1 #include "git-compat-util.h"
10 #include "sha1-lookup.h"
11 #include "commit-graph.h"
12 #include "object-store.h"
15 #include "replace-object.h"
18 #include "commit-slab.h"
20 #include "json-writer.h"
23 void git_test_write_commit_graph_or_die(void)
26 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0))
29 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS, 0))
30 flags = COMMIT_GRAPH_WRITE_BLOOM_FILTERS;
32 if (write_commit_graph_reachable(the_repository->objects->odb,
34 die("failed to write commit-graph under GIT_TEST_COMMIT_GRAPH");
37 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
38 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
39 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
40 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
41 #define GRAPH_CHUNKID_EXTRAEDGES 0x45444745 /* "EDGE" */
42 #define GRAPH_CHUNKID_BLOOMINDEXES 0x42494458 /* "BIDX" */
43 #define GRAPH_CHUNKID_BLOOMDATA 0x42444154 /* "BDAT" */
44 #define GRAPH_CHUNKID_BASE 0x42415345 /* "BASE" */
45 #define MAX_NUM_CHUNKS 7
47 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
49 #define GRAPH_VERSION_1 0x1
50 #define GRAPH_VERSION GRAPH_VERSION_1
52 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
53 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
54 #define GRAPH_PARENT_NONE 0x70000000
56 #define GRAPH_LAST_EDGE 0x80000000
58 #define GRAPH_HEADER_SIZE 8
59 #define GRAPH_FANOUT_SIZE (4 * 256)
60 #define GRAPH_CHUNKLOOKUP_WIDTH 12
61 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
62 + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
64 /* Remember to update object flag allocation in object.h */
65 #define REACHABLE (1u<<15)
67 /* Keep track of the order in which commits are added to our list. */
68 define_commit_slab(commit_pos, int);
69 static struct commit_pos commit_pos = COMMIT_SLAB_INIT(1, commit_pos);
71 static void set_commit_pos(struct repository *r, const struct object_id *oid)
73 static int32_t max_pos;
74 struct commit *commit = lookup_commit(r, oid);
77 return; /* should never happen, but be lenient */
79 *commit_pos_at(&commit_pos, commit) = max_pos++;
82 static int commit_pos_cmp(const void *va, const void *vb)
84 const struct commit *a = *(const struct commit **)va;
85 const struct commit *b = *(const struct commit **)vb;
86 return commit_pos_at(&commit_pos, a) -
87 commit_pos_at(&commit_pos, b);
90 define_commit_slab(commit_graph_data_slab, struct commit_graph_data);
91 static struct commit_graph_data_slab commit_graph_data_slab =
92 COMMIT_SLAB_INIT(1, commit_graph_data_slab);
94 uint32_t commit_graph_position(const struct commit *c)
96 struct commit_graph_data *data =
97 commit_graph_data_slab_peek(&commit_graph_data_slab, c);
99 return data ? data->graph_pos : COMMIT_NOT_FROM_GRAPH;
102 uint32_t commit_graph_generation(const struct commit *c)
104 struct commit_graph_data *data =
105 commit_graph_data_slab_peek(&commit_graph_data_slab, c);
108 return GENERATION_NUMBER_INFINITY;
109 else if (data->graph_pos == COMMIT_NOT_FROM_GRAPH)
110 return GENERATION_NUMBER_INFINITY;
112 return data->generation;
115 static struct commit_graph_data *commit_graph_data_at(const struct commit *c)
117 unsigned int i, nth_slab;
118 struct commit_graph_data *data =
119 commit_graph_data_slab_peek(&commit_graph_data_slab, c);
124 nth_slab = c->index / commit_graph_data_slab.slab_size;
125 data = commit_graph_data_slab_at(&commit_graph_data_slab, c);
128 * commit-slab initializes elements with zero, overwrite this with
129 * COMMIT_NOT_FROM_GRAPH for graph_pos.
131 * We avoid initializing generation with checking if graph position
132 * is not COMMIT_NOT_FROM_GRAPH.
134 for (i = 0; i < commit_graph_data_slab.slab_size; i++) {
135 commit_graph_data_slab.slab[nth_slab][i].graph_pos =
136 COMMIT_NOT_FROM_GRAPH;
142 static int commit_gen_cmp(const void *va, const void *vb)
144 const struct commit *a = *(const struct commit **)va;
145 const struct commit *b = *(const struct commit **)vb;
147 uint32_t generation_a = commit_graph_generation(a);
148 uint32_t generation_b = commit_graph_generation(b);
149 /* lower generation commits first */
150 if (generation_a < generation_b)
152 else if (generation_a > generation_b)
155 /* use date as a heuristic when generations are equal */
156 if (a->date < b->date)
158 else if (a->date > b->date)
163 char *get_commit_graph_filename(struct object_directory *obj_dir)
165 return xstrfmt("%s/info/commit-graph", obj_dir->path);
168 static char *get_split_graph_filename(struct object_directory *odb,
171 return xstrfmt("%s/info/commit-graphs/graph-%s.graph", odb->path,
175 static char *get_chain_filename(struct object_directory *odb)
177 return xstrfmt("%s/info/commit-graphs/commit-graph-chain", odb->path);
180 static uint8_t oid_version(void)
185 static struct commit_graph *alloc_commit_graph(void)
187 struct commit_graph *g = xcalloc(1, sizeof(*g));
192 extern int read_replace_refs;
194 static int commit_graph_compatible(struct repository *r)
199 if (read_replace_refs) {
200 prepare_replace_object(r);
201 if (hashmap_get_size(&r->objects->replace_map->map))
205 prepare_commit_graft(r);
206 if (r->parsed_objects &&
207 (r->parsed_objects->grafts_nr || r->parsed_objects->substituted_parent))
209 if (is_repository_shallow(r))
215 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
217 *fd = git_open(graph_file);
220 if (fstat(*fd, st)) {
227 struct commit_graph *load_commit_graph_one_fd_st(struct repository *r,
228 int fd, struct stat *st,
229 struct object_directory *odb)
233 struct commit_graph *ret;
235 graph_size = xsize_t(st->st_size);
237 if (graph_size < GRAPH_MIN_SIZE) {
239 error(_("commit-graph file is too small"));
242 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
244 ret = parse_commit_graph(r, graph_map, graph_size);
249 munmap(graph_map, graph_size);
254 static int verify_commit_graph_lite(struct commit_graph *g)
257 * Basic validation shared between parse_commit_graph()
258 * which'll be called every time the graph is used, and the
259 * much more expensive verify_commit_graph() used by
260 * "commit-graph verify".
