commit-graph: store graph in struct object_store
[git/git.git] / commit-graph.c
CommitLineData
08fd81c9
DS
1#include "cache.h"
2#include "config.h"
33286dcd 3#include "dir.h"
08fd81c9
DS
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"
59fb8770 10#include "refs.h"
08fd81c9
DS
11#include "revision.h"
12#include "sha1-lookup.h"
13#include "commit-graph.h"
b10edb2d 14#include "object-store.h"
96af91d4 15#include "alloc.h"
08fd81c9
DS
16
17#define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
18#define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
19#define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
20#define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
21#define GRAPH_CHUNKID_LARGEEDGES 0x45444745 /* "EDGE" */
22
23#define GRAPH_DATA_WIDTH 36
24
25#define GRAPH_VERSION_1 0x1
26#define GRAPH_VERSION GRAPH_VERSION_1
27
28#define GRAPH_OID_VERSION_SHA1 1
29#define GRAPH_OID_LEN_SHA1 GIT_SHA1_RAWSZ
30#define GRAPH_OID_VERSION GRAPH_OID_VERSION_SHA1
31#define GRAPH_OID_LEN GRAPH_OID_LEN_SHA1
32
33#define GRAPH_OCTOPUS_EDGES_NEEDED 0x80000000
34#define GRAPH_PARENT_MISSING 0x7fffffff
35#define GRAPH_EDGE_LAST_MASK 0x7fffffff
36#define GRAPH_PARENT_NONE 0x70000000
37
38#define GRAPH_LAST_EDGE 0x80000000
39
0e3b97cc 40#define GRAPH_HEADER_SIZE 8
08fd81c9
DS
41#define GRAPH_FANOUT_SIZE (4 * 256)
42#define GRAPH_CHUNKLOOKUP_WIDTH 12
0e3b97cc
DS
43#define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
44 + GRAPH_FANOUT_SIZE + GRAPH_OID_LEN)
08fd81c9 45
2a2e32bd 46char *get_commit_graph_filename(const char *obj_dir)
08fd81c9
DS
47{
48 return xstrfmt("%s/info/commit-graph", obj_dir);
49}
50
2a2e32bd
DS
51static struct commit_graph *alloc_commit_graph(void)
52{
53 struct commit_graph *g = xcalloc(1, sizeof(*g));
54 g->graph_fd = -1;
55
56 return g;
57}
58
59struct commit_graph *load_commit_graph_one(const char *graph_file)
60{
61 void *graph_map;
62 const unsigned char *data, *chunk_lookup;
63 size_t graph_size;
64 struct stat st;
65 uint32_t i;
66 struct commit_graph *graph;
67 int fd = git_open(graph_file);
68 uint64_t last_chunk_offset;
69 uint32_t last_chunk_id;
70 uint32_t graph_signature;
71 unsigned char graph_version, hash_version;
72
73 if (fd < 0)
74 return NULL;
75 if (fstat(fd, &st)) {
76 close(fd);
77 return NULL;
78 }
79 graph_size = xsize_t(st.st_size);
80
81 if (graph_size < GRAPH_MIN_SIZE) {
82 close(fd);
83 die("graph file %s is too small", graph_file);
84 }
85 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
86 data = (const unsigned char *)graph_map;
87
88 graph_signature = get_be32(data);
89 if (graph_signature != GRAPH_SIGNATURE) {
90 error("graph signature %X does not match signature %X",
91 graph_signature, GRAPH_SIGNATURE);
92 goto cleanup_fail;
93 }
94
95 graph_version = *(unsigned char*)(data + 4);
96 if (graph_version != GRAPH_VERSION) {
97 error("graph version %X does not match version %X",
98 graph_version, GRAPH_VERSION);
99 goto cleanup_fail;
100 }
101
102 hash_version = *(unsigned char*)(data + 5);
103 if (hash_version != GRAPH_OID_VERSION) {
104 error("hash version %X does not match version %X",
105 hash_version, GRAPH_OID_VERSION);
106 goto cleanup_fail;
107 }
108
109 graph = alloc_commit_graph();
110
111 graph->hash_len = GRAPH_OID_LEN;
112 graph->num_chunks = *(unsigned char*)(data + 6);
113 graph->graph_fd = fd;
114 graph->data = graph_map;
115 graph->data_len = graph_size;
116
117 last_chunk_id = 0;
118 last_chunk_offset = 8;
119 chunk_lookup = data + 8;
120 for (i = 0; i < graph->num_chunks; i++) {
121 uint32_t chunk_id = get_be32(chunk_lookup + 0);
122 uint64_t chunk_offset = get_be64(chunk_lookup + 4);
123 int chunk_repeated = 0;
124
125 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
126
127 if (chunk_offset > graph_size - GIT_MAX_RAWSZ) {
128 error("improper chunk offset %08x%08x", (uint32_t)(chunk_offset >> 32),
129 (uint32_t)chunk_offset);
