2 .\" Copyright 1993, 1994, 1995 by Theodore Ts'o. All Rights Reserved.
3 .\" This file may be copied under the terms of the GNU Public License.
5 .TH EXT4 5 "@E2FSPROGS_MONTH@ @E2FSPROGS_YEAR@" "E2fsprogs version @E2FSPROGS_VERSION@"
7 ext2 \- the second extended file system
9 ext3 \- the third extended file system
11 ext4 \- the fourth extended file system
13 The second, third, and fourth extended file systems, or ext2, ext3, and
14 ext4 as they are commonly known, are Linux file systems that have
15 historically been the default file system for many Linux distributions.
16 They are general purpose file systems that have been designed for
17 extensibility and backwards compatibility. In particular, file systems
18 previously intended for use with the ext2 and ext3 file systems can be
19 mounted using the ext4 file system driver, and indeed in many modern
20 Linux distributions, the ext4 file system driver has been configured
21 to handle mount requests for ext2 and ext3 file systems.
22 .SH FILE SYSTEM FEATURES
23 A file system formatted for ext2, ext3, or ext4 can have some
24 collection of the following file system feature flags enabled. Some of
25 these features are not supported by all implementations of the ext2,
26 ext3, and ext4 file system drivers, depending on Linux kernel version in
27 use. On other operating systems, such as the GNU/HURD or FreeBSD, only
28 a very restrictive set of file system features may be supported in their
29 implementations of ext2.
33 Enables the file system to be larger than 2^32 blocks. This feature is set
34 automatically, as needed, but it can be useful to specify this feature
35 explicitly if the file system might need to be resized larger than 2^32
36 blocks, even if it was smaller than that threshold when it was
37 originally created. Note that some older kernels and older versions
38 of e2fsprogs will not support file systems with this ext4 feature enabled.
42 This ext4 feature enables clustered block allocation, so that the unit of
43 allocation is a power of two number of blocks. That is, each bit in the
44 what had traditionally been known as the block allocation bitmap now
45 indicates whether a cluster is in use or not, where a cluster is by
46 default composed of 16 blocks. This feature can decrease the time
47 spent on doing block allocation and brings smaller fragmentation, especially
48 for large files. The size can be specified using the
53 The bigalloc feature is still under development, and may not be fully
54 supported with your kernel or may have various bugs. Please see the web
55 page http://ext4.wiki.kernel.org/index.php/Bigalloc for details.
56 May clash with delayed allocation (see
60 This feature requires that the
66 Use hashed b-trees to speed up name lookups in large directories. This
67 feature is supported by ext3 and ext4 file systems, and is ignored by
72 Normally, ext4 allows an inode to have no more than 65,000 hard links.
73 This applies to regular files as well as directories, which means that
74 there can be no more than 64,998 subdirectories in a directory (because
75 each of the '.' and '..' entries, as well as the directory entry for the
76 directory in its parent directory counts as a hard link). This feature
77 lifts this limit by causing ext4 to use a link count of 1 to indicate
78 that the number of hard links to a directory is not known when the link
79 count might exceed the maximum count limit.
83 Normally, a file's extended attributes and associated metadata must fit within
84 the inode or the inode's associated extended attribute block. This feature
85 allows the value of each extended attribute to be placed in the data blocks of a
86 separate inode if necessary, increasing the limit on the size and number of
87 extended attributes per file.
91 This ext4 feature provides file-system level encryption of data blocks
92 and file names. The inode metadata (timestamps, file size, user/group
97 This feature is most useful on file systems with multiple users, or
98 where not all files should be encrypted. In many use cases, especially
99 on single-user systems, encryption at the block device layer using
100 dm-crypt may provide much better security.
104 This feature enables the use of extended attributes. This feature is
105 supported by ext2, ext3, and ext4.
109 This ext4 feature allows the mapping of logical block numbers for a
110 particular inode to physical blocks on the storage device to be stored
111 using an extent tree, which is a more efficient data structure than the
112 traditional indirect block scheme used by the ext2 and ext3 file
113 systems. The use of the extent tree decreases metadata block overhead,
114 improves file system performance, and decreases the needed to run
121 are accepted as valid names for this feature for
122 historical/backwards compatibility reasons.)
126 This ext4 feature reserves a specific amount of space in each inode for
127 extended metadata such as nanosecond timestamps and file creation time,
128 even if the current kernel does not currently need to reserve this much
129 space. Without this feature, the kernel will reserve the amount of
130 space for features it currently needs, and the rest may be
131 consumed by extended attributes.
133 For this feature to be useful the inode size must be 256 bytes in size
138 This feature enables the storage of file type information in directory
139 entries. This feature is supported by ext2, ext3, and ext4.
143 This ext4 feature allows the per-block group metadata (allocation
146 to be placed anywhere on the storage media. In addition,
148 will place the per-block group metadata together starting at the first
149 block group of each "flex_bg group". The size of the flex_bg group
150 can be specified using the
156 This ext4 feature provides file system level character encoding support
157 for directories with the casefold (+F) flag enabled. This feature is
158 name-preserving on the disk, but it allows applications to lookup for a
159 file in the file system using an encoding equivalent version of the file
164 Create a journal to ensure filesystem consistency even across unclean
165 shutdowns. Setting the filesystem feature is equivalent to using the
168 .BR mke2fs " or " tune2fs.
169 This feature is supported by ext3 and ext4, and ignored by the
170 ext2 file system driver.
174 This ext4 feature allows files to be larger than 2 terabytes in size.
177 Allow data to be stored in the inode and extended attribute area.
181 This feature is enabled on the superblock found on an external journal
182 device. The block size for the external journal must be the same as the
183 file system which uses it.
185 The external journal device can be used by a file system by specifying
188 .BR device= <external-device>
196 This feature increases the limit on the number of files per directory by
197 raising the maximum size of directories and, for hashed b-tree directories (see
199 the maximum height of the hashed b-tree used to store the directory entries.
203 This feature flag is set automatically by modern kernels when a file
204 larger than 2 gigabytes is created. Very old kernels could not
205 handle large files, so this feature flag was used to prohibit those
206 kernels from mounting file systems that they could not understand.
210 This ext4 feature enables metadata checksumming. This feature stores
211 checksums for all of the filesystem metadata (superblock, group
212 descriptor blocks, inode and block bitmaps, directories, and
213 extent tree blocks). The checksum algorithm used for the metadata
214 blocks is different than the one used for group descriptors with the
216 feature. These two features are incompatible and
218 will be used preferentially instead of
221 .B metadata_csum_seed
223 This feature allows the filesystem to store the metadata checksum seed in the
224 superblock, which allows the administrator to change the UUID of a filesystem
227 feature while it is mounted.
231 This ext4 feature allows file systems to be resized on-line without explicitly
232 needing to reserve space for growth in the size of the block group
233 descriptors. This scheme is also used to resize file systems which are
234 larger than 2^32 blocks. It is not recommended that this feature be set
235 when a file system is created, since this alternate method of storing
236 the block group descriptors will slow down the time needed to mount the
237 file system, and newer kernels can automatically set this feature as
238 necessary when doing an online resize and no more reserved space is
239 available in the resize inode.
243 This ext4 feature provides multiple mount protection (MMP). MMP helps to
244 protect the filesystem from being multiply mounted and is useful in
245 shared storage environments.
249 This ext4 feature provides project quota support. With this feature,
250 the project ID of inode will be managed when the filesystem is mounted.
254 Create quota inodes (inode #3 for userquota and inode
255 #4 for group quota) and set them in the superblock.
256 With this feature, the quotas will be enabled
257 automatically when the filesystem is mounted.
259 Causes the quota files (i.e., user.quota and
260 group.quota which existed
261 in the older quota design) to be hidden inodes.
