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 #include "commit-slab.h"
22 void git_test_write_commit_graph_or_die(void)
25 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0))
28 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS, 0))
29 flags = COMMIT_GRAPH_WRITE_BLOOM_FILTERS;
31 if (write_commit_graph_reachable(the_repository->objects->odb,
33 die("failed to write commit-graph under GIT_TEST_COMMIT_GRAPH");
36 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
37 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
38 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
39 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
40 #define GRAPH_CHUNKID_EXTRAEDGES 0x45444745 /* "EDGE" */
41 #define GRAPH_CHUNKID_BLOOMINDEXES 0x42494458 /* "BIDX" */
42 #define GRAPH_CHUNKID_BLOOMDATA 0x42444154 /* "BDAT" */
43 #define GRAPH_CHUNKID_BASE 0x42415345 /* "BASE" */
44 #define MAX_NUM_CHUNKS 7
46 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
48 #define GRAPH_VERSION_1 0x1
49 #define GRAPH_VERSION GRAPH_VERSION_1
51 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
52 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
53 #define GRAPH_PARENT_NONE 0x70000000
55 #define GRAPH_LAST_EDGE 0x80000000
57 #define GRAPH_HEADER_SIZE 8
58 #define GRAPH_FANOUT_SIZE (4 * 256)
59 #define GRAPH_CHUNKLOOKUP_WIDTH 12
60 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
61 + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
63 /* Remember to update object flag allocation in object.h */
64 #define REACHABLE (1u<<15)
66 /* Keep track of the order in which commits are added to our list. */
67 define_commit_slab(commit_pos, int);
68 static struct commit_pos commit_pos = COMMIT_SLAB_INIT(1, commit_pos);
70 static void set_commit_pos(struct repository *r, const struct object_id *oid)
72 static int32_t max_pos;
73 struct commit *commit = lookup_commit(r, oid);
76 return; /* should never happen, but be lenient */
78 *commit_pos_at(&commit_pos, commit) = max_pos++;
81 static int commit_pos_cmp(const void *va, const void *vb)
83 const struct commit *a = *(const struct commit **)va;
84 const struct commit *b = *(const struct commit **)vb;
85 return commit_pos_at(&commit_pos, a) -
86 commit_pos_at(&commit_pos, b);
89 static int commit_gen_cmp(const void *va, const void *vb)
91 const struct commit *a = *(const struct commit **)va;
92 const struct commit *b = *(const struct commit **)vb;
94 /* lower generation commits first */
95 if (a->generation < b->generation)
97 else if (a->generation > b->generation)
100 /* use date as a heuristic when generations are equal */
101 if (a->date < b->date)
103 else if (a->date > b->date)
108 char *get_commit_graph_filename(struct object_directory *obj_dir)
110 return xstrfmt("%s/info/commit-graph", obj_dir->path);
113 static char *get_split_graph_filename(struct object_directory *odb,
116 return xstrfmt("%s/info/commit-graphs/graph-%s.graph", odb->path,
120 static char *get_chain_filename(struct object_directory *odb)
122 return xstrfmt("%s/info/commit-graphs/commit-graph-chain", odb->path);
125 static uint8_t oid_version(void)
130 static struct commit_graph *alloc_commit_graph(void)
132 struct commit_graph *g = xcalloc(1, sizeof(*g));
137 extern int read_replace_refs;
139 static int commit_graph_compatible(struct repository *r)
144 if (read_replace_refs) {
145 prepare_replace_object(r);
146 if (hashmap_get_size(&r->objects->replace_map->map))
150 prepare_commit_graft(r);
151 if (r->parsed_objects && r->parsed_objects->grafts_nr)
153 if (is_repository_shallow(r))
159 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
161 *fd = git_open(graph_file);
164 if (fstat(*fd, st)) {
171 struct commit_graph *load_commit_graph_one_fd_st(int fd, struct stat *st,
172 struct object_directory *odb)
176 struct commit_graph *ret;
178 graph_size = xsize_t(st->st_size);
180 if (graph_size < GRAPH_MIN_SIZE) {
182 error(_("commit-graph file is too small"));
185 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
187 ret = parse_commit_graph(graph_map, graph_size);
192 munmap(graph_map, graph_size);
197 static int verify_commit_graph_lite(struct commit_graph *g)
200 * Basic validation shared between parse_commit_graph()
201 * which'll be called every time the graph is used, and the
202 * much more expensive verify_commit_graph() used by
203 * "commit-graph verify".
205 * There should only be very basic checks here to ensure that
206 * we don't e.g. segfault in fill_commit_in_graph(), but
207 * because this is a very hot codepath nothing that e.g. loops
208 * over g->num_commits, or runs a checksum on the commit-graph
211 if (!g->chunk_oid_fanout) {
212 error("commit-graph is missing the OID Fanout chunk");
215 if (!g->chunk_oid_lookup) {
216 error("commit-graph is missing the OID Lookup chunk");
219 if (!g->chunk_commit_data) {
220 error("commit-graph is missing the Commit Data chunk");
227 struct commit_graph *parse_commit_graph(void *graph_map, size_t graph_size)
229 const unsigned char *data, *chunk_lookup;
231 struct commit_graph *graph;
232 uint64_t last_chunk_offset;
233 uint32_t last_chunk_id;
234 uint32_t graph_signature;
235 unsigned char graph_version, hash_version;
240 if (graph_size < GRAPH_MIN_SIZE)
243 data = (const unsigned char *)graph_map;
245 graph_signature = get_be32(data);
246 if (graph_signature != GRAPH_SIGNATURE) {
247 error(_("commit-graph signature %X does not match signature %X"),
248 graph_signature, GRAPH_SIGNATURE);
252 graph_version = *(unsigned char*)(data + 4);
253 if (graph_version != GRAPH_VERSION) {
254 error(_("commit-graph version %X does not match version %X"),
255 graph_version, GRAPH_VERSION);
259 hash_version = *(unsigned char*)(data + 5);
260 if (hash_version != oid_version()) {
261 error(_("commit-graph hash version %X does not match version %X"),
262 hash_version, oid_version());
266 graph = alloc_commit_graph();
268 graph->hash_len = the_hash_algo->rawsz;
269 graph->num_chunks = *(unsigned char*)(data + 6);
270 graph->data = graph_map;
271 graph->data_len = graph_size;
274 last_chunk_offset = 8;
275 chunk_lookup = data + 8;
276 for (i = 0; i < graph->num_chunks; i++) {
278 uint64_t chunk_offset;
279 int chunk_repeated = 0;
281 if (data + graph_size - chunk_lookup <
282 GRAPH_CHUNKLOOKUP_WIDTH) {
283 error(_("commit-graph chunk lookup table entry missing; file may be incomplete"));
288 chunk_id = get_be32(chunk_lookup + 0);
289 chunk_offset = get_be64(chunk_lookup + 4);
291 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
293 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
294 error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
295 (uint32_t)chunk_offset);
301 case GRAPH_CHUNKID_OIDFANOUT:
302 if (graph->chunk_oid_fanout)
305 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
308 case GRAPH_CHUNKID_OIDLOOKUP:
309 if (graph->chunk_oid_lookup)
312 graph->chunk_oid_lookup = data + chunk_offset;
315 case GRAPH_CHUNKID_DATA:
316 if (graph->chunk_commit_data)
319 graph->chunk_commit_data = data + chunk_offset;
322 case GRAPH_CHUNKID_EXTRAEDGES:
323 if (graph->chunk_extra_edges)
326 graph->chunk_extra_edges = data + chunk_offset;
329 case GRAPH_CHUNKID_BASE:
330 if (graph->chunk_base_graphs)
333 graph->chunk_base_graphs = data + chunk_offset;
336 case GRAPH_CHUNKID_BLOOMINDEXES:
337 if (graph->chunk_bloom_indexes)
340 graph->chunk_bloom_indexes = data + chunk_offset;
343 case GRAPH_CHUNKID_BLOOMDATA:
344 if (graph->chunk_bloom_data)
