From: Andi Kleen Subject: Re: [RFC PATCH 15/35] subarch support for controlling interrupt delivery Date: Wed, 10 May 2006 12:54:26 +0200 Message-ID: <200605101254.26870.ak@suse.de> References: <20060509084945.373541000@sous-sol.org> <200605092356.28818.ak@suse.de> <20060510103520.GX7834@cl.cam.ac.uk> Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20060510103520.GX7834@cl.cam.ac.uk> Content-Disposition: inline List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Christian Limpach Cc: xen-devel@lists.xensource.com, Ian Pratt , linux-kernel@vger.kernel.org, "Martin J. Bligh" , Chris Wright , virtualization@lists.osdl.org List-Id: xen-devel@lists.xenproject.org es, which is why I measured that one as well. > > Now, the original concern was that we have the five operations implemented > as multi-line macros and doing a hybrid solution doesn't really address > that. If it's straight-forward to convert to an inline do it. If not keep it as a macro. After all code style is just a tool, not something self serving. > > Also, it's not quite clear to me what's the best way to turn three of > the five into functions, whether inline or not. > > For measuring the sizes, I did the following: > add void ___restore_flags(unsigned long *x) with the implementation > and then: > #define __restore_flags(x) ___restore_flags(&(x)) Yes that is the standard way to do it > Alternatively, would it make sense to change __restore_flags to take > a pointer to flags instead? That would be quite an invasive change... No. -Andi