Merge branch 'ms/fetch-follow-tag-optim'
[git.git] / commit-graph.c
blob19cb1fecfa34ba5945bb6baab6f4c0a4b096cc78
1 #include "cache.h"
2 #include "config.h"
3 #include "dir.h"
4 #include "git-compat-util.h"
5 #include "lockfile.h"
6 #include "pack.h"
7 #include "packfile.h"
8 #include "commit.h"
9 #include "object.h"
10 #include "refs.h"
11 #include "revision.h"
12 #include "sha1-lookup.h"
13 #include "commit-graph.h"
14 #include "object-store.h"
15 #include "alloc.h"
16 #include "hashmap.h"
17 #include "replace-object.h"
18 #include "progress.h"
20 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
21 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
22 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
23 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
24 #define GRAPH_CHUNKID_EXTRAEDGES 0x45444745 /* "EDGE" */
25 #define GRAPH_CHUNKID_BASE 0x42415345 /* "BASE" */
27 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
29 #define GRAPH_VERSION_1 0x1
30 #define GRAPH_VERSION GRAPH_VERSION_1
32 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
33 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
34 #define GRAPH_PARENT_NONE 0x70000000
36 #define GRAPH_LAST_EDGE 0x80000000
38 #define GRAPH_HEADER_SIZE 8
39 #define GRAPH_FANOUT_SIZE (4 * 256)
40 #define GRAPH_CHUNKLOOKUP_WIDTH 12
41 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
42 + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
44 char *get_commit_graph_filename(const char *obj_dir)
46 char *filename = xstrfmt("%s/info/commit-graph", obj_dir);
47 char *normalized = xmalloc(strlen(filename) + 1);
48 normalize_path_copy(normalized, filename);
49 free(filename);
50 return normalized;
53 static char *get_split_graph_filename(const char *obj_dir,
54 const char *oid_hex)
56 char *filename = xstrfmt("%s/info/commit-graphs/graph-%s.graph",
57 obj_dir,
58 oid_hex);
59 char *normalized = xmalloc(strlen(filename) + 1);
60 normalize_path_copy(normalized, filename);
61 free(filename);
62 return normalized;
65 static char *get_chain_filename(const char *obj_dir)
67 return xstrfmt("%s/info/commit-graphs/commit-graph-chain", obj_dir);
70 static uint8_t oid_version(void)
72 return 1;
75 static struct commit_graph *alloc_commit_graph(void)
77 struct commit_graph *g = xcalloc(1, sizeof(*g));
78 g->graph_fd = -1;
80 return g;
83 extern int read_replace_refs;
85 static int commit_graph_compatible(struct repository *r)
87 if (!r->gitdir)
88 return 0;
90 if (read_replace_refs) {
91 prepare_replace_object(r);
92 if (hashmap_get_size(&r->objects->replace_map->map))
93 return 0;
96 prepare_commit_graft(r);
97 if (r->parsed_objects && r->parsed_objects->grafts_nr)
98 return 0;
99 if (is_repository_shallow(r))
100 return 0;
102 return 1;
105 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
107 *fd = git_open(graph_file);
108 if (*fd < 0)
109 return 0;
110 if (fstat(*fd, st)) {
111 close(*fd);
112 return 0;
114 return 1;
117 struct commit_graph *load_commit_graph_one_fd_st(int fd, struct stat *st)
119 void *graph_map;
120 size_t graph_size;
121 struct commit_graph *ret;
123 graph_size = xsize_t(st->st_size);
125 if (graph_size < GRAPH_MIN_SIZE) {
126 close(fd);
127 error(_("commit-graph file is too small"));
128 return NULL;
130 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
131 ret = parse_commit_graph(graph_map, fd, graph_size);
133 if (!ret) {
134 munmap(graph_map, graph_size);
135 close(fd);
138 return ret;
141 static int verify_commit_graph_lite(struct commit_graph *g)
144 * Basic validation shared between parse_commit_graph()
145 * which'll be called every time the graph is used, and the
146 * much more expensive verify_commit_graph() used by
147 * "commit-graph verify".
149 * There should only be very basic checks here to ensure that
150 * we don't e.g. segfault in fill_commit_in_graph(), but
151 * because this is a very hot codepath nothing that e.g. loops
152 * over g->num_commits, or runs a checksum on the commit-graph
153 * itself.
155 if (!g->chunk_oid_fanout) {
156 error("commit-graph is missing the OID Fanout chunk");
157 return 1;
159 if (!g->chunk_oid_lookup) {
160 error("commit-graph is missing the OID Lookup chunk");
161 return 1;
163 if (!g->chunk_commit_data) {
164 error("commit-graph is missing the Commit Data chunk");
165 return 1;
168 return 0;
171 struct commit_graph *parse_commit_graph(void *graph_map, int fd,
172 size_t graph_size)
174 const unsigned char *data, *chunk_lookup;
175 uint32_t i;
176 struct commit_graph *graph;
177 uint64_t last_chunk_offset;
178 uint32_t last_chunk_id;
179 uint32_t graph_signature;
180 unsigned char graph_version, hash_version;
182 if (!graph_map)
183 return NULL;
185 if (graph_size < GRAPH_MIN_SIZE)
186 return NULL;
188 data = (const unsigned char *)graph_map;
190 graph_signature = get_be32(data);
191 if (graph_signature != GRAPH_SIGNATURE) {
192 error(_("commit-graph signature %X does not match signature %X"),
193 graph_signature, GRAPH_SIGNATURE);
194 return NULL;
197 graph_version = *(unsigned char*)(data + 4);
198 if (graph_version != GRAPH_VERSION) {
199 error(_("commit-graph version %X does not match version %X"),
200 graph_version, GRAPH_VERSION);
201 return NULL;
204 hash_version = *(unsigned char*)(data + 5);
205 if (hash_version != oid_version()) {
206 error(_("commit-graph hash version %X does not match version %X"),
207 hash_version, oid_version());
208 return NULL;
211 graph = alloc_commit_graph();
213 graph->hash_len = the_hash_algo->rawsz;
214 graph->num_chunks = *(unsigned char*)(data + 6);
215 graph->graph_fd = fd;
216 graph->data = graph_map;
217 graph->data_len = graph_size;
219 last_chunk_id = 0;
220 last_chunk_offset = 8;
221 chunk_lookup = data + 8;
222 for (i = 0; i < graph->num_chunks; i++) {
223 uint32_t chunk_id;
224 uint64_t chunk_offset;
225 int chunk_repeated = 0;
227 if (data + graph_size - chunk_lookup <
228 GRAPH_CHUNKLOOKUP_WIDTH) {
229 error(_("commit-graph chunk lookup table entry missing; file may be incomplete"));
230 free(graph);
231 return NULL;
234 chunk_id = get_be32(chunk_lookup + 0);
235 chunk_offset = get_be64(chunk_lookup + 4);
237 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
239 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
240 error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
241 (uint32_t)chunk_offset);
242 free(graph);
243 return NULL;
246 switch (chunk_id) {
247 case GRAPH_CHUNKID_OIDFANOUT:
248 if (graph->chunk_oid_fanout)
249 chunk_repeated = 1;
250 else
251 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
252 break;
254 case GRAPH_CHUNKID_OIDLOOKUP:
255 if (graph->chunk_oid_lookup)
256 chunk_repeated = 1;
257 else
258 graph->chunk_oid_lookup = data + chunk_offset;
259 break;
261 case GRAPH_CHUNKID_DATA:
262 if (graph->chunk_commit_data)
263 chunk_repeated = 1;
264 else
265 graph->chunk_commit_data = data + chunk_offset;
266 break;
268 case GRAPH_CHUNKID_EXTRAEDGES:
269 if (graph->chunk_extra_edges)
270 chunk_repeated = 1;
271 else
272 graph->chunk_extra_edges = data + chunk_offset;
273 break;
275 case GRAPH_CHUNKID_BASE:
276 if (graph->chunk_base_graphs)
277 chunk_repeated = 1;
278 else
279 graph->chunk_base_graphs = data + chunk_offset;
282 if (chunk_repeated) {
283 error(_("commit-graph chunk id %08x appears multiple times"), chunk_id);
284 free(graph);
285 return NULL;
288 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
290 graph->num_commits = (chunk_offset - last_chunk_offset)
291 / graph->hash_len;
294 last_chunk_id = chunk_id;
295 last_chunk_offset = chunk_offset;
