Merge branch 'jk/ewah-bounds-check'
[alt-git.git] / commit-graph.c
blob4c6127088ff96282fbcdc98ef5767281e52846e9
1 #include "cache.h"
2 #include "config.h"
3 #include "git-compat-util.h"
4 #include "lockfile.h"
5 #include "pack.h"
6 #include "packfile.h"
7 #include "commit.h"
8 #include "object.h"
9 #include "revision.h"
10 #include "sha1-lookup.h"
11 #include "commit-graph.h"
12 #include "object-store.h"
14 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
15 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
16 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
17 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
18 #define GRAPH_CHUNKID_LARGEEDGES 0x45444745 /* "EDGE" */
20 #define GRAPH_DATA_WIDTH 36
22 #define GRAPH_VERSION_1 0x1
23 #define GRAPH_VERSION GRAPH_VERSION_1
25 #define GRAPH_OID_VERSION_SHA1 1
26 #define GRAPH_OID_LEN_SHA1 GIT_SHA1_RAWSZ
27 #define GRAPH_OID_VERSION GRAPH_OID_VERSION_SHA1
28 #define GRAPH_OID_LEN GRAPH_OID_LEN_SHA1
30 #define GRAPH_OCTOPUS_EDGES_NEEDED 0x80000000
31 #define GRAPH_PARENT_MISSING 0x7fffffff
32 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
33 #define GRAPH_PARENT_NONE 0x70000000
35 #define GRAPH_LAST_EDGE 0x80000000
37 #define GRAPH_FANOUT_SIZE (4 * 256)
38 #define GRAPH_CHUNKLOOKUP_WIDTH 12
39 #define GRAPH_MIN_SIZE (5 * GRAPH_CHUNKLOOKUP_WIDTH + GRAPH_FANOUT_SIZE + \
40 GRAPH_OID_LEN + 8)
42 char *get_commit_graph_filename(const char *obj_dir)
44 return xstrfmt("%s/info/commit-graph", obj_dir);
47 static struct commit_graph *alloc_commit_graph(void)
49 struct commit_graph *g = xcalloc(1, sizeof(*g));
50 g->graph_fd = -1;
52 return g;
55 struct commit_graph *load_commit_graph_one(const char *graph_file)
57 void *graph_map;
58 const unsigned char *data, *chunk_lookup;
59 size_t graph_size;
60 struct stat st;
61 uint32_t i;
62 struct commit_graph *graph;
63 int fd = git_open(graph_file);
64 uint64_t last_chunk_offset;
65 uint32_t last_chunk_id;
66 uint32_t graph_signature;
67 unsigned char graph_version, hash_version;
69 if (fd < 0)
70 return NULL;
71 if (fstat(fd, &st)) {
72 close(fd);
73 return NULL;
75 graph_size = xsize_t(st.st_size);
77 if (graph_size < GRAPH_MIN_SIZE) {
78 close(fd);
79 die("graph file %s is too small", graph_file);
81 graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
82 data = (const unsigned char *)graph_map;
84 graph_signature = get_be32(data);
85 if (graph_signature != GRAPH_SIGNATURE) {
86 error("graph signature %X does not match signature %X",
87 graph_signature, GRAPH_SIGNATURE);
88 goto cleanup_fail;
91 graph_version = *(unsigned char*)(data + 4);
92 if (graph_version != GRAPH_VERSION) {
93 error("graph version %X does not match version %X",
94 graph_version, GRAPH_VERSION);
95 goto cleanup_fail;
98 hash_version = *(unsigned char*)(data + 5);
99 if (hash_version != GRAPH_OID_VERSION) {
100 error("hash version %X does not match version %X",
101 hash_version, GRAPH_OID_VERSION);
102 goto cleanup_fail;
105 graph = alloc_commit_graph();
107 graph->hash_len = GRAPH_OID_LEN;
108 graph->num_chunks = *(unsigned char*)(data + 6);
109 graph->graph_fd = fd;
110 graph->data = graph_map;
111 graph->data_len = graph_size;
113 last_chunk_id = 0;
114 last_chunk_offset = 8;
115 chunk_lookup = data + 8;
116 for (i = 0; i < graph->num_chunks; i++) {
117 uint32_t chunk_id = get_be32(chunk_lookup + 0);
118 uint64_t chunk_offset = get_be64(chunk_lookup + 4);
119 int chunk_repeated = 0;
121 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
123 if (chunk_offset > graph_size - GIT_MAX_RAWSZ) {
124 error("improper chunk offset %08x%08x", (uint32_t)(chunk_offset >> 32),
125 (uint32_t)chunk_offset);
126 goto cleanup_fail;
129 switch (chunk_id) {
130 case GRAPH_CHUNKID_OIDFANOUT:
131 if (graph->chunk_oid_fanout)
132 chunk_repeated = 1;
133 else
134 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
135 break;
137 case GRAPH_CHUNKID_OIDLOOKUP:
