Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <1559178307-6835-1-git-send-email-wanpengli@tencent.com>
 <20190610143420.GA6594@flask> <20190611011100.GB24835@linux.intel.com>
 <CANRm+Cwv5jqxBW=Ss5nkX7kZM3_Y-Ucs66yx5+wN09=W4pUdzA@mail.gmail.com>
 <F136E492-5350-49EE-A856-FBAEDB12FF99@gmail.com> <CANRm+CyZcvuT80ixp9f0FNmjN+rTUtw8MshtBG0Uk4L1B1UjDw@mail.gmail.com>
 <153047ED-75E2-4E70-BC33-C5FF27C08638@gmail.com>
In-Reply-To: <153047ED-75E2-4E70-BC33-C5FF27C08638@gmail.com>
From:   Wanpeng Li <kernellwp@gmail.com>
Date:   Wed, 12 Jun 2019 09:18:51 +0800
Message-ID: <CANRm+Cx6Z=jxLaXqwhBDpVTsKH8mgoo4iC=U8GbAAJz-5gk5ZA@mail.gmail.com>
Subject: Re: [PATCH v3 0/3] KVM: Yield to IPI target if necessary
To:     Nadav Amit <nadav.amit@gmail.com>
Cc:     Sean Christopherson <sean.j.christopherson@intel.com>,
        =?UTF-8?B?UmFkaW0gS3LEjW3DocWZ?= <rkrcmar@redhat.com>,
        LKML <linux-kernel@vger.kernel.org>, kvm <kvm@vger.kernel.org>,
        Paolo Bonzini <pbonzini@redhat.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, 12 Jun 2019 at 00:57, Nadav Amit <nadav.amit@gmail.com> wrote:
>
> > On Jun 11, 2019, at 3:02 AM, Wanpeng Li <kernellwp@gmail.com> wrote:
> >
> > On Tue, 11 Jun 2019 at 09:48, Nadav Amit <nadav.amit@gmail.com> wrote:
> >>> On Jun 10, 2019, at 6:45 PM, Wanpeng Li <kernellwp@gmail.com> wrote:
> >>>
> >>> On Tue, 11 Jun 2019 at 09:11, Sean Christopherson
> >>> <sean.j.christopherson@intel.com> wrote:
> >>>> On Mon, Jun 10, 2019 at 04:34:20PM +0200, Radim Kr=C4=8Dm=C3=A1=C5=
=99 wrote:
> >>>>> 2019-05-30 09:05+0800, Wanpeng Li:
> >>>>>> The idea is from Xen, when sending a call-function IPI-many to vCP=
Us,
> >>>>>> yield if any of the IPI target vCPUs was preempted. 17% performanc=
e
> >>>>>> increasement of ebizzy benchmark can be observed in an over-subscr=
ibe
> >>>>>> environment. (w/ kvm-pv-tlb disabled, testing TLB flush call-funct=
ion
> >>>>>> IPI-many since call-function is not easy to be trigged by userspac=
e
> >>>>>> workload).
> >>>>>
> >>>>> Have you checked if we could gain performance by having the yield a=
s an
> >>>>> extension to our PV IPI call?
> >>>>>
> >>>>> It would allow us to skip the VM entry/exit overhead on the caller.
> >>>>> (The benefit of that might be negligible and it also poses a
> >>>>> complication when splitting the target mask into several PV IPI
> >>>>> hypercalls.)
> >>>>
> >>>> Tangetially related to splitting PV IPI hypercalls, are there any ma=
jor
> >>>> hurdles to supporting shorthand?  Not having to generate the mask fo=
r
> >>>> ->send_IPI_allbutself and ->kvm_send_ipi_all seems like an easy to w=
ay
> >>>> shave cycles for affected flows.
> >>>
> >>> Not sure why shorthand is not used for native x2apic mode.
> >>
> >> Why do you say so? native_send_call_func_ipi() checks if allbutself
> >> shorthand should be used and does so (even though the check can be mor=
e
> >> efficient - I=E2=80=99m looking at that code right now=E2=80=A6)
> >
> > Please continue to follow the apic/x2apic driver. Just apic_flat set
> > APIC_DEST_ALLBUT/APIC_DEST_ALLINC to ICR.
>
> Indeed - I was sure by the name that it does it correctly. That=E2=80=99s=
 stupid.
>
> I=E2=80=99ll add it to the patch-set I am working on (TLB shootdown impro=
vements),
> if you don=E2=80=99t mind.

Original for hotplug cpu safe.
https://lwn.net/Articles/138365/
https://lwn.net/Articles/138368/
Not sure shortcut native support is acceptable, I will play my
kvm_send_ipi_allbutself and kvm_send_ipi_all. :)

Regards,
Wanpeng Li