Re: [PATCH v5 0/4] iio: Add support for rotation from north

[Srinivas Pandruvada]

Hi,

Since you have only one attribute, it is working for you.
There is a dereference error in *channels.

Check the attached diff, which will fix this.

Thanks,
Srinivas

On 07/09/2014 03:12 PM, Reyad Attiyat wrote:
> Hey Srinivas,
>
> I did look into the panic you sent. I wasn't sure exactly what caused
> the NULL pointer.
> I tested it with out applying the last rotation from north patch, so
> no hid usages are found, my device only has rotation_from_north, and
> hid_parse_report() would return -EINVAL.
> I added another check to make sure a iio channel were set up, and
> return -EINVAL if not, but couldn't think of anything else.
>
> Could you test this version with dynamic debugging and see if it is
> finding any hid usage attributes.
> Any ideas what could cause this? I think I'm handling errors properly
> by returning what parse_report returns (-EINVAL or -ENOMEM) in probe
>
> Thanks,
> Reyad Attiyat
>
> On Wed, Jul 9, 2014 at 2:45 PM, Srinivas Pandruvada
> <srinivas.pandruvada@xxxxxxxxxxxxxxx> wrote:
> > On 07/09/2014 12:30 PM, Reyad Attiyat wrote:
> > > This series of patches modifies magn-3d driver to handle the rotation
> > > from north usage. This is done by scanning the report and then building
> > > the iio arrays (vals and channels) dynamically.
> > >
> > > Changes from V4
> > > I use the address field of struct iio_chan_spec to hold the array index
> > > of the usage attribute. The scan_index field is generated when creating
> > > an iio channel.
> > >
> > > Reyad Attiyat (4):
> > >     iio: Documentation: Add documentation for rotation from north sensor
> > >       usage attributes
> > >     iio: types: Added support for rotation from north usage attributes
> > >     iio: hid-sensor-magn-3d: Scan for usage attributes before setting up
> > >       iio channels
> > >     iio: hid-sensor-magn-3d: Add support for rotation from north
> > >
> > >    Documentation/ABI/testing/sysfs-bus-iio       |  82 +++++++++++
> > >    drivers/iio/industrialio-core.c               |   4 +
> > >    drivers/iio/magnetometer/hid-sensor-magn-3d.c | 199
> > > ++++++++++++++++++++------
> > >    include/linux/iio/types.h                     |   4 +
> > >    4 files changed, 245 insertions(+), 44 deletions(-)
> > Did you get chance to look at the cause of panic?
> >
> > Thanks,
> > Srinivas

diff --git a/drivers/iio/magnetometer/hid-sensor-magn-3d.c b/drivers/iio/magnetometer/hid-sensor-magn-3d.c
index 070d20e..63f23af 100644
--- a/drivers/iio/magnetometer/hid-sensor-magn-3d.c
+++ b/drivers/iio/magnetometer/hid-sensor-magn-3d.c
@@ -271,6 +271,7 @@ static int magn_3d_parse_report(struct platform_device *pdev,
 	int ret = 0;
 	int i;
 	int attr_count = 0;
+	struct iio_chan_spec *_channels;
 
 	/* Scan for each usage attribute supported */
 	for (i = 0; i < MAGN_3D_CHANNEL_MAX; i++) {
@@ -300,10 +301,10 @@ static int magn_3d_parse_report(struct platform_device *pdev,
 		return  -EINVAL;
 
 	/* Setup IIO channel array */
-	*channels = devm_kcalloc(&pdev->dev, attr_count,
+	_channels = devm_kcalloc(&pdev->dev, attr_count,
 								sizeof(struct iio_chan_spec),
 								GFP_KERNEL);
-	if (!*channels) {
+	if (!_channels) {
 		dev_err(&pdev->dev, "failed to allocate space for iio channels\n");
 		return -ENOMEM;
 	}
@@ -322,14 +323,16 @@ static int magn_3d_parse_report(struct platform_device *pdev,
 	{
 		if (st->magn[i].index >= 0) {
 			/* Setup IIO channel struct */
-			*channels[*chan_count] = magn_3d_channels[i];
+			_channels[*chan_count] = magn_3d_channels[i];
 
 			st->magn_val_addr[i] = &(st->iio_val[*chan_count]);
-			magn_3d_adjust_channel_bit_mask(*channels, *chan_count, st->magn[i].size);
+			magn_3d_adjust_channel_bit_mask(_channels, *chan_count, st->magn[i].size);
 			(*chan_count)++;
 		}
 	}
 
+	*channels = _channels;
+
 	st->scale_precision = hid_sensor_format_scale(
 				HID_USAGE_SENSOR_COMPASS_3D,
 				&st->magn[CHANNEL_SCAN_INDEX_X],