347 uint32_t hash_version;
348 graph->chunk_bloom_data = data + chunk_offset;
349 hash_version = get_be32(data + chunk_offset);
351 if (hash_version != 1)
354 graph->bloom_filter_settings = xmalloc(sizeof(struct bloom_filter_settings));
355 graph->bloom_filter_settings->hash_version = hash_version;
356 graph->bloom_filter_settings->num_hashes = get_be32(data + chunk_offset + 4);
357 graph->bloom_filter_settings->bits_per_entry = get_be32(data + chunk_offset + 8);
362 if (chunk_repeated) {
363 error(_("commit-graph chunk id %08x appears multiple times"), chunk_id);
368 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
370 graph->num_commits = (chunk_offset - last_chunk_offset)
374 last_chunk_id = chunk_id;
375 last_chunk_offset = chunk_offset;
378 if (graph->chunk_bloom_indexes && graph->chunk_bloom_data) {
379 init_bloom_filters();
381 /* We need both the bloom chunks to exist together. Else ignore the data */
382 graph->chunk_bloom_indexes = NULL;
383 graph->chunk_bloom_data = NULL;
384 graph->bloom_filter_settings = NULL;
387 hashcpy(graph->oid.hash, graph->data + graph->data_len - graph->hash_len);
389 if (verify_commit_graph_lite(graph)) {
397 static struct commit_graph *load_commit_graph_one(const char *graph_file,
398 struct object_directory *odb)
403 struct commit_graph *g;
404 int open_ok = open_commit_graph(graph_file, &fd, &st);
409 g = load_commit_graph_one_fd_st(fd, &st, odb);
412 g->filename = xstrdup(graph_file);
417 static struct commit_graph *load_commit_graph_v1(struct repository *r,
418 struct object_directory *odb)
420 char *graph_name = get_commit_graph_filename(odb);
421 struct commit_graph *g = load_commit_graph_one(graph_name, odb);
427 static int add_graph_to_chain(struct commit_graph *g,
428 struct commit_graph *chain,
429 struct object_id *oids,
432 struct commit_graph *cur_g = chain;
434 if (n && !g->chunk_base_graphs) {
435 warning(_("commit-graph has no base graphs chunk"));
443 !oideq(&oids[n], &cur_g->oid) ||
444 !hasheq(oids[n].hash, g->chunk_base_graphs + g->hash_len * n)) {
445 warning(_("commit-graph chain does not match"));
449 cur_g = cur_g->base_graph;
452 g->base_graph = chain;
455 g->num_commits_in_base = chain->num_commits + chain->num_commits_in_base;
460 static struct commit_graph *load_commit_graph_chain(struct repository *r,
461 struct object_directory *odb)
463 struct commit_graph *graph_chain = NULL;
464 struct strbuf line = STRBUF_INIT;
466 struct object_id *oids;
467 int i = 0, valid = 1, count;
468 char *chain_name = get_chain_filename(odb);
472 fp = fopen(chain_name, "r");
473 stat_res = stat(chain_name, &st);
478 st.st_size <= the_hash_algo->hexsz)
481 count = st.st_size / (the_hash_algo->hexsz + 1);
482 oids = xcalloc(count, sizeof(struct object_id));
486 for (i = 0; i < count; i++) {
487 struct object_directory *odb;
489 if (strbuf_getline_lf(&line, fp) == EOF)
492 if (get_oid_hex(line.buf, &oids[i])) {
493 warning(_("invalid commit-graph chain: line '%s' not a hash"),
500 for (odb = r->objects->odb; odb; odb = odb->next) {
501 char *graph_name = get_split_graph_filename(odb, line.buf);
502 struct commit_graph *g = load_commit_graph_one(graph_name, odb);
507 if (add_graph_to_chain(g, graph_chain, oids, i)) {
517 warning(_("unable to find all commit-graph files"));
524 strbuf_release(&line);
529 struct commit_graph *read_commit_graph_one(struct repository *r,
530 struct object_directory *odb)
532 struct commit_graph *g = load_commit_graph_v1(r, odb);
535 g = load_commit_graph_chain(r, odb);
540 static void prepare_commit_graph_one(struct repository *r,
541 struct object_directory *odb)
544 if (r->objects->commit_graph)
547 r->objects->commit_graph = read_commit_graph_one(r, odb);
551 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
553 * On the first invocation, this function attempts to load the commit
554 * graph if the_repository is configured to have one.
556 static int prepare_commit_graph(struct repository *r)
558 struct object_directory *odb;
561 * This must come before the "already attempted?" check below, because
562 * we want to disable even an already-loaded graph file.
564 if (r->commit_graph_disabled)
567 if (r->objects->commit_graph_attempted)
568 return !!r->objects->commit_graph;
569 r->objects->commit_graph_attempted = 1;
571 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
572 die("dying as requested by the '%s' variable on commit-graph load!",
573 GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD);
575 prepare_repo_settings(r);
577 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
578 r->settings.core_commit_graph != 1)
580 * This repository is not configured to use commit graphs, so
581 * do not load one. (But report commit_graph_attempted anyway
582 * so that commit graph loading is not attempted again for this
587 if (!commit_graph_compatible(r))
591 for (odb = r->objects->odb;
592 !r->objects->commit_graph && odb;
594 prepare_commit_graph_one(r, odb);
595 return !!r->objects->commit_graph;
598 int generation_numbers_enabled(struct repository *r)
600 uint32_t first_generation;
601 struct commit_graph *g;
602 if (!prepare_commit_graph(r))
605 g = r->objects->commit_graph;
610 first_generation = get_be32(g->chunk_commit_data +
611 g->hash_len + 8) >> 2;
613 return !!first_generation;
616 static void close_commit_graph_one(struct commit_graph *g)
621 close_commit_graph_one(g->base_graph);
622 free_commit_graph(g);
625 void close_commit_graph(struct raw_object_store *o)
627 close_commit_graph_one(o->commit_graph);
628 o->commit_graph = NULL;
631 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
633 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
634 g->chunk_oid_lookup, g->hash_len, pos);
637 static void load_oid_from_graph(struct commit_graph *g,
639 struct object_id *oid)
643 while (g && pos < g->num_commits_in_base)
647 BUG("NULL commit-graph");
649 if (pos >= g->num_commits + g->num_commits_in_base)
650 die(_("invalid commit position. commit-graph is likely corrupt"));
652 lex_index = pos - g->num_commits_in_base;
654 hashcpy(oid->hash, g->chunk_oid_lookup + g->hash_len * lex_index);
657 static struct commit_list **insert_parent_or_die(struct repository *r,
658 struct commit_graph *g,
660 struct commit_list **pptr)
663 struct object_id oid;
665 if (pos >= g->num_commits + g->num_commits_in_base)
666 die("invalid parent position %"PRIu32, pos);
668 load_oid_from_graph(g, pos, &oid);
669 c = lookup_commit(r, &oid);
671 die(_("could not find commit %s"), oid_to_hex(&oid));
673 return &commit_list_insert(c, pptr)->next;
676 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
678 const unsigned char *commit_data;
681 while (pos < g->num_commits_in_base)
684 lex_index = pos - g->num_commits_in_base;
685 commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * lex_index;
686 item->graph_pos = pos;
687 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
690 static inline void set_commit_tree(struct commit *c, struct tree *t)
695 static int fill_commit_in_graph(struct repository *r,
697 struct commit_graph *g, uint32_t pos)
700 uint32_t *parent_data_ptr;
701 uint64_t date_low, date_high;
702 struct commit_list **pptr;
703 const unsigned char *commit_data;
706 while (pos < g->num_commits_in_base)
709 if (pos >= g->num_commits + g->num_commits_in_base)
710 die(_("invalid commit position. commit-graph is likely corrupt"));
713 * Store the "full" position, but then use the
714 * "local" position for the rest of the calculation.
716 item->graph_pos = pos;
717 lex_index = pos - g->num_commits_in_base;
719 commit_data = g->chunk_commit_data + (g->hash_len + 16) * lex_index;
721 item->object.parsed = 1;
723 set_commit_tree(item, NULL);
725 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
726 date_low = get_be32(commit_data + g->hash_len + 12);
727 item->date = (timestamp_t)((date_high << 32) | date_low);
729 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
731 pptr = &item->parents;
733 edge_value = get_be32(commit_data + g->hash_len);
734 if (edge_value == GRAPH_PARENT_NONE)
736 pptr = insert_parent_or_die(r, g, edge_value, pptr);
738 edge_value = get_be32(commit_data + g->hash_len + 4);
739 if (edge_value == GRAPH_PARENT_NONE)
741 if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
742 pptr = insert_parent_or_die(r, g, edge_value, pptr);
746 parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
747 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
749 edge_value = get_be32(parent_data_ptr);
750 pptr = insert_parent_or_die(r, g,
751 edge_value & GRAPH_EDGE_LAST_MASK,
754 } while (!(edge_value & GRAPH_LAST_EDGE));
759 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
761 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
762 *pos = item->graph_pos;
765 struct commit_graph *cur_g = g;
768 while (cur_g && !bsearch_graph(cur_g, &(item->object.oid), &lex_index))
769 cur_g = cur_g->base_graph;
772 *pos = lex_index + cur_g->num_commits_in_base;
780 static int parse_commit_in_graph_one(struct repository *r,
781 struct commit_graph *g,
786 if (item->object.parsed)
789 if (find_commit_in_graph(item, g, &pos))
790 return fill_commit_in_graph(r, item, g, pos);
795 int parse_commit_in_graph(struct repository *r, struct commit *item)
797 if (!prepare_commit_graph(r))
799 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
802 void load_commit_graph_info(struct repository *r, struct commit *item)
805 if (!prepare_commit_graph(r))
807 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
808 fill_commit_graph_info(item, r->objects->commit_graph, pos);
811 static struct tree *load_tree_for_commit(struct repository *r,
812 struct commit_graph *g,
815 struct object_id oid;
816 const unsigned char *commit_data;
818 while (c->graph_pos < g->num_commits_in_base)
821 commit_data = g->chunk_commit_data +
822 GRAPH_DATA_WIDTH * (c->graph_pos - g->num_commits_in_base);
824 hashcpy(oid.hash, commit_data);
825 set_commit_tree(c, lookup_tree(r, &oid));
827 return c->maybe_tree;
830 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
831 struct commit_graph *g,
832 const struct commit *c)
835 return c->maybe_tree;
836 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
837 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
839 return load_tree_for_commit(r, g, (struct commit *)c);
842 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
844 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
847 struct packed_commit_list {
848 struct commit **list;
853 struct packed_oid_list {
854 struct object_id *list;
859 struct write_commit_graph_context {
860 struct repository *r;
861 struct object_directory *odb;
863 struct packed_oid_list oids;
864 struct packed_commit_list commits;
866 unsigned long approx_nr_objects;
867 struct progress *progress;
869 uint64_t progress_cnt;
871 char *base_graph_name;
872 int num_commit_graphs_before;
873 int num_commit_graphs_after;
874 char **commit_graph_filenames_before;
875 char **commit_graph_filenames_after;
876 char **commit_graph_hash_after;
877 uint32_t new_num_commits_in_base;
878 struct commit_graph *new_base_graph;
887 const struct split_commit_graph_opts *split_opts;
888 size_t total_bloom_filter_data_size;
891 static void write_graph_chunk_fanout(struct hashfile *f,
892 struct write_commit_graph_context *ctx)
895 struct commit **list = ctx->commits.list;
898 * Write the first-level table (the list is sorted,
899 * but we use a 256-entry lookup to be able to avoid
900 * having to do eight extra binary search iterations).
902 for (i = 0; i < 256; i++) {
903 while (count < ctx->commits.nr) {
904 if ((*list)->object.oid.hash[0] != i)
906 display_progress(ctx->progress, ++ctx->progress_cnt);
911 hashwrite_be32(f, count);
915 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
916 struct write_commit_graph_context *ctx)
918 struct commit **list = ctx->commits.list;
920 for (count = 0; count < ctx->commits.nr; count++, list++) {
921 display_progress(ctx->progress, ++ctx->progress_cnt);
922 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
926 static const unsigned char *commit_to_sha1(size_t index, void *table)
928 struct commit **commits = table;
929 return commits[index]->object.oid.hash;
932 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
933 struct write_commit_graph_context *ctx)
935 struct commit **list = ctx->commits.list;
936 struct commit **last = ctx->commits.list + ctx->commits.nr;
937 uint32_t num_extra_edges = 0;
939 while (list < last) {
940 struct commit_list *parent;
941 struct object_id *tree;
943 uint32_t packedDate[2];
944 display_progress(ctx->progress, ++ctx->progress_cnt);
946 if (parse_commit_no_graph(*list))
947 die(_("unable to parse commit %s"),
948 oid_to_hex(&(*list)->object.oid));
949 tree = get_commit_tree_oid(*list);
950 hashwrite(f, tree->hash, hash_len);
952 parent = (*list)->parents;
955 edge_value = GRAPH_PARENT_NONE;
957 edge_value = sha1_pos(parent->item->object.oid.hash,
963 edge_value += ctx->new_num_commits_in_base;
964 else if (ctx->new_base_graph) {
966 if (find_commit_in_graph(parent->item,
973 BUG("missing parent %s for commit %s",
974 oid_to_hex(&parent->item->object.oid),
975 oid_to_hex(&(*list)->object.oid));
978 hashwrite_be32(f, edge_value);
981 parent = parent->next;
984 edge_value = GRAPH_PARENT_NONE;
985 else if (parent->next)
986 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
988 edge_value = sha1_pos(parent->item->object.oid.hash,
994 edge_value += ctx->new_num_commits_in_base;
995 else if (ctx->new_base_graph) {
997 if (find_commit_in_graph(parent->item,
1004 BUG("missing parent %s for commit %s",
1005 oid_to_hex(&parent->item->object.oid),
1006 oid_to_hex(&(*list)->object.oid));
1009 hashwrite_be32(f, edge_value);
1011 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
1014 parent = parent->next;
1018 if (sizeof((*list)->date) > 4)
1019 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
1023 packedDate[0] |= htonl((*list)->generation << 2);
1025 packedDate[1] = htonl((*list)->date);
1026 hashwrite(f, packedDate, 8);
1032 static void write_graph_chunk_extra_edges(struct hashfile *f,
1033 struct write_commit_graph_context *ctx)
1035 struct commit **list = ctx->commits.list;
1036 struct commit **last = ctx->commits.list + ctx->commits.nr;
1037 struct commit_list *parent;
1039 while (list < last) {
1040 int num_parents = 0;
1042 display_progress(ctx->progress, ++ctx->progress_cnt);
1044 for (parent = (*list)->parents; num_parents < 3 && parent;
1045 parent = parent->next)
1048 if (num_parents <= 2) {
