?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
bloom: fix whitespace around tab length
[git.git] / commit-graph.c
blob196e817a84c04259d6d6a07b06e0be5abe232298
1 #include "cache.h"
2 #include "config.h"
3 #include "dir.h"
4 #include "git-compat-util.h"
5 #include "lockfile.h"
6 #include "pack.h"
7 #include "packfile.h"
8 #include "commit.h"
9 #include "object.h"
10 #include "refs.h"
11 #include "revision.h"
12 #include "sha1-lookup.h"
13 #include "commit-graph.h"
14 #include "object-store.h"
15 #include "alloc.h"
16 #include "hashmap.h"
17 #include "replace-object.h"
18 #include "progress.h"
19 #include "bloom.h"
20 #include "commit-slab.h"
21
22 void git_test_write_commit_graph_or_die(void)
23 {
24 int flags = 0;
25 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0))
26 return;
27
28 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS, 0))
29 flags = COMMIT_GRAPH_WRITE_BLOOM_FILTERS;
30
31 if (write_commit_graph_reachable(the_repository->objects->odb,
32 flags, NULL))
33 die("failed to write commit-graph under GIT_TEST_COMMIT_GRAPH");
34 }
35
36 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
37 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
38 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
39 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
40 #define GRAPH_CHUNKID_EXTRAEDGES 0x45444745 /* "EDGE" */
41 #define GRAPH_CHUNKID_BLOOMINDEXES 0x42494458 /* "BIDX" */
42 #define GRAPH_CHUNKID_BLOOMDATA 0x42444154 /* "BDAT" */
43 #define GRAPH_CHUNKID_BASE 0x42415345 /* "BASE" */
44 #define MAX_NUM_CHUNKS 7
45
46 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
47
48 #define GRAPH_VERSION_1 0x1
49 #define GRAPH_VERSION GRAPH_VERSION_1
50
51 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
52 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
53 #define GRAPH_PARENT_NONE 0x70000000
54
55 #define GRAPH_LAST_EDGE 0x80000000
56
57 #define GRAPH_HEADER_SIZE 8
58 #define GRAPH_FANOUT_SIZE (4 * 256)
59 #define GRAPH_CHUNKLOOKUP_WIDTH 12
60 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
61 + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
62
63 /* Remember to update object flag allocation in object.h */
64 #define REACHABLE (1u<<15)
65
66 /* Keep track of the order in which commits are added to our list. */
67 define_commit_slab(commit_pos, int);
68 static struct commit_pos commit_pos = COMMIT_SLAB_INIT(1, commit_pos);
69
70 static void set_commit_pos(struct repository *r, const struct object_id *oid)
71 {
72 static int32_t max_pos;
73 struct commit *commit = lookup_commit(r, oid);
74
75 if (!commit)
76 return; /* should never happen, but be lenient */
77
78 *commit_pos_at(&commit_pos, commit) = max_pos++;
79 }
80
81 static int commit_pos_cmp(const void *va, const void *vb)
82 {
83 const struct commit *a = *(const struct commit **)va;
84 const struct commit *b = *(const struct commit **)vb;
85 return commit_pos_at(&commit_pos, a) -
86 commit_pos_at(&commit_pos, b);
87 }
88
89 static int commit_gen_cmp(const void *va, const void *vb)
90 {
91 const struct commit *a = *(const struct commit **)va;
92 const struct commit *b = *(const struct commit **)vb;
93
94 /* lower generation commits first */
95 if (a->generation < b->generation)
96 return -1;
97 else if (a->generation > b->generation)
98 return 1;
99
100 /* use date as a heuristic when generations are equal */
101 if (a->date < b->date)
102 return -1;
103 else if (a->date > b->date)
104 return 1;
105 return 0;
106 }
107
108 char *get_commit_graph_filename(struct object_directory *obj_dir)
109 {
110 return xstrfmt("%s/info/commit-graph", obj_dir->path);
111 }
112
113 static char *get_split_graph_filename(struct object_directory *odb,
114 const char *oid_hex)
115 {
116 return xstrfmt("%s/info/commit-graphs/graph-%s.graph", odb->path,
117 oid_hex);
118 }
119
120 static char *get_chain_filename(struct object_directory *odb)
121 {
122 return xstrfmt("%s/info/commit-graphs/commit-graph-chain", odb->path);
123 }
124
125 static uint8_t oid_version(void)
126 {
127 return 1;
128 }
129
130 static struct commit_graph *alloc_commit_graph(void)
131 {
132 struct commit_graph *g = xcalloc(1, sizeof(*g));
133 g->graph_fd = -1;
134
135 return g;
136 }
137
138 extern int read_replace_refs;
139
140 static int commit_graph_compatible(struct repository *r)
141 {
142 if (!r->gitdir)
143 return 0;
144
145 if (read_replace_refs) {
146 prepare_replace_object(r);
147 if (hashmap_get_size(&r->objects->replace_map->map))
148 return 0;
149 }
150
151 prepare_commit_graft(r);
152 if (r->parsed_objects && r->parsed_objects->grafts_nr)
153 return 0;
154 if (is_repository_shallow(r))
155 return 0;
156
157 return 1;
158 }
159
160 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
161 {
162 *fd = git_open(graph_file);
163 if (*fd < 0)
164 return 0;
165 if (fstat(*fd, st)) {
166 close(*fd);
167 return 0;
168 }
169 return 1;
170 }
171
172 struct commit_graph *load_commit_graph_one_fd_st(int fd, struct stat *st,
173 struct object_directory *odb)
174 {
175 void *graph_map;
176 size_t graph_size;
177 struct commit_graph *ret;
178
179 graph_size = xsize_t(st->st_size);
180
181 if (graph_size < GRAPH_MIN_SIZE) {
182 close(fd);
183 error(_("commit-graph file is too small"));
184 return NULL;
185 }
186 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
187 ret = parse_commit_graph(graph_map, fd, graph_size);
188
189 if (ret)
190 ret->odb = odb;
191 else {
192 munmap(graph_map, graph_size);
193 close(fd);
194 }
195
196 return ret;
197 }
198
199 static int verify_commit_graph_lite(struct commit_graph *g)
200 {
201 /*
202 * Basic validation shared between parse_commit_graph()
203 * which'll be called every time the graph is used, and the
204 * much more expensive verify_commit_graph() used by
205 * "commit-graph verify".
206 *
207 * There should only be very basic checks here to ensure that
208 * we don't e.g. segfault in fill_commit_in_graph(), but
209 * because this is a very hot codepath nothing that e.g. loops
210 * over g->num_commits, or runs a checksum on the commit-graph
211 * itself.
