commit-graph write: add more descriptive progress output
[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 36
27
28 #define GRAPH_VERSION_1 0x1
29 #define GRAPH_VERSION GRAPH_VERSION_1
30
31 #define GRAPH_OID_VERSION_SHA1 1
32 #define GRAPH_OID_LEN_SHA1 GIT_SHA1_RAWSZ
33 #define GRAPH_OID_VERSION GRAPH_OID_VERSION_SHA1
34 #define GRAPH_OID_LEN GRAPH_OID_LEN_SHA1
35
36 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
37 #define GRAPH_PARENT_MISSING 0x7fffffff
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 + GRAPH_OID_LEN)
48
49 char *get_commit_graph_filename(const char *obj_dir)
50 {
51 return xstrfmt("%s/info/commit-graph", obj_dir);
52 }
53
54 static struct commit_graph *alloc_commit_graph(void)
55 {
56 struct commit_graph *g = xcalloc(1, sizeof(*g));
57 g->graph_fd = -1;
58
59 return g;
60 }
61
62 extern int read_replace_refs;
63
64 static int commit_graph_compatible(struct repository *r)
65 {
66 if (!r->gitdir)
67 return 0;
68
69 if (read_replace_refs) {
70 prepare_replace_object(r);
71 if (hashmap_get_size(&r->objects->replace_map->map))
72 return 0;
73 }
74
75 prepare_commit_graft(r);
76 if (r->parsed_objects && r->parsed_objects->grafts_nr)
77 return 0;
78 if (is_repository_shallow(r))
79 return 0;
80
81 return 1;
82 }
83
84 struct commit_graph *load_commit_graph_one(const char *graph_file)
85 {
86 void *graph_map;
87 const unsigned char *data, *chunk_lookup;
88 size_t graph_size;
89 struct stat st;
90 uint32_t i;
91 struct commit_graph *graph;
92 int fd = git_open(graph_file);
93 uint64_t last_chunk_offset;
94 uint32_t last_chunk_id;
95 uint32_t graph_signature;
96 unsigned char graph_version, hash_version;
97
98 if (fd < 0)
99 return NULL;
100 if (fstat(fd, &st)) {
101 close(fd);
102 return NULL;
103 }
104 graph_size = xsize_t(st.st_size);
105
106 if (graph_size < GRAPH_MIN_SIZE) {
107 close(fd);
108 die(_("graph file %s is too small"), graph_file);
109 }
110 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
111 data = (const unsigned char *)graph_map;
112
113 graph_signature = get_be32(data);
114 if (graph_signature != GRAPH_SIGNATURE) {
115 error(_("graph signature %X does not match signature %X"),
116 graph_signature, GRAPH_SIGNATURE);
117 goto cleanup_fail;
118 }
119
120 graph_version = *(unsigned char*)(data + 4);
121 if (graph_version != GRAPH_VERSION) {
122 error(_("graph version %X does not match version %X"),
123 graph_version, GRAPH_VERSION);
124 goto cleanup_fail;
125 }
126
127 hash_version = *(unsigned char*)(data + 5);
128 if (hash_version != GRAPH_OID_VERSION) {
129 error(_("hash version %X does not match version %X"),
130 hash_version, GRAPH_OID_VERSION);
131 goto cleanup_fail;
132 }
133
134 graph = alloc_commit_graph();
135
136 graph->hash_len = GRAPH_OID_LEN;
137 graph->num_chunks = *(unsigned char*)(data + 6);
138 graph->graph_fd = fd;
139 graph->data = graph_map;
140 graph->data_len = graph_size;
141
142 last_chunk_id = 0;
143 last_chunk_offset = 8;
144 chunk_lookup = data + 8;
145 for (i = 0; i < graph->num_chunks; i++) {
146 uint32_t chunk_id = get_be32(chunk_lookup + 0);
147 uint64_t chunk_offset = get_be64(chunk_lookup + 4);
148 int chunk_repeated = 0;
149
150 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
151
152 if (chunk_offset > graph_size - GIT_MAX_RAWSZ) {
153 error(_("improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
154 (uint32_t)chunk_offset);
155 goto cleanup_fail;
156 }
157
158 switch (chunk_id) {
159 case GRAPH_CHUNKID_OIDFANOUT:
160 if (graph->chunk_oid_fanout)
161 chunk_repeated = 1;
162 else
163 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
164 break;
165
166 case GRAPH_CHUNKID_OIDLOOKUP:
167 if (graph->chunk_oid_lookup)
168 chunk_repeated = 1;
169 else
170 graph->chunk_oid_lookup = data + chunk_offset;
171 break;
172
173 case GRAPH_CHUNKID_DATA:
174 if (graph->chunk_commit_data)
175 chunk_repeated = 1;
176 else
177 graph->chunk_commit_data = data + chunk_offset;
178 break;
179
180 case GRAPH_CHUNKID_EXTRAEDGES:
181 if (graph->chunk_extra_edges)
182 chunk_repeated = 1;
183 else
184 graph->chunk_extra_edges = data + chunk_offset;
185 break;
186 }
187
188 if (chunk_repeated) {
189 error(_("chunk id %08x appears multiple times"), chunk_id);
190 goto cleanup_fail;
191 }
192
