diff mbox series

[v1] opp: Release current OPP properly

Message ID 20210812101353.14318-1-digetx@gmail.com
State New
Headers show
Series [v1] opp: Release current OPP properly | expand

Commit Message

Dmitry Osipenko Aug. 12, 2021, 10:13 a.m. UTC
The current_opp is released only when whole OPP table is released,
otherwise it's only marked as removed by dev_pm_opp_remove_table().
Functions like dev_pm_opp_put_clkname() and dev_pm_opp_put_supported_hw()
are checking whether OPP table is empty and it's not if current_opp is
set since it holds the refcount of OPP, this produces a noisy warning
from these functions about busy OPP table. Release current_opp when
OPP table is removed to fix it.

Cc: stable@vger.kernel.org
Fixes: 81c4d8a3c414 ("opp: Keep track of currently programmed OPP")
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
---
 drivers/opp/core.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

Comments

Viresh Kumar Aug. 13, 2021, 4:12 a.m. UTC | #1
On 12-08-21, 13:13, Dmitry Osipenko wrote:
> The current_opp is released only when whole OPP table is released,

> otherwise it's only marked as removed by dev_pm_opp_remove_table().

> Functions like dev_pm_opp_put_clkname() and dev_pm_opp_put_supported_hw()

> are checking whether OPP table is empty and it's not if current_opp is

> set since it holds the refcount of OPP, this produces a noisy warning

> from these functions about busy OPP table. Release current_opp when

> OPP table is removed to fix it.

> 

> Cc: stable@vger.kernel.org

> Fixes: 81c4d8a3c414 ("opp: Keep track of currently programmed OPP")

> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>

> ---

>  drivers/opp/core.c | 9 ++++++---

>  1 file changed, 6 insertions(+), 3 deletions(-)

> 

> diff --git a/drivers/opp/core.c b/drivers/opp/core.c

> index b335c077f215..73da968b5c86 100644

> --- a/drivers/opp/core.c

> +++ b/drivers/opp/core.c

> @@ -1378,9 +1378,6 @@ static void _opp_table_kref_release(struct kref *kref)

>  	list_del(&opp_table->node);

>  	mutex_unlock(&opp_table_lock);

>  

> -	if (opp_table->current_opp)

> -		dev_pm_opp_put(opp_table->current_opp);

> -

>  	_of_clear_opp_table(opp_table);

>  

>  	/* Release clk */

> @@ -2901,6 +2898,12 @@ void dev_pm_opp_remove_table(struct device *dev)

>  	if (_opp_remove_all_static(opp_table))

>  		dev_pm_opp_put_opp_table(opp_table);

>  

> +	/* Drop reference taken by _find_current_opp() */

> +	if (opp_table->current_opp) {

> +		dev_pm_opp_put(opp_table->current_opp);

> +		opp_table->current_opp = NULL;

> +	}

> +


It is better to drop the reference when the OPP table is really
getting removed.

I think the WARN_ON() in the put_* functions can be dropped. It is
important to check this when the stuff is getting set, like in
dev_pm_opp_set_supported_hw(), but removal is just fine.

-- 
viresh
Dmitry Osipenko Aug. 13, 2021, 5:23 p.m. UTC | #2
13.08.2021 07:12, Viresh Kumar пишет:
> On 12-08-21, 13:13, Dmitry Osipenko wrote:

>> The current_opp is released only when whole OPP table is released,

>> otherwise it's only marked as removed by dev_pm_opp_remove_table().

>> Functions like dev_pm_opp_put_clkname() and dev_pm_opp_put_supported_hw()

>> are checking whether OPP table is empty and it's not if current_opp is

>> set since it holds the refcount of OPP, this produces a noisy warning

>> from these functions about busy OPP table. Release current_opp when

>> OPP table is removed to fix it.

>>

>> Cc: stable@vger.kernel.org

>> Fixes: 81c4d8a3c414 ("opp: Keep track of currently programmed OPP")

>> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>

>> ---

>>  drivers/opp/core.c | 9 ++++++---

>>  1 file changed, 6 insertions(+), 3 deletions(-)

>>

>> diff --git a/drivers/opp/core.c b/drivers/opp/core.c

>> index b335c077f215..73da968b5c86 100644

>> --- a/drivers/opp/core.c

>> +++ b/drivers/opp/core.c

>> @@ -1378,9 +1378,6 @@ static void _opp_table_kref_release(struct kref *kref)

>>  	list_del(&opp_table->node);

>>  	mutex_unlock(&opp_table_lock);

>>  

>> -	if (opp_table->current_opp)

>> -		dev_pm_opp_put(opp_table->current_opp);

>> -

>>  	_of_clear_opp_table(opp_table);

>>  

>>  	/* Release clk */

>> @@ -2901,6 +2898,12 @@ void dev_pm_opp_remove_table(struct device *dev)

>>  	if (_opp_remove_all_static(opp_table))

>>  		dev_pm_opp_put_opp_table(opp_table);

>>  

>> +	/* Drop reference taken by _find_current_opp() */

>> +	if (opp_table->current_opp) {

>> +		dev_pm_opp_put(opp_table->current_opp);

>> +		opp_table->current_opp = NULL;

>> +	}

>> +

> 

> It is better to drop the reference when the OPP table is really

> getting removed.

> 

> I think the WARN_ON() in the put_* functions can be dropped. It is

> important to check this when the stuff is getting set, like in

> dev_pm_opp_set_supported_hw(), but removal is just fine.

> 


Alright, I'll make another patch then.
diff mbox series

Patch

diff --git a/drivers/opp/core.c b/drivers/opp/core.c
index b335c077f215..73da968b5c86 100644
--- a/drivers/opp/core.c
+++ b/drivers/opp/core.c
@@ -1378,9 +1378,6 @@  static void _opp_table_kref_release(struct kref *kref)
 	list_del(&opp_table->node);
 	mutex_unlock(&opp_table_lock);
 
-	if (opp_table->current_opp)
-		dev_pm_opp_put(opp_table->current_opp);
-
 	_of_clear_opp_table(opp_table);
 
 	/* Release clk */
@@ -2901,6 +2898,12 @@  void dev_pm_opp_remove_table(struct device *dev)
 	if (_opp_remove_all_static(opp_table))
 		dev_pm_opp_put_opp_table(opp_table);
 
+	/* Drop reference taken by _find_current_opp() */
+	if (opp_table->current_opp) {
+		dev_pm_opp_put(opp_table->current_opp);
+		opp_table->current_opp = NULL;
+	}
+
 	/* Drop reference taken by _find_opp_table() */
 	dev_pm_opp_put_opp_table(opp_table);
 }