2 The Second Extended Filesystem
3 ==============================
5 ext2 was originally released in January 1993. Written by R\'emy Card,
6 Theodore Ts'o and Stephen Tweedie, it was a major rewrite of the
7 Extended Filesystem. It is currently still (April 2001) the predominant
8 filesystem in use by Linux. There are also implementations available
9 for NetBSD, FreeBSD, the GNU HURD, Windows 95/98/NT, OS/2 and RISC OS.
14 When mounting an ext2 filesystem, the following options are accepted.
15 Defaults are marked with (*).
17 bsddf (*) Makes `df' act like BSD.
18 minixdf Makes `df' act like Minix.
20 barrier=1 This enables/disables barriers. barrier=0 disables it,
23 orlov (*) This enables the new Orlov block allocator. It's
26 oldalloc This disables the Orlov block allocator and
27 enables the old block allocator. Orlov should
28 have better performance, we'd like to get some
29 feedback if it's the contrary for you.
31 user_xattr (*) Enables POSIX Extended Attributes. It's enabled by
32 default, however you need to confifure its support
33 (CONFIG_EXT2_FS_XATTR). This is neccesary if you want
34 to use POSIX Acces Control Lists support. You can visit
35 http://acl.bestbits.at to know more about POSIX Extended
38 nouser_xattr Disables POSIX Extended Attributes.
40 acl (*) Enables POSIX Access Control Lists support. This is
41 enabled by default, however you need to configure
42 its support (CONFIG_EXT2_FS_POSIX_ACL). If you want
43 to know more about ACLs visit http://acl.bestbits.at
45 noacl This option disables POSIX Access Control List support.
47 check=none, nocheck (*) Don't do extra checking of bitmaps on mount
48 (check=normal and check=strict options removed)
50 debug Extra debugging information is sent to the
51 kernel syslog. Useful for developers.
53 errors=continue (*) Keep going on a filesystem error.
54 errors=remount-ro Remount the filesystem read-only on an error.
55 errors=panic Panic and halt the machine if an error occurs.
57 grpid, bsdgroups Give objects the same group ID as their parent.
58 nogrpid, sysvgroups (*) New objects have the group ID of their creator.
60 resuid=n The user ID which may use the reserved blocks.
61 resgid=n The group ID which may use the reserved blocks.
63 sb=n Use alternate superblock at this location.
65 grpquota,noquota,quota,usrquota Quota options are silently ignored by ext2.
71 ext2 shares many properties with traditional Unix filesystems. It has
72 the concepts of blocks, inodes and directories. It has space in the
73 specification for Access Control Lists (ACLs), fragments, undeletion and
74 compression though these are not yet implemented (some are available as
75 separate patches). There is also a versioning mechanism to allow new
76 features (such as journalling) to be added in a maximally compatible
82 The space in the device or file is split up into blocks. These are
83 a fixed size, of 1024, 2048 or 4096 bytes (8192 bytes on Alpha systems),
84 which is decided when the filesystem is created. Smaller blocks mean
85 less wasted space per file, but require slightly more accounting overhead,
86 and also impose other limits on the size of files and the filesystem.
91 Blocks are clustered into block groups in order to reduce fragmentation
92 and minimise the amount of head seeking when reading a large amount
93 of consecutive data. Information about each block group is kept in a
94 descriptor table stored in the block(s) immediately after the superblock.
95 Two blocks near the start of each group are reserved for the block usage
96 bitmap and the inode usage bitmap which show which blocks and inodes
97 are in use. Since each bitmap is limited to a single block, this means
98 that the maximum size of a block group is 8 times the size of a block.
100 The block(s) following the bitmaps in each block group are designated
101 as the inode table for that block group and the remainder are the data
102 blocks. The block allocation algorithm attempts to allocate data blocks
103 in the same block group as the inode which contains them.
108 The superblock contains all the information about the configuration of
109 the filing system. The primary copy of the superblock is stored at an
110 offset of 1024 bytes from the start of the device, and it is essential
111 to mounting the filesystem. Since it is so important, backup copies of
112 the superblock are stored in block groups throughout the filesystem.
113 The first version of ext2 (revision 0) stores a copy at the start of
114 every block group, along with backups of the group descriptor block(s).
