commit-graph: drop COMMIT_GRAPH_WRITE_CHECK_OIDS flag
[git] / commit-graph.c
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"
19 #include "bloom.h"
20 #include "commit-slab.h"
21
22 void git_test_write_commit_graph_or_die(void)
23 {
24         int flags = 0;
25         if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0))
26                 return;
27
28         if (git_env_bool(GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS, 0))
29                 flags = COMMIT_GRAPH_WRITE_BLOOM_FILTERS;
30
31         if (write_commit_graph_reachable(the_repository->objects->odb,
32                                          flags, NULL))
33                 die("failed to write commit-graph under GIT_TEST_COMMIT_GRAPH");
34 }
35
36 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
37 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
38 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
39 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
40 #define GRAPH_CHUNKID_EXTRAEDGES 0x45444745 /* "EDGE" */
41 #define GRAPH_CHUNKID_BLOOMINDEXES 0x42494458 /* "BIDX" */
42 #define GRAPH_CHUNKID_BLOOMDATA 0x42444154 /* "BDAT" */
43 #define GRAPH_CHUNKID_BASE 0x42415345 /* "BASE" */
44 #define MAX_NUM_CHUNKS 7
45
46 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
47
48 #define GRAPH_VERSION_1 0x1
49 #define GRAPH_VERSION GRAPH_VERSION_1
50
51 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
52 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
53 #define GRAPH_PARENT_NONE 0x70000000
54
55 #define GRAPH_LAST_EDGE 0x80000000
56
57 #define GRAPH_HEADER_SIZE 8
58 #define GRAPH_FANOUT_SIZE (4 * 256)
59 #define GRAPH_CHUNKLOOKUP_WIDTH 12
60 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
61                         + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
62
63 /* Remember to update object flag allocation in object.h */
64 #define REACHABLE       (1u<<15)
65
66 /* Keep track of the order in which commits are added to our list. */
67 define_commit_slab(commit_pos, int);
68 static struct commit_pos commit_pos = COMMIT_SLAB_INIT(1, commit_pos);
69
70 static void set_commit_pos(struct repository *r, const struct object_id *oid)
71 {
72         static int32_t max_pos;
73         struct commit *commit = lookup_commit(r, oid);
74
75         if (!commit)
76                 return; /* should never happen, but be lenient */
77
78         *commit_pos_at(&commit_pos, commit) = max_pos++;
79 }
80
81 static int commit_pos_cmp(const void *va, const void *vb)
82 {
83         const struct commit *a = *(const struct commit **)va;
84         const struct commit *b = *(const struct commit **)vb;
85         return commit_pos_at(&commit_pos, a) -
86                commit_pos_at(&commit_pos, b);
87 }
88
89 static int commit_gen_cmp(const void *va, const void *vb)
90 {
91         const struct commit *a = *(const struct commit **)va;
92         const struct commit *b = *(const struct commit **)vb;
93
94         /* lower generation commits first */
95         if (a->generation < b->generation)
96                 return -1;
97         else if (a->generation > b->generation)
98                 return 1;
99
100         /* use date as a heuristic when generations are equal */
101         if (a->date < b->date)
102                 return -1;
103         else if (a->date > b->date)
104                 return 1;
105         return 0;
106 }
107
108 char *get_commit_graph_filename(struct object_directory *obj_dir)
109 {
110         return xstrfmt("%s/info/commit-graph", obj_dir->path);
111 }
112
113 static char *get_split_graph_filename(struct object_directory *odb,
114                                       const char *oid_hex)
115 {
116         return xstrfmt("%s/info/commit-graphs/graph-%s.graph", odb->path,
117                        oid_hex);
118 }
119
120 static char *get_chain_filename(struct object_directory *odb)
121 {
122         return xstrfmt("%s/info/commit-graphs/commit-graph-chain", odb->path);
123 }
124
125 static uint8_t oid_version(void)
126 {
127         return 1;
128 }
129
130 static struct commit_graph *alloc_commit_graph(void)
131 {
132         struct commit_graph *g = xcalloc(1, sizeof(*g));
133
134         return g;
135 }
136
137 extern int read_replace_refs;
138
139 static int commit_graph_compatible(struct repository *r)
140 {
141         if (!r->gitdir)
142                 return 0;
143
144         if (read_replace_refs) {
145                 prepare_replace_object(r);
146                 if (hashmap_get_size(&r->objects->replace_map->map))
147                         return 0;
148         }
149
150         prepare_commit_graft(r);
151         if (r->parsed_objects && r->parsed_objects->grafts_nr)
152                 return 0;
153         if (is_repository_shallow(r))
154                 return 0;
155
156         return 1;
157 }
158
159 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
160 {
161         *fd = git_open(graph_file);
162         if (*fd < 0)
163                 return 0;
164         if (fstat(*fd, st)) {
165                 close(*fd);
166                 return 0;
167         }
168         return 1;
169 }
170
171 struct commit_graph *load_commit_graph_one_fd_st(int fd, struct stat *st,
172                                                  struct object_directory *odb)
173 {
174         void *graph_map;
175         size_t graph_size;
176         struct commit_graph *ret;
177
178         graph_size = xsize_t(st->st_size);
179
180         if (graph_size < GRAPH_MIN_SIZE) {
181                 close(fd);
182                 error(_("commit-graph file is too small"));
183                 return NULL;
184         }
185         graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
186         close(fd);
187         ret = parse_commit_graph(graph_map, graph_size);
188
189         if (ret)
190                 ret->odb = odb;
191         else
192                 munmap(graph_map, graph_size);
193
194         return ret;
195 }
196
197 static int verify_commit_graph_lite(struct commit_graph *g)
198 {
199         /*
200          * Basic validation shared between parse_commit_graph()
201          * which'll be called every time the graph is used, and the
202          * much more expensive verify_commit_graph() used by
203          * "commit-graph verify".
204          *
205          * There should only be very basic checks here to ensure that
206          * we don't e.g. segfault in fill_commit_in_graph(), but
207          * because this is a very hot codepath nothing that e.g. loops
208          * over g->num_commits, or runs a checksum on the commit-graph
209          * itself.
210          */
211         if (!g->chunk_oid_fanout) {
212                 error("commit-graph is missing the OID Fanout chunk");
213                 return 1;
214         }
215         if (!g->chunk_oid_lookup) {
216                 error("commit-graph is missing the OID Lookup chunk");
217                 return 1;
218         }
219         if (!g->chunk_commit_data) {
220                 error("commit-graph is missing the Commit Data chunk");
221                 return 1;
222         }
223
224         return 0;
225 }
226
227 struct commit_graph *parse_commit_graph(void *graph_map, size_t graph_size)
228 {
229         const unsigned char *data, *chunk_lookup;
230         uint32_t i;
231         struct commit_graph *graph;
232         uint64_t last_chunk_offset;
233         uint32_t last_chunk_id;
234         uint32_t graph_signature;
235         unsigned char graph_version, hash_version;
236
237         if (!graph_map)
238                 return NULL;
239
240         if (graph_size < GRAPH_MIN_SIZE)
241                 return NULL;
242
243         data = (const unsigned char *)graph_map;
244
245         graph_signature = get_be32(data);
246         if (graph_signature != GRAPH_SIGNATURE) {
247                 error(_("commit-graph signature %X does not match signature %X"),
248                       graph_signature, GRAPH_SIGNATURE);
249                 return NULL;
250         }
251
252         graph_version = *(unsigned char*)(data + 4);
253         if (graph_version != GRAPH_VERSION) {
254                 error(_("commit-graph version %X does not match version %X"),
255                       graph_version, GRAPH_VERSION);
256                 return NULL;
257         }
258
259         hash_version = *(unsigned char*)(data + 5);
260         if (hash_version != oid_version()) {
261                 error(_("commit-graph hash version %X does not match version %X"),
262                       hash_version, oid_version());
263                 return NULL;
264         }
265
266         graph = alloc_commit_graph();
267
268         graph->hash_len = the_hash_algo->rawsz;
269         graph->num_chunks = *(unsigned char*)(data + 6);
270         graph->data = graph_map;
271         graph->data_len = graph_size;
272
273         last_chunk_id = 0;
274         last_chunk_offset = 8;
275         chunk_lookup = data + 8;
276         for (i = 0; i < graph->num_chunks; i++) {
277                 uint32_t chunk_id;
278                 uint64_t chunk_offset;
279                 int chunk_repeated = 0;
280
281                 if (data + graph_size - chunk_lookup <
282                     GRAPH_CHUNKLOOKUP_WIDTH) {
283                         error(_("commit-graph chunk lookup table entry missing; file may be incomplete"));
284                         free(graph);
285                         return NULL;
286                 }
287
288                 chunk_id = get_be32(chunk_lookup + 0);
289                 chunk_offset = get_be64(chunk_lookup + 4);
290
291                 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
292
293                 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
294                         error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
295                               (uint32_t)chunk_offset);
296                         free(graph);
297                         return NULL;
298                 }
299
300                 switch (chunk_id) {
301                 case GRAPH_CHUNKID_OIDFANOUT:
302                         if (graph->chunk_oid_fanout)
303                                 chunk_repeated = 1;
304                         else
305                                 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
306                         break;
307
308                 case GRAPH_CHUNKID_OIDLOOKUP:
309                         if (graph->chunk_oid_lookup)
310                                 chunk_repeated = 1;
311                         else
312                                 graph->chunk_oid_lookup = data + chunk_offset;
313                         break;
314
315                 case GRAPH_CHUNKID_DATA:
316                         if (graph->chunk_commit_data)
317                                 chunk_repeated = 1;
318                         else
319                                 graph->chunk_commit_data = data + chunk_offset;
320                         break;
321
322                 case GRAPH_CHUNKID_EXTRAEDGES:
323                         if (graph->chunk_extra_edges)
324                                 chunk_repeated = 1;
325                         else
326                                 graph->chunk_extra_edges = data + chunk_offset;
327                         break;
328
329                 case GRAPH_CHUNKID_BASE:
330                         if (graph->chunk_base_graphs)
331                                 chunk_repeated = 1;
332                         else
333                                 graph->chunk_base_graphs = data + chunk_offset;
334                         break;
335
336                 case GRAPH_CHUNKID_BLOOMINDEXES:
337                         if (graph->chunk_bloom_indexes)
338                                 chunk_repeated = 1;
339                         else
340                                 graph->chunk_bloom_indexes = data + chunk_offset;
341                         break;
342
343                 case GRAPH_CHUNKID_BLOOMDATA:
344                         if (graph->chunk_bloom_data)
345                                 chunk_repeated = 1;
346                         else {
347                                 uint32_t hash_version;
348                                 graph->chunk_bloom_data = data + chunk_offset;
349                                 hash_version = get_be32(data + chunk_offset);
350
351                                 if (hash_version != 1)
352                                         break;
353
354                                 graph->bloom_filter_settings = xmalloc(sizeof(struct bloom_filter_settings));
355                                 graph->bloom_filter_settings->hash_version = hash_version;
356                                 graph->bloom_filter_settings->num_hashes = get_be32(data + chunk_offset + 4);
357                                 graph->bloom_filter_settings->bits_per_entry = get_be32(data + chunk_offset + 8);
358                         }
359                         break;
360                 }
361
362                 if (chunk_repeated) {
363                         error(_("commit-graph chunk id %08x appears multiple times"), chunk_id);
364                         free(graph);
365                         return NULL;
366                 }
367
368                 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
369                 {
370                         graph->num_commits = (chunk_offset - last_chunk_offset)
371                                              / graph->hash_len;
372                 }
373
374                 last_chunk_id = chunk_id;
375                 last_chunk_offset = chunk_offset;
376         }
377
378         if (graph->chunk_bloom_indexes && graph->chunk_bloom_data) {
379                 init_bloom_filters();
380         } else {
381                 /* We need both the bloom chunks to exist together. Else ignore the data */
382                 graph->chunk_bloom_indexes = NULL;
383                 graph->chunk_bloom_data = NULL;
384                 graph->bloom_filter_settings = NULL;
385         }
386
387         hashcpy(graph->oid.hash, graph->data + graph->data_len - graph->hash_len);
388
389         if (verify_commit_graph_lite(graph)) {
390                 free(graph);
391                 return NULL;
392         }
393
394         return graph;
395 }
396
397 static struct commit_graph *load_commit_graph_one(const char *graph_file,
398                                                   struct object_directory *odb)
399 {
400
401         struct stat st;
402         int fd;
403         struct commit_graph *g;
404         int open_ok = open_commit_graph(graph_file, &fd, &st);
405
406         if (!open_ok)
407                 return NULL;
408
409         g = load_commit_graph_one_fd_st(fd, &st, odb);
410
411         if (g)
412                 g->filename = xstrdup(graph_file);
413
414         return g;
415 }
416
417 static struct commit_graph *load_commit_graph_v1(struct repository *r,
418                                                  struct object_directory *odb)
419 {
420         char *graph_name = get_commit_graph_filename(odb);
421         struct commit_graph *g = load_commit_graph_one(graph_name, odb);
422         free(graph_name);
423
424         return g;
425 }
426
427 static int add_graph_to_chain(struct commit_graph *g,
428                               struct commit_graph *chain,
429                               struct object_id *oids,
430                               int n)
431 {
432         struct commit_graph *cur_g = chain;
433
434         if (n && !g->chunk_base_graphs) {
435                 warning(_("commit-graph has no base graphs chunk"));
436                 return 0;
437         }
438
439         while (n) {
440                 n--;
441
442                 if (!cur_g ||
443                     !oideq(&oids[n], &cur_g->oid) ||
444                     !hasheq(oids[n].hash, g->chunk_base_graphs + g->hash_len * n)) {
445                         warning(_("commit-graph chain does not match"));
446                         return 0;
447                 }
448
449                 cur_g = cur_g->base_graph;
450         }
451
452         g->base_graph = chain;
453
454         if (chain)
455                 g->num_commits_in_base = chain->num_commits + chain->num_commits_in_base;
456
457         return 1;
458 }
459
460 static struct commit_graph *load_commit_graph_chain(struct repository *r,
461                                                     struct object_directory *odb)
462 {
463         struct commit_graph *graph_chain = NULL;
464         struct strbuf line = STRBUF_INIT;
465         struct stat st;
466         struct object_id *oids;
467         int i = 0, valid = 1, count;
468         char *chain_name = get_chain_filename(odb);
469         FILE *fp;
470         int stat_res;
471
472         fp = fopen(chain_name, "r");
473         stat_res = stat(chain_name, &st);
474         free(chain_name);
475
476         if (!fp ||
477             stat_res ||
478             st.st_size <= the_hash_algo->hexsz)
479                 return NULL;
480
481         count = st.st_size / (the_hash_algo->hexsz + 1);
482         oids = xcalloc(count, sizeof(struct object_id));
483
484         prepare_alt_odb(r);
485
486         for (i = 0; i < count; i++) {
487                 struct object_directory *odb;
488
489                 if (strbuf_getline_lf(&line, fp) == EOF)
490                         break;
491
492                 if (get_oid_hex(line.buf, &oids[i])) {
493                         warning(_("invalid commit-graph chain: line '%s' not a hash"),
494                                 line.buf);
495                         valid = 0;
496                         break;
497                 }
498
499                 valid = 0;
500                 for (odb = r->objects->odb; odb; odb = odb->next) {
501                         char *graph_name = get_split_graph_filename(odb, line.buf);
502                         struct commit_graph *g = load_commit_graph_one(graph_name, odb);
503
504                         free(graph_name);
505
506                         if (g) {
507                                 if (add_graph_to_chain(g, graph_chain, oids, i)) {
508                                         graph_chain = g;
509                                         valid = 1;
510                                 }
511
512                                 break;
513                         }
514                 }
515
516                 if (!valid) {
517                         warning(_("unable to find all commit-graph files"));
518                         break;
519                 }
520         }
521
522         free(oids);
523         fclose(fp);
524         strbuf_release(&line);
525
526         return graph_chain;
527 }
528
529 struct commit_graph *read_commit_graph_one(struct repository *r,
530                                            struct object_directory *odb)
531 {
532         struct commit_graph *g = load_commit_graph_v1(r, odb);
533
534         if (!g)
535                 g = load_commit_graph_chain(r, odb);
536
537         return g;
538 }
539
540 static void prepare_commit_graph_one(struct repository *r,
541                                      struct object_directory *odb)
542 {
543
544         if (r->objects->commit_graph)
545                 return;
546
547         r->objects->commit_graph = read_commit_graph_one(r, odb);
548 }
549
550 /*
551  * Return 1 if commit_graph is non-NULL, and 0 otherwise.
552  *
553  * On the first invocation, this function attempts to load the commit
554  * graph if the_repository is configured to have one.
555  */
556 static int prepare_commit_graph(struct repository *r)
557 {
558         struct object_directory *odb;
559
560         /*
561          * This must come before the "already attempted?" check below, because
562          * we want to disable even an already-loaded graph file.
563          */
564         if (r->commit_graph_disabled)
565                 return 0;
566
567         if (r->objects->commit_graph_attempted)
568                 return !!r->objects->commit_graph;
569         r->objects->commit_graph_attempted = 1;
570
571         if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
572                 die("dying as requested by the '%s' variable on commit-graph load!",
573                     GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD);
574
575         prepare_repo_settings(r);
576
577         if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
578             r->settings.core_commit_graph != 1)
579                 /*
580                  * This repository is not configured to use commit graphs, so
581                  * do not load one. (But report commit_graph_attempted anyway
582                  * so that commit graph loading is not attempted again for this
583                  * repository.)
