On Mon, 2024-11-04 at 17:02 +0100, Gregory CLEMENT wrote:

Marek Behún [off-list ref] writes:

quoted

When global LED brightness is changed by pressing the front button on
Turris Omnia, the MCU can produce an interrupt to the CPU. Add the
description of this interrupt to the LED controller node.

Signed-off-by: Marek Behún <kabel@kernel.org>

As the binding had been acked, and this change won't produce any
regression I can already get it.

Applied on mvebu/dt

Thanks,

Gregory

This patch made it into v6.13-rc1, and completely breaks the LEDS on my Turris Omnia CZNIC11.
The entries in /sys/class/leds disappeared, and instead the kernel says:

[   11.746116] i2c 1-002b: deferred probe pending: (reason unknown)

If I just revert the patch on top of v6.13-rc1, all is good again. Kernel config attached, for reference.

Best regards, Klaus

#regzbot introduced: 68bc6a71f56875a7dbde4f44aa4404c214651e1c

quoted

---
 arch/arm/boot/dts/marvell/armada-385-turris-omnia.dts | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/arm/boot/dts/marvell/armada-385-turris-omnia.dts b/arch/arm/boot/dts/marvell/armada-385-turris-omnia.dts
index 43202890c959..83fe00abd652 100644
--- a/arch/arm/boot/dts/marvell/armada-385-turris-omnia.dts
+++ b/arch/arm/boot/dts/marvell/armada-385-turris-omnia.dts

@@ -251,6 +251,7 @@ mcu: system-controller@2a {

 			led-controller@2b {
 				compatible = "cznic,turris-omnia-leds";
 				reg = <0x2b>;
+				interrupts-extended = <&mcu 11 IRQ_TYPE_NONE>;
 				#address-cells = <1>;
 				#size-cells = <0>;
 				status = "okay";
-- 
2.45.2