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 verify_commit_graph_lite(struct commit_graph
*g
)
282 * Basic validation shared between parse_commit_graph()
283 * which'll be called every time the graph is used, and the
284 * much more expensive verify_commit_graph() used by
285 * "commit-graph verify".
287 * There should only be very basic checks here to ensure that
288 * we don't e.g. segfault in fill_commit_in_graph(), but
289 * because this is a very hot codepath nothing that e.g. loops
290 * over g->num_commits, or runs a checksum on the commit-graph
293 if (!g
->chunk_oid_fanout
) {
294 error("commit-graph is missing the OID Fanout chunk");
297 if (!g
->chunk_oid_lookup
) {
298 error("commit-graph is missing the OID Lookup chunk");
301 if (!g
->chunk_commit_data
) {
302 error("commit-graph is missing the Commit Data chunk");
306 for (i
= 0; i
< 255; i
++) {
307 uint32_t oid_fanout1
= ntohl(g
->chunk_oid_fanout
[i
]);
308 uint32_t oid_fanout2
= ntohl(g
->chunk_oid_fanout
[i
+ 1]);
310 if (oid_fanout1
> oid_fanout2
) {
311 error("commit-graph fanout values out of order");
315 if (ntohl(g
->chunk_oid_fanout
[255]) != g
->num_commits
) {
316 error("commit-graph oid table and fanout disagree on size");
323 static int graph_read_oid_fanout(const unsigned char *chunk_start
,
324 size_t chunk_size
, void *data
)
326 struct commit_graph
*g
= data
;
327 if (chunk_size
!= 256 * sizeof(uint32_t))
328 return error("commit-graph oid fanout chunk is wrong size");
329 g
->chunk_oid_fanout
= (const uint32_t *)chunk_start
;
333 static int graph_read_oid_lookup(const unsigned char *chunk_start
,
334 size_t chunk_size
, void *data
)
336 struct commit_graph
*g
= data
;
337 g
->chunk_oid_lookup
= chunk_start
;
338 g
->num_commits
= chunk_size
/ g
->hash_len
;
342 static int graph_read_commit_data(const unsigned char *chunk_start
,
343 size_t chunk_size
, void *data
)
345 struct commit_graph
*g
= data
;
346 if (chunk_size
!= g
->num_commits
* GRAPH_DATA_WIDTH
)
347 return error("commit-graph commit data chunk is wrong size");
348 g
->chunk_commit_data
= chunk_start
;
352 static int graph_read_generation_data(const unsigned char *chunk_start
,
353 size_t chunk_size
, void *data
)
355 struct commit_graph
*g
= data
;
356 if (chunk_size
!= g
->num_commits
* sizeof(uint32_t))
357 return error("commit-graph generations chunk is wrong size");
358 g
->chunk_generation_data
= chunk_start
;
362 static int graph_read_bloom_index(const unsigned char *chunk_start
,
363 size_t chunk_size
, void *data
)
365 struct commit_graph
*g
= data
;
366 if (chunk_size
!= g
->num_commits
* 4) {
367 warning("commit-graph changed-path index chunk is too small");
370 g
->chunk_bloom_indexes
= chunk_start
;
374 static int graph_read_bloom_data(const unsigned char *chunk_start
,
375 size_t chunk_size
, void *data
)
377 struct commit_graph
*g
= data
;
378 uint32_t hash_version
;
380 if (chunk_size
< BLOOMDATA_CHUNK_HEADER_SIZE
) {
381 warning("ignoring too-small changed-path chunk"
382 " (%"PRIuMAX
" < %"PRIuMAX
") in commit-graph file",
383 (uintmax_t)chunk_size
,
384 (uintmax_t)BLOOMDATA_CHUNK_HEADER_SIZE
);
388 g
->chunk_bloom_data
= chunk_start
;
389 g
->chunk_bloom_data_size
= chunk_size
;
390 hash_version
= get_be32(chunk_start
);
392 if (hash_version
!= 1)
395 g
->bloom_filter_settings
= xmalloc(sizeof(struct bloom_filter_settings
));
396 g
->bloom_filter_settings
->hash_version
= hash_version
;
397 g
->bloom_filter_settings
->num_hashes
= get_be32(chunk_start
+ 4);
398 g
->bloom_filter_settings
->bits_per_entry
= get_be32(chunk_start
+ 8);
399 g
->bloom_filter_settings
->max_changed_paths
= DEFAULT_BLOOM_MAX_CHANGES
;
404 struct commit_graph
*parse_commit_graph(struct repo_settings
*s
,
405 void *graph_map
, size_t graph_size
)
407 const unsigned char *data
;
408 struct commit_graph
*graph
;
409 uint32_t graph_signature
;
410 unsigned char graph_version
, hash_version
;
411 struct chunkfile
*cf
= NULL
;
416 if (graph_size
< GRAPH_MIN_SIZE
)
419 data
= (const unsigned char *)graph_map
;
421 graph_signature
= get_be32(data
);
422 if (graph_signature
!= GRAPH_SIGNATURE
) {
423 error(_("commit-graph signature %X does not match signature %X"),
424 graph_signature
, GRAPH_SIGNATURE
);
428 graph_version
= *(unsigned char*)(data
+ 4);
429 if (graph_version
!= GRAPH_VERSION
) {
430 error(_("commit-graph version %X does not match version %X"),
431 graph_version
, GRAPH_VERSION
);
435 hash_version
= *(unsigned char*)(data
+ 5);
436 if (hash_version
!= oid_version(the_hash_algo
)) {
437 error(_("commit-graph hash version %X does not match version %X"),
438 hash_version
, oid_version(the_hash_algo
));
442 graph
= alloc_commit_graph();
444 graph
->hash_len
= the_hash_algo
->rawsz
;
445 graph
->num_chunks
= *(unsigned char*)(data
+ 6);
446 graph
->data
= graph_map
;
447 graph
->data_len
= graph_size
;
449 if (graph_size
< GRAPH_HEADER_SIZE
+
450 (graph
->num_chunks
+ 1) * CHUNK_TOC_ENTRY_SIZE
+
451 GRAPH_FANOUT_SIZE
+ the_hash_algo
->rawsz
) {
452 error(_("commit-graph file is too small to hold %u chunks"),
458 cf
= init_chunkfile(NULL
);
460 if (read_table_of_contents(cf
, graph
->data
, graph_size
,
461 GRAPH_HEADER_SIZE
, graph
->num_chunks
, 1))
462 goto free_and_return
;
464 read_chunk(cf
, GRAPH_CHUNKID_OIDFANOUT
, graph_read_oid_fanout
, graph
);
465 read_chunk(cf
, GRAPH_CHUNKID_OIDLOOKUP
, graph_read_oid_lookup
, graph
);
466 read_chunk(cf
, GRAPH_CHUNKID_DATA
, graph_read_commit_data
, graph
);
467 pair_chunk(cf
, GRAPH_CHUNKID_EXTRAEDGES
, &graph
->chunk_extra_edges
,
468 &graph
->chunk_extra_edges_size
);
469 pair_chunk(cf
, GRAPH_CHUNKID_BASE
, &graph
->chunk_base_graphs
,
470 &graph
->chunk_base_graphs_size
);
472 if (s
->commit_graph_generation_version
>= 2) {
473 read_chunk(cf
, GRAPH_CHUNKID_GENERATION_DATA
,
474 graph_read_generation_data
, graph
);
475 pair_chunk(cf
, GRAPH_CHUNKID_GENERATION_DATA_OVERFLOW
,
476 &graph
->chunk_generation_data_overflow
,
477 &graph
->chunk_generation_data_overflow_size
);
479 if (graph
->chunk_generation_data
)
480 graph
->read_generation_data
= 1;
483 if (s
->commit_graph_read_changed_paths
) {
484 read_chunk(cf
, GRAPH_CHUNKID_BLOOMINDEXES
,
485 graph_read_bloom_index
, graph
);
486 read_chunk(cf
, GRAPH_CHUNKID_BLOOMDATA
,
487 graph_read_bloom_data
, graph
);
490 if (graph
->chunk_bloom_indexes
&& graph
->chunk_bloom_data
) {
491 init_bloom_filters();
493 /* We need both the bloom chunks to exist together. Else ignore the data */
494 graph
->chunk_bloom_indexes
= NULL
;
495 graph
->chunk_bloom_data
= NULL
;
496 FREE_AND_NULL(graph
->bloom_filter_settings
);
499 oidread(&graph
->oid
, graph
->data
+ graph
->data_len
- graph
->hash_len
);
501 if (verify_commit_graph_lite(graph
))
502 goto free_and_return
;
509 free(graph
->bloom_filter_settings
);
514 static struct commit_graph
*load_commit_graph_one(struct repository
*r
,
515 const char *graph_file
,
516 struct object_directory
*odb
)
521 struct commit_graph
*g
;
522 int open_ok
= open_commit_graph(graph_file
, &fd
, &st
);
527 g
= load_commit_graph_one_fd_st(r
, fd
, &st
, odb
);
530 g
->filename
= xstrdup(graph_file
);
535 static struct commit_graph
*load_commit_graph_v1(struct repository
*r
,
536 struct object_directory
*odb
)
538 char *graph_name
= get_commit_graph_filename(odb
);
539 struct commit_graph
*g
= load_commit_graph_one(r
, graph_name
, odb
);
546 * returns 1 if and only if all graphs in the chain have
547 * corrected commit dates stored in the generation_data chunk.