115 Because this can consume a considerable amount of space for large
116 filesystems, later revisions can optionally reduce the number of backup
117 copies by only putting backups in specific groups (this is the sparse
118 superblock feature). The groups chosen are 0, 1 and powers of 3, 5 and 7.
120 The information in the superblock contains fields such as the total
121 number of inodes and blocks in the filesystem and how many are free,
122 how many inodes and blocks are in each block group, when the filesystem
123 was mounted (and if it was cleanly unmounted), when it was modified,
124 what version of the filesystem it is (see the Revisions section below)
125 and which OS created it.
127 If the filesystem is revision 1 or higher, then there are extra fields,
128 such as a volume name, a unique identification number, the inode size,
129 and space for optional filesystem features to store configuration info.
131 All fields in the superblock (as in all other ext2 structures) are stored
132 on the disc in little endian format, so a filesystem is portable between
133 machines without having to know what machine it was created on.
138 The inode (index node) is a fundamental concept in the ext2 filesystem.
139 Each object in the filesystem is represented by an inode. The inode
140 structure contains pointers to the filesystem blocks which contain the
141 data held in the object and all of the metadata about an object except
142 its name. The metadata about an object includes the permissions, owner,
143 group, flags, size, number of blocks used, access time, change time,
144 modification time, deletion time, number of links, fragments, version
145 (for NFS) and extended attributes (EAs) and/or Access Control Lists (ACLs).
147 There are some reserved fields which are currently unused in the inode
148 structure and several which are overloaded. One field is reserved for the
149 directory ACL if the inode is a directory and alternately for the top 32
150 bits of the file size if the inode is a regular file (allowing file sizes
151 larger than 2GB). The translator field is unused under Linux, but is used
152 by the HURD to reference the inode of a program which will be used to
153 interpret this object. Most of the remaining reserved fields have been
154 used up for both Linux and the HURD for larger owner and group fields,
155 The HURD also has a larger mode field so it uses another of the remaining
156 fields to store the extra more bits.
158 There are pointers to the first 12 blocks which contain the file's data
159 in the inode. There is a pointer to an indirect block (which contains
160 pointers to the next set of blocks), a pointer to a doubly-indirect
161 block (which contains pointers to indirect blocks) and a pointer to a
162 trebly-indirect block (which contains pointers to doubly-indirect blocks).
164 The flags field contains some ext2-specific flags which aren't catered
165 for by the standard chmod flags. These flags can be listed with lsattr
166 and changed with the chattr command, and allow specific filesystem
167 behaviour on a per-file basis. There are flags for secure deletion,
168 undeletable, compression, synchronous updates, immutability, append-only,
169 dumpable, no-atime, indexed directories, and data-journaling. Not all
170 of these are supported yet.
175 A directory is a filesystem object and has an inode just like a file.
176 It is a specially formatted file containing records which associate
177 each name with an inode number. Later revisions of the filesystem also
178 encode the type of the object (file, directory, symlink, device, fifo,
179 socket) to avoid the need to check the inode itself for this information
180 (support for taking advantage of this feature does not yet exist in
183 The inode allocation code tries to assign inodes which are in the same
184 block group as the directory in which they are first created.
186 The current implementation of ext2 uses a singly-linked list to store
187 the filenames in the directory; a pending enhancement uses hashing of the
188 filenames to allow lookup without the need to scan the entire directory.
190 The current implementation never removes empty directory blocks once they
191 have been allocated to hold more files.
196 Symbolic links are also filesystem objects with inodes. They deserve
197 special mention because the data for them is stored within the inode
198 itself if the symlink is less than 60 bytes long. It uses the fields
199 which would normally be used to store the pointers to data blocks.
200 This is a worthwhile optimisation as it we avoid allocating a full
201 block for the symlink, and most symlinks are less than 60 characters long.
203 Character and block special devices never have data blocks assigned to
204 them. Instead, their device number is stored in the inode, again reusing
205 the fields which would be used to point to the data blocks.
210 In ext2, there is a mechanism for reserving a certain number of blocks
211 for a particular user (normally the super-user). This is intended to
212 allow for the system to continue functioning even if non-priveleged users
213 fill up all the space available to them (this is independent of filesystem
214 quotas). It also keeps the filesystem from filling up entirely which
215 helps combat fragmentation.
