Re: Kernel 2.0.29 - Ultrastor SCSI messages
Wed, 5 Mar 1997 22:23:26 +0100

>"Kernel 2.0.29 - Ultrastor SCSI messages"
>I compiled a new kernel 2.0.29. I have a SCSI adaptor (Ultrastor 14F), a
>SCSI HD and a SCSI CDROM. When I reboot with the new kernel, I get a
>lots of the following messages.
>U14F0: ihdlr, mbox 3, err 0x92:0, target 0.0:0, pid 909, count 1153
>I get these messages after ending an "startx" session.
>Can someone tell me what these errors mean? What do I do to get rid of
>these errors?
>Any help,pointers or advice is greatly appreciated?
>Thank you in advance.
Any time an i/o operation completes in error, the interrupt handler
in u14-34f prints a line of warning like the above.
On error 0x92 (data over/underrun) the driver retries (up to 64 times)
until the operation succeeds, otherwise it reports the error to the
scsi upper layer. If there were no further messages from the driver
or the scsi subsystem, you can safely ignore the warning: the error
has been recovered.
This kind of error is likely with old firmware revisions and also
with the latest one in case of heavy sequential i/o (using big
record sizes). Basically it means that the board has been stressed
at its limits and a little over, and the driver is doing its best
in order to avoid that the upper scsi layer could gets upset and take
inappropriate abort/reset actions in this transient situation.
Users can notice this errors when upgrading from 1.2.13 to 2.0.x,
since the performance improvement of 2.0.x trigger more problems
in the Ultrastor firmware.