584                  */
585                 return 0;
586
587         if (!commit_graph_compatible(r))
588                 return 0;
589
590         prepare_alt_odb(r);
591         for (odb = r->objects->odb;
592              !r->objects->commit_graph && odb;
593              odb = odb->next)
594                 prepare_commit_graph_one(r, odb);
595         return !!r->objects->commit_graph;
596 }
597
598 int generation_numbers_enabled(struct repository *r)
599 {
600         uint32_t first_generation;
601         struct commit_graph *g;
602         if (!prepare_commit_graph(r))
603                return 0;
604
605         g = r->objects->commit_graph;
606
607         if (!g->num_commits)
608                 return 0;
609
610         first_generation = get_be32(g->chunk_commit_data +
611                                     g->hash_len + 8) >> 2;
612
613         return !!first_generation;
614 }
615
616 static void close_commit_graph_one(struct commit_graph *g)
617 {
618         if (!g)
619                 return;
620
621         close_commit_graph_one(g->base_graph);
622         free_commit_graph(g);
623 }
624
625 void close_commit_graph(struct raw_object_store *o)
626 {
627         close_commit_graph_one(o->commit_graph);
628         o->commit_graph = NULL;
629 }
630
631 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
632 {
633         return bsearch_hash(oid->hash, g->chunk_oid_fanout,
634                             g->chunk_oid_lookup, g->hash_len, pos);
635 }
636
637 static void load_oid_from_graph(struct commit_graph *g,
638                                 uint32_t pos,
639                                 struct object_id *oid)
640 {
641         uint32_t lex_index;
642
643         while (g && pos < g->num_commits_in_base)
644                 g = g->base_graph;
645
646         if (!g)
647                 BUG("NULL commit-graph");
648
649         if (pos >= g->num_commits + g->num_commits_in_base)
650                 die(_("invalid commit position. commit-graph is likely corrupt"));
651
652         lex_index = pos - g->num_commits_in_base;
653
654         hashcpy(oid->hash, g->chunk_oid_lookup + g->hash_len * lex_index);
655 }
656
657 static struct commit_list **insert_parent_or_die(struct repository *r,
658                                                  struct commit_graph *g,
659                                                  uint32_t pos,
660                                                  struct commit_list **pptr)
661 {
662         struct commit *c;
663         struct object_id oid;
664
665         if (pos >= g->num_commits + g->num_commits_in_base)
666                 die("invalid parent position %"PRIu32, pos);
667
668         load_oid_from_graph(g, pos, &oid);
669         c = lookup_commit(r, &oid);
670         if (!c)
671                 die(_("could not find commit %s"), oid_to_hex(&oid));
672         c->graph_pos = pos;
673         return &commit_list_insert(c, pptr)->next;
674 }
675
676 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
677 {
678         const unsigned char *commit_data;
679         uint32_t lex_index;
680
681         while (pos < g->num_commits_in_base)
682                 g = g->base_graph;
683
684         lex_index = pos - g->num_commits_in_base;
685         commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * lex_index;
686         item->graph_pos = pos;
687         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
688 }
689
690 static inline void set_commit_tree(struct commit *c, struct tree *t)
691 {
692         c->maybe_tree = t;
693 }
694
695 static int fill_commit_in_graph(struct repository *r,
696                                 struct commit *item,
697                                 struct commit_graph *g, uint32_t pos)
698 {
699         uint32_t edge_value;
700         uint32_t *parent_data_ptr;
701         uint64_t date_low, date_high;
702         struct commit_list **pptr;
703         const unsigned char *commit_data;
704         uint32_t lex_index;
705
706         while (pos < g->num_commits_in_base)
707                 g = g->base_graph;
708
709         if (pos >= g->num_commits + g->num_commits_in_base)
710                 die(_("invalid commit position. commit-graph is likely corrupt"));
711
712         /*
713          * Store the "full" position, but then use the
714          * "local" position for the rest of the calculation.
715          */
716         item->graph_pos = pos;
717         lex_index = pos - g->num_commits_in_base;
718
719         commit_data = g->chunk_commit_data + (g->hash_len + 16) * lex_index;
720
721         item->object.parsed = 1;
722
723         set_commit_tree(item, NULL);
724
725         date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
726         date_low = get_be32(commit_data + g->hash_len + 12);
727         item->date = (timestamp_t)((date_high << 32) | date_low);
728
729         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
730
731         pptr = &item->parents;
732
733         edge_value = get_be32(commit_data + g->hash_len);
734         if (edge_value == GRAPH_PARENT_NONE)
735                 return 1;
736         pptr = insert_parent_or_die(r, g, edge_value, pptr);
737
738         edge_value = get_be32(commit_data + g->hash_len + 4);
739         if (edge_value == GRAPH_PARENT_NONE)
740                 return 1;
741         if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
742                 pptr = insert_parent_or_die(r, g, edge_value, pptr);
743                 return 1;
744         }
745
746         parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
747                           4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
748         do {
749                 edge_value = get_be32(parent_data_ptr);
750                 pptr = insert_parent_or_die(r, g,
751                                             edge_value & GRAPH_EDGE_LAST_MASK,
752                                             pptr);
753                 parent_data_ptr++;
754         } while (!(edge_value & GRAPH_LAST_EDGE));
755
756         return 1;
757 }
758
759 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
760 {
761         if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
762                 *pos = item->graph_pos;
763                 return 1;
764         } else {
765                 struct commit_graph *cur_g = g;
766                 uint32_t lex_index;
767
768                 while (cur_g && !bsearch_graph(cur_g, &(item->object.oid), &lex_index))
769                         cur_g = cur_g->base_graph;
770
771                 if (cur_g) {
772                         *pos = lex_index + cur_g->num_commits_in_base;
773                         return 1;
774                 }
775
776                 return 0;
777         }
778 }
779
780 static int parse_commit_in_graph_one(struct repository *r,
781                                      struct commit_graph *g,
782                                      struct commit *item)
783 {
784         uint32_t pos;
785
786         if (item->object.parsed)
787                 return 1;
788
789         if (find_commit_in_graph(item, g, &pos))
790                 return fill_commit_in_graph(r, item, g, pos);
791
792         return 0;
793 }
794
795 int parse_commit_in_graph(struct repository *r, struct commit *item)
796 {
797         if (!prepare_commit_graph(r))
798                 return 0;
799         return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
800 }
801
802 void load_commit_graph_info(struct repository *r, struct commit *item)
803 {
804         uint32_t pos;
805         if (!prepare_commit_graph(r))
806                 return;
807         if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
808                 fill_commit_graph_info(item, r->objects->commit_graph, pos);
809 }
810
811 static struct tree *load_tree_for_commit(struct repository *r,
812                                          struct commit_graph *g,
813                                          struct commit *c)
814 {
815         struct object_id oid;
816         const unsigned char *commit_data;
817
818         while (c->graph_pos < g->num_commits_in_base)
819                 g = g->base_graph;
820
821         commit_data = g->chunk_commit_data +
822                         GRAPH_DATA_WIDTH * (c->graph_pos - g->num_commits_in_base);
823
824         hashcpy(oid.hash, commit_data);
825         set_commit_tree(c, lookup_tree(r, &oid));
826
827         return c->maybe_tree;
828 }
829
830 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
831                                                  struct commit_graph *g,
832                                                  const struct commit *c)
833 {
834         if (c->maybe_tree)
835                 return c->maybe_tree;
836         if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
837                 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
838
839         return load_tree_for_commit(r, g, (struct commit *)c);
840 }
841
842 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
843 {
844         return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
845 }
846
847 struct packed_commit_list {
848         struct commit **list;
849         int nr;
850         int alloc;
851 };
852
853 struct packed_oid_list {
854         struct object_id *list;
855         int nr;
856         int alloc;
857 };
858
859 struct write_commit_graph_context {
860         struct repository *r;
861         struct object_directory *odb;
862         char *graph_name;
863         struct packed_oid_list oids;
864         struct packed_commit_list commits;
865         int num_extra_edges;
866         unsigned long approx_nr_objects;
867         struct progress *progress;
868         int progress_done;
869         uint64_t progress_cnt;
870
871         char *base_graph_name;
872         int num_commit_graphs_before;
873         int num_commit_graphs_after;
874         char **commit_graph_filenames_before;
875         char **commit_graph_filenames_after;
876         char **commit_graph_hash_after;
877         uint32_t new_num_commits_in_base;
878         struct commit_graph *new_base_graph;
879
880         unsigned append:1,
881                  report_progress:1,
882                  split:1,
883                  changed_paths:1,
884                  order_by_pack:1;
885
886         const struct split_commit_graph_opts *split_opts;
887         size_t total_bloom_filter_data_size;
888 };
889
890 static void write_graph_chunk_fanout(struct hashfile *f,
891                                      struct write_commit_graph_context *ctx)
892 {
893         int i, count = 0;
894         struct commit **list = ctx->commits.list;
895
896         /*
897          * Write the first-level table (the list is sorted,
898          * but we use a 256-entry lookup to be able to avoid
899          * having to do eight extra binary search iterations).
