1 .. SPDX-License-Identifier: GPL-2.0
2
3 ========================
4 ext4 General Information
5 ========================
6
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
10 feature requirements.
11
12 Mailing list: linux-ext4@vger.kernel.org
13 Web site: http://ext4.wiki.kernel.org
14
15
16 Quick usage instructions
17 ========================
18
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
22
23 - The latest version of e2fsprogs can be found at:
24
25 https://www.kernel.org/pub/linux/kernel/people/tytso/e2fsprogs/
26
27 or
28
29 http://sourceforge.net/project/showfiles.php?group_id=2406
30
31 or grab the latest git repository from:
32
33 https://git.kernel.org/pub/scm/fs/ext2/e2fsprogs.git
34
35 - Create a new filesystem using the ext4 filesystem type:
36
37 # mke2fs -t ext4 /dev/hda1
38
39 Or to configure an existing ext3 filesystem to support extents:
40
41 # tune2fs -O extents /dev/hda1
42
43 If the filesystem was created with 128 byte inodes, it can be
44 converted to use 256 byte for greater efficiency via:
45
46 # tune2fs -I 256 /dev/hda1
47
48 - Mounting:
49
50 # mount -t ext4 /dev/hda1 /wherever
51
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.
68
69 Features
70 ========
71
72 Currently Available
73 -------------------
74
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
87 flex_bg feature
88 * large file support
89 * inode allocation using large virtual block groups via flex_bg
90 * delayed allocation
91 * large block (up to pagesize) support
92 * efficient new ordered mode in JBD2 and ext4 (avoid using buffer head to force
93 the ordering)
94 * Case-insensitive file name lookups
95 * file-based encryption support (fscrypt)
96 * file-based verity support (fsverity)
97
98 [1] Filesystems with a block size of 1k may see a limit imposed by the
99 directory hash tree having a maximum depth of two.
100
101 case-insensitive file name lookups
102 ======================================================
103
104 The case-insensitive file name lookup feature is supported on a
105 per-directory basis, allowing the user to mix case-insensitive and
106 case-sensitive directories in the same filesystem. It is enabled by
107 flipping the +F inode attribute of an empty directory. The
108 case-insensitive string match operation is only defined when we know how
109 text in encoded in a byte sequence. For that reason, in order to enable
110 case-insensitive directories, the filesystem must have the
111 casefold feature, which stores the filesystem-wide encoding
112 model used. By default, the charset adopted is the latest version of
113 Unicode (12.1.0, by the time of this writing), encoded in the UTF-8
114 form. The comparison algorithm is implemented by normalizing the
115 strings to the Canonical decomposition form, as defined by Unicode,
116 followed by a byte per byte comparison.
117
118 The case-awareness is name-preserving on the disk, meaning that the file
119 name provided by userspace is a byte-per-byte match to what is actually
120 written in the disk. The Unicode normalization format used by the
121 kernel is thus an internal representation, and not exposed to the
122 userspace nor to the disk, with the important exception of disk hashes,
123 used on large case-insensitive directories with DX feature. On DX
124 directories, the hash must be calculated using the casefolded version of
125 the filename, meaning that the normalization format used actually has an
126 impact on where the directory entry is stored.
127
128 When we change from viewing filenames as opaque byte sequences to seeing
129 them as encoded strings we need to address what happens when a program
130 tries to create a file with an invalid name. The Unicode subsystem
131 within the kernel leaves the decision of what to do in this case to the
132 filesystem, which select its preferred behavior by enabling/disabling
133 the strict mode. When Ext4 encounters one of those strings and the
134 filesystem did not require strict mode, it falls back to considering the
135 entire string as an opaque byte sequence, which still allows the user to
136 operate on that file, but the case-insensitive lookups won't work.
