]>
Commit | Line | Data |
---|---|---|
7d0cdf82 | 1 | /* |
359fc2d2 | 2 | * Copyright (C) the libgit2 contributors. All rights reserved. |
7d0cdf82 | 3 | * |
bb742ede VM |
4 | * This file is part of libgit2, distributed under the GNU GPL v2 with |
5 | * a Linking Exception. For full terms see the included COPYING file. | |
7d0cdf82 CMN |
6 | */ |
7 | ||
a15c550d | 8 | #include "common.h" |
7d0cdf82 CMN |
9 | #include "odb.h" |
10 | #include "pack.h" | |
11 | #include "delta-apply.h" | |
a070f152 | 12 | #include "sha1_lookup.h" |
a15c550d VM |
13 | #include "mwindow.h" |
14 | #include "fileops.h" | |
b7f167da | 15 | #include "oid.h" |
7d0cdf82 | 16 | |
0c3bae62 | 17 | #include <zlib.h> |
7d0cdf82 | 18 | |
a070f152 | 19 | static int packfile_open(struct git_pack_file *p); |
e1de726c | 20 | static git_off_t nth_packed_object_offset(const struct git_pack_file *p, uint32_t n); |
a070f152 CMN |
21 | int packfile_unpack_compressed( |
22 | git_rawobj *obj, | |
23 | struct git_pack_file *p, | |
24 | git_mwindow **w_curs, | |
e1de726c | 25 | git_off_t *curpos, |
a070f152 CMN |
26 | size_t size, |
27 | git_otype type); | |
28 | ||
29 | /* Can find the offset of an object given | |
30 | * a prefix of an identifier. | |
904b67e6 | 31 | * Throws GIT_EAMBIGUOUSOIDPREFIX if short oid |
a070f152 CMN |
32 | * is ambiguous within the pack. |
33 | * This method assumes that len is between | |
34 | * GIT_OID_MINPREFIXLEN and GIT_OID_HEXSZ. | |
35 | */ | |
36 | static int pack_entry_find_offset( | |
e1de726c | 37 | git_off_t *offset_out, |
a070f152 CMN |
38 | git_oid *found_oid, |
39 | struct git_pack_file *p, | |
40 | const git_oid *short_oid, | |
b8457baa | 41 | size_t len); |
a070f152 | 42 | |
e1de726c RB |
43 | static int packfile_error(const char *message) |
44 | { | |
45 | giterr_set(GITERR_ODB, "Invalid pack file - %s", message); | |
46 | return -1; | |
47 | } | |
48 | ||
c8f79c2b CMN |
49 | /******************** |
50 | * Delta base cache | |
51 | ********************/ | |
c0f4a011 | 52 | |
525d961c | 53 | static git_pack_cache_entry *new_cache_object(git_rawobj *source) |
c0f4a011 | 54 | { |
c8f79c2b | 55 | git_pack_cache_entry *e = git__calloc(1, sizeof(git_pack_cache_entry)); |
c0f4a011 CMN |
56 | if (!e) |
57 | return NULL; | |
58 | ||
c0f4a011 CMN |
59 | memcpy(&e->raw, source, sizeof(git_rawobj)); |
60 | ||
61 | return e; | |
62 | } | |
63 | ||
64 | static void free_cache_object(void *o) | |
65 | { | |
66 | git_pack_cache_entry *e = (git_pack_cache_entry *)o; | |
67 | ||
68 | if (e != NULL) { | |
0ed75620 | 69 | assert(e->refcount.val == 0); |
c0f4a011 CMN |
70 | git__free(e->raw.data); |
71 | git__free(e); | |
72 | } | |
73 | } | |
74 | ||
c8f79c2b CMN |
75 | static void cache_free(git_pack_cache *cache) |
76 | { | |
77 | khiter_t k; | |
78 | ||
79 | if (cache->entries) { | |
80 | for (k = kh_begin(cache->entries); k != kh_end(cache->entries); k++) { | |
81 | if (kh_exist(cache->entries, k)) | |
82 | free_cache_object(kh_value(cache->entries, k)); | |
83 | } | |
84 | ||
85 | git_offmap_free(cache->entries); | |
aa3bf89d | 86 | git_mutex_free(&cache->lock); |
c8f79c2b | 87 | } |
f658dc43 RB |
88 | |
89 | memset(cache, 0, sizeof(*cache)); | |
c8f79c2b CMN |
90 | } |
91 | ||
92 | static int cache_init(git_pack_cache *cache) | |
93 | { | |
f658dc43 RB |
94 | memset(cache, 0, sizeof(*cache)); |
95 | ||
c8f79c2b CMN |
96 | cache->entries = git_offmap_alloc(); |
97 | GITERR_CHECK_ALLOC(cache->entries); | |
f658dc43 | 98 | |
c8f79c2b | 99 | cache->memory_limit = GIT_PACK_CACHE_MEMORY_LIMIT; |
1a42dd17 RB |
100 | |
101 | if (git_mutex_init(&cache->lock)) { | |
102 | giterr_set(GITERR_OS, "Failed to initialize pack cache mutex"); | |
103 | ||
104 | git__free(cache->entries); | |
105 | cache->entries = NULL; | |
106 | ||
107 | return -1; | |
108 | } | |
c8f79c2b CMN |
109 | |
110 | return 0; | |
111 | } | |
112 | ||
09e29e47 | 113 | static git_pack_cache_entry *cache_get(git_pack_cache *cache, git_off_t offset) |
c8f79c2b CMN |
114 | { |
115 | khiter_t k; | |
116 | git_pack_cache_entry *entry = NULL; | |
117 | ||
09e29e47 CMN |
118 | if (git_mutex_lock(&cache->lock) < 0) |
119 | return NULL; | |
120 | ||
c8f79c2b CMN |
121 | k = kh_get(off, cache->entries, offset); |
122 | if (k != kh_end(cache->entries)) { /* found it */ | |
123 | entry = kh_value(cache->entries, k); | |
124 | git_atomic_inc(&entry->refcount); | |
0ed75620 | 125 | entry->last_usage = cache->use_ctr++; |
c8f79c2b CMN |
126 | } |
127 | git_mutex_unlock(&cache->lock); | |
128 | ||
129 | return entry; | |
130 | } | |
131 | ||
0ed75620 CMN |
132 | /* Run with the cache lock held */ |
133 | static void free_lowest_entry(git_pack_cache *cache) | |
134 | { | |
ed6648ba CMN |
135 | git_pack_cache_entry *entry; |
136 | khiter_t k; | |
137 | ||
0ed75620 CMN |
138 | for (k = kh_begin(cache->entries); k != kh_end(cache->entries); k++) { |
139 | if (!kh_exist(cache->entries, k)) | |
140 | continue; | |
141 | ||
142 | entry = kh_value(cache->entries, k); | |
ed6648ba | 143 | |
9c62aaab CMN |
144 | if (entry && entry->refcount.val == 0) { |
145 | cache->memory_used -= entry->raw.len; | |
146 | kh_del(off, cache->entries, k); | |
147 | free_cache_object(entry); | |
ed6648ba CMN |
148 | } |
149 | } | |
0ed75620 CMN |
150 | } |
151 | ||
c8f79c2b CMN |
152 | static int cache_add(git_pack_cache *cache, git_rawobj *base, git_off_t offset) |
153 | { | |
154 | git_pack_cache_entry *entry; | |
155 | int error, exists = 0; | |
156 | khiter_t k; | |
157 | ||
0ed75620 CMN |
158 | if (base->len > GIT_PACK_CACHE_SIZE_LIMIT) |
159 | return -1; | |
160 | ||
c8f79c2b CMN |
161 | entry = new_cache_object(base); |
162 | if (entry) { | |
09e29e47 CMN |
163 | if (git_mutex_lock(&cache->lock) < 0) { |
164 | giterr_set(GITERR_OS, "failed to lock cache"); | |
165 | return -1; | |
166 | } | |
c8f79c2b CMN |
167 | /* Add it to the cache if nobody else has */ |
168 | exists = kh_get(off, cache->entries, offset) != kh_end(cache->entries); | |
169 | if (!exists) { | |
0ed75620 CMN |
170 | while (cache->memory_used + base->len > cache->memory_limit) |
171 | free_lowest_entry(cache); | |
172 | ||
c8f79c2b CMN |
173 | k = kh_put(off, cache->entries, offset, &error); |
174 | assert(error != 0); | |
175 | kh_value(cache->entries, k) = entry; | |
0ed75620 | 176 | cache->memory_used += entry->raw.len; |
c8f79c2b CMN |
177 | } |
178 | git_mutex_unlock(&cache->lock); | |
179 | /* Somebody beat us to adding it into the cache */ | |
180 | if (exists) { | |
181 | git__free(entry); | |
182 | return -1; | |
183 | } | |
184 | } | |
185 | ||
186 | return 0; | |
187 | } | |
188 | ||
a070f152 CMN |
189 | /*********************************************************** |
190 | * | |
191 | * PACK INDEX METHODS | |
192 | * | |
193 | ***********************************************************/ | |
194 | ||
195 | static void pack_index_free(struct git_pack_file *p) | |
196 | { | |
60ecdf59 DMB |
197 | if (p->oids) { |
198 | git__free(p->oids); | |
199 | p->oids = NULL; | |
200 | } | |
a070f152 CMN |
201 | if (p->index_map.data) { |
202 | git_futils_mmap_free(&p->index_map); | |
203 | p->index_map.data = NULL; | |
204 | } | |
205 | } | |
206 | ||
87d9869f | 207 | static int pack_index_check(const char *path, struct git_pack_file *p) |
a070f152 CMN |
208 | { |
209 | struct git_pack_idx_header *hdr; | |
210 | uint32_t version, nr, i, *index; | |
a070f152 CMN |
211 | void *idx_map; |
212 | size_t idx_size; | |
a070f152 | 213 | struct stat st; |
a070f152 | 214 | int error; |
e1de726c RB |
215 | /* TODO: properly open the file without access time using O_NOATIME */ |
216 | git_file fd = git_futils_open_ro(path); | |
a070f152 | 217 | if (fd < 0) |
e1de726c | 218 | return fd; |
a070f152 | 219 | |
9d2f841a RB |
220 | if (p_fstat(fd, &st) < 0) { |
221 | p_close(fd); | |
222 | giterr_set(GITERR_OS, "Unable to stat pack index '%s'", path); | |
223 | return -1; | |
224 | } | |
225 | ||
226 | if (!S_ISREG(st.st_mode) || | |
e1de726c RB |
227 | !git__is_sizet(st.st_size) || |
228 | (idx_size = (size_t)st.st_size) < 4 * 256 + 20 + 20) | |
229 | { | |
a070f152 | 230 | p_close(fd); |
9d2f841a | 231 | giterr_set(GITERR_ODB, "Invalid pack index '%s'", path); |
e1de726c | 232 | return -1; |
a070f152 CMN |
233 | } |
234 | ||
235 | error = git_futils_mmap_ro(&p->index_map, fd, 0, idx_size); | |
e1de726c | 236 | |
a070f152 CMN |
237 | p_close(fd); |
238 | ||
e1de726c RB |
239 | if (error < 0) |
240 | return error; | |
a070f152 CMN |
241 | |
242 | hdr = idx_map = p->index_map.data; | |
243 | ||
244 | if (hdr->idx_signature == htonl(PACK_IDX_SIGNATURE)) { | |
245 | version = ntohl(hdr->idx_version); | |
246 | ||
247 | if (version < 2 || version > 2) { | |
248 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 249 | return packfile_error("unsupported index version"); |
a070f152 CMN |
250 | } |
251 | ||
252 | } else | |
253 | version = 1; | |
254 | ||
255 | nr = 0; | |
256 | index = idx_map; | |
257 | ||
258 | if (version > 1) | |
87d9869f | 259 | index += 2; /* skip index header */ |
a070f152 CMN |
260 | |
261 | for (i = 0; i < 256; i++) { | |
262 | uint32_t n = ntohl(index[i]); | |
263 | if (n < nr) { | |
264 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 265 | return packfile_error("index is non-monotonic"); |
a070f152 CMN |
266 | } |
267 | nr = n; | |
268 | } | |
269 | ||
270 | if (version == 1) { | |
271 | /* | |
272 | * Total size: | |
87d9869f VM |
273 | * - 256 index entries 4 bytes each |
274 | * - 24-byte entries * nr (20-byte sha1 + 4-byte offset) | |
275 | * - 20-byte SHA1 of the packfile | |
276 | * - 20-byte SHA1 file checksum | |
a070f152 CMN |
277 | */ |
278 | if (idx_size != 4*256 + nr * 24 + 20 + 20) { | |
279 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 280 | return packfile_error("index is corrupted"); |
a070f152 CMN |
281 | } |
282 | } else if (version == 2) { | |
283 | /* | |
284 | * Minimum size: | |
87d9869f VM |
285 | * - 8 bytes of header |
286 | * - 256 index entries 4 bytes each | |
287 | * - 20-byte sha1 entry * nr | |
288 | * - 4-byte crc entry * nr | |
289 | * - 4-byte offset entry * nr | |
290 | * - 20-byte SHA1 of the packfile | |
291 | * - 20-byte SHA1 file checksum | |
a070f152 CMN |
292 | * And after the 4-byte offset table might be a |
293 | * variable sized table containing 8-byte entries | |
294 | * for offsets larger than 2^31. | |
295 | */ | |
296 | unsigned long min_size = 8 + 4*256 + nr*(20 + 4 + 4) + 20 + 20; | |
297 | unsigned long max_size = min_size; | |
298 | ||
299 | if (nr) | |
300 | max_size += (nr - 1)*8; | |
301 | ||
302 | if (idx_size < min_size || idx_size > max_size) { | |
303 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 304 | return packfile_error("wrong index size"); |
a070f152 | 305 | } |
a070f152 CMN |
306 | } |
307 | ||
a070f152 | 308 | p->num_objects = nr; |
0ddfcb40 | 309 | p->index_version = version; |
e1de726c | 310 | return 0; |
a070f152 CMN |
311 | } |
312 | ||
313 | static int pack_index_open(struct git_pack_file *p) | |
314 | { | |
315 | char *idx_name; | |
24c70804 RB |
316 | int error = 0; |
