commit-graph: create write_commit_graph_context
[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
20 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
21 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
22 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
23 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
24 #define GRAPH_CHUNKID_EXTRAEDGES 0x45444745 /* "EDGE" */
25
26 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
27
28 #define GRAPH_VERSION_1 0x1
29 #define GRAPH_VERSION GRAPH_VERSION_1
30
31 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
32 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
33 #define GRAPH_PARENT_NONE 0x70000000
34
35 #define GRAPH_LAST_EDGE 0x80000000
36
37 #define GRAPH_HEADER_SIZE 8
38 #define GRAPH_FANOUT_SIZE (4 * 256)
39 #define GRAPH_CHUNKLOOKUP_WIDTH 12
40 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
41                         + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
42
43 char *get_commit_graph_filename(const char *obj_dir)
44 {
45         return xstrfmt("%s/info/commit-graph", obj_dir);
46 }
47
48 static uint8_t oid_version(void)
49 {
50         return 1;
51 }
52
53 static struct commit_graph *alloc_commit_graph(void)
54 {
55         struct commit_graph *g = xcalloc(1, sizeof(*g));
56         g->graph_fd = -1;
57
58         return g;
59 }
60
61 extern int read_replace_refs;
62
63 static int commit_graph_compatible(struct repository *r)
64 {
65         if (!r->gitdir)
66                 return 0;
67
68         if (read_replace_refs) {
69                 prepare_replace_object(r);
70                 if (hashmap_get_size(&r->objects->replace_map->map))
71                         return 0;
72         }
73
74         prepare_commit_graft(r);
75         if (r->parsed_objects && r->parsed_objects->grafts_nr)
76                 return 0;
77         if (is_repository_shallow(r))
78                 return 0;
79
80         return 1;
81 }
82
83 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
84 {
85         *fd = git_open(graph_file);
86         if (*fd < 0)
87                 return 0;
88         if (fstat(*fd, st)) {
89                 close(*fd);
90                 return 0;
91         }
92         return 1;
93 }
94
95 struct commit_graph *load_commit_graph_one_fd_st(int fd, struct stat *st)
96 {
97         void *graph_map;
98         size_t graph_size;
99         struct commit_graph *ret;
100
101         graph_size = xsize_t(st->st_size);
102
103         if (graph_size < GRAPH_MIN_SIZE) {
104                 close(fd);
105                 error(_("commit-graph file is too small"));
106                 return NULL;
107         }
108         graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
109         ret = parse_commit_graph(graph_map, fd, graph_size);
110
111         if (!ret) {
112                 munmap(graph_map, graph_size);
113                 close(fd);
114         }
115
116         return ret;
117 }
118
119 static int verify_commit_graph_lite(struct commit_graph *g)
120 {
121         /*
122          * Basic validation shared between parse_commit_graph()
123          * which'll be called every time the graph is used, and the
124          * much more expensive verify_commit_graph() used by
125          * "commit-graph verify".
126          *
127          * There should only be very basic checks here to ensure that
128          * we don't e.g. segfault in fill_commit_in_graph(), but
129          * because this is a very hot codepath nothing that e.g. loops
130          * over g->num_commits, or runs a checksum on the commit-graph
131          * itself.
132          */
133         if (!g->chunk_oid_fanout) {
134                 error("commit-graph is missing the OID Fanout chunk");
135                 return 1;
136         }
137         if (!g->chunk_oid_lookup) {
138                 error("commit-graph is missing the OID Lookup chunk");
139                 return 1;
140         }
141         if (!g->chunk_commit_data) {
142                 error("commit-graph is missing the Commit Data chunk");
143                 return 1;
144         }
145
146         return 0;
147 }
148
149 struct commit_graph *parse_commit_graph(void *graph_map, int fd,
150                                         size_t graph_size)
151 {
152         const unsigned char *data, *chunk_lookup;
153         uint32_t i;
154         struct commit_graph *graph;
155         uint64_t last_chunk_offset;
156         uint32_t last_chunk_id;
157         uint32_t graph_signature;
158         unsigned char graph_version, hash_version;
159
160         if (!graph_map)
161                 return NULL;
162
163         if (graph_size < GRAPH_MIN_SIZE)
164                 return NULL;
165
166         data = (const unsigned char *)graph_map;
167
168         graph_signature = get_be32(data);
169         if (graph_signature != GRAPH_SIGNATURE) {
170                 error(_("commit-graph signature %X does not match signature %X"),
171                       graph_signature, GRAPH_SIGNATURE);
172                 return NULL;
173         }
174
175         graph_version = *(unsigned char*)(data + 4);
176         if (graph_version != GRAPH_VERSION) {
177                 error(_("commit-graph version %X does not match version %X"),
178                       graph_version, GRAPH_VERSION);
179                 return NULL;
180         }
181
182         hash_version = *(unsigned char*)(data + 5);
183         if (hash_version != oid_version()) {
184                 error(_("commit-graph hash version %X does not match version %X"),
185                       hash_version, oid_version());
186                 return NULL;
187         }
188
189         graph = alloc_commit_graph();
190
191         graph->hash_len = the_hash_algo->rawsz;
192         graph->num_chunks = *(unsigned char*)(data + 6);
193         graph->graph_fd = fd;
194         graph->data = graph_map;
195         graph->data_len = graph_size;
196
197         last_chunk_id = 0;
198         last_chunk_offset = 8;
199         chunk_lookup = data + 8;
200         for (i = 0; i < graph->num_chunks; i++) {
201                 uint32_t chunk_id;
202                 uint64_t chunk_offset;
203                 int chunk_repeated = 0;
204
205                 if (data + graph_size - chunk_lookup <
206                     GRAPH_CHUNKLOOKUP_WIDTH) {
207                         error(_("commit-graph chunk lookup table entry missing; file may be incomplete"));
208                         free(graph);
209                         return NULL;
210                 }
211
212                 chunk_id = get_be32(chunk_lookup + 0);
213                 chunk_offset = get_be64(chunk_lookup + 4);
214
215                 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
216
217                 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
218                         error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
219                               (uint32_t)chunk_offset);
220                         free(graph);
221                         return NULL;
222                 }
223
224                 switch (chunk_id) {
225                 case GRAPH_CHUNKID_OIDFANOUT:
226                         if (graph->chunk_oid_fanout)
227                                 chunk_repeated = 1;
228                         else
229                                 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
230                         break;
231
232                 case GRAPH_CHUNKID_OIDLOOKUP:
233                         if (graph->chunk_oid_lookup)
234                                 chunk_repeated = 1;
235                         else
236                                 graph->chunk_oid_lookup = data + chunk_offset;
237                         break;
238
239                 case GRAPH_CHUNKID_DATA:
240                         if (graph->chunk_commit_data)
241                                 chunk_repeated = 1;
242                         else
243                                 graph->chunk_commit_data = data + chunk_offset;
244                         break;
245
246                 case GRAPH_CHUNKID_EXTRAEDGES:
247                         if (graph->chunk_extra_edges)
248                                 chunk_repeated = 1;
249                         else
250                                 graph->chunk_extra_edges = data + chunk_offset;
251                         break;
252                 }
253
254                 if (chunk_repeated) {
255                         error(_("commit-graph chunk id %08x appears multiple times"), chunk_id);
256                         free(graph);
257                         return NULL;
258                 }
259
260                 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
261                 {
262                         graph->num_commits = (chunk_offset - last_chunk_offset)
263                                              / graph->hash_len;
264                 }
265
266                 last_chunk_id = chunk_id;
267                 last_chunk_offset = chunk_offset;
268         }
269
270         if (verify_commit_graph_lite(graph))
271                 return NULL;
272
273         return graph;
274 }
275
276 static struct commit_graph *load_commit_graph_one(const char *graph_file)
277 {
278
279         struct stat st;
280         int fd;
281         int open_ok = open_commit_graph(graph_file, &fd, &st);
282
283         if (!open_ok)
284                 return NULL;
285
286         return load_commit_graph_one_fd_st(fd, &st);
287 }
288
289 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
290 {
291         char *graph_name;
292
293         if (r->objects->commit_graph)
294                 return;
295
296         graph_name = get_commit_graph_filename(obj_dir);
297         r->objects->commit_graph =
298                 load_commit_graph_one(graph_name);
299
300         FREE_AND_NULL(graph_name);
301 }
302
303 /*
304  * Return 1 if commit_graph is non-NULL, and 0 otherwise.
305  *
306  * On the first invocation, this function attemps to load the commit
307  * graph if the_repository is configured to have one.
308  */
309 static int prepare_commit_graph(struct repository *r)
310 {
311         struct object_directory *odb;
312         int config_value;
313
314         if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
315                 die("dying as requested by the '%s' variable on commit-graph load!",
316                     GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD);
317
318         if (r->objects->commit_graph_attempted)
319                 return !!r->objects->commit_graph;
320         r->objects->commit_graph_attempted = 1;
321
322         if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
323             (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
324             !config_value))
325                 /*
326                  * This repository is not configured to use commit graphs, so
327                  * do not load one. (But report commit_graph_attempted anyway
328                  * so that commit graph loading is not attempted again for this
329                  * repository.)
330                  */
331                 return 0;
332
333         if (!commit_graph_compatible(r))
334                 return 0;
335
336         prepare_alt_odb(r);
337         for (odb = r->objects->odb;
338              !r->objects->commit_graph && odb;
339              odb = odb->next)
340                 prepare_commit_graph_one(r, odb->path);
341         return !!r->objects->commit_graph;
342 }
343
344 int generation_numbers_enabled(struct repository *r)
345 {
346         uint32_t first_generation;
347         struct commit_graph *g;
348         if (!prepare_commit_graph(r))
349                return 0;
350
351         g = r->objects->commit_graph;
352
353         if (!g->num_commits)
354                 return 0;
355
356         first_generation = get_be32(g->chunk_commit_data +
357                                     g->hash_len + 8) >> 2;
358
359         return !!first_generation;
360 }
361
362 void close_commit_graph(struct repository *r)
363 {
364         free_commit_graph(r->objects->commit_graph);
365         r->objects->commit_graph = NULL;
366 }
367
368 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
369 {
370         return bsearch_hash(oid->hash, g->chunk_oid_fanout,
371                             g->chunk_oid_lookup, g->hash_len, pos);
372 }
373
374 static struct commit_list **insert_parent_or_die(struct repository *r,
375                                                  struct commit_graph *g,
376                                                  uint64_t pos,
377                                                  struct commit_list **pptr)
378 {
379         struct commit *c;
380         struct object_id oid;
381
382         if (pos >= g->num_commits)
383                 die("invalid parent position %"PRIu64, pos);
384
385         hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
386         c = lookup_commit(r, &oid);
387         if (!c)
388                 die(_("could not find commit %s"), oid_to_hex(&oid));
389         c->graph_pos = pos;
390         return &commit_list_insert(c, pptr)->next;
391 }
392
393 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
394 {
395         const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
396         item->graph_pos = pos;
397         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
398 }
399
400 static int fill_commit_in_graph(struct repository *r,
401                                 struct commit *item,
402                                 struct commit_graph *g, uint32_t pos)
403 {
404         uint32_t edge_value;
405         uint32_t *parent_data_ptr;
406         uint64_t date_low, date_high;
407         struct commit_list **pptr;
408         const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
409
410         item->object.parsed = 1;
411         item->graph_pos = pos;
412
413         item->maybe_tree = NULL;
414
415         date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
416         date_low = get_be32(commit_data + g->hash_len + 12);
417         item->date = (timestamp_t)((date_high << 32) | date_low);
418
419         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
420
421         pptr = &item->parents;
422
423         edge_value = get_be32(commit_data + g->hash_len);
424         if (edge_value == GRAPH_PARENT_NONE)
425                 return 1;
426         pptr = insert_parent_or_die(r, g, edge_value, pptr);
427
428         edge_value = get_be32(commit_data + g->hash_len + 4);
429         if (edge_value == GRAPH_PARENT_NONE)
430                 return 1;
431         if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
432                 pptr = insert_parent_or_die(r, g, edge_value, pptr);
433                 return 1;
434         }
435
436         parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
437                           4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
438         do {
439                 edge_value = get_be32(parent_data_ptr);
440                 pptr = insert_parent_or_die(r, g,
441                                             edge_value & GRAPH_EDGE_LAST_MASK,
442                                             pptr);
443                 parent_data_ptr++;
444         } while (!(edge_value & GRAPH_LAST_EDGE));
445
446         return 1;
447 }
448
449 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
450 {
451         if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
452                 *pos = item->graph_pos;
453                 return 1;
454         } else {
455                 return bsearch_graph(g, &(item->object.oid), pos);
456         }
457 }
458
459 static int parse_commit_in_graph_one(struct repository *r,
460                                      struct commit_graph *g,
461                                      struct commit *item)
462 {
463         uint32_t pos;
464
465         if (item->object.parsed)
466                 return 1;
467
468         if (find_commit_in_graph(item, g, &pos))
469                 return fill_commit_in_graph(r, item, g, pos);
470
471         return 0;
472 }
473
474 int parse_commit_in_graph(struct repository *r, struct commit *item)
475 {
476         if (!prepare_commit_graph(r))
477                 return 0;
478         return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
479 }
480
481 void load_commit_graph_info(struct repository *r, struct commit *item)
482 {
483         uint32_t pos;
484         if (!prepare_commit_graph(r))
485                 return;
486         if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
487                 fill_commit_graph_info(item, r->objects->commit_graph, pos);
488 }
489
490 static struct tree *load_tree_for_commit(struct repository *r,
491                                          struct commit_graph *g,
492                                          struct commit *c)
493 {
494         struct object_id oid;
495         const unsigned char *commit_data = g->chunk_commit_data +
496                                            GRAPH_DATA_WIDTH * (c->graph_pos);
497
498         hashcpy(oid.hash, commit_data);
499         c->maybe_tree = lookup_tree(r, &oid);
500
501         return c->maybe_tree;
502 }
503
504 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
505                                                  struct commit_graph *g,
506                                                  const struct commit *c)
507 {
508         if (c->maybe_tree)
509                 return c->maybe_tree;
510         if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
511                 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
512
513         return load_tree_for_commit(r, g, (struct commit *)c);
514 }
515
516 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
517 {
518         return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
519 }
520
521 struct packed_commit_list {
522         struct commit **list;
523         int nr;
524         int alloc;
525 };
526
527 struct packed_oid_list {
528         struct object_id *list;
529         int nr;
530         int alloc;
531 };
532
533 struct write_commit_graph_context {
534         struct repository *r;
535         const char *obj_dir;
536         char *graph_name;
537         struct packed_oid_list oids;
538         struct packed_commit_list commits;
539         int num_extra_edges;
540         unsigned long approx_nr_objects;
541         struct progress *progress;
542         int progress_done;
543         uint64_t progress_cnt;
544         unsigned append:1,
545                  report_progress:1;
546 };
547
548 static void write_graph_chunk_fanout(struct hashfile *f,
549                                      struct write_commit_graph_context *ctx)
550 {
551         int i, count = 0;
552         struct commit **list = ctx->commits.list;
553
554         /*
555          * Write the first-level table (the list is sorted,
556          * but we use a 256-entry lookup to be able to avoid
557          * having to do eight extra binary search iterations).
558          */
559         for (i = 0; i < 256; i++) {
560                 while (count < ctx->commits.nr) {
561                         if ((*list)->object.oid.hash[0] != i)
562                                 break;
563                         display_progress(ctx->progress, ++ctx->progress_cnt);
564                         count++;
565                         list++;
566                 }
567
568                 hashwrite_be32(f, count);
569         }
570 }
571
572 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
573                                    struct write_commit_graph_context *ctx)
574 {
575         struct commit **list = ctx->commits.list;
576         int count;
577         for (count = 0; count < ctx->commits.nr; count++, list++) {
578                 display_progress(ctx->progress, ++ctx->progress_cnt);
579                 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
580         }
581 }
582
583 static const unsigned char *commit_to_sha1(size_t index, void *table)
584 {
585         struct commit **commits = table;
586         return commits[index]->object.oid.hash;
587 }
588
589 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
590                                    struct write_commit_graph_context *ctx)
591 {
592         struct commit **list = ctx->commits.list;
593         struct commit **last = ctx->commits.list + ctx->commits.nr;
594         uint32_t num_extra_edges = 0;
595
596         while (list < last) {
597                 struct commit_list *parent;
598                 int edge_value;
599                 uint32_t packedDate[2];
600                 display_progress(ctx->progress, ++ctx->progress_cnt);
601
602                 parse_commit_no_graph(*list);
603                 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
604
605                 parent = (*list)->parents;
606
607                 if (!parent)
608                         edge_value = GRAPH_PARENT_NONE;
609                 else {
610                         edge_value = sha1_pos(parent->item->object.oid.hash,
611                                               ctx->commits.list,
612                                               ctx->commits.nr,
613                                               commit_to_sha1);
614
615                         if (edge_value < 0)
616                                 BUG("missing parent %s for commit %s",
617                                     oid_to_hex(&parent->item->object.oid),
618                                     oid_to_hex(&(*list)->object.oid));
619                 }
620
621                 hashwrite_be32(f, edge_value);
622
623                 if (parent)
624                         parent = parent->next;
625
626                 if (!parent)
627                         edge_value = GRAPH_PARENT_NONE;
628                 else if (parent->next)
629                         edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
630                 else {
631                         edge_value = sha1_pos(parent->item->object.oid.hash,
632                                               ctx->commits.list,
633                                               ctx->commits.nr,
634                                               commit_to_sha1);
635                         if (edge_value < 0)
636                                 BUG("missing parent %s for commit %s",
637                                     oid_to_hex(&parent->item->object.oid),
638                                     oid_to_hex(&(*list)->object.oid));
639                 }
640
641                 hashwrite_be32(f, edge_value);
642
643                 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
644                         do {
645                                 num_extra_edges++;
646                                 parent = parent->next;
647                         } while (parent);
648                 }
649
650                 if (sizeof((*list)->date) > 4)
651                         packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
652                 else
653                         packedDate[0] = 0;
654
655                 packedDate[0] |= htonl((*list)->generation << 2);
656
657                 packedDate[1] = htonl((*list)->date);
658                 hashwrite(f, packedDate, 8);
659
660                 list++;
661         }
662 }
663
664 static void write_graph_chunk_extra_edges(struct hashfile *f,
665                                           struct write_commit_graph_context *ctx)
666 {
667         struct commit **list = ctx->commits.list;
668         struct commit **last = ctx->commits.list + ctx->commits.nr;
669         struct commit_list *parent;
670
671         while (list < last) {
672                 int num_parents = 0;
673
674                 display_progress(ctx->progress, ++ctx->progress_cnt);
675
676                 for (parent = (*list)->parents; num_parents < 3 && parent;
677                      parent = parent->next)
678                         num_parents++;
679
680                 if (num_parents <= 2) {
681                         list++;
682                         continue;
683                 }
684
685                 /* Since num_parents > 2, this initializer is safe. */
686                 for (parent = (*list)->parents->next; parent; parent = parent->next) {
687                         int edge_value = sha1_pos(parent->item->object.oid.hash,
688                                                   ctx->commits.list,
689                                                   ctx->commits.nr,
690                                                   commit_to_sha1);
691
692                         if (edge_value < 0)
693                                 BUG("missing parent %s for commit %s",
694                                     oid_to_hex(&parent->item->object.oid),
695                                     oid_to_hex(&(*list)->object.oid));
696                         else if (!parent->next)
697                                 edge_value |= GRAPH_LAST_EDGE;
698
699                         hashwrite_be32(f, edge_value);
700                 }
701
702                 list++;
703         }
704 }
705
706 static int commit_compare(const void *_a, const void *_b)
707 {
708         const struct object_id *a = (const struct object_id *)_a;
709         const struct object_id *b = (const struct object_id *)_b;
710         return oidcmp(a, b);
711 }
712
713 static int add_packed_commits(const struct object_id *oid,
714                               struct packed_git *pack,
715                               uint32_t pos,
716                               void *data)
717 {
718         struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
719         enum object_type type;
720         off_t offset = nth_packed_object_offset(pack, pos);
721         struct object_info oi = OBJECT_INFO_INIT;
722
723         if (ctx->progress)
724                 display_progress(ctx->progress, ++ctx->progress_done);
725
726         oi.typep = &type;
727         if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
728                 die(_("unable to get type of object %s"), oid_to_hex(oid));
729
730         if (type != OBJ_COMMIT)
731                 return 0;
732
733         ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
734         oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
735         ctx->oids.nr++;
736
737         return 0;
738 }
739
740 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
741 {
742         struct commit_list *parent;
743         for (parent = commit->parents; parent; parent = parent->next) {
744                 if (!(parent->item->object.flags & UNINTERESTING)) {
745                         ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
746                         oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
747                         ctx->oids.nr++;
748                         parent->item->object.flags |= UNINTERESTING;
749                 }
750         }
751 }
752
753 static void close_reachable(struct write_commit_graph_context *ctx)
754 {
755         int i;
756         struct commit *commit;
757
758         if (ctx->report_progress)
759                 ctx->progress = start_delayed_progress(
760                                         _("Loading known commits in commit graph"),
761                                         ctx->oids.nr);
762         for (i = 0; i < ctx->oids.nr; i++) {
763                 display_progress(ctx->progress, i + 1);
764                 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
765                 if (commit)
766                         commit->object.flags |= UNINTERESTING;
767         }
768         stop_progress(&ctx->progress);
769
770         /*
771          * As this loop runs, ctx->oids.nr may grow, but not more
772          * than the number of missing commits in the reachable
773          * closure.