193 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
194 {
195 graph->num_commits = (chunk_offset - last_chunk_offset)
196 / graph->hash_len;
197 }
198
199 last_chunk_id = chunk_id;
200 last_chunk_offset = chunk_offset;
201 }
202
203 return graph;
204
205 cleanup_fail:
206 munmap(graph_map, graph_size);
207 close(fd);
208 exit(1);
209 }
210
211 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
212 {
213 char *graph_name;
214
215 if (r->objects->commit_graph)
216 return;
217
218 graph_name = get_commit_graph_filename(obj_dir);
219 r->objects->commit_graph =
220 load_commit_graph_one(graph_name);
221
222 FREE_AND_NULL(graph_name);
223 }
224
225 /*
226 * Return 1 if commit_graph is non-NULL, and 0 otherwise.
227 *
228 * On the first invocation, this function attemps to load the commit
229 * graph if the_repository is configured to have one.
230 */
231 static int prepare_commit_graph(struct repository *r)
232 {
233 struct object_directory *odb;
234 int config_value;
235
236 if (r->objects->commit_graph_attempted)
237 return !!r->objects->commit_graph;
238 r->objects->commit_graph_attempted = 1;
239
240 if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
241 (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
242 !config_value))
243 /*
244 * This repository is not configured to use commit graphs, so
245 * do not load one. (But report commit_graph_attempted anyway
246 * so that commit graph loading is not attempted again for this
247 * repository.)
248 */
249 return 0;
250
251 if (!commit_graph_compatible(r))
252 return 0;
253
254 prepare_alt_odb(r);
255 for (odb = r->objects->odb;
256 !r->objects->commit_graph && odb;
257 odb = odb->next)
258 prepare_commit_graph_one(r, odb->path);
259 return !!r->objects->commit_graph;
260 }
261
262 int generation_numbers_enabled(struct repository *r)
263 {
264 uint32_t first_generation;
265 struct commit_graph *g;
266 if (!prepare_commit_graph(r))
267 return 0;
268
269 g = r->objects->commit_graph;
270
271 if (!g->num_commits)
272 return 0;
273
274 first_generation = get_be32(g->chunk_commit_data +
275 g->hash_len + 8) >> 2;
276
277 return !!first_generation;
278 }
279
280 void close_commit_graph(struct repository *r)
281 {
282 free_commit_graph(r->objects->commit_graph);
283 r->objects->commit_graph = NULL;
284 }
285
286 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
287 {
288 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
289 g->chunk_oid_lookup, g->hash_len, pos);
290 }
291
292 static struct commit_list **insert_parent_or_die(struct commit_graph *g,
293 uint64_t pos,
294 struct commit_list **pptr)
295 {
296 struct commit *c;
297 struct object_id oid;
298
299 if (pos >= g->num_commits)
300 die("invalid parent position %"PRIu64, pos);
301
302 hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
303 c = lookup_commit(the_repository, &oid);
304 if (!c)
305 die(_("could not find commit %s"), oid_to_hex(&oid));
306 c->graph_pos = pos;
307 return &commit_list_insert(c, pptr)->next;
308 }
309
310 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
311 {
312 const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
313 item->graph_pos = pos;
314 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
315 }
316
317 static int fill_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t pos)
318 {
319 uint32_t edge_value;
320 uint32_t *parent_data_ptr;
321 uint64_t date_low, date_high;
322 struct commit_list **pptr;
323 const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
324
325 item->object.parsed = 1;
326 item->graph_pos = pos;
327
328 item->maybe_tree = NULL;
329
330 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
331 date_low = get_be32(commit_data + g->hash_len + 12);
332 item->date = (timestamp_t)((date_high << 32) | date_low);
333
334 item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
335
336 pptr = &item->parents;
337
338 edge_value = get_be32(commit_data + g->hash_len);
339 if (edge_value == GRAPH_PARENT_NONE)
340 return 1;
341 pptr = insert_parent_or_die(g, edge_value, pptr);
342
343 edge_value = get_be32(commit_data + g->hash_len + 4);
344 if (edge_value == GRAPH_PARENT_NONE)
345 return 1;
346 if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
347 pptr = insert_parent_or_die(g, edge_value, pptr);
348 return 1;
349 }
350
