From: Willy Tarreau Subject: Re: Broken userspace crypto in linux-4.1.18 Date: Thu, 18 Feb 2016 00:33:53 +0100 Message-ID: <20160217233353.GC31125@1wt.eu> References: <56C47DF9.6030704@whissi.de> <56C485AA.5060303@oracle.com> <56C49096.6040904@whissi.de> <56C4F068.8060806@oracle.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: "herbert@gondor.apana.org.au" , "dvyukov@google.com" , "stable@vger.kernel.org" , "linux-crypto@vger.kernel.org" To: Sasha Levin , "Thomas D." Return-path: Content-Disposition: inline In-Reply-To: <56C4F068.8060806@oracle.com> Sender: stable-owner@vger.kernel.org List-Id: linux-crypto.vger.kernel.org On Wed, Feb 17, 2016 at 05:12:56PM -0500, Sasha Levin wrote: > On 02/17/2016 10:24 AM, Thomas D. wrote: > > Hi, > > > > Sasha Levin wrote: > >> > So either the upstream patch is broken, or the 4.1 backport is wrong/missing > >> > dependency/missing fix. > >> > > >> > Any chance you could try 4.5-rc3 and see if that works for you? That'll narrow > >> > it down a lot. > > 4.5-rc3 works for me. > > > > Linux-4.4.0 works, too. > > Herbert, > > Is there a dependency I missed in 4.1? I don't really see anything that > could have gone wrong there. Or maybe Thomas can run a bisect ? Thomas, do you feel comfortable with doing this ? There are a few patches between 4.1.17 and 4.1.18, in 8 steps it should be done, and only 4 if you limit yourself to the crypto directory. Stable versions are fast at bisecting because each patch touches a very isolated area so only the first build takes a bit of time but the other ones are often pretty fast. Finding the faulty patch can help spotting if something is wrong in its backport. Regards, Willy