298 hashcpy(graph->oid.hash, graph->data + graph->data_len - graph->hash_len);
300 if (verify_commit_graph_lite(graph)) {
301 free(graph);
302 return NULL;
305 return graph;
308 static struct commit_graph *load_commit_graph_one(const char *graph_file)
311 struct stat st;
312 int fd;
313 struct commit_graph *g;
314 int open_ok = open_commit_graph(graph_file, &fd, &st);
316 if (!open_ok)
317 return NULL;
319 g = load_commit_graph_one_fd_st(fd, &st);
321 if (g)
322 g->filename = xstrdup(graph_file);
324 return g;
327 static struct commit_graph *load_commit_graph_v1(struct repository *r, const char *obj_dir)
329 char *graph_name = get_commit_graph_filename(obj_dir);
330 struct commit_graph *g = load_commit_graph_one(graph_name);
331 free(graph_name);
333 if (g)
334 g->obj_dir = obj_dir;
336 return g;
339 static int add_graph_to_chain(struct commit_graph *g,
340 struct commit_graph *chain,
341 struct object_id *oids,
342 int n)
344 struct commit_graph *cur_g = chain;
346 if (n && !g->chunk_base_graphs) {
347 warning(_("commit-graph has no base graphs chunk"));
348 return 0;
351 while (n) {
352 n--;
354 if (!cur_g ||
355 !oideq(&oids[n], &cur_g->oid) ||
356 !hasheq(oids[n].hash, g->chunk_base_graphs + g->hash_len * n)) {
357 warning(_("commit-graph chain does not match"));
358 return 0;
361 cur_g = cur_g->base_graph;
364 g->base_graph = chain;
366 if (chain)
367 g->num_commits_in_base = chain->num_commits + chain->num_commits_in_base;
369 return 1;
372 static struct commit_graph *load_commit_graph_chain(struct repository *r, const char *obj_dir)
374 struct commit_graph *graph_chain = NULL;
375 struct strbuf line = STRBUF_INIT;
376 struct stat st;
377 struct object_id *oids;
378 int i = 0, valid = 1, count;
379 char *chain_name = get_chain_filename(obj_dir);
380 FILE *fp;
381 int stat_res;
383 fp = fopen(chain_name, "r");
384 stat_res = stat(chain_name, &st);
385 free(chain_name);
387 if (!fp ||
388 stat_res ||
389 st.st_size <= the_hash_algo->hexsz)
390 return NULL;
392 count = st.st_size / (the_hash_algo->hexsz + 1);
393 oids = xcalloc(count, sizeof(struct object_id));
395 prepare_alt_odb(r);
397 for (i = 0; i < count; i++) {
398 struct object_directory *odb;
400 if (strbuf_getline_lf(&line, fp) == EOF)
401 break;
403 if (get_oid_hex(line.buf, &oids[i])) {
404 warning(_("invalid commit-graph chain: line '%s' not a hash"),
405 line.buf);
406 valid = 0;
407 break;
410 valid = 0;
411 for (odb = r->objects->odb; odb; odb = odb->next) {
412 char *graph_name = get_split_graph_filename(odb->path, line.buf);
413 struct commit_graph *g = load_commit_graph_one(graph_name);
415 free(graph_name);
417 if (g) {
418 g->obj_dir = odb->path;
420 if (add_graph_to_chain(g, graph_chain, oids, i)) {
421 graph_chain = g;
422 valid = 1;
425 break;
429 if (!valid) {
430 warning(_("unable to find all commit-graph files"));
431 break;
435 free(oids);
436 fclose(fp);
437 strbuf_release(&line);
439 return graph_chain;
442 struct commit_graph *read_commit_graph_one(struct repository *r, const char *obj_dir)
444 struct commit_graph *g = load_commit_graph_v1(r, obj_dir);
446 if (!g)
447 g = load_commit_graph_chain(r, obj_dir);
449 return g;
452 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
455 if (r->objects->commit_graph)
456 return;
458 r->objects->commit_graph = read_commit_graph_one(r, obj_dir);
462 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
464 * On the first invocation, this function attemps to load the commit
465 * graph if the_repository is configured to have one.
467 static int prepare_commit_graph(struct repository *r)
469 struct object_directory *odb;
472 * This must come before the "already attempted?" check below, because
473 * we want to disable even an already-loaded graph file.
475 if (r->commit_graph_disabled)
476 return 0;
478 if (r->objects->commit_graph_attempted)
479 return !!r->objects->commit_graph;
480 r->objects->commit_graph_attempted = 1;
482 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
483 die("dying as requested by the '%s' variable on commit-graph load!",
484 GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD);
486 prepare_repo_settings(r);
488 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
489 r->settings.core_commit_graph != 1)
491 * This repository is not configured to use commit graphs, so
492 * do not load one. (But report commit_graph_attempted anyway
493 * so that commit graph loading is not attempted again for this
494 * repository.)
496 return 0;
498 if (!commit_graph_compatible(r))
499 return 0;
501 prepare_alt_odb(r);
502 for (odb = r->objects->odb;
503 !r->objects->commit_graph && odb;
504 odb = odb->next)
505 prepare_commit_graph_one(r, odb->path);
506 return !!r->objects->commit_graph;
509 int generation_numbers_enabled(struct repository *r)
511 uint32_t first_generation;
512 struct commit_graph *g;
513 if (!prepare_commit_graph(r))
514 return 0;
516 g = r->objects->commit_graph;
518 if (!g->num_commits)
519 return 0;
521 first_generation = get_be32(g->chunk_commit_data +
522 g->hash_len + 8) >> 2;
524 return !!first_generation;
527 static void close_commit_graph_one(struct commit_graph *g)
529 if (!g)
530 return;
532 close_commit_graph_one(g->base_graph);
533 free_commit_graph(g);
536 void close_commit_graph(struct raw_object_store *o)
538 close_commit_graph_one(o->commit_graph);
539 o->commit_graph = NULL;
542 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
544 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
545 g->chunk_oid_lookup, g->hash_len, pos);
548 static void load_oid_from_graph(struct commit_graph *g,
549 uint32_t pos,
550 struct object_id *oid)
552 uint32_t lex_index;
554 while (g && pos < g->num_commits_in_base)
555 g = g->base_graph;
557 if (!g)
558 BUG("NULL commit-graph");
560 if (pos >= g->num_commits + g->num_commits_in_base)
561 die(_("invalid commit position. commit-graph is likely corrupt"));
563 lex_index = pos - g->num_commits_in_base;
565 hashcpy(oid->hash, g->chunk_oid_lookup + g->hash_len * lex_index);
568 static struct commit_list **insert_parent_or_die(struct repository *r,
569 struct commit_graph *g,
570 uint32_t pos,
571 struct commit_list **pptr)
573 struct commit *c;
574 struct object_id oid;
576 if (pos >= g->num_commits + g->num_commits_in_base)
577 die("invalid parent position %"PRIu32, pos);
579 load_oid_from_graph(g, pos, &oid);
580 c = lookup_commit(r, &oid);
581 if (!c)
582 die(_("could not find commit %s"), oid_to_hex(&oid));
583 c->graph_pos = pos;
584 return &commit_list_insert(c, pptr)->next;
587 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
589 const unsigned char *commit_data;
590 uint32_t lex_index;
592 while (pos < g->num_commits_in_base)
593 g = g->base_graph;
595 lex_index = pos - g->num_commits_in_base;
596 commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * lex_index;
597 item->graph_pos = pos;
598 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
601 static inline void set_commit_tree(struct commit *c, struct tree *t)
603 c->maybe_tree = t;
606 static int fill_commit_in_graph(struct repository *r,
607 struct commit *item,
608 struct commit_graph *g, uint32_t pos)
610 uint32_t edge_value;
611 uint32_t *parent_data_ptr;
612 uint64_t date_low, date_high;
613 struct commit_list **pptr;
614 const unsigned char *commit_data;
615 uint32_t lex_index;
617 while (pos < g->num_commits_in_base)
618 g = g->base_graph;
620 if (pos >= g->num_commits + g->num_commits_in_base)
621 die(_("invalid commit position. commit-graph is likely corrupt"));
624 * Store the "full" position, but then use the
625 * "local" position for the rest of the calculation.
