l10n: bg.po: Updated Bulgarian translation (5490t)
[git.git] / commit-graph.c
blobc11b59f28b3164cee21ab972088c32082d8f5bb6
1 #include "git-compat-util.h"
2 #include "config.h"
3 #include "lockfile.h"
4 #include "pack.h"
5 #include "packfile.h"
6 #include "commit.h"
7 #include "object.h"
8 #include "refs.h"
9 #include "revision.h"
10 #include "hash-lookup.h"
11 #include "commit-graph.h"
12 #include "object-store.h"
13 #include "alloc.h"
14 #include "hashmap.h"
15 #include "replace-object.h"
16 #include "progress.h"
17 #include "bloom.h"
18 #include "commit-slab.h"
19 #include "shallow.h"
20 #include "json-writer.h"
21 #include "trace2.h"
22 #include "chunk-format.h"
24 void git_test_write_commit_graph_or_die(void)
26 int flags = 0;
27 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0))
28 return;
30 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS, 0))
31 flags = COMMIT_GRAPH_WRITE_BLOOM_FILTERS;
33 if (write_commit_graph_reachable(the_repository->objects->odb,
34 flags, NULL))
35 die("failed to write commit-graph under GIT_TEST_COMMIT_GRAPH");
38 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
39 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
40 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
41 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
42 #define GRAPH_CHUNKID_GENERATION_DATA 0x47444132 /* "GDA2" */
43 #define GRAPH_CHUNKID_GENERATION_DATA_OVERFLOW 0x47444f32 /* "GDO2" */
44 #define GRAPH_CHUNKID_EXTRAEDGES 0x45444745 /* "EDGE" */
45 #define GRAPH_CHUNKID_BLOOMINDEXES 0x42494458 /* "BIDX" */
46 #define GRAPH_CHUNKID_BLOOMDATA 0x42444154 /* "BDAT" */
47 #define GRAPH_CHUNKID_BASE 0x42415345 /* "BASE" */
49 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
51 #define GRAPH_VERSION_1 0x1
52 #define GRAPH_VERSION GRAPH_VERSION_1
54 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
55 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
56 #define GRAPH_PARENT_NONE 0x70000000
58 #define GRAPH_LAST_EDGE 0x80000000
60 #define GRAPH_HEADER_SIZE 8
61 #define GRAPH_FANOUT_SIZE (4 * 256)
62 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * CHUNK_TOC_ENTRY_SIZE \
63 + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
65 #define CORRECTED_COMMIT_DATE_OFFSET_OVERFLOW (1ULL << 31)
67 /* Remember to update object flag allocation in object.h */
68 #define REACHABLE (1u<<15)
70 define_commit_slab(topo_level_slab, uint32_t);
72 /* Keep track of the order in which commits are added to our list. */
73 define_commit_slab(commit_pos, int);
74 static struct commit_pos commit_pos = COMMIT_SLAB_INIT(1, commit_pos);
76 static void set_commit_pos(struct repository *r, const struct object_id *oid)
78 static int32_t max_pos;
79 struct commit *commit = lookup_commit(r, oid);
81 if (!commit)
82 return; /* should never happen, but be lenient */
84 *commit_pos_at(&commit_pos, commit) = max_pos++;
87 static int commit_pos_cmp(const void *va, const void *vb)
89 const struct commit *a = *(const struct commit **)va;
90 const struct commit *b = *(const struct commit **)vb;
91 return commit_pos_at(&commit_pos, a) -
92 commit_pos_at(&commit_pos, b);
95 define_commit_slab(commit_graph_data_slab, struct commit_graph_data);
96 static struct commit_graph_data_slab commit_graph_data_slab =
97 COMMIT_SLAB_INIT(1, commit_graph_data_slab);
99 static int get_configured_generation_version(struct repository *r)
101 int version = 2;
102 repo_config_get_int(r, "commitgraph.generationversion", &version);
103 return version;
106 uint32_t commit_graph_position(const struct commit *c)
108 struct commit_graph_data *data =
109 commit_graph_data_slab_peek(&commit_graph_data_slab, c);
111 return data ? data->graph_pos : COMMIT_NOT_FROM_GRAPH;
114 timestamp_t commit_graph_generation(const struct commit *c)
116 struct commit_graph_data *data =
117 commit_graph_data_slab_peek(&commit_graph_data_slab, c);
119 if (!data)
120 return GENERATION_NUMBER_INFINITY;
121 else if (data->graph_pos == COMMIT_NOT_FROM_GRAPH)
122 return GENERATION_NUMBER_INFINITY;
124 return data->generation;
127 static struct commit_graph_data *commit_graph_data_at(const struct commit *c)
129 unsigned int i, nth_slab;
130 struct commit_graph_data *data =
131 commit_graph_data_slab_peek(&commit_graph_data_slab, c);
133 if (data)
134 return data;
136 nth_slab = c->index / commit_graph_data_slab.slab_size;
137 data = commit_graph_data_slab_at(&commit_graph_data_slab, c);
140 * commit-slab initializes elements with zero, overwrite this with
141 * COMMIT_NOT_FROM_GRAPH for graph_pos.
143 * We avoid initializing generation with checking if graph position
144 * is not COMMIT_NOT_FROM_GRAPH.
146 for (i = 0; i < commit_graph_data_slab.slab_size; i++) {
147 commit_graph_data_slab.slab[nth_slab][i].graph_pos =
148 COMMIT_NOT_FROM_GRAPH;
151 return data;
155 * Should be used only while writing commit-graph as it compares
156 * generation value of commits by directly accessing commit-slab.
158 static int commit_gen_cmp(const void *va, const void *vb)
160 const struct commit *a = *(const struct commit **)va;
161 const struct commit *b = *(const struct commit **)vb;
163 const timestamp_t generation_a = commit_graph_data_at(a)->generation;
164 const timestamp_t generation_b = commit_graph_data_at(b)->generation;
165 /* lower generation commits first */
166 if (generation_a < generation_b)
167 return -1;
168 else if (generation_a > generation_b)
169 return 1;
171 /* use date as a heuristic when generations are equal */
172 if (a->date < b->date)
173 return -1;
174 else if (a->date > b->date)
175 return 1;
176 return 0;
179 char *get_commit_graph_filename(struct object_directory *obj_dir)
181 return xstrfmt("%s/info/commit-graph", obj_dir->path);
184 static char *get_split_graph_filename(struct object_directory *odb,
185 const char *oid_hex)
187 return xstrfmt("%s/info/commit-graphs/graph-%s.graph", odb->path,
188 oid_hex);
191 char *get_commit_graph_chain_filename(struct object_directory *odb)
193 return xstrfmt("%s/info/commit-graphs/commit-graph-chain", odb->path);
196 static struct commit_graph *alloc_commit_graph(void)
198 struct commit_graph *g = xcalloc(1, sizeof(*g));
200 return g;
203 extern int read_replace_refs;
205 static int commit_graph_compatible(struct repository *r)
207 if (!r->gitdir)
208 return 0;
210 if (read_replace_refs) {
211 prepare_replace_object(r);
212 if (hashmap_get_size(&r->objects->replace_map->map))
213 return 0;
216 prepare_commit_graft(r);
217 if (r->parsed_objects &&
218 (r->parsed_objects->grafts_nr || r->parsed_objects->substituted_parent))
219 return 0;
220 if (is_repository_shallow(r))
221 return 0;
223 return 1;
226 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
228 *fd = git_open(graph_file);
229 if (*fd < 0)
230 return 0;
231 if (fstat(*fd, st)) {
232 close(*fd);
233 return 0;
235 return 1;
238 struct commit_graph *load_commit_graph_one_fd_st(struct repository *r,
239 int fd, struct stat *st,
240 struct object_directory *odb)
242 void *graph_map;
243 size_t graph_size;
244 struct commit_graph *ret;
246 graph_size = xsize_t(st->st_size);
248 if (graph_size < GRAPH_MIN_SIZE) {
249 close(fd);
250 error(_("commit-graph file is too small"));
251 return NULL;
253 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
254 close(fd);
255 prepare_repo_settings(r);
256 ret = parse_commit_graph(&r->settings, graph_map, graph_size);
258 if (ret)
259 ret->odb = odb;
260 else
261 munmap(graph_map, graph_size);
263 return ret;
266 static int verify_commit_graph_lite(struct commit_graph *g)
269 * Basic validation shared between parse_commit_graph()
270 * which'll be called every time the graph is used, and the
271 * much more expensive verify_commit_graph() used by
272 * "commit-graph verify".
274 * There should only be very basic checks here to ensure that
275 * we don't e.g. segfault in fill_commit_in_graph(), but
276 * because this is a very hot codepath nothing that e.g. loops
277 * over g->num_commits, or runs a checksum on the commit-graph
278 * itself.
