From mboxrd@z Thu Jan 1 00:00:00 1970 From: Vlad Yasevich Subject: Re: [PATCH] macvtap: fix two races Date: Fri, 09 Aug 2013 14:01:00 -0400 Message-ID: <52052E5C.6060000@redhat.com> References: <20130807164319.20aa2333@thhw500> <1375889211.4004.28.camel@edumazet-glaptop> <20130808102551.6aa47852@thhw500> <1375968072.4004.92.camel@edumazet-glaptop> <20130808155634.239dbc2d@thhw500> <1375974374.4004.103.camel@edumazet-glaptop> <520523DA.1000400@redhat.com> <1376070067.20509.1.camel@edumazet-glaptop> Reply-To: vyasevic@redhat.com Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Cc: Thomas Huth , David Miller , netdev@vger.kernel.org To: Eric Dumazet Return-path: Received: from mx1.redhat.com ([209.132.183.28]:55026 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758547Ab3HISBF (ORCPT ); Fri, 9 Aug 2013 14:01:05 -0400 In-Reply-To: <1376070067.20509.1.camel@edumazet-glaptop> Sender: netdev-owner@vger.kernel.org List-ID: On 08/09/2013 01:41 PM, Eric Dumazet wrote: > On Fri, 2013-08-09 at 13:16 -0400, Vlad Yasevich wrote: > >> I was looking at this a bit more and I think this call to >> macvlan_count_rx() is double counting the packets. >> >> In macvlan_handle_frame(), we call macvlan_count_rx() after we call >> vlan->receive(). For macvtap, receive() function is essentially >> macvtap_forward() which just tacks the data onto the queue. >> >> Then, the above code counts the data again as we pull it off the queue >> socket queue to give to the user. > > Hmm, it seems a different issue, and probably needs a patch on its own. > > When was this problem added ? > > Looks like both macvlan and macvtap packet counting has been there since the beginning... -vlad