tjh.dev
Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
1
2Ext3 Filesystem
3===============
4
5Ext3 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,
7Andrew Morton, Alexander Viro, Ted Ts'o and Stephen Tweedie.
8
9Ext3 is the ext2 filesystem enhanced with journalling capabilities.
10
11Options
12=======
13
14When mounting an ext3 filesystem, the following option are accepted:
15(*) == default
16
17ro 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
22journal=update Update the ext3 file system's journal to the current
23 format.
24
25journal=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.
28
29journal_dev=devnum When the external journal device's major/minor numbers
30 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.
34
35noload Don't load the journal on mounting. Note that this forces
36 mount of inconsistent filesystem, which can lead to
37 various problems.
38
39data=journal All data are committed into the journal prior to being
40 written into the main file system.
41
42data=ordered (*) All data are forced directly out to the main file
43 system prior to its metadata being committed to the
44 journal.
45
46data=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.
49
50commit=nrsec (*) Ext3 can be told to sync all its data and metadata
51 every 'nrsec' seconds. The default value is 5 seconds.
52 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).
59 Setting it to very large values will improve
60 performance.
61
62barrier=1 This enables/disables barriers. barrier=0 disables
63 it, barrier=1 enables it.
64
65orlov (*) This enables the new Orlov block allocator. It is
66 enabled by default.
67
68oldalloc This disables the Orlov block allocator and enables
69 the old block allocator. Orlov should have better
70 performance - we'd like to get some feedback if it's
71 the contrary for you.
72
73user_xattr Enables Extended User Attributes. Additionally, you
74 need to have extended attribute support enabled in the
75 kernel configuration (CONFIG_EXT3_FS_XATTR). See the
76 attr(5) manual page and http://acl.bestbits.at/ to
77 learn more about extended attributes.
78
79nouser_xattr Disables Extended User Attributes.
80
81acl Enables POSIX Access Control Lists support.
82 Additionally, you need to have ACL support enabled in
83 the kernel configuration (CONFIG_EXT3_FS_POSIX_ACL).
84 See the acl(5) manual page and http://acl.bestbits.at/
85 for more information.
86
87noacl This option disables POSIX Access Control List
88 support.
89
90reservation
91
92noreservation
93
94bsddf (*) Make 'df' act like BSD.
95minixdf Make 'df' act like Minix.
96
97check=none Don't do extra checking of bitmaps on mount.
98nocheck
99
100debug Extra debugging information is sent to syslog.
101
102errors=remount-ro Remount the filesystem read-only on an error.
103errors=continue Keep going on a filesystem error.
104errors=panic Panic and halt the machine if an error occurs.
105 (These mount options override the errors behavior
106 specified in the superblock, which can be
107 configured using tune2fs.)
108
109data_err=ignore(*) Just print an error message if an error occurs
110 in a file data buffer in ordered mode.
111data_err=abort Abort the journal if an error occurs in a file
112 data buffer in ordered mode.
113
114grpid Give objects the same group ID as their creator.
115bsdgroups
116
117nogrpid (*) New objects have the group ID of their creator.
118sysvgroups
119
120resgid=n The group ID which may use the reserved blocks.
121
122resuid=n The user ID which may use the reserved blocks.
123
124sb=n Use alternate superblock at this location.
125
126quota
127noquota
128grpquota
129usrquota
130
131bh (*) ext3 associates buffer heads to data pages to
132nobh (a) cache disk block mapping information
133 (b) link pages into transaction to provide
134 ordering guarantees.
135 "bh" option forces use of buffer heads.
136 "nobh" option tries to avoid associating buffer
137 heads (supported only for "writeback" mode).
138
139
140Specification
141=============
142Ext3 shares all disk implementation with the ext2 filesystem, and adds
143transactions capabilities to ext2. Journaling is done by the Journaling Block
144Device layer.
145
146Journaling Block Device layer
147-----------------------------
148The Journaling Block Device layer (JBD) isn't ext3 specific. It was designed
149to add journaling capabilities to a block device. The ext3 filesystem code
150will inform the JBD of modifications it is performing (called a transaction).
151The journal supports the transactions start and stop, and in case of a crash,
152the journal can replay the transactions to quickly put the partition back into
153a consistent state.
154
155Handles represent a single atomic update to a filesystem. JBD can handle an
156external journal on a block device.
157
158Data Mode
159---------
160There are 3 different data modes:
161
162* writeback mode
163In data=writeback mode, ext3 does not journal data at all. This mode provides
164a similar level of journaling as that of XFS, JFS, and ReiserFS in its default
165mode - metadata journaling. A crash+recovery can cause incorrect data to
166appear in files which were written shortly before the crash. This mode will
167typically provide the best ext3 performance.
168
169* ordered mode
170In data=ordered mode, ext3 only officially journals metadata, but it logically
171groups metadata and data blocks into a single unit called a transaction. When
172it's time to write the new metadata out to disk, the associated data blocks
173are written first. In general, this mode performs slightly slower than
174writeback but significantly faster than journal mode.
175
176* journal mode
177data=journal mode provides full data and metadata journaling. All new data is
178written to the journal first, and then to its final location.
179In the event of a crash, the journal can be replayed, bringing both data and
180metadata into a consistent state. This mode is the slowest except when data
181needs to be read from and written to disk at the same time where it
182outperforms all other modes.
183
184Compatibility
185-------------
186
187Ext2 partitions can be easily convert to ext3, with `tune2fs -j <dev>`.
188Ext3 is fully compatible with Ext2. Ext3 partitions can easily be mounted as
189Ext2.
190
191
192External Tools
193==============
194See manual pages to learn more.
195
196tune2fs: create a ext3 journal on a ext2 partition with the -j flag.
197mke2fs: create a ext3 partition with the -j flag.
198debugfs: ext2 and ext3 file system debugger.
199ext2online: online (mounted) ext2 and ext3 filesystem resizer
200
201
202References
203==========
204
205kernel source: <file:fs/ext3/>
206 <file:fs/jbd/>
207
208programs: http://e2fsprogs.sourceforge.net/
209 http://ext2resize.sourceforge.net
210
211useful links: http://www.ibm.com/developerworks/library/l-fs7.html
212 http://www.ibm.com/developerworks/library/l-fs8.html