627 item->graph_pos = pos;
628 lex_index = pos - g->num_commits_in_base;
630 commit_data = g->chunk_commit_data + (g->hash_len + 16) * lex_index;
632 item->object.parsed = 1;
634 set_commit_tree(item, NULL);
636 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
637 date_low = get_be32(commit_data + g->hash_len + 12);
638 item->date = (timestamp_t)((date_high << 32) | date_low);
640 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
642 pptr = &item->parents;
644 edge_value = get_be32(commit_data + g->hash_len);
645 if (edge_value == GRAPH_PARENT_NONE)
646 return 1;
647 pptr = insert_parent_or_die(r, g, edge_value, pptr);
649 edge_value = get_be32(commit_data + g->hash_len + 4);
650 if (edge_value == GRAPH_PARENT_NONE)
651 return 1;
652 if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
653 pptr = insert_parent_or_die(r, g, edge_value, pptr);
654 return 1;
657 parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
658 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
659 do {
660 edge_value = get_be32(parent_data_ptr);
661 pptr = insert_parent_or_die(r, g,
662 edge_value & GRAPH_EDGE_LAST_MASK,
663 pptr);
664 parent_data_ptr++;
665 } while (!(edge_value & GRAPH_LAST_EDGE));
667 return 1;
670 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
672 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
673 *pos = item->graph_pos;
674 return 1;
675 } else {
676 struct commit_graph *cur_g = g;
677 uint32_t lex_index;
679 while (cur_g && !bsearch_graph(cur_g, &(item->object.oid), &lex_index))
680 cur_g = cur_g->base_graph;
682 if (cur_g) {
683 *pos = lex_index + cur_g->num_commits_in_base;
684 return 1;
687 return 0;
691 static int parse_commit_in_graph_one(struct repository *r,
692 struct commit_graph *g,
693 struct commit *item)
695 uint32_t pos;
697 if (item->object.parsed)
698 return 1;
700 if (find_commit_in_graph(item, g, &pos))
701 return fill_commit_in_graph(r, item, g, pos);
703 return 0;
706 int parse_commit_in_graph(struct repository *r, struct commit *item)
708 if (!prepare_commit_graph(r))
709 return 0;
710 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
713 void load_commit_graph_info(struct repository *r, struct commit *item)
715 uint32_t pos;
716 if (!prepare_commit_graph(r))
717 return;
718 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
719 fill_commit_graph_info(item, r->objects->commit_graph, pos);
722 static struct tree *load_tree_for_commit(struct repository *r,
723 struct commit_graph *g,
724 struct commit *c)
726 struct object_id oid;
727 const unsigned char *commit_data;
729 while (c->graph_pos < g->num_commits_in_base)
730 g = g->base_graph;
732 commit_data = g->chunk_commit_data +
733 GRAPH_DATA_WIDTH * (c->graph_pos - g->num_commits_in_base);
735 hashcpy(oid.hash, commit_data);
736 set_commit_tree(c, lookup_tree(r, &oid));
738 return c->maybe_tree;
741 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
742 struct commit_graph *g,
743 const struct commit *c)
745 if (c->maybe_tree)
746 return c->maybe_tree;
747 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
748 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
750 return load_tree_for_commit(r, g, (struct commit *)c);
753 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
755 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
758 struct packed_commit_list {
759 struct commit **list;
760 int nr;
761 int alloc;
764 struct packed_oid_list {
765 struct object_id *list;
766 int nr;
767 int alloc;
770 struct write_commit_graph_context {
771 struct repository *r;
772 char *obj_dir;
773 char *graph_name;
774 struct packed_oid_list oids;
775 struct packed_commit_list commits;
776 int num_extra_edges;
777 unsigned long approx_nr_objects;
778 struct progress *progress;
779 int progress_done;
780 uint64_t progress_cnt;
782 char *base_graph_name;
783 int num_commit_graphs_before;
784 int num_commit_graphs_after;
785 char **commit_graph_filenames_before;
786 char **commit_graph_filenames_after;
787 char **commit_graph_hash_after;
788 uint32_t new_num_commits_in_base;
789 struct commit_graph *new_base_graph;
791 unsigned append:1,
792 report_progress:1,
793 split:1,
794 check_oids:1;
796 const struct split_commit_graph_opts *split_opts;
799 static void write_graph_chunk_fanout(struct hashfile *f,
800 struct write_commit_graph_context *ctx)
802 int i, count = 0;
803 struct commit **list = ctx->commits.list;
806 * Write the first-level table (the list is sorted,
807 * but we use a 256-entry lookup to be able to avoid
808 * having to do eight extra binary search iterations).
