> > The best solution I've come up with to use myself currently is
> > to use a serial port for the debugger.
> 
> Yeah--unfortunately not too handy for the average person dealing with a
> single-user workstation getting kernel panics and just wanting to get a
> register dump or stack trace...

You don't need ddb to do that.  you could do gdb on a crash dump.
Indeed, that's probably a lot easier for "the average person," because
i'm _quite_ sure that the average person is more familiar with GDB
than with ddb.  8-)

Sure, for debugging a kernel that's in the process of dying, ddb on
console is better than gdb on a crash dump (but remote kgdb on the
live kernel is better yet).  However, do (perhaps: should) average
users be doing that?


One of the things I like about Digital UNIX (formerly DEC OSF/1) is
that after it crashes, when rebooting and saving the kernel core dump,
it also extracts 'interesting' bits of information into a text file
for easy perusal.



chris