317 | size_t name_len, base_len; | |
318 | ||
0ddfcb40 | 319 | if (p->index_version > -1) |
53607868 | 320 | return 0; |
a070f152 | 321 | |
24c70804 RB |
322 | name_len = strlen(p->pack_name); |
323 | assert(name_len > strlen(".pack")); /* checked by git_pack_file alloc */ | |
e1de726c | 324 | |
53607868 RB |
325 | if ((idx_name = git__malloc(name_len)) == NULL) |
326 | return -1; | |
44ef8b1b | 327 | |
24c70804 RB |
328 | base_len = name_len - strlen(".pack"); |
329 | memcpy(idx_name, p->pack_name, base_len); | |
330 | memcpy(idx_name + base_len, ".idx", sizeof(".idx")); | |
a070f152 | 331 | |
050af8bb RD |
332 | if ((error = git_mutex_lock(&p->lock)) < 0) { |
333 | git__free(idx_name); | |
53607868 | 334 | return error; |
050af8bb | 335 | } |
53607868 | 336 | |
0ddfcb40 | 337 | if (p->index_version == -1) |
53607868 | 338 | error = pack_index_check(idx_name, p); |
24c70804 | 339 | |
3286c408 | 340 | git__free(idx_name); |
a070f152 | 341 | |
24c70804 RB |
342 | git_mutex_unlock(&p->lock); |
343 | ||
e1de726c | 344 | return error; |
a070f152 CMN |
345 | } |
346 | ||
347 | static unsigned char *pack_window_open( | |
348 | struct git_pack_file *p, | |
7d0cdf82 | 349 | git_mwindow **w_cursor, |
e1de726c | 350 | git_off_t offset, |
7d0cdf82 CMN |
351 | unsigned int *left) |
352 | { | |
e1de726c | 353 | if (p->mwf.fd == -1 && packfile_open(p) < 0) |
7d0cdf82 CMN |
354 | return NULL; |
355 | ||
356 | /* Since packfiles end in a hash of their content and it's | |
357 | * pointless to ask for an offset into the middle of that | |
358 | * hash, and the pack_window_contains function above wouldn't match | |
359 | * don't allow an offset too close to the end of the file. | |
360 | */ | |
361 | if (offset > (p->mwf.size - 20)) | |
362 | return NULL; | |
363 | ||
364 | return git_mwindow_open(&p->mwf, w_cursor, offset, 20, left); | |
365 | } | |
366 | ||
51e82492 CMN |
367 | /* |
368 | * The per-object header is a pretty dense thing, which is | |
369 | * - first byte: low four bits are "size", | |
370 | * then three bits of "type", | |
371 | * with the high bit being "size continues". | |
372 | * - each byte afterwards: low seven bits are size continuation, | |
373 | * with the high bit being "size continues" | |
374 | */ | |
51a3dfb5 | 375 | size_t git_packfile__object_header(unsigned char *hdr, size_t size, git_otype type) |
51e82492 CMN |
376 | { |
377 | unsigned char *hdr_base; | |
378 | unsigned char c; | |
379 | ||
380 | assert(type >= GIT_OBJ_COMMIT && type <= GIT_OBJ_REF_DELTA); | |
381 | ||
382 | /* TODO: add support for chunked objects; see git.git 6c0d19b1 */ | |
383 | ||
384 | c = (unsigned char)((type << 4) | (size & 15)); | |
385 | size >>= 4; | |
386 | hdr_base = hdr; | |
387 | ||
388 | while (size) { | |
389 | *hdr++ = c | 0x80; | |
390 | c = size & 0x7f; | |
391 | size >>= 7; | |
392 | } | |
393 | *hdr++ = c; | |
394 | ||
51a3dfb5 | 395 | return (hdr - hdr_base); |
51e82492 CMN |
396 | } |
397 | ||
398 | ||
45d773ef CMN |
399 | static int packfile_unpack_header1( |
400 | unsigned long *usedp, | |
7d0cdf82 CMN |
401 | size_t *sizep, |
402 | git_otype *type, | |
403 | const unsigned char *buf, | |
404 | unsigned long len) | |
405 | { | |
406 | unsigned shift; | |
407 | unsigned long size, c; | |
408 | unsigned long used = 0; | |
2aeadb9c | 409 | |
7d0cdf82 CMN |
410 | c = buf[used++]; |
411 | *type = (c >> 4) & 7; | |
412 | size = c & 15; | |
413 | shift = 4; | |
414 | while (c & 0x80) { | |
45d773ef | 415 | if (len <= used) |
904b67e6 | 416 | return GIT_EBUFS; |
45d773ef CMN |
417 | |
418 | if (bitsizeof(long) <= shift) { | |
419 | *usedp = 0; | |
420 | return -1; | |
421 | } | |
7d0cdf82 CMN |
422 | |
423 | c = buf[used++]; | |
424 | size += (c & 0x7f) << shift; | |
425 | shift += 7; | |
426 | } | |
427 | ||
428 | *sizep = (size_t)size; | |
45d773ef CMN |
429 | *usedp = used; |
430 | return 0; | |
7d0cdf82 CMN |
431 | } |
432 | ||
433 | int git_packfile_unpack_header( | |
434 | size_t *size_p, | |
435 | git_otype *type_p, | |
436 | git_mwindow_file *mwf, | |
437 | git_mwindow **w_curs, | |
e1de726c | 438 | git_off_t *curpos) |
7d0cdf82 CMN |
439 | { |
440 | unsigned char *base; | |
441 | unsigned int left; | |
442 | unsigned long used; | |
45d773ef | 443 | int ret; |
7d0cdf82 CMN |
444 | |
445 | /* pack_window_open() assures us we have [base, base + 20) available | |
87d9869f VM |
446 | * as a range that we can look at at. (Its actually the hash |
447 | * size that is assured.) With our object header encoding | |
7d0cdf82 CMN |
448 | * the maximum deflated object size is 2^137, which is just |
449 | * insane, so we know won't exceed what we have been given. | |
450 | */ | |
24c70804 | 451 | /* base = pack_window_open(p, w_curs, *curpos, &left); */ |
7d0cdf82 CMN |
452 | base = git_mwindow_open(mwf, w_curs, *curpos, 20, &left); |
453 | if (base == NULL) | |
904b67e6 | 454 | return GIT_EBUFS; |
2aeadb9c | 455 | |
9d2f841a | 456 | ret = packfile_unpack_header1(&used, size_p, type_p, base, left); |
45d773ef | 457 | git_mwindow_close(w_curs); |
904b67e6 | 458 | if (ret == GIT_EBUFS) |
45d773ef CMN |
459 | return ret; |
460 | else if (ret < 0) | |
e1de726c | 461 | return packfile_error("header length is zero"); |
7d0cdf82 CMN |
462 | |
463 | *curpos += used; | |
e1de726c | 464 | return 0; |
7d0cdf82 CMN |
465 | } |
466 | ||
44f9f547 DMB |
467 | int git_packfile_resolve_header( |
468 | size_t *size_p, | |
469 | git_otype *type_p, | |
470 | struct git_pack_file *p, | |
471 | git_off_t offset) | |
472 | { | |
