Re: [Xen-devel] Debuging dom0, oops while kexec

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 21/06/13 13:01, Daniil Cherednik wrote:
> Hello!
>
> We use ubuntu 12.04 on dom0 xen-4.1, linux-3.5
> We have some issue with our hardware driver, and I need to get dom0
> coredump. But when I was trying to configure kdump it was unsuccessful
> and I used kexec -l ... to diagnostic the issue I got oops like
> http://bugzilla.xensource.com/bugzilla/show_bug.cgi?id=1807
> Also I noticed recently it was discussing about implementation kexec
> mechanism in hypervisor without involving dom0 kernel...
> So I will be grateful if someone can take me any advises about using
> kexec with xen and about bug1807
>
> Best regards,
> Daniil Cherednik

First of all, you want -p for a crash kernel, not -l (which is a load
kernel).

That link shows a linux-3.5 bug.  At a guess machine_kexec_prepare is
not suitably sanitising some pointers before trying to use them.

There are patch series for Xen (and libxc) and kexec tools.  The Xen
patch series is being deferred until Xen 4.4 development starts.  The
kexec tools series also needs to wait until the libxc patches are available.

The slight problem with this approach is that you need to patch the
kernel to unconditionally issue a kexec hypercall as it no longer knows
whether a crash image has been loaded or not.

~Andrew

>
>
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxx
> http://lists.xen.org/xen-devel


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel