Re: input layer strangeness

[Vojtech Pavlik <vojtech@suse.cz>]

On Tue, Oct 01, 2002 at 11:12:04PM -0700, Andrew Morton wrote:
> Vojtech Pavlik wrote:
> > 
> > On Tue, Oct 01, 2002 at 09:23:43PM -0700, Andrew Morton wrote:
> > >
> > > It's been doing this ever since the input layer changes:
> > >
> > > - open a few xterms
> > > - press the spacebar, leave pressed
> > > - start waggling the mouse about
> > > - stop pressing spacebar, keep waggling the mouse about,
> > >   across the xterms
> > >
> > > The keystrokes *never* stop coming.  Just the continuous mouse
> > > activity causes a stream of keyboard input, at seemingly the normal
> > > autorepeat rate. I can keep them coming for 30 seconds, just by
> > > moving the mouse.
> > 
> > Do they stop coming when you stop moving the mouse or they don't stop at
> > all? The first would be quite interesting, the second would probably be
> > a missed key release event due to keyboard controller overload by the
> > mouse.
> 
> They stop immediately if I stop moving the mouse.
> 
> > > In practice, it's irritating because it's quite easy to get a
> > > stream of erroneous input dumped into the wrong windows.
> > >
> > > It's a vanilla dual pentium with an AT keyboard and a PS/2
> > > mouse.
> > 
> > Can you check if it happens also on UP? Just want to know if it might be
> > a SMP issue ...
> 
> Will do.

Results? It also might be the keyboard controller in your machine
starves the keyboard input pipe when the mouse is moving and flooding it
with data. This way the release event from the keyboard doesn't come
until you stop moving the mouse. Can you verify that with
I8042_DEBUG_INPUT? Thanks.

-- 
Vojtech Pavlik
SuSE Labs