid_to_hex(&cur_oid));
2025 /* parse parent in case it is in a base graph */
2026 parse_commit_in_graph_one(r, g, graph_parents->item);
2028 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
2029 graph_report(_("commit-graph parent for %s is %s != %s"),
2030 oid_to_hex(&cur_oid),
2031 oid_to_hex(&graph_parents->item->object.oid),
2032 oid_to_hex(&odb_parents->item->object.oid));
2034 if (graph_parents->item->generation > max_generation)
2035 max_generation = graph_parents->item->generation;
2037 graph_parents = graph_parents->next;
2038 odb_parents = odb_parents->next;
2041 if (odb_parents != NULL)
2042 graph_report(_("commit-graph parent list for commit %s terminates early"),
2043 oid_to_hex(&cur_oid));
2045 if (!graph_commit->generation) {
2046 if (generation_zero == GENERATION_NUMBER_EXISTS)
2047 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2048 oid_to_hex(&cur_oid));
2049 generation_zero = GENERATION_ZERO_EXISTS;
2050 } else if (generation_zero == GENERATION_ZERO_EXISTS)
2051 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2052 oid_to_hex(&cur_oid));
2054 if (generation_zero == GENERATION_ZERO_EXISTS)
2058 * If one of our parents has generation GENERATION_NUMBER_MAX, then
2059 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
2060 * extra logic in the following condition.
2062 if (max_generation == GENERATION_NUMBER_MAX)
2065 if (graph_commit->generation != max_generation + 1)
2066 graph_report(_("commit-graph generation for commit %s is %u != %u"),
2067 oid_to_hex(&cur_oid),
2068 graph_commit->generation,
2069 max_generation + 1);
2071 if (graph_commit->date != odb_commit->date)
2072 graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
2073 oid_to_hex(&cur_oid),
2077 stop_progress(&progress);
2079 local_error = verify_commit_graph_error;
2081 if (!(flags & COMMIT_GRAPH_VERIFY_SHALLOW) && g->base_graph)
2082 local_error |= verify_commit_graph(r, g->base_graph, flags);
2087 void free_commit_graph(struct commit_graph *g)
2091 if (g->graph_fd >= 0) {
2092 munmap((void *)g->data, g->data_len);