Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit f2802e7f authored by Don Zickus's avatar Don Zickus Committed by Andi Kleen
Browse files

[PATCH] Add SMP support on x86_64 to reservation framework



This patch includes the changes to make the nmi watchdog on x86_64 SMP
aware.  A bunch of code was moved around to make it simpler to read.  In
addition, it is now possible to determine if a particular NMI was the result
of the watchdog or not.  This feature allows the kernel to filter out
unknown NMIs easier.

Signed-off-by: default avatarDon Zickus <dzickus@redhat.com>
Signed-off-by: default avatarAndi Kleen <ak@suse.de>
parent cb9c448c
Loading
Loading
Loading
Loading
+1 −2
Original line number Original line Diff line number Diff line
@@ -479,8 +479,7 @@ void __cpuinit setup_local_APIC (void)
	}
	}


	nmi_watchdog_default();
	nmi_watchdog_default();
	if (nmi_watchdog == NMI_LOCAL_APIC)
	setup_apic_nmi_watchdog(NULL);
		setup_apic_nmi_watchdog();
	apic_pm_activate();
	apic_pm_activate();
}
}


+287 −139
Original line number Original line Diff line number Diff line
@@ -56,53 +56,29 @@ static unsigned int lapic_nmi_owner;
#define LAPIC_NMI_RESERVED	(1<<1)
#define LAPIC_NMI_RESERVED	(1<<1)


/* nmi_active:
/* nmi_active:
 * +1: the lapic NMI watchdog is active, but can be disabled
 * >0: the lapic NMI watchdog is active, but can be disabled
 *  0: the lapic NMI watchdog has not been set up, and cannot
 * <0: the lapic NMI watchdog has not been set up, and cannot
 *     be enabled
 *     be enabled
 * -1: the lapic NMI watchdog is disabled, but can be enabled
 *  0: the lapic NMI watchdog is disabled, but can be enabled
 */
 */
int nmi_active;		/* oprofile uses this */
atomic_t nmi_active = ATOMIC_INIT(0);		/* oprofile uses this */
int panic_on_timeout;
int panic_on_timeout;


unsigned int nmi_watchdog = NMI_DEFAULT;
unsigned int nmi_watchdog = NMI_DEFAULT;
static unsigned int nmi_hz = HZ;
static unsigned int nmi_hz = HZ;
static unsigned int nmi_perfctr_msr;	/* the MSR to reset in NMI handler */
static unsigned int nmi_p4_cccr_val;


/* Note that these events don't tick when the CPU idles. This means
struct nmi_watchdog_ctlblk {
   the frequency varies with CPU load. */
	int enabled;
	u64 check_bit;
	unsigned int cccr_msr;
	unsigned int perfctr_msr;  /* the MSR to reset in NMI handler */
	unsigned int evntsel_msr;  /* the MSR to select the events to handle */
};
static DEFINE_PER_CPU(struct nmi_watchdog_ctlblk, nmi_watchdog_ctlblk);


#define K7_EVNTSEL_ENABLE	(1 << 22)
/* local prototypes */
#define K7_EVNTSEL_INT		(1 << 20)
static void stop_apic_nmi_watchdog(void *unused);
#define K7_EVNTSEL_OS		(1 << 17)
static int unknown_nmi_panic_callback(struct pt_regs *regs, int cpu);
#define K7_EVNTSEL_USR		(1 << 16)
#define K7_EVENT_CYCLES_PROCESSOR_IS_RUNNING	0x76
#define K7_NMI_EVENT		K7_EVENT_CYCLES_PROCESSOR_IS_RUNNING

