On Tue, Aug 14, 2007 at 01:11:45AM +0200, Martin Husemann wrote:
> I have no idea why it seems to work for others (or so I've been told) - I'd
> expect it to blow up on any machine with serial console on a com device.

Ok, after sleep, coffee, and some people pointing it out to me off list:
the obivous difference is the presence of the sc->sc_enable function.

After looking at the code again I think the right solution is to remove all
locking (including the splserial) from com_enable_debugport() and make sure
all callers do it properly (the com_config call in attach_subr does not do
it currently - probably won't matter there, but I guess consistency is better).

Martin