Tim,
I found that your small patch (changeset 15291) caused failure with
Windows 2003 64Bit and Vista 64Bit under nested paging. It turned out
that intercepting CR0 read should not be disabled. Please see the change
below. Other guests seem running well.
Thanks,
-Wei
========Replace
vmcb->cr_intercepts &= ~(CR_INTERCEPT_CR0_READ
|CR_INTERCEPT_CR3_READ
|CR_INTERCEPT_CR4_READ);
with
vmcb->cr_intercepts &= ~(CR_INTERCEPT_CR3_READ
|CR_INTERCEPT_CR4_READ);
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel
At 11:45 -0500 on 13 Jul (1184327144), Huang2, Wei wrote:> I found that your small patch (changeset 15291) caused failure with > Windows 2003 64Bit and Vista 64Bit under nested paging. It turned out > that intercepting CR0 read should not be disabled.Why not? Is the guest''s CR0 not synchronised with cpu_shadow_cr0? Or is there some other reason? Cheers, Tim -- Tim Deegan <Tim.Deegan@xensource.com>, XenSource UK Limited Registered office c/o EC2Y 5EB, UK; company number 05334508 _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Yes, I think it is related to the synchronization between vmcb->cr0 and cpu_shadow_cr0. Need more investigation... Maybe these two guests do internal consistency checking and bail out whenever they find errors. Not sure yet. -Wei Tim Deegan wrote:> At 11:45 -0500 on 13 Jul (1184327144), Huang2, Wei wrote: >> I found that your small patch (changeset 15291) caused failure with >> Windows 2003 64Bit and Vista 64Bit under nested paging. It turned out >> that intercepting CR0 read should not be disabled. > > Why not? Is the guest''s CR0 not synchronised with cpu_shadow_cr0? > Or is there some other reason? > > Cheers, > > Tim_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
At 11:58 -0500 on 13 Jul (1184327893), Huang2, Wei wrote:> Yes, I think it is related to the synchronization between vmcb->cr0 and > cpu_shadow_cr0. Need more investigation... Maybe these two guests do > internal consistency checking and bail out whenever they find errors. > Not sure yet.Just had a look -- our lazy FPU code lets CR0.TS get out of sync with the guest''s one, so you''re right, we do need to intercept read-cr0. I''ll revert that change. Thanks, Tim. -- Tim Deegan <Tim.Deegan@xensource.com>, XenSource UK Limited Registered office c/o EC2Y 5EB, UK; company number 05334508 _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel