?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
revision: complicated pathspecs disable filters
[git.git] / commit-graph.c
blob77668629e27dbbae783cfd2364fee78d41bf446d
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 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
23 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
24 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
25 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
26 #define GRAPH_CHUNKID_EXTRAEDGES 0x45444745 /* "EDGE" */
27 #define GRAPH_CHUNKID_BLOOMINDEXES 0x42494458 /* "BIDX" */
28 #define GRAPH_CHUNKID_BLOOMDATA 0x42444154 /* "BDAT" */
29 #define GRAPH_CHUNKID_BASE 0x42415345 /* "BASE" */
30 #define MAX_NUM_CHUNKS 7
31
32 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
33
34 #define GRAPH_VERSION_1 0x1
35 #define GRAPH_VERSION GRAPH_VERSION_1
36
37 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
38 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
39 #define GRAPH_PARENT_NONE 0x70000000
40
41 #define GRAPH_LAST_EDGE 0x80000000
42
43 #define GRAPH_HEADER_SIZE 8
44 #define GRAPH_FANOUT_SIZE (4 * 256)
45 #define GRAPH_CHUNKLOOKUP_WIDTH 12
46 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
47 + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
48
49 /* Remember to update object flag allocation in object.h */
50 #define REACHABLE (1u<<15)
51
52 /* Keep track of the order in which commits are added to our list. */
53 define_commit_slab(commit_pos, int);
54 static struct commit_pos commit_pos = COMMIT_SLAB_INIT(1, commit_pos);
55
56 static void set_commit_pos(struct repository *r, const struct object_id *oid)
57 {
58 static int32_t max_pos;
59 struct commit *commit = lookup_commit(r, oid);
60
61 if (!commit)
62 return; /* should never happen, but be lenient */
63
64 *commit_pos_at(&commit_pos, commit) = max_pos++;
65 }
66
67 static int commit_pos_cmp(const void *va, const void *vb)
68 {
69 const struct commit *a = *(const struct commit **)va;
70 const struct commit *b = *(const struct commit **)vb;
71 return commit_pos_at(&commit_pos, a) -
72 commit_pos_at(&commit_pos, b);
73 }
74
75 static int commit_gen_cmp(const void *va, const void *vb)
76 {
77 const struct commit *a = *(const struct commit **)va;
78 const struct commit *b = *(const struct commit **)vb;
79
80 /* lower generation commits first */
81 if (a->generation < b->generation)
82 return -1;
83 else if (a->generation > b->generation)
84 return 1;
85
86 /* use date as a heuristic when generations are equal */
87 if (a->date < b->date)
88 return -1;
89 else if (a->date > b->date)
90 return 1;
91 return 0;
92 }
93
94 char *get_commit_graph_filename(struct object_directory *obj_dir)
95 {
96 return xstrfmt("%s/info/commit-graph", obj_dir->path);
97 }
98
99 static char *get_split_graph_filename(struct object_directory *odb,
100 const char *oid_hex)
101 {
102 return xstrfmt("%s/info/commit-graphs/graph-%s.graph", odb->path,
103 oid_hex);
104 }
105
106 static char *get_chain_filename(struct object_directory *odb)
107 {
108 return xstrfmt("%s/info/commit-graphs/commit-graph-chain", odb->path);
109 }
110
111 static uint8_t oid_version(void)
112 {
113 return 1;
114 }
115
116 static struct commit_graph *alloc_commit_graph(void)
117 {
118 struct commit_graph *g = xcalloc(1, sizeof(*g));
119 g->graph_fd = -1;
120
121 return g;
122 }
123
124 extern int read_replace_refs;
125
126 static int commit_graph_compatible(struct repository *r)
127 {
128 if (!r->gitdir)
129 return 0;
130
131 if (read_replace_refs) {
132 prepare_replace_object(r);
133 if (hashmap_get_size(&r->objects->replace_map->map))
134 return 0;
135 }
136
137 prepare_commit_graft(r);
138 if (r->parsed_objects && r->parsed_objects->grafts_nr)
139 return 0;
140 if (is_repository_shallow(r))
141 return 0;
142
143 return 1;
144 }
145
146 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
147 {
148 *fd = git_open(graph_file);
149 if (*fd < 0)
150 return 0;
151 if (fstat(*fd, st)) {
152 close(*fd);
153 return 0;
154 }
155 return 1;
156 }
157
158 struct commit_graph *load_commit_graph_one_fd_st(int fd, struct stat *st,
159 struct object_directory *odb)
160 {
161 void *graph_map;
162 size_t graph_size;
163 struct commit_graph *ret;
164
165 graph_size = xsize_t(st->st_size);
166
167 if (graph_size < GRAPH_MIN_SIZE) {
168 close(fd);
169 error(_("commit-graph file is too small"));
170 return NULL;
171 }
172 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
173 ret = parse_commit_graph(graph_map, fd, graph_size);
174
175 if (ret)
176 ret->odb = odb;
177 else {
178 munmap(graph_map, graph_size);
179 close(fd);
180 }
181
182 return ret;
183 }
184
185 static int verify_commit_graph_lite(struct commit_graph *g)
186 {
187 /*
188 * Basic validation shared between parse_commit_graph()
189 * which'll be called every time the graph is used, and the
190 * much more expensive verify_commit_graph() used by
191 * "commit-graph verify".
192 *
193 * There should only be very basic checks here to ensure that
194 * we don't e.g. segfault in fill_commit_in_graph(), but
195 * because this is a very hot codepath nothing that e.g. loops
196 * over g->num_commits, or runs a checksum on the commit-graph
197 * itself.
198 */
199 if (!g->chunk_oid_fanout) {
200 error("commit-graph is missing the OID Fanout chunk");
201 return 1;
202 }
203 if (!g->chunk_oid_lookup) {
204 error("commit-graph is missing the OID Lookup chunk");
205 return 1;
206 }
207 if (!g->chunk_commit_data) {
208 error("commit-graph is missing the Commit Data chunk");
209 return 1;
210 }
211
212 return 0;
213 }
214
215 struct commit_graph *parse_commit_graph(void *graph_map, int fd,
216 size_t graph_size)
217 {
218 const unsigned char *data, *chunk_lookup;
219 uint32_t i;
220 struct commit_graph *graph;
221 uint64_t last_chunk_offset;
222 uint32_t last_chunk_id;
223 uint32_t graph_signature;
224 unsigned char graph_version, hash_version;
225
226 if (!graph_map)
227 return NULL;
228
229 if (graph_size < GRAPH_MIN_SIZE)
230 return NULL;
231
232 data = (const unsigned char *)graph_map;
233
234 graph_signature = get_be32(data);
235 if (graph_signature != GRAPH_SIGNATURE) {
236 error(_("commit-graph signature %X does not match signature %X"),
237 graph_signature, GRAPH_SIGNATURE);
238 return NULL;
239 }
240
241 graph_version = *(unsigned char*)(data + 4);
242 if (graph_version != GRAPH_VERSION) {
243 error(_("commit-graph version %X does not match version %X"),
244 graph_version, GRAPH_VERSION);
245 return NULL;
246 }
247
248 hash_version = *(unsigned char*)(data + 5);
249 if (hash_version != oid_version()) {
250 error(_("commit-graph hash version %X does not match version %X"),
251 hash_version, oid_version());
252 return NULL;
253 }
254
255 graph = alloc_commit_graph();
256
257 graph->hash_len = the_hash_algo->rawsz;
258 graph->num_chunks = *(unsigned char*)(data + 6);
259 graph->graph_fd = fd;
260 graph->data = graph_map;
261 graph->data_len = graph_size;
262
263 last_chunk_id = 0;
264 last_chunk_offset = 8;
265 chunk_lookup = data + 8;
266 for (i = 0; i < graph->num_chunks; i++) {
267 uint32_t chunk_id;
268 uint64_t chunk_offset;
269 int chunk_repeated = 0;
270
271 if (data + graph_size - chunk_lookup <
272 GRAPH_CHUNKLOOKUP_WIDTH) {
273 error(_("commit-graph chunk lookup table entry missing; file may be incomplete"));
274 free(graph);
275 return NULL;
276 }
277
278 chunk_id = get_be32(chunk_lookup + 0);
279 chunk_offset = get_be64(chunk_lookup + 4);
280
