2 .\" Title: gitformat-index
3 .\" Author: [FIXME: author] [see http://www.docbook.org/tdg5/en/html/author]
4 .\" Generator: DocBook XSL Stylesheets v1.79.2 <http://docbook.sf.net/>
7 .\" Source: Git 2.46.1.603.g94b60adee3
10 .TH "GITFORMAT\-INDEX" "5" "2024-09-20" "Git 2\&.46\&.1\&.603\&.g94b60a" "Git Manual"
11 .\" -----------------------------------------------------------------
12 .\" * Define some portability stuff
13 .\" -----------------------------------------------------------------
14 .\" ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
15 .\" http://bugs.debian.org/507673
16 .\" http://lists.gnu.org/archive/html/groff/2009-02/msg00013.html
17 .\" ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
20 .\" -----------------------------------------------------------------
21 .\" * set default formatting
22 .\" -----------------------------------------------------------------
23 .\" disable hyphenation
25 .\" disable justification (adjust text to left margin only)
27 .\" -----------------------------------------------------------------
28 .\" * MAIN CONTENT STARTS HERE *
29 .\" -----------------------------------------------------------------
31 gitformat-index \- Git index format
40 .SH "THE GIT INDEX FILE HAS THE FOLLOWING FORMAT"
46 All binary numbers are in network byte order\&.
47 In a repository using the traditional SHA\-1, checksums and object IDs
48 (object names) mentioned below are all computed using SHA\-1\&. Similarly,
49 in SHA\-256 repositories, these values are computed using SHA\-256\&.
50 Version 2 is described here unless stated otherwise\&.
64 A 12\-byte header consisting of
71 The signature is { \*(AqD\*(Aq, \*(AqI\*(Aq, \*(AqR\*(Aq, \*(AqC\*(Aq } (stands for "dircache")
81 4\-byte version number:
82 The current supported versions are 2, 3 and 4\&.
92 32\-bit number of index entries\&.
107 A number of sorted index entries (see below)\&.
124 Extensions are identified by signature\&. Optional extensions can
125 be ignored if Git does not understand them\&.
135 4\-byte extension signature\&. If the first byte is \*(AqA\*(Aq\&.\&.\*(AqZ\*(Aq the
136 extension is optional and can be ignored\&.
146 32\-bit size of the extension
171 Hash checksum over the content of the index file before this checksum\&.
179 Index entries are sorted in ascending order on the name field,
180 interpreted as a string of unsigned bytes (i\&.e\&. memcmp() order, no
181 localization, no special casing of directory separator \*(Aq/\*(Aq)\&. Entries
182 with the same name are sorted by their stage field\&.
192 An index entry typically represents a file\&. However, if sparse\-checkout
193 is enabled in cone mode (`core\&.sparseCheckoutCone` is enabled) and the
194 `extensions\&.sparseIndex` extension is enabled, then the index may
195 contain entries for directories outside of the sparse\-checkout definition\&.
196 These entries have mode `040000`, include the `SKIP_WORKTREE` bit, and
197 the path ends in a directory separator\&.
207 32\-bit ctime seconds, the last time a file\*(Aqs metadata changed
218 32\-bit ctime nanosecond fractions
229 32\-bit mtime seconds, the last time a file\*(Aqs data changed
240 32\-bit mtime nanosecond fractions
273 32\-bit mode, split into (high to low bits)
283 16\-bit unused, must be zero
294 valid values in binary are 1000 (regular file), 1010 (symbolic link)
305 3\-bit unused, must be zero
315 9\-bit unix permission\&. Only 0755 and 0644 are valid for regular files\&.
316 Symbolic links and gitlinks have value 0 in this field\&.
349 This is the on\-disk size from stat(2), truncated to 32\-bit\&.
359 Object name for the represented object
369 A 16\-bit \*(Aqflags\*(Aq field split into (high to low bits)
379 1\-bit assume\-valid flag
389 1\-bit extended flag (must be zero in version 2)
399 2\-bit stage (during merge)
409 12\-bit name length if the length is less than 0xFFF; otherwise 0xFFF
410 is stored in this field\&.
420 (Version 3 or later) A 16\-bit field, only applicable if the
421 "extended flag" above is 1, split into (high to low bits)\&.
431 1\-bit reserved for future
441 1\-bit skip\-worktree flag (used by sparse checkout)
451 1\-bit intent\-to\-add flag (used by "git add \-N")
461 13\-bit unused, must be zero
471 Entry path name (variable length) relative to top level directory
472 (without leading slash)\&. \*(Aq/\*(Aq is used as path separator\&. The special
473 path components "\&.", "\&.\&." and "\&.git" (without quotes) are disallowed\&.
474 Trailing slash is also disallowed\&.
484 The exact encoding is undefined, but the \*(Aq\&.\*(Aq and \*(Aq/\*(Aq characters
485 are encoded in 7\-bit ASCII and the encoding cannot contain a NUL
486 byte (iow, this is a UNIX pathname)\&.
496 (Version 4) In version 4, the entry path name is prefix\-compressed
497 relative to the path name for the previous entry (the very first
498 entry is encoded as if the path name for the previous entry is an
499 empty string)\&. At the beginning of an entry, an integer N in the
500 variable width encoding (the same encoding as the offset is encoded
501 for OFS_DELTA pack entries; see linkgit:gitformat\-pack[5]) is stored, followed
502 by a NUL\-terminated string S\&. Removing N bytes from the end of the
503 path name for the previous entry, and replacing it with the string S
504 yields the path name for this entry\&.
514 1\-8 nul bytes as necessary to pad the entry to a multiple of eight bytes
515 while keeping the name NUL\-terminated\&.
525 (Version 4) In version 4, the padding after the pathname does not
536 Interpretation of index entries in split index mode is completely
537 different\&. See below for details\&.
549 Since the index does not record entries for directories, the cache
550 entries cannot describe tree objects that already exist in the object
551 database for regions of the index that are unchanged from an existing
552 commit\&. The cache tree extension stores a recursive tree structure that
553 describes the trees that already exist and completely match sections of
554 the cache entries\&. This speeds up tree object generation from the index
555 for a new commit by only computing the trees that are "new" to that
556 commit\&. It also assists when comparing the index to another tree, such
557 as `HEAD^{tree}`, since sections of the index can be skipped when a tree
558 comparison demonstrates equality\&.
568 The recursive tree structure uses nodes that store a number of cache
569 entries, a list of subnodes, and an object ID (OID)\&. The OID references
570 the existing tree for that node, if it is known to exist\&. The subnodes
571 correspond to subdirectories that themselves have cache tree nodes\&. The
572 number of cache entries corresponds to the number of cache entries in
573 the index that describe paths within that tree\*(Aqs directory\&.
583 The extension tracks the full directory structure in the cache tree
584 extension, but this is generally smaller than the full cache entry list\&.
594 When a path is updated in index, Git invalidates all nodes of the
595 recursive cache tree corresponding to the parent directories of that
596 path\&. We store these tree nodes as being "invalid" by using "\-1" as the
597 number of cache entries\&. Invalid nodes still store a span of index
598 entries, allowing Git to focus its efforts when reconstructing a full
609 The signature for this extension is { \*(AqT\*(Aq, \*(AqR\*(Aq, \*(AqE\*(Aq, \*(AqE\*(Aq }\&.
619 A series of entries fill the entire extension; each of which
634 NUL\-terminated path component (relative to its parent directory);
645 ASCII decimal number of entries in the index that is covered by the tree this entry represents (entry_count);
667 ASCII decimal number that represents the number of subtrees this tree has;
678 A newline (ASCII 10); and
689 Object name for the object that would result from writing this span of index as a tree\&.
695 An entry can be in an invalidated state and is represented by having
696 a negative number in the entry_count field\&. In this case, there is no
697 object name and the next entry starts immediately after the newline\&.
698 When writing an invalid entry, \-1 should always be used as entry_count\&.
708 The entries are written out in the top\-down, depth\-first order\&. The
709 first entry represents the root level of the repository, followed by the
710 first subtree\-\-let\*(Aqs call this A\-\-of the root level (with its name
711 relative to the root level), followed by the first subtree of A (with
712 its name relative to A), and so on\&. The specified number of subtrees
713 indicates when the current level of the recursive stack is complete\&.
