t6030: make test work with SHA-256
[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                 free(graph);
272                 return NULL;
273         }
274
275         return graph;
276 }
277
278 static struct commit_graph *load_commit_graph_one(const char *graph_file)
279 {
280
281         struct stat st;
282         int fd;
283         int open_ok = open_commit_graph(graph_file, &fd, &st);
284
285         if (!open_ok)
286                 return NULL;
287
288         return load_commit_graph_one_fd_st(fd, &st);
289 }
290
291 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
292 {
293         char *graph_name;
294
295         if (r->objects->commit_graph)
296                 return;
297
298         graph_name = get_commit_graph_filename(obj_dir);
299         r->objects->commit_graph =
300                 load_commit_graph_one(graph_name);
301
302         FREE_AND_NULL(graph_name);
303 }
304
305 /*
306  * Return 1 if commit_graph is non-NULL, and 0 otherwise.
307  *
308  * On the first invocation, this function attemps to load the commit
309  * graph if the_repository is configured to have one.
310  */
311 static int prepare_commit_graph(struct repository *r)
312 {
313         struct object_directory *odb;
314         int config_value;
315
316         if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
317                 die("dying as requested by the '%s' variable on commit-graph load!",
318                     GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD);
319
320         if (r->objects->commit_graph_attempted)
321                 return !!r->objects->commit_graph;
322         r->objects->commit_graph_attempted = 1;
323
324         if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
325             (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
326             !config_value))
327                 /*
328                  * This repository is not configured to use commit graphs, so
329                  * do not load one. (But report commit_graph_attempted anyway
330                  * so that commit graph loading is not attempted again for this
331                  * repository.)
332                  */
333                 return 0;
334
335         if (!commit_graph_compatible(r))
336                 return 0;
337
338         prepare_alt_odb(r);
339         for (odb = r->objects->odb;
340              !r->objects->commit_graph && odb;
341              odb = odb->next)
342                 prepare_commit_graph_one(r, odb->path);
343         return !!r->objects->commit_graph;
344 }
345
346 int generation_numbers_enabled(struct repository *r)
347 {
348         uint32_t first_generation;
349         struct commit_graph *g;
350         if (!prepare_commit_graph(r))
351                return 0;
352
353         g = r->objects->commit_graph;
354
355         if (!g->num_commits)
356                 return 0;
357
358         first_generation = get_be32(g->chunk_commit_data +
359                                     g->hash_len + 8) >> 2;
360
361         return !!first_generation;
362 }
363
364 void close_commit_graph(struct repository *r)
365 {
366         free_commit_graph(r->objects->commit_graph);
367         r->objects->commit_graph = NULL;
368 }
369
370 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
371 {
372         return bsearch_hash(oid->hash, g->chunk_oid_fanout,
373                             g->chunk_oid_lookup, g->hash_len, pos);
374 }
375
376 static struct commit_list **insert_parent_or_die(struct repository *r,
377                                                  struct commit_graph *g,
378                                                  uint64_t pos,
379                                                  struct commit_list **pptr)
380 {
381         struct commit *c;
382         struct object_id oid;
383
384         if (pos >= g->num_commits)
385                 die("invalid parent position %"PRIu64, pos);
386
387         hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
388         c = lookup_commit(r, &oid);
389         if (!c)
390                 die(_("could not find commit %s"), oid_to_hex(&oid));
391         c->graph_pos = pos;
392         return &commit_list_insert(c, pptr)->next;
393 }
394
395 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
396 {
397         const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
398         item->graph_pos = pos;
399         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
400 }
401
402 static inline void set_commit_tree(struct commit *c, struct tree *t)
403 {
404         c->maybe_tree = t;
405 }
406
407 static int fill_commit_in_graph(struct repository *r,
408                                 struct commit *item,
409                                 struct commit_graph *g, uint32_t pos)
410 {
411         uint32_t edge_value;
412         uint32_t *parent_data_ptr;
