t/perf: add perf tests for fetches from a bitmapped server
[git/git.git] / pack-bitmap.c
CommitLineData
fff42755
VM
1#include "cache.h"
2#include "commit.h"
3#include "tag.h"
4#include "diff.h"
5#include "revision.h"
6#include "progress.h"
7#include "list-objects.h"
8#include "pack.h"
9#include "pack-bitmap.h"
10#include "pack-revindex.h"
11#include "pack-objects.h"
0317f455 12#include "packfile.h"
a80d72db
SB
13#include "repository.h"
14#include "object-store.h"
fff42755
VM
15
16/*
17 * An entry on the bitmap index, representing the bitmap for a given
18 * commit.
19 */
20struct stored_bitmap {
21 unsigned char sha1[20];
22 struct ewah_bitmap *root;
23 struct stored_bitmap *xor;
24 int flags;
25};
26
27/*
3ae5fa07 28 * The active bitmap index for a repository. By design, repositories only have
fff42755
VM
29 * a single bitmap index available (the index for the biggest packfile in
30 * the repository), since bitmap indexes need full closure.
31 *
32 * If there is more than one bitmap index available (e.g. because of alternates),
33 * the active bitmap index is the largest one.
34 */
3ae5fa07 35struct bitmap_index {
fff42755
VM
36 /* Packfile to which this bitmap index belongs to */
37 struct packed_git *pack;
38
fff42755
VM
39 /*
40 * Mark the first `reuse_objects` in the packfile as reused:
41 * they will be sent as-is without using them for repacking
42 * calculations
43 */
44 uint32_t reuse_objects;
45
46 /* mmapped buffer of the whole bitmap index */
47 unsigned char *map;
48 size_t map_size; /* size of the mmaped buffer */
49 size_t map_pos; /* current position when loading the index */
50
51 /*
52 * Type indexes.
53 *
54 * Each bitmap marks which objects in the packfile are of the given
55 * type. This provides type information when yielding the objects from
56 * the packfile during a walk, which allows for better delta bases.
57 */
58 struct ewah_bitmap *commits;
59 struct ewah_bitmap *trees;
60 struct ewah_bitmap *blobs;
61 struct ewah_bitmap *tags;
62
25143a54 63 /* Map from SHA1 -> `stored_bitmap` for all the bitmapped commits */
fff42755
VM
64 khash_sha1 *bitmaps;
65
66 /* Number of bitmapped commits */
67 uint32_t entry_count;
68
f3c23db2 69 /* If not NULL, this is a name-hash cache pointing into map. */
ae4f07fb
VM
70 uint32_t *hashes;
71
fff42755
VM
72 /*
73 * Extended index.
74 *
75 * When trying to perform bitmap operations with objects that are not
76 * packed in `pack`, these objects are added to this "fake index" and
77 * are assumed to appear at the end of the packfile for all operations
78 */
79 struct eindex {
80 struct object **objects;
81 uint32_t *hashes;
82 uint32_t count, alloc;
83 khash_sha1_pos *positions;
84 } ext_index;
85
86 /* Bitmap result of the last performed walk */
87 struct bitmap *result;
88
89 /* Version of the bitmap index */
90 unsigned int version;
91
92 unsigned loaded : 1;
3ae5fa07 93};
fff42755
VM
94
95static struct ewah_bitmap *lookup_stored_bitmap(struct stored_bitmap *st)
96{
97 struct ewah_bitmap *parent;
98 struct ewah_bitmap *composed;
99
100 if (st->xor == NULL)
101 return st->root;
102
103 composed = ewah_pool_new();
104 parent = lookup_stored_bitmap(st->xor);
105 ewah_xor(st->root, parent, composed);
106
107 ewah_pool_free(st->root);
108 st->root = composed;
109 st->xor = NULL;
110
111 return composed;
112}
113
114/*
115 * Read a bitmap from the current read position on the mmaped
116 * index, and increase the read position accordingly
117 */
118static struct ewah_bitmap *read_bitmap_1(struct bitmap_index *index)
119{
120 struct ewah_bitmap *b = ewah_pool_new();
121
1140bf01 122 ssize_t bitmap_size = ewah_read_mmap(b,
fff42755
VM
123 index->map + index->map_pos,
124 index->map_size - index->map_pos);
125
126 if (bitmap_size < 0) {
127 error("Failed to load bitmap index (corrupted?)");
128 ewah_pool_free(b);
129 return NULL;
130 }
131
132 index->map_pos += bitmap_size;
133 return b;
134}
135
136static int load_bitmap_header(struct bitmap_index *index)
137{
138 struct bitmap_disk_header *header = (void *)index->map;
139
140 if (index->map_size < sizeof(*header) + 20)
141 return error("Corrupted bitmap index (missing header data)");
