]>
Commit | Line | Data |
---|---|---|
b6fdc44c | 1 | #include "git-compat-util.h" |
fff42755 | 2 | #include "commit.h" |
f394e093 | 3 | #include "gettext.h" |
41771fa4 | 4 | #include "hex.h" |
baf20c39 | 5 | #include "strbuf.h" |
fff42755 VM |
6 | #include "tag.h" |
7 | #include "diff.h" | |
8 | #include "revision.h" | |
9 | #include "progress.h" | |
10 | #include "list-objects.h" | |
11 | #include "pack.h" | |
12 | #include "pack-bitmap.h" | |
13 | #include "pack-revindex.h" | |
14 | #include "pack-objects.h" | |
0317f455 | 15 | #include "packfile.h" |
a80d72db | 16 | #include "repository.h" |
74ea5c95 | 17 | #include "trace2.h" |
87bed179 | 18 | #include "object-file.h" |
a034e910 | 19 | #include "object-store-ll.h" |
6663ae0a | 20 | #include "list-objects-filter-options.h" |
0f533c72 | 21 | #include "midx.h" |
3f267a11 | 22 | #include "config.h" |
fff42755 VM |
23 | |
24 | /* | |
25 | * An entry on the bitmap index, representing the bitmap for a given | |
26 | * commit. | |
27 | */ | |
28 | struct stored_bitmap { | |
53636539 | 29 | struct object_id oid; |
fff42755 VM |
30 | struct ewah_bitmap *root; |
31 | struct stored_bitmap *xor; | |
32 | int flags; | |
33 | }; | |
34 | ||
35 | /* | |
3ae5fa07 | 36 | * The active bitmap index for a repository. By design, repositories only have |
fff42755 VM |
37 | * a single bitmap index available (the index for the biggest packfile in |
38 | * the repository), since bitmap indexes need full closure. | |
39 | * | |
40 | * If there is more than one bitmap index available (e.g. because of alternates), | |
41 | * the active bitmap index is the largest one. | |
42 | */ | |
3ae5fa07 | 43 | struct bitmap_index { |
0f533c72 TB |
44 | /* |
45 | * The pack or multi-pack index (MIDX) that this bitmap index belongs | |
46 | * to. | |
47 | * | |
48 | * Exactly one of these must be non-NULL; this specifies the object | |
49 | * order used to interpret this bitmap. | |
50 | */ | |
fff42755 | 51 | struct packed_git *pack; |
0f533c72 | 52 | struct multi_pack_index *midx; |
fff42755 | 53 | |
fff42755 VM |
54 | /* |
55 | * Mark the first `reuse_objects` in the packfile as reused: | |
56 | * they will be sent as-is without using them for repacking | |
57 | * calculations | |
58 | */ | |
59 | uint32_t reuse_objects; | |
60 | ||
61 | /* mmapped buffer of the whole bitmap index */ | |
62 | unsigned char *map; | |
63 | size_t map_size; /* size of the mmaped buffer */ | |
64 | size_t map_pos; /* current position when loading the index */ | |
65 | ||
66 | /* | |
67 | * Type indexes. | |
68 | * | |
69 | * Each bitmap marks which objects in the packfile are of the given | |
70 | * type. This provides type information when yielding the objects from | |
71 | * the packfile during a walk, which allows for better delta bases. | |
72 | */ | |
73 | struct ewah_bitmap *commits; | |
74 | struct ewah_bitmap *trees; | |
75 | struct ewah_bitmap *blobs; | |
76 | struct ewah_bitmap *tags; | |
77 | ||
3c771448 | 78 | /* Map from object ID -> `stored_bitmap` for all the bitmapped commits */ |
79 | kh_oid_map_t *bitmaps; | |
fff42755 VM |
80 | |
81 | /* Number of bitmapped commits */ | |
82 | uint32_t entry_count; | |
83 | ||
f3c23db2 | 84 | /* If not NULL, this is a name-hash cache pointing into map. */ |
ae4f07fb VM |
85 | uint32_t *hashes; |
86 | ||
0f533c72 TB |
87 | /* The checksum of the packfile or MIDX; points into map. */ |
88 | const unsigned char *checksum; | |
89 | ||
28cd7306 AC |
90 | /* |
91 | * If not NULL, this point into the commit table extension | |
92 | * (within the memory mapped region `map`). | |
93 | */ | |
94 | unsigned char *table_lookup; | |
95 | ||
fff42755 VM |
96 | /* |
97 | * Extended index. | |
98 | * | |
99 | * When trying to perform bitmap operations with objects that are not | |
100 | * packed in `pack`, these objects are added to this "fake index" and | |
101 | * are assumed to appear at the end of the packfile for all operations | |
102 | */ | |
103 | struct eindex { | |
104 | struct object **objects; | |
105 | uint32_t *hashes; | |
106 | uint32_t count, alloc; | |
3c771448 | 107 | kh_oid_pos_t *positions; |
fff42755 VM |
108 | } ext_index; |
109 | ||
110 | /* Bitmap result of the last performed walk */ | |
111 | struct bitmap *result; | |
112 | ||
30cdc33f JK |
113 | /* "have" bitmap from the last performed walk */ |
114 | struct bitmap *haves; | |
115 | ||
fff42755 VM |
116 | /* Version of the bitmap index */ |
117 | unsigned int version; | |
3ae5fa07 | 118 | }; |
fff42755 VM |
119 | |
120 | static struct ewah_bitmap *lookup_stored_bitmap(struct stored_bitmap *st) | |
121 | { | |
122 | struct ewah_bitmap *parent; | |
123 | struct ewah_bitmap *composed; | |
124 | ||
afe8a907 | 125 | if (!st->xor) |
fff42755 VM |
126 | return st->root; |
127 | ||
128 | composed = ewah_pool_new(); | |
129 | parent = lookup_stored_bitmap(st->xor); | |
130 | ewah_xor(st->root, parent, composed); | |
131 | ||
132 | ewah_pool_free(st->root); | |
133 | st->root = composed; | |
134 | st->xor = NULL; | |
135 | ||
136 | return composed; | |
137 | } | |
138 | ||
139 | /* | |
140 | * Read a bitmap from the current read position on the mmaped | |
141 | * index, and increase the read position accordingly | |
142 | */ | |
143 | static struct ewah_bitmap *read_bitmap_1(struct bitmap_index *index) | |
144 | { | |
145 | struct ewah_bitmap *b = ewah_pool_new(); | |
146 | ||
1140bf01 | 147 | ssize_t bitmap_size = ewah_read_mmap(b, |
fff42755 VM |
148 | index->map + index->map_pos, |
149 | index->map_size - index->map_pos); | |
150 | ||
151 | if (bitmap_size < 0) { | |
9975975d | 152 | error(_("failed to load bitmap index (corrupted?)")); |
fff42755 VM |
153 | ewah_pool_free(b); |
154 | return NULL; | |
155 | } | |
156 | ||
157 | index->map_pos += bitmap_size; | |
158 | return b; | |
159 | } | |
160 | ||
ed184620 TB |
161 | static uint32_t bitmap_num_objects(struct bitmap_index *index) |
162 | { | |
0f533c72 TB |
163 | if (index->midx) |
164 | return index->midx->num_objects; | |
ed184620 TB |
165 | return index->pack->num_objects; |
166 | } | |
167 | ||
fff42755 VM |
168 | static int load_bitmap_header(struct bitmap_index *index) |
169 | { | |
170 | struct bitmap_disk_header *header = (void *)index->map; | |
ca510902 | 171 | size_t header_size = sizeof(*header) - GIT_MAX_RAWSZ + the_hash_algo->rawsz; |
fff42755 | 172 | |
ca510902 | 173 | if (index->map_size < header_size + the_hash_algo->rawsz) |
9975975d | 174 | return error(_("corrupted bitmap index (too small)")); |
fff42755 VM |
175 | |
176 | if (memcmp(header->magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE)) != 0) | |
9975975d | 177 | return error(_("corrupted bitmap index file (wrong header)")); |
fff42755 VM |
178 | |
179 | index->version = ntohs(header->version); | |
180 | if (index->version != 1) | |
9975975d | 181 | return error(_("unsupported version '%d' for bitmap index file"), index->version); |
fff42755 VM |
182 | |
183 | /* Parse known bitmap format options */ | |
184 | { | |
185 | uint32_t flags = ntohs(header->options); | |
ed184620 | 186 | size_t cache_size = st_mult(bitmap_num_objects(index), sizeof(uint32_t)); |
ec6c7b43 | 187 | unsigned char *index_end = index->map + index->map_size - the_hash_algo->rawsz; |
fff42755 VM |
188 | |
189 | if ((flags & BITMAP_OPT_FULL_DAG) == 0) | |
baf20c39 | 190 | BUG("unsupported options for bitmap index file " |
fff42755 | 191 | "(Git requires BITMAP_OPT_FULL_DAG)"); |
ae4f07fb VM |
192 | |
193 | if (flags & BITMAP_OPT_HASH_CACHE) { | |
ec6c7b43 | 194 | if (cache_size > index_end - index->map - header_size) |
9975975d | 195 | return error(_("corrupted bitmap index file (too short to fit hash cache)")); |
ec6c7b43 JK |
196 | index->hashes = (void *)(index_end - cache_size); |
197 | index_end -= cache_size; | |
ae4f07fb | 198 | } |
28cd7306 AC |
199 | |
200 | if (flags & BITMAP_OPT_LOOKUP_TABLE) { | |
201 | size_t table_size = st_mult(ntohl(header->entry_count), | |
202 | BITMAP_LOOKUP_TABLE_TRIPLET_WIDTH); | |
203 | if (table_size > index_end - index->map - header_size) | |
204 | return error(_("corrupted bitmap index file (too short to fit lookup table)")); | |
205 | if (git_env_bool("GIT_TEST_READ_COMMIT_TABLE", 1)) | |
206 | index->table_lookup = (void *)(index_end - table_size); | |
207 | index_end -= table_size; | |
208 | } | |
fff42755 VM |
209 | } |
210 | ||
211 | index->entry_count = ntohl(header->entry_count); | |
0f533c72 | 212 | index->checksum = header->checksum; |
ca510902 | 213 | index->map_pos += header_size; |
fff42755 VM |
214 | return 0; |
215 | } | |
216 | ||
217 | static struct stored_bitmap *store_bitmap(struct bitmap_index *index, | |
218 | struct ewah_bitmap *root, | |
500e4f23 | 219 | const struct object_id *oid, |
fff42755 VM |
220 | struct stored_bitmap *xor_with, |
221 | int flags) | |
222 | { | |
223 | struct stored_bitmap *stored; | |
224 | khiter_t hash_pos; | |
225 | int ret; | |
226 | ||
227 | stored = xmalloc(sizeof(struct stored_bitmap)); | |
228 | stored->root = root; | |
229 | stored->xor = xor_with; | |
230 | stored->flags = flags; | |
500e4f23 | 231 | oidcpy(&stored->oid, oid); |
fff42755 | 232 | |
3c771448 | 233 | hash_pos = kh_put_oid_map(index->bitmaps, stored->oid, &ret); |
fff42755 | 234 | |
28cd7306 AC |
235 | /* |
236 | * A 0 return code means the insertion succeeded with no changes, | |
237 | * because the SHA1 already existed on the map. This is bad, there | |
238 | * shouldn't be duplicated commits in the index. | |
239 | */ | |
fff42755 | 240 | if (ret == 0) { |
9975975d | 241 | error(_("duplicate entry in bitmap index: '%s'"), oid_to_hex(oid)); |
fff42755 VM |
242 | return NULL; |
243 | } | |
244 | ||
245 | kh_value(index->bitmaps, hash_pos) = stored; | |
246 | return stored; | |
247 | } | |
248 | ||
b5007211 KB |
249 | static inline uint32_t read_be32(const unsigned char *buffer, size_t *pos) |
250 | { | |
251 | uint32_t result = get_be32(buffer + *pos); | |
252 | (*pos) += sizeof(result); | |
253 | return result; | |
254 | } | |
255 | ||
256 | static inline uint8_t read_u8(const unsigned char *buffer, size_t *pos) | |
257 | { | |
258 | return buffer[(*pos)++]; | |
259 | } | |
260 | ||
599dc766 RS |
261 | #define MAX_XOR_OFFSET 160 |
262 | ||
6b4277e6 TB |
263 | static int nth_bitmap_object_oid(struct bitmap_index *index, |
264 | struct object_id *oid, | |
265 | uint32_t n) | |
266 | { | |
0f533c72 TB |
267 | if (index->midx) |
268 | return nth_midxed_object_oid(oid, index->midx, n) ? 0 : -1; | |
6b4277e6 TB |
269 | return nth_packed_object_id(oid, index->pack, n); |
270 | } | |
271 | ||
fff42755 VM |
272 | static int load_bitmap_entries_v1(struct bitmap_index *index) |
273 | { | |
fff42755 | 274 | uint32_t i; |
599dc766 | 275 | struct stored_bitmap *recent_bitmaps[MAX_XOR_OFFSET] = { NULL }; |
fff42755 VM |
276 | |
277 | for (i = 0; i < index->entry_count; ++i) { | |
278 | int xor_offset, flags; | |
279 | struct ewah_bitmap *bitmap = NULL; | |
280 | struct stored_bitmap *xor_bitmap = NULL; | |
281 | uint32_t commit_idx_pos; | |
500e4f23 | 282 | struct object_id oid; |
fff42755 | 283 | |
c6b0c391 | 284 | if (index->map_size - index->map_pos < 6) |
9975975d | 285 | return error(_("corrupt ewah bitmap: truncated header for entry %d"), i); |
c6b0c391 | 286 | |
b5007211 KB |
287 | commit_idx_pos = read_be32(index->map, &index->map_pos); |
288 | xor_offset = read_u8(index->map, &index->map_pos); | |
289 | flags = read_u8(index->map, &index->map_pos); | |
fff42755 | 290 | |
6b4277e6 | 291 | if (nth_bitmap_object_oid(index, &oid, commit_idx_pos) < 0) |
9975975d | 292 | return error(_("corrupt ewah bitmap: commit index %u out of range"), |
c6b0c391 | 293 | (unsigned)commit_idx_pos); |
fff42755 | 294 | |
fff42755 VM |
295 | bitmap = read_bitmap_1(index); |
296 | if (!bitmap) | |
297 | return -1; | |
298 | ||
299 | if (xor_offset > MAX_XOR_OFFSET || xor_offset > i) | |
9975975d | 300 | return error(_("corrupted bitmap pack index")); |
fff42755 VM |
301 | |
302 | if (xor_offset > 0) { | |
303 | xor_bitmap = recent_bitmaps[(i - xor_offset) % MAX_XOR_OFFSET]; | |
304 | ||
afe8a907 | 305 | if (!xor_bitmap) |
9975975d | 306 | return error(_("invalid XOR offset in bitmap pack index")); |
fff42755 VM |
307 | } |
308 | ||
309 | recent_bitmaps[i % MAX_XOR_OFFSET] = store_bitmap( | |
500e4f23 | 310 | index, bitmap, &oid, xor_bitmap, flags); |
fff42755 VM |
311 | } |
312 | ||
313 | return 0; | |
314 | } | |
315 | ||
0f533c72 TB |
316 | char *midx_bitmap_filename(struct multi_pack_index *midx) |
317 | { | |
60980aed TB |
318 | struct strbuf buf = STRBUF_INIT; |
319 | ||
320 | get_midx_filename(&buf, midx->object_dir); | |
321 | strbuf_addf(&buf, "-%s.bitmap", hash_to_hex(get_midx_checksum(midx))); | |
322 | ||
323 | return strbuf_detach(&buf, NULL); | |
0f533c72 TB |
324 | } |
325 | ||
326 | char *pack_bitmap_filename(struct packed_git *p) | |
cb468050 | 327 | { |
9ae97018 | 328 | size_t len; |
cb468050 | 329 | |
9ae97018 | 330 | if (!strip_suffix(p->pack_name, ".pack", &len)) |
033abf97 | 331 | BUG("pack_name does not end in .pack"); |
9ae97018 | 332 | return xstrfmt("%.*s.bitmap", (int)len, p->pack_name); |
cb468050 JH |
333 | } |
334 | ||
0f533c72 TB |
335 | static int open_midx_bitmap_1(struct bitmap_index *bitmap_git, |
336 | struct multi_pack_index *midx) | |
337 | { | |
338 | struct stat st; | |
349c26ff TL |
339 | char *bitmap_name = midx_bitmap_filename(midx); |
340 | int fd = git_open(bitmap_name); | |
44f9fd64 TB |
341 | uint32_t i; |
342 | struct packed_git *preferred; | |
0f533c72 | 343 | |
6411cc08 TL |
344 | if (fd < 0) { |
345 | if (errno != ENOENT) | |
346 | warning_errno("cannot open '%s'", bitmap_name); | |
347 | free(bitmap_name); | |
0f533c72 | 348 | return -1; |
6411cc08 TL |
349 | } |
350 | free(bitmap_name); | |
0f533c72 TB |
351 | |
352 | if (fstat(fd, &st)) { | |
9005eb02 | 353 | error_errno(_("cannot fstat bitmap file")); |
0f533c72 TB |
354 | close(fd); |
355 | return -1; | |
356 | } | |
357 | ||
358 | if (bitmap_git->pack || bitmap_git->midx) { | |
60980aed TB |
359 | struct strbuf buf = STRBUF_INIT; |
360 | get_midx_filename(&buf, midx->object_dir); | |
8ddc0663 TL |
361 | trace2_data_string("bitmap", the_repository, |
362 | "ignoring extra midx bitmap file", buf.buf); | |
0f533c72 | 363 | close(fd); |
60980aed | 364 | strbuf_release(&buf); |
0f533c72 TB |
365 | return -1; |
366 | } | |
367 | ||
368 | bitmap_git->midx = midx; | |
369 | bitmap_git->map_size = xsize_t(st.st_size); | |
370 | bitmap_git->map_pos = 0; | |
371 | bitmap_git->map = xmmap(NULL, bitmap_git->map_size, PROT_READ, | |
372 | MAP_PRIVATE, fd, 0); | |
373 | close(fd); | |
374 | ||
375 | if (load_bitmap_header(bitmap_git) < 0) | |
376 | goto cleanup; | |
377 | ||
9005eb02 TL |
378 | if (!hasheq(get_midx_checksum(bitmap_git->midx), bitmap_git->checksum)) { |
379 | error(_("checksum doesn't match in MIDX and bitmap")); | |
0f533c72 | 380 | goto cleanup; |
9005eb02 | 381 | } |
0f533c72 | 382 | |
5a6072f6 | 383 | if (load_midx_revindex(bitmap_git->midx)) { |
0f533c72 TB |
384 | warning(_("multi-pack bitmap is missing required reverse index")); |
385 | goto cleanup; | |
386 | } | |
44f9fd64 TB |
387 | |
388 | for (i = 0; i < bitmap_git->midx->num_packs; i++) { | |
06f38678 TB |
389 | if (prepare_midx_pack(the_repository, bitmap_git->midx, i)) { |
390 | warning(_("could not open pack %s"), | |
391 | bitmap_git->midx->pack_names[i]); | |
392 | goto cleanup; | |
393 | } | |
44f9fd64 TB |
394 | } |
395 | ||
396 | preferred = bitmap_git->midx->packs[midx_preferred_pack(bitmap_git)]; | |
397 | if (!is_pack_valid(preferred)) { | |
398 | warning(_("preferred pack (%s) is invalid"), | |
399 | preferred->pack_name); | |
400 | goto cleanup; | |
401 | } | |
402 | ||
0f533c72 TB |
403 | return 0; |
404 | ||
405 | cleanup: | |
406 | munmap(bitmap_git->map, bitmap_git->map_size); | |
407 | bitmap_git->map_size = 0; | |
f8b60cf9 | 408 | bitmap_git->map_pos = 0; |
0f533c72 | 409 | bitmap_git->map = NULL; |
f8b60cf9 | 410 | bitmap_git->midx = NULL; |
0f533c72 TB |
411 | return -1; |
412 | } | |
413 | ||
3ae5fa07 | 414 | static int open_pack_bitmap_1(struct bitmap_index *bitmap_git, struct packed_git *packfile) |
fff42755 VM |
415 | { |
416 | int fd; | |
417 | struct stat st; | |
349c26ff | 418 | char *bitmap_name; |
fff42755 | 419 | |
349c26ff TL |
420 | bitmap_name = pack_bitmap_filename(packfile); |
421 | fd = git_open(bitmap_name); | |
fff42755 | 422 | |
6411cc08 TL |
423 | if (fd < 0) { |
424 | if (errno != ENOENT) | |
425 | warning_errno("cannot open '%s'", bitmap_name); | |
426 | free(bitmap_name); | |
fff42755 | 427 | return -1; |
6411cc08 TL |
428 | } |
429 | free(bitmap_name); | |
fff42755 VM |
430 | |
431 | if (fstat(fd, &st)) { | |
9005eb02 | 432 | error_errno(_("cannot fstat bitmap file")); |
fff42755 VM |
433 | close(fd); |
434 | return -1; | |
435 | } | |
436 | ||
0f533c72 | 437 | if (bitmap_git->pack || bitmap_git->midx) { |
8ddc0663 TL |
438 | trace2_data_string("bitmap", the_repository, |
439 | "ignoring extra bitmap file", packfile->pack_name); | |
fff42755 VM |
440 | close(fd); |
441 | return -1; | |
442 | } | |
443 | ||
dc1daacd JK |
444 | if (!is_pack_valid(packfile)) { |
445 | close(fd); | |
446 | return -1; | |
447 | } | |
448 | ||
3ae5fa07 JT |
449 | bitmap_git->pack = packfile; |
450 | bitmap_git->map_size = xsize_t(st.st_size); | |
451 | bitmap_git->map = xmmap(NULL, bitmap_git->map_size, PROT_READ, MAP_PRIVATE, fd, 0); | |
452 | bitmap_git->map_pos = 0; | |
fff42755 VM |
453 | close(fd); |
454 | ||
3ae5fa07 JT |
455 | if (load_bitmap_header(bitmap_git) < 0) { |
456 | munmap(bitmap_git->map, bitmap_git->map_size); | |
457 | bitmap_git->map = NULL; | |
458 | bitmap_git->map_size = 0; | |
f8b60cf9 TB |
459 | bitmap_git->map_pos = 0; |
460 | bitmap_git->pack = NULL; | |
fff42755 VM |
461 | return -1; |
462 | } | |
463 | ||
8ddc0663 TL |
464 | trace2_data_string("bitmap", the_repository, "opened bitmap file", |
465 | packfile->pack_name); | |
fff42755 VM |
466 | return 0; |
467 | } | |
468 | ||
65308ad8 | 469 | static int load_reverse_index(struct repository *r, struct bitmap_index *bitmap_git) |
0f533c72 TB |
470 | { |
471 | if (bitmap_is_midx(bitmap_git)) { | |
472 | uint32_t i; | |
473 | int ret; | |
474 | ||
475 | /* | |
476 | * The multi-pack-index's .rev file is already loaded via | |
477 | * open_pack_bitmap_1(). | |
478 | * | |
479 | * But we still need to open the individual pack .rev files, | |
480 | * since we will need to make use of them in pack-objects. | |
481 | */ | |
482 | for (i = 0; i < bitmap_git->midx->num_packs; i++) { | |
65308ad8 | 483 | ret = load_pack_revindex(r, bitmap_git->midx->packs[i]); |
0f533c72 TB |
484 | if (ret) |
485 | return ret; | |
486 | } | |
487 | return 0; | |
488 | } | |
65308ad8 | 489 | return load_pack_revindex(r, bitmap_git->pack); |
0f533c72 TB |
490 | } |
491 | ||
65308ad8 | 492 | static int load_bitmap(struct repository *r, struct bitmap_index *bitmap_git) |
fff42755 | 493 | { |
199c86be | 494 | assert(bitmap_git->map); |
fff42755 | 495 | |
3c771448 | 496 | bitmap_git->bitmaps = kh_init_oid_map(); |
497 | bitmap_git->ext_index.positions = kh_init_oid_pos(); | |
0f533c72 | 498 | |
65308ad8 | 499 | if (load_reverse_index(r, bitmap_git)) |
4828ce98 | 500 | goto failed; |
fff42755 | 501 | |
3ae5fa07 JT |
502 | if (!(bitmap_git->commits = read_bitmap_1(bitmap_git)) || |
503 | !(bitmap_git->trees = read_bitmap_1(bitmap_git)) || | |
504 | !(bitmap_git->blobs = read_bitmap_1(bitmap_git)) || | |
505 | !(bitmap_git->tags = read_bitmap_1(bitmap_git))) | |
fff42755 VM |
506 | goto failed; |
507 | ||
28cd7306 | 508 | if (!bitmap_git->table_lookup && load_bitmap_entries_v1(bitmap_git) < 0) |
fff42755 VM |
509 | goto failed; |
510 | ||
fff42755 VM |
511 | return 0; |
512 | ||
513 | failed: | |
3ae5fa07 JT |
514 | munmap(bitmap_git->map, bitmap_git->map_size); |
515 | bitmap_git->map = NULL; | |
516 | bitmap_git->map_size = 0; | |
bb514de3 JK |
517 | |
518 | kh_destroy_oid_map(bitmap_git->bitmaps); | |
519 | bitmap_git->bitmaps = NULL; | |
520 | ||
521 | kh_destroy_oid_pos(bitmap_git->ext_index.positions); | |
522 | bitmap_git->ext_index.positions = NULL; | |
523 | ||
fff42755 VM |
524 | return -1; |
525 | } | |
526 | ||
7c141127 NTND |
527 | static int open_pack_bitmap(struct repository *r, |
528 | struct bitmap_index *bitmap_git) | |
fff42755 VM |
529 | { |
530 | struct packed_git *p; | |
531 | int ret = -1; | |
532 | ||
7c141127 | 533 | for (p = get_all_packs(r); p; p = p->next) { |
833f4c05 | 534 | if (open_pack_bitmap_1(bitmap_git, p) == 0) { |
fff42755 | 535 | ret = 0; |
833f4c05 JK |
536 | /* |
537 | * The only reason to keep looking is to report | |
538 | * duplicates. | |
539 | */ | |
540 | if (!trace2_is_enabled()) | |
541 | break; | |
542 | } | |
fff42755 VM |
543 | } |
544 | ||
545 | return ret; | |
546 | } | |
547 | ||
0f533c72 TB |
548 | static int open_midx_bitmap(struct repository *r, |
549 | struct bitmap_index *bitmap_git) | |
550 | { | |
5dcee7c7 | 551 | int ret = -1; |
0f533c72 TB |
552 | struct multi_pack_index *midx; |
553 | ||
554 | assert(!bitmap_git->map); | |
555 | ||
556 | for (midx = get_multi_pack_index(r); midx; midx = midx->next) { | |
557 | if (!open_midx_bitmap_1(bitmap_git, midx)) | |
5dcee7c7 | 558 | ret = 0; |
0f533c72 | 559 | } |
5dcee7c7 | 560 | return ret; |
0f533c72 TB |
561 | } |
562 | ||
563 | static int open_bitmap(struct repository *r, | |
564 | struct bitmap_index *bitmap_git) | |
565 | { | |
c8f43570 JK |
566 | int found; |
567 | ||
0f533c72 TB |
568 | assert(!bitmap_git->map); |
569 | ||
c8f43570 JK |
570 | found = !open_midx_bitmap(r, bitmap_git); |
571 | ||
572 | /* | |
573 | * these will all be skipped if we opened a midx bitmap; but run it | |
574 | * anyway if tracing is enabled to report the duplicates | |
575 | */ | |
576 | if (!found || trace2_is_enabled()) | |
577 | found |= !open_pack_bitmap(r, bitmap_git); | |
578 | ||
579 | return found ? 0 : -1; | |
0f533c72 TB |
580 | } |
581 | ||
7c141127 | 582 | struct bitmap_index *prepare_bitmap_git(struct repository *r) |
fff42755 | 583 | { |
3ae5fa07 | 584 | struct bitmap_index *bitmap_git = xcalloc(1, sizeof(*bitmap_git)); |
fff42755 | 585 | |
65308ad8 | 586 | if (!open_bitmap(r, bitmap_git) && !load_bitmap(r, bitmap_git)) |
0f533c72 TB |
587 | return bitmap_git; |
588 | ||
589 | free_bitmap_index(bitmap_git); | |
590 | return NULL; | |
591 | } | |
592 | ||
bfbb60d3 | 593 | struct bitmap_index *prepare_midx_bitmap_git(struct multi_pack_index *midx) |
0f533c72 | 594 | { |
65308ad8 | 595 | struct repository *r = the_repository; |
0f533c72 TB |
596 | struct bitmap_index *bitmap_git = xcalloc(1, sizeof(*bitmap_git)); |
597 | ||
65308ad8 | 598 | if (!open_midx_bitmap_1(bitmap_git, midx) && !load_bitmap(r, bitmap_git)) |
3ae5fa07 | 599 | return bitmap_git; |
fff42755 | 600 | |
f3c23db2 | 601 | free_bitmap_index(bitmap_git); |
3ae5fa07 | 602 | return NULL; |
fff42755 VM |
603 | } |
604 | ||
605 | struct include_data { | |
3ae5fa07 | 606 | struct bitmap_index *bitmap_git; |
fff42755 VM |
607 | struct bitmap *base; |
608 | struct bitmap *seen; | |
609 | }; | |
610 | ||
28cd7306 AC |
611 | struct bitmap_lookup_table_triplet { |
612 | uint32_t commit_pos; | |
613 | uint64_t offset; | |
614 | uint32_t xor_row; | |
615 | }; | |
616 | ||
617 | struct bitmap_lookup_table_xor_item { | |
618 | struct object_id oid; | |
619 | uint64_t offset; | |
620 | }; | |
621 | ||
622 | /* | |
623 | * Given a `triplet` struct pointer and pointer `p`, this | |
624 | * function reads the triplet beginning at `p` into the struct. | |
625 | * Note that this function assumes that there is enough memory | |
626 | * left for filling the `triplet` struct from `p`. | |
627 | */ | |
628 | static int bitmap_lookup_table_get_triplet_by_pointer(struct bitmap_lookup_table_triplet *triplet, | |
629 | const unsigned char *p) | |
630 | { | |
631 | if (!triplet) | |
632 | return -1; | |
633 | ||
634 | triplet->commit_pos = get_be32(p); | |
635 | p += sizeof(uint32_t); | |
636 | triplet->offset = get_be64(p); | |
637 | p += sizeof(uint64_t); | |
638 | triplet->xor_row = get_be32(p); | |
639 | return 0; | |
640 | } | |
641 | ||
642 | /* | |
643 | * This function gets the raw triplet from `row`'th row in the | |
644 | * lookup table and fills that data to the `triplet`. | |
645 | */ | |
646 | static int bitmap_lookup_table_get_triplet(struct bitmap_index *bitmap_git, | |
647 | uint32_t pos, | |
648 | struct bitmap_lookup_table_triplet *triplet) | |
649 | { | |
650 | unsigned char *p = NULL; | |
651 | if (pos >= bitmap_git->entry_count) | |
652 | return error(_("corrupt bitmap lookup table: triplet position out of index")); | |
653 | ||
654 | p = bitmap_git->table_lookup + st_mult(pos, BITMAP_LOOKUP_TABLE_TRIPLET_WIDTH); | |
655 | ||
656 | return bitmap_lookup_table_get_triplet_by_pointer(triplet, p); | |
657 | } | |
658 | ||
659 | /* | |
660 | * Searches for a matching triplet. `commit_pos` is a pointer | |
661 | * to the wanted commit position value. `table_entry` points to | |
662 | * a triplet in lookup table. The first 4 bytes of each | |
663 | * triplet (pointed by `table_entry`) are compared with `*commit_pos`. | |
664 | */ | |
665 | static int triplet_cmp(const void *commit_pos, const void *table_entry) | |
666 | { | |
667 | ||
668 | uint32_t a = *(uint32_t *)commit_pos; | |
669 | uint32_t b = get_be32(table_entry); | |
670 | if (a > b) | |
671 | return 1; | |
672 | else if (a < b) | |
673 | return -1; | |
674 | ||
675 | return 0; | |
676 | } | |
677 | ||
678 | static uint32_t bitmap_bsearch_pos(struct bitmap_index *bitmap_git, | |
679 | struct object_id *oid, | |
680 | uint32_t *result) | |
681 | { | |
682 | int found; | |
683 | ||
684 | if (bitmap_is_midx(bitmap_git)) | |
685 | found = bsearch_midx(oid, bitmap_git->midx, result); | |
686 | else | |
687 | found = bsearch_pack(oid, bitmap_git->pack, result); | |
688 | ||
689 | return found; | |
690 | } | |
691 | ||
692 | /* | |
693 | * `bsearch_triplet_by_pos` function searches for the raw triplet | |
694 | * having commit position same as `commit_pos` and fills `triplet` | |
695 | * object from the raw triplet. Returns 1 on success and 0 on | |
696 | * failure. | |
697 | */ | |
698 | static int bitmap_bsearch_triplet_by_pos(uint32_t commit_pos, | |
699 | struct bitmap_index *bitmap_git, | |
700 | struct bitmap_lookup_table_triplet *triplet) | |
701 | { | |
702 | unsigned char *p = bsearch(&commit_pos, bitmap_git->table_lookup, bitmap_git->entry_count, | |
703 | BITMAP_LOOKUP_TABLE_TRIPLET_WIDTH, triplet_cmp); | |
704 | ||
705 | if (!p) | |
706 | return -1; | |
707 | ||
708 | return bitmap_lookup_table_get_triplet_by_pointer(triplet, p); | |
709 | } | |
710 | ||
711 | static struct stored_bitmap *lazy_bitmap_for_commit(struct bitmap_index *bitmap_git, | |
712 | struct commit *commit) | |
713 | { | |
714 | uint32_t commit_pos, xor_row; | |
715 | uint64_t offset; | |
716 | int flags; | |
717 | struct bitmap_lookup_table_triplet triplet; | |
718 | struct object_id *oid = &commit->object.oid; | |
719 | struct ewah_bitmap *bitmap; | |
720 | struct stored_bitmap *xor_bitmap = NULL; | |
721 | const int bitmap_header_size = 6; | |
722 | static struct bitmap_lookup_table_xor_item *xor_items = NULL; | |
723 | static size_t xor_items_nr = 0, xor_items_alloc = 0; | |
724 | static int is_corrupt = 0; | |
725 | int xor_flags; | |
726 | khiter_t hash_pos; | |
727 | struct bitmap_lookup_table_xor_item *xor_item; | |
728 | ||
729 | if (is_corrupt) | |
730 | return NULL; | |
731 | ||
732 | if (!bitmap_bsearch_pos(bitmap_git, oid, &commit_pos)) | |
733 | return NULL; | |
734 | ||
735 | if (bitmap_bsearch_triplet_by_pos(commit_pos, bitmap_git, &triplet) < 0) | |
736 | return NULL; | |
737 | ||
738 | xor_items_nr = 0; | |
739 | offset = triplet.offset; | |
740 | xor_row = triplet.xor_row; | |
741 | ||
742 | while (xor_row != 0xffffffff) { | |
743 | ALLOC_GROW(xor_items, xor_items_nr + 1, xor_items_alloc); | |
744 | ||
745 | if (xor_items_nr + 1 >= bitmap_git->entry_count) { | |
711340c7 | 746 | error(_("corrupt bitmap lookup table: xor chain exceeds entry count")); |
28cd7306 AC |
747 | goto corrupt; |
748 | } | |
749 | ||
750 | if (bitmap_lookup_table_get_triplet(bitmap_git, xor_row, &triplet) < 0) | |
751 | goto corrupt; | |
752 | ||
753 | xor_item = &xor_items[xor_items_nr]; | |
754 | xor_item->offset = triplet.offset; | |
755 | ||
756 | if (nth_bitmap_object_oid(bitmap_git, &xor_item->oid, triplet.commit_pos) < 0) { | |
757 | error(_("corrupt bitmap lookup table: commit index %u out of range"), | |
758 | triplet.commit_pos); | |
759 | goto corrupt; | |
760 | } | |
761 | ||
762 | hash_pos = kh_get_oid_map(bitmap_git->bitmaps, xor_item->oid); | |
763 | ||
764 | /* | |
765 | * If desired bitmap is already stored, we don't need | |
766 | * to iterate further. Because we know that bitmaps | |
767 | * that are needed to be parsed to parse this bitmap | |
768 | * has already been stored. So, assign this stored bitmap | |
769 | * to the xor_bitmap. | |
770 | */ | |
771 | if (hash_pos < kh_end(bitmap_git->bitmaps) && | |
772 | (xor_bitmap = kh_value(bitmap_git->bitmaps, hash_pos))) | |
773 | break; | |
774 | xor_items_nr++; | |
775 | xor_row = triplet.xor_row; | |
776 | } | |
777 | ||
778 | while (xor_items_nr) { | |
779 | xor_item = &xor_items[xor_items_nr - 1]; | |
780 | bitmap_git->map_pos = xor_item->offset; | |
781 | if (bitmap_git->map_size - bitmap_git->map_pos < bitmap_header_size) { | |
782 | error(_("corrupt ewah bitmap: truncated header for bitmap of commit \"%s\""), | |
783 | oid_to_hex(&xor_item->oid)); | |
784 | goto corrupt; | |
785 | } | |
786 | ||
787 | bitmap_git->map_pos += sizeof(uint32_t) + sizeof(uint8_t); | |
788 | xor_flags = read_u8(bitmap_git->map, &bitmap_git->map_pos); | |
789 | bitmap = read_bitmap_1(bitmap_git); | |
790 | ||
791 | if (!bitmap) | |
792 | goto corrupt; | |
793 | ||
794 | xor_bitmap = store_bitmap(bitmap_git, bitmap, &xor_item->oid, xor_bitmap, xor_flags); | |
795 | xor_items_nr--; | |
796 | } | |
797 | ||
798 | bitmap_git->map_pos = offset; | |
799 | if (bitmap_git->map_size - bitmap_git->map_pos < bitmap_header_size) { | |
800 | error(_("corrupt ewah bitmap: truncated header for bitmap of commit \"%s\""), | |
801 | oid_to_hex(oid)); | |
802 | goto corrupt; | |
803 | } | |
804 | ||
805 | /* | |
806 | * Don't bother reading the commit's index position or its xor | |
807 | * offset: | |
808 | * | |
809 | * - The commit's index position is irrelevant to us, since | |
810 | * load_bitmap_entries_v1 only uses it to learn the object | |
811 | * id which is used to compute the hashmap's key. We already | |
812 | * have an object id, so no need to look it up again. | |
813 | * | |
814 | * - The xor_offset is unusable for us, since it specifies how | |
815 | * many entries previous to ours we should look at. This | |
816 | * makes sense when reading the bitmaps sequentially (as in | |
817 | * load_bitmap_entries_v1()), since we can keep track of | |
818 | * each bitmap as we read them. | |
819 | * | |
820 | * But it can't work for us, since the bitmap's don't have a | |
821 | * fixed size. So we learn the position of the xor'd bitmap | |
822 | * from the commit table (and resolve it to a bitmap in the | |
823 | * above if-statement). | |
824 | * | |
825 | * Instead, we can skip ahead and immediately read the flags and | |
826 | * ewah bitmap. | |
827 | */ | |
828 | bitmap_git->map_pos += sizeof(uint32_t) + sizeof(uint8_t); | |
829 | flags = read_u8(bitmap_git->map, &bitmap_git->map_pos); | |
830 | bitmap = read_bitmap_1(bitmap_git); | |
831 | ||
832 | if (!bitmap) | |
833 | goto corrupt; | |
834 | ||
835 | return store_bitmap(bitmap_git, bitmap, oid, xor_bitmap, flags); | |
836 | ||
837 | corrupt: | |
838 | free(xor_items); | |
839 | is_corrupt = 1; | |
840 | return NULL; | |
841 | } | |
842 | ||
98c31f36 TB |
843 | struct ewah_bitmap *bitmap_for_commit(struct bitmap_index *bitmap_git, |
844 | struct commit *commit) | |
845 | { | |
846 | khiter_t hash_pos = kh_get_oid_map(bitmap_git->bitmaps, | |
847 | commit->object.oid); | |
28cd7306 AC |
848 | if (hash_pos >= kh_end(bitmap_git->bitmaps)) { |
849 | struct stored_bitmap *bitmap = NULL; | |
850 | if (!bitmap_git->table_lookup) | |
851 | return NULL; | |
852 | ||
89a1ab8f | 853 | /* this is a fairly hot codepath - no trace2_region please */ |
28cd7306 AC |
854 | /* NEEDSWORK: cache misses aren't recorded */ |
855 | bitmap = lazy_bitmap_for_commit(bitmap_git, commit); | |
28cd7306 AC |
856 | if (!bitmap) |
857 | return NULL; | |
858 | return lookup_stored_bitmap(bitmap); | |
859 | } | |
98c31f36 TB |
860 | return lookup_stored_bitmap(kh_value(bitmap_git->bitmaps, hash_pos)); |
861 | } | |
862 | ||
3ae5fa07 | 863 | static inline int bitmap_position_extended(struct bitmap_index *bitmap_git, |
3c771448 | 864 | const struct object_id *oid) |
fff42755 | 865 | { |
4ed43d16 | 866 | kh_oid_pos_t *positions = bitmap_git->ext_index.positions; |
3c771448 | 867 | khiter_t pos = kh_get_oid_pos(positions, *oid); |
fff42755 VM |
868 | |
869 | if (pos < kh_end(positions)) { | |
870 | int bitmap_pos = kh_value(positions, pos); | |
ed184620 | 871 | return bitmap_pos + bitmap_num_objects(bitmap_git); |
fff42755 VM |
872 | } |
873 | ||
874 | return -1; | |
875 | } | |
876 | ||
3ae5fa07 | 877 | static inline int bitmap_position_packfile(struct bitmap_index *bitmap_git, |
3c771448 | 878 | const struct object_id *oid) |
fff42755 | 879 | { |
57665086 | 880 | uint32_t pos; |
3c771448 | 881 | off_t offset = find_pack_entry_one(oid->hash, bitmap_git->pack); |
fff42755 VM |
882 | if (!offset) |
883 | return -1; | |
884 | ||
57665086 TB |
885 | if (offset_to_pack_pos(bitmap_git->pack, offset, &pos) < 0) |
886 | return -1; | |
887 | return pos; | |
fff42755 VM |
888 | } |
889 | ||
0f533c72 TB |
890 | static int bitmap_position_midx(struct bitmap_index *bitmap_git, |
891 | const struct object_id *oid) | |
892 | { | |
893 | uint32_t want, got; | |
894 | if (!bsearch_midx(oid, bitmap_git->midx, &want)) | |
895 | return -1; | |
896 | ||
897 | if (midx_to_pack_pos(bitmap_git->midx, want, &got) < 0) | |
898 | return -1; | |
899 | return got; | |
900 | } | |
901 | ||
3ae5fa07 | 902 | static int bitmap_position(struct bitmap_index *bitmap_git, |
3c771448 | 903 | const struct object_id *oid) |
fff42755 | 904 | { |
0f533c72 TB |
905 | int pos; |
906 | if (bitmap_is_midx(bitmap_git)) | |
907 | pos = bitmap_position_midx(bitmap_git, oid); | |
908 | else | |
909 | pos = bitmap_position_packfile(bitmap_git, oid); | |
3c771448 | 910 | return (pos >= 0) ? pos : bitmap_position_extended(bitmap_git, oid); |
fff42755 VM |
911 | } |
912 | ||
3ae5fa07 JT |
913 | static int ext_index_add_object(struct bitmap_index *bitmap_git, |
914 | struct object *object, const char *name) | |
fff42755 | 915 | { |
3ae5fa07 | 916 | struct eindex *eindex = &bitmap_git->ext_index; |
fff42755 VM |
917 | |
918 | khiter_t hash_pos; | |
919 | int hash_ret; | |
920 | int bitmap_pos; | |
921 | ||
3c771448 | 922 | hash_pos = kh_put_oid_pos(eindex->positions, object->oid, &hash_ret); |
fff42755 VM |
923 | if (hash_ret > 0) { |
924 | if (eindex->count >= eindex->alloc) { | |
925 | eindex->alloc = (eindex->alloc + 16) * 3 / 2; | |
2756ca43 RS |
926 | REALLOC_ARRAY(eindex->objects, eindex->alloc); |
927 | REALLOC_ARRAY(eindex->hashes, eindex->alloc); | |
fff42755 VM |
928 | } |
929 | ||
930 | bitmap_pos = eindex->count; | |
931 | eindex->objects[eindex->count] = object; | |
932 | eindex->hashes[eindex->count] = pack_name_hash(name); | |
933 | kh_value(eindex->positions, hash_pos) = bitmap_pos; | |
934 | eindex->count++; | |
935 | } else { | |
936 | bitmap_pos = kh_value(eindex->positions, hash_pos); | |
937 | } | |
938 | ||
ed184620 | 939 | return bitmap_pos + bitmap_num_objects(bitmap_git); |
fff42755 VM |
940 | } |
941 | ||
3ae5fa07 JT |
942 | struct bitmap_show_data { |
943 | struct bitmap_index *bitmap_git; | |
944 | struct bitmap *base; | |
945 | }; | |
946 | ||
947 | static void show_object(struct object *object, const char *name, void *data_) | |
fff42755 | 948 | { |
3ae5fa07 | 949 | struct bitmap_show_data *data = data_; |
fff42755 VM |
950 | int bitmap_pos; |
951 | ||
3c771448 | 952 | bitmap_pos = bitmap_position(data->bitmap_git, &object->oid); |
fff42755 | 953 | |
de1e67d0 | 954 | if (bitmap_pos < 0) |
3ae5fa07 JT |
955 | bitmap_pos = ext_index_add_object(data->bitmap_git, object, |
956 | name); | |
fff42755 | 957 | |
3ae5fa07 | 958 | bitmap_set(data->base, bitmap_pos); |
fff42755 VM |
959 | } |
960 | ||
c50dca2a JK |
961 | static void show_commit(struct commit *commit UNUSED, |
962 | void *data UNUSED) | |
fff42755 VM |
963 | { |
964 | } | |
965 | ||
3ae5fa07 JT |
966 | static int add_to_include_set(struct bitmap_index *bitmap_git, |
967 | struct include_data *data, | |
98c31f36 | 968 | struct commit *commit, |
fff42755 VM |
969 | int bitmap_pos) |
970 | { | |
98c31f36 | 971 | struct ewah_bitmap *partial; |
fff42755 VM |
972 | |
973 | if (data->seen && bitmap_get(data->seen, bitmap_pos)) | |
974 | return 0; | |
975 | ||
976 | if (bitmap_get(data->base, bitmap_pos)) | |
977 | return 0; | |
978 | ||
98c31f36 TB |
979 | partial = bitmap_for_commit(bitmap_git, commit); |
980 | if (partial) { | |
981 | bitmap_or_ewah(data->base, partial); | |
fff42755 VM |
982 | return 0; |
983 | } | |
984 | ||
985 | bitmap_set(data->base, bitmap_pos); | |
986 | return 1; | |
987 | } | |
988 | ||
989 | static int should_include(struct commit *commit, void *_data) | |
990 | { | |
991 | struct include_data *data = _data; | |
992 | int bitmap_pos; | |
993 | ||
3c771448 | 994 | bitmap_pos = bitmap_position(data->bitmap_git, &commit->object.oid); |
fff42755 | 995 | if (bitmap_pos < 0) |
3ae5fa07 JT |
996 | bitmap_pos = ext_index_add_object(data->bitmap_git, |
997 | (struct object *)commit, | |
998 | NULL); | |
fff42755 | 999 | |
98c31f36 | 1000 | if (!add_to_include_set(data->bitmap_git, data, commit, bitmap_pos)) { |
fff42755 VM |
1001 | struct commit_list *parent = commit->parents; |
1002 | ||
1003 | while (parent) { | |
1004 | parent->item->object.flags |= SEEN; | |
1005 | parent = parent->next; | |
1006 | } | |
1007 | ||
1008 | return 0; | |
1009 | } | |
1010 | ||
1011 | return 1; | |
1012 | } | |
1013 | ||
aa9ad6fe JK |
1014 | static int should_include_obj(struct object *obj, void *_data) |
1015 | { | |
1016 | struct include_data *data = _data; | |
1017 | int bitmap_pos; | |
1018 | ||
1019 | bitmap_pos = bitmap_position(data->bitmap_git, &obj->oid); | |
1020 | if (bitmap_pos < 0) | |
1021 | return 1; | |
1022 | if ((data->seen && bitmap_get(data->seen, bitmap_pos)) || | |
1023 | bitmap_get(data->base, bitmap_pos)) { | |
1024 | obj->flags |= SEEN; | |
1025 | return 0; | |
1026 | } | |
1027 | return 1; | |
1028 | } | |
1029 | ||
83578051 TB |
1030 | static int add_commit_to_bitmap(struct bitmap_index *bitmap_git, |
1031 | struct bitmap **base, | |
1032 | struct commit *commit) | |
1033 | { | |
1034 | struct ewah_bitmap *or_with = bitmap_for_commit(bitmap_git, commit); | |
1035 | ||
1036 | if (!or_with) | |
1037 | return 0; | |
1038 | ||
afe8a907 | 1039 | if (!*base) |
83578051 TB |
1040 | *base = ewah_to_bitmap(or_with); |
1041 | else | |
1042 | bitmap_or_ewah(*base, or_with); | |
1043 | ||
1044 | return 1; | |
1045 | } | |
1046 | ||
47ff853f TB |
1047 | static struct bitmap *fill_in_bitmap(struct bitmap_index *bitmap_git, |
1048 | struct rev_info *revs, | |
1049 | struct bitmap *base, | |
1050 | struct bitmap *seen) | |
1051 | { | |
1052 | struct include_data incdata; | |
1053 | struct bitmap_show_data show_data; | |
1054 | ||
1055 | if (!base) | |
1056 | base = bitmap_new(); | |
1057 | ||
1058 | incdata.bitmap_git = bitmap_git; | |
1059 | incdata.base = base; | |
1060 | incdata.seen = seen; | |
1061 | ||
1062 | revs->include_check = should_include; | |
1063 | revs->include_check_obj = should_include_obj; | |
1064 | revs->include_check_data = &incdata; | |
1065 | ||
1066 | if (prepare_revision_walk(revs)) | |
1067 | die(_("revision walk setup failed")); | |
1068 | ||
1069 | show_data.bitmap_git = bitmap_git; | |
1070 | show_data.base = base; | |
1071 | ||
1072 | traverse_commit_list(revs, show_commit, show_object, &show_data); | |
1073 | ||
1074 | revs->include_check = NULL; | |
1075 | revs->include_check_obj = NULL; | |
1076 | revs->include_check_data = NULL; | |
1077 | ||
1078 | return base; | |
1079 | } | |
1080 | ||
b0afdce5 TB |
1081 | struct bitmap_boundary_cb { |
1082 | struct bitmap_index *bitmap_git; | |
1083 | struct bitmap *base; | |
1084 | ||
1085 | struct object_array boundary; | |
1086 | }; | |
1087 | ||
1088 | static void show_boundary_commit(struct commit *commit, void *_data) | |
1089 | { | |
1090 | struct bitmap_boundary_cb *data = _data; | |
1091 | ||
1092 | if (commit->object.flags & BOUNDARY) | |
1093 | add_object_array(&commit->object, "", &data->boundary); | |
1094 | ||
1095 | if (commit->object.flags & UNINTERESTING) { | |
1096 | if (bitmap_walk_contains(data->bitmap_git, data->base, | |
1097 | &commit->object.oid)) | |
1098 | return; | |
1099 | ||
1100 | add_commit_to_bitmap(data->bitmap_git, &data->base, commit); | |
1101 | } | |
1102 | } | |
1103 | ||
d79b9f7c JK |
1104 | static void show_boundary_object(struct object *object UNUSED, |
1105 | const char *name UNUSED, | |
1106 | void *data UNUSED) | |
b0afdce5 TB |
1107 | { |
1108 | BUG("should not be called"); | |
1109 | } | |
1110 | ||
1111 | static struct bitmap *find_boundary_objects(struct bitmap_index *bitmap_git, | |
1112 | struct rev_info *revs, | |
1113 | struct object_list *roots) | |
1114 | { | |
1115 | struct bitmap_boundary_cb cb; | |
1116 | struct object_list *root; | |
1117 | unsigned int i; | |
1118 | unsigned int tmp_blobs, tmp_trees, tmp_tags; | |
1119 | int any_missing = 0; | |
1120 | ||
1121 | cb.bitmap_git = bitmap_git; | |
1122 | cb.base = bitmap_new(); | |
1123 | object_array_init(&cb.boundary); | |
1124 | ||
1125 | revs->ignore_missing_links = 1; | |
1126 | ||
1127 | /* | |
1128 | * OR in any existing reachability bitmaps among `roots` into | |
1129 | * `cb.base`. | |
1130 | */ | |
1131 | for (root = roots; root; root = root->next) { | |
1132 | struct object *object = root->item; | |
1133 | if (object->type != OBJ_COMMIT || | |
1134 | bitmap_walk_contains(bitmap_git, cb.base, &object->oid)) | |
1135 | continue; | |
1136 | ||
1137 | if (add_commit_to_bitmap(bitmap_git, &cb.base, | |
1138 | (struct commit *)object)) | |
1139 | continue; | |
1140 | ||
1141 | any_missing = 1; | |
1142 | } | |
1143 | ||
1144 | if (!any_missing) | |
1145 | goto cleanup; | |
1146 | ||
1147 | tmp_blobs = revs->blob_objects; | |
1148 | tmp_trees = revs->tree_objects; | |
1149 | tmp_tags = revs->blob_objects; | |
1150 | revs->blob_objects = 0; | |
1151 | revs->tree_objects = 0; | |
1152 | revs->tag_objects = 0; | |
1153 | ||
1154 | /* | |
1155 | * We didn't have complete coverage of the roots. First setup a | |
1156 | * revision walk to (a) OR in any bitmaps that are UNINTERESTING | |
1157 | * between the tips and boundary, and (b) record the boundary. | |
1158 | */ | |
1159 | trace2_region_enter("pack-bitmap", "boundary-prepare", the_repository); | |
1160 | if (prepare_revision_walk(revs)) | |
1161 | die("revision walk setup failed"); | |
1162 | trace2_region_leave("pack-bitmap", "boundary-prepare", the_repository); | |
1163 | ||
1164 | trace2_region_enter("pack-bitmap", "boundary-traverse", the_repository); | |
1165 | revs->boundary = 1; | |
1166 | traverse_commit_list_filtered(revs, | |
1167 | show_boundary_commit, | |
1168 | show_boundary_object, | |
1169 | &cb, NULL); | |
1170 | revs->boundary = 0; | |
1171 | trace2_region_leave("pack-bitmap", "boundary-traverse", the_repository); | |
1172 | ||
1173 | revs->blob_objects = tmp_blobs; | |
1174 | revs->tree_objects = tmp_trees; | |
1175 | revs->tag_objects = tmp_tags; | |
1176 | ||
1177 | reset_revision_walk(); | |
1178 | clear_object_flags(UNINTERESTING); | |
1179 | ||
1180 | /* | |
1181 | * Then add the boundary commit(s) as fill-in traversal tips. | |
1182 | */ | |
1183 | trace2_region_enter("pack-bitmap", "boundary-fill-in", the_repository); | |
1184 | for (i = 0; i < cb.boundary.nr; i++) { | |
1185 | struct object *obj = cb.boundary.objects[i].item; | |
1186 | if (bitmap_walk_contains(bitmap_git, cb.base, &obj->oid)) | |
1187 | obj->flags |= SEEN; | |
1188 | else | |
1189 | add_pending_object(revs, obj, ""); | |
1190 | } | |
1191 | if (revs->pending.nr) | |
1192 | cb.base = fill_in_bitmap(bitmap_git, revs, cb.base, NULL); | |
1193 | trace2_region_leave("pack-bitmap", "boundary-fill-in", the_repository); | |
1194 | ||
1195 | cleanup: | |
1196 | object_array_clear(&cb.boundary); | |
1197 | revs->ignore_missing_links = 0; | |
1198 | ||
1199 | return cb.base; | |
1200 | } | |
1201 | ||
3ae5fa07 JT |
1202 | static struct bitmap *find_objects(struct bitmap_index *bitmap_git, |
1203 | struct rev_info *revs, | |
fff42755 | 1204 | struct object_list *roots, |
09d4a79e | 1205 | struct bitmap *seen) |
fff42755 VM |
1206 | { |
1207 | struct bitmap *base = NULL; | |
1208 | int needs_walk = 0; | |
1209 | ||
1210 | struct object_list *not_mapped = NULL; | |
1211 | ||
1212 | /* | |
1213 | * Go through all the roots for the walk. The ones that have bitmaps | |
1214 | * on the bitmap index will be `or`ed together to form an initial | |
1215 | * global reachability analysis. | |
1216 | * | |
1217 | * The ones without bitmaps in the index will be stored in the | |
1218 | * `not_mapped_list` for further processing. | |
1219 | */ | |
1220 | while (roots) { | |
1221 | struct object *object = roots->item; | |
1222 | roots = roots->next; | |
1223 | ||
83578051 TB |
1224 | if (object->type == OBJ_COMMIT && |
1225 | add_commit_to_bitmap(bitmap_git, &base, (struct commit *)object)) { | |
1226 | object->flags |= SEEN; | |
1227 | continue; | |
fff42755 VM |
1228 | } |
1229 | ||
1230 | object_list_insert(object, ¬_mapped); | |
1231 | } | |
1232 | ||
1233 | /* | |
1234 | * Best case scenario: We found bitmaps for all the roots, | |
1235 | * so the resulting `or` bitmap has the full reachability analysis | |
1236 | */ | |
afe8a907 | 1237 | if (!not_mapped) |
fff42755 VM |
1238 | return base; |
1239 | ||
1240 | roots = not_mapped; | |
1241 | ||
1242 | /* | |
1243 | * Let's iterate through all the roots that don't have bitmaps to | |
1244 | * check if we can determine them to be reachable from the existing | |
1245 | * global bitmap. | |
1246 | * | |
1247 | * If we cannot find them in the existing global bitmap, we'll need | |
1248 | * to push them to an actual walk and run it until we can confirm | |
1249 | * they are reachable | |
1250 | */ | |
1251 | while (roots) { | |
1252 | struct object *object = roots->item; | |
1253 | int pos; | |
1254 | ||
1255 | roots = roots->next; | |
3c771448 | 1256 | pos = bitmap_position(bitmap_git, &object->oid); |
fff42755 VM |
1257 | |
1258 | if (pos < 0 || base == NULL || !bitmap_get(base, pos)) { | |
1259 | object->flags &= ~UNINTERESTING; | |
1260 | add_pending_object(revs, object, ""); | |
1261 | needs_walk = 1; | |
1262 | } else { | |
1263 | object->flags |= SEEN; | |
1264 | } | |
1265 | } | |
1266 | ||
1267 | if (needs_walk) { | |
b0afdce5 TB |
1268 | /* |
1269 | * This fill-in traversal may walk over some objects | |
1270 | * again, since we have already traversed in order to | |
1271 | * find the boundary. | |
1272 | * | |
1273 | * But this extra walk should be extremely cheap, since | |
1274 | * all commit objects are loaded into memory, and | |
1275 | * because we skip walking to parents that are | |
1276 | * UNINTERESTING, since it will be marked in the haves | |
1277 | * bitmap already (or it has an on-disk bitmap, since | |
1278 | * OR-ing it in covers all of its ancestors). | |
1279 | */ | |
47ff853f | 1280 | base = fill_in_bitmap(bitmap_git, revs, base, seen); |
fff42755 VM |
1281 | } |
1282 | ||
1283 | return base; | |
1284 | } | |
1285 | ||
3ae5fa07 | 1286 | static void show_extended_objects(struct bitmap_index *bitmap_git, |
4eb707eb | 1287 | struct rev_info *revs, |
fff42755 VM |
1288 | show_reachable_fn show_reach) |
1289 | { | |
3ae5fa07 JT |
1290 | struct bitmap *objects = bitmap_git->result; |
1291 | struct eindex *eindex = &bitmap_git->ext_index; | |
fff42755 VM |
1292 | uint32_t i; |
1293 | ||
1294 | for (i = 0; i < eindex->count; ++i) { | |
1295 | struct object *obj; | |
1296 | ||
0948c501 | 1297 | if (!bitmap_get(objects, st_add(bitmap_num_objects(bitmap_git), i))) |
