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

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).

------------[ cut here ]------------
WARNING: at drivers/gpio/gpiolib.c:1258 gpio_free+0x100/0x12c()
Modules linked in:
[<c0061208>] (unwind_backtrace+0x0/0xe4) from [<c0091f58>](warn_slowpath_common+0x4c/0x64)
[<c0091f58>] (warn_slowpath_common+0x4c/0x64) from [<c0091f88>](warn_slowpath_null+0x18/0x1c)
[<c0091f88>] (warn_slowpath_null+0x18/0x1c) from [<c024e610>](gpio_free+0x100/0x12c)
[<c024e610>] (gpio_free+0x100/0x12c) from [<c03e9fbc>](ads7846_probe+0xa38/0xc5c)
[<c03e9fbc>] (ads7846_probe+0xa38/0xc5c) from [<c02cff14>](spi_drv_probe+0x18/0x1c)
[<c02cff14>] (spi_drv_probe+0x18/0x1c) from [<c028bca4>](driver_probe_device+0xc8/0x184)
[<c028bca4>] (driver_probe_device+0xc8/0x184) from [<c028bdc8>](__driver_attach+0x68/0x8c)
[<c028bdc8>] (__driver_attach+0x68/0x8c) from [<c028b4c8>](bus_for_each_dev+0x48/0x74)
[<c028b4c8>] (bus_for_each_dev+0x48/0x74) from [<c028ae08>](bus_add_driver+0xa0/0x220)
[<c028ae08>] (bus_add_driver+0xa0/0x220) from [<c028c0c0>](driver_register+0xa8/0x134)
[<c028c0c0>] (driver_register+0xa8/0x134) from [<c0050550>](do_one_initcall+0xcc/0x1a4)
[<c0050550>] (do_one_initcall+0xcc/0x1a4) from [<c00084e4>](kernel_init+0x14c/0x214)
[<c00084e4>] (kernel_init+0x14c/0x214) from [<c005b494>](kernel_thread_exit+0x0/0x8)
---[ end trace 4053287f8a5ec18f ]---

Initialize gpio_pendown in ads7846_probe to -EINVAL before
ads7846_setup_pendown function and using gpio_is_valid function
in conditional check removes the above backtrace
warning.

Signed-off-by: Sourav Poddar <redacted>
Signed-off-by: Kishon Vijay Abraham I <redacted>
Cc: Dmitry Torokhov <redacted>

---
 Links related to the previous discussions:
 http://www.spinics.net/lists/linux-omap/msg45093.html
 http://www.mail-archive.com/linux-omap@vger.kernel.org/msg43506.html

 drivers/input/touchscreen/ads7846.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

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;
 	}
 

@@ -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))
 		gpio_free(ts->gpio_pendown);
  err_cleanup_filter:
 	if (ts->filter_cleanup)

-- 
1.7.0.4

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

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...

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.

You could do the same in the remove-path.

Regards,

   Wolfram

-- 
Pengutronix e.K.                           | Wolfram Sang                |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |

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.

quoted

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)

Ack.

quoted

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?

I thought it to be more readable to have one-entry-one-OK-exit. But
actually I don't mind that much.

quoted

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_... ;)

Yes, I meant that :)

-- 
Pengutronix e.K.                           | Wolfram Sang                |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |