Hifn driver in SMP (was Re: GELI - disk encryption GEOM class committed.)

[Sam Leffler]

Pawel Jakub Dawidek wrote:
> On Mon, Aug 08, 2005 at 09:55:45AM +0200, Michael Reifenberger wrote:
> +> On Sun, 7 Aug 2005, Mike Tancsa wrote:
> +> 
> +> >Date: Sun, 07 Aug 2005 23:37:02 -0400
> +> >From: Mike Tancsa <mike at sentex.net>
> +> >To: Pawel Jakub Dawidek <pjd at freebsd.org>
> +> >Cc: freebsd-current at freebsd.org
> +> >Subject: Hifn driver in SMP (was Re: GELI - disk encryption GEOM class
> +> >    committed.)
> +> >Actually, it seems I can easily lock up the box with
> +> >/usr/bin/openssl aes-128-cbc -in big -k pass | ssh -c aes128-cbc mdtancsa at 127.0.0.1 "cat - >  /home/mdtancsa/targetfile.enc"
> +> >Is the hifn driver perhaps not fully smp safe ?
> +> 
> +> BTW:
> +> I've had hard lockups using the previous version of GELI and hifn on UP and amd64 (if that matters) too.
> +> newfs worked but iozone locked up allways...
> 
> You're probably seeing the same problem I have.
> 

Are all the lockups on amd systems?  I definitely tested hifn on a 
dual-xeon w/o issues but phk and others have reported various issues 
with hifn-based cards on operteron systems (though not hard lockups from 
what I recall).

Given that I don't have an smp amd system someone w/ such a setup will 
need to collect info to help.  Can you break to the debugger and inspect 
state?  There's also debug.hifn I think (or something like it) to get 
debug msgs to the console.

	Sam