212 */
213 if (!g->chunk_oid_fanout) {
214 error("commit-graph is missing the OID Fanout chunk");
215 return 1;
216 }
217 if (!g->chunk_oid_lookup) {
218 error("commit-graph is missing the OID Lookup chunk");
219 return 1;
220 }
221 if (!g->chunk_commit_data) {
222 error("commit-graph is missing the Commit Data chunk");
223 return 1;
224 }
225
226 return 0;
227 }
228
229 struct commit_graph *parse_commit_graph(void *graph_map, int fd,
230 size_t graph_size)
231 {
232 const unsigned char *data, *chunk_lookup;
233 uint32_t i;
234 struct commit_graph *graph;
235 uint64_t last_chunk_offset;
236 uint32_t last_chunk_id;
237 uint32_t graph_signature;
238 unsigned char graph_version, hash_version;
239
240 if (!graph_map)
241 return NULL;
242
243 if (graph_size < GRAPH_MIN_SIZE)
244 return NULL;
245
246 data = (const unsigned char *)graph_map;
247
248 graph_signature = get_be32(data);
249 if (graph_signature != GRAPH_SIGNATURE) {
250 error(_("commit-graph signature %X does not match signature %X"),
251 graph_signature, GRAPH_SIGNATURE);
252 return NULL;
253 }
254
255 graph_version = *(unsigned char*)(data + 4);
256 if (graph_version != GRAPH_VERSION) {
257 error(_("commit-graph version %X does not match version %X"),
258 graph_version, GRAPH_VERSION);
259 return NULL;
260 }
261
262 hash_version = *(unsigned char*)(data + 5);
263 if (hash_version != oid_version()) {
264 error(_("commit-graph hash version %X does not match version %X"),
265 hash_version, oid_version());
266 return NULL;
267 }
268
269 graph = alloc_commit_graph();
270
271 graph->hash_len = the_hash_algo->rawsz;
272 graph->num_chunks = *(unsigned char*)(data + 6);
273 graph->graph_fd = fd;
274 graph->data = graph_map;
275 graph->data_len = graph_size;
276
277 last_chunk_id = 0;
278 last_chunk_offset = 8;
279 chunk_lookup = data + 8;
280 for (i = 0; i < graph->num_chunks; i++) {
281 uint32_t chunk_id;
282 uint64_t chunk_offset;
283 int chunk_repeated = 0;
284
285 if (data + graph_size - chunk_lookup <
286 GRAPH_CHUNKLOOKUP_WIDTH) {
287 error(_("commit-graph chunk lookup table entry missing; file may be incomplete"));
288 free(graph);
289 return NULL;
290 }
291
292 chunk_id = get_be32(chunk_lookup + 0);
293 chunk_offset = get_be64(chunk_lookup + 4);
294
295 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
296
297 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
298 error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
299 (uint32_t)chunk_offset);
300 free(graph);
301 return NULL;
302 }
303
304 switch (chunk_id) {
305 case GRAPH_CHUNKID_OIDFANOUT:
306 if (graph->chunk_oid_fanout)
307 chunk_repeated = 1;
308 else
309 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
310 break;
311
312 case GRAPH_CHUNKID_OIDLOOKUP:
313 if (graph->chunk_oid_lookup)
314 chunk_repeated = 1;
315 else
316 graph->chunk_oid_lookup = data + chunk_offset;
317 break;
318
319 case GRAPH_CHUNKID_DATA:
320 if (graph->chunk_commit_data)
321 chunk_repeated = 1;
322 else
323 graph->chunk_commit_data = data + chunk_offset;
324 break;
325
326 case GRAPH_CHUNKID_EXTRAEDGES:
327 if (graph->chunk_extra_edges)
328 chunk_repeated = 1;
329 else
330 graph->chunk_extra_edges = data + chunk_offset;
331 break;
332
333 case GRAPH_CHUNKID_BASE:
334 if (graph->chunk_base_graphs)
335 chunk_repeated = 1;
336 else
337 graph->chunk_base_graphs = data + chunk_offset;
338 break;
339
340 case GRAPH_CHUNKID_BLOOMINDEXES:
341 if (graph->chunk_bloom_indexes)
342 chunk_repeated = 1;
343 else
344 graph->chunk_bloom_indexes = data + chunk_offset;
345 break;
346
347 case GRAPH_CHUNKID_BLOOMDATA:
348 if (graph->chunk_bloom_data)
349 chunk_repeated = 1;
350 else {
351 uint32_t hash_version;
352 graph->chunk_bloom_data = data + chunk_offset;
353 hash_version = get_be32(data + chunk_offset);
354
355 if (hash_version != 1)
356 break;
357
358 graph->bloom_filter_settings = xmalloc(sizeof(struct bloom_filter_settings));
359 graph->bloom_filter_settings->hash_version = hash_version;
360 graph->bloom_filter_settings->num_hashes = get_be32(data + chunk_offset + 4);
361 graph->bloom_filter_settings->bits_per_entry = get_be32(data + chunk_offset + 8);
362 }
363 break;
364 }
365
366 if (chunk_repeated) {
367 error(_("commit-graph chunk id %08x appears multiple times"), chunk_id);
368 free(graph);
369 return NULL;
370 }
371
372 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
373 {
374 graph->num_commits = (chunk_offset - last_chunk_offset)
375 / graph->hash_len;
376 }
377
378 last_chunk_id = chunk_id;
379 last_chunk_offset = chunk_offset;
380 }
381
382 if (graph->chunk_bloom_indexes && graph->chunk_bloom_data) {
383 init_bloom_filters();
384 } else {
385 /* We need both the bloom chunks to exist together. Else ignore the data */
386 graph->chunk_bloom_indexes = NULL;
387 graph->chunk_bloom_data = NULL;
388 graph->bloom_filter_settings = NULL;
389 }
390
391 hashcpy(graph->oid.hash, graph->data + graph->data_len - graph->hash_len);
392
393 if (verify_commit_graph_lite(graph)) {
394 free(graph);
395 return NULL;
396 }
397
398 return graph;
399 }
400
401 static struct commit_graph *load_commit_graph_one(const char *graph_file,
402 struct object_directory *odb)
403 {
404
405 struct stat st;
406 int fd;
407 struct commit_graph *g;
408 int open_ok = open_commit_graph(graph_file, &fd, &st);
409
410 if (!open_ok)
411 return NULL;
412
413 g = load_commit_graph_one_fd_st(fd, &st, odb);
414
415 if (g)
416 g->filename = xstrdup(graph_file);
417
418 return g;
419 }
420
421 static struct commit_graph *load_commit_graph_v1(struct repository *r,
422 struct object_directory *odb)
423 {
424 char *graph_name = get_commit_graph_filename(odb);
425 struct commit_graph *g = load_commit_graph_one(graph_name, odb);
426 free(graph_name);
427
428 return g;
429 }
430
431 static int add_graph_to_chain(struct commit_graph *g,
432 struct commit_graph *chain,
433 struct object_id *oids,
434 int n)
435 {
436 struct commit_graph *cur_g = chain;
437
438 if (n && !g->chunk_base_graphs) {
439 warning(_("commit-graph has no base graphs chunk"));
440 return 0;
441 }
442
443 while (n) {
444 n--;
445
446 if (!cur_g ||
447 !oideq(&oids[n], &cur_g->oid) ||
448 !hasheq(oids[n].hash, g->chunk_base_graphs + g->hash_len * n)) {
449 warning(_("commit-graph chain does not match"));
450 return 0;
451 }
452
453 cur_g = cur_g->base_graph;
454 }
455
456 g->base_graph = chain;
457
458 if (chain)
459 g->num_commits_in_base = chain->num_commits + chain->num_commits_in_base;
460
461 return 1;
462 }
463
464 static struct commit_graph *load_commit_graph_chain(struct repository *r,
465 struct object_directory *odb)
466 {
467 struct commit_graph *graph_chain = NULL;
468 struct strbuf line = STRBUF_INIT;
469 struct stat st;
470 struct object_id *oids;
471 int i = 0, valid = 1, count;
472 char *chain_name = get_chain_filename(odb);
473 FILE *fp;
474 int stat_res;
475
476 fp = fopen(chain_name, "r");
477 stat_res = stat(chain_name, &st);
478 free(chain_name);
479
480 if (!fp ||
481 stat_res ||
482 st.st_size <= the_hash_algo->hexsz)
483 return NULL;
484
485 count = st.st_size / (the_hash_algo->hexsz + 1);
486 oids = xcalloc(count, sizeof(struct object_id));
487
488 prepare_alt_odb(r);
489
490 for (i = 0; i < count; i++) {
491 struct object_directory *odb;
492
493 if (strbuf_getline_lf(&line, fp) == EOF)
494 break;
495
496 if (get_oid_hex(line.buf, &oids[i])) {
497 warning(_("invalid commit-graph chain: line '%s' not a hash"),
498 line.buf);
499 valid = 0;
500 break;
501 }
502
503 valid = 0;
504 for (odb = r->objects->odb; odb; odb = odb->next) {
505 char *graph_name = get_split_graph_filename(odb, line.buf);
506 struct commit_graph *g = load_commit_graph_one(graph_name, odb);
507
508 free(graph_name);
509
510 if (g) {
511 if (add_graph_to_chain(g, graph_chain, oids, i)) {
512 graph_chain = g;
513 valid = 1;
514 }
515
516 break;
517 }
518 }
519
520 if (!valid) {
521 warning(_("unable to find all commit-graph files"));
522 break;
523 }
524 }
525
526 free(oids);
527 fclose(fp);
528 strbuf_release(&line);
529
530 return graph_chain;
531 }
532
533 struct commit_graph *read_commit_graph_one(struct repository *r,
534 struct object_directory *odb)
535 {
536 struct commit_graph *g = load_commit_graph_v1(r, odb);
537
538 if (!g)
539 g = load_commit_graph_chain(r, odb);
540
541 return g;
542 }
543
544 static void prepare_commit_graph_one(struct repository *r,
545 struct object_directory *odb)
546 {
547
548 if (r->objects->commit_graph)
549 return;
550
551 r->objects->commit_graph = read_commit_graph_one(r, odb);
552 }
553
554 /*
555 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
556 *
557 * On the first invocation, this function attempts to load the commit
558 * graph if the_repository is configured to have one.
559 */
560 static int prepare_commit_graph(struct repository *r)
561 {
562 struct object_directory *odb;
563
564 /*
565 * This must come before the "already attempted?" check below, because
566 * we want to disable even an already-loaded graph file.
567 */
568 if (r->commit_graph_disabled)
569 return 0;
570
571 if (r->objects->commit_graph_attempted)
572 return !!r->objects->commit_graph;
573 r->objects->commit_graph_attempted = 1;
574
575 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
576 die("dying as requested by the '%s' variable on commit-graph load!",
577 GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD);
578
579 prepare_repo_settings(r);
580
581 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
582 r->settings.core_commit_graph != 1)
583 /*
584 * This repository is not configured to use commit graphs, so
585 * do not load one. (But report commit_graph_attempted anyway
586 * so that commit graph loading is not attempted again for this
587 * repository.)
