1SG_FORMAT(8) SG3_UTILS SG_FORMAT(8)
2
6 sg_format - format, resize a SCSI disk or format a tape
7
9 sg_format [--cmplst={0|1}] [--count=COUNT] [--dcrt] [--dry-run]
10 [--early] [--ffmt=FFMT] [--fmtpinfo=FPI] [--format] [--help] [--ip-def]
11 [--long] [--mode=MP] [--pfu=PFU] [--pie=PIE] [--pinfo] [--poll=PT]
12 [--quick] [--resize] [--rto_req] [--security] [--six] [--size=LB_SZ]
13 [--tape=FM] [--timeout=SECS] [--verbose] [--verify] [--version]
14 [--wait] DEVICE
15
17 Not all SCSI direct access devices need to be formatted and some have
18 vendor specific formatting procedures. SCSI disks with rotating media
19 are probably the largest group that do support a 'standard' format
20 operation. They are typically factory formatted to a block size of 512
21 bytes with the largest number of blocks that the manufacturer recom‐
22 mends. The manufacturer's recommendation typically leaves aside a cer‐
23 tain number of tracks, spread across the media, for reassignment of
24 blocks to logical block addresses during the life of the disk.
25 This utility can format modern SCSI disks and potentially change their
27 block size (if permitted) and the block count (i.e. number of accessi‐
28 ble blocks on the media also known as "resizing"). Resizing a disk to
29 less than the manufacturer's recommended block count is sometimes
30 called "short stroking" (see NOTES section). Resizing the block count
31 while not changing the block size may not require a format operation.
32 The SBC-2 standard (see www.t10.org) has obsoleted the "format device"
33 mode page. Many of the low level details found in that mode page are
34 now left up to the discretion of the manufacturer. There is a Format
35 Status log page which reports on the previous successful format opera‐
36 tion(s).
37 When this utility is used without options (i.e. it is only given a
39 DEVICE argument) it prints out the existing block size and block count
40 derived from two sources. These two sources are a block descriptor in
41 the response to a MODE SENSE command and the response to a READ CAPAC‐
42 ITY command. The reason for this double check is to detect a "format
43 corrupt" state (see the NOTES section). This usage will not modify the
44 disk.
45 When this utility is used with the "--format" (or "-F") option it will
47 attempt to format the given DEVICE. In the absence of the --quick
48 option there is a 15 second pause during which time the user is invited
49 thrice (5 seconds apart) to abort sg_format. This occurs just prior the
50 SCSI FORMAT UNIT command being issued. See the NOTES section for more
51 information.
52 Protection information (PI) is optional and is made up of one or more
54 protection intervals, each made up of 8 bytes associated with a logical
55 block. When PI is active each logical block will have 1, 2, 4, 8, etc
56 protection intervals (i.e. a power of two), interleaved with (and fol‐
57 lowing) the user data to which they refer. Four protection types are
58 defined with protection type 0 being no protection intervals. See the
59 PROTECTION INFORMATION section below for more information.
60 When the --tape=FM option is given then the SCSI FORMAT MEDIUM command
62 is sent to the DEVICE. FORMAT MEDIUM is defined in the SSC documents at
63 T10 and prepares a volume for use. That may include partitioning the
64 medium. See the section below on TAPE for more information.
65
67 Arguments to long options are mandatory for short options as well. The
68 options are arranged in alphabetical order based on the long option
69 name.
70 -C, --cmplst={0|1}
72 sets the CMPLST ("complete list") bit in the FORMAT UNIT cdb to
73 0 or 1. If the value is 0 then the existing GLIST (grown list)
74 is taken into account. If the value is 1 then the existing
75 GLIST is ignored. CMPLST defaults to 1 apart from when the
76 --ffmt=FFMT option's value is non-zero in which case CMPLST
77 defaults to 0. See the LISTS section below. In most cases this
78 bit should be left at its default value.
79 -c, --count=COUNT
81 where COUNT is the number of blocks to be formatted or media to
82 be resized to. Can be used with either --format or --resize.