142
143 if (memcmp(header->magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE)) != 0)
144 return error("Corrupted bitmap index file (wrong header)");
145
146 index->version = ntohs(header->version);
147 if (index->version != 1)
148 return error("Unsupported version for bitmap index file (%d)", index->version);
149
150 /* Parse known bitmap format options */
151 {
152 uint32_t flags = ntohs(header->options);
153
154 if ((flags & BITMAP_OPT_FULL_DAG) == 0)
155 return error("Unsupported options for bitmap index file "
156 "(Git requires BITMAP_OPT_FULL_DAG)");
ae4f07fb
VM
157
158 if (flags & BITMAP_OPT_HASH_CACHE) {
159 unsigned char *end = index->map + index->map_size - 20;
160 index->hashes = ((uint32_t *)end) - index->pack->num_objects;
161 }
fff42755
VM
162 }
163
164 index->entry_count = ntohl(header->entry_count);
165 index->map_pos += sizeof(*header);
166 return 0;
167}
168
169static struct stored_bitmap *store_bitmap(struct bitmap_index *index,
170 struct ewah_bitmap *root,
171 const unsigned char *sha1,
172 struct stored_bitmap *xor_with,
173 int flags)
174{
175 struct stored_bitmap *stored;
176 khiter_t hash_pos;
177 int ret;
178
179 stored = xmalloc(sizeof(struct stored_bitmap));
180 stored->root = root;
181 stored->xor = xor_with;
182 stored->flags = flags;
183 hashcpy(stored->sha1, sha1);
184
185 hash_pos = kh_put_sha1(index->bitmaps, stored->sha1, &ret);
186
187 /* a 0 return code means the insertion succeeded with no changes,
188 * because the SHA1 already existed on the map. this is bad, there
189 * shouldn't be duplicated commits in the index */
190 if (ret == 0) {
191 error("Duplicate entry in bitmap index: %s", sha1_to_hex(sha1));
192 return NULL;
193 }
194
195 kh_value(index->bitmaps, hash_pos) = stored;
196 return stored;
197}
198
b5007211
KB
199static inline uint32_t read_be32(const unsigned char *buffer, size_t *pos)
200{
201 uint32_t result = get_be32(buffer + *pos);
202 (*pos) += sizeof(result);
203 return result;
204}
205
206static inline uint8_t read_u8(const unsigned char *buffer, size_t *pos)
207{
208 return buffer[(*pos)++];
209}
210
599dc766
RS
211#define MAX_XOR_OFFSET 160
212
fff42755
VM
213static int load_bitmap_entries_v1(struct bitmap_index *index)
214{
fff42755 215 uint32_t i;
599dc766 216 struct stored_bitmap *recent_bitmaps[MAX_XOR_OFFSET] = { NULL };
fff42755
VM
217
218 for (i = 0; i < index->entry_count; ++i) {
219 int xor_offset, flags;
220 struct ewah_bitmap *bitmap = NULL;
221 struct stored_bitmap *xor_bitmap = NULL;
222 uint32_t commit_idx_pos;
223 const unsigned char *sha1;
224
b5007211
KB
225 commit_idx_pos = read_be32(index->map, &index->map_pos);
226 xor_offset = read_u8(index->map, &index->map_pos);
227 flags = read_u8(index->map, &index->map_pos);
fff42755 228
fff42755
VM
229 sha1 = nth_packed_object_sha1(index->pack, commit_idx_pos);
230
fff42755
VM
231 bitmap = read_bitmap_1(index);
232 if (!bitmap)
233 return -1;
234
235 if (xor_offset > MAX_XOR_OFFSET || xor_offset > i)
236 return error("Corrupted bitmap pack index");
237
238 if (xor_offset > 0) {
239 xor_bitmap = recent_bitmaps[(i - xor_offset) % MAX_XOR_OFFSET];
240
241 if (xor_bitmap == NULL)
242 return error("Invalid XOR offset in bitmap pack index");
243 }
244
245 recent_bitmaps[i % MAX_XOR_OFFSET] = store_bitmap(
246 index, bitmap, sha1, xor_bitmap, flags);
247 }
248
249 return 0;
250}
251
cb468050
JH
252static char *pack_bitmap_filename(struct packed_git *p)
253{
9ae97018 254 size_t len;
cb468050 255
9ae97018 256 if (!strip_suffix(p->pack_name, ".pack", &len))
033abf97 257 BUG("pack_name does not end in .pack");
9ae97018 258 return xstrfmt("%.*s.bitmap", (int)len, p->pack_name);
cb468050
JH
259}
260
3ae5fa07 261static int open_pack_bitmap_1(struct bitmap_index *bitmap_git, struct packed_git *packfile)
fff42755
VM
262{
263 int fd;
264 struct stat st;
265 char *idx_name;
266
267 if (open_pack_index(packfile))
268 return -1;
269
270 idx_name = pack_bitmap_filename(packfile);
a5436b57 271 fd = git_open(idx_name);
fff42755
VM
272 free(idx_name);
273
274 if (fd < 0)
275 return -1;
276
277 if (fstat(fd, &st)) {
278 close(fd);
279 return -1;
