]>
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 CMN |
87 | } |
88 | } | |
89 | ||
90 | static int cache_init(git_pack_cache *cache) | |
91 | { | |
92 | memset(cache, 0, sizeof(git_pack_cache)); | |
93 | cache->entries = git_offmap_alloc(); | |
94 | GITERR_CHECK_ALLOC(cache->entries); | |
95 | cache->memory_limit = GIT_PACK_CACHE_MEMORY_LIMIT; | |
96 | git_mutex_init(&cache->lock); | |
97 | ||
98 | return 0; | |
99 | } | |
100 | ||
09e29e47 | 101 | static git_pack_cache_entry *cache_get(git_pack_cache *cache, git_off_t offset) |
c8f79c2b CMN |
102 | { |
103 | khiter_t k; | |
104 | git_pack_cache_entry *entry = NULL; | |
105 | ||
09e29e47 CMN |
106 | if (git_mutex_lock(&cache->lock) < 0) |
107 | return NULL; | |
108 | ||
c8f79c2b CMN |
109 | k = kh_get(off, cache->entries, offset); |
110 | if (k != kh_end(cache->entries)) { /* found it */ | |
111 | entry = kh_value(cache->entries, k); | |
112 | git_atomic_inc(&entry->refcount); | |
0ed75620 | 113 | entry->last_usage = cache->use_ctr++; |
c8f79c2b CMN |
114 | } |
115 | git_mutex_unlock(&cache->lock); | |
116 | ||
117 | return entry; | |
118 | } | |
119 | ||
0ed75620 CMN |
120 | /* Run with the cache lock held */ |
121 | static void free_lowest_entry(git_pack_cache *cache) | |
122 | { | |
ed6648ba CMN |
123 | git_pack_cache_entry *entry; |
124 | khiter_t k; | |
125 | ||
0ed75620 CMN |
126 | for (k = kh_begin(cache->entries); k != kh_end(cache->entries); k++) { |
127 | if (!kh_exist(cache->entries, k)) | |
128 | continue; | |
129 | ||
130 | entry = kh_value(cache->entries, k); | |
ed6648ba | 131 | |
9c62aaab CMN |
132 | if (entry && entry->refcount.val == 0) { |
133 | cache->memory_used -= entry->raw.len; | |
134 | kh_del(off, cache->entries, k); | |
135 | free_cache_object(entry); | |
ed6648ba CMN |
136 | } |
137 | } | |
0ed75620 CMN |
138 | } |
139 | ||
c8f79c2b CMN |
140 | static int cache_add(git_pack_cache *cache, git_rawobj *base, git_off_t offset) |
141 | { | |
142 | git_pack_cache_entry *entry; | |
143 | int error, exists = 0; | |
144 | khiter_t k; | |
145 | ||
0ed75620 CMN |
146 | if (base->len > GIT_PACK_CACHE_SIZE_LIMIT) |
147 | return -1; | |
148 | ||
c8f79c2b CMN |
149 | entry = new_cache_object(base); |
150 | if (entry) { | |
09e29e47 CMN |
151 | if (git_mutex_lock(&cache->lock) < 0) { |
152 | giterr_set(GITERR_OS, "failed to lock cache"); | |
153 | return -1; | |
154 | } | |
c8f79c2b CMN |
155 | /* Add it to the cache if nobody else has */ |
156 | exists = kh_get(off, cache->entries, offset) != kh_end(cache->entries); | |
157 | if (!exists) { | |
0ed75620 CMN |
158 | while (cache->memory_used + base->len > cache->memory_limit) |
159 | free_lowest_entry(cache); | |
160 | ||
c8f79c2b CMN |
161 | k = kh_put(off, cache->entries, offset, &error); |
162 | assert(error != 0); | |
163 | kh_value(cache->entries, k) = entry; | |
0ed75620 | 164 | cache->memory_used += entry->raw.len; |
c8f79c2b CMN |
165 | } |
166 | git_mutex_unlock(&cache->lock); | |
167 | /* Somebody beat us to adding it into the cache */ | |
168 | if (exists) { | |
169 | git__free(entry); | |
170 | return -1; | |
171 | } | |
172 | } | |
173 | ||
174 | return 0; | |
175 | } | |
176 | ||
a070f152 CMN |
177 | /*********************************************************** |
178 | * | |
179 | * PACK INDEX METHODS | |
180 | * | |
181 | ***********************************************************/ | |
182 | ||
183 | static void pack_index_free(struct git_pack_file *p) | |
184 | { | |
60ecdf59 DMB |
185 | if (p->oids) { |
186 | git__free(p->oids); | |
187 | p->oids = NULL; | |
188 | } | |
a070f152 CMN |
189 | if (p->index_map.data) { |
190 | git_futils_mmap_free(&p->index_map); | |
191 | p->index_map.data = NULL; | |
192 | } | |
193 | } | |
194 | ||
87d9869f | 195 | static int pack_index_check(const char *path, struct git_pack_file *p) |
a070f152 CMN |
196 | { |
197 | struct git_pack_idx_header *hdr; | |
198 | uint32_t version, nr, i, *index; | |
a070f152 CMN |
199 | void *idx_map; |
200 | size_t idx_size; | |
a070f152 | 201 | struct stat st; |
a070f152 | 202 | int error; |
e1de726c RB |
203 | /* TODO: properly open the file without access time using O_NOATIME */ |
204 | git_file fd = git_futils_open_ro(path); | |
a070f152 | 205 | if (fd < 0) |
e1de726c | 206 | return fd; |
a070f152 | 207 | |
9d2f841a RB |
208 | if (p_fstat(fd, &st) < 0) { |
209 | p_close(fd); | |
210 | giterr_set(GITERR_OS, "Unable to stat pack index '%s'", path); | |
211 | return -1; | |
212 | } | |
213 | ||
214 | if (!S_ISREG(st.st_mode) || | |
e1de726c RB |
215 | !git__is_sizet(st.st_size) || |
216 | (idx_size = (size_t)st.st_size) < 4 * 256 + 20 + 20) | |
217 | { | |
a070f152 | 218 | p_close(fd); |
9d2f841a | 219 | giterr_set(GITERR_ODB, "Invalid pack index '%s'", path); |
e1de726c | 220 | return -1; |
a070f152 CMN |
221 | } |
222 | ||
223 | error = git_futils_mmap_ro(&p->index_map, fd, 0, idx_size); | |
e1de726c | 224 | |
a070f152 CMN |
225 | p_close(fd); |
226 | ||
e1de726c RB |
227 | if (error < 0) |
228 | return error; | |
a070f152 CMN |
229 | |
230 | hdr = idx_map = p->index_map.data; | |
231 | ||
232 | if (hdr->idx_signature == htonl(PACK_IDX_SIGNATURE)) { | |
233 | version = ntohl(hdr->idx_version); | |
234 | ||
235 | if (version < 2 || version > 2) { | |
236 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 237 | return packfile_error("unsupported index version"); |
a070f152 CMN |
238 | } |
239 | ||
240 | } else | |
241 | version = 1; | |
242 | ||
243 | nr = 0; | |
244 | index = idx_map; | |
245 | ||
246 | if (version > 1) | |
87d9869f | 247 | index += 2; /* skip index header */ |
a070f152 CMN |
248 | |
249 | for (i = 0; i < 256; i++) { | |
250 | uint32_t n = ntohl(index[i]); | |
251 | if (n < nr) { | |
252 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 253 | return packfile_error("index is non-monotonic"); |
a070f152 CMN |
254 | } |
255 | nr = n; | |
256 | } | |
257 | ||
258 | if (version == 1) { | |
259 | /* | |
260 | * Total size: | |
87d9869f VM |
261 | * - 256 index entries 4 bytes each |
262 | * - 24-byte entries * nr (20-byte sha1 + 4-byte offset) | |
263 | * - 20-byte SHA1 of the packfile | |
264 | * - 20-byte SHA1 file checksum | |
a070f152 CMN |
265 | */ |
266 | if (idx_size != 4*256 + nr * 24 + 20 + 20) { | |
267 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 268 | return packfile_error("index is corrupted"); |
a070f152 CMN |
269 | } |
270 | } else if (version == 2) { | |
271 | /* | |
272 | * Minimum size: | |
87d9869f VM |
273 | * - 8 bytes of header |
274 | * - 256 index entries 4 bytes each | |
275 | * - 20-byte sha1 entry * nr | |
276 | * - 4-byte crc entry * nr | |
277 | * - 4-byte offset entry * nr | |
278 | * - 20-byte SHA1 of the packfile | |
279 | * - 20-byte SHA1 file checksum | |
a070f152 CMN |
280 | * And after the 4-byte offset table might be a |
281 | * variable sized table containing 8-byte entries | |
282 | * for offsets larger than 2^31. | |
283 | */ | |
284 | unsigned long min_size = 8 + 4*256 + nr*(20 + 4 + 4) + 20 + 20; | |
285 | unsigned long max_size = min_size; | |
286 | ||
287 | if (nr) | |
288 | max_size += (nr - 1)*8; | |
289 | ||
290 | if (idx_size < min_size || idx_size > max_size) { | |
291 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 292 | return packfile_error("wrong index size"); |
a070f152 | 293 | } |
a070f152 CMN |
294 | } |
295 | ||
296 | p->index_version = version; | |
297 | p->num_objects = nr; | |
e1de726c | 298 | return 0; |
a070f152 CMN |
299 | } |
300 | ||
301 | static int pack_index_open(struct git_pack_file *p) | |
302 | { | |
303 | char *idx_name; | |
24c70804 RB |
304 | int error = 0; |
305 | size_t name_len, base_len; | |
306 | ||
a070f152 | 307 | if (p->index_map.data) |
53607868 | 308 | return 0; |
a070f152 | 309 | |
24c70804 RB |
310 | name_len = strlen(p->pack_name); |
311 | assert(name_len > strlen(".pack")); /* checked by git_pack_file alloc */ | |
e1de726c | 312 | |
53607868 RB |
313 | if ((idx_name = git__malloc(name_len)) == NULL) |
314 | return -1; | |
44ef8b1b | 315 | |
24c70804 RB |
316 | base_len = name_len - strlen(".pack"); |
317 | memcpy(idx_name, p->pack_name, base_len); | |
318 | memcpy(idx_name + base_len, ".idx", sizeof(".idx")); | |
a070f152 | 319 | |
53607868 RB |
320 | if ((error = git_mutex_lock(&p->lock)) < 0) |
321 | return error; | |
322 | ||
323 | if (!p->index_map.data) | |
324 | error = pack_index_check(idx_name, p); | |
24c70804 | 325 | |
3286c408 | 326 | git__free(idx_name); |
a070f152 | 327 | |
24c70804 RB |
328 | git_mutex_unlock(&p->lock); |
329 | ||
e1de726c | 330 | return error; |
a070f152 CMN |
331 | } |
332 | ||
333 | static unsigned char *pack_window_open( | |
334 | struct git_pack_file *p, | |
7d0cdf82 | 335 | git_mwindow **w_cursor, |
e1de726c | 336 | git_off_t offset, |
7d0cdf82 CMN |
337 | unsigned int *left) |
338 | { | |
e1de726c | 339 | if (p->mwf.fd == -1 && packfile_open(p) < 0) |
7d0cdf82 CMN |
340 | return NULL; |
341 | ||
342 | /* Since packfiles end in a hash of their content and it's | |
343 | * pointless to ask for an offset into the middle of that | |
344 | * hash, and the pack_window_contains function above wouldn't match | |
345 | * don't allow an offset too close to the end of the file. | |
346 | */ | |
347 | if (offset > (p->mwf.size - 20)) | |
348 | return NULL; | |
349 | ||
350 | return git_mwindow_open(&p->mwf, w_cursor, offset, 20, left); | |
351 | } | |
352 | ||
45d773ef CMN |
353 | static int packfile_unpack_header1( |
354 | unsigned long *usedp, | |
7d0cdf82 CMN |
355 | size_t *sizep, |
356 | git_otype *type, | |
357 | const unsigned char *buf, | |
358 | unsigned long len) | |
359 | { | |
360 | unsigned shift; | |
361 | unsigned long size, c; | |
362 | unsigned long used = 0; | |
2aeadb9c | 363 | |
7d0cdf82 CMN |
364 | c = buf[used++]; |
