1mkfs.xfs(8) System Manager's Manual mkfs.xfs(8)
2
3
4
6 mkfs.xfs - construct an XFS filesystem
7
9 mkfs.xfs [ -b block_size ] [ -d data_section_options ] [ -f ] [ -i
10 inode_options ] [ -l log_section_options ] [ -n naming_options ] [ -p
11 protofile ] [ -q ] [ -r realtime_section_options ] [ -s sector_size ] [
12 -L label ] [ -N ] [ -K ] device
13
15 mkfs.xfs constructs an XFS filesystem by writing on a special file
16 using the values found in the arguments of the command line. It is
17 invoked automatically by mkfs(8) when it is given the -t xfs option.
18
19 In its simplest (and most commonly used form), the size of the filesys‐
20 tem is determined from the disk driver. As an example, to make a
21 filesystem with an internal log on the first partition on the first
22 SCSI disk, use:
23
24 mkfs.xfs /dev/sda1
25
26 The metadata log can be placed on another device to reduce the number
27 of disk seeks. To create a filesystem on the first partition on the
28 first SCSI disk with a 10000 block log located on the first partition
29 on the second SCSI disk, use:
30
31 mkfs.xfs -l logdev=/dev/sdb1,size=10000b /dev/sda1
32
33 Each of the option elements in the argument list above can be given as
34 multiple comma-separated suboptions if multiple suboptions apply to the
35 same option. Equivalently, each main option can be given multiple
36 times with different suboptions. For example, -l internal,size=10000b
37 and -l internal -l size=10000b are equivalent.
38
39 In the descriptions below, sizes are given in sectors, bytes, blocks,
40 kilobytes, megabytes, gigabytes, etc. Sizes are treated as hexadecimal
41 if prefixed by 0x or 0X, octal if prefixed by 0, or decimal otherwise.
42 The following lists possible multiplication suffixes:
43 s - multiply by sector size (default = 512, see -s option
44 below).
45 b - multiply by filesystem block size (default = 4K, see -b
46 option below).
47 k - multiply by one kilobyte (1,024 bytes).
48 m - multiply by one megabyte (1,048,576 bytes).
49 g - multiply by one gigabyte (1,073,741,824 bytes).
50 t - multiply by one terabyte (1,099,511,627,776 bytes).
51 p - multiply by one petabyte (1,024 terabytes).
52 e - multiply by one exabyte (1,048,576 terabytes).
53
55 -b block_size_options
56 This option specifies the fundamental block size of the filesys‐
57 tem. The valid block_size_options are: log=value or size=value
58 and only one can be supplied. The block size is specified
59 either as a base two logarithm value with log=, or in bytes with
60 size=. The default value is 4096 bytes (4 KiB), the minimum is
61 512, and the maximum is 65536 (64 KiB). XFS on Linux currently
62 only supports pagesize or smaller blocks.
63
64 -d data_section_options
65 These options specify the location, size, and other parameters
66 of the data section of the filesystem. The valid data_sec‐
67 tion_options are:
68
69 agcount=value
70 This is used to specify the number of allocation
71 groups. The data section of the filesystem is
72 divided into allocation groups to improve the per‐
73 formance of XFS. More allocation groups imply that
74 more parallelism can be achieved when allocating
75 blocks and inodes. The minimum allocation group size
76 is 16 MiB; the maximum size is just under 1 TiB.
77 The data section of the filesystem is divided into
78 value allocation groups (default value is scaled
79 automatically based on the underlying device size).
80
81 agsize=value
82 This is an alternative to using the agcount subop‐
83 tion. The value is the desired size of the alloca‐
84 tion group expressed in bytes (usually using the m
85 or g suffixes). This value must be a multiple of
86 the filesystem block size, and must be at least
87 16MiB, and no more than 1TiB, and may be automati‐
88 cally adjusted to properly align with the stripe
89 geometry. The agcount and agsize suboptions are
90 mutually exclusive.
91
92 name=value
93 This can be used to specify the name of the special
94 file containing the filesystem. In this case, the
95 log section must be specified as internal (with a
96 size, see the -l option below) and there can be no
97 real-time section.
