Subject: Re: siop crash... was Re: rename of scsipi kernel thread
To: John Clark <>
From: Manuel Bouyer <>
List: tech-kern
Date: 04/01/2002 21:35:30
On Sun, Mar 31, 2002 at 12:00:35PM -0800, John Clark wrote:
> Am Sonntag den, 31. März 2002, um 12:11, schrieb Manuel Bouyer:
> >Hi,
> >would anyone object if I rename the per-channel scsipi kernel thread 
> >from
> >controller:channel to the scsipi channel name (e.g. siop0:0 -> scsibus0,
> >ahc0:1 -> scsibus1, pciide0:0 -> atapibus0, etc ...) ?
> This post reminded me of something. I have had the siop driver crash
> when a device is attached to my 53c1010 interface. I've not got the exact
> absolute reason why, but the symptom is this. When the interface chip
> thinks the bus is differential, but the actual device is single-ended, 
> there
> is a mismatch error. Rather than resetting the SE/LVD mode, the driver
> causes a kernel panic.
> This also occurs in the Linux version as well. I'll look into it more, 
> but I
> though I'd mention it, and see if anyone has had a similar strange 
> problem.
> (In the linux driver I got to the point where the scripts interrupt code 
> tried to
> continue the scripts, after which the panic... haven't looked in that 
> detail
> into the netbsd version.)
> The interface works correctly with a device that always is LVD, and I 
> thought
> that if the drive forced SE before the interface chip was 
> initialized(via jumpers
> or whatnot, that worked too, but I'll have to retrace my steps.

I just received a NCR53c1010. I should be able to check this in the next
few days (as well as adding support for Ultra160 :)

Manuel Bouyer <>