1 #include "git-compat-util.h"
10 #include "sha1-lookup.h"
11 #include "commit-graph.h"
12 #include "object-store.h"
15 #include "replace-object.h"
18 #include "commit-slab.h"
20 void git_test_write_commit_graph_or_die(void)
23 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH
, 0))
26 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS
, 0))
27 flags
= COMMIT_GRAPH_WRITE_BLOOM_FILTERS
;
29 if (write_commit_graph_reachable(the_repository
->objects
->odb
,
31 die("failed to write commit-graph under GIT_TEST_COMMIT_GRAPH");
34 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
35 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
36 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
37 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
38 #define GRAPH_CHUNKID_EXTRAEDGES 0x45444745 /* "EDGE" */
39 #define GRAPH_CHUNKID_BLOOMINDEXES 0x42494458 /* "BIDX" */
40 #define GRAPH_CHUNKID_BLOOMDATA 0x42444154 /* "BDAT" */
41 #define GRAPH_CHUNKID_BASE 0x42415345 /* "BASE" */
42 #define MAX_NUM_CHUNKS 7
44 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
46 #define GRAPH_VERSION_1 0x1
47 #define GRAPH_VERSION GRAPH_VERSION_1
49 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
50 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
51 #define GRAPH_PARENT_NONE 0x70000000
53 #define GRAPH_LAST_EDGE 0x80000000
55 #define GRAPH_HEADER_SIZE 8
56 #define GRAPH_FANOUT_SIZE (4 * 256)
57 #define GRAPH_CHUNKLOOKUP_WIDTH 12
58 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
59 + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
61 /* Remember to update object flag allocation in object.h */
62 #define REACHABLE (1u<<15)
64 /* Keep track of the order in which commits are added to our list. */
65 define_commit_slab(commit_pos
, int);
66 static struct commit_pos commit_pos
= COMMIT_SLAB_INIT(1, commit_pos
);
68 static void set_commit_pos(struct repository
*r
, const struct object_id
*oid
)
70 static int32_t max_pos
;
71 struct commit
*commit
= lookup_commit(r
, oid
);
74 return; /* should never happen, but be lenient */
76 *commit_pos_at(&commit_pos
, commit
) = max_pos
++;
79 static int commit_pos_cmp(const void *va
, const void *vb
)
81 const struct commit
*a
= *(const struct commit
**)va
;
82 const struct commit
*b
= *(const struct commit
**)vb
;
83 return commit_pos_at(&commit_pos
, a
) -
84 commit_pos_at(&commit_pos
, b
);
87 static int commit_gen_cmp(const void *va
, const void *vb
)
89 const struct commit
*a
= *(const struct commit
**)va
;
90 const struct commit
*b
= *(const struct commit
**)vb
;
92 /* lower generation commits first */
93 if (a
->generation
< b
->generation
)
95 else if (a
->generation
> b
->generation
)
98 /* use date as a heuristic when generations are equal */
99 if (a
->date
< b
->date
)
101 else if (a
->date
> b
->date
)
106 char *get_commit_graph_filename(struct object_directory
*obj_dir
)
108 return xstrfmt("%s/info/commit-graph", obj_dir
->path
);
111 static char *get_split_graph_filename(struct object_directory
*odb
,
114 return xstrfmt("%s/info/commit-graphs/graph-%s.graph", odb
->path
,
118 static char *get_chain_filename(struct object_directory
*odb
)
120 return xstrfmt("%s/info/commit-graphs/commit-graph-chain", odb
->path
);
123 static uint8_t oid_version(void)
128 static struct commit_graph
*alloc_commit_graph(void)
130 struct commit_graph
*g
= xcalloc(1, sizeof(*g
));
136 extern int read_replace_refs
;
138 static int commit_graph_compatible(struct repository
*r
)
143 if (read_replace_refs
) {
144 prepare_replace_object(r
);
145 if (hashmap_get_size(&r
->objects
->replace_map
->map
))
149 prepare_commit_graft(r
);
150 if (r
->parsed_objects
&& r
->parsed_objects
->grafts_nr
)
152 if (is_repository_shallow(r
))
158 int open_commit_graph(const char *graph_file
, int *fd
, struct stat
*st
)
160 *fd
= git_open(graph_file
);
163 if (fstat(*fd
, st
)) {
170 struct commit_graph
*load_commit_graph_one_fd_st(int fd
, struct stat
*st
,
171 struct object_directory
*odb
)
175 struct commit_graph
*ret
;
177 graph_size
= xsize_t(st
->st_size
);
179 if (graph_size
< GRAPH_MIN_SIZE
) {
181 error(_("commit-graph file is too small"));
184 graph_map
= xmmap(NULL
, graph_size
, PROT_READ
, MAP_PRIVATE
, fd
, 0);
185 ret
= parse_commit_graph(graph_map
, fd
, graph_size
);
190 munmap(graph_map
, graph_size
);
197 static int verify_commit_graph_lite(struct commit_graph
*g
)
200 * Basic validation shared between parse_commit_graph()
201 * which'll be called every time the graph is used, and the
202 * much more expensive verify_commit_graph() used by
203 * "commit-graph verify".
205 * There should only be very basic checks here to ensure that
206 * we don't e.g. segfault in fill_commit_in_graph(), but
207 * because this is a very hot codepath nothing that e.g. loops
208 * over g->num_commits, or runs a checksum on the commit-graph
211 if (!g
->chunk_oid_fanout
) {
212 error("commit-graph is missing the OID Fanout chunk");
215 if (!g
->chunk_oid_lookup
) {
216 error("commit-graph is missing the OID Lookup chunk");
219 if (!g
->chunk_commit_data
) {
220 error("commit-graph is missing the Commit Data chunk");
227 struct commit_graph
*parse_commit_graph(void *graph_map
, int fd
,
230 const unsigned char *data
, *chunk_lookup
;
232 struct commit_graph
*graph
;
233 uint64_t next_chunk_offset
;
234 uint32_t graph_signature
;
235 unsigned char graph_version
, hash_version
;
240 if (graph_size
< GRAPH_MIN_SIZE
)
243 data
= (const unsigned char *)graph_map
;
245 graph_signature
= get_be32(data
);
246 if (graph_signature
!= GRAPH_SIGNATURE
) {
247 error(_("commit-graph signature %X does not match signature %X"),
248 graph_signature
, GRAPH_SIGNATURE
);
252 graph_version
= *(unsigned char*)(data
+ 4);
253 if (graph_version
!= GRAPH_VERSION
) {
254 error(_("commit-graph version %X does not match version %X"),
255 graph_version
, GRAPH_VERSION
);
259 hash_version
= *(unsigned char*)(data
+ 5);
260 if (hash_version
!= oid_version()) {
261 error(_("commit-graph hash version %X does not match version %X"),
262 hash_version
, oid_version());
266 graph
= alloc_commit_graph();
268 graph
->hash_len
= the_hash_algo
->rawsz
;
269 graph
->num_chunks
= *(unsigned char*)(data
+ 6);
270 graph
->graph_fd
= fd
;
271 graph
->data
= graph_map
;
272 graph
->data_len
= graph_size
;
274 if (graph_size
< GRAPH_HEADER_SIZE
+
275 (graph
->num_chunks
+ 1) * GRAPH_CHUNKLOOKUP_WIDTH
+
276 GRAPH_FANOUT_SIZE
+ the_hash_algo
->rawsz
) {
277 error(_("commit-graph file is too small to hold %u chunks"),
283 chunk_lookup
= data
+ 8;
284 next_chunk_offset
= get_be64(chunk_lookup
+ 4);
285 for (i
= 0; i
< graph
->num_chunks
; i
++) {
287 uint64_t chunk_offset
= next_chunk_offset
;
288 int chunk_repeated
= 0;
290 chunk_id
= get_be32(chunk_lookup
+ 0);
292 chunk_lookup
+= GRAPH_CHUNKLOOKUP_WIDTH
;
293 next_chunk_offset
= get_be64(chunk_lookup
+ 4);
295 if (chunk_offset
> graph_size
- the_hash_algo
->rawsz
) {
296 error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset
>> 32),
297 (uint32_t)chunk_offset
);
303 case GRAPH_CHUNKID_OIDFANOUT
:
304 if (graph
->chunk_oid_fanout
)
307 graph
->chunk_oid_fanout
= (uint32_t*)(data
+ chunk_offset
);
310 case GRAPH_CHUNKID_OIDLOOKUP
:
311 if (graph
->chunk_oid_lookup
)
314 graph
->chunk_oid_lookup
= data
+ chunk_offset
;
315 graph
->num_commits
= (next_chunk_offset
- chunk_offset
)
320 case GRAPH_CHUNKID_DATA
:
321 if (graph
->chunk_commit_data
)
324 graph
->chunk_commit_data
= data
+ chunk_offset
;
327 case GRAPH_CHUNKID_EXTRAEDGES
:
328 if (graph
->chunk_extra_edges
)
331 graph
->chunk_extra_edges
= data
+ chunk_offset
;
334 case GRAPH_CHUNKID_BASE
:
335 if (graph
->chunk_base_graphs
)
338 graph
->chunk_base_graphs
= data
+ chunk_offset
;
341 case GRAPH_CHUNKID_BLOOMINDEXES
:
342 if (graph
->chunk_bloom_indexes
)
345 graph
->chunk_bloom_indexes
= data
+ chunk_offset
;
348 case GRAPH_CHUNKID_BLOOMDATA
:
349 if (graph
->chunk_bloom_data
)
352 uint32_t hash_version
;
353 graph
->chunk_bloom_data
= data
+ chunk_offset
;
354 hash_version
= get_be32(data
+ chunk_offset
);
356 if (hash_version
!= 1)
359 graph
->bloom_filter_settings
= xmalloc(sizeof(struct bloom_filter_settings
));
360 graph
->bloom_filter_settings
->hash_version
= hash_version
;
361 graph
->bloom_filter_settings
->num_hashes
= get_be32(data
+ chunk_offset
+ 4);
362 graph
->bloom_filter_settings
->bits_per_entry
= get_be32(data
+ chunk_offset
+ 8);
367 if (chunk_repeated
) {
368 error(_("commit-graph chunk id %08x appears multiple times"), chunk_id
);
374 if (graph
->chunk_bloom_indexes
&& graph
->chunk_bloom_data
) {
375 init_bloom_filters();
377 /* We need both the bloom chunks to exist together. Else ignore the data */
378 graph
->chunk_bloom_indexes
= NULL
;
379 graph
->chunk_bloom_data
= NULL
;
380 graph
->bloom_filter_settings
= NULL
;
383 hashcpy(graph
->oid
.hash
, graph
->data
+ graph
->data_len
- graph
->hash_len
);
385 if (verify_commit_graph_lite(graph
)) {
393 static struct commit_graph
*load_commit_graph_one(const char *graph_file
,
394 struct object_directory
*odb
)
399 struct commit_graph
*g
;
400 int open_ok
= open_commit_graph(graph_file
, &fd
, &st
);
405 g
= load_commit_graph_one_fd_st(fd
, &st
, odb
);
408 g
->filename
= xstrdup(graph_file
);
413 static struct commit_graph
*load_commit_graph_v1(struct repository
*r
,
414 struct object_directory
*odb
)
416 char *graph_name
= get_commit_graph_filename(odb
);
417 struct commit_graph
*g
= load_commit_graph_one(graph_name
, odb
);
423 static int add_graph_to_chain(struct commit_graph
*g
,
424 struct commit_graph
*chain
,
425 struct object_id
*oids
,
428 struct commit_graph
*cur_g
= chain
;
430 if (n
&& !g
->chunk_base_graphs
) {
431 warning(_("commit-graph has no base graphs chunk"));
439 !oideq(&oids
[n
], &cur_g
->oid
) ||
440 !hasheq(oids
[n
].hash
, g
->chunk_base_graphs
+ g
->hash_len
* n
)) {
441 warning(_("commit-graph chain does not match"));
445 cur_g
= cur_g
->base_graph
;
448 g
->base_graph
= chain
;
451 g
->num_commits_in_base
= chain
->num_commits
+ chain
->num_commits_in_base
;
456 static struct commit_graph
*load_commit_graph_chain(struct repository
*r
,
457 struct object_directory
*odb
)
459 struct commit_graph
*graph_chain
= NULL
;
460 struct strbuf line
= STRBUF_INIT
;
462 struct object_id
*oids
;
463 int i
= 0, valid
= 1, count
;
464 char *chain_name
= get_chain_filename(odb
);
468 fp
= fopen(chain_name
, "r");
469 stat_res
= stat(chain_name
, &st
);
474 st
.st_size
<= the_hash_algo
->hexsz
)
477 count
= st
.st_size
/ (the_hash_algo
->hexsz
+ 1);
478 oids
= xcalloc(count
, sizeof(struct object_id
));
482 for (i
= 0; i
< count
; i
++) {
483 struct object_directory
*odb
;
485 if (strbuf_getline_lf(&line
, fp
) == EOF
)
488 if (get_oid_hex(line
.buf
, &oids
[i
])) {
489 warning(_("invalid commit-graph chain: line '%s' not a hash"),
496 for (odb
= r
->objects
->odb
; odb
; odb
= odb
->next
) {
497 char *graph_name
= get_split_graph_filename(odb
, line
.buf
);
498 struct commit_graph
*g
= load_commit_graph_one(graph_name
, odb
);
503 if (add_graph_to_chain(g
, graph_chain
, oids
, i
)) {
513 warning(_("unable to find all commit-graph files"));
520 strbuf_release(&line
);
525 struct commit_graph
*read_commit_graph_one(struct repository
*r
,
526 struct object_directory
*odb
)
528 struct commit_graph
*g
= load_commit_graph_v1(r
, odb
);
531 g
= load_commit_graph_chain(r
, odb
);
536 static void prepare_commit_graph_one(struct repository
*r
,
537 struct object_directory
*odb
)
540 if (r
->objects
->commit_graph
)
543 r
->objects
->commit_graph
= read_commit_graph_one(r
, odb
);
547 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
549 * On the first invocation, this function attempts to load the commit
550 * graph if the_repository is configured to have one.
552 static int prepare_commit_graph(struct repository
*r
)
554 struct object_directory
*odb
;
557 * This must come before the "already attempted?" check below, because
558 * we want to disable even an already-loaded graph file.
560 if (r
->commit_graph_disabled
)
563 if (r
->objects
->commit_graph_attempted
)
564 return !!r
->objects
->commit_graph
;
565 r
->objects
->commit_graph_attempted
= 1;
567 prepare_repo_settings(r
);
569 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH
, 0) &&
570 r
->settings
.core_commit_graph
!= 1)
572 * This repository is not configured to use commit graphs, so
573 * do not load one. (But report commit_graph_attempted anyway
574 * so that commit graph loading is not attempted again for this
579 if (!commit_graph_compatible(r
))
583 for (odb
= r
->objects
->odb
;
584 !r
->objects
->commit_graph
&& odb
;
586 prepare_commit_graph_one(r
, odb
);
587 return !!r
->objects
->commit_graph
;
590 int generation_numbers_enabled(struct repository
*r
)
592 uint32_t first_generation
;
593 struct commit_graph
*g
;
594 if (!prepare_commit_graph(r
))
597 g
= r
->objects
->commit_graph
;
602 first_generation
= get_be32(g
->chunk_commit_data
+
603 g
->hash_len
+ 8) >> 2;
605 return !!first_generation
;
608 static void close_commit_graph_one(struct commit_graph
*g
)
613 close_commit_graph_one(g
->base_graph
);
614 free_commit_graph(g
);
617 void close_commit_graph(struct raw_object_store
*o
)
619 close_commit_graph_one(o
->commit_graph
);
620 o
->commit_graph
= NULL
;
623 static int bsearch_graph(struct commit_graph
*g
, struct object_id
*oid
, uint32_t *pos
)
625 return bsearch_hash(oid
->hash
, g
->chunk_oid_fanout
,
626 g
->chunk_oid_lookup
, g
->hash_len
, pos
);
629 static void load_oid_from_graph(struct commit_graph
*g
,
631 struct object_id
*oid
)
635 while (g
&& pos
< g
->num_commits_in_base
)
639 BUG("NULL commit-graph");
641 if (pos
>= g
->num_commits
+ g
->num_commits_in_base
)
642 die(_("invalid commit position. commit-graph is likely corrupt"));
644 lex_index
= pos
- g
->num_commits_in_base
;
646 hashcpy(oid
->hash
, g
->chunk_oid_lookup
+ g
->hash_len
* lex_index
);
649 static struct commit_list
**insert_parent_or_die(struct repository
*r
,
650 struct commit_graph
*g
,
652 struct commit_list
**pptr
)
655 struct object_id oid
;
657 if (pos
>= g
->num_commits
+ g
->num_commits_in_base
)
658 die("invalid parent position %"PRIu32
, pos
);
660 load_oid_from_graph(g
, pos
, &oid
);
661 c
= lookup_commit(r
, &oid
);
663 die(_("could not find commit %s"), oid_to_hex(&oid
));
665 return &commit_list_insert(c
, pptr
)->next
;
668 static void fill_commit_graph_info(struct commit
*item
, struct commit_graph
*g
, uint32_t pos
)
670 const unsigned char *commit_data
;
673 while (pos
< g
->num_commits_in_base
)
676 lex_index
= pos
- g
->num_commits_in_base
;
677 commit_data
= g
->chunk_commit_data
+ GRAPH_DATA_WIDTH
* lex_index
;
678 item
->graph_pos
= pos
;
679 item
->generation
= get_be32(commit_data
+ g
->hash_len
+ 8) >> 2;
682 static inline void set_commit_tree(struct commit
*c
, struct tree
*t
)
687 static int fill_commit_in_graph(struct repository
*r
,
689 struct commit_graph
*g
, uint32_t pos
)
692 uint32_t *parent_data_ptr
;
693 uint64_t date_low
, date_high
;
694 struct commit_list
**pptr
;
695 const unsigned char *commit_data
;
698 while (pos
< g
->num_commits_in_base
)
701 if (pos
>= g
->num_commits
+ g
->num_commits_in_base
)
702 die(_("invalid commit position. commit-graph is likely corrupt"));
705 * Store the "full" position, but then use the
706 * "local" position for the rest of the calculation.