138 if (graph->chunk_oid_lookup)
139 chunk_repeated = 1;
140 else
141 graph->chunk_oid_lookup = data + chunk_offset;
142 break;
144 case GRAPH_CHUNKID_DATA:
145 if (graph->chunk_commit_data)
146 chunk_repeated = 1;
147 else
148 graph->chunk_commit_data = data + chunk_offset;
149 break;
151 case GRAPH_CHUNKID_LARGEEDGES:
152 if (graph->chunk_large_edges)
153 chunk_repeated = 1;
154 else
155 graph->chunk_large_edges = data + chunk_offset;
156 break;
159 if (chunk_repeated) {
160 error("chunk id %08x appears multiple times", chunk_id);
161 goto cleanup_fail;
164 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
166 graph->num_commits = (chunk_offset - last_chunk_offset)
167 / graph->hash_len;
170 last_chunk_id = chunk_id;
171 last_chunk_offset = chunk_offset;
174 return graph;
176 cleanup_fail:
177 munmap(graph_map, graph_size);
178 close(fd);
179 exit(1);
182 /* global storage */
183 static struct commit_graph *commit_graph = NULL;
185 static void prepare_commit_graph_one(const char *obj_dir)
187 char *graph_name;
189 if (commit_graph)
190 return;
192 graph_name = get_commit_graph_filename(obj_dir);
193 commit_graph = load_commit_graph_one(graph_name);
195 FREE_AND_NULL(graph_name);
198 static int prepare_commit_graph_run_once = 0;
199 static void prepare_commit_graph(void)
201 struct alternate_object_database *alt;
202 char *obj_dir;
204 if (prepare_commit_graph_run_once)
205 return;
206 prepare_commit_graph_run_once = 1;
208 obj_dir = get_object_directory();
209 prepare_commit_graph_one(obj_dir);
210 prepare_alt_odb(the_repository);
211 for (alt = the_repository->objects->alt_odb_list;
212 !commit_graph && alt;
213 alt = alt->next)
214 prepare_commit_graph_one(alt->path);
217 static void close_commit_graph(void)
219 if (!commit_graph)
220 return;
222 if (commit_graph->graph_fd >= 0) {
223 munmap((void *)commit_graph->data, commit_graph->data_len);
224 commit_graph->data = NULL;
225 close(commit_graph->graph_fd);
228 FREE_AND_NULL(commit_graph);
231 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
233 return bsearch_hash(oid->hash, g->chunk_oid_fanout,
234 g->chunk_oid_lookup, g->hash_len, pos);
237 static struct commit_list **insert_parent_or_die(struct commit_graph *g,
238 uint64_t pos,
239 struct commit_list **pptr)
241 struct commit *c;
242 struct object_id oid;
243 hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
244 c = lookup_commit(&oid);
245 if (!c)
246 die("could not find commit %s", oid_to_hex(&oid));
247 c->graph_pos = pos;
248 return &commit_list_insert(c, pptr)->next;
251 static int fill_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t pos)
253 uint32_t edge_value;
254 uint32_t *parent_data_ptr;
255 uint64_t date_low, date_high;
256 struct commit_list **pptr;
257 const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
259 item->object.parsed = 1;
260 item->graph_pos = pos;
262 item->maybe_tree = NULL;
264 date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
265 date_low = get_be32(commit_data + g->hash_len + 12);
266 item->date = (timestamp_t)((date_high << 32) | date_low);
268 pptr = &item->parents;
270 edge_value = get_be32(commit_data + g->hash_len);
271 if (edge_value == GRAPH_PARENT_NONE)
272 return 1;
273 pptr = insert_parent_or_die(g, edge_value, pptr);
275 edge_value = get_be32(commit_data + g->hash_len + 4);
276 if (edge_value == GRAPH_PARENT_NONE)
277 return 1;
278 if (!(edge_value & GRAPH_OCTOPUS_EDGES_NEEDED)) {
279 pptr = insert_parent_or_die(g, edge_value, pptr);
280 return 1;
283 parent_data_ptr = (uint32_t*)(g->chunk_large_edges +
284 4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
285 do {
286 edge_value = get_be32(parent_data_ptr);
287 pptr = insert_parent_or_die(g,
288 edge_value & GRAPH_EDGE_LAST_MASK,
289 pptr);
290 parent_data_ptr++;
291 } while (!(edge_value & GRAPH_LAST_EDGE));
293 return 1;
296 int parse_commit_in_graph(struct commit *item)
