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;
Subject: Re: [PATCH 1/2] KVM: X86: Single target IPI fastpath
To:     Wanpeng Li <kernellwp@gmail.com>, linux-kernel@vger.kernel.org,
        kvm@vger.kernel.org
Cc:     =?UTF-8?B?UmFkaW0gS3LEjW3DocWZ?= <rkrcmar@redhat.com>,
        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>
References: <1573283135-5502-1-git-send-email-wanpengli@tencent.com>
From:   Paolo Bonzini <pbonzini@redhat.com>
Message-ID: <6c2c7bbb-39f4-2a77-632e-7730e9887fc5@redhat.com>
Date:   Thu, 14 Nov 2019 12:58:56 +0100
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101
 Thunderbird/60.8.0
MIME-Version: 1.0
In-Reply-To: <1573283135-5502-1-git-send-email-wanpengli@tencent.com>
Content-Language: en-US
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Ok, it's not _so_ ugly after all.

> ---
>  arch/x86/kvm/vmx/vmx.c   | 39 +++++++++++++++++++++++++++++++++++++--
>  include/linux/kvm_host.h |  1 +
>  2 files changed, 38 insertions(+), 2 deletions(-)
>=20
> diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
> index 5d21a4a..5c67061 100644
> --- a/arch/x86/kvm/vmx/vmx.c
> +++ b/arch/x86/kvm/vmx/vmx.c
> @@ -5924,7 +5924,9 @@ static int vmx_handle_exit(struct kvm_vcpu *vcpu)
>  =09=09}
>  =09}
> =20
> -=09if (exit_reason < kvm_vmx_max_exit_handlers
> +=09if (vcpu->fast_vmexit)
> +=09=09return 1;
> +=09else if (exit_reason < kvm_vmx_max_exit_handlers

Instead of a separate vcpu->fast_vmexit, perhaps you can set exit_reason
to vmx->exit_reason to -1 if the fast path succeeds.

> +=09=09=09if (ret =3D=3D 0)
> +=09=09=09=09ret =3D kvm_skip_emulated_instruction(vcpu);

Please move the "kvm_skip_emulated_instruction(vcpu)" to
vmx_handle_exit, so that this basically is

#define EXIT_REASON_NEED_SKIP_EMULATED_INSN -1

=09if (ret =3D=3D 0)
=09=09vcpu->exit_reason =3D EXIT_REASON_NEED_SKIP_EMULATED_INSN;

and handle_ipi_fastpath can return void.

Thanks,

Paolo

> +=09=09};
> +=09};
> +
> +=09return ret;
> +}
> +
>  static void vmx_vcpu_run(struct kvm_vcpu *vcpu)
>  {
>  =09struct vcpu_vmx *vmx =3D to_vmx(vcpu);
> @@ -6615,6 +6645,12 @@ static void vmx_vcpu_run(struct kvm_vcpu *vcpu)
>  =09=09=09=09  | (1 << VCPU_EXREG_CR3));
>  =09vcpu->arch.regs_dirty =3D 0;
> =20
> +=09vmx->exit_reason =3D vmx->fail ? 0xdead : vmcs_read32(VM_EXIT_REASON)=
;
> +=09vcpu->fast_vmexit =3D false;
> +=09if (!is_guest_mode(vcpu) &&
> +=09=09vmx->exit_reason =3D=3D EXIT_REASON_MSR_WRITE)
> +=09=09vcpu->fast_vmexit =3D handle_ipi_fastpath(vcpu);

This should be done later, at least after kvm_put_guest_xcr0, because
running with partially-loaded guest state is harder to audit.  The best
place to put it actually is right after the existing vmx->exit_reason
assignment, where we already handle EXIT_REASON_MCE_DURING_VMENTRY.

>  =09pt_guest_exit(vmx);
> =20
>  =09/*
> @@ -6634,7 +6670,6 @@ static void vmx_vcpu_run(struct kvm_vcpu *vcpu)
>  =09vmx->nested.nested_run_pending =3D 0;
>  =09vmx->idt_vectoring_info =3D 0;
> =20
> -=09vmx->exit_reason =3D vmx->fail ? 0xdead : vmcs_read32(VM_EXIT_REASON)=
;
>  =09if ((u16)vmx->exit_reason =3D=3D EXIT_REASON_MCE_DURING_VMENTRY)
>  =09=09kvm_machine_check();
> =20
> diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
> index 719fc3e..7a7358b 100644
> --- a/include/linux/kvm_host.h
> +++ b/include/linux/kvm_host.h
> @@ -319,6 +319,7 @@ struct kvm_vcpu {
>  #endif
>  =09bool preempted;
>  =09bool ready;
> +=09bool fast_vmexit;
>  =09struct kvm_vcpu_arch arch;
>  =09struct dentry *debugfs_dentry;
>  };
>=20