#define MSR_P4_MISC_ENABLE	0x1A0
#define MSR_P4_MISC_ENABLE_PERF_AVAIL	(1<<7)
#define MSR_P4_MISC_ENABLE_PEBS_UNAVAIL	(1<<12)
#define MSR_P4_PERFCTR0		0x300
#define MSR_P4_CCCR0		0x360
#define P4_ESCR_EVENT_SELECT(N)	((N)<<25)
#define P4_ESCR_OS		(1<<3)
#define P4_ESCR_USR		(1<<2)
#define P4_CCCR_OVF_PMI0	(1<<26)
#define P4_CCCR_OVF_PMI1	(1<<27)
#define P4_CCCR_THRESHOLD(N)	((N)<<20)
#define P4_CCCR_COMPLEMENT	(1<<19)
#define P4_CCCR_COMPARE		(1<<18)
#define P4_CCCR_REQUIRED	(3<<16)
#define P4_CCCR_ESCR_SELECT(N)	((N)<<13)
#define P4_CCCR_ENABLE		(1<<12)
/* Set up IQ_COUNTER0 to behave like a clock, by having IQ_CCCR0 filter
   CRU_ESCR0 (with any non-null event selector) through a complemented
   max threshold. [IA32-Vol3, Section 14.9.9] */
#define MSR_P4_IQ_COUNTER0	0x30C
#define P4_NMI_CRU_ESCR0	(P4_ESCR_EVENT_SELECT(0x3F)|P4_ESCR_OS|P4_ESCR_USR)
#define P4_NMI_IQ_CCCR0	\
	(P4_CCCR_OVF_PMI0|P4_CCCR_THRESHOLD(15)|P4_CCCR_COMPLEMENT|	\
	 P4_CCCR_COMPARE|P4_CCCR_REQUIRED|P4_CCCR_ESCR_SELECT(4)|P4_CCCR_ENABLE)


/* converts an msr to an appropriate reservation bit */
/* converts an msr to an appropriate reservation bit */
static inline unsigned int nmi_perfctr_msr_to_bit(unsigned int msr)
static inline unsigned int nmi_perfctr_msr_to_bit(unsigned int msr)
@@ -241,6 +217,12 @@ int __init check_nmi_watchdog (void)
	int *counts;
	int *counts;
	int cpu;
	int cpu;


	if ((nmi_watchdog == NMI_NONE) || (nmi_watchdog == NMI_DEFAULT))
		return 0;

	if (!atomic_read(&nmi_active))
		return 0;

	counts = kmalloc(NR_CPUS * sizeof(int), GFP_KERNEL);
	counts = kmalloc(NR_CPUS * sizeof(int), GFP_KERNEL);
	if (!counts)
	if (!counts)
		return -1;
		return -1;
@@ -258,19 +240,22 @@ int __init check_nmi_watchdog (void)
	mdelay((10*1000)/nmi_hz); // wait 10 ticks
	mdelay((10*1000)/nmi_hz); // wait 10 ticks


	for_each_online_cpu(cpu) {
	for_each_online_cpu(cpu) {
		if (!per_cpu(nmi_watchdog_ctlblk, cpu).enabled)
			continue;
		if (cpu_pda(cpu)->__nmi_count - counts[cpu] <= 5) {
		if (cpu_pda(cpu)->__nmi_count - counts[cpu] <= 5) {
			endflag = 1;
			printk("CPU#%d: NMI appears to be stuck (%d->%d)!\n",
			printk("CPU#%d: NMI appears to be stuck (%d->%d)!\n",
			       cpu,
			       cpu,
			       counts[cpu],
			       counts[cpu],
			       cpu_pda(cpu)->__nmi_count);
			       cpu_pda(cpu)->__nmi_count);
			nmi_active = 0;
			per_cpu(nmi_watchdog_ctlblk, cpu).enabled = 0;
			lapic_nmi_owner &= ~LAPIC_NMI_WATCHDOG;
			atomic_dec(&nmi_active);
			nmi_perfctr_msr = 0;
		}
	}
	if (!atomic_read(&nmi_active)) {
		kfree(counts);
		kfree(counts);
		atomic_set(&nmi_active, -1);
		return -1;
		return -1;
	}
	}
	}
	endflag = 1;
	endflag = 1;
	printk("OK.\n");
	printk("OK.\n");


@@ -297,8 +282,11 @@ int __init setup_nmi_watchdog(char *str)


	get_option(&str, &nmi);
	get_option(&str, &nmi);


	if (nmi >= NMI_INVALID)
	if ((nmi >= NMI_INVALID) || (nmi < NMI_NONE))
		return 0;
		return 0;

	if ((nmi == NMI_LOCAL_APIC) && (nmi_known_cpu() == 0))
		return 0;  /* no lapic support */
	nmi_watchdog = nmi;
	nmi_watchdog = nmi;
	return 1;
	return 1;
}
}
@@ -307,31 +295,30 @@ __setup("nmi_watchdog=", setup_nmi_watchdog);


static void disable_lapic_nmi_watchdog(void)
static void disable_lapic_nmi_watchdog(void)
{
{
	if (nmi_active <= 0)
	BUG_ON(nmi_watchdog != NMI_LOCAL_APIC);

	if (atomic_read(&nmi_active) <= 0)
		return;
		return;
	switch (boot_cpu_data.x86_vendor) {

	case X86_VENDOR_AMD:
	on_each_cpu(stop_apic_nmi_watchdog, NULL, 0, 1);
		wrmsr(MSR_K7_EVNTSEL0, 0, 0);

		break;
	BUG_ON(atomic_read(&nmi_active) != 0);
	case X86_VENDOR_INTEL:
		if (boot_cpu_data.x86 == 15) {
			wrmsr(MSR_P4_IQ_CCCR0, 0, 0);
			wrmsr(MSR_P4_CRU_ESCR0, 0, 0);
		}
		break;
	}
	nmi_active = -1;
	/* tell do_nmi() and others that we're not active any more */
	nmi_watchdog = 0;
}
}


static void enable_lapic_nmi_watchdog(void)
static void enable_lapic_nmi_watchdog(void)
{
{
	if (nmi_active < 0) {
	BUG_ON(nmi_watchdog != NMI_LOCAL_APIC);
		nmi_watchdog = NMI_LOCAL_APIC;

	/* are we already enabled */
	if (atomic_read(&nmi_active) != 0)
		return;

	/* are we lapic aware */
	if (nmi_known_cpu() <= 0)
		return;

	on_each_cpu(setup_apic_nmi_watchdog, NULL, 0, 1);
	touch_nmi_watchdog();
	touch_nmi_watchdog();
		setup_apic_nmi_watchdog();
	}
}
}


int reserve_lapic_nmi(void)
int reserve_lapic_nmi(void)
@@ -363,21 +350,24 @@ void release_lapic_nmi(void)


void disable_timer_nmi_watchdog(void)
void disable_timer_nmi_watchdog(void)
{
{
	if ((nmi_watchdog != NMI_IO_APIC) || (nmi_active <= 0))
	BUG_ON(nmi_watchdog != NMI_IO_APIC);

	if (atomic_read(&nmi_active) <= 0)
		return;
		return;


	disable_irq(0);
	disable_irq(0);
	unset_nmi_callback();
	on_each_cpu(stop_apic_nmi_watchdog, NULL, 0, 1);
	nmi_active = -1;

	nmi_watchdog = NMI_NONE;
	BUG_ON(atomic_read(&nmi_active) != 0);
}
}


void enable_timer_nmi_watchdog(void)
void enable_timer_nmi_watchdog(void)
{
{
	if (nmi_active < 0) {
	BUG_ON(nmi_watchdog != NMI_IO_APIC);
		nmi_watchdog = NMI_IO_APIC;

	if (atomic_read(&nmi_active) == 0) {
		touch_nmi_watchdog();
		touch_nmi_watchdog();
		nmi_active = 1;
		on_each_cpu(setup_apic_nmi_watchdog, NULL, 0, 1);
		enable_irq(0);
		enable_irq(0);
	}
	}
}
}
@@ -388,7 +378,7 @@ static int nmi_pm_active; /* nmi_active before suspend */