708 item
->graph_pos
= pos
;
709 lex_index
= pos
- g
->num_commits_in_base
;
711 commit_data
= g
->chunk_commit_data
+ (g
->hash_len
+ 16) * lex_index
;
713 item
->object
.parsed
= 1;
715 set_commit_tree(item
, NULL
);
717 date_high
= get_be32(commit_data
+ g
->hash_len
+ 8) & 0x3;
718 date_low
= get_be32(commit_data
+ g
->hash_len
+ 12);
719 item
->date
= (timestamp_t
)((date_high
<< 32) | date_low
);
721 item
->generation
= get_be32(commit_data
+ g
->hash_len
+ 8) >> 2;
723 pptr
= &item
->parents
;
725 edge_value
= get_be32(commit_data
+ g
->hash_len
);
726 if (edge_value
== GRAPH_PARENT_NONE
)
728 pptr
= insert_parent_or_die(r
, g
, edge_value
, pptr
);
730 edge_value
= get_be32(commit_data
+ g
->hash_len
+ 4);
731 if (edge_value
== GRAPH_PARENT_NONE
)
733 if (!(edge_value
& GRAPH_EXTRA_EDGES_NEEDED
)) {
734 pptr
= insert_parent_or_die(r
, g
, edge_value
, pptr
);
738 parent_data_ptr
= (uint32_t*)(g
->chunk_extra_edges
+
739 4 * (uint64_t)(edge_value
& GRAPH_EDGE_LAST_MASK
));
741 edge_value
= get_be32(parent_data_ptr
);
742 pptr
= insert_parent_or_die(r
, g
,
743 edge_value
& GRAPH_EDGE_LAST_MASK
,
746 } while (!(edge_value
& GRAPH_LAST_EDGE
));
751 static int find_commit_in_graph(struct commit
*item
, struct commit_graph
*g
, uint32_t *pos
)
753 if (item
->graph_pos
!= COMMIT_NOT_FROM_GRAPH
) {
754 *pos
= item
->graph_pos
;
757 struct commit_graph
*cur_g
= g
;
760 while (cur_g
&& !bsearch_graph(cur_g
, &(item
->object
.oid
), &lex_index
))
761 cur_g
= cur_g
->base_graph
;
764 *pos
= lex_index
+ cur_g
->num_commits_in_base
;
772 static int parse_commit_in_graph_one(struct repository
*r
,
773 struct commit_graph
*g
,
778 if (item
->object
.parsed
)
781 if (find_commit_in_graph(item
, g
, &pos
))
782 return fill_commit_in_graph(r
, item
, g
, pos
);
787 int parse_commit_in_graph(struct repository
*r
, struct commit
*item
)
789 static int checked_env
= 0;
792 git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_PARSE
, 0))
793 die("dying as requested by the '%s' variable on commit-graph parse!",
794 GIT_TEST_COMMIT_GRAPH_DIE_ON_PARSE
);
797 if (!prepare_commit_graph(r
))
799 return parse_commit_in_graph_one(r
, r
->objects
->commit_graph
, item
);
802 void load_commit_graph_info(struct repository
*r
, struct commit
*item
)
805 if (!prepare_commit_graph(r
))
807 if (find_commit_in_graph(item
, r
->objects
->commit_graph
, &pos
))
808 fill_commit_graph_info(item
, r
->objects
->commit_graph
, pos
);
811 static struct tree
*load_tree_for_commit(struct repository
*r
,
812 struct commit_graph
*g
,
815 struct object_id oid
;
816 const unsigned char *commit_data
;
818 while (c
->graph_pos
< g
->num_commits_in_base
)
821 commit_data
= g
->chunk_commit_data
+
822 GRAPH_DATA_WIDTH
* (c
->graph_pos
- g
->num_commits_in_base
);
824 hashcpy(oid
.hash
, commit_data
);
825 set_commit_tree(c
, lookup_tree(r
, &oid
));
827 return c
->maybe_tree
;
830 static struct tree
*get_commit_tree_in_graph_one(struct repository
*r
,
831 struct commit_graph
*g
,
832 const struct commit
*c
)
835 return c
->maybe_tree
;
836 if (c
->graph_pos
== COMMIT_NOT_FROM_GRAPH
)
837 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
839 return load_tree_for_commit(r
, g
, (struct commit
*)c
);
842 struct tree
*get_commit_tree_in_graph(struct repository
*r
, const struct commit
*c
)
844 return get_commit_tree_in_graph_one(r
, r
->objects
->commit_graph
, c
);
847 struct packed_commit_list
{
848 struct commit
**list
;
853 struct packed_oid_list
{
854 struct object_id
*list
;
859 struct write_commit_graph_context
{
860 struct repository
*r
;
861 struct object_directory
*odb
;
863 struct packed_oid_list oids
;
864 struct packed_commit_list commits
;
866 unsigned long approx_nr_objects
;
867 struct progress
*progress
;
869 uint64_t progress_cnt
;
871 char *base_graph_name
;
872 int num_commit_graphs_before
;
873 int num_commit_graphs_after
;
874 char **commit_graph_filenames_before
;
875 char **commit_graph_filenames_after
;
876 char **commit_graph_hash_after
;
877 uint32_t new_num_commits_in_base
;
878 struct commit_graph
*new_base_graph
;
887 const struct split_commit_graph_opts
*split_opts
;
888 size_t total_bloom_filter_data_size
;
889 const struct bloom_filter_settings
*bloom_settings
;
892 static void write_graph_chunk_fanout(struct hashfile
*f
,
893 struct write_commit_graph_context
*ctx
)
896 struct commit
**list
= ctx
->commits
.list
;
899 * Write the first-level table (the list is sorted,
900 * but we use a 256-entry lookup to be able to avoid
901 * having to do eight extra binary search iterations).
903 for (i
= 0; i
< 256; i
++) {
904 while (count
< ctx
->commits
.nr
) {
905 if ((*list
)->object
.oid
.hash
[0] != i
)
907 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
912 hashwrite_be32(f
, count
);
916 static void write_graph_chunk_oids(struct hashfile
*f
, int hash_len
,
917 struct write_commit_graph_context
*ctx
)
919 struct commit
**list
= ctx
->commits
.list
;
921 for (count
= 0; count
< ctx
->commits
.nr
; count
++, list
++) {
922 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
923 hashwrite(f
, (*list
)->object
.oid
.hash
, (int)hash_len
);
927 static const unsigned char *commit_to_sha1(size_t index
, void *table
)
929 struct commit
**commits
= table
;
930 return commits
[index
]->object
.oid
.hash
;
933 static void write_graph_chunk_data(struct hashfile
*f
, int hash_len
,
934 struct write_commit_graph_context
*ctx
)
936 struct commit
**list
= ctx
->commits
.list
;
937 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
938 uint32_t num_extra_edges
= 0;
940 while (list
< last
) {
941 struct commit_list
*parent
;
942 struct object_id
*tree
;
944 uint32_t packedDate
[2];
945 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
947 if (parse_commit_no_graph(*list
))
948 die(_("unable to parse commit %s"),
949 oid_to_hex(&(*list
)->object
.oid
));
950 tree
= get_commit_tree_oid(*list
);
951 hashwrite(f
, tree
->hash
, hash_len
);
953 parent
= (*list
)->parents
;
956 edge_value
= GRAPH_PARENT_NONE
;
958 edge_value
= sha1_pos(parent
->item
->object
.oid
.hash
,
964 edge_value
+= ctx
->new_num_commits_in_base
;
967 if (find_commit_in_graph(parent
->item
,
974 BUG("missing parent %s for commit %s",
975 oid_to_hex(&parent
->item
->object
.oid
),
976 oid_to_hex(&(*list
)->object
.oid
));
979 hashwrite_be32(f
, edge_value
);
982 parent
= parent
->next
;
985 edge_value
= GRAPH_PARENT_NONE
;
986 else if (parent
->next
)
987 edge_value
= GRAPH_EXTRA_EDGES_NEEDED
| num_extra_edges
;
989 edge_value
= sha1_pos(parent
->item
->object
.oid
.hash
,
995 edge_value
+= ctx
->new_num_commits_in_base
;
998 if (find_commit_in_graph(parent
->item
,
1005 BUG("missing parent %s for commit %s",
1006 oid_to_hex(&parent
->item
->object
.oid
),
1007 oid_to_hex(&(*list
)->object
.oid
));
1010 hashwrite_be32(f
, edge_value
);
1012 if (edge_value
& GRAPH_EXTRA_EDGES_NEEDED
) {
1015 parent
= parent
->next
;
1019 if (sizeof((*list
)->date
) > 4)
1020 packedDate
[0] = htonl(((*list
)->date
>> 32) & 0x3);
1024 packedDate
[0] |= htonl((*list
)->generation
<< 2);
1026 packedDate
[1] = htonl((*list
)->date
);
1027 hashwrite(f
, packedDate