351 parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
352 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
353 do {
354 edge_value = get_be32(parent_data_ptr);
355 pptr = insert_parent_or_die(g,
356 edge_value & GRAPH_EDGE_LAST_MASK,
357 pptr);
358 parent_data_ptr++;
359 } while (!(edge_value & GRAPH_LAST_EDGE));
360
361 return 1;
362 }
363
364 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
365 {
366 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
367 *pos = item->graph_pos;
368 return 1;
369 } else {
370 return bsearch_graph(g, &(item->object.oid), pos);
371 }
372 }
373
374 static int parse_commit_in_graph_one(struct commit_graph *g, struct commit *item)
375 {
376 uint32_t pos;
377
378 if (item->object.parsed)
379 return 1;
380
381 if (find_commit_in_graph(item, g, &pos))
382 return fill_commit_in_graph(item, g, pos);
383
384 return 0;
385 }
386
387 int parse_commit_in_graph(struct repository *r, struct commit *item)
388 {
389 if (!prepare_commit_graph(r))
390 return 0;
391 return parse_commit_in_graph_one(r->objects->commit_graph, item);
392 }
393
394 void load_commit_graph_info(struct repository *r, struct commit *item)
395 {
396 uint32_t pos;
397 if (!prepare_commit_graph(r))
398 return;
399 if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
400 fill_commit_graph_info(item, r->objects->commit_graph, pos);
401 }
402
403 static struct tree *load_tree_for_commit(struct commit_graph *g, struct commit *c)
404 {
405 struct object_id oid;
406 const unsigned char *commit_data = g->chunk_commit_data +
407 GRAPH_DATA_WIDTH * (c->graph_pos);
408
409 hashcpy(oid.hash, commit_data);
410 c->maybe_tree = lookup_tree(the_repository, &oid);
411
412 return c->maybe_tree;
413 }
414
415 static struct tree *get_commit_tree_in_graph_one(struct commit_graph *g,
416 const struct commit *c)
417 {
418 if (c->maybe_tree)
419 return c->maybe_tree;
420 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
421 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
422
423 return load_tree_for_commit(g, (struct commit *)c);
424 }
425
426 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
427 {
428 return get_commit_tree_in_graph_one(r->objects->commit_graph, c);
429 }
430
431 static void write_graph_chunk_fanout(struct hashfile *f,
432 struct commit **commits,
433 int nr_commits,
434 struct progress *progress,
435 uint64_t *progress_cnt)
436 {
437 int i, count = 0;
438 struct commit **list = commits;
439
440 /*
441 * Write the first-level table (the list is sorted,
442 * but we use a 256-entry lookup to be able to avoid
443 * having to do eight extra binary search iterations).
444 */
445 for (i = 0; i < 256; i++) {
446 while (count < nr_commits) {
447 if ((*list)->object.oid.hash[0] != i)
448 break;
449 display_progress(progress, ++*progress_cnt);
450 count++;
451 list++;
452 }
453
454 hashwrite_be32(f, count);
455 }
456 }
457
458 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
459 struct commit **commits, int nr_commits,
460 struct progress *progress,
461 uint64_t *progress_cnt)
462 {
463 struct commit **list = commits;
464 int count;
465 for (count = 0; count < nr_commits; count++, list++) {
466 display_progress(progress, ++*progress_cnt);
467 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
468 }
469 }
470
471 static const unsigned char *commit_to_sha1(size_t index, void *table)
472 {
473 struct commit **commits = table;
474 return commits[index]->object.oid.hash;
475 }
476
477 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
478 struct commit **commits, int nr_commits,
479 struct progress *progress,
480 uint64_t *progress_cnt)
481 {
482 struct commit **list = commits;
483 struct commit **last = commits + nr_commits;
484 uint32_t num_extra_edges = 0;
485
486 while (list < last) {
487 struct commit_list *parent;
488 int edge_value;
489 uint32_t packedDate[2];
490 display_progress(progress, ++*progress_cnt);
491
492 parse_commit(*list);
493 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
494
495 parent = (*list)->parents;
496
497 if (!parent)
498 edge_value = GRAPH_PARENT_NONE;
499 else {
500 edge_value = sha1_pos(parent->item->object.oid.hash,
501 commits,
502 nr_commits,
503 commit_to_sha1);
504
505 if (edge_value < 0)
506 edge_value = GRAPH_PARENT_MISSING;
507 }
