Xen users - Nov 2005 - FreeBSD DomU Fatal trap 12 in cpu_switch_load_gs

I see the following with a FreeBSD 5.3 DomU on 2.0.7 

Fatal trap 12: page fault while in kernel mode
fault virtual address   = 0xca7a0088
fault code              = supervisor read, page not present
instruction pointer     = 0x819:0xc0183343
stack pointer           = 0x821:0xca069d30
frame pointer           = 0x821:0x0
code segment            = base 0x235, limit 0x8485e7c3, type 0x9
                        = DPL 0, pres 1, def32 1, gran 0
processor eflags        = interrupt enabled, resume, IOPL = 0
current process         = 1943 (cc)
trap number             = 12
panic: page fault

%nm -n XENUkernelNFS |grep c0183343
c0183343 T cpu_switch_load_gs

This is with a kernel compiled to run diskless over nfs and no kdb. With 
KDB enabled the machine just hangs (I assume the debugger is trying to 
attach to a non-existent serial port)

Should I be running 2.0.7 2.0-testing or 3.x to stand the best chance of getting
a stable FreeBSD DomU? Is this already fixed somewhere?

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

This is a bug that I had thought that I had fixed. I haven't done anything
with 5.3 on 2.0.7 in quite some time. Up until the xenbus changes I was
tracking 3.0 and was going to check into 6.0 CVS.

I ended up taking a break after xenbus went in. I didn't anticipate the
level of churn that came with the xenbus changes. However, judging from the
6-month slip in the planned release date of 3.0, no one else did either.

Is this a pre-built binary? Or one that you compiled? If it is one that you
compiled then your best bet may be waiting until FreeBSD works on 3.0 again.
I'd like to say soon, but it isn't at the front of my current queue.

-Kip



On 11/9/05, Geoff Buckingham <geoffb@chuggalug.clues.com>
wrote:
>
> I see the following with a FreeBSD 5.3 DomU on 2.0.7
>
> Fatal trap 12: page fault while in kernel mode
> fault virtual address = 0xca7a0088
> fault code = supervisor read, page not present
> instruction pointer = 0x819:0xc0183343
> stack pointer = 0x821:0xca069d30
> frame pointer = 0x821:0x0
> code segment = base 0x235, limit 0x8485e7c3, type 0x9
> = DPL 0, pres 1, def32 1, gran 0
> processor eflags = interrupt enabled, resume, IOPL = 0
> current process = 1943 (cc)
> trap number = 12
> panic: page fault
>
> %nm -n XENUkernelNFS |grep c0183343
> c0183343 T cpu_switch_load_gs
>
> This is with a kernel compiled to run diskless over nfs and no kdb. With
> KDB enabled the machine just hangs (I assume the debugger is trying to
> attach to a non-existent serial port)
>
> Should I be running 2.0.7 2.0-testing or 3.x to stand the best chance of
> getting a stable FreeBSD DomU? Is this already fixed somewhere?
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
>

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Wed, Nov 09, 2005 at 01:46:42PM -0600, Kip Macy
wrote:
> This is a bug that I had thought that I had fixed. I haven''t done
anything
> with 5.3 on 2.0.7 in quite some time. Up until the xenbus changes I was
> tracking 3.0 and was going to check into 6.0 CVS.
It is possible I have re-introduced it as I was initally unsure of which 
source to be using for my own kernels. I think I built this through 
following what I now realise are out-dated instructions on the xenobsd pages.

I have had crashes with kernels built from the included sparse freebsd 
directory but as they have kdb enabled I don''t see any output on the
console,
they may have a different cause.

I will rebuild this kernel config from the sparse directory provided in 2.0.7
and see if I get the same or different errors.

PS Would you accept patches to your build scripts to work on
''stock'' FreeBSD
systems (no wget, bash in /usr/local/bin etc.) 

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users