1.. SPDX-License-Identifier: GPL-2.0
2
3======================
4The SGI XFS Filesystem
5======================
6
7XFS is a high performance journaling filesystem which originated
8on the SGI IRIX platform.  It is completely multi-threaded, can
9support large files and large filesystems, extended attributes,
10variable block sizes, is extent based, and makes extensive use of
11Btrees (directories, extents, free space) to aid both performance
12and scalability.
13
14Refer to the documentation at https://xfs.wiki.kernel.org/
15for further details.  This implementation is on-disk compatible
16with the IRIX version of XFS.
17
18
19Mount Options
20=============
21
22When mounting an XFS filesystem, the following options are accepted.
23
24  allocsize=size
25	Sets the buffered I/O end-of-file preallocation size when
26	doing delayed allocation writeout (default size is 64KiB).
27	Valid values for this option are page size (typically 4KiB)
28	through to 1GiB, inclusive, in power-of-2 increments.
29
30	The default behaviour is for dynamic end-of-file
31	preallocation size, which uses a set of heuristics to
32	optimise the preallocation size based on the current
33	allocation patterns within the file and the access patterns
34	to the file. Specifying a fixed ``allocsize`` value turns off
35	the dynamic behaviour.
36
37  attr2 or noattr2
38	The options enable/disable an "opportunistic" improvement to
39	be made in the way inline extended attributes are stored
40	on-disk.  When the new form is used for the first time when
41	``attr2`` is selected (either when setting or removing extended
42	attributes) the on-disk superblock feature bit field will be
43	updated to reflect this format being in use.
44
45	The default behaviour is determined by the on-disk feature
46	bit indicating that ``attr2`` behaviour is active. If either
47	mount option is set, then that becomes the new default used
48	by the filesystem.
49
50	CRC enabled filesystems always use the ``attr2`` format, and so
51	will reject the ``noattr2`` mount option if it is set.
52
53  discard or nodiscard (default)
54	Enable/disable the issuing of commands to let the block
55	device reclaim space freed by the filesystem.  This is
56	useful for SSD devices, thinly provisioned LUNs and virtual
57	machine images, but may have a performance impact.
58
59	Note: It is currently recommended that you use the ``fstrim``
60	application to ``discard`` unused blocks rather than the ``discard``
61	mount option because the performance impact of this option
62	is quite severe.
63
64  grpid/bsdgroups or nogrpid/sysvgroups (default)
65	These options define what group ID a newly created file
66	gets.  When ``grpid`` is set, it takes the group ID of the
67	directory in which it is created; otherwise it takes the
68	``fsgid`` of the current process, unless the directory has the
69	``setgid`` bit set, in which case it takes the ``gid`` from the
70	parent directory, and also gets the ``setgid`` bit set if it is
71	a directory itself.
72
73  filestreams
74	Make the data allocator use the filestreams allocation mode
75	across the entire filesystem rather than just on directories
76	configured to use it.
77
78  ikeep or noikeep (default)
79	When ``ikeep`` is specified, XFS does not delete empty inode
80	clusters and keeps them around on disk.  When ``noikeep`` is
81	specified, empty inode clusters are returned to the free
82	space pool.
83
84  inode32 or inode64 (default)
85	When ``inode32`` is specified, it indicates that XFS limits
86	inode creation to locations which will not result in inode
87	numbers with more than 32 bits of significance.
88
89	When ``inode64`` is specified, it indicates that XFS is allowed
90	to create inodes at any location in the filesystem,
91	including those which will result in inode numbers occupying
92	more than 32 bits of significance.
93
94	``inode32`` is provided for backwards compatibility with older
95	systems and applications, since 64 bits inode numbers might
96	cause problems for some applications that cannot handle
97	large inode numbers.  If applications are in use which do
98	not handle inode numbers bigger than 32 bits, the ``inode32``
99	option should be specified.
100
101  largeio or nolargeio (default)
102	If ``nolargeio`` is specified, the optimal I/O reported in
103	``st_blksize`` by **stat(2)** will be as small as possible to allow
104	user applications to avoid inefficient read/modify/write
105	I/O.  This is typically the page size of the machine, as
106	this is the granularity of the page cache.