static int lapic_nmi_suspend(struct sys_device *dev, pm_message_t state)
static int lapic_nmi_suspend(struct sys_device *dev, pm_message_t state)
{
{
	nmi_pm_active = nmi_active;
	nmi_pm_active = atomic_read(&nmi_active);
	disable_lapic_nmi_watchdog();
	disable_lapic_nmi_watchdog();
	return 0;
	return 0;
}
}
@@ -415,7 +405,13 @@ static int __init init_lapic_nmi_sysfs(void)
{
{
	int error;
	int error;


	if (nmi_active == 0 || nmi_watchdog != NMI_LOCAL_APIC)
	/* 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;
		return 0;


	error = sysdev_class_register(&nmi_sysclass);
	error = sysdev_class_register(&nmi_sysclass);
@@ -428,100 +424,232 @@ late_initcall(init_lapic_nmi_sysfs);


#endif	/* CONFIG_PM */
#endif	/* CONFIG_PM */


/*
 * Activate the NMI watchdog via the local APIC.
 * Original code written by Keith Owens.
 */

/* Note that these events don't tick when the CPU idles. This means
   the frequency varies with CPU load. */

#define K7_EVNTSEL_ENABLE	(1 << 22)
#define K7_EVNTSEL_INT		(1 << 20)
#define K7_EVNTSEL_OS		(1 << 17)
#define K7_EVNTSEL_USR		(1 << 16)
#define K7_EVENT_CYCLES_PROCESSOR_IS_RUNNING	0x76
#define K7_NMI_EVENT		K7_EVENT_CYCLES_PROCESSOR_IS_RUNNING

static int setup_k7_watchdog(void)
static int setup_k7_watchdog(void)
{
{
	unsigned int perfctr_msr, evntsel_msr;
	unsigned int evntsel;
	unsigned int evntsel;
	struct nmi_watchdog_ctlblk *wd = &__get_cpu_var(nmi_watchdog_ctlblk);


	nmi_perfctr_msr = MSR_K7_PERFCTR0;
	perfctr_msr = MSR_K7_PERFCTR0;

	evntsel_msr = MSR_K7_EVNTSEL0;
	if (!reserve_perfctr_nmi(nmi_perfctr_msr))
	if (!reserve_perfctr_nmi(perfctr_msr))
		goto fail;
		goto fail;


	if (!reserve_evntsel_nmi(MSR_K7_EVNTSEL0))
	if (!reserve_evntsel_nmi(evntsel_msr))
		goto fail1;
		goto fail1;


	/* Simulator may not support it */
	/* Simulator may not support it */
	if (checking_wrmsrl(MSR_K7_EVNTSEL0, 0UL))
	if (checking_wrmsrl(evntsel_msr, 0UL))
		goto fail2;
		goto fail2;
	wrmsrl(MSR_K7_PERFCTR0, 0UL);
	wrmsrl(perfctr_msr, 0UL);


	evntsel = K7_EVNTSEL_INT
	evntsel = K7_EVNTSEL_INT
		| K7_EVNTSEL_OS
		| K7_EVNTSEL_OS
		| K7_EVNTSEL_USR
		| K7_EVNTSEL_USR
		| K7_NMI_EVENT;
		| K7_NMI_EVENT;


	wrmsr(MSR_K7_EVNTSEL0, evntsel, 0);
	/* setup the timer */
	wrmsrl(MSR_K7_PERFCTR0, -((u64)cpu_khz * 1000 / nmi_hz));
	wrmsr(evntsel_msr, evntsel, 0);
	wrmsrl(perfctr_msr, -((u64)cpu_khz * 1000 / nmi_hz));
	apic_write(APIC_LVTPC, APIC_DM_NMI);
	apic_write(APIC_LVTPC, APIC_DM_NMI);
	evntsel |= K7_EVNTSEL_ENABLE;
	evntsel |= K7_EVNTSEL_ENABLE;
	wrmsr(MSR_K7_EVNTSEL0, evntsel, 0);
	wrmsr(evntsel_msr, evntsel, 0);

