1 #include "git-compat-util.h"
11 #include "hash-lookup.h"
12 #include "commit-graph.h"
13 #include "object-file.h"
14 #include "object-store-ll.h"
15 #include "oid-array.h"
19 #include "replace-object.h"
22 #include "commit-slab.h"
24 #include "json-writer.h"
27 #include "chunk-format.h"
29 void git_test_write_commit_graph_or_die(void)
32 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH
, 0))
35 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS
, 0))
36 flags
= COMMIT_GRAPH_WRITE_BLOOM_FILTERS
;
38 if (write_commit_graph_reachable(the_repository
->objects
->odb
,
40 die("failed to write commit-graph under GIT_TEST_COMMIT_GRAPH");
43 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
44 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
45 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
46 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
47 #define GRAPH_CHUNKID_GENERATION_DATA 0x47444132 /* "GDA2" */
48 #define GRAPH_CHUNKID_GENERATION_DATA_OVERFLOW 0x47444f32 /* "GDO2" */
49 #define GRAPH_CHUNKID_EXTRAEDGES 0x45444745 /* "EDGE" */
50 #define GRAPH_CHUNKID_BLOOMINDEXES 0x42494458 /* "BIDX" */
51 #define GRAPH_CHUNKID_BLOOMDATA 0x42444154 /* "BDAT" */
52 #define GRAPH_CHUNKID_BASE 0x42415345 /* "BASE" */
54 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
56 #define GRAPH_VERSION_1 0x1
57 #define GRAPH_VERSION GRAPH_VERSION_1
59 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
60 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
61 #define GRAPH_PARENT_NONE 0x70000000
63 #define GRAPH_LAST_EDGE 0x80000000
65 #define GRAPH_HEADER_SIZE 8
66 #define GRAPH_FANOUT_SIZE (4 * 256)
67 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * CHUNK_TOC_ENTRY_SIZE \
68 + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
70 #define CORRECTED_COMMIT_DATE_OFFSET_OVERFLOW (1ULL << 31)
72 /* Remember to update object flag allocation in object.h */
73 #define REACHABLE (1u<<15)
75 define_commit_slab(topo_level_slab
, uint32_t);
77 /* Keep track of the order in which commits are added to our list. */
78 define_commit_slab(commit_pos
, int);
79 static struct commit_pos commit_pos
= COMMIT_SLAB_INIT(1, commit_pos
);
81 static void set_commit_pos(struct repository
*r
, const struct object_id
*oid
)
83 static int32_t max_pos
;
84 struct commit
*commit
= lookup_commit(r
, oid
);
87 return; /* should never happen, but be lenient */
89 *commit_pos_at(&commit_pos
, commit
) = max_pos
++;
92 static int commit_pos_cmp(const void *va
, const void *vb
)
94 const struct commit
*a
= *(const struct commit
**)va
;
95 const struct commit
*b
= *(const struct commit
**)vb
;
96 return commit_pos_at(&commit_pos
, a
) -
97 commit_pos_at(&commit_pos
, b
);
100 define_commit_slab(commit_graph_data_slab
, struct commit_graph_data
);
101 static struct commit_graph_data_slab commit_graph_data_slab
=
102 COMMIT_SLAB_INIT(1, commit_graph_data_slab
);
104 static int get_configured_generation_version(struct repository
*r
)
107 repo_config_get_int(r
, "commitgraph.generationversion", &version
);
111 uint32_t commit_graph_position(const struct commit
*c
)
113 struct commit_graph_data
*data
=
114 commit_graph_data_slab_peek(&commit_graph_data_slab
, c
);
116 return data
? data
->graph_pos
: COMMIT_NOT_FROM_GRAPH
;
119 timestamp_t
commit_graph_generation(const struct commit
*c
)
121 struct commit_graph_data
*data
=
122 commit_graph_data_slab_peek(&commit_graph_data_slab
, c
);
124 if (data
&& data
->generation
)
125 return data
->generation
;
127 return GENERATION_NUMBER_INFINITY
;
130 static timestamp_t
commit_graph_generation_from_graph(const struct commit
*c
)
132 struct commit_graph_data
*data
=
133 commit_graph_data_slab_peek(&commit_graph_data_slab
, c
);
135 if (!data
|| data
->graph_pos
== COMMIT_NOT_FROM_GRAPH
)
136 return GENERATION_NUMBER_INFINITY
;
137 return data
->generation
;
140 static struct commit_graph_data
*commit_graph_data_at(const struct commit
*c
)
142 unsigned int i
, nth_slab
;
143 struct commit_graph_data
*data
=
144 commit_graph_data_slab_peek(&commit_graph_data_slab
, c
);
149 nth_slab
= c
->index
/ commit_graph_data_slab
.slab_size
;
150 data
= commit_graph_data_slab_at(&commit_graph_data_slab
, c
);
153 * commit-slab initializes elements with zero, overwrite this with
154 * COMMIT_NOT_FROM_GRAPH for graph_pos.
156 * We avoid initializing generation with checking if graph position
157 * is not COMMIT_NOT_FROM_GRAPH.
159 for (i
= 0; i
< commit_graph_data_slab
.slab_size
; i
++) {
160 commit_graph_data_slab
.slab
[nth_slab
][i
].graph_pos
=
161 COMMIT_NOT_FROM_GRAPH
;
168 * Should be used only while writing commit-graph as it compares
169 * generation value of commits by directly accessing commit-slab.
171 static int commit_gen_cmp(const void *va
, const void *vb
)
173 const struct commit
*a
= *(const struct commit
**)va
;
174 const struct commit
*b
= *(const struct commit
**)vb
;
176 const timestamp_t generation_a
= commit_graph_data_at(a
)->generation
;
177 const timestamp_t generation_b
= commit_graph_data_at(b
)->generation
;
178 /* lower generation commits first */
179 if (generation_a
< generation_b
)
181 else if (generation_a
> generation_b
)
184 /* use date as a heuristic when generations are equal */
185 if (a
->date
< b
->date
)
187 else if (a
->date
> b
->date
)
192 char *get_commit_graph_filename(struct object_directory
*obj_dir
)
194 return xstrfmt("%s/info/commit-graph", obj_dir
->path
);
197 static char *get_split_graph_filename(struct object_directory
*odb
,
200 return xstrfmt("%s/info/commit-graphs/graph-%s.graph", odb
->path
,
204 char *get_commit_graph_chain_filename(struct object_directory
*odb
)
206 return xstrfmt("%s/info/commit-graphs/commit-graph-chain", odb
->path
);
209 static struct commit_graph
*alloc_commit_graph(void)
211 struct commit_graph
*g
= xcalloc(1, sizeof(*g
));
216 static int commit_graph_compatible(struct repository
*r
)
221 if (replace_refs_enabled(r
)) {
222 prepare_replace_object(r
);
223 if (hashmap_get_size(&r
->objects
->replace_map
->map
))
227 prepare_commit_graft(r
);
228 if (r
->parsed_objects
&&
229 (r
->parsed_objects
->grafts_nr
|| r
->parsed_objects
->substituted_parent
))
231 if (is_repository_shallow(r
))
237 int open_commit_graph(const char *graph_file
, int *fd
, struct stat
*st
)
239 *fd
= git_open(graph_file
);
242 if (fstat(*fd
, st
)) {
249 struct commit_graph
*load_commit_graph_one_fd_st(struct repository
*r
,
250 int fd
, struct stat
*st
,
251 struct object_directory
*odb
)
255 struct commit_graph
*ret
;
257 graph_size
= xsize_t(st
->st_size
);
259 if (graph_size
< GRAPH_MIN_SIZE
) {
261 error(_("commit-graph file is too small"));
264 graph_map
= xmmap(NULL
, graph_size
, PROT_READ
, MAP_PRIVATE
, fd
, 0);
266 prepare_repo_settings(r
);
267 ret
= parse_commit_graph(&r
->settings
, graph_map
, graph_size
);
272 munmap(graph_map
, graph_size
);
277 static int graph_read_oid_fanout(const unsigned char *chunk_start
,
278 size_t chunk_size
, void *data
)
280 struct commit_graph
*g
= data
;
283 if (chunk_size
!= 256 * sizeof(uint32_t))
284 return error(_("commit-graph oid fanout chunk is wrong size"));
285 g
->chunk_oid_fanout
= (const uint32_t *)chunk_start
;
286 g
->num_commits
= ntohl(g
->chunk_oid_fanout
[255]);
288 for (i
= 0; i
< 255; i
++) {
289 uint32_t oid_fanout1
= ntohl(g
->chunk_oid_fanout
[i
]);
290 uint32_t oid_fanout2
= ntohl(g
->chunk_oid_fanout
[i
+ 1]);
292 if (oid_fanout1
> oid_fanout2
) {
293 error(_("commit-graph fanout values out of order"));
301 static int graph_read_oid_lookup(const unsigned char *chunk_start
,
302 size_t chunk_size
, void *data
)
304 struct commit_graph
*g
= data
;
305 g
->chunk_oid_lookup
= chunk_start
;
306 if (chunk_size
/ g
->hash_len
!= g
->num_commits
)
307 return error(_("commit-graph OID lookup chunk is the wrong size"));
311 static int graph_read_commit_data(const unsigned char *chunk_start
,
312 size_t chunk_size
, void *data
)
314 struct commit_graph
*g
= data
;
315 if (chunk_size
/ GRAPH_DATA_WIDTH
!= g
->num_commits
)
316 return error(_("commit-graph commit data chunk is wrong size"));
317 g
->chunk_commit_data
= chunk_start
;
321 static int graph_read_generation_data(const unsigned char *chunk_start
,
322 size_t chunk_size
, void *data
)
324 struct commit_graph
*g
= data
;
325 if (chunk_size
/ sizeof(uint32_t) != g
->num_commits
)
326 return error(_("commit-graph generations chunk is wrong size"));
327 g
->chunk_generation_data
= chunk_start
;
331 static int graph_read_bloom_index(const unsigned char *chunk_start
,
332 size_t chunk_size
, void *data
)
334 struct commit_graph
*g
= data
;
335 if (chunk_size
/ 4 != g
->num_commits
) {
336 warning(_("commit-graph changed-path index chunk is too small"));
339 g
->chunk_bloom_indexes
= chunk_start
;
343 static int graph_read_bloom_data(const unsigned char *chunk_start
,
344 size_t chunk_size
, void *data
)
346 struct commit_graph
*g
= data
;
347 uint32_t hash_version
;
349 if (chunk_size
< BLOOMDATA_CHUNK_HEADER_SIZE
) {
350 warning(_("ignoring too-small changed-path chunk"
351 " (%"PRIuMAX
" < %"PRIuMAX
") in commit-graph file"),
352 (uintmax_t)chunk_size
,
353 (uintmax_t)BLOOMDATA_CHUNK_HEADER_SIZE
);
357 g
->chunk_bloom_data
= chunk_start
;
358 g
->chunk_bloom_data_size
= chunk_size
;
359 hash_version
= get_be32(chunk_start
);
361 if (hash_version
!= 1)
364 g
->bloom_filter_settings
= xmalloc(sizeof(struct bloom_filter_settings
));
365 g
->bloom_filter_settings
->hash_version
= hash_version
;
366 g
->bloom_filter_settings
->num_hashes
= get_be32(chunk_start
+ 4);
367 g
->bloom_filter_settings
->bits_per_entry
= get_be32(chunk_start
+ 8);
368 g
->bloom_filter_settings
->max_changed_paths
= DEFAULT_BLOOM_MAX_CHANGES
;
373 struct commit_graph
*parse_commit_graph(struct repo_settings
*s
,
374 void *graph_map
, size_t graph_size
)
376 const unsigned char *data
;
377 struct commit_graph
*graph
;
378 uint32_t graph_signature
;
379 unsigned char graph_version
, hash_version
;
380 struct chunkfile
*cf
= NULL
;
385 if (graph_size
< GRAPH_MIN_SIZE
)
388 data
= (const unsigned char *)graph_map
;
390 graph_signature
= get_be32(data
);
391 if (graph_signature
!= GRAPH_SIGNATURE
) {
392 error(_("commit-graph signature %X does not match signature %X"),
393 graph_signature
, GRAPH_SIGNATURE
);
397 graph_version
= *(unsigned char*)(data
+ 4);
398 if (graph_version
!= GRAPH_VERSION
) {
399 error(_("commit-graph version %X does not match version %X"),
400 graph_version
, GRAPH_VERSION
);
404 hash_version
= *(unsigned char*)(data
+ 5);
405 if (hash_version
!= oid_version(the_hash_algo
)) {
406 error(_("commit-graph hash version %X does not match version %X"),
407 hash_version
, oid_version(the_hash_algo
));
411 graph
= alloc_commit_graph();
413 graph
->hash_len
= the_hash_algo
->rawsz
;
414 graph
->num_chunks
= *(unsigned char*)(data
+ 6);
415 graph
->data
= graph_map
;
416 graph
->data_len
= graph_size
;
418 if (graph_size
< GRAPH_HEADER_SIZE
+
419 (graph
->num_chunks
+ 1) * CHUNK_TOC_ENTRY_SIZE
+
420 GRAPH_FANOUT_SIZE
+ the_hash_algo
->rawsz
) {
421 error(_("commit-graph file is too small to hold %u chunks"),
427 cf
= init_chunkfile(NULL
);
429 if (read_table_of_contents(cf
, graph
->data
, graph_size
,
430 GRAPH_HEADER_SIZE
, graph
->num_chunks
, 1))
431 goto free_and_return
;
433 if (read_chunk(cf
, GRAPH_CHUNKID_OIDFANOUT
, graph_read_oid_fanout
, graph
)) {
434 error(_("commit-graph required OID fanout chunk missing or corrupted"));
435 goto free_and_return
;
437 if (read_chunk(cf
, GRAPH_CHUNKID_OIDLOOKUP
, graph_read_oid_lookup
, graph
)) {
438 error(_("commit-graph required OID lookup chunk missing or corrupted"));
439 goto free_and_return
;
441 if (read_chunk(cf
, GRAPH_CHUNKID_DATA
, graph_read_commit_data
, graph
)) {
442 error(_("commit-graph required commit data chunk missing or corrupted"));
443 goto free_and_return
;
446 pair_chunk(cf
, GRAPH_CHUNKID_EXTRAEDGES
, &graph
->chunk_extra_edges
,
447 &graph
->chunk_extra_edges_size
);
448 pair_chunk(cf
, GRAPH_CHUNKID_BASE
, &graph
->chunk_base_graphs
,
449 &graph
->chunk_base_graphs_size
);
451 if (s
->commit_graph_generation_version
>= 2) {
452 read_chunk(cf
, GRAPH_CHUNKID_GENERATION_DATA
,
453 graph_read_generation_data
, graph
);
454 pair_chunk(cf
, GRAPH_CHUNKID_GENERATION_DATA_OVERFLOW
,
455 &graph
->chunk_generation_data_overflow
,
456 &graph
->chunk_generation_data_overflow_size
);
458 if (graph
->chunk_generation_data
)
459 graph
->read_generation_data
= 1;
462 if (s
->commit_graph_read_changed_paths
) {
463 read_chunk(cf
, GRAPH_CHUNKID_BLOOMINDEXES
,
464 graph_read_bloom_index
, graph
);
465 read_chunk(cf
, GRAPH_CHUNKID_BLOOMDATA
,
466 graph_read_bloom_data
, graph
);
469 if (graph
->chunk_bloom_indexes
&& graph
->chunk_bloom_data
) {
470 init_bloom_filters();
472 /* We need both the bloom chunks to exist together. Else ignore the data */
473 graph
->chunk_bloom_indexes
= NULL
;
474 graph
->chunk_bloom_data
= NULL
;
475 FREE_AND_NULL(graph
->bloom_filter_settings
);
478 oidread(&graph
->oid
, graph
->data
+ graph
->data_len
- graph
->hash_len
);
485 free(graph
->bloom_filter_settings
);
490 static struct commit_graph
*load_commit_graph_one(struct repository
*r
,
491 const char *graph_file
,
492 struct object_directory
*odb
)
497 struct commit_graph
*g
;
498 int open_ok
= open_commit_graph(graph_file
, &fd
, &st
);
503 g
= load_commit_graph_one_fd_st(r
, fd
, &st
, odb
);
506 g
->filename
= xstrdup(graph_file
);
511 static struct commit_graph
*load_commit_graph_v1(struct repository
*r
,
512 struct object_directory
*odb
)
514 char *graph_name
= get_commit_graph_filename(odb
);
515 struct commit_graph
*g
= load_commit_graph_one(r
, graph_name
, odb
);
522 * returns 1 if and only if all graphs in the chain have
523 * corrected commit dates stored in the generation_data chunk.
525 static int validate_mixed_generation_chain(struct commit_graph
*g
)
527 int read_generation_data
= 1;
528 struct commit_graph
*p
= g
;
530 while (read_generation_data
&& p
) {
531 read_generation_data
= p
->read_generation_data
;
535 if (read_generation_data
)
539 g
->read_generation_data
= 0;
546 static int add_graph_to_chain(struct commit_graph
*g
,
547 struct commit_graph
*chain
,
548 struct object_id
*oids
,
551 struct commit_graph
*cur_g
= chain
;
553 if (n
&& !g
->chunk_base_graphs
) {
554 warning(_("commit-graph has no base graphs chunk"));
558 if (g
->chunk_base_graphs_size
/ g
->hash_len
< n
) {
559 warning(_("commit-graph base graphs chunk is too small"));
567 !oideq(&oids
[n
], &cur_g
->oid
) ||
568 !hasheq(oids
[n
].hash
, g
->chunk_base_graphs
+ st_mult(g
->hash_len
, n
))) {
569 warning(_("commit-graph chain does not match"));
573 cur_g
= cur_g
->base_graph
;
577 if (unsigned_add_overflows(chain
->num_commits
,
578 chain
->num_commits_in_base
)) {
579 warning(_("commit count in base graph too high: %"PRIuMAX
),
580 (uintmax_t)chain
->num_commits_in_base
);
583 g
->num_commits_in_base
= chain
->num_commits
+ chain
->num_commits_in_base
;
586 g
->base_graph
= chain
;
591 int open_commit_graph_chain(const char *chain_file
,
592 int *fd
, struct stat
*st
)
594 *fd
= git_open(chain_file
);
597 if (fstat(*fd
, st
)) {
601 if (st
->st_size
< the_hash_algo
->hexsz
) {
604 /* treat empty files the same as missing */
607 warning(_("commit-graph chain file too small"));
615 struct commit_graph
*load_commit_graph_chain_fd_st(struct repository
*r
,
616 int fd
, struct stat
*st
,
617 int *incomplete_chain
)
619 struct commit_graph
*graph_chain
= NULL
;
620 struct strbuf line
= STRBUF_INIT
;
621 struct object_id
*oids
;
622 int i
= 0, valid
= 1, count
;
623 FILE *fp
= xfdopen(fd
, "r");
625 count
= st
->st_size
/ (the_hash_algo
->hexsz
+ 1);
626 CALLOC_ARRAY(oids
, count
);
630 for (i
= 0; i
< count
; i
++) {
631 struct object_directory
*odb
;
633 if (strbuf_getline_lf(&line
, fp
) == EOF
)
636 if (get_oid_hex(line
.buf
, &oids
[i
])) {
637 warning(_("invalid commit-graph chain: line '%s' not a hash"),
644 for (odb
= r
->objects
->odb
; odb
; odb
= odb
->next
) {
645 char *graph_name
= get_split_graph_filename(odb
, line
.buf
);
646 struct commit_graph
*g
= load_commit_graph_one(r
, graph_name
, odb
);
651 if (add_graph_to_chain(g
, graph_chain
, oids
, i
)) {
655 free_commit_graph(g
);
663 warning(_("unable to find all commit-graph files"));
668 validate_mixed_generation_chain(graph_chain
);
672 strbuf_release(&line
);
674 *incomplete_chain
= !valid
;
678 static struct commit_graph
*load_commit_graph_chain(struct repository
*r
,
679 struct object_directory
*odb
)
681 char *chain_file
= get_commit_graph_chain_filename(odb
);
684 struct commit_graph
*g
= NULL
;
686 if (open_commit_graph_chain(chain_file
, &fd
, &st
)) {
688 /* ownership of fd is taken over by load function */
689 g
= load_commit_graph_chain_fd_st(r
, fd
, &st
, &incomplete
);
696 struct commit_graph
*read_commit_graph_one(struct repository
*r
,
697 struct object_directory
*odb
)
699 struct commit_graph
*g
= load_commit_graph_v1(r
, odb
);
702 g
= load_commit_graph_chain(r
, odb
);
707 static void prepare_commit_graph_one(struct repository
*r
,
708 struct object_directory
*odb
)
711 if (r
->objects
->commit_graph
)
714 r
->objects
->commit_graph
= read_commit_graph_one(r
, odb
);
718 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
720 * On the first invocation, this function attempts to load the commit
721 * graph if the_repository is configured to have one.
723 static int prepare_commit_graph(struct repository
*r
)
725 struct object_directory
*odb
;
728 * Early return if there is no git dir or if the commit graph is
731 * This must come before the "already attempted?" check below, because
732 * we want to disable even an already-loaded graph file.
