commit-graph: extract copy_oids_to_commits()
[git/git.git] / commit-graph.c
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
20 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
21 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
22 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
23 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
24 #define GRAPH_CHUNKID_EXTRAEDGES 0x45444745 /* "EDGE" */
25
26 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
27
28 #define GRAPH_VERSION_1 0x1
29 #define GRAPH_VERSION GRAPH_VERSION_1
30
31 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
32 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
33 #define GRAPH_PARENT_NONE 0x70000000
34
35 #define GRAPH_LAST_EDGE 0x80000000
36
37 #define GRAPH_HEADER_SIZE 8
38 #define GRAPH_FANOUT_SIZE (4 * 256)
39 #define GRAPH_CHUNKLOOKUP_WIDTH 12
40 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
41 + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
42
43 char *get_commit_graph_filename(const char *obj_dir)
44 {
45 return xstrfmt("%s/info/commit-graph", obj_dir);
46 }
47
48 static uint8_t oid_version(void)
49 {
50 return 1;
51 }
52
53 static struct commit_graph *alloc_commit_graph(void)
54 {
55 struct commit_graph *g = xcalloc(1, sizeof(*g));
56 g->graph_fd = -1;
57
58 return g;
59 }
60
61 extern int read_replace_refs;
62
63 static int commit_graph_compatible(struct repository *r)
64 {
65 if (!r->gitdir)
66 return 0;
67
68 if (read_replace_refs) {
69 prepare_replace_object(r);
70 if (hashmap_get_size(&r->objects->replace_map->map))
71 return 0;
72 }
73
74 prepare_commit_graft(r);
75 if (r->parsed_objects && r->parsed_objects->grafts_nr)
76 return 0;
77 if (is_repository_shallow(r))
78 return 0;
79
80 return 1;
81 }
82
83 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
84 {
85 *fd = git_open(graph_file);
86 if (*fd < 0)
87 return 0;
88 if (fstat(*fd, st)) {
89 close(*fd);
90 return 0;
91 }
92 return 1;
93 }
94
95 struct commit_graph *load_commit_graph_one_fd_st(int fd, struct stat *st)
96 {
97 void *graph_map;
98 size_t graph_size;
99 struct commit_graph *ret;
100
101 graph_size = xsize_t(st->st_size);
102
103 if (graph_size < GRAPH_MIN_SIZE) {
104 close(fd);
105 error(_("commit-graph file is too small"));
106 return NULL;
107 }
108 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
109 ret = parse_commit_graph(graph_map, fd, graph_size);
110
111 if (!ret) {
112 munmap(graph_map, graph_size);
113 close(fd);
114 }
115
116 return ret;
117 }
118
119 static int verify_commit_graph_lite(struct commit_graph *g)
120 {
121 /*
122 * Basic validation shared between parse_commit_graph()
123 * which'll be called every time the graph is used, and the
124 * much more expensive verify_commit_graph() used by
125 * "commit-graph verify".
126 *
127 * There should only be very basic checks here to ensure that
128 * we don't e.g. segfault in fill_commit_in_graph(), but
129 * because this is a very hot codepath nothing that e.g. loops
130 * over g->num_commits, or runs a checksum on the commit-graph
131 * itself.
132 */
133 if (!g->chunk_oid_fanout) {
134 error("commit-graph is missing the OID Fanout chunk");
135 return 1;
136 }
137 if (!g->chunk_oid_lookup) {
138 error("commit-graph is missing the OID Lookup chunk");
139 return 1;
140 }
141 if (!g->chunk_commit_data) {
142 error("commit-graph is missing the Commit Data chunk");
143 return 1;
144 }
145
146 return 0;
147 }
148
149 struct commit_graph *parse_commit_graph(void *graph_map, int fd,
150 size_t graph_size)
151 {
152 const unsigned char *data, *chunk_lookup;
153 uint32_t i;
154 struct commit_graph *graph;
155 uint64_t last_chunk_offset;
156 uint32_t last_chunk_id;
157 uint32_t graph_signature;
158 unsigned char graph_version, hash_version;
159
160 if (!graph_map)
161 return NULL;
162
163 if (graph_size < GRAPH_MIN_SIZE)
164 return NULL;
165
166 data = (const unsigned char *)graph_map;
167
168 graph_signature = get_be32(data);
169 if (graph_signature != GRAPH_SIGNATURE) {
170 error(_("commit-graph signature %X does not match signature %X"),
171 graph_signature, GRAPH_SIGNATURE);
172 return NULL;
173 }
174
175 graph_version = *(unsigned char*)(data + 4);
176 if (graph_version != GRAPH_VERSION) {
177 error(_("commit-graph version %X does not match version %X"),
178 graph_version, GRAPH_VERSION);
179 return NULL;
180 }
181
182 hash_version = *(unsigned char*)(data + 5);
183 if (hash_version != oid_version()) {
184 error(_("commit-graph hash version %X does not match version %X"),
185 hash_version, oid_version());
186 return NULL;
187 }
188
189 graph = alloc_commit_graph();
190
191 graph->hash_len = the_hash_algo->rawsz;
192 graph->num_chunks = *(unsigned char*)(data + 6);
193 graph->graph_fd = fd;
194 graph->data = graph_map;
195 graph->data_len = graph_size;
196
197 last_chunk_id = 0;
198 last_chunk_offset = 8;
199 chunk_lookup = data + 8;
200 for (i = 0; i < graph->num_chunks; i++) {
201 uint32_t chunk_id;
202 uint64_t chunk_offset;
203 int chunk_repeated = 0;
204
205 if (data + graph_size - chunk_lookup <
206 GRAPH_CHUNKLOOKUP_WIDTH) {
207 error(_("commit-graph chunk lookup table entry missing; file may be incomplete"));
208 free(graph);
209 return NULL;
210 }
211
212 chunk_id = get_be32(chunk_lookup + 0);
213 chunk_offset = get_be64(chunk_lookup + 4);
214
215 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
216
217 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
218 error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
219 (uint32_t)chunk_offset);
220 free(graph);
221 return NULL;
222 }
223
224 switch (chunk_id) {
225 case GRAPH_CHUNKID_OIDFANOUT:
226 if (graph->chunk_oid_fanout)
227 chunk_repeated = 1;
228 else
229 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
230 break;
231
232 case GRAPH_CHUNKID_OIDLOOKUP:
233 if (graph->chunk_oid_lookup)
234 chunk_repeated = 1;
235 else
236 graph->chunk_oid_lookup = data + chunk_offset;
237 break;
238
239 case GRAPH_CHUNKID_DATA:
240 if (graph->chunk_commit_data)
241 chunk_repeated = 1;
242 else
243 graph->chunk_commit_data = data + chunk_offset;
244 break;
245
246 case GRAPH_CHUNKID_EXTRAEDGES:
247 if (graph->chunk_extra_edges)
248 chunk_repeated = 1;
249 else
250 graph->chunk_extra_edges = data + chunk_offset;
251 break;
252 }
253
254 if (chunk_repeated) {
255 error(_("commit-graph chunk id %08x appears multiple times"), chunk_id);
256 free(graph);
257 return NULL;
258 }
259
260 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
261 {
262 graph->num_commits = (chunk_offset - last_chunk_offset)
263 / graph->hash_len;
264 }
265
266 last_chunk_id = chunk_id;
267 last_chunk_offset = chunk_offset;
268 }
269
270 if (verify_commit_graph_lite(graph))
271 return NULL;
272
273 return graph;
274 }
275
276 static struct commit_graph *load_commit_graph_one(const char *graph_file)
277 {
278
279 struct stat st;
280 int fd;
281 int open_ok = open_commit_graph(graph_file, &fd, &st);
282
283 if (!open_ok)
284 return NULL;
285
286 return load_commit_graph_one_fd_st(fd, &st);
287 }
288
289 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
290 {
291 char *graph_name;
292
293 if (r->objects->commit_graph)
294 return;
295
296 graph_name = get_commit_graph_filename(obj_dir);
297 r->objects->commit_graph =
298 load_commit_graph_one(graph_name);
299
300 FREE_AND_NULL(graph_name);
301 }
302
303 /*
304 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
305 *
306 * On the first invocation, this function attemps to load the commit
307 * graph if the_repository is configured to have one.
308 */
309 static int prepare_commit_graph(struct repository *r)
310 {
311 struct object_directory *odb;
312 int config_value;
313
314 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
315 die("dying as requested by the '%s' variable on commit-graph load!",
316 GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD);
317
318 if (r->objects->commit_graph_attempted)
319 return !!r->objects->commit_graph;
320 r->objects->commit_graph_attempted = 1;
321
322 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
323 (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
324 !config_value))
325 /*
326 * This repository is not configured to use commit graphs, so
327 * do not load one. (But report commit_graph_attempted anyway
328 * so that commit graph loading is not attempted again for this
329 * repository.)
330 */
331 return 0;
332
333 if (!commit_graph_compatible(r))
334 return 0;
335
336 prepare_alt_odb(r);
337 for (odb = r->objects->odb;
338 !r->objects->commit_graph && odb;
339 odb = odb->next)
340 prepare_commit_graph_one(r, odb->path);
341 return !!r->objects->commit_graph;
342 }
343
344 int generation_numbers_enabled(struct repository *r)
345 {
346 uint32_t first_generation;
347 struct commit_graph *g;
348 if (!prepare_commit_graph(r))
349 return 0;
350
351 g = r->objects->commit_graph;
352
353 if (!g->num_commits)
354 return 0;
355
356 first_generation = get_be32(g->chunk_commit_data +
357 g->hash_len + 8) >> 2;
358
359 return !!first_generation;
360 }
361
362 void close_commit_graph(struct repository *r)
363 {
364 free_commit_graph(r->objects->commit_graph);
365 r->objects->commit_graph = NULL;
366 }
367
368 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
369 {
370 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
371 g->chunk_oid_lookup, g->hash_len, pos);
372 }
373
374 static struct commit_list **insert_parent_or_die(struct repository *r,
375 struct commit_graph *g,
376 uint64_t pos,
377 struct commit_list **pptr)
378 {
379 struct commit *c;
380 struct object_id oid;
381
382 if (pos >= g->num_commits)
383 die("invalid parent position %"PRIu64, pos);
384
385 hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
386 c = lookup_commit(r, &oid);
387 if (!c)
388 die(_("could not find commit %s"), oid_to_hex(&oid));
389 c->graph_pos = pos;
390 return &commit_list_insert(c, pptr)->next;
391 }
392
393 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
394 {
395 const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
396 item->graph_pos = pos;
397 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
398 }
399
400 static int fill_commit_in_graph(struct repository *r,
401 struct commit *item,
402 struct commit_graph *g, uint32_t pos)
403 {
404 uint32_t edge_value;
405 uint32_t *parent_data_ptr;
406 uint64_t date_low, date_high;
407 struct commit_list **pptr;
408 const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
409
410 item->object.parsed = 1;
411 item->graph_pos = pos;
412
413 item->maybe_tree = NULL;
414
415 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
416 date_low = get_be32(commit_data + g->hash_len + 12);
417 item->date = (timestamp_t)((date_high << 32) | date_low);
418
419 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
420
421 pptr = &item->parents;
422
423 edge_value = get_be32(commit_data + g->hash_len);
424 if (edge_value == GRAPH_PARENT_NONE)
425 return 1;
426 pptr = insert_parent_or_die(r, g, edge_value, pptr);
427
428 edge_value = get_be32(commit_data + g->hash_len + 4);
429 if (edge_value == GRAPH_PARENT_NONE)
430 return 1;
431 if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
432 pptr = insert_parent_or_die(r, g, edge_value, pptr);
433 return 1;
434 }
435
436 parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
437 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
438 do {
439 edge_value = get_be32(parent_data_ptr);
440 pptr = insert_parent_or_die(r, g,
441 edge_value & GRAPH_EDGE_LAST_MASK,
442 pptr);
443 parent_data_ptr++;
444 } while (!(edge_value & GRAPH_LAST_EDGE));
445
446 return 1;
447 }
448
449 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
450 {
451 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
452 *pos = item->graph_pos;
453 return 1;
454 } else {
455 return bsearch_graph(g, &(item->object.oid), pos);
456 }
457 }
458
459 static int parse_commit_in_graph_one(struct repository *r,
460 struct commit_graph *g,
461 struct commit *item)
462 {
463 uint32_t pos;
464
465 if (item->object.parsed)
466 return 1;
467
468 if (find_commit_in_graph(item, g, &pos))
469 return fill_commit_in_graph(r, item, g, pos);
470
471 return 0;
472 }
473
474 int parse_commit_in_graph(struct repository *r, struct commit *item)
475 {
476 if (!prepare_commit_graph(r))
477 return 0;
478 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
479 }
480
481 void load_commit_graph_info(struct repository *r, struct commit *item)
482 {
483 uint32_t pos;
484 if (!prepare_commit_graph(r))
485 return;
486 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
487 fill_commit_graph_info(item, r->objects->commit_graph, pos);
488 }
489
490 static struct tree *load_tree_for_commit(struct repository *r,
491 struct commit_graph *g,
492 struct commit *c)
493 {
494 struct object_id oid;
495 const unsigned char *commit_data = g->chunk_commit_data +
496 GRAPH_DATA_WIDTH * (c->graph_pos);
497
498 hashcpy(oid.hash, commit_data);
499 c->maybe_tree = lookup_tree(r, &oid);
500
501 return c->maybe_tree;
502 }
503
504 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
505 struct commit_graph *g,
506 const struct commit *c)
507 {
508 if (c->maybe_tree)
509 return c->maybe_tree;
510 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
511 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
512
513 return load_tree_for_commit(r, g, (struct commit *)c);
514 }
515
516 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
517 {
518 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
519 }
520
521 struct packed_commit_list {
522 struct commit **list;
523 int nr;
524 int alloc;
525 };
526
527 struct packed_oid_list {
528 struct object_id *list;
529 int nr;
530 int alloc;
531 };
532
533 struct write_commit_graph_context {
534 struct repository *r;
535 const char *obj_dir;
536 char *graph_name;
537 struct packed_oid_list oids;
538 struct packed_commit_list commits;
539 int num_extra_edges;
540 unsigned long approx_nr_objects;
541 struct progress *progress;
542 int progress_done;
543 uint64_t progress_cnt;
544 unsigned append:1,
545 report_progress:1;
546 };
547
548 static void write_graph_chunk_fanout(struct hashfile *f,
549 struct write_commit_graph_context *ctx)
550 {
551 int i, count = 0;
552 struct commit **list = ctx->commits.list;
553
554 /*
555 * Write the first-level table (the list is sorted,
556 * but we use a 256-entry lookup to be able to avoid
557 * having to do eight extra binary search iterations).
