linux-kernel - Re: [PATCH] PM: runtime: document common mistake with pm_runtime_get

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAJZ5v0iUQBfrTtVmfrrDixZnnr1_THgaM1+mFu4TRT+OOYb2mw@mail.gmail.com>
Date:   Fri, 23 Apr 2021 16:08:23 +0200
From:   "Rafael J. Wysocki" <rafael@...nel.org>
To:     Krzysztof Kozlowski <krzysztof.kozlowski@...onical.com>
Cc:     "Rafael J. Wysocki" <rjw@...ysocki.net>,
        Len Brown <len.brown@...el.com>, Pavel Machek <pavel@....cz>,
        Linux PM <linux-pm@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Marek Szyprowski <m.szyprowski@...sung.com>
Subject: Re: [PATCH] PM: runtime: document common mistake with pm_runtime_get_sync()

On Thu, Apr 22, 2021 at 6:46 PM Krzysztof Kozlowski
<krzysztof.kozlowski@...onical.com> wrote:
>
> pm_runtime_get_sync(), contradictory to intuition, does not drop the
> runtime PM usage counter on errors which lead to several wrong usages in
> drivers (missing the put).  pm_runtime_resume_and_get() was added as a
> better implementation so document the preference of using it, hoping it
> will stop bad patterns.
>
> Suggested-by: Marek Szyprowski <m.szyprowski@...sung.com>
> Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@...onical.com>
> ---
>  Documentation/power/runtime_pm.rst | 4 +++-
>  include/linux/pm_runtime.h         | 3 +++
>  2 files changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/Documentation/power/runtime_pm.rst b/Documentation/power/runtime_pm.rst
> index 18ae21bf7f92..478f08942811 100644
> --- a/Documentation/power/runtime_pm.rst
> +++ b/Documentation/power/runtime_pm.rst
> @@ -378,7 +378,9 @@ drivers/base/power/runtime.c and include/linux/pm_runtime.h:
>
>    `int pm_runtime_get_sync(struct device *dev);`
>      - increment the device's usage counter, run pm_runtime_resume(dev) and
> -      return its result
> +      return its result;
> +      be aware that it does not drop the device's usage counter on errors so
> +      usage of pm_runtime_resume_and_get(dev) usually results in cleaner code

Whether or not it results in cleaner code depends on what that code does.

If the code is

pm_runtime_get_sync(dev);

<Do something that will fail if the device is in a low-power state,
but there is no way to handle the failure gracefully anyway>

pm_runtime_put(dev);

then having to use pm_runtime_resume_and_get() instead of the
pm_runtime_get_sync() would be a nuisance.

However, if the code wants to check the return value, that is:

error = pm_runtime_resume_and_get(dev);
if (error)
        return error;

<Do something that will crash and burn the system if the device is in
a low-power state>

pm_runtime_put(dev);

then obviously pm_runtime_resume_and_get() should be your choice.

The rule of thumb seems to be whether or not the return value is going
to be used.

>    `int pm_runtime_get_if_in_use(struct device *dev);`
>      - return -EINVAL if 'power.disable_depth' is nonzero; otherwise, if the
> diff --git a/include/linux/pm_runtime.h b/include/linux/pm_runtime.h
> index 6c08a085367b..0dfd23d2cfc3 100644
> --- a/include/linux/pm_runtime.h
> +++ b/include/linux/pm_runtime.h
> @@ -380,6 +380,9 @@ static inline int pm_runtime_get(struct device *dev)
>   * The possible return values of this function are the same as for
>   * pm_runtime_resume() and the runtime PM usage counter of @dev remains
>   * incremented in all cases, even if it returns an error code.
> + * Lack of decrementing the runtime PM usage counter on errors is a common
> + * mistake, so consider using pm_runtime_resume_and_get() instead for a cleaner
> + * code.
>   */
>  static inline int pm_runtime_get_sync(struct device *dev)
>  {
> --
> 2.25.1
>