473 | git_mwindow *w_curs = NULL; | |
474 | git_off_t curpos = offset; | |
475 | size_t size; | |
476 | git_otype type; | |
477 | git_off_t base_offset; | |
478 | int error; | |
479 | ||
480 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
481 | git_mwindow_close(&w_curs); | |
482 | if (error < 0) | |
483 | return error; | |
484 | ||
485 | if (type == GIT_OBJ_OFS_DELTA || type == GIT_OBJ_REF_DELTA) { | |
486 | size_t base_size; | |
487 | git_rawobj delta; | |
488 | base_offset = get_delta_base(p, &w_curs, &curpos, type, offset); | |
489 | git_mwindow_close(&w_curs); | |
490 | error = packfile_unpack_compressed(&delta, p, &w_curs, &curpos, size, type); | |
491 | git_mwindow_close(&w_curs); | |
492 | if (error < 0) | |
493 | return error; | |
494 | error = git__delta_read_header(delta.data, delta.len, &base_size, size_p); | |
495 | git__free(delta.data); | |
496 | if (error < 0) | |
497 | return error; | |
498 | } else | |
499 | *size_p = size; | |
500 | ||
501 | while (type == GIT_OBJ_OFS_DELTA || type == GIT_OBJ_REF_DELTA) { | |
502 | curpos = base_offset; | |
503 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
504 | git_mwindow_close(&w_curs); | |
505 | if (error < 0) | |
506 | return error; | |
507 | if (type != GIT_OBJ_OFS_DELTA && type != GIT_OBJ_REF_DELTA) | |
508 | break; | |
509 | base_offset = get_delta_base(p, &w_curs, &curpos, type, base_offset); | |
510 | git_mwindow_close(&w_curs); | |
511 | } | |
512 | *type_p = type; | |
513 | ||
514 | return error; | |
515 | } | |
516 | ||
a070f152 | 517 | static int packfile_unpack_delta( |
7d0cdf82 | 518 | git_rawobj *obj, |
a070f152 | 519 | struct git_pack_file *p, |
7d0cdf82 | 520 | git_mwindow **w_curs, |
e1de726c | 521 | git_off_t *curpos, |
7d0cdf82 CMN |
522 | size_t delta_size, |
523 | git_otype delta_type, | |
e1de726c | 524 | git_off_t obj_offset) |
7d0cdf82 | 525 | { |
c0f4a011 | 526 | git_off_t base_offset, base_key; |
7d0cdf82 | 527 | git_rawobj base, delta; |
525d961c | 528 | git_pack_cache_entry *cached = NULL; |
c0f4a011 | 529 | int error, found_base = 0; |
7d0cdf82 | 530 | |
b5b474dd | 531 | base_offset = get_delta_base(p, w_curs, curpos, delta_type, obj_offset); |
45d773ef | 532 | git_mwindow_close(w_curs); |
7d0cdf82 | 533 | if (base_offset == 0) |
e1de726c RB |
534 | return packfile_error("delta offset is zero"); |
535 | if (base_offset < 0) /* must actually be an error code */ | |
536 | return (int)base_offset; | |
7d0cdf82 | 537 | |
c8f79c2b CMN |
538 | if (!p->bases.entries && (cache_init(&p->bases) < 0)) |
539 | return -1; | |
7d0cdf82 | 540 | |
c0f4a011 | 541 | base_key = base_offset; /* git_packfile_unpack modifies base_offset */ |
c8f79c2b | 542 | if ((cached = cache_get(&p->bases, base_offset)) != NULL) { |
c0f4a011 | 543 | memcpy(&base, &cached->raw, sizeof(git_rawobj)); |
c8f79c2b | 544 | found_base = 1; |
525d961c | 545 | } |
525d961c CMN |
546 | |
547 | if (!cached) { /* have to inflate it */ | |
c0f4a011 CMN |
548 | error = git_packfile_unpack(&base, p, &base_offset); |
549 | ||
550 | /* | |
551 | * TODO: git.git tries to load the base from other packfiles | |
552 | * or loose objects. | |
553 | * | |
554 | * We'll need to do this in order to support thin packs. | |
555 | */ | |
556 | if (error < 0) | |
557 | return error; | |
558 | } | |
7d0cdf82 CMN |
559 | |
560 | error = packfile_unpack_compressed(&delta, p, w_curs, curpos, delta_size, delta_type); | |
45d773ef | 561 | git_mwindow_close(w_curs); |
c0f4a011 | 562 | |
e1de726c | 563 | if (error < 0) { |
c0f4a011 CMN |
564 | if (!found_base) |
565 | git__free(base.data); | |
e1de726c | 566 | return error; |
7d0cdf82 CMN |
567 | } |
568 | ||
569 | obj->type = base.type; | |
e1de726c | 570 | error = git__delta_apply(obj, base.data, base.len, delta.data, delta.len); |
c0f4a011 CMN |
571 | if (error < 0) |
572 | goto on_error; | |
573 | ||
c8f79c2b | 574 | if (found_base) |
525d961c | 575 | git_atomic_dec(&cached->refcount); |
c8f79c2b CMN |
576 | else if (cache_add(&p->bases, &base, base_key) < 0) |
577 | git__free(base.data); | |
7d0cdf82 | 578 | |
c0f4a011 | 579 | on_error: |
3286c408 | 580 | git__free(delta.data); |
7d0cdf82 | 581 | |
7d0cdf82 CMN |
582 | return error; /* error set by git__delta_apply */ |
583 | } | |
584 | ||
a070f152 | 585 | int git_packfile_unpack( |
e1de726c RB |
586 | git_rawobj *obj, |
587 | struct git_pack_file *p, | |
588 | git_off_t *obj_offset) | |
7d0cdf82 CMN |
589 | { |
590 | git_mwindow *w_curs = NULL; | |
e1de726c | 591 | git_off_t curpos = *obj_offset; |
7d0cdf82 CMN |
592 | int error; |
593 | ||
594 | size_t size = 0; | |
595 | git_otype type; | |
596 | ||
597 | /* | |
598 | * TODO: optionally check the CRC on the packfile | |
599 | */ | |
600 | ||
601 | obj->data = NULL; | |
602 | obj->len = 0; | |
603 | obj->type = GIT_OBJ_BAD; | |
604 | ||
605 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
45d773ef CMN |
606 | git_mwindow_close(&w_curs); |
607 | ||
e1de726c RB |
608 | if (error < 0) |
609 | return error; | |
7d0cdf82 CMN |
610 | |
611 | switch (type) { | |
612 | case GIT_OBJ_OFS_DELTA: | |
613 | case GIT_OBJ_REF_DELTA: | |
614 | error = packfile_unpack_delta( | |
b5b474dd CMN |
615 | obj, p, &w_curs, &curpos, |
616 | size, type, *obj_offset); | |
7d0cdf82 CMN |
617 | break; |
618 | ||
619 | case GIT_OBJ_COMMIT: | |
620 | case GIT_OBJ_TREE: | |
621 | case GIT_OBJ_BLOB: | |
622 | case GIT_OBJ_TAG: | |
623 | error = packfile_unpack_compressed( | |
b5b474dd | 624 | obj, p, &w_curs, &curpos, |
7d0cdf82 CMN |
625 | size, type); |
626 | break; | |
627 | ||
628 | default: | |
e1de726c | 629 | error = packfile_error("invalid packfile type in header");; |
7d0cdf82 CMN |
630 | break; |
631 | } | |
632 | ||
b5b474dd | 633 | *obj_offset = curpos; |
e1de726c | 634 | return error; |
7d0cdf82 CMN |
635 | } |
636 | ||
282283ac RB |
637 | static void *use_git_alloc(void *opaq, unsigned int count, unsigned int size) |
638 | { | |
639 | GIT_UNUSED(opaq); | |
640 | return git__calloc(count, size); | |
641 | } | |
642 | ||
643 | static void use_git_free(void *opaq, void *ptr) | |
644 | { | |
645 | GIT_UNUSED(opaq); | |
646 | git__free(ptr); | |
647 | } | |
648 | ||
46635339 CMN |
649 | int git_packfile_stream_open(git_packfile_stream *obj, struct git_pack_file *p, git_off_t curpos) |
650 | { | |
651 | int st; | |
652 | ||
653 | memset(obj, 0, sizeof(git_packfile_stream)); | |
654 | obj->curpos = curpos; | |
655 | obj->p = p; | |
656 | obj->zstream.zalloc = use_git_alloc; | |
657 | obj->zstream.zfree = use_git_free; | |
658 | obj->zstream.next_in = Z_NULL; | |
659 | obj->zstream.next_out = Z_NULL; | |
660 | st = inflateInit(&obj->zstream); | |
661 | if (st != Z_OK) { | |
662 | git__free(obj); | |
663 | giterr_set(GITERR_ZLIB, "Failed to inflate packfile"); | |
664 | return -1; | |
665 | } | |
666 | ||
667 | return 0; | |
668 | } | |
669 | ||
670 | ssize_t git_packfile_stream_read(git_packfile_stream *obj, void *buffer, size_t len) | |
671 | { | |
672 | unsigned char *in; | |
673 | size_t written; | |
674 | int st; | |
675 | ||
676 | if (obj->done) | |
677 | return 0; | |
678 | ||
679 | in = pack_window_open(obj->p, &obj->mw, obj->curpos, &obj->zstream.avail_in); | |
680 | if (in == NULL) | |
681 | return GIT_EBUFS; | |
682 | ||
683 | obj->zstream.next_out = buffer; | |
090d5e1f | 684 | obj->zstream.avail_out = (unsigned int)len; |
46635339 CMN |
685 | obj->zstream.next_in = in; |
686 | ||
687 | st = inflate(&obj->zstream, Z_SYNC_FLUSH); | |
688 | git_mwindow_close(&obj->mw); | |
689 | ||
690 | obj->curpos += obj->zstream.next_in - in; | |
691 | written = len - obj->zstream.avail_out; | |
692 | ||
693 | if (st != Z_OK && st != Z_STREAM_END) { | |
694 | giterr_set(GITERR_ZLIB, "Failed to inflate packfile"); | |
695 | return -1; | |
696 | } | |
697 | ||
698 | if (st == Z_STREAM_END) | |
699 | obj->done = 1; | |
700 | ||
701 | ||
702 | /* If we didn't write anything out but we're not done, we need more data */ | |
703 | if (!written && st != Z_STREAM_END) | |
704 | return GIT_EBUFS; | |
705 | ||
706 | return written; | |
707 | ||
708 | } | |
709 | ||
710 | void git_packfile_stream_free(git_packfile_stream *obj) | |
711 | { | |
712 | inflateEnd(&obj->zstream); | |
713 | } | |
714 | ||
7d0cdf82 | 715 | int packfile_unpack_compressed( |
e1de726c RB |
716 | git_rawobj *obj, |
717 | struct git_pack_file *p, | |
718 | git_mwindow **w_curs, | |
719 | git_off_t *curpos, | |
720 | size_t size, | |
721 | git_otype type) | |
7d0cdf82 CMN |
722 | { |
723 | int st; | |
724 | z_stream stream; | |
725 | unsigned char *buffer, *in; | |
726 | ||
e1de726c RB |
727 | buffer = git__calloc(1, size + 1); |
728 | GITERR_CHECK_ALLOC(buffer); | |
7d0cdf82 CMN |
729 | |
730 | memset(&stream, 0, sizeof(stream)); | |
731 | stream.next_out = buffer; | |
1c3fac4d | 732 | stream.avail_out = (uInt)size + 1; |
282283ac RB |
733 | stream.zalloc = use_git_alloc; |
734 | stream.zfree = use_git_free; | |
7d0cdf82 CMN |
735 | |
736 | st = inflateInit(&stream); | |
737 | if (st != Z_OK) { | |
3286c408 | 738 | git__free(buffer); |
e1de726c | 739 | giterr_set(GITERR_ZLIB, "Failed to inflate packfile"); |
45d773ef | 740 | |
e1de726c | 741 | return -1; |
7d0cdf82 CMN |
742 | } |
743 | ||
744 | do { | |
b5b474dd | 745 | in = pack_window_open(p, w_curs, *curpos, &stream.avail_in); |
7d0cdf82 CMN |
746 | stream.next_in = in; |
747 | st = inflate(&stream, Z_FINISH); | |
45d773ef | 748 | git_mwindow_close(w_curs); |
7d0cdf82 CMN |
749 | |
750 | if (!stream.avail_out) | |
751 | break; /* the payload is larger than it should be */ | |
752 | ||
45d773ef CMN |
753 | if (st == Z_BUF_ERROR && in == NULL) { |
754 | inflateEnd(&stream); | |
755 | git__free(buffer); | |
904b67e6 | 756 | return GIT_EBUFS; |
45d773ef CMN |
757 | } |
758 | ||
b5b474dd | 759 | *curpos += stream.next_in - in; |
7d0cdf82 CMN |
760 | } while (st == Z_OK || st == Z_BUF_ERROR); |
761 | ||
762 | inflateEnd(&stream); | |
763 | ||
764 | if ((st != Z_STREAM_END) || stream.total_out != size) { | |
3286c408 | 765 | git__free(buffer); |
e1de726c RB |
766 | giterr_set(GITERR_ZLIB, "Failed to inflate packfile"); |
767 | return -1; | |
7d0cdf82 CMN |
768 | } |
769 | ||
770 | obj->type = type; | |
771 | obj->len = size; | |
772 | obj->data = buffer; | |
e1de726c | 773 | return 0; |
7d0cdf82 CMN |
774 | } |
775 | ||
b5b474dd CMN |
776 | /* |
777 | * curpos is where the data starts, delta_obj_offset is the where the | |
778 | * header starts | |
779 | */ | |
e1de726c RB |
780 | git_off_t get_delta_base( |
781 | struct git_pack_file *p, | |
782 | git_mwindow **w_curs, | |
783 | git_off_t *curpos, | |
784 | git_otype type, | |
785 | git_off_t delta_obj_offset) | |
7d0cdf82 | 786 | { |
45d773ef CMN |
787 | unsigned int left = 0; |
788 | unsigned char *base_info; | |
e1de726c | 789 | git_off_t base_offset; |
7d0cdf82 CMN |
790 | git_oid unused; |
791 | ||
45d773ef CMN |
792 | base_info = pack_window_open(p, w_curs, *curpos, &left); |
793 | /* Assumption: the only reason this would fail is because the file is too small */ | |
794 | if (base_info == NULL) | |
904b67e6 | 795 | return GIT_EBUFS; |
7d0cdf82 CMN |
796 | /* pack_window_open() assured us we have [base_info, base_info + 20) |
797 | * as a range that we can look at without walking off the | |
87d9869f VM |
798 | * end of the mapped window. Its actually the hash size |
799 | * that is assured. An OFS_DELTA longer than the hash size | |
7d0cdf82 CMN |
800 | * is stupid, as then a REF_DELTA would be smaller to store. |
801 | */ | |
802 | if (type == GIT_OBJ_OFS_DELTA) { | |
803 | unsigned used = 0; | |
804 | unsigned char c = base_info[used++]; | |
805 | base_offset = c & 127; | |
806 | while (c & 128) { | |
45d773ef | 807 | if (left <= used) |
904b67e6 | 808 | return GIT_EBUFS; |
7d0cdf82 CMN |
809 | base_offset += 1; |
810 | if (!base_offset || MSB(base_offset, 7)) | |
87d9869f | 811 | return 0; /* overflow */ |
7d0cdf82 CMN |
812 | c = base_info[used++]; |
813 | base_offset = (base_offset << 7) + (c & 127); | |
814 | } | |
815 | base_offset = delta_obj_offset - base_offset; | |
816 | if (base_offset <= 0 || base_offset >= delta_obj_offset) | |
87d9869f | 817 | return 0; /* out of bound */ |
7d0cdf82 CMN |
818 | *curpos += used; |
819 | } else if (type == GIT_OBJ_REF_DELTA) { | |
c1af5a39 CMN |
820 | /* If we have the cooperative cache, search in it first */ |
821 | if (p->has_cache) { | |
0e040c03 CMN |
822 | khiter_t k; |
823 | git_oid oid; | |
c1af5a39 | 824 | |
0e040c03 CMN |
825 | git_oid_fromraw(&oid, base_info); |
826 | k = kh_get(oid, p->idx_cache, &oid); | |
827 | if (k != kh_end(p->idx_cache)) { | |
c1af5a39 | 828 | *curpos += 20; |
0e040c03 | 829 | return ((struct git_pack_entry *)kh_value(p->idx_cache, k))->offset; |
c1af5a39 CMN |
830 | } |
831 | } | |
7d0cdf82 | 832 | /* The base entry _must_ be in the same pack */ |
e1de726c RB |
833 | if (pack_entry_find_offset(&base_offset, &unused, p, (git_oid *)base_info, GIT_OID_HEXSZ) < 0) |
834 | return packfile_error("base entry delta is not in the same pack"); | |
7d0cdf82 CMN |
835 | *curpos += 20; |
836 | } else | |
837 | return 0; | |
838 | ||
839 | return base_offset; | |
840 | } | |
a070f152 CMN |
841 | |
842 | /*********************************************************** | |
843 | * | |
844 | * PACKFILE METHODS | |
845 | * | |
846 | ***********************************************************/ | |
847 | ||
96c9b9f0 | 848 | void git_packfile_free(struct git_pack_file *p) |
a070f152 | 849 | { |
24c70804 RB |
850 | if (!p) |
851 | return; | |
852 | ||
c8f79c2b | 853 | cache_free(&p->bases); |
c0f4a011 | 854 | |
a070f152 CMN |
855 | git_mwindow_free_all(&p->mwf); |
856 | ||
3a2d48d5 | 857 | if (p->mwf.fd >= 0) |
a070f152 CMN |
858 | p_close(p->mwf.fd); |
859 | ||
860 | pack_index_free(p); | |
861 | ||
3286c408 | 862 | git__free(p->bad_object_sha1); |
24c70804 | 863 | |
24c70804 | 864 | git_mutex_free(&p->lock); |
3286c408 | 865 | git__free(p); |
a070f152 CMN |
866 | } |
867 | ||
868 | static int packfile_open(struct git_pack_file *p) | |
869 | { | |
870 | struct stat st; | |
871 | struct git_pack_header hdr; | |
872 | git_oid sha1; | |
873 | unsigned char *idx_sha1; | |
874 | ||
0ddfcb40 | 875 | if (p->index_version == -1 && pack_index_open(p) < 0) |
282283ac | 876 | return git_odb__error_notfound("failed to open packfile", NULL); |
a070f152 | 877 | |
9d2f841a RB |
878 | /* if mwf opened by another thread, return now */ |
879 | if (git_mutex_lock(&p->lock) < 0) | |
880 | return packfile_error("failed to get lock for open"); | |
881 | ||
882 | if (p->mwf.fd >= 0) { | |
883 | git_mutex_unlock(&p->lock); | |
884 | return 0; | |
885 | } | |
886 | ||
a070f152 | 887 | /* TODO: open with noatime */ |
e1de726c | 888 | p->mwf.fd = git_futils_open_ro(p->pack_name); |
9d2f841a RB |
889 | if (p->mwf.fd < 0) |
890 | goto cleanup; | |
a070f152 | 891 | |
e1de726c RB |
892 | if (p_fstat(p->mwf.fd, &st) < 0 || |
893 | git_mwindow_file_register(&p->mwf) < 0) | |
894 | goto cleanup; | |
a070f152 CMN |
895 | |
896 | /* If we created the struct before we had the pack we lack size. */ | |
897 | if (!p->mwf.size) { | |
898 | if (!S_ISREG(st.st_mode)) | |
899 | goto cleanup; | |
e1de726c | 900 | p->mwf.size = (git_off_t)st.st_size; |
a070f152 CMN |
901 | } else if (p->mwf.size != st.st_size) |
902 | goto cleanup; | |
903 | ||
904 | #if 0 | |
905 | /* We leave these file descriptors open with sliding mmap; | |
906 | * there is no point keeping them open across exec(), though. | |
907 | */ | |
908 | fd_flag = fcntl(p->mwf.fd, F_GETFD, 0); | |
909 | if (fd_flag < 0) | |
e1de726c | 910 | goto cleanup; |
a070f152 CMN |
911 | |
912 | fd_flag |= FD_CLOEXEC; | |
913 | if (fcntl(p->pack_fd, F_SETFD, fd_flag) == -1) | |
e1de726c | 914 | goto cleanup; |
a070f152 CMN |
915 | #endif |
916 | ||
917 | /* Verify we recognize this pack file format. */ | |
e1de726c RB |
918 | if (p_read(p->mwf.fd, &hdr, sizeof(hdr)) < 0 || |
919 | hdr.hdr_signature != htonl(PACK_SIGNATURE) || | |
920 | !pack_version_ok(hdr.hdr_version)) | |
a070f152 CMN |
921 | goto cleanup; |
922 | ||
923 | /* Verify the pack matches its index. */ | |
e1de726c RB |
924 | if (p->num_objects != ntohl(hdr.hdr_entries) || |
925 | p_lseek(p->mwf.fd, p->mwf.size - GIT_OID_RAWSZ, SEEK_SET) == -1 || | |
926 | p_read(p->mwf.fd, sha1.id, GIT_OID_RAWSZ) < 0) | |
a070f152 CMN |
927 | goto cleanup; |
928 | ||
929 | idx_sha1 = ((unsigned char *)p->index_map.data) + p->index_map.len - 40; | |
930 | ||
9d2f841a RB |
931 | if (git_oid__cmp(&sha1, (git_oid *)idx_sha1) != 0) |
932 | goto cleanup; | |
933 | ||
934 | git_mutex_unlock(&p->lock); | |
935 | return 0; | |
a070f152 CMN |
