From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752812AbdEHGSu (ORCPT ); Mon, 8 May 2017 02:18:50 -0400 Received: from mail-wr0-f196.google.com ([209.85.128.196]:34828 "EHLO mail-wr0-f196.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751267AbdEHGSt (ORCPT ); Mon, 8 May 2017 02:18:49 -0400 Date: Mon, 8 May 2017 08:18:45 +0200 From: Ingo Molnar To: alexander.levin@verizon.com Cc: "mingo@redhat.com" , "linux-kernel@vger.kernel.org" , "ben@decadent.org.uk" , Arnaldo Carvalho de Melo Subject: Re: [GIT PULL] liblockdep for 4.12 Message-ID: <20170508061845.rhoq73yentafplwv@gmail.com> References: <20170505025244.b7slcr4cbywlvf5e@sasha-lappy> <20170505061138.mrxapt6k2k4k7igq@gmail.com> <20170507230444.hpglxzcdtiusxng7@sasha-lappy> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20170507230444.hpglxzcdtiusxng7@sasha-lappy> User-Agent: NeoMutt/20170113 (1.7.2) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org * alexander.levin@verizon.com wrote: > On Fri, May 05, 2017 at 08:11:38AM +0200, Ingo Molnar wrote: > > * alexander.levin@verizon.com wrote: > > Could you please include all the patches as part of the pull request, so that > > individual patches can be discussed? > > Sure, I'll resend. > > > In particular I'm wondering about: > > > > > tools/include/linux/jhash.h | 125 +++++++++++++++++++++++ > > > > include/linux/jhash.h is 175 lines long - why not make an exact copy and keep it > > synchronized the way perf does it? > > Including jhash() vs only jhash2() required pulling additional headers in. > Since we don't need jhash() I preferred to drop it. > > Let me know if you'd prefer to do it the other way and include everything. How many headers would that be? We already have tools/include/../bitops.h, so we'd only need include/linux/unaligned/packed_struct.h, which looks straightforward and smallish. ( That header includes kernel.h, but we have that in tools as well, so in theory it could work. ) The principle would be to 'COW-share' as much code as possible, at least for obvious arithmetics helpers - which hashes do appear to be. Thanks, Ingo