Hi!

On Thu 2017-06-29 09:31:02, Pali Rohár wrote:
> On Thursday 29 June 2017 00:44:27 Bastien Nocera wrote:
> > On Wed, 2017-06-28 at 22:15 +0200, Pavel Machek wrote:
> > > 
> > <snip>
> > > While policy normally belongs to userspace, I'd argue this is
> > > workaround for a hardware bug, and in-kernel solution would be
> > > acceptable.
> > > 
> > > Anyway, disable attribute would be nice first step.
> > 
> > It's already fixed for those of us on recent distributions. The
> > "ID_INPUT_TOUCHPAD_INTEGRATION=internal" touchpads will be disabled
> > when the lid is closed, when libinput is used to process the events.
> 
> But this does not fix other usage of /dev/input/* and also does not fix
> pressing spurious keys in linux virtual tty (ctrl+alt+f1). So it is not
> a fix.
> 
> Also important question is: How you detect which input device is
> "internal", non-removable part of notebook and which one is external?
> 
> You can have external USB touchpad, and also you can have external PS/2
> keyboard connected to docking station (which was e.g. my situation).
> 
> Also there are PS/2 to active USB converters, to make whole situation
> complicated.
> 
> And moreover some internal notebook keyboards are connected via USB and
> some touchpads via i2c/smbus.
> 
> I think this detection is not easy or at least I have no idea how to do
> properly. Existence of PS/2 keyboard does not mean it is internal and
> existence of USB keyboard does not mean it is external.
> 
> Maybe ACPI/DSDT provides some information? (No idea, just asking)

I'm not sure it is complex. You simply add DMI blacklist of the bad
systems, with IDs of bad devices.

									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html