281 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
282
283 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
284 error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
285 (uint32_t)chunk_offset);
286 free(graph);
287 return NULL;
288 }
289
290 switch (chunk_id) {
291 case GRAPH_CHUNKID_OIDFANOUT:
292 if (graph->chunk_oid_fanout)
293 chunk_repeated = 1;
294 else
295 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
296 break;
297
298 case GRAPH_CHUNKID_OIDLOOKUP:
299 if (graph->chunk_oid_lookup)
300 chunk_repeated = 1;
301 else
302 graph->chunk_oid_lookup = data + chunk_offset;
303 break;
304
305 case GRAPH_CHUNKID_DATA:
306 if (graph->chunk_commit_data)
307 chunk_repeated = 1;
308 else
309 graph->chunk_commit_data = data + chunk_offset;
310 break;
311
312 case GRAPH_CHUNKID_EXTRAEDGES:
313 if (graph->chunk_extra_edges)
314 chunk_repeated = 1;
315 else
316 graph->chunk_extra_edges = data + chunk_offset;
317 break;
318
319 case GRAPH_CHUNKID_BASE:
320 if (graph->chunk_base_graphs)
321 chunk_repeated = 1;
322 else
323 graph->chunk_base_graphs = data + chunk_offset;
324 break;
325
326 case GRAPH_CHUNKID_BLOOMINDEXES:
327 if (graph->chunk_bloom_indexes)
328 chunk_repeated = 1;
329 else
330 graph->chunk_bloom_indexes = data + chunk_offset;
331 break;
332
333 case GRAPH_CHUNKID_BLOOMDATA:
334 if (graph->chunk_bloom_data)
335 chunk_repeated = 1;
336 else {
337 uint32_t hash_version;
338 graph->chunk_bloom_data = data + chunk_offset;
339 hash_version = get_be32(data + chunk_offset);
340
341 if (hash_version != 1)
342 break;
343
344 graph->bloom_filter_settings = xmalloc(sizeof(struct bloom_filter_settings));
345 graph->bloom_filter_settings->hash_version = hash_version;
346 graph->bloom_filter_settings->num_hashes = get_be32(data + chunk_offset + 4);
347 graph->bloom_filter_settings->bits_per_entry = get_be32(data + chunk_offset + 8);
348 }
349 break;
350 }
351
352 if (chunk_repeated) {
353 error(_("commit-graph chunk id %08x appears multiple times"), chunk_id);
354 free(graph);
355 return NULL;
356 }
357
358 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
359 {
360 graph->num_commits = (chunk_offset - last_chunk_offset)
361 / graph->hash_len;
362 }
363
364 last_chunk_id = chunk_id;
365 last_chunk_offset = chunk_offset;
366 }
367
368 if (graph->chunk_bloom_indexes && graph->chunk_bloom_data) {
369 init_bloom_filters();
370 } else {
371 /* We need both the bloom chunks to exist together. Else ignore the data */
372 graph->chunk_bloom_indexes = NULL;
373 graph->chunk_bloom_data = NULL;
374 graph->bloom_filter_settings = NULL;
375 }
376
377 hashcpy(graph->oid.hash, graph->data + graph->data_len - graph->hash_len);
378
379 if (verify_commit_graph_lite(graph)) {
380 free(graph);
381 return NULL;
382 }
383
384 return graph;
385 }
386
387 static struct commit_graph *load_commit_graph_one(const char *graph_file,
388 struct object_directory *odb)
389 {
390
391 struct stat st;
392 int fd;
393 struct commit_graph *g;
394 int open_ok = open_commit_graph(graph_file, &fd, &st);
395
396 if (!open_ok)
397 return NULL;
398
399 g = load_commit_graph_one_fd_st(fd, &st, odb);
400
401 if (g)
402 g->filename = xstrdup(graph_file);
403
404 return g;
405 }
406
407 static struct commit_graph *load_commit_graph_v1(struct repository *r,
408 struct object_directory *odb)
409 {
410 char *graph_name = get_commit_graph_filename(odb);
411 struct commit_graph *g = load_commit_graph_one(graph_name, odb);
412 free(graph_name);
413
414 return g;
415 }
416
417 static int add_graph_to_chain(struct commit_graph *g,
418 struct commit_graph *chain,
419 struct object_id *oids,
420 int n)
421 {
422 struct commit_graph *cur_g = chain;
423
424 if (n && !g->chunk_base_graphs) {
425 warning(_("commit-graph has no base graphs chunk"));
426 return 0;
427 }
428
429 while (n) {
430 n--;
431
432 if (!cur_g ||
433 !oideq(&oids[n], &cur_g->oid) ||
434 !hasheq(oids[n].hash, g->chunk_base_graphs + g->hash_len * n)) {
435 warning(_("commit-graph chain does not match"));
436 return 0;
437 }
438
439 cur_g = cur_g->base_graph;
440 }
441
442 g->base_graph = chain;
443
444 if (chain)
445 g->num_commits_in_base = chain->num_commits + chain->num_commits_in_base;
446
447 return 1;
448 }
449
450 static struct commit_graph *load_commit_graph_chain(struct repository *r,
451 struct object_directory *odb)
452 {
453 struct commit_graph *graph_chain = NULL;
454 struct strbuf line = STRBUF_INIT;
455 struct stat st;
456 struct object_id *oids;
457 int i = 0, valid = 1, count;
458 char *chain_name = get_chain_filename(odb);
459 FILE *fp;
460 int stat_res;
461
462 fp = fopen(chain_name, "r");
463 stat_res = stat(chain_name, &st);
464 free(chain_name);
465
466 if (!fp ||
467 stat_res ||
468 st.st_size <= the_hash_algo->hexsz)
469 return NULL;
470
471 count = st.st_size / (the_hash_algo->hexsz + 1);
472 oids = xcalloc(count, sizeof(struct object_id));
473
474 prepare_alt_odb(r);
475
476 for (i = 0; i < count; i++) {
477 struct object_directory *odb;
478
479 if (strbuf_getline_lf(&line, fp) == EOF)
480 break;
481
482 if (get_oid_hex(line.buf, &oids[i])) {
483 warning(_("invalid commit-graph chain: line '%s' not a hash"),
484 line.buf);
485 valid = 0;
486 break;
487 }
488
489 valid = 0;
490 for (odb = r->objects->odb; odb; odb = odb->next) {
491 char *graph_name = get_split_graph_filename(odb, line.buf);
492 struct commit_graph *g = load_commit_graph_one(graph_name, odb);
493
494 free(graph_name);
495
496 if (g) {
497 if (add_graph_to_chain(g, graph_chain, oids, i)) {
498 graph_chain = g;
499 valid = 1;
500 }
501
502 break;
503 }
504 }
505
506 if (!valid) {
507 warning(_("unable to find all commit-graph files"));
508 break;
509 }
510 }
511
512 free(oids);
513 fclose(fp);
514 strbuf_release(&line);
515
516 return graph_chain;
517 }
518
519 struct commit_graph *read_commit_graph_one(struct repository *r,
520 struct object_directory *odb)
521 {
522 struct commit_graph *g = load_commit_graph_v1(r, odb);
523
524 if (!g)
525 g = load_commit_graph_chain(r, odb);
526
527 return g;
528 }
529
530 static void prepare_commit_graph_one(struct repository *r,
531 struct object_directory *odb)
532 {
533
534 if (r->objects->commit_graph)
535 return;
536
537 r->objects->commit_graph = read_commit_graph_one(r, odb);
538 }
539
540 /*
541 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
542 *
543 * On the first invocation, this function attempts to load the commit
544 * graph if the_repository is configured to have one.
545 */
546 static int prepare_commit_graph(struct repository *r)
547 {
548 struct object_directory *odb;
549
550 /*
551 * This must come before the "already attempted?" check below, because
552 * we want to disable even an already-loaded graph file.
553 */
554 if (r->commit_graph_disabled)
555 return 0;
556
557 if (r->objects->commit_graph_attempted)
558 return !!r->objects->commit_graph;
559 r->objects->commit_graph_attempted = 1;
560
561 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
562 die("dying as requested by the '%s' variable on commit-graph load!",
563 GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD);
564
565 prepare_repo_settings(r);
566
567 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
568 r->settings.core_commit_graph != 1)
569 /*
570 * This repository is not configured to use commit graphs, so
571 * do not load one. (But report commit_graph_attempted anyway
572 * so that commit graph loading is not attempted again for this
573 * repository.)
574 */
575 return 0;
576
577 if (!commit_graph_compatible(r))
578 return 0;
579
580 prepare_alt_odb(r);
581 for (odb = r->objects->odb;
582 !r->objects->commit_graph && odb;
583 odb = odb->next)
584 prepare_commit_graph_one(r, odb);
585 return !!r->objects->commit_graph;
586 }
587
588 int generation_numbers_enabled(struct repository *r)
589 {
590 uint32_t first_generation;
591 struct commit_graph *g;
592 if (!prepare_commit_graph(r))
593 return 0;
594
595 g = r->objects->commit_graph;
596
597 if (!g->num_commits)
598 return 0;
599
600 first_generation = get_be32(g->chunk_commit_data +
601 g->hash_len + 8) >> 2;
602
603 return !!first_generation;
604 }
605
606 static void close_commit_graph_one(struct commit_graph *g)
607 {
608 if (!g)
609 return;
610
611 close_commit_graph_one(g->base_graph);
612 free_commit_graph(g);
613 }
614
615 void close_commit_graph(struct raw_object_store *o)
616 {
617 close_commit_graph_one(o->commit_graph);
618 o->commit_graph = NULL;
619 }
620
621 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
622 {
623 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
624 g->chunk_oid_lookup, g->hash_len, pos);
625 }
626
627 static void load_oid_from_graph(struct commit_graph *g,
628 uint32_t pos,
629 struct object_id *oid)
630 {
631 uint32_t lex_index;
632
633 while (g && pos < g->num_commits_in_base)
634 g = g->base_graph;
635
636 if (!g)
637 BUG("NULL commit-graph");
638
639 if (pos >= g->num_commits + g->num_commits_in_base)
640 die(_("invalid commit position. commit-graph is likely corrupt"));
641
642 lex_index = pos - g->num_commits_in_base;
643
644 hashcpy(oid->hash, g->chunk_oid_lookup + g->hash_len * lex_index);
645 }
646
647 static struct commit_list **insert_parent_or_die(struct repository *r,
648 struct commit_graph *g,
649 uint32_t pos,
650 struct commit_list **pptr)
651 {
652 struct commit *c;
653 struct object_id oid;
654
655 if (pos >= g->num_commits + g->num_commits_in_base)
656 die("invalid parent position %"PRIu32, pos);
657
658 load_oid_from_graph(g, pos, &oid);
659 c = lookup_commit(r, &oid);
660 if (!c)
661 die(_("could not find commit %s"), oid_to_hex(&oid));
662 c->graph_pos = pos;
663 return &commit_list_insert(c, pptr)->next;
664 }
665
666 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
667 {
668 const unsigned char *commit_data;
669 uint32_t lex_index;
670
671 while (pos < g->num_commits_in_base)
672 g = g->base_graph;
673
674 lex_index = pos - g->num_commits_in_base;
675 commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * lex_index;
676 item->graph_pos = pos;
677 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
678 }
679
680 static inline void set_commit_tree(struct commit *c, struct tree *t)
681 {
682 c->maybe_tree = t;
683 }
684
685 static int fill_commit_in_graph(struct repository *r,
686 struct commit *item,
687 struct commit_graph *g, uint32_t pos)
688 {
689 uint32_t edge_value;
690 uint32_t *parent_data_ptr;
691 uint64_t date_low, date_high;
692 struct commit_list **pptr;
693 const unsigned char *commit_data;
694 uint32_t lex_index;
695
696 while (pos < g->num_commits_in_base)
697 g = g->base_graph;
698
699 if (pos >= g->num_commits + g->num_commits_in_base)
700 die(_("invalid commit position. commit-graph is likely corrupt"));
701
702 /*
703 * Store the "full" position, but then use the
704 * "local" position for the rest of the calculation.
705 */
706 item->graph_pos = pos;
707 lex_index = pos - g->num_commits_in_base;
708
709 commit_data = g->chunk_commit_data + (g->hash_len + 16) * lex_index;
710
711 item->object.parsed = 1;
712
713 set_commit_tree(item, NULL);
714
715 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
716 date_low = get_be32(commit_data + g->hash_len + 12);
717 item->date = (timestamp_t)((date_high << 32) | date_low);
718
719 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
720
721 pptr = &item->parents;
722
723 edge_value = get_be32(commit_data + g->hash_len);
724 if (edge_value == GRAPH_PARENT_NONE)
725 return 1;
726 pptr = insert_parent_or_die(r, g, edge_value, pptr);
727
728 edge_value = get_be32(commit_data + g->hash_len + 4);
729 if (edge_value == GRAPH_PARENT_NONE)
730 return 1;
731 if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
732 pptr = insert_parent_or_die(r, g, edge_value, pptr);
733 return 1;
734 }
735
736 parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
737 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
738 do {
739 edge_value = get_be32(parent_data_ptr);
740 pptr = insert_parent_or_die(r, g,
741 edge_value & GRAPH_EDGE_LAST_MASK,
742 pptr);
743 parent_data_ptr++;
744 } while (!(edge_value & GRAPH_LAST_EDGE));
745
746 return 1;
747 }
748
749 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
750 {
751 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
752 *pos = item->graph_pos;
753 return 1;
754 } else {
755 struct commit_graph *cur_g = g;
756 uint32_t lex_index;
757
758 while (cur_g && !bsearch_graph(cur_g, &(item->object.oid), &lex_index))
759 cur_g = cur_g->base_graph;
760
761 if (cur_g) {
762 *pos = lex_index + cur_g->num_commits_in_base;
763 return 1;
764 }
765
766 return 0;
767 }
768 }
769
770 static int parse_commit_in_graph_one(struct repository *r,
771 struct commit_graph *g,
772 struct commit *item)
773 {
774 uint32_t pos;
775
776 if (item->object.parsed)
777 return 1;
778
779 if (find_commit_in_graph(item, g, &pos))
780 return fill_commit_in_graph(r, item, g, pos);
781
782 return 0;
783 }
784
785 int parse_commit_in_graph(struct repository *r, struct commit *item)
786 {
787 if (!prepare_commit_graph(r))
788 return 0;
789 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
790 }
791
792 void load_commit_graph_info(struct repository *r, struct commit *item)
793 {
794 uint32_t pos;
795 if (!prepare_commit_graph(r))
796 return;
797 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
798 fill_commit_graph_info(item, r->objects->commit_graph, pos);
799 }
800
801 static struct tree *load_tree_for_commit(struct repository *r,
802 struct commit_graph *g,
803 struct commit *c)
804 {
805 struct object_id oid;
806 const unsigned char *commit_data;
807
808 while (c->graph_pos < g->num_commits_in_base)
809 g = g->base_graph;
810
811 commit_data = g->chunk_commit_data +
812 GRAPH_DATA_WIDTH * (c->graph_pos - g->num_commits_in_base);
813
814 hashcpy(oid.hash, commit_data);
815 set_commit_tree(c, lookup_tree(r, &oid));
816
817 return c->maybe_tree;
818 }
819
820 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
821 struct commit_graph *g,
822 const struct commit *c)
823 {
824 if (c->maybe_tree)
825 return c->maybe_tree;
826 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
827 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
828
829 return load_tree_for_commit(r, g, (struct commit *)c);
830 }
831
832 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
833 {
834 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
835 }
836
837 struct packed_commit_list {
838 struct commit **list;
839 int nr;
840 int alloc;
841 };
842
843 struct packed_oid_list {
844 struct object_id *list;
845 int nr;
846 int alloc;
847 };
848
849 struct write_commit_graph_context {
850 struct repository *r;
851 struct object_directory *odb;
852 char *graph_name;
853 struct packed_oid_list oids;
854 struct packed_commit_list commits;
855 int num_extra_edges;
856 unsigned long approx_nr_objects;
857 struct progress *progress;
858 int progress_done;
859 uint64_t progress_cnt;
860
861 char *base_graph_name;
862 int num_commit_graphs_before;
863 int num_commit_graphs_after;
864 char **commit_graph_filenames_before;
865 char **commit_graph_filenames_after;
866 char **commit_graph_hash_after;
867 uint32_t new_num_commits_in_base;
868 struct commit_graph *new_base_graph;
869
870 unsigned append:1,
871 report_progress:1,
872 split:1,
873 check_oids:1,
874 changed_paths:1,
875 order_by_pack:1;
876
877 const struct split_commit_graph_opts *split_opts;
878 size_t total_bloom_filter_data_size;
879 };
880
881 static void write_graph_chunk_fanout(struct hashfile *f,
882 struct write_commit_graph_context *ctx)
883 {
884 int i, count = 0;
885 struct commit **list = ctx->commits.list;
886
887 /*
888 * Write the first-level table (the list is sorted,
889 * but we use a 256-entry lookup to be able to avoid
890 * having to do eight extra binary search iterations).
891 */
892 for (i = 0; i < 256; i++) {
893 while (count < ctx->commits.nr) {
894 if ((*list)->object.oid.hash[0] != i)
895 break;
896 display_progress(ctx->progress, ++ctx->progress_cnt);
897 count++;
898 list++;
899 }
900
901 hashwrite_be32(f, count);
902 }
903 }
904
905 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
906 struct write_commit_graph_context *ctx)
907 {
908 struct commit **list = ctx->commits.list;
909 int count;
910 for (count = 0; count < ctx->commits.nr; count++, list++) {
911 display_progress(ctx->progress, ++ctx->progress_cnt);
912 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
913 }
914 }
915
916 static const unsigned char *commit_to_sha1(size_t index, void *table)
917 {
918 struct commit **commits = table;
919 return commits[index]->object.oid.hash;
920 }
921
922 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
923 struct write_commit_graph_context *ctx)
924 {
925 struct commit **list = ctx->commits.list;
926 struct commit **last = ctx->commits.list + ctx->commits.nr;
927 uint32_t num_extra_edges = 0;
928
929 while (list < last) {
930 struct commit_list *parent;
931 struct object_id *tree;
932 int edge_value;
933 uint32_t packedDate[2];
934 display_progress(ctx->progress, ++ctx->progress_cnt);
935
936 if (parse_commit_no_graph(*list))
937 die(_("unable to parse commit %s"),
938 oid_to_hex(&(*list)->object.oid));
939 tree = get_commit_tree_oid(*list);
940 hashwrite(f, tree->hash, hash_len);
941
942 parent = (*list)->parents;
943
944 if (!parent)
945 edge_value = GRAPH_PARENT_NONE;
946 else {
947 edge_value = sha1_pos(parent->item->object.oid.hash,
948 ctx->commits.list,
949 ctx->commits.nr,
950 commit_to_sha1);
951
952 if (edge_value >= 0)
953 edge_value += ctx->new_num_commits_in_base;
954 else {
955 uint32_t pos;
956 if (find_commit_in_graph(parent->item,
957 ctx->new_base_graph,
958 &pos))
959 edge_value = pos;
960 }
961
962 if (edge_value < 0)
963 BUG("missing parent %s for commit %s",
964 oid_to_hex(&parent->item->object.oid),
965 oid_to_hex(&(*list)->object.oid));
966 }
967
968 hashwrite_be32(f, edge_value);
969
970 if (parent)
971 parent = parent->next;
972
973 if (!parent)
974 edge_value = GRAPH_PARENT_NONE;
975 else if (parent->next)
976 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
977 else {
978 edge_value = sha1_pos(parent->item->object.oid.hash,
979 ctx->commits.list,
980 ctx->commits.nr,
981 commit_to_sha1);
982
983 if (edge_value >= 0)
984 edge_value += ctx->new_num_commits_in_base;
985 else {
986 uint32_t pos;
987 if (find_commit_in_graph(parent->item,
988 ctx->new_base_graph,
989 &pos))
990 edge_value = pos;
991 }
992
993 if (edge_value < 0)
994 BUG("missing parent %s for commit %s",
995 oid_to_hex(&parent->item->object.oid),
996 oid_to_hex(&(*list)->object.oid));
997 }
998
999 hashwrite_be32(f, edge_value);
1000
1001 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
1002 do {
1003 num_extra_edges++;
1004 parent = parent->next;
1005 } while (parent);
1006 }
1007
1008 if (sizeof((*list)->date) > 4)
1009 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
1010 else
1011 packedDate[0] = 0;
1012
1013 packedDate[0] |= htonl((*list)->generation << 2);
1014
1015 packedDate[1] = htonl((*list)->date);
1016 hashwrite(f, packedDate, 8);
1017
1018 list++;
1019 }
1020 }
1021
1022 static void write_graph_chunk_extra_edges(struct hashfile *f,
1023 struct write_commit_graph_context *ctx)
1024 {
1025 struct commit **list = ctx->commits.list;
1026 struct commit **last = ctx->commits.list + ctx->commits.nr;
1027 struct commit_list *parent;
1028
1029 while (list < last) {
1030 int num_parents = 0;
1031
1032 display_progress(ctx->progress, ++ctx->progress_cnt);
1033
1034 for (parent = (*list)->parents; num_parents < 3 && parent;
1035 parent = parent->next)
1036 num_parents++;
1037
1038 if (num_parents <= 2) {
1039 list++;
1040 continue;
1041 }
1042
1043 /* Since num_parents > 2, this initializer is safe. */
1044 for (parent = (*list)->parents->next; parent; parent = parent->next) {
1045 int edge_value = sha1_pos(parent->item->object.oid.hash,
1046 ctx->commits.list,
1047 ctx->commits.nr,
1048 commit_to_sha1);
1049
1050 if (edge_value >= 0)
1051 edge_value += ctx->new_num_commits_in_base;
1052 else {
1053 uint32_t pos;
1054 if (find_commit_in_graph(parent->item,
1055 ctx->new_base_graph,
1056 &pos))
1057 edge_value = pos;
1058 }
1059
1060 if (edge_value < 0)
1061 BUG("missing parent %s for commit %s",
1062 oid_to_hex(&parent->item->object.oid),
1063 oid_to_hex(&(*list)->object.oid));
1064 else if (!parent->next)
1065 edge_value |= GRAPH_LAST_EDGE;
1066
1067 hashwrite_be32(f, edge_value);
1068 }
1069
1070 list++;
1071 }
1072 }
1073
1074 static void write_graph_chunk_bloom_indexes(struct hashfile *f,
1075 struct write_commit_graph_context *ctx)
1076 {
1077 struct commit **list = ctx->commits.list;
1078 struct commit **last = ctx->commits.list + ctx->commits.nr;
1079 uint32_t cur_pos = 0;
1080 struct progress *progress = NULL;
1081 int i = 0;
1082
1083 if (ctx->report_progress)
1084 progress = start_delayed_progress(
1085 _("Writing changed paths Bloom filters index"),
1086 ctx->commits.nr);
1087
1088 while (list < last) {
1089 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list, 0);
1090 cur_pos += filter->len;
1091 display_progress(progress, ++i);
1092 hashwrite_be32(f, cur_pos);
1093 list++;
1094 }
1095
1096 stop_progress(&progress);
1097 }
1098
1099 static void write_graph_chunk_bloom_data(struct hashfile *f,
1100 struct write_commit_graph_context *ctx,
1101 const struct bloom_filter_settings *settings)
1102 {
1103 struct commit **list = ctx->commits.list;
1104 struct commit **last = ctx->commits.list + ctx->commits.nr;
1105 struct progress *progress = NULL;
1106 int i = 0;
1107
1108 if (ctx->report_progress)
1109 progress = start_delayed_progress(
1110 _("Writing changed paths Bloom filters data"),
1111 ctx->commits.nr);
1112
1113 hashwrite_be32(f, settings->hash_version);
1114 hashwrite_be32(f, settings->num_hashes);
1115 hashwrite_be32(f, settings->bits_per_entry);
1116
1117 while (list < last) {
1118 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list, 0);
1119 display_progress(progress, ++i);
1120 hashwrite(f, filter->data, filter->len * sizeof(unsigned char));
1121 list++;
1122 }
1123
1124 stop_progress(&progress);
1125 }
1126
1127 static int oid_compare(const void *_a, const void *_b)
1128 {
1129 const struct object_id *a = (const struct object_id *)_a;
1130 const struct object_id *b = (const struct object_id *)_b;
1131 return oidcmp(a, b);
1132 }
1133
1134 static int add_packed_commits(const struct object_id *oid,
1135 struct packed_git *pack,
1136 uint32_t pos,
1137 void *data)
1138 {
1139 struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
1140 enum object_type type;
1141 off_t offset = nth_packed_object_offset(pack, pos);
1142 struct object_info oi = OBJECT_INFO_INIT;
1143
1144 if (ctx->progress)
1145 display_progress(ctx->progress, ++ctx->progress_done);
1146
1147 oi.typep = &type;
1148 if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
1149 die(_("unable to get type of object %s"), oid_to_hex(oid));
1150
1151 if (type != OBJ_COMMIT)
1152 return 0;
1153
1154 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1155 oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
1156 ctx->oids.nr++;
1157
1158 set_commit_pos(ctx->r, oid);
1159
1160 return 0;
1161 }
1162
1163 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
1164 {
1165 struct commit_list *parent;
1166 for (parent = commit->parents; parent; parent = parent->next) {
1167 if (!(parent->item->object.flags & REACHABLE)) {
1168 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1169 oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
1170 ctx->oids.nr++;
1171 parent->item->object.flags |= REACHABLE;
1172 }
1173 }
1174 }
1175
1176 static void close_reachable(struct write_commit_graph_context *ctx)
1177 {
1178 int i;
1179 struct commit *commit;
1180
1181 if (ctx->report_progress)
1182 ctx->progress = start_delayed_progress(
1183 _("Loading known commits in commit graph"),
1184 ctx->oids.nr);
1185 for (i = 0; i < ctx->oids.nr; i++) {
1186 display_progress(ctx->progress, i + 1);
1187 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1188 if (commit)
1189 commit->object.flags |= REACHABLE;
1190 }
1191 stop_progress(&ctx->progress);
1192
1193 /*
1194 * As this loop runs, ctx->oids.nr may grow, but not more
1195 * than the number of missing commits in the reachable
1196 * closure.