558 */
559 for (i = 0; i < 256; i++) {
560 while (count < ctx->commits.nr) {
561 if ((*list)->object.oid.hash[0] != i)
562 break;
563 display_progress(ctx->progress, ++ctx->progress_cnt);
564 count++;
565 list++;
566 }
567
568 hashwrite_be32(f, count);
569 }
570 }
571
572 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
573 struct write_commit_graph_context *ctx)
574 {
575 struct commit **list = ctx->commits.list;
576 int count;
577 for (count = 0; count < ctx->commits.nr; count++, list++) {
578 display_progress(ctx->progress, ++ctx->progress_cnt);
579 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
580 }
581 }
582
583 static const unsigned char *commit_to_sha1(size_t index, void *table)
584 {
585 struct commit **commits = table;
586 return commits[index]->object.oid.hash;
587 }
588
589 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
590 struct write_commit_graph_context *ctx)
591 {
592 struct commit **list = ctx->commits.list;
593 struct commit **last = ctx->commits.list + ctx->commits.nr;
594 uint32_t num_extra_edges = 0;
595
596 while (list < last) {
597 struct commit_list *parent;
598 int edge_value;
599 uint32_t packedDate[2];
600 display_progress(ctx->progress, ++ctx->progress_cnt);
601
602 parse_commit_no_graph(*list);
603 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
604
605 parent = (*list)->parents;
606
607 if (!parent)
608 edge_value = GRAPH_PARENT_NONE;
609 else {
610 edge_value = sha1_pos(parent->item->object.oid.hash,
611 ctx->commits.list,
612 ctx->commits.nr,
613 commit_to_sha1);
614
615 if (edge_value < 0)
616 BUG("missing parent %s for commit %s",
617 oid_to_hex(&parent->item->object.oid),
618 oid_to_hex(&(*list)->object.oid));
619 }
620
621 hashwrite_be32(f, edge_value);
622
623 if (parent)
624 parent = parent->next;
625
626 if (!parent)
627 edge_value = GRAPH_PARENT_NONE;
628 else if (parent->next)
629 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
630 else {
631 edge_value = sha1_pos(parent->item->object.oid.hash,
632 ctx->commits.list,
633 ctx->commits.nr,
634 commit_to_sha1);
635 if (edge_value < 0)
636 BUG("missing parent %s for commit %s",
637 oid_to_hex(&parent->item->object.oid),
638 oid_to_hex(&(*list)->object.oid));
639 }
640
641 hashwrite_be32(f, edge_value);
642
643 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
644 do {
645 num_extra_edges++;
646 parent = parent->next;
647 } while (parent);
648 }
649
650 if (sizeof((*list)->date) > 4)
651 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
652 else
653 packedDate[0] = 0;
654
655 packedDate[0] |= htonl((*list)->generation << 2);
656
657 packedDate[1] = htonl((*list)->date);
658 hashwrite(f, packedDate, 8);
659
660 list++;
661 }
662 }
663
664 static void write_graph_chunk_extra_edges(struct hashfile *f,
665 struct write_commit_graph_context *ctx)
666 {
667 struct commit **list = ctx->commits.list;
668 struct commit **last = ctx->commits.list + ctx->commits.nr;
669 struct commit_list *parent;
670
671 while (list < last) {
672 int num_parents = 0;
673
674 display_progress(ctx->progress, ++ctx->progress_cnt);
675
676 for (parent = (*list)->parents; num_parents < 3 && parent;
677 parent = parent->next)
678 num_parents++;
679
680 if (num_parents <= 2) {
681 list++;
682 continue;
683 }
684
685 /* Since num_parents > 2, this initializer is safe. */
686 for (parent = (*list)->parents->next; parent; parent = parent->next) {
687 int edge_value = sha1_pos(parent->item->object.oid.hash,
688 ctx->commits.list,
689 ctx->commits.nr,
690 commit_to_sha1);
691
692 if (edge_value < 0)
693 BUG("missing parent %s for commit %s",
694 oid_to_hex(&parent->item->object.oid),
695 oid_to_hex(&(*list)->object.oid));
696 else if (!parent->next)
697 edge_value |= GRAPH_LAST_EDGE;
698
699 hashwrite_be32(f, edge_value);
700 }
701
702 list++;
703 }
704 }
705
706 static int commit_compare(const void *_a, const void *_b)
707 {
708 const struct object_id *a = (const struct object_id *)_a;
709 const struct object_id *b = (const struct object_id *)_b;
710 return oidcmp(a, b);
711 }
712
713 static int add_packed_commits(const struct object_id *oid,
714 struct packed_git *pack,
715 uint32_t pos,
716 void *data)
717 {
718 struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
719 enum object_type type;
720 off_t offset = nth_packed_object_offset(pack, pos);
721 struct object_info oi = OBJECT_INFO_INIT;
722
723 if (ctx->progress)
724 display_progress(ctx->progress, ++ctx->progress_done);
725
726 oi.typep = &type;
727 if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
728 die(_("unable to get type of object %s"), oid_to_hex(oid));
729
730 if (type != OBJ_COMMIT)
731 return 0;
732
733 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
734 oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
735 ctx->oids.nr++;
736
737 return 0;
738 }
739
740 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
741 {
742 struct commit_list *parent;
743 for (parent = commit->parents; parent; parent = parent->next) {
744 if (!(parent->item->object.flags & UNINTERESTING)) {
745 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
746 oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
747 ctx->oids.nr++;
748 parent->item->object.flags |= UNINTERESTING;
749 }
750 }
751 }
752
753 static void close_reachable(struct write_commit_graph_context *ctx)
754 {
755 int i;
756 struct commit *commit;
757
758 if (ctx->report_progress)
759 ctx->progress = start_delayed_progress(
760 _("Loading known commits in commit graph"),
761 ctx->oids.nr);
762 for (i = 0; i < ctx->oids.nr; i++) {
763 display_progress(ctx->progress, i + 1);
764 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
765 if (commit)
766 commit->object.flags |= UNINTERESTING;
767 }
768 stop_progress(&ctx->progress);
769
770 /*
771 * As this loop runs, ctx->oids.nr may grow, but not more
772 * than the number of missing commits in the reachable
773 * closure.
