]>
Commit | Line | Data |
---|---|---|
7d0cdf82 | 1 | /* |
359fc2d2 | 2 | * Copyright (C) the libgit2 contributors. All rights reserved. |
7d0cdf82 | 3 | * |
bb742ede VM |
4 | * This file is part of libgit2, distributed under the GNU GPL v2 with |
5 | * a Linking Exception. For full terms see the included COPYING file. | |
7d0cdf82 CMN |
6 | */ |
7 | ||
a15c550d | 8 | #include "common.h" |
7d0cdf82 CMN |
9 | #include "odb.h" |
10 | #include "pack.h" | |
11 | #include "delta-apply.h" | |
a070f152 | 12 | #include "sha1_lookup.h" |
a15c550d VM |
13 | #include "mwindow.h" |
14 | #include "fileops.h" | |
b7f167da | 15 | #include "oid.h" |
7d0cdf82 | 16 | |
0c3bae62 | 17 | #include <zlib.h> |
7d0cdf82 | 18 | |
a070f152 | 19 | static int packfile_open(struct git_pack_file *p); |
e1de726c | 20 | static git_off_t nth_packed_object_offset(const struct git_pack_file *p, uint32_t n); |
a070f152 CMN |
21 | int packfile_unpack_compressed( |
22 | git_rawobj *obj, | |
23 | struct git_pack_file *p, | |
24 | git_mwindow **w_curs, | |
e1de726c | 25 | git_off_t *curpos, |
a070f152 CMN |
26 | size_t size, |
27 | git_otype type); | |
28 | ||
29 | /* Can find the offset of an object given | |
30 | * a prefix of an identifier. | |
904b67e6 | 31 | * Throws GIT_EAMBIGUOUSOIDPREFIX if short oid |
a070f152 CMN |
32 | * is ambiguous within the pack. |
33 | * This method assumes that len is between | |
34 | * GIT_OID_MINPREFIXLEN and GIT_OID_HEXSZ. | |
35 | */ | |
36 | static int pack_entry_find_offset( | |
e1de726c | 37 | git_off_t *offset_out, |
a070f152 CMN |
38 | git_oid *found_oid, |
39 | struct git_pack_file *p, | |
40 | const git_oid *short_oid, | |
b8457baa | 41 | size_t len); |
a070f152 | 42 | |
e1de726c RB |
43 | static int packfile_error(const char *message) |
44 | { | |
45 | giterr_set(GITERR_ODB, "Invalid pack file - %s", message); | |
46 | return -1; | |
47 | } | |
48 | ||
c8f79c2b CMN |
49 | /******************** |
50 | * Delta base cache | |
51 | ********************/ | |
c0f4a011 | 52 | |
525d961c | 53 | static git_pack_cache_entry *new_cache_object(git_rawobj *source) |
c0f4a011 | 54 | { |
c8f79c2b | 55 | git_pack_cache_entry *e = git__calloc(1, sizeof(git_pack_cache_entry)); |
c0f4a011 CMN |
56 | if (!e) |
57 | return NULL; | |
58 | ||
c0f4a011 CMN |
59 | memcpy(&e->raw, source, sizeof(git_rawobj)); |
60 | ||
61 | return e; | |
62 | } | |
63 | ||
64 | static void free_cache_object(void *o) | |
65 | { | |
66 | git_pack_cache_entry *e = (git_pack_cache_entry *)o; | |
67 | ||
68 | if (e != NULL) { | |
0ed75620 | 69 | assert(e->refcount.val == 0); |
c0f4a011 CMN |
70 | git__free(e->raw.data); |
71 | git__free(e); | |
72 | } | |
73 | } | |
74 | ||
c8f79c2b CMN |
75 | static void cache_free(git_pack_cache *cache) |
76 | { | |
77 | khiter_t k; | |
78 | ||
79 | if (cache->entries) { | |
80 | for (k = kh_begin(cache->entries); k != kh_end(cache->entries); k++) { | |
81 | if (kh_exist(cache->entries, k)) | |
82 | free_cache_object(kh_value(cache->entries, k)); | |
83 | } | |
84 | ||
85 | git_offmap_free(cache->entries); | |
649214be | 86 | cache->entries = NULL; |
c8f79c2b CMN |
87 | } |
88 | } | |
89 | ||
90 | static int cache_init(git_pack_cache *cache) | |
91 | { | |
c8f79c2b CMN |
92 | cache->entries = git_offmap_alloc(); |
93 | GITERR_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)) { | |
98 | giterr_set(GITERR_OS, "Failed to initialize pack cache mutex"); | |
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 CMN |
110 | { |
111 | khiter_t k; | |
112 | git_pack_cache_entry *entry = NULL; | |
113 | ||
09e29e47 CMN |
114 | if (git_mutex_lock(&cache->lock) < 0) |
115 | return NULL; | |
116 | ||
c8f79c2b CMN |
117 | k = kh_get(off, cache->entries, offset); |
118 | if (k != kh_end(cache->entries)) { /* found it */ | |
119 | entry = kh_value(cache->entries, k); | |
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 | { | |
ed6648ba CMN |
131 | git_pack_cache_entry *entry; |
132 | khiter_t k; | |
133 | ||
0ed75620 CMN |
134 | for (k = kh_begin(cache->entries); k != kh_end(cache->entries); k++) { |
135 | if (!kh_exist(cache->entries, k)) | |
136 | continue; | |
137 | ||
138 | entry = kh_value(cache->entries, k); | |
ed6648ba | 139 | |
9c62aaab CMN |
140 | if (entry && entry->refcount.val == 0) { |
141 | cache->memory_used -= entry->raw.len; | |
142 | kh_del(off, cache->entries, k); | |
143 | free_cache_object(entry); | |
ed6648ba CMN |
144 | } |
145 | } | |
0ed75620 CMN |
146 | } |
147 | ||
c8f79c2b CMN |
148 | static int cache_add(git_pack_cache *cache, git_rawobj *base, git_off_t offset) |
149 | { | |
150 | git_pack_cache_entry *entry; | |
151 | int error, exists = 0; | |
152 | khiter_t k; | |
153 | ||
0ed75620 CMN |
154 | if (base->len > GIT_PACK_CACHE_SIZE_LIMIT) |
155 | return -1; | |
156 | ||
c8f79c2b CMN |
157 | entry = new_cache_object(base); |
158 | if (entry) { | |
09e29e47 CMN |
159 | if (git_mutex_lock(&cache->lock) < 0) { |
160 | giterr_set(GITERR_OS, "failed to lock cache"); | |
6f73e026 | 161 | git__free(entry); |
09e29e47 CMN |
162 | return -1; |
163 | } | |
c8f79c2b CMN |
164 | /* Add it to the cache if nobody else has */ |
165 | exists = kh_get(off, cache->entries, offset) != kh_end(cache->entries); | |
166 | if (!exists) { | |
0ed75620 CMN |
167 | while (cache->memory_used + base->len > cache->memory_limit) |
168 | free_lowest_entry(cache); | |
169 | ||
c8f79c2b CMN |
170 | k = kh_put(off, cache->entries, offset, &error); |
171 | assert(error != 0); | |
172 | kh_value(cache->entries, k) = entry; | |
0ed75620 | 173 | cache->memory_used += entry->raw.len; |
c8f79c2b CMN |
174 | } |
175 | git_mutex_unlock(&cache->lock); | |
176 | /* Somebody beat us to adding it into the cache */ | |
177 | if (exists) { | |
178 | git__free(entry); | |
179 | return -1; | |
180 | } | |
181 | } | |
182 | ||
183 | return 0; | |
184 | } | |
185 | ||
a070f152 CMN |
186 | /*********************************************************** |
187 | * | |
188 | * PACK INDEX METHODS | |
189 | * | |
190 | ***********************************************************/ | |
191 | ||
192 | static void pack_index_free(struct git_pack_file *p) | |
193 | { | |
60ecdf59 DMB |
194 | if (p->oids) { |
195 | git__free(p->oids); | |
196 | p->oids = NULL; | |
197 | } | |
a070f152 CMN |
198 | if (p->index_map.data) { |
199 | git_futils_mmap_free(&p->index_map); | |
200 | p->index_map.data = NULL; | |
201 | } | |
202 | } | |
203 | ||
87d9869f | 204 | static int pack_index_check(const char *path, struct git_pack_file *p) |
a070f152 CMN |
205 | { |
206 | struct git_pack_idx_header *hdr; | |
207 | uint32_t version, nr, i, *index; | |
a070f152 CMN |
208 | void *idx_map; |
209 | size_t idx_size; | |
a070f152 | 210 | struct stat st; |
a070f152 | 211 | int error; |
e1de726c RB |
212 | /* TODO: properly open the file without access time using O_NOATIME */ |
213 | git_file fd = git_futils_open_ro(path); | |
a070f152 | 214 | if (fd < 0) |
e1de726c | 215 | return fd; |
a070f152 | 216 | |
9d2f841a RB |
217 | if (p_fstat(fd, &st) < 0) { |
218 | p_close(fd); | |
219 | giterr_set(GITERR_OS, "Unable to stat pack index '%s'", path); | |
220 | return -1; | |
221 | } | |
222 | ||
223 | if (!S_ISREG(st.st_mode) || | |
e1de726c RB |
224 | !git__is_sizet(st.st_size) || |
225 | (idx_size = (size_t)st.st_size) < 4 * 256 + 20 + 20) | |
226 | { | |
a070f152 | 227 | p_close(fd); |
9d2f841a | 228 | giterr_set(GITERR_ODB, "Invalid pack index '%s'", path); |
e1de726c | 229 | return -1; |
a070f152 CMN |
230 | } |
231 | ||
232 | error = git_futils_mmap_ro(&p->index_map, fd, 0, idx_size); | |
e1de726c | 233 | |
a070f152 CMN |
234 | p_close(fd); |
235 | ||
e1de726c RB |
236 | if (error < 0) |
237 | return error; | |
a070f152 CMN |
238 | |
239 | hdr = idx_map = p->index_map.data; | |
240 | ||
241 | if (hdr->idx_signature == htonl(PACK_IDX_SIGNATURE)) { | |
242 | version = ntohl(hdr->idx_version); | |
243 | ||
244 | if (version < 2 || version > 2) { | |
245 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 246 | return packfile_error("unsupported index version"); |
a070f152 CMN |
247 | } |
248 | ||
249 | } else | |
250 | version = 1; | |
251 | ||
252 | nr = 0; | |
253 | index = idx_map; | |
254 | ||
255 | if (version > 1) | |
87d9869f | 256 | index += 2; /* skip index header */ |
a070f152 CMN |
257 | |
258 | for (i = 0; i < 256; i++) { | |
259 | uint32_t n = ntohl(index[i]); | |
260 | if (n < nr) { | |
261 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 262 | return packfile_error("index is non-monotonic"); |
a070f152 CMN |
263 | } |
264 | nr = n; | |
265 | } | |
266 | ||
267 | if (version == 1) { | |
268 | /* | |
269 | * Total size: | |
87d9869f VM |
270 | * - 256 index entries 4 bytes each |
271 | * - 24-byte entries * nr (20-byte sha1 + 4-byte offset) | |
272 | * - 20-byte SHA1 of the packfile | |
273 | * - 20-byte SHA1 file checksum | |
a070f152 CMN |
274 | */ |
275 | if (idx_size != 4*256 + nr * 24 + 20 + 20) { | |
276 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 277 | return packfile_error("index is corrupted"); |
a070f152 CMN |
278 | } |
279 | } else if (version == 2) { | |
280 | /* | |
281 | * Minimum size: | |
87d9869f VM |
282 | * - 8 bytes of header |
283 | * - 256 index entries 4 bytes each | |
284 | * - 20-byte sha1 entry * nr | |
285 | * - 4-byte crc entry * nr | |
286 | * - 4-byte offset entry * nr | |
287 | * - 20-byte SHA1 of the packfile | |
288 | * - 20-byte SHA1 file checksum | |
a070f152 CMN |
289 | * And after the 4-byte offset table might be a |
290 | * variable sized table containing 8-byte entries | |
291 | * for offsets larger than 2^31. | |
292 | */ | |
293 | unsigned long min_size = 8 + 4*256 + nr*(20 + 4 + 4) + 20 + 20; | |
294 | unsigned long max_size = min_size; | |
295 | ||
296 | if (nr) | |
297 | max_size += (nr - 1)*8; | |
298 | ||
299 | if (idx_size < min_size || idx_size > max_size) { | |
300 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 301 | return packfile_error("wrong index size"); |
a070f152 | 302 | } |
a070f152 CMN |
303 | } |
304 | ||
a070f152 | 305 | p->num_objects = nr; |
0ddfcb40 | 306 | p->index_version = version; |
e1de726c | 307 | return 0; |
a070f152 CMN |
308 | } |
309 | ||
310 | static int pack_index_open(struct git_pack_file *p) | |
311 | { | |
312 | char *idx_name; | |
24c70804 RB |
313 | int error = 0; |
314 | size_t name_len, base_len; | |
315 | ||
0ddfcb40 | 316 | if (p->index_version > -1) |
53607868 | 317 | return 0; |
a070f152 | 318 | |
24c70804 RB |
319 | name_len = strlen(p->pack_name); |
320 | assert(name_len > strlen(".pack")); /* checked by git_pack_file alloc */ | |
e1de726c | 321 | |
53607868 RB |
322 | if ((idx_name = git__malloc(name_len)) == NULL) |
323 | return -1; | |
44ef8b1b | 324 | |
24c70804 RB |
325 | base_len = name_len - strlen(".pack"); |
326 | memcpy(idx_name, p->pack_name, base_len); | |
327 | memcpy(idx_name + base_len, ".idx", sizeof(".idx")); | |
a070f152 | 328 | |
050af8bb RD |
329 | if ((error = git_mutex_lock(&p->lock)) < 0) { |
330 | git__free(idx_name); | |
53607868 | 331 | return error; |
050af8bb | 332 | } |
53607868 | 333 | |
0ddfcb40 | 334 | if (p->index_version == -1) |
53607868 | 335 | error = pack_index_check(idx_name, p); |
24c70804 | 336 | |
3286c408 | 337 | git__free(idx_name); |
a070f152 | 338 | |
24c70804 RB |
339 | git_mutex_unlock(&p->lock); |
340 | ||
e1de726c | 341 | return error; |
a070f152 CMN |
342 | } |
343 | ||
344 | static unsigned char *pack_window_open( | |
345 | struct git_pack_file *p, | |
7d0cdf82 | 346 | git_mwindow **w_cursor, |
e1de726c | 347 | git_off_t offset, |
7d0cdf82 CMN |
348 | unsigned int *left) |
349 | { | |
e1de726c | 350 | if (p->mwf.fd == -1 && packfile_open(p) < 0) |
7d0cdf82 CMN |
351 | return NULL; |
352 | ||
353 | /* Since packfiles end in a hash of their content and it's | |
354 | * pointless to ask for an offset into the middle of that | |
355 | * hash, and the pack_window_contains function above wouldn't match | |
356 | * don't allow an offset too close to the end of the file. | |
357 | */ | |
358 | if (offset > (p->mwf.size - 20)) | |
359 | return NULL; | |
360 | ||
361 | return git_mwindow_open(&p->mwf, w_cursor, offset, 20, left); | |
362 | } | |
363 | ||
51e82492 CMN |