107
108	If ``largeio`` is specified, a filesystem that was created with a
109	``swidth`` specified will return the ``swidth`` value (in bytes)
110	in ``st_blksize``. If the filesystem does not have a ``swidth``
111	specified but does specify an ``allocsize`` then ``allocsize``
112	(in bytes) will be returned instead. Otherwise the behaviour
113	is the same as if ``nolargeio`` was specified.
114
115  logbufs=value
116	Set the number of in-memory log buffers.  Valid numbers
117	range from 2-8 inclusive.
118
119	The default value is 8 buffers.
120
121	If the memory cost of 8 log buffers is too high on small
122	systems, then it may be reduced at some cost to performance
123	on metadata intensive workloads. The ``logbsize`` option below
124	controls the size of each buffer and so is also relevant to
125	this case.
126
127  logbsize=value
128	Set the size of each in-memory log buffer.  The size may be
129	specified in bytes, or in kilobytes with a "k" suffix.
130	Valid sizes for version 1 and version 2 logs are 16384 (16k)
131	and 32768 (32k).  Valid sizes for version 2 logs also
132	include 65536 (64k), 131072 (128k) and 262144 (256k). The
133	logbsize must be an integer multiple of the log
134	stripe unit configured at **mkfs(8)** time.
135
136	The default value for version 1 logs is 32768, while the
137	default value for version 2 logs is MAX(32768, log_sunit).
138
139  logdev=device and rtdev=device
140	Use an external log (metadata journal) and/or real-time device.
141	An XFS filesystem has up to three parts: a data section, a log
142	section, and a real-time section.  The real-time section is
143	optional, and the log section can be separate from the data
144	section or contained within it.
145
146  noalign
147	Data allocations will not be aligned at stripe unit
148	boundaries. This is only relevant to filesystems created
149	with non-zero data alignment parameters (``sunit``, ``swidth``) by
150	**mkfs(8)**.
151
152  norecovery
153	The filesystem will be mounted without running log recovery.
154	If the filesystem was not cleanly unmounted, it is likely to
155	be inconsistent when mounted in ``norecovery`` mode.
156	Some files or directories may not be accessible because of this.
157	Filesystems mounted ``norecovery`` must be mounted read-only or
158	the mount will fail.
159
160  nouuid
161	Don't check for double mounted file systems using the file
162	system ``uuid``.  This is useful to mount LVM snapshot volumes,
163	and often used in combination with ``norecovery`` for mounting
164	read-only snapshots.
165
166  noquota
167	Forcibly turns off all quota accounting and enforcement
168	within the filesystem.
169
170  uquota/usrquota/uqnoenforce/quota
171	User disk quota accounting enabled, and limits (optionally)
172	enforced.  Refer to **xfs_quota(8)** for further details.
173
174  gquota/grpquota/gqnoenforce
175	Group disk quota accounting enabled and limits (optionally)
176	enforced.  Refer to **xfs_quota(8)** for further details.
177
178  pquota/prjquota/pqnoenforce
179	Project disk quota accounting enabled and limits (optionally)
180	enforced.  Refer to **xfs_quota(8)** for further details.
181
182  sunit=value and swidth=value
183	Used to specify the stripe unit and width for a RAID device
184	or a stripe volume.  "value" must be specified in 512-byte
185	block units. These options are only relevant to filesystems
186	that were created with non-zero data alignment parameters.
187
188	The ``sunit`` and ``swidth`` parameters specified must be compatible
189	with the existing filesystem alignment characteristics.  In
190	general, that means the only valid changes to ``sunit`` are
191	increasing it by a power-of-2 multiple. Valid ``swidth`` values
192	are any integer multiple of a valid ``sunit`` value.
193
194	Typically the only time these mount options are necessary if
195	after an underlying RAID device has had it's geometry
196	modified, such as adding a new disk to a RAID5 lun and
197	reshaping it.
198
199  swalloc
200	Data allocations will be rounded up to stripe width boundaries
201	when the current end of file is being extended and the file
202	size is larger than the stripe width size.
203
204  wsync
205	When specified, all filesystem namespace operations are
206	executed synchronously. This ensures that when the namespace
207	operation (create, unlink, etc) completes, the change to the
208	namespace is on stable storage. This is useful in HA setups
209	where failover must not result in clients seeing
210	inconsistent namespace presentation during or after a
211	failover event.
212
213Deprecation of V4 Format
214========================
215
216The V4 filesystem format lacks certain features that are supported by
217the V5 format, such as metadata checksumming, strengthened metadata
218verification, and the ability to store timestamps past the year 2038.
219Because of this, the V4 format is deprecated.  All users should upgrade
220by backing up their files, reformatting, and restoring from the backup.
221
222Administrators and users can detect a V4 filesystem by running xfs_info
223against a filesystem mountpoint and checking for a string containing
224"crc=".  If no such string is found, please upgrade xfsprogs to the
225latest version and try again.
226
227The deprecation will take place in two parts.  Support for mounting V4
228filesystems can now be disabled at kernel build time via Kconfig option.
229The option will default to yes until September 2025, at which time it
230will be changed to default to no.  In September 2030, support will be
231removed from the codebase entirely.
232
233Note: Distributors may choose to withdraw V4 format support earlier than
234the dates listed above.
235
236Deprecated Mount Options
237========================
238
239===========================     ================
240  Name				Removal Schedule
241===========================     ================
242Mounting with V4 filesystem     September 2030
243ikeep/noikeep			September 2025
244attr2/noattr2			September 2025
245===========================     ================
246
247
248Removed Mount Options
249=====================
250
251===========================     =======
252  Name				Removed
253===========================	=======
254  delaylog/nodelaylog		v4.0
255  ihashsize			v4.0
256  irixsgid			v4.0
257  osyncisdsync/osyncisosync	v4.0
258  barrier			v4.19
259  nobarrier			v4.19
260===========================     =======
261
262sysctls
263=======
264
265The following sysctls are available for the XFS filesystem:
266
267  fs.xfs.stats_clear		(Min: 0  Default: 0  Max: 1)
268	Setting this to "1" clears accumulated XFS statistics
269	in /proc/fs/xfs/stat.  It then immediately resets to "0".
270
271  fs.xfs.xfssyncd_centisecs	(Min: 100  Default: 3000  Max: 720000)
272	The interval at which the filesystem flushes metadata
273	out to disk and runs internal cache cleanup routines.
274
275  fs.xfs.filestream_centisecs	(Min: 1  Default: 3000  Max: 360000)
276	The interval at which the filesystem ages filestreams cache
277	references and returns timed-out AGs back to the free stream
278	pool.
279
280  fs.xfs.speculative_prealloc_lifetime
281	(Units: seconds   Min: 1  Default: 300  Max: 86400)
282	The interval at which the background scanning for inodes
283	with unused speculative preallocation runs. The scan
284	removes unused preallocation from clean inodes and releases
285	the unused space back to the free pool.
286
287  fs.xfs.speculative_cow_prealloc_lifetime
288	This is an alias for speculative_prealloc_lifetime.
289
290  fs.xfs.error_level		(Min: 0  Default: 3  Max: 11)
291	A volume knob for error reporting when internal errors occur.
292	This will generate detailed messages & backtraces for filesystem
293	shutdowns, for example.  Current threshold values are:
294
295		XFS_ERRLEVEL_OFF:       0
296		XFS_ERRLEVEL_LOW:       1
297		XFS_ERRLEVEL_HIGH:      5
298
299  fs.xfs.panic_mask		(Min: 0  Default: 0  Max: 256)
300	Causes certain error conditions to call BUG(). Value is a bitmask;
301	OR together the tags which represent errors which should cause panics:
302
303		XFS_NO_PTAG                     0
304		XFS_PTAG_IFLUSH                 0x00000001
305		XFS_PTAG_LOGRES                 0x00000002
306		XFS_PTAG_AILDELETE              0x00000004
307		XFS_PTAG_ERROR_REPORT           0x00000008
308		XFS_PTAG_SHUTDOWN_CORRUPT       0x00000010
309		XFS_PTAG_SHUTDOWN_IOERROR       0x00000020
310		XFS_PTAG_SHUTDOWN_LOGERROR      0x00000040
311		XFS_PTAG_FSBLOCK_ZERO           0x00000080
312		XFS_PTAG_VERIFIER_ERROR         0x00000100
313
314	This option is intended for debugging only.
315
316  fs.xfs.irix_symlink_mode	(Min: 0  Default: 0  Max: 1)
317	Controls whether symlinks are created with mode 0777 (default)
318	or whether their mode is affected by the umask (irix mode).
319
320  fs.xfs.irix_sgid_inherit	(Min: 0  Default: 0  Max: 1)
321	Controls files created in SGID directories.
322	If the group ID of the new file does not match the effective group
323	ID or one of the supplementary group IDs of the parent dir, the
324	ISGID bit is cleared if the irix_sgid_inherit compatibility sysctl
325	is set.
326
327  fs.xfs.inherit_sync		(Min: 0  Default: 1  Max: 1)
328	Setting this to "1" will cause the "sync" flag set
329	by the **xfs_io(8)** chattr command on a directory to be
330	inherited by files in that directory.
331
332  fs.xfs.inherit_nodump		(Min: 0  Default: 1  Max: 1)
333	Setting this to "1" will cause the "nodump" flag set
334	by the **xfs_io(8)** chattr command on a directory to be
335	inherited by files in that directory.
336
337  fs.xfs.inherit_noatime	(Min: 0  Default: 1  Max: 1)
338	Setting this to "1" will cause the "noatime" flag set
339	by the **xfs_io(8)** chattr command on a directory to be
340	inherited by files in that directory.
341
342  fs.xfs.inherit_nosymlinks	(Min: 0  Default: 1  Max: 1)
343	Setting this to "1" will cause the "nosymlinks" flag set
344	by the **xfs_io(8)** chattr command on a directory to be
345	inherited by files in that directory.
346
347  fs.xfs.inherit_nodefrag	(Min: 0  Default: 1  Max: 1)
348	Setting this to "1" will cause the "nodefrag" flag set
349	by the **xfs_io(8)** chattr command on a directory to be
350	inherited by files in that directory.
351
352  fs.xfs.rotorstep		(Min: 1  Default: 1  Max: 256)
353	In "inode32" allocation mode, this option determines how many
354	files the allocator attempts to allocate in the same allocation
355	group before moving to the next allocation group.  The intent
356	is to control the rate at which the allocator moves between
357	allocation groups when allocating extents for new files.
358
359Deprecated Sysctls
360==================
361
362===========================================     ================
363  Name                                          Removal Schedule
364===========================================     ================
365fs.xfs.irix_sgid_inherit                        September 2025
366fs.xfs.irix_symlink_mode                        September 2025
367fs.xfs.speculative_cow_prealloc_lifetime        September 2025
368===========================================     ================
369
370
371Removed Sysctls
372===============
373
374=============================	=======
375  Name				Removed
376=============================	=======
377  fs.xfs.xfsbufd_centisec	v4.0
378  fs.xfs.age_buffer_centisecs	v4.0
379=============================	=======
380
381Error handling
382==============
383
384XFS can act differently according to the type of error found during its
385operation. The implementation introduces the following concepts to the error
386handler:
387
388 -failure speed:
389	Defines how fast XFS should propagate an error upwards when a specific
390	error is found during the filesystem operation. It can propagate
391	immediately, after a defined number of retries, after a set time period,
392	or simply retry forever.
393
394 -error classes:
395	Specifies the subsystem the error configuration will apply to, such as
396	metadata IO or memory allocation. Different subsystems will have
397	different error handlers for which behaviour can be configured.
398
399 -error handlers:
400	Defines the behavior for a specific error.
401
402The filesystem behavior during an error can be set via ``sysfs`` files. Each
403error handler works independently - the first condition met by an error handler
404for a specific class will cause the error to be propagated rather than reset and
405retried.
406
407The action taken by the filesystem when the error is propagated is context
408dependent - it may cause a shut down in the case of an unrecoverable error,
409it may be reported back to userspace, or it may even be ignored because
410there's nothing useful we can with the error or anyone we can report it to (e.g.
411during unmount).
412
413The configuration files are organized into the following hierarchy for each
414mounted filesystem:
415
416  /sys/fs/xfs/<dev>/error/<class>/<error>/
417
418Where:
419  <dev>
420	The short device name of the mounted filesystem. This is the same device
421	name that shows up in XFS kernel error messages as "XFS(<dev>): ..."
422
423  <class>
424	The subsystem the error configuration belongs to. As of 4.9, the defined
425	classes are:
426
427		- "metadata": applies metadata buffer write IO
428
429  <error>
430	The individual error handler configurations.
431
432
433Each filesystem has "global" error configuration options defined in their top
434level directory:
435
436  /sys/fs/xfs/<dev>/error/
437
438  fail_at_unmount		(Min:  0  Default:  1  Max: 1)
439	Defines the filesystem error behavior at unmount time.
440
441	If set to a value of 1, XFS will override all other error configurations
442	during unmount and replace them with "immediate fail" characteristics.
