At 08:40 +0100 on 08 Aug (1312792831), Jan Beulich wrote:
> >>> On 03.08.11 at 19:58, "Kay, Allen M" <allen.m.kay@xxxxxxxxx> wrote:
> > Off hand, I don't see any reason iommu_setup() cannot be moved after
> > do_presmp_initcalls() in __start_xen().
> >
> > Theoretically, iommu_setup() just needs to be called before any device
> > operation. We should give it a try.
>
> So who's going to try this out then, and submit a patch?
I am, once I get back to the office (probably Wednesday)
Tim.
> > Allen
> >
> > -----Original Message-----
> > From: Tim Deegan [mailto:tim@xxxxxxx]
> > Sent: Wednesday, August 03, 2011 10:13 AM
> > To: Ren, Yongjie
> > Cc: Xen Devel; Kay, Allen M; Li, Xin
> > Subject: Re: [Xen-devel] [bug] 'VT-d 1G super page' feature is blocked
> >
> > At 10:08 +0800 on 03 Aug (1312366080), Ren, Yongjie wrote:
> >> Hi Tim,
> >> Any comment on this issue ?
> >> It seems that cs# 23247 introduced this issue.
> >
> > Hi,
> >
> > Allen points out that the EPT capability variable that is checked
> > doesn't get initialized until VMX is set up, which right now happens
> > _after_ the IOMMU is configured.
> >
> > Allen, is there any reason that the call to iommu_enable() shouldn't be
> > moved to after do_presmp_initcalls()? If so, then I guess we should
> > open-code the hardware capability check in the vtD init code.
> >
> > Tim.
>
>
>
>
--
Tim Deegan <tim@xxxxxxx>
Principal Software Engineer, Xen Platform Team
Citrix Systems UK Ltd. (Company #02937203, SL9 0BG)
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|
Home