Merge branch 'jc/post-c89-rules-doc' into maint
[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 free(graph);
272 return NULL;
273 }
274
275 return graph;
276 }
277
278 static struct commit_graph *load_commit_graph_one(const char *graph_file)
279 {
280
281 struct stat st;
282 int fd;
283 int open_ok = open_commit_graph(graph_file, &fd, &st);
284
285 if (!open_ok)
286 return NULL;
287
288 return load_commit_graph_one_fd_st(fd, &st);
289 }
290
291 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
292 {
293 char *graph_name;
294
295 if (r->objects->commit_graph)
296 return;
297
298 graph_name = get_commit_graph_filename(obj_dir);
299 r->objects->commit_graph =
300 load_commit_graph_one(graph_name);
301
302 FREE_AND_NULL(graph_name);
303 }
304
305 /*
306 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
307 *
308 * On the first invocation, this function attemps to load the commit
309 * graph if the_repository is configured to have one.
310 */
311 static int prepare_commit_graph(struct repository *r)
312 {
313 struct object_directory *odb;
314 int config_value;
315
316 if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
317 die("dying as requested by the '%s' variable on commit-graph load!",
318 GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD);
319
320 if (r->objects->commit_graph_attempted)
321 return !!r->objects->commit_graph;
322 r->objects->commit_graph_attempted = 1;
323
324 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
325 (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
326 !config_value))
327 /*
328 * This repository is not configured to use commit graphs, so
329 * do not load one. (But report commit_graph_attempted anyway
330 * so that commit graph loading is not attempted again for this
331 * repository.)
332 */
333 return 0;
334
335 if (!commit_graph_compatible(r))
336 return 0;
337
338 prepare_alt_odb(r);
339 for (odb = r->objects->odb;
340 !r->objects->commit_graph && odb;
341 odb = odb->next)
342 prepare_commit_graph_one(r, odb->path);
343 return !!r->objects->commit_graph;
344 }
345
346 int generation_numbers_enabled(struct repository *r)
347 {
348 uint32_t first_generation;
349 struct commit_graph *g;
350 if (!prepare_commit_graph(r))
351 return 0;
352
353 g = r->objects->commit_graph;
354
355 if (!g->num_commits)
356 return 0;
357
358 first_generation = get_be32(g->chunk_commit_data +
359 g->hash_len + 8) >> 2;
360
361 return !!first_generation;
362 }
363
364 void close_commit_graph(struct raw_object_store *o)
365 {
366 free_commit_graph(o->commit_graph);
367 o->commit_graph = NULL;
368 }
369
370 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
371 {
372 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
373 g->chunk_oid_lookup, g->hash_len, pos);
374 }
375
376 static struct commit_list **insert_parent_or_die(struct repository *r,
377 struct commit_graph *g,
378 uint64_t pos,
379 struct commit_list **pptr)
380 {
381 struct commit *c;
382 struct object_id oid;
383
384 if (pos >= g->num_commits)
385 die("invalid parent position %"PRIu64, pos);
386
387 hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
388 c = lookup_commit(r, &oid);
389 if (!c)
390 die(_("could not find commit %s"), oid_to_hex(&oid));
391 c->graph_pos = pos;
392 return &commit_list_insert(c, pptr)->next;
393 }
394
395 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
396 {
397 const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
398 item->graph_pos = pos;
399 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
400 }
401
402 static inline void set_commit_tree(struct commit *c, struct tree *t)
403 {
404 c->maybe_tree = t;
405 }
406
407 static int fill_commit_in_graph(struct repository *r,
408 struct commit *item,
409 struct commit_graph *g, uint32_t pos)
410 {
411 uint32_t edge_value;
412 uint32_t *parent_data_ptr;
413 uint64_t date_low, date_high;
414 struct commit_list **pptr;
415 const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
416
417 item->object.parsed = 1;
418 item->graph_pos = pos;
419
420 set_commit_tree(item, NULL);
421
422 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
423 date_low = get_be32(commit_data + g->hash_len + 12);
424 item->date = (timestamp_t)((date_high << 32) | date_low);
425
426 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
427
428 pptr = &item->parents;
429
430 edge_value = get_be32(commit_data + g->hash_len);
431 if (edge_value == GRAPH_PARENT_NONE)
432 return 1;
433 pptr = insert_parent_or_die(r, g, edge_value, pptr);
434
435 edge_value = get_be32(commit_data + g->hash_len + 4);
436 if (edge_value == GRAPH_PARENT_NONE)
437 return 1;
438 if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
439 pptr = insert_parent_or_die(r, g, edge_value, pptr);
440 return 1;
441 }
442
443 parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
444 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
445 do {
446 edge_value = get_be32(parent_data_ptr);
447 pptr = insert_parent_or_die(r, g,
448 edge_value & GRAPH_EDGE_LAST_MASK,
449 pptr);
450 parent_data_ptr++;
451 } while (!(edge_value & GRAPH_LAST_EDGE));
452
453 return 1;
454 }
455
456 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
457 {
458 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
459 *pos = item->graph_pos;
460 return 1;
461 } else {
462 return bsearch_graph(g, &(item->object.oid), pos);
463 }
464 }
465
466 static int parse_commit_in_graph_one(struct repository *r,
467 struct commit_graph *g,
468 struct commit *item)
469 {
470 uint32_t pos;
471
472 if (item->object.parsed)
473 return 1;
474
475 if (find_commit_in_graph(item, g, &pos))
476 return fill_commit_in_graph(r, item, g, pos);
477
478 return 0;
479 }
480
481 int parse_commit_in_graph(struct repository *r, struct commit *item)
482 {
483 if (!prepare_commit_graph(r))
484 return 0;
485 return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
486 }
487
488 void load_commit_graph_info(struct repository *r, struct commit *item)
489 {
490 uint32_t pos;
491 if (!prepare_commit_graph(r))
492 return;
493 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
494 fill_commit_graph_info(item, r->objects->commit_graph, pos);
495 }
496
497 static struct tree *load_tree_for_commit(struct repository *r,
498 struct commit_graph *g,
499 struct commit *c)
500 {
501 struct object_id oid;
502 const unsigned char *commit_data = g->chunk_commit_data +
503 GRAPH_DATA_WIDTH * (c->graph_pos);
504
505 hashcpy(oid.hash, commit_data);
506 set_commit_tree(c, lookup_tree(r, &oid));
507
508 return c->maybe_tree;
509 }
510
511 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
512 struct commit_graph *g,
513 const struct commit *c)
514 {
515 if (c->maybe_tree)
516 return c->maybe_tree;
517 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
518 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
519
520 return load_tree_for_commit(r, g, (struct commit *)c);
521 }
522
523 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
524 {
525 return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
526 }
527
528 struct packed_commit_list {
529 struct commit **list;
530 int nr;
531 int alloc;
532 };
533
534 struct packed_oid_list {
535 struct object_id *list;
536 int nr;
537 int alloc;
538 };
539
540 struct write_commit_graph_context {
541 struct repository *r;
542 const char *obj_dir;
543 char *graph_name;
544 struct packed_oid_list oids;
545 struct packed_commit_list commits;
546 int num_extra_edges;
547 unsigned long approx_nr_objects;
548 struct progress *progress;
549 int progress_done;
550 uint64_t progress_cnt;
551 unsigned append:1,
552 report_progress:1;
553 };
554
555 static void write_graph_chunk_fanout(struct hashfile *f,
556 struct write_commit_graph_context *ctx)
557 {
558 int i, count = 0;
559 struct commit **list = ctx->commits.list;
560
561 /*
562 * Write the first-level table (the list is sorted,
563 * but we use a 256-entry lookup to be able to avoid
564 * having to do eight extra binary search iterations).
565 */
566 for (i = 0; i < 256; i++) {
567 while (count < ctx->commits.nr) {
568 if ((*list)->object.oid.hash[0] != i)
569 break;
570 display_progress(ctx->progress, ++ctx->progress_cnt);
571 count++;
572 list++;
573 }
574
575 hashwrite_be32(f, count);
576 }
577 }
578
579 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
580 struct write_commit_graph_context *ctx)
581 {
582 struct commit **list = ctx->commits.list;
583 int count;
584 for (count = 0; count < ctx->commits.nr; count++, list++) {
585 display_progress(ctx->progress, ++ctx->progress_cnt);
586 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
587 }
588 }
589
590 static const unsigned char *commit_to_sha1(size_t index, void *table)
591 {
592 struct commit **commits = table;
593 return commits[index]->object.oid.hash;
594 }
595
596 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
597 struct write_commit_graph_context *ctx)
598 {
599 struct commit **list = ctx->commits.list;
600 struct commit **last = ctx->commits.list + ctx->commits.nr;
601 uint32_t num_extra_edges = 0;
602
603 while (list < last) {
604 struct commit_list *parent;
605 int edge_value;
606 uint32_t packedDate[2];
607 display_progress(ctx->progress, ++ctx->progress_cnt);
608
609 parse_commit_no_graph(*list);
610 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
611
612 parent = (*list)->parents;
613
614 if (!parent)
615 edge_value = GRAPH_PARENT_NONE;
616 else {
617 edge_value = sha1_pos(parent->item->object.oid.hash,
618 ctx->commits.list,
619 ctx->commits.nr,
620 commit_to_sha1);
621
622 if (edge_value < 0)
623 BUG("missing parent %s for commit %s",
624 oid_to_hex(&parent->item->object.oid),
625 oid_to_hex(&(*list)->object.oid));
626 }
627
628 hashwrite_be32(f, edge_value);
629
630 if (parent)
631 parent = parent->next;
632
633 if (!parent)
634 edge_value = GRAPH_PARENT_NONE;
635 else if (parent->next)
636 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
637 else {
638 edge_value = sha1_pos(parent->item->object.oid.hash,
639 ctx->commits.list,
640 ctx->commits.nr,
641 commit_to_sha1);
642 if (edge_value < 0)
643 BUG("missing parent %s for commit %s",
644 oid_to_hex(&parent->item->object.oid),
645 oid_to_hex(&(*list)->object.oid));
646 }
647
648 hashwrite_be32(f, edge_value);
649
650 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
651 do {
652 num_extra_edges++;
653 parent = parent->next;
654 } while (parent);
655 }
656
657 if (sizeof((*list)->date) > 4)
658 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
659 else
660 packedDate[0] = 0;
661
662 packedDate[0] |= htonl((*list)->generation << 2);
663
664 packedDate[1] = htonl((*list)->date);
665 hashwrite(f, packedDate, 8);
666
667 list++;
668 }
669 }
670
671 static void write_graph_chunk_extra_edges(struct hashfile *f,
672 struct write_commit_graph_context *ctx)
673 {
674 struct commit **list = ctx->commits.list;
675 struct commit **last = ctx->commits.list + ctx->commits.nr;
676 struct commit_list *parent;
677
678 while (list < last) {
679 int num_parents = 0;
680
681 display_progress(ctx->progress, ++ctx->progress_cnt);
682
683 for (parent = (*list)->parents; num_parents < 3 && parent;
684 parent = parent->next)
685 num_parents++;
686
687 if (num_parents <= 2) {
688 list++;
689 continue;
690 }
691
692 /* Since num_parents > 2, this initializer is safe. */
693 for (parent = (*list)->parents->next; parent; parent = parent->next) {
694 int edge_value = sha1_pos(parent->item->object.oid.hash,
695 ctx->commits.list,
696 ctx->commits.nr,
697 commit_to_sha1);
698
699 if (edge_value < 0)
700 BUG("missing parent %s for commit %s",
701 oid_to_hex(&parent->item->object.oid),
702 oid_to_hex(&(*list)->object.oid));
703 else if (!parent->next)
704 edge_value |= GRAPH_LAST_EDGE;
705
706 hashwrite_be32(f, edge_value);
707 }
708
709 list++;
710 }
711 }
712
713 static int commit_compare(const void *_a, const void *_b)
714 {
715 const struct object_id *a = (const struct object_id *)_a;
716 const struct object_id *b = (const struct object_id *)_b;
