From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ryan Mallon Subject: Re: [PATCH 3/3 v2] at91/atmel-mci: inclusion of sd/mmc driver in at91sam9g45 chip and board Date: Tue, 03 Nov 2009 15:30:56 +1300 Message-ID: <4AEF95E0.7040602@bluewatersys.com> References: <20090930155557.7dae503b@hskinnemoen-d830> <20091028083548.38f1d80c@hskinnemoen-d830> <1256763057.4362.36.camel@ben-desktop> <4AEF12B7.3040704@atmel.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <4AEF12B7.3040704@atmel.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: linux-arm-kernel-bounces@lists.infradead.org Errors-To: linux-arm-kernel-bounces+linux-arm-kernel=m.gmane.org@lists.infradead.org To: Nicolas Ferre Cc: =?ISO-8859-1?Q?=22=5C=22Thiago_A=2E=5C=22_Corr=EAa=22?= , David Brownell , linux-mmc@vger.kernel.org, kernel@avr32linux.org, linux-kernel@vger.kernel.org, David Brownell , Andrew Victor , Ben Nizette , akpm@linux-foundation.org, linux-arm-kernel@lists.infradead.org List-Id: linux-mmc@vger.kernel.org Nicolas Ferre wrote: > Ben Nizette : > >> On Wed, 2009-10-28 at 21:53 +0200, Andrew Victor wrote: >> >>> hi, >>> >>> >>>> Then I think it would be best to use GPIO_PIN_NONE. Makes it clear >>>> what is expected and avoids confusion on what should be the proper >>>> value. >>>> I hope I'm not saying non-sense, but even if I am, I guess you can see >>>> that I'm advocating against the magic numbers :) >>>> >>> What magic numbers ? >>> >> I think Thiago was referring to the "-1" in the original patch as the >> magic number. >> >> Leaving the field blank to be initialised to 0 is certainly the >> cleanest, I agree, but it doesn't actually /work/. On many archs 0 is a >> valid gpio number; the gpio_is_valid check used throughout the kernel >> (including atmel-mci.c) looks like >> >> static inline int gpio_is_valid(int number) >> { >> /* only some non-negative numbers are valid */ >> return ((unsigned)number) < ARCH_NR_GPIOS; >> } >> > > I understand that the better way to solve this issue is to: > - keep the AT91 way of specifying not connected pins (= 0) > - code the gpio_is_valid() function for at91 that tests this way of > handling not connected gpio > It doesn't appear that the gpio_is_valid function can be overridden by a platform specific version. However, as you point out, on AT91 it appears broken since anything less than AT91_PIN_PA0 (32) is not a valid gpio. IIRC, we can't mark static inline functions as weak, and we don't want to turn gpio_is_valid into an actual function call. We could do some preprocessor magic, but that gets a bit messy. CC'ed David Brownell, who does most of the gpiolib stuff. Any ideas? ~Ryan -- Bluewater Systems Ltd - ARM Technology Solution Centre Ryan Mallon 5 Amuri Park, 404 Barbadoes St ryan@bluewatersys.com PO Box 13 889, Christchurch 8013 http://www.bluewatersys.com New Zealand Phone: +64 3 3779127 Freecall: Australia 1800 148 751 Fax: +64 3 3779135 USA 1800 261 2934