262 * There should only be very basic checks here to ensure that
263 * we don't e.g. segfault in fill_commit_in_graph(), but
264 * because this is a very hot codepath nothing that e.g. loops
265 * over g->num_commits, or runs a checksum on the commit-graph
268 if (!g->chunk_oid_fanout) {
269 error("commit-graph is missing the OID Fanout chunk");
272 if (!g->chunk_oid_lookup) {
273 error("commit-graph is missing the OID Lookup chunk");
276 if (!g->chunk_commit_data) {
277 error("commit-graph is missing the Commit Data chunk");
284 struct commit_graph *parse_commit_graph(struct repository *r,
285 void *graph_map, size_t graph_size)
287 const unsigned char *data, *chunk_lookup;
289 struct commit_graph *graph;
290 uint64_t next_chunk_offset;
291 uint32_t graph_signature;
292 unsigned char graph_version, hash_version;
297 if (graph_size < GRAPH_MIN_SIZE)
300 data = (const unsigned char *)graph_map;
302 graph_signature = get_be32(data);
303 if (graph_signature != GRAPH_SIGNATURE) {
304 error(_("commit-graph signature %X does not match signature %X"),
305 graph_signature, GRAPH_SIGNATURE);
309 graph_version = *(unsigned char*)(data + 4);
310 if (graph_version != GRAPH_VERSION) {
311 error(_("commit-graph version %X does not match version %X"),
312 graph_version, GRAPH_VERSION);
316 hash_version = *(unsigned char*)(data + 5);
317 if (hash_version != oid_version()) {
318 error(_("commit-graph hash version %X does not match version %X"),
319 hash_version, oid_version());
323 graph = alloc_commit_graph();
325 graph->hash_len = the_hash_algo->rawsz;
326 graph->num_chunks = *(unsigned char*)(data + 6);
327 graph->data = graph_map;
328 graph->data_len = graph_size;
330 if (graph_size < GRAPH_HEADER_SIZE +
331 (graph->num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH +
332 GRAPH_FANOUT_SIZE + the_hash_algo->rawsz) {
333 error(_("commit-graph file is too small to hold %u chunks"),
339 chunk_lookup = data + 8;
340 next_chunk_offset = get_be64(chunk_lookup + 4);
341 for (i = 0; i < graph->num_chunks; i++) {
343 uint64_t chunk_offset = next_chunk_offset;
344 int chunk_repeated = 0;
346 chunk_id = get_be32(chunk_lookup + 0);
348 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
349 next_chunk_offset = get_be64(chunk_lookup + 4);
351 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
352 error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
353 (uint32_t)chunk_offset);
354 goto free_and_return;
358 case GRAPH_CHUNKID_OIDFANOUT:
359 if (graph->chunk_oid_fanout)
362 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
365 case GRAPH_CHUNKID_OIDLOOKUP:
366 if (graph->chunk_oid_lookup)
369 graph->chunk_oid_lookup = data + chunk_offset;
370 graph->num_commits = (next_chunk_offset - chunk_offset)
375 case GRAPH_CHUNKID_DATA:
376 if (graph->chunk_commit_data)
379 graph->chunk_commit_data = data + chunk_offset;
382 case GRAPH_CHUNKID_EXTRAEDGES:
383 if (graph->chunk_extra_edges)
386 graph->chunk_extra_edges = data + chunk_offset;
389 case GRAPH_CHUNKID_BASE:
390 if (graph->chunk_base_graphs)
393 graph->chunk_base_graphs = data + chunk_offset;
396 case GRAPH_CHUNKID_BLOOMINDEXES:
397 if (graph->chunk_bloom_indexes)
400 graph->chunk_bloom_indexes = data + chunk_offset;
403 case GRAPH_CHUNKID_BLOOMDATA:
404 if (graph->chunk_bloom_data)
407 uint32_t hash_version;
408 graph->chunk_bloom_data = data + chunk_offset;
409 hash_version = get_be32(data + chunk_offset);
411 if (hash_version != 1)
414 graph->bloom_filter_settings = xmalloc(sizeof(struct bloom_filter_settings));
415 graph->bloom_filter_settings->hash_version = hash_version;
416 graph->bloom_filter_settings->num_hashes = get_be32(data + chunk_offset + 4);
417 graph->bloom_filter_settings->bits_per_entry = get_be32(data + chunk_offset + 8);
422 if (chunk_repeated) {
423 error(_("commit-graph chunk id %08x appears multiple times"), chunk_id);
424 goto free_and_return;
428 if (graph->chunk_bloom_indexes && graph->chunk_bloom_data) {
429 init_bloom_filters();
431 /* We need both the bloom chunks to exist together. Else ignore the data */
432 graph->chunk_bloom_indexes = NULL;
433 graph->chunk_bloom_data = NULL;
434 FREE_AND_NULL(graph->bloom_filter_settings);
437 hashcpy(graph->oid.hash, graph->data + graph->data_len - graph->hash_len);
439 if (verify_commit_graph_lite(graph))
440 goto free_and_return;
445 free(graph->bloom_filter_settings);
450 static struct commit_graph *load_commit_graph_one(struct repository *r,
451 const char *graph_file,
452 struct object_directory *odb)
457 struct commit_graph *g;
458 int open_ok = open_commit_graph(graph_file, &fd, &st);
463 g = load_commit_graph_one_fd_st(r, fd, &st, odb);
466 g->filename = xstrdup(graph_file);
471 static struct commit_graph *load_commit_graph_v1(struct repository *r,
472 struct object_directory *odb)
474 char *graph_name = get_commit_graph_filename(odb);
475 struct commit_graph *g = load_commit_graph_one(r, graph_name, odb);
481 static int add_graph_to_chain(struct commit_graph *g,
482 struct commit_graph *chain,
483 struct object_id *oids,
486 struct commit_graph *cur_g = chain;
488 if (n && !g->chunk_base_graphs) {
489 warning(_("commit-graph has no base graphs chunk"));
497 !oideq(&oids[n], &cur_g->oid) ||
498 !hasheq(oids[n].hash, g->chunk_base_graphs + g->hash_len * n)) {
499 warning(_("commit-graph chain does not match"));
503 cur_g = cur_g->base_graph;
506 g->base_graph = chain;
509 g->num_commits_in_base = chain->num_commits + chain->num_commits_in_base;
514 static struct commit_graph *load_commit_graph_chain(struct repository *r,
515 struct object_directory *odb)
517 struct commit_graph *graph_chain = NULL;
518 struct strbuf line = STRBUF_INIT;
520 struct object_id *oids;
521 int i = 0, valid = 1, count;
522 char *chain_name = get_chain_filename(odb);
526 fp = fopen(chain_name, "r");
527 stat_res = stat(chain_name, &st);
532 st.st_size <= the_hash_algo->hexsz)
535 count = st.st_size / (the_hash_algo->hexsz + 1);
536 oids = xcalloc(count, sizeof(struct object_id));
540 for (i = 0; i < count; i++) {
541 struct object_directory *odb;
543 if (strbuf_getline_lf(&line, fp) == EOF)
546 if (get_oid_hex(line.buf, &oids[i])) {
547 warning(_("invalid commit-graph chain: line '%s' not a hash"),
554 for (odb = r->objects->odb; odb; odb = odb->next) {
555 char *graph_name = get_split_graph_filename(odb, line.buf);
556 struct commit_graph *g = load_commit_graph_one(r, graph_name, odb);
561 if (add_graph_to_chain(g, graph_chain, oids, i)) {
571 warning(_("unable to find all commit-graph files"));
578 strbuf_release(&line);
583 struct commit_graph *read_commit_graph_one(struct repository *r,
584 struct object_directory *odb)
586 struct commit_graph *g = load_commit_graph_v1(r, odb);
589 g = load_commit_graph_chain(r, odb);
594 static void prepare_commit_graph_one(struct repository *r,
595 struct object_directory *odb)
598 if (r->objects->commit_graph)
601 r->objects->commit_graph = read_commit_graph_one(r, odb);
605 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
607 * On the first invocation, this function attempts to load the commit
608 * graph if the_repository is configured to have one.
610 static int prepare_commit_graph(struct repository *r)
612 struct object_directory *odb;
615 * This must come before the "already attempted?" check below, because
616 * we want to disable even an already-loaded graph file.
618 if (r->commit_graph_disabled)
621 if (r->objects->commit_graph_attempted)
622 return !!r->objects->commit_graph;
623 r->objects->commit_graph_attempted = 1;
625 prepare_repo_settings(r);
627 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
628 r->settings.core_commit_graph != 1)
630 * This repository is not configured to use commit graphs, so
631 * do not load one. (But report commit_graph_attempted anyway
632 * so that commit graph loading is not attempted again for this
637 if (!commit_graph_compatible(r))
641 for (odb = r->objects->odb;
642 !r->objects->commit_graph && odb;
644 prepare_commit_graph_one(r, odb);
645 return !!r->objects->commit_graph;
648 int generation_numbers_enabled(struct repository *r)
650 uint32_t first_generation;
651 struct commit_graph *g;
652 if (!prepare_commit_graph(r))
655 g = r->objects->commit_graph;
660 first_generation = get_be32(g->chunk_commit_data +
661 g->hash_len + 8) >> 2;
663 return !!first_generation;
666 struct bloom_filter_settings *get_bloom_filter_settings(struct repository *r)
668 struct commit_graph *g = r->objects->commit_graph;
670 if (g->bloom_filter_settings)
671 return g->bloom_filter_settings;
677 static void close_commit_graph_one(struct commit_graph *g)
682 close_commit_graph_one(g->base_graph);
683 free_commit_graph(g);
686 void close_commit_graph(struct raw_object_store *o)
688 close_commit_graph_one(o->commit_graph);
689 o->commit_graph = NULL;
692 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
694 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
695 g->chunk_oid_lookup, g->hash_len, pos);
698 static void load_oid_from_graph(struct commit_graph *g,
700 struct object_id *oid)
704 while (g && pos < g->num_commits_in_base)
708 BUG("NULL commit-graph");
710 if (pos >= g->num_commits + g->num_commits_in_base)
711 die(_("invalid commit position. commit-graph is likely corrupt"));
713 lex_index = pos - g->num_commits_in_base;
715 hashcpy(oid->hash, g->chunk_oid_lookup + g->hash_len * lex_index);
718 static struct commit_list **insert_parent_or_die(struct repository *r,
719 struct commit_graph *g,
721 struct commit_list **pptr)
724 struct object_id oid;
726 if (pos >= g->num_commits + g->num_commits_in_base)
727 die("invalid parent position %"PRIu32, pos);
729 load_oid_from_graph(g, pos, &oid);
730 c = lookup_commit(r, &oid);
732 die(_("could not find commit %s"), oid_to_hex(&oid));
733 commit_graph_data_at(c)->graph_pos = pos;
734 return &commit_list_insert(c, pptr)->next;
737 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
739 const unsigned char *commit_data;
740 struct commit_graph_data *graph_data;
743 while (pos < g->num_commits_in_base)
746 lex_index = pos - g->num_commits_in_base;
747 commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * lex_index;
749 graph_data = commit_graph_data_at(item);
750 graph_data->graph_pos = pos;
751 graph_data->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
754 static inline void set_commit_tree(struct commit *c, struct tree *t)
759 static int fill_commit_in_graph(struct repository *r,
761 struct commit_graph *g, uint32_t pos)
764 uint32_t *parent_data_ptr;
765 uint64_t date_low, date_high;
766 struct commit_list **pptr;
767 struct commit_graph_data *graph_data;
768 const unsigned char *commit_data;
771 while (pos < g->num_commits_in_base)
774 if (pos >= g->num_commits + g->num_commits_in_base)
775 die(_("invalid commit position. commit-graph is likely corrupt"));
778 * Store the "full" position, but then use the
779 * "local" position for the rest of the calculation.
781 graph_data = commit_graph_data_at(item);
782 graph_data->graph_pos = pos;
783 lex_index = pos - g->num_commits_in_base;
785 commit_data = g->chunk_commit_data + (g->hash_len + 16) * lex_index;
787 item->object.parsed = 1;
789 set_commit_tree(item, NULL);
791 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
792 date_low = get_be32(commit_data + g->hash_len + 12);
793 item->date = (timestamp_t)((date_high << 32) | date_low);
795 graph_data->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
797 pptr = &item->parents;
799 edge_value = get_be32(commit_data + g->hash_len);
800 if (edge_value == GRAPH_PARENT_NONE)
802 pptr = insert_parent_or_die(r, g, edge_value, pptr);
804 edge_value = get_be32(commit_data + g->hash_len + 4);
805 if (edge_value == GRAPH_PARENT_NONE)
807 if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
808 pptr = insert_parent_or_die(r, g, edge_value, pptr);
812 parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
813 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
815 edge_value = get_be32(parent_data_ptr);
816 pptr = insert_parent_or_die(r, g,
817 edge_value & GRAPH_EDGE_LAST_MASK,
820 } while (!(edge_value & GRAPH_LAST_EDGE));
825 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
827 uint32_t graph_pos = commit_graph_position(item);
828 if (graph_pos != COMMIT_NOT_FROM_GRAPH) {
832 struct commit_graph *cur_g = g;
835 while (cur_g && !bsearch_graph(cur_g, &(item->object.oid), &lex_index))
836 cur_g = cur_g->base_graph;
839 *pos = lex_index + cur_g->num_commits_in_base;
847 static int parse_commit_in_graph_one(struct repository *r,
848 struct commit_graph *g,
853 if (item->object.parsed)
856 if (find_commit_in_graph(item, g, &pos))
857 return fill_commit_in_graph(r, item, g, pos);
862 int parse_commit_in_graph(struct repository *r, struct commit *item)
864 static int checked_env = 0;
867 git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_PARSE, 0))
868 die("dying as requested by the '%s' variable on commit-graph parse!",
869 GIT_TEST_COMMIT_GRAPH_DIE_ON_PARSE);
872 if (!prepare_commit_graph(r))
874 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
877 void load_commit_graph_info(struct repository *r, struct commit *item)
880 if (!prepare_commit_graph(r))
882 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
883 fill_commit_graph_info(item, r->objects->commit_graph, pos);
886 static struct tree *load_tree_for_commit(struct repository *r,
887 struct commit_graph *g,
890 struct object_id oid;
891 const unsigned char *commit_data;
892 uint32_t graph_pos = commit_graph_position(c);
894 while (graph_pos < g->num_commits_in_base)
897 commit_data = g->chunk_commit_data +
898 GRAPH_DATA_WIDTH * (graph_pos - g->num_commits_in_base);
900 hashcpy(oid.hash, commit_data);
901 set_commit_tree(c, lookup_tree(r, &oid));
903 return c->maybe_tree;
906 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
907 struct commit_graph *g,
908 const struct commit *c)
911 return c->maybe_tree;
912 if (commit_graph_position(c) == COMMIT_NOT_FROM_GRAPH)
913 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
915 return load_tree_for_commit(r, g, (struct commit *)c);
918 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
920 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
923 struct packed_commit_list {
924 struct commit **list;
929 struct packed_oid_list {
930 struct object_id *list;
935 struct write_commit_graph_context {
936 struct repository *r;
937 struct object_directory *odb;
939 struct packed_oid_list oids;
940 struct packed_commit_list commits;
942 unsigned long approx_nr_objects;
943 struct progress *progress;
945 uint64_t progress_cnt;
947 char *base_graph_name;
948 int num_commit_graphs_before;
949 int num_commit_graphs_after;
950 char **commit_graph_filenames_before;
951 char **commit_graph_filenames_after;
952 char **commit_graph_hash_after;
953 uint32_t new_num_commits_in_base;
954 struct commit_graph *new_base_graph;
962 const struct split_commit_graph_opts *split_opts;
963 size_t total_bloom_filter_data_size;
964 const struct bloom_filter_settings *bloom_settings;
967 static int write_graph_chunk_fanout(struct hashfile *f,
968 struct write_commit_graph_context *ctx)
971 struct commit **list = ctx->commits.list;
974 * Write the first-level table (the list is sorted,
975 * but we use a 256-entry lookup to be able to avoid
976 * having to do eight extra binary search iterations).
978 for (i = 0; i < 256; i++) {
979 while (count < ctx->commits.nr) {
980 if ((*list)->object.oid.hash[0] != i)
982 display_progress(ctx->progress, ++ctx->progress_cnt);
987 hashwrite_be32(f, count);
993 static int write_graph_chunk_oids(struct hashfile *f,
994 struct write_commit_graph_context *ctx)
996 struct commit **list = ctx->commits.list;
998 for (count = 0; count < ctx->commits.nr; count++, list++) {
999 display_progress(ctx->progress, ++ctx->progress_cnt);
1000 hashwrite(f, (*list)->object.oid.hash, the_hash_algo->rawsz);
1006 static const unsigned char *commit_to_sha1(size_t index, void *table)
1008 struct commit **commits = table;
1009 return commits[index]->object.oid.hash;
1012 static int write_graph_chunk_data(struct hashfile *f,
1013 struct write_commit_graph_context *ctx)
1015 struct commit **list = ctx->commits.list;
1016 struct commit **last = ctx->commits.list + ctx->commits.nr;
1017 uint32_t num_extra_edges = 0;
1019 while (list < last) {
1020 struct commit_list *parent;
1021 struct object_id *tree;
1023 uint32_t packedDate[2];
1024 display_progress(ctx->progress, ++ctx->progress_cnt);
1026 if (parse_commit_no_graph(*list))
1027 die(_("unable to parse commit %s"),
1028 oid_to_hex(&(*list)->object.oid));
1029 tree = get_commit_tree_oid(*list);
1030 hashwrite(f, tree->hash, the_hash_algo->rawsz);
1032 parent = (*list)->parents;
1035 edge_value = GRAPH_PARENT_NONE;
1037 edge_value = sha1_pos(parent->item->object.oid.hash,
1042 if (edge_value >= 0)
1043 edge_value += ctx->new_num_commits_in_base;
1044 else if (ctx->new_base_graph) {
1046 if (find_commit_in_graph(parent->item,
1047 ctx->new_base_graph,
1053 BUG("missing parent %s for commit %s",
1054 oid_to_hex(&parent->item->object.oid),
1055 oid_to_hex(&(*list)->object.oid));
1058 hashwrite_be32(f, edge_value);
1061 parent = parent->next;
1064 edge_value = GRAPH_PARENT_NONE;
1065 else if (parent->next)
1066 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
1068 edge_value = sha1_pos(parent->item->object.oid.hash,
1073 if (edge_value >= 0)
1074 edge_value += ctx->new_num_commits_in_base;
1075 else if (ctx->new_base_graph) {
1077 if (find_commit_in_graph(parent->item,
1078 ctx->new_base_graph,
1084 BUG("missing parent %s for commit %s",
1085 oid_to_hex(&parent->item->object.oid),
1086 oid_to_hex(&(*list)->object.oid));
1089 hashwrite_be32(f, edge_value);
1091 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
1094 parent = parent->next;
1098 if (sizeof((*list)->date) > 4)
1099 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
1103 packedDate[0] |= htonl(commit_graph_data_at(*list)->generation << 2);
1105 packedDate[1] = htonl((*list)->date);
1106 hashwrite(f, packedDate, 8);
1114 static int write_graph_chunk_extra_edges(struct hashfile *f,
1115 struct write_commit_graph_context *ctx)
1117 struct commit **list = ctx->commits.list;
1118 struct commit **last = ctx->commits.list + ctx->commits.nr;
1119 struct commit_list *parent;
1121 while (list < last) {
1122 int num_parents = 0;
1124 display_progress(ctx->progress, ++ctx->progress_cnt);
1126 for (parent = (*list)->parents; num_parents < 3 && parent;
1127 parent = parent->next)
1130 if (num_parents <= 2) {
1135 /* Since num_parents > 2, this initializer is safe. */
1136 for (parent = (*list)->parents->next; parent; parent = parent->next) {
1137 int edge_value = sha1_pos(parent->item->object.oid.hash,
1142 if (edge_value >= 0)
1143 edge_value += ctx->new_num_commits_in_base;
1144 else if (ctx->new_base_graph) {
1146 if (find_commit_in_graph(parent->item,
1147 ctx->new_base_graph,
1153 BUG("missing parent %s for commit %s",
1154 oid_to_hex(&parent->item->object.oid),
1155 oid_to_hex(&(*list)->object.oid));
1156 else if (!parent->next)
1157 edge_value |= GRAPH_LAST_EDGE;
1159 hashwrite_be32(f, edge_value);
1168 static int write_graph_chunk_bloom_indexes(struct hashfile *f,
1169 struct write_commit_graph_context *ctx)
1171 struct commit **list = ctx->commits.list;
1172 struct commit **last = ctx->commits.list + ctx->commits.nr;
1173 uint32_t cur_pos = 0;
1175 while (list < last) {
1176 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list, 0);
1177 size_t len = filter ? filter->len : 0;
1179 display_progress(ctx->progress, ++ctx->progress_cnt);
1180 hashwrite_be32(f, cur_pos);
1187 static void trace2_bloom_filter_settings(struct write_commit_graph_context *ctx)
1189 struct json_writer jw = JSON_WRITER_INIT;
1191 jw_object_begin(&jw, 0);
1192 jw_object_intmax(&jw, "hash_version", ctx->bloom_settings->hash_version);
1193 jw_object_intmax(&jw, "num_hashes", ctx->bloom_settings->num_hashes);
1194 jw_object_intmax(&jw, "bits_per_entry", ctx->bloom_settings->bits_per_entry);
1197 trace2_data_json("bloom", ctx->r, "settings", &jw);
1202 static int write_graph_chunk_bloom_data(struct hashfile *f,
1203 struct write_commit_graph_context *ctx)
1205 struct commit **list = ctx->commits.list;
1206 struct commit **last = ctx->commits.list + ctx->commits.nr;
1208 trace2_bloom_filter_settings(ctx);
1210 hashwrite_be32(f, ctx->bloom_settings->hash_version);
1211 hashwrite_be32(f, ctx->bloom_settings->num_hashes);
1212 hashwrite_be32(f, ctx->bloom_settings->bits_per_entry);
1214 while (list < last) {
1215 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list, 0);
1216 size_t len = filter ? filter->len : 0;
1218 display_progress(ctx->progress, ++ctx->progress_cnt);
1220 hashwrite(f, filter->data, len * sizeof(unsigned char));
1227 static int oid_compare(const void *_a, const void *_b)
1229 const struct object_id *a = (const struct object_id *)_a;
1230 const struct object_id *b = (const struct object_id *)_b;
1231 return oidcmp(a, b);
1234 static int add_packed_commits(const struct object_id *oid,
1235 struct packed_git *pack,
1239 struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
1240 enum object_type type;
1241 off_t offset = nth_packed_object_offset(pack, pos);
1242 struct object_info oi = OBJECT_INFO_INIT;
1245 display_progress(ctx->progress, ++ctx->progress_done);
1248 if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
1249 die(_("unable to get type of object %s"), oid_to_hex(oid));
1251 if (type != OBJ_COMMIT)
1254 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1255 oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
1258 set_commit_pos(ctx->r, oid);
1263 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
1265 struct commit_list *parent;
1266 for (parent = commit->parents; parent; parent = parent->next) {
1267 if (!(parent->item->object.flags & REACHABLE)) {
1268 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1269 oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
1271 parent->item->object.flags |= REACHABLE;
1276 static void close_reachable(struct write_commit_graph_context *ctx)
1279 struct commit *commit;
1280 enum commit_graph_split_flags flags = ctx->split_opts ?
1281 ctx->split_opts->flags : COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1283 if (ctx->report_progress)
1284 ctx->progress = start_delayed_progress(
1285 _("Loading known commits in commit graph"),
1287 for (i = 0; i < ctx->oids.nr; i++) {
1288 display_progress(ctx->progress, i + 1);
1289 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1291 commit->object.flags |= REACHABLE;
1293 stop_progress(&ctx->progress);
1296 * As this loop runs, ctx->oids.nr may grow, but not more
1297 * than the number of missing commits in the reachable
1300 if (ctx->report_progress)
1301 ctx->progress = start_delayed_progress(
1302 _("Expanding reachable commits in commit graph"),
1304 for (i = 0; i < ctx->oids.nr; i++) {
1305 display_progress(ctx->progress, i + 1);
1306 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1311 if ((!parse_commit(commit) &&
1312 commit_graph_position(commit) == COMMIT_NOT_FROM_GRAPH) ||
1313 flags == COMMIT_GRAPH_SPLIT_REPLACE)
1314 add_missing_parents(ctx, commit);
1315 } else if (!parse_commit_no_graph(commit))
1316 add_missing_parents(ctx, commit);
1318 stop_progress(&ctx->progress);
1320 if (ctx->report_progress)
1321 ctx->progress = start_delayed_progress(
1322 _("Clearing commit marks in commit graph"),
1324 for (i = 0; i < ctx->oids.nr; i++) {
1325 display_progress(ctx->progress, i + 1);
1326 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1329 commit->object.flags &= ~REACHABLE;
1331 stop_progress(&ctx->progress);
1334 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1337 struct commit_list *list = NULL;
1339 if (ctx->report_progress)
1340 ctx->progress = start_delayed_progress(
1341 _("Computing commit graph generation numbers"),
1343 for (i = 0; i < ctx->commits.nr; i++) {
1344 uint32_t generation = commit_graph_data_at(ctx->commits.list[i])->generation;
1346 display_progress(ctx->progress, i + 1);
1347 if (generation != GENERATION_NUMBER_INFINITY &&
1348 generation != GENERATION_NUMBER_ZERO)
1351 commit_list_insert(ctx->commits.list[i], &list);
1353 struct commit *current = list->item;
1354 struct commit_list *parent;
1355 int all_parents_computed = 1;
1356 uint32_t max_generation = 0;
1358 for (parent = current->parents; parent; parent = parent->next) {
1359 generation = commit_graph_data_at(parent->item)->generation;
1361 if (generation == GENERATION_NUMBER_INFINITY ||
1362 generation == GENERATION_NUMBER_ZERO) {
1363 all_parents_computed = 0;
1364 commit_list_insert(parent->item, &list);
1366 } else if (generation > max_generation) {
1367 max_generation = generation;
1371 if (all_parents_computed) {
1372 struct commit_graph_data *data = commit_graph_data_at(current);
1374 data->generation = max_generation + 1;
1377 if (data->generation > GENERATION_NUMBER_MAX)
1378 data->generation = GENERATION_NUMBER_MAX;
1382 stop_progress(&ctx->progress);
1385 static void compute_bloom_filters(struct write_commit_graph_context *ctx)
1388 struct progress *progress = NULL;
1389 struct commit **sorted_commits;
1391 init_bloom_filters();
1393 if (ctx->report_progress)
1394 progress = start_delayed_progress(
1395 _("Computing commit changed paths Bloom filters"),
1398 ALLOC_ARRAY(sorted_commits, ctx->commits.nr);
1399 COPY_ARRAY(sorted_commits, ctx->commits.list, ctx->commits.nr);
1401 if (ctx->order_by_pack)
1402 QSORT(sorted_commits, ctx->commits.nr, commit_pos_cmp);
1404 QSORT(sorted_commits, ctx->commits.nr, commit_gen_cmp);
1406 for (i = 0; i < ctx->commits.nr; i++) {
1407 struct commit *c = sorted_commits[i];
1408 struct bloom_filter *filter = get_bloom_filter(ctx->r, c, 1);
1409 ctx->total_bloom_filter_data_size += sizeof(unsigned char) * filter->len;
1410 display_progress(progress, i + 1);
1413 free(sorted_commits);
1414 stop_progress(&progress);
1417 struct refs_cb_data {
1418 struct oidset *commits;
1419 struct progress *progress;
1422 static int add_ref_to_set(const char *refname,
1423 const struct object_id *oid,
1424 int flags, void *cb_data)
1426 struct object_id peeled;
1427 struct refs_cb_data *data = (struct refs_cb_data *)cb_data;
1429 if (!peel_ref(refname, &peeled))
1431 if (oid_object_info(the_repository, oid, NULL) == OBJ_COMMIT)
1432 oidset_insert(data->commits, oid);
1434 display_progress(data->progress, oidset_size(data->commits));
1439 int write_commit_graph_reachable(struct object_directory *odb,
1440 enum commit_graph_write_flags flags,
1441 const struct split_commit_graph_opts *split_opts)
1443 struct oidset commits = OIDSET_INIT;
1444 struct refs_cb_data data;
1447 memset(&data, 0, sizeof(data));
1448 data.commits = &commits;
1449 if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
1450 data.progress = start_delayed_progress(
1451 _("Collecting referenced commits"), 0);
1453 for_each_ref(add_ref_to_set, &data);
1455 stop_progress(&data.progress);
1457 result = write_commit_graph(odb, NULL, &commits,
1460 oidset_clear(&commits);
1464 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
1465 struct string_list *pack_indexes)
1468 struct strbuf progress_title = STRBUF_INIT;
1469 struct strbuf packname = STRBUF_INIT;
1472 strbuf_addf(&packname, "%s/pack/", ctx->odb->path);
1473 dirlen = packname.len;
1474 if (ctx->report_progress) {
1475 strbuf_addf(&progress_title,
1476 Q_("Finding commits for commit graph in %d pack",
1477 "Finding commits for commit graph in %d packs",
1480 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1481 ctx->progress_done = 0;
1483 for (i = 0; i < pack_indexes->nr; i++) {
1484 struct packed_git *p;
1485 strbuf_setlen(&packname, dirlen);
1486 strbuf_addstr(&packname, pack_indexes->items[i].string);
1487 p = add_packed_git(packname.buf, packname.len, 1);
1489 error(_("error adding pack %s"), packname.buf);
1492 if (open_pack_index(p)) {
1493 error(_("error opening index for %s"), packname.buf);
1496 for_each_object_in_pack(p, add_packed_commits, ctx,
1497 FOR_EACH_OBJECT_PACK_ORDER);
1502 stop_progress(&ctx->progress);
1503 strbuf_release(&progress_title);
1504 strbuf_release(&packname);
1509 static int fill_oids_from_commits(struct write_commit_graph_context *ctx,
1510 struct oidset *commits)
1512 struct oidset_iter iter;
1513 struct object_id *oid;
1515 if (!oidset_size(commits))
1518 oidset_iter_init(commits, &iter);
1519 while ((oid = oidset_iter_next(&iter))) {
1520 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1521 oidcpy(&ctx->oids.list[ctx->oids.nr], oid);
1528 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1530 if (ctx->report_progress)
1531 ctx->progress = start_delayed_progress(
1532 _("Finding commits for commit graph among packed objects"),
1533 ctx->approx_nr_objects);
1534 for_each_packed_object(add_packed_commits, ctx,
1535 FOR_EACH_OBJECT_PACK_ORDER);
1536 if (ctx->progress_done < ctx->approx_nr_objects)
1537 display_progress(ctx->progress, ctx->approx_nr_objects);
1538 stop_progress(&ctx->progress);
1541 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
1543 uint32_t i, count_distinct = 1;
1545 if (ctx->report_progress)
1546 ctx->progress = start_delayed_progress(
1547 _("Counting distinct commits in commit graph"),
1549 display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1550 QSORT(ctx->oids.list, ctx->oids.nr, oid_compare);
1552 for (i = 1; i < ctx->oids.nr; i++) {
1553 display_progress(ctx->progress, i + 1);
1554 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i])) {
1556 struct commit *c = lookup_commit(ctx->r, &ctx->oids.list[i]);
1558 if (!c || commit_graph_position(c) != COMMIT_NOT_FROM_GRAPH)
1565 stop_progress(&ctx->progress);
1567 return count_distinct;
1570 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1573 enum commit_graph_split_flags flags = ctx->split_opts ?
