Unnecessary work and noise from mce code in suspend/resume path

When we suspend a laptop we offline all but one processor. But the mce code registers on a notify chain so it can clean up some sysfs entries. Part of that code calls device_unregister() which will fire kobject_uevent() which might wake up some user code that is watching for such things.

Patch below from Aiden Park works around the issue by avoiding the device_unregister()/device_register() by just keeping track of the original device registrations.

1) Is there a better way to avoid the kobject_uevent()?
2) What user code actually uses all these sysfs files? Some test code in mcelog(8) uses:
/sys/devices/system/machinecheck/machinecheck0/trigger
Some code in mce-test package adjusts
/sys/devices/system/machinecheck/machinecheck0/tolerant
But neither of these make use of all the per-cpu instances created/destroyed on every suspend/resume (or logical processor offline). Should we just move these out of per-cpu directories and scrap this whole block of code?

3) checkpatch isn't happy about use of NR_CPUS I suspect the usage here warrants for_each_possible_cpu()

index 68317c80de7f..e49302f24ce5

arch/x86/kernel/cpu/mcheck/mce.c +2 -0

@@ -113,6 +113,8 @@ static DEFINE_PER_CPU(struct work_struct, mce_work);
 
 static void (*quirk_no_way_out)(int bank, struct mce *m, struct pt_regs *regs);
 
+static struct device *mce_devices[NR_CPUS];
+
 /*
  * CPU/chipset specific EDAC code can register a notifier call here to print
  * MCE errors in a human-readable form.

arch/x86/kernel/cpu/mcheck/mce.c +7 -1

+		dev = mce_devices[i];
+		if (dev) {
+			device_unregister(dev);
+			mce_devices[i] = NULL;
+		}
+	}
 }
 
 /*

@@ -2272,23 +2285,27 @@ static void mce_device_release(struct device *dev)
 static int mce_device_create(unsigned int cpu)
 {
 	struct device *dev;
-	int err;
+	int err = 0;
 	int i, j;
 
 	if (!mce_available(&boot_cpu_data))
 		return -EIO;

arch/x86/kernel/cpu/mcheck/mce.c +5 -0

+		if (!dev)
+			return -ENOMEM;
+		dev->id  = cpu;
+		dev->bus = &mce_subsys;
+		dev->release = &mce_device_release;
 

Quoted 5 lines

When we suspend a laptop we offline all but one processor. But the mce code registers on a notify chain so it can clean up some sysfs entries. Part of that code calls device_unregister() which will fire kobject_uevent() which might wake up some user code that is watching for such things.

The issue being?

It is not clear from the text what actually is the problem.