1053 /* Since num_parents > 2, this initializer is safe. */
1054 for (parent = (*list)->parents->next; parent; parent = parent->next) {
1055 int edge_value = sha1_pos(parent->item->object.oid.hash,
1060 if (edge_value >= 0)
1061 edge_value += ctx->new_num_commits_in_base;
1062 else if (ctx->new_base_graph) {
1064 if (find_commit_in_graph(parent->item,
1065 ctx->new_base_graph,
1071 BUG("missing parent %s for commit %s",
1072 oid_to_hex(&parent->item->object.oid),
1073 oid_to_hex(&(*list)->object.oid));
1074 else if (!parent->next)
1075 edge_value |= GRAPH_LAST_EDGE;
1077 hashwrite_be32(f, edge_value);
1084 static void write_graph_chunk_bloom_indexes(struct hashfile *f,
1085 struct write_commit_graph_context *ctx)
1087 struct commit **list = ctx->commits.list;
1088 struct commit **last = ctx->commits.list + ctx->commits.nr;
1089 uint32_t cur_pos = 0;
1090 struct progress *progress = NULL;
1093 if (ctx->report_progress)
1094 progress = start_delayed_progress(
1095 _("Writing changed paths Bloom filters index"),
1098 while (list < last) {
1099 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list, 0);
1100 cur_pos += filter->len;
1101 display_progress(progress, ++i);
1102 hashwrite_be32(f, cur_pos);
1106 stop_progress(&progress);
1109 static void write_graph_chunk_bloom_data(struct hashfile *f,
1110 struct write_commit_graph_context *ctx,
1111 const struct bloom_filter_settings *settings)
1113 struct commit **list = ctx->commits.list;
1114 struct commit **last = ctx->commits.list + ctx->commits.nr;
1115 struct progress *progress = NULL;
1118 if (ctx->report_progress)
1119 progress = start_delayed_progress(
1120 _("Writing changed paths Bloom filters data"),
1123 hashwrite_be32(f, settings->hash_version);
1124 hashwrite_be32(f, settings->num_hashes);
1125 hashwrite_be32(f, settings->bits_per_entry);
1127 while (list < last) {
1128 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list, 0);
1129 display_progress(progress, ++i);
1130 hashwrite(f, filter->data, filter->len * sizeof(unsigned char));
1134 stop_progress(&progress);
1137 static int oid_compare(const void *_a, const void *_b)
1139 const struct object_id *a = (const struct object_id *)_a;
1140 const struct object_id *b = (const struct object_id *)_b;
1141 return oidcmp(a, b);
1144 static int add_packed_commits(const struct object_id *oid,
1145 struct packed_git *pack,
1149 struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
1150 enum object_type type;
1151 off_t offset = nth_packed_object_offset(pack, pos);
1152 struct object_info oi = OBJECT_INFO_INIT;
1155 display_progress(ctx->progress, ++ctx->progress_done);
1158 if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
1159 die(_("unable to get type of object %s"), oid_to_hex(oid));
1161 if (type != OBJ_COMMIT)
1164 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1165 oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
1168 set_commit_pos(ctx->r, oid);
1173 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
1175 struct commit_list *parent;
1176 for (parent = commit->parents; parent; parent = parent->next) {
1177 if (!(parent->item->object.flags & REACHABLE)) {
1178 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1179 oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
1181 parent->item->object.flags |= REACHABLE;
1186 static void close_reachable(struct write_commit_graph_context *ctx)
1189 struct commit *commit;
1190 enum commit_graph_split_flags flags = ctx->split_opts ?
1191 ctx->split_opts->flags : COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1193 if (ctx->report_progress)
1194 ctx->progress = start_delayed_progress(
1195 _("Loading known commits in commit graph"),
1197 for (i = 0; i < ctx->oids.nr; i++) {
1198 display_progress(ctx->progress, i + 1);
1199 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1201 commit->object.flags |= REACHABLE;
1203 stop_progress(&ctx->progress);
1206 * As this loop runs, ctx->oids.nr may grow, but not more
1207 * than the number of missing commits in the reachable
1210 if (ctx->report_progress)
1211 ctx->progress = start_delayed_progress(
1212 _("Expanding reachable commits in commit graph"),
1214 for (i = 0; i < ctx->oids.nr; i++) {
1215 display_progress(ctx->progress, i + 1);
1216 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1221 if ((!parse_commit(commit) &&
1222 commit->graph_pos == COMMIT_NOT_FROM_GRAPH) ||
1223 flags == COMMIT_GRAPH_SPLIT_REPLACE)
1224 add_missing_parents(ctx, commit);
1225 } else if (!parse_commit_no_graph(commit))
1226 add_missing_parents(ctx, commit);
1228 stop_progress(&ctx->progress);
1230 if (ctx->report_progress)
1231 ctx->progress = start_delayed_progress(
1232 _("Clearing commit marks in commit graph"),
1234 for (i = 0; i < ctx->oids.nr; i++) {
1235 display_progress(ctx->progress, i + 1);
1236 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1239 commit->object.flags &= ~REACHABLE;
1241 stop_progress(&ctx->progress);
1244 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1247 struct commit_list *list = NULL;
1249 if (ctx->report_progress)
1250 ctx->progress = start_delayed_progress(
1251 _("Computing commit graph generation numbers"),
1253 for (i = 0; i < ctx->commits.nr; i++) {
1254 display_progress(ctx->progress, i + 1);
1255 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
1256 ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
1259 commit_list_insert(ctx->commits.list[i], &list);
1261 struct commit *current = list->item;
1262 struct commit_list *parent;
1263 int all_parents_computed = 1;
1264 uint32_t max_generation = 0;
1266 for (parent = current->parents; parent; parent = parent->next) {
1267 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
1268 parent->item->generation == GENERATION_NUMBER_ZERO) {
1269 all_parents_computed = 0;
1270 commit_list_insert(parent->item, &list);
1272 } else if (parent->item->generation > max_generation) {
1273 max_generation = parent->item->generation;
1277 if (all_parents_computed) {
1278 current->generation = max_generation + 1;
1281 if (current->generation > GENERATION_NUMBER_MAX)
1282 current->generation = GENERATION_NUMBER_MAX;
1286 stop_progress(&ctx->progress);
1289 static void compute_bloom_filters(struct write_commit_graph_context *ctx)
1292 struct progress *progress = NULL;
1293 struct commit **sorted_commits;
1295 init_bloom_filters();
1297 if (ctx->report_progress)
1298 progress = start_delayed_progress(
1299 _("Computing commit changed paths Bloom filters"),
1302 ALLOC_ARRAY(sorted_commits, ctx->commits.nr);
1303 COPY_ARRAY(sorted_commits, ctx->commits.list, ctx->commits.nr);
1305 if (ctx->order_by_pack)
1306 QSORT(sorted_commits, ctx->commits.nr, commit_pos_cmp);
1308 QSORT(sorted_commits, ctx->commits.nr, commit_gen_cmp);
1310 for (i = 0; i < ctx->commits.nr; i++) {
1311 struct commit *c = sorted_commits[i];
1312 struct bloom_filter *filter = get_bloom_filter(ctx->r, c, 1);
1313 ctx->total_bloom_filter_data_size += sizeof(unsigned char) * filter->len;
1314 display_progress(progress, i + 1);