810 for (i = 0; i < 256; i++) {
811 while (count < ctx->commits.nr) {
812 if ((*list)->object.oid.hash[0] != i)
813 break;
814 display_progress(ctx->progress, ++ctx->progress_cnt);
815 count++;
816 list++;
819 hashwrite_be32(f, count);
823 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
824 struct write_commit_graph_context *ctx)
826 struct commit **list = ctx->commits.list;
827 int count;
828 for (count = 0; count < ctx->commits.nr; count++, list++) {
829 display_progress(ctx->progress, ++ctx->progress_cnt);
830 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
834 static const unsigned char *commit_to_sha1(size_t index, void *table)
836 struct commit **commits = table;
837 return commits[index]->object.oid.hash;
840 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
841 struct write_commit_graph_context *ctx)
843 struct commit **list = ctx->commits.list;
844 struct commit **last = ctx->commits.list + ctx->commits.nr;
845 uint32_t num_extra_edges = 0;
847 while (list < last) {
848 struct commit_list *parent;
849 int edge_value;
850 uint32_t packedDate[2];
851 display_progress(ctx->progress, ++ctx->progress_cnt);
853 parse_commit_no_graph(*list);
854 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
856 parent = (*list)->parents;
858 if (!parent)
859 edge_value = GRAPH_PARENT_NONE;
860 else {
861 edge_value = sha1_pos(parent->item->object.oid.hash,
862 ctx->commits.list,
863 ctx->commits.nr,
864 commit_to_sha1);
866 if (edge_value >= 0)
867 edge_value += ctx->new_num_commits_in_base;
868 else {
869 uint32_t pos;
870 if (find_commit_in_graph(parent->item,
871 ctx->new_base_graph,
872 &pos))
873 edge_value = pos;
876 if (edge_value < 0)
877 BUG("missing parent %s for commit %s",
878 oid_to_hex(&parent->item->object.oid),
879 oid_to_hex(&(*list)->object.oid));
882 hashwrite_be32(f, edge_value);
884 if (parent)
885 parent = parent->next;
887 if (!parent)
888 edge_value = GRAPH_PARENT_NONE;
889 else if (parent->next)
890 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
891 else {
892 edge_value = sha1_pos(parent->item->object.oid.hash,
893 ctx->commits.list,
894 ctx->commits.nr,
895 commit_to_sha1);
897 if (edge_value >= 0)
898 edge_value += ctx->new_num_commits_in_base;
899 else {
900 uint32_t pos;
901 if (find_commit_in_graph(parent->item,
902 ctx->new_base_graph,
903 &pos))
904 edge_value = pos;
907 if (edge_value < 0)
908 BUG("missing parent %s for commit %s",
909 oid_to_hex(&parent->item->object.oid),
910 oid_to_hex(&(*list)->object.oid));
913 hashwrite_be32(f, edge_value);
915 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
916 do {
917 num_extra_edges++;
918 parent = parent->next;
919 } while (parent);
922 if (sizeof((*list)->date) > 4)
923 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
924 else
925 packedDate[0] = 0;
927 packedDate[0] |= htonl((*list)->generation << 2);
929 packedDate[1] = htonl((*list)->date);
930 hashwrite(f, packedDate, 8);
932 list++;
936 static void write_graph_chunk_extra_edges(struct hashfile *f,
937 struct write_commit_graph_context *ctx)
939 struct commit **list = ctx->commits.list;
940 struct commit **last = ctx->commits.list + ctx->commits.nr;
941 struct commit_list *parent;
943 while (list < last) {
944 int num_parents = 0;
946 display_progress(ctx->progress, ++ctx->progress_cnt);
948 for (parent = (*list)->parents; num_parents < 3 && parent;
949 parent = parent->next)
950 num_parents++;
952 if (num_parents <= 2) {
953 list++;
954 continue;
957 /* Since num_parents > 2, this initializer is safe. */
958 for (parent = (*list)->parents->next; parent; parent = parent->next) {
959 int edge_value = sha1_pos(parent->item->object.oid.hash,
960 ctx->commits.list,
961 ctx->commits.nr,
962 commit_to_sha1);
964 if (edge_value >= 0)
965 edge_value += ctx->new_num_commits_in_base;
966 else {
967 uint32_t pos;
968 if (find_commit_in_graph(parent->item,
969 ctx->new_base_graph,
970 &pos))
971 edge_value = pos;
974 if (edge_value < 0)
975 BUG("missing parent %s for commit %s",
976 oid_to_hex(&parent->item->object.oid),
977 oid_to_hex(&(*list)->object.oid));
978 else if (!parent->next)
979 edge_value |= GRAPH_LAST_EDGE;
981 hashwrite_be32(f, edge_value);
984 list++;
988 static int oid_compare(const void *_a, const void *_b)
990 const struct object_id *a = (const struct object_id *)_a;
991 const struct object_id *b = (const struct object_id *)_b;
992 return oidcmp(a, b);
995 static int add_packed_commits(const struct object_id *oid,
996 struct packed_git *pack,
997 uint32_t pos,
998 void *data)
1000 struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
1001 enum object_type type;
1002 off_t offset = nth_packed_object_offset(pack, pos);
1003 struct object_info oi = OBJECT_INFO_INIT;
1005 if (ctx->progress)
1006 display_progress(ctx->progress, ++ctx->progress_done);
1008 oi.typep = &type;
1009 if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
1010 die(_("unable to get type of object %s"), oid_to_hex(oid));
1012 if (type != OBJ_COMMIT)
1013 return 0;
1015 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1016 oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
1017 ctx->oids.nr++;
1019 return 0;
1022 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
1024 struct commit_list *parent;
1025 for (parent = commit->parents; parent; parent = parent->next) {
1026 if (!(parent->item->object.flags & UNINTERESTING)) {
1027 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1028 oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
1029 ctx->oids.nr++;
1030 parent->item->object.flags |= UNINTERESTING;
1035 static void close_reachable(struct write_commit_graph_context *ctx)
1037 int i;
1038 struct commit *commit;
1040 if (ctx->report_progress)
1041 ctx->progress = start_delayed_progress(
1042 _("Loading known commits in commit graph"),
1043 ctx->oids.nr);
1044 for (i = 0; i < ctx->oids.nr; i++) {
1045 display_progress(ctx->progress, i + 1);
1046 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1047 if (commit)
1048 commit->object.flags |= UNINTERESTING;
1050 stop_progress(&ctx->progress);
1053 * As this loop runs, ctx->oids.nr may grow, but not more
1054 * than the number of missing commits in the reachable
1055 * closure.
1057 if (ctx->report_progress)
1058 ctx->progress = start_delayed_progress(
1059 _("Expanding reachable commits in commit graph"),
1061 for (i = 0; i < ctx->oids.nr; i++) {
1062 display_progress(ctx->progress, i + 1);
1063 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1065 if (!commit)
1066 continue;
1067 if (ctx->split) {
1068 if (!parse_commit(commit) &&
1069 commit->graph_pos == COMMIT_NOT_FROM_GRAPH)
1070 add_missing_parents(ctx, commit);
1071 } else if (!parse_commit_no_graph(commit))
1072 add_missing_parents(ctx, commit);
1074 stop_progress(&ctx->progress);
1076 if (ctx->report_progress)
1077 ctx->progress = start_delayed_progress(
1078 _("Clearing commit marks in commit graph"),
1079 ctx->oids.nr);
1080 for (i = 0; i < ctx->oids.nr; i++) {
1081 display_progress(ctx->progress, i + 1);
1082 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1084 if (commit)
1085 commit->object.flags &= ~UNINTERESTING;
1087 stop_progress(&ctx->progress);
1090 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1092 int i;
1093 struct commit_list *list = NULL;
1095 if (ctx->report_progress)
1096 ctx->progress = start_progress(
1097 _("Computing commit graph generation numbers"),
1098 ctx->commits.nr);
1099 for (i = 0; i < ctx->commits.nr; i++) {
1100 display_progress(ctx->progress, i + 1);
1101 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
