From mboxrd@z Thu Jan 1 00:00:00 1970 From: Arnd Bergmann Subject: Re: [PATCH/RFC] mmc: add DT bindings for more MMC capability flags Date: Wed, 6 Feb 2013 22:00:35 +0000 Message-ID: <201302062200.35499.arnd@arndb.de> References: <1923973.xHEn0Jh3IU@wuerfel> Mime-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Return-path: Received: from moutng.kundenserver.de ([212.227.126.186]:64061 "EHLO moutng.kundenserver.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758454Ab3BFWAt (ORCPT ); Wed, 6 Feb 2013 17:00:49 -0500 In-Reply-To: Sender: linux-mmc-owner@vger.kernel.org List-Id: linux-mmc@vger.kernel.org To: Guennadi Liakhovetski Cc: Mark Rutland , "linux-mmc@vger.kernel.org" , "devicetree-discuss@lists.ozlabs.org" , Chris Ball , Magnus Damm On Wednesday 06 February 2013, Guennadi Liakhovetski wrote: > > On Thu, 7 Feb 2013, Arnd Bergmann wrote: > > > On Wednesday 06 February 2013 17:25:42 Guennadi Liakhovetski wrote: > > > > > > Thank for pointing me out at that thread. However, I don't think > > > MMC_CAP_POWER_OFF_CARD has anything to do with compatibility or hardware > > > revisions. At least I haven't yet come across any sd/mmc hosts, that also > > > supply card power. You could "derive" this flag from the presence of a > > > regulator, capable of changing its status (switching on / off), but even > > > then you're not guaranteed, that you actually can (and want to) power the > > > card off at run-time - the regulator can be shared etc. So, an explicit > > > flag is needed. > > > > It sounds like something that should be handled in a controller specific > > way I think. E.g. on SDHCI, there seems to always be a method to power > > down the card using the SDHCI_POWER_CONTROL register, even without > > any external regulators. > > If I understand correctly, that register only controls card bus power. > Further sdhci.c uses regulators (host->vmmc) to power up and down the > card. Ok, that may be true. So a device that can only power down the bus but not the card itself should not set MMC_CAP_POWER_OFF_CARD then? I only saw that it is set unconditionally for the PCI case, which does not use regulators. Arnd