774 */
775 if (ctx->report_progress)
776 ctx->progress = start_delayed_progress(
777 _("Expanding reachable commits in commit graph"),
778 ctx->oids.nr);
779 for (i = 0; i < ctx->oids.nr; i++) {
780 display_progress(ctx->progress, i + 1);
781 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
782
783 if (commit && !parse_commit_no_graph(commit))
784 add_missing_parents(ctx, commit);
785 }
786 stop_progress(&ctx->progress);
787
788 if (ctx->report_progress)
789 ctx->progress = start_delayed_progress(
790 _("Clearing commit marks in commit graph"),
791 ctx->oids.nr);
792 for (i = 0; i < ctx->oids.nr; i++) {
793 display_progress(ctx->progress, i + 1);
794 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
795
796 if (commit)
797 commit->object.flags &= ~UNINTERESTING;
798 }
799 stop_progress(&ctx->progress);
800 }
801
802 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
803 {
804 int i;
805 struct commit_list *list = NULL;
806
807 if (ctx->report_progress)
808 ctx->progress = start_progress(
809 _("Computing commit graph generation numbers"),
810 ctx->commits.nr);
811 for (i = 0; i < ctx->commits.nr; i++) {
812 display_progress(ctx->progress, i + 1);
813 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
814 ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
815 continue;
816
817 commit_list_insert(ctx->commits.list[i], &list);
818 while (list) {
819 struct commit *current = list->item;
820 struct commit_list *parent;
821 int all_parents_computed = 1;
822 uint32_t max_generation = 0;
823
824 for (parent = current->parents; parent; parent = parent->next) {
825 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
826 parent->item->generation == GENERATION_NUMBER_ZERO) {
827 all_parents_computed = 0;
828 commit_list_insert(parent->item, &list);
829 break;
830 } else if (parent->item->generation > max_generation) {
831 max_generation = parent->item->generation;
832 }
833 }
834
835 if (all_parents_computed) {
836 current->generation = max_generation + 1;
837 pop_commit(&list);
838
839 if (current->generation > GENERATION_NUMBER_MAX)
840 current->generation = GENERATION_NUMBER_MAX;
841 }
842 }
843 }
844 stop_progress(&ctx->progress);
845 }
846
847 static int add_ref_to_list(const char *refname,
848 const struct object_id *oid,
849 int flags, void *cb_data)
850 {
851 struct string_list *list = (struct string_list *)cb_data;
852
853 string_list_append(list, oid_to_hex(oid));
854 return 0;
855 }
856
857 int write_commit_graph_reachable(const char *obj_dir, unsigned int flags)
858 {
859 struct string_list list = STRING_LIST_INIT_DUP;
860 int result;
861
862 for_each_ref(add_ref_to_list, &list);
863 result = write_commit_graph(obj_dir, NULL, &list,
864 flags);
865
866 string_list_clear(&list, 0);
867 return result;
868 }
869
870 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
871 struct string_list *pack_indexes)
872 {
873 uint32_t i;
874 struct strbuf progress_title = STRBUF_INIT;
875 struct strbuf packname = STRBUF_INIT;
876 int dirlen;
877
878 strbuf_addf(&packname, "%s/pack/", ctx->obj_dir);
879 dirlen = packname.len;
880 if (ctx->report_progress) {
881 strbuf_addf(&progress_title,
882 Q_("Finding commits for commit graph in %d pack",
883 "Finding commits for commit graph in %d packs",
884 pack_indexes->nr),
885 pack_indexes->nr);
886 ctx->progress = start_delayed_progress(progress_title.buf, 0);
887 ctx->progress_done = 0;
888 }
889 for (i = 0; i < pack_indexes->nr; i++) {
890 struct packed_git *p;
891 strbuf_setlen(&packname, dirlen);
892 strbuf_addstr(&packname, pack_indexes->items[i].string);
893 p = add_packed_git(packname.buf, packname.len, 1);
894 if (!p) {
895 error(_("error adding pack %s"), packname.buf);
896 return -1;
897 }
898 if (open_pack_index(p)) {
899 error(_("error opening index for %s"), packname.buf);
900 return -1;
901 }
902 for_each_object_in_pack(p, add_packed_commits, ctx,
903 FOR_EACH_OBJECT_PACK_ORDER);
904 close_pack(p);
905 free(p);
906 }
907
908 stop_progress(&ctx->progress);
909 strbuf_reset(&progress_title);
910 strbuf_release(&packname);
911
912 return 0;
913 }
914
915 static void fill_oids_from_commit_hex(struct write_commit_graph_context *ctx,
916 struct string_list *commit_hex)
917 {
918 uint32_t i;
919 struct strbuf progress_title = STRBUF_INIT;
920
921 if (ctx->report_progress) {
922 strbuf_addf(&progress_title,
923 Q_("Finding commits for commit graph from %d ref",
924 "Finding commits for commit graph from %d refs",
925 commit_hex->nr),
926 commit_hex->nr);
927 ctx->progress = start_delayed_progress(
928 progress_title.buf,
929 commit_hex->nr);
930 }
931 for (i = 0; i < commit_hex->nr; i++) {
932 const char *end;
933 struct object_id oid;
934 struct commit *result;
935
936 display_progress(ctx->progress, i + 1);
937 if (commit_hex->items[i].string &&
938 parse_oid_hex(commit_hex->items[i].string, &oid, &end))
939 continue;
940
941 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
942
943 if (result) {
944 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
945 oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
946 ctx->oids.nr++;
947 }
948 }
949 stop_progress(&ctx->progress);
950 strbuf_release(&progress_title);
951 }
952
953 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
954 {
955 if (ctx->report_progress)
956 ctx->progress = start_delayed_progress(
957 _("Finding commits for commit graph among packed objects"),
958 ctx->approx_nr_objects);
959 for_each_packed_object(add_packed_commits, ctx,
960 FOR_EACH_OBJECT_PACK_ORDER);
961 if (ctx->progress_done < ctx->approx_nr_objects)
962 display_progress(ctx->progress, ctx->approx_nr_objects);
963 stop_progress(&ctx->progress);
964 }
965
966 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
967 {
968 uint32_t i, count_distinct = 1;
969
970 if (ctx->report_progress)
971 ctx->progress = start_delayed_progress(
972 _("Counting distinct commits in commit graph"),
973 ctx->oids.nr);
974 display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
975 QSORT(ctx->oids.list, ctx->oids.nr, commit_compare);
976
977 for (i = 1; i < ctx->oids.nr; i++) {
978 display_progress(ctx->progress, i + 1);
979 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
980 count_distinct++;
981 }
982 stop_progress(&ctx->progress);
983
984 return count_distinct;
985 }
986
987 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
988 {
989 uint32_t i;
990 struct commit_list *parent;
991
992 ctx->num_extra_edges = 0;
993 if (ctx->report_progress)
994 ctx->progress = start_delayed_progress(
995 _("Finding extra edges in commit graph"),
996 ctx->oids.nr);
997 for (i = 0; i < ctx->oids.nr; i++) {
998 int num_parents = 0;
999 display_progress(ctx->progress, i + 1);
1000 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1001 continue;
1002
1003 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1004 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1005
1006 for (parent = ctx->commits.list[ctx->commits.nr]->parents;
1007 parent; parent = parent->next)
1008 num_parents++;
1009
1010 if (num_parents > 2)
1011 ctx->num_extra_edges += num_parents - 1;
1012
1013 ctx->commits.nr++;
1014 }
1015 stop_progress(&ctx->progress);
1016 }
1017
1018 int write_commit_graph(const char *obj_dir,
1019 struct string_list *pack_indexes,
1020 struct string_list *commit_hex,
1021 unsigned int flags)
1022 {
1023 struct write_commit_graph_context *ctx;
1024 struct hashfile *f;
1025 uint32_t i, count_distinct = 0;
1026 char *graph_name = NULL;
1027 struct lock_file lk = LOCK_INIT;
1028 uint32_t chunk_ids[5];
1029 uint64_t chunk_offsets[5];
1030 int num_chunks;
1031 const unsigned hashsz = the_hash_algo->rawsz;
1032 struct strbuf progress_title = STRBUF_INIT;
1033 int res = 0;
1034
1035 if (!commit_graph_compatible(the_repository))
1036 return 0;
1037
1038 ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
1039 ctx->r = the_repository;
1040 ctx->obj_dir = obj_dir;
1041 ctx->append = flags & COMMIT_GRAPH_APPEND ? 1 : 0;
1042 ctx->report_progress = flags & COMMIT_GRAPH_PROGRESS ? 1 : 0;
1043
1044 ctx->approx_nr_objects = approximate_object_count();
1045 ctx->oids.alloc = ctx->approx_nr_objects / 32;
1046
1047 if (ctx->append) {