	wd->perfctr_msr = perfctr_msr;
	wd->evntsel_msr = evntsel_msr;
	wd->cccr_msr = 0;  //unused
	wd->check_bit = 1ULL<<63;
	return 1;
	return 1;
fail2:
fail2:
	release_evntsel_nmi(MSR_K7_EVNTSEL0);
	release_evntsel_nmi(evntsel_msr);
fail1:
fail1:
	release_perfctr_nmi(nmi_perfctr_msr);
	release_perfctr_nmi(perfctr_msr);
fail:
fail:
	return 0;
	return 0;
}
}


static void stop_k7_watchdog(void)
{
	struct nmi_watchdog_ctlblk *wd = &__get_cpu_var(nmi_watchdog_ctlblk);

	wrmsr(wd->evntsel_msr, 0, 0);

	release_evntsel_nmi(wd->evntsel_msr);
	release_perfctr_nmi(wd->perfctr_msr);
}

/* Note that these events don't tick when the CPU idles. This means
   the frequency varies with CPU load. */

#define MSR_P4_MISC_ENABLE_PERF_AVAIL	(1<<7)
#define P4_ESCR_EVENT_SELECT(N)	((N)<<25)
#define P4_ESCR_OS		(1<<3)
#define P4_ESCR_USR		(1<<2)
#define P4_CCCR_OVF_PMI0	(1<<26)
#define P4_CCCR_OVF_PMI1	(1<<27)
#define P4_CCCR_THRESHOLD(N)	((N)<<20)
#define P4_CCCR_COMPLEMENT	(1<<19)
#define P4_CCCR_COMPARE		(1<<18)
#define P4_CCCR_REQUIRED	(3<<16)
#define P4_CCCR_ESCR_SELECT(N)	((N)<<13)
#define P4_CCCR_ENABLE		(1<<12)
#define P4_CCCR_OVF 		(1<<31)
/* Set up IQ_COUNTER0 to behave like a clock, by having IQ_CCCR0 filter
   CRU_ESCR0 (with any non-null event selector) through a complemented
   max threshold. [IA32-Vol3, Section 14.9.9] */


static int setup_p4_watchdog(void)
static int setup_p4_watchdog(void)
{
{
	unsigned int perfctr_msr, evntsel_msr, cccr_msr;
	unsigned int evntsel, cccr_val;
	unsigned int misc_enable, dummy;
	unsigned int misc_enable, dummy;
	unsigned int ht_num;
	struct nmi_watchdog_ctlblk *wd = &__get_cpu_var(nmi_watchdog_ctlblk);


	rdmsr(MSR_P4_MISC_ENABLE, misc_enable, dummy);
	rdmsr(MSR_IA32_MISC_ENABLE, misc_enable, dummy);
	if (!(misc_enable & MSR_P4_MISC_ENABLE_PERF_AVAIL))
	if (!(misc_enable & MSR_P4_MISC_ENABLE_PERF_AVAIL))
		return 0;
		return 0;


	nmi_perfctr_msr = MSR_P4_IQ_COUNTER0;
	nmi_p4_cccr_val = P4_NMI_IQ_CCCR0;
#ifdef CONFIG_SMP
#ifdef CONFIG_SMP
	if (smp_num_siblings == 2)
	/* detect which hyperthread we are on */
		nmi_p4_cccr_val |= P4_CCCR_OVF_PMI1;
	if (smp_num_siblings == 2) {
		unsigned int ebx, apicid;

        	ebx = cpuid_ebx(1);
	        apicid = (ebx >> 24) & 0xff;
        	ht_num = apicid & 1;
	} else
#endif
#endif
		ht_num = 0;

	/* performance counters are shared resources
	 * assign each hyperthread its own set
	 * (re-use the ESCR0 register, seems safe
	 * and keeps the cccr_val the same)
	 */
	if (!ht_num) {
		/* logical cpu 0 */
		perfctr_msr = MSR_P4_IQ_PERFCTR0;
		evntsel_msr = MSR_P4_CRU_ESCR0;
		cccr_msr = MSR_P4_IQ_CCCR0;
		cccr_val = P4_CCCR_OVF_PMI0 | P4_CCCR_ESCR_SELECT(4);
	} else {
		/* logical cpu 1 */
		perfctr_msr = MSR_P4_IQ_PERFCTR1;
		evntsel_msr = MSR_P4_CRU_ESCR0;
		cccr_msr = MSR_P4_IQ_CCCR1;
		cccr_val = P4_CCCR_OVF_PMI1 | P4_CCCR_ESCR_SELECT(4);
	}


	if (!reserve_perfctr_nmi(nmi_perfctr_msr))
	if (!reserve_perfctr_nmi(perfctr_msr))
		goto fail;
		goto fail;


	if (!reserve_evntsel_nmi(MSR_P4_CRU_ESCR0))
	if (!reserve_evntsel_nmi(evntsel_msr))
		goto fail1;
		goto fail1;


	wrmsr(MSR_P4_CRU_ESCR0, P4_NMI_CRU_ESCR0, 0);
	evntsel = P4_ESCR_EVENT_SELECT(0x3F)
	wrmsr(MSR_P4_IQ_CCCR0, P4_NMI_IQ_CCCR0 & ~P4_CCCR_ENABLE, 0);
	 	| P4_ESCR_OS
	Dprintk("setting P4_IQ_COUNTER0 to 0x%08lx\n", -(cpu_khz * 1000UL / nmi_hz));
		| P4_ESCR_USR;
	wrmsrl(MSR_P4_IQ_COUNTER0, -((u64)cpu_khz * 1000 / nmi_hz));

	cccr_val |= P4_CCCR_THRESHOLD(15)
		 | P4_CCCR_COMPLEMENT
		 | P4_CCCR_COMPARE
		 | P4_CCCR_REQUIRED;

	wrmsr(evntsel_msr, evntsel, 0);
	wrmsr(cccr_msr, cccr_val, 0);
	wrmsrl(perfctr_msr, -((u64)cpu_khz * 1000 / nmi_hz));
	apic_write(APIC_LVTPC, APIC_DM_NMI);
	apic_write(APIC_LVTPC, APIC_DM_NMI);
	wrmsr(MSR_P4_IQ_CCCR0, nmi_p4_cccr_val, 0);
	cccr_val |= P4_CCCR_ENABLE;
	wrmsr(cccr_msr, cccr_val, 0);

	wd->perfctr_msr = perfctr_msr;
	wd->evntsel_msr = evntsel_msr;
	wd->cccr_msr = cccr_msr;
	wd->check_bit = 1ULL<<39;
	return 1;
	return 1;
fail1:
fail1:
	release_perfctr_nmi(nmi_perfctr_msr);
	release_perfctr_nmi(perfctr_msr);
fail:
fail:
	return 0;
	return 0;
}
}


void setup_apic_nmi_watchdog(void)
static void stop_p4_watchdog(void)
{
{
	struct nmi_watchdog_ctlblk *wd = &__get_cpu_var(nmi_watchdog_ctlblk);

	wrmsr(wd->cccr_msr, 0, 0);
	wrmsr(wd->evntsel_msr, 0, 0);