130 goto cleanup_fail;
131 }
132
133 switch (chunk_id) {
134 case GRAPH_CHUNKID_OIDFANOUT:
135 if (graph->chunk_oid_fanout)
136 chunk_repeated = 1;
137 else
138 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
139 break;
140
141 case GRAPH_CHUNKID_OIDLOOKUP:
142 if (graph->chunk_oid_lookup)
143 chunk_repeated = 1;
144 else
145 graph->chunk_oid_lookup = data + chunk_offset;
146 break;
147
148 case GRAPH_CHUNKID_DATA:
149 if (graph->chunk_commit_data)
150 chunk_repeated = 1;
151 else
152 graph->chunk_commit_data = data + chunk_offset;
153 break;
154
155 case GRAPH_CHUNKID_LARGEEDGES:
156 if (graph->chunk_large_edges)
157 chunk_repeated = 1;
158 else
159 graph->chunk_large_edges = data + chunk_offset;
160 break;
161 }
162
163 if (chunk_repeated) {
164 error("chunk id %08x appears multiple times", chunk_id);
165 goto cleanup_fail;
166 }
167
168 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
169 {
170 graph->num_commits = (chunk_offset - last_chunk_offset)
171 / graph->hash_len;
172 }
173
174 last_chunk_id = chunk_id;
175 last_chunk_offset = chunk_offset;
176 }
177
178 return graph;
179
180cleanup_fail:
181 munmap(graph_map, graph_size);
182 close(fd);
183 exit(1);
184}
185
177722b3
DS
186static void prepare_commit_graph_one(const char *obj_dir)
187{
188 char *graph_name;
189
85277506 190 if (the_repository->objects->commit_graph)
177722b3
DS
191 return;
192
193 graph_name = get_commit_graph_filename(obj_dir);
85277506
JT
194 the_repository->objects->commit_graph =
195 load_commit_graph_one(graph_name);
177722b3
DS
196
197 FREE_AND_NULL(graph_name);
198}
199
5faf357b
JT
200/*
201 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
202 *
203 * On the first invocation, this function attemps to load the commit
204 * graph if the_repository is configured to have one.
205 */
206static int prepare_commit_graph(void)
177722b3
DS
207{
208 struct alternate_object_database *alt;
209 char *obj_dir;
210
85277506
JT
211 if (the_repository->objects->commit_graph_attempted)
212 return !!the_repository->objects->commit_graph;
213 the_repository->objects->commit_graph_attempted = 1;
177722b3 214
5faf357b
JT
215 if (!core_commit_graph)
216 return 0;
217
177722b3
DS
218 obj_dir = get_object_directory();
219 prepare_commit_graph_one(obj_dir);
b10edb2d
JH
220 prepare_alt_odb(the_repository);
221 for (alt = the_repository->objects->alt_odb_list;
85277506 222 !the_repository->objects->commit_graph && alt;
b10edb2d 223 alt = alt->next)
177722b3 224 prepare_commit_graph_one(alt->path);
85277506 225 return !!the_repository->objects->commit_graph;
177722b3
DS
226}
227
228static void close_commit_graph(void)
229{
85277506
JT
230 free_commit_graph(the_repository->objects->commit_graph);
231 the_repository->objects->commit_graph = NULL;
177722b3
DS
232}
233
234static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
235{
236 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
237 g->chunk_oid_lookup, g->hash_len, pos);
238}
239
240static struct commit_list **insert_parent_or_die(struct commit_graph *g,
241 uint64_t pos,
242 struct commit_list **pptr)
243{
244 struct commit *c;
245 struct object_id oid;
96af91d4 246
53614b13
DS
247 if (pos >= g->num_commits)
248 die("invalid parent position %"PRIu64, pos);
249
177722b3 250 hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
c1f5eb49 251 c = lookup_commit(the_repository, &oid);
177722b3
DS
252 if (!c)
253 die("could not find commit %s", oid_to_hex(&oid));
254 c->graph_pos = pos;
255 return &commit_list_insert(c, pptr)->next;
256}
257
e2838d85
DS
258static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
259{
260 const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
261 item->graph_pos = pos;
262 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
263}
264
177722b3
DS
265static int fill_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t pos)
266{
177722b3
DS
267 uint32_t edge_value;
268 uint32_t *parent_data_ptr;
269 uint64_t date_low, date_high;
270 struct commit_list **pptr;
271 const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
272
273 item->object.parsed = 1;
274 item->graph_pos = pos;
275
7b8a21db 276 item->maybe_tree = NULL;
177722b3
DS
277
278 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
279 date_low = get_be32(commit_data + g->hash_len + 12);
280 item->date = (timestamp_t)((date_high << 32) | date_low);
281
83073cc9
DS
282 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
283
177722b3
DS
284 pptr = &item->parents;
285
286 edge_value = get_be32(commit_data + g->hash_len);
287 if (edge_value == GRAPH_PARENT_NONE)
288 return 1;
289 pptr = insert_parent_or_die(g, edge_value, pptr);
290
291 edge_value = get_be32(commit_data + g->hash_len + 4);
292 if (edge_value == GRAPH_PARENT_NONE)
293 return 1;
294 if (!(edge_value & GRAPH_OCTOPUS_EDGES_NEEDED)) {
295 pptr = insert_parent_or_die(g, edge_value, pptr);
296 return 1;
297 }
298
299 parent_data_ptr = (uint32_t*)(g->chunk_large_edges +
300 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
301 do {
302 edge_value = get_be32(parent_data_ptr);
303 pptr = insert_parent_or_die(g,
304 edge_value & GRAPH_EDGE_LAST_MASK,
305 pptr);
306 parent_data_ptr++;
307 } while (!(edge_value & GRAPH_LAST_EDGE));
308
309 return 1;
310}
311
e2838d85
DS
312static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
313{
314 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
315 *pos = item->graph_pos;
316 return 1;
317 } else {
318 return bsearch_graph(g, &(item->object.oid), pos);
319 }
320}
321
ee797053 322static int parse_commit_in_graph_one(struct commit_graph *g, struct commit *item)
177722b3 323{
e2838d85
DS
324 uint32_t pos;
325
177722b3
DS
326 if (!core_commit_graph)
327 return 0;
328 if (item->object.parsed)
329 return 1;
ee797053
DS
330
331 if (find_commit_in_graph(item, g, &pos))
332 return fill_commit_in_graph(item, g, pos);
333
334 return 0;
335}
336
337int parse_commit_in_graph(struct commit *item)
338{
5faf357b 339 if (!prepare_commit_graph())
ee797053 340 return 0;
85277506 341 return parse_commit_in_graph_one(the_repository->objects->commit_graph, item);
177722b3
DS
342}
343
e2838d85
DS
344void load_commit_graph_info(struct commit *item)
345{
346 uint32_t pos;
5faf357b 347 if (!prepare_commit_graph())
e2838d85 348 return;
85277506
JT
349 if (find_commit_in_graph(item, the_repository->objects->commit_graph, &pos))
350 fill_commit_graph_info(item, the_repository->objects->commit_graph, pos);
e2838d85
DS
351}
352
7b8a21db
DS
353static struct tree *load_tree_for_commit(struct commit_graph *g, struct commit *c)
354{
355 struct object_id oid;
356 const unsigned char *commit_data = g->chunk_commit_data +
357 GRAPH_DATA_WIDTH * (c->graph_pos);
358
359 hashcpy(oid.hash, commit_data);
f86bcc7b 360 c->maybe_tree = lookup_tree(the_repository, &oid);
7b8a21db
DS
361
362 return c->maybe_tree;
363}
364
0cbef8f8
DS
365static struct tree *get_commit_tree_in_graph_one(struct commit_graph *g,
366 const struct commit *c)
7b8a21db
DS
367{
368 if (c->maybe_tree)
369 return c->maybe_tree;
370 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
0cbef8f8
DS
371 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
372
373 return load_tree_for_commit(g, (struct commit *)c);
374}
7b8a21db 375
0cbef8f8
DS
376struct tree *get_commit_tree_in_graph(const struct commit *c)
377{
85277506 378 return get_commit_tree_in_graph_one(the_repository->objects->commit_graph, c);
7b8a21db
DS
379}
380
08fd81c9
DS
381static void write_graph_chunk_fanout(struct hashfile *f,
382 struct commit **commits,
383 int nr_commits)
384{
385 int i, count = 0;
386 struct commit **list = commits;
387
388 /*
389 * Write the first-level table (the list is sorted,
390 * but we use a 256-entry lookup to be able to avoid
391 * having to do eight extra binary search iterations).
392 */
393 for (i = 0; i < 256; i++) {
394 while (count < nr_commits) {
395 if ((*list)->object.oid.hash[0] != i)
396 break;
397 count++;
398 list++;
399 }
400
401 hashwrite_be32(f, count);
402 }
403}
404
405static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
406 struct commit **commits, int nr_commits)
407{
408 struct commit **list = commits;
409 int count;
410 for (count = 0; count < nr_commits; count++, list++)
411 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
412}
413
414static const unsigned char *commit_to_sha1(size_t index, void *table)
415{
416 struct commit **commits = table;
417 return commits[index]->object.oid.hash;
418}
419
420static void write_graph_chunk_data(struct hashfile *f, int hash_len,
421 struct commit **commits, int nr_commits)
422{
423 struct commit **list = commits;
424 struct commit **last = commits + nr_commits;
425 uint32_t num_extra_edges = 0;
426
427 while (list < last) {
428 struct commit_list *parent;
429 int edge_value;
430 uint32_t packedDate[2];
431
432 parse_commit(*list);
2e27bd77 433 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
08fd81c9
DS
434
435 parent = (*list)->parents;
436
437 if (!parent)
438 edge_value = GRAPH_PARENT_NONE;
439 else {
440 edge_value = sha1_pos(parent->item->object.oid.hash,
441 commits,
442 nr_commits,
443 commit_to_sha1);
444
445 if (edge_value < 0)
446 edge_value = GRAPH_PARENT_MISSING;
447 }
448
449 hashwrite_be32(f, edge_value);
450
451 if (parent)
452 parent = parent->next;
453
454 if (!parent)
455 edge_value = GRAPH_PARENT_NONE;
456 else if (parent->next)
457 edge_value = GRAPH_OCTOPUS_EDGES_NEEDED | num_extra_edges;
458 else {
459 edge_value = sha1_pos(parent->item->object.oid.hash,
460 commits,
461 nr_commits,
462 commit_to_sha1);
463 if (edge_value < 0)
464 edge_value = GRAPH_PARENT_MISSING;
465 }
466
467 hashwrite_be32(f, edge_value);
468
469 if (edge_value & GRAPH_OCTOPUS_EDGES_NEEDED) {
470 do {
471 num_extra_edges++;
472 parent = parent->next;
473 } while (parent);
474 }
475
476 if (sizeof((*list)->date) > 4)
477 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
478 else
479 packedDate[0] = 0;
480
3258c663
DS
481 packedDate[0] |= htonl((*list)->generation << 2);
482
08fd81c9
DS
483 packedDate[1] = htonl((*list)->date);
484 hashwrite(f, packedDate, 8);
485
486 list++;
487 }
488}
489
490static void write_graph_chunk_large_edges(struct hashfile *f,
491 struct commit **commits,
492 int nr_commits)
493{
494 struct commit **list = commits;
495 struct commit **last = commits + nr_commits;
496 struct commit_list *parent;
497
498 while (list < last) {
499 int num_parents = 0;
500 for (parent = (*list)->parents; num_parents < 3 && parent;
501 parent = parent->next)
502 num_parents++;
503
504 if (num_parents <= 2) {
505 list++;
506 continue;
507 }
508
509 /* Since num_parents > 2, this initializer is safe. */
510 for (parent = (*list)->parents->next; parent; parent = parent->next) {
511 int edge_value = sha1_pos(parent->item->object.oid.hash,
512 commits,
513 nr_commits,
514 commit_to_sha1);
515
516 if (edge_value < 0)
517 edge_value = GRAPH_PARENT_MISSING;
518 else if (!parent->next)
519 edge_value |= GRAPH_LAST_EDGE;
520
521 hashwrite_be32(f, edge_value);
522 }
523
524 list++;
525 }
526}
527
528static int commit_compare(const void *_a, const void *_b)
529{
530 const struct object_id *a = (const struct object_id *)_a;
531 const struct object_id *b = (const struct object_id *)_b;
532 return oidcmp(a, b);
533}
534
535struct packed_commit_list {
536 struct commit **list;
537 int nr;
538 int alloc;
539};
540
541struct packed_oid_list {
542 struct object_id *list;
543 int nr;
544 int alloc;
545};
546
547static int add_packed_commits(const struct object_id *oid,
548 struct packed_git *pack,
549 uint32_t pos,
550 void *data)
551{
552 struct packed_oid_list *list = (struct packed_oid_list*)data;
553 enum object_type type;
554 off_t offset = nth_packed_object_offset(pack, pos);
555 struct object_info oi = OBJECT_INFO_INIT;
556
557 oi.typep = &type;
fcb6df32 558 if (packed_object_info(the_repository, pack, offset, &oi) < 0)
08fd81c9
DS
559 die("unable to get type of object %s", oid_to_hex(oid));
560
561 if (type != OBJ_COMMIT)
562 return 0;
563
564 ALLOC_GROW(list->list, list->nr + 1, list->alloc);
565 oidcpy(&(list->list[list->nr]), oid);
566 list->nr++;
567
568 return 0;
569}
570
4f2542b4
DS
571static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
572{
573 struct commit_list *parent;
574 for (parent = commit->parents; parent; parent = parent->next) {
575 if (!(parent->item->object.flags & UNINTERESTING)) {
576 ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
577 oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
578 oids->nr++;
579 parent->item->object.flags |= UNINTERESTING;
580 }
581 }
582}
583
584static void close_reachable(struct packed_oid_list *oids)
585{
586 int i;
587 struct commit *commit;
588
589 for (i = 0; i < oids->nr; i++) {
c1f5eb49 590 commit = lookup_commit(the_repository, &oids->list[i]);
4f2542b4
DS
591 if (commit)
592 commit->object.flags |= UNINTERESTING;
593 }
594
595 /*
596 * As this loop runs, oids->nr may grow, but not more
597 * than the number of missing commits in the reachable
598 * closure.
599 */
600 for (i = 0; i < oids->nr; i++) {
c1f5eb49 601 commit = lookup_commit(the_repository, &oids->list[i]);
4f2542b4
DS
602
603 if (commit && !parse_commit(commit))
604 add_missing_parents(oids, commit);
605 }
606
607 for (i = 0; i < oids->nr; i++) {
c1f5eb49 608 commit = lookup_commit(the_repository, &oids->list[i]);
4f2542b4
DS
609
610 if (commit)
611 commit->object.flags &= ~UNINTERESTING;
612 }
613}
614
3258c663
DS
615static void compute_generation_numbers(struct packed_commit_list* commits)
616{
617 int i;
618 struct commit_list *list = NULL;
619
620 for (i = 0; i < commits->nr; i++) {
621 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
622 commits->list[i]->generation != GENERATION_NUMBER_ZERO)
623 continue;
624
625 commit_list_insert(commits->list[i], &list);
626 while (list) {
627 struct commit *current = list->item;
628 struct commit_list *parent;
629 int all_parents_computed = 1;
630 uint32_t max_generation = 0;
631
632 for (parent = current->parents; parent; parent = parent->next) {
633 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
634 parent->item->generation == GENERATION_NUMBER_ZERO) {
635 all_parents_computed = 0;
636 commit_list_insert(parent->item, &list);
637 break;
638 } else if (parent->item->generation > max_generation) {
639 max_generation = parent->item->generation;
640 }
641 }
642
643 if (all_parents_computed) {
644 current->generation = max_generation + 1;
645 pop_commit(&list);
646
647 if (current->generation > GENERATION_NUMBER_MAX)
648 current->generation = GENERATION_NUMBER_MAX;
649 }
650 }
651 }
652}
653
59fb8770
DS
654static int add_ref_to_list(const char *refname,
655 const struct object_id *oid,
656 int flags, void *cb_data)
657{
658 struct string_list *list = (struct string_list *)cb_data;
659
660 string_list_append(list, oid_to_hex(oid));
661 return 0;
662}
663
664void write_commit_graph_reachable(const char *obj_dir, int append)
665{
666 struct string_list list;
667
668 string_list_init(&list, 1);
669 for_each_ref(add_ref_to_list, &list);
670 write_commit_graph(obj_dir, NULL, &list, append);
671}
672
049d51a2 673void write_commit_graph(const char *obj_dir,
d88b14b3
DS
674 struct string_list *pack_indexes,
675 struct string_list *commit_hex,
7547b95b 676 int append)
08fd81c9
DS
677{
678 struct packed_oid_list oids;
679 struct packed_commit_list commits;
680 struct hashfile *f;
681 uint32_t i, count_distinct = 0;
682 char *graph_name;
08fd81c9
DS
683 struct lock_file lk = LOCK_INIT;
684 uint32_t chunk_ids[5];
685 uint64_t chunk_offsets[5];
686 int num_chunks;
687 int num_extra_edges;
688 struct commit_list *parent;
689
690 oids.nr = 0;
691 oids.alloc = approximate_object_count() / 4;
692
7547b95b
DS
693 if (append) {
694 prepare_commit_graph_one(obj_dir);
85277506
JT
695 if (the_repository->objects->commit_graph)
696 oids.alloc += the_repository->objects->commit_graph->num_commits;
7547b95b
DS
697 }
698
08fd81c9
DS
699 if (oids.alloc < 1024)
700 oids.alloc = 1024;
701 ALLOC_ARRAY(oids.list, oids.alloc);
702
85277506
JT
703 if (append && the_repository->objects->commit_graph) {
704 struct commit_graph *commit_graph =
705 the_repository->objects->commit_graph;
7547b95b
DS
706 for (i = 0; i < commit_graph->num_commits; i++) {
707 const unsigned char *hash = commit_graph->chunk_oid_lookup +
708 commit_graph->hash_len * i;
709 hashcpy(oids.list[oids.nr++].hash, hash);
710 }
711 }
712
049d51a2
DS
713 if (pack_indexes) {
714 struct strbuf packname = STRBUF_INIT;
715 int dirlen;
716 strbuf_addf(&packname, "%s/pack/", obj_dir);
717 dirlen = packname.len;
d88b14b3 718 for (i = 0; i < pack_indexes->nr; i++) {
049d51a2
DS
719 struct packed_git *p;
720 strbuf_setlen(&packname, dirlen);
d88b14b3 721 strbuf_addstr(&packname, pack_indexes->items[i].string);
049d51a2
DS
722 p = add_packed_git(packname.buf, packname.len, 1);
723 if (!p)
724 die("error adding pack %s", packname.buf);
725 if (open_pack_index(p))
726 die("error opening index for %s", packname.buf);
727 for_each_object_in_pack(p, add_packed_commits, &oids);
728 close_pack(p);
729 }
730 strbuf_release(&packname);
3d5df01b
DS
731 }
732
733 if (commit_hex) {
d88b14b3 734 for (i = 0; i < commit_hex->nr; i++) {
3d5df01b
DS
735 const char *end;
736 struct object_id oid;
737 struct commit *result;
738
d88b14b3
DS
739 if (commit_hex->items[i].string &&
740 parse_oid_hex(commit_hex->items[i].string, &oid, &end))
3d5df01b
DS
741 continue;
742
21e1ee8f 743 result = lookup_commit_reference_gently(the_repository, &oid, 1);
3d5df01b
DS
744
745 if (result) {
746 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
747 oidcpy(&oids.list[oids.nr], &(result->object.oid));
748 oids.nr++;
749 }
750 }
751 }
752
753 if (!pack_indexes && !commit_hex)
049d51a2
DS
754 for_each_packed_object(add_packed_commits, &oids, 0);
755
4f2542b4 756 close_reachable(&oids);
08fd81c9
DS
757
758 QSORT(oids.list, oids.nr, commit_compare);
759
760 count_distinct = 1;
761 for (i = 1; i < oids.nr; i++) {
762 if (oidcmp(&oids.list[i-1], &oids.list[i]))
763 count_distinct++;
764 }
765
766 if (count_distinct >= GRAPH_PARENT_MISSING)
767 die(_("the commit graph format cannot write %d commits"), count_distinct);
768
769 commits.nr = 0;
770 commits.alloc = count_distinct;
771 ALLOC_ARRAY(commits.list, commits.alloc);
772
773 num_extra_edges = 0;
774 for (i = 0; i < oids.nr; i++) {
775 int num_parents = 0;
776 if (i > 0 && !oidcmp(&oids.list[i-1], &oids.list[i]))
777 continue;
778
c1f5eb49 779 commits.list[commits.nr] = lookup_commit(the_repository, &oids.list[i]);
08fd81c9
DS
780 parse_commit(commits.list[commits.nr]);
781
782 for (parent = commits.list[commits.nr]->parents;
783 parent; parent = parent->next)
784 num_parents++;
785
786 if (num_parents > 2)
787 num_extra_edges += num_parents - 1;
788
789 commits.nr++;
790 }
791 num_chunks = num_extra_edges ? 4 : 3;
792
793 if (commits.nr >= GRAPH_PARENT_MISSING)
794 die(_("too many commits to write graph"));
795
3258c663 796 compute_generation_numbers(&commits);
08fd81c9 797
08fd81c9 798 graph_name = get_commit_graph_filename(obj_dir);
33286dcd
DS
799 if (safe_create_leading_directories(graph_name))
800 die_errno(_("unable to create leading directories of %s"),
801 graph_name);
08fd81c9 802
33286dcd 803 hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
08fd81c9
DS
804 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
805
806 hashwrite_be32(f, GRAPH_SIGNATURE);
807
