2 * Copyright (C) 2009-2011 the libgit2 contributors
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.
10 #include "git2/repository.h"
15 #include "delta-apply.h"
16 #include "sha1_lookup.h"
20 #include "git2/odb_backend.h"
23 git_odb_backend parent
;
25 struct git_pack_file
*last_found
;
27 time_t pack_folder_mtime
;
31 * The wonderful tale of a Packed Object lookup query
32 * ===================================================
33 * A riveting and epic story of epicness and ASCII
34 * art, presented by yours truly,
38 * Chapter 1: Once upon a time...
39 * Initialization of the Pack Backend
40 * --------------------------------------------------
42 * # git_odb_backend_pack
43 * | Creates the pack backend structure, initializes the
44 * | callback pointers to our default read() and exist() methods,
45 * | and tries to preload all the known packfiles in the ODB.
47 * |-# packfile_load_all
48 * | Tries to find the `pack` folder, if it exists. ODBs without
49 * | a pack folder are ignored altogether. If there's a `pack` folder
50 * | we run a `dirent` callback through every file in the pack folder
51 * | to find our packfiles. The packfiles are then sorted according
52 * | to a sorting callback.
54 * |-# packfile_load__cb
55 * | | This callback is called from `dirent` with every single file
56 * | | inside the pack folder. We find the packs by actually locating
57 * | | their index (ends in ".idx"). From that index, we verify that
58 * | | the corresponding packfile exists and is valid, and if so, we
59 * | | add it to the pack list.
61 * | |-# packfile_check
62 * | Make sure that there's a packfile to back this index, and store
63 * | some very basic information regarding the packfile itself,
64 * | such as the full path, the size, and the modification time.
65 * | We don't actually open the packfile to check for internal consistency.
67 * |-# packfile_sort__cb
68 * Sort all the preloaded packs according to some specific criteria:
69 * we prioritize the "newer" packs because it's more likely they
70 * contain the objects we are looking for, and we prioritize local
71 * packs over remote ones.
75 * Chapter 2: To be, or not to be...
76 * A standard packed `exist` query for an OID
77 * --------------------------------------------------
79 * # pack_backend__exists
80 * | Check if the given SHA1 oid exists in any of the packs
81 * | that have been loaded for our ODB.
84 * | Iterate through all the packs that have been preloaded
85 * | (starting by the pack where the latest object was found)
86 * | to try to find the OID in one of them.
88 * |-# pack_entry_find1
89 * | Check the index of an individual pack to see if the SHA1
90 * | OID can be found. If we can find the offset to that SHA1
91 * | inside of the index, that means the object is contained
92 * | inside of the packfile and we can stop searching.
93 * | Before returning, we verify that the packfile behing the
94 * | index we are searching still exists on disk.
96 * |-# pack_entry_find_offset
97 * | | Mmap the actual index file to disk if it hasn't been opened
98 * | | yet, and run a binary search through it to find the OID.
99 * | | See <http://book.git-scm.com/7_the_packfile.html> for specifics
100 * | | on the Packfile Index format and how do we find entries in it.
102 * | |-# pack_index_open
103 * | | Guess the name of the index based on the full path to the
104 * | | packfile, open it and verify its contents. Only if the index
105 * | | has not been opened already.
107 * | |-# pack_index_check
108 * | Mmap the index file and do a quick run through the header
109 * | to guess the index version (right now we support v1 and v2),
110 * | and to verify that the size of the index makes sense.
113 * See `packfile_open` in Chapter 3
117 * Chapter 3: The neverending story...
118 * A standard packed `lookup` query for an OID
119 * --------------------------------------------------
125 /***********************************************************
127 * FORWARD DECLARATIONS
129 ***********************************************************/
131 static void pack_window_free_all(struct pack_backend
*backend
, struct git_pack_file
*p
);
132 static int pack_window_contains(git_mwindow
*win
, off_t offset
);
134 static int packfile_sort__cb(const void *a_
, const void *b_
);
136 static int packfile_load__cb(void *_data
, char *path
);
137 static int packfile_refresh_all(struct pack_backend
*backend
);
139 static int pack_entry_find(struct git_pack_entry
*e
,
140 struct pack_backend
*backend
, const git_oid
*oid
);
142 /* Can find the offset of an object given
143 * a prefix of an identifier.
144 * Throws GIT_EAMBIGUOUSOIDPREFIX if short oid
146 * This method assumes that len is between
147 * GIT_OID_MINPREFIXLEN and GIT_OID_HEXSZ.
