From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mark Brown Subject: Re: [PATCH] mmc: move regulator handling to core Date: Thu, 3 Dec 2009 13:40:32 +0000 Message-ID: <20091203134031.GA1311@rakim.wolfsonmicro.main> References: <1259844390-10541-1-git-send-email-daniel@caiaq.de> <20091203130627.GA31254@rakim.wolfsonmicro.main> <20091203131423.GV14091@buzzloop.caiaq.de> <20091203132241.GB31533@rakim.wolfsonmicro.main> <20091203133200.GX14091@buzzloop.caiaq.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from opensource.wolfsonmicro.com ([80.75.67.52]:43347 "EHLO opensource2.wolfsonmicro.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1755239AbZLCNk1 (ORCPT ); Thu, 3 Dec 2009 08:40:27 -0500 Content-Disposition: inline In-Reply-To: <20091203133200.GX14091@buzzloop.caiaq.de> Sender: linux-mmc-owner@vger.kernel.org List-Id: linux-mmc@vger.kernel.org To: Daniel Mack Cc: linux-kernel@vger.kernel.org, Liam Girdwood , Pierre Ossman , Andrew Morton , Matt Fleming , Adrian Hunter , David Brownell , Russell King , Linus Walleij , Eric Miao , Robert Jarzmik , Cliff Brake , Jarkko Lavinen , linux-mmc@vger.kernel.org, linux-arm-kernel@lists.infradead.org On Thu, Dec 03, 2009 at 02:32:00PM +0100, Daniel Mack wrote: > On Thu, Dec 03, 2009 at 01:22:41PM +0000, Mark Brown wrote: > > On Thu, Dec 03, 2009 at 02:14:23PM +0100, Daniel Mack wrote: > > > I would expect the power to be killed when the last user stops using it. > > > Which should result in the same effect if you only have one host, one > > > regulator, and one user. > > Yes, it's always fine in that case (modulo always_on and/or regulators > > without power control). > Well, it didn't for me and always_on, though, due to the return values I > described. I mean your new code is fine. > > This goes back to the thing about using > > regulator_get_exclusive(), the message given was that the MMC drivers > > really needed to be able to guarantee that the power would be removed > > when that was requested. > > Like I say, if there isn't a *strict* requirement but it's only > > desirable (possibly strongly desirable) then your approach is obviously > > preferable. > The mmci people would need to answer that. To me, the code just looked > like a power saving feature. > If this driver needs it, the only tweak to my patch to let that > particular call site use regulator_get_exclusive, and the core will > still do the right thing. For this case, the behaviour should be exactly > the same than it currently is, correct? No, you'll also need to update the way the driver bootstraps the reference count since with regulator_get_exclusive() the reference count is initialised to 1 if the regulator is enabled when it is claimed.