From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753522AbbJNRGF (ORCPT ); Wed, 14 Oct 2015 13:06:05 -0400 Received: from e32.co.us.ibm.com ([32.97.110.150]:43562 "EHLO e32.co.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752222AbbJNRGC (ORCPT ); Wed, 14 Oct 2015 13:06:02 -0400 X-IBM-Helo: d03dlp01.boulder.ibm.com X-IBM-MailFrom: paulmck@linux.vnet.ibm.com X-IBM-RcptTo: linux-kernel@vger.kernel.org;linux-tip-commits@vger.kernel.org Date: Wed, 14 Oct 2015 10:06:05 -0700 From: "Paul E. McKenney" To: Dmitry Vyukov Cc: Andrey Ryabinin , tip-bot for Andrey Ryabinin , linux-tip-commits@vger.kernel.org, kasan-dev , Ingo Molnar , Kostya Serebryany , Borislav Petkov , Andrew Morton , LKML , Peter Zijlstra , Andy Lutomirski , Linus Torvalds , Thomas Gleixner , Sasha Levin , Denys Vlasenko , Wolfram Gloger , Andrey Konovalov , "H. Peter Anvin" , Alexander Potapenko Subject: Re: [tip:locking/urgent] compiler, atomics: Provide READ_ONCE_NOCHECK () Message-ID: <20151014170605.GZ3910@linux.vnet.ibm.com> Reply-To: paulmck@linux.vnet.ibm.com References: <1444750088-24444-2-git-send-email-aryabinin@virtuozzo.com> <20151014154532.GV3910@linux.vnet.ibm.com> <561E807C.50109@virtuozzo.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.21 (2010-09-15) X-TM-AS-MML: disable X-Content-Scanned: Fidelis XPS MAILER x-cbid: 15101417-0005-0000-0000-000018F58E06 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Oct 14, 2015 at 06:29:59PM +0200, Dmitry Vyukov wrote: > On Wed, Oct 14, 2015 at 6:19 PM, Andrey Ryabinin > wrote: > > > > > > On 10/14/2015 06:50 PM, Dmitry Vyukov wrote: > >> On Wed, Oct 14, 2015 at 5:45 PM, Paul E. McKenney > >> wrote: > >>> On Wed, Oct 14, 2015 at 08:28:43AM -0700, tip-bot for Andrey Ryabinin wrote: > >>>> Commit-ID: 4115ffdf4d6f8986a7abe1dd522c163f599bc0e6 > >>>> Gitweb: http://git.kernel.org/tip/4115ffdf4d6f8986a7abe1dd522c163f599bc0e6 > >>>> Author: Andrey Ryabinin > >>>> AuthorDate: Tue, 13 Oct 2015 18:28:07 +0300 > >>>> Committer: Ingo Molnar > >>>> CommitDate: Wed, 14 Oct 2015 16:44:06 +0200 > >>>> > >>>> compiler, atomics: Provide READ_ONCE_NOCHECK() > >>>> > >>>> Some code may perform racy by design memory reads. This could be > >>>> harmless, yet such code may produce KASAN warnings. > >>>> > >>>> To hide such accesses from KASAN this patch introduces > >>>> READ_ONCE_NOCHECK() macro. KASAN will not check the memory > >>>> accessed by READ_ONCE_NOCHECK(). > >>>> > >>>> This patch creates __read_once_size_nocheck() a clone of > >>>> __read_once_size_check() (renamed __read_once_size()). > >>>> The only difference between them is 'no_sanitized_address' > >>>> attribute appended to '*_nocheck' function. This attribute tells > >>>> the compiler that instrumentation of memory accesses should not > >>>> be applied to that function. We declare it as static > >>>> '__maybe_unsed' because GCC is not capable to inline such > >>>> function: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67368 > >>>> > >>>> With KASAN=n READ_ONCE_NOCHECK() is just a clone of READ_ONCE(). > >>> > >>> So I add READ_ONCE_NOCHECK() for accesses for which the compiler cannot > >>> prove safe address for KASAN's benefit, but READ_ONCE() suffices for > >>> the data-race-detection logic in KTSAN, correct? > >> > >> KTSAN also needs READ_ONCE_NOCHECK() here. > > > > Does it? What's the difference between READ_ONCE_NOCHECK() and READ_ONCE() with KTSAN=y? > > AFAIK READ_ONCE() is sufficient to hide race from KTSAN. It doesn't *require* READ_ONCE_NOCHECK(), right? > > > For not there is no difference, because you just added > READ_ONCE_NOCHECK and we have not yet supported it. > But my plan is to completely ignore accessed from READ_ONCE_NOCHECK in > KTSAN so that they never lead to race reports. > > READ_ONCE in get_wchan still can lead to a data race report, because > it is READ_ONCE in get_wchan versus a normal write to stack in the > other thread. That is not atomic and not generally safe. Where possible, it would be better to make the normal write instead be WRITE_ONCE(). That might well not be possible here, but let's not be too aggressive about silencing KTSAN. Thanx, Paul