, 8);
1033 static void write_graph_chunk_extra_edges(struct hashfile
*f
,
1034 struct write_commit_graph_context
*ctx
)
1036 struct commit
**list
= ctx
->commits
.list
;
1037 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
1038 struct commit_list
*parent
;
1040 while (list
< last
) {
1041 int num_parents
= 0;
1043 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1045 for (parent
= (*list
)->parents
; num_parents
< 3 && parent
;
1046 parent
= parent
->next
)
1049 if (num_parents
<= 2) {
1054 /* Since num_parents > 2, this initializer is safe. */
1055 for (parent
= (*list
)->parents
->next
; parent
; parent
= parent
->next
) {
1056 int edge_value
= sha1_pos(parent
->item
->object
.oid
.hash
,
1061 if (edge_value
>= 0)
1062 edge_value
+= ctx
->new_num_commits_in_base
;
1065 if (find_commit_in_graph(parent
->item
,
1066 ctx
->new_base_graph
,
1072 BUG("missing parent %s for commit %s",
1073 oid_to_hex(&parent
->item
->object
.oid
),
1074 oid_to_hex(&(*list
)->object
.oid
));
1075 else if (!parent
->next
)
1076 edge_value
|= GRAPH_LAST_EDGE
;
1078 hashwrite_be32(f
, edge_value
);
1085 static void write_graph_chunk_bloom_indexes(struct hashfile
*f
,
1086 struct write_commit_graph_context
*ctx
)
1088 struct commit
**list
= ctx
->commits
.list
;
1089 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
1090 uint32_t cur_pos
= 0;
1091 struct progress
*progress
= NULL
;
1094 if (ctx
->report_progress
)
1095 progress
= start_delayed_progress(
1096 _("Writing changed paths Bloom filters index"),
1099 while (list
< last
) {
1100 struct bloom_filter
*filter
= get_bloom_filter(ctx
->r
, *list
, 0);
1101 cur_pos
+= filter
->len
;
1102 display_progress(progress
, ++i
);
1103 hashwrite_be32(f
, cur_pos
);
1107 stop_progress(&progress
);
1110 static void write_graph_chunk_bloom_data(struct hashfile
*f
,
1111 struct write_commit_graph_context
*ctx
)
1113 struct commit
**list
= ctx
->commits
.list
;
1114 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
1115 struct progress
*progress
= NULL
;
1118 if (ctx
->report_progress
)
1119 progress
= start_delayed_progress(
1120 _("Writing changed paths Bloom filters data"),
1123 hashwrite_be32(f
, ctx
->bloom_settings
->hash_version
);
1124 hashwrite_be32(f
, ctx
->bloom_settings
->num_hashes
);
1125 hashwrite_be32(f
, ctx
->bloom_settings
->bits_per_entry
);
1127 while (list
< last
) {
1128 struct bloom_filter
*filter
= get_bloom_filter(ctx
->r
, *list
, 0);
1129 display_progress(progress
, ++i
);
1130 hashwrite(f
, filter
->data
, filter
->len
* sizeof(unsigned char));
1134 stop_progress(&progress
);
1137 static int oid_compare(const void *_a
, const void *_b
)
1139 const struct object_id
*a
= (const struct object_id
*)_a
;
1140 const struct object_id
*b
= (const struct object_id
*)_b
;
1141 return oidcmp(a
, b
);
1144 static int add_packed_commits(const struct object_id
*oid
,
1145 struct packed_git
*pack
,
1149 struct write_commit_graph_context
*ctx
= (struct write_commit_graph_context
*)data
;
1150 enum object_type type
;
1151 off_t offset
= nth_packed_object_offset(pack
, pos
);
1152 struct object_info oi
= OBJECT_INFO_INIT
;
1155 display_progress(ctx
->progress
, ++ctx
->progress_done
);
1158 if (packed_object_info(ctx
->r
, pack
, offset
, &oi
) < 0)
1159 die(_("unable to get type of object %s"), oid_to_hex(oid
));
1161 if (type
!= OBJ_COMMIT
)
1164 ALLOC_GROW(ctx
->oids
.list
, ctx
->oids
.nr
+ 1, ctx
->oids
.alloc
);
1165 oidcpy(&(ctx
->oids
.list
[ctx
->oids
.nr
]), oid
);
1168 set_commit_pos(ctx
->r
, oid
);
1173 static void add_missing_parents(struct write_commit_graph_context
*ctx
, struct commit
*commit
)
1175 struct commit_list
*parent
;
1176 for (parent
= commit
->parents
; parent
; parent
= parent
->next
) {
1177 if (!(parent
->item
->object
.flags
& REACHABLE
)) {
1178 ALLOC_GROW(ctx
->oids
.list
, ctx
->oids
.nr
+ 1, ctx
->oids
.alloc
);
1179 oidcpy(&ctx
->oids
.list
[ctx
->oids
.nr
], &(parent
->item
->object
.oid
));
1181 parent
->item
->object
.flags
|= REACHABLE
;
1186 static void close_reachable(struct write_commit_graph_context
*ctx
)
1189 struct commit
*commit
;
1191 if (ctx
->report_progress
)
1192 ctx
->progress
= start_delayed_progress(
1193 _("Loading known commits in commit graph"),
1195 for (i
= 0; i
< ctx
->oids
.nr
; i
++) {
1196 display_progress(ctx
->progress
, i
+ 1);
1197 commit
= lookup_commit(ctx
->r
, &ctx
->oids
.list
[i
]);
1199 commit
->object
.flags
|= REACHABLE
;
1201 stop_progress(&ctx
->progress
);
1204 * As this loop runs, ctx->oids.nr may grow, but not more
1205 * than the number of missing commits in the reachable
1208 if (ctx
->report_progress
)
1209 ctx
->progress
= start_delayed_progress(
1210 _("Expanding reachable commits in commit graph"),
1212 for (i
= 0; i
< ctx
->oids
.nr
; i
++) {
1213 display_progress(ctx
->progress
, i
+ 1);
1214 commit
= lookup_commit(ctx
->r
, &ctx
->oids
.list
[i
]);
1219 if (!parse_commit(commit
) &&
1220 commit
->graph_pos
== COMMIT_NOT_FROM_GRAPH
)
1221 add_missing_parents(ctx
, commit
);
1222 } else if (!parse_commit_no_graph(commit
))
1223 add_missing_parents(ctx
, commit
);
1225 stop_progress(&ctx
->progress
);
1227 if (ctx
->report_progress
)
1228 ctx
->progress
= start_delayed_progress(
1229 _("Clearing commit marks in commit graph"),
1231 for (i
= 0; i
< ctx
->oids
.nr
; i
++) {
1232 display_progress(ctx
->progress
, i
+ 1);
1233 commit
= lookup_commit(ctx
->r
, &ctx
->oids
.list
[i
]);
1236 commit
->object
.flags
&= ~REACHABLE
;
1238 stop_progress(&ctx
->progress
);
1241 static void compute_generation_numbers(struct write_commit_graph_context
*ctx
)
1244 struct commit_list
*list
= NULL
;
1246 if (ctx
->report_progress
)
1247 ctx
->progress
= start_delayed_progress(
1248 _("Computing commit graph generation numbers"),
1250 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1251 display_progress(ctx
->progress
, i
+ 1);
1252 if (ctx
->commits
.list
[i
]->generation
!= GENERATION_NUMBER_INFINITY
&&
1253 ctx
->commits
.list
[i
]->generation
!= GENERATION_NUMBER_ZERO
)
1256 commit_list_insert(ctx
->commits
.list
[i
], &list
);
1258 struct commit
*current
= list
->item
;
1259 struct commit_list
*parent
;
1260 int all_parents_computed
= 1;
1261 uint32_t max_generation
= 0;
1263 for (parent
= current
->parents
; parent
; parent
= parent
->next
) {
1264 if (parent
->item
->generation
== GENERATION_NUMBER_INFINITY
||
1265 parent
->item
->generation
== GENERATION_NUMBER_ZERO
) {
1266 all_parents_computed
= 0;
1267 commit_list_insert(parent
->item
, &list
);
1269 } else if (parent
->item
->generation
> max_generation
) {
1270 max_generation
= parent
->item
->generation
;
1274 if (all_parents_computed
) {
1275 current
->generation
= max_generation
+ 1;
1278 if (current
->generation
> GENERATION_NUMBER_MAX
)
1279 current
->generation
= GENERATION_NUMBER_MAX
;
1283 stop_progress(&ctx
->progress
);
1286 static void compute_bloom_filters(struct write_commit_graph_context
*ctx
)
1289 struct progress
*progress
= NULL
;
1290 struct commit
**sorted_commits
;
1292 init_bloom_filters();
1294 if (ctx
->report_progress
)
1295 progress
= start_delayed_progress(
