[PATCH] regulator: regulator disable supply fix

[PATCH] regulator: regulator disable supply fix

[Mattias Wallin]

This patch fixes a disable failure when regulator supply is used.
A while loop in regulator disable checks for supply pointer != NULL
but the pointer is not always updated, resulting in the while loop
running too many times causing a disable failure.

Signed-off-by: Mattias Wallin <mattias.wallin@stericsson.com>
---
 drivers/regulator/core.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/drivers/regulator/core.c b/drivers/regulator/core.c
index f1d10c9..dc7d36e 100644
--- a/drivers/regulator/core.c
+++ b/drivers/regulator/core.c
@@ -1424,6 +1424,7 @@ int regulator_disable(struct regulator *regulator)
 	/* decrease our supplies ref count and disable if required */
 	while (supply_rdev != NULL) {
 		rdev = supply_rdev;
+		supply_rdev = NULL;
 
 		mutex_lock(&rdev->mutex);
 		_regulator_disable(rdev, &supply_rdev);
-- 
1.6.3.3

Re: [PATCH] regulator: regulator disable supply fix

[Mark Brown]

On Tue, Nov 02, 2010 at 01:22:29PM +0100, Mattias Wallin wrote:
> This patch fixes a disable failure when regulator supply is used.
> A while loop in regulator disable checks for supply pointer != NULL
> but the pointer is not always updated, resulting in the while loop
> running too many times causing a disable failure.

> Signed-off-by: Mattias Wallin <mattias.wallin@stericsson.com>

Hrm.  This does fix the problem which is needed so:

Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

> ---
>  drivers/regulator/core.c |    1 +
>  1 files changed, 1 insertions(+), 0 deletions(-)
> 
> diff --git a/drivers/regulator/core.c b/drivers/regulator/core.c
> index f1d10c9..dc7d36e 100644
> --- a/drivers/regulator/core.c
> +++ b/drivers/regulator/core.c
> @@ -1424,6 +1424,7 @@ int regulator_disable(struct regulator *regulator)
>  	/* decrease our supplies ref count and disable if required */
>  	while (supply_rdev != NULL) {
>  		rdev = supply_rdev;
> +		supply_rdev = NULL;
>  
>  		mutex_lock(&rdev->mutex);
>  		_regulator_disable(rdev, &supply_rdev);

but this does smell a bit like it ought to be in the _regulator_disable()
call rather than the caller.

Re: [PATCH] regulator: regulator disable supply fix

[Mattias Wallin]

On 11/02/2010 02:19 PM, Mark Brown wrote:
> On Tue, Nov 02, 2010 at 01:22:29PM +0100, Mattias Wallin wrote:
>> This patch fixes a disable failure when regulator supply is used.
>> A while loop in regulator disable checks for supply pointer != NULL
>> but the pointer is not always updated, resulting in the while loop
>> running too many times causing a disable failure.
> 
>> Signed-off-by: Mattias Wallin <mattias.wallin@stericsson.com>
> 
> Hrm.  This does fix the problem which is needed so:
> 
> Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
> 
>> ---
>>  drivers/regulator/core.c |    1 +
>>  1 files changed, 1 insertions(+), 0 deletions(-)
>>
>> diff --git a/drivers/regulator/core.c b/drivers/regulator/core.c
>> index f1d10c9..dc7d36e 100644
>> --- a/drivers/regulator/core.c
>> +++ b/drivers/regulator/core.c
>> @@ -1424,6 +1424,7 @@ int regulator_disable(struct regulator *regulator)
>>  	/* decrease our supplies ref count and disable if required */
>>  	while (supply_rdev != NULL) {
>>  		rdev = supply_rdev;
>> +		supply_rdev = NULL;
>>  
>>  		mutex_lock(&rdev->mutex);
>>  		_regulator_disable(rdev, &supply_rdev);
> 
> but this does smell a bit like it ought to be in the _regulator_disable()
> call rather than the caller.

Would you prefer below solution instead?

diff --git a/drivers/regulator/core.c b/drivers/regulator/core.c
index dc7d36e..51f66b2 100644
--- a/drivers/regulator/core.c
+++ b/drivers/regulator/core.c
@@ -1359,6 +1359,7 @@ static int _regulator_disable(struct regulator_dev *rdev,
                struct regulator_dev **supply_rdev_ptr)
 {
        int ret = 0;
+       *supply_rdev_ptr = NULL;
 
        if (WARN(rdev->use_count <= 0,
                        "unbalanced disables for %s\n",

Re: [PATCH] regulator: regulator disable supply fix

[Mark Brown]

On Tue, Nov 02, 2010 at 02:45:26PM +0100, Mattias Wallin wrote:

> > but this does smell a bit like it ought to be in the _regulator_disable()
> > call rather than the caller.

> Would you prefer below solution instead?

Yes.