From mboxrd@z Thu Jan  1 00:00:00 1970
From: David Ahern <dsahern@gmail.com>
Subject: Re: xdp and fragments with virtio
Date: Fri, 18 May 2018 10:42:51 -0600
Message-ID: <5ed42a82-56f6-d8ea-57b8-8f2229857dd1@gmail.com>
References: <e35c149e-cb15-bb8c-2e03-d1641d21d694@gmail.com>
 <4d6952b4-b252-2fe5-8893-a3e2329bd34b@redhat.com>
 <ff050e6b-cd64-d2cc-e8fa-03a07b7cf247@gmail.com>
 <fbcf2fcc-0e40-6f0f-f8a5-8255a59eaec5@redhat.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 8bit
To: Jason Wang <jasowang@redhat.com>,
        "netdev@vger.kernel.org" <netdev@vger.kernel.org>
Return-path: <netdev-owner@vger.kernel.org>
Received: from mail-pf0-f196.google.com ([209.85.192.196]:34378 "EHLO
        mail-pf0-f196.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
        with ESMTP id S1752165AbeERQmz (ORCPT
        <rfc822;netdev@vger.kernel.org>); Fri, 18 May 2018 12:42:55 -0400
Received: by mail-pf0-f196.google.com with SMTP id a14-v6so4026615pfi.1
        for <netdev@vger.kernel.org>; Fri, 18 May 2018 09:42:55 -0700 (PDT)
In-Reply-To: <fbcf2fcc-0e40-6f0f-f8a5-8255a59eaec5@redhat.com>
Content-Language: en-US
Sender: netdev-owner@vger.kernel.org
List-ID: <netdev.vger.kernel.org>

On 5/17/18 3:24 AM, Jason Wang wrote:
> It looks like we wrongly drop packets after linearizing the packets
> during XDP_REDIRECT.
> 
> Please try the patch (but I do spot some other issues, will post a series):
> 
> Thanks
> 
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index f34794a..59702f9 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -800,7 +800,7 @@ static struct sk_buff *receive_mergeable(struct
> net_device *dev,
>                         }
>                         *xdp_xmit = true;
>                         if (unlikely(xdp_page != page))
> -                               goto err_xdp;
> +                               put_page(page);
>                         rcu_read_unlock();
>                         goto xdp_xmit;
>                 default:

Yes, that does solve the problem of fragments getting lost.