1296 _("Computing commit changed paths Bloom filters"),
1299 ALLOC_ARRAY(sorted_commits
, ctx
->commits
.nr
);
1300 COPY_ARRAY(sorted_commits
, ctx
->commits
.list
, ctx
->commits
.nr
);
1302 if (ctx
->order_by_pack
)
1303 QSORT(sorted_commits
, ctx
->commits
.nr
, commit_pos_cmp
);
1305 QSORT(sorted_commits
, ctx
->commits
.nr
, commit_gen_cmp
);
1307 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1308 struct commit
*c
= sorted_commits
[i
];
1309 struct bloom_filter
*filter
= get_bloom_filter(ctx
->r
, c
, 1);
1310 ctx
->total_bloom_filter_data_size
+= sizeof(unsigned char) * filter
->len
;
1311 display_progress(progress
, i
+ 1);
1314 free(sorted_commits
);
1315 stop_progress(&progress
);
1318 static int add_ref_to_list(const char *refname
,
1319 const struct object_id
*oid
,
1320 int flags
, void *cb_data
)
1322 struct string_list
*list
= (struct string_list
*)cb_data
;
1324 string_list_append(list
, oid_to_hex(oid
));
1328 int write_commit_graph_reachable(struct object_directory
*odb
,
1329 enum commit_graph_write_flags flags
,
1330 const struct split_commit_graph_opts
*split_opts
)
1332 struct string_list list
= STRING_LIST_INIT_DUP
;
1335 for_each_ref(add_ref_to_list
, &list
);
1336 result
= write_commit_graph(odb
, NULL
, &list
,
1339 string_list_clear(&list
, 0);
1343 static int fill_oids_from_packs(struct write_commit_graph_context
*ctx
,
1344 struct string_list
*pack_indexes
)
1347 struct strbuf progress_title
= STRBUF_INIT
;
1348 struct strbuf packname
= STRBUF_INIT
;
1351 strbuf_addf(&packname
, "%s/pack/", ctx
->odb
->path
);
1352 dirlen
= packname
.len
;
1353 if (ctx
->report_progress
) {
1354 strbuf_addf(&progress_title
,
1355 Q_("Finding commits for commit graph in %d pack",
1356 "Finding commits for commit graph in %d packs",
1359 ctx
->progress
= start_delayed_progress(progress_title
.buf
, 0);
1360 ctx
->progress_done
= 0;
1362 for (i
= 0; i
< pack_indexes
->nr
; i
++) {
1363 struct packed_git
*p
;
1364 strbuf_setlen(&packname
, dirlen
);
1365 strbuf_addstr(&packname
, pack_indexes
->items
[i
].string
);
1366 p
= add_packed_git(packname
.buf
, packname
.len
, 1);
1368 error(_("error adding pack %s"), packname
.buf
);
1371 if (open_pack_index(p
)) {
1372 error(_("error opening index for %s"), packname
.buf
);
1375 for_each_object_in_pack(p
, add_packed_commits
, ctx
,
1376 FOR_EACH_OBJECT_PACK_ORDER
);
1381 stop_progress(&ctx
->progress
);
1382 strbuf_release(&progress_title
);
1383 strbuf_release(&packname
);
1388 static int fill_oids_from_commit_hex(struct write_commit_graph_context
*ctx
,
1389 struct string_list
*commit_hex
)
1392 struct strbuf progress_title
= STRBUF_INIT
;
1394 if (ctx
->report_progress
) {
1395 strbuf_addf(&progress_title
,
1396 Q_("Finding commits for commit graph from %d ref",
1397 "Finding commits for commit graph from %d refs",
1400 ctx
->progress
= start_delayed_progress(
1404 for (i
= 0; i
< commit_hex
->nr
; i
++) {
1406 struct object_id oid
;
1407 struct commit
*result
;
1409 display_progress(ctx
->progress
, i
+ 1);
1410 if (!parse_oid_hex(commit_hex
->items
[i
].string
, &oid
, &end
) &&
1411 (result
= lookup_commit_reference_gently(ctx
->r
, &oid
, 1))) {
1412 ALLOC_GROW(ctx
->oids
.list
, ctx
->oids
.nr
+ 1, ctx
->oids
.alloc
);
1413 oidcpy(&ctx
->oids
.list
[ctx
->oids
.nr
], &(result
->object
.oid
));
1415 } else if (ctx
->check_oids
) {
1416 error(_("invalid commit object id: %s"),
1417 commit_hex
->items
[i
].string
);
1421 stop_progress(&ctx
->progress
);
1422 strbuf_release(&progress_title
);
1427 static void fill_oids_from_all_packs(struct write_commit_graph_context
*ctx
)
1429 if (ctx
->report_progress
)
1430 ctx
->progress
= start_delayed_progress(
1431 _("Finding commits for commit graph among packed objects"),
1432 ctx
->approx_nr_objects
);
1433 for_each_packed_object(add_packed_commits
, ctx
,
1434 FOR_EACH_OBJECT_PACK_ORDER
);
1435 if (ctx
->progress_done
< ctx
->approx_nr_objects
)
1436 display_progress(ctx
->progress
, ctx
->approx_nr_objects
);
1437 stop_progress(&ctx
->progress
);
1440 static uint32_t count_distinct_commits(struct write_commit_graph_context
*ctx
)
1442 uint32_t i
, count_distinct
= 1;
1444 if (ctx
->report_progress
)
1445 ctx
->progress
= start_delayed_progress(
1446 _("Counting distinct commits in commit graph"),
1448 display_progress(ctx
->progress
, 0); /* TODO: Measure QSORT() progress */
1449 QSORT(ctx
->oids
.list
, ctx
->oids
.nr
, oid_compare
);
1451 for (i
= 1; i
< ctx
->oids
.nr
; i
++) {
1452 display_progress(ctx
->progress
, i
+ 1);
1453 if (!oideq(&ctx
->oids
.list
[i
- 1], &ctx
->oids
.list
[i
])) {
1455 struct commit
*c
= lookup_commit(ctx
->r
, &ctx
->oids
.list
[i
]);
1457 if (!c
|| c
->graph_pos
!= COMMIT_NOT_FROM_GRAPH
)
1464 stop_progress(&ctx
->progress
);
1466 return count_distinct
;
1469 static void copy_oids_to_commits(struct write_commit_graph_context
*ctx
)
1473 ctx
->num_extra_edges
= 0;
1474 if (ctx
->report_progress
)
1475 ctx
->progress
= start_delayed_progress(
1476 _("Finding extra edges in commit graph"),
1478 for (i
= 0; i
< ctx
->oids
.nr
; i
++) {
1479 unsigned int num_parents
;
1481 display_progress(ctx
->progress
, i
+ 1);
1482 if (i
> 0 && oideq(&ctx
->oids
.list
[i
- 1], &ctx
->oids
.list
[i
]))
1485 ALLOC_GROW(ctx
->commits
.list
, ctx
->commits
.nr
+ 1, ctx
->commits
.alloc
);
1486 ctx
->commits
.list
[ctx
->commits
.nr
] = lookup_commit(ctx
->r
, &ctx
->oids
.list
[i
]);
1489 ctx
->commits
.list
[ctx
->commits
.nr
]->graph_pos
!= COMMIT_NOT_FROM_GRAPH
)
1492 parse_commit_no_graph(ctx
->commits
.list
[ctx
->commits
.nr
]);
1494 num_parents
= commit_list_count(ctx
->commits
.list
[ctx
->commits
.nr
]->parents
);
1495 if (num_parents
> 2)
1496 ctx
->num_extra_edges
+= num_parents
- 1;
1500 stop_progress(&ctx
->progress
);
1503 static int write_graph_chunk_base_1(struct hashfile
*f
,
1504 struct commit_graph
*g
)
1511 num
= write_graph_chunk_base_1(f
, g
->base_graph
);
1512 hashwrite(f
, g
->oid
.hash
, the_hash_algo
->rawsz
);
1516 static int write_graph_chunk_base(struct hashfile
*f
,
1517 struct write_commit_graph_context
*ctx
)
1519 int num
= write_graph_chunk_base_1(f
, ctx
->new_base_graph
);
1521 if (num
!= ctx
->num_commit_graphs_after
- 1) {
1522 error(_("failed to write correct number of base graph ids"));
1534 static int write_commit_graph_file(struct write_commit_graph_context
*ctx
)
1539 struct lock_file lk
= LOCK_INIT
;
1540 struct chunk_info chunks
[MAX_NUM_CHUNKS
+ 1];
1541 const unsigned hashsz
= the_hash_algo
->rawsz
;
1542 struct strbuf progress_title
= STRBUF_INIT
;
1544 uint64_t chunk_offset
;
1545 struct object_id file_hash
;
1546 const struct bloom_filter_settings bloom_settings
= DEFAULT_BLOOM_FILTER_SETTINGS
;
1548 ctx
->bloom_settings
= &bloom_settings
;
1551 struct strbuf tmp_file
= STRBUF_INIT
;
1553 strbuf_addf(&tmp_file
,
1554 "%s/info/commit-graphs/tmp_graph_XXXXXX",
1556 ctx
->graph_name
= strbuf_detach(&tmp_file
, NULL
);
1558 ctx
->graph_name
= get_commit_graph_filename(ctx
->odb
);
1561 if (safe_create_leading_directories(ctx
->graph_name
)) {
1562 UNLEAK(ctx
->graph_name
);
1563 error(_("unable to create leading directories of %s"),