549 static int validate_mixed_generation_chain(struct commit_graph
*g
)
551 int read_generation_data
= 1;
552 struct commit_graph
*p
= g
;
554 while (read_generation_data
&& p
) {
555 read_generation_data
= p
->read_generation_data
;
559 if (read_generation_data
)
563 g
->read_generation_data
= 0;
570 static int add_graph_to_chain(struct commit_graph
*g
,
571 struct commit_graph
*chain
,
572 struct object_id
*oids
,
575 struct commit_graph
*cur_g
= chain
;
577 if (n
&& !g
->chunk_base_graphs
) {
578 warning(_("commit-graph has no base graphs chunk"));
582 if (g
->chunk_base_graphs_size
/ g
->hash_len
< n
) {
583 warning(_("commit-graph base graphs chunk is too small"));
591 !oideq(&oids
[n
], &cur_g
->oid
) ||
592 !hasheq(oids
[n
].hash
, g
->chunk_base_graphs
+ st_mult(g
->hash_len
, n
))) {
593 warning(_("commit-graph chain does not match"));
597 cur_g
= cur_g
->base_graph
;
601 if (unsigned_add_overflows(chain
->num_commits
,
602 chain
->num_commits_in_base
)) {
603 warning(_("commit count in base graph too high: %"PRIuMAX
),
604 (uintmax_t)chain
->num_commits_in_base
);
607 g
->num_commits_in_base
= chain
->num_commits
+ chain
->num_commits_in_base
;
610 g
->base_graph
= chain
;
615 int open_commit_graph_chain(const char *chain_file
,
616 int *fd
, struct stat
*st
)
618 *fd
= git_open(chain_file
);
621 if (fstat(*fd
, st
)) {
625 if (st
->st_size
< the_hash_algo
->hexsz
) {
628 /* treat empty files the same as missing */
631 warning("commit-graph chain file too small");
639 struct commit_graph
*load_commit_graph_chain_fd_st(struct repository
*r
,
640 int fd
, struct stat
*st
,
641 int *incomplete_chain
)
643 struct commit_graph
*graph_chain
= NULL
;
644 struct strbuf line
= STRBUF_INIT
;
645 struct object_id
*oids
;
646 int i
= 0, valid
= 1, count
;
647 FILE *fp
= xfdopen(fd
, "r");
649 count
= st
->st_size
/ (the_hash_algo
->hexsz
+ 1);
650 CALLOC_ARRAY(oids
, count
);
654 for (i
= 0; i
< count
; i
++) {
655 struct object_directory
*odb
;
657 if (strbuf_getline_lf(&line
, fp
) == EOF
)
660 if (get_oid_hex(line
.buf
, &oids
[i
])) {
661 warning(_("invalid commit-graph chain: line '%s' not a hash"),
668 for (odb
= r
->objects
->odb
; odb
; odb
= odb
->next
) {
669 char *graph_name
= get_split_graph_filename(odb
, line
.buf
);
670 struct commit_graph
*g
= load_commit_graph_one(r
, graph_name
, odb
);
675 if (add_graph_to_chain(g
, graph_chain
, oids
, i
)) {
679 free_commit_graph(g
);
687 warning(_("unable to find all commit-graph files"));
692 validate_mixed_generation_chain(graph_chain
);
696 strbuf_release(&line
);
698 *incomplete_chain
= !valid
;
702 static struct commit_graph
*load_commit_graph_chain(struct repository
*r
,
703 struct object_directory
*odb
)
705 char *chain_file
= get_commit_graph_chain_filename(odb
);
708 struct commit_graph
*g
= NULL
;
710 if (open_commit_graph_chain(chain_file
, &fd
, &st
)) {
712 /* ownership of fd is taken over by load function */
713 g
= load_commit_graph_chain_fd_st(r
, fd
, &st
, &incomplete
);
720 struct commit_graph
*read_commit_graph_one(struct repository
*r
,
721 struct object_directory
*odb
)
723 struct commit_graph
*g
= load_commit_graph_v1(r
, odb
);
726 g
= load_commit_graph_chain(r
, odb
);
731 static void prepare_commit_graph_one(struct repository
*r
,
732 struct object_directory
*odb
)
735 if (r
->objects
->commit_graph
)
738 r
->objects
->commit_graph
= read_commit_graph_one(r
, odb
);
742 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
744 * On the first invocation, this function attempts to load the commit
745 * graph if the_repository is configured to have one.
747 static int prepare_commit_graph(struct repository
*r
)
749 struct object_directory
*odb
;
752 * Early return if there is no git dir or if the commit graph is
755 * This must come before the "already attempted?" check below, because
756 * we want to disable even an already-loaded graph file.
758 if (!r
->gitdir
|| r
->commit_graph_disabled
)
761 if (r
->objects
->commit_graph_attempted
)
762 return !!r
->objects
->commit_graph
;
763 r
->objects
->commit_graph_attempted
= 1;
765 prepare_repo_settings(r
);
767 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH
, 0) &&
768 r
->settings
.core_commit_graph
!= 1)
770 * This repository is not configured to use commit graphs, so
771 * do not load one. (But report commit_graph_attempted anyway
772 * so that commit graph loading is not attempted again for this
777 if (!commit_graph_compatible(r
))
781 for (odb
= r
->objects
->odb
;
782 !r
->objects
->commit_graph
&& odb
;
784 prepare_commit_graph_one(r
, odb
);
785 return !!r
->objects
->commit_graph
;
788 int generation_numbers_enabled(struct repository
*r
)
790 uint32_t first_generation
;
791 struct commit_graph
*g
;
792 if (!prepare_commit_graph(r
))
795 g
= r
->objects
->commit_graph
;
800 first_generation
= get_be32(g
->chunk_commit_data
+
801 g
->hash_len
+ 8) >> 2;
803 return !!first_generation
;
806 int corrected_commit_dates_enabled(struct repository
*r
)
808 struct commit_graph
*g
;
809 if (!prepare_commit_graph(r
))
812 g
= r
->objects
->commit_graph
;
817 return g
->read_generation_data
;
820 struct bloom_filter_settings
*get_bloom_filter_settings(struct repository
*r
)
822 struct commit_graph
*g
= r
->objects
->commit_graph
;
824 if (g
->bloom_filter_settings
)
825 return g
->bloom_filter_settings
;
831 void close_commit_graph(struct raw_object_store
*o
)
833 clear_commit_graph_data_slab(&commit_graph_data_slab
);
834 free_commit_graph(o
->commit_graph
);
835 o
->commit_graph
= NULL
;
838 static int bsearch_graph(struct commit_graph
*g
, const struct object_id
*oid
, uint32_t *pos
)
840 return bsearch_hash(oid
->hash
, g
->chunk_oid_fanout
,
841 g
->chunk_oid_lookup
, g
->hash_len
, pos
);
844 static void load_oid_from_graph(struct commit_graph
*g
,
846 struct object_id
*oid
)
850 while (g
&& pos
< g
->num_commits_in_base
)
854 BUG("NULL commit-graph");
856 if (pos
>= g
->num_commits
+ g
->num_commits_in_base
)
857 die(_("invalid commit position. commit-graph is likely corrupt"));
859 lex_index
= pos
- g
->num_commits_in_base
;
861 oidread(oid
, g
->chunk_oid_lookup
+ st_mult(g
->hash_len
, lex_index
));
864 static struct commit_list
**insert_parent_or_die(struct repository
*r
,
865 struct commit_graph
*g
,
867 struct commit_list
**pptr
)
870 struct object_id oid
;
872 if (pos
>= g
->num_commits
+ g
->num_commits_in_base
)
873 die("invalid parent position %"PRIu32
, pos
);
875 load_oid_from_graph(g
, pos
, &oid
);
876 c
= lookup_commit(r
, &oid
);
878 die(_("could not find commit %s"), oid_to_hex(&oid
));
879 commit_graph_data_at(c
)->graph_pos
= pos
;
880 return &commit_list_insert(c
, pptr
)->next
;
883 static void fill_commit_graph_info(struct commit
*item
, struct commit_graph
*g
, uint32_t pos
)
885 const unsigned char *commit_data
;
886 struct commit_graph_data
*graph_data
;
887 uint32_t lex_index
, offset_pos
;
888 uint64_t date_high
, date_low
, offset
;
890 while (pos
< g
->num_commits_in_base
)
893 if (pos
>= g
->num_commits
+ g
->num_commits_in_base
)
894 die(_("invalid commit position. commit-graph is likely corrupt"));
896 lex_index
= pos
- g
->num_commits_in_base
;
897 commit_data
= g
->chunk_commit_data
+ st_mult(GRAPH_DATA_WIDTH
, lex_index
);
899 graph_data
= commit_graph_data_at(item
);
900 graph_data
->graph_pos
= pos
;
902 date_high
= get_be32(commit_data
+ g
->hash_len
+ 8) & 0x3;
903 date_low
= get_be32(commit_data
+ g
->hash_len
+ 12);
904 item
->date
= (timestamp_t
)((date_high
<< 32) | date_low
);
906 if (g
->read_generation_data
) {
907 offset
= (timestamp_t
)get_be32(g
->chunk_generation_data
+ st_mult(sizeof(uint32_t), lex_index
));
909 if (offset
& CORRECTED_COMMIT_DATE_OFFSET_OVERFLOW
) {
910 if (!g
->chunk_generation_data_overflow
)
911 die(_("commit-graph requires overflow generation data but has none"));
913 offset_pos
= offset
^ CORRECTED_COMMIT_DATE_OFFSET_OVERFLOW
;
914 if (g
->chunk_generation_data_overflow_size
/ sizeof(uint64_t) <= offset_pos
)
915 die(_("commit-graph overflow generation data is too small"));
916 graph_data
->generation
= item
->date
+
917 get_be64(g
->chunk_generation_data_overflow
+ sizeof(uint64_t) * offset_pos
);
919 graph_data
->generation
= item
->date
+ offset
;
921 graph_data
->generation
= get_be32(commit_data
+ g
->hash_len
+ 8) >> 2;
924 *topo_level_slab_at(g
->topo_levels
, item
) = get_be32(commit_data
+ g
->hash_len
+ 8) >> 2;
927 static inline void set_commit_tree(struct commit
*c
, struct tree
*t
)
932 static int fill_commit_in_graph(struct repository
*r
,
934 struct commit_graph
*g
, uint32_t pos
)
937 uint32_t parent_data_pos
;
938 struct commit_list
**pptr
;
939 const unsigned char *commit_data
;
942 while (pos
< g
->num_commits_in_base
)
945 fill_commit_graph_info(item
, g
, pos
);
947 lex_index
= pos
- g
->num_commits_in_base
;
948 commit_data
= g
->chunk_commit_data
+ st_mult(g
->hash_len
+ 16, lex_index
);
950 item
->object
.parsed
= 1;
952 set_commit_tree(item
, NULL
);
954 pptr
= &item
->parents
;
956 edge_value
= get_be32(commit_data
+ g
->hash_len
);
957 if (edge_value
== GRAPH_PARENT_NONE
)
959 pptr
= insert_parent_or_die(r
, g
, edge_value
, pptr
);
961 edge_value
= get_be32(commit_data
+ g
->hash_len
+ 4);
962 if (edge_value
== GRAPH_PARENT_NONE
)
964 if (!(edge_value
& GRAPH_EXTRA_EDGES_NEEDED
)) {
965 pptr
= insert_parent_or_die(r
, g
, edge_value
, pptr
);
969 parent_data_pos
= edge_value
& GRAPH_EDGE_LAST_MASK
;
971 if (g
->chunk_extra_edges_size
/ sizeof(uint32_t) <= parent_data_pos
) {
