the way around this bug:
After building serveral kernels with different options and using
different devices, the following works:
1) use /dev/lpa as device
2) don't delete the 'irg ?' in the kernel config file
printing should work then ...
I'll investigate further, why /dev/lpt0 won't work ...
EXAMINE on SRM after halt gave:
 PMEM: 0 000000000
don't know if there were more or less zeros, but there was nothing
else.
Another interesting thing:
if I do a 'ktrace -t iencs cat anytext > /dev/lpt0' the system won't
halt! But it won't print anything either ... a ctrl-C terminates it
...mmmh printcap os now OK .. 
whoa !! this is curious - nearly absurd !!
printer lpt is /dev/lpt
printer lp is /dev/lpa

'lpr -Plpt anytext' works !!!
'lpr -Plp any' works (expected)
BUT: lpr on lpt is AWFULL LAME -- 2 line per minute !!!
QUESTION NOW ARISING: why is the polling device faster the 
                      interrupt device OR HAVE THEY BEEN
                      INTERCHANGED ???? this could perhaps
                      explain, why a 'cat anything > /dev/lpt'
                      doesn't work - cat puts to much things
                      out to the polling device who gets 
                      scrambled. lpr does a better job on
                      this. This is just my guess ...
MARTIN