* Virtualization + CPU w/crypto acceleration instructions
@ 2007-03-12 19:35 Molle Bestefich
2007-03-12 19:44 ` Jeremy Fitzhardinge
` (2 more replies)
0 siblings, 3 replies; 5+ messages in thread
From: Molle Bestefich @ 2007-03-12 19:35 UTC (permalink / raw)
To: xen-users, virtualization
In a virtualized environment running Xen or VMware, can the guest
operating systems utilize hardware crypto acceleration such as the
"PadLock" AES acceleration built into VIA CPUs?
(Wasn't able to find this info in any of ~400 Google hits, so I'm
shooting it out a bit to let as many as possible benefit from the
answer.)
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Virtualization + CPU w/crypto acceleration instructions
2007-03-12 19:35 Virtualization + CPU w/crypto acceleration instructions Molle Bestefich
@ 2007-03-12 19:44 ` Jeremy Fitzhardinge
2007-03-12 19:50 ` Kraska, Joe A (US SSA)
2007-03-12 23:46 ` Zachary Amsden
2 siblings, 0 replies; 5+ messages in thread
From: Jeremy Fitzhardinge @ 2007-03-12 19:44 UTC (permalink / raw)
To: Molle Bestefich; +Cc: virtualization, xen-users
Molle Bestefich wrote:
> In a virtualized environment running Xen or VMware, can the guest
> operating systems utilize hardware crypto acceleration such as the
> "PadLock" AES acceleration built into VIA CPUs?
>
If they're unprivileged instructions, then they should just work. If
its done via MSRs or some other ring0 mechanism, it might be more tricky.
J
^ permalink raw reply [flat|nested] 5+ messages in thread
* RE: Virtualization + CPU w/crypto acceleration instructions
2007-03-12 19:35 Virtualization + CPU w/crypto acceleration instructions Molle Bestefich
2007-03-12 19:44 ` Jeremy Fitzhardinge
@ 2007-03-12 19:50 ` Kraska, Joe A (US SSA)
2007-03-12 23:46 ` Zachary Amsden
2 siblings, 0 replies; 5+ messages in thread
From: Kraska, Joe A (US SSA) @ 2007-03-12 19:50 UTC (permalink / raw)
To: Molle Bestefich, xen-users, virtualization
This seems more like a developer question. You might try the xen-devel
list. Anyway, I believe that all hardware instructions on any x86 cpu
are supposed to be executable by arbitrary assembly. Those that cannot
be properly kernel trapped may experience performance loss, but I
think that's SUPPOSED to be the only limitation.
Would be nice if someone with exact knowledge chipped in, of course.
Joe.
> -----Original Message-----
> From: xen-users-bounces@lists.xensource.com [mailto:xen-users-
> bounces@lists.xensource.com] On Behalf Of Molle Bestefich
> Sent: Monday, March 12, 2007 12:36 PM
> To: xen-users@lists.xensource.com; virtualization@lists.osdl.org
> Subject: [Xen-users] Virtualization + CPU w/crypto acceleration
instructions
>
> In a virtualized environment running Xen or VMware, can the guest
> operating systems utilize hardware crypto acceleration such as the
> "PadLock" AES acceleration built into VIA CPUs?
>
> (Wasn't able to find this info in any of ~400 Google hits, so I'm
> shooting it out a bit to let as many as possible benefit from the
> answer.)
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Virtualization + CPU w/crypto acceleration instructions
2007-03-12 19:35 Virtualization + CPU w/crypto acceleration instructions Molle Bestefich
2007-03-12 19:44 ` Jeremy Fitzhardinge
2007-03-12 19:50 ` Kraska, Joe A (US SSA)
@ 2007-03-12 23:46 ` Zachary Amsden
2007-03-13 0:04 ` Jeremy Fitzhardinge
2 siblings, 1 reply; 5+ messages in thread
From: Zachary Amsden @ 2007-03-12 23:46 UTC (permalink / raw)
To: Molle Bestefich; +Cc: virtualization, xen-users
Molle Bestefich wrote:
> In a virtualized environment running Xen or VMware, can the guest
> operating systems utilize hardware crypto acceleration such as the
> "PadLock" AES acceleration built into VIA CPUs?
VMware doesn't support VIA CPUs, and I think we would have compatibility
issues with these instructions. In general, most hypervisors would need
to support decoding of the new instructions for complete emulation in
various fault and corner cases. In general, I would expect some trouble
just running a hypervisor or VIA CPUs, as the test matrix is so low I
can't point even to one working example. Perhaps someone can correct me
if they have got one up and running.
UML should work fine though.
Zach
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Virtualization + CPU w/crypto acceleration instructions
2007-03-12 23:46 ` Zachary Amsden
@ 2007-03-13 0:04 ` Jeremy Fitzhardinge
0 siblings, 0 replies; 5+ messages in thread
From: Jeremy Fitzhardinge @ 2007-03-13 0:04 UTC (permalink / raw)
To: Zachary Amsden; +Cc: virtualization, Molle Bestefich, xen-users
Zachary Amsden wrote:
> VMware doesn't support VIA CPUs, and I think we would have compatibility
> issues with these instructions. In general, most hypervisors would need
> to support decoding of the new instructions for complete emulation in
> various fault and corner cases.
Using a crypto instruction to update a pagetable would be... interesting ;)
J
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2007-03-13 0:04 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2007-03-12 19:35 Virtualization + CPU w/crypto acceleration instructions Molle Bestefich
2007-03-12 19:44 ` Jeremy Fitzhardinge
2007-03-12 19:50 ` Kraska, Joe A (US SSA)
2007-03-12 23:46 ` Zachary Amsden
2007-03-13 0:04 ` Jeremy Fitzhardinge
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).