4 #include "git-compat-util.h"
12 #include "sha1-lookup.h"
13 #include "commit-graph.h"
14 #include "object-store.h"
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" */
23 #define GRAPH_DATA_WIDTH 36
25 #define GRAPH_VERSION_1 0x1
26 #define GRAPH_VERSION GRAPH_VERSION_1
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
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
38 #define GRAPH_LAST_EDGE 0x80000000
40 #define GRAPH_HEADER_SIZE 8
41 #define GRAPH_FANOUT_SIZE (4 * 256)
42 #define GRAPH_CHUNKLOOKUP_WIDTH 12
43 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
44 + GRAPH_FANOUT_SIZE + GRAPH_OID_LEN)
46 char *get_commit_graph_filename(const char *obj_dir
)
48 return xstrfmt("%s/info/commit-graph", obj_dir
);
51 static struct commit_graph
*alloc_commit_graph(void)
53 struct commit_graph
*g
= xcalloc(1, sizeof(*g
));
59 struct commit_graph
*load_commit_graph_one(const char *graph_file
)
62 const unsigned char *data
, *chunk_lookup
;
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
;
79 graph_size
= xsize_t(st
.st_size
);
81 if (graph_size
< GRAPH_MIN_SIZE
) {
83 die(_("graph file %s is too small"), graph_file
);
85 graph_map
= xmmap(NULL
, graph_size
, PROT_READ
, MAP_PRIVATE
, fd
, 0);
86 data
= (const unsigned char *)graph_map
;
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
);
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
);
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
);
109 graph
= alloc_commit_graph();
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
;
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;
125 chunk_lookup
+= GRAPH_CHUNKLOOKUP_WIDTH
;
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
);
134 case GRAPH_CHUNKID_OIDFANOUT
:
135 if (graph
->chunk_oid_fanout
)
138 graph
->chunk_oid_fanout
= (uint32_t*)(data
+ chunk_offset
);
141 case GRAPH_CHUNKID_OIDLOOKUP
:
142 if (graph
->chunk_oid_lookup
)
145 graph
->chunk_oid_lookup
= data
+ chunk_offset
;
148 case GRAPH_CHUNKID_DATA
:
149 if (graph
->chunk_commit_data
)
152 graph
->chunk_commit_data
= data
+ chunk_offset
;
155 case GRAPH_CHUNKID_LARGEEDGES
:
156 if (graph
->chunk_large_edges
)
159 graph
->chunk_large_edges
= data
+ chunk_offset
;
163 if (chunk_repeated
) {
164 error(_("chunk id %08x appears multiple times"), chunk_id
);
168 if (last_chunk_id
== GRAPH_CHUNKID_OIDLOOKUP
)
170 graph
->num_commits
= (chunk_offset
- last_chunk_offset
)
174 last_chunk_id
= chunk_id
;
175 last_chunk_offset
= chunk_offset
;
181 munmap(graph_map
, graph_size
);
186 static void prepare_commit_graph_one(struct repository
*r
, const char *obj_dir
)
190 if (r
->objects
->commit_graph
)
193 graph_name
= get_commit_graph_filename(obj_dir
);
194 r
->objects
->commit_graph
=
195 load_commit_graph_one(graph_name
);
197 FREE_AND_NULL(graph_name
);
201 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
203 * On the first invocation, this function attemps to load the commit
204 * graph if the_repository is configured to have one.