364 | /* |
365 | * The per-object header is a pretty dense thing, which is | |
366 | * - first byte: low four bits are "size", | |
367 | * then three bits of "type", | |
368 | * with the high bit being "size continues". | |
369 | * - each byte afterwards: low seven bits are size continuation, | |
370 | * with the high bit being "size continues" | |
371 | */ | |
51a3dfb5 | 372 | size_t git_packfile__object_header(unsigned char *hdr, size_t size, git_otype type) |
51e82492 CMN |
373 | { |
374 | unsigned char *hdr_base; | |
375 | unsigned char c; | |
376 | ||
377 | assert(type >= GIT_OBJ_COMMIT && type <= GIT_OBJ_REF_DELTA); | |
378 | ||
379 | /* TODO: add support for chunked objects; see git.git 6c0d19b1 */ | |
380 | ||
381 | c = (unsigned char)((type << 4) | (size & 15)); | |
382 | size >>= 4; | |
383 | hdr_base = hdr; | |
384 | ||
385 | while (size) { | |
386 | *hdr++ = c | 0x80; | |
387 | c = size & 0x7f; | |
388 | size >>= 7; | |
389 | } | |
390 | *hdr++ = c; | |
391 | ||
51a3dfb5 | 392 | return (hdr - hdr_base); |
51e82492 CMN |
393 | } |
394 | ||
395 | ||
45d773ef CMN |
396 | static int packfile_unpack_header1( |
397 | unsigned long *usedp, | |
7d0cdf82 CMN |
398 | size_t *sizep, |
399 | git_otype *type, | |
400 | const unsigned char *buf, | |
401 | unsigned long len) | |
402 | { | |
403 | unsigned shift; | |
404 | unsigned long size, c; | |
405 | unsigned long used = 0; | |
2aeadb9c | 406 | |
7d0cdf82 CMN |
407 | c = buf[used++]; |
408 | *type = (c >> 4) & 7; | |
409 | size = c & 15; | |
410 | shift = 4; | |
411 | while (c & 0x80) { | |
ec7e680c RP |
412 | if (len <= used) { |
413 | giterr_set(GITERR_ODB, "buffer too small"); | |
904b67e6 | 414 | return GIT_EBUFS; |
ec7e680c | 415 | } |
45d773ef CMN |
416 | |
417 | if (bitsizeof(long) <= shift) { | |
418 | *usedp = 0; | |
ec7e680c | 419 | giterr_set(GITERR_ODB, "packfile corrupted"); |
45d773ef CMN |
420 | return -1; |
421 | } | |
7d0cdf82 CMN |
422 | |
423 | c = buf[used++]; | |
424 | size += (c & 0x7f) << shift; | |
425 | shift += 7; | |
426 | } | |
427 | ||
428 | *sizep = (size_t)size; | |
45d773ef CMN |
429 | *usedp = used; |
430 | return 0; | |
7d0cdf82 CMN |
431 | } |
432 | ||
433 | int git_packfile_unpack_header( | |
434 | size_t *size_p, | |
435 | git_otype *type_p, | |
436 | git_mwindow_file *mwf, | |
437 | git_mwindow **w_curs, | |
e1de726c | 438 | git_off_t *curpos) |
7d0cdf82 CMN |
439 | { |
440 | unsigned char *base; | |
441 | unsigned int left; | |
442 | unsigned long used; | |
45d773ef | 443 | int ret; |
7d0cdf82 CMN |
444 | |
445 | /* pack_window_open() assures us we have [base, base + 20) available | |
87d9869f VM |
446 | * as a range that we can look at at. (Its actually the hash |
447 | * size that is assured.) With our object header encoding | |
7d0cdf82 CMN |
448 | * the maximum deflated object size is 2^137, which is just |
449 | * insane, so we know won't exceed what we have been given. | |
450 | */ | |
24c70804 | 451 | /* base = pack_window_open(p, w_curs, *curpos, &left); */ |
7d0cdf82 CMN |
452 | base = git_mwindow_open(mwf, w_curs, *curpos, 20, &left); |
453 | if (base == NULL) | |
904b67e6 | 454 | return GIT_EBUFS; |
2aeadb9c | 455 | |
9d2f841a | 456 | ret = packfile_unpack_header1(&used, size_p, type_p, base, left); |
45d773ef | 457 | git_mwindow_close(w_curs); |
904b67e6 | 458 | if (ret == GIT_EBUFS) |
45d773ef CMN |
459 | return ret; |
460 | else if (ret < 0) | |
e1de726c | 461 | return packfile_error("header length is zero"); |
7d0cdf82 CMN |
462 | |
463 | *curpos += used; | |
e1de726c | 464 | return 0; |
7d0cdf82 CMN |
465 | } |
466 | ||
44f9f547 DMB |
467 | int git_packfile_resolve_header( |
468 | size_t *size_p, | |
469 | git_otype *type_p, | |
470 | struct git_pack_file *p, | |
471 | git_off_t offset) | |
472 | { | |
473 | git_mwindow *w_curs = NULL; | |
474 | git_off_t curpos = offset; | |
475 | size_t size; | |
476 | git_otype type; | |
477 | git_off_t base_offset; | |
478 | int error; | |
479 | ||
480 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
481 | git_mwindow_close(&w_curs); | |
482 | if (error < 0) | |
483 | return error; | |
484 | ||
485 | if (type == GIT_OBJ_OFS_DELTA || type == GIT_OBJ_REF_DELTA) { | |
486 | size_t base_size; | |
487 | git_rawobj delta; | |
488 | base_offset = get_delta_base(p, &w_curs, &curpos, type, offset); | |
489 | git_mwindow_close(&w_curs); | |
490 | error = packfile_unpack_compressed(&delta, p, &w_curs, &curpos, size, type); | |
491 | git_mwindow_close(&w_curs); | |
492 | if (error < 0) | |
493 | return error; | |
494 | error = git__delta_read_header(delta.data, delta.len, &base_size, size_p); | |
495 | git__free(delta.data); | |
496 | if (error < 0) | |
497 | return error; | |
498 | } else | |
499 | *size_p = size; | |
500 | ||
501 | while (type == GIT_OBJ_OFS_DELTA || type == GIT_OBJ_REF_DELTA) { | |
502 | curpos = base_offset; | |
503 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
504 | git_mwindow_close(&w_curs); | |
505 | if (error < 0) | |
506 | return error; | |
507 | if (type != GIT_OBJ_OFS_DELTA && type != GIT_OBJ_REF_DELTA) | |
508 | break; | |
509 | base_offset = get_delta_base(p, &w_curs, &curpos, type, base_offset); | |
510 | git_mwindow_close(&w_curs); | |
511 | } | |
512 | *type_p = type; | |
513 | ||
514 | return error; | |
515 | } | |
516 | ||
15bcced2 CMN |
517 | #define SMALL_STACK_SIZE 64 |
518 | ||
a3ffbf23 CMN |
519 | /** |
520 | * Generate the chain of dependencies which we need to get to the | |
521 | * object at `off`. `chain` is used a stack, popping gives the right | |
522 | * order to apply deltas on. If an object is found in the pack's base | |
523 | * cache, we stop calculating there. | |
524 | */ | |
15bcced2 CMN |
525 | static int pack_dependency_chain(git_dependency_chain *chain_out, |
526 | git_pack_cache_entry **cached_out, git_off_t *cached_off, | |
527 | struct pack_chain_elem *small_stack, size_t *stack_sz, | |
528 | struct git_pack_file *p, git_off_t obj_offset) | |
a3ffbf23 CMN |
529 | { |
530 | git_dependency_chain chain = GIT_ARRAY_INIT; | |
531 | git_mwindow *w_curs = NULL; | |
