Could it be that remote kernel debugging over a serial line
doesn't always work as it should? For me, it hits breakpoints
and I can examine stack frames and variables. But a "step" and
a "jump" won't do what I'd expect them to do (they all seem to
"continue").

Of course, since I'm debugging this kernel, it may be trashing
some vital data structures of the kgdb stubs. Or is it the phase
of the moon (The Moon is Waxing Crescent (20% of Full))?


Martin