How to force a bus reset?
Jim Thompson
jim.thompson at pobox.com
Fri Jan 8 07:16:02 PST 1999
>>>>> "Brian" == Brian Ristuccia <brianr at osiris.978.org> writes:
Brian> I've been having similar woes with my Archive DAT
Brian> changer. When it hits end-of-tape it gets into an error
Brian> condition that can only be cleared by removing and
Brian> re-inserting the magazine.
I have similar problems with my tape drive, a refurbed Exabyte 8200.
When it hits EOT, some mt commands cause it to hang. The drive shows no
apparent activity, the command never completes, and can't be interrupted.
Cycling power to the drive sometimes recovers the situation -- the mt
command finishes. At other times, the mt command is still hung after
power cycle. The system log shows no SCSI command timeouts, nor bus
resets.
The commands that are most likely to lock up the driver from EOT are
commands that would try to seek past EOT: "mt fsf 1", "mt eod". The
rewind commands "mt rewind" and "mt offline" are OK; I haven't tried a
bsf command. Frankly, this seems more like an st driver problem than an
aic7xxx problem. Any clues?
Incidently this tape drive, or one of the linux drivers, was a real
nightmare when I was using my Buslogic SCSI card. The drive itself
worked fine, but using it appeared to cause kernel memory corruption. I
got warnings from the paging system complaining about attempts to load
nonexistent pages, errors reported by the VFS system, corrupt ext2
partitions, kernel panics, system hangs, and system resets. Turning off
disconnect for the drive appeared to fix (or mask) the problem, but of
course also made the system unusably slow when the drive was in use.
Since going to the aic7xxx driver I haven't had any problem except the
relatively painless drive hangs mentioned above. Thanks, Doug, for
making my drive usable.
Jim
--
_ |
_| ~- Jim Thompson | Attaboy, Clarence!
\, _} jim.thompson at pobox.com |
\( http://pobox.com/~kzinti |
To Unsubscribe: send mail to majordomo at FreeBSD.org
with "unsubscribe aic7xxx" in the body of the message
More information about the aic7xxx
mailing list