280 }
281
3ae5fa07 282 if (bitmap_git->pack) {
fff42755
VM
283 warning("ignoring extra bitmap file: %s", packfile->pack_name);
284 close(fd);
285 return -1;
286 }
287
3ae5fa07
JT
288 bitmap_git->pack = packfile;
289 bitmap_git->map_size = xsize_t(st.st_size);
290 bitmap_git->map = xmmap(NULL, bitmap_git->map_size, PROT_READ, MAP_PRIVATE, fd, 0);
291 bitmap_git->map_pos = 0;
fff42755
VM
292 close(fd);
293
3ae5fa07
JT
294 if (load_bitmap_header(bitmap_git) < 0) {
295 munmap(bitmap_git->map, bitmap_git->map_size);
296 bitmap_git->map = NULL;
297 bitmap_git->map_size = 0;
fff42755
VM
298 return -1;
299 }
300
301 return 0;
302}
303
3ae5fa07 304static int load_pack_bitmap(struct bitmap_index *bitmap_git)
fff42755 305{
3ae5fa07 306 assert(bitmap_git->map && !bitmap_git->loaded);
fff42755 307
3ae5fa07
JT
308 bitmap_git->bitmaps = kh_init_sha1();
309 bitmap_git->ext_index.positions = kh_init_sha1_pos();
310 load_pack_revindex(bitmap_git->pack);
fff42755 311
3ae5fa07
JT
312 if (!(bitmap_git->commits = read_bitmap_1(bitmap_git)) ||
313 !(bitmap_git->trees = read_bitmap_1(bitmap_git)) ||
314 !(bitmap_git->blobs = read_bitmap_1(bitmap_git)) ||
315 !(bitmap_git->tags = read_bitmap_1(bitmap_git)))
fff42755
VM
316 goto failed;
317
3ae5fa07 318 if (load_bitmap_entries_v1(bitmap_git) < 0)
fff42755
VM
319 goto failed;
320
3ae5fa07 321 bitmap_git->loaded = 1;
fff42755
VM
322 return 0;
323
324failed:
3ae5fa07
JT
325 munmap(bitmap_git->map, bitmap_git->map_size);
326 bitmap_git->map = NULL;
327 bitmap_git->map_size = 0;
fff42755
VM
328 return -1;
329}
330
3ae5fa07 331static int open_pack_bitmap(struct bitmap_index *bitmap_git)
fff42755
VM
332{
333 struct packed_git *p;
334 int ret = -1;
335
3ae5fa07 336 assert(!bitmap_git->map && !bitmap_git->loaded);
fff42755 337
a80d72db 338 for (p = get_packed_git(the_repository); p; p = p->next) {
3ae5fa07 339 if (open_pack_bitmap_1(bitmap_git, p) == 0)
fff42755
VM
340 ret = 0;
341 }
342
343 return ret;
344}
345
3ae5fa07 346struct bitmap_index *prepare_bitmap_git(void)
fff42755 347{
3ae5fa07 348 struct bitmap_index *bitmap_git = xcalloc(1, sizeof(*bitmap_git));
fff42755 349
3ae5fa07
JT
350 if (!open_pack_bitmap(bitmap_git) && !load_pack_bitmap(bitmap_git))
351 return bitmap_git;
fff42755 352
f3c23db2 353 free_bitmap_index(bitmap_git);
3ae5fa07 354 return NULL;
fff42755
VM
355}
356
357struct include_data {
3ae5fa07 358 struct bitmap_index *bitmap_git;
fff42755
VM
359 struct bitmap *base;
360 struct bitmap *seen;
361};
362
3ae5fa07
JT
363static inline int bitmap_position_extended(struct bitmap_index *bitmap_git,
364 const unsigned char *sha1)
fff42755 365{
3ae5fa07 366 khash_sha1_pos *positions = bitmap_git->ext_index.positions;
fff42755
VM
367 khiter_t pos = kh_get_sha1_pos(positions, sha1);
368
369 if (pos < kh_end(positions)) {
370 int bitmap_pos = kh_value(positions, pos);
3ae5fa07 371 return bitmap_pos + bitmap_git->pack->num_objects;
fff42755
VM
372 }
373
374 return -1;
375}
376
3ae5fa07
JT
377static inline int bitmap_position_packfile(struct bitmap_index *bitmap_git,
378 const unsigned char *sha1)
fff42755 379{
3ae5fa07 380 off_t offset = find_pack_entry_one(sha1, bitmap_git->pack);
fff42755
VM
381 if (!offset)
382 return -1;
383
3ae5fa07 384 return find_revindex_position(bitmap_git->pack, offset);
fff42755
VM
385}
386
3ae5fa07
JT
387static int bitmap_position(struct bitmap_index *bitmap_git,
388 const unsigned char *sha1)
fff42755 389{
3ae5fa07
JT
390 int pos = bitmap_position_packfile(bitmap_git, sha1);
391 return (pos >= 0) ? pos : bitmap_position_extended(bitmap_git, sha1);
fff42755
VM
392}
393
3ae5fa07
JT
394static int ext_index_add_object(struct bitmap_index *bitmap_git,
395 struct object *object, const char *name)
fff42755 396{
3ae5fa07 397 struct eindex *eindex = &bitmap_git->ext_index;
fff42755
VM
398
399 khiter_t hash_pos;
400 int hash_ret;
401 int bitmap_pos;
402
ed1c9977 403 hash_pos = kh_put_sha1_pos(eindex->positions, object->oid.hash, &hash_ret);
fff42755
VM
404 if (hash_ret > 0) {
405 if (eindex->count >= eindex->alloc) {
406 eindex->alloc = (eindex->alloc + 16) * 3 / 2;
2756ca43
RS
407 REALLOC_ARRAY(eindex->objects, eindex->alloc);
