Re: [Xen-devel] [for-4.7] x86/emulate: synchronize LOCKed instruction emulation

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 27/04/2016 07:25, Jan Beulich wrote:
>>>> On 26.04.16 at 19:39, <andrew.cooper3@xxxxxxxxxx> wrote:
>> On 26/04/16 18:23, Razvan Cojocaru wrote:
>>> Regarding this version of the patch, Jan has asked for more information
>>> on the performance impact, but I'm not sure how to obtain it in a
>>> rigorous manner. If it is decided that a version of this patch is
>>> desirable, I can go on fixing the issues we've found and address the
>>> comments we've had so far and submit a new version.
>> XenServer did performance testing.  No observable impact for normal VM
>> workloads (which is to be expected, as an OS wouldn't normally LOCK the
>> instructions it uses for MMIO).  The per-cpu rwlocks have ~0 overhead
>> when the lock isn't held for writing.
> So how about PV guests doing locked page table updates, or the
> impact on log-dirty mode?

As I sad - no observable change.

Emulated pagetable writes are already a slowpath for PV guests, so
avoided  (Hypercalls are several times faster).  Logdirty doesn't
directly cause any instructions to be emulated.  For hap guests, the
EPT/NPT violation is resolved and the guest re-entered.  For PV guests
and shadow hvm, we do get emulated pagetable writes, but how many of
those are actually locked?  The majority in Xen at least and just
explicit-width mov's.

~Andrew


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel