x86, UV: Fix smp_processor_id() use in a preemptable region
authorcpw@sgi.com <cpw@sgi.com>
Tue, 21 Jun 2011 12:21:26 +0000 (07:21 -0500)
committerIngo Molnar <mingo@elte.hu>
Tue, 21 Jun 2011 12:50:31 +0000 (14:50 +0200)
Fix a call by tunables_write() to smp_processor_id() within a
preemptable region.

Call get_cpu()/put_cpu() around the region where the returned
cpu number is actually used, which makes it non-preemptable.

A DEBUG_PREEMPT warning is prevented.

UV does not support cpu hotplug yet, but this is a step toward
that ability as well.

Signed-off-by: Cliff Wickman <cpw@sgi.com>
Link: http://lkml.kernel.org/r/20110621122242.086384966@sgi.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
arch/x86/platform/uv/tlb_uv.c

index 68e467f69fec8b022446de87ff60d4327b38f893..34be6509330956fa9e990147ae6fd39045ac4a11 100644 (file)
@@ -1334,9 +1334,10 @@ static ssize_t tunables_write(struct file *file, const char __user *user,
 
        instr[count] = '\0';
 
-       bcp = &per_cpu(bau_control, smp_processor_id());
-
+       cpu = get_cpu();
+       bcp = &per_cpu(bau_control, cpu);
        ret = parse_tunables_write(bcp, instr, count);
+       put_cpu();
        if (ret)
                return ret;