1574 ctx->split_opts->flags : COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1576 ctx->num_extra_edges = 0;
1577 if (ctx->report_progress)
1578 ctx->progress = start_delayed_progress(
1579 _("Finding extra edges in commit graph"),
1581 for (i = 0; i < ctx->oids.nr; i++) {
1582 unsigned int num_parents;
1584 display_progress(ctx->progress, i + 1);
1585 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1588 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + 1, ctx->commits.alloc);
1589 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1591 if (ctx->split && flags != COMMIT_GRAPH_SPLIT_REPLACE &&
1592 commit_graph_position(ctx->commits.list[ctx->commits.nr]) != COMMIT_NOT_FROM_GRAPH)
1595 if (ctx->split && flags == COMMIT_GRAPH_SPLIT_REPLACE)
1596 parse_commit(ctx->commits.list[ctx->commits.nr]);
1598 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1600 num_parents = commit_list_count(ctx->commits.list[ctx->commits.nr]->parents);
1601 if (num_parents > 2)
1602 ctx->num_extra_edges += num_parents - 1;
1606 stop_progress(&ctx->progress);
1609 static int write_graph_chunk_base_1(struct hashfile *f,
1610 struct commit_graph *g)
1617 num = write_graph_chunk_base_1(f, g->base_graph);
1618 hashwrite(f, g->oid.hash, the_hash_algo->rawsz);
1622 static int write_graph_chunk_base(struct hashfile *f,
1623 struct write_commit_graph_context *ctx)
1625 int num = write_graph_chunk_base_1(f, ctx->new_base_graph);
1627 if (num != ctx->num_commit_graphs_after - 1) {
1628 error(_("failed to write correct number of base graph ids"));
1635 typedef int (*chunk_write_fn)(struct hashfile *f,
1636 struct write_commit_graph_context *ctx);
1641 chunk_write_fn write_fn;
1644 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1649 struct lock_file lk = LOCK_INIT;
1650 struct chunk_info chunks[MAX_NUM_CHUNKS + 1];
1651 const unsigned hashsz = the_hash_algo->rawsz;
1652 struct strbuf progress_title = STRBUF_INIT;
1654 uint64_t chunk_offset;
1655 struct object_id file_hash;
1656 struct bloom_filter_settings bloom_settings = DEFAULT_BLOOM_FILTER_SETTINGS;
1658 if (!ctx->bloom_settings) {
1659 bloom_settings.bits_per_entry = git_env_ulong("GIT_TEST_BLOOM_SETTINGS_BITS_PER_ENTRY",
1660 bloom_settings.bits_per_entry);
1661 bloom_settings.num_hashes = git_env_ulong("GIT_TEST_BLOOM_SETTINGS_NUM_HASHES",
1662 bloom_settings.num_hashes);
1663 ctx->bloom_settings = &bloom_settings;
1667 struct strbuf tmp_file = STRBUF_INIT;
1669 strbuf_addf(&tmp_file,
1670 "%s/info/commit-graphs/tmp_graph_XXXXXX",
1672 ctx->graph_name = strbuf_detach(&tmp_file, NULL);
1674 ctx->graph_name = get_commit_graph_filename(ctx->odb);
1677 if (safe_create_leading_directories(ctx->graph_name)) {
1678 UNLEAK(ctx->graph_name);
1679 error(_("unable to create leading directories of %s"),
1685 char *lock_name = get_chain_filename(ctx->odb);
1687 hold_lock_file_for_update_mode(&lk, lock_name,
1688 LOCK_DIE_ON_ERROR, 0444);
1690 fd = git_mkstemp_mode(ctx->graph_name, 0444);
1692 error(_("unable to create temporary graph layer"));
1696 if (adjust_shared_perm(ctx->graph_name)) {
1697 error(_("unable to adjust shared permissions for '%s'"),
1702 f = hashfd(fd, ctx->graph_name);
1704 hold_lock_file_for_update_mode(&lk, ctx->graph_name,
1705 LOCK_DIE_ON_ERROR, 0444);
1706 fd = lk.tempfile->fd;
1707 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1710 chunks[0].id = GRAPH_CHUNKID_OIDFANOUT;
1711 chunks[0].size = GRAPH_FANOUT_SIZE;
1712 chunks[0].write_fn = write_graph_chunk_fanout;
1713 chunks[1].id = GRAPH_CHUNKID_OIDLOOKUP;
1714 chunks[1].size = hashsz * ctx->commits.nr;
1715 chunks[1].write_fn = write_graph_chunk_oids;
1716 chunks[2].id = GRAPH_CHUNKID_DATA;
1717 chunks[2].size = (hashsz + 16) * ctx->commits.nr;
1718 chunks[2].write_fn = write_graph_chunk_data;
1719 if (ctx->num_extra_edges) {
1720 chunks[num_chunks].id = GRAPH_CHUNKID_EXTRAEDGES;
1721 chunks[num_chunks].size = 4 * ctx->num_extra_edges;
1722 chunks[num_chunks].write_fn = write_graph_chunk_extra_edges;
1725 if (ctx->changed_paths) {
1726 chunks[num_chunks].id = GRAPH_CHUNKID_BLOOMINDEXES;
1727 chunks[num_chunks].size = sizeof(uint32_t) * ctx->commits.nr;
1728 chunks[num_chunks].write_fn = write_graph_chunk_bloom_indexes;
1730 chunks[num_chunks].id = GRAPH_CHUNKID_BLOOMDATA;
1731 chunks[num_chunks].size = sizeof(uint32_t) * 3
1732 + ctx->total_bloom_filter_data_size;
1733 chunks[num_chunks].write_fn = write_graph_chunk_bloom_data;
1736 if (ctx->num_commit_graphs_after > 1) {
1737 chunks[num_chunks].id = GRAPH_CHUNKID_BASE;
1738 chunks[num_chunks].size = hashsz * (ctx->num_commit_graphs_after - 1);
1739 chunks[num_chunks].write_fn = write_graph_chunk_base;
1743 chunks[num_chunks].id = 0;
1744 chunks[num_chunks].size = 0;
1746 hashwrite_be32(f, GRAPH_SIGNATURE);
1748 hashwrite_u8(f, GRAPH_VERSION);
1749 hashwrite_u8(f, oid_version());
1750 hashwrite_u8(f, num_chunks);
1751 hashwrite_u8(f, ctx->num_commit_graphs_after - 1);
1753 chunk_offset = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1754 for (i = 0; i <= num_chunks; i++) {
1755 uint32_t chunk_write[3];
1757 chunk_write[0] = htonl(chunks[i].id);
1758 chunk_write[1] = htonl(chunk_offset >> 32);
1759 chunk_write[2] = htonl(chunk_offset & 0xffffffff);
1760 hashwrite(f, chunk_write, 12);
1762 chunk_offset += chunks[i].size;
1765 if (ctx->report_progress) {
1766 strbuf_addf(&progress_title,
1767 Q_("Writing out commit graph in %d pass",
1768 "Writing out commit graph in %d passes",
1771 ctx->progress = start_delayed_progress(
1773 num_chunks * ctx->commits.nr);
1776 for (i = 0; i < num_chunks; i++) {
1777 uint64_t start_offset = f->total + f->offset;
1779 if (chunks[i].write_fn(f, ctx))
1782 if (f->total + f->offset != start_offset + chunks[i].size)
1783 BUG("expected to write %"PRId64" bytes to chunk %"PRIx32", but wrote %"PRId64" instead",
1784 chunks[i].size, chunks[i].id,
1785 f->total + f->offset - start_offset);
1788 stop_progress(&ctx->progress);
1789 strbuf_release(&progress_title);
1791 if (ctx->split && ctx->base_graph_name && ctx->num_commit_graphs_after > 1) {
1792 char *new_base_hash = xstrdup(oid_to_hex(&ctx->new_base_graph->oid));
1793 char *new_base_name = get_split_graph_filename(ctx->new_base_graph->odb, new_base_hash);
1795 free(ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2]);
1796 free(ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2]);
1797 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2] = new_base_name;
1798 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2] = new_base_hash;
1801 close_commit_graph(ctx->r->objects);
1802 finalize_hashfile(f, file_hash.hash, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1805 FILE *chainf = fdopen_lock_file(&lk, "w");
1806 char *final_graph_name;
1812 error(_("unable to open commit-graph chain file"));
1816 if (ctx->base_graph_name) {
1818 int idx = ctx->num_commit_graphs_after - 1;
1819 if (ctx->num_commit_graphs_after > 1)
1822 dest = ctx->commit_graph_filenames_after[idx];
1824 if (strcmp(ctx->base_graph_name, dest)) {
1825 result = rename(ctx->base_graph_name, dest);
1828 error(_("failed to rename base commit-graph file"));
1833 char *graph_name = get_commit_graph_filename(ctx->odb);
1837 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1] = xstrdup(oid_to_hex(&file_hash));
1838 final_graph_name = get_split_graph_filename(ctx->odb,
1839 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1]);
1840 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 1] = final_graph_name;
1842 result = rename(ctx->graph_name, final_graph_name);
1844 for (i = 0; i < ctx->num_commit_graphs_after; i++)
1845 fprintf(lk.tempfile->fp, "%s\n", ctx->commit_graph_hash_after[i]);
1848 error(_("failed to rename temporary commit-graph file"));
1853 commit_lock_file(&lk);