588 */
589 return 0;
590
591 if (!commit_graph_compatible(r))
592 return 0;
593
594 prepare_alt_odb(r);
595 for (odb = r->objects->odb;
596 !r->objects->commit_graph && odb;
597 odb = odb->next)
598 prepare_commit_graph_one(r, odb);
599 return !!r->objects->commit_graph;
600 }
601
602 int generation_numbers_enabled(struct repository *r)
603 {
604 uint32_t first_generation;
605 struct commit_graph *g;
606 if (!prepare_commit_graph(r))
607 return 0;
608
609 g = r->objects->commit_graph;
610
611 if (!g->num_commits)
612 return 0;
613
614 first_generation = get_be32(g->chunk_commit_data +
615 g->hash_len + 8) >> 2;
616
617 return !!first_generation;
618 }
619
620 static void close_commit_graph_one(struct commit_graph *g)
621 {
622 if (!g)
623 return;
624
625 close_commit_graph_one(g->base_graph);
626 free_commit_graph(g);
627 }
628
629 void close_commit_graph(struct raw_object_store *o)
630 {
631 close_commit_graph_one(o->commit_graph);
632 o->commit_graph = NULL;
633 }
634
635 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
636 {
637 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
638 g->chunk_oid_lookup, g->hash_len, pos);
639 }
640
641 static void load_oid_from_graph(struct commit_graph *g,
642 uint32_t pos,
643 struct object_id *oid)
644 {
645 uint32_t lex_index;
646
647 while (g && pos < g->num_commits_in_base)
648 g = g->base_graph;
649
650 if (!g)
651 BUG("NULL commit-graph");
652
653 if (pos >= g->num_commits + g->num_commits_in_base)
654 die(_("invalid commit position. commit-graph is likely corrupt"));
655
656 lex_index = pos - g->num_commits_in_base;
657
658 hashcpy(oid->hash, g->chunk_oid_lookup + g->hash_len * lex_index);
659 }
660
661 static struct commit_list **insert_parent_or_die(struct repository *r,
662 struct commit_graph *g,
663 uint32_t pos,
664 struct commit_list **pptr)
665 {
666 struct commit *c;
667 struct object_id oid;
668
669 if (pos >= g->num_commits + g->num_commits_in_base)
670 die("invalid parent position %"PRIu32, pos);
671
672 load_oid_from_graph(g, pos, &oid);
673 c = lookup_commit(r, &oid);
674 if (!c)
675 die(_("could not find commit %s"), oid_to_hex(&oid));
676 c->graph_pos = pos;
677 return &commit_list_insert(c, pptr)->next;
678 }
679
680 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
681 {
682 const unsigned char *commit_data;
683 uint32_t lex_index;
684
685 while (pos < g->num_commits_in_base)
686 g = g->base_graph;
687
688 lex_index = pos - g->num_commits_in_base;
689 commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * lex_index;
690 item->graph_pos = pos;
691 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
692 }
693
694 static inline void set_commit_tree(struct commit *c, struct tree *t)
695 {
696 c->maybe_tree = t;
697 }
698
699 static int fill_commit_in_graph(struct repository *r,
700 struct commit *item,
701 struct commit_graph *g, uint32_t pos)
702 {
703 uint32_t edge_value;
704 uint32_t *parent_data_ptr;
705 uint64_t date_low, date_high;
706 struct commit_list **pptr;
707 const unsigned char *commit_data;
708 uint32_t lex_index;
709
710 while (pos < g->num_commits_in_base)
711 g = g->base_graph;
712
713 if (pos >= g->num_commits + g->num_commits_in_base)
714 die(_("invalid commit position. commit-graph is likely corrupt"));
715
716 /*
717 * Store the "full" position, but then use the
718 * "local" position for the rest of the calculation.
719 */
720 item->graph_pos = pos;
721 lex_index = pos - g->num_commits_in_base;
722
723 commit_data = g->chunk_commit_data + (g->hash_len + 16) * lex_index;
724
725 item->object.parsed = 1;
726
727 set_commit_tree(item, NULL);
728
729 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
730 date_low = get_be32(commit_data + g->hash_len + 12);
731 item->date = (timestamp_t)((date_high << 32) | date_low);
732
733 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
734
735 pptr = &item->parents;
736
737 edge_value = get_be32(commit_data + g->hash_len);
738 if (edge_value == GRAPH_PARENT_NONE)
739 return 1;
740 pptr = insert_parent_or_die(r, g, edge_value, pptr);
741
742 edge_value = get_be32(commit_data + g->hash_len + 4);
743 if (edge_value == GRAPH_PARENT_NONE)
744 return 1;
745 if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
746 pptr = insert_parent_or_die(r, g, edge_value, pptr);
747 return 1;
748 }
749
750 parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
751 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
752 do {
753 edge_value = get_be32(parent_data_ptr);
754 pptr = insert_parent_or_die(r, g,
755 edge_value & GRAPH_EDGE_LAST_MASK,
756 pptr);
757 parent_data_ptr++;
758 } while (!(edge_value & GRAPH_LAST_EDGE));
759
760 return 1;
761 }
762
763 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
764 {
765 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
766 *pos = item->graph_pos;
767 return 1;
768 } else {
769 struct commit_graph *cur_g = g;
770 uint32_t lex_index;
771
772 while (cur_g && !bsearch_graph(cur_g, &(item->object.oid), &lex_index))
773 cur_g = cur_g->base_graph;
774
775 if (cur_g) {
776 *pos = lex_index + cur_g->num_commits_in_base;
777 return 1;
778 }
779
780 return 0;
781 }
782 }
783
784 static int parse_commit_in_graph_one(struct repository *r,
785 struct commit_graph *g,
786 struct commit *item)
787 {
788 uint32_t pos;
789
790 if (item->object.parsed)
791 return 1;
792
793 if (find_commit_in_graph(item, g, &pos))
794 return fill_commit_in_graph(r, item, g, pos);
795
796 return 0;
797 }
798
799 int parse_commit_in_graph(struct repository *r, struct commit *item)
800 {
801 if (!prepare_commit_graph(r))
802 return 0;
803 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
804 }
805
806 void load_commit_graph_info(struct repository *r, struct commit *item)
807 {
808 uint32_t pos;
809 if (!prepare_commit_graph(r))
810 return;
811 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
812 fill_commit_graph_info(item, r->objects->commit_graph, pos);
813 }
814
815 static struct tree *load_tree_for_commit(struct repository *r,
816 struct commit_graph *g,
817 struct commit *c)
818 {
819 struct object_id oid;
820 const unsigned char *commit_data;
821
822 while (c->graph_pos < g->num_commits_in_base)
823 g = g->base_graph;
824
825 commit_data = g->chunk_commit_data +
826 GRAPH_DATA_WIDTH * (c->graph_pos - g->num_commits_in_base);
827
828 hashcpy(oid.hash, commit_data);
829 set_commit_tree(c, lookup_tree(r, &oid));
830
831 return c->maybe_tree;
832 }
833
834 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
835 struct commit_graph *g,
836 const struct commit *c)
837 {
838 if (c->maybe_tree)
839 return c->maybe_tree;
840 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
841 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
842
843 return load_tree_for_commit(r, g, (struct commit *)c);
844 }
845
846 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
847 {
848 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
849 }
850
851 struct packed_commit_list {
852 struct commit **list;
853 int nr;
854 int alloc;
855 };
856
857 struct packed_oid_list {
858 struct object_id *list;
859 int nr;
860 int alloc;
861 };
862
863 struct write_commit_graph_context {
864 struct repository *r;
865 struct object_directory *odb;
866 char *graph_name;
867 struct packed_oid_list oids;
868 struct packed_commit_list commits;
869 int num_extra_edges;
870 unsigned long approx_nr_objects;
871 struct progress *progress;
872 int progress_done;
873 uint64_t progress_cnt;
874
875 char *base_graph_name;
876 int num_commit_graphs_before;
877 int num_commit_graphs_after;
878 char **commit_graph_filenames_before;
879 char **commit_graph_filenames_after;
880 char **commit_graph_hash_after;
881 uint32_t new_num_commits_in_base;
882 struct commit_graph *new_base_graph;
883
884 unsigned append:1,
885 report_progress:1,
886 split:1,
887 check_oids:1,
888 changed_paths:1,
889 order_by_pack:1;
890
891 const struct split_commit_graph_opts *split_opts;
892 size_t total_bloom_filter_data_size;
893 };
894
895 static void write_graph_chunk_fanout(struct hashfile *f,
896 struct write_commit_graph_context *ctx)
897 {
898 int i, count = 0;
899 struct commit **list = ctx->commits.list;
900
901 /*
902 * Write the first-level table (the list is sorted,
903 * but we use a 256-entry lookup to be able to avoid
904 * having to do eight extra binary search iterations).
