Return-Path: linux-nfs-owner@vger.kernel.org Received: from nm5.bullet.mail.bf1.yahoo.com ([98.139.212.164]:21664 "HELO nm5.bullet.mail.bf1.yahoo.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with SMTP id S1760216Ab3DCGmn convert rfc822-to-8bit (ORCPT ); Wed, 3 Apr 2013 02:42:43 -0400 References: <1364317202.2660.132.camel@willson.li.ssimo.org> <515B2E7D.1050005@RedHat.com> Message-ID: <1364970564.49524.YahooMailNeo@web161304.mail.bf1.yahoo.com> Date: Tue, 2 Apr 2013 23:29:24 -0700 (PDT) From: Alex Dubov Reply-To: Alex Dubov Subject: Re: Allow building nfs-utils directly against GSSAPI To: Steve Dickson Cc: "linux-nfs@vger.kernel.org" In-Reply-To: <515B2E7D.1050005@RedHat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Sender: linux-nfs-owner@vger.kernel.org List-ID: Hi, >>? >> I can augment my patch to make use of your changes. Would you be willing to >> consider it? > I when ahead and took Simo's patches but I would like to continue to work > with on get your Heimdal build working again... What exactly broke? > I thought you were CCed of follow up messages as well. Basically, Simo's patches fix most of the problems with building on Heimdal (these problems came?from libgssglue direction). Only 3 problems remained, which I addressed in the follow-up patch: > 1. On some systems, only libroken.so is available (small fix to kerberos5.m4) >? > 2. krb5_util.c:check_for_target - Heimdal variant constructs a "pattern" >? ? principal and uses krb5_cc_retrieve_cred to get a matching credential. >? ? This should work on mit-krb5, so old method of iterating over every >? ? credential in cache may possibly be dropped outright and "#$if" guard >? ? omitted. >? ? For the sake of the above I reformatted the old approach to make it a bit >? ? more clear what's going on there. >? > 3. krb5_util.c:gssd_k5_err_msg - krb5_get_err_text is marked as deprecated, >? ? at least on Heimdal. If krb5_get_error_message is available, it should not >? ? be reached at all, thus "#elif" guard. Per issue 2, Simo told me he's going to look at it himself, with a view to remove "#if" branching altogether (present code at that location is mit-krb5 implementation specific). Issue 3 is somewhat not clear to me: should krb5_get_err_text stay at all? It's deprecated on Heimdal and apparently is not supported on mit-krb5 at all. Which other kerberos libraries may need to be supported by the code?