1569 char *lock_name
= get_chain_filename(ctx
->odb
);
1571 hold_lock_file_for_update(&lk
, lock_name
, LOCK_DIE_ON_ERROR
);
1573 fd
= git_mkstemp_mode(ctx
->graph_name
, 0444);
1575 error(_("unable to create '%s'"), ctx
->graph_name
);
1579 f
= hashfd(fd
, ctx
->graph_name
);
1581 hold_lock_file_for_update(&lk
, ctx
->graph_name
, LOCK_DIE_ON_ERROR
);
1582 fd
= lk
.tempfile
->fd
;
1583 f
= hashfd(lk
.tempfile
->fd
, lk
.tempfile
->filename
.buf
);
1586 chunks
[0].id
= GRAPH_CHUNKID_OIDFANOUT
;
1587 chunks
[0].size
= GRAPH_FANOUT_SIZE
;
1588 chunks
[1].id
= GRAPH_CHUNKID_OIDLOOKUP
;
1589 chunks
[1].size
= hashsz
* ctx
->commits
.nr
;
1590 chunks
[2].id
= GRAPH_CHUNKID_DATA
;
1591 chunks
[2].size
= (hashsz
+ 16) * ctx
->commits
.nr
;
1592 if (ctx
->num_extra_edges
) {
1593 chunks
[num_chunks
].id
= GRAPH_CHUNKID_EXTRAEDGES
;
1594 chunks
[num_chunks
].size
= 4 * ctx
->num_extra_edges
;
1597 if (ctx
->changed_paths
) {
1598 chunks
[num_chunks
].id
= GRAPH_CHUNKID_BLOOMINDEXES
;
1599 chunks
[num_chunks
].size
= sizeof(uint32_t) * ctx
->commits
.nr
;
1601 chunks
[num_chunks
].id
= GRAPH_CHUNKID_BLOOMDATA
;
1602 chunks
[num_chunks
].size
= sizeof(uint32_t) * 3
1603 + ctx
->total_bloom_filter_data_size
;
1606 if (ctx
->num_commit_graphs_after
> 1) {
1607 chunks
[num_chunks
].id
= GRAPH_CHUNKID_BASE
;
1608 chunks
[num_chunks
].size
= hashsz
* (ctx
->num_commit_graphs_after
- 1);
1612 chunks
[num_chunks
].id
= 0;
1613 chunks
[num_chunks
].size
= 0;
1615 hashwrite_be32(f
, GRAPH_SIGNATURE
);
1617 hashwrite_u8(f
, GRAPH_VERSION
);
1618 hashwrite_u8(f
, oid_version());
1619 hashwrite_u8(f
, num_chunks
);
1620 hashwrite_u8(f
, ctx
->num_commit_graphs_after
- 1);
1622 chunk_offset
= 8 + (num_chunks
+ 1) * GRAPH_CHUNKLOOKUP_WIDTH
;
1623 for (i
= 0; i
<= num_chunks
; i
++) {
1624 uint32_t chunk_write
[3];
1626 chunk_write
[0] = htonl(chunks
[i
].id
);
1627 chunk_write
[1] = htonl(chunk_offset
>> 32);
1628 chunk_write
[2] = htonl(chunk_offset
& 0xffffffff);
1629 hashwrite(f
, chunk_write
, 12);
1631 chunk_offset
+= chunks
[i
].size
;
1634 if (ctx
->report_progress
) {
1635 strbuf_addf(&progress_title
,
1636 Q_("Writing out commit graph in %d pass",
1637 "Writing out commit graph in %d passes",
1640 ctx
->progress
= start_delayed_progress(
1642 num_chunks
* ctx
->commits
.nr
);
1644 write_graph_chunk_fanout(f
, ctx
);
1645 write_graph_chunk_oids(f
, hashsz
, ctx
);
1646 write_graph_chunk_data(f
, hashsz
, ctx
);
1647 if (ctx
->num_extra_edges
)
1648 write_graph_chunk_extra_edges(f
, ctx
);
1649 if (ctx
->changed_paths
) {
1650 write_graph_chunk_bloom_indexes(f
, ctx
);
1651 write_graph_chunk_bloom_data(f
, ctx
);
1653 if (ctx
->num_commit_graphs_after
> 1 &&
1654 write_graph_chunk_base(f
, ctx
)) {
1657 stop_progress(&ctx
->progress
);
1658 strbuf_release(&progress_title
);
1660 if (ctx
->split
&& ctx
->base_graph_name
&& ctx
->num_commit_graphs_after
> 1) {
1661 char *new_base_hash
= xstrdup(oid_to_hex(&ctx
->new_base_graph
->oid
));
1662 char *new_base_name
= get_split_graph_filename(ctx
->new_base_graph
->odb
, new_base_hash
);
1664 free(ctx
->commit_graph_filenames_after
[ctx
->num_commit_graphs_after
- 2]);
1665 free(ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 2]);
1666 ctx
->commit_graph_filenames_after
[ctx
->num_commit_graphs_after
- 2] = new_base_name
;
1667 ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 2] = new_base_hash
;
1670 close_commit_graph(ctx
->r
->objects
);
1671 finalize_hashfile(f
, file_hash
.hash
, CSUM_HASH_IN_STREAM
| CSUM_FSYNC
);
1674 FILE *chainf
= fdopen_lock_file(&lk
, "w");
1675 char *final_graph_name
;
1681 error(_("unable to open commit-graph chain file"));
1685 if (ctx
->base_graph_name
) {
1686 const char *dest
= ctx
->commit_graph_filenames_after
[
1687 ctx
->num_commit_graphs_after
- 2];
1689 if (strcmp(ctx
->base_graph_name
, dest
)) {
1690 result
= rename(ctx
->base_graph_name
, dest
);
1693 error(_("failed to rename base commit-graph file"));
1698 char *graph_name
= get_commit_graph_filename(ctx
->odb
);
1702 ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 1] = xstrdup(oid_to_hex(&file_hash
));
1703 final_graph_name
= get_split_graph_filename(ctx
->odb
,
1704 ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 1]);
1705 ctx
->commit_graph_filenames_after
[ctx
->num_commit_graphs_after
- 1] = final_graph_name
;
1707 result
= rename(ctx
->graph_name
, final_graph_name
);
1709 for (i
= 0; i
< ctx
->num_commit_graphs_after
; i
++)
1710 fprintf(lk
.tempfile
->fp
, "%s\n", ctx
->commit_graph_hash_after
[i
]);
1713 error(_("failed to rename temporary commit-graph file"));
1718 commit_lock_file(&lk
);
1723 static void split_graph_merge_strategy(struct write_commit_graph_context
*ctx
)
1725 struct commit_graph
*g
;
1726 uint32_t num_commits
;
1729 int max_commits
= 0;
1732 if (ctx
->split_opts
) {
1733 max_commits
= ctx
->split_opts
->max_commits
;
1735 if (ctx
->split_opts
->size_multiple
)
1736 size_mult
= ctx
->split_opts
->size_multiple
;
1739 g
= ctx
->r
->objects
->commit_graph
;
1740 num_commits
= ctx
->commits
.nr
;
1741 ctx
->num_commit_graphs_after
= ctx
->num_commit_graphs_before
+ 1;
1743 while (g
&& (g
->num_commits
<= size_mult
* num_commits
||
1744 (max_commits
&& num_commits
> max_commits
))) {
1745 if (g
->odb
!= ctx
->odb
)
1748 num_commits
+= g
->num_commits
;
1751 ctx
->num_commit_graphs_after
--;
1754 ctx
->new_base_graph
= g
;
1756 if (ctx
->num_commit_graphs_after
== 2) {
1757 char *old_graph_name
= get_commit_graph_filename(g
->odb
);
1759 if (!strcmp(g
->filename
, old_graph_name
) &&
1760 g
->odb
!= ctx
->odb
) {
1761 ctx
->num_commit_graphs_after
= 1;
1762 ctx
->new_base_graph
= NULL
;
1765 free(old_graph_name
);
1768 ALLOC_ARRAY(ctx
->commit_graph_filenames_after
, ctx
->num_commit_graphs_after
);
1769 ALLOC_ARRAY(ctx
->commit_graph_hash_after
, ctx
->num_commit_graphs_after
);
1771 for (i
= 0; i
< ctx
->num_commit_graphs_after
&&
1772 i
< ctx
->num_commit_graphs_before
; i
++)
1773 ctx
->commit_graph_filenames_after
[i
] = xstrdup(ctx
->commit_graph_filenames_before
[i
]);
1775 i
= ctx
->num_commit_graphs_before
- 1;
1776 g
= ctx
->r
->objects
->commit_graph
;
1779 if (i
< ctx
->num_commit_graphs_after
)
1780 ctx
->commit_graph_hash_after
[i
] = xstrdup(oid_to_hex(&g
->oid
));
1787 static void merge_commit_graph(struct write_commit_graph_context
*ctx
,
1788 struct commit_graph
*g
)
1791 uint32_t offset
= g
->num_commits_in_base
;
1793 ALLOC_GROW(ctx
->commits
.list
, ctx
->commits
.nr
+ g
->num_commits
, ctx
->commits
.alloc
);
1795 for (i
= 0; i
< g
->num_commits
; i
++) {
1796 struct object_id oid
;
1797 struct commit
*result
;
1799 display_progress(ctx
->progress
, i
+ 1);
1801 load_oid_from_graph(g
, i
+ offset
, &oid
);
1803 /* only add commits if they still exist in the repo */
1804 result
= lookup_commit_reference_gently(ctx
->r
, &oid
, 1);
1807 ctx
->commits
.list
[ctx
->commits
.nr
] = result
;
1813 static int commit_compare(const void *_a
, const void *_b
)
1815 const struct commit
*a
= *(const struct commit
**)_a
;
1816 const struct commit
*b
= *(const struct commit
**)_b
;
1817 return oidcmp(&a
->object
.oid
, &b