808 hashwrite_u8(f, GRAPH_VERSION);
809 hashwrite_u8(f, GRAPH_OID_VERSION);
810 hashwrite_u8(f, num_chunks);
811 hashwrite_u8(f, 0); /* unused padding byte */
812
813 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
814 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
815 chunk_ids[2] = GRAPH_CHUNKID_DATA;
816 if (num_extra_edges)
817 chunk_ids[3] = GRAPH_CHUNKID_LARGEEDGES;
818 else
819 chunk_ids[3] = 0;
820 chunk_ids[4] = 0;
821
822 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
823 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
824 chunk_offsets[2] = chunk_offsets[1] + GRAPH_OID_LEN * commits.nr;
825 chunk_offsets[3] = chunk_offsets[2] + (GRAPH_OID_LEN + 16) * commits.nr;
826 chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
827
828 for (i = 0; i <= num_chunks; i++) {
829 uint32_t chunk_write[3];
830
831 chunk_write[0] = htonl(chunk_ids[i]);
832 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
833 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
834 hashwrite(f, chunk_write, 12);
835 }
836
837 write_graph_chunk_fanout(f, commits.list, commits.nr);
838 write_graph_chunk_oids(f, GRAPH_OID_LEN, commits.list, commits.nr);
839 write_graph_chunk_data(f, GRAPH_OID_LEN, commits.list, commits.nr);
840 write_graph_chunk_large_edges(f, commits.list, commits.nr);
841
177722b3 842 close_commit_graph();
08fd81c9
DS
843 finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
844 commit_lock_file(&lk);
845
846 free(oids.list);
847 oids.alloc = 0;
848 oids.nr = 0;
849}
283e68c7 850
41df0e30 851#define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
283e68c7
DS
852static int verify_commit_graph_error;
853
854static void graph_report(const char *fmt, ...)
855{
856 va_list ap;
857
858 verify_commit_graph_error = 1;
859 va_start(ap, fmt);
860 vfprintf(stderr, fmt, ap);
861 fprintf(stderr, "\n");
862 va_end(ap);
863}
864
1373e547
DS
865#define GENERATION_ZERO_EXISTS 1
866#define GENERATION_NUMBER_EXISTS 2
867
283e68c7
DS
868int verify_commit_graph(struct repository *r, struct commit_graph *g)
869{
9bda8467 870 uint32_t i, cur_fanout_pos = 0;
41df0e30 871 struct object_id prev_oid, cur_oid, checksum;
1373e547 872 int generation_zero = 0;
41df0e30
DS
873 struct hashfile *f;
874 int devnull;
9bda8467 875
283e68c7
DS
876 if (!g) {
877 graph_report("no commit-graph file loaded");
878 return 1;
879 }
880
2bd0365f
DS
881 verify_commit_graph_error = 0;
882
883 if (!g->chunk_oid_fanout)
884 graph_report("commit-graph is missing the OID Fanout chunk");
885 if (!g->chunk_oid_lookup)
886 graph_report("commit-graph is missing the OID Lookup chunk");
887 if (!g->chunk_commit_data)
888 graph_report("commit-graph is missing the Commit Data chunk");
889
9bda8467
DS
890 if (verify_commit_graph_error)
891 return verify_commit_graph_error;
892
41df0e30
DS
893 devnull = open("/dev/null", O_WRONLY);
894 f = hashfd(devnull, NULL);
895 hashwrite(f, g->data, g->data_len - g->hash_len);
896 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
897 if (hashcmp(checksum.hash, g->data + g->data_len - g->hash_len)) {
898 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
899 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
900 }
901
9bda8467 902 for (i = 0; i < g->num_commits; i++) {
2e3c0737
DS
903 struct commit *graph_commit;
904
9bda8467
DS
905 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
906
907 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
908 graph_report("commit-graph has incorrect OID order: %s then %s",
909 oid_to_hex(&prev_oid),
910 oid_to_hex(&cur_oid));
911
912 oidcpy(&prev_oid, &cur_oid);
913
914 while (cur_oid.hash[0] > cur_fanout_pos) {
915 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
916
917 if (i != fanout_value)
918 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
919 cur_fanout_pos, fanout_value, i);
920 cur_fanout_pos++;
921 }
2e3c0737 922
