Can  someone confirm that the cursor fix works properly 
	a) when X11 is started -- the Xserver gets to show a cursor
	b) after X11 exits

testing a kernel without the pmcursorOn() call I added would be nice,
but isn't necesary.

I'll commit the fix as soon as i hear it hasn't broken X.

To Thorsten, or anyone else with a 3100 mfb:  you can always try doing

	ln -s /dev/fb0 /dev/pm
	ln -s /dev/fb0 /dev/pm0

until a  rebuilt Xmfbpmax is available.

thanks
--Jonathan