Re: [PATCH 1/2] drm/meson: dw-hdmi: Disable clocks on driver teardown

From: Marc Zyngier <maz@kernel.org>
To: Jerome Brunet <jbrunet@baylibre.com>
Cc: Neil Armstrong <narmstrong@baylibre.com>,
	Martin Blumenstingl <martin.blumenstingl@googlemail.com>,
	Kevin Hilman <khilman@baylibre.com>,
	linux-kernel@vger.kernel.org, dri-devel@lists.freedesktop.org,
	Guillaume Tucker <guillaume.tucker@collabora.com>,
	linux-amlogic@lists.infradead.org, kernel-team@android.com,
	linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH 1/2] drm/meson: dw-hdmi: Disable clocks on driver teardown
Date: Mon, 23 Nov 2020 14:15:45 +0000	[thread overview]
Message-ID: <763f505eececa3b02bd794aec5e36d22@kernel.org> (raw)
In-Reply-To: <1jwnyc41c6.fsf@starbuckisacylon.baylibre.com>

On 2020-11-23 14:03, Jerome Brunet wrote:
> On Fri 20 Nov 2020 at 10:42, Marc Zyngier <maz@kernel.org> wrote:
> 
>> The HDMI driver request clocks early, but never disable them, leaving
>> the clocks on even when the driver is removed.
>> 
>> Fix it by slightly refactoring the clock code, and register a devm
>> action that will eventually disable/unprepare the enabled clocks.
>> 
>> Signed-off-by: Marc Zyngier <maz@kernel.org>
>> ---
>>  drivers/gpu/drm/meson/meson_dw_hdmi.c | 43 
>> ++++++++++++++++++---------
>>  1 file changed, 29 insertions(+), 14 deletions(-)
>> 
>> diff --git a/drivers/gpu/drm/meson/meson_dw_hdmi.c 
>> b/drivers/gpu/drm/meson/meson_dw_hdmi.c
>> index 7f8eea494147..29623b309cb1 100644
>> --- a/drivers/gpu/drm/meson/meson_dw_hdmi.c
>> +++ b/drivers/gpu/drm/meson/meson_dw_hdmi.c
>> @@ -145,8 +145,6 @@ struct meson_dw_hdmi {
>>  	struct reset_control *hdmitx_apb;
>>  	struct reset_control *hdmitx_ctrl;
>>  	struct reset_control *hdmitx_phy;
>> -	struct clk *hdmi_pclk;
>> -	struct clk *venci_clk;
>>  	struct regulator *hdmi_supply;
>>  	u32 irq_stat;
>>  	struct dw_hdmi *hdmi;
>> @@ -946,6 +944,29 @@ static void meson_disable_regulator(void *data)
>>  	regulator_disable(data);
>>  }
>> 
>> +static void meson_disable_clk(void *data)
>> +{
>> +	clk_disable_unprepare(data);
>> +}
>> +
>> +static int meson_enable_clk(struct device *dev, char *name)
>> +{
>> +	struct clk *clk;
>> +	int ret;
>> +
>> +	clk = devm_clk_get(dev, name);
>> +	if (IS_ERR(clk)) {
>> +		dev_err(dev, "Unable to get %s pclk\n", name);
>> +		return PTR_ERR(clk);
>> +	}
>> +
>> +	ret = clk_prepare_enable(clk);
>> +	if (!ret)
>> +		ret = devm_add_action_or_reset(dev, meson_disable_clk,
>> clk);
> 
> Thanks for fixing this Marc.
> 
> FYI, while it is fine to declare a function to disable the clocks, a 
> quick
> cast may avoid it
> 
> devm_add_action_or_reset(dev, (void(*)(void *))clk_disable_unprepare, 
> clk);

While this works for now, a change to the clk_disable_unprepare()
prototype (such as adding a second argument) would now go completely
unnoticed (after all, you've cast the function, it *must* be correct,
right?), and someone would spend a few hours trying to track down memory
corruption or some other interesting results.

Yes, casting C functions can be hilarious.

I can see a few uses of this hack in the tree, and I have my pop-corn
ready.

Thanks,

         M.
-- 
Jazz is not dead. It just smells funny...

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel