Merge branch 'kt/add-i-progress'
[git.git] / commit-graph.c
blobfc4a43b8d6eb112f1c845882b32411c37a402e3b
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 struct object_id *tree;
850 int edge_value;
851 uint32_t packedDate[2];
852 display_progress(ctx->progress, ++ctx->progress_cnt);
854 if (parse_commit_no_graph(*list))
855 die(_("unable to parse commit %s"),
856 oid_to_hex(&(*list)->object.oid));
857 tree = get_commit_tree_oid(*list);
858 if (!tree)
859 die(_("unable to get tree for %s"),
860 oid_to_hex(&(*list)->object.oid));
861 hashwrite(f, tree->hash, hash_len);
863 parent = (*list)->parents;
865 if (!parent)
866 edge_value = GRAPH_PARENT_NONE;
867 else {
868 edge_value = sha1_pos(parent->item->object.oid.hash,
869 ctx->commits.list,
870 ctx->commits.nr,
871 commit_to_sha1);
873 if (edge_value >= 0)
874 edge_value += ctx->new_num_commits_in_base;
875 else {
876 uint32_t pos;
877 if (find_commit_in_graph(parent->item,
878 ctx->new_base_graph,
879 &pos))
880 edge_value = pos;
883 if (edge_value < 0)
884 BUG("missing parent %s for commit %s",
885 oid_to_hex(&parent->item->object.oid),
886 oid_to_hex(&(*list)->object.oid));
889 hashwrite_be32(f, edge_value);
891 if (parent)
892 parent = parent->next;
894 if (!parent)
895 edge_value = GRAPH_PARENT_NONE;
896 else if (parent->next)
897 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
898 else {
899 edge_value = sha1_pos(parent->item->object.oid.hash,
900 ctx->commits.list,
901 ctx->commits.nr,
902 commit_to_sha1);
904 if (edge_value >= 0)
905 edge_value += ctx->new_num_commits_in_base;
906 else {
907 uint32_t pos;
908 if (find_commit_in_graph(parent->item,
909 ctx->new_base_graph,
910 &pos))
911 edge_value = pos;
914 if (edge_value < 0)
915 BUG("missing parent %s for commit %s",
916 oid_to_hex(&parent->item->object.oid),
917 oid_to_hex(&(*list)->object.oid));
920 hashwrite_be32(f, edge_value);
922 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
923 do {
924 num_extra_edges++;
925 parent = parent->next;
926 } while (parent);
929 if (sizeof((*list)->date) > 4)
930 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
931 else
932 packedDate[0] = 0;
934 packedDate[0] |= htonl((*list)->generation << 2);
936 packedDate[1] = htonl((*list)->date);
937 hashwrite(f, packedDate, 8);
939 list++;
943 static void write_graph_chunk_extra_edges(struct hashfile *f,
944 struct write_commit_graph_context *ctx)
946 struct commit **list = ctx->commits.list;
947 struct commit **last = ctx->commits.list + ctx->commits.nr;
948 struct commit_list *parent;
950 while (list < last) {
951 int num_parents = 0;
953 display_progress(ctx->progress, ++ctx->progress_cnt);
955 for (parent = (*list)->parents; num_parents < 3 && parent;
956 parent = parent->next)
957 num_parents++;
959 if (num_parents <= 2) {
960 list++;
961 continue;
964 /* Since num_parents > 2, this initializer is safe. */
965 for (parent = (*list)->parents->next; parent; parent = parent->next) {
966 int edge_value = sha1_pos(parent->item->object.oid.hash,
967 ctx->commits.list,
968 ctx->commits.nr,
969 commit_to_sha1);
971 if (edge_value >= 0)
972 edge_value += ctx->new_num_commits_in_base;
973 else {
974 uint32_t pos;
975 if (find_commit_in_graph(parent->item,
976 ctx->new_base_graph,
977 &pos))
978 edge_value = pos;
981 if (edge_value < 0)
982 BUG("missing parent %s for commit %s",
983 oid_to_hex(&parent->item->object.oid),
984 oid_to_hex(&(*list)->object.oid));
985 else if (!parent->next)
986 edge_value |= GRAPH_LAST_EDGE;
988 hashwrite_be32(f, edge_value);
991 list++;
995 static int oid_compare(const void *_a, const void *_b)
997 const struct object_id *a = (const struct object_id *)_a;
998 const struct object_id *b = (const struct object_id *)_b;
999 return oidcmp(a, b);
1002 static int add_packed_commits(const struct object_id *oid,
1003 struct packed_git *pack,
1004 uint32_t pos,
1005 void *data)
1007 struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
1008 enum object_type type;
1009 off_t offset = nth_packed_object_offset(pack, pos);
1010 struct object_info oi = OBJECT_INFO_INIT;
1012 if (ctx->progress)
1013 display_progress(ctx->progress, ++ctx->progress_done);
1015 oi.typep = &type;
1016 if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
1017 die(_("unable to get type of object %s"), oid_to_hex(oid));
1019 if (type != OBJ_COMMIT)
1020 return 0;
1022 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1023 oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
1024 ctx->oids.nr++;
1026 return 0;
1029 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
1031 struct commit_list *parent;
1032 for (parent = commit->parents; parent; parent = parent->next) {
1033 if (!(parent->item->object.flags & UNINTERESTING)) {
1034 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1035 oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
1036 ctx->oids.nr++;
1037 parent->item->object.flags |= UNINTERESTING;
1042 static void close_reachable(struct write_commit_graph_context *ctx)
1044 int i;
1045 struct commit *commit;
1047 if (ctx->report_progress)
1048 ctx->progress = start_delayed_progress(
1049 _("Loading known commits in commit graph"),
1050 ctx->oids.nr);
1051 for (i = 0; i < ctx->oids.nr; i++) {
1052 display_progress(ctx->progress, i + 1);
1053 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1054 if (commit)
1055 commit->object.flags |= UNINTERESTING;
1057 stop_progress(&ctx->progress);
1060 * As this loop runs, ctx->oids.nr may grow, but not more
1061 * than the number of missing commits in the reachable
1062 * closure.
