From mboxrd@z Thu Jan  1 00:00:00 1970
From: Jiri Pirko <jiri@resnulli.us>
Subject: Re: [patch net-next RFC 3/3] switchdev: introduce deferred variants
 of obj_add/del helpers
Date: Thu, 8 Oct 2015 15:09:11 +0200
Message-ID: <20151008130910.GJ2186@nanopsycho.orion>
References: <1444242652-17260-1-git-send-email-jiri@resnulli.us>
 <1444242652-17260-4-git-send-email-jiri@resnulli.us>
 <CAJ3xEMhEupHZvdQfBucemQwyjtQUcwwc5D5MhYqpctJGqtcTsA@mail.gmail.com>
 <20151008082858.GC2186@nanopsycho.orion>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: Linux Netdev List <netdev@vger.kernel.org>,
	David Miller <davem@davemloft.net>, idosch@mellanox.com,
	Elad Raz <eladr@mellanox.com>,
	Scott Feldman <sfeldma@gmail.com>,
	Florian Fainelli <f.fainelli@gmail.com>,
	Guenter Roeck <linux@roeck-us.net>,
	Vivien Didelot <vivien.didelot@savoirfairelinux.com>,
	Andrew Lunn <andrew@lunn.ch>,
	john fastabend <john.fastabend@gmail.com>,
	David Laight <David.Laight@aculab.com>
To: Or Gerlitz <gerlitz.or@gmail.com>
Return-path: <netdev-owner@vger.kernel.org>
Received: from mail-wi0-f172.google.com ([209.85.212.172]:36634 "EHLO
	mail-wi0-f172.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1752288AbbJHNJN (ORCPT
	<rfc822;netdev@vger.kernel.org>); Thu, 8 Oct 2015 09:09:13 -0400
Received: by wicgb1 with SMTP id gb1so24916958wic.1
        for <netdev@vger.kernel.org>; Thu, 08 Oct 2015 06:09:12 -0700 (PDT)
Content-Disposition: inline
In-Reply-To: <20151008082858.GC2186@nanopsycho.orion>
Sender: netdev-owner@vger.kernel.org
List-ID: <netdev.vger.kernel.org>

Thu, Oct 08, 2015 at 10:28:58AM CEST, jiri@resnulli.us wrote:
>Thu, Oct 08, 2015 at 08:45:58AM CEST, gerlitz.or@gmail.com wrote:
>>On Wed, Oct 7, 2015 at 9:30 PM, Jiri Pirko <jiri@resnulli.us> wrote:
>>> From: Jiri Pirko <jiri@mellanox.com>
>>>
>>> Similar to the attr usecase, the caller knows if he is holding RTNL and is
>>> in atomic section. So let the called to decide the correct call variant.
>>>
>>> This allows drivers to sleep inside their ops and wait for hw to get the
>>> operation status. Then the status is propagated into switchdev core.
>>> This avoids silent errors in drivers.
>>>
>>> Signed-off-by: Jiri Pirko <jiri@mellanox.com>
>>
>>> diff --git a/net/bridge/br_fdb.c b/net/bridge/br_fdb.c
>>> index 7f7d551..2086767 100644
>>> --- a/net/bridge/br_fdb.c
>>> +++ b/net/bridge/br_fdb.c
>>> @@ -139,7 +139,7 @@ static void fdb_del_external_learn(struct net_bridge_fdb_entry *f)
>>>                 .vid = f->vlan_id,
>>>         };
>>>
>>> -       switchdev_port_obj_del(f->dst->dev, &fdb.obj);
>>> +       switchdev_port_obj_del_deferred(f->dst->dev, &fdb.obj);
>>>  }
>>
>>
>>>  static void fdb_delete(struct net_bridge *br, struct net_bridge_fdb_entry *f)
>>> diff --git a/net/switchdev/switchdev.c b/net/switchdev/switchdev.c
>>> index c29f4ee..49e6e6f 100644
>>> --- a/net/switchdev/switchdev.c
>>> +++ b/net/switchdev/switchdev.c
>>> @@ -362,6 +362,75 @@ int switchdev_port_obj_add(struct net_device *dev,
>>>  }
>>>  EXPORT_SYMBOL_GPL(switchdev_port_obj_add);
>>>
>>> +struct switchdev_obj_work {
>>> +       struct work_struct work;
>>> +       struct net_device *dev;
>>> +       struct switchdev_obj obj;
>>> +       bool add; /* add of del */
>>> +};
>>> +
>>> +static void switchdev_port_obj_work(struct work_struct *work)
>>> +{
>>> +       struct switchdev_obj_work *ow =
>>> +                       container_of(work, struct switchdev_obj_work, work);
>>> +       int err;
>>> +
>>> +       rtnl_lock();
>>> +       if (ow->add)
>>> +               err = switchdev_port_obj_add(ow->dev, &ow->obj);
>>> +       else
>>> +               err = switchdev_port_obj_del(ow->dev, &ow->obj);
>>> +       if (err && err != -EOPNOTSUPP)
>>> +               netdev_err(ow->dev, "failed (err=%d) to %s object (id=%d)\n",
>>> +                          err, ow->add ? "add" : "del", ow->obj.id);
>>
>>This introduced a regression to the 2-phase commit scheme, since the
>>prepare commit can fail
>>and that would go un-noticed toward the upper layer, agree?
>
>Well, no. This still does the transaction for all lower devices in one
>go. No change in that.
>
Now I get it, yes you are right. But currently there is no code in
kernel which would control retval of deferred attr_set or obj_add/del