1 #include "git-compat-util.h"
10 #include "hash-lookup.h"
11 #include "commit-graph.h"
12 #include "object-store.h"
15 #include "replace-object.h"
18 #include "commit-slab.h"
20 #include "json-writer.h"
22 #include "chunk-format.h"
24 void git_test_write_commit_graph_or_die(void)
27 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH
, 0))
30 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS
, 0))
31 flags
= COMMIT_GRAPH_WRITE_BLOOM_FILTERS
;
33 if (write_commit_graph_reachable(the_repository
->objects
->odb
,
35 die("failed to write commit-graph under GIT_TEST_COMMIT_GRAPH");
38 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
39 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
40 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
41 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
42 #define GRAPH_CHUNKID_GENERATION_DATA 0x47444132 /* "GDA2" */
43 #define GRAPH_CHUNKID_GENERATION_DATA_OVERFLOW 0x47444f32 /* "GDO2" */
44 #define GRAPH_CHUNKID_EXTRAEDGES 0x45444745 /* "EDGE" */
45 #define GRAPH_CHUNKID_BLOOMINDEXES 0x42494458 /* "BIDX" */
46 #define GRAPH_CHUNKID_BLOOMDATA 0x42444154 /* "BDAT" */
47 #define GRAPH_CHUNKID_BASE 0x42415345 /* "BASE" */
49 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
51 #define GRAPH_VERSION_1 0x1
52 #define GRAPH_VERSION GRAPH_VERSION_1
54 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
55 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
56 #define GRAPH_PARENT_NONE 0x70000000
58 #define GRAPH_LAST_EDGE 0x80000000
60 #define GRAPH_HEADER_SIZE 8
61 #define GRAPH_FANOUT_SIZE (4 * 256)
62 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * CHUNK_TOC_ENTRY_SIZE \
63 + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
65 #define CORRECTED_COMMIT_DATE_OFFSET_OVERFLOW (1ULL << 31)
67 /* Remember to update object flag allocation in object.h */
68 #define REACHABLE (1u<<15)
70 define_commit_slab(topo_level_slab
, uint32_t);
72 /* Keep track of the order in which commits are added to our list. */
73 define_commit_slab(commit_pos
, int);
74 static struct commit_pos commit_pos
= COMMIT_SLAB_INIT(1, commit_pos
);
76 static void set_commit_pos(struct repository
*r
, const struct object_id
*oid
)
78 static int32_t max_pos
;
79 struct commit
*commit
= lookup_commit(r
, oid
);
82 return; /* should never happen, but be lenient */
84 *commit_pos_at(&commit_pos
, commit
) = max_pos
++;
87 static int commit_pos_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
;
91 return commit_pos_at(&commit_pos
, a
) -
92 commit_pos_at(&commit_pos
, b
);
95 define_commit_slab(commit_graph_data_slab
, struct commit_graph_data
);
96 static struct commit_graph_data_slab commit_graph_data_slab
=
97 COMMIT_SLAB_INIT(1, commit_graph_data_slab
);
99 static int get_configured_generation_version(struct repository
*r
)
102 repo_config_get_int(r
, "commitgraph.generationversion", &version
);
106 uint32_t commit_graph_position(const struct commit
*c
)
108 struct commit_graph_data
*data
=
109 commit_graph_data_slab_peek(&commit_graph_data_slab
, c
);
111 return data
? data
->graph_pos
: COMMIT_NOT_FROM_GRAPH
;
114 timestamp_t
commit_graph_generation(const struct commit
*c
)
116 struct commit_graph_data
*data
=
117 commit_graph_data_slab_peek(&commit_graph_data_slab
, c
);
120 return GENERATION_NUMBER_INFINITY
;
121 else if (data
->graph_pos
== COMMIT_NOT_FROM_GRAPH
)
122 return GENERATION_NUMBER_INFINITY
;
124 return data
->generation
;
127 static struct commit_graph_data
*commit_graph_data_at(const struct commit
*c
)
129 unsigned int i
, nth_slab
;
130 struct commit_graph_data
*data
=
131 commit_graph_data_slab_peek(&commit_graph_data_slab
, c
);
136 nth_slab
= c
->index
/ commit_graph_data_slab
.slab_size
;
137 data
= commit_graph_data_slab_at(&commit_graph_data_slab
, c
);
140 * commit-slab initializes elements with zero, overwrite this with
141 * COMMIT_NOT_FROM_GRAPH for graph_pos.
143 * We avoid initializing generation with checking if graph position
144 * is not COMMIT_NOT_FROM_GRAPH.
146 for (i
= 0; i
< commit_graph_data_slab
.slab_size
; i
++) {
147 commit_graph_data_slab
.slab
[nth_slab
][i
].graph_pos
=
148 COMMIT_NOT_FROM_GRAPH
;
155 * Should be used only while writing commit-graph as it compares
156 * generation value of commits by directly accessing commit-slab.
158 static int commit_gen_cmp(const void *va
, const void *vb
)
160 const struct commit
*a
= *(const struct commit
**)va
;
161 const struct commit
*b
= *(const struct commit
**)vb
;
163 const timestamp_t generation_a
= commit_graph_data_at(a
)->generation
;
164 const timestamp_t generation_b
= commit_graph_data_at(b
)->generation
;
165 /* lower generation commits first */
166 if (generation_a
< generation_b
)
168 else if (generation_a
> generation_b
)
171 /* use date as a heuristic when generations are equal */
172 if (a
->date
< b
->date
)
174 else if (a
->date
> b
->date
)
179 char *get_commit_graph_filename(struct object_directory
*obj_dir
)
181 return xstrfmt("%s/info/commit-graph", obj_dir
->path
);
184 static char *get_split_graph_filename(struct object_directory
*odb
,
187 return xstrfmt("%s/info/commit-graphs/graph-%s.graph", odb
->path
,
191 char *get_commit_graph_chain_filename(struct object_directory
*odb
)
193 return xstrfmt("%s/info/commit-graphs/commit-graph-chain", odb
->path
);
196 static struct commit_graph
*alloc_commit_graph(void)
198 struct commit_graph
*g
= xcalloc(1, sizeof(*g
));
203 extern int read_replace_refs
;
205 static int commit_graph_compatible(struct repository
*r
)
210 if (read_replace_refs
) {
211 prepare_replace_object(r
);
212 if (hashmap_get_size(&r
->objects
->replace_map
->map
))
216 prepare_commit_graft(r
);
217 if (r
->parsed_objects
&&
218 (r
->parsed_objects
->grafts_nr
|| r
->parsed_objects
->substituted_parent
))
220 if (is_repository_shallow(r
))
226 int open_commit_graph(const char *graph_file
, int *fd
, struct stat
*st
)
228 *fd
= git_open(graph_file
);
231 if (fstat(*fd
, st
)) {
238 struct commit_graph
*load_commit_graph_one_fd_st(struct repository
*r
,
239 int fd
, struct stat
*st
,
240 struct object_directory
*odb
)
244 struct commit_graph
*ret
;
246 graph_size
= xsize_t(st
->st_size
);
248 if (graph_size
< GRAPH_MIN_SIZE
) {
250 error(_("commit-graph file is too small"));
253 graph_map
= xmmap(NULL
, graph_size
, PROT_READ
, MAP_PRIVATE
, fd
, 0);
255 prepare_repo_settings(r
);
256 ret
= parse_commit_graph(&r
->settings
, graph_map
, graph_size
);
261 munmap(graph_map
, graph_size
);
266 static int verify_commit_graph_lite(struct commit_graph
*g
)
269 * Basic validation shared between parse_commit_graph()
270 * which'll be called every time the graph is used, and the
271 * much more expensive verify_commit_graph() used by
272 * "commit-graph verify".
