scsiinfo - query information from a scsi device

Authors

       Eric Youngdale.
       Michael Weller <eowmob@exp-math.uni-essen.de>, Versions 1.5 & 1.7

scsiinfo 1.7                                     23 August 1997                                      scsiinfo(8)

Bugs

       Restrictions of the SCSI_IOCTL_SEND_COMMANDioctl(2) call make it impossible to send or receive more than
       4096 bytes of arguments. This could be avoided by using the proper generic scsi device /dev/sg*  instead,
       at  least where the kernel is compiled to support it. Most of the time this is not needed though and thus
       I'm myself to lazy to do it. It will basically just truncate the vendor specified primary  defect  lists.
       Thus I'm too lazy to fix it.

Description

scsiinfo  queries information from an scsi target. This means generally using the INQUIRY scsi command or
       reading out SCSI-II mode pages (the number of the mode pages and corresponding sections  of  the  SCSI-II
       sections  is  given  below).  It  allows  also to modify some of these settings on the scsi device (if it
       supports it).

       Except for the -v and -l options you must specify exactly one scsi device to work on. You may specify any
       linux scsi device disk, tape, cdrom, generic scsi.

       Some scsi devices (typically non removable disks) will allow to store  your  modifications  in  some  non
       volatile  memory. Some of these settings (for example those dealing with the layout of logical blocks and
       sectors set aside as replacements for erroneous blocks) might render the disk unusable until a low  level
       format.

Files

/dev/sd*/dev/sg*/dev/scd*/dev/st*/dev/nst*/dev/rmt*/dev/nrmt*

Name

       scsiinfo - query information from a scsi device

Options

InformationavailablefrommostSCSIdevices(includesSCSI-I)-i display all information from the INQUIRY scsi command.

-s displays the unit serial number using the INQUIRY scsi command.

-d display factory and grown defect lists (typically for disks only).

It is currently only possible to return defect information up to 4096 bytes. Longer defect lists
are truncated. See the BUGS section.

-farg specify the format in which to return the defect information. The target may decide to fail
reporting defect information in unsupported formats or decide to return data in a different
format. scsiinfo supports all SCSI-II specified defect formats:

-Flogical
logical blocks. Use of this format is discouraged as the assignment of logical blocks
varies according to format parameters and status of the defect list, hence is no unique
specification of defects.

-Fphysical
physical blocks. Return defect as cylinder, head, physical sector triples.

-Findex
defect bytes from index. Return defect as cylinder, head, byte offset from index. The
SCSI-II standard is not very clear on this to me. It is unclear to me if there is a single
bad byte, this offset away from the index hole on the disk (this is only figuratively,
there won't be a hole as used to be on 5 1/4" floppy disks), or if all bytes from the index
to this position are considered to be bad.

SCSI-IImodepages-C displays information from Control Mode Page. (Page 0Ah, section 7.3.3.1)

-D displays information from Disconnect-Reconnect Page. (Page 02h, section 7.3.3.2)

-p displays information from Peripheral Device Page. (Page 09h, section 7.3.3.3)

-c displays information from Caching Page. (Page 08h, section 8.3.3.1)

-f displays information from Format Device Page. (Page 03h, section 8.3.3.3)

-n displays information from Notch and Partition Page. (Page 0Ch, section 8.3.3.5)

A huge scsi disk might be divided into several notches. These are regions of logical blocks or
cylinders on the disk. Each such notch might have different values for the other mode pages.

Typically a modern disk will have several notches and have more sectors per track on the inner
tracks/notches on the disk and more sectors per track on the outer (longer) tracks for optimal
capacity. Also different amounts of reserved backup sectors may be available in the notches
depending on their capacity.

-e displays information from Error Recovery page. (Page 01h, section 8.3.3.6)

-g displays information from Rigid Disk Drive Geometry Page. (Page 04h, section 8.3.3.7)

-V displays information from Verify Error Recovery Page. (Page 07h, section 8.3.3.8)

Selectmodepageset
By default the current settings are queried from the devices. You can however specify one of these:

-M displays manufacturer defaults instead of current values.

-S displays defaults saved in NVRAM instead of current values.

-m displays modifiable fields instead of current values (All bits set in modifiable fields).

Miscellaneous-v Show scsiinfo version.

-vv Dump sense buffer in case of error.

-a All of the above (expect listing defects).

-l List scsi devices known to the system.

-L List mode pages pages supported by this scsiinfo version and target (notched pages and active
notch are also returned).

-X displays output suitable for the X-based interface. Instead of nice explanations, just the bare
values are written to stdout.

-R Replace parameters. Use with -X and specify the values to set on the command line in the order and
format as -X uses to report them. (Expert use only, definitely use the Tcl/Tk interface scsi-config(8)tomodifysettings.)

Use this in conjunction with -S to modify the NVRAM settings.

-X and -R can be used only with one of the display page options.

-m and -M cannot be used with -R.

You may use -M, -S with -L though it will make no difference. As a special goodie when using -LXR then a
/bin/sh script is written to stdout that will restore the current settings of the target when executed.
You can use one of -M, -S with -LXR to save the corresponding values.