?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
commit-graph: simplify write_commit_graph_file() #1
[git/debian.git] / commit-graph.c
blob79cddabcd120732c404e473c1fbe512c36b89c62
1 #include "git-compat-util.h"
2 #include "config.h"
3 #include "lockfile.h"
4 #include "pack.h"
5 #include "packfile.h"
6 #include "commit.h"
7 #include "object.h"
8 #include "refs.h"
9 #include "revision.h"
10 #include "sha1-lookup.h"
11 #include "commit-graph.h"
12 #include "object-store.h"
13 #include "alloc.h"
14 #include "hashmap.h"
15 #include "replace-object.h"
16 #include "progress.h"
17 #include "bloom.h"
18 #include "commit-slab.h"
19
20 void git_test_write_commit_graph_or_die(void)
21 {
22 int flags = 0;
23 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0))
24 return;
25
26 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS, 0))
27 flags = COMMIT_GRAPH_WRITE_BLOOM_FILTERS;
28
29 if (write_commit_graph_reachable(the_repository->objects->odb,
30 flags, NULL))
31 die("failed to write commit-graph under GIT_TEST_COMMIT_GRAPH");
32 }
33
34 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
35 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
36 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
37 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
38 #define GRAPH_CHUNKID_EXTRAEDGES 0x45444745 /* "EDGE" */
39 #define GRAPH_CHUNKID_BLOOMINDEXES 0x42494458 /* "BIDX" */
40 #define GRAPH_CHUNKID_BLOOMDATA 0x42444154 /* "BDAT" */
41 #define GRAPH_CHUNKID_BASE 0x42415345 /* "BASE" */
42 #define MAX_NUM_CHUNKS 7
43
44 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
45
46 #define GRAPH_VERSION_1 0x1
47 #define GRAPH_VERSION GRAPH_VERSION_1
48
49 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
50 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
51 #define GRAPH_PARENT_NONE 0x70000000
52
53 #define GRAPH_LAST_EDGE 0x80000000
54
55 #define GRAPH_HEADER_SIZE 8
56 #define GRAPH_FANOUT_SIZE (4 * 256)
57 #define GRAPH_CHUNKLOOKUP_WIDTH 12
58 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
59 + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
60
61 /* Remember to update object flag allocation in object.h */
62 #define REACHABLE (1u<<15)
63
64 /* Keep track of the order in which commits are added to our list. */
65 define_commit_slab(commit_pos, int);
66 static struct commit_pos commit_pos = COMMIT_SLAB_INIT(1, commit_pos);
67
68 static void set_commit_pos(struct repository *r, const struct object_id *oid)
69 {
70 static int32_t max_pos;
71 struct commit *commit = lookup_commit(r, oid);
72
73 if (!commit)
74 return; /* should never happen, but be lenient */
75
76 *commit_pos_at(&commit_pos, commit) = max_pos++;
77 }
78
79 static int commit_pos_cmp(const void *va, const void *vb)
80 {
81 const struct commit *a = *(const struct commit **)va;
82 const struct commit *b = *(const struct commit **)vb;
83 return commit_pos_at(&commit_pos, a) -
84 commit_pos_at(&commit_pos, b);
85 }
86
87 static int commit_gen_cmp(const void *va, const void *vb)
88 {
89 const struct commit *a = *(const struct commit **)va;
90 const struct commit *b = *(const struct commit **)vb;
91
92 /* lower generation commits first */
93 if (a->generation < b->generation)
94 return -1;
95 else if (a->generation > b->generation)
96 return 1;
97
98 /* use date as a heuristic when generations are equal */
99 if (a->date < b->date)
100 return -1;
101 else if (a->date > b->date)
102 return 1;
103 return 0;
104 }
105
106 char *get_commit_graph_filename(struct object_directory *obj_dir)
107 {
108 return xstrfmt("%s/info/commit-graph", obj_dir->path);
109 }
110
111 static char *get_split_graph_filename(struct object_directory *odb,
112 const char *oid_hex)
113 {
114 return xstrfmt("%s/info/commit-graphs/graph-%s.graph", odb->path,
115 oid_hex);
116 }
117
118 static char *get_chain_filename(struct object_directory *odb)
119 {
120 return xstrfmt("%s/info/commit-graphs/commit-graph-chain", odb->path);
121 }
122
123 static uint8_t oid_version(void)
124 {
125 return 1;
126 }
127
128 static struct commit_graph *alloc_commit_graph(void)
129 {
130 struct commit_graph *g = xcalloc(1, sizeof(*g));
131 g->graph_fd = -1;
132
133 return g;
134 }
135
136 extern int read_replace_refs;
137
138 static int commit_graph_compatible(struct repository *r)
139 {
140 if (!r->gitdir)
141 return 0;
142
143 if (read_replace_refs) {
144 prepare_replace_object(r);
145 if (hashmap_get_size(&r->objects->replace_map->map))
146 return 0;
147 }
148
149 prepare_commit_graft(r);
150 if (r->parsed_objects && r->parsed_objects->grafts_nr)
151 return 0;
152 if (is_repository_shallow(r))
153 return 0;
154
155 return 1;
156 }
157
158 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
159 {
160 *fd = git_open(graph_file);
161 if (*fd < 0)
162 return 0;
163 if (fstat(*fd, st)) {
164 close(*fd);
165 return 0;
166 }
167 return 1;
168 }
169
170 struct commit_graph *load_commit_graph_one_fd_st(int fd, struct stat *st,
171 struct object_directory *odb)
172 {
173 void *graph_map;
174 size_t graph_size;
175 struct commit_graph *ret;
176
177 graph_size = xsize_t(st->st_size);
178
179 if (graph_size < GRAPH_MIN_SIZE) {
180 close(fd);
181 error(_("commit-graph file is too small"));
182 return NULL;
183 }
184 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
185 ret = parse_commit_graph(graph_map, fd, graph_size);
186
187 if (ret)
188 ret->odb = odb;
189 else {
190 munmap(graph_map, graph_size);
191 close(fd);
192 }
193
194 return ret;
195 }
196
197 static int verify_commit_graph_lite(struct commit_graph *g)
198 {
199 /*
200 * Basic validation shared between parse_commit_graph()
201 * which'll be called every time the graph is used, and the
202 * much more expensive verify_commit_graph() used by
203 * "commit-graph verify".
204 *
205 * There should only be very basic checks here to ensure that
206 * we don't e.g. segfault in fill_commit_in_graph(), but
207 * because this is a very hot codepath nothing that e.g. loops
208 * over g->num_commits, or runs a checksum on the commit-graph
209 * itself.
210 */
211 if (!g->chunk_oid_fanout) {
212 error("commit-graph is missing the OID Fanout chunk");
213 return 1;
214 }
215 if (!g->chunk_oid_lookup) {
216 error("commit-graph is missing the OID Lookup chunk");
217 return 1;
218 }
219 if (!g->chunk_commit_data) {
220 error("commit-graph is missing the Commit Data chunk");
221 return 1;
222 }
223
224 return 0;
225 }
226
227 struct commit_graph *parse_commit_graph(void *graph_map, int fd,
228 size_t graph_size)
229 {
230 const unsigned char *data, *chunk_lookup;
231 uint32_t i;
232 struct commit_graph *graph;
233 uint64_t next_chunk_offset;
234 uint32_t graph_signature;
235 unsigned char graph_version, hash_version;
236
237 if (!graph_map)
238 return NULL;
239
240 if (graph_size < GRAPH_MIN_SIZE)
241 return NULL;
242
243 data = (const unsigned char *)graph_map;
244
245 graph_signature = get_be32(data);
246 if (graph_signature != GRAPH_SIGNATURE) {
247 error(_("commit-graph signature %X does not match signature %X"),
248 graph_signature, GRAPH_SIGNATURE);
249 return NULL;
250 }
251
252 graph_version = *(unsigned char*)(data + 4);
253 if (graph_version != GRAPH_VERSION) {
254 error(_("commit-graph version %X does not match version %X"),
255 graph_version, GRAPH_VERSION);
256 return NULL;
257 }
258
259 hash_version = *(unsigned char*)(data + 5);
260 if (hash_version != oid_version()) {
261 error(_("commit-graph hash version %X does not match version %X"),
262 hash_version, oid_version());
263 return NULL;
264 }
265
266 graph = alloc_commit_graph();
267
268 graph->hash_len = the_hash_algo->rawsz;
269 graph->num_chunks = *(unsigned char*)(data + 6);
270 graph->graph_fd = fd;
271 graph->data = graph_map;
272 graph->data_len = graph_size;
273
274 if (graph_size < GRAPH_HEADER_SIZE +
275 (graph->num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH +
276 GRAPH_FANOUT_SIZE + the_hash_algo->rawsz) {
277 error(_("commit-graph file is too small to hold %u chunks"),
278 graph->num_chunks);
279 free(graph);
280 return NULL;
281 }
282
283 chunk_lookup = data + 8;
284 next_chunk_offset = get_be64(chunk_lookup + 4);
285 for (i = 0; i < graph->num_chunks; i++) {
286 uint32_t chunk_id;
287 uint64_t chunk_offset = next_chunk_offset;
288 int chunk_repeated = 0;
289
290 chunk_id = get_be32(chunk_lookup + 0);
291
292 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
293 next_chunk_offset = get_be64(chunk_lookup + 4);
294
295 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
296 error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
297 (uint32_t)chunk_offset);
298 free(graph);
299 return NULL;
300 }
301
302 switch (chunk_id) {
303 case GRAPH_CHUNKID_OIDFANOUT:
304 if (graph->chunk_oid_fanout)
305 chunk_repeated = 1;
306 else
307 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
308 break;
309
310 case GRAPH_CHUNKID_OIDLOOKUP:
311 if (graph->chunk_oid_lookup)
312 chunk_repeated = 1;
313 else {
314 graph->chunk_oid_lookup = data + chunk_offset;
315 graph->num_commits = (next_chunk_offset - chunk_offset)
316 / graph->hash_len;
317 }
318 break;
319
320 case GRAPH_CHUNKID_DATA:
321 if (graph->chunk_commit_data)
322 chunk_repeated = 1;
323 else
324 graph->chunk_commit_data = data + chunk_offset;
325 break;
326
327 case GRAPH_CHUNKID_EXTRAEDGES:
328 if (graph->chunk_extra_edges)
329 chunk_repeated = 1;
330 else
331 graph->chunk_extra_edges = data + chunk_offset;
332 break;
333
334 case GRAPH_CHUNKID_BASE:
335 if (graph->chunk_base_graphs)
336 chunk_repeated = 1;
337 else
338 graph->chunk_base_graphs = data + chunk_offset;
339 break;
340
341 case GRAPH_CHUNKID_BLOOMINDEXES:
342 if (graph->chunk_bloom_indexes)
343 chunk_repeated = 1;
344 else
345 graph->chunk_bloom_indexes = data + chunk_offset;
346 break;
347
348 case GRAPH_CHUNKID_BLOOMDATA:
349 if (graph->chunk_bloom_data)
350 chunk_repeated = 1;
351 else {
352 uint32_t hash_version;
353 graph->chunk_bloom_data = data + chunk_offset;
354 hash_version = get_be32(data + chunk_offset);
355
356 if (hash_version != 1)
357 break;
358
359 graph->bloom_filter_settings = xmalloc(sizeof(struct bloom_filter_settings));
360 graph->bloom_filter_settings->hash_version = hash_version;
361 graph->bloom_filter_settings->num_hashes = get_be32(data + chunk_offset + 4);
362 graph->bloom_filter_settings->bits_per_entry = get_be32(data + chunk_offset + 8);
363 }
364 break;
365 }
366
367 if (chunk_repeated) {
368 error(_("commit-graph chunk id %08x appears multiple times"), chunk_id);
369 free(graph);
370 return NULL;
371 }
372 }
373
374 if (graph->chunk_bloom_indexes && graph->chunk_bloom_data) {
375 init_bloom_filters();
376 } else {
377 /* We need both the bloom chunks to exist together. Else ignore the data */
378 graph->chunk_bloom_indexes = NULL;
379 graph->chunk_bloom_data = NULL;
380 graph->bloom_filter_settings = NULL;
381 }
382
383 hashcpy(graph->oid.hash, graph->data + graph->data_len - graph->hash_len);
384
385 if (verify_commit_graph_lite(graph)) {
386 free(graph);
387 return NULL;
388 }
389
390 return graph;
391 }
392
393 static struct commit_graph *load_commit_graph_one(const char *graph_file,
394 struct object_directory *odb)
395 {
396
397 struct stat st;
398 int fd;
399 struct commit_graph *g;
400 int open_ok = open_commit_graph(graph_file, &fd, &st);
401
402 if (!open_ok)
403 return NULL;
404
405 g = load_commit_graph_one_fd_st(fd, &st, odb);
406
407 if (g)
408 g->filename = xstrdup(graph_file);
409
410 return g;
411 }
412
413 static struct commit_graph *load_commit_graph_v1(struct repository *r,
414 struct object_directory *odb)
415 {
416 char *graph_name = get_commit_graph_filename(odb);
417 struct commit_graph *g = load_commit_graph_one(graph_name, odb);
418 free(graph_name);
419
420 return g;
421 }
422
423 static int add_graph_to_chain(struct commit_graph *g,
424 struct commit_graph *chain,
425 struct object_id *oids,
426 int n)
427 {
428 struct commit_graph *cur_g = chain;
429
430 if (n && !g->chunk_base_graphs) {
431 warning(_("commit-graph has no base graphs chunk"));
432 return 0;
433 }
434
435 while (n) {
436 n--;
437
438 if (!cur_g ||
439 !oideq(&oids[n], &cur_g->oid) ||
440 !hasheq(oids[n].hash, g->chunk_base_graphs + g->hash_len * n)) {
441 warning(_("commit-graph chain does not match"));
442 return 0;
443 }
444
445 cur_g = cur_g->base_graph;
446 }
447
448 g->base_graph = chain;
449
450 if (chain)
451 g->num_commits_in_base = chain->num_commits + chain->num_commits_in_base;
452
453 return 1;
454 }
455
456 static struct commit_graph *load_commit_graph_chain(struct repository *r,
457 struct object_directory *odb)
458 {
459 struct commit_graph *graph_chain = NULL;
460 struct strbuf line = STRBUF_INIT;
461 struct stat st;
462 struct object_id *oids;
463 int i = 0, valid = 1, count;
464 char *chain_name = get_chain_filename(odb);
465 FILE *fp;
466 int stat_res;
467
468 fp = fopen(chain_name, "r");
469 stat_res = stat(chain_name, &st);
470 free(chain_name);
471
472 if (!fp ||
473 stat_res ||
474 st.st_size <= the_hash_algo->hexsz)
475 return NULL;
476
477 count = st.st_size / (the_hash_algo->hexsz + 1);
478 oids = xcalloc(count, sizeof(struct object_id));
479
480 prepare_alt_odb(r);
481
482 for (i = 0; i < count; i++) {
483 struct object_directory *odb;
484
485 if (strbuf_getline_lf(&line, fp) == EOF)
486 break;
487
488 if (get_oid_hex(line.buf, &oids[i])) {
489 warning(_("invalid commit-graph chain: line '%s' not a hash"),
490 line.buf);
491 valid = 0;
492 break;
493 }
494
495 valid = 0;
496 for (odb = r->objects->odb; odb; odb = odb->next) {
497 char *graph_name = get_split_graph_filename(odb, line.buf);
498 struct commit_graph *g = load_commit_graph_one(graph_name, odb);
499
500 free(graph_name);
501
502 if (g) {
503 if (add_graph_to_chain(g, graph_chain, oids, i)) {
504 graph_chain = g;
505 valid = 1;
506 }
507
508 break;
509 }
510 }
511
512 if (!valid) {
513 warning(_("unable to find all commit-graph files"));
514 break;
515 }
516 }
517
518 free(oids);
519 fclose(fp);
520 strbuf_release(&line);
521
522 return graph_chain;
523 }
524
525 struct commit_graph *read_commit_graph_one(struct repository *r,
526 struct object_directory *odb)
527 {
528 struct commit_graph *g = load_commit_graph_v1(r, odb);
529
530 if (!g)
531 g = load_commit_graph_chain(r, odb);
532
533 return g;
534 }
535
536 static void prepare_commit_graph_one(struct repository *r,
537 struct object_directory *odb)
538 {
539
540 if (r->objects->commit_graph)
541 return;
542
543 r->objects->commit_graph = read_commit_graph_one(r, odb);
544 }
545
546 /*
547 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
548 *
549 * On the first invocation, this function attempts to load the commit
550 * graph if the_repository is configured to have one.
551 */
552 static int prepare_commit_graph(struct repository *r)
553 {
554 struct object_directory *odb;
555
556 /*
557 * This must come before the "already attempted?" check below, because
558 * we want to disable even an already-loaded graph file.
559 */
560 if (r->commit_graph_disabled)
561 return 0;
562
563 if (r->objects->commit_graph_attempted)
564 return !!r->objects->commit_graph;
565 r->objects->commit_graph_attempted = 1;
566
567 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
568 die("dying as requested by the '%s' variable on commit-graph load!",
569 GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD);
570
571 prepare_repo_settings(r);
572
573 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
574 r->settings.core_commit_graph != 1)
575 /*
576 * This repository is not configured to use commit graphs, so
577 * do not load one. (But report commit_graph_attempted anyway
578 * so that commit graph loading is not attempted again for this
579 * repository.)
580 */
581 return 0;
582
583 if (!commit_graph_compatible(r))
584 return 0;
585
586 prepare_alt_odb(r);
587 for (odb = r->objects->odb;
588 !r->objects->commit_graph && odb;
589 odb = odb->next)
590 prepare_commit_graph_one(r, odb);
591 return !!r->objects->commit_graph;
592 }
593
594 int generation_numbers_enabled(struct repository *r)
595 {
596 uint32_t first_generation;
597 struct commit_graph *g;
598 if (!prepare_commit_graph(r))
599 return 0;
600
601 g = r->objects->commit_graph;
602
603 if (!g->num_commits)
604 return 0;
605
606 first_generation = get_be32(g->chunk_commit_data +
607 g->hash_len + 8) >> 2;
608
609 return !!first_generation;
610 }
611
612 static void close_commit_graph_one(struct commit_graph *g)
613 {
614 if (!g)
615 return;
616
617 close_commit_graph_one(g->base_graph);
618 free_commit_graph(g);
619 }
620
621 void close_commit_graph(struct raw_object_store *o)
622 {
623 close_commit_graph_one(o->commit_graph);
624 o->commit_graph = NULL;
625 }
626
627 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
628 {
629 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
630 g->chunk_oid_lookup, g->hash_len, pos);
631 }
632
633 static void load_oid_from_graph(struct commit_graph *g,
634 uint32_t pos,
635 struct object_id *oid)
636 {
637 uint32_t lex_index;
638
639 while (g && pos < g->num_commits_in_base)
640 g = g->base_graph;
641
642 if (!g)
643 BUG("NULL commit-graph");
644
645 if (pos >= g->num_commits + g->num_commits_in_base)
646 die(_("invalid commit position. commit-graph is likely corrupt"));
647
648 lex_index = pos - g->num_commits_in_base;
649
650 hashcpy(oid->hash, g->chunk_oid_lookup + g->hash_len * lex_index);
651 }
652
653 static struct commit_list **insert_parent_or_die(struct repository *r,
654 struct commit_graph *g,
655 uint32_t pos,
656 struct commit_list **pptr)
657 {
658 struct commit *c;
659 struct object_id oid;
660
661 if (pos >= g->num_commits + g->num_commits_in_base)
662 die("invalid parent position %"PRIu32, pos);
663
664 load_oid_from_graph(g, pos, &oid);
665 c = lookup_commit(r, &oid);
666 if (!c)
667 die(_("could not find commit %s"), oid_to_hex(&oid));
668 c->graph_pos = pos;
669 return &commit_list_insert(c, pptr)->next;
670 }
671
672 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
673 {
674 const unsigned char *commit_data;
675 uint32_t lex_index;
676
677 while (pos < g->num_commits_in_base)
678 g = g->base_graph;
679
680 lex_index = pos - g->num_commits_in_base;
681 commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * lex_index;
682 item->graph_pos = pos;
683 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
684 }
685
686 static inline void set_commit_tree(struct commit *c, struct tree *t)
687 {
688 c->maybe_tree = t;
689 }
690
691 static int fill_commit_in_graph(struct repository *r,
692 struct commit *item,
693 struct commit_graph *g, uint32_t pos)
694 {
695 uint32_t edge_value;
696 uint32_t *parent_data_ptr;
697 uint64_t date_low, date_high;
698 struct commit_list **pptr;
699 const unsigned char *commit_data;
700 uint32_t lex_index;
701
702 while (pos < g->num_commits_in_base)
703 g = g->base_graph;
704
705 if (pos >= g->num_commits + g->num_commits_in_base)
706 die(_("invalid commit position. commit-graph is likely corrupt"));
707
708 /*
709 * Store the "full" position, but then use the
710 * "local" position for the rest of the calculation.
711 */
712 item->graph_pos = pos;
713 lex_index = pos - g->num_commits_in_base;
714
715 commit_data = g->chunk_commit_data + (g->hash_len + 16) * lex_index;
716
717 item->object.parsed = 1;
718
719 set_commit_tree(item, NULL);
720
721 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
722 date_low = get_be32(commit_data + g->hash_len + 12);
723 item->date = (timestamp_t)((date_high << 32) | date_low);
724
725 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
726
727 pptr = &item->parents;
728
729 edge_value = get_be32(commit_data + g->hash_len);
730 if (edge_value == GRAPH_PARENT_NONE)
731 return 1;
732 pptr = insert_parent_or_die(r, g, edge_value, pptr);
733
734 edge_value = get_be32(commit_data + g->hash_len + 4);
735 if (edge_value == GRAPH_PARENT_NONE)
736 return 1;
737 if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
738 pptr = insert_parent_or_die(r, g, edge_value, pptr);
739 return 1;
740 }
741
742 parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
743 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
744 do {
745 edge_value = get_be32(parent_data_ptr);
746 pptr = insert_parent_or_die(r, g,
747 edge_value & GRAPH_EDGE_LAST_MASK,
748 pptr);
749 parent_data_ptr++;
750 } while (!(edge_value & GRAPH_LAST_EDGE));
751
752 return 1;
753 }
754
755 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
756 {
757 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
758 *pos = item->graph_pos;
759 return 1;
760 } else {
761 struct commit_graph *cur_g = g;
762 uint32_t lex_index;
763
764 while (cur_g && !bsearch_graph(cur_g, &(item->object.oid), &lex_index))
765 cur_g = cur_g->base_graph;
766
767 if (cur_g) {
768 *pos = lex_index + cur_g->num_commits_in_base;
769 return 1;
770 }
771
772 return 0;
773 }
774 }
775
776 static int parse_commit_in_graph_one(struct repository *r,
777 struct commit_graph *g,
778 struct commit *item)
779 {
780 uint32_t pos;
781
782 if (item->object.parsed)
783 return 1;
784
785 if (find_commit_in_graph(item, g, &pos))
786 return fill_commit_in_graph(r, item, g, pos);
787
788 return 0;
789 }
790
791 int parse_commit_in_graph(struct repository *r, struct commit *item)
792 {
793 if (!prepare_commit_graph(r))
794 return 0;
795 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
796 }
797
798 void load_commit_graph_info(struct repository *r, struct commit *item)
799 {
800 uint32_t pos;
801 if (!prepare_commit_graph(r))
802 return;
803 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
804 fill_commit_graph_info(item, r->objects->commit_graph, pos);
805 }
806
807 static struct tree *load_tree_for_commit(struct repository *r,
808 struct commit_graph *g,
809 struct commit *c)
810 {
811 struct object_id oid;
812 const unsigned char *commit_data;
813
814 while (c->graph_pos < g->num_commits_in_base)
815 g = g->base_graph;
816
817 commit_data = g->chunk_commit_data +
818 GRAPH_DATA_WIDTH * (c->graph_pos - g->num_commits_in_base);
819
820 hashcpy(oid.hash, commit_data);
821 set_commit_tree(c, lookup_tree(r, &oid));
822
823 return c->maybe_tree;
824 }
825
826 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
827 struct commit_graph *g,
828 const struct commit *c)
829 {
830 if (c->maybe_tree)
831 return c->maybe_tree;
832 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
833 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
834
835 return load_tree_for_commit(r, g, (struct commit *)c);
836 }
837
838 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
839 {
840 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
841 }
842
843 struct packed_commit_list {
844 struct commit **list;
845 int nr;
846 int alloc;
847 };
848
849 struct packed_oid_list {
850 struct object_id *list;
851 int nr;
852 int alloc;
853 };
854
855 struct write_commit_graph_context {
856 struct repository *r;
857 struct object_directory *odb;
858 char *graph_name;
859 struct packed_oid_list oids;
860 struct packed_commit_list commits;
861 int num_extra_edges;
862 unsigned long approx_nr_objects;
863 struct progress *progress;
864 int progress_done;
865 uint64_t progress_cnt;
866
867 char *base_graph_name;
868 int num_commit_graphs_before;
869 int num_commit_graphs_after;
870 char **commit_graph_filenames_before;
871 char **commit_graph_filenames_after;
872 char **commit_graph_hash_after;
873 uint32_t new_num_commits_in_base;
874 struct commit_graph *new_base_graph;
875
876 unsigned append:1,
877 report_progress:1,
878 split:1,
879 check_oids:1,
880 changed_paths:1,
881 order_by_pack:1;
882
883 const struct split_commit_graph_opts *split_opts;
884 size_t total_bloom_filter_data_size;
885 };
886
887 static void write_graph_chunk_fanout(struct hashfile *f,
888 struct write_commit_graph_context *ctx)
889 {
890 int i, count = 0;
891 struct commit **list = ctx->commits.list;
892
893 /*
894 * Write the first-level table (the list is sorted,
895 * but we use a 256-entry lookup to be able to avoid
896 * having to do eight extra binary search iterations).
897 */
898 for (i = 0; i < 256; i++) {
899 while (count < ctx->commits.nr) {
900 if ((*list)->object.oid.hash[0] != i)
901 break;
902 display_progress(ctx->progress, ++ctx->progress_cnt);
903 count++;
904 list++;
905 }
906
907 hashwrite_be32(f, count);
908 }
909 }
910
911 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
912 struct write_commit_graph_context *ctx)
913 {
914 struct commit **list = ctx->commits.list;
915 int count;
916 for (count = 0; count < ctx->commits.nr; count++, list++) {
917 display_progress(ctx->progress, ++ctx->progress_cnt);
918 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
919 }
920 }
921
922 static const unsigned char *commit_to_sha1(size_t index, void *table)
923 {
924 struct commit **commits = table;
925 return commits[index]->object.oid.hash;
926 }
927
928 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
929 struct write_commit_graph_context *ctx)
930 {
931 struct commit **list = ctx->commits.list;
932 struct commit **last = ctx->commits.list + ctx->commits.nr;
933 uint32_t num_extra_edges = 0;
934
935 while (list < last) {
936 struct commit_list *parent;
937 struct object_id *tree;
938 int edge_value;
939 uint32_t packedDate[2];
940 display_progress(ctx->progress, ++ctx->progress_cnt);
941
942 if (parse_commit_no_graph(*list))
943 die(_("unable to parse commit %s"),
944 oid_to_hex(&(*list)->object.oid));
945 tree = get_commit_tree_oid(*list);
946 hashwrite(f, tree->hash, hash_len);
947
948 parent = (*list)->parents;
949
950 if (!parent)
951 edge_value = GRAPH_PARENT_NONE;
952 else {
953 edge_value = sha1_pos(parent->item->object.oid.hash,
954 ctx->commits.list,
955 ctx->commits.nr,
956 commit_to_sha1);
957
958 if (edge_value >= 0)
959 edge_value += ctx->new_num_commits_in_base;
960 else {
961 uint32_t pos;
962 if (find_commit_in_graph(parent->item,
963 ctx->new_base_graph,
964 &pos))
965 edge_value = pos;
966 }
967
968 if (edge_value < 0)
969 BUG("missing parent %s for commit %s",
970 oid_to_hex(&parent->item->object.oid),
971 oid_to_hex(&(*list)->object.oid));
972 }
973
974 hashwrite_be32(f, edge_value);
975
976 if (parent)
977 parent = parent->next;
978
979 if (!parent)
980 edge_value = GRAPH_PARENT_NONE;
981 else if (parent->next)
982 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
983 else {
984 edge_value = sha1_pos(parent->item->object.oid.hash,
985 ctx->commits.list,
986 ctx->commits.nr,
987 commit_to_sha1);
988
989 if (edge_value >= 0)
990 edge_value += ctx->new_num_commits_in_base;
991 else {
992 uint32_t pos;
993 if (find_commit_in_graph(parent->item,
994 ctx->new_base_graph,
995 &pos))
996 edge_value = pos;
997 }
998
999 if (edge_value < 0)
1000 BUG("missing parent %s for commit %s",
1001 oid_to_hex(&parent->item->object.oid),
1002 oid_to_hex(&(*list)->object.oid));
1003 }
1004
1005 hashwrite_be32(f, edge_value);
1006
1007 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
1008 do {
1009 num_extra_edges++;
1010 parent = parent->next;
1011 } while (parent);
1012 }
1013
1014 if (sizeof((*list)->date) > 4)
1015 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
1016 else
1017 packedDate[0] = 0;
1018
1019 packedDate[0] |= htonl((*list)->generation << 2);
1020
1021 packedDate[1] = htonl((*list)->date);
1022 hashwrite(f, packedDate, 8);
1023
1024 list++;
1025 }
1026 }
1027
1028 static void write_graph_chunk_extra_edges(struct hashfile *f,
1029 struct write_commit_graph_context *ctx)
1030 {
1031 struct commit **list = ctx->commits.list;
1032 struct commit **last = ctx->commits.list + ctx->commits.nr;
1033 struct commit_list *parent;
1034
1035 while (list < last) {
1036 int num_parents = 0;
1037
1038 display_progress(ctx->progress, ++ctx->progress_cnt);
1039
1040 for (parent = (*list)->parents; num_parents < 3 && parent;
1041 parent = parent->next)
1042 num_parents++;
1043
1044 if (num_parents <= 2) {
1045 list++;
1046 continue;
1047 }
1048
1049 /* Since num_parents > 2, this initializer is safe. */
1050 for (parent = (*list)->parents->next; parent; parent = parent->next) {
1051 int edge_value = sha1_pos(parent->item->object.oid.hash,
1052 ctx->commits.list,
1053 ctx->commits.nr,
1054 commit_to_sha1);
1055
1056 if (edge_value >= 0)
1057 edge_value += ctx->new_num_commits_in_base;
1058 else {
1059 uint32_t pos;
1060 if (find_commit_in_graph(parent->item,
1061 ctx->new_base_graph,
1062 &pos))
1063 edge_value = pos;
1064 }
1065
1066 if (edge_value < 0)
1067 BUG("missing parent %s for commit %s",
1068 oid_to_hex(&parent->item->object.oid),
1069 oid_to_hex(&(*list)->object.oid));
1070 else if (!parent->next)
1071 edge_value |= GRAPH_LAST_EDGE;
1072
1073 hashwrite_be32(f, edge_value);
1074 }
1075
1076 list++;
1077 }
1078 }
1079
1080 static void write_graph_chunk_bloom_indexes(struct hashfile *f,
1081 struct write_commit_graph_context *ctx)
1082 {
1083 struct commit **list = ctx->commits.list;
1084 struct commit **last = ctx->commits.list + ctx->commits.nr;
1085 uint32_t cur_pos = 0;
1086 struct progress *progress = NULL;
1087 int i = 0;
1088
1089 if (ctx->report_progress)
1090 progress = start_delayed_progress(
1091 _("Writing changed paths Bloom filters index"),
1092 ctx->commits.nr);
1093
1094 while (list < last) {
1095 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list, 0);
1096 cur_pos += filter->len;
1097 display_progress(progress, ++i);
1098 hashwrite_be32(f, cur_pos);
1099 list++;
1100 }
1101
1102 stop_progress(&progress);
1103 }
1104
1105 static void write_graph_chunk_bloom_data(struct hashfile *f,
1106 struct write_commit_graph_context *ctx,
1107 const struct bloom_filter_settings *settings)
1108 {
1109 struct commit **list = ctx->commits.list;
1110 struct commit **last = ctx->commits.list + ctx->commits.nr;
1111 struct progress *progress = NULL;
1112 int i = 0;
1113
1114 if (ctx->report_progress)
1115 progress = start_delayed_progress(
1116 _("Writing changed paths Bloom filters data"),
1117 ctx->commits.nr);
1118
1119 hashwrite_be32(f, settings->hash_version);
1120 hashwrite_be32(f, settings->num_hashes);
1121 hashwrite_be32(f, settings->bits_per_entry);
1122
1123 while (list < last) {
1124 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list, 0);
1125 display_progress(progress, ++i);
1126 hashwrite(f, filter->data, filter->len * sizeof(unsigned char));
1127 list++;
1128 }
1129
1130 stop_progress(&progress);
1131 }
1132
1133 static int oid_compare(const void *_a, const void *_b)
1134 {
1135 const struct object_id *a = (const struct object_id *)_a;
1136 const struct object_id *b = (const struct object_id *)_b;
1137 return oidcmp(a, b);
1138 }
1139
1140 static int add_packed_commits(const struct object_id *oid,
1141 struct packed_git *pack,
1142 uint32_t pos,
1143 void *data)
1144 {
1145 struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
1146 enum object_type type;
1147 off_t offset = nth_packed_object_offset(pack, pos);
1148 struct object_info oi = OBJECT_INFO_INIT;
1149
1150 if (ctx->progress)
1151 display_progress(ctx->progress, ++ctx->progress_done);
1152
1153 oi.typep = &type;
1154 if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
1155 die(_("unable to get type of object %s"), oid_to_hex(oid));
1156
1157 if (type != OBJ_COMMIT)
1158 return 0;
1159
1160 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1161 oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
1162 ctx->oids.nr++;
1163
1164 set_commit_pos(ctx->r, oid);
1165
1166 return 0;
1167 }
1168
1169 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
1170 {
1171 struct commit_list *parent;
1172 for (parent = commit->parents; parent; parent = parent->next) {
1173 if (!(parent->item->object.flags & REACHABLE)) {
1174 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1175 oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
1176 ctx->oids.nr++;
1177 parent->item->object.flags |= REACHABLE;
1178 }
1179 }
1180 }
1181
1182 static void close_reachable(struct write_commit_graph_context *ctx)
1183 {
1184 int i;
1185 struct commit *commit;
1186
1187 if (ctx->report_progress)
1188 ctx->progress = start_delayed_progress(
1189 _("Loading known commits in commit graph"),
1190 ctx->oids.nr);
1191 for (i = 0; i < ctx->oids.nr; i++) {
1192 display_progress(ctx->progress, i + 1);
1193 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1194 if (commit)
1195 commit->object.flags |= REACHABLE;
1196 }
1197 stop_progress(&ctx->progress);
1198
1199 /*
1200 * As this loop runs, ctx->oids.nr may grow, but not more
1201 * than the number of missing commits in the reachable
1202 * closure.
1203 */
1204 if (ctx->report_progress)
1205 ctx->progress = start_delayed_progress(
1206 _("Expanding reachable commits in commit graph"),
1207 0);
1208 for (i = 0; i < ctx->oids.nr; i++) {
1209 display_progress(ctx->progress, i + 1);
1210 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1211
1212 if (!commit)
1213 continue;
1214 if (ctx->split) {
1215 if (!parse_commit(commit) &&
1216 commit->graph_pos == COMMIT_NOT_FROM_GRAPH)
1217 add_missing_parents(ctx, commit);
1218 } else if (!parse_commit_no_graph(commit))
1219 add_missing_parents(ctx, commit);
1220 }
1221 stop_progress(&ctx->progress);
1222
1223 if (ctx->report_progress)
1224 ctx->progress = start_delayed_progress(
1225 _("Clearing commit marks in commit graph"),
1226 ctx->oids.nr);
1227 for (i = 0; i < ctx->oids.nr; i++) {
1228 display_progress(ctx->progress, i + 1);
1229 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1230
1231 if (commit)
1232 commit->object.flags &= ~REACHABLE;
1233 }
1234 stop_progress(&ctx->progress);
1235 }
1236
1237 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1238 {
1239 int i;
1240 struct commit_list *list = NULL;
1241
1242 if (ctx->report_progress)
1243 ctx->progress = start_delayed_progress(
1244 _("Computing commit graph generation numbers"),
1245 ctx->commits.nr);
1246 for (i = 0; i < ctx->commits.nr; i++) {
1247 display_progress(ctx->progress, i + 1);
1248 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
1249 ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
1250 continue;
1251
1252 commit_list_insert(ctx->commits.list[i], &list);
1253 while (list) {
1254 struct commit *current = list->item;
1255 struct commit_list *parent;
1256 int all_parents_computed = 1;
1257 uint32_t max_generation = 0;
1258
1259 for (parent = current->parents; parent; parent = parent->next) {
1260 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
1261 parent->item->generation == GENERATION_NUMBER_ZERO) {
1262 all_parents_computed = 0;
1263 commit_list_insert(parent->item, &list);
1264 break;
1265 } else if (parent->item->generation > max_generation) {
1266 max_generation = parent->item->generation;
1267 }
1268 }
1269
1270 if (all_parents_computed) {
1271 current->generation = max_generation + 1;
1272 pop_commit(&list);
1273
1274 if (current->generation > GENERATION_NUMBER_MAX)
1275 current->generation = GENERATION_NUMBER_MAX;
1276 }
1277 }
1278 }
1279 stop_progress(&ctx->progress);
1280 }
1281
1282 static void compute_bloom_filters(struct write_commit_graph_context *ctx)
1283 {
1284 int i;
1285 struct progress *progress = NULL;
1286 struct commit **sorted_commits;
1287
1288 init_bloom_filters();
1289
1290 if (ctx->report_progress)
1291 progress = start_delayed_progress(
1292 _("Computing commit changed paths Bloom filters"),
1293 ctx->commits.nr);
1294
1295 ALLOC_ARRAY(sorted_commits, ctx->commits.nr);
1296 COPY_ARRAY(sorted_commits, ctx->commits.list, ctx->commits.nr);
1297
1298 if (ctx->order_by_pack)
1299 QSORT(sorted_commits, ctx->commits.nr, commit_pos_cmp);
1300 else
1301 QSORT(sorted_commits, ctx->commits.nr, commit_gen_cmp);
1302
1303 for (i = 0; i < ctx->commits.nr; i++) {
1304 struct commit *c = sorted_commits[i];
1305 struct bloom_filter *filter = get_bloom_filter(ctx->r, c, 1);
1306 ctx->total_bloom_filter_data_size += sizeof(unsigned char) * filter->len;
1307 display_progress(progress, i + 1);
1308 }
1309
1310 free(sorted_commits);
1311 stop_progress(&progress);
1312 }
1313
1314 static int add_ref_to_list(const char *refname,
1315 const struct object_id *oid,
1316 int flags, void *cb_data)
1317 {
1318 struct string_list *list = (struct string_list *)cb_data;
1319
1320 string_list_append(list, oid_to_hex(oid));
1321 return 0;
1322 }
1323
1324 int write_commit_graph_reachable(struct object_directory *odb,
1325 enum commit_graph_write_flags flags,
1326 const struct split_commit_graph_opts *split_opts)
1327 {
1328 struct string_list list = STRING_LIST_INIT_DUP;
1329 int result;
1330
1331 for_each_ref(add_ref_to_list, &list);
1332 result = write_commit_graph(odb, NULL, &list,
1333 flags, split_opts);
1334
1335 string_list_clear(&list, 0);
1336 return result;
1337 }
1338
1339 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
1340 struct string_list *pack_indexes)
1341 {
1342 uint32_t i;
1343 struct strbuf progress_title = STRBUF_INIT;
1344 struct strbuf packname = STRBUF_INIT;
1345 int dirlen;
1346
1347 strbuf_addf(&packname, "%s/pack/", ctx->odb->path);
1348 dirlen = packname.len;
1349 if (ctx->report_progress) {
1350 strbuf_addf(&progress_title,
1351 Q_("Finding commits for commit graph in %d pack",
1352 "Finding commits for commit graph in %d packs",
1353 pack_indexes->nr),
1354 pack_indexes->nr);
1355 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1356 ctx->progress_done = 0;
1357 }
1358 for (i = 0; i < pack_indexes->nr; i++) {
1359 struct packed_git *p;
1360 strbuf_setlen(&packname, dirlen);
1361 strbuf_addstr(&packname, pack_indexes->items[i].string);
1362 p = add_packed_git(packname.buf, packname.len, 1);
1363 if (!p) {
1364 error(_("error adding pack %s"), packname.buf);
1365 return -1;
1366 }
1367 if (open_pack_index(p)) {
1368 error(_("error opening index for %s"), packname.buf);
1369 return -1;
1370 }
1371 for_each_object_in_pack(p, add_packed_commits, ctx,
1372 FOR_EACH_OBJECT_PACK_ORDER);
1373 close_pack(p);
1374 free(p);
1375 }
1376
1377 stop_progress(&ctx->progress);
1378 strbuf_release(&progress_title);
1379 strbuf_release(&packname);
1380
1381 return 0;
1382 }
1383
1384 static int fill_oids_from_commit_hex(struct write_commit_graph_context *ctx,
1385 struct string_list *commit_hex)
1386 {
1387 uint32_t i;
1388 struct strbuf progress_title = STRBUF_INIT;
1389
1390 if (ctx->report_progress) {
1391 strbuf_addf(&progress_title,
1392 Q_("Finding commits for commit graph from %d ref",
1393 "Finding commits for commit graph from %d refs",
1394 commit_hex->nr),
1395 commit_hex->nr);
1396 ctx->progress = start_delayed_progress(
1397 progress_title.buf,
1398 commit_hex->nr);
1399 }
1400 for (i = 0; i < commit_hex->nr; i++) {
1401 const char *end;
1402 struct object_id oid;
1403 struct commit *result;
1404
1405 display_progress(ctx->progress, i + 1);
1406 if (!parse_oid_hex(commit_hex->items[i].string, &oid, &end) &&
1407 (result = lookup_commit_reference_gently(ctx->r, &oid, 1))) {
1408 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1409 oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
1410 ctx->oids.nr++;
1411 } else if (ctx->check_oids) {
1412 error(_("invalid commit object id: %s"),
1413 commit_hex->items[i].string);
1414 return -1;
1415 }
1416 }
1417 stop_progress(&ctx->progress);
1418 strbuf_release(&progress_title);
1419
1420 return 0;
1421 }
1422
1423 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1424 {
1425 if (ctx->report_progress)
1426 ctx->progress = start_delayed_progress(
1427 _("Finding commits for commit graph among packed objects"),
1428 ctx->approx_nr_objects);
1429 for_each_packed_object(add_packed_commits, ctx,
1430 FOR_EACH_OBJECT_PACK_ORDER);
1431 if (ctx->progress_done < ctx->approx_nr_objects)
1432 display_progress(ctx->progress, ctx->approx_nr_objects);
1433 stop_progress(&ctx->progress);
1434 }
1435
1436 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
1437 {
1438 uint32_t i, count_distinct = 1;
1439
1440 if (ctx->report_progress)
1441 ctx->progress = start_delayed_progress(
1442 _("Counting distinct commits in commit graph"),
1443 ctx->oids.nr);
1444 display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1445 QSORT(ctx->oids.list, ctx->oids.nr, oid_compare);
1446
1447 for (i = 1; i < ctx->oids.nr; i++) {
1448 display_progress(ctx->progress, i + 1);
1449 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i])) {
1450 if (ctx->split) {
1451 struct commit *c = lookup_commit(ctx->r, &ctx->oids.list[i]);
1452
1453 if (!c || c->graph_pos != COMMIT_NOT_FROM_GRAPH)
1454 continue;
1455 }
1456
1457 count_distinct++;
1458 }
1459 }
1460 stop_progress(&ctx->progress);
1461
1462 return count_distinct;
1463 }
1464
1465 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1466 {
1467 uint32_t i;
1468
1469 ctx->num_extra_edges = 0;
1470 if (ctx->report_progress)
1471 ctx->progress = start_delayed_progress(
1472 _("Finding extra edges in commit graph"),
1473 ctx->oids.nr);
1474 for (i = 0; i < ctx->oids.nr; i++) {
1475 unsigned int num_parents;
1476
1477 display_progress(ctx->progress, i + 1);
1478 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1479 continue;
1480
1481 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + 1, ctx->commits.alloc);
1482 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1483
1484 if (ctx->split &&
1485 ctx->commits.list[ctx->commits.nr]->graph_pos != COMMIT_NOT_FROM_GRAPH)
1486 continue;
1487
1488 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1489
1490 num_parents = commit_list_count(ctx->commits.list[ctx->commits.nr]->parents);
1491 if (num_parents > 2)
1492 ctx->num_extra_edges += num_parents - 1;
