Re: page fault trap installing NetBSD 5.0.1

[Sarton O'Brien <bsd-xen%roguewrt.org@localhost>]

Hi Jeremy,

On 25/09/2009 3:31 AM, Jeremy C. Reed wrote:
> Please Cc me on replies.
>
> If this is already covered, please point me to it. I am not sure if I am
> using the correct list as I don't know Xen. (I have used remotely hosted
> NetBSD systems running on Xen but I never installed them.)
>
> I am trying to install using the amd64cd-5.0.1.iso
>
> I don't want to install a Xen server. (As you can tell, I don't know the
> terminology.) There is already a Debian Linux server with Linux
> 2.6.26-1-xen-amd64 kernel running xenwatch, xenbus, xenrestored, xend,
> xenconsoled, and many qemu-dm processes. I think that is xen-3.2-1. (It
> is hosting OpenSolaris, OpenBSD, Windows, Debian, and others.)
>
> I don't know Xen and I installed by logging into the Debian system and
> running:
>
>         lvcreate -n netbsd-5-amd64-1 -L 8G virt2
>
> I downloaded the NetBSD amd64cd-5.0.1.iso image.
>
> I setup a /etc/xen/vm/base/netbsd-5-amd64-1 file. If you need to see
> this, let me know.
>
> I ran:
>
>         xm create /etc/xen/vm/base/netbsd-5-amd64-1
>
> A qemu-dm process was started. I setup some SSH tunnels and connected to
> it with vncviewer. It showed the NetBSD 5.0.1 installer. I went through
> the many steps and it formatted its (virtual) wd0.
>
> But when it got to the untar step in sysinst (extracting
> kern-GENERIC.tgz), it aborted with:
>
> kernel: page fault trap, code=0
> Stopped in pid 0.4 (system) at  0xffffffff804f163e:    repe insl
> %dx,%es:(%rdi)
> db{0}>  bt
> ?() at 0xffffffff804f163e
> ?() at 0xffffffff804fefc7
> ?() at 0xffffffff802c3207
> ?() at 0xffffffff804e0e4d
> ?() at 0xffffffff80105a4f
> Bad frame pointer: 0xffff800001215ac40
>
> Any ideas how I can troubleshoot this further?
>
> Or am I doing this wrong?
>
> (By the way, my goal is to install NetBSD/amd64 5.x so I can build some
> pkgsrc packages for X11_TYPE=modular. I will be glad to share these by
> uploading to the NetBSD ftp.)

Although you might be doing something wrong, it would be interesting to 
know if a current snapshot has the same issue. Maybe we're missing a 
pullup ...

Normally the first issue someone encounters with NetBSD/xen is a sparse 
file problem but seeing as linux is the dom0, I'm not sure if that may 
translate to issues within a NetBSD HVM guest. This problem would 
normally occur during formatting IIRC.

Sorry, I'm not much help ... it seems strange a release would panic 
under relatively normal operation. Hopefully it's already been addressed.

Sarton