83 With --format this option need not be given in which case it is
84 assumed to be zero.
85 With --format the interpretation of COUNT is:
86 (COUNT > 0) : only format the first COUNT blocks and READ
87 CAPACITY will report COUNT blocks after format
88 (COUNT = 0) and block size unchanged : use existing block
89 count
90 (COUNT = 0) and block size changed : recommended maximum block
91 count for new block size
92 (COUNT = -1) : use recommended maximum block count
93 (COUNT < -1) : illegal
94 With --resize this option must be given and COUNT has this
95 interpretation:
96 (COUNT > 0) : after resize READ CAPACITY will report COUNT
97 blocks
98 (COUNT = 0) : after resize READ CAPACITY will report 0 blocks
99 (COUNT = -1) : after resize READ CAPACITY will report its max‐
100 imum number of blocks
101 (COUNT < -1) : illegal
102 In both cases if the given COUNT exceeds the maximum number of
103 blocks (for the block size) then the disk reports an error. See
104 NOTES section below.
105 -D, --dcrt
107 this option sets the DCRT bit in the FORMAT UNIT command's
108 parameter list header. It will "disable certification". Certifi‐
109 cation verifies that blocks are usable during the format
110 process. Using this option may speed the format. The default
111 action of this utility (i.e. when this option is not given) is
112 to clear the DCRT bit thereby requesting "media certification".
113 When the DCRT bit is set, the FOV bit must also be set hence
114 sg_format does that.
115 -d, --dry-run
117 this option will parse the command line, do all the preparation
118 but bypass the actual FORMAT UNIT or FORMAT MEDIUM commands.
119 Also if the options would cause the logical block size to
120 change, then the MODE SELECT command that would do that is also
121 bypassed when the dry run option is given.
122 -e, --early
124 during a format operation, The default action of this utility is
125 to poll the disk every 60 seconds (or every 10 seconds if FFMT
126 is non-zero) to determine the progress of the format operation
127 until it is finished. When this option is given this utility
128 will exit "early", that is as soon as the format operation has
129 commenced. Then the user can monitor the progress of the ongoing
130 format operation with other utilities (e.g. sg_turs(8) or
131 sg_requests(8)). This option and --wait are mutually exclusive.
132 -t, --ffmt=FFMT
134 FFMT (fast format) is placed in a field of the same name in the
135 FORMAT UNIT cdb. The field was introduced in SBC-4 revision 10.
136 The default value is 0 which implies the former action which is
137 typically to overwrite all blocks on the DEVICE. That can take a
138 long time (e.g. with hard disks over 10 TB in size that can be
139 days). With FFMT set that time may be reduced to minutes. So it
140 is worth trying if it is available.
141 FFMT has values 1 and 2 for fast format with 3 being reserved
142 currently. These two values include this description: "The
143 device server initializes the medium ... without overwriting the
144 medium (i.e. resources for managing medium access are initial‐
145 ized and the medium is not written)". The difference between 1
146 and 2 concerns read operations on LBAs to which no data has been
147 written to, after the fast format. When FFMT is 1 the read oper‐
148 ation should return "unspecified logical block data" and com‐
149 plete without error. When FFMT is 2 the read operation may yield
150 check condition status with a sense key set to hardware error,
151 medium error or command aborted. See SBC-4 revision 15 section
152 4.35 for more details.
153 -f, --fmtpinfo=FPI
155 sets the FMTPINFO field in the FORMAT UNIT cdb to a value
156 between 0 and 3. The default value is 0. The FMTPINFO field
157 from SBC-3 revision 16 is a 2 bit field (bits 7 and 6 of byte 1
158 in the cdb). Prior to that revision it was a single bit field
159 (bit 7 of byte 1 in the cdb) and there was an accompanying bit
160 called RTO_REQ (bit 6 of byte 1 in the cdb). The deprecated
161 options "--pinfo" and "--rto-req" represent the older usage.
162 This option should be used in their place. See the PROTECTION
163 INFORMATION section below for more information.
164 -F, --format
166 issue a SCSI FORMAT UNIT command. This will destroy all the
167 data held on the media. This option is required to change the
168 block size of a disk. In the absence of the --quick option, the
169 user is given a 15 second count down to ponder the wisdom of
170 doing this, during which time control-C (amongst other Unix com‐
171 mands) can be used to kill this process before it does any dam‐
172 age.
173 When used three times (or more) the preliminary MODE SENSE and
174 SELECT commands are bypassed, leaving only the initial INQUIRY
175 and FORMAT UNIT commands. This is for emergency use (e.g. when
176 the MODE SENSE/SELECT commands are not working) and cannot
177 change the logical block size.
178 See NOTES section for implementation details and EXAMPLES sec‐
179 tion for typical use.
180 -h, --help
182 print out the usage information then exit.
183 -I, --ip-def
185 sets the default Initialization Pattern. Some disks (SSDs) use
186 this to flag that a format should fully provision (i.e. asso‐
187 ciate a physical block with every logical block). The same disks
188 (SSDs) might thin provision if this option is not given. If this
189 option is given then the --security option cannot be given. Also
190 accepts --ip_def for this option.
191 -l, --long
193 the default action of this utility is to assume 32 bit logical
194 block addresses. With 512 byte block size this permits more than
195 2 terabytes (almost 2 ** 41 bytes) on a single disk. This option
196 selects commands and parameters that allow for 64 bit logical
197 block addresses. Specifically this option sets the "longlba"
198 flag in the MODE SENSE (10) command and uses READ CAPACITY (16)
199 rather than READ CAPACITY (10). If this option is not given and
200 READ CAPACITY (10) or MODE SELECT detects a disk the needs more
201 than 32 bits to represent its logical blocks then it is set
202 internally. This option does not set the LONGLIST bit in the
203 FORMAT UNIT command. The LONGLIST bit is set as required depend‐
204 ing other parameters (e.g. when '--pie=PIE' is greater than
205 zero).
206 -M, --mode=MP
208 MP is a mode page number (0 to 62 inclusive) that will be used
209 for reading and perhaps changing the device logical block size.
210 The default is 1 which is the Read-Write Error Recovery mode
211 page.
212 Preferably the chosen (or default) mode page should be saveable
213 (i.e. accept the SP bit set in the MODE SELECT command used
214 when the logical block size is being changed). Recent version of
215 this utility will retry a MODE SELECT if the SP=1 variant fails
216 with a sense key of ILLEGAL REQUEST. That retry will use the
217 same MODE SELECT command but with SP=0 .
218 -P, --pfu=PFU
220 sets the "Protection Field Usage" field in the parameter block
221 associated with a FORMAT UNIT command to PFU. The default value
222 is 0, the only other defined value currently is 1. See the PRO‐
223 TECTION INFORMATION section below for more information.
224 -q, --pie=PIE
226 sets the "Protection Interval Exponent" field in the parameter
227 block associated with a FORMAT UNIT command to PIE. The default
228 value is 0. PIE can only be non-zero with protection types 2
229 and 3. The value of 0 is typical for 512 byte blocks; with 4096
230 byte blocks a value of 3 may be appropriate (i.e. 8 protection
231 intervals interleaved with 4096 bytes of user data). A device
232 may not support any non-zero values. This field first appeared
233 in SBC-3 revision 18.
234 -p, --pinfo
236 this option is deprecated, use the --fmtpinfo=FPI option
237 instead. If used, then it sets bit 7 of byte 1 in the FORMAT
238 UNIT cdb and that is equivalent to setting --fmtpinfo=2. [So if
239 --pinfo is used (plus --fmtpinfo=FPI and --pfu=PFU are not given
240 or their arguments are 0) then protection type 1 is selected.]
241 -x, --poll=PT
243 where PT is the type of poll used. If PT is 0 then a TEST UNIT
244 READY command is used, otherwise a REQUEST SENSE command is
245 used. The default is currently 0 but this will change to 1 in
246 the near future. See the NOTES sections below.
