From mboxrd@z Thu Jan 1 00:00:00 1970 From: Arnd Bergmann Subject: Re: [PATCH 1/3] ARM: tegra: Add AHB driver Date: Mon, 23 Apr 2012 18:32:15 +0000 Message-ID: <201204231832.16268.arnd@arndb.de> References: <1335181043-15348-1-git-send-email-hdoyu@nvidia.com> <201204231233.09423.arnd@arndb.de> <4F9598AD.2030308@wwwdotorg.org> Mime-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-15" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <4F9598AD.2030308-3lzwWm7+Weoh9ZMKESR00Q@public.gmane.org> Sender: linux-tegra-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Stephen Warren Cc: Hiroshi DOYU , linux-tegra-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, Colin Cross , Olof Johansson , Russell King , linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org, linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-Id: linux-tegra@vger.kernel.org On Monday 23 April 2012, Stephen Warren wrote: > I don't know whether it has to be early, but if this driver isn't going > to be instantiated from device tree, I'd prefer it to be initialized by > an explicit function call from the boards or common code rather than an > initcall. It seems a lot easier to keep track of all the initialization > by doing it explicitly. That said, it's quite possible this API couuld > be called from say init_machine instead of init_early though. Having it called from init_machine sounds ok, as long as it still scans the device tree for the addresses rather than hardcoding. Arnd