1858 static void split_graph_merge_strategy(struct write_commit_graph_context *ctx)
1860 struct commit_graph *g;
1861 uint32_t num_commits;
1862 enum commit_graph_split_flags flags = COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1865 int max_commits = 0;
1868 if (ctx->split_opts) {
1869 max_commits = ctx->split_opts->max_commits;
1871 if (ctx->split_opts->size_multiple)
1872 size_mult = ctx->split_opts->size_multiple;
1874 flags = ctx->split_opts->flags;
1877 g = ctx->r->objects->commit_graph;
1878 num_commits = ctx->commits.nr;
1879 if (flags == COMMIT_GRAPH_SPLIT_REPLACE)
1880 ctx->num_commit_graphs_after = 1;
1882 ctx->num_commit_graphs_after = ctx->num_commit_graphs_before + 1;
1884 if (flags != COMMIT_GRAPH_SPLIT_MERGE_PROHIBITED &&
1885 flags != COMMIT_GRAPH_SPLIT_REPLACE) {
1886 while (g && (g->num_commits <= size_mult * num_commits ||
1887 (max_commits && num_commits > max_commits))) {
1888 if (g->odb != ctx->odb)
1891 num_commits += g->num_commits;
1894 ctx->num_commit_graphs_after--;
1898 if (flags != COMMIT_GRAPH_SPLIT_REPLACE)
1899 ctx->new_base_graph = g;
1900 else if (ctx->num_commit_graphs_after != 1)
1901 BUG("split_graph_merge_strategy: num_commit_graphs_after "
1902 "should be 1 with --split=replace");
1904 if (ctx->num_commit_graphs_after == 2) {
1905 char *old_graph_name = get_commit_graph_filename(g->odb);
1907 if (!strcmp(g->filename, old_graph_name) &&
1908 g->odb != ctx->odb) {
1909 ctx->num_commit_graphs_after = 1;
1910 ctx->new_base_graph = NULL;
1913 free(old_graph_name);
1916 CALLOC_ARRAY(ctx->commit_graph_filenames_after, ctx->num_commit_graphs_after);
1917 CALLOC_ARRAY(ctx->commit_graph_hash_after, ctx->num_commit_graphs_after);
1919 for (i = 0; i < ctx->num_commit_graphs_after &&
1920 i < ctx->num_commit_graphs_before; i++)
1921 ctx->commit_graph_filenames_after[i] = xstrdup(ctx->commit_graph_filenames_before[i]);
1923 i = ctx->num_commit_graphs_before - 1;
1924 g = ctx->r->objects->commit_graph;
1927 if (i < ctx->num_commit_graphs_after)
1928 ctx->commit_graph_hash_after[i] = xstrdup(oid_to_hex(&g->oid));
1935 static void merge_commit_graph(struct write_commit_graph_context *ctx,
1936 struct commit_graph *g)
1939 uint32_t offset = g->num_commits_in_base;
1941 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + g->num_commits, ctx->commits.alloc);
1943 for (i = 0; i < g->num_commits; i++) {
1944 struct object_id oid;
1945 struct commit *result;
1947 display_progress(ctx->progress, i + 1);
1949 load_oid_from_graph(g, i + offset, &oid);
1951 /* only add commits if they still exist in the repo */
1952 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1955 ctx->commits.list[ctx->commits.nr] = result;
1961 static int commit_compare(const void *_a, const void *_b)
1963 const struct commit *a = *(const struct commit **)_a;
1964 const struct commit *b = *(const struct commit **)_b;
1965 return oidcmp(&a->object.oid, &b->object.oid);
1968 static void sort_and_scan_merged_commits(struct write_commit_graph_context *ctx)
1972 if (ctx->report_progress)
1973 ctx->progress = start_delayed_progress(
1974 _("Scanning merged commits"),
1977 QSORT(ctx->commits.list, ctx->commits.nr, commit_compare);
1979 ctx->num_extra_edges = 0;
1980 for (i = 0; i < ctx->commits.nr; i++) {
1981 display_progress(ctx->progress, i);
1983 if (i && oideq(&ctx->commits.list[i - 1]->object.oid,
1984 &ctx->commits.list[i]->object.oid)) {
1985 die(_("unexpected duplicate commit id %s"),
1986 oid_to_hex(&ctx->commits.list[i]->object.oid));
1988 unsigned int num_parents;
1990 num_parents = commit_list_count(ctx->commits.list[i]->parents);
1991 if (num_parents > 2)
1992 ctx->num_extra_edges += num_parents - 1;
1996 stop_progress(&ctx->progress);
1999 static void merge_commit_graphs(struct write_commit_graph_context *ctx)
2001 struct commit_graph *g = ctx->r->objects->commit_graph;
2002 uint32_t current_graph_number = ctx->num_commit_graphs_before;
2004 while (g && current_graph_number >= ctx->num_commit_graphs_after) {
2005 current_graph_number--;
2007 if (ctx->report_progress)
2008 ctx->progress = start_delayed_progress(_("Merging commit-graph"), 0);
2010 merge_commit_graph(ctx, g);
2011 stop_progress(&ctx->progress);
2017 ctx->new_base_graph = g;
2018 ctx->new_num_commits_in_base = g->num_commits + g->num_commits_in_base;
2021 if (ctx->new_base_graph)
2022 ctx->base_graph_name = xstrdup(ctx->new_base_graph->filename);
2024 sort_and_scan_merged_commits(ctx);
2027 static void mark_commit_graphs(struct write_commit_graph_context *ctx)
2030 time_t now = time(NULL);
2032 for (i = ctx->num_commit_graphs_after - 1; i < ctx->num_commit_graphs_before; i++) {
2034 struct utimbuf updated_time;
2036 stat(ctx->commit_graph_filenames_before[i], &st);
2038 updated_time.actime = st.st_atime;
2039 updated_time.modtime = now;
2040 utime(ctx->commit_graph_filenames_before[i], &updated_time);
2044 static void expire_commit_graphs(struct write_commit_graph_context *ctx)
2046 struct strbuf path = STRBUF_INIT;
2050 timestamp_t expire_time = time(NULL);
2052 if (ctx->split_opts && ctx->split_opts->expire_time)
2053 expire_time = ctx->split_opts->expire_time;
2055 char *chain_file_name = get_chain_filename(ctx->odb);
2056 unlink(chain_file_name);
2057 free(chain_file_name);
2058 ctx->num_commit_graphs_after = 0;
2061 strbuf_addstr(&path, ctx->odb->path);
2062 strbuf_addstr(&path, "/info/commit-graphs");
2063 dir = opendir(path.buf);
2068 strbuf_addch(&path, '/');
2069 dirnamelen = path.len;
2070 while ((de = readdir(dir)) != NULL) {
2072 uint32_t i, found = 0;
2074 strbuf_setlen(&path, dirnamelen);
2075 strbuf_addstr(&path, de->d_name);
2077 stat(path.buf, &st);
2079 if (st.st_mtime > expire_time)
2081 if (path.len < 6 || strcmp(path.buf + path.len - 6, ".graph"))
2084 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
2085 if (!strcmp(ctx->commit_graph_filenames_after[i],
2097 strbuf_release(&path);
2100 int write_commit_graph(struct object_directory *odb,
2101 struct string_list *pack_indexes,
2102 struct oidset *commits,
2103 enum commit_graph_write_flags flags,
2104 const struct split_commit_graph_opts *split_opts)
2106 struct write_commit_graph_context *ctx;
2107 uint32_t i, count_distinct = 0;
2111 if (!commit_graph_compatible(the_repository))
2114 ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
2115 ctx->r = the_repository;
2117 ctx->append = flags & COMMIT_GRAPH_WRITE_APPEND ? 1 : 0;
2118 ctx->report_progress = flags & COMMIT_GRAPH_WRITE_PROGRESS ? 1 : 0;
2119 ctx->split = flags & COMMIT_GRAPH_WRITE_SPLIT ? 1 : 0;
2120 ctx->split_opts = split_opts;
2121 ctx->total_bloom_filter_data_size = 0;
2123 if (flags & COMMIT_GRAPH_WRITE_BLOOM_FILTERS)
2124 ctx->changed_paths = 1;
2125 if (!(flags & COMMIT_GRAPH_NO_WRITE_BLOOM_FILTERS)) {
2126 struct commit_graph *g;
2127 prepare_commit_graph_one(ctx->r, ctx->odb);
2129 g = ctx->r->objects->commit_graph;
2131 /* We have changed-paths already. Keep them in the next graph */
2132 if (g && g->chunk_bloom_data) {
2133 ctx->changed_paths = 1;
2134 ctx->bloom_settings = g->bloom_filter_settings;
2139 struct commit_graph *g;
2140 prepare_commit_graph(ctx->r);
2142 g = ctx->r->objects->commit_graph;
2145 ctx->num_commit_graphs_before++;
2149 if (ctx->num_commit_graphs_before) {
2150 ALLOC_ARRAY(ctx->commit_graph_filenames_before, ctx->num_commit_graphs_before);
2151 i = ctx->num_commit_graphs_before;
2152 g = ctx->r->objects->commit_graph;
2155 ctx->commit_graph_filenames_before[--i] = xstrdup(g->filename);
2160 if (ctx->split_opts)
2161 replace = ctx->split_opts->flags & COMMIT_GRAPH_SPLIT_REPLACE;
2164 ctx->approx_nr_objects = approximate_object_count();
2165 ctx->oids.alloc = ctx->approx_nr_objects / 32;
2167 if (ctx->split && split_opts && ctx->oids.alloc > split_opts->max_commits)
2168 ctx->oids.alloc = split_opts->max_commits;
