]>
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 | ||
7d0cdf82 | 8 | #include "pack.h" |
eae0bfdc PP |
9 | |
10 | #include "odb.h" | |
6a2d2f8a | 11 | #include "delta.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); |
b644e223 | 21 | static int packfile_unpack_compressed( |
a070f152 CMN |
22 | git_rawobj *obj, |
23 | struct git_pack_file *p, | |
24 | git_mwindow **w_curs, | |
e1de726c | 25 | git_off_t *curpos, |
a070f152 | 26 | size_t size, |
ac3d33df | 27 | git_object_t type); |
a070f152 CMN |
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 | { | |
ac3d33df | 45 | git_error_set(GIT_ERROR_ODB, "invalid pack file - %s", message); |
e1de726c RB |
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 | ||
8588cb0c | 59 | git_atomic_inc(&e->refcount); |
c0f4a011 CMN |
60 | memcpy(&e->raw, source, sizeof(git_rawobj)); |
61 | ||
62 | return e; | |
63 | } | |
64 | ||
65 | static void free_cache_object(void *o) | |
66 | { | |
67 | git_pack_cache_entry *e = (git_pack_cache_entry *)o; | |
68 | ||
69 | if (e != NULL) { | |
0ed75620 | 70 | assert(e->refcount.val == 0); |
c0f4a011 CMN |
71 | git__free(e->raw.data); |
72 | git__free(e); | |
73 | } | |
74 | } | |
75 | ||
c8f79c2b CMN |
76 | static void cache_free(git_pack_cache *cache) |
77 | { | |
9694d9ba | 78 | git_pack_cache_entry *entry; |
c8f79c2b CMN |
79 | |
80 | if (cache->entries) { | |
9694d9ba PS |
81 | git_offmap_foreach_value(cache->entries, entry, { |
82 | free_cache_object(entry); | |
83 | }); | |
c8f79c2b CMN |
84 | |
85 | git_offmap_free(cache->entries); | |
649214be | 86 | cache->entries = NULL; |
c8f79c2b CMN |
87 | } |
88 | } | |
89 | ||
90 | static int cache_init(git_pack_cache *cache) | |
91 | { | |
c8f79c2b | 92 | cache->entries = git_offmap_alloc(); |
ac3d33df | 93 | GIT_ERROR_CHECK_ALLOC(cache->entries); |
f658dc43 | 94 | |
c8f79c2b | 95 | cache->memory_limit = GIT_PACK_CACHE_MEMORY_LIMIT; |
1a42dd17 RB |
96 | |
97 | if (git_mutex_init(&cache->lock)) { | |
ac3d33df | 98 | git_error_set(GIT_ERROR_OS, "failed to initialize pack cache mutex"); |
1a42dd17 RB |
99 | |
100 | git__free(cache->entries); | |
101 | cache->entries = NULL; | |
102 | ||
103 | return -1; | |
104 | } | |
c8f79c2b CMN |
105 | |
106 | return 0; | |
107 | } | |
108 | ||
09e29e47 | 109 | static git_pack_cache_entry *cache_get(git_pack_cache *cache, git_off_t offset) |
c8f79c2b | 110 | { |
c8f79c2b | 111 | git_pack_cache_entry *entry = NULL; |
ac3d33df | 112 | size_t k; |
c8f79c2b | 113 | |
09e29e47 CMN |
114 | if (git_mutex_lock(&cache->lock) < 0) |
115 | return NULL; | |
116 | ||
a853c527 | 117 | k = git_offmap_lookup_index(cache->entries, offset); |
64e46dc3 | 118 | if (git_offmap_valid_index(cache->entries, k)) { /* found it */ |
cb18386f | 119 | entry = git_offmap_value_at(cache->entries, k); |
c8f79c2b | 120 | git_atomic_inc(&entry->refcount); |
0ed75620 | 121 | entry->last_usage = cache->use_ctr++; |
c8f79c2b CMN |
122 | } |
123 | git_mutex_unlock(&cache->lock); | |
124 | ||
125 | return entry; | |
126 | } | |
127 | ||
0ed75620 CMN |
128 | /* Run with the cache lock held */ |
129 | static void free_lowest_entry(git_pack_cache *cache) | |
130 | { | |
685f2251 | 131 | git_off_t offset; |
ed6648ba | 132 | git_pack_cache_entry *entry; |
ed6648ba | 133 | |
685f2251 | 134 | git_offmap_foreach(cache->entries, offset, entry, { |
9c62aaab CMN |
135 | if (entry && entry->refcount.val == 0) { |
136 | cache->memory_used -= entry->raw.len; | |
685f2251 | 137 | git_offmap_delete(cache->entries, offset); |
9c62aaab | 138 | free_cache_object(entry); |
ed6648ba | 139 | } |
9694d9ba | 140 | }); |
0ed75620 CMN |
141 | } |
142 | ||
8588cb0c JH |
143 | static int cache_add( |
144 | git_pack_cache_entry **cached_out, | |
145 | git_pack_cache *cache, | |
146 | git_rawobj *base, | |
147 | git_off_t offset) | |
c8f79c2b CMN |
148 | { |
149 | git_pack_cache_entry *entry; | |
150 | int error, exists = 0; | |
ac3d33df | 151 | size_t k; |
c8f79c2b | 152 | |
0ed75620 CMN |
153 | if (base->len > GIT_PACK_CACHE_SIZE_LIMIT) |
154 | return -1; | |
155 | ||
c8f79c2b CMN |
156 | entry = new_cache_object(base); |
157 | if (entry) { | |
09e29e47 | 158 | if (git_mutex_lock(&cache->lock) < 0) { |
ac3d33df | 159 | git_error_set(GIT_ERROR_OS, "failed to lock cache"); |
6f73e026 | 160 | git__free(entry); |
09e29e47 CMN |
161 | return -1; |
162 | } | |
c8f79c2b | 163 | /* Add it to the cache if nobody else has */ |
036daa59 | 164 | exists = git_offmap_exists(cache->entries, offset); |
c8f79c2b | 165 | if (!exists) { |
0ed75620 CMN |
166 | while (cache->memory_used + base->len > cache->memory_limit) |
167 | free_lowest_entry(cache); | |
168 | ||
f31cb45a | 169 | k = git_offmap_put(cache->entries, offset, &error); |
c8f79c2b | 170 | assert(error != 0); |
85d2748c | 171 | git_offmap_set_value_at(cache->entries, k, entry); |
0ed75620 | 172 | cache->memory_used += entry->raw.len; |
8588cb0c JH |
173 | |
174 | *cached_out = entry; | |
c8f79c2b CMN |
175 | } |
176 | git_mutex_unlock(&cache->lock); | |
177 | /* Somebody beat us to adding it into the cache */ | |
178 | if (exists) { | |
179 | git__free(entry); | |
180 | return -1; | |
181 | } | |
182 | } | |
183 | ||
184 | return 0; | |
185 | } | |
186 | ||
a070f152 CMN |
187 | /*********************************************************** |
188 | * | |
189 | * PACK INDEX METHODS | |
190 | * | |
191 | ***********************************************************/ | |
192 | ||
193 | static void pack_index_free(struct git_pack_file *p) | |
194 | { | |
60ecdf59 DMB |
195 | if (p->oids) { |
196 | git__free(p->oids); | |
197 | p->oids = NULL; | |
198 | } | |
a070f152 CMN |
199 | if (p->index_map.data) { |
200 | git_futils_mmap_free(&p->index_map); | |
201 | p->index_map.data = NULL; | |
202 | } | |
203 | } | |
204 | ||
87d9869f | 205 | static int pack_index_check(const char *path, struct git_pack_file *p) |
a070f152 CMN |
206 | { |
207 | struct git_pack_idx_header *hdr; | |
208 | uint32_t version, nr, i, *index; | |
a070f152 CMN |
209 | void *idx_map; |
210 | size_t idx_size; | |
a070f152 | 211 | struct stat st; |
a070f152 | 212 | int error; |
e1de726c RB |
213 | /* TODO: properly open the file without access time using O_NOATIME */ |
214 | git_file fd = git_futils_open_ro(path); | |
a070f152 | 215 | if (fd < 0) |
e1de726c | 216 | return fd; |
a070f152 | 217 | |