717 return oidcmp(a, b);
718 }
719
720 static int add_packed_commits(const struct object_id *oid,
721 struct packed_git *pack,
722 uint32_t pos,
723 void *data)
724 {
725 struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
726 enum object_type type;
727 off_t offset = nth_packed_object_offset(pack, pos);
728 struct object_info oi = OBJECT_INFO_INIT;
729
730 if (ctx->progress)
731 display_progress(ctx->progress, ++ctx->progress_done);
732
733 oi.typep = &type;
734 if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
735 die(_("unable to get type of object %s"), oid_to_hex(oid));
736
737 if (type != OBJ_COMMIT)
738 return 0;
739
740 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
741 oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
742 ctx->oids.nr++;
743
744 return 0;
745 }
746
747 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
748 {
749 struct commit_list *parent;
750 for (parent = commit->parents; parent; parent = parent->next) {
751 if (!(parent->item->object.flags & UNINTERESTING)) {
752 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
753 oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
754 ctx->oids.nr++;
755 parent->item->object.flags |= UNINTERESTING;
756 }
757 }
758 }
759
760 static void close_reachable(struct write_commit_graph_context *ctx)
761 {
762 int i;
763 struct commit *commit;
764
765 if (ctx->report_progress)
766 ctx->progress = start_delayed_progress(
767 _("Loading known commits in commit graph"),
768 ctx->oids.nr);
769 for (i = 0; i < ctx->oids.nr; i++) {
770 display_progress(ctx->progress, i + 1);
771 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
772 if (commit)
773 commit->object.flags |= UNINTERESTING;
774 }
775 stop_progress(&ctx->progress);
776
777 /*
778 * As this loop runs, ctx->oids.nr may grow, but not more
779 * than the number of missing commits in the reachable
780 * closure.
781 */
782 if (ctx->report_progress)
783 ctx->progress = start_delayed_progress(
784 _("Expanding reachable commits in commit graph"),
785 ctx->oids.nr);
786 for (i = 0; i < ctx->oids.nr; i++) {
787 display_progress(ctx->progress, i + 1);
788 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
789
790 if (commit && !parse_commit_no_graph(commit))
791 add_missing_parents(ctx, commit);
792 }
793 stop_progress(&ctx->progress);
794
795 if (ctx->report_progress)
796 ctx->progress = start_delayed_progress(
797 _("Clearing commit marks in commit graph"),
798 ctx->oids.nr);
799 for (i = 0; i < ctx->oids.nr; i++) {
800 display_progress(ctx->progress, i + 1);
801 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
802
803 if (commit)
804 commit->object.flags &= ~UNINTERESTING;
805 }
806 stop_progress(&ctx->progress);
807 }
808
809 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
810 {
811 int i;
812 struct commit_list *list = NULL;
813
814 if (ctx->report_progress)
815 ctx->progress = start_progress(
816 _("Computing commit graph generation numbers"),
817 ctx->commits.nr);
818 for (i = 0; i < ctx->commits.nr; i++) {
819 display_progress(ctx->progress, i + 1);
820 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
821 ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
822 continue;
823
824 commit_list_insert(ctx->commits.list[i], &list);
825 while (list) {
826 struct commit *current = list->item;
827 struct commit_list *parent;
828 int all_parents_computed = 1;
829 uint32_t max_generation = 0;
830
831 for (parent = current->parents; parent; parent = parent->next) {
832 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
833 parent->item->generation == GENERATION_NUMBER_ZERO) {
834 all_parents_computed = 0;
835 commit_list_insert(parent->item, &list);
836 break;
837 } else if (parent->item->generation > max_generation) {
838 max_generation = parent->item->generation;
839 }
840 }
841
842 if (all_parents_computed) {
843 current->generation = max_generation + 1;
844 pop_commit(&list);
845
846 if (current->generation > GENERATION_NUMBER_MAX)
847 current->generation = GENERATION_NUMBER_MAX;
848 }
849 }
850 }
851 stop_progress(&ctx->progress);
852 }
853
854 static int add_ref_to_list(const char *refname,
855 const struct object_id *oid,
856 int flags, void *cb_data)
857 {
858 struct string_list *list = (struct string_list *)cb_data;
859
860 string_list_append(list, oid_to_hex(oid));
861 return 0;
862 }
863
864 int write_commit_graph_reachable(const char *obj_dir, unsigned int flags)
865 {
866 struct string_list list = STRING_LIST_INIT_DUP;
867 int result;
868
869 for_each_ref(add_ref_to_list, &list);
870 result = write_commit_graph(obj_dir, NULL, &list,
871 flags);
872
873 string_list_clear(&list, 0);
874 return result;
875 }
876
877 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
878 struct string_list *pack_indexes)
879 {
880 uint32_t i;
881 struct strbuf progress_title = STRBUF_INIT;
882 struct strbuf packname = STRBUF_INIT;
883 int dirlen;
884
885 strbuf_addf(&packname, "%s/pack/", ctx->obj_dir);
886 dirlen = packname.len;
887 if (ctx->report_progress) {
888 strbuf_addf(&progress_title,
889 Q_("Finding commits for commit graph in %d pack",
890 "Finding commits for commit graph in %d packs",
891 pack_indexes->nr),
892 pack_indexes->nr);
893 ctx->progress = start_delayed_progress(progress_title.buf, 0);
894 ctx->progress_done = 0;
895 }
896 for (i = 0; i < pack_indexes->nr; i++) {
897 struct packed_git *p;
898 strbuf_setlen(&packname, dirlen);
899 strbuf_addstr(&packname, pack_indexes->items[i].string);
900 p = add_packed_git(packname.buf, packname.len, 1);
901 if (!p) {
902 error(_("error adding pack %s"), packname.buf);
903 return -1;
904 }
905 if (open_pack_index(p)) {
906 error(_("error opening index for %s"), packname.buf);
907 return -1;
908 }
909 for_each_object_in_pack(p, add_packed_commits, ctx,
910 FOR_EACH_OBJECT_PACK_ORDER);
911 close_pack(p);
912 free(p);
913 }
914
915 stop_progress(&ctx->progress);
916 strbuf_reset(&progress_title);
917 strbuf_release(&packname);
918
919 return 0;
920 }
921
922 static void fill_oids_from_commit_hex(struct write_commit_graph_context *ctx,
923 struct string_list *commit_hex)
924 {
925 uint32_t i;
926 struct strbuf progress_title = STRBUF_INIT;
927
928 if (ctx->report_progress) {
929 strbuf_addf(&progress_title,
930 Q_("Finding commits for commit graph from %d ref",
931 "Finding commits for commit graph from %d refs",
932 commit_hex->nr),
933 commit_hex->nr);
934 ctx->progress = start_delayed_progress(
935 progress_title.buf,
936 commit_hex->nr);
937 }
938 for (i = 0; i < commit_hex->nr; i++) {
939 const char *end;
940 struct object_id oid;
941 struct commit *result;
942
943 display_progress(ctx->progress, i + 1);
944 if (commit_hex->items[i].string &&
945 parse_oid_hex(commit_hex->items[i].string, &oid, &end))
946 continue;
947
948 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
949
950 if (result) {
951 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
952 oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
953 ctx->oids.nr++;
954 }
955 }
956 stop_progress(&ctx->progress);
957 strbuf_release(&progress_title);
958 }
959
960 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
961 {
962 if (ctx->report_progress)
963 ctx->progress = start_delayed_progress(
964 _("Finding commits for commit graph among packed objects"),
965 ctx->approx_nr_objects);
966 for_each_packed_object(add_packed_commits, ctx,
967 FOR_EACH_OBJECT_PACK_ORDER);
968 if (ctx->progress_done < ctx->approx_nr_objects)
969 display_progress(ctx->progress, ctx->approx_nr_objects);
970 stop_progress(&ctx->progress);
971 }
972
973 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
974 {
975 uint32_t i, count_distinct = 1;
976
977 if (ctx->report_progress)
978 ctx->progress = start_delayed_progress(
979 _("Counting distinct commits in commit graph"),
980 ctx->oids.nr);
981 display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
982 QSORT(ctx->oids.list, ctx->oids.nr, commit_compare);
983
984 for (i = 1; i < ctx->oids.nr; i++) {