408 REALLOC_ARRAY(eindex->hashes, eindex->alloc);
fff42755
VM
409 }
410
411 bitmap_pos = eindex->count;
412 eindex->objects[eindex->count] = object;
413 eindex->hashes[eindex->count] = pack_name_hash(name);
414 kh_value(eindex->positions, hash_pos) = bitmap_pos;
415 eindex->count++;
416 } else {
417 bitmap_pos = kh_value(eindex->positions, hash_pos);
418 }
419
3ae5fa07 420 return bitmap_pos + bitmap_git->pack->num_objects;
fff42755
VM
421}
422
3ae5fa07
JT
423struct bitmap_show_data {
424 struct bitmap_index *bitmap_git;
425 struct bitmap *base;
426};
427
428static void show_object(struct object *object, const char *name, void *data_)
fff42755 429{
3ae5fa07 430 struct bitmap_show_data *data = data_;
fff42755
VM
431 int bitmap_pos;
432
3ae5fa07 433 bitmap_pos = bitmap_position(data->bitmap_git, object->oid.hash);
fff42755 434
de1e67d0 435 if (bitmap_pos < 0)
3ae5fa07
JT
436 bitmap_pos = ext_index_add_object(data->bitmap_git, object,
437 name);
fff42755 438
3ae5fa07 439 bitmap_set(data->base, bitmap_pos);
fff42755
VM
440}
441
442static void show_commit(struct commit *commit, void *data)
443{
444}
445
3ae5fa07
JT
446static int add_to_include_set(struct bitmap_index *bitmap_git,
447 struct include_data *data,
fff42755
VM
448 const unsigned char *sha1,
449 int bitmap_pos)
450{
451 khiter_t hash_pos;
452
453 if (data->seen && bitmap_get(data->seen, bitmap_pos))
454 return 0;
455
456 if (bitmap_get(data->base, bitmap_pos))
457 return 0;
458
3ae5fa07
JT
459 hash_pos = kh_get_sha1(bitmap_git->bitmaps, sha1);
460 if (hash_pos < kh_end(bitmap_git->bitmaps)) {
461 struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, hash_pos);
fff42755
VM
462 bitmap_or_ewah(data->base, lookup_stored_bitmap(st));
463 return 0;
464 }
465
466 bitmap_set(data->base, bitmap_pos);
467 return 1;
468}
469
470static int should_include(struct commit *commit, void *_data)
471{
472 struct include_data *data = _data;
473 int bitmap_pos;
474
3ae5fa07 475 bitmap_pos = bitmap_position(data->bitmap_git, commit->object.oid.hash);
fff42755 476 if (bitmap_pos < 0)
3ae5fa07
JT
477 bitmap_pos = ext_index_add_object(data->bitmap_git,
478 (struct object *)commit,
479 NULL);
fff42755 480
3ae5fa07
JT
481 if (!add_to_include_set(data->bitmap_git, data, commit->object.oid.hash,
482 bitmap_pos)) {
fff42755
VM
483 struct commit_list *parent = commit->parents;
484
485 while (parent) {
486 parent->item->object.flags |= SEEN;
487 parent = parent->next;
488 }
489
490 return 0;
491 }
492
493 return 1;
494}
495
3ae5fa07
JT
496static struct bitmap *find_objects(struct bitmap_index *bitmap_git,
497 struct rev_info *revs,
fff42755
VM
498 struct object_list *roots,
499 struct bitmap *seen)
500{
501 struct bitmap *base = NULL;
502 int needs_walk = 0;
503
504 struct object_list *not_mapped = NULL;
505
506 /*
507 * Go through all the roots for the walk. The ones that have bitmaps
508 * on the bitmap index will be `or`ed together to form an initial
509 * global reachability analysis.
510 *
511 * The ones without bitmaps in the index will be stored in the
512 * `not_mapped_list` for further processing.
513 */
514 while (roots) {
515 struct object *object = roots->item;
516 roots = roots->next;
517
518 if (object->type == OBJ_COMMIT) {
3ae5fa07 519 khiter_t pos = kh_get_sha1(bitmap_git->bitmaps, object->oid.hash);
fff42755 520
3ae5fa07
JT
521 if (pos < kh_end(bitmap_git->bitmaps)) {
522 struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, pos);
fff42755
VM
523 struct ewah_bitmap *or_with = lookup_stored_bitmap(st);
524
525 if (base == NULL)
526 base = ewah_to_bitmap(or_with);
527 else
528 bitmap_or_ewah(base, or_with);
529
530 object->flags |= SEEN;
531 continue;
532 }
533 }
534
535 object_list_insert(object, &not_mapped);
536 }
537
538 /*
539 * Best case scenario: We found bitmaps for all the roots,
540 * so the resulting `or` bitmap has the full reachability analysis
541 */
542 if (not_mapped == NULL)
543 return base;
544
545 roots = not_mapped;
546
547 /*
548 * Let's iterate through all the roots that don't have bitmaps to
549 * check if we can determine them to be reachable from the existing
550 * global bitmap.
