Merge branch 'bc/hash-independent-tests-part-4'
[git.git] / commit-graph.c
blobb3c4de79b6da4502726dbec5e21b11734d76e28e
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);
438 return graph_chain;
441 struct commit_graph *read_commit_graph_one(struct repository *r, const char *obj_dir)
443 struct commit_graph *g = load_commit_graph_v1(r, obj_dir);
445 if (!g)
446 g = load_commit_graph_chain(r, obj_dir);
448 return g;
451 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
454 if (r->objects->commit_graph)
455 return;
457 r->objects->commit_graph = read_commit_graph_one(r, obj_dir);
461 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
463 * On the first invocation, this function attemps to load the commit
464 * graph if the_repository is configured to have one.
466 static int prepare_commit_graph(struct repository *r)
468 struct object_directory *odb;
469 int config_value;
471 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
472 die("dying as requested by the '%s' variable on commit-graph load!",
473 GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD);
475 if (r->objects->commit_graph_attempted)
476 return !!r->objects->commit_graph;
477 r->objects->commit_graph_attempted = 1;
479 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
480 (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
481 !config_value))
483 * This repository is not configured to use commit graphs, so
484 * do not load one. (But report commit_graph_attempted anyway
485 * so that commit graph loading is not attempted again for this
486 * repository.)
488 return 0;
490 if (!commit_graph_compatible(r))
491 return 0;
493 prepare_alt_odb(r);
494 for (odb = r->objects->odb;
495 !r->objects->commit_graph && odb;
496 odb = odb->next)
497 prepare_commit_graph_one(r, odb->path);
498 return !!r->objects->commit_graph;
501 int generation_numbers_enabled(struct repository *r)
503 uint32_t first_generation;
504 struct commit_graph *g;
505 if (!prepare_commit_graph(r))
506 return 0;
508 g = r->objects->commit_graph;
510 if (!g->num_commits)
511 return 0;
513 first_generation = get_be32(g->chunk_commit_data +
514 g->hash_len + 8) >> 2;
516 return !!first_generation;
519 static void close_commit_graph_one(struct commit_graph *g)
521 if (!g)
522 return;
524 close_commit_graph_one(g->base_graph);
525 free_commit_graph(g);
528 void close_commit_graph(struct raw_object_store *o)
530 close_commit_graph_one(o->commit_graph);
531 o->commit_graph = NULL;
534 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
536 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
537 g->chunk_oid_lookup, g->hash_len, pos);
540 static void load_oid_from_graph(struct commit_graph *g,
541 uint32_t pos,
542 struct object_id *oid)
544 uint32_t lex_index;
546 while (g && pos < g->num_commits_in_base)
547 g = g->base_graph;
549 if (!g)
550 BUG("NULL commit-graph");
552 if (pos >= g->num_commits + g->num_commits_in_base)
553 die(_("invalid commit position. commit-graph is likely corrupt"));
555 lex_index = pos - g->num_commits_in_base;
557 hashcpy(oid->hash, g->chunk_oid_lookup + g->hash_len * lex_index);
560 static struct commit_list **insert_parent_or_die(struct repository *r,
561 struct commit_graph *g,
562 uint32_t pos,
563 struct commit_list **pptr)
565 struct commit *c;
566 struct object_id oid;
568 if (pos >= g->num_commits + g->num_commits_in_base)
569 die("invalid parent position %"PRIu32, pos);
571 load_oid_from_graph(g, pos, &oid);
572 c = lookup_commit(r, &oid);
573 if (!c)
574 die(_("could not find commit %s"), oid_to_hex(&oid));
575 c->graph_pos = pos;
576 return &commit_list_insert(c, pptr)->next;
579 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
581 const unsigned char *commit_data;
582 uint32_t lex_index;
584 while (pos < g->num_commits_in_base)
585 g = g->base_graph;
587 lex_index = pos - g->num_commits_in_base;
588 commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * lex_index;
589 item->graph_pos = pos;
590 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
593 static inline void set_commit_tree(struct commit *c, struct tree *t)
595 c->maybe_tree = t;
598 static int fill_commit_in_graph(struct repository *r,
599 struct commit *item,
600 struct commit_graph *g, uint32_t pos)
602 uint32_t edge_value;
603 uint32_t *parent_data_ptr;
604 uint64_t date_low, date_high;
605 struct commit_list **pptr;
606 const unsigned char *commit_data;
607 uint32_t lex_index;
609 while (pos < g->num_commits_in_base)
610 g = g->base_graph;
612 if (pos >= g->num_commits + g->num_commits_in_base)
613 die(_("invalid commit position. commit-graph is likely corrupt"));
616 * Store the "full" position, but then use the
617 * "local" position for the rest of the calculation.
