Re: [Xen-devel] CR4.TSD in guest kernel vs userland

>>> Keir Fraser <keir.fraser@xxxxxxxxxxxxx> 22.09.09 08:34 >>>
>It's only likely to be reasonable for PV 64-bit guests, which transition via
>the hypervisor already for guest-kernel <-> guest-app transitions (i.e.,
>those transition times suck already).

But still, I don't think adding a serializing instruction that likely (I don't
know of officially available numbers) takes hundreds of clocks is a good
thing to do.

Jan

>On 21/09/2009 23:49, "Dan Magenheimer" <dan.magenheimer@xxxxxxxxxx> wrote:
>> I was fretting about the overhead of emulating TSC in
>> guest kernels.  Is it possible (and relatively inexpensive)
>> to change CR4.TSD on every transition between guest
>> kernel and userland?  I suspect not, but thought
>> I'd check.  If so, what about HVM vs PV and 32-bit
>> vs 64-bit guest.



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

<Prev in Thread]	Current Thread	[Next in Thread>
[Xen-devel] CR4.TSD in guest kernel vs userland, Dan Magenheimer Re: [Xen-devel] CR4.TSD in guest kernel vs userland, Keir Fraser Re: [Xen-devel] CR4.TSD in guest kernel vs userland, Jan Beulich <=

Previous by Date:

[Xen-devel] Re: [PATCH] EPT: Flush running cpus, add mask to flush when scheduled in, Jan Beulich

Next by Date:

Re: [Xen-devel] rdtscP and xen (and maybe the app-tsc answer I've been looking for), Jan Beulich

Previous by Thread:

Re: [Xen-devel] CR4.TSD in guest kernel vs userland, Keir Fraser

Next by Thread:

Re: [Xen-devel] Can the qemu-xen git repository be moved to a 'git://' protocol? It can save a lot of time..., Brendan Cully

Indexes:

[Date] [Thread] [Top] [All Lists]