Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <1587468026-15753-1-git-send-email-wanpengli@tencent.com>
 <1587468026-15753-3-git-send-email-wanpengli@tencent.com> <68eb0e46-4c2a-0292-3dfa-db2ae2b2b13d@redhat.com>
In-Reply-To: <68eb0e46-4c2a-0292-3dfa-db2ae2b2b13d@redhat.com>
From:   Wanpeng Li <kernellwp@gmail.com>
Date:   Wed, 22 Apr 2020 09:01:52 +0800
Message-ID: <CANRm+CwXhe+TdB8JpQ78qR-sO6FB_cMjqHxj5fSyKEMPkMVm8g@mail.gmail.com>
Subject: Re: [PATCH 2/2] KVM: VMX: Handle preemption timer fastpath
To:     Paolo Bonzini <pbonzini@redhat.com>
Cc:     LKML <linux-kernel@vger.kernel.org>, kvm <kvm@vger.kernel.org>,
        Sean Christopherson <sean.j.christopherson@intel.com>,
        Vitaly Kuznetsov <vkuznets@redhat.com>,
        Wanpeng Li <wanpengli@tencent.com>,
        Jim Mattson <jmattson@google.com>,
        Joerg Roedel <joro@8bytes.org>,
        Haiwei Li <lihaiwei@tencent.com>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Tue, 21 Apr 2020 at 19:40, Paolo Bonzini <pbonzini@redhat.com> wrote:
>
> On 21/04/20 13:20, Wanpeng Li wrote:
> > +
> > +     if (!vmx->req_immediate_exit &&
> > +             !unlikely(vmx->loaded_vmcs->hv_timer_soft_disabled)) {
> > +             if (!vmx_interrupt_allowed(vcpu) ||

For non-APICv case, we need to request interrupt-window.

> > +                     !apic_lvtt_tscdeadline(apic) ||

Now just add fastpath for tscdeadline mode.

> > +                     vmx->rmode.vm86_active ||
> > +                     is_smm(vcpu) ||
> > +                     !kvm_apic_hw_enabled(apic))

These stuff can be removed, kvm_apic_hw_enable() is check in
vmx_fast_deliver_interrupt().

> > +                     return EXIT_FASTPATH_NONE;
> > +
> > +             if (!apic->lapic_timer.hv_timer_in_use)
> > +                     return EXIT_FASTPATH_CONT_RUN;
> > +
> > +             WARN_ON(swait_active(&vcpu->wq));
> > +             vmx_cancel_hv_timer(vcpu);
> > +             apic->lapic_timer.hv_timer_in_use = false;
> > +
> > +             if (atomic_read(&apic->lapic_timer.pending))
> > +                     return EXIT_FASTPATH_CONT_RUN;

Other two checks are the same in kvm_lapic_expired_hv_timer().

    Wanpeng

> > +
> > +             ktimer->expired_tscdeadline = ktimer->tscdeadline;
> > +             vmx_fast_deliver_interrupt(vcpu);
> > +             ktimer->tscdeadline = 0;
> > +             return EXIT_FASTPATH_CONT_RUN;
> > +     }
> > +
>
> Can you explain all the checks you have here, and why you need something
> more complex than apic_timer_expired (possibly by adding some
> optimizations to kvm_apic_local_deliver)?  This code is impossible to
> maintain.
>
> Paolo
>