619 item->graph_pos = pos;
620 lex_index = pos - g->num_commits_in_base;
622 commit_data = g->chunk_commit_data + (g->hash_len + 16) * lex_index;
624 item->object.parsed = 1;
626 set_commit_tree(item, NULL);
628 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
629 date_low = get_be32(commit_data + g->hash_len + 12);
630 item->date = (timestamp_t)((date_high << 32) | date_low);
632 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
634 pptr = &item->parents;
636 edge_value = get_be32(commit_data + g->hash_len);
637 if (edge_value == GRAPH_PARENT_NONE)
638 return 1;
639 pptr = insert_parent_or_die(r, g, edge_value, pptr);
641 edge_value = get_be32(commit_data + g->hash_len + 4);
642 if (edge_value == GRAPH_PARENT_NONE)
643 return 1;
644 if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
645 pptr = insert_parent_or_die(r, g, edge_value, pptr);
646 return 1;
649 parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
650 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
651 do {
652 edge_value = get_be32(parent_data_ptr);
653 pptr = insert_parent_or_die(r, g,
654 edge_value & GRAPH_EDGE_LAST_MASK,
655 pptr);
656 parent_data_ptr++;
657 } while (!(edge_value & GRAPH_LAST_EDGE));
659 return 1;
662 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
664 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
665 *pos = item->graph_pos;
666 return 1;
667 } else {
668 struct commit_graph *cur_g = g;
669 uint32_t lex_index;
671 while (cur_g && !bsearch_graph(cur_g, &(item->object.oid), &lex_index))
672 cur_g = cur_g->base_graph;
674 if (cur_g) {
675 *pos = lex_index + cur_g->num_commits_in_base;
676 return 1;
679 return 0;
683 static int parse_commit_in_graph_one(struct repository *r,
684 struct commit_graph *g,
685 struct commit *item)
687 uint32_t pos;
689 if (item->object.parsed)
690 return 1;
692 if (find_commit_in_graph(item, g, &pos))
693 return fill_commit_in_graph(r, item, g, pos);
695 return 0;
698 int parse_commit_in_graph(struct repository *r, struct commit *item)
700 if (!prepare_commit_graph(r))
701 return 0;
702 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
705 void load_commit_graph_info(struct repository *r, struct commit *item)
707 uint32_t pos;
708 if (!prepare_commit_graph(r))
709 return;
710 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
711 fill_commit_graph_info(item, r->objects->commit_graph, pos);
714 static struct tree *load_tree_for_commit(struct repository *r,
715 struct commit_graph *g,
716 struct commit *c)
718 struct object_id oid;
719 const unsigned char *commit_data;
721 while (c->graph_pos < g->num_commits_in_base)
722 g = g->base_graph;
724 commit_data = g->chunk_commit_data +
725 GRAPH_DATA_WIDTH * (c->graph_pos - g->num_commits_in_base);
727 hashcpy(oid.hash, commit_data);
728 set_commit_tree(c, lookup_tree(r, &oid));
730 return c->maybe_tree;
733 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
734 struct commit_graph *g,
735 const struct commit *c)
737 if (c->maybe_tree)
738 return c->maybe_tree;
739 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
740 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
742 return load_tree_for_commit(r, g, (struct commit *)c);
745 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
747 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
750 struct packed_commit_list {
751 struct commit **list;
752 int nr;
753 int alloc;
756 struct packed_oid_list {
757 struct object_id *list;
758 int nr;
759 int alloc;
762 struct write_commit_graph_context {
763 struct repository *r;
764 char *obj_dir;
765 char *graph_name;
766 struct packed_oid_list oids;
767 struct packed_commit_list commits;
768 int num_extra_edges;
769 unsigned long approx_nr_objects;
770 struct progress *progress;
771 int progress_done;
772 uint64_t progress_cnt;
774 char *base_graph_name;
775 int num_commit_graphs_before;
776 int num_commit_graphs_after;
777 char **commit_graph_filenames_before;
778 char **commit_graph_filenames_after;
779 char **commit_graph_hash_after;
780 uint32_t new_num_commits_in_base;
781 struct commit_graph *new_base_graph;
783 unsigned append:1,
784 report_progress:1,
785 split:1;
787 const struct split_commit_graph_opts *split_opts;
790 static void write_graph_chunk_fanout(struct hashfile *f,
791 struct write_commit_graph_context *ctx)
793 int i, count = 0;
794 struct commit **list = ctx->commits.list;
797 * Write the first-level table (the list is sorted,
798 * but we use a 256-entry lookup to be able to avoid
799 * having to do eight extra binary search iterations).
801 for (i = 0; i < 256; i++) {
802 while (count < ctx->commits.nr) {
803 if ((*list)->object.oid.hash[0] != i)
804 break;
805 display_progress(ctx->progress, ++ctx->progress_cnt);
806 count++;
807 list++;
810 hashwrite_be32(f, count);
814 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
815 struct write_commit_graph_context *ctx)
817 struct commit **list = ctx->commits.list;
818 int count;
819 for (count = 0; count < ctx->commits.nr; count++, list++) {
820 display_progress(ctx->progress, ++ctx->progress_cnt);
821 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
825 static const unsigned char *commit_to_sha1(size_t index, void *table)
827 struct commit **commits = table;
828 return commits[index]->object.oid.hash;
831 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
832 struct write_commit_graph_context *ctx)
834 struct commit **list = ctx->commits.list;
835 struct commit **last = ctx->commits.list + ctx->commits.nr;
836 uint32_t num_extra_edges = 0;
838 while (list < last) {
839 struct commit_list *parent;
840 int edge_value;
841 uint32_t packedDate[2];
842 display_progress(ctx->progress, ++ctx->progress_cnt);
844 parse_commit_no_graph(*list);
845 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