206 static int prepare_commit_graph(struct repository
*r
)
208 struct alternate_object_database
*alt
;
212 if (r
->objects
->commit_graph_attempted
)
213 return !!r
->objects
->commit_graph
;
214 r
->objects
->commit_graph_attempted
= 1;
216 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH
, 0) &&
217 (repo_config_get_bool(r
, "core.commitgraph", &config_value
) ||
220 * This repository is not configured to use commit graphs, so
221 * do not load one. (But report commit_graph_attempted anyway
222 * so that commit graph loading is not attempted again for this
227 obj_dir
= r
->objects
->objectdir
;
228 prepare_commit_graph_one(r
, obj_dir
);
230 for (alt
= r
->objects
->alt_odb_list
;
231 !r
->objects
->commit_graph
&& alt
;
233 prepare_commit_graph_one(r
, alt
->path
);
234 return !!r
->objects
->commit_graph
;
237 static void close_commit_graph(void)
239 free_commit_graph(the_repository
->objects
->commit_graph
);
240 the_repository
->objects
->commit_graph
= NULL
;
243 static int bsearch_graph(struct commit_graph
*g
, struct object_id
*oid
, uint32_t *pos
)
245 return bsearch_hash(oid
->hash
, g
->chunk_oid_fanout
,
246 g
->chunk_oid_lookup
, g
->hash_len
, pos
);
249 static struct commit_list
**insert_parent_or_die(struct commit_graph
*g
,
251 struct commit_list
**pptr
)
254 struct object_id oid
;
256 if (pos
>= g
->num_commits
)
257 die("invalid parent position %"PRIu64
, pos
);
259 hashcpy(oid
.hash
, g
->chunk_oid_lookup
+ g
->hash_len
* pos
);
260 c
= lookup_commit(the_repository
, &oid
);
262 die(_("could not find commit %s"), oid_to_hex(&oid
));
264 return &commit_list_insert(c
, pptr
)->next
;
267 static void fill_commit_graph_info(struct commit
*item
, struct commit_graph
*g
, uint32_t pos
)
269 const unsigned char *commit_data
= g
->chunk_commit_data
+ GRAPH_DATA_WIDTH
* pos
;
270 item
->graph_pos
= pos
;
271 item
->generation
= get_be32(commit_data
+ g
->hash_len
+ 8) >> 2;
274 static int fill_commit_in_graph(struct commit
*item
, struct commit_graph
*g
, uint32_t pos
)
277 uint32_t *parent_data_ptr
;
278 uint64_t date_low
, date_high
;
279 struct commit_list
**pptr
;
280 const unsigned char *commit_data
= g
->chunk_commit_data
+ (g
->hash_len
+ 16) * pos
;
282 item
->object
.parsed
= 1;
283 item
->graph_pos
= pos
;
285 item
->maybe_tree
= NULL
;
287 date_high
= get_be32(commit_data
+ g
->hash_len
+ 8) & 0x3;
288 date_low
= get_be32(commit_data
+ g
->hash_len
+ 12);
289 item
->date
= (timestamp_t
)((date_high
<< 32) | date_low
);
291 item
->generation
= get_be32(commit_data
+ g
->hash_len
+ 8) >> 2;
293 pptr
= &item
->parents
;
295 edge_value
= get_be32(commit_data
+ g
->hash_len
);
296 if (edge_value
== GRAPH_PARENT_NONE
)
298 pptr
= insert_parent_or_die(g
, edge_value
, pptr
);
300 edge_value
= get_be32(commit_data
+ g
->hash_len
+ 4);
301 if (edge_value
== GRAPH_PARENT_NONE
)
303 if (!(edge_value
& GRAPH_OCTOPUS_EDGES_NEEDED
)) {
304 pptr
= insert_parent_or_die(g
, edge_value
, pptr
);
308 parent_data_ptr
= (uint32_t*)(g
->chunk_large_edges
+
309 4 * (uint64_t)(edge_value
& GRAPH_EDGE_LAST_MASK
));
311 edge_value
= get_be32(parent_data_ptr
);
312 pptr
= insert_parent_or_die(g
,
313 edge_value
& GRAPH_EDGE_LAST_MASK
,
316 } while (!(edge_value
& GRAPH_LAST_EDGE
));
321 static int find_commit_in_graph(struct commit
*item
, struct commit_graph
*g
, uint32_t *pos
)
323 if (item
->graph_pos
!= COMMIT_NOT_FROM_GRAPH
) {
324 *pos
= item
->graph_pos
;
327 return bsearch_graph(g
, &(item
->object
.oid
), pos
);
331 static int parse_commit_in_graph_one(struct commit_graph
*g
, struct commit
*item
)
335 if (item
->object
.parsed
)
338 if (find_commit_in_graph(item
, g
, &pos
))
339 return fill_commit_in_graph(item
, g
, pos
);
344 int parse_commit_in_graph(struct repository
*r
, struct commit
*item
)
346 if (!prepare_commit_graph(r
))
348 return parse_commit_in_graph_one(r
->objects
->commit_graph
, item
);
351 void load_commit_graph_info(struct repository
*r
, struct commit
*item
)
354 if (!prepare_commit_graph(r
))
356 if (find_commit_in_graph(item
, r
->objects
->commit_graph
, &pos
))
357 fill_commit_graph_info(item
, r
->objects
->commit_graph
, pos
);
360 static struct tree
*load_tree_for_commit(struct commit_graph
*g
, struct commit
*c
)
362 struct object_id oid
;
363 const unsigned char *commit_data
= g
->chunk_commit_data
+
364 GRAPH_DATA_WIDTH
* (c
->graph_pos
);
366 hashcpy(oid
.hash
, commit_data
);
367 c
->maybe_tree
= lookup_tree(the_repository
, &oid
);
369 return c
->maybe_tree
;
372 static struct tree
*get_commit_tree_in_graph_one(struct commit_graph
*g
,
373 const struct commit
*c
)
376 return c
->maybe_tree
;
377 if (c
->graph_pos
== COMMIT_NOT_FROM_GRAPH
)
378 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
380 return load_tree_for_commit(g
, (struct commit
*)c
);
383 struct tree
*get_commit_tree_in_graph(struct repository
*r
, const struct commit
*c
)
385 return get_commit_tree_in_graph_one(r
->objects
->commit_graph
, c
);
388 static void write_graph_chunk_fanout(struct hashfile
*f
,
389 struct commit
**commits
,
393 struct commit
**list
= commits
;
396 * Write the first-level table (the list is sorted,
397 * but we use a 256-entry lookup to be able to avoid
398 * having to do eight extra binary search iterations).
400 for (i
= 0; i
< 256; i
++) {
401 while (count
< nr_commits
) {
402 if ((*list
)->object
.oid
.hash
[0] != i
)
408 hashwrite_be32(f
, count
);
412 static void write_graph_chunk_oids(struct hashfile
*f
, int hash_len
,
413 struct commit
**commits
, int nr_commits
)
415 struct commit
**list
= commits
;
417 for (count
= 0; count
< nr_commits
; count
++, list
++)
418 hashwrite(f
, (*list
)->object
.oid
.hash
, (int)hash_len
);
421 static const unsigned char *commit_to_sha1(size_t index
, void *table
)
423 struct commit
**commits
= table
;
424 return commits
[index
]->object
.oid
.hash
;
427 static void write_graph_chunk_data(struct hashfile
*f
, int hash_len
,
428 struct commit
**commits
, int nr_commits
)
430 struct commit
**list
= commits
;
431 struct commit
**last
= commits
+ nr_commits
;
432 uint32_t num_extra_edges
= 0;
434 while (list
< last
) {
435 struct commit_list
*parent
;
437 uint32_t packedDate
[2];
440 hashwrite(f
, get_commit_tree_oid(*list
)->hash
, hash_len
);
442 parent
= (*list
)->parents
;
445 edge_value
= GRAPH_PARENT_NONE
;
447 edge_value
= sha1_pos(parent
->item
->object
.oid
.hash
,
453 edge_value
= GRAPH_PARENT_MISSING
;
456 hashwrite_be32(f
, edge_value
);
459 parent
= parent
->next
;
462 edge_value
= GRAPH_PARENT_NONE
;
463 else if (parent
->next
)
464 edge_value
= GRAPH_OCTOPUS_EDGES_NEEDED
| num_extra_edges
;
466 edge_value
= sha1_pos(parent
->item
->object
.oid
.hash
,
471 edge_value
= GRAPH_PARENT_MISSING
;
474 hashwrite_be32(f
, edge_value
);
476 if (edge_value
& GRAPH_OCTOPUS_EDGES_NEEDED
) {
479 parent
= parent
->next
;
483 if (sizeof((*list
)->date
) > 4)
484 packedDate
[0] = htonl(((*list
)->date
>> 32) & 0x3);
488 packedDate
[0] |= htonl((*list
)->generation
<< 2);
490 packedDate
[1] = htonl((*list
)->date
);
491 hashwrite(f
, packedDate
, 8);
497 static void write_graph_chunk_large_edges(struct hashfile
*f
,
498 struct commit
**commits
,
501 struct commit
**list
= commits
;
502 struct commit
**last
= commits
+ nr_commits
;
503 struct commit_list
*parent
;
505 while (list
< last
) {
507 for (parent
= (*list
)->parents
; num_parents
< 3 && parent
;
508 parent
= parent
->next
)
511 if (num_parents
<= 2) {
516 /* Since num_parents > 2, this initializer is safe. */
517 for (parent
= (*list
)->parents
->next
; parent
; parent
= parent
->next
) {
518 int edge_value
= sha1_pos(parent
->item
->object
.oid
.hash
,
524 edge_value
= GRAPH_PARENT_MISSING
;
525 else if (!parent
->next
)
526 edge_value
|= GRAPH_LAST_EDGE
;
528 hashwrite_be32(f
, edge_value
);
535 static int commit_compare(const void *_a
, const void *_b
)
537 const struct object_id
*a
= (const struct object_id
*)_a
;
538 const struct object_id
*b
= (const struct object_id
*)_b
;
542 struct packed_commit_list
{
543 struct commit
**list
;
548 struct packed_oid_list
{
549 struct object_id
*list
;
554 static int add_packed_commits(const struct object_id
*oid
,
555 struct packed_git
*pack
,
559 struct packed_oid_list
*list
= (struct packed_oid_list
*)data
;
560 enum object_type type
;
561 off_t offset
= nth_packed_object_offset(pack
, pos
);
562 struct object_info oi
= OBJECT_INFO_INIT
;
565 if (packed_object_info(the_repository
, pack
, offset
, &oi
) < 0)
566 die(_("unable to get type of object %s"), oid_to_hex(oid
));
568 if (type
!= OBJ_COMMIT
)
571 ALLOC_GROW(list
->list
, list
->nr
+ 1, list
->alloc
);
572 oidcpy(&(list
->list
[list
->nr
]), oid
);
578 static void add_missing_parents(struct packed_oid_list
*oids
, struct commit
*commit
)
580 struct commit_list
*parent
;
581 for (parent
= commit
->parents
; parent
; parent
= parent
->next
) {
582 if (!(parent
->item
->object
.flags
& UNINTERESTING
)) {
583 ALLOC_GROW(oids
->list
, oids
->nr
+ 1, oids
->alloc
);
584 oidcpy(&oids
->list
[oids
->nr
], &(parent
->item
->object
.oid
));
586 parent
->item
->object
.flags
|= UNINTERESTING
;
591 static void close_reachable(struct packed_oid_list
*oids
)
594 struct commit
*commit
;
596 for (i
= 0; i
< oids
->nr
; i
++) {
597 commit
= lookup_commit(the_repository
, &oids
->list
[i
]);
599 commit
->object
.flags
|= UNINTERESTING
;
603 * As this loop runs, oids->nr may grow, but not more
604 * than the number of missing commits in the reachable
607 for (i
= 0; i
< oids
->nr
; i
++) {
608 commit
= lookup_commit(the_repository
, &oids
->list
[i
]);
610 if (commit
&& !parse_commit(commit
))
611 add_missing_parents(oids
, commit
);
614 for (i
= 0; i
< oids
->nr
; i
++) {
615 commit
= lookup_commit(the_repository
, &oids
->list
[i
]);
618 commit
->object
.flags