274 * There should only be very basic checks here to ensure that
275 * we don't e.g. segfault in fill_commit_in_graph(), but
276 * because this is a very hot codepath nothing that e.g. loops
277 * over g->num_commits, or runs a checksum on the commit-graph
280 if (!g
->chunk_oid_fanout
) {
281 error("commit-graph is missing the OID Fanout chunk");
284 if (!g
->chunk_oid_lookup
) {
285 error("commit-graph is missing the OID Lookup chunk");
288 if (!g
->chunk_commit_data
) {
289 error("commit-graph is missing the Commit Data chunk");
296 static int graph_read_oid_lookup(const unsigned char *chunk_start
,
297 size_t chunk_size
, void *data
)
299 struct commit_graph
*g
= data
;
300 g
->chunk_oid_lookup
= chunk_start
;
301 g
->num_commits
= chunk_size
/ g
->hash_len
;
305 static int graph_read_bloom_data(const unsigned char *chunk_start
,
306 size_t chunk_size
, void *data
)
308 struct commit_graph
*g
= data
;
309 uint32_t hash_version
;
310 g
->chunk_bloom_data
= chunk_start
;
311 hash_version
= get_be32(chunk_start
);
313 if (hash_version
!= 1)
316 g
->bloom_filter_settings
= xmalloc(sizeof(struct bloom_filter_settings
));
317 g
->bloom_filter_settings
->hash_version
= hash_version
;
318 g
->bloom_filter_settings
->num_hashes
= get_be32(chunk_start
+ 4);
319 g
->bloom_filter_settings
->bits_per_entry
= get_be32(chunk_start
+ 8);
320 g
->bloom_filter_settings
->max_changed_paths
= DEFAULT_BLOOM_MAX_CHANGES
;
325 struct commit_graph
*parse_commit_graph(struct repo_settings
*s
,
326 void *graph_map
, size_t graph_size
)
328 const unsigned char *data
;
329 struct commit_graph
*graph
;
330 uint32_t graph_signature
;
331 unsigned char graph_version
, hash_version
;
332 struct chunkfile
*cf
= NULL
;
337 if (graph_size
< GRAPH_MIN_SIZE
)
340 data
= (const unsigned char *)graph_map
;
342 graph_signature
= get_be32(data
);
343 if (graph_signature
!= GRAPH_SIGNATURE
) {
344 error(_("commit-graph signature %X does not match signature %X"),
345 graph_signature
, GRAPH_SIGNATURE
);
349 graph_version
= *(unsigned char*)(data
+ 4);
350 if (graph_version
!= GRAPH_VERSION
) {
351 error(_("commit-graph version %X does not match version %X"),
352 graph_version
, GRAPH_VERSION
);
356 hash_version
= *(unsigned char*)(data
+ 5);
357 if (hash_version
!= oid_version(the_hash_algo
)) {
358 error(_("commit-graph hash version %X does not match version %X"),
359 hash_version
, oid_version(the_hash_algo
));
363 graph
= alloc_commit_graph();
365 graph
->hash_len
= the_hash_algo
->rawsz
;
366 graph
->num_chunks
= *(unsigned char*)(data
+ 6);
367 graph
->data
= graph_map
;
368 graph
->data_len
= graph_size
;
370 if (graph_size
< GRAPH_HEADER_SIZE
+
371 (graph
->num_chunks
+ 1) * CHUNK_TOC_ENTRY_SIZE
+
372 GRAPH_FANOUT_SIZE
+ the_hash_algo
->rawsz
) {
373 error(_("commit-graph file is too small to hold %u chunks"),
379 cf
= init_chunkfile(NULL
);
381 if (read_table_of_contents(cf
, graph
->data
, graph_size
,
382 GRAPH_HEADER_SIZE
, graph
->num_chunks
))
383 goto free_and_return
;
385 pair_chunk(cf
, GRAPH_CHUNKID_OIDFANOUT
,
386 (const unsigned char **)&graph
->chunk_oid_fanout
);
387 read_chunk(cf
, GRAPH_CHUNKID_OIDLOOKUP
, graph_read_oid_lookup
, graph
);
388 pair_chunk(cf
, GRAPH_CHUNKID_DATA
, &graph
->chunk_commit_data
);
389 pair_chunk(cf
, GRAPH_CHUNKID_EXTRAEDGES
, &graph
->chunk_extra_edges
);
390 pair_chunk(cf
, GRAPH_CHUNKID_BASE
, &graph
->chunk_base_graphs
);
392 if (s
->commit_graph_generation_version
>= 2) {
393 pair_chunk(cf
, GRAPH_CHUNKID_GENERATION_DATA
,
394 &graph
->chunk_generation_data
);
395 pair_chunk(cf
, GRAPH_CHUNKID_GENERATION_DATA_OVERFLOW
,
396 &graph
->chunk_generation_data_overflow
);
398 if (graph
->chunk_generation_data
)
399 graph
->read_generation_data
= 1;
402 if (s
->commit_graph_read_changed_paths
) {
403 pair_chunk(cf
, GRAPH_CHUNKID_BLOOMINDEXES
,
404 &graph
->chunk_bloom_indexes
);
405 read_chunk(cf
, GRAPH_CHUNKID_BLOOMDATA
,
406 graph_read_bloom_data
, graph
);
409 if (graph
->chunk_bloom_indexes
&& graph
->chunk_bloom_data
) {
410 init_bloom_filters();
412 /* We need both the bloom chunks to exist together. Else ignore the data */
413 graph
->chunk_bloom_indexes
= NULL
;
414 graph
->chunk_bloom_data
= NULL
;
415 FREE_AND_NULL(graph
->bloom_filter_settings
);
418 oidread(&graph
->oid
, graph
->data
+ graph
->data_len
- graph
->hash_len
);
420 if (verify_commit_graph_lite(graph
))
421 goto free_and_return
;
428 free(graph
->bloom_filter_settings
);
433 static struct commit_graph
*load_commit_graph_one(struct repository
*r
,
434 const char *graph_file
,
435 struct object_directory
*odb
)
440 struct commit_graph
*g
;
441 int open_ok
= open_commit_graph(graph_file
, &fd
, &st
);
446 g
= load_commit_graph_one_fd_st(r
, fd
, &st
, odb
);
449 g
->filename
= xstrdup(graph_file
);
454 static struct commit_graph
*load_commit_graph_v1(struct repository
*r
,
455 struct object_directory
*odb
)
457 char *graph_name
= get_commit_graph_filename(odb
);
458 struct commit_graph
*g
= load_commit_graph_one(r
, graph_name
, odb
);
464 static int add_graph_to_chain(struct commit_graph
*g
,
465 struct commit_graph
*chain
,
466 struct object_id
*oids
,
469 struct commit_graph
*cur_g
= chain
;
471 if (n
&& !g
->chunk_base_graphs
) {
472 warning(_("commit-graph has no base graphs chunk"));
480 !oideq(&oids
[n
], &cur_g
->oid
) ||
481 !hasheq(oids
[n
].hash
, g
->chunk_base_graphs
+ g
->hash_len
* n
)) {
482 warning(_("commit-graph chain does not match"));
486 cur_g
= cur_g
->base_graph
;
489 g
->base_graph
= chain
;
492 g
->num_commits_in_base
= chain
->num_commits
+ chain
->num_commits_in_base
;
497 static struct commit_graph
*load_commit_graph_chain(struct repository
*r
,
498 struct object_directory
*odb
)
500 struct commit_graph
*graph_chain
= NULL
;
501 struct strbuf line
= STRBUF_INIT
;
503 struct object_id
*oids
;
504 int i
= 0, valid
= 1, count
;
505 char *chain_name
= get_commit_graph_chain_filename(odb
);
509 fp
= fopen(chain_name
, "r");
510 stat_res
= stat(chain_name
, &st
);
516 st
.st_size
<= the_hash_algo
->hexsz
) {
521 count
= st
.st_size
/ (the_hash_algo
->hexsz
+ 1);
522 CALLOC_ARRAY(oids
, count
);
526 for (i
= 0; i
< count
; i
++) {
527 struct object_directory
*odb
;
529 if (strbuf_getline_lf(&line
, fp
) == EOF
)
532 if (get_oid_hex(line
.buf
, &oids
[i
])) {
533 warning(_("invalid commit-graph chain: line '%s' not a hash"),
540 for (odb
= r
->objects
->odb
; odb
; odb
= odb
->next
) {
541 char *graph_name
= get_split_graph_filename(odb
, line
.buf
);
542 struct commit_graph
*g
= load_commit_graph_one(r
, graph_name
, odb
);
547 if (add_graph_to_chain(g
, graph_chain
, oids
, i
)) {
557 warning(_("unable to find all commit-graph files"));
564 strbuf_release(&line
);
570 * returns 1 if and only if all graphs in the chain have
571 * corrected commit dates stored in the generation_data chunk.
573 static int validate_mixed_generation_chain(struct commit_graph
*g
)
575 int read_generation_data
= 1;
576 struct commit_graph
*p
= g
;
578 while (read_generation_data
&& p
) {
579 read_generation_data
= p
->read_generation_data
;
583 if (read_generation_data
)
587 g
->read_generation_data
= 0;
594 struct commit_graph
*read_commit_graph_one(struct repository
*r
,
595 struct object_directory
*odb
)
597 struct commit_graph
*g
= load_commit_graph_v1(r
, odb
);
600 g
= load_commit_graph_chain(r
, odb
);
602 validate_mixed_generation_chain(g
);
607 static void prepare_commit_graph_one(struct repository
*r
,
608 struct object_directory
*odb
)
611 if (r
->objects
->commit_graph
)
614 r
->objects
->commit_graph
= read_commit_graph_one(r
, odb
);
618 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
620 * On the first invocation, this function attempts to load the commit
621 * graph if the_repository is configured to have one.
623 static int prepare_commit_graph(struct repository
*r
)
625 struct object_directory
*odb
;
628 * Early return if there is no git dir or if the commit graph is
631 * This must come before the "already attempted?" check below, because
632 * we want to disable even an already-loaded graph file.
634 if (!r
->gitdir
|| r
->commit_graph_disabled
)
637 if (r
->objects
->commit_graph_attempted
)
638 return !!r
->objects
->commit_graph
;
639 r
->objects
->commit_graph_attempted
= 1;
641 prepare_repo_settings(r
);
643 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH
, 0) &&
644 r
->settings
.core_commit_graph
!= 1)
646 * This repository is not configured to use commit graphs, so
647 * do not load one. (But report commit_graph_attempted anyway
648 * so that commit graph loading is not attempted again for this
653 if (!commit_graph_compatible(r
))
657 for (odb
= r
->objects
->odb
;
658 !r
->objects
->commit_graph
&& odb
;
660 prepare_commit_graph_one(r
, odb
);
661 return !!r
->objects
->commit_graph
;
664 int generation_numbers_enabled(struct repository
*r
)
666 uint32_t first_generation
;
667 struct commit_graph
*g
;
668 if (!prepare_commit_graph(r
))
671 g
= r
->objects
->commit_graph
;
676 first_generation
= get_be32(g
->chunk_commit_data
+
677 g
->hash_len
+ 8) >> 2;
679 return !!first_generation
;
682 int corrected_commit_dates_enabled(struct repository
*r
)
684 struct commit_graph
*g
;
685 if (!prepare_commit_graph(r
))
688 g
= r
->objects
->commit_graph
;
693 return g
->read_generation_data
;
696 struct bloom_filter_settings
*get_bloom_filter_settings(struct repository
*r
)
698 struct commit_graph
*g
= r
->objects
->commit_graph
;
700 if (g
->bloom_filter_settings
)
701 return g
->bloom_filter_settings
;
707 static void close_commit_graph_one(struct commit_graph
*g
)
712 clear_commit_graph_data_slab(&commit_graph_data_slab
);
713 close_commit_graph_one(g
->base_graph
);
714 free_commit_graph(g
);
717 void close_commit_graph(struct raw_object_store
*o
)
719 close_commit_graph_one(o
->commit_graph
);
720 o
->commit_graph
= NULL
;
723 static int bsearch_graph(struct commit_graph
*g
, const struct object_id
*oid
, uint32_t *pos
)
725 return bsearch_hash(oid
->hash
, g
->chunk_oid_fanout
,
726 g
->chunk_oid_lookup
, g
->hash_len
, pos
);
729 static void load_oid_from_graph(struct commit_graph
*g
,
731 struct object_id
*oid
)
735 while (g
&& pos
< g
->num_commits_in_base
)
739 BUG("NULL commit-graph");
741 if (pos
>= g
->num_commits
+ g
->num_commits_in_base
)
742 die(_("invalid commit position. commit-graph is likely corrupt"));
744 lex_index
= pos
- g
->num_commits_in_base
;
746 oidread(oid
, g
->chunk_oid_lookup
+ g
->hash_len
* lex_index
);
749 static struct commit_list
**insert_parent_or_die(struct repository
*r
,
750 struct commit_graph
*g
,
752 struct commit_list
**pptr
)
755 struct object_id oid
;
757 if (pos
>= g
->num_commits
+ g
->num_commits_in_base
)
758 die("invalid parent position %"PRIu32
, pos
);
760 load_oid_from_graph(g
, pos
, &oid
);
761 c
= lookup_commit(r
, &oid
);
763 die(_("could not find commit %s"), oid_to_hex(&oid
));
764 commit_graph_data_at(c
)->graph_pos
= pos
;
765 return &commit_list_insert(c
, pptr
)->next
;
768 static void fill_commit_graph_info(struct commit
*item
, struct commit_graph
*g
, uint32_t pos
)
770 const unsigned char *commit_data
;
771 struct commit_graph_data
*graph_data
;
772 uint32_t lex_index
, offset_pos
;
773 uint64_t date_high
, date_low
, offset
;
775 while (pos
< g
->num_commits_in_base
)
778 if (pos
>= g
->num_commits
+ g
->num_commits_in_base
)
779 die(_("invalid commit position. commit-graph is likely corrupt"));
781 lex_index
= pos
- g
->num_commits_in_base
;
782 commit_data
= g
->chunk_commit_data
+ GRAPH_DATA_WIDTH
* lex_index
;
784 graph_data
= commit_graph_data_at(item
);
785 graph_data
->graph_pos
= pos
;
787 date_high
= get_be32(commit_data
+ g
->hash_len
+ 8) & 0x3;
788 date_low
= get_be32(commit_data
+ g
->hash_len
+ 12);
789 item
->date
= (timestamp_t
)((date_high
<< 32) | date_low
);
791 if (g
->read_generation_data
) {
792 offset
= (timestamp_t
)get_be32(g
->chunk_generation_data
+ sizeof(uint32_t) * lex_index
);
794 if (offset
& CORRECTED_COMMIT_DATE_OFFSET_OVERFLOW
) {
795 if (!g
->chunk_generation_data_overflow
)
796 die(_("commit-graph requires overflow generation data but has none"));
798 offset_pos
= offset
^ CORRECTED_COMMIT_DATE_OFFSET_OVERFLOW
;
799 graph_data
->generation
= item
->date
+ get_be64(g
->chunk_generation_data_overflow
+ 8 * offset_pos
);
801 graph_data
->generation
= item
->date
+ offset
;
803 graph_data
->generation
= get_be32(commit_data
+ g
->hash_len
+ 8) >> 2;
806 *topo_level_slab_at(g
->topo_levels
, item
) = get_be32(commit_data
+ g
->hash_len
+ 8) >> 2;
809 static inline void set_commit_tree(struct commit
*c
, struct tree
*t
)
814 static int fill_commit_in_graph(struct repository
*r
,
816 struct commit_graph
*g
, uint32_t pos
)
819 uint32_t *parent_data_ptr
;
820 struct commit_list
**pptr
;
821 const unsigned char *commit_data
;
824 while (pos
< g
->num_commits_in_base
)
827 fill_commit_graph_info(item
, g
, pos
);
829 lex_index
= pos
- g
->num_commits_in_base
;
830 commit_data
= g
->chunk_commit_data
+ (g
->hash_len
+ 16) * lex_index
;
832 item
->object
.parsed
= 1;
834 set_commit_tree(item
, NULL
);
836 pptr
= &item
->parents
;
838 edge_value
= get_be32(commit_data
+ g
->hash_len
);
839 if (edge_value
== GRAPH_PARENT_NONE
)
841 pptr
= insert_parent_or_die(r
, g
, edge_value
, pptr
);
843 edge_value
= get_be32(commit_data
+ g
->hash_len
+ 4);
844 if (edge_value
== GRAPH_PARENT_NONE
)
846 if (!(edge_value
& GRAPH_EXTRA_EDGES_NEEDED
)) {
847 pptr
= insert_parent_or_die(r
, g
, edge_value
, pptr
);
851 parent_data_ptr
= (uint32_t*)(g
->chunk_extra_edges
+
852 4 * (uint64_t)(edge_value
& GRAPH_EDGE_LAST_MASK
));
854 edge_value
= get_be32(parent_data_ptr
);
855 pptr
= insert_parent_or_die(r
, g
,
856 edge_value
& GRAPH_EDGE_LAST_MASK
,
859 } while (!(edge_value
& GRAPH_LAST_EDGE
));
864 static int search_commit_pos_in_graph(const struct object_id
*id
, struct commit_graph
*g
, uint32_t *pos
)
866 struct commit_graph
*cur_g
= g
;
869 while (cur_g
&& !bsearch_graph(cur_g
, id
, &lex_index
))
870 cur_g
= cur_g
->base_graph
;
873 *pos
= lex_index
+ cur_g
->num_commits_in_base
;
880 static int find_commit_pos_in_graph(struct commit
*item
, struct commit_graph
*g
, uint32_t *pos
)
882 uint32_t graph_pos
= commit_graph_position(item
);
883 if (graph_pos
!= COMMIT_NOT_FROM_GRAPH
) {
887 return search_commit_pos_in_graph(&item
->object
.oid
, g
, pos
);
891 struct commit
*lookup_commit_in_graph(struct repository
*repo
, const struct object_id
*id
)
893 struct commit
*commit
;
896 if (!repo
->objects
->commit_graph
)
898 if (!search_commit_pos_in_graph(id
, repo
->objects
->commit_graph
, &pos
))
900 if (!has_object(repo
, id
, 0))
903 commit
= lookup_commit(repo
, id
);
906 if (commit
->object
.parsed
)
909 if (!fill_commit_in_graph(repo
, commit
, repo
->objects
->commit_graph
, pos
))
915 static int parse_commit_in_graph_one(struct repository
*r
,
916 struct commit_graph
*g
,
921 if (item
->object
.parsed
)
924 if (find_commit_pos_in_graph(item
, g
, &pos
))
925 return fill_commit_in_graph(r
, item
, g
, pos
);
930 int parse_commit_in_graph(struct repository
*r
, struct commit
*item
)
932 static int checked_env
= 0;
935 git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_PARSE
, 0))
936 die("dying as requested by the '%s' variable on commit-graph parse!",
937 GIT_TEST_COMMIT_GRAPH_DIE_ON_PARSE
);
940 if (!prepare_commit_graph(r
))
942 return parse_commit_in_graph_one(r
, r
->objects
->commit_graph
, item
);
945 void load_commit_graph_info(struct repository
*r
, struct commit
*item
)
948 if (!prepare_commit_graph(r
))
950 if (find_commit_pos_in_graph(item
, r
->objects
->commit_graph
, &pos
))
951 fill_commit_graph_info(item
, r
->objects
->commit_graph
, pos
);
954 static struct tree
*load_tree_for_commit(struct repository
*r
,
955 struct commit_graph
*g
,
958 struct object_id oid
;
959 const unsigned char *commit_data
;
960 uint32_t graph_pos
= commit_graph_position(c
);
962 while (graph_pos
< g
->num_commits_in_base
)
965 commit_data
= g
->chunk_commit_data
+
966 GRAPH_DATA_WIDTH
* (graph_pos
- g
->num_commits_in_base
);
968 oidread(&oid
, commit_data
);
969 set_commit_tree(c
, lookup_tree(r
, &oid
));
971 return c
->maybe_tree
;
974 static struct tree
*get_commit_tree_in_graph_one(struct repository
*r
,
975 struct commit_graph
*g
,
976 const struct commit
*c
)
979 return c
->maybe_tree
;
980 if (commit_graph_position(c
) == COMMIT_NOT_FROM_GRAPH
)
981 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
983 return load_tree_for_commit(r
, g
, (struct commit
*)c
);
986 struct tree
*get_commit_tree_in_graph(struct repository
*r
, const struct commit
*c
)
988 return get_commit_tree_in_graph_one(r
, r
->objects
->commit_graph
, c
);
991 struct packed_commit_list
{
992 struct commit
**list
;
997 struct write_commit_graph_context
{
998 struct repository
*r
;
999 struct object_directory
*odb
;
1001 struct oid_array oids
;
1002 struct packed_commit_list commits
;
1003 int num_extra_edges
;
1004 int num_generation_data_overflows
;
1005 unsigned long approx_nr_objects
;
1006 struct progress
*progress
;
1008 uint64_t progress_cnt
;
1010 char *base_graph_name
;
1011 int num_commit_graphs_before
;
1012 int num_commit_graphs_after
;
1013 char **commit_graph_filenames_before
;
1014 char **commit_graph_filenames_after
;
1015 char **commit_graph_hash_after
;
1016 uint32_t new_num_commits_in_base
;
1017 struct commit_graph
*new_base_graph
;
1024 write_generation_data
:1,
1025 trust_generation_numbers
:1;
1027 struct topo_level_slab
*topo_levels
;
1028 const struct commit_graph_opts
*opts
;
1029 size_t total_bloom_filter_data_size
;
1030 const struct bloom_filter_settings
*bloom_settings
;
1032 int count_bloom_filter_computed
;
1033 int count_bloom_filter_not_computed
;
1034 int count_bloom_filter_trunc_empty
;
1035 int count_bloom_filter_trunc_large
;
1038 static int write_graph_chunk_fanout(struct hashfile
*f
,
1041 struct write_commit_graph_context
*ctx
= data
;
1043 struct commit
**list
= ctx
->commits
.list
;
1046 * Write the first-level table (the list is sorted,
1047 * but we use a 256-entry lookup to be able to avoid
1048 * having to do eight extra binary search iterations).
1050 for (i
= 0; i
< 256; i
++) {
1051 while (count
< ctx
->commits
.nr
) {
1052 if ((*list
)->object
.oid
.hash
[0] != i
)
1054 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1059 hashwrite_be32(f
, count
);
1065 static int write_graph_chunk_oids(struct hashfile
*f
,
1068 struct write_commit_graph_context
*ctx
= data
;
1069 struct commit
**list
= ctx
->commits
.list
;
1071 for (count
= 0; count
< ctx
->commits
.nr
; count
++, list
++) {
1072 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1073 hashwrite(f
, (*list
)->object
.oid
.hash
, the_hash_algo
->rawsz
);
1079 static const struct object_id
*commit_to_oid(size_t index
, const void *table
)
1081 const struct commit
* const *commits
= table
;
1082 return &commits
[index
]->object
.oid
;
1085 static int write_graph_chunk_data(struct hashfile
*f
,
1088 struct write_commit_graph_context
*ctx
= data
;
1089 struct commit
**list
= ctx
->commits
.list
;
1090 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
1091 uint32_t num_extra_edges
= 0;
1093 while (list
< last
) {
1094 struct commit_list
*parent
;
1095 struct object_id
*tree
;
1097 uint32_t packedDate
[2];
1098 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1100 if (repo_parse_commit_no_graph(ctx
->r
, *list
))
1101 die(_("unable to parse commit %s"),
1102 oid_to_hex(&(*list
)->object
.oid
));
1103 tree
= get_commit_tree_oid(*list
);
1104 hashwrite(f
, tree
->hash
, the_hash_algo
->rawsz
);
1106 parent
= (*list
)->parents
;
1109 edge_value
= GRAPH_PARENT_NONE
;
1111 edge_value
= oid_pos(&parent
->item
->object
.oid
,
1116 if (edge_value
>= 0)
1117 edge_value
+= ctx
->new_num_commits_in_base
;
1118 else if (ctx
->new_base_graph
) {
1120 if (find_commit_pos_in_graph(parent
->item
,
1121 ctx
->new_base_graph
,
1127 BUG("missing parent %s for commit %s",
1128 oid_to_hex(&parent
->item
->object
.oid
),
1129 oid_to_hex(&(*list
)->object
.oid
));
1132 hashwrite_be32(f
, edge_value
);
1135 parent
= parent
->next
;
1138 edge_value
= GRAPH_PARENT_NONE
;
1139 else if (parent
->next
)
1140 edge_value
= GRAPH_EXTRA_EDGES_NEEDED
| num_extra_edges
;
1142 edge_value
= oid_pos(&parent
->item
->object
.oid
,
1147 if (edge_value
>= 0)
1148 edge_value
+= ctx
->new_num_commits_in_base
;
1149 else if (ctx
->new_base_graph
) {
1151 if (find_commit_pos_in_graph(parent
->item
,
1152 ctx
->new_base_graph
,
1158 BUG("missing parent %s for commit %s",
1159 oid_to_hex(&parent
->item
->object
.oid
),
1160 oid_to_hex(&(*list
)->object
.oid
));
1163 hashwrite_be32(f
, edge_value
);
1165 if (edge_value
& GRAPH_EXTRA_EDGES_NEEDED
) {
1168 parent
= parent
->next
;
1172 if (sizeof((*list
)->date
) > 4)
1173 packedDate
[0] = htonl(((*list
)->date
>> 32) & 0x3);
1177 packedDate
[0] |= htonl(*topo_level_slab_at(ctx
->topo_levels
, *list
) << 2);
1179 packedDate
[1] = htonl((*list
)->date
);
1180 hashwrite(f
, packedDate
, 8);
1188 static int write_graph_chunk_generation_data(struct hashfile
*f
,
1191 struct write_commit_graph_context
*ctx
= data
;
1192 int i
, num_generation_data_overflows
= 0;
1194 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1195 struct commit
*c
= ctx
->commits
.list
[i
];
1197 repo_parse_commit(ctx
->r
, c
);
1198 offset
= commit_graph_data_at(c
)->generation
- c
->date
;
1199 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1201 if (offset
> GENERATION_NUMBER_V2_OFFSET_MAX
) {
1202 offset
= CORRECTED_COMMIT_DATE_OFFSET_OVERFLOW
| num_generation_data_overflows
;
1203 num_generation_data_overflows
++;
1206 hashwrite_be32(f
, offset
);
1212 static int write_graph_chunk_generation_data_overflow(struct hashfile
*f
,
1215 struct write_commit_graph_context
*ctx
= data
;
1217 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1218 struct commit
*c
= ctx
->commits
.list
[i
];
1219 timestamp_t offset
= commit_graph_data_at(c
)->generation
- c
->date
;
1220 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1222 if (offset
> GENERATION_NUMBER_V2_OFFSET_MAX
) {
1223 hashwrite_be32(f
, offset
>> 32);
1224 hashwrite_be32(f
, (uint32_t) offset
);
1231 static int write_graph_chunk_extra_edges(struct hashfile
*f
,
1234 struct write_commit_graph_context
*ctx
= data
;
1235 struct commit
**list
= ctx
->commits
.list
;
1236 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
1237 struct commit_list
*parent
;
1239 while (list
< last
) {
1240 int num_parents
= 0;
1242 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1244 for (parent
= (*list
)->parents
; num_parents
< 3 && parent
;
1245 parent
= parent
->next
)
1248 if (num_parents
<= 2) {
1253 /* Since num_parents > 2, this initializer is safe. */
1254 for (parent
= (*list
)->parents
->next
; parent
; parent
= parent
->next
) {
1255 int edge_value
= oid_pos(&parent
->item
->object
.oid
,
1260 if (edge_value
>= 0)
1261 edge_value
+= ctx
->new_num_commits_in_base
;
1262 else if (ctx
->new_base_graph
) {
1264 if (find_commit_pos_in_graph(parent
->item
,
1265 ctx
->new_base_graph
,
1271 BUG("missing parent %s for commit %s",
1272 oid_to_hex(&parent
->item
->object
.oid
),
1273 oid_to_hex(&(*list
)->object
.oid
));
1274 else if (!parent
->next
)
1275 edge_value
|= GRAPH_LAST_EDGE
;
1277 hashwrite_be32(f
, edge_value
);
1286 static int write_graph_chunk_bloom_indexes(struct hashfile
*f
,
1289 struct write_commit_graph_context
*ctx
= data
;
1290 struct commit
**list
= ctx
->commits
.list
;
1291 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
1292 uint32_t cur_pos
= 0;
1294 while (list
< last
) {
1295 struct bloom_filter
*filter
= get_bloom_filter(ctx
->r
, *list
);
1296 size_t len
= filter
? filter
->len
: 0;
1298 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1299 hashwrite_be32(f
, cur_pos
);
1306 static void trace2_bloom_filter_settings(struct write_commit_graph_context
*ctx
)
1308 struct json_writer jw
= JSON_WRITER_INIT
;
1310 jw_object_begin(&jw
, 0);
1311 jw_object_intmax(&jw
, "hash_version", ctx
->bloom_settings
->hash_version
);
1312 jw_object_intmax(&jw
, "num_hashes", ctx
->bloom_settings
->num_hashes
);
1313 jw_object_intmax(&jw
, "bits_per_entry", ctx
->bloom_settings
->bits_per_entry
);
1314 jw_object_intmax(&jw
, "max_changed_paths", ctx
->bloom_settings
->max_changed_paths
);
1317 trace2_data_json("bloom", ctx
->r
, "settings", &jw
);
1322 static int write_graph_chunk_bloom_data(struct hashfile
*f
,
1325 struct write_commit_graph_context
*ctx
= data
;
1326 struct commit
**list
= ctx
->commits
.list
;
1327 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
1329 trace2_bloom_filter_settings(ctx
);
1331 hashwrite_be32(f
, ctx
->bloom_settings
->hash_version
);
1332 hashwrite_be32(f
, ctx
->bloom_settings
->num_hashes
);
1333 hashwrite_be32(f
, ctx
->bloom_settings
->bits_per_entry
);
1335 while (list
< last
) {
1336 struct bloom_filter
*filter
= get_bloom_filter(ctx
->r
, *list
);
1337 size_t len
= filter
? filter
->len
: 0;
1339 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1341 hashwrite(f
, filter
->data
, len
* sizeof(unsigned char));
1348 static int add_packed_commits(const struct object_id
*oid
,
1349 struct packed_git
*pack
,
1353 struct write_commit_graph_context
*ctx
= (struct write_commit_graph_context
*)data
;
1354 enum object_type type
;
1355 off_t offset
= nth_packed_object_offset(pack
, pos
);
1356 struct object_info oi
= OBJECT_INFO_INIT
;
1359 display_progress(ctx
->progress
, ++ctx
->progress_done
);
1362 if (packed_object_info(ctx
->r
, pack
, offset
, &oi
) < 0)
1363 die(_("unable to get type of object %s"), oid_to_hex(oid
));
1365 if (type
!= OBJ_COMMIT
)
1368 oid_array_append(&ctx
->oids
, oid
);
1369 set_commit_pos(ctx
->r
, oid
);
1374 static void add_missing_parents(struct write_commit_graph_context
*ctx
, struct commit
*commit
)
1376 struct commit_list
*parent
;
1377 for (parent
= commit
->parents
; parent
; parent
= parent
->next
) {
1378 if (!(parent
->item
->object
.flags
& REACHABLE
)) {
1379 oid_array_append(&ctx
->oids
, &parent
->item
->object
.oid
);
1380 parent
->item
->object
.flags
|= REACHABLE
;
1385 static void close_reachable(struct write_commit_graph_context
*ctx
)
1388 struct commit
*commit
;
1389 enum commit_graph_split_flags flags
= ctx
->opts
?
