[PATCH][2.6] hardware breakpoint fix/support for xmon

[Anton Blanchard]

> I have seen a number of instances during bringup where one of the CPUs
> will hang in a rtas call (due to a FW DSI), and then xmon would come in
> and disable_surveillance and hang on the rtas lock.  I would suggest
> removing the disable_surveillance() call since there are no boxes that
> I'm aware of that enable surveillance by default.  Maybe make it a
> separate command to disable surveillance similar to KDB.

Interesting, I enable it on all our machines and would have thought we
should be doing it automatically. From a RAS perspective we want to
reboot the machine if it locks up.

Maybe we can have a spin_trylock rtas call and whinge if the lock was
already taken. Once we switch to stopping all cpus on xmon entry we
want to do this to avoid deadlocks anyway.

Anton

** Sent via the linuxppc64-dev mail list. See http://lists.linuxppc.org/