From mboxrd@z Thu Jan  1 00:00:00 1970
From: Paolo Abeni <pabeni@redhat.com>
Subject: Re: unexpected GRO/veth behavior
Date: Tue, 11 Sep 2018 13:44:56 +0200
Message-ID: <3ab5755e75cfe2bfaeeb133b53512da43cfdb8b0.camel@redhat.com>
References: <4106d3f7eee7f0186fcfdd0331cdafeecd3240c0.camel@redhat.com>
         <bcea5cafe139e2c561f036f5b3e2598d7d0dd970.camel@redhat.com>
         <ecb079cc-033d-0726-75ce-00eada2ae765@gmail.com>
Mime-Version: 1.0
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
Cc: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
To: Eric Dumazet <eric.dumazet@gmail.com>, netdev@vger.kernel.org
Return-path: <netdev-owner@vger.kernel.org>
Received: from mx3-rdu2.redhat.com ([66.187.233.73]:46694 "EHLO mx1.redhat.com"
        rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP
        id S1726554AbeIKQnz (ORCPT <rfc822;netdev@vger.kernel.org>);
        Tue, 11 Sep 2018 12:43:55 -0400
In-Reply-To: <ecb079cc-033d-0726-75ce-00eada2ae765@gmail.com>
Sender: netdev-owner@vger.kernel.org
List-ID: <netdev.vger.kernel.org>

On Tue, 2018-09-11 at 03:27 -0700, Eric Dumazet wrote:
> On 09/10/2018 11:54 PM, Paolo Abeni wrote:
> > I *think* we can hit a similar condition with a tun device in IFF_NAPI
> > mode.
> 
> Why ?
> 
> tun_get_user() does not attach skb to a socket, that would be quite useless since skb
> 
	
I think we can hit this code path:

tun_get_user() -> tun_alloc_skb() -> sock_alloc_send_pskb() ->
skb_set_owner_w() 

Than later napi and GRO.

Cheers,

Paolo