1493
1494 ctx->commits.nr++;
1495 }
1496 stop_progress(&ctx->progress);
1497 }
1498
1499 static int write_graph_chunk_base_1(struct hashfile *f,
1500 struct commit_graph *g)
1501 {
1502 int num = 0;
1503
1504 if (!g)
1505 return 0;
1506
1507 num = write_graph_chunk_base_1(f, g->base_graph);
1508 hashwrite(f, g->oid.hash, the_hash_algo->rawsz);
1509 return num + 1;
1510 }
1511
1512 static int write_graph_chunk_base(struct hashfile *f,
1513 struct write_commit_graph_context *ctx)
1514 {
1515 int num = write_graph_chunk_base_1(f, ctx->new_base_graph);
1516
1517 if (num != ctx->num_commit_graphs_after - 1) {
1518 error(_("failed to write correct number of base graph ids"));
1519 return -1;
1520 }
1521
1522 return 0;
1523 }
1524
1525 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1526 {
1527 uint32_t i;
1528 int fd;
1529 struct hashfile *f;
1530 struct lock_file lk = LOCK_INIT;
1531 uint32_t chunk_ids[MAX_NUM_CHUNKS + 1];
1532 uint64_t chunk_sizes[MAX_NUM_CHUNKS + 1];
1533 const unsigned hashsz = the_hash_algo->rawsz;
1534 struct strbuf progress_title = STRBUF_INIT;
1535 int num_chunks = 3;
1536 uint64_t chunk_offset;
1537 struct object_id file_hash;
1538 const struct bloom_filter_settings bloom_settings = DEFAULT_BLOOM_FILTER_SETTINGS;
1539
1540 if (ctx->split) {
1541 struct strbuf tmp_file = STRBUF_INIT;
1542
1543 strbuf_addf(&tmp_file,
1544 "%s/info/commit-graphs/tmp_graph_XXXXXX",
1545 ctx->odb->path);
1546 ctx->graph_name = strbuf_detach(&tmp_file, NULL);
1547 } else {
1548 ctx->graph_name = get_commit_graph_filename(ctx->odb);
1549 }
1550
1551 if (safe_create_leading_directories(ctx->graph_name)) {
1552 UNLEAK(ctx->graph_name);
1553 error(_("unable to create leading directories of %s"),
1554 ctx->graph_name);
1555 return -1;
1556 }
1557
1558 if (ctx->split) {
1559 char *lock_name = get_chain_filename(ctx->odb);
1560
1561 hold_lock_file_for_update(&lk, lock_name, LOCK_DIE_ON_ERROR);
1562
1563 fd = git_mkstemp_mode(ctx->graph_name, 0444);
1564 if (fd < 0) {
1565 error(_("unable to create '%s'"), ctx->graph_name);
1566 return -1;
1567 }
1568
1569 f = hashfd(fd, ctx->graph_name);
1570 } else {
1571 hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
1572 fd = lk.tempfile->fd;
1573 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1574 }
1575
1576 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1577 chunk_sizes[0] = GRAPH_FANOUT_SIZE;
1578 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1579 chunk_sizes[1] = hashsz * ctx->commits.nr;
1580 chunk_ids[2] = GRAPH_CHUNKID_DATA;
1581 chunk_sizes[2] = (hashsz + 16) * ctx->commits.nr;
1582
1583 if (ctx->num_extra_edges) {
1584 chunk_ids[num_chunks] = GRAPH_CHUNKID_EXTRAEDGES;
1585 chunk_sizes[num_chunks] = 4 * ctx->num_extra_edges;
1586 num_chunks++;
1587 }
1588 if (ctx->changed_paths) {
1589 chunk_ids[num_chunks] = GRAPH_CHUNKID_BLOOMINDEXES;
1590 chunk_sizes[num_chunks] = sizeof(uint32_t) * ctx->commits.nr;
1591 num_chunks++;
1592 chunk_ids[num_chunks] = GRAPH_CHUNKID_BLOOMDATA;
1593 chunk_sizes[num_chunks] = sizeof(uint32_t) * 3
1594 + ctx->total_bloom_filter_data_size;
1595 num_chunks++;
1596 }
1597 if (ctx->num_commit_graphs_after > 1) {
1598 chunk_ids[num_chunks] = GRAPH_CHUNKID_BASE;
1599 chunk_sizes[num_chunks] = hashsz * (ctx->num_commit_graphs_after - 1);
1600 num_chunks++;
1601 }
1602
1603 chunk_ids[num_chunks] = 0;
1604 chunk_sizes[num_chunks] = 0;
1605
1606 hashwrite_be32(f, GRAPH_SIGNATURE);
1607
1608 hashwrite_u8(f, GRAPH_VERSION);
1609 hashwrite_u8(f, oid_version());
1610 hashwrite_u8(f, num_chunks);
1611 hashwrite_u8(f, ctx->num_commit_graphs_after - 1);
1612
1613 chunk_offset = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1614 for (i = 0; i <= num_chunks; i++) {
1615 uint32_t chunk_write[3];
1616
1617 chunk_write[0] = htonl(chunk_ids[i]);
1618 chunk_write[1] = htonl(chunk_offset >> 32);
1619 chunk_write[2] = htonl(chunk_offset & 0xffffffff);
1620 hashwrite(f, chunk_write, 12);
1621
1622 chunk_offset += chunk_sizes[i];
1623 }
1624
1625 if (ctx->report_progress) {
1626 strbuf_addf(&progress_title,
1627 Q_("Writing out commit graph in %d pass",
1628 "Writing out commit graph in %d passes",
1629 num_chunks),
1630 num_chunks);
1631 ctx->progress = start_delayed_progress(
1632 progress_title.buf,
1633 num_chunks * ctx->commits.nr);
1634 }
1635 write_graph_chunk_fanout(f, ctx);
1636 write_graph_chunk_oids(f, hashsz, ctx);
1637 write_graph_chunk_data(f, hashsz, ctx);
1638 if (ctx->num_extra_edges)
1639 write_graph_chunk_extra_edges(f, ctx);
1640 if (ctx->changed_paths) {
1641 write_graph_chunk_bloom_indexes(f, ctx);
1642 write_graph_chunk_bloom_data(f, ctx, &bloom_settings);
1643 }
1644 if (ctx->num_commit_graphs_after > 1 &&
1645 write_graph_chunk_base(f, ctx)) {
1646 return -1;
1647 }
1648 stop_progress(&ctx->progress);
1649 strbuf_release(&progress_title);
1650
1651 if (ctx->split && ctx->base_graph_name && ctx->num_commit_graphs_after > 1) {
1652 char *new_base_hash = xstrdup(oid_to_hex(&ctx->new_base_graph->oid));
1653 char *new_base_name = get_split_graph_filename(ctx->new_base_graph->odb, new_base_hash);
1654
1655 free(ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2]);
1656 free(ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2]);
1657 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2] = new_base_name;
1658 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2] = new_base_hash;
1659 }
1660
1661 close_commit_graph(ctx->r->objects);
1662 finalize_hashfile(f, file_hash.hash, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1663
1664 if (ctx->split) {
1665 FILE *chainf = fdopen_lock_file(&lk, "w");
1666 char *final_graph_name;
1667 int result;
1668
1669 close(fd);
1670
1671 if (!chainf) {
1672 error(_("unable to open commit-graph chain file"));
1673 return -1;
1674 }
1675
1676 if (ctx->base_graph_name) {
1677 const char *dest = ctx->commit_graph_filenames_after[
1678 ctx->num_commit_graphs_after - 2];
1679
1680 if (strcmp(ctx->base_graph_name, dest)) {
1681 result = rename(ctx->base_graph_name, dest);
1682
1683 if (result) {
1684 error(_("failed to rename base commit-graph file"));
1685 return -1;
1686 }
1687 }
1688 } else {
1689 char *graph_name = get_commit_graph_filename(ctx->odb);
1690 unlink(graph_name);
1691 }
1692
1693 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1] = xstrdup(oid_to_hex(&file_hash));
1694 final_graph_name = get_split_graph_filename(ctx->odb,
1695 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1]);
1696 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 1] = final_graph_name;
1697
1698 result = rename(ctx->graph_name, final_graph_name);
1699
1700 for (i = 0; i < ctx->num_commit_graphs_after; i++)
1701 fprintf(lk.tempfile->fp, "%s\n", ctx->commit_graph_hash_after[i]);
1702
1703 if (result) {
1704 error(_("failed to rename temporary commit-graph file"));
1705 return -1;
1706 }
1707 }
1708
1709 commit_lock_file(&lk);
1710
1711 return 0;
1712 }
1713
1714 static void split_graph_merge_strategy(struct write_commit_graph_context *ctx)
1715 {
1716 struct commit_graph *g;
1717 uint32_t num_commits;
1718 uint32_t i;
1719
1720 int max_commits = 0;
1721 int size_mult = 2;
1722
1723 if (ctx->split_opts) {
1724 max_commits = ctx->split_opts->max_commits;
1725
1726 if (ctx->split_opts->size_multiple)
1727 size_mult = ctx->split_opts->size_multiple;
1728 }
1729
1730 g = ctx->r->objects->commit_graph;
1731 num_commits = ctx->commits.nr;
1732 ctx->num_commit_graphs_after = ctx->num_commit_graphs_before + 1;
1733
1734 while (g && (g->num_commits <= size_mult * num_commits ||
1735 (max_commits && num_commits > max_commits))) {
1736 if (g->odb != ctx->odb)
1737 break;
1738
1739 num_commits += g->num_commits;
1740 g = g->base_graph;
1741
1742 ctx->num_commit_graphs_after--;
1743 }
1744
1745 ctx->new_base_graph = g;
1746
1747 if (ctx->num_commit_graphs_after == 2) {
1748 char *old_graph_name = get_commit_graph_filename(g->odb);
1749
1750 if (!strcmp(g->filename, old_graph_name) &&
1751 g->odb != ctx->odb) {
1752 ctx->num_commit_graphs_after = 1;
1753 ctx->new_base_graph = NULL;
1754 }
1755
1756 free(old_graph_name);
1757 }
1758
1759 ALLOC_ARRAY(ctx->commit_graph_filenames_after, ctx->num_commit_graphs_after);
1760 ALLOC_ARRAY(ctx->commit_graph_hash_after, ctx->num_commit_graphs_after);
1761
1762 for (i = 0; i < ctx->num_commit_graphs_after &&
1763 i < ctx->num_commit_graphs_before; i++)
1764 ctx->commit_graph_filenames_after[i] = xstrdup(ctx->commit_graph_filenames_before[i]);
1765
1766 i = ctx->num_commit_graphs_before - 1;
1767 g = ctx->r->objects->commit_graph;
1768
1769 while (g) {
1770 if (i < ctx->num_commit_graphs_after)
1771 ctx->commit_graph_hash_after[i] = xstrdup(oid_to_hex(&g->oid));
1772
1773 i--;
1774 g = g->base_graph;
1775 }
1776 }
1777
1778 static void merge_commit_graph(struct write_commit_graph_context *ctx,
1779 struct commit_graph *g)
1780 {
1781 uint32_t i;
1782 uint32_t offset = g->num_commits_in_base;
1783
1784 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + g->num_commits, ctx->commits.alloc);
1785
1786 for (i = 0; i < g->num_commits; i++) {
1787 struct object_id oid;
1788 struct commit *result;
1789
1790 display_progress(ctx->progress, i + 1);
1791
1792 load_oid_from_graph(g, i + offset, &oid);
1793
1794 /* only add commits if they still exist in the repo */
1795 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1796
1797 if (result) {
1798 ctx->commits.list[ctx->commits.nr] = result;
1799 ctx->commits.nr++;
1800 }
1801 }
1802 }
1803
1804 static int commit_compare(const void *_a, const void *_b)
1805 {
1806 const struct commit *a = *(const struct commit **)_a;
1807 const struct commit *b = *(const struct commit **)_b;
1808 return oidcmp(&a->object.oid, &b->object.oid);
1809 }
1810
1811 static void sort_and_scan_merged_commits(struct write_commit_graph_context *ctx)
1812 {
1813 uint32_t i;
1814
1815 if (ctx->report_progress)
1816 ctx->progress = start_delayed_progress(
1817 _("Scanning merged commits"),
1818 ctx->commits.nr);
1819
1820 QSORT(ctx->commits.list, ctx->commits.nr, commit_compare);
1821
1822 ctx->num_extra_edges = 0;
1823 for (i = 0; i < ctx->commits.nr; i++) {
1824 display_progress(ctx->progress, i);
1825
1826 if (i && oideq(&ctx->commits.list[i - 1]->object.oid,
1827 &ctx->commits.list[i]->object.oid)) {
1828 die(_("unexpected duplicate commit id %s"),
1829 oid_to_hex(&ctx->commits.list[i]->object.oid));
1830 } else {
1831 unsigned int num_parents;
1832
1833 num_parents = commit_list_count(ctx->commits.list[i]->parents);
1834 if (num_parents > 2)
1835 ctx->num_extra_edges += num_parents - 1;
1836 }
1837 }
1838
1839 stop_progress(&ctx->progress);
1840 }
1841
1842 static void merge_commit_graphs(struct write_commit_graph_context *ctx)
1843 {
1844 struct commit_graph *g = ctx->r->objects->commit_graph;
1845 uint32_t current_graph_number = ctx->num_commit_graphs_before;
1846
1847 while (g && current_graph_number >= ctx->num_commit_graphs_after) {
1848 current_graph_number--;
1849
1850 if (ctx->report_progress)
1851 ctx->progress = start_delayed_progress(_("Merging commit-graph"), 0);
1852
1853 merge_commit_graph(ctx, g);
1854 stop_progress(&ctx->progress);
1855
1856 g = g->base_graph;
1857 }
1858
1859 if (g) {
1860 ctx->new_base_graph = g;
1861 ctx->new_num_commits_in_base = g->num_commits + g->num_commits_in_base;
1862 }
1863
1864 if (ctx->new_base_graph)
1865 ctx->base_graph_name = xstrdup(ctx->new_base_graph->filename);
1866
1867 sort_and_scan_merged_commits(ctx);
1868 }
1869
1870 static void mark_commit_graphs(struct write_commit_graph_context *ctx)
1871 {
1872 uint32_t i;
1873 time_t now = time(NULL);
1874
1875 for (i = ctx->num_commit_graphs_after - 1; i < ctx->num_commit_graphs_before; i++) {
1876 struct stat st;
1877 struct utimbuf updated_time;
1878
1879 stat(ctx->commit_graph_filenames_before[i], &st);
1880
1881 updated_time.actime = st.st_atime;
1882 updated_time.modtime = now;
1883 utime(ctx->commit_graph_filenames_before[i], &updated_time);
1884 }
1885 }
1886
1887 static void expire_commit_graphs(struct write_commit_graph_context *ctx)
1888 {
1889 struct strbuf path = STRBUF_INIT;
1890 DIR *dir;
1891 struct dirent *de;
1892 size_t dirnamelen;
1893 timestamp_t expire_time = time(NULL);
1894
1895 if (ctx->split_opts && ctx->split_opts->expire_time)
1896 expire_time -= ctx->split_opts->expire_time;
1897 if (!ctx->split) {
1898 char *chain_file_name = get_chain_filename(ctx->odb);
1899 unlink(chain_file_name);
1900 free(chain_file_name);
1901 ctx->num_commit_graphs_after = 0;
1902 }
1903
1904 strbuf_addstr(&path, ctx->odb->path);
1905 strbuf_addstr(&path, "/info/commit-graphs");
1906 dir = opendir(path.buf);
1907
1908 if (!dir)
1909 goto out;
1910
1911 strbuf_addch(&path, '/');
1912 dirnamelen = path.len;
1913 while ((de = readdir(dir)) != NULL) {
1914 struct stat st;
1915 uint32_t i, found = 0;
1916
1917 strbuf_setlen(&path, dirnamelen);
1918 strbuf_addstr(&path, de->d_name);
1919
1920 stat(path.buf, &st);
1921
1922 if (st.st_mtime > expire_time)
1923 continue;
1924 if (path.len < 6 || strcmp(path.buf + path.len - 6, ".graph"))
1925 continue;
1926
1927 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
1928 if (!strcmp(ctx->commit_graph_filenames_after[i],
1929 path.buf)) {
1930 found = 1;
1931 break;
1932 }
1933 }
1934
1935 if (!found)
1936 unlink(path.buf);
1937 }
1938
1939 out:
1940 strbuf_release(&path);
1941 }
1942
1943 int write_commit_graph(struct object_directory *odb,
1944 struct string_list *pack_indexes,
1945 struct string_list *commit_hex,
1946 enum commit_graph_write_flags flags,
1947 const struct split_commit_graph_opts *split_opts)
1948 {
1949 struct write_commit_graph_context *ctx;
1950 uint32_t i, count_distinct = 0;
1951 int res = 0;
1952
1953 if (!commit_graph_compatible(the_repository))
1954 return 0;
1955
1956 ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
1957 ctx->r = the_repository;
1958 ctx->odb = odb;
1959 ctx->append = flags & COMMIT_GRAPH_WRITE_APPEND ? 1 : 0;
1960 ctx->report_progress = flags & COMMIT_GRAPH_WRITE_PROGRESS ? 1 : 0;
1961 ctx->split = flags & COMMIT_GRAPH_WRITE_SPLIT ? 1 : 0;
1962 ctx->check_oids = flags & COMMIT_GRAPH_WRITE_CHECK_OIDS ? 1 : 0;
1963 ctx->split_opts = split_opts;
1964 ctx->changed_paths = flags & COMMIT_GRAPH_WRITE_BLOOM_FILTERS ? 1 : 0;
1965 ctx->total_bloom_filter_data_size = 0;
1966
1967 if (ctx->split) {
1968 struct commit_graph *g;
1969 prepare_commit_graph(ctx->r);
1970
1971 g = ctx->r->objects->commit_graph;
1972
1973 while (g) {
1974 ctx->num_commit_graphs_before++;
1975 g = g->base_graph;
1976 }
1977
1978 if (ctx->num_commit_graphs_before) {
1979 ALLOC_ARRAY(ctx->commit_graph_filenames_before, ctx->num_commit_graphs_before);
1980 i = ctx->num_commit_graphs_before;
1981 g = ctx->r->objects->commit_graph;
1982
1983 while (g) {
1984 ctx->commit_graph_filenames_before[--i] = xstrdup(g->filename);
1985 g = g->base_graph;
1986 }
1987 }
1988 }
1989
1990 ctx->approx_nr_objects = approximate_object_count();
1991 ctx->oids.alloc = ctx->approx_nr_objects / 32;
1992
1993 if (ctx->split && split_opts && ctx->oids.alloc > split_opts->max_commits)
1994 ctx->oids.alloc = split_opts->max_commits;
1995
1996 if (ctx->append) {
1997 prepare_commit_graph_one(ctx->r, ctx->odb);
1998 if (ctx->r->objects->commit_graph)
1999 ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
2000 }
2001
2002 if (ctx->oids.alloc < 1024)
2003 ctx->oids.alloc = 1024;
2004 ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
2005
2006 if (ctx->append && ctx->r->objects->commit_graph) {
2007 struct commit_graph *g = ctx->r->objects->commit_graph;
2008 for (i = 0; i < g->num_commits; i++) {
2009 const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
2010 hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
2011 }
2012 }
2013
2014 if (pack_indexes) {
2015 ctx->order_by_pack = 1;
2016 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
2017 goto cleanup;
2018 }
2019
2020 if (commit_hex) {
2021 if ((res = fill_oids_from_commit_hex(ctx, commit_hex)))
2022 goto cleanup;
2023 }
2024
2025 if (!pack_indexes && !commit_hex) {
2026 ctx->order_by_pack = 1;
2027 fill_oids_from_all_packs(ctx);
2028 }
2029
2030 close_reachable(ctx);
2031
2032 count_distinct = count_distinct_commits(ctx);
2033
2034 if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
2035 error(_("the commit graph format cannot write %d commits"), count_distinct);
2036 res = -1;
2037 goto cleanup;
2038 }
2039
2040 ctx->commits.alloc = count_distinct;
2041 ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
2042
2043 copy_oids_to_commits(ctx);
2044
2045 if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
2046 error(_("too many commits to write graph"));
2047 res = -1;
2048 goto cleanup;
2049 }
2050
2051 if (!ctx->commits.nr)
2052 goto cleanup;
2053
2054 if (ctx->split) {
2055 split_graph_merge_strategy(ctx);
2056
2057 merge_commit_graphs(ctx);
2058 } else
2059 ctx->num_commit_graphs_after = 1;
2060
2061 compute_generation_numbers(ctx);
2062
2063 if (ctx->changed_paths)
2064 compute_bloom_filters(ctx);
2065
2066 res = write_commit_graph_file(ctx);
2067
2068 if (ctx->split)
2069 mark_commit_graphs(ctx);
2070
2071 expire_commit_graphs(ctx);
2072
2073 cleanup:
2074 free(ctx->graph_name);
2075 free(ctx->commits.list);
2076 free(ctx->oids.list);
2077
2078 if (ctx->commit_graph_filenames_after) {
2079 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
2080 free(ctx->commit_graph_filenames_after[i]);
2081 free(ctx->commit_graph_hash_after[i]);
2082 }
2083
2084 for (i = 0; i < ctx->num_commit_graphs_before; i++)
2085 free(ctx->commit_graph_filenames_before[i]);
2086
2087 free(ctx->commit_graph_filenames_after);
2088 free(ctx->commit_graph_filenames_before);
2089 free(ctx->commit_graph_hash_after);
2090 }
2091
2092 free(ctx);
2093
2094 return res;
2095 }
2096
2097 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
2098 static int verify_commit_graph_error;
2099
2100 static void graph_report(const char *fmt, ...)
2101 {
2102 va_list ap;
2103
2104 verify_commit_graph_error = 1;
2105 va_start(ap, fmt);
2106 vfprintf(stderr, fmt, ap);
2107 fprintf(stderr, "\n");
2108 va_end(ap);
2109 }
2110
2111 #define GENERATION_ZERO_EXISTS 1
2112 #define GENERATION_NUMBER_EXISTS 2
2113
2114 int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
2115 {
2116 uint32_t i, cur_fanout_pos = 0;
2117 struct object_id prev_oid, cur_oid, checksum;
2118 int generation_zero = 0;
2119 struct hashfile *f;
2120 int devnull;
2121 struct progress *progress = NULL;
2122 int local_error = 0;
2123
2124 if (!g) {
2125 graph_report("no commit-graph file loaded");
2126 return 1;
2127 }
2128
2129 verify_commit_graph_error = verify_commit_graph_lite(g);
2130 if (verify_commit_graph_error)
2131 return verify_commit_graph_error;
2132
2133 devnull = open("/dev/null", O_WRONLY);
2134 f = hashfd(devnull, NULL);
2135 hashwrite(f, g->data, g->data_len - g->hash_len);
2136 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
2137 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
2138 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
2139 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
2140 }
2141
2142 for (i = 0; i < g->num_commits; i++) {
2143 struct commit *graph_commit;
2144
2145 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2146
2147 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
2148 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
2149 oid_to_hex(&prev_oid),
2150 oid_to_hex(&cur_oid));
2151
2152 oidcpy(&prev_oid, &cur_oid);
2153
2154 while (cur_oid.hash[0] > cur_fanout_pos) {
2155 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2156
2157 if (i != fanout_value)
2158 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2159 cur_fanout_pos, fanout_value, i);
2160 cur_fanout_pos++;
2161 }
2162
2163 graph_commit = lookup_commit(r, &cur_oid);
2164 if (!parse_commit_in_graph_one(r, g, graph_commit))
2165 graph_report(_("failed to parse commit %s from commit-graph"),
2166 oid_to_hex(&cur_oid));
2167 }
2168
2169 while (cur_fanout_pos < 256) {
2170 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2171
2172 if (g->num_commits != fanout_value)
2173 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2174 cur_fanout_pos, fanout_value, i);
2175
2176 cur_fanout_pos++;
2177 }
2178
2179 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
2180 return verify_commit_graph_error;
2181
2182 if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
2183 progress = start_progress(_("Verifying commits in commit graph"),
2184 g->num_commits);
2185
2186 for (i = 0; i < g->num_commits; i++) {
2187 struct commit *graph_commit, *odb_commit;
2188 struct commit_list *graph_parents, *odb_parents;
2189 uint32_t max_generation = 0;
2190
2191 display_progress(progress, i + 1);
2192 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2193
2194 graph_commit = lookup_commit(r, &cur_oid);
2195 odb_commit = (struct commit *)create_object(r, &cur_oid, alloc_commit_node(r));
2196 if (parse_commit_internal(odb_commit, 0, 0)) {
2197 graph_report(_("failed to parse commit %s from object database for commit-graph"),
2198 oid_to_hex(&cur_oid));
2199 continue;
2200 }
2201
2202 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
2203 get_commit_tree_oid(odb_commit)))
2204 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2205 oid_to_hex(&cur_oid),
2206 oid_to_hex(get_commit_tree_oid(graph_commit)),
2207 oid_to_hex(get_commit_tree_oid(odb_commit)));
2208
2209 graph_parents = graph_commit->parents;
2210 odb_parents = odb_commit->parents;
2211
2212 while (graph_parents) {
2213 if (odb_parents == NULL) {
2214 graph_report(_("commit-graph parent list for commit %s is too long"),
2215 oid_to_hex(&cur_oid));
2216 break;
2217 }
2218
2219 /* parse parent in case it is in a base graph */
2220 parse_commit_in_graph_one(r, g, graph_parents->item);
2221
2222 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
2223 graph_report(_("commit-graph parent for %s is %s != %s"),
2224 oid_to_hex(&cur_oid),
2225 oid_to_hex(&graph_parents->item->object.oid),
2226 oid_to_hex(&odb_parents->item->object.oid));
2227
2228 if (graph_parents->item->generation > max_generation)
2229 max_generation = graph_parents->item->generation;
2230
2231 graph_parents = graph_parents->next;
2232 odb_parents = odb_parents->next;
2233 }
2234
2235 if (odb_parents != NULL)
2236 graph_report(_("commit-graph parent list for commit %s terminates early"),
2237 oid_to_hex(&cur_oid));
2238
2239 if (!graph_commit->generation) {
2240 if (generation_zero == GENERATION_NUMBER_EXISTS)
2241 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2242 oid_to_hex(&cur_oid));
2243 generation_zero = GENERATION_ZERO_EXISTS;
2244 } else if (generation_zero == GENERATION_ZERO_EXISTS)
2245 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2246 oid_to_hex(&cur_oid));
2247
2248 if (generation_zero == GENERATION_ZERO_EXISTS)
2249 continue;
2250
2251 /*
2252 * If one of our parents has generation GENERATION_NUMBER_MAX, then
2253 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
2254 * extra logic in the following condition.
2255 */
2256 if (max_generation == GENERATION_NUMBER_MAX)
2257 max_generation--;
2258
2259 if (graph_commit->generation != max_generation + 1)
2260 graph_report(_("commit-graph generation for commit %s is %u != %u"),
2261 oid_to_hex(&cur_oid),
2262 graph_commit->generation,
2263 max_generation + 1);
2264
2265 if (graph_commit->date != odb_commit->date)
2266 graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
2267 oid_to_hex(&cur_oid),
2268 graph_commit->date,
2269 odb_commit->date);
2270 }
2271 stop_progress(&progress);
2272
2273 local_error = verify_commit_graph_error;
2274
2275 if (!(flags & COMMIT_GRAPH_VERIFY_SHALLOW) && g->base_graph)
2276 local_error |= verify_commit_graph(r, g->base_graph, flags);
2277
2278 return local_error;
2279 }
2280
2281 void free_commit_graph(struct commit_graph *g)
2282 {
2283 if (!g)
2284 return;
2285 if (g->graph_fd >= 0) {
2286 munmap((void *)g->data, g->data_len);
2287 g->data = NULL;
2288 close(g->graph_fd);
2289 }
2290 free(g->filename);
2291 free(g->bloom_filter_settings);
2292 free(g);
2293 }
2294
2295 void disable_commit_graph(struct repository *r)
2296 {
2297 r->commit_graph_disabled = 1;
2298 }
git for Debian