Jonathan's review fixes on v7. Same functionality, except the flush hook,
which now holds the device mode while it drains.

Since v7:
 - the timestamp scan index lands in patch 5, where the enum is
 - the FIFO drain reads with one loop counter, declarations split
 - dps310_fifo_hold_free() is gone. It was a bare kfree() once the
   clearing went, and the alloc side sets the fields instead
 - the flush hook holds the mode across the drain, so a concurrent
   buffer disable cannot free active_scan_mask under it. Found by
   Sashiko on v7. Done with IIO_DEV_GUARD_CURRENT_MODE() and not the
   buffer mode claim, which patch 10 warns on, see the reply on 9/10

Retested on a DPS310 on a BeagleBone Black: six of six buffer enables
produce data, and 64 scan reads come back whole at watermark 1, 8 and 24.
Built with clang 23 and CONFIG_WARN_CONTEXT_ANALYSIS, no warnings.

Patch 10 opts into clang context analysis and comes last, so it can be
dropped alone.

Deferred to a cleanup series: min_t() to min(), 1000 to KILO,
30 * USEC_PER_MSEC, David's define sort, c00/c10 via get_unaligned_be24().

Rupesh Majhi (10):
  iio: pressure: dps310: fix CFG_REG bit definitions
  iio: pressure: dps310: use a local device pointer in probe
  iio: pressure: dps310: use get_unaligned_be24() for the 24-bit results
  iio: pressure: dps310: take the lock once per raw read
  iio: pressure: dps310: add triggered buffer support
  iio: core: add an accessor for scan_timestamp
  iio: pressure: dps310: read buffered samples from the hardware FIFO
  iio: pressure: dps310: derive the drain interval from the watermark
  iio: pressure: dps310: implement .hwfifo_flush_to_buffer()
  iio: pressure: dps310: check the lock markings with context analysis

 drivers/iio/pressure/Kconfig  |   2 +
 drivers/iio/pressure/Makefile |   2 +
 drivers/iio/pressure/dps310.c | 726 ++++++++++++++++++++++++++++++----
 include/linux/iio/iio.h       |   9 +
 4 files changed, 652 insertions(+), 87 deletions(-)

-- 
2.43.0