From mboxrd@z Thu Jan 1 00:00:00 1970 From: kgene.kim@samsung.com (Kukjin Kim) Date: Thu, 09 Feb 2012 20:44:22 +0900 Subject: [PATCH 1/2] ARM: EXYNOS: add support GPIO for EXYNOS5250 In-Reply-To: <20120202182726.GO15343@ponder.secretlab.ca> References: <1328025040-4751-1-git-send-email-kgene.kim@samsung.com> <1328025040-4751-2-git-send-email-kgene.kim@samsung.com> <20120131224043.GD8338@n2100.arm.linux.org.uk> <20120202182726.GO15343@ponder.secretlab.ca> Message-ID: <06ad01cce720$2acadf50$80609df0$%kim@samsung.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org Grant Likely wrote: > > On Tue, Jan 31, 2012 at 10:40:43PM +0000, Russell King - ARM Linux wrote: > > On Wed, Feb 01, 2012 at 12:50:39AM +0900, Kukjin Kim wrote: > > > From: Sangsu Park > > > > > > This patch adds follwing. > > > - IO-map for EXYNOS5250 GPIO support > > > - EXYNOS5250 GPIO bank size/number definitions > > > - memory map definition for S5P GPIO4 > > > > > > Signed-off-by: Sangsu Park > > > Signed-off-by: Kukjin Kim > > > > Do you actually need these static mapping definitions? The samsung > > gpiolib initialization is called from a core_initcall(), and at this > > time ioremap() is fully capable of working. > > Indeed. Please get rid of the S5P_VA_GPIO* macros and use ioremap to > set up the registers mappings. I understand that the existing code > already does that, but I don't want to see any more added. Also, it > is easy to fix by adding the ioremap calls to samsung_gpio_init() > Yes, I agree. Will be updated with ioremap(). Thanks. Best regards, Kgene. -- Kukjin Kim , Senior Engineer, SW Solution Development Team, Samsung Electronics Co., Ltd.