1317 free(sorted_commits);
1318 stop_progress(&progress);
1321 static int add_ref_to_set(const char *refname,
1322 const struct object_id *oid,
1323 int flags, void *cb_data)
1325 struct oidset *commits = (struct oidset *)cb_data;
1327 oidset_insert(commits, oid);
1331 int write_commit_graph_reachable(struct object_directory *odb,
1332 enum commit_graph_write_flags flags,
1333 const struct split_commit_graph_opts *split_opts)
1335 struct oidset commits = OIDSET_INIT;
1338 for_each_ref(add_ref_to_set, &commits);
1339 result = write_commit_graph(odb, NULL, &commits,
1342 oidset_clear(&commits);
1346 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
1347 struct string_list *pack_indexes)
1350 struct strbuf progress_title = STRBUF_INIT;
1351 struct strbuf packname = STRBUF_INIT;
1354 strbuf_addf(&packname, "%s/pack/", ctx->odb->path);
1355 dirlen = packname.len;
1356 if (ctx->report_progress) {
1357 strbuf_addf(&progress_title,
1358 Q_("Finding commits for commit graph in %d pack",
1359 "Finding commits for commit graph in %d packs",
1362 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1363 ctx->progress_done = 0;
1365 for (i = 0; i < pack_indexes->nr; i++) {
1366 struct packed_git *p;
1367 strbuf_setlen(&packname, dirlen);
1368 strbuf_addstr(&packname, pack_indexes->items[i].string);
1369 p = add_packed_git(packname.buf, packname.len, 1);
1371 error(_("error adding pack %s"), packname.buf);
1374 if (open_pack_index(p)) {
1375 error(_("error opening index for %s"), packname.buf);
1378 for_each_object_in_pack(p, add_packed_commits, ctx,
1379 FOR_EACH_OBJECT_PACK_ORDER);
1384 stop_progress(&ctx->progress);
1385 strbuf_release(&progress_title);
1386 strbuf_release(&packname);
1391 static int fill_oids_from_commits(struct write_commit_graph_context *ctx,
1392 struct oidset *commits)
1395 struct strbuf progress_title = STRBUF_INIT;
1396 struct oidset_iter iter;
1397 struct object_id *oid;
1399 if (!oidset_size(commits))
1402 if (ctx->report_progress) {
1403 strbuf_addf(&progress_title,
1404 Q_("Finding commits for commit graph from %d ref",
1405 "Finding commits for commit graph from %d refs",
1406 oidset_size(commits)),
1407 oidset_size(commits));
1408 ctx->progress = start_delayed_progress(
1410 oidset_size(commits));
1413 oidset_iter_init(commits, &iter);
1414 while ((oid = oidset_iter_next(&iter))) {
1415 struct commit *result;
1417 display_progress(ctx->progress, ++i);
1419 result = lookup_commit_reference_gently(ctx->r, oid, 1);
1421 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1422 oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
1424 } else if (ctx->check_oids) {
1425 error(_("invalid commit object id: %s"),
1431 stop_progress(&ctx->progress);
1432 strbuf_release(&progress_title);
1437 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1439 if (ctx->report_progress)
1440 ctx->progress = start_delayed_progress(
1441 _("Finding commits for commit graph among packed objects"),
1442 ctx->approx_nr_objects);
1443 for_each_packed_object(add_packed_commits, ctx,
1444 FOR_EACH_OBJECT_PACK_ORDER);
1445 if (ctx->progress_done < ctx->approx_nr_objects)
1446 display_progress(ctx->progress, ctx->approx_nr_objects);
1447 stop_progress(&ctx->progress);
1450 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
1452 uint32_t i, count_distinct = 1;
1454 if (ctx->report_progress)
1455 ctx->progress = start_delayed_progress(
1456 _("Counting distinct commits in commit graph"),
1458 display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1459 QSORT(ctx->oids.list, ctx->oids.nr, oid_compare);
1461 for (i = 1; i < ctx->oids.nr; i++) {
1462 display_progress(ctx->progress, i + 1);
1463 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i])) {
1465 struct commit *c = lookup_commit(ctx->r, &ctx->oids.list[i]);
1467 if (!c || c->graph_pos != COMMIT_NOT_FROM_GRAPH)
1474 stop_progress(&ctx->progress);
1476 return count_distinct;
1479 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1482 enum commit_graph_split_flags flags = ctx->split_opts ?
1483 ctx->split_opts->flags : COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1485 ctx->num_extra_edges = 0;
1486 if (ctx->report_progress)
1487 ctx->progress = start_delayed_progress(
1488 _("Finding extra edges in commit graph"),
1490 for (i = 0; i < ctx->oids.nr; i++) {
1491 unsigned int num_parents;
1493 display_progress(ctx->progress, i + 1);
1494 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1497 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + 1, ctx->commits.alloc);
1498 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1500 if (ctx->split && flags != COMMIT_GRAPH_SPLIT_REPLACE &&
1501 ctx->commits.list[ctx->commits.nr]->graph_pos != COMMIT_NOT_FROM_GRAPH)
1504 if (ctx->split && flags == COMMIT_GRAPH_SPLIT_REPLACE)
1505 parse_commit(ctx->commits.list[ctx->commits.nr]);
1507 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1509 num_parents = commit_list_count(ctx->commits.list[ctx->commits.nr]->parents);
1510 if (num_parents > 2)
1511 ctx->num_extra_edges += num_parents - 1;
1515 stop_progress(&ctx->progress);
1518 static int write_graph_chunk_base_1(struct hashfile *f,
1519 struct commit_graph *g)
1526 num = write_graph_chunk_base_1(f, g->base_graph);
1527 hashwrite(f, g->oid.hash, the_hash_algo->rawsz);
1531 static int write_graph_chunk_base(struct hashfile *f,
1532 struct write_commit_graph_context *ctx)
1534 int num = write_graph_chunk_base_1(f, ctx->new_base_graph);
1536 if (num != ctx->num_commit_graphs_after - 1) {
1537 error(_("failed to write correct number of base graph ids"));
1544 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1549 struct lock_file lk = LOCK_INIT;
1550 uint32_t chunk_ids[MAX_NUM_CHUNKS + 1];
1551 uint64_t chunk_offsets[MAX_NUM_CHUNKS + 1];
1552 const unsigned hashsz = the_hash_algo->rawsz;
1553 struct strbuf progress_title = STRBUF_INIT;
1555 struct object_id file_hash;
1556 const struct bloom_filter_settings bloom_settings = DEFAULT_BLOOM_FILTER_SETTINGS;
1559 struct strbuf tmp_file = STRBUF_INIT;
1561 strbuf_addf(&tmp_file,
1562 "%s/info/commit-graphs/tmp_graph_XXXXXX",
1564 ctx->graph_name = strbuf_detach(&tmp_file, NULL);
1566 ctx->graph_name = get_commit_graph_filename(ctx->odb);
1569 if (safe_create_leading_directories(ctx->graph_name)) {
1570 UNLEAK(ctx->graph_name);
1571 error(_("unable to create leading directories of %s"),
1577 char *lock_name = get_chain_filename(ctx->odb);
1579 hold_lock_file_for_update_mode(&lk, lock_name,
1580 LOCK_DIE_ON_ERROR, 0444);
1582 fd = git_mkstemp_mode(ctx->graph_name, 0444);
1584 error(_("unable to create temporary graph layer"));
1588 if (adjust_shared_perm(ctx->graph_name)) {
1589 error(_("unable to adjust shared permissions for '%s'"),
1594 f = hashfd(fd, ctx->graph_name);
1596 hold_lock_file_for_update_mode(&lk, ctx->graph_name,
1597 LOCK_DIE_ON_ERROR, 0444);
1598 fd = lk.tempfile->fd;
1599 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1602 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1603 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1604 chunk_ids[2] = GRAPH_CHUNKID_DATA;
