Skip to contentSkip to conversation

locking/lockdep: Zap lock classes even with lock debugging disabled

PATCH Superseded Bart Van Assche 2 participants 5 messages Mar 26, 2019 – Apr 3, 2019 Locking primitives

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 locking/lockdep: Zap lock classes even with lock debugging disabled

Commit a0b0fd53e1e6 ("locking/lockdep: Free lock classes that are no longer in use") changed the behavior of lockdep_free_key_range() from unconditionally zapping lock classes into only zapping lock classes if debug_lock == true. Since the new behavior can cause cat /proc/lockdep to crash due to a NULL pointer dereference, restore the pre-v5.1 behavior.

Cc Thomas Gleixner <[email protected]>
Cc Will Deacon <[email protected]>
Cc Waiman Long <[email protected]>
Cc shenghui <[email protected]>
Fixes a0b0fd53e1e6 ("locking/lockdep: Free lock classes that are no longer in use") # v5.1-rc1.
---
 kernel/locking/lockdep.c | 23 ++++++-----------------
 1 file changed, 6 insertions(+), 17 deletions(-)
kernel/locking/lockdep.c +6 -17
index 34cdcbedda49..70480e4f8f5d 100644
@@ -4689,8 +4689,7 @@ static void free_zapped_rcu(struct rcu_head *ch)
 		return;
 
 	raw_local_irq_save(flags);
-	if (!graph_lock())
-		goto out_irq;
+	arch_spin_lock(&lockdep_lock);
 
 	/* closed head */
 	pf = delayed_free.pf + (delayed_free.index ^ 1);
@@ -4702,8 +4701,7 @@ static void free_zapped_rcu(struct rcu_head *ch)
 	 */
 	call_rcu_zapped(delayed_free.pf + delayed_free.index);
 
-	graph_unlock();
-out_irq:
+	arch_spin_unlock(&lockdep_lock);
 	raw_local_irq_restore(flags);
 }
 
@@ -4744,21 +4742,15 @@ static void lockdep_free_key_range_reg(void *start, unsigned long size)
 {
 	struct pending_free *pf;
 	unsigned long flags;
-	int locked;
 
 	init_data_structures_once();
 
 	raw_local_irq_save(flags);
-	locked = graph_lock();
-	if (!locked)
-		goto out_irq;
-
+	arch_spin_lock(&lockdep_lock);
 	pf = get_pending_free();
 	__lockdep_free_key_range(pf, start, size);
 	call_rcu_zapped(pf);
-
-	graph_unlock();
-out_irq:
+	arch_spin_unlock(&lockdep_lock);
 	raw_local_irq_restore(flags);
 
 	/*
@@ -4911,9 +4903,7 @@ void lockdep_unregister_key(struct lock_class_key *key)
 		return;
 
 	raw_local_irq_save(flags);
-	if (!graph_lock())
-		goto out_irq;
-
+	arch_spin_lock(&lockdep_lock);
 	pf = get_pending_free();
 	hlist_for_each_entry_rcu(k, hash_head, hash_entry) {
 		if (k == key) {
@@ -4925,8 +4915,7 @@ void lockdep_unregister_key(struct lock_class_key *key)
 	WARN_ON_ONCE(!found);
 	__lockdep_free_key_range(pf, key, 1);
 	call_rcu_zapped(pf);
-	graph_unlock();
-out_irq:
+	arch_spin_unlock(&lockdep_lock);
 	raw_local_irq_restore(flags);
 
 	/* Wait until is_dynamic_key() has finished accessing k->hash_entry. */
Signature · 1 line
2.21.0.196.g041f5ea1cf98
#1 lore raw
Quoted 5 lines · Bart Van Assche

Commit a0b0fd53e1e6 ("locking/lockdep: Free lock classes that are no longer in use") changed the behavior of lockdep_free_key_range() from unconditionally zapping lock classes into only zapping lock classes if debug_lock == true. Since the new behavior can cause cat /proc/lockdep to crash due to a NULL pointer dereference, restore the pre-v5.1 behavior.

Can you elaborate on this NULL dereference please, and why this patch fixes it?

Quoted 16 lines
Cc Thomas Gleixner <[email protected]>
Cc Will Deacon <[email protected]>
Cc Waiman Long <[email protected]>
Cc shenghui <[email protected]>
Fixes a0b0fd53e1e6 ("locking/lockdep: Free lock classes that are no longer in use") # v5.1-rc1.
---
 kernel/locking/lockdep.c | 23 ++++++-----------------
 1 file changed, 6 insertions(+), 17 deletions(-)
kernel/locking/lockdep.c +0 -0
index 34cdcbedda49..70480e4f8f5d 100644

kernel/locking/lockdep.c static void free_zapped_rcu(struct rcu_head *ch) · line 4692

Quoted 6 lines
kernel/locking/lockdep.c +1 -2
@@ -4689,8 +4689,7 @@ static void free_zapped_rcu(struct rcu_head *ch)
 		return;
 
 	raw_local_irq_save(flags);
-	if (!graph_lock())
-		goto out_irq;
+	arch_spin_lock(&lockdep_lock);

This also throws out the recursion counting. Is that ok?

Will

#2 lore raw
Quoted 9 lines · Will Deacon
Quoted 5 lines · Bart Van Assche

Commit a0b0fd53e1e6 ("locking/lockdep: Free lock classes that are no longer in use" changed the behavior of lockdep_free_key_range() from unconditionally zapping lock classes into only zapping lock classes if debug_lock ==true. Since the new behavior can cause cat /proc/lockdep to crash due to a NULL pointer dereference, restore the pre-v5.1 behavior.

Can you elaborate on this NULL dereference please, and why this patch fixes it?

Hi Will,

Not zapping lock classes if debug_lock ==false leaves dangling pointers in
several lockdep datastructures, e.g. lock_class::name in the all_lock_classes
list. The shell command "cat /proc/lockdep"causes the kernel to iterate the
all_lock_classes list. Hence the "unable to handle kernel paging request"issue that Shenghui encountered by running cat /proc/lockdep. Please let me know if you would like me to repost this patch with a more detailed
description.

Quoted 5 lines
Quoted 5 lines
Cc Thomas Gleixner <[email protected]>>>Cc: Will Deacon <[email protected]>>>Cc: Waiman Long <[email protected]>>>Cc: shenghui <[email protected]>>>Reported-by: shenghui <[email protected]>>>Fixes: a0b0fd53e1e6 ("locking/lockdep: Free lock classes that are no longer in use" #v5.1-rc1.

kernel/locking/lockdep.c |23 +-+-+-+-+-+------------------

1 file changed, 6 insertions(+-), 17 deletions(-)
Quoted 9 lines

kernel/locking/lockdep.c

Quoted 8 lines
kernel/locking/lockdep.c +1 -1
index 34cdcbedda49..70480e4f8f5d 100644
--- a/kernel/locking/lockdep.c
+-+-+- b/kernel/locking/lockdep.c

@@-4689,8 +-4689,7 @@static void free_zapped_rcu(struct rcu_head *ch)

		return;>> 
	raw_local_irq_save(flags);>>-	if (!graph_lock())
kernel/locking/lockdep.c +0 -1
-		goto out_irq;>>+-	arch_spin_lock(&lockdep_lock);>

This also throws out the recursion counting. Is that ok?

I think that that's OK. My understanding is that lockdep keeps track of recursion to avoid that lockdep_lock is locked recursively. However, none of the functions modified by this patch are called with that lock held.

Thanks,

Bart.

#3 lore raw
Quoted 11 lines · Bart Van Assche
Quoted 9 lines · Will Deacon
Quoted 5 lines · Bart Van Assche

Commit a0b0fd53e1e6 ("locking/lockdep: Free lock classes that are no longer in use") changed the behavior of lockdep_free_key_range() from unconditionally zapping lock classes into only zapping lock classes if debug_lock == true. Since the new behavior can cause cat /proc/lockdep to crash due to a NULL pointer dereference, restore the pre-v5.1 behavior.

Can you elaborate on this NULL dereference please, and why this patch fixes it?

Quoted 7 lines

Not zapping lock classes if debug_lock == false leaves dangling pointers in several lockdep datastructures, e.g. lock_class::name in the all_lock_classes list. The shell command "cat /proc/lockdep" causes the kernel to iterate the all_lock_classes list. Hence the "unable to handle kernel paging request" issue that Shenghui encountered by running cat /proc/lockdep. Please let me know if you would like me to repost this patch with a more detailed description.

That would help me, at least (maybe with the crash log). Thanks.

Quoted 25 lines
Quoted 18 lines
Quoted 18 lines
Cc Thomas Gleixner <[email protected]>
Cc Will Deacon <[email protected]>
Cc Waiman Long <[email protected]>
Cc shenghui <[email protected]>
Fixes a0b0fd53e1e6 ("locking/lockdep: Free lock classes that are no longer in use") # v5.1-rc1.
---
 kernel/locking/lockdep.c | 23 ++++++-----------------
 1 file changed, 6 insertions(+), 17 deletions(-)
kernel/locking/lockdep.c +0 -0
index 34cdcbedda49..70480e4f8f5d 100644
@@ -4689,8 +4689,7 @@ static void free_zapped_rcu(struct rcu_head *ch)
 		return;
 
Quoted 6 lines

kernel/locking/lockdep.c static void free_zapped_rcu(struct rcu_head *ch) · line 4692

Quoted 4 lines
kernel/locking/lockdep.c +1 -2
 	raw_local_irq_save(flags);
-	if (!graph_lock())
-		goto out_irq;
+	arch_spin_lock(&lockdep_lock);

This also throws out the recursion counting. Is that ok?

Quoted 3 lines

I think that that's OK. My understanding is that lockdep keeps track of recursion to avoid that lockdep_lock is locked recursively. However, none of the functions modified by this patch are called with that lock held.

Might be worth adding a comment to that effect, so people don't change that in future.

Will

#4 lore raw
In reply to Will Deacon
Quoted 7 lines · Will Deacon

Review

Quoted 6 lines

--- a/kernel/locking/lockdep.c
+-+-+- b/kernel/locking/lockdep.c
@@-4689,8 +-4689,7 @@static void free_zapped_rcu(struct rcu_head *ch)

		return;>> 
	raw_local_irq_save(flags);>>-	if (!graph_lock())
(quoted diff) +0 -1
-		goto out_irq;>>+-	arch_spin_lock(&lockdep_lock);>

This also throws out the recursion counting. Is that ok?

Apparently not, as the build bot just figured out. I will restore recursion counting. See also
https://lore.kernel.org/lkml/5ca4ea66.eVtXtf5CmqRjP7xF%[email protected]/.

Bart.