Jack
I made a few changes to your patch to get KDB to work properly (on a
vendor kernel). They mostly center around 3 things.
1. NMI handling didn't seem to work in my kernel (whic:h is not
kernel.org)
2. software or hardware breakpoints would always print the message
1 out of 4 cpus in kdb, waiting for the rest, timeout in 10 second(s)
...3 cpus are not in kdb, their state is unknown
with a 10 second delay. ss and go do the same sort of thing.
3. Finally get_wchan doesn't really work (atleast in my kernel) because
either a task is TASK_RUNNING, or schedule is at the top of the task
frame. Hence get_wchan always returns 0, which makes the backtrace
comamnds not operate properly.
also, I believe there is a missing kdb_ipi call in traps.c, but i'm not
entirely sure about this one.
I'd appreciate comments on this patch as I'm new to x86_64 arch.
After this patch, kdb works like a champ in mv's vendor kernel. Good
work on the port.
regards,
-steve
-- Binary/unsupported file stripped by Ecartis --
-- Type: text/x-patch
-- File: linux-2.6.10-sd.patch
---------------------------
Use http://oss.sgi.com/ecartis to modify your settings or to unsubscribe.
|