774          */
775         if (ctx->report_progress)
776                 ctx->progress = start_delayed_progress(
777                                         _("Expanding reachable commits in commit graph"),
778                                         ctx->oids.nr);
779         for (i = 0; i < ctx->oids.nr; i++) {
780                 display_progress(ctx->progress, i + 1);
781                 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
782
783                 if (commit && !parse_commit_no_graph(commit))
784                         add_missing_parents(ctx, commit);
785         }
786         stop_progress(&ctx->progress);
787
788         if (ctx->report_progress)
789                 ctx->progress = start_delayed_progress(
790                                         _("Clearing commit marks in commit graph"),
791                                         ctx->oids.nr);
792         for (i = 0; i < ctx->oids.nr; i++) {
793                 display_progress(ctx->progress, i + 1);
794                 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
795
796                 if (commit)
797                         commit->object.flags &= ~UNINTERESTING;
798         }
799         stop_progress(&ctx->progress);
800 }
801
802 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
803 {
804         int i;
805         struct commit_list *list = NULL;
806
807         if (ctx->report_progress)
808                 ctx->progress = start_progress(
809                                         _("Computing commit graph generation numbers"),
810                                         ctx->commits.nr);
811         for (i = 0; i < ctx->commits.nr; i++) {
812                 display_progress(ctx->progress, i + 1);
813                 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
814                     ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
815                         continue;
816
817                 commit_list_insert(ctx->commits.list[i], &list);
818                 while (list) {
819                         struct commit *current = list->item;
820                         struct commit_list *parent;
821                         int all_parents_computed = 1;
822                         uint32_t max_generation = 0;
823
824                         for (parent = current->parents; parent; parent = parent->next) {
825                                 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
826                                     parent->item->generation == GENERATION_NUMBER_ZERO) {
827                                         all_parents_computed = 0;
828                                         commit_list_insert(parent->item, &list);
829                                         break;
830                                 } else if (parent->item->generation > max_generation) {
831                                         max_generation = parent->item->generation;
832                                 }
833                         }
834
835                         if (all_parents_computed) {
836                                 current->generation = max_generation + 1;
837                                 pop_commit(&list);
838
839                                 if (current->generation > GENERATION_NUMBER_MAX)
840                                         current->generation = GENERATION_NUMBER_MAX;
841                         }
842                 }
843         }
844         stop_progress(&ctx->progress);
845 }
846
847 static int add_ref_to_list(const char *refname,
848                            const struct object_id *oid,
849                            int flags, void *cb_data)
850 {
851         struct string_list *list = (struct string_list *)cb_data;
852
853         string_list_append(list, oid_to_hex(oid));
854         return 0;
855 }
856
857 int write_commit_graph_reachable(const char *obj_dir, unsigned int flags)
858 {
859         struct string_list list = STRING_LIST_INIT_DUP;
860         int result;
861
862         for_each_ref(add_ref_to_list, &list);
863         result = write_commit_graph(obj_dir, NULL, &list,
864                                     flags);
865
866         string_list_clear(&list, 0);
867         return result;
868 }
869
870 int write_commit_graph(const char *obj_dir,
871                        struct string_list *pack_indexes,
872                        struct string_list *commit_hex,
873                        unsigned int flags)
874 {
875         struct write_commit_graph_context *ctx;
876         struct hashfile *f;
877         uint32_t i, count_distinct = 0;
878         char *graph_name = NULL;
879         struct lock_file lk = LOCK_INIT;
880         uint32_t chunk_ids[5];
881         uint64_t chunk_offsets[5];
882         int num_chunks;
883         struct commit_list *parent;
884         const unsigned hashsz = the_hash_algo->rawsz;
885         struct strbuf progress_title = STRBUF_INIT;
886         int res = 0;
887
888         if (!commit_graph_compatible(the_repository))
889                 return 0;
890
891         ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
892         ctx->r = the_repository;
893         ctx->obj_dir = obj_dir;
894         ctx->append = flags & COMMIT_GRAPH_APPEND ? 1 : 0;
895         ctx->report_progress = flags & COMMIT_GRAPH_PROGRESS ? 1 : 0;
896
897         ctx->approx_nr_objects = approximate_object_count();
898         ctx->oids.alloc = ctx->approx_nr_objects / 32;
899
900         if (ctx->append) {