&= ~UNINTERESTING
;
622 static void compute_generation_numbers(struct packed_commit_list
* commits
)
625 struct commit_list
*list
= NULL
;
627 for (i
= 0; i
< commits
->nr
; i
++) {
628 if (commits
->list
[i
]->generation
!= GENERATION_NUMBER_INFINITY
&&
629 commits
->list
[i
]->generation
!= GENERATION_NUMBER_ZERO
)
632 commit_list_insert(commits
->list
[i
], &list
);
634 struct commit
*current
= list
->item
;
635 struct commit_list
*parent
;
636 int all_parents_computed
= 1;
637 uint32_t max_generation
= 0;
639 for (parent
= current
->parents
; parent
; parent
= parent
->next
) {
640 if (parent
->item
->generation
== GENERATION_NUMBER_INFINITY
||
641 parent
->item
->generation
== GENERATION_NUMBER_ZERO
) {
642 all_parents_computed
= 0;
643 commit_list_insert(parent
->item
, &list
);
645 } else if (parent
->item
->generation
> max_generation
) {
646 max_generation
= parent
->item
->generation
;
650 if (all_parents_computed
) {
651 current
->generation
= max_generation
+ 1;
654 if (current
->generation
> GENERATION_NUMBER_MAX
)
655 current
->generation
= GENERATION_NUMBER_MAX
;
661 static int add_ref_to_list(const char *refname
,
662 const struct object_id
*oid
,
663 int flags
, void *cb_data
)
665 struct string_list
*list
= (struct string_list
*)cb_data
;
667 string_list_append(list
, oid_to_hex(oid
));
671 void write_commit_graph_reachable(const char *obj_dir
, int append
)
673 struct string_list list
;
675 string_list_init(&list
, 1);
676 for_each_ref(add_ref_to_list
, &list
);
677 write_commit_graph(obj_dir
, NULL
, &list
, append
);
680 void write_commit_graph(const char *obj_dir
,
681 struct string_list
*pack_indexes
,
682 struct string_list
*commit_hex
,
685 struct packed_oid_list oids
;
686 struct packed_commit_list commits
;
688 uint32_t i
, count_distinct
= 0;
690 struct lock_file lk
= LOCK_INIT
;
691 uint32_t chunk_ids
[5];
692 uint64_t chunk_offsets
[5];
695 struct commit_list
*parent
;
698 oids
.alloc
= approximate_object_count() / 4;
701 prepare_commit_graph_one(the_repository
, obj_dir
);
702 if (the_repository
->objects
->commit_graph
)
703 oids
.alloc
+= the_repository
->objects
->commit_graph
->num_commits
;
706 if (oids
.alloc
< 1024)
708 ALLOC_ARRAY(oids
.list
, oids
.alloc
);
710 if (append
&& the_repository
->objects
->commit_graph
) {
711 struct commit_graph
*commit_graph
=
712 the_repository
->objects
->commit_graph
;
713 for (i
= 0; i
< commit_graph
->num_commits
; i
++) {
714 const unsigned char *hash
= commit_graph
->chunk_oid_lookup
+
715 commit_graph
->hash_len
* i
;
716 hashcpy(oids
.list
[oids
.nr
++].hash
, hash
);
721 struct strbuf packname
= STRBUF_INIT
;
723 strbuf_addf(&packname
, "%s/pack/", obj_dir
);
724 dirlen
= packname
.len
;
725 for (i
= 0; i
< pack_indexes
->nr
; i
++) {
726 struct packed_git
*p
;
727 strbuf_setlen(&packname
, dirlen
);
728 strbuf_addstr(&packname
, pack_indexes
->items
[i
].string
);
729 p
= add_packed_git(packname
.buf
, packname
.len
, 1);
731 die(_("error adding pack %s"), packname
.buf
);
732 if (open_pack_index(p
))
733 die(_("error opening index for %s"), packname
.buf
);
734 for_each_object_in_pack(p
, add_packed_commits
, &oids
, 0);
737 strbuf_release(&packname
);
741 for (i
= 0; i
< commit_hex
->nr
; i
++) {
743 struct object_id oid
;