9d2f841a RB |
218 | if (p_fstat(fd, &st) < 0) { |
219 | p_close(fd); | |
ac3d33df | 220 | git_error_set(GIT_ERROR_OS, "unable to stat pack index '%s'", path); |
9d2f841a RB |
221 | return -1; |
222 | } | |
223 | ||
224 | if (!S_ISREG(st.st_mode) || | |
e1de726c RB |
225 | !git__is_sizet(st.st_size) || |
226 | (idx_size = (size_t)st.st_size) < 4 * 256 + 20 + 20) | |
227 | { | |
a070f152 | 228 | p_close(fd); |
ac3d33df | 229 | git_error_set(GIT_ERROR_ODB, "invalid pack index '%s'", path); |
e1de726c | 230 | return -1; |
a070f152 CMN |
231 | } |
232 | ||
233 | error = git_futils_mmap_ro(&p->index_map, fd, 0, idx_size); | |
e1de726c | 234 | |
a070f152 CMN |
235 | p_close(fd); |
236 | ||
e1de726c RB |
237 | if (error < 0) |
238 | return error; | |
a070f152 CMN |
239 | |
240 | hdr = idx_map = p->index_map.data; | |
241 | ||
242 | if (hdr->idx_signature == htonl(PACK_IDX_SIGNATURE)) { | |
243 | version = ntohl(hdr->idx_version); | |
244 | ||
245 | if (version < 2 || version > 2) { | |
246 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 247 | return packfile_error("unsupported index version"); |
a070f152 CMN |
248 | } |
249 | ||
250 | } else | |
251 | version = 1; | |
252 | ||
253 | nr = 0; | |
254 | index = idx_map; | |
255 | ||
256 | if (version > 1) | |
87d9869f | 257 | index += 2; /* skip index header */ |
a070f152 CMN |
258 | |
259 | for (i = 0; i < 256; i++) { | |
260 | uint32_t n = ntohl(index[i]); | |
261 | if (n < nr) { | |
262 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 263 | return packfile_error("index is non-monotonic"); |
a070f152 CMN |
264 | } |
265 | nr = n; | |
266 | } | |
267 | ||
268 | if (version == 1) { | |
269 | /* | |
270 | * Total size: | |
87d9869f VM |
271 | * - 256 index entries 4 bytes each |
272 | * - 24-byte entries * nr (20-byte sha1 + 4-byte offset) | |
273 | * - 20-byte SHA1 of the packfile | |
274 | * - 20-byte SHA1 file checksum | |
a070f152 CMN |
275 | */ |
276 | if (idx_size != 4*256 + nr * 24 + 20 + 20) { | |
277 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 278 | return packfile_error("index is corrupted"); |
a070f152 CMN |
279 | } |
280 | } else if (version == 2) { | |
281 | /* | |
282 | * Minimum size: | |
87d9869f VM |
283 | * - 8 bytes of header |
284 | * - 256 index entries 4 bytes each | |
285 | * - 20-byte sha1 entry * nr | |
286 | * - 4-byte crc entry * nr | |
287 | * - 4-byte offset entry * nr | |
288 | * - 20-byte SHA1 of the packfile | |
289 | * - 20-byte SHA1 file checksum | |
a070f152 CMN |
290 | * And after the 4-byte offset table might be a |
291 | * variable sized table containing 8-byte entries | |
292 | * for offsets larger than 2^31. | |
293 | */ | |
294 | unsigned long min_size = 8 + 4*256 + nr*(20 + 4 + 4) + 20 + 20; | |
295 | unsigned long max_size = min_size; | |
296 | ||
297 | if (nr) | |
298 | max_size += (nr - 1)*8; | |
299 | ||
300 | if (idx_size < min_size || idx_size > max_size) { | |
301 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 302 | return packfile_error("wrong index size"); |
a070f152 | 303 | } |
a070f152 CMN |
304 | } |
305 | ||
a070f152 | 306 | p->num_objects = nr; |
0ddfcb40 | 307 | p->index_version = version; |
e1de726c | 308 | return 0; |
a070f152 CMN |
309 | } |
310 | ||
311 | static int pack_index_open(struct git_pack_file *p) | |
312 | { | |
24c70804 | 313 | int error = 0; |
878293f7 | 314 | size_t name_len; |
a693b873 | 315 | git_buf idx_name; |
24c70804 | 316 | |
0ddfcb40 | 317 | if (p->index_version > -1) |
53607868 | 318 | return 0; |
a070f152 | 319 | |
24c70804 RB |
320 | name_len = strlen(p->pack_name); |
321 | assert(name_len > strlen(".pack")); /* checked by git_pack_file alloc */ | |
e1de726c | 322 | |
a693b873 PS |
323 | if (git_buf_init(&idx_name, name_len) < 0) |
324 | return -1; | |
325 | ||
878293f7 CMN |
326 | git_buf_put(&idx_name, p->pack_name, name_len - strlen(".pack")); |
327 | git_buf_puts(&idx_name, ".idx"); | |
328 | if (git_buf_oom(&idx_name)) { | |
ac3d33df | 329 | git_buf_dispose(&idx_name); |
53607868 | 330 | return -1; |
878293f7 | 331 | } |
a070f152 | 332 | |
050af8bb | 333 | if ((error = git_mutex_lock(&p->lock)) < 0) { |
ac3d33df | 334 | git_buf_dispose(&idx_name); |
53607868 | 335 | return error; |
050af8bb | 336 | } |
53607868 | 337 | |
0ddfcb40 | 338 | if (p->index_version == -1) |
878293f7 | 339 | error = pack_index_check(idx_name.ptr, p); |
24c70804 | 340 | |
ac3d33df | 341 | git_buf_dispose(&idx_name); |
a070f152 | 342 | |
24c70804 RB |
343 | git_mutex_unlock(&p->lock); |
344 | ||
e1de726c | 345 | return error; |
a070f152 CMN |
346 | } |
347 | ||
348 | static unsigned char *pack_window_open( | |
349 | struct git_pack_file *p, | |
7d0cdf82 | 350 | git_mwindow **w_cursor, |
e1de726c | 351 | git_off_t offset, |
7d0cdf82 CMN |
352 | unsigned int *left) |
353 | { | |
e1de726c | 354 | if (p->mwf.fd == -1 && packfile_open(p) < 0) |
7d0cdf82 CMN |
355 | return NULL; |
356 | ||
357 | /* Since packfiles end in a hash of their content and it's | |
358 | * pointless to ask for an offset into the middle of that | |
359 | * hash, and the pack_window_contains function above wouldn't match | |
360 | * don't allow an offset too close to the end of the file. | |
6d97beb9 CMN |
361 | * |
362 | * Don't allow a negative offset, as that means we've wrapped | |
363 | * around. | |
7d0cdf82 CMN |
364 | */ |
365 | if (offset > (p->mwf.size - 20)) | |
366 | return NULL; | |
6d97beb9 CMN |
367 | if (offset < 0) |
368 | return NULL; | |
7d0cdf82 CMN |
369 | |
370 | return git_mwindow_open(&p->mwf, w_cursor, offset, 20, left); | |
371 | } | |
372 | ||
51e82492 CMN |
373 | /* |
374 | * The per-object header is a pretty dense thing, which is | |
375 | * - first byte: low four bits are "size", | |
376 | * then three bits of "type", | |
377 | * with the high bit being "size continues". | |
378 | * - each byte afterwards: low seven bits are size continuation, | |
379 | * with the high bit being "size continues" | |
380 | */ | |
ac3d33df | 381 | size_t git_packfile__object_header(unsigned char *hdr, size_t size, git_object_t type) |
51e82492 CMN |
382 | { |
383 | unsigned char *hdr_base; | |
384 | unsigned char c; | |
385 | ||
ac3d33df | 386 | assert(type >= GIT_OBJECT_COMMIT && type <= GIT_OBJECT_REF_DELTA); |
51e82492 CMN |
387 | |
388 | /* TODO: add support for chunked objects; see git.git 6c0d19b1 */ | |
