]>
Commit | Line | Data |
---|---|---|
48a8c26c | 1 | Git index format |
8c7d0517 NTND |
2 | ================ |
3 | ||
2de9b711 | 4 | == The Git index file has the following format |
8c7d0517 | 5 | |
123712ba MÅ |
6 | All binary numbers are in network byte order. |
7 | In a repository using the traditional SHA-1, checksums and object IDs | |
8 | (object names) mentioned below are all computed using SHA-1. Similarly, | |
9 | in SHA-256 repositories, these values are computed using SHA-256. | |
10 | Version 2 is described here unless stated otherwise. | |
8c7d0517 NTND |
11 | |
12 | - A 12-byte header consisting of | |
13 | ||
14 | 4-byte signature: | |
23fcc98f | 15 | The signature is { 'D', 'I', 'R', 'C' } (stands for "dircache") |
8c7d0517 NTND |
16 | |
17 | 4-byte version number: | |
300e39f6 | 18 | The current supported versions are 2, 3 and 4. |
8c7d0517 NTND |
19 | |
20 | 32-bit number of index entries. | |
21 | ||
23fcc98f | 22 | - A number of sorted index entries (see below). |
8c7d0517 NTND |
23 | |
24 | - Extensions | |
25 | ||
26 | Extensions are identified by signature. Optional extensions can | |
48a8c26c | 27 | be ignored if Git does not understand them. |
8c7d0517 | 28 | |
845d15d4 | 29 | Git currently supports cache tree and resolve undo extensions. |
8c7d0517 NTND |
30 | |
31 | 4-byte extension signature. If the first byte is 'A'..'Z' the | |
32 | extension is optional and can be ignored. | |
33 | ||
34 | 32-bit size of the extension | |
35 | ||
36 | Extension data | |
37 | ||
123712ba | 38 | - Hash checksum over the content of the index file before this checksum. |
8c7d0517 NTND |
39 | |
40 | == Index entry | |
41 | ||
42 | Index entries are sorted in ascending order on the name field, | |
23fcc98f JH |
43 | interpreted as a string of unsigned bytes (i.e. memcmp() order, no |
44 | localization, no special casing of directory separator '/'). Entries | |
45 | with the same name are sorted by their stage field. | |
8c7d0517 NTND |
46 | |
47 | 32-bit ctime seconds, the last time a file's metadata changed | |
48 | this is stat(2) data | |
49 | ||
50 | 32-bit ctime nanosecond fractions | |
51 | this is stat(2) data | |
52 | ||
53 | 32-bit mtime seconds, the last time a file's data changed | |
54 | this is stat(2) data | |
55 | ||
56 | 32-bit mtime nanosecond fractions | |
57 | this is stat(2) data | |
58 | ||
59 | 32-bit dev | |
60 | this is stat(2) data | |
61 | ||
62 | 32-bit ino | |
63 | this is stat(2) data | |
64 | ||
65 | 32-bit mode, split into (high to low bits) | |
66 | ||
67 | 4-bit object type | |
23fcc98f | 68 | valid values in binary are 1000 (regular file), 1010 (symbolic link) |
8c7d0517 NTND |
69 | and 1110 (gitlink) |
70 | ||
71 | 3-bit unused | |
72 | ||
23fcc98f JH |
73 | 9-bit unix permission. Only 0755 and 0644 are valid for regular files. |
74 | Symbolic links and gitlinks have value 0 in this field. | |
8c7d0517 NTND |
75 | |
76 | 32-bit uid | |
77 | this is stat(2) data | |
78 | ||
79 | 32-bit gid | |
80 | this is stat(2) data | |
81 | ||
82 | 32-bit file size | |
23fcc98f | 83 | This is the on-disk size from stat(2), truncated to 32-bit. |
8c7d0517 | 84 | |
123712ba | 85 | Object name for the represented object |
8c7d0517 | 86 | |
23fcc98f | 87 | A 16-bit 'flags' field split into (high to low bits) |
8c7d0517 NTND |
88 | |
89 | 1-bit assume-valid flag | |
90 | ||
91 | 1-bit extended flag (must be zero in version 2) | |
92 | ||
93 | 2-bit stage (during merge) | |
94 | ||
23fcc98f JH |
95 | 12-bit name length if the length is less than 0xFFF; otherwise 0xFFF |
96 | is stored in this field. | |
8c7d0517 | 97 | |
300e39f6 NTND |
98 | (Version 3 or later) A 16-bit field, only applicable if the |
99 | "extended flag" above is 1, split into (high to low bits). | |
8c7d0517 NTND |
100 | |
101 | 1-bit reserved for future | |
102 | ||
103 | 1-bit skip-worktree flag (used by sparse checkout) | |
104 | ||
105 | 1-bit intent-to-add flag (used by "git add -N") | |
106 | ||
107 | 13-bit unused, must be zero | |
108 | ||
109 | Entry path name (variable length) relative to top level directory | |
110 | (without leading slash). '/' is used as path separator. The special | |
23fcc98f | 111 | path components ".", ".." and ".git" (without quotes) are disallowed. |
8c7d0517 NTND |
112 | Trailing slash is also disallowed. |
113 | ||
114 | The exact encoding is undefined, but the '.' and '/' characters | |
23fcc98f JH |
115 | are encoded in 7-bit ASCII and the encoding cannot contain a NUL |
116 | byte (iow, this is a UNIX pathname). | |
8c7d0517 | 117 | |
afd7bd22 JH |
118 | (Version 4) In version 4, the entry path name is prefix-compressed |
119 | relative to the path name for the previous entry (the very first | |
120 | entry is encoded as if the path name for the previous entry is an | |
121 | empty string). At the beginning of an entry, an integer N in the | |
122 | variable width encoding (the same encoding as the offset is encoded | |
123 | for OFS_DELTA pack entries; see pack-format.txt) is stored, followed | |
124 | by a NUL-terminated string S. Removing N bytes from the end of the | |
125 | path name for the previous entry, and replacing it with the string S | |
126 | yields the path name for this entry. | |
127 | ||
8c7d0517 NTND |
128 | 1-8 nul bytes as necessary to pad the entry to a multiple of eight bytes |
129 | while keeping the name NUL-terminated. | |
130 | ||
afd7bd22 JH |
131 | (Version 4) In version 4, the padding after the pathname does not |
132 | exist. | |
133 | ||
5fc2fc8f NTND |
134 | Interpretation of index entries in split index mode is completely |
135 | different. See below for details. | |
136 | ||
8c7d0517 NTND |
137 | == Extensions |
138 | ||
845d15d4 | 139 | === Cache tree |
8c7d0517 | 140 | |
22ad8600 DS |
141 | Since the index does not record entries for directories, the cache |
142 | entries cannot describe tree objects that already exist in the object | |
143 | database for regions of the index that are unchanged from an existing | |
144 | commit. The cache tree extension stores a recursive tree structure that | |
145 | describes the trees that already exist and completely match sections of | |
146 | the cache entries. This speeds up tree object generation from the index | |
147 | for a new commit by only computing the trees that are "new" to that | |
148 | commit. It also assists when comparing the index to another tree, such | |
149 | as `HEAD^{tree}`, since sections of the index can be skipped when a tree | |
150 | comparison demonstrates equality. | |
151 | ||
152 | The recursive tree structure uses nodes that store a number of cache | |
153 | entries, a list of subnodes, and an object ID (OID). The OID references | |
154 | the existing tree for that node, if it is known to exist. The subnodes | |
155 | correspond to subdirectories that themselves have cache tree nodes. The | |
156 | number of cache entries corresponds to the number of cache entries in | |
157 | the index that describe paths within that tree's directory. | |
158 | ||
159 | The extension tracks the full directory structure in the cache tree | |
160 | extension, but this is generally smaller than the full cache entry list. | |
161 | ||
162 | When a path is updated in index, Git invalidates all nodes of the | |
