powerpc/book3s: Fix endianess issue for HMI handling on napping cpus.

(NOTE: This patch depends on upstream HMI handling patchset at
 https://lists.ozlabs.org/pipermail/linuxppc-dev/2014-July/119731.html

)

The current HMI handling on napping cpus does not take care of endianess
issue. On LE host kernel when we wake up from nap due to HMI interrupt we
would checkstop while jumping into opal call. There is a similar issue in
case of fast sleep wakeup where the code invokes opal_resync_tb opal call
without handling LE issue. This patch fixes that as well.

With this patch applied, HMIs handling on LE host kernel works fine.

Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>