905 */
906 for (i = 0; i < 256; i++) {
907 while (count < ctx->commits.nr) {
908 if ((*list)->object.oid.hash[0] != i)
909 break;
910 display_progress(ctx->progress, ++ctx->progress_cnt);
911 count++;
912 list++;
913 }
914
915 hashwrite_be32(f, count);
916 }
917 }
918
919 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
920 struct write_commit_graph_context *ctx)
921 {
922 struct commit **list = ctx->commits.list;
923 int count;
924 for (count = 0; count < ctx->commits.nr; count++, list++) {
925 display_progress(ctx->progress, ++ctx->progress_cnt);
926 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
927 }
928 }
929
930 static const unsigned char *commit_to_sha1(size_t index, void *table)
931 {
932 struct commit **commits = table;
933 return commits[index]->object.oid.hash;
934 }
935
936 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
937 struct write_commit_graph_context *ctx)
938 {
939 struct commit **list = ctx->commits.list;
940 struct commit **last = ctx->commits.list + ctx->commits.nr;
941 uint32_t num_extra_edges = 0;
942
943 while (list < last) {
944 struct commit_list *parent;
945 struct object_id *tree;
946 int edge_value;
947 uint32_t packedDate[2];
948 display_progress(ctx->progress, ++ctx->progress_cnt);
949
950 if (parse_commit_no_graph(*list))
951 die(_("unable to parse commit %s"),
952 oid_to_hex(&(*list)->object.oid));
953 tree = get_commit_tree_oid(*list);
954 hashwrite(f, tree->hash, hash_len);
955
956 parent = (*list)->parents;
957
958 if (!parent)
959 edge_value = GRAPH_PARENT_NONE;
960 else {
961 edge_value = sha1_pos(parent->item->object.oid.hash,
962 ctx->commits.list,
963 ctx->commits.nr,
964 commit_to_sha1);
965
966 if (edge_value >= 0)
967 edge_value += ctx->new_num_commits_in_base;
968 else {
969 uint32_t pos;
970 if (find_commit_in_graph(parent->item,
971 ctx->new_base_graph,
972 &pos))
973 edge_value = pos;
974 }
975
976 if (edge_value < 0)
977 BUG("missing parent %s for commit %s",
978 oid_to_hex(&parent->item->object.oid),
979 oid_to_hex(&(*list)->object.oid));
980 }
981
982 hashwrite_be32(f, edge_value);
983
984 if (parent)
985 parent = parent->next;
986
987 if (!parent)
988 edge_value = GRAPH_PARENT_NONE;
989 else if (parent->next)
990 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
991 else {
992 edge_value = sha1_pos(parent->item->object.oid.hash,
993 ctx->commits.list,
994 ctx->commits.nr,
995 commit_to_sha1);
996
997 if (edge_value >= 0)
998 edge_value += ctx->new_num_commits_in_base;
999 else {
1000 uint32_t pos;
1001 if (find_commit_in_graph(parent->item,
1002 ctx->new_base_graph,
1003 &pos))
1004 edge_value = pos;
1005 }
1006
1007 if (edge_value < 0)
1008 BUG("missing parent %s for commit %s",
1009 oid_to_hex(&parent->item->object.oid),
1010 oid_to_hex(&(*list)->object.oid));
1011 }
1012
1013 hashwrite_be32(f, edge_value);
1014
1015 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
1016 do {
1017 num_extra_edges++;
1018 parent = parent->next;
1019 } while (parent);
1020 }
1021
1022 if (sizeof((*list)->date) > 4)
1023 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
1024 else
1025 packedDate[0] = 0;
1026
1027 packedDate[0] |= htonl((*list)->generation << 2);
1028
1029 packedDate[1] = htonl((*list)->date);
1030 hashwrite(f, packedDate, 8);
1031
1032 list++;
1033 }
1034 }
1035
1036 static void write_graph_chunk_extra_edges(struct hashfile *f,
1037 struct write_commit_graph_context *ctx)
1038 {
1039 struct commit **list = ctx->commits.list;
1040 struct commit **last = ctx->commits.list + ctx->commits.nr;
1041 struct commit_list *parent;
1042
1043 while (list < last) {
1044 int num_parents = 0;
1045
1046 display_progress(ctx->progress, ++ctx->progress_cnt);
1047
1048 for (parent = (*list)->parents; num_parents < 3 && parent;
1049 parent = parent->next)
1050 num_parents++;
1051
1052 if (num_parents <= 2) {
1053 list++;
1054 continue;
1055 }
1056
1057 /* Since num_parents > 2, this initializer is safe. */
1058 for (parent = (*list)->parents->next; parent; parent = parent->next) {
1059 int edge_value = sha1_pos(parent->item->object.oid.hash,
1060 ctx->commits.list,
1061 ctx->commits.nr,
1062 commit_to_sha1);
1063
1064 if (edge_value >= 0)
1065 edge_value += ctx->new_num_commits_in_base;
1066 else {
1067 uint32_t pos;
1068 if (find_commit_in_graph(parent->item,
1069 ctx->new_base_graph,
1070 &pos))
1071 edge_value = pos;
1072 }
1073
1074 if (edge_value < 0)
1075 BUG("missing parent %s for commit %s",
1076 oid_to_hex(&parent->item->object.oid),
1077 oid_to_hex(&(*list)->object.oid));
1078 else if (!parent->next)
1079 edge_value |= GRAPH_LAST_EDGE;
1080
1081 hashwrite_be32(f, edge_value);
1082 }
1083
1084 list++;
1085 }
1086 }
1087
1088 static void write_graph_chunk_bloom_indexes(struct hashfile *f,
1089 struct write_commit_graph_context *ctx)
1090 {
1091 struct commit **list = ctx->commits.list;
1092 struct commit **last = ctx->commits.list + ctx->commits.nr;
1093 uint32_t cur_pos = 0;
1094 struct progress *progress = NULL;
1095 int i = 0;
1096
1097 if (ctx->report_progress)
1098 progress = start_delayed_progress(
1099 _("Writing changed paths Bloom filters index"),
1100 ctx->commits.nr);
1101
1102 while (list < last) {
1103 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list, 0);
1104 cur_pos += filter->len;
1105 display_progress(progress, ++i);
1106 hashwrite_be32(f, cur_pos);
1107 list++;
1108 }
1109
1110 stop_progress(&progress);
1111 }
1112
1113 static void write_graph_chunk_bloom_data(struct hashfile *f,
1114 struct write_commit_graph_context *ctx,
1115 const struct bloom_filter_settings *settings)
1116 {
1117 struct commit **list = ctx->commits.list;
1118 struct commit **last = ctx->commits.list + ctx->commits.nr;
1119 struct progress *progress = NULL;
1120 int i = 0;
1121
1122 if (ctx->report_progress)
1123 progress = start_delayed_progress(
1124 _("Writing changed paths Bloom filters data"),
1125 ctx->commits.nr);
1126
1127 hashwrite_be32(f, settings->hash_version);
1128 hashwrite_be32(f, settings->num_hashes);
1129 hashwrite_be32(f, settings->bits_per_entry);
1130
1131 while (list < last) {
1132 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list, 0);
1133 display_progress(progress, ++i);
1134 hashwrite(f, filter->data, filter->len * sizeof(unsigned char));
1135 list++;
1136 }
1137
1138 stop_progress(&progress);
1139 }
1140
1141 static int oid_compare(const void *_a, const void *_b)
1142 {
1143 const struct object_id *a = (const struct object_id *)_a;
1144 const struct object_id *b = (const struct object_id *)_b;
1145 return oidcmp(a, b);
1146 }
1147
1148 static int add_packed_commits(const struct object_id *oid,
1149 struct packed_git *pack,
1150 uint32_t pos,
1151 void *data)
1152 {
1153 struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
1154 enum object_type type;
1155 off_t offset = nth_packed_object_offset(pack, pos);
1156 struct object_info oi = OBJECT_INFO_INIT;
1157
1158 if (ctx->progress)
1159 display_progress(ctx->progress, ++ctx->progress_done);
1160
1161 oi.typep = &type;
1162 if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
1163 die(_("unable to get type of object %s"), oid_to_hex(oid));
1164
1165 if (type != OBJ_COMMIT)
1166 return 0;
1167
1168 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1169 oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
1170 ctx->oids.nr++;
1171
1172 set_commit_pos(ctx->r, oid);
1173
1174 return 0;
1175 }
1176
1177 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
1178 {
1179 struct commit_list *parent;
1180 for (parent = commit->parents; parent; parent = parent->next) {
1181 if (!(parent->item->object.flags & REACHABLE)) {
1182 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1183 oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
1184 ctx->oids.nr++;
1185 parent->item->object.flags |= REACHABLE;
1186 }
1187 }
1188 }
1189
1190 static void close_reachable(struct write_commit_graph_context *ctx)
1191 {
1192 int i;
1193 struct commit *commit;
1194
1195 if (ctx->report_progress)
1196 ctx->progress = start_delayed_progress(
1197 _("Loading known commits in commit graph"),
1198 ctx->oids.nr);
1199 for (i = 0; i < ctx->oids.nr; i++) {
1200 display_progress(ctx->progress, i + 1);
1201 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1202 if (commit)
1203 commit->object.flags |= REACHABLE;
1204 }
1205 stop_progress(&ctx->progress);
1206
1207 /*
1208 * As this loop runs, ctx->oids.nr may grow, but not more
1209 * than the number of missing commits in the reachable
1210 * closure.
