MIME-Version: 1.0
In-Reply-To: <f1038a3e-57fb-ec01-26a0-452a11dfcf3a@linaro.org>
References: <20211117020346.4088302-1-swboyd@chromium.org> <76b103ec-7034-e6c1-1ab4-174cf16f9fc8@linaro.org>
 <CAE-0n53HNSRTdADO1dbQTyLafyajUTatMq5tsLeNDLQ4g95YpA@mail.gmail.com> <f1038a3e-57fb-ec01-26a0-452a11dfcf3a@linaro.org>
From:   Stephen Boyd <swboyd@chromium.org>
User-Agent: alot/0.9.1
Date:   Fri, 19 Nov 2021 01:57:03 +0100
Message-ID: <CAE-0n51TYW26W8+_A1Ar7eUGLeUoV2eE=8YAkD4N7d5EqoNNRg@mail.gmail.com>
Subject: Re: [PATCH] cpufreq: qcom-hw: Use optional irq API
To:     "Rafael J . Wysocki" <rafael@kernel.org>,
        Thara Gopinath <thara.gopinath@linaro.org>,
        Viresh Kumar <viresh.kumar@linaro.org>
Cc:     linux-kernel@vger.kernel.org, linux-arm-msm@vger.kernel.org,
        linux-pm@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Quoting Thara Gopinath (2021-11-18 06:08:04)
>
>
> On 11/17/21 11:32 PM, Stephen Boyd wrote:
> > Quoting Thara Gopinath (2021-11-17 18:55:17)
> >> Hello Stephen,
> >>
> >> Thanks for the patch
> >>
> >> On 11/16/21 9:03 PM, Stephen Boyd wrote:
> >>> Use platform_get_irq_optional() to avoid a noisy error message when the
> >>> irq isn't specified. The irq is definitely optional given that we only
> >>> care about errors that are -EPROBE_DEFER here.
> >>>
> >>> Cc: Thara Gopinath <thara.gopinath@linaro.org>
> >>> Signed-off-by: Stephen Boyd <swboyd@chromium.org>
> >>> ---
> >>>    drivers/cpufreq/qcom-cpufreq-hw.c | 8 +++++---
> >>>    1 file changed, 5 insertions(+), 3 deletions(-)
> >>>
> >>> diff --git a/drivers/cpufreq/qcom-cpufreq-hw.c b/drivers/cpufreq/qcom-cpufreq-hw.c
> >>> index a2be0df7e174..b442d4983a22 100644
> >>> --- a/drivers/cpufreq/qcom-cpufreq-hw.c
> >>> +++ b/drivers/cpufreq/qcom-cpufreq-hw.c
> >>> @@ -382,9 +382,11 @@ static int qcom_cpufreq_hw_lmh_init(struct cpufreq_policy *policy, int index)
> >>>         * Look for LMh interrupt. If no interrupt line is specified /
> >>>         * if there is an error, allow cpufreq to be enabled as usual.
> >>>         */
> >>> -     data->throttle_irq = platform_get_irq(pdev, index);
> >>> -     if (data->throttle_irq <= 0)
> >>> -             return data->throttle_irq == -EPROBE_DEFER ? -EPROBE_DEFER : 0;
> >>> +     data->throttle_irq = platform_get_irq_optional(pdev, index);
> >>> +     if (data->throttle_irq == -ENXIO)
> >>> +             return 0;
> >>> +     if (data->throttle_irq < 0)
> >>> +             return data->throttle_irq;
> >>
> >> Here the idea is to return only -EPROBE_DEFER error. Else return a 0 ,
> >> so that cpufreq is enabled even if lmh interrupt is inaccessible. The
> >> above check returns errors other than -EPROBE_DEFER as well. So I would
> >> say make irq optional and keep the below check
> >>
> >> if (data->throttle_irq <= 0)
> >>          return data->throttle_irq == -EPROBE_DEFER ? -EPROBE_DEFER : 0;
> >
> > I'd like to catch other errors, for example, DT has an irq specified
> > that is outside the range of irqs available. If the DT is correct, then
> > it will either have a valid irq and this will return a >= 0 value or
> > nothing will be specified and we'll get back -ENXIO now. Do you have
> > some scenario where my patch fails to work?
>
> Exactly. Like in the scenario you mentioned above, I do not want cpufreq
> to be disabled. This interrupt is a throttle notification interrupt. The
> action taken on basis of this is to send thermal pressure signal to
> scheduler so that scheduler places tasks better. Even if the dt has
> messed up this interrupt, I think cpufreq should still be enabled. May
> be we can print a warn and still return 0 to enable cpufreq.
>

If the DT is messed up then we have problems that we should fix instead
of silently ignore in this driver. Is there some bad DT out there that
you're worried about? I don't believe this patch breaks anything but if
you can point to something that it breaks I'd be glad to investigate.