413         uint64_t date_low, date_high;
414         struct commit_list **pptr;
415         const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
416
417         item->object.parsed = 1;
418         item->graph_pos = pos;
419
420         set_commit_tree(item, NULL);
421
422         date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
423         date_low = get_be32(commit_data + g->hash_len + 12);
424         item->date = (timestamp_t)((date_high << 32) | date_low);
425
426         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
427
428         pptr = &item->parents;
429
430         edge_value = get_be32(commit_data + g->hash_len);
431         if (edge_value == GRAPH_PARENT_NONE)
432                 return 1;
433         pptr = insert_parent_or_die(r, g, edge_value, pptr);
434
435         edge_value = get_be32(commit_data + g->hash_len + 4);
436         if (edge_value == GRAPH_PARENT_NONE)
437                 return 1;
438         if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
439                 pptr = insert_parent_or_die(r, g, edge_value, pptr);
440                 return 1;
441         }
442
443         parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
444                           4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
445         do {
446                 edge_value = get_be32(parent_data_ptr);
447                 pptr = insert_parent_or_die(r, g,
448                                             edge_value & GRAPH_EDGE_LAST_MASK,
449                                             pptr);
450                 parent_data_ptr++;
451         } while (!(edge_value & GRAPH_LAST_EDGE));
452
453         return 1;
454 }
455
456 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
457 {
458         if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
459                 *pos = item->graph_pos;
460                 return 1;
461         } else {
462                 return bsearch_graph(g, &(item->object.oid), pos);
463         }
464 }
465
466 static int parse_commit_in_graph_one(struct repository *r,
467                                      struct commit_graph *g,
468                                      struct commit *item)
469 {
470         uint32_t pos;
471
472         if (item->object.parsed)
473                 return 1;
474
475         if (find_commit_in_graph(item, g, &pos))
476                 return fill_commit_in_graph(r, item, g, pos);
477
478         return 0;
479 }
480
481 int parse_commit_in_graph(struct repository *r, struct commit *item)
482 {
483         if (!prepare_commit_graph(r))
484                 return 0;
485         return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
486 }
487
488 void load_commit_graph_info(struct repository *r, struct commit *item)
489 {
490         uint32_t pos;
491         if (!prepare_commit_graph(r))
492                 return;
493         if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
494                 fill_commit_graph_info(item, r->objects->commit_graph, pos);
495 }
496
497 static struct tree *load_tree_for_commit(struct repository *r,
498                                          struct commit_graph *g,
499                                          struct commit *c)
500 {
501         struct object_id oid;
502         const unsigned char *commit_data = g->chunk_commit_data +
503                                            GRAPH_DATA_WIDTH * (c->graph_pos);
504
505         hashcpy(oid.hash, commit_data);
506         set_commit_tree(c, lookup_tree(r, &oid));
507
508         return c->maybe_tree;
509 }
510
511 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
512                                                  struct commit_graph *g,
513                                                  const struct commit *c)
514 {
515         if (c->maybe_tree)
516                 return c->maybe_tree;
517         if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
518                 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
519
520         return load_tree_for_commit(r, g, (struct commit *)c);
521 }
522
523 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
524 {
525         return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
526 }
527
528 static void write_graph_chunk_fanout(struct hashfile *f,
529                                      struct commit **commits,
530                                      int nr_commits,
531                                      struct progress *progress,
532                                      uint64_t *progress_cnt)
533 {
534         int i, count = 0;
535         struct commit **list = commits;
536
537         /*
538          * Write the first-level table (the list is sorted,
539          * but we use a 256-entry lookup to be able to avoid
540          * having to do eight extra binary search iterations).