744 struct commit
*result
;
746 if (commit_hex
->items
[i
].string
&&
747 parse_oid_hex(commit_hex
->items
[i
].string
, &oid
, &end
))
750 result
= lookup_commit_reference_gently(the_repository
, &oid
, 1);
753 ALLOC_GROW(oids
.list
, oids
.nr
+ 1, oids
.alloc
);
754 oidcpy(&oids
.list
[oids
.nr
], &(result
->object
.oid
));
760 if (!pack_indexes
&& !commit_hex
)
761 for_each_packed_object(add_packed_commits
, &oids
, 0);
763 close_reachable(&oids
);
765 QSORT(oids
.list
, oids
.nr
, commit_compare
);
768 for (i
= 1; i
< oids
.nr
; i
++) {
769 if (oidcmp(&oids
.list
[i
-1], &oids
.list
[i
]))
773 if (count_distinct
>= GRAPH_PARENT_MISSING
)
774 die(_("the commit graph format cannot write %d commits"), count_distinct
);
777 commits
.alloc
= count_distinct
;
778 ALLOC_ARRAY(commits
.list
, commits
.alloc
);
781 for (i
= 0; i
< oids
.nr
; i
++) {
783 if (i
> 0 && !oidcmp(&oids
.list
[i
-1], &oids
.list
[i
]))
786 commits
.list
[commits
.nr
] = lookup_commit(the_repository
, &oids
.list
[i
]);
787 parse_commit(commits
.list
[commits
.nr
]);
789 for (parent
= commits
.list
[commits
.nr
]->parents
;
790 parent
; parent
= parent
->next
)
794 num_extra_edges
+= num_parents
- 1;
798 num_chunks
= num_extra_edges
? 4 : 3;
800 if (commits
.nr
>= GRAPH_PARENT_MISSING
)
801 die(_("too many commits to write graph"));
803 compute_generation_numbers(&commits
);
805 graph_name
= get_commit_graph_filename(obj_dir
);
806 if (safe_create_leading_directories(graph_name
))
807 die_errno(_("unable to create leading directories of %s"),
810 hold_lock_file_for_update(&lk
, graph_name
, LOCK_DIE_ON_ERROR
);
811 f
= hashfd(lk
.tempfile
->fd
, lk
.tempfile
->filename
.buf
);
813 hashwrite_be32(f
, GRAPH_SIGNATURE
);
815 hashwrite_u8(f
, GRAPH_VERSION
);
816 hashwrite_u8(f
, GRAPH_OID_VERSION
);
817 hashwrite_u8(f
, num_chunks
);
818 hashwrite_u8(f
, 0); /* unused padding byte */
820 chunk_ids
[0] = GRAPH_CHUNKID_OIDFANOUT
;
821 chunk_ids
[1] = GRAPH_CHUNKID_OIDLOOKUP
;
822 chunk_ids
[2] = GRAPH_CHUNKID_DATA
;
824 chunk_ids
[3] = GRAPH_CHUNKID_LARGEEDGES
;
829 chunk_offsets
[0] = 8 + (num_chunks
+ 1) * GRAPH_CHUNKLOOKUP_WIDTH
;
830 chunk_offsets
[1] = chunk_offsets
[0] + GRAPH_FANOUT_SIZE
;
831 chunk_offsets
[2] = chunk_offsets
[1] + GRAPH_OID_LEN
* commits
.nr
;
832 chunk_offsets
[3] = chunk_offsets
[2] + (GRAPH_OID_LEN
+ 16) * commits
.nr
;
833 chunk_offsets
[4] = chunk_offsets
[3] + 4 * num_extra_edges
;
835 for (i
= 0; i
<= num_chunks
; i
++) {
836 uint32_t chunk_write
[3];
838 chunk_write
[0] = htonl(chunk_ids
[i
]);
839 chunk_write
[1] = htonl(chunk_offsets
[i
] >> 32);
840 chunk_write
[2] = htonl(chunk_offsets
[i
] & 0xffffffff);
841 hashwrite(f
, chunk_write
, 12);
844 write_graph_chunk_fanout(f
, commits
.list
, commits
.nr
);
845 write_graph_chunk_oids(f
, GRAPH_OID_LEN
, commits
.list
, commits
.nr
);
846 write_graph_chunk_data(f
, GRAPH_OID_LEN
, commits
.list
, commits
.nr
);
847 write_graph_chunk_large_edges(f
, commits
.list
, commits
.nr
);
849 close_commit_graph();
850 finalize_hashfile(f
, NULL
, CSUM_HASH_IN_STREAM
| CSUM_FSYNC
);
851 commit_lock_file(&lk
);
858 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
859 static int verify_commit_graph_error
;
861 static void graph_report(const char *fmt
, ...)
