From mboxrd@z Thu Jan 1 00:00:00 1970 From: Javier Martinez Canillas Subject: Re: [PATCH v2 06/12] mmc: pwrseq: add support for power-on sequencing through DT Date: Mon, 11 Jan 2016 13:02:46 -0300 Message-ID: <5693D226.2000307@osg.samsung.com> References: <1452155155-16232-1-git-send-email-wxt@rock-chips.com> <1452155155-16232-7-git-send-email-wxt@rock-chips.com> <568FAA07.1010509@osg.samsung.com> <6960259.dpUH8KWsNS@phil> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Return-path: Received: from lists.s-osg.org ([54.187.51.154]:55208 "EHLO lists.s-osg.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757640AbcAKQCy (ORCPT ); Mon, 11 Jan 2016 11:02:54 -0500 In-Reply-To: <6960259.dpUH8KWsNS@phil> Sender: linux-mmc-owner@vger.kernel.org List-Id: linux-mmc@vger.kernel.org To: Heiko Stuebner Cc: Caesar Wang , Ulf Hansson , linux-rockchip@lists.infradead.org, leecam@google.com, leozwang@google.com, keescook@google.com, Srinivas Kandagatla , linux-mmc@vger.kernel.org, linux-kernel@vger.kernel.org, Alexandre Courbot , NeilBrown Hello Heiko, On 01/08/2016 11:42 PM, Heiko Stuebner wrote: > Am Freitag, 8. Januar 2016, 09:22:31 schrieb Javier Martinez Canillas: >>> For example: >>> we need enable wifi module power to via the WL_REG_ON >>> pin, we need enable it as the regulator if this pin is connected to >>> the gpio of cpu. >> >> This part confuses me, so does your chip have an actual regulator that >> needs to be enabled or is just a fake regulator whose gpio property is >> used not to enable the regulator but to toggle the WL_REG_ON pin of >> the WiFi chip? > > another option would be to use the reset-gpio-handles. rk3288-veyron and I > think some Exynos as well use it that way. > Yes I know, my point was that the reset-gpios property should be used instead of a fake regulator if what's needed is to toggle a chip pin. > >>> Maybe, someone will say that can pull up/down from dts. >>> Unfortunately some SoCs can't support pinctrl pull up/down in >>> internal. >> >> Can you please elaborate on this? AFAIU this limitation is the reason >> why you went with the regulator approach so I think it deserve a more >> deep explanation. > > On the rk3036 each pin has an individual unchangable pull direction. So it's > either no bias or pulling in the predefined direction (the pin_default bias > option). > I think each change has to be justified on its own so I would say that having a regulator enabled as a part of a SDIO chip's power sequencing is something needed for many platforms, and that this provider should be extended to support that (something like commit msg in patch 05/12). And then in the kylin DTS change (patch 08/12), I would explain why a chained regulators approach is used/needed instead of the reset-gpios due any platform limitations. > > Heiko > Best regards, -- Javier Martinez Canillas Open Source Group Samsung Research America