265 This file system feature indicates that space has been reserved so that
266 the block group descriptor table can be extended while resizing a mounted
267 file system. The online resize operation
268 is carried out by the kernel, triggered by
272 will attempt to reserve enough space so that the
273 filesystem may grow to 1024 times its initial size. This can be changed
278 This feature requires that the
286 This file system feature is set on all modern ext2, ext3, and ext4 file
287 systems. It indicates that backup copies of the superblock and block
288 group descriptors are present only in a few block groups, not all of
293 This feature indicates that there will only be at most two backup
294 superblocks and block group descriptors. The block groups used to store
295 the backup superblock(s) and blockgroup descriptor(s) are stored in the
296 superblock, but typically, one will be located at the beginning of block
297 group #1, and one in the last block group in the file system. This
298 feature is essentially a more extreme version of sparse_super and is
299 designed to allow a much larger percentage of the disk to have
300 contiguous blocks available for data files.
304 This ext4 file system feature indicates that the block group descriptors
305 will be protected using checksums, making it safe for
307 to create a file system without initializing all of the block groups.
308 The kernel will keep a high watermark of unused inodes, and initialize
309 inode tables and blocks lazily. This feature speeds up the time to check
310 the file system using
312 and it also speeds up the time required for
314 to create the file system.
316 This section describes mount options which are specific to ext2, ext3,
317 and ext4. Other generic mount options may be used as well; see
320 .SH "Mount options for ext2"
321 The `ext2' filesystem is the standard Linux filesystem.
322 Since Linux 2.5.46, for most mount options the default
323 is determined by the filesystem superblock. Set them with
327 Support POSIX Access Control Lists (or not). See the
332 Set the behavior for the
336 behavior is to return in the
338 field the total number of blocks of the filesystem, while the
340 behavior (which is the default) is to subtract the overhead blocks
341 used by the ext2 filesystem and not available for file storage. Thus
343 % mount /k \-o minixdf; df /k; umount /k
348 Filesystem#1024-blocks#Used#Available#Capacity#Mounted on
349 /dev/sda6#2630655#86954#2412169#3%#/k
352 % mount /k \-o bsddf; df /k; umount /k
357 Filesystem#1024-blocks#Used#Available#Capacity#Mounted on
358 /dev/sda6#2543714#13#2412169#0%#/k
361 (Note that this example shows that one can add command line options
362 to the options given in
365 .BR check=none " or " nocheck
366 No checking is done at mount time. This is the default. This is fast.
369 every now and then, e.g.\& at boot time. The non-default behavior is unsupported
370 (check=normal and check=strict options have been removed). Note that these mount options
371 don't have to be supported if ext4 kernel driver is used for ext2 and ext3 filesystems.
374 Print debugging info upon each (re)mount.
376 .BR errors= { continue | remount-ro | panic }
377 Define the behavior when an error is encountered.
378 (Either ignore errors and just mark the filesystem erroneous and continue,
379 or remount the filesystem read-only, or panic and halt the system.)
380 The default is set in the filesystem superblock, and can be
384 .BR grpid | bsdgroups " and " nogrpid | sysvgroups
385 These options define what group id a newly created file gets.
388 is set, it takes the group id of the directory in which it is created;
389 otherwise (the default) it takes the fsgid of the current process, unless
390 the directory has the setgid bit set, in which case it takes the gid
391 from the parent directory, and also gets the setgid bit set
392 if it is a directory itself.
394 .BR grpquota | noquota | quota | usrquota
395 The usrquota (same as quota) mount option enables user quota support on the
396 filesystem. grpquota enables group quotas support. You need the quota utilities
397 to actually enable and manage the quota system.
400 Disables 32-bit UIDs and GIDs. This is for interoperability with older
401 kernels which only store and expect 16-bit values.
403 .BR oldalloc " or " orlov
404 Use old allocator or Orlov allocator for new inodes. Orlov is default.
406 \fBresgid=\fP\,\fIn\fP and \fBresuid=\fP\,\fIn\fP
407 The ext2 filesystem reserves a certain percentage of the available
408 space (by default 5%, see
412 These options determine who can use the reserved blocks.
413 (Roughly: whoever has the specified uid, or belongs to the specified group.)
416 Instead of using the normal superblock, use an alternative superblock
419 This option is normally used when the primary superblock has been
420 corrupted. The location of backup superblocks is dependent on the
421 filesystem's blocksize, the number of blocks per group, and features
425 Additional backup superblocks can be determined by using the
429 option to print out where the superblocks exist, supposing
431 is supplied with arguments that are consistent with the filesystem's layout
432 (e.g. blocksize, blocks per group,
436 The block number here uses 1\ k units. Thus, if you want to use logical
437 block 32768 on a filesystem with 4\ k blocks, use "sb=131072".
439 .BR user_xattr | nouser_xattr
440 Support "user." extended attributes (or not).
443 .SH "Mount options for ext3"
444 The ext3 filesystem is a version of the ext2 filesystem which has been
445 enhanced with journaling. It supports the same options as ext2 as
446 well as the following additions:
448 .BR journal_dev=devnum / journal_path=path
449 When the external journal device's major/minor numbers
450 have changed, these options allow the user to specify
451 the new journal location. The journal device is
452 identified either through its new major/minor numbers encoded
453 in devnum, or via a path to the device.
455 .BR norecovery / noload
456 Don't load the journal on mounting. Note that
457 if the filesystem was not unmounted cleanly,
458 skipping the journal replay will lead to the
459 filesystem containing inconsistencies that can
460 lead to any number of problems.
462 .BR data= { journal | ordered | writeback }
463 Specifies the journaling mode for file data. Metadata is always journaled.
464 To use modes other than
466 on the root filesystem, pass the mode to the kernel as boot parameter, e.g.\&
467 .IR rootflags=data=journal .
471 All data is committed into the journal prior to being written into the
475 This is the default mode. All data is forced directly out to the main file
476 system prior to its metadata being committed to the journal.
479 Data ordering is not preserved \(en data may be written into the main
480 filesystem after its metadata has been committed to the journal.
481 This is rumoured to be the highest-throughput option. It guarantees
482 internal filesystem integrity, however it can allow old data to appear
483 in files after a crash and journal recovery.
487 Just print an error message if an error occurs in a file data buffer in
491 Abort the journal if an error occurs in a file data buffer in ordered mode.
493 .BR barrier=0 " / " barrier=1 "
494 This disables / enables the use of write barriers in the jbd code. barrier=0
495 disables, barrier=1 enables (default). This also requires an IO stack which can
496 support barriers, and if jbd gets an error on a barrier write, it will disable
497 barriers again with a warning. Write barriers enforce proper on-disk ordering
498 of journal commits, making volatile disk write caches safe to use, at some
499 performance penalty. If your disks are battery-backed in one way or another,
500 disabling barriers may safely improve performance.
503 Start a journal commit every
505 seconds. The default value is 5 seconds. Zero means default.
508 Enable Extended User Attributes. See the
512 .BR jqfmt= { vfsold | vfsv0 | vfsv1 }
513 Apart from the old quota system (as in ext2, jqfmt=vfsold aka version 1 quota)
514 ext3 also supports journaled quotas (version 2 quota). jqfmt=vfsv0 or
515 jqfmt=vfsv1 enables journaled quotas. Journaled quotas have the advantage that
516 even after a crash no quota check is required. When the
518 filesystem feature is enabled, journaled quotas are used automatically, and
519 this mount option is ignored.
521 .BR usrjquota=aquota.user | grpjquota=aquota.group
522 For journaled quotas (jqfmt=vfsv0 or jqfmt=vfsv1), the mount options
523 usrjquota=aquota.user and grpjquota=aquota.group are required to tell the
524 quota system which quota database files to use. When the
526 filesystem feature is enabled, journaled quotas are used automatically, and
527 this mount option is ignored.
529 .SH "Mount options for ext4"
530 The ext4 filesystem is an advanced level of the ext3 filesystem which
531 incorporates scalability and reliability enhancements for supporting large
535 .B journal_dev, journal_path, norecovery, noload, data, commit, orlov,
536 .B oldalloc, [no]user_xattr, [no]acl, bsddf, minixdf, debug, errors,
537 .B data_err, grpid, bsdgroups, nogrpid, sysvgroups, resgid, resuid, sb,
538 .B quota, noquota, nouid32, grpquota, usrquota, usrjquota, grpjquota,
539 .B and jqfmt are backwardly compatible with ext3 or ext2.
541 .B journal_checksum | nojournal_checksum
542 The journal_checksum option enables checksumming of the journal transactions.
543 This will allow the recovery code in e2fsck and the kernel to detect corruption
544 in the kernel. It is a compatible change and will be ignored by older kernels.
546 .B journal_async_commit
547 Commit block can be written to disk without waiting for descriptor blocks. If
548 enabled older kernels cannot mount the device.
549 This will enable 'journal_checksum' internally.
551 .BR barrier=0 " / " barrier=1 " / " barrier " / " nobarrier
552 These mount options have the same effect as in ext3. The mount options
553 "barrier" and "nobarrier" are added for consistency with other ext4 mount
556 The ext4 filesystem enables write barriers by default.
558 .BI inode_readahead_blks= n
559 This tuning parameter controls the maximum number of inode table blocks that
560 ext4's inode table readahead algorithm will pre-read into the buffer cache.
561 The value must be a power of 2. The default value is 32 blocks.
564 Number of filesystem blocks that mballoc will try to use for allocation size
565 and alignment. For RAID5/6 systems this should be the number of data disks *
566 RAID chunk size in filesystem blocks.
569 Deferring block allocation until write-out time.
572 Disable delayed allocation. Blocks are allocated when data is copied from user
575 .BI max_batch_time= usec
576 Maximum amount of time ext4 should wait for additional filesystem operations to
577 be batch together with a synchronous write operation. Since a synchronous
578 write operation is going to force a commit and then a wait for the I/O
579 complete, it doesn't cost much, and can be a huge throughput win, we wait for a
580 small amount of time to see if any other transactions can piggyback on the
581 synchronous write. The algorithm used is designed to automatically tune for
582 the speed of the disk, by measuring the amount of time (on average) that it
583 takes to finish committing a transaction. Call this time the "commit time".
584 If the time that the transaction has been running is less than the commit time,
585 ext4 will try sleeping for the commit time to see if other operations will join
586 the transaction. The commit time is capped by the max_batch_time, which
587 defaults to 15000\ \[mc]s (15\ ms). This optimization can be turned off entirely by
588 setting max_batch_time to 0.
590 .BI min_batch_time= usec
591 This parameter sets the commit time (as described above) to be at least
592 min_batch_time. It defaults to zero microseconds. Increasing this parameter
593 may improve the throughput of multi-threaded, synchronous workloads on very
594 fast disks, at the cost of increasing latency.
596 .BI journal_ioprio= prio
597 The I/O priority (from 0 to 7, where 0 is the highest priority) which should be
598 used for I/O operations submitted by kjournald2 during a commit operation.
599 This defaults to 3, which is a slightly higher priority than the default I/O
603 Simulate the effects of calling ext4_abort() for
604 debugging purposes. This is normally used while
605 remounting a filesystem which is already mounted.
607 .BR auto_da_alloc | noauto_da_alloc
608 Many broken applications don't use fsync() when
609 replacing existing files via patterns such as
611 fd = open("foo.new")/write(fd,...)/close(fd)/ rename("foo.new", "foo")
615 fd = open("foo", O_TRUNC)/write(fd,...)/close(fd).
617 If auto_da_alloc is enabled, ext4 will detect the replace-via-rename and
618 replace-via-truncate patterns and force that any delayed allocation blocks are
619 allocated such that at the next journal commit, in the default data=ordered
620 mode, the data blocks of the new file are forced to disk before the rename()
621 operation is committed. This provides roughly the same level of guarantees as
622 ext3, and avoids the "zero-length" problem that can happen when a system
623 crashes before the delayed allocation blocks are forced to disk.
626 Do not initialize any uninitialized inode table blocks in the background. This
627 feature may be used by installation CD's so that the install process can
628 complete as quickly as possible; the inode table initialization process would
629 then be deferred until the next time the filesystem is mounted.
632 The lazy itable init code will wait n times the number of milliseconds it took
633 to zero out the previous block group's inode table. This minimizes the impact on
634 system performance while the filesystem's inode table is being initialized.
636 .BR discard / nodiscard
637 Controls whether ext4 should issue discard/TRIM commands to the underlying
638 block device when blocks are freed. This is useful for SSD devices and
639 sparse/thinly-provisioned LUNs, but it is off by default until sufficient
640 testing has been done.