508
509 hashwrite_be32(f, edge_value);
510
511 if (parent)
512 parent = parent->next;
513
514 if (!parent)
515 edge_value = GRAPH_PARENT_NONE;
516 else if (parent->next)
517 edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
518 else {
519 edge_value = sha1_pos(parent->item->object.oid.hash,
520 commits,
521 nr_commits,
522 commit_to_sha1);
523 if (edge_value < 0)
524 edge_value = GRAPH_PARENT_MISSING;
525 }
526
527 hashwrite_be32(f, edge_value);
528
529 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
530 do {
531 num_extra_edges++;
532 parent = parent->next;
533 } while (parent);
534 }
535
536 if (sizeof((*list)->date) > 4)
537 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
538 else
539 packedDate[0] = 0;
540
541 packedDate[0] |= htonl((*list)->generation << 2);
542
543 packedDate[1] = htonl((*list)->date);
544 hashwrite(f, packedDate, 8);
545
546 list++;
547 }
548 }
549
550 static void write_graph_chunk_extra_edges(struct hashfile *f,
551 struct commit **commits,
552 int nr_commits,
553 struct progress *progress,
554 uint64_t *progress_cnt)
555 {
556 struct commit **list = commits;
557 struct commit **last = commits + nr_commits;
558 struct commit_list *parent;
559
560 while (list < last) {
561 int num_parents = 0;
562
563 display_progress(progress, ++*progress_cnt);
564
565 for (parent = (*list)->parents; num_parents < 3 && parent;
566 parent = parent->next)
567 num_parents++;
568
569 if (num_parents <= 2) {
570 list++;
571 continue;
572 }
573
574 /* Since num_parents > 2, this initializer is safe. */
575 for (parent = (*list)->parents->next; parent; parent = parent->next) {
576 int edge_value = sha1_pos(parent->item->object.oid.hash,
577 commits,
578 nr_commits,
579 commit_to_sha1);
580
581 if (edge_value < 0)
582 edge_value = GRAPH_PARENT_MISSING;
583 else if (!parent->next)
584 edge_value |= GRAPH_LAST_EDGE;
585
586 hashwrite_be32(f, edge_value);
587 }
588
589 list++;
590 }
591 }
592
593 static int commit_compare(const void *_a, const void *_b)
594 {
595 const struct object_id *a = (const struct object_id *)_a;
596 const struct object_id *b = (const struct object_id *)_b;
597 return oidcmp(a, b);
598 }
599
600 struct packed_commit_list {
601 struct commit **list;
602 int nr;
603 int alloc;
604 };
605
606 struct packed_oid_list {
607 struct object_id *list;
608 int nr;
609 int alloc;
610 struct progress *progress;
611 int progress_done;
612 };
613
614 static int add_packed_commits(const struct object_id *oid,
615 struct packed_git *pack,
616 uint32_t pos,
617 void *data)
618 {
619 struct packed_oid_list *list = (struct packed_oid_list*)data;
620 enum object_type type;
621 off_t offset = nth_packed_object_offset(pack, pos);
622 struct object_info oi = OBJECT_INFO_INIT;
623
624 if (list->progress)
625 display_progress(list->progress, ++list->progress_done);
626
627 oi.typep = &type;
628 if (packed_object_info(the_repository, pack, offset, &oi) < 0)
629 die(_("unable to get type of object %s"), oid_to_hex(oid));
630
631 if (type != OBJ_COMMIT)
632 return 0;
633
634 ALLOC_GROW(list->list, list->nr + 1, list->alloc);
635 oidcpy(&(list->list[list->nr]), oid);
636 list->nr++;
637
638 return 0;
639 }
640
641 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
642 {
643 struct commit_list *parent;
644 for (parent = commit->parents; parent; parent = parent->next) {
645 if (!(parent->item->object.flags & UNINTERESTING)) {
646 ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
647 oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
648 oids->nr++;
649 parent->item->object.flags |= UNINTERESTING;
650 }
651 }
652 }
653
654 static void close_reachable(struct packed_oid_list *oids, int report_progress)
655 {
656 int i, j;
657 struct commit *commit;
658 struct progress *progress = NULL;
659
660 if (report_progress)
661 progress = start_delayed_progress(
662 _("Loading known commits in commit graph"), j = 0);
663 for (i = 0; i < oids->nr; i++) {
664 display_progress(progress, ++j);
665 commit = lookup_commit(the_repository, &oids->list[i]);
666 if (commit)
667 commit->object.flags |= UNINTERESTING;
668 }
669 stop_progress(&progress);
670
671 /*
672 * As this loop runs, oids->nr may grow, but not more
673 * than the number of missing commits in the reachable
674 * closure.
