Re: [Xen-devel] (XEN) traps.c:3156: GPF messages in xen dmesg

[Sander Eikelenboom <linux@xxxxxxxxxxxxxx>]

Tuesday, December 18, 2012, 8:49:18 AM, you wrote:

>>>> On 17.12.12 at 18:55, Sander Eikelenboom <linux@xxxxxxxxxxxxxx> wrote:

>> Monday, December 17, 2012, 5:57:24 PM, you wrote:
>> 
>>>>>> On 17.12.12 at 17:48, Valtteri Kiviniemi <kiviniemi.valtteri@xxxxxxxxx> 
>>>>>> wrote:
>>>> I'm running Xen 4.2.0 with Linux kernel 3.7.0 and I'm seeing a flood of
>>>> these messages in xen dmesg:
>>>> 
>>>> (XEN) traps.c:3156: GPF (0060): ffff82c480159247 -> ffff82c4802170e4
>>>> (XEN) traps.c:3156: GPF (0060): ffff82c480159247 -> ffff82c4802170e4
>>>> (XEN) traps.c:3156: GPF (0060): ffff82c480159247 -> ffff82c4802170e4
>>>> (XEN) traps.c:3156: GPF (0060): ffff82c480159247 -> ffff82c4802170e4
>>>> (XEN) traps.c:3156: GPF (0060): ffff82c480159247 -> ffff82c4802170e4
>>>> (XEN) traps.c:3156: GPF (0060): ffff82c480159247 -> ffff82c4802170e4
>>>> (XEN) traps.c:3156: GPF (0060): ffff82c480159247 -> ffff82c4802170e4
>>>> 
>>>> What does that mean and is it something that I should worry about?
>> 
>>> It means that the hypervisor recovered from #GP faults several
>>> times. What exactly it was that faulted you'd need to look up by
>>> translating the addresses above and resolving them to source
>>> locations. That'll also tell you whether you ought to be worried.
>> 
>>> A common example of these happening is Xen carrying out MSR
>>> accesses on behalf of the kernel, when the MSR actually isn't
>>> implemented (i.e. the kernel itself also is prepared to handle
>>> faults upon accessing them).
>> 
>> Perhaps related to a previous discussion (with a open end ..)
>> http://lists.xen.org/archives/html/xen-devel/2012-09/msg01599.html 

> Not impossible, but iirc those would generally be accompanied
> by SIGSEGV-s and/or kernel crashes in some guest.

> Jan

Valtteri,

Are you by any chance using pci-passthrough to one of your domains ?

--

Sander


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