]> git.proxmox.com Git - mirror_ubuntu-zesty-kernel.git/blob - Documentation/filesystems/ext4.txt
Merge tag 'v3.9-rc3' into drm-intel-next-queued
[mirror_ubuntu-zesty-kernel.git] / Documentation / filesystems / ext4.txt
1
2 Ext4 Filesystem
3 ===============
4
5 Ext4 is an an advanced level of the ext3 filesystem which incorporates
6 scalability and reliability enhancements for supporting large filesystems
7 (64 bit) in keeping with increasing disk capacities and state-of-the-art
8 feature requirements.
9
10 Mailing list: linux-ext4@vger.kernel.org
11 Web site: http://ext4.wiki.kernel.org
12
13
14 1. Quick usage instructions:
15 ===========================
16
17 Note: More extensive information for getting started with ext4 can be
18 found at the ext4 wiki site at the URL:
19 http://ext4.wiki.kernel.org/index.php/Ext4_Howto
20
21 - Compile and install the latest version of e2fsprogs (as of this
22 writing version 1.41.3) from:
23
24 http://sourceforge.net/project/showfiles.php?group_id=2406
25
26 or
27
28 ftp://ftp.kernel.org/pub/linux/kernel/people/tytso/e2fsprogs/
29
30 or grab the latest git repository from:
31
32 git://git.kernel.org/pub/scm/fs/ext2/e2fsprogs.git
33
34 - Note that it is highly important to install the mke2fs.conf file
35 that comes with the e2fsprogs 1.41.x sources in /etc/mke2fs.conf. If
36 you have edited the /etc/mke2fs.conf file installed on your system,
37 you will need to merge your changes with the version from e2fsprogs
38 1.41.x.
39
40 - Create a new filesystem using the ext4 filesystem type:
41
42 # mke2fs -t ext4 /dev/hda1
43
44 Or to configure an existing ext3 filesystem to support extents:
45
46 # tune2fs -O extents /dev/hda1
47
48 If the filesystem was created with 128 byte inodes, it can be
49 converted to use 256 byte for greater efficiency via:
50
51 # tune2fs -I 256 /dev/hda1
52
53 (Note: we currently do not have tools to convert an ext4
54 filesystem back to ext3; so please do not do try this on production
55 filesystems.)
56
57 - Mounting:
58
59 # mount -t ext4 /dev/hda1 /wherever
60
61 - When comparing performance with other filesystems, it's always
62 important to try multiple workloads; very often a subtle change in a
63 workload parameter can completely change the ranking of which
64 filesystems do well compared to others. When comparing versus ext3,
65 note that ext4 enables write barriers by default, while ext3 does
66 not enable write barriers by default. So it is useful to use
67 explicitly specify whether barriers are enabled or not when via the
68 '-o barriers=[0|1]' mount option for both ext3 and ext4 filesystems
69 for a fair comparison. When tuning ext3 for best benchmark numbers,
70 it is often worthwhile to try changing the data journaling mode; '-o
71 data=writeback' can be faster for some workloads. (Note however that
72 running mounted with data=writeback can potentially leave stale data
73 exposed in recently written files in case of an unclean shutdown,
74 which could be a security exposure in some situations.) Configuring
75 the filesystem with a large journal can also be helpful for
76 metadata-intensive workloads.
77
78 2. Features
79 ===========
80
81 2.1 Currently available
82
83 * ability to use filesystems > 16TB (e2fsprogs support not available yet)
84 * extent format reduces metadata overhead (RAM, IO for access, transactions)
85 * extent format more robust in face of on-disk corruption due to magics,
86 * internal redundancy in tree
87 * improved file allocation (multi-block alloc)
88 * lift 32000 subdirectory limit imposed by i_links_count[1]
89 * nsec timestamps for mtime, atime, ctime, create time
90 * inode version field on disk (NFSv4, Lustre)
91 * reduced e2fsck time via uninit_bg feature
92 * journal checksumming for robustness, performance
93 * persistent file preallocation (e.g for streaming media, databases)
94 * ability to pack bitmaps and inode tables into larger virtual groups via the
95 flex_bg feature
96 * large file support
97 * Inode allocation using large virtual block groups via flex_bg
98 * delayed allocation
99 * large block (up to pagesize) support
100 * efficient new ordered mode in JBD2 and ext4(avoid using buffer head to force
101 the ordering)
102
103 [1] Filesystems with a block size of 1k may see a limit imposed by the
104 directory hash tree having a maximum depth of two.
