On 2024/12/18 19:44, Dmitry Baryshkov wrote:
On Wed, Dec 18, 2024 at 11:18:01AM +0800, fange zhang wrote:
On 2024/12/13 18:19, Dmitry Baryshkov wrote:
On Fri, 13 Dec 2024 at 11:21, fange zhang <quic_fangez@xxxxxxxxxxx> wrote:
On 2024/12/10 19:02, Dmitry Baryshkov wrote:
On Tue, Dec 10, 2024 at 02:54:00PM +0800, Fange Zhang wrote:
+ i2c@0 {
+ reg = <0>;
+ #address-cells = <1>;
+ #size-cells = <0>;
+
+ anx7625@58 {
+ compatible = "analogix,anx7625";
+ reg = <0x58>;
+ interrupt-parent = <&ioexp>;
+ interrupts = <0 0>;
will change it to interrupts-extended in next patch
- interrupt-parent = <&ioexp>;
- interrupts = <0 0>;
+ interrupts-extended = <&ioexp 0 IRQ_TYPE_NONE>;
Yes, much better. BTW: are you sure that it's really IRQ_TYPE_NONE?
We extensively tested FALLING and BOTH type, and they all work. However, I
believe it’s better to use the default type, which is the same as the
downstream approach. This way, it will be more stable.
Following downstream is a lame reason. Downstream kernels are frequently
wrong in many ways. So please check the actual documentation for ANX7625
and specify correct interrupt type.
after checked ANX7625 and semtech,sx1509q documentation,
will do these change:
1.change NONE to FALLING for ioexp
- interrupts-extended = <&tlmm 58 IRQ_TYPE_NONE>;
+ interrupts-extended = <&tlmm 58 IRQ_TYPE_EDGE_FALLING>;
2.remove interrupts-extended from anx7625@58