985 display_progress(ctx->progress, i + 1);
986 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
987 count_distinct++;
988 }
989 stop_progress(&ctx->progress);
990
991 return count_distinct;
992 }
993
994 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
995 {
996 uint32_t i;
997 struct commit_list *parent;
998
999 ctx->num_extra_edges = 0;
1000 if (ctx->report_progress)
1001 ctx->progress = start_delayed_progress(
1002 _("Finding extra edges in commit graph"),
1003 ctx->oids.nr);
1004 for (i = 0; i < ctx->oids.nr; i++) {
1005 int num_parents = 0;
1006 display_progress(ctx->progress, i + 1);
1007 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1008 continue;
1009
1010 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1011 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1012
1013 for (parent = ctx->commits.list[ctx->commits.nr]->parents;
1014 parent; parent = parent->next)
1015 num_parents++;
1016
1017 if (num_parents > 2)
1018 ctx->num_extra_edges += num_parents - 1;
1019
1020 ctx->commits.nr++;
1021 }
1022 stop_progress(&ctx->progress);
1023 }
1024
1025 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1026 {
1027 uint32_t i;
1028 struct hashfile *f;
1029 struct lock_file lk = LOCK_INIT;
1030 uint32_t chunk_ids[5];
1031 uint64_t chunk_offsets[5];
1032 const unsigned hashsz = the_hash_algo->rawsz;
1033 struct strbuf progress_title = STRBUF_INIT;
1034 int num_chunks = ctx->num_extra_edges ? 4 : 3;
1035
1036 ctx->graph_name = get_commit_graph_filename(ctx->obj_dir);
1037 if (safe_create_leading_directories(ctx->graph_name)) {
1038 UNLEAK(ctx->graph_name);
1039 error(_("unable to create leading directories of %s"),
1040 ctx->graph_name);
1041 return -1;
1042 }
1043
1044 hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
1045 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1046
1047 hashwrite_be32(f, GRAPH_SIGNATURE);
1048
1049 hashwrite_u8(f, GRAPH_VERSION);
1050 hashwrite_u8(f, oid_version());
1051 hashwrite_u8(f, num_chunks);
1052 hashwrite_u8(f, 0); /* unused padding byte */
1053
1054 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1055 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1056 chunk_ids[2] = GRAPH_CHUNKID_DATA;
1057 if (ctx->num_extra_edges)
1058 chunk_ids[3] = GRAPH_CHUNKID_EXTRAEDGES;
1059 else
1060 chunk_ids[3] = 0;
1061 chunk_ids[4] = 0;
1062
1063 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1064 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1065 chunk_offsets[2] = chunk_offsets[1] + hashsz * ctx->commits.nr;
1066 chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * ctx->commits.nr;
1067 chunk_offsets[4] = chunk_offsets[3] + 4 * ctx->num_extra_edges;
1068
1069 for (i = 0; i <= num_chunks; i++) {
1070 uint32_t chunk_write[3];
1071
1072 chunk_write[0] = htonl(chunk_ids[i]);
1073 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1074 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1075 hashwrite(f, chunk_write, 12);
1076 }
1077
1078 if (ctx->report_progress) {
1079 strbuf_addf(&progress_title,
1080 Q_("Writing out commit graph in %d pass",
1081 "Writing out commit graph in %d passes",
1082 num_chunks),
1083 num_chunks);
1084 ctx->progress = start_delayed_progress(
1085 progress_title.buf,
1086 num_chunks * ctx->commits.nr);
1087 }
1088 write_graph_chunk_fanout(f, ctx);
1089 write_graph_chunk_oids(f, hashsz, ctx);
1090 write_graph_chunk_data(f, hashsz, ctx);
1091 if (ctx->num_extra_edges)
1092 write_graph_chunk_extra_edges(f, ctx);
1093 stop_progress(&ctx->progress);
1094 strbuf_release(&progress_title);
1095
1096 close_commit_graph(ctx->r->objects);
1097 finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1098 commit_lock_file(&lk);
1099
1100 return 0;
1101 }
1102
1103 int write_commit_graph(const char *obj_dir,
1104 struct string_list *pack_indexes,
1105 struct string_list *commit_hex,
1106 unsigned int flags)
1107 {
1108 struct write_commit_graph_context *ctx;
1109 uint32_t i, count_distinct = 0;
1110 int res = 0;
1111
