Return-path: Received: from bu3sch.de ([62.75.166.246]:53802 "EHLO vs166246.vserver.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750933AbZCSS4o (ORCPT ); Thu, 19 Mar 2009 14:56:44 -0400 From: Michael Buesch To: Randy Dunlap Subject: Re: linux-next: Tree for March 19 (wireless/p54) Date: Thu, 19 Mar 2009 19:55:00 +0100 Cc: Stephen Rothwell , linux-next@vger.kernel.org, LKML , "linux-wireless@vger.kernel.org" References: <20090319221024.5e2ad6e5.sfr@canb.auug.org.au> <200903191850.17825.mb@bu3sch.de> <49C293CC.6020009@oracle.com> In-Reply-To: <49C293CC.6020009@oracle.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Message-Id: <200903191955.00718.mb@bu3sch.de> (sfid-20090319_195647_938265_FE74A6AC) Sender: linux-wireless-owner@vger.kernel.org List-ID: On Thursday 19 March 2009 19:49:48 Randy Dunlap wrote: > Michael Buesch wrote: > > On Thursday 19 March 2009 18:28:30 Randy Dunlap wrote: > >> Stephen Rothwell wrote: > >>> Hi all, > >>> > >>> Changes since 20090318: > >> > >> ERROR: "led_classdev_register" [drivers/net/wireless/p54/p54common.ko] undefined! > >> ERROR: "led_classdev_unregister" [drivers/net/wireless/p54/p54common.ko] undefined! > >> > >> One possible option (fix) is to have CONFIG_MAC80211_LEDS also select > >> LEDS_CLASS (that would build the missing functions). > > > > Does p54 select CONFIG_MAC80211_LEDS? If that's the case, I think it should > > No, it doesn't. p54common.c just has a large block of code that is > surrounded by > #ifdef CONFIG_MAC80211_LEDS > ... > #endif Yeah. I do think that it'd be wrong for mac80211 to select LEDS_CLASS. mac80211 implements the triggers and the lowlevel drivers implement the LED class device. So mac80211 does not depend on LEDS_CLASS at all. So p54 should probably introduce an automagic config option like b43 does, or it should do something (ugly) like this: #if defined(CONFIG_MAC80211_LEDS) && (defined(CONFIG_LEDS_CLASS) || defined(CONFIG_LEDS_CLASS_MODULE)) ... #endif -- Greetings, Michael.