1102 ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
1103 continue;
1105 commit_list_insert(ctx->commits.list[i], &list);
1106 while (list) {
1107 struct commit *current = list->item;
1108 struct commit_list *parent;
1109 int all_parents_computed = 1;
1110 uint32_t max_generation = 0;
1112 for (parent = current->parents; parent; parent = parent->next) {
1113 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
1114 parent->item->generation == GENERATION_NUMBER_ZERO) {
1115 all_parents_computed = 0;
1116 commit_list_insert(parent->item, &list);
1117 break;
1118 } else if (parent->item->generation > max_generation) {
1119 max_generation = parent->item->generation;
1123 if (all_parents_computed) {
1124 current->generation = max_generation + 1;
1125 pop_commit(&list);
1127 if (current->generation > GENERATION_NUMBER_MAX)
1128 current->generation = GENERATION_NUMBER_MAX;
1132 stop_progress(&ctx->progress);
1135 static int add_ref_to_list(const char *refname,
1136 const struct object_id *oid,
1137 int flags, void *cb_data)
1139 struct string_list *list = (struct string_list *)cb_data;
1141 string_list_append(list, oid_to_hex(oid));
1142 return 0;
1145 int write_commit_graph_reachable(const char *obj_dir,
1146 enum commit_graph_write_flags flags,
1147 const struct split_commit_graph_opts *split_opts)
1149 struct string_list list = STRING_LIST_INIT_DUP;
1150 int result;
1152 for_each_ref(add_ref_to_list, &list);
1153 result = write_commit_graph(obj_dir, NULL, &list,
1154 flags, split_opts);
1156 string_list_clear(&list, 0);
1157 return result;
1160 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
1161 struct string_list *pack_indexes)
1163 uint32_t i;
1164 struct strbuf progress_title = STRBUF_INIT;
1165 struct strbuf packname = STRBUF_INIT;
1166 int dirlen;
1168 strbuf_addf(&packname, "%s/pack/", ctx->obj_dir);
1169 dirlen = packname.len;
1170 if (ctx->report_progress) {
1171 strbuf_addf(&progress_title,
1172 Q_("Finding commits for commit graph in %d pack",
1173 "Finding commits for commit graph in %d packs",
1174 pack_indexes->nr),
1175 pack_indexes->nr);
1176 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1177 ctx->progress_done = 0;
1179 for (i = 0; i < pack_indexes->nr; i++) {
1180 struct packed_git *p;
1181 strbuf_setlen(&packname, dirlen);
1182 strbuf_addstr(&packname, pack_indexes->items[i].string);
1183 p = add_packed_git(packname.buf, packname.len, 1);
1184 if (!p) {
1185 error(_("error adding pack %s"), packname.buf);
1186 return -1;
1188 if (open_pack_index(p)) {
1189 error(_("error opening index for %s"), packname.buf);
1190 return -1;
1192 for_each_object_in_pack(p, add_packed_commits, ctx,
1193 FOR_EACH_OBJECT_PACK_ORDER);
1194 close_pack(p);
1195 free(p);
1198 stop_progress(&ctx->progress);
1199 strbuf_release(&progress_title);
1200 strbuf_release(&packname);
1202 return 0;
1205 static int fill_oids_from_commit_hex(struct write_commit_graph_context *ctx,
1206 struct string_list *commit_hex)
1208 uint32_t i;
1209 struct strbuf progress_title = STRBUF_INIT;
1211 if (ctx->report_progress) {
1212 strbuf_addf(&progress_title,
1213 Q_("Finding commits for commit graph from %d ref",
1214 "Finding commits for commit graph from %d refs",
1215 commit_hex->nr),
1216 commit_hex->nr);
1217 ctx->progress = start_delayed_progress(
1218 progress_title.buf,
1219 commit_hex->nr);
1221 for (i = 0; i < commit_hex->nr; i++) {
1222 const char *end;
1223 struct object_id oid;
1224 struct commit *result;
1226 display_progress(ctx->progress, i + 1);
1227 if (!parse_oid_hex(commit_hex->items[i].string, &oid, &end) &&
1228 (result = lookup_commit_reference_gently(ctx->r, &oid, 1))) {
1229 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1230 oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
1231 ctx->oids.nr++;
1232 } else if (ctx->check_oids) {
1233 error(_("invalid commit object id: %s"),
1234 commit_hex->items[i].string);
1235 return -1;
1238 stop_progress(&ctx->progress);
1239 strbuf_release(&progress_title);
1241 return 0;
1244 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1246 if (ctx->report_progress)
1247 ctx->progress = start_delayed_progress(
1248 _("Finding commits for commit graph among packed objects"),
1249 ctx->approx_nr_objects);
1250 for_each_packed_object(add_packed_commits, ctx,
1251 FOR_EACH_OBJECT_PACK_ORDER);
1252 if (ctx->progress_done < ctx->approx_nr_objects)
1253 display_progress(ctx->progress, ctx->approx_nr_objects);
1254 stop_progress(&ctx->progress);
1257 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
1259 uint32_t i, count_distinct = 1;
1261 if (ctx->report_progress)
1262 ctx->progress = start_delayed_progress(
1263 _("Counting distinct commits in commit graph"),
1264 ctx->oids.nr);
1265 display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1266 QSORT(ctx->oids.list, ctx->oids.nr, oid_compare);
1268 for (i = 1; i < ctx->oids.nr; i++) {
1269 display_progress(ctx->progress, i + 1);
1270 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i])) {
1271 if (ctx->split) {
1272 struct commit *c = lookup_commit(ctx->r, &ctx->oids.list[i]);
1274 if (!c || c->graph_pos != COMMIT_NOT_FROM_GRAPH)
1275 continue;
1278 count_distinct++;
1281 stop_progress(&ctx->progress);
1283 return count_distinct;
1286 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1288 uint32_t i;
1290 ctx->num_extra_edges = 0;
1291 if (ctx->report_progress)
1292 ctx->progress = start_delayed_progress(
1293 _("Finding extra edges in commit graph"),
1294 ctx->oids.nr);
1295 for (i = 0; i < ctx->oids.nr; i++) {
1296 unsigned int num_parents;
1298 display_progress(ctx->progress, i + 1);
1299 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1300 continue;
1302 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + 1, ctx->commits.alloc);
1303 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1305 if (ctx->split &&
1306 ctx->commits.list[ctx->commits.nr]->graph_pos != COMMIT_NOT_FROM_GRAPH)
1307 continue;
1309 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1311 num_parents = commit_list_count(ctx->commits.list[ctx->commits.nr]->parents);
1312 if (num_parents > 2)
1313 ctx->num_extra_edges += num_parents - 1;
1315 ctx->commits.nr++;
1317 stop_progress(&ctx->progress);
1320 static int write_graph_chunk_base_1(struct hashfile *f,
1321 struct commit_graph *g)
1323 int num = 0;
1325 if (!g)
1326 return 0;
1328 num = write_graph_chunk_base_1(f, g->base_graph);
1329 hashwrite(f, g->oid.hash, the_hash_algo->rawsz);
1330 return num + 1;
1333 static int write_graph_chunk_base(struct hashfile *f,
1334 struct write_commit_graph_context *ctx)
1336 int num = write_graph_chunk_base_1(f, ctx->new_base_graph);
1338 if (num != ctx->num_commit_graphs_after - 1) {
1339 error(_("failed to write correct number of base graph ids"));
1340 return -1;
1343 return 0;
1346 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1348 uint32_t i;
1349 int fd;
1350 struct hashfile *f;
1351 struct lock_file lk = LOCK_INIT;
1352 uint32_t chunk_ids[6];
1353 uint64_t chunk_offsets[6];
1354 const unsigned hashsz = the_hash_algo->rawsz;
1355 struct strbuf progress_title = STRBUF_INIT;
1356 int num_chunks = 3;
1357 struct object_id file_hash;
1359 if (ctx->split) {
1360 struct strbuf tmp_file = STRBUF_INIT;
1362 strbuf_addf(&tmp_file,
1363 "%s/info/commit-graphs/tmp_graph_XXXXXX",
1364 ctx->obj_dir);
1365 ctx->graph_name = strbuf_detach(&tmp_file, NULL);
1366 } else {
1367 ctx->graph_name = get_commit_graph_filename(ctx->obj_dir);
1370 if (safe_create_leading_directories(ctx->graph_name)) {
1371 UNLEAK(ctx->graph_name);
