From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Subject: Re: [PATCH] clk: x86: Do not gate clocks enabled by the firmware To: Enric Balletbo Serra References: <20170710203801.17496-1-carlo@caione.org> <20170710235403.GB25390@fury> <1ac00dd5-038c-7ef7-44e7-4c95fea8042d@linux.intel.com> <1500904960.29303.204.camel@linux.intel.com> <20eea5c5-9f95-3273-97f6-193c56fe59fb@linux.intel.com> Cc: Andy Shevchenko , Darren Hart , Carlo Caione , Michael Turquette , linux-clk@vger.kernel.org, Stephen Boyd , linux@endlessm.com, Carlo Caione From: Pierre-Louis Bossart Message-ID: Date: Mon, 24 Jul 2017 18:44:20 +0200 MIME-Version: 1.0 In-Reply-To: Content-Type: text/plain; charset=utf-8; format=flowed List-ID: On 7/24/17 6:33 PM, Enric Balletbo Serra wrote: > Hi, > > 2017-07-24 18:26 GMT+02:00 Pierre-Louis Bossart > : >> On 7/24/17 5:02 PM, Enric Balletbo Serra wrote: >>> >>> 2017-07-24 16:07 GMT+02:00 Pierre-Louis Bossart >>> : >>>> >>>> On 7/24/17 4:02 PM, Andy Shevchenko wrote: >>>>> >>>>> >>>>> On Mon, 2017-07-24 at 15:49 +0200, Pierre-Louis Bossart wrote: >>>>>> >>>>>> >>>>>> On 7/11/17 1:54 AM, Darren Hart wrote: >>>>>>> >>>>>>> >>>>>>> On Tue, Jul 11, 2017 at 12:31:28AM +0200, Enric Balletbo Serra >>>>>>> wrote: >>>>> >>>>> >>>>> >>>>>>>> Tested-by: Enric Balletbo i Serra >>>>>>> >>>>>>> >>>>>>> >>>>>>> Excellent, thank you Enric. Pierre, any objections to this going in >>>>>>> to >>>>>>> 4.13 now and stable back to 4.12? >>>>>> >>>>>> >>>>>> >>>>>> I saw an update of this patch being merged while I was away on an >>>>>> extended summer break, but out of curiosity what was the problem with >>>>>> no >>>>>> audio on Baytrail? the code was added precisely to add support for >>>>>> 19.2 >>>>>> MHz on Baytrail and finally solve audio issues. I have no reports of >>>>>> broken functionality with this 282a4e4 commit... Enric, do you have a >>>>>> bugzilla or pointers? >>>>> >>>>> >>>>> >>>>> TL;DR: We should respect firmware settings before blindly shut down >>>>> "unused" resources. (We used to have similar stuff in GPIO/pinctrl >>>>> driver for direct IRQ pins, where we shut up the pins which are used >>>>> directly by IOAPIC) >>>> >>>> >>>> >>>> I wasn't arguing against those patches, just curious as to which >>>> platforms >>>> were broken by setting pmc_plat_3 blindly in the driver. >>>> >>> >>> After your patch the clock framework disables the required clocks >>> because the driver does not request then and audio stopped to >>> work.This is not the solution as is the codec who *must* request these >>> clocks. We should implement this in the codec driver.. But now, at >>> least, if firmware enables the clocks audio continues working on >>> Baytrail. >> >> >> Clocks are not enabled/managed in any of the BIOSes I've seen for Baytrail, >> are you talking about a Chromebook by any chance with the legacy non-DPCM >> byt-max98080 driver? >> > > Yes the issue is with a Chromebook with the legacy byt-max98080 driver. ah, yes I never tested on those (couldn't make the device boot). We may be able to do more simplifications, in machine drivers I test for Baytrail vs. Cherrytrail to manage the clocks at the kernel level, but maybe I should just look at what the firmware does and use the kernel clock management as a fallback.