pack-bitmap: convert struct stored_bitmap to object_id
[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 {
53636539 21 struct object_id oid;
fff42755
VM
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
30cdc33f
JK
89 /* "have" bitmap from the last performed walk */
90 struct bitmap *haves;
91
fff42755
VM
92 /* Version of the bitmap index */
93 unsigned int version;
3ae5fa07 94};
fff42755
VM
95
96static struct ewah_bitmap *lookup_stored_bitmap(struct stored_bitmap *st)
97{
98 struct ewah_bitmap *parent;
99 struct ewah_bitmap *composed;
100
101 if (st->xor == NULL)
102 return st->root;
103
104 composed = ewah_pool_new();
105 parent = lookup_stored_bitmap(st->xor);
106 ewah_xor(st->root, parent, composed);
107
108 ewah_pool_free(st->root);
109 st->root = composed;
110 st->xor = NULL;
111
112 return composed;
113}
114
115/*
116 * Read a bitmap from the current read position on the mmaped
117 * index, and increase the read position accordingly
118 */
119static struct ewah_bitmap *read_bitmap_1(struct bitmap_index *index)
120{
121 struct ewah_bitmap *b = ewah_pool_new();
122
1140bf01 123 ssize_t bitmap_size = ewah_read_mmap(b,
fff42755
VM
124 index->map + index->map_pos,
125 index->map_size - index->map_pos);
126
127 if (bitmap_size < 0) {
128 error("Failed to load bitmap index (corrupted?)");
129 ewah_pool_free(b);
130 return NULL;
131 }
132
133 index->map_pos += bitmap_size;
134 return b;
135}
136
137static int load_bitmap_header(struct bitmap_index *index)
138{
139 struct bitmap_disk_header *header = (void *)index->map;
140
141 if (index->map_size < sizeof(*header) + 20)
142 return error("Corrupted bitmap index (missing header data)");
143
144 if (memcmp(header->magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE)) != 0)
145 return error("Corrupted bitmap index file (wrong header)");
146
147 index->version = ntohs(header->version);
148 if (index->version != 1)
149 return error("Unsupported version for bitmap index file (%d)", index->version);
150
151 /* Parse known bitmap format options */
152 {
153 uint32_t flags = ntohs(header->options);
154
155 if ((flags & BITMAP_OPT_FULL_DAG) == 0)
156 return error("Unsupported options for bitmap index file "
157 "(Git requires BITMAP_OPT_FULL_DAG)");
ae4f07fb
VM
158
159 if (flags & BITMAP_OPT_HASH_CACHE) {
160 unsigned char *end = index->map + index->map_size - 20;
161 index->hashes = ((uint32_t *)end) - index->pack->num_objects;
162 }
fff42755
VM
163 }
164
165 index->entry_count = ntohl(header->entry_count);
0f4d6cad 166 index->map_pos += sizeof(*header) - GIT_MAX_RAWSZ + the_hash_algo->rawsz;
fff42755
VM
167 return 0;
168}
169
170static struct stored_bitmap *store_bitmap(struct bitmap_index *index,
171 struct ewah_bitmap *root,
172 const unsigned char *sha1,
173 struct stored_bitmap *xor_with,
174 int flags)
175{
176 struct stored_bitmap *stored;
177 khiter_t hash_pos;
178 int ret;
179
180 stored = xmalloc(sizeof(struct stored_bitmap));
181 stored->root = root;
182 stored->xor = xor_with;
183 stored->flags = flags;
53636539 184 oidread(&stored->oid, sha1);
fff42755 185
53636539 186 hash_pos = kh_put_sha1(index->bitmaps, stored->oid.hash, &ret);
fff42755
VM
187
188 /* a 0 return code means the insertion succeeded with no changes,
189 * because the SHA1 already existed on the map. this is bad, there
190 * shouldn't be duplicated commits in the index */
191 if (ret == 0) {
192 error("Duplicate entry in bitmap index: %s", sha1_to_hex(sha1));
193 return NULL;
194 }
195
196 kh_value(index->bitmaps, hash_pos) = stored;
197 return stored;
198}
199
b5007211
KB
200static inline uint32_t read_be32(const unsigned char *buffer, size_t *pos)
201{
202 uint32_t result = get_be32(buffer + *pos);
203 (*pos) += sizeof(result);
204 return result;
205}
206
207static inline uint8_t read_u8(const unsigned char *buffer, size_t *pos)
208{
209 return buffer[(*pos)++];
210}
211
599dc766
RS
212#define MAX_XOR_OFFSET 160
213
fff42755
VM
214static int load_bitmap_entries_v1(struct bitmap_index *index)
215{
fff42755 216 uint32_t i;
599dc766 217 struct stored_bitmap *recent_bitmaps[MAX_XOR_OFFSET] = { NULL };
fff42755
VM
218
219 for (i = 0; i < index->entry_count; ++i) {
220 int xor_offset, flags;
221 struct ewah_bitmap *bitmap = NULL;
222 struct stored_bitmap *xor_bitmap = NULL;
