> Date: Mon, 14 Oct 2024 11:53:57 +0000
> From: Miod Vallat <miod@online.fr>
> 
> > > Johannes should really be using the GENERIC.MP config.
> > 
> > Apologies for the incompetence, I didn't know that this would make a
> > difference.
> 
> What Mark means is that the patch modifies the GENERIC and GENERIC.MP
> kernel configuration to add a "pckbc at acpi" line. If your CUSTOM
> kernel lacks this line, then the acpi attachment will obviously never
> occur.
> 
> >             Anyways, I'm happy to report that I'm writing this email
> > from my laptop keyboard :) However, on a few occasions now (maybe 5 or 6
> > times) while writing this email, keyboard input froze for like a second
> > or so and suddenly the last pressed key got repeated many times. After
> > these hiccups I can resume typing normally. Have you experienced this
> > too?
> 
> No. I am puzzled as your DSDT bits reported an edge-triggered interrupt,
> so I was expecting pckbc@isa to work and pckbc@acpi to work as badly as
> pckbc@isa. Must be something magic (but then, it's ACPI, so it's a black
> box anyway)

The polarity seems to be different though.  I believe normal ISA
interrupts are edge-triggered and "active-high".  But the DSDT for
this machine says the interrupt is "active-low".  So using pckbdc@acpi
will make a difference here.

I used quotation marks around "active-high" and "active-low", because
those terms don't really make sense for edge-triggered interrupts.
They really mean rising edge or falling edge in the context of
edge-triggered interrupts.  And what goes up must come down, so every
rising edge must be followed by a falling edge.  But there probably is
some subtlety in the detection logic that requires things to
configured properly.