who broke keyboards again?

[Matthias Andree]

Alfred Perlstein <alfred at freebsd.org> writes:

> I have a client that has several dell machines, when he boots without
> a keyboard he gets:
>
>   atkbdc0: <Keyboard controller (i8042)> at port 0x60,0x64 on isa0
>
> but no:
>   atkbd0: <AT Keyboard> flags 0x1 irq 1 on atkbdc0
>   kbd0 at atkbd0
>
> This used to work, then was broken by someone trying to "do the right
> thing".
>
> We need the keyboards to be hot swappable.  I recommend that the next
> dork that breaks this should have to reenact the lye scene from Fight
> Club.
>
> No other OS has this problem.  Also, breaking this in -stable is
> super lame, c'mon can't we do better?

In this context, detection of Fujitsu-Siemens-Computers KBPC S2
keyboards (looks like normal PS/2 stuff) has also been broken on
-CURRENT somewhen after 5.1. If some of such breakage has propagated
into -STABLE (via MFC), then the person who did the original change
ought to sit down and clear things up.

However, I have been unable to identify a "suspicious" commit.

If however you need to boot without keyboard and plug one in later, then
you cannot go with the default flags of 0x1 for this will disable the
_driver_ when no keyboard is found at boot time. Change it to 0x0, 0x2 or
0x3, whichever works well for you.

I wonder what this whole mess with the keyboard detection/nondetection
and flags=0x1 is about. 0x0 should be the default, if someone knows he
won't ever need the driver (headless), he can remove it from the kernel.

-- 
Matthias Andree

Encrypt your mail: my GnuPG key ID is 0x052E7D95