1 .. SPDX-License-Identifier: GPL-2.0
3 ========================
5 ========================
7 Ext4 is an advanced level of the ext3 filesystem which incorporates
8 scalability and reliability enhancements for supporting large filesystems
9 (64 bit) in keeping with increasing disk capacities and state-of-the-art
12 Mailing list: linux-ext4@vger.kernel.org
13 Web site: http://ext4.wiki.kernel.org
16 Quick usage instructions
17 ========================
19 Note: More extensive information for getting started with ext4 can be
20 found at the ext4 wiki site at the URL:
21 http://ext4.wiki.kernel.org/index.php/Ext4_Howto
23 - The latest version of e2fsprogs can be found at:
25 https://www.kernel.org/pub/linux/kernel/people/tytso/e2fsprogs/
29 http://sourceforge.net/project/showfiles.php?group_id=2406
31 or grab the latest git repository from:
33 https://git.kernel.org/pub/scm/fs/ext2/e2fsprogs.git
35 - Create a new filesystem using the ext4 filesystem type:
37 # mke2fs -t ext4 /dev/hda1
39 Or to configure an existing ext3 filesystem to support extents:
41 # tune2fs -O extents /dev/hda1
43 If the filesystem was created with 128 byte inodes, it can be
44 converted to use 256 byte for greater efficiency via:
46 # tune2fs -I 256 /dev/hda1
50 # mount -t ext4 /dev/hda1 /wherever
52 - When comparing performance with other filesystems, it's always
53 important to try multiple workloads; very often a subtle change in a
54 workload parameter can completely change the ranking of which
55 filesystems do well compared to others. When comparing versus ext3,
56 note that ext4 enables write barriers by default, while ext3 does
57 not enable write barriers by default. So it is useful to use
58 explicitly specify whether barriers are enabled or not when via the
59 '-o barriers=[0|1]' mount option for both ext3 and ext4 filesystems
60 for a fair comparison. When tuning ext3 for best benchmark numbers,
61 it is often worthwhile to try changing the data journaling mode; '-o
62 data=writeback' can be faster for some workloads. (Note however that
63 running mounted with data=writeback can potentially leave stale data
64 exposed in recently written files in case of an unclean shutdown,
65 which could be a security exposure in some situations.) Configuring
66 the filesystem with a large journal can also be helpful for
67 metadata-intensive workloads.
75 * ability to use filesystems > 16TB (e2fsprogs support not available yet)
76 * extent format reduces metadata overhead (RAM, IO for access, transactions)
77 * extent format more robust in face of on-disk corruption due to magics,
78 * internal redundancy in tree
79 * improved file allocation (multi-block alloc)
80 * lift 32000 subdirectory limit imposed by i_links_count[1]
81 * nsec timestamps for mtime, atime, ctime, create time
82 * inode version field on disk (NFSv4, Lustre)
83 * reduced e2fsck time via uninit_bg feature
84 * journal checksumming for robustness, performance
85 * persistent file preallocation (e.g for streaming media, databases)
86 * ability to pack bitmaps and inode tables into larger virtual groups via the
89 * inode allocation using large virtual block groups via flex_bg
91 * large block (up to pagesize) support
92 * efficient new ordered mode in JBD2 and ext4 (avoid using buffer head to force
95 [1] Filesystems with a block size of 1k may see a limit imposed by the
96 directory hash tree having a maximum depth of two.
101 When mounting an ext4 filesystem, the following option are accepted:
104 ======================= =======================================================
105 Mount Option Description
106 ======================= =======================================================
107 ro Mount filesystem read only. Note that ext4 will
108 replay the journal (and thus write to the
109 partition) even when mounted "read only". The
110 mount options "ro,noload" can be used to prevent
111 writes to the filesystem.
113 journal_checksum Enable checksumming of the journal transactions.
114 This will allow the recovery code in e2fsck and the
115 kernel to detect corruption in the kernel. It is a
116 compatible change and will be ignored by older kernels.
118 journal_async_commit Commit block can be written to disk without waiting
119 for descriptor blocks. If enabled older kernels cannot
120 mount the device. This will enable 'journal_checksum'
124 journal_dev=devnum When the external journal device's major/minor numbers
125 have changed, these options allow the user to specify
126 the new journal location. The journal device is
127 identified through either its new major/minor numbers
128 encoded in devnum, or via a path to the device.