1048 prepare_commit_graph_one(ctx->r, ctx->obj_dir);
1049 if (ctx->r->objects->commit_graph)
1050 ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
1051 }
1052
1053 if (ctx->oids.alloc < 1024)
1054 ctx->oids.alloc = 1024;
1055 ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
1056
1057 if (ctx->append && ctx->r->objects->commit_graph) {
1058 struct commit_graph *g = ctx->r->objects->commit_graph;
1059 for (i = 0; i < g->num_commits; i++) {
1060 const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
1061 hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
1062 }
1063 }
1064
1065 if (pack_indexes) {
1066 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
1067 goto cleanup;
1068 }
1069
1070 if (commit_hex)
1071 fill_oids_from_commit_hex(ctx, commit_hex);
1072
1073 if (!pack_indexes && !commit_hex)
1074 fill_oids_from_all_packs(ctx);
1075
1076 close_reachable(ctx);
1077
1078 count_distinct = count_distinct_commits(ctx);
1079
1080 if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
1081 error(_("the commit graph format cannot write %d commits"), count_distinct);
1082 res = -1;
1083 goto cleanup;
1084 }
1085
1086 ctx->commits.alloc = count_distinct;
1087 ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
1088
1089 copy_oids_to_commits(ctx);
1090
1091 if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
1092 error(_("too many commits to write graph"));
1093 res = -1;
1094 goto cleanup;
1095 }
1096
1097 compute_generation_numbers(ctx);
1098
1099 num_chunks = ctx->num_extra_edges ? 4 : 3;
1100
1101 ctx->graph_name = get_commit_graph_filename(ctx->obj_dir);
1102 if (safe_create_leading_directories(ctx->graph_name)) {
1103 UNLEAK(ctx->graph_name);
1104 error(_("unable to create leading directories of %s"),
1105 ctx->graph_name);
1106 res = -1;
1107 goto cleanup;
1108 }
1109
1110 hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
1111 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1112
1113 hashwrite_be32(f, GRAPH_SIGNATURE);
1114
1115 hashwrite_u8(f, GRAPH_VERSION);
1116 hashwrite_u8(f, oid_version());
1117 hashwrite_u8(f, num_chunks);
1118 hashwrite_u8(f, 0); /* unused padding byte */
1119
1120 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1121 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1122 chunk_ids[2] = GRAPH_CHUNKID_DATA;
1123 if (ctx->num_extra_edges)
1124 chunk_ids[3] = GRAPH_CHUNKID_EXTRAEDGES;
1125 else
1126 chunk_ids[3] = 0;
1127 chunk_ids[4] = 0;
1128
1129 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1130 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1131 chunk_offsets[2] = chunk_offsets[1] + hashsz * ctx->commits.nr;
1132 chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * ctx->commits.nr;
1133 chunk_offsets[4] = chunk_offsets[3] + 4 * ctx->num_extra_edges;
1134
1135 for (i = 0; i <= num_chunks; i++) {
1136 uint32_t chunk_write[3];
1137
1138 chunk_write[0] = htonl(chunk_ids[i]);
1139 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1140 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1141 hashwrite(f, chunk_write, 12);
1142 }
1143
1144 if (ctx->report_progress) {
1145 strbuf_addf(&progress_title,
1146 Q_("Writing out commit graph in %d pass",
1147 "Writing out commit graph in %d passes",
1148 num_chunks),
1149 num_chunks);
1150 ctx->progress = start_delayed_progress(
1151 progress_title.buf,
1152 num_chunks * ctx->commits.nr);
1153 }
1154 write_graph_chunk_fanout(f, ctx);
1155 write_graph_chunk_oids(f, hashsz, ctx);
1156 write_graph_chunk_data(f, hashsz, ctx);
1157 if (ctx->num_extra_edges)
1158 write_graph_chunk_extra_edges(f, ctx);
1159 stop_progress(&ctx->progress);
1160 strbuf_release(&progress_title);
1161
1162 close_commit_graph(ctx->r);
1163 finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1164 commit_lock_file(&lk);
1165
1166 cleanup:
1167 free(graph_name);
1168 free(ctx->commits.list);
1169 free(ctx->oids.list);
1170 free(ctx);
1171
1172 return res;
1173 }
1174
1175 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1176 static int verify_commit_graph_error;
1177
1178 static void graph_report(const char *fmt, ...)
1179 {
1180 va_list ap;
1181
1182 verify_commit_graph_error = 1;
1183 va_start(ap, fmt);
1184 vfprintf(stderr, fmt, ap);
1185 fprintf(stderr, "\n");
1186 va_end(ap);
1187 }
1188
1189 #define GENERATION_ZERO_EXISTS 1
1190 #define GENERATION_NUMBER_EXISTS 2
1191
1192 int verify_commit_graph(struct repository *r, struct commit_graph *g)
1193 {
1194 uint32_t i, cur_fanout_pos = 0;
1195 struct object_id prev_oid, cur_oid, checksum;
1196 int generation_zero = 0;
1197 struct hashfile *f;
1198 int devnull;
1199 struct progress *progress = NULL;
1200
1201 if (!g) {
1202 graph_report("no commit-graph file loaded");
1203 return 1;
1204 }
1205
1206 verify_commit_graph_error = verify_commit_graph_lite(g);
1207 if (verify_commit_graph_error)
1208 return verify_commit_graph_error;
1209
1210 devnull = open("/dev/null", O_WRONLY);
1211 f = hashfd(devnull, NULL);
1212 hashwrite(f, g->data, g->data_len - g->hash_len);
1213 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1214 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1215 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1216 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1217 }
1218
1219 for (i = 0; i < g->num_commits; i++) {
1220 struct commit *graph_commit;
1221
1222 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1223
1224 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1225 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
1226 oid_to_hex(&prev_oid),
1227 oid_to_hex(&cur_oid));
1228
1229 oidcpy(&prev_oid, &cur_oid);
1230
1231 while (cur_oid.hash[0] > cur_fanout_pos) {
1232 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1233
1234 if (i != fanout_value)
1235 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1236 cur_fanout_pos, fanout_value, i);
1237 cur_fanout_pos++;
1238 }
1239
1240 graph_commit = lookup_commit(r, &cur_oid);
1241 if (!parse_commit_in_graph_one(r, g, graph_commit))
1242 graph_report(_("failed to parse commit %s from commit-graph"),
1243 oid_to_hex(&cur_oid));
1244 }
1245
1246 while (cur_fanout_pos < 256) {
1247 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1248
1249 if (g->num_commits != fanout_value)
1250 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1251 cur_fanout_pos, fanout_value, i);
1252
1253 cur_fanout_pos++;
1254 }
1255
1256 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1257 return verify_commit_graph_error;
1258
1259 progress = start_progress(_("Verifying commits in commit graph"),
1260 g->num_commits);
1261 for (i = 0; i < g->num_commits; i++) {
1262 struct commit *graph_commit, *odb_commit;
1263 struct commit_list *graph_parents, *odb_parents;
1264 uint32_t max_generation = 0;
1265
1266 display_progress(progress, i + 1);
1267 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1268
1269 graph_commit = lookup_commit(r, &cur_oid);
1270 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1271 if (parse_commit_internal(odb_commit, 0, 0)) {
1272 graph_report(_("failed to parse commit %s from object database for commit-graph"),
1273 oid_to_hex(&cur_oid));
1274 continue;
1275 }
1276
1277 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
1278 get_commit_tree_oid(odb_commit)))
1279 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
1280 oid_to_hex(&cur_oid),
1281 oid_to_hex(get_commit_tree_oid(graph_commit)),
1282 oid_to_hex(get_commit_tree_oid(odb_commit)));
1283
1284 graph_parents = graph_commit->parents;
1285 odb_parents = odb_commit->parents;
1286
1287 while (graph_parents) {
1288 if (odb_parents == NULL) {
1289 graph_report(_("commit-graph parent list for commit %s is too long"),
1290 oid_to_hex(&cur_oid));
1291 break;
1292 }
1293
1294 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1295 graph_report(_("commit-graph parent for %s is %s != %s"),
1296 oid_to_hex(&cur_oid),
1297 oid_to_hex(&graph_parents->item->object.oid),
1298 oid_to_hex(&odb_parents->item->object.oid));
1299
1300 if (graph_parents->item->generation > max_generation)
1301 max_generation = graph_parents->item->generation;
1302
1303 graph_parents = graph_parents->next;
1304 odb_parents = odb_parents->next;
1305 }
1306
1307 if (odb_parents != NULL)
1308 graph_report(_("commit-graph parent list for commit %s terminates early"),
1309 oid_to_hex(&cur_oid));
1310
1311 if (!graph_commit->generation) {
1312 if (generation_zero == GENERATION_NUMBER_EXISTS)
1313 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
1314 oid_to_hex(&cur_oid));
1315 generation_zero = GENERATION_ZERO_EXISTS;
1316 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1317 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
1318 oid_to_hex(&cur_oid));
1319
1320 if (generation_zero == GENERATION_ZERO_EXISTS)
1321 continue;
1322
1323 /*
1324 * If one of our parents has generation GENERATION_NUMBER_MAX, then
1325 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1326 * extra logic in the following condition.
1327 */
1328 if (max_generation == GENERATION_NUMBER_MAX)
1329 max_generation--;
1330
1331 if (graph_commit->generation != max_generation + 1)
1332 graph_report(_("commit-graph generation for commit %s is %u != %u"),
1333 oid_to_hex(&cur_oid),
1334 graph_commit->generation,
1335 max_generation + 1);
1336
1337 if (graph_commit->date != odb_commit->date)
1338 graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
1339 oid_to_hex(&cur_oid),
1340 graph_commit->date,
1341 odb_commit->date);
1342 }
1343 stop_progress(&progress);
1344
1345 return verify_commit_graph_error;
1346 }
1347
1348 void free_commit_graph(struct commit_graph *g)
1349 {
1350 if (!g)
1351 return;
1352 if (g->graph_fd >= 0) {
1353 munmap((void *)g->data, g->data_len);
1354 g->data = NULL;
1355 close(g->graph_fd);
1356 }
1357 free(g);
1358 }