allow cpu_down() to be called earlier

The function's use of the stop-machine logic has so far prevented its
use ahead of the processing of the "ordinary" initcalls. Since at this
early time we're in a controlled environment anyway, there's no need for
such a heavy tool. Additionally this ought to have less of a performance
impact especially on large systems, compared to the alternative of
making stop-machine functionality available earlier.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: 5894c0a2da
master date: 2018-07-16 15:15:12 +0200

xen/common/cpu.c

@@ -68,12 +68,17 @@ void __init register_cpu_notifier(struct notifier_block *nb)
     spin_unlock(&cpu_add_remove_lock);
 }
 
-static int take_cpu_down(void *unused)
+static void _take_cpu_down(void *unused)
 {
     void *hcpu = (void *)(long)smp_processor_id();
     int notifier_rc = notifier_call_chain(&cpu_chain, CPU_DYING, hcpu, NULL);
     BUG_ON(notifier_rc != NOTIFY_DONE);
     __cpu_disable();
+}
+
+static int take_cpu_down(void *arg)
+{
+    _take_cpu_down(arg);
     return 0;
 }
 
@@ -99,7 +104,9 @@ int cpu_down(unsigned int cpu)
         goto fail;
     }
 
-    if ( (err = stop_machine_run(take_cpu_down, NULL, cpu)) < 0 )
+    if ( unlikely(system_state < SYS_STATE_active) )
+        on_selected_cpus(cpumask_of(cpu), _take_cpu_down, NULL, 1);
+    else if ( (err = stop_machine_run(take_cpu_down, NULL, cpu)) < 0 )
         goto fail;
 
     __cpu_die(cpu);