On Fri, 23 Nov 2007, Matthias Drochner wrote:

> I somehow suspect that there is some interaction between
> the callout changes and some other kernel problem which
> sends my box into the eternal spinlock.
>
> Any ideas?

it looks reasonable but I'm not sure that sc->sc_dev is initialised?
perhaps this will work..

--- if_wpi.c.orig	2007-11-23 20:17:16.000000000 +0000
+++ if_wpi.c	2007-11-23 20:16:49.000000000 +0000
@@ -201,6 +201,7 @@
 	pcireg_t data;
 	int error, ac, revision;

+	sc->sc_dev = self;
 	sc->sc_pct = pa->pa_pc;
 	sc->sc_pcitag = pa->pa_tag;


iain