1372 error(_("unable to create leading directories of %s"),
1373 ctx->graph_name);
1374 return -1;
1377 if (ctx->split) {
1378 char *lock_name = get_chain_filename(ctx->obj_dir);
1380 hold_lock_file_for_update(&lk, lock_name, LOCK_DIE_ON_ERROR);
1382 fd = git_mkstemp_mode(ctx->graph_name, 0444);
1383 if (fd < 0) {
1384 error(_("unable to create '%s'"), ctx->graph_name);
1385 return -1;
1388 f = hashfd(fd, ctx->graph_name);
1389 } else {
1390 hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
1391 fd = lk.tempfile->fd;
1392 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1395 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1396 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1397 chunk_ids[2] = GRAPH_CHUNKID_DATA;
1398 if (ctx->num_extra_edges) {
1399 chunk_ids[num_chunks] = GRAPH_CHUNKID_EXTRAEDGES;
1400 num_chunks++;
1402 if (ctx->num_commit_graphs_after > 1) {
1403 chunk_ids[num_chunks] = GRAPH_CHUNKID_BASE;
1404 num_chunks++;
1407 chunk_ids[num_chunks] = 0;
1409 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1410 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1411 chunk_offsets[2] = chunk_offsets[1] + hashsz * ctx->commits.nr;
1412 chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * ctx->commits.nr;
1414 num_chunks = 3;
1415 if (ctx->num_extra_edges) {
1416 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1417 4 * ctx->num_extra_edges;
1418 num_chunks++;
1420 if (ctx->num_commit_graphs_after > 1) {
1421 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1422 hashsz * (ctx->num_commit_graphs_after - 1);
1423 num_chunks++;
1426 hashwrite_be32(f, GRAPH_SIGNATURE);
1428 hashwrite_u8(f, GRAPH_VERSION);
1429 hashwrite_u8(f, oid_version());
1430 hashwrite_u8(f, num_chunks);
1431 hashwrite_u8(f, ctx->num_commit_graphs_after - 1);
1433 for (i = 0; i <= num_chunks; i++) {
1434 uint32_t chunk_write[3];
1436 chunk_write[0] = htonl(chunk_ids[i]);
1437 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1438 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1439 hashwrite(f, chunk_write, 12);
1442 if (ctx->report_progress) {
1443 strbuf_addf(&progress_title,
1444 Q_("Writing out commit graph in %d pass",
1445 "Writing out commit graph in %d passes",
1446 num_chunks),
1447 num_chunks);
1448 ctx->progress = start_delayed_progress(
1449 progress_title.buf,
1450 num_chunks * ctx->commits.nr);
1452 write_graph_chunk_fanout(f, ctx);
1453 write_graph_chunk_oids(f, hashsz, ctx);
1454 write_graph_chunk_data(f, hashsz, ctx);
1455 if (ctx->num_extra_edges)
1456 write_graph_chunk_extra_edges(f, ctx);
1457 if (ctx->num_commit_graphs_after > 1 &&
1458 write_graph_chunk_base(f, ctx)) {
1459 return -1;
1461 stop_progress(&ctx->progress);
1462 strbuf_release(&progress_title);
1464 if (ctx->split && ctx->base_graph_name && ctx->num_commit_graphs_after > 1) {
1465 char *new_base_hash = xstrdup(oid_to_hex(&ctx->new_base_graph->oid));
1466 char *new_base_name = get_split_graph_filename(ctx->new_base_graph->obj_dir, new_base_hash);
1468 free(ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2]);
1469 free(ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2]);
1470 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2] = new_base_name;
1471 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2] = new_base_hash;
1474 close_commit_graph(ctx->r->objects);
1475 finalize_hashfile(f, file_hash.hash, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1477 if (ctx->split) {
1478 FILE *chainf = fdopen_lock_file(&lk, "w");
1479 char *final_graph_name;
1480 int result;
1482 close(fd);
1484 if (!chainf) {
1485 error(_("unable to open commit-graph chain file"));
1486 return -1;
1489 if (ctx->base_graph_name) {
1490 const char *dest = ctx->commit_graph_filenames_after[
1491 ctx->num_commit_graphs_after - 2];
1493 if (strcmp(ctx->base_graph_name, dest)) {
1494 result = rename(ctx->base_graph_name, dest);
1496 if (result) {
1497 error(_("failed to rename base commit-graph file"));
1498 return -1;
1501 } else {
1502 char *graph_name = get_commit_graph_filename(ctx->obj_dir);
1503 unlink(graph_name);
1506 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1] = xstrdup(oid_to_hex(&file_hash));
1507 final_graph_name = get_split_graph_filename(ctx->obj_dir,
1508 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1]);
1509 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 1] = final_graph_name;
1511 result = rename(ctx->graph_name, final_graph_name);
1513 for (i = 0; i < ctx->num_commit_graphs_after; i++)
1514 fprintf(lk.tempfile->fp, "%s\n", ctx->commit_graph_hash_after[i]);
1516 if (result) {
1517 error(_("failed to rename temporary commit-graph file"));
1518 return -1;
1522 commit_lock_file(&lk);
1524 return 0;
1527 static void split_graph_merge_strategy(struct write_commit_graph_context *ctx)
1529 struct commit_graph *g = ctx->r->objects->commit_graph;
1530 uint32_t num_commits = ctx->commits.nr;
1531 uint32_t i;
1533 int max_commits = 0;
1534 int size_mult = 2;
1536 if (ctx->split_opts) {
1537 max_commits = ctx->split_opts->max_commits;
1538 size_mult = ctx->split_opts->size_multiple;
1541 g = ctx->r->objects->commit_graph;
1542 ctx->num_commit_graphs_after = ctx->num_commit_graphs_before + 1;
1544 while (g && (g->num_commits <= size_mult * num_commits ||
1545 (max_commits && num_commits > max_commits))) {
1546 if (strcmp(g->obj_dir, ctx->obj_dir))
1547 break;
1549 num_commits += g->num_commits;
1550 g = g->base_graph;
1552 ctx->num_commit_graphs_after--;
1555 ctx->new_base_graph = g;
1557 if (ctx->num_commit_graphs_after == 2) {
1558 char *old_graph_name = get_commit_graph_filename(g->obj_dir);
1560 if (!strcmp(g->filename, old_graph_name) &&
1561 strcmp(g->obj_dir, ctx->obj_dir)) {
1562 ctx->num_commit_graphs_after = 1;
1563 ctx->new_base_graph = NULL;
1566 free(old_graph_name);
1569 ALLOC_ARRAY(ctx->commit_graph_filenames_after, ctx->num_commit_graphs_after);
1570 ALLOC_ARRAY(ctx->commit_graph_hash_after, ctx->num_commit_graphs_after);
1572 for (i = 0; i < ctx->num_commit_graphs_after &&
1573 i < ctx->num_commit_graphs_before; i++)
1574 ctx->commit_graph_filenames_after[i] = xstrdup(ctx->commit_graph_filenames_before[i]);
1576 i = ctx->num_commit_graphs_before - 1;
1577 g = ctx->r->objects->commit_graph;
1579 while (g) {
1580 if (i < ctx->num_commit_graphs_after)
1581 ctx->commit_graph_hash_after[i] = xstrdup(oid_to_hex(&g->oid));
1583 i--;
1584 g = g->base_graph;
1588 static void merge_commit_graph(struct write_commit_graph_context *ctx,
1589 struct commit_graph *g)
1591 uint32_t i;
1592 uint32_t offset = g->num_commits_in_base;
1594 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + g->num_commits, ctx->commits.alloc);
1596 for (i = 0; i < g->num_commits; i++) {
1597 struct object_id oid;
1598 struct commit *result;
1600 display_progress(ctx->progress, i + 1);
1602 load_oid_from_graph(g, i + offset, &oid);
1604 /* only add commits if they still exist in the repo */
1605 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1607 if (result) {
1608 ctx->commits.list[ctx->commits.nr] = result;
1609 ctx->commits.nr++;
1614 static int commit_compare(const void *_a, const void *_b)
1616 const struct commit *a = *(const struct commit **)_a;
1617 const struct commit *b = *(const struct commit **)_b;
1618 return oidcmp(&a->object.oid, &b->object.oid);
1621 static void sort_and_scan_merged_commits(struct write_commit_graph_context *ctx)
1623 uint32_t i;
1625 if (ctx->report_progress)
1626 ctx->progress = start_delayed_progress(
1627 _("Scanning merged commits"),
1628 ctx->commits.nr);
