From mboxrd@z Thu Jan 1 00:00:00 1970 From: Laurent Pinchart Subject: Re: [PATCH 13/26] ARM: omap3.dtsi: add omapdss information Date: Fri, 13 Dec 2013 04:27:12 +0100 Message-ID: <1933495.zHe4JMYZyB@avalon> References: <1386160133-24026-1-git-send-email-tomi.valkeinen@ti.com> <52A9760A.3000801@ti.com> <20131212215913.GB29072@muru.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7Bit Return-path: In-Reply-To: <20131212215913.GB29072@muru.com> Sender: linux-omap-owner@vger.kernel.org To: Tony Lindgren Cc: Tomi Valkeinen , linux-omap@vger.kernel.org, linux-fbdev@vger.kernel.org, devicetree@vger.kernel.org, Archit Taneja , Darren Etheridge List-Id: devicetree@vger.kernel.org Hi Tony, On Thursday 12 December 2013 21:59:13 Tony Lindgren wrote: > On Thu, Dec 12, 2013 at 10:38:34AM +0200, Tomi Valkeinen wrote: > > On 2013-12-12 01:44, Laurent Pinchart wrote: > > > > So, are they independent? I don't know =). I think they lean on the > > independent side. dss_core is always needed for the submodules to work, > > but for example DSI could be used without DISPC, using system DMA to > > transfer data from memory to DSI. Not a very useful thing to do, but > > still, there are dedicated DMA channels for that. > > If they have separate hwmod entries, they should be considered separate > independent devices for sure. > > To summarize, here are few reasons why they need to be treated as > separate devices: Are you talking generally here, or about the DSS modules in particular ? > 1. The modules maybe clocked/powered/idled separately and can have their > own idle configuration so they can do the hardware based idling > separately. I don't think this applies to the DSS modules. > 2. Doing a readback after a write to one module will not flush the write > to the other modules on the (bus depending on the SoC version AFAIK). > That can lead to nasty bugs caused by the ordering. How does separate devices solve this ? > 3. If the devices are described in a different way in the .dts files > from the hwmod data, we will not have 1-to-1 mapping and will never > be able to replace ti,hwmods with just the compatible string. -- Regards, Laurent Pinchart