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
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=none, nocheck (*) Don't do extra checking of bitmaps on mount
21
(check=normal and check=strict options removed)
23
debug Extra debugging information is sent to the
24
kernel syslog. Useful for developers.
26
errors=continue Keep going on a filesystem error.
27
errors=remount-ro Remount the filesystem read-only on an error.
28
errors=panic Panic and halt the machine if an error occurs.
30
grpid, bsdgroups Give objects the same group ID as their parent.
31
nogrpid, sysvgroups New objects have the group ID of their creator.
33
nouid32 Use 16-bit UIDs and GIDs.
35
oldalloc Enable the old block allocator. Orlov should
36
have better performance, we'd like to get some
37
feedback if it's the contrary for you.
38
orlov (*) Use the Orlov block allocator.
39
(See http://lwn.net/Articles/14633/ and
40
http://lwn.net/Articles/14446/.)
42
resuid=n The user ID which may use the reserved blocks.
43
resgid=n The group ID which may use the reserved blocks.
45
sb=n Use alternate superblock at this location.
47
user_xattr Enable "user." POSIX Extended Attributes
48
(requires CONFIG_EXT2_FS_XATTR).
49
See also http://acl.bestbits.at
50
nouser_xattr Don't support "user." extended attributes.
52
acl Enable POSIX Access Control Lists support
53
(requires CONFIG_EXT2_FS_POSIX_ACL).
54
See also http://acl.bestbits.at
55
noacl Don't support POSIX ACLs.
57
nobh Do not attach buffer_heads to file pagecache.
59
xip Use execute in place (no caching) if possible
61
grpquota,noquota,quota,usrquota Quota options are silently ignored by ext2.
67
ext2 shares many properties with traditional Unix filesystems. It has
68
the concepts of blocks, inodes and directories. It has space in the
69
specification for Access Control Lists (ACLs), fragments, undeletion and
70
compression though these are not yet implemented (some are available as
71
separate patches). There is also a versioning mechanism to allow new
72
features (such as journalling) to be added in a maximally compatible
78
The space in the device or file is split up into blocks. These are
79
a fixed size, of 1024, 2048 or 4096 bytes (8192 bytes on Alpha systems),
80
which is decided when the filesystem is created. Smaller blocks mean
81
less wasted space per file, but require slightly more accounting overhead,
82
and also impose other limits on the size of files and the filesystem.
87
Blocks are clustered into block groups in order to reduce fragmentation
88
and minimise the amount of head seeking when reading a large amount
89
of consecutive data. Information about each block group is kept in a
90
descriptor table stored in the block(s) immediately after the superblock.
91
Two blocks near the start of each group are reserved for the block usage
92
bitmap and the inode usage bitmap which show which blocks and inodes
93
are in use. Since each bitmap is limited to a single block, this means
94
that the maximum size of a block group is 8 times the size of a block.
96
The block(s) following the bitmaps in each block group are designated
97
as the inode table for that block group and the remainder are the data
98
blocks. The block allocation algorithm attempts to allocate data blocks
99
in the same block group as the inode which contains them.
104
The superblock contains all the information about the configuration of
105
the filing system. The primary copy of the superblock is stored at an
106
offset of 1024 bytes from the start of the device, and it is essential
107
to mounting the filesystem. Since it is so important, backup copies of
108
the superblock are stored in block groups throughout the filesystem.
109
The first version of ext2 (revision 0) stores a copy at the start of
110
every block group, along with backups of the group descriptor block(s).
111
Because this can consume a considerable amount of space for large
112
filesystems, later revisions can optionally reduce the number of backup
113
copies by only putting backups in specific groups (this is the sparse
114
superblock feature). The groups chosen are 0, 1 and powers of 3, 5 and 7.
116
The information in the superblock contains fields such as the total
117
number of inodes and blocks in the filesystem and how many are free,
118
how many inodes and blocks are in each block group, when the filesystem
119
was mounted (and if it was cleanly unmounted), when it was modified,
120
what version of the filesystem it is (see the Revisions section below)
121
and which OS created it.
123
If the filesystem is revision 1 or higher, then there are extra fields,
124
such as a volume name, a unique identification number, the inode size,
125
and space for optional filesystem features to store configuration info.
127
All fields in the superblock (as in all other ext2 structures) are stored
128
on the disc in little endian format, so a filesystem is portable between
129
machines without having to know what machine it was created on.
134
The inode (index node) is a fundamental concept in the ext2 filesystem.
135
Each object in the filesystem is represented by an inode. The inode
136
structure contains pointers to the filesystem blocks which contain the
137
data held in the object and all of the metadata about an object except
138
its name. The metadata about an object includes the permissions, owner,
139
group, flags, size, number of blocks used, access time, change time,
140
modification time, deletion time, number of links, fragments, version
141
(for NFS) and extended attributes (EAs) and/or Access Control Lists (ACLs).
143
There are some reserved fields which are currently unused in the inode
144
structure and several which are overloaded. One field is reserved for the
145
directory ACL if the inode is a directory and alternately for the top 32
146
bits of the file size if the inode is a regular file (allowing file sizes
147
larger than 2GB). The translator field is unused under Linux, but is used
148
by the HURD to reference the inode of a program which will be used to
149
interpret this object. Most of the remaining reserved fields have been
150
used up for both Linux and the HURD for larger owner and group fields,
151
The HURD also has a larger mode field so it uses another of the remaining
152
fields to store the extra more bits.
154
There are pointers to the first 12 blocks which contain the file's data
155
in the inode. There is a pointer to an indirect block (which contains
156
pointers to the next set of blocks), a pointer to a doubly-indirect
157
block (which contains pointers to indirect blocks) and a pointer to a
158
trebly-indirect block (which contains pointers to doubly-indirect blocks).
160
The flags field contains some ext2-specific flags which aren't catered
161
for by the standard chmod flags. These flags can be listed with lsattr
162
and changed with the chattr command, and allow specific filesystem
163
behaviour on a per-file basis. There are flags for secure deletion,
164
undeletable, compression, synchronous updates, immutability, append-only,
165
dumpable, no-atime, indexed directories, and data-journaling. Not all
166
of these are supported yet.
171
A directory is a filesystem object and has an inode just like a file.
172
It is a specially formatted file containing records which associate
173
each name with an inode number. Later revisions of the filesystem also
174
encode the type of the object (file, directory, symlink, device, fifo,
175
socket) to avoid the need to check the inode itself for this information
176
(support for taking advantage of this feature does not yet exist in
179
The inode allocation code tries to assign inodes which are in the same
180
block group as the directory in which they are first created.
182
The current implementation of ext2 uses a singly-linked list to store
183
the filenames in the directory; a pending enhancement uses hashing of the
184
filenames to allow lookup without the need to scan the entire directory.
186
The current implementation never removes empty directory blocks once they
187
have been allocated to hold more files.
192
Symbolic links are also filesystem objects with inodes. They deserve
193
special mention because the data for them is stored within the inode
194
itself if the symlink is less than 60 bytes long. It uses the fields
195
which would normally be used to store the pointers to data blocks.
196
This is a worthwhile optimisation as it we avoid allocating a full
197
block for the symlink, and most symlinks are less than 60 characters long.
199
Character and block special devices never have data blocks assigned to
200
them. Instead, their device number is stored in the inode, again reusing
201
the fields which would be used to point to the data blocks.
206
In ext2, there is a mechanism for reserving a certain number of blocks
207
for a particular user (normally the super-user). This is intended to
208
allow for the system to continue functioning even if non-privileged users
209
fill up all the space available to them (this is independent of filesystem
210
quotas). It also keeps the filesystem from filling up entirely which
211
helps combat fragmentation.
216
At boot time, most systems run a consistency check (e2fsck) on their
217
filesystems. The superblock of the ext2 filesystem contains several
218
fields which indicate whether fsck should actually run (since checking
219
the filesystem at boot can take a long time if it is large). fsck will
220
run if the filesystem was not cleanly unmounted, if the maximum mount
221
count has been exceeded or if the maximum time between checks has been
224
Feature Compatibility
225
---------------------
227
The compatibility feature mechanism used in ext2 is sophisticated.
228
It safely allows features to be added to the filesystem, without
229
unnecessarily sacrificing compatibility with older versions of the
230
filesystem code. The feature compatibility mechanism is not supported by
231
the original revision 0 (EXT2_GOOD_OLD_REV) of ext2, but was introduced in
232
revision 1. There are three 32-bit fields, one for compatible features
233
(COMPAT), one for read-only compatible (RO_COMPAT) features and one for
234
incompatible (INCOMPAT) features.
236
These feature flags have specific meanings for the kernel as follows:
238
A COMPAT flag indicates that a feature is present in the filesystem,
239
but the on-disk format is 100% compatible with older on-disk formats, so
240
a kernel which didn't know anything about this feature could read/write
241
the filesystem without any chance of corrupting the filesystem (or even
242
making it inconsistent). This is essentially just a flag which says
243
"this filesystem has a (hidden) feature" that the kernel or e2fsck may
244
want to be aware of (more on e2fsck and feature flags later). The ext3
245
HAS_JOURNAL feature is a COMPAT flag because the ext3 journal is simply
246
a regular file with data blocks in it so the kernel does not need to
247
take any special notice of it if it doesn't understand ext3 journaling.
249
An RO_COMPAT flag indicates that the on-disk format is 100% compatible
250
with older on-disk formats for reading (i.e. the feature does not change
251
the visible on-disk format). However, an old kernel writing to such a
252
filesystem would/could corrupt the filesystem, so this is prevented. The
253
most common such feature, SPARSE_SUPER, is an RO_COMPAT feature because
254
sparse groups allow file data blocks where superblock/group descriptor
255
backups used to live, and ext2_free_blocks() refuses to free these blocks,
256
which would leading to inconsistent bitmaps. An old kernel would also
257
get an error if it tried to free a series of blocks which crossed a group
258
boundary, but this is a legitimate layout in a SPARSE_SUPER filesystem.
260
An INCOMPAT flag indicates the on-disk format has changed in some
261
way that makes it unreadable by older kernels, or would otherwise
262
cause a problem if an old kernel tried to mount it. FILETYPE is an
263
INCOMPAT flag because older kernels would think a filename was longer
264
than 256 characters, which would lead to corrupt directory listings.
265
The COMPRESSION flag is an obvious INCOMPAT flag - if the kernel
266
doesn't understand compression, you would just get garbage back from
267
read() instead of it automatically decompressing your data. The ext3
268
RECOVER flag is needed to prevent a kernel which does not understand the
269
ext3 journal from mounting the filesystem without replaying the journal.
271
For e2fsck, it needs to be more strict with the handling of these
272
flags than the kernel. If it doesn't understand ANY of the COMPAT,
273
RO_COMPAT, or INCOMPAT flags it will refuse to check the filesystem,
274
because it has no way of verifying whether a given feature is valid
275
or not. Allowing e2fsck to succeed on a filesystem with an unknown
276
feature is a false sense of security for the user. Refusing to check
277
a filesystem with unknown features is a good incentive for the user to
278
update to the latest e2fsck. This also means that anyone adding feature
279
flags to ext2 also needs to update e2fsck to verify these features.
284
It is frequently claimed that the ext2 implementation of writing
285
asynchronous metadata is faster than the ffs synchronous metadata
286
scheme but less reliable. Both methods are equally resolvable by their
287
respective fsck programs.
289
If you're exceptionally paranoid, there are 3 ways of making metadata
290
writes synchronous on ext2:
292
per-file if you have the program source: use the O_SYNC flag to open()
293
per-file if you don't have the source: use "chattr +S" on the file
294
per-filesystem: add the "sync" option to mount (or in /etc/fstab)
296
the first and last are not ext2 specific but do force the metadata to
297
be written synchronously. See also Journaling below.