280 if (!g->chunk_oid_fanout) {
281 error("commit-graph is missing the OID Fanout chunk");
282 return 1;
284 if (!g->chunk_oid_lookup) {
285 error("commit-graph is missing the OID Lookup chunk");
286 return 1;
288 if (!g->chunk_commit_data) {
289 error("commit-graph is missing the Commit Data chunk");
290 return 1;
293 return 0;
296 static int graph_read_oid_lookup(const unsigned char *chunk_start,
297 size_t chunk_size, void *data)
299 struct commit_graph *g = data;
300 g->chunk_oid_lookup = chunk_start;
301 g->num_commits = chunk_size / g->hash_len;
302 return 0;
305 static int graph_read_bloom_data(const unsigned char *chunk_start,
306 size_t chunk_size, void *data)
308 struct commit_graph *g = data;
309 uint32_t hash_version;
310 g->chunk_bloom_data = chunk_start;
311 hash_version = get_be32(chunk_start);
313 if (hash_version != 1)
314 return 0;
316 g->bloom_filter_settings = xmalloc(sizeof(struct bloom_filter_settings));
317 g->bloom_filter_settings->hash_version = hash_version;
318 g->bloom_filter_settings->num_hashes = get_be32(chunk_start + 4);
319 g->bloom_filter_settings->bits_per_entry = get_be32(chunk_start + 8);
320 g->bloom_filter_settings->max_changed_paths = DEFAULT_BLOOM_MAX_CHANGES;
322 return 0;
325 struct commit_graph *parse_commit_graph(struct repo_settings *s,
326 void *graph_map, size_t graph_size)
328 const unsigned char *data;
329 struct commit_graph *graph;
330 uint32_t graph_signature;
331 unsigned char graph_version, hash_version;
332 struct chunkfile *cf = NULL;
334 if (!graph_map)
335 return NULL;
337 if (graph_size < GRAPH_MIN_SIZE)
338 return NULL;
340 data = (const unsigned char *)graph_map;
342 graph_signature = get_be32(data);
343 if (graph_signature != GRAPH_SIGNATURE) {
344 error(_("commit-graph signature %X does not match signature %X"),
345 graph_signature, GRAPH_SIGNATURE);
346 return NULL;
349 graph_version = *(unsigned char*)(data + 4);
350 if (graph_version != GRAPH_VERSION) {
351 error(_("commit-graph version %X does not match version %X"),
352 graph_version, GRAPH_VERSION);
353 return NULL;
356 hash_version = *(unsigned char*)(data + 5);
357 if (hash_version != oid_version(the_hash_algo)) {
358 error(_("commit-graph hash version %X does not match version %X"),
359 hash_version, oid_version(the_hash_algo));
360 return NULL;
363 graph = alloc_commit_graph();
365 graph->hash_len = the_hash_algo->rawsz;
366 graph->num_chunks = *(unsigned char*)(data + 6);
367 graph->data = graph_map;
368 graph->data_len = graph_size;
370 if (graph_size < GRAPH_HEADER_SIZE +
371 (graph->num_chunks + 1) * CHUNK_TOC_ENTRY_SIZE +
372 GRAPH_FANOUT_SIZE + the_hash_algo->rawsz) {
373 error(_("commit-graph file is too small to hold %u chunks"),
374 graph->num_chunks);
375 free(graph);
376 return NULL;
379 cf = init_chunkfile(NULL);
381 if (read_table_of_contents(cf, graph->data, graph_size,
382 GRAPH_HEADER_SIZE, graph->num_chunks))
383 goto free_and_return;
385 pair_chunk(cf, GRAPH_CHUNKID_OIDFANOUT,
386 (const unsigned char **)&graph->chunk_oid_fanout);
387 read_chunk(cf, GRAPH_CHUNKID_OIDLOOKUP, graph_read_oid_lookup, graph);
388 pair_chunk(cf, GRAPH_CHUNKID_DATA, &graph->chunk_commit_data);
389 pair_chunk(cf, GRAPH_CHUNKID_EXTRAEDGES, &graph->chunk_extra_edges);
390 pair_chunk(cf, GRAPH_CHUNKID_BASE, &graph->chunk_base_graphs);
392 if (s->commit_graph_generation_version >= 2) {
393 pair_chunk(cf, GRAPH_CHUNKID_GENERATION_DATA,
394 &graph->chunk_generation_data);
395 pair_chunk(cf, GRAPH_CHUNKID_GENERATION_DATA_OVERFLOW,
396 &graph->chunk_generation_data_overflow);
398 if (graph->chunk_generation_data)
399 graph->read_generation_data = 1;
402 if (s->commit_graph_read_changed_paths) {
403 pair_chunk(cf, GRAPH_CHUNKID_BLOOMINDEXES,
404 &graph->chunk_bloom_indexes);
405 read_chunk(cf, GRAPH_CHUNKID_BLOOMDATA,
406 graph_read_bloom_data, graph);
409 if (graph->chunk_bloom_indexes && graph->chunk_bloom_data) {
410 init_bloom_filters();
411 } else {
412 /* We need both the bloom chunks to exist together. Else ignore the data */
413 graph->chunk_bloom_indexes = NULL;
414 graph->chunk_bloom_data = NULL;
415 FREE_AND_NULL(graph->bloom_filter_settings);
418 oidread(&graph->oid, graph->data + graph->data_len - graph->hash_len);
420 if (verify_commit_graph_lite(graph))
421 goto free_and_return;
423 free_chunkfile(cf);
424 return graph;
426 free_and_return:
427 free_chunkfile(cf);
428 free(graph->bloom_filter_settings);
429 free(graph);
430 return NULL;
433 static struct commit_graph *load_commit_graph_one(struct repository *r,
434 const char *graph_file,
435 struct object_directory *odb)
438 struct stat st;
439 int fd;
440 struct commit_graph *g;
441 int open_ok = open_commit_graph(graph_file, &fd, &st);
443 if (!open_ok)
444 return NULL;
446 g = load_commit_graph_one_fd_st(r, fd, &st, odb);
448 if (g)
449 g->filename = xstrdup(graph_file);
451 return g;
454 static struct commit_graph *load_commit_graph_v1(struct repository *r,
455 struct object_directory *odb)
457 char *graph_name = get_commit_graph_filename(odb);
458 struct commit_graph *g = load_commit_graph_one(r, graph_name, odb);
459 free(graph_name);
461 return g;
464 static int add_graph_to_chain(struct commit_graph *g,
465 struct commit_graph *chain,
466 struct object_id *oids,
467 int n)
469 struct commit_graph *cur_g = chain;
471 if (n && !g->chunk_base_graphs) {
472 warning(_("commit-graph has no base graphs chunk"));
473 return 0;
476 while (n) {
477 n--;
479 if (!cur_g ||
480 !oideq(&oids[n], &cur_g->oid) ||
481 !hasheq(oids[n].hash, g->chunk_base_graphs + g->hash_len * n)) {
482 warning(_("commit-graph chain does not match"));
483 return 0;
486 cur_g = cur_g->base_graph;
489 g->base_graph = chain;
491 if (chain)
492 g->num_commits_in_base = chain->num_commits + chain->num_commits_in_base;
494 return 1;
497 static struct commit_graph *load_commit_graph_chain(struct repository *r,
498 struct object_directory *odb)
500 struct commit_graph *graph_chain = NULL;
501 struct strbuf line = STRBUF_INIT;
502 struct stat st;
503 struct object_id *oids;
504 int i = 0, valid = 1, count;
505 char *chain_name = get_commit_graph_chain_filename(odb);
506 FILE *fp;
507 int stat_res;
509 fp = fopen(chain_name, "r");
510 stat_res = stat(chain_name, &st);
511 free(chain_name);
513 if (!fp)
514 return NULL;
515 if (stat_res ||
516 st.st_size <= the_hash_algo->hexsz) {
517 fclose(fp);
518 return NULL;
521 count = st.st_size / (the_hash_algo->hexsz + 1);
522 CALLOC_ARRAY(oids, count);
524 prepare_alt_odb(r);
526 for (i = 0; i < count; i++) {
527 struct object_directory *odb;
529 if (strbuf_getline_lf(&line, fp) == EOF)
530 break;
532 if (get_oid_hex(line.buf, &oids[i])) {
533 warning(_("invalid commit-graph chain: line '%s' not a hash"),
534 line.buf);
535 valid = 0;
536 break;
539 valid = 0;
540 for (odb = r->objects->odb; odb; odb = odb->next) {
541 char *graph_name = get_split_graph_filename(odb, line.buf);
542 struct commit_graph *g = load_commit_graph_one(r, graph_name, odb);
544 free(graph_name);
546 if (g) {
547 if (add_graph_to_chain(g, graph_chain, oids, i)) {
548 graph_chain = g;
549 valid = 1;
552 break;
556 if (!valid) {
557 warning(_("unable to find all commit-graph files"));
558 break;
562 free(oids);
563 fclose(fp);
564 strbuf_release(&line);
566 return graph_chain;
570 * returns 1 if and only if all graphs in the chain have
571 * corrected commit dates stored in the generation_data chunk.
573 static int validate_mixed_generation_chain(struct commit_graph *g)
575 int read_generation_data = 1;
576 struct commit_graph *p = g;
578 while (read_generation_data && p) {
579 read_generation_data = p->read_generation_data;
580 p = p->base_graph;
583 if (read_generation_data)
584 return 1;
586 while (g) {
587 g->read_generation_data = 0;
588 g = g->base_graph;
591 return 0;
594 struct commit_graph *read_commit_graph_one(struct repository *r,
595 struct object_directory *odb)
597 struct commit_graph *g = load_commit_graph_v1(r, odb);
599 if (!g)
600 g = load_commit_graph_chain(r, odb);
602 validate_mixed_generation_chain(g);
604 return g;
607 static void prepare_commit_graph_one(struct repository *r,
608 struct object_directory *odb)
611 if (r->objects->commit_graph)
612 return;
614 r->objects->commit_graph = read_commit_graph_one(r, odb);
618 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
620 * On the first invocation, this function attempts to load the commit
621 * graph if the_repository is configured to have one.
623 static int prepare_commit_graph(struct repository *r)
625 struct object_directory *odb;
628 * Early return if there is no git dir or if the commit graph is
629 * disabled.
631 * This must come before the "already attempted?" check below, because
632 * we want to disable even an already-loaded graph file.
634 if (!r->gitdir || r->commit_graph_disabled)
635 return 0;
637 if (r->objects->commit_graph_attempted)
638 return !!r->objects->commit_graph;
639 r->objects->commit_graph_attempted = 1;
641 prepare_repo_settings(r);
643 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
644 r->settings.core_commit_graph != 1)
646 * This repository is not configured to use commit graphs, so
647 * do not load one. (But report commit_graph_attempted anyway
648 * so that commit graph loading is not attempted again for this
649 * repository.)
651 return 0;
653 if (!commit_graph_compatible(r))
654 return 0;
656 prepare_alt_odb(r);
657 for (odb = r->objects->odb;
658 !r->objects->commit_graph && odb;
659 odb = odb->next)
660 prepare_commit_graph_one(r, odb);
661 return !!r->objects->commit_graph;
664 int generation_numbers_enabled(struct repository *r)
666 uint32_t first_generation;
667 struct commit_graph *g;
668 if (!prepare_commit_graph(r))
669 return 0;
671 g = r->objects->commit_graph;
673 if (!g->num_commits)
674 return 0;
676 first_generation = get_be32(g->chunk_commit_data +
677 g->hash_len + 8) >> 2;
679 return !!first_generation;
682 int corrected_commit_dates_enabled(struct repository *r)
684 struct commit_graph *g;
685 if (!prepare_commit_graph(r))
686 return 0;
688 g = r->objects->commit_graph;
690 if (!g->num_commits)
691 return 0;
693 return g->read_generation_data;
696 struct bloom_filter_settings *get_bloom_filter_settings(struct repository *r)
698 struct commit_graph *g = r->objects->commit_graph;
699 while (g) {
700 if (g->bloom_filter_settings)
701 return g->bloom_filter_settings;
702 g = g->base_graph;
704 return NULL;
707 static void close_commit_graph_one(struct commit_graph *g)
709 if (!g)
710 return;
712 clear_commit_graph_data_slab(&commit_graph_data_slab);
713 close_commit_graph_one(g->base_graph);
714 free_commit_graph(g);
717 void close_commit_graph(struct raw_object_store *o)
719 close_commit_graph_one(o->commit_graph);
720 o->commit_graph = NULL;
723 static int bsearch_graph(struct commit_graph *g, const struct object_id *oid, uint32_t *pos)
725 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
726 g->chunk_oid_lookup, g->hash_len, pos);
729 static void load_oid_from_graph(struct commit_graph *g,
730 uint32_t pos,
731 struct object_id *oid)
733 uint32_t lex_index;
735 while (g && pos < g->num_commits_in_base)
736 g = g->base_graph;
738 if (!g)
739 BUG("NULL commit-graph");
741 if (pos >= g->num_commits + g->num_commits_in_base)
742 die(_("invalid commit position. commit-graph is likely corrupt"));
744 lex_index = pos - g->num_commits_in_base;
746 oidread(oid, g->chunk_oid_lookup + g->hash_len * lex_index);
749 static struct commit_list **insert_parent_or_die(struct repository *r,
750 struct commit_graph *g,
751 uint32_t pos,
752 struct commit_list **pptr)
754 struct commit *c;
755 struct object_id oid;
757 if (pos >= g->num_commits + g->num_commits_in_base)
758 die("invalid parent position %"PRIu32, pos);
760 load_oid_from_graph(g, pos, &oid);
761 c = lookup_commit(r, &oid);
762 if (!c)
763 die(_("could not find commit %s"), oid_to_hex(&oid));
764 commit_graph_data_at(c)->graph_pos = pos;
765 return &commit_list_insert(c, pptr)->next;
768 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
770 const unsigned char *commit_data;
771 struct commit_graph_data *graph_data;
772 uint32_t lex_index, offset_pos;
773 uint64_t date_high, date_low, offset;
775 while (pos < g->num_commits_in_base)
776 g = g->base_graph;
778 if (pos >= g->num_commits + g->num_commits_in_base)
779 die(_("invalid commit position. commit-graph is likely corrupt"));
781 lex_index = pos - g->num_commits_in_base;
782 commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * lex_index;
784 graph_data = commit_graph_data_at(item);
785 graph_data->graph_pos = pos;
787 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
788 date_low = get_be32(commit_data + g->hash_len + 12);
789 item->date = (timestamp_t)((date_high << 32) | date_low);
791 if (g->read_generation_data) {
792 offset = (timestamp_t)get_be32(g->chunk_generation_data + sizeof(uint32_t) * lex_index);
794 if (offset & CORRECTED_COMMIT_DATE_OFFSET_OVERFLOW) {
795 if (!g->chunk_generation_data_overflow)
796 die(_("commit-graph requires overflow generation data but has none"));
798 offset_pos = offset ^ CORRECTED_COMMIT_DATE_OFFSET_OVERFLOW;
799 graph_data->generation = item->date + get_be64(g->chunk_generation_data_overflow + 8 * offset_pos);
800 } else
801 graph_data->generation = item->date + offset;
802 } else
803 graph_data->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
805 if (g->topo_levels)
806 *topo_level_slab_at(g->topo_levels, item) = get_be32(commit_data + g->hash_len + 8) >> 2;
809 static inline void set_commit_tree(struct commit *c, struct tree *t)
811 c->maybe_tree = t;
814 static int fill_commit_in_graph(struct repository *r,
815 struct commit *item,
816 struct commit_graph *g, uint32_t pos)
818 uint32_t edge_value;
819 uint32_t *parent_data_ptr;
820 struct commit_list **pptr;
821 const unsigned char *commit_data;
822 uint32_t lex_index;
824 while (pos < g->num_commits_in_base)
825 g = g->base_graph;
827 fill_commit_graph_info(item, g, pos);
829 lex_index = pos - g->num_commits_in_base;
830 commit_data = g->chunk_commit_data + (g->hash_len + 16) * lex_index;
832 item->object.parsed = 1;
834 set_commit_tree(item, NULL);
836 pptr = &item->parents;
838 edge_value = get_be32(commit_data + g->hash_len);
839 if (edge_value == GRAPH_PARENT_NONE)
840 return 1;
841 pptr = insert_parent_or_die(r, g, edge_value, pptr);
843 edge_value = get_be32(commit_data + g->hash_len + 4);
844 if (edge_value == GRAPH_PARENT_NONE)
845 return 1;
846 if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
847 pptr = insert_parent_or_die(r, g, edge_value, pptr);
848 return 1;
851 parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
852 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
853 do {
854 edge_value = get_be32(parent_data_ptr);
855 pptr = insert_parent_or_die(r, g,
856 edge_value & GRAPH_EDGE_LAST_MASK,
857 pptr);
858 parent_data_ptr++;
859 } while (!(edge_value & GRAPH_LAST_EDGE));
861 return 1;
864 static int search_commit_pos_in_graph(const struct object_id *id, struct commit_graph *g, uint32_t *pos)
866 struct commit_graph *cur_g = g;
867 uint32_t lex_index;
869 while (cur_g && !bsearch_graph(cur_g, id, &lex_index))
870 cur_g = cur_g->base_graph;
872 if (cur_g) {
873 *pos = lex_index + cur_g->num_commits_in_base;
874 return 1;
877 return 0;
880 static int find_commit_pos_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
882 uint32_t graph_pos = commit_graph_position(item);
883 if (graph_pos != COMMIT_NOT_FROM_GRAPH) {
884 *pos = graph_pos;
885 return 1;
886 } else {
887 return search_commit_pos_in_graph(&item->object.oid, g, pos);
891 int repo_find_commit_pos_in_graph(struct repository *r, struct commit *c,
892 uint32_t *pos)
894 if (!prepare_commit_graph(r))
895 return 0;
896 return find_commit_pos_in_graph(c, r->objects->commit_graph, pos);
899 struct commit *lookup_commit_in_graph(struct repository *repo, const struct object_id *id)
901 struct commit *commit;
902 uint32_t pos;
904 if (!prepare_commit_graph(repo))
905 return NULL;
906 if (!search_commit_pos_in_graph(id, repo->objects->commit_graph, &pos))
907 return NULL;
908 if (!has_object(repo, id, 0))
909 return NULL;
911 commit = lookup_commit(repo, id);
912 if (!commit)
913 return NULL;
914 if (commit->object.parsed)
915 return commit;
917 if (!fill_commit_in_graph(repo, commit, repo->objects->commit_graph, pos))
918 return NULL;
920 return commit;
923 static int parse_commit_in_graph_one(struct repository *r,
924 struct commit_graph *g,
925 struct commit *item)
927 uint32_t pos;
929 if (item->object.parsed)
930 return 1;
932 if (find_commit_pos_in_graph(item, g, &pos))
933 return fill_commit_in_graph(r, item, g, pos);
935 return 0;
938 int parse_commit_in_graph(struct repository *r, struct commit *item)
940 static int checked_env = 0;
942 if (!checked_env &&
943 git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_PARSE, 0))
944 die("dying as requested by the '%s' variable on commit-graph parse!",
945 GIT_TEST_COMMIT_GRAPH_DIE_ON_PARSE);
946 checked_env = 1;
948 if (!prepare_commit_graph(r))
949 return 0;
950 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
953 void load_commit_graph_info(struct repository *r, struct commit *item)
955 uint32_t pos;
956 if (repo_find_commit_pos_in_graph(r, item, &pos))
957 fill_commit_graph_info(item, r->objects->commit_graph, pos);
960 static struct tree *load_tree_for_commit(struct repository *r,
961 struct commit_graph *g,
962 struct commit *c)
964 struct object_id oid;
965 const unsigned char *commit_data;
966 uint32_t graph_pos = commit_graph_position(c);
968 while (graph_pos < g->num_commits_in_base)
969 g = g->base_graph;
971 commit_data = g->chunk_commit_data +
972 GRAPH_DATA_WIDTH * (graph_pos - g->num_commits_in_base);
974 oidread(&oid, commit_data);
975 set_commit_tree(c, lookup_tree(r, &oid));
977 return c->maybe_tree;
980 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
981 struct commit_graph *g,
982 const struct commit *c)
984 if (c->maybe_tree)
985 return c->maybe_tree;
986 if (commit_graph_position(c) == COMMIT_NOT_FROM_GRAPH)
987 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
989 return load_tree_for_commit(r, g, (struct commit *)c);
992 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
994 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
997 struct packed_commit_list {
998 struct commit **list;
999 size_t nr;
1000 size_t alloc;
1003 struct write_commit_graph_context {
1004 struct repository *r;
1005 struct object_directory *odb;
1006 char *graph_name;
1007 struct oid_array oids;
1008 struct packed_commit_list commits;
1009 int num_extra_edges;
1010 int num_generation_data_overflows;
1011 unsigned long approx_nr_objects;
1012 struct progress *progress;
1013 int progress_done;
1014 uint64_t progress_cnt;
1016 char *base_graph_name;
1017 int num_commit_graphs_before;
1018 int num_commit_graphs_after;
1019 char **commit_graph_filenames_before;
1020 char **commit_graph_filenames_after;
1021 char **commit_graph_hash_after;
1022 uint32_t new_num_commits_in_base;
1023 struct commit_graph *new_base_graph;
1025 unsigned append:1,
1026 report_progress:1,
1027 split:1,
1028 changed_paths:1,
1029 order_by_pack:1,
1030 write_generation_data:1,
1031 trust_generation_numbers:1;
1033 struct topo_level_slab *topo_levels;
1034 const struct commit_graph_opts *opts;
1035 size_t total_bloom_filter_data_size;
1036 const struct bloom_filter_settings *bloom_settings;
1038 int count_bloom_filter_computed;
1039 int count_bloom_filter_not_computed;
1040 int count_bloom_filter_trunc_empty;
1041 int count_bloom_filter_trunc_large;
1044 static int write_graph_chunk_fanout(struct hashfile *f,
1045 void *data)
1047 struct write_commit_graph_context *ctx = data;
1048 int i, count = 0;
1049 struct commit **list = ctx->commits.list;
1052 * Write the first-level table (the list is sorted,
1053 * but we use a 256-entry lookup to be able to avoid
1054 * having to do eight extra binary search iterations).
