Re: rc4 and later log message: gpiochip_add_data_with_key: get_direction failed

From: Genes Lists
Date: Sun Mar 16 2025 - 08:49:11 EST


On Tue, 2025-03-11 at 10:40 -0700, Bartosz Golaszewski wrote:
On Tue, 11 Mar 2025 15:03:59 +0100, Genes Lists <lists@xxxxxxxxxxxx> said:
On Sat, 2025-03-08 at 15:45 -0500, Genes Lists wrote:
......


There are two problems here. The issue you're seeing is fixed in next but
not in mainline due to my omission. I will send a patch for that.

On the other hand, the pinctrl driver in question should be fixed too.
Can you try the following change:

diff --git a/drivers/pinctrl/intel/pinctrl-intel.c
b/drivers/pinctrl/intel/pinctrl-intel.c
index d889c7c878e2..0c6925b53d9f 100644
--- a/drivers/pinctrl/intel/pinctrl-intel.c
+++ b/drivers/pinctrl/intel/pinctrl-intel.c
@@ -1068,7 +1068,11 @@ static int intel_gpio_get_direction(struct
gpio_chip *chip, unsigned int offset)

        pin = intel_gpio_to_pin(pctrl, offset, NULL, NULL);
        if (pin < 0)
-               return -EINVAL;
+               /*
+                * For pins configured to functions other than GPIO, default
+                * to the safe INPUT value.
+                */
+               return GPIO_LINE_DIRECTION_IN;

        reg = intel_get_padcfg(pctrl, pin, PADCFG0);
        if (!reg)

?

FYI: This was uncovered by commit 9d846b1aebbe ("gpiolib: check the
return value of gpio_chip::get_direction()").

Bart

Hi Bart - I don't see this pincntrl patch in mainline yet -  what's your thinking on this?

thanks!


-- 
Gene

Attachment: signature.asc
Description: This is a digitally signed message part