734 if (!r
->gitdir
|| r
->commit_graph_disabled
)
737 if (r
->objects
->commit_graph_attempted
)
738 return !!r
->objects
->commit_graph
;
739 r
->objects
->commit_graph_attempted
= 1;
741 prepare_repo_settings(r
);
743 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH
, 0) &&
744 r
->settings
.core_commit_graph
!= 1)
746 * This repository is not configured to use commit graphs, so
747 * do not load one. (But report commit_graph_attempted anyway
748 * so that commit graph loading is not attempted again for this
753 if (!commit_graph_compatible(r
))
757 for (odb
= r
->objects
->odb
;
758 !r
->objects
->commit_graph
&& odb
;
760 prepare_commit_graph_one(r
, odb
);
761 return !!r
->objects
->commit_graph
;
764 int generation_numbers_enabled(struct repository
*r
)
766 uint32_t first_generation
;
767 struct commit_graph
*g
;
768 if (!prepare_commit_graph(r
))
771 g
= r
->objects
->commit_graph
;
776 first_generation
= get_be32(g
->chunk_commit_data
+
777 g
->hash_len
+ 8) >> 2;
779 return !!first_generation
;
782 int corrected_commit_dates_enabled(struct repository
*r
)
784 struct commit_graph
*g
;
785 if (!prepare_commit_graph(r
))
788 g
= r
->objects
->commit_graph
;
793 return g
->read_generation_data
;
796 struct bloom_filter_settings
*get_bloom_filter_settings(struct repository
*r
)
798 struct commit_graph
*g
= r
->objects
->commit_graph
;
800 if (g
->bloom_filter_settings
)
801 return g
->bloom_filter_settings
;
807 void close_commit_graph(struct raw_object_store
*o
)
809 if (!o
->commit_graph
)
812 clear_commit_graph_data_slab(&commit_graph_data_slab
);
813 free_commit_graph(o
->commit_graph
);
814 o
->commit_graph
= NULL
;
817 static int bsearch_graph(struct commit_graph
*g
, const struct object_id
*oid
, uint32_t *pos
)
819 return bsearch_hash(oid
->hash
, g
->chunk_oid_fanout
,
820 g
->chunk_oid_lookup
, g
->hash_len
, pos
);
823 static void load_oid_from_graph(struct commit_graph
*g
,
825 struct object_id
*oid
)
829 while (g
&& pos
< g
->num_commits_in_base
)
833 BUG("NULL commit-graph");
835 if (pos
>= g
->num_commits
+ g
->num_commits_in_base
)
836 die(_("invalid commit position. commit-graph is likely corrupt"));
838 lex_index
= pos
- g
->num_commits_in_base
;
840 oidread(oid
, g
->chunk_oid_lookup
+ st_mult(g
->hash_len
, lex_index
));
843 static struct commit_list
**insert_parent_or_die(struct repository
*r
,
844 struct commit_graph
*g
,
846 struct commit_list
**pptr
)
849 struct object_id oid
;
851 if (pos
>= g
->num_commits
+ g
->num_commits_in_base
)
852 die("invalid parent position %"PRIu32
, pos
);
854 load_oid_from_graph(g
, pos
, &oid
);
855 c
= lookup_commit(r
, &oid
);
857 die(_("could not find commit %s"), oid_to_hex(&oid
));
858 commit_graph_data_at(c
)->graph_pos
= pos
;
859 return &commit_list_insert(c
, pptr
)->next
;
862 static void fill_commit_graph_info(struct commit
*item
, struct commit_graph
*g
, uint32_t pos
)
864 const unsigned char *commit_data
;
865 struct commit_graph_data
*graph_data
;
866 uint32_t lex_index
, offset_pos
;
867 uint64_t date_high
, date_low
, offset
;
869 while (pos
< g
->num_commits_in_base
)
872 if (pos
>= g
->num_commits
+ g
->num_commits_in_base
)
873 die(_("invalid commit position. commit-graph is likely corrupt"));
875 lex_index
= pos
- g
->num_commits_in_base
;
876 commit_data
= g
->chunk_commit_data
+ st_mult(GRAPH_DATA_WIDTH
, lex_index
);
878 graph_data
= commit_graph_data_at(item
);
879 graph_data
->graph_pos
= pos
;
881 date_high
= get_be32(commit_data
+ g
->hash_len
+ 8) & 0x3;
882 date_low
= get_be32(commit_data
+ g
->hash_len
+ 12);
883 item
->date
= (timestamp_t
)((date_high
<< 32) | date_low
);
885 if (g
->read_generation_data
) {
886 offset
= (timestamp_t
)get_be32(g
->chunk_generation_data
+ st_mult(sizeof(uint32_t), lex_index
));
888 if (offset
& CORRECTED_COMMIT_DATE_OFFSET_OVERFLOW
) {
889 if (!g
->chunk_generation_data_overflow
)
890 die(_("commit-graph requires overflow generation data but has none"));
892 offset_pos
= offset
^ CORRECTED_COMMIT_DATE_OFFSET_OVERFLOW
;
893 if (g
->chunk_generation_data_overflow_size
/ sizeof(uint64_t) <= offset_pos
)
894 die(_("commit-graph overflow generation data is too small"));
895 graph_data
->generation
= item
->date
+
896 get_be64(g
->chunk_generation_data_overflow
+ sizeof(uint64_t) * offset_pos
);
898 graph_data
->generation
= item
->date
+ offset
;
900 graph_data
->generation
= get_be32(commit_data
+ g
->hash_len
+ 8) >> 2;
903 *topo_level_slab_at(g
->topo_levels
, item
) = get_be32(commit_data
+ g
->hash_len
+ 8) >> 2;
906 static inline void set_commit_tree(struct commit
*c
, struct tree
*t
)
911 static int fill_commit_in_graph(struct repository
*r
,
913 struct commit_graph
*g
, uint32_t pos
)
916 uint32_t parent_data_pos
;
917 struct commit_list
**pptr
;
918 const unsigned char *commit_data
;
921 while (pos
< g
->num_commits_in_base
)
924 fill_commit_graph_info(item
, g
, pos
);
926 lex_index
= pos
- g
->num_commits_in_base
;
927 commit_data
= g
->chunk_commit_data
+ st_mult(g
->hash_len
+ 16, lex_index
);
929 item
->object
.parsed
= 1;
931 set_commit_tree(item
, NULL
);
933 pptr
= &item
->parents
;
935 edge_value
= get_be32(commit_data
+ g
->hash_len
);
936 if (edge_value
== GRAPH_PARENT_NONE
)
938 pptr
= insert_parent_or_die(r
, g
, edge_value
, pptr
);
940 edge_value
= get_be32(commit_data
+ g
->hash_len
+ 4);
941 if (edge_value
== GRAPH_PARENT_NONE
)
943 if (!(edge_value
& GRAPH_EXTRA_EDGES_NEEDED
)) {
944 pptr
= insert_parent_or_die(r
, g
, edge_value
, pptr
);
948 parent_data_pos
= edge_value
& GRAPH_EDGE_LAST_MASK
;
950 if (g
->chunk_extra_edges_size
/ sizeof(uint32_t) <= parent_data_pos
) {
951 error(_("commit-graph extra-edges pointer out of bounds"));
952 free_commit_list(item
->parents
);
953 item
->parents
= NULL
;
954 item
->object
.parsed
= 0;
957 edge_value
= get_be32(g
->chunk_extra_edges
+
958 sizeof(uint32_t) * parent_data_pos
);
959 pptr
= insert_parent_or_die(r
, g
,
960 edge_value
& GRAPH_EDGE_LAST_MASK
,
963 } while (!(edge_value
& GRAPH_LAST_EDGE
));
968 static int search_commit_pos_in_graph(const struct object_id
*id
, struct commit_graph
*g
, uint32_t *pos
)
970 struct commit_graph
*cur_g
= g
;
973 while (cur_g
&& !bsearch_graph(cur_g
, id
, &lex_index
))
974 cur_g
= cur_g
->base_graph
;
977 *pos
= lex_index
+ cur_g
->num_commits_in_base
;
984 static int find_commit_pos_in_graph(struct commit
*item
, struct commit_graph
*g
, uint32_t *pos
)
986 uint32_t graph_pos
= commit_graph_position(item
);
987 if (graph_pos
!= COMMIT_NOT_FROM_GRAPH
) {
991 return search_commit_pos_in_graph(&item
->object
.oid
, g
, pos
);
995 int repo_find_commit_pos_in_graph(struct repository
*r
, struct commit
*c
,
998 if (!prepare_commit_graph(r
))
1000 return find_commit_pos_in_graph(c
, r
->objects
->commit_graph
, pos
);
1003 struct commit
*lookup_commit_in_graph(struct repository
*repo
, const struct object_id
*id
)
1005 static int commit_graph_paranoia
= -1;
1006 struct commit
*commit
;
1009 if (commit_graph_paranoia
== -1)
1010 commit_graph_paranoia
= git_env_bool(GIT_COMMIT_GRAPH_PARANOIA
, 0);
1012 if (!prepare_commit_graph(repo
))
1014 if (!search_commit_pos_in_graph(id
, repo
->objects
->commit_graph
, &pos
))
1016 if (commit_graph_paranoia
&& !has_object(repo
, id
, 0))
1019 commit
= lookup_commit(repo
, id
);
1022 if (commit
->object
.parsed
)
1025 if (!fill_commit_in_graph(repo
, commit
, repo
->objects
->commit_graph
, pos
))
1031 static int parse_commit_in_graph_one(struct repository
*r
,
1032 struct commit_graph
*g
,
1033 struct commit
*item
)
1037 if (item
->object
.parsed
)
1040 if (find_commit_pos_in_graph(item
, g
, &pos
))
1041 return fill_commit_in_graph(r
, item
, g
, pos
);
1046 int parse_commit_in_graph(struct repository
*r
, struct commit
*item
)
1048 static int checked_env
= 0;
1051 git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_PARSE
, 0))
1052 die("dying as requested by the '%s' variable on commit-graph parse!",
1053 GIT_TEST_COMMIT_GRAPH_DIE_ON_PARSE
);
1056 if (!prepare_commit_graph(r
))
1058 return parse_commit_in_graph_one(r
, r
->objects
->commit_graph
, item
);
1061 void load_commit_graph_info(struct repository
*r
, struct commit
*item
)
1064 if (repo_find_commit_pos_in_graph(r
, item
, &pos
))
1065 fill_commit_graph_info(item
, r
->objects
->commit_graph
, pos
);
1068 static struct tree
*load_tree_for_commit(struct repository
*r
,
1069 struct commit_graph
*g
,
1072 struct object_id oid
;
1073 const unsigned char *commit_data
;
1074 uint32_t graph_pos
= commit_graph_position(c
);
1076 while (graph_pos
< g
->num_commits_in_base
)
1079 commit_data
= g
->chunk_commit_data
+
1080 st_mult(GRAPH_DATA_WIDTH
, graph_pos
- g
->num_commits_in_base
);
1082 oidread(&oid
, commit_data
);
1083 set_commit_tree(c
, lookup_tree(r
, &oid
));
1085 return c
->maybe_tree
;
1088 static struct tree
*get_commit_tree_in_graph_one(struct repository
*r
,
1089 struct commit_graph
*g
,
1090 const struct commit
*c
)
1093 return c
->maybe_tree
;
1094 if (commit_graph_position(c
) == COMMIT_NOT_FROM_GRAPH
)
1095 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
1097 return load_tree_for_commit(r
, g
, (struct commit
*)c
);
1100 struct tree
*get_commit_tree_in_graph(struct repository
*r
, const struct commit
*c
)
1102 return get_commit_tree_in_graph_one(r
, r
->objects
->commit_graph
, c
);
1105 struct packed_commit_list
{
1106 struct commit
**list
;
1111 struct write_commit_graph_context
{
1112 struct repository
*r
;
1113 struct object_directory
*odb
;
1115 struct oid_array oids
;
1116 struct packed_commit_list commits
;
1117 int num_extra_edges
;
1118 int num_generation_data_overflows
;
1119 unsigned long approx_nr_objects
;
1120 struct progress
*progress
;
1122 uint64_t progress_cnt
;
1124 char *base_graph_name
;
1125 int num_commit_graphs_before
;
1126 int num_commit_graphs_after
;
1127 char **commit_graph_filenames_before
;
1128 char **commit_graph_filenames_after
;
1129 char **commit_graph_hash_after
;
1130 uint32_t new_num_commits_in_base
;
1131 struct commit_graph
*new_base_graph
;
1138 write_generation_data
:1,
1139 trust_generation_numbers
:1;
1141 struct topo_level_slab
*topo_levels
;
1142 const struct commit_graph_opts
*opts
;
1143 size_t total_bloom_filter_data_size
;
1144 const struct bloom_filter_settings
*bloom_settings
;
1146 int count_bloom_filter_computed
;
1147 int count_bloom_filter_not_computed
;
1148 int count_bloom_filter_trunc_empty
;
1149 int count_bloom_filter_trunc_large
;
1152 static int write_graph_chunk_fanout(struct hashfile
*f
,
1155 struct write_commit_graph_context
*ctx
= data
;
1157 struct commit
**list
= ctx
->commits
.list
;
1160 * Write the first-level table (the list is sorted,
1161 * but we use a 256-entry lookup to be able to avoid
1162 * having to do eight extra binary search iterations).
1164 for (i
= 0; i
< 256; i
++) {
1165 while (count
< ctx
->commits
.nr
) {
1166 if ((*list
)->object
.oid
.hash
[0] != i
)
1168 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1173 hashwrite_be32(f
, count
);
1179 static int write_graph_chunk_oids(struct hashfile
*f
,
1182 struct write_commit_graph_context
*ctx
= data
;
1183 struct commit
**list
= ctx
->commits
.list
;
1185 for (count
= 0; count
< ctx
->commits
.nr
; count
++, list
++) {
1186 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1187 hashwrite(f
, (*list
)->object
.oid
.hash
, the_hash_algo
->rawsz
);
1193 static const struct object_id
*commit_to_oid(size_t index
, const void *table
)
1195 const struct commit
* const *commits
= table
;
1196 return &commits
[index
]->object
.oid
;
1199 static int write_graph_chunk_data(struct hashfile
*f
,
1202 struct write_commit_graph_context
*ctx
= data
;
1203 struct commit
**list
= ctx
->commits
.list
;
1204 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
1205 uint32_t num_extra_edges
= 0;
1207 while (list
< last
) {
1208 struct commit_list
*parent
;
1209 struct object_id
*tree
;
1211 uint32_t packedDate
[2];
1212 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1214 if (repo_parse_commit_no_graph(ctx
->r
, *list
))
1215 die(_("unable to parse commit %s"),
1216 oid_to_hex(&(*list
)->object
.oid
));
1217 tree
= get_commit_tree_oid(*list
);
1218 hashwrite(f
, tree
->hash
, the_hash_algo
->rawsz
);
1220 parent
= (*list
)->parents
;
1223 edge_value
= GRAPH_PARENT_NONE
;
1225 edge_value
= oid_pos(&parent
->item
->object
.oid
,
1230 if (edge_value
>= 0)
1231 edge_value
+= ctx
->new_num_commits_in_base
;
1232 else if (ctx
->new_base_graph
) {
1234 if (find_commit_pos_in_graph(parent
->item
,
1235 ctx
->new_base_graph
,
1241 BUG("missing parent %s for commit %s",
1242 oid_to_hex(&parent
->item
->object
.oid
),
1243 oid_to_hex(&(*list
)->object
.oid
));
1246 hashwrite_be32(f
, edge_value
);
1249 parent
= parent
->next
;
1252 edge_value
= GRAPH_PARENT_NONE
;
1253 else if (parent
->next
)
1254 edge_value
= GRAPH_EXTRA_EDGES_NEEDED
| num_extra_edges
;
1256 edge_value
= oid_pos(&parent
->item
->object
.oid
,
1261 if (edge_value
>= 0)
1262 edge_value
+= ctx
->new_num_commits_in_base
;
1263 else if (ctx
->new_base_graph
) {
1265 if (find_commit_pos_in_graph(parent
->item
,
1266 ctx
->new_base_graph
,
1272 BUG("missing parent %s for commit %s",
1273 oid_to_hex(&parent
->item
->object
.oid
),
1274 oid_to_hex(&(*list
)->object
.oid
));
1277 hashwrite_be32(f
, edge_value
);
1279 if (edge_value
& GRAPH_EXTRA_EDGES_NEEDED
) {
1282 parent
= parent
->next
;
1286 if (sizeof((*list
)->date
) > 4)
1287 packedDate
[0] = htonl(((*list
)->date
>> 32) & 0x3);
1291 packedDate
[0] |= htonl(*topo_level_slab_at(ctx
->topo_levels
, *list
) << 2);
1293 packedDate
[1] = htonl((*list
)->date
);
1294 hashwrite(f
, packedDate
, 8);
1302 static int write_graph_chunk_generation_data(struct hashfile
*f
,
1305 struct write_commit_graph_context
*ctx
= data
;
1306 int i
, num_generation_data_overflows
= 0;
1308 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1309 struct commit
*c
= ctx
->commits
.list
[i
];
1311 repo_parse_commit(ctx
->r
, c
);
1312 offset
= commit_graph_data_at(c
)->generation
- c
->date
;
1313 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1315 if (offset
> GENERATION_NUMBER_V2_OFFSET_MAX
) {
1316 offset
= CORRECTED_COMMIT_DATE_OFFSET_OVERFLOW
| num_generation_data_overflows
;
1317 num_generation_data_overflows
++;
1320 hashwrite_be32(f
, offset
);
1326 static int write_graph_chunk_generation_data_overflow(struct hashfile
*f
,
1329 struct write_commit_graph_context
*ctx
= data
;
1331 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1332 struct commit
*c
= ctx
->commits
.list
[i
];
1333 timestamp_t offset
= commit_graph_data_at(c
)->generation
- c
->date
;
1334 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1336 if (offset
> GENERATION_NUMBER_V2_OFFSET_MAX
) {
1337 hashwrite_be32(f
, offset
>> 32);
1338 hashwrite_be32(f
, (uint32_t) offset
);
1345 static int write_graph_chunk_extra_edges(struct hashfile
*f
,
1348 struct write_commit_graph_context
*ctx
= data
;
1349 struct commit
**list
= ctx
->commits
.list
;
1350 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
1351 struct commit_list
*parent
;
1353 while (list
< last
) {
1354 int num_parents
= 0;
1356 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1358 for (parent
= (*list
)->parents
; num_parents
< 3 && parent
;
1359 parent
= parent
->next
)
1362 if (num_parents
<= 2) {
1367 /* Since num_parents > 2, this initializer is safe. */
1368 for (parent
= (*list
)->parents
->next
; parent
; parent
= parent
->next
) {
1369 int edge_value
= oid_pos(&parent
->item
->object
.oid
,
1374 if (edge_value
>= 0)
1375 edge_value
+= ctx
->new_num_commits_in_base
;
1376 else if (ctx
->new_base_graph
) {
1378 if (find_commit_pos_in_graph(parent
->item
,
1379 ctx
->new_base_graph
,
1385 BUG("missing parent %s for commit %s",
1386 oid_to_hex(&parent
->item
->object
.oid
),
1387 oid_to_hex(&(*list
)->object
.oid
));
1388 else if (!parent
->next
)
1389 edge_value
|= GRAPH_LAST_EDGE
;
1391 hashwrite_be32(f
, edge_value
);
1400 static int write_graph_chunk_bloom_indexes(struct hashfile
*f
,
1403 struct write_commit_graph_context
*ctx
= data
;
1404 struct commit
**list
= ctx
->commits
.list
;
1405 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
1406 uint32_t cur_pos
= 0;
1408 while (list
< last
) {
1409 struct bloom_filter
*filter
= get_bloom_filter(ctx
->r
, *list
);
1410 size_t len
= filter
? filter
->len
: 0;
1412 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1413 hashwrite_be32(f
, cur_pos
);
1420 static void trace2_bloom_filter_settings(struct write_commit_graph_context
*ctx
)
1422 struct json_writer jw
= JSON_WRITER_INIT
;
1424 jw_object_begin(&jw
, 0);
1425 jw_object_intmax(&jw
, "hash_version", ctx
->bloom_settings
->hash_version
);
1426 jw_object_intmax(&jw
, "num_hashes", ctx
->bloom_settings
->num_hashes
);
1427 jw_object_intmax(&jw
, "bits_per_entry", ctx
->bloom_settings
->bits_per_entry
);
1428 jw_object_intmax(&jw
, "max_changed_paths", ctx
->bloom_settings
->max_changed_paths
);
1431 trace2_data_json("bloom", ctx
->r
, "settings", &jw
);
1436 static int write_graph_chunk_bloom_data(struct hashfile
*f
,
1439 struct write_commit_graph_context
*ctx
= data
;
1440 struct commit
**list
= ctx
->commits
.list
;
1441 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
1443 trace2_bloom_filter_settings(ctx
);
1445 hashwrite_be32(f
, ctx
->bloom_settings
->hash_version
);
1446 hashwrite_be32(f
, ctx
->bloom_settings
->num_hashes
);
1447 hashwrite_be32(f
, ctx
->bloom_settings
->bits_per_entry
);
1449 while (list
< last
) {
1450 struct bloom_filter
*filter
= get_bloom_filter(ctx
->r
, *list
);
1451 size_t len
= filter
? filter
->len
: 0;
1453 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1455 hashwrite(f
, filter
->data
, len
* sizeof(unsigned char));
1462 static int add_packed_commits(const struct object_id
*oid
,
1463 struct packed_git
*pack
,
1467 struct write_commit_graph_context
*ctx
= (struct write_commit_graph_context
*)data
;
1468 enum object_type type
;
1469 off_t offset
= nth_packed_object_offset(pack
, pos
);
1470 struct object_info oi
= OBJECT_INFO_INIT
;
1473 display_progress(ctx
->progress
, ++ctx
->progress_done
);
1476 if (packed_object_info(ctx
->r
, pack
, offset
, &oi
) < 0)
1477 die(_("unable to get type of object %s"), oid_to_hex(oid
));
1479 if (type
!= OBJ_COMMIT
)
1482 oid_array_append(&ctx
->oids
, oid
);
1483 set_commit_pos(ctx
->r
, oid
);
1488 static void add_missing_parents(struct write_commit_graph_context
*ctx
, struct commit
*commit
)
1490 struct commit_list
*parent
;
1491 for (parent
= commit
->parents
; parent
; parent
= parent
->next
) {
1492 if (!(parent
->item
->object
.flags
& REACHABLE
)) {
1493 oid_array_append(&ctx
->oids
, &parent
->item
->object
.oid
);
1494 parent
->item
->object
.flags
|= REACHABLE
;
1499 static void close_reachable(struct write_commit_graph_context
*ctx
)
1502 struct commit
*commit
;
1503 enum commit_graph_split_flags flags
= ctx
->opts
?
