From mboxrd@z Thu Jan 1 00:00:00 1970 From: Uwe =?iso-8859-1?Q?Kleine-K=F6nig?= Subject: Re: [PATCH] sunrpc: prompt for RPCSEC_GSS_KRB5 even if NFS_V4 is enabled Date: Mon, 27 Sep 2010 14:54:04 +0200 Message-ID: <20100927125404.GB6271@pengutronix.de> References: <20100909181454.GB14915@pengutronix.de> <1285584120-16860-1-git-send-email-u.kleine-koenig@pengutronix.de> <1285587572.19362.2.camel@heimdal.trondhjem.org> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: linux-kernel@vger.kernel.org, "J. Bruce Fields" , Neil Brown , "David S. Miller" , Randy Dunlap , linux-nfs@vger.kernel.org, netdev@vger.kernel.org To: Trond Myklebust Return-path: Content-Disposition: inline In-Reply-To: <1285587572.19362.2.camel@heimdal.trondhjem.org> Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org Hi Trond, On Mon, Sep 27, 2010 at 07:39:32AM -0400, Trond Myklebust wrote: > On Mon, 2010-09-27 at 12:41 +0200, Uwe Kleine-K=F6nig wrote: > > NFS_V4 works fine without RPCSEC_GSS_KRB5 (even without CRYPTO). > > This dependency was introduced in > >=20 > > df486a2 (NFS: Fix the selection of security flavours in Kconfig) > >=20 > > to fix a build failure as RPCSEC_GSS_KRB5 was thought to be needed = for > > NFS_V4. The fix didn't work completely as NFS_V4 didn't enforce CR= YPTO > > and so the select on RPCSEC_GSS_KRB5 didn't work in all situations = (e.g. > > arm/mx1_defconfig). > >=20 > > This was rectified by > >=20 > > 827e345 (SUNRPC: Fix the NFSv4 and RPCSEC_GSS Kconfig dependencies= ) > >=20 > > but the magic for RPCSEC_GSS_KRB5 introduced by df486a2 wasn't reve= rted. > >=20 > > Cc: Trond Myklebust > > Signed-off-by: Uwe Kleine-K=F6nig > > --- > > Hello, > >=20 > > after Trond sent me the patch that later ended in 827e345702 I sugg= ested > > to fold the patch below into it[1], but without reaction and succes= s as I > > noticed just now. :-( > >=20 > > Best regards > > Uwe >=20 > That's because you completely fail to justify why should we change th= e > behaviour to suddenly make RPCSEC_GSS_KRB5 optional for NFSv4. That h= as > never been the case before. My intention is not to make "RPCSEC_GSS_KRB5 optional for NFSv4". Firs= t I saw a build failure and then I wondered if the fix was optimal. Afte= r reading the log of 827e345 I thought NFSv4 doesn't depend on RPCSEC_GSS_KRB5, still more considering that 827e345 was your fix after I suggested to select CRYPTO to enforce RPCSEC_GSS_KRB5 again. Currently you can have NFSv4 without RPCSEC_GSS_KRB5 because if you don't have CRYPTO RPCSEC_GSS_KRB5 is off, too, even if it defaults to yes and there's no prompt. (Selecting would not work, too.) And note that RPCSEC_GSS_KRB5 already selects SUNRPC_GSS, so 827e345 doesn't do anything useful if NFS_V4 really needs RPCSEC_GSS_KRB5. So either we should really enforce RPCSEC_GSS_KRB5 if NFS_V4 is selecte= d (by letting one of these select CRYPTO, see e.g. my first patch, or by letting NFS_V4 depend on CRYPTO) or make it optional in all cases (as i= t is already now in some cases). Best regards Uwe --=20 Pengutronix e.K. | Uwe Kleine-K=F6nig = | Industrial Linux Solutions | http://www.pengutronix.de/= |