149 static int pack_entry_find_prefix(struct git_pack_entry
*e
,
150 struct pack_backend
*backend
,
151 const git_oid
*short_oid
,
156 /***********************************************************
158 * PACK WINDOW MANAGEMENT
160 ***********************************************************/
162 GIT_INLINE(void) pack_window_free_all(struct pack_backend
*GIT_UNUSED(backend
), struct git_pack_file
*p
)
164 GIT_UNUSED_ARG(backend
);
165 git_mwindow_free_all(&p
->mwf
);
168 GIT_INLINE(int) pack_window_contains(git_mwindow
*win
, off_t offset
)
170 /* We must promise at least 20 bytes (one hash) after the
171 * offset is available from this window, otherwise the offset
172 * is not actually in this window and a different window (which
173 * has that one hash excess) must be used. This is to support
174 * the object header and delta base parsing routines below.
176 return git_mwindow_contains(win
, offset
+ 20);
179 static int packfile_sort__cb(const void *a_
, const void *b_
)
181 const struct git_pack_file
*a
= a_
;
182 const struct git_pack_file
*b
= b_
;
186 * Local packs tend to contain objects specific to our
187 * variant of the project than remote ones. In addition,
188 * remote ones could be on a network mounted filesystem.
189 * Favor local ones for these reasons.
191 st
= a
->pack_local
- b
->pack_local
;
196 * Younger packs tend to contain more recent objects,
197 * and more recent objects tend to get accessed more
200 if (a
->mtime
< b
->mtime
)
202 else if (a
->mtime
== b
->mtime
)
210 static int packfile_load__cb(void *_data
, char *path
)
212 struct pack_backend
*backend
= (struct pack_backend
*)_data
;
213 struct git_pack_file
*pack
;
217 if (git__suffixcmp(path
, ".idx") != 0)
218 return GIT_SUCCESS
; /* not an index */
220 for (i
= 0; i
< backend
->packs
.length
; ++i
) {
221 struct git_pack_file
*p
= git_vector_get(&backend
->packs
, i
);
222 if (memcmp(p
->pack_name
, path
, strlen(path
) - strlen(".idx")) == 0)
226 error
= git_packfile_check(&pack
, path
);
227 if (error
== GIT_ENOTFOUND
) {
228 /* ignore missing .pack file as git does */
230 } else if (error
< GIT_SUCCESS
)
231 return git__rethrow(error
, "Failed to load packfile");
233 if (git_vector_insert(&backend
->packs
, pack
) < GIT_SUCCESS
) {
241 static int packfile_refresh_all(struct pack_backend
*backend
)
246 if (backend
->pack_folder
== NULL
)
249 if (p_stat(backend
->pack_folder
, &st
) < 0 || !S_ISDIR(st
.st_mode
))
250 return git__throw(GIT_ENOTFOUND
, "Failed to refresh packfiles. Backend not found");
252 if (st
.st_mtime
!= backend
->pack_folder_mtime
) {
253 char path
[GIT_PATH_MAX
];
254 strcpy(path
, backend
->pack_folder
);
256 /* reload all packs */
257 error
= git_futils_direach(path
, GIT_PATH_MAX
, packfile_load__cb
, (void *)backend
);
258 if (error
< GIT_SUCCESS
)
259 return git__rethrow(error
, "Failed to refresh packfiles");
261 git_vector_sort(&backend
->packs
);
262 backend
->pack_folder_mtime
= st
.st_mtime
;
268 static int pack_entry_find(struct git_pack_entry
*e
, struct pack_backend
*backend
, const git_oid
*oid
)
273 if ((error
= packfile_refresh_all(backend
)) < GIT_SUCCESS
)
274 return git__rethrow(error
, "Failed to find pack entry");
276 if (backend
->last_found
&&
277 git_pack_entry_find(e
, backend
->last_found
, oid
, GIT_OID_HEXSZ
) == GIT_SUCCESS
)
280 for (i
= 0; i
< backend
->packs
.length
; ++i
) {
281 struct git_pack_file
*p
;
283 p
= git_vector_get(&backend
->packs
, i
);
284 if (p
== backend
->last_found
)
287 if (git_pack_entry_find(e
, p
, oid
, GIT_OID_HEXSZ
) == GIT_SUCCESS
) {
288 backend
->last_found
= p
;
293 return git__throw(GIT_ENOTFOUND
, "Failed to find pack entry");
296 static int pack_entry_find_prefix(
297 struct git_pack_entry
*e
,
298 struct pack_backend
*backend
,
299 const git_oid
*short_oid
,
306 if ((error
= packfile_refresh_all(backend
)) < GIT_SUCCESS
)
307 return git__rethrow(error
, "Failed to find pack entry");
309 if (backend
->last_found
) {
310 error
= git_pack_entry_find(e
, backend
->last_found
, short_oid
, len
);
311 if (error
== GIT_EAMBIGUOUSOIDPREFIX
) {
