From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ben Myers Subject: Re: [RFC v2] Unicode/UTF-8 support for XFS Date: Tue, 23 Sep 2014 15:45:13 -0500 Message-ID: <20140923204512.GI4482@sgi.com> References: <20140918195650.GI19952@sgi.com> <87lhpbhfgg.fsf@tassilo.jf.intel.com> <20140922184145.GH4482@sgi.com> <20140922192958.GJ4120@two.firstfloor.org> <54219C17.3090104@sgi.com> <20140923201540.GB15923@two.firstfloor.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Olaf Weber , linux-fsdevel@vger.kernel.org, tinguely@sgi.com, xfs@oss.sgi.com To: Andi Kleen Return-path: Received: from relay3.sgi.com ([192.48.152.1]:42487 "EHLO relay.sgi.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1753090AbaIWUpO (ORCPT ); Tue, 23 Sep 2014 16:45:14 -0400 Content-Disposition: inline In-Reply-To: <20140923201540.GB15923@two.firstfloor.org> Sender: linux-fsdevel-owner@vger.kernel.org List-ID: Hi Andi, On Tue, Sep 23, 2014 at 10:15:40PM +0200, Andi Kleen wrote: > > You only pay the space cost if you use it, similar to the nls tables. > > The way Linux module loading works these things are loaded by default, > not when someone needs it. > > So no, you (or rather every unfortunate Linux XFS user) would pay it always, > unless you black list the module or rebuild the kernel. The suggestion to use symbol_get was made in response to the initial post of the series. In my testing the normalization module did not load until I attempted to mount a filesystem which required it. Seems to work ok. See patch 10, "xfs: implement demand load of utf8norm.ko". Thanks, Ben