From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:36535)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <famz@redhat.com>) id 1YJIm9-0005BN-Pi
	for qemu-devel@nongnu.org; Thu, 05 Feb 2015 04:31:06 -0500
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <famz@redhat.com>) id 1YJIm6-0001Nz-KE
	for qemu-devel@nongnu.org; Thu, 05 Feb 2015 04:31:05 -0500
Received: from mx1.redhat.com ([209.132.183.28]:60984)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <famz@redhat.com>) id 1YJIm6-0001Nv-DF
	for qemu-devel@nongnu.org; Thu, 05 Feb 2015 04:31:02 -0500
Date: Thu, 5 Feb 2015 17:30:55 +0800
From: Fam Zheng <famz@redhat.com>
Message-ID: <20150205093055.GD5048@ad.nay.redhat.com>
References: <1422967948-3261-1-git-send-email-pbonzini@redhat.com>
	<1422967948-3261-6-git-send-email-pbonzini@redhat.com>
	<20150205062301.GA18713@ad.nay.redhat.com>
	<54D32BBC.3030302@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <54D32BBC.3030302@redhat.com>
Subject: Re: [Qemu-devel] [PATCH 5/9] exec: protect mru_block with RCU
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: Paolo Bonzini <pbonzini@redhat.com>
Cc: peter.maydell@linaro.org, qemu-devel@nongnu.org, fred.konrad@greensocs.com

On Thu, 02/05 09:37, Paolo Bonzini wrote:
> 
> 
> On 05/02/2015 07:23, Fam Zheng wrote:
> >> > @@ -1381,14 +1397,16 @@ static ram_addr_t ram_block_add(RAMBlock *new_block, Error **errp)
> >> >          QTAILQ_INSERT_TAIL(&ram_list.blocks, new_block, next);
> >> >      }
> >> >      ram_list.mru_block = NULL;
> >> > +    atomic_rcu_set(&ram_list.version, ram_list.version + 1);
> >> >  
> >> > -    ram_list.version++;
> > Why is this not atomic_inc
> 
> Because writes are protected by the ramlist lock.  atomic_inc is more
> expensive.

OK!

> 
> > (or why is atomic_rcu_set necessary here)?
> 
> I probably should move it to patch 9; it is needed to update the list
> before ram_list.version.
> 
> If you prefer I can change it to
> 
> 	smp_wmb();
> 	atomic_set(&ram_list.version, ram_list.version + 1);
> 
> ?
> 

Yes, this looks more obvious :)

Fam