Re: [PATCH v3 1/2] ads7846: fix gpio free without requesting

Hi,

On 02/03/11 17:47, Wolfram Sang wrote:

On Thu, Feb 03, 2011 at 08:51:26PM +0530, Sourav Poddar wrote:

quoted

gpio_pendown in ads7846_probe is not getting initalized (defaulted to 0)
resulting in gpio_free being called without a gpio_request. This
results in the following backtrace in bootup (at least on an OMAP3430 SDP).

I wonder if it makes sense to merge both patches under the name of "fix
gpio-handling" or similar. Not sure, though...

I'd rather not do that, because this patch fixes the request/free problem
and the second is changing the functionality (e.g. configures the gpio as input)

quoted

diff --git a/drivers/input/touchscreen/ads7846.c b/drivers/input/touchscreen/ads7846.c
index 14ea54b..ce5baee 100644
--- a/drivers/input/touchscreen/ads7846.c
+++ b/drivers/input/touchscreen/ads7846.c

@@ -952,6 +952,7 @@ static int __devinit ads7846_setup_pendown(struct spi_device *spi, struct ads784
 
 	if (pdata->get_pendown_state) {
 		ts->get_pendown_state = pdata->get_pendown_state;
+		ts->gpio_pendown = -EINVAL;
 		return 0;
 	}

Will probably work, but maybe it is better to reorganize the code to
just have one success-exit-point. That would be mean adding an else
branch to this if-block.

This is something that can be done, though I fear the code readability
will suffer. Is it worth?

quoted

@@ -1353,7 +1354,7 @@ static int __devinit ads7846_probe(struct spi_device *spi)
  err_put_regulator:
 	regulator_put(ts->reg);
  err_free_gpio:
-	if (ts->gpio_pendown != -1)
+	if (gpio_is_valid(ts->gpio_pendown))

You could do the same in the remove-path.

You mean, _should_... ;)
Otherwise, the patch is not complete.

-- 
Regards,
Igor.