247 -Q, --quick
249 the default action (i.e. when the option is not given) is to
250 give the user 15 seconds to reconsider doing a format operation
251 on the DEVICE. When this option is given that step (i.e. the 15
252 second warning period) is skipped.
253 -r, --resize
255 rather than format the disk, it can be resized. This means
256 changing the number of blocks on the device reported by the READ
257 CAPACITY command. This option should be used with the
258 --count=COUNT option. The contents of all logical blocks on the
259 media remain unchanged when this option is used. This means that
260 any resize operation can be reversed. This option cannot be used
261 together with either --format or a --size=LB_SZ whose argument
262 is different to the existing block size.
263 -R, --rto_req
265 The option is deprecated, use the --fmtpinfo=FPI option instead.
266 If used, then it sets bit 6 of byte 1 in the FORMAT UNIT cdb.
267 -S, --security
269 sets the "Security Initialization" (SI) bit in the FORMAT UNIT
270 command's initialization pattern descriptor within the parameter
271 list. According to SBC-3 the default initialization pattern
272 "shall be written using a security erasure write technique". See
273 the NOTES section on the SCSI SANITIZE command. If this option
274 is given then the --ip_def option cannot be given.
275 -6, --six
277 Use 6 byte variants of MODE SENSE and MODE SELECT. The default
278 action is to use the 10 byte variants. Some MO drives need this
279 option set when doing a format.
280 -s, --size=LB_SZ
282 where LB_SZ is the logical block size (i.e. number of user bytes
283 in each block) to format the device to. The default value is
284 whatever is currently reported by the block descriptor in a MODE
285 SENSE command. If the block size given by this option is differ‐
286 ent from the current value then a MODE SELECT command is used to
287 change it prior to the FORMAT UNIT command being started (as
288 recommended in the SBC standards). Some SCSI disks have 512 byte
289 logical blocks by default and allow an alternate logical block
290 size of 4096 bytes. If the given size in unacceptable to the
291 disk, most likely an "Invalid field in parameter list" message
292 will appear in sense data (requires the use of '-v' to decode
293 sense data).
294 Note that formatting a disk to add or remove protection informa‐
295 tion is not regarded as a change to its logical block size so
296 this option should not be used.
297 -T, --tape=FM
299 will send a FORMAT MEDIUM command to the DEVICE with its FORMAT
300 field set to FM. This option is used to prepare a tape (i.e. the
301 "medium") in a tape drive for use. Values for FM include 0 to do
302 the "default" format; 1 to partition a volume and 2 to do a
303 default format then partition.
304 -m, --timeout=SECS
306 where SECS is the FORMAT UNIT or FORMAT MEDIUM command timeout
307 in seconds. SECS will only be used if it exceeds the internal
308 timeout which is 20 seconds if the IMMED bit is set and 72000
309 seconds (20 hours) or higher if the IMMED bit is not set. If the
310 disk size exceeds 4 TB then the timeout value is increased to
311 144000 seconds (40 hours). And if it is greater than 8 TB then
312 the timeout value is increased to 288000 seconds (80 hours). If
313 the timeout is exceeded then the operating system will typically
314 abort the command. Aborting a command may escalate to a LUN
315 reset (or worse). A timeout may also leave the disk or tape for‐
316 mat operation incomplete. And that may result in the disk or
317 tape being in a "format corrupt" state requiring another format
318 to remedy the situation. So for various reasons timeouts are
319 best avoided.
320 -v, --verbose
322 increase the level of verbosity, (i.e. debug output). "-vvv"
323 gives a lot more debug output.
324 -y, --verify
326 set the VERIFY bit in the FORMAT MEDIUM cdb. The default is that
327 the VERIFY bit is clear. This option is only appropriate for
328 tapes.
329 -V, --version
331 print the version string and then exit.