1197 */
1198 if (ctx->report_progress)
1199 ctx->progress = start_delayed_progress(
1200 _("Expanding reachable commits in commit graph"),
1201 0);
1202 for (i = 0; i < ctx->oids.nr; i++) {
1203 display_progress(ctx->progress, i + 1);
1204 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1205
1206 if (!commit)
1207 continue;
1208 if (ctx->split) {
1209 if (!parse_commit(commit) &&
1210 commit->graph_pos == COMMIT_NOT_FROM_GRAPH)
1211 add_missing_parents(ctx, commit);
1212 } else if (!parse_commit_no_graph(commit))
1213 add_missing_parents(ctx, commit);
1214 }
1215 stop_progress(&ctx->progress);
1216
1217 if (ctx->report_progress)
1218 ctx->progress = start_delayed_progress(
1219 _("Clearing commit marks in commit graph"),
1220 ctx->oids.nr);
1221 for (i = 0; i < ctx->oids.nr; i++) {
1222 display_progress(ctx->progress, i + 1);
1223 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1224
1225 if (commit)
1226 commit->object.flags &= ~REACHABLE;
1227 }
1228 stop_progress(&ctx->progress);
1229 }
1230
1231 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1232 {
1233 int i;
1234 struct commit_list *list = NULL;
1235
1236 if (ctx->report_progress)
1237 ctx->progress = start_delayed_progress(
1238 _("Computing commit graph generation numbers"),
1239 ctx->commits.nr);
1240 for (i = 0; i < ctx->commits.nr; i++) {
1241 display_progress(ctx->progress, i + 1);
1242 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
1243 ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
1244 continue;
1245
1246 commit_list_insert(ctx->commits.list[i], &list);
1247 while (list) {
1248 struct commit *current = list->item;
1249 struct commit_list *parent;
1250 int all_parents_computed = 1;
1251 uint32_t max_generation = 0;
1252
1253 for (parent = current->parents; parent; parent = parent->next) {
1254 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
1255 parent->item->generation == GENERATION_NUMBER_ZERO) {
1256 all_parents_computed = 0;
1257 commit_list_insert(parent->item, &list);
1258 break;
1259 } else if (parent->item->generation > max_generation) {
1260 max_generation = parent->item->generation;
1261 }
1262 }
1263
1264 if (all_parents_computed) {
1265 current->generation = max_generation + 1;
1266 pop_commit(&list);
1267
1268 if (current->generation > GENERATION_NUMBER_MAX)
1269 current->generation = GENERATION_NUMBER_MAX;
1270 }
1271 }
1272 }
1273 stop_progress(&ctx->progress);
1274 }
1275
1276 static void compute_bloom_filters(struct write_commit_graph_context *ctx)
1277 {
1278 int i;
1279 struct progress *progress = NULL;
1280 struct commit **sorted_commits;
1281
1282 init_bloom_filters();
1283
1284 if (ctx->report_progress)
1285 progress = start_delayed_progress(
1286 _("Computing commit changed paths Bloom filters"),
1287 ctx->commits.nr);
1288
1289 ALLOC_ARRAY(sorted_commits, ctx->commits.nr);
1290 COPY_ARRAY(sorted_commits, ctx->commits.list, ctx->commits.nr);
1291
1292 if (ctx->order_by_pack)
1293 QSORT(sorted_commits, ctx->commits.nr, commit_pos_cmp);
1294 else
1295 QSORT(sorted_commits, ctx->commits.nr, commit_gen_cmp);
1296
1297 for (i = 0; i < ctx->commits.nr; i++) {
1298 struct commit *c = sorted_commits[i];
1299 struct bloom_filter *filter = get_bloom_filter(ctx->r, c, 1);
1300 ctx->total_bloom_filter_data_size += sizeof(unsigned char) * filter->len;
1301 display_progress(progress, i + 1);
1302 }
1303
1304 free(sorted_commits);
1305 stop_progress(&progress);
1306 }
1307
1308 static int add_ref_to_list(const char *refname,
1309 const struct object_id *oid,
1310 int flags, void *cb_data)
1311 {
1312 struct string_list *list = (struct string_list *)cb_data;
1313
1314 string_list_append(list, oid_to_hex(oid));
1315 return 0;
1316 }
1317
1318 int write_commit_graph_reachable(struct object_directory *odb,
1319 enum commit_graph_write_flags flags,
1320 const struct split_commit_graph_opts *split_opts)
1321 {
1322 struct string_list list = STRING_LIST_INIT_DUP;
1323 int result;
1324
1325 for_each_ref(add_ref_to_list, &list);
1326 result = write_commit_graph(odb, NULL, &list,
1327 flags, split_opts);
1328
1329 string_list_clear(&list, 0);
1330 return result;
1331 }
1332
1333 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
1334 struct string_list *pack_indexes)
1335 {
1336 uint32_t i;
1337 struct strbuf progress_title = STRBUF_INIT;
1338 struct strbuf packname = STRBUF_INIT;
1339 int dirlen;
1340
1341 strbuf_addf(&packname, "%s/pack/", ctx->odb->path);
1342 dirlen = packname.len;
1343 if (ctx->report_progress) {
1344 strbuf_addf(&progress_title,
1345 Q_("Finding commits for commit graph in %d pack",
1346 "Finding commits for commit graph in %d packs",
1347 pack_indexes->nr),
1348 pack_indexes->nr);
1349 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1350 ctx->progress_done = 0;
1351 }
1352 for (i = 0; i < pack_indexes->nr; i++) {
1353 struct packed_git *p;
1354 strbuf_setlen(&packname, dirlen);
1355 strbuf_addstr(&packname, pack_indexes->items[i].string);
1356 p = add_packed_git(packname.buf, packname.len, 1);
1357 if (!p) {
1358 error(_("error adding pack %s"), packname.buf);
1359 return -1;
1360 }
1361 if (open_pack_index(p)) {
1362 error(_("error opening index for %s"), packname.buf);
1363 return -1;
1364 }
1365 for_each_object_in_pack(p, add_packed_commits, ctx,
1366 FOR_EACH_OBJECT_PACK_ORDER);
1367 close_pack(p);
1368 free(p);
1369 }
1370
1371 stop_progress(&ctx->progress);
1372 strbuf_release(&progress_title);
1373 strbuf_release(&packname);
1374
1375 return 0;
1376 }
1377
1378 static int fill_oids_from_commit_hex(struct write_commit_graph_context *ctx,
1379 struct string_list *commit_hex)
1380 {
1381 uint32_t i;
1382 struct strbuf progress_title = STRBUF_INIT;
1383
1384 if (ctx->report_progress) {
1385 strbuf_addf(&progress_title,
1386 Q_("Finding commits for commit graph from %d ref",
1387 "Finding commits for commit graph from %d refs",
1388 commit_hex->nr),
1389 commit_hex->nr);
1390 ctx->progress = start_delayed_progress(
1391 progress_title.buf,
1392 commit_hex->nr);
1393 }
1394 for (i = 0; i < commit_hex->nr; i++) {
1395 const char *end;
1396 struct object_id oid;
1397 struct commit *result;
1398
1399 display_progress(ctx->progress, i + 1);
1400 if (!parse_oid_hex(commit_hex->items[i].string, &oid, &end) &&
1401 (result = lookup_commit_reference_gently(ctx->r, &oid, 1))) {
1402 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1403 oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
1404 ctx->oids.nr++;
1405 } else if (ctx->check_oids) {
1406 error(_("invalid commit object id: %s"),
1407 commit_hex->items[i].string);
1408 return -1;
1409 }
1410 }
1411 stop_progress(&ctx->progress);
1412 strbuf_release(&progress_title);
1413
1414 return 0;
1415 }
1416
1417 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1418 {
1419 if (ctx->report_progress)
1420 ctx->progress = start_delayed_progress(
1421 _("Finding commits for commit graph among packed objects"),
1422 ctx->approx_nr_objects);
1423 for_each_packed_object(add_packed_commits, ctx,
1424 FOR_EACH_OBJECT_PACK_ORDER);
1425 if (ctx->progress_done < ctx->approx_nr_objects)
1426 display_progress(ctx->progress, ctx->approx_nr_objects);
1427 stop_progress(&ctx->progress);
1428 }
1429
1430 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
1431 {
1432 uint32_t i, count_distinct = 1;
1433
1434 if (ctx->report_progress)
1435 ctx->progress = start_delayed_progress(
1436 _("Counting distinct commits in commit graph"),
1437 ctx->oids.nr);
1438 display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1439 QSORT(ctx->oids.list, ctx->oids.nr, oid_compare);
1440
1441 for (i = 1; i < ctx->oids.nr; i++) {
1442 display_progress(ctx->progress, i + 1);
1443 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i])) {
1444 if (ctx->split) {
1445 struct commit *c = lookup_commit(ctx->r, &ctx->oids.list[i]);
1446
1447 if (!c || c->graph_pos != COMMIT_NOT_FROM_GRAPH)
1448 continue;
1449 }
1450
1451 count_distinct++;
1452 }
1453 }
1454 stop_progress(&ctx->progress);
1455
1456 return count_distinct;
1457 }
1458
1459 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1460 {
1461 uint32_t i;
1462
1463 ctx->num_extra_edges = 0;
1464 if (ctx->report_progress)
1465 ctx->progress = start_delayed_progress(
1466 _("Finding extra edges in commit graph"),
1467 ctx->oids.nr);
1468 for (i = 0; i < ctx->oids.nr; i++) {
1469 unsigned int num_parents;
1470
1471 display_progress(ctx->progress, i + 1);
1472 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1473 continue;
1474
1475 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + 1, ctx->commits.alloc);
1476 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1477
1478 if (ctx->split &&
1479 ctx->commits.list[ctx->commits.nr]->graph_pos != COMMIT_NOT_FROM_GRAPH)
1480 continue;
1481
1482 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1483
1484 num_parents = commit_list_count(ctx->commits.list[ctx->commits.nr]->parents);
1485 if (num_parents > 2)
1486 ctx->num_extra_edges += num_parents - 1;
1487
1488 ctx->commits.nr++;
1489 }
1490 stop_progress(&ctx->progress);
1491 }
1492
