From: Christophe Leroy Subject: Re: [PATCH 4/9] powerpc: inline ip_fast_csum() Date: Mon, 29 Feb 2016 08:25:59 +0100 Message-ID: <56D3F287.6020707@c-s.fr> References: <807dbc9449077e36752c649c09ae1c0d70e45254.1442876807.git.christophe.leroy@c-s.fr> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Benjamin Herrenschmidt , Paul Mackerras , Michael Ellerman , scottwood@freescale.com, linux-kernel@vger.kernel.org, linuxppc-dev@lists.ozlabs.org, netdev@vger.kernel.org To: Denis Kirjanov Return-path: In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org Le 23/09/2015 07:43, Denis Kirjanov a =C3=A9crit : > On 9/22/15, Christophe Leroy wrote: >> In several architectures, ip_fast_csum() is inlined >> There are functions like ip_send_check() which do nothing >> much more than calling ip_fast_csum(). >> Inlining ip_fast_csum() allows the compiler to optimise better > Hi Christophe, > I did try it and see no difference on ppc64. Did you test with sockli= b > with modified loopback and if so do you have any numbers? Hi Denis, I put a mftbl at start and end of ip_send_check() and tested on a MPC88= 5: * Without ip_fast_csum() inlined, approxymatly 7 TB ticks are spent in=20 ip_send_check() * With ip_fast_csum() inlined, approxymatly 5,4 TB ticks are spent in=20 ip_send_check() So it is about 23% time reduction. Christophe