541          */
542         for (i = 0; i < 256; i++) {
543                 while (count < nr_commits) {
544                         if ((*list)->object.oid.hash[0] != i)
545                                 break;
546                         display_progress(progress, ++*progress_cnt);
547                         count++;
548                         list++;
549                 }
550
551                 hashwrite_be32(f, count);
552         }
553 }
554
555 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
556                                    struct commit **commits, int nr_commits,
557                                    struct progress *progress,
558                                    uint64_t *progress_cnt)
559 {
560         struct commit **list = commits;
561         int count;
562         for (count = 0; count < nr_commits; count++, list++) {
563                 display_progress(progress, ++*progress_cnt);
564                 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
565         }
566 }
567
568 static const unsigned char *commit_to_sha1(size_t index, void *table)
569 {
570         struct commit **commits = table;
571         return commits[index]->object.oid.hash;
572 }
573
574 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
575                                    struct commit **commits, int nr_commits,
576                                    struct progress *progress,
577                                    uint64_t *progress_cnt)
578 {
579         struct commit **list = commits;
580         struct commit **last = commits + nr_commits;
581         uint32_t num_extra_edges = 0;
582
583         while (list < last) {
584                 struct commit_list *parent;
585                 int edge_value;
586                 uint32_t packedDate[2];
587                 display_progress(progress, ++*progress_cnt);
588
589                 parse_commit_no_graph(*list);
590                 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
591
592                 parent = (*list)->parents;
593
594                 if (!parent)
595                         edge_value = GRAPH_PARENT_NONE;
596                 else {
597                         edge_value = sha1_pos(parent->item->object.oid.hash,
598                                               commits,
599                                               nr_commits,
600                                               commit_to_sha1);
601
602                         if (edge_value < 0)
603                                 BUG("missing parent %s for commit %s",
604                                     oid_to_hex(&parent->item->object.oid),
605                                     oid_to_hex(&(*list)->object.oid));
606                 }
607
608                 hashwrite_be32(f, edge_value);
609
610                 if (parent)
611                         parent = parent->next;
612
613                 if (!parent)
614                         edge_value = GRAPH_PARENT_NONE;
615                 else if (parent->next)
616                         edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
617                 else {
618                         edge_value = sha1_pos(parent->item->object.oid.hash,
619                                               commits,
620                                               nr_commits,
621                                               commit_to_sha1);
622                         if (edge_value < 0)
623                                 BUG("missing parent %s for commit %s",
624                                     oid_to_hex(&parent->item->object.oid),
625                                     oid_to_hex(&(*list)->object.oid));
626                 }
627
628                 hashwrite_be32(f, edge_value);
629
630                 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
631                         do {
632                                 num_extra_edges++;
633                                 parent = parent->next;
634                         } while (parent);
635                 }
636
637                 if (sizeof((*list)->date) > 4)
638                         packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
639                 else
640                         packedDate[0] = 0;
641
642                 packedDate[0] |= htonl((*list)->generation << 2);
643
644                 packedDate[1] = htonl((*list)->date);
645                 hashwrite(f, packedDate, 8);
646
647                 list++;
648         }
649 }
650
651 static void write_graph_chunk_extra_edges(struct hashfile *f,
652                                           struct commit **commits,
653                                           int nr_commits,
654                                           struct progress *progress,
655                                           uint64_t *progress_cnt)
656 {
657         struct commit **list = commits;
658         struct commit **last = commits + nr_commits;
659         struct commit_list *parent;
660
661         while (list < last) {
662                 int num_parents = 0;
663
664                 display_progress(progress, ++*progress_cnt);
665
666                 for (parent = (*list)->parents; num_parents < 3 && parent;
667                      parent = parent->next)
668                         num_parents++;
669
670                 if (num_parents <= 2) {
671                         list++;
672                         continue;
673                 }
674
675                 /* Since num_parents > 2, this initializer is safe. */
676                 for (parent = (*list)->parents->next; parent; parent = parent->next) {
677                         int edge_value = sha1_pos(parent->item->object.oid.hash,
678                                                   commits,
679                                                   nr_commits,
680                                                   commit_to_sha1);
681
682                         if (edge_value < 0)
683                                 BUG("missing parent %s for commit %s",
684                                     oid_to_hex(&parent->item->object.oid),
685                                     oid_to_hex(&(*list)->object.oid));
686                         else if (!parent->next)
687                                 edge_value |= GRAPH_LAST_EDGE;
688
689                         hashwrite_be32(f, edge_value);
690                 }
691
692                 list++;
693         }
694 }
695
696 static int commit_compare(const void *_a, const void *_b)
697 {
698         const struct object_id *a = (const struct object_id *)_a;
699         const struct object_id *b = (const struct object_id *)_b;
700         return oidcmp(a, b);
701 }
702
703 struct packed_commit_list {
704         struct commit **list;
705         int nr;
706         int alloc;
707 };
708
709 struct packed_oid_list {
710         struct object_id *list;
711         int nr;
712         int alloc;
713         struct progress *progress;
714         int progress_done;
715 };
716
717 static int add_packed_commits(const struct object_id *oid,
718                               struct packed_git *pack,
719                               uint32_t pos,
720                               void *data)
721 {
722         struct packed_oid_list *list = (struct packed_oid_list*)data;
723         enum object_type type;
724         off_t offset = nth_packed_object_offset(pack, pos);
725         struct object_info oi = OBJECT_INFO_INIT;
726
727         if (list->progress)
728                 display_progress(list->progress, ++list->progress_done);
729
730         oi.typep = &type;
731         if (packed_object_info(the_repository, pack, offset, &oi) < 0)
732                 die(_("unable to get type of object %s"), oid_to_hex(oid));
733
734         if (type != OBJ_COMMIT)
735                 return 0;
736
737         ALLOC_GROW(list->list, list->nr + 1, list->alloc);
738         oidcpy(&(list->list[list->nr]), oid);
739         list->nr++;
740
741         return 0;
742 }
743
744 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
745 {
746         struct commit_list *parent;
747         for (parent = commit->parents; parent; parent = parent->next) {
748                 if (!(parent->item->object.flags & UNINTERESTING)) {
749                         ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
750                         oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
751                         oids->nr++;
752                         parent->item->object.flags |= UNINTERESTING;
753                 }
754         }
755 }
756
757 static void close_reachable(struct packed_oid_list *oids, int report_progress)
758 {
759         int i;
760         struct commit *commit;
761         struct progress *progress = NULL;
762
763         if (report_progress)
764                 progress = start_delayed_progress(
765                         _("Loading known commits in commit graph"), oids->nr);
766         for (i = 0; i < oids->nr; i++) {
767                 display_progress(progress, i + 1);
768                 commit = lookup_commit(the_repository, &oids->list[i]);
769                 if (commit)
770                         commit->object.flags |= UNINTERESTING;
771         }
772         stop_progress(&progress);
773
774         /*
775          * As this loop runs, oids->nr may grow, but not more
776          * than the number of missing commits in the reachable
777          * closure.
778          */
779         if (report_progress)
780                 progress = start_delayed_progress(
781                         _("Expanding reachable commits in commit graph"), oids->nr);
782         for (i = 0; i < oids->nr; i++) {
783                 display_progress(progress, i + 1);
784                 commit = lookup_commit(the_repository, &oids->list[i]);
785
786                 if (commit && !parse_commit_no_graph(commit))
787                         add_missing_parents(oids, commit);
788         }
789         stop_progress(&progress);
790
791         if (report_progress)
792                 progress = start_delayed_progress(
793                         _("Clearing commit marks in commit graph"), oids->nr);
794         for (i = 0; i < oids->nr; i++) {
795                 display_progress(progress, i + 1);
796                 commit = lookup_commit(the_repository, &oids->list[i]);
797
798                 if (commit)
799                         commit->object.flags &= ~UNINTERESTING;
800         }
801         stop_progress(&progress);
802 }
803
804 static void compute_generation_numbers(struct packed_commit_list* commits,
805                                        int report_progress)
806 {
807         int i;
808         struct commit_list *list = NULL;
809         struct progress *progress = NULL;
810
811         if (report_progress)
812                 progress = start_progress(
813                         _("Computing commit graph generation numbers"),
814                         commits->nr);
815         for (i = 0; i < commits->nr; i++) {
816                 display_progress(progress, i + 1);
817                 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
818                     commits->list[i]->generation != GENERATION_NUMBER_ZERO)