1064 if (ctx->report_progress)
1065 ctx->progress = start_delayed_progress(
1066 _("Expanding reachable commits in commit graph"),
1068 for (i = 0; i < ctx->oids.nr; i++) {
1069 display_progress(ctx->progress, i + 1);
1070 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1072 if (!commit)
1073 continue;
1074 if (ctx->split) {
1075 if (!parse_commit(commit) &&
1076 commit->graph_pos == COMMIT_NOT_FROM_GRAPH)
1077 add_missing_parents(ctx, commit);
1078 } else if (!parse_commit_no_graph(commit))
1079 add_missing_parents(ctx, commit);
1081 stop_progress(&ctx->progress);
1083 if (ctx->report_progress)
1084 ctx->progress = start_delayed_progress(
1085 _("Clearing commit marks in commit graph"),
1086 ctx->oids.nr);
1087 for (i = 0; i < ctx->oids.nr; i++) {
1088 display_progress(ctx->progress, i + 1);
1089 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1091 if (commit)
1092 commit->object.flags &= ~UNINTERESTING;
1094 stop_progress(&ctx->progress);
1097 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1099 int i;
1100 struct commit_list *list = NULL;
1102 if (ctx->report_progress)
1103 ctx->progress = start_progress(
1104 _("Computing commit graph generation numbers"),
1105 ctx->commits.nr);
1106 for (i = 0; i < ctx->commits.nr; i++) {
1107 display_progress(ctx->progress, i + 1);
1108 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
1109 ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
1110 continue;
1112 commit_list_insert(ctx->commits.list[i], &list);
1113 while (list) {
1114 struct commit *current = list->item;
1115 struct commit_list *parent;
1116 int all_parents_computed = 1;
1117 uint32_t max_generation = 0;
1119 for (parent = current->parents; parent; parent = parent->next) {
1120 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
1121 parent->item->generation == GENERATION_NUMBER_ZERO) {
1122 all_parents_computed = 0;
1123 commit_list_insert(parent->item, &list);
1124 break;
1125 } else if (parent->item->generation > max_generation) {
1126 max_generation = parent->item->generation;
1130 if (all_parents_computed) {
1131 current->generation = max_generation + 1;
1132 pop_commit(&list);
1134 if (current->generation > GENERATION_NUMBER_MAX)
1135 current->generation = GENERATION_NUMBER_MAX;
1139 stop_progress(&ctx->progress);
1142 static int add_ref_to_list(const char *refname,
1143 const struct object_id *oid,
1144 int flags, void *cb_data)
1146 struct string_list *list = (struct string_list *)cb_data;
1148 string_list_append(list, oid_to_hex(oid));
1149 return 0;
1152 int write_commit_graph_reachable(const char *obj_dir,
1153 enum commit_graph_write_flags flags,
1154 const struct split_commit_graph_opts *split_opts)
1156 struct string_list list = STRING_LIST_INIT_DUP;
1157 int result;
1159 for_each_ref(add_ref_to_list, &list);
1160 result = write_commit_graph(obj_dir, NULL, &list,
1161 flags, split_opts);
1163 string_list_clear(&list, 0);
1164 return result;
1167 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
1168 struct string_list *pack_indexes)
1170 uint32_t i;
1171 struct strbuf progress_title = STRBUF_INIT;
1172 struct strbuf packname = STRBUF_INIT;
1173 int dirlen;
1175 strbuf_addf(&packname, "%s/pack/", ctx->obj_dir);
1176 dirlen = packname.len;
1177 if (ctx->report_progress) {
1178 strbuf_addf(&progress_title,
1179 Q_("Finding commits for commit graph in %d pack",
1180 "Finding commits for commit graph in %d packs",
1181 pack_indexes->nr),
1182 pack_indexes->nr);
1183 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1184 ctx->progress_done = 0;
1186 for (i = 0; i < pack_indexes->nr; i++) {
1187 struct packed_git *p;
1188 strbuf_setlen(&packname, dirlen);
1189 strbuf_addstr(&packname, pack_indexes->items[i].string);
1190 p = add_packed_git(packname.buf, packname.len, 1);
1191 if (!p) {
1192 error(_("error adding pack %s"), packname.buf);
1193 return -1;
1195 if (open_pack_index(p)) {
1196 error(_("error opening index for %s"), packname.buf);
1197 return -1;
1199 for_each_object_in_pack(p, add_packed_commits, ctx,