389 | ||
390 | c = (unsigned char)((type << 4) | (size & 15)); | |
391 | size >>= 4; | |
392 | hdr_base = hdr; | |
393 | ||
394 | while (size) { | |
395 | *hdr++ = c | 0x80; | |
396 | c = size & 0x7f; | |
397 | size >>= 7; | |
398 | } | |
399 | *hdr++ = c; | |
400 | ||
51a3dfb5 | 401 | return (hdr - hdr_base); |
51e82492 CMN |
402 | } |
403 | ||
404 | ||
45d773ef CMN |
405 | static int packfile_unpack_header1( |
406 | unsigned long *usedp, | |
7d0cdf82 | 407 | size_t *sizep, |
ac3d33df | 408 | git_object_t *type, |
7d0cdf82 CMN |
409 | const unsigned char *buf, |
410 | unsigned long len) | |
411 | { | |
412 | unsigned shift; | |
413 | unsigned long size, c; | |
414 | unsigned long used = 0; | |
2aeadb9c | 415 | |
7d0cdf82 CMN |
416 | c = buf[used++]; |
417 | *type = (c >> 4) & 7; | |
418 | size = c & 15; | |
419 | shift = 4; | |
420 | while (c & 0x80) { | |
ec7e680c | 421 | if (len <= used) { |
ac3d33df | 422 | git_error_set(GIT_ERROR_ODB, "buffer too small"); |
904b67e6 | 423 | return GIT_EBUFS; |
ec7e680c | 424 | } |
45d773ef CMN |
425 | |
426 | if (bitsizeof(long) <= shift) { | |
427 | *usedp = 0; | |
ac3d33df | 428 | git_error_set(GIT_ERROR_ODB, "packfile corrupted"); |
45d773ef CMN |
429 | return -1; |
430 | } | |
7d0cdf82 CMN |
431 | |
432 | c = buf[used++]; | |
433 | size += (c & 0x7f) << shift; | |
434 | shift += 7; | |
435 | } | |
436 | ||
437 | *sizep = (size_t)size; | |
45d773ef CMN |
438 | *usedp = used; |
439 | return 0; | |
7d0cdf82 CMN |
440 | } |
441 | ||
442 | int git_packfile_unpack_header( | |
443 | size_t *size_p, | |
ac3d33df | 444 | git_object_t *type_p, |
7d0cdf82 CMN |
445 | git_mwindow_file *mwf, |
446 | git_mwindow **w_curs, | |
e1de726c | 447 | git_off_t *curpos) |
7d0cdf82 CMN |
448 | { |
449 | unsigned char *base; | |
450 | unsigned int left; | |
451 | unsigned long used; | |
45d773ef | 452 | int ret; |
7d0cdf82 CMN |
453 | |
454 | /* pack_window_open() assures us we have [base, base + 20) available | |
87d9869f VM |
455 | * as a range that we can look at at. (Its actually the hash |
456 | * size that is assured.) With our object header encoding | |
7d0cdf82 CMN |
457 | * the maximum deflated object size is 2^137, which is just |
458 | * insane, so we know won't exceed what we have been given. | |
459 | */ | |
24c70804 | 460 | /* base = pack_window_open(p, w_curs, *curpos, &left); */ |
7d0cdf82 CMN |
461 | base = git_mwindow_open(mwf, w_curs, *curpos, 20, &left); |
462 | if (base == NULL) | |
904b67e6 | 463 | return GIT_EBUFS; |
2aeadb9c | 464 | |
9d2f841a | 465 | ret = packfile_unpack_header1(&used, size_p, type_p, base, left); |
45d773ef | 466 | git_mwindow_close(w_curs); |
904b67e6 | 467 | if (ret == GIT_EBUFS) |
45d773ef CMN |
468 | return ret; |
469 | else if (ret < 0) | |
e1de726c | 470 | return packfile_error("header length is zero"); |
7d0cdf82 CMN |
471 | |
472 | *curpos += used; | |
e1de726c | 473 | return 0; |
7d0cdf82 CMN |
474 | } |
475 | ||
44f9f547 DMB |
476 | int git_packfile_resolve_header( |
477 | size_t *size_p, | |
ac3d33df | 478 | git_object_t *type_p, |
44f9f547 DMB |
479 | struct git_pack_file *p, |
480 | git_off_t offset) | |
481 | { | |
482 | git_mwindow *w_curs = NULL; | |
483 | git_off_t curpos = offset; | |
484 | size_t size; | |
ac3d33df | 485 | git_object_t type; |
44f9f547 DMB |
486 | git_off_t base_offset; |
487 | int error; | |
488 | ||
489 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
44f9f547 DMB |
490 | if (error < 0) |
491 | return error; | |
492 | ||
ac3d33df | 493 | if (type == GIT_OBJECT_OFS_DELTA || type == GIT_OBJECT_REF_DELTA) { |
44f9f547 | 494 | size_t base_size; |
a97b769a CMN |
495 | git_packfile_stream stream; |
496 | ||
44f9f547 DMB |
497 | base_offset = get_delta_base(p, &w_curs, &curpos, type, offset); |
498 | git_mwindow_close(&w_curs); | |
a97b769a | 499 | if ((error = git_packfile_stream_open(&stream, p, curpos)) < 0) |
44f9f547 | 500 | return error; |
6a2d2f8a | 501 | error = git_delta_read_header_fromstream(&base_size, size_p, &stream); |
ac3d33df | 502 | git_packfile_stream_dispose(&stream); |
44f9f547 DMB |
503 | if (error < 0) |
504 | return error; | |
34b32053 | 505 | } else { |
44f9f547 | 506 | *size_p = size; |
34b32053 PS |
507 | base_offset = 0; |
508 | } | |
44f9f547 | 509 | |
ac3d33df | 510 | while (type == GIT_OBJECT_OFS_DELTA || type == GIT_OBJECT_REF_DELTA) { |
44f9f547 DMB |
511 | curpos = base_offset; |
512 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
44f9f547 DMB |
513 | if (error < 0) |
514 | return error; | |
ac3d33df | 515 | if (type != GIT_OBJECT_OFS_DELTA && type != GIT_OBJECT_REF_DELTA) |
44f9f547 DMB |
516 | break; |
517 | base_offset = get_delta_base(p, &w_curs, &curpos, type, base_offset); | |
518 | git_mwindow_close(&w_curs); | |
519 | } | |
520 | *type_p = type; | |
521 | ||
522 | return error; | |
523 | } | |
524 | ||
15bcced2 CMN |
525 | #define SMALL_STACK_SIZE 64 |
526 | ||
a3ffbf23 CMN |
527 | /** |
528 | * Generate the chain of dependencies which we need to get to the | |
529 | * object at `off`. `chain` is used a stack, popping gives the right | |
530 | * order to apply deltas on. If an object is found in the pack's base | |
531 | * cache, we stop calculating there. | |
532 | */ | |
15bcced2 CMN |
533 | static int pack_dependency_chain(git_dependency_chain *chain_out, |
534 | git_pack_cache_entry **cached_out, git_off_t *cached_off, | |
535 | struct pack_chain_elem *small_stack, size_t *stack_sz, | |
536 | struct git_pack_file *p, git_off_t obj_offset) | |
a3ffbf23 CMN |
537 | { |
538 | git_dependency_chain chain = GIT_ARRAY_INIT; | |
539 | git_mwindow *w_curs = NULL; | |
540 | git_off_t curpos = obj_offset, base_offset; | |
15bcced2 CMN |
541 | int error = 0, use_heap = 0; |
542 | size_t size, elem_pos; | |
ac3d33df | 543 | git_object_t type; |
a3ffbf23 | 544 | |
15bcced2 | 545 | elem_pos = 0; |
a3ffbf23 CMN |
546 | while (true) { |
547 | struct pack_chain_elem *elem; | |
548 | git_pack_cache_entry *cached = NULL; | |
549 | ||
550 | /* if we have a base cached, we can stop here instead */ | |
551 | if ((cached = cache_get(&p->bases, obj_offset)) != NULL) { | |
552 | *cached_out = cached; | |
553 | *cached_off = obj_offset; | |
554 | break; | |
555 | } | |
556 | ||
15bcced2 CMN |
557 | /* if we run out of space on the small stack, use the array */ |
558 | if (elem_pos == SMALL_STACK_SIZE) { | |
559 | git_array_init_to_size(chain, elem_pos); | |
ac3d33df | 560 | GIT_ERROR_CHECK_ARRAY(chain); |
15bcced2 CMN |
561 | memcpy(chain.ptr, small_stack, elem_pos * sizeof(struct pack_chain_elem)); |
562 | chain.size = elem_pos; | |
563 | use_heap = 1; | |
564 | } | |
565 | ||
a3ffbf23 | 566 | curpos = obj_offset; |
15bcced2 CMN |
567 | if (!use_heap) { |
568 | elem = &small_stack[elem_pos]; | |
569 | } else { | |
570 | elem = git_array_alloc(chain); | |
571 | if (!elem) { | |
572 | error = -1; | |
573 | goto on_error; | |
574 | } | |
a3ffbf23 CMN |
575 | } |
576 | ||
577 | elem->base_key = obj_offset; | |
578 | ||
579 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
a3ffbf23 CMN |
580 | |
581 | if (error < 0) | |
582 | goto on_error; | |
583 | ||
584 | elem->offset = curpos; | |
585 | elem->size = size; | |
586 | elem->type = type; | |
587 | elem->base_key = obj_offset; | |
588 | ||
ac3d33df | 589 | if (type != GIT_OBJECT_OFS_DELTA && type != GIT_OBJECT_REF_DELTA) |
a3ffbf23 CMN |
590 | break; |
591 | ||
592 | base_offset = get_delta_base(p, &w_curs, &curpos, type, obj_offset); | |
593 | git_mwindow_close(&w_curs); | |
594 | ||
595 | if (base_offset == 0) { | |
596 | error = packfile_error("delta offset is zero"); | |
597 | goto on_error; | |
598 | } | |
599 | if (base_offset < 0) { /* must actually be an error code */ | |
600 | error = (int)base_offset; | |
601 | goto on_error; | |
602 | } | |
603 | ||
604 | /* we need to pass the pos *after* the delta-base bit */ | |
605 | elem->offset = curpos; | |
606 | ||
607 | /* go through the loop again, but with the new object */ | |
608 | obj_offset = base_offset; | |
15bcced2 | 609 | elem_pos++; |
a3ffbf23 CMN |
610 | } |
611 | ||
ac3d33df | 612 | |
15bcced2 | 613 | *stack_sz = elem_pos + 1; |
a3ffbf23 CMN |
614 | *chain_out = chain; |
615 | return error; | |
616 | ||
617 | on_error: | |
618 | git_array_clear(chain); | |
619 | return error; | |
620 | } | |
621 | ||
a070f152 | 622 | int git_packfile_unpack( |
e1de726c RB |
623 | git_rawobj *obj, |
624 | struct git_pack_file *p, | |
625 | git_off_t *obj_offset) | |
7d0cdf82 CMN |
626 | { |
627 | git_mwindow *w_curs = NULL; | |
e1de726c | 628 | git_off_t curpos = *obj_offset; |
a332e91c CMN |
629 | int error, free_base = 0; |
630 | git_dependency_chain chain = GIT_ARRAY_INIT; | |
15bcced2 | 631 | struct pack_chain_elem *elem = NULL, *stack; |
a332e91c | 632 | git_pack_cache_entry *cached = NULL; |
15bcced2 | 633 | struct pack_chain_elem small_stack[SMALL_STACK_SIZE]; |
f1453c59 | 634 | size_t stack_size = 0, elem_pos, alloclen; |
ac3d33df | 635 | git_object_t base_type; |
7d0cdf82 CMN |
636 | |
637 | /* | |
638 | * TODO: optionally check the CRC on the packfile | |
639 | */ | |
640 | ||
15bcced2 | 641 | error = pack_dependency_chain(&chain, &cached, obj_offset, small_stack, &stack_size, p, *obj_offset); |
2acdf4b8 CMN |
642 | if (error < 0) |
643 | return error; | |
644 | ||
7d0cdf82 CMN |
645 | obj->data = NULL; |
646 | obj->len = 0; | |
ac3d33df | 647 | obj->type = GIT_OBJECT_INVALID; |
7d0cdf82 | 648 | |
15bcced2 CMN |
649 | /* let's point to the right stack */ |
650 | stack = chain.ptr ? chain.ptr : small_stack; | |
651 | ||
652 | elem_pos = stack_size; | |
a3ffbf23 | 653 | if (cached) { |
a332e91c | 654 | memcpy(obj, &cached->raw, sizeof(git_rawobj)); |
a3ffbf23 | 655 | base_type = obj->type; |
15bcced2 | 656 | elem_pos--; /* stack_size includes the base, which isn't actually there */ |
a3ffbf23 | 657 | } else { |
15bcced2 | 658 | elem = &stack[--elem_pos]; |
a3ffbf23 | 659 | base_type = elem->type; |
a332e91c | 660 | } |
45d773ef | 661 | |
9dbd150f | 662 | switch (base_type) { |
ac3d33df JK |
663 | case GIT_OBJECT_COMMIT: |
664 | case GIT_OBJECT_TREE: | |
665 | case GIT_OBJECT_BLOB: | |
666 | case GIT_OBJECT_TAG: | |
a3ffbf23 | 667 | if (!cached) { |
9dbd150f CMN |
668 | curpos = elem->offset; |
669 | error = packfile_unpack_compressed(obj, p, &w_curs, &curpos, elem->size, elem->type); | |
670 | git_mwindow_close(&w_curs); | |
a3ffbf23 | 671 | base_type = elem->type; |
9dbd150f CMN |
672 | } |
673 | if (error < 0) | |
674 | goto cleanup; | |
675 | break; | |
ac3d33df JK |
676 | case GIT_OBJECT_OFS_DELTA: |
677 | case GIT_OBJECT_REF_DELTA: | |
9dbd150f CMN |
678 | error = packfile_error("dependency chain ends in a delta"); |
679 | goto cleanup; | |
680 | default: | |
681 | error = packfile_error("invalid packfile type in header"); | |
682 | goto cleanup; | |
683 | } | |
684 | ||
a332e91c | 685 | /* |
c968ce2c | 686 | * Finding the object we want a cached base element is |
a332e91c CMN |
687 | * problematic, as we need to make sure we don't accidentally |
688 | * give the caller the cached object, which it would then feel | |
689 | * free to free, so we need to copy the data. | |
690 | */ | |
15bcced2 | 691 | if (cached && stack_size == 1) { |
a332e91c | 692 | void *data = obj->data; |
392702ee | 693 | |
ac3d33df | 694 | GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, obj->len, 1); |
f1453c59 | 695 | obj->data = git__malloc(alloclen); |
ac3d33df | 696 | GIT_ERROR_CHECK_ALLOC(obj->data); |
392702ee | 697 | |
a332e91c CMN |
698 | memcpy(obj->data, data, obj->len + 1); |
699 | git_atomic_dec(&cached->refcount); | |
700 | goto cleanup; | |
701 | } | |
702 | ||
2acdf4b8 | 703 | /* we now apply each consecutive delta until we run out */ |
15bcced2 | 704 | while (elem_pos > 0 && !error) { |
2acdf4b8 CMN |
705 | git_rawobj base, delta; |
706 | ||
c968ce2c CMN |
707 | /* |
708 | * We can now try to add the base to the cache, as | |
709 | * long as it's not already the cached one. | |
710 | */ | |
711 | if (!cached) | |
8588cb0c | 712 | free_base = !!cache_add(&cached, &p->bases, obj, elem->base_key); |
c968ce2c | 713 | |
15bcced2 | 714 | elem = &stack[elem_pos - 1]; |
2acdf4b8 CMN |
715 | curpos = elem->offset; |
716 | error = packfile_unpack_compressed(&delta, p, &w_curs, &curpos, elem->size, elem->type); | |
717 | git_mwindow_close(&w_curs); | |
718 | ||
eae0bfdc PP |
719 | if (error < 0) { |
720 | /* We have transferred ownership of the data to the cache. */ | |
721 | obj->data = NULL; | |
2acdf4b8 | 722 | break; |
eae0bfdc | 723 | } |
2acdf4b8 CMN |
724 | |
725 | /* the current object becomes the new base, on which we apply the delta */ | |
726 | base = *obj; | |
727 | obj->data = NULL; | |
728 | obj->len = 0; | |
ac3d33df | 729 | obj->type = GIT_OBJECT_INVALID; |
2acdf4b8 | 730 | |
6a2d2f8a | 731 | error = git_delta_apply(&obj->data, &obj->len, base.data, base.len, delta.data, delta.len); |
a332e91c | 732 | obj->type = base_type; |
6a2d2f8a | 733 | |
a332e91c CMN |
734 | /* |
735 | * We usually don't want to free the base at this | |
736 | * point, as we put it into the cache in the previous | |
737 | * iteration. free_base lets us know that we got the | |
738 | * base object directly from the packfile, so we can free it. | |
739 | */ | |
2acdf4b8 | 740 | git__free(delta.data); |
a332e91c CMN |
741 | if (free_base) { |
742 | free_base = 0; | |
743 | git__free(base.data); | |
744 | } | |
745 | ||
746 | if (cached) { | |
747 | git_atomic_dec(&cached->refcount); | |
748 | cached = NULL; | |
749 | } | |
2acdf4b8 CMN |
750 | |
751 | if (error < 0) | |
752 | break; | |
7d0cdf82 | 753 | |
15bcced2 | 754 | elem_pos--; |
7d0cdf82 CMN |
755 | } |
756 | ||
2acdf4b8 | 757 | cleanup: |
ff5eea06 | 758 | if (error < 0) { |
a332e91c | 759 | git__free(obj->data); |
ff5eea06 PS |
760 | if (cached) |
761 | git_atomic_dec(&cached->refcount); | |
762 | } | |
a332e91c | 763 | |
a3ffbf23 | 764 | if (elem) |
b3d3459f | 765 | *obj_offset = curpos; |
a332e91c | 766 | |
2acdf4b8 | 767 | git_array_clear(chain); |
e1de726c | 768 | return error; |
7d0cdf82 CMN |
769 | } |
770 | ||
282283ac RB |
771 | static void *use_git_alloc(void *opaq, unsigned int count, unsigned int size) |
772 | { | |
773 | GIT_UNUSED(opaq); | |
774 | return git__calloc(count, size); | |
775 | } | |
776 | ||
777 | static void use_git_free(void *opaq, void *ptr) | |
778 | { | |
779 | GIT_UNUSED(opaq); | |
780 | git__free(ptr); | |
781 | } | |
782 | ||
46635339 CMN |
783 | int git_packfile_stream_open(git_packfile_stream *obj, struct git_pack_file *p, git_off_t curpos) |
784 | { | |
785 | int st; | |
786 | ||
787 | memset(obj, 0, sizeof(git_packfile_stream)); | |
788 | obj->curpos = curpos; | |
789 | obj->p = p; | |
790 | obj->zstream.zalloc = use_git_alloc; | |
791 | obj->zstream.zfree = use_git_free; | |
792 | obj->zstream.next_in = Z_NULL; | |
793 | obj->zstream.next_out = Z_NULL; | |
794 | st = inflateInit(&obj->zstream); | |
795 | if (st != Z_OK) { | |
ac3d33df | 796 | git_error_set(GIT_ERROR_ZLIB, "failed to init packfile stream"); |
46635339 CMN |
797 | return -1; |
798 | } | |
799 | ||
800 | return 0; | |
801 | } | |
802 | ||
803 | ssize_t git_packfile_stream_read(git_packfile_stream *obj, void *buffer, size_t len) | |
804 | { | |
805 | unsigned char *in; | |
806 | size_t written; | |
807 | int st; | |
808 | ||
809 | if (obj->done) | |
810 | return 0; | |
811 | ||
812 | in = pack_window_open(obj->p, &obj->mw, obj->curpos, &obj->zstream.avail_in); | |
813 | if (in == NULL) | |
814 | return GIT_EBUFS; | |
815 | ||
816 | obj->zstream.next_out = buffer; | |
090d5e1f | 817 | obj->zstream.avail_out = (unsigned int)len; |
46635339 CMN |
818 | obj->zstream.next_in = in; |
819 | ||
820 | st = inflate(&obj->zstream, Z_SYNC_FLUSH); | |
821 | git_mwindow_close(&obj->mw); | |
822 | ||
823 | obj->curpos += obj->zstream.next_in - in; | |
824 | written = len - obj->zstream.avail_out; | |
825 | ||
826 | if (st != Z_OK && st != Z_STREAM_END) { | |
ac3d33df | 827 | git_error_set(GIT_ERROR_ZLIB, "error reading from the zlib stream"); |
46635339 CMN |
828 | return -1; |
829 | } | |
830 | ||
831 | if (st == Z_STREAM_END) | |
832 | obj->done = 1; | |
833 | ||
834 | ||
835 | /* If we didn't write anything out but we're not done, we need more data */ | |
836 | if (!written && st != Z_STREAM_END) | |
837 | return GIT_EBUFS; | |
838 | ||
839 | return written; | |
840 | ||
841 | } | |
842 | ||
ac3d33df | 843 | void git_packfile_stream_dispose(git_packfile_stream *obj) |
46635339 CMN |
844 | { |
845 | inflateEnd(&obj->zstream); | |
846 | } | |
847 | ||
b644e223 | 848 | static int packfile_unpack_compressed( |
e1de726c RB |
849 | git_rawobj *obj, |
850 | struct git_pack_file *p, | |
851 | git_mwindow **w_curs, | |
852 | git_off_t *curpos, | |
853 | size_t size, | |
ac3d33df | 854 | git_object_t type) |
7d0cdf82 | 855 | { |
f1453c59 | 856 | size_t buf_size; |
7d0cdf82 CMN |
857 | int st; |
858 | z_stream stream; | |
859 | unsigned char *buffer, *in; | |
860 | ||
ac3d33df | 861 | GIT_ERROR_CHECK_ALLOC_ADD(&buf_size, size, 1); |
f1453c59 | 862 | buffer = git__calloc(1, buf_size); |
ac3d33df | 863 | GIT_ERROR_CHECK_ALLOC(buffer); |
7d0cdf82 CMN |
864 | |
865 | memset(&stream, 0, sizeof(stream)); | |
866 | stream.next_out = buffer; | |
f1453c59 | 867 | stream.avail_out = (uInt)buf_size; |
282283ac RB |
868 | stream.zalloc = use_git_alloc; |
869 | stream.zfree = use_git_free; | |
7d0cdf82 CMN |
870 | |
871 | st = inflateInit(&stream); | |
872 | if (st != Z_OK) { | |
3286c408 | 873 | git__free(buffer); |
ac3d33df | 874 | git_error_set(GIT_ERROR_ZLIB, "failed to init zlib stream on unpack"); |
45d773ef | 875 | |
e1de726c | 876 | return -1; |
7d0cdf82 CMN |
877 | } |
878 | ||
879 | do { | |
b5b474dd | 880 | in = pack_window_open(p, w_curs, *curpos, &stream.avail_in); |
7d0cdf82 CMN |
881 | stream.next_in = in; |
882 | st = inflate(&stream, Z_FINISH); | |
45d773ef | 883 | git_mwindow_close(w_curs); |
7d0cdf82 CMN |
884 | |
885 | if (!stream.avail_out) | |
886 | break; /* the payload is larger than it should be */ | |
887 | ||
45d773ef CMN |
888 | if (st == Z_BUF_ERROR && in == NULL) { |
889 | inflateEnd(&stream); | |
890 | git__free(buffer); | |
904b67e6 | 891 | return GIT_EBUFS; |
45d773ef CMN |
892 | } |
893 | ||
b5b474dd | 894 | *curpos += stream.next_in - in; |
7d0cdf82 CMN |
895 | } while (st == Z_OK || st == Z_BUF_ERROR); |
896 | ||
897 | inflateEnd(&stream); | |
898 | ||
899 | if ((st != Z_STREAM_END) || stream.total_out != size) { | |
3286c408 | 900 | git__free(buffer); |
ac3d33df | 901 | git_error_set(GIT_ERROR_ZLIB, "error inflating zlib stream"); |
e1de726c | 902 | return -1; |
7d0cdf82 CMN |
903 | } |
904 | ||
905 | obj->type = type; | |
906 | obj->len = size; | |
907 | obj->data = buffer; | |
e1de726c | 908 | return 0; |
7d0cdf82 CMN |
909 | } |
910 | ||
b5b474dd CMN |
911 | /* |
912 | * curpos is where the data starts, delta_obj_offset is the where the | |
913 | * header starts | |
914 | */ | |
e1de726c RB |
915 | git_off_t get_delta_base( |
916 | struct git_pack_file *p, | |
917 | git_mwindow **w_curs, | |
918 | git_off_t *curpos, | |
ac3d33df | 919 | git_object_t type, |
e1de726c | 920 | git_off_t delta_obj_offset) |
7d0cdf82 | 921 | { |
45d773ef CMN |
922 | unsigned int left = 0; |
923 | unsigned char *base_info; | |
e1de726c | 924 | git_off_t base_offset; |
7d0cdf82 CMN |
925 | git_oid unused; |
926 | ||
45d773ef CMN |
927 | base_info = pack_window_open(p, w_curs, *curpos, &left); |
928 | /* Assumption: the only reason this would fail is because the file is too small */ | |
929 | if (base_info == NULL) | |
904b67e6 | 930 | return GIT_EBUFS; |
7d0cdf82 CMN |
931 | /* pack_window_open() assured us we have [base_info, base_info + 20) |
932 | * as a range that we can look at without walking off the | |
87d9869f VM |
933 | * end of the mapped window. Its actually the hash size |
934 | * that is assured. An OFS_DELTA longer than the hash size | |
7d0cdf82 CMN |
935 | * is stupid, as then a REF_DELTA would be smaller to store. |
936 | */ | |
ac3d33df | 937 | if (type == GIT_OBJECT_OFS_DELTA) { |
7d0cdf82 CMN |
938 | unsigned used = 0; |
939 | unsigned char c = base_info[used++]; | |
eae0bfdc | 940 | size_t unsigned_base_offset = c & 127; |
7d0cdf82 | 941 | while (c & 128) { |
45d773ef | 942 | if (left <= used) |
904b67e6 | 943 | return GIT_EBUFS; |
eae0bfdc PP |
944 | unsigned_base_offset += 1; |
945 | if (!unsigned_base_offset || MSB(unsigned_base_offset, 7)) | |
87d9869f | 946 | return 0; /* overflow */ |
7d0cdf82 | 947 | c = base_info[used++]; |
eae0bfdc | 948 | unsigned_base_offset = (unsigned_base_offset << 7) + (c & 127); |
7d0cdf82 | 949 | } |
eae0bfdc | 950 | if (unsigned_base_offset == 0 || (size_t)delta_obj_offset <= unsigned_base_offset) |
87d9869f | 951 | return 0; /* out of bound */ |
eae0bfdc | 952 | base_offset = delta_obj_offset - unsigned_base_offset; |
7d0cdf82 | 953 | *curpos += used; |
ac3d33df | 954 | } else if (type == GIT_OBJECT_REF_DELTA) { |
c1af5a39 CMN |
955 | /* If we have the cooperative cache, search in it first */ |
956 | if (p->has_cache) { | |
0e040c03 | 957 | git_oid oid; |
ac3d33df | 958 | size_t k; |
c1af5a39 | 959 | |
0e040c03 | 960 | git_oid_fromraw(&oid, base_info); |
a853c527 | 961 | k = git_oidmap_lookup_index(p->idx_cache, &oid); |
64e46dc3 | 962 | if (git_oidmap_valid_index(p->idx_cache, k)) { |
c1af5a39 | 963 | *curpos += 20; |
cb18386f | 964 | return ((struct git_pack_entry *)git_oidmap_value_at(p->idx_cache, k))->offset; |
38c10ecd ET |
965 | } else { |
966 | /* If we're building an index, don't try to find the pack | |
967 | * entry; we just haven't seen it yet. We'll make | |
968 | * progress again in the next loop. | |
969 | */ | |
970 | return GIT_PASSTHROUGH; | |
c1af5a39 CMN |
971 | } |
972 | } | |
38c10ecd | 973 | |
7d0cdf82 | 974 | /* The base entry _must_ be in the same pack */ |
e1de726c RB |
975 | if (pack_entry_find_offset(&base_offset, &unused, p, (git_oid *)base_info, GIT_OID_HEXSZ) < 0) |
976 | return packfile_error("base entry delta is not in the same pack"); | |
7d0cdf82 CMN |
977 | *curpos += 20; |
978 | } else | |
979 | return 0; | |
980 | ||
981 | return base_offset; | |
982 | } | |
a070f152 CMN |
983 | |
984 | /*********************************************************** | |
985 | * | |
986 | * PACKFILE METHODS | |
987 | * | |
988 | ***********************************************************/ | |
989 | ||
bf339ab0 ET |
990 | void git_packfile_close(struct git_pack_file *p, bool unlink_packfile) |
991 | { | |
992 | if (p->mwf.fd >= 0) { | |
993 | git_mwindow_free_all_locked(&p->mwf); | |
994 | p_close(p->mwf.fd); | |
995 | p->mwf.fd = -1; | |
996 | } | |
997 | ||
998 | if (unlink_packfile) | |
999 | p_unlink(p->pack_name); | |
1000 | } | |
1001 | ||
96c9b9f0 | 1002 | void git_packfile_free(struct git_pack_file *p) |
a070f152 | 1003 | { |
24c70804 RB |
1004 | if (!p) |
1005 | return; | |
1006 | ||
c8f79c2b | 1007 | cache_free(&p->bases); |
c0f4a011 | 1008 | |
bf339ab0 | 1009 | git_packfile_close(p, false); |
a070f152 CMN |
1010 | |
1011 | pack_index_free(p); | |
1012 | ||
3286c408 | 1013 | git__free(p->bad_object_sha1); |
24c70804 | 1014 | |
24c70804 | 1015 | git_mutex_free(&p->lock); |
649214be | 1016 | git_mutex_free(&p->bases.lock); |
3286c408 | 1017 | git__free(p); |
a070f152 CMN |
1018 | } |
1019 | ||
1020 | static int packfile_open(struct git_pack_file *p) | |
1021 | { | |
1022 | struct stat st; | |
1023 | struct git_pack_header hdr; | |
1024 | git_oid sha1; | |
1025 | unsigned char *idx_sha1; | |
1026 | ||
0ddfcb40 | 1027 | if (p->index_version == -1 && pack_index_open(p) < 0) |
e10144ae | 1028 | return git_odb__error_notfound("failed to open packfile", NULL, 0); |
a070f152 | 1029 | |
9d2f841a RB |
1030 | /* if mwf opened by another thread, return now */ |
1031 | if (git_mutex_lock(&p->lock) < 0) | |
1032 | return packfile_error("failed to get lock for open"); | |
1033 | ||
1034 | if (p->mwf.fd >= 0) { | |
1035 | git_mutex_unlock(&p->lock); | |
1036 | return 0; | |
1037 | } | |
1038 | ||
a070f152 | 1039 | /* TODO: open with noatime */ |
e1de726c | 1040 | p->mwf.fd = git_futils_open_ro(p->pack_name); |
9d2f841a RB |
1041 | if (p->mwf.fd < 0) |
1042 | goto cleanup; | |
a070f152 | 1043 | |
e1de726c RB |
1044 | if (p_fstat(p->mwf.fd, &st) < 0 || |
1045 | git_mwindow_file_register(&p->mwf) < 0) | |
1046 | goto cleanup; | |
a070f152 CMN |
1047 | |
1048 | /* If we created the struct before we had the pack we lack size. */ | |
1049 | if (!p->mwf.size) { | |
1050 | if (!S_ISREG(st.st_mode)) | |
1051 | goto cleanup; | |
e1de726c | 1052 | p->mwf.size = (git_off_t)st.st_size; |
a070f152 CMN |
1053 | } else if (p->mwf.size != st.st_size) |
1054 | goto cleanup; | |
1055 | ||
1056 | #if 0 | |
1057 | /* We leave these file descriptors open with sliding mmap; | |
1058 | * there is no point keeping them open across exec(), though. | |
1059 | */ | |
1060 | fd_flag = fcntl(p->mwf.fd, F_GETFD, 0); | |
1061 | if (fd_flag < 0) | |
e1de726c | 1062 | goto cleanup; |
a070f152 CMN |
1063 | |
1064 | fd_flag |= FD_CLOEXEC; | |
1065 | if (fcntl(p->pack_fd, F_SETFD, fd_flag) == -1) | |
e1de726c | 1066 | goto cleanup; |
a070f152 CMN |
1067 | #endif |
1068 | ||
1069 | /* Verify we recognize this pack file format. */ | |
e1de726c RB |
1070 | if (p_read(p->mwf.fd, &hdr, sizeof(hdr)) < 0 || |