819                         continue;
820
821                 commit_list_insert(commits->list[i], &list);
822                 while (list) {
823                         struct commit *current = list->item;
824                         struct commit_list *parent;
825                         int all_parents_computed = 1;
826                         uint32_t max_generation = 0;
827
828                         for (parent = current->parents; parent; parent = parent->next) {
829                                 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
830                                     parent->item->generation == GENERATION_NUMBER_ZERO) {
831                                         all_parents_computed = 0;
832                                         commit_list_insert(parent->item, &list);
833                                         break;
834                                 } else if (parent->item->generation > max_generation) {
835                                         max_generation = parent->item->generation;
836                                 }
837                         }
838
839                         if (all_parents_computed) {
840                                 current->generation = max_generation + 1;
841                                 pop_commit(&list);
842
843                                 if (current->generation > GENERATION_NUMBER_MAX)
844                                         current->generation = GENERATION_NUMBER_MAX;
845                         }
846                 }
847         }
848         stop_progress(&progress);
849 }
850
851 static int add_ref_to_list(const char *refname,
852                            const struct object_id *oid,
853                            int flags, void *cb_data)
854 {
855         struct string_list *list = (struct string_list *)cb_data;
856
857         string_list_append(list, oid_to_hex(oid));
858         return 0;
859 }
860
861 void write_commit_graph_reachable(const char *obj_dir, int append,
862                                   int report_progress)
863 {
864         struct string_list list = STRING_LIST_INIT_DUP;
865
866         for_each_ref(add_ref_to_list, &list);
867         write_commit_graph(obj_dir, NULL, &list, append, report_progress);
868
869         string_list_clear(&list, 0);
870 }
871
872 void write_commit_graph(const char *obj_dir,
873                         struct string_list *pack_indexes,
874                         struct string_list *commit_hex,
875                         int append, int report_progress)
876 {
877         struct packed_oid_list oids;
878         struct packed_commit_list commits;
879         struct hashfile *f;
880         uint32_t i, count_distinct = 0;
881         char *graph_name;
882         struct lock_file lk = LOCK_INIT;
883         uint32_t chunk_ids[5];
884         uint64_t chunk_offsets[5];
885         int num_chunks;
886         int num_extra_edges;
887         struct commit_list *parent;
888         struct progress *progress = NULL;
889         const unsigned hashsz = the_hash_algo->rawsz;
890         uint64_t progress_cnt = 0;
891         struct strbuf progress_title = STRBUF_INIT;
892         unsigned long approx_nr_objects;
893
894         if (!commit_graph_compatible(the_repository))
895                 return;
896
897         oids.nr = 0;
898         approx_nr_objects = approximate_object_count();
899         oids.alloc = approx_nr_objects / 32;
900         oids.progress = NULL;
901         oids.progress_done = 0;
902
903         if (append) {
904                 prepare_commit_graph_one(the_repository, obj_dir);
905                 if (the_repository->objects->commit_graph)
906                         oids.alloc += the_repository->objects->commit_graph->num_commits;
907         }
908
909         if (oids.alloc < 1024)
910                 oids.alloc = 1024;
911         ALLOC_ARRAY(oids.list, oids.alloc);
912
913         if (append && the_repository->objects->commit_graph) {
914                 struct commit_graph *commit_graph =
915                         the_repository->objects->commit_graph;
916                 for (i = 0; i < commit_graph->num_commits; i++) {
917                         const unsigned char *hash = commit_graph->chunk_oid_lookup +
918                                 commit_graph->hash_len * i;
919                         hashcpy(oids.list[oids.nr++].hash, hash);
920                 }
921         }
922
923         if (pack_indexes) {
924                 struct strbuf packname = STRBUF_INIT;
925                 int dirlen;
926                 strbuf_addf(&packname, "%s/pack/", obj_dir);
927                 dirlen = packname.len;
928                 if (report_progress) {
929                         strbuf_addf(&progress_title,
930                                     Q_("Finding commits for commit graph in %d pack",
931                                        "Finding commits for commit graph in %d packs",
932                                        pack_indexes->nr),
933                                     pack_indexes->nr);
934                         oids.progress = start_delayed_progress(progress_title.buf, 0);
935                         oids.progress_done = 0;
936                 }
937                 for (i = 0; i < pack_indexes->nr; i++) {
938                         struct packed_git *p;
939                         strbuf_setlen(&packname, dirlen);
940                         strbuf_addstr(&packname, pack_indexes->items[i].string);
941                         p = add_packed_git(packname.buf, packname.len, 1);
942                         if (!p)
943                                 die(_("error adding pack %s"), packname.buf);
944                         if (open_pack_index(p))
945                                 die(_("error opening index for %s"), packname.buf);
946                         for_each_object_in_pack(p, add_packed_commits, &oids,
947                                                 FOR_EACH_OBJECT_PACK_ORDER);
948                         close_pack(p);
949                         free(p);
950                 }
951                 stop_progress(&oids.progress);
952                 strbuf_reset(&progress_title);
953                 strbuf_release(&packname);
954         }
955
956         if (commit_hex) {
957                 if (report_progress) {
958                         strbuf_addf(&progress_title,
959                                     Q_("Finding commits for commit graph from %d ref",
960                                        "Finding commits for commit graph from %d refs",
961                                        commit_hex->nr),
962                                     commit_hex->nr);
963                         progress = start_delayed_progress(progress_title.buf,
964                                                           commit_hex->nr);
965                 }
966                 for (i = 0; i < commit_hex->nr; i++) {
967                         const char *end;
968                         struct object_id oid;
969                         struct commit *result;
970
971                         display_progress(progress, i + 1);
972                         if (commit_hex->items[i].string &&
973                             parse_oid_hex(commit_hex->items[i].string, &oid, &end))
974                                 continue;
975
976                         result = lookup_commit_reference_gently(the_repository, &oid, 1);
977
978                         if (result) {
979                                 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
980                                 oidcpy(&oids.list[oids.nr], &(result->object.oid));
981                                 oids.nr++;
982                         }
983                 }
984                 stop_progress(&progress);
985                 strbuf_reset(&progress_title);
986         }
987
988         if (!pack_indexes && !commit_hex) {
989                 if (report_progress)
990                         oids.progress = start_delayed_progress(
991                                 _("Finding commits for commit graph among packed objects"),
992                                 approx_nr_objects);
993                 for_each_packed_object(add_packed_commits, &oids,
994                                        FOR_EACH_OBJECT_PACK_ORDER);
995                 if (oids.progress_done < approx_nr_objects)
996                         display_progress(oids.progress, approx_nr_objects);
997                 stop_progress(&oids.progress);
998         }
999
1000         close_reachable(&oids, report_progress);
1001
1002         if (report_progress)
1003                 progress = start_delayed_progress(
1004                         _("Counting distinct commits in commit graph"),
1005                         oids.nr);
1006         display_progress(progress, 0); /* TODO: Measure QSORT() progress */
1007         QSORT(oids.list, oids.nr, commit_compare);
1008         count_distinct = 1;
1009         for (i = 1; i < oids.nr; i++) {
1010                 display_progress(progress, i + 1);
1011                 if (!oideq(&oids.list[i - 1], &oids.list[i]))
1012                         count_distinct++;
1013         }
1014         stop_progress(&progress);
1015
1016         if (count_distinct >= GRAPH_EDGE_LAST_MASK)
1017                 die(_("the commit graph format cannot write %d commits"), count_distinct);
1018
1019         commits.nr = 0;
1020         commits.alloc = count_distinct;
1021         ALLOC_ARRAY(commits.list, commits.alloc);
1022
1023         num_extra_edges = 0;
1024         if (report_progress)
1025                 progress = start_delayed_progress(
1026                         _("Finding extra edges in commit graph"),
1027                         oids.nr);
1028         for (i = 0; i < oids.nr; i++) {
1029                 int num_parents = 0;
1030                 display_progress(progress, i + 1);
1031                 if (i > 0 && oideq(&oids.list[i - 1], &oids.list[i]))
1032                         continue;
1033
1034                 commits.list[commits.nr] = lookup_commit(the_repository, &oids.list[i]);
1035                 parse_commit_no_graph(commits.list[commits.nr]);
1036
1037                 for (parent = commits.list[commits.nr]->parents;
1038                      parent; parent = parent->next)
1039                         num_parents++;
1040
1041                 if (num_parents > 2)
1042                         num_extra_edges += num_parents - 1;
1043
1044                 commits.nr++;
1045         }
1046         num_chunks = num_extra_edges ? 4 : 3;
1047         stop_progress(&progress);
1048
1049         if (commits.nr >= GRAPH_EDGE_LAST_MASK)
1050                 die(_("too many commits to write graph"));
1051
1052         compute_generation_numbers(&commits, report_progress);
1053
1054         graph_name = get_commit_graph_filename(obj_dir);
1055         if (safe_create_leading_directories(graph_name)) {
1056                 UNLEAK(graph_name);
1057                 die_errno(_("unable to create leading directories of %s"),
1058                           graph_name);
1059         }
1060
1061         hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
1062         f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1063
1064         hashwrite_be32(f, GRAPH_SIGNATURE);
1065
1066         hashwrite_u8(f, GRAPH_VERSION);
1067         hashwrite_u8(f, oid_version());
1068         hashwrite_u8(f, num_chunks);
1069         hashwrite_u8(f, 0); /* unused padding byte */
1070
1071         chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1072         chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1073         chunk_ids[2] = GRAPH_CHUNKID_DATA;
1074         if (num_extra_edges)
1075                 chunk_ids[3] = GRAPH_CHUNKID_EXTRAEDGES;
1076         else
1077                 chunk_ids[3] = 0;
1078         chunk_ids[4] = 0;
1079
1080         chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1081         chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1082         chunk_offsets[2] = chunk_offsets[1] + hashsz * commits.nr;
1083         chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * commits.nr;
1084         chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
1085
1086         for (i = 0; i <= num_chunks; i++) {
1087                 uint32_t chunk_write[3];
1088
1089                 chunk_write[0] = htonl(chunk_ids[i]);
1090                 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1091                 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1092                 hashwrite(f, chunk_write, 12);
1093         }
1094
1095         if (report_progress) {
1096                 strbuf_addf(&progress_title,
1097                             Q_("Writing out commit graph in %d pass",
1098                                "Writing out commit graph in %d passes",
1099                                num_chunks),
1100                             num_chunks);
1101                 progress = start_delayed_progress(
1102                         progress_title.buf,
1103                         num_chunks * commits.nr);
1104         }
1105         write_graph_chunk_fanout(f, commits.list, commits.nr, progress, &progress_cnt);
1106         write_graph_chunk_oids(f, hashsz, commits.list, commits.nr, progress, &progress_cnt);
1107         write_graph_chunk_data(f, hashsz, commits.list, commits.nr, progress, &progress_cnt);
1108         if (num_extra_edges)
1109                 write_graph_chunk_extra_edges(f, commits.list, commits.nr, progress, &progress_cnt);
1110         stop_progress(&progress);
1111         strbuf_release(&progress_title);
1112
1113         close_commit_graph(the_repository);
1114         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1115         commit_lock_file(&lk);
1116
1117         free(graph_name);
1118         free(commits.list);
1119         free(oids.list);
1120 }
1121
1122 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1123 static int verify_commit_graph_error;
1124
1125 static void graph_report(const char *fmt, ...)
1126 {
1127         va_list ap;
1128
1129         verify_commit_graph_error = 1;
1130         va_start(ap, fmt);
1131         vfprintf(stderr, fmt, ap);
1132         fprintf(stderr, "\n");
1133         va_end(ap);
1134 }
1135
1136 #define GENERATION_ZERO_EXISTS 1
1137 #define GENERATION_NUMBER_EXISTS 2
1138
1139 int verify_commit_graph(struct repository *r, struct commit_graph *g)
1140 {
1141         uint32_t i, cur_fanout_pos = 0;
1142         struct object_id prev_oid, cur_oid, checksum;
1143         int generation_zero = 0;
1144         struct hashfile *f;
1145         int devnull;
1146         struct progress *progress = NULL;
1147
1148         if (!g) {
1149                 graph_report("no commit-graph file loaded");
1150                 return 1;
1151         }
1152
1153         verify_commit_graph_error = verify_commit_graph_lite(g);
1154         if (verify_commit_graph_error)
1155                 return verify_commit_graph_error;
1156
1157         devnull = open("/dev/null", O_WRONLY);
1158         f = hashfd(devnull, NULL);
1159         hashwrite(f, g->data, g->data_len - g->hash_len);
1160         finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1161         if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1162                 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1163                 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1164         }
1165
1166         for (i = 0; i < g->num_commits; i++) {
1167                 struct commit *graph_commit;
1168
1169                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1170
1171                 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1172                         graph_report(_("commit-graph has incorrect OID order: %s then %s"),
1173                                      oid_to_hex(&prev_oid),
1174                                      oid_to_hex(&cur_oid));
1175
1176                 oidcpy(&prev_oid, &cur_oid);
1177
1178                 while (cur_oid.hash[0] > cur_fanout_pos) {
1179                         uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1180
1181                         if (i != fanout_value)
1182                                 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1183                                              cur_fanout_pos, fanout_value, i);
1184                         cur_fanout_pos++;
1185                 }
1186
1187                 graph_commit = lookup_commit(r, &cur_oid);
1188                 if (!parse_commit_in_graph_one(r, g, graph_commit))
1189                         graph_report(_("failed to parse commit %s from commit-graph"),
1190                                      oid_to_hex(&cur_oid));
1191         }
1192
1193         while (cur_fanout_pos < 256) {
1194                 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1195
1196                 if (g->num_commits != fanout_value)
1197                         graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1198                                      cur_fanout_pos, fanout_value, i);
1199
1200                 cur_fanout_pos++;
1201         }
1202
1203         if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1204                 return verify_commit_graph_error;
1205
1206         progress = start_progress(_("Verifying commits in commit graph"),
1207                                   g->num_commits);
1208         for (i = 0; i < g->num_commits; i++) {
1209                 struct commit *graph_commit, *odb_commit;
1210                 struct commit_list *graph_parents, *odb_parents;
1211                 uint32_t max_generation = 0;
1212
1213                 display_progress(progress, i + 1);
1214                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1215
1216                 graph_commit = lookup_commit(r, &cur_oid);
1217                 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1218                 if (parse_commit_internal(odb_commit, 0, 0)) {
1219                         graph_report(_("failed to parse commit %s from object database for commit-graph"),
1220                                      oid_to_hex(&cur_oid));
1221                         continue;
1222                 }
1223
1224                 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
1225                            get_commit_tree_oid(odb_commit)))
1226                         graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
1227                                      oid_to_hex(&cur_oid),
1228                                      oid_to_hex(get_commit_tree_oid(graph_commit)),
1229                                      oid_to_hex(get_commit_tree_oid(odb_commit)));
1230
1231                 graph_parents = graph_commit->parents;
1232                 odb_parents = odb_commit->parents;
1233
1234                 while (graph_parents) {
1235                         if (odb_parents == NULL) {
1236                                 graph_report(_("commit-graph parent list for commit %s is too long"),
1237                                              oid_to_hex(&cur_oid));
1238                                 break;
1239                         }
1240
1241                         if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1242                                 graph_report(_("commit-graph parent for %s is %s != %s"),
1243                                              oid_to_hex(&cur_oid),
1244                                              oid_to_hex(&graph_parents->item->object.oid),
1245                                              oid_to_hex(&odb_parents->item->object.oid));
1246
1247                         if (graph_parents->item->generation > max_generation)
1248                                 max_generation = graph_parents->item->generation;
1249
1250                         graph_parents = graph_parents->next;
1251                         odb_parents = odb_parents->next;
1252                 }
1253
1254                 if (odb_parents != NULL)
1255                         graph_report(_("commit-graph parent list for commit %s terminates early"),
1256                                      oid_to_hex(&cur_oid));
1257
1258                 if (!graph_commit->generation) {
1259                         if (generation_zero == GENERATION_NUMBER_EXISTS)
1260                                 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
1261                                              oid_to_hex(&cur_oid));
1262                         generation_zero = GENERATION_ZERO_EXISTS;
1263                 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1264                         graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
1265                                      oid_to_hex(&cur_oid));
1266
1267                 if (generation_zero == GENERATION_ZERO_EXISTS)
1268                         continue;
1269
1270                 /*
1271                  * If one of our parents has generation GENERATION_NUMBER_MAX, then
1272                  * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1273                  * extra logic in the following condition.
1274                  */
1275                 if (max_generation == GENERATION_NUMBER_MAX)
1276                         max_generation--;
1277
1278                 if (graph_commit->generation != max_generation + 1)
1279                         graph_report(_("commit-graph generation for commit %s is %u != %u"),
1280                                      oid_to_hex(&cur_oid),
1281                                      graph_commit->generation,
1282                                      max_generation + 1);
1283
1284                 if (graph_commit->date != odb_commit->date)
1285                         graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
1286                                      oid_to_hex(&cur_oid),
1287                                      graph_commit->date,
1288                                      odb_commit->date);
1289         }
1290         stop_progress(&progress);
1291
1292         return verify_commit_graph_error;
1293 }
1294
1295 void free_commit_graph(struct commit_graph *g)
1296 {
1297         if (!g)
1298                 return;
1299         if (g->graph_fd >= 0) {
1300                 munmap((void *)g->data, g->data_len);
1301                 g->data = NULL;
1302                 close(g->graph_fd);
1303         }
1304         free(g);
1305 }