From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <bruce.richardson@intel.com>
Received: from mga07.intel.com (mga07.intel.com [134.134.136.100])
 by dpdk.org (Postfix) with ESMTP id E90261B217
 for <dev@dpdk.org>; Mon,  6 Nov 2017 10:25:45 +0100 (CET)
Received: from orsmga004.jf.intel.com ([10.7.209.38])
 by orsmga105.jf.intel.com with ESMTP; 06 Nov 2017 01:25:44 -0800
X-ExtLoop1: 1
X-IronPort-AV: E=Sophos;i="5.44,351,1505804400"; d="scan'208";a="146400248"
Received: from bricha3-mobl3.ger.corp.intel.com ([10.237.221.32])
 by orsmga004.jf.intel.com with SMTP; 06 Nov 2017 01:25:42 -0800
Received: by  (sSMTP sendmail emulation); Mon, 06 Nov 2017 09:25:42 +0000
Date: Mon, 6 Nov 2017 09:25:41 +0000
From: Bruce Richardson <bruce.richardson@intel.com>
To: Mike Stolarchuk <mike.stolarchuk@bigswitch.com>
Cc: dev@dpdk.org
Message-ID: <20171106092541.GA21408@bricha3-MOBL3.ger.corp.intel.com>
References: <CA+Qp4XwDBVuFBoJe2xsw74PV+=K_2F5XxfOw9UqoLH072npbiQ@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CA+Qp4XwDBVuFBoJe2xsw74PV+=K_2F5XxfOw9UqoLH072npbiQ@mail.gmail.com>
Organization: Intel Research and Development Ireland Ltd.
User-Agent: Mutt/1.9.1 (2017-09-22)
Subject: Re: [dpdk-dev] What's the point of RTE_MBUF_PREFETCH_TO_FREE()?
X-BeenThere: dev@dpdk.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: DPDK patches and discussions <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: Mon, 06 Nov 2017 09:25:46 -0000

On Fri, Nov 03, 2017 at 01:28:18PM -0700, Mike Stolarchuk wrote:
> It seems to be used to prefetch the mbuf in preparation for a free of an
> mbuf.
> 
> But RTE_MBUF_PREFETCH_TO_FREE() currently does a prefetch for the
> ffirst cache line in the mbuf,  but doesn't a free do a put to the mbuf's
> mempool?
> 
> And isn't the mempool reference in the mbuf's second cache line?
> 
Yes, that looks like a bug. Doing "git blame" shows that that macro
comes from the original versions of DPDK pre-dpdk.org, and hasn't been
modified since, so it was missed when the mbuf was expanded from 1 line
to 2. I also note that the high-performance ixgbe/i40e vector drivers do
not use this macro, only the slow paths in those drivers do, so I wonder
if it can be removed. If not, it should be fixed.

	/Bruce