Subject: Re: experiences with 3.0_BETA
To: Pavel Cahyna <pavel.cahyna@st.mff.cuni.cz>
From: Manuel Bouyer <bouyer@antioche.lip6.fr>
List: port-xen
Date: 05/23/2005 14:28:53
On Sun, May 22, 2005 at 01:47:51PM +0200, Pavel Cahyna wrote:
> Hello,
>
> I just booted the 3.0_BETA kernel from the daily autobuilds and it works!
> Thanks a lot for thit work.
>
> I encountered several issues, however:
>
> 1/ most serious: Xfree86 crashed with signal 11 when switching desktops in
> windowmaker. This never happened with my 2.0_BETA/i386 kernel. The whole
> userland is 2.0_BETA, but this shouldn't matter, no? What could be done to
> debug this?
It could be because NetBSD/Xen doesn't support MTTR yet. Known issue, and
I think there is even a PR open about this
>
> 2/ some utilities print something like "ksyms: device not configured" when
> booting, and vmstat doesn't work. I found the answer in the post
> Message-ID: <20050409182722.GA10451@antioche.eu.org>
> - it is necessary to use a patched Xen to load the symbol table for domain
> 0, which one can get by ftp from pub/NetBSD/arch/xen/xen20.
> Could this information be added to the Xen HOWTO
> (http://www.netbsd.org/Ports/xen/howto.html#netbsd-domain0), please? For
> example: "Until a version of Xen which supports loading the symbol table
> is released, it is better to use a patched version available from
> ftp://ftp.NetBSD.org/pub/NetBSD/arch/xen/xen20/xen205-loadksym.gz
> Otherwise, commands like netstat or vmstat will probably fail and symbolic
> kernel debugging in ddb will be impossible."
>
> 3/ This is probably harmless, but the SCSI bus scanning is very verbose:
> scsibus0: waiting 2 seconds for devices to settle...
> ahc0:A:2:0: Ensuring async
This is because the kernel is compiled with 'options DEBUG'.
--
Manuel Bouyer <bouyer@antioche.eu.org>
NetBSD: 26 ans d'experience feront toujours la difference
--