]>
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 | |
c369b379 SW |
19 | GIT__USE_OFFMAP |
20 | GIT__USE_OIDMAP | |
b63b76e0 | 21 | |
a070f152 | 22 | static int packfile_open(struct git_pack_file *p); |
e1de726c | 23 | static git_off_t nth_packed_object_offset(const struct git_pack_file *p, uint32_t n); |
b644e223 | 24 | static int packfile_unpack_compressed( |
a070f152 CMN |
25 | git_rawobj *obj, |
26 | struct git_pack_file *p, | |
27 | git_mwindow **w_curs, | |
e1de726c | 28 | git_off_t *curpos, |
a070f152 CMN |
29 | size_t size, |
30 | git_otype type); | |
31 | ||
32 | /* Can find the offset of an object given | |
33 | * a prefix of an identifier. | |
904b67e6 | 34 | * Throws GIT_EAMBIGUOUSOIDPREFIX if short oid |
a070f152 CMN |
35 | * is ambiguous within the pack. |
36 | * This method assumes that len is between | |
37 | * GIT_OID_MINPREFIXLEN and GIT_OID_HEXSZ. | |
38 | */ | |
39 | static int pack_entry_find_offset( | |
e1de726c | 40 | git_off_t *offset_out, |
a070f152 CMN |
41 | git_oid *found_oid, |
42 | struct git_pack_file *p, | |
43 | const git_oid *short_oid, | |
b8457baa | 44 | size_t len); |
a070f152 | 45 | |
e1de726c RB |
46 | static int packfile_error(const char *message) |
47 | { | |
48 | giterr_set(GITERR_ODB, "Invalid pack file - %s", message); | |
49 | return -1; | |
50 | } | |
51 | ||
c8f79c2b CMN |
52 | /******************** |
53 | * Delta base cache | |
54 | ********************/ | |
c0f4a011 | 55 | |
525d961c | 56 | static git_pack_cache_entry *new_cache_object(git_rawobj *source) |
c0f4a011 | 57 | { |
c8f79c2b | 58 | git_pack_cache_entry *e = git__calloc(1, sizeof(git_pack_cache_entry)); |
c0f4a011 CMN |
59 | if (!e) |
60 | return NULL; | |
61 | ||
8588cb0c | 62 | git_atomic_inc(&e->refcount); |
c0f4a011 CMN |
63 | memcpy(&e->raw, source, sizeof(git_rawobj)); |
64 | ||
65 | return e; | |
66 | } | |
67 | ||
68 | static void free_cache_object(void *o) | |
69 | { | |
70 | git_pack_cache_entry *e = (git_pack_cache_entry *)o; | |
71 | ||
72 | if (e != NULL) { | |
0ed75620 | 73 | assert(e->refcount.val == 0); |
c0f4a011 CMN |
74 | git__free(e->raw.data); |
75 | git__free(e); | |
76 | } | |
77 | } | |
78 | ||
c8f79c2b CMN |
79 | static void cache_free(git_pack_cache *cache) |
80 | { | |
81 | khiter_t k; | |
82 | ||
83 | if (cache->entries) { | |
84 | for (k = kh_begin(cache->entries); k != kh_end(cache->entries); k++) { | |
85 | if (kh_exist(cache->entries, k)) | |
86 | free_cache_object(kh_value(cache->entries, k)); | |
87 | } | |
88 | ||
89 | git_offmap_free(cache->entries); | |
649214be | 90 | cache->entries = NULL; |
c8f79c2b CMN |
91 | } |
92 | } | |
93 | ||
94 | static int cache_init(git_pack_cache *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 | ||
8588cb0c JH |
152 | static int cache_add( |
153 | git_pack_cache_entry **cached_out, | |
154 | git_pack_cache *cache, | |
155 | git_rawobj *base, | |
156 | git_off_t offset) | |
c8f79c2b CMN |
157 | { |
158 | git_pack_cache_entry *entry; | |
159 | int error, exists = 0; | |
160 | khiter_t k; | |
161 | ||
0ed75620 CMN |
162 | if (base->len > GIT_PACK_CACHE_SIZE_LIMIT) |
163 | return -1; | |
164 | ||
c8f79c2b CMN |
165 | entry = new_cache_object(base); |
166 | if (entry) { | |
09e29e47 CMN |
167 | if (git_mutex_lock(&cache->lock) < 0) { |
168 | giterr_set(GITERR_OS, "failed to lock cache"); | |
6f73e026 | 169 | git__free(entry); |
09e29e47 CMN |
170 | return -1; |
171 | } | |
c8f79c2b CMN |
172 | /* Add it to the cache if nobody else has */ |
173 | exists = kh_get(off, cache->entries, offset) != kh_end(cache->entries); | |
174 | if (!exists) { | |
0ed75620 CMN |
175 | while (cache->memory_used + base->len > cache->memory_limit) |
176 | free_lowest_entry(cache); | |
177 | ||
c8f79c2b CMN |
178 | k = kh_put(off, cache->entries, offset, &error); |
179 | assert(error != 0); | |
180 | kh_value(cache->entries, k) = entry; | |
0ed75620 | 181 | cache->memory_used += entry->raw.len; |
8588cb0c JH |
182 | |
183 | *cached_out = entry; | |
c8f79c2b CMN |
184 | } |
185 | git_mutex_unlock(&cache->lock); | |
186 | /* Somebody beat us to adding it into the cache */ | |
187 | if (exists) { | |
188 | git__free(entry); | |
189 | return -1; | |
190 | } | |
191 | } | |
192 | ||
193 | return 0; | |
194 | } | |
195 | ||
a070f152 CMN |
196 | /*********************************************************** |
197 | * | |
198 | * PACK INDEX METHODS | |
199 | * | |
200 | ***********************************************************/ | |
201 | ||
202 | static void pack_index_free(struct git_pack_file *p) | |
203 | { | |
60ecdf59 DMB |
204 | if (p->oids) { |
205 | git__free(p->oids); | |
206 | p->oids = NULL; | |
207 | } | |
a070f152 CMN |
208 | if (p->index_map.data) { |
209 | git_futils_mmap_free(&p->index_map); | |
210 | p->index_map.data = NULL; | |
211 | } | |
212 | } | |
213 | ||
87d9869f | 214 | static int pack_index_check(const char *path, struct git_pack_file *p) |
a070f152 CMN |
215 | { |
216 | struct git_pack_idx_header *hdr; | |
217 | uint32_t version, nr, i, *index; | |
a070f152 CMN |
218 | void *idx_map; |
219 | size_t idx_size; | |
a070f152 | 220 | struct stat st; |
a070f152 | 221 | int error; |
e1de726c RB |
222 | /* TODO: properly open the file without access time using O_NOATIME */ |
223 | git_file fd = git_futils_open_ro(path); | |
a070f152 | 224 | if (fd < 0) |
e1de726c | 225 | return fd; |
a070f152 | 226 | |
9d2f841a RB |
227 | if (p_fstat(fd, &st) < 0) { |
228 | p_close(fd); | |
229 | giterr_set(GITERR_OS, "Unable to stat pack index '%s'", path); | |
230 | return -1; | |
231 | } | |
232 | ||
233 | if (!S_ISREG(st.st_mode) || | |
e1de726c RB |
234 | !git__is_sizet(st.st_size) || |
235 | (idx_size = (size_t)st.st_size) < 4 * 256 + 20 + 20) | |
236 | { | |
a070f152 | 237 | p_close(fd); |
9d2f841a | 238 | giterr_set(GITERR_ODB, "Invalid pack index '%s'", path); |
e1de726c | 239 | return -1; |
a070f152 CMN |
240 | } |
241 | ||
242 | error = git_futils_mmap_ro(&p->index_map, fd, 0, idx_size); | |
e1de726c | 243 | |
a070f152 CMN |
244 | p_close(fd); |
245 | ||
e1de726c RB |
246 | if (error < 0) |
247 | return error; | |
a070f152 CMN |
248 | |
249 | hdr = idx_map = p->index_map.data; | |
250 | ||
251 | if (hdr->idx_signature == htonl(PACK_IDX_SIGNATURE)) { | |
252 | version = ntohl(hdr->idx_version); | |
253 | ||
254 | if (version < 2 || version > 2) { | |
255 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 256 | return packfile_error("unsupported index version"); |
a070f152 CMN |
257 | } |
258 | ||
259 | } else | |
260 | version = 1; | |
261 | ||
262 | nr = 0; | |
263 | index = idx_map; | |
264 | ||
265 | if (version > 1) | |
87d9869f | 266 | index += 2; /* skip index header */ |
a070f152 CMN |
267 | |
268 | for (i = 0; i < 256; i++) { | |
269 | uint32_t n = ntohl(index[i]); | |
270 | if (n < nr) { | |
271 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 272 | return packfile_error("index is non-monotonic"); |
a070f152 CMN |
273 | } |
274 | nr = n; | |
275 | } | |
276 | ||
277 | if (version == 1) { | |
278 | /* | |
279 | * Total size: | |
87d9869f VM |
280 | * - 256 index entries 4 bytes each |
281 | * - 24-byte entries * nr (20-byte sha1 + 4-byte offset) | |
282 | * - 20-byte SHA1 of the packfile | |
283 | * - 20-byte SHA1 file checksum | |
a070f152 CMN |
284 | */ |
285 | if (idx_size != 4*256 + nr * 24 + 20 + 20) { | |
286 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 287 | return packfile_error("index is corrupted"); |
a070f152 CMN |
288 | } |
289 | } else if (version == 2) { | |
290 | /* | |
291 | * Minimum size: | |
87d9869f VM |
292 | * - 8 bytes of header |
293 | * - 256 index entries 4 bytes each | |
294 | * - 20-byte sha1 entry * nr | |
295 | * - 4-byte crc entry * nr | |
296 | * - 4-byte offset entry * nr | |
297 | * - 20-byte SHA1 of the packfile | |
298 | * - 20-byte SHA1 file checksum | |
a070f152 CMN |
299 | * And after the 4-byte offset table might be a |
300 | * variable sized table containing 8-byte entries | |
301 | * for offsets larger than 2^31. | |
302 | */ | |
303 | unsigned long min_size = 8 + 4*256 + nr*(20 + 4 + 4) + 20 + 20; | |
304 | unsigned long max_size = min_size; | |
305 | ||
306 | if (nr) | |
307 | max_size += (nr - 1)*8; | |
308 | ||
309 | if (idx_size < min_size || idx_size > max_size) { | |
310 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 311 | return packfile_error("wrong index size"); |
a070f152 | 312 | } |
a070f152 CMN |
313 | } |
314 | ||
a070f152 | 315 | p->num_objects = nr; |
0ddfcb40 | 316 | p->index_version = version; |
e1de726c | 317 | return 0; |
a070f152 CMN |
318 | } |
319 | ||
320 | static int pack_index_open(struct git_pack_file *p) | |
321 | { | |
24c70804 | 322 | int error = 0; |
878293f7 CMN |
323 | size_t name_len; |
324 | git_buf idx_name = GIT_BUF_INIT; | |
24c70804 | 325 | |
0ddfcb40 | 326 | if (p->index_version > -1) |
53607868 | 327 | return 0; |
a070f152 | 328 | |
24c70804 RB |
329 | name_len = strlen(p->pack_name); |
330 | assert(name_len > strlen(".pack")); /* checked by git_pack_file alloc */ | |
e1de726c | 331 | |
878293f7 CMN |
332 | git_buf_grow(&idx_name, name_len); |
333 | git_buf_put(&idx_name, p->pack_name, name_len - strlen(".pack")); | |
334 | git_buf_puts(&idx_name, ".idx"); | |
335 | if (git_buf_oom(&idx_name)) { | |
336 | giterr_set_oom(); | |
53607868 | 337 | return -1; |
878293f7 | 338 | } |
a070f152 | 339 | |
050af8bb | 340 | if ((error = git_mutex_lock(&p->lock)) < 0) { |
878293f7 | 341 | git_buf_free(&idx_name); |
53607868 | 342 | return error; |
050af8bb | 343 | } |
53607868 | 344 | |
0ddfcb40 | 345 | if (p->index_version == -1) |
878293f7 | 346 | error = pack_index_check(idx_name.ptr, p); |
24c70804 | 347 | |
878293f7 | 348 | git_buf_free(&idx_name); |
a070f152 | 349 | |
24c70804 RB |
350 | git_mutex_unlock(&p->lock); |
351 | ||
e1de726c | 352 | return error; |
a070f152 CMN |
353 | } |
354 | ||
355 | static unsigned char *pack_window_open( | |
356 | struct git_pack_file *p, | |
7d0cdf82 | 357 | git_mwindow **w_cursor, |
e1de726c | 358 | git_off_t offset, |
7d0cdf82 CMN |
359 | unsigned int *left) |
360 | { | |
e1de726c | 361 | if (p->mwf.fd == -1 && packfile_open(p) < 0) |
7d0cdf82 CMN |
362 | return NULL; |
363 | ||
364 | /* Since packfiles end in a hash of their content and it's | |
365 | * pointless to ask for an offset into the middle of that | |
366 | * hash, and the pack_window_contains function above wouldn't match | |
367 | * don't allow an offset too close to the end of the file. | |
6d97beb9 CMN |
368 | * |
369 | * Don't allow a negative offset, as that means we've wrapped | |
370 | * around. | |
7d0cdf82 CMN |
371 | */ |
372 | if (offset > (p->mwf.size - 20)) | |
373 | return NULL; | |
6d97beb9 CMN |
374 | if (offset < 0) |
375 | return NULL; | |
7d0cdf82 CMN |
376 | |
377 | return git_mwindow_open(&p->mwf, w_cursor, offset, 20, left); | |
378 | } | |
379 | ||
51e82492 CMN |
380 | /* |
381 | * The per-object header is a pretty dense thing, which is | |
382 | * - first byte: low four bits are "size", | |
383 | * then three bits of "type", | |
384 | * with the high bit being "size continues". | |
385 | * - each byte afterwards: low seven bits are size continuation, | |
386 | * with the high bit being "size continues" | |
387 | */ | |
51a3dfb5 | 388 | size_t git_packfile__object_header(unsigned char *hdr, size_t size, git_otype type) |
51e82492 CMN |
389 | { |
390 | unsigned char *hdr_base; | |
391 | unsigned char c; | |
392 | ||
393 | assert(type >= GIT_OBJ_COMMIT && type <= GIT_OBJ_REF_DELTA); | |
394 | ||
395 | /* TODO: add support for chunked objects; see git.git 6c0d19b1 */ | |
396 | ||
397 | c = (unsigned char)((type << 4) | (size & 15)); | |
398 | size >>= 4; | |
399 | hdr_base = hdr; | |
400 | ||
401 | while (size) { | |
402 | *hdr++ = c | 0x80; | |
403 | c = size & 0x7f; | |
404 | size >>= 7; | |
405 | } | |
406 | *hdr++ = c; | |
407 | ||
51a3dfb5 | 408 | return (hdr - hdr_base); |
51e82492 CMN |
409 | } |
410 | ||
411 | ||
45d773ef CMN |
412 | static int packfile_unpack_header1( |
413 | unsigned long *usedp, | |
7d0cdf82 CMN |
414 | size_t *sizep, |
415 | git_otype *type, | |
416 | const unsigned char *buf, | |
417 | unsigned long len) | |
418 | { | |
419 | unsigned shift; | |
420 | unsigned long size, c; | |
421 | unsigned long used = 0; | |
2aeadb9c | 422 | |
7d0cdf82 CMN |
423 | c = buf[used++]; |
424 | *type = (c >> 4) & 7; | |
425 | size = c & 15; | |
426 | shift = 4; | |
427 | while (c & 0x80) { | |
ec7e680c RP |
428 | if (len <= used) { |
429 | giterr_set(GITERR_ODB, "buffer too small"); | |
904b67e6 | 430 | return GIT_EBUFS; |
ec7e680c | 431 | } |
45d773ef CMN |
432 | |
433 | if (bitsizeof(long) <= shift) { | |
434 | *usedp = 0; | |
ec7e680c | 435 | giterr_set(GITERR_ODB, "packfile corrupted"); |
45d773ef CMN |
436 | return -1; |
437 | } | |
7d0cdf82 CMN |
438 | |
439 | c = buf[used++]; | |
440 | size += (c & 0x7f) << shift; | |
441 | shift += 7; | |
442 | } | |
443 | ||
444 | *sizep = (size_t)size; | |
45d773ef CMN |
445 | *usedp = used; |
446 | return 0; | |
7d0cdf82 CMN |
447 | } |
448 | ||
449 | int git_packfile_unpack_header( | |
450 | size_t *size_p, | |
451 | git_otype *type_p, | |
452 | git_mwindow_file *mwf, | |
453 | git_mwindow **w_curs, | |
e1de726c | 454 | git_off_t *curpos) |
7d0cdf82 CMN |
455 | { |
456 | unsigned char *base; | |
457 | unsigned int left; | |
458 | unsigned long used; | |
45d773ef | 459 | int ret; |
7d0cdf82 CMN |
460 | |
461 | /* pack_window_open() assures us we have [base, base + 20) available | |
87d9869f VM |
462 | * as a range that we can look at at. (Its actually the hash |
463 | * size that is assured.) With our object header encoding | |
7d0cdf82 CMN |
464 | * the maximum deflated object size is 2^137, which is just |
465 | * insane, so we know won't exceed what we have been given. | |
466 | */ | |
24c70804 | 467 | /* base = pack_window_open(p, w_curs, *curpos, &left); */ |
7d0cdf82 CMN |
468 | base = git_mwindow_open(mwf, w_curs, *curpos, 20, &left); |
469 | if (base == NULL) | |
904b67e6 | 470 | return GIT_EBUFS; |
2aeadb9c | 471 | |
9d2f841a | 472 | ret = packfile_unpack_header1(&used, size_p, type_p, base, left); |
45d773ef | 473 | git_mwindow_close(w_curs); |
904b67e6 | 474 | if (ret == GIT_EBUFS) |
45d773ef CMN |
475 | return ret; |
476 | else if (ret < 0) | |
e1de726c | 477 | return packfile_error("header length is zero"); |
7d0cdf82 CMN |
478 | |
479 | *curpos += used; | |
e1de726c | 480 | return 0; |
7d0cdf82 CMN |
481 | } |
482 | ||
44f9f547 DMB |
483 | int git_packfile_resolve_header( |
484 | size_t *size_p, | |
485 | git_otype *type_p, | |
486 | struct git_pack_file *p, | |
487 | git_off_t offset) | |
488 | { | |
489 | git_mwindow *w_curs = NULL; | |
490 | git_off_t curpos = offset; | |
491 | size_t size; | |
492 | git_otype type; | |
493 | git_off_t base_offset; | |
494 | int error; | |
495 | ||
496 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
44f9f547 DMB |
497 | if (error < 0) |
498 | return error; | |
499 | ||
500 | if (type == GIT_OBJ_OFS_DELTA || type == GIT_OBJ_REF_DELTA) { | |
501 | size_t base_size; | |
502 | git_rawobj delta; | |
503 | base_offset = get_delta_base(p, &w_curs, &curpos, type, offset); | |
504 | git_mwindow_close(&w_curs); | |
505 | error = packfile_unpack_compressed(&delta, p, &w_curs, &curpos, size, type); | |
506 | git_mwindow_close(&w_curs); | |
507 | if (error < 0) | |
508 | return error; | |
509 | error = git__delta_read_header(delta.data, delta.len, &base_size, size_p); | |
510 | git__free(delta.data); | |
511 | if (error < 0) | |
512 | return error; | |
513 | } else | |
514 | *size_p = size; | |
515 | ||
516 | while (type == GIT_OBJ_OFS_DELTA || type == GIT_OBJ_REF_DELTA) { | |
517 | curpos = base_offset; | |
518 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
44f9f547 DMB |
519 | if (error < 0) |
520 | return error; | |
521 | if (type != GIT_OBJ_OFS_DELTA && type != GIT_OBJ_REF_DELTA) | |
522 | break; | |
523 | base_offset = get_delta_base(p, &w_curs, &curpos, type, base_offset); | |
524 | git_mwindow_close(&w_curs); | |
525 | } | |
526 | *type_p = type; | |
527 | ||
528 | return error; | |
529 | } | |
530 | ||
15bcced2 CMN |
531 | #define SMALL_STACK_SIZE 64 |
532 | ||
a3ffbf23 CMN |
533 | /** |
534 | * Generate the chain of dependencies which we need to get to the | |
535 | * object at `off`. `chain` is used a stack, popping gives the right | |
536 | * order to apply deltas on. If an object is found in the pack's base | |
537 | * cache, we stop calculating there. | |
538 | */ | |
15bcced2 CMN |
539 | static int pack_dependency_chain(git_dependency_chain *chain_out, |
540 | git_pack_cache_entry **cached_out, git_off_t *cached_off, | |
541 | struct pack_chain_elem *small_stack, size_t *stack_sz, | |
542 | struct git_pack_file *p, git_off_t obj_offset) | |
a3ffbf23 CMN |
543 | { |
544 | git_dependency_chain chain = GIT_ARRAY_INIT; | |
545 | git_mwindow *w_curs = NULL; | |
546 | git_off_t curpos = obj_offset, base_offset; | |
15bcced2 CMN |
547 | int error = 0, use_heap = 0; |
548 | size_t size, elem_pos; | |
a3ffbf23 CMN |
549 | git_otype type; |
550 | ||
15bcced2 | 551 | elem_pos = 0; |
a3ffbf23 CMN |
552 | while (true) { |
553 | struct pack_chain_elem *elem; | |
554 | git_pack_cache_entry *cached = NULL; | |
555 | ||
556 | /* if we have a base cached, we can stop here instead */ | |
557 | if ((cached = cache_get(&p->bases, obj_offset)) != NULL) { | |
558 | *cached_out = cached; | |
559 | *cached_off = obj_offset; | |
560 | break; | |
561 | } | |
562 | ||
15bcced2 CMN |
563 | /* if we run out of space on the small stack, use the array */ |
564 | if (elem_pos == SMALL_STACK_SIZE) { | |
565 | git_array_init_to_size(chain, elem_pos); | |
566 | GITERR_CHECK_ARRAY(chain); | |
567 | memcpy(chain.ptr, small_stack, elem_pos * sizeof(struct pack_chain_elem)); | |
568 | chain.size = elem_pos; | |
569 | use_heap = 1; | |
570 | } | |
571 | ||
a3ffbf23 | 572 | curpos = obj_offset; |
15bcced2 CMN |
573 | if (!use_heap) { |
574 | elem = &small_stack[elem_pos]; | |
575 | } else { | |
576 | elem = git_array_alloc(chain); | |
577 | if (!elem) { | |
578 | error = -1; | |
579 | goto on_error; | |
580 | } | |
a3ffbf23 CMN |
581 | } |
582 | ||
583 | elem->base_key = obj_offset; | |
584 | ||
585 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
a3ffbf23 CMN |
586 | |
587 | if (error < 0) | |
588 | goto on_error; | |
589 | ||
590 | elem->offset = curpos; | |
591 | elem->size = size; | |
592 | elem->type = type; | |
593 | elem->base_key = obj_offset; | |
594 | ||
595 | if (type != GIT_OBJ_OFS_DELTA && type != GIT_OBJ_REF_DELTA) | |
596 | break; | |
597 | ||
598 | base_offset = get_delta_base(p, &w_curs, &curpos, type, obj_offset); | |
599 | git_mwindow_close(&w_curs); | |
600 | ||
601 | if (base_offset == 0) { | |
602 | error = packfile_error("delta offset is zero"); | |
603 | goto on_error; | |
604 | } | |
605 | if (base_offset < 0) { /* must actually be an error code */ | |
606 | error = (int)base_offset; | |
607 | goto on_error; | |
608 | } | |
609 | ||
610 | /* we need to pass the pos *after* the delta-base bit */ | |
611 | elem->offset = curpos; | |
612 | ||
613 | /* go through the loop again, but with the new object */ | |
614 | obj_offset = base_offset; | |
15bcced2 | 615 | elem_pos++; |
a3ffbf23 CMN |
616 | } |
617 | ||
15bcced2 CMN |
618 | |
619 | *stack_sz = elem_pos + 1; | |
a3ffbf23 CMN |
620 | *chain_out = chain; |
621 | return error; | |
622 | ||
623 | on_error: | |
624 | git_array_clear(chain); | |
625 | return error; | |
626 | } | |
627 | ||
a070f152 | 628 | int git_packfile_unpack( |
e1de726c RB |
629 | git_rawobj *obj, |
630 | struct git_pack_file *p, | |
631 | git_off_t *obj_offset) | |
7d0cdf82 CMN |
632 | { |
633 | git_mwindow *w_curs = NULL; | |
e1de726c | 634 | git_off_t curpos = *obj_offset; |
a332e91c CMN |
635 | int error, free_base = 0; |
636 | git_dependency_chain chain = GIT_ARRAY_INIT; | |
15bcced2 | 637 | struct pack_chain_elem *elem = NULL, *stack; |
a332e91c | 638 | git_pack_cache_entry *cached = NULL; |
15bcced2 | 639 | struct pack_chain_elem small_stack[SMALL_STACK_SIZE]; |
f1453c59 | 640 | size_t stack_size = 0, elem_pos, alloclen; |
2acdf4b8 | 641 | git_otype base_type; |
7d0cdf82 CMN |
642 | |
643 | /* | |
644 | * TODO: optionally check the CRC on the packfile | |
645 | */ | |
646 | ||
15bcced2 | 647 | error = pack_dependency_chain(&chain, &cached, obj_offset, small_stack, &stack_size, p, *obj_offset); |
2acdf4b8 CMN |
648 | if (error < 0) |
649 | return error; | |
650 | ||
7d0cdf82 CMN |
651 | obj->data = NULL; |
652 | obj->len = 0; | |
653 | obj->type = GIT_OBJ_BAD; | |
654 | ||
15bcced2 CMN |
655 | /* let's point to the right stack */ |
656 | stack = chain.ptr ? chain.ptr : small_stack; | |
657 | ||
658 | elem_pos = stack_size; | |
a3ffbf23 | 659 | if (cached) { |
a332e91c | 660 | memcpy(obj, &cached->raw, sizeof(git_rawobj)); |
a3ffbf23 | 661 | base_type = obj->type; |
15bcced2 | 662 | elem_pos--; /* stack_size includes the base, which isn't actually there */ |
a3ffbf23 | 663 | } else { |
15bcced2 | 664 | elem = &stack[--elem_pos]; |
a3ffbf23 | 665 | base_type = elem->type; |
a332e91c | 666 | } |
45d773ef | 667 | |
9dbd150f CMN |
668 | switch (base_type) { |
669 | case GIT_OBJ_COMMIT: | |
670 | case GIT_OBJ_TREE: | |
671 | case GIT_OBJ_BLOB: | |
672 | case GIT_OBJ_TAG: | |
a3ffbf23 | 673 | if (!cached) { |
9dbd150f CMN |
674 | curpos = elem->offset; |
675 | error = packfile_unpack_compressed(obj, p, &w_curs, &curpos, elem->size, elem->type); | |
676 | git_mwindow_close(&w_curs); | |
a3ffbf23 | 677 | base_type = elem->type; |
9dbd150f CMN |
678 | } |
679 | if (error < 0) | |
680 | goto cleanup; | |
681 | break; | |
682 | case GIT_OBJ_OFS_DELTA: | |
683 | case GIT_OBJ_REF_DELTA: | |
684 | error = packfile_error("dependency chain ends in a delta"); | |
685 | goto cleanup; | |
686 | default: | |
687 | error = packfile_error("invalid packfile type in header"); | |
688 | goto cleanup; | |
689 | } | |
690 | ||
a332e91c | 691 | /* |
c968ce2c | 692 | * Finding the object we want a cached base element is |
a332e91c CMN |
693 | * problematic, as we need to make sure we don't accidentally |
694 | * give the caller the cached object, which it would then feel | |
695 | * free to free, so we need to copy the data. | |
696 | */ | |
15bcced2 | 697 | if (cached && stack_size == 1) { |
a332e91c | 698 | void *data = obj->data; |
392702ee | 699 | |
f1453c59 ET |
700 | GITERR_CHECK_ALLOC_ADD(&alloclen, obj->len, 1); |
701 | obj->data = git__malloc(alloclen); | |
a332e91c | 702 | GITERR_CHECK_ALLOC(obj->data); |
392702ee | 703 | |
a332e91c CMN |
704 | memcpy(obj->data, data, obj->len + 1); |
705 | git_atomic_dec(&cached->refcount); | |
706 | goto cleanup; | |
707 | } | |
708 | ||
2acdf4b8 | 709 | /* we now apply each consecutive delta until we run out */ |
15bcced2 | 710 | while (elem_pos > 0 && !error) { |
2acdf4b8 CMN |
711 | git_rawobj base, delta; |
712 | ||
c968ce2c CMN |
713 | /* |
714 | * We can now try to add the base to the cache, as | |
715 | * long as it's not already the cached one. | |
716 | */ | |
717 | if (!cached) | |
8588cb0c | 718 | free_base = !!cache_add(&cached, &p->bases, obj, elem->base_key); |
c968ce2c | 719 | |
15bcced2 | 720 | elem = &stack[elem_pos - 1]; |
2acdf4b8 CMN |
721 | curpos = elem->offset; |
722 | error = packfile_unpack_compressed(&delta, p, &w_curs, &curpos, elem->size, elem->type); | |
723 | git_mwindow_close(&w_curs); | |
724 | ||
725 | if (error < 0) | |
726 | break; | |
727 | ||
728 | /* the current object becomes the new base, on which we apply the delta */ | |
729 | base = *obj; | |
730 | obj->data = NULL; | |
731 | obj->len = 0; | |
732 | obj->type = GIT_OBJ_BAD; | |
733 | ||
734 | error = git__delta_apply(obj, base.data, base.len, delta.data, delta.len); | |
a332e91c CMN |
735 | obj->type = base_type; |
736 | /* | |
737 | * We usually don't want to free the base at this | |
738 | * point, as we put it into the cache in the previous | |
739 | * iteration. free_base lets us know that we got the | |
740 | * base object directly from the packfile, so we can free it. | |
741 | */ | |
2acdf4b8 | 742 | git__free(delta.data); |
a332e91c CMN |
743 | if (free_base) { |
744 | free_base = 0; | |
745 | git__free(base.data); | |
746 | } | |
747 | ||
748 | if (cached) { | |
749 | git_atomic_dec(&cached->refcount); | |
750 | cached = NULL; | |
751 | } | |
2acdf4b8 CMN |
752 | |
753 | if (error < 0) | |
754 | break; | |
7d0cdf82 | 755 | |
15bcced2 | 756 | elem_pos--; |
7d0cdf82 CMN |
757 | } |
758 | ||
2acdf4b8 | 759 | cleanup: |
a332e91c CMN |
760 | if (error < 0) |
761 | git__free(obj->data); | |
762 | ||
a3ffbf23 | 763 | if (elem) |
b3d3459f | 764 | *obj_offset = curpos; |
a332e91c | 765 | |
2acdf4b8 | 766 | git_array_clear(chain); |
e1de726c | 767 | return error; |
7d0cdf82 CMN |
768 | } |
769 | ||
282283ac RB |
770 | static void *use_git_alloc(void *opaq, unsigned int count, unsigned int size) |
771 | { | |
772 | GIT_UNUSED(opaq); | |
773 | return git__calloc(count, size); | |
774 | } | |
775 | ||
776 | static void use_git_free(void *opaq, void *ptr) | |
777 | { | |
778 | GIT_UNUSED(opaq); | |
779 | git__free(ptr); | |
780 | } | |
781 | ||
46635339 CMN |
782 | int git_packfile_stream_open(git_packfile_stream *obj, struct git_pack_file *p, git_off_t curpos) |
783 | { | |
784 | int st; | |
785 | ||
786 | memset(obj, 0, sizeof(git_packfile_stream)); | |
787 | obj->curpos = curpos; | |
788 | obj->p = p; | |
789 | obj->zstream.zalloc = use_git_alloc; | |
790 | obj->zstream.zfree = use_git_free; | |
791 | obj->zstream.next_in = Z_NULL; | |
792 | obj->zstream.next_out = Z_NULL; | |
793 | st = inflateInit(&obj->zstream); | |
794 | if (st != Z_OK) { | |
ae081739 | 795 | giterr_set(GITERR_ZLIB, "failed to init packfile stream"); |
46635339 CMN |
796 | return -1; |
797 | } | |
798 | ||
799 | return 0; | |
800 | } | |
801 | ||
802 | ssize_t git_packfile_stream_read(git_packfile_stream *obj, void *buffer, size_t len) | |
803 | { | |
804 | unsigned char *in; | |
805 | size_t written; | |
806 | int st; | |
807 | ||
808 | if (obj->done) | |
809 | return 0; | |
810 | ||
811 | in = pack_window_open(obj->p, &obj->mw, obj->curpos, &obj->zstream.avail_in); | |
812 | if (in == NULL) | |
813 | return GIT_EBUFS; | |
814 | ||
815 | obj->zstream.next_out = buffer; | |
090d5e1f | 816 | obj->zstream.avail_out = (unsigned int)len; |
46635339 CMN |
817 | obj->zstream.next_in = in; |
818 | ||
819 | st = inflate(&obj->zstream, Z_SYNC_FLUSH); | |
820 | git_mwindow_close(&obj->mw); | |
821 | ||
822 | obj->curpos += obj->zstream.next_in - in; | |
823 | written = len - obj->zstream.avail_out; | |
824 | ||
825 | if (st != Z_OK && st != Z_STREAM_END) { | |
ae081739 | 826 | giterr_set(GITERR_ZLIB, "error reading from the zlib stream"); |
46635339 CMN |
827 | return -1; |
828 | } | |
829 | ||
830 | if (st == Z_STREAM_END) | |
831 | obj->done = 1; | |
832 | ||
833 | ||
834 | /* If we didn't write anything out but we're not done, we need more data */ | |
835 | if (!written && st != Z_STREAM_END) | |
836 | return GIT_EBUFS; | |
837 | ||
838 | return written; | |
839 | ||
840 | } | |
841 | ||
842 | void git_packfile_stream_free(git_packfile_stream *obj) | |
843 | { | |
844 | inflateEnd(&obj->zstream); | |
845 | } | |
846 | ||
b644e223 | 847 | static int packfile_unpack_compressed( |
e1de726c RB |
848 | git_rawobj *obj, |
849 | struct git_pack_file *p, | |
850 | git_mwindow **w_curs, | |
851 | git_off_t *curpos, | |
852 | size_t size, | |
853 | git_otype type) | |
7d0cdf82 | 854 | { |
f1453c59 | 855 | size_t buf_size; |
7d0cdf82 CMN |
856 | int st; |
857 | z_stream stream; | |
858 | unsigned char *buffer, *in; | |
859 | ||
f1453c59 ET |
860 | GITERR_CHECK_ALLOC_ADD(&buf_size, size, 1); |
861 | buffer = git__calloc(1, buf_size); | |
e1de726c | 862 | GITERR_CHECK_ALLOC(buffer); |
7d0cdf82 CMN |
863 | |
864 | memset(&stream, 0, sizeof(stream)); | |
865 | stream.next_out = buffer; | |
f1453c59 | 866 | stream.avail_out = (uInt)buf_size; |
282283ac RB |
867 | stream.zalloc = use_git_alloc; |
868 | stream.zfree = use_git_free; | |
7d0cdf82 CMN |
869 | |
870 | st = inflateInit(&stream); | |
871 | if (st != Z_OK) { | |
3286c408 | 872 | git__free(buffer); |
ae081739 | 873 | giterr_set(GITERR_ZLIB, "failed to init zlib stream on unpack"); |
45d773ef | 874 | |
e1de726c | 875 | return -1; |
7d0cdf82 CMN |
876 | } |
877 | ||
878 | do { | |
b5b474dd | 879 | in = pack_window_open(p, w_curs, *curpos, &stream.avail_in); |
7d0cdf82 CMN |
880 | stream.next_in = in; |
881 | st = inflate(&stream, Z_FINISH); | |
45d773ef | 882 | git_mwindow_close(w_curs); |
7d0cdf82 CMN |
883 | |
884 | if (!stream.avail_out) | |
885 | break; /* the payload is larger than it should be */ | |
886 | ||
45d773ef CMN |
887 | if (st == Z_BUF_ERROR && in == NULL) { |
888 | inflateEnd(&stream); | |
889 | git__free(buffer); | |
904b67e6 | 890 | return GIT_EBUFS; |
45d773ef CMN |
891 | } |
892 | ||
b5b474dd | 893 | *curpos += stream.next_in - in; |
7d0cdf82 CMN |
894 | } while (st == Z_OK || st == Z_BUF_ERROR); |
895 | ||
896 | inflateEnd(&stream); | |
897 | ||
898 | if ((st != Z_STREAM_END) || stream.total_out != size) { | |
3286c408 | 899 | git__free(buffer); |
ae081739 | 900 | giterr_set(GITERR_ZLIB, "error inflating zlib stream"); |
e1de726c | 901 | return -1; |
7d0cdf82 CMN |
902 | } |
903 | ||
904 | obj->type = type; | |
905 | obj->len = size; | |
906 | obj->data = buffer; | |
e1de726c | 907 | return 0; |
7d0cdf82 CMN |
908 | } |
909 | ||
b5b474dd CMN |
910 | /* |
911 | * curpos is where the data starts, delta_obj_offset is the where the | |
912 | * header starts | |
913 | */ | |
e1de726c RB |
914 | git_off_t get_delta_base( |
915 | struct git_pack_file *p, | |
916 | git_mwindow **w_curs, | |
917 | git_off_t *curpos, | |
918 | git_otype type, | |
919 | git_off_t delta_obj_offset) | |
7d0cdf82 | 920 | { |
45d773ef CMN |
921 | unsigned int left = 0; |
922 | unsigned char *base_info; | |
e1de726c | 923 | git_off_t base_offset; |
7d0cdf82 CMN |
924 | git_oid unused; |
925 | ||
45d773ef CMN |
926 | base_info = pack_window_open(p, w_curs, *curpos, &left); |
927 | /* Assumption: the only reason this would fail is because the file is too small */ | |
928 | if (base_info == NULL) | |
904b67e6 | 929 | return GIT_EBUFS; |
7d0cdf82 CMN |
930 | /* pack_window_open() assured us we have [base_info, base_info + 20) |
931 | * as a range that we can look at without walking off the | |
87d9869f VM |
932 | * end of the mapped window. Its actually the hash size |
933 | * that is assured. An OFS_DELTA longer than the hash size | |
7d0cdf82 CMN |
934 | * is stupid, as then a REF_DELTA would be smaller to store. |
935 | */ | |
936 | if (type == GIT_OBJ_OFS_DELTA) { | |
937 | unsigned used = 0; | |
938 | unsigned char c = base_info[used++]; | |
939 | base_offset = c & 127; | |
940 | while (c & 128) { | |
45d773ef | 941 | if (left <= used) |
904b67e6 | 942 | return GIT_EBUFS; |
7d0cdf82 CMN |
943 | base_offset += 1; |
944 | if (!base_offset || MSB(base_offset, 7)) | |
87d9869f | 945 | return 0; /* overflow */ |
7d0cdf82 CMN |
946 | c = base_info[used++]; |
947 | base_offset = (base_offset << 7) + (c & 127); | |
948 | } | |
949 | base_offset = delta_obj_offset - base_offset; | |
950 | if (base_offset <= 0 || base_offset >= delta_obj_offset) | |
87d9869f | 951 | return 0; /* out of bound */ |
7d0cdf82 CMN |
952 | *curpos += used; |
953 | } else if (type == GIT_OBJ_REF_DELTA) { | |
c1af5a39 CMN |
954 | /* If we have the cooperative cache, search in it first */ |
955 | if (p->has_cache) { | |
0e040c03 CMN |
956 | khiter_t k; |
957 | git_oid oid; | |
c1af5a39 | 958 | |
0e040c03 CMN |
959 | git_oid_fromraw(&oid, base_info); |
960 | k = kh_get(oid, p->idx_cache, &oid); | |
961 | if (k != kh_end(p->idx_cache)) { | |
c1af5a39 | 962 | *curpos += 20; |
0e040c03 | 963 | return ((struct git_pack_entry *)kh_value(p->idx_cache, k))->offset; |
38c10ecd ET |
964 | } else { |
965 | /* If we're building an index, don't try to find the pack | |
966 | * entry; we just haven't seen it yet. We'll make | |
967 | * progress again in the next loop. | |
968 | */ | |
969 | return GIT_PASSTHROUGH; | |
c1af5a39 CMN |
970 | } |
971 | } | |
38c10ecd | 972 | |
7d0cdf82 | 973 | /* The base entry _must_ be in the same pack */ |
e1de726c RB |
974 | if (pack_entry_find_offset(&base_offset, &unused, p, (git_oid *)base_info, GIT_OID_HEXSZ) < 0) |
975 | return packfile_error("base entry delta is not in the same pack"); | |
7d0cdf82 CMN |
976 | *curpos += 20; |
977 | } else | |
978 | return 0; | |
979 | ||
980 | return base_offset; | |
981 | } | |
a070f152 CMN |
982 | |
983 | /*********************************************************** | |
984 | * | |
985 | * PACKFILE METHODS | |
986 | * | |
987 | ***********************************************************/ | |
988 | ||
96c9b9f0 | 989 | void git_packfile_free(struct git_pack_file *p) |
a070f152 | 990 | { |
24c70804 RB |
991 | if (!p) |
992 | return; | |
993 | ||
c8f79c2b | 994 | cache_free(&p->bases); |
c0f4a011 | 995 | |
5e0f47c3 CMN |
996 | if (p->mwf.fd >= 0) { |
997 | git_mwindow_free_all_locked(&p->mwf); | |
a070f152 | 998 | p_close(p->mwf.fd); |
5e0f47c3 | 999 | } |
a070f152 CMN |
1000 | |
1001 | pack_index_free(p); | |
1002 | ||
3286c408 | 1003 | git__free(p->bad_object_sha1); |
24c70804 | 1004 | |
24c70804 | 1005 | git_mutex_free(&p->lock); |
649214be | 1006 | git_mutex_free(&p->bases.lock); |
3286c408 | 1007 | git__free(p); |
a070f152 CMN |
1008 | } |
1009 | ||
1010 | static int packfile_open(struct git_pack_file *p) | |
1011 | { | |
1012 | struct stat st; | |
1013 | struct git_pack_header hdr; | |
1014 | git_oid sha1; | |
1015 | unsigned char *idx_sha1; | |
1016 | ||
0ddfcb40 | 1017 | if (p->index_version == -1 && pack_index_open(p) < 0) |
e10144ae | 1018 | return git_odb__error_notfound("failed to open packfile", NULL, 0); |
a070f152 | 1019 | |
9d2f841a RB |
1020 | /* if mwf opened by another thread, return now */ |
1021 | if (git_mutex_lock(&p->lock) < 0) | |
1022 | return packfile_error("failed to get lock for open"); | |
1023 | ||
1024 | if (p->mwf.fd >= 0) { | |
1025 | git_mutex_unlock(&p->lock); | |
1026 | return 0; | |
1027 | } | |
1028 | ||
a070f152 | 1029 | /* TODO: open with noatime */ |
e1de726c | 1030 | p->mwf.fd = git_futils_open_ro(p->pack_name); |
9d2f841a RB |
1031 | if (p->mwf.fd < 0) |
1032 | goto cleanup; | |
a070f152 | 1033 | |
e1de726c RB |
1034 | if (p_fstat(p->mwf.fd, &st) < 0 || |
1035 | git_mwindow_file_register(&p->mwf) < 0) | |
1036 | goto cleanup; | |
a070f152 CMN |
1037 | |
1038 | /* If we created the struct before we had the pack we lack size. */ | |
1039 | if (!p->mwf.size) { | |
1040 | if (!S_ISREG(st.st_mode)) | |
1041 | goto cleanup; | |
e1de726c | 1042 | p->mwf.size = (git_off_t)st.st_size; |
a070f152 CMN |
1043 | } else if (p->mwf.size != st.st_size) |
1044 | goto cleanup; | |
1045 | ||
1046 | #if 0 | |
1047 | /* We leave these file descriptors open with sliding mmap; | |
1048 | * there is no point keeping them open across exec(), though. | |
1049 | */ | |
1050 | fd_flag = fcntl(p->mwf.fd, F_GETFD, 0); | |
1051 | if (fd_flag < 0) | |
e1de726c | 1052 | goto cleanup; |
a070f152 CMN |
1053 | |
1054 | fd_flag |= FD_CLOEXEC; | |
1055 | if (fcntl(p->pack_fd, F_SETFD, fd_flag) == -1) | |
e1de726c | 1056 | goto cleanup; |
a070f152 CMN |
1057 | #endif |
1058 | ||
1059 | /* Verify we recognize this pack file format. */ | |
e1de726c RB |
1060 | if (p_read(p->mwf.fd, &hdr, sizeof(hdr)) < 0 || |
1061 | hdr.hdr_signature != htonl(PACK_SIGNATURE) || | |
1062 | !pack_version_ok(hdr.hdr_version)) | |
a070f152 CMN |
1063 | goto cleanup; |
1064 | ||
1065 | /* Verify the pack matches its index. */ | |
e1de726c RB |
1066 | if (p->num_objects != ntohl(hdr.hdr_entries) || |
1067 | p_lseek(p->mwf.fd, p->mwf.size - GIT_OID_RAWSZ, SEEK_SET) == -1 || | |
1068 | p_read(p->mwf.fd, sha1.id, GIT_OID_RAWSZ) < 0) | |
a070f152 CMN |
1069 | goto cleanup; |
1070 | ||
1071 | idx_sha1 = ((unsigned char *)p->index_map.data) + p->index_map.len - 40; | |
1072 | ||
9d2f841a RB |
1073 | if (git_oid__cmp(&sha1, (git_oid *)idx_sha1) != 0) |
1074 | goto cleanup; | |
1075 | ||
1076 | git_mutex_unlock(&p->lock); | |
1077 | return 0; | |
a070f152 CMN |
1078 | |
1079 | cleanup: | |
e1de726c | 1080 | giterr_set(GITERR_OS, "Invalid packfile '%s'", p->pack_name); |
9d2f841a | 1081 | |
3a2d48d5 SS |
1082 | if (p->mwf.fd >= 0) |
1083 | p_close(p->mwf.fd); | |
a070f152 | 1084 | p->mwf.fd = -1; |
9d2f841a RB |
1085 | |
1086 | git_mutex_unlock(&p->lock); | |
1087 | ||
e1de726c | 1088 | return -1; |
a070f152 CMN |
1089 | } |
1090 | ||
b3b66c57 CMN |
1091 | int git_packfile__name(char **out, const char *path) |
1092 | { | |
1093 | size_t path_len; | |
1094 | git_buf buf = GIT_BUF_INIT; | |
1095 | ||
1096 | path_len = strlen(path); | |
1097 | ||
1098 | if (path_len < strlen(".idx")) | |
e10144ae | 1099 | return git_odb__error_notfound("invalid packfile path", NULL, 0); |
b3b66c57 CMN |
1100 | |
1101 | if (git_buf_printf(&buf, "%.*s.pack", (int)(path_len - strlen(".idx")), path) < 0) | |
1102 | return -1; | |
1103 | ||
1104 | *out = git_buf_detach(&buf); | |
1105 | return 0; | |
1106 | } | |
1107 | ||
5d2d21e5 | 1108 | int git_packfile_alloc(struct git_pack_file **pack_out, const char *path) |
a070f152 CMN |
1109 | { |
1110 | struct stat st; | |
1111 | struct git_pack_file *p; | |
f1453c59 | 1112 | size_t path_len = path ? strlen(path) : 0, alloc_len; |
a070f152 CMN |
1113 | |
1114 | *pack_out = NULL; | |
24c70804 | 1115 | |
38eef611 | 1116 | if (path_len < strlen(".idx")) |
e10144ae | 1117 | return git_odb__error_notfound("invalid packfile path", NULL, 0); |
24c70804 | 1118 | |
f1453c59 ET |
1119 | GITERR_CHECK_ALLOC_ADD(&alloc_len, sizeof(*p), path_len); |
1120 | GITERR_CHECK_ALLOC_ADD(&alloc_len, alloc_len, 2); | |
392702ee | 1121 | |
f1453c59 | 1122 | p = git__calloc(1, alloc_len); |
e1de726c | 1123 | GITERR_CHECK_ALLOC(p); |
a070f152 | 1124 | |
38eef611 RB |
1125 | memcpy(p->pack_name, path, path_len + 1); |
1126 | ||
a070f152 CMN |
1127 | /* |
1128 | * Make sure a corresponding .pack file exists and that | |
1129 | * the index looks sane. | |
1130 | */ | |
38eef611 RB |
1131 | if (git__suffixcmp(path, ".idx") == 0) { |
1132 | size_t root_len = path_len - strlen(".idx"); | |
1133 | ||
1134 | memcpy(p->pack_name + root_len, ".keep", sizeof(".keep")); | |
1135 | if (git_path_exists(p->pack_name) == true) | |
1136 | p->pack_keep = 1; | |
a070f152 | 1137 | |
38eef611 | 1138 | memcpy(p->pack_name + root_len, ".pack", sizeof(".pack")); |
38eef611 | 1139 | } |
a070f152 | 1140 | |
e1de726c | 1141 | if (p_stat(p->pack_name, &st) < 0 || !S_ISREG(st.st_mode)) { |
3286c408 | 1142 | git__free(p); |
e10144ae | 1143 | return git_odb__error_notfound("packfile not found", NULL, 0); |
a070f152 CMN |
1144 | } |
1145 | ||
1146 | /* ok, it looks sane as far as we can check without | |
1147 | * actually mapping the pack file. | |
1148 | */ | |
38eef611 | 1149 | p->mwf.fd = -1; |
1af56d7d | 1150 | p->mwf.size = st.st_size; |
a070f152 CMN |
1151 | p->pack_local = 1; |
1152 | p->mtime = (git_time_t)st.st_mtime; | |
0ddfcb40 | 1153 | p->index_version = -1; |
a070f152 | 1154 | |
1a42dd17 RB |
1155 | if (git_mutex_init(&p->lock)) { |
1156 | giterr_set(GITERR_OS, "Failed to initialize packfile mutex"); | |
1157 | git__free(p); | |
1158 | return -1; | |
1159 | } | |
38eef611 | 1160 | |
649214be CMN |
1161 | if (cache_init(&p->bases) < 0) { |
1162 | git__free(p); | |
1163 | return -1; | |
1164 | } | |
1165 | ||
a070f152 | 1166 | *pack_out = p; |
e1de726c RB |
1167 | |
1168 | return 0; | |
a070f152 CMN |
1169 | } |
1170 | ||
1171 | /*********************************************************** | |
1172 | * | |
1173 | * PACKFILE ENTRY SEARCH INTERNALS | |
1174 | * | |
1175 | ***********************************************************/ | |
1176 | ||
e1de726c | 1177 | static git_off_t nth_packed_object_offset(const struct git_pack_file *p, uint32_t n) |
a070f152 CMN |
1178 | { |
1179 | const unsigned char *index = p->index_map.data; | |
ea9e00cb | 1180 | const unsigned char *end = index + p->index_map.len; |
a070f152 CMN |
1181 | index += 4 * 256; |
1182 | if (p->index_version == 1) { | |
1183 | return ntohl(*((uint32_t *)(index + 24 * n))); | |
1184 | } else { | |
1185 | uint32_t off; | |
1186 | index += 8 + p->num_objects * (20 + 4); | |
1187 | off = ntohl(*((uint32_t *)(index + 4 * n))); | |
1188 | if (!(off & 0x80000000)) | |
1189 | return off; | |
1190 | index += p->num_objects * 4 + (off & 0x7fffffff) * 8; | |
ea9e00cb CMN |
1191 | |
1192 | /* Make sure we're not being sent out of bounds */ | |
1193 | if (index >= end - 8) | |
1194 | return -1; | |
1195 | ||
a070f152 | 1196 | return (((uint64_t)ntohl(*((uint32_t *)(index + 0)))) << 32) | |
87d9869f | 1197 | ntohl(*((uint32_t *)(index + 4))); |
a070f152 CMN |
1198 | } |
1199 | } | |
1200 | ||
60ecdf59 DMB |
1201 | static int git__memcmp4(const void *a, const void *b) { |
1202 | return memcmp(a, b, 4); | |
1203 | } | |
1204 | ||
521aedad | 1205 | int git_pack_foreach_entry( |
5dca2010 | 1206 | struct git_pack_file *p, |
c3fb7d04 | 1207 | git_odb_foreach_cb cb, |
5dca2010 | 1208 | void *data) |
521aedad CMN |
1209 | { |
1210 | const unsigned char *index = p->index_map.data, *current; | |
521aedad | 1211 | uint32_t i; |
dab89f9b | 1212 | int error = 0; |
521aedad CMN |
1213 | |
1214 | if (index == NULL) { | |
521aedad CMN |
1215 | if ((error = pack_index_open(p)) < 0) |
1216 | return error; | |
1217 | ||
1218 | assert(p->index_map.data); | |
1219 | ||
1220 | index = p->index_map.data; | |
1221 | } | |
1222 | ||
1223 | if (p->index_version > 1) { | |
1224 | index += 8; | |
1225 | } | |
1226 | ||
1227 | index += 4 * 256; | |
1228 | ||
60ecdf59 DMB |
1229 | if (p->oids == NULL) { |
1230 | git_vector offsets, oids; | |
521aedad | 1231 | |
60ecdf59 DMB |
1232 | if ((error = git_vector_init(&oids, p->num_objects, NULL))) |
1233 | return error; | |
1234 | ||
1235 | if ((error = git_vector_init(&offsets, p->num_objects, git__memcmp4))) | |
1236 | return error; | |
5dca2010 | 1237 | |
60ecdf59 DMB |
1238 | if (p->index_version > 1) { |
1239 | const unsigned char *off = index + 24 * p->num_objects; | |
1240 | for (i = 0; i < p->num_objects; i++) | |
1241 | git_vector_insert(&offsets, (void*)&off[4 * i]); | |
1242 | git_vector_sort(&offsets); | |
1243 | git_vector_foreach(&offsets, i, current) | |
1244 | git_vector_insert(&oids, (void*)&index[5 * (current - off)]); | |
1245 | } else { | |
1246 | for (i = 0; i < p->num_objects; i++) | |
1247 | git_vector_insert(&offsets, (void*)&index[24 * i]); | |
1248 | git_vector_sort(&offsets); | |
1249 | git_vector_foreach(&offsets, i, current) | |
1250 | git_vector_insert(&oids, (void*)¤t[4]); | |
1251 | } | |
25e0b157 | 1252 | |
60ecdf59 | 1253 | git_vector_free(&offsets); |
25e0b157 | 1254 | p->oids = (git_oid **)git_vector_detach(NULL, NULL, &oids); |
521aedad CMN |
1255 | } |
1256 | ||
60ecdf59 | 1257 | for (i = 0; i < p->num_objects; i++) |
26c1cb91 RB |
1258 | if ((error = cb(p->oids[i], data)) != 0) |
1259 | return giterr_set_after_callback(error); | |
60ecdf59 | 1260 | |
25e0b157 | 1261 | return error; |
521aedad CMN |
1262 | } |
1263 | ||
a070f152 | 1264 | static int pack_entry_find_offset( |
e1de726c RB |
1265 | git_off_t *offset_out, |
1266 | git_oid *found_oid, | |
1267 | struct git_pack_file *p, | |
1268 | const git_oid *short_oid, | |
b8457baa | 1269 | size_t len) |
a070f152 | 1270 | { |
34bd5999 CMN |
1271 | const uint32_t *level1_ofs = p->index_map.data; |
1272 | const unsigned char *index = p->index_map.data; | |
a070f152 CMN |
1273 | unsigned hi, lo, stride; |
1274 | int pos, found = 0; | |
ea9e00cb | 1275 | git_off_t offset; |
a070f152 CMN |
1276 | const unsigned char *current = 0; |
1277 | ||
1278 | *offset_out = 0; | |
1279 | ||
0ddfcb40 | 1280 | if (p->index_version == -1) { |
34bd5999 | 1281 | int error; |
a070f152 | 1282 | |
34bd5999 CMN |
1283 | if ((error = pack_index_open(p)) < 0) |
1284 | return error; | |
1285 | assert(p->index_map.data); | |
8c535f3f | 1286 | |
34bd5999 CMN |
1287 | index = p->index_map.data; |
1288 | level1_ofs = p->index_map.data; | |
1289 | } | |
a070f152 CMN |
1290 | |
1291 | if (p->index_version > 1) { | |
1292 | level1_ofs += 2; | |
1293 | index += 8; | |
1294 | } | |
1295 | ||
1296 | index += 4 * 256; | |
1297 | hi = ntohl(level1_ofs[(int)short_oid->id[0]]); | |
1298 | lo = ((short_oid->id[0] == 0x0) ? 0 : ntohl(level1_ofs[(int)short_oid->id[0] - 1])); | |
1299 | ||
1300 | if (p->index_version > 1) { | |
1301 | stride = 20; | |
1302 | } else { | |
1303 | stride = 24; | |
1304 | index += 4; | |
1305 | } | |
1306 | ||
1307 | #ifdef INDEX_DEBUG_LOOKUP | |
1308 | printf("%02x%02x%02x... lo %u hi %u nr %d\n", | |
1309 | short_oid->id[0], short_oid->id[1], short_oid->id[2], lo, hi, p->num_objects); | |
1310 | #endif | |
1311 | ||
67591c8c | 1312 | #ifdef GIT_USE_LOOKUP |
87d9869f | 1313 | pos = sha1_entry_pos(index, stride, 0, lo, hi, p->num_objects, short_oid->id); |
67591c8c VM |
1314 | #else |
1315 | pos = sha1_position(index, stride, lo, hi, short_oid->id); | |
1316 | #endif | |
a070f152 CMN |
1317 | |
1318 | if (pos >= 0) { | |
1319 | /* An object matching exactly the oid was found */ | |
1320 | found = 1; | |
1321 | current = index + pos * stride; | |
1322 | } else { | |
1323 | /* No object was found */ | |
1324 | /* pos refers to the object with the "closest" oid to short_oid */ | |
1325 | pos = - 1 - pos; | |
1326 | if (pos < (int)p->num_objects) { | |
1327 | current = index + pos * stride; | |
1328 | ||
282283ac | 1329 | if (!git_oid_ncmp(short_oid, (const git_oid *)current, len)) |
a070f152 | 1330 | found = 1; |
a070f152 CMN |
1331 | } |
1332 | } | |
1333 | ||
b2a2702d | 1334 | if (found && len != GIT_OID_HEXSZ && pos + 1 < (int)p->num_objects) { |
a070f152 CMN |
1335 | /* Check for ambiguousity */ |
1336 | const unsigned char *next = current + stride; | |
1337 | ||
1338 | if (!git_oid_ncmp(short_oid, (const git_oid *)next, len)) { | |
1339 | found = 2; | |
1340 | } | |
1341 | } | |
1342 | ||
e1de726c | 1343 | if (!found) |
e10144ae | 1344 | return git_odb__error_notfound("failed to find offset for pack entry", short_oid, len); |
e1de726c RB |
1345 | if (found > 1) |
1346 | return git_odb__error_ambiguous("found multiple offsets for pack entry"); | |
24c70804 | 1347 | |
ea9e00cb CMN |
1348 | if ((offset = nth_packed_object_offset(p, pos)) < 0) { |
1349 | giterr_set(GITERR_ODB, "packfile index is corrupt"); | |
1350 | return -1; | |
1351 | } | |
1352 | ||
1353 | *offset_out = offset; | |
e1de726c | 1354 | git_oid_fromraw(found_oid, current); |
a070f152 CMN |
1355 | |
1356 | #ifdef INDEX_DEBUG_LOOKUP | |
e1de726c | 1357 | { |
a070f152 CMN |
1358 | unsigned char hex_sha1[GIT_OID_HEXSZ + 1]; |
1359 | git_oid_fmt(hex_sha1, found_oid); | |
1360 | hex_sha1[GIT_OID_HEXSZ] = '\0'; | |
1361 | printf("found lo=%d %s\n", lo, hex_sha1); | |
a070f152 | 1362 | } |
e1de726c | 1363 | #endif |
24c70804 | 1364 | |
e1de726c | 1365 | return 0; |
a070f152 CMN |
1366 | } |
1367 | ||
1368 | int git_pack_entry_find( | |
1369 | struct git_pack_entry *e, | |
1370 | struct git_pack_file *p, | |
1371 | const git_oid *short_oid, | |
b8457baa | 1372 | size_t len) |
a070f152 | 1373 | { |
e1de726c | 1374 | git_off_t offset; |
a070f152 CMN |
1375 | git_oid found_oid; |
1376 | int error; | |
1377 | ||
1378 | assert(p); | |
1379 | ||
1380 | if (len == GIT_OID_HEXSZ && p->num_bad_objects) { | |
1381 | unsigned i; | |
1382 | for (i = 0; i < p->num_bad_objects; i++) | |
b7f167da | 1383 | if (git_oid__cmp(short_oid, &p->bad_object_sha1[i]) == 0) |
e1de726c | 1384 | return packfile_error("bad object found in packfile"); |
a070f152 CMN |
1385 | } |
1386 | ||
1387 | error = pack_entry_find_offset(&offset, &found_oid, p, short_oid, len); | |
e1de726c RB |
1388 | if (error < 0) |
1389 | return error; | |
a070f152 CMN |
1390 | |
1391 | /* we found a unique entry in the index; | |
1392 | * make sure the packfile backing the index | |
1393 | * still exists on disk */ | |
e1de726c RB |
1394 | if (p->mwf.fd == -1 && (error = packfile_open(p)) < 0) |
1395 | return error; | |
a070f152 CMN |
1396 | |
1397 | e->offset = offset; | |
1398 | e->p = p; | |
1399 | ||
1400 | git_oid_cpy(&e->sha1, &found_oid); | |
e1de726c | 1401 | return 0; |
a070f152 | 1402 | } |