Re: gnumach2 and pcmcia

[Roland McGrath]

I think the first problem is just some sketchy interactions between gdb and
the kernel's own fault handling.  Try picking different spots for your
breakpoints and for where you try to step/next; sometimes you need to set a
bkpt at a later likely spot and continue past a chunk instead of trying to
step/next through it.

The latter problem is a real issue, I think.  I believe it bit me last time
I tried to hammer on the network too, but I haven't tried it (at all) in a
while.  It has been a long time now since I had my head fully wrapped
around all the interrupt handling issues, and there were definitely some
subtle complications in what I decided to do and what assumptions I decided
to code in.