[PATCH linux dev-4.10 v6 2/2] drivers/leds: pca955x count LED lines as GPIO

C├ędric Le Goater clg at kaod.org
Mon Jul 31 17:26:16 AEST 2017


On 07/31/2017 04:16 AM, Joel Stanley wrote:
> On Sun, Jul 30, 2017 at 10:20 AM, Brad Bishop
> <bradleyb at fuzziesquirrel.com> wrote:
>>
>>> On Jul 29, 2017, at 2:36 PM, Christopher Bostic <cbostic at linux.vnet.ibm.com> wrote:
>>>
>>> In order for gpio-keys-polled driver to properly register pins
>>> connected to the pca955x chip it is necessary to count 'LED' lines
>>> as part of the total gpio count.  The assocated gpio_chip struct
>>> for the pca955x is then set to all available I/O (LEDS and 'GPIO').
>>> Failure to do so will cause any pin # > GPIO pin count to fail
>>> gpio-keys-polled probe/registration as it falls outside the bounds
>>> of what the gpio_chip struct indicates is available.
>>>
>>> Signed-off-by: Christopher Bostic <cbostic at linux.vnet.ibm.com>
>> Tested-by: Brad Bishop <bradleyb at fuzziesquirrel.com>
>>
>>> ---
>>> drivers/leds/leds-pca955x.c | 2 ++
>>> 1 file changed, 2 insertions(+)
>>>
>>> diff --git a/drivers/leds/leds-pca955x.c b/drivers/leds/leds-pca955x.c
>>> index 9216742..4fa58b9 100644
>>> --- a/drivers/leds/leds-pca955x.c
>>> +++ b/drivers/leds/leds-pca955x.c
>>> @@ -472,6 +472,8 @@ static int pca955x_probe(struct i2c_client *client,
>>>                       ngpios++;
>>>                       break;
>>>               case PCA955X_TYPE_LED:
>>> +                     ngpios++;
>>> +
> 
> Hey Cedric, does this look like the correct fix?
> 

At first glance, I would say no. I will take a look.

C.


More information about the openbmc mailing list