298 if (!core_commit_graph)
299 return 0;
300 if (item->object.parsed)
301 return 1;
303 prepare_commit_graph();
304 if (commit_graph) {
305 uint32_t pos;
306 int found;
307 if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
308 pos = item->graph_pos;
309 found = 1;
310 } else {
311 found = bsearch_graph(commit_graph, &(item->object.oid), &pos);
314 if (found)
315 return fill_commit_in_graph(item, commit_graph, pos);
318 return 0;
321 static struct tree *load_tree_for_commit(struct commit_graph *g, struct commit *c)
323 struct object_id oid;
324 const unsigned char *commit_data = g->chunk_commit_data +
325 GRAPH_DATA_WIDTH * (c->graph_pos);
327 hashcpy(oid.hash, commit_data);
328 c->maybe_tree = lookup_tree(&oid);
330 return c->maybe_tree;
333 struct tree *get_commit_tree_in_graph(const struct commit *c)
335 if (c->maybe_tree)
336 return c->maybe_tree;
337 if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
338 BUG("get_commit_tree_in_graph called from non-commit-graph commit");
340 return load_tree_for_commit(commit_graph, (struct commit *)c);
343 static void write_graph_chunk_fanout(struct hashfile *f,
344 struct commit **commits,
345 int nr_commits)
347 int i, count = 0;
348 struct commit **list = commits;
351 * Write the first-level table (the list is sorted,
352 * but we use a 256-entry lookup to be able to avoid
353 * having to do eight extra binary search iterations).
355 for (i = 0; i < 256; i++) {
356 while (count < nr_commits) {
357 if ((*list)->object.oid.hash[0] != i)
358 break;
359 count++;
360 list++;
363 hashwrite_be32(f, count);
367 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
368 struct commit **commits, int nr_commits)
370 struct commit **list = commits;
371 int count;
372 for (count = 0; count < nr_commits; count++, list++)
373 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
376 static const unsigned char *commit_to_sha1(size_t index, void *table)
378 struct commit **commits = table;
379 return commits[index]->object.oid.hash;
382 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
383 struct commit **commits, int nr_commits)
385 struct commit **list = commits;
386 struct commit **last = commits + nr_commits;
387 uint32_t num_extra_edges = 0;
389 while (list < last) {
390 struct commit_list *parent;
391 int edge_value;
392 uint32_t packedDate[2];
394 parse_commit(*list);
395 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
397 parent = (*list)->parents;
399 if (!parent)
400 edge_value = GRAPH_PARENT_NONE;
401 else {
402 edge_value = sha1_pos(parent->item->object.oid.hash,
403 commits,
404 nr_commits,
405 commit_to_sha1);
407 if (edge_value < 0)
408 edge_value = GRAPH_PARENT_MISSING;
411 hashwrite_be32(f, edge_value);
413 if (parent)
414 parent = parent->next;
416 if (!parent)
417 edge_value = GRAPH_PARENT_NONE;
418 else if (parent->next)
419 edge_value = GRAPH_OCTOPUS_EDGES_NEEDED | num_extra_edges;
420 else {
421 edge_value = sha1_pos(parent->item->object.oid.hash,
422 commits,
423 nr_commits,
424 commit_to_sha1);
425 if (edge_value < 0)
426 edge_value = GRAPH_PARENT_MISSING;
429 hashwrite_be32(f, edge_value);
431 if (edge_value & GRAPH_OCTOPUS_EDGES_NEEDED) {
432 do {
433 num_extra_edges++;
434 parent = parent->next;
435 } while (parent);
438 if (sizeof((*list)->date) > 4)
439 packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
440 else
441 packedDate[0] = 0;
443 packedDate[1] = htonl((*list)->date);
444 hashwrite(f, packedDate, 8);
446 list++;
450 static void write_graph_chunk_large_edges(struct hashfile *f,
451 struct commit **commits,
452 int nr_commits)
454 struct commit **list = commits;
455 struct commit **last = commits + nr_commits;
456 struct commit_list *parent;
458 while (list < last) {
459 int num_parents = 0;
460 for (parent = (*list)->parents; num_parents < 3 && parent;
461 parent = parent->next)
462 num_parents++;
464 if (num_parents <= 2) {