163 | recursive cache tree corresponding to the parent directories of that | |
164 | path. We store these tree nodes as being "invalid" by using "-1" as the | |
165 | number of cache entries. Invalid nodes still store a span of index | |
166 | entries, allowing Git to focus its efforts when reconstructing a full | |
167 | cache tree. | |
8c7d0517 | 168 | |
23fcc98f | 169 | The signature for this extension is { 'T', 'R', 'E', 'E' }. |
8c7d0517 | 170 | |
23fcc98f JH |
171 | A series of entries fill the entire extension; each of which |
172 | consists of: | |
8c7d0517 | 173 | |
23fcc98f | 174 | - NUL-terminated path component (relative to its parent directory); |
8c7d0517 | 175 | |
23fcc98f JH |
176 | - ASCII decimal number of entries in the index that is covered by the |
177 | tree this entry represents (entry_count); | |
8c7d0517 | 178 | |
23fcc98f | 179 | - A space (ASCII 32); |
8c7d0517 | 180 | |
23fcc98f JH |
181 | - ASCII decimal number that represents the number of subtrees this |
182 | tree has; | |
8c7d0517 | 183 | |
23fcc98f JH |
184 | - A newline (ASCII 10); and |
185 | ||
123712ba MÅ |
186 | - Object name for the object that would result from writing this span |
187 | of index as a tree. | |
23fcc98f | 188 | |
e44b6df9 | 189 | An entry can be in an invalidated state and is represented by having |
4a6385fe NTND |
190 | a negative number in the entry_count field. In this case, there is no |
191 | object name and the next entry starts immediately after the newline. | |
192 | When writing an invalid entry, -1 should always be used as entry_count. | |
23fcc98f JH |
193 | |
194 | The entries are written out in the top-down, depth-first order. The | |
195 | first entry represents the root level of the repository, followed by the | |
3b19dba7 | 196 | first subtree--let's call this A--of the root level (with its name |
23fcc98f | 197 | relative to the root level), followed by the first subtree of A (with |
4bdde337 DS |
198 | its name relative to A), and so on. The specified number of subtrees |
199 | indicates when the current level of the recursive stack is complete. | |
8c7d0517 NTND |
200 | |
201 | === Resolve undo | |
202 | ||
23fcc98f | 203 | A conflict is represented in the index as a set of higher stage entries. |
8c7d0517 | 204 | When a conflict is resolved (e.g. with "git add path"), these higher |
17b83d71 | 205 | stage entries will be removed and a stage-0 entry with proper resolution |
23fcc98f | 206 | is added. |
8c7d0517 | 207 | |
23fcc98f JH |
208 | When these higher stage entries are removed, they are saved in the |
209 | resolve undo extension, so that conflicts can be recreated (e.g. with | |
210 | "git checkout -m"), in case users want to redo a conflict resolution | |
211 | from scratch. | |
8c7d0517 | 212 | |
23fcc98f | 213 | The signature for this extension is { 'R', 'E', 'U', 'C' }. |
8c7d0517 | 214 | |
23fcc98f JH |
215 | A series of entries fill the entire extension; each of which |
216 | consists of: | |
8c7d0517 | 217 | |
23fcc98f JH |
218 | - NUL-terminated pathname the entry describes (relative to the root of |
219 | the repository, i.e. full pathname); | |
8c7d0517 | 220 | |
23fcc98f JH |
221 | - Three NUL-terminated ASCII octal numbers, entry mode of entries in |
222 | stage 1 to 3 (a missing stage is represented by "0" in this field); | |
223 | and | |
8c7d0517 | 224 | |
123712ba | 225 | - At most three object names of the entry in stages from 1 to 3 |
23fcc98f | 226 | (nothing is written for a missing stage). |
8c7d0517 | 227 | |
5fc2fc8f NTND |
228 | === Split index |
229 | ||
230 | In split index mode, the majority of index entries could be stored | |
231 | in a separate file. This extension records the changes to be made on | |
232 | top of that to produce the final index. | |
233 | ||
f2667a83 | 234 | The signature for this extension is { 'l', 'i', 'n', 'k' }. |
5fc2fc8f NTND |
235 | |
236 | The extension consists of: | |
237 | ||
123712ba MÅ |
238 | - Hash of the shared index file. The shared index file path |
239 | is $GIT_DIR/sharedindex.<hash>. If all bits are zero, the | |
5fc2fc8f NTND |
240 | index does not require a shared index file. |
241 | ||
242 | - An ewah-encoded delete bitmap, each bit represents an entry in the | |
243 | shared index. If a bit is set, its corresponding entry in the | |
244 | shared index will be removed from the final index. Note, because | |
245 | a delete operation changes index entry positions, but we do need | |
246 | original positions in replace phase, it's best to just mark | |
247 | entries for removal, then do a mass deletion after replacement. | |
248 | ||
249 | - An ewah-encoded replace bitmap, each bit represents an entry in | |
250 | the shared index. If a bit is set, its corresponding entry in the | |
251 | shared index will be replaced with an entry in this index | |
252 | file. All replaced entries are stored in sorted order in this | |
253 | index. The first "1" bit in the replace bitmap corresponds to the | |
254 | first index entry, the second "1" bit to the second entry and so | |
255 | on. Replaced entries may have empty path names to save space. | |
256 | ||
257 | The remaining index entries after replaced ones will be added to the | |
f745acb0 | 258 | final index. These added entries are also sorted by entry name then |
5fc2fc8f | 259 | stage. |
83c094ad NTND |
260 | |
261 | == Untracked cache | |
262 | ||
263 | Untracked cache saves the untracked file list and necessary data to | |
264 | verify the cache. The signature for this extension is { 'U', 'N', | |
265 | 'T', 'R' }. | |
266 | ||
267 | The extension starts with | |
268 | ||
1e8fef60 NTND |
269 | - A sequence of NUL-terminated strings, preceded by the size of the |
270 | sequence in variable width encoding. Each string describes the | |
271 | environment where the cache can be used. | |
272 | ||
83c094ad NTND |
273 | - Stat data of $GIT_DIR/info/exclude. See "Index entry" section from |
274 | ctime field until "file size". | |
275 | ||
7dd0eaa3 | 276 | - Stat data of core.excludesFile |
83c094ad NTND |
277 | |
278 | - 32-bit dir_flags (see struct dir_struct) | |
279 | ||
123712ba | 280 | - Hash of $GIT_DIR/info/exclude. A null hash means the file |
83c094ad NTND |
281 | does not exist. |
282 | ||
7dd0eaa3 | 283 | - Hash of core.excludesFile. A null hash means the file does |
83c094ad NTND |
284 | not exist. |
285 | ||
286 | - NUL-terminated string of per-dir exclude file name. This usually | |
287 | is ".gitignore". | |
288 | ||
289 | - The number of following directory blocks, variable width | |
290 | encoding. If this number is zero, the extension ends here with a | |
291 | following NUL. | |
292 | ||
293 | - A number of directory blocks in depth-first-search order, each | |
294 | consists of | |
295 | ||
296 | - The number of untracked entries, variable width encoding. | |
297 | ||
298 | - The number of sub-directory blocks, variable width encoding. | |
299 | ||
300 | - The directory name terminated by NUL. | |
301 | ||
da4c5ada | 302 | - A number of untracked file/dir names terminated by NUL. |
83c094ad NTND |
303 | |
304 | The remaining data of each directory block is grouped by type: | |
305 | ||
306 | - An ewah bitmap, the n-th bit marks whether the n-th directory has | |
307 | valid untracked cache entries. | |
308 | ||