551 *
552 * If we cannot find them in the existing global bitmap, we'll need
553 * to push them to an actual walk and run it until we can confirm
554 * they are reachable
555 */
556 while (roots) {
557 struct object *object = roots->item;
558 int pos;
559
560 roots = roots->next;
3ae5fa07 561 pos = bitmap_position(bitmap_git, object->oid.hash);
fff42755
VM
562
563 if (pos < 0 || base == NULL || !bitmap_get(base, pos)) {
564 object->flags &= ~UNINTERESTING;
565 add_pending_object(revs, object, "");
566 needs_walk = 1;
567 } else {
568 object->flags |= SEEN;
569 }
570 }
571
572 if (needs_walk) {
573 struct include_data incdata;
3ae5fa07 574 struct bitmap_show_data show_data;
fff42755
VM
575
576 if (base == NULL)
577 base = bitmap_new();
578
3ae5fa07 579 incdata.bitmap_git = bitmap_git;
fff42755
VM
580 incdata.base = base;
581 incdata.seen = seen;
582
583 revs->include_check = should_include;
584 revs->include_check_data = &incdata;
585
586 if (prepare_revision_walk(revs))
587 die("revision walk setup failed");
588
3ae5fa07
JT
589 show_data.bitmap_git = bitmap_git;
590 show_data.base = base;
591
592 traverse_commit_list(revs, show_commit, show_object,
593 &show_data);
fff42755
VM
594 }
595
596 return base;
597}
598
3ae5fa07 599static void show_extended_objects(struct bitmap_index *bitmap_git,
fff42755
VM
600 show_reachable_fn show_reach)
601{
3ae5fa07
JT
602 struct bitmap *objects = bitmap_git->result;
603 struct eindex *eindex = &bitmap_git->ext_index;
fff42755
VM
604 uint32_t i;
605
606 for (i = 0; i < eindex->count; ++i) {
607 struct object *obj;
608
3ae5fa07 609 if (!bitmap_get(objects, bitmap_git->pack->num_objects + i))
fff42755
VM
610 continue;
611
612 obj = eindex->objects[i];
20664967 613 show_reach(&obj->oid, obj->type, 0, eindex->hashes[i], NULL, 0);
fff42755
VM
614 }
615}
616
617static void show_objects_for_type(
3ae5fa07 618 struct bitmap_index *bitmap_git,
fff42755
VM
619 struct ewah_bitmap *type_filter,
620 enum object_type object_type,
621 show_reachable_fn show_reach)
622{
623 size_t pos = 0, i = 0;
624 uint32_t offset;
625
626 struct ewah_iterator it;
627 eword_t filter;
628
3ae5fa07
JT
629 struct bitmap *objects = bitmap_git->result;
630
631 if (bitmap_git->reuse_objects == bitmap_git->pack->num_objects)
fff42755
VM
632 return;
633
634 ewah_iterator_init(&it, type_filter);
635
636 while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
637 eword_t word = objects->words[i] & filter;
638
34b935c0 639 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
20664967 640 struct object_id oid;
fff42755
VM
641 struct revindex_entry *entry;
642 uint32_t hash = 0;
643
644 if ((word >> offset) == 0)
645 break;
646
647 offset += ewah_bit_ctz64(word >> offset);
648
3ae5fa07 649 if (pos + offset < bitmap_git->reuse_objects)
fff42755
VM
650 continue;
651
3ae5fa07
JT
652 entry = &bitmap_git->pack->revindex[pos + offset];
653 nth_packed_object_oid(&oid, bitmap_git->pack, entry->nr);
fff42755 654
3ae5fa07
JT
655 if (bitmap_git->hashes)
656 hash = get_be32(bitmap_git->hashes + entry->nr);
ae4f07fb 657
3ae5fa07 658 show_reach(&oid, object_type, 0, hash, bitmap_git->pack, entry->offset);
fff42755
VM
659 }
660
34b935c0 661 pos += BITS_IN_EWORD;
fff42755
VM
662 i++;
663 }
664}
665
3ae5fa07
JT
666static int in_bitmapped_pack(struct bitmap_index *bitmap_git,
667 struct object_list *roots)
fff42755
VM
668{
669 while (roots) {
670 struct object *object = roots->item;
671 roots = roots->next;
672
3ae5fa07 673 if (find_pack_entry_one(object->oid.hash, bitmap_git->pack) > 0)
fff42755
VM
674 return 1;
675 }
676
677 return 0;
678}
679
3ae5fa07 680struct bitmap_index *prepare_bitmap_walk(struct rev_info *revs)
fff42755
VM
681{
682 unsigned int i;
fff42755
VM
683
684 struct object_list *wants = NULL;
685 struct object_list *haves = NULL;
686
687 struct bitmap *wants_bitmap = NULL;
688 struct bitmap *haves_bitmap = NULL;
689
3ae5fa07
JT
690 struct bitmap_index *bitmap_git = xcalloc(1, sizeof(*bitmap_git));
691 /* try to open a bitmapped pack, but don't parse it yet
692 * because we may not need to use it */
693 if (open_pack_bitmap(bitmap_git) < 0)
f3c23db2 694 goto cleanup;
fff42755 695
4d01a7fa
696 for (i = 0; i < revs->pending.nr; ++i) {
