From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stephen Boyd Subject: Re: warnings with clk_disable_unprepare() Date: Thu, 8 Dec 2016 13:18:31 -0800 Message-ID: <20161208211831.GL5423@codeaurora.org> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: Sender: linux-clk-owner@vger.kernel.org To: Pierre-Louis Bossart Cc: "alsa-devel@alsa-project.org" , linux-clk@vger.kernel.org List-Id: alsa-devel@alsa-project.org On 12/05, Pierre-Louis Bossart wrote: > Hi, > In the modified cht-bsw_rt5645 machine driver for Asus X205TA and > T100HA, I disable the audio mclk with a call to > clk_disable_unprepare() when I see a SND_SOC_DAPM_EVENT_OFF event in > the platform_clock_control hook. > > For some reason I seem to get multiple SND_SOC_DAPM_EVENT_OFF events > and the clock framework complains with dmesg warnings that the > refcount is already zero (which has no functional side effect). > > WARNING: CPU: 0 PID: 2002 at drivers/clk/clk.c:476 clk_unprepare+0x1a/0x21 > > WARNING: CPU: 2 PID: 2095 at drivers/clk/clk.c:594 > clk_core_disable_lock+0x12/0x1b > > I could fix this in three different ways: > - remove multiple SND_SOC_DAPM_EVENT_OFF events (not sure what the > root cause is, maybe issues with jack detection?) It would be good to know what the root cause is here before thinking about doing anything else. > - call a clk_is_enabled() or clk_refcount() function before calling > clk_disable_unprepare() (but they don't seem to exist in > include/linux/clk.h) > - count the references myself (which does exactly the same as what > the clk framework already does so it's not terribly elegant) > Both of these sound like hacks around the problem of unbalanced off event calls. -- Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project