1
2Ext4 Filesystem
3===============
4
5Ext4 is an an advanced level of the ext3 filesystem which incorporates
6scalability and reliability enhancements for supporting large filesystems
7(64 bit) in keeping with increasing disk capacities and state-of-the-art
8feature requirements.
9
10Mailing list:	linux-ext4@vger.kernel.org
11Web site:	http://ext4.wiki.kernel.org
12
13
141. Quick usage instructions:
15===========================
16
17Note: 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
782. Features
79===========
80
812.1 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
104directory hash tree having a maximum depth of two.
105
1062.2 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
114There are several others under discussion, whether they all make it in is
115partly a function of how much time everyone has to work on them. Features like
116metadata checksumming have been discussed and planned for a bit but no patches
117exist yet so I'm not sure they're in the near-term roadmap.
118
119The big performance win will come with mballoc, delalloc and flex_bg
120grouping 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
1253. Options
126==========
127
128When mounting an ext4 filesystem, the following option are accepted:
129(*) == default
130
131ro                   	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
137journal_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
142journal_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
147journal=update		Update the ext4 file system's journal to the current
148			format.
149
150journal_dev=devnum	When the external journal device's major/minor numbers
151			have changed, this option allows the user to specify
152			the new journal location.  The journal device is
153			identified through its new major/minor numbers encoded
154			in devnum.
155
156norecovery		Don't load the journal on mounting.  Note that
157noload			if the filesystem was not unmounted cleanly,
158                     	skipping the journal replay will lead to the
159                     	filesystem containing inconsistencies that can
160                     	lead to any number of problems.
161
162data=journal		All data are committed into the journal prior to being
163			written into the main file system.  Enabling
164			this mode will disable delayed allocation and
165			O_DIRECT support.
166
167data=ordered	(*)	All data are forced directly out to the main file
168			system prior to its metadata being committed to the
169			journal.
170
171data=writeback		Data ordering is not preserved, data may be written
172			into the main file system after its metadata has been
173			committed to the journal.
174
175commit=nrsec	(*)	Ext4 can be told to sync all its data and metadata
176			every 'nrsec' seconds. The default value is 5 seconds.
177			This means that if you lose your power, you will lose
178			as much as the latest 5 seconds of work (your
179			filesystem will not be damaged though, thanks to the
180			journaling).  This default value (or any low value)
181			will hurt performance, but it's good for data-safety.
182			Setting it to 0 will have the same effect as leaving
183			it at the default (5 seconds).
184			Setting it to very large values will improve
185			performance.
186
187barrier=<0|1(*)>	This enables/disables the use of write barriers in
188barrier(*)		the jbd code.  barrier=0 disables, barrier=1 enables.
189nobarrier		This also requires an IO stack which can support
190			barriers, and if jbd gets an error on a barrier
191			write, it will disable again with a warning.
192			Write barriers enforce proper on-disk ordering
193			of journal commits, making volatile disk write caches
194			safe to use, at some performance penalty.  If
195			your disks are battery-backed in one way or another,
196			disabling barriers may safely improve performance.
197			The mount options "barrier" and "nobarrier" can
198			also be used to enable or disable barriers, for
199			consistency with other ext4 mount options.
200
201inode_readahead_blks=n	This tuning parameter controls the maximum
202			number of inode table blocks that ext4's inode
203			table readahead algorithm will pre-read into
204			the buffer cache.  The default value is 32 blocks.
205
206nouser_xattr		Disables Extended User Attributes. If you have extended
207			attribute support enabled in the kernel configuration
208			(CONFIG_EXT4_FS_XATTR), extended attribute support
209			is enabled by default on mount. See the attr(5) manual
210			page and http://acl.bestbits.at/ for more information
211			about extended attributes.
212
213noacl			This option disables POSIX Access Control List
214			support. If ACL support is enabled in the kernel
215			configuration (CONFIG_EXT4_FS_POSIX_ACL), ACL is
216			enabled by default on mount. See the acl(5) manual
217			page and http://acl.bestbits.at/ for more information
218			about acl.
219
220bsddf		(*)	Make 'df' act like BSD.
221minixdf			Make 'df' act like Minix.
222
223debug			Extra debugging information is sent to syslog.
224
225abort			Simulate the effects of calling ext4_abort() for
226			debugging purposes.  This is normally used while
227			remounting a filesystem which is already mounted.
228
229errors=remount-ro	Remount the filesystem read-only on an error.
230errors=continue		Keep going on a filesystem error.
231errors=panic		Panic and halt the machine if an error occurs.
232                        (These mount options override the errors behavior
233                        specified in the superblock, which can be configured
234                        using tune2fs)
235
236data_err=ignore(*)	Just print an error message if an error occurs
237			in a file data buffer in ordered mode.
238data_err=abort		Abort the journal if an error occurs in a file
239			data buffer in ordered mode.
240
241grpid			Give objects the same group ID as their creator.
242bsdgroups
243
244nogrpid		(*)	New objects have the group ID of their creator.
245sysvgroups
246
247resgid=n		The group ID which may use the reserved blocks.
248
249resuid=n		The user ID which may use the reserved blocks.
250
251sb=n			Use alternate superblock at this location.
252
253quota			These options are ignored by the filesystem. They
254noquota			are used only by quota tools to recognize volumes
255grpquota		where quota should be turned on. See documentation
256usrquota		in the quota-tools package for more details
257			(http://sourceforge.net/projects/linuxquota).
258
259jqfmt=<quota type>	These options tell filesystem details about quota
260usrjquota=<file>	so that quota information can be properly updated
261grpjquota=<file>	during journal replay. They replace the above
262			quota options. See documentation in the quota-tools
263			package for more details
264			(http://sourceforge.net/projects/linuxquota).
265
266stripe=n		Number of filesystem blocks that mballoc will try
267			to use for allocation size and alignment. For RAID5/6
268			systems this should be the number of data
269			disks *  RAID chunk size in file system blocks.
270
271delalloc	(*)	Defer block allocation until just before ext4
272			writes out the block(s) in question.  This
273			allows ext4 to better allocation decisions
274			more efficiently.
275nodelalloc		Disable delayed allocation.  Blocks are allocated
276			when the data is copied from userspace to the
277			page cache, either via the write(2) system call
278			or when an mmap'ed page which was previously
279			unallocated is written for the first time.
280
281max_batch_time=usec	Maximum amount of time ext4 should wait for
282			additional filesystem operations to be batch
283			together with a synchronous write operation.
284			Since a synchronous write operation is going to
285			force a commit and then a wait for the I/O
286			complete, it doesn't cost much, and can be a
287			huge throughput win, we wait for a small amount
288			of time to see if any other transactions can
289			piggyback on the synchronous write.   The
290			algorithm used is designed to automatically tune
291			for the speed of the disk, by measuring the
292			amount of time (on average) that it takes to
293			finish committing a transaction.  Call this time
294			the "commit time".  If the time that the
295			transaction has been running is less than the
296			commit time, ext4 will try sleeping for the
297			commit time to see if other operations will join
298			the transaction.   The commit time is capped by
299			the max_batch_time, which defaults to 15000us
300			(15ms).   This optimization can be turned off
301			entirely by setting max_batch_time to 0.
302
303min_batch_time=usec	This parameter sets the commit time (as
304			described above) to be at least min_batch_time.
305			It defaults to zero microseconds.  Increasing
306			this parameter may improve the throughput of
307			multi-threaded, synchronous workloads on very
308			fast disks, at the cost of increasing latency.
309
310journal_ioprio=prio	The I/O priority (from 0 to 7, where 0 is the
311			highest priorty) which should be used for I/O
312			operations submitted by kjournald2 during a
313			commit operation.  This defaults to 3, which is
314			a slightly higher priority than the default I/O
315			priority.