1211 */
1212 if (ctx->report_progress)
1213 ctx->progress = start_delayed_progress(
1214 _("Expanding reachable commits in commit graph"),
1215 0);
1216 for (i = 0; i < ctx->oids.nr; i++) {
1217 display_progress(ctx->progress, i + 1);
1218 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1219
1220 if (!commit)
1221 continue;
1222 if (ctx->split) {
1223 if (!parse_commit(commit) &&
1224 commit->graph_pos == COMMIT_NOT_FROM_GRAPH)
1225 add_missing_parents(ctx, commit);
1226 } else if (!parse_commit_no_graph(commit))
1227 add_missing_parents(ctx, commit);
1228 }
1229 stop_progress(&ctx->progress);
1230
1231 if (ctx->report_progress)
1232 ctx->progress = start_delayed_progress(
1233 _("Clearing commit marks in commit graph"),
1234 ctx->oids.nr);
1235 for (i = 0; i < ctx->oids.nr; i++) {
1236 display_progress(ctx->progress, i + 1);
1237 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1238
1239 if (commit)
1240 commit->object.flags &= ~REACHABLE;
1241 }
1242 stop_progress(&ctx->progress);
1243 }
1244
1245 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1246 {
1247 int i;
1248 struct commit_list *list = NULL;
1249
1250 if (ctx->report_progress)
1251 ctx->progress = start_delayed_progress(
1252 _("Computing commit graph generation numbers"),
1253 ctx->commits.nr);
1254 for (i = 0; i < ctx->commits.nr; i++) {
1255 display_progress(ctx->progress, i + 1);
1256 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
1257 ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
1258 continue;
1259
1260 commit_list_insert(ctx->commits.list[i], &list);
1261 while (list) {
1262 struct commit *current = list->item;
1263 struct commit_list *parent;
1264 int all_parents_computed = 1;
1265 uint32_t max_generation = 0;
1266
1267 for (parent = current->parents; parent; parent = parent->next) {
1268 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
1269 parent->item->generation == GENERATION_NUMBER_ZERO) {
1270 all_parents_computed = 0;
1271 commit_list_insert(parent->item, &list);
1272 break;
1273 } else if (parent->item->generation > max_generation) {
1274 max_generation = parent->item->generation;
1275 }
1276 }
1277
1278 if (all_parents_computed) {
1279 current->generation = max_generation + 1;
1280 pop_commit(&list);
1281
1282 if (current->generation > GENERATION_NUMBER_MAX)
1283 current->generation = GENERATION_NUMBER_MAX;
1284 }
1285 }
1286 }
1287 stop_progress(&ctx->progress);
1288 }
1289
1290 static void compute_bloom_filters(struct write_commit_graph_context *ctx)
1291 {
1292 int i;
1293 struct progress *progress = NULL;
1294 struct commit **sorted_commits;
1295
1296 init_bloom_filters();
1297
1298 if (ctx->report_progress)
1299 progress = start_delayed_progress(
1300 _("Computing commit changed paths Bloom filters"),
1301 ctx->commits.nr);
1302
1303 ALLOC_ARRAY(sorted_commits, ctx->commits.nr);
1304 COPY_ARRAY(sorted_commits, ctx->commits.list, ctx->commits.nr);
1305
1306 if (ctx->order_by_pack)
1307 QSORT(sorted_commits, ctx->commits.nr, commit_pos_cmp);
1308 else
1309 QSORT(sorted_commits, ctx->commits.nr, commit_gen_cmp);
1310
1311 for (i = 0; i < ctx->commits.nr; i++) {
1312 struct commit *c = sorted_commits[i];
1313 struct bloom_filter *filter = get_bloom_filter(ctx->r, c, 1);
1314 ctx->total_bloom_filter_data_size += sizeof(unsigned char) * filter->len;
1315 display_progress(progress, i + 1);
1316 }
1317
1318 free(sorted_commits);
1319 stop_progress(&progress);
1320 }
1321
1322 static int add_ref_to_list(const char *refname,
1323 const struct object_id *oid,
1324 int flags, void *cb_data)
1325 {
1326 struct string_list *list = (struct string_list *)cb_data;
1327
1328 string_list_append(list, oid_to_hex(oid));
1329 return 0;
1330 }
1331
1332 int write_commit_graph_reachable(struct object_directory *odb,
1333 enum commit_graph_write_flags flags,
1334 const struct split_commit_graph_opts *split_opts)
1335 {
1336 struct string_list list = STRING_LIST_INIT_DUP;
1337 int result;
1338
1339 for_each_ref(add_ref_to_list, &list);
1340 result = write_commit_graph(odb, NULL, &list,
1341 flags, split_opts);
1342
1343 string_list_clear(&list, 0);
1344 return result;
1345 }
1346
1347 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
1348 struct string_list *pack_indexes)
1349 {
1350 uint32_t i;
1351 struct strbuf progress_title = STRBUF_INIT;
1352 struct strbuf packname = STRBUF_INIT;
1353 int dirlen;
1354
1355 strbuf_addf(&packname, "%s/pack/", ctx->odb->path);
1356 dirlen = packname.len;
1357 if (ctx->report_progress) {
1358 strbuf_addf(&progress_title,
1359 Q_("Finding commits for commit graph in %d pack",
1360 "Finding commits for commit graph in %d packs",
1361 pack_indexes->nr),
1362 pack_indexes->nr);
1363 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1364 ctx->progress_done = 0;
1365 }
1366 for (i = 0; i < pack_indexes->nr; i++) {
1367 struct packed_git *p;
1368 strbuf_setlen(&packname, dirlen);
1369 strbuf_addstr(&packname, pack_indexes->items[i].string);
1370 p = add_packed_git(packname.buf, packname.len, 1);
1371 if (!p) {
1372 error(_("error adding pack %s"), packname.buf);
1373 return -1;
1374 }
1375 if (open_pack_index(p)) {
1376 error(_("error opening index for %s"), packname.buf);
1377 return -1;
1378 }
1379 for_each_object_in_pack(p, add_packed_commits, ctx,
1380 FOR_EACH_OBJECT_PACK_ORDER);
1381 close_pack(p);
1382 free(p);
1383 }
1384
1385 stop_progress(&ctx->progress);
1386 strbuf_release(&progress_title);
1387 strbuf_release(&packname);
1388
1389 return 0;
1390 }
1391
1392 static int fill_oids_from_commit_hex(struct write_commit_graph_context *ctx,
1393 struct string_list *commit_hex)
1394 {
1395 uint32_t i;
1396 struct strbuf progress_title = STRBUF_INIT;
1397
1398 if (ctx->report_progress) {
1399 strbuf_addf(&progress_title,
1400 Q_("Finding commits for commit graph from %d ref",
1401 "Finding commits for commit graph from %d refs",
1402 commit_hex->nr),
1403 commit_hex->nr);
1404 ctx->progress = start_delayed_progress(
1405 progress_title.buf,
1406 commit_hex->nr);
1407 }
1408 for (i = 0; i < commit_hex->nr; i++) {
1409 const char *end;
1410 struct object_id oid;
1411 struct commit *result;
1412
1413 display_progress(ctx->progress, i + 1);
1414 if (!parse_oid_hex(commit_hex->items[i].string, &oid, &end) &&
1415 (result = lookup_commit_reference_gently(ctx->r, &oid, 1))) {
1416 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1417 oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
1418 ctx->oids.nr++;
1419 } else if (ctx->check_oids) {
1420 error(_("invalid commit object id: %s"),
1421 commit_hex->items[i].string);
1422 return -1;
1423 }
1424 }
1425 stop_progress(&ctx->progress);
1426 strbuf_release(&progress_title);
1427
1428 return 0;
1429 }
1430
1431 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1432 {
1433 if (ctx->report_progress)
1434 ctx->progress = start_delayed_progress(
1435 _("Finding commits for commit graph among packed objects"),
1436 ctx->approx_nr_objects);
1437 for_each_packed_object(add_packed_commits, ctx,
1438 FOR_EACH_OBJECT_PACK_ORDER);
1439 if (ctx->progress_done < ctx->approx_nr_objects)
1440 display_progress(ctx->progress, ctx->approx_nr_objects);
1441 stop_progress(&ctx->progress);
1442 }
1443
1444 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
1445 {
1446 uint32_t i, count_distinct = 1;
1447
1448 if (ctx->report_progress)
1449 ctx->progress = start_delayed_progress(
1450 _("Counting distinct commits in commit graph"),
1451 ctx->oids.nr);
1452 display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1453 QSORT(ctx->oids.list, ctx->oids.nr, oid_compare);
1454
1455 for (i = 1; i < ctx->oids.nr; i++) {
1456 display_progress(ctx->progress, i + 1);
1457 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i])) {
1458 if (ctx->split) {
1459 struct commit *c = lookup_commit(ctx->r, &ctx->oids.list[i]);
1460
1461 if (!c || c->graph_pos != COMMIT_NOT_FROM_GRAPH)
1462 continue;
1463 }
1464
1465 count_distinct++;
1466 }
1467 }
1468 stop_progress(&ctx->progress);
1469
1470 return count_distinct;
1471 }
1472
1473 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1474 {
1475 uint32_t i;
1476
1477 ctx->num_extra_edges = 0;
1478 if (ctx->report_progress)
1479 ctx->progress = start_delayed_progress(
1480 _("Finding extra edges in commit graph"),
1481 ctx->oids.nr);
1482 for (i = 0; i < ctx->oids.nr; i++) {
1483 unsigned int num_parents;
1484
1485 display_progress(ctx->progress, i + 1);
1486 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1487 continue;
1488
1489 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + 1, ctx->commits.alloc);
1490 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1491
1492 if (ctx->split &&
1493 ctx->commits.list[ctx->commits.nr]->graph_pos != COMMIT_NOT_FROM_GRAPH)
1494 continue;
1495
1496 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1497
1498 num_parents = commit_list_count(ctx->commits.list[ctx->commits.nr]->parents);
1499 if (num_parents > 2)
1500 ctx->num_extra_edges += num_parents - 1;
1501
1502 ctx->commits.nr++;
1503 }
1504 stop_progress(&ctx->progress);
1505 }
1506
