]>
Commit | Line | Data |
---|---|---|
7d0cdf82 | 1 | /* |
5e0de328 | 2 | * Copyright (C) 2009-2012 the libgit2 contributors |
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" | |
7d0cdf82 CMN |
15 | |
16 | #include "git2/oid.h" | |
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, | |
41 | unsigned int len); | |
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 | ||
a070f152 CMN |
49 | /*********************************************************** |
50 | * | |
51 | * PACK INDEX METHODS | |
52 | * | |
53 | ***********************************************************/ | |
54 | ||
55 | static void pack_index_free(struct git_pack_file *p) | |
56 | { | |
57 | if (p->index_map.data) { | |
58 | git_futils_mmap_free(&p->index_map); | |
59 | p->index_map.data = NULL; | |
60 | } | |
61 | } | |
62 | ||
87d9869f | 63 | static int pack_index_check(const char *path, struct git_pack_file *p) |
a070f152 CMN |
64 | { |
65 | struct git_pack_idx_header *hdr; | |
66 | uint32_t version, nr, i, *index; | |
a070f152 CMN |
67 | void *idx_map; |
68 | size_t idx_size; | |
a070f152 | 69 | struct stat st; |
a070f152 | 70 | int error; |
e1de726c RB |
71 | /* TODO: properly open the file without access time using O_NOATIME */ |
72 | git_file fd = git_futils_open_ro(path); | |
a070f152 | 73 | if (fd < 0) |
e1de726c | 74 | return fd; |
a070f152 | 75 | |
e1de726c RB |
76 | if (p_fstat(fd, &st) < 0 || |
77 | !S_ISREG(st.st_mode) || | |
78 | !git__is_sizet(st.st_size) || | |
79 | (idx_size = (size_t)st.st_size) < 4 * 256 + 20 + 20) | |
80 | { | |
a070f152 | 81 | p_close(fd); |
e1de726c RB |
82 | giterr_set(GITERR_OS, "Failed to check pack index."); |
83 | return -1; | |
a070f152 CMN |
84 | } |
85 | ||
86 | error = git_futils_mmap_ro(&p->index_map, fd, 0, idx_size); | |
e1de726c | 87 | |
a070f152 CMN |
88 | p_close(fd); |
89 | ||
e1de726c RB |
90 | if (error < 0) |
91 | return error; | |
a070f152 CMN |
92 | |
93 | hdr = idx_map = p->index_map.data; | |
94 | ||
95 | if (hdr->idx_signature == htonl(PACK_IDX_SIGNATURE)) { | |
96 | version = ntohl(hdr->idx_version); | |
97 | ||
98 | if (version < 2 || version > 2) { | |
99 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 100 | return packfile_error("unsupported index version"); |
a070f152 CMN |
101 | } |
102 | ||
103 | } else | |
104 | version = 1; | |
105 | ||
106 | nr = 0; | |
107 | index = idx_map; | |
108 | ||
109 | if (version > 1) | |
87d9869f | 110 | index += 2; /* skip index header */ |
a070f152 CMN |
111 | |
112 | for (i = 0; i < 256; i++) { | |
113 | uint32_t n = ntohl(index[i]); | |
114 | if (n < nr) { | |
115 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 116 | return packfile_error("index is non-monotonic"); |
a070f152 CMN |
117 | } |
118 | nr = n; | |
119 | } | |
120 | ||
121 | if (version == 1) { | |
122 | /* | |
123 | * Total size: | |
87d9869f VM |
124 | * - 256 index entries 4 bytes each |
125 | * - 24-byte entries * nr (20-byte sha1 + 4-byte offset) | |
126 | * - 20-byte SHA1 of the packfile | |
127 | * - 20-byte SHA1 file checksum | |
a070f152 CMN |
128 | */ |
129 | if (idx_size != 4*256 + nr * 24 + 20 + 20) { | |
130 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 131 | return packfile_error("index is corrupted"); |
a070f152 CMN |
132 | } |
133 | } else if (version == 2) { | |
134 | /* | |
135 | * Minimum size: | |
87d9869f VM |
136 | * - 8 bytes of header |
137 | * - 256 index entries 4 bytes each | |
138 | * - 20-byte sha1 entry * nr | |
139 | * - 4-byte crc entry * nr | |
140 | * - 4-byte offset entry * nr | |
141 | * - 20-byte SHA1 of the packfile | |
142 | * - 20-byte SHA1 file checksum | |
a070f152 CMN |
143 | * And after the 4-byte offset table might be a |
144 | * variable sized table containing 8-byte entries | |
145 | * for offsets larger than 2^31. | |
146 | */ | |
147 | unsigned long min_size = 8 + 4*256 + nr*(20 + 4 + 4) + 20 + 20; | |
148 | unsigned long max_size = min_size; | |
149 | ||
150 | if (nr) | |
151 | max_size += (nr - 1)*8; | |
152 | ||
153 | if (idx_size < min_size || idx_size > max_size) { | |
154 | git_futils_mmap_free(&p->index_map); | |
e1de726c | 155 | return packfile_error("wrong index size"); |
a070f152 | 156 | } |
a070f152 CMN |
157 | } |
158 | ||
159 | p->index_version = version; | |
160 | p->num_objects = nr; | |
e1de726c | 161 | return 0; |
a070f152 CMN |
162 | } |
163 | ||
164 | static int pack_index_open(struct git_pack_file *p) | |
165 | { | |
166 | char *idx_name; | |
167 | int error; | |
44ef8b1b | 168 | size_t name_len, offset; |
a070f152 CMN |
169 | |
170 | if (p->index_map.data) | |
e1de726c | 171 | return 0; |
a070f152 CMN |
172 | |
173 | idx_name = git__strdup(p->pack_name); | |
e1de726c RB |
174 | GITERR_CHECK_ALLOC(idx_name); |
175 | ||
44ef8b1b RB |
176 | name_len = strlen(idx_name); |
177 | offset = name_len - strlen(".pack"); | |
178 | assert(offset < name_len); /* make sure no underflow */ | |
179 | ||
180 | strncpy(idx_name + offset, ".idx", name_len - offset); | |
a070f152 CMN |
181 | |
182 | error = pack_index_check(idx_name, p); | |
3286c408 | 183 | git__free(idx_name); |
a070f152 | 184 | |
e1de726c | 185 | return error; |
a070f152 CMN |
186 | } |
187 | ||
188 | static unsigned char *pack_window_open( | |
189 | struct git_pack_file *p, | |
7d0cdf82 | 190 | git_mwindow **w_cursor, |
e1de726c | 191 | git_off_t offset, |
7d0cdf82 CMN |
192 | unsigned int *left) |
193 | { | |
e1de726c | 194 | if (p->mwf.fd == -1 && packfile_open(p) < 0) |
7d0cdf82 CMN |
195 | return NULL; |
196 | ||
197 | /* Since packfiles end in a hash of their content and it's | |
198 | * pointless to ask for an offset into the middle of that | |
199 | * hash, and the pack_window_contains function above wouldn't match | |
200 | * don't allow an offset too close to the end of the file. | |
201 | */ | |
202 | if (offset > (p->mwf.size - 20)) | |
203 | return NULL; | |
204 | ||
205 | return git_mwindow_open(&p->mwf, w_cursor, offset, 20, left); | |
206 | } | |
207 | ||
45d773ef CMN |
208 | static int packfile_unpack_header1( |
209 | unsigned long *usedp, | |
7d0cdf82 CMN |
210 | size_t *sizep, |
211 | git_otype *type, | |
212 | const unsigned char *buf, | |
213 | unsigned long len) | |
214 | { | |
215 | unsigned shift; | |
216 | unsigned long size, c; | |
217 | unsigned long used = 0; | |
2aeadb9c | 218 | |
7d0cdf82 CMN |
219 | c = buf[used++]; |
220 | *type = (c >> 4) & 7; | |
221 | size = c & 15; | |
222 | shift = 4; | |
223 | while (c & 0x80) { | |
45d773ef | 224 | if (len <= used) |
904b67e6 | 225 | return GIT_EBUFS; |
45d773ef CMN |
226 | |
227 | if (bitsizeof(long) <= shift) { | |
228 | *usedp = 0; | |
229 | return -1; | |
230 | } | |
7d0cdf82 CMN |
231 | |
232 | c = buf[used++]; | |
233 | size += (c & 0x7f) << shift; | |
234 | shift += 7; | |
235 | } | |
236 | ||
237 | *sizep = (size_t)size; | |
45d773ef CMN |
238 | *usedp = used; |
239 | return 0; | |
7d0cdf82 CMN |
240 | } |
241 | ||
242 | int git_packfile_unpack_header( | |
243 | size_t *size_p, | |
244 | git_otype *type_p, | |
245 | git_mwindow_file *mwf, | |
246 | git_mwindow **w_curs, | |
e1de726c | 247 | git_off_t *curpos) |
7d0cdf82 CMN |
248 | { |
249 | unsigned char *base; | |
250 | unsigned int left; | |
251 | unsigned long used; | |
45d773ef | 252 | int ret; |
7d0cdf82 CMN |
253 | |
254 | /* pack_window_open() assures us we have [base, base + 20) available | |
87d9869f VM |
255 | * as a range that we can look at at. (Its actually the hash |
256 | * size that is assured.) With our object header encoding | |
7d0cdf82 CMN |
257 | * the maximum deflated object size is 2^137, which is just |
258 | * insane, so we know won't exceed what we have been given. | |
259 | */ | |
260 | // base = pack_window_open(p, w_curs, *curpos, &left); | |
261 | base = git_mwindow_open(mwf, w_curs, *curpos, 20, &left); | |
262 | if (base == NULL) | |
904b67e6 | 263 | return GIT_EBUFS; |
2aeadb9c CY |
264 | |
265 | ret = packfile_unpack_header1(&used, size_p, type_p, base, left); | |
45d773ef | 266 | git_mwindow_close(w_curs); |
904b67e6 | 267 | if (ret == GIT_EBUFS) |
45d773ef CMN |
268 | return ret; |
269 | else if (ret < 0) | |
e1de726c | 270 | return packfile_error("header length is zero"); |
7d0cdf82 CMN |
271 | |
272 | *curpos += used; | |
e1de726c | 273 | return 0; |
7d0cdf82 CMN |
274 | } |
275 | ||
a070f152 | 276 | static int packfile_unpack_delta( |
7d0cdf82 | 277 | git_rawobj *obj, |
a070f152 | 278 | struct git_pack_file *p, |
7d0cdf82 | 279 | git_mwindow **w_curs, |
e1de726c | 280 | git_off_t *curpos, |
7d0cdf82 CMN |
281 | size_t delta_size, |
282 | git_otype delta_type, | |
e1de726c | 283 | git_off_t obj_offset) |
7d0cdf82 | 284 | { |
e1de726c | 285 | git_off_t base_offset; |
7d0cdf82 CMN |
286 | git_rawobj base, delta; |
287 | int error; | |
288 | ||
b5b474dd | 289 | base_offset = get_delta_base(p, w_curs, curpos, delta_type, obj_offset); |
45d773ef | 290 | git_mwindow_close(w_curs); |
7d0cdf82 | 291 | if (base_offset == 0) |
e1de726c RB |
292 | return packfile_error("delta offset is zero"); |
293 | if (base_offset < 0) /* must actually be an error code */ | |
294 | return (int)base_offset; | |
7d0cdf82 | 295 | |
b5b474dd | 296 | error = git_packfile_unpack(&base, p, &base_offset); |
7d0cdf82 CMN |
297 | |
298 | /* | |
299 | * TODO: git.git tries to load the base from other packfiles | |
300 | * or loose objects. | |
301 | * | |
302 | * We'll need to do this in order to support thin packs. | |
303 | */ | |
e1de726c RB |
304 | if (error < 0) |
305 | return error; | |
7d0cdf82 CMN |
306 | |
307 | error = packfile_unpack_compressed(&delta, p, w_curs, curpos, delta_size, delta_type); | |
45d773ef | 308 | git_mwindow_close(w_curs); |
e1de726c | 309 | if (error < 0) { |
3286c408 | 310 | git__free(base.data); |
e1de726c | 311 | return error; |
7d0cdf82 CMN |
312 | } |
313 | ||
314 | obj->type = base.type; | |
e1de726c | 315 | error = git__delta_apply(obj, base.data, base.len, delta.data, delta.len); |
7d0cdf82 | 316 | |
3286c408 VM |
317 | git__free(base.data); |
318 | git__free(delta.data); | |
7d0cdf82 CMN |
319 | |
320 | /* TODO: we might want to cache this shit. eventually */ | |
321 | //add_delta_base_cache(p, base_offset, base, base_size, *type); | |
e1de726c | 322 | |
7d0cdf82 CMN |
323 | return error; /* error set by git__delta_apply */ |
324 | } | |
325 | ||
a070f152 | 326 | int git_packfile_unpack( |
e1de726c RB |
327 | git_rawobj *obj, |
328 | struct git_pack_file *p, | |
329 | git_off_t *obj_offset) | |
7d0cdf82 CMN |
330 | { |
331 | git_mwindow *w_curs = NULL; | |
e1de726c | 332 | git_off_t curpos = *obj_offset; |
7d0cdf82 CMN |
333 | int error; |
334 | ||
335 | size_t size = 0; | |
336 | git_otype type; | |
337 | ||
338 | /* | |
339 | * TODO: optionally check the CRC on the packfile | |
340 | */ | |
341 | ||
342 | obj->data = NULL; | |
343 | obj->len = 0; | |
344 | obj->type = GIT_OBJ_BAD; | |
345 | ||
346 | error = git_packfile_unpack_header(&size, &type, &p->mwf, &w_curs, &curpos); | |
45d773ef CMN |
347 | git_mwindow_close(&w_curs); |
348 | ||
e1de726c RB |
349 | if (error < 0) |
350 | return error; | |
7d0cdf82 CMN |
351 | |
352 | switch (type) { | |
353 | case GIT_OBJ_OFS_DELTA: | |
354 | case GIT_OBJ_REF_DELTA: | |
355 | error = packfile_unpack_delta( | |
b5b474dd CMN |
356 | obj, p, &w_curs, &curpos, |
357 | size, type, *obj_offset); | |
7d0cdf82 CMN |
358 | break; |
359 | ||
360 | case GIT_OBJ_COMMIT: | |
361 | case GIT_OBJ_TREE: | |
362 | case GIT_OBJ_BLOB: | |
363 | case GIT_OBJ_TAG: | |
364 | error = packfile_unpack_compressed( | |
b5b474dd | 365 | obj, p, &w_curs, &curpos, |
7d0cdf82 CMN |
366 | size, type); |
367 | break; | |
368 | ||
369 | default: | |
e1de726c | 370 | error = packfile_error("invalid packfile type in header");; |
7d0cdf82 CMN |
371 | break; |
372 | } | |
373 | ||
b5b474dd | 374 | *obj_offset = curpos; |
e1de726c | 375 | return error; |
7d0cdf82 CMN |
376 | } |
377 | ||
282283ac RB |
378 | static void *use_git_alloc(void *opaq, unsigned int count, unsigned int size) |
379 | { | |
380 | GIT_UNUSED(opaq); | |
381 | return git__calloc(count, size); | |
382 | } | |
383 | ||
384 | static void use_git_free(void *opaq, void *ptr) | |
385 | { | |
386 | GIT_UNUSED(opaq); | |
387 | git__free(ptr); | |
388 | } | |
389 | ||
7d0cdf82 | 390 | int packfile_unpack_compressed( |
e1de726c RB |
391 | git_rawobj *obj, |
392 | struct git_pack_file *p, | |
393 | git_mwindow **w_curs, | |
394 | git_off_t *curpos, | |
395 | size_t size, | |
396 | git_otype type) | |
7d0cdf82 CMN |
397 | { |
398 | int st; | |
399 | z_stream stream; | |
400 | unsigned char *buffer, *in; | |
401 | ||
e1de726c RB |
402 | buffer = git__calloc(1, size + 1); |
403 | GITERR_CHECK_ALLOC(buffer); | |
7d0cdf82 CMN |
404 | |
405 | memset(&stream, 0, sizeof(stream)); | |
406 | stream.next_out = buffer; | |
1c3fac4d | 407 | stream.avail_out = (uInt)size + 1; |
282283ac RB |
408 | stream.zalloc = use_git_alloc; |
409 | stream.zfree = use_git_free; | |
7d0cdf82 CMN |
410 | |
411 | st = inflateInit(&stream); | |
412 | if (st != Z_OK) { | |
3286c408 | 413 | git__free(buffer); |
e1de726c | 414 | giterr_set(GITERR_ZLIB, "Failed to inflate packfile"); |
45d773ef | 415 | |
e1de726c | 416 | return -1; |
7d0cdf82 CMN |
417 | } |
418 | ||
419 | do { | |
b5b474dd | 420 | in = pack_window_open(p, w_curs, *curpos, &stream.avail_in); |
7d0cdf82 CMN |
421 | stream.next_in = in; |
422 | st = inflate(&stream, Z_FINISH); | |
45d773ef | 423 | git_mwindow_close(w_curs); |
7d0cdf82 CMN |
424 | |
425 | if (!stream.avail_out) | |
426 | break; /* the payload is larger than it should be */ | |
427 | ||
45d773ef CMN |
428 | if (st == Z_BUF_ERROR && in == NULL) { |
429 | inflateEnd(&stream); | |
430 | git__free(buffer); | |
904b67e6 | 431 | return GIT_EBUFS; |
45d773ef CMN |
432 | } |
433 | ||
b5b474dd | 434 | *curpos += stream.next_in - in; |
7d0cdf82 CMN |
435 | } while (st == Z_OK || st == Z_BUF_ERROR); |
436 | ||
437 | inflateEnd(&stream); | |
438 | ||
439 | if ((st != Z_STREAM_END) || stream.total_out != size) { | |