1390 ctx
->opts
->split_flags
: COMMIT_GRAPH_SPLIT_UNSPECIFIED
;
1392 if (ctx
->report_progress
)
1393 ctx
->progress
= start_delayed_progress(
1394 _("Loading known commits in commit graph"),
1396 for (i
= 0; i
< ctx
->oids
.nr
; i
++) {
1397 display_progress(ctx
->progress
, i
+ 1);
1398 commit
= lookup_commit(ctx
->r
, &ctx
->oids
.oid
[i
]);
1400 commit
->object
.flags
|= REACHABLE
;
1402 stop_progress(&ctx
->progress
);
1405 * As this loop runs, ctx->oids.nr may grow, but not more
1406 * than the number of missing commits in the reachable
1409 if (ctx
->report_progress
)
1410 ctx
->progress
= start_delayed_progress(
1411 _("Expanding reachable commits in commit graph"),
1413 for (i
= 0; i
< ctx
->oids
.nr
; i
++) {
1414 display_progress(ctx
->progress
, i
+ 1);
1415 commit
= lookup_commit(ctx
->r
, &ctx
->oids
.oid
[i
]);
1420 if ((!repo_parse_commit(ctx
->r
, commit
) &&
1421 commit_graph_position(commit
) == COMMIT_NOT_FROM_GRAPH
) ||
1422 flags
== COMMIT_GRAPH_SPLIT_REPLACE
)
1423 add_missing_parents(ctx
, commit
);
1424 } else if (!repo_parse_commit_no_graph(ctx
->r
, commit
))
1425 add_missing_parents(ctx
, commit
);
1427 stop_progress(&ctx
->progress
);
1429 if (ctx
->report_progress
)
1430 ctx
->progress
= start_delayed_progress(
1431 _("Clearing commit marks in commit graph"),
1433 for (i
= 0; i
< ctx
->oids
.nr
; i
++) {
1434 display_progress(ctx
->progress
, i
+ 1);
1435 commit
= lookup_commit(ctx
->r
, &ctx
->oids
.oid
[i
]);
1438 commit
->object
.flags
&= ~REACHABLE
;
1440 stop_progress(&ctx
->progress
);
1443 static void compute_topological_levels(struct write_commit_graph_context
*ctx
)
1446 struct commit_list
*list
= NULL
;
1448 if (ctx
->report_progress
)
1449 ctx
->progress
= start_delayed_progress(
1450 _("Computing commit graph topological levels"),
1452 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1453 struct commit
*c
= ctx
->commits
.list
[i
];
1456 repo_parse_commit(ctx
->r
, c
);
1457 level
= *topo_level_slab_at(ctx
->topo_levels
, c
);
1459 display_progress(ctx
->progress
, i
+ 1);
1460 if (level
!= GENERATION_NUMBER_ZERO
)
1463 commit_list_insert(c
, &list
);
1465 struct commit
*current
= list
->item
;
1466 struct commit_list
*parent
;
1467 int all_parents_computed
= 1;
1468 uint32_t max_level
= 0;
1470 for (parent
= current
->parents
; parent
; parent
= parent
->next
) {
1471 repo_parse_commit(ctx
->r
, parent
->item
);
1472 level
= *topo_level_slab_at(ctx
->topo_levels
, parent
->item
);
1474 if (level
== GENERATION_NUMBER_ZERO
) {
1475 all_parents_computed
= 0;
1476 commit_list_insert(parent
->item
, &list
);
1480 if (level
> max_level
)
1484 if (all_parents_computed
) {
1487 if (max_level
> GENERATION_NUMBER_V1_MAX
- 1)
1488 max_level
= GENERATION_NUMBER_V1_MAX
- 1;
1489 *topo_level_slab_at(ctx
->topo_levels
, current
) = max_level
+ 1;
1493 stop_progress(&ctx
->progress
);
1496 static void compute_generation_numbers(struct write_commit_graph_context
*ctx
)
1499 struct commit_list
*list
= NULL
;
1501 if (ctx
->report_progress
)
1502 ctx
->progress
= start_delayed_progress(
1503 _("Computing commit graph generation numbers"),
1506 if (!ctx
->trust_generation_numbers
) {
1507 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1508 struct commit
*c
= ctx
->commits
.list
[i
];
1509 repo_parse_commit(ctx
->r
, c
);
1510 commit_graph_data_at(c
)->generation
= GENERATION_NUMBER_ZERO
;
1514 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1515 struct commit
*c
= ctx
->commits
.list
[i
];
1516 timestamp_t corrected_commit_date
;
1518 repo_parse_commit(ctx
->r
, c
);
1519 corrected_commit_date
= commit_graph_data_at(c
)->generation
;
1521 display_progress(ctx
->progress
, i
+ 1);
1522 if (corrected_commit_date
!= GENERATION_NUMBER_ZERO
)
1525 commit_list_insert(c
, &list
);
1527 struct commit
*current
= list
->item
;
1528 struct commit_list
*parent
;
1529 int all_parents_computed
= 1;
1530 timestamp_t max_corrected_commit_date
= 0;
1532 for (parent
= current
->parents
; parent
; parent
= parent
->next
) {
1533 repo_parse_commit(ctx
->r
, parent
->item
);
1534 corrected_commit_date
= commit_graph_data_at(parent
->item
)->generation
;
1536 if (corrected_commit_date
== GENERATION_NUMBER_ZERO
) {
1537 all_parents_computed
= 0;
1538 commit_list_insert(parent
->item
, &list
);
1542 if (corrected_commit_date
> max_corrected_commit_date
)
1543 max_corrected_commit_date
= corrected_commit_date
;
1546 if (all_parents_computed
) {
1549 if (current
->date
&& current
->date
> max_corrected_commit_date
)
1550 max_corrected_commit_date
= current
->date
- 1;
1551 commit_graph_data_at(current
)->generation
= max_corrected_commit_date
+ 1;
1556 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1557 struct commit
*c
= ctx
->commits
.list
[i
];
1558 timestamp_t offset
= commit_graph_data_at(c
)->generation
- c
->date
;
1559 if (offset
> GENERATION_NUMBER_V2_OFFSET_MAX
)
1560 ctx
->num_generation_data_overflows
++;
1562 stop_progress(&ctx
->progress
);
1565 static void trace2_bloom_filter_write_statistics(struct write_commit_graph_context
*ctx
)
1567 trace2_data_intmax("commit-graph", ctx
->r
, "filter-computed",
1568 ctx
->count_bloom_filter_computed
);
1569 trace2_data_intmax("commit-graph", ctx
->r
, "filter-not-computed",
1570 ctx
->count_bloom_filter_not_computed
);
1571 trace2_data_intmax("commit-graph", ctx
->r
, "filter-trunc-empty",
1572 ctx
->count_bloom_filter_trunc_empty
);
1573 trace2_data_intmax("commit-graph", ctx
->r
, "filter-trunc-large",
1574 ctx
->count_bloom_filter_trunc_large
);
1577 static void compute_bloom_filters(struct write_commit_graph_context
*ctx
)
1580 struct progress
*progress
= NULL
;
1581 struct commit
**sorted_commits
;
1582 int max_new_filters
;
1584 init_bloom_filters();
1586 if (ctx
->report_progress
)
1587 progress
= start_delayed_progress(
1588 _("Computing commit changed paths Bloom filters"),
1591 ALLOC_ARRAY(sorted_commits
, ctx
->commits
.nr
);
1592 COPY_ARRAY(sorted_commits
, ctx
->commits
.list
, ctx
->commits
.nr
);
1594 if (ctx
->order_by_pack
)
1595 QSORT(sorted_commits
, ctx
->commits
.nr
, commit_pos_cmp
);
1597 QSORT(sorted_commits
, ctx
->commits
.nr
, commit_gen_cmp
);
1599 max_new_filters
= ctx
->opts
&& ctx
->opts
->max_new_filters
>= 0 ?
1600 ctx
->opts
->max_new_filters
: ctx
->commits
.nr
;
1602 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1603 enum bloom_filter_computed computed
= 0;
1604 struct commit
*c
= sorted_commits
[i
];
1605 struct bloom_filter
*filter
= get_or_compute_bloom_filter(
1608 ctx
->count_bloom_filter_computed
< max_new_filters
,
1609 ctx
->bloom_settings
,
1611 if (computed
& BLOOM_COMPUTED
) {
1612 ctx
->count_bloom_filter_computed
++;
1613 if (computed
& BLOOM_TRUNC_EMPTY
)
1614 ctx
->count_bloom_filter_trunc_empty
++;
1615 if (computed
& BLOOM_TRUNC_LARGE
)
1616 ctx
->count_bloom_filter_trunc_large
++;
1617 } else if (computed
& BLOOM_NOT_COMPUTED
)
1618 ctx
->count_bloom_filter_not_computed
++;
1619 ctx
->total_bloom_filter_data_size
+= filter
1620 ? sizeof(unsigned char) * filter
->len
: 0;
1621 display_progress(progress
, i
+ 1);
1624 if (trace2_is_enabled())
1625 trace2_bloom_filter_write_statistics(ctx
);
1627 free(sorted_commits
);
1628 stop_progress(&progress
);
1631 struct refs_cb_data
{
1632 struct oidset
*commits
;
1633 struct progress
*progress
;
1636 static int add_ref_to_set(const char *refname
,
1637 const struct object_id
*oid
,
1638 int flags
, void *cb_data
)
1640 struct object_id peeled
;
1641 struct refs_cb_data
*data
= (struct refs_cb_data
*)cb_data
;
1643 if (!peel_iterated_oid(oid
, &peeled
))
1645 if (oid_object_info(the_repository
, oid
, NULL
) == OBJ_COMMIT
)
1646 oidset_insert(data
->commits
, oid
);
1648 display_progress(data
->progress
, oidset_size(data
->commits
));
1653 int write_commit_graph_reachable(struct object_directory
*odb
,
1654 enum commit_graph_write_flags flags
,
1655 const struct commit_graph_opts
*opts
)
1657 struct oidset commits
= OIDSET_INIT
;
1658 struct refs_cb_data data
;
1661 memset(&data
, 0, sizeof(data
));
1662 data
.commits
= &commits
;
1663 if (flags
& COMMIT_GRAPH_WRITE_PROGRESS
)
1664 data
.progress
= start_delayed_progress(
1665 _("Collecting referenced commits"), 0);
1667 for_each_ref(add_ref_to_set
, &data
);
1669 stop_progress(&data
.progress
);
1671 result
= write_commit_graph(odb
, NULL
, &commits
,
1674 oidset_clear(&commits
);
1678 static int fill_oids_from_packs(struct write_commit_graph_context
*ctx
,
1679 const struct string_list
*pack_indexes
)
1682 struct strbuf progress_title
= STRBUF_INIT
;
1683 struct strbuf packname
= STRBUF_INIT
;
1687 strbuf_addf(&packname
, "%s/pack/", ctx
->odb
->path
);
1688 dirlen
= packname
.len
;
1689 if (ctx
->report_progress
) {
1690 strbuf_addf(&progress_title
,
1691 Q_("Finding commits for commit graph in %"PRIuMAX
" pack",
1692 "Finding commits for commit graph in %"PRIuMAX
" packs",
1694 (uintmax_t)pack_indexes
->nr
);
1695 ctx
->progress
= start_delayed_progress(progress_title
.buf
, 0);
1696 ctx
->progress_done
= 0;
1698 for (i
= 0; i
< pack_indexes
->nr
; i
++) {
1699 struct packed_git
*p
;
1700 strbuf_setlen(&packname
, dirlen
);
1701 strbuf_addstr(&packname
, pack_indexes
->items
[i
].string
);
1702 p
= add_packed_git(packname
.buf
, packname
.len
, 1);
1704 ret
= error(_("error adding pack %s"), packname
.buf
);
1707 if (open_pack_index(p
)) {
1708 ret
= error(_("error opening index for %s"), packname
.buf
);
1711 for_each_object_in_pack(p
, add_packed_commits
, ctx
,
1712 FOR_EACH_OBJECT_PACK_ORDER
);
1718 stop_progress(&ctx
->progress
);
1719 strbuf_release(&progress_title
);
1720 strbuf_release(&packname
);
1725 static int fill_oids_from_commits(struct write_commit_graph_context
*ctx
,
1726 struct oidset
*commits
)
1728 struct oidset_iter iter
;
1729 struct object_id
*oid
;
1731 if (!oidset_size(commits
))
1734 oidset_iter_init(commits
, &iter
);
1735 while ((oid
= oidset_iter_next(&iter
))) {
1736 oid_array_append(&ctx
->oids
, oid
);
1742 static void fill_oids_from_all_packs(struct write_commit_graph_context
*ctx
)
1744 if (ctx
->report_progress
)
1745 ctx
->progress
= start_delayed_progress(
1746 _("Finding commits for commit graph among packed objects"),
1747 ctx
->approx_nr_objects
);
1748 for_each_packed_object(add_packed_commits
, ctx
,
1749 FOR_EACH_OBJECT_PACK_ORDER
);
1750 if (ctx
->progress_done
< ctx
->approx_nr_objects
)
1751 display_progress(ctx
->progress
, ctx
->approx_nr_objects
);
1752 stop_progress(&ctx
->progress
);
1755 static void copy_oids_to_commits(struct write_commit_graph_context
*ctx
)
1758 enum commit_graph_split_flags flags
= ctx
->opts
?
1759 ctx
->opts
->split_flags
: COMMIT_GRAPH_SPLIT_UNSPECIFIED
;
1761 ctx
->num_extra_edges
= 0;
1762 if (ctx
->report_progress
)
1763 ctx
->progress
= start_delayed_progress(
1764 _("Finding extra edges in commit graph"),
1766 oid_array_sort(&ctx
->oids
);
1767 for (i
= 0; i
< ctx
->oids
.nr
; i
= oid_array_next_unique(&ctx
->oids
, i
)) {
1768 unsigned int num_parents
;
1770 display_progress(ctx
->progress
, i
+ 1);
1772 ALLOC_GROW(ctx
->commits
.list
, ctx
->commits
.nr
+ 1, ctx
->commits
.alloc
);
1773 ctx
->commits
.list
[ctx
->commits
.nr
] = lookup_commit(ctx
->r
, &ctx
->oids
.oid
[i
]);
1775 if (ctx
->split
&& flags
!= COMMIT_GRAPH_SPLIT_REPLACE
&&
1776 commit_graph_position(ctx
->commits
.list
[ctx
->commits
.nr
]) != COMMIT_NOT_FROM_GRAPH
)
1779 if (ctx
->split
&& flags
== COMMIT_GRAPH_SPLIT_REPLACE
)
1780 repo_parse_commit(ctx
->r
, ctx
->commits
.list
[ctx
->commits
.nr
]);
1782 repo_parse_commit_no_graph(ctx
->r
, ctx
->commits
.list
[ctx
->commits
.nr
]);
1784 num_parents
= commit_list_count(ctx
->commits
.list
[ctx
->commits
.nr
]->parents
);
1785 if (num_parents
> 2)
1786 ctx
->num_extra_edges
+= num_parents
- 1;
1790 stop_progress(&ctx
->progress
);
1793 static int write_graph_chunk_base_1(struct hashfile
*f
,
1794 struct commit_graph
*g
)
1801 num
= write_graph_chunk_base_1(f
, g
->base_graph
);
1802 hashwrite(f
, g
->oid
.hash
, the_hash_algo
->rawsz
);
1806 static int write_graph_chunk_base(struct hashfile
*f
,
1809 struct write_commit_graph_context
*ctx
= data
;
1810 int num
= write_graph_chunk_base_1(f
, ctx
->new_base_graph
);
1812 if (num
!= ctx
->num_commit_graphs_after
- 1) {
1813 error(_("failed to write correct number of base graph ids"));
1820 static int write_commit_graph_file(struct write_commit_graph_context
*ctx
)
1825 struct lock_file lk
= LOCK_INIT
;
1826 const unsigned hashsz
= the_hash_algo
->rawsz
;
1827 struct strbuf progress_title
= STRBUF_INIT
;
1828 struct chunkfile
*cf
;
1829 unsigned char file_hash
[GIT_MAX_RAWSZ
];
1832 struct strbuf tmp_file
= STRBUF_INIT
;
1834 strbuf_addf(&tmp_file
,
1835 "%s/info/commit-graphs/tmp_graph_XXXXXX",
1837 ctx
->graph_name
= strbuf_detach(&tmp_file
, NULL
);
1839 ctx
->graph_name
= get_commit_graph_filename(ctx
->odb
);
1842 if (safe_create_leading_directories(ctx
->graph_name
)) {
1843 UNLEAK(ctx
->graph_name
);
1844 error(_("unable to create leading directories of %s"),
1850 char *lock_name
= get_commit_graph_chain_filename(ctx
->odb
);
1852 hold_lock_file_for_update_mode(&lk
, lock_name
,
1853 LOCK_DIE_ON_ERROR
, 0444);
1856 fd
= git_mkstemp_mode(ctx
->graph_name
, 0444);
1858 error(_("unable to create temporary graph layer"));
1862 if (adjust_shared_perm(ctx
->graph_name
)) {
1863 error(_("unable to adjust shared permissions for '%s'"),
1868 f
= hashfd(fd
, ctx
->graph_name
);
1870 hold_lock_file_for_update_mode(&lk
, ctx
->graph_name
,
1871 LOCK_DIE_ON_ERROR
, 0444);
1872 fd
= get_lock_file_fd(&lk
);
1873 f
= hashfd(fd
, get_lock_file_path(&lk
));
1876 cf
= init_chunkfile(f
);
1878 add_chunk(cf
, GRAPH_CHUNKID_OIDFANOUT
, GRAPH_FANOUT_SIZE
,
1879 write_graph_chunk_fanout
);
1880 add_chunk(cf
, GRAPH_CHUNKID_OIDLOOKUP
, hashsz
* ctx
->commits
.nr
,
1881 write_graph_chunk_oids
);
1882 add_chunk(cf
, GRAPH_CHUNKID_DATA
, (hashsz
+ 16) * ctx
->commits
.nr
,
1883 write_graph_chunk_data
);
1885 if (ctx
->write_generation_data
)
1886 add_chunk(cf
, GRAPH_CHUNKID_GENERATION_DATA
,
1887 sizeof(uint32_t) * ctx
->commits
.nr
,
1888 write_graph_chunk_generation_data
);
1889 if (ctx
->num_generation_data_overflows
)
1890 add_chunk(cf
, GRAPH_CHUNKID_GENERATION_DATA_OVERFLOW
,
1891 sizeof(timestamp_t
) * ctx
->num_generation_data_overflows
,
1892 write_graph_chunk_generation_data_overflow
);
1893 if (ctx
->num_extra_edges
)
1894 add_chunk(cf
, GRAPH_CHUNKID_EXTRAEDGES
,
1895 4 * ctx
->num_extra_edges
,
1896 write_graph_chunk_extra_edges
);
1897 if (ctx
->changed_paths
) {
1898 add_chunk(cf
, GRAPH_CHUNKID_BLOOMINDEXES
,
1899 sizeof(uint32_t) * ctx
->commits
.nr
,
1900 write_graph_chunk_bloom_indexes
);
1901 add_chunk(cf
, GRAPH_CHUNKID_BLOOMDATA
,
1902 sizeof(uint32_t) * 3
1903 + ctx
->total_bloom_filter_data_size
,
1904 write_graph_chunk_bloom_data
);
1906 if (ctx
->num_commit_graphs_after
> 1)
1907 add_chunk(cf
, GRAPH_CHUNKID_BASE
,
1908 hashsz
* (ctx
->num_commit_graphs_after
- 1),
1909 write_graph_chunk_base
);
1911 hashwrite_be32(f
, GRAPH_SIGNATURE
);
1913 hashwrite_u8(f
, GRAPH_VERSION
);
1914 hashwrite_u8(f
, oid_version(the_hash_algo
));
1915 hashwrite_u8(f
, get_num_chunks(cf
));
1916 hashwrite_u8(f
, ctx
->num_commit_graphs_after
- 1);
1918 if (ctx
->report_progress
) {
1919 strbuf_addf(&progress_title
,
1920 Q_("Writing out commit graph in %d pass",
1921 "Writing out commit graph in %d passes",
1922 get_num_chunks(cf
)),
1923 get_num_chunks(cf
));
1924 ctx
->progress
= start_delayed_progress(
1926 get_num_chunks(cf
) * ctx
->commits
.nr
);
1929 write_chunkfile(cf
, ctx
);
1931 stop_progress(&ctx
->progress
);
1932 strbuf_release(&progress_title
);
1934 if (ctx
->split
&& ctx
->base_graph_name
&& ctx
->num_commit_graphs_after
> 1) {
1935 char *new_base_hash
= xstrdup(oid_to_hex(&ctx
->new_base_graph
->oid
));
1936 char *new_base_name
= get_split_graph_filename(ctx
->new_base_graph
->odb
, new_base_hash
);
1938 free(ctx
->commit_graph_filenames_after
[ctx
->num_commit_graphs_after
- 2]);
1939 free(ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 2]);
1940 ctx
->commit_graph_filenames_after
[ctx
->num_commit_graphs_after
- 2] = new_base_name
;
1941 ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 2] = new_base_hash
;
1944 close_commit_graph(ctx
->r
->objects
);
1945 finalize_hashfile(f
, file_hash
, FSYNC_COMPONENT_COMMIT_GRAPH
,
1946 CSUM_HASH_IN_STREAM
| CSUM_FSYNC
);
1950 FILE *chainf
= fdopen_lock_file(&lk
, "w");
1951 char *final_graph_name
;
1957 error(_("unable to open commit-graph chain file"));
1961 if (ctx
->base_graph_name
) {
1963 int idx
= ctx
->num_commit_graphs_after
- 1;
1964 if (ctx
->num_commit_graphs_after
> 1)
1967 dest
= ctx
->commit_graph_filenames_after
[idx
];
1969 if (strcmp(ctx
->base_graph_name
, dest
)) {
1970 result
= rename(ctx
->base_graph_name
, dest
);
1973 error(_("failed to rename base commit-graph file"));
1978 char *graph_name
= get_commit_graph_filename(ctx
->odb
);
1983 ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 1] = xstrdup(hash_to_hex(file_hash
));
1984 final_graph_name
= get_split_graph_filename(ctx
->odb
,
1985 ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 1]);
1986 ctx
->commit_graph_filenames_after
[ctx
->num_commit_graphs_after
- 1] = final_graph_name
;
1988 result
= rename(ctx
->graph_name
, final_graph_name
);
1990 for (i
= 0; i
< ctx
->num_commit_graphs_after
; i
++)
1991 fprintf(get_lock_file_fp(&lk
), "%s\n", ctx
->commit_graph_hash_after
[i
]);
1994 error(_("failed to rename temporary commit-graph file"));
1999 commit_lock_file(&lk
);
2004 static void split_graph_merge_strategy(struct write_commit_graph_context
*ctx
)
2006 struct commit_graph
*g
;
2007 uint32_t num_commits
;
2008 enum commit_graph_split_flags flags
= COMMIT_GRAPH_SPLIT_UNSPECIFIED
;
2011 int max_commits
= 0;
2015 max_commits
= ctx
->opts
->max_commits
;
2017 if (ctx
->opts
->size_multiple
)
2018 size_mult
= ctx
->opts
->size_multiple
;
2020 flags
= ctx
->opts
->split_flags
;
2023 g
= ctx
->r
->objects
->commit_graph
;
2024 num_commits
= ctx
->commits
.nr
;
2025 if (flags
== COMMIT_GRAPH_SPLIT_REPLACE
)
2026 ctx
->num_commit_graphs_after
= 1;
2028 ctx
->num_commit_graphs_after
= ctx
->num_commit_graphs_before
+ 1;
2030 if (flags
!= COMMIT_GRAPH_SPLIT_MERGE_PROHIBITED
&&
2031 flags
!= COMMIT_GRAPH_SPLIT_REPLACE
) {
2032 while (g
&& (g
->num_commits
<= size_mult
* num_commits
||
2033 (max_commits
&& num_commits
> max_commits
))) {
2034 if (g
->odb
!= ctx
->odb
)
2037 num_commits
+= g
->num_commits
;
2040 ctx
->num_commit_graphs_after
--;
2044 if (flags
!= COMMIT_GRAPH_SPLIT_REPLACE
)
2045 ctx
->new_base_graph
= g
;
2046 else if (ctx
->num_commit_graphs_after
!= 1)
2047 BUG("split_graph_merge_strategy: num_commit_graphs_after "
2048 "should be 1 with --split=replace");
2050 if (ctx
->num_commit_graphs_after
== 2) {
2051 char *old_graph_name
= get_commit_graph_filename(g
->odb
);
2053 if (!strcmp(g
->filename
, old_graph_name
) &&
2054 g
->odb
!= ctx
->odb
) {
2055 ctx
->num_commit_graphs_after
= 1;
2056 ctx
->new_base_graph
= NULL
;
2059 free(old_graph_name
);
2062 CALLOC_ARRAY(ctx
->commit_graph_filenames_after
, ctx
->num_commit_graphs_after
);
2063 CALLOC_ARRAY(ctx
->commit_graph_hash_after
, ctx
->num_commit_graphs_after
);
2065 for (i
= 0; i
< ctx
->num_commit_graphs_after
&&
2066 i
< ctx
->num_commit_graphs_before
; i
++)
2067 ctx
->commit_graph_filenames_after
[i
] = xstrdup(ctx
->commit_graph_filenames_before
[i
]);
2069 i
= ctx
->num_commit_graphs_before
- 1;
2070 g
= ctx
->r
->objects
->commit_graph
;
2073 if (i
< ctx
->num_commit_graphs_after
)
2074 ctx
->commit_graph_hash_after
[i
] = xstrdup(oid_to_hex(&g
->oid
));
2077 * If the topmost remaining layer has generation data chunk, the
2078 * resultant layer also has generation data chunk.
