On Thu, 26 Jun 1997, Rick Hawkins wrote:
>
>
> > This means you finally got the CD drive working?
>
> Another one, actually. I plugged it into the controller, inserted the
> module, and mounted with no problem. THe other drive is indeed bad :(
> Gee, it was all so easy with a working piece :)
>
> However, it does strike me as a problem that the bad drive can hang the
> system. Is this a bug, or acceptable behavior?
>
Without sounding to contradictory, I don't think it is a bug, but it isn't
acceptable behavior either.
While there are many kinds of hardware failure that will "bring the
machine down" reguardless of the OS, and these problems certainly can't be
fixed by a more "robust" driver. There are many failures that these
drivers should be able to recover from that they don't seem able to do.
I had an aha1542 controler that got partially fried by the death of the CD
drive that was attatched to it. Upon replacing the drive, the new drives
would eventually fail, until I replaced the controler. When they failed
the controler would find itself in conflict between an I/O error and the
device timeout that caused it. After several retries at reseting the scsi
device, which pushed the drive into a locked state, the driver would lock
up so tight that nothing else could happen in the system and a power-down
reboot was required to correct the problem. If the SCSI subsystem were
more robust, this would have resolved itself into an unrecoverable error
and the process could have died gracefully, leaving the system in a useful
state.
The reason non of this is a bug is that the design of the software
"requires" that the hardware work in the advertised fashion. While this
may sound like an excuse for failure to provide adequate error recovery,
if you think about it some, you will conlcude that in every error recovery
path there is an additional chance for failure due to hardware
deficiencies. This recursion makes it impossible to cover all the bases,
and you are left with the software we have. Sometimes it can manage under
extreme conditions and sometimes it can't.
Glad to hear you found the problem,
Dwarf
--
_-_-_-_-_-_- _-_-_-_-_-_-_-
aka Dale Scheetz Phone: 1 (904) 656-9769
Flexible Software 11000 McCrackin Road
e-mail: [EMAIL PROTECTED] Tallahassee, FL 32308
_-_-_-_-_-_- If you don't see what you want, just ask _-_-_-_-_-_-_-
--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
[EMAIL PROTECTED] .
Trouble? e-mail to [EMAIL PROTECTED] .
