From mboxrd@z Thu Jan  1 00:00:00 1970
From: Eric Kinzie <ehkinzie@gmail.com>
Subject: Re: [PATCH v2 4/6] bond mode 4: allow external state
 machine
Date: Tue, 1 Mar 2016 12:40:46 -0500
Message-ID: <20160301174042.GA14082@roosta.home>
References: <8CEF83825BEC744B83065625E567D7C219FB2C51@IRSMSX108.ger.corp.intel.com>
 <1455909473-13539-1-git-send-email-ehkinzie@gmail.com>
 <1455909473-13539-5-git-send-email-ehkinzie@gmail.com>
 <56CB0709.6010407@redhat.com>
 <8CEF83825BEC744B83065625E567D7C219FB4852@IRSMSX108.ger.corp.intel.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: "dev@dpdk.org" <dev@dpdk.org>
To: "Iremonger, Bernard" <bernard.iremonger@intel.com>
Return-path: <dev-bounces@dpdk.org>
Received: from mail-qg0-f67.google.com (mail-qg0-f67.google.com
 [209.85.192.67]) by dpdk.org (Postfix) with ESMTP id D2623AA39
 for <dev@dpdk.org>; Tue,  1 Mar 2016 18:40:49 +0100 (CET)
Received: by mail-qg0-f67.google.com with SMTP id p68so8319181qge.2
 for <dev@dpdk.org>; Tue, 01 Mar 2016 09:40:49 -0800 (PST)
Content-Disposition: inline
In-Reply-To: <8CEF83825BEC744B83065625E567D7C219FB4852@IRSMSX108.ger.corp.intel.com>
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>
Errors-To: dev-bounces@dpdk.org
Sender: "dev" <dev-bounces@dpdk.org>

On Thu Feb 25 15:22:35 +0000 2016, Iremonger, Bernard wrote:
> Hi Eric,
> <snip>
> 
> > > @@ -157,6 +159,7 @@ struct rte_eth_bond_8023ad_conf {
> > >   	uint32_t tx_period_ms;
> > >   	uint32_t rx_marker_period_ms;
> > >   	uint32_t update_timeout_ms;
> > > +	rte_eth_bond_8023ad_ext_slowrx_fn slowrx_cb;
> > >   };
> > 
> > This still is a likely an ABI break, previously discussed around here:
> > http://dpdk.org/ml/archives/dev/2015-November/027321.html
> > 
> > It might not be embedded anywhere in DPDK codebase, but there's no
> > telling what others might have done with it (have an array of them, embed in
> > other structs etc).
> > 
> > Also ultimately ABI compatibility goes both ways: when the library soname
> > does not change then an application is free to assume both downgrading and
> > upgrading are safe. In this case, upgrading *might* be okay, downgrading
> > certainly is not. So by that measure it definitely is an ABI break.
> > 
> > [...]
> > > diff --git a/drivers/net/bonding/rte_eth_bond_version.map
> > > b/drivers/net/bonding/rte_eth_bond_version.map
> > > index 22bd920..33d73ff 100644
> > > --- a/drivers/net/bonding/rte_eth_bond_version.map
> > > +++ b/drivers/net/bonding/rte_eth_bond_version.map
> > > @@ -27,3 +27,9 @@ DPDK_2.1 {
> > >   	rte_eth_bond_free;
> > >
> > >   } DPDK_2.0;
> > > +
> > > +DPDK_2.2 {
> > > +	rte_eth_bond_8023ad_ext_collect;
> > > +	rte_eth_bond_8023ad_ext_distrib;
> > > +	rte_eth_bond_8023ad_ext_slowtx;
> > > +} DPDK_2.1;
> > >
> > 
> > These symbols are not part of DPDK 2.2, the version here is wrong.
> > Technically it would not actually matter much but better not to confuse
> > things unnecessarily.
> > 
> > 	- Panu -
> 
> It looks like Panu's points are valid, a V3 of this patch set which takes care of these issues will be needed.
> 
> Patches 1/6, 5/6 and 6/6 of the patch set are bug fixes, so each patch should contain a fixes line.
> Patches 2/6, 3/6 and 4/6 are a new feature, the release notes should be updated for this feature.
> 
> Could I suggest splitting the patch set into two patch sets, a bug fix patch set and a new feature patch set.
> 
> Regards,
> 
> Bernard.

Bernard, a v3 is on the way.  I included only things that are fixes,
but the patch set doesn't quite match the set of patch numbers you listed
above.  1/6 (bond: use existing enslaved device queues) is an improvement,
but doesn't really fix anything that was broken, so I left that out.
2/6 (bond mode 4: copy entire config structure) and 3/6 (bond mode 4:
do not ignore multicast) fix bugs and are included.


Eric