105
106 2.2 Candidate features for future inclusion
107
108 * Online defrag (patches available but not well tested)
109 * reduced mke2fs time via lazy itable initialization in conjunction with
110 the uninit_bg feature (capability to do this is available in e2fsprogs
111 but a kernel thread to do lazy zeroing of unused inode table blocks
112 after filesystem is first mounted is required for safety)
113
114 There are several others under discussion, whether they all make it in is
115 partly a function of how much time everyone has to work on them. Features like
116 metadata checksumming have been discussed and planned for a bit but no patches
117 exist yet so I'm not sure they're in the near-term roadmap.
118
119 The big performance win will come with mballoc, delalloc and flex_bg
120 grouping of bitmaps and inode tables. Some test results available here:
121
122 - http://www.bullopensource.org/ext4/20080818-ffsb/ffsb-write-2.6.27-rc1.html
123 - http://www.bullopensource.org/ext4/20080818-ffsb/ffsb-readwrite-2.6.27-rc1.html
124
125 3. Options
126 ==========
127
128 When mounting an ext4 filesystem, the following option are accepted:
129 (*) == default
130
131 ro Mount filesystem read only. Note that ext4 will
132 replay the journal (and thus write to the
133 partition) even when mounted "read only". The
134 mount options "ro,noload" can be used to prevent
135 writes to the filesystem.
136
137 journal_checksum Enable checksumming of the journal transactions.
138 This will allow the recovery code in e2fsck and the
139 kernel to detect corruption in the kernel. It is a
140 compatible change and will be ignored by older kernels.
141
142 journal_async_commit Commit block can be written to disk without waiting
143 for descriptor blocks. If enabled older kernels cannot
144 mount the device. This will enable 'journal_checksum'
145 internally.
146
147 journal_dev=devnum When the external journal device's major/minor numbers
148 have changed, this option allows the user to specify
149 the new journal location. The journal device is
150 identified through its new major/minor numbers encoded
151 in devnum.
152
153 norecovery Don't load the journal on mounting. Note that
154 noload if the filesystem was not unmounted cleanly,
155 skipping the journal replay will lead to the
156 filesystem containing inconsistencies that can
157 lead to any number of problems.
158
159 data=journal All data are committed into the journal prior to being
160 written into the main file system. Enabling
161 this mode will disable delayed allocation and
162 O_DIRECT support.
163
164 data=ordered (*) All data are forced directly out to the main file
165 system prior to its metadata being committed to the
166 journal.
167
168 data=writeback Data ordering is not preserved, data may be written
169 into the main file system after its metadata has been
170 committed to the journal.
171
172 commit=nrsec (*) Ext4 can be told to sync all its data and metadata
173 every 'nrsec' seconds. The default value is 5 seconds.
174 This means that if you lose your power, you will lose
175 as much as the latest 5 seconds of work (your
176 filesystem will not be damaged though, thanks to the
177 journaling). This default value (or any low value)
178 will hurt performance, but it's good for data-safety.
179 Setting it to 0 will have the same effect as leaving
180 it at the default (5 seconds).
181 Setting it to very large values will improve
182 performance.
183
184 barrier=<0|1(*)> This enables/disables the use of write barriers in
185 barrier(*) the jbd code. barrier=0 disables, barrier=1 enables.
186 nobarrier This also requires an IO stack which can support
187 barriers, and if jbd gets an error on a barrier
188 write, it will disable again with a warning.
189 Write barriers enforce proper on-disk ordering
190 of journal commits, making volatile disk write caches
191 safe to use, at some performance penalty. If
192 your disks are battery-backed in one way or another,
193 disabling barriers may safely improve performance.