847 parent = (*list)->parents;
849 if (!parent)
850 edge_value = GRAPH_PARENT_NONE;
851 else {
852 edge_value = sha1_pos(parent->item->object.oid.hash,
853 ctx->commits.list,
854 ctx->commits.nr,
855 commit_to_sha1);
857 if (edge_value >= 0)
858 edge_value += ctx->new_num_commits_in_base;
859 else {
860 uint32_t pos;
861 if (find_commit_in_graph(parent->item,
862 ctx->new_base_graph,
863 &pos))
864 edge_value = pos;
867 if (edge_value < 0)
868 BUG("missing parent %s for commit %s",
869 oid_to_hex(&parent->item->object.oid),
870 oid_to_hex(&(*list)->object.oid));
873 hashwrite_be32(f, edge_value);
875 if (parent)
876 parent = parent->next;
878 if (!parent)
879 edge_value = GRAPH_PARENT_NONE;
880 else if (parent->next)
881 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
882 else {
883 edge_value = sha1_pos(parent->item->object.oid.hash,
884 ctx->commits.list,
885 ctx->commits.nr,
886 commit_to_sha1);
888 if (edge_value >= 0)
889 edge_value += ctx->new_num_commits_in_base;
890 else {
891 uint32_t pos;
892 if (find_commit_in_graph(parent->item,
893 ctx->new_base_graph,
894 &pos))
895 edge_value = pos;
898 if (edge_value < 0)
899 BUG("missing parent %s for commit %s",
900 oid_to_hex(&parent->item->object.oid),
901 oid_to_hex(&(*list)->object.oid));
904 hashwrite_be32(f, edge_value);
906 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
907 do {
908 num_extra_edges++;
909 parent = parent->next;
910 } while (parent);
913 if (sizeof((*list)->date) > 4)
914 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
915 else
916 packedDate[0] = 0;
918 packedDate[0] |= htonl((*list)->generation << 2);
920 packedDate[1] = htonl((*list)->date);
921 hashwrite(f, packedDate, 8);
923 list++;
927 static void write_graph_chunk_extra_edges(struct hashfile *f,
928 struct write_commit_graph_context *ctx)
930 struct commit **list = ctx->commits.list;
931 struct commit **last = ctx->commits.list + ctx->commits.nr;
932 struct commit_list *parent;
934 while (list < last) {
935 int num_parents = 0;
937 display_progress(ctx->progress, ++ctx->progress_cnt);
939 for (parent = (*list)->parents; num_parents < 3 && parent;
940 parent = parent->next)
941 num_parents++;
943 if (num_parents <= 2) {
944 list++;
945 continue;
948 /* Since num_parents > 2, this initializer is safe. */
949 for (parent = (*list)->parents->next; parent; parent = parent->next) {
950 int edge_value = sha1_pos(parent->item->object.oid.hash,
951 ctx->commits.list,
952 ctx->commits.nr,
953 commit_to_sha1);
955 if (edge_value >= 0)
956 edge_value += ctx->new_num_commits_in_base;
957 else {
958 uint32_t pos;
959 if (find_commit_in_graph(parent->item,
960 ctx->new_base_graph,
961 &pos))
962 edge_value = pos;
965 if (edge_value < 0)
966 BUG("missing parent %s for commit %s",
967 oid_to_hex(&parent->item->object.oid),
968 oid_to_hex(&(*list)->object.oid));
969 else if (!parent->next)
970 edge_value |= GRAPH_LAST_EDGE;
972 hashwrite_be32(f, edge_value);
975 list++;
979 static int oid_compare(const void *_a, const void *_b)
981 const struct object_id *a = (const struct object_id *)_a;
982 const struct object_id *b = (const struct object_id *)_b;
983 return oidcmp(a, b);
986 static int add_packed_commits(const struct object_id *oid,
987 struct packed_git *pack,
988 uint32_t pos,
989 void *data)
991 struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
992 enum object_type type;
993 off_t offset = nth_packed_object_offset(pack, pos);
994 struct object_info oi = OBJECT_INFO_INIT;
996 if (ctx->progress)
997 display_progress(ctx->progress, ++ctx->progress_done);
999 oi.typep = &type;
1000 if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
1001 die(_("unable to get type of object %s"), oid_to_hex(oid));
1003 if (type != OBJ_COMMIT)
1004 return 0;
1006 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1007 oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
1008 ctx->oids.nr++;
1010 return 0;
1013 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
1015 struct commit_list *parent;
1016 for (parent = commit->parents; parent; parent = parent->next) {
1017 if (!(parent->item->object.flags & UNINTERESTING)) {
1018 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1019 oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
1020 ctx->oids.nr++;
1021 parent->item->object.flags |= UNINTERESTING;
1026 static void close_reachable(struct write_commit_graph_context *ctx)
1028 int i;
1029 struct commit *commit;
1031 if (ctx->report_progress)
1032 ctx->progress = start_delayed_progress(
1033 _("Loading known commits in commit graph"),
1034 ctx->oids.nr);
1035 for (i = 0; i < ctx->oids.nr; i++) {
1036 display_progress(ctx->progress, i + 1);
1037 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1038 if (commit)
1039 commit->object.flags |= UNINTERESTING;
1041 stop_progress(&ctx->progress);
1044 * As this loop runs, ctx->oids.nr may grow, but not more
1045 * than the number of missing commits in the reachable
1046 * closure.
1048 if (ctx->report_progress)
1049 ctx->progress = start_delayed_progress(
1050 _("Expanding reachable commits in commit graph"),
1051 ctx->oids.nr);
1052 for (i = 0; i < ctx->oids.nr; i++) {
1053 display_progress(ctx->progress, i + 1);
1054 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1056 if (!commit)
1057 continue;
1058 if (ctx->split) {
1059 if (!parse_commit(commit) &&
1060 commit->graph_pos == COMMIT_NOT_FROM_GRAPH)
1061 add_missing_parents(ctx, commit);
1062 } else if (!parse_commit_no_graph(commit))
1063 add_missing_parents(ctx, commit);
1065 stop_progress(&ctx->progress);
1067 if (ctx->report_progress)
1068 ctx->progress = start_delayed_progress(
1069 _("Clearing commit marks in commit graph"),
1070 ctx->oids.nr);
1071 for (i = 0; i < ctx->oids.nr; i++) {
1072 display_progress(ctx->progress, i + 1);
