[PATCH v3 1/3] PM / OPP: add dev_pm_opp_get_suspend_opp() helper

Viresh Kumar viresh.kumar at linaro.org
Sun Sep 6 22:56:11 PDT 2015


On 03-09-15, 20:11, Bartlomiej Zolnierkiewicz wrote:
> Add dev_pm_opp_get_suspend_opp() helper to obtain suspend opp.
> 
> Cc: Viresh Kumar <viresh.kumar at linaro.org>
> Cc: Thomas Abraham <thomas.ab at samsung.com>
> Cc: Javier Martinez Canillas <javier at osg.samsung.com>
> Cc: Krzysztof Kozlowski <k.kozlowski at samsung.com>
> Cc: Marek Szyprowski <m.szyprowski at samsung.com>
> Cc: Tobias Jakobi <tjakobi at math.uni-bielefeld.de>
> Signed-off-by: Bartlomiej Zolnierkiewicz <b.zolnierkie at samsung.com>
> ---
>  drivers/base/power/opp.c | 28 ++++++++++++++++++++++++++++
>  include/linux/pm_opp.h   |  6 ++++++
>  2 files changed, 34 insertions(+)
> 
> diff --git a/drivers/base/power/opp.c b/drivers/base/power/opp.c
> index eb25449..eaafca7 100644
> --- a/drivers/base/power/opp.c
> +++ b/drivers/base/power/opp.c
> @@ -341,6 +341,34 @@ unsigned long dev_pm_opp_get_max_clock_latency(struct device *dev)
>  EXPORT_SYMBOL_GPL(dev_pm_opp_get_max_clock_latency);
>  
>  /**
> + * dev_pm_opp_get_suspend_opp() - Get suspend opp
> + * @dev:	device for which we do this operation
> + *
> + * Return: This function returns pointer to the suspend opp if it is
> + * defined, otherwise it returns NULL.
> + *
> + * Locking: This function takes rcu_read_lock().
> + */
> +struct dev_pm_opp *dev_pm_opp_get_suspend_opp(struct device *dev)
> +{
> +	struct device_opp *dev_opp;
> +	struct dev_pm_opp *suspend_opp;
> +
> +	rcu_read_lock();

This is wrong. Please have a look at other exported APIs that return
an OPP. RCU protected structures must be accessed from within RCU
locks _ONLY_. But your function returns the pointer after dropping
the locks. Which essentially means that the OPP can be freed by core,
by the time you start using it. :)

So, in such cases, its upto the caller to ensure that locks are taken
properly.

You can look at dev_pm_opp_find_freq_exact() as an example.

-- 
viresh



More information about the linux-arm-kernel mailing list