From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Michael S. Tsirkin" Subject: [PATCH RFC] kvm: optimize out smp_mb using srcu_read_unlock Date: Wed, 30 Oct 2013 21:09:29 +0200 Message-ID: <20131030190929.GA7153@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: kvm@vger.kernel.org, gleb@redhat.com, pbonzini@redhat.com To: paulmck@linux.vnet.ibm.com, linux-kernel Return-path: Content-Disposition: inline Sender: linux-kernel-owner@vger.kernel.org List-Id: kvm.vger.kernel.org I noticed that srcu_read_lock/unlock both have a memory barrier, so just by moving srcu_read_unlock earlier we can get rid of one call to smp_mb(). Unsurprisingly, the gain is small but measureable using the unit test microbenchmark: before vmcall 1407 after vmcall 1357 Signed-off-by: Michael S. Tsirkin -- I didn't stress test this yet, sending out for early review/flames. Paul, could you review this patch please? Documentation/memory-barriers.txt says that unlock has a weaker uni-directional barrier, but in practice srcu_read_unlock calls smp_mb(). Is it OK to rely on this? If not, can I add smp_mb__after_srcu_read_unlock (making it an empty macro for now) so we can avoid an actual extra smp_mb()? Thanks. diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index 8617c9d..a48fb36 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -5949,8 +5949,10 @@ restore: /* We should set ->mode before check ->requests, * see the comment in make_all_cpus_request. + * + * srcu_read_unlock below acts as a memory barrier. */ - smp_mb(); + srcu_read_unlock(&vcpu->kvm->srcu, vcpu->srcu_idx); local_irq_disable(); @@ -5960,12 +5962,11 @@ restore: smp_wmb(); local_irq_enable(); preempt_enable(); + vcpu->srcu_idx = srcu_read_lock(&vcpu->kvm->srcu); r = 1; goto cancel_injection; } - srcu_read_unlock(&vcpu->kvm->srcu, vcpu->srcu_idx); - if (req_immediate_exit) smp_send_reschedule(vcpu->cpu); -- MST