From mboxrd@z Thu Jan 1 00:00:00 1970 From: daniel.thompson@linaro.org (Daniel Thompson) Date: Fri, 22 May 2015 15:14:53 +0100 Subject: [PATCH v8 14/16] ARM: dts: Introduce STM32F429 MCU In-Reply-To: References: <1431158038-3813-1-git-send-email-mcoquelin.stm32@gmail.com> <2282066.NWoIT9ZyLc@wuerfel> <13641152.Yt4ZI3oT6L@wuerfel> <1432285588.3929.28.camel@pengutronix.de> <20150522091822.GF8557@lukather> <1432289231.3929.60.camel@pengutronix.de> <555F2A8A.5020205@suse.de> Message-ID: <555F39DD.9060209@linaro.org> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On 22/05/15 14:57, Maxime Coquelin wrote: > 2015-05-22 15:09 GMT+02:00 Andreas F?rber : >> As you should know, I did have an RCC clk driver, and there is no such >> issue. The two drivers use different mechanisms for initialization. And >> I'm pretty sure that I've already remarked that on the list, too. > > Yes, you use of_iomap in your clock driver [0]. > Daniel, would you accept to do the same? > That would remove one difference between stm32/sunxi/socfpga reset drivers. In fact, that is exactly what I am currently doing, though I was planning for it to be temporary. It seems a bit weird to me that one driver (which requests too much register space) only works because another driver chooses not to request any. BTW in drivers/clk there are ~110 of_iomaps and only 10 of_io_request_and_maps... so I don't think anyone will yell at me for using of_iomap(). Daniel.