725 A conflict is represented in the index as a set of higher stage entries\&.
726 When a conflict is resolved (e\&.g\&. with "git add path"), these higher
727 stage entries will be removed and a stage\-0 entry with proper resolution
738 When these higher stage entries are removed, they are saved in the
739 resolve undo extension, so that conflicts can be recreated (e\&.g\&. with
740 "git checkout \-m"), in case users want to redo a conflict resolution
751 The signature for this extension is { \*(AqR\*(Aq, \*(AqE\*(Aq, \*(AqU\*(Aq, \*(AqC\*(Aq }\&.
761 A series of entries fill the entire extension; each of which
776 NUL\-terminated pathname the entry describes (relative to the root of the repository, i\&.e\&. full pathname);
787 Three NUL\-terminated ASCII octal numbers, entry mode of entries in stage 1 to 3 (a missing stage is represented by "0" in this field); and
798 At most three object names of the entry in stages from 1 to 3 (nothing is written for a missing stage)\&.
806 In split index mode, the majority of index entries could be stored
807 in a separate file\&. This extension records the changes to be made on
808 top of that to produce the final index\&.
818 The signature for this extension is { \*(Aql\*(Aq, \*(Aqi\*(Aq, \*(Aqn\*(Aq, \*(Aqk\*(Aq }\&.
828 The extension consists of:
842 Hash of the shared index file\&. The shared index file path is $GIT_DIR/sharedindex\&.<hash>\&. If all bits are zero, the index does not require a shared index file\&.
853 An ewah\-encoded delete bitmap, each bit represents an entry in the shared index\&. If a bit is set, its corresponding entry in the shared index will be removed from the final index\&. Note, because a delete operation changes index entry positions, but we do need original positions in replace phase, it\(cqs best to just mark entries for removal, then do a mass deletion after replacement\&.
864 An ewah\-encoded replace bitmap, each bit represents an entry in the shared index\&. If a bit is set, its corresponding entry in the shared index will be replaced with an entry in this index file\&. All replaced entries are stored in sorted order in this index\&. The first "1" bit in the replace bitmap corresponds to the first index entry, the second "1" bit to the second entry and so on\&. Replaced entries may have empty path names to save space\&.
870 The remaining index entries after replaced ones will be added to the
871 final index\&. These added entries are also sorted by entry name then
878 .SH "UNTRACKED CACHE"
884 Untracked cache saves the untracked file list and necessary data to
885 verify the cache\&. The signature for this extension is { \*(AqU\*(Aq, \*(AqN\*(Aq,
886 \*(AqT\*(Aq, \*(AqR\*(Aq }\&.
896 The extension starts with
910 A sequence of NUL\-terminated strings, preceded by the size of the sequence in variable width encoding\&. Each string describes the environment where the cache can be used\&.
921 Stat data of $GIT_DIR/info/exclude\&. See "Index entry" section from ctime field until "file size"\&.
932 Stat data of core\&.excludesFile
943 32\-bit dir_flags (see struct dir_struct)
954 Hash of $GIT_DIR/info/exclude\&. A null hash means the file does not exist\&.
965 Hash of core\&.excludesFile\&. A null hash means the file does not exist\&.
976 NUL\-terminated string of per\-dir exclude file name\&. This usually is "\&.gitignore"\&.
987 The number of following directory blocks, variable width encoding\&. If this number is zero, the extension ends here with a following NUL\&.
998 A number of directory blocks in depth\-first\-search order, each consists of
1003 \h'-04'\(bu\h'+03'\c
1009 The number of untracked entries, variable width encoding\&.
1014 \h'-04'\(bu\h'+03'\c
1020 The number of sub\-directory blocks, variable width encoding\&.
1025 \h'-04'\(bu\h'+03'\c
1031 The directory name terminated by NUL\&.
1036 \h'-04'\(bu\h'+03'\c
1042 A number of untracked file/dir names terminated by NUL\&.
1045 The remaining data of each directory block is grouped by type:
1049 \h'-04'\(bu\h'+03'\c
1055 An ewah bitmap, the n\-th bit marks whether the n\-th directory has valid untracked cache entries\&.
1060 \h'-04'\(bu\h'+03'\c
1066 An ewah bitmap, the n\-th bit records "check\-only" bit of read_directory_recursive() for the n\-th directory\&.
1071 \h'-04'\(bu\h'+03'\c
1077 An ewah bitmap, the n\-th bit indicates whether hash and stat data is valid for the n\-th directory and exists in the next data\&.
