DragonFly On-Line Manual Pages
SCGCHECK(1) Schily's USER COMMANDS SCGCHECK(1)
scgcheck - check and validate the ABI of libscg
scgcheck [ options ]
Scgcheck is used to check and verify the Application Binary Interface
Most users do not need to care about device naming at all, as in -auto
mode, scgcheck implements auto target support and automagically finds a
test drive in case that exactly one CD-ROM type drive is available in
Print version information and exit.
-auto Instead of asking to confirm each test before runing it,
scgcheck tries to do a fully automated test.
Set the SCSI target for the device, see notes above. A typical
target device specification is dev=1,6,0 . If a filename must
be provided together with the numerical target specification,
the filename is implementation specific. The correct filename
in this case can be found in the system specific manuals of the
target operating system. On a FreeBSD system without CAM
support, you need to use the control device (e.g.
/dev/rcd0.ctl). A correct device specification in this case may
be dev=/dev/rcd0.ctl:@ .
General SCSI addressing
The target device to the dev= option refers to
scsibus/target/lun of the device. Communication on SunOS is done
with the SCSI general driver scg. Other operating systems are
using a library simulation of this driver. Possible syntax is:
dev= scsibus,target,lun or dev= target,lun. In the latter case,
the device has to be connected to the default SCSI bus of the
machine. Scsibus, target and lun are integer numbers. Some
operating systems or SCSI transport implementations may require
to specify a filename in addition. In this case the correct
syntax for the device is: dev= devicename:scsibus,target,lun or
dev= devicename:target,lun. If the name of the device node that
has been specified on such a system refers to exactly one SCSI
device, a shorthand in the form dev= devicename:@ or dev=
devicename:@,lun may be used instead of dev=
Remote SCSI addressing
To access remote SCSI devices, you need to prepend the SCSI
device name by a remote device indicator. The remote device
indicator is either REMOTE:user@host: or REMOTE:host: A valid
remote SCSI device name may be: REMOTE:user@host: to allow
remote SCSI bus scanning or REMOTE:user@host:1,0,0 to access the
SCSI device at host connected to SCSI bus # 1,target 0, lun 0.
In order to allow remote access to a specific host, the rscsi(1)
program needs to be present and configured on the host.
Alternate SCSI transports
ATAPI drives are just SCSI drives that inherently use the ATA
packet interface as SCSI command transport layer build into the
IDE (ATA) transport. You may need to specify an alternate
transport layer on the command line if your OS does not
implement a fully integrated kernel driver subsystem that allows
to access any drive using SCSI commands via a single unique user
To access SCSI devices via alternate transport layers, you need
to prepend the SCSI device name by a transport layer indicator.
The transport layer indicator may be something like USCSI: or
ATAPI:. To get a list of supported transport layers for your
platform, use dev= HELP:
To make scgcheck portable to all UNIX platforms, the syntax dev=
devicename:scsibus,target,lun is preferred as it hides OS
specific knowledge about device names from the user. A specific
OS may not necessarily support a way to specify a real device
file name nor a way to specify scsibus,target,lun.
Scsibus 0 is the default SCSI bus on the machine. Watch the boot
messages for more information or look into /var/run/dmesg.boot
for more information about the SCSI configuration of your
machine. If you have problems to figure out what values for
scsibus,target,lun should be used, try the -scanbus option of
scgcheck described below.
If no dev= option is present, or if it only contains a transport
specifyer but no address notation, scgcheck tries to scan the
SCSI address space for CD-ROM drives. If exactly one is found,
this is used by default.
Set the default SCSI command timeout value to # seconds. The
default SCSI command timeout is the minimum timeout used for
sending SCSI commands. If a SCSI command fails due to a
timeout, you may try to raise the default SCSI command timeout
above the timeout value of the failed command. If the command
runs correctly with a raised command timeout, please report the
better timeout value and the corresponding command to the author
of the program. If no timeout option is present, a default
timeout of 40 seconds is used.
Set the misc debug value to # (with debug=#) or increment the
misc debug level by one (with -d). If you specify -dd, this
equals to debug=2. This may help to find problems while opening
a driver for libscg. as well as with sector sizes and sector
types. Using -debug slows down the process and may be the
reason for a buffer underrun.
Tell the scg-driver to modify the kernel debug value while SCSI
commands are running.
Do not print out a status report for failed SCSI commands.
-v Increment the level of general verbosity by one. This is used
e.g. to display the progress of the process.
-V Increment the verbose level with respect of SCSI command
transport by one. This helps to debug problems during the
process, that occur in the CD-Recorder. If you get
incomprehensible error messages you should use this flag to get
more detailed output. -VV will show data buffer content in
addition. Using -V or -VV slows down the process.
f=file Specify the log file to be used instead of check.log.
cdrecord(1), readcd(1), mkisofs(8).
When using scgcheck with the broken Linux SCSI generic driver. You
should note that scgcheck uses a hack, that tries to emulate the
functionality of the scg driver. Unfortunately, the sg driver on Linux
has several severe bugs:
o It cannot see if a SCSI command could not be sent at all.
o It cannot get the SCSI status byte. Scgcheck for that reason
cannot report failing SCSI commands in some situations.
o It cannot get real DMA count of transfer. Scgcheck cannot tell
you if there is an DMA residual count.
o It cannot get number of bytes valid in auto sense data.
Scgcheck cannot tell you if device transfers no sense data at
o It fetches to few data in auto request sense (CCS/SCSI-2/SCSI-3
needs >= 18).
A typical error message for a SCSI command looks like:
scgcheck: I/O error. test unit ready: scsi sendcmd: no error
CDB: 00 20 00 00 00 00
status: 0x2 (CHECK CONDITION)
Sense Bytes: 70 00 05 00 00 00 00 0A 00 00 00 00 25 00 00 00 00 00
Sense Key: 0x5 Illegal Request, Segment 0
Sense Code: 0x25 Qual 0x00 (logical unit not supported) Fru 0x0
Sense flags: Blk 0 (not valid)
cmd finished after 0.002s timeout 40s
The first line gives information about the transport of the command.
The text after the first colon gives the error text for the system call
from the view of the kernel. It usually is: I/O error unless other
problems happen. The next words contain a short description for the
SCSI command that fails. The rest of the line tells you if there were
any problems for the transport of the command over the SCSI bus. fatal
error means that it was not possible to transport the command (i.e. no
device present at the requested SCSI address).
The second line prints the SCSI command descriptor block for the failed
The third line gives information on the SCSI status code returned by
the command, if the transport of the command succeeds. This is error
information from the SCSI device.
The fourth line is a hex dump of the auto request sense information for
The fifth line is the error text for the sense key if available,
followed by the segment number that is only valid if the command was a
copy command. If the error message is not directly related to the
current command, the text deferred error is appended.
The sixth line is the error text for the sense code and the sense
qualifier if available. If the type of the device is known, the sense
data is decoded from tables in scsierrs.c . The text is followed by
the error value for a field replaceable unit.
The seventh line prints the block number that is related to the failed
command and text for several error flags. The block number may not be
The eight line reports the timeout set up for this command and the time
that the command realy needed to complete.
Additional information can be found on:
If you have support questions, send them to:
If you have definitely found a bug, send a mail to:
To subscribe, use:
Joerg Schilling Version 3.0 SCGCHECK(1)