936 | |
937 | cleanup: | |
e1de726c | 938 | giterr_set(GITERR_OS, "Invalid packfile '%s'", p->pack_name); |
9d2f841a | 939 | |
3a2d48d5 SS |
940 | if (p->mwf.fd >= 0) |
941 | p_close(p->mwf.fd); | |
a070f152 | 942 | p->mwf.fd = -1; |
9d2f841a RB |
943 | |
944 | git_mutex_unlock(&p->lock); | |
945 | ||
e1de726c | 946 | return -1; |
a070f152 CMN |
947 | } |
948 | ||
5d2d21e5 | 949 | int git_packfile_alloc(struct git_pack_file **pack_out, const char *path) |
a070f152 CMN |
950 | { |
951 | struct stat st; | |
952 | struct git_pack_file *p; | |
38eef611 | 953 | size_t path_len = path ? strlen(path) : 0; |
a070f152 CMN |
954 | |
955 | *pack_out = NULL; | |
24c70804 | 956 | |
38eef611 | 957 | if (path_len < strlen(".idx")) |
24c70804 RB |
958 | return git_odb__error_notfound("invalid packfile path", NULL); |
959 | ||
38eef611 | 960 | p = git__calloc(1, sizeof(*p) + path_len + 2); |
e1de726c | 961 | GITERR_CHECK_ALLOC(p); |
a070f152 | 962 | |
38eef611 RB |
963 | memcpy(p->pack_name, path, path_len + 1); |
964 | ||
a070f152 CMN |
965 | /* |
966 | * Make sure a corresponding .pack file exists and that | |
967 | * the index looks sane. | |
968 | */ | |
38eef611 RB |
969 | if (git__suffixcmp(path, ".idx") == 0) { |
970 | size_t root_len = path_len - strlen(".idx"); | |
971 | ||
972 | memcpy(p->pack_name + root_len, ".keep", sizeof(".keep")); | |
973 | if (git_path_exists(p->pack_name) == true) | |
974 | p->pack_keep = 1; | |
a070f152 | 975 | |
38eef611 RB |
976 | memcpy(p->pack_name + root_len, ".pack", sizeof(".pack")); |
977 | path_len = path_len - strlen(".idx") + strlen(".pack"); | |
978 | } | |
a070f152 | 979 | |
e1de726c | 980 | if (p_stat(p->pack_name, &st) < 0 || !S_ISREG(st.st_mode)) { |
3286c408 | 981 | git__free(p); |
282283ac | 982 | return git_odb__error_notfound("packfile not found", NULL); |
a070f152 CMN |
983 | } |
984 | ||
985 | /* ok, it looks sane as far as we can check without | |
986 | * actually mapping the pack file. | |
987 | */ | |
38eef611 | 988 | p->mwf.fd = -1; |
1af56d7d | 989 | p->mwf.size = st.st_size; |
a070f152 CMN |
990 | p->pack_local = 1; |
991 | p->mtime = (git_time_t)st.st_mtime; | |
0ddfcb40 | 992 | p->index_version = -1; |
a070f152 | 993 | |
1a42dd17 RB |
994 | if (git_mutex_init(&p->lock)) { |
995 | giterr_set(GITERR_OS, "Failed to initialize packfile mutex"); | |
996 | git__free(p); | |
997 | return -1; | |
998 | } | |
38eef611 | 999 | |
a070f152 CMN |
1000 | /* see if we can parse the sha1 oid in the packfile name */ |
1001 | if (path_len < 40 || | |
e1de726c | 1002 | git_oid_fromstr(&p->sha1, path + path_len - GIT_OID_HEXSZ) < 0) |
a070f152 CMN |
1003 | memset(&p->sha1, 0x0, GIT_OID_RAWSZ); |
1004 | ||
1005 | *pack_out = p; | |
e1de726c RB |
1006 | |
1007 | return 0; | |
a070f152 CMN |
1008 | } |
1009 | ||
1010 | /*********************************************************** | |
1011 | * | |
1012 | * PACKFILE ENTRY SEARCH INTERNALS | |
1013 | * | |
1014 | ***********************************************************/ | |
1015 | ||
e1de726c | 1016 | static git_off_t nth_packed_object_offset(const struct git_pack_file *p, uint32_t n) |
a070f152 CMN |
1017 | { |
1018 | const unsigned char *index = p->index_map.data; | |
1019 | index += 4 * 256; | |
1020 | if (p->index_version == 1) { | |
1021 | return ntohl(*((uint32_t *)(index + 24 * n))); | |
1022 | } else { | |
1023 | uint32_t off; | |
1024 | index += 8 + p->num_objects * (20 + 4); | |
1025 | off = ntohl(*((uint32_t *)(index + 4 * n))); | |
1026 | if (!(off & 0x80000000)) | |
1027 | return off; | |
1028 | index += p->num_objects * 4 + (off & 0x7fffffff) * 8; | |
1029 | return (((uint64_t)ntohl(*((uint32_t *)(index + 0)))) << 32) | | |
87d9869f | 1030 | ntohl(*((uint32_t *)(index + 4))); |
a070f152 CMN |
1031 | } |
1032 | } | |
1033 | ||
60ecdf59 DMB |
1034 | static int git__memcmp4(const void *a, const void *b) { |
1035 | return memcmp(a, b, 4); | |
1036 | } | |
1037 | ||
521aedad | 1038 | int git_pack_foreach_entry( |
5dca2010 | 1039 | struct git_pack_file *p, |
c3fb7d04 | 1040 | git_odb_foreach_cb cb, |
5dca2010 | 1041 | void *data) |
521aedad CMN |
1042 | { |
1043 | const unsigned char *index = p->index_map.data, *current; | |
521aedad | 1044 | uint32_t i; |
dab89f9b | 1045 | int error = 0; |
521aedad CMN |
1046 | |
1047 | if (index == NULL) { | |
521aedad CMN |
1048 | if ((error = pack_index_open(p)) < 0) |
1049 | return error; | |
1050 | ||
1051 | assert(p->index_map.data); | |
1052 | ||
1053 | index = p->index_map.data; | |
1054 | } | |
1055 | ||
1056 | if (p->index_version > 1) { | |
1057 | index += 8; | |
1058 | } | |
1059 | ||
1060 | index += 4 * 256; | |
1061 | ||
60ecdf59 DMB |
1062 | if (p->oids == NULL) { |
1063 | git_vector offsets, oids; | |
521aedad | 1064 | |
60ecdf59 DMB |
1065 | if ((error = git_vector_init(&oids, p->num_objects, NULL))) |
1066 | return error; | |
1067 | ||
1068 | if ((error = git_vector_init(&offsets, p->num_objects, git__memcmp4))) | |
1069 | return error; | |
5dca2010 | 1070 | |
60ecdf59 DMB |
1071 | if (p->index_version > 1) { |
1072 | const unsigned char *off = index + 24 * p->num_objects; | |
1073 | for (i = 0; i < p->num_objects; i++) | |
1074 | git_vector_insert(&offsets, (void*)&off[4 * i]); | |
1075 | git_vector_sort(&offsets); | |
1076 | git_vector_foreach(&offsets, i, current) | |
1077 | git_vector_insert(&oids, (void*)&index[5 * (current - off)]); | |
1078 | } else { | |
1079 | for (i = 0; i < p->num_objects; i++) | |
1080 | git_vector_insert(&offsets, (void*)&index[24 * i]); | |