1073 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1075 if (commit)
1076 commit->object.flags &= ~UNINTERESTING;
1078 stop_progress(&ctx->progress);
1081 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1083 int i;
1084 struct commit_list *list = NULL;
1086 if (ctx->report_progress)
1087 ctx->progress = start_progress(
1088 _("Computing commit graph generation numbers"),
1089 ctx->commits.nr);
1090 for (i = 0; i < ctx->commits.nr; i++) {
1091 display_progress(ctx->progress, i + 1);
1092 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
1093 ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
1094 continue;
1096 commit_list_insert(ctx->commits.list[i], &list);
1097 while (list) {
1098 struct commit *current = list->item;
1099 struct commit_list *parent;
1100 int all_parents_computed = 1;
1101 uint32_t max_generation = 0;
1103 for (parent = current->parents; parent; parent = parent->next) {
1104 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
1105 parent->item->generation == GENERATION_NUMBER_ZERO) {
1106 all_parents_computed = 0;
1107 commit_list_insert(parent->item, &list);
1108 break;
1109 } else if (parent->item->generation > max_generation) {
1110 max_generation = parent->item->generation;
1114 if (all_parents_computed) {
1115 current->generation = max_generation + 1;
1116 pop_commit(&list);
1118 if (current->generation > GENERATION_NUMBER_MAX)
1119 current->generation = GENERATION_NUMBER_MAX;
1123 stop_progress(&ctx->progress);
1126 static int add_ref_to_list(const char *refname,
1127 const struct object_id *oid,
1128 int flags, void *cb_data)
1130 struct string_list *list = (struct string_list *)cb_data;
1132 string_list_append(list, oid_to_hex(oid));
1133 return 0;
1136 int write_commit_graph_reachable(const char *obj_dir, unsigned int flags,
1137 const struct split_commit_graph_opts *split_opts)
1139 struct string_list list = STRING_LIST_INIT_DUP;
1140 int result;
1142 for_each_ref(add_ref_to_list, &list);
1143 result = write_commit_graph(obj_dir, NULL, &list,
1144 flags, split_opts);
1146 string_list_clear(&list, 0);
1147 return result;
1150 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
1151 struct string_list *pack_indexes)
1153 uint32_t i;
1154 struct strbuf progress_title = STRBUF_INIT;
1155 struct strbuf packname = STRBUF_INIT;
1156 int dirlen;
1158 strbuf_addf(&packname, "%s/pack/", ctx->obj_dir);
1159 dirlen = packname.len;
1160 if (ctx->report_progress) {
1161 strbuf_addf(&progress_title,
1162 Q_("Finding commits for commit graph in %d pack",
1163 "Finding commits for commit graph in %d packs",
1164 pack_indexes->nr),
1165 pack_indexes->nr);
1166 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1167 ctx->progress_done = 0;
1169 for (i = 0; i < pack_indexes->nr; i++) {
1170 struct packed_git *p;
1171 strbuf_setlen(&packname, dirlen);
1172 strbuf_addstr(&packname, pack_indexes->items[i].string);
1173 p = add_packed_git(packname.buf, packname.len, 1);
1174 if (!p) {
1175 error(_("error adding pack %s"), packname.buf);
1176 return -1;
1178 if (open_pack_index(p)) {
1179 error(_("error opening index for %s"), packname.buf);
1180 return -1;
1182 for_each_object_in_pack(p, add_packed_commits, ctx,
1183 FOR_EACH_OBJECT_PACK_ORDER);
1184 close_pack(p);
1185 free(p);
1188 stop_progress(&ctx->progress);
1189 strbuf_reset(&progress_title);
1190 strbuf_release(&packname);
1192 return 0;
1195 static void fill_oids_from_commit_hex(struct write_commit_graph_context *ctx,
1196 struct string_list *commit_hex)
1198 uint32_t i;
1199 struct strbuf progress_title = STRBUF_INIT;
1201 if (ctx->report_progress) {
1202 strbuf_addf(&progress_title,
1203 Q_("Finding commits for commit graph from %d ref",
1204 "Finding commits for commit graph from %d refs",
1205 commit_hex->nr),
1206 commit_hex->nr);
1207 ctx->progress = start_delayed_progress(
1208 progress_title.buf,
1209 commit_hex->nr);
1211 for (i = 0; i < commit_hex->nr; i++) {
1212 const char *end;
1213 struct object_id oid;
1214 struct commit *result;
1216 display_progress(ctx->progress, i + 1);
1217 if (commit_hex->items[i].string &&
1218 parse_oid_hex(commit_hex->items[i].string, &oid, &end))
1219 continue;
1221 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1223 if (result) {
1224 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1225 oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
1226 ctx->oids.nr++;
1229 stop_progress(&ctx->progress);
1230 strbuf_release(&progress_title);
1233 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1235 if (ctx->report_progress)
1236 ctx->progress = start_delayed_progress(
1237 _("Finding commits for commit graph among packed objects"),
1238 ctx->approx_nr_objects);
1239 for_each_packed_object(add_packed_commits, ctx,
1240 FOR_EACH_OBJECT_PACK_ORDER);
1241 if (ctx->progress_done < ctx->approx_nr_objects)
1242 display_progress(ctx->progress, ctx->approx_nr_objects);
1243 stop_progress(&ctx->progress);
1246 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
1248 uint32_t i, count_distinct = 1;
1250 if (ctx->report_progress)
1251 ctx->progress = start_delayed_progress(
1252 _("Counting distinct commits in commit graph"),
1253 ctx->oids.nr);
1254 display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1255 QSORT(ctx->oids.list, ctx->oids.nr, oid_compare);
1257 for (i = 1; i < ctx->oids.nr; i++) {
1258 display_progress(ctx->progress, i + 1);
1259 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i])) {
1260 if (ctx->split) {
1261 struct commit *c = lookup_commit(ctx->r, &ctx->oids.list[i]);
1263 if (!c || c->graph_pos != COMMIT_NOT_FROM_GRAPH)
1264 continue;
1267 count_distinct++;
1270 stop_progress(&ctx->progress);
1272 return count_distinct;
1275 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1277 uint32_t i;
