From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:52511)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <peterz@infradead.org>) id 1a9Wdi-000550-00
	for qemu-devel@nongnu.org; Thu, 17 Dec 2015 06:22:30 -0500
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <peterz@infradead.org>) id 1a9Wde-0003w5-S7
	for qemu-devel@nongnu.org; Thu, 17 Dec 2015 06:22:29 -0500
Received: from bombadil.infradead.org ([2001:1868:205::9]:43634)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <peterz@infradead.org>) id 1a9Wde-0003vi-NP
	for qemu-devel@nongnu.org; Thu, 17 Dec 2015 06:22:26 -0500
Date: Thu, 17 Dec 2015 12:22:22 +0100
From: Peter Zijlstra <peterz@infradead.org>
Message-ID: <20151217112222.GC6375@twins.programming.kicks-ass.net>
References: <1450347932-16325-1-git-send-email-mst@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <1450347932-16325-1-git-send-email-mst@redhat.com>
Subject: Re: [Qemu-devel] [PATCH] virtio_ring: use smp_store_mb
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: "Michael S. Tsirkin" <mst@redhat.com>
Cc: qemu-devel@nongnu.org, Jason Wang <jasowang@redhat.com>, linux-kernel@vger.kernel.org, virtualization@lists.linux-foundation.org

On Thu, Dec 17, 2015 at 12:32:53PM +0200, Michael S. Tsirkin wrote:
> Seems to give a speedup on my box but I'm less sure about this one.  E.g. as
> xchng faster than mfence on all/most intel CPUs? Anyone has an opinion?

Would help if you Cc people who would actually know this :-)

Yes, we've recently established that xchg is indeed faster than mfence
on at least recent machines, see:

  lkml.kernel.org/r/CA+55aFynbkeuUGs9s-q+fLY6MeRBA6MjEyWWbbe7A5AaqsAknw@mail.gmail.com

> +static inline void virtio_store_mb(bool weak_barriers,
> +				   __virtio16 *p, __virtio16 v)
> +{
> +#ifdef CONFIG_SMP
> +	if (weak_barriers)
> +		smp_store_mb(*p, v);
> +	else
> +#endif
> +	{
> +		WRITE_ONCE(*p, v);
> +		mb();
> +	}
> +}

Note that virtio_mb() is weirdly inconsistent with virtio_[rw]mb() in
that they use dma_* ops for weak_barriers, while virtio_mb() uses
smp_mb().

As previously stated, smp_mb() does not cover the same memory domains as
dma_mb() would.