98
99 file[=value]
100 This is used to specify that the file given by the
101 name suboption is a regular file. The value is
102 either 0 or 1, with 1 signifying that the file is
103 regular. This suboption is used only to make a
104 filesystem image. If the value is omitted then 1 is
105 assumed.
106
107 size=value
108 This is used to specify the size of the data sec‐
109 tion. This suboption is required if -d file[=1] is
110 given. Otherwise, it is only needed if the filesys‐
111 tem should occupy less space than the size of the
112 special file.
113
114 sunit=value
115 This is used to specify the stripe unit for a RAID
116 device or a logical volume. The value has to be
117 specified in 512-byte block units. Use the su subop‐
118 tion to specify the stripe unit size in bytes. This
119 suboption ensures that data allocations will be
120 stripe unit aligned when the current end of file is
121 being extended and the file size is larger than
122 512KiB. Also inode allocations and the internal log
123 will be stripe unit aligned.
124
125 su=value
126 This is an alternative to using sunit. The su sub‐
127 option is used to specify the stripe unit for a RAID
128 device or a striped logical volume. The value has to
129 be specified in bytes, (usually using the m or g
130 suffixes). This value must be a multiple of the
131 filesystem block size.
132
133 swidth=value
134 This is used to specify the stripe width for a RAID
135 device or a striped logical volume. The value has to
136 be specified in 512-byte block units. Use the sw
137 suboption to specify the stripe width size in bytes.
138 This suboption is required if -d sunit has been
139 specified and it has to be a multiple of the -d
140 sunit suboption.
141
142 sw=value
143 suboption is an alternative to using swidth. The sw
144 suboption is used to specify the stripe width for a
145 RAID device or striped logical volume. The value is
146 expressed as a multiplier of the stripe unit, usu‐
147 ally the same as the number of stripe members in the
148 logical volume configuration, or data disks in a
149 RAID device.
150
151 When a filesystem is created on a logical volume
152 device, mkfs.xfs will automatically query the logi‐
153 cal volume for appropriate sunit and swidth values.
154
155 -f Force overwrite when an existing filesystem is detected on the
156 device. By default, mkfs.xfs will not write to the device if it
157 suspects that there is a filesystem or partition table on the
158 device already.
159
160 -i inode_options
161 This option specifies the inode size of the filesystem, and
162 other inode allocation parameters. The XFS inode contains a
163 fixed-size part and a variable-size part. The variable-size
164 part, whose size is affected by this option, can contain: direc‐
165 tory data, for small directories; attribute data, for small
166 attribute sets; symbolic link data, for small symbolic links;
167 the extent list for the file, for files with a small number of
168 extents; and the root of a tree describing the location of
169 extents for the file, for files with a large number of extents.
170
171 The valid inode_options are:
172
173 size=value | log=value | perblock=value
174 The inode size is specified either as a value in
175 bytes with size=, a base two logarithm value with
176 log=, or as the number fitting in a filesystem block
177 with perblock=. The mininum (and default) value is
178 256 bytes. The maximum value is 2048 (2 KiB) sub‐
179 ject to the restriction that the inode size cannot
180 exceed one half of the filesystem block size.
181
182 XFS uses 64-bit inode numbers internally; however,
183 the number of significant bits in an inode number is
184 affected by filesystem geometry. In practice,
185 filesystem size and inode size are the predominant
186 factors. The Linux kernel (on 32 bit hardware plat‐
187 forms) and most applications cannot currently handle
188 inode numbers greater than 32 significant bits, so
189 if no inode size is given on the command line,
190 mkfs.xfs will attempt to choose a size such that
191 inode numbers will be < 32 bits. If an inode size
192 is specified, or if a filesystem is sufficently
193 large, mkfs.xfs will warn if this will create inode
194 numbers > 32 significant bits.
195
196 maxpct=value
197 This specifies the maximum percentage of space in
198 the filesystem that can be allocated to inodes. The
199 default value is 25% for filesystems under 1TB, 5%
200 for filesystems under 50TB and 1% for filesystems
201 over 50TB.