1278 struct commit_list *parent;
1280 ctx->num_extra_edges = 0;
1281 if (ctx->report_progress)
1282 ctx->progress = start_delayed_progress(
1283 _("Finding extra edges in commit graph"),
1284 ctx->oids.nr);
1285 for (i = 0; i < ctx->oids.nr; i++) {
1286 int num_parents = 0;
1287 display_progress(ctx->progress, i + 1);
1288 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1289 continue;
1291 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + 1, ctx->commits.alloc);
1292 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1294 if (ctx->split &&
1295 ctx->commits.list[ctx->commits.nr]->graph_pos != COMMIT_NOT_FROM_GRAPH)
1296 continue;
1298 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1300 for (parent = ctx->commits.list[ctx->commits.nr]->parents;
1301 parent; parent = parent->next)
1302 num_parents++;
1304 if (num_parents > 2)
1305 ctx->num_extra_edges += num_parents - 1;
1307 ctx->commits.nr++;
1309 stop_progress(&ctx->progress);
1312 static int write_graph_chunk_base_1(struct hashfile *f,
1313 struct commit_graph *g)
1315 int num = 0;
1317 if (!g)
1318 return 0;
1320 num = write_graph_chunk_base_1(f, g->base_graph);
1321 hashwrite(f, g->oid.hash, the_hash_algo->rawsz);
1322 return num + 1;
1325 static int write_graph_chunk_base(struct hashfile *f,
1326 struct write_commit_graph_context *ctx)
1328 int num = write_graph_chunk_base_1(f, ctx->new_base_graph);
1330 if (num != ctx->num_commit_graphs_after - 1) {
1331 error(_("failed to write correct number of base graph ids"));
1332 return -1;
1335 return 0;
1338 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1340 uint32_t i;
1341 int fd;
1342 struct hashfile *f;
1343 struct lock_file lk = LOCK_INIT;
1344 uint32_t chunk_ids[6];
1345 uint64_t chunk_offsets[6];
1346 const unsigned hashsz = the_hash_algo->rawsz;
1347 struct strbuf progress_title = STRBUF_INIT;
1348 int num_chunks = 3;
1349 struct object_id file_hash;
1351 if (ctx->split) {
1352 struct strbuf tmp_file = STRBUF_INIT;
1354 strbuf_addf(&tmp_file,
1355 "%s/info/commit-graphs/tmp_graph_XXXXXX",
1356 ctx->obj_dir);
1357 ctx->graph_name = strbuf_detach(&tmp_file, NULL);
1358 } else {
1359 ctx->graph_name = get_commit_graph_filename(ctx->obj_dir);
1362 if (safe_create_leading_directories(ctx->graph_name)) {
1363 UNLEAK(ctx->graph_name);
1364 error(_("unable to create leading directories of %s"),
1365 ctx->graph_name);
1366 return -1;
1369 if (ctx->split) {
1370 char *lock_name = get_chain_filename(ctx->obj_dir);
1372 hold_lock_file_for_update(&lk, lock_name, LOCK_DIE_ON_ERROR);
1374 fd = git_mkstemp_mode(ctx->graph_name, 0444);
1375 if (fd < 0) {
1376 error(_("unable to create '%s'"), ctx->graph_name);
1377 return -1;
1380 f = hashfd(fd, ctx->graph_name);
1381 } else {
1382 hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
1383 fd = lk.tempfile->fd;
1384 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1387 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1388 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1389 chunk_ids[2] = GRAPH_CHUNKID_DATA;
1390 if (ctx->num_extra_edges) {
1391 chunk_ids[num_chunks] = GRAPH_CHUNKID_EXTRAEDGES;
1392 num_chunks++;
1394 if (ctx->num_commit_graphs_after > 1) {
1395 chunk_ids[num_chunks] = GRAPH_CHUNKID_BASE;
1396 num_chunks++;
1399 chunk_ids[num_chunks] = 0;
1401 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1402 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1403 chunk_offsets[2] = chunk_offsets[1] + hashsz * ctx->commits.nr;
1404 chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * ctx->commits.nr;
1406 num_chunks = 3;
1407 if (ctx->num_extra_edges) {
1408 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1409 4 * ctx->num_extra_edges;
1410 num_chunks++;
1412 if (ctx->num_commit_graphs_after > 1) {
1413 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1414 hashsz * (ctx->num_commit_graphs_after - 1);
1415 num_chunks++;
1418 hashwrite_be32(f, GRAPH_SIGNATURE);
1420 hashwrite_u8(f, GRAPH_VERSION);
1421 hashwrite_u8(f, oid_version());
1422 hashwrite_u8(f, num_chunks);
1423 hashwrite_u8(f, ctx->num_commit_graphs_after - 1);
1425 for (i = 0; i <= num_chunks; i++) {
1426 uint32_t chunk_write[3];
1428 chunk_write[0] = htonl(chunk_ids[i]);
1429 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1430 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1431 hashwrite(f, chunk_write, 12);
1434 if (ctx->report_progress) {
1435 strbuf_addf(&progress_title,
1436 Q_("Writing out commit graph in %d pass",
1437 "Writing out commit graph in %d passes",
1438 num_chunks),
1439 num_chunks);
1440 ctx->progress = start_delayed_progress(
1441 progress_title.buf,
1442 num_chunks * ctx->commits.nr);
1444 write_graph_chunk_fanout(f, ctx);
1445 write_graph_chunk_oids(f, hashsz, ctx);
1446 write_graph_chunk_data(f, hashsz, ctx);
1447 if (ctx->num_extra_edges)
1448 write_graph_chunk_extra_edges(f, ctx);
1449 if (ctx->num_commit_graphs_after > 1 &&
1450 write_graph_chunk_base(f, ctx)) {
1451 return -1;
1453 stop_progress(&ctx->progress);
1454 strbuf_release(&progress_title);
1456 if (ctx->split && ctx->base_graph_name && ctx->num_commit_graphs_after > 1) {
1457 char *new_base_hash = xstrdup(oid_to_hex(&ctx->new_base_graph->oid));
1458 char *new_base_name = get_split_graph_filename(ctx->new_base_graph->obj_dir, new_base_hash);
1460 free(ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2]);
1461 free(ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2]);
1462 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2] = new_base_name;
1463 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2] = new_base_hash;
1466 close_commit_graph(ctx->r->objects);
1467 finalize_hashfile(f, file_hash.hash, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1469 if (ctx->split) {
1470 FILE *chainf = fdopen_lock_file(&lk, "w");
