Bug#594089: keyboard-configuration: caps lock keycode problem

[Samuel Thibault <sthibault@debian.org>]

Tom Vier, le Fri 03 Sep 2010 16:35:47 -0400, a écrit :
> On 09/02/2010 03:46 PM, Samuel Thibault wrote:
> >Hello,
> >
> >Tom Vier, le Thu 02 Sep 2010 15:11:49 -0400, a écrit :
> >>It says UNICODE mode and the codes for caps lock are:
> >>
> >>press: 0x3a 0xe0 0x66
> >>repeat: 0x3a
> >>release: 0xba 0xe0 0x
> >
> >0xe6 I guess?
> 
> Yes.
> 
> >>I also just discovered the right shift key goes back a page in
> >>iceweasel. Here are the right-shift codes:
> >>
> >>showkey -s:
> >>press: 0x36 0xe0 0x6a
> >>repeat: 0x36
> >>release: 0xb6 0xe0 0xea
> >
> >All this looks like one of the few keyboard glitches that linux-2.6
> >fixes for some keyboards known to be bogus, while other keyboards do use
> >these scancodes to report e.g. multimedia keys.  Could you try
> >showkey -s with the older, working kernel?
> 
> Stable (2.6.26) which worked fine gives the same as Testing (2.6.32):

When you say "Stable", you mean both kernel and Xorg?  Could you also
run xev on your stable box?

> press: 0x3a 0xe0 0x66
> repeat: 0x3a
> release: 0xba 0xe0 0xe6
> 
> ...which makes me think it's an xorg problem, but I'm not familiar with 
> how the whole keyboard stack works.

It's a big mess :)

> I wonder why the logitech generates so many codes.

It looks to me like a keyboard bug.  Do you see any pattern, e.g. does
the left shift key produce one too?  Do the keys which do produce extra
scancodes form any particular shape on the keyboard?

Such bug should probably be fixed in the kernel, just like it is already
in atkbd.c for some keyboards.

Samuel