465 list++;
466 continue;
469 /* Since num_parents > 2, this initializer is safe. */
470 for (parent = (*list)->parents->next; parent; parent = parent->next) {
471 int edge_value = sha1_pos(parent->item->object.oid.hash,
472 commits,
473 nr_commits,
474 commit_to_sha1);
476 if (edge_value < 0)
477 edge_value = GRAPH_PARENT_MISSING;
478 else if (!parent->next)
479 edge_value |= GRAPH_LAST_EDGE;
481 hashwrite_be32(f, edge_value);
484 list++;
488 static int commit_compare(const void *_a, const void *_b)
490 const struct object_id *a = (const struct object_id *)_a;
491 const struct object_id *b = (const struct object_id *)_b;
492 return oidcmp(a, b);
495 struct packed_commit_list {
496 struct commit **list;
497 int nr;
498 int alloc;
501 struct packed_oid_list {
502 struct object_id *list;
503 int nr;
504 int alloc;
507 static int add_packed_commits(const struct object_id *oid,
508 struct packed_git *pack,
509 uint32_t pos,
510 void *data)
512 struct packed_oid_list *list = (struct packed_oid_list*)data;
513 enum object_type type;
514 off_t offset = nth_packed_object_offset(pack, pos);
515 struct object_info oi = OBJECT_INFO_INIT;
517 oi.typep = &type;
518 if (packed_object_info(the_repository, pack, offset, &oi) < 0)
519 die("unable to get type of object %s", oid_to_hex(oid));
521 if (type != OBJ_COMMIT)
522 return 0;
524 ALLOC_GROW(list->list, list->nr + 1, list->alloc);
525 oidcpy(&(list->list[list->nr]), oid);
526 list->nr++;
528 return 0;
531 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
533 struct commit_list *parent;
534 for (parent = commit->parents; parent; parent = parent->next) {
535 if (!(parent->item->object.flags & UNINTERESTING)) {
536 ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
537 oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
538 oids->nr++;
539 parent->item->object.flags |= UNINTERESTING;
544 static void close_reachable(struct packed_oid_list *oids)
546 int i;
547 struct commit *commit;
549 for (i = 0; i < oids->nr; i++) {
550 commit = lookup_commit(&oids->list[i]);
551 if (commit)
552 commit->object.flags |= UNINTERESTING;
556 * As this loop runs, oids->nr may grow, but not more
557 * than the number of missing commits in the reachable
558 * closure.
560 for (i = 0; i < oids->nr; i++) {
561 commit = lookup_commit(&oids->list[i]);
563 if (commit && !parse_commit(commit))
564 add_missing_parents(oids, commit);
567 for (i = 0; i < oids->nr; i++) {
568 commit = lookup_commit(&oids->list[i]);
570 if (commit)
571 commit->object.flags &= ~UNINTERESTING;
575 void write_commit_graph(const char *obj_dir,
576 const char **pack_indexes,
577 int nr_packs,
578 const char **commit_hex,
579 int nr_commits,
580 int append)
582 struct packed_oid_list oids;
583 struct packed_commit_list commits;
584 struct hashfile *f;
585 uint32_t i, count_distinct = 0;
586 char *graph_name;
587 int fd;
588 struct lock_file lk = LOCK_INIT;
589 uint32_t chunk_ids[5];
590 uint64_t chunk_offsets[5];
591 int num_chunks;
592 int num_extra_edges;
593 struct commit_list *parent;
595 oids.nr = 0;
596 oids.alloc = approximate_object_count() / 4;
598 if (append) {
599 prepare_commit_graph_one(obj_dir);
600 if (commit_graph)
601 oids.alloc += commit_graph->num_commits;
604 if (oids.alloc < 1024)
605 oids.alloc = 1024;
606 ALLOC_ARRAY(oids.list, oids.alloc);
608 if (append && commit_graph) {
609 for (i = 0; i < commit_graph->num_commits; i++) {
610 const unsigned char *hash = commit_graph->chunk_oid_lookup +
611 commit_graph->hash_len * i;
612 hashcpy(oids.list[oids.nr++].hash, hash);
616 if (pack_indexes) {
617 struct strbuf packname = STRBUF_INIT;
618 int dirlen;
619 strbuf_addf(&packname, "%s/pack/", obj_dir);
620 dirlen = packname.len;
621 for (i = 0; i < nr_packs; i++) {
622 struct packed_git *p;
623 strbuf_setlen(&packname, dirlen);
624 strbuf_addstr(&packname, pack_indexes[i]);
625 p = add_packed_git(packname.buf, packname.len, 1);