194 The mount options "barrier" and "nobarrier" can
195 also be used to enable or disable barriers, for
196 consistency with other ext4 mount options.
197
198 inode_readahead_blks=n This tuning parameter controls the maximum
199 number of inode table blocks that ext4's inode
200 table readahead algorithm will pre-read into
201 the buffer cache. The default value is 32 blocks.
202
203 nouser_xattr Disables Extended User Attributes. See the
204 attr(5) manual page and http://acl.bestbits.at/
205 for more information about extended attributes.
206
207 noacl This option disables POSIX Access Control List
208 support. If ACL support is enabled in the kernel
209 configuration (CONFIG_EXT4_FS_POSIX_ACL), ACL is
210 enabled by default on mount. See the acl(5) manual
211 page and http://acl.bestbits.at/ for more information
212 about acl.
213
214 bsddf (*) Make 'df' act like BSD.
215 minixdf Make 'df' act like Minix.
216
217 debug Extra debugging information is sent to syslog.
218
219 abort Simulate the effects of calling ext4_abort() for
220 debugging purposes. This is normally used while
221 remounting a filesystem which is already mounted.
222
223 errors=remount-ro Remount the filesystem read-only on an error.
224 errors=continue Keep going on a filesystem error.
225 errors=panic Panic and halt the machine if an error occurs.
226 (These mount options override the errors behavior
227 specified in the superblock, which can be configured
228 using tune2fs)
229
230 data_err=ignore(*) Just print an error message if an error occurs
231 in a file data buffer in ordered mode.
232 data_err=abort Abort the journal if an error occurs in a file
233 data buffer in ordered mode.
234
235 grpid Give objects the same group ID as their creator.
236 bsdgroups
237
238 nogrpid (*) New objects have the group ID of their creator.
239 sysvgroups
240
241 resgid=n The group ID which may use the reserved blocks.
242
243 resuid=n The user ID which may use the reserved blocks.
244
245 sb=n Use alternate superblock at this location.
246
247 quota These options are ignored by the filesystem. They
248 noquota are used only by quota tools to recognize volumes
249 grpquota where quota should be turned on. See documentation
250 usrquota in the quota-tools package for more details
251 (http://sourceforge.net/projects/linuxquota).
252
253 jqfmt=<quota type> These options tell filesystem details about quota
254 usrjquota=<file> so that quota information can be properly updated
255 grpjquota=<file> during journal replay. They replace the above
256 quota options. See documentation in the quota-tools
257 package for more details
258 (http://sourceforge.net/projects/linuxquota).
259
260 stripe=n Number of filesystem blocks that mballoc will try
261 to use for allocation size and alignment. For RAID5/6
262 systems this should be the number of data
263 disks * RAID chunk size in file system blocks.
264
265 delalloc (*) Defer block allocation until just before ext4
266 writes out the block(s) in question. This
267 allows ext4 to better allocation decisions
268 more efficiently.
269 nodelalloc Disable delayed allocation. Blocks are allocated
270 when the data is copied from userspace to the
271 page cache, either via the write(2) system call
272 or when an mmap'ed page which was previously
273 unallocated is written for the first time.
274
275 max_batch_time=usec Maximum amount of time ext4 should wait for
276 additional filesystem operations to be batch
277 together with a synchronous write operation.
278 Since a synchronous write operation is going to
279 force a commit and then a wait for the I/O
280 complete, it doesn't cost much, and can be a
281 huge throughput win, we wait for a small amount
282 of time to see if any other transactions can
283 piggyback on the synchronous write. The
284 algorithm used is designed to automatically tune
285 for the speed of the disk, by measuring the
286 amount of time (on average) that it takes to
287 finish committing a transaction. Call this time
288 the "commit time". If the time that the
289 transaction has been running is less than the
290 commit time, ext4 will try sleeping for the
291 commit time to see if other operations will join
292 the transaction. The commit time is capped by
293 the max_batch_time, which defaults to 15000us
294 (15ms). This optimization can be turned off
295 entirely by setting max_batch_time to 0.