2080 if (i
== ctx
->num_commit_graphs_after
- 2)
2081 ctx
->write_generation_data
= !!g
->chunk_generation_data
;
2088 static void merge_commit_graph(struct write_commit_graph_context
*ctx
,
2089 struct commit_graph
*g
)
2092 uint32_t offset
= g
->num_commits_in_base
;
2094 ALLOC_GROW(ctx
->commits
.list
, ctx
->commits
.nr
+ g
->num_commits
, ctx
->commits
.alloc
);
2096 for (i
= 0; i
< g
->num_commits
; i
++) {
2097 struct object_id oid
;
2098 struct commit
*result
;
2100 display_progress(ctx
->progress
, i
+ 1);
2102 load_oid_from_graph(g
, i
+ offset
, &oid
);
2104 /* only add commits if they still exist in the repo */
2105 result
= lookup_commit_reference_gently(ctx
->r
, &oid
, 1);
2108 ctx
->commits
.list
[ctx
->commits
.nr
] = result
;
2114 static int commit_compare(const void *_a
, const void *_b
)
2116 const struct commit
*a
= *(const struct commit
**)_a
;
2117 const struct commit
*b
= *(const struct commit
**)_b
;
2118 return oidcmp(&a
->object
.oid
, &b
->object
.oid
);
2121 static void sort_and_scan_merged_commits(struct write_commit_graph_context
*ctx
)
2123 uint32_t i
, dedup_i
= 0;
2125 if (ctx
->report_progress
)
2126 ctx
->progress
= start_delayed_progress(
2127 _("Scanning merged commits"),
2130 QSORT(ctx
->commits
.list
, ctx
->commits
.nr
, commit_compare
);
2132 ctx
->num_extra_edges
= 0;
2133 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
2134 display_progress(ctx
->progress
, i
+ 1);
2136 if (i
&& oideq(&ctx
->commits
.list
[i
- 1]->object
.oid
,
2137 &ctx
->commits
.list
[i
]->object
.oid
)) {
2139 * Silently ignore duplicates. These were likely
2140 * created due to a commit appearing in multiple
2141 * layers of the chain, which is unexpected but
2142 * not invalid. We should make sure there is a
2143 * unique copy in the new layer.
2146 unsigned int num_parents
;
2148 ctx
->commits
.list
[dedup_i
] = ctx
->commits
.list
[i
];
2151 num_parents
= commit_list_count(ctx
->commits
.list
[i
]->parents
);
2152 if (num_parents
> 2)
2153 ctx
->num_extra_edges
+= num_parents
- 1;
2157 ctx
->commits
.nr
= dedup_i
;
2159 stop_progress(&ctx
->progress
);
2162 static void merge_commit_graphs(struct write_commit_graph_context
*ctx
)
2164 struct commit_graph
*g
= ctx
->r
->objects
->commit_graph
;
2165 uint32_t current_graph_number
= ctx
->num_commit_graphs_before
;
2167 while (g
&& current_graph_number
>= ctx
->num_commit_graphs_after
) {
2168 current_graph_number
--;
2170 if (ctx
->report_progress
)
2171 ctx
->progress
= start_delayed_progress(_("Merging commit-graph"), 0);
2173 merge_commit_graph(ctx
, g
);
2174 stop_progress(&ctx
->progress
);
2180 ctx
->new_base_graph
= g
;
2181 ctx
->new_num_commits_in_base
= g
->num_commits
+ g
->num_commits_in_base
;
2184 if (ctx
->new_base_graph
)
2185 ctx
->base_graph_name
= xstrdup(ctx
->new_base_graph
->filename
);
2187 sort_and_scan_merged_commits(ctx
);
2190 static void mark_commit_graphs(struct write_commit_graph_context
*ctx
)
2193 time_t now
= time(NULL
);
2195 for (i
= ctx
->num_commit_graphs_after
- 1; i
< ctx
->num_commit_graphs_before
; i
++) {
2197 struct utimbuf updated_time
;
2199 if (stat(ctx
->commit_graph_filenames_before
[i
], &st
) < 0)
2202 updated_time
.actime
= st
.st_atime
;
2203 updated_time
.modtime
= now
;
2204 utime(ctx
->commit_graph_filenames_before
[i
], &updated_time
);
2208 static void expire_commit_graphs(struct write_commit_graph_context
*ctx
)
2210 struct strbuf path
= STRBUF_INIT
;
2214 timestamp_t expire_time
= time(NULL
);
2216 if (ctx
->opts
&& ctx
->opts
->expire_time
)
2217 expire_time
= ctx
->opts
->expire_time
;
2219 char *chain_file_name
= get_commit_graph_chain_filename(ctx
->odb
);
2220 unlink(chain_file_name
);
2221 free(chain_file_name
);
2222 ctx
->num_commit_graphs_after
= 0;
2225 strbuf_addstr(&path
, ctx
->odb
->path
);
2226 strbuf_addstr(&path
, "/info/commit-graphs");
2227 dir
= opendir(path
.buf
);
2232 strbuf_addch(&path
, '/');
2233 dirnamelen
= path
.len
;
2234 while ((de
= readdir(dir
)) != NULL
) {
2236 uint32_t i
, found
= 0;
2238 strbuf_setlen(&path
, dirnamelen
);
2239 strbuf_addstr(&path
, de
->d_name
);
2241 if (stat(path
.buf
, &st
) < 0)
2244 if (st
.st_mtime
> expire_time
)
2246 if (path
.len
< 6 || strcmp(path
.buf
+ path
.len
- 6, ".graph"))
2249 for (i
= 0; i
< ctx
->num_commit_graphs_after
; i
++) {
2250 if (!strcmp(ctx
->commit_graph_filenames_after
[i
],
2262 strbuf_release(&path
);
2265 int write_commit_graph(struct object_directory
*odb
,
2266 const struct string_list
*const pack_indexes
,
2267 struct oidset
*commits
,
2268 enum commit_graph_write_flags flags
,
2269 const struct commit_graph_opts
*opts
)
2271 struct repository
*r
= the_repository
;
2272 struct write_commit_graph_context
*ctx
;
2276 struct bloom_filter_settings bloom_settings
= DEFAULT_BLOOM_FILTER_SETTINGS
;
2277 struct topo_level_slab topo_levels
;
2279 prepare_repo_settings(r
);
2280 if (!r
->settings
.core_commit_graph
) {
2281 warning(_("attempting to write a commit-graph, but 'core.commitGraph' is disabled"));
2284 if (!commit_graph_compatible(r
))
2287 CALLOC_ARRAY(ctx
, 1);
2290 ctx
->append
= flags
& COMMIT_GRAPH_WRITE_APPEND
? 1 : 0;
2291 ctx
->report_progress
= flags
& COMMIT_GRAPH_WRITE_PROGRESS
? 1 : 0;
2292 ctx
->split
= flags
& COMMIT_GRAPH_WRITE_SPLIT
? 1 : 0;
2294 ctx
->total_bloom_filter_data_size
= 0;
2295 ctx
->write_generation_data
= (get_configured_generation_version(r
) == 2);
2296 ctx
->num_generation_data_overflows
= 0;
2298 bloom_settings
.bits_per_entry
= git_env_ulong("GIT_TEST_BLOOM_SETTINGS_BITS_PER_ENTRY",
2299 bloom_settings
.bits_per_entry
);
2300 bloom_settings
.num_hashes
= git_env_ulong("GIT_TEST_BLOOM_SETTINGS_NUM_HASHES",
2301 bloom_settings
.num_hashes
);
2302 bloom_settings
.max_changed_paths
= git_env_ulong("GIT_TEST_BLOOM_SETTINGS_MAX_CHANGED_PATHS",
2303 bloom_settings
.max_changed_paths
);
2304 ctx
->bloom_settings
= &bloom_settings
;
2306 init_topo_level_slab(&topo_levels
);
2307 ctx
->topo_levels
= &topo_levels
;
2309 prepare_commit_graph(ctx
->r
);
2310 if (ctx
->r
->objects
->commit_graph
) {
2311 struct commit_graph
*g
= ctx
->r
->objects
->commit_graph
;
2314 g
->topo_levels
= &topo_levels
;
2319 if (flags
& COMMIT_GRAPH_WRITE_BLOOM_FILTERS
)
2320 ctx
->changed_paths
= 1;
2321 if (!(flags
& COMMIT_GRAPH_NO_WRITE_BLOOM_FILTERS
)) {
2322 struct commit_graph
*g
;
2324 g
= ctx
->r
->objects
->commit_graph
;
2326 /* We have changed-paths already. Keep them in the next graph */
2327 if (g
&& g
->chunk_bloom_data
) {
2328 ctx
->changed_paths
= 1;
2329 ctx
->bloom_settings
= g
->bloom_filter_settings
;
2334 struct commit_graph
*g
= ctx
->r
->objects
->commit_graph
;
2337 ctx
->num_commit_graphs_before
++;
2341 if (ctx
->num_commit_graphs_before
) {
2342 ALLOC_ARRAY(ctx
->commit_graph_filenames_before
, ctx
->num_commit_graphs_before
);
2343 i
= ctx
->num_commit_graphs_before
;
2344 g
= ctx
->r
->objects
->commit_graph
;
2347 ctx
->commit_graph_filenames_before
[--i
] = xstrdup(g
->filename
);
2353 replace
= ctx
->opts
->split_flags
& COMMIT_GRAPH_SPLIT_REPLACE
;
2356 ctx
->approx_nr_objects
= approximate_object_count();
2358 if (ctx
->append
&& ctx
->r
->objects
->commit_graph
) {
2359 struct commit_graph
*g
= ctx
->r
->objects
->commit_graph
;
2360 for (i
= 0; i
< g
->num_commits
; i
++) {
2361 struct object_id oid
;
2362 oidread(&oid
, g
->chunk_oid_lookup
+ g
->hash_len
* i
);
2363 oid_array_append(&ctx
->oids
, &oid
);
2368 ctx
->order_by_pack
= 1;
2369 if ((res
= fill_oids_from_packs(ctx
