Return-path: Received: from mail.hevs.ch ([153.109.23.10]:50051 "EHLO mail.hevs.ch" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750986AbYFTHJh (ORCPT ); Fri, 20 Jun 2008 03:09:37 -0400 From: Marc Pignat To: libertas-dev@lists.infradead.org Subject: Re: [PATCH] libertas: make mesh code configurable Date: Fri, 20 Jun 2008 08:48:27 +0200 Cc: Holger Schurig , "John W. Linville" , Dan Williams , linux-wireless@vger.kernel.org References: <200806191704.10297.hs4233@mail.mn-solutions.de> In-Reply-To: <200806191704.10297.hs4233@mail.mn-solutions.de> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Message-Id: <200806200848.28077.marc.pignat@hevs.ch> (sfid-20080620_090939_949864_AEF2A656) Sender: linux-wireless-owner@vger.kernel.org List-ID: Hi! On Thursday 19 June 2008, Holger Schurig wrote: > There are no known firmwares for CF and SDIO based devices that > support MESH. For those devices, mesh-support in the driver is just > a bit of bloat. Moreover, they're mostly used in embedded devices, > where space counts. > > This patch saves about 10 kB in .text and 288 bytes in .data: > > 129542 4012 28 133582 209ce libertas.ko > 119279 3724 28 123031 1e097 nomesh/libertas.ko > > 8142 840 0 8982 2316 libertas_cs.ko > 8134 840 0 8974 230e nomesh/libertas_cs.ko > > 9516 700 8 10224 27f0 libertas_sdio.ko > 9516 700 8 10224 27f0 nomesh/libertas_sdio.ko > > 10333 860 0 11193 2bb9 usb8xxx.ko > 10333 860 0 11193 2bb9 nomesh/usb8xxx.ko > > Signed-off-by: Holger Schurig Looking at the numbers, this seems a good idea. But, looking at the code, there are so many #ifdefs! These #ifdefs have a cost: * #ifdefs decrease code readability * every future patch should be (compile AND runtime) tested with the mesh code enable and disable. IHMO, the cost versus benefits is far to high. Best regards Marc