302
There are various limits imposed by the on-disk layout of ext2. Other
303
limits are imposed by the current implementation of the kernel code.
304
Many of the limits are determined at the time the filesystem is first
305
created, and depend upon the block size chosen. The ratio of inodes to
306
data blocks is fixed at filesystem creation time, so the only way to
307
increase the number of inodes is to increase the size of the filesystem.
308
No tools currently exist which can change the ratio of inodes to blocks.
310
Most of these limits could be overcome with slight changes in the on-disk
311
format and using a compatibility flag to signal the format change (at
312
the expense of some compatibility).
314
Filesystem block size: 1kB 2kB 4kB 8kB
316
File size limit: 16GB 256GB 2048GB 2048GB
317
Filesystem size limit: 2047GB 8192GB 16384GB 32768GB
319
There is a 2.4 kernel limit of 2048GB for a single block device, so no
320
filesystem larger than that can be created at this time. There is also
321
an upper limit on the block size imposed by the page size of the kernel,
322
so 8kB blocks are only allowed on Alpha systems (and other architectures
323
which support larger pages).
325
There is an upper limit of 32000 subdirectories in a single directory.
327
There is a "soft" upper limit of about 10-15k files in a single directory
328
with the current linear linked-list directory implementation. This limit
329
stems from performance problems when creating and deleting (and also
330
finding) files in such large directories. Using a hashed directory index
331
(under development) allows 100k-1M+ files in a single directory without
332
performance problems (although RAM size becomes an issue at this point).
334
The (meaningless) absolute upper limit of files in a single directory
335
(imposed by the file size, the realistic limit is obviously much less)
336
is over 130 trillion files. It would be higher except there are not
337
enough 4-character names to make up unique directory entries, so they
338
have to be 8 character filenames, even then we are fairly close to
339
running out of unique filenames.
344
A journaling extension to the ext2 code has been developed by Stephen
345
Tweedie. It avoids the risks of metadata corruption and the need to
346
wait for e2fsck to complete after a crash, without requiring a change
347
to the on-disk ext2 layout. In a nutshell, the journal is a regular
348
file which stores whole metadata (and optionally data) blocks that have
349
been modified, prior to writing them into the filesystem. This means
350
it is possible to add a journal to an existing ext2 filesystem without
351
the need for data conversion.
353
When changes to the filesystem (e.g. a file is renamed) they are stored in
354
a transaction in the journal and can either be complete or incomplete at
355
the time of a crash. If a transaction is complete at the time of a crash
356
(or in the normal case where the system does not crash), then any blocks
357
in that transaction are guaranteed to represent a valid filesystem state,
358
and are copied into the filesystem. If a transaction is incomplete at
359
the time of the crash, then there is no guarantee of consistency for
360
the blocks in that transaction so they are discarded (which means any
361
filesystem changes they represent are also lost).
362
Check Documentation/filesystems/ext3.txt if you want to read more about
368
The kernel source file:/usr/src/linux/fs/ext2/
369
e2fsprogs (e2fsck) http://e2fsprogs.sourceforge.net/
370
Design & Implementation http://e2fsprogs.sourceforge.net/ext2intro.html
371
Journaling (ext3) ftp://ftp.uk.linux.org/pub/linux/sct/fs/jfs/
372
Filesystem Resizing http://ext2resize.sourceforge.net/
373
Compression (*) http://e2compr.sourceforge.net/
376
Windows 95/98/NT/2000 http://www.chrysocome.net/explore2fs
377
Windows 95 (*) http://www.yipton.net/content.html#FSDEXT2
378
DOS client (*) ftp://metalab.unc.edu/pub/Linux/system/filesystems/ext2/
379
OS/2 (+) ftp://metalab.unc.edu/pub/Linux/system/filesystems/ext2/
380
RISC OS client http://www.esw-heim.tu-clausthal.de/~marco/smorbrod/IscaFS/
382
(*) no longer actively developed/supported (as of Apr 2001)
383
(+) no longer actively developed/supported (as of Mar 2009)