1504 ctx
->opts
->split_flags
: COMMIT_GRAPH_SPLIT_UNSPECIFIED
;
1506 if (ctx
->report_progress
)
1507 ctx
->progress
= start_delayed_progress(
1508 _("Loading known commits in commit graph"),
1510 for (i
= 0; i
< ctx
->oids
.nr
; i
++) {
1511 display_progress(ctx
->progress
, i
+ 1);
1512 commit
= lookup_commit(ctx
->r
, &ctx
->oids
.oid
[i
]);
1514 commit
->object
.flags
|= REACHABLE
;
1516 stop_progress(&ctx
->progress
);
1519 * As this loop runs, ctx->oids.nr may grow, but not more
1520 * than the number of missing commits in the reachable
1523 if (ctx
->report_progress
)
1524 ctx
->progress
= start_delayed_progress(
1525 _("Expanding reachable commits in commit graph"),
1527 for (i
= 0; i
< ctx
->oids
.nr
; i
++) {
1528 display_progress(ctx
->progress
, i
+ 1);
1529 commit
= lookup_commit(ctx
->r
, &ctx
->oids
.oid
[i
]);
1534 if ((!repo_parse_commit(ctx
->r
, commit
) &&
1535 commit_graph_position(commit
) == COMMIT_NOT_FROM_GRAPH
) ||
1536 flags
== COMMIT_GRAPH_SPLIT_REPLACE
)
1537 add_missing_parents(ctx
, commit
);
1538 } else if (!repo_parse_commit_no_graph(ctx
->r
, commit
))
1539 add_missing_parents(ctx
, commit
);
1541 stop_progress(&ctx
->progress
);
1543 if (ctx
->report_progress
)
1544 ctx
->progress
= start_delayed_progress(
1545 _("Clearing commit marks in commit graph"),
1547 for (i
= 0; i
< ctx
->oids
.nr
; i
++) {
1548 display_progress(ctx
->progress
, i
+ 1);
1549 commit
= lookup_commit(ctx
->r
, &ctx
->oids
.oid
[i
]);
1552 commit
->object
.flags
&= ~REACHABLE
;
1554 stop_progress(&ctx
->progress
);
1557 struct compute_generation_info
{
1558 struct repository
*r
;
1559 struct packed_commit_list
*commits
;
1560 struct progress
*progress
;
1563 timestamp_t (*get_generation
)(struct commit
*c
, void *data
);
1564 void (*set_generation
)(struct commit
*c
, timestamp_t gen
, void *data
);
1568 static timestamp_t
compute_generation_from_max(struct commit
*c
,
1569 timestamp_t max_gen
,
1570 int generation_version
)
1572 switch (generation_version
) {
1573 case 1: /* topological levels */
1574 if (max_gen
> GENERATION_NUMBER_V1_MAX
- 1)
1575 max_gen
= GENERATION_NUMBER_V1_MAX
- 1;
1578 case 2: /* corrected commit date */
1579 if (c
->date
&& c
->date
> max_gen
)
1580 max_gen
= c
->date
- 1;
1584 BUG("attempting unimplemented version");
1588 static void compute_reachable_generation_numbers(
1589 struct compute_generation_info
*info
,
1590 int generation_version
)
1593 struct commit_list
*list
= NULL
;
1595 for (i
= 0; i
< info
->commits
->nr
; i
++) {
1596 struct commit
*c
= info
->commits
->list
[i
];
1598 repo_parse_commit(info
->r
, c
);
1599 gen
= info
->get_generation(c
, info
->data
);
1600 display_progress(info
->progress
, info
->progress_cnt
+ 1);
1602 if (gen
!= GENERATION_NUMBER_ZERO
&& gen
!= GENERATION_NUMBER_INFINITY
)
1605 commit_list_insert(c
, &list
);
1607 struct commit
*current
= list
->item
;
1608 struct commit_list
*parent
;
1609 int all_parents_computed
= 1;
1610 uint32_t max_gen
= 0;
1612 for (parent
= current
->parents
; parent
; parent
= parent
->next
) {
1613 repo_parse_commit(info
->r
, parent
->item
);
1614 gen
= info
->get_generation(parent
->item
, info
->data
);
1616 if (gen
== GENERATION_NUMBER_ZERO
) {
1617 all_parents_computed
= 0;
1618 commit_list_insert(parent
->item
, &list
);
1626 if (all_parents_computed
) {
1628 gen
= compute_generation_from_max(
1630 generation_version
);
1631 info
->set_generation(current
, gen
, info
->data
);
1637 static timestamp_t
get_topo_level(struct commit
*c
, void *data
)
1639 struct write_commit_graph_context
*ctx
= data
;
1640 return *topo_level_slab_at(ctx
->topo_levels
, c
);
1643 static void set_topo_level(struct commit
*c
, timestamp_t t
, void *data
)
1645 struct write_commit_graph_context
*ctx
= data
;
1646 *topo_level_slab_at(ctx
->topo_levels
, c
) = (uint32_t)t
;
1649 static void compute_topological_levels(struct write_commit_graph_context
*ctx
)
1651 struct compute_generation_info info
= {
1653 .commits
= &ctx
->commits
,
1654 .get_generation
= get_topo_level
,
1655 .set_generation
= set_topo_level
,
1659 if (ctx
->report_progress
)
1660 info
.progress
= ctx
->progress
1661 = start_delayed_progress(
1662 _("Computing commit graph topological levels"),
1665 compute_reachable_generation_numbers(&info
, 1);
1667 stop_progress(&ctx
->progress
);
1670 static timestamp_t
get_generation_from_graph_data(struct commit
*c
,
1673 return commit_graph_data_at(c
)->generation
;
1676 static void set_generation_v2(struct commit
*c
, timestamp_t t
,
1679 struct commit_graph_data
*g
= commit_graph_data_at(c
);
1683 static void compute_generation_numbers(struct write_commit_graph_context
*ctx
)
1686 struct compute_generation_info info
= {
1688 .commits
= &ctx
->commits
,
1689 .get_generation
= get_generation_from_graph_data
,
1690 .set_generation
= set_generation_v2
,
1693 if (ctx
->report_progress
)
1694 info
.progress
= ctx
->progress
1695 = start_delayed_progress(
1696 _("Computing commit graph generation numbers"),
1699 if (!ctx
->trust_generation_numbers
) {
1700 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1701 struct commit
*c
= ctx
->commits
.list
[i
];
1702 repo_parse_commit(ctx
->r
, c
);
1703 commit_graph_data_at(c
)->generation
= GENERATION_NUMBER_ZERO
;
1707 compute_reachable_generation_numbers(&info
, 2);
1709 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1710 struct commit
*c
= ctx
->commits
.list
[i
];
1711 timestamp_t offset
= commit_graph_data_at(c
)->generation
- c
->date
;
1712 if (offset
> GENERATION_NUMBER_V2_OFFSET_MAX
)
1713 ctx
->num_generation_data_overflows
++;
1715 stop_progress(&ctx
->progress
);
1718 static void set_generation_in_graph_data(struct commit
*c
, timestamp_t t
,
1721 commit_graph_data_at(c
)->generation
= t
;
1725 * After this method, all commits reachable from those in the given
1726 * list will have non-zero, non-infinite generation numbers.
1728 void ensure_generations_valid(struct repository
*r
,
1729 struct commit
**commits
, size_t nr
)
1731 int generation_version
= get_configured_generation_version(r
);
1732 struct packed_commit_list list
= {
1737 struct compute_generation_info info
= {
1740 .get_generation
= get_generation_from_graph_data
,
1741 .set_generation
= set_generation_in_graph_data
,
1744 compute_reachable_generation_numbers(&info
, generation_version
);
1747 static void trace2_bloom_filter_write_statistics(struct write_commit_graph_context
*ctx
)
1749 trace2_data_intmax("commit-graph", ctx
->r
, "filter-computed",
1750 ctx
->count_bloom_filter_computed
);
1751 trace2_data_intmax("commit-graph", ctx
->r
, "filter-not-computed",
1752 ctx
->count_bloom_filter_not_computed
);
1753 trace2_data_intmax("commit-graph", ctx
->r
, "filter-trunc-empty",
1754 ctx
->count_bloom_filter_trunc_empty
);
1755 trace2_data_intmax("commit-graph", ctx
->r
, "filter-trunc-large",
1756 ctx
->count_bloom_filter_trunc_large
);
1759 static void compute_bloom_filters(struct write_commit_graph_context
*ctx
)
1762 struct progress
*progress
= NULL
;
1763 struct commit
**sorted_commits
;
1764 int max_new_filters
;
1766 init_bloom_filters();
1768 if (ctx
->report_progress
)
1769 progress
= start_delayed_progress(
1770 _("Computing commit changed paths Bloom filters"),
1773 DUP_ARRAY(sorted_commits
, ctx
->commits
.list
, ctx
->commits
.nr
);
1775 if (ctx
->order_by_pack
)
1776 QSORT(sorted_commits
, ctx
->commits
.nr
, commit_pos_cmp
);
1778 QSORT(sorted_commits
, ctx
->commits
.nr
, commit_gen_cmp
);
1780 max_new_filters
= ctx
->opts
&& ctx
->opts
->max_new_filters
>= 0 ?
1781 ctx
->opts
->max_new_filters
: ctx
->commits
.nr
;
1783 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1784 enum bloom_filter_computed computed
= 0;
1785 struct commit
*c
= sorted_commits
[i
];
1786 struct bloom_filter
*filter
= get_or_compute_bloom_filter(
1789 ctx
->count_bloom_filter_computed
< max_new_filters
,
1790 ctx
->bloom_settings
,
1792 if (computed
& BLOOM_COMPUTED
) {
1793 ctx
->count_bloom_filter_computed
++;
1794 if (computed
& BLOOM_TRUNC_EMPTY
)
1795 ctx
->count_bloom_filter_trunc_empty
++;
1796 if (computed
& BLOOM_TRUNC_LARGE
)
1797 ctx
->count_bloom_filter_trunc_large
++;
1798 } else if (computed
& BLOOM_NOT_COMPUTED
)
1799 ctx
->count_bloom_filter_not_computed
++;
1800 ctx
->total_bloom_filter_data_size
+= filter
1801 ? sizeof(unsigned char) * filter
->len
: 0;
1802 display_progress(progress
, i
+ 1);
1805 if (trace2_is_enabled())
1806 trace2_bloom_filter_write_statistics(ctx
);
1808 free(sorted_commits
);
1809 stop_progress(&progress
);
1812 struct refs_cb_data
{
1813 struct oidset
*commits
;
1814 struct progress
*progress
;
1817 static int add_ref_to_set(const char *refname UNUSED
,
1818 const struct object_id
*oid
,
1819 int flags UNUSED
, void *cb_data
)
1821 struct object_id peeled
;
1822 struct refs_cb_data
*data
= (struct refs_cb_data
*)cb_data
;
1824 if (!peel_iterated_oid(oid
, &peeled
))
1826 if (oid_object_info(the_repository
, oid
, NULL
) == OBJ_COMMIT
)
1827 oidset_insert(data
->commits
, oid
);
1829 display_progress(data
->progress
, oidset_size(data
->commits
));
1834 int write_commit_graph_reachable(struct object_directory
*odb
,
1835 enum commit_graph_write_flags flags
,
1836 const struct commit_graph_opts
*opts
)
1838 struct oidset commits
= OIDSET_INIT
;
1839 struct refs_cb_data data
;
1842 memset(&data
, 0, sizeof(data
));
1843 data
.commits
= &commits
;
1844 if (flags
& COMMIT_GRAPH_WRITE_PROGRESS
)
1845 data
.progress
= start_delayed_progress(
1846 _("Collecting referenced commits"), 0);
1848 refs_for_each_ref(get_main_ref_store(the_repository
), add_ref_to_set
,
1851 stop_progress(&data
.progress
);
1853 result
= write_commit_graph(odb
, NULL
, &commits
,
1856 oidset_clear(&commits
);
1860 static int fill_oids_from_packs(struct write_commit_graph_context
*ctx
,
1861 const struct string_list
*pack_indexes
)
1864 struct strbuf progress_title
= STRBUF_INIT
;
1865 struct strbuf packname
= STRBUF_INIT
;
1869 strbuf_addf(&packname
, "%s/pack/", ctx
->odb
->path
);
1870 dirlen
= packname
.len
;
1871 if (ctx
->report_progress
) {
1872 strbuf_addf(&progress_title
,
1873 Q_("Finding commits for commit graph in %"PRIuMAX
" pack",
1874 "Finding commits for commit graph in %"PRIuMAX
" packs",
1876 (uintmax_t)pack_indexes
->nr
);
1877 ctx
->progress
= start_delayed_progress(progress_title
.buf
, 0);
1878 ctx
->progress_done
= 0;
1880 for (i
= 0; i
< pack_indexes
->nr
; i
++) {
1881 struct packed_git
*p
;
1882 strbuf_setlen(&packname
, dirlen
);
1883 strbuf_addstr(&packname
, pack_indexes
->items
[i
].string
);
1884 p
= add_packed_git(packname
.buf
, packname
.len
, 1);
1886 ret
= error(_("error adding pack %s"), packname
.buf
);
1889 if (open_pack_index(p
)) {
1890 ret
= error(_("error opening index for %s"), packname
.buf
);
1893 for_each_object_in_pack(p
, add_packed_commits
, ctx
,
1894 FOR_EACH_OBJECT_PACK_ORDER
);
1900 stop_progress(&ctx
->progress
);
1901 strbuf_release(&progress_title
);
1902 strbuf_release(&packname
);
1907 static int fill_oids_from_commits(struct write_commit_graph_context
*ctx
,
1908 struct oidset
*commits
)
1910 struct oidset_iter iter
;
1911 struct object_id
*oid
;
1913 if (!oidset_size(commits
))
1916 oidset_iter_init(commits
, &iter
);
1917 while ((oid
= oidset_iter_next(&iter
))) {
1918 oid_array_append(&ctx
->oids
, oid
);
1924 static void fill_oids_from_all_packs(struct write_commit_graph_context
*ctx
)
1926 if (ctx
->report_progress
)
1927 ctx
->progress
= start_delayed_progress(
1928 _("Finding commits for commit graph among packed objects"),
1929 ctx
->approx_nr_objects
);
1930 for_each_packed_object(add_packed_commits
, ctx
,
1931 FOR_EACH_OBJECT_PACK_ORDER
);
1932 if (ctx
->progress_done
< ctx
->approx_nr_objects
)
1933 display_progress(ctx
->progress
, ctx
->approx_nr_objects
);
1934 stop_progress(&ctx
->progress
);
1937 static void copy_oids_to_commits(struct write_commit_graph_context
*ctx
)
1940 enum commit_graph_split_flags flags
= ctx
->opts
?
