[4029] in linux-scsi channel archive

home help back first fref pref prev next nref lref last post

Re: SCSI Timeout on AIC7xxx when PPA is active.

daemon@ATHENA.MIT.EDU (Mark Garlanger)
Thu May 28 21:24:32 1998

Date: 	Thu, 28 May 1998 20:16:59 -0500
From: Mark Garlanger <garlangr@cyberramp.net>
To: David Campbell <campbell@gear.torque.net>
CC: linux-scsi@vger.rutgers.edu

David Campbell wrote:

> > Date:          Thu, 28 May 1998 06:59:03 -0500
> > From:          Mark Garlanger <garlangr@cyberramp.net>
> > To:            David Campbell <campbell@gear.torque.net>
> > Cc:            linux-parport@torque.net, linux-scsi@vger.rutgers.edu
> > Subject:       Re: SCSI Timeout on AIC7xxx when PPA is active.
>
> *snip*
>
> > 1) Zip drive unplugged but driver still in the kernel
>
> *snip*
> SMP     => OK
> !SMP    => OK (??)

Since SMP worked, I felt !SMP would work without issue...


>
>
> > 2) Zip drive plugged in, turned on but no disk in drive
>
> *snip*
> SMP     => Crash
> !SMP    => ????

Tried the !SMP a few minutes ago, it crashes too.

>
>
> > 3) Zip drive plugged in, turned on but with a disk in drive
>
> *snip*
> SMP     => Crash
> !SMP    => ????

!SMP also crashes..

>
>
> *light bulb over head*
>
> The problem may reside in the the AIC7xxx driver. When *any* low level
> driver completes a command then the mid-level driver checks *all* low
> level drivers to see if they can accept another command.
>
> Perhaps when ppa returns from a command it sees that the AIC7xxx is free
> (incorrectly) and sends it a SCSI command which jams the queue (this
> would fit the symptoms).
>
> Another posibility is that ppa calls the completion function in an
> interrupt (OK - software interrupt. tq_timer to be exact). The AIC7xxx may
> receive its next instruction in this state and do something strange with
> it.

Would it be possible to move this function completion call to outside the
interruptin order to test this out.

        Thanks,
            Mark

>
>
> I will take a quick peek at the AIC code, don't expect any miracles.
>
> David Campbell
> =======================================================
> campbell@torque.net
>
> "All parallel ports are equal - Some are more equal than others"




-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.rutgers.edu

home help back first fref pref prev next nref lref last post