220 At boot time, most systems run a consistency check (e2fsck) on their
221 filesystems. The superblock of the ext2 filesystem contains several
222 fields which indicate whether fsck should actually run (since checking
223 the filesystem at boot can take a long time if it is large). fsck will
224 run if the filesystem was not cleanly unmounted, if the maximum mount
225 count has been exceeded or if the maximum time between checks has been
228 Feature Compatibility
229 ---------------------
231 The compatibility feature mechanism used in ext2 is sophisticated.
232 It safely allows features to be added to the filesystem, without
233 unnecessarily sacrificing compatibility with older versions of the
234 filesystem code. The feature compatibility mechanism is not supported by
235 the original revision 0 (EXT2_GOOD_OLD_REV) of ext2, but was introduced in
236 revision 1. There are three 32-bit fields, one for compatible features
237 (COMPAT), one for read-only compatible (RO_COMPAT) features and one for
238 incompatible (INCOMPAT) features.
240 These feature flags have specific meanings for the kernel as follows:
242 A COMPAT flag indicates that a feature is present in the filesystem,
243 but the on-disk format is 100% compatible with older on-disk formats, so
244 a kernel which didn't know anything about this feature could read/write
245 the filesystem without any chance of corrupting the filesystem (or even
246 making it inconsistent). This is essentially just a flag which says
247 "this filesystem has a (hidden) feature" that the kernel or e2fsck may
248 want to be aware of (more on e2fsck and feature flags later). The ext3
249 HAS_JOURNAL feature is a COMPAT flag because the ext3 journal is simply
250 a regular file with data blocks in it so the kernel does not need to
251 take any special notice of it if it doesn't understand ext3 journaling.
253 An RO_COMPAT flag indicates that the on-disk format is 100% compatible
254 with older on-disk formats for reading (i.e. the feature does not change
255 the visible on-disk format). However, an old kernel writing to such a
256 filesystem would/could corrupt the filesystem, so this is prevented. The
257 most common such feature, SPARSE_SUPER, is an RO_COMPAT feature because
258 sparse groups allow file data blocks where superblock/group descriptor
259 backups used to live, and ext2_free_blocks() refuses to free these blocks,
260 which would leading to inconsistent bitmaps. An old kernel would also
261 get an error if it tried to free a series of blocks which crossed a group
262 boundary, but this is a legitimate layout in a SPARSE_SUPER filesystem.
264 An INCOMPAT flag indicates the on-disk format has changed in some
265 way that makes it unreadable by older kernels, or would otherwise
266 cause a problem if an old kernel tried to mount it. FILETYPE is an
267 INCOMPAT flag because older kernels would think a filename was longer
268 than 256 characters, which would lead to corrupt directory listings.
269 The COMPRESSION flag is an obvious INCOMPAT flag - if the kernel
270 doesn't understand compression, you would just get garbage back from
271 read() instead of it automatically decompressing your data. The ext3
272 RECOVER flag is needed to prevent a kernel which does not understand the
273 ext3 journal from mounting the filesystem without replaying the journal.
275 For e2fsck, it needs to be more strict with the handling of these
276 flags than the kernel. If it doesn't understand ANY of the COMPAT,
277 RO_COMPAT, or INCOMPAT flags it will refuse to check the filesystem,
278 because it has no way of verifying whether a given feature is valid
279 or not. Allowing e2fsck to succeed on a filesystem with an unknown
280 feature is a false sense of security for the user. Refusing to check
281 a filesystem with unknown features is a good incentive for the user to
282 update to the latest e2fsck. This also means that anyone adding feature
283 flags to ext2 also needs to update e2fsck to verify these features.
288 It is frequently claimed that the ext2 implementation of writing
289 asynchronous metadata is faster than the ffs synchronous metadata
290 scheme but less reliable. Both methods are equally resolvable by their
291 respective fsck programs.
293 If you're exceptionally paranoid, there are 3 ways of making metadata
294 writes synchronous on ext2:
296 per-file if you have the program source: use the O_SYNC flag to open()
297 per-file if you don't have the source: use "chattr +S" on the file
298 per-filesystem: add the "sync" option to mount (or in /etc/fstab)
300 the first and last are not ext2 specific but do force the metadata to
301 be written synchronously. See also Journaling below.
306 There are various limits imposed by the on-disk layout of ext2. Other
307 limits are imposed by the current implementation of the kernel code.
308 Many of the limits are determined at the time the filesystem is first
309 created, and depend upon the block size chosen. The ratio of inodes to
310 data blocks is fixed at filesystem creation time, so the only way to
311 increase the number of inodes is to increase the size of the filesystem.
312 No tools currently exist which can change the ratio of inodes to blocks.
314 Most of these limits could be overcome with slight changes in the on-disk
315 format and using a compatibility flag to signal the format change (at
316 the expense of some compatibility).
318 Filesystem block size: 1kB 2kB 4kB 8kB
320 File size limit: 16GB 256GB 2048GB 2048GB
321 Filesystem size limit: 2047GB 8192GB 16384GB 32768GB
323 There is a 2.4 kernel limit of 2048GB for a single block device, so no
324 filesystem larger than that can be created at this time. There is also
325 an upper limit on the block size imposed by the page size of the kernel,
326 so 8kB blocks are only allowed on Alpha systems (and other architectures
327 which support larger pages).
329 There is an upper limit of 32768 subdirectories in a single directory.
331 There is a "soft" upper limit of about 10-15k files in a single directory
332 with the current linear linked-list directory implementation. This limit
333 stems from performance problems when creating and deleting (and also
334 finding) files in such large directories. Using a hashed directory index
335 (under development) allows 100k-1M+ files in a single directory without
336 performance problems (although RAM size becomes an issue at this point).
338 The (meaningless) absolute upper limit of files in a single directory
339 (imposed by the file size, the realistic limit is obviously much less)
340 is over 130 trillion files. It would be higher except there are not
341 enough 4-character names to make up unique directory entries, so they
342 have to be 8 character filenames, even then we are fairly close to
343 running out of unique filenames.
348 A journaling extension to the ext2 code has been developed by Stephen
349 Tweedie. It avoids the risks of metadata corruption and the need to
350 wait for e2fsck to complete after a crash, without requiring a change
351 to the on-disk ext2 layout. In a nutshell, the journal is a regular
352 file which stores whole metadata (and optionally data) blocks that have
353 been modified, prior to writing them into the filesystem. This means
354 it is possible to add a journal to an existing ext2 filesystem without
355 the need for data conversion.
357 When changes to the filesystem (e.g. a file is renamed) they are stored in
358 a transaction in the journal and can either be complete or incomplete at
359 the time of a crash. If a transaction is complete at the time of a crash
360 (or in the normal case where the system does not crash), then any blocks
361 in that transaction are guaranteed to represent a valid filesystem state,
362 and are copied into the filesystem. If a transaction is incomplete at
363 the time of the crash, then there is no guarantee of consistency for
364 the blocks in that transaction so they are discarded (which means any
365 filesystem changes they represent are also lost).
366 Check Documentation/filesystems/ext3.txt if you want to read more about
372 The kernel source file:/usr/src/linux/fs/ext2/
373 e2fsprogs (e2fsck) http://e2fsprogs.sourceforge.net/
374 Design & Implementation http://e2fsprogs.sourceforge.net/ext2intro.html
375 Journaling (ext3) ftp://ftp.uk.linux.org/pub/linux/sct/fs/jfs/
376 Hashed Directories http://kernelnewbies.org/~phillips/htree/
377 Filesystem Resizing http://ext2resize.sourceforge.net/
378 Compression (*) http://www.netspace.net.au/~reiter/e2compr/
381 Windows 95/98/NT/2000 http://uranus.it.swin.edu.au/~jn/linux/Explore2fs.htm
382 Windows 95 (*) http://www.yipton.demon.co.uk/content.html#FSDEXT2
383 DOS client (*) ftp://metalab.unc.edu/pub/Linux/system/filesystems/ext2/
384 OS/2 http://perso.wanadoo.fr/matthieu.willm/ext2-os2/
385 RISC OS client ftp://ftp.barnet.ac.uk/pub/acorn/armlinux/iscafs/
387 (*) no longer actively developed/supported (as of Apr 2001)
git://git.onelab.eu / linux-2.6.git / blob