316
317auto_da_alloc(*)	Many broken applications don't use fsync() when
318noauto_da_alloc		replacing existing files via patterns such as
319			fd = open("foo.new")/write(fd,..)/close(fd)/
320			rename("foo.new", "foo"), or worse yet,
321			fd = open("foo", O_TRUNC)/write(fd,..)/close(fd).
322			If auto_da_alloc is enabled, ext4 will detect
323			the replace-via-rename and replace-via-truncate
324			patterns and force that any delayed allocation
325			blocks are allocated such that at the next
326			journal commit, in the default data=ordered
327			mode, the data blocks of the new file are forced
328			to disk before the rename() operation is
329			committed.  This provides roughly the same level
330			of guarantees as ext3, and avoids the
331			"zero-length" problem that can happen when a
332			system crashes before the delayed allocation
333			blocks are forced to disk.
334
335noinit_itable		Do not initialize any uninitialized inode table
336			blocks in the background.  This feature may be
337			used by installation CD's so that the install
338			process can complete as quickly as possible; the
339			inode table initialization process would then be
340			deferred until the next time the  file system
341			is unmounted.
342
343init_itable=n		The lazy itable init code will wait n times the
344			number of milliseconds it took to zero out the
345			previous block group's inode table.  This
346			minimizes the impact on the systme performance
347			while file system's inode table is being initialized.
348
349discard			Controls whether ext4 should issue discard/TRIM
350nodiscard(*)		commands to the underlying block device when
351			blocks are freed.  This is useful for SSD devices
352			and sparse/thinly-provisioned LUNs, but it is off
353			by default until sufficient testing has been done.
354
355nouid32			Disables 32-bit UIDs and GIDs.  This is for
356			interoperability  with  older kernels which only
357			store and expect 16-bit values.
358
359resize			Allows to resize filesystem to the end of the last
360			existing block group, further resize has to be done
361			with resize2fs either online, or offline. It can be
362			used only with conjunction with remount.
363
364block_validity		This options allows to enables/disables the in-kernel
365noblock_validity	facility for tracking filesystem metadata blocks
366			within internal data structures. This allows multi-
367			block allocator and other routines to quickly locate
368			extents which might overlap with filesystem metadata
369			blocks. This option is intended for debugging
370			purposes and since it negatively affects the
371			performance, it is off by default.
372
373dioread_lock		Controls whether or not ext4 should use the DIO read
374dioread_nolock		locking. If the dioread_nolock option is specified
375			ext4 will allocate uninitialized extent before buffer
376			write and convert the extent to initialized after IO
377			completes. This approach allows ext4 code to avoid
378			using inode mutex, which improves scalability on high
379			speed storages. However this does not work with
380			data journaling and dioread_nolock option will be
381			ignored with kernel warning. Note that dioread_nolock
382			code path is only used for extent-based files.
383			Because of the restrictions this options comprises
384			it is off by default (e.g. dioread_lock).
385
386i_version		Enable 64-bit inode version support. This option is
387			off by default.
388
389Data Mode
390=========
391There are 3 different data modes:
392
393* writeback mode
394In data=writeback mode, ext4 does not journal data at all.  This mode provides
395a similar level of journaling as that of XFS, JFS, and ReiserFS in its default
396mode - metadata journaling.  A crash+recovery can cause incorrect data to
397appear in files which were written shortly before the crash.  This mode will
398typically provide the best ext4 performance.
399
400* ordered mode
401In data=ordered mode, ext4 only officially journals metadata, but it logically
402groups metadata information related to data changes with the data blocks into a
403single unit called a transaction.  When it's time to write the new metadata
404out to disk, the associated data blocks are written first.  In general,
405this mode performs slightly slower than writeback but significantly faster than journal mode.
406
407* journal mode
408data=journal mode provides full data and metadata journaling.  All new data is
409written to the journal first, and then to its final location.
410In the event of a crash, the journal can be replayed, bringing both data and
411metadata into a consistent state.  This mode is the slowest except when data
412needs to be read from and written to disk at the same time where it
413outperforms all others modes.  Enabling this mode will disable delayed
414allocation and O_DIRECT support.
415
416/proc entries
417=============
418
419Information about mounted ext4 file systems can be found in
420/proc/fs/ext4.  Each mounted filesystem will have a directory in
421/proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or
422/proc/fs/ext4/dm-0).   The files in each per-device directory are shown
423in table below.
424
425Files in /proc/fs/ext4/<devname>
426..............................................................................
427 File            Content
428 mb_groups       details of multiblock allocator buddy cache of free blocks
429..............................................................................
430
431/sys entries
432============
433
434Information about mounted ext4 file systems can be found in
435/sys/fs/ext4.  Each mounted filesystem will have a directory in
436/sys/fs/ext4 based on its device name (i.e., /sys/fs/ext4/hdc or
437/sys/fs/ext4/dm-0).   The files in each per-device directory are shown
438in table below.
439
440Files in /sys/fs/ext4/<devname>
441(see also Documentation/ABI/testing/sysfs-fs-ext4)
442..............................................................................
443 File                         Content
444
445 delayed_allocation_blocks    This file is read-only and shows the number of
446                              blocks that are dirty in the page cache, but
447                              which do not have their location in the
448                              filesystem allocated yet.
449
450 inode_goal                   Tuning parameter which (if non-zero) controls
451                              the goal inode used by the inode allocator in
452                              preference to all other allocation heuristics.
453                              This is intended for debugging use only, and
454                              should be 0 on production systems.
455
456 inode_readahead_blks         Tuning parameter which controls the maximum
457                              number of inode table blocks that ext4's inode
458                              table readahead algorithm will pre-read into
459                              the buffer cache
460
461 lifetime_write_kbytes        This file is read-only and shows the number of
462                              kilobytes of data that have been written to this
463                              filesystem since it was created.
464
465 max_writeback_mb_bump        The maximum number of megabytes the writeback
466                              code will try to write out before move on to
467                              another inode.
468
469 mb_group_prealloc            The multiblock allocator will round up allocation
470                              requests to a multiple of this tuning parameter if
471                              the stripe size is not set in the ext4 superblock
472
473 mb_max_to_scan               The maximum number of extents the multiblock
474                              allocator will search to find the best extent
475
476 mb_min_to_scan               The minimum number of extents the multiblock
477                              allocator will search to find the best extent
478
479 mb_order2_req                Tuning parameter which controls the minimum size
480                              for requests (as a power of 2) where the buddy
481                              cache is used
482
483 mb_stats                     Controls whether the multiblock allocator should
484                              collect statistics, which are shown during the
485                              unmount. 1 means to collect statistics, 0 means
486                              not to collect statistics
487
488 mb_stream_req                Files which have fewer blocks than this tunable
489                              parameter will have their blocks allocated out
490                              of a block group specific preallocation pool, so
491                              that small files are packed closely together.
492                              Each large file will have its blocks allocated
493                              out of its own unique preallocation pool.
494
495 session_write_kbytes         This file is read-only and shows the number of
496                              kilobytes of data that have been written to this
497                              filesystem since it was mounted.
498..............................................................................
499
500Ioctls
501======
502
503There is some Ext4 specific functionality which can be accessed by applications
504through the system call interfaces. The list of all Ext4 specific ioctls are
505shown in the table below.
506
507Table of Ext4 specific ioctls
508..............................................................................
509 Ioctl			      Description
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.
514
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.
519
520 EXT4_IOC_GETVERSION
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
527			      FS_IOC_GETVERSION.
528
529 EXT4_IOC_SETVERSION
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.
534
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.
542
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.
552
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).
563
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
575			      to work.
576
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.
584
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.
590
591..............................................................................
592
593References
594==========
595
596kernel source:	<file:fs/ext4/>
597		<file:fs/jbd2/>
598
599programs:	http://e2fsprogs.sourceforge.net/
600
601useful links:	http://fedoraproject.org/wiki/ext3-devel
602		http://www.bullopensource.org/ext4/
603		http://ext4.wiki.kernel.org/index.php/Main_Page
604		http://fedoraproject.org/wiki/Features/Ext4
605