> Date: Sun, 22 Dec 2024 11:45:39 +0100
> From: Landry Breuil <landry@openbsd.org>
> 
> Le Sun, Dec 22, 2024 at 10:51:35AM +0100, Marcus Glocker a écrit :
> > patrick@ found a NetBSD commit which figured out how to map the GPIO
> > pins based on the ACPI tables for X1E devices:
> > 
> > https://github.com/NetBSD/src/commit/575554bd388d93cfff9174abadf2aab93fbfdf8e
> > 
> > Currently we're only using hard coded mapping in qcgpio(4), which
> > might not do the correct mapping, or miss some mappings.
> > 
> > Following a diff which tries to adapt the NetBSD implementation in
> > qcpgio(4) for X1E devices.
> > 
> > Before I throw this diff in my attic, maybe some X1E users want to
> > regression test it, or see if it even fixes some GPIO mappings.
> 
> what could be a symptom of a non-working gpio mapping in dmesg ? missing
> devices ? disappearing devices with the diff ? will try it on the hp
> omnibook.

Note that this diff only changes things when booted with ACPI.  So to
test one a machine that normally boots with a device tree, you have to
enter "mach acpi" at the bootloader "boot>" prompt.

I'd say the feedback we're after is a dmesg diff with/without the
patch.  And a check of whether the machine still boots and whether the
keyboard still works.