675 */
676 if (report_progress)
677 progress = start_delayed_progress(
678 _("Expanding reachable commits in commit graph"), j = 0);
679 for (i = 0; i < oids->nr; i++) {
680 display_progress(progress, ++j);
681 commit = lookup_commit(the_repository, &oids->list[i]);
682
683 if (commit && !parse_commit(commit))
684 add_missing_parents(oids, commit);
685 }
686 stop_progress(&progress);
687
688 if (report_progress)
689 progress = start_delayed_progress(
690 _("Clearing commit marks in commit graph"), j = 0);
691 for (i = 0; i < oids->nr; i++) {
692 display_progress(progress, ++j);
693 commit = lookup_commit(the_repository, &oids->list[i]);
694
695 if (commit)
696 commit->object.flags &= ~UNINTERESTING;
697 }
698 stop_progress(&progress);
699 }
700
701 static void compute_generation_numbers(struct packed_commit_list* commits,
702 int report_progress)
703 {
704 int i;
705 struct commit_list *list = NULL;
706 struct progress *progress = NULL;
707
708 if (report_progress)
709 progress = start_progress(
710 _("Computing commit graph generation numbers"),
711 commits->nr);
712 for (i = 0; i < commits->nr; i++) {
713 display_progress(progress, i + 1);
714 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
715 commits->list[i]->generation != GENERATION_NUMBER_ZERO)
716 continue;
717
718 commit_list_insert(commits->list[i], &list);
719 while (list) {
720 struct commit *current = list->item;
721 struct commit_list *parent;
722 int all_parents_computed = 1;
723 uint32_t max_generation = 0;
724
725 for (parent = current->parents; parent; parent = parent->next) {
726 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
727 parent->item->generation == GENERATION_NUMBER_ZERO) {
728 all_parents_computed = 0;
729 commit_list_insert(parent->item, &list);
730 break;
731 } else if (parent->item->generation > max_generation) {
732 max_generation = parent->item->generation;
733 }
734 }
735
736 if (all_parents_computed) {
737 current->generation = max_generation + 1;
738 pop_commit(&list);
739
740 if (current->generation > GENERATION_NUMBER_MAX)
741 current->generation = GENERATION_NUMBER_MAX;
742 }
743 }
744 }
745 stop_progress(&progress);
746 }
747
748 static int add_ref_to_list(const char *refname,
749 const struct object_id *oid,
750 int flags, void *cb_data)
751 {
752 struct string_list *list = (struct string_list *)cb_data;
753
754 string_list_append(list, oid_to_hex(oid));
755 return 0;
756 }
757
758 void write_commit_graph_reachable(const char *obj_dir, int append,
759 int report_progress)
760 {
761 struct string_list list = STRING_LIST_INIT_DUP;
762
763 for_each_ref(add_ref_to_list, &list);
764 write_commit_graph(obj_dir, NULL, &list, append, report_progress);
765
766 string_list_clear(&list, 0);
767 }
768
769 void write_commit_graph(const char *obj_dir,
770 struct string_list *pack_indexes,
771 struct string_list *commit_hex,
772 int append, int report_progress)
773 {
774 struct packed_oid_list oids;
775 struct packed_commit_list commits;
776 struct hashfile *f;
777 uint32_t i, count_distinct = 0;
778 char *graph_name;
779 struct lock_file lk = LOCK_INIT;
780 uint32_t chunk_ids[5];
781 uint64_t chunk_offsets[5];
782 int num_chunks;
783 int num_extra_edges;
784 struct commit_list *parent;
785 struct progress *progress = NULL;
786 uint64_t progress_cnt = 0;
787 struct strbuf progress_title = STRBUF_INIT;
788 unsigned long approx_nr_objects;
789
790 if (!commit_graph_compatible(the_repository))
791 return;
792
793 oids.nr = 0;
794 approx_nr_objects = approximate_object_count();
795 oids.alloc = approx_nr_objects / 32;
796 oids.progress = NULL;
797 oids.progress_done = 0;
798
799 if (append) {
800 prepare_commit_graph_one(the_repository, obj_dir);
801 if (the_repository->objects->commit_graph)
802 oids.alloc += the_repository->objects->commit_graph->num_commits;
803 }
804
805 if (oids.alloc < 1024)
806 oids.alloc = 1024;
807 ALLOC_ARRAY(oids.list, oids.alloc);
808
809 if (append && the_repository->objects->commit_graph) {
810 struct commit_graph *commit_graph =
811 the_repository->objects->commit_graph;
812 for (i = 0; i < commit_graph->num_commits; i++) {
