From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paolo Bonzini Subject: Re: [PATCH] KVM: SVM: fix interrupt injection (apic->isr_count always 0) Date: Thu, 26 Feb 2015 15:46:36 +0100 Message-ID: <54EF31CC.6070003@redhat.com> References: <20150225160056.GF3226@pd.tnic> <20150225194140.GA9777@potion.brq.redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Tiejun Chen , kvm ML , lkml To: =?UTF-8?B?UmFkaW0gS3LEjW3DocWZ?= , Borislav Petkov Return-path: In-Reply-To: <20150225194140.GA9777@potion.brq.redhat.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: kvm.vger.kernel.org On 25/02/2015 20:41, Radim Kr=C4=8Dm=C3=A1=C5=99 wrote: > 2015-02-25 17:00+0100, Borislav Petkov: >> Hi, >> >> commit in $Subject breaks my kvm guest on AMD host, causing it to do= the >> following below. Mouse doesn't work anymore in the guest, network is >> gone too. >> >> Reverting it fixes the issue. >=20 > Thanks, >=20 > this patch should fix it. >=20 > ---8<--- > In commit b4eef9b36db4, we started to use hwapic_isr_update() !=3D NU= LL > instead of kvm_apic_vid_enabled(vcpu->kvm). This didn't work because > SVM had it defined and "apicv" path in apic_{set,clear}_isr() does no= t > change apic->isr_count, because it should always be 1. The initial > value of apic->isr_count was based on kvm_apic_vid_enabled(vcpu->kvm)= , > which is always 0 for SVM, so KVM could have injected interrupts when= it > shouldn't. >=20 > Fix it by setting SVM's hwapic_isr_update to NULL and make the initia= l > isr_count depend on hwapic_isr_update() for good measure. >=20 > Fixes: b4eef9b36db4 ("kvm: x86: vmx: NULL out hwapic_isr_update() in = case of !enable_apicv") > Reported-by: Borislav Petkov > Signed-off-by: Radim Kr=C4=8Dm=C3=A1=C5=99 > --- > arch/x86/kvm/lapic.c | 4 ++-- > arch/x86/kvm/svm.c | 7 +------ > 2 files changed, 3 insertions(+), 8 deletions(-) >=20 > diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c > index e55b5fc344eb..bd4e34de24c7 100644 > --- a/arch/x86/kvm/lapic.c > +++ b/arch/x86/kvm/lapic.c > @@ -1572,7 +1572,7 @@ void kvm_lapic_reset(struct kvm_vcpu *vcpu) > apic_set_reg(apic, APIC_TMR + 0x10 * i, 0); > } > apic->irr_pending =3D kvm_apic_vid_enabled(vcpu->kvm); > - apic->isr_count =3D kvm_apic_vid_enabled(vcpu->kvm); > + apic->isr_count =3D kvm_x86_ops->hwapic_isr_update ? 1 : 0; > apic->highest_isr_cache =3D -1; > update_divide_count(apic); > atomic_set(&apic->lapic_timer.pending, 0); > @@ -1782,7 +1782,7 @@ void kvm_apic_post_state_restore(struct kvm_vcp= u *vcpu, > update_divide_count(apic); > start_apic_timer(apic); > apic->irr_pending =3D true; > - apic->isr_count =3D kvm_apic_vid_enabled(vcpu->kvm) ? > + apic->isr_count =3D kvm_x86_ops->hwapic_isr_update ? > 1 : count_vectors(apic->regs + APIC_ISR); > apic->highest_isr_cache =3D -1; > if (kvm_x86_ops->hwapic_irr_update) > diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c > index d319e0c24758..54c7b36ad12d 100644 > --- a/arch/x86/kvm/svm.c > +++ b/arch/x86/kvm/svm.c > @@ -3649,11 +3649,6 @@ static void svm_load_eoi_exitmap(struct kvm_vc= pu *vcpu, u64 *eoi_exit_bitmap) > return; > } > =20 > -static void svm_hwapic_isr_update(struct kvm *kvm, int isr) > -{ > - return; > -} > - > static void svm_sync_pir_to_irr(struct kvm_vcpu *vcpu) > { > return; > @@ -4403,7 +4398,7 @@ static struct kvm_x86_ops svm_x86_ops =3D { > .set_virtual_x2apic_mode =3D svm_set_virtual_x2apic_mode, > .vm_has_apicv =3D svm_vm_has_apicv, > .load_eoi_exitmap =3D svm_load_eoi_exitmap, > - .hwapic_isr_update =3D svm_hwapic_isr_update, > + .hwapic_isr_update =3D NULL, Please remove the line altogether. Paolo