865 verify_commit_graph_error
= 1;
867 vfprintf(stderr
, fmt
, ap
);
868 fprintf(stderr
, "\n");
872 #define GENERATION_ZERO_EXISTS 1
873 #define GENERATION_NUMBER_EXISTS 2
875 int verify_commit_graph(struct repository
*r
, struct commit_graph
*g
)
877 uint32_t i
, cur_fanout_pos
= 0;
878 struct object_id prev_oid
, cur_oid
, checksum
;
879 int generation_zero
= 0;
884 graph_report("no commit-graph file loaded");
888 verify_commit_graph_error
= 0;
890 if (!g
->chunk_oid_fanout
)
891 graph_report("commit-graph is missing the OID Fanout chunk");
892 if (!g
->chunk_oid_lookup
)
893 graph_report("commit-graph is missing the OID Lookup chunk");
894 if (!g
->chunk_commit_data
)
895 graph_report("commit-graph is missing the Commit Data chunk");
897 if (verify_commit_graph_error
)
898 return verify_commit_graph_error
;
900 devnull
= open("/dev/null", O_WRONLY
);
901 f
= hashfd(devnull
, NULL
);
902 hashwrite(f
, g
->data
, g
->data_len
- g
->hash_len
);
903 finalize_hashfile(f
, checksum
.hash
, CSUM_CLOSE
);
904 if (hashcmp(checksum
.hash
, g
->data
+ g
->data_len
- g
->hash_len
)) {
905 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
906 verify_commit_graph_error
= VERIFY_COMMIT_GRAPH_ERROR_HASH
;
909 for (i
= 0; i
< g
->num_commits
; i
++) {
910 struct commit
*graph_commit
;
912 hashcpy(cur_oid
.hash
, g
->chunk_oid_lookup
+ g
->hash_len
* i
);
914 if (i
&& oidcmp(&prev_oid
, &cur_oid
) >= 0)
915 graph_report("commit-graph has incorrect OID order: %s then %s",
916 oid_to_hex(&prev_oid
),
917 oid_to_hex(&cur_oid
));
919 oidcpy(&prev_oid
, &cur_oid
);
921 while (cur_oid
.hash
[0] > cur_fanout_pos
) {
922 uint32_t fanout_value
= get_be32(g
->chunk_oid_fanout
+ cur_fanout_pos
);
924 if (i
!= fanout_value
)
925 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
926 cur_fanout_pos
, fanout_value
, i
);
930 graph_commit
= lookup_commit(r
, &cur_oid
);
931 if (!parse_commit_in_graph_one(g
, graph_commit
))
932 graph_report("failed to parse %s from commit-graph",
933 oid_to_hex(&cur_oid
));
936 while (cur_fanout_pos
< 256) {
937 uint32_t fanout_value
= get_be32(g
->chunk_oid_fanout
+ cur_fanout_pos
);
939 if (g
->num_commits
!= fanout_value
)
940 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
941 cur_fanout_pos
, fanout_value
, i
);
946 if (verify_commit_graph_error
& ~VERIFY_COMMIT_GRAPH_ERROR_HASH
)
947 return verify_commit_graph_error
;
949 for (i
= 0; i
< g
->num_commits
; i
++) {
950 struct commit
*graph_commit
, *odb_commit
;
951 struct commit_list
*graph_parents
, *odb_parents
;
952 uint32_t max_generation
= 0;
954 hashcpy(cur_oid
.hash
, g
->chunk_oid_lookup
+ g
->hash_len
* i
);
956 graph_commit
