From: Trond Myklebust Subject: Re: [PATCH 5/5] NFS: NFS_V4 and NFSD_V4 should depend on SUNRPC_GSS Date: Mon, 17 Mar 2008 14:31:44 -0400 Message-ID: <1205778704.11303.95.camel@heimdal.trondhjem.org> References: <20080314181525.29889.35370.stgit@manray.1015granger.net> <1205769601.11303.10.camel@heimdal.trondhjem.org> <0951705D-9F24-4E6F-B034-255C1A6655DB@oracle.com> Mime-Version: 1.0 Content-Type: text/plain Cc: linux-nfs@vger.kernel.org To: Chuck Lever Return-path: Received: from mx2.netapp.com ([216.240.18.37]:56259 "EHLO mx2.netapp.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752420AbYCQSb7 (ORCPT ); Mon, 17 Mar 2008 14:31:59 -0400 In-Reply-To: <0951705D-9F24-4E6F-B034-255C1A6655DB@oracle.com> Sender: linux-nfs-owner@vger.kernel.org List-ID: On Mon, 2008-03-17 at 12:59 -0400, Chuck Lever wrote: > > Selecting SUNRPC_GSS is _not_ equivalent to what we have today. > > Currently, if you select NFSv4, it will force compilation of the > > KerberosV mechanism too. AFAICS, this patch regresses that behaviour. > > > Enforcing that behavior breaks the rules of Kconfig. The current _implementation_ may break the rules, but the behaviour is definitely correct and should not be changed. > The purpose of Kconfig is to allow the kernel to build correctly. > The resulting configuration does not have to make sense from a run- > time perspective. While Kconfig may allow it, that still doesn't make it a good policy. > Normally the way this sort of thing is handled is that a note is left > in the help text that says "Oh, and you will find enabling Kerberos V > support useful since the spec requires it." Adding a "default m && > (NFSV4 || NFSD_V4)" or something like that under Kerberos GSS might > work as well. config RPCSEC_GSS_KRB5 tristate "Secure RPC: Kerberos V mechanism (EXPERIMENTAL)" depends on SUNRPC && EXPERIMENTAL default y if (NFS_V4 && NFS_FS=y) || (NFSD_V4 && NFSD=y) default m if (NFS_V4 && NFS_FS=m) || (NFSD_V4 && NFSD=m) select ... > But the point is, what I'm removing is not a build dependency; what > I'm replacing it with is a build dependency. See above. "It compiles but doesn't run" isn't acceptable. -- Trond Myklebust Linux NFS client maintainer NetApp Trond.Myklebust@netapp.com www.netapp.com