1071 | hdr.hdr_signature != htonl(PACK_SIGNATURE) || | |
1072 | !pack_version_ok(hdr.hdr_version)) | |
a070f152 CMN |
1073 | goto cleanup; |
1074 | ||
1075 | /* Verify the pack matches its index. */ | |
e1de726c RB |
1076 | if (p->num_objects != ntohl(hdr.hdr_entries) || |
1077 | p_lseek(p->mwf.fd, p->mwf.size - GIT_OID_RAWSZ, SEEK_SET) == -1 || | |
1078 | p_read(p->mwf.fd, sha1.id, GIT_OID_RAWSZ) < 0) | |
a070f152 CMN |
1079 | goto cleanup; |
1080 | ||
1081 | idx_sha1 = ((unsigned char *)p->index_map.data) + p->index_map.len - 40; | |
1082 | ||
9d2f841a RB |
1083 | if (git_oid__cmp(&sha1, (git_oid *)idx_sha1) != 0) |
1084 | goto cleanup; | |
1085 | ||
1086 | git_mutex_unlock(&p->lock); | |
1087 | return 0; | |
a070f152 CMN |
1088 | |
1089 | cleanup: | |
ac3d33df | 1090 | git_error_set(GIT_ERROR_OS, "invalid packfile '%s'", p->pack_name); |
9d2f841a | 1091 | |
3a2d48d5 SS |
1092 | if (p->mwf.fd >= 0) |
1093 | p_close(p->mwf.fd); | |
a070f152 | 1094 | p->mwf.fd = -1; |
9d2f841a RB |
1095 | |
1096 | git_mutex_unlock(&p->lock); | |
1097 | ||
e1de726c | 1098 | return -1; |
a070f152 CMN |
1099 | } |
1100 | ||
b3b66c57 CMN |
1101 | int git_packfile__name(char **out, const char *path) |
1102 | { | |
1103 | size_t path_len; | |
1104 | git_buf buf = GIT_BUF_INIT; | |
1105 | ||
1106 | path_len = strlen(path); | |
1107 | ||
1108 | if (path_len < strlen(".idx")) | |
e10144ae | 1109 | return git_odb__error_notfound("invalid packfile path", NULL, 0); |
b3b66c57 CMN |
1110 | |
1111 | if (git_buf_printf(&buf, "%.*s.pack", (int)(path_len - strlen(".idx")), path) < 0) | |
1112 | return -1; | |
1113 | ||
1114 | *out = git_buf_detach(&buf); | |
1115 | return 0; | |
1116 | } | |
1117 | ||
5d2d21e5 | 1118 | int git_packfile_alloc(struct git_pack_file **pack_out, const char *path) |
a070f152 CMN |
1119 | { |
1120 | struct stat st; | |
1121 | struct git_pack_file *p; | |
f1453c59 | 1122 | size_t path_len = path ? strlen(path) : 0, alloc_len; |
a070f152 CMN |
1123 | |
1124 | *pack_out = NULL; | |
24c70804 | 1125 | |
38eef611 | 1126 | if (path_len < strlen(".idx")) |
e10144ae | 1127 | return git_odb__error_notfound("invalid packfile path", NULL, 0); |
24c70804 | 1128 | |
ac3d33df JK |
1129 | GIT_ERROR_CHECK_ALLOC_ADD(&alloc_len, sizeof(*p), path_len); |
1130 | GIT_ERROR_CHECK_ALLOC_ADD(&alloc_len, alloc_len, 2); | |
392702ee | 1131 | |
f1453c59 | 1132 | p = git__calloc(1, alloc_len); |
ac3d33df | 1133 | GIT_ERROR_CHECK_ALLOC(p); |
a070f152 | 1134 | |
38eef611 RB |
1135 | memcpy(p->pack_name, path, path_len + 1); |
1136 | ||
a070f152 CMN |
1137 | /* |
1138 | * Make sure a corresponding .pack file exists and that | |
1139 | * the index looks sane. | |
1140 | */ | |
38eef611 RB |
1141 | if (git__suffixcmp(path, ".idx") == 0) { |
1142 | size_t root_len = path_len - strlen(".idx"); | |
1143 | ||
1144 | memcpy(p->pack_name + root_len, ".keep", sizeof(".keep")); | |
1145 | if (git_path_exists(p->pack_name) == true) | |
1146 | p->pack_keep = 1; | |
a070f152 | 1147 | |
38eef611 | 1148 | memcpy(p->pack_name + root_len, ".pack", sizeof(".pack")); |
38eef611 | 1149 | } |
a070f152 | 1150 | |
e1de726c | 1151 | if (p_stat(p->pack_name, &st) < 0 || !S_ISREG(st.st_mode)) { |
3286c408 | 1152 | git__free(p); |
e10144ae | 1153 | return git_odb__error_notfound("packfile not found", NULL, 0); |
a070f152 CMN |
1154 | } |
1155 | ||
1156 | /* ok, it looks sane as far as we can check without | |
1157 | * actually mapping the pack file. | |
1158 | */ | |
38eef611 | 1159 | p->mwf.fd = -1; |
1af56d7d | 1160 | p->mwf.size = st.st_size; |
a070f152 CMN |
1161 | p->pack_local = 1; |
1162 | p->mtime = (git_time_t)st.st_mtime; | |
0ddfcb40 | 1163 | p->index_version = -1; |
a070f152 | 1164 | |
1a42dd17 | 1165 | if (git_mutex_init(&p->lock)) { |
ac3d33df | 1166 | git_error_set(GIT_ERROR_OS, "failed to initialize packfile mutex"); |
1a42dd17 RB |
1167 | git__free(p); |
1168 | return -1; | |
1169 | } | |
38eef611 | 1170 | |
649214be CMN |
1171 | if (cache_init(&p->bases) < 0) { |
1172 | git__free(p); | |
1173 | return -1; | |
1174 | } | |
1175 | ||
a070f152 | 1176 | *pack_out = p; |
e1de726c RB |
1177 | |
1178 | return 0; | |
a070f152 CMN |
1179 | } |
1180 | ||
1181 | /*********************************************************** | |
1182 | * | |
1183 | * PACKFILE ENTRY SEARCH INTERNALS | |
1184 | * | |
1185 | ***********************************************************/ | |
1186 | ||
e1de726c | 1187 | static git_off_t nth_packed_object_offset(const struct git_pack_file *p, uint32_t n) |
a070f152 CMN |
1188 | { |
1189 | const unsigned char *index = p->index_map.data; | |
ea9e00cb | 1190 | const unsigned char *end = index + p->index_map.len; |
a070f152 CMN |
1191 | index += 4 * 256; |
1192 | if (p->index_version == 1) { | |
1193 | return ntohl(*((uint32_t *)(index + 24 * n))); | |
1194 | } else { | |
1195 | uint32_t off; | |
1196 | index += 8 + p->num_objects * (20 + 4); | |
1197 | off = ntohl(*((uint32_t *)(index + 4 * n))); | |
1198 | if (!(off & 0x80000000)) | |
1199 | return off; | |
1200 | index += p->num_objects * 4 + (off & 0x7fffffff) * 8; | |
ea9e00cb CMN |
1201 | |
1202 | /* Make sure we're not being sent out of bounds */ | |
1203 | if (index >= end - 8) | |
1204 | return -1; | |
1205 | ||
a070f152 | 1206 | return (((uint64_t)ntohl(*((uint32_t *)(index + 0)))) << 32) | |
87d9869f | 1207 | ntohl(*((uint32_t *)(index + 4))); |
a070f152 CMN |
1208 | } |
1209 | } | |
1210 | ||
60ecdf59 DMB |
1211 | static int git__memcmp4(const void *a, const void *b) { |
1212 | return memcmp(a, b, 4); | |
1213 | } | |
1214 | ||
521aedad | 1215 | int git_pack_foreach_entry( |
5dca2010 | 1216 | struct git_pack_file *p, |
c3fb7d04 | 1217 | git_odb_foreach_cb cb, |
5dca2010 | 1218 | void *data) |
521aedad CMN |
1219 | { |
1220 | const unsigned char *index = p->index_map.data, *current; | |
521aedad | 1221 | uint32_t i; |
dab89f9b | 1222 | int error = 0; |
521aedad CMN |
1223 | |
1224 | if (index == NULL) { | |
521aedad CMN |
1225 | if ((error = pack_index_open(p)) < 0) |
1226 | return error; | |
1227 | ||
1228 | assert(p->index_map.data); | |
1229 | ||
1230 | index = p->index_map.data; | |
1231 | } | |
1232 | ||
1233 | if (p->index_version > 1) { | |
1234 | index += 8; | |
1235 | } | |
1236 | ||
1237 | index += 4 * 256; | |
1238 | ||
60ecdf59 DMB |
1239 | if (p->oids == NULL) { |
1240 | git_vector offsets, oids; | |
521aedad | 1241 | |
60ecdf59 DMB |
1242 | if ((error = git_vector_init(&oids, p->num_objects, NULL))) |
1243 | return error; | |
1244 | ||
1245 | if ((error = git_vector_init(&offsets, p->num_objects, git__memcmp4))) | |
1246 | return error; | |
5dca2010 | 1247 | |
60ecdf59 DMB |
1248 | if (p->index_version > 1) { |
1249 | const unsigned char *off = index + 24 * p->num_objects; | |
1250 | for (i = 0; i < p->num_objects; i++) | |
1251 | git_vector_insert(&offsets, (void*)&off[4 * i]); | |
1252 | git_vector_sort(&offsets); | |
1253 | git_vector_foreach(&offsets, i, current) | |
1254 | git_vector_insert(&oids, (void*)&index[5 * (current - off)]); | |
1255 | } else { | |
1256 | for (i = 0; i < p->num_objects; i++) | |
1257 | git_vector_insert(&offsets, (void*)&index[24 * i]); | |
1258 | git_vector_sort(&offsets); | |
1259 | git_vector_foreach(&offsets, i, current) | |
1260 | git_vector_insert(&oids, (void*)¤t[4]); | |
1261 | } | |
25e0b157 | 1262 | |
60ecdf59 | 1263 | git_vector_free(&offsets); |
25e0b157 | 1264 | p->oids = (git_oid **)git_vector_detach(NULL, NULL, &oids); |
521aedad CMN |
1265 | } |
1266 | ||
60ecdf59 | 1267 | for (i = 0; i < p->num_objects; i++) |
26c1cb91 | 1268 | if ((error = cb(p->oids[i], data)) != 0) |
ac3d33df | 1269 | return git_error_set_after_callback(error); |
60ecdf59 | 1270 | |
25e0b157 | 1271 | return error; |
521aedad CMN |
1272 | } |
1273 | ||
a070f152 | 1274 | static int pack_entry_find_offset( |
e1de726c RB |
1275 | git_off_t *offset_out, |
1276 | git_oid *found_oid, | |
1277 | struct git_pack_file *p, | |
1278 | const git_oid *short_oid, | |
b8457baa | 1279 | size_t len) |
a070f152 | 1280 | { |
0cf15e39 PS |
1281 | const uint32_t *level1_ofs; |
1282 | const unsigned char *index; | |
a070f152 CMN |
1283 | unsigned hi, lo, stride; |
1284 | int pos, found = 0; | |
ea9e00cb | 1285 | git_off_t offset; |
a070f152 CMN |
1286 | const unsigned char *current = 0; |
1287 | ||
1288 | *offset_out = 0; | |
1289 | ||
0ddfcb40 | 1290 | if (p->index_version == -1) { |
34bd5999 | 1291 | int error; |
a070f152 | 1292 | |
34bd5999 CMN |
1293 | if ((error = pack_index_open(p)) < 0) |
1294 | return error; | |
1295 | assert(p->index_map.data); | |
34bd5999 | 1296 | } |
a070f152 | 1297 | |
0cf15e39 PS |
1298 | index = p->index_map.data; |
1299 | level1_ofs = p->index_map.data; | |
1300 | ||
a070f152 CMN |
1301 | if (p->index_version > 1) { |
1302 | level1_ofs += 2; | |
1303 | index += 8; | |
1304 | } | |
1305 | ||
1306 | index += 4 * 256; | |
1307 | hi = ntohl(level1_ofs[(int)short_oid->id[0]]); | |
1308 | lo = ((short_oid->id[0] == 0x0) ? 0 : ntohl(level1_ofs[(int)short_oid->id[0] - 1])); | |
1309 | ||
1310 | if (p->index_version > 1) { | |
1311 | stride = 20; | |
1312 | } else { | |
1313 | stride = 24; | |
1314 | index += 4; | |
1315 | } | |
1316 | ||
1317 | #ifdef INDEX_DEBUG_LOOKUP | |
1318 | printf("%02x%02x%02x... lo %u hi %u nr %d\n", | |
1319 | short_oid->id[0], short_oid->id[1], short_oid->id[2], lo, hi, p->num_objects); | |
1320 | #endif | |
1321 | ||
67591c8c | 1322 | pos = sha1_position(index, stride, lo, hi, short_oid->id); |
a070f152 CMN |
1323 | |
1324 | if (pos >= 0) { | |
1325 | /* An object matching exactly the oid was found */ | |
1326 | found = 1; | |
1327 | current = index + pos * stride; | |
1328 | } else { | |
1329 | /* No object was found */ | |
1330 | /* pos refers to the object with the "closest" oid to short_oid */ | |
1331 | pos = - 1 - pos; | |
1332 | if (pos < (int)p->num_objects) { | |
1333 | current = index + pos * stride; | |
1334 | ||
282283ac | 1335 | if (!git_oid_ncmp(short_oid, (const git_oid *)current, len)) |
a070f152 | 1336 | found = 1; |
a070f152 CMN |
1337 | } |
1338 | } | |
1339 | ||
b2a2702d | 1340 | if (found && len != GIT_OID_HEXSZ && pos + 1 < (int)p->num_objects) { |
a070f152 CMN |
1341 | /* Check for ambiguousity */ |
1342 | const unsigned char *next = current + stride; | |
1343 | ||
1344 | if (!git_oid_ncmp(short_oid, (const git_oid *)next, len)) { | |
1345 | found = 2; | |
1346 | } | |
1347 | } | |
1348 | ||
e1de726c | 1349 | if (!found) |
e10144ae | 1350 | return git_odb__error_notfound("failed to find offset for pack entry", short_oid, len); |
e1de726c RB |
1351 | if (found > 1) |
1352 | return git_odb__error_ambiguous("found multiple offsets for pack entry"); | |
24c70804 | 1353 | |
ea9e00cb | 1354 | if ((offset = nth_packed_object_offset(p, pos)) < 0) { |
ac3d33df | 1355 | git_error_set(GIT_ERROR_ODB, "packfile index is corrupt"); |
ea9e00cb CMN |
1356 | return -1; |
1357 | } | |
1358 | ||
1359 | *offset_out = offset; | |
e1de726c | 1360 | git_oid_fromraw(found_oid, current); |
a070f152 CMN |
1361 | |
1362 | #ifdef INDEX_DEBUG_LOOKUP | |
e1de726c | 1363 | { |
a070f152 CMN |
1364 | unsigned char hex_sha1[GIT_OID_HEXSZ + 1]; |
1365 | git_oid_fmt(hex_sha1, found_oid); | |
1366 | hex_sha1[GIT_OID_HEXSZ] = '\0'; | |
1367 | printf("found lo=%d %s\n", lo, hex_sha1); | |
a070f152 | 1368 | } |
e1de726c | 1369 | #endif |
24c70804 | 1370 | |
e1de726c | 1371 | return 0; |
a070f152 CMN |
1372 | } |
1373 | ||
1374 | int git_pack_entry_find( | |
1375 | struct git_pack_entry *e, | |
1376 | struct git_pack_file *p, | |
1377 | const git_oid *short_oid, | |
b8457baa | 1378 | size_t len) |
a070f152 | 1379 | { |
e1de726c | 1380 | git_off_t offset; |
a070f152 CMN |
1381 | git_oid found_oid; |
1382 | int error; | |
1383 | ||
1384 | assert(p); | |
1385 | ||
1386 | if (len == GIT_OID_HEXSZ && p->num_bad_objects) { | |
1387 | unsigned i; | |
1388 | for (i = 0; i < p->num_bad_objects; i++) | |
b7f167da | 1389 | if (git_oid__cmp(short_oid, &p->bad_object_sha1[i]) == 0) |
e1de726c | 1390 | return packfile_error("bad object found in packfile"); |
a070f152 CMN |
1391 | } |
1392 | ||
1393 | error = pack_entry_find_offset(&offset, &found_oid, p, short_oid, len); | |
e1de726c RB |
1394 | if (error < 0) |
1395 | return error; | |
a070f152 CMN |
1396 | |
1397 | /* we found a unique entry in the index; | |
1398 | * make sure the packfile backing the index | |
1399 | * still exists on disk */ | |
e1de726c RB |
1400 | if (p->mwf.fd == -1 && (error = packfile_open(p)) < 0) |
1401 | return error; | |
a070f152 CMN |
1402 | |
1403 | e->offset = offset; | |
1404 | e->p = p; | |
1405 | ||
1406 | git_oid_cpy(&e->sha1, &found_oid); | |
e1de726c | 1407 | return 0; |
a070f152 | 1408 | } |