On Mon, Feb 21, 2005 at 12:52:21PM +1100, Simon Burge wrote:
> I'm not sure why you're not getting anywhere when you go to multiuser
> mode.  Do you see anything at all after the "root file system" prompt,
> or is it completely silent once /etc/rc starts up?  Could this be some
> wierd serial config (parity maybe) problem?

I see all of the output from rc.d stuff, and the date at the end,
but I never see anything from getty.  while single-user, I can echo
stuff to /dev/console and /dev/tty00 fine, but echoing to /dev/tty02
doesn't show up.

I tried dropping into ddb to try to poke around from there, but that
just got me a stream of traps:

telnet> send bre
Stopped at      0xe00eb6bc:     exit    []
kernel: undefined instruction trap, code=0
fatal undefined instruction in supervisor mode
trap type=10, pc=0xe004fac7, tear=0x8086100, msr=0x9b
panic: trap
Stopped at 0xe00eb6bc:     exit    []
kernel: undefined instruction trap, code=0
fatal undefined instruction in supervisor mode
trap type=10, pc=0xe004fac7, tear=0x8086100, msr=0x9b
panic: trap
...

0xe00eb6bc is in cpu_Debugger(), 0xe004fac7 is in db_command_loop():

e004fac4 <_db_command_loop>:
e004fac4:       82 f8 24        enter   [r3,r4,r5,r6,r7], 36
e004fac7:       be c4 25        movl    f4, tos


> > I tried building the lpt driver to test my changes in that file,
> > but it gets syntax errors.  my change there is pretty simple,
> > just to require that the addr and irq be specified in the config file.
> > I guess no one has tried to use this driver for a long time.
> > 
> > is -current working any better than this for anyone else?
> 
> As I mentioned, I'm running fine multi-user now (with a kernel with your
> patches).

ok, so I guess the more important question for me is, do we need more
testing of those changes, or can I just check them in now?  :-)

-Chuck