1056 for (i = 0; i < 256; i++) {
1057 while (count < ctx->commits.nr) {
1058 if ((*list)->object.oid.hash[0] != i)
1059 break;
1060 display_progress(ctx->progress, ++ctx->progress_cnt);
1061 count++;
1062 list++;
1065 hashwrite_be32(f, count);
1068 return 0;
1071 static int write_graph_chunk_oids(struct hashfile *f,
1072 void *data)
1074 struct write_commit_graph_context *ctx = data;
1075 struct commit **list = ctx->commits.list;
1076 int count;
1077 for (count = 0; count < ctx->commits.nr; count++, list++) {
1078 display_progress(ctx->progress, ++ctx->progress_cnt);
1079 hashwrite(f, (*list)->object.oid.hash, the_hash_algo->rawsz);
1082 return 0;
1085 static const struct object_id *commit_to_oid(size_t index, const void *table)
1087 const struct commit * const *commits = table;
1088 return &commits[index]->object.oid;
1091 static int write_graph_chunk_data(struct hashfile *f,
1092 void *data)
1094 struct write_commit_graph_context *ctx = data;
1095 struct commit **list = ctx->commits.list;
1096 struct commit **last = ctx->commits.list + ctx->commits.nr;
1097 uint32_t num_extra_edges = 0;
1099 while (list < last) {
1100 struct commit_list *parent;
1101 struct object_id *tree;
1102 int edge_value;
1103 uint32_t packedDate[2];
1104 display_progress(ctx->progress, ++ctx->progress_cnt);
1106 if (repo_parse_commit_no_graph(ctx->r, *list))
1107 die(_("unable to parse commit %s"),
1108 oid_to_hex(&(*list)->object.oid));
1109 tree = get_commit_tree_oid(*list);
1110 hashwrite(f, tree->hash, the_hash_algo->rawsz);
1112 parent = (*list)->parents;
1114 if (!parent)
1115 edge_value = GRAPH_PARENT_NONE;
1116 else {
1117 edge_value = oid_pos(&parent->item->object.oid,
1118 ctx->commits.list,
1119 ctx->commits.nr,
1120 commit_to_oid);
1122 if (edge_value >= 0)
1123 edge_value += ctx->new_num_commits_in_base;
1124 else if (ctx->new_base_graph) {
1125 uint32_t pos;
1126 if (find_commit_pos_in_graph(parent->item,
1127 ctx->new_base_graph,
1128 &pos))
1129 edge_value = pos;
1132 if (edge_value < 0)
1133 BUG("missing parent %s for commit %s",
1134 oid_to_hex(&parent->item->object.oid),
1135 oid_to_hex(&(*list)->object.oid));
1138 hashwrite_be32(f, edge_value);
1140 if (parent)
1141 parent = parent->next;
1143 if (!parent)
1144 edge_value = GRAPH_PARENT_NONE;
1145 else if (parent->next)
1146 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
1147 else {
1148 edge_value = oid_pos(&parent->item->object.oid,
1149 ctx->commits.list,
1150 ctx->commits.nr,
1151 commit_to_oid);
1153 if (edge_value >= 0)
1154 edge_value += ctx->new_num_commits_in_base;
1155 else if (ctx->new_base_graph) {
1156 uint32_t pos;
1157 if (find_commit_pos_in_graph(parent->item,
1158 ctx->new_base_graph,
1159 &pos))
1160 edge_value = pos;
1163 if (edge_value < 0)
1164 BUG("missing parent %s for commit %s",
1165 oid_to_hex(&parent->item->object.oid),
1166 oid_to_hex(&(*list)->object.oid));
1169 hashwrite_be32(f, edge_value);
1171 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
1172 do {
1173 num_extra_edges++;
1174 parent = parent->next;
1175 } while (parent);
1178 if (sizeof((*list)->date) > 4)
1179 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
1180 else
1181 packedDate[0] = 0;
1183 packedDate[0] |= htonl(*topo_level_slab_at(ctx->topo_levels, *list) << 2);
1185 packedDate[1] = htonl((*list)->date);
1186 hashwrite(f, packedDate, 8);
1188 list++;
1191 return 0;
1194 static int write_graph_chunk_generation_data(struct hashfile *f,
1195 void *data)
1197 struct write_commit_graph_context *ctx = data;
1198 int i, num_generation_data_overflows = 0;
1200 for (i = 0; i < ctx->commits.nr; i++) {
1201 struct commit *c = ctx->commits.list[i];
1202 timestamp_t offset;
1203 repo_parse_commit(ctx->r, c);
1204 offset = commit_graph_data_at(c)->generation - c->date;
1205 display_progress(ctx->progress, ++ctx->progress_cnt);
1207 if (offset > GENERATION_NUMBER_V2_OFFSET_MAX) {
1208 offset = CORRECTED_COMMIT_DATE_OFFSET_OVERFLOW | num_generation_data_overflows;
1209 num_generation_data_overflows++;
1212 hashwrite_be32(f, offset);
1215 return 0;
1218 static int write_graph_chunk_generation_data_overflow(struct hashfile *f,
1219 void *data)
1221 struct write_commit_graph_context *ctx = data;
1222 int i;
1223 for (i = 0; i < ctx->commits.nr; i++) {
1224 struct commit *c = ctx->commits.list[i];
1225 timestamp_t offset = commit_graph_data_at(c)->generation - c->date;
1226 display_progress(ctx->progress, ++ctx->progress_cnt);
1228 if (offset > GENERATION_NUMBER_V2_OFFSET_MAX) {
1229 hashwrite_be32(f, offset >> 32);
1230 hashwrite_be32(f, (uint32_t) offset);
1234 return 0;
1237 static int write_graph_chunk_extra_edges(struct hashfile *f,
1238 void *data)
1240 struct write_commit_graph_context *ctx = data;
1241 struct commit **list = ctx->commits.list;
1242 struct commit **last = ctx->commits.list + ctx->commits.nr;
1243 struct commit_list *parent;
1245 while (list < last) {
1246 int num_parents = 0;
1248 display_progress(ctx->progress, ++ctx->progress_cnt);
1250 for (parent = (*list)->parents; num_parents < 3 && parent;
1251 parent = parent->next)
1252 num_parents++;
1254 if (num_parents <= 2) {
1255 list++;
1256 continue;
1259 /* Since num_parents > 2, this initializer is safe. */
1260 for (parent = (*list)->parents->next; parent; parent = parent->next) {
1261 int edge_value = oid_pos(&parent->item->object.oid,
1262 ctx->commits.list,
1263 ctx->commits.nr,
1264 commit_to_oid);
1266 if (edge_value >= 0)
1267 edge_value += ctx->new_num_commits_in_base;
1268 else if (ctx->new_base_graph) {
1269 uint32_t pos;
1270 if (find_commit_pos_in_graph(parent->item,
1271 ctx->new_base_graph,
1272 &pos))
1273 edge_value = pos;
1276 if (edge_value < 0)
1277 BUG("missing parent %s for commit %s",
1278 oid_to_hex(&parent->item->object.oid),
1279 oid_to_hex(&(*list)->object.oid));
1280 else if (!parent->next)
1281 edge_value |= GRAPH_LAST_EDGE;
1283 hashwrite_be32(f, edge_value);
1286 list++;
1289 return 0;
1292 static int write_graph_chunk_bloom_indexes(struct hashfile *f,
1293 void *data)
1295 struct write_commit_graph_context *ctx = data;
1296 struct commit **list = ctx->commits.list;
1297 struct commit **last = ctx->commits.list + ctx->commits.nr;
1298 uint32_t cur_pos = 0;
1300 while (list < last) {
1301 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list);
1302 size_t len = filter ? filter->len : 0;
1303 cur_pos += len;
1304 display_progress(ctx->progress, ++ctx->progress_cnt);
1305 hashwrite_be32(f, cur_pos);
1306 list++;
1309 return 0;
1312 static void trace2_bloom_filter_settings(struct write_commit_graph_context *ctx)
1314 struct json_writer jw = JSON_WRITER_INIT;
1316 jw_object_begin(&jw, 0);
1317 jw_object_intmax(&jw, "hash_version", ctx->bloom_settings->hash_version);
1318 jw_object_intmax(&jw, "num_hashes", ctx->bloom_settings->num_hashes);
1319 jw_object_intmax(&jw, "bits_per_entry", ctx->bloom_settings->bits_per_entry);
1320 jw_object_intmax(&jw, "max_changed_paths", ctx->bloom_settings->max_changed_paths);
1321 jw_end(&jw);
1323 trace2_data_json("bloom", ctx->r, "settings", &jw);
1325 jw_release(&jw);
1328 static int write_graph_chunk_bloom_data(struct hashfile *f,
1329 void *data)
1331 struct write_commit_graph_context *ctx = data;
1332 struct commit **list = ctx->commits.list;
1333 struct commit **last = ctx->commits.list + ctx->commits.nr;
1335 trace2_bloom_filter_settings(ctx);
1337 hashwrite_be32(f, ctx->bloom_settings->hash_version);
1338 hashwrite_be32(f, ctx->bloom_settings->num_hashes);
1339 hashwrite_be32(f, ctx->bloom_settings->bits_per_entry);
1341 while (list < last) {
1342 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list);
1343 size_t len = filter ? filter->len : 0;
1345 display_progress(ctx->progress, ++ctx->progress_cnt);
1346 if (len)
1347 hashwrite(f, filter->data, len * sizeof(unsigned char));
1348 list++;
1351 return 0;
1354 static int add_packed_commits(const struct object_id *oid,
1355 struct packed_git *pack,
1356 uint32_t pos,
1357 void *data)
1359 struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
1360 enum object_type type;
1361 off_t offset = nth_packed_object_offset(pack, pos);
1362 struct object_info oi = OBJECT_INFO_INIT;
1364 if (ctx->progress)
1365 display_progress(ctx->progress, ++ctx->progress_done);
1367 oi.typep = &type;
1368 if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
1369 die(_("unable to get type of object %s"), oid_to_hex(oid));
1371 if (type != OBJ_COMMIT)
1372 return 0;
1374 oid_array_append(&ctx->oids, oid);
1375 set_commit_pos(ctx->r, oid);
1377 return 0;
1380 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
1382 struct commit_list *parent;
1383 for (parent = commit->parents; parent; parent = parent->next) {
1384 if (!(parent->item->object.flags & REACHABLE)) {
1385 oid_array_append(&ctx->oids, &parent->item->object.oid);
1386 parent->item->object.flags |= REACHABLE;
1391 static void close_reachable(struct write_commit_graph_context *ctx)
1393 int i;
1394 struct commit *commit;
1395 enum commit_graph_split_flags flags = ctx->opts ?
1396 ctx->opts->split_flags : COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1398 if (ctx->report_progress)
1399 ctx->progress = start_delayed_progress(
1400 _("Loading known commits in commit graph"),
1401 ctx->oids.nr);
1402 for (i = 0; i < ctx->oids.nr; i++) {
1403 display_progress(ctx->progress, i + 1);
1404 commit = lookup_commit(ctx->r, &ctx->oids.oid[i]);
1405 if (commit)
1406 commit->object.flags |= REACHABLE;
1408 stop_progress(&ctx->progress);
1411 * As this loop runs, ctx->oids.nr may grow, but not more
1412 * than the number of missing commits in the reachable
1413 * closure.
1415 if (ctx->report_progress)
1416 ctx->progress = start_delayed_progress(
1417 _("Expanding reachable commits in commit graph"),
1419 for (i = 0; i < ctx->oids.nr; i++) {
1420 display_progress(ctx->progress, i + 1);
1421 commit = lookup_commit(ctx->r, &ctx->oids.oid[i]);
1423 if (!commit)
1424 continue;
1425 if (ctx->split) {
1426 if ((!repo_parse_commit(ctx->r, commit) &&
1427 commit_graph_position(commit) == COMMIT_NOT_FROM_GRAPH) ||
1428 flags == COMMIT_GRAPH_SPLIT_REPLACE)
1429 add_missing_parents(ctx, commit);
1430 } else if (!repo_parse_commit_no_graph(ctx->r, commit))
1431 add_missing_parents(ctx, commit);
1433 stop_progress(&ctx->progress);
1435 if (ctx->report_progress)
1436 ctx->progress = start_delayed_progress(
1437 _("Clearing commit marks in commit graph"),
1438 ctx->oids.nr);
1439 for (i = 0; i < ctx->oids.nr; i++) {
1440 display_progress(ctx->progress, i + 1);
1441 commit = lookup_commit(ctx->r, &ctx->oids.oid[i]);
1443 if (commit)
1444 commit->object.flags &= ~REACHABLE;
1446 stop_progress(&ctx->progress);
1449 static void compute_topological_levels(struct write_commit_graph_context *ctx)
1451 int i;
1452 struct commit_list *list = NULL;
1454 if (ctx->report_progress)
1455 ctx->progress = start_delayed_progress(
1456 _("Computing commit graph topological levels"),
1457 ctx->commits.nr);
1458 for (i = 0; i < ctx->commits.nr; i++) {
1459 struct commit *c = ctx->commits.list[i];
1460 uint32_t level;
1462 repo_parse_commit(ctx->r, c);
1463 level = *topo_level_slab_at(ctx->topo_levels, c);
1465 display_progress(ctx->progress, i + 1);
1466 if (level != GENERATION_NUMBER_ZERO)
1467 continue;
1469 commit_list_insert(c, &list);
1470 while (list) {
1471 struct commit *current = list->item;
1472 struct commit_list *parent;
1473 int all_parents_computed = 1;
1474 uint32_t max_level = 0;
1476 for (parent = current->parents; parent; parent = parent->next) {
1477 repo_parse_commit(ctx->r, parent->item);
1478 level = *topo_level_slab_at(ctx->topo_levels, parent->item);
1480 if (level == GENERATION_NUMBER_ZERO) {
1481 all_parents_computed = 0;
1482 commit_list_insert(parent->item, &list);
1483 break;
1486 if (level > max_level)
1487 max_level = level;
1490 if (all_parents_computed) {
1491 pop_commit(&list);
1493 if (max_level > GENERATION_NUMBER_V1_MAX - 1)
1494 max_level = GENERATION_NUMBER_V1_MAX - 1;
1495 *topo_level_slab_at(ctx->topo_levels, current) = max_level + 1;
1499 stop_progress(&ctx->progress);
1502 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1504 int i;
1505 struct commit_list *list = NULL;
1507 if (ctx->report_progress)
1508 ctx->progress = start_delayed_progress(
1509 _("Computing commit graph generation numbers"),
1510 ctx->commits.nr);
1512 if (!ctx->trust_generation_numbers) {
1513 for (i = 0; i < ctx->commits.nr; i++) {
1514 struct commit *c = ctx->commits.list[i];
1515 repo_parse_commit(ctx->r, c);
1516 commit_graph_data_at(c)->generation = GENERATION_NUMBER_ZERO;
1520 for (i = 0; i < ctx->commits.nr; i++) {
1521 struct commit *c = ctx->commits.list[i];
1522 timestamp_t corrected_commit_date;
1524 repo_parse_commit(ctx->r, c);
1525 corrected_commit_date = commit_graph_data_at(c)->generation;
1527 display_progress(ctx->progress, i + 1);
1528 if (corrected_commit_date != GENERATION_NUMBER_ZERO)
1529 continue;
1531 commit_list_insert(c, &list);
1532 while (list) {
1533 struct commit *current = list->item;
1534 struct commit_list *parent;
1535 int all_parents_computed = 1;
1536 timestamp_t max_corrected_commit_date = 0;
1538 for (parent = current->parents; parent; parent = parent->next) {
1539 repo_parse_commit(ctx->r, parent->item);
1540 corrected_commit_date = commit_graph_data_at(parent->item)->generation;
1542 if (corrected_commit_date == GENERATION_NUMBER_ZERO) {
1543 all_parents_computed = 0;
1544 commit_list_insert(parent->item, &list);
1545 break;
1548 if (corrected_commit_date > max_corrected_commit_date)
1549 max_corrected_commit_date = corrected_commit_date;
1552 if (all_parents_computed) {
1553 pop_commit(&list);
1555 if (current->date && current->date > max_corrected_commit_date)
1556 max_corrected_commit_date = current->date - 1;
1557 commit_graph_data_at(current)->generation = max_corrected_commit_date + 1;
1562 for (i = 0; i < ctx->commits.nr; i++) {
1563 struct commit *c = ctx->commits.list[i];
1564 timestamp_t offset = commit_graph_data_at(c)->generation - c->date;
1565 if (offset > GENERATION_NUMBER_V2_OFFSET_MAX)
1566 ctx->num_generation_data_overflows++;
1568 stop_progress(&ctx->progress);
1571 static void trace2_bloom_filter_write_statistics(struct write_commit_graph_context *ctx)
1573 trace2_data_intmax("commit-graph", ctx->r, "filter-computed",
1574 ctx->count_bloom_filter_computed);
1575 trace2_data_intmax("commit-graph", ctx->r, "filter-not-computed",
1576 ctx->count_bloom_filter_not_computed);
1577 trace2_data_intmax("commit-graph", ctx->r, "filter-trunc-empty",
1578 ctx->count_bloom_filter_trunc_empty);
1579 trace2_data_intmax("commit-graph", ctx->r, "filter-trunc-large",
1580 ctx->count_bloom_filter_trunc_large);
1583 static void compute_bloom_filters(struct write_commit_graph_context *ctx)
1585 int i;
1586 struct progress *progress = NULL;
1587 struct commit **sorted_commits;
1588 int max_new_filters;
1590 init_bloom_filters();
1592 if (ctx->report_progress)
1593 progress = start_delayed_progress(
1594 _("Computing commit changed paths Bloom filters"),
1595 ctx->commits.nr);
1597 DUP_ARRAY(sorted_commits, ctx->commits.list, ctx->commits.nr);
1599 if (ctx->order_by_pack)
1600 QSORT(sorted_commits, ctx->commits.nr, commit_pos_cmp);
1601 else
1602 QSORT(sorted_commits, ctx->commits.nr, commit_gen_cmp);
1604 max_new_filters = ctx->opts && ctx->opts->max_new_filters >= 0 ?
1605 ctx->opts->max_new_filters : ctx->commits.nr;
1607 for (i = 0; i < ctx->commits.nr; i++) {
1608 enum bloom_filter_computed computed = 0;
1609 struct commit *c = sorted_commits[i];
1610 struct bloom_filter *filter = get_or_compute_bloom_filter(
1611 ctx->r,
1613 ctx->count_bloom_filter_computed < max_new_filters,
1614 ctx->bloom_settings,
1615 &computed);
1616 if (computed & BLOOM_COMPUTED) {
1617 ctx->count_bloom_filter_computed++;
1618 if (computed & BLOOM_TRUNC_EMPTY)
1619 ctx->count_bloom_filter_trunc_empty++;
1620 if (computed & BLOOM_TRUNC_LARGE)
1621 ctx->count_bloom_filter_trunc_large++;
1622 } else if (computed & BLOOM_NOT_COMPUTED)
1623 ctx->count_bloom_filter_not_computed++;
1624 ctx->total_bloom_filter_data_size += filter
1625 ? sizeof(unsigned char) * filter->len : 0;
1626 display_progress(progress, i + 1);
1629 if (trace2_is_enabled())
1630 trace2_bloom_filter_write_statistics(ctx);
1632 free(sorted_commits);
1633 stop_progress(&progress);
1636 struct refs_cb_data {
1637 struct oidset *commits;
1638 struct progress *progress;
1641 static int add_ref_to_set(const char *refname UNUSED,
1642 const struct object_id *oid,
1643 int flags UNUSED, void *cb_data)
1645 struct object_id peeled;
1646 struct refs_cb_data *data = (struct refs_cb_data *)cb_data;
1648 if (!peel_iterated_oid(oid, &peeled))
1649 oid = &peeled;
1650 if (oid_object_info(the_repository, oid, NULL) == OBJ_COMMIT)
1651 oidset_insert(data->commits, oid);
1653 display_progress(data->progress, oidset_size(data->commits));
1655 return 0;
1658 int write_commit_graph_reachable(struct object_directory *odb,
1659 enum commit_graph_write_flags flags,
1660 const struct commit_graph_opts *opts)
1662 struct oidset commits = OIDSET_INIT;
1663 struct refs_cb_data data;
1664 int result;
1666 memset(&data, 0, sizeof(data));
1667 data.commits = &commits;
1668 if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
1669 data.progress = start_delayed_progress(
1670 _("Collecting referenced commits"), 0);
1672 for_each_ref(add_ref_to_set, &data);
1674 stop_progress(&data.progress);
1676 result = write_commit_graph(odb, NULL, &commits,
1677 flags, opts);
1679 oidset_clear(&commits);
1680 return result;
1683 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
1684 const struct string_list *pack_indexes)
1686 uint32_t i;
1687 struct strbuf progress_title = STRBUF_INIT;
1688 struct strbuf packname = STRBUF_INIT;
1689 int dirlen;
1690 int ret = 0;
1692 strbuf_addf(&packname, "%s/pack/", ctx->odb->path);
1693 dirlen = packname.len;
1694 if (ctx->report_progress) {
1695 strbuf_addf(&progress_title,
1696 Q_("Finding commits for commit graph in %"PRIuMAX" pack",
1697 "Finding commits for commit graph in %"PRIuMAX" packs",
1698 pack_indexes->nr),
1699 (uintmax_t)pack_indexes->nr);
1700 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1701 ctx->progress_done = 0;
1703 for (i = 0; i < pack_indexes->nr; i++) {
1704 struct packed_git *p;
1705 strbuf_setlen(&packname, dirlen);
1706 strbuf_addstr(&packname, pack_indexes->items[i].string);
1707 p = add_packed_git(packname.buf, packname.len, 1);
1708 if (!p) {
1709 ret = error(_("error adding pack %s"), packname.buf);
1710 goto cleanup;
1712 if (open_pack_index(p)) {
1713 ret = error(_("error opening index for %s"), packname.buf);
1714 goto cleanup;
1716 for_each_object_in_pack(p, add_packed_commits, ctx,
1717 FOR_EACH_OBJECT_PACK_ORDER);
1718 close_pack(p);
1719 free(p);
1722 cleanup:
1723 stop_progress(&ctx->progress);
1724 strbuf_release(&progress_title);
1725 strbuf_release(&packname);
1727 return ret;
1730 static int fill_oids_from_commits(struct write_commit_graph_context *ctx,
1731 struct oidset *commits)
1733 struct oidset_iter iter;
1734 struct object_id *oid;
1736 if (!oidset_size(commits))
1737 return 0;
1739 oidset_iter_init(commits, &iter);
1740 while ((oid = oidset_iter_next(&iter))) {
1741 oid_array_append(&ctx->oids, oid);
1744 return 0;
1747 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1749 if (ctx->report_progress)
1750 ctx->progress = start_delayed_progress(
1751 _("Finding commits for commit graph among packed objects"),
1752 ctx->approx_nr_objects);
1753 for_each_packed_object(add_packed_commits, ctx,
1754 FOR_EACH_OBJECT_PACK_ORDER);
1755 if (ctx->progress_done < ctx->approx_nr_objects)
1756 display_progress(ctx->progress, ctx->approx_nr_objects);
1757 stop_progress(&ctx->progress);
1760 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1762 uint32_t i;
1763 enum commit_graph_split_flags flags = ctx->opts ?
