| blob | d16334ec48ba02cc719d9072dd800f12d9ed17cb |
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.
11 Options
12 =======
14 Most defaults are determined by the filesystem superblock, and can be
15 set using tune2fs(8). Kernel-determined defaults are indicated by (*).
17 bsddf (*) Makes `df' act like BSD.
18 minixdf Makes `df' act like Minix.
20 check Check block and inode bitmaps at mount time
21 (requires CONFIG_EXT2_CHECK).
22 check=none, nocheck (*) Don't do extra checking of bitmaps on mount
23 (check=normal and check=strict options removed)
25 debug Extra debugging information is sent to the
26 kernel syslog. Useful for developers.
28 errors=continue Keep going on a filesystem error.
29 errors=remount-ro Remount the filesystem read-only on an error.
30 errors=panic Panic and halt the machine if an error occurs.
32 grpid, bsdgroups Give objects the same group ID as their parent.
33 nogrpid, sysvgroups New objects have the group ID of their creator.
35 nouid32 Use 16-bit UIDs and GIDs.
37 oldalloc Enable the old block allocator. Orlov should
38 have better performance, we'd like to get some
39 feedback if it's the contrary for you.
40 orlov (*) Use the Orlov block allocator.
41 (See http://lwn.net/Articles/14633/ and
42 http://lwn.net/Articles/14446/.)
44 resuid=n The user ID which may use the reserved blocks.
45 resgid=n The group ID which may use the reserved blocks.
47 sb=n Use alternate superblock at this location.
49 user_xattr Enable "user." POSIX Extended Attributes
50 (requires CONFIG_EXT2_FS_XATTR).
51 See also http://acl.bestbits.at
52 nouser_xattr Don't support "user." extended attributes.
54 acl Enable POSIX Access Control Lists support
55 (requires CONFIG_EXT2_FS_POSIX_ACL).
56 See also http://acl.bestbits.at
57 noacl Don't support POSIX ACLs.
59 nobh Do not attach buffer_heads to file pagecache.
61 xip Use execute in place (no caching) if possible
63 grpquota,noquota,quota,usrquota Quota options are silently ignored by ext2.
66 Specification
67 =============
69 ext2 shares many properties with traditional Unix filesystems. It has
70 the concepts of blocks, inodes and directories. It has space in the
71 specification for Access Control Lists (ACLs), fragments, undeletion and
72 compression though these are not yet implemented (some are available as
73 separate patches). There is also a versioning mechanism to allow new
74 features (such as journalling) to be added in a maximally compatible
75 manner.
77 Blocks
78 ------
80 The space in the device or file is split up into blocks. These are
81 a fixed size, of 1024, 2048 or 4096 bytes (8192 bytes on Alpha systems),
82 which is decided when the filesystem is created. Smaller blocks mean
83 less wasted space per file, but require slightly more accounting overhead,
84 and also impose other limits on the size of files and the filesystem.
86 Block Groups
87 ------------
89 Blocks are clustered into block groups in order to reduce fragmentation
90 and minimise the amount of head seeking when reading a large amount
91 of consecutive data. Information about each block group is kept in a
92 descriptor table stored in the block(s) immediately after the superblock.
93 Two blocks near the start of each group are reserved for the block usage
94 bitmap and the inode usage bitmap which show which blocks and inodes
95 are in use. Since each bitmap is limited to a single block, this means
96 that the maximum size of a block group is 8 times the size of a block.
98 The block(s) following the bitmaps in each block group are designated
99 as the inode table for that block group and the remainder are the data
100 blocks. The block allocation algorithm attempts to allocate data blocks
101 in the same block group as the inode which contains them.
103 The Superblock
104 --------------
106 The superblock contains all the information about the configuration of
107 the filing system. The primary copy of the superblock is stored at an
108 offset of 1024 bytes from the start of the device, and it is essential
109 to mounting the filesystem. Since it is so important, backup copies of
110 the superblock are stored in block groups throughout the filesystem.
111 The first version of ext2 (revision 0) stores a copy at the start of
112 every block group, along with backups of the group descriptor block(s).
113 Because this can consume a considerable amount of space for large
114 filesystems, later revisions can optionally reduce the number of backup
115 copies by only putting backups in specific groups (this is the sparse
116 superblock feature). The groups chosen are 0, 1 and powers of 3, 5 and 7.
118 The information in the superblock contains fields such as the total
119 number of inodes and blocks in the filesystem and how many are free,
120 how many inodes and blocks are in each block group, when the filesystem
121 was mounted (and if it was cleanly unmounted), when it was modified,
122 what version of the filesystem it is (see the Revisions section below)
123 and which OS created it.
125 If the filesystem is revision 1 or higher, then there are extra fields,
126 such as a volume name, a unique identification number, the inode size,
127 and space for optional filesystem features to store configuration info.
129 All fields in the superblock (as in all other ext2 structures) are stored
130 on the disc in little endian format, so a filesystem is portable between
131 machines without having to know what machine it was created on.
133 Inodes
134 ------
136 The inode (index node) is a fundamental concept in the ext2 filesystem.
137 Each object in the filesystem is represented by an inode. The inode
138 structure contains pointers to the filesystem blocks which contain the
139 data held in the object and all of the metadata about an object except
140 its name. The metadata about an object includes the permissions, owner,
141 group, flags, size, number of blocks used, access time, change time,
142 modification time, deletion time, number of links, fragments, version
143 (for NFS) and extended attributes (EAs) and/or Access Control Lists (ACLs).
145 There are some reserved fields which are currently unused in the inode
146 structure and several which are overloaded. One field is reserved for the
147 directory ACL if the inode is a directory and alternately for the top 32
148 bits of the file size if the inode is a regular file (allowing file sizes
149 larger than 2GB). The translator field is unused under Linux, but is used
150 by the HURD to reference the inode of a program which will be used to
151 interpret this object. Most of the remaining reserved fields have been
152 used up for both Linux and the HURD for larger owner and group fields,
153 The HURD also has a larger mode field so it uses another of the remaining
154 fields to store the extra more bits.
156 There are pointers to the first 12 blocks which contain the file's data
157 in the inode. There is a pointer to an indirect block (which contains
158 pointers to the next set of blocks), a pointer to a doubly-indirect
159 block (which contains pointers to indirect blocks) and a pointer to a
160 trebly-indirect block (which contains pointers to doubly-indirect blocks).
162 The flags field contains some ext2-specific flags which aren't catered
163 for by the standard chmod flags. These flags can be listed with lsattr
164 and changed with the chattr command, and allow specific filesystem
165 behaviour on a per-file basis. There are flags for secure deletion,
166 undeletable, compression, synchronous updates, immutability, append-only,
167 dumpable, no-atime, indexed directories, and data-journaling. Not all
168 of these are supported yet.
170 Directories
171 -----------
173 A directory is a filesystem object and has an inode just like a file.
174 It is a specially formatted file containing records which associate
175 each name with an inode number. Later revisions of the filesystem also
176 encode the type of the object (file, directory, symlink, device, fifo,
177 socket) to avoid the need to check the inode itself for this information
178 (support for taking advantage of this feature does not yet exist in
179 Glibc 2.2).
181 The inode allocation code tries to assign inodes which are in the same
182 block group as the directory in which they are first created.
184 The current implementation of ext2 uses a singly-linked list to store
185 the filenames in the directory; a pending enhancement uses hashing of the
186 filenames to allow lookup without the need to scan the entire directory.
188 The current implementation never removes empty directory blocks once they
189 have been allocated to hold more files.
191 Special files
192 -------------
194 Symbolic links are also filesystem objects with inodes. They deserve
195 special mention because the data for them is stored within the inode
196 itself if the symlink is less than 60 bytes long. It uses the fields
197 which would normally be used to store the pointers to data blocks.
198 This is a worthwhile optimisation as it we avoid allocating a full
199 block for the symlink, and most symlinks are less than 60 characters long.
201 Character and block special devices never have data blocks assigned to
202 them. Instead, their device number is stored in the inode, again reusing
203 the fields which would be used to point to the data blocks.
205 Reserved Space
206 --------------
208 In ext2, there is a mechanism for reserving a certain number of blocks
209 for a particular user (normally the super-user). This is intended to
210 allow for the system to continue functioning even if non-priveleged users
211 fill up all the space available to them (this is independent of filesystem
212 quotas). It also keeps the filesystem from filling up entirely which
213 helps combat fragmentation.
215 Filesystem check
216 ----------------
218 At boot time, most systems run a consistency check (e2fsck) on their
219 filesystems. The superblock of the ext2 filesystem contains several
220 fields which indicate whether fsck should actually run (since checking
221 the filesystem at boot can take a long time if it is large). fsck will
222 run if the filesystem was not cleanly unmounted, if the maximum mount
223 count has been exceeded or if the maximum time between checks has been
224 exceeded.
226 Feature Compatibility
227 ---------------------
229 The compatibility feature mechanism used in ext2 is sophisticated.
230 It safely allows features to be added to the filesystem, without
231 unnecessarily sacrificing compatibility with older versions of the
232 filesystem code. The feature compatibility mechanism is not supported by
233 the original revision 0 (EXT2_GOOD_OLD_REV) of ext2, but was introduced in
234 revision 1. There are three 32-bit fields, one for compatible features
235 (COMPAT), one for read-only compatible (RO_COMPAT) features and one for
236 incompatible (INCOMPAT) features.
238 These feature flags have specific meanings for the kernel as follows:
240 A COMPAT flag indicates that a feature is present in the filesystem,
241 but the on-disk format is 100% compatible with older on-disk formats, so
242 a kernel which didn't know anything about this feature could read/write
243 the filesystem without any chance of corrupting the filesystem (or even
244 making it inconsistent). This is essentially just a flag which says
245 "this filesystem has a (hidden) feature" that the kernel or e2fsck may
246 want to be aware of (more on e2fsck and feature flags later). The ext3
247 HAS_JOURNAL feature is a COMPAT flag because the ext3 journal is simply
248 a regular file with data blocks in it so the kernel does not need to
249 take any special notice of it if it doesn't understand ext3 journaling.
251 An RO_COMPAT flag indicates that the on-disk format is 100% compatible
252 with older on-disk formats for reading (i.e. the feature does not change
253 the visible on-disk format). However, an old kernel writing to such a
254 filesystem would/could corrupt the filesystem, so this is prevented. The
255 most common such feature, SPARSE_SUPER, is an RO_COMPAT feature because
256 sparse groups allow file data blocks where superblock/group descriptor
257 backups used to live, and ext2_free_blocks() refuses to free these blocks,
258 which would leading to inconsistent bitmaps. An old kernel would also
259 get an error if it tried to free a series of blocks which crossed a group
260 boundary, but this is a legitimate layout in a SPARSE_SUPER filesystem.
262 An INCOMPAT flag indicates the on-disk format has changed in some
263 way that makes it unreadable by older kernels, or would otherwise
264 cause a problem if an old kernel tried to mount it. FILETYPE is an
265 INCOMPAT flag because older kernels would think a filename was longer
266 than 256 characters, which would lead to corrupt directory listings.
267 The COMPRESSION flag is an obvious INCOMPAT flag - if the kernel
268 doesn't understand compression, you would just get garbage back from
269 read() instead of it automatically decompressing your data. The ext3
270 RECOVER flag is needed to prevent a kernel which does not understand the
271 ext3 journal from mounting the filesystem without replaying the journal.
273 For e2fsck, it needs to be more strict with the handling of these
274 flags than the kernel. If it doesn't understand ANY of the COMPAT,
275 RO_COMPAT, or INCOMPAT flags it will refuse to check the filesystem,
276 because it has no way of verifying whether a given feature is valid
277 or not. Allowing e2fsck to succeed on a filesystem with an unknown
278 feature is a false sense of security for the user. Refusing to check
279 a filesystem with unknown features is a good incentive for the user to
280 update to the latest e2fsck. This also means that anyone adding feature
281 flags to ext2 also needs to update e2fsck to verify these features.
283 Metadata
284 --------
286 It is frequently claimed that the ext2 implementation of writing
287 asynchronous metadata is faster than the ffs synchronous metadata
288 scheme but less reliable. Both methods are equally resolvable by their
289 respective fsck programs.
291 If you're exceptionally paranoid, there are 3 ways of making metadata
292 writes synchronous on ext2:
294 per-file if you have the program source: use the O_SYNC flag to open()
295 per-file if you don't have the source: use "chattr +S" on the file
296 per-filesystem: add the "sync" option to mount (or in /etc/fstab)
298 the first and last are not ext2 specific but do force the metadata to
299 be written synchronously. See also Journaling below.
301 Limitations
302 -----------
304 There are various limits imposed by the on-disk layout of ext2. Other
305 limits are imposed by the current implementation of the kernel code.
306 Many of the limits are determined at the time the filesystem is first
307 created, and depend upon the block size chosen. The ratio of inodes to
308 data blocks is fixed at filesystem creation time, so the only way to
309 increase the number of inodes is to increase the size of the filesystem.
310 No tools currently exist which can change the ratio of inodes to blocks.
312 Most of these limits could be overcome with slight changes in the on-disk
313 format and using a compatibility flag to signal the format change (at
314 the expense of some compatibility).
316 Filesystem block size: 1kB 2kB 4kB 8kB
318 File size limit: 16GB 256GB 2048GB 2048GB
319 Filesystem size limit: 2047GB 8192GB 16384GB 32768GB
321 There is a 2.4 kernel limit of 2048GB for a single block device, so no
322 filesystem larger than that can be created at this time. There is also
323 an upper limit on the block size imposed by the page size of the kernel,
324 so 8kB blocks are only allowed on Alpha systems (and other architectures
325 which support larger pages).
327 There is an upper limit of 32768 subdirectories in a single directory.
329 There is a "soft" upper limit of about 10-15k files in a single directory
330 with the current linear linked-list directory implementation. This limit
331 stems from performance problems when creating and deleting (and also
332 finding) files in such large directories. Using a hashed directory index
333 (under development) allows 100k-1M+ files in a single directory without
334 performance problems (although RAM size becomes an issue at this point).
336 The (meaningless) absolute upper limit of files in a single directory
337 (imposed by the file size, the realistic limit is obviously much less)
338 is over 130 trillion files. It would be higher except there are not
339 enough 4-character names to make up unique directory entries, so they
340 have to be 8 character filenames, even then we are fairly close to
341 running out of unique filenames.
343 Journaling
344 ----------
346 A journaling extension to the ext2 code has been developed by Stephen
347 Tweedie. It avoids the risks of metadata corruption and the need to
348 wait for e2fsck to complete after a crash, without requiring a change
349 to the on-disk ext2 layout. In a nutshell, the journal is a regular
350 file which stores whole metadata (and optionally data) blocks that have
351 been modified, prior to writing them into the filesystem. This means
352 it is possible to add a journal to an existing ext2 filesystem without
353 the need for data conversion.
355 When changes to the filesystem (e.g. a file is renamed) they are stored in
356 a transaction in the journal and can either be complete or incomplete at
357 the time of a crash. If a transaction is complete at the time of a crash
358 (or in the normal case where the system does not crash), then any blocks
359 in that transaction are guaranteed to represent a valid filesystem state,
360 and are copied into the filesystem. If a transaction is incomplete at
361 the time of the crash, then there is no guarantee of consistency for
362 the blocks in that transaction so they are discarded (which means any
363 filesystem changes they represent are also lost).
364 Check Documentation/filesystems/ext3.txt if you want to read more about
365 ext3 and journaling.
367 References
368 ==========
370 The kernel source file:/usr/src/linux/fs/ext2/
371 e2fsprogs (e2fsck) http://e2fsprogs.sourceforge.net/
372 Design & Implementation http://e2fsprogs.sourceforge.net/ext2intro.html
373 Journaling (ext3) ftp://ftp.uk.linux.org/pub/linux/sct/fs/jfs/
374 Hashed Directories http://kernelnewbies.org/~phillips/htree/
375 Filesystem Resizing http://ext2resize.sourceforge.net/
376 Compression (*) http://www.netspace.net.au/~reiter/e2compr/
378 Implementations for:
379 Windows 95/98/NT/2000 http://uranus.it.swin.edu.au/~jn/linux/Explore2fs.htm
380 Windows 95 (*) http://www.yipton.demon.co.uk/content.html#FSDEXT2
381 DOS client (*) ftp://metalab.unc.edu/pub/Linux/system/filesystems/ext2/
382 OS/2 http://perso.wanadoo.fr/matthieu.willm/ext2-os2/
383 RISC OS client ftp://ftp.barnet.ac.uk/pub/acorn/armlinux/iscafs/
385 (*) no longer actively developed/supported (as of Apr 2001)