From mboxrd@z Thu Jan  1 00:00:00 1970
From: Sheng Yang <sheng@linux.intel.com>
Subject: Re: KVM: Merge MSI handling to kvm_set_irq
Date: Fri, 9 Jan 2009 09:32:47 +0800
Message-ID: <200901090932.48148.sheng@linux.intel.com>
References: <1231324966-22286-1-git-send-email-sheng@linux.intel.com> <1231324966-22286-6-git-send-email-sheng@linux.intel.com> <20090108135441.GA4628@silverwood.ncultra.org>
Mime-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Cc: Avi Kivity <avi@redhat.com>, Marcelo Tosatti <mtosatti@redhat.com>,
	kvm@vger.kernel.org
To: ncmike@ncultra.org
Return-path: <kvm-owner@vger.kernel.org>
Received: from mga09.intel.com ([134.134.136.24]:29630 "EHLO mga09.intel.com"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S1751068AbZAIBcw (ORCPT <rfc822;kvm@vger.kernel.org>);
	Thu, 8 Jan 2009 20:32:52 -0500
In-Reply-To: <20090108135441.GA4628@silverwood.ncultra.org>
Content-Disposition: inline
Sender: kvm-owner@vger.kernel.org
List-ID: <kvm.vger.kernel.org>

On Thursday 08 January 2009 21:54:41 Mike Day wrote:
> On 07/01/09 18:42 +0800, Sheng Yang wrote:
> > Using kvm_set_irq to handle all interrupt injection.
> >
> > Signed-off-by: Sheng Yang <sheng@linux.intel.com>
> > ---
> >
> > +static void gsi_dispatch(struct kvm *kvm, u32 gsi)
>
> ...
>
> > +		case IOAPIC_FIXED:
> > +			for (vcpu_id = 0; deliver_bitmask != 0; vcpu_id++) {
> > +				if (!(deliver_bitmask & (1 << vcpu_id)))
> > +					continue;
> > +				deliver_bitmask &= ~(1 << vcpu_id);
> > +				vcpu = ioapic->kvm->vcpus[vcpu_id];
> > +				if (vcpu)
> > +					kvm_apic_set_irq(vcpu, vector,
> > +							trig_mode);
> > +			}
> > +			break;
> > +		default:
> > +			break;
> > +		}
>
> In cases such as the for() loop above, which are numerous in the
> patchset, I wonder if using bitops would be slightly better:
>
> 		case IOAPIC_FIXED:
> 			while (deliver_bitmask != 0) {
> 				vcpu_id = ffs(deliver_bitmask);
> 				__clear_bit(vcpu_id - 1, &deliver_bitmask);
> 				vcpu = ioapic->kvm->vcpus[vcpu_id - 1];
> 				if (vcpu)
> 					kvm_apic_set_irq(vcpu, vector,
> 							 trig_mode);
> 			} ;
>
>
> I did a quick check and the second example compiles to a more
> consise set of assembler instructions. The current code uses bitops in
> cases like this.
>

Yes, that's what I did for bitmap changing in the following patches. Please 
refer to "[PATCH 10/10] KVM: bit ops for deliver_bitmap" I sent before.

-- 
regards
Yang, Sheng

> Mike