1493 static int write_graph_chunk_base_1(struct hashfile *f,
1494 struct commit_graph *g)
1495 {
1496 int num = 0;
1497
1498 if (!g)
1499 return 0;
1500
1501 num = write_graph_chunk_base_1(f, g->base_graph);
1502 hashwrite(f, g->oid.hash, the_hash_algo->rawsz);
1503 return num + 1;
1504 }
1505
1506 static int write_graph_chunk_base(struct hashfile *f,
1507 struct write_commit_graph_context *ctx)
1508 {
1509 int num = write_graph_chunk_base_1(f, ctx->new_base_graph);
1510
1511 if (num != ctx->num_commit_graphs_after - 1) {
1512 error(_("failed to write correct number of base graph ids"));
1513 return -1;
1514 }
1515
1516 return 0;
1517 }
1518
1519 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1520 {
1521 uint32_t i;
1522 int fd;
1523 struct hashfile *f;
1524 struct lock_file lk = LOCK_INIT;
1525 uint32_t chunk_ids[MAX_NUM_CHUNKS + 1];
1526 uint64_t chunk_offsets[MAX_NUM_CHUNKS + 1];
1527 const unsigned hashsz = the_hash_algo->rawsz;
1528 struct strbuf progress_title = STRBUF_INIT;
1529 int num_chunks = 3;
1530 struct object_id file_hash;
1531 const struct bloom_filter_settings bloom_settings = DEFAULT_BLOOM_FILTER_SETTINGS;
1532
1533 if (ctx->split) {
1534 struct strbuf tmp_file = STRBUF_INIT;
1535
1536 strbuf_addf(&tmp_file,
1537 "%s/info/commit-graphs/tmp_graph_XXXXXX",
1538 ctx->odb->path);
1539 ctx->graph_name = strbuf_detach(&tmp_file, NULL);
1540 } else {
1541 ctx->graph_name = get_commit_graph_filename(ctx->odb);
1542 }
1543
1544 if (safe_create_leading_directories(ctx->graph_name)) {
1545 UNLEAK(ctx->graph_name);
1546 error(_("unable to create leading directories of %s"),
1547 ctx->graph_name);
1548 return -1;
1549 }
1550
1551 if (ctx->split) {
1552 char *lock_name = get_chain_filename(ctx->odb);
1553
1554 hold_lock_file_for_update(&lk, lock_name, LOCK_DIE_ON_ERROR);
1555
1556 fd = git_mkstemp_mode(ctx->graph_name, 0444);
1557 if (fd < 0) {
1558 error(_("unable to create '%s'"), ctx->graph_name);
1559 return -1;
1560 }
1561
1562 f = hashfd(fd, ctx->graph_name);
1563 } else {
1564 hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
1565 fd = lk.tempfile->fd;
1566 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1567 }
1568
1569 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1570 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1571 chunk_ids[2] = GRAPH_CHUNKID_DATA;
1572 if (ctx->num_extra_edges) {
1573 chunk_ids[num_chunks] = GRAPH_CHUNKID_EXTRAEDGES;
1574 num_chunks++;
1575 }
1576 if (ctx->changed_paths) {
1577 chunk_ids[num_chunks] = GRAPH_CHUNKID_BLOOMINDEXES;
1578 num_chunks++;
1579 chunk_ids[num_chunks] = GRAPH_CHUNKID_BLOOMDATA;
1580 num_chunks++;
1581 }
1582 if (ctx->num_commit_graphs_after > 1) {
1583 chunk_ids[num_chunks] = GRAPH_CHUNKID_BASE;
1584 num_chunks++;
1585 }
1586
1587 chunk_ids[num_chunks] = 0;
1588
1589 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1590 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1591 chunk_offsets[2] = chunk_offsets[1] + hashsz * ctx->commits.nr;
1592 chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * ctx->commits.nr;
1593
1594 num_chunks = 3;
1595 if (ctx->num_extra_edges) {
1596 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1597 4 * ctx->num_extra_edges;
1598 num_chunks++;
1599 }
1600 if (ctx->changed_paths) {
1601 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1602 sizeof(uint32_t) * ctx->commits.nr;
1603 num_chunks++;
1604
1605 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1606 sizeof(uint32_t) * 3 + ctx->total_bloom_filter_data_size;
1607 num_chunks++;
1608 }
1609 if (ctx->num_commit_graphs_after > 1) {
1610 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1611 hashsz * (ctx->num_commit_graphs_after - 1);
1612 num_chunks++;
1613 }
1614
1615 hashwrite_be32(f, GRAPH_SIGNATURE);
1616
1617 hashwrite_u8(f, GRAPH_VERSION);
1618 hashwrite_u8(f, oid_version());
1619 hashwrite_u8(f, num_chunks);
1620 hashwrite_u8(f, ctx->num_commit_graphs_after - 1);
1621
1622 for (i = 0; i <= num_chunks; i++) {
1623 uint32_t chunk_write[3];
1624
1625 chunk_write[0] = htonl(chunk_ids[i]);
1626 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1627 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1628 hashwrite(f, chunk_write, 12);
1629 }
1630
1631 if (ctx->report_progress) {
1632 strbuf_addf(&progress_title,
1633 Q_("Writing out commit graph in %d pass",
1634 "Writing out commit graph in %d passes",
1635 num_chunks),
1636 num_chunks);
1637 ctx->progress = start_delayed_progress(
1638 progress_title.buf,
1639 num_chunks * ctx->commits.nr);
1640 }
1641 write_graph_chunk_fanout(f, ctx);
1642 write_graph_chunk_oids(f, hashsz, ctx);
1643 write_graph_chunk_data(f, hashsz, ctx);
1644 if (ctx->num_extra_edges)
1645 write_graph_chunk_extra_edges(f, ctx);
1646 if (ctx->changed_paths) {
1647 write_graph_chunk_bloom_indexes(f, ctx);
1648 write_graph_chunk_bloom_data(f, ctx, &bloom_settings);
1649 }
1650 if (ctx->num_commit_graphs_after > 1 &&
1651 write_graph_chunk_base(f, ctx)) {
1652 return -1;
1653 }
1654 stop_progress(&ctx->progress);
1655 strbuf_release(&progress_title);
1656
1657 if (ctx->split && ctx->base_graph_name && ctx->num_commit_graphs_after > 1) {
1658 char *new_base_hash = xstrdup(oid_to_hex(&ctx->new_base_graph->oid));
1659 char *new_base_name = get_split_graph_filename(ctx->new_base_graph->odb, new_base_hash);
1660
1661 free(ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2]);
1662 free(ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2]);
1663 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2] = new_base_name;
1664 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2] = new_base_hash;
1665 }
1666
1667 close_commit_graph(ctx->r->objects);
1668 finalize_hashfile(f, file_hash.hash, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1669
1670 if (ctx->split) {
1671 FILE *chainf = fdopen_lock_file(&lk, "w");
1672 char *final_graph_name;
1673 int result;
1674
1675 close(fd);
1676
1677 if (!chainf) {
1678 error(_("unable to open commit-graph chain file"));
1679 return -1;
1680 }
1681
1682 if (ctx->base_graph_name) {
1683 const char *dest = ctx->commit_graph_filenames_after[
1684 ctx->num_commit_graphs_after - 2];
1685
1686 if (strcmp(ctx->base_graph_name, dest)) {
1687 result = rename(ctx->base_graph_name, dest);
1688
1689 if (result) {
1690 error(_("failed to rename base commit-graph file"));
1691 return -1;
1692 }
1693 }
1694 } else {
1695 char *graph_name = get_commit_graph_filename(ctx->odb);
1696 unlink(graph_name);
1697 }
1698
1699 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1] = xstrdup(oid_to_hex(&file_hash));
1700 final_graph_name = get_split_graph_filename(ctx->odb,
1701 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1]);
1702 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 1] = final_graph_name;
1703
1704 result = rename(ctx->graph_name, final_graph_name);
1705
1706 for (i = 0; i < ctx->num_commit_graphs_after; i++)
1707 fprintf(lk.tempfile->fp, "%s\n", ctx->commit_graph_hash_after[i]);
1708
1709 if (result) {
1710 error(_("failed to rename temporary commit-graph file"));
1711 return -1;
1712 }
1713 }
1714
1715 commit_lock_file(&lk);
1716
1717 return 0;
1718 }
1719
1720 static void split_graph_merge_strategy(struct write_commit_graph_context *ctx)
1721 {
1722 struct commit_graph *g;
1723 uint32_t num_commits;
1724 uint32_t i;
1725
1726 int max_commits = 0;
1727 int size_mult = 2;
1728
1729 if (ctx->split_opts) {
1730 max_commits = ctx->split_opts->max_commits;
1731
1732 if (ctx->split_opts->size_multiple)
1733 size_mult = ctx->split_opts->size_multiple;
1734 }
1735
1736 g = ctx->r->objects->commit_graph;
1737 num_commits = ctx->commits.nr;
1738 ctx->num_commit_graphs_after = ctx->num_commit_graphs_before + 1;
1739
1740 while (g && (g->num_commits <= size_mult * num_commits ||
1741 (max_commits && num_commits > max_commits))) {
1742 if (g->odb != ctx->odb)
1743 break;
1744
1745 num_commits += g->num_commits;
1746 g = g->base_graph;
1747
1748 ctx->num_commit_graphs_after--;
1749 }
1750
1751 ctx->new_base_graph = g;
1752
1753 if (ctx->num_commit_graphs_after == 2) {
1754 char *old_graph_name = get_commit_graph_filename(g->odb);
1755
1756 if (!strcmp(g->filename, old_graph_name) &&
1757 g->odb != ctx->odb) {
1758 ctx->num_commit_graphs_after = 1;
1759 ctx->new_base_graph = NULL;
1760 }
1761
1762 free(old_graph_name);
1763 }
1764
1765 ALLOC_ARRAY(ctx->commit_graph_filenames_after, ctx->num_commit_graphs_after);
1766 ALLOC_ARRAY(ctx->commit_graph_hash_after, ctx->num_commit_graphs_after);
1767
1768 for (i = 0; i < ctx->num_commit_graphs_after &&
1769 i < ctx->num_commit_graphs_before; i++)
1770 ctx->commit_graph_filenames_after[i] = xstrdup(ctx->commit_graph_filenames_before[i]);
1771
1772 i = ctx->num_commit_graphs_before - 1;
1773 g = ctx->r->objects->commit_graph;
1774
1775 while (g) {
1776 if (i < ctx->num_commit_graphs_after)
1777 ctx->commit_graph_hash_after[i] = xstrdup(oid_to_hex(&g->oid));
1778
1779 i--;
1780 g = g->base_graph;
1781 }
1782 }
1783
1784 static void merge_commit_graph(struct write_commit_graph_context *ctx,
1785 struct commit_graph *g)
1786 {