3286c408 | 440 | git__free(buffer); |
e1de726c RB |
441 | giterr_set(GITERR_ZLIB, "Failed to inflate packfile"); |
442 | return -1; | |
7d0cdf82 CMN |
443 | } |
444 | ||
445 | obj->type = type; | |
446 | obj->len = size; | |
447 | obj->data = buffer; | |
e1de726c | 448 | return 0; |
7d0cdf82 CMN |
449 | } |
450 | ||
b5b474dd CMN |
451 | /* |
452 | * curpos is where the data starts, delta_obj_offset is the where the | |
453 | * header starts | |
454 | */ | |
e1de726c RB |
455 | git_off_t get_delta_base( |
456 | struct git_pack_file *p, | |
457 | git_mwindow **w_curs, | |
458 | git_off_t *curpos, | |
459 | git_otype type, | |
460 | git_off_t delta_obj_offset) | |
7d0cdf82 | 461 | { |
45d773ef CMN |
462 | unsigned int left = 0; |
463 | unsigned char *base_info; | |
e1de726c | 464 | git_off_t base_offset; |
7d0cdf82 CMN |
465 | git_oid unused; |
466 | ||
45d773ef CMN |
467 | base_info = pack_window_open(p, w_curs, *curpos, &left); |
468 | /* Assumption: the only reason this would fail is because the file is too small */ | |
469 | if (base_info == NULL) | |
904b67e6 | 470 | return GIT_EBUFS; |
7d0cdf82 CMN |
471 | /* pack_window_open() assured us we have [base_info, base_info + 20) |
472 | * as a range that we can look at without walking off the | |
87d9869f VM |
473 | * end of the mapped window. Its actually the hash size |
474 | * that is assured. An OFS_DELTA longer than the hash size | |
7d0cdf82 CMN |
475 | * is stupid, as then a REF_DELTA would be smaller to store. |
476 | */ | |
477 | if (type == GIT_OBJ_OFS_DELTA) { | |
478 | unsigned used = 0; | |
479 | unsigned char c = base_info[used++]; | |
480 | base_offset = c & 127; | |
481 | while (c & 128) { | |
45d773ef | 482 | if (left <= used) |
904b67e6 | 483 | return GIT_EBUFS; |
7d0cdf82 CMN |
484 | base_offset += 1; |
485 | if (!base_offset || MSB(base_offset, 7)) | |
87d9869f | 486 | return 0; /* overflow */ |
7d0cdf82 CMN |
487 | c = base_info[used++]; |
488 | base_offset = (base_offset << 7) + (c & 127); | |
489 | } | |
490 | base_offset = delta_obj_offset - base_offset; | |
491 | if (base_offset <= 0 || base_offset >= delta_obj_offset) | |
87d9869f | 492 | return 0; /* out of bound */ |
7d0cdf82 CMN |
493 | *curpos += used; |
494 | } else if (type == GIT_OBJ_REF_DELTA) { | |
c1af5a39 CMN |
495 | /* If we have the cooperative cache, search in it first */ |
496 | if (p->has_cache) { | |
497 | int pos; | |
498 | struct git_pack_entry key; | |
499 | ||
500 | git_oid_fromraw(&key.sha1, base_info); | |
501 | pos = git_vector_bsearch(&p->cache, &key); | |
502 | if (pos >= 0) { | |
503 | *curpos += 20; | |
504 | return ((struct git_pack_entry *)git_vector_get(&p->cache, pos))->offset; | |
505 | } | |
506 | } | |
7d0cdf82 | 507 | /* The base entry _must_ be in the same pack */ |
e1de726c RB |
508 | if (pack_entry_find_offset(&base_offset, &unused, p, (git_oid *)base_info, GIT_OID_HEXSZ) < 0) |
509 | return packfile_error("base entry delta is not in the same pack"); | |
7d0cdf82 CMN |
510 | *curpos += 20; |
511 | } else | |
512 | return 0; | |
513 | ||
514 | return base_offset; | |
515 | } | |
a070f152 CMN |
516 | |
517 | /*********************************************************** | |
518 | * | |
519 | * PACKFILE METHODS | |
520 | * | |
521 | ***********************************************************/ | |
522 | ||
44ef8b1b | 523 | static struct git_pack_file *packfile_alloc(size_t extra) |
a070f152 | 524 | { |
e1de726c RB |
525 | struct git_pack_file *p = git__calloc(1, sizeof(*p) + extra); |