901                 prepare_commit_graph_one(ctx->r, ctx->obj_dir);
902                 if (ctx->r->objects->commit_graph)
903                         ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
904         }
905
906         if (ctx->oids.alloc < 1024)
907                 ctx->oids.alloc = 1024;
908         ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
909
910         if (ctx->append && ctx->r->objects->commit_graph) {
911                 struct commit_graph *g = ctx->r->objects->commit_graph;
912                 for (i = 0; i < g->num_commits; i++) {
913                         const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
914                         hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
915                 }
916         }
917
918         if (pack_indexes) {
919                 struct strbuf packname = STRBUF_INIT;
920                 int dirlen;
921                 strbuf_addf(&packname, "%s/pack/", obj_dir);
922                 dirlen = packname.len;
923                 if (ctx->report_progress) {
924                         strbuf_addf(&progress_title,
925                                     Q_("Finding commits for commit graph in %d pack",
926                                        "Finding commits for commit graph in %d packs",
927                                        pack_indexes->nr),
928                                     pack_indexes->nr);
929                         ctx->progress = start_delayed_progress(progress_title.buf, 0);
930                         ctx->progress_done = 0;
931                 }
932                 for (i = 0; i < pack_indexes->nr; i++) {
933                         struct packed_git *p;
934                         strbuf_setlen(&packname, dirlen);
935                         strbuf_addstr(&packname, pack_indexes->items[i].string);
936                         p = add_packed_git(packname.buf, packname.len, 1);
937                         if (!p) {
938                                 error(_("error adding pack %s"), packname.buf);
939                                 res = -1;
940                                 goto cleanup;
941                         }
942                         if (open_pack_index(p)) {
943                                 error(_("error opening index for %s"), packname.buf);
944                                 res = -1;
945                                 goto cleanup;
946                         }
947                         for_each_object_in_pack(p, add_packed_commits, ctx,
948                                                 FOR_EACH_OBJECT_PACK_ORDER);
949                         close_pack(p);
950                         free(p);
951                 }
952                 stop_progress(&ctx->progress);
953                 strbuf_reset(&progress_title);
954                 strbuf_release(&packname);
955         }
956
957         if (commit_hex) {
958                 if (ctx->report_progress) {
959                         strbuf_addf(&progress_title,
960                                     Q_("Finding commits for commit graph from %d ref",
961                                        "Finding commits for commit graph from %d refs",
962                                        commit_hex->nr),
963                                     commit_hex->nr);
964                         ctx->progress = start_delayed_progress(
965                                                 progress_title.buf,
966                                                 commit_hex->nr);
967                 }
968                 for (i = 0; i < commit_hex->nr; i++) {
969                         const char *end;
970                         struct object_id oid;
971                         struct commit *result;
972
973                         display_progress(ctx->progress, i + 1);
974                         if (commit_hex->items[i].string &&
975                             parse_oid_hex(commit_hex->items[i].string, &oid, &end))
976                                 continue;
977
978                         result = lookup_commit_reference_gently(ctx->r, &oid, 1);
979
980                         if (result) {
981                                 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
982                                 oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
983                                 ctx->oids.nr++;
984                         }
985                 }
986                 stop_progress(&ctx->progress);
987                 strbuf_reset(&progress_title);
988         }
989
990         if (!pack_indexes && !commit_hex) {
991                 if (ctx->report_progress)
992                         ctx->progress = start_delayed_progress(
993                                 _("Finding commits for commit graph among packed objects"),
994                                 ctx->approx_nr_objects);
995                 for_each_packed_object(add_packed_commits, ctx,
996                                        FOR_EACH_OBJECT_PACK_ORDER);
997                 if (ctx->progress_done < ctx->approx_nr_objects)
998                         display_progress(ctx->progress, ctx->approx_nr_objects);
999                 stop_progress(&ctx->progress);
1000         }
1001
1002         close_reachable(ctx);
1003
1004         if (ctx->report_progress)
1005                 ctx->progress = start_delayed_progress(
1006                         _("Counting distinct commits in commit graph"),
1007                         ctx->oids.nr);