642 .BR block_validity / noblock_validity
643 This option enables/disables the in-kernel facility for tracking
644 filesystem metadata blocks within internal data structures. This allows multi-\c
645 block allocator and other routines to quickly locate extents which might
646 overlap with filesystem metadata blocks. This option is intended for debugging
647 purposes and since it negatively affects the performance, it is off by default.
649 .BR dioread_lock / dioread_nolock
650 Controls whether or not ext4 should use the DIO read locking. If the
651 dioread_nolock option is specified ext4 will allocate uninitialized extent
652 before buffer write and convert the extent to initialized after IO completes.
653 This approach allows ext4 code to avoid using inode mutex, which improves
654 scalability on high speed storages. However this does not work with data
655 journaling and dioread_nolock option will be ignored with kernel warning.
656 Note that dioread_nolock code path is only used for extent-based files.
657 Because of the restrictions this options comprises it is off by default
658 (e.g.\& dioread_lock).
661 This limits the size of the directories so that any attempt to expand them
662 beyond the specified limit in kilobytes will cause an ENOSPC error. This is
663 useful in memory-constrained environments, where a very large directory can
664 cause severe performance problems or even provoke the Out Of Memory killer. (For
665 example, if there is only 512\ MB memory available, a 176\ MB directory may
666 seriously cramp the system's style.)
669 Enable 64-bit inode version support. This option is off by default.
672 This option disables use of mbcache for extended attribute deduplication. On
673 systems where extended attributes are rarely or never shared between files,
674 use of mbcache for deduplication adds unnecessary computational overhead.
677 The prjquota mount option enables project quota support on the filesystem.
678 You need the quota utilities to actually enable and manage the quota system.
679 This mount option requires the
684 The ext2, ext3, and ext4 filesystems support setting the following file
685 attributes on Linux systems using the
689 .BR a " - append only"
691 .BR A " - no atime updates"
695 .BR D " - synchronous directory updates"
699 .BR S " - synchronous updates"
701 .BR u " - undeletable"
703 In addition, the ext3 and ext4 filesystems support the following flag:
705 .BR j " - data journaling"
707 Finally, the ext4 filesystem also supports the following flag:
709 .BR e " - extents format"
711 For descriptions of these attribute flags, please refer to the
715 This section lists the file system driver (e.g., ext2, ext3, ext4) and
716 upstream kernel version where a particular file system feature was
717 supported. Note that in some cases the feature was present in earlier
718 kernel versions, but there were known, serious bugs. In other cases the
719 feature may still be considered in an experimental state. Finally, note
720 that some distributions may have backported features into older kernels;
721 in particular the kernel versions in certain "enterprise distributions"
722 can be extremely misleading.
723 .IP "\fBfiletype\fR" 2in
725 .IP "\fBsparse_super\fR" 2in
727 .IP "\fBlarge_file\fR" 2in
729 .IP "\fBhas_journal\fR" 2in
731 .IP "\fBext_attr\fR" 2in
733 .IP "\fBdir_index\fR" 2in
735 .IP "\fBresize_inode\fR" 2in
736 ext3, 2.6.10 (online resizing)
737 .IP "\fB64bit\fR" 2in
739 .IP "\fBdir_nlink\fR" 2in
741 .IP "\fBextent\fR" 2in
743 .IP "\fBextra_isize\fR" 2in
745 .IP "\fBflex_bg\fR" 2in
747 .IP "\fBhuge_file\fR" 2in
749 .IP "\fBmeta_bg\fR" 2in
751 .IP "\fBuninit_bg\fR" 2in
755 .IP "\fBbigalloc\fR" 2in
757 .IP "\fBquota\fR" 2in
759 .IP "\fBinline_data\fR" 2in
761 .IP "\fBsparse_super2\fR" 2in
763 .IP "\fBmetadata_csum\fR" 2in
765 .IP "\fBencrypt\fR" 2in
767 .IP "\fBmetadata_csum_seed\fR" 2i
769 .IP "\fBproject\fR" 2i
771 .IP "\fBea_inode\fR" 2i
773 .IP "\fBlarge_dir\fR" 2i