On MPC86xx the watchdog could be enabled only at power-on-reset, and could
not be disabled afterwards.  We must ping the watchdog from the kernel
until the userspace handles it.

MPC83xx CPUs are only differ in a way that watchdog could be disabled
once, but after it was enabled via software it becomes just the same as
MPC86xx.

Thus, to support MPC86xx I added the kernel timer which pings the watchdog
until the userspace opens it.

Since we implemented the timer, now we're able to implement proper
handling for the CONFIG_WATCHDOG_NOWAYOUT case, for MPC83xx and MPC86xx.

Also move the probe code into subsys_initcall, because we want start
pinging the watchdog ASAP, and misc devices are available in
subsys_initcall.
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Cc: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>