Keyboard dead on bootup on -test6.

Keyboard dead on bootup on -test6.

[Rob Landley]

Since I switched to test6, my stuck-key problems went away.  (Thank you.  Good 
work.  Judging by the repeat speed, it's back to using hardware repeat.)

However, a problem I'd seen before resurfaced.  Sometimes when I boot up the 
keyboard is completely dead.  It just did this to me (had to use the power 
button, because of course ctrl-alt-del did nothing), and I compared the boot 
logs:

This is what it looks like when it works normally:

Sep 30 16:18:28 localhost kernel: serio: i8042 AUX port at 0x60,0x64 irq 12
Sep 30 16:18:28 localhost kernel: input: AT Translated Set 2 keyboard on 
isa0060/serio0
Sep 30 16:18:28 localhost kernel: serio: i8042 KBD port at 0x60,0x64 irq 1

This was the failure:

Sep 30 16:17:31 localhost kernel: atkbd.c: Unknown key pressed (raw set 0, 
code 0xfc, data 0xfc, on isa0060/serio1).
Sep 30 16:17:31 localhost kernel: serio: i8042 AUX port at 0x60,0x64 irq 12
Sep 30 16:17:31 localhost kernel: serio: i8042 KBD port at 0x60,0x64 irq 1

Under -test5, that failure would have left me with a stuck key endlessly 
repeating (and an otherwise dead keyboard).  Now at least the stuck key part 
has gone away, but the keyboard is still dead until I power cycle the 
machine.

I continue to be using a thinkpad iSeries, I believe it's a 1200C...

Rob

Re: Keyboard dead on bootup on -test6.

[Andries Brouwer]

On Tue, Sep 30, 2003 at 04:32:01PM -0500, Rob Landley wrote:

> This was the failure:
> 
> Sep 30 16:17:31 localhost kernel: atkbd.c: Unknown key pressed (raw set 0, 
> code 0xfc, data 0xfc, on isa0060/serio1).
> Sep 30 16:17:31 localhost kernel: serio: i8042 AUX port at 0x60,0x64 irq 12
> Sep 30 16:17:31 localhost kernel: serio: i8042 KBD port at 0x60,0x64 irq 1
> 
> Under -test5, that failure would have left me with a stuck key endlessly 
> repeating (and an otherwise dead keyboard).  Now at least the stuck key part 
> has gone away, but the keyboard is still dead until I power cycle the 
> machine.

I suppose this is the kernel trying to set LEDs on the mouse,
and the mouse complains.

Andries


(0xfc is a typical mouse error code; also "set 0" suggests this)

Re: Keyboard dead on bootup on -test6.

[Rob Landley]

On Tuesday 30 September 2003 19:52, Andries Brouwer wrote:
> On Tue, Sep 30, 2003 at 04:32:01PM -0500, Rob Landley wrote:
> > This was the failure:
> >
> > Sep 30 16:17:31 localhost kernel: atkbd.c: Unknown key pressed (raw set
> > 0, code 0xfc, data 0xfc, on isa0060/serio1).
> > Sep 30 16:17:31 localhost kernel: serio: i8042 AUX port at 0x60,0x64 irq
> > 12 Sep 30 16:17:31 localhost kernel: serio: i8042 KBD port at 0x60,0x64
> > irq 1
> >
> > Under -test5, that failure would have left me with a stuck key endlessly
> > repeating (and an otherwise dead keyboard).  Now at least the stuck key
> > part has gone away, but the keyboard is still dead until I power cycle
> > the machine.
>
> I suppose this is the kernel trying to set LEDs on the mouse,
> and the mouse complains.
>
> Andries

It's a laptop nipple mouse, a miniature joystick between the G and H keys.  
There are no LED's anywhere near it.  So I'm not surprised it complains.  But 
why does that kill my keyboard?  (You'll notice that the keyboard 
identification line goes missing when this error occurs...)

Rob

Re: Keyboard dead on bootup on -test6.

[Andries Brouwer]

On Tue, Sep 30, 2003 at 08:21:56PM -0500, Rob Landley wrote:

> > > Sep 30 16:17:31 localhost kernel: atkbd.c: Unknown key pressed (raw set
> > > 0, code 0xfc, data 0xfc, on isa0060/serio1).

> > I suppose this is the kernel trying to set LEDs on the mouse,
> > and the mouse complains.

> There are no LED's anywhere near it.  So I'm not surprised it complains.
> But why does that kill my keyboard?

I can conjecture, but reading the facts is easier if you have a debug log
(#define DEBUG in i8042.c).

[In case you already sent one, point at the URL - I've seen so many recently
I lost track who reported what.]

Re: Keyboard dead on bootup on -test6.

[Rob Landley]

On Wednesday 01 October 2003 13:36, Andries Brouwer wrote:
> On Tue, Sep 30, 2003 at 08:21:56PM -0500, Rob Landley wrote:
> > > > Sep 30 16:17:31 localhost kernel: atkbd.c: Unknown key pressed (raw
> > > > set 0, code 0xfc, data 0xfc, on isa0060/serio1).
> > >
> > > I suppose this is the kernel trying to set LEDs on the mouse,
> > > and the mouse complains.
> >
> > There are no LED's anywhere near it.  So I'm not surprised it complains.
> > But why does that kill my keyboard?
>
> I can conjecture, but reading the facts is easier if you have a debug log
> (#define DEBUG in i8042.c).
>
> [In case you already sent one, point at the URL - I've seen so many
> recently I lost track who reported what.]

I'll #define that net recompile and let you know.

Will this make /var/log/messages bigger, or does it do something else?

Rob