From mboxrd@z Thu Jan 1 00:00:00 1970 From: Simon Horman Date: Wed, 17 Jul 2013 23:01:13 +0000 Subject: Re: [PATCH v4 1/3] ARM: shmobile: r8a7790: add Ether support Message-Id: <20130717230113.GA8154@verge.net.au> List-Id: References: <1372663234-13486-1-git-send-email-horms+renesas@verge.net.au> <1372663234-13486-2-git-send-email-horms+renesas@verge.net.au> <20130717091103.GD15981@e106331-lin.cambridge.arm.com> In-Reply-To: <20130717091103.GD15981@e106331-lin.cambridge.arm.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: linux-arm-kernel@lists.infradead.org On Wed, Jul 17, 2013 at 10:11:03AM +0100, Mark Rutland wrote: [snip] > Why are we hardcoding the address and interrupt of a device rather than > describing it in the devicetree? This is the most trivial case to > describe. The approach that shmobile is taking is to either initialise a device entirely using C (non-DT) or entirely using DT. And moreover to either initialise a board using C or DT (though in the latter case some devices still end up being initialised in C to provide a minimally working system). At this time for more recent boards we typically provide both a C and DT versions to initialise a board. The C version typically is able to initialise a fuller set of hardware while the DT version catches up. In this case the device is being initialised in C. And this code is intended ti be used when a board and thus its SoC is initialised using C. Once the SH ethernet driver and all its pre-requisite drivers and infrastructure can be initialised using DT then that will be used when initialising the board using DT.