1471 char *final_graph_name;
1472 int result;
1474 close(fd);
1476 if (!chainf) {
1477 error(_("unable to open commit-graph chain file"));
1478 return -1;
1481 if (ctx->base_graph_name) {
1482 const char *dest = ctx->commit_graph_filenames_after[
1483 ctx->num_commit_graphs_after - 2];
1485 if (strcmp(ctx->base_graph_name, dest)) {
1486 result = rename(ctx->base_graph_name, dest);
1488 if (result) {
1489 error(_("failed to rename base commit-graph file"));
1490 return -1;
1493 } else {
1494 char *graph_name = get_commit_graph_filename(ctx->obj_dir);
1495 unlink(graph_name);
1498 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1] = xstrdup(oid_to_hex(&file_hash));
1499 final_graph_name = get_split_graph_filename(ctx->obj_dir,
1500 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1]);
1501 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 1] = final_graph_name;
1503 result = rename(ctx->graph_name, final_graph_name);
1505 for (i = 0; i < ctx->num_commit_graphs_after; i++)
1506 fprintf(lk.tempfile->fp, "%s\n", ctx->commit_graph_hash_after[i]);
1508 if (result) {
1509 error(_("failed to rename temporary commit-graph file"));
1510 return -1;
1514 commit_lock_file(&lk);
1516 return 0;
1519 static void split_graph_merge_strategy(struct write_commit_graph_context *ctx)
1521 struct commit_graph *g = ctx->r->objects->commit_graph;
1522 uint32_t num_commits = ctx->commits.nr;
1523 uint32_t i;
1525 int max_commits = 0;
1526 int size_mult = 2;
1528 if (ctx->split_opts) {
1529 max_commits = ctx->split_opts->max_commits;
1530 size_mult = ctx->split_opts->size_multiple;
1533 g = ctx->r->objects->commit_graph;
1534 ctx->num_commit_graphs_after = ctx->num_commit_graphs_before + 1;
1536 while (g && (g->num_commits <= size_mult * num_commits ||
1537 (max_commits && num_commits > max_commits))) {
1538 if (strcmp(g->obj_dir, ctx->obj_dir))
1539 break;
1541 num_commits += g->num_commits;
1542 g = g->base_graph;
1544 ctx->num_commit_graphs_after--;
1547 ctx->new_base_graph = g;
1549 if (ctx->num_commit_graphs_after == 2) {
1550 char *old_graph_name = get_commit_graph_filename(g->obj_dir);
1552 if (!strcmp(g->filename, old_graph_name) &&
1553 strcmp(g->obj_dir, ctx->obj_dir)) {
1554 ctx->num_commit_graphs_after = 1;
1555 ctx->new_base_graph = NULL;
1558 free(old_graph_name);
1561 ALLOC_ARRAY(ctx->commit_graph_filenames_after, ctx->num_commit_graphs_after);
1562 ALLOC_ARRAY(ctx->commit_graph_hash_after, ctx->num_commit_graphs_after);
1564 for (i = 0; i < ctx->num_commit_graphs_after &&
1565 i < ctx->num_commit_graphs_before; i++)
1566 ctx->commit_graph_filenames_after[i] = xstrdup(ctx->commit_graph_filenames_before[i]);
1568 i = ctx->num_commit_graphs_before - 1;
1569 g = ctx->r->objects->commit_graph;
1571 while (g) {
1572 if (i < ctx->num_commit_graphs_after)
1573 ctx->commit_graph_hash_after[i] = xstrdup(oid_to_hex(&g->oid));
1575 i--;
1576 g = g->base_graph;
1580 static void merge_commit_graph(struct write_commit_graph_context *ctx,
1581 struct commit_graph *g)
1583 uint32_t i;
1584 uint32_t offset = g->num_commits_in_base;
1586 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + g->num_commits, ctx->commits.alloc);
1588 for (i = 0; i < g->num_commits; i++) {
1589 struct object_id oid;
1590 struct commit *result;
1592 display_progress(ctx->progress, i + 1);
1594 load_oid_from_graph(g, i + offset, &oid);
1596 /* only add commits if they still exist in the repo */
1597 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1599 if (result) {
1600 ctx->commits.list[ctx->commits.nr] = result;
1601 ctx->commits.nr++;
1606 static int commit_compare(const void *_a, const void *_b)
1608 const struct commit *a = *(const struct commit **)_a;
1609 const struct commit *b = *(const struct commit **)_b;
1610 return oidcmp(&a->object.oid, &b->object.oid);
1613 static void sort_and_scan_merged_commits(struct write_commit_graph_context *ctx)
1615 uint32_t i, num_parents;
1616 struct commit_list *parent;
1618 if (ctx->report_progress)
1619 ctx->progress = start_delayed_progress(
1620 _("Scanning merged commits"),
1621 ctx->commits.nr);
1623 QSORT(ctx->commits.list, ctx->commits.nr, commit_compare);
1625 ctx->num_extra_edges = 0;
1626 for (i = 0; i < ctx->commits.nr; i++) {
1627 display_progress(ctx->progress, i);
1629 if (i && oideq(&ctx->commits.list[i - 1]->object.oid,
1630 &ctx->commits.list[i]->object.oid)) {
1631 die(_("unexpected duplicate commit id %s"),
1632 oid_to_hex(&ctx->commits.list[i]->object.oid));
1633 } else {
1634 num_parents = 0;
1635 for (parent = ctx->commits.list[i]->parents; parent; parent = parent->next)
1636 num_parents++;
1638 if (num_parents > 2)
1639 ctx->num_extra_edges += num_parents - 2;
1643 stop_progress(&ctx->progress);
1646 static void merge_commit_graphs(struct write_commit_graph_context *ctx)
1648 struct commit_graph *g = ctx->r->objects->commit_graph;
1649 uint32_t current_graph_number = ctx->num_commit_graphs_before;
1650 struct strbuf progress_title = STRBUF_INIT;
1652 while (g && current_graph_number >= ctx->num_commit_graphs_after) {
1653 current_graph_number--;
1655 if (ctx->report_progress) {
1656 strbuf_addstr(&progress_title, _("Merging commit-graph"));
1657 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1660 merge_commit_graph(ctx, g);
1661 stop_progress(&ctx->progress);
1662 strbuf_release(&progress_title);
1664 g = g->base_graph;
1667 if (g) {
1668 ctx->new_base_graph = g;
1669 ctx->new_num_commits_in_base = g->num_commits + g->num_commits_in_base;
1672 if (ctx->new_base_graph)
1673 ctx->base_graph_name = xstrdup(ctx->new_base_graph->filename);
1675 sort_and_scan_merged_commits(ctx);
1678 static void mark_commit_graphs(struct write_commit_graph_context *ctx)
1680 uint32_t i;
1681 time_t now = time(NULL);