1507 static int write_graph_chunk_base_1(struct hashfile *f,
1508 struct commit_graph *g)
1509 {
1510 int num = 0;
1511
1512 if (!g)
1513 return 0;
1514
1515 num = write_graph_chunk_base_1(f, g->base_graph);
1516 hashwrite(f, g->oid.hash, the_hash_algo->rawsz);
1517 return num + 1;
1518 }
1519
1520 static int write_graph_chunk_base(struct hashfile *f,
1521 struct write_commit_graph_context *ctx)
1522 {
1523 int num = write_graph_chunk_base_1(f, ctx->new_base_graph);
1524
1525 if (num != ctx->num_commit_graphs_after - 1) {
1526 error(_("failed to write correct number of base graph ids"));
1527 return -1;
1528 }
1529
1530 return 0;
1531 }
1532
1533 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1534 {
1535 uint32_t i;
1536 int fd;
1537 struct hashfile *f;
1538 struct lock_file lk = LOCK_INIT;
1539 uint32_t chunk_ids[MAX_NUM_CHUNKS + 1];
1540 uint64_t chunk_offsets[MAX_NUM_CHUNKS + 1];
1541 const unsigned hashsz = the_hash_algo->rawsz;
1542 struct strbuf progress_title = STRBUF_INIT;
1543 int num_chunks = 3;
1544 struct object_id file_hash;
1545 const struct bloom_filter_settings bloom_settings = DEFAULT_BLOOM_FILTER_SETTINGS;
1546
1547 if (ctx->split) {
1548 struct strbuf tmp_file = STRBUF_INIT;
1549
1550 strbuf_addf(&tmp_file,
1551 "%s/info/commit-graphs/tmp_graph_XXXXXX",
1552 ctx->odb->path);
1553 ctx->graph_name = strbuf_detach(&tmp_file, NULL);
1554 } else {
1555 ctx->graph_name = get_commit_graph_filename(ctx->odb);
1556 }
1557
1558 if (safe_create_leading_directories(ctx->graph_name)) {
1559 UNLEAK(ctx->graph_name);
1560 error(_("unable to create leading directories of %s"),
1561 ctx->graph_name);
1562 return -1;
1563 }
1564
1565 if (ctx->split) {
1566 char *lock_name = get_chain_filename(ctx->odb);
1567
1568 hold_lock_file_for_update(&lk, lock_name, LOCK_DIE_ON_ERROR);
1569
1570 fd = git_mkstemp_mode(ctx->graph_name, 0444);
1571 if (fd < 0) {
1572 error(_("unable to create '%s'"), ctx->graph_name);
1573 return -1;
1574 }
1575
1576 f = hashfd(fd, ctx->graph_name);
1577 } else {
1578 hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
1579 fd = lk.tempfile->fd;
1580 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1581 }
1582
1583 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1584 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1585 chunk_ids[2] = GRAPH_CHUNKID_DATA;
1586 if (ctx->num_extra_edges) {
1587 chunk_ids[num_chunks] = GRAPH_CHUNKID_EXTRAEDGES;
1588 num_chunks++;
1589 }
1590 if (ctx->changed_paths) {
1591 chunk_ids[num_chunks] = GRAPH_CHUNKID_BLOOMINDEXES;
1592 num_chunks++;
1593 chunk_ids[num_chunks] = GRAPH_CHUNKID_BLOOMDATA;
1594 num_chunks++;
1595 }
1596 if (ctx->num_commit_graphs_after > 1) {
1597 chunk_ids[num_chunks] = GRAPH_CHUNKID_BASE;
1598 num_chunks++;
1599 }
1600
1601 chunk_ids[num_chunks] = 0;
1602
1603 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1604 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1605 chunk_offsets[2] = chunk_offsets[1] + hashsz * ctx->commits.nr;
1606 chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * ctx->commits.nr;
1607
1608 num_chunks = 3;
1609 if (ctx->num_extra_edges) {
1610 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1611 4 * ctx->num_extra_edges;
1612 num_chunks++;
1613 }
1614 if (ctx->changed_paths) {
1615 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1616 sizeof(uint32_t) * ctx->commits.nr;
1617 num_chunks++;
1618
1619 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1620 sizeof(uint32_t) * 3 + ctx->total_bloom_filter_data_size;
1621 num_chunks++;
1622 }
1623 if (ctx->num_commit_graphs_after > 1) {
1624 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1625 hashsz * (ctx->num_commit_graphs_after - 1);
1626 num_chunks++;
1627 }
1628
1629 hashwrite_be32(f, GRAPH_SIGNATURE);
1630
1631 hashwrite_u8(f, GRAPH_VERSION);
1632 hashwrite_u8(f, oid_version());
1633 hashwrite_u8(f, num_chunks);
1634 hashwrite_u8(f, ctx->num_commit_graphs_after - 1);
1635
1636 for (i = 0; i <= num_chunks; i++) {
1637 uint32_t chunk_write[3];
1638
1639 chunk_write[0] = htonl(chunk_ids[i]);
1640 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1641 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1642 hashwrite(f, chunk_write, 12);
1643 }
1644
1645 if (ctx->report_progress) {
1646 strbuf_addf(&progress_title,
1647 Q_("Writing out commit graph in %d pass",
1648 "Writing out commit graph in %d passes",
1649 num_chunks),
1650 num_chunks);
1651 ctx->progress = start_delayed_progress(
1652 progress_title.buf,
1653 num_chunks * ctx->commits.nr);
1654 }
1655 write_graph_chunk_fanout(f, ctx);
1656 write_graph_chunk_oids(f, hashsz, ctx);
1657 write_graph_chunk_data(f, hashsz, ctx);
1658 if (ctx->num_extra_edges)
1659 write_graph_chunk_extra_edges(f, ctx);
1660 if (ctx->changed_paths) {
1661 write_graph_chunk_bloom_indexes(f, ctx);
1662 write_graph_chunk_bloom_data(f, ctx, &bloom_settings);
1663 }
1664 if (ctx->num_commit_graphs_after > 1 &&
1665 write_graph_chunk_base(f, ctx)) {
1666 return -1;
1667 }
1668 stop_progress(&ctx->progress);
1669 strbuf_release(&progress_title);
1670
1671 if (ctx->split && ctx->base_graph_name && ctx->num_commit_graphs_after > 1) {
1672 char *new_base_hash = xstrdup(oid_to_hex(&ctx->new_base_graph->oid));
1673 char *new_base_name = get_split_graph_filename(ctx->new_base_graph->odb, new_base_hash);
1674
1675 free(ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2]);
1676 free(ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2]);
1677 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2] = new_base_name;
1678 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2] = new_base_hash;
1679 }
1680
1681 close_commit_graph(ctx->r->objects);
1682 finalize_hashfile(f, file_hash.hash, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1683
1684 if (ctx->split) {
1685 FILE *chainf = fdopen_lock_file(&lk, "w");
1686 char *final_graph_name;
1687 int result;
1688
1689 close(fd);
1690
1691 if (!chainf) {
1692 error(_("unable to open commit-graph chain file"));
1693 return -1;
1694 }
1695
1696 if (ctx->base_graph_name) {
1697 const char *dest = ctx->commit_graph_filenames_after[
1698 ctx->num_commit_graphs_after - 2];
1699
1700 if (strcmp(ctx->base_graph_name, dest)) {
1701 result = rename(ctx->base_graph_name, dest);
1702
1703 if (result) {
1704 error(_("failed to rename base commit-graph file"));
1705 return -1;
1706 }
1707 }
1708 } else {
1709 char *graph_name = get_commit_graph_filename(ctx->odb);
1710 unlink(graph_name);
1711 }
1712
1713 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1] = xstrdup(oid_to_hex(&file_hash));
1714 final_graph_name = get_split_graph_filename(ctx->odb,
1715 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1]);
1716 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 1] = final_graph_name;
1717
1718 result = rename(ctx->graph_name, final_graph_name);
1719
1720 for (i = 0; i < ctx->num_commit_graphs_after; i++)
1721 fprintf(lk.tempfile->fp, "%s\n", ctx->commit_graph_hash_after[i]);
1722
1723 if (result) {
1724 error(_("failed to rename temporary commit-graph file"));
1725 return -1;
1726 }
1727 }
1728
1729 commit_lock_file(&lk);
1730
1731 return 0;
1732 }
1733
1734 static void split_graph_merge_strategy(struct write_commit_graph_context *ctx)
1735 {
1736 struct commit_graph *g;
1737 uint32_t num_commits;
1738 uint32_t i;
1739
1740 int max_commits = 0;
1741 int size_mult = 2;
1742
1743 if (ctx->split_opts) {
1744 max_commits = ctx->split_opts->max_commits;
1745
1746 if (ctx->split_opts->size_multiple)
1747 size_mult = ctx->split_opts->size_multiple;
1748 }
1749
1750 g = ctx->r->objects->commit_graph;
1751 num_commits = ctx->commits.nr;
1752 ctx->num_commit_graphs_after = ctx->num_commit_graphs_before + 1;
1753
1754 while (g && (g->num_commits <= size_mult * num_commits ||
1755 (max_commits && num_commits > max_commits))) {
1756 if (g->odb != ctx->odb)
1757 break;
1758
1759 num_commits += g->num_commits;
1760 g = g->base_graph;
1761
1762 ctx->num_commit_graphs_after--;
1763 }
1764
1765 ctx->new_base_graph = g;
1766
1767 if (ctx->num_commit_graphs_after == 2) {
1768 char *old_graph_name = get_commit_graph_filename(g->odb);
1769
1770 if (!strcmp(g->filename, old_graph_name) &&
1771 g->odb != ctx->odb) {
1772 ctx->num_commit_graphs_after = 1;
1773 ctx->new_base_graph = NULL;
1774 }
1775
1776 free(old_graph_name);
1777 }
1778
1779 ALLOC_ARRAY(ctx->commit_graph_filenames_after, ctx->num_commit_graphs_after);
1780 ALLOC_ARRAY(ctx->commit_graph_hash_after, ctx->num_commit_graphs_after);
1781
1782 for (i = 0; i < ctx->num_commit_graphs_after &&
1783 i < ctx->num_commit_graphs_before; i++)
1784 ctx->commit_graph_filenames_after[i] = xstrdup(ctx->commit_graph_filenames_before[i]);
1785
1786 i = ctx->num_commit_graphs_before - 1;
1787 g = ctx->r->objects->commit_graph;
1788
1789 while (g) {
1790 if (i < ctx->num_commit_graphs_after)
1791 ctx->commit_graph_hash_after[i] = xstrdup(oid_to_hex(&g->oid));
1792
1793 i--;
1794 g = g->base_graph;
1795 }
1796 }
1797
1798 static void merge_commit_graph(struct write_commit_graph_context *ctx,
1799 struct commit_graph *g)
1800 {
