Re: gpiolib and pinmux interaction

[Linus Walleij <linus.walleij@xxxxxxxxxx>]

Hi Ramon,

Stefan Wahren is working on this for the BCM2835. It is a bit complex,
check our discussion threads:
https://marc.info/?l=linux-gpio&m=154805472108758&w=2

This is the latest version of his patch:
https://marc.info/?l=linux-gpio&m=154805472508764&w=2

I hope Stefan is pursuing this because I agree and I really
like the improvements being done.

Yours,
Linus Walleij


On Tue, Mar 12, 2019 at 10:43 AM Fried, Ramon
<ramon.fried@xxxxxxxxxxxxxxx> wrote:
>
> Hi Linus,
> Have a small question.
>
> In case where pinmux is used with gpiolibd, shouldn't the consumer be
> set to "used" or something if the pin was muxed and acquired by
> different driver ?
>
> The example below refers to pinctrl driver with ability to mux every pin
> as a gpio.
> The output of gpioinfo is misleading, as it appears that all the gpio
> muxable pins are free to be used.
>
> # gpioinfo
> gpiochip0 - 29 lines:
>          line   0:      unnamed       unused   input  active-high
>          line   1:      unnamed       unused   input  active-high
>          line   2:      unnamed       unused   input  active-high
>
> # gpioset gpiochip0 2=1
> [    6.642630] dum-pinctrl 1400000.pinctrl: pin PA2 already requested by
> a00000.uart; cannot claim for gpio-low:469
> [    6.642668] dum-pinctrl 1400000.pinctrl: pin-2 (gpio-low:469) status -22
> gpioset: error setting the GPIO line values: Invalid argument
>
> Thanks,
> Ramon.
>