Re: _cpu_down deadlock [was Re: 2.6.19-rc1-mm1]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]


On Thursday October 12, [email protected] wrote:
> On Thu, 12 Oct 2006 17:53:11 +1000
> Neil Brown <[email protected]> wrote:
> > I think I'm in favour of the following. 
> Would be simpler to take cpu_add_remove_lock in
> [un]register_cpu_notifier().  I actually thought I'd done that to fix this
> bug but must have forgotten or lost the patch :(
> We can then convert all the notifier chains in there to raw_*.

The two philosophers gaped at him.

"Bloody hell," said Majikthise, "now that is what I call
thinking. Here Vroomfondel, why do we never think of things like

"Dunno," said Vroomfondel in an awed whisper, "think our brains must
be too highly trained Majikthise." 

[ ]

I guess you'll be wanting this then, unless you have done it already.


Subject: Convert cpu hotplug notifiers to use raw_notifier instead of blocking_notifier

The use of blocking notifier by _cpu_up and _cpu_down in cpu.c
has two problem.

1/ An interaction with the workqueue notifier causes lockdep to
   spit a warning.
2/ A notifier could conceivable be added or removed while _cpu_up or
   _cpu_down are in process.  As each notifier is called twice 
   (prepare then commit/abort) this could be unhealthy.

To fix to we simply take cpu_add_remove_lock while adding
or removing notifiers to/from the list.

This makes the 'blocking' usage unnecessary as all accesses to
cpu_chain are now protected by cpu_add_remove_lock.  So
change "blocking" to "raw" in all relevant places.
This fixes 1.

Credit: Andrew Morton
Cc:  [email protected] (maintainer)
Cc: Michal Piotrowski <[email protected]> (reporter)
Signed-off-by: Neil Brown <[email protected]>

### Diffstat output
 ./kernel/cpu.c |   24 +++++++++++++++---------
 1 file changed, 15 insertions(+), 9 deletions(-)

diff .prev/kernel/cpu.c ./kernel/cpu.c
--- .prev/kernel/cpu.c	2006-10-13 14:30:56.000000000 +1000
+++ ./kernel/cpu.c	2006-10-13 14:33:49.000000000 +1000
@@ -19,7 +19,7 @@
 static DEFINE_MUTEX(cpu_add_remove_lock);
 static DEFINE_MUTEX(cpu_bitmask_lock);
-static __cpuinitdata BLOCKING_NOTIFIER_HEAD(cpu_chain);
+static __cpuinitdata RAW_NOTIFIER_HEAD(cpu_chain);
 /* If set, cpu_up and cpu_down will return -EBUSY and do nothing.
  * Should always be manipulated under cpu_add_remove_lock
@@ -68,7 +68,11 @@ EXPORT_SYMBOL_GPL(unlock_cpu_hotplug);
 /* Need to know about CPUs going up/down? */
 int __cpuinit register_cpu_notifier(struct notifier_block *nb)
-	return blocking_notifier_chain_register(&cpu_chain, nb);
+	int ret;
+	mutex_lock(&cpu_add_remove_lock);
+	ret = raw_notifier_chain_register(&cpu_chain, nb);
+	mutex_unlock(&cpu_add_remove_lock);
+	return ret;
@@ -77,7 +81,9 @@ EXPORT_SYMBOL(register_cpu_notifier);
 void unregister_cpu_notifier(struct notifier_block *nb)
-	blocking_notifier_chain_unregister(&cpu_chain, nb);
+	mutex_lock(&cpu_add_remove_lock);
+	raw_notifier_chain_unregister(&cpu_chain, nb);
+	mutex_unlock(&cpu_add_remove_lock);
@@ -126,7 +132,7 @@ static int _cpu_down(unsigned int cpu)
 	if (!cpu_online(cpu))
 		return -EINVAL;
-	err = blocking_notifier_call_chain(&cpu_chain, CPU_DOWN_PREPARE,
+	err = raw_notifier_call_chain(&cpu_chain, CPU_DOWN_PREPARE,
 						(void *)(long)cpu);
 	if (err == NOTIFY_BAD) {
 		printk("%s: attempt to take down CPU %u failed\n",
@@ -146,7 +152,7 @@ static int _cpu_down(unsigned int cpu)
 	if (IS_ERR(p)) {
 		/* CPU didn't die: tell everyone.  Can't complain. */
-		if (blocking_notifier_call_chain(&cpu_chain, CPU_DOWN_FAILED,
+		if (raw_notifier_call_chain(&cpu_chain, CPU_DOWN_FAILED,
 				(void *)(long)cpu) == NOTIFY_BAD)
@@ -169,7 +175,7 @@ static int _cpu_down(unsigned int cpu)
 	/* CPU is completely dead: tell everyone.  Too late to complain. */
-	if (blocking_notifier_call_chain(&cpu_chain, CPU_DEAD,
+	if (raw_notifier_call_chain(&cpu_chain, CPU_DEAD,
 			(void *)(long)cpu) == NOTIFY_BAD)
@@ -206,7 +212,7 @@ static int __devinit _cpu_up(unsigned in
 	if (cpu_online(cpu) || !cpu_present(cpu))
 		return -EINVAL;
-	ret = blocking_notifier_call_chain(&cpu_chain, CPU_UP_PREPARE, hcpu);
+	ret = raw_notifier_call_chain(&cpu_chain, CPU_UP_PREPARE, hcpu);
 	if (ret == NOTIFY_BAD) {
 		printk("%s: attempt to bring up CPU %u failed\n",
 				__FUNCTION__, cpu);
@@ -223,11 +229,11 @@ static int __devinit _cpu_up(unsigned in
 	/* Now call notifier in preparation. */
-	blocking_notifier_call_chain(&cpu_chain, CPU_ONLINE, hcpu);
+	raw_notifier_call_chain(&cpu_chain, CPU_ONLINE, hcpu);
 	if (ret != 0)
-		blocking_notifier_call_chain(&cpu_chain,
+		raw_notifier_call_chain(&cpu_chain,
 				CPU_UP_CANCELED, hcpu);
 	return ret;
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at
Please read the FAQ at

[Index of Archives]     [Kernel Newbies]     [Netfilter]     [Bugtraq]     [Photo]     [Stuff]     [Gimp]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Video 4 Linux]     [Linux for the blind]     [Linux Resources]
  Powered by Linux