1605 if (ctx->num_extra_edges) {
1606 chunk_ids[num_chunks] = GRAPH_CHUNKID_EXTRAEDGES;
1609 if (ctx->changed_paths) {
1610 chunk_ids[num_chunks] = GRAPH_CHUNKID_BLOOMINDEXES;
1612 chunk_ids[num_chunks] = GRAPH_CHUNKID_BLOOMDATA;
1615 if (ctx->num_commit_graphs_after > 1) {
1616 chunk_ids[num_chunks] = GRAPH_CHUNKID_BASE;
1620 chunk_ids[num_chunks] = 0;
1622 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1623 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1624 chunk_offsets[2] = chunk_offsets[1] + hashsz * ctx->commits.nr;
1625 chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * ctx->commits.nr;
1628 if (ctx->num_extra_edges) {
1629 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1630 4 * ctx->num_extra_edges;
1633 if (ctx->changed_paths) {
1634 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1635 sizeof(uint32_t) * ctx->commits.nr;
1638 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1639 sizeof(uint32_t) * 3 + ctx->total_bloom_filter_data_size;
1642 if (ctx->num_commit_graphs_after > 1) {
1643 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1644 hashsz * (ctx->num_commit_graphs_after - 1);
1648 hashwrite_be32(f, GRAPH_SIGNATURE);
1650 hashwrite_u8(f, GRAPH_VERSION);
1651 hashwrite_u8(f, oid_version());
1652 hashwrite_u8(f, num_chunks);
1653 hashwrite_u8(f, ctx->num_commit_graphs_after - 1);
1655 for (i = 0; i <= num_chunks; i++) {
1656 uint32_t chunk_write[3];
1658 chunk_write[0] = htonl(chunk_ids[i]);
1659 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1660 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1661 hashwrite(f, chunk_write, 12);
1664 if (ctx->report_progress) {
1665 strbuf_addf(&progress_title,
1666 Q_("Writing out commit graph in %d pass",
1667 "Writing out commit graph in %d passes",
1670 ctx->progress = start_delayed_progress(
1672 num_chunks * ctx->commits.nr);
1674 write_graph_chunk_fanout(f, ctx);
1675 write_graph_chunk_oids(f, hashsz, ctx);
1676 write_graph_chunk_data(f, hashsz, ctx);
1677 if (ctx->num_extra_edges)
1678 write_graph_chunk_extra_edges(f, ctx);
1679 if (ctx->changed_paths) {
1680 write_graph_chunk_bloom_indexes(f, ctx);
1681 write_graph_chunk_bloom_data(f, ctx, &bloom_settings);
1683 if (ctx->num_commit_graphs_after > 1 &&
1684 write_graph_chunk_base(f, ctx)) {
1687 stop_progress(&ctx->progress);
1688 strbuf_release(&progress_title);
1690 if (ctx->split && ctx->base_graph_name && ctx->num_commit_graphs_after > 1) {
1691 char *new_base_hash = xstrdup(oid_to_hex(&ctx->new_base_graph->oid));
1692 char *new_base_name = get_split_graph_filename(ctx->new_base_graph->odb, new_base_hash);
1694 free(ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2]);
1695 free(ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2]);
1696 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2] = new_base_name;
1697 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2] = new_base_hash;
1700 close_commit_graph(ctx->r->objects);
1701 finalize_hashfile(f, file_hash.hash, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1704 FILE *chainf = fdopen_lock_file(&lk, "w");
1705 char *final_graph_name;
1711 error(_("unable to open commit-graph chain file"));
1715 if (ctx->base_graph_name) {
1717 int idx = ctx->num_commit_graphs_after - 1;
1718 if (ctx->num_commit_graphs_after > 1)
1721 dest = ctx->commit_graph_filenames_after[idx];
1723 if (strcmp(ctx->base_graph_name, dest)) {
1724 result = rename(ctx->base_graph_name, dest);
1727 error(_("failed to rename base commit-graph file"));
1732 char *graph_name = get_commit_graph_filename(ctx->odb);
1736 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1] = xstrdup(oid_to_hex(&file_hash));
1737 final_graph_name = get_split_graph_filename(ctx->odb,
1738 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1]);
1739 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 1] = final_graph_name;
1741 result = rename(ctx->graph_name, final_graph_name);
1743 for (i = 0; i < ctx->num_commit_graphs_after; i++)
1744 fprintf(lk.tempfile->fp, "%s\n", ctx->commit_graph_hash_after[i]);
1747 error(_("failed to rename temporary commit-graph file"));
1752 commit_lock_file(&lk);
1757 static void split_graph_merge_strategy(struct write_commit_graph_context *ctx)
1759 struct commit_graph *g;
1760 uint32_t num_commits;
1761 enum commit_graph_split_flags flags = COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1764 int max_commits = 0;
1767 if (ctx->split_opts) {
1768 max_commits = ctx->split_opts->max_commits;
1770 if (ctx->split_opts->size_multiple)
1771 size_mult = ctx->split_opts->size_multiple;
1773 flags = ctx->split_opts->flags;
1776 g = ctx->r->objects->commit_graph;
1777 num_commits = ctx->commits.nr;
1778 if (flags == COMMIT_GRAPH_SPLIT_REPLACE)
1779 ctx->num_commit_graphs_after = 1;
1781 ctx->num_commit_graphs_after = ctx->num_commit_graphs_before + 1;
1783 if (flags != COMMIT_GRAPH_SPLIT_MERGE_PROHIBITED &&
1784 flags != COMMIT_GRAPH_SPLIT_REPLACE) {
1785 while (g && (g->num_commits <= size_mult * num_commits ||
1786 (max_commits && num_commits > max_commits))) {
1787 if (g->odb != ctx->odb)
1790 num_commits += g->num_commits;
1793 ctx->num_commit_graphs_after--;
1797 if (flags != COMMIT_GRAPH_SPLIT_REPLACE)
1798 ctx->new_base_graph = g;
1799 else if (ctx->num_commit_graphs_after != 1)
1800 BUG("split_graph_merge_strategy: num_commit_graphs_after "
1801 "should be 1 with --split=replace");
1803 if (ctx->num_commit_graphs_after == 2) {
1804 char *old_graph_name = get_commit_graph_filename(g->odb);
1806 if (!strcmp(g->filename, old_graph_name) &&
1807 g->odb != ctx->odb) {
1808 ctx->num_commit_graphs_after = 1;
1809 ctx->new_base_graph = NULL;
1812 free(old_graph_name);
1815 CALLOC_ARRAY(ctx->commit_graph_filenames_after, ctx->num_commit_graphs_after);
1816 CALLOC_ARRAY(ctx->commit_graph_hash_after, ctx->num_commit_graphs_after);
1818 for (i = 0; i < ctx->num_commit_graphs_after &&
1819 i < ctx->num_commit_graphs_before; i++)
1820 ctx->commit_graph_filenames_after[i] = xstrdup(ctx->commit_graph_filenames_before[i]);
1822 i = ctx->num_commit_graphs_before - 1;
1823 g = ctx->r->objects->commit_graph;
1826 if (i < ctx->num_commit_graphs_after)
1827 ctx->commit_graph_hash_after[i] = xstrdup(oid_to_hex(&g->oid));
1834 static void merge_commit_graph(struct write_commit_graph_context *ctx,
1835 struct commit_graph *g)
1838 uint32_t offset = g->num_commits_in_base;
1840 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + g->num_commits, ctx->commits.alloc);
1842 for (i = 0; i < g->num_commits; i++) {
1843 struct object_id oid;
1844 struct commit *result;
1846 display_progress(ctx->progress, i + 1);
1848 load_oid_from_graph(g, i + offset, &oid);
1850 /* only add commits if they still exist in the repo */
1851 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1854 ctx->commits.list[ctx->commits.nr] = result;
1860 static int commit_compare(const void *_a, const void *_b)
1862 const struct commit *a = *(const struct commit **)_a;
1863 const struct commit *b = *(const struct commit **)_b;
1864 return oidcmp(&a->object.oid, &b->object.oid);
1867 static void sort_and_scan_merged_commits(struct write_commit_graph_context *ctx)
1871 if (ctx->report_progress)
1872 ctx->progress = start_delayed_progress(
1873 _("Scanning merged commits"),
1876 QSORT(ctx->commits.list, ctx->commits.nr, commit_compare);
1878 ctx->num_extra_edges = 0;
1879 for (i = 0; i < ctx->commits.nr; i++) {
1880 display_progress(ctx->progress, i);
1882 if (i && oideq(&ctx->commits.list[i - 1]->object.oid,
1883 &ctx->commits.list[i]->object.oid)) {
1884 die(_("unexpected duplicate commit id %s"),
1885 oid_to_hex(&ctx->commits.list[i]->object.oid));
1887 unsigned int num_parents;
1889 num_parents = commit_list_count(ctx->commits.list[i]->parents);
1890 if (num_parents > 2)
1891 ctx->num_extra_edges += num_parents - 1;
1895 stop_progress(&ctx->progress);
1898 static void merge_commit_graphs(struct write_commit_graph_context *ctx)
1900 struct commit_graph *g = ctx->r->objects->commit_graph;
1901 uint32_t current_graph_number = ctx->num_commit_graphs_before;
1903 while (g && current_graph_number >= ctx->num_commit_graphs_after) {
1904 current_graph_number--;
1906 if (ctx->report_progress)
1907 ctx->progress = start_delayed_progress(_("Merging commit-graph"), 0);
1909 merge_commit_graph(ctx, g);
1910 stop_progress(&ctx->progress);
1916 ctx->new_base_graph = g;
1917 ctx->new_num_commits_in_base = g->num_commits + g->num_commits_in_base;
1920 if (ctx->new_base_graph)
1921 ctx->base_graph_name = xstrdup(ctx->new_base_graph->filename);
1923 sort_and_scan_merged_commits(ctx);
1926 static void mark_commit_graphs(struct write_commit_graph_context *ctx)
1929 time_t now = time(NULL);
1931 for (i = ctx->num_commit_graphs_after - 1; i < ctx->num_commit_graphs_before; i++) {
1933 struct utimbuf updated_time;
1935 stat(ctx->commit_graph_filenames_before[i], &st);
1937 updated_time.actime = st.st_atime;
1938 updated_time.modtime = now;
1939 utime(ctx->commit_graph_filenames_before[i], &updated_time);
1943 static void expire_commit_graphs(struct write_commit_graph_context *ctx)
1945 struct strbuf path = STRBUF_INIT;
1949 timestamp_t expire_time = time(NULL);
1951 if (ctx->split_opts && ctx->split_opts->expire_time)
1952 expire_time = ctx->split_opts->expire_time;
1954 char *chain_file_name = get_chain_filename(ctx->odb);
1955 unlink(chain_file_name);
1956 free(chain_file_name);
1957 ctx->num_commit_graphs_after = 0;
1960 strbuf_addstr(&path, ctx->odb->path);
1961 strbuf_addstr(&path, "/info/commit-graphs");
1962 dir = opendir(path.buf);
1967 strbuf_addch(&path, '/');
1968 dirnamelen = path.len;
1969 while ((de = readdir(dir)) != NULL) {
1971 uint32_t i, found = 0;
1973 strbuf_setlen(&path, dirnamelen);
1974 strbuf_addstr(&path, de->d_name);
1976 stat(path.buf, &st);
1978 if (st.st_mtime > expire_time)
1980 if (path.len < 6 || strcmp(path.buf + path.len - 6, ".graph"))
1983 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
1984 if (!strcmp(ctx->commit_graph_filenames_after[i],
1996 strbuf_release(&path);
1999 int write_commit_graph(struct object_directory *odb,
2000 struct string_list *pack_indexes,
2001 struct oidset *commits,
2002 enum commit_graph_write_flags flags,
2003 const struct split_commit_graph_opts *split_opts)
2005 struct write_commit_graph_context *ctx;
2006 uint32_t i, count_distinct = 0;
2010 if (!commit_graph_compatible(the_repository))
2013 ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
2014 ctx->r = the_repository;
2016 ctx->append = flags & COMMIT_GRAPH_WRITE_APPEND ? 1 : 0;
2017 ctx->report_progress = flags & COMMIT_GRAPH_WRITE_PROGRESS ? 1 : 0;
2018 ctx->split = flags & COMMIT_GRAPH_WRITE_SPLIT ? 1 : 0;
2019 ctx->check_oids = flags & COMMIT_GRAPH_WRITE_CHECK_OIDS ? 1 : 0;
2020 ctx->split_opts = split_opts;
2021 ctx->changed_paths = flags & COMMIT_GRAPH_WRITE_BLOOM_FILTERS ? 1 : 0;
2022 ctx->total_bloom_filter_data_size = 0;
2025 struct commit_graph *g;
2026 prepare_commit_graph(ctx->r);
2028 g = ctx->r->objects->commit_graph;
2031 ctx->num_commit_graphs_before++;
2035 if (ctx->num_commit_graphs_before) {
2036 ALLOC_ARRAY(ctx->commit_graph_filenames_before, ctx->num_commit_graphs_before);
2037 i = ctx->num_commit_graphs_before;
2038 g = ctx->r->objects->commit_graph;
2041 ctx->commit_graph_filenames_before[--i] = xstrdup(g->filename);
2046 if (ctx->split_opts)
2047 replace = ctx->split_opts->flags & COMMIT_GRAPH_SPLIT_REPLACE;
2050 ctx->approx_nr_objects = approximate_object_count();
2051 ctx->oids.alloc = ctx->approx_nr_objects / 32;
2053 if (ctx->split && split_opts && ctx->oids.alloc > split_opts->max_commits)
2054 ctx->oids.alloc = split_opts->max_commits;
2057 prepare_commit_graph_one(ctx->r, ctx->odb);
2058 if (ctx->r->objects->commit_graph)
2059 ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
2062 if (ctx->oids.alloc < 1024)
2063 ctx->oids.alloc = 1024;
2064 ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
2066 if (ctx->append && ctx->r->objects->commit_graph) {
2067 struct commit_graph *g = ctx->r->objects->commit_graph;
2068 for (i = 0; i < g->num_commits; i++) {
2069 const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
2070 hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
2075 ctx->order_by_pack = 1;
2076 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
2081 if ((res = fill_oids_from_commits(ctx, commits)))
2085 if (!pack_indexes && !commits) {
2086 ctx->order_by_pack = 1;
2087 fill_oids_from_all_packs(ctx);
2090 close_reachable(ctx);
2092 count_distinct = count_distinct_commits(ctx);
2094 if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
2095 error(_("the commit graph format cannot write %d commits"), count_distinct);
2100 ctx->commits.alloc = count_distinct;
2101 ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
2103 copy_oids_to_commits(ctx);
2105 if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
2106 error(_("too many commits to write graph"));
2111 if (!ctx->commits.nr && !replace)
2115 split_graph_merge_strategy(ctx);
2118 merge_commit_graphs(ctx);
2120 ctx->num_commit_graphs_after = 1;
2122 compute_generation_numbers(ctx);
2124 if (ctx->changed_paths)
2125 compute_bloom_filters(ctx);
2127 res = write_commit_graph_file(ctx);
2130 mark_commit_graphs(ctx);
2132 expire_commit_graphs(ctx);
2135 free(ctx->graph_name);
2136 free(ctx->commits.list);
2137 free(ctx->oids.list);
2139 if (ctx->commit_graph_filenames_after) {
2140 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
2141 free(ctx->commit_graph_filenames_after[i]);
2142 free(ctx->commit_graph_hash_after[i]);
2145 for (i = 0; i < ctx->num_commit_graphs_before; i++)
2146 free(ctx->commit_graph_filenames_before[i]);
2148 free(ctx->commit_graph_filenames_after);
2149 free(ctx->commit_graph_filenames_before);
2150 free(ctx->commit_graph_hash_after);
2158 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
2159 static int verify_commit_graph_error;
2161 static void graph_report(const char *fmt, ...)