1082 \h'-04'\(bu\h'+03'\c
1088 An array of stat data\&. The n\-th data corresponds with the n\-th "one" bit in the previous ewah bitmap\&.
1093 \h'-04'\(bu\h'+03'\c
1099 An array of hashes\&. The n\-th hash corresponds with the n\-th "one" bit in the previous ewah bitmap\&.
1104 \h'-04'\(bu\h'+03'\c
1112 .SH "FILE SYSTEM MONITOR CACHE"
1118 The file system monitor cache tracks files for which the core\&.fsmonitor
1119 hook has told us about changes\&. The signature for this extension is
1120 { \*(AqF\*(Aq, \*(AqS\*(Aq, \*(AqM\*(Aq, \*(AqN\*(Aq }\&.
1130 The extension starts with
1138 \h'-04'\(bu\h'+03'\c
1144 32\-bit version number: the current supported versions are 1 and 2\&.
1149 \h'-04'\(bu\h'+03'\c
1155 (Version 1) 64\-bit time: the extension data reflects all changes through the given time which is stored as the nanoseconds elapsed since midnight, January 1, 1970\&.
1160 \h'-04'\(bu\h'+03'\c
1166 (Version 2) A null terminated string: an opaque token defined by the file system monitor application\&. The extension data reflects all changes relative to that token\&.
1171 \h'-04'\(bu\h'+03'\c
1177 32\-bit bitmap size: the size of the CE_FSMONITOR_VALID bitmap\&.
1182 \h'-04'\(bu\h'+03'\c
1188 An ewah bitmap, the n\-th bit indicates whether the n\-th index entry is not CE_FSMONITOR_VALID\&.
1190 .SH "END OF INDEX ENTRY"
1196 The End of Index Entry (EOIE) is used to locate the end of the variable
1197 length index entries and the beginning of the extensions\&. Code can take
1198 advantage of this to quickly locate the index extensions without having
1199 to parse through all of the index entries\&.
1209 Because it must be able to be loaded before the variable length cache
1210 entries and other index extensions, this extension must be written last\&.
1211 The signature for this extension is { \*(AqE\*(Aq, \*(AqO\*(Aq, \*(AqI\*(Aq, \*(AqE\*(Aq }\&.
1221 The extension consists of:
1229 \h'-04'\(bu\h'+03'\c
1235 32\-bit offset to the end of the index entries
1240 \h'-04'\(bu\h'+03'\c
1246 Hash over the extension types and their sizes (but not their contents)\&. E\&.g\&. if we have "TREE" extension that is N\-bytes long, "REUC" extension that is M\-bytes long, followed by "EOIE", then the hash would be:
1252 Hash("TREE" + <binary\-representation\-of\-N> +
1253 "REUC" + <binary\-representation\-of\-M>)
1259 .SH "INDEX ENTRY OFFSET TABLE"
1265 The Index Entry Offset Table (IEOT) is used to help address the CPU
1266 cost of loading the index by enabling multi\-threading the process of
1267 converting cache entries from the on\-disk format to the in\-memory format\&.
1268 The signature for this extension is { \*(AqI\*(Aq, \*(AqE\*(Aq, \*(AqO\*(Aq, \*(AqT\*(Aq }\&.
1278 The extension consists of:
1286 \h'-04'\(bu\h'+03'\c
1292 32\-bit version (currently 1)
1297 \h'-04'\(bu\h'+03'\c
1303 A number of index offset entries each consisting of:
1308 \h'-04'\(bu\h'+03'\c
1314 32\-bit offset from the beginning of the file to the first cache entry in this block of entries\&.
1319 \h'-04'\(bu\h'+03'\c
1325 32\-bit count of cache entries in this block
1327 .SH "SPARSE DIRECTORY ENTRIES"
1333 When using sparse\-checkout in cone mode, some entire directories within
1334 the index can be summarized by pointing to a tree object instead of the
1335 entire expanded list of paths within that tree\&. An index containing such
1336 entries is a "sparse index"\&. Index format versions 4 and less were not
1337 implemented with such entries in mind\&. Thus, for these versions, an
1338 index containing sparse directory entries will include this extension
1339 with signature { \*(Aqs\*(Aq, \*(Aqd\*(Aq, \*(Aqi\*(Aq, \*(Aqr\*(Aq }\&. Like the split\-index extension,
1340 tools should avoid interacting with a sparse index unless they understand
1348 Part of the \fBgit\fR(1) suite