Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230113180235.1604526-1-daniel.lezcano@linaro.org>
 <20230113180235.1604526-2-daniel.lezcano@linaro.org> <CAJZ5v0jDF9qrQAQM1AhT-Q4A3Nzyht9XxZoZyS1afoAPt1h_=A@mail.gmail.com>
 <e4468710-1336-847e-89e4-51c745a0a79b@linaro.org> <CAJZ5v0gdXJuNr9jUm6rPdPCuBjc8jwgcZQwtyxurd-jr50btLw@mail.gmail.com>
 <b007d54e-737f-b0cc-2fb3-45a06d5fa044@linaro.org>
In-Reply-To: <b007d54e-737f-b0cc-2fb3-45a06d5fa044@linaro.org>
From:   "Rafael J. Wysocki" <rafael@kernel.org>
Date:   Fri, 20 Jan 2023 19:47:49 +0100
Message-ID: <CAJZ5v0j=4u6eKBiEEBsLJySidSi+VPcGc5Qw0_qK4RVvX4=pzw@mail.gmail.com>
Subject: Re: [PATCH v5 1/3] thermal/acpi: Add ACPI trip point routines
To:     Daniel Lezcano <daniel.lezcano@linaro.org>
Cc:     "Rafael J. Wysocki" <rafael@kernel.org>,
        srinivas.pandruvada@linux.intel.com, linux-pm@vger.kernel.org,
        linux-kernel@vger.kernel.org, linux-acpi@vger.kernel.org,
        rui.zhang@intel.com, christophe.jaillet@wanadoo.fr
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 8BIT
Precedence: bulk

On Fri, Jan 20, 2023 at 7:27 PM Daniel Lezcano
<daniel.lezcano@linaro.org> wrote:
>
> On 20/01/2023 19:15, Rafael J. Wysocki wrote:
> > On Fri, Jan 20, 2023 at 7:08 PM Daniel Lezcano
> > <daniel.lezcano@linaro.org> wrote:
> >>
> >>
> >> Hi Rafael,
> >>
> >>
> >> On 19/01/2023 14:15, Rafael J. Wysocki wrote:
> >>
> >> [ ... ]
> >>
> >>>> +static int thermal_acpi_get_temperature_object(struct acpi_device *adev,
> >>>> +                                              char *object, int *temperature)
> >>>
> >>> So this would become thermal_acpi_get_temp_object(). or even
> >>> thermal_acpi_get_temp() because it really returns the temperature
> >>> value.
> >>>
> >>> I also don't particularly like returning values via pointers, which is
> >>> entirely avoidable here, because the temperature value obtained from
> >>> the ACPI control methods must be a positive number.
> >>>
> >>> So I would make it
> >>>
> >>> static int thermal_acpi_get_temp(struct acpi_device *adev, char *object_name)
> >>> {
> >>
> >> We are converting decikelvin -> millicelsius. Even it is very unlikely,
> >> the result could be less than zero (eg. -1°C). We won't be able to
> >> differentiate -ENODATA with a negative value, no ?
> >>
> >> In the future, it is possible we will have to deal with cold trip points
> >> in order to warm a board. May be we should don't care for now ?
> >
> > My point is that the ACPI specification mandates that the return
> > values be in deciK and so always non-negative.
>
> I understand that but the code does:
>
> static int thermal_acpi_get_temp(struct acpi_device *adev, char
> *object_name)
> {
>         ...
>
>         return deci_kelvin_to_millicelsius(temp);
> }
>
> All the callers do:
>
> ...
>
>          ret = thermal_acpi_get_temp(adev, name);
>          if (ret < 0)
>                 return ret;
>         /* This could be an error
>          * or negative millicelsius temperature
>          */
>
>         /* here we already have millicelsius */
>          trip->temperature = ret;
> ...
>
> So I guess we want to do:
>
> ...
>
>          ret = thermal_acpi_get_temp(adev, name);
>          if (ret < 0)
>                 return ret;
>
>         /* we convert here instead in thermal_acpi_get_temp() */
>          trip->temperature = deci_kelvin_to_millicelsius(ret);
> ...
>
> Sounds good ?

Yes, it does.  Convert when it is known to be valid.