[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <948ba34a-1e05-45c4-8ba7-66c72bdb6fa5@amlogic.com>
Date: Mon, 17 Jun 2024 16:44:13 +0800
From: Junyi Zhao <junyi.zhao@...ogic.com>
To: Jerome Brunet <jbrunet@...libre.com>,
Uwe Kleine-König <u.kleine-koenig@...libre.com>
Cc: Kelvin Zhang via B4 Relay <devnull+kelvin.zhang.amlogic.com@...nel.org>,
Neil Armstrong <neil.armstrong@...aro.org>,
Kevin Hilman <khilman@...libre.com>,
Martin Blumenstingl <martin.blumenstingl@...glemail.com>,
Rob Herring <robh@...nel.org>, Krzysztof Kozlowski <krzk+dt@...nel.org>,
Conor Dooley <conor+dt@...nel.org>, kelvin.zhang@...ogic.com,
linux-pwm@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
linux-amlogic@...ts.infradead.org, linux-kernel@...r.kernel.org,
devicetree@...r.kernel.org
Subject: Re: [PATCH v8 1/2] pwm: meson: Add support for Amlogic S4 PWM
On 2024/6/14 15:24, Jerome Brunet wrote:
> [ EXTERNAL EMAIL ]
>
> On Thu 13 Jun 2024 at 22:46, Uwe Kleine-König <u.kleine-koenig@...libre.com> wrote:
>
>> Hello,
>>
>> On Thu, Jun 13, 2024 at 05:54:31PM +0200, Jerome Brunet wrote:
>>>> + for (i = 0; i < MESON_NUM_PWMS; i++) {
>>>> + meson->channels[i].clk = of_clk_get(np, i);
>>>> + if (IS_ERR(meson->channels[i].clk))
>>>> + return dev_err_probe(dev,
>>>> + PTR_ERR(meson->channels[i].clk),
>>>> + "Failed to get clk\n");
>>>
>>> here it is ok but ..
>>>
>>>> +
>>>> + ret = devm_add_action_or_reset(dev, meson_pwm_s4_put_clk,
>>>> + meson->channels[i].clk);
>>>> + if (ret)
>>>> + return dev_err_probe(dev, ret,
>>>> + "Failed to add clk_put action\n");
>>>
>>> ... here, devm_add_action_or_reset cannot defer, so dev_err_probe is not useful.
>>> dev_err() if you must print something. Just "return ret;" would be fine
>>> by me
>>
>> I don't agree. dev_err_probe() has several purposes. Only one of them is
>> handling -EPROBE_DEFER. See also commit
>> 532888a59505da2a3fbb4abac6adad381cedb374.
>
> I was stuck on the -EPROBE_DEFER usage. Thanks for the heads up
>
>>
>> So yes, please use dev_err_probe() also to handle
>> devm_add_action_or_reset().
>
> My point here is also that devm_add_action_or_reset() can only fail on
> memory allocation, like (devm_)kzalloc. Looking around the kernel, we
> tend to not add messages for that and just return the error code,
> presumably because those same 'out of memory' messages would proliferate
> everywhere.
>
>>
>> Best regards
>> Uwe
>
> --
> Jerome
Hi Uwe, I didnt get the clear point.
So, if we need "return ret" directly? or keep "dev_err_probe()" to print?
--
Best regards
Junyi
Powered by blists - more mailing lists