130 norecovery Don't load the journal on mounting. Note that
131 noload if the filesystem was not unmounted cleanly,
132 skipping the journal replay will lead to the
133 filesystem containing inconsistencies that can
134 lead to any number of problems.
136 data=journal All data are committed into the journal prior to being
137 written into the main file system. Enabling
138 this mode will disable delayed allocation and
141 data=ordered (*) All data are forced directly out to the main file
142 system prior to its metadata being committed to the
145 data=writeback Data ordering is not preserved, data may be written
146 into the main file system after its metadata has been
147 committed to the journal.
149 commit=nrsec (*) Ext4 can be told to sync all its data and metadata
150 every 'nrsec' seconds. The default value is 5 seconds.
151 This means that if you lose your power, you will lose
152 as much as the latest 5 seconds of work (your
153 filesystem will not be damaged though, thanks to the
154 journaling). This default value (or any low value)
155 will hurt performance, but it's good for data-safety.
156 Setting it to 0 will have the same effect as leaving
157 it at the default (5 seconds).
158 Setting it to very large values will improve
161 barrier=<0|1(*)> This enables/disables the use of write barriers in
162 barrier(*) the jbd code. barrier=0 disables, barrier=1 enables.
163 nobarrier This also requires an IO stack which can support
164 barriers, and if jbd gets an error on a barrier
165 write, it will disable again with a warning.
166 Write barriers enforce proper on-disk ordering
167 of journal commits, making volatile disk write caches
168 safe to use, at some performance penalty. If
169 your disks are battery-backed in one way or another,
170 disabling barriers may safely improve performance.
171 The mount options "barrier" and "nobarrier" can
172 also be used to enable or disable barriers, for
173 consistency with other ext4 mount options.
175 inode_readahead_blks=n This tuning parameter controls the maximum
176 number of inode table blocks that ext4's inode
177 table readahead algorithm will pre-read into
178 the buffer cache. The default value is 32 blocks.
180 nouser_xattr Disables Extended User Attributes. See the
181 attr(5) manual page for more information about
184 noacl This option disables POSIX Access Control List
185 support. If ACL support is enabled in the kernel
186 configuration (CONFIG_EXT4_FS_POSIX_ACL), ACL is
187 enabled by default on mount. See the acl(5) manual
188 page for more information about acl.
190 bsddf (*) Make 'df' act like BSD.
191 minixdf Make 'df' act like Minix.
193 debug Extra debugging information is sent to syslog.
195 abort Simulate the effects of calling ext4_abort() for
196 debugging purposes. This is normally used while
197 remounting a filesystem which is already mounted.
199 errors=remount-ro Remount the filesystem read-only on an error.
200 errors=continue Keep going on a filesystem error.
201 errors=panic Panic and halt the machine if an error occurs.
202 (These mount options override the errors behavior
203 specified in the superblock, which can be configured
206 data_err=ignore(*) Just print an error message if an error occurs
207 in a file data buffer in ordered mode.
208 data_err=abort Abort the journal if an error occurs in a file
209 data buffer in ordered mode.
211 grpid New objects have the group ID of their parent.
214 nogrpid (*) New objects have the group ID of their creator.
217 resgid=n The group ID which may use the reserved blocks.
219 resuid=n The user ID which may use the reserved blocks.
221 sb=n Use alternate superblock at this location.
223 quota These options are ignored by the filesystem. They
224 noquota are used only by quota tools to recognize volumes
225 grpquota where quota should be turned on. See documentation
226 usrquota in the quota-tools package for more details
227 (http://sourceforge.net/projects/linuxquota).
229 jqfmt=<quota type> These options tell filesystem details about quota
230 usrjquota=<file> so that quota information can be properly updated
231 grpjquota=<file> during journal replay. They replace the above
232 quota options. See documentation in the quota-tools
233 package for more details
234 (http://sourceforge.net/projects/linuxquota).
236 stripe=n Number of filesystem blocks that mballoc will try
237 to use for allocation size and alignment. For RAID5/6
238 systems this should be the number of data
239 disks * RAID chunk size in file system blocks.
241 delalloc (*) Defer block allocation until just before ext4
242 writes out the block(s) in question. This
243 allows ext4 to better allocation decisions
245 nodelalloc Disable delayed allocation. Blocks are allocated
246 when the data is copied from userspace to the
247 page cache, either via the write(2) system call
248 or when an mmap'ed page which was previously
249 unallocated is written for the first time.
251 max_batch_time=usec Maximum amount of time ext4 should wait for
252 additional filesystem operations to be batch
253 together with a synchronous write operation.
254 Since a synchronous write operation is going to
255 force a commit and then a wait for the I/O
256 complete, it doesn't cost much, and can be a
257 huge throughput win, we wait for a small amount
258 of time to see if any other transactions can
259 piggyback on the synchronous write. The
260 algorithm used is designed to automatically tune
261 for the speed of the disk, by measuring the
262 amount of time (on average) that it takes to
263 finish committing a transaction. Call this time
264 the "commit time". If the time that the
265 transaction has been running is less than the
266 commit time, ext4 will try sleeping for the
267 commit time to see if other operations will join
268 the transaction. The commit time is capped by
269 the max_batch_time, which defaults to 15000us
270 (15ms). This optimization can be turned off
271 entirely by setting max_batch_time to 0.
273 min_batch_time=usec This parameter sets the commit time (as
274 described above) to be at least min_batch_time.
275 It defaults to zero microseconds. Increasing
276 this parameter may improve the throughput of
277 multi-threaded, synchronous workloads on very
278 fast disks, at the cost of increasing latency.
280 journal_ioprio=prio The I/O priority (from 0 to 7, where 0 is the
281 highest priority) which should be used for I/O
282 operations submitted by kjournald2 during a
283 commit operation. This defaults to 3, which is
284 a slightly higher priority than the default I/O
287 auto_da_alloc(*) Many broken applications don't use fsync() when
288 noauto_da_alloc replacing existing files via patterns such as
289 fd = open("foo.new")/write(fd,..)/close(fd)/
290 rename("foo.new", "foo"), or worse yet,
291 fd = open("foo", O_TRUNC)/write(fd,..)/close(fd).
292 If auto_da_alloc is enabled, ext4 will detect
293 the replace-via-rename and replace-via-truncate
294 patterns and force that any delayed allocation
295 blocks are allocated such that at the next
296 journal commit, in the default data=ordered
297 mode, the data blocks of the new file are forced
298 to disk before the rename() operation is
299 committed. This provides roughly the same level
300 of guarantees as ext3, and avoids the
301 "zero-length" problem that can happen when a
302 system crashes before the delayed allocation
303 blocks are forced to disk.
305 noinit_itable Do not initialize any uninitialized inode table
306 blocks in the background. This feature may be
307 used by installation CD's so that the install
308 process can complete as quickly as possible; the
309 inode table initialization process would then be
310 deferred until the next time the file system
313 init_itable=n The lazy itable init code will wait n times the
314 number of milliseconds it took to zero out the
315 previous block group's inode table. This
316 minimizes the impact on the system performance
317 while file system's inode table is being initialized.
319 discard Controls whether ext4 should issue discard/TRIM
320 nodiscard(*) commands to the underlying block device when
321 blocks are freed. This is useful for SSD devices
322 and sparse/thinly-provisioned LUNs, but it is off
323 by default until sufficient testing has been done.
325 nouid32 Disables 32-bit UIDs and GIDs. This is for
326 interoperability with older kernels which only
327 store and expect 16-bit values.
329 block_validity(*) These options enable or disable the in-kernel
330 noblock_validity facility for tracking filesystem metadata blocks
331 within internal data structures. This allows multi-
332 block allocator and other routines to notice
333 bugs or corrupted allocation bitmaps which cause
334 blocks to be allocated which overlap with
335 filesystem metadata blocks.
337 dioread_lock Controls whether or not ext4 should use the DIO read
338 dioread_nolock locking. If the dioread_nolock option is specified
339 ext4 will allocate uninitialized extent before buffer
340 write and convert the extent to initialized after IO
341 completes. This approach allows ext4 code to avoid
342 using inode mutex, which improves scalability on high
343 speed storages. However this does not work with
344 data journaling and dioread_nolock option will be
345 ignored with kernel warning. Note that dioread_nolock
346 code path is only used for extent-based files.
347 Because of the restrictions this options comprises
348 it is off by default (e.g. dioread_lock).
350 max_dir_size_kb=n This limits the size of directories so that any
351 attempt to expand them beyond the specified
352 limit in kilobytes will cause an ENOSPC error.
353 This is useful in memory constrained
354 environments, where a very large directory can
355 cause severe performance problems or even
356 provoke the Out Of Memory killer. (For example,
357 if there is only 512mb memory available, a 176mb
358 directory may seriously cramp the system's style.)
