From: "J. Bruce Fields" Subject: Re: [PATCH 2/5] nfsd: Fix independence of a few nfsd related headers Date: Thu, 5 Nov 2009 11:33:33 -0500 Message-ID: <20091105163333.GB29981@fieldses.org> References: <4ADEC1EF.8040107@panasas.com> <1256112873-32495-1-git-send-email-bharrosh@panasas.com> <1256171298.6809.1.camel@heimdal.trondhjem.org> <4AE01569.9000002@panasas.com> <1256220146.6402.23.camel@heimdal.trondhjem.org> <4AE08165.2040100@panasas.com> <20091104220927.GS11637@fieldses.org> <4AF2914F.60803@panasas.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Trond Myklebust , Benny Halevy , pNFS Mailing List , NFS list , Andy Adamson To: Boaz Harrosh Return-path: Received: from fieldses.org ([174.143.236.118]:52883 "EHLO fieldses.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756623AbZKEQd0 (ORCPT ); Thu, 5 Nov 2009 11:33:26 -0500 In-Reply-To: <4AF2914F.60803@panasas.com> Sender: linux-nfs-owner@vger.kernel.org List-ID: On Thu, Nov 05, 2009 at 10:48:15AM +0200, Boaz Harrosh wrote: > On 11/05/2009 12:09 AM, J. Bruce Fields wrote: > > On Thu, Oct 22, 2009 at 05:59:33PM +0200, Boaz Harrosh wrote: > >> On 10/22/2009 04:02 PM, Trond Myklebust wrote: > >>> No. What I'm saying is that this doesn't have to be an absolute rule. > >>> The Kernel style guide assumes that everything in 'include/*' is going > >>> to be exported all around the kernel. > >>> The problem is that we put a lot of stuff which is private to fs/nfs and > >>> fs/nfsd in there. Those header files do not have to absolutely follow > >>> the style guide rule, 'cos we know what is being included before and > >>> after them... > >>> > >> > >> I'm not sure I understand > >> You are saying that the patches are very good, but only > >> the rule I stated originally could be relaxed a little with private > >> headers where we might get lazy, if the effects are very local? > >> > >> Well, that's not a problem then, right? just that I can relax a bit > >> if I want. > >> > >> But I disagree: see 3, 4, 5 above and that last patch I submitted. That patch > >> is only the beginning. 85% of all source files in nfs/nfsd could receive the > >> same love. I only done these I touched. Code tends to stay much-much longer > >> then we spend time on it. Better get it in shape the first time. > > > > I'm assuming Trond's objection is just to the patch changelog > > (specifically, to the statement that any header "should be compilation > > independent"), not to these specific changes. > > > > --b. > > Speaking of which, Bruce I have a question. > > There are a few files in include/linux/ that define xdr definitions > these are used by exportfs nfs and nfsd. Some of it gets exposed > to filesystems. Which specifically? > With the pNFS tree we are adding lots more of these, > specifically I'm now to move the pnfs_osd_xdr stuff as well, and blocks. > > Could I open up a new include/linux/exportfs/ folder and put there any thing > xdr and exportfs related? Well, if it's just a few things, maybe include/linux/exportfs.h? But, sure, makes sense if there's includes that are specific to pNFS and that wouldn't be needed by any filesystem that included exportfs.h. > What should be the scope of the move, should any include/linux/ common > files used both by nfs & nfsd be moved there? I don't see why we'd want to do that, but maybe I'm missing something. --b.