365 | *type = (c >> 4) & 7; | |
366 | size = c & 15; | |
367 | shift = 4; | |
368 | while (c & 0x80) { | |
45d773ef | 369 | if (len <= used) |
904b67e6 | 370 | return GIT_EBUFS; |
45d773ef CMN |
371 | |
372 | if (bitsizeof(long) <= shift) { | |
373 | *usedp = 0; | |
374 | return -1; | |
375 | } | |
7d0cdf82 CMN |
376 | |
377 | c = buf[used++]; | |
378 | size += (c & 0x7f) << shift; | |
379 | shift += 7; | |
380 | } | |
381 | ||
382 | *sizep = (size_t)size; | |
45d773ef CMN |
383 | *usedp = used; |
384 | return 0; | |
7d0cdf82 CMN |
385 | } |
386 | ||
387 | int git_packfile_unpack_header( | |
388 | size_t *size_p, | |
389 | git_otype *type_p, | |
390 | git_mwindow_file *mwf, | |
391 | git_mwindow **w_curs, | |
e1de726c | 392 | git_off_t *curpos) |
7d0cdf82 CMN |
393 | { |
394 | unsigned char *base; | |
395 | unsigned int left; | |
396 | unsigned long used; | |
45d773ef | 397 | int ret; |
7d0cdf82 CMN |
398 | |
399 | /* pack_window_open() assures us we have [base, base + 20) available | |
87d9869f VM |
400 | * as a range that we can look at at. (Its actually the hash |
401 | * size that is assured.) With our object header encoding | |
7d0cdf82 CMN |
402 | * the maximum deflated object size is 2^137, which is just |
403 | * insane, so we know won't exceed what we have been given. | |
404 | */ | |
24c70804 | 405 | /* base = pack_window_open(p, w_curs, *curpos, &left); */ |
7d0cdf82 CMN |
406 | base = git_mwindow_open(mwf, w_curs, *curpos, 20, &left); |
407 | if (base == NULL) | |
904b67e6 | 408 | return GIT_EBUFS; |
2aeadb9c | 409 | |
9d2f841a | 410 | ret = packfile_unpack_header1(&used, size_p, type_p, base, left); |
45d773ef | 411 | git_mwindow_close(w_curs); |
904b67e6 | 412 | if (ret == GIT_EBUFS) |
45d773ef CMN |
413 | return ret; |
414 | else if (ret < 0) | |
e1de726c | 415 | return packfile_error("header length is zero"); |
7d0cdf82 CMN |
416 | |
417 | *curpos += used; | |
e1de726c | 418 | return 0; |
7d0cdf82 CMN |
419 | } |
420 | ||
44f9f547 DMB |
421 | int git_packfile_resolve_header( |
422 | size_t *size_p, | |
423 | git_otype *type_p, | |
424 | struct git_pack_file *p, | |
425 | git_off_t offset) | |
426 | { | |
427 | git_mwindow *w_curs = NULL; | |
428 | git_off_t curpos = offset; | |
429 | size_t size; | |
430 | git_otype type; | |
431 | git_off_t base_offset; | |
432 | int error; | |
433 | ||
434 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
435 | git_mwindow_close(&w_curs); | |
436 | if (error < 0) | |
437 | return error; | |
438 | ||
439 | if (type == GIT_OBJ_OFS_DELTA || type == GIT_OBJ_REF_DELTA) { | |
440 | size_t base_size; | |
441 | git_rawobj delta; | |
442 | base_offset = get_delta_base(p, &w_curs, &curpos, type, offset); | |
443 | git_mwindow_close(&w_curs); | |
444 | error = packfile_unpack_compressed(&delta, p, &w_curs, &curpos, size, type); | |
445 | git_mwindow_close(&w_curs); | |
446 | if (error < 0) | |
447 | return error; | |
448 | error = git__delta_read_header(delta.data, delta.len, &base_size, size_p); | |
449 | git__free(delta.data); | |
450 | if (error < 0) | |
451 | return error; | |
452 | } else | |
453 | *size_p = size; | |
454 | ||
455 | while (type == GIT_OBJ_OFS_DELTA || type == GIT_OBJ_REF_DELTA) { | |
456 | curpos = base_offset; | |
457 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
458 | git_mwindow_close(&w_curs); | |
459 | if (error < 0) | |
460 | return error; | |
461 | if (type != GIT_OBJ_OFS_DELTA && type != GIT_OBJ_REF_DELTA) | |
462 | break; | |
463 | base_offset = get_delta_base(p, &w_curs, &curpos, type, base_offset); | |
464 | git_mwindow_close(&w_curs); | |
465 | } | |
466 | *type_p = type; | |
467 | ||
468 | return error; | |
469 | } | |
470 | ||
a070f152 | 471 | static int packfile_unpack_delta( |
7d0cdf82 | 472 | git_rawobj *obj, |
a070f152 | 473 | struct git_pack_file *p, |
7d0cdf82 | 474 | git_mwindow **w_curs, |
e1de726c | 475 | git_off_t *curpos, |
7d0cdf82 CMN |
476 | size_t delta_size, |
477 | git_otype delta_type, | |
e1de726c | 478 | git_off_t obj_offset) |
7d0cdf82 | 479 | { |
c0f4a011 | 480 | git_off_t base_offset, base_key; |
7d0cdf82 | 481 | git_rawobj base, delta; |
525d961c | 482 | git_pack_cache_entry *cached = NULL; |
c0f4a011 | 483 | int error, found_base = 0; |
7d0cdf82 | 484 | |
b5b474dd | 485 | base_offset = get_delta_base(p, w_curs, curpos, delta_type, obj_offset); |
45d773ef | 486 | git_mwindow_close(w_curs); |
7d0cdf82 | 487 | if (base_offset == 0) |
e1de726c RB |
488 | return packfile_error("delta offset is zero"); |
489 | if (base_offset < 0) /* must actually be an error code */ | |
490 | return (int)base_offset; | |
7d0cdf82 | 491 | |
c8f79c2b CMN |
492 | if (!p->bases.entries && (cache_init(&p->bases) < 0)) |
493 | return -1; | |
7d0cdf82 | 494 | |
c0f4a011 | 495 | base_key = base_offset; /* git_packfile_unpack modifies base_offset */ |
c8f79c2b | 496 | if ((cached = cache_get(&p->bases, base_offset)) != NULL) { |
