Merge tag 'iio-for-3.14b' of git://git.kernel.org/pub/scm/linux/kernel/git/jic23/iio into staging-next

Jonathan writes:

2nd round of new IIO drivers, features and cleanups for the 3.14 cycle.

New drivers

* HID inclinometer driver.

* DHT11 humidity driver.  Note that previous humidity drivers have been in
  hwmon, but no one was ever entirely happy with that, and they should find
  a more comfortable home in IIO (their original placement in hwmon was my
  fault - oops).  As this is our first humidity driver, core support is also
  added.

New features

* Two of mxs-lradc channels are internally wired to a temperature sensor,
  make this explicit in the driver by providing the relevant temperature
  channel.

* Add support for blocking IO on buffers.

* Add a data_available call back to the interface between buffer implementations
  and the core.  This is much cleaner than the old, 'stufftoread' flag.
  Implemented in the kfifo buffer.

Cleanups

* Last user of the old event configuration interface is converted and the
  old interface dropped.  Nice to be rid of this thanks to Lars-Peter's hard
  work!

* Replace all remaining instances of the IIO_ST macro with explicit filling
  of the scan_type structure within struct iio_chan_spec.  This macro was a
  bad idea, that rapidly ceased to cover all elements of the structure.
  Miss reading of the macro arguements has led to a number of bugs so lets
  just get rid of it. The final removal patch is awaiting for some fixes
  to make their way into mainline.
  In a couple of drivers, no elements of scan_type were even being used so
  in those case, it has been dropped entirely.

* Drop a couple of of_match_ptr helper uses in drivers where devicetree is
  not optional and hence the structures being protected by this always exist.

* Fix up some cases where data was read from a device in a particular
  byte order, but he code placed it into a s16 or similar.  These were
  highlighted by Sparse.

* Use the new ATTRIBUTE_GROUPS macro to drop some boiler plate in the triggers
  core code.

* ad7746 and ad7280a - stop storing buffers on the stack, giving cleaner code
  and possibly avoiding issues with i2c bus drivers that assume they can dma
  directly into the buffer.  Note that this cannot currently happen as the the
  i2c_smbus_read_i2c_block_data function has a memcpy from the buffer actually
  passed to the bus driver.  I missed this element of the commit message
  and don't think it is major enough to rebase the iio tree.

* ad5791 and ad5504 stop storing buffers on the stack for an SPI driver.
  Unlike the i2c drivers, this is a real issue for SPI drivers which can dma
  directly into the buffer supplied.

1 files changed, 6 insertions, 1 deletions

diff --git a/drivers/iio/dac/ad5764.c b/drivers/iio/dac/ad5764.c
index a8ff5b2ed13..d0d38165339 100644
--- a/drivers/iio/dac/ad5764.c
+++ b/drivers/iio/dac/ad5764.c
@@ -83,7 +83,12 @@ enum ad5764_type {
 		BIT(IIO_CHAN_INFO_CALIBSCALE) |			\
 		BIT(IIO_CHAN_INFO_CALIBBIAS),			\
 	.info_mask_shared_by_type = BIT(IIO_CHAN_INFO_OFFSET),	\
-	.scan_type = IIO_ST('u', (_bits), 16, 16 - (_bits))	\
+	.scan_type = {						\
+		.sign = 'u',					\
+		.realbits = (_bits),				\
+		.storagebits = 16,				\
+		.shift = 16 - (_bits),				\
+	},							\
 }
 
 #define DECLARE_AD5764_CHANNELS(_name, _bits) \