Re: [PATCH 1/2] input: adp5588-keys - get value from data out when dir out

[Jean-Francois Dagenais <jeff.dagenais@gmail.com>]

On Apr 21, 2012, at 2:47, Dmitry Torokhov wrote:

> Hi Jean-François,
> 
> On Fri, Apr 20, 2012 at 11:32:00AM -0400, Jean-François Dagenais wrote:
>> As discussed here: http://ez.analog.com/message/35852,
>> the 5587 revC and 5588 revB spec sheets contain a mistake
>> in the GPIO_DAT_STATx register description.
>> 
>> According to R.Shnell at ADI, as well as my own
>> observations, it should read:
>> "GPIO data status (shows GPIO state when read for inputs)".
>> 
>> This commit changes the get value function accordingly.
>> 
>> A similar patch for gpio-adp5588 follows.
>> 
>> Signed-off-by: Jean-François Dagenais <jeff.dagenais@gmail.com>
>> Acked-by: Michael Hennerich <michael.hennerich@analog.com>
>> ---
>> drivers/input/keyboard/adp5588-keys.c |    7 +++++++
>> 1 files changed, 7 insertions(+), 0 deletions(-)
>> 
>> diff --git a/drivers/input/keyboard/adp5588-keys.c b/drivers/input/keyboard/adp5588-keys.c
>> index 6412ced..b7a0f1a 100644
>> --- a/drivers/input/keyboard/adp5588-keys.c
>> +++ b/drivers/input/keyboard/adp5588-keys.c
>> @@ -78,6 +78,13 @@ static int adp5588_gpio_get_value(struct gpio_chip *chip, unsigned off)
>> 	unsigned int bank = ADP5588_BANK(kpad->gpiomap[off]);
>> 	unsigned int bit = ADP5588_BIT(kpad->gpiomap[off]);
>> 
>> +	mutex_lock(&kpad->gpio_lock);
>> +	if (kpad->dir[bank] & bit) {
>> +		int result = !!(kpad->dat_out[bank] & bit);
>> +		mutex_unlock(&kpad->gpio_lock);
>> +		return result;
>> +	}
>> +	mutex_unlock(&kpad->gpio_lock);
>> 	return !!(adp5588_read(kpad->client, GPIO_DAT_STAT1 + bank) & bit);
> 
> This locking looks wrong as it is possible for adp5588_gpio_get_value()
> to get scheduled out after checking kpad->dir[bank] and releasing
> kpad->gpio_lock while another thread executes
> adp5588_gpio_direction_output() and modifies kpad->dir[bank].
> 
> You should keep mutex for the entire duration; something like this:
> 
> 	mutex_lock(&kpad->gpio_lock);
> 	if (kpad->dir[bank] & bit)
> 		val = kpad->dat_out[bank];
> 	else
> 		val = adp5588_read(kpad->client, GPIO_DAT_STAT1 + bank);
> 	mutex_unlock(&kpad->gpio_lock);
> 
> 	return !!(val & bit);
> 
Agreed, I think I wanted to minimize refactoring a bit too much and got distracted. I will send V2 shortly for both drivers.
> Thanks.
Thank you!
> 
> -- 
> Dmitry