626 if (!p)
627 die("error adding pack %s", packname.buf);
628 if (open_pack_index(p))
629 die("error opening index for %s", packname.buf);
630 for_each_object_in_pack(p, add_packed_commits, &oids);
631 close_pack(p);
633 strbuf_release(&packname);
636 if (commit_hex) {
637 for (i = 0; i < nr_commits; i++) {
638 const char *end;
639 struct object_id oid;
640 struct commit *result;
642 if (commit_hex[i] && parse_oid_hex(commit_hex[i], &oid, &end))
643 continue;
645 result = lookup_commit_reference_gently(&oid, 1);
647 if (result) {
648 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
649 oidcpy(&oids.list[oids.nr], &(result->object.oid));
650 oids.nr++;
655 if (!pack_indexes && !commit_hex)
656 for_each_packed_object(add_packed_commits, &oids, 0);
658 close_reachable(&oids);
660 QSORT(oids.list, oids.nr, commit_compare);
662 count_distinct = 1;
663 for (i = 1; i < oids.nr; i++) {
664 if (oidcmp(&oids.list[i-1], &oids.list[i]))
665 count_distinct++;
668 if (count_distinct >= GRAPH_PARENT_MISSING)
669 die(_("the commit graph format cannot write %d commits"), count_distinct);
671 commits.nr = 0;
672 commits.alloc = count_distinct;
673 ALLOC_ARRAY(commits.list, commits.alloc);
675 num_extra_edges = 0;
676 for (i = 0; i < oids.nr; i++) {
677 int num_parents = 0;
678 if (i > 0 && !oidcmp(&oids.list[i-1], &oids.list[i]))
679 continue;
681 commits.list[commits.nr] = lookup_commit(&oids.list[i]);
682 parse_commit(commits.list[commits.nr]);
684 for (parent = commits.list[commits.nr]->parents;
685 parent; parent = parent->next)
686 num_parents++;
688 if (num_parents > 2)
689 num_extra_edges += num_parents - 1;
691 commits.nr++;
693 num_chunks = num_extra_edges ? 4 : 3;
695 if (commits.nr >= GRAPH_PARENT_MISSING)
696 die(_("too many commits to write graph"));
698 graph_name = get_commit_graph_filename(obj_dir);
699 fd = hold_lock_file_for_update(&lk, graph_name, 0);
701 if (fd < 0) {
702 struct strbuf folder = STRBUF_INIT;
703 strbuf_addstr(&folder, graph_name);
704 strbuf_setlen(&folder, strrchr(folder.buf, '/') - folder.buf);
706 if (mkdir(folder.buf, 0777) < 0)
707 die_errno(_("cannot mkdir %s"), folder.buf);
708 strbuf_release(&folder);
710 fd = hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
712 if (fd < 0)
713 die_errno("unable to create '%s'", graph_name);
716 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
718 hashwrite_be32(f, GRAPH_SIGNATURE);
720 hashwrite_u8(f, GRAPH_VERSION);
721 hashwrite_u8(f, GRAPH_OID_VERSION);
722 hashwrite_u8(f, num_chunks);
723 hashwrite_u8(f, 0); /* unused padding byte */
725 chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
726 chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
727 chunk_ids[2] = GRAPH_CHUNKID_DATA;
728 if (num_extra_edges)
729 chunk_ids[3] = GRAPH_CHUNKID_LARGEEDGES;
730 else
731 chunk_ids[3] = 0;
732 chunk_ids[4] = 0;
734 chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
735 chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
736 chunk_offsets[2] = chunk_offsets[1] + GRAPH_OID_LEN * commits.nr;
737 chunk_offsets[3] = chunk_offsets[2] + (GRAPH_OID_LEN + 16) * commits.nr;
738 chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
740 for (i = 0; i <= num_chunks; i++) {
741 uint32_t chunk_write[3];
743 chunk_write[0] = htonl(chunk_ids[i]);
744 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
745 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
746 hashwrite(f, chunk_write, 12);
749 write_graph_chunk_fanout(f, commits.list, commits.nr);
750 write_graph_chunk_oids(f, GRAPH_OID_LEN, commits.list, commits.nr);
751 write_graph_chunk_data(f, GRAPH_OID_LEN, commits.list, commits.nr);
752 write_graph_chunk_large_edges(f, commits.list, commits.nr);
754 close_commit_graph();
755 finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
756 commit_lock_file(&lk);
758 free(oids.list);
759 oids.alloc = 0;
760 oids.nr = 0;