, pack_indexes
)))
2374 if ((res
= fill_oids_from_commits(ctx
, commits
)))
2378 if (!pack_indexes
&& !commits
) {
2379 ctx
->order_by_pack
= 1;
2380 fill_oids_from_all_packs(ctx
);
2383 close_reachable(ctx
);
2385 copy_oids_to_commits(ctx
);
2387 if (ctx
->commits
.nr
>= GRAPH_EDGE_LAST_MASK
) {
2388 error(_("too many commits to write graph"));
2393 if (!ctx
->commits
.nr
&& !replace
)
2397 split_graph_merge_strategy(ctx
);
2400 merge_commit_graphs(ctx
);
2402 ctx
->num_commit_graphs_after
= 1;
2404 ctx
->trust_generation_numbers
= validate_mixed_generation_chain(ctx
->r
->objects
->commit_graph
);
2406 compute_topological_levels(ctx
);
2407 if (ctx
->write_generation_data
)
2408 compute_generation_numbers(ctx
);
2410 if (ctx
->changed_paths
)
2411 compute_bloom_filters(ctx
);
2413 res
= write_commit_graph_file(ctx
);
2416 mark_commit_graphs(ctx
);
2418 expire_commit_graphs(ctx
);
2421 free(ctx
->graph_name
);
2422 free(ctx
->commits
.list
);
2423 oid_array_clear(&ctx
->oids
);
2424 clear_topo_level_slab(&topo_levels
);
2426 if (ctx
->commit_graph_filenames_after
) {
2427 for (i
= 0; i
< ctx
->num_commit_graphs_after
; i
++) {
2428 free(ctx
->commit_graph_filenames_after
[i
]);
2429 free(ctx
->commit_graph_hash_after
[i
]);
2432 for (i
= 0; i
< ctx
->num_commit_graphs_before
; i
++)
2433 free(ctx
->commit_graph_filenames_before
[i
]);
2435 free(ctx
->commit_graph_filenames_after
);
2436 free(ctx
->commit_graph_filenames_before
);
2437 free(ctx
->commit_graph_hash_after
);
2445 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
2446 static int verify_commit_graph_error
;
2448 __attribute__((format (printf
, 1, 2)))
2449 static void graph_report(const char *fmt
, ...)
2453 verify_commit_graph_error
= 1;
2455 vfprintf(stderr
, fmt
, ap
);
2456 fprintf(stderr
, "\n");
2460 #define GENERATION_ZERO_EXISTS 1
2461 #define GENERATION_NUMBER_EXISTS 2
2463 static int commit_graph_checksum_valid(struct commit_graph
*g
)
2465 return hashfile_checksum_valid(g
->data
, g
->data_len
);
2468 int verify_commit_graph(struct repository
*r
, struct commit_graph
*g
, int flags
)
2470 uint32_t i
, cur_fanout_pos
= 0;
2471 struct object_id prev_oid
, cur_oid
;
2472 int generation_zero
= 0;
2473 struct progress
*progress
= NULL
;
2474 int local_error
= 0;
2477 graph_report("no commit-graph file loaded");
2481 verify_commit_graph_error
= verify_commit_graph_lite(g
);
2482 if (verify_commit_graph_error
)
2483 return verify_commit_graph_error
;
2485 if (!commit_graph_checksum_valid(g
)) {
2486 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
2487 verify_commit_graph_error
= VERIFY_COMMIT_GRAPH_ERROR_HASH
;
2490 for (i
= 0; i
< g
->num_commits
; i
++) {
2491 struct commit
*graph_commit
;
2493 oidread(&cur_oid
, g
->chunk_oid_lookup
+ g
->hash_len
* i
);
2495 if (i
&& oidcmp(&prev_oid
, &cur_oid
) >= 0)
2496 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
2497 oid_to_hex(&prev_oid
),
2498 oid_to_hex(&cur_oid
));
2500 oidcpy(&prev_oid
, &cur_oid
);
2502 while (cur_oid
.hash
[0] > cur_fanout_pos
) {
2503 uint32_t fanout_value
= get_be32(g
->chunk_oid_fanout
+ cur_fanout_pos
);
2505 if (i
!= fanout_value
)
2506 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2507 cur_fanout_pos
, fanout_value
, i
);
2511 graph_commit
= lookup_commit(r
, &cur_oid
);
2512 if (!parse_commit_in_graph_one(r
, g
, graph_commit
))
2513 graph_report(_("failed to parse commit %s from commit-graph"),
2514 oid_to_hex(&cur_oid
));
2517 while (cur_fanout_pos
< 256) {
2518 uint32_t fanout_value
= get_be32(g
->chunk_oid_fanout
+ cur_fanout_pos
);
2520 if (g
->num_commits
!= fanout_value
)
2521 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2522 cur_fanout_pos
, fanout_value
, i
);
2527 if (verify_commit_graph_error
& ~VERIFY_COMMIT_GRAPH_ERROR_HASH
)
2528 return verify_commit_graph_error
;
2530 if (flags
& COMMIT_GRAPH_WRITE_PROGRESS
)
2531 progress
= start_progress(_("Verifying commits in commit graph"),
2534 for (i
= 0; i
< g
->num_commits
; i
++) {
2535 struct commit
*graph_commit
, *odb_commit
;
2536 struct commit_list
*graph_parents
, *odb_parents
;
2537 timestamp_t max_generation
= 0;
2538 timestamp_t generation
;
2540 display_progress(progress
, i
+ 1);
2541 oidread(&cur_oid
, g
->chunk_oid_lookup
+ g
->hash_len
* i
);
2543 graph_commit
= lookup_commit(r
, &cur_oid
);
2544 odb_commit
= (struct commit
*)create_object(r
, &cur_oid
, alloc_commit_node(r
));
2545 if (parse_commit_internal(odb_commit
, 0, 0)) {
2546 graph_report(_("failed to parse commit %s from object database for commit-graph"),
2547 oid_to_hex(&cur_oid
));
2551 if (!oideq(&get_commit_tree_in_graph_one(r
, g
, graph_commit
)->object
.oid
,
2552 get_commit_tree_oid(odb_commit
)))
2553 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2554 oid_to_hex(&cur_oid
),
2555 oid_to_hex(get_commit_tree_oid(graph_commit
)),
2556 oid_to_hex(get_commit_tree_oid(odb_commit
)));
2558 graph_parents
= graph_commit
->parents
;
2559 odb_parents
= odb_commit
->parents
;
2561 while (graph_parents
) {
2563 graph_report(_("commit-graph parent list for commit %s is too long"),
2564 oid_to_hex(&cur_oid
));
2568 /* parse parent in case it is in a base graph */
2569 parse_commit_in_graph_one(r
, g
, graph_parents
->item
);
2571 if (!oideq(&graph_parents
->item
->object
.oid
, &odb_parents
->item
->object
.oid
))
2572 graph_report(_("commit-graph parent for %s is %s != %s"),
2573 oid_to_hex(&cur_oid
),
2574 oid_to_hex(&graph_parents
->item
->object
.oid
),
2575 oid_to_hex(&odb_parents
->item
->object
.oid
));
2577 generation
= commit_graph_generation(graph_parents
->item
);
2578 if (generation
> max_generation
)
2579 max_generation
= generation
;
2581 graph_parents
= graph_parents
->next
;
2582 odb_parents
= odb_parents
->next
;
2586 graph_report(_("commit-graph parent list for commit %s terminates early"),
2587 oid_to_hex(&cur_oid
));
2589 if (!commit_graph_generation(graph_commit
)) {
2590 if (generation_zero
== GENERATION_NUMBER_EXISTS
)
2591 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2592 oid_to_hex(&cur_oid
));
2593 generation_zero
= GENERATION_ZERO_EXISTS
;
2594 } else if (generation_zero
== GENERATION_ZERO_EXISTS
)
2595 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2596 oid_to_hex(&cur_oid
));
2598 if (generation_zero
== GENERATION_ZERO_EXISTS
)
2602 * If we are using topological level and one of our parents has
2603 * generation GENERATION_NUMBER_V1_MAX, then our generation is
2604 * also GENERATION_NUMBER_V1_MAX. Decrement to avoid extra logic
2605 * in the following condition.
2607 if (!g
->read_generation_data
&& max_generation
== GENERATION_NUMBER_V1_MAX
)
2610 generation
= commit_graph_generation(graph_commit
);
2611 if (generation
< max_generation
+ 1)
2612 graph_report(_("commit-graph generation for commit %s is %"PRItime
" < %"PRItime
),
2613 oid_to_hex(&cur_oid
),
2615 max_generation
+ 1);
2617 if (graph_commit
->date
!= odb_commit
->date
)
2618 graph_report(_("commit date for commit %s in commit-graph is %"PRItime
" != %"PRItime
),
2619 oid_to_hex(&cur_oid
),
2623 stop_progress(&progress
);
2625 local_error
= verify_commit_graph_error
;
2627 if (!(flags
& COMMIT_GRAPH_VERIFY_SHALLOW
) && g
->base_graph
)
2628 local_error
|= verify_commit_graph(r
, g
->base_graph
, flags
);
2633 void free_commit_graph(struct commit_graph
*g
)
2638 munmap((void *)g
->data
, g
->data_len
);
2642 free(g
->bloom_filter_settings
);
2646 void disable_commit_graph(struct repository
*r
)
2648 r
->commit_graph_disabled
= 1;