1200 FOR_EACH_OBJECT_PACK_ORDER);
1201 close_pack(p);
1202 free(p);
1205 stop_progress(&ctx->progress);
1206 strbuf_release(&progress_title);
1207 strbuf_release(&packname);
1209 return 0;
1212 static int fill_oids_from_commit_hex(struct write_commit_graph_context *ctx,
1213 struct string_list *commit_hex)
1215 uint32_t i;
1216 struct strbuf progress_title = STRBUF_INIT;
1218 if (ctx->report_progress) {
1219 strbuf_addf(&progress_title,
1220 Q_("Finding commits for commit graph from %d ref",
1221 "Finding commits for commit graph from %d refs",
1222 commit_hex->nr),
1223 commit_hex->nr);
1224 ctx->progress = start_delayed_progress(
1225 progress_title.buf,
1226 commit_hex->nr);
1228 for (i = 0; i < commit_hex->nr; i++) {
1229 const char *end;
1230 struct object_id oid;
1231 struct commit *result;
1233 display_progress(ctx->progress, i + 1);
1234 if (!parse_oid_hex(commit_hex->items[i].string, &oid, &end) &&
1235 (result = lookup_commit_reference_gently(ctx->r, &oid, 1))) {
1236 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1237 oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
1238 ctx->oids.nr++;
1239 } else if (ctx->check_oids) {
1240 error(_("invalid commit object id: %s"),
1241 commit_hex->items[i].string);
1242 return -1;
1245 stop_progress(&ctx->progress);
1246 strbuf_release(&progress_title);
1248 return 0;
1251 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1253 if (ctx->report_progress)
1254 ctx->progress = start_delayed_progress(
1255 _("Finding commits for commit graph among packed objects"),
1256 ctx->approx_nr_objects);
1257 for_each_packed_object(add_packed_commits, ctx,
1258 FOR_EACH_OBJECT_PACK_ORDER);
1259 if (ctx->progress_done < ctx->approx_nr_objects)
1260 display_progress(ctx->progress, ctx->approx_nr_objects);
1261 stop_progress(&ctx->progress);
1264 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
1266 uint32_t i, count_distinct = 1;
1268 if (ctx->report_progress)
1269 ctx->progress = start_delayed_progress(
1270 _("Counting distinct commits in commit graph"),
1271 ctx->oids.nr);
1272 display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1273 QSORT(ctx->oids.list, ctx->oids.nr, oid_compare);
1275 for (i = 1; i < ctx->oids.nr; i++) {
1276 display_progress(ctx->progress, i + 1);
1277 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i])) {
1278 if (ctx->split) {
1279 struct commit *c = lookup_commit(ctx->r, &ctx->oids.list[i]);
1281 if (!c || c->graph_pos != COMMIT_NOT_FROM_GRAPH)
1282 continue;
1285 count_distinct++;
1288 stop_progress(&ctx->progress);
1290 return count_distinct;
1293 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1295 uint32_t i;
1297 ctx->num_extra_edges = 0;
1298 if (ctx->report_progress)
1299 ctx->progress = start_delayed_progress(
1300 _("Finding extra edges in commit graph"),
1301 ctx->oids.nr);
1302 for (i = 0; i < ctx->oids.nr; i++) {
1303 unsigned int num_parents;
1305 display_progress(ctx->progress, i + 1);
1306 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1307 continue;
1309 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + 1, ctx->commits.alloc);
1310 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1312 if (ctx->split &&
1313 ctx->commits.list[ctx->commits.nr]->graph_pos != COMMIT_NOT_FROM_GRAPH)
1314 continue;
1316 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1318 num_parents = commit_list_count(ctx->commits.list[ctx->commits.nr]->parents);
1319 if (num_parents > 2)
1320 ctx->num_extra_edges += num_parents - 1;
1322 ctx->commits.nr++;
1324 stop_progress(&ctx->progress);
1327 static int write_graph_chunk_base_1(struct hashfile *f,
1328 struct commit_graph *g)
1330 int num = 0;
1332 if (!g)
1333 return 0;
1335 num = write_graph_chunk_base_1(f, g->base_graph);
1336 hashwrite(f, g->oid.hash, the_hash_algo->rawsz);
1337 return num + 1;
1340 static int write_graph_chunk_base(struct hashfile *f,
1341 struct write_commit_graph_context *ctx)
1343 int num = write_graph_chunk_base_1(f, ctx->new_base_graph);
1345 if (num != ctx->num_commit_graphs_after - 1) {
1346 error(_("failed to write correct number of base graph ids"));
1347 return -1;
1350 return 0;
1353 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1355 uint32_t i;
1356 int fd;
1357 struct hashfile *f;
1358 struct lock_file lk = LOCK_INIT;
1359 uint32_t chunk_ids[6];
1360 uint64_t chunk_offsets[6];
1361 const unsigned hashsz = the_hash_algo->rawsz;
1362 struct strbuf progress_title = STRBUF_INIT;
1363 int num_chunks = 3;
1364 struct object_id file_hash;
1366 if (ctx->split) {
1367 struct strbuf tmp_file = STRBUF_INIT;
1369 strbuf_addf(&tmp_file,
1370 "%s/info/commit-graphs/tmp_graph_XXXXXX",
1371 ctx->obj_dir);
1372 ctx->graph_name = strbuf_detach(&tmp_file, NULL);
1373 } else {
1374 ctx->graph_name = get_commit_graph_filename(ctx->obj_dir);
1377 if (safe_create_leading_directories(ctx->graph_name)) {
1378 UNLEAK(ctx->graph_name);
1379 error(_("unable to create leading directories of %s"),
1380 ctx->graph_name);
1381 return -1;
1384 if (ctx->split) {
1385 char *lock_name = get_chain_filename(ctx->obj_dir);
1387 hold_lock_file_for_update(&lk, lock_name, LOCK_DIE_ON_ERROR);
1389 fd = git_mkstemp_mode(ctx->graph_name, 0444);
1390 if (fd < 0) {
1391 error(_("unable to create '%s'"), ctx->graph_name);
1392 return -1;
1395 f = hashfd(fd, ctx->graph_name);
1396 } else {
1397 hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
1398 fd = lk.tempfile->fd;
1399 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1402 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1403 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1404 chunk_ids[2] = GRAPH_CHUNKID_DATA;
1405 if (ctx->num_extra_edges) {
1406 chunk_ids[num_chunks] = GRAPH_CHUNKID_EXTRAEDGES;
1407 num_chunks++;
1409 if (ctx->num_commit_graphs_after > 1) {
1410 chunk_ids[num_chunks] = GRAPH_CHUNKID_BASE;
1411 num_chunks++;
1414 chunk_ids[num_chunks] = 0;
1416 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1417 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1418 chunk_offsets[2] = chunk_offsets[1] + hashsz * ctx->commits.nr;
1419 chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * ctx->commits.nr;
1421 num_chunks = 3;
1422 if (ctx->num_extra_edges) {
1423 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1424 4 * ctx->num_extra_edges;
1425 num_chunks++;
1427 if (ctx->num_commit_graphs_after > 1) {
1428 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1429 hashsz * (ctx->num_commit_graphs_after - 1);
1430 num_chunks++;
1433 hashwrite_be32(f, GRAPH_SIGNATURE);
1435 hashwrite_u8(f, GRAPH_VERSION);
1436 hashwrite_u8(f, oid_version());
1437 hashwrite_u8(f, num_chunks);
1438 hashwrite_u8(f, ctx->num_commit_graphs_after - 1);
1440 for (i = 0; i <= num_chunks; i++) {
1441 uint32_t chunk_write[3];
1443 chunk_write[0] = htonl(chunk_ids[i]);
1444 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1445 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1446 hashwrite(f, chunk_write, 12);
1449 if (ctx->report_progress) {
1450 strbuf_addf(&progress_title,
1451 Q_("Writing out commit graph in %d pass",
1452 "Writing out commit graph in %d passes",
1453 num_chunks),
1454 num_chunks);
1455 ctx->progress = start_delayed_progress(
1456 progress_title.buf,
1457 num_chunks * ctx->commits.nr);
1459 write_graph_chunk_fanout(f, ctx);
1460 write_graph_chunk_oids(f, hashsz, ctx);
1461 write_graph_chunk_data(f, hashsz, ctx);
1462 if (ctx->num_extra_edges)
1463 write_graph_chunk_extra_edges(f, ctx);
1464 if (ctx->num_commit_graphs_after > 1 &&
1465 write_graph_chunk_base(f, ctx)) {
1466 return -1;
1468 stop_progress(&ctx->progress);
1469 strbuf_release(&progress_title);
1471 if (ctx->split && ctx->base_graph_name && ctx->num_commit_graphs_after > 1) {
1472 char *new_base_hash = xstrdup(oid_to_hex(&ctx->new_base_graph->oid));
1473 char *new_base_name = get_split_graph_filename(ctx->new_base_graph->obj_dir, new_base_hash);
1475 free(ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2]);
