Hi, >As I said, maybe the problem is fixed now, but at the same time there >is something nice about the interrupt type only being specified in one >place. ...and IIRC the device tree validator gets upset if you don't >specify the interrupt type there, so removing it from the source code >seems nice...
My device tree does have interrupts already specified for the node, so allowing the driver to use that does seem like a good idea. > [1] > https://stackoverflow.com/questions/40011799/mapping-device-tree-interrupt-flags-to-devm-request-irq The link you sent suggests just using "0" for the flag value lets the device tree flags to be used. I couldn't do this because the IRQF_ONESHOT needed to be passed as a flag or I'd get stack traces every time the irq was triggered. Several of the other drivers in drivers/gpu/drm/bridge only pass IRQ_ONESHOT so I think those might be using the device tree flags with the oneshot flag. As far as I could tell, only passing the oneshot flag and setting the interrupt level high in the device tree works.