1941 ctx
->opts
->split_flags
: COMMIT_GRAPH_SPLIT_UNSPECIFIED
;
1943 ctx
->num_extra_edges
= 0;
1944 if (ctx
->report_progress
)
1945 ctx
->progress
= start_delayed_progress(
1946 _("Finding extra edges in commit graph"),
1948 oid_array_sort(&ctx
->oids
);
1949 for (i
= 0; i
< ctx
->oids
.nr
; i
= oid_array_next_unique(&ctx
->oids
, i
)) {
1950 unsigned int num_parents
;
1952 display_progress(ctx
->progress
, i
+ 1);
1954 ALLOC_GROW(ctx
->commits
.list
, ctx
->commits
.nr
+ 1, ctx
->commits
.alloc
);
1955 ctx
->commits
.list
[ctx
->commits
.nr
] = lookup_commit(ctx
->r
, &ctx
->oids
.oid
[i
]);
1957 if (ctx
->split
&& flags
!= COMMIT_GRAPH_SPLIT_REPLACE
&&
1958 commit_graph_position(ctx
->commits
.list
[ctx
->commits
.nr
]) != COMMIT_NOT_FROM_GRAPH
)
1961 if (ctx
->split
&& flags
== COMMIT_GRAPH_SPLIT_REPLACE
)
1962 repo_parse_commit(ctx
->r
, ctx
->commits
.list
[ctx
->commits
.nr
]);
1964 repo_parse_commit_no_graph(ctx
->r
, ctx
->commits
.list
[ctx
->commits
.nr
]);
1966 num_parents
= commit_list_count(ctx
->commits
.list
[ctx
->commits
.nr
]->parents
);
1967 if (num_parents
> 2)
1968 ctx
->num_extra_edges
+= num_parents
- 1;
1972 stop_progress(&ctx
->progress
);
1975 static int write_graph_chunk_base_1(struct hashfile
*f
,
1976 struct commit_graph
*g
)
1983 num
= write_graph_chunk_base_1(f
, g
->base_graph
);
1984 hashwrite(f
, g
->oid
.hash
, the_hash_algo
->rawsz
);
1988 static int write_graph_chunk_base(struct hashfile
*f
,
1991 struct write_commit_graph_context
*ctx
= data
;
1992 int num
= write_graph_chunk_base_1(f
, ctx
->new_base_graph
);
1994 if (num
!= ctx
->num_commit_graphs_after
- 1) {
1995 error(_("failed to write correct number of base graph ids"));
2002 static int write_commit_graph_file(struct write_commit_graph_context
*ctx
)
2007 struct lock_file lk
= LOCK_INIT
;
2008 const unsigned hashsz
= the_hash_algo
->rawsz
;
2009 struct strbuf progress_title
= STRBUF_INIT
;
2010 struct chunkfile
*cf
;
2011 unsigned char file_hash
[GIT_MAX_RAWSZ
];
2014 struct strbuf tmp_file
= STRBUF_INIT
;
2016 strbuf_addf(&tmp_file
,
2017 "%s/info/commit-graphs/tmp_graph_XXXXXX",
2019 ctx
->graph_name
= strbuf_detach(&tmp_file
, NULL
);
2021 ctx
->graph_name
= get_commit_graph_filename(ctx
->odb
);
2024 if (safe_create_leading_directories(ctx
->graph_name
)) {
2025 UNLEAK(ctx
->graph_name
);
2026 error(_("unable to create leading directories of %s"),
2032 char *lock_name
= get_commit_graph_chain_filename(ctx
->odb
);
2034 hold_lock_file_for_update_mode(&lk
, lock_name
,
2035 LOCK_DIE_ON_ERROR
, 0444);
2038 fd
= git_mkstemp_mode(ctx
->graph_name
, 0444);
2040 error(_("unable to create temporary graph layer"));
2044 if (adjust_shared_perm(ctx
->graph_name
)) {
2045 error(_("unable to adjust shared permissions for '%s'"),
2050 f
= hashfd(fd
, ctx
->graph_name
);
2052 hold_lock_file_for_update_mode(&lk
, ctx
->graph_name
,
2053 LOCK_DIE_ON_ERROR
, 0444);
2054 fd
= get_lock_file_fd(&lk
);
2055 f
= hashfd(fd
, get_lock_file_path(&lk
));
2058 cf
= init_chunkfile(f
);
2060 add_chunk(cf
, GRAPH_CHUNKID_OIDFANOUT
, GRAPH_FANOUT_SIZE
,
2061 write_graph_chunk_fanout
);
2062 add_chunk(cf
, GRAPH_CHUNKID_OIDLOOKUP
, st_mult(hashsz
, ctx
->commits
.nr
),
2063 write_graph_chunk_oids
);
2064 add_chunk(cf
, GRAPH_CHUNKID_DATA
, st_mult(hashsz
+ 16, ctx
->commits
.nr
),
2065 write_graph_chunk_data
);
2067 if (ctx
->write_generation_data
)
2068 add_chunk(cf
, GRAPH_CHUNKID_GENERATION_DATA
,
2069 st_mult(sizeof(uint32_t), ctx
->commits
.nr
),
2070 write_graph_chunk_generation_data
);
2071 if (ctx
->num_generation_data_overflows
)
2072 add_chunk(cf
, GRAPH_CHUNKID_GENERATION_DATA_OVERFLOW
,
2073 st_mult(sizeof(timestamp_t
), ctx
->num_generation_data_overflows
),
2074 write_graph_chunk_generation_data_overflow
);
2075 if (ctx
->num_extra_edges
)
2076 add_chunk(cf
, GRAPH_CHUNKID_EXTRAEDGES
,
2077 st_mult(4, ctx
->num_extra_edges
),
2078 write_graph_chunk_extra_edges
);
2079 if (ctx
->changed_paths
) {
2080 add_chunk(cf
, GRAPH_CHUNKID_BLOOMINDEXES
,
2081 st_mult(sizeof(uint32_t), ctx
->commits
.nr
),
2082 write_graph_chunk_bloom_indexes
);
2083 add_chunk(cf
, GRAPH_CHUNKID_BLOOMDATA
,
2084 st_add(sizeof(uint32_t) * 3,
2085 ctx
->total_bloom_filter_data_size
),
2086 write_graph_chunk_bloom_data
);
2088 if (ctx
->num_commit_graphs_after
> 1)
2089 add_chunk(cf
, GRAPH_CHUNKID_BASE
,
2090 st_mult(hashsz
, ctx
->num_commit_graphs_after
- 1),
2091 write_graph_chunk_base
);
2093 hashwrite_be32(f
, GRAPH_SIGNATURE
);
2095 hashwrite_u8(f
, GRAPH_VERSION
);
2096 hashwrite_u8(f
, oid_version(the_hash_algo
));
2097 hashwrite_u8(f
, get_num_chunks(cf
));
2098 hashwrite_u8(f
, ctx
->num_commit_graphs_after
- 1);
2100 if (ctx
->report_progress
) {
2101 strbuf_addf(&progress_title
,
2102 Q_("Writing out commit graph in %d pass",
2103 "Writing out commit graph in %d passes",
2104 get_num_chunks(cf
)),
2105 get_num_chunks(cf
));
2106 ctx
->progress
= start_delayed_progress(
2108 st_mult(get_num_chunks(cf
), ctx
->commits
.nr
));
2111 write_chunkfile(cf
, ctx
);
2113 stop_progress(&ctx
->progress
);
2114 strbuf_release(&progress_title
);
2116 if (ctx
->split
&& ctx
->base_graph_name
&& ctx
->num_commit_graphs_after
> 1) {
2117 char *new_base_hash
= xstrdup(oid_to_hex(&ctx
->new_base_graph
->oid
));
2118 char *new_base_name
= get_split_graph_filename(ctx
->new_base_graph
->odb
, new_base_hash
);
2120 free(ctx
->commit_graph_filenames_after
[ctx
->num_commit_graphs_after
- 2]);
2121 free(ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 2]);
2122 ctx
->commit_graph_filenames_after
[ctx
->num_commit_graphs_after
- 2] = new_base_name
;
2123 ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 2] = new_base_hash
;
2126 close_commit_graph(ctx
->r
->objects
);
2127 finalize_hashfile(f
, file_hash
, FSYNC_COMPONENT_COMMIT_GRAPH
,
2128 CSUM_HASH_IN_STREAM
| CSUM_FSYNC
);
2132 FILE *chainf
= fdopen_lock_file(&lk
, "w");
2133 char *final_graph_name
;
2139 error(_("unable to open commit-graph chain file"));
2143 if (ctx
->base_graph_name
) {
2145 int idx
= ctx
->num_commit_graphs_after
- 1;
2146 if (ctx
->num_commit_graphs_after
> 1)
2149 dest
= ctx
->commit_graph_filenames_after
[idx
];
2151 if (strcmp(ctx
->base_graph_name
, dest
)) {
2152 result
= rename(ctx
->base_graph_name
, dest
);
2155 error(_("failed to rename base commit-graph file"));
2160 char *graph_name
= get_commit_graph_filename(ctx
->odb
);
2165 free(ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 1]);
2166 ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 1] = xstrdup(hash_to_hex(file_hash
));
2167 final_graph_name
= get_split_graph_filename(ctx
->odb
,
2168 ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 1]);
2169 free(ctx
->commit_graph_filenames_after
[ctx
->num_commit_graphs_after
- 1]);
2170 ctx
->commit_graph_filenames_after
[ctx
->num_commit_graphs_after
- 1] = final_graph_name
;
2172 result
= rename(ctx
->graph_name
, final_graph_name
);
2174 for (i
= 0; i
< ctx
->num_commit_graphs_after
; i
++)
2175 fprintf(get_lock_file_fp(&lk
), "%s\n", ctx
->commit_graph_hash_after
[i
]);
2178 error(_("failed to rename temporary commit-graph file"));
2183 commit_lock_file(&lk
);
2188 static void split_graph_merge_strategy(struct write_commit_graph_context
*ctx
)
2190 struct commit_graph
*g
;
2191 uint32_t num_commits
;
2192 enum commit_graph_split_flags flags
= COMMIT_GRAPH_SPLIT_UNSPECIFIED
;
2195 int max_commits
= 0;
2199 max_commits
= ctx
->opts
->max_commits
;
2201 if (ctx
->opts
->size_multiple
)
2202 size_mult
= ctx
->opts
->size_multiple
;
2204 flags
= ctx
->opts
->split_flags
;
2207 g
= ctx
->r
->objects
->commit_graph
;
2208 num_commits
= ctx
->commits
.nr
;
2209 if (flags
== COMMIT_GRAPH_SPLIT_REPLACE
)
2210 ctx
->num_commit_graphs_after
= 1;
2212 ctx
->num_commit_graphs_after
= ctx
->num_commit_graphs_before
+ 1;
2214 if (flags
!= COMMIT_GRAPH_SPLIT_MERGE_PROHIBITED
&&
2215 flags
!= COMMIT_GRAPH_SPLIT_REPLACE
) {
2216 while (g
&& (g
->num_commits
<= st_mult(size_mult
, num_commits
) ||
2217 (max_commits
&& num_commits
> max_commits
))) {
2218 if (g
->odb
!= ctx
->odb
)
2221 if (unsigned_add_overflows(num_commits
, g
->num_commits
))
2222 die(_("cannot merge graphs with %"PRIuMAX
", "
2223 "%"PRIuMAX
" commits"),
2224 (uintmax_t)num_commits
,
2225 (uintmax_t)g
->num_commits
);
2226 num_commits
+= g
->num_commits
;
2229 ctx
->num_commit_graphs_after
--;
2233 if (flags
!= COMMIT_GRAPH_SPLIT_REPLACE
)
2234 ctx
->new_base_graph
= g
;
2235 else if (ctx
->num_commit_graphs_after
!= 1)
2236 BUG("split_graph_merge_strategy: num_commit_graphs_after "
2237 "should be 1 with --split=replace");
2239 if (ctx
->num_commit_graphs_after
== 2) {
2240 char *old_graph_name
= get_commit_graph_filename(g
->odb
);
2242 if (!strcmp(g
->filename
, old_graph_name
) &&
2243 g
->odb
!= ctx
->odb
) {
2244 ctx
->num_commit_graphs_after
= 1;
2245 ctx
->new_base_graph
= NULL
;
2248 free(old_graph_name
);
2251 CALLOC_ARRAY(ctx
->commit_graph_filenames_after
, ctx
->num_commit_graphs_after
);
2252 CALLOC_ARRAY(ctx
->commit_graph_hash_after
, ctx
->num_commit_graphs_after
);
2254 for (i
= 0; i
< ctx
->num_commit_graphs_after
&&
2255 i
< ctx
->num_commit_graphs_before
; i
++)
2256 ctx
->commit_graph_filenames_after
[i
] = xstrdup(ctx
->commit_graph_filenames_before
[i
]);
2258 i
= ctx
->num_commit_graphs_before
- 1;
2259 g
= ctx
->r
->objects
->commit_graph
;
2262 if (i
< ctx
->num_commit_graphs_after
)
2263 ctx
->commit_graph_hash_after
[i
] = xstrdup(oid_to_hex(&g
->oid
));
2266 * If the topmost remaining layer has generation data chunk, the
2267 * resultant layer also has generation data chunk.
