From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linville@tuxdriver.com>
Received: from smtp.tuxdriver.com (charlotte.tuxdriver.com [70.61.120.58])
 by dpdk.org (Postfix) with ESMTP id EAE56B3A8
 for <dev@dpdk.org>; Fri, 11 Jul 2014 17:44:41 +0200 (CEST)
Received: from uucp by smtp.tuxdriver.com with local-rmail (Exim 4.63)
 (envelope-from <linville@tuxdriver.com>)
 id 1X5d0V-0003iT-Sj; Fri, 11 Jul 2014 11:45:07 -0400
Received: from linville-x1.hq.tuxdriver.com (localhost.localdomain [127.0.0.1])
 by linville-x1.hq.tuxdriver.com (8.14.8/8.14.6) with ESMTP id s6BFXAdH028213; 
 Fri, 11 Jul 2014 11:33:11 -0400
Received: (from linville@localhost)
 by linville-x1.hq.tuxdriver.com (8.14.8/8.14.8/Submit) id s6BFXAoj028212;
 Fri, 11 Jul 2014 11:33:10 -0400
Date: Fri, 11 Jul 2014 11:33:10 -0400
From: "John W. Linville" <linville@tuxdriver.com>
To: "Venkatesan, Venky" <venky.venkatesan@intel.com>
Message-ID: <20140711153309.GD25478@tuxdriver.com>
References: <1405024369-30058-1-git-send-email-linville@tuxdriver.com>
 <20140711061147.06c12136@samsung-9>
 <20140711144912.GA25478@tuxdriver.com>
 <59AF69C657FD0841A61C55336867B5B0343ACD8B@IRSMSX103.ger.corp.intel.com>
 <1FD9B82B8BF2CF418D9A1000154491D974124465@ORSMSX102.amr.corp.intel.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <1FD9B82B8BF2CF418D9A1000154491D974124465@ORSMSX102.amr.corp.intel.com>
User-Agent: Mutt/1.5.23 (2014-03-12)
Cc: "dev@dpdk.org" <dev@dpdk.org>
Subject: Re: [dpdk-dev] [PATCH] librte_pmd_packet: add PMD for
 AF_PACKET-based virtual devices
X-BeenThere: dev@dpdk.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: patches and discussions about DPDK <dev.dpdk.org>
List-Unsubscribe: <http://dpdk.org/ml/options/dev>,
 <mailto:dev-request@dpdk.org?subject=unsubscribe>
List-Archive: <http://dpdk.org/ml/archives/dev/>
List-Post: <mailto:dev@dpdk.org>
List-Help: <mailto:dev-request@dpdk.org?subject=help>
List-Subscribe: <http://dpdk.org/ml/listinfo/dev>,
 <mailto:dev-request@dpdk.org?subject=subscribe>
X-List-Received-Date: Fri, 11 Jul 2014 15:44:42 -0000

On Fri, Jul 11, 2014 at 03:29:17PM +0000, Venkatesan, Venky wrote:
> > -----Original Message-----
> > From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of John W. Linville
> > Sent: Friday, July 11, 2014 7:49 AM
> > To: Stephen Hemminger
> > Cc: dev@dpdk.org
> > Subject: Re: [dpdk-dev] [PATCH] librte_pmd_packet: add PMD for 
> > AF_PACKET- based virtual devices
> > 
> > On Fri, Jul 11, 2014 at 06:11:47AM -0700, Stephen Hemminger wrote:
> > > On Thu, 10 Jul 2014 16:32:49 -0400
> > > "John W. Linville" <linville@tuxdriver.com> wrote:
> > >
> > > > This is a Linux-specific virtual PMD driver backed by an AF_PACKET 

<snip>

> > > > +struct pkt_rx_queue {
> > > > +	int sockfd;
> > > > +
> > > > +	struct iovec *rd;
> > > > +	uint8_t *map;
> > > > +	unsigned int framecount;
> > > > +	unsigned int framenum;
> > > > +
> > > > +	struct rte_mempool *mb_pool;
> > > > +
> > > > +	volatile unsigned long rx_pkts;
> > > > +	volatile unsigned long err_pkts;
> > >
> > > Use of volatile will generate slow code, don't think it is 
> > > necessary, especially when only one CPU can use a queue at a time.
> > 
> > That is a good point, worth checking out.  FWIW, those lines are 
> > boilerplate originally copied from the pcap PMD. :-)
> > 
> 
> > Yes, I agree it's worth checking out if there is a performance impact, but if we assume that the stats for RX/TX are possibly going to be read by another core, they really should be volatile for correctness
> 
> Accessing the rx_queue structure directly for stats is unlikely to happen from a second core; we should probably change the PCAP PMD as well (thanks for pointing that out John). 

"Unlikely" doesn't sound completely safe... :-)

-- 
John W. Linville		Someday the world will need a hero, and you
linville@tuxdriver.com			might be all we have.  Be ready.