2165 verify_commit_graph_error = 1;
2167 vfprintf(stderr, fmt, ap);
2168 fprintf(stderr, "\n");
2172 #define GENERATION_ZERO_EXISTS 1
2173 #define GENERATION_NUMBER_EXISTS 2
2175 int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
2177 uint32_t i, cur_fanout_pos = 0;
2178 struct object_id prev_oid, cur_oid, checksum;
2179 int generation_zero = 0;
2182 struct progress *progress = NULL;
2183 int local_error = 0;
2186 graph_report("no commit-graph file loaded");
2190 verify_commit_graph_error = verify_commit_graph_lite(g);
2191 if (verify_commit_graph_error)
2192 return verify_commit_graph_error;
2194 devnull = open("/dev/null", O_WRONLY);
2195 f = hashfd(devnull, NULL);
2196 hashwrite(f, g->data, g->data_len - g->hash_len);
2197 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
2198 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
2199 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
2200 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
2203 for (i = 0; i < g->num_commits; i++) {
2204 struct commit *graph_commit;
2206 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2208 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
2209 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
2210 oid_to_hex(&prev_oid),
2211 oid_to_hex(&cur_oid));
2213 oidcpy(&prev_oid, &cur_oid);
2215 while (cur_oid.hash[0] > cur_fanout_pos) {
2216 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2218 if (i != fanout_value)
2219 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2220 cur_fanout_pos, fanout_value, i);
2224 graph_commit = lookup_commit(r, &cur_oid);
2225 if (!parse_commit_in_graph_one(r, g, graph_commit))
2226 graph_report(_("failed to parse commit %s from commit-graph"),
2227 oid_to_hex(&cur_oid));
2230 while (cur_fanout_pos < 256) {
2231 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2233 if (g->num_commits != fanout_value)
2234 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2235 cur_fanout_pos, fanout_value, i);
2240 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
2241 return verify_commit_graph_error;
2243 if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
2244 progress = start_progress(_("Verifying commits in commit graph"),
2247 for (i = 0; i < g->num_commits; i++) {
2248 struct commit *graph_commit, *odb_commit;
2249 struct commit_list *graph_parents, *odb_parents;
2250 uint32_t max_generation = 0;
2252 display_progress(progress, i + 1);
2253 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2255 graph_commit = lookup_commit(r, &cur_oid);
2256 odb_commit = (struct commit *)create_object(r, &cur_oid, alloc_commit_node(r));
2257 if (parse_commit_internal(odb_commit, 0, 0)) {
2258 graph_report(_("failed to parse commit %s from object database for commit-graph"),
2259 oid_to_hex(&cur_oid));
2263 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
2264 get_commit_tree_oid(odb_commit)))
2265 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2266 oid_to_hex(&cur_oid),
2267 oid_to_hex(get_commit_tree_oid(graph_commit)),
2268 oid_to_hex(get_commit_tree_oid(odb_commit)));
2270 graph_parents = graph_commit->parents;
2271 odb_parents = odb_commit->parents;
2273 while (graph_parents) {
2274 if (odb_parents == NULL) {
2275 graph_report(_("commit-graph parent list for commit %s is too long"),
2276 oid_to_hex(&cur_oid));
2280 /* parse parent in case it is in a base graph */
2281 parse_commit_in_graph_one(r, g, graph_parents->item);
2283 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
2284 graph_report(_("commit-graph parent for %s is %s != %s"),
2285 oid_to_hex(&cur_oid),
2286 oid_to_hex(&graph_parents->item->object.oid),
2287 oid_to_hex(&odb_parents->item->object.oid));
2289 if (graph_parents->item->generation > max_generation)
2290 max_generation = graph_parents->item->generation;
2292 graph_parents = graph_parents->next;
2293 odb_parents = odb_parents->next;
2296 if (odb_parents != NULL)
2297 graph_report(_("commit-graph parent list for commit %s terminates early"),
2298 oid_to_hex(&cur_oid));
2300 if (!graph_commit->generation) {
2301 if (generation_zero == GENERATION_NUMBER_EXISTS)
2302 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2303 oid_to_hex(&cur_oid));
2304 generation_zero = GENERATION_ZERO_EXISTS;
2305 } else if (generation_zero == GENERATION_ZERO_EXISTS)
2306 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2307 oid_to_hex(&cur_oid));
2309 if (generation_zero == GENERATION_ZERO_EXISTS)
2313 * If one of our parents has generation GENERATION_NUMBER_MAX, then
2314 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
2315 * extra logic in the following condition.
2317 if (max_generation == GENERATION_NUMBER_MAX)
2320 if (graph_commit->generation != max_generation + 1)
2321 graph_report(_("commit-graph generation for commit %s is %u != %u"),
2322 oid_to_hex(&cur_oid),
2323 graph_commit->generation,
2324 max_generation + 1);
2326 if (graph_commit->date != odb_commit->date)
2327 graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
2328 oid_to_hex(&cur_oid),
2332 stop_progress(&progress);
2334 local_error = verify_commit_graph_error;
2336 if (!(flags & COMMIT_GRAPH_VERIFY_SHALLOW) && g->base_graph)
2337 local_error |= verify_commit_graph(r, g->base_graph, flags);
2342 void free_commit_graph(struct commit_graph *g)
2347 munmap((void *)g->data, g->data_len);
2351 free(g->bloom_filter_settings);
2355 void disable_commit_graph(struct repository *r)
2357 r->commit_graph_disabled = 1;