1630 QSORT(ctx->commits.list, ctx->commits.nr, commit_compare);
1632 ctx->num_extra_edges = 0;
1633 for (i = 0; i < ctx->commits.nr; i++) {
1634 display_progress(ctx->progress, i);
1636 if (i && oideq(&ctx->commits.list[i - 1]->object.oid,
1637 &ctx->commits.list[i]->object.oid)) {
1638 die(_("unexpected duplicate commit id %s"),
1639 oid_to_hex(&ctx->commits.list[i]->object.oid));
1640 } else {
1641 unsigned int num_parents;
1643 num_parents = commit_list_count(ctx->commits.list[i]->parents);
1644 if (num_parents > 2)
1645 ctx->num_extra_edges += num_parents - 1;
1649 stop_progress(&ctx->progress);
1652 static void merge_commit_graphs(struct write_commit_graph_context *ctx)
1654 struct commit_graph *g = ctx->r->objects->commit_graph;
1655 uint32_t current_graph_number = ctx->num_commit_graphs_before;
1656 struct strbuf progress_title = STRBUF_INIT;
1658 while (g && current_graph_number >= ctx->num_commit_graphs_after) {
1659 current_graph_number--;
1661 if (ctx->report_progress) {
1662 strbuf_addstr(&progress_title, _("Merging commit-graph"));
1663 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1666 merge_commit_graph(ctx, g);
1667 stop_progress(&ctx->progress);
1668 strbuf_release(&progress_title);
1670 g = g->base_graph;
1673 if (g) {
1674 ctx->new_base_graph = g;
1675 ctx->new_num_commits_in_base = g->num_commits + g->num_commits_in_base;
1678 if (ctx->new_base_graph)
1679 ctx->base_graph_name = xstrdup(ctx->new_base_graph->filename);
1681 sort_and_scan_merged_commits(ctx);
1684 static void mark_commit_graphs(struct write_commit_graph_context *ctx)
1686 uint32_t i;
1687 time_t now = time(NULL);
1689 for (i = ctx->num_commit_graphs_after - 1; i < ctx->num_commit_graphs_before; i++) {
1690 struct stat st;
1691 struct utimbuf updated_time;
1693 stat(ctx->commit_graph_filenames_before[i], &st);
1695 updated_time.actime = st.st_atime;
1696 updated_time.modtime = now;
1697 utime(ctx->commit_graph_filenames_before[i], &updated_time);
1701 static void expire_commit_graphs(struct write_commit_graph_context *ctx)
1703 struct strbuf path = STRBUF_INIT;
1704 DIR *dir;
1705 struct dirent *de;
1706 size_t dirnamelen;
1707 timestamp_t expire_time = time(NULL);
1709 if (ctx->split_opts && ctx->split_opts->expire_time)
1710 expire_time -= ctx->split_opts->expire_time;
1711 if (!ctx->split) {
1712 char *chain_file_name = get_chain_filename(ctx->obj_dir);
1713 unlink(chain_file_name);
1714 free(chain_file_name);
1715 ctx->num_commit_graphs_after = 0;
1718 strbuf_addstr(&path, ctx->obj_dir);
1719 strbuf_addstr(&path, "/info/commit-graphs");
1720 dir = opendir(path.buf);
1722 if (!dir)
1723 goto out;
1725 strbuf_addch(&path, '/');
1726 dirnamelen = path.len;
1727 while ((de = readdir(dir)) != NULL) {
1728 struct stat st;
1729 uint32_t i, found = 0;
1731 strbuf_setlen(&path, dirnamelen);
1732 strbuf_addstr(&path, de->d_name);
1734 stat(path.buf, &st);
1736 if (st.st_mtime > expire_time)
1737 continue;
1738 if (path.len < 6 || strcmp(path.buf + path.len - 6, ".graph"))
1739 continue;
1741 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
1742 if (!strcmp(ctx->commit_graph_filenames_after[i],
1743 path.buf)) {
1744 found = 1;
1745 break;
1749 if (!found)
1750 unlink(path.buf);
1753 out:
1754 strbuf_release(&path);
1757 int write_commit_graph(const char *obj_dir,
1758 struct string_list *pack_indexes,
1759 struct string_list *commit_hex,
1760 enum commit_graph_write_flags flags,
1761 const struct split_commit_graph_opts *split_opts)
1763 struct write_commit_graph_context *ctx;
1764 uint32_t i, count_distinct = 0;
1765 size_t len;
1766 int res = 0;
1768 if (!commit_graph_compatible(the_repository))
1769 return 0;
1771 ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
1772 ctx->r = the_repository;
1774 /* normalize object dir with no trailing slash */
1775 ctx->obj_dir = xmallocz(strlen(obj_dir) + 1);
1776 normalize_path_copy(ctx->obj_dir, obj_dir);
1777 len = strlen(ctx->obj_dir);
1778 if (len && ctx->obj_dir[len - 1] == '/')
1779 ctx->obj_dir[len - 1] = 0;
1781 ctx->append = flags & COMMIT_GRAPH_WRITE_APPEND ? 1 : 0;
1782 ctx->report_progress = flags & COMMIT_GRAPH_WRITE_PROGRESS ? 1 : 0;
1783 ctx->split = flags & COMMIT_GRAPH_WRITE_SPLIT ? 1 : 0;
1784 ctx->check_oids = flags & COMMIT_GRAPH_WRITE_CHECK_OIDS ? 1 : 0;
1785 ctx->split_opts = split_opts;
1787 if (ctx->split) {
1788 struct commit_graph *g;
1789 prepare_commit_graph(ctx->r);
1791 g = ctx->r->objects->commit_graph;
1793 while (g) {
1794 ctx->num_commit_graphs_before++;
1795 g = g->base_graph;
1798 if (ctx->num_commit_graphs_before) {
1799 ALLOC_ARRAY(ctx->commit_graph_filenames_before, ctx->num_commit_graphs_before);
1800 i = ctx->num_commit_graphs_before;
1801 g = ctx->r->objects->commit_graph;
1803 while (g) {
1804 ctx->commit_graph_filenames_before[--i] = xstrdup(g->filename);
1805 g = g->base_graph;
1810 ctx->approx_nr_objects = approximate_object_count();
1811 ctx->oids.alloc = ctx->approx_nr_objects / 32;
1813 if (ctx->split && split_opts && ctx->oids.alloc > split_opts->max_commits)
1814 ctx->oids.alloc = split_opts->max_commits;
1816 if (ctx->append) {
1817 prepare_commit_graph_one(ctx->r, ctx->obj_dir);
1818 if (ctx->r->objects->commit_graph)
1819 ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
1822 if (ctx->oids.alloc < 1024)
1823 ctx->oids.alloc = 1024;
1824 ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
1826 if (ctx->append && ctx->r->objects->commit_graph) {
1827 struct commit_graph *g = ctx->r->objects->commit_graph;
1828 for (i = 0; i < g->num_commits; i++) {
1829 const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
1830 hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
1834 if (pack_indexes) {
1835 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
1836 goto cleanup;
1839 if (commit_hex) {
1840 if ((res = fill_oids_from_commit_hex(ctx, commit_hex)))
1841 goto cleanup;
1844 if (!pack_indexes && !commit_hex)
1845 fill_oids_from_all_packs(ctx);
1847 close_reachable(ctx);
1849 count_distinct = count_distinct_commits(ctx);
1851 if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
1852 error(_("the commit graph format cannot write %d commits"), count_distinct);
1853 res = -1;
1854 goto cleanup;
1857 ctx->commits.alloc = count_distinct;
1858 ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
1860 copy_oids_to_commits(ctx);
1862 if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
1863 error(_("too many commits to write graph"));
1864 res = -1;
1865 goto cleanup;
1868 if (!ctx->commits.nr)
1869 goto cleanup;
1871 if (ctx->split) {
1872 split_graph_merge_strategy(ctx);
1874 merge_commit_graphs(ctx);
1875 } else
1876 ctx->num_commit_graphs_after = 1;
1878 compute_generation_numbers(ctx);
1880 res = write_commit_graph_file(ctx);
1882 if (ctx->split)
1883 mark_commit_graphs(ctx);
1885 expire_commit_graphs(ctx);
1887 cleanup:
1888 free(ctx->graph_name);
1889 free(ctx->commits.list);
1890 free(ctx->oids.list);
1891 free(ctx->obj_dir);
1893 if (ctx->commit_graph_filenames_after) {
1894 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
1895 free(ctx->commit_graph_filenames_after[i]);
1896 free(ctx->commit_graph_hash_after[i]);
1899 for (i = 0; i < ctx->num_commit_graphs_before; i++)
1900 free(ctx->commit_graph_filenames_before[i]);
1902 free(ctx->commit_graph_filenames_after);
1903 free(ctx->commit_graph_filenames_before);
1904 free(ctx->commit_graph_hash_after);
1907 free(ctx);
1909 return res;
1912 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1913 static int verify_commit_graph_error;
1915 static void graph_report(const char *fmt, ...)
