blob: 272f80d5f966741c567b3259ac6755a9055373c4 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001
2Ext3 Filesystem
3===============
4
Jesper Juhlc63ca3c2006-01-09 20:53:57 -08005Ext3 was originally released in September 1999. Written by Stephen Tweedie
6for the 2.2 branch, and ported to 2.4 kernels by Peter Braam, Andreas Dilger,
Linus Torvalds1da177e2005-04-16 15:20:36 -07007Andrew Morton, Alexander Viro, Ted Ts'o and Stephen Tweedie.
8
Jesper Juhlc63ca3c2006-01-09 20:53:57 -08009Ext3 is the ext2 filesystem enhanced with journalling capabilities.
Linus Torvalds1da177e2005-04-16 15:20:36 -070010
11Options
12=======
13
14When mounting an ext3 filesystem, the following option are accepted:
15(*) == default
16
Pavel Macheke3375ac2009-01-10 00:47:21 +010017ro Mount filesystem read only. Note that ext3 will replay
18 the journal (and thus write to the partition) even when
19 mounted "read only". Mount options "ro,noload" can be
20 used to prevent writes to the filesystem.
21
Jesper Juhlc63ca3c2006-01-09 20:53:57 -080022journal=update Update the ext3 file system's journal to the current
23 format.
Linus Torvalds1da177e2005-04-16 15:20:36 -070024
Jesper Juhlc63ca3c2006-01-09 20:53:57 -080025journal=inum When a journal already exists, this option is ignored.
26 Otherwise, it specifies the number of the inode which
27 will represent the ext3 file system's journal file.
Linus Torvalds1da177e2005-04-16 15:20:36 -070028
Johann Lombardi71b96252006-01-08 01:03:20 -080029journal_dev=devnum When the external journal device's major/minor numbers
Jesper Juhlc63ca3c2006-01-09 20:53:57 -080030 have changed, this option allows the user to specify
31 the new journal location. The journal device is
32 identified through its new major/minor numbers encoded
33 in devnum.
Johann Lombardi71b96252006-01-08 01:03:20 -080034
Eric Sandeendee1d3b2009-11-16 16:50:49 -060035norecovery Don't load the journal on mounting. Note that this forces
36noload mount of inconsistent filesystem, which can lead to
Pavel Macheke3375ac2009-01-10 00:47:21 +010037 various problems.
Linus Torvalds1da177e2005-04-16 15:20:36 -070038
Jesper Juhlc63ca3c2006-01-09 20:53:57 -080039data=journal All data are committed into the journal prior to being
40 written into the main file system.
Linus Torvalds1da177e2005-04-16 15:20:36 -070041
42data=ordered (*) All data are forced directly out to the main file
Jesper Juhlc63ca3c2006-01-09 20:53:57 -080043 system prior to its metadata being committed to the
44 journal.
Linus Torvalds1da177e2005-04-16 15:20:36 -070045
Jesper Juhlc63ca3c2006-01-09 20:53:57 -080046data=writeback Data ordering is not preserved, data may be written
47 into the main file system after its metadata has been
48 committed to the journal.
Linus Torvalds1da177e2005-04-16 15:20:36 -070049
50commit=nrsec (*) Ext3 can be told to sync all its data and metadata
51 every 'nrsec' seconds. The default value is 5 seconds.
Jesper Juhlc63ca3c2006-01-09 20:53:57 -080052 This means that if you lose your power, you will lose
53 as much as the latest 5 seconds of work (your
54 filesystem will not be damaged though, thanks to the
55 journaling). This default value (or any low value)
56 will hurt performance, but it's good for data-safety.
57 Setting it to 0 will have the same effect as leaving
58 it at the default (5 seconds).
Linus Torvalds1da177e2005-04-16 15:20:36 -070059 Setting it to very large values will improve
60 performance.
61
Eric Sandeen0636c732010-04-30 11:09:34 -050062barrier=<0(*)|1> This enables/disables the use of write barriers in
63barrier the jbd code. barrier=0 disables, barrier=1 enables.
64nobarrier (*) This also requires an IO stack which can support
65 barriers, and if jbd gets an error on a barrier
66 write, it will disable again with a warning.
67 Write barriers enforce proper on-disk ordering
68 of journal commits, making volatile disk write caches
69 safe to use, at some performance penalty. If
70 your disks are battery-backed in one way or another,
71 disabling barriers may safely improve performance.