813 const unsigned char *hash = commit_graph->chunk_oid_lookup +
814 commit_graph->hash_len * i;
815 hashcpy(oids.list[oids.nr++].hash, hash);
816 }
817 }
818
819 if (pack_indexes) {
820 struct strbuf packname = STRBUF_INIT;
821 int dirlen;
822 strbuf_addf(&packname, "%s/pack/", obj_dir);
823 dirlen = packname.len;
824 if (report_progress) {
825 strbuf_addf(&progress_title,
826 Q_("Finding commits for commit graph in %d pack",
827 "Finding commits for commit graph in %d packs",
828 pack_indexes->nr),
829 pack_indexes->nr);
830 oids.progress = start_delayed_progress(progress_title.buf, 0);
831 oids.progress_done = 0;
832 }
833 for (i = 0; i < pack_indexes->nr; i++) {
834 struct packed_git *p;
835 strbuf_setlen(&packname, dirlen);
836 strbuf_addstr(&packname, pack_indexes->items[i].string);
837 p = add_packed_git(packname.buf, packname.len, 1);
838 if (!p)
839 die(_("error adding pack %s"), packname.buf);
840 if (open_pack_index(p))
841 die(_("error opening index for %s"), packname.buf);
842 for_each_object_in_pack(p, add_packed_commits, &oids,
843 FOR_EACH_OBJECT_PACK_ORDER);
844 close_pack(p);
845 free(p);
846 }
847 stop_progress(&oids.progress);
848 strbuf_reset(&progress_title);
849 strbuf_release(&packname);
850 }
851
852 if (commit_hex) {
853 if (report_progress) {
854 strbuf_addf(&progress_title,
855 Q_("Finding commits for commit graph from %d ref",
856 "Finding commits for commit graph from %d refs",
857 commit_hex->nr),
858 commit_hex->nr);
859 progress = start_delayed_progress(progress_title.buf,
860 commit_hex->nr);
861 }
862 for (i = 0; i < commit_hex->nr; i++) {
863 const char *end;
864 struct object_id oid;
865 struct commit *result;
866
867 display_progress(progress, i + 1);
868 if (commit_hex->items[i].string &&
869 parse_oid_hex(commit_hex->items[i].string, &oid, &end))
870 continue;
871
872 result = lookup_commit_reference_gently(the_repository, &oid, 1);
873
874 if (result) {
875 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
876 oidcpy(&oids.list[oids.nr], &(result->object.oid));
877 oids.nr++;
878 }
879 }
880 stop_progress(&progress);
881 strbuf_reset(&progress_title);
882 }
883
884 if (!pack_indexes && !commit_hex) {
885 if (report_progress)
886 oids.progress = start_delayed_progress(
887 _("Finding commits for commit graph among packed objects"),
888 approx_nr_objects);
889 for_each_packed_object(add_packed_commits, &oids,
890 FOR_EACH_OBJECT_PACK_ORDER);
891 if (oids.progress_done < approx_nr_objects)
892 display_progress(oids.progress, approx_nr_objects);
893 stop_progress(&oids.progress);
894 }
895
896 close_reachable(&oids, report_progress);
897
898 QSORT(oids.list, oids.nr, commit_compare);
899
900 count_distinct = 1;
901 for (i = 1; i < oids.nr; i++) {
902 if (!oideq(&oids.list[i - 1], &oids.list[i]))
903 count_distinct++;
904 }
905
906 if (count_distinct >= GRAPH_PARENT_MISSING)
907 die(_("the commit graph format cannot write %d commits"), count_distinct);
908
909 commits.nr = 0;
910 commits.alloc = count_distinct;
911 ALLOC_ARRAY(commits.list, commits.alloc);
912
913 num_extra_edges = 0;
914 for (i = 0; i < oids.nr; i++) {
915 int num_parents = 0;
916 if (i > 0 && oideq(&oids.list[i - 1], &oids.list[i]))
917 continue;
918
919 commits.list[commits.nr] = lookup_commit(the_repository, &oids.list[i]);
920 parse_commit(commits.list[commits.nr]);
921
922 for (parent = commits.list[commits.nr]->parents;
923 parent; parent = parent->next)
924 num_parents++;
925
926 if (num_parents > 2)
927 num_extra_edges += num_parents - 1;
928
929 commits.nr++;
930 }
931 num_chunks = num_extra_edges ? 4 : 3;
932
933 if (commits.nr >= GRAPH_PARENT_MISSING)
934 die(_("too many commits to write graph"));
935
936 compute_generation_numbers(&commits, report_progress);
937
938 graph_name = get_commit_graph_filename(obj_dir);
939 if (safe_create_leading_directories(graph_name)) {
940 UNLEAK(graph_name);
941 die_errno(_("unable to create leading directories of %s"),
942 graph_name);
943 }
944
945 hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
946 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
