Re: [Xen-devel] domU can't start, Non-priv warnings

[Jerone Young <jyoung5@xxxxxxxxxx>]

On Wed, 2005-08-17 at 06:59 -0500, Jerone Young wrote:
> On 8/17/05, Petersson, Mats <mats.petersson@xxxxxxx> wrote:
> > 
> > 
> > > -----Original Message-----
> > > From: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx
> > > [mailto:xen-devel-bounces@xxxxxxxxxxxxxxxxxxx] On Behalf Of
> > > Jerone Young
> > > Sent: 16 August 2005 20:33
> > > To: Andrew Theurer
> > > Cc: xen-devel@xxxxxxxxxxxxxxxxxxx
> > > Subject: Re: [Xen-devel] domU can't start, Non-priv warnings
> > >
> > > On Tue, 2005-08-16 at 14:19 -0500, Andrew Theurer wrote:
> > > > Can't seem to fully boot a domU on either EM64T or Opteron
> > > with x86_64
> > > > xen-unstable.  Both systems' report this from xen:
> > > >
> > > > (XEN) (file=traps.c, line=872) Non-priv domain attempted
> > > > WRMSR(00000000c0000100.
> > > > (XEN) (file=traps.c, line=872) Non-priv domain attempted
> > > > WRMSR(00000000c0000102.
> > > > (XEN) (file=traps.c, line=880) Non-priv domain attempted
> > > > RDMSR(00000000c0000080.
> > > >
> > > > They boot partially, ending with:
> > > >
> > > > xen_mem: Initialising balloon driver.
> > > > Grant table initialized
> > > > IA32 emulation $Id: sys_ia32.c,v 1.32 2002/03/24 13:02:28 ak Exp $
> > > > audit: initializing netlink socket (disabled)
> > > > audit(1124251914.348:0): initialized
> > > > VFS: Disk quotas dquot_6.5.1
> > > > Dquot-cache hash table entries: 512 (order 0, 4096 bytes)
> > > >
> > > >
> > > > This is with the default kernel configs xen-unstable
> > > provides.  NX is
> > > > enabled on the EM64T.  I could not find a similar BIOS option for
> > > > Opteron.  Even through they are "stuck", cpu time for these domains
> > > > increments steadily.  Any ideas?
> > >
> > > Checking it out...looks like Xen is pretty screwed on Opteron
> > > machines at the moment (at least on my dual Opteron
> > > workstation). There seems to be some timer issue...takes
> > > forever fro the machine to boot up.
> > >
> > > Oh..AMD does not have an option for the NX bit...the cpu
> > > always can understand it.
> > 
> > That is incorrect. AMD's 64-bit processors use NXE, bit 11, in EFER (MSR
> > 0xC0000080) to determine if NX is "in use" or not.
> 
> You apparently miss understand this. Intel CPUs have an option in the
> bios where if bit 11 is set it has no idea what to do with it. Where
> as on AMD 64-bit CPUs if bit 11 is set then it always does. My big
> complaint is why Intel made this possible in the first place.
Wow I just read this ...don't write emails really early in the
morning :-)

Ok the problem is that in Xen x86-64 the NX bit is set on memory pages
and your Intel EM64T CPU cannot understand this....xen go boom! So there
is an option in the bios for Intel EM64T CPUs that is called "disable
execute bit" (at least on IBM machines). This name is a bit
misleading...it enables the NX feature on the CPU (does not activate
it). So what this option does is allow the activation of NX bit
enforcement on the CPU as well as allow the CPU to understand the NX bit
set on memory pages. Where as on AMD K8 CPUs always have the ability to
have the NX bit feature there used and understands the bit set on memory
pages.

Much better..

> 
> > However, I'm not aware of any way to disable the ability to set this bit
> > in the MSR, if that's what you mean by "AMD does not have an option for
> > the NX bit".
> > 
> > --
> > Mats
> > > >
> > > > -Andrew
> > > >
> > > > _______________________________________________
> > > > Xen-devel mailing list
> > > > Xen-devel@xxxxxxxxxxxxxxxxxxx
> > > > http://lists.xensource.com/xen-devel
> > > >
> > > --
> > > Jerone Young
> > > IBM Linux Technology Center
> > > jyoung5@xxxxxxxxxx
> > > 512-838-1157 (T/L: 678-1157)
> > >
> > >
> > > _______________________________________________
> > > 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
> >
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxxxxxxxx
> http://lists.xensource.com/xen-devel
> 
-- 
Jerone Young
IBM Linux Technology Center
jyoung5@xxxxxxxxxx
512-838-1157 (T/L: 678-1157)


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