Subject: Re: Random timeouts in 1.0-ALPHA...
To: Mark Steven de Sagun-Tamola <buckwild@u.washington.edu>
From: Michael L. VanLoon -- Iowa State University <michaelv@iastate.edu>
List: current-users
Date: 07/21/1994 14:48:00
>I'm getting random timeouts and lost interrupts whenever disk access gets 
>heavy.  The error message reads as follows:
>
>wdc0: lost interrupt: status 58<seekdone,drq> error 0
>wd: wdcstart: timeout waiting for drq: status 0 error 0
>
>This message just repeats itself over and over until I do a ^C to stop 

This sounds like you may have a bad spot on your hard drive, and the
drive is trying to recalibrate or move the spot, while in the mean
time the wd driver is timing out waiting for the drive.  You can
verify this by 1) borrowing an IDE drive from a friend who'll let you
reformat it to put NetBSD on it, and see if that drive works better,
or 2) get a DOS exhaustive low-level formatter like SpinRite, or the
one in PC Tools, and let it run all night on your drive.  It will tell
you if you have bad, or even marginal, spots on your drive.

Either way, the problems aren't caused by software -- the problems are
due to the low quality of much of the IDE hardware on the market (IDE
is an extremely loosely defined "standard").  I suspect it's your
drive that's giving you fits.

>I think it may be the bounce buffer code in the new wd driver that may be 
>causing this, but I'm not so sure.

IDE drives don't use DMA (at least under the wd driver), so they don't
have bounce buffers.

				--Michael

-----------------------------------------------------------------------------
 Michael L. VanLoon                 Iowa State University Computation Center
    michaelv@iastate.edu                    Project Vincent Systems Staff
  Free your mind and your machine -- NetBSD free Un*x for PC/Mac/Amiga/etc.
-----------------------------------------------------------------------------

------------------------------------------------------------------------------