72 The mount options "barrier" and "nobarrier" can
73 also be used to enable or disable barriers, for
74 consistency with other ext3 mount options.
Linus Torvalds1da177e2005-04-16 15:20:36 -070075
Jesper Juhlc63ca3c2006-01-09 20:53:57 -080076orlov (*) This enables the new Orlov block allocator. It is
77 enabled by default.
Linus Torvalds1da177e2005-04-16 15:20:36 -070078
Jesper Juhlc63ca3c2006-01-09 20:53:57 -080079oldalloc This disables the Orlov block allocator and enables
80 the old block allocator. Orlov should have better
81 performance - we'd like to get some feedback if it's
82 the contrary for you.
Linus Torvalds1da177e2005-04-16 15:20:36 -070083
Jesper Juhlc63ca3c2006-01-09 20:53:57 -080084user_xattr Enables Extended User Attributes. Additionally, you
85 need to have extended attribute support enabled in the
86 kernel configuration (CONFIG_EXT3_FS_XATTR). See the
87 attr(5) manual page and http://acl.bestbits.at/ to
88 learn more about extended attributes.
Linus Torvalds1da177e2005-04-16 15:20:36 -070089
Andreas Gruenbacher85b87242005-12-12 00:37:04 -080090nouser_xattr Disables Extended User Attributes.
Linus Torvalds1da177e2005-04-16 15:20:36 -070091
Jesper Juhlc63ca3c2006-01-09 20:53:57 -080092acl Enables POSIX Access Control Lists support.
93 Additionally, you need to have ACL support enabled in
94 the kernel configuration (CONFIG_EXT3_FS_POSIX_ACL).
95 See the acl(5) manual page and http://acl.bestbits.at/
96 for more information.
Linus Torvalds1da177e2005-04-16 15:20:36 -070097
Jesper Juhlc63ca3c2006-01-09 20:53:57 -080098noacl This option disables POSIX Access Control List
99 support.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700100
101reservation
102
103noreservation
104
Linus Torvalds1da177e2005-04-16 15:20:36 -0700105bsddf (*) Make 'df' act like BSD.
106minixdf Make 'df' act like Minix.
107
108check=none Don't do extra checking of bitmaps on mount.
Jesper Juhlc63ca3c2006-01-09 20:53:57 -0800109nocheck
Linus Torvalds1da177e2005-04-16 15:20:36 -0700110
111debug Extra debugging information is sent to syslog.
112
Pavel Macheke3375ac2009-01-10 00:47:21 +0100113errors=remount-ro Remount the filesystem read-only on an error.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700114errors=continue Keep going on a filesystem error.
115errors=panic Panic and halt the machine if an error occurs.
Pavel Macheke3375ac2009-01-10 00:47:21 +0100116 (These mount options override the errors behavior
117 specified in the superblock, which can be
118 configured using tune2fs.)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700119
Hidehiro Kawai0e4fb5e2008-10-18 20:27:57 -0700120data_err=ignore(*) Just print an error message if an error occurs
121 in a file data buffer in ordered mode.
122data_err=abort Abort the journal if an error occurs in a file
123 data buffer in ordered mode.
124
Linus Torvalds1da177e2005-04-16 15:20:36 -0700125grpid Give objects the same group ID as their creator.
Jesper Juhlc63ca3c2006-01-09 20:53:57 -0800126bsdgroups
Linus Torvalds1da177e2005-04-16 15:20:36 -0700127
128nogrpid (*) New objects have the group ID of their creator.
129sysvgroups
130
131resgid=n The group ID which may use the reserved blocks.
132
133resuid=n The user ID which may use the reserved blocks.
134
135sb=n Use alternate superblock at this location.
136
Jan Kara6dbce522009-09-17 17:37:12 +0200137quota These options are ignored by the filesystem. They
138noquota are used only by quota tools to recognize volumes
139grpquota where quota should be turned on. See documentation
140usrquota in the quota-tools package for more details
141 (http://sourceforge.net/projects/linuxquota).
142
143jqfmt=<quota type> These options tell filesystem details about quota
144usrjquota=<file> so that quota information can be properly updated
145grpjquota=<file> during journal replay. They replace the above
146 quota options. See documentation in the quota-tools
147 package for more details
148 (http://sourceforge.net/projects/linuxquota).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700149
Badari Pulavartyade1a292006-06-26 00:25:04 -0700150bh (*) ext3 associates buffer heads to data pages to
151nobh (a) cache disk block mapping information
152 (b) link pages into transaction to provide
153 ordering guarantees.