296
297 min_batch_time=usec This parameter sets the commit time (as
298 described above) to be at least min_batch_time.
299 It defaults to zero microseconds. Increasing
300 this parameter may improve the throughput of
301 multi-threaded, synchronous workloads on very
302 fast disks, at the cost of increasing latency.
303
304 journal_ioprio=prio The I/O priority (from 0 to 7, where 0 is the
305 highest priority) which should be used for I/O
306 operations submitted by kjournald2 during a
307 commit operation. This defaults to 3, which is
308 a slightly higher priority than the default I/O
309 priority.
310
311 auto_da_alloc(*) Many broken applications don't use fsync() when
312 noauto_da_alloc replacing existing files via patterns such as
313 fd = open("foo.new")/write(fd,..)/close(fd)/
314 rename("foo.new", "foo"), or worse yet,
315 fd = open("foo", O_TRUNC)/write(fd,..)/close(fd).
316 If auto_da_alloc is enabled, ext4 will detect
317 the replace-via-rename and replace-via-truncate
318 patterns and force that any delayed allocation
319 blocks are allocated such that at the next
320 journal commit, in the default data=ordered
321 mode, the data blocks of the new file are forced
322 to disk before the rename() operation is
323 committed. This provides roughly the same level
324 of guarantees as ext3, and avoids the
325 "zero-length" problem that can happen when a
326 system crashes before the delayed allocation
327 blocks are forced to disk.
328
329 noinit_itable Do not initialize any uninitialized inode table
330 blocks in the background. This feature may be
331 used by installation CD's so that the install
332 process can complete as quickly as possible; the
333 inode table initialization process would then be
334 deferred until the next time the file system
335 is unmounted.
336
337 init_itable=n The lazy itable init code will wait n times the
338 number of milliseconds it took to zero out the
339 previous block group's inode table. This
340 minimizes the impact on the system performance
341 while file system's inode table is being initialized.
342
343 discard Controls whether ext4 should issue discard/TRIM
344 nodiscard(*) commands to the underlying block device when
345 blocks are freed. This is useful for SSD devices
346 and sparse/thinly-provisioned LUNs, but it is off
347 by default until sufficient testing has been done.
348
349 nouid32 Disables 32-bit UIDs and GIDs. This is for
350 interoperability with older kernels which only
351 store and expect 16-bit values.
352
353 block_validity This options allows to enables/disables the in-kernel
354 noblock_validity facility for tracking filesystem metadata blocks
355 within internal data structures. This allows multi-
356 block allocator and other routines to quickly locate
357 extents which might overlap with filesystem metadata
358 blocks. This option is intended for debugging
359 purposes and since it negatively affects the
360 performance, it is off by default.
361
362 dioread_lock Controls whether or not ext4 should use the DIO read
363 dioread_nolock locking. If the dioread_nolock option is specified
364 ext4 will allocate uninitialized extent before buffer
365 write and convert the extent to initialized after IO
366 completes. This approach allows ext4 code to avoid
367 using inode mutex, which improves scalability on high
368 speed storages. However this does not work with
369 data journaling and dioread_nolock option will be
370 ignored with kernel warning. Note that dioread_nolock
371 code path is only used for extent-based files.
372 Because of the restrictions this options comprises
373 it is off by default (e.g. dioread_lock).
374
375 max_dir_size_kb=n This limits the size of directories so that any
376 attempt to expand them beyond the specified
377 limit in kilobytes will cause an ENOSPC error.
378 This is useful in memory constrained
379 environments, where a very large directory can
380 cause severe performance problems or even
381 provoke the Out Of Memory killer. (For example,
382 if there is only 512mb memory available, a 176mb
383 directory may seriously cramp the system's style.)
384
385 i_version Enable 64-bit inode version support. This option is
386 off by default.
387
388 Data Mode
389 =========
390 There are 3 different data modes:
391
392 * writeback mode
393 In data=writeback mode, ext4 does not journal data at all. This mode provides
394 a similar level of journaling as that of XFS, JFS, and ReiserFS in its default
395 mode - metadata journaling. A crash+recovery can cause incorrect data to
396 appear in files which were written shortly before the crash. This mode will
397 typically provide the best ext4 performance.
398
399 * ordered mode
400 In data=ordered mode, ext4 only officially journals metadata, but it logically
401 groups metadata information related to data changes with the data blocks into a
402 single unit called a transaction. When it's time to write the new metadata
403 out to disk, the associated data blocks are written first. In general,
404 this mode performs slightly slower than writeback but significantly faster than journal mode.
405
406 * journal mode
407 data=journal mode provides full data and metadata journaling. All new data is
408 written to the journal first, and then to its final location.
409 In the event of a crash, the journal can be replayed, bringing both data and
410 metadata into a consistent state. This mode is the slowest except when data
411 needs to be read from and written to disk at the same time where it
412 outperforms all others modes. Enabling this mode will disable delayed
413 allocation and O_DIRECT support.
414
415 /proc entries
416 =============
417
418 Information about mounted ext4 file systems can be found in
419 /proc/fs/ext4. Each mounted filesystem will have a directory in
420 /proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or
421 /proc/fs/ext4/dm-0). The files in each per-device directory are shown
422 in table below.
423
424 Files in /proc/fs/ext4/<devname>
425 ..............................................................................
426 File Content
427 mb_groups details of multiblock allocator buddy cache of free blocks
428 ..............................................................................
429
430 /sys entries
431 ============
432
433 Information about mounted ext4 file systems can be found in
434 /sys/fs/ext4. Each mounted filesystem will have a directory in
435 /sys/fs/ext4 based on its device name (i.e., /sys/fs/ext4/hdc or
436 /sys/fs/ext4/dm-0). The files in each per-device directory are shown
437 in table below.
438
439 Files in /sys/fs/ext4/<devname>
440 (see also Documentation/ABI/testing/sysfs-fs-ext4)
441 ..............................................................................
442 File Content
443
444 delayed_allocation_blocks This file is read-only and shows the number of
445 blocks that are dirty in the page cache, but
446 which do not have their location in the
447 filesystem allocated yet.
448
449 inode_goal Tuning parameter which (if non-zero) controls
450 the goal inode used by the inode allocator in
451 preference to all other allocation heuristics.
452 This is intended for debugging use only, and
453 should be 0 on production systems.
454
455 inode_readahead_blks Tuning parameter which controls the maximum
456 number of inode table blocks that ext4's inode
457 table readahead algorithm will pre-read into
458 the buffer cache
459
460 lifetime_write_kbytes This file is read-only and shows the number of
461 kilobytes of data that have been written to this
462 filesystem since it was created.
463
464 max_writeback_mb_bump The maximum number of megabytes the writeback
465 code will try to write out before move on to
466 another inode.
467
468 mb_group_prealloc The multiblock allocator will round up allocation
469 requests to a multiple of this tuning parameter if
470 the stripe size is not set in the ext4 superblock
471
472 mb_max_to_scan The maximum number of extents the multiblock
473 allocator will search to find the best extent
474
475 mb_min_to_scan The minimum number of extents the multiblock
476 allocator will search to find the best extent
477
478 mb_order2_req Tuning parameter which controls the minimum size
479 for requests (as a power of 2) where the buddy
480 cache is used
481
482 mb_stats Controls whether the multiblock allocator should
483 collect statistics, which are shown during the
484 unmount. 1 means to collect statistics, 0 means
485 not to collect statistics
486
487 mb_stream_req Files which have fewer blocks than this tunable
488 parameter will have their blocks allocated out
489 of a block group specific preallocation pool, so
490 that small files are packed closely together.
491 Each large file will have its blocks allocated
492 out of its own unique preallocation pool.
493
494 session_write_kbytes This file is read-only and shows the number of
495 kilobytes of data that have been written to this
496 filesystem since it was mounted.
497 ..............................................................................
498
499 Ioctls
500 ======
501
502 There is some Ext4 specific functionality which can be accessed by applications
503 through the system call interfaces. The list of all Ext4 specific ioctls are
504 shown in the table below.
505
506 Table of Ext4 specific ioctls
507 ..............................................................................
508 Ioctl Description
509 EXT4_IOC_GETFLAGS Get additional attributes associated with inode.
510 The ioctl argument is an integer bitfield, with
511 bit values described in ext4.h. This ioctl is an
512 alias for FS_IOC_GETFLAGS.
513
514 EXT4_IOC_SETFLAGS Set additional attributes associated with inode.
515 The ioctl argument is an integer bitfield, with
516 bit values described in ext4.h. This ioctl is an
517 alias for FS_IOC_SETFLAGS.
518
519 EXT4_IOC_GETVERSION
520 EXT4_IOC_GETVERSION_OLD
521 Get the inode i_generation number stored for
522 each inode. The i_generation number is normally
523 changed only when new inode is created and it is
524 particularly useful for network filesystems. The
525 '_OLD' version of this ioctl is an alias for
526 FS_IOC_GETVERSION.
527
528 EXT4_IOC_SETVERSION
529 EXT4_IOC_SETVERSION_OLD
530 Set the inode i_generation number stored for
531 each inode. The '_OLD' version of this ioctl
532 is an alias for FS_IOC_SETVERSION.
533
534 EXT4_IOC_GROUP_EXTEND This ioctl has the same purpose as the resize
535 mount option. It allows to resize filesystem
536 to the end of the last existing block group,
537 further resize has to be done with resize2fs,
538 either online, or offline. The argument points
539 to the unsigned logn number representing the
540 filesystem new block count.
541
542 EXT4_IOC_MOVE_EXT Move the block extents from orig_fd (the one
543 this ioctl is pointing to) to the donor_fd (the
544 one specified in move_extent structure passed
545 as an argument to this ioctl). Then, exchange
546 inode metadata between orig_fd and donor_fd.
547 This is especially useful for online
548 defragmentation, because the allocator has the
549 opportunity to allocate moved blocks better,
550 ideally into one contiguous extent.
551
552 EXT4_IOC_GROUP_ADD Add a new group descriptor to an existing or
553 new group descriptor block. The new group
554 descriptor is described by ext4_new_group_input
555 structure, which is passed as an argument to
556 this ioctl. This is especially useful in
557 conjunction with EXT4_IOC_GROUP_EXTEND,
558 which allows online resize of the filesystem
559 to the end of the last existing block group.
560 Those two ioctls combined is used in userspace
561 online resize tool (e.g. resize2fs).
562
563 EXT4_IOC_MIGRATE This ioctl operates on the filesystem itself.
564 It converts (migrates) ext3 indirect block mapped
565 inode to ext4 extent mapped inode by walking
566 through indirect block mapping of the original
567 inode and converting contiguous block ranges
568 into ext4 extents of the temporary inode. Then,
569 inodes are swapped. This ioctl might help, when
570 migrating from ext3 to ext4 filesystem, however
571 suggestion is to create fresh ext4 filesystem
572 and copy data from the backup. Note, that
573 filesystem has to support extents for this ioctl
574 to work.
575
576 EXT4_IOC_ALLOC_DA_BLKS Force all of the delay allocated blocks to be
577 allocated to preserve application-expected ext3
578 behaviour. Note that this will also start
579 triggering a write of the data blocks, but this
580 behaviour may change in the future as it is
581 not necessary and has been done this way only
582 for sake of simplicity.
583
584 EXT4_IOC_RESIZE_FS Resize the filesystem to a new size. The number
585 of blocks of resized filesystem is passed in via
586 64 bit integer argument. The kernel allocates
587 bitmaps and inode table, the userspace tool thus
588 just passes the new number of blocks.
589
590 ..............................................................................
591
592 References
593 ==========
594
595 kernel source: <file:fs/ext4/>
596 <file:fs/jbd2/>
597
598 programs: http://e2fsprogs.sourceforge.net/
599
600 useful links: http://fedoraproject.org/wiki/ext3-devel
601 http://www.bullopensource.org/ext4/
602 http://ext4.wiki.kernel.org/index.php/Main_Page
603 http://fedoraproject.org/wiki/Features/Ext4