1GITFORMAT-INDEX(5) Git Manual GITFORMAT-INDEX(5)
2
6 gitformat-index - Git index format
7
9 $GIT_DIR/index
10
12 Git index format
13
15 All binary numbers are in network byte order.
16 In a repository using the traditional SHA-1, checksums and object IDs
17 (object names) mentioned below are all computed using SHA-1. Similarly,
18 in SHA-256 repositories, these values are computed using SHA-256.
19 Version 2 is described here unless stated otherwise.
20 • A 12-byte header consisting of
22 4-byte signature:
24 The signature is { 'D', 'I', 'R', 'C' } (stands for "dircache")
25 4-byte version number:
27 The current supported versions are 2, 3 and 4.
28 32-bit number of index entries.
30 • A number of sorted index entries (see below).
32 • Extensions
34 Extensions are identified by signature. Optional extensions can
36 be ignored if Git does not understand them.
37 4-byte extension signature. If the first byte is 'A'..'Z' the
39 extension is optional and can be ignored.
40 32-bit size of the extension
42 Extension data
44 • Hash checksum over the content of the index file before this
46 checksum.
47
49 Index entries are sorted in ascending order on the name field,
50 interpreted as a string of unsigned bytes (i.e. memcmp() order, no
51 localization, no special casing of directory separator '/'). Entries
52 with the same name are sorted by their stage field.
53 An index entry typically represents a file. However, if sparse-checkout
55 is enabled in cone mode (`core.sparseCheckoutCone` is enabled) and the
56 `extensions.sparseIndex` extension is enabled, then the index may
57 contain entries for directories outside of the sparse-checkout definition.
58 These entries have mode `040000`, include the `SKIP_WORKTREE` bit, and
59 the path ends in a directory separator.
60 32-bit ctime seconds, the last time a file's metadata changed
62 this is stat(2) data
63 32-bit ctime nanosecond fractions
65 this is stat(2) data
66 32-bit mtime seconds, the last time a file's data changed
68 this is stat(2) data
69 32-bit mtime nanosecond fractions
71 this is stat(2) data
72 32-bit dev
74 this is stat(2) data
75 32-bit ino
77 this is stat(2) data
78 32-bit mode, split into (high to low bits)
80 4-bit object type
82 valid values in binary are 1000 (regular file), 1010 (symbolic link)
83 and 1110 (gitlink)
84 3-bit unused
86 9-bit unix permission. Only 0755 and 0644 are valid for regular files.
88 Symbolic links and gitlinks have value 0 in this field.
89 32-bit uid
91 this is stat(2) data
92 32-bit gid
94 this is stat(2) data
95 32-bit file size
97 This is the on-disk size from stat(2), truncated to 32-bit.
98 Object name for the represented object
100 A 16-bit 'flags' field split into (high to low bits)
102 1-bit assume-valid flag
104 1-bit extended flag (must be zero in version 2)
106 2-bit stage (during merge)
108 12-bit name length if the length is less than 0xFFF; otherwise 0xFFF
110 is stored in this field.
111 (Version 3 or later) A 16-bit field, only applicable if the
113 "extended flag" above is 1, split into (high to low bits).
114 1-bit reserved for future
116 1-bit skip-worktree flag (used by sparse checkout)
118 1-bit intent-to-add flag (used by "git add -N")
120 13-bit unused, must be zero
122 Entry path name (variable length) relative to top level directory
124 (without leading slash). '/' is used as path separator. The special
125 path components ".", ".." and ".git" (without quotes) are disallowed.
126 Trailing slash is also disallowed.
127 The exact encoding is undefined, but the '.' and '/' characters
129 are encoded in 7-bit ASCII and the encoding cannot contain a NUL
130 byte (iow, this is a UNIX pathname).
131 (Version 4) In version 4, the entry path name is prefix-compressed
133 relative to the path name for the previous entry (the very first
134 entry is encoded as if the path name for the previous entry is an
135 empty string). At the beginning of an entry, an integer N in the
136 variable width encoding (the same encoding as the offset is encoded
137 for OFS_DELTA pack entries; see linkgit:gitformat-pack[5]) is stored, followed
138 by a NUL-terminated string S. Removing N bytes from the end of the
139 path name for the previous entry, and replacing it with the string S
140 yields the path name for this entry.
141 1-8 nul bytes as necessary to pad the entry to a multiple of eight bytes
143 while keeping the name NUL-terminated.
144 (Version 4) In version 4, the padding after the pathname does not
146 exist.
147 Interpretation of index entries in split index mode is completely
149 different. See below for details.
150
152 Cache tree
153 Since the index does not record entries for directories, the cache
154 entries cannot describe tree objects that already exist in the object
155 database for regions of the index that are unchanged from an existing
156 commit. The cache tree extension stores a recursive tree structure that
157 describes the trees that already exist and completely match sections of
158 the cache entries. This speeds up tree object generation from the index
159 for a new commit by only computing the trees that are "new" to that
160 commit. It also assists when comparing the index to another tree, such
161 as `HEAD^{tree}`, since sections of the index can be skipped when a tree
162 comparison demonstrates equality.
163 The recursive tree structure uses nodes that store a number of cache
165 entries, a list of subnodes, and an object ID (OID). The OID references
166 the existing tree for that node, if it is known to exist. The subnodes
167 correspond to subdirectories that themselves have cache tree nodes. The
168 number of cache entries corresponds to the number of cache entries in
169 the index that describe paths within that tree's directory.
170 The extension tracks the full directory structure in the cache tree
172 extension, but this is generally smaller than the full cache entry list.
173 When a path is updated in index, Git invalidates all nodes of the
175 recursive cache tree corresponding to the parent directories of that
176 path. We store these tree nodes as being "invalid" by using "-1" as the
177 number of cache entries. Invalid nodes still store a span of index
178 entries, allowing Git to focus its efforts when reconstructing a full
179 cache tree.
180 The signature for this extension is { 'T', 'R', 'E', 'E' }.
182 A series of entries fill the entire extension; each of which
184 consists of:
185 • NUL-terminated path component (relative to its parent directory);
187 • ASCII decimal number of entries in the index that is covered by the
189 tree this entry represents (entry_count);
190 • A space (ASCII 32);
192 • ASCII decimal number that represents the number of subtrees this
194 tree has;
195 • A newline (ASCII 10); and
197 • Object name for the object that would result from writing this span
199 of index as a tree.
200 An entry can be in an invalidated state and is represented by having
202 a negative number in the entry_count field. In this case, there is no
203 object name and the next entry starts immediately after the newline.
204 When writing an invalid entry, -1 should always be used as entry_count.
205 The entries are written out in the top-down, depth-first order. The
207 first entry represents the root level of the repository, followed by the
208 first subtree--let's call this A--of the root level (with its name
209 relative to the root level), followed by the first subtree of A (with
210 its name relative to A), and so on. The specified number of subtrees
211 indicates when the current level of the recursive stack is complete.
212 Resolve undo
214 A conflict is represented in the index as a set of higher stage entries.
215 When a conflict is resolved (e.g. with "git add path"), these higher
216 stage entries will be removed and a stage-0 entry with proper resolution
217 is added.
218 When these higher stage entries are removed, they are saved in the
220 resolve undo extension, so that conflicts can be recreated (e.g. with
221 "git checkout -m"), in case users want to redo a conflict resolution
222 from scratch.
223 The signature for this extension is { 'R', 'E', 'U', 'C' }.
225 A series of entries fill the entire extension; each of which
227 consists of:
228 • NUL-terminated pathname the entry describes (relative to the root
230 of the repository, i.e. full pathname);
231 • Three NUL-terminated ASCII octal numbers, entry mode of entries in
233 stage 1 to 3 (a missing stage is represented by "0" in this field);
234 and
235 • At most three object names of the entry in stages from 1 to 3
237 (nothing is written for a missing stage).
238 Split index
240 In split index mode, the majority of index entries could be stored
241 in a separate file. This extension records the changes to be made on
242 top of that to produce the final index.
243 The signature for this extension is { 'l', 'i', 'n', 'k' }.
245 The extension consists of:
247 • Hash of the shared index file. The shared index file path is
249 $GIT_DIR/sharedindex.<hash>. If all bits are zero, the index does
250 not require a shared index file.
251 • An ewah-encoded delete bitmap, each bit represents an entry in the
253 shared index. If a bit is set, its corresponding entry in the
254 shared index will be removed from the final index. Note, because a
255 delete operation changes index entry positions, but we do need
256 original positions in replace phase, it’s best to just mark entries
257 for removal, then do a mass deletion after replacement.
258 • An ewah-encoded replace bitmap, each bit represents an entry in the
260 shared index. If a bit is set, its corresponding entry in the
261 shared index will be replaced with an entry in this index file. All
262 replaced entries are stored in sorted order in this index. The
263 first "1" bit in the replace bitmap corresponds to the first index
264 entry, the second "1" bit to the second entry and so on. Replaced
265 entries may have empty path names to save space.
266 The remaining index entries after replaced ones will be added to the
268 final index. These added entries are also sorted by entry name then
269 stage.
