From mboxrd@z Thu Jan 1 00:00:00 1970 From: arnd@arndb.de (Arnd Bergmann) Date: Fri, 26 Sep 2014 11:14:30 +0200 Subject: [PATCH v5] mfd: syscon: Decouple syscon interface from platform devices In-Reply-To: References: <1411360807-7750-1-git-send-email-pankaj.dubey@samsung.com> <2553182.6YgvmDLuEA@wuerfel> Message-ID: <3291391.H71MjevFsA@wuerfel> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Friday 26 September 2014 09:48:24 Joachim Eastwood wrote: > On 26 September 2014 09:16, Arnd Bergmann wrote: > > On Friday 26 September 2014 07:34:12 Joachim Eastwood wrote: > >> I am working on Cortex-M4 no-MMU platform that isn't upstream yet, btw. > >> > > > > Sorry for drifting off-topic, but this is very interesting to me. Can you > > say which one you are working on and what your timeline is for submitting > > it upstream? > > It's NXP LPC18xx/43xx which is Cortex-M3/M4. > > 3.19 or 3.20 might be target. Ah, very nice! > Right now everything is in a github repository here: > https://github.com/manabian/linux-lpc > > Most stuff are working now and I am in the process of clean it up and > adding documentation. Ok, looks like you are making good progress. I noticed three high-level issues that you may want to address: - the watchdog driver should use the generic watchdog framework rather than registering a misc device. - the ehci glue can probably go away if you make very small changes to the generic ehci platform driver - I don't like the way that the stmmac glue drivers are added, I thought we had fixed this before but I think I need to dig up old emails. The driver should really be a loadable module that hooks calls into the common code rather than being linked into one module. I also have a plan for doing multiplatform builds of nommu kernels, for build testing mostly, I wouldn't expect you to run that configuration. No need for you to address that yourself though, we'll get there. Arnd