972 error("commit-graph extra-edges pointer out of bounds");
973 free_commit_list(item
->parents
);
974 item
->parents
= NULL
;
975 item
->object
.parsed
= 0;
978 edge_value
= get_be32(g
->chunk_extra_edges
+
979 sizeof(uint32_t) * parent_data_pos
);
980 pptr
= insert_parent_or_die(r
, g
,
981 edge_value
& GRAPH_EDGE_LAST_MASK
,
984 } while (!(edge_value
& GRAPH_LAST_EDGE
));
989 static int search_commit_pos_in_graph(const struct object_id
*id
, struct commit_graph
*g
, uint32_t *pos
)
991 struct commit_graph
*cur_g
= g
;
994 while (cur_g
&& !bsearch_graph(cur_g
, id
, &lex_index
))
995 cur_g
= cur_g
->base_graph
;
998 *pos
= lex_index
+ cur_g
->num_commits_in_base
;
1005 static int find_commit_pos_in_graph(struct commit
*item
, struct commit_graph
*g
, uint32_t *pos
)
1007 uint32_t graph_pos
= commit_graph_position(item
);
1008 if (graph_pos
!= COMMIT_NOT_FROM_GRAPH
) {
1012 return search_commit_pos_in_graph(&item
->object
.oid
, g
, pos
);
1016 int repo_find_commit_pos_in_graph(struct repository
*r
, struct commit
*c
,
1019 if (!prepare_commit_graph(r
))
1021 return find_commit_pos_in_graph(c
, r
->objects
->commit_graph
, pos
);
1024 struct commit
*lookup_commit_in_graph(struct repository
*repo
, const struct object_id
*id
)
1026 static int commit_graph_paranoia
= -1;
1027 struct commit
*commit
;
1030 if (commit_graph_paranoia
== -1)
1031 commit_graph_paranoia
= git_env_bool(GIT_COMMIT_GRAPH_PARANOIA
, 1);
1033 if (!prepare_commit_graph(repo
))
1035 if (!search_commit_pos_in_graph(id
, repo
->objects
->commit_graph
, &pos
))
1037 if (commit_graph_paranoia
&& !has_object(repo
, id
, 0))
1040 commit
= lookup_commit(repo
, id
);
1043 if (commit
->object
.parsed
)
1046 if (!fill_commit_in_graph(repo
, commit
, repo
->objects
->commit_graph
, pos
))
1052 static int parse_commit_in_graph_one(struct repository
*r
,
1053 struct commit_graph
*g
,
1054 struct commit
*item
)
1058 if (item
->object
.parsed
)
1061 if (find_commit_pos_in_graph(item
, g
, &pos
))
1062 return fill_commit_in_graph(r
, item
, g
, pos
);
1067 int parse_commit_in_graph(struct repository
*r
, struct commit
*item
)
1069 static int checked_env
= 0;
1072 git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_PARSE
, 0))
1073 die("dying as requested by the '%s' variable on commit-graph parse!",
1074 GIT_TEST_COMMIT_GRAPH_DIE_ON_PARSE
);
1077 if (!prepare_commit_graph(r
))
1079 return parse_commit_in_graph_one(r
, r
->objects
->commit_graph
, item
);
1082 void load_commit_graph_info(struct repository
*r
, struct commit
*item
)
1085 if (repo_find_commit_pos_in_graph(r
, item
, &pos
))
1086 fill_commit_graph_info(item
, r
->objects
->commit_graph
, pos
);
1089 static struct tree
*load_tree_for_commit(struct repository
*r
,
1090 struct commit_graph
*g
,
1093 struct object_id oid
;
1094 const unsigned char *commit_data
;
1095 uint32_t graph_pos
= commit_graph_position(c
);
1097 while (graph_pos
< g
->num_commits_in_base
)
1100 commit_data
= g
->chunk_commit_data
+
1101 st_mult(GRAPH_DATA_WIDTH
, graph_pos
- g
->num_commits_in_base
);
1103 oidread(&oid
, commit_data
);
1104 set_commit_tree(c
, lookup_tree(r
, &oid
));
1106 return c
->maybe_tree
;
1109 static struct tree
*get_commit_tree_in_graph_one(struct repository
*r
,
1110 struct commit_graph
*g
,
1111 const struct commit
*c
)
1114 return c
->maybe_tree
;
1115 if (commit_graph_position(c
) == COMMIT_NOT_FROM_GRAPH
)
1116 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
1118 return load_tree_for_commit(r
, g
, (struct commit
*)c
);
1121 struct tree
*get_commit_tree_in_graph(struct repository
*r
, const struct commit
*c
)
1123 return get_commit_tree_in_graph_one(r
, r
->objects
->commit_graph
, c
);
1126 struct packed_commit_list
{
1127 struct commit
**list
;
1132 struct write_commit_graph_context
{
1133 struct repository
*r
;
1134 struct object_directory
*odb
;
1136 struct oid_array oids
;
1137 struct packed_commit_list commits
;
1138 int num_extra_edges
;
1139 int num_generation_data_overflows
;
1140 unsigned long approx_nr_objects
;
1141 struct progress
*progress
;
1143 uint64_t progress_cnt
;
1145 char *base_graph_name
;
1146 int num_commit_graphs_before
;
1147 int num_commit_graphs_after
;
1148 char **commit_graph_filenames_before
;
1149 char **commit_graph_filenames_after
;
1150 char **commit_graph_hash_after
;
1151 uint32_t new_num_commits_in_base
;
1152 struct commit_graph
*new_base_graph
;
1159 write_generation_data
:1,
1160 trust_generation_numbers
:1;
1162 struct topo_level_slab
*topo_levels
;
1163 const struct commit_graph_opts
*opts
;
1164 size_t total_bloom_filter_data_size
;
1165 const struct bloom_filter_settings
*bloom_settings
;
1167 int count_bloom_filter_computed
;
1168 int count_bloom_filter_not_computed
;
1169 int count_bloom_filter_trunc_empty
;
1170 int count_bloom_filter_trunc_large
;
1173 static int write_graph_chunk_fanout(struct hashfile
*f
,
1176 struct write_commit_graph_context
*ctx
= data
;
1178 struct commit
**list
= ctx
->commits
.list
;
1181 * Write the first-level table (the list is sorted,
1182 * but we use a 256-entry lookup to be able to avoid
1183 * having to do eight extra binary search iterations).
1185 for (i
= 0; i
< 256; i
++) {
1186 while (count
< ctx
->commits
.nr
) {
1187 if ((*list
)->object
.oid
.hash
[0] != i
)
1189 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1194 hashwrite_be32(f
, count
);
1200 static int write_graph_chunk_oids(struct hashfile
*f
,
1203 struct write_commit_graph_context
*ctx
= data
;
1204 struct commit
**list
= ctx
->commits
.list
;
1206 for (count
= 0; count
< ctx
->commits
.nr
; count
++, list
++) {
1207 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1208 hashwrite(f
, (*list
)->object
.oid
.hash
, the_hash_algo
->rawsz
);
1214 static const struct object_id
*commit_to_oid(size_t index
, const void *table
)
1216 const struct commit
* const *commits
= table
;
1217 return &commits
[index
]->object
.oid
;
1220 static int write_graph_chunk_data(struct hashfile
*f
,
1223 struct write_commit_graph_context
*ctx
= data
;
1224 struct commit
**list
= ctx
->commits
.list
;
1225 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
1226 uint32_t num_extra_edges
= 0;
1228 while (list
< last
) {
1229 struct commit_list
*parent
;
1230 struct object_id
*tree
;
1232 uint32_t packedDate
[2];
1233 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1235 if (repo_parse_commit_no_graph(ctx
->r
, *list
))
1236 die(_("unable to parse commit %s"),
1237 oid_to_hex(&(*list
)->object
.oid
));
1238 tree
= get_commit_tree_oid(*list
);
1239 hashwrite(f
, tree
->hash
, the_hash_algo
->rawsz
);
1241 parent
= (*list
)->parents
;
1244 edge_value
= GRAPH_PARENT_NONE
;
1246 edge_value
= oid_pos(&parent
->item
->object
.oid
,
1251 if (edge_value
>= 0)
1252 edge_value
+= ctx
->new_num_commits_in_base
;
1253 else if (ctx
->new_base_graph
) {
1255 if (find_commit_pos_in_graph(parent
->item
,
1256 ctx
->new_base_graph
,
1262 BUG("missing parent %s for commit %s",
1263 oid_to_hex(&parent
->item
->object
.oid
),
1264 oid_to_hex(&(*list
)->object
.oid
));
1267 hashwrite_be32(f
, edge_value
);
1270 parent
= parent
->next
;
1273 edge_value
= GRAPH_PARENT_NONE
;
1274 else if (parent
->next
)
1275 edge_value
= GRAPH_EXTRA_EDGES_NEEDED
| num_extra_edges
;
1277 edge_value
= oid_pos(&parent
->item
->object
.oid
,
1282 if (edge_value
>= 0)
1283 edge_value
+= ctx
->new_num_commits_in_base
;
1284 else if (ctx
->new_base_graph
) {
1286 if (find_commit_pos_in_graph(parent
->item
,
1287 ctx
->new_base_graph
,
1293 BUG("missing parent %s for commit %s",
1294 oid_to_hex(&parent
->item
->object
.oid
),
1295 oid_to_hex(&(*list
)->object
.oid
));
1298 hashwrite_be32(f
, edge_value
);
1300 if (edge_value
& GRAPH_EXTRA_EDGES_NEEDED
) {
1303 parent
= parent
->next
;
1307 if (sizeof((*list
)->date
) > 4)
1308 packedDate
[0] = htonl(((*list
)->date
>> 32) & 0x3);
1312 packedDate
[0] |= htonl(*topo_level_slab_at(ctx
->topo_levels
, *list
) << 2);
1314 packedDate
[1] = htonl((*list
)->date
);
1315 hashwrite(f
, packedDate
, 8);
1323 static int write_graph_chunk_generation_data(struct hashfile
*f
,
1326 struct write_commit_graph_context
*ctx
= data
;
1327 int i
, num_generation_data_overflows
= 0;
1329 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1330 struct commit
*c
= ctx
->commits
.list
[i
];
1332 repo_parse_commit(ctx
->r
, c
);
1333 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 offset
= CORRECTED_COMMIT_DATE_OFFSET_OVERFLOW
| num_generation_data_overflows
;
1338 num_generation_data_overflows
++;
1341 hashwrite_be32(f
, offset
);
1347 static int write_graph_chunk_generation_data_overflow(struct hashfile
*f
,
1350 struct write_commit_graph_context
*ctx
= data
;
1352 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1353 struct commit
*c
= ctx
->commits
.list
[i
];
1354 timestamp_t offset
= commit_graph_data_at(c
)->generation
- c
->date
;
1355 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1357 if (offset
> GENERATION_NUMBER_V2_OFFSET_MAX
) {
1358 hashwrite_be32(f
, offset
>> 32);
1359 hashwrite_be32(f
, (uint32_t) offset
);
1366 static int write_graph_chunk_extra_edges(struct hashfile
*f
,
1369 struct write_commit_graph_context
*ctx
= data
;
1370 struct commit
**list
= ctx
->commits
.list
;
1371 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
1372 struct commit_list
*parent
;
1374 while (list
< last
) {
1375 int num_parents
= 0;
1377 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1379 for (parent
= (*list
)->parents
; num_parents
< 3 && parent
;
1380 parent
= parent
->next
)
1383 if (num_parents
<= 2) {
1388 /* Since num_parents > 2, this initializer is safe. */
1389 for (parent
= (*list
)->parents
->next
; parent
; parent
= parent
->next
) {
1390 int edge_value
= oid_pos(&parent
->item
->object
.oid
,
1395 if (edge_value
>= 0)
1396 edge_value
+= ctx
->new_num_commits_in_base
;
1397 else if (ctx
->new_base_graph
) {
1399 if (find_commit_pos_in_graph(parent
->item
,
1400 ctx
->new_base_graph
,
1406 BUG("missing parent %s for commit %s",
1407 oid_to_hex(&parent
->item
->object
.oid
),
1408 oid_to_hex(&(*list
)->object
.oid
));
1409 else if (!parent
->next
)
1410 edge_value
|= GRAPH_LAST_EDGE
;
1412 hashwrite_be32(f
, edge_value
);
1421 static int write_graph_chunk_bloom_indexes(struct hashfile
*f
,
1424 struct write_commit_graph_context
*ctx
= data
;
1425 struct commit
**list
= ctx
->commits
.list
;
1426 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
1427 uint32_t cur_pos
= 0;
1429 while (list
< last
) {
1430 struct bloom_filter
*filter
= get_bloom_filter(ctx
->r
, *list
);
1431 size_t len
= filter
? filter
->len
: 0;
1433 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1434 hashwrite_be32(f
, cur_pos
);
1441 static void trace2_bloom_filter_settings(struct write_commit_graph_context
*ctx
)
1443 struct json_writer jw
= JSON_WRITER_INIT
;
1445 jw_object_begin(&jw
, 0);
1446 jw_object_intmax(&jw
, "hash_version", ctx
->bloom_settings
->hash_version
);
1447 jw_object_intmax(&jw
, "num_hashes", ctx
->bloom_settings
->num_hashes
);
1448 jw_object_intmax(&jw
, "bits_per_entry", ctx
->bloom_settings
->bits_per_entry
);
1449 jw_object_intmax(&jw
, "max_changed_paths", ctx
->bloom_settings
->max_changed_paths
);
1452 trace2_data_json("bloom", ctx
->r
, "settings", &jw
);
1457 static int write_graph_chunk_bloom_data(struct hashfile
*f
,
1460 struct write_commit_graph_context
*ctx
= data
;
1461 struct commit
**list
= ctx
->commits
.list
;
1462 struct commit
**last
= ctx
->commits
.list
+ ctx
->commits
.nr
;
1464 trace2_bloom_filter_settings(ctx
);
1466 hashwrite_be32(f
, ctx
->bloom_settings
->hash_version
);
1467 hashwrite_be32(f
, ctx
->bloom_settings
->num_hashes
);
1468 hashwrite_be32(f
, ctx
->bloom_settings
->bits_per_entry
);
1470 while (list
< last
) {
1471 struct bloom_filter
*filter
= get_bloom_filter(ctx
->r
, *list
);
1472 size_t len
= filter
? filter
->len
: 0;
1474 display_progress(ctx
->progress
, ++ctx
->progress_cnt
);
1476 hashwrite(f
, filter
->data
, len
* sizeof(unsigned char));
1483 static int add_packed_commits(const struct object_id
*oid
,
1484 struct packed_git
*pack
,
1488 struct write_commit_graph_context
*ctx
= (struct write_commit_graph_context
*)data
;
1489 enum object_type type
;
1490 off_t offset
= nth_packed_object_offset(pack
, pos
);
1491 struct object_info oi
= OBJECT_INFO_INIT
;
1494 display_progress(ctx
->progress
, ++ctx
->progress_done
);
1497 if (packed_object_info(ctx
->r
, pack
, offset
, &oi
) < 0)
1498 die(_("unable to get type of object %s"), oid_to_hex(oid
));
1500 if (type
!= OBJ_COMMIT
)
1503 oid_array_append(&ctx
->oids
, oid
);
1504 set_commit_pos(ctx
->r
, oid
);
1509 static void add_missing_parents(struct write_commit_graph_context
*ctx
, struct commit
*commit
)
1511 struct commit_list
*parent
;
1512 for (parent
= commit
->parents
; parent
; parent
= parent
->next
) {
1513 if (!(parent
->item
->object
.flags
& REACHABLE
)) {
1514 oid_array_append(&ctx
->oids
, &parent
->item
->object
.oid
);
1515 parent
->item
->object
.flags
|= REACHABLE
;
1520 static void close_reachable(struct write_commit_graph_context
*ctx
)
1523 struct commit
*commit
;
1524 enum commit_graph_split_flags flags
= ctx
->opts
?
1525 ctx
->opts
->split_flags
: COMMIT_GRAPH_SPLIT_UNSPECIFIED
;
1527 if (ctx
->report_progress
)
1528 ctx
->progress
= start_delayed_progress(
1529 _("Loading known commits in commit graph"),
1531 for (i
= 0; i
< ctx
->oids
.nr
; i
++) {
1532 display_progress(ctx
->progress
, i
+ 1);
1533 commit
= lookup_commit(ctx
->r
, &ctx
->oids
.oid
[i
]);
1535 commit
->object
.flags
|= REACHABLE
;
1537 stop_progress(&ctx
->progress
);
1540 * As this loop runs, ctx->oids.nr may grow, but not more
1541 * than the number of missing commits in the reachable
1544 if (ctx
->report_progress
)
1545 ctx
->progress
= start_delayed_progress(
1546 _("Expanding reachable commits in commit graph"),
1548 for (i
= 0; i
< ctx
->oids
.nr
; i
++) {
1549 display_progress(ctx
->progress
, i
+ 1);
1550 commit
= lookup_commit(ctx
->r
, &ctx
->oids
.oid
[i
]);
1555 if ((!repo_parse_commit(ctx
->r
, commit
) &&
1556 commit_graph_position(commit
) == COMMIT_NOT_FROM_GRAPH
) ||
1557 flags
== COMMIT_GRAPH_SPLIT_REPLACE
)
1558 add_missing_parents(ctx
, commit
);
1559 } else if (!repo_parse_commit_no_graph(ctx
->r
, commit
))
1560 add_missing_parents(ctx
, commit
);
1562 stop_progress(&ctx
->progress
);
1564 if (ctx
->report_progress
)
1565 ctx
->progress
= start_delayed_progress(
1566 _("Clearing commit marks in commit graph"),
1568 for (i
= 0; i
< ctx
->oids
.nr
; i
++) {
1569 display_progress(ctx
->progress
, i
+ 1);
1570 commit
= lookup_commit(ctx
->r
, &ctx
->oids
.oid
[i
]);
1573 commit
->object
.flags
&= ~REACHABLE
;
1575 stop_progress(&ctx
->progress
);
1578 struct compute_generation_info
{
1579 struct repository
*r
;
1580 struct packed_commit_list
*commits
;
1581 struct progress
*progress
;
1584 timestamp_t (*get_generation
)(struct commit
*c
, void *data
);
1585 void (*set_generation
)(struct commit
*c
, timestamp_t gen
, void *data
);
1589 static timestamp_t
compute_generation_from_max(struct commit
*c
,
1590 timestamp_t max_gen
,
1591 int generation_version
)
1593 switch (generation_version
) {
1594 case 1: /* topological levels */
1595 if (max_gen
> GENERATION_NUMBER_V1_MAX
- 1)
1596 max_gen
= GENERATION_NUMBER_V1_MAX
- 1;
1599 case 2: /* corrected commit date */
1600 if (c
->date
&& c
->date
> max_gen
)
1601 max_gen
= c
->date
- 1;
1605 BUG("attempting unimplemented version");
1609 static void compute_reachable_generation_numbers(
1610 struct compute_generation_info
*info
,
1611 int generation_version
)
1614 struct commit_list
*list
= NULL
;
1616 for (i
= 0; i
< info
->commits
->nr
; i
++) {
1617 struct commit
*c
= info
->commits
->list
[i
];
1619 repo_parse_commit(info
->r
, c
);
1620 gen
= info
->get_generation(c
, info
->data
);
1621 display_progress(info
->progress
, info
->progress_cnt
+ 1);
1623 if (gen
!= GENERATION_NUMBER_ZERO
&& gen
!= GENERATION_NUMBER_INFINITY
)
1626 commit_list_insert(c
, &list
);
1628 struct commit
*current
= list
->item
;
1629 struct commit_list
*parent
;
1630 int all_parents_computed
= 1;
1631 uint32_t max_gen
= 0;
1633 for (parent
= current
->parents
; parent
; parent
= parent
->next
) {
1634 repo_parse_commit(info
->r
, parent
->item
);
1635 gen
= info
->get_generation(parent
->item
, info
->data
);
1637 if (gen
== GENERATION_NUMBER_ZERO
) {
1638 all_parents_computed
= 0;
1639 commit_list_insert(parent
->item
, &list
);
1647 if (all_parents_computed
) {
1649 gen
= compute_generation_from_max(
1651 generation_version
);
1652 info
->set_generation(current
, gen
, info
->data
);
1658 static timestamp_t
get_topo_level(struct commit
*c
, void *data
)
1660 struct write_commit_graph_context
*ctx
= data
;
1661 return *topo_level_slab_at(ctx
->topo_levels
, c
);
1664 static void set_topo_level(struct commit
*c
, timestamp_t t
, void *data
)
1666 struct write_commit_graph_context
*ctx
= data
;
1667 *topo_level_slab_at(ctx
->topo_levels
, c
) = (uint32_t)t
;
1670 static void compute_topological_levels(struct write_commit_graph_context
*ctx
)
1672 struct compute_generation_info info
= {
1674 .commits
= &ctx
->commits
,
1675 .get_generation
= get_topo_level
,
1676 .set_generation
= set_topo_level
,
1680 if (ctx
->report_progress
)
1681 info
.progress
= ctx
->progress
1682 = start_delayed_progress(
1683 _("Computing commit graph topological levels"),
1686 compute_reachable_generation_numbers(&info
, 1);
1688 stop_progress(&ctx
->progress
);
1691 static timestamp_t
get_generation_from_graph_data(struct commit
*c
,
1694 return commit_graph_data_at(c
)->generation
;
1697 static void set_generation_v2(struct commit
*c
, timestamp_t t
,
1700 struct commit_graph_data
*g
= commit_graph_data_at(c
);
1704 static void compute_generation_numbers(struct write_commit_graph_context
*ctx
)
1707 struct compute_generation_info info
= {
1709 .commits
= &ctx
->commits
,
1710 .get_generation
= get_generation_from_graph_data
,
1711 .set_generation
= set_generation_v2
,
1714 if (ctx
->report_progress
)
1715 info
.progress
= ctx
->progress
1716 = start_delayed_progress(
1717 _("Computing commit graph generation numbers"),
1720 if (!ctx
->trust_generation_numbers
) {
1721 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1722 struct commit
*c
= ctx
->commits
.list
[i
];
1723 repo_parse_commit(ctx
->r
, c
);
1724 commit_graph_data_at(c
)->generation
= GENERATION_NUMBER_ZERO
;
1728 compute_reachable_generation_numbers(&info
, 2);
1730 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1731 struct commit
*c
= ctx
->commits
.list
[i
];
1732 timestamp_t offset
= commit_graph_data_at(c
)->generation
- c
->date
;
1733 if (offset
> GENERATION_NUMBER_V2_OFFSET_MAX
)
1734 ctx
->num_generation_data_overflows
++;
1736 stop_progress(&ctx
->progress
);
1739 static void set_generation_in_graph_data(struct commit
*c
, timestamp_t t
,
1742 commit_graph_data_at(c
)->generation
= t
;
1746 * After this method, all commits reachable from those in the given
1747 * list will have non-zero, non-infinite generation numbers.