->object
.oid
);
1820 static void sort_and_scan_merged_commits(struct write_commit_graph_context
*ctx
)
1824 if (ctx
->report_progress
)
1825 ctx
->progress
= start_delayed_progress(
1826 _("Scanning merged commits"),
1829 QSORT(ctx
->commits
.list
, ctx
->commits
.nr
, commit_compare
);
1831 ctx
->num_extra_edges
= 0;
1832 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1833 display_progress(ctx
->progress
, i
);
1835 if (i
&& oideq(&ctx
->commits
.list
[i
- 1]->object
.oid
,
1836 &ctx
->commits
.list
[i
]->object
.oid
)) {
1837 die(_("unexpected duplicate commit id %s"),
1838 oid_to_hex(&ctx
->commits
.list
[i
]->object
.oid
));
1840 unsigned int num_parents
;
1842 num_parents
= commit_list_count(ctx
->commits
.list
[i
]->parents
);
1843 if (num_parents
> 2)
1844 ctx
->num_extra_edges
+= num_parents
- 1;
1848 stop_progress(&ctx
->progress
);
1851 static void merge_commit_graphs(struct write_commit_graph_context
*ctx
)
1853 struct commit_graph
*g
= ctx
->r
->objects
->commit_graph
;
1854 uint32_t current_graph_number
= ctx
->num_commit_graphs_before
;
1856 while (g
&& current_graph_number
>= ctx
->num_commit_graphs_after
) {
1857 current_graph_number
--;
1859 if (ctx
->report_progress
)
1860 ctx
->progress
= start_delayed_progress(_("Merging commit-graph"), 0);
1862 merge_commit_graph(ctx
, g
);
1863 stop_progress(&ctx
->progress
);
1869 ctx
->new_base_graph
= g
;
1870 ctx
->new_num_commits_in_base
= g
->num_commits
+ g
->num_commits_in_base
;
1873 if (ctx
->new_base_graph
)
1874 ctx
->base_graph_name
= xstrdup(ctx
->new_base_graph
->filename
);
1876 sort_and_scan_merged_commits(ctx
);
1879 static void mark_commit_graphs(struct write_commit_graph_context
*ctx
)
1882 time_t now
= time(NULL
);
1884 for (i
= ctx
->num_commit_graphs_after
- 1; i
< ctx
->num_commit_graphs_before
; i
++) {
1886 struct utimbuf updated_time
;
1888 stat(ctx
->commit_graph_filenames_before
[i
], &st
);
1890 updated_time
.actime
= st
.st_atime
;
1891 updated_time
.modtime
= now
;
1892 utime(ctx
->commit_graph_filenames_before
[i
], &updated_time
);
1896 static void expire_commit_graphs(struct write_commit_graph_context
*ctx
)
1898 struct strbuf path
= STRBUF_INIT
;
1902 timestamp_t expire_time
= time(NULL
);
1904 if (ctx
->split_opts
&& ctx
->split_opts
->expire_time
)
1905 expire_time
-= ctx
->split_opts
->expire_time
;
1907 char *chain_file_name
= get_chain_filename(ctx
->odb
);
1908 unlink(chain_file_name
);
1909 free(chain_file_name
);
1910 ctx
->num_commit_graphs_after
= 0;
1913 strbuf_addstr(&path
, ctx
->odb
->path
);
1914 strbuf_addstr(&path
, "/info/commit-graphs");
1915 dir
= opendir(path
.buf
);
1920 strbuf_addch(&path
, '/');
1921 dirnamelen
= path
.len
;
1922 while ((de
= readdir(dir
)) != NULL
) {
1924 uint32_t i
, found
= 0;
1926 strbuf_setlen(&path
, dirnamelen
);
1927 strbuf_addstr(&path
, de
->d_name
);
1929 stat(path
.buf
, &st
);
1931 if (st
.st_mtime
> expire_time
)
1933 if (path
.len
< 6 || strcmp(path
.buf
+ path
.len
- 6, ".graph"))
1936 for (i
= 0; i
< ctx
->num_commit_graphs_after
; i
++) {
1937 if (!strcmp(ctx
->commit_graph_filenames_after
[i
],
1949 strbuf_release(&path
);
1952 int write_commit_graph(struct object_directory
*odb
,
1953 struct string_list
*pack_indexes
,
1954 struct string_list
*commit_hex
,
1955 enum commit_graph_write_flags flags
,
1956 const struct split_commit_graph_opts
*split_opts
)
1958 struct write_commit_graph_context
*ctx
;
1959 uint32_t i
, count_distinct
= 0;
1962 if (!commit_graph_compatible(the_repository
))
1965 ctx
= xcalloc(1, sizeof(struct write_commit_graph_context
));
1966 ctx
->r
= the_repository
;
1968 ctx
->append
= flags
& COMMIT_GRAPH_WRITE_APPEND
? 1 : 0;
1969 ctx
->report_progress
= flags
& COMMIT_GRAPH_WRITE_PROGRESS
? 1 : 0;
1970 ctx
->split
= flags
& COMMIT_GRAPH_WRITE_SPLIT
? 1 : 0;
1971 ctx
->check_oids
= flags
& COMMIT_GRAPH_WRITE_CHECK_OIDS
? 1 : 0;
1972 ctx
->split_opts
= split_opts
;
1973 ctx
->changed_paths
= flags
& COMMIT_GRAPH_WRITE_BLOOM_FILTERS
? 1 : 0;
1974 ctx
->total_bloom_filter_data_size
= 0;
1977 struct commit_graph
*g
;
1978 prepare_commit_graph(ctx
->r
);
1980 g
= ctx
->r
->objects
->commit_graph
;
1983 ctx
->num_commit_graphs_before
++;
1987 if (ctx
->num_commit_graphs_before
) {
1988 ALLOC_ARRAY(ctx
->commit_graph_filenames_before
, ctx
->num_commit_graphs_before
);
1989 i
= ctx
->num_commit_graphs_before
;
1990 g
= ctx
->r
->objects
->commit_graph
;
1993 ctx
->commit_graph_filenames_before
[--i
] = xstrdup(g
->filename
);
1999 ctx
->approx_nr_objects
= approximate_object_count();
2000 ctx
->oids
.alloc
= ctx
->approx_nr_objects
/ 32;
2002 if (ctx
->split
&& split_opts
&& ctx
->oids
.alloc
> split_opts
->max_commits
)
2003 ctx
->oids
.alloc
= split_opts
->max_commits
;
2006 prepare_commit_graph_one(ctx
->r
, ctx
->odb
);
2007 if (ctx
->r
->objects
->commit_graph
)
2008 ctx
->oids
.alloc
+= ctx
->r
->objects
->commit_graph
->num_commits
;
2011 if (ctx
->oids
.alloc
< 1024)
2012 ctx
->oids
.alloc
= 1024;
2013 ALLOC_ARRAY(ctx
->oids
.list
, ctx
->oids
.alloc
);
2015 if (ctx
->append
&& ctx
->r
->objects
->commit_graph
) {
2016 struct commit_graph
*g
= ctx
->r
->objects
->commit_graph
;
2017 for (i
= 0; i
< g
->num_commits
; i
++) {
2018 const unsigned char *hash
= g
->chunk_oid_lookup
+ g
->hash_len
* i
;
2019 hashcpy(ctx
->oids
.list
[ctx
->oids
.nr
++].hash
, hash
);
2024 ctx
->order_by_pack
= 1;
2025 if ((res
= fill_oids_from_packs(ctx
, pack_indexes
)))
2030 if ((res
= fill_oids_from_commit_hex(ctx
, commit_hex
)))
2034 if (!pack_indexes
&& !commit_hex
) {
2035 ctx
->order_by_pack
= 1;
2036 fill_oids_from_all_packs(ctx
);
2039 close_reachable(ctx
);
2041 count_distinct
= count_distinct_commits(ctx
);
2043 if (count_distinct
>= GRAPH_EDGE_LAST_MASK
) {
2044 error(_("the commit graph format cannot write %d commits"), count_distinct
);
2049 ctx
->commits
.alloc
= count_distinct
;
2050 ALLOC_ARRAY(ctx
->commits
.list
, ctx
->commits
.alloc
);
2052 copy_oids_to_commits(ctx
);
2054 if (ctx
->commits
.nr
>= GRAPH_EDGE_LAST_MASK
) {
2055 error(_("too many commits to write graph"));
2060 if (!ctx
->commits
.nr
)
2064 split_graph_merge_strategy(ctx
);
2066 merge_commit_graphs(ctx
);
2068 ctx
->num_commit_graphs_after
= 1;
2070 compute_generation_numbers(ctx
);
2072 if (ctx
->changed_paths
)
2073 compute_bloom_filters(ctx
);
2075 res
= write_commit_graph_file(ctx
);
2078 mark_commit_graphs(ctx
);
2080 expire_commit_graphs(ctx
);
2083 free(ctx
->graph_name
);
2084 free(ctx
->commits
.list
);
2085 free(ctx
->oids
.list
);
2087 if (ctx
->commit_graph_filenames_after
) {
2088 for (i
= 0; i
< ctx
->num_commit_graphs_after
; i
++) {
2089 free(ctx
->commit_graph_filenames_after
[i
]);
2090 free(ctx
->commit_graph_hash_after
[i
]);
2093 for (i
= 0; i
< ctx
->num_commit_graphs_before
; i
++)
2094 free(ctx
->commit_graph_filenames_before
[i
]);
2096 free(ctx
->commit_graph_filenames_after
);
2097 free(ctx
->commit_graph_filenames_before
);
2098 free(ctx
->commit_graph_hash_after
);
2106 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
2107 static int verify_commit_graph_error
;
2109 static void graph_report(const char *fmt
, ...)