1081 | git_vector_sort(&offsets); | |
1082 | git_vector_foreach(&offsets, i, current) | |
1083 | git_vector_insert(&oids, (void*)¤t[4]); | |
1084 | } | |
25e0b157 | 1085 | |
60ecdf59 | 1086 | git_vector_free(&offsets); |
25e0b157 | 1087 | p->oids = (git_oid **)git_vector_detach(NULL, NULL, &oids); |
521aedad CMN |
1088 | } |
1089 | ||
60ecdf59 | 1090 | for (i = 0; i < p->num_objects; i++) |
c7b3e1b3 RB |
1091 | if ((error = cb(p->oids[i], data)) != 0) { |
1092 | GITERR_CALLBACK(error); | |
25e0b157 | 1093 | break; |
c7b3e1b3 | 1094 | } |
60ecdf59 | 1095 | |
25e0b157 | 1096 | return error; |
521aedad CMN |
1097 | } |
1098 | ||
a070f152 | 1099 | static int pack_entry_find_offset( |
e1de726c RB |
1100 | git_off_t *offset_out, |
1101 | git_oid *found_oid, | |
1102 | struct git_pack_file *p, | |
1103 | const git_oid *short_oid, | |
b8457baa | 1104 | size_t len) |
a070f152 | 1105 | { |
34bd5999 CMN |
1106 | const uint32_t *level1_ofs = p->index_map.data; |
1107 | const unsigned char *index = p->index_map.data; | |
a070f152 CMN |
1108 | unsigned hi, lo, stride; |
1109 | int pos, found = 0; | |
1110 | const unsigned char *current = 0; | |
1111 | ||
1112 | *offset_out = 0; | |
1113 | ||
0ddfcb40 | 1114 | if (p->index_version == -1) { |
34bd5999 | 1115 | int error; |
a070f152 | 1116 | |
34bd5999 CMN |
1117 | if ((error = pack_index_open(p)) < 0) |
1118 | return error; | |
1119 | assert(p->index_map.data); | |
8c535f3f | 1120 | |
34bd5999 CMN |
1121 | index = p->index_map.data; |
1122 | level1_ofs = p->index_map.data; | |
1123 | } | |
a070f152 CMN |
1124 | |
1125 | if (p->index_version > 1) { | |
1126 | level1_ofs += 2; | |
1127 | index += 8; | |
1128 | } | |
1129 | ||
1130 | index += 4 * 256; | |
1131 | hi = ntohl(level1_ofs[(int)short_oid->id[0]]); | |
1132 | lo = ((short_oid->id[0] == 0x0) ? 0 : ntohl(level1_ofs[(int)short_oid->id[0] - 1])); | |
1133 | ||
1134 | if (p->index_version > 1) { | |
1135 | stride = 20; | |
1136 | } else { | |
1137 | stride = 24; | |
1138 | index += 4; | |
1139 | } | |
1140 | ||
1141 | #ifdef INDEX_DEBUG_LOOKUP | |
1142 | printf("%02x%02x%02x... lo %u hi %u nr %d\n", | |
1143 | short_oid->id[0], short_oid->id[1], short_oid->id[2], lo, hi, p->num_objects); | |
1144 | #endif | |
1145 | ||
67591c8c | 1146 | #ifdef GIT_USE_LOOKUP |
87d9869f | 1147 | pos = sha1_entry_pos(index, stride, 0, lo, hi, p->num_objects, short_oid->id); |
67591c8c VM |
1148 | #else |
1149 | pos = sha1_position(index, stride, lo, hi, short_oid->id); | |
1150 | #endif | |
a070f152 CMN |
1151 | |
1152 | if (pos >= 0) { | |
1153 | /* An object matching exactly the oid was found */ | |
1154 | found = 1; | |
1155 | current = index + pos * stride; | |
1156 | } else { | |
1157 | /* No object was found */ | |
1158 | /* pos refers to the object with the "closest" oid to short_oid */ | |
1159 | pos = - 1 - pos; | |
1160 | if (pos < (int)p->num_objects) { | |
1161 | current = index + pos * stride; | |
1162 | ||
282283ac | 1163 | if (!git_oid_ncmp(short_oid, (const git_oid *)current, len)) |
a070f152 | 1164 | found = 1; |
a070f152 CMN |
1165 | } |
1166 | } | |
1167 | ||
b2a2702d | 1168 | if (found && len != GIT_OID_HEXSZ && pos + 1 < (int)p->num_objects) { |
a070f152 CMN |
1169 | /* Check for ambiguousity */ |
1170 | const unsigned char *next = current + stride; | |
1171 | ||
1172 | if (!git_oid_ncmp(short_oid, (const git_oid *)next, len)) { | |
1173 | found = 2; | |
1174 | } | |
1175 | } | |
1176 | ||
e1de726c | 1177 | if (!found) |
282283ac | 1178 | return git_odb__error_notfound("failed to find offset for pack entry", short_oid); |
e1de726c RB |
1179 | if (found > 1) |
1180 | return git_odb__error_ambiguous("found multiple offsets for pack entry"); | |
24c70804 | 1181 | |
e1de726c RB |
1182 | *offset_out = nth_packed_object_offset(p, pos); |
1183 | git_oid_fromraw(found_oid, current); | |
a070f152 CMN |
1184 | |
1185 | #ifdef INDEX_DEBUG_LOOKUP | |
e1de726c | 1186 | { |
a070f152 CMN |
1187 | unsigned char hex_sha1[GIT_OID_HEXSZ + 1]; |
1188 | git_oid_fmt(hex_sha1, found_oid); | |
1189 | hex_sha1[GIT_OID_HEXSZ] = '\0'; | |
1190 | printf("found lo=%d %s\n", lo, hex_sha1); | |
a070f152 | 1191 | } |
e1de726c | 1192 | #endif |
24c70804 | 1193 | |
e1de726c | 1194 | return 0; |
a070f152 CMN |
1195 | } |
1196 | ||
1197 | int git_pack_entry_find( | |
1198 | struct git_pack_entry *e, | |
1199 | struct git_pack_file *p, | |
1200 | const git_oid *short_oid, | |
b8457baa | 1201 | size_t len) |
a070f152 | 1202 | { |
e1de726c | 1203 | git_off_t offset; |
a070f152 CMN |
1204 | git_oid found_oid; |
1205 | int error; | |
1206 | ||
1207 | assert(p); | |
1208 | ||
1209 | if (len == GIT_OID_HEXSZ && p->num_bad_objects) { | |
1210 | unsigned i; | |
1211 | for (i = 0; i < p->num_bad_objects; i++) | |
b7f167da | 1212 | if (git_oid__cmp(short_oid, &p->bad_object_sha1[i]) == 0) |
e1de726c | 1213 | return packfile_error("bad object found in packfile"); |
a070f152 CMN |
1214 | } |
1215 | ||
1216 | error = pack_entry_find_offset(&offset, &found_oid, p, short_oid, len); | |
e1de726c RB |
1217 | if (error < 0) |
1218 | return error; | |
a070f152 CMN |
1219 | |
1220 | /* we found a unique entry in the index; | |
1221 | * make sure the packfile backing the index | |
1222 | * still exists on disk */ | |
e1de726c RB |
1223 | if (p->mwf.fd == -1 && (error = packfile_open(p)) < 0) |
1224 | return error; | |
a070f152 CMN |
1225 | |
1226 | e->offset = offset; | |
1227 | e->p = p; | |
1228 | ||
1229 | git_oid_cpy(&e->sha1, &found_oid); | |
e1de726c | 1230 | return 0; |
a070f152 | 1231 | } |