Merge branch 'x86/nmi' into x86/devel

Conflicts:

	arch/x86/kernel/nmi.c
	arch/x86/kernel/nmi_32.c

Signed-off-by: Ingo Molnar <mingo@elte.hu>

6 files changed, 164 insertions, 561 deletions

diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
index d1d4ee895270..8ef4dbf32cf9 100644
--- a/arch/x86/kernel/Makefile
+++ b/arch/x86/kernel/Makefile
@@ -52,7 +52,7 @@ obj-$(CONFIG_X86_32_SMP)	+= smpcommon.o
 obj-$(CONFIG_X86_64_SMP)	+= tsc_sync.o smpcommon.o
 obj-$(CONFIG_X86_TRAMPOLINE)	+= trampoline_$(BITS).o
 obj-$(CONFIG_X86_MPPARSE)	+= mpparse.o
-obj-$(CONFIG_X86_LOCAL_APIC)	+= apic_$(BITS).o nmi_$(BITS).o
+obj-$(CONFIG_X86_LOCAL_APIC)	+= apic_$(BITS).o nmi.o
 obj-$(CONFIG_X86_IO_APIC)	+= io_apic_$(BITS).o
 obj-$(CONFIG_X86_REBOOTFIXUPS)	+= reboot_fixups_32.o
 obj-$(CONFIG_KEXEC)		+= machine_kexec_$(BITS).o
diff --git a/arch/x86/kernel/cpu/perfctr-watchdog.c b/arch/x86/kernel/cpu/perfctr-watchdog.c
index f9ae93adffe5..ddda4b64f545 100644
--- a/arch/x86/kernel/cpu/perfctr-watchdog.c
+++ b/arch/x86/kernel/cpu/perfctr-watchdog.c
@@ -181,7 +181,9 @@ void disable_lapic_nmi_watchdog(void)
 		return;
 
 	on_each_cpu(stop_apic_nmi_watchdog, NULL, 0, 1);
-	wd_ops->unreserve();
+
+	if (wd_ops)
+		wd_ops->unreserve();
 
 	BUG_ON(atomic_read(&nmi_active) != 0);
 }
diff --git a/arch/x86/kernel/nmi_64.c b/arch/x86/kernel/nmi.c
index d62f3b66b529..9ebf71323c9a 100644
--- a/arch/x86/kernel/nmi_64.c
+++ b/arch/x86/kernel/nmi.c
@@ -6,10 +6,13 @@
  *  Fixes:
  *  Mikael Pettersson	: AMD K7 support for local APIC NMI watchdog.
  *  Mikael Pettersson	: Power Management for local APIC NMI watchdog.
+ *  Mikael Pettersson	: Pentium 4 support for local APIC NMI watchdog.
  *  Pavel Machek and
  *  Mikael Pettersson	: PM converted to driver model. Disable/enable API.
  */
 
+#include <asm/apic.h>
+
 #include <linux/nmi.h>
 #include <linux/mm.h>
 #include <linux/delay.h>
@@ -17,22 +20,26 @@
 #include <linux/module.h>
 #include <linux/sysdev.h>
 #include <linux/sysctl.h>
+#include <linux/percpu.h>
 #include <linux/kprobes.h>
 #include <linux/cpumask.h>
+#include <linux/kernel_stat.h>
 #include <linux/kdebug.h>
+#include <linux/smp.h>
 
 #include <asm/i8259.h>
 #include <asm/io_apic.h>
 #include <asm/smp.h>
 #include <asm/nmi.h>
 #include <asm/proto.h>
+#include <asm/timer.h>
+
 #include <asm/mce.h>
 
 #include <mach_traps.h>
 
 int unknown_nmi_panic;
 int nmi_watchdog_enabled;
-int panic_on_unrecovered_nmi;
 
 static cpumask_t backtrace_mask = CPU_MASK_NONE;
 
@@ -43,12 +50,47 @@ static cpumask_t backtrace_mask = CPU_MASK_NONE;
  *  0: the lapic NMI watchdog is disabled, but can be enabled
  */
 atomic_t nmi_active = ATOMIC_INIT(0);		/* oprofile uses this */
-static int panic_on_timeout;
+EXPORT_SYMBOL(nmi_active);
 
 unsigned int nmi_watchdog = NMI_DEFAULT;
-static unsigned int nmi_hz = HZ;
+EXPORT_SYMBOL(nmi_watchdog);
+
+static int panic_on_timeout;
 
+static unsigned int nmi_hz = HZ;
 static DEFINE_PER_CPU(short, wd_enabled);
+static int endflag __initdata;
+
+static inline unsigned int get_nmi_count(int cpu)
+{
+#ifdef CONFIG_X86_64
+	return cpu_pda(cpu)->__nmi_count;
+#else
+	return nmi_count(cpu);
+#endif
+}
+
+static inline int mce_in_progress(void)
+{
+#if defined(CONFIG_X86_64) && defined(CONFIG_X86_MCE)
+	return atomic_read(&mce_entry) > 0;
+#endif
+	return 0;
+}
+
+/*
+ * Take the local apic timer and PIT/HPET into account. We don't
+ * know which one is active, when we have highres/dyntick on
+ */
+static inline unsigned int get_timer_irqs(int cpu)
+{
+#ifdef CONFIG_X86_64
+	return read_pda(apic_timer_irqs) + read_pda(irq0_irqs);
+#else
+	return per_cpu(irq_stat, cpu).apic_timer_irqs +
+		per_cpu(irq_stat, cpu).irq0_irqs;
+#endif
+}
 
 /* Run after command line and cpu_init init, but before all other checks */
 void nmi_watchdog_default(void)
@@ -58,22 +100,23 @@ void nmi_watchdog_default(void)
 	nmi_watchdog = NMI_NONE;
 }
 
-static int endflag __initdata = 0;
-
 #ifdef CONFIG_SMP
-/* The performance counters used by NMI_LOCAL_APIC don't trigger when
+/*
+ * 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
  * CPUs during the test make them busy.
  */
 static __init void nmi_cpu_busy(void *data)
 {
 	local_irq_enable_in_hardirq();
-	/* Intentionally don't use cpu_relax here. This is
-	   to make sure that the performance counter really ticks,
-	   even if there is a simulator or similar that catches the
-	   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. */
+	/*
+	 * Intentionally don't use cpu_relax here. This is
+	 * to make sure that the performance counter really ticks,
+	 * even if there is a simulator or similar that catches the
+	 * 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)
 		mb();
 }
@@ -81,10 +124,10 @@ static __init void nmi_cpu_busy(void *data)
 
 int __init check_nmi_watchdog(void)
 {
-	int *prev_nmi_count;
+	unsigned int *prev_nmi_count;
 	int cpu;
 
-	if ((nmi_watchdog == NMI_NONE) || (nmi_watchdog == NMI_DISABLED))
+	if (nmi_watchdog == NMI_NONE || nmi_watchdog == NMI_DISABLED)
 		return 0;
 
 	if (!atomic_read(&nmi_active))
@@ -101,20 +144,20 @@ int __init check_nmi_watchdog(void)
 		smp_call_function(nmi_cpu_busy, (void *)&endflag, 0, 0);
 #endif
 
-	for (cpu = 0; cpu < nr_cpu_ids; cpu++)
-		prev_nmi_count[cpu] = cpu_pda(cpu)->__nmi_count;
+	for_each_possible_cpu(cpu)
+		prev_nmi_count[cpu] = get_nmi_count(cpu);
 	local_irq_enable();
-	mdelay((20*1000)/nmi_hz); // wait 20 ticks
+	mdelay((20 * 1000) / nmi_hz); /* wait 20 ticks */
 
 	for_each_online_cpu(cpu) {
 		if (!per_cpu(wd_enabled, cpu))
 			continue;
-		if (cpu_pda(cpu)->__nmi_count - prev_nmi_count[cpu] <= 5) {
+		if (get_nmi_count(cpu) - prev_nmi_count[cpu] <= 5) {
 			printk(KERN_WARNING "WARNING: CPU#%d: NMI "
-			       "appears to be stuck (%d->%d)!\n",
+				"appears to be stuck (%d->%d)!\n",
 				cpu,
 				prev_nmi_count[cpu],
-				cpu_pda(cpu)->__nmi_count);
+				get_nmi_count(cpu));
 			per_cpu(wd_enabled, cpu) = 0;
 			atomic_dec(&nmi_active);
 		}
@@ -127,8 +170,10 @@ int __init check_nmi_watchdog(void)
 	}
 	printk("OK.\n");
 
-	/* now that we know it works we can reduce NMI frequency to
-	   something more reasonable; makes a difference in some configs */
+	/*
+	 * now that we know it works we can reduce NMI frequency to
+	 * something more reasonable; makes a difference in some configs
+	 */
 	if (nmi_watchdog == NMI_LOCAL_APIC)
 		nmi_hz = lapic_adjust_nmi_hz(1);
 
@@ -137,7 +182,6 @@ int __init check_nmi_watchdog(void)
 error:
 	if (nmi_watchdog == NMI_IO_APIC && !timer_through_8259)
 		disable_8259A_irq(0);
-
 	return -1;
 }
 
@@ -145,7 +189,7 @@ static int __init setup_nmi_watchdog(char *str)
 {
 	int nmi;
 
-	if (!strncmp(str,"panic",5)) {
+	if (!strncmp(str, "panic", 5)) {
 		panic_on_timeout = 1;
 		str = strchr(str, ',');
 		if (!str)
@@ -155,15 +199,17 @@ static int __init setup_nmi_watchdog(char *str)
 
 	get_option(&str, &nmi);
 
-	if ((nmi >= NMI_INVALID) || (nmi < NMI_NONE))
+	if (nmi >= NMI_INVALID || nmi < NMI_NONE)
 		return 0;
 
 	nmi_watchdog = nmi;
 	return 1;
 }
-
 __setup("nmi_watchdog=", setup_nmi_watchdog);
 
+/*
+ * Suspend/resume support
+ */
 #ifdef CONFIG_PM
 
 static int nmi_pm_active; /* nmi_active before suspend */
@@ -202,7 +248,8 @@ static int __init init_lapic_nmi_sysfs(void)
 {
 	int error;
 
-	/* should really be a BUG_ON but b/c this is an
+	/*
+	 * should really be a BUG_ON but b/c this is an
 	 * init call, it just doesn't work.  -dcz
 	 */
 	if (nmi_watchdog != NMI_LOCAL_APIC)
@@ -216,6 +263,7 @@ static int __init init_lapic_nmi_sysfs(void)
 		error = sysdev_register(&device_lapic_nmi);
 	return error;
 }
+
 /* must come after the local APIC's device_initcall() */
 late_initcall(init_lapic_nmi_sysfs);
 
@@ -223,7 +271,7 @@ late_initcall(init_lapic_nmi_sysfs);
 
 static void __acpi_nmi_enable(void *__unused)
 {
-	apic_write(APIC_LVT0, APIC_DM_NMI);
+	apic_write_around(APIC_LVT0, APIC_DM_NMI);
 }
 
 /*
@@ -237,7 +285,7 @@ void acpi_nmi_enable(void)
 
 static void __acpi_nmi_disable(void *__unused)
 {
-	apic_write(APIC_LVT0, APIC_DM_NMI | APIC_LVT_MASKED);
+	apic_write_around(APIC_LVT0, APIC_DM_NMI | APIC_LVT_MASKED);
 }
 
 /*
@@ -256,11 +304,12 @@ void setup_apic_nmi_watchdog(void *unused)
 
 	/* cheap hack to support suspend/resume */
 	/* if cpu0 is not active neither should the other cpus */
-	if ((smp_processor_id() != 0) && (atomic_read(&nmi_active) <= 0))
+	if (smp_processor_id() != 0 && atomic_read(&nmi_active) <= 0)
 		return;
 
 	switch (nmi_watchdog) {
 	case NMI_LOCAL_APIC:
+		 /* enable it before to avoid race with handler */
 		__get_cpu_var(wd_enabled) = 1;
 		if (lapic_watchdog_init(nmi_hz) < 0) {
 			__get_cpu_var(wd_enabled) = 0;
@@ -276,9 +325,9 @@ void setup_apic_nmi_watchdog(void *unused)
 void stop_apic_nmi_watchdog(void *unused)
 {
 	/* only support LOCAL and IO APICs for now */
-	if ((nmi_watchdog != NMI_LOCAL_APIC) &&
-	    (nmi_watchdog != NMI_IO_APIC))
-	    	return;
+	if (nmi_watchdog != NMI_LOCAL_APIC &&
+	    nmi_watchdog != NMI_IO_APIC)
+		return;
 	if (__get_cpu_var(wd_enabled) == 0)
 		return;
 	if (nmi_watchdog == NMI_LOCAL_APIC)
@@ -294,6 +343,11 @@ void stop_apic_nmi_watchdog(void *unused)
  *
  * as these watchdog NMI IRQs are generated on every CPU, we only
  * have to check the current processor.
+ *
+ * since NMIs don't listen to _any_ locks, we have to be extremely
+ * careful not to rely on unsafe variables. The printk might lock
+ * up though, so we have to break up any console locks first ...
+ * [when there will be more tty-related locks, break them up here too!]
  */
 
 static DEFINE_PER_CPU(unsigned, last_irq_sum);
@@ -302,11 +356,12 @@ static DEFINE_PER_CPU(int, nmi_touch);
 
 void touch_nmi_watchdog(void)
 {
-	if (nmi_watchdog > 0) {
+	if (nmi_watchdog == NMI_LOCAL_APIC ||
+		nmi_watchdog == NMI_IO_APIC) {
 		unsigned cpu;
 
 		/*
- 		 * Tell other CPUs to reset their alert counters. We cannot
+		 * Tell other CPUs to reset their alert counters. We cannot
 		 * do it ourselves because the alert count increase is not
 		 * atomic.
 		 */
@@ -316,6 +371,9 @@ void touch_nmi_watchdog(void)
 		}
 	}
 
+	/*
+	 * Tickle the softlockup detector too:
+	 */
 	touch_softlockup_watchdog();
 }
 EXPORT_SYMBOL(touch_nmi_watchdog);
@@ -323,7 +381,12 @@ EXPORT_SYMBOL(touch_nmi_watchdog);
 notrace __kprobes int
 nmi_watchdog_tick(struct pt_regs *regs, unsigned reason)
 {
-	int sum;
+	/*
+	 * Since current_thread_info()-> is always on the stack, and we
+	 * always switch the stack NMI-atomically, it's safe to use
+	 * smp_processor_id().
+	 */
+	unsigned int sum;
 	int touched = 0;
 	int cpu = smp_processor_id();
 	int rc = 0;
@@ -335,7 +398,8 @@ nmi_watchdog_tick(struct pt_regs *regs, unsigned reason)
 		touched = 1;
 	}
 
-	sum = read_pda(apic_timer_irqs) + read_pda(irq0_irqs);
+	sum = get_timer_irqs(cpu);
+
 	if (__get_cpu_var(nmi_touch)) {
 		__get_cpu_var(nmi_touch) = 0;
 		touched = 1;
@@ -345,28 +409,29 @@ nmi_watchdog_tick(struct pt_regs *regs, unsigned reason)
 		static DEFINE_SPINLOCK(lock);	/* Serialise the printks */
 
 		spin_lock(&lock);
-		printk("NMI backtrace for cpu %d\n", cpu);
+		printk(KERN_WARNING "NMI backtrace for cpu %d\n", cpu);
 		dump_stack();
 		spin_unlock(&lock);
 		cpu_clear(cpu, backtrace_mask);
 	}
 
-#ifdef CONFIG_X86_MCE
-	/* Could check oops_in_progress here too, but it's safer
-	   not too */
-	if (atomic_read(&mce_entry) > 0)
+	/* Could check oops_in_progress here too, but it's safer not to */
+	if (mce_in_progress())
 		touched = 1;
-#endif
-	/* if the apic timer isn't firing, this cpu isn't doing much */
+
+	/* if the none of the timers isn't firing, this cpu isn't doing much */
 	if (!touched && __get_cpu_var(last_irq_sum) == sum) {
 		/*
 		 * Ayiee, looks like this CPU is stuck ...
 		 * wait a few IRQs (5 seconds) before doing the oops ...
 		 */
 		local_inc(&__get_cpu_var(alert_counter));
-		if (local_read(&__get_cpu_var(alert_counter)) == 5*nmi_hz)
-			die_nmi("NMI Watchdog detected LOCKUP on CPU %d\n", regs,
-				panic_on_timeout);
+		if (local_read(&__get_cpu_var(alert_counter)) == 5 * nmi_hz)
+			/*
+			 * die_nmi will return ONLY if NOTIFY_STOP happens..
+			 */
+			die_nmi("BUG: NMI Watchdog detected LOCKUP",
+				regs, panic_on_timeout);
 	} else {
 		__get_cpu_var(last_irq_sum) = sum;
 		local_set(&__get_cpu_var(alert_counter), 0);
@@ -380,7 +445,8 @@ nmi_watchdog_tick(struct pt_regs *regs, unsigned reason)
 		rc |= lapic_wd_event(nmi_hz);
 		break;
 	case NMI_IO_APIC:
-		/* don't know how to accurately check for this.
+		/*
+		 * don't know how to accurately check for this.
 		 * just assume it was a watchdog timer interrupt
 		 * This matches the old behaviour.
 		 */
@@ -390,30 +456,6 @@ nmi_watchdog_tick(struct pt_regs *regs, unsigned reason)
 	return rc;
 }
 
-static unsigned ignore_nmis;
-
-asmlinkage notrace __kprobes void
-do_nmi(struct pt_regs *regs, long error_code)
-{
-	nmi_enter();
-	add_pda(__nmi_count,1);
-	if (!ignore_nmis)
-		default_do_nmi(regs);
-	nmi_exit();
-}
-
-void stop_nmi(void)
-{
-	acpi_nmi_disable();
-	ignore_nmis++;
-}
-
-void restart_nmi(void)
-{
-	ignore_nmis--;
-	acpi_nmi_enable();
-}
-
 #ifdef CONFIG_SYSCTL
 
 static int unknown_nmi_panic_callback(struct pt_regs *regs, int cpu)
@@ -422,7 +464,7 @@ static int unknown_nmi_panic_callback(struct pt_regs *regs, int cpu)
 	char buf[64];
 
 	sprintf(buf, "NMI received for unknown reason %02x\n", reason);
-	die_nmi(buf, regs, 1);	/* Always panic here */
+	die_nmi(buf, regs, 1); /* Always panic here */
 	return 0;
 }
 
@@ -441,27 +483,37 @@ int proc_nmi_enabled(struct ctl_table *table, int write, struct file *file,
 		return 0;
 
 	if (atomic_read(&nmi_active) < 0 || nmi_watchdog == NMI_DISABLED) {
-		printk( KERN_WARNING "NMI watchdog is permanently disabled\n");
+		printk(KERN_WARNING
+			"NMI watchdog is permanently disabled\n");
 		return -EIO;
 	}
 
 	/* if nmi_watchdog is not set yet, then set it */
 	nmi_watchdog_default();
 