360 i_version Enable 64-bit inode version support. This option is
363 dax Use direct access (no page cache). See
364 Documentation/filesystems/dax.txt. Note that
365 this option is incompatible with data=journal.
366 ======================= =======================================================
370 There are 3 different data modes:
374 In data=writeback mode, ext4 does not journal data at all. This mode provides
375 a similar level of journaling as that of XFS, JFS, and ReiserFS in its default
376 mode - metadata journaling. A crash+recovery can cause incorrect data to
377 appear in files which were written shortly before the crash. This mode will
378 typically provide the best ext4 performance.
382 In data=ordered mode, ext4 only officially journals metadata, but it logically
383 groups metadata information related to data changes with the data blocks into
384 a single unit called a transaction. When it's time to write the new metadata
385 out to disk, the associated data blocks are written first. In general, this
386 mode performs slightly slower than writeback but significantly faster than
391 data=journal mode provides full data and metadata journaling. All new data is
392 written to the journal first, and then to its final location. In the event of
393 a crash, the journal can be replayed, bringing both data and metadata into a
394 consistent state. This mode is the slowest except when data needs to be read
395 from and written to disk at the same time where it outperforms all others
396 modes. Enabling this mode will disable delayed allocation and O_DIRECT
402 Information about mounted ext4 file systems can be found in
403 /proc/fs/ext4. Each mounted filesystem will have a directory in
404 /proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or
405 /proc/fs/ext4/dm-0). The files in each per-device directory are shown
408 Files in /proc/fs/ext4/<devname>
410 ================ =======
412 ================ =======
413 mb_groups details of multiblock allocator buddy cache of free blocks
414 ================ =======
419 Information about mounted ext4 file systems can be found in
420 /sys/fs/ext4. Each mounted filesystem will have a directory in
421 /sys/fs/ext4 based on its device name (i.e., /sys/fs/ext4/hdc or
422 /sys/fs/ext4/dm-0). The files in each per-device directory are shown
425 Files in /sys/fs/ext4/<devname>:
427 (see also Documentation/ABI/testing/sysfs-fs-ext4)
429 ============================= =================================================
431 ============================= =================================================
432 delayed_allocation_blocks This file is read-only and shows the number of
433 blocks that are dirty in the page cache, but
434 which do not have their location in the
435 filesystem allocated yet.
437 inode_goal Tuning parameter which (if non-zero) controls
438 the goal inode used by the inode allocator in
439 preference to all other allocation heuristics.
440 This is intended for debugging use only, and
441 should be 0 on production systems.
443 inode_readahead_blks Tuning parameter which controls the maximum
444 number of inode table blocks that ext4's inode
445 table readahead algorithm will pre-read into
448 lifetime_write_kbytes This file is read-only and shows the number of
449 kilobytes of data that have been written to this
450 filesystem since it was created.
452 max_writeback_mb_bump The maximum number of megabytes the writeback
453 code will try to write out before move on to
456 mb_group_prealloc The multiblock allocator will round up allocation
457 requests to a multiple of this tuning parameter if
458 the stripe size is not set in the ext4 superblock
460 mb_max_to_scan The maximum number of extents the multiblock
461 allocator will search to find the best extent
463 mb_min_to_scan The minimum number of extents the multiblock
464 allocator will search to find the best extent
466 mb_order2_req Tuning parameter which controls the minimum size
467 for requests (as a power of 2) where the buddy
470 mb_stats Controls whether the multiblock allocator should
471 collect statistics, which are shown during the
472 unmount. 1 means to collect statistics, 0 means
473 not to collect statistics
475 mb_stream_req Files which have fewer blocks than this tunable
476 parameter will have their blocks allocated out
477 of a block group specific preallocation pool, so
478 that small files are packed closely together.
479 Each large file will have its blocks allocated
480 out of its own unique preallocation pool.
482 session_write_kbytes This file is read-only and shows the number of
483 kilobytes of data that have been written to this
484 filesystem since it was mounted.
486 reserved_clusters This is RW file and contains number of reserved
487 clusters in the file system which will be used
488 in the specific situations to avoid costly
489 zeroout, unexpected ENOSPC, or possible data
490 loss. The default is 2% or 4096 clusters,
491 whichever is smaller and this can be changed
492 however it can never exceed number of clusters
493 in the file system. If there is not enough space
494 for the reserved space when mounting the file
495 mount will _not_ fail.