1749 void ensure_generations_valid(struct repository
*r
,
1750 struct commit
**commits
, size_t nr
)
1752 int generation_version
= get_configured_generation_version(r
);
1753 struct packed_commit_list list
= {
1758 struct compute_generation_info info
= {
1761 .get_generation
= get_generation_from_graph_data
,
1762 .set_generation
= set_generation_in_graph_data
,
1765 compute_reachable_generation_numbers(&info
, generation_version
);
1768 static void trace2_bloom_filter_write_statistics(struct write_commit_graph_context
*ctx
)
1770 trace2_data_intmax("commit-graph", ctx
->r
, "filter-computed",
1771 ctx
->count_bloom_filter_computed
);
1772 trace2_data_intmax("commit-graph", ctx
->r
, "filter-not-computed",
1773 ctx
->count_bloom_filter_not_computed
);
1774 trace2_data_intmax("commit-graph", ctx
->r
, "filter-trunc-empty",
1775 ctx
->count_bloom_filter_trunc_empty
);
1776 trace2_data_intmax("commit-graph", ctx
->r
, "filter-trunc-large",
1777 ctx
->count_bloom_filter_trunc_large
);
1780 static void compute_bloom_filters(struct write_commit_graph_context
*ctx
)
1783 struct progress
*progress
= NULL
;
1784 struct commit
**sorted_commits
;
1785 int max_new_filters
;
1787 init_bloom_filters();
1789 if (ctx
->report_progress
)
1790 progress
= start_delayed_progress(
1791 _("Computing commit changed paths Bloom filters"),
1794 DUP_ARRAY(sorted_commits
, ctx
->commits
.list
, ctx
->commits
.nr
);
1796 if (ctx
->order_by_pack
)
1797 QSORT(sorted_commits
, ctx
->commits
.nr
, commit_pos_cmp
);
1799 QSORT(sorted_commits
, ctx
->commits
.nr
, commit_gen_cmp
);
1801 max_new_filters
= ctx
->opts
&& ctx
->opts
->max_new_filters
>= 0 ?
1802 ctx
->opts
->max_new_filters
: ctx
->commits
.nr
;
1804 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
1805 enum bloom_filter_computed computed
= 0;
1806 struct commit
*c
= sorted_commits
[i
];
1807 struct bloom_filter
*filter
= get_or_compute_bloom_filter(
1810 ctx
->count_bloom_filter_computed
< max_new_filters
,
1811 ctx
->bloom_settings
,
1813 if (computed
& BLOOM_COMPUTED
) {
1814 ctx
->count_bloom_filter_computed
++;
1815 if (computed
& BLOOM_TRUNC_EMPTY
)
1816 ctx
->count_bloom_filter_trunc_empty
++;
1817 if (computed
& BLOOM_TRUNC_LARGE
)
1818 ctx
->count_bloom_filter_trunc_large
++;
1819 } else if (computed
& BLOOM_NOT_COMPUTED
)
1820 ctx
->count_bloom_filter_not_computed
++;
1821 ctx
->total_bloom_filter_data_size
+= filter
1822 ? sizeof(unsigned char) * filter
->len
: 0;
1823 display_progress(progress
, i
+ 1);
1826 if (trace2_is_enabled())
1827 trace2_bloom_filter_write_statistics(ctx
);
1829 free(sorted_commits
);
1830 stop_progress(&progress
);
1833 struct refs_cb_data
{
1834 struct oidset
*commits
;
1835 struct progress
*progress
;
1838 static int add_ref_to_set(const char *refname UNUSED
,
1839 const struct object_id
*oid
,
1840 int flags UNUSED
, void *cb_data
)
1842 struct object_id peeled
;
1843 struct refs_cb_data
*data
= (struct refs_cb_data
*)cb_data
;
1845 if (!peel_iterated_oid(oid
, &peeled
))
1847 if (oid_object_info(the_repository
, oid
, NULL
) == OBJ_COMMIT
)
1848 oidset_insert(data
->commits
, oid
);
1850 display_progress(data
->progress
, oidset_size(data
->commits
));
1855 int write_commit_graph_reachable(struct object_directory
*odb
,
1856 enum commit_graph_write_flags flags
,
1857 const struct commit_graph_opts
*opts
)
1859 struct oidset commits
= OIDSET_INIT
;
1860 struct refs_cb_data data
;
1863 memset(&data
, 0, sizeof(data
));
1864 data
.commits
= &commits
;
1865 if (flags
& COMMIT_GRAPH_WRITE_PROGRESS
)
1866 data
.progress
= start_delayed_progress(
1867 _("Collecting referenced commits"), 0);
1869 for_each_ref(add_ref_to_set
, &data
);
1871 stop_progress(&data
.progress
);
1873 result
= write_commit_graph(odb
, NULL
, &commits
,
1876 oidset_clear(&commits
);
1880 static int fill_oids_from_packs(struct write_commit_graph_context
*ctx
,
1881 const struct string_list
*pack_indexes
)
1884 struct strbuf progress_title
= STRBUF_INIT
;
1885 struct strbuf packname
= STRBUF_INIT
;
1889 strbuf_addf(&packname
, "%s/pack/", ctx
->odb
->path
);
1890 dirlen
= packname
.len
;
1891 if (ctx
->report_progress
) {
1892 strbuf_addf(&progress_title
,
1893 Q_("Finding commits for commit graph in %"PRIuMAX
" pack",
1894 "Finding commits for commit graph in %"PRIuMAX
" packs",
1896 (uintmax_t)pack_indexes
->nr
);
1897 ctx
->progress
= start_delayed_progress(progress_title
.buf
, 0);
1898 ctx
->progress_done
= 0;
1900 for (i
= 0; i
< pack_indexes
->nr
; i
++) {
1901 struct packed_git
*p
;
1902 strbuf_setlen(&packname
, dirlen
);
1903 strbuf_addstr(&packname
, pack_indexes
->items
[i
].string
);
1904 p
= add_packed_git(packname
.buf
, packname
.len
, 1);
1906 ret
= error(_("error adding pack %s"), packname
.buf
);
1909 if (open_pack_index(p
)) {
1910 ret
= error(_("error opening index for %s"), packname
.buf
);
1913 for_each_object_in_pack(p
, add_packed_commits
, ctx
,
1914 FOR_EACH_OBJECT_PACK_ORDER
);
1920 stop_progress(&ctx
->progress
);
1921 strbuf_release(&progress_title
);
1922 strbuf_release(&packname
);
1927 static int fill_oids_from_commits(struct write_commit_graph_context
*ctx
,
1928 struct oidset
*commits
)
1930 struct oidset_iter iter
;
1931 struct object_id
*oid
;
1933 if (!oidset_size(commits
))
1936 oidset_iter_init(commits
, &iter
);
1937 while ((oid
= oidset_iter_next(&iter
))) {
1938 oid_array_append(&ctx
->oids
, oid
);
1944 static void fill_oids_from_all_packs(struct write_commit_graph_context
*ctx
)
1946 if (ctx
->report_progress
)
1947 ctx
->progress
= start_delayed_progress(
1948 _("Finding commits for commit graph among packed objects"),
1949 ctx
->approx_nr_objects
);
1950 for_each_packed_object(add_packed_commits
, ctx
,
1951 FOR_EACH_OBJECT_PACK_ORDER
);
1952 if (ctx
->progress_done
< ctx
->approx_nr_objects
)
1953 display_progress(ctx
->progress
, ctx
->approx_nr_objects
);
1954 stop_progress(&ctx
->progress
);
1957 static void copy_oids_to_commits(struct write_commit_graph_context
*ctx
)
1960 enum commit_graph_split_flags flags
= ctx
->opts
?