+#ifdef CONFIG_X86_32
+	if (nmi_watchdog == NMI_NONE) {
+		if (lapic_watchdog_ok())
+			nmi_watchdog = NMI_LOCAL_APIC;
+		else
+			nmi_watchdog = NMI_IO_APIC;
+	}
+#endif
+
 	if (nmi_watchdog == NMI_LOCAL_APIC) {
 		if (nmi_watchdog_enabled)
 			enable_lapic_nmi_watchdog();
 		else
 			disable_lapic_nmi_watchdog();
 	} else {
-		printk( KERN_WARNING
+		printk(KERN_WARNING
 			"NMI watchdog doesn't know what hardware to touch\n");
 		return -EIO;
 	}
 	return 0;
 }
 
-#endif
+#endif /* CONFIG_SYSCTL */
 
 int do_nmi_callback(struct pt_regs *regs, int cpu)
 {
@@ -484,6 +536,3 @@ void __trigger_all_cpu_backtrace(void)
 		mdelay(1);
 	}
 }
-
-EXPORT_SYMBOL(nmi_active);
-EXPORT_SYMBOL(nmi_watchdog);
diff --git a/arch/x86/kernel/nmi_32.c b/arch/x86/kernel/nmi_32.c
deleted file mode 100644
index 6580dae46277..000000000000
--- a/arch/x86/kernel/nmi_32.c
+++ /dev/null
@@ -1,476 +0,0 @@
-/*
- *  NMI watchdog support on APIC systems
- *
- *  Started by Ingo Molnar <mingo@redhat.com>
- *
- *  Fixes:
- *  Mikael Pettersson	: AMD K7 support for local APIC NMI watchdog.
- *  Mikael Pettersson	: Power Management for local APIC NMI watchdog.
- *  Mikael Pettersson	: Pentium 4 support for local APIC NMI watchdog.
- *  Pavel Machek and
- *  Mikael Pettersson	: PM converted to driver model. Disable/enable API.
- */
-
-#include <linux/delay.h>
-#include <linux/interrupt.h>
-#include <linux/module.h>
-#include <linux/nmi.h>
-#include <linux/sysdev.h>
-#include <linux/sysctl.h>
-#include <linux/percpu.h>
-#include <linux/kprobes.h>
-#include <linux/cpumask.h>
-#include <linux/kernel_stat.h>
-#include <linux/kdebug.h>
-#include <linux/slab.h>
-
-#include <asm/i8259.h>
-#include <asm/io_apic.h>
-#include <asm/smp.h>
-#include <asm/nmi.h>
-#include <asm/timer.h>
-
-#include "mach_traps.h"
-
-int unknown_nmi_panic;
-int nmi_watchdog_enabled;
-
-static cpumask_t backtrace_mask = CPU_MASK_NONE;
-
-/* nmi_active:
- * >0: the lapic NMI watchdog is active, but can be disabled
- * <0: the lapic NMI watchdog has not been set up, and cannot
- *     be enabled
- *  0: the lapic NMI watchdog is disabled, but can be enabled
- */
-atomic_t nmi_active = ATOMIC_INIT(0);		/* oprofile uses this */
-
-unsigned int nmi_watchdog = NMI_DEFAULT;
-static unsigned int nmi_hz = HZ;
-
-static DEFINE_PER_CPU(short, wd_enabled);
-
-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
- * CPUs during the test make them busy.
- */
-static __init void nmi_cpu_busy(void *data)
-{
-	local_irq_enable_in_hardirq();
-	/* Intentionally don't use cpu_relax here. This is
-	   to make sure that the performance counter really ticks,
-	   even if there is a simulator or similar that catches the
-	   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)
-		mb();
-}
-#endif
-
-int __init check_nmi_watchdog(void)
-{
-	unsigned int *prev_nmi_count;
-	int cpu;
-
-	if ((nmi_watchdog == NMI_NONE) || (nmi_watchdog == NMI_DISABLED))
-		return 0;
-
-	if (!atomic_read(&nmi_active))
-		return 0;
-
-	prev_nmi_count = kmalloc(NR_CPUS * sizeof(int), GFP_KERNEL);
-	if (!prev_nmi_count)
-		goto error;
-
-	printk(KERN_INFO "Testing NMI watchdog ... ");
-
-#ifdef CONFIG_SMP
-	if (nmi_watchdog == NMI_LOCAL_APIC)
-		smp_call_function(nmi_cpu_busy, (void *)&endflag, 0, 0);
-#endif
-
-	for_each_possible_cpu(cpu)
-		prev_nmi_count[cpu] = nmi_count(cpu);
-	local_irq_enable();
-	mdelay((20*1000)/nmi_hz); // wait 20 ticks
-
-	for_each_possible_cpu(cpu) {
-#ifdef CONFIG_SMP
-		/* Check cpu_callin_map here because that is set
-		   after the timer is started. */
-		if (!cpu_isset(cpu, cpu_callin_map))
-			continue;
-#endif
-		if (!per_cpu(wd_enabled, cpu))
-			continue;
-		if (nmi_count(cpu) - prev_nmi_count[cpu] <= 5) {
-			printk(KERN_WARNING "WARNING: CPU#%d: NMI "
-				"appears to be stuck (%d->%d)!\n",
-				cpu,
-				prev_nmi_count[cpu],
-				nmi_count(cpu));
-			per_cpu(wd_enabled, cpu) = 0;
-			atomic_dec(&nmi_active);
-		}
-	}
-	endflag = 1;
-	if (!atomic_read(&nmi_active)) {
-		kfree(prev_nmi_count);
-		atomic_set(&nmi_active, -1);
-		goto error;
-	}
-	printk("OK.\n");
-
-	/* now that we know it works we can reduce NMI frequency to
-	   something more reasonable; makes a difference in some configs */
-	if (nmi_watchdog == NMI_LOCAL_APIC)
-		nmi_hz = lapic_adjust_nmi_hz(1);
-
-	kfree(prev_nmi_count);
-	return 0;
-error:
-	if (nmi_watchdog == NMI_IO_APIC && !timer_through_8259)
-		disable_8259A_irq(0);
-	timer_ack = 0;
-
-	return -1;
-}
-
-static int __init setup_nmi_watchdog(char *str)
-{
-	int nmi;
-
-	get_option(&str, &nmi);
-
-	if ((nmi >= NMI_INVALID) || (nmi < NMI_NONE))
-		return 0;
-
-	nmi_watchdog = nmi;
-	return 1;
-}
-
-__setup("nmi_watchdog=", setup_nmi_watchdog);
-
-
-/* Suspend/resume support */
-
-#ifdef CONFIG_PM
-
-static int nmi_pm_active; /* nmi_active before suspend */
-
-static int lapic_nmi_suspend(struct sys_device *dev, pm_message_t state)
-{
-	/* only CPU0 goes here, other CPUs should be offline */
-	nmi_pm_active = atomic_read(&nmi_active);
-	stop_apic_nmi_watchdog(NULL);
-	BUG_ON(atomic_read(&nmi_active) != 0);
-	return 0;
-}
-
-static int lapic_nmi_resume(struct sys_device *dev)
-{
-	/* only CPU0 goes here, other CPUs should be offline */
-	if (nmi_pm_active > 0) {
-		setup_apic_nmi_watchdog(NULL);
-		touch_nmi_watchdog();
-	}
-	return 0;
-}
-
-
-static struct sysdev_class nmi_sysclass = {
-	.name		= "lapic_nmi",
-	.resume		= lapic_nmi_resume,
-	.suspend	= lapic_nmi_suspend,
-};
-
-static struct sys_device device_lapic_nmi = {
-	.id	= 0,
-	.cls	= &nmi_sysclass,
-};
-
-static int __init init_lapic_nmi_sysfs(void)
-{
-	int error;
-
-	/* should really be a BUG_ON but b/c this is an
-	 * init call, it just doesn't work.  -dcz
-	 */
-	if (nmi_watchdog != NMI_LOCAL_APIC)
-		return 0;
-
-	if (atomic_read(&nmi_active) < 0)
-		return 0;
-
-	error = sysdev_class_register(&nmi_sysclass);
-	if (!error)
-		error = sysdev_register(&device_lapic_nmi);
-	return error;
-}
-/* must come after the local APIC's device_initcall() */
-late_initcall(init_lapic_nmi_sysfs);
-
-#endif	/* CONFIG_PM */
-
-static void __acpi_nmi_enable(void *__unused)
-{
-	apic_write_around(APIC_LVT0, APIC_DM_NMI);
-}
-
-/*
- * Enable timer based NMIs on all CPUs:
- */
-void acpi_nmi_enable(void)
-{
-	if (atomic_read(&nmi_active) && nmi_watchdog == NMI_IO_APIC)
-		on_each_cpu(__acpi_nmi_enable, NULL, 0, 1);
-}
-
-static void __acpi_nmi_disable(void *__unused)
-{
-	apic_write(APIC_LVT0, APIC_DM_NMI | APIC_LVT_MASKED);
-}
-
-/*
- * Disable timer based NMIs on all CPUs:
- */
-void acpi_nmi_disable(void)
-{
-	if (atomic_read(&nmi_active) && nmi_watchdog == NMI_IO_APIC)
-		on_each_cpu(__acpi_nmi_disable, NULL, 0, 1);
-}
-
-void setup_apic_nmi_watchdog(void *unused)
-{
-	if (__get_cpu_var(wd_enabled))
-		return;
-
-	/* cheap hack to support suspend/resume */
-	/* if cpu0 is not active neither should the other cpus */
-	if ((smp_processor_id() != 0) && (atomic_read(&nmi_active) <= 0))
-		return;
-
-	switch (nmi_watchdog) {
-	case NMI_LOCAL_APIC:
-		__get_cpu_var(wd_enabled) = 1; /* enable it before to avoid race with handler */
-		if (lapic_watchdog_init(nmi_hz) < 0) {
-			__get_cpu_var(wd_enabled) = 0;
-			return;
-		}
-		/* FALL THROUGH */
-	case NMI_IO_APIC:
-		__get_cpu_var(wd_enabled) = 1;
-		atomic_inc(&nmi_active);
-	}
-}
-
-void stop_apic_nmi_watchdog(void *unused)
-{
-	/* only support LOCAL and IO APICs for now */
-	if ((nmi_watchdog != NMI_LOCAL_APIC) &&
-	    (nmi_watchdog != NMI_IO_APIC))
-	    	return;
-	if (__get_cpu_var(wd_enabled) == 0)
-		return;
-	if (nmi_watchdog == NMI_LOCAL_APIC)
-		lapic_watchdog_stop();
-	__get_cpu_var(wd_enabled) = 0;
-	atomic_dec(&nmi_active);
-}
-
-/*
- * the best way to detect whether a CPU has a 'hard lockup' problem
- * is to check it's local APIC timer IRQ counts. If they are not
- * changing then that CPU has some problem.
- *
- * as these watchdog NMI IRQs are generated on every CPU, we only
- * have to check the current processor.
- *
- * since NMIs don't listen to _any_ locks, we have to be extremely
- * careful not to rely on unsafe variables. The printk might lock
- * up though, so we have to break up any console locks first ...
- * [when there will be more tty-related locks, break them up
- *  here too!]
- */
-
-static unsigned int
-	last_irq_sums [NR_CPUS],
-	alert_counter [NR_CPUS];
-
-void touch_nmi_watchdog(void)
-{
-	if (nmi_watchdog > 0) {
-		unsigned cpu;
-
-		/*
-		 * Just reset the alert counters, (other CPUs might be
-		 * spinning on locks we hold):
-		 */
-		for_each_present_cpu(cpu) {
-			if (alert_counter[cpu])
-				alert_counter[cpu] = 0;
-		}
-	}
-
-	/*
-	 * Tickle the softlockup detector too:
-	 */
-	touch_softlockup_watchdog();
-}
-EXPORT_SYMBOL(touch_nmi_watchdog);
-
-extern void die_nmi(struct pt_regs *, const char *msg);
-
-notrace __kprobes int
-nmi_watchdog_tick(struct pt_regs *regs, unsigned reason)
-{
-
-	/*
-	 * Since current_thread_info()-> is always on the stack, and we
-	 * always switch the stack NMI-atomically, it's safe to use
-	 * smp_processor_id().
-	 */
-	unsigned int sum;
-	int touched = 0;
-	int cpu = smp_processor_id();
-	int rc = 0;
-
-	/* check for other users first */
-	if (notify_die(DIE_NMI, "nmi", regs, reason, 2, SIGINT)
-			== NOTIFY_STOP) {
-		rc = 1;
-		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);
-	}
-
-	/*
-	 * Take the local apic timer and PIT/HPET into account. We don't
-	 * know which one is active, when we have highres/dyntick on
-	 */
-	sum = per_cpu(irq_stat, cpu).apic_timer_irqs +
-		per_cpu(irq_stat, cpu).irq0_irqs;
-
-	/* if the none of the timers isn't firing, this cpu isn't doing much */
-	if (!touched && last_irq_sums[cpu] == sum) {
-		/*
-		 * Ayiee, looks like this CPU is stuck ...
-		 * wait a few IRQs (5 seconds) before doing the oops ...
-		 */
-		alert_counter[cpu]++;
-		if (alert_counter[cpu] == 5*nmi_hz)
-			/*
-			 * die_nmi will return ONLY if NOTIFY_STOP happens..
-			 */
-			die_nmi(regs, "BUG: NMI Watchdog detected LOCKUP");
-	} else {
-		last_irq_sums[cpu] = sum;
-		alert_counter[cpu] = 0;
-	}
-	/* see if the nmi watchdog went off */
-	if (!__get_cpu_var(wd_enabled))
-		return rc;
-	switch (nmi_watchdog) {
-	case NMI_LOCAL_APIC:
-		rc |= lapic_wd_event(nmi_hz);
-		break;
-	case NMI_IO_APIC:
-		/* don't know how to accurately check for this.
-		 * just assume it was a watchdog timer interrupt
-		 * This matches the old behaviour.
-		 */
-		rc = 1;
-		break;
-	}
-	return rc;
-}
-
-#ifdef CONFIG_SYSCTL
-
-static int unknown_nmi_panic_callback(struct pt_regs *regs, int cpu)
-{
-	unsigned char reason = get_nmi_reason();
-	char buf[64];
-
-	sprintf(buf, "NMI received for unknown reason %02x\n", reason);
-	die_nmi(regs, buf);
-	return 0;
-}
-
-/*
- * proc handler for /proc/sys/kernel/nmi
- */
-int proc_nmi_enabled(struct ctl_table *table, int write, struct file *file,
-			void __user *buffer, size_t *length, loff_t *ppos)
-{
-	int old_state;
-
-	nmi_watchdog_enabled = (atomic_read(&nmi_active) > 0) ? 1 : 0;
-	old_state = nmi_watchdog_enabled;
-	proc_dointvec(table, write, file, buffer, length, ppos);
-	if (!!old_state == !!nmi_watchdog_enabled)
-		return 0;
-
-	if (atomic_read(&nmi_active) < 0 || nmi_watchdog == NMI_DISABLED) {
-		printk( KERN_WARNING "NMI watchdog is permanently disabled\n");
-		return -EIO;
-	}
-
-	if (nmi_watchdog == NMI_DEFAULT) {
-		if (lapic_watchdog_ok())
-			nmi_watchdog = NMI_LOCAL_APIC;
-		else
-			nmi_watchdog = NMI_IO_APIC;
-	}
-
-	if (nmi_watchdog == NMI_LOCAL_APIC) {
-		if (nmi_watchdog_enabled)
-			enable_lapic_nmi_watchdog();
-		else
-			disable_lapic_nmi_watchdog();
-	} else {
-		printk( KERN_WARNING
-			"NMI watchdog doesn't know what hardware to touch\n");
-		return -EIO;
-	}
-	return 0;
-}
-
-#endif
-
-int do_nmi_callback(struct pt_regs *regs, int cpu)
-{
-#ifdef CONFIG_SYSCTL
-	if (unknown_nmi_panic)
-		return unknown_nmi_panic_callback(regs, cpu);
-#endif
-	return 0;
-}
-
-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);
diff --git a/arch/x86/kernel/traps_32.c b/arch/x86/kernel/traps_32.c
index 08d752de4eee..cf37d20b1ba7 100644
--- a/arch/x86/kernel/traps_32.c
+++ b/arch/x86/kernel/traps_32.c
@@ -756,9 +756,9 @@ unknown_nmi_error(unsigned char reason, struct pt_regs *regs)
 
 static DEFINE_SPINLOCK(nmi_print_lock);
 
-void notrace __kprobes die_nmi(struct pt_regs *regs, const char *msg)
+void notrace __kprobes die_nmi(char *str, struct pt_regs *regs, int do_panic)
 {
-	if (notify_die(DIE_NMIWATCHDOG, msg, regs, 0, 2, SIGINT) == NOTIFY_STOP)
+	if (notify_die(DIE_NMIWATCHDOG, str, regs, 0, 2, SIGINT) == NOTIFY_STOP)
 		return;
 
 	spin_lock(&nmi_print_lock);
@@ -767,10 +767,12 @@ void notrace __kprobes die_nmi(struct pt_regs *regs, const char *msg)
 	* to get a message out:
 	*/
 	bust_spinlocks(1);
-	printk(KERN_EMERG "%s", msg);
+	printk(KERN_EMERG "%s", str);
 	printk(" on CPU%d, ip %08lx, registers:\n",
 		smp_processor_id(), regs->ip);
 	show_registers(regs);
+	if (do_panic)
+		panic("Non maskable interrupt");
 	console_silent();
 	spin_unlock(&nmi_print_lock);
 	bust_spinlocks(0);
diff --git a/arch/x86/kernel/traps_64.c b/arch/x86/kernel/traps_64.c
index ec6d3b2130c4..80ba6d37bfe0 100644
--- a/arch/x86/kernel/traps_64.c
+++ b/arch/x86/kernel/traps_64.c
@@ -75,7 +75,9 @@ asmlinkage void alignment_check(void);
 asmlinkage void machine_check(void);
 asmlinkage void spurious_interrupt_bug(void);
 
+int panic_on_unrecovered_nmi;
 static unsigned int code_bytes = 64;
+static unsigned ignore_nmis;
 
 static inline void conditional_sti(struct pt_regs *regs)
 {
@@ -613,7 +615,9 @@ die_nmi(char *str, struct pt_regs *regs, int do_panic)
 	 * We are in trouble anyway, lets at least try
 	 * to get a message out.
 	 */
-	printk(str, smp_processor_id());
+	printk(KERN_EMERG "%s", str);
+	printk(" on CPU%d, ip %08lx, registers:\n",
+		smp_processor_id(), regs->ip);
 	show_registers(regs);
 	if (kexec_should_crash(current))
 		crash_kexec(regs);
@@ -862,6 +866,28 @@ asmlinkage notrace  __kprobes void default_do_nmi(struct pt_regs *regs)
 		io_check_error(reason, regs);
 }
 
+asmlinkage notrace __kprobes void
+do_nmi(struct pt_regs *regs, long error_code)
+{
+	nmi_enter();
+	add_pda(__nmi_count, 1);
+	if (!ignore_nmis)
+		default_do_nmi(regs);
+	nmi_exit();
+}
+
+void stop_nmi(void)
+{
+	acpi_nmi_disable();
+	ignore_nmis++;
+}
+
+void restart_nmi(void)
+{
+	ignore_nmis--;
+	acpi_nmi_enable();
+}
+
 /* runs on IST stack. */
 asmlinkage void __kprobes do_int3(struct pt_regs * regs, long error_code)
 {