From mboxrd@z Thu Jan 1 00:00:00 1970 From: Arnd Bergmann Subject: Re: [PATCH v3 06/10] clk: exynos5420: register clocks using common clock framework Date: Tue, 18 Jun 2013 19:13:45 +0200 Message-ID: <201306181913.45484.arnd@arndb.de> References: <1371466836-4111-1-git-send-email-chander.kashyap@linaro.org> <201306181601.16795.arnd@arndb.de> <9285029.CIld1HyBWB@amdc1227> Mime-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Return-path: Received: from moutng.kundenserver.de ([212.227.17.10]:53012 "EHLO moutng.kundenserver.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754497Ab3FRROC (ORCPT ); Tue, 18 Jun 2013 13:14:02 -0400 In-Reply-To: <9285029.CIld1HyBWB@amdc1227> Sender: linux-samsung-soc-owner@vger.kernel.org List-Id: linux-samsung-soc@vger.kernel.org To: Tomasz Figa Cc: Chander Kashyap , "linux-arm-kernel@lists.infradead.org" , Mark Rutland , "kgene.kim@samsung.com" , "linux-serial@vger.kernel.org" , "linux-samsung-soc@vger.kernel.org" , Thomas Abraham , "thomas.ab@samsung.com" , "s.nawrocki@samsung.com" On Tuesday 18 June 2013, Tomasz Figa wrote: > This is a binding that has been defined for Samsung Common Clock Framework > drivers. Exynos4 and Exynos5250 use the same convention. The numbers are > defined in a way that should allow adding further clocks of particular types in > future as need for such shows up. I see, it's probably too late to change that then. Maybe if there is an Exynos6 this should be done differently though. > Physically there is one clock controller (CMU) which has a lot of dividers, > muxes and gates and so it is represented as a single device node. Some platforms just put device nodes for their clocks under the clock controller node, which would have helped here. Another option would be to use #clock-cells=<2> and make the first cell the type of the clock, and the second one the number. Also, if I understand things correctly, simple fixed rate clocks would normally not go into the clock driver at all when they are not controller by it. Arnd From mboxrd@z Thu Jan 1 00:00:00 1970 From: arnd@arndb.de (Arnd Bergmann) Date: Tue, 18 Jun 2013 19:13:45 +0200 Subject: [PATCH v3 06/10] clk: exynos5420: register clocks using common clock framework In-Reply-To: <9285029.CIld1HyBWB@amdc1227> References: <1371466836-4111-1-git-send-email-chander.kashyap@linaro.org> <201306181601.16795.arnd@arndb.de> <9285029.CIld1HyBWB@amdc1227> Message-ID: <201306181913.45484.arnd@arndb.de> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Tuesday 18 June 2013, Tomasz Figa wrote: > This is a binding that has been defined for Samsung Common Clock Framework > drivers. Exynos4 and Exynos5250 use the same convention. The numbers are > defined in a way that should allow adding further clocks of particular types in > future as need for such shows up. I see, it's probably too late to change that then. Maybe if there is an Exynos6 this should be done differently though. > Physically there is one clock controller (CMU) which has a lot of dividers, > muxes and gates and so it is represented as a single device node. Some platforms just put device nodes for their clocks under the clock controller node, which would have helped here. Another option would be to use #clock-cells=<2> and make the first cell the type of the clock, and the second one the number. Also, if I understand things correctly, simple fixed rate clocks would normally not go into the clock driver at all when they are not controller by it. Arnd