On 20/04/2023 14:30, Krzysztof Kozlowski wrote: > On 20/04/2023 13:58, Mark Brown wrote: >> On Thu, Apr 20, 2023 at 12:16:12PM +0200, Krzysztof Kozlowski wrote: >> >>> - gpio_direction_output(wcd938x->reset_gpio, 0); >>> - /* 20us sleep required after pulling the reset gpio to LOW */ >>> + gpiod_set_value_cansleep(wcd938x->reset_gpio, 1); >>> + /* 20us sleep required after asserting the reset gpio */ >> >> This is inverting the sense of the GPIO in the API from active low to >> active high which will mean we're introducing a new reliance on having >> the signal described as active low in DT. That's an ABI concern. > > It's bringing it to the correct level. Old code was not respecting the > DTS thus if such DTS came with inverted design, the driver would not work. > > We were already fixing the upstream DTS users and I thought all of them > are fixed since long time (half a year) or even correct from the > beginning. Now I found one more case with incorrect level, which I will fix. No, my bad - all upstream DTSes are corrected since half year. > >> >> I remain deeply unconvinced that remapping active low outputs like this >> in the GPIO API is helping. > > The code is mapping them to correct state. The previous state was > incorrect and did not allow to handle active high (which can happen). > This is the effort to make code correct - driver and DTS. Best regards, Krzysztof