137
138 Options
139 =======
140
141 When mounting an ext4 filesystem, the following option are accepted:
142 (*) == default
143
144 ro
145 Mount filesystem read only. Note that ext4 will replay the journal (and
146 thus write to the partition) even when mounted "read only". The mount
147 options "ro,noload" can be used to prevent writes to the filesystem.
148
149 journal_checksum
150 Enable checksumming of the journal transactions. This will allow the
151 recovery code in e2fsck and the kernel to detect corruption in the
152 kernel. It is a compatible change and will be ignored by older
153 kernels.
154
155 journal_async_commit
156 Commit block can be written to disk without waiting for descriptor
157 blocks. If enabled older kernels cannot mount the device. This will
158 enable 'journal_checksum' internally.
159
160 journal_path=path, journal_dev=devnum
161 When the external journal device's major/minor numbers have changed,
162 these options allow the user to specify the new journal location. The
163 journal device is identified through either its new major/minor numbers
164 encoded in devnum, or via a path to the device.
165
166 norecovery, noload
167 Don't load the journal on mounting. Note that if the filesystem was
168 not unmounted cleanly, skipping the journal replay will lead to the
169 filesystem containing inconsistencies that can lead to any number of
170 problems.
171
172 data=journal
173 All data are committed into the journal prior to being written into the
174 main file system. Enabling this mode will disable delayed allocation
175 and O_DIRECT support.
176
177 data=ordered (*)
178 All data are forced directly out to the main file system prior to its
179 metadata being committed to the journal.
180
181 data=writeback
182 Data ordering is not preserved, data may be written into the main file
183 system after its metadata has been committed to the journal.
184
185 commit=nrsec (*)
186 This setting limits the maximum age of the running transaction to
187 'nrsec' seconds. The default value is 5 seconds. This means that if
188 you lose your power, you will lose as much as the latest 5 seconds of
189 metadata changes (your filesystem will not be damaged though, thanks
190 to the journaling). This default value (or any low value) will hurt
191 performance, but it's good for data-safety. Setting it to 0 will have
192 the same effect as leaving it at the default (5 seconds). Setting it
193 to very large values will improve performance. Note that due to
194 delayed allocation even older data can be lost on power failure since
195 writeback of those data begins only after time set in
196 /proc/sys/vm/dirty_expire_centisecs.
197
198 barrier=<0|1(*)>, barrier(*), nobarrier
199 This enables/disables the use of write barriers in the jbd code.
200 barrier=0 disables, barrier=1 enables. This also requires an IO stack
201 which can support barriers, and if jbd gets an error on a barrier
202 write, it will disable again with a warning. Write barriers enforce
203 proper on-disk ordering of journal commits, making volatile disk write
204 caches safe to use, at some performance penalty. If your disks are
205 battery-backed in one way or another, disabling barriers may safely
206 improve performance. The mount options "barrier" and "nobarrier" can
207 also be used to enable or disable barriers, for consistency with other
208 ext4 mount options.
209
210 inode_readahead_blks=n
211 This tuning parameter controls the maximum number of inode table blocks
212 that ext4's inode table readahead algorithm will pre-read into the
213 buffer cache. The default value is 32 blocks.
214
215 bsddf (*)
216 Make 'df' act like BSD.
217
218 minixdf
219 Make 'df' act like Minix.
220
221 debug
222 Extra debugging information is sent to syslog.
223
224 abort
225 Simulate the effects of calling ext4_abort() for debugging purposes.
226 This is normally used while remounting a filesystem which is already
227 mounted.
228
229 errors=remount-ro
230 Remount the filesystem read-only on an error.
231
232 errors=continue
233 Keep going on a filesystem error.
234
235 errors=panic
236 Panic and halt the machine if an error occurs. (These mount options
237 override the errors behavior specified in the superblock, which can be
238 configured using tune2fs)
239
240 data_err=ignore(*)
241 Just print an error message if an error occurs in a file data buffer in
242 ordered mode.
243 data_err=abort
244 Abort the journal if an error occurs in a file data buffer in ordered
245 mode.
246
247 grpid | bsdgroups
248 New objects have the group ID of their parent.
249
250 nogrpid (*) | sysvgroups
251 New objects have the group ID of their creator.
252
253 resgid=n
254 The group ID which may use the reserved blocks.
255
256 resuid=n
257 The user ID which may use the reserved blocks.
258
259 sb=
260 Use alternate superblock at this location.
261
262 quota, noquota, grpquota, usrquota
263 These options are ignored by the filesystem. They are used only by
264 quota tools to recognize volumes where quota should be turned on. See
265 documentation in the quota-tools package for more details
266 (http://sourceforge.net/projects/linuxquota).
267
268 jqfmt=<quota type>, usrjquota=<file>, grpjquota=<file>
269 These options tell filesystem details about quota so that quota
270 information can be properly updated during journal replay. They replace
271 the above quota options. See documentation in the quota-tools package
272 for more details (http://sourceforge.net/projects/linuxquota).
273
274 stripe=n
275 Number of filesystem blocks that mballoc will try to use for allocation
276 size and alignment. For RAID5/6 systems this should be the number of
277 data disks * RAID chunk size in file system blocks.
278
279 delalloc (*)
280 Defer block allocation until just before ext4 writes out the block(s)
281 in question. This allows ext4 to better allocation decisions more
282 efficiently.
283
284 nodelalloc
285 Disable delayed allocation. Blocks are allocated when the data is
286 copied from userspace to the page cache, either via the write(2) system
287 call or when an mmap'ed page which was previously unallocated is
288 written for the first time.
289
290 max_batch_time=usec
291 Maximum amount of time ext4 should wait for additional filesystem
292 operations to be batch together with a synchronous write operation.
293 Since a synchronous write operation is going to force a commit and then
294 a wait for the I/O complete, it doesn't cost much, and can be a huge
295 throughput win, we wait for a small amount of time to see if any other
296 transactions can piggyback on the synchronous write. The algorithm
297 used is designed to automatically tune for the speed of the disk, by
298 measuring the amount of time (on average) that it takes to finish
299 committing a transaction. Call this time the "commit time". If the
300 time that the transaction has been running is less than the commit
301 time, ext4 will try sleeping for the commit time to see if other
302 operations will join the transaction. The commit time is capped by
303 the max_batch_time, which defaults to 15000us (15ms). This
304 optimization can be turned off entirely by setting max_batch_time to 0.
305
306 min_batch_time=usec
307 This parameter sets the commit time (as described above) to be at least
308 min_batch_time. It defaults to zero microseconds. Increasing this
309 parameter may improve the throughput of multi-threaded, synchronous
310 workloads on very fast disks, at the cost of increasing latency.
311
312 journal_ioprio=prio
313 The I/O priority (from 0 to 7, where 0 is the highest priority) which
314 should be used for I/O operations submitted by kjournald2 during a
315 commit operation. This defaults to 3, which is a slightly higher
316 priority than the default I/O priority.
317
318 auto_da_alloc(*), noauto_da_alloc
319 Many broken applications don't use fsync() when replacing existing
320 files via patterns such as fd = open("foo.new")/write(fd,..)/close(fd)/
321 rename("foo.new", "foo"), or worse yet, fd = open("foo",
322 O_TRUNC)/write(fd,..)/close(fd). If auto_da_alloc is enabled, ext4
323 will detect the replace-via-rename and replace-via-truncate patterns
324 and force that any delayed allocation blocks are allocated such that at
325 the next journal commit, in the default data=ordered mode, the data
326 blocks of the new file are forced to disk before the rename() operation
327 is committed. This provides roughly the same level of guarantees as
328 ext3, and avoids the "zero-length" problem that can happen when a
329 system crashes before the delayed allocation blocks are forced to disk.
330
331 noinit_itable
332 Do not initialize any uninitialized inode table blocks in the
333 background. This feature may be used by installation CD's so that the
334 install process can complete as quickly as possible; the inode table
335 initialization process would then be deferred until the next time the
336 file system is unmounted.