532 | git_off_t curpos = obj_offset, base_offset; | |
15bcced2 CMN |
533 | int error = 0, use_heap = 0; |
534 | size_t size, elem_pos; | |
a3ffbf23 CMN |
535 | git_otype type; |
536 | ||
15bcced2 | 537 | elem_pos = 0; |
a3ffbf23 CMN |
538 | while (true) { |
539 | struct pack_chain_elem *elem; | |
540 | git_pack_cache_entry *cached = NULL; | |
541 | ||
542 | /* if we have a base cached, we can stop here instead */ | |
543 | if ((cached = cache_get(&p->bases, obj_offset)) != NULL) { | |
544 | *cached_out = cached; | |
545 | *cached_off = obj_offset; | |
546 | break; | |
547 | } | |
548 | ||
15bcced2 CMN |
549 | /* if we run out of space on the small stack, use the array */ |
550 | if (elem_pos == SMALL_STACK_SIZE) { | |
551 | git_array_init_to_size(chain, elem_pos); | |
552 | GITERR_CHECK_ARRAY(chain); | |
553 | memcpy(chain.ptr, small_stack, elem_pos * sizeof(struct pack_chain_elem)); | |
554 | chain.size = elem_pos; | |
555 | use_heap = 1; | |
556 | } | |
557 | ||
a3ffbf23 | 558 | curpos = obj_offset; |
15bcced2 CMN |
559 | if (!use_heap) { |
560 | elem = &small_stack[elem_pos]; | |
561 | } else { | |
562 | elem = git_array_alloc(chain); | |
563 | if (!elem) { | |
564 | error = -1; | |
565 | goto on_error; | |
566 | } | |
a3ffbf23 CMN |
567 | } |
568 | ||
569 | elem->base_key = obj_offset; | |
570 | ||
571 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
572 | git_mwindow_close(&w_curs); | |
573 | ||
574 | if (error < 0) | |
575 | goto on_error; | |
576 | ||
577 | elem->offset = curpos; | |
578 | elem->size = size; | |
579 | elem->type = type; | |
580 | elem->base_key = obj_offset; | |
581 | ||
582 | if (type != GIT_OBJ_OFS_DELTA && type != GIT_OBJ_REF_DELTA) | |
583 | break; | |
584 | ||
585 | base_offset = get_delta_base(p, &w_curs, &curpos, type, obj_offset); | |
586 | git_mwindow_close(&w_curs); | |
587 | ||
588 | if (base_offset == 0) { | |
589 | error = packfile_error("delta offset is zero"); | |
590 | goto on_error; | |
591 | } | |
592 | if (base_offset < 0) { /* must actually be an error code */ | |
593 | error = (int)base_offset; | |
594 | goto on_error; | |
595 | } | |
596 | ||
597 | /* we need to pass the pos *after* the delta-base bit */ | |
598 | elem->offset = curpos; | |
599 | ||
600 | /* go through the loop again, but with the new object */ | |
601 | obj_offset = base_offset; | |
15bcced2 | 602 | elem_pos++; |
a3ffbf23 CMN |
603 | } |
604 | ||
15bcced2 CMN |
605 | |
606 | *stack_sz = elem_pos + 1; | |
a3ffbf23 CMN |
607 | *chain_out = chain; |
608 | return error; | |
609 | ||
610 | on_error: | |
611 | git_array_clear(chain); | |
612 | return error; | |
613 | } | |
614 | ||
a070f152 | 615 | int git_packfile_unpack( |
e1de726c RB |
616 | git_rawobj *obj, |
617 | struct git_pack_file *p, | |
618 | git_off_t *obj_offset) | |
7d0cdf82 CMN |
619 | { |
620 | git_mwindow *w_curs = NULL; | |
e1de726c | 621 | git_off_t curpos = *obj_offset; |
a332e91c CMN |
622 | int error, free_base = 0; |
623 | git_dependency_chain chain = GIT_ARRAY_INIT; | |
15bcced2 | 624 | struct pack_chain_elem *elem = NULL, *stack; |
a332e91c | 625 | git_pack_cache_entry *cached = NULL; |
15bcced2 | 626 | struct pack_chain_elem small_stack[SMALL_STACK_SIZE]; |
e640a77c | 627 | size_t stack_size = 0, elem_pos; |
2acdf4b8 | 628 | git_otype base_type; |
7d0cdf82 CMN |
629 | |
630 | /* | |
631 | * TODO: optionally check the CRC on the packfile | |
632 | */ | |
633 | ||
15bcced2 | 634 | error = pack_dependency_chain(&chain, &cached, obj_offset, small_stack, &stack_size, p, *obj_offset); |
2acdf4b8 CMN |
635 | if (error < 0) |
636 | return error; | |
637 | ||
7d0cdf82 CMN |
638 | obj->data = NULL; |
639 | obj->len = 0; | |
640 | obj->type = GIT_OBJ_BAD; | |
641 | ||
15bcced2 CMN |
642 | /* let's point to the right stack */ |
643 | stack = chain.ptr ? chain.ptr : small_stack; | |
644 | ||
645 | elem_pos = stack_size; | |
a3ffbf23 | 646 | if (cached) { |
a332e91c | 647 | memcpy(obj, &cached->raw, sizeof(git_rawobj)); |
a3ffbf23 | 648 | base_type = obj->type; |
15bcced2 | 649 | elem_pos--; /* stack_size includes the base, which isn't actually there */ |
a3ffbf23 | 650 | } else { |
15bcced2 | 651 | elem = &stack[--elem_pos]; |
a3ffbf23 | 652 | base_type = elem->type; |
a332e91c | 653 | } |
45d773ef | 654 | |
9dbd150f CMN |
655 | switch (base_type) { |
656 | case GIT_OBJ_COMMIT: | |
657 | case GIT_OBJ_TREE: | |
658 | case GIT_OBJ_BLOB: | |
659 | case GIT_OBJ_TAG: | |
a3ffbf23 | 660 | if (!cached) { |
9dbd150f CMN |
661 | curpos = elem->offset; |
662 | error = packfile_unpack_compressed(obj, p, &w_curs, &curpos, elem->size, elem->type); | |
663 | git_mwindow_close(&w_curs); | |
a3ffbf23 | 664 | base_type = elem->type; |
9dbd150f CMN |
665 | } |
666 | if (error < 0) | |
667 | goto cleanup; | |
668 | break; | |
669 | case GIT_OBJ_OFS_DELTA: | |
670 | case GIT_OBJ_REF_DELTA: | |
671 | error = packfile_error("dependency chain ends in a delta"); | |
672 | goto cleanup; | |
673 | default: | |
674 | error = packfile_error("invalid packfile type in header"); | |
675 | goto cleanup; | |
676 | } | |
677 | ||
a332e91c | 678 | /* |
c968ce2c | 679 | * Finding the object we want a cached base element is |
a332e91c CMN |
680 | * problematic, as we need to make sure we don't accidentally |
681 | * give the caller the cached object, which it would then feel | |
682 | * free to free, so we need to copy the data. | |
683 | */ | |
15bcced2 | 684 | if (cached && stack_size == 1) { |
a332e91c | 685 | void *data = obj->data; |
392702ee ET |
686 | |
687 | GITERR_CHECK_ALLOC_ADD(obj->len, 1); | |
a332e91c CMN |
688 | obj->data = git__malloc(obj->len + 1); |
689 | GITERR_CHECK_ALLOC(obj->data); | |
392702ee | 690 | |
a332e91c CMN |
691 | memcpy(obj->data, data, obj->len + 1); |
692 | git_atomic_dec(&cached->refcount); | |
693 | goto cleanup; | |
694 | } | |
695 | ||
2acdf4b8 | 696 | /* we now apply each consecutive delta until we run out */ |
15bcced2 | 697 | while (elem_pos > 0 && !error) { |
2acdf4b8 CMN |
698 | git_rawobj base, delta; |
699 | ||
c968ce2c CMN |
700 | /* |
701 | * We can now try to add the base to the cache, as | |
