On 12/10/18 10:43, Marcel Ziswiler wrote: > On Thu, 2018-10-11 at 19:47 +0200, Linus Walleij wrote: >> On Thu, Oct 11, 2018 at 5:34 PM Marcel Ziswiler >> <marcel.ziswiler@xxxxxxxxxxx> wrote: >> >>> I guess that is also what broke HDMI on Apalis/Colibri T30 causing >>> me >>> to submit a fix [1]. I may also help testing. >> >> I see there are many ways to skin this cat. > > Yes, as a matter of fact I screened the kernel concerning this multi > gpio stuff but could not quite find many examples and no mentioning > anywhere whether or not this is actually allowed. So I kind of assumed > that this may just not really be allowed and cooked up my patch which > is anyway kind of a cleaner solution. I mean explicitly modelling the > GPIO into some intermediate regulator supplying the others. See the function 'regulator_ena_gpio_request()', it states that the same GPIO pin can be shared among regulators. We had the same situation for Tegra124 Jetson TK1 but I don't think that adding a pseudo intermediate regulator is cleaner. If the GPIO controls more than one regulator, I don't see why is it necessary to change the DT. There are several other people reporting the same problem with various different boards. So this does seem to be a common usage. Cheers Jon -- nvpublic