337
338 init_itable=n
339 The lazy itable init code will wait n times the number of milliseconds
340 it took to zero out the previous block group's inode table. This
341 minimizes the impact on the system performance while file system's
342 inode table is being initialized.
343
344 discard, nodiscard(*)
345 Controls whether ext4 should issue discard/TRIM commands to the
346 underlying block device when blocks are freed. This is useful for SSD
347 devices and sparse/thinly-provisioned LUNs, but it is off by default
348 until sufficient testing has been done.
349
350 nouid32
351 Disables 32-bit UIDs and GIDs. This is for interoperability with
352 older kernels which only store and expect 16-bit values.
353
354 block_validity(*), noblock_validity
355 These options enable or disable the in-kernel facility for tracking
356 filesystem metadata blocks within internal data structures. This
357 allows multi- block allocator and other routines to notice bugs or
358 corrupted allocation bitmaps which cause blocks to be allocated which
359 overlap with filesystem metadata blocks.
360
361 dioread_lock, dioread_nolock
362 Controls whether or not ext4 should use the DIO read locking. If the
363 dioread_nolock option is specified ext4 will allocate uninitialized
364 extent before buffer write and convert the extent to initialized after
365 IO completes. This approach allows ext4 code to avoid using inode
366 mutex, which improves scalability on high speed storages. However this
367 does not work with data journaling and dioread_nolock option will be
368 ignored with kernel warning. Note that dioread_nolock code path is only
369 used for extent-based files. Because of the restrictions this options
370 comprises it is off by default (e.g. dioread_lock).
371
372 max_dir_size_kb=n
373 This limits the size of directories so that any attempt to expand them
374 beyond the specified limit in kilobytes will cause an ENOSPC error.
375 This is useful in memory constrained environments, where a very large
376 directory can cause severe performance problems or even provoke the Out
377 Of Memory killer. (For example, if there is only 512mb memory
378 available, a 176mb directory may seriously cramp the system's style.)
379
380 i_version
381 Enable 64-bit inode version support. This option is off by default.
382
383 dax
384 Use direct access (no page cache). See
385 Documentation/filesystems/dax.rst. Note that this option is
386 incompatible with data=journal.
387
388 inlinecrypt
389 When possible, encrypt/decrypt the contents of encrypted files using the
390 blk-crypto framework rather than filesystem-layer encryption. This
391 allows the use of inline encryption hardware. The on-disk format is
392 unaffected. For more details, see
393 Documentation/block/inline-encryption.rst.
394
395 Data Mode
396 =========
397 There are 3 different data modes:
398
399 * writeback mode
400
401 In data=writeback mode, ext4 does not journal data at all. This mode provides
402 a similar level of journaling as that of XFS, JFS, and ReiserFS in its default
403 mode - metadata journaling. A crash+recovery can cause incorrect data to
404 appear in files which were written shortly before the crash. This mode will
405 typically provide the best ext4 performance.
406
407 * ordered mode
408
409 In data=ordered mode, ext4 only officially journals metadata, but it logically
410 groups metadata information related to data changes with the data blocks into
411 a single unit called a transaction. When it's time to write the new metadata
412 out to disk, the associated data blocks are written first. In general, this
413 mode performs slightly slower than writeback but significantly faster than
414 journal mode.
415
416 * journal mode
417
418 data=journal mode provides full data and metadata journaling. All new data is
419 written to the journal first, and then to its final location. In the event of
420 a crash, the journal can be replayed, bringing both data and metadata into a
421 consistent state. This mode is the slowest except when data needs to be read
422 from and written to disk at the same time where it outperforms all others
423 modes. Enabling this mode will disable delayed allocation and O_DIRECT
424 support.