2269 if (i
== ctx
->num_commit_graphs_after
- 2)
2270 ctx
->write_generation_data
= !!g
->chunk_generation_data
;
2277 static void merge_commit_graph(struct write_commit_graph_context
*ctx
,
2278 struct commit_graph
*g
)
2281 uint32_t offset
= g
->num_commits_in_base
;
2283 if (unsigned_add_overflows(ctx
->commits
.nr
, g
->num_commits
))
2284 die(_("cannot merge graph %s, too many commits: %"PRIuMAX
),
2285 oid_to_hex(&g
->oid
),
2286 (uintmax_t)st_add(ctx
->commits
.nr
, g
->num_commits
));
2288 ALLOC_GROW(ctx
->commits
.list
, ctx
->commits
.nr
+ g
->num_commits
, ctx
->commits
.alloc
);
2290 for (i
= 0; i
< g
->num_commits
; i
++) {
2291 struct object_id oid
;
2292 struct commit
*result
;
2294 display_progress(ctx
->progress
, i
+ 1);
2296 load_oid_from_graph(g
, i
+ offset
, &oid
);
2298 /* only add commits if they still exist in the repo */
2299 result
= lookup_commit_reference_gently(ctx
->r
, &oid
, 1);
2302 ctx
->commits
.list
[ctx
->commits
.nr
] = result
;
2308 static int commit_compare(const void *_a
, const void *_b
)
2310 const struct commit
*a
= *(const struct commit
**)_a
;
2311 const struct commit
*b
= *(const struct commit
**)_b
;
2312 return oidcmp(&a
->object
.oid
, &b
->object
.oid
);
2315 static void sort_and_scan_merged_commits(struct write_commit_graph_context
*ctx
)
2317 uint32_t i
, dedup_i
= 0;
2319 if (ctx
->report_progress
)
2320 ctx
->progress
= start_delayed_progress(
2321 _("Scanning merged commits"),
2324 QSORT(ctx
->commits
.list
, ctx
->commits
.nr
, commit_compare
);
2326 ctx
->num_extra_edges
= 0;
2327 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
2328 display_progress(ctx
->progress
, i
+ 1);
2330 if (i
&& oideq(&ctx
->commits
.list
[i
- 1]->object
.oid
,
2331 &ctx
->commits
.list
[i
]->object
.oid
)) {
2333 * Silently ignore duplicates. These were likely
2334 * created due to a commit appearing in multiple
2335 * layers of the chain, which is unexpected but
2336 * not invalid. We should make sure there is a
2337 * unique copy in the new layer.
2340 unsigned int num_parents
;
2342 ctx
->commits
.list
[dedup_i
] = ctx
->commits
.list
[i
];
2345 num_parents
= commit_list_count(ctx
->commits
.list
[i
]->parents
);
2346 if (num_parents
> 2)
2347 ctx
->num_extra_edges
+= num_parents
- 1;
2351 ctx
->commits
.nr
= dedup_i
;
2353 stop_progress(&ctx
->progress
);
2356 static void merge_commit_graphs(struct write_commit_graph_context
*ctx
)
2358 struct commit_graph
*g
= ctx
->r
->objects
->commit_graph
;
2359 uint32_t current_graph_number
= ctx
->num_commit_graphs_before
;
2361 while (g
&& current_graph_number
>= ctx
->num_commit_graphs_after
) {
2362 current_graph_number
--;
2364 if (ctx
->report_progress
)
2365 ctx
->progress
= start_delayed_progress(_("Merging commit-graph"), 0);
2367 merge_commit_graph(ctx
, g
);
2368 stop_progress(&ctx
->progress
);
2374 ctx
->new_base_graph
= g
;
2375 ctx
->new_num_commits_in_base
= g
->num_commits
+ g
->num_commits_in_base
;
2378 if (ctx
->new_base_graph
)
2379 ctx
->base_graph_name
= xstrdup(ctx
->new_base_graph
->filename
);
2381 sort_and_scan_merged_commits(ctx
);
2384 static void mark_commit_graphs(struct write_commit_graph_context
*ctx
)
2387 time_t now
= time(NULL
);
2389 for (i
= ctx
->num_commit_graphs_after
- 1; i
< ctx
->num_commit_graphs_before
; i
++) {
2391 struct utimbuf updated_time
;
2393 if (stat(ctx
->commit_graph_filenames_before
[i
], &st
) < 0)
2396 updated_time
.actime
= st
.st_atime
;
2397 updated_time
.modtime
= now
;
2398 utime(ctx
->commit_graph_filenames_before
[i
], &updated_time
);
2402 static void expire_commit_graphs(struct write_commit_graph_context
*ctx
)
2404 struct strbuf path
= STRBUF_INIT
;
2408 timestamp_t expire_time
= time(NULL
);
2410 if (ctx
->opts
&& ctx
->opts
->expire_time
)
2411 expire_time
= ctx
->opts
->expire_time
;
2413 char *chain_file_name
= get_commit_graph_chain_filename(ctx
->odb
);
2414 unlink(chain_file_name
);
2415 free(chain_file_name
);
2416 ctx
->num_commit_graphs_after
= 0;
2419 strbuf_addstr(&path
, ctx
->odb
->path
);
2420 strbuf_addstr(&path
, "/info/commit-graphs");
2421 dir
= opendir(path
.buf
);
2426 strbuf_addch(&path
, '/');
2427 dirnamelen
= path
.len
;
2428 while ((de
= readdir(dir
)) != NULL
) {
2430 uint32_t i
, found
= 0;
2432 strbuf_setlen(&path
, dirnamelen
);
2433 strbuf_addstr(&path
, de
->d_name
);
2435 if (stat(path
.buf
, &st
) < 0)
2438 if (st
.st_mtime
> expire_time
)
2440 if (path
.len
< 6 || strcmp(path
.buf
+ path
.len
- 6, ".graph"))
2443 for (i
= 0; i
< ctx
->num_commit_graphs_after
; i
++) {
2444 if (!strcmp(ctx
->commit_graph_filenames_after
[i
],
2458 strbuf_release(&path
);
2461 int write_commit_graph(struct object_directory
*odb
,
2462 const struct string_list
*const pack_indexes
,
2463 struct oidset
*commits
,
2464 enum commit_graph_write_flags flags
,
2465 const struct commit_graph_opts
*opts
)
2467 struct repository
*r
= the_repository
;
2468 struct write_commit_graph_context
*ctx
;
2472 struct bloom_filter_settings bloom_settings
= DEFAULT_BLOOM_FILTER_SETTINGS
;
2473 struct topo_level_slab topo_levels
;
2475 prepare_repo_settings(r
);
2476 if (!r
->settings
.core_commit_graph
) {
2477 warning(_("attempting to write a commit-graph, but 'core.commitGraph' is disabled"));
2480 if (!commit_graph_compatible(r
))
2483 CALLOC_ARRAY(ctx
, 1);
2486 ctx
->append
= flags
& COMMIT_GRAPH_WRITE_APPEND
? 1 : 0;
2487 ctx
->report_progress
= flags
& COMMIT_GRAPH_WRITE_PROGRESS
? 1 : 0;
2488 ctx
->split
= flags
& COMMIT_GRAPH_WRITE_SPLIT
? 1 : 0;
2490 ctx
->total_bloom_filter_data_size
= 0;
2491 ctx
->write_generation_data
= (get_configured_generation_version(r
) == 2);
2492 ctx
->num_generation_data_overflows
= 0;
2494 bloom_settings
.bits_per_entry
= git_env_ulong("GIT_TEST_BLOOM_SETTINGS_BITS_PER_ENTRY",
2495 bloom_settings
.bits_per_entry
);
2496 bloom_settings
.num_hashes
= git_env_ulong("GIT_TEST_BLOOM_SETTINGS_NUM_HASHES",
2497 bloom_settings
.num_hashes
);
2498 bloom_settings
.max_changed_paths
= git_env_ulong("GIT_TEST_BLOOM_SETTINGS_MAX_CHANGED_PATHS",
2499 bloom_settings
.max_changed_paths
);
2500 ctx
->bloom_settings
= &bloom_settings
;
2502 init_topo_level_slab(&topo_levels
);
2503 ctx
->topo_levels
= &topo_levels
;
2505 prepare_commit_graph(ctx
->r
);
2506 if (ctx
->r
->objects
->commit_graph
) {
2507 struct commit_graph
*g
= ctx
->r
->objects
->commit_graph
;
2510 g
->topo_levels
= &topo_levels
;
2515 if (flags
& COMMIT_GRAPH_WRITE_BLOOM_FILTERS
)
2516 ctx
->changed_paths
= 1;
2517 if (!(flags
& COMMIT_GRAPH_NO_WRITE_BLOOM_FILTERS
)) {
2518 struct commit_graph
*g
;
2520 g
= ctx
->r
->objects
->commit_graph
;
2522 /* We have changed-paths already. Keep them in the next graph */
2523 if (g
&& g
->chunk_bloom_data
) {
2524 ctx
->changed_paths
= 1;
2525 ctx
->bloom_settings
= g
->bloom_filter_settings
;
2530 struct commit_graph
*g
= ctx
->r
->objects
->commit_graph
;
2533 ctx
->num_commit_graphs_before
++;
2537 if (ctx
->num_commit_graphs_before
) {
2538 ALLOC_ARRAY(ctx
->commit_graph_filenames_before
, ctx
->num_commit_graphs_before
);
2539 i
= ctx
->num_commit_graphs_before
;
2540 g
= ctx
->r
->objects
->commit_graph
;
2543 ctx
->commit_graph_filenames_before
[--i
] = xstrdup(g
->filename
);
2549 replace
= ctx
->opts
->split_flags
& COMMIT_GRAPH_SPLIT_REPLACE
;
2552 ctx
->approx_nr_objects
= repo_approximate_object_count(the_repository
);
2554 if (ctx
->append
&& ctx
->r
->objects
->commit_graph
) {
2555 struct commit_graph
*g
= ctx
->r
->objects
->commit_graph
;
2556 for (i
= 0; i
< g
->num_commits
; i
++) {
2557 struct object_id oid
;
2558 oidread(&oid
, g
->chunk_oid_lookup
+ st_mult(g
->hash_len
, i
));
2559 oid_array_append(&ctx
->oids
, &oid
);
2564 ctx