1917 va_list ap;
1919 verify_commit_graph_error = 1;
1920 va_start(ap, fmt);
1921 vfprintf(stderr, fmt, ap);
1922 fprintf(stderr, "\n");
1923 va_end(ap);
1926 #define GENERATION_ZERO_EXISTS 1
1927 #define GENERATION_NUMBER_EXISTS 2
1929 int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
1931 uint32_t i, cur_fanout_pos = 0;
1932 struct object_id prev_oid, cur_oid, checksum;
1933 int generation_zero = 0;
1934 struct hashfile *f;
1935 int devnull;
1936 struct progress *progress = NULL;
1937 int local_error = 0;
1939 if (!g) {
1940 graph_report("no commit-graph file loaded");
1941 return 1;
1944 verify_commit_graph_error = verify_commit_graph_lite(g);
1945 if (verify_commit_graph_error)
1946 return verify_commit_graph_error;
1948 devnull = open("/dev/null", O_WRONLY);
1949 f = hashfd(devnull, NULL);
1950 hashwrite(f, g->data, g->data_len - g->hash_len);
1951 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1952 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1953 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1954 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1957 for (i = 0; i < g->num_commits; i++) {
1958 struct commit *graph_commit;
1960 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1962 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1963 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
1964 oid_to_hex(&prev_oid),
1965 oid_to_hex(&cur_oid));
1967 oidcpy(&prev_oid, &cur_oid);
1969 while (cur_oid.hash[0] > cur_fanout_pos) {
1970 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1972 if (i != fanout_value)
1973 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1974 cur_fanout_pos, fanout_value, i);
1975 cur_fanout_pos++;
1978 graph_commit = lookup_commit(r, &cur_oid);
1979 if (!parse_commit_in_graph_one(r, g, graph_commit))
1980 graph_report(_("failed to parse commit %s from commit-graph"),
1981 oid_to_hex(&cur_oid));
1984 while (cur_fanout_pos < 256) {
1985 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1987 if (g->num_commits != fanout_value)
1988 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1989 cur_fanout_pos, fanout_value, i);
1991 cur_fanout_pos++;
1994 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1995 return verify_commit_graph_error;
1997 progress = start_progress(_("Verifying commits in commit graph"),
1998 g->num_commits);
1999 for (i = 0; i < g->num_commits; i++) {
2000 struct commit *graph_commit, *odb_commit;
2001 struct commit_list *graph_parents, *odb_parents;
2002 uint32_t max_generation = 0;
2004 display_progress(progress, i + 1);
2005 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2007 graph_commit = lookup_commit(r, &cur_oid);
2008 odb_commit = (struct commit *)create_object(r, &cur_oid, alloc_commit_node(r));
2009 if (parse_commit_internal(odb_commit, 0, 0)) {
2010 graph_report(_("failed to parse commit %s from object database for commit-graph"),
2011 oid_to_hex(&cur_oid));
2012 continue;
2015 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
2016 get_commit_tree_oid(odb_commit)))
2017 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2018 oid_to_hex(&cur_oid),
2019 oid_to_hex(get_commit_tree_oid(graph_commit)),
2020 oid_to_hex(get_commit_tree_oid(odb_commit)));
2022 graph_parents = graph_commit->parents;
2023 odb_parents = odb_commit->parents;
2025 while (graph_parents) {
2026 if (odb_parents == NULL) {
2027 graph_report(_("commit-graph parent list for commit %s is too long"),
2028 oid_to_hex(&cur_oid));
2029 break;
2032 /* parse parent in case it is in a base graph */
2033 parse_commit_in_graph_one(r, g, graph_parents->item);
2035 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
2036 graph_report(_("commit-graph parent for %s is %s != %s"),
2037 oid_to_hex(&cur_oid),
2038 oid_to_hex(&graph_parents->item->object.oid),
2039 oid_to_hex(&odb_parents->item->object.oid));
2041 if (graph_parents->item->generation > max_generation)
2042 max_generation = graph_parents->item->generation;
2044 graph_parents = graph_parents->next;
2045 odb_parents = odb_parents->next;
2048 if (odb_parents != NULL)
2049 graph_report(_("commit-graph parent list for commit %s terminates early"),
2050 oid_to_hex(&cur_oid));
2052 if (!graph_commit->generation) {
2053 if (generation_zero == GENERATION_NUMBER_EXISTS)
2054 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2055 oid_to_hex(&cur_oid));
2056 generation_zero = GENERATION_ZERO_EXISTS;
2057 } else if (generation_zero == GENERATION_ZERO_EXISTS)
2058 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2059 oid_to_hex(&cur_oid));
2061 if (generation_zero == GENERATION_ZERO_EXISTS)
2062 continue;
2065 * If one of our parents has generation GENERATION_NUMBER_MAX, then
2066 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
2067 * extra logic in the following condition.
2069 if (max_generation == GENERATION_NUMBER_MAX)
2070 max_generation--;
2072 if (graph_commit->generation != max_generation + 1)
2073 graph_report(_("commit-graph generation for commit %s is %u != %u"),
2074 oid_to_hex(&cur_oid),
2075 graph_commit->generation,
2076 max_generation + 1);
2078 if (graph_commit->date != odb_commit->date)
2079 graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
2080 oid_to_hex(&cur_oid),
2081 graph_commit->date,
2082 odb_commit->date);
2084 stop_progress(&progress);
2086 local_error = verify_commit_graph_error;
2088 if (!(flags & COMMIT_GRAPH_VERIFY_SHALLOW) && g->base_graph)
2089 local_error |= verify_commit_graph(r, g->base_graph, flags);
2091 return local_error;
2094 void free_commit_graph(struct commit_graph *g)
2096 if (!g)
2097 return;
2098 if (g->graph_fd >= 0) {
2099 munmap((void *)g->data, g->data_len);
2100 g->data = NULL;
2101 close(g->graph_fd);
2103 free(g->filename);
2104 free(g);
2107 void disable_commit_graph(struct repository *r)
2109 r->commit_graph_disabled = 1;