From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753364AbdHIPIn (ORCPT ); Wed, 9 Aug 2017 11:08:43 -0400 Received: from merlin.infradead.org ([205.233.59.134]:54550 "EHLO merlin.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753020AbdHIPIm (ORCPT ); Wed, 9 Aug 2017 11:08:42 -0400 Date: Wed, 9 Aug 2017 17:08:34 +0200 From: Peter Zijlstra To: Mel Gorman Cc: Mark Rutland , Thomas Gleixner , Linus Torvalds , Ingo Molnar , Davidlohr Bueso , Hugh Dickins , Sebastian Andrzej Siewior , linux-kernel@vger.kernel.org Subject: Re: [PATCH] futex: Remove unnecessary warning from get_futex_key Message-ID: <20170809150834.ostlfg3kw2ecwlgf@hirez.programming.kicks-ass.net> References: <20170809072711.np2ayrasrogph5tm@suse.de> <20170809140519.GE29494@leverpostej> <20170809144309.no7x6hpj6rgwfnvg@suse.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20170809144309.no7x6hpj6rgwfnvg@suse.de> User-Agent: NeoMutt/20170609 (1.8.3) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Aug 09, 2017 at 03:43:09PM +0100, Mel Gorman wrote: > On Wed, Aug 09, 2017 at 03:05:19PM +0100, Mark Rutland wrote: > > > diff --git a/kernel/futex.c b/kernel/futex.c > > > index 16dbe4c93895..f50b434756c1 100644 > > > --- a/kernel/futex.c > > > +++ b/kernel/futex.c > > > @@ -670,13 +670,14 @@ get_futex_key(u32 __user *uaddr, int fshared, union futex_key *key, int rw) > > > * this reference was taken by ihold under the page lock > > > * pinning the inode in place so i_lock was unnecessary. The > > > * only way for this check to fail is if the inode was > > > - * truncated in parallel so warn for now if this happens. > > > + * truncated in parallel which is almost certainly an > > > + * application bug. In such a case, just retry. > > > * > > > * We are not calling into get_futex_key_refs() in file-backed > > > * cases, therefore a successful atomic_inc return below will > > > * guarantee that get_futex_key() will still imply smp_mb(); (B). > > > */ > > > - if (WARN_ON_ONCE(!atomic_inc_not_zero(&inode->i_count))) { > > > + if (!atomic_inc_not_zero(&inode->i_count)) { > > > > I applied the same diff yesterday, and haven't seen anything go wrong > > with my test case and/or with Syzkaller running, so FWIW: > > > > Tested-by: Mark Rutland > > > > Thanks for putting this together! > > > > No problem. FWIW, I had the test case running for 12 hours in a loop as > well and other than having to adjust the number of threads doing futex() > to trigger the warning without the patch, I observed no other problems. > If Thomas is happy, I hope this can be merged for 4.13 (or picked up > directly by Linus if he feels like it). Even if it's delayed, I'll resubmit > to -stable manually if the "Cc: stable" gets stripped along the way. Probably best if Linus picks this up directly as Thomas is on holidays. In any case, Acked-by: Peter Zijlstra (Intel)