1476 free(ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2]);
1477 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2] = new_base_name;
1478 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2] = new_base_hash;
1481 close_commit_graph(ctx->r->objects);
1482 finalize_hashfile(f, file_hash.hash, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1484 if (ctx->split) {
1485 FILE *chainf = fdopen_lock_file(&lk, "w");
1486 char *final_graph_name;
1487 int result;
1489 close(fd);
1491 if (!chainf) {
1492 error(_("unable to open commit-graph chain file"));
1493 return -1;
1496 if (ctx->base_graph_name) {
1497 const char *dest = ctx->commit_graph_filenames_after[
1498 ctx->num_commit_graphs_after - 2];
1500 if (strcmp(ctx->base_graph_name, dest)) {
1501 result = rename(ctx->base_graph_name, dest);
1503 if (result) {
1504 error(_("failed to rename base commit-graph file"));
1505 return -1;
1508 } else {
1509 char *graph_name = get_commit_graph_filename(ctx->obj_dir);
1510 unlink(graph_name);
1513 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1] = xstrdup(oid_to_hex(&file_hash));
1514 final_graph_name = get_split_graph_filename(ctx->obj_dir,
1515 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1]);
1516 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 1] = final_graph_name;
1518 result = rename(ctx->graph_name, final_graph_name);
1520 for (i = 0; i < ctx->num_commit_graphs_after; i++)
1521 fprintf(lk.tempfile->fp, "%s\n", ctx->commit_graph_hash_after[i]);
1523 if (result) {
1524 error(_("failed to rename temporary commit-graph file"));
1525 return -1;
1529 commit_lock_file(&lk);
1531 return 0;
1534 static void split_graph_merge_strategy(struct write_commit_graph_context *ctx)
1536 struct commit_graph *g;
1537 uint32_t num_commits;
1538 uint32_t i;
1540 int max_commits = 0;
1541 int size_mult = 2;
1543 if (ctx->split_opts) {
1544 max_commits = ctx->split_opts->max_commits;
1545 size_mult = ctx->split_opts->size_multiple;
1548 g = ctx->r->objects->commit_graph;
1549 num_commits = ctx->commits.nr;
1550 ctx->num_commit_graphs_after = ctx->num_commit_graphs_before + 1;
1552 while (g && (g->num_commits <= size_mult * num_commits ||
1553 (max_commits && num_commits > max_commits))) {
1554 if (strcmp(g->obj_dir, ctx->obj_dir))
1555 break;
1557 num_commits += g->num_commits;
1558 g = g->base_graph;
1560 ctx->num_commit_graphs_after--;
1563 ctx->new_base_graph = g;
1565 if (ctx->num_commit_graphs_after == 2) {
1566 char *old_graph_name = get_commit_graph_filename(g->obj_dir);
1568 if (!strcmp(g->filename, old_graph_name) &&
1569 strcmp(g->obj_dir, ctx->obj_dir)) {
1570 ctx->num_commit_graphs_after = 1;
1571 ctx->new_base_graph = NULL;
1574 free(old_graph_name);
1577 ALLOC_ARRAY(ctx->commit_graph_filenames_after, ctx->num_commit_graphs_after);
1578 ALLOC_ARRAY(ctx->commit_graph_hash_after, ctx->num_commit_graphs_after);
1580 for (i = 0; i < ctx->num_commit_graphs_after &&
1581 i < ctx->num_commit_graphs_before; i++)
1582 ctx->commit_graph_filenames_after[i] = xstrdup(ctx->commit_graph_filenames_before[i]);
1584 i = ctx->num_commit_graphs_before - 1;
1585 g = ctx->r->objects->commit_graph;
1587 while (g) {
1588 if (i < ctx->num_commit_graphs_after)
1589 ctx->commit_graph_hash_after[i] = xstrdup(oid_to_hex(&g->oid));
1591 i--;
1592 g = g->base_graph;
1596 static void merge_commit_graph(struct write_commit_graph_context *ctx,
1597 struct commit_graph *g)
1599 uint32_t i;
1600 uint32_t offset = g->num_commits_in_base;
1602 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + g->num_commits, ctx->commits.alloc);
1604 for (i = 0; i < g->num_commits; i++) {
1605 struct object_id oid;
1606 struct commit *result;
1608 display_progress(ctx->progress, i + 1);
1610 load_oid_from_graph(g, i + offset, &oid);
1612 /* only add commits if they still exist in the repo */
1613 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1615 if (result) {
1616 ctx->commits.list[ctx->commits.nr] = result;
1617 ctx->commits.nr++;
1622 static int commit_compare(const void *_a, const void *_b)