2113 verify_commit_graph_error
= 1;
2115 vfprintf(stderr
, fmt
, ap
);
2116 fprintf(stderr
, "\n");
2120 #define GENERATION_ZERO_EXISTS 1
2121 #define GENERATION_NUMBER_EXISTS 2
2123 int verify_commit_graph(struct repository
*r
, struct commit_graph
*g
, int flags
)
2125 uint32_t i
, cur_fanout_pos
= 0;
2126 struct object_id prev_oid
, cur_oid
, checksum
;
2127 int generation_zero
= 0;
2130 struct progress
*progress
= NULL
;
2131 int local_error
= 0;
2134 graph_report("no commit-graph file loaded");
2138 verify_commit_graph_error
= verify_commit_graph_lite(g
);
2139 if (verify_commit_graph_error
)
2140 return verify_commit_graph_error
;
2142 devnull
= open("/dev/null", O_WRONLY
);
2143 f
= hashfd(devnull
, NULL
);
2144 hashwrite(f
, g
->data
, g
->data_len
- g
->hash_len
);
2145 finalize_hashfile(f
, checksum
.hash
, CSUM_CLOSE
);
2146 if (!hasheq(checksum
.hash
, g
->data
+ g
->data_len
- g
->hash_len
)) {
2147 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
2148 verify_commit_graph_error
= VERIFY_COMMIT_GRAPH_ERROR_HASH
;
2151 for (i
= 0; i
< g
->num_commits
; i
++) {
2152 struct commit
*graph_commit
;
2154 hashcpy(cur_oid
.hash
, g
->chunk_oid_lookup
+ g
->hash_len
* i
);
2156 if (i
&& oidcmp(&prev_oid
, &cur_oid
) >= 0)
2157 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
2158 oid_to_hex(&prev_oid
),
2159 oid_to_hex(&cur_oid
));
2161 oidcpy(&prev_oid
, &cur_oid
);
2163 while (cur_oid
.hash
[0] > cur_fanout_pos
) {
2164 uint32_t fanout_value
= get_be32(g
->chunk_oid_fanout
+ cur_fanout_pos
);
2166 if (i
!= fanout_value
)
2167 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2168 cur_fanout_pos
, fanout_value
, i
);
2172 graph_commit
= lookup_commit(r
, &cur_oid
);
2173 if (!parse_commit_in_graph_one(r
, g
, graph_commit
))
2174 graph_report(_("failed to parse commit %s from commit-graph"),
2175 oid_to_hex(&cur_oid
));
2178 while (cur_fanout_pos
< 256) {
2179 uint32_t fanout_value
= get_be32(g
->chunk_oid_fanout
+ cur_fanout_pos
);
2181 if (g
->num_commits
!= fanout_value
)
2182 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2183 cur_fanout_pos
, fanout_value
, i
);
2188 if (verify_commit_graph_error
& ~VERIFY_COMMIT_GRAPH_ERROR_HASH
)
2189 return verify_commit_graph_error
;
2191 if (flags
& COMMIT_GRAPH_WRITE_PROGRESS
)
2192 progress
= start_progress(_("Verifying commits in commit graph"),
2195 for (i
= 0; i
< g
->num_commits
; i
++) {
2196 struct commit
*graph_commit
, *odb_commit
;
2197 struct commit_list
*graph_parents
, *odb_parents
;
2198 uint32_t max_generation
= 0;
2200 display_progress(progress
, i
+ 1);
2201 hashcpy(cur_oid
.hash
, g
->chunk_oid_lookup
+ g
->hash_len
* i
);
2203 graph_commit
= lookup_commit(r
, &cur_oid
);
2204 odb_commit
= (struct commit
*)create_object(r
, &cur_oid
, alloc_commit_node(r
));
2205 if (parse_commit_internal(odb_commit
, 0, 0)) {
2206 graph_report(_("failed to parse commit %s from object database for commit-graph"),
2207 oid_to_hex(&cur_oid
));
2211 if (!oideq(&get_commit_tree_in_graph_one(r
, g
, graph_commit
)->object
.oid
,
2212 get_commit_tree_oid(odb_commit
)))
2213 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2214 oid_to_hex(&cur_oid
),
2215 oid_to_hex(get_commit_tree_oid(graph_commit
)),
2216 oid_to_hex(get_commit_tree_oid(odb_commit
)));
2218 graph_parents
= graph_commit
->parents
;
2219 odb_parents
= odb_commit
->parents
;
2221 while (graph_parents
) {
2222 if (odb_parents
== NULL
) {
2223 graph_report(_("commit-graph parent list for commit %s is too long"),
2224 oid_to_hex(&cur_oid
));
2228 /* parse parent in case it is in a base graph */
2229 parse_commit_in_graph_one(r
, g
, graph_parents
->item
);
2231 if (!oideq(&graph_parents
->item
->object
.oid
, &odb_parents
->item
->object
.oid
))
2232 graph_report(_("commit-graph parent for %s is %s != %s"),
2233 oid_to_hex(&cur_oid
),
2234 oid_to_hex(&graph_parents
->item
->object
.oid
),
2235 oid_to_hex(&odb_parents
->item
->object
.oid
));
2237 if (graph_parents
->item
->generation
> max_generation
)
2238 max_generation
= graph_parents
->item
->generation
;
2240 graph_parents
= graph_parents
->next
;
2241 odb_parents
= odb_parents
->next
;
2244 if (odb_parents
!= NULL
)
2245 graph_report(_("commit-graph parent list for commit %s terminates early"),
2246 oid_to_hex(&cur_oid
));
2248 if (!graph_commit
->generation
) {
2249 if (generation_zero
== GENERATION_NUMBER_EXISTS
)
2250 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2251 oid_to_hex(&cur_oid
));
2252 generation_zero
= GENERATION_ZERO_EXISTS
;
2253 } else if (generation_zero
== GENERATION_ZERO_EXISTS
)
2254 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2255 oid_to_hex(&cur_oid
));
2257 if (generation_zero
== GENERATION_ZERO_EXISTS
)
2261 * If one of our parents has generation GENERATION_NUMBER_MAX, then
2262 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
2263 * extra logic in the following condition.
2265 if (max_generation
== GENERATION_NUMBER_MAX
)
2268 if (graph_commit
->generation
!= max_generation
+ 1)
2269 graph_report(_("commit-graph generation for commit %s is %u != %u"),
2270 oid_to_hex(&cur_oid
),
2271 graph_commit
->generation
,
2272 max_generation
+ 1);
2274 if (graph_commit
->date
!= odb_commit
->date
)
2275 graph_report(_("commit date for commit %s in commit-graph is %"PRItime
" != %"PRItime
),
2276 oid_to_hex(&cur_oid
),
2280 stop_progress(&progress
);
2282 local_error
= verify_commit_graph_error
;
2284 if (!(flags
& COMMIT_GRAPH_VERIFY_SHALLOW
) && g
->base_graph
)
2285 local_error
|= verify_commit_graph(r
, g
->base_graph
, flags
);
2290 void free_commit_graph(struct commit_graph
*g
)
2294 if (g
->graph_fd
>= 0) {
2295 munmap((void *)g
->data
, g
->data_len
);
2300 free(g
->bloom_filter_settings
);
2304 void disable_commit_graph(struct repository
*r
)
2306 r
->commit_graph_disabled
= 1;