From mboxrd@z Thu Jan  1 00:00:00 1970
From: Radim =?utf-8?B?S3LEjW3DocWZ?= <rkrcmar@redhat.com>
Subject: Re: [PATCH] KVM: x86: call irq notifiers with directed EOI
Date: Fri, 20 Mar 2015 15:43:02 +0100
Message-ID: <20150320144302.GA14772@potion.brq.redhat.com>
References: <1426703902-16818-1-git-send-email-rkrcmar@redhat.com>
 <20150319214449.GA4264@amt.cnet>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: QUOTED-PRINTABLE
Cc: linux-kernel@vger.kernel.org, kvm@vger.kernel.org,
	Paolo Bonzini <pbonzini@redhat.com>, chao.zhou@intel.com
To: Marcelo Tosatti <mtosatti@redhat.com>
Return-path: <linux-kernel-owner@vger.kernel.org>
Content-Disposition: inline
In-Reply-To: <20150319214449.GA4264@amt.cnet>
Sender: linux-kernel-owner@vger.kernel.org
List-Id: kvm.vger.kernel.org

2015-03-19 18:44-0300, Marcelo Tosatti:
> On Wed, Mar 18, 2015 at 07:38:22PM +0100, Radim Kr=C4=8Dm=C3=A1=C5=99=
 wrote:
> > kvm_ioapic_update_eoi() wasn't called if directed EOI was enabled.
> > We need to do that for irq notifiers.  (Like with edge interrupts.)
> >=20
> > Fix it by skipping EOI broadcast only.
> >=20
> > Bug: https://bugzilla.kernel.org/show_bug.cgi?id=3D82211
> > Signed-off-by: Radim Kr=C4=8Dm=C3=A1=C5=99 <rkrcmar@redhat.com>
> > ---
> > diff --git a/arch/x86/kvm/ioapic.c b/arch/x86/kvm/ioapic.c
> > @@ -443,7 +444,8 @@ static void __kvm_ioapic_update_eoi(struct kvm_=
vcpu *vcpu,
> > -		if (trigger_mode !=3D IOAPIC_LEVEL_TRIG)
> > +		if (trigger_mode !=3D IOAPIC_LEVEL_TRIG ||
> > +		    kvm_apic_get_reg(apic, APIC_SPIV) & APIC_SPIV_DIRECTED_EOI)
> >  			continue;
>=20
> Don't you have to handle kvm_ioapic_eoi_inject_work as well?

It works without that: ent->fields.remote_irr =3D=3D 1, thus
kvm_ioapic_eoi_inject_work() will do nothing.
Adding a check would be better for clarity, though.

We could add the EOI register (implement IO-APIC version 0x20), because
kernels are forced to do ugly hacks otherwise (switching to
edge-triggered mode and back).
We also clear remote_irr on a different occasion (just a write to
ioreg).

I'll take a closer look at the second one.

> >  		ASSERT(ent->fields.trig_mode =3D=3D IOAPIC_LEVEL_TRIG);
>=20
> This assert can now fail?

I think it can't (nothing changed), but that is how asserts should be.
It checks a different variable than the condition above.
('trigger_mode' is sourced from APIC_TMR, which should correctly match
 'ent->fields.trig_mode'.)

The assert would be more useful before 'continue;', and modified:
  ASSERT(ent->fields.trig_mode =3D=3D trigger_mode)

Thanks for the review, I'll incorporate the your comments to v2.