From mboxrd@z Thu Jan 1 00:00:00 1970 From: Randy Dunlap Subject: Re: [PATCH -next] v4l2: handle unregister for non-I2C builds Date: Fri, 22 May 2009 08:57:08 -0700 Message-ID: <4A16CB54.9000705@oracle.com> References: <20090511161442.3e9d9cb9.sfr@canb.auug.org.au> <4A085455.5040108@oracle.com> <20090522054847.GB14059@linux-sh.org> <20090522175554.19465733.sfr@canb.auug.org.au> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: Received: from rcsinet12.oracle.com ([148.87.113.124]:37428 "EHLO rgminet12.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757316AbZEVPyw (ORCPT ); Fri, 22 May 2009 11:54:52 -0400 In-Reply-To: <20090522175554.19465733.sfr@canb.auug.org.au> Sender: linux-next-owner@vger.kernel.org List-ID: To: Stephen Rothwell Cc: Mauro Carvalho Chehab , Paul Mundt , linux-next@vger.kernel.org, LKML , linux-media@vger.kernel.org, Hans Verkuil Stephen Rothwell wrote: > On Fri, 22 May 2009 14:48:47 +0900 Paul Mundt wrote: >> On Mon, May 11, 2009 at 09:37:41AM -0700, Randy Dunlap wrote: >>> From: Randy Dunlap >>> >>> Build fails when CONFIG_I2C=n, so handle that case in the if block: >>> >>> drivers/built-in.o: In function `v4l2_device_unregister': >>> (.text+0x157821): undefined reference to `i2c_unregister_device' >>> >>> Signed-off-by: Randy Dunlap >> This patch still has not been applied as far as I can tell, and builds >> are still broken as a result, almost 2 weeks after the fact. > > In fact there has been no updates to the v4l-dvb tree at all since > May 11. Mauro? I got an 'hg' autocommit message for the patch on May-11, but that's all I've seen/heard about it. I'm really concerned about how slowly patches are merged for linux-next trees... and not just for drivers/media/ > I have reverted the patch that caused the build breakage ... (commit > d5bc7940d39649210f1affac1fa32f253cc45a81 "V4L/DVB (11673): v4l2-device: > unregister i2c_clients when unregistering the v4l2_device"). > > [By the way, an alternative fix might be to just define > V4L2_SUBDEV_FL_IS_I2C to be zero if CONFIG_I2C and CONFIG_I2C_MODULE are > not defined (gcc should then just elide the offending code).] -- ~Randy LPC 2009, Sept. 23-25, Portland, Oregon http://linuxplumbersconf.org/2009/