1683 for (i = ctx->num_commit_graphs_after - 1; i < ctx->num_commit_graphs_before; i++) {
1684 struct stat st;
1685 struct utimbuf updated_time;
1687 stat(ctx->commit_graph_filenames_before[i], &st);
1689 updated_time.actime = st.st_atime;
1690 updated_time.modtime = now;
1691 utime(ctx->commit_graph_filenames_before[i], &updated_time);
1695 static void expire_commit_graphs(struct write_commit_graph_context *ctx)
1697 struct strbuf path = STRBUF_INIT;
1698 DIR *dir;
1699 struct dirent *de;
1700 size_t dirnamelen;
1701 timestamp_t expire_time = time(NULL);
1703 if (ctx->split_opts && ctx->split_opts->expire_time)
1704 expire_time -= ctx->split_opts->expire_time;
1705 if (!ctx->split) {
1706 char *chain_file_name = get_chain_filename(ctx->obj_dir);
1707 unlink(chain_file_name);
1708 free(chain_file_name);
1709 ctx->num_commit_graphs_after = 0;
1712 strbuf_addstr(&path, ctx->obj_dir);
1713 strbuf_addstr(&path, "/info/commit-graphs");
1714 dir = opendir(path.buf);
1716 if (!dir) {
1717 strbuf_release(&path);
1718 return;
1721 strbuf_addch(&path, '/');
1722 dirnamelen = path.len;
1723 while ((de = readdir(dir)) != NULL) {
1724 struct stat st;
1725 uint32_t i, found = 0;
1727 strbuf_setlen(&path, dirnamelen);
1728 strbuf_addstr(&path, de->d_name);
1730 stat(path.buf, &st);
1732 if (st.st_mtime > expire_time)
1733 continue;
1734 if (path.len < 6 || strcmp(path.buf + path.len - 6, ".graph"))
1735 continue;
1737 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
1738 if (!strcmp(ctx->commit_graph_filenames_after[i],
1739 path.buf)) {
1740 found = 1;
1741 break;
1745 if (!found)
1746 unlink(path.buf);
1750 int write_commit_graph(const char *obj_dir,
1751 struct string_list *pack_indexes,
1752 struct string_list *commit_hex,
1753 unsigned int flags,
1754 const struct split_commit_graph_opts *split_opts)
1756 struct write_commit_graph_context *ctx;
1757 uint32_t i, count_distinct = 0;
1758 size_t len;
1759 int res = 0;
1761 if (!commit_graph_compatible(the_repository))
1762 return 0;
1764 ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
1765 ctx->r = the_repository;
1767 /* normalize object dir with no trailing slash */
1768 ctx->obj_dir = xmallocz(strlen(obj_dir) + 1);
1769 normalize_path_copy(ctx->obj_dir, obj_dir);
1770 len = strlen(ctx->obj_dir);
1771 if (len && ctx->obj_dir[len - 1] == '/')
1772 ctx->obj_dir[len - 1] = 0;
1774 ctx->append = flags & COMMIT_GRAPH_APPEND ? 1 : 0;
1775 ctx->report_progress = flags & COMMIT_GRAPH_PROGRESS ? 1 : 0;
1776 ctx->split = flags & COMMIT_GRAPH_SPLIT ? 1 : 0;
1777 ctx->split_opts = split_opts;
1779 if (ctx->split) {
1780 struct commit_graph *g;
1781 prepare_commit_graph(ctx->r);
1783 g = ctx->r->objects->commit_graph;
1785 while (g) {
1786 ctx->num_commit_graphs_before++;
1787 g = g->base_graph;
1790 if (ctx->num_commit_graphs_before) {
1791 ALLOC_ARRAY(ctx->commit_graph_filenames_before, ctx->num_commit_graphs_before);
1792 i = ctx->num_commit_graphs_before;
1793 g = ctx->r->objects->commit_graph;
1795 while (g) {
1796 ctx->commit_graph_filenames_before[--i] = xstrdup(g->filename);
1797 g = g->base_graph;
1802 ctx->approx_nr_objects = approximate_object_count();
1803 ctx->oids.alloc = ctx->approx_nr_objects / 32;
1805 if (ctx->split && split_opts && ctx->oids.alloc > split_opts->max_commits)
1806 ctx->oids.alloc = split_opts->max_commits;
1808 if (ctx->append) {
1809 prepare_commit_graph_one(ctx->r, ctx->obj_dir);
1810 if (ctx->r->objects->commit_graph)
1811 ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
1814 if (ctx->oids.alloc < 1024)
1815 ctx->oids.alloc = 1024;
1816 ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
1818 if (ctx->append && ctx->r->objects->commit_graph) {
1819 struct commit_graph *g = ctx->r->objects->commit_graph;
1820 for (i = 0; i < g->num_commits; i++) {
1821 const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
1822 hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
1826 if (pack_indexes) {
1827 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
1828 goto cleanup;
1831 if (commit_hex)
1832 fill_oids_from_commit_hex(ctx, commit_hex);
1834 if (!pack_indexes && !commit_hex)
1835 fill_oids_from_all_packs(ctx);
1837 close_reachable(ctx);
1839 count_distinct = count_distinct_commits(ctx);
1841 if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
1842 error(_("the commit graph format cannot write %d commits"), count_distinct);
1843 res = -1;
1844 goto cleanup;
1847 ctx->commits.alloc = count_distinct;
1848 ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
1850 copy_oids_to_commits(ctx);
1852 if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
1853 error(_("too many commits to write graph"));
1854 res = -1;
1855 goto cleanup;
1858 if (!ctx->commits.nr)
1859 goto cleanup;
1861 if (ctx->split) {
1862 split_graph_merge_strategy(ctx);
1864 merge_commit_graphs(ctx);
1865 } else
1866 ctx->num_commit_graphs_after = 1;
1868 compute_generation_numbers(ctx);
1870 res = write_commit_graph_file(ctx);
1872 if (ctx->split)
1873 mark_commit_graphs(ctx);
1875 expire_commit_graphs(ctx);
1877 cleanup:
1878 free(ctx->graph_name);
1879 free(ctx->commits.list);
1880 free(ctx->oids.list);
1881 free(ctx->obj_dir);
1883 if (ctx->commit_graph_filenames_after) {
1884 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
1885 free(ctx->commit_graph_filenames_after[i]);
1886 free(ctx->commit_graph_hash_after[i]);
1889 for (i = 0; i < ctx->num_commit_graphs_before; i++)
1890 free(ctx->commit_graph_filenames_before[i]);
1892 free(ctx->commit_graph_filenames_after);
1893 free(ctx->commit_graph_filenames_before);
1894 free(ctx->commit_graph_hash_after);
1897 free(ctx);
1899 return res;
1902 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1903 static int verify_commit_graph_error;
1905 static void graph_report(const char *fmt, ...)
1907 va_list ap;
1909 verify_commit_graph_error = 1;
1910 va_start(ap, fmt);
1911 vfprintf(stderr, fmt, ap);
1912 fprintf(stderr, "\n");
1913 va_end(ap);
1916 #define GENERATION_ZERO_EXISTS 1
1917 #define GENERATION_NUMBER_EXISTS 2
1919 int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
1921 uint32_t i, cur_fanout_pos = 0;
1922 struct object_id prev_oid, cur_oid, checksum;
1923 int generation_zero = 0;
1924 struct hashfile *f;
1925 int devnull;
1926 struct progress *progress = NULL;
1927 int local_error = 0;
1929 if (!g) {
1930 graph_report("no commit-graph file loaded");
1931 return 1;
1934 verify_commit_graph_error = verify_commit_graph_lite(g);
1935 if (verify_commit_graph_error)
1936 return verify_commit_graph_error;
1938 devnull = open("/dev/null", O_WRONLY);
1939 f = hashfd(devnull, NULL);
1940 hashwrite(f, g->data, g->data_len - g->hash_len);
1941 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1942 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1943 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1944 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1947 for (i = 0; i < g->num_commits; i++) {
1948 struct commit *graph_commit;
1950 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1952 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1953 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
1954 oid_to_hex(&prev_oid),
1955 oid_to_hex(&cur_oid));
1957 oidcpy(&prev_oid, &cur_oid);
1959 while (cur_oid.hash[0] > cur_fanout_pos) {
1960 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1962 if (i != fanout_value)
1963 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1964 cur_fanout_pos, fanout_value, i);
1965 cur_fanout_pos++;
1968 graph_commit = lookup_commit(r, &cur_oid);
1969 if (!parse_commit_in_graph_one(r, g, graph_commit))
1970 graph_report(_("failed to parse commit %s from commit-graph"),
1971 oid_to_hex(&cur_oid));
1974 while (cur_fanout_pos < 256) {
1975 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1977 if (g->num_commits != fanout_value)
1978 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1979 cur_fanout_pos, fanout_value, i);
1981 cur_fanout_pos++;
1984 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1985 return verify_commit_graph_error;
1987 progress = start_progress(_("Verifying commits in commit graph"),
1988 g->num_commits);
1989 for (i = 0; i < g->num_commits; i++) {
1990 struct commit *graph_commit, *odb_commit;
1991 struct commit_list *graph_parents, *odb_parents;
1992 uint32_t max_generation = 0;
1994 display_progress(progress, i + 1);
1995 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1997 graph_commit = lookup_commit(r, &cur_oid);
1998 odb_commit = (struct commit *)create_object(r, &cur_oid, alloc_commit_node(r));
1999 if (parse_commit_internal(odb_commit, 0, 0)) {
2000 graph_report(_("failed to parse commit %s from object database for commit-graph"),
2001 oid_to_hex(&cur_oid));
2002 continue;
2005 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
2006 get_commit_tree_oid(odb_commit)))
2007 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2008 oid_to_hex(&cur_oid),
2009 oid_to_hex(get_commit_tree_oid(graph_commit)),
2010 oid_to_hex(get_commit_tree_oid(odb_commit)));
2012 graph_parents = graph_commit->parents;
2013 odb_parents = odb_commit->parents;
2015 while (graph_parents) {
2016 if (odb_parents == NULL) {
2017 graph_report(_("commit-graph parent list for commit %s is too long"),
2018 oid_to_hex(&cur_oid));
2019 break;
2022 /* parse parent in case it is in a base graph */
2023 parse_commit_in_graph_one(r, g, graph_parents->item);
2025 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
2026 graph_report(_("commit-graph parent for %s is %s != %s"),
2027 oid_to_hex(&cur_oid),
2028 oid_to_hex(&graph_parents->item->object.oid),
2029 oid_to_hex(&odb_parents->item->object.oid));
2031 if (graph_parents->item->generation > max_generation)
2032 max_generation = graph_parents->item->generation;
2034 graph_parents = graph_parents->next;
2035 odb_parents = odb_parents->next;
2038 if (odb_parents != NULL)
2039 graph_report(_("commit-graph parent list for commit %s terminates early"),
2040 oid_to_hex(&cur_oid));
2042 if (!graph_commit->generation) {
2043 if (generation_zero == GENERATION_NUMBER_EXISTS)
2044 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2045 oid_to_hex(&cur_oid));
2046 generation_zero = GENERATION_ZERO_EXISTS;
2047 } else if (generation_zero == GENERATION_ZERO_EXISTS)
2048 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2049 oid_to_hex(&cur_oid));
2051 if (generation_zero == GENERATION_ZERO_EXISTS)
2052 continue;
2055 * If one of our parents has generation GENERATION_NUMBER_MAX, then
2056 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
2057 * extra logic in the following condition.
2059 if (max_generation == GENERATION_NUMBER_MAX)
2060 max_generation--;
2062 if (graph_commit->generation != max_generation + 1)
2063 graph_report(_("commit-graph generation for commit %s is %u != %u"),
2064 oid_to_hex(&cur_oid),
2065 graph_commit->generation,
2066 max_generation + 1);
2068 if (graph_commit->date != odb_commit->date)
2069 graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
2070 oid_to_hex(&cur_oid),
2071 graph_commit->date,
2072 odb_commit->date);
2074 stop_progress(&progress);
2076 local_error = verify_commit_graph_error;
2078 if (!(flags & COMMIT_GRAPH_VERIFY_SHALLOW) && g->base_graph)
2079 local_error |= verify_commit_graph(r, g->base_graph, flags);
2081 return local_error;
2084 void free_commit_graph(struct commit_graph *g)
2086 if (!g)
2087 return;
2088 if (g->graph_fd >= 0) {
2089 munmap((void *)g->data, g->data_len);
2090 g->data = NULL;
2091 close(g->graph_fd);
2093 free(g->filename);
2094 free(g);