	release_evntsel_nmi(wd->evntsel_msr);
	release_perfctr_nmi(wd->perfctr_msr);
}

void setup_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) {
		switch (boot_cpu_data.x86_vendor) {
		switch (boot_cpu_data.x86_vendor) {
		case X86_VENDOR_AMD:
		case X86_VENDOR_AMD:
		if (boot_cpu_data.x86 != 15)
			return;
			if (strstr(boot_cpu_data.x86_model_id, "Screwdriver"))
			if (strstr(boot_cpu_data.x86_model_id, "Screwdriver"))
				return;
				return;
			if (!setup_k7_watchdog())
			if (!setup_k7_watchdog())
				return;
				return;
			break;
			break;
		case X86_VENDOR_INTEL:
		case X86_VENDOR_INTEL:
		if (boot_cpu_data.x86 != 15)
			return;
			if (!setup_p4_watchdog())
			if (!setup_p4_watchdog())
				return;
				return;
			break;
			break;
		default:
			return;
		}
	}
	__get_cpu_var(nmi_watchdog_ctlblk.enabled) = 1;
	atomic_inc(&nmi_active);
}

static 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) {
		switch (boot_cpu_data.x86_vendor) {
		case X86_VENDOR_AMD:
			if (strstr(boot_cpu_data.x86_model_id, "Screwdriver"))
				return;
			stop_k7_watchdog();
			break;
		case X86_VENDOR_INTEL:
			stop_p4_watchdog();
			break;
		default:
		default:
			return;
			return;
		}
		}
	lapic_nmi_owner = LAPIC_NMI_WATCHDOG;
	}
	nmi_active = 1;
	__get_cpu_var(nmi_watchdog_ctlblk.enabled) = 0;
	atomic_dec(&nmi_active);
}
}


/*
/*
@@ -558,38 +686,52 @@ void __kprobes nmi_watchdog_tick(struct pt_regs * regs, unsigned reason)
{
{
	int sum;
	int sum;
	int touched = 0;
	int touched = 0;
	struct nmi_watchdog_ctlblk *wd = &__get_cpu_var(nmi_watchdog_ctlblk);
	u64 dummy;

	/* check for other users first */
	if (notify_die(DIE_NMI, "nmi", regs, reason, 2, SIGINT)
			== NOTIFY_STOP) {
		touched = 1;
	}


	sum = read_pda(apic_timer_irqs);
	sum = read_pda(apic_timer_irqs);
	if (__get_cpu_var(nmi_touch)) {
	if (__get_cpu_var(nmi_touch)) {
		__get_cpu_var(nmi_touch) = 0;
		__get_cpu_var(nmi_touch) = 0;
		touched = 1;
		touched = 1;
	}
	}

#ifdef CONFIG_X86_MCE
#ifdef CONFIG_X86_MCE
	/* Could check oops_in_progress here too, but it's safer
	/* Could check oops_in_progress here too, but it's safer
	   not too */
	   not too */
	if (atomic_read(&mce_entry) > 0)
	if (atomic_read(&mce_entry) > 0)
		touched = 1;
		touched = 1;
#endif
#endif
	/* if the apic timer isn't firing, this cpu isn't doing much */
	if (!touched && __get_cpu_var(last_irq_sum) == sum) {
	if (!touched && __get_cpu_var(last_irq_sum) == sum) {
		/*
		/*
		 * Ayiee, looks like this CPU is stuck ...
		 * Ayiee, looks like this CPU is stuck ...
		 * wait a few IRQs (5 seconds) before doing the oops ...
		 * wait a few IRQs (5 seconds) before doing the oops ...
		 */
		 */
		local_inc(&__get_cpu_var(alert_counter));
		local_inc(&__get_cpu_var(alert_counter));
		if (local_read(&__get_cpu_var(alert_counter)) == 5*nmi_hz) {
		if (local_read(&__get_cpu_var(alert_counter)) == 5*nmi_hz)
			if (notify_die(DIE_NMI, "nmi", regs, reason, 2, SIGINT)
							== NOTIFY_STOP) {
				local_set(&__get_cpu_var(alert_counter), 0);
				return;
			}
			die_nmi("NMI Watchdog detected LOCKUP on CPU %d\n", regs);
			die_nmi("NMI Watchdog detected LOCKUP on CPU %d\n", regs);
		}
	} else {
	} else {
		__get_cpu_var(last_irq_sum) = sum;
		__get_cpu_var(last_irq_sum) = sum;
		local_set(&__get_cpu_var(alert_counter), 0);
		local_set(&__get_cpu_var(alert_counter), 0);
	}
	}
	if (nmi_perfctr_msr) {