1801 uint32_t i;
1802 uint32_t offset = g->num_commits_in_base;
1803
1804 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + g->num_commits, ctx->commits.alloc);
1805
1806 for (i = 0; i < g->num_commits; i++) {
1807 struct object_id oid;
1808 struct commit *result;
1809
1810 display_progress(ctx->progress, i + 1);
1811
1812 load_oid_from_graph(g, i + offset, &oid);
1813
1814 /* only add commits if they still exist in the repo */
1815 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1816
1817 if (result) {
1818 ctx->commits.list[ctx->commits.nr] = result;
1819 ctx->commits.nr++;
1820 }
1821 }
1822 }
1823
1824 static int commit_compare(const void *_a, const void *_b)
1825 {
1826 const struct commit *a = *(const struct commit **)_a;
1827 const struct commit *b = *(const struct commit **)_b;
1828 return oidcmp(&a->object.oid, &b->object.oid);
1829 }
1830
1831 static void sort_and_scan_merged_commits(struct write_commit_graph_context *ctx)
1832 {
1833 uint32_t i;
1834
1835 if (ctx->report_progress)
1836 ctx->progress = start_delayed_progress(
1837 _("Scanning merged commits"),
1838 ctx->commits.nr);
1839
1840 QSORT(ctx->commits.list, ctx->commits.nr, commit_compare);
1841
1842 ctx->num_extra_edges = 0;
1843 for (i = 0; i < ctx->commits.nr; i++) {
1844 display_progress(ctx->progress, i);
1845
1846 if (i && oideq(&ctx->commits.list[i - 1]->object.oid,
1847 &ctx->commits.list[i]->object.oid)) {
1848 die(_("unexpected duplicate commit id %s"),
1849 oid_to_hex(&ctx->commits.list[i]->object.oid));
1850 } else {
1851 unsigned int num_parents;
1852
1853 num_parents = commit_list_count(ctx->commits.list[i]->parents);
1854 if (num_parents > 2)
1855 ctx->num_extra_edges += num_parents - 1;
1856 }
1857 }
1858
1859 stop_progress(&ctx->progress);
1860 }
1861
1862 static void merge_commit_graphs(struct write_commit_graph_context *ctx)
1863 {
1864 struct commit_graph *g = ctx->r->objects->commit_graph;
1865 uint32_t current_graph_number = ctx->num_commit_graphs_before;
1866
1867 while (g && current_graph_number >= ctx->num_commit_graphs_after) {
1868 current_graph_number--;
1869
1870 if (ctx->report_progress)
1871 ctx->progress = start_delayed_progress(_("Merging commit-graph"), 0);
1872
1873 merge_commit_graph(ctx, g);
1874 stop_progress(&ctx->progress);
1875
1876 g = g->base_graph;
1877 }
1878
1879 if (g) {
1880 ctx->new_base_graph = g;
1881 ctx->new_num_commits_in_base = g->num_commits + g->num_commits_in_base;
1882 }
1883
1884 if (ctx->new_base_graph)
1885 ctx->base_graph_name = xstrdup(ctx->new_base_graph->filename);
1886
1887 sort_and_scan_merged_commits(ctx);
1888 }
1889
1890 static void mark_commit_graphs(struct write_commit_graph_context *ctx)
1891 {
1892 uint32_t i;
1893 time_t now = time(NULL);
1894
1895 for (i = ctx->num_commit_graphs_after - 1; i < ctx->num_commit_graphs_before; i++) {
1896 struct stat st;
1897 struct utimbuf updated_time;
1898
1899 stat(ctx->commit_graph_filenames_before[i], &st);
1900
1901 updated_time.actime = st.st_atime;
1902 updated_time.modtime = now;
1903 utime(ctx->commit_graph_filenames_before[i], &updated_time);
1904 }
1905 }
1906
1907 static void expire_commit_graphs(struct write_commit_graph_context *ctx)
1908 {
1909 struct strbuf path = STRBUF_INIT;
1910 DIR *dir;
1911 struct dirent *de;
1912 size_t dirnamelen;
1913 timestamp_t expire_time = time(NULL);
1914
1915 if (ctx->split_opts && ctx->split_opts->expire_time)
1916 expire_time -= ctx->split_opts->expire_time;
1917 if (!ctx->split) {
1918 char *chain_file_name = get_chain_filename(ctx->odb);
1919 unlink(chain_file_name);
1920 free(chain_file_name);
1921 ctx->num_commit_graphs_after = 0;
1922 }
1923
1924 strbuf_addstr(&path, ctx->odb->path);
1925 strbuf_addstr(&path, "/info/commit-graphs");
1926 dir = opendir(path.buf);
1927
1928 if (!dir)
1929 goto out;
1930
1931 strbuf_addch(&path, '/');
1932 dirnamelen = path.len;
1933 while ((de = readdir(dir)) != NULL) {
1934 struct stat st;
1935 uint32_t i, found = 0;
1936
1937 strbuf_setlen(&path, dirnamelen);
1938 strbuf_addstr(&path, de->d_name);
1939
1940 stat(path.buf, &st);
1941
1942 if (st.st_mtime > expire_time)
1943 continue;
1944 if (path.len < 6 || strcmp(path.buf + path.len - 6, ".graph"))
1945 continue;
1946
1947 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
1948 if (!strcmp(ctx->commit_graph_filenames_after[i],
1949 path.buf)) {
1950 found = 1;
1951 break;
1952 }
1953 }
1954
1955 if (!found)
1956 unlink(path.buf);
1957 }
1958
1959 out:
1960 strbuf_release(&path);
1961 }
1962
1963 int write_commit_graph(struct object_directory *odb,
1964 struct string_list *pack_indexes,
1965 struct string_list *commit_hex,
1966 enum commit_graph_write_flags flags,
1967 const struct split_commit_graph_opts *split_opts)
1968 {
1969 struct write_commit_graph_context *ctx;
1970 uint32_t i, count_distinct = 0;
1971 int res = 0;
1972
1973 if (!commit_graph_compatible(the_repository))
1974 return 0;
1975
1976 ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
1977 ctx->r = the_repository;
1978 ctx->odb = odb;
1979 ctx->append = flags & COMMIT_GRAPH_WRITE_APPEND ? 1 : 0;
1980 ctx->report_progress = flags & COMMIT_GRAPH_WRITE_PROGRESS ? 1 : 0;
1981 ctx->split = flags & COMMIT_GRAPH_WRITE_SPLIT ? 1 : 0;
1982 ctx->check_oids = flags & COMMIT_GRAPH_WRITE_CHECK_OIDS ? 1 : 0;
1983 ctx->split_opts = split_opts;
1984 ctx->changed_paths = flags & COMMIT_GRAPH_WRITE_BLOOM_FILTERS ? 1 : 0;
1985 ctx->total_bloom_filter_data_size = 0;
1986
1987 if (ctx->split) {
1988 struct commit_graph *g;
1989 prepare_commit_graph(ctx->r);
1990
1991 g = ctx->r->objects->commit_graph;
1992
1993 while (g) {
1994 ctx->num_commit_graphs_before++;
1995 g = g->base_graph;
1996 }
1997
1998 if (ctx->num_commit_graphs_before) {
1999 ALLOC_ARRAY(ctx->commit_graph_filenames_before, ctx->num_commit_graphs_before);
2000 i = ctx->num_commit_graphs_before;
2001 g = ctx->r->objects->commit_graph;
2002
2003 while (g) {
2004 ctx->commit_graph_filenames_before[--i] = xstrdup(g->filename);
2005 g = g->base_graph;
2006 }
2007 }
2008 }
2009
2010 ctx->approx_nr_objects = approximate_object_count();
2011 ctx->oids.alloc = ctx->approx_nr_objects / 32;
2012
2013 if (ctx->split && split_opts && ctx->oids.alloc > split_opts->max_commits)
2014 ctx->oids.alloc = split_opts->max_commits;
2015
2016 if (ctx->append) {
2017 prepare_commit_graph_one(ctx->r, ctx->odb);
2018 if (ctx->r->objects->commit_graph)
2019 ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
2020 }
2021
2022 if (ctx->oids.alloc < 1024)
2023 ctx->oids.alloc = 1024;
2024 ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
2025
2026 if (ctx->append && ctx->r->objects->commit_graph) {
2027 struct commit_graph *g = ctx->r->objects->commit_graph;
2028 for (i = 0; i < g->num_commits; i++) {
2029 const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
2030 hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
2031 }
2032 }
2033
2034 if (pack_indexes) {
2035 ctx->order_by_pack = 1;
2036 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
2037 goto cleanup;
2038 }
2039
2040 if (commit_hex) {
2041 if ((res = fill_oids_from_commit_hex(ctx, commit_hex)))
2042 goto cleanup;
2043 }
2044
2045 if (!pack_indexes && !commit_hex) {
2046 ctx->order_by_pack = 1;
2047 fill_oids_from_all_packs(ctx);
2048 }
2049
2050 close_reachable(ctx);
2051
2052 count_distinct = count_distinct_commits(ctx);
2053
2054 if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
2055 error(_("the commit graph format cannot write %d commits"), count_distinct);
2056 res = -1;
2057 goto cleanup;
2058 }
2059
2060 ctx->commits.alloc = count_distinct;
2061 ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
2062
2063 copy_oids_to_commits(ctx);
2064
2065 if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
2066 error(_("too many commits to write graph"));
2067 res = -1;
2068 goto cleanup;
2069 }
2070
2071 if (!ctx->commits.nr)
2072 goto cleanup;
2073
2074 if (ctx->split) {
2075 split_graph_merge_strategy(ctx);
2076
2077 merge_commit_graphs(ctx);
2078 } else
2079 ctx->num_commit_graphs_after = 1;
2080
2081 compute_generation_numbers(ctx);
2082
2083 if (ctx->changed_paths)
2084 compute_bloom_filters(ctx);
2085
2086 res = write_commit_graph_file(ctx);
2087
2088 if (ctx->split)
2089 mark_commit_graphs(ctx);
2090
2091 expire_commit_graphs(ctx);
2092
2093 cleanup:
2094 free(ctx->graph_name);
2095 free(ctx->commits.list);
2096 free(ctx->oids.list);
2097
2098 if (ctx->commit_graph_filenames_after) {
2099 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
2100 free(ctx->commit_graph_filenames_after[i]);
2101 free(ctx->commit_graph_hash_after[i]);
2102 }
2103
2104 for (i = 0; i < ctx->num_commit_graphs_before; i++)
2105 free(ctx->commit_graph_filenames_before[i]);
2106
2107 free(ctx->commit_graph_filenames_after);
2108 free(ctx->commit_graph_filenames_before);
2109 free(ctx->commit_graph_hash_after);
2110 }
2111
2112 free(ctx);
2113
2114 return res;
2115 }
2116
2117 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
2118 static int verify_commit_graph_error;
2119
2120 static void graph_report(const char *fmt, ...)