223 uint32_t commit_idx_pos;
224 const unsigned char *sha1;
225
b5007211
KB
226 commit_idx_pos = read_be32(index->map, &index->map_pos);
227 xor_offset = read_u8(index->map, &index->map_pos);
228 flags = read_u8(index->map, &index->map_pos);
fff42755 229
fff42755
VM
230 sha1 = nth_packed_object_sha1(index->pack, commit_idx_pos);
231
fff42755
VM
232 bitmap = read_bitmap_1(index);
233 if (!bitmap)
234 return -1;
235
236 if (xor_offset > MAX_XOR_OFFSET || xor_offset > i)
237 return error("Corrupted bitmap pack index");
238
239 if (xor_offset > 0) {
240 xor_bitmap = recent_bitmaps[(i - xor_offset) % MAX_XOR_OFFSET];
241
242 if (xor_bitmap == NULL)
243 return error("Invalid XOR offset in bitmap pack index");
244 }
245
246 recent_bitmaps[i % MAX_XOR_OFFSET] = store_bitmap(
247 index, bitmap, sha1, xor_bitmap, flags);
248 }
249
250 return 0;
251}
252
cb468050
JH
253static char *pack_bitmap_filename(struct packed_git *p)
254{
9ae97018 255 size_t len;
cb468050 256
9ae97018 257 if (!strip_suffix(p->pack_name, ".pack", &len))
033abf97 258 BUG("pack_name does not end in .pack");
9ae97018 259 return xstrfmt("%.*s.bitmap", (int)len, p->pack_name);
cb468050
JH
260}
261
3ae5fa07 262static int open_pack_bitmap_1(struct bitmap_index *bitmap_git, struct packed_git *packfile)
fff42755
VM
263{
264 int fd;
265 struct stat st;
266 char *idx_name;
267
268 if (open_pack_index(packfile))
269 return -1;
270
271 idx_name = pack_bitmap_filename(packfile);
a5436b57 272 fd = git_open(idx_name);
fff42755
VM
273 free(idx_name);
274
275 if (fd < 0)
276 return -1;
277
278 if (fstat(fd, &st)) {
279 close(fd);
280 return -1;
281 }
282
3ae5fa07 283 if (bitmap_git->pack) {
fff42755
VM
284 warning("ignoring extra bitmap file: %s", packfile->pack_name);
285 close(fd);
286 return -1;
287 }
288
3ae5fa07
JT
289 bitmap_git->pack = packfile;
290 bitmap_git->map_size = xsize_t(st.st_size);
291 bitmap_git->map = xmmap(NULL, bitmap_git->map_size, PROT_READ, MAP_PRIVATE, fd, 0);
292 bitmap_git->map_pos = 0;
fff42755
VM
293 close(fd);
294
3ae5fa07
JT
295 if (load_bitmap_header(bitmap_git) < 0) {
296 munmap(bitmap_git->map, bitmap_git->map_size);
297 bitmap_git->map = NULL;
298 bitmap_git->map_size = 0;
fff42755
VM
299 return -1;
300 }
301
302 return 0;
303}
304
3ae5fa07 305static int load_pack_bitmap(struct bitmap_index *bitmap_git)
fff42755 306{
199c86be 307 assert(bitmap_git->map);
fff42755 308
3ae5fa07
JT
309 bitmap_git->bitmaps = kh_init_sha1();
310 bitmap_git->ext_index.positions = kh_init_sha1_pos();
311 load_pack_revindex(bitmap_git->pack);
fff42755 312
3ae5fa07
JT
313 if (!(bitmap_git->commits = read_bitmap_1(bitmap_git)) ||
314 !(bitmap_git->trees = read_bitmap_1(bitmap_git)) ||
315 !(bitmap_git->blobs = read_bitmap_1(bitmap_git)) ||
316 !(bitmap_git->tags = read_bitmap_1(bitmap_git)))
fff42755
VM
317 goto failed;
318
3ae5fa07 319 if (load_bitmap_entries_v1(bitmap_git) < 0)
fff42755
VM
320 goto failed;
321
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
7c141127
NTND
331static int open_pack_bitmap(struct repository *r,
332 struct bitmap_index *bitmap_git)
fff42755
VM
333{
334 struct packed_git *p;
335 int ret = -1;
336
199c86be 337 assert(!bitmap_git->map);
fff42755 338
7c141127 339 for (p = get_all_packs(r); p; p = p->next) {
3ae5fa07 340 if (open_pack_bitmap_1(bitmap_git, p) == 0)
fff42755
VM
341 ret = 0;
342 }
343
344 return ret;
345}
346
7c141127 347struct bitmap_index *prepare_bitmap_git(struct repository *r)
fff42755 348{
3ae5fa07 349 struct bitmap_index *bitmap_git = xcalloc(1, sizeof(*bitmap_git));
fff42755 350
7c141127 351 if (!open_pack_bitmap(r, bitmap_git) && !load_pack_bitmap(bitmap_git))
3ae5fa07 352 return bitmap_git;
fff42755 353
f3c23db2 354 free_bitmap_index(bitmap_git);
3ae5fa07 355 return NULL;
fff42755
VM
356}
357
358struct include_data {
3ae5fa07 359 struct bitmap_index *bitmap_git;
fff42755
VM
360 struct bitmap *base;
361 struct bitmap *seen;
362};
363
3ae5fa07
JT
364static inline int bitmap_position_extended(struct bitmap_index *bitmap_git,
365 const unsigned char *sha1)
fff42755 366{
3ae5fa07 367 khash_sha1_pos *positions = bitmap_git->ext_index.positions;
fff42755
VM
368 khiter_t pos = kh_get_sha1_pos(positions, sha1);
369
370 if (pos < kh_end(positions)) {
371 int bitmap_pos = kh_value(positions, pos);
3ae5fa07 372 return bitmap_pos + bitmap_git->pack->num_objects;
fff42755
VM
373 }
374
375 return -1;
376}
377
3ae5fa07
JT
378static inline int bitmap_position_packfile(struct bitmap_index *bitmap_git,
379 const unsigned char *sha1)
fff42755 380{
3ae5fa07 381 off_t offset = find_pack_entry_one(sha1, bitmap_git->pack);
fff42755
VM
382 if (!offset)
383 return -1;
384
3ae5fa07 385 return find_revindex_position(bitmap_git->pack, offset);
fff42755
VM
386}
387
3ae5fa07
JT
388static int bitmap_position(struct bitmap_index *bitmap_git,
389 const unsigned char *sha1)
fff42755 390{
3ae5fa07
JT
391 int pos = bitmap_position_packfile(bitmap_git, sha1);
392 return (pos >= 0) ? pos : bitmap_position_extended(bitmap_git, sha1);
fff42755
VM
393}
394
3ae5fa07
JT
395static int ext_index_add_object(struct bitmap_index *bitmap_git,
396 struct object *object, const char *name)
fff42755 397{
3ae5fa07 398 struct eindex *eindex = &bitmap_git->ext_index;
fff42755
VM
399
400 khiter_t hash_pos;
401 int hash_ret;
402 int bitmap_pos;
403
ed1c9977 404 hash_pos = kh_put_sha1_pos(eindex->positions, object->oid.hash, &hash_ret);
fff42755
VM
405 if (hash_ret > 0) {
406 if (eindex->count >= eindex->alloc) {
407 eindex->alloc = (eindex->alloc + 16) * 3 / 2;
2756ca43
RS
408 REALLOC_ARRAY(eindex->objects, eindex->alloc);
409 REALLOC_ARRAY(eindex->hashes, eindex->alloc);
fff42755
VM
410 }
411
412 bitmap_pos = eindex->count;
413 eindex->objects[eindex->count] = object;
414 eindex->hashes[eindex->count] = pack_name_hash(name);
415 kh_value(eindex->positions, hash_pos) = bitmap_pos;
416 eindex->count++;
417 } else {
418 bitmap_pos = kh_value(eindex->positions, hash_pos);
419 }
420
3ae5fa07 421 return bitmap_pos + bitmap_git->pack->num_objects;
fff42755
VM
422}
423
3ae5fa07
JT
424struct bitmap_show_data {
425 struct bitmap_index *bitmap_git;
426 struct bitmap *base;
427};
428
429static void show_object(struct object *object, const char *name, void *data_)
fff42755 430{
3ae5fa07 431 struct bitmap_show_data *data = data_;
fff42755
VM
432 int bitmap_pos;
433
3ae5fa07 434 bitmap_pos = bitmap_position(data->bitmap_git, object->oid.hash);
fff42755 435
de1e67d0 436 if (bitmap_pos < 0)
3ae5fa07
JT
437 bitmap_pos = ext_index_add_object(data->bitmap_git, object,
438 name);
fff42755 439
3ae5fa07 440 bitmap_set(data->base, bitmap_pos);
fff42755
VM
441}
442
443static void show_commit(struct commit *commit, void *data)
444{
445}
446
3ae5fa07
JT
447static int add_to_include_set(struct bitmap_index *bitmap_git,
448 struct include_data *data,
fff42755
VM
449 const unsigned char *sha1,
450 int bitmap_pos)
451{
452 khiter_t hash_pos;
453
454 if (data->seen && bitmap_get(data->seen, bitmap_pos))
455 return 0;
456
457 if (bitmap_get(data->base, bitmap_pos))
458 return 0;
459
3ae5fa07
JT
460 hash_pos = kh_get_sha1(bitmap_git->bitmaps, sha1);
461 if (hash_pos < kh_end(bitmap_git->bitmaps)) {
462 struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, hash_pos);
fff42755
VM
463 bitmap_or_ewah(data->base, lookup_stored_bitmap(st));
464 return 0;
465 }
466
467 bitmap_set(data->base, bitmap_pos);
468 return 1;
469}
470
471static int should_include(struct commit *commit, void *_data)
472{
473 struct include_data *data = _data;
474 int bitmap_pos;
475
3ae5fa07 476 bitmap_pos = bitmap_position(data->bitmap_git, commit->object.oid.hash);
fff42755 477 if (bitmap_pos < 0)
3ae5fa07
JT
478 bitmap_pos = ext_index_add_object(data->bitmap_git,
479 (struct object *)commit,
480 NULL);
fff42755 481
3ae5fa07
JT
482 if (!add_to_include_set(data->bitmap_git, data, commit->object.oid.hash,
483 bitmap_pos)) {
fff42755
VM
484 struct commit_list *parent = commit->parents;
485
486 while (parent) {
487 parent->item->object.flags |= SEEN;
488 parent = parent->next;
489 }
490
491 return 0;
492 }
493
494 return 1;
495}
496
3ae5fa07
JT
497static struct bitmap *find_objects(struct bitmap_index *bitmap_git,
498 struct rev_info *revs,
fff42755
VM
499 struct object_list *roots,
500 struct bitmap *seen)
501{
502 struct bitmap *base = NULL;
503 int needs_walk = 0;
504
505 struct object_list *not_mapped = NULL;
506
507 /*
508 * Go through all the roots for the walk. The ones that have bitmaps
509 * on the bitmap index will be `or`ed together to form an initial
510 * global reachability analysis.
511 *
512 * The ones without bitmaps in the index will be stored in the
513 * `not_mapped_list` for further processing.
514 */
515 while (roots) {
516 struct object *object = roots->item;
517 roots = roots->next;
518
519 if (object->type == OBJ_COMMIT) {
3ae5fa07 520 khiter_t pos = kh_get_sha1(bitmap_git->bitmaps, object->oid.hash);
fff42755 521
3ae5fa07
JT
522 if (pos < kh_end(bitmap_git->bitmaps)) {
523 struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, pos);
fff42755
VM
524 struct ewah_bitmap *or_with = lookup_stored_bitmap(st);
525
526 if (base == NULL)
527 base = ewah_to_bitmap(or_with);
528 else
529 bitmap_or_ewah(base, or_with);
530
531 object->flags |= SEEN;
532 continue;
533 }
534 }
535
536 object_list_insert(object, &not_mapped);
537 }
538
539 /*
540 * Best case scenario: We found bitmaps for all the roots,
541 * so the resulting `or` bitmap has the full reachability analysis
542 */
543 if (not_mapped == NULL)
544 return base;
545
546 roots = not_mapped;
547
548 /*
549 * Let's iterate through all the roots that don't have bitmaps to
550 * check if we can determine them to be reachable from the existing
551 * global bitmap.
552 *
553 * If we cannot find them in the existing global bitmap, we'll need
554 * to push them to an actual walk and run it until we can confirm
555 * they are reachable
556 */
557 while (roots) {
558 struct object *object = roots->item;
559 int pos;
560
561 roots = roots->next;
3ae5fa07 562 pos = bitmap_position(bitmap_git, object->oid.hash);
fff42755
VM
563
564 if (pos < 0 || base == NULL || !bitmap_get(base, pos)) {
565 object->flags &= ~UNINTERESTING;
566 add_pending_object(revs, object, "");
567 needs_walk = 1;
568 } else {
569 object->flags |= SEEN;
570 }
571 }
572
573 if (needs_walk) {
574 struct include_data incdata;
3ae5fa07 575 struct bitmap_show_data show_data;
fff42755
VM
576
577 if (base == NULL)
578 base = bitmap_new();
579
3ae5fa07 580 incdata.bitmap_git = bitmap_git;
fff42755
VM
581 incdata.base = base;
582 incdata.seen = seen;
583
584 revs->include_check = should_include;
585 revs->include_check_data = &incdata;
586
587 if (prepare_revision_walk(revs))
588 die("revision walk setup failed");
589
3ae5fa07
JT
590 show_data.bitmap_git = bitmap_git;
591 show_data.base = base;
592
593 traverse_commit_list(revs, show_commit, show_object,
594 &show_data);
fff42755
VM
595 }
596
597 return base;
598}
599
3ae5fa07 600static void show_extended_objects(struct bitmap_index *bitmap_git,
fff42755
VM
601 show_reachable_fn show_reach)
602{
3ae5fa07
JT
603 struct bitmap *objects = bitmap_git->result;
604 struct eindex *eindex = &bitmap_git->ext_index;
fff42755
VM
605 uint32_t i;
606
607 for (i = 0; i < eindex->count; ++i) {
608 struct object *obj;
609
3ae5fa07 610 if (!bitmap_get(objects, bitmap_git->pack->num_objects + i))
fff42755
VM
611 continue;
612
613 obj = eindex->objects[i];
20664967 614 show_reach(&obj->oid, obj->type, 0, eindex->hashes[i], NULL, 0);
fff42755
VM
615 }
616}
617
618static void show_objects_for_type(
3ae5fa07 619 struct bitmap_index *bitmap_git,
fff42755
VM
620 struct ewah_bitmap *type_filter,
621 enum object_type object_type,
622 show_reachable_fn show_reach)
623{
624 size_t pos = 0, i = 0;
625 uint32_t offset;
626
627 struct ewah_iterator it;
628 eword_t filter;
629
3ae5fa07
JT
630 struct bitmap *objects = bitmap_git->result;
631
632 if (bitmap_git->reuse_objects == bitmap_git->pack->num_objects)
fff42755
VM
633 return;
634
635 ewah_iterator_init(&it, type_filter);
636
637 while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
638 eword_t word = objects->words[i] & filter;
639
34b935c0 640 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
20664967 641 struct object_id oid;
fff42755
VM
642 struct revindex_entry *entry;
643 uint32_t hash = 0;
644
645 if ((word >> offset) == 0)
646 break;
647
648 offset += ewah_bit_ctz64(word >> offset);
649
3ae5fa07 650 if (pos + offset < bitmap_git->reuse_objects)
fff42755
VM
651 continue;
652
3ae5fa07
JT
653 entry = &bitmap_git->pack->revindex[pos + offset];
654 nth_packed_object_oid(&oid, bitmap_git->pack, entry->nr);
fff42755 655
3ae5fa07
JT
656 if (bitmap_git->hashes)
657 hash = get_be32(bitmap_git->hashes + entry->nr);
ae4f07fb 658
3ae5fa07 659 show_reach(&oid, object_type, 0, hash, bitmap_git->pack, entry->offset);
fff42755
VM
660 }
661
34b935c0 662 pos += BITS_IN_EWORD;
fff42755
VM
663 i++;
664 }
665}
666
3ae5fa07
JT
667static int in_bitmapped_pack(struct bitmap_index *bitmap_git,
668 struct object_list *roots)
fff42755
VM
669{
670 while (roots) {
671 struct object *object = roots->item;
672 roots = roots->next;
673
3ae5fa07 674 if (find_pack_entry_one(object->oid.hash, bitmap_git->pack) > 0)
fff42755
VM
675 return 1;
676 }
677
678 return 0;
679}
680
3ae5fa07 681struct bitmap_index *prepare_bitmap_walk(struct rev_info *revs)
fff42755
VM
682{
683 unsigned int i;
fff42755
VM
684
685 struct object_list *wants = NULL;
686 struct object_list *haves = NULL;
687
688 struct bitmap *wants_bitmap = NULL;
689 struct bitmap *haves_bitmap = NULL;
690
3ae5fa07
JT
691 struct bitmap_index *bitmap_git = xcalloc(1, sizeof(*bitmap_git));
692 /* try to open a bitmapped pack, but don't parse it yet
693 * because we may not need to use it */
7c141127 694 if (open_pack_bitmap(revs->repo, bitmap_git) < 0)
f3c23db2 695 goto cleanup;
fff42755 696
4d01a7fa
697 for (i = 0; i < revs->pending.nr; ++i) {
698 struct object *object = revs->pending.objects[i].item;
fff42755
VM
699
700 if (object->type == OBJ_NONE)
c251c83d 701 parse_object_or_die(&object->oid, NULL);
fff42755
VM
702
703 while (object->type == OBJ_TAG) {
704 struct tag *tag = (struct tag *) object;
705
706 if (object->flags & UNINTERESTING)
707 object_list_insert(object, &haves);
708 else
709 object_list_insert(object, &wants);
710
711 if (!tag->tagged)
712 die("bad tag");
c251c83d 713 object = parse_object_or_die(&tag->tagged->oid, NULL);
fff42755
VM
714 }
715
716 if (object->flags & UNINTERESTING)
717 object_list_insert(object, &haves);
718 else
719 object_list_insert(object, &wants);
720 }
721
722 /*
723 * if we have a HAVES list, but none of those haves is contained
724 * in the packfile that has a bitmap, we don't have anything to
725 * optimize here
726 */
3ae5fa07 727 if (haves && !in_bitmapped_pack(bitmap_git, haves))
f3c23db2 728 goto cleanup;
fff42755
VM
729
730 /* if we don't want anything, we're done here */
731 if (!wants)
f3c23db2 732 goto cleanup;
fff42755
VM
733
734 /*
735 * now we're going to use bitmaps, so load the actual bitmap entries
736 * from disk. this is the point of no return; after this the rev_list
737 * becomes invalidated and we must perform the revwalk through bitmaps
738 */
199c86be 739 if (load_pack_bitmap(bitmap_git) < 0)
f3c23db2 740 goto cleanup;
fff42755 741
4d01a7fa 742 object_array_clear(&revs->pending);
fff42755
VM
743
744 if (haves) {
2db1a43f 745 revs->ignore_missing_links = 1;
3ae5fa07 746 haves_bitmap = find_objects(bitmap_git, revs, haves, NULL);
fff42755 747 reset_revision_walk();
2db1a43f 748 revs->ignore_missing_links = 0;
fff42755
VM
749
750 if (haves_bitmap == NULL)
033abf97 751 BUG("failed to perform bitmap walk");
fff42755
VM
752 }
753
3ae5fa07 754 wants_bitmap = find_objects(bitmap_git, revs, wants, haves_bitmap);
fff42755
VM
755
756 if (!wants_bitmap)
033abf97 757 BUG("failed to perform bitmap walk");
fff42755
VM
758
759 if (haves_bitmap)
760 bitmap_and_not(wants_bitmap, haves_bitmap);
761
3ae5fa07 762 bitmap_git->result = wants_bitmap;
30cdc33f 763 bitmap_git->haves = haves_bitmap;
fff42755 764
3ae5fa07 765 return bitmap_git;
f3c23db2
JT
766
767cleanup:
768 free_bitmap_index(bitmap_git);
769 return NULL;
fff42755
VM
770}
771
3ae5fa07
JT
772int reuse_partial_packfile_from_bitmap(struct bitmap_index *bitmap_git,
773 struct packed_git **packfile,
fff42755
VM
774 uint32_t *entries,
775 off_t *up_to)
776{
777 /*
778 * Reuse the packfile content if we need more than
779 * 90% of its objects
780 */
781 static const double REUSE_PERCENT = 0.9;
782
3ae5fa07 783 struct bitmap *result = bitmap_git->result;
fff42755
VM
784 uint32_t reuse_threshold;
785 uint32_t i, reuse_objects = 0;
786
787 assert(result);
788
789 for (i = 0; i < result->word_alloc; ++i) {
790 if (result->words[i] != (eword_t)~0) {
791 reuse_objects += ewah_bit_ctz64(~result->words[i]);
792 break;
793 }
794
34b935c0 795 reuse_objects += BITS_IN_EWORD;
fff42755
VM
796 }
797
798#ifdef GIT_BITMAP_DEBUG
799 {
800 const unsigned char *sha1;
801 struct revindex_entry *entry;
802
3ae5fa07
JT
803 entry = &bitmap_git->reverse_index->revindex[reuse_objects];
804 sha1 = nth_packed_object_sha1(bitmap_git->pack, entry->nr);
fff42755
VM
805
806 fprintf(stderr, "Failed to reuse at %d (%016llx)\n",
807 reuse_objects, result->words[i]);
808 fprintf(stderr, " %s\n", sha1_to_hex(sha1));
809 }
810#endif
811
812 if (!reuse_objects)
813 return -1;
814
3ae5fa07
JT
815 if (reuse_objects >= bitmap_git->pack->num_objects) {
816 bitmap_git->reuse_objects = *entries = bitmap_git->pack->num_objects;
fff42755 817 *up_to = -1; /* reuse the full pack */
3ae5fa07 818 *packfile = bitmap_git->pack;
fff42755
VM
819 return 0;
820 }
821
3ae5fa07 822 reuse_threshold = bitmap_popcount(bitmap_git->result) * REUSE_PERCENT;
fff42755
VM
823
824 if (reuse_objects < reuse_threshold)
825 return -1;
826
3ae5fa07
JT
827 bitmap_git->reuse_objects = *entries = reuse_objects;
828 *up_to = bitmap_git->pack->revindex[reuse_objects].offset;
829 *packfile = bitmap_git->pack;
fff42755
VM
830
831 return 0;
832}
833
3ae5fa07
JT
834void traverse_bitmap_commit_list(struct bitmap_index *bitmap_git,
835 show_reachable_fn show_reachable)
fff42755 836{
3ae5fa07 837 assert(bitmap_git->result);
fff42755 838
3ae5fa07 839 show_objects_for_type(bitmap_git, bitmap_git->commits,
fff42755 840 OBJ_COMMIT, show_reachable);
3ae5fa07 841 show_objects_for_type(bitmap_git, bitmap_git->trees,
fff42755 842 OBJ_TREE, show_reachable);
3ae5fa07 843 show_objects_for_type(bitmap_git, bitmap_git->blobs,
fff42755 844 OBJ_BLOB, show_reachable);
3ae5fa07 845 show_objects_for_type(bitmap_git, bitmap_git->tags,
fff42755
VM
846 OBJ_TAG, show_reachable);
847
3ae5fa07 848 show_extended_objects(bitmap_git, show_reachable);
fff42755
VM
849}
850
3ae5fa07 851static uint32_t count_object_type(struct bitmap_index *bitmap_git,
fff42755
VM
852 enum object_type type)
853{
3ae5fa07
JT
854 struct bitmap *objects = bitmap_git->result;
855 struct eindex *eindex = &bitmap_git->ext_index;
fff42755
VM
856
857 uint32_t i = 0, count = 0;
858 struct ewah_iterator it;
859 eword_t filter;
860
861 switch (type) {
862 case OBJ_COMMIT:
3ae5fa07 863 ewah_iterator_init(&it, bitmap_git->commits);
fff42755
VM
864 break;
865
866 case OBJ_TREE:
3ae5fa07 867 ewah_iterator_init(&it, bitmap_git->trees);
fff42755
VM
868 break;
869
870 case OBJ_BLOB:
3ae5fa07 871 ewah_iterator_init(&it, bitmap_git->blobs);
fff42755
VM
872 break;
873
874 case OBJ_TAG:
3ae5fa07 875 ewah_iterator_init(&it, bitmap_git->tags);
fff42755
VM
876 break;
877
878 default:
879 return 0;
880 }
881
882 while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) {
883 eword_t word = objects->words[i++] & filter;
884 count += ewah_bit_popcount64(word);
885 }
886
887 for (i = 0; i < eindex->count; ++i) {
888 if (eindex->objects[i]->type == type &&
3ae5fa07 889 bitmap_get(objects, bitmap_git->pack->num_objects + i))
fff42755
VM
890 count++;
891 }
892
893 return count;
894}
895
3ae5fa07
JT
896void count_bitmap_commit_list(struct bitmap_index *bitmap_git,
897 uint32_t *commits, uint32_t *trees,
fff42755
VM
898 uint32_t *blobs, uint32_t *tags)
899{
3ae5fa07 900 assert(bitmap_git->result);
fff42755
VM
901
902 if (commits)
3ae5fa07 903 *commits = count_object_type(bitmap_git, OBJ_COMMIT);
fff42755
VM
904
905 if (trees)
3ae5fa07 906 *trees = count_object_type(bitmap_git, OBJ_TREE);
fff42755
VM
907
908 if (blobs)
3ae5fa07 909 *blobs = count_object_type(bitmap_git, OBJ_BLOB);
fff42755
VM
910
911 if (tags)
3ae5fa07 912 *tags = count_object_type(bitmap_git, OBJ_TAG);
fff42755
VM
913}
914
915struct bitmap_test_data {
3ae5fa07 916 struct bitmap_index *bitmap_git;
fff42755
VM
917 struct bitmap *base;
918 struct progress *prg;
919 size_t seen;
920};
921
de1e67d0
JK
922static void test_show_object(struct object *object, const char *name,
923 void *data)
fff42755
VM
924{
925 struct bitmap_test_data *tdata = data;
926 int bitmap_pos;
927
3ae5fa07 928 bitmap_pos = bitmap_position(tdata->bitmap_git, object->oid.hash);
fff42755 929 if (bitmap_pos < 0)
f2fd0760 930 die("Object not in bitmap: %s\n", oid_to_hex(&object->oid));
fff42755
VM
931
932 bitmap_set(tdata->base, bitmap_pos);
933 display_progress(tdata->prg, ++tdata->seen);
934}
935
936static void test_show_commit(struct commit *commit, void *data)
937{
938 struct bitmap_test_data *tdata = data;
939 int bitmap_pos;
940
3ae5fa07
JT
941 bitmap_pos = bitmap_position(tdata->bitmap_git,
942 commit->object.oid.hash);
fff42755 943 if (bitmap_pos < 0)
f2fd0760 944 die("Object not in bitmap: %s\n", oid_to_hex(&commit->object.oid));
fff42755
VM
945
946 bitmap_set(tdata->base, bitmap_pos);
947 display_progress(tdata->prg, ++tdata->seen);
948}
949
950void test_bitmap_walk(struct rev_info *revs)
951{
952 struct object *root;
953 struct bitmap *result = NULL;
954 khiter_t pos;
955 size_t result_popcnt;
956 struct bitmap_test_data tdata;
3ae5fa07 957 struct bitmap_index *bitmap_git;
fff42755 958
7c141127 959 if (!(bitmap_git = prepare_bitmap_git(revs->repo)))
fff42755
VM
960 die("failed to load bitmap indexes");
961
962 if (revs->pending.nr != 1)
963 die("you must specify exactly one commit to test");
964
965 fprintf(stderr, "Bitmap v%d test (%d entries loaded)\n",
3ae5fa07 966 bitmap_git->version, bitmap_git->entry_count);
fff42755
VM
967
968 root = revs->pending.objects[0].item;
3ae5fa07 969 pos = kh_get_sha1(bitmap_git->bitmaps, root->oid.hash);
fff42755 970
3ae5fa07
JT
971 if (pos < kh_end(bitmap_git->bitmaps)) {
972 struct stored_bitmap *st = kh_value(bitmap_git->bitmaps, pos);
fff42755
VM
973 struct ewah_bitmap *bm = lookup_stored_bitmap(st);
974
975 fprintf(stderr, "Found bitmap for %s. %d bits / %08x checksum\n",
f2fd0760 976 oid_to_hex(&root->oid), (int)bm->bit_size, ewah_checksum(bm));
fff42755
VM
977
978 result = ewah_to_bitmap(bm);
979 }
980
981 if (result == NULL)
f2fd0760 982 die("Commit %s doesn't have an indexed bitmap", oid_to_hex(&root->oid));
fff42755
VM
983
984 revs->tag_objects = 1;
985 revs->tree_objects = 1;
986 revs->blob_objects = 1;
987
988 result_popcnt = bitmap_popcount(result);
989
990 if (prepare_revision_walk(revs))
991 die("revision walk setup failed");
992
3ae5fa07 993 tdata.bitmap_git = bitmap_git;
fff42755
VM
994 tdata.base = bitmap_new();
995 tdata.prg = start_progress("Verifying bitmap entries", result_popcnt);
996 tdata.seen = 0;
997
998 traverse_commit_list(revs, &test_show_commit, &test_show_object, &tdata);
999
1000 stop_progress(&tdata.prg);
1001
1002 if (bitmap_equals(result, tdata.base))
1003 fprintf(stderr, "OK!\n");
1004 else
1005 fprintf(stderr, "Mismatch!\n");
f86a3747 1006
f3c23db2 1007 free_bitmap_index(bitmap_git);
fff42755 1008}
7cc8f971
VM
1009
1010static int rebuild_bitmap(uint32_t *reposition,
1011 struct ewah_bitmap *source,
1012 struct bitmap *dest)
1013{
1014 uint32_t pos = 0;
1015 struct ewah_iterator it;
1016 eword_t word;
1017
1018 ewah_iterator_init(&it, source);
1019
1020 while (ewah_iterator_next(&word, &it)) {
1021 uint32_t offset, bit_pos;
1022
34b935c0 1023 for (offset = 0; offset < BITS_IN_EWORD; ++offset) {
7cc8f971
VM
1024 if ((word >> offset) == 0)
1025 break;
1026
1027 offset += ewah_bit_ctz64(word >> offset);
1028
1029 bit_pos = reposition[pos + offset];
1030 if (bit_pos > 0)
1031 bitmap_set(dest, bit_pos - 1);
1032 else /* can't reuse, we don't have the object */
1033 return -1;
1034 }
1035
34b935c0 1036 pos += BITS_IN_EWORD;
7cc8f971
VM
1037 }
1038 return 0;
1039}
1040
3ae5fa07
JT
1041int rebuild_existing_bitmaps(struct bitmap_index *bitmap_git,
1042 struct packing_data *mapping,
7cc8f971
VM
1043 khash_sha1 *reused_bitmaps,
1044 int show_progress)
1045{
1046 uint32_t i, num_objects;
1047 uint32_t *reposition;
1048 struct bitmap *rebuild;
1049 struct stored_bitmap *stored;
1050 struct progress *progress = NULL;
1051
1052 khiter_t hash_pos;
1053 int hash_ret;
1054
3ae5fa07 1055 num_objects = bitmap_git->pack->num_objects;
7cc8f971
VM
1056 reposition = xcalloc(num_objects, sizeof(uint32_t));
1057
1058 for (i = 0; i < num_objects; ++i) {
1059 const unsigned char *sha1;
1060 struct revindex_entry *entry;
1061 struct object_entry *oe;
1062
3ae5fa07
JT
1063 entry = &bitmap_git->pack->revindex[i];
1064 sha1 = nth_packed_object_sha1(bitmap_git->pack, entry->nr);
7cc8f971
VM
1065 oe = packlist_find(mapping, sha1, NULL);
1066
1067 if (oe)
06af3bba 1068 reposition[i] = oe_in_pack_pos(mapping, oe) + 1;
7cc8f971
VM
1069 }
1070
1071 rebuild = bitmap_new();
1072 i = 0;
1073
1074 if (show_progress)
1075 progress = start_progress("Reusing bitmaps", 0);
1076
3ae5fa07 1077 kh_foreach_value(bitmap_git->bitmaps, stored, {
7cc8f971
VM
1078 if (stored->flags & BITMAP_FLAG_REUSE) {
1079 if (!rebuild_bitmap(reposition,
1080 lookup_stored_bitmap(stored),
1081 rebuild)) {
1082 hash_pos = kh_put_sha1(reused_bitmaps,
53636539 1083 stored->oid.hash,
7cc8f971
VM
1084 &hash_ret);
1085 kh_value(reused_bitmaps, hash_pos) =
1086 bitmap_to_ewah(rebuild);
1087 }
1088 bitmap_reset(rebuild);
1089 display_progress(progress, ++i);
1090 }
1091 });
1092
1093 stop_progress(&progress);
1094
1095 free(reposition);
1096 bitmap_free(rebuild);
1097 return 0;
1098}
f3c23db2
JT
1099
1100void free_bitmap_index(struct bitmap_index *b)
1101{
1102 if (!b)
1103 return;
1104
1105 if (b->map)
1106 munmap(b->map, b->map_size);
1107 ewah_pool_free(b->commits);
1108 ewah_pool_free(b->trees);
1109 ewah_pool_free(b->blobs);
1110 ewah_pool_free(b->tags);
1111 kh_destroy_sha1(b->bitmaps);
1112 free(b->ext_index.objects);
1113 free(b->ext_index.hashes);
1114 bitmap_free(b->result);
30cdc33f 1115 bitmap_free(b->haves);
f3c23db2
JT
1116 free(b);
1117}
30cdc33f
JK
1118
1119int bitmap_has_sha1_in_uninteresting(struct bitmap_index *bitmap_git,
1120 const unsigned char *sha1)
1121{
1122 int pos;
1123
1124 if (!bitmap_git)
1125 return 0; /* no bitmap loaded */
30cdc33f
JK
1126 if (!bitmap_git->haves)
1127 return 0; /* walk had no "haves" */
1128
1129 pos = bitmap_position_packfile(bitmap_git, sha1);
1130 if (pos < 0)
1131 return 0;
1132
1133 return bitmap_get(bitmap_git->haves, pos);
1134}