c0f4a011 | 497 | memcpy(&base, &cached->raw, sizeof(git_rawobj)); |
c8f79c2b | 498 | found_base = 1; |
525d961c | 499 | } |
525d961c CMN |
500 | |
501 | if (!cached) { /* have to inflate it */ | |
c0f4a011 CMN |
502 | error = git_packfile_unpack(&base, p, &base_offset); |
503 | ||
504 | /* | |
505 | * TODO: git.git tries to load the base from other packfiles | |
506 | * or loose objects. | |
507 | * | |
508 | * We'll need to do this in order to support thin packs. | |
509 | */ | |
510 | if (error < 0) | |
511 | return error; | |
512 | } | |
7d0cdf82 CMN |
513 | |
514 | error = packfile_unpack_compressed(&delta, p, w_curs, curpos, delta_size, delta_type); | |
45d773ef | 515 | git_mwindow_close(w_curs); |
c0f4a011 | 516 | |
e1de726c | 517 | if (error < 0) { |
c0f4a011 CMN |
518 | if (!found_base) |
519 | git__free(base.data); | |
e1de726c | 520 | return error; |
7d0cdf82 CMN |
521 | } |
522 | ||
523 | obj->type = base.type; | |
e1de726c | 524 | error = git__delta_apply(obj, base.data, base.len, delta.data, delta.len); |
c0f4a011 CMN |
525 | if (error < 0) |
526 | goto on_error; | |
527 | ||
c8f79c2b | 528 | if (found_base) |
525d961c | 529 | git_atomic_dec(&cached->refcount); |
c8f79c2b CMN |
530 | else if (cache_add(&p->bases, &base, base_key) < 0) |
531 | git__free(base.data); | |
7d0cdf82 | 532 | |
c0f4a011 | 533 | on_error: |
3286c408 | 534 | git__free(delta.data); |
7d0cdf82 | 535 | |
7d0cdf82 CMN |
536 | return error; /* error set by git__delta_apply */ |
537 | } | |
538 | ||
a070f152 | 539 | int git_packfile_unpack( |
e1de726c RB |
540 | git_rawobj *obj, |
541 | struct git_pack_file *p, | |
542 | git_off_t *obj_offset) | |
7d0cdf82 CMN |
543 | { |
544 | git_mwindow *w_curs = NULL; | |
e1de726c | 545 | git_off_t curpos = *obj_offset; |
7d0cdf82 CMN |
546 | int error; |
547 | ||
548 | size_t size = 0; | |
549 | git_otype type; | |
550 | ||
551 | /* | |
552 | * TODO: optionally check the CRC on the packfile | |
553 | */ | |
554 | ||
555 | obj->data = NULL; | |
556 | obj->len = 0; | |
557 | obj->type = GIT_OBJ_BAD; | |
558 | ||
559 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
45d773ef CMN |
560 | git_mwindow_close(&w_curs); |
561 | ||
e1de726c RB |
562 | if (error < 0) |
563 | return error; | |
7d0cdf82 CMN |
564 | |
565 | switch (type) { | |
566 | case GIT_OBJ_OFS_DELTA: | |
567 | case GIT_OBJ_REF_DELTA: | |
568 | error = packfile_unpack_delta( | |
b5b474dd CMN |
569 | obj, p, &w_curs, &curpos, |
570 | size, type, *obj_offset); | |
7d0cdf82 CMN |
571 | break; |
572 | ||
573 | case GIT_OBJ_COMMIT: | |
574 | case GIT_OBJ_TREE: | |
575 | case GIT_OBJ_BLOB: | |
576 | case GIT_OBJ_TAG: | |
577 | error = packfile_unpack_compressed( | |
b5b474dd | 578 | obj, p, &w_curs, &curpos, |
7d0cdf82 CMN |
579 | size, type); |
580 | break; | |
581 | ||
582 | default: | |
e1de726c | 583 | error = packfile_error("invalid packfile type in header");; |
7d0cdf82 CMN |
584 | break; |
585 | } | |
586 | ||
b5b474dd | 587 | *obj_offset = curpos; |
e1de726c | 588 | return error; |
7d0cdf82 CMN |
589 | } |
590 | ||
282283ac RB |
591 | static void *use_git_alloc(void *opaq, unsigned int count, unsigned int size) |
592 | { | |
593 | GIT_UNUSED(opaq); | |
594 | return git__calloc(count, size); | |
595 | } | |
596 | ||
597 | static void use_git_free(void *opaq, void *ptr) | |
598 | { | |
599 | GIT_UNUSED(opaq); | |
600 | git__free(ptr); | |
601 | } | |
602 | ||
46635339 CMN |
603 | int git_packfile_stream_open(git_packfile_stream *obj, struct git_pack_file *p, git_off_t curpos) |
604 | { | |
605 | int st; | |
606 | ||
607 | memset(obj, 0, sizeof(git_packfile_stream)); | |
608 | obj->curpos = curpos; | |
609 | obj->p = p; | |
610 | obj->zstream.zalloc = use_git_alloc; | |
611 | obj->zstream.zfree = use_git_free; | |
612 | obj->zstream.next_in = Z_NULL; | |
613 | obj->zstream.next_out = Z_NULL; | |
614 | st = inflateInit(&obj->zstream); | |
615 | if (st != Z_OK) { | |
616 | git__free(obj); | |
617 | giterr_set(GITERR_ZLIB, "Failed to inflate packfile"); | |
618 | return -1; | |
619 | } | |
620 | ||
621 | return 0; | |
622 | } | |
623 | ||
624 | ssize_t git_packfile_stream_read(git_packfile_stream *obj, void *buffer, size_t len) | |
625 | { | |
626 | unsigned char *in; | |
627 | size_t written; | |
628 | int st; | |
629 | ||
630 | if (obj->done) | |
631 | return 0; | |
632 | ||
633 | in = pack_window_open(obj->p, &obj->mw, obj->curpos, &obj->zstream.avail_in); | |
634 | if (in == NULL) | |
635 | return GIT_EBUFS; | |
636 | ||
637 | obj->zstream.next_out = buffer; | |
090d5e1f | 638 | obj->zstream.avail_out = (unsigned int)len; |
46635339 CMN |
639 | obj->zstream.next_in = in; |
640 | ||
641 | st = inflate(&obj->zstream, Z_SYNC_FLUSH); | |
642 | git_mwindow_close(&obj->mw); | |
643 | ||
644 | obj->curpos += obj->zstream.next_in - in; | |
645 | written = len - obj->zstream.avail_out; | |
646 | ||
647 | if (st != Z_OK && st != Z_STREAM_END) { | |
648 | giterr_set(GITERR_ZLIB, "Failed to inflate packfile"); | |
649 | return -1; | |
650 | } | |
651 | ||
652 | if (st == Z_STREAM_END) | |
653 | obj->done = 1; | |
654 | ||
655 | ||
656 | /* If we didn't write anything out but we're not done, we need more data */ | |
657 | if (!written && st != Z_STREAM_END) | |
658 | return GIT_EBUFS; | |
659 | ||
660 | return written; | |
661 | ||
662 | } | |
663 | ||
664 | void git_packfile_stream_free(git_packfile_stream *obj) | |
665 | { | |
666 | inflateEnd(&obj->zstream); | |
667 | } | |
668 | ||
7d0cdf82 | 669 | int packfile_unpack_compressed( |
e1de726c RB |
670 | git_rawobj *obj, |
671 | struct git_pack_file *p, | |
672 | git_mwindow **w_curs, | |
673 | git_off_t *curpos, | |
674 | size_t size, | |
675 | git_otype type) | |
7d0cdf82 CMN |
676 | { |
677 | int st; | |
678 | z_stream stream; | |
679 | unsigned char *buffer, *in; | |
680 | ||
e1de726c RB |
681 | buffer = git__calloc(1, size + 1); |
682 | GITERR_CHECK_ALLOC(buffer); | |
7d0cdf82 CMN |
683 | |
684 | memset(&stream, 0, sizeof(stream)); | |
685 | stream.next_out = buffer; | |
1c3fac4d | 686 | stream.avail_out = (uInt)size + 1; |
282283ac RB |
687 | stream.zalloc = use_git_alloc; |
688 | stream.zfree = use_git_free; | |
7d0cdf82 CMN |
689 | |
690 | st = inflateInit(&stream); | |
691 | if (st != Z_OK) { | |
3286c408 | 692 | git__free(buffer); |
e1de726c | 693 | giterr_set(GITERR_ZLIB, "Failed to inflate packfile"); |
45d773ef | 694 | |
e1de726c | 695 | return -1; |
7d0cdf82 CMN |
696 | } |
697 | ||
698 | do { | |
b5b474dd | 699 | in = pack_window_open(p, w_curs, *curpos, &stream.avail_in); |
7d0cdf82 CMN |
700 | stream.next_in = in; |
701 | st = inflate(&stream, Z_FINISH); | |
45d773ef | 702 | git_mwindow_close(w_curs); |
7d0cdf82 CMN |
703 | |
704 | if (!stream.avail_out) | |
705 | break; /* the payload is larger than it should be */ | |
706 | ||
45d773ef CMN |
707 | if (st == Z_BUF_ERROR && in == NULL) { |
708 | inflateEnd(&stream); | |
709 | git__free(buffer); | |
904b67e6 | 710 | return GIT_EBUFS; |
45d773ef CMN |
711 | } |
712 | ||
b5b474dd | 713 | *curpos += stream.next_in - in; |
7d0cdf82 CMN |
714 | } while (st == Z_OK || st == Z_BUF_ERROR); |
715 | ||
716 | inflateEnd(&stream); | |
717 | ||
718 | if ((st != Z_STREAM_END) || stream.total_out != size) { | |
3286c408 | 719 | git__free(buffer); |
e1de726c RB |
720 | giterr_set(GITERR_ZLIB, "Failed to inflate packfile"); |
721 | return -1; | |
7d0cdf82 CMN |
722 | } |
723 | ||
724 | obj->type = type; | |
725 | obj->len = size; | |
726 | obj->data = buffer; | |
e1de726c | 727 | return 0; |
7d0cdf82 CMN |
728 | } |
729 | ||
b5b474dd CMN |
730 | /* |
731 | * curpos is where the data starts, delta_obj_offset is the where the | |
732 | * header starts | |
733 | */ | |
e1de726c RB |
734 | git_off_t get_delta_base( |
735 | struct git_pack_file *p, | |
736 | git_mwindow **w_curs, | |
737 | git_off_t *curpos, | |
738 | git_otype type, | |
739 | git_off_t delta_obj_offset) | |
7d0cdf82 | 740 | { |
45d773ef CMN |
741 | unsigned int left = 0; |
742 | unsigned char *base_info; | |
e1de726c | 743 | git_off_t base_offset; |
7d0cdf82 CMN |
744 | git_oid unused; |
745 | ||
45d773ef CMN |
746 | base_info = pack_window_open(p, w_curs, *curpos, &left); |
747 | /* Assumption: the only reason this would fail is because the file is too small */ | |
748 | if (base_info == NULL) | |
904b67e6 | 749 | return GIT_EBUFS; |
7d0cdf82 CMN |
750 | /* pack_window_open() assured us we have [base_info, base_info + 20) |
751 | * as a range that we can look at without walking off the | |
87d9869f VM |
752 | * end of the mapped window. Its actually the hash size |
753 | * that is assured. An OFS_DELTA longer than the hash size | |
7d0cdf82 CMN |
754 | * is stupid, as then a REF_DELTA would be smaller to store. |
755 | */ | |
756 | if (type == GIT_OBJ_OFS_DELTA) { | |
757 | unsigned used = 0; | |
758 | unsigned char c = base_info[used++]; | |
759 | base_offset = c & 127; | |
760 | while (c & 128) { | |
45d773ef | 761 | if (left <= used) |
904b67e6 | 762 | return GIT_EBUFS; |
7d0cdf82 CMN |
763 | base_offset += 1; |
764 | if (!base_offset || MSB(base_offset, 7)) | |
87d9869f | 765 | return 0; /* overflow */ |
7d0cdf82 CMN |
766 | c = base_info[used++]; |
767 | base_offset = (base_offset << 7) + (c & 127); | |
768 | } | |
769 | base_offset = delta_obj_offset - base_offset; | |
770 | if (base_offset <= 0 || base_offset >= delta_obj_offset) | |
87d9869f | 771 | return 0; /* out of bound */ |
7d0cdf82 CMN |
772 | *curpos += used; |
773 | } else if (type == GIT_OBJ_REF_DELTA) { | |
c1af5a39 CMN |
774 | /* If we have the cooperative cache, search in it first */ |
775 | if (p->has_cache) { | |
0e040c03 CMN |
776 | khiter_t k; |
777 | git_oid oid; | |
c1af5a39 | 778 | |
0e040c03 CMN |
779 | git_oid_fromraw(&oid, base_info); |
780 | k = kh_get(oid, p->idx_cache, &oid); | |
