From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx2.suse.de ([195.135.220.15]:33376 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758753AbcCDJQT (ORCPT ); Fri, 4 Mar 2016 04:16:19 -0500 Date: Fri, 4 Mar 2016 10:16:02 +0100 From: David Sterba To: Marc Kleine-Budde Cc: Will Deacon , Chris Mason , Josef Bacik , "linux-arm-kernel@lists.infradead.org" , linux-btrfs@vger.kernel.org Subject: Re: btrfs_get_token_64() alignment problem on ARM (was: Re: DWord alignment on ARMv7) Message-ID: <20160304091602.GH29310@suse.cz> Reply-To: dsterba@suse.cz References: <56D8BA3F.7050508@pengutronix.de> <20160303235426.GA11237@arm.com> <56D940E8.2030202@pengutronix.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <56D940E8.2030202@pengutronix.de> Sender: linux-btrfs-owner@vger.kernel.org List-ID: On Fri, Mar 04, 2016 at 09:01:44AM +0100, Marc Kleine-Budde wrote: > Hello, > > On 03/04/2016 12:54 AM, Will Deacon wrote: > > Hi Marc, > > > > On Thu, Mar 03, 2016 at 11:27:11PM +0100, Marc Kleine-Budde wrote: > >> I'm using btrfs on am ARMv7 and it turns out, that the kernel has to > >> fixup a lot of kernel originated alignment issues. > >> > >> See /proc/cpu/alignment (~4h of uptime): > >>> System: 22304815 (btrfs_get_token_64+0x13c/0x148 [btrfs]) > >> > >> For example, when compiling the kernel on a btrfs volume the counter > >> increases by 100...1000 per second. > >> > >> The function shown "btrfs_get_token_64()" is defined here: > >>> http://lxr.free-electrons.com/source/fs/btrfs/struct-funcs.c#L53 > >> ...it already uses get_unaligned_leXX accessors. > >> > >> Quoting a comment in arch/arm/mm/alignment.c: > >> > >> * ARMv6 and later CPUs can perform unaligned accesses for > >> * most single load and store instructions up to word size. > >> * LDM, STM, LDRD and STRD still need to be handled. > >> > >> But on a 32bit ARMv7 64bits are not word-sized. > >> > >> Is the exception and fixup overhead neglectable? Do we have to introduce > >> something like HAVE_EFFICIENT_UNALIGNED_64BIT_ACCESS? > > > > Ouch, that trap/emulate is certainly going to have an effect on your > > performance. I doubt that HAVE_EFFICIENT_UNALIGNED_ACCESS applies to > > types bigger than the native word size on many architectures, so my > > hunch is that the btrfs code should be checking BITS_PER_LONG or similar > > to establish whether or not to break the access up into word accesses. > > I've added the btrfs maintainers on Cc. Can this be done transparently via the the get_unaligned_le* helpers? This seems to be too arch-specific to fix it in btrfs.