1764 ctx->opts->split_flags : COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1766 ctx->num_extra_edges = 0;
1767 if (ctx->report_progress)
1768 ctx->progress = start_delayed_progress(
1769 _("Finding extra edges in commit graph"),
1770 ctx->oids.nr);
1771 oid_array_sort(&ctx->oids);
1772 for (i = 0; i < ctx->oids.nr; i = oid_array_next_unique(&ctx->oids, i)) {
1773 unsigned int num_parents;
1775 display_progress(ctx->progress, i + 1);
1777 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + 1, ctx->commits.alloc);
1778 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.oid[i]);
1780 if (ctx->split && flags != COMMIT_GRAPH_SPLIT_REPLACE &&
1781 commit_graph_position(ctx->commits.list[ctx->commits.nr]) != COMMIT_NOT_FROM_GRAPH)
1782 continue;
1784 if (ctx->split && flags == COMMIT_GRAPH_SPLIT_REPLACE)
1785 repo_parse_commit(ctx->r, ctx->commits.list[ctx->commits.nr]);
1786 else
1787 repo_parse_commit_no_graph(ctx->r, ctx->commits.list[ctx->commits.nr]);
1789 num_parents = commit_list_count(ctx->commits.list[ctx->commits.nr]->parents);
1790 if (num_parents > 2)
1791 ctx->num_extra_edges += num_parents - 1;
1793 ctx->commits.nr++;
1795 stop_progress(&ctx->progress);
1798 static int write_graph_chunk_base_1(struct hashfile *f,
1799 struct commit_graph *g)
1801 int num = 0;
1803 if (!g)
1804 return 0;
1806 num = write_graph_chunk_base_1(f, g->base_graph);
1807 hashwrite(f, g->oid.hash, the_hash_algo->rawsz);
1808 return num + 1;
1811 static int write_graph_chunk_base(struct hashfile *f,
1812 void *data)
1814 struct write_commit_graph_context *ctx = data;
1815 int num = write_graph_chunk_base_1(f, ctx->new_base_graph);
1817 if (num != ctx->num_commit_graphs_after - 1) {
1818 error(_("failed to write correct number of base graph ids"));
1819 return -1;
1822 return 0;
1825 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1827 uint32_t i;
1828 int fd;
1829 struct hashfile *f;
1830 struct lock_file lk = LOCK_INIT;
1831 const unsigned hashsz = the_hash_algo->rawsz;
1832 struct strbuf progress_title = STRBUF_INIT;
1833 struct chunkfile *cf;
1834 unsigned char file_hash[GIT_MAX_RAWSZ];
1836 if (ctx->split) {
1837 struct strbuf tmp_file = STRBUF_INIT;
1839 strbuf_addf(&tmp_file,
1840 "%s/info/commit-graphs/tmp_graph_XXXXXX",
1841 ctx->odb->path);
1842 ctx->graph_name = strbuf_detach(&tmp_file, NULL);
1843 } else {
1844 ctx->graph_name = get_commit_graph_filename(ctx->odb);
1847 if (safe_create_leading_directories(ctx->graph_name)) {
1848 UNLEAK(ctx->graph_name);
1849 error(_("unable to create leading directories of %s"),
1850 ctx->graph_name);
1851 return -1;
1854 if (ctx->split) {
1855 char *lock_name = get_commit_graph_chain_filename(ctx->odb);
1857 hold_lock_file_for_update_mode(&lk, lock_name,
1858 LOCK_DIE_ON_ERROR, 0444);
1859 free(lock_name);
1861 fd = git_mkstemp_mode(ctx->graph_name, 0444);
1862 if (fd < 0) {
1863 error(_("unable to create temporary graph layer"));
1864 return -1;
1867 if (adjust_shared_perm(ctx->graph_name)) {
1868 error(_("unable to adjust shared permissions for '%s'"),
1869 ctx->graph_name);
1870 return -1;
1873 f = hashfd(fd, ctx->graph_name);
1874 } else {
1875 hold_lock_file_for_update_mode(&lk, ctx->graph_name,
1876 LOCK_DIE_ON_ERROR, 0444);
1877 fd = get_lock_file_fd(&lk);
1878 f = hashfd(fd, get_lock_file_path(&lk));
1881 cf = init_chunkfile(f);
1883 add_chunk(cf, GRAPH_CHUNKID_OIDFANOUT, GRAPH_FANOUT_SIZE,
1884 write_graph_chunk_fanout);
1885 add_chunk(cf, GRAPH_CHUNKID_OIDLOOKUP, hashsz * ctx->commits.nr,
1886 write_graph_chunk_oids);
1887 add_chunk(cf, GRAPH_CHUNKID_DATA, (hashsz + 16) * ctx->commits.nr,
1888 write_graph_chunk_data);
1890 if (ctx->write_generation_data)
1891 add_chunk(cf, GRAPH_CHUNKID_GENERATION_DATA,
1892 sizeof(uint32_t) * ctx->commits.nr,
1893 write_graph_chunk_generation_data);
1894 if (ctx->num_generation_data_overflows)
1895 add_chunk(cf, GRAPH_CHUNKID_GENERATION_DATA_OVERFLOW,
1896 sizeof(timestamp_t) * ctx->num_generation_data_overflows,
1897 write_graph_chunk_generation_data_overflow);
1898 if (ctx->num_extra_edges)
1899 add_chunk(cf, GRAPH_CHUNKID_EXTRAEDGES,
1900 4 * ctx->num_extra_edges,
1901 write_graph_chunk_extra_edges);
1902 if (ctx->changed_paths) {
1903 add_chunk(cf, GRAPH_CHUNKID_BLOOMINDEXES,
1904 sizeof(uint32_t) * ctx->commits.nr,
1905 write_graph_chunk_bloom_indexes);
1906 add_chunk(cf, GRAPH_CHUNKID_BLOOMDATA,
1907 sizeof(uint32_t) * 3
1908 + ctx->total_bloom_filter_data_size,
1909 write_graph_chunk_bloom_data);
1911 if (ctx->num_commit_graphs_after > 1)
1912 add_chunk(cf, GRAPH_CHUNKID_BASE,
1913 hashsz * (ctx->num_commit_graphs_after - 1),
1914 write_graph_chunk_base);
1916 hashwrite_be32(f, GRAPH_SIGNATURE);
1918 hashwrite_u8(f, GRAPH_VERSION);
1919 hashwrite_u8(f, oid_version(the_hash_algo));
1920 hashwrite_u8(f, get_num_chunks(cf));
1921 hashwrite_u8(f, ctx->num_commit_graphs_after - 1);
1923 if (ctx->report_progress) {
1924 strbuf_addf(&progress_title,
1925 Q_("Writing out commit graph in %d pass",
1926 "Writing out commit graph in %d passes",
1927 get_num_chunks(cf)),
1928 get_num_chunks(cf));
1929 ctx->progress = start_delayed_progress(
1930 progress_title.buf,
1931 get_num_chunks(cf) * ctx->commits.nr);
1934 write_chunkfile(cf, ctx);
1936 stop_progress(&ctx->progress);
1937 strbuf_release(&progress_title);
1939 if (ctx->split && ctx->base_graph_name && ctx->num_commit_graphs_after > 1) {
1940 char *new_base_hash = xstrdup(oid_to_hex(&ctx->new_base_graph->oid));
1941 char *new_base_name = get_split_graph_filename(ctx->new_base_graph->odb, new_base_hash);
1943 free(ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2]);
1944 free(ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2]);
1945 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2] = new_base_name;
1946 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2] = new_base_hash;
1949 close_commit_graph(ctx->r->objects);
1950 finalize_hashfile(f, file_hash, FSYNC_COMPONENT_COMMIT_GRAPH,
1951 CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1952 free_chunkfile(cf);
1954 if (ctx->split) {
1955 FILE *chainf = fdopen_lock_file(&lk, "w");
1956 char *final_graph_name;
1957 int result;
1959 close(fd);
1961 if (!chainf) {
1962 error(_("unable to open commit-graph chain file"));
1963 return -1;
1966 if (ctx->base_graph_name) {
1967 const char *dest;
1968 int idx = ctx->num_commit_graphs_after - 1;
1969 if (ctx->num_commit_graphs_after > 1)
1970 idx--;
1972 dest = ctx->commit_graph_filenames_after[idx];
1974 if (strcmp(ctx->base_graph_name, dest)) {
1975 result = rename(ctx->base_graph_name, dest);
1977 if (result) {
1978 error(_("failed to rename base commit-graph file"));
1979 return -1;
1982 } else {
1983 char *graph_name = get_commit_graph_filename(ctx->odb);
1984 unlink(graph_name);
1985 free(graph_name);
1988 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1] = xstrdup(hash_to_hex(file_hash));
1989 final_graph_name = get_split_graph_filename(ctx->odb,
1990 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1]);
1991 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 1] = final_graph_name;
1993 result = rename(ctx->graph_name, final_graph_name);
1995 for (i = 0; i < ctx->num_commit_graphs_after; i++)
1996 fprintf(get_lock_file_fp(&lk), "%s\n", ctx->commit_graph_hash_after[i]);
1998 if (result) {
1999 error(_("failed to rename temporary commit-graph file"));
2000 return -1;
2004 commit_lock_file(&lk);
2006 return 0;
2009 static void split_graph_merge_strategy(struct write_commit_graph_context *ctx)
2011 struct commit_graph *g;
2012 uint32_t num_commits;
2013 enum commit_graph_split_flags flags = COMMIT_GRAPH_SPLIT_UNSPECIFIED;
2014 uint32_t i;
2016 int max_commits = 0;
2017 int size_mult = 2;
2019 if (ctx->opts) {
2020 max_commits = ctx->opts->max_commits;
2022 if (ctx->opts->size_multiple)
2023 size_mult = ctx->opts->size_multiple;
2025 flags = ctx->opts->split_flags;
2028 g = ctx->r->objects->commit_graph;
2029 num_commits = ctx->commits.nr;
2030 if (flags == COMMIT_GRAPH_SPLIT_REPLACE)
2031 ctx->num_commit_graphs_after = 1;
2032 else
2033 ctx->num_commit_graphs_after = ctx->num_commit_graphs_before + 1;
2035 if (flags != COMMIT_GRAPH_SPLIT_MERGE_PROHIBITED &&
2036 flags != COMMIT_GRAPH_SPLIT_REPLACE) {
2037 while (g && (g->num_commits <= size_mult * num_commits ||
2038 (max_commits && num_commits > max_commits))) {
2039 if (g->odb != ctx->odb)
2040 break;
2042 num_commits += g->num_commits;
2043 g = g->base_graph;
2045 ctx->num_commit_graphs_after--;
2049 if (flags != COMMIT_GRAPH_SPLIT_REPLACE)
2050 ctx->new_base_graph = g;
2051 else if (ctx->num_commit_graphs_after != 1)
2052 BUG("split_graph_merge_strategy: num_commit_graphs_after "
2053 "should be 1 with --split=replace");
2055 if (ctx->num_commit_graphs_after == 2) {
2056 char *old_graph_name = get_commit_graph_filename(g->odb);
2058 if (!strcmp(g->filename, old_graph_name) &&
2059 g->odb != ctx->odb) {
2060 ctx->num_commit_graphs_after = 1;
2061 ctx->new_base_graph = NULL;
2064 free(old_graph_name);
2067 CALLOC_ARRAY(ctx->commit_graph_filenames_after, ctx->num_commit_graphs_after);
2068 CALLOC_ARRAY(ctx->commit_graph_hash_after, ctx->num_commit_graphs_after);
2070 for (i = 0; i < ctx->num_commit_graphs_after &&
2071 i < ctx->num_commit_graphs_before; i++)
2072 ctx->commit_graph_filenames_after[i] = xstrdup(ctx->commit_graph_filenames_before[i]);
2074 i = ctx->num_commit_graphs_before - 1;
2075 g = ctx->r->objects->commit_graph;
2077 while (g) {
2078 if (i < ctx->num_commit_graphs_after)
2079 ctx->commit_graph_hash_after[i] = xstrdup(oid_to_hex(&g->oid));
2082 * If the topmost remaining layer has generation data chunk, the
2083 * resultant layer also has generation data chunk.
2085 if (i == ctx->num_commit_graphs_after - 2)
2086 ctx->write_generation_data = !!g->chunk_generation_data;
2088 i--;
2089 g = g->base_graph;
2093 static void merge_commit_graph(struct write_commit_graph_context *ctx,
2094 struct commit_graph *g)
2096 uint32_t i;
2097 uint32_t offset = g->num_commits_in_base;
2099 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + g->num_commits, ctx->commits.alloc);
2101 for (i = 0; i < g->num_commits; i++) {
2102 struct object_id oid;
2103 struct commit *result;
2105 display_progress(ctx->progress, i + 1);
2107 load_oid_from_graph(g, i + offset, &oid);
2109 /* only add commits if they still exist in the repo */
2110 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
2112 if (result) {
2113 ctx->commits.list[ctx->commits.nr] = result;
2114 ctx->commits.nr++;
2119 static int commit_compare(const void *_a, const void *_b)
2121 const struct commit *a = *(const struct commit **)_a;
2122 const struct commit *b = *(const struct commit **)_b;
2123 return oidcmp(&a->object.oid, &b->object.oid);
2126 static void sort_and_scan_merged_commits(struct write_commit_graph_context *ctx)
2128 uint32_t i, dedup_i = 0;
2130 if (ctx->report_progress)
2131 ctx->progress = start_delayed_progress(
2132 _("Scanning merged commits"),
2133 ctx->commits.nr);
2135 QSORT(ctx->commits.list, ctx->commits.nr, commit_compare);
2137 ctx->num_extra_edges = 0;
2138 for (i = 0; i < ctx->commits.nr; i++) {
2139 display_progress(ctx->progress, i + 1);
2141 if (i && oideq(&ctx->commits.list[i - 1]->object.oid,
2142 &ctx->commits.list[i]->object.oid)) {
2144 * Silently ignore duplicates. These were likely
2145 * created due to a commit appearing in multiple
2146 * layers of the chain, which is unexpected but
2147 * not invalid. We should make sure there is a
2148 * unique copy in the new layer.
2150 } else {
2151 unsigned int num_parents;
2153 ctx->commits.list[dedup_i] = ctx->commits.list[i];
2154 dedup_i++;
2156 num_parents = commit_list_count(ctx->commits.list[i]->parents);
2157 if (num_parents > 2)
2158 ctx->num_extra_edges += num_parents - 1;
2162 ctx->commits.nr = dedup_i;
2164 stop_progress(&ctx->progress);
2167 static void merge_commit_graphs(struct write_commit_graph_context *ctx)
2169 struct commit_graph *g = ctx->r->objects->commit_graph;
2170 uint32_t current_graph_number = ctx->num_commit_graphs_before;
2172 while (g && current_graph_number >= ctx->num_commit_graphs_after) {
2173 current_graph_number--;
2175 if (ctx->report_progress)
2176 ctx->progress = start_delayed_progress(_("Merging commit-graph"), 0);
2178 merge_commit_graph(ctx, g);
2179 stop_progress(&ctx->progress);
2181 g = g->base_graph;
2184 if (g) {
2185 ctx->new_base_graph = g;
2186 ctx->new_num_commits_in_base = g->num_commits + g->num_commits_in_base;
2189 if (ctx->new_base_graph)
2190 ctx->base_graph_name = xstrdup(ctx->new_base_graph->filename);
2192 sort_and_scan_merged_commits(ctx);
2195 static void mark_commit_graphs(struct write_commit_graph_context *ctx)
2197 uint32_t i;
2198 time_t now = time(NULL);
2200 for (i = ctx->num_commit_graphs_after - 1; i < ctx->num_commit_graphs_before; i++) {
2201 struct stat st;
2202 struct utimbuf updated_time;
2204 if (stat(ctx->commit_graph_filenames_before[i], &st) < 0)
2205 continue;
2207 updated_time.actime = st.st_atime;
2208 updated_time.modtime = now;
2209 utime(ctx->commit_graph_filenames_before[i], &updated_time);
2213 static void expire_commit_graphs(struct write_commit_graph_context *ctx)
2215 struct strbuf path = STRBUF_INIT;
2216 DIR *dir;
2217 struct dirent *de;
2218 size_t dirnamelen;
2219 timestamp_t expire_time = time(NULL);
2221 if (ctx->opts && ctx->opts->expire_time)
2222 expire_time = ctx->opts->expire_time;
2223 if (!ctx->split) {
2224 char *chain_file_name = get_commit_graph_chain_filename(ctx->odb);
2225 unlink(chain_file_name);
2226 free(chain_file_name);
2227 ctx->num_commit_graphs_after = 0;
2230 strbuf_addstr(&path, ctx->odb->path);
2231 strbuf_addstr(&path, "/info/commit-graphs");
2232 dir = opendir(path.buf);
2234 if (!dir)
2235 goto out;
2237 strbuf_addch(&path, '/');
2238 dirnamelen = path.len;
2239 while ((de = readdir(dir)) != NULL) {
2240 struct stat st;
2241 uint32_t i, found = 0;
2243 strbuf_setlen(&path, dirnamelen);
2244 strbuf_addstr(&path, de->d_name);
2246 if (stat(path.buf, &st) < 0)
2247 continue;
2249 if (st.st_mtime > expire_time)
2250 continue;
2251 if (path.len < 6 || strcmp(path.buf + path.len - 6, ".graph"))
2252 continue;
2254 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
2255 if (!strcmp(ctx->commit_graph_filenames_after[i],
2256 path.buf)) {
2257 found = 1;
2258 break;
2262 if (!found)
2263 unlink(path.buf);
2266 out:
2267 if(dir)
2268 closedir(dir);
2269 strbuf_release(&path);
2272 int write_commit_graph(struct object_directory *odb,
2273 const struct string_list *const pack_indexes,
2274 struct oidset *commits,
2275 enum commit_graph_write_flags flags,
2276 const struct commit_graph_opts *opts)
2278 struct repository *r = the_repository;
2279 struct write_commit_graph_context *ctx;
2280 uint32_t i;
2281 int res = 0;
2282 int replace = 0;
2283 struct bloom_filter_settings bloom_settings = DEFAULT_BLOOM_FILTER_SETTINGS;
2284 struct topo_level_slab topo_levels;
2286 prepare_repo_settings(r);
2287 if (!r->settings.core_commit_graph) {
2288 warning(_("attempting to write a commit-graph, but 'core.commitGraph' is disabled"));
2289 return 0;
2291 if (!commit_graph_compatible(r))
2292 return 0;
2294 CALLOC_ARRAY(ctx, 1);
2295 ctx->r = r;
2296 ctx->odb = odb;
2297 ctx->append = flags & COMMIT_GRAPH_WRITE_APPEND ? 1 : 0;
2298 ctx->report_progress = flags & COMMIT_GRAPH_WRITE_PROGRESS ? 1 : 0;
2299 ctx->split = flags & COMMIT_GRAPH_WRITE_SPLIT ? 1 : 0;
2300 ctx->opts = opts;
2301 ctx->total_bloom_filter_data_size = 0;
2302 ctx->write_generation_data = (get_configured_generation_version(r) == 2);
2303 ctx->num_generation_data_overflows = 0;
2305 bloom_settings.bits_per_entry = git_env_ulong("GIT_TEST_BLOOM_SETTINGS_BITS_PER_ENTRY",
2306 bloom_settings.bits_per_entry);
2307 bloom_settings.num_hashes = git_env_ulong("GIT_TEST_BLOOM_SETTINGS_NUM_HASHES",
2308 bloom_settings.num_hashes);
2309 bloom_settings.max_changed_paths = git_env_ulong("GIT_TEST_BLOOM_SETTINGS_MAX_CHANGED_PATHS",
2310 bloom_settings.max_changed_paths);
2311 ctx->bloom_settings = &bloom_settings;
2313 init_topo_level_slab(&topo_levels);
2314 ctx->topo_levels = &topo_levels;
2316 prepare_commit_graph(ctx->r);
2317 if (ctx->r->objects->commit_graph) {
2318 struct commit_graph *g = ctx->r->objects->commit_graph;
2320 while (g) {
2321 g->topo_levels = &topo_levels;
2322 g = g->base_graph;
2326 if (flags & COMMIT_GRAPH_WRITE_BLOOM_FILTERS)
2327 ctx->changed_paths = 1;
2328 if (!(flags & COMMIT_GRAPH_NO_WRITE_BLOOM_FILTERS)) {
2329 struct commit_graph *g;
2331 g = ctx->r->objects->commit_graph;
2333 /* We have changed-paths already. Keep them in the next graph */
2334 if (g && g->chunk_bloom_data) {
2335 ctx->changed_paths = 1;
2336 ctx->bloom_settings = g->bloom_filter_settings;
2340 if (ctx->split) {
2341 struct commit_graph *g = ctx->r->objects->commit_graph;
2343 while (g) {
2344 ctx->num_commit_graphs_before++;
2345 g = g->base_graph;
2348 if (ctx->num_commit_graphs_before) {
2349 ALLOC_ARRAY(ctx->commit_graph_filenames_before, ctx->num_commit_graphs_before);
2350 i = ctx->num_commit_graphs_before;
2351 g = ctx->r->objects->commit_graph;
2353 while (g) {
2354 ctx->commit_graph_filenames_before[--i] = xstrdup(g->filename);
2355 g = g->base_graph;
2359 if (ctx->opts)
2360 replace = ctx->opts->split_flags & COMMIT_GRAPH_SPLIT_REPLACE;
2363 ctx->approx_nr_objects = approximate_object_count();
2365 if (ctx->append && ctx->r->objects->commit_graph) {
2366 struct commit_graph *g = ctx->r->objects->commit_graph;
2367 for (i = 0; i < g->num_commits; i++) {
2368 struct object_id oid;
2369 oidread(&oid, g->chunk_oid_lookup + g->hash_len * i);
2370 oid_array_append(&ctx->oids, &oid);
