From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S936473Ab3DHPgV (ORCPT ); Mon, 8 Apr 2013 11:36:21 -0400 Received: from mail-ea0-f173.google.com ([209.85.215.173]:45729 "EHLO mail-ea0-f173.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S936341Ab3DHPgT (ORCPT ); Mon, 8 Apr 2013 11:36:19 -0400 Date: Mon, 8 Apr 2013 17:36:14 +0200 From: Ingo Molnar To: Paul Bolle Cc: Thomas Gleixner , Martin Schwidefsky , Heiko Carstens , Ingo Molnar , Peter Zijlstra , linux390@de.ibm.com, linux-s390@vger.kernel.org, linux-kernel@vger.kernel.org Subject: Re: [PATCH] locking: remove three unused Kconfig symbols Message-ID: <20130408153614.GB17500@gmail.com> References: <1365428154.1830.99.camel@x61.thuisdomein> <20130408151609.GA13640@gmail.com> <1365435015.1830.113.camel@x61.thuisdomein> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1365435015.1830.113.camel@x61.thuisdomein> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org * Paul Bolle wrote: > On Mon, 2013-04-08 at 17:16 +0200, Ingo Molnar wrote: > > * Paul Bolle wrote: > > > The Kconfig symbols ARCH_INLINE_READ_UNLOCK_IRQ, > > > ARCH_INLINE_SPIN_UNLOCK_IRQ, and ARCH_INLINE_WRITE_UNLOCK_IRQ were added > > > in v2.6.33, but have never actually been used. It is safe to remove > > > these. > > > > > Erm, if you look at the code in question you'll see that they are unused I think > > due to a bug: > > > > config INLINE_READ_UNLOCK_IRQ > > def_bool y > > depends on !PREEMPT || ARCH_INLINE_READ_UNLOCK_BH > > > > Shouldn't that depend on ARCH_INLINE_READ_UNLOCK_IRQ? > > > > Similarly for the others. > > Yes, that seems correct. I must admit that I didn't spot that > possibility, but then again this Kconfig file is near unreadable (too > much repetition for human readers). Yeah, Kconfig isn't a very sophisticated language - it's the COBOL of config languages. > Would you have any idea why this problem wasn't noticed in 16 (!) releases? > Fedora 17, which I did this patch on, has Kconfig symbolk PREEMPT not set. Is it > perhaps commit to not enable that option? AFAICS it's a latent bug: it can only be noticed if these switches are set to different values by an arch - but the defaults and s390 (the two main cases) never do that. So it's not causing any runtime problems today - but should be fixed nevertheless, now that you noticed it! :-) Thanks, Ingo