Hi Rafael,

quoted

quoted

Would a patch to handle this specific case be acceptable in the kernel though?

We have quirks. :-)

That said I think intel-hid could check the 0xCE event unconditionally
in the "wakeup" mode, but that will be a separate patch on top of my
series.

I guess I'll wait for your s2-idle-dell-test patch series to land first, to
discuss then how to handle this other Dell model (with Alex Hung?).

In the meantime, I'll update the bug report I had created:
https://bugzilla.kernel.org/show_bug.cgi?id=195897

It seems your series should land in 4.13 so I'm looking at proposing a
patch on top of it, as discussed.

However, I realize that I only know how to handle the wakeup case but
not the suspend one which doesn't work either. Indeed I've always
triggered suspend via the command line so far.

In intel-hid.c I see no obvious code referring to suspend. With added
debug logs, I can see that notify_handler receives the same 0xce / 0xcf
Notify events upon power button press when the system is running.

What would be the expected way then to trigger a normal suspend
sequence, when catching these uncommon events? Should a more
standard event be created instead? If that's a right approach, is
there existing kernel code to do so ?

Thank you,
Jerome