From mboxrd@z Thu Jan  1 00:00:00 1970
From: Jarno Rajahalme <jarno.rajahalme@gmail.com>
Subject: Re: Long delays creating a netns after deleting one (possibly RCU related)
Date: Thu, 17 Nov 2016 16:31:19 -0800
Message-ID: <5697DF47-5CCB-41F3-9BD1-62DE42AD61B8@gmail.com>
References: <CA+pO-2f+7QfsLiFwvGcNqTrQ0t8B5+w0vZHFrshWsSWgDE6iUg@mail.gmail.com> <CAM_iQpUNRyv3wHi3cJpcH6avo+D7Gxwqp+iybUxYc7PCo3WUPw@mail.gmail.com> <20161110212404.GB4127@linux.vnet.ibm.com> <CA+pO-2ddoJYvhVoV3Bo+-3K7V63eY9sbKzKwYrut65GZXs6_1g@mail.gmail.com> <20161112002347.GL4127@linux.vnet.ibm.com> <CAM_iQpUDnPfVayQv7Q+ZtC_2_ZbX9MRX=514gFpF3E6XY+GtSA@mail.gmail.com> <CAM_iQpUq14GsvGyz2xA1PkDq5YO743T68Zh2zm=NJZQ9S2Ahqg@mail.gmail.com> <20161114162417.GJ4127@linux.vnet.ibm.com> <CAM_iQpXL+JaVG86+h2ucYs4Dm0zJKHq+4Nm+gk75wESGOzTmJQ@mail.gmail.com> <20161114181425.GN4127@linux.vnet.ibm.com> <87y40lhfrt.fsf@xmission.com> <1479163571.8455.83.camel@edumazet-glaptop3.roam.corp.google.com> <1479164967.8455.87.camel@edumazet-glaptop3.roam.corp.google.com>
Mime-Version: 1.0 (Mac OS X Mail 9.3 \(3124\))
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: quoted-printable
Cc: "Eric W. Biederman" <ebiederm@xmission.com>,
        "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>,
        Cong Wang <xiyou.wangcong@gmail.com>,
        Rolf Neugebauer <rolf.neugebauer@docker.com>,
        LKML <linux-kernel@vger.kernel.org>,
        Linux Kernel Network Developers <netdev@vger.kernel.org>,
        Justin Cormack <justin.cormack@docker.com>,
        Ian Campbell <ian.campbell@docker.com>,
        Eric Dumazet <edumazet@google.com>
To: Eric Dumazet <eric.dumazet@gmail.com>
Return-path: <linux-kernel-owner@vger.kernel.org>
In-Reply-To: <1479164967.8455.87.camel@edumazet-glaptop3.roam.corp.google.com>
Sender: linux-kernel-owner@vger.kernel.org
List-Id: netdev.vger.kernel.org


> On Nov 14, 2016, at 3:09 PM, Eric Dumazet <eric.dumazet@gmail.com> =
wrote:
>=20
> On Mon, 2016-11-14 at 14:46 -0800, Eric Dumazet wrote:
>> On Mon, 2016-11-14 at 16:12 -0600, Eric W. Biederman wrote:
>>=20
>>> synchronize_rcu_expidited is not enough if you have multiple network
>>> devices in play.
>>>=20
>>> Looking at the code it comes down to this commit, and it appears =
there
>>> is a promise add rcu grace period combining by Eric Dumazet.
>>>=20
>>> Eric since people are hitting noticable stalls because of the rcu =
grace
>>> period taking a long time do you think you could look at this code =
path
>>> a bit more?
>>>=20
>>> commit 93d05d4a320cb16712bb3d57a9658f395d8cecb9
>>> Author: Eric Dumazet <edumazet@google.com>
>>> Date:   Wed Nov 18 06:31:03 2015 -0800
>>=20
>> Absolutely, I will take a loop asap.
>=20
> The worst offender should be fixed by the following patch.
>=20
> busy poll needs to poll the physical device, not a virtual one...
>=20
> diff --git a/include/net/gro_cells.h b/include/net/gro_cells.h
> index =
d15214d673b2e8e08fd6437b572278fb1359f10d..2a1abbf8da74368cd01adc40cef6c064=
4e059ef2 100644
> --- a/include/net/gro_cells.h
> +++ b/include/net/gro_cells.h
> @@ -68,6 +68,9 @@ static inline int gro_cells_init(struct gro_cells =
*gcells, struct net_device *de
> 		struct gro_cell *cell =3D per_cpu_ptr(gcells->cells, i);
>=20
> 		__skb_queue_head_init(&cell->napi_skbs);
> +
> +		set_bit(NAPI_STATE_NO_BUSY_POLL, &cell->napi.state);
> +
> 		netif_napi_add(dev, &cell->napi, gro_cell_poll, 64);
> 		napi_enable(&cell->napi);
> 	}
>=20

This solved a ~20 second slowdown between OVS datapath unit tests for =
me.

  Jarno