202
203 In the default inode allocation mode, inode blocks
204 are chosen such that inode numbers will not exceed
205 32 bits, which restricts the inode blocks to the
206 lower portion of the filesystem. The data block
207 allocator will avoid these low blocks to accommodate
208 the specified maxpct, so a high value may result in
209 a filesystem with nothing but inodes in a signifi‐
210 cant portion of the lower blocks of the filesystem.
211 (This restriction is not present when the filesystem
212 is mounted with the inode64 option on 64-bit plat‐
213 forms).
214
215 Setting the value to 0 means that essentially all of
216 the filesystem can become inode blocks, subject to
217 inode32 restrictions.
218
219 This value can be modified with xfs_growfs(8).
220
221 align[=value]
222 This is used to specify that inode allocation is or
223 is not aligned. The value is either 0 or 1, with 1
224 signifying that inodes are allocated aligned. If
225 the value is omitted, 1 is assumed. The default is
226 that inodes are aligned. Aligned inode access is
227 normally more efficient than unaligned access;
228 alignment must be established at the time the
229 filesystem is created, since inodes are allocated at
230 that time. This option can be used to turn off
231 inode alignment when the filesystem needs to be
232 mountable by a version of IRIX that does not have
233 the inode alignment feature (any release of IRIX
234 before 6.2, and IRIX 6.2 without XFS patches).
235
236 attr=value
237 This is used to specify the version of extended
238 attribute inline allocation policy to be used. By
239 default, this is 2, which uses an efficient algo‐
240 rithm for managing the available inline inode space
241 between attribute and extent data.
242
243 The previous version 1, which has fixed regions for
244 attribute and extent data, is kept for backwards
245 compatibility with kernels older than version
246 2.6.16.
247
248 -l log_section_options
249 These options specify the location, size, and other parameters
250 of the log section of the filesystem. The valid log_sec‐
251 tion_options are:
252
253 internal[=value]
254 This is used to specify that the log section is a
255 piece of the data section instead of being another
256 device or logical volume. The value is either 0 or
257 1, with 1 signifying that the log is internal. If
258 the value is omitted, 1 is assumed.
259
260 logdev=device
261 This is used to specify that the log section should
262 reside on the device separate from the data section.
263 The internal=1 and logdev options are mutually
264 exclusive.
265
266 size=value
267 This is used to specify the size of the log section.
268
269 If the log is contained within the data section and
270 size isn't specified, mkfs.xfs will try to select a
271 suitable log size depending on the size of the
272 filesystem. The actual logsize depends on the
273 filesystem block size and the directory block size.
274
275 Otherwise, the size suboption is only needed if the
276 log section of the filesystem should occupy less
277 space than the size of the special file. The value
278 is specified in bytes or blocks, with a b suffix
279 meaning multiplication by the filesystem block size,
280 as described above. The overriding minimum value for
281 size is 512 blocks. With some combinations of
282 filesystem block size, inode size, and directory
283 block size, the minimum log size is larger than 512
284 blocks.
285
286 version=value
287 This specifies the version of the log. The current
288 default is 2, which allows for larger log buffer
289 sizes, as well as supporting stripe-aligned log
290 writes (see the sunit and su options, below).
291
292 The previous version 1, which is limited to 32k log
293 buffers and does not support stripe-aligned writes,
294 is kept for backwards compatibility with very old
295 2.4 kernels.
296
297 sunit=value
298 This specifies the alignment to be used for log
299 writes. The value has to be specified in 512-byte
300 block units. Use the su suboption to specify the log
301 stripe unit size in bytes. Log writes will be
302 aligned on this boundary, and rounded up to this
303 boundary. This gives major improvements in perfor‐
304 mance on some configurations such as software RAID5
305 when the sunit is specified as the filesystem block
306 size. The equivalent byte value must be a multiple
307 of the filesystem block size. Version 2 logs are
308 automatically selected if the log sunit suboption is
309 specified.