1961 ctx
->opts
->split_flags
: COMMIT_GRAPH_SPLIT_UNSPECIFIED
;
1963 ctx
->num_extra_edges
= 0;
1964 if (ctx
->report_progress
)
1965 ctx
->progress
= start_delayed_progress(
1966 _("Finding extra edges in commit graph"),
1968 oid_array_sort(&ctx
->oids
);
1969 for (i
= 0; i
< ctx
->oids
.nr
; i
= oid_array_next_unique(&ctx
->oids
, i
)) {
1970 unsigned int num_parents
;
1972 display_progress(ctx
->progress
, i
+ 1);
1974 ALLOC_GROW(ctx
->commits
.list
, ctx
->commits
.nr
+ 1, ctx
->commits
.alloc
);
1975 ctx
->commits
.list
[ctx
->commits
.nr
] = lookup_commit(ctx
->r
, &ctx
->oids
.oid
[i
]);
1977 if (ctx
->split
&& flags
!= COMMIT_GRAPH_SPLIT_REPLACE
&&
1978 commit_graph_position(ctx
->commits
.list
[ctx
->commits
.nr
]) != COMMIT_NOT_FROM_GRAPH
)
1981 if (ctx
->split
&& flags
== COMMIT_GRAPH_SPLIT_REPLACE
)
1982 repo_parse_commit(ctx
->r
, ctx
->commits
.list
[ctx
->commits
.nr
]);
1984 repo_parse_commit_no_graph(ctx
->r
, ctx
->commits
.list
[ctx
->commits
.nr
]);
1986 num_parents
= commit_list_count(ctx
->commits
.list
[ctx
->commits
.nr
]->parents
);
1987 if (num_parents
> 2)
1988 ctx
->num_extra_edges
+= num_parents
- 1;
1992 stop_progress(&ctx
->progress
);
1995 static int write_graph_chunk_base_1(struct hashfile
*f
,
1996 struct commit_graph
*g
)
2003 num
= write_graph_chunk_base_1(f
, g
->base_graph
);
2004 hashwrite(f
, g
->oid
.hash
, the_hash_algo
->rawsz
);
2008 static int write_graph_chunk_base(struct hashfile
*f
,
2011 struct write_commit_graph_context
*ctx
= data
;
2012 int num
= write_graph_chunk_base_1(f
, ctx
->new_base_graph
);
2014 if (num
!= ctx
->num_commit_graphs_after
- 1) {
2015 error(_("failed to write correct number of base graph ids"));
2022 static int write_commit_graph_file(struct write_commit_graph_context
*ctx
)
2027 struct lock_file lk
= LOCK_INIT
;
2028 const unsigned hashsz
= the_hash_algo
->rawsz
;
2029 struct strbuf progress_title
= STRBUF_INIT
;
2030 struct chunkfile
*cf
;
2031 unsigned char file_hash
[GIT_MAX_RAWSZ
];
2034 struct strbuf tmp_file
= STRBUF_INIT
;
2036 strbuf_addf(&tmp_file
,
2037 "%s/info/commit-graphs/tmp_graph_XXXXXX",
2039 ctx
->graph_name
= strbuf_detach(&tmp_file
, NULL
);
2041 ctx
->graph_name
= get_commit_graph_filename(ctx
->odb
);
2044 if (safe_create_leading_directories(ctx
->graph_name
)) {
2045 UNLEAK(ctx
->graph_name
);
2046 error(_("unable to create leading directories of %s"),
2052 char *lock_name
= get_commit_graph_chain_filename(ctx
->odb
);
2054 hold_lock_file_for_update_mode(&lk
, lock_name
,
2055 LOCK_DIE_ON_ERROR
, 0444);
2058 fd
= git_mkstemp_mode(ctx
->graph_name
, 0444);
2060 error(_("unable to create temporary graph layer"));
2064 if (adjust_shared_perm(ctx
->graph_name
)) {
2065 error(_("unable to adjust shared permissions for '%s'"),
2070 f
= hashfd(fd
, ctx
->graph_name
);
2072 hold_lock_file_for_update_mode(&lk
, ctx
->graph_name
,
2073 LOCK_DIE_ON_ERROR
, 0444);
2074 fd
= get_lock_file_fd(&lk
);
2075 f
= hashfd(fd
, get_lock_file_path(&lk
));
2078 cf
= init_chunkfile(f
);
2080 add_chunk(cf
, GRAPH_CHUNKID_OIDFANOUT
, GRAPH_FANOUT_SIZE
,
2081 write_graph_chunk_fanout
);
2082 add_chunk(cf
, GRAPH_CHUNKID_OIDLOOKUP
, st_mult(hashsz
, ctx
->commits
.nr
),
2083 write_graph_chunk_oids
);
2084 add_chunk(cf
, GRAPH_CHUNKID_DATA
, st_mult(hashsz
+ 16, ctx
->commits
.nr
),
2085 write_graph_chunk_data
);
2087 if (ctx
->write_generation_data
)
2088 add_chunk(cf
, GRAPH_CHUNKID_GENERATION_DATA
,
2089 st_mult(sizeof(uint32_t), ctx
->commits
.nr
),
2090 write_graph_chunk_generation_data
);
2091 if (ctx
->num_generation_data_overflows
)
2092 add_chunk(cf
, GRAPH_CHUNKID_GENERATION_DATA_OVERFLOW
,
2093 st_mult(sizeof(timestamp_t
), ctx
->num_generation_data_overflows
),
2094 write_graph_chunk_generation_data_overflow
);
2095 if (ctx
->num_extra_edges
)
2096 add_chunk(cf
, GRAPH_CHUNKID_EXTRAEDGES
,
2097 st_mult(4, ctx
->num_extra_edges
),
2098 write_graph_chunk_extra_edges
);
2099 if (ctx
->changed_paths
) {
2100 add_chunk(cf
, GRAPH_CHUNKID_BLOOMINDEXES
,
2101 st_mult(sizeof(uint32_t), ctx
->commits
.nr
),
2102 write_graph_chunk_bloom_indexes
);
2103 add_chunk(cf
, GRAPH_CHUNKID_BLOOMDATA
,
2104 st_add(sizeof(uint32_t) * 3,
2105 ctx
->total_bloom_filter_data_size
),
2106 write_graph_chunk_bloom_data
);
2108 if (ctx
->num_commit_graphs_after
> 1)
2109 add_chunk(cf
, GRAPH_CHUNKID_BASE
,
2110 st_mult(hashsz
, ctx
->num_commit_graphs_after
- 1),
2111 write_graph_chunk_base
);
2113 hashwrite_be32(f
, GRAPH_SIGNATURE
);
2115 hashwrite_u8(f
, GRAPH_VERSION
);
2116 hashwrite_u8(f
, oid_version(the_hash_algo
));
2117 hashwrite_u8(f
, get_num_chunks(cf
));
2118 hashwrite_u8(f
, ctx
->num_commit_graphs_after
- 1);
2120 if (ctx
->report_progress
) {
2121 strbuf_addf(&progress_title
,
2122 Q_("Writing out commit graph in %d pass",
2123 "Writing out commit graph in %d passes",
2124 get_num_chunks(cf
)),
2125 get_num_chunks(cf
));
2126 ctx
->progress
= start_delayed_progress(
2128 st_mult(get_num_chunks(cf
), ctx
->commits
.nr
));
2131 write_chunkfile(cf
, ctx
);
2133 stop_progress(&ctx
->progress
);
2134 strbuf_release(&progress_title
);
2136 if (ctx
->split
&& ctx
->base_graph_name
&& ctx
->num_commit_graphs_after
> 1) {
2137 char *new_base_hash
= xstrdup(oid_to_hex(&ctx
->new_base_graph
->oid
));
2138 char *new_base_name
= get_split_graph_filename(ctx
->new_base_graph
->odb
, new_base_hash
);
2140 free(ctx
->commit_graph_filenames_after
[ctx
->num_commit_graphs_after
- 2]);
2141 free(ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 2]);
2142 ctx
->commit_graph_filenames_after
[ctx
->num_commit_graphs_after
- 2] = new_base_name
;
2143 ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 2] = new_base_hash
;
2146 close_commit_graph(ctx
->r
->objects
);
2147 finalize_hashfile(f
, file_hash
, FSYNC_COMPONENT_COMMIT_GRAPH
,
2148 CSUM_HASH_IN_STREAM
| CSUM_FSYNC
);
2152 FILE *chainf
= fdopen_lock_file(&lk
, "w");
2153 char *final_graph_name
;
2159 error(_("unable to open commit-graph chain file"));
2163 if (ctx
->base_graph_name
) {
2165 int idx
= ctx
->num_commit_graphs_after
- 1;
2166 if (ctx
->num_commit_graphs_after
> 1)
2169 dest
= ctx
->commit_graph_filenames_after
[idx
];
2171 if (strcmp(ctx
->base_graph_name
, dest
)) {
2172 result
= rename(ctx
->base_graph_name
, dest
);
2175 error(_("failed to rename base commit-graph file"));
2180 char *graph_name
= get_commit_graph_filename(ctx
->odb
);
2185 free(ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 1]);
2186 ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 1] = xstrdup(hash_to_hex(file_hash
));
2187 final_graph_name
= get_split_graph_filename(ctx
->odb
,
2188 ctx
->commit_graph_hash_after
[ctx
->num_commit_graphs_after
- 1]);
2189 free(ctx
->commit_graph_filenames_after
[ctx
->num_commit_graphs_after
- 1]);
2190 ctx
->commit_graph_filenames_after
[ctx
->num_commit_graphs_after
- 1] = final_graph_name
;
2192 result
= rename(ctx
->graph_name
, final_graph_name
);
2194 for (i
= 0; i
< ctx
->num_commit_graphs_after
; i
++)
2195 fprintf(get_lock_file_fp(&lk
), "%s\n", ctx
->commit_graph_hash_after
[i
]);
2198 error(_("failed to rename temporary commit-graph file"));
2203 commit_lock_file(&lk
);
2208 static void split_graph_merge_strategy(struct write_commit_graph_context
*ctx
)
2210 struct commit_graph
*g
;
2211 uint32_t num_commits
;
2212 enum commit_graph_split_flags flags
= COMMIT_GRAPH_SPLIT_UNSPECIFIED
;
2215 int max_commits
= 0;
2219 max_commits
= ctx
->opts
->max_commits
;
2221 if (ctx
->opts
->size_multiple
)
2222 size_mult
= ctx
->opts
->size_multiple
;
2224 flags
= ctx
->opts
->split_flags
;
2227 g
= ctx
->r
->objects
->commit_graph
;
2228 num_commits
= ctx
->commits
.nr
;
2229 if (flags
== COMMIT_GRAPH_SPLIT_REPLACE
)
2230 ctx
->num_commit_graphs_after
= 1;
2232 ctx
->num_commit_graphs_after
= ctx
->num_commit_graphs_before
+ 1;
2234 if (flags
!= COMMIT_GRAPH_SPLIT_MERGE_PROHIBITED
&&
2235 flags
!= COMMIT_GRAPH_SPLIT_REPLACE
) {
2236 while (g
&& (g
->num_commits
<= st_mult(size_mult
, num_commits
) ||
2237 (max_commits
&& num_commits
> max_commits
))) {
2238 if (g
->odb
!= ctx
->odb
)
2241 if (unsigned_add_overflows(num_commits
, g
->num_commits
))
2242 die(_("cannot merge graphs with %"PRIuMAX
", "
2243 "%"PRIuMAX
" commits"),
2244 (uintmax_t)num_commits
,
2245 (uintmax_t)g
->num_commits
);
2246 num_commits
+= g
->num_commits
;
2249 ctx
->num_commit_graphs_after
--;
2253 if (flags
!= COMMIT_GRAPH_SPLIT_REPLACE
)
2254 ctx
->new_base_graph
= g
;
2255 else if (ctx
->num_commit_graphs_after
!= 1)
2256 BUG("split_graph_merge_strategy: num_commit_graphs_after "
2257 "should be 1 with --split=replace");
2259 if (ctx
->num_commit_graphs_after
== 2) {
2260 char *old_graph_name
= get_commit_graph_filename(g
->odb
);
2262 if (!strcmp(g
->filename
, old_graph_name
) &&
2263 g
->odb
!= ctx
->odb
) {
2264 ctx
->num_commit_graphs_after
= 1;
2265 ctx
->new_base_graph
= NULL
;
2268 free(old_graph_name
);
2271 CALLOC_ARRAY(ctx
->commit_graph_filenames_after
, ctx
->num_commit_graphs_after
);
2272 CALLOC_ARRAY(ctx
->commit_graph_hash_after
, ctx
->num_commit_graphs_after
);
2274 for (i
= 0; i
< ctx
->num_commit_graphs_after
&&
2275 i
< ctx
->num_commit_graphs_before
; i
++)
2276 ctx
->commit_graph_filenames_after
[i
] = xstrdup(ctx
->commit_graph_filenames_before
[i
]);
2278 i
= ctx
->num_commit_graphs_before
- 1;
2279 g
= ctx
->r
->objects
->commit_graph
;
2282 if (i
< ctx
->num_commit_graphs_after
)
2283 ctx
->commit_graph_hash_after
[i
] = xstrdup(oid_to_hex(&g
->oid
));
2286 * If the topmost remaining layer has generation data chunk, the
2287 * resultant layer also has generation data chunk.
2289 if (i
== ctx
->num_commit_graphs_after
- 2)
2290 ctx
->write_generation_data
= !!g
->chunk_generation_data
;
2297 static void merge_commit_graph(struct write_commit_graph_context
*ctx
,
2298 struct commit_graph
*g
)
2301 uint32_t offset
= g
->num_commits_in_base
;
2303 if (unsigned_add_overflows(ctx
->commits
.nr
, g
->num_commits
))
2304 die(_("cannot merge graph %s, too many commits: %"PRIuMAX
),
2305 oid_to_hex(&g
->oid
),
2306 (uintmax_t)st_add(ctx
->commits
.nr
, g
->num_commits
));
2308 ALLOC_GROW(ctx
->commits
.list
, ctx
->commits
.nr
+ g
->num_commits
, ctx
->commits
.alloc
);
2310 for (i
= 0; i
< g
->num_commits
; i
++) {
2311 struct object_id oid
;
2312 struct commit
*result
;
2314 display_progress(ctx
->progress
, i
+ 1);
2316 load_oid_from_graph(g
, i
+ offset
, &oid
);
2318 /* only add commits if they still exist in the repo */
2319 result
= lookup_commit_reference_gently(ctx
->r
, &oid
, 1);
2322 ctx
->commits
.list
[ctx
->commits
.nr
] = result
;
2328 static int commit_compare(const void *_a
, const void *_b
)
2330 const struct commit
*a
= *(const struct commit
**)_a
;
2331 const struct commit
*b
= *(const struct commit
**)_b
;
2332 return oidcmp(&a
->object
.oid
, &b
->object
.oid
);
2335 static void sort_and_scan_merged_commits(struct write_commit_graph_context
*ctx
)
2337 uint32_t i
, dedup_i
= 0;
2339 if (ctx
->report_progress
)
2340 ctx
->progress
= start_delayed_progress(
2341 _("Scanning merged commits"),
2344 QSORT(ctx
->commits
.list
, ctx
->commits
.nr
, commit_compare
);
2346 ctx
->num_extra_edges
= 0;
2347 for (i
= 0; i
< ctx
->commits
.nr
; i
++) {
2348 display_progress(ctx
->progress
, i
+ 1);
2350 if (i
&& oideq(&ctx
->commits
.list
[i
- 1]->object
.oid
,
2351 &ctx
->commits
.list
[i
]->object
.oid
)) {
2353 * Silently ignore duplicates. These were likely
2354 * created due to a commit appearing in multiple
2355 * layers of the chain, which is unexpected but
2356 * not invalid. We should make sure there is a
2357 * unique copy in the new layer.
2360 unsigned int num_parents
;
2362 ctx
->commits
.list
[dedup_i
] = ctx
->commits
.list
[i
];
2365 num_parents
= commit_list_count(ctx
->commits
.list
[i
]->parents
);
2366 if (num_parents
> 2)
2367 ctx
->num_extra_edges
+= num_parents
- 1;
2371 ctx
->commits
.nr
= dedup_i
;
2373 stop_progress(&ctx
->progress
);
2376 static void merge_commit_graphs(struct write_commit_graph_context
*ctx
)
2378 struct commit_graph
*g
= ctx
->r
->objects
->commit_graph
;
2379 uint32_t current_graph_number
= ctx
->num_commit_graphs_before
;
2381 while (g
&& current_graph_number
>= ctx
->num_commit_graphs_after
) {
2382 current_graph_number
--;
2384 if (ctx
->report_progress
)
2385 ctx
->progress
= start_delayed_progress(_("Merging commit-graph"), 0);
2387 merge_commit_graph(ctx
, g
);
2388 stop_progress(&ctx
->progress
);
2394 ctx
->new_base_graph
= g
;
2395 ctx
->new_num_commits_in_base
= g
->num_commits
+ g
->num_commits_in_base
;
2398 if (ctx
->new_base_graph
)
2399 ctx
->base_graph_name
= xstrdup(ctx
->new_base_graph
->filename
);
2401 sort_and_scan_merged_commits(ctx
);
2404 static void mark_commit_graphs(struct write_commit_graph_context
*ctx
)
2407 time_t now
= time(NULL
);
2409 for (i
= ctx
->num_commit_graphs_after
- 1; i
< ctx
->num_commit_graphs_before
; i
++) {
2411 struct utimbuf updated_time
;
2413 if (stat(ctx
->commit_graph_filenames_before
[i
], &st
) < 0)
2416 updated_time
.actime
= st
.st_atime
;
2417 updated_time
.modtime
= now
;
2418 utime(ctx
->commit_graph_filenames_before
[i
], &updated_time
);
2422 static void expire_commit_graphs(struct write_commit_graph_context
*ctx
)
2424 struct strbuf path
= STRBUF_INIT
;
2428 timestamp_t expire_time
= time(NULL
);
2430 if (ctx
->opts
&& ctx
->opts
->expire_time
)
2431 expire_time
= ctx
->opts
->expire_time
;
2433 char *chain_file_name
= get_commit_graph_chain_filename(ctx
->odb
);
2434 unlink(chain_file_name
);
2435 free(chain_file_name
);
2436 ctx
->num_commit_graphs_after
= 0;
2439 strbuf_addstr(&path
, ctx
->odb
->path
);
2440 strbuf_addstr(&path
, "/info/commit-graphs");
2441 dir
= opendir(path
.buf
);
2446 strbuf_addch(&path
, '/');
2447 dirnamelen
= path
.len
;
2448 while ((de
= readdir(dir
)) != NULL
) {
2450 uint32_t i
, found
= 0;
2452 strbuf_setlen(&path
, dirnamelen
);
2453 strbuf_addstr(&path
, de
->d_name
);
2455 if (stat(path
.buf
, &st
) < 0)
2458 if (st
.st_mtime
> expire_time
)
2460 if (path
.len
< 6 || strcmp(path
.buf
+ path
.len
- 6, ".graph"))
2463 for (i
= 0; i
< ctx
->num_commit_graphs_after
; i
++) {
2464 if (!strcmp(ctx
->commit_graph_filenames_after
[i
],
2478 strbuf_release(&path
);
2481 int write_commit_graph(struct object_directory
*odb
,
2482 const struct string_list
*const pack_indexes
,
2483 struct oidset
*commits
,
2484 enum commit_graph_write_flags flags
,
2485 const struct commit_graph_opts
*opts
)
2487 struct repository
*r
= the_repository
;
2488 struct write_commit_graph_context
*ctx
;
2492 struct bloom_filter_settings bloom_settings
= DEFAULT_BLOOM_FILTER_SETTINGS
;
2493 struct topo_level_slab topo_levels
;
2495 prepare_repo_settings(r
);
2496 if (!r
->settings
.core_commit_graph
) {
2497 warning(_("attempting to write a commit-graph, but 'core.commitGraph' is disabled"));
2500 if (!commit_graph_compatible(r
))
2503 CALLOC_ARRAY(ctx
, 1);
2506 ctx
->append
= flags
& COMMIT_GRAPH_WRITE_APPEND
? 1 : 0;
2507 ctx
->report_progress
= flags
& COMMIT_GRAPH_WRITE_PROGRESS
? 1 : 0;
2508 ctx
->split
= flags
& COMMIT_GRAPH_WRITE_SPLIT
? 1 : 0;
2510 ctx
->total_bloom_filter_data_size
= 0;
2511 ctx
->write_generation_data
= (get_configured_generation_version(r
) == 2);
2512 ctx
->num_generation_data_overflows
= 0;
2514 bloom_settings
.bits_per_entry
= git_env_ulong("GIT_TEST_BLOOM_SETTINGS_BITS_PER_ENTRY",
2515 bloom_settings
.bits_per_entry
);
2516 bloom_settings
.num_hashes
= git_env_ulong("GIT_TEST_BLOOM_SETTINGS_NUM_HASHES",
2517 bloom_settings
.num_hashes
);
2518 bloom_settings
.max_changed_paths
= git_env_ulong("GIT_TEST_BLOOM_SETTINGS_MAX_CHANGED_PATHS",
2519 bloom_settings
.max_changed_paths
);
2520 ctx
->bloom_settings
= &bloom_settings
;
2522 init_topo_level_slab(&topo_levels
);
2523 ctx
->topo_levels
= &topo_levels
;
2525 prepare_commit_graph(ctx
->r
);
2526 if (ctx
->r
->objects
->commit_graph
) {
2527 struct commit_graph
*g
= ctx
->r
->objects
->commit_graph
;
2530 g
->topo_levels
= &topo_levels
;
2535 if (flags
& COMMIT_GRAPH_WRITE_BLOOM_FILTERS
)
2536 ctx
->changed_paths
= 1;
2537 if (!(flags
& COMMIT_GRAPH_NO_WRITE_BLOOM_FILTERS
)) {
2538 struct commit_graph
*g
;
2540 g
= ctx
->r
->objects
->commit_graph
;
2542 /* We have changed-paths already. Keep them in the next graph */
2543 if (g
&& g
->chunk_bloom_data
) {
2544 ctx
->changed_paths
= 1;
2545 ctx
->bloom_settings
= g
->bloom_filter_settings
;
2550 struct commit_graph
*g
= ctx
->r
->objects
->commit_graph
;
2553 ctx
->num_commit_graphs_before
++;
2557 if (ctx
->num_commit_graphs_before
) {
2558 ALLOC_ARRAY(ctx
->commit_graph_filenames_before
, ctx
->num_commit_graphs_before
);
2559 i
= ctx
->num_commit_graphs_before
;
2560 g
= ctx
->r
->objects
->commit_graph
;
2563 ctx
->commit_graph_filenames_before
[--i
] = xstrdup(g
->filename
);
2569 replace
= ctx
->opts
->split_flags
& COMMIT_GRAPH_SPLIT_REPLACE
;
2572 ctx
->approx_nr_objects
= repo_approximate_object_count(the_repository
);
2574 if (ctx
->append
&& ctx
->r
->objects
->commit_graph
) {
2575 struct commit_graph
*g
= ctx
->r
->objects
->commit_graph
;
2576 for (i
= 0; i
< g
->num_commits
; i
++) {
2577 struct object_id oid
;
2578 oidread(&oid
, g
->chunk_oid_lookup
+ st_mult(g
->hash_len
, i
));
2579 oid_array_append(&ctx
->oids
, &oid
);
2584 ctx
->order_by_pack
= 1;
2585 if ((res
= fill_oids_from_packs(ctx
, pack_indexes
)))
2590 if ((res
= fill_oids_from_commits(ctx
, commits
)))
2594 if (!pack_indexes
&& !commits
) {
2595 ctx
->order_by_pack
= 1;
2596 fill_oids_from_all_packs(ctx
);
2599 close_reachable(ctx
);
2601 copy_oids_to_commits(ctx
);
2603 if (ctx
->commits
.nr
>= GRAPH_EDGE_LAST_MASK
) {
2604 error(_("too many commits to write graph"));
2609 if (!ctx
->commits
.nr
&& !replace
)
2613 split_graph_merge_strategy(ctx
);
2616 merge_commit_graphs(ctx
);
2618 ctx
->num_commit_graphs_after
= 1;
2620 ctx
->trust_generation_numbers
= validate_mixed_generation_chain(ctx
->r
->objects
->commit_graph
);
2622 compute_topological_levels(ctx
);
2623 if (ctx
->write_generation_data
)
2624 compute_generation_numbers(ctx
);
2626 if (ctx
->changed_paths
)
2627 compute_bloom_filters(ctx
);
2629 res
= write_commit_graph_file(ctx
);
2632 mark_commit_graphs(ctx
);
2634 expire_commit_graphs(ctx
);
2637 free(ctx
->graph_name
);
2638 free(ctx
->base_graph_name
);
2639 free(ctx
->commits
.list
);
2640 oid_array_clear(&ctx
->oids
);
2641 clear_topo_level_slab(&topo_levels
);
2643 if (ctx
->commit_graph_filenames_after
) {
2644 for (i
= 0; i
< ctx
->num_commit_graphs_after
; i
++) {
2645 free(ctx
->commit_graph_filenames_after
[i
]);
2646 free(ctx
->commit_graph_hash_after
[i
]);
2649 for (i
= 0; i
< ctx
->num_commit_graphs_before
; i
++)
2650 free(ctx
->commit_graph_filenames_before
[i
]);
2652 free(ctx
->commit_graph_filenames_after
);
2653 free(ctx
->commit_graph_filenames_before
);
2654 free(ctx
->commit_graph_hash_after
);
2662 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
2663 static int verify_commit_graph_error
;
2665 __attribute__((format (printf
, 1, 2)))
2666 static void graph_report(const char *fmt
, ...)