697 struct object *object = revs->pending.objects[i].item;
fff42755
VM
698
699 if (object->type == OBJ_NONE)
c251c83d 700 parse_object_or_die(&object->oid, NULL);
fff42755
VM
701
702 while (object->type == OBJ_TAG) {
703 struct tag *tag = (struct tag *) object;
704
705 if (object->flags & UNINTERESTING)
706 object_list_insert(object, &haves);
707 else
708 object_list_insert(object, &wants);
709
710 if (!tag->tagged)
711 die("bad tag");
c251c83d 712 object = parse_object_or_die(&tag->tagged->oid, NULL);
fff42755
VM
713 }
714
715 if (object->flags & UNINTERESTING)
716 object_list_insert(object, &haves);
717 else
718 object_list_insert(object, &wants);
719 }
720
721 /*
722 * if we have a HAVES list, but none of those haves is contained
723 * in the packfile that has a bitmap, we don't have anything to
724 * optimize here
725 */
3ae5fa07 726 if (haves && !in_bitmapped_pack(bitmap_git, haves))
f3c23db2 727 goto cleanup;
fff42755
VM
728
729 /* if we don't want anything, we're done here */
730 if (!wants)
f3c23db2 731 goto cleanup;
fff42755
VM
732
733 /*
734 * now we're going to use bitmaps, so load the actual bitmap entries
735 * from disk. this is the point of no return; after this the rev_list
736 * becomes invalidated and we must perform the revwalk through bitmaps
737 */
3ae5fa07 738 if (!bitmap_git->loaded && load_pack_bitmap(bitmap_git) < 0)
f3c23db2 739 goto cleanup;
fff42755 740
4d01a7fa 741 object_array_clear(&revs->pending);
fff42755
VM
742
743 if (haves) {
2db1a43f 744 revs->ignore_missing_links = 1;
3ae5fa07 745 haves_bitmap = find_objects(bitmap_git, revs, haves, NULL);
fff42755 746 reset_revision_walk();
2db1a43f 747 revs->ignore_missing_links = 0;
fff42755
VM
748
749 if (haves_bitmap == NULL)
033abf97 750 BUG("failed to perform bitmap walk");
fff42755
VM
751 }
752
3ae5fa07 753 wants_bitmap = find_objects(bitmap_git, revs, wants, haves_bitmap);
fff42755
VM
754
755 if (!wants_bitmap)
033abf97 756 BUG("failed to perform bitmap walk");
fff42755
VM
757
758 if (haves_bitmap)
759 bitmap_and_not(wants_bitmap, haves_bitmap);
760
3ae5fa07 761 bitmap_git->result = wants_bitmap;
fff42755
VM
762
763 bitmap_free(haves_bitmap);
3ae5fa07 764 return bitmap_git;
f3c23db2
JT
765
766cleanup:
767 free_bitmap_index(bitmap_git);
768 return NULL;
fff42755
VM
769}
770
3ae5fa07
JT
771int reuse_partial_packfile_from_bitmap(struct bitmap_index *bitmap_git,
772 struct packed_git **packfile,
fff42755
VM
773 uint32_t *entries,
774 off_t *up_to)
775{
776 /*
777 * Reuse the packfile content if we need more than
778 * 90% of its objects
779 */
780 static const double REUSE_PERCENT = 0.9;
781
3ae5fa07 782 struct bitmap *result = bitmap_git->result;
fff42755
VM
783 uint32_t reuse_threshold;
784 uint32_t i, reuse_objects = 0;
785
786 assert(result);
787
788 for (i = 0; i < result->word_alloc; ++i) {
789 if (result->words[i] != (eword_t)~0) {
790 reuse_objects += ewah_bit_ctz64(~result->words[i]);
791 break;
792 }
793
34b935c0 794 reuse_objects += BITS_IN_EWORD;
fff42755
VM
795 }
796
797#ifdef GIT_BITMAP_DEBUG
798 {
799 const unsigned char *sha1;
800 struct revindex_entry *entry;
801
3ae5fa07
JT
802 entry = &bitmap_git->reverse_index->revindex[reuse_objects];
803 sha1 = nth_packed_object_sha1(bitmap_git->pack, entry->nr);
fff42755
VM
804
805 fprintf(stderr, "Failed to reuse at %d (%016llx)\n",
806 reuse_objects, result->words[i]);
807 fprintf(stderr, " %s\n", sha1_to_hex(sha1));
808 }
809#endif
810
811 if (!reuse_objects)
812 return -1;
813
3ae5fa07
JT
814 if (reuse_objects >= bitmap_git->pack->num_objects) {
815 bitmap_git->reuse_objects = *entries = bitmap_git->pack->num_objects;
fff42755 816 *up_to = -1; /* reuse the full pack */
3ae5fa07 817 *packfile = bitmap_git->pack;
fff42755
VM
818 return 0;
819 }
820
3ae5fa07 821 reuse_threshold = bitmap_popcount(bitmap_git->result) * REUSE_PERCENT;
fff42755
VM
822
823 if (reuse_objects < reuse_threshold)
824 return -1;
825
3ae5fa07
JT
826 bitmap_git->reuse_objects = *entries = reuse_objects;
827 *up_to = bitmap_git->pack->revindex[reuse_objects].offset;
828 *packfile = bitmap_git->pack;
fff42755
VM
829
830 return 0;
831}
832
3ae5fa07
JT
833void traverse_bitmap_commit_list(struct bitmap_index *bitmap_git,
834 show_reachable_fn show_reachable)
