WARNING - OLD ARCHIVES

This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
   
 
 
Xen 
 
Home Products Support Community News
 
   
 

xen-devel

Re: [Xen-devel] Re: xen: use paging_mode_hap()

At 08:53 +0100 on 07 Apr (1270630406), Xu, Dongxiao wrote:
> We found that guests with Intel EPT could not boot up with latest
> upstream Xen.
> 
> The root cause is: 
> When calling p2m_init() function, the PG_HAP_enable hasn't been
> written to paging mode, which Tim has pointed out in this mail.

True - I thought that reshuffling hap_enable() would be good enough, but
of course p2m_init() gets called before any hap code, so this patch is
needed.

Acked-by: Tim Deegan <Tim.Deegan@xxxxxxxxxx>

> After adding the following patch, HVM guest with hap=1 works well.
> 
> Signed-off-by: Dongxiao Xu <dongxiao.xu@xxxxxxxxx>
> 
> diff -r adce8bc43fcc xen/arch/x86/mm/p2m.c
> --- a/xen/arch/x86/mm/p2m.c     Tue Apr 06 07:16:47 2010 +0100
> +++ b/xen/arch/x86/mm/p2m.c     Wed Apr 07 23:48:14 2010 +0800
> @@ -1712,7 +1712,7 @@ int p2m_init(struct domain *d)
>      p2m->get_entry_current = p2m_gfn_to_mfn_current;
>      p2m->change_entry_type_global = p2m_change_type_global;
> 
> -    if ( is_hvm_domain(d) && paging_mode_hap(d) &&
> +    if ( is_hvm_domain(d) && d->arch.hvm_domain.hap_enabled &&
>           (boot_cpu_data.x86_vendor == X86_VENDOR_INTEL) )
>          ept_p2m_init(d);
> 
> 
> Thanks,
> Dongxiao
> 
> 
> Christoph Egger wrote:
> > On Wednesday 31 March 2010 11:29:32 Tim Deegan wrote:
> >> At 16:51 +0100 on 30 Mar (1269967863), Christoph Egger wrote:
> >>> Hi!
> >>> 
> >>> Attached patch makes xen consequently use paging_mode_hap().
> >>> 
> >>> Signed-off-by: Christoph Egger <Christoph.Egger@xxxxxxx>
> >>> 
> >>> P.S.: This is a resend. Needs ack from Tim.
> >> 
> >> Nack, at least in its current form - paging_mode_hap(d) is not the
> >> same as d->arch.hvm_domain.hap_enabled.  Especially in paging.c,
> >> where we test hap_enabled to gate whether we should call the
> >> function that makes paging_mode_hap(d) true! 
> >> 
> >> I think the p2m.c changes are wrong too, since the p2m init code is
> >> called from hap_enable before it writes PG_HAP_enable to the paging
> >> mode.  (That's probably a bug in its own right; it should set the
> >> mode and then call the p2m code, and shadow should do likewise, and
> >> then these changes would be correct and desirable).
> >> 
> >> The other substitutions are probably OK.
> >> 
> >> Cheers,
> >> 
> >> Tim.
> > 
> > Thanks for you comments. Attached is my updated patch.
> > 
> > Signed-off-by: Christoph Egger <Christoph.Egger@xxxxxxx>
> 

-- 
Tim Deegan <Tim.Deegan@xxxxxxxxxx>
Principal Software Engineer, XenServer Engineering
Citrix Systems UK Ltd.  (Company #02937203, SL9 0BG)

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel

<Prev in Thread] Current Thread [Next in Thread>