* About MONITOR/MWAIT in i386 CPU model
@ 2019-11-13 8:42 Tao Xu
2019-11-13 22:47 ` Eduardo Habkost
0 siblings, 1 reply; 3+ messages in thread
From: Tao Xu @ 2019-11-13 8:42 UTC (permalink / raw)
To: ehabkost; +Cc: qemu-devel@nongnu.org, Guo, Xuelian
Hi Eduardo，
After kvm use "-overcommit cpu-pm=on" to expose MONITOR/MWAIT
(commit id 6f131f13e68d648a8e4f083c667ab1acd88ce4cd), the MONITOR/MWAIT
feature in CPU model (phenom core2duo coreduo n270 Opteron_G3 EPYC
Snowridge Denverton) may be unused. For example, when we boot a guest
with Denverton cpu model, guest cannot detect MONITOR and boot with no
warning. Should we remove this feature from some CPU model?
Tested by Guo, Xuelian <xuelian.guo@intel.com>
Tao Xu
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: About MONITOR/MWAIT in i386 CPU model
2019-11-13 8:42 About MONITOR/MWAIT in i386 CPU model Tao Xu
@ 2019-11-13 22:47 ` Eduardo Habkost
2019-11-14 0:50 ` Tao Xu
0 siblings, 1 reply; 3+ messages in thread
From: Eduardo Habkost @ 2019-11-13 22:47 UTC (permalink / raw)
To: Tao Xu; +Cc: Lai, Paul C, qemu-devel@nongnu.org, Guo, Xuelian
On Wed, Nov 13, 2019 at 04:42:25PM +0800, Tao Xu wrote:
> Hi Eduardo，
>
> After kvm use "-overcommit cpu-pm=on" to expose MONITOR/MWAIT
> (commit id 6f131f13e68d648a8e4f083c667ab1acd88ce4cd), the MONITOR/MWAIT
> feature in CPU model (phenom core2duo coreduo n270 Opteron_G3 EPYC Snowridge
> Denverton) may be unused. For example, when we boot a guest with Denverton
> cpu model, guest cannot detect MONITOR and boot with no warning. Should we
> remove this feature from some CPU model?
Good catch, thanks!
Yes, we should remove them from Opteron_G3, EPYC, Snowridge, and
Denverton, at least. The other older CPU models can be left
alone: they are more useful for use with TCG than with KVM, and
TCG supports MONITOR/MWAIT.
I would like to understand why this wasn't detected during
testing by Intel. I suggest always testing CPU models using the
"enforce" flag to make sure warnings don't go unnoticed.
>
> Tested by Guo, Xuelian <xuelian.guo@intel.com>
>
> Tao Xu
>
--
Eduardo
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: About MONITOR/MWAIT in i386 CPU model
2019-11-13 22:47 ` Eduardo Habkost
@ 2019-11-14 0:50 ` Tao Xu
0 siblings, 0 replies; 3+ messages in thread
From: Tao Xu @ 2019-11-14 0:50 UTC (permalink / raw)
To: Eduardo Habkost; +Cc: Lai, Paul C, qemu-devel@nongnu.org, Guo, Xuelian
On 11/14/2019 6:47 AM, Eduardo Habkost wrote:
> On Wed, Nov 13, 2019 at 04:42:25PM +0800, Tao Xu wrote:
>> Hi Eduardo，
>>
>> After kvm use "-overcommit cpu-pm=on" to expose MONITOR/MWAIT
>> (commit id 6f131f13e68d648a8e4f083c667ab1acd88ce4cd), the MONITOR/MWAIT
>> feature in CPU model (phenom core2duo coreduo n270 Opteron_G3 EPYC Snowridge
>> Denverton) may be unused. For example, when we boot a guest with Denverton
>> cpu model, guest cannot detect MONITOR and boot with no warning. Should we
>> remove this feature from some CPU model?
>
> Good catch, thanks!
>
> Yes, we should remove them from Opteron_G3, EPYC, Snowridge, and
> Denverton, at least. The other older CPU models can be left
> alone: they are more useful for use with TCG than with KVM, and
> TCG supports MONITOR/MWAIT.
>
> I would like to understand why this wasn't detected during
> testing by Intel. I suggest always testing CPU models using the
> "enforce" flag to make sure warnings don't go unnoticed.
>
OK we will improve the testing.
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2019-11-14 0:55 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2019-11-13 8:42 About MONITOR/MWAIT in i386 CPU model Tao Xu
2019-11-13 22:47 ` Eduardo Habkost
2019-11-14 0:50 ` Tao Xu
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).