x86, mce: disable machine checks on offlined CPUs

Impact: Lower priority bug fix

Offlined CPUs could still get machine checks, but the machine check handler
cannot handle them properly, leading to an unconditional crash. Disable
machine checks on CPUs that are going down.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>

1 files changed, 23 insertions, 0 deletions

diff --git a/arch/x86/kernel/cpu/mcheck/mce_64.c b/arch/x86/kernel/cpu/mcheck/mce_64.c
index 4e2b1bc5131c..1db94c0d5aaf 100644
--- a/arch/x86/kernel/cpu/mcheck/mce_64.c
+++ b/arch/x86/kernel/cpu/mcheck/mce_64.c
@@ -906,6 +906,27 @@ static __cpuinit void mce_remove_device(unsigned int cpu)
 	cpu_clear(cpu, mce_device_initialized);
 }
 
+/* Make sure there are no machine checks on offlined CPUs. */
+static void __cpuexit mce_disable_cpu(void *h)
+{
+	int i;
+
+	if (!mce_available(&current_cpu_data))
+		return;
+	for (i = 0; i < banks; i++)
+		wrmsrl(MSR_IA32_MC0_CTL + i*4, 0);
+}
+
+static void __cpuexit mce_reenable_cpu(void *h)
+{
+	int i;
+
+	if (!mce_available(&current_cpu_data))
+		return;
+	for (i = 0; i < banks; i++)
+		wrmsrl(MSR_IA32_MC0_CTL + i*4, bank[i]);
+}
+
 /* Get notified when a cpu comes on/off. Be hotplug friendly. */
 static int __cpuinit mce_cpu_callback(struct notifier_block *nfb,
 				      unsigned long action, void *hcpu)
@@ -929,11 +950,13 @@ static int __cpuinit mce_cpu_callback(struct notifier_block *nfb,
 	case CPU_DOWN_PREPARE:
 	case CPU_DOWN_PREPARE_FROZEN:
 		del_timer_sync(t);
+		smp_call_function_single(cpu, mce_disable_cpu, NULL, 1);
 		break;
 	case CPU_DOWN_FAILED:
 	case CPU_DOWN_FAILED_FROZEN:
 		t->expires = round_jiffies_relative(jiffies + next_interval);
 		add_timer_on(t, cpu);
+		smp_call_function_single(cpu, mce_reenable_cpu, NULL, 1);
 		break;
 	}
 	return NOTIFY_OK;