= lookup_commit(r
, &cur_oid
);
957 odb_commit
= (struct commit
*)create_object(r
, cur_oid
.hash
, alloc_commit_node(r
));
958 if (parse_commit_internal(odb_commit
, 0, 0)) {
959 graph_report("failed to parse %s from object database",
960 oid_to_hex(&cur_oid
));
964 if (oidcmp(&get_commit_tree_in_graph_one(g
, graph_commit
)->object
.oid
,
965 get_commit_tree_oid(odb_commit
)))
966 graph_report("root tree OID for commit %s in commit-graph is %s != %s",
967 oid_to_hex(&cur_oid
),
968 oid_to_hex(get_commit_tree_oid(graph_commit
)),
969 oid_to_hex(get_commit_tree_oid(odb_commit
)));
971 graph_parents
= graph_commit
->parents
;
972 odb_parents
= odb_commit
->parents
;
974 while (graph_parents
) {
975 if (odb_parents
== NULL
) {
976 graph_report("commit-graph parent list for commit %s is too long",
977 oid_to_hex(&cur_oid
));
981 if (oidcmp(&graph_parents
->item
->object
.oid
, &odb_parents
->item
->object
.oid
))
982 graph_report("commit-graph parent for %s is %s != %s",
983 oid_to_hex(&cur_oid
),
984 oid_to_hex(&graph_parents
->item
->object
.oid
),
985 oid_to_hex(&odb_parents
->item
->object
.oid
));
987 if (graph_parents
->item
->generation
> max_generation
)
988 max_generation
= graph_parents
->item
->generation
;
990 graph_parents
= graph_parents
->next
;
991 odb_parents
= odb_parents
->next
;
994 if (odb_parents
!= NULL
)
995 graph_report("commit-graph parent list for commit %s terminates early",
996 oid_to_hex(&cur_oid
));
998 if (!graph_commit
->generation
) {
999 if (generation_zero
== GENERATION_NUMBER_EXISTS
)
1000 graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
1001 oid_to_hex(&cur_oid
));
1002 generation_zero
= GENERATION_ZERO_EXISTS
;
1003 } else if (generation_zero
== GENERATION_ZERO_EXISTS
)
1004 graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
1005 oid_to_hex(&cur_oid
));
1007 if (generation_zero
== GENERATION_ZERO_EXISTS
)
1011 * If one of our parents has generation GENERATION_NUMBER_MAX, then
1012 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1013 * extra logic in the following condition.
1015 if (max_generation
== GENERATION_NUMBER_MAX
)
1018 if (graph_commit
->generation
!= max_generation
+ 1)
1019 graph_report("commit-graph generation for commit %s is %u != %u",
1020 oid_to_hex(&cur_oid
),
1021 graph_commit
->generation
,
1022 max_generation
+ 1);
1024 if (graph_commit
->date
!= odb_commit
->date
)
1025 graph_report("commit date for commit %s in commit-graph is %"PRItime
" != %"PRItime
,
1026 oid_to_hex(&cur_oid
),
1031 return verify_commit_graph_error
;
1034 void free_commit_graph(struct commit_graph
*g
)
1038 if (g
->graph_fd
>= 0) {
1039 munmap((void *)g
->data
, g
->data_len
);