Received-SPF: pass (google.com: domain of linux-kernel+bounces-153684-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
Precedence: bulk
MIME-Version: 1.0
References: <20240422060835.71708-1-W_Armin@gmx.de> <ZiYTJOmwDVaws3lh@surfacebook.localdomain>
In-Reply-To: <ZiYTJOmwDVaws3lh@surfacebook.localdomain>
From: "Rafael J. Wysocki" <rafael@kernel.org>
Date: Mon, 22 Apr 2024 18:15:13 +0200
Message-ID: <CAJZ5v0joULbyKQJ4suqKj7_UZwaA63M+N4AaWSm7vOdM4OZBEA@mail.gmail.com>
Subject: Re: [PATCH RESEND v5] ACPI: fan: Add hwmon support
To: Andy Shevchenko <andy.shevchenko@gmail.com>, Armin Wolf <w_armin@gmx.de>
Cc: mlj@danelec.com, rafael.j.wysocki@intel.com, lenb@kernel.org, 
	jdelvare@suse.com, linux@roeck-us.net, linux@weissschuh.net, 
	ilpo.jarvinen@linux.intel.com, linux-acpi@vger.kernel.org, 
	linux-hwmon@vger.kernel.org, linux-kernel@vger.kernel.org, 
	platform-driver-x86@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Mon, Apr 22, 2024 at 9:36=E2=80=AFAM Andy Shevchenko
<andy.shevchenko@gmail.com> wrote:
>
> Mon, Apr 22, 2024 at 08:08:35AM +0200, Armin Wolf kirjoitti:
> > Currently, the driver does only support a custom sysfs
> > interface to allow userspace to read the fan speed.
> > Add support for the standard hwmon interface so users
> > can read the fan speed with standard tools like "sensors".
> >
> > Tested with a custom ACPI SSDT.
>
> ...
>
> > +/*
> > + * fan_hwmon.c - hwmon interface for the ACPI Fan driver
>
> No file name in the file, it makes an unneeded burden if file is going to=
 be
> renamed.
>
> > + *
> > + * Copyright (C) 2024 Armin Wolf <W_Armin@gmx.de>
> > + */
>
> ...
>
> > +#include <linux/acpi.h>
> > +#include <linux/hwmon.h>
>
> Please, follow IWYU pronciple, e.g., missing err.h here.
>
> > +#include <linux/limits.h>
> > +#include <linux/units.h>
>
> ...
>
> > +/* Returned when the ACPI fan does not support speed reporting */
> > +#define FAN_SPEED_UNAVAILABLE        0xffffffff
> > +#define FAN_POWER_UNAVAILABLE        0xffffffff
>
> Shouldn't these be rather as ~0 of the respective types or _MAX (from lim=
its.h)
> or something like that?
>
> ...
>
> > +static struct acpi_fan_fps *acpi_fan_get_current_fps(struct acpi_fan *=
fan, u64 control)
> > +{
> > +     int i;
>
> unsigned
>
> > +     for (i =3D 0; i < fan->fps_count; i++) {
> > +             if (fan->fps[i].control =3D=3D control)
> > +                     return &fan->fps[i];
> > +     }
> > +
> > +     return NULL;
> > +}
>
> ...
>
> > +static umode_t acpi_fan_is_visible(const void *drvdata, enum hwmon_sen=
sor_types type, u32 attr,
> > +                                int channel)
> > +{
> > +     const struct acpi_fan *fan =3D drvdata;
> > +     int i;
>
> unsigned
>
> > +     switch (type) {
> > +     case hwmon_fan:
> > +             switch (attr) {
> > +             case hwmon_fan_input:
> > +                     return 0444;
> > +             case hwmon_fan_target:
> > +                     /* When in fine grain control mode, not every fan=
 control value
> > +                      * has an associated fan performance state.
> > +                      */
> > +                     if (fan->fif.fine_grain_ctrl)
> > +                             return 0;
> > +
> > +                     return 0444;
> > +             default:
>
> > +                     break;
> > +             }
> > +             break;
>
> Instead of two breaks, just return 0 from 'default' case.

I agree here, but for a different reason.

If all of the other cases return from the function, the default one
should return either unless there is a specific reason not to.  IIUC,
there's no such reason in this case.

> > +     case hwmon_power:
> > +             switch (attr) {
> > +             case hwmon_power_input:
> > +                     /* When in fine grain control mode, not every fan=
 control value
> > +                      * has an associated fan performance state.
> > +                      */
>
> /*
>  * Use correct style for multi-line
>  * comment blocks. As in this example.
>  */

Yes, please.

Thanks!