1112 if (!commit_graph_compatible(the_repository))
1113 return 0;
1114
1115 ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
1116 ctx->r = the_repository;
1117 ctx->obj_dir = obj_dir;
1118 ctx->append = flags & COMMIT_GRAPH_APPEND ? 1 : 0;
1119 ctx->report_progress = flags & COMMIT_GRAPH_PROGRESS ? 1 : 0;
1120
1121 ctx->approx_nr_objects = approximate_object_count();
1122 ctx->oids.alloc = ctx->approx_nr_objects / 32;
1123
1124 if (ctx->append) {
1125 prepare_commit_graph_one(ctx->r, ctx->obj_dir);
1126 if (ctx->r->objects->commit_graph)
1127 ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
1128 }
1129
1130 if (ctx->oids.alloc < 1024)
1131 ctx->oids.alloc = 1024;
1132 ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
1133
1134 if (ctx->append && ctx->r->objects->commit_graph) {
1135 struct commit_graph *g = ctx->r->objects->commit_graph;
1136 for (i = 0; i < g->num_commits; i++) {
1137 const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
1138 hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
1139 }
1140 }
1141
1142 if (pack_indexes) {
1143 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
1144 goto cleanup;
1145 }
1146
1147 if (commit_hex)
1148 fill_oids_from_commit_hex(ctx, commit_hex);
1149
1150 if (!pack_indexes && !commit_hex)
1151 fill_oids_from_all_packs(ctx);
1152
1153 close_reachable(ctx);
1154
1155 count_distinct = count_distinct_commits(ctx);
1156
1157 if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
1158 error(_("the commit graph format cannot write %d commits"), count_distinct);
1159 res = -1;
1160 goto cleanup;
1161 }
1162
1163 ctx->commits.alloc = count_distinct;
1164 ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
1165
1166 copy_oids_to_commits(ctx);
1167
1168 if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
1169 error(_("too many commits to write graph"));
1170 res = -1;
1171 goto cleanup;
1172 }
1173
1174 compute_generation_numbers(ctx);
1175
1176 res = write_commit_graph_file(ctx);
1177
1178 cleanup:
1179 free(ctx->graph_name);
1180 free(ctx->commits.list);
1181 free(ctx->oids.list);
1182 free(ctx);
1183
1184 return res;
1185 }
1186
1187 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1188 static int verify_commit_graph_error;
1189
1190 static void graph_report(const char *fmt, ...)
1191 {
1192 va_list ap;
1193
1194 verify_commit_graph_error = 1;
1195 va_start(ap, fmt);
1196 vfprintf(stderr, fmt, ap);
1197 fprintf(stderr, "\n");
1198 va_end(ap);
1199 }
1200
1201 #define GENERATION_ZERO_EXISTS 1
1202 #define GENERATION_NUMBER_EXISTS 2
1203
1204 int verify_commit_graph(struct repository *r, struct commit_graph *g)
1205 {
1206 uint32_t i, cur_fanout_pos = 0;
1207 struct object_id prev_oid, cur_oid, checksum;
1208 int generation_zero = 0;
1209 struct hashfile *f;
1210 int devnull;
1211 struct progress *progress = NULL;
1212
1213 if (!g) {
1214 graph_report("no commit-graph file loaded");
1215 return 1;
1216 }
1217
1218 verify_commit_graph_error = verify_commit_graph_lite(g);
1219 if (verify_commit_graph_error)
1220 return verify_commit_graph_error;
1221
1222 devnull = open("/dev/null", O_WRONLY);
1223 f = hashfd(devnull, NULL);
1224 hashwrite(f, g->data, g->data_len - g->hash_len);
1225 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1226 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1227 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1228 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1229 }
1230
1231 for (i = 0; i < g->num_commits; i++) {
1232 struct commit *graph_commit;
1233
1234 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1235
1236 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1237 graph_report(_("commit-graph has incorrect OID order: %s then %s"),
1238 oid_to_hex(&prev_oid),
1239 oid_to_hex(&cur_oid));
1240
1241 oidcpy(&prev_oid, &cur_oid);
1242
1243 while (cur_oid.hash[0] > cur_fanout_pos) {
1244 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1245
1246 if (i != fanout_value)
