Xen devel - Jun 2013 - dom0 crash without any message

Hello,

I''m unable to boot a dom0 3.4.x kernel on top of xen4.1.x
Xen seems to boot fine but the dom0 is crashing without any error:

(XEN) Xen trace buffers: disabled
(XEN) Std. Loglevel: All
(XEN) Guest Loglevel: All
(XEN) *** Serial input -> DOM0 (type ''CTRL-a'' three times
to switch
input to Xen)
(XEN) Freed 232kB init memory.
mapping kernel into physical memory
about to get started...
(XEN) Domain 0 crashed: ''noreboot'' set - not rebooting.


I have attached my dom0 config and the complete boot log. I never went
through this kind of issue, i.e dom0 crashing silently.
The exact same hypervisor+kernel is working fine on older machines.
Any hint?

Best regards,

--
William



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

On Mon, 2013-06-03 at 18:14 +0200, William Dauchy wrote:
> Hello,
> 
> I''m unable to boot a dom0 3.4.x kernel on top of xen4.1.x
> Xen seems to boot fine but the dom0 is crashing without any error:
> 
> (XEN) Xen trace buffers: disabled
> (XEN) Std. Loglevel: All
> (XEN) Guest Loglevel: All
> (XEN) *** Serial input -> DOM0 (type ''CTRL-a'' three
times to switch
> input to Xen)
> (XEN) Freed 232kB init memory.
> mapping kernel into physical memory
> about to get started...
> (XEN) Domain 0 crashed: ''noreboot'' set - not rebooting.
> 
> 
> I have attached my dom0 config and the complete boot log. I never went
> through this kind of issue, i.e dom0 crashing silently.
> The exact same hypervisor+kernel is working fine on older machines.
> Any hint?
earlyprintk=xen on the dom0 kernel command line might add something
useful to the logs.
> 
> Best regards,
> 
> --
> William
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel

Hello
On Mon, Jun 3, 2013 at 6:37 PM, Ian Campbell <Ian.Campbell@citrix.com>
wrote:
> earlyprintk=xen on the dom0 kernel command line might add something
> useful to the logs.
true, I forgot about that option. It''s now crashing on "Max
apic_id
exceeded!". Of course the boot continues with nolapic option.
Do you know how can I workaround that without nolapic option?

--
William


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

On 03/06/13 18:38, William Dauchy wrote:
> Hello
> On Mon, Jun 3, 2013 at 6:37 PM, Ian Campbell
<Ian.Campbell@citrix.com> wrote:
>> earlyprintk=xen on the dom0 kernel command line might add something
>> useful to the logs.
> true, I forgot about that option. It''s now crashing on "Max
apic_id
> exceeded!". Of course the boot continues with nolapic option.
> Do you know how can I workaround that without nolapic option?
>
> --
> William
You need to boot with dom0_max_vcpus=16 or so

If I remember our internal tickets correctly, some PVops gets confused
with more than 16 vcpus.

~Andrew

On Mon, Jun 3, 2013 at 7:45 PM, Andrew Cooper <andrew.cooper3@citrix.com>
wrote:
> You need to boot with dom0_max_vcpus=16 or so
> If I remember our internal tickets correctly, some PVops gets confused
> with more than 16 vcpus.
I tried 8 or 16, it doesn''t change anything without nolapic.

--
William

On Mon, Jun 03, 2013 at 07:38:55PM +0200, William Dauchy
wrote:
> Hello
> On Mon, Jun 3, 2013 at 6:37 PM, Ian Campbell
<Ian.Campbell@citrix.com> wrote:
> > earlyprintk=xen on the dom0 kernel command line might add something
> > useful to the logs.
> 
> true, I forgot about that option. It''s now crashing on "Max
apic_id
> exceeded!". Of course the boot continues with nolapic option.
> Do you know how can I workaround that without nolapic option?
Do you see this if you boot the same kernel, but without Xen?

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

On Tue, Jun 4, 2013 at 4:05 PM, Konrad Rzeszutek Wilk
<konrad.wilk@oracle.com> wrote:
> Do you see this if you boot the same kernel, but without Xen?
I haven''t tested with a i386 baremetal kernel but with an amd64
baremetal kernel it boots fine (thanks to the X2APIC in amd64). I
guess that on i386 I''ve reached the 0xff APIC limit.
My dom0 is now booting when disabling two cores in the bios config; it
means reducing the cpu from 32 to 24. I hoped for a cmdline solution
instead of changing the bios config but did not find any. Did I miss
one? dom0_max_vcpus did not worked.

--
William

On Tue, Jun 04, 2013 at 04:21:45PM +0200, William Dauchy
wrote:
> On Tue, Jun 4, 2013 at 4:05 PM, Konrad Rzeszutek Wilk
> <konrad.wilk@oracle.com> wrote:
> > Do you see this if you boot the same kernel, but without Xen?
> 
> I haven''t tested with a i386 baremetal kernel but with an amd64
> baremetal kernel it boots fine (thanks to the X2APIC in amd64). I
> guess that on i386 I''ve reached the 0xff APIC limit.
I would think that the 64-bit kernel under Xen would work as well?
> My dom0 is now booting when disabling two cores in the bios config; it
> means reducing the cpu from 32 to 24. I hoped for a cmdline solution
> instead of changing the bios config but did not find any. Did I miss
> one? dom0_max_vcpus did not worked.
There is also the Linux ones - nr_cpus= or some variant of it.

But instead of using work-arounds I would like to understand why you are
hitting this and fix it.

Hence the question about baremetal - and if you could kindly also attach
the serial log of that bootup - it could in figuring out what is going
on.
> 
> --
> William
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel
>

On Tue, Jun 4, 2013 at 6:26 PM, Konrad Rzeszutek Wilk
<konrad.wilk@oracle.com> wrote:
> I would think that the 64-bit kernel under Xen would work as well?
haven''t tested but I guess so regarding the issue.
> There is also the Linux ones - nr_cpus= or some variant of it.
yup, it only worked through bios configuration.
> But instead of using work-arounds I would like to understand why you are
> hitting this and fix it.
> Hence the question about baremetal - and if you could kindly also attach
> the serial log of that bootup - it could in figuring out what is going
> on.
you mean an i386 baremetal log?

--
William

On Wed, Jun 05, 2013 at 02:52:49PM +0200, William Dauchy
wrote:
> On Tue, Jun 4, 2013 at 6:26 PM, Konrad Rzeszutek Wilk
> <konrad.wilk@oracle.com> wrote:
> > I would think that the 64-bit kernel under Xen would work as well?
> 
> haven''t tested but I guess so regarding the issue.
> 
> > There is also the Linux ones - nr_cpus= or some variant of it.
> 
> yup, it only worked through bios configuration.
> 
> > But instead of using work-arounds I would like to understand why you
are
> > hitting this and fix it.
> > Hence the question about baremetal - and if you could kindly also
attach
> > the serial log of that bootup - it could in figuring out what is going
> > on.
> 
> you mean an i386 baremetal log?
Yes.
> 
> --
> William