900          */
901         for (i = 0; i < 256; i++) {
902                 while (count < ctx->commits.nr) {
903                         if ((*list)->object.oid.hash[0] != i)
904                                 break;
905                         display_progress(ctx->progress, ++ctx->progress_cnt);
906                         count++;
907                         list++;
908                 }
909
910                 hashwrite_be32(f, count);
911         }
912 }
913
914 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
915                                    struct write_commit_graph_context *ctx)
916 {
917         struct commit **list = ctx->commits.list;
918         int count;
919         for (count = 0; count < ctx->commits.nr; count++, list++) {
920                 display_progress(ctx->progress, ++ctx->progress_cnt);
921                 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
922         }
923 }
924
925 static const unsigned char *commit_to_sha1(size_t index, void *table)
926 {
927         struct commit **commits = table;
928         return commits[index]->object.oid.hash;
929 }
930
931 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
932                                    struct write_commit_graph_context *ctx)
933 {
934         struct commit **list = ctx->commits.list;
935         struct commit **last = ctx->commits.list + ctx->commits.nr;
936         uint32_t num_extra_edges = 0;
937
938         while (list < last) {
939                 struct commit_list *parent;
940                 struct object_id *tree;
941                 int edge_value;
942                 uint32_t packedDate[2];
943                 display_progress(ctx->progress, ++ctx->progress_cnt);
944
945                 if (parse_commit_no_graph(*list))
946                         die(_("unable to parse commit %s"),
947                                 oid_to_hex(&(*list)->object.oid));
948                 tree = get_commit_tree_oid(*list);
949                 hashwrite(f, tree->hash, hash_len);
950
951                 parent = (*list)->parents;
952
953                 if (!parent)
954                         edge_value = GRAPH_PARENT_NONE;
955                 else {
956                         edge_value = sha1_pos(parent->item->object.oid.hash,
957                                               ctx->commits.list,
958                                               ctx->commits.nr,
959                                               commit_to_sha1);
960
961                         if (edge_value >= 0)
962                                 edge_value += ctx->new_num_commits_in_base;
963                         else if (ctx->new_base_graph) {
964                                 uint32_t pos;
965                                 if (find_commit_in_graph(parent->item,
966                                                          ctx->new_base_graph,
967                                                          &pos))
968                                         edge_value = pos;
969                         }
970
971                         if (edge_value < 0)
972                                 BUG("missing parent %s for commit %s",
973                                     oid_to_hex(&parent->item->object.oid),
974                                     oid_to_hex(&(*list)->object.oid));
975                 }
976
977                 hashwrite_be32(f, edge_value);
978
979                 if (parent)
980                         parent = parent->next;
981
982                 if (!parent)
983                         edge_value = GRAPH_PARENT_NONE;
984                 else if (parent->next)
985                         edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
986                 else {
987                         edge_value = sha1_pos(parent->item->object.oid.hash,
988                                               ctx->commits.list,
989                                               ctx->commits.nr,
990                                               commit_to_sha1);
991
992                         if (edge_value >= 0)
993                                 edge_value += ctx->new_num_commits_in_base;
994                         else if (ctx->new_base_graph) {
995                                 uint32_t pos;
996                                 if (find_commit_in_graph(parent->item,
997                                                          ctx->new_base_graph,
998                                                          &pos))
999                                         edge_value = pos;
1000                         }
1001
1002                         if (edge_value < 0)
1003                                 BUG("missing parent %s for commit %s",
1004                                     oid_to_hex(&parent->item->object.oid),
1005                                     oid_to_hex(&(*list)->object.oid));
1006                 }
1007
1008                 hashwrite_be32(f, edge_value);
1009
1010                 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
1011                         do {
1012                                 num_extra_edges++;
1013                                 parent = parent->next;
1014                         } while (parent);
1015                 }
1016
1017                 if (sizeof((*list)->date) > 4)
1018                         packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
1019                 else
1020                         packedDate[0] = 0;
1021
1022                 packedDate[0] |= htonl((*list)->generation << 2);
1023
1024                 packedDate[1] = htonl((*list)->date);
1025                 hashwrite(f, packedDate, 8);
1026
1027                 list++;
1028         }
1029 }
1030
1031 static void write_graph_chunk_extra_edges(struct hashfile *f,
1032                                           struct write_commit_graph_context *ctx)
1033 {
1034         struct commit **list = ctx->commits.list;
1035         struct commit **last = ctx->commits.list + ctx->commits.nr;
1036         struct commit_list *parent;
1037
1038         while (list < last) {
1039                 int num_parents = 0;
1040
1041                 display_progress(ctx->progress, ++ctx->progress_cnt);
1042
1043                 for (parent = (*list)->parents; num_parents < 3 && parent;
1044                      parent = parent->next)
1045                         num_parents++;
1046
1047                 if (num_parents <= 2) {
1048                         list++;
1049                         continue;
1050                 }
1051
1052                 /* Since num_parents > 2, this initializer is safe. */
1053                 for (parent = (*list)->parents->next; parent; parent = parent->next) {
1054                         int edge_value = sha1_pos(parent->item->object.oid.hash,
1055                                                   ctx->commits.list,
1056                                                   ctx->commits.nr,
1057                                                   commit_to_sha1);
1058
1059                         if (edge_value >= 0)
1060                                 edge_value += ctx->new_num_commits_in_base;
1061                         else if (ctx->new_base_graph) {
1062                                 uint32_t pos;
1063                                 if (find_commit_in_graph(parent->item,
1064                                                          ctx->new_base_graph,
1065                                                          &pos))
1066                                         edge_value = pos;
1067                         }
1068
1069                         if (edge_value < 0)
1070                                 BUG("missing parent %s for commit %s",
1071                                     oid_to_hex(&parent->item->object.oid),
1072                                     oid_to_hex(&(*list)->object.oid));
1073                         else if (!parent->next)
1074                                 edge_value |= GRAPH_LAST_EDGE;
1075
1076                         hashwrite_be32(f, edge_value);
1077                 }
1078
1079                 list++;
1080         }
1081 }
1082
1083 static void write_graph_chunk_bloom_indexes(struct hashfile *f,
1084                                             struct write_commit_graph_context *ctx)
1085 {
1086         struct commit **list = ctx->commits.list;
1087         struct commit **last = ctx->commits.list + ctx->commits.nr;
1088         uint32_t cur_pos = 0;
1089         struct progress *progress = NULL;
1090         int i = 0;
1091
1092         if (ctx->report_progress)
1093                 progress = start_delayed_progress(
1094                         _("Writing changed paths Bloom filters index"),
1095                         ctx->commits.nr);
1096
1097         while (list < last) {
1098                 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list, 0);
1099                 cur_pos += filter->len;
1100                 display_progress(progress, ++i);
1101                 hashwrite_be32(f, cur_pos);
1102                 list++;
1103         }
1104
1105         stop_progress(&progress);
1106 }
1107
1108 static void write_graph_chunk_bloom_data(struct hashfile *f,
1109                                          struct write_commit_graph_context *ctx,
1110                                          const struct bloom_filter_settings *settings)
1111 {
1112         struct commit **list = ctx->commits.list;
1113         struct commit **last = ctx->commits.list + ctx->commits.nr;
1114         struct progress *progress = NULL;
1115         int i = 0;
1116
1117         if (ctx->report_progress)
1118                 progress = start_delayed_progress(
1119                         _("Writing changed paths Bloom filters data"),
1120                         ctx->commits.nr);
1121
1122         hashwrite_be32(f, settings->hash_version);
1123         hashwrite_be32(f, settings->num_hashes);
1124         hashwrite_be32(f, settings->bits_per_entry);
1125
1126         while (list < last) {
1127                 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list, 0);
1128                 display_progress(progress, ++i);
1129                 hashwrite(f, filter->data, filter->len * sizeof(unsigned char));
1130                 list++;
1131         }
1132
1133         stop_progress(&progress);
1134 }
1135
1136 static int oid_compare(const void *_a, const void *_b)
1137 {
1138         const struct object_id *a = (const struct object_id *)_a;
1139         const struct object_id *b = (const struct object_id *)_b;
1140         return oidcmp(a, b);
1141 }
1142
1143 static int add_packed_commits(const struct object_id *oid,
1144                               struct packed_git *pack,
1145                               uint32_t pos,
1146                               void *data)
1147 {
1148         struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
1149         enum object_type type;
1150         off_t offset = nth_packed_object_offset(pack, pos);
1151         struct object_info oi = OBJECT_INFO_INIT;
1152
1153         if (ctx->progress)
1154                 display_progress(ctx->progress, ++ctx->progress_done);
1155
1156         oi.typep = &type;
1157         if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
1158                 die(_("unable to get type of object %s"), oid_to_hex(oid));
1159
1160         if (type != OBJ_COMMIT)
1161                 return 0;
1162
1163         ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1164         oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
1165         ctx->oids.nr++;
1166
1167         set_commit_pos(ctx->r, oid);
1168
1169         return 0;
1170 }
1171
1172 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
1173 {
1174         struct commit_list *parent;
1175         for (parent = commit->parents; parent; parent = parent->next) {
1176                 if (!(parent->item->object.flags & REACHABLE)) {
1177                         ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1178                         oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
1179                         ctx->oids.nr++;
1180                         parent->item->object.flags |= REACHABLE;
1181                 }
1182         }
1183 }
1184
1185 static void close_reachable(struct write_commit_graph_context *ctx)
1186 {
1187         int i;
1188         struct commit *commit;
1189         enum commit_graph_split_flags flags = ctx->split_opts ?
1190                 ctx->split_opts->flags : COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1191
1192         if (ctx->report_progress)
1193                 ctx->progress = start_delayed_progress(
1194                                         _("Loading known commits in commit graph"),
1195                                         ctx->oids.nr);
1196         for (i = 0; i < ctx->oids.nr; i++) {
1197                 display_progress(ctx->progress, i + 1);
1198                 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1199                 if (commit)
1200                         commit->object.flags |= REACHABLE;
1201         }
1202         stop_progress(&ctx->progress);
1203
1204         /*
1205          * As this loop runs, ctx->oids.nr may grow, but not more
1206          * than the number of missing commits in the reachable
1207          * closure.
1208          */
1209         if (ctx->report_progress)
1210                 ctx->progress = start_delayed_progress(
1211                                         _("Expanding reachable commits in commit graph"),
1212                                         0);
1213         for (i = 0; i < ctx->oids.nr; i++) {
1214                 display_progress(ctx->progress, i + 1);
1215                 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1216
1217                 if (!commit)
1218                         continue;
1219                 if (ctx->split) {
1220                         if ((!parse_commit(commit) &&
1221                              commit->graph_pos == COMMIT_NOT_FROM_GRAPH) ||
1222                             flags == COMMIT_GRAPH_SPLIT_REPLACE)
1223                                 add_missing_parents(ctx, commit);
1224                 } else if (!parse_commit_no_graph(commit))
1225                         add_missing_parents(ctx, commit);
1226         }
1227         stop_progress(&ctx->progress);
1228
1229         if (ctx->report_progress)
1230                 ctx->progress = start_delayed_progress(
1231                                         _("Clearing commit marks in commit graph"),
1232                                         ctx->oids.nr);
1233         for (i = 0; i < ctx->oids.nr; i++) {
1234                 display_progress(ctx->progress, i + 1);
1235                 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1236
1237                 if (commit)
1238                         commit->object.flags &= ~REACHABLE;
1239         }
1240         stop_progress(&ctx->progress);
1241 }
1242
1243 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1244 {
1245         int i;
1246         struct commit_list *list = NULL;
1247
1248         if (ctx->report_progress)
1249                 ctx->progress = start_delayed_progress(
1250                                         _("Computing commit graph generation numbers"),
1251                                         ctx->commits.nr);
1252         for (i = 0; i < ctx->commits.nr; i++) {
1253                 display_progress(ctx->progress, i + 1);
1254                 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
1255                     ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
1256                         continue;
1257
1258                 commit_list_insert(ctx->commits.list[i], &list);
1259                 while (list) {
1260                         struct commit *current = list->item;
1261                         struct commit_list *parent;
1262                         int all_parents_computed = 1;
1263                         uint32_t max_generation = 0;
1264
1265                         for (parent = current->parents; parent; parent = parent->next) {
1266                                 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
1267                                     parent->item->generation == GENERATION_NUMBER_ZERO) {
1268                                         all_parents_computed = 0;
1269                                         commit_list_insert(parent->item, &list);
1270                                         break;
1271                                 } else if (parent->item->generation > max_generation) {
1272                                         max_generation = parent->item->generation;
1273                                 }
1274                         }
1275
1276                         if (all_parents_computed) {
1277                                 current->generation = max_generation + 1;
1278                                 pop_commit(&list);
1279
1280                                 if (current->generation > GENERATION_NUMBER_MAX)
1281                                         current->generation = GENERATION_NUMBER_MAX;
1282                         }
1283                 }
1284         }
1285         stop_progress(&ctx->progress);
1286 }
1287
1288 static void compute_bloom_filters(struct write_commit_graph_context *ctx)
1289 {
1290         int i;
1291         struct progress *progress = NULL;
1292         struct commit **sorted_commits;
1293
1294         init_bloom_filters();
1295
1296         if (ctx->report_progress)
1297                 progress = start_delayed_progress(
1298                         _("Computing commit changed paths Bloom filters"),
1299                         ctx->commits.nr);
1300
1301         ALLOC_ARRAY(sorted_commits, ctx->commits.nr);
1302         COPY_ARRAY(sorted_commits, ctx->commits.list, ctx->commits.nr);
1303
1304         if (ctx->order_by_pack)
1305                 QSORT(sorted_commits, ctx->commits.nr, commit_pos_cmp);
1306         else
1307                 QSORT(sorted_commits, ctx->commits.nr, commit_gen_cmp);
1308
1309         for (i = 0; i < ctx->commits.nr; i++) {
1310                 struct commit *c = sorted_commits[i];
1311                 struct bloom_filter *filter = get_bloom_filter(ctx->r, c, 1);
1312                 ctx->total_bloom_filter_data_size += sizeof(unsigned char) * filter->len;
1313                 display_progress(progress, i + 1);
1314         }
1315
1316         free(sorted_commits);
1317         stop_progress(&progress);
1318 }
1319
1320 struct refs_cb_data {
1321         struct oidset *commits;
1322         struct progress *progress;
1323 };
1324
1325 static int add_ref_to_set(const char *refname,
1326                           const struct object_id *oid,
1327                           int flags, void *cb_data)
1328 {
1329         struct object_id peeled;
1330         struct refs_cb_data *data = (struct refs_cb_data *)cb_data;
1331
1332         if (!peel_ref(refname, &peeled))
1333                 oid = &peeled;
1334         if (oid_object_info(the_repository, oid, NULL) == OBJ_COMMIT)
1335                 oidset_insert(data->commits, oid);
1336
1337         display_progress(data->progress, oidset_size(data->commits));
1338
1339         return 0;
1340 }
1341
1342 int write_commit_graph_reachable(struct object_directory *odb,
1343                                  enum commit_graph_write_flags flags,
1344                                  const struct split_commit_graph_opts *split_opts)
1345 {
1346         struct oidset commits = OIDSET_INIT;
1347         struct refs_cb_data data;
1348         int result;
1349
1350         memset(&data, 0, sizeof(data));
1351         data.commits = &commits;
1352         if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
1353                 data.progress = start_delayed_progress(
1354                         _("Collecting referenced commits"), 0);
1355
1356         for_each_ref(add_ref_to_set, &data);
1357         result = write_commit_graph(odb, NULL, &commits,
1358                                     flags, split_opts);
1359
1360         oidset_clear(&commits);
1361         if (data.progress)
1362                 stop_progress(&data.progress);
1363         return result;
1364 }
1365
1366 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
1367                                 struct string_list *pack_indexes)
1368 {
1369         uint32_t i;
1370         struct strbuf progress_title = STRBUF_INIT;
1371         struct strbuf packname = STRBUF_INIT;
1372         int dirlen;
1373
1374         strbuf_addf(&packname, "%s/pack/", ctx->odb->path);
1375         dirlen = packname.len;
1376         if (ctx->report_progress) {
1377                 strbuf_addf(&progress_title,
1378                             Q_("Finding commits for commit graph in %d pack",
1379                                "Finding commits for commit graph in %d packs",
1380                                pack_indexes->nr),
1381                             pack_indexes->nr);
1382                 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1383                 ctx->progress_done = 0;
1384         }
1385         for (i = 0; i < pack_indexes->nr; i++) {
1386                 struct packed_git *p;
1387                 strbuf_setlen(&packname, dirlen);
1388                 strbuf_addstr(&packname, pack_indexes->items[i].string);
1389                 p = add_packed_git(packname.buf, packname.len, 1);
1390                 if (!p) {
1391                         error(_("error adding pack %s"), packname.buf);
1392                         return -1;
1393                 }
1394                 if (open_pack_index(p)) {
1395                         error(_("error opening index for %s"), packname.buf);
1396                         return -1;
1397                 }
1398                 for_each_object_in_pack(p, add_packed_commits, ctx,
1399                                         FOR_EACH_OBJECT_PACK_ORDER);
1400                 close_pack(p);
1401                 free(p);
1402         }
1403
1404         stop_progress(&ctx->progress);
1405         strbuf_release(&progress_title);
1406         strbuf_release(&packname);
1407
1408         return 0;
1409 }
1410
1411 static int fill_oids_from_commits(struct write_commit_graph_context *ctx,
1412                                   struct oidset *commits)
1413 {
1414         struct oidset_iter iter;
1415         struct object_id *oid;
1416
1417         if (!oidset_size(commits))
1418                 return 0;
1419
1420         oidset_iter_init(commits, &iter);
1421         while ((oid = oidset_iter_next(&iter))) {
1422                 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1423                 oidcpy(&ctx->oids.list[ctx->oids.nr], oid);
1424                 ctx->oids.nr++;
1425         }
1426
1427         return 0;
1428 }
1429
1430 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1431 {
1432         if (ctx->report_progress)
1433                 ctx->progress = start_delayed_progress(
1434                         _("Finding commits for commit graph among packed objects"),
1435                         ctx->approx_nr_objects);
1436         for_each_packed_object(add_packed_commits, ctx,
1437                                FOR_EACH_OBJECT_PACK_ORDER);
1438         if (ctx->progress_done < ctx->approx_nr_objects)
1439                 display_progress(ctx->progress, ctx->approx_nr_objects);
1440         stop_progress(&ctx->progress);
1441 }
1442
1443 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
1444 {
1445         uint32_t i, count_distinct = 1;
1446
1447         if (ctx->report_progress)
1448                 ctx->progress = start_delayed_progress(
1449                         _("Counting distinct commits in commit graph"),
1450                         ctx->oids.nr);
1451         display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1452         QSORT(ctx->oids.list, ctx->oids.nr, oid_compare);
1453
1454         for (i = 1; i < ctx->oids.nr; i++) {
1455                 display_progress(ctx->progress, i + 1);
1456                 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i])) {
1457                         if (ctx->split) {
1458                                 struct commit *c = lookup_commit(ctx->r, &ctx->oids.list[i]);
1459
1460                                 if (!c || c->graph_pos != COMMIT_NOT_FROM_GRAPH)
1461                                         continue;
1462                         }
1463
1464                         count_distinct++;
1465                 }
1466         }
1467         stop_progress(&ctx->progress);
1468
1469         return count_distinct;
1470 }
1471
1472 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1473 {
1474         uint32_t i;
1475         enum commit_graph_split_flags flags = ctx->split_opts ?
