On Wed, Dec 29, 2021 at 10:11:57AM -0300, Fabio Estevam wrote:
On Wed, Dec 29, 2021 at 9:43 AM Mark Brown broonie@kernel.org wrote:
There might be more out of tree users of course - there's no requirement for people to upstream their DTs. Probably better to play it safe.
If someone correctly describes the gpio_reset as active-low, then the driver will not work.
If there is any out-of-tree user of the gpio_reset property, they are passing the incorrect polarity in the device tree and things are working by pure luck. (wrong polarity in dts + wrong polarity handling in the driver = working state)
Yeah a fair bit of this sort of thing kinda happened in the early gpiod conversion, and it is indeed a bit sad.
Ok, if this can't be fixed, then let's drop patches 2 and 3, which is unfortunate.
I think we can still keep patch 3 here, still valuable to put the device back into reset, even if we are using the backwards reset.
Thanks, Charles