> Also, yesterday I wrote(more like cloned) the keyboard device driver into a
 > controller driver which behaves like a keyboard and I'm getting segfaults of
 > the same kind from the kernel... It was a quick job to get the dreamcast
 > controller up and working... the driver understands key presses and such,
 > however, it is unable to write information to the console.  Whenever it
 > wskbd_input's nothing happens.  I think what might be happing is that for
 > some reason durring wskbd_attach it calls wskbd_load_keymap and in the
 > wskbd_load_keymap it segfaults, thus the keyboard map isn't getting loaded or
 > attached or something.  Anyway, I'm wondering if/how/why the kernel is
 > allowed to fault like this and keep operating... Also if the keyboard map
 > isn't getting loaded why isn't there an error message or something...

Check mkbd's revision you are using. This behavior seems the same problem
which I encontered one week ago. This problem was fixed with rev. 1.6
(this fix was commited three days ago). After this fix, you can use
"options SCIFCONSOLE".

Old mkbd.c had been broken. Even if an another device was selected as
console, mkbd0 set itself as console. This is too bad. Only one device can
act as console.

Perhaps you connected both keyboard and controller. To connect both,
some code have to be written to select which device will act as console.

----------------------------------------------------------
		SAITOH Masanobu (masanobu@iij.ad.jp
				  msaitoh@netbsd.org)