On Mon, Dec 17, 2018 at 07:03:31PM +0100, Stephan Gerhold wrote:
Side note: I have considered fixing this in the DSDT table a few times before but have never tried it because it kind of feels wrong to me. It would probably work, but I believe the kernel is at fault here:
I've been fixing mistakes and adding missing GPIOs to the DSDT, but is this really the case here? Everything that is needed for the driver exists in the ACPI table. If this was a BYT-CR device, it would work as-is, with no additional modifications.
Now, in order to fulfill the current expectations for BYT-T devices, I would need to add 4 additional IRQs that would never be used. But which IRQs would I list there? Dummy/invalid IRQs? To me, that does not sound like the DSDT will become any more valid.
Let me know what you think! :)
This thread appears to have died without a conclusion I can see? I don't have strong feelings here, it seems like it's a choice between different evils so people working on the platform should make the call.