1008         display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1009         QSORT(ctx->oids.list, ctx->oids.nr, commit_compare);
1010         count_distinct = 1;
1011         for (i = 1; i < ctx->oids.nr; i++) {
1012                 display_progress(ctx->progress, i + 1);
1013                 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1014                         count_distinct++;
1015         }
1016         stop_progress(&ctx->progress);
1017
1018         if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
1019                 error(_("the commit graph format cannot write %d commits"), count_distinct);
1020                 res = -1;
1021                 goto cleanup;
1022         }
1023
1024         ctx->commits.alloc = count_distinct;
1025         ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
1026
1027         ctx->num_extra_edges = 0;
1028         if (ctx->report_progress)
1029                 ctx->progress = start_delayed_progress(
1030                         _("Finding extra edges in commit graph"),
1031                         ctx->oids.nr);
1032         for (i = 0; i < ctx->oids.nr; i++) {
1033                 int num_parents = 0;
1034                 display_progress(ctx->progress, i + 1);
1035                 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1036                         continue;
1037
1038                 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1039                 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1040
1041                 for (parent = ctx->commits.list[ctx->commits.nr]->parents;
1042                      parent; parent = parent->next)
1043                         num_parents++;
1044
1045                 if (num_parents > 2)
1046                         ctx->num_extra_edges += num_parents - 1;
1047
1048                 ctx->commits.nr++;
1049         }
1050         stop_progress(&ctx->progress);
1051
1052         if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
1053                 error(_("too many commits to write graph"));
1054                 res = -1;
1055                 goto cleanup;
1056         }
1057
1058         compute_generation_numbers(ctx);
1059
1060         num_chunks = ctx->num_extra_edges ? 4 : 3;
1061
1062         ctx->graph_name = get_commit_graph_filename(ctx->obj_dir);
1063         if (safe_create_leading_directories(ctx->graph_name)) {
1064                 UNLEAK(ctx->graph_name);
1065                 error(_("unable to create leading directories of %s"),
1066                         ctx->graph_name);
1067                 res = -1;
1068                 goto cleanup;
1069         }
1070
1071         hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
1072         f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1073
1074         hashwrite_be32(f, GRAPH_SIGNATURE);
1075
1076         hashwrite_u8(f, GRAPH_VERSION);
1077         hashwrite_u8(f, oid_version());
1078         hashwrite_u8(f, num_chunks);
1079         hashwrite_u8(f, 0); /* unused padding byte */
1080
1081         chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1082         chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1083         chunk_ids[2] = GRAPH_CHUNKID_DATA;
1084         if (ctx->num_extra_edges)
1085                 chunk_ids[3] = GRAPH_CHUNKID_EXTRAEDGES;
1086         else
1087                 chunk_ids[3] = 0;
1088         chunk_ids[4] = 0;
1089
1090         chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1091         chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1092         chunk_offsets[2] = chunk_offsets[1] + hashsz * ctx->commits.nr;
1093         chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * ctx->commits.nr;
1094         chunk_offsets[4] = chunk_offsets[3] + 4 * ctx->num_extra_edges;
1095
1096         for (i = 0; i <= num_chunks; i++) {
1097                 uint32_t chunk_write[3];
1098
1099                 chunk_write[0] = htonl(chunk_ids[i]);
1100                 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1101                 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1102                 hashwrite(f, chunk_write, 12);
1103         }
1104
1105         if (ctx->report_progress) {
1106                 strbuf_addf(&progress_title,
1107                             Q_("Writing out commit graph in %d pass",
1108                                "Writing out commit graph in %d passes",
1109                                num_chunks),
1110                             num_chunks);
1111                 ctx->progress = start_delayed_progress(
1112                         progress_title.buf,
1113                         num_chunks * ctx->commits.nr);
1114         }
1115         write_graph_chunk_fanout(f, ctx);
1116         write_graph_chunk_oids(f, hashsz, ctx);
1117         write_graph_chunk_data(f, hashsz, ctx);
1118         if (ctx->num_extra_edges)
1119                 write_graph_chunk_extra_edges(f, ctx);
1120         stop_progress(&ctx->progress);
1121         strbuf_release(&progress_title);
1122
1123         close_commit_graph(ctx->r);
1124         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1125         commit_lock_file(&lk);
1126
1127 cleanup:
1128         free(graph_name);
1129         free(ctx->commits.list);
1130         free(ctx->oids.list);
1131         free(ctx);
1132
1133         return res;
1134 }
1135
1136 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1137 static int verify_commit_graph_error;
1138
1139 static void graph_report(const char *fmt, ...)
