Re: 2.6.0-test5 atkbd.c: Unknown key (100% reproduceable)

[Vojtech Pavlik]

On Mon, Sep 15, 2003 at 11:45:25PM +1000, Dmitri Katchalov wrote:
> > On Sun, Sep 14, 2003 at 06:12:49PM +1000, Dmitri Katchalov wrote:
> > > Quoting Andries Brouwer <aebr@xxxxxxxxxx>:
> >
> > > > On Sat, Sep 13, 2003 at 06:51:14PM +1000, Dmitri Katchalov wrote:
> > > >
> > > > > I'm consistently getting this error:
> > > > >
> > > > > atkbd.c: Unknown key (set 2, scancode 0xab, on isa0060/serio0)
> pressed.
> > > > > This happens whenever I type 'f' in "<F7>usbdevfs".
> >
> > It seems most likely that this keyboard is broken.
> > Instead of 0xa1 (f release) you get 0xab (\ release).
> > By some coincidence 0xab is f release in untranslated scancode set 2.
> > (But you are in translated scancode set 2, otherwise the other letters
> > would also have produced different codes.)
> >
> > Could you try to run "showkey -s" on the console under 2.4.*?
> > Hit and release the f a few times. Type <F7>usbdevfs.
> >
> > Just a broken key is something I have seen lots of times.
> > Since for most operating systems make codes are important
> > while break codes (other than those for Shift, Ctrl, Alt)
> > are not, a key with broken release code is usually harmless.
> >
> > This case seems interesting because, if I understand you correctly,
> > the f in itself is not always broken, but this error occurs after
> > a particular sequence of keystrokes.
> 
> I've just run showkey. This is WEIRD! I've never seen anything
> like it (almost)! It is definitely a bug in my keyboard.
> Whenever 'f' occurs in the stream exactly 12 scancodes
> after <F7> it reports wrong release code. It even has a "queue"
> so that multiple instances of the bug can be pipelined :)

Interesting. Can you check if it goes away with a different keyboard?
How about when you disable ACPI?

> Interestingly the bug has no ill effects on 2.4.18 at all. It just works.
> In 2.6.0-test5 I'm getting a message right across the screen followed
> by zillions of 'f's.
> 
> Andries, thanks for your time and for the hints you gave me. I'll put a
> workaround in my kernel. I don't think it will be useful to anyone else
> though:)

Next available 2.6 should have that worked around.

-- 
Vojtech Pavlik
SuSE Labs, SuSE CR
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/