270
272 Untracked cache saves the untracked file list and necessary data to
273 verify the cache. The signature for this extension is { 'U', 'N',
274 'T', 'R' }.
275 The extension starts with
277 • A sequence of NUL-terminated strings, preceded by the size of the
279 sequence in variable width encoding. Each string describes the
280 environment where the cache can be used.
281 • Stat data of $GIT_DIR/info/exclude. See "Index entry" section from
283 ctime field until "file size".
284 • Stat data of core.excludesFile
286 • 32-bit dir_flags (see struct dir_struct)
288 • Hash of $GIT_DIR/info/exclude. A null hash means the file does not
290 exist.
291 • Hash of core.excludesFile. A null hash means the file does not
293 exist.
294 • NUL-terminated string of per-dir exclude file name. This usually is
296 ".gitignore".
297 • The number of following directory blocks, variable width encoding.
299 If this number is zero, the extension ends here with a following
300 NUL.
301 • A number of directory blocks in depth-first-search order, each
303 consists of
304 • The number of untracked entries, variable width encoding.
306 • The number of sub-directory blocks, variable width encoding.
308 • The directory name terminated by NUL.
310 • A number of untracked file/dir names terminated by NUL.
312 The remaining data of each directory block is grouped by type:
314 • An ewah bitmap, the n-th bit marks whether the n-th directory has
316 valid untracked cache entries.
317 • An ewah bitmap, the n-th bit records "check-only" bit of
319 read_directory_recursive() for the n-th directory.
320 • An ewah bitmap, the n-th bit indicates whether hash and stat data
322 is valid for the n-th directory and exists in the next data.
323 • An array of stat data. The n-th data corresponds with the n-th
325 "one" bit in the previous ewah bitmap.
326 • An array of hashes. The n-th hash corresponds with the n-th "one"
328 bit in the previous ewah bitmap.
329 • One NUL.
331
333 The file system monitor cache tracks files for which the core.fsmonitor
334 hook has told us about changes. The signature for this extension is
335 { 'F', 'S', 'M', 'N' }.
336 The extension starts with
338 • 32-bit version number: the current supported versions are 1 and 2.
340 • (Version 1) 64-bit time: the extension data reflects all changes
342 through the given time which is stored as the nanoseconds elapsed
343 since midnight, January 1, 1970.
344 • (Version 2) A null terminated string: an opaque token defined by
346 the file system monitor application. The extension data reflects
347 all changes relative to that token.
348 • 32-bit bitmap size: the size of the CE_FSMONITOR_VALID bitmap.
350 • An ewah bitmap, the n-th bit indicates whether the n-th index entry
352 is not CE_FSMONITOR_VALID.
353
355 The End of Index Entry (EOIE) is used to locate the end of the variable
356 length index entries and the beginning of the extensions. Code can take
357 advantage of this to quickly locate the index extensions without having
358 to parse through all of the index entries.
359 Because it must be able to be loaded before the variable length cache
361 entries and other index extensions, this extension must be written last.
362 The signature for this extension is { 'E', 'O', 'I', 'E' }.
363 The extension consists of:
365 • 32-bit offset to the end of the index entries
367 • Hash over the extension types and their sizes (but not their
369 contents). E.g. if we have "TREE" extension that is N-bytes long,
370 "REUC" extension that is M-bytes long, followed by "EOIE", then the
371 hash would be:
372 Hash("TREE" + <binary representation of N> +
374 "REUC" + <binary representation of M>)
375
377 The Index Entry Offset Table (IEOT) is used to help address the CPU
378 cost of loading the index by enabling multi-threading the process of
379 converting cache entries from the on-disk format to the in-memory format.
380 The signature for this extension is { 'I', 'E', 'O', 'T' }.
381 The extension consists of:
383 • 32-bit version (currently 1)
385 • A number of index offset entries each consisting of:
387 • 32-bit offset from the beginning of the file to the first cache
389 entry in this block of entries.
390 • 32-bit count of cache entries in this block
392
394 When using sparse-checkout in cone mode, some entire directories within
395 the index can be summarized by pointing to a tree object instead of the
396 entire expanded list of paths within that tree. An index containing such
397 entries is a "sparse index". Index format versions 4 and less were not
398 implemented with such entries in mind. Thus, for these versions, an
399 index containing sparse directory entries will include this extension
400 with signature { 's', 'd', 'i', 'r' }. Like the split-index extension,
401 tools should avoid interacting with a sparse index unless they understand
402 this extension.
403
405 Part of the git(1) suite
406Git 2.39.1 2023-01-13 GITFORMAT-INDEX(5)