443	i.e. no retries, no retry timeout. This will always allow unmount to
444	succeed when there are persistent errors present.
445
446	If set to 0, the configured retry behaviour will continue until all
447	retries and/or timeouts have been exhausted. This will delay unmount
448	completion when there are persistent errors, and it may prevent the
449	filesystem from ever unmounting fully in the case of "retry forever"
450	handler configurations.
451
452	Note: there is no guarantee that fail_at_unmount can be set while an
453	unmount is in progress. It is possible that the ``sysfs`` entries are
454	removed by the unmounting filesystem before a "retry forever" error
455	handler configuration causes unmount to hang, and hence the filesystem
456	must be configured appropriately before unmount begins to prevent
457	unmount hangs.
458
459Each filesystem has specific error class handlers that define the error
460propagation behaviour for specific errors. There is also a "default" error
461handler defined, which defines the behaviour for all errors that don't have
462specific handlers defined. Where multiple retry constraints are configured for
463a single error, the first retry configuration that expires will cause the error
464to be propagated. The handler configurations are found in the directory:
465
466  /sys/fs/xfs/<dev>/error/<class>/<error>/
467
468  max_retries			(Min: -1  Default: Varies  Max: INTMAX)
469	Defines the allowed number of retries of a specific error before
470	the filesystem will propagate the error. The retry count for a given
471	error context (e.g. a specific metadata buffer) is reset every time
472	there is a successful completion of the operation.
473
474	Setting the value to "-1" will cause XFS to retry forever for this
475	specific error.
476
477	Setting the value to "0" will cause XFS to fail immediately when the
478	specific error is reported.
479
480	Setting the value to "N" (where 0 < N < Max) will make XFS retry the
481	operation "N" times before propagating the error.
482
483  retry_timeout_seconds		(Min:  -1  Default:  Varies  Max: 1 day)
484	Define the amount of time (in seconds) that the filesystem is
485	allowed to retry its operations when the specific error is
486	found.
487
488	Setting the value to "-1" will allow XFS to retry forever for this
489	specific error.
490
491	Setting the value to "0" will cause XFS to fail immediately when the
492	specific error is reported.
493
494	Setting the value to "N" (where 0 < N < Max) will allow XFS to retry the
495	operation for up to "N" seconds before propagating the error.
496
497**Note:** The default behaviour for a specific error handler is dependent on both
498the class and error context. For example, the default values for
499"metadata/ENODEV" are "0" rather than "-1" so that this error handler defaults
500to "fail immediately" behaviour. This is done because ENODEV is a fatal,
501unrecoverable error no matter how many times the metadata IO is retried.
502
503Workqueue Concurrency
504=====================
505
506XFS uses kernel workqueues to parallelize metadata update processes.  This
507enables it to take advantage of storage hardware that can service many IO
508operations simultaneously.  This interface exposes internal implementation
509details of XFS, and as such is explicitly not part of any userspace API/ABI
510guarantee the kernel may give userspace.  These are undocumented features of
511the generic workqueue implementation XFS uses for concurrency, and they are
512provided here purely for diagnostic and tuning purposes and may change at any
513time in the future.
514
515The control knobs for a filesystem's workqueues are organized by task at hand
516and the short name of the data device.  They all can be found in:
517
518  /sys/bus/workqueue/devices/${task}!${device}
519
520================  ===========
521  Task            Description
522================  ===========
523  xfs_iwalk-$pid  Inode scans of the entire filesystem. Currently limited to
524                  mount time quotacheck.
525  xfs-blockgc     Background garbage collection of disk space that have been
526                  speculatively allocated beyond EOF or for staging copy on
527                  write operations.
528================  ===========
529
530For example, the knobs for the quotacheck workqueue for /dev/nvme0n1 would be
531found in /sys/bus/workqueue/devices/xfs_iwalk-1111!nvme0n1/.
532
533The interesting knobs for XFS workqueues are as follows:
534
535============     ===========
536  Knob           Description
537============     ===========
538  max_active     Maximum number of background threads that can be started to
539                 run the work.
540  cpumask        CPUs upon which the threads are allowed to run.
541  nice           Relative priority of scheduling the threads.  These are the
542                 same nice levels that can be applied to userspace processes.
543============     ===========
544