2670 verify_commit_graph_error
= 1;
2672 vfprintf(stderr
, fmt
, ap
);
2673 fprintf(stderr
, "\n");
2677 static int commit_graph_checksum_valid(struct commit_graph
*g
)
2679 return hashfile_checksum_valid(g
->data
, g
->data_len
);
2682 static int verify_one_commit_graph(struct repository
*r
,
2683 struct commit_graph
*g
,
2684 struct progress
*progress
,
2687 uint32_t i
, cur_fanout_pos
= 0;
2688 struct object_id prev_oid
, cur_oid
;
2689 struct commit
*seen_gen_zero
= NULL
;
2690 struct commit
*seen_gen_non_zero
= NULL
;
2692 verify_commit_graph_error
= verify_commit_graph_lite(g
);
2693 if (verify_commit_graph_error
)
2694 return verify_commit_graph_error
;
2696 if (!commit_graph_checksum_valid(g
)) {
2697 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
2698 verify_commit_graph_error
= VERIFY_COMMIT_GRAPH_ERROR_HASH
;
2701 for (i
= 0; i
< g
->num_commits
; i
++) {
2702 struct commit
*graph_commit
;
2704 oidread(&cur_oid
, g
->chunk_oid_lookup
+ st_mult(g
->hash_len
, i
));
2706 if (i
&& oidcmp(&prev_oid
, &cur_oid
) >= 0)
2707 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
2708 oid_to_hex(&prev_oid
),
2709 oid_to_hex(&cur_oid
));
2711 oidcpy(&prev_oid
, &cur_oid
);
2713 while (cur_oid
.hash
[0] > cur_fanout_pos
) {
2714 uint32_t fanout_value
= get_be32(g
->chunk_oid_fanout
+ cur_fanout_pos
);
2716 if (i
!= fanout_value
)
2717 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2718 cur_fanout_pos
, fanout_value
, i
);
2722 graph_commit
= lookup_commit(r
, &cur_oid
);
2723 if (!parse_commit_in_graph_one(r
, g
, graph_commit
))
2724 graph_report(_("failed to parse commit %s from commit-graph"),
2725 oid_to_hex(&cur_oid
));
2728 while (cur_fanout_pos
< 256) {
2729 uint32_t fanout_value
= get_be32(g
->chunk_oid_fanout
+ cur_fanout_pos
);
2731 if (g
->num_commits
!= fanout_value
)
2732 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2733 cur_fanout_pos
, fanout_value
, i
);
2738 if (verify_commit_graph_error
& ~VERIFY_COMMIT_GRAPH_ERROR_HASH
)
2739 return verify_commit_graph_error
;
2741 for (i
= 0; i
< g
->num_commits
; i
++) {
2742 struct commit
*graph_commit
, *odb_commit
;
2743 struct commit_list
*graph_parents
, *odb_parents
;
2744 timestamp_t max_generation
= 0;
2745 timestamp_t generation
;
2747 display_progress(progress
, ++(*seen
));
2748 oidread(&cur_oid
, g
->chunk_oid_lookup
+ st_mult(g
->hash_len
, i
));
2750 graph_commit
= lookup_commit(r
, &cur_oid
);
2751 odb_commit
= (struct commit
*)create_object(r
, &cur_oid
, alloc_commit_node(r
));
2752 if (repo_parse_commit_internal(r
, odb_commit
, 0, 0)) {
2753 graph_report(_("failed to parse commit %s from object database for commit-graph"),
2754 oid_to_hex(&cur_oid
));
2758 if (!oideq(&get_commit_tree_in_graph_one(r
, g
, graph_commit
)->object
.oid
,
2759 get_commit_tree_oid(odb_commit
)))
2760 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2761 oid_to_hex(&cur_oid
),
2762 oid_to_hex(get_commit_tree_oid(graph_commit
)),
2763 oid_to_hex(get_commit_tree_oid(odb_commit
)));
2765 graph_parents
= graph_commit
->parents
;
2766 odb_parents
= odb_commit
->parents
;
2768 while (graph_parents
) {
2770 graph_report(_("commit-graph parent list for commit %s is too long"),
2771 oid_to_hex(&cur_oid
));
2775 /* parse parent in case it is in a base graph */
2776 parse_commit_in_graph_one(r
, g
, graph_parents
->item
);
2778 if (!oideq(&graph_parents
->item
->object
.oid
, &odb_parents
->item
->object
.oid
))
2779 graph_report(_("commit-graph parent for %s is %s != %s"),
2780 oid_to_hex(&cur_oid
),
2781 oid_to_hex(&graph_parents
->item
->object
.oid
),
2782 oid_to_hex(&odb_parents
->item
->object
.oid
));
2784 generation
= commit_graph_generation_from_graph(graph_parents
->item
);
2785 if (generation
> max_generation
)
2786 max_generation
= generation
;
2788 graph_parents
= graph_parents
->next
;
2789 odb_parents
= odb_parents
->next
;
2793 graph_report(_("commit-graph parent list for commit %s terminates early"),
2794 oid_to_hex(&cur_oid
));
2796 if (commit_graph_generation_from_graph(graph_commit
))
2797 seen_gen_non_zero
= graph_commit
;
2799 seen_gen_zero
= graph_commit
;
2805 * If we are using topological level and one of our parents has
2806 * generation GENERATION_NUMBER_V1_MAX, then our generation is
2807 * also GENERATION_NUMBER_V1_MAX. Decrement to avoid extra logic
2808 * in the following condition.
2810 if (!g
->read_generation_data
&& max_generation
== GENERATION_NUMBER_V1_MAX
)
2813 generation
= commit_graph_generation(graph_commit
);
2814 if (generation
< max_generation
+ 1)
2815 graph_report(_("commit-graph generation for commit %s is %"PRItime
" < %"PRItime
),
2816 oid_to_hex(&cur_oid
),
2818 max_generation
+ 1);
2820 if (graph_commit
->date
!= odb_commit
->date
)
2821 graph_report(_("commit date for commit %s in commit-graph is %"PRItime
" != %"PRItime
),
2822 oid_to_hex(&cur_oid
),
2827 if (seen_gen_zero
&& seen_gen_non_zero
)
2828 graph_report(_("commit-graph has both zero and non-zero "
2829 "generations (e.g., commits '%s' and '%s')"),
2830 oid_to_hex(&seen_gen_zero
->object
.oid
),
2831 oid_to_hex(&seen_gen_non_zero
->object
.oid
));
2833 return verify_commit_graph_error
;
2836 int verify_commit_graph(struct repository
*r
, struct commit_graph
*g
, int flags
)
2838 struct progress
*progress
= NULL
;
2839 int local_error
= 0;
2843 graph_report("no commit-graph file loaded");
2847 if (flags
& COMMIT_GRAPH_WRITE_PROGRESS
) {
2848 uint64_t total
= g
->num_commits
;
2849 if (!(flags
& COMMIT_GRAPH_VERIFY_SHALLOW
))
2850 total
+= g
->num_commits_in_base
;
2852 progress
= start_progress(_("Verifying commits in commit graph"),
2856 for (; g
; g
= g
->base_graph
) {
2857 local_error
|= verify_one_commit_graph(r
, g
, progress
, &seen
);
2858 if (flags
& COMMIT_GRAPH_VERIFY_SHALLOW
)
2862 stop_progress(&progress
);
2867 void free_commit_graph(struct commit_graph
*g
)
2870 struct commit_graph
*next
= g
->base_graph
;
2873 munmap((void *)g
->data
, g
->data_len
);
2875 free(g
->bloom_filter_settings
);
2882 void disable_commit_graph(struct repository
*r
)
2884 r
->commit_graph_disabled
= 1;