332 -w, --wait
334 the default format action is to set the "IMMED" bit in the FOR‐
335 MAT UNIT command's (short) parameter header. If this option
336 (i.e. --wait) is given then the "IMMED" bit is not set. If
337 --wait is given then the FORMAT UNIT or FORMAT MEDIUM command
338 waits until the format operation completes before returning its
339 response. This can be many hours on large disks. See the --time‐
340 out=SECS option.
341
343 The SBC-3 draft (revision 20) defines PLIST, CLIST, DLIST and GLIST in
344 section 4.10 on "Medium defects". Briefly, the PLIST is the "primary"
345 list of manufacturer detected defects, the CLIST ("certification" list)
346 contains those detected during the format operation, the DLIST is a
347 list of defects that can be given to the format operation. The GLIST is
348 the grown list which starts in the format process as CLIST+DLIST and
349 can "grow" later due to automatic reallocation (see the ARRE and AWRE
350 bits in the Read-Write Error Recovery mode page (see sdparm(8))) and
351 use of the SCSI REASSIGN BLOCKS command (see sg_reassign(8)).
352 By the SBC-3 standard (following draft revision 36) the CLIST and DLIST
354 had been removed, leaving PLIST and GLIST. Only PLIST and GLIST are
355 found in the SBC-4 drafts.
356 The CMPLST bit (controlled by the --cmplst=0|1 option) determines
358 whether the existing GLIST, when the format operation is invoked, is
359 taken into account. The sg_format utility sets the FOV bit to zero
360 which causes DPRY=0, so the PLIST is taken into account, and DCRT=0, so
361 the CLIST is generated and used during the format process.
362 The sg_format utility does not permit a user to provide a defect list
364 (i.e. DLIST).
365
367 Protection Information (PI) is additional information held with logical
368 blocks so that an application and/or host bus adapter can check the
369 correctness of those logical blocks. PI is placed in one or more pro‐
370 tection intervals interleaved in each logical block. Each protection
371 interval follows the user data to which it refers. A protection inter‐
372 val contains 8 bytes made up of a 2 byte "logical block guard" (CRC), a
373 2 byte "logical block application guard", and a 4 byte "logical block
374 reference tag". Devices with 512 byte logical block size typically have
375 one protection interval appended, making its logical block data 520
376 bytes long. Devices with 4096 byte logical block size often have 8 pro‐
377 tection intervals spread across its logical block data for a total size
378 of 4160 bytes. Note that for all other purposes the logical block size
379 is considered to be 512 and 4096 bytes respectively.
380 The SBC-3 standard have added several "protection types" to the PI
382 introduced in the SBC-2 standard. SBC-3 defines 4 protection types
383 (types 0 to 3) with protection type 0 meaning no PI is maintained.
384 While a device may support one or more protection types, it can only be
385 formatted with 1 of the 4. To change a device's protection type, it
386 must be re-formatted. For more information see the Protection Informa‐
387 tion in section 4.22 of draft SBC-4 revision 15.
388 A device that supports PI information (i.e. supports one or more pro‐
390 tection types 1, 2 and 3) sets the "PROTECT" bit in its standard
391 INQUIRY response. It also sets the SPT field in the EXTENDED INQUIRY
392 VPD page response to indicate which protection types it supports. Given
393 PROTECT=1 then SPT=0 implies the device supports PI type 1 only, SPT=1
394 implies the device supports PI types 1 and 2, and various other
395 non-obvious mappings up to SPT=7 which implies protection types 1, 2
396 and 3 are supported. The current protection type of a disk can be found
397 in the "P_TYPE" and "PROT_EN" fields in the response of a READ CAPACITY
398 (16) command (e.g. with the 'sg_readcap --long' utility).