1476                 ctx->split_opts->flags : COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1477
1478         ctx->num_extra_edges = 0;
1479         if (ctx->report_progress)
1480                 ctx->progress = start_delayed_progress(
1481                         _("Finding extra edges in commit graph"),
1482                         ctx->oids.nr);
1483         for (i = 0; i < ctx->oids.nr; i++) {
1484                 unsigned int num_parents;
1485
1486                 display_progress(ctx->progress, i + 1);
1487                 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1488                         continue;
1489
1490                 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + 1, ctx->commits.alloc);
1491                 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1492
1493                 if (ctx->split && flags != COMMIT_GRAPH_SPLIT_REPLACE &&
1494                     ctx->commits.list[ctx->commits.nr]->graph_pos != COMMIT_NOT_FROM_GRAPH)
1495                         continue;
1496
1497                 if (ctx->split && flags == COMMIT_GRAPH_SPLIT_REPLACE)
1498                         parse_commit(ctx->commits.list[ctx->commits.nr]);
1499                 else
1500                         parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1501
1502                 num_parents = commit_list_count(ctx->commits.list[ctx->commits.nr]->parents);
1503                 if (num_parents > 2)
1504                         ctx->num_extra_edges += num_parents - 1;
1505
1506                 ctx->commits.nr++;
1507         }
1508         stop_progress(&ctx->progress);
1509 }
1510
1511 static int write_graph_chunk_base_1(struct hashfile *f,
1512                                     struct commit_graph *g)
1513 {
1514         int num = 0;
1515
1516         if (!g)
1517                 return 0;
1518
1519         num = write_graph_chunk_base_1(f, g->base_graph);
1520         hashwrite(f, g->oid.hash, the_hash_algo->rawsz);
1521         return num + 1;
1522 }
1523
1524 static int write_graph_chunk_base(struct hashfile *f,
1525                                   struct write_commit_graph_context *ctx)
1526 {
1527         int num = write_graph_chunk_base_1(f, ctx->new_base_graph);
1528
1529         if (num != ctx->num_commit_graphs_after - 1) {
1530                 error(_("failed to write correct number of base graph ids"));
1531                 return -1;
1532         }
1533
1534         return 0;
1535 }
1536
1537 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1538 {
1539         uint32_t i;
1540         int fd;
1541         struct hashfile *f;
1542         struct lock_file lk = LOCK_INIT;
1543         uint32_t chunk_ids[MAX_NUM_CHUNKS + 1];
1544         uint64_t chunk_offsets[MAX_NUM_CHUNKS + 1];
1545         const unsigned hashsz = the_hash_algo->rawsz;
1546         struct strbuf progress_title = STRBUF_INIT;
1547         int num_chunks = 3;
1548         struct object_id file_hash;
1549         const struct bloom_filter_settings bloom_settings = DEFAULT_BLOOM_FILTER_SETTINGS;
1550
1551         if (ctx->split) {
1552                 struct strbuf tmp_file = STRBUF_INIT;
1553
1554                 strbuf_addf(&tmp_file,
1555                             "%s/info/commit-graphs/tmp_graph_XXXXXX",
1556                             ctx->odb->path);
1557                 ctx->graph_name = strbuf_detach(&tmp_file, NULL);
1558         } else {
1559                 ctx->graph_name = get_commit_graph_filename(ctx->odb);
1560         }
1561
1562         if (safe_create_leading_directories(ctx->graph_name)) {
1563                 UNLEAK(ctx->graph_name);
1564                 error(_("unable to create leading directories of %s"),
1565                         ctx->graph_name);
1566                 return -1;
1567         }
1568
1569         if (ctx->split) {
1570                 char *lock_name = get_chain_filename(ctx->odb);
1571
1572                 hold_lock_file_for_update(&lk, lock_name, LOCK_DIE_ON_ERROR);
1573
1574                 fd = git_mkstemp_mode(ctx->graph_name, 0444);
1575                 if (fd < 0) {
1576                         error(_("unable to create temporary graph layer"));
1577                         return -1;
1578                 }
1579
1580                 f = hashfd(fd, ctx->graph_name);
1581         } else {
1582                 hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
1583                 fd = lk.tempfile->fd;
1584                 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1585         }
1586
1587         chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1588         chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1589         chunk_ids[2] = GRAPH_CHUNKID_DATA;
1590         if (ctx->num_extra_edges) {
1591                 chunk_ids[num_chunks] = GRAPH_CHUNKID_EXTRAEDGES;
1592                 num_chunks++;
1593         }
1594         if (ctx->changed_paths) {
1595                 chunk_ids[num_chunks] = GRAPH_CHUNKID_BLOOMINDEXES;
1596                 num_chunks++;
1597                 chunk_ids[num_chunks] = GRAPH_CHUNKID_BLOOMDATA;
1598                 num_chunks++;
1599         }
1600         if (ctx->num_commit_graphs_after > 1) {
1601                 chunk_ids[num_chunks] = GRAPH_CHUNKID_BASE;
1602                 num_chunks++;
1603         }
1604
1605         chunk_ids[num_chunks] = 0;
1606
1607         chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1608         chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1609         chunk_offsets[2] = chunk_offsets[1] + hashsz * ctx->commits.nr;
1610         chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * ctx->commits.nr;
1611
1612         num_chunks = 3;
1613         if (ctx->num_extra_edges) {
1614                 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1615                                                 4 * ctx->num_extra_edges;
1616                 num_chunks++;
1617         }
1618         if (ctx->changed_paths) {
1619                 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1620                                                 sizeof(uint32_t) * ctx->commits.nr;
1621                 num_chunks++;
1622
1623                 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1624                                                 sizeof(uint32_t) * 3 + ctx->total_bloom_filter_data_size;
1625                 num_chunks++;
1626         }
1627         if (ctx->num_commit_graphs_after > 1) {
1628                 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1629                                                 hashsz * (ctx->num_commit_graphs_after - 1);
1630                 num_chunks++;
1631         }
1632
1633         hashwrite_be32(f, GRAPH_SIGNATURE);
1634
1635         hashwrite_u8(f, GRAPH_VERSION);
1636         hashwrite_u8(f, oid_version());
1637         hashwrite_u8(f, num_chunks);
1638         hashwrite_u8(f, ctx->num_commit_graphs_after - 1);
1639
1640         for (i = 0; i <= num_chunks; i++) {
1641                 uint32_t chunk_write[3];
1642
1643                 chunk_write[0] = htonl(chunk_ids[i]);
1644                 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1645                 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1646                 hashwrite(f, chunk_write, 12);
1647         }
1648
1649         if (ctx->report_progress) {
1650                 strbuf_addf(&progress_title,
1651                             Q_("Writing out commit graph in %d pass",
1652                                "Writing out commit graph in %d passes",
1653                                num_chunks),
1654                             num_chunks);
1655                 ctx->progress = start_delayed_progress(
1656                         progress_title.buf,
1657                         num_chunks * ctx->commits.nr);
1658         }
1659         write_graph_chunk_fanout(f, ctx);
1660         write_graph_chunk_oids(f, hashsz, ctx);
1661         write_graph_chunk_data(f, hashsz, ctx);
1662         if (ctx->num_extra_edges)
1663                 write_graph_chunk_extra_edges(f, ctx);
1664         if (ctx->changed_paths) {
1665                 write_graph_chunk_bloom_indexes(f, ctx);
1666                 write_graph_chunk_bloom_data(f, ctx, &bloom_settings);
1667         }
1668         if (ctx->num_commit_graphs_after > 1 &&
1669             write_graph_chunk_base(f, ctx)) {
1670                 return -1;
1671         }
1672         stop_progress(&ctx->progress);
1673         strbuf_release(&progress_title);
1674
1675         if (ctx->split && ctx->base_graph_name && ctx->num_commit_graphs_after > 1) {
1676                 char *new_base_hash = xstrdup(oid_to_hex(&ctx->new_base_graph->oid));
1677                 char *new_base_name = get_split_graph_filename(ctx->new_base_graph->odb, new_base_hash);
1678
1679                 free(ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2]);
1680                 free(ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2]);
1681                 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2] = new_base_name;
1682                 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2] = new_base_hash;
1683         }
1684
1685         close_commit_graph(ctx->r->objects);
1686         finalize_hashfile(f, file_hash.hash, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1687
1688         if (ctx->split) {
1689                 FILE *chainf = fdopen_lock_file(&lk, "w");
1690                 char *final_graph_name;
1691                 int result;
1692
1693                 close(fd);
1694
1695                 if (!chainf) {
1696                         error(_("unable to open commit-graph chain file"));
1697                         return -1;
1698                 }
1699
1700                 if (ctx->base_graph_name) {
1701                         const char *dest;
1702                         int idx = ctx->num_commit_graphs_after - 1;
1703                         if (ctx->num_commit_graphs_after > 1)
1704                                 idx--;
1705
1706                         dest = ctx->commit_graph_filenames_after[idx];
1707
1708                         if (strcmp(ctx->base_graph_name, dest)) {
1709                                 result = rename(ctx->base_graph_name, dest);
1710
1711                                 if (result) {
1712                                         error(_("failed to rename base commit-graph file"));
1713                                         return -1;
1714                                 }
1715                         }
1716                 } else {
1717                         char *graph_name = get_commit_graph_filename(ctx->odb);
1718                         unlink(graph_name);
1719                 }
1720
1721                 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1] = xstrdup(oid_to_hex(&file_hash));
1722                 final_graph_name = get_split_graph_filename(ctx->odb,
1723                                         ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1]);
1724                 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 1] = final_graph_name;
1725
1726                 result = rename(ctx->graph_name, final_graph_name);
1727
1728                 for (i = 0; i < ctx->num_commit_graphs_after; i++)
1729                         fprintf(lk.tempfile->fp, "%s\n", ctx->commit_graph_hash_after[i]);
1730
1731                 if (result) {
1732                         error(_("failed to rename temporary commit-graph file"));
1733                         return -1;
1734                 }
1735         }
1736
1737         commit_lock_file(&lk);
1738
1739         return 0;
1740 }
1741
1742 static void split_graph_merge_strategy(struct write_commit_graph_context *ctx)
1743 {
1744         struct commit_graph *g;
1745         uint32_t num_commits;
1746         enum commit_graph_split_flags flags = COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1747         uint32_t i;
1748
1749         int max_commits = 0;
1750         int size_mult = 2;
1751
1752         if (ctx->split_opts) {
1753                 max_commits = ctx->split_opts->max_commits;
1754
1755                 if (ctx->split_opts->size_multiple)
1756                         size_mult = ctx->split_opts->size_multiple;
1757
1758                 flags = ctx->split_opts->flags;
1759         }
1760
1761         g = ctx->r->objects->commit_graph;
1762         num_commits = ctx->commits.nr;
1763         if (flags == COMMIT_GRAPH_SPLIT_REPLACE)
1764                 ctx->num_commit_graphs_after = 1;
1765         else
1766                 ctx->num_commit_graphs_after = ctx->num_commit_graphs_before + 1;
1767
1768         if (flags != COMMIT_GRAPH_SPLIT_MERGE_PROHIBITED &&
1769             flags != COMMIT_GRAPH_SPLIT_REPLACE) {
1770                 while (g && (g->num_commits <= size_mult * num_commits ||
1771                             (max_commits && num_commits > max_commits))) {
1772                         if (g->odb != ctx->odb)
1773                                 break;
1774
1775                         num_commits += g->num_commits;
1776                         g = g->base_graph;
1777
1778                         ctx->num_commit_graphs_after--;
1779                 }
1780         }
1781
1782         if (flags != COMMIT_GRAPH_SPLIT_REPLACE)
1783                 ctx->new_base_graph = g;
1784         else if (ctx->num_commit_graphs_after != 1)
1785                 BUG("split_graph_merge_strategy: num_commit_graphs_after "
1786                     "should be 1 with --split=replace");
1787
1788         if (ctx->num_commit_graphs_after == 2) {
1789                 char *old_graph_name = get_commit_graph_filename(g->odb);
1790
1791                 if (!strcmp(g->filename, old_graph_name) &&
1792                     g->odb != ctx->odb) {
1793                         ctx->num_commit_graphs_after = 1;
1794                         ctx->new_base_graph = NULL;
1795                 }
1796
1797                 free(old_graph_name);
1798         }
1799
1800         CALLOC_ARRAY(ctx->commit_graph_filenames_after, ctx->num_commit_graphs_after);
1801         CALLOC_ARRAY(ctx->commit_graph_hash_after, ctx->num_commit_graphs_after);
1802
1803         for (i = 0; i < ctx->num_commit_graphs_after &&
1804                     i < ctx->num_commit_graphs_before; i++)
1805                 ctx->commit_graph_filenames_after[i] = xstrdup(ctx->commit_graph_filenames_before[i]);
1806
1807         i = ctx->num_commit_graphs_before - 1;
1808         g = ctx->r->objects->commit_graph;
1809
1810         while (g) {
1811                 if (i < ctx->num_commit_graphs_after)
1812                         ctx->commit_graph_hash_after[i] = xstrdup(oid_to_hex(&g->oid));
1813
1814                 i--;
1815                 g = g->base_graph;
1816         }
1817 }
1818
1819 static void merge_commit_graph(struct write_commit_graph_context *ctx,
1820                                struct commit_graph *g)
1821 {
1822         uint32_t i;
1823         uint32_t offset = g->num_commits_in_base;
1824
1825         ALLOC_GROW(ctx->commits.list, ctx->commits.nr + g->num_commits, ctx->commits.alloc);
1826
1827         for (i = 0; i < g->num_commits; i++) {
1828                 struct object_id oid;
1829                 struct commit *result;
1830
1831                 display_progress(ctx->progress, i + 1);
1832
1833                 load_oid_from_graph(g, i + offset, &oid);
1834
1835                 /* only add commits if they still exist in the repo */
1836                 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1837
1838                 if (result) {
1839                         ctx->commits.list[ctx->commits.nr] = result;
1840                         ctx->commits.nr++;
1841                 }
1842         }
1843 }
1844
1845 static int commit_compare(const void *_a, const void *_b)
1846 {
1847         const struct commit *a = *(const struct commit **)_a;
1848         const struct commit *b = *(const struct commit **)_b;
1849         return oidcmp(&a->object.oid, &b->object.oid);
1850 }
1851
1852 static void sort_and_scan_merged_commits(struct write_commit_graph_context *ctx)
1853 {
1854         uint32_t i;
1855
1856         if (ctx->report_progress)
1857                 ctx->progress = start_delayed_progress(
1858                                         _("Scanning merged commits"),
1859                                         ctx->commits.nr);
1860
1861         QSORT(ctx->commits.list, ctx->commits.nr, commit_compare);
1862
1863         ctx->num_extra_edges = 0;
1864         for (i = 0; i < ctx->commits.nr; i++) {
1865                 display_progress(ctx->progress, i);
1866
1867                 if (i && oideq(&ctx->commits.list[i - 1]->object.oid,
1868                           &ctx->commits.list[i]->object.oid)) {
1869                         die(_("unexpected duplicate commit id %s"),
1870                             oid_to_hex(&ctx->commits.list[i]->object.oid));
1871                 } else {
1872                         unsigned int num_parents;
1873
1874                         num_parents = commit_list_count(ctx->commits.list[i]->parents);
1875                         if (num_parents > 2)
1876                                 ctx->num_extra_edges += num_parents - 1;
1877                 }
1878         }
1879
1880         stop_progress(&ctx->progress);
1881 }
1882
1883 static void merge_commit_graphs(struct write_commit_graph_context *ctx)
1884 {
1885         struct commit_graph *g = ctx->r->objects->commit_graph;
1886         uint32_t current_graph_number = ctx->num_commit_graphs_before;
1887
1888         while (g && current_graph_number >= ctx->num_commit_graphs_after) {
1889                 current_graph_number--;
1890
1891                 if (ctx->report_progress)
1892                         ctx->progress = start_delayed_progress(_("Merging commit-graph"), 0);
1893
1894                 merge_commit_graph(ctx, g);
1895                 stop_progress(&ctx->progress);
1896
1897                 g = g->base_graph;
1898         }
1899
1900         if (g) {
1901                 ctx->new_base_graph = g;
1902                 ctx->new_num_commits_in_base = g->num_commits + g->num_commits_in_base;
1903         }
1904
1905         if (ctx->new_base_graph)
1906                 ctx->base_graph_name = xstrdup(ctx->new_base_graph->filename);
1907
1908         sort_and_scan_merged_commits(ctx);
1909 }
1910
1911 static void mark_commit_graphs(struct write_commit_graph_context *ctx)
1912 {
1913         uint32_t i;
1914         time_t now = time(NULL);
1915
1916         for (i = ctx->num_commit_graphs_after - 1; i < ctx->num_commit_graphs_before; i++) {
1917                 struct stat st;
1918                 struct utimbuf updated_time;
1919
1920                 stat(ctx->commit_graph_filenames_before[i], &st);
1921
1922                 updated_time.actime = st.st_atime;
1923                 updated_time.modtime = now;
1924                 utime(ctx->commit_graph_filenames_before[i], &updated_time);
1925         }
1926 }
1927
1928 static void expire_commit_graphs(struct write_commit_graph_context *ctx)
1929 {
1930         struct strbuf path = STRBUF_INIT;
1931         DIR *dir;
1932         struct dirent *de;
1933         size_t dirnamelen;
1934         timestamp_t expire_time = time(NULL);
1935
1936         if (ctx->split_opts && ctx->split_opts->expire_time)
1937                 expire_time = ctx->split_opts->expire_time;
1938         if (!ctx->split) {
1939                 char *chain_file_name = get_chain_filename(ctx->odb);
1940                 unlink(chain_file_name);
1941                 free(chain_file_name);
1942                 ctx->num_commit_graphs_after = 0;
1943         }
1944
1945         strbuf_addstr(&path, ctx->odb->path);
1946         strbuf_addstr(&path, "/info/commit-graphs");
1947         dir = opendir(path.buf);
1948
1949         if (!dir)
1950                 goto out;
1951
1952         strbuf_addch(&path, '/');
1953         dirnamelen = path.len;
1954         while ((de = readdir(dir)) != NULL) {
1955                 struct stat st;
1956                 uint32_t i, found = 0;
1957
1958                 strbuf_setlen(&path, dirnamelen);
1959                 strbuf_addstr(&path, de->d_name);
1960
1961                 stat(path.buf, &st);
1962
1963                 if (st.st_mtime > expire_time)
1964                         continue;
1965                 if (path.len < 6 || strcmp(path.buf + path.len - 6, ".graph"))
1966                         continue;
1967
1968                 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
1969                         if (!strcmp(ctx->commit_graph_filenames_after[i],
1970                                     path.buf)) {
1971                                 found = 1;
1972                                 break;
1973                         }
1974                 }
1975
1976                 if (!found)
1977                         unlink(path.buf);
1978         }
1979
1980 out:
1981         strbuf_release(&path);
1982 }
1983
1984 int write_commit_graph(struct object_directory *odb,
1985                        struct string_list *pack_indexes,
1986                        struct oidset *commits,
1987                        enum commit_graph_write_flags flags,
1988                        const struct split_commit_graph_opts *split_opts)
1989 {
1990         struct write_commit_graph_context *ctx;
1991         uint32_t i, count_distinct = 0;
1992         int res = 0;
1993         int replace = 0;
1994
1995         if (!commit_graph_compatible(the_repository))
1996                 return 0;
1997
1998         ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
1999         ctx->r = the_repository;
2000         ctx->odb = odb;
2001         ctx->append = flags & COMMIT_GRAPH_WRITE_APPEND ? 1 : 0;
2002         ctx->report_progress = flags & COMMIT_GRAPH_WRITE_PROGRESS ? 1 : 0;
2003         ctx->split = flags & COMMIT_GRAPH_WRITE_SPLIT ? 1 : 0;
2004         ctx->split_opts = split_opts;
2005         ctx->changed_paths = flags & COMMIT_GRAPH_WRITE_BLOOM_FILTERS ? 1 : 0;
2006         ctx->total_bloom_filter_data_size = 0;
2007
2008         if (ctx->split) {
2009                 struct commit_graph *g;
2010                 prepare_commit_graph(ctx->r);
2011
2012                 g = ctx->r->objects->commit_graph;
2013
2014                 while (g) {
2015                         ctx->num_commit_graphs_before++;
2016                         g = g->base_graph;
2017                 }
2018
2019                 if (ctx->num_commit_graphs_before) {
2020                         ALLOC_ARRAY(ctx->commit_graph_filenames_before, ctx->num_commit_graphs_before);
2021                         i = ctx->num_commit_graphs_before;
2022                         g = ctx->r->objects->commit_graph;
2023
2024                         while (g) {
2025                                 ctx->commit_graph_filenames_before[--i] = xstrdup(g->filename);
2026                                 g = g->base_graph;
2027                         }
2028                 }
2029
2030                 if (ctx->split_opts)
2031                         replace = ctx->split_opts->flags & COMMIT_GRAPH_SPLIT_REPLACE;
2032         }
2033
2034         ctx->approx_nr_objects = approximate_object_count();
2035         ctx->oids.alloc = ctx->approx_nr_objects / 32;
2036
2037         if (ctx->split && split_opts && ctx->oids.alloc > split_opts->max_commits)
2038                 ctx->oids.alloc = split_opts->max_commits;
2039
2040         if (ctx->append) {
2041                 prepare_commit_graph_one(ctx->r, ctx->odb);
2042                 if (ctx->r->objects->commit_graph)
2043                         ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
2044         }
2045
2046         if (ctx->oids.alloc < 1024)
2047                 ctx->oids.alloc = 1024;
2048         ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
2049
2050         if (ctx->append && ctx->r->objects->commit_graph) {
2051                 struct commit_graph *g = ctx->r->objects->commit_graph;
2052                 for (i = 0; i < g->num_commits; i++) {
2053                         const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
2054                         hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
2055                 }
2056         }
2057
2058         if (pack_indexes) {
2059                 ctx->order_by_pack = 1;
2060                 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
2061                         goto cleanup;
2062         }
2063
2064         if (commits) {
2065                 if ((res = fill_oids_from_commits(ctx, commits)))
2066                         goto cleanup;
2067         }
2068
2069         if (!pack_indexes && !commits) {
2070                 ctx->order_by_pack = 1;
2071                 fill_oids_from_all_packs(ctx);
2072         }
2073
2074         close_reachable(ctx);
2075
2076         count_distinct = count_distinct_commits(ctx);
2077
2078         if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
2079                 error(_("the commit graph format cannot write %d commits"), count_distinct);
2080                 res = -1;
2081                 goto cleanup;
2082         }
2083
2084         ctx->commits.alloc = count_distinct;
2085         ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
2086
2087         copy_oids_to_commits(ctx);
2088
2089         if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
2090                 error(_("too many commits to write graph"));
2091                 res = -1;
2092                 goto cleanup;
2093         }
2094
2095         if (!ctx->commits.nr && !replace)
2096                 goto cleanup;
2097
2098         if (ctx->split) {
2099                 split_graph_merge_strategy(ctx);
2100
2101                 if (!replace)
2102                         merge_commit_graphs(ctx);
2103         } else
2104                 ctx->num_commit_graphs_after = 1;
2105
2106         compute_generation_numbers(ctx);
2107
2108         if (ctx->changed_paths)
2109                 compute_bloom_filters(ctx);
2110
2111         res = write_commit_graph_file(ctx);
2112
2113         if (ctx->split)
2114                 mark_commit_graphs(ctx);
2115
2116         expire_commit_graphs(ctx);
2117
2118 cleanup:
2119         free(ctx->graph_name);
2120         free(ctx->commits.list);
2121         free(ctx->oids.list);
2122
2123         if (ctx->commit_graph_filenames_after) {
2124                 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
2125                         free(ctx->commit_graph_filenames_after[i]);
2126                         free(ctx->commit_graph_hash_after[i]);
2127                 }
2128
2129                 for (i = 0; i < ctx->num_commit_graphs_before; i++)
2130                         free(ctx->commit_graph_filenames_before[i]);
2131
2132                 free(ctx->commit_graph_filenames_after);
2133                 free(ctx->commit_graph_filenames_before);
2134                 free(ctx->commit_graph_hash_after);
2135         }
2136
2137         free(ctx);
2138
2139         return res;
2140 }
2141
2142 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
2143 static int verify_commit_graph_error;
2144
2145 static void graph_report(const char *fmt, ...)