1140 {
1141         va_list ap;
1142
1143         verify_commit_graph_error = 1;
1144         va_start(ap, fmt);
1145         vfprintf(stderr, fmt, ap);
1146         fprintf(stderr, "\n");
1147         va_end(ap);
1148 }
1149
1150 #define GENERATION_ZERO_EXISTS 1
1151 #define GENERATION_NUMBER_EXISTS 2
1152
1153 int verify_commit_graph(struct repository *r, struct commit_graph *g)
1154 {
1155         uint32_t i, cur_fanout_pos = 0;
1156         struct object_id prev_oid, cur_oid, checksum;
1157         int generation_zero = 0;
1158         struct hashfile *f;
1159         int devnull;
1160         struct progress *progress = NULL;
1161
1162         if (!g) {
1163                 graph_report("no commit-graph file loaded");
1164                 return 1;
1165         }
1166
1167         verify_commit_graph_error = verify_commit_graph_lite(g);
1168         if (verify_commit_graph_error)
1169                 return verify_commit_graph_error;
1170
1171         devnull = open("/dev/null", O_WRONLY);
1172         f = hashfd(devnull, NULL);
1173         hashwrite(f, g->data, g->data_len - g->hash_len);
1174         finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1175         if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1176                 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1177                 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1178         }
1179
1180         for (i = 0; i < g->num_commits; i++) {
1181                 struct commit *graph_commit;
1182
1183                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1184
1185                 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1186                         graph_report(_("commit-graph has incorrect OID order: %s then %s"),
1187                                      oid_to_hex(&prev_oid),
1188                                      oid_to_hex(&cur_oid));
1189
1190                 oidcpy(&prev_oid, &cur_oid);
1191
1192                 while (cur_oid.hash[0] > cur_fanout_pos) {
1193                         uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1194
1195                         if (i != fanout_value)
1196                                 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1197                                              cur_fanout_pos, fanout_value, i);
1198                         cur_fanout_pos++;
1199                 }
1200
1201                 graph_commit = lookup_commit(r, &cur_oid);
1202                 if (!parse_commit_in_graph_one(r, g, graph_commit))
1203                         graph_report(_("failed to parse commit %s from commit-graph"),
1204                                      oid_to_hex(&cur_oid));
1205         }
1206
1207         while (cur_fanout_pos < 256) {
1208                 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1209
1210                 if (g->num_commits != fanout_value)
1211                         graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1212                                      cur_fanout_pos, fanout_value, i);
1213
1214                 cur_fanout_pos++;
1215         }
1216
1217         if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1218                 return verify_commit_graph_error;
1219
1220         progress = start_progress(_("Verifying commits in commit graph"),
1221                                   g->num_commits);
1222         for (i = 0; i < g->num_commits; i++) {
1223                 struct commit *graph_commit, *odb_commit;
1224                 struct commit_list *graph_parents, *odb_parents;
1225                 uint32_t max_generation = 0;
1226
1227                 display_progress(progress, i + 1);
1228                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1229
1230                 graph_commit = lookup_commit(r, &cur_oid);
1231                 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1232                 if (parse_commit_internal(odb_commit, 0, 0)) {
1233                         graph_report(_("failed to parse commit %s from object database for commit-graph"),
1234                                      oid_to_hex(&cur_oid));
1235                         continue;
1236                 }
1237
1238                 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
1239                            get_commit_tree_oid(odb_commit)))
1240                         graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
1241                                      oid_to_hex(&cur_oid),
1242                                      oid_to_hex(get_commit_tree_oid(graph_commit)),
1243                                      oid_to_hex(get_commit_tree_oid(odb_commit)));
1244
1245                 graph_parents = graph_commit->parents;
1246                 odb_parents = odb_commit->parents;
1247
1248                 while (graph_parents) {
1249                         if (odb_parents == NULL) {
1250                                 graph_report(_("commit-graph parent list for commit %s is too long"),
1251                                              oid_to_hex(&cur_oid));
1252                                 break;
1253                         }
1254
1255                         if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1256                                 graph_report(_("commit-graph parent for %s is %s != %s"),
1257                                              oid_to_hex(&cur_oid),
1258                                              oid_to_hex(&graph_parents->item->object.oid),
1259                                              oid_to_hex(&odb_parents->item->object.oid));
1260
1261                         if (graph_parents->item->generation > max_generation)
1262                                 max_generation = graph_parents->item->generation;
1263
1264                         graph_parents = graph_parents->next;
1265                         odb_parents = odb_parents->next;
1266                 }
1267
1268                 if (odb_parents != NULL)
1269                         graph_report(_("commit-graph parent list for commit %s terminates early"),
1270                                      oid_to_hex(&cur_oid));
1271
1272                 if (!graph_commit->generation) {
1273                         if (generation_zero == GENERATION_NUMBER_EXISTS)
1274                                 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
1275                                              oid_to_hex(&cur_oid));
1276                         generation_zero = GENERATION_ZERO_EXISTS;
1277                 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1278                         graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
1279                                      oid_to_hex(&cur_oid));
1280
1281                 if (generation_zero == GENERATION_ZERO_EXISTS)
1282                         continue;
1283
1284                 /*
1285                  * If one of our parents has generation GENERATION_NUMBER_MAX, then
1286                  * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1287                  * extra logic in the following condition.
1288                  */
1289                 if (max_generation == GENERATION_NUMBER_MAX)
1290                         max_generation--;
1291
1292                 if (graph_commit->generation != max_generation + 1)
1293                         graph_report(_("commit-graph generation for commit %s is %u != %u"),
1294                                      oid_to_hex(&cur_oid),
1295                                      graph_commit->generation,
1296                                      max_generation + 1);
1297
1298                 if (graph_commit->date != odb_commit->date)
1299                         graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
1300                                      oid_to_hex(&cur_oid),
1301                                      graph_commit->date,
1302                                      odb_commit->date);
1303         }
1304         stop_progress(&progress);
1305
1306         return verify_commit_graph_error;
1307 }
1308
1309 void free_commit_graph(struct commit_graph *g)
1310 {
1311         if (!g)
1312                 return;
1313         if (g->graph_fd >= 0) {
1314                 munmap((void *)g->data, g->data_len);
1315                 g->data = NULL;
1316                 close(g->graph_fd);
1317         }
1318         free(g);
1319 }