On Fri, 7 Apr 2023 18:04:35 +0100 Jonathan Cameron jic23@kernel.org wrote:
On Wed, 5 Apr 2023 23:22:26 +0200 Patrik Dahlström risca@dalakolonin.se wrote:
This series is based on linux-next/master [1] and [2].
The palmas gpadc block has support for monitoring up to 2 ADC channels and issue an interrupt if they reach past a set threshold. This can be configured statically with device tree today, but it only gets enabled when reaching sleep mode. Also, it doesn't look like anyone is using it.
Instead of this one special case, change the code so userspace can configure the ADC channels to their own needs through the iio events subsystem. The high and low threshold values can be set for every channel, but only 2 thresholds can be enabled at a time. Trying to enable more than 2 thresholds will result in an error.
The configured thresholds will wake up the system from sleep mode if wakeup is enabled in /sys/devices/.../power/wakeup.
The old platform data was removed.
Thresholds, events, and wakeup were tested on omap5-uevm board. It wakes up from sleep mode when wakeup is enabled and a threshold is passed. A userspace tool for monitoring events and adjusting thresholds can be found at [3].
V2 -> V3:
- Rebased to linux-next.
As per reply to the earlier thread. Don't base on linux-next. It can be very unstable though not so much later in a cycle like this.
If there isn't a lot of churn going on in the driver, fine to base on previous release kernel or rc1 (good to say if it is an rc1)
If there is churn underway (which is true here) then iio/togreg + extra patches lists that need to be applied listed in this cover letter.
Just goes to show I focused on the change log and skipped the rest :) As you have it here is fine though change log could have mentioned the extra patch as well even if just "Rebased to linux-next + devm patch."
In this case linux-next is close enough for this driver to the iio/togreg tree that it doesn't matter that it shouldn't be used as a base (no impact in this particular case I think).
Anyhow, all good. I noticed I'd misinterpreted what you'd done here when I saw the context in one of the patches. oops :)
Jonathan
I'm also fine with you just adding the devm patch to this series as the first patch.
Jonathan
- Avoid reconfiguring events on error and when old == new value.
V1 -> V2:
- Begin by removing adc_wakeupX_data instead of doing it last.
- Split changes in smaller patches
[1] git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git [2] https://lore.kernel.org/linux-iio/20230318163039.56115-1-jic23@kernel.org/ [3] https://github.com/Risca/pyra_vol_mon
Patrik Dahlström (7): iio: adc: palmas: remove adc_wakeupX_data iio: adc: palmas: replace "wakeup" with "event" iio: adc: palmas: use iio_event_direction for threshold polarity iio: adc: palmas: move eventX_enable into palmas_adc_event iio: adc: palmas: always reset events on unload iio: adc: palmas: add support for iio threshold events iio: adc: palmas: don't alter event config on suspend/resume
drivers/iio/adc/palmas_gpadc.c | 559 +++++++++++++++++++++++++++------ include/linux/mfd/palmas.h | 8 - 2 files changed, 464 insertions(+), 103 deletions(-)
base-commit: 8417c8f5007bf4567ccffda850a3157c7d905f67 prerequisite-patch-id: b0418c707db13f514400956596e9ebe91c25bba0