1 files changed, 39 insertions, 11 deletions

diff --git a/arch/i386/kernel/nmi.c b/arch/i386/kernel/nmi.c
index eaafe233a5da..a5e34d655965 100644
--- a/arch/i386/kernel/nmi.c
+++ b/arch/i386/kernel/nmi.c
@@ -22,6 +22,7 @@
 #include <linux/percpu.h>
 #include <linux/dmi.h>
 #include <linux/kprobes.h>
+#include <linux/cpumask.h>
 
 #include <asm/smp.h>
 #include <asm/nmi.h>
@@ -42,6 +43,8 @@ int nmi_watchdog_enabled;
 static DEFINE_PER_CPU(unsigned long, perfctr_nmi_owner);
 static DEFINE_PER_CPU(unsigned long, evntsel_nmi_owner[3]);
 
+static cpumask_t backtrace_mask = CPU_MASK_NONE;
+
 /* this number is calculated from Intel's MSR_P4_CRU_ESCR5 register and it's
  * offset from MSR_P4_BSU_ESCR0.  It will be the max for all platforms (for now)
  */
@@ -192,6 +195,8 @@ static __cpuinit inline int nmi_known_cpu(void)
         return 0;
 }
 
+static int endflag __initdata = 0;
+
 #ifdef CONFIG_SMP
 /* The performance counters used by NMI_LOCAL_APIC don't trigger when
  * the CPU is idle. To make sure the NMI watchdog really ticks on all
@@ -199,7 +204,6 @@ static __cpuinit inline int nmi_known_cpu(void)
  */
 static __init void nmi_cpu_busy(void *data)
 {
-        volatile int *endflag = data;
         local_irq_enable_in_hardirq();
         /* Intentionally don't use cpu_relax here. This is
            to make sure that the performance counter really ticks,
@@ -207,14 +211,13 @@ static __init void nmi_cpu_busy(void *data)
            pause instruction. On a real HT machine this is fine because
            all other CPUs are busy with "useless" delay loops and don't
            care if they get somewhat less cycles. */
-        while (*endflag == 0)
-                barrier();
+        while (endflag == 0)
+                mb();
 }
 #endif
 
 static int __init check_nmi_watchdog(void)
 {
-        volatile int endflag = 0;
         unsigned int *prev_nmi_count;
         int cpu;
 
@@ -867,14 +870,16 @@ static unsigned int
 
 void touch_nmi_watchdog (void)
 {
-        int i;
+        if (nmi_watchdog > 0) {
+                unsigned cpu;
 
-        /*
-         * Just reset the alert counters, (other CPUs might be
-         * spinning on locks we hold):
-         */
-        for_each_possible_cpu(i)
-                alert_counter[i] = 0;
+                /*
+                 * Just reset the alert counters, (other CPUs might be
+                 * spinning on locks we hold):
+                 */
+                for_each_present_cpu (cpu)
+                        alert_counter[cpu] = 0;
+        }
 
         /*
          * Tickle the softlockup detector too:
@@ -907,6 +912,16 @@ __kprobes int nmi_watchdog_tick(struct pt_regs * regs, unsigned reason)
                 touched = 1;
         }
 
+        if (cpu_isset(cpu, backtrace_mask)) {
+                static DEFINE_SPINLOCK(lock);   /* Serialise the printks */
+
+                spin_lock(&lock);
+                printk("NMI backtrace for cpu %d\n", cpu);
+                dump_stack();
+                spin_unlock(&lock);
+                cpu_clear(cpu, backtrace_mask);
+        }
+
         sum = per_cpu(irq_stat, cpu).apic_timer_irqs;
 
         /* if the apic timer isn't firing, this cpu isn't doing much */
@@ -1033,6 +1048,19 @@ int proc_nmi_enabled(struct ctl_table *table, int write, struct file *file,
 
 #endif
 
+void __trigger_all_cpu_backtrace(void)
+{
+        int i;
+
+        backtrace_mask = cpu_online_map;
+        /* Wait for up to 10 seconds for all CPUs to do the backtrace */
+        for (i = 0; i < 10 * 1000; i++) {
+                if (cpus_empty(backtrace_mask))
+                        break;
+                mdelay(1);
+        }
+}
+
 EXPORT_SYMBOL(nmi_active);
 EXPORT_SYMBOL(nmi_watchdog);
 EXPORT_SYMBOL(avail_to_resrv_perfctr_nmi);