309 | - An ewah bitmap, the n-th bit records "check-only" bit of | |
310 | read_directory_recursive() for the n-th directory. | |
311 | ||
123712ba | 312 | - An ewah bitmap, the n-th bit indicates whether hash and stat data |
83c094ad NTND |
313 | is valid for the n-th directory and exists in the next data. |
314 | ||
315 | - An array of stat data. The n-th data corresponds with the n-th | |
316 | "one" bit in the previous ewah bitmap. | |
317 | ||
123712ba | 318 | - An array of hashes. The n-th hash corresponds with the n-th "one" bit |
83c094ad NTND |
319 | in the previous ewah bitmap. |
320 | ||
321 | - One NUL. | |
780494b1 BP |
322 | |
323 | == File System Monitor cache | |
324 | ||
325 | The file system monitor cache tracks files for which the core.fsmonitor | |
326 | hook has told us about changes. The signature for this extension is | |
327 | { 'F', 'S', 'M', 'N' }. | |
328 | ||
329 | The extension starts with | |
330 | ||
5885367e | 331 | - 32-bit version number: the current supported versions are 1 and 2. |
780494b1 | 332 | |
5885367e JH |
333 | - (Version 1) |
334 | 64-bit time: the extension data reflects all changes through the given | |
780494b1 BP |
335 | time which is stored as the nanoseconds elapsed since midnight, |
336 | January 1, 1970. | |
337 | ||
5885367e JH |
338 | - (Version 2) |
339 | A null terminated string: an opaque token defined by the file system | |
340 | monitor application. The extension data reflects all changes relative | |
341 | to that token. | |
342 | ||
780494b1 BP |
343 | - 32-bit bitmap size: the size of the CE_FSMONITOR_VALID bitmap. |
344 | ||
345 | - An ewah bitmap, the n-th bit indicates whether the n-th index entry | |
346 | is not CE_FSMONITOR_VALID. | |
3b1d9e04 BP |
347 | |
348 | == End of Index Entry | |
349 | ||
350 | The End of Index Entry (EOIE) is used to locate the end of the variable | |
031fd4b9 | 351 | length index entries and the beginning of the extensions. Code can take |
3b1d9e04 BP |
352 | advantage of this to quickly locate the index extensions without having |
353 | to parse through all of the index entries. | |
354 | ||
355 | Because it must be able to be loaded before the variable length cache | |
356 | entries and other index extensions, this extension must be written last. | |
357 | The signature for this extension is { 'E', 'O', 'I', 'E' }. | |
358 | ||
359 | The extension consists of: | |
360 | ||
361 | - 32-bit offset to the end of the index entries | |
362 | ||
123712ba | 363 | - Hash over the extension types and their sizes (but not |
3b1d9e04 BP |
364 | their contents). E.g. if we have "TREE" extension that is N-bytes |
365 | long, "REUC" extension that is M-bytes long, followed by "EOIE", | |
366 | then the hash would be: | |
367 | ||
123712ba | 368 | Hash("TREE" + <binary representation of N> + |
3b1d9e04 | 369 | "REUC" + <binary representation of M>) |
3255089a BP |
370 | |
371 | == Index Entry Offset Table | |
372 | ||
373 | The Index Entry Offset Table (IEOT) is used to help address the CPU | |
374 | cost of loading the index by enabling multi-threading the process of | |
375 | converting cache entries from the on-disk format to the in-memory format. | |
376 | The signature for this extension is { 'I', 'E', 'O', 'T' }. | |
377 | ||
378 | The extension consists of: | |
379 | ||
380 | - 32-bit version (currently 1) | |
381 | ||
382 | - A number of index offset entries each consisting of: | |
383 | ||
031fd4b9 | 384 | - 32-bit offset from the beginning of the file to the first cache entry |
3255089a BP |
385 | in this block of entries. |
386 | ||
387 | - 32-bit count of cache entries in this block |