From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752469Ab2KSMT2 (ORCPT ); Mon, 19 Nov 2012 07:19:28 -0500 Received: from arroyo.ext.ti.com ([192.94.94.40]:42443 "EHLO arroyo.ext.ti.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752421Ab2KSMT0 (ORCPT ); Mon, 19 Nov 2012 07:19:26 -0500 Message-ID: <50AA23C4.1000706@ti.com> Date: Mon, 19 Nov 2012 13:19:16 +0100 From: Peter Ujfalusi User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:16.0) Gecko/20121105 Thunderbird/16.0.1 MIME-Version: 1.0 To: Linus Walleij CC: Sascha Hauer , Grant Likely , , Subject: Re: [PATCH] gpio: twl4030: Use only TWL4030_MODULE_LED for LED configuration References: <1352799313-21019-1-git-send-email-peter.ujfalusi@ti.com> <50A9F335.80107@ti.com> In-Reply-To: Content-Type: text/plain; charset="ISO-8859-1" Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi Linus, On 11/19/2012 11:40 AM, Linus Walleij wrote: > On Mon, Nov 19, 2012 at 9:52 AM, Peter Ujfalusi wrote: > >> I was actually tempted to remove the whole LED (PWM) thing from the >> gpio-twl4030 driver. It was a big surprise to me to see something like this in >> there. > > It looks wrong. In theory I think this should be ripped out and moved to > drivers/leds/leds-twl4030.c. But that would only be in case it was *always* > exclusively used for a LED and as you say: > >> It turns out that on BeagleBoard the USB host enable signal is connected to >> LEDA (PWMA) of twl4030... It is an enable signal. Seriously. So what we do >> here is either configure the PWMs as full on, or turn it off. > > Sounds like some hardware engineer has been having fun or was > just out of GPIOs to use... So this LED PWM is also used as a > GPIO. My guess it was out of fun. On BealgeBoard GPIO13 from twl4030 is free (LEDSYNC/GPIO.13) so it could have been used for this, or GPIO16/17 as well. > I think this part of the driver should be moved to > drivers/pwm/pwm-twl4030.c and modeled as a PWM. If some > platform need to use the PWM as if it was a GPIO then that's just a > special usecase. (Setting duty cycles to something like > INT_MAX and just switching polarity to toggle it...) > > If it needs to be used by a LED there needs to be some generic > LED type just using a standard pwm_request() to get some > specific PWM, such as leds/leds-pwm.c or so. This way the > PWM can be used for LEDs if need be or other things... I have already sent a series to support the PWMs in twl4030/6030: https://lkml.org/lkml/2012/11/8/219 It adds support for both type of PWMs (the one named as PWM and the ones named as LED since they are just PWMs). We already have leds/leds-pwm driver in upstream so we are going to use that if there is a led connected to one of them: https://lkml.org/lkml/2012/11/12/222 We can use the backlight-pwm if the LCD backlight has been hooked to one of the PWM (as it is done on SDP4430). This is why I was thinking of adding gpio/gpio-pwm driver. With that we can use the PWM as GPO line. > Sounds correct, Sascha? > >> Either way this is wrong IMHO to handle the LEDA/B via the gpio-twl4030 driver. > > It's confusing indeed. > > So what we're dealing with is: a LED-specific PWM, being used > as a PWM with eternal dutycycle and then being used as GPIO. > > Well, we get to deal with it ... :-/ gpio/gpio-pwm driver? -- Péter