947
948 hashwrite_be32(f, GRAPH_SIGNATURE);
949
950 hashwrite_u8(f, GRAPH_VERSION);
951 hashwrite_u8(f, GRAPH_OID_VERSION);
952 hashwrite_u8(f, num_chunks);
953 hashwrite_u8(f, 0); /* unused padding byte */
954
955 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
956 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
957 chunk_ids[2] = GRAPH_CHUNKID_DATA;
958 if (num_extra_edges)
959 chunk_ids[3] = GRAPH_CHUNKID_EXTRAEDGES;
960 else
961 chunk_ids[3] = 0;
962 chunk_ids[4] = 0;
963
964 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
965 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
966 chunk_offsets[2] = chunk_offsets[1] + GRAPH_OID_LEN * commits.nr;
967 chunk_offsets[3] = chunk_offsets[2] + (GRAPH_OID_LEN + 16) * commits.nr;
968 chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
969
970 for (i = 0; i <= num_chunks; i++) {
971 uint32_t chunk_write[3];
972
973 chunk_write[0] = htonl(chunk_ids[i]);
974 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
975 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
976 hashwrite(f, chunk_write, 12);
977 }
978
979 if (report_progress) {
980 strbuf_addf(&progress_title,
981 Q_("Writing out commit graph in %d pass",
982 "Writing out commit graph in %d passes",
983 num_chunks),
984 num_chunks);
985 progress = start_delayed_progress(
986 progress_title.buf,
987 num_chunks * commits.nr);
988 }
989 write_graph_chunk_fanout(f, commits.list, commits.nr, progress, &progress_cnt);
990 write_graph_chunk_oids(f, GRAPH_OID_LEN, commits.list, commits.nr, progress, &progress_cnt);
991 write_graph_chunk_data(f, GRAPH_OID_LEN, commits.list, commits.nr, progress, &progress_cnt);
992 if (num_extra_edges)
993 write_graph_chunk_extra_edges(f, commits.list, commits.nr, progress, &progress_cnt);
994 stop_progress(&progress);
995 strbuf_release(&progress_title);
996
997 close_commit_graph(the_repository);
998 finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
999 commit_lock_file(&lk);
1000
1001 free(graph_name);
1002 free(commits.list);
1003 free(oids.list);
1004 }
1005
1006 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1007 static int verify_commit_graph_error;
1008
1009 static void graph_report(const char *fmt, ...)
1010 {
1011 va_list ap;
1012
1013 verify_commit_graph_error = 1;
1014 va_start(ap, fmt);
1015 vfprintf(stderr, fmt, ap);
1016 fprintf(stderr, "\n");
1017 va_end(ap);
1018 }
1019
1020 #define GENERATION_ZERO_EXISTS 1
1021 #define GENERATION_NUMBER_EXISTS 2
1022
1023 int verify_commit_graph(struct repository *r, struct commit_graph *g)
1024 {
1025 uint32_t i, cur_fanout_pos = 0;
1026 struct object_id prev_oid, cur_oid, checksum;
1027 int generation_zero = 0;
1028 struct hashfile *f;
1029 int devnull;
1030 struct progress *progress = NULL;
1031
1032 if (!g) {
1033 graph_report("no commit-graph file loaded");
1034 return 1;
1035 }
1036
1037 verify_commit_graph_error = 0;
1038
1039 if (!g->chunk_oid_fanout)
1040 graph_report("commit-graph is missing the OID Fanout chunk");
1041 if (!g->chunk_oid_lookup)
1042 graph_report("commit-graph is missing the OID Lookup chunk");
1043 if (!g->chunk_commit_data)
1044 graph_report("commit-graph is missing the Commit Data chunk");
1045
1046 if (verify_commit_graph_error)
1047 return verify_commit_graph_error;
1048
1049 devnull = open("/dev/null", O_WRONLY);
1050 f = hashfd(devnull, NULL);
1051 hashwrite(f, g->data, g->data_len - g->hash_len);
1052 finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1053 if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1054 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1055 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1056 }
1057
1058 for (i = 0; i < g->num_commits; i++) {
1059 struct commit *graph_commit;
1060
1061 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1062
1063 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1064 graph_report("commit-graph has incorrect OID order: %s then %s",
1065 oid_to_hex(&prev_oid),
1066 oid_to_hex(&cur_oid));
1067
1068 oidcpy(&prev_oid, &cur_oid);
1069
1070 while (cur_oid.hash[0] > cur_fanout_pos) {
