From mboxrd@z Thu Jan  1 00:00:00 1970
From: Kashyap Chamarthy <kchamart@redhat.com>
Subject: Re: [nVMX] With 3.20.0-0.rc0.git5.1 on L0, booting L2 guest results
 in L1 *rebooting*
Date: Thu, 19 Feb 2015 22:10:11 +0100
Message-ID: <20150219211011.GE28728@tesla.redhat.com>
References: <20150216204013.GI21838@tesla.redhat.com>
 <54E2D966.9070706@siemens.com>
 <20150217112426.GL21838@tesla.redhat.com>
 <54E4C0FD.4090906@redhat.com>
 <20150219120719.GJ11686@tesla.redhat.com>
 <20150219150159.GA19057@potion.brq.redhat.com>
 <20150219160221.GB19057@potion.brq.redhat.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: QUOTED-PRINTABLE
Cc: Paolo Bonzini <pbonzini@redhat.com>,
	Jan Kiszka <jan.kiszka@siemens.com>, kvm@vger.kernel.org,
	dgilbert@redhat.com
To: Radim =?utf-8?B?S3LEjW3DocWZ?= <rkrcmar@redhat.com>
Return-path: <kvm-owner@vger.kernel.org>
Received: from mx1.redhat.com ([209.132.183.28]:52860 "EHLO mx1.redhat.com"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S1752595AbbBSVK0 (ORCPT <rfc822;kvm@vger.kernel.org>);
	Thu, 19 Feb 2015 16:10:26 -0500
Content-Disposition: inline
In-Reply-To: <20150219160221.GB19057@potion.brq.redhat.com>
Sender: kvm-owner@vger.kernel.org
List-ID: <kvm.vger.kernel.org>

On Thu, Feb 19, 2015 at 05:02:22PM +0100, Radim Kr=C4=8Dm=C3=A1=C5=99 w=
rote:
> 2015-02-19 16:01+0100, Radim Kr=C4=8Dm=C3=A1=C5=99:
> > 2015-02-19 13:07+0100, Kashyap Chamarthy:
> > 5f3d5799974b8 KVM: nVMX: Rework event injection and recovery:
> >   This concept is based on the rule that a pending vmlaunch/vmresum=
e is
> >   not canceled. Otherwise, we would risk to lose injected events or=
 leak
> >   them into the wrong queues. Encode this rule via a WARN_ON_ONCE a=
t the
> >   entry of nested_vmx_vmexit.
> >=20
> > I wonder if we have broken the invariant since 3.9 ...
>=20
> e011c663b9c786d115c0f45e5b0bfae0c39428d4
> KVM: nVMX: Check all exceptions for intercept during delivery to L2
>=20
>   All exceptions should be checked for intercept during delivery to L=
2,
>   but we check only #PF currently. Drop nested_run_pending while we a=
re
>   at it since exception cannot be injected during vmentry anyway.
>=20
> The last sentence is not true.
>=20
> Can you try if the following patch works?

Sure, will test a Kernel built with the below patch and report back.

Thanks for taking a look.

--
/kashyap


> (I know little about nested, so it might be introducing another bug.)
>=20
> Thanks.
>=20
> ---8<---
> KVM: nVMX: fix L2 to L1 interrupt leak
>=20
> When vmx->nested.nested_run_pending is set, we aren't expected to exi=
t
> to L1, but nested_vmx_check_exception() could, since e011c663b9c7.
> Prevent that.
>=20
> Fixes: e011c663b9c7 ("Check all exceptions for intercept during deliv=
ery to L2")
> Signed-off-by: Radim Kr=C4=8Dm=C3=A1=C5=99 <rkrcmar@redhat.com>
> ---
>  arch/x86/kvm/vmx.c | 3 +++
>  1 file changed, 3 insertions(+)
>=20
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index 3f73bfad0349..389166a1b79a 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -2098,6 +2098,9 @@ static int nested_vmx_check_exception(struct kv=
m_vcpu *vcpu, unsigned nr)
>  {
>  	struct vmcs12 *vmcs12 =3D get_vmcs12(vcpu);
> =20
> +	if (to_vmx(vcpu)->nested.nested_run_pending)
> +		return 0;
> +
>  	if (!(vmcs12->exception_bitmap & (1u << nr)))
>  		return 0;