702 | * long as it's not already the cached one. | |
703 | */ | |
704 | if (!cached) | |
705 | free_base = !!cache_add(&p->bases, obj, elem->base_key); | |
706 | ||
15bcced2 | 707 | elem = &stack[elem_pos - 1]; |
2acdf4b8 CMN |
708 | curpos = elem->offset; |
709 | error = packfile_unpack_compressed(&delta, p, &w_curs, &curpos, elem->size, elem->type); | |
710 | git_mwindow_close(&w_curs); | |
711 | ||
712 | if (error < 0) | |
713 | break; | |
714 | ||
715 | /* the current object becomes the new base, on which we apply the delta */ | |
716 | base = *obj; | |
717 | obj->data = NULL; | |
718 | obj->len = 0; | |
719 | obj->type = GIT_OBJ_BAD; | |
720 | ||
721 | error = git__delta_apply(obj, base.data, base.len, delta.data, delta.len); | |
a332e91c CMN |
722 | obj->type = base_type; |
723 | /* | |
724 | * We usually don't want to free the base at this | |
725 | * point, as we put it into the cache in the previous | |
726 | * iteration. free_base lets us know that we got the | |
727 | * base object directly from the packfile, so we can free it. | |
728 | */ | |
2acdf4b8 | 729 | git__free(delta.data); |
a332e91c CMN |
730 | if (free_base) { |
731 | free_base = 0; | |
732 | git__free(base.data); | |
733 | } | |
734 | ||
735 | if (cached) { | |
736 | git_atomic_dec(&cached->refcount); | |
737 | cached = NULL; | |
738 | } | |
2acdf4b8 CMN |
739 | |
740 | if (error < 0) | |
741 | break; | |
7d0cdf82 | 742 | |
15bcced2 | 743 | elem_pos--; |
7d0cdf82 CMN |
744 | } |
745 | ||
2acdf4b8 | 746 | cleanup: |
a332e91c CMN |
747 | if (error < 0) |
748 | git__free(obj->data); | |
749 | ||
a3ffbf23 | 750 | if (elem) |
b3d3459f | 751 | *obj_offset = curpos; |
a332e91c | 752 | |
2acdf4b8 | 753 | git_array_clear(chain); |
e1de726c | 754 | return error; |
7d0cdf82 CMN |
755 | } |
756 | ||
282283ac RB |
757 | static void *use_git_alloc(void *opaq, unsigned int count, unsigned int size) |
758 | { | |
759 | GIT_UNUSED(opaq); | |
760 | return git__calloc(count, size); | |
761 | } | |
762 | ||
763 | static void use_git_free(void *opaq, void *ptr) | |
764 | { | |
765 | GIT_UNUSED(opaq); | |
766 | git__free(ptr); | |
767 | } | |
768 | ||
46635339 CMN |
769 | int git_packfile_stream_open(git_packfile_stream *obj, struct git_pack_file *p, git_off_t curpos) |
770 | { | |
771 | int st; | |
772 | ||
773 | memset(obj, 0, sizeof(git_packfile_stream)); | |
774 | obj->curpos = curpos; | |
775 | obj->p = p; | |
776 | obj->zstream.zalloc = use_git_alloc; | |
777 | obj->zstream.zfree = use_git_free; | |
778 | obj->zstream.next_in = Z_NULL; | |
779 | obj->zstream.next_out = Z_NULL; | |
780 | st = inflateInit(&obj->zstream); | |
781 | if (st != Z_OK) { | |
782 | git__free(obj); | |
ae081739 | 783 | giterr_set(GITERR_ZLIB, "failed to init packfile stream"); |
46635339 CMN |
784 | return -1; |
785 | } | |
786 | ||
787 | return 0; | |
788 | } | |
789 | ||
790 | ssize_t git_packfile_stream_read(git_packfile_stream *obj, void *buffer, size_t len) | |
791 | { | |
792 | unsigned char *in; | |
793 | size_t written; | |
794 | int st; | |
795 | ||
796 | if (obj->done) | |
797 | return 0; | |
798 | ||
799 | in = pack_window_open(obj->p, &obj->mw, obj->curpos, &obj->zstream.avail_in); | |
800 | if (in == NULL) | |
801 | return GIT_EBUFS; | |
802 | ||
803 | obj->zstream.next_out = buffer; | |
090d5e1f | 804 | obj->zstream.avail_out = (unsigned int)len; |
46635339 CMN |
805 | obj->zstream.next_in = in; |
806 | ||
807 | st = inflate(&obj->zstream, Z_SYNC_FLUSH); | |
808 | git_mwindow_close(&obj->mw); | |
809 | ||
810 | obj->curpos += obj->zstream.next_in - in; | |
811 | written = len - obj->zstream.avail_out; | |
812 | ||
813 | if (st != Z_OK && st != Z_STREAM_END) { | |
ae081739 | 814 | giterr_set(GITERR_ZLIB, "error reading from the zlib stream"); |
46635339 CMN |
815 | return -1; |
816 | } | |
817 | ||
818 | if (st == Z_STREAM_END) | |
819 | obj->done = 1; | |
820 | ||
821 | ||
822 | /* If we didn't write anything out but we're not done, we need more data */ | |
823 | if (!written && st != Z_STREAM_END) | |
824 | return GIT_EBUFS; | |
825 | ||
826 | return written; | |
827 | ||
828 | } | |
829 | ||
830 | void git_packfile_stream_free(git_packfile_stream *obj) | |
831 | { | |
832 | inflateEnd(&obj->zstream); | |
833 | } | |
834 | ||
7d0cdf82 | 835 | int packfile_unpack_compressed( |
e1de726c RB |
836 | git_rawobj *obj, |
837 | struct git_pack_file *p, | |
838 | git_mwindow **w_curs, | |
839 | git_off_t *curpos, | |
840 | size_t size, | |
841 | git_otype type) | |
7d0cdf82 CMN |
842 | { |
843 | int st; | |
844 | z_stream stream; | |
845 | unsigned char *buffer, *in; | |
846 | ||
392702ee | 847 | GITERR_CHECK_ALLOC_ADD(size, 1); |
e1de726c RB |
848 | buffer = git__calloc(1, size + 1); |
849 | GITERR_CHECK_ALLOC(buffer); | |
7d0cdf82 CMN |
850 | |
851 | memset(&stream, 0, sizeof(stream)); | |
852 | stream.next_out = buffer; | |
1c3fac4d | 853 | stream.avail_out = (uInt)size + 1; |
282283ac RB |
854 | stream.zalloc = use_git_alloc; |
855 | stream.zfree = use_git_free; | |
7d0cdf82 CMN |
856 | |
857 | st = inflateInit(&stream); | |
858 | if (st != Z_OK) { | |
3286c408 | 859 | git__free(buffer); |
ae081739 | 860 | giterr_set(GITERR_ZLIB, "failed to init zlib stream on unpack"); |
45d773ef | 861 | |
e1de726c | 862 | return -1; |
7d0cdf82 CMN |
863 | } |
864 | ||
865 | do { | |
b5b474dd | 866 | in = pack_window_open(p, w_curs, *curpos, &stream.avail_in); |
7d0cdf82 CMN |
867 | stream.next_in = in; |
868 | st = inflate(&stream, Z_FINISH); | |
45d773ef | 869 | git_mwindow_close(w_curs); |
7d0cdf82 CMN |
870 | |
871 | if (!stream.avail_out) | |
872 | break; /* the payload is larger than it should be */ | |
873 | ||
45d773ef CMN |
874 | if (st == Z_BUF_ERROR && in == NULL) { |
875 | inflateEnd(&stream); | |
876 | git__free(buffer); | |
904b67e6 | 877 | return GIT_EBUFS; |
45d773ef CMN |
878 | } |
879 | ||
b5b474dd | 880 | *curpos += stream.next_in - in; |
7d0cdf82 CMN |
881 | } while (st == Z_OK || st == Z_BUF_ERROR); |
882 | ||
883 | inflateEnd(&stream); | |
884 | ||
885 | if ((st != Z_STREAM_END) || stream.total_out != size) { | |
3286c408 | 886 | git__free(buffer); |
ae081739 | 887 | giterr_set(GITERR_ZLIB, "error inflating zlib stream"); |
e1de726c | 888 | return -1; |
7d0cdf82 CMN |
889 | } |
890 | ||
891 | obj->type = type; | |
892 | obj->len = size; | |
893 | obj->data = buffer; | |
e1de726c | 894 | return 0; |
7d0cdf82 CMN |
895 | } |
896 | ||
b5b474dd CMN |
897 | /* |
898 | * curpos is where the data starts, delta_obj_offset is the where the | |
899 | * header starts | |
900 | */ | |
e1de726c RB |
901 | git_off_t get_delta_base( |
902 | struct git_pack_file *p, | |
903 | git_mwindow **w_curs, | |
904 | git_off_t *curpos, | |
905 | git_otype type, | |
906 | git_off_t delta_obj_offset) | |
7d0cdf82 | 907 | { |
45d773ef CMN |
908 | unsigned int left = 0; |
909 | unsigned char *base_info; | |
e1de726c | 910 | git_off_t base_offset; |
7d0cdf82 CMN |
911 | git_oid unused; |
912 | ||
45d773ef CMN |
913 | base_info = pack_window_open(p, w_curs, *curpos, &left); |
914 | /* Assumption: the only reason this would fail is because the file is too small */ | |
915 | if (base_info == NULL) | |
904b67e6 | 916 | return GIT_EBUFS; |
7d0cdf82 CMN |
917 | /* pack_window_open() assured us we have [base_info, base_info + 20) |
918 | * as a range that we can look at without walking off the | |
87d9869f VM |
919 | * end of the mapped window. Its actually the hash size |
920 | * that is assured. An OFS_DELTA longer than the hash size | |
7d0cdf82 CMN |
921 | * is stupid, as then a REF_DELTA would be smaller to store. |
922 | */ | |
923 | if (type == GIT_OBJ_OFS_DELTA) { | |
924 | unsigned used = 0; | |
925 | unsigned char c = base_info[used++]; | |
926 | base_offset = c & 127; | |
927 | while (c & 128) { | |
45d773ef | 928 | if (left <= used) |
904b67e6 | 929 | return GIT_EBUFS; |
7d0cdf82 CMN |
930 | base_offset += 1; |
931 | if (!base_offset || MSB(base_offset, 7)) | |
87d9869f | 932 | return 0; /* overflow */ |
7d0cdf82 CMN |
933 | c = base_info[used++]; |
934 | base_offset = (base_offset << 7) + (c & 127); | |
935 | } | |
936 | base_offset = delta_obj_offset - base_offset; | |
937 | if (base_offset <= 0 || base_offset >= delta_obj_offset) | |
87d9869f | 938 | return 0; /* out of bound */ |
7d0cdf82 CMN |
939 | *curpos += used; |
940 | } else if (type == GIT_OBJ_REF_DELTA) { | |
c1af5a39 CMN |
941 | /* If we have the cooperative cache, search in it first */ |
942 | if (p->has_cache) { | |
0e040c03 CMN |
943 | khiter_t k; |
944 | git_oid oid; | |
c1af5a39 | 945 | |
0e040c03 CMN |
946 | git_oid_fromraw(&oid, base_info); |
947 | k = kh_get(oid, p->idx_cache, &oid); | |
948 | if (k != kh_end(p->idx_cache)) { | |
c1af5a39 | 949 | *curpos += 20; |
0e040c03 | 950 | return ((struct git_pack_entry *)kh_value(p->idx_cache, k))->offset; |
c1af5a39 CMN |
951 | } |
952 | } | |
7d0cdf82 | 953 | /* The base entry _must_ be in the same pack */ |
e1de726c RB |
954 | if (pack_entry_find_offset(&base_offset, &unused, p, (git_oid *)base_info, GIT_OID_HEXSZ) < 0) |
955 | return packfile_error("base entry delta is not in the same pack"); | |
7d0cdf82 CMN |
956 | *curpos += 20; |
957 | } else | |
958 | return 0; | |
959 | ||
960 | return base_offset; | |
961 | } | |
a070f152 CMN |
962 | |
963 | /*********************************************************** | |
964 | * | |
965 | * PACKFILE METHODS | |
966 | * | |
967 | ***********************************************************/ | |
968 | ||
96c9b9f0 | 969 | void git_packfile_free(struct git_pack_file *p) |
a070f152 | 970 | { |
24c70804 RB |
971 | if (!p) |
972 | return; | |
973 | ||
c8f79c2b | 974 | cache_free(&p->bases); |
c0f4a011 | 975 | |
5e0f47c3 CMN |
976 | if (p->mwf.fd >= 0) { |
977 | git_mwindow_free_all_locked(&p->mwf); | |
a070f152 | 978 | p_close(p->mwf.fd); |
5e0f47c3 | 979 | } |
a070f152 CMN |
980 | |
981 | pack_index_free(p); | |
982 | ||
3286c408 | 983 | git__free(p->bad_object_sha1); |
24c70804 | 984 | |
24c70804 | 985 | git_mutex_free(&p->lock); |
649214be | 986 | git_mutex_free(&p->bases.lock); |
3286c408 | 987 | git__free(p); |
a070f152 CMN |
988 | } |
989 | ||
990 | static int packfile_open(struct git_pack_file *p) | |
991 | { | |
992 | struct stat st; | |
993 | struct git_pack_header hdr; | |
994 | git_oid sha1; | |
995 | unsigned char *idx_sha1; | |
996 | ||
0ddfcb40 | 997 | if (p->index_version == -1 && pack_index_open(p) < 0) |
282283ac | 998 | return git_odb__error_notfound("failed to open packfile", NULL); |
a070f152 | 999 | |
9d2f841a RB |
1000 | /* if mwf opened by another thread, return now */ |
1001 | if (git_mutex_lock(&p->lock) < 0) | |
1002 | return packfile_error("failed to get lock for open"); | |
1003 | ||
1004 | if (p->mwf.fd >= 0) { | |
1005 | git_mutex_unlock(&p->lock); | |
1006 | return 0; | |
1007 | } | |
1008 | ||
a070f152 | 1009 | /* TODO: open with noatime */ |
e1de726c | 1010 | p->mwf.fd = git_futils_open_ro(p->pack_name); |
9d2f841a RB |
1011 | if (p->mwf.fd < 0) |
1012 | goto cleanup; | |
a070f152 | 1013 | |
e1de726c RB |
1014 | if (p_fstat(p->mwf.fd, &st) < 0 || |
1015 | git_mwindow_file_register(&p->mwf) < 0) | |
1016 | goto cleanup; | |
a070f152 CMN |
1017 | |
1018 | /* If we created the struct before we had the pack we lack size. */ | |
1019 | if (!p->mwf.size) { | |
1020 | if (!S_ISREG(st.st_mode)) | |
1021 | goto cleanup; | |
e1de726c | 1022 | p->mwf.size = (git_off_t)st.st_size; |
a070f152 CMN |
1023 | } else if (p->mwf.size != st.st_size) |
1024 | goto cleanup; | |
1025 | ||
1026 | #if 0 | |
1027 | /* We leave these file descriptors open with sliding mmap; | |
1028 | * there is no point keeping them open across exec(), though. | |
1029 | */ | |
1030 | fd_flag = fcntl(p->mwf.fd, F_GETFD, 0); | |
1031 | if (fd_flag < 0) | |
e1de726c | 1032 | goto cleanup; |
a070f152 CMN |
1033 | |
1034 | fd_flag |= FD_CLOEXEC; | |
1035 | if (fcntl(p->pack_fd, F_SETFD, fd_flag) == -1) | |
e1de726c | 1036 | goto cleanup; |
a070f152 CMN |
1037 | #endif |
1038 | ||
1039 | /* Verify we recognize this pack file format. */ | |
e1de726c RB |
1040 | if (p_read(p->mwf.fd, &hdr, sizeof(hdr)) < 0 || |
1041 | hdr.hdr_signature != htonl(PACK_SIGNATURE) || | |
1042 | !pack_version_ok(hdr.hdr_version)) | |
a070f152 CMN |
1043 | goto cleanup; |
1044 | ||
1045 | /* Verify the pack matches its index. */ | |
e1de726c RB |
1046 | if (p->num_objects != ntohl(hdr.hdr_entries) || |
1047 | p_lseek(p->mwf.fd, p->mwf.size - GIT_OID_RAWSZ, SEEK_SET) == -1 || | |
1048 | p_read(p->mwf.fd, sha1.id, GIT_OID_RAWSZ) < 0) | |
a070f152 CMN |
1049 | goto cleanup; |
1050 | ||
1051 | idx_sha1 = ((unsigned char *)p->index_map.data) + p->index_map.len - 40; | |
1052 | ||
9d2f841a RB |
1053 | if (git_oid__cmp(&sha1, (git_oid *)idx_sha1) != 0) |
1054 | goto cleanup; | |
1055 | ||
1056 | git_mutex_unlock(&p->lock); | |
1057 | return 0; | |
a070f152 CMN |
1058 | |
1059 | cleanup: | |
e1de726c | 1060 | giterr_set(GITERR_OS, "Invalid packfile '%s'", p->pack_name); |
9d2f841a | 1061 | |
3a2d48d5 SS |
1062 | if (p->mwf.fd >= 0) |
1063 | p_close(p->mwf.fd); | |
a070f152 | 1064 | p->mwf.fd = -1; |
9d2f841a RB |
1065 | |
1066 | git_mutex_unlock(&p->lock); | |
1067 | ||
e1de726c | 1068 | return -1; |
a070f152 CMN |
1069 | } |
1070 | ||
b3b66c57 CMN |
1071 | int git_packfile__name(char **out, const char *path) |
1072 | { | |
1073 | size_t path_len; | |
1074 | git_buf buf = GIT_BUF_INIT; | |
1075 | ||
1076 | path_len = strlen(path); | |
1077 | ||
1078 | if (path_len < strlen(".idx")) | |
1079 | return git_odb__error_notfound("invalid packfile path", NULL); | |
1080 | ||
1081 | if (git_buf_printf(&buf, "%.*s.pack", (int)(path_len - strlen(".idx")), path) < 0) | |
1082 | return -1; | |
1083 | ||
1084 | *out = git_buf_detach(&buf); | |
1085 | return 0; | |
1086 | } | |
1087 | ||
5d2d21e5 | 1088 | int git_packfile_alloc(struct git_pack_file **pack_out, const char *path) |
a070f152 CMN |
1089 | { |
1090 | struct stat st; | |
1091 | struct git_pack_file *p; | |
38eef611 | 1092 | size_t path_len = path ? strlen(path) : 0; |
a070f152 CMN |
1093 | |
1094 | *pack_out = NULL; | |
24c70804 | 1095 | |
38eef611 | 1096 | if (path_len < strlen(".idx")) |
24c70804 RB |
1097 | return git_odb__error_notfound("invalid packfile path", NULL); |
1098 | ||
392702ee ET |
1099 | GITERR_CHECK_ALLOC_ADD(sizeof(*p), path_len); |
1100 | GITERR_CHECK_ALLOC_ADD(sizeof(*p) + path_len, 2); | |
1101 | ||
38eef611 | 1102 | p = git__calloc(1, sizeof(*p) + path_len + 2); |
e1de726c | 1103 | GITERR_CHECK_ALLOC(p); |
a070f152 | 1104 | |
38eef611 RB |
1105 | memcpy(p->pack_name, path, path_len + 1); |
1106 | ||
a070f152 CMN |
1107 | /* |
1108 | * Make sure a corresponding .pack file exists and that | |
1109 | * the index looks sane. | |
1110 | */ | |
38eef611 RB |
1111 | if (git__suffixcmp(path, ".idx") == 0) { |
1112 | size_t root_len = path_len - strlen(".idx"); | |
1113 | ||
1114 | memcpy(p->pack_name + root_len, ".keep", sizeof(".keep")); | |
1115 | if (git_path_exists(p->pack_name) == true) | |
1116 | p->pack_keep = 1; | |
a070f152 | 1117 | |
38eef611 | 1118 | memcpy(p->pack_name + root_len, ".pack", sizeof(".pack")); |
38eef611 | 1119 | } |
a070f152 | 1120 | |
e1de726c | 1121 | if (p_stat(p->pack_name, &st) < 0 || !S_ISREG(st.st_mode)) { |
3286c408 | 1122 | git__free(p); |
282283ac | 1123 | return git_odb__error_notfound("packfile not found", NULL); |
a070f152 CMN |
1124 | } |
1125 | ||
1126 | /* ok, it looks sane as far as we can check without | |
1127 | * actually mapping the pack file. | |
1128 | */ | |
38eef611 | 1129 | p->mwf.fd = -1; |
1af56d7d | 1130 | p->mwf.size = st.st_size; |
a070f152 CMN |
1131 | p->pack_local = 1; |
1132 | p->mtime = (git_time_t)st.st_mtime; | |
0ddfcb40 | 1133 | p->index_version = -1; |
a070f152 | 1134 | |
1a42dd17 RB |
1135 | if (git_mutex_init(&p->lock)) { |
1136 | giterr_set(GITERR_OS, "Failed to initialize packfile mutex"); | |
1137 | git__free(p); | |
1138 | return -1; | |
1139 | } | |
38eef611 | 1140 | |
649214be CMN |
1141 | if (cache_init(&p->bases) < 0) { |
1142 | git__free(p); | |
1143 | return -1; | |
1144 | } | |
1145 | ||
a070f152 | 1146 | *pack_out = p; |
e1de726c RB |
1147 | |
1148 | return 0; | |
a070f152 CMN |
1149 | } |
1150 | ||
1151 | /*********************************************************** | |
1152 | * | |
1153 | * PACKFILE ENTRY SEARCH INTERNALS | |
1154 | * | |
1155 | ***********************************************************/ | |
1156 | ||
e1de726c | 1157 | static git_off_t nth_packed_object_offset(const struct git_pack_file *p, uint32_t n) |
a070f152 CMN |
1158 | { |
1159 | const unsigned char *index = p->index_map.data; | |
1160 | index += 4 * 256; | |
1161 | if (p->index_version == 1) { | |
1162 | return ntohl(*((uint32_t *)(index + 24 * n))); | |
1163 | } else { | |
1164 | uint32_t off; | |
1165 | index += 8 + p->num_objects * (20 + 4); | |
1166 | off = ntohl(*((uint32_t *)(index + 4 * n))); | |
1167 | if (!(off & 0x80000000)) | |
1168 | return off; | |
1169 | index += p->num_objects * 4 + (off & 0x7fffffff) * 8; | |
1170 | return (((uint64_t)ntohl(*((uint32_t *)(index + 0)))) << 32) | | |
87d9869f | 1171 | ntohl(*((uint32_t *)(index + 4))); |
a070f152 CMN |
1172 | } |
1173 | } | |
1174 | ||
60ecdf59 DMB |
1175 | static int git__memcmp4(const void *a, const void *b) { |
1176 | return memcmp(a, b, 4); | |
1177 | } | |
1178 | ||
521aedad | 1179 | int git_pack_foreach_entry( |
5dca2010 | 1180 | struct git_pack_file *p, |
c3fb7d04 | 1181 | git_odb_foreach_cb cb, |
5dca2010 | 1182 | void *data) |
521aedad CMN |
1183 | { |
1184 | const unsigned char *index = p->index_map.data, *current; | |
521aedad | 1185 | uint32_t i; |
dab89f9b | 1186 | int error = 0; |
521aedad CMN |
1187 | |
1188 | if (index == NULL) { | |
521aedad CMN |
1189 | if ((error = pack_index_open(p)) < 0) |
1190 | return error; | |
1191 | ||
1192 | assert(p->index_map.data); | |
1193 | ||
1194 | index = p->index_map.data; | |
1195 | } | |
1196 | ||
1197 | if (p->index_version > 1) { | |
1198 | index += 8; | |
1199 | } | |
1200 | ||
1201 | index += 4 * 256; | |
1202 | ||
60ecdf59 DMB |
1203 | if (p->oids == NULL) { |
1204 | git_vector offsets, oids; | |
521aedad | 1205 | |
60ecdf59 DMB |
1206 | if ((error = git_vector_init(&oids, p->num_objects, NULL))) |
1207 | return error; | |
1208 | ||
1209 | if ((error = git_vector_init(&offsets, p->num_objects, git__memcmp4))) | |