fff42755 835{
3ae5fa07 836 assert(bitmap_git->result);
fff42755 837
3ae5fa07 838 show_objects_for_type(bitmap_git, bitmap_git->commits,
fff42755 839 OBJ_COMMIT, show_reachable);
3ae5fa07 840 show_objects_for_type(bitmap_git, bitmap_git->trees,
fff42755 841 OBJ_TREE, show_reachable);
3ae5fa07 842 show_objects_for_type(bitmap_git, bitmap_git->blobs,
fff42755 843 OBJ_BLOB, show_reachable);
3ae5fa07 844 show_objects_for_type(bitmap_git, bitmap_git->tags,
fff42755
VM
845 OBJ_TAG, show_reachable);
846
3ae5fa07 847 show_extended_objects(bitmap_git, show_reachable);
fff42755 848
3ae5fa07
JT
849 bitmap_free(bitmap_git->result);
850 bitmap_git->result = NULL;
fff42755
VM
851}
852
3ae5fa07 853static uint32_t count_object_type(struct bitmap_index *bitmap_git,
fff42755
VM
854 enum object_type type)
855{
3ae5fa07
JT
856 struct bitmap *objects = bitmap_git->result;
857 struct eindex *eindex = &bitmap_git->ext_index;
fff42755
VM
858
859 uint32_t i = 0, count = 0;
860 struct ewah_iterator it;
861 eword_t filter;
862
863 switch (type) {
864 case OBJ_COMMIT:
3ae5fa07 865 ewah_iterator_init(&it, bitmap_git->commits);
fff42755
VM
866 break;
867
868 case OBJ_TREE:
3ae5fa07 869 ewah_iterator_init(&it, bitmap_git->trees);
fff42755
VM
870 break;
871
872 case OBJ_BLOB:
3ae5fa07 873 ewah_iterator_init(&it, bitmap_git->blobs);
fff42755
VM
874 break;
875
876 case OBJ_TAG:
3ae5fa07 877 ewah_iterator_init(&it, bitmap_git->tags);
fff42755
VM
878 break;
879
880 default:
881 return 0;
882 }
883
884 while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
885 eword_t word = objects->words[i++] & filter;
886 count += ewah_bit_popcount64(word);
887 }
888
889 for (i = 0; i < eindex->count; ++i) {
890 if (eindex->objects[i]->type == type &&
3ae5fa07 891 bitmap_get(objects, bitmap_git->pack->num_objects + i))
fff42755
VM
892 count++;
893 }
894
895 return count;
896}
897
3ae5fa07
JT
898void count_bitmap_commit_list(struct bitmap_index *bitmap_git,
899 uint32_t *commits, uint32_t *trees,
fff42755
VM
900 uint32_t *blobs, uint32_t *tags)
901{
3ae5fa07 902 assert(bitmap_git->result);
fff42755
VM
903
904 if (commits)
3ae5fa07 905 *commits = count_object_type(bitmap_git, OBJ_COMMIT);
fff42755
VM
906
907 if (trees)
3ae5fa07 908 *trees = count_object_type(bitmap_git, OBJ_TREE);
fff42755
VM
909
910 if (blobs)
3ae5fa07 911 *blobs = count_object_type(bitmap_git, OBJ_BLOB);
fff42755
VM
912
913 if (tags)
3ae5fa07 914 *tags = count_object_type(bitmap_git, OBJ_TAG);
fff42755
VM
915}
916
917struct bitmap_test_data {
3ae5fa07 918 struct bitmap_index *bitmap_git;
fff42755
VM
919 struct bitmap *base;
920 struct progress *prg;
921 size_t seen;
922};
923
de1e67d0
JK
924static void test_show_object(struct object *object, const char *name,
925 void *data)
fff42755
VM
926{
927 struct bitmap_test_data *tdata = data;
928 int bitmap_pos;
929
3ae5fa07 930 bitmap_pos = bitmap_position(tdata->bitmap_git, object->oid.hash);
fff42755 931 if (bitmap_pos < 0)
f2fd0760 932 die("Object not in bitmap: %s\n", oid_to_hex(&object->oid));
fff42755
VM
933
934 bitmap_set(tdata->base, bitmap_pos);
935 display_progress(tdata->prg, ++tdata->seen);
936}
937
938static void test_show_commit(struct commit *commit, void *data)
939{
940 struct bitmap_test_data *tdata = data;
941 int bitmap_pos;
942
3ae5fa07
JT
943 bitmap_pos = bitmap_position(tdata->bitmap_git,
944 commit->object.oid.hash);
fff42755 945 if (bitmap_pos < 0)
f2fd0760 946 die("Object not in bitmap: %s\n", oid_to_hex(&commit->object.oid));
fff42755
VM
947
948 bitmap_set(tdata->base, bitmap_pos);
949 display_progress(tdata->prg, ++tdata->seen);
950}
951
952void test_bitmap_walk(struct rev_info *revs)
953{
954 struct object *root;
955 struct bitmap *result = NULL;
956 khiter_t pos;
957 size_t result_popcnt;
958 struct bitmap_test_data tdata;
3ae5fa07 959 struct bitmap_index *bitmap_git;
fff42755 960
3ae5fa07 961 if (!(bitmap_git = prepare_bitmap_git()))
fff42755
VM
962 die("failed to load bitmap indexes");
963
964 if (revs->pending.nr != 1)
965 die("you must specify exactly one commit to test");
966
967 fprintf(stderr, "Bitmap v%d test (%d entries loaded)\n",
3ae5fa07 968 bitmap_git->version, bitmap_git->entry_count);
fff42755
VM
969
