On 30/11/2024 20:26, Jonathan Cameron wrote:
On Sat, 30 Nov 2024 18:15:06 +0000
Jonathan Cameron <jic23@xxxxxxxxxx> wrote:
On Thu, 28 Nov 2024 11:03:40 +0200
Matti Vaittinen <mazziesaccount@xxxxxxxxx> wrote:
Many of the Kionix/ROHM accelerometers have a "PC1 - bit" which enables
the accelerometer. While a sensor configuration like ODR, g-range, FIFO
status etc. are changed, the PC1 bit must be cleared (sensor must be
disabled). (See the description for different CNTL registers [1])
In order to ensure this the kx022a driver uses a mutex, which is locked
when the PC1 bit is cleared, and held for the duration of the
configuration, and released after PC1 bit is set again (enabling the
sensor).
The locking and PC1 bit toggling was implemented using functions:
kx022a_turn_off_lock() and kx022a_turn_on_unlock().
Based on a discussions [2], the IIO subsystem prefers open-coding the
locking with scoped_guard() over these functions.
Drop the kx022a_turn_off_lock() and kx022a_turn_on_unlock() and use
scoped_guard() instead.
[1]: https://fscdn.rohm.com/kionix/en/datasheet/kx022acr-z-e.pdf
[2]: https://lore.kernel.org/all/20241126175550.4a8bedf3@jic23-huawei/
Signed-off-by: Matti Vaittinen <mazziesaccount@xxxxxxxxx>
---
Revision history:
v2 => v3:
- New patch
NOTE: This patch uses the if_not_cond_guard() which is currently missing
the iio_testing.
https://lore.kernel.org/all/20241001-cleanup-if_not_cond_guard-v1-1-7753810b0f7a@xxxxxxxxxxxx/T/#m69982b23da9f71e72d84855b34e9b142cb3a1920
Looks good to me. If no one else comments, I'll pick this up when
I have the precursor available (so hopefully just after rc1)
or maybe not.
https://lore.kernel.org/all/CAHk-=whn07tnDosPfn+UcAtWHBcLg=KqA16SHVv0GV4t8P1fHw@xxxxxxxxxxxxxx/
Seems Linus is unconvinced.
Hmmm. We might have to roll back the uses of cond_guard() entirely.
Which will be a pain. Ah well. Sometimes an idea turns out to not be as useful
as it initially seemed.
46 instances to get rid of in the tree today...
Ouch! :( Sorry to hear Jonathan.
Yours,
-- Matti
