From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mathieu Desnoyers Date: Sat, 22 Jan 2011 17:15:49 +0000 Subject: Re: [patch 2/3] tracing: fix sparc64 alignment crash with Message-Id: <20110122171549.GB31308@Krystal> List-Id: References: <20110121203630.725922272@efficios.com> <20110121203642.884088920@efficios.com> <20110121.160509.146106348.davem@davemloft.net> In-Reply-To: <20110121.160509.146106348.davem@davemloft.net> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: David Miller Cc: linux-kernel@vger.kernel.org, rostedt@goodmis.org, fweisbec@gmail.com, mingo@elte.hu, richm@oldelvet.org.uk, ben@decadent.org.uk, sparclinux@vger.kernel.org, stable * David Miller (davem@davemloft.net) wrote: > From: Mathieu Desnoyers > Date: Fri, 21 Jan 2011 15:36:32 -0500 > > > Problem description: > > > > gcc happily align structures defined statically on 32-byte. Ftrace trace events > > and Tracepoints both statically define structures into sections (using the > > "section" attribute), to then assign these to symbols with the linker scripts to > > iterate the these sections as an array. > > > > However, gcc uses different alignments for these structures when they are > > defined statically and when they are globally visible and/or in an array. > > Therefore iteration on these arrays sees "holes" of padding. > > > > Use the __u64_aligned for type declarations and variable definitions to ensure > > that gcc: > > > > a) iterates on the correctly aligned type. (type attribute) > > b) generates the definitions within the sections with the appropriate alignment. > > (variable attribute) > > > > The Ftrace code introduced the "aligned(4)" variable attribute in commit > > 1473e4417c79f12d91ef91a469699bfa911f510f to try to work around this problem that > > showed up on x86_64, but it causes unaligned accesses on sparc64, and is > > generally a bad idea on 64-bit if RCU pointers are contained within the > > structure. Moreover, it did not use the same attribute as type attribute, which > > could cause the iteration on the extern array structure not to match the > > variable definitions for some structure sizes. > > > > We should also ensure proper alignment of each Ftrace section in > > include/asm-generic/vmlinux.lds.h. > > > > Moving all STRUCT_ALIGN() for FTRACE_EVENTS() and TRACE_SYSCALLS() into the > > definitions, so the alignment is only done if these infrastructures are > > configured in. Use U64_ALIGN instead of STRUCT_ALIGN. > > > > Also align TRACE_PRINTKS on U64_ALIGN to make sure the beginning of the section > > is aligned on pointer size. > > > > Signed-off-by: Mathieu Desnoyers > > Acked-by: David S. Miller Thanks! This patch, just like 1/3, should also go into -stable, for both .36 and .37. As I'm thinking, the following patch (for Tracepoints) should only go into the -tip tree, as it does not fix a problem -- it merely reduces the data footprint of the tracepoints structures. I'm therefore not forwarding patch 3/3 for inclusion into -stable. Mathieu -- Mathieu Desnoyers Operating System Efficiency R&D Consultant EfficiOS Inc. http://www.efficios.com From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752246Ab1AVRPw (ORCPT ); Sat, 22 Jan 2011 12:15:52 -0500 Received: from mail.openrapids.net ([64.15.138.104]:57544 "EHLO blackscsi.openrapids.net" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1751487Ab1AVRPv (ORCPT ); Sat, 22 Jan 2011 12:15:51 -0500 Date: Sat, 22 Jan 2011 12:15:49 -0500 From: Mathieu Desnoyers To: David Miller Cc: linux-kernel@vger.kernel.org, rostedt@goodmis.org, fweisbec@gmail.com, mingo@elte.hu, richm@oldelvet.org.uk, ben@decadent.org.uk, sparclinux@vger.kernel.org, stable Subject: Re: [patch 2/3] tracing: fix sparc64 alignment crash with __u64_aligned/U64_ALIGN() Message-ID: <20110122171549.GB31308@Krystal> References: <20110121203630.725922272@efficios.com> <20110121203642.884088920@efficios.com> <20110121.160509.146106348.davem@davemloft.net> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20110121.160509.146106348.davem@davemloft.net> X-Editor: vi X-Info: http://www.efficios.com X-Operating-System: Linux/2.6.26-2-686 (i686) X-Uptime: 12:11:38 up 59 days, 22:14, 5 users, load average: 0.13, 0.10, 0.04 User-Agent: Mutt/1.5.18 (2008-05-17) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org * David Miller (davem@davemloft.net) wrote: > From: Mathieu Desnoyers > Date: Fri, 21 Jan 2011 15:36:32 -0500 > > > Problem description: > > > > gcc happily align structures defined statically on 32-byte. Ftrace trace events > > and Tracepoints both statically define structures into sections (using the > > "section" attribute), to then assign these to symbols with the linker scripts to > > iterate the these sections as an array. > > > > However, gcc uses different alignments for these structures when they are > > defined statically and when they are globally visible and/or in an array. > > Therefore iteration on these arrays sees "holes" of padding. > > > > Use the __u64_aligned for type declarations and variable definitions to ensure > > that gcc: > > > > a) iterates on the correctly aligned type. (type attribute) > > b) generates the definitions within the sections with the appropriate alignment. > > (variable attribute) > > > > The Ftrace code introduced the "aligned(4)" variable attribute in commit > > 1473e4417c79f12d91ef91a469699bfa911f510f to try to work around this problem that > > showed up on x86_64, but it causes unaligned accesses on sparc64, and is > > generally a bad idea on 64-bit if RCU pointers are contained within the > > structure. Moreover, it did not use the same attribute as type attribute, which > > could cause the iteration on the extern array structure not to match the > > variable definitions for some structure sizes. > > > > We should also ensure proper alignment of each Ftrace section in > > include/asm-generic/vmlinux.lds.h. > > > > Moving all STRUCT_ALIGN() for FTRACE_EVENTS() and TRACE_SYSCALLS() into the > > definitions, so the alignment is only done if these infrastructures are > > configured in. Use U64_ALIGN instead of STRUCT_ALIGN. > > > > Also align TRACE_PRINTKS on U64_ALIGN to make sure the beginning of the section > > is aligned on pointer size. > > > > Signed-off-by: Mathieu Desnoyers > > Acked-by: David S. Miller Thanks! This patch, just like 1/3, should also go into -stable, for both .36 and .37. As I'm thinking, the following patch (for Tracepoints) should only go into the -tip tree, as it does not fix a problem -- it merely reduces the data footprint of the tracepoints structures. I'm therefore not forwarding patch 3/3 for inclusion into -stable. Mathieu -- Mathieu Desnoyers Operating System Efficiency R&D Consultant EfficiOS Inc. http://www.efficios.com