RE: [Xen-devel] HVM PAE >4Gig on PAE HV support/failure

["Woller, Thomas" <thomas.woller@xxxxxxx>]

> Should still work with up to 8GB, by design.
Ok thanks. We'll look at it then and figure out which c/s is the
culprit. 
We are getting below error in p2m with shadow (64b hv works fine).  
HVM Pae on pae hv w/ Hap on AMD-V returns the intended error >4Gig -
without below error, (and logs xm dmesg properly), so the hap= guest
config option seems to be working as intended.

... 
(XEN) p2mdebug: guest_physmap_add_entry(): adding gfn=0xeffff
mfn=0x3e4c3
(XEN) p2mdebug: guest_physmap_add_entry(): adding gfn=0x100000
mfn=0x3e4c2
(XEN) p2mdebug: p2m_find_entry(): gfn=0x100000 out of range
(gfn_remainder=0x100000 shift=18 index=0x4 max=0x4)
(XEN) pg error: guest_physmap_add_entry(): p2m EINVAL 2

tom

> -----Original Message-----
> From: Keir Fraser [mailto:keir.fraser@xxxxxxxxxxxxx] 
> Sent: Thursday, March 13, 2008 10:43 AM
> To: Woller, Thomas; xen-devel@xxxxxxxxxxxxxxxxxxx
> Subject: Re: [Xen-devel] HVM PAE >4Gig on PAE HV support/failure
> 
> Should still work with up to 8GB, by design.
> 
>  -- Keir
> 
> On 13/3/08 15:19, "Woller, Thomas" <thomas.woller@xxxxxxx> wrote:
> 
> > Our testing with 17193 indicates that pae HVM guests (using shadow
> > paging) on pae hv fail if attempting to use >4Gig of memory (3840).
> > Is this change on purpose?  Or should HVM shadow guests (hap=0 in 
> > guest
> > config) be allowed to use up to 8Gig as previously supported.
> > 
> > We have only tested on AMD-V, so don't know about VT boxes.
> > Thanks for any info,
> > tom
> > 
> > 
> > _______________________________________________
> > Xen-devel mailing list
> > Xen-devel@xxxxxxxxxxxxxxxxxxx
> > http://lists.xensource.com/xen-devel
> 
> 
> 
> 


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