Re: [Xen-devel] [V8 PATCH 0/8] pvh dom0....

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 22.03.14 at 02:39, <mukesh.rathor@xxxxxxxxxx> wrote:
> Patch 2: comment changes (based on feedback from Konrad):
> 
>   -            /* Unused RAM areas are marked UNUSABLE, so skip it too */
>   +            /* Unused RAM areas are marked UNUSABLE, so skip them too */
> 
> -    /* If the e820 ended under 4GB, we must map the remaining space upto 4GB 
> */
> +    /* 
> +     * Some BIOSes may not report io space above ram that is less than 4GB. 
> So
> +     * we map any non-ram upto 4GB.
> +     */
> 
> Jan, I retained Reviewed-by. If above voids the "reviewed-by", I'll just undo
> the comment changes. Please lmk if that's the case.

Definitely not - you'd have to make rather extensive changes to
comments to void a Reviewed-by (or a patch would have to consist
mostly of comments or changes thereto).

> Caveat:
> 
> With these patches, an HVM guest during boot intermittantly takes
> unresolved fault in load_module() path. This for PVH dom0 and not PV, and
> latest upstream kernel. I was not able to reproduce with older stock 
> fedora 19 kernel. So, it appears the problem might be on the guest side, 
> and not xen. Hence I decided to go ahead and submit these while I debug 
> that.

Without proper understanding of this I don't see a way for us to apply
the series, and it should therefore have been tagged as RFC. Just to
avoid any misunderstanding - even if the root cause was on the guest
side, if without your patches the issue never surfaces, this would still
need to be treated as a regression, and a way would need to be found
to make those guests work flawlessly again. Furthermore, with the
issue being in HVM guests, it is rather unlikely (but admittedly not entirely
impossible) that the issue is in the guest, as similar problems aren't - aiui-
known from Linux running on native hardware.

Apart from that, assuming you spent quite some time in isolating the
problem - did you determine which of the patches in this series
introduces it? Knowing that might help in determining possible reasons
(including having others help you with that).

Jan


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