fff42755 VM |
1298 | continue; |
1299 | ||
1300 | obj = eindex->objects[i]; | |
4eb707eb JK |
1301 | if ((obj->type == OBJ_BLOB && !revs->blob_objects) || |
1302 | (obj->type == OBJ_TREE && !revs->tree_objects) || | |
1303 | (obj->type == OBJ_TAG && !revs->tag_objects)) | |
1304 | continue; | |
1305 | ||
20664967 | 1306 | show_reach(&obj->oid, obj->type, 0, eindex->hashes[i], NULL, 0); |
fff42755 VM |
1307 | } |
1308 | } | |
1309 | ||
551cf8b6 JK |
1310 | static void init_type_iterator(struct ewah_iterator *it, |
1311 | struct bitmap_index *bitmap_git, | |
1312 | enum object_type type) | |
1313 | { | |
1314 | switch (type) { | |
1315 | case OBJ_COMMIT: | |
1316 | ewah_iterator_init(it, bitmap_git->commits); | |
1317 | break; | |
1318 | ||
1319 | case OBJ_TREE: | |
1320 | ewah_iterator_init(it, bitmap_git->trees); | |
1321 | break; | |
1322 | ||
1323 | case OBJ_BLOB: | |
1324 | ewah_iterator_init(it, bitmap_git->blobs); | |
1325 | break; | |
1326 | ||
1327 | case OBJ_TAG: | |
1328 | ewah_iterator_init(it, bitmap_git->tags); | |
1329 | break; | |
1330 | ||
1331 | default: | |
1332 | BUG("object type %d not stored by bitmap type index", type); | |
1333 | break; | |
1334 | } | |
1335 | } | |
1336 | ||
fff42755 | 1337 | static void show_objects_for_type( |
3ae5fa07 | 1338 | struct bitmap_index *bitmap_git, |
fff42755 VM |
1339 | enum object_type object_type, |
1340 | show_reachable_fn show_reach) | |
1341 | { | |
d2ea0310 | 1342 | size_t i = 0; |
fff42755 VM |
1343 | uint32_t offset; |
1344 | ||
1345 | struct ewah_iterator it; | |
1346 | eword_t filter; | |
1347 | ||
3ae5fa07 JT |
1348 | struct bitmap *objects = bitmap_git->result; |
1349 | ||
551cf8b6 | 1350 | init_type_iterator(&it, bitmap_git, object_type); |
fff42755 | 1351 | |
d2ea0310 JK |
1352 | for (i = 0; i < objects->word_alloc && |
1353 | ewah_iterator_next(&filter, &it); i++) { | |
fff42755 | 1354 | eword_t word = objects->words[i] & filter; |
d2ea0310 JK |
1355 | size_t pos = (i * BITS_IN_EWORD); |
1356 | ||
1357 | if (!word) | |
1358 | continue; | |
fff42755 | 1359 | |
34b935c0 | 1360 | for (offset = 0; offset < BITS_IN_EWORD; ++offset) { |
0f533c72 | 1361 | struct packed_git *pack; |
20664967 | 1362 | struct object_id oid; |
cf98f2e8 TB |
1363 | uint32_t hash = 0, index_pos; |
1364 | off_t ofs; | |
fff42755 VM |
1365 | |
1366 | if ((word >> offset) == 0) | |
1367 | break; | |
1368 | ||
1369 | offset += ewah_bit_ctz64(word >> offset); | |
1370 | ||
0f533c72 TB |
1371 | if (bitmap_is_midx(bitmap_git)) { |
1372 | struct multi_pack_index *m = bitmap_git->midx; | |
1373 | uint32_t pack_id; | |
1374 | ||
1375 | index_pos = pack_pos_to_midx(m, pos + offset); | |
1376 | ofs = nth_midxed_offset(m, index_pos); | |
1377 | nth_midxed_object_oid(&oid, m, index_pos); | |
1378 | ||
1379 | pack_id = nth_midxed_pack_int_id(m, index_pos); | |
1380 | pack = bitmap_git->midx->packs[pack_id]; | |
1381 | } else { | |
1382 | index_pos = pack_pos_to_index(bitmap_git->pack, pos + offset); | |
1383 | ofs = pack_pos_to_offset(bitmap_git->pack, pos + offset); | |
1384 | nth_bitmap_object_oid(bitmap_git, &oid, index_pos); | |
1385 | ||
1386 | pack = bitmap_git->pack; | |
1387 | } | |
fff42755 | 1388 | |
3ae5fa07 | 1389 | if (bitmap_git->hashes) |
cf98f2e8 | 1390 | hash = get_be32(bitmap_git->hashes + index_pos); |
ae4f07fb | 1391 | |
0f533c72 | 1392 | show_reach(&oid, object_type, 0, hash, pack, ofs); |
fff42755 | 1393 | } |
fff42755 VM |
1394 | } |
1395 | } | |
1396 | ||
3ae5fa07 JT |
1397 | static int in_bitmapped_pack(struct bitmap_index *bitmap_git, |
1398 | struct object_list *roots) | |
fff42755 VM |
1399 | { |
1400 | while (roots) { | |
1401 | struct object *object = roots->item; | |
1402 | roots = roots->next; | |
1403 | ||
0f533c72 TB |
1404 | if (bitmap_is_midx(bitmap_git)) { |
1405 | if (bsearch_midx(&object->oid, bitmap_git->midx, NULL)) | |
1406 | return 1; | |
1407 | } else { | |
1408 | if (find_pack_entry_one(object->oid.hash, bitmap_git->pack) > 0) | |
1409 | return 1; | |
1410 | } | |
fff42755 VM |
1411 | } |
1412 | ||
1413 | return 0; | |
1414 | } | |
1415 | ||
856e12c1 TB |
1416 | static struct bitmap *find_tip_objects(struct bitmap_index *bitmap_git, |
1417 | struct object_list *tip_objects, | |
1418 | enum object_type type) | |
4f3bd560 JK |
1419 | { |
1420 | struct bitmap *result = bitmap_new(); | |
1421 | struct object_list *p; | |
1422 | ||
1423 | for (p = tip_objects; p; p = p->next) { | |
1424 | int pos; | |
1425 | ||
856e12c1 | 1426 | if (p->item->type != type) |
4f3bd560 JK |
1427 | continue; |
1428 | ||
1429 | pos = bitmap_position(bitmap_git, &p->item->oid); | |
1430 | if (pos < 0) | |
1431 | continue; | |
1432 | ||
1433 | bitmap_set(result, pos); | |
1434 | } | |
1435 | ||
1436 | return result; | |
1437 | } | |
1438 | ||
856e12c1 TB |
1439 | static void filter_bitmap_exclude_type(struct bitmap_index *bitmap_git, |
1440 | struct object_list *tip_objects, | |
1441 | struct bitmap *to_filter, | |
1442 | enum object_type type) | |
4f3bd560 JK |
1443 | { |
1444 | struct eindex *eindex = &bitmap_git->ext_index; | |
1445 | struct bitmap *tips; | |
1446 | struct ewah_iterator it; | |
1447 | eword_t mask; | |
1448 | uint32_t i; | |
1449 | ||
1450 | /* | |
1451 | * The non-bitmap version of this filter never removes | |
856e12c1 | 1452 | * objects which the other side specifically asked for, |
4f3bd560 JK |
1453 | * so we must match that behavior. |
1454 | */ | |
856e12c1 | 1455 | tips = find_tip_objects(bitmap_git, tip_objects, type); |
4f3bd560 JK |
1456 | |
1457 | /* | |
ddcb189d TB |
1458 | * We can use the type-level bitmap for 'type' to work in whole |
1459 | * words for the objects that are actually in the bitmapped | |
1460 | * packfile. | |
4f3bd560 | 1461 | */ |
856e12c1 | 1462 | for (i = 0, init_type_iterator(&it, bitmap_git, type); |
4f3bd560 JK |
1463 | i < to_filter->word_alloc && ewah_iterator_next(&mask, &it); |
1464 | i++) { | |
1465 | if (i < tips->word_alloc) | |
1466 | mask &= ~tips->words[i]; | |
1467 | to_filter->words[i] &= ~mask; | |
1468 | } | |
1469 | ||
1470 | /* | |
ddcb189d TB |
1471 | * Clear any objects that weren't in the packfile (and so would |
1472 | * not have been caught by the loop above. We'll have to check | |
1473 | * them individually. | |
4f3bd560 JK |
1474 | */ |
1475 | for (i = 0; i < eindex->count; i++) { | |
0948c501 | 1476 | size_t pos = st_add(i, bitmap_num_objects(bitmap_git)); |
856e12c1 | 1477 | if (eindex->objects[i]->type == type && |
4f3bd560 JK |
1478 | bitmap_get(to_filter, pos) && |
1479 | !bitmap_get(tips, pos)) | |
1480 | bitmap_unset(to_filter, pos); | |
1481 | } | |
1482 | ||
1483 | bitmap_free(tips); | |
1484 | } | |
1485 | ||
856e12c1 TB |
1486 | static void filter_bitmap_blob_none(struct bitmap_index *bitmap_git, |
1487 | struct object_list *tip_objects, | |
1488 | struct bitmap *to_filter) | |
1489 | { | |
1490 | filter_bitmap_exclude_type(bitmap_git, tip_objects, to_filter, | |
1491 | OBJ_BLOB); | |
1492 | } | |
1493 | ||
84243da1 JK |
1494 | static unsigned long get_size_by_pos(struct bitmap_index *bitmap_git, |
1495 | uint32_t pos) | |
1496 | { | |
84243da1 JK |
1497 | unsigned long size; |
1498 | struct object_info oi = OBJECT_INFO_INIT; | |
1499 | ||
1500 | oi.sizep = &size; | |
1501 | ||
ed184620 | 1502 | if (pos < bitmap_num_objects(bitmap_git)) { |
0f533c72 TB |
1503 | struct packed_git *pack; |
1504 | off_t ofs; | |
1505 | ||
1506 | if (bitmap_is_midx(bitmap_git)) { | |
1507 | uint32_t midx_pos = pack_pos_to_midx(bitmap_git->midx, pos); | |
1508 | uint32_t pack_id = nth_midxed_pack_int_id(bitmap_git->midx, midx_pos); | |
1509 | ||
1510 | pack = bitmap_git->midx->packs[pack_id]; | |
1511 | ofs = nth_midxed_offset(bitmap_git->midx, midx_pos); | |
1512 | } else { | |
1513 | pack = bitmap_git->pack; | |
1514 | ofs = pack_pos_to_offset(pack, pos); | |
1515 | } | |
1516 | ||
a78a9032 | 1517 | if (packed_object_info(the_repository, pack, ofs, &oi) < 0) { |
84243da1 | 1518 | struct object_id oid; |
6b4277e6 TB |
1519 | nth_bitmap_object_oid(bitmap_git, &oid, |
1520 | pack_pos_to_index(pack, pos)); | |
84243da1 JK |
1521 | die(_("unable to get size of %s"), oid_to_hex(&oid)); |
1522 | } | |
1523 | } else { | |
1524 | struct eindex *eindex = &bitmap_git->ext_index; | |
ed184620 | 1525 | struct object *obj = eindex->objects[pos - bitmap_num_objects(bitmap_git)]; |
84243da1 JK |
1526 | if (oid_object_info_extended(the_repository, &obj->oid, &oi, 0) < 0) |
1527 | die(_("unable to get size of %s"), oid_to_hex(&obj->oid)); | |
1528 | } | |
1529 | ||
1530 | return size; | |
1531 | } | |
1532 | ||
1533 | static void filter_bitmap_blob_limit(struct bitmap_index *bitmap_git, | |
1534 | struct object_list *tip_objects, | |
1535 | struct bitmap *to_filter, | |
1536 | unsigned long limit) | |
1537 | { | |
1538 | struct eindex *eindex = &bitmap_git->ext_index; | |
1539 | struct bitmap *tips; | |
1540 | struct ewah_iterator it; | |
1541 | eword_t mask; | |
1542 | uint32_t i; | |
1543 | ||
856e12c1 | 1544 | tips = find_tip_objects(bitmap_git, tip_objects, OBJ_BLOB); |
84243da1 JK |
1545 | |
1546 | for (i = 0, init_type_iterator(&it, bitmap_git, OBJ_BLOB); | |
1547 | i < to_filter->word_alloc && ewah_iterator_next(&mask, &it); | |
1548 | i++) { | |
1549 | eword_t word = to_filter->words[i] & mask; | |
1550 | unsigned offset; | |
1551 | ||
1552 | for (offset = 0; offset < BITS_IN_EWORD; offset++) { | |
1553 | uint32_t pos; | |
1554 | ||
1555 | if ((word >> offset) == 0) | |
1556 | break; | |
1557 | offset += ewah_bit_ctz64(word >> offset); | |
1558 | pos = i * BITS_IN_EWORD + offset; | |
1559 | ||
1560 | if (!bitmap_get(tips, pos) && | |
1561 | get_size_by_pos(bitmap_git, pos) >= limit) | |
1562 | bitmap_unset(to_filter, pos); | |
1563 | } | |
1564 | } | |
1565 | ||
1566 | for (i = 0; i < eindex->count; i++) { | |
0948c501 | 1567 | size_t pos = st_add(i, bitmap_num_objects(bitmap_git)); |
84243da1 JK |
1568 | if (eindex->objects[i]->type == OBJ_BLOB && |
1569 | bitmap_get(to_filter, pos) && | |
1570 | !bitmap_get(tips, pos) && | |
1571 | get_size_by_pos(bitmap_git, pos) >= limit) | |
1572 | bitmap_unset(to_filter, pos); | |
1573 | } | |
1574 | ||
1575 | bitmap_free(tips); | |
1576 | } | |
1577 | ||
b0a8d482 TB |
1578 | static void filter_bitmap_tree_depth(struct bitmap_index *bitmap_git, |
1579 | struct object_list *tip_objects, | |
1580 | struct bitmap *to_filter, | |
1581 | unsigned long limit) | |
1582 | { | |
1583 | if (limit) | |
1584 | BUG("filter_bitmap_tree_depth given non-zero limit"); | |
1585 | ||
1586 | filter_bitmap_exclude_type(bitmap_git, tip_objects, to_filter, | |
1587 | OBJ_TREE); | |
1588 | filter_bitmap_exclude_type(bitmap_git, tip_objects, to_filter, | |
1589 | OBJ_BLOB); | |
1590 | } | |
1591 | ||
7ab6aafa PS |
1592 | static void filter_bitmap_object_type(struct bitmap_index *bitmap_git, |
1593 | struct object_list *tip_objects, | |
1594 | struct bitmap *to_filter, | |
1595 | enum object_type object_type) | |
1596 | { | |
1597 | if (object_type < OBJ_COMMIT || object_type > OBJ_TAG) | |
1598 | BUG("filter_bitmap_object_type given invalid object"); | |
1599 | ||
1600 | if (object_type != OBJ_TAG) | |
1601 | filter_bitmap_exclude_type(bitmap_git, tip_objects, to_filter, OBJ_TAG); | |
1602 | if (object_type != OBJ_COMMIT) | |
1603 | filter_bitmap_exclude_type(bitmap_git, tip_objects, to_filter, OBJ_COMMIT); | |
1604 | if (object_type != OBJ_TREE) | |
1605 | filter_bitmap_exclude_type(bitmap_git, tip_objects, to_filter, OBJ_TREE); | |
1606 | if (object_type != OBJ_BLOB) | |
1607 | filter_bitmap_exclude_type(bitmap_git, tip_objects, to_filter, OBJ_BLOB); | |
1608 | } | |
1609 | ||
6663ae0a JK |
1610 | static int filter_bitmap(struct bitmap_index *bitmap_git, |
1611 | struct object_list *tip_objects, | |
1612 | struct bitmap *to_filter, | |
1613 | struct list_objects_filter_options *filter) | |
1614 | { | |
1615 | if (!filter || filter->choice == LOFC_DISABLED) | |
1616 | return 0; | |
1617 | ||
4f3bd560 JK |
1618 | if (filter->choice == LOFC_BLOB_NONE) { |
1619 | if (bitmap_git) | |
1620 | filter_bitmap_blob_none(bitmap_git, tip_objects, | |
1621 | to_filter); | |
1622 | return 0; | |
1623 | } | |
1624 | ||
84243da1 JK |
1625 | if (filter->choice == LOFC_BLOB_LIMIT) { |
1626 | if (bitmap_git) | |
1627 | filter_bitmap_blob_limit(bitmap_git, tip_objects, | |
1628 | to_filter, | |
1629 | filter->blob_limit_value); | |
1630 | return 0; | |
1631 | } | |
1632 | ||
b0a8d482 TB |
1633 | if (filter->choice == LOFC_TREE_DEPTH && |
1634 | filter->tree_exclude_depth == 0) { | |
1635 | if (bitmap_git) | |
1636 | filter_bitmap_tree_depth(bitmap_git, tip_objects, | |
1637 | to_filter, | |
1638 | filter->tree_exclude_depth); | |
1639 | return 0; | |
1640 | } | |
1641 | ||
7ab6aafa PS |
1642 | if (filter->choice == LOFC_OBJECT_TYPE) { |
1643 | if (bitmap_git) | |
1644 | filter_bitmap_object_type(bitmap_git, tip_objects, | |
1645 | to_filter, | |
1646 | filter->object_type); | |
1647 | return 0; | |
1648 | } | |
1649 | ||
169a15eb PS |
1650 | if (filter->choice == LOFC_COMBINE) { |
1651 | int i; | |
1652 | for (i = 0; i < filter->sub_nr; i++) { | |
1653 | if (filter_bitmap(bitmap_git, tip_objects, to_filter, | |
1654 | &filter->sub[i]) < 0) | |
1655 | return -1; | |
1656 | } | |
1657 | return 0; | |
1658 | } | |
1659 | ||
6663ae0a JK |
1660 | /* filter choice not handled */ |
1661 | return -1; | |
1662 | } | |
1663 | ||
1664 | static int can_filter_bitmap(struct list_objects_filter_options *filter) | |
1665 | { | |
1666 | return !filter_bitmap(NULL, NULL, NULL, filter); | |
1667 | } | |
1668 | ||
1669 | struct bitmap_index *prepare_bitmap_walk(struct rev_info *revs, | |
9cf68b27 | 1670 | int filter_provided_objects) |
fff42755 VM |
1671 | { |
1672 | unsigned int i; | |
b0afdce5 | 1673 | int use_boundary_traversal; |
fff42755 VM |
1674 | |
1675 | struct object_list *wants = NULL; | |
1676 | struct object_list *haves = NULL; | |
1677 | ||
1678 | struct bitmap *wants_bitmap = NULL; | |
1679 | struct bitmap *haves_bitmap = NULL; | |
1680 | ||
d90fe06e JK |
1681 | struct bitmap_index *bitmap_git; |
1682 | ||
1683 | /* | |
1684 | * We can't do pathspec limiting with bitmaps, because we don't know | |
1685 | * which commits are associated with which object changes (let alone | |
1686 | * even which objects are associated with which paths). | |
1687 | */ | |
1688 | if (revs->prune) | |
1689 | return NULL; | |
1690 | ||
09d4a79e | 1691 | if (!can_filter_bitmap(&revs->filter)) |
6663ae0a JK |
1692 | return NULL; |
1693 | ||
3ae5fa07 JT |
1694 | /* try to open a bitmapped pack, but don't parse it yet |
1695 | * because we may not need to use it */ | |
ca56dadb | 1696 | CALLOC_ARRAY(bitmap_git, 1); |
0f533c72 | 1697 | if (open_bitmap(revs->repo, bitmap_git) < 0) |
f3c23db2 | 1698 | goto cleanup; |
fff42755 | 1699 | |
4d01a7fa MÅ |
1700 | for (i = 0; i < revs->pending.nr; ++i) { |
1701 | struct object *object = revs->pending.objects[i].item; | |
fff42755 VM |
1702 | |
1703 | if (object->type == OBJ_NONE) | |
c251c83d | 1704 | parse_object_or_die(&object->oid, NULL); |
fff42755 VM |
1705 | |
1706 | while (object->type == OBJ_TAG) { | |
1707 | struct tag *tag = (struct tag *) object; | |
1708 | ||
1709 | if (object->flags & UNINTERESTING) | |
1710 | object_list_insert(object, &haves); | |
1711 | else | |
1712 | object_list_insert(object, &wants); | |
1713 | ||
dad3f060 | 1714 | object = parse_object_or_die(get_tagged_oid(tag), NULL); |
540cdc11 | 1715 | object->flags |= (tag->object.flags & UNINTERESTING); |
fff42755 VM |
1716 | } |
1717 | ||
1718 | if (object->flags & UNINTERESTING) | |
1719 | object_list_insert(object, &haves); | |
1720 | else | |
1721 | object_list_insert(object, &wants); | |
1722 | } | |
1723 | ||
b0afdce5 TB |
1724 | use_boundary_traversal = git_env_bool(GIT_TEST_PACK_USE_BITMAP_BOUNDARY_TRAVERSAL, -1); |
1725 | if (use_boundary_traversal < 0) { | |
1726 | prepare_repo_settings(revs->repo); | |
1727 | use_boundary_traversal = revs->repo->settings.pack_use_bitmap_boundary_traversal; | |
1728 | } | |
1729 | ||
1730 | if (!use_boundary_traversal) { | |
1731 | /* | |
1732 | * if we have a HAVES list, but none of those haves is contained | |
1733 | * in the packfile that has a bitmap, we don't have anything to | |
1734 | * optimize here | |
1735 | */ | |
1736 | if (haves && !in_bitmapped_pack(bitmap_git, haves)) | |
1737 | goto cleanup; | |
1738 | } | |
fff42755 VM |
1739 | |
1740 | /* if we don't want anything, we're done here */ | |
1741 | if (!wants) | |
f3c23db2 | 1742 | goto cleanup; |
fff42755 VM |
1743 | |
1744 | /* | |
1745 | * now we're going to use bitmaps, so load the actual bitmap entries | |
1746 | * from disk. this is the point of no return; after this the rev_list | |
1747 | * becomes invalidated and we must perform the revwalk through bitmaps | |
1748 | */ | |
65308ad8 | 1749 | if (load_bitmap(revs->repo, bitmap_git) < 0) |
f3c23db2 | 1750 | goto cleanup; |
fff42755 | 1751 | |
b0afdce5 TB |
1752 | if (!use_boundary_traversal) |
1753 | object_array_clear(&revs->pending); | |
fff42755 VM |
1754 | |
1755 | if (haves) { | |
b0afdce5 TB |
1756 | if (use_boundary_traversal) { |
1757 | trace2_region_enter("pack-bitmap", "haves/boundary", the_repository); | |
1758 | haves_bitmap = find_boundary_objects(bitmap_git, revs, haves); | |
1759 | trace2_region_leave("pack-bitmap", "haves/boundary", the_repository); | |
1760 | } else { | |
1761 | trace2_region_enter("pack-bitmap", "haves/classic", the_repository); | |
1762 | revs->ignore_missing_links = 1; | |
1763 | haves_bitmap = find_objects(bitmap_git, revs, haves, NULL); | |
1764 | reset_revision_walk(); | |
1765 | revs->ignore_missing_links = 0; | |
1766 | trace2_region_leave("pack-bitmap", "haves/classic", the_repository); | |
1767 | } | |
fff42755 | 1768 | |
afe8a907 | 1769 | if (!haves_bitmap) |
033abf97 | 1770 | BUG("failed to perform bitmap walk"); |
fff42755 VM |
1771 | } |
1772 | ||
b0afdce5 TB |
1773 | if (use_boundary_traversal) { |
1774 | object_array_clear(&revs->pending); | |
1775 | reset_revision_walk(); | |
1776 | } | |
1777 | ||
09d4a79e | 1778 | wants_bitmap = find_objects(bitmap_git, revs, wants, haves_bitmap); |
fff42755 VM |
1779 | |
1780 | if (!wants_bitmap) | |
033abf97 | 1781 | BUG("failed to perform bitmap walk"); |
fff42755 VM |
1782 | |
1783 | if (haves_bitmap) | |
1784 | bitmap_and_not(wants_bitmap, haves_bitmap); | |
1785 | ||
09d4a79e DS |
1786 | filter_bitmap(bitmap_git, |
1787 | (revs->filter.choice && filter_provided_objects) ? NULL : wants, | |
1788 | wants_bitmap, | |
1789 | &revs->filter); | |
6663ae0a | 1790 | |
3ae5fa07 | 1791 | bitmap_git->result = wants_bitmap; |
30cdc33f | 1792 | bitmap_git->haves = haves_bitmap; |
fff42755 | 1793 | |
acac50dd JK |
1794 | object_list_free(&wants); |
1795 | object_list_free(&haves); | |
1796 | ||
3ae5fa07 | 1797 | return bitmap_git; |
f3c23db2 JT |
1798 | |
1799 | cleanup: | |
1800 | free_bitmap_index(bitmap_git); | |
acac50dd JK |
1801 | object_list_free(&wants); |
1802 | object_list_free(&haves); | |
f3c23db2 | 1803 | return NULL; |
fff42755 VM |
1804 | } |
1805 | ||
a5f9f24a TB |
1806 | /* |
1807 | * -1 means "stop trying further objects"; 0 means we may or may not have | |
1808 | * reused, but you can keep feeding bits. | |
1809 | */ | |
73cd7d94 | 1810 | static int try_partial_reuse(struct packed_git *pack, |
a5f9f24a TB |
1811 | size_t pos, |
1812 | struct bitmap *reuse, | |
1813 | struct pack_window **w_curs) | |
fff42755 | 1814 | { |
0f533c72 | 1815 | off_t offset, delta_obj_offset; |
bb514de3 JK |
1816 | enum object_type type; |
1817 | unsigned long size; | |
1818 | ||
0f533c72 TB |
1819 | /* |
1820 | * try_partial_reuse() is called either on (a) objects in the | |
1821 | * bitmapped pack (in the case of a single-pack bitmap) or (b) | |
1822 | * objects in the preferred pack of a multi-pack bitmap. | |
1823 | * Importantly, the latter can pretend as if only a single pack | |
1824 | * exists because: | |
1825 | * | |
1826 | * - The first pack->num_objects bits of a MIDX bitmap are | |
1827 | * reserved for the preferred pack, and | |
1828 | * | |
1829 | * - Ties due to duplicate objects are always resolved in | |
1830 | * favor of the preferred pack. | |
1831 | * | |
1832 | * Therefore we do not need to ever ask the MIDX for its copy of | |
1833 | * an object by OID, since it will always select it from the | |
1834 | * preferred pack. Likewise, the selected copy of the base | |
1835 | * object for any deltas will reside in the same pack. | |
1836 | * | |
1837 | * This means that we can reuse pos when looking up the bit in | |
1838 | * the reuse bitmap, too, since bits corresponding to the | |
1839 | * preferred pack precede all bits from other packs. | |
1840 | */ | |
1841 | ||
1842 | if (pos >= pack->num_objects) | |
1843 | return -1; /* not actually in the pack or MIDX preferred pack */ | |
bb514de3 | 1844 | |
0f533c72 TB |
1845 | offset = delta_obj_offset = pack_pos_to_offset(pack, pos); |
1846 | type = unpack_object_header(pack, w_curs, &offset, &size); | |
bb514de3 | 1847 | if (type < 0) |
a5f9f24a | 1848 | return -1; /* broken packfile, punt */ |
bb514de3 JK |
1849 | |
1850 | if (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA) { | |
1851 | off_t base_offset; | |
011f3fd5 | 1852 | uint32_t base_pos; |
bb514de3 JK |
1853 | |
1854 | /* | |
1855 | * Find the position of the base object so we can look it up | |
1856 | * in our bitmaps. If we can't come up with an offset, or if | |
1857 | * that offset is not in the revidx, the pack is corrupt. | |
1858 | * There's nothing we can do, so just punt on this object, | |
1859 | * and the normal slow path will complain about it in | |
1860 | * more detail. | |
1861 | */ | |
0f533c72 TB |
1862 | base_offset = get_delta_base(pack, w_curs, &offset, type, |
1863 | delta_obj_offset); | |
bb514de3 | 1864 | if (!base_offset) |
a5f9f24a | 1865 | return 0; |
0f533c72 | 1866 | if (offset_to_pack_pos(pack, base_offset, &base_pos) < 0) |
a5f9f24a | 1867 | return 0; |
bb514de3 JK |
1868 | |
1869 | /* | |
1870 | * We assume delta dependencies always point backwards. This | |
1871 | * lets us do a single pass, and is basically always true | |
1872 | * due to the way OFS_DELTAs work. You would not typically | |
1873 | * find REF_DELTA in a bitmapped pack, since we only bitmap | |
1874 | * packs we write fresh, and OFS_DELTA is the default). But | |
1875 | * let's double check to make sure the pack wasn't written with | |
1876 | * odd parameters. | |
1877 | */ | |
1878 | if (base_pos >= pos) | |
a5f9f24a | 1879 | return 0; |
bb514de3 JK |
1880 | |
1881 | /* | |
1882 | * And finally, if we're not sending the base as part of our | |
1883 | * reuse chunk, then don't send this object either. The base | |
1884 | * would come after us, along with other objects not | |
1885 | * necessarily in the pack, which means we'd need to convert | |
1886 | * to REF_DELTA on the fly. Better to just let the normal | |
1887 | * object_entry code path handle it. | |
1888 | */ | |
1889 | if (!bitmap_get(reuse, base_pos)) | |
a5f9f24a | 1890 | return 0; |
bb514de3 JK |
1891 | } |
1892 | ||
fff42755 | 1893 | /* |
bb514de3 | 1894 | * If we got here, then the object is OK to reuse. Mark it. |
fff42755 | 1895 | */ |
bb514de3 | 1896 | bitmap_set(reuse, pos); |
a5f9f24a | 1897 | return 0; |
bb514de3 | 1898 | } |
fff42755 | 1899 | |
6d08b9d4 | 1900 | uint32_t midx_preferred_pack(struct bitmap_index *bitmap_git) |
0f533c72 TB |
1901 | { |
1902 | struct multi_pack_index *m = bitmap_git->midx; | |
1903 | if (!m) | |
1904 | BUG("midx_preferred_pack: requires non-empty MIDX"); | |
1905 | return nth_midxed_pack_int_id(m, pack_pos_to_midx(bitmap_git->midx, 0)); | |
1906 | } | |
1907 | ||
bb514de3 JK |
1908 | int reuse_partial_packfile_from_bitmap(struct bitmap_index *bitmap_git, |
1909 | struct packed_git **packfile_out, | |
1910 | uint32_t *entries, | |
1911 | struct bitmap **reuse_out) | |
1912 | { | |
65308ad8 | 1913 | struct repository *r = the_repository; |
0f533c72 | 1914 | struct packed_git *pack; |
3ae5fa07 | 1915 | struct bitmap *result = bitmap_git->result; |
bb514de3 JK |
1916 | struct bitmap *reuse; |
1917 | struct pack_window *w_curs = NULL; | |
1918 | size_t i = 0; | |
1919 | uint32_t offset; | |
0f533c72 | 1920 | uint32_t objects_nr; |
fff42755 VM |
1921 | |
1922 | assert(result); | |
1923 | ||
65308ad8 | 1924 | load_reverse_index(r, bitmap_git); |
0f533c72 TB |
1925 | |
1926 | if (bitmap_is_midx(bitmap_git)) | |
1927 | pack = bitmap_git->midx->packs[midx_preferred_pack(bitmap_git)]; | |
1928 | else | |
1929 | pack = bitmap_git->pack; | |
1930 | objects_nr = pack->num_objects; | |
1931 | ||
bb514de3 JK |
1932 | while (i < result->word_alloc && result->words[i] == (eword_t)~0) |
1933 | i++; | |
fff42755 | 1934 | |
0f533c72 TB |
1935 | /* |
1936 | * Don't mark objects not in the packfile or preferred pack. This bitmap | |
1937 | * marks objects eligible for reuse, but the pack-reuse code only | |
1938 | * understands how to reuse a single pack. Since the preferred pack is | |
1939 | * guaranteed to have all bases for its deltas (in a multi-pack bitmap), | |
1940 | * we use it instead of another pack. In single-pack bitmaps, the choice | |
1941 | * is made for us. | |
1942 | */ | |
ed184620 TB |
1943 | if (i > objects_nr / BITS_IN_EWORD) |
1944 | i = objects_nr / BITS_IN_EWORD; | |
fff42755 | 1945 | |
bb514de3 JK |
1946 | reuse = bitmap_word_alloc(i); |
1947 | memset(reuse->words, 0xFF, i * sizeof(eword_t)); | |
fff42755 | 1948 | |
bb514de3 JK |
1949 | for (; i < result->word_alloc; ++i) { |
1950 | eword_t word = result->words[i]; | |
1951 | size_t pos = (i * BITS_IN_EWORD); | |
fff42755 | 1952 | |
bb514de3 JK |
1953 | for (offset = 0; offset < BITS_IN_EWORD; ++offset) { |
1954 | if ((word >> offset) == 0) | |
1955 | break; | |
fff42755 | 1956 | |
bb514de3 | 1957 | offset += ewah_bit_ctz64(word >> offset); |
73cd7d94 | 1958 | if (try_partial_reuse(pack, pos + offset, |
0f533c72 | 1959 | reuse, &w_curs) < 0) { |
a5f9f24a TB |
1960 | /* |
1961 | * try_partial_reuse indicated we couldn't reuse | |
1962 | * any bits, so there is no point in trying more | |
1963 | * bits in the current word, or any other words | |
1964 | * in result. | |
1965 | * | |
1966 | * Jump out of both loops to avoid future | |
1967 | * unnecessary calls to try_partial_reuse. | |
1968 | */ | |
1969 | goto done; | |
1970 | } | |
bb514de3 | 1971 | } |
fff42755 | 1972 | } |
fff42755 | 1973 | |
a5f9f24a | 1974 | done: |
bb514de3 | 1975 | unuse_pack(&w_curs); |
fff42755 | 1976 | |
bb514de3 JK |
1977 | *entries = bitmap_popcount(reuse); |
1978 | if (!*entries) { | |
1979 | bitmap_free(reuse); | |
fff42755 | 1980 | return -1; |
fff42755 VM |
1981 | } |
1982 | ||
bb514de3 JK |
1983 | /* |
1984 | * Drop any reused objects from the result, since they will not | |
1985 | * need to be handled separately. | |
1986 | */ | |
1987 | bitmap_and_not(result, reuse); | |
0f533c72 | 1988 | *packfile_out = pack; |
bb514de3 | 1989 | *reuse_out = reuse; |
fff42755 VM |
1990 | return 0; |
1991 | } | |
fff42755 | 1992 | |
40d18ff8 JK |
1993 | int bitmap_walk_contains(struct bitmap_index *bitmap_git, |
1994 | struct bitmap *bitmap, const struct object_id *oid) | |
1995 | { | |
1996 | int idx; | |
fff42755 | 1997 | |
40d18ff8 JK |
1998 | if (!bitmap) |
1999 | return 0; | |
fff42755 | 2000 | |
40d18ff8 JK |
2001 | idx = bitmap_position(bitmap_git, oid); |
2002 | return idx >= 0 && bitmap_get(bitmap, idx); | |
fff42755 VM |
2003 | } |
2004 | ||
3ae5fa07 | 2005 | void traverse_bitmap_commit_list(struct bitmap_index *bitmap_git, |
4eb707eb | 2006 | struct rev_info *revs, |
3ae5fa07 | 2007 | show_reachable_fn show_reachable) |
fff42755 | 2008 | { |
3ae5fa07 | 2009 | assert(bitmap_git->result); |
fff42755 | 2010 | |
551cf8b6 | 2011 | show_objects_for_type(bitmap_git, OBJ_COMMIT, show_reachable); |
4eb707eb JK |
2012 | if (revs->tree_objects) |
2013 | show_objects_for_type(bitmap_git, OBJ_TREE, show_reachable); | |
2014 | if (revs->blob_objects) | |
2015 | show_objects_for_type(bitmap_git, OBJ_BLOB, show_reachable); | |
2016 | if (revs->tag_objects) | |
2017 | show_objects_for_type(bitmap_git, OBJ_TAG, show_reachable); | |
fff42755 | 2018 | |
4eb707eb | 2019 | show_extended_objects(bitmap_git, revs, show_reachable); |
fff42755 VM |
2020 | } |
2021 | ||
3ae5fa07 | 2022 | static uint32_t count_object_type(struct bitmap_index *bitmap_git, |
fff42755 VM |
2023 | enum object_type type) |
2024 | { | |
3ae5fa07 JT |
2025 | struct bitmap *objects = bitmap_git->result; |
2026 | struct eindex *eindex = &bitmap_git->ext_index; | |
fff42755 VM |
2027 | |
2028 | uint32_t i = 0, count = 0; | |
2029 | struct ewah_iterator it; | |
2030 | eword_t filter; | |
2031 | ||
551cf8b6 | 2032 | init_type_iterator(&it, bitmap_git, type); |
fff42755 VM |
2033 | |
2034 | while (i < objects->word_alloc && ewah_iterator_next(&filter, &it)) { | |
2035 | eword_t word = objects->words[i++] & filter; | |
2036 | count += ewah_bit_popcount64(word); | |
2037 | } | |
2038 | ||
2039 | for (i = 0; i < eindex->count; ++i) { | |
2040 | if (eindex->objects[i]->type == type && | |
0948c501 TB |
2041 | bitmap_get(objects, |
2042 | st_add(bitmap_num_objects(bitmap_git), i))) | |
fff42755 VM |
2043 | count++; |
2044 | } | |
2045 | ||
2046 | return count; | |
2047 | } | |
2048 | ||
3ae5fa07 JT |
2049 | void count_bitmap_commit_list(struct bitmap_index *bitmap_git, |
2050 | uint32_t *commits, uint32_t *trees, | |
fff42755 VM |
2051 | uint32_t *blobs, uint32_t *tags) |
2052 | { | |
3ae5fa07 | 2053 | assert(bitmap_git->result); |
fff42755 VM |
2054 | |
2055 | if (commits) | |
3ae5fa07 | 2056 | *commits = count_object_type(bitmap_git, OBJ_COMMIT); |
fff42755 VM |
2057 | |
2058 | if (trees) | |
3ae5fa07 | 2059 | *trees = count_object_type(bitmap_git, OBJ_TREE); |
fff42755 VM |
2060 | |
2061 | if (blobs) | |
3ae5fa07 | 2062 | *blobs = count_object_type(bitmap_git, OBJ_BLOB); |
fff42755 VM |
2063 | |
2064 | if (tags) | |
3ae5fa07 | 2065 | *tags = count_object_type(bitmap_git, OBJ_TAG); |
fff42755 VM |
2066 | } |
2067 | ||
2068 | struct bitmap_test_data { | |
3ae5fa07 | 2069 | struct bitmap_index *bitmap_git; |
fff42755 | 2070 | struct bitmap *base; |
fa95666a TB |
2071 | struct bitmap *commits; |
2072 | struct bitmap *trees; | |
2073 | struct bitmap *blobs; | |
2074 | struct bitmap *tags; | |
fff42755 VM |
2075 | struct progress *prg; |
2076 | size_t seen; | |
2077 | }; | |
2078 | ||
fa95666a TB |
2079 | static void test_bitmap_type(struct bitmap_test_data *tdata, |
2080 | struct object *obj, int pos) | |
2081 | { | |
2082 | enum object_type bitmap_type = OBJ_NONE; | |
2083 | int bitmaps_nr = 0; | |
2084 | ||
2085 | if (bitmap_get(tdata->commits, pos)) { | |
2086 | bitmap_type = OBJ_COMMIT; | |
2087 | bitmaps_nr++; | |
2088 | } | |
2089 | if (bitmap_get(tdata->trees, pos)) { | |
2090 | bitmap_type = OBJ_TREE; | |
2091 | bitmaps_nr++; | |
2092 | } | |
2093 | if (bitmap_get(tdata->blobs, pos)) { | |
2094 | bitmap_type = OBJ_BLOB; | |
2095 | bitmaps_nr++; | |
2096 | } | |
2097 | if (bitmap_get(tdata->tags, pos)) { | |
2098 | bitmap_type = OBJ_TAG; | |
2099 | bitmaps_nr++; | |
2100 | } | |
2101 | ||
2102 | if (bitmap_type == OBJ_NONE) | |
9975975d | 2103 | die(_("object '%s' not found in type bitmaps"), |
fa95666a TB |
2104 | oid_to_hex(&obj->oid)); |
2105 | ||
2106 | if (bitmaps_nr > 1) | |
9975975d | 2107 | die(_("object '%s' does not have a unique type"), |
fa95666a TB |
2108 | oid_to_hex(&obj->oid)); |
2109 | ||
2110 | if (bitmap_type != obj->type) | |
9975975d | 2111 | die(_("object '%s': real type '%s', expected: '%s'"), |
fa95666a TB |
2112 | oid_to_hex(&obj->oid), |
2113 | type_name(obj->type), | |
2114 | type_name(bitmap_type)); | |
2115 | } | |
2116 | ||
c50dca2a JK |
2117 | static void test_show_object(struct object *object, |
2118 | const char *name UNUSED, | |
de1e67d0 | 2119 | void *data) |
fff42755 VM |
2120 | { |
2121 | struct bitmap_test_data *tdata = data; | |
2122 | int bitmap_pos; | |
2123 | ||
3c771448 | 2124 | bitmap_pos = bitmap_position(tdata->bitmap_git, &object->oid); |
fff42755 | 2125 | if (bitmap_pos < 0) |
9975975d | 2126 | die(_("object not in bitmap: '%s'"), oid_to_hex(&object->oid)); |
fa95666a | 2127 | test_bitmap_type(tdata, object, bitmap_pos); |
fff42755 VM |
2128 | |
2129 | bitmap_set(tdata->base, bitmap_pos); | |
2130 | display_progress(tdata->prg, ++tdata->seen); | |
2131 | } | |
2132 | ||
2133 | static void test_show_commit(struct commit *commit, void *data) | |
2134 | { | |
2135 | struct bitmap_test_data *tdata = data; | |
2136 | int bitmap_pos; | |
2137 | ||
3ae5fa07 | 2138 | bitmap_pos = bitmap_position(tdata->bitmap_git, |
3c771448 | 2139 | &commit->object.oid); |
fff42755 | 2140 | if (bitmap_pos < 0) |
9975975d | 2141 | die(_("object not in bitmap: '%s'"), oid_to_hex(&commit->object.oid)); |
fa95666a | 2142 | test_bitmap_type(tdata, &commit->object, bitmap_pos); |
fff42755 VM |
2143 | |
2144 | bitmap_set(tdata->base, bitmap_pos); | |
2145 | display_progress(tdata->prg, ++tdata->seen); | |
2146 | } | |
2147 | ||
2148 | void test_bitmap_walk(struct rev_info *revs) | |
2149 | { | |
2150 | struct object *root; | |
2151 | struct bitmap *result = NULL; | |
fff42755 VM |
2152 | size_t result_popcnt; |
2153 | struct bitmap_test_data tdata; | |
3ae5fa07 | 2154 | struct bitmap_index *bitmap_git; |
98c31f36 | 2155 | struct ewah_bitmap *bm; |
fff42755 | 2156 | |
7c141127 | 2157 | if (!(bitmap_git = prepare_bitmap_git(revs->repo))) |
9975975d | 2158 | die(_("failed to load bitmap indexes")); |
fff42755 VM |
2159 | |
2160 | if (revs->pending.nr != 1) | |
9975975d | 2161 | die(_("you must specify exactly one commit to test")); |
fff42755 | 2162 | |
28cd7306 AC |
2163 | fprintf_ln(stderr, "Bitmap v%d test (%d entries%s)", |
2164 | bitmap_git->version, | |
2165 | bitmap_git->entry_count, | |
2166 | bitmap_git->table_lookup ? "" : " loaded"); | |
fff42755 VM |
2167 | |
2168 | root = revs->pending.objects[0].item; | |
98c31f36 | 2169 | bm = bitmap_for_commit(bitmap_git, (struct commit *)root); |
fff42755 | 2170 | |
98c31f36 | 2171 | if (bm) { |
baf20c39 | 2172 | fprintf_ln(stderr, "Found bitmap for '%s'. %d bits / %08x checksum", |
f2fd0760 | 2173 | oid_to_hex(&root->oid), (int)bm->bit_size, ewah_checksum(bm)); |
fff42755 VM |
2174 | |
2175 | result = ewah_to_bitmap(bm); | |
2176 | } | |
2177 | ||
afe8a907 | 2178 | if (!result) |
9975975d | 2179 | die(_("commit '%s' doesn't have an indexed bitmap"), oid_to_hex(&root->oid)); |
fff42755 VM |
2180 | |
2181 | revs->tag_objects = 1; | |
2182 | revs->tree_objects = 1; | |
2183 | revs->blob_objects = 1; | |
2184 | ||
2185 | result_popcnt = bitmap_popcount(result); | |
2186 | ||
2187 | if (prepare_revision_walk(revs)) | |
9975975d | 2188 | die(_("revision walk setup failed")); |
fff42755 | 2189 | |
3ae5fa07 | 2190 | tdata.bitmap_git = bitmap_git; |
fff42755 | 2191 | tdata.base = bitmap_new(); |
fa95666a TB |
2192 | tdata.commits = ewah_to_bitmap(bitmap_git->commits); |
2193 | tdata.trees = ewah_to_bitmap(bitmap_git->trees); | |
2194 | tdata.blobs = ewah_to_bitmap(bitmap_git->blobs); | |
2195 | tdata.tags = ewah_to_bitmap(bitmap_git->tags); | |
fff42755 VM |
2196 | tdata.prg = start_progress("Verifying bitmap entries", result_popcnt); |
2197 | tdata.seen = 0; | |
2198 | ||
2199 | traverse_commit_list(revs, &test_show_commit, &test_show_object, &tdata); | |
2200 | ||
2201 | stop_progress(&tdata.prg); | |
2202 | ||
2203 | if (bitmap_equals(result, tdata.base)) | |
baf20c39 | 2204 | fprintf_ln(stderr, "OK!"); |
fff42755 | 2205 | else |
9975975d | 2206 | die(_("mismatch in bitmap results")); |
f86a3747 | 2207 | |
02281511 TB |
2208 | bitmap_free(result); |
2209 | bitmap_free(tdata.base); | |
2210 | bitmap_free(tdata.commits); | |
2211 | bitmap_free(tdata.trees); | |
2212 | bitmap_free(tdata.blobs); | |
2213 | bitmap_free(tdata.tags); | |
f3c23db2 | 2214 | free_bitmap_index(bitmap_git); |
fff42755 | 2215 | } |
7cc8f971 | 2216 | |
dff5e49e TB |
2217 | int test_bitmap_commits(struct repository *r) |
2218 | { | |
dff5e49e TB |
2219 | struct object_id oid; |
2220 | MAYBE_UNUSED void *value; | |
28cd7306 | 2221 | struct bitmap_index *bitmap_git = prepare_bitmap_git(r); |
dff5e49e TB |
2222 | |
2223 | if (!bitmap_git) | |
9975975d | 2224 | die(_("failed to load bitmap indexes")); |
dff5e49e | 2225 | |
28cd7306 AC |
2226 | /* |
2227 | * As this function is only used to print bitmap selected | |
2228 | * commits, we don't have to read the commit table. | |
2229 | */ | |
2230 | if (bitmap_git->table_lookup) { | |
2231 | if (load_bitmap_entries_v1(bitmap_git) < 0) | |
2232 | die(_("failed to load bitmap indexes")); | |
2233 | } | |
2234 | ||
dff5e49e | 2235 | kh_foreach(bitmap_git->bitmaps, oid, value, { |
baf20c39 | 2236 | printf_ln("%s", oid_to_hex(&oid)); |
dff5e49e TB |
2237 | }); |
2238 | ||
2239 | free_bitmap_index(bitmap_git); | |
2240 | ||
2241 | return 0; | |
2242 | } | |
2243 | ||
a05f02b1 TB |
2244 | int test_bitmap_hashes(struct repository *r) |
2245 | { | |
2246 | struct bitmap_index *bitmap_git = prepare_bitmap_git(r); | |
2247 | struct object_id oid; | |
2248 | uint32_t i, index_pos; | |
2249 | ||
875da7f0 | 2250 | if (!bitmap_git || !bitmap_git->hashes) |
a05f02b1 TB |
2251 | goto cleanup; |
2252 | ||
2253 | for (i = 0; i < bitmap_num_objects(bitmap_git); i++) { | |
2254 | if (bitmap_is_midx(bitmap_git)) | |
2255 | index_pos = pack_pos_to_midx(bitmap_git->midx, i); | |
2256 | else | |
2257 | index_pos = pack_pos_to_index(bitmap_git->pack, i); | |
2258 | ||
2259 | nth_bitmap_object_oid(bitmap_git, &oid, index_pos); | |
2260 | ||
baf20c39 | 2261 | printf_ln("%s %"PRIu32"", |
a05f02b1 TB |
2262 | oid_to_hex(&oid), get_be32(bitmap_git->hashes + index_pos)); |
2263 | } | |
2264 | ||
2265 | cleanup: | |
2266 | free_bitmap_index(bitmap_git); | |
2267 | ||
2268 | return 0; | |
2269 | } | |
2270 | ||
449fa5ee JK |
2271 | int rebuild_bitmap(const uint32_t *reposition, |
2272 | struct ewah_bitmap *source, | |
2273 | struct bitmap *dest) | |
7cc8f971 VM |
2274 | { |
2275 | uint32_t pos = 0; | |
2276 | struct ewah_iterator it; | |
2277 | eword_t word; | |
2278 | ||
2279 | ewah_iterator_init(&it, source); | |
2280 | ||
2281 | while (ewah_iterator_next(&word, &it)) { | |
2282 | uint32_t offset, bit_pos; | |
2283 | ||
34b935c0 | 2284 | for (offset = 0; offset < BITS_IN_EWORD; ++offset) { |
7cc8f971 VM |
2285 | if ((word >> offset) == 0) |
2286 | break; | |
2287 | ||
2288 | offset += ewah_bit_ctz64(word >> offset); | |
2289 | ||
2290 | bit_pos = reposition[pos + offset]; | |
2291 | if (bit_pos > 0) | |
2292 | bitmap_set(dest, bit_pos - 1); | |
2293 | else /* can't reuse, we don't have the object */ | |
2294 | return -1; | |
2295 | } | |
2296 | ||
34b935c0 | 2297 | pos += BITS_IN_EWORD; |
7cc8f971 VM |
2298 | } |
2299 | return 0; | |
2300 | } | |
2301 | ||
449fa5ee JK |
2302 | uint32_t *create_bitmap_mapping(struct bitmap_index *bitmap_git, |
2303 | struct packing_data *mapping) | |
7cc8f971 | 2304 | { |
65308ad8 | 2305 | struct repository *r = the_repository; |
7cc8f971 VM |
2306 | uint32_t i, num_objects; |
2307 | uint32_t *reposition; | |
7cc8f971 | 2308 | |
0f533c72 | 2309 | if (!bitmap_is_midx(bitmap_git)) |
65308ad8 | 2310 | load_reverse_index(r, bitmap_git); |
5a6072f6 | 2311 | else if (load_midx_revindex(bitmap_git->midx)) |
0f533c72 TB |
2312 | BUG("rebuild_existing_bitmaps: missing required rev-cache " |
2313 | "extension"); | |
2314 | ||
ed184620 | 2315 | num_objects = bitmap_num_objects(bitmap_git); |
ca56dadb | 2316 | CALLOC_ARRAY(reposition, num_objects); |
7cc8f971 VM |
2317 | |
2318 | for (i = 0; i < num_objects; ++i) { | |
3df28cae | 2319 | struct object_id oid; |
7cc8f971 | 2320 | struct object_entry *oe; |
8de300e1 | 2321 | uint32_t index_pos; |
7cc8f971 | 2322 | |
0f533c72 | 2323 | if (bitmap_is_midx(bitmap_git)) |
8de300e1 | 2324 | index_pos = pack_pos_to_midx(bitmap_git->midx, i); |
0f533c72 | 2325 | else |
8de300e1 TB |
2326 | index_pos = pack_pos_to_index(bitmap_git->pack, i); |
2327 | nth_bitmap_object_oid(bitmap_git, &oid, index_pos); | |
3a37876b | 2328 | oe = packlist_find(mapping, &oid); |
7cc8f971 | 2329 | |
8de300e1 | 2330 | if (oe) { |
06af3bba | 2331 | reposition[i] = oe_in_pack_pos(mapping, oe) + 1; |
8de300e1 TB |
2332 | if (bitmap_git->hashes && !oe->hash) |
2333 | oe->hash = get_be32(bitmap_git->hashes + index_pos); | |
2334 | } | |
7cc8f971 VM |
2335 | } |
2336 | ||
449fa5ee | 2337 | return reposition; |
7cc8f971 | 2338 | } |
f3c23db2 JT |
2339 | |
2340 | void free_bitmap_index(struct bitmap_index *b) | |
2341 | { | |
2342 | if (!b) | |
2343 | return; | |
2344 | ||
2345 | if (b->map) | |
2346 | munmap(b->map, b->map_size); | |
2347 | ewah_pool_free(b->commits); | |
2348 | ewah_pool_free(b->trees); | |
2349 | ewah_pool_free(b->blobs); | |
2350 | ewah_pool_free(b->tags); | |
655b8561 TB |
2351 | if (b->bitmaps) { |
2352 | struct stored_bitmap *sb; | |
2353 | kh_foreach_value(b->bitmaps, sb, { | |
2354 | ewah_pool_free(sb->root); | |
2355 | free(sb); | |
2356 | }); | |
2357 | } | |
3c771448 | 2358 | kh_destroy_oid_map(b->bitmaps); |
f3c23db2 JT |
2359 | free(b->ext_index.objects); |
2360 | free(b->ext_index.hashes); | |
655b8561 | 2361 | kh_destroy_oid_pos(b->ext_index.positions); |
f3c23db2 | 2362 | bitmap_free(b->result); |
30cdc33f | 2363 | bitmap_free(b->haves); |
0f533c72 TB |
2364 | if (bitmap_is_midx(b)) { |
2365 | /* | |
2366 | * Multi-pack bitmaps need to have resources associated with | |
2367 | * their on-disk reverse indexes unmapped so that stale .rev and | |
2368 | * .bitmap files can be removed. | |
2369 | * | |
2370 | * Unlike pack-based bitmaps, multi-pack bitmaps can be read and | |
2371 | * written in the same 'git multi-pack-index write --bitmap' | |
2372 | * process. Close resources so they can be removed safely on | |
2373 | * platforms like Windows. | |
2374 | */ | |
2375 | close_midx_revindex(b->midx); | |
2376 | } | |
f3c23db2 JT |
2377 | free(b); |
2378 | } | |
30cdc33f | 2379 | |
3c771448 | 2380 | int bitmap_has_oid_in_uninteresting(struct bitmap_index *bitmap_git, |
2381 | const struct object_id *oid) | |
30cdc33f | 2382 | { |
8ebf5296 JK |
2383 | return bitmap_git && |
2384 | bitmap_walk_contains(bitmap_git, bitmap_git->haves, oid); | |
30cdc33f | 2385 | } |
16950f83 JK |
2386 | |
2387 | static off_t get_disk_usage_for_type(struct bitmap_index *bitmap_git, | |
2388 | enum object_type object_type) | |
2389 | { | |
2390 | struct bitmap *result = bitmap_git->result; | |
16950f83 JK |
2391 | off_t total = 0; |
2392 | struct ewah_iterator it; | |
2393 | eword_t filter; | |
2394 | size_t i; | |
2395 | ||
2396 | init_type_iterator(&it, bitmap_git, object_type); | |
2397 | for (i = 0; i < result->word_alloc && | |
2398 | ewah_iterator_next(&filter, &it); i++) { | |
2399 | eword_t word = result->words[i] & filter; | |
2400 | size_t base = (i * BITS_IN_EWORD); | |
2401 | unsigned offset; | |
2402 | ||
2403 | if (!word) | |
2404 | continue; | |
2405 | ||
2406 | for (offset = 0; offset < BITS_IN_EWORD; offset++) { | |
16950f83 JK |
2407 | if ((word >> offset) == 0) |
2408 | break; | |
2409 | ||
2410 | offset += ewah_bit_ctz64(word >> offset); | |
0f533c72 TB |
2411 | |
2412 | if (bitmap_is_midx(bitmap_git)) { | |
2413 | uint32_t pack_pos; | |
2414 | uint32_t midx_pos = pack_pos_to_midx(bitmap_git->midx, base + offset); | |
2415 | off_t offset = nth_midxed_offset(bitmap_git->midx, midx_pos); | |
2416 | ||
2417 | uint32_t pack_id = nth_midxed_pack_int_id(bitmap_git->midx, midx_pos); | |
2418 | struct packed_git *pack = bitmap_git->midx->packs[pack_id]; | |
2419 | ||
2420 | if (offset_to_pack_pos(pack, offset, &pack_pos) < 0) { | |
2421 | struct object_id oid; | |
2422 | nth_midxed_object_oid(&oid, bitmap_git->midx, midx_pos); | |
2423 | ||
baf20c39 | 2424 | die(_("could not find '%s' in pack '%s' at offset %"PRIuMAX), |
0f533c72 TB |
2425 | oid_to_hex(&oid), |
2426 | pack->pack_name, | |
2427 | (uintmax_t)offset); | |
2428 | } | |
2429 | ||
2430 | total += pack_pos_to_offset(pack, pack_pos + 1) - offset; | |
2431 | } else { | |
2432 | size_t pos = base + offset; | |
2433 | total += pack_pos_to_offset(bitmap_git->pack, pos + 1) - | |
2434 | pack_pos_to_offset(bitmap_git->pack, pos); | |
2435 | } | |
16950f83 JK |
2436 | } |
2437 | } | |
2438 | ||
2439 | return total; | |
2440 | } | |
2441 | ||
2442 | static off_t get_disk_usage_for_extended(struct bitmap_index *bitmap_git) | |
2443 | { | |
2444 | struct bitmap *result = bitmap_git->result; | |
16950f83 JK |
2445 | struct eindex *eindex = &bitmap_git->ext_index; |
2446 | off_t total = 0; | |
2447 | struct object_info oi = OBJECT_INFO_INIT; | |
2448 | off_t object_size; | |
2449 | size_t i; | |
2450 | ||
2451 | oi.disk_sizep = &object_size; | |
2452 | ||
2453 | for (i = 0; i < eindex->count; i++) { | |
2454 | struct object *obj = eindex->objects[i]; | |
2455 | ||
0948c501 TB |
2456 | if (!bitmap_get(result, |
2457 | st_add(bitmap_num_objects(bitmap_git), i))) | |
16950f83 JK |
2458 | continue; |
2459 | ||
2460 | if (oid_object_info_extended(the_repository, &obj->oid, &oi, 0) < 0) | |
baf20c39 | 2461 | die(_("unable to get disk usage of '%s'"), |
16950f83 JK |
2462 | oid_to_hex(&obj->oid)); |
2463 | ||
2464 | total += object_size; | |
2465 | } | |
2466 | return total; | |
2467 | } | |
2468 | ||
2469 | off_t get_disk_usage_from_bitmap(struct bitmap_index *bitmap_git, | |
2470 | struct rev_info *revs) | |
2471 | { | |
2472 | off_t total = 0; | |
2473 | ||
2474 | total += get_disk_usage_for_type(bitmap_git, OBJ_COMMIT); | |
2475 | if (revs->tree_objects) | |
2476 | total += get_disk_usage_for_type(bitmap_git, OBJ_TREE); | |
2477 | if (revs->blob_objects) | |
2478 | total += get_disk_usage_for_type(bitmap_git, OBJ_BLOB); | |
2479 | if (revs->tag_objects) | |
2480 | total += get_disk_usage_for_type(bitmap_git, OBJ_TAG); | |
2481 | ||
2482 | total += get_disk_usage_for_extended(bitmap_git); | |
2483 | ||
2484 | return total; | |
2485 | } | |
3f267a11 | 2486 | |
0f533c72 TB |
2487 | int bitmap_is_midx(struct bitmap_index *bitmap_git) |
2488 | { | |
2489 | return !!bitmap_git->midx; | |
2490 | } | |
2491 | ||
3f267a11 TB |
2492 | const struct string_list *bitmap_preferred_tips(struct repository *r) |
2493 | { | |
a4286193 ÆAB |
2494 | const struct string_list *dest; |
2495 | ||
9e2d884d | 2496 | if (!repo_config_get_string_multi(r, "pack.preferbitmaptips", &dest)) |
a4286193 ÆAB |
2497 | return dest; |
2498 | return NULL; | |
3f267a11 | 2499 | } |
711260fd TB |
2500 | |
2501 | int bitmap_is_preferred_refname(struct repository *r, const char *refname) | |
2502 | { | |
2503 | const struct string_list *preferred_tips = bitmap_preferred_tips(r); | |
2504 | struct string_list_item *item; | |
2505 | ||
2506 | if (!preferred_tips) | |
2507 | return 0; | |
2508 | ||
2509 | for_each_string_list_item(item, preferred_tips) { | |
2510 | if (starts_with(refname, item->string)) | |
2511 | return 1; | |
2512 | } | |
2513 | ||
2514 | return 0; | |
2515 | } | |
756f1bcd DS |
2516 | |
2517 | static int verify_bitmap_file(const char *name) | |
2518 | { | |
2519 | struct stat st; | |
2520 | unsigned char *data; | |
2521 | int fd = git_open(name); | |
2522 | int res = 0; | |
2523 | ||
2524 | /* It is OK to not have the file. */ | |
2525 | if (fd < 0 || fstat(fd, &st)) { | |
2526 | if (fd >= 0) | |
2527 | close(fd); | |
2528 | return 0; | |
2529 | } | |
2530 | ||
2531 | data = xmmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0); | |
2532 | close(fd); | |
2533 | if (!hashfile_checksum_valid(data, st.st_size)) | |
2534 | res = error(_("bitmap file '%s' has invalid checksum"), | |
2535 | name); | |
2536 | ||
2537 | munmap(data, st.st_size); | |
2538 | return res; | |
2539 | } | |
2540 | ||
2541 | int verify_bitmap_files(struct repository *r) | |
2542 | { | |
2543 | int res = 0; | |
2544 | ||
2545 | for (struct multi_pack_index *m = get_multi_pack_index(r); | |
2546 | m; m = m->next) { | |
2547 | char *midx_bitmap_name = midx_bitmap_filename(m); | |
2548 | res |= verify_bitmap_file(midx_bitmap_name); | |
2549 | free(midx_bitmap_name); | |
2550 | } | |
2551 | ||
2552 | for (struct packed_git *p = get_all_packs(r); | |
2553 | p; p = p->next) { | |
2554 | char *pack_bitmap_name = pack_bitmap_filename(p); | |
2555 | res |= verify_bitmap_file(pack_bitmap_name); | |
2556 | free(pack_bitmap_name); | |
2557 | } | |
2558 | ||
2559 | return res; | |
2560 | } |