timer: Fix hotplug regression

The recent timer wheel rework removed the get/put_cpu_var() pair in
the hotplug migration code, which results in:

BUG: using smp_processor_id() in preemptible [00000000] code: hib.sh/2845
...
[<ffffffff810d4fa3>] timer_cpu_notify+0x53/0x12

That hunk is a leftover from an earlier iteration and went unnoticed
so far.

Restore the previous code which was obviously correct.

Fixes: 0eeda71 'timer: Replace timer base by a cpu index'
Reported-and_tested-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

kernel/time/timer.c

@@ -1566,7 +1566,7 @@ static void migrate_timers(int cpu)
 
 	BUG_ON(cpu_online(cpu));
 	old_base = per_cpu_ptr(&tvec_bases, cpu);
-	new_base = this_cpu_ptr(&tvec_bases);
+	new_base = get_cpu_ptr(&tvec_bases);
 	/*
 	 * The caller is globally serialized and nobody else
 	 * takes two locks at once, deadlock is not possible.
@@ -1590,6 +1590,7 @@ static void migrate_timers(int cpu)
 
 	spin_unlock(&old_base->lock);
 	spin_unlock_irq(&new_base->lock);
+	put_cpu_ptr(&tvec_bases);
 }
 
 static int timer_cpu_notify(struct notifier_block *self,