425
426 /proc entries
427 =============
428
429 Information about mounted ext4 file systems can be found in
430 /proc/fs/ext4. Each mounted filesystem will have a directory in
431 /proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or
432 /proc/fs/ext4/dm-0). The files in each per-device directory are shown
433 in table below.
434
435 Files in /proc/fs/ext4/<devname>
436
437 mb_groups
438 details of multiblock allocator buddy cache of free blocks
439
440 /sys entries
441 ============
442
443 Information about mounted ext4 file systems can be found in
444 /sys/fs/ext4. Each mounted filesystem will have a directory in
445 /sys/fs/ext4 based on its device name (i.e., /sys/fs/ext4/hdc or
446 /sys/fs/ext4/dm-0). The files in each per-device directory are shown
447 in table below.
448
449 Files in /sys/fs/ext4/<devname>:
450
451 (see also Documentation/ABI/testing/sysfs-fs-ext4)
452
453 delayed_allocation_blocks
454 This file is read-only and shows the number of blocks that are dirty in
455 the page cache, but which do not have their location in the filesystem
456 allocated yet.
457
458 inode_goal
459 Tuning parameter which (if non-zero) controls the goal inode used by
460 the inode allocator in preference to all other allocation heuristics.
461 This is intended for debugging use only, and should be 0 on production
462 systems.
463
464 inode_readahead_blks
465 Tuning parameter which controls the maximum number of inode table
466 blocks that ext4's inode table readahead algorithm will pre-read into
467 the buffer cache.
468
469 lifetime_write_kbytes
470 This file is read-only and shows the number of kilobytes of data that
471 have been written to this filesystem since it was created.
472
473 max_writeback_mb_bump
474 The maximum number of megabytes the writeback code will try to write
475 out before move on to another inode.
476
477 mb_group_prealloc
478 The multiblock allocator will round up allocation requests to a
479 multiple of this tuning parameter if the stripe size is not set in the
480 ext4 superblock
481
482 mb_max_to_scan
483 The maximum number of extents the multiblock allocator will search to
484 find the best extent.
485
486 mb_min_to_scan
487 The minimum number of extents the multiblock allocator will search to
488 find the best extent.
489
490 mb_order2_req
491 Tuning parameter which controls the minimum size for requests (as a
492 power of 2) where the buddy cache is used.
493
494 mb_stats
495 Controls whether the multiblock allocator should collect statistics,
496 which are shown during the unmount. 1 means to collect statistics, 0
497 means not to collect statistics.
498
499 mb_stream_req
500 Files which have fewer blocks than this tunable parameter will have
501 their blocks allocated out of a block group specific preallocation
502 pool, so that small files are packed closely together. Each large file
503 will have its blocks allocated out of its own unique preallocation
504 pool.
505
506 session_write_kbytes
507 This file is read-only and shows the number of kilobytes of data that
508 have been written to this filesystem since it was mounted.
509
510 reserved_clusters
511 This is RW file and contains number of reserved clusters in the file
512 system which will be used in the specific situations to avoid costly
513 zeroout, unexpected ENOSPC, or possible data loss. The default is 2% or
514 4096 clusters, whichever is smaller and this can be changed however it
515 can never exceed number of clusters in the file system. If there is not
516 enough space for the reserved space when mounting the file mount will
517 _not_ fail.
518
519 Ioctls
520 ======
521
522 Ext4 implements various ioctls which can be used by applications to access
523 ext4-specific functionality. An incomplete list of these ioctls is shown in the
524 table below. This list includes truly ext4-specific ioctls (``EXT4_IOC_*``) as
525 well as ioctls that may have been ext4-specific originally but are now supported
526 by some other filesystem(s) too (``FS_IOC_*``).
527
528 Table of Ext4 ioctls
529
530 FS_IOC_GETFLAGS
531 Get additional attributes associated with inode. The ioctl argument is
532 an integer bitfield, with bit values described in ext4.h.
533
534 FS_IOC_SETFLAGS
535 Set additional attributes associated with inode. The ioctl argument is
536 an integer bitfield, with bit values described in ext4.h.
537
538 EXT4_IOC_GETVERSION, EXT4_IOC_GETVERSION_OLD
539 Get the inode i_generation number stored for each inode. The
540 i_generation number is normally changed only when new inode is created
541 and it is particularly useful for network filesystems. The '_OLD'
542 version of this ioctl is an alias for FS_IOC_GETVERSION.
543
544 EXT4_IOC_SETVERSION, EXT4_IOC_SETVERSION_OLD
545 Set the inode i_generation number stored for each inode. The '_OLD'
546 version of this ioctl is an alias for FS_IOC_SETVERSION.
547
548 EXT4_IOC_GROUP_EXTEND
549 This ioctl has the same purpose as the resize mount option. It allows
550 to resize filesystem to the end of the last existing block group,
551 further resize has to be done with resize2fs, either online, or
552 offline. The argument points to the unsigned logn number representing
553 the filesystem new block count.
554
555 EXT4_IOC_MOVE_EXT
556 Move the block extents from orig_fd (the one this ioctl is pointing to)
557 to the donor_fd (the one specified in move_extent structure passed as
558 an argument to this ioctl). Then, exchange inode metadata between
559 orig_fd and donor_fd. This is especially useful for online
560 defragmentation, because the allocator has the opportunity to allocate
561 moved blocks better, ideally into one contiguous extent.
562
563 EXT4_IOC_GROUP_ADD
564 Add a new group descriptor to an existing or new group descriptor
565 block. The new group descriptor is described by ext4_new_group_input
566 structure, which is passed as an argument to this ioctl. This is
567 especially useful in conjunction with EXT4_IOC_GROUP_EXTEND, which
568 allows online resize of the filesystem to the end of the last existing
569 block group. Those two ioctls combined is used in userspace online
570 resize tool (e.g. resize2fs).
571
572 EXT4_IOC_MIGRATE
573 This ioctl operates on the filesystem itself. It converts (migrates)
574 ext3 indirect block mapped inode to ext4 extent mapped inode by walking
575 through indirect block mapping of the original inode and converting
576 contiguous block ranges into ext4 extents of the temporary inode. Then,
577 inodes are swapped. This ioctl might help, when migrating from ext3 to
578 ext4 filesystem, however suggestion is to create fresh ext4 filesystem
579 and copy data from the backup. Note, that filesystem has to support
580 extents for this ioctl to work.
581
582 EXT4_IOC_ALLOC_DA_BLKS
583 Force all of the delay allocated blocks to be allocated to preserve
584 application-expected ext3 behaviour. Note that this will also start
585 triggering a write of the data blocks, but this behaviour may change in
586 the future as it is not necessary and has been done this way only for
587 sake of simplicity.
588
589 EXT4_IOC_RESIZE_FS
590 Resize the filesystem to a new size. The number of blocks of resized
591 filesystem is passed in via 64 bit integer argument. The kernel
592 allocates bitmaps and inode table, the userspace tool thus just passes
593 the new number of blocks.
594
595 EXT4_IOC_SWAP_BOOT
596 Swap i_blocks and associated attributes (like i_blocks, i_size,
597 i_flags, ...) from the specified inode with inode EXT4_BOOT_LOADER_INO
598 (#5). This is typically used to store a boot loader in a secure part of
599 the filesystem, where it can't be changed by a normal user by accident.
600 The data blocks of the previous boot loader will be associated with the
601 given inode.
602
603 References
604 ==========
605
606 kernel source: <file:fs/ext4/>
607 <file:fs/jbd2/>
608
609 programs: http://e2fsprogs.sourceforge.net/
610
611 useful links: https://fedoraproject.org/wiki/ext3-devel
612 http://www.bullopensource.org/ext4/
613 http://ext4.wiki.kernel.org/index.php/Main_Page
614 https://fedoraproject.org/wiki/Features/Ext4
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.