From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751096AbWAJPCm (ORCPT ); Tue, 10 Jan 2006 10:02:42 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751099AbWAJPCm (ORCPT ); Tue, 10 Jan 2006 10:02:42 -0500 Received: from cabal.ca ([134.117.69.58]:6813 "EHLO fattire.cabal.ca") by vger.kernel.org with ESMTP id S1751096AbWAJPCl (ORCPT ); Tue, 10 Jan 2006 10:02:41 -0500 Date: Tue, 10 Jan 2006 10:01:41 -0500 From: Kyle McMartin To: Christoph Hellwig Cc: akpm@osdl.org, carlos@parisc-linux.org, willy@parisc-linux.org, linux-kernel@vger.kernel.org, parisc-linux@lists.parisc-linux.org Subject: Re: [parisc-linux] [PATCH 1/5] Add generic compat_siginfo_t Message-ID: <20060110150141.GE28306@quicksilver.road.mcmartin.ca> References: <20060108193755.GH3782@tachyon.int.mcmartin.ca> <20060109141355.GA22296@lst.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20060109141355.GA22296@lst.de> User-Agent: Mutt/1.5.11 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Mon, Jan 09, 2006 at 03:13:55PM +0100, Christoph Hellwig wrote: > -mm already has a much better implementation for compat_sys_timer_create > that doesn't require all the sigevent churn in this patch (which btw > doesn't seem to be mentioned in the changelog at all). Ok, I see it in -mm now, and it does seem to be nicer code, I'll try it out on parisc64. > But even with > that remove there seems to be a lot of useless ifdef and indirection > in this patch. I agree, I'm really just trying to shepard this home so we don't have to maintain it out of tree. I'm not overly attached to the code, if I can make parisc64 work with your compat signal bits, I'll be just as happy. The one thing from this patchset I'd like to keep is the is_compat_task() as it does provide nice cleanups Cheers, Kyle > Over the next days I'll send out my generic compat > singal bits which don't require all this, but otoh require every > architecture to supply helpers. If you can make those generic without > all the ifdef an additional header bits all power to you!