781 | if (k != kh_end(p->idx_cache)) { | |
c1af5a39 | 782 | *curpos += 20; |
0e040c03 | 783 | return ((struct git_pack_entry *)kh_value(p->idx_cache, k))->offset; |
c1af5a39 CMN |
784 | } |
785 | } | |
7d0cdf82 | 786 | /* The base entry _must_ be in the same pack */ |
e1de726c RB |
787 | if (pack_entry_find_offset(&base_offset, &unused, p, (git_oid *)base_info, GIT_OID_HEXSZ) < 0) |
788 | return packfile_error("base entry delta is not in the same pack"); | |
7d0cdf82 CMN |
789 | *curpos += 20; |
790 | } else | |
791 | return 0; | |
792 | ||
793 | return base_offset; | |
794 | } | |
a070f152 CMN |
795 | |
796 | /*********************************************************** | |
797 | * | |
798 | * PACKFILE METHODS | |
799 | * | |
800 | ***********************************************************/ | |
801 | ||
96c9b9f0 | 802 | void git_packfile_free(struct git_pack_file *p) |
a070f152 | 803 | { |
24c70804 RB |
804 | if (!p) |
805 | return; | |
806 | ||
c8f79c2b | 807 | cache_free(&p->bases); |
c0f4a011 | 808 | |
a070f152 CMN |
809 | git_mwindow_free_all(&p->mwf); |
810 | ||
811 | if (p->mwf.fd != -1) | |
812 | p_close(p->mwf.fd); | |
813 | ||
814 | pack_index_free(p); | |
815 | ||
3286c408 | 816 | git__free(p->bad_object_sha1); |
24c70804 | 817 | |
24c70804 | 818 | git_mutex_free(&p->lock); |
3286c408 | 819 | git__free(p); |
a070f152 CMN |
820 | } |
821 | ||
822 | static int packfile_open(struct git_pack_file *p) | |
823 | { | |
824 | struct stat st; | |
825 | struct git_pack_header hdr; | |
826 | git_oid sha1; | |
827 | unsigned char *idx_sha1; | |
828 | ||
0d0fa7c3 | 829 | if (!p->index_map.data && pack_index_open(p) < 0) |
282283ac | 830 | return git_odb__error_notfound("failed to open packfile", NULL); |
a070f152 | 831 | |
9d2f841a RB |
832 | /* if mwf opened by another thread, return now */ |
833 | if (git_mutex_lock(&p->lock) < 0) | |
834 | return packfile_error("failed to get lock for open"); | |
835 | ||
836 | if (p->mwf.fd >= 0) { | |
837 | git_mutex_unlock(&p->lock); | |
838 | return 0; | |
839 | } | |
840 | ||
a070f152 | 841 | /* TODO: open with noatime */ |
e1de726c | 842 | p->mwf.fd = git_futils_open_ro(p->pack_name); |
9d2f841a RB |
843 | if (p->mwf.fd < 0) |
844 | goto cleanup; | |
a070f152 | 845 | |
e1de726c RB |
846 | if (p_fstat(p->mwf.fd, &st) < 0 || |
847 | git_mwindow_file_register(&p->mwf) < 0) | |
848 | goto cleanup; | |
a070f152 CMN |
849 | |
850 | /* If we created the struct before we had the pack we lack size. */ | |
851 | if (!p->mwf.size) { | |
852 | if (!S_ISREG(st.st_mode)) | |
853 | goto cleanup; | |
e1de726c | 854 | p->mwf.size = (git_off_t)st.st_size; |
a070f152 CMN |
855 | } else if (p->mwf.size != st.st_size) |
856 | goto cleanup; | |
857 | ||
858 | #if 0 | |
859 | /* We leave these file descriptors open with sliding mmap; | |
860 | * there is no point keeping them open across exec(), though. | |
861 | */ | |
862 | fd_flag = fcntl(p->mwf.fd, F_GETFD, 0); | |
863 | if (fd_flag < 0) | |
e1de726c | 864 | goto cleanup; |
a070f152 CMN |
865 | |
866 | fd_flag |= FD_CLOEXEC; | |
867 | if (fcntl(p->pack_fd, F_SETFD, fd_flag) == -1) | |
e1de726c | 868 | goto cleanup; |
a070f152 CMN |
869 | #endif |
870 | ||
871 | /* Verify we recognize this pack file format. */ | |
e1de726c RB |
872 | if (p_read(p->mwf.fd, &hdr, sizeof(hdr)) < 0 || |
873 | hdr.hdr_signature != htonl(PACK_SIGNATURE) || | |
874 | !pack_version_ok(hdr.hdr_version)) | |
a070f152 CMN |
875 | goto cleanup; |
876 | ||
877 | /* Verify the pack matches its index. */ | |
e1de726c RB |
878 | if (p->num_objects != ntohl(hdr.hdr_entries) || |
879 | p_lseek(p->mwf.fd, p->mwf.size - GIT_OID_RAWSZ, SEEK_SET) == -1 || | |
880 | p_read(p->mwf.fd, sha1.id, GIT_OID_RAWSZ) < 0) | |
a070f152 CMN |
881 | goto cleanup; |
882 | ||
883 | idx_sha1 = ((unsigned char *)p->index_map.data) + p->index_map.len - 40; | |
884 | ||
9d2f841a RB |
885 | if (git_oid__cmp(&sha1, (git_oid *)idx_sha1) != 0) |
886 | goto cleanup; | |
887 | ||
888 | git_mutex_unlock(&p->lock); | |
889 | return 0; | |
a070f152 CMN |
890 | |
891 | cleanup: | |
e1de726c | 892 | giterr_set(GITERR_OS, "Invalid packfile '%s'", p->pack_name); |
9d2f841a | 893 | |
a070f152 CMN |
894 | p_close(p->mwf.fd); |
895 | p->mwf.fd = -1; | |
9d2f841a RB |
896 | |
897 | git_mutex_unlock(&p->lock); | |
898 | ||
e1de726c | 899 | return -1; |
a070f152 CMN |
900 | } |
901 | ||
5d2d21e5 | 902 | int git_packfile_alloc(struct git_pack_file **pack_out, const char *path) |
a070f152 CMN |
903 | { |
904 | struct stat st; | |
905 | struct git_pack_file *p; | |
38eef611 | 906 | size_t path_len = path ? strlen(path) : 0; |
a070f152 CMN |
907 | |
908 | *pack_out = NULL; | |
24c70804 | 909 | |
38eef611 | 910 | if (path_len < strlen(".idx")) |
24c70804 RB |
911 | return git_odb__error_notfound("invalid packfile path", NULL); |
912 | ||
38eef611 | 913 | p = git__calloc(1, sizeof(*p) + path_len + 2); |
e1de726c | 914 | GITERR_CHECK_ALLOC(p); |
a070f152 | 915 | |
38eef611 RB |
916 | memcpy(p->pack_name, path, path_len + 1); |
917 | ||
a070f152 CMN |
918 | /* |
919 | * Make sure a corresponding .pack file exists and that | |
920 | * the index looks sane. | |
921 | */ | |
38eef611 RB |
922 | if (git__suffixcmp(path, ".idx") == 0) { |
923 | size_t root_len = path_len - strlen(".idx"); | |
924 | ||
925 | memcpy(p->pack_name + root_len, ".keep", sizeof(".keep")); | |
926 | if (git_path_exists(p->pack_name) == true) | |
927 | p->pack_keep = 1; | |
a070f152 | 928 | |
38eef611 RB |
929 | memcpy(p->pack_name + root_len, ".pack", sizeof(".pack")); |
930 | path_len = path_len - strlen(".idx") + strlen(".pack"); | |
931 | } | |
a070f152 | 932 | |
e1de726c | 933 | if (p_stat(p->pack_name, &st) < 0 || !S_ISREG(st.st_mode)) { |
3286c408 | 934 | git__free(p); |
282283ac | 935 | return git_odb__error_notfound("packfile not found", NULL); |
a070f152 CMN |
936 | } |
937 | ||
938 | /* ok, it looks sane as far as we can check without | |
939 | * actually mapping the pack file. | |
940 | */ | |
38eef611 | 941 | p->mwf.fd = -1; |
1af56d7d | 942 | p->mwf.size = st.st_size; |
a070f152 CMN |
943 | p->pack_local = 1; |
944 | p->mtime = (git_time_t)st.st_mtime; | |
945 | ||
38eef611 RB |
946 | git_mutex_init(&p->lock); |
947 | ||
a070f152 CMN |
948 | /* see if we can parse the sha1 oid in the packfile name */ |
949 | if (path_len < 40 || | |
e1de726c | 950 | git_oid_fromstr(&p->sha1, path + path_len - GIT_OID_HEXSZ) < 0) |
a070f152 CMN |
951 | memset(&p->sha1, 0x0, GIT_OID_RAWSZ); |
952 | ||
953 | *pack_out = p; | |
e1de726c RB |
954 | |
955 | return 0; | |
a070f152 CMN |
956 | } |
957 | ||
958 | /*********************************************************** | |
959 | * | |
960 | * PACKFILE ENTRY SEARCH INTERNALS | |
961 | * | |
962 | ***********************************************************/ | |
963 | ||
e1de726c | 964 | static git_off_t nth_packed_object_offset(const struct git_pack_file *p, uint32_t n) |
a070f152 CMN |
965 | { |
966 | const unsigned char *index = p->index_map.data; | |
967 | index += 4 * 256; | |
968 | if (p->index_version == 1) { | |
969 | return ntohl(*((uint32_t *)(index + 24 * n))); | |
970 | } else { | |
971 | uint32_t off; | |
972 | index += 8 + p->num_objects * (20 + 4); | |
973 | off = ntohl(*((uint32_t *)(index + 4 * n))); | |
974 | if (!(off & 0x80000000)) | |
975 | return off; | |
976 | index += p->num_objects * 4 + (off & 0x7fffffff) * 8; | |
977 | return (((uint64_t)ntohl(*((uint32_t *)(index + 0)))) << 32) | | |
87d9869f | 978 | ntohl(*((uint32_t *)(index + 4))); |
a070f152 CMN |
979 | } |
980 | } | |
981 | ||
60ecdf59 DMB |
982 | static int git__memcmp4(const void *a, const void *b) { |
983 | return memcmp(a, b, 4); | |
984 | } | |
985 | ||
521aedad | 986 | int git_pack_foreach_entry( |
5dca2010 | 987 | struct git_pack_file *p, |
c3fb7d04 | 988 | git_odb_foreach_cb cb, |
5dca2010 | 989 | void *data) |
521aedad CMN |
990 | { |
991 | const unsigned char *index = p->index_map.data, *current; | |
521aedad CMN |
992 | uint32_t i; |
993 | ||
994 | if (index == NULL) { | |
995 | int error; | |
996 | ||
997 | if ((error = pack_index_open(p)) < 0) | |
998 | return error; | |
999 | ||
1000 | assert(p->index_map.data); | |
1001 | ||
1002 | index = p->index_map.data; | |
1003 | } | |
1004 | ||
1005 | if (p->index_version > 1) { | |
1006 | index += 8; | |
1007 | } | |
1008 | ||
1009 | index += 4 * 256; | |
1010 | ||
60ecdf59 DMB |
1011 | if (p->oids == NULL) { |
1012 | git_vector offsets, oids; | |
1013 | int error; | |
521aedad | 1014 | |
60ecdf59 DMB |
1015 | if ((error = git_vector_init(&oids, p->num_objects, NULL))) |
1016 | return error; | |
1017 | ||
1018 | if ((error = git_vector_init(&offsets, p->num_objects, git__memcmp4))) | |
1019 | return error; | |
5dca2010 | 1020 | |
60ecdf59 DMB |
1021 | if (p->index_version > 1) { |
1022 | const unsigned char *off = index + 24 * p->num_objects; | |
1023 | for (i = 0; i < p->num_objects; i++) | |
1024 | git_vector_insert(&offsets, (void*)&off[4 * i]); | |
1025 | git_vector_sort(&offsets); | |
1026 | git_vector_foreach(&offsets, i, current) | |
1027 | git_vector_insert(&oids, (void*)&index[5 * (current - off)]); | |
1028 | } else { | |
1029 | for (i = 0; i < p->num_objects; i++) | |
1030 | git_vector_insert(&offsets, (void*)&index[24 * i]); | |
1031 | git_vector_sort(&offsets); | |
1032 | git_vector_foreach(&offsets, i, current) | |
1033 | git_vector_insert(&oids, (void*)¤t[4]); | |
1034 | } | |
1035 | git_vector_free(&offsets); | |
1036 | p->oids = (git_oid **)oids.contents; | |
521aedad CMN |
1037 | } |
1038 | ||
60ecdf59 DMB |
1039 | for (i = 0; i < p->num_objects; i++) |
1040 | if (cb(p->oids[i], data)) | |
1041 | return GIT_EUSER; | |
1042 | ||
521aedad CMN |
1043 | return 0; |
1044 | } | |
1045 | ||
a070f152 | 1046 | static int pack_entry_find_offset( |