1210 | return error; | |
5dca2010 | 1211 | |
60ecdf59 DMB |
1212 | if (p->index_version > 1) { |
1213 | const unsigned char *off = index + 24 * p->num_objects; | |
1214 | for (i = 0; i < p->num_objects; i++) | |
1215 | git_vector_insert(&offsets, (void*)&off[4 * i]); | |
1216 | git_vector_sort(&offsets); | |
1217 | git_vector_foreach(&offsets, i, current) | |
1218 | git_vector_insert(&oids, (void*)&index[5 * (current - off)]); | |
1219 | } else { | |
1220 | for (i = 0; i < p->num_objects; i++) | |
1221 | git_vector_insert(&offsets, (void*)&index[24 * i]); | |
1222 | git_vector_sort(&offsets); | |
1223 | git_vector_foreach(&offsets, i, current) | |
1224 | git_vector_insert(&oids, (void*)¤t[4]); | |
1225 | } | |
25e0b157 | 1226 | |
60ecdf59 | 1227 | git_vector_free(&offsets); |
25e0b157 | 1228 | p->oids = (git_oid **)git_vector_detach(NULL, NULL, &oids); |
521aedad CMN |
1229 | } |
1230 | ||
60ecdf59 | 1231 | for (i = 0; i < p->num_objects; i++) |
26c1cb91 RB |
1232 | if ((error = cb(p->oids[i], data)) != 0) |
1233 | return giterr_set_after_callback(error); | |
60ecdf59 | 1234 | |
25e0b157 | 1235 | return error; |
521aedad CMN |
1236 | } |
1237 | ||
a070f152 | 1238 | static int pack_entry_find_offset( |
e1de726c RB |
1239 | git_off_t *offset_out, |
1240 | git_oid *found_oid, | |
1241 | struct git_pack_file *p, | |
1242 | const git_oid *short_oid, | |
b8457baa | 1243 | size_t len) |
a070f152 | 1244 | { |
34bd5999 CMN |
1245 | const uint32_t *level1_ofs = p->index_map.data; |
1246 | const unsigned char *index = p->index_map.data; | |
a070f152 CMN |
1247 | unsigned hi, lo, stride; |
1248 | int pos, found = 0; | |
1249 | const unsigned char *current = 0; | |
1250 | ||
1251 | *offset_out = 0; | |
1252 | ||
0ddfcb40 | 1253 | if (p->index_version == -1) { |
34bd5999 | 1254 | int error; |
a070f152 | 1255 | |
34bd5999 CMN |
1256 | if ((error = pack_index_open(p)) < 0) |
1257 | return error; | |
1258 | assert(p->index_map.data); | |
8c535f3f | 1259 | |
34bd5999 CMN |
1260 | index = p->index_map.data; |
1261 | level1_ofs = p->index_map.data; | |
1262 | } | |
a070f152 CMN |
1263 | |
1264 | if (p->index_version > 1) { | |
1265 | level1_ofs += 2; | |
1266 | index += 8; | |
1267 | } | |
1268 | ||
1269 | index += 4 * 256; | |
1270 | hi = ntohl(level1_ofs[(int)short_oid->id[0]]); | |
1271 | lo = ((short_oid->id[0] == 0x0) ? 0 : ntohl(level1_ofs[(int)short_oid->id[0] - 1])); | |
1272 | ||
1273 | if (p->index_version > 1) { | |
1274 | stride = 20; | |
1275 | } else { | |
1276 | stride = 24; | |
1277 | index += 4; | |
1278 | } | |
1279 | ||
1280 | #ifdef INDEX_DEBUG_LOOKUP | |
1281 | printf("%02x%02x%02x... lo %u hi %u nr %d\n", | |
1282 | short_oid->id[0], short_oid->id[1], short_oid->id[2], lo, hi, p->num_objects); | |
1283 | #endif | |
1284 | ||
67591c8c | 1285 | #ifdef GIT_USE_LOOKUP |
87d9869f | 1286 | pos = sha1_entry_pos(index, stride, 0, lo, hi, p->num_objects, short_oid->id); |
67591c8c VM |
1287 | #else |
1288 | pos = sha1_position(index, stride, lo, hi, short_oid->id); | |
1289 | #endif | |
a070f152 CMN |
1290 | |
1291 | if (pos >= 0) { | |
1292 | /* An object matching exactly the oid was found */ | |
1293 | found = 1; | |
1294 | current = index + pos * stride; | |
1295 | } else { | |
1296 | /* No object was found */ | |
1297 | /* pos refers to the object with the "closest" oid to short_oid */ | |
1298 | pos = - 1 - pos; | |
1299 | if (pos < (int)p->num_objects) { | |
1300 | current = index + pos * stride; | |
1301 | ||
282283ac | 1302 | if (!git_oid_ncmp(short_oid, (const git_oid *)current, len)) |
a070f152 | 1303 | found = 1; |
a070f152 CMN |
1304 | } |
1305 | } | |
1306 | ||
b2a2702d | 1307 | if (found && len != GIT_OID_HEXSZ && pos + 1 < (int)p->num_objects) { |
a070f152 CMN |
1308 | /* Check for ambiguousity */ |
1309 | const unsigned char *next = current + stride; | |
1310 | ||
1311 | if (!git_oid_ncmp(short_oid, (const git_oid *)next, len)) { | |
1312 | found = 2; | |
1313 | } | |
1314 | } | |
1315 | ||
e1de726c | 1316 | if (!found) |
282283ac | 1317 | return git_odb__error_notfound("failed to find offset for pack entry", short_oid); |
e1de726c RB |
1318 | if (found > 1) |
1319 | return git_odb__error_ambiguous("found multiple offsets for pack entry"); | |
24c70804 | 1320 | |
e1de726c RB |
1321 | *offset_out = nth_packed_object_offset(p, pos); |
1322 | git_oid_fromraw(found_oid, current); | |
a070f152 CMN |
1323 | |
1324 | #ifdef INDEX_DEBUG_LOOKUP | |
e1de726c | 1325 | { |
a070f152 CMN |
1326 | unsigned char hex_sha1[GIT_OID_HEXSZ + 1]; |
1327 | git_oid_fmt(hex_sha1, found_oid); | |
1328 | hex_sha1[GIT_OID_HEXSZ] = '\0'; | |
1329 | printf("found lo=%d %s\n", lo, hex_sha1); | |
a070f152 | 1330 | } |
e1de726c | 1331 | #endif |
24c70804 | 1332 | |
e1de726c | 1333 | return 0; |
a070f152 CMN |
1334 | } |
1335 | ||
1336 | int git_pack_entry_find( | |
1337 | struct git_pack_entry *e, | |
1338 | struct git_pack_file *p, | |
1339 | const git_oid *short_oid, | |
b8457baa | 1340 | size_t len) |
a070f152 | 1341 | { |
e1de726c | 1342 | git_off_t offset; |
a070f152 CMN |
1343 | git_oid found_oid; |
1344 | int error; | |
1345 | ||
1346 | assert(p); | |
1347 | ||
1348 | if (len == GIT_OID_HEXSZ && p->num_bad_objects) { | |
1349 | unsigned i; | |
1350 | for (i = 0; i < p->num_bad_objects; i++) | |
b7f167da | 1351 | if (git_oid__cmp(short_oid, &p->bad_object_sha1[i]) == 0) |
e1de726c | 1352 | return packfile_error("bad object found in packfile"); |
a070f152 CMN |
1353 | } |
1354 | ||
1355 | error = pack_entry_find_offset(&offset, &found_oid, p, short_oid, len); | |
e1de726c RB |
1356 | if (error < 0) |
1357 | return error; | |
a070f152 CMN |
1358 | |
1359 | /* we found a unique entry in the index; | |
1360 | * make sure the packfile backing the index | |
1361 | * still exists on disk */ | |
e1de726c RB |
1362 | if (p->mwf.fd == -1 && (error = packfile_open(p)) < 0) |
1363 | return error; | |
a070f152 CMN |
1364 | |
1365 | e->offset = offset; | |
1366 | e->p = p; | |
1367 | ||
1368 | git_oid_cpy(&e->sha1, &found_oid); | |
e1de726c | 1369 | return 0; |
a070f152 | 1370 | } |