82952964 923 graph_commit = lookup_commit(r, &cur_oid);
2e3c0737
DS
924 if (!parse_commit_in_graph_one(g, graph_commit))
925 graph_report("failed to parse %s from commit-graph",
926 oid_to_hex(&cur_oid));
9bda8467
DS
927 }
928
929 while (cur_fanout_pos < 256) {
930 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
931
932 if (g->num_commits != fanout_value)
933 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
934 cur_fanout_pos, fanout_value, i);
935
936 cur_fanout_pos++;
937 }
938
41df0e30 939 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
96af91d4
DS
940 return verify_commit_graph_error;
941
942 for (i = 0; i < g->num_commits; i++) {
2e3c0737 943 struct commit *graph_commit, *odb_commit;
53614b13 944 struct commit_list *graph_parents, *odb_parents;
1373e547 945 uint32_t max_generation = 0;
96af91d4
DS
946
947 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
948
82952964 949 graph_commit = lookup_commit(r, &cur_oid);
96af91d4
DS
950 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
951 if (parse_commit_internal(odb_commit, 0, 0)) {
952 graph_report("failed to parse %s from object database",
953 oid_to_hex(&cur_oid));
954 continue;
955 }
2e3c0737
DS
956
957 if (oidcmp(&get_commit_tree_in_graph_one(g, graph_commit)->object.oid,
958 get_commit_tree_oid(odb_commit)))
959 graph_report("root tree OID for commit %s in commit-graph is %s != %s",
960 oid_to_hex(&cur_oid),
961 oid_to_hex(get_commit_tree_oid(graph_commit)),
962 oid_to_hex(get_commit_tree_oid(odb_commit)));
53614b13
DS
963
964 graph_parents = graph_commit->parents;
965 odb_parents = odb_commit->parents;
966
967 while (graph_parents) {
968 if (odb_parents == NULL) {
969 graph_report("commit-graph parent list for commit %s is too long",
970 oid_to_hex(&cur_oid));
971 break;
972 }
973
974 if (oidcmp(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
975 graph_report("commit-graph parent for %s is %s != %s",
976 oid_to_hex(&cur_oid),
977 oid_to_hex(&graph_parents->item->object.oid),
978 oid_to_hex(&odb_parents->item->object.oid));
979
1373e547
DS
980 if (graph_parents->item->generation > max_generation)
981 max_generation = graph_parents->item->generation;
982
53614b13
DS
983 graph_parents = graph_parents->next;
984 odb_parents = odb_parents->next;
985 }
986
987 if (odb_parents != NULL)
988 graph_report("commit-graph parent list for commit %s terminates early",
989 oid_to_hex(&cur_oid));
1373e547
DS
990
991 if (!graph_commit->generation) {
992 if (generation_zero == GENERATION_NUMBER_EXISTS)
993 graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
994 oid_to_hex(&cur_oid));
995 generation_zero = GENERATION_ZERO_EXISTS;
996 } else if (generation_zero == GENERATION_ZERO_EXISTS)
997 graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
998 oid_to_hex(&cur_oid));
999
1000 if (generation_zero == GENERATION_ZERO_EXISTS)
1001 continue;
1002
1003 /*
1004 * If one of our parents has generation GENERATION_NUMBER_MAX, then
1005 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1006 * extra logic in the following condition.
1007 */
1008 if (max_generation == GENERATION_NUMBER_MAX)
1009 max_generation--;
1010
1011 if (graph_commit->generation != max_generation + 1)
1012 graph_report("commit-graph generation for commit %s is %u != %u",
1013 oid_to_hex(&cur_oid),
1014 graph_commit->generation,
1015 max_generation + 1);
88968ebf
DS
1016
1017 if (graph_commit->date != odb_commit->date)
1018 graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
1019 oid_to_hex(&cur_oid),
1020 graph_commit->date,
1021 odb_commit->date);
96af91d4
DS
1022 }
1023
283e68c7
DS
1024 return verify_commit_graph_error;
1025}
c3756d5b
JT
1026
1027void free_commit_graph(struct commit_graph *g)
1028{
1029 if (!g)
1030 return;
1031 if (g->graph_fd >= 0) {
1032 munmap((void *)g->data, g->data_len);
1033 g->data = NULL;
1034 close(g->graph_fd);
1035 }
1036 free(g);
1037}