2374 if (pack_indexes) {
2375 ctx->order_by_pack = 1;
2376 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
2377 goto cleanup;
2380 if (commits) {
2381 if ((res = fill_oids_from_commits(ctx, commits)))
2382 goto cleanup;
2385 if (!pack_indexes && !commits) {
2386 ctx->order_by_pack = 1;
2387 fill_oids_from_all_packs(ctx);
2390 close_reachable(ctx);
2392 copy_oids_to_commits(ctx);
2394 if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
2395 error(_("too many commits to write graph"));
2396 res = -1;
2397 goto cleanup;
2400 if (!ctx->commits.nr && !replace)
2401 goto cleanup;
2403 if (ctx->split) {
2404 split_graph_merge_strategy(ctx);
2406 if (!replace)
2407 merge_commit_graphs(ctx);
2408 } else
2409 ctx->num_commit_graphs_after = 1;
2411 ctx->trust_generation_numbers = validate_mixed_generation_chain(ctx->r->objects->commit_graph);
2413 compute_topological_levels(ctx);
2414 if (ctx->write_generation_data)
2415 compute_generation_numbers(ctx);
2417 if (ctx->changed_paths)
2418 compute_bloom_filters(ctx);
2420 res = write_commit_graph_file(ctx);
2422 if (ctx->split)
2423 mark_commit_graphs(ctx);
2425 expire_commit_graphs(ctx);
2427 cleanup:
2428 free(ctx->graph_name);
2429 free(ctx->commits.list);
2430 oid_array_clear(&ctx->oids);
2431 clear_topo_level_slab(&topo_levels);
2433 if (ctx->commit_graph_filenames_after) {
2434 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
2435 free(ctx->commit_graph_filenames_after[i]);
2436 free(ctx->commit_graph_hash_after[i]);
2439 for (i = 0; i < ctx->num_commit_graphs_before; i++)
2440 free(ctx->commit_graph_filenames_before[i]);
2442 free(ctx->commit_graph_filenames_after);
2443 free(ctx->commit_graph_filenames_before);
2444 free(ctx->commit_graph_hash_after);
2447 free(ctx);
2449 return res;
2452 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
2453 static int verify_commit_graph_error;
2455 __attribute__((format (printf, 1, 2)))
2456 static void graph_report(const char *fmt, ...)
2458 va_list ap;
2460 verify_commit_graph_error = 1;
2461 va_start(ap, fmt);
2462 vfprintf(stderr, fmt, ap);
2463 fprintf(stderr, "\n");
2464 va_end(ap);
2467 #define GENERATION_ZERO_EXISTS 1
2468 #define GENERATION_NUMBER_EXISTS 2
2470 static int commit_graph_checksum_valid(struct commit_graph *g)
2472 return hashfile_checksum_valid(g->data, g->data_len);
2475 int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
2477 uint32_t i, cur_fanout_pos = 0;
2478 struct object_id prev_oid, cur_oid;
2479 int generation_zero = 0;
2480 struct progress *progress = NULL;
2481 int local_error = 0;
2483 if (!g) {
2484 graph_report("no commit-graph file loaded");
2485 return 1;
2488 verify_commit_graph_error = verify_commit_graph_lite(g);
2489 if (verify_commit_graph_error)
2490 return verify_commit_graph_error;
2492 if (!commit_graph_checksum_valid(g)) {
2493 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
2494 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
2497 for (i = 0; i < g->num_commits; i++) {
2498 struct commit *graph_commit;
2500 oidread(&cur_oid, g->chunk_oid_lookup + g->hash_len * i);
2502 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
2503 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
2504 oid_to_hex(&prev_oid),
2505 oid_to_hex(&cur_oid));
2507 oidcpy(&prev_oid, &cur_oid);
2509 while (cur_oid.hash[0] > cur_fanout_pos) {
2510 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2512 if (i != fanout_value)
2513 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2514 cur_fanout_pos, fanout_value, i);
2515 cur_fanout_pos++;
2518 graph_commit = lookup_commit(r, &cur_oid);
2519 if (!parse_commit_in_graph_one(r, g, graph_commit))
2520 graph_report(_("failed to parse commit %s from commit-graph"),
2521 oid_to_hex(&cur_oid));
2524 while (cur_fanout_pos < 256) {
2525 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2527 if (g->num_commits != fanout_value)
2528 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2529 cur_fanout_pos, fanout_value, i);
2531 cur_fanout_pos++;
2534 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
2535 return verify_commit_graph_error;
2537 if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
2538 progress = start_progress(_("Verifying commits in commit graph"),
2539 g->num_commits);
2541 for (i = 0; i < g->num_commits; i++) {
2542 struct commit *graph_commit, *odb_commit;
2543 struct commit_list *graph_parents, *odb_parents;
2544 timestamp_t max_generation = 0;
2545 timestamp_t generation;
2547 display_progress(progress, i + 1);
2548 oidread(&cur_oid, g->chunk_oid_lookup + g->hash_len * i);
2550 graph_commit = lookup_commit(r, &cur_oid);
2551 odb_commit = (struct commit *)create_object(r, &cur_oid, alloc_commit_node(r));
2552 if (parse_commit_internal(odb_commit, 0, 0)) {
2553 graph_report(_("failed to parse commit %s from object database for commit-graph"),
2554 oid_to_hex(&cur_oid));
2555 continue;
2558 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
2559 get_commit_tree_oid(odb_commit)))
2560 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2561 oid_to_hex(&cur_oid),
2562 oid_to_hex(get_commit_tree_oid(graph_commit)),
2563 oid_to_hex(get_commit_tree_oid(odb_commit)));
2565 graph_parents = graph_commit->parents;
2566 odb_parents = odb_commit->parents;
2568 while (graph_parents) {
2569 if (!odb_parents) {
2570 graph_report(_("commit-graph parent list for commit %s is too long"),
2571 oid_to_hex(&cur_oid));
2572 break;
2575 /* parse parent in case it is in a base graph */
2576 parse_commit_in_graph_one(r, g, graph_parents->item);
2578 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
2579 graph_report(_("commit-graph parent for %s is %s != %s"),
2580 oid_to_hex(&cur_oid),
2581 oid_to_hex(&graph_parents->item->object.oid),
2582 oid_to_hex(&odb_parents->item->object.oid));
2584 generation = commit_graph_generation(graph_parents->item);
2585 if (generation > max_generation)
2586 max_generation = generation;
2588 graph_parents = graph_parents->next;
2589 odb_parents = odb_parents->next;
2592 if (odb_parents)
2593 graph_report(_("commit-graph parent list for commit %s terminates early"),
2594 oid_to_hex(&cur_oid));
2596 if (!commit_graph_generation(graph_commit)) {
2597 if (generation_zero == GENERATION_NUMBER_EXISTS)
2598 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2599 oid_to_hex(&cur_oid));
2600 generation_zero = GENERATION_ZERO_EXISTS;
2601 } else if (generation_zero == GENERATION_ZERO_EXISTS)
2602 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2603 oid_to_hex(&cur_oid));
2605 if (generation_zero == GENERATION_ZERO_EXISTS)
2606 continue;
2609 * If we are using topological level and one of our parents has
2610 * generation GENERATION_NUMBER_V1_MAX, then our generation is
2611 * also GENERATION_NUMBER_V1_MAX. Decrement to avoid extra logic
2612 * in the following condition.
2614 if (!g->read_generation_data && max_generation == GENERATION_NUMBER_V1_MAX)
2615 max_generation--;
2617 generation = commit_graph_generation(graph_commit);
2618 if (generation < max_generation + 1)
2619 graph_report(_("commit-graph generation for commit %s is %"PRItime" < %"PRItime),
2620 oid_to_hex(&cur_oid),
2621 generation,
2622 max_generation + 1);
2624 if (graph_commit->date != odb_commit->date)
2625 graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
2626 oid_to_hex(&cur_oid),
2627 graph_commit->date,
2628 odb_commit->date);
2630 stop_progress(&progress);
2632 local_error = verify_commit_graph_error;
2634 if (!(flags & COMMIT_GRAPH_VERIFY_SHALLOW) && g->base_graph)
2635 local_error |= verify_commit_graph(r, g->base_graph, flags);
2637 return local_error;
2640 void free_commit_graph(struct commit_graph *g)
2642 if (!g)
2643 return;
2644 if (g->data) {
2645 munmap((void *)g->data, g->data_len);
2646 g->data = NULL;
2648 free(g->filename);
2649 free(g->bloom_filter_settings);
2650 free(g);
2653 void disable_commit_graph(struct repository *r)
2655 r->commit_graph_disabled = 1;