From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from cvs.linux-mips.org (eddie.linux-mips.org [78.24.191.182]) by ozlabs.org (Postfix) with ESMTP id 98145B70FE for ; Fri, 3 Sep 2010 02:43:39 +1000 (EST) Received: from localhost.localdomain ([127.0.0.1]:33471 "EHLO localhost.localdomain" rhost-flags-OK-OK-OK-OK) by eddie.linux-mips.org with ESMTP id S1491020Ab0IBQnf (ORCPT ); Thu, 2 Sep 2010 18:43:35 +0200 Date: Thu, 2 Sep 2010 17:43:35 +0100 (BST) From: "Maciej W. Rozycki" To: David Miller Subject: Re: ERR_PTR pattern in phylib In-Reply-To: <20100902.093451.189703514.davem@davemloft.net> Message-ID: References: <20100902.093451.189703514.davem@davemloft.net> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Cc: netdev@vger.kernel.org, afleming@freescale.com, linuxppc-dev@lists.ozlabs.org List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Thu, 2 Sep 2010, David Miller wrote: > > Arguably using a union here would make things cleaner and any reasonable > > ABI will place small unions used as arguments or return values in > > registers, but I'm not sure if the cost of the rewrite is worth the > > Well, sparc 32-bit's ABI for one is "not reasonable" and these return > values will always get stack slots and passed back by reference. So that makes it two -- together with MIPS o32. ;) Still used by everybody concerned as contraptions tend to stick. :/ That doesn't make it reasonable though. Nor me to push the union approach -- that's an elegant approach that would probably give us less than take in return. Maciej