526 | if (p != NULL) | |
527 | p->mwf.fd = -1; | |
a070f152 CMN |
528 | return p; |
529 | } | |
530 | ||
531 | ||
532 | void packfile_free(struct git_pack_file *p) | |
533 | { | |
534 | assert(p); | |
535 | ||
536 | /* clear_delta_base_cache(); */ | |
537 | git_mwindow_free_all(&p->mwf); | |
1d8943c6 | 538 | git_mwindow_file_deregister(&p->mwf); |
a070f152 CMN |
539 | |
540 | if (p->mwf.fd != -1) | |
541 | p_close(p->mwf.fd); | |
542 | ||
543 | pack_index_free(p); | |
544 | ||
3286c408 VM |
545 | git__free(p->bad_object_sha1); |
546 | git__free(p); | |
a070f152 CMN |
547 | } |
548 | ||
549 | static int packfile_open(struct git_pack_file *p) | |
550 | { | |
551 | struct stat st; | |
552 | struct git_pack_header hdr; | |
553 | git_oid sha1; | |
554 | unsigned char *idx_sha1; | |
555 | ||
e1de726c RB |
556 | assert(p->index_map.data); |
557 | ||
0d0fa7c3 | 558 | if (!p->index_map.data && pack_index_open(p) < 0) |
282283ac | 559 | return git_odb__error_notfound("failed to open packfile", NULL); |
a070f152 CMN |
560 | |
561 | /* TODO: open with noatime */ | |
e1de726c RB |
562 | p->mwf.fd = git_futils_open_ro(p->pack_name); |
563 | if (p->mwf.fd < 0) | |
564 | return p->mwf.fd; | |
a070f152 | 565 | |
e1de726c RB |
566 | if (p_fstat(p->mwf.fd, &st) < 0 || |
567 | git_mwindow_file_register(&p->mwf) < 0) | |
568 | goto cleanup; | |
a070f152 CMN |
569 | |
570 | /* If we created the struct before we had the pack we lack size. */ | |
571 | if (!p->mwf.size) { | |
572 | if (!S_ISREG(st.st_mode)) | |
573 | goto cleanup; | |
e1de726c | 574 | p->mwf.size = (git_off_t)st.st_size; |
a070f152 CMN |
575 | } else if (p->mwf.size != st.st_size) |
576 | goto cleanup; | |
577 | ||
578 | #if 0 | |
579 | /* We leave these file descriptors open with sliding mmap; | |
580 | * there is no point keeping them open across exec(), though. | |
581 | */ | |
582 | fd_flag = fcntl(p->mwf.fd, F_GETFD, 0); | |
583 | if (fd_flag < 0) | |
e1de726c | 584 | goto cleanup; |
a070f152 CMN |
585 | |
586 | fd_flag |= FD_CLOEXEC; | |
587 | if (fcntl(p->pack_fd, F_SETFD, fd_flag) == -1) | |
e1de726c | 588 | goto cleanup; |
a070f152 CMN |
589 | #endif |
590 | ||
591 | /* Verify we recognize this pack file format. */ | |
e1de726c RB |
592 | if (p_read(p->mwf.fd, &hdr, sizeof(hdr)) < 0 || |
593 | hdr.hdr_signature != htonl(PACK_SIGNATURE) || | |
594 | !pack_version_ok(hdr.hdr_version)) | |
a070f152 CMN |
595 | goto cleanup; |
596 | ||
597 | /* Verify the pack matches its index. */ | |
e1de726c RB |
598 | if (p->num_objects != ntohl(hdr.hdr_entries) || |
599 | p_lseek(p->mwf.fd, p->mwf.size - GIT_OID_RAWSZ, SEEK_SET) == -1 || | |
600 | p_read(p->mwf.fd, sha1.id, GIT_OID_RAWSZ) < 0) | |
a070f152 CMN |
601 | goto cleanup; |
602 | ||
603 | idx_sha1 = ((unsigned char *)p->index_map.data) + p->index_map.len - 40; | |
604 | ||
e1de726c RB |
605 | if (git_oid_cmp(&sha1, (git_oid *)idx_sha1) == 0) |
606 | return 0; | |
a070f152 CMN |
607 | |
608 | cleanup: | |
e1de726c | 609 | giterr_set(GITERR_OS, "Invalid packfile '%s'", p->pack_name); |
a070f152 CMN |
610 | p_close(p->mwf.fd); |
611 | p->mwf.fd = -1; | |
e1de726c | 612 | return -1; |
a070f152 CMN |
613 | } |
614 | ||
615 | int git_packfile_check(struct git_pack_file **pack_out, const char *path) | |
616 | { | |
617 | struct stat st; | |
618 | struct git_pack_file *p; | |
619 | size_t path_len; | |
620 | ||
621 | *pack_out = NULL; | |
622 | path_len = strlen(path); | |
623 | p = packfile_alloc(path_len + 2); | |
e1de726c | 624 | GITERR_CHECK_ALLOC(p); |
a070f152 CMN |
625 | |
626 | /* | |
627 | * Make sure a corresponding .pack file exists and that | |
628 | * the index looks sane. | |
629 | */ | |
932669b8 | 630 | path_len -= strlen(".idx"); |
a070f152 | 631 | if (path_len < 1) { |
3286c408 | 632 | git__free(p); |
282283ac | 633 | return git_odb__error_notfound("invalid packfile path", NULL); |
a070f152 CMN |
634 | } |
635 | ||
636 | memcpy(p->pack_name, path, path_len); | |
637 | ||
638 | strcpy(p->pack_name + path_len, ".keep"); | |
1a481123 | 639 | if (git_path_exists(p->pack_name) == true) |
a070f152 CMN |
640 | p->pack_keep = 1; |
641 | ||
642 | strcpy(p->pack_name + path_len, ".pack"); | |
e1de726c | 643 | if (p_stat(p->pack_name, &st) < 0 || !S_ISREG(st.st_mode)) { |
3286c408 | 644 | git__free(p); |
282283ac | 645 | return git_odb__error_notfound("packfile not found", NULL); |
a070f152 CMN |
646 | } |
647 | ||
648 | /* ok, it looks sane as far as we can check without | |
649 | * actually mapping the pack file. | |
650 | */ | |
1af56d7d | 651 | p->mwf.size = st.st_size; |
a070f152 CMN |
652 | p->pack_local = 1; |
653 | p->mtime = (git_time_t)st.st_mtime; | |
654 | ||
655 | /* see if we can parse the sha1 oid in the packfile name */ | |
656 | if (path_len < 40 || | |
e1de726c | 657 | git_oid_fromstr(&p->sha1, path + path_len - GIT_OID_HEXSZ) < 0) |
a070f152 CMN |
658 | memset(&p->sha1, 0x0, GIT_OID_RAWSZ); |
659 | ||
660 | *pack_out = p; | |
e1de726c RB |
661 | |
662 | return 0; | |
a070f152 CMN |
663 | } |
664 | ||
665 | /*********************************************************** | |
666 | * | |
667 | * PACKFILE ENTRY SEARCH INTERNALS | |
668 | * | |
669 | ***********************************************************/ | |
670 | ||
e1de726c | 671 | static git_off_t nth_packed_object_offset(const struct git_pack_file *p, uint32_t n) |
a070f152 CMN |
672 | { |
673 | const unsigned char *index = p->index_map.data; | |
674 | index += 4 * 256; | |
675 | if (p->index_version == 1) { | |
676 | return ntohl(*((uint32_t *)(index + 24 * n))); | |
677 | } else { | |
678 | uint32_t off; | |
679 | index += 8 + p->num_objects * (20 + 4); | |
680 | off = ntohl(*((uint32_t *)(index + 4 * n))); | |
681 | if (!(off & 0x80000000)) | |
682 | return off; | |
683 | index += p->num_objects * 4 + (off & 0x7fffffff) * 8; | |
684 | return (((uint64_t)ntohl(*((uint32_t *)(index + 0)))) << 32) | | |
87d9869f | 685 | ntohl(*((uint32_t *)(index + 4))); |
a070f152 CMN |
686 | } |
687 | } | |
688 | ||
689 | static int pack_entry_find_offset( | |
e1de726c RB |
690 | git_off_t *offset_out, |
691 | git_oid *found_oid, | |
692 | struct git_pack_file *p, | |
693 | const git_oid *short_oid, | |
694 | unsigned int len) | |
a070f152 CMN |
695 | { |
696 | const uint32_t *level1_ofs = p->index_map.data; | |
697 | const unsigned char *index = p->index_map.data; | |
698 | unsigned hi, lo, stride; | |
699 | int pos, found = 0; | |
700 | const unsigned char *current = 0; | |
701 | ||
702 | *offset_out = 0; | |
703 | ||
704 | if (index == NULL) { | |
705 | int error; | |
706 | ||
e1de726c RB |
707 | if ((error = pack_index_open(p)) < 0) |
708 | return error; | |
a070f152 CMN |
709 | |
710 | assert(p->index_map.data); | |
711 | ||
712 | index = p->index_map.data; | |
713 | level1_ofs = p->index_map.data; | |
714 | } | |
715 | ||
716 | if (p->index_version > 1) { | |
717 | level1_ofs += 2; | |
718 | index += 8; | |
719 | } | |
720 | ||
721 | index += 4 * 256; | |
722 | hi = ntohl(level1_ofs[(int)short_oid->id[0]]); | |
723 | lo = ((short_oid->id[0] == 0x0) ? 0 : ntohl(level1_ofs[(int)short_oid->id[0] - 1])); | |
724 | ||
725 | if (p->index_version > 1) { | |
726 | stride = 20; | |
727 | } else { | |
728 | stride = 24; | |
729 | index += 4; | |
730 | } | |
731 | ||
732 | #ifdef INDEX_DEBUG_LOOKUP | |
733 | printf("%02x%02x%02x... lo %u hi %u nr %d\n", | |
734 | short_oid->id[0], short_oid->id[1], short_oid->id[2], lo, hi, p->num_objects); | |
735 | #endif | |
736 | ||
737 | /* Use git.git lookup code */ | |
87d9869f | 738 | pos = sha1_entry_pos(index, stride, 0, lo, hi, p->num_objects, short_oid->id); |
a070f152 CMN |
739 | |
740 | if (pos >= 0) { | |
741 | /* An object matching exactly the oid was found */ | |
742 | found = 1; | |
743 | current = index + pos * stride; | |
744 | } else { | |
745 | /* No object was found */ | |
746 | /* pos refers to the object with the "closest" oid to short_oid */ | |
747 | pos = - 1 - pos; | |
748 | if (pos < (int)p->num_objects) { | |
749 | current = index + pos * stride; | |
750 | ||
282283ac | 751 | if (!git_oid_ncmp(short_oid, (const git_oid *)current, len)) |
a070f152 | 752 | found = 1; |
a070f152 CMN |
753 | } |
754 | } | |
755 | ||
b2a2702d | 756 | if (found && len != GIT_OID_HEXSZ && pos + 1 < (int)p->num_objects) { |
a070f152 CMN |
757 | /* Check for ambiguousity */ |
758 | const unsigned char *next = current + stride; | |
759 | ||
760 | if (!git_oid_ncmp(short_oid, (const git_oid *)next, len)) { | |
761 | found = 2; | |
762 | } | |
763 | } | |
764 | ||
e1de726c | 765 | if (!found) |
282283ac | 766 | return git_odb__error_notfound("failed to find offset for pack entry", short_oid); |
e1de726c RB |
767 | if (found > 1) |
768 | return git_odb__error_ambiguous("found multiple offsets for pack entry"); | |
769 | *offset_out = nth_packed_object_offset(p, pos); | |
770 | git_oid_fromraw(found_oid, current); | |
a070f152 CMN |
771 | |
772 | #ifdef INDEX_DEBUG_LOOKUP | |
e1de726c | 773 | { |
a070f152 CMN |
774 | unsigned char hex_sha1[GIT_OID_HEXSZ + 1]; |
775 | git_oid_fmt(hex_sha1, found_oid); | |
776 | hex_sha1[GIT_OID_HEXSZ] = '\0'; | |
777 | printf("found lo=%d %s\n", lo, hex_sha1); | |
a070f152 | 778 | } |
e1de726c RB |
779 | #endif |
780 | return 0; | |
a070f152 CMN |
781 | } |
782 | ||
783 | int git_pack_entry_find( | |
784 | struct git_pack_entry *e, | |
785 | struct git_pack_file *p, | |
786 | const git_oid *short_oid, | |
787 | unsigned int len) | |
788 | { | |
e1de726c | 789 | git_off_t offset; |
a070f152 CMN |
790 | git_oid found_oid; |
791 | int error; | |
792 | ||
793 | assert(p); | |
794 | ||
795 | if (len == GIT_OID_HEXSZ && p->num_bad_objects) { | |
796 | unsigned i; | |
797 | for (i = 0; i < p->num_bad_objects; i++) | |
798 | if (git_oid_cmp(short_oid, &p->bad_object_sha1[i]) == 0) | |
e1de726c | 799 | return packfile_error("bad object found in packfile"); |
a070f152 CMN |
800 | } |
801 | ||
802 | error = pack_entry_find_offset(&offset, &found_oid, p, short_oid, len); | |
e1de726c RB |
803 | if (error < 0) |
804 | return error; | |
a070f152 CMN |
805 | |
806 | /* we found a unique entry in the index; | |
807 | * make sure the packfile backing the index | |
808 | * still exists on disk */ | |
e1de726c RB |
809 | if (p->mwf.fd == -1 && (error = packfile_open(p)) < 0) |
810 | return error; | |
a070f152 CMN |
811 | |
812 | e->offset = offset; | |
813 | e->p = p; | |
814 | ||
815 | git_oid_cpy(&e->sha1, &found_oid); | |
e1de726c | 816 | return 0; |
a070f152 | 817 | } |