A Andreas Dannenberg #
Message details and related content
This patch series extends the driver to also support bq24250/bq24251.
The bq24250/251/257 devices have a very similar feature set and are
virtually identical from a control register point of view so it made
sense to extend the existing driver rather than submitting a new driver.
In addition to the new device support the driver is also extended to
allow access to some device features previously hidden. Basic and
potentially dangerous charger config parameters affecting the actual
charging of the Li-Ion battery are only configurable through firmware
rather than sysfs properties. However some newly introduced properties
are exposed through sysfs properties as access to them may be desired
from userspace. For example, it is now possible to manually configure
the maximum current drawn from the input source to accommodate different
chargers (0.5A, 1.5A, 2.0A and so on) based on system knowledge a
userspace application may have rather than rely on the auto-detection
mechanism that may not work in all possible scenarios.
Note that most patches have dependencies on other patches in the series.
v8:
- Rebased onto Sebastian's "next" tree and dropped the "power: bq24257:
Remove IRQ config through stat-gpios" patch from the series as it's
already present
- Reworded ti,current-limit DT doc description
- Removed linux/gpio.h include from the platform data patch
- Successfully re-tested the patch that implements the use of the
managed power supply register functions regarding possible side
effects and updated the patch's comment for better clarity. Also
updated the description of that patch to explain the "why".
- Added handling of -EPROBE_DEFER during GPIO pin probe. This error now
leads to an abort of the driver probe and the Kernel will re-try to
probe the driver later (successfully tested).
- Added some of the previous Reviewed-by's from Laurentiu and Krzysztof
(only if there were trivial changes stemming from the discussions or
no changes at all)
v7:
- Fixed an issue that would have impacted running bisect on the series
and double checked the entire series builds one by one by running
a git rebase -i --exec="make ..." ... on the Kernel tree