Commit 8735728e authored by Rafael J. Wysocki's avatar Rafael J. Wysocki Committed by Ingo Molnar
Browse files

x86 MCE: Fix CPU hotplug problem with multiple multicore AMD CPUs 



During CPU hot-remove the sysfs directory created by
threshold_create_bank(), defined in
arch/x86/kernel/cpu/mcheck/mce_amd_64.c, has to be removed before
its parent directory, created by mce_create_device(), defined in
arch/x86/kernel/cpu/mcheck/mce_64.c .  Moreover, when the CPU in
question is hotplugged again, obviously the latter has to be created
before the former.  At present, the right ordering is not enforced,
because all of these operations are carried out by CPU hotplug
notifiers which are not appropriately ordered with respect to each
other.  This leads to serious problems on systems with two or more
multicore AMD CPUs, among other things during suspend and hibernation.

Fix the problem by placing threshold bank CPU hotplug callbacks in
mce_cpu_callback(), so that they are invoked at the right places,
if defined.  Additionally, use kobject_del() to remove the sysfs
directory associated with the kobject created by
kobject_create_and_add() in threshold_create_bank(), to prevent the
kernel from crashing during CPU hotplug operations on systems with
two or more multicore AMD CPUs.

This patch fixes bug #11337.

Signed-off-by: default avatarRafael J. Wysocki <rjw@sisk.pl>
Acked-by: default avatarAndi Kleen <andi@firstfloor.org>
Tested-by: default avatarMark Langsdorf <mark.langsdorf@amd.com>
Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
parent 91ede005

arch/x86/kernel/cpu/mcheck/mce_64.c

+5 −0
Original line number Diff line number Diff line
@@ -759,6 +759,7 @@ static struct sysdev_class mce_sysclass = { 
};



DEFINE_PER_CPU(struct sys_device, device_mce);

void (*threshold_cpu_callback)(unsigned long action, unsigned int cpu) __cpuinitdata;



/* Why are there no generic functions for this? */

#define ACCESSOR(name, var, start) \
@@ -883,9 +884,13 @@ static int __cpuinit mce_cpu_callback(struct notifier_block *nfb, 
	case CPU_ONLINE:

	case CPU_ONLINE_FROZEN:

		mce_create_device(cpu);

		if (threshold_cpu_callback)

			threshold_cpu_callback(action, cpu);

		break;

	case CPU_DEAD:

	case CPU_DEAD_FROZEN:

		if (threshold_cpu_callback)

			threshold_cpu_callback(action, cpu);

		mce_remove_device(cpu);

		break;

	}

arch/x86/kernel/cpu/mcheck/mce_amd_64.c

+5 −13
Original line number Diff line number Diff line
@@ -628,6 +628,7 @@ static void threshold_remove_bank(unsigned int cpu, int bank) 
	deallocate_threshold_block(cpu, bank);



free_out:

	kobject_del(b->kobj);

	kobject_put(b->kobj);

	kfree(b);

	per_cpu(threshold_banks, cpu)[bank] = NULL;
@@ -645,14 +646,11 @@ static void threshold_remove_device(unsigned int cpu) 
}



/* get notified when a cpu comes on/off */

static int __cpuinit threshold_cpu_callback(struct notifier_block *nfb,

					    unsigned long action, void *hcpu)

static void __cpuinit amd_64_threshold_cpu_callback(unsigned long action,

						     unsigned int cpu)

{

	/* cpu was unsigned int to begin with */

	unsigned int cpu = (unsigned long)hcpu;



	if (cpu >= NR_CPUS)

		goto out;

		return;



	switch (action) {

	case CPU_ONLINE:
@@ -666,14 +664,8 @@ static int __cpuinit threshold_cpu_callback(struct notifier_block *nfb, 
	default:

		break;

	}

      out:

	return NOTIFY_OK;

}



static struct notifier_block threshold_cpu_notifier __cpuinitdata = {

	.notifier_call = threshold_cpu_callback,

};



static __init int threshold_init_device(void)

{

	unsigned lcpu = 0;
@@ -684,7 +676,7 @@ static __init int threshold_init_device(void) 
		if (err)

			return err;

	}

	register_hotcpu_notifier(&threshold_cpu_notifier);

	threshold_cpu_callback = amd_64_threshold_cpu_callback;

	return 0;

}

include/asm-x86/mce.h

+1 −0
Original line number Diff line number Diff line
@@ -92,6 +92,7 @@ extern int mce_disabled; 


void mce_log(struct mce *m);

DECLARE_PER_CPU(struct sys_device, device_mce);

extern void (*threshold_cpu_callback)(unsigned long action, unsigned int cpu);



#ifdef CONFIG_X86_MCE_INTEL

void mce_intel_feature_init(struct cpuinfo_x86 *c);

CRA Git | Maintained and supported by SUSTech CRA and CCSE