399 Given that a device supports a particular protection type, a user can
401 then choose to format that disk with that protection type by setting
402 the "FMTPINFO" and "Protection Field Usage" fields in the FORMAT UNIT
403 command. Those fields correspond to the --fmtpinfo=FPI and the
404 --pfu=PFU options in this utility. The list below shows the four pro‐
405 tection types followed by the options of this utility needed to select
406 them:
407 0 : --fmtpinfo=0 --pfu=0
408 1 : --fmtpinfo=2 --pfu=0
409 2 : --fmtpinfo=3 --pfu=0
410 3 : --fmtpinfo=3 --pfu=1
411 The default value of FPI (in --fmtpinfo=FPI) is 0 and the default value
412 of PFU (in --pfu=PFU) is 0. So if neither --fmtpinfo=FPI nor --pfu=PFU
413 are given then protection type 0 (i.e. no protection information) is
414 chosen.
415
417 After a format that changes the logical block size or the number of
418 logical blocks on a disk, the operating system may need to be told to
419 re-initialize its setting for that disk. In Linux that can be done
420 with:
421 echo 1 > /sys/block/sd{letter(s)}/device/rescan
422 where "letter(s)" will be between 'a' and 'zzz'. The lsscsi utility in
423 Linux can be used to check the various namings of a disk.
424 The SBC-2 standard states that the REQUEST SENSE command should be used
426 for obtaining progress indication when the format command is underway.
427 However, tests on a selection of disks shows that TEST UNIT READY com‐
428 mands yield progress indications (but not REQUEST SENSE commands). So
429 the current version of this utility defaults to using TEST UNIT READY
430 commands to poll the disk to find out the progress of the format. The
431 --poll=PT option has been added to control this.
432 When the --format option is given without the --wait option then the
434 SCSI FORMAT UNIT command is issued with the IMMED bit set which causes
435 the SCSI command to return after it has started the format operation.
436 The --early option will cause sg_format to exit at that point. Other‐
437 wise the DEVICE is polled every 60 seconds or every 10 seconds if FFMT
438 is non-zero. The poll is with TEST UNIT READY or REQUEST SENSE commands
439 until one reports an "all clear" (i.e. the format operation has com‐
440 pleted). Normally these polling commands will result in a progress
441 indicator (expressed as a percentage) being output to the screen. If
442 the user gets bored watching the progress report then sg_format process
443 can be terminated (e.g. with control-C) without affecting the format
444 operation which continues. However a target or device reset (or a power
445 cycle) will probably cause the device to become "format corrupt".
446 When the --format (or --tape) and --wait options are both given then
448 this utility may take a long time to return. In this case care should
449 be taken not to send any other SCSI commands to the disk as it may not
450 respond leaving those commands queued behind the active format command.
451 This may cause a timeout in the OS driver (in a lot shorter period than
452 20 hours applicable to some format operations). This may result in the
453 OS resetting the disk leaving the format operation incomplete. This may
454 leave the disk in a "format corrupt" state requiring another format to
455 remedy the situation. Modern SCSI devices should yield a "not ready"
456 sense key with an additional sense indicating a format is in progress.
457 With older devices the user should take precautions that nothing
458 attempts to access a device while it is being formatted.
459 When the block size (i.e. the number of bytes in each block) is changed
461 on a disk two SCSI commands must be sent: a MODE SELECT to change the
462 block size followed by a FORMAT command. If the MODE SELECT command
463 succeeds and the FORMAT fails then the disk may be in a state that the
464 standard calls "format corrupt". A block descriptor in a subsequent
465 MODE SENSE will report the requested new block size while a READ CAPAC‐
466 ITY command will report the existing (i.e. previous) block size. Alter‐
467 natively the READ CAPACITY command may fail, reporting the device is
468 not ready, potentially requiring a format. The solution to this situa‐
469 tion is to do a format again (and this time the new block size does not
470 have to be given) or change the block size back to the original size.
471 The SBC-2 standard states that the block count can be set back to the
473 manufacturer's maximum recommended value in a format or resize opera‐
474 tion. This can be done by placing an address of 0xffffffff (or the 64
475 bit equivalent) in the appropriate block descriptor field to a MODE
476 SELECT command. In signed (two's complement) arithmetic that value cor‐
477 responds to '-1'. So a --count=-1 causes the block count to be set back
478 to the manufacturer's maximum recommended value. To see exactly which
479 SCSI commands are being executed and parameters passed add the "-vvv"
480 option to the sg_format command line.
481 The FMTDATA field shown in the FORMAT UNIT cdb does not have a corre‐
483 sponding option in this utility. When set in the cdb it indicates an
484 additional parameter list will be sent to the DEVICE along with the
485 cdb. It is set as required, basically when any field in the parameter
486 list header is set.
487 Short stroking is a technique to trade off capacity for performance on
489 hard disks. "Hard" disk is often used to mean a storage device with
490 spinning platters which contain the user data. Solid State Disk (SSD)
491 is the newer form of storage device that contains no moving parts. Hard
492 disk performance is usually highest on the outer tracks (usually the
493 lower logical block addresses) so by resizing or reformatting a disk to
494 a smaller capacity, average performance will usually be increased.
495 Other utilities may be useful in finding information associated with
497 formatting. These include sg_inq(8) to fetch standard INQUIRY informa‐
498 tion (e.g. the PROTECT bit) and to fetch the EXTENDED INQUIRY VPD page
499 (e.g. RTO and GRD_CHK bits). The sdparm(8) utility can be used to
500 access and potentially change the now obsolete format mode page.
501 scsiformat is another utility available for formatting SCSI disks with
503 Linux. It dates from 1997 (most recent update) and may be useful for
504 disks whose firmware is of that vintage.
505 The COUNT numeric argument may include a multiplicative suffix or be
507 given in hexadecimal. See the "NUMERIC ARGUMENTS" section in the
508 sg3_utils(8) man page.
509 The SCSI SANITIZE command was introduced in SBC-3 revision 27. It is
511 closely related to the ATA sanitize disk feature set and can be used to
512 remove all existing data from a disk. Sanitize is more likely to be
513 implemented on modern disks (including SSDs) than FORMAT UNIT's secu‐
514 rity initialization feature (see the --security option) and in some
515 cases much faster.
516 SSDs that support thin provisioning will typically unmap all logical
518 blocks during a format. The reason is to improve the SSD's endurance.
519 Also thin provisioned formats typically complete faster than fully pro‐
520 visioned ones on the same disk (see the --ip_def option). In either
521 case format operations on SSDs tend to be a lot faster than they are on
522 hard disks with spinning media.
523
525 Tape system use a variant of the FORMAT UNIT command used on disks.
526 Tape systems use the FORMAT MEDIUM command which is simpler with only
527 three fields in the cdb typically used. Apart from sharing the same
528 opcode the cdbs of FORMAT UNIT and FORMAT MEDIUM are quite different.
529 FORMAT MEDIUM's fields are VERIFY, IMMED and FORMAT (with TRANSFER
530 LENGTH always set to 0). The VERIFY bit field is set with the --verify
531 option. The IMMED bit is manipulated by the --wait option in the same
532 way it is for disks; one difference is that if the --poll=PT option is
533 not given then it defaults to PT of 1 which means the poll is done with
534 REQUEST SENSE commands.
535 The argument given to the --tape=FM option is used to set the FORMAT
537 field. FM can take values from "-1" to "15" where "-1" (the default)
538 means don't do a tape format; value "8" to "15" are for vendor specific
539 formats. The --early option may also be used to set the IMMED bit and
540 then exit this utility (rather than poll periodically until it is fin‐
541 ished). In this case the tape drive will still be busy doing the format
542 for some time but, according to T10, should still respond in full to
543 the INQUIRY and REPORT LUNS commands. Other commands (including REQUEST
544 SENSE) should yield a "not ready" sense key with an additional sense
545 code of "Logical unit not ready, format in progress". Additionally
546 REQUEST SENSE should contain a progress indication in its sense data.
547 When FM is 1 or 2 then the settings in the Medium partition mode page
549 control the partitioning. That mode page can be viewed and modified
550 with the sdparm utility.
551 Prior to invoking this utility the tape may need to be positioned to
553 the beginning of partition 0. In Linux that can typically be done with
554 the mt utility (e.g. 'mt -f /dev/st0 rewind').
555
557 These examples use Linux device names. For suitable device names in
558 other supported Operating Systems see the sg3_utils(8) man page.
559 In the first example below simply find out the existing block count and
561 size derived from two sources: a block descriptor in a MODE SELECT com‐
562 mand response and from the response of a READ CAPACITY commands. No
563 changes are made:
564 # sg_format /dev/sdm
566 Now a simple format, leaving the block count and size as they were pre‐
568 viously. The FORMAT UNIT command is executed in IMMED mode and the
569 device is polled every 60 seconds to print out a progress indication:
570 # sg_format --format /dev/sdm
572 Now the same format, but waiting (passively) until the format operation
574 is complete:
575 # sg_format --format --wait /dev/sdm
577 Next is a format in which the block size is changed to 520 bytes and
579 the block count is set to the manufacturer's maximum value (for that
580 block size). Note, not all disks support changing the block size:
581 # sg_format --format --size=520 /dev/sdm
583 Now a resize operation so that only the first 0x10000 (65536) blocks on
585 a disk are accessible. The remaining blocks remain unaltered.
586 # sg_format --resize --count=0x10000 /dev/sdm
588 Now resize the disk back to its normal (maximum) block count:
590 # sg_format --resize --count=-1 /dev/sdm
592 One reason to format a SCSI disk is to add protection information.
594 First check which protection types are supported by a disk (by checking
595 the SPT field in the Extended inquiry VPD page together with the Pro‐
596 tect bit in the standard inquiry response):
597 # sg_vpd -p ei -l /dev/sdb
599 extended INQUIRY data VPD page:
600 ACTIVATE_MICROCODE=0
601 SPT=1 [protection types 1 and 2 supported]
602 ....
603 Format with type 1 protection:
605 # sg_format --format --fmtpinfo=2 /dev/sdm
607 After a successful format with type 1 protection, READ CAPACITY(16)
609 should show something like this:
610 # sg_readcap -l /dev/sdm
612 Read Capacity results:
613 Protection: prot_en=1, p_type=0, p_i_exponent=0 [type 1 protec‐
614 tion]
615 Logical block provisioning: lbpme=0, lbprz=0
616 ....
617 To format with type 3 protection:
619 # sg_format --format --fmtpinfo=3 --pfu=1 /dev/sdm
621 For the disk shown above this will probably fail because the Extended
623 inquiry VPD page showed only types 1 and 2 protection are supported.
624 Here are examples of using fast format (FFMT field in FORMAT UNIT cdb)
626 to quickly switch between 512 and 4096 byte logical block size. Assume
627 disk starts with 4096 byte logical block size and all important data
628 has been backed up.
629 # sg_format --format --ffmt=1 --size=512 /dev/sdd
631 Now /dev/sdd should have 512 byte logical block size. And to switch it
633 back:
634 # sg_format --format --ffmt=1 --size=4096 /dev/sdd
636
638 The exit status of sg_format is 0 when it is successful. Otherwise see
639 the sg3_utils(8) man page. Unless the --wait option is given, the exit
640 status may not reflect the success of otherwise of the format. Using
641 sg_turs(8) and sg_readcap(8) after the format operation may be wise.
642
644 Written by Grant Grundler, James Bottomley and Douglas Gilbert.
645
647 Report bugs to <dgilbert at interlog dot com>.
648
650 Copyright © 2005-2018 Grant Grundler, James Bottomley and Douglas
651 Gilbert
652 This software is distributed under the GPL version 2. There is NO war‐
653 ranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PUR‐
654 POSE.
655
657 sg_turs(8), sg_requests(8), sg_inq(8), sg_modes(8), sg_vpd(8), sg_reas‐
658 sign(8), sg_readcap(8), sg3_utils(8), sg_sanitize(8) [all in
659 sg3_utils], lsscsi(8), mt(mt-st), sdparm(8), scsiformat (old),
660 hdparm(8)
661sg3_utils-1.43 September 2018 SG_FORMAT(8)