1787 uint32_t i;
1788 uint32_t offset = g->num_commits_in_base;
1789
1790 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + g->num_commits, ctx->commits.alloc);
1791
1792 for (i = 0; i < g->num_commits; i++) {
1793 struct object_id oid;
1794 struct commit *result;
1795
1796 display_progress(ctx->progress, i + 1);
1797
1798 load_oid_from_graph(g, i + offset, &oid);
1799
1800 /* only add commits if they still exist in the repo */
1801 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1802
1803 if (result) {
1804 ctx->commits.list[ctx->commits.nr] = result;
1805 ctx->commits.nr++;
1806 }
1807 }
1808 }
1809
1810 static int commit_compare(const void *_a, const void *_b)
1811 {
1812 const struct commit *a = *(const struct commit **)_a;
1813 const struct commit *b = *(const struct commit **)_b;
1814 return oidcmp(&a->object.oid, &b->object.oid);
1815 }
1816
1817 static void sort_and_scan_merged_commits(struct write_commit_graph_context *ctx)
1818 {
1819 uint32_t i;
1820
1821 if (ctx->report_progress)
1822 ctx->progress = start_delayed_progress(
1823 _("Scanning merged commits"),
1824 ctx->commits.nr);
1825
1826 QSORT(ctx->commits.list, ctx->commits.nr, commit_compare);
1827
1828 ctx->num_extra_edges = 0;
1829 for (i = 0; i < ctx->commits.nr; i++) {
1830 display_progress(ctx->progress, i);
1831
1832 if (i && oideq(&ctx->commits.list[i - 1]->object.oid,
1833 &ctx->commits.list[i]->object.oid)) {
1834 die(_("unexpected duplicate commit id %s"),
1835 oid_to_hex(&ctx->commits.list[i]->object.oid));
1836 } else {
1837 unsigned int num_parents;
1838
1839 num_parents = commit_list_count(ctx->commits.list[i]->parents);
1840 if (num_parents > 2)
1841 ctx->num_extra_edges += num_parents - 1;
1842 }
1843 }
1844
1845 stop_progress(&ctx->progress);
1846 }
1847
1848 static void merge_commit_graphs(struct write_commit_graph_context *ctx)
1849 {
1850 struct commit_graph *g = ctx->r->objects->commit_graph;
1851 uint32_t current_graph_number = ctx->num_commit_graphs_before;
1852
1853 while (g && current_graph_number >= ctx->num_commit_graphs_after) {
1854 current_graph_number--;
1855
1856 if (ctx->report_progress)
1857 ctx->progress = start_delayed_progress(_("Merging commit-graph"), 0);
1858
1859 merge_commit_graph(ctx, g);
1860 stop_progress(&ctx->progress);
1861
1862 g = g->base_graph;
1863 }
1864
1865 if (g) {
1866 ctx->new_base_graph = g;
1867 ctx->new_num_commits_in_base = g->num_commits + g->num_commits_in_base;
1868 }
1869
1870 if (ctx->new_base_graph)
1871 ctx->base_graph_name = xstrdup(ctx->new_base_graph->filename);
1872
1873 sort_and_scan_merged_commits(ctx);
1874 }
1875
1876 static void mark_commit_graphs(struct write_commit_graph_context *ctx)
1877 {
1878 uint32_t i;
1879 time_t now = time(NULL);
1880
1881 for (i = ctx->num_commit_graphs_after - 1; i < ctx->num_commit_graphs_before; i++) {
1882 struct stat st;
1883 struct utimbuf updated_time;
1884
1885 stat(ctx->commit_graph_filenames_before[i], &st);
1886
1887 updated_time.actime = st.st_atime;
1888 updated_time.modtime = now;
1889 utime(ctx->commit_graph_filenames_before[i], &updated_time);
1890 }
1891 }
1892
1893 static void expire_commit_graphs(struct write_commit_graph_context *ctx)
1894 {
1895 struct strbuf path = STRBUF_INIT;
1896 DIR *dir;
1897 struct dirent *de;
1898 size_t dirnamelen;
1899 timestamp_t expire_time = time(NULL);
1900
1901 if (ctx->split_opts && ctx->split_opts->expire_time)
1902 expire_time -= ctx->split_opts->expire_time;
1903 if (!ctx->split) {
1904 char *chain_file_name = get_chain_filename(ctx->odb);
1905 unlink(chain_file_name);
1906 free(chain_file_name);
1907 ctx->num_commit_graphs_after = 0;
1908 }
1909
1910 strbuf_addstr(&path, ctx->odb->path);
1911 strbuf_addstr(&path, "/info/commit-graphs");
1912 dir = opendir(path.buf);
1913
1914 if (!dir)
1915 goto out;
1916
1917 strbuf_addch(&path, '/');
1918 dirnamelen = path.len;
1919 while ((de = readdir(dir)) != NULL) {
1920 struct stat st;
1921 uint32_t i, found = 0;
1922
1923 strbuf_setlen(&path, dirnamelen);
1924 strbuf_addstr(&path, de->d_name);
1925
1926 stat(path.buf, &st);
1927
1928 if (st.st_mtime > expire_time)
1929 continue;
1930 if (path.len < 6 || strcmp(path.buf + path.len - 6, ".graph"))
1931 continue;
1932
1933 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
1934 if (!strcmp(ctx->commit_graph_filenames_after[i],
1935 path.buf)) {
1936 found = 1;
1937 break;
1938 }
1939 }
1940
1941 if (!found)
1942 unlink(path.buf);
1943 }
1944
1945 out:
1946 strbuf_release(&path);
1947 }
1948
1949 int write_commit_graph(struct object_directory *odb,
1950 struct string_list *pack_indexes,
1951 struct string_list *commit_hex,
1952 enum commit_graph_write_flags flags,
1953 const struct split_commit_graph_opts *split_opts)
1954 {
1955 struct write_commit_graph_context *ctx;
1956 uint32_t i, count_distinct = 0;
1957 int res = 0;
1958
1959 if (!commit_graph_compatible(the_repository))
1960 return 0;
1961
1962 ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
1963 ctx->r = the_repository;
1964 ctx->odb = odb;
1965 ctx->append = flags & COMMIT_GRAPH_WRITE_APPEND ? 1 : 0;
1966 ctx->report_progress = flags & COMMIT_GRAPH_WRITE_PROGRESS ? 1 : 0;
1967 ctx->split = flags & COMMIT_GRAPH_WRITE_SPLIT ? 1 : 0;
1968 ctx->check_oids = flags & COMMIT_GRAPH_WRITE_CHECK_OIDS ? 1 : 0;
1969 ctx->split_opts = split_opts;
1970 ctx->changed_paths = flags & COMMIT_GRAPH_WRITE_BLOOM_FILTERS ? 1 : 0;
1971 ctx->total_bloom_filter_data_size = 0;
1972
1973 if (ctx->split) {
1974 struct commit_graph *g;
1975 prepare_commit_graph(ctx->r);
1976
1977 g = ctx->r->objects->commit_graph;
1978
1979 while (g) {
1980 ctx->num_commit_graphs_before++;
1981 g = g->base_graph;
1982 }
1983
1984 if (ctx->num_commit_graphs_before) {
1985 ALLOC_ARRAY(ctx->commit_graph_filenames_before, ctx->num_commit_graphs_before);
1986 i = ctx->num_commit_graphs_before;
1987 g = ctx->r->objects->commit_graph;
1988
1989 while (g) {
1990 ctx->commit_graph_filenames_before[--i] = xstrdup(g->filename);
1991 g = g->base_graph;
1992 }
1993 }
1994 }
1995
1996 ctx->approx_nr_objects = approximate_object_count();
1997 ctx->oids.alloc = ctx->approx_nr_objects / 32;
1998
1999 if (ctx->split && split_opts && ctx->oids.alloc > split_opts->max_commits)
2000 ctx->oids.alloc = split_opts->max_commits;
2001
2002 if (ctx->append) {
2003 prepare_commit_graph_one(ctx->r, ctx->odb);
2004 if (ctx->r->objects->commit_graph)
2005 ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
2006 }
2007
2008 if (ctx->oids.alloc < 1024)
2009 ctx->oids.alloc = 1024;
2010 ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
2011
2012 if (ctx->append && ctx->r->objects->commit_graph) {
2013 struct commit_graph *g = ctx->r->objects->commit_graph;
2014 for (i = 0; i < g->num_commits; i++) {
2015 const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
2016 hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
2017 }
2018 }
2019
2020 if (pack_indexes) {
2021 ctx->order_by_pack = 1;
2022 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
2023 goto cleanup;
2024 }
2025
2026 if (commit_hex) {
2027 if ((res = fill_oids_from_commit_hex(ctx, commit_hex)))
2028 goto cleanup;
2029 }
2030
2031 if (!pack_indexes && !commit_hex) {
2032 ctx->order_by_pack = 1;
2033 fill_oids_from_all_packs(ctx);
2034 }
2035
2036 close_reachable(ctx);
2037
2038 count_distinct = count_distinct_commits(ctx);
2039
2040 if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
2041 error(_("the commit graph format cannot write %d commits"), count_distinct);
2042 res = -1;
2043 goto cleanup;
2044 }
2045
2046 ctx->commits.alloc = count_distinct;
2047 ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
2048
2049 copy_oids_to_commits(ctx);
2050
2051 if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
2052 error(_("too many commits to write graph"));
2053 res = -1;
2054 goto cleanup;
2055 }
2056
2057 if (!ctx->commits.nr)
2058 goto cleanup;
2059
2060 if (ctx->split) {
2061 split_graph_merge_strategy(ctx);
2062
2063 merge_commit_graphs(ctx);
2064 } else
2065 ctx->num_commit_graphs_after = 1;
2066
2067 compute_generation_numbers(ctx);
2068
2069 if (ctx->changed_paths)
2070 compute_bloom_filters(ctx);
2071
2072 res = write_commit_graph_file(ctx);
2073
2074 if (ctx->split)
2075 mark_commit_graphs(ctx);
2076
2077 expire_commit_graphs(ctx);
2078
2079 cleanup:
2080 free(ctx->graph_name);
2081 free(ctx->commits.list);
2082 free(ctx->oids.list);
2083
2084 if (ctx->commit_graph_filenames_after) {
2085 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
2086 free(ctx->commit_graph_filenames_after[i]);
2087 free(ctx->commit_graph_hash_after[i]);
2088 }
2089
2090 for (i = 0; i < ctx->num_commit_graphs_before; i++)
2091 free(ctx->commit_graph_filenames_before[i]);
2092
2093 free(ctx->commit_graph_filenames_after);
2094 free(ctx->commit_graph_filenames_before);
2095 free(ctx->commit_graph_hash_after);
2096 }
2097
2098 free(ctx);
2099
2100 return res;
2101 }
2102
2103 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
2104 static int verify_commit_graph_error;
2105
2106 static void graph_report(const char *fmt, ...)
2107 {
2108 va_list ap;
2109
2110 verify_commit_graph_error = 1;
2111 va_start(ap, fmt);
2112 vfprintf(stderr, fmt, ap);
2113 fprintf(stderr, "\n");
2114 va_end(ap);
2115 }
2116
2117 #define GENERATION_ZERO_EXISTS 1
2118 #define GENERATION_NUMBER_EXISTS 2
2119
2120 int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
2121 {
2122 uint32_t i, cur_fanout_pos = 0;
2123 struct object_id prev_oid, cur_oid, checksum;
2124 int generation_zero = 0;
2125 struct hashfile *f;
2126 int devnull;
2127 struct progress *progress = NULL;
2128 int local_error = 0;
2129
2130 if (!g) {
2131 graph_report("no commit-graph file loaded");
2132 return 1;
2133 }
2134
2135 verify_commit_graph_error = verify_commit_graph_lite(g);
2136 if (verify_commit_graph_error)
2137 return verify_commit_graph_error;
2138
2139 devnull = open("/dev/null", O_WRONLY);
2140 f = hashfd(devnull, NULL);
2141 hashwrite(f, g->data, g->data_len - g->hash_len);
2142 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
2143 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
2144 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
2145 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
2146 }
2147
2148 for (i = 0; i < g->num_commits; i++) {
2149 struct commit *graph_commit;
2150
2151 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2152
2153 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
2154 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
2155 oid_to_hex(&prev_oid),
2156 oid_to_hex(&cur_oid));
2157
2158 oidcpy(&prev_oid, &cur_oid);
2159
2160 while (cur_oid.hash[0] > cur_fanout_pos) {
2161 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2162
2163 if (i != fanout_value)
2164 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2165 cur_fanout_pos, fanout_value, i);
2166 cur_fanout_pos++;
2167 }
2168
2169 graph_commit = lookup_commit(r, &cur_oid);
2170 if (!parse_commit_in_graph_one(r, g, graph_commit))
2171 graph_report(_("failed to parse commit %s from commit-graph"),
2172 oid_to_hex(&cur_oid));
2173 }
2174
2175 while (cur_fanout_pos < 256) {
2176 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2177
2178 if (g->num_commits != fanout_value)
2179 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2180 cur_fanout_pos, fanout_value, i);
2181
2182 cur_fanout_pos++;
2183 }
2184
2185 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
2186 return verify_commit_graph_error;
2187
2188 if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
2189 progress = start_progress(_("Verifying commits in commit graph"),
2190 g->num_commits);
2191
2192 for (i = 0; i < g->num_commits; i++) {
2193 struct commit *graph_commit, *odb_commit;
2194 struct commit_list *graph_parents, *odb_parents;
2195 uint32_t max_generation = 0;
2196
2197 display_progress(progress, i + 1);
2198 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2199
2200 graph_commit = lookup_commit(r, &cur_oid);
2201 odb_commit = (struct commit *)create_object(r, &cur_oid, alloc_commit_node(r));
2202 if (parse_commit_internal(odb_commit, 0, 0)) {
2203 graph_report(_("failed to parse commit %s from object database for commit-graph"),
2204 oid_to_hex(&cur_oid));
2205 continue;
2206 }
2207
2208 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
2209 get_commit_tree_oid(odb_commit)))
2210 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2211 oid_to_hex(&cur_oid),
2212 oid_to_hex(get_commit_tree_oid(graph_commit)),
2213 oid_to_hex(get_commit_tree_oid(odb_commit)));
2214
2215 graph_parents = graph_commit->parents;
2216 odb_parents = odb_commit->parents;
2217
2218 while (graph_parents) {
2219 if (odb_parents == NULL) {
2220 graph_report(_("commit-graph parent list for commit %s is too long"),
2221 oid_to_hex(&cur_oid));
2222 break;
2223 }
2224
2225 /* parse parent in case it is in a base graph */
2226 parse_commit_in_graph_one(r, g, graph_parents->item);
2227
2228 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
2229 graph_report(_("commit-graph parent for %s is %s != %s"),
2230 oid_to_hex(&cur_oid),
2231 oid_to_hex(&graph_parents->item->object.oid),
2232 oid_to_hex(&odb_parents->item->object.oid));
2233
2234 if (graph_parents->item->generation > max_generation)
2235 max_generation = graph_parents->item->generation;
2236
2237 graph_parents = graph_parents->next;
2238 odb_parents = odb_parents->next;
2239 }
2240
2241 if (odb_parents != NULL)
2242 graph_report(_("commit-graph parent list for commit %s terminates early"),
2243 oid_to_hex(&cur_oid));
2244
2245 if (!graph_commit->generation) {
2246 if (generation_zero == GENERATION_NUMBER_EXISTS)
2247 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2248 oid_to_hex(&cur_oid));
2249 generation_zero = GENERATION_ZERO_EXISTS;
2250 } else if (generation_zero == GENERATION_ZERO_EXISTS)
2251 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2252 oid_to_hex(&cur_oid));
2253
2254 if (generation_zero == GENERATION_ZERO_EXISTS)
2255 continue;
2256
2257 /*
2258 * If one of our parents has generation GENERATION_NUMBER_MAX, then
2259 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
2260 * extra logic in the following condition.
2261 */
2262 if (max_generation == GENERATION_NUMBER_MAX)
2263 max_generation--;
2264
2265 if (graph_commit->generation != max_generation + 1)
2266 graph_report(_("commit-graph generation for commit %s is %u != %u"),
2267 oid_to_hex(&cur_oid),
2268 graph_commit->generation,
2269 max_generation + 1);
2270
2271 if (graph_commit->date != odb_commit->date)
2272 graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
2273 oid_to_hex(&cur_oid),
2274 graph_commit->date,
2275 odb_commit->date);
2276 }
2277 stop_progress(&progress);
2278
2279 local_error = verify_commit_graph_error;
2280
2281 if (!(flags & COMMIT_GRAPH_VERIFY_SHALLOW) && g->base_graph)
2282 local_error |= verify_commit_graph(r, g->base_graph, flags);
2283
2284 return local_error;
2285 }
2286
2287 void free_commit_graph(struct commit_graph *g)
2288 {
2289 if (!g)
2290 return;
2291 if (g->graph_fd >= 0) {
2292 munmap((void *)g->data, g->data_len);
2293 g->data = NULL;
2294 close(g->graph_fd);
2295 }
2296 free(g->filename);
2297 free(g->bloom_filter_settings);
2298 free(g);
2299 }
2300
2301 void disable_commit_graph(struct repository *r)
2302 {
2303 r->commit_graph_disabled = 1;
2304 }
The Best VCS Around