e1de726c RB |
1047 | git_off_t *offset_out, |
1048 | git_oid *found_oid, | |
1049 | struct git_pack_file *p, | |
1050 | const git_oid *short_oid, | |
b8457baa | 1051 | size_t len) |
a070f152 | 1052 | { |
8c535f3f RB |
1053 | const uint32_t *level1_ofs; |
1054 | const unsigned char *index; | |
a070f152 CMN |
1055 | unsigned hi, lo, stride; |
1056 | int pos, found = 0; | |
1057 | const unsigned char *current = 0; | |
1058 | ||
1059 | *offset_out = 0; | |
1060 | ||
8c535f3f RB |
1061 | if (!p->index_map.data && pack_index_open(p) < 0) |
1062 | return git_odb__error_notfound("failed to open packfile", NULL); | |
a070f152 | 1063 | |
8c535f3f RB |
1064 | if (git_mutex_lock(&p->lock) < 0) |
1065 | return packfile_error("failed to get lock for finding entry offset"); | |
a070f152 | 1066 | |
8c535f3f RB |
1067 | assert(p->index_map.data); |
1068 | ||
1069 | index = p->index_map.data; | |
1070 | level1_ofs = p->index_map.data; | |
a070f152 CMN |
1071 | |
1072 | if (p->index_version > 1) { | |
1073 | level1_ofs += 2; | |
1074 | index += 8; | |
1075 | } | |
1076 | ||
1077 | index += 4 * 256; | |
1078 | hi = ntohl(level1_ofs[(int)short_oid->id[0]]); | |
1079 | lo = ((short_oid->id[0] == 0x0) ? 0 : ntohl(level1_ofs[(int)short_oid->id[0] - 1])); | |
1080 | ||
1081 | if (p->index_version > 1) { | |
1082 | stride = 20; | |
1083 | } else { | |
1084 | stride = 24; | |
1085 | index += 4; | |
1086 | } | |
1087 | ||
1088 | #ifdef INDEX_DEBUG_LOOKUP | |
1089 | printf("%02x%02x%02x... lo %u hi %u nr %d\n", | |
1090 | short_oid->id[0], short_oid->id[1], short_oid->id[2], lo, hi, p->num_objects); | |
1091 | #endif | |
1092 | ||
1093 | /* Use git.git lookup code */ | |
87d9869f | 1094 | pos = sha1_entry_pos(index, stride, 0, lo, hi, p->num_objects, short_oid->id); |
a070f152 | 1095 | |
8c535f3f RB |
1096 | git_mutex_unlock(&p->lock); |
1097 | ||
a070f152 CMN |
1098 | if (pos >= 0) { |
1099 | /* An object matching exactly the oid was found */ | |
1100 | found = 1; | |
1101 | current = index + pos * stride; | |
1102 | } else { | |
1103 | /* No object was found */ | |
1104 | /* pos refers to the object with the "closest" oid to short_oid */ | |
1105 | pos = - 1 - pos; | |
1106 | if (pos < (int)p->num_objects) { | |
1107 | current = index + pos * stride; | |
1108 | ||
282283ac | 1109 | if (!git_oid_ncmp(short_oid, (const git_oid *)current, len)) |
a070f152 | 1110 | found = 1; |
a070f152 CMN |
1111 | } |
1112 | } | |
1113 | ||
b2a2702d | 1114 | if (found && len != GIT_OID_HEXSZ && pos + 1 < (int)p->num_objects) { |
a070f152 CMN |
1115 | /* Check for ambiguousity */ |
1116 | const unsigned char *next = current + stride; | |
1117 | ||
1118 | if (!git_oid_ncmp(short_oid, (const git_oid *)next, len)) { | |
1119 | found = 2; | |
1120 | } | |
1121 | } | |
1122 | ||
e1de726c | 1123 | if (!found) |
282283ac | 1124 | return git_odb__error_notfound("failed to find offset for pack entry", short_oid); |
e1de726c RB |
1125 | if (found > 1) |
1126 | return git_odb__error_ambiguous("found multiple offsets for pack entry"); | |
24c70804 | 1127 | |
e1de726c RB |
1128 | *offset_out = nth_packed_object_offset(p, pos); |
1129 | git_oid_fromraw(found_oid, current); | |
a070f152 CMN |
1130 | |
1131 | #ifdef INDEX_DEBUG_LOOKUP | |
e1de726c | 1132 | { |
a070f152 CMN |
1133 | unsigned char hex_sha1[GIT_OID_HEXSZ + 1]; |
1134 | git_oid_fmt(hex_sha1, found_oid); | |
1135 | hex_sha1[GIT_OID_HEXSZ] = '\0'; | |
1136 | printf("found lo=%d %s\n", lo, hex_sha1); | |
a070f152 | 1137 | } |
e1de726c | 1138 | #endif |
24c70804 | 1139 | |
e1de726c | 1140 | return 0; |
a070f152 CMN |
1141 | } |
1142 | ||
1143 | int git_pack_entry_find( | |
1144 | struct git_pack_entry *e, | |
1145 | struct git_pack_file *p, | |
1146 | const git_oid *short_oid, | |
b8457baa | 1147 | size_t len) |
a070f152 | 1148 | { |
e1de726c | 1149 | git_off_t offset; |
a070f152 CMN |
1150 | git_oid found_oid; |
1151 | int error; | |
1152 | ||
1153 | assert(p); | |
1154 | ||
1155 | if (len == GIT_OID_HEXSZ && p->num_bad_objects) { | |
1156 | unsigned i; | |
1157 | for (i = 0; i < p->num_bad_objects; i++) | |
b7f167da | 1158 | if (git_oid__cmp(short_oid, &p->bad_object_sha1[i]) == 0) |
e1de726c | 1159 | return packfile_error("bad object found in packfile"); |
a070f152 CMN |
1160 | } |
1161 | ||
1162 | error = pack_entry_find_offset(&offset, &found_oid, p, short_oid, len); | |
e1de726c RB |
1163 | if (error < 0) |
1164 | return error; | |
a070f152 CMN |
1165 | |
1166 | /* we found a unique entry in the index; | |
1167 | * make sure the packfile backing the index | |
1168 | * still exists on disk */ | |
e1de726c RB |
1169 | if (p->mwf.fd == -1 && (error = packfile_open(p)) < 0) |
1170 | return error; | |
a070f152 CMN |
1171 | |
1172 | e->offset = offset; | |
1173 | e->p = p; | |
1174 | ||
1175 | git_oid_cpy(&e->sha1, &found_oid); | |
e1de726c | 1176 | return 0; |
a070f152 | 1177 | } |