->order_by_pack
= 1;
2565 if ((res
= fill_oids_from_packs(ctx
, pack_indexes
)))
2570 if ((res
= fill_oids_from_commits(ctx
, commits
)))
2574 if (!pack_indexes
&& !commits
) {
2575 ctx
->order_by_pack
= 1;
2576 fill_oids_from_all_packs(ctx
);
2579 close_reachable(ctx
);
2581 copy_oids_to_commits(ctx
);
2583 if (ctx
->commits
.nr
>= GRAPH_EDGE_LAST_MASK
) {
2584 error(_("too many commits to write graph"));
2589 if (!ctx
->commits
.nr
&& !replace
)
2593 split_graph_merge_strategy(ctx
);
2596 merge_commit_graphs(ctx
);
2598 ctx
->num_commit_graphs_after
= 1;
2600 ctx
->trust_generation_numbers
= validate_mixed_generation_chain(ctx
->r
->objects
->commit_graph
);
2602 compute_topological_levels(ctx
);
2603 if (ctx
->write_generation_data
)
2604 compute_generation_numbers(ctx
);
2606 if (ctx
->changed_paths
)
2607 compute_bloom_filters(ctx
);
2609 res
= write_commit_graph_file(ctx
);
2612 mark_commit_graphs(ctx
);
2614 expire_commit_graphs(ctx
);
2617 free(ctx
->graph_name
);
2618 free(ctx
->base_graph_name
);
2619 free(ctx
->commits
.list
);
2620 oid_array_clear(&ctx
->oids
);
2621 clear_topo_level_slab(&topo_levels
);
2623 for (i
= 0; i
< ctx
->num_commit_graphs_before
; i
++)
2624 free(ctx
->commit_graph_filenames_before
[i
]);
2625 free(ctx
->commit_graph_filenames_before
);
2627 for (i
= 0; i
< ctx
->num_commit_graphs_after
; i
++) {
2628 free(ctx
->commit_graph_filenames_after
[i
]);
2629 free(ctx
->commit_graph_hash_after
[i
]);
2631 free(ctx
->commit_graph_filenames_after
);
2632 free(ctx
->commit_graph_hash_after
);
2639 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
2640 static int verify_commit_graph_error
;
2642 __attribute__((format (printf
, 1, 2)))
2643 static void graph_report(const char *fmt
, ...)
2647 verify_commit_graph_error
= 1;
2649 vfprintf(stderr
, fmt
, ap
);
2650 fprintf(stderr
, "\n");
2654 static int commit_graph_checksum_valid(struct commit_graph
*g
)
2656 return hashfile_checksum_valid(g
->data
, g
->data_len
);
2659 static int verify_one_commit_graph(struct repository
*r
,
2660 struct commit_graph
*g
,
2661 struct progress
*progress
,
2664 uint32_t i
, cur_fanout_pos
= 0;
2665 struct object_id prev_oid
, cur_oid
;
2666 struct commit
*seen_gen_zero
= NULL
;
2667 struct commit
*seen_gen_non_zero
= NULL
;
2669 if (!commit_graph_checksum_valid(g
)) {
2670 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
2671 verify_commit_graph_error
= VERIFY_COMMIT_GRAPH_ERROR_HASH
;
2674 for (i
= 0; i
< g
->num_commits
; i
++) {
2675 struct commit
*graph_commit
;
2677 oidread(&cur_oid
, g
->chunk_oid_lookup
+ st_mult(g
->hash_len
, i
));
2679 if (i
&& oidcmp(&prev_oid
, &cur_oid
) >= 0)
2680 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
2681 oid_to_hex(&prev_oid
),
2682 oid_to_hex(&cur_oid
));
2684 oidcpy(&prev_oid
, &cur_oid
);
2686 while (cur_oid
.hash
[0] > cur_fanout_pos
) {
2687 uint32_t fanout_value
= get_be32(g
->chunk_oid_fanout
+ cur_fanout_pos
);
2689 if (i
!= fanout_value
)
2690 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2691 cur_fanout_pos
, fanout_value
, i
);
2695 graph_commit
= lookup_commit(r
, &cur_oid
);
2696 if (!parse_commit_in_graph_one(r
, g
, graph_commit
))
2697 graph_report(_("failed to parse commit %s from commit-graph"),
2698 oid_to_hex(&cur_oid
));
2701 while (cur_fanout_pos
< 256) {
2702 uint32_t fanout_value
= get_be32(g
->chunk_oid_fanout
+ cur_fanout_pos
);
2704 if (g
->num_commits
!= fanout_value
)
2705 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2706 cur_fanout_pos
, fanout_value
, i
);
2711 if (verify_commit_graph_error
& ~VERIFY_COMMIT_GRAPH_ERROR_HASH
)
2712 return verify_commit_graph_error
;
2714 for (i
= 0; i
< g
->num_commits
; i
++) {
2715 struct commit
*graph_commit
, *odb_commit
;
2716 struct commit_list
*graph_parents
, *odb_parents
;
2717 timestamp_t max_generation
= 0;
2718 timestamp_t generation
;
2720 display_progress(progress
, ++(*seen
));
2721 oidread(&cur_oid
, g
->chunk_oid_lookup
+ st_mult(g
->hash_len
, i
));
2723 graph_commit
= lookup_commit(r
, &cur_oid
);
2724 odb_commit
= (struct commit
*)create_object(r
, &cur_oid
, alloc_commit_node(r
));
2725 if (repo_parse_commit_internal(r
, odb_commit
, 0, 0)) {
2726 graph_report(_("failed to parse commit %s from object database for commit-graph"),
2727 oid_to_hex(&cur_oid
));
2731 if (!oideq(&get_commit_tree_in_graph_one(r
, g
, graph_commit
)->object
.oid
,
2732 get_commit_tree_oid(odb_commit
)))
2733 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2734 oid_to_hex(&cur_oid
),
2735 oid_to_hex(get_commit_tree_oid(graph_commit
)),
2736 oid_to_hex(get_commit_tree_oid(odb_commit
)));
2738 graph_parents
= graph_commit
->parents
;
2739 odb_parents
= odb_commit
->parents
;
2741 while (graph_parents
) {
2743 graph_report(_("commit-graph parent list for commit %s is too long"),
2744 oid_to_hex(&cur_oid
));
2748 /* parse parent in case it is in a base graph */
2749 parse_commit_in_graph_one(r
, g
, graph_parents
->item
);
2751 if (!oideq(&graph_parents
->item
->object
.oid
, &odb_parents
->item
->object
.oid
))
2752 graph_report(_("commit-graph parent for %s is %s != %s"),
2753 oid_to_hex(&cur_oid
),
2754 oid_to_hex(&graph_parents
->item
->object
.oid
),
2755 oid_to_hex(&odb_parents
->item
->object
.oid
));
2757 generation
= commit_graph_generation_from_graph(graph_parents
->item
);
2758 if (generation
> max_generation
)
2759 max_generation
= generation
;
2761 graph_parents
= graph_parents
->next
;
2762 odb_parents
= odb_parents
->next
;
2766 graph_report(_("commit-graph parent list for commit %s terminates early"),
2767 oid_to_hex(&cur_oid
));
2769 if (commit_graph_generation_from_graph(graph_commit
))
2770 seen_gen_non_zero
= graph_commit
;
2772 seen_gen_zero
= graph_commit
;
2778 * If we are using topological level and one of our parents has
2779 * generation GENERATION_NUMBER_V1_MAX, then our generation is
2780 * also GENERATION_NUMBER_V1_MAX. Decrement to avoid extra logic
2781 * in the following condition.
2783 if (!g
->read_generation_data
&& max_generation
== GENERATION_NUMBER_V1_MAX
)
2786 generation
= commit_graph_generation(graph_commit
);
2787 if (generation
< max_generation
+ 1)
2788 graph_report(_("commit-graph generation for commit %s is %"PRItime
" < %"PRItime
),
2789 oid_to_hex(&cur_oid
),
2791 max_generation
+ 1);
2793 if (graph_commit
->date
!= odb_commit
->date
)
2794 graph_report(_("commit date for commit %s in commit-graph is %"PRItime
" != %"PRItime
),
2795 oid_to_hex(&cur_oid
),
2800 if (seen_gen_zero
&& seen_gen_non_zero
)
2801 graph_report(_("commit-graph has both zero and non-zero "
2802 "generations (e.g., commits '%s' and '%s')"),
2803 oid_to_hex(&seen_gen_zero
->object
.oid
),
2804 oid_to_hex(&seen_gen_non_zero
->object
.oid
));
2806 return verify_commit_graph_error
;
2809 int verify_commit_graph(struct repository
*r
, struct commit_graph
*g
, int flags
)
2811 struct progress
*progress
= NULL
;
2812 int local_error
= 0;
2816 graph_report("no commit-graph file loaded");
2820 if (flags
& COMMIT_GRAPH_WRITE_PROGRESS
) {
2821 uint64_t total
= g
->num_commits
;
2822 if (!(flags
& COMMIT_GRAPH_VERIFY_SHALLOW
))
2823 total
+= g
->num_commits_in_base
;
2825 progress
= start_progress(_("Verifying commits in commit graph"),
2829 for (; g
; g
= g
->base_graph
) {
2830 local_error
|= verify_one_commit_graph(r
, g
, progress
, &seen
);
2831 if (flags
& COMMIT_GRAPH_VERIFY_SHALLOW
)
2835 stop_progress(&progress
);
2840 void free_commit_graph(struct commit_graph
*g
)
2843 struct commit_graph
*next
= g
->base_graph
;
2846 munmap((void *)g
->data
, g
->data_len
);
2848 free(g
->bloom_filter_settings
);
2855 void disable_commit_graph(struct repository
*r
)
2857 r
->commit_graph_disabled
= 1;