 		if (nmi_perfctr_msr == MSR_P4_IQ_COUNTER0) {
	/* see if the nmi watchdog went off */
	if (wd->enabled) {
		if (nmi_watchdog == NMI_LOCAL_APIC) {
			rdmsrl(wd->perfctr_msr, dummy);
			if (dummy & wd->check_bit){
				/* this wasn't a watchdog timer interrupt */
				goto done;
			}

			/* only Intel uses the cccr msr */
	 		if (wd->cccr_msr != 0) {
	 			/*
	 			/*
	 			 * P4 quirks:
	 			 * P4 quirks:
	 			 * - An overflown perfctr will assert its interrupt
	 			 * - An overflown perfctr will assert its interrupt
@@ -597,12 +739,18 @@ void __kprobes nmi_watchdog_tick(struct pt_regs * regs, unsigned reason)
	 			 * - LVTPC is masked on interrupt and must be
	 			 * - LVTPC is masked on interrupt and must be
	 			 *   unmasked by the LVTPC handler.
	 			 *   unmasked by the LVTPC handler.
	 			 */
	 			 */
 			wrmsr(MSR_P4_IQ_CCCR0, nmi_p4_cccr_val, 0);
				rdmsrl(wd->cccr_msr, dummy);
				dummy &= ~P4_CCCR_OVF;
	 			wrmsrl(wd->cccr_msr, dummy);
	 			apic_write(APIC_LVTPC, APIC_DM_NMI);
	 			apic_write(APIC_LVTPC, APIC_DM_NMI);
	 		}
	 		}
		wrmsrl(nmi_perfctr_msr, -((u64)cpu_khz * 1000 / nmi_hz));
			/* start the cycle over again */
			wrmsrl(wd->perfctr_msr, -((u64)cpu_khz * 1000 / nmi_hz));
		}
		}
	}
	}
done:
	return;
}


static __kprobes int dummy_nmi_callback(struct pt_regs * regs, int cpu)
static __kprobes int dummy_nmi_callback(struct pt_regs * regs, int cpu)
{
{
+2 −1
Original line number Original line Diff line number Diff line
@@ -63,7 +63,7 @@ extern void release_perfctr_nmi(unsigned int);
extern int reserve_evntsel_nmi(unsigned int);
extern int reserve_evntsel_nmi(unsigned int);
extern void release_evntsel_nmi(unsigned int);
extern void release_evntsel_nmi(unsigned int);


extern void setup_apic_nmi_watchdog (void);
extern void setup_apic_nmi_watchdog (void *);
extern int reserve_lapic_nmi(void);
extern int reserve_lapic_nmi(void);
extern void release_lapic_nmi(void);
extern void release_lapic_nmi(void);
extern void disable_timer_nmi_watchdog(void);
extern void disable_timer_nmi_watchdog(void);
@@ -73,6 +73,7 @@ extern void nmi_watchdog_tick (struct pt_regs * regs, unsigned reason);
extern void nmi_watchdog_default(void);
extern void nmi_watchdog_default(void);
extern int setup_nmi_watchdog(char *);
extern int setup_nmi_watchdog(char *);


extern atomic_t nmi_active;
extern unsigned int nmi_watchdog;
extern unsigned int nmi_watchdog;
#define NMI_DEFAULT	-1
#define NMI_DEFAULT	-1
#define NMI_NONE	0
#define NMI_NONE	0