1624 const struct commit *a = *(const struct commit **)_a;
1625 const struct commit *b = *(const struct commit **)_b;
1626 return oidcmp(&a->object.oid, &b->object.oid);
1629 static void sort_and_scan_merged_commits(struct write_commit_graph_context *ctx)
1631 uint32_t i;
1633 if (ctx->report_progress)
1634 ctx->progress = start_delayed_progress(
1635 _("Scanning merged commits"),
1636 ctx->commits.nr);
1638 QSORT(ctx->commits.list, ctx->commits.nr, commit_compare);
1640 ctx->num_extra_edges = 0;
1641 for (i = 0; i < ctx->commits.nr; i++) {
1642 display_progress(ctx->progress, i);
1644 if (i && oideq(&ctx->commits.list[i - 1]->object.oid,
1645 &ctx->commits.list[i]->object.oid)) {
1646 die(_("unexpected duplicate commit id %s"),
1647 oid_to_hex(&ctx->commits.list[i]->object.oid));
1648 } else {
1649 unsigned int num_parents;
1651 num_parents = commit_list_count(ctx->commits.list[i]->parents);
1652 if (num_parents > 2)
1653 ctx->num_extra_edges += num_parents - 1;
1657 stop_progress(&ctx->progress);
1660 static void merge_commit_graphs(struct write_commit_graph_context *ctx)
1662 struct commit_graph *g = ctx->r->objects->commit_graph;
1663 uint32_t current_graph_number = ctx->num_commit_graphs_before;
1664 struct strbuf progress_title = STRBUF_INIT;
1666 while (g && current_graph_number >= ctx->num_commit_graphs_after) {
1667 current_graph_number--;
1669 if (ctx->report_progress) {
1670 strbuf_addstr(&progress_title, _("Merging commit-graph"));
1671 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1674 merge_commit_graph(ctx, g);
1675 stop_progress(&ctx->progress);
1676 strbuf_release(&progress_title);
1678 g = g->base_graph;
1681 if (g) {
1682 ctx->new_base_graph = g;
1683 ctx->new_num_commits_in_base = g->num_commits + g->num_commits_in_base;
1686 if (ctx->new_base_graph)
1687 ctx->base_graph_name = xstrdup(ctx->new_base_graph->filename);
1689 sort_and_scan_merged_commits(ctx);
1692 static void mark_commit_graphs(struct write_commit_graph_context *ctx)
1694 uint32_t i;
1695 time_t now = time(NULL);
1697 for (i = ctx->num_commit_graphs_after - 1; i < ctx->num_commit_graphs_before; i++) {
1698 struct stat st;
1699 struct utimbuf updated_time;
1701 stat(ctx->commit_graph_filenames_before[i], &st);
1703 updated_time.actime = st.st_atime;
1704 updated_time.modtime = now;
1705 utime(ctx->commit_graph_filenames_before[i], &updated_time);
1709 static void expire_commit_graphs(struct write_commit_graph_context *ctx)
1711 struct strbuf path = STRBUF_INIT;
1712 DIR *dir;
1713 struct dirent *de;
1714 size_t dirnamelen;
1715 timestamp_t expire_time = time(NULL);
1717 if (ctx->split_opts && ctx->split_opts->expire_time)
1718 expire_time -= ctx->split_opts->expire_time;
1719 if (!ctx->split) {
1720 char *chain_file_name = get_chain_filename(ctx->obj_dir);
1721 unlink(chain_file_name);
1722 free(chain_file_name);
1723 ctx->num_commit_graphs_after = 0;
1726 strbuf_addstr(&path, ctx->obj_dir);
1727 strbuf_addstr(&path, "/info/commit-graphs");
1728 dir = opendir(path.buf);
1730 if (!dir)
1731 goto out;
1733 strbuf_addch(&path, '/');
1734 dirnamelen = path.len;
1735 while ((de = readdir(dir)) != NULL) {
1736 struct stat st;
1737 uint32_t i, found = 0;
1739 strbuf_setlen(&path, dirnamelen);
1740 strbuf_addstr(&path, de->d_name);
1742 stat(path.buf, &st);
1744 if (st.st_mtime > expire_time)
1745 continue;
1746 if (path.len < 6 || strcmp(path.buf + path.len - 6, ".graph"))
1747 continue;
1749 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
1750 if (!strcmp(ctx->commit_graph_filenames_after[i],
1751 path.buf)) {
1752 found = 1;
1753 break;
1757 if (!found)
1758 unlink(path.buf);
1761 out:
1762 strbuf_release(&path);
1765 int write_commit_graph(const char *obj_dir,
1766 struct string_list *pack_indexes,
1767 struct string_list *commit_hex,
1768 enum commit_graph_write_flags flags,
1769 const struct split_commit_graph_opts *split_opts)
1771 struct write_commit_graph_context *ctx;
1772 uint32_t i, count_distinct = 0;
1773 size_t len;
1774 int res = 0;
1776 if (!commit_graph_compatible(the_repository))
1777 return 0;
1779 ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
1780 ctx->r = the_repository;
1782 /* normalize object dir with no trailing slash */
1783 ctx->obj_dir = xmallocz(strlen(obj_dir) + 1);
1784 normalize_path_copy(ctx->obj_dir, obj_dir);
1785 len = strlen(ctx->obj_dir);
1786 if (len && ctx->obj_dir[len - 1] == '/')
1787 ctx->obj_dir[len - 1] = 0;
1789 ctx->append = flags & COMMIT_GRAPH_WRITE_APPEND ? 1 : 0;
1790 ctx->report_progress = flags & COMMIT_GRAPH_WRITE_PROGRESS ? 1 : 0;
1791 ctx->split = flags & COMMIT_GRAPH_WRITE_SPLIT ? 1 : 0;
1792 ctx->check_oids = flags & COMMIT_GRAPH_WRITE_CHECK_OIDS ? 1 : 0;
1793 ctx->split_opts = split_opts;
1795 if (ctx->split) {
1796 struct commit_graph *g;
1797 prepare_commit_graph(ctx->r);
1799 g = ctx->r->objects->commit_graph;
1801 while (g) {
1802 ctx->num_commit_graphs_before++;
1803 g = g->base_graph;
1806 if (ctx->num_commit_graphs_before) {
1807 ALLOC_ARRAY(ctx->commit_graph_filenames_before, ctx->num_commit_graphs_before);
1808 i = ctx->num_commit_graphs_before;
1809 g = ctx->r->objects->commit_graph;
1811 while (g) {
1812 ctx->commit_graph_filenames_before[--i] = xstrdup(g->filename);
1813 g = g->base_graph;
1818 ctx->approx_nr_objects = approximate_object_count();
1819 ctx->oids.alloc = ctx->approx_nr_objects / 32;
1821 if (ctx->split && split_opts && ctx->oids.alloc > split_opts->max_commits)
1822 ctx->oids.alloc = split_opts->max_commits;
1824 if (ctx->append) {
1825 prepare_commit_graph_one(ctx->r, ctx->obj_dir);
1826 if (ctx->r->objects->commit_graph)
1827 ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
1830 if (ctx->oids.alloc < 1024)
1831 ctx->oids.alloc = 1024;
1832 ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
1834 if (ctx->append && ctx->r->objects->commit_graph) {
1835 struct commit_graph *g = ctx->r->objects->commit_graph;
1836 for (i = 0; i < g->num_commits; i++) {
1837 const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
1838 hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
1842 if (pack_indexes) {
1843 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
1844 goto cleanup;
1847 if (commit_hex) {
1848 if ((res = fill_oids_from_commit_hex(ctx, commit_hex)))
1849 goto cleanup;
1852 if (!pack_indexes && !commit_hex)
1853 fill_oids_from_all_packs(ctx);
1855 close_reachable(ctx);
1857 count_distinct = count_distinct_commits(ctx);
1859 if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
1860 error(_("the commit graph format cannot write %d commits"), count_distinct);
1861 res = -1;
1862 goto cleanup;
1865 ctx->commits.alloc = count_distinct;
1866 ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
1868 copy_oids_to_commits(ctx);
1870 if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
1871 error(_("too many commits to write graph"));
1872 res = -1;
1873 goto cleanup;
1876 if (!ctx->commits.nr)
1877 goto cleanup;
1879 if (ctx->split) {
1880 split_graph_merge_strategy(ctx);
1882 merge_commit_graphs(ctx);
1883 } else
1884 ctx->num_commit_graphs_after = 1;
1886 compute_generation_numbers(ctx);
1888 res = write_commit_graph_file(ctx);
1890 if (ctx->split)
1891 mark_commit_graphs(ctx);
1893 expire_commit_graphs(ctx);
1895 cleanup:
1896 free(ctx->graph_name);
1897 free(ctx->commits.list);
1898 free(ctx->oids.list);
1899 free(ctx->obj_dir);
1901 if (ctx->commit_graph_filenames_after) {
1902 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
1903 free(ctx->commit_graph_filenames_after[i]);
1904 free(ctx->commit_graph_hash_after[i]);
1907 for (i = 0; i < ctx->num_commit_graphs_before; i++)
1908 free(ctx->commit_graph_filenames_before[i]);
1910 free(ctx->commit_graph_filenames_after);
1911 free(ctx->commit_graph_filenames_before);
1912 free(ctx->commit_graph_hash_after);
1915 free(ctx);
1917 return res;
1920 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1921 static int verify_commit_graph_error;
1923 static void graph_report(const char *fmt, ...)
1925 va_list ap;
1927 verify_commit_graph_error = 1;
1928 va_start(ap, fmt);
1929 vfprintf(stderr, fmt, ap);
1930 fprintf(stderr, "\n");
1931 va_end(ap);
1934 #define GENERATION_ZERO_EXISTS 1
1935 #define GENERATION_NUMBER_EXISTS 2
1937 int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
1939 uint32_t i, cur_fanout_pos = 0;
1940 struct object_id prev_oid, cur_oid, checksum;
1941 int generation_zero = 0;
1942 struct hashfile *f;
1943 int devnull;
1944 struct progress *progress = NULL;
1945 int local_error = 0;
1947 if (!g) {
1948 graph_report("no commit-graph file loaded");
1949 return 1;
1952 verify_commit_graph_error = verify_commit_graph_lite(g);
1953 if (verify_commit_graph_error)
1954 return verify_commit_graph_error;
1956 devnull = open("/dev/null", O_WRONLY);
1957 f = hashfd(devnull, NULL);
1958 hashwrite(f, g->data, g->data_len - g->hash_len);
1959 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1960 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1961 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1962 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1965 for (i = 0; i < g->num_commits; i++) {
1966 struct commit *graph_commit;
1968 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1970 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1971 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
1972 oid_to_hex(&prev_oid),
1973 oid_to_hex(&cur_oid));
1975 oidcpy(&prev_oid, &cur_oid);
1977 while (cur_oid.hash[0] > cur_fanout_pos) {
1978 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1980 if (i != fanout_value)
1981 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1982 cur_fanout_pos, fanout_value, i);
1983 cur_fanout_pos++;
1986 graph_commit = lookup_commit(r, &cur_oid);
1987 if (!parse_commit_in_graph_one(r, g, graph_commit))
1988 graph_report(_("failed to parse commit %s from commit-graph"),
1989 oid_to_hex(&cur_oid));
1992 while (cur_fanout_pos < 256) {
1993 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1995 if (g->num_commits != fanout_value)
1996 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1997 cur_fanout_pos, fanout_value, i);
1999 cur_fanout_pos++;
2002 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
2003 return verify_commit_graph_error;
2005 if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
2006 progress = start_progress(_("Verifying commits in commit graph"),
2007 g->num_commits);
2009 for (i = 0; i < g->num_commits; i++) {
2010 struct commit *graph_commit, *odb_commit;
2011 struct commit_list *graph_parents, *odb_parents;
2012 uint32_t max_generation = 0;
2014 display_progress(progress, i + 1);
2015 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2017 graph_commit = lookup_commit(r, &cur_oid);
2018 odb_commit = (struct commit *)create_object(r, &cur_oid, alloc_commit_node(r));
2019 if (parse_commit_internal(odb_commit, 0, 0)) {
2020 graph_report(_("failed to parse commit %s from object database for commit-graph"),
2021 oid_to_hex(&cur_oid));
2022 continue;
2025 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
2026 get_commit_tree_oid(odb_commit)))
2027 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2028 oid_to_hex(&cur_oid),
2029 oid_to_hex(get_commit_tree_oid(graph_commit)),
2030 oid_to_hex(get_commit_tree_oid(odb_commit)));
2032 graph_parents = graph_commit->parents;
2033 odb_parents = odb_commit->parents;
2035 while (graph_parents) {
2036 if (odb_parents == NULL) {
2037 graph_report(_("commit-graph parent list for commit %s is too long"),
2038 oid_to_hex(&cur_oid));
2039 break;
2042 /* parse parent in case it is in a base graph */
2043 parse_commit_in_graph_one(r, g, graph_parents->item);
2045 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
2046 graph_report(_("commit-graph parent for %s is %s != %s"),
2047 oid_to_hex(&cur_oid),
2048 oid_to_hex(&graph_parents->item->object.oid),
2049 oid_to_hex(&odb_parents->item->object.oid));
2051 if (graph_parents->item->generation > max_generation)
2052 max_generation = graph_parents->item->generation;
2054 graph_parents = graph_parents->next;
2055 odb_parents = odb_parents->next;
2058 if (odb_parents != NULL)
2059 graph_report(_("commit-graph parent list for commit %s terminates early"),
2060 oid_to_hex(&cur_oid));
2062 if (!graph_commit->generation) {
2063 if (generation_zero == GENERATION_NUMBER_EXISTS)
2064 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2065 oid_to_hex(&cur_oid));
2066 generation_zero = GENERATION_ZERO_EXISTS;
2067 } else if (generation_zero == GENERATION_ZERO_EXISTS)
2068 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2069 oid_to_hex(&cur_oid));
2071 if (generation_zero == GENERATION_ZERO_EXISTS)
2072 continue;
2075 * If one of our parents has generation GENERATION_NUMBER_MAX, then
2076 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
2077 * extra logic in the following condition.
2079 if (max_generation == GENERATION_NUMBER_MAX)
2080 max_generation--;
2082 if (graph_commit->generation != max_generation + 1)
2083 graph_report(_("commit-graph generation for commit %s is %u != %u"),
2084 oid_to_hex(&cur_oid),
2085 graph_commit->generation,
2086 max_generation + 1);
2088 if (graph_commit->date != odb_commit->date)
2089 graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
2090 oid_to_hex(&cur_oid),
2091 graph_commit->date,
2092 odb_commit->date);
2094 stop_progress(&progress);
2096 local_error = verify_commit_graph_error;
2098 if (!(flags & COMMIT_GRAPH_VERIFY_SHALLOW) && g->base_graph)
2099 local_error |= verify_commit_graph(r, g->base_graph, flags);
2101 return local_error;
2104 void free_commit_graph(struct commit_graph *g)
2106 if (!g)
2107 return;
2108 if (g->graph_fd >= 0) {
2109 munmap((void *)g->data, g->data_len);
2110 g->data = NULL;
2111 close(g->graph_fd);
2113 free(g->filename);
2114 free(g);
2117 void disable_commit_graph(struct repository *r)
2119 r->commit_graph_disabled = 1;