154 "bh" option forces use of buffer heads.
155 "nobh" option tries to avoid associating buffer
156 heads (supported only for "writeback" mode).
157
Linus Torvalds1da177e2005-04-16 15:20:36 -0700158
159Specification
160=============
Jesper Juhlc63ca3c2006-01-09 20:53:57 -0800161Ext3 shares all disk implementation with the ext2 filesystem, and adds
162transactions capabilities to ext2. Journaling is done by the Journaling Block
163Device layer.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700164
165Journaling Block Device layer
166-----------------------------
Shaun Zinck73563372007-10-20 02:38:36 +0200167The Journaling Block Device layer (JBD) isn't ext3 specific. It was designed
168to add journaling capabilities to a block device. The ext3 filesystem code
169will inform the JBD of modifications it is performing (called a transaction).
170The journal supports the transactions start and stop, and in case of a crash,
171the journal can replay the transactions to quickly put the partition back into
172a consistent state.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700173
Jesper Juhlc63ca3c2006-01-09 20:53:57 -0800174Handles represent a single atomic update to a filesystem. JBD can handle an
175external journal on a block device.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700176
177Data Mode
178---------
Jesper Juhlc63ca3c2006-01-09 20:53:57 -0800179There are 3 different data modes:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700180
181* writeback mode
Jesper Juhlc63ca3c2006-01-09 20:53:57 -0800182In data=writeback mode, ext3 does not journal data at all. This mode provides
183a similar level of journaling as that of XFS, JFS, and ReiserFS in its default
184mode - metadata journaling. A crash+recovery can cause incorrect data to
185appear in files which were written shortly before the crash. This mode will
186typically provide the best ext3 performance.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700187
188* ordered mode
Jesper Juhlc63ca3c2006-01-09 20:53:57 -0800189In data=ordered mode, ext3 only officially journals metadata, but it logically
190groups metadata and data blocks into a single unit called a transaction. When
191it's time to write the new metadata out to disk, the associated data blocks
192are written first. In general, this mode performs slightly slower than
193writeback but significantly faster than journal mode.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700194
195* journal mode
Jesper Juhlc63ca3c2006-01-09 20:53:57 -0800196data=journal mode provides full data and metadata journaling. All new data is
197written to the journal first, and then to its final location.
198In the event of a crash, the journal can be replayed, bringing both data and
199metadata into a consistent state. This mode is the slowest except when data
200needs to be read from and written to disk at the same time where it
Shaun Zinck73563372007-10-20 02:38:36 +0200201outperforms all other modes.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700202
203Compatibility
204-------------
205
206Ext2 partitions can be easily convert to ext3, with `tune2fs -j <dev>`.
Jesper Juhlc63ca3c2006-01-09 20:53:57 -0800207Ext3 is fully compatible with Ext2. Ext3 partitions can easily be mounted as
208Ext2.
209
Linus Torvalds1da177e2005-04-16 15:20:36 -0700210
211External Tools
212==============
Jesper Juhlc63ca3c2006-01-09 20:53:57 -0800213See manual pages to learn more.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700214
Jesper Juhlc63ca3c2006-01-09 20:53:57 -0800215tune2fs: create a ext3 journal on a ext2 partition with the -j flag.
216mke2fs: create a ext3 partition with the -j flag.
217debugfs: ext2 and ext3 file system debugger.
Tore Andersone56d5ae2006-01-11 12:17:31 -0800218ext2online: online (mounted) ext2 and ext3 filesystem resizer
Jesper Juhlc63ca3c2006-01-09 20:53:57 -0800219
Linus Torvalds1da177e2005-04-16 15:20:36 -0700220
221References
222==========
223
Jesper Juhlc63ca3c2006-01-09 20:53:57 -0800224kernel source: <file:fs/ext3/>
225 <file:fs/jbd/>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700226
Jesper Juhlc63ca3c2006-01-09 20:53:57 -0800227programs: http://e2fsprogs.sourceforge.net/
Tore Andersone56d5ae2006-01-11 12:17:31 -0800228 http://ext2resize.sourceforge.net
Linus Torvalds1da177e2005-04-16 15:20:36 -0700229
Jody McIntyreab03eca2009-03-12 17:39:23 -0400230useful links: http://www.ibm.com/developerworks/library/l-fs7.html
231 http://www.ibm.com/developerworks/library/l-fs8.html