1071 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1072
1073 if (i != fanout_value)
1074 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1075 cur_fanout_pos, fanout_value, i);
1076 cur_fanout_pos++;
1077 }
1078
1079 graph_commit = lookup_commit(r, &cur_oid);
1080 if (!parse_commit_in_graph_one(g, graph_commit))
1081 graph_report("failed to parse %s from commit-graph",
1082 oid_to_hex(&cur_oid));
1083 }
1084
1085 while (cur_fanout_pos < 256) {
1086 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1087
1088 if (g->num_commits != fanout_value)
1089 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1090 cur_fanout_pos, fanout_value, i);
1091
1092 cur_fanout_pos++;
1093 }
1094
1095 if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1096 return verify_commit_graph_error;
1097
1098 progress = start_progress(_("Verifying commits in commit graph"),
1099 g->num_commits);
1100 for (i = 0; i < g->num_commits; i++) {
1101 struct commit *graph_commit, *odb_commit;
1102 struct commit_list *graph_parents, *odb_parents;
1103 uint32_t max_generation = 0;
1104
1105 display_progress(progress, i + 1);
1106 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1107
1108 graph_commit = lookup_commit(r, &cur_oid);
1109 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1110 if (parse_commit_internal(odb_commit, 0, 0)) {
1111 graph_report("failed to parse %s from object database",
1112 oid_to_hex(&cur_oid));
1113 continue;
1114 }
1115
1116 if (!oideq(&get_commit_tree_in_graph_one(g, graph_commit)->object.oid,
1117 get_commit_tree_oid(odb_commit)))
1118 graph_report("root tree OID for commit %s in commit-graph is %s != %s",
1119 oid_to_hex(&cur_oid),
1120 oid_to_hex(get_commit_tree_oid(graph_commit)),
1121 oid_to_hex(get_commit_tree_oid(odb_commit)));
1122
1123 graph_parents = graph_commit->parents;
1124 odb_parents = odb_commit->parents;
1125
1126 while (graph_parents) {
1127 if (odb_parents == NULL) {
1128 graph_report("commit-graph parent list for commit %s is too long",
1129 oid_to_hex(&cur_oid));
1130 break;
1131 }
1132
1133 if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1134 graph_report("commit-graph parent for %s is %s != %s",
1135 oid_to_hex(&cur_oid),
1136 oid_to_hex(&graph_parents->item->object.oid),
1137 oid_to_hex(&odb_parents->item->object.oid));
1138
1139 if (graph_parents->item->generation > max_generation)
1140 max_generation = graph_parents->item->generation;
1141
1142 graph_parents = graph_parents->next;
1143 odb_parents = odb_parents->next;
1144 }
1145
1146 if (odb_parents != NULL)
1147 graph_report("commit-graph parent list for commit %s terminates early",
1148 oid_to_hex(&cur_oid));
1149
1150 if (!graph_commit->generation) {
1151 if (generation_zero == GENERATION_NUMBER_EXISTS)
1152 graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
1153 oid_to_hex(&cur_oid));
1154 generation_zero = GENERATION_ZERO_EXISTS;
1155 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1156 graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
1157 oid_to_hex(&cur_oid));
1158
1159 if (generation_zero == GENERATION_ZERO_EXISTS)
1160 continue;
1161
1162 /*
1163 * If one of our parents has generation GENERATION_NUMBER_MAX, then
1164 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1165 * extra logic in the following condition.
1166 */
1167 if (max_generation == GENERATION_NUMBER_MAX)
1168 max_generation--;
1169
1170 if (graph_commit->generation != max_generation + 1)
1171 graph_report("commit-graph generation for commit %s is %u != %u",
1172 oid_to_hex(&cur_oid),
1173 graph_commit->generation,
1174 max_generation + 1);
1175
1176 if (graph_commit->date != odb_commit->date)
1177 graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
1178 oid_to_hex(&cur_oid),
1179 graph_commit->date,
1180 odb_commit->date);
1181 }
1182 stop_progress(&progress);
1183
1184 return verify_commit_graph_error;
1185 }
1186
1187 void free_commit_graph(struct commit_graph *g)
1188 {
1189 if (!g)
1190 return;
1191 if (g->graph_fd >= 0) {
1192 munmap((void *)g->data, g->data_len);
1193 g->data = NULL;
1194 close(g->graph_fd);
1195 }
1196 free(g);
1197 }