2146 {
2147         va_list ap;
2148
2149         verify_commit_graph_error = 1;
2150         va_start(ap, fmt);
2151         vfprintf(stderr, fmt, ap);
2152         fprintf(stderr, "\n");
2153         va_end(ap);
2154 }
2155
2156 #define GENERATION_ZERO_EXISTS 1
2157 #define GENERATION_NUMBER_EXISTS 2
2158
2159 int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
2160 {
2161         uint32_t i, cur_fanout_pos = 0;
2162         struct object_id prev_oid, cur_oid, checksum;
2163         int generation_zero = 0;
2164         struct hashfile *f;
2165         int devnull;
2166         struct progress *progress = NULL;
2167         int local_error = 0;
2168
2169         if (!g) {
2170                 graph_report("no commit-graph file loaded");
2171                 return 1;
2172         }
2173
2174         verify_commit_graph_error = verify_commit_graph_lite(g);
2175         if (verify_commit_graph_error)
2176                 return verify_commit_graph_error;
2177
2178         devnull = open("/dev/null", O_WRONLY);
2179         f = hashfd(devnull, NULL);
2180         hashwrite(f, g->data, g->data_len - g->hash_len);
2181         finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
2182         if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
2183                 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
2184                 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
2185         }
2186
2187         for (i = 0; i < g->num_commits; i++) {
2188                 struct commit *graph_commit;
2189
2190                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2191
2192                 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
2193                         graph_report(_("commit-graph has incorrect OID order: %s then %s"),
2194                                      oid_to_hex(&prev_oid),
2195                                      oid_to_hex(&cur_oid));
2196
2197                 oidcpy(&prev_oid, &cur_oid);
2198
2199                 while (cur_oid.hash[0] > cur_fanout_pos) {
2200                         uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2201
2202                         if (i != fanout_value)
2203                                 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2204                                              cur_fanout_pos, fanout_value, i);
2205                         cur_fanout_pos++;
2206                 }
2207
2208                 graph_commit = lookup_commit(r, &cur_oid);
2209                 if (!parse_commit_in_graph_one(r, g, graph_commit))
2210                         graph_report(_("failed to parse commit %s from commit-graph"),
2211                                      oid_to_hex(&cur_oid));
2212         }
2213
2214         while (cur_fanout_pos < 256) {
2215                 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2216
2217                 if (g->num_commits != fanout_value)
2218                         graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2219                                      cur_fanout_pos, fanout_value, i);
2220
2221                 cur_fanout_pos++;
2222         }
2223
2224         if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
2225                 return verify_commit_graph_error;
2226
2227         if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
2228                 progress = start_progress(_("Verifying commits in commit graph"),
2229                                         g->num_commits);
2230
2231         for (i = 0; i < g->num_commits; i++) {
2232                 struct commit *graph_commit, *odb_commit;
2233                 struct commit_list *graph_parents, *odb_parents;
2234                 uint32_t max_generation = 0;
2235
2236                 display_progress(progress, i + 1);
2237                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2238
2239                 graph_commit = lookup_commit(r, &cur_oid);
2240                 odb_commit = (struct commit *)create_object(r, &cur_oid, alloc_commit_node(r));
2241                 if (parse_commit_internal(odb_commit, 0, 0)) {
2242                         graph_report(_("failed to parse commit %s from object database for commit-graph"),
2243                                      oid_to_hex(&cur_oid));
2244                         continue;
2245                 }
2246
2247                 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
2248                            get_commit_tree_oid(odb_commit)))
2249                         graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2250                                      oid_to_hex(&cur_oid),
2251                                      oid_to_hex(get_commit_tree_oid(graph_commit)),
2252                                      oid_to_hex(get_commit_tree_oid(odb_commit)));
2253
2254                 graph_parents = graph_commit->parents;
2255                 odb_parents = odb_commit->parents;
2256
2257                 while (graph_parents) {
2258                         if (odb_parents == NULL) {
2259                                 graph_report(_("commit-graph parent list for commit %s is too long"),
2260                                              oid_to_hex(&cur_oid));
2261                                 break;
2262                         }
2263
2264                         /* parse parent in case it is in a base graph */
2265                         parse_commit_in_graph_one(r, g, graph_parents->item);
2266
2267                         if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
2268                                 graph_report(_("commit-graph parent for %s is %s != %s"),
2269                                              oid_to_hex(&cur_oid),
2270                                              oid_to_hex(&graph_parents->item->object.oid),
2271                                              oid_to_hex(&odb_parents->item->object.oid));
2272
2273                         if (graph_parents->item->generation > max_generation)
2274                                 max_generation = graph_parents->item->generation;
2275
2276                         graph_parents = graph_parents->next;
2277                         odb_parents = odb_parents->next;
2278                 }
2279
2280                 if (odb_parents != NULL)
2281                         graph_report(_("commit-graph parent list for commit %s terminates early"),
2282                                      oid_to_hex(&cur_oid));
2283
2284                 if (!graph_commit->generation) {
2285                         if (generation_zero == GENERATION_NUMBER_EXISTS)
2286                                 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2287                                              oid_to_hex(&cur_oid));
2288                         generation_zero = GENERATION_ZERO_EXISTS;
2289                 } else if (generation_zero == GENERATION_ZERO_EXISTS)
2290                         graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2291                                      oid_to_hex(&cur_oid));
2292
2293                 if (generation_zero == GENERATION_ZERO_EXISTS)
2294                         continue;
2295
2296                 /*
2297                  * If one of our parents has generation GENERATION_NUMBER_MAX, then
2298                  * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
2299                  * extra logic in the following condition.
2300                  */
2301                 if (max_generation == GENERATION_NUMBER_MAX)
2302                         max_generation--;
2303
2304                 if (graph_commit->generation != max_generation + 1)
2305                         graph_report(_("commit-graph generation for commit %s is %u != %u"),
2306                                      oid_to_hex(&cur_oid),
2307                                      graph_commit->generation,
2308                                      max_generation + 1);
2309
2310                 if (graph_commit->date != odb_commit->date)
2311                         graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
2312                                      oid_to_hex(&cur_oid),
2313                                      graph_commit->date,
2314                                      odb_commit->date);
2315         }
2316         stop_progress(&progress);
2317
2318         local_error = verify_commit_graph_error;
2319
2320         if (!(flags & COMMIT_GRAPH_VERIFY_SHALLOW) && g->base_graph)
2321                 local_error |= verify_commit_graph(r, g->base_graph, flags);
2322
2323         return local_error;
2324 }
2325
2326 void free_commit_graph(struct commit_graph *g)
2327 {
2328         if (!g)
2329                 return;
2330         if (g->data) {
2331                 munmap((void *)g->data, g->data_len);
2332                 g->data = NULL;
2333         }
2334         free(g->filename);
2335         free(g->bloom_filter_settings);
2336         free(g);
2337 }
2338
2339 void disable_commit_graph(struct repository *r)
2340 {
2341         r->commit_graph_disabled = 1;
2342 }