From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751865AbaIJO1m (ORCPT ); Wed, 10 Sep 2014 10:27:42 -0400 Received: from mx1.redhat.com ([209.132.183.28]:56285 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751370AbaIJO1l (ORCPT ); Wed, 10 Sep 2014 10:27:41 -0400 Date: Wed, 10 Sep 2014 10:27:12 -0400 From: Dave Jones To: "Theodore Ts'o" , Dan Carpenter , Andrew Morton , Jiri Kosina , Christoph Lameter , Pekka Enberg , David Rientjes , Joonsoo Kim , linux-kernel@vger.kernel.org, linux-mm@kvack.org Subject: Re: [PATCH] mm/sl[aou]b: make kfree() aware of error pointers Message-ID: <20140910142712.GA10785@redhat.com> Mail-Followup-To: Dave Jones , Theodore Ts'o , Dan Carpenter , Andrew Morton , Jiri Kosina , Christoph Lameter , Pekka Enberg , David Rientjes , Joonsoo Kim , linux-kernel@vger.kernel.org, linux-mm@kvack.org References: <20140909162114.44b3e98cf925f125e84a8a06@linux-foundation.org> <20140909221138.2587d864.akpm@linux-foundation.org> <20140910063630.GM6549@mwanda> <20140910135649.GB31903@thunk.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20140910135649.GB31903@thunk.org> User-Agent: Mutt/1.5.23 (2014-03-12) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Sep 10, 2014 at 09:56:49AM -0400, Theodore Ts'o wrote: > The ironic thing is that I asked Dan to add the feature to smatch > because I found two such bugs in ext4, and I suspected there would be > more. Sure enough, it found four more such bugs, including two in a > recent commit where I had found the first two bugs --- and I had > missed the other two even though I was specifically looking for such > instances. Oops. :-) > > Maybe we can add a debugging config option? I think having static > checkers plus some kmalloc failure testing should be sufficient to > prevent these sorts of problem from showing up. > > It would seem to me that this is the sort of thing that a static > checker should find reliably; Coverity has found things that were more > complex than what this should require, I think. I don't know if they > would be willing to add something this kernel-specific, though. (I've > added Dave Jones to the thread since he's been working a lot with > Coverity; Dave, what do you think?) It *might* be possible to rig up something using their modelling functionality, but I've not managed to make that work to my ends in the past. I suspect a runtime check would be more fruitful faster than they could implement kernel specific checkers & roll them out. Dave