Re: [RFC PATCH 3/4] KVM: x86: Add EOI exit bitmap inference

From: Steve Rutherford <srutherford@google.com>
To: Paolo Bonzini <pbonzini@redhat.com>
Cc: Jan Kiszka <jan.kiszka@siemens.com>,
	kvm@vger.kernel.org, ahonig@google.com
Subject: Re: [RFC PATCH 3/4] KVM: x86: Add EOI exit bitmap inference
Date: Wed, 13 May 2015 15:21:00 -0700	[thread overview]
Message-ID: <20150513222100.GC23867@google.com> (raw)
In-Reply-To: <555305A5.8060601@redhat.com>

On Wed, May 13, 2015 at 10:04:53AM +0200, Paolo Bonzini wrote:
> 
> 
> On 13/05/2015 08:12, Jan Kiszka wrote:
> >> +void kvm_scan_ioapic_routes(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap)
> >> +{
> >> +	struct kvm *kvm = vcpu->kvm;
> >> +	struct kvm_kernel_irq_routing_entry *entry;
> >> +	struct kvm_irq_routing_table *table;
> >> +	u32 i, nr_rt_entries;
> >> +
> >> +	mutex_lock(&kvm->irq_lock);
> 
> This only needs irq_srcu protection, not irq_lock, so the lookup cost
> becomes much smaller (all CPUs can proceed in parallel).
> 
> You would need to put an smp_mb here, to ensure that irq_routing is read
> after KVM_SCAN_IOAPIC is cleared.  You can introduce
> smb_mb__after_srcu_read_lock in order to elide it.
> 
> The matching memory barrier would be a smp_mb__before_atomic in
> kvm_make_scan_ioapic_request.
> 

Makes sense, I'll update this for the next iteration.

Steve