From: James Morris Subject: Re: [GIT PULL] KEYS: Fixes and crypto fixes Date: Thu, 28 Sep 2017 12:08:36 +1000 (AEST) Message-ID: References: <28036.1506547164@warthog.procyon.org.uk> <20170928001529.GA120911@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Cc: David Howells , Eric Biggers , "Jason A. Donenfeld" , Michael Halcrow , keyrings@vger.kernel.org, linux-security-module@vger.kernel.org, linux-crypto@vger.kernel.org, linux-kernel@vger.kernel.org, Herbert Xu To: Eric Biggers Return-path: Received: from namei.org ([65.99.196.166]:50717 "EHLO namei.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752494AbdI1CIw (ORCPT ); Wed, 27 Sep 2017 22:08:52 -0400 In-Reply-To: <20170928001529.GA120911@gmail.com> Sender: linux-crypto-owner@vger.kernel.org List-ID: On Wed, 27 Sep 2017, Eric Biggers wrote: > On Thu, Sep 28, 2017 at 09:14:58AM +1000, James Morris wrote: > > On Wed, 27 Sep 2017, David Howells wrote: > > > > > (2) Fixing big_key to use safe crypto from Jason A. Donenfeld. > > > > > > > I'm concerned about the lack of crypto review mentioned by Jason -- I > > wonder if we can get this rewrite any more review from crypto folk. > > > > Also, are there any tests for this code? If not, it would be good to make > > some. > > > > There is a test for the big_key key type in the keyutils test suite. I also > manually tested Jason's change. And as far as I can tell there isn't actually a > whole lot to test besides adding a big_key larger than BIG_KEY_FILE_THRESHOLD > bytes, reading it back, and verifying that the data is unchanged --- since that > covers the code that was changed. An earlier version of the patch produced a > warning with CONFIG_DEBUG_SG=y since it put the aead_request on the stack, but > that's been fixed. > Ok, thanks a lot. > It would be great if someone else would comment on the crypto too, but for what > it's worth I'm satisfied with the crypto changes. GCM is a much better choice > than ECB as long as we don't repeat (key, IV) pairs --- which we don't. And in > any case ECB mode makes no sense in this context; you'd need a *very* good > reason to actually choose to encrypt something with ECB mode. Unfortunately it > tends to be a favorite of people who don't understand encryption modes... Adding Herbert. -- James Morris