310
311 The su suboption is an alternative to using sunit.
312
313 su=value
314 This is used to specify the log stripe. The value
315 has to be specified in bytes, (usually using the s
316 or b suffixes). This value must be a multiple of the
317 filesystem block size. Version 2 logs are automati‐
318 cally selected if the log su suboption is specified.
319
320 lazy-count=value
321 This changes the method of logging various persis‐
322 tent counters in the superblock. Under metadata
323 intensive workloads, these counters are updated and
324 logged frequently enough that the superblock updates
325 become a serialisation point in the filesystem. The
326 value can be either 0 or 1.
327
328 With lazy-count=1, the superblock is not modified or
329 logged on every change of the persistent counters.
330 Instead, enough information is kept in other parts
331 of the filesystem to be able to maintain the persis‐
332 tent counter values without needed to keep them in
333 the superblock. This gives significant improvements
334 in performance on some configurations. The default
335 value is 1 (on) so you must specify lazy-count=0 if
336 you want to disable this feature for older kernels
337 which don't support it.
338
339 -n naming_options
340 These options specify the version and size parameters for the
341 naming (directory) area of the filesystem. The valid nam‐
342 ing_options are:
343
344 size=value | log=value
345 The block size is specified either as a value in
346 bytes with size=, or as a base two logarithm value
347 with log=. The block size must be a power of 2 and
348 cannot be less than the filesystem block size. The
349 default size value for version 2 directories is 4096
350 bytes (4 KiB), unless the filesystem block size is
351 larger than 4096, in which case the default value is
352 the filesystem block size. For version 1 directo‐
353 ries the block size is the same as the filesystem
354 block size.
355
356 version=value
357 The naming (directory) version value can be either 2
358 or 'ci', defaulting to 2 if unspecified. With ver‐
359 sion 2 directories, the directory block size can be
360 any power of 2 size from the filesystem block size
361 up to 65536.
362
363 The version=ci option enables ASCII only case-insen‐
364 sitive filename lookup and version 2 directories.
365 Filenames are case-preserving, that is, the names
366 are stored in directories using the case they were
367 created with.
368
369 Note: Version 1 directories are not supported.
370
371 -p protofile
372 If the optional -p protofile argument is given, mkfs.xfs uses
373 protofile as a prototype file and takes its directions from that
374 file. The blocks and inodes specifiers in the protofile are
375 provided for backwards compatibility, but are otherwise unused.
376 The syntax of the protofile is defined by a number of tokens
377 separated by spaces or newlines. Note that the line numbers are
378 not part of the syntax but are meant to help you in the follow‐
379 ing discussion of the file contents.
380
381 1 /stand/diskboot
382 2 4872 110
383 3 d--777 3 1
384 4 usr d--777 3 1
385 5 sh ---755 3 1 /bin/sh
386 6 ken d--755 6 1
387 7 $
388 8 b0 b--644 3 1 0 0
389 9 c0 c--644 3 1 0 0
390 10 fifo p--644 3 1
391 11 slink l--644 3 1 /a/symbolic/link
392 12 : This is a comment line
393 13 $
394 14 $
395
396 Line 1 is a dummy string. (It was formerly the bootfilename.)
397 It is present for backward compatibility; boot blocks are not
398 used on SGI systems.
399
400 Note that some string of characters must be present as the first
401 line of the proto file to cause it to be parsed correctly; the
402 value of this string is immaterial since it is ignored.
403
404 Line 2 contains two numeric values (formerly the numbers of
405 blocks and inodes). These are also merely for backward compati‐
406 bility: two numeric values must appear at this point for the
407 proto file to be correctly parsed, but their values are immate‐
408 rial since they are ignored.
409
410 The lines 3 through 11 specify the files and directories you
411 want to include in this filesystem. Line 3 defines the root
412 directory. Other directories and files that you want in the
413 filesystem are indicated by lines 4 through 6 and lines 8
414 through 10. Line 11 contains symbolic link syntax.
415
416 Notice the dollar sign ($) syntax on line 7. This syntax directs
417 the mkfs.xfs command to terminate the branch of the filesystem
418 it is currently on and then continue from the directory speci‐
419 fied by the next line, in this case line 8. It must be the last
420 character on a line. The colon on line 12 introduces a comment;
421 all characters up until the following newline are ignored. Note
422 that this means you cannot have a file in a prototype file whose
423 name contains a colon. The $ on lines 13 and 14 end the
424 process, since no additional specifications follow.
425
426 File specifications provide the following:
427
428 * file mode
429 * user ID
430 * group ID
431 * the file's beginning contents
432
433 A 6-character string defines the mode for a file. The first
434 character of this string defines the file type. The character
435 range for this first character is -bcdpl. A file may be a regu‐
436 lar file, a block special file, a character special file, direc‐
437 tory files, named pipes (first-in, first out files), and sym‐
438 bolic links. The second character of the mode string is used to
439 specify setuserID mode, in which case it is u. If setuserID
440 mode is not specified, the second character is -. The third
441 character of the mode string is used to specify the setgroupID
442 mode, in which case it is g. If setgroupID mode is not speci‐
443 fied, the third character is -. The remaining characters of the
444 mode string are a three digit octal number. This octal number
445 defines the owner, group, and other read, write, and execute
446 permissions for the file, respectively. For more information on
447 file permissions, see the chmod(1) command.
448
449 Following the mode character string are two decimal number
450 tokens that specify the user and group IDs of the file's owner.
451
452 In a regular file, the next token specifies the pathname from
453 which the contents and size of the file are copied. In a block
454 or character special file, the next token are two decimal num‐
455 bers that specify the major and minor device numbers. When a
456 file is a symbolic link, the next token specifies the contents
457 of the link.
458
459 When the file is a directory, the mkfs.xfs command creates the
460 entries dot (.) and dot-dot (..) and then reads the list of
461 names and file specifications in a recursive manner for all of
462 the entries in the directory. A scan of the protofile is always
463 terminated with the dollar ( $ ) token.
464
465 -q Quiet option. Normally mkfs.xfs prints the parameters of the
466 filesystem to be constructed; the -q flag suppresses this.
467
468 -r realtime_section_options
469 These options specify the location, size, and other parameters
470 of the real-time section of the filesystem. The valid real‐
471 time_section_options are:
472
473 rtdev=device
474 This is used to specify the device which should con‐
475 tain the real-time section of the filesystem. The
476 suboption value is the name of a block device.
477
478 extsize=value
479 This is used to specify the size of the blocks in
480 the real-time section of the filesystem. This value
481 must be a multiple of the filesystem block size. The
482 minimum allowed size is the filesystem block size or
483 4 KiB (whichever is larger); the default size is the
484 stripe width for striped volumes or 64 KiB for non-
485 striped volumes; the maximum allowed size is 1 GiB.
486 The real-time extent size should be carefully chosen
487 to match the parameters of the physical media used.
488
489 size=value
490 This is used to specify the size of the real-time
491 section. This suboption is only needed if the real-
492 time section of the filesystem should occupy less
493 space than the size of the partition or logical vol‐
494 ume containing the section.
495
496 -s sector_size
497 This option specifies the fundamental sector size of the
498 filesystem. The sector_size is specified either as a value in
499 bytes with size=value or as a base two logarithm value with
500 log=value. The default sector_size is 512 bytes. The minimum
501 value for sector size is 512; the maximum is 32768 (32 KiB). The
502 sector_size must be a power of 2 size and cannot be made larger
503 than the filesystem block size.
504
505 -L label
506 Set the filesystem label. XFS filesystem labels can be at most
507 12 characters long; if label is longer than 12 characters,
508 mkfs.xfs will not proceed with creating the filesystem. Refer
509 to the mount(8) and xfs_admin(8) manual entries for additional
510 information.
511
512 -N Causes the file system parameters to be printed out without
513 really creating the file system.
514
515 -K Do not attempt to discard blocks at mkfs time.
516
518 xfs(5), mkfs(8), mount(8), xfs_info(8), xfs_admin(8).
519
521 With a prototype file, it is not possible to specify hard links.
522
523
524
525 mkfs.xfs(8)