970 root = revs->pending.objects[0].item;
3ae5fa07 971 pos = kh_get_sha1(bitmap_git->bitmaps, root->oid.hash);
fff42755 972
3ae5fa07
JT
973 if (pos < kh_end(bitmap_git->bitmaps)) {
974 struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, pos);
fff42755
VM
975 struct ewah_bitmap *bm = lookup_stored_bitmap(st);
976
977 fprintf(stderr, "Found bitmap for %s. %d bits / %08x checksum\n",
f2fd0760 978 oid_to_hex(&root->oid), (int)bm->bit_size, ewah_checksum(bm));
fff42755
VM
979
980 result = ewah_to_bitmap(bm);
981 }
982
983 if (result == NULL)
f2fd0760 984 die("Commit %s doesn't have an indexed bitmap", oid_to_hex(&root->oid));
fff42755
VM
985
986 revs->tag_objects = 1;
987 revs->tree_objects = 1;
988 revs->blob_objects = 1;
989
990 result_popcnt = bitmap_popcount(result);
991
992 if (prepare_revision_walk(revs))
993 die("revision walk setup failed");
994
3ae5fa07 995 tdata.bitmap_git = bitmap_git;
fff42755
VM
996 tdata.base = bitmap_new();
997 tdata.prg = start_progress("Verifying bitmap entries", result_popcnt);
998 tdata.seen = 0;
999
1000 traverse_commit_list(revs, &test_show_commit, &test_show_object, &tdata);
1001
1002 stop_progress(&tdata.prg);
1003
1004 if (bitmap_equals(result, tdata.base))
1005 fprintf(stderr, "OK!\n");
1006 else
1007 fprintf(stderr, "Mismatch!\n");
f86a3747 1008
f3c23db2 1009 free_bitmap_index(bitmap_git);
fff42755 1010}
7cc8f971
VM
1011
1012static int rebuild_bitmap(uint32_t *reposition,
1013 struct ewah_bitmap *source,
1014 struct bitmap *dest)
1015{
1016 uint32_t pos = 0;
1017 struct ewah_iterator it;
1018 eword_t word;
1019
1020 ewah_iterator_init(&it, source);
1021
1022 while (ewah_iterator_next(&word, &it)) {
1023 uint32_t offset, bit_pos;
1024
34b935c0 1025 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
7cc8f971
VM
1026 if ((word >> offset) == 0)
1027 break;
1028
1029 offset += ewah_bit_ctz64(word >> offset);
1030
1031 bit_pos = reposition[pos + offset];
1032 if (bit_pos > 0)
1033 bitmap_set(dest, bit_pos - 1);
1034 else /* can't reuse, we don't have the object */
1035 return -1;
1036 }
1037
34b935c0 1038 pos += BITS_IN_EWORD;
7cc8f971
VM
1039 }
1040 return 0;
1041}
1042
3ae5fa07
JT
1043int rebuild_existing_bitmaps(struct bitmap_index *bitmap_git,
1044 struct packing_data *mapping,
7cc8f971
VM
1045 khash_sha1 *reused_bitmaps,
1046 int show_progress)
1047{
1048 uint32_t i, num_objects;
1049 uint32_t *reposition;
1050 struct bitmap *rebuild;
1051 struct stored_bitmap *stored;
1052 struct progress *progress = NULL;
1053
1054 khiter_t hash_pos;
1055 int hash_ret;
1056
3ae5fa07 1057 num_objects = bitmap_git->pack->num_objects;
7cc8f971
VM
1058 reposition = xcalloc(num_objects, sizeof(uint32_t));
1059
1060 for (i = 0; i < num_objects; ++i) {
1061 const unsigned char *sha1;
1062 struct revindex_entry *entry;
1063 struct object_entry *oe;
1064
3ae5fa07
JT
1065 entry = &bitmap_git->pack->revindex[i];
1066 sha1 = nth_packed_object_sha1(bitmap_git->pack, entry->nr);
7cc8f971
VM
1067 oe = packlist_find(mapping, sha1, NULL);
1068
1069 if (oe)
06af3bba 1070 reposition[i] = oe_in_pack_pos(mapping, oe) + 1;
7cc8f971
VM
1071 }
1072
1073 rebuild = bitmap_new();
1074 i = 0;
1075
1076 if (show_progress)
1077 progress = start_progress("Reusing bitmaps", 0);
1078
3ae5fa07 1079 kh_foreach_value(bitmap_git->bitmaps, stored, {
7cc8f971
VM
1080 if (stored->flags & BITMAP_FLAG_REUSE) {
1081 if (!rebuild_bitmap(reposition,
1082 lookup_stored_bitmap(stored),
1083 rebuild)) {
1084 hash_pos = kh_put_sha1(reused_bitmaps,
1085 stored->sha1,
1086 &hash_ret);
1087 kh_value(reused_bitmaps, hash_pos) =
1088 bitmap_to_ewah(rebuild);
1089 }
1090 bitmap_reset(rebuild);
1091 display_progress(progress, ++i);
1092 }
1093 });
1094
1095 stop_progress(&progress);
1096
1097 free(reposition);
1098 bitmap_free(rebuild);
1099 return 0;
1100}
f3c23db2
JT
1101
1102void free_bitmap_index(struct bitmap_index *b)
1103{
1104 if (!b)
1105 return;
1106
1107 if (b->map)
1108 munmap(b->map, b->map_size);
1109 ewah_pool_free(b->commits);
1110 ewah_pool_free(b->trees);
1111 ewah_pool_free(b->blobs);
1112 ewah_pool_free(b->tags);
1113 kh_destroy_sha1(b->bitmaps);
1114 free(b->ext_index.objects);
1115 free(b->ext_index.hashes);
1116 bitmap_free(b->result);
1117 free(b);
1118}