Re: pinctrl: qcom: ipq4019: Use of gpio-hog's

[Christian Lamparter <chunkeey@xxxxxxxxx>]

On Montag, 26. März 2018 14:44:52 CEST Sven Eckelmann wrote:
> I have here a board (IPQ4019 based) which requires an GPIO set to output + low 
> to enable the USB. This should be defined via the DTS and thus a gpio-hog 
> seemed to be appropriate in the tlmm:
> 
>     &tlmm {
>         serial_pins: serial_pinmux {
>                 mux {
>                         pins = "gpio16", "gpio17";
>                         function = "blsp_uart0";
>                         bias-disable;
>                 };
>         };
>     
>         spi_0_pins: spi_0_pinmux {
>                 pin {
>                         function = "blsp_spi0";
>                         pins = "gpio13", "gpio14", "gpio15";
>                         drive-strength = <12>;
>                         bias-disable;
>                 };
>                 pin_cs {
>                         function = "gpio";
>                         pins = "gpio12";
>                         drive-strength = <2>;
>                         bias-disable;
>                         output-high;
>                 };
>         };
>     
>         enable-usb-power {
>                 gpio-hog;
>                 gpios = <58 GPIO_ACTIVE_LOW>;
>                 output-low;
>                 line-name = "enable USB2 power";
>         };
>     };
> 
> But this seems to be incompatible with the general design because it now kills 
> the bootup completely.
> [...]
>     [    1.298883] requesting hog GPIO enable USB2 power (chip 1000000.pinctrl, offset 58) failed, -517
>     [    1.299609] gpiochip_add_data: GPIOs 0..99 (1000000.pinctrl) failed to register
>     [    1.308589] ipq4019-pinctrl 1000000.pinctrl: Failed register gpiochip
>     [    1.316586] msm_serial 78af000.serial: could not find pctldev for node /soc/pinctrl@1000000/serial_pinmux, deferring probe
>     [    1.322415] spi_qup 78b5000.spi: could not find pctldev for node /soc/pinctrl@1000000/spi_0_pinmux, deferri
> 
> What is the correct way for qcom,ipq4019-pinctrl to configure a pin during 
> bootup (gpio-hog)? I can use the deprecated gpio sysfs infrastructure for
> now but this seems to be only a dirty hack.
> 
I found the bug/issue and it's sort of specific to OF, pinctrl and gpio-hog.

However I do think that all pinctrl drivers are affect that call
gpiochip_add_pin_range() *after* (devm_)gpiochip_add_data().

According to:
<https://elixir.bootlin.com/linux/v4.16-rc7/ident/gpiochip_add_pin_range>
This seems to be quite a lot actually:
(pinctrl-baytrail, pinctrl-cherryview, pinctrl-intel, not sure if Intel
really is affected. I know there's at least one x86 with OF. But I don't
know if it has a pinctrl. @Andy?)
pinctrl-mt7622, pinctrl-mtk-common.c, pinctrl-abx500.c, pinctrl-msm.c,
pinctrl-as3722.c, pinctrl-at91-pio4.c, pinctrl-axp209.c, pinctrl-coh901.c,
pinctrl-digicolor.c, pinctrl-pistachio.c, pinctrl-sx150x.c

For ipq4019 (and all the others too?), there is this an workaround
that can be done:

Add the gpio-ranges property to the pinctrl node in the qcom-ipq4019.dtsi:

tlmm: pinctrl@1000000 {
        gpio-ranges = <&tlmm 0 0 99>;
};

and remove the gpiochip_add_pin_range() call in:
<https://elixir.bootlin.com/linux/v4.16-rc7/source/drivers/pinctrl/qcom/pinctrl-msm.c#L834>


This will fix the issue of the missing pin ranges, since the ranges
are now being added by of_gpiochip_add_pin_range()
<https://elixir.bootlin.com/linux/v4.16-rc7/source/drivers/gpio/gpiolib-of.c#L618>

which is called before of_gpiochip_scan_gpios() in the same function.
<https://elixir.bootlin.com/linux/v4.16-rc7/source/drivers/gpio/gpiolib-of.c#L629>

[for reference: of_gpiochip_scan_gpios() is doing the the gpio-hog setup for OF.
<https://elixir.bootlin.com/linux/v4.16-rc7/source/drivers/gpio/gpiolib-of.c#L372>]

Regards,
Christian
--
To unsubscribe from this list: send the line "unsubscribe linux-gpio" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html