2171 prepare_commit_graph_one(ctx->r, ctx->odb);
2172 if (ctx->r->objects->commit_graph)
2173 ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
2176 if (ctx->oids.alloc < 1024)
2177 ctx->oids.alloc = 1024;
2178 ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
2180 if (ctx->append && ctx->r->objects->commit_graph) {
2181 struct commit_graph *g = ctx->r->objects->commit_graph;
2182 for (i = 0; i < g->num_commits; i++) {
2183 const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
2184 hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
2189 ctx->order_by_pack = 1;
2190 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
2195 if ((res = fill_oids_from_commits(ctx, commits)))
2199 if (!pack_indexes && !commits) {
2200 ctx->order_by_pack = 1;
2201 fill_oids_from_all_packs(ctx);
2204 close_reachable(ctx);
2206 count_distinct = count_distinct_commits(ctx);
2208 if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
2209 error(_("the commit graph format cannot write %d commits"), count_distinct);
2214 ctx->commits.alloc = count_distinct;
2215 ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
2217 copy_oids_to_commits(ctx);
2219 if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
2220 error(_("too many commits to write graph"));
2225 if (!ctx->commits.nr && !replace)
2229 split_graph_merge_strategy(ctx);
2232 merge_commit_graphs(ctx);
2234 ctx->num_commit_graphs_after = 1;
2236 compute_generation_numbers(ctx);
2238 if (ctx->changed_paths)
2239 compute_bloom_filters(ctx);
2241 res = write_commit_graph_file(ctx);
2244 mark_commit_graphs(ctx);
2246 expire_commit_graphs(ctx);
2249 free(ctx->graph_name);
2250 free(ctx->commits.list);
2251 free(ctx->oids.list);
2253 if (ctx->commit_graph_filenames_after) {
2254 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
2255 free(ctx->commit_graph_filenames_after[i]);
2256 free(ctx->commit_graph_hash_after[i]);
2259 for (i = 0; i < ctx->num_commit_graphs_before; i++)
2260 free(ctx->commit_graph_filenames_before[i]);
2262 free(ctx->commit_graph_filenames_after);
2263 free(ctx->commit_graph_filenames_before);
2264 free(ctx->commit_graph_hash_after);
2272 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
2273 static int verify_commit_graph_error;
2275 static void graph_report(const char *fmt, ...)
2279 verify_commit_graph_error = 1;
2281 vfprintf(stderr, fmt, ap);
2282 fprintf(stderr, "\n");
2286 #define GENERATION_ZERO_EXISTS 1
2287 #define GENERATION_NUMBER_EXISTS 2
2289 int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
2291 uint32_t i, cur_fanout_pos = 0;
2292 struct object_id prev_oid, cur_oid, checksum;
2293 int generation_zero = 0;
2296 struct progress *progress = NULL;
2297 int local_error = 0;
2300 graph_report("no commit-graph file loaded");
2304 verify_commit_graph_error = verify_commit_graph_lite(g);
2305 if (verify_commit_graph_error)
2306 return verify_commit_graph_error;
2308 devnull = open("/dev/null", O_WRONLY);
2309 f = hashfd(devnull, NULL);
2310 hashwrite(f, g->data, g->data_len - g->hash_len);
2311 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
2312 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
2313 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
2314 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
2317 for (i = 0; i < g->num_commits; i++) {
2318 struct commit *graph_commit;
2320 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2322 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
2323 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
2324 oid_to_hex(&prev_oid),
2325 oid_to_hex(&cur_oid));
2327 oidcpy(&prev_oid, &cur_oid);
2329 while (cur_oid.hash[0] > cur_fanout_pos) {
2330 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2332 if (i != fanout_value)
2333 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2334 cur_fanout_pos, fanout_value, i);
2338 graph_commit = lookup_commit(r, &cur_oid);
2339 if (!parse_commit_in_graph_one(r, g, graph_commit))
2340 graph_report(_("failed to parse commit %s from commit-graph"),
2341 oid_to_hex(&cur_oid));
2344 while (cur_fanout_pos < 256) {
2345 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2347 if (g->num_commits != fanout_value)
2348 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2349 cur_fanout_pos, fanout_value, i);
2354 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
2355 return verify_commit_graph_error;
2357 if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
2358 progress = start_progress(_("Verifying commits in commit graph"),
2361 for (i = 0; i < g->num_commits; i++) {
2362 struct commit *graph_commit, *odb_commit;
2363 struct commit_list *graph_parents, *odb_parents;
2364 uint32_t max_generation = 0;
2365 uint32_t generation;
2367 display_progress(progress, i + 1);
2368 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2370 graph_commit = lookup_commit(r, &cur_oid);
2371 odb_commit = (struct commit *)create_object(r, &cur_oid, alloc_commit_node(r));
2372 if (parse_commit_internal(odb_commit, 0, 0)) {
2373 graph_report(_("failed to parse commit %s from object database for commit-graph"),
2374 oid_to_hex(&cur_oid));
2378 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
2379 get_commit_tree_oid(odb_commit)))
2380 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2381 oid_to_hex(&cur_oid),
2382 oid_to_hex(get_commit_tree_oid(graph_commit)),
2383 oid_to_hex(get_commit_tree_oid(odb_commit)));
2385 graph_parents = graph_commit->parents;
2386 odb_parents = odb_commit->parents;
2388 while (graph_parents) {
2389 if (odb_parents == NULL) {
2390 graph_report(_("commit-graph parent list for commit %s is too long"),
2391 oid_to_hex(&cur_oid));
2395 /* parse parent in case it is in a base graph */
2396 parse_commit_in_graph_one(r, g, graph_parents->item);
2398 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
2399 graph_report(_("commit-graph parent for %s is %s != %s"),
2400 oid_to_hex(&cur_oid),
2401 oid_to_hex(&graph_parents->item->object.oid),
2402 oid_to_hex(&odb_parents->item->object.oid));
2404 generation = commit_graph_generation(graph_parents->item);
2405 if (generation > max_generation)
2406 max_generation = generation;
2408 graph_parents = graph_parents->next;
2409 odb_parents = odb_parents->next;
2412 if (odb_parents != NULL)
2413 graph_report(_("commit-graph parent list for commit %s terminates early"),
2414 oid_to_hex(&cur_oid));
2416 if (!commit_graph_generation(graph_commit)) {
2417 if (generation_zero == GENERATION_NUMBER_EXISTS)
2418 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2419 oid_to_hex(&cur_oid));
2420 generation_zero = GENERATION_ZERO_EXISTS;
2421 } else if (generation_zero == GENERATION_ZERO_EXISTS)
2422 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2423 oid_to_hex(&cur_oid));
2425 if (generation_zero == GENERATION_ZERO_EXISTS)
2429 * If one of our parents has generation GENERATION_NUMBER_MAX, then
2430 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
2431 * extra logic in the following condition.
2433 if (max_generation == GENERATION_NUMBER_MAX)
2436 generation = commit_graph_generation(graph_commit);
2437 if (generation != max_generation + 1)
2438 graph_report(_("commit-graph generation for commit %s is %u != %u"),
2439 oid_to_hex(&cur_oid),
2441 max_generation + 1);
2443 if (graph_commit->date != odb_commit->date)
2444 graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
2445 oid_to_hex(&cur_oid),
2449 stop_progress(&progress);
2451 local_error = verify_commit_graph_error;
2453 if (!(flags & COMMIT_GRAPH_VERIFY_SHALLOW) && g->base_graph)
2454 local_error |= verify_commit_graph(r, g->base_graph, flags);
2459 void free_commit_graph(struct commit_graph *g)
2464 munmap((void *)g->data, g->data_len);
2468 free(g->bloom_filter_settings);
2472 void disable_commit_graph(struct repository *r)
2474 r->commit_graph_disabled = 1;