2121 {
2122 va_list ap;
2123
2124 verify_commit_graph_error = 1;
2125 va_start(ap, fmt);
2126 vfprintf(stderr, fmt, ap);
2127 fprintf(stderr, "\n");
2128 va_end(ap);
2129 }
2130
2131 #define GENERATION_ZERO_EXISTS 1
2132 #define GENERATION_NUMBER_EXISTS 2
2133
2134 int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
2135 {
2136 uint32_t i, cur_fanout_pos = 0;
2137 struct object_id prev_oid, cur_oid, checksum;
2138 int generation_zero = 0;
2139 struct hashfile *f;
2140 int devnull;
2141 struct progress *progress = NULL;
2142 int local_error = 0;
2143
2144 if (!g) {
2145 graph_report("no commit-graph file loaded");
2146 return 1;
2147 }
2148
2149 verify_commit_graph_error = verify_commit_graph_lite(g);
2150 if (verify_commit_graph_error)
2151 return verify_commit_graph_error;
2152
2153 devnull = open("/dev/null", O_WRONLY);
2154 f = hashfd(devnull, NULL);
2155 hashwrite(f, g->data, g->data_len - g->hash_len);
2156 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
2157 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
2158 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
2159 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
2160 }
2161
2162 for (i = 0; i < g->num_commits; i++) {
2163 struct commit *graph_commit;
2164
2165 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2166
2167 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
2168 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
2169 oid_to_hex(&prev_oid),
2170 oid_to_hex(&cur_oid));
2171
2172 oidcpy(&prev_oid, &cur_oid);
2173
2174 while (cur_oid.hash[0] > cur_fanout_pos) {
2175 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2176
2177 if (i != fanout_value)
2178 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2179 cur_fanout_pos, fanout_value, i);
2180 cur_fanout_pos++;
2181 }
2182
2183 graph_commit = lookup_commit(r, &cur_oid);
2184 if (!parse_commit_in_graph_one(r, g, graph_commit))
2185 graph_report(_("failed to parse commit %s from commit-graph"),
2186 oid_to_hex(&cur_oid));
2187 }
2188
2189 while (cur_fanout_pos < 256) {
2190 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2191
2192 if (g->num_commits != fanout_value)
2193 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2194 cur_fanout_pos, fanout_value, i);
2195
2196 cur_fanout_pos++;
2197 }
2198
2199 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
2200 return verify_commit_graph_error;
2201
2202 if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
2203 progress = start_progress(_("Verifying commits in commit graph"),
2204 g->num_commits);
2205
2206 for (i = 0; i < g->num_commits; i++) {
2207 struct commit *graph_commit, *odb_commit;
2208 struct commit_list *graph_parents, *odb_parents;
2209 uint32_t max_generation = 0;
2210
2211 display_progress(progress, i + 1);
2212 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2213
2214 graph_commit = lookup_commit(r, &cur_oid);
2215 odb_commit = (struct commit *)create_object(r, &cur_oid, alloc_commit_node(r));
2216 if (parse_commit_internal(odb_commit, 0, 0)) {
2217 graph_report(_("failed to parse commit %s from object database for commit-graph"),
2218 oid_to_hex(&cur_oid));
2219 continue;
2220 }
2221
2222 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
2223 get_commit_tree_oid(odb_commit)))
2224 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2225 oid_to_hex(&cur_oid),
2226 oid_to_hex(get_commit_tree_oid(graph_commit)),
2227 oid_to_hex(get_commit_tree_oid(odb_commit)));
2228
2229 graph_parents = graph_commit->parents;
2230 odb_parents = odb_commit->parents;
2231
2232 while (graph_parents) {
2233 if (odb_parents == NULL) {
2234 graph_report(_("commit-graph parent list for commit %s is too long"),
2235 oid_to_hex(&cur_oid));
2236 break;
2237 }
2238
2239 /* parse parent in case it is in a base graph */
2240 parse_commit_in_graph_one(r, g, graph_parents->item);
2241
2242 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
2243 graph_report(_("commit-graph parent for %s is %s != %s"),
2244 oid_to_hex(&cur_oid),
2245 oid_to_hex(&graph_parents->item->object.oid),
2246 oid_to_hex(&odb_parents->item->object.oid));
2247
2248 if (graph_parents->item->generation > max_generation)
2249 max_generation = graph_parents->item->generation;
2250
2251 graph_parents = graph_parents->next;
2252 odb_parents = odb_parents->next;
2253 }
2254
2255 if (odb_parents != NULL)
2256 graph_report(_("commit-graph parent list for commit %s terminates early"),
2257 oid_to_hex(&cur_oid));
2258
2259 if (!graph_commit->generation) {
2260 if (generation_zero == GENERATION_NUMBER_EXISTS)
2261 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2262 oid_to_hex(&cur_oid));
2263 generation_zero = GENERATION_ZERO_EXISTS;
2264 } else if (generation_zero == GENERATION_ZERO_EXISTS)
2265 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2266 oid_to_hex(&cur_oid));
2267
2268 if (generation_zero == GENERATION_ZERO_EXISTS)
2269 continue;
2270
2271 /*
2272 * If one of our parents has generation GENERATION_NUMBER_MAX, then
2273 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
2274 * extra logic in the following condition.
2275 */
2276 if (max_generation == GENERATION_NUMBER_MAX)
2277 max_generation--;
2278
2279 if (graph_commit->generation != max_generation + 1)
2280 graph_report(_("commit-graph generation for commit %s is %u != %u"),
2281 oid_to_hex(&cur_oid),
2282 graph_commit->generation,
2283 max_generation + 1);
2284
2285 if (graph_commit->date != odb_commit->date)
2286 graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
2287 oid_to_hex(&cur_oid),
2288 graph_commit->date,
2289 odb_commit->date);
2290 }
2291 stop_progress(&progress);
2292
2293 local_error = verify_commit_graph_error;
2294
2295 if (!(flags & COMMIT_GRAPH_VERIFY_SHALLOW) && g->base_graph)
2296 local_error |= verify_commit_graph(r, g->base_graph, flags);
2297
2298 return local_error;
2299 }
2300
2301 void free_commit_graph(struct commit_graph *g)
2302 {
2303 if (!g)
2304 return;
2305 if (g->graph_fd >= 0) {
2306 munmap((void *)g->data, g->data_len);
2307 g->data = NULL;
2308 close(g->graph_fd);
2309 }
2310 free(g->filename);
2311 free(g->bloom_filter_settings);
2312 free(g);
2313 }
2314
2315 void disable_commit_graph(struct repository *r)
2316 {
2317 r->commit_graph_disabled = 1;
2318 }
The Best VCS Around