From mboxrd@z Thu Jan 1 00:00:00 1970 From: khilman@deeprootsystems.com (Kevin Hilman) Date: Fri, 10 Dec 2010 08:56:28 -0800 Subject: [PATCH v7 11/12] davinci: add tnetv107x evm backlight device In-Reply-To: (Sekhar Nori's message of "Fri, 10 Dec 2010 21:31:59 +0530") References: <1291733522-3626-1-git-send-email-cyril@ti.com> <1291733522-3626-12-git-send-email-cyril@ti.com> <4D00EEA1.8070201@ti.com> Message-ID: <87aakdd0sj.fsf@deeprootsystems.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org "Nori, Sekhar" writes: > On Thu, Dec 09, 2010 at 20:28:41, Chemparathy, Cyril wrote: >> On 12/09/2010 06:00 AM, Nori, Sekhar wrote: >> > On Thu, Dec 09, 2010 at 14:25:49, Nori, Sekhar wrote: >> > >> >> This call should simply return if machine is not tnetv107x EVM. >> >> >> >> I didn't follow the entire series but wondering why >> >> platform device registration should be a late init call. >> >> Typically the driver probe can be made a late init call >> >> in case of init sequence dependencies. >> > >> > I meant driver init, not probe. >> >> This was done to handle driver dependencies, and is meant to be >> temporary (until a real driver dependency framework comes in). >> >> Please see [1] for a bit more of background on changing drivers to use >> module_init(). > > Hmm, okay. Please include a check for machine in that case. > Sekhar is right. Today, we don't (can't) build this machine with others for a couple reasons, but we're moving towards removing those restrictions, so when this machine is included in a binary with others, it needs to check. Thanks, Kevin