From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: [PATCH net-next 0/4] net: bridge: simplify receive path and consolidate forwarding paths Date: Sat, 16 Jul 2016 19:58:02 -0700 (PDT) Message-ID: <20160716.195802.931580320600384868.davem@davemloft.net> References: <1468465802-4571-1-git-send-email-nikolay@cumulusnetworks.com> Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org, roopa@cumulusnetworks.com, stephen@networkplumber.org, bridge@lists.linux-foundation.org To: nikolay@cumulusnetworks.com Return-path: Received: from shards.monkeyblade.net ([184.105.139.130]:44092 "EHLO shards.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751871AbcGQC6G (ORCPT ); Sat, 16 Jul 2016 22:58:06 -0400 In-Reply-To: <1468465802-4571-1-git-send-email-nikolay@cumulusnetworks.com> Sender: netdev-owner@vger.kernel.org List-ID: From: Nikolay Aleksandrov Date: Thu, 14 Jul 2016 06:09:58 +0300 > This set tries to simplify the receive and forwarding paths. Patch 01 is > a trivial style adjustment, patch 02 removes one conditional from the > unicast fast path, patch 03 removes another conditional and more imporantly > removes the skb0/skb2 ambiguity about locally receiving the skb and > switches to a boolean called "local_rcv". > Patch 04 is the most important change which consolidates the forwarding > paths for locally originated and forwarded packets into __br_forward. This > allows us to remove the function pointers giving a minor performance boost, > more importantly it makes it much easier to reason about the forwarding > path and reduces the code duplication that was needed when making changes. > Also it allows the receive path to fully setup the environment prior to > calling any forwarding functions (i.e. to properly set unicast, local_rcv > and search for unicast/mcast dst). > Functionally everything should stay the same after this set. > > I've done basic tests with unicast/multicast/broadcast Tx/Rx. Please > review carefully. I've reviewed this twice and can't find any problems, so applied to net-next, thanks.