lockups with 5.0.7 & 2.0.33
Patrick Michael Kane
modus at pr.es.to
Thu Feb 19 08:00:50 PST 1998
On Thu, Feb 19, 1998 at 06:32:08AM -0600, Doug Ledford wrote:
> Patrick Michael Kane wrote:
> >
> > Greets.
> >
> > I am experiencing lockups (no log messages) on a Linux system, running
> > 2.0.33, using either the stock 2.0.33 driver or the new 5.0.7 driver using
> > an Adaptec 2940W and an IBM DCHS04U wide internal drive.
>
> > Tag Queue Enable Flags: 0x0000
> > Ordered Queue Tag Flags: 0x0000
> > Tagged Queue By Device array for aic7xxx host instance 0:
> > {255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255}
> > Actual queue depth per device for aic7xxx host instance 0:
> > {1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}
> >
> > The system locks, with the drive access light on, whenever I do extended
> > writes, normally within a few minutes of starting the operation. No errors
> > are logged out to any of the logfiles or printed on the screen.
>
> I think you have a hardware problem. Try disabling your L2 cache in your
> BIOS and see what happens. I've never heard of the current driver (or
> sequencer) locking up the SCSI bus when you have tagged queueing disabled
> and only one device, there simply isn't enough going on with the bus. Your
> current setup is exactly that, tagged queueing on the IBM drive is disabled,
> you're only using one SCB ever, and you're locking up during large writes
> (aka, heavy DMA activity across the PCI bus). That's why I suspect either
> cache or RAM.
Hmmm. I'll try this.
> > Oddly enough, the system seemed to work fine under a slackware
> > installation that was recently replaced with redhat 5.0.
>
> Older, slower version of the driver, less likely to max out bus transfer
> speeds I would guess :)
Nope. The old install was running 2.0.33, as well. That's what is baffling
me. If it is indeed a hardware problem, why didn't it manifest itself under
the previous install?
--
Patrick Michael Kane
<modus at pr.es.to>
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