Subject: Re: Is NetBSD domain0 ready for production server?
To: Martti Kuparinen <martti.kuparinen@iki.fi>
From: Manuel Bouyer <bouyer@antioche.lip6.fr>
List: port-xen
Date: 03/30/2005 10:16:44
On Wed, Mar 30, 2005 at 10:34:58AM +0300, Martti Kuparinen wrote:
> Manuel Bouyer wrote:
> 
> >>xvif1.0: Ethernet address aa:00:00:4a:0c:48
> >>xbdback VBD grow domain 1: unknwon device 0x44e98
> >>xbdback VBD grow domain 1: unknwon device 0x44e98
> >
> >
> >This looks completely wrong, the device should be 0x0405 for /dev/wd0f.
> >It's not even a sign extention bug or an uninitalised field, is't 
> >completely
> >different.
> 
> >What version of the xen kernel and the tools do you have ?
> 
> # uname -a
> NetBSD xen0 3.99.2 NetBSD 3.99.2 (XEN0) #0: Mon Mar 28 18:38:51 EEST 
> 2005  root@xen0:/usr/src/sys/arch/i386/compile/XEN0 i386
> 
> # pkg_info | grep xen
> xentools20-2.0.3nb3 Userland Tools for Xen

Hum, I'm still at 2.99.16 or 3.0_BETA here.
I'm building a 3.99.2 now.

> 
> >Can you try to ktrace xm and xend to see what it's doing with /dev/wd0f ?
> 
> # ktrace /usr/pkg/sbin/xend start
> # ktrace xm create -c /xen/conf/linux1
> 
> # kdump (see http://users.piuha.net/martti/tmp/xen.log)

Doing is this way, kdump only outputs the trace for xm, as the first
ktrace.out file is overwritten.
It looks like the real work is done by xend anymay.
So you want only:
ktrace -i /usr/pkg/sbin/xend start
(-i so that we get the childs are traced too).

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--