496 ============================= =================================================
501 There is some Ext4 specific functionality which can be accessed by applications
502 through the system call interfaces. The list of all Ext4 specific ioctls are
503 shown in the table below.
505 Table of Ext4 specific ioctls
507 ============================= =================================================
509 ============================= =================================================
510 EXT4_IOC_GETFLAGS Get additional attributes associated with inode.
511 The ioctl argument is an integer bitfield, with
512 bit values described in ext4.h. This ioctl is an
513 alias for FS_IOC_GETFLAGS.
515 EXT4_IOC_SETFLAGS Set additional attributes associated with inode.
516 The ioctl argument is an integer bitfield, with
517 bit values described in ext4.h. This ioctl is an
518 alias for FS_IOC_SETFLAGS.
521 EXT4_IOC_GETVERSION_OLD
522 Get the inode i_generation number stored for
523 each inode. The i_generation number is normally
524 changed only when new inode is created and it is
525 particularly useful for network filesystems. The
526 '_OLD' version of this ioctl is an alias for
530 EXT4_IOC_SETVERSION_OLD
531 Set the inode i_generation number stored for
532 each inode. The '_OLD' version of this ioctl
533 is an alias for FS_IOC_SETVERSION.
535 EXT4_IOC_GROUP_EXTEND This ioctl has the same purpose as the resize
536 mount option. It allows to resize filesystem
537 to the end of the last existing block group,
538 further resize has to be done with resize2fs,
539 either online, or offline. The argument points
540 to the unsigned logn number representing the
541 filesystem new block count.
543 EXT4_IOC_MOVE_EXT Move the block extents from orig_fd (the one
544 this ioctl is pointing to) to the donor_fd (the
545 one specified in move_extent structure passed
546 as an argument to this ioctl). Then, exchange
547 inode metadata between orig_fd and donor_fd.
548 This is especially useful for online
549 defragmentation, because the allocator has the
550 opportunity to allocate moved blocks better,
551 ideally into one contiguous extent.
553 EXT4_IOC_GROUP_ADD Add a new group descriptor to an existing or
554 new group descriptor block. The new group
555 descriptor is described by ext4_new_group_input
556 structure, which is passed as an argument to
557 this ioctl. This is especially useful in
558 conjunction with EXT4_IOC_GROUP_EXTEND,
559 which allows online resize of the filesystem
560 to the end of the last existing block group.
561 Those two ioctls combined is used in userspace
562 online resize tool (e.g. resize2fs).
564 EXT4_IOC_MIGRATE This ioctl operates on the filesystem itself.
565 It converts (migrates) ext3 indirect block mapped
566 inode to ext4 extent mapped inode by walking
567 through indirect block mapping of the original
568 inode and converting contiguous block ranges
569 into ext4 extents of the temporary inode. Then,
570 inodes are swapped. This ioctl might help, when
571 migrating from ext3 to ext4 filesystem, however
572 suggestion is to create fresh ext4 filesystem
573 and copy data from the backup. Note, that
574 filesystem has to support extents for this ioctl
577 EXT4_IOC_ALLOC_DA_BLKS Force all of the delay allocated blocks to be
578 allocated to preserve application-expected ext3
579 behaviour. Note that this will also start
580 triggering a write of the data blocks, but this
581 behaviour may change in the future as it is
582 not necessary and has been done this way only
583 for sake of simplicity.
585 EXT4_IOC_RESIZE_FS Resize the filesystem to a new size. The number
586 of blocks of resized filesystem is passed in via
587 64 bit integer argument. The kernel allocates
588 bitmaps and inode table, the userspace tool thus
589 just passes the new number of blocks.
591 EXT4_IOC_SWAP_BOOT Swap i_blocks and associated attributes
592 (like i_blocks, i_size, i_flags, ...) from
593 the specified inode with inode
594 EXT4_BOOT_LOADER_INO (#5). This is typically
595 used to store a boot loader in a secure part of
596 the filesystem, where it can't be changed by a
597 normal user by accident.
598 The data blocks of the previous boot loader
599 will be associated with the given inode.
600 ============================= =================================================
605 kernel source: <file:fs/ext4/>
608 programs: http://e2fsprogs.sourceforge.net/
610 useful links: http://fedoraproject.org/wiki/ext3-devel
611 http://www.bullopensource.org/ext4/
612 http://ext4.wiki.kernel.org/index.php/Main_Page
613 http://fedoraproject.org/wiki/Features/Ext4
git.samba.org / sfrench / cifs-2.6.git / blob