312 return git__rethrow(error
, "Failed to find pack entry. Ambiguous sha1 prefix");
313 } else if (error
== GIT_SUCCESS
) {
318 for (i
= 0; i
< backend
->packs
.length
; ++i
) {
319 struct git_pack_file
*p
;
321 p
= git_vector_get(&backend
->packs
, i
);
322 if (p
== backend
->last_found
)
325 error
= git_pack_entry_find(e
, p
, short_oid
, len
);
326 if (error
== GIT_EAMBIGUOUSOIDPREFIX
) {
327 return git__rethrow(error
, "Failed to find pack entry. Ambiguous sha1 prefix");
328 } else if (error
== GIT_SUCCESS
) {
332 backend
->last_found
= p
;
337 return git__rethrow(GIT_ENOTFOUND
, "Failed to find pack entry");
338 } else if (found
> 1) {
339 return git__rethrow(GIT_EAMBIGUOUSOIDPREFIX
, "Failed to find pack entry. Ambiguous sha1 prefix");
347 /***********************************************************
349 * PACKED BACKEND PUBLIC API
351 * Implement the git_odb_backend API calls
353 ***********************************************************/
356 int pack_backend__read_header(git_rawobj *obj, git_odb_backend *backend, const git_oid *oid)
358 pack_location location;
360 assert(obj && backend && oid);
362 if (locate_packfile(&location, (struct pack_backend *)backend, oid) < 0)
363 return GIT_ENOTFOUND;
365 return read_header_packed(obj, &location);
369 static int pack_backend__read(void **buffer_p
, size_t *len_p
, git_otype
*type_p
, git_odb_backend
*backend
, const git_oid
*oid
)
371 struct git_pack_entry e
;
375 if ((error
= pack_entry_find(&e
, (struct pack_backend
*)backend
, oid
)) < GIT_SUCCESS
)
376 return git__rethrow(error
, "Failed to read pack backend");
378 if ((error
= git_packfile_unpack(&raw
, e
.p
, &e
.offset
)) < GIT_SUCCESS
)
379 return git__rethrow(error
, "Failed to read pack backend");
381 *buffer_p
= raw
.data
;
388 static int pack_backend__read_prefix(
393 git_odb_backend
*backend
,
394 const git_oid
*short_oid
,
397 if (len
< GIT_OID_MINPREFIXLEN
)
398 return git__throw(GIT_EAMBIGUOUSOIDPREFIX
, "Failed to read pack backend. Prefix length is lower than %d.", GIT_OID_MINPREFIXLEN
);
400 if (len
>= GIT_OID_HEXSZ
) {
401 /* We can fall back to regular read method */
402 int error
= pack_backend__read(buffer_p
, len_p
, type_p
, backend
, short_oid
);
403 if (error
== GIT_SUCCESS
)
404 git_oid_cpy(out_oid
, short_oid
);
408 struct git_pack_entry e
;
412 if ((error
= pack_entry_find_prefix(&e
, (struct pack_backend
*)backend
, short_oid
, len
)) < GIT_SUCCESS
)
413 return git__rethrow(error
, "Failed to read pack backend");
415 if ((error
= git_packfile_unpack(&raw
, e
.p
, &e
.offset
)) < GIT_SUCCESS
)
416 return git__rethrow(error
, "Failed to read pack backend");
418 *buffer_p
= raw
.data
;
421 git_oid_cpy(out_oid
, &e
.sha1
);
427 static int pack_backend__exists(git_odb_backend
*backend
, const git_oid
*oid
)
429 struct git_pack_entry e
;
430 return pack_entry_find(&e
, (struct pack_backend
*)backend
, oid
) == GIT_SUCCESS
;
433 static void pack_backend__free(git_odb_backend
*_backend
)
435 struct pack_backend
*backend
;
440 backend
= (struct pack_backend
*)_backend
;
442 for (i
= 0; i
< backend
->packs
.length
; ++i
) {
443 struct git_pack_file
*p
= git_vector_get(&backend
->packs
, i
);
447 git_vector_free(&backend
->packs
);
448 git__free(backend
->pack_folder
);
452 int git_odb_backend_pack(git_odb_backend
**backend_out
, const char *objects_dir
)
454 struct pack_backend
*backend
;
455 char path
[GIT_PATH_MAX
];
457 backend
= git__calloc(1, sizeof(struct pack_backend
));
461 if (git_vector_init(&backend
->packs
, 8, packfile_sort__cb
) < GIT_SUCCESS
) {
466 git_path_join(path
, objects_dir
, "pack");
467 if (git_futils_isdir(path
) == GIT_SUCCESS
) {
468 backend
->pack_folder
= git__strdup(path
);
469 backend
->pack_folder_mtime
= 0;
471 if (backend
->pack_folder
== NULL
) {
477 backend
->parent
.read
= &pack_backend__read
;
478 backend
->parent
.read_prefix
= &pack_backend__read_prefix
;
479 backend
->parent
.read_header
= NULL
;
480 backend
->parent
.exists
= &pack_backend__exists
;
481 backend
->parent
.free
= &pack_backend__free
;
483 *backend_out
= (git_odb_backend
*)backend
;