1247 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1248 cur_fanout_pos, fanout_value, i);
1249 cur_fanout_pos++;
1250 }
1251
1252 graph_commit = lookup_commit(r, &cur_oid);
1253 if (!parse_commit_in_graph_one(r, g, graph_commit))
1254 graph_report(_("failed to parse commit %s from commit-graph"),
1255 oid_to_hex(&cur_oid));
1256 }
1257
1258 while (cur_fanout_pos < 256) {
1259 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1260
1261 if (g->num_commits != fanout_value)
1262 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1263 cur_fanout_pos, fanout_value, i);
1264
1265 cur_fanout_pos++;
1266 }
1267
1268 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1269 return verify_commit_graph_error;
1270
1271 progress = start_progress(_("Verifying commits in commit graph"),
1272 g->num_commits);
1273 for (i = 0; i < g->num_commits; i++) {
1274 struct commit *graph_commit, *odb_commit;
1275 struct commit_list *graph_parents, *odb_parents;
1276 uint32_t max_generation = 0;
1277
1278 display_progress(progress, i + 1);
1279 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1280
1281 graph_commit = lookup_commit(r, &cur_oid);
1282 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1283 if (parse_commit_internal(odb_commit, 0, 0)) {
1284 graph_report(_("failed to parse commit %s from object database for commit-graph"),
1285 oid_to_hex(&cur_oid));
1286 continue;
1287 }
1288
1289 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
1290 get_commit_tree_oid(odb_commit)))
1291 graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
1292 oid_to_hex(&cur_oid),
1293 oid_to_hex(get_commit_tree_oid(graph_commit)),
1294 oid_to_hex(get_commit_tree_oid(odb_commit)));
1295
1296 graph_parents = graph_commit->parents;
1297 odb_parents = odb_commit->parents;
1298
1299 while (graph_parents) {
1300 if (odb_parents == NULL) {
1301 graph_report(_("commit-graph parent list for commit %s is too long"),
1302 oid_to_hex(&cur_oid));
1303 break;
1304 }
1305
1306 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1307 graph_report(_("commit-graph parent for %s is %s != %s"),
1308 oid_to_hex(&cur_oid),
1309 oid_to_hex(&graph_parents->item->object.oid),
1310 oid_to_hex(&odb_parents->item->object.oid));
1311
1312 if (graph_parents->item->generation > max_generation)
1313 max_generation = graph_parents->item->generation;
1314
1315 graph_parents = graph_parents->next;
1316 odb_parents = odb_parents->next;
1317 }
1318
1319 if (odb_parents != NULL)
1320 graph_report(_("commit-graph parent list for commit %s terminates early"),
1321 oid_to_hex(&cur_oid));
1322
1323 if (!graph_commit->generation) {
1324 if (generation_zero == GENERATION_NUMBER_EXISTS)
1325 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
1326 oid_to_hex(&cur_oid));
1327 generation_zero = GENERATION_ZERO_EXISTS;
1328 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1329 graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
1330 oid_to_hex(&cur_oid));
1331
1332 if (generation_zero == GENERATION_ZERO_EXISTS)
1333 continue;
1334
1335 /*
1336 * If one of our parents has generation GENERATION_NUMBER_MAX, then
1337 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1338 * extra logic in the following condition.
1339 */
1340 if (max_generation == GENERATION_NUMBER_MAX)
1341 max_generation--;
1342
1343 if (graph_commit->generation != max_generation + 1)
1344 graph_report(_("commit-graph generation for commit %s is %u != %u"),
1345 oid_to_hex(&cur_oid),
1346 graph_commit->generation,
1347 max_generation + 1);
1348
1349 if (graph_commit->date != odb_commit->date)
1350 graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
1351 oid_to_hex(&cur_oid),
1352 graph_commit->date,
1353 odb_commit->date);
1354 }
1355 stop_progress(&progress);
1356
1357 return verify_commit_graph_error;
1358 }
1359
1360 void free_commit_graph(struct commit_graph *g)
1361 {
1362 if (!g)
1363 return;
1364 if (g->graph_fd >= 0) {
1365 munmap((void *)g->data, g->data_len);
1366 g->data = NULL;
1367 close(g->graph_fd);
1368 }
1369 free(g);
1370 }