Re: [PATCH v5 03/30] thermal/core: Add a generic thermal_zone_set_trip() function

From: "Rafael J. Wysocki" <rafael@kernel.org>
To: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	Linux PM <linux-pm@vger.kernel.org>,
	"Zhang, Rui" <rui.zhang@intel.com>,
	Amit Kucheria <amitk@kernel.org>
Subject: Re: [PATCH v5 03/30] thermal/core: Add a generic thermal_zone_set_trip() function
Date: Tue, 27 Sep 2022 12:38:42 +0200	[thread overview]
Message-ID: <CAJZ5v0gATxtX5RW0oHbhT_hjUoEC3V39tQpJi74eg8iXhrwZKg@mail.gmail.com> (raw)
In-Reply-To: <ee9ea160-ae77-112b-5302-74179e372387@linaro.org>

On Tue, Sep 27, 2022 at 12:11 AM Daniel Lezcano
<daniel.lezcano@linaro.org> wrote:
>
> On 26/09/2022 21:25, Rafael J. Wysocki wrote:
>
> [ ... ]
>
> >> +       if ((t.temperature != trip->temperature) && tz->ops->set_trip_temp) {
> >
> > The inner parens are not needed here and below.
> >
> >> +
> >
> > And the extra empty line is not needed here (and below) too IMO.
> >
> >> +               ret = tz->ops->set_trip_temp(tz, trip_id, trip->temperature);
> >> +               if (ret)
> >> +                       goto out;
> >> +       }
> >
>
> Without the parens, the following happens:
>
>
> warning: this ‘if’ clause does not guard... [-Wmisleading-indentation]
>   1229 |         if ((t.temperature != trip->temperature) &&
> tz->ops->set_trip_temp)
>        |         ^~
> note: ...this statement, but the latter is misleadingly indented as if
> it were guarded by the ‘if’
>   1231 |                 if (ret)
>        |                 ^~

This is about indentation, though, so it looks like white space is
mangled somehow.

As a matter of correctness, the inner parens are not needed.

> >> +       if ((t.hysteresis != trip->hysteresis) && tz->ops->set_trip_hyst) {
> >> +
> >> +               ret = tz->ops->set_trip_hyst(tz, trip_id, trip->hysteresis);
> >> +               if (ret)
> >> +                       goto out;
> >> +       }
> >> +
> >> +       if (((t.temperature != trip->temperature) ||
> >> +            (t.hysteresis != trip->hysteresis)) && tz->trips)
> >> +               tz->trips[trip_id] = *trip;
> >
> > I would write this as
> >
> > if (tz->trips && (t.temperature != trip->temperature || t.hysteresis
> > != trip->hysteresis))
> >          tz->trips[trip_id] = *trip;
>
> Ok, sure
>
> > But
> >
> > 1. Do we want to copy the trip type here too?
>
> The function thermal_zone_set_trip() is called from thermal_sysfs.c, it
> is the unique call site. However, I think it is a good idea to check the
> type of the trip point is not changed, even if it is not possible with
> the actual code.
>
> > 2. If tz->trips is set, do we still want to invoke ->set_trip_temp()
> > or ->set_trip_hyst() if they are present?
>
> No but there are bogus drivers setting the interrupt with these ops
> instead of using the set_trips ops (eg. [1][2][3]). So in order to keep
> those working ATM, I'm keeping them and when all the drivers will be
> changed, I'll wipe out the set_trip_* ops from everywhere.

Do those drivers set tz->trips?  If not, the tz->trips check can go
before the ops ones.

> [1] drivers/thermal/samsung/exynos_tmu.c
> [2] drivers/thermal/qcom/qcom-spmi-temp-alarm.c
> [3] drivers/thermal/imx_thermal.c