Skip to contentSkip to conversation

locking/lockdep: Avoid potential access of invalid memory in lock_class

PATCH Applied Waiman Long 3 participants 5 messages Jan 3, 2022 – Jan 26, 2022 Locking primitives

lore
Share & tools
lore

Waiman Long

author reviewer
#0 lore raw

PATCH 1/1 locking/lockdep: Avoid potential access of invalid memory in lock_class

It was found that reading /proc/lockdep after a lockdep splat may potentially cause an access to freed memory if lockdep_unregister_key() is called after the splat but before access to /proc/lockdep [1]. This is due to the fact that graph_lock() call in lockdep_unregister_key() fails after the clearing of debug_locks by the splat process.

After lockdep_unregister_key() is called, the lock_name may be freed but the corresponding lock_class structure still have a reference to it. That invalid memory pointer will then be accessed when /proc/lockdep is read by a user and a use-after-free (UAF) error will be reported if KASAN is enabled.

To fix this problem, lockdep_unregister_key() is now modified to always search for a matching key irrespective of the debug_locks state and zap the corresponding lock class if a matching one is found.

[1] https://lore.kernel.org/lkml/[email protected]/

Fixes 8b39adbee805 ("locking/lockdep: Make lockdep_unregister_key() honor 'debug_locks' again")
---
 kernel/locking/lockdep.c | 24 +++++++++++++++---------
 1 file changed, 15 insertions(+), 9 deletions(-)
kernel/locking/lockdep.c +15 -9
index 2270ec68f10a..d4252b5c9863 100644
@@ -6288,7 +6288,13 @@ void lockdep_reset_lock(struct lockdep_map *lock)
 		lockdep_reset_lock_reg(lock);
 }
 
-/* Unregister a dynamically allocated key. */
+/*
+ * Unregister a dynamically allocated key.
+ *
+ * Unlike lockdep_register_key(), a search is always done to find a matching
+ * key irrespective of debug_locks to avoid potential invalid access to freed
+ * memory in lock_class entry.
+ */
 void lockdep_unregister_key(struct lock_class_key *key)
 {
 	struct hlist_head *hash_head = keyhashentry(key);
@@ -6303,10 +6309,8 @@ void lockdep_unregister_key(struct lock_class_key *key)
 		return;
 
 	raw_local_irq_save(flags);
-	if (!graph_lock())
-		goto out_irq;
+	lockdep_lock();
 
-	pf = get_pending_free();
 	hlist_for_each_entry_rcu(k, hash_head, hash_entry) {
 		if (k == key) {
 			hlist_del_rcu(&k->hash_entry);
@@ -6314,11 +6318,13 @@ void lockdep_unregister_key(struct lock_class_key *key)
 			break;
 		}
 	}
-	WARN_ON_ONCE(!found);
-	__lockdep_free_key_range(pf, key, 1);
-	call_rcu_zapped(pf);
-	graph_unlock();
-out_irq:
+	WARN_ON_ONCE(!found && debug_locks);
+	if (found) {
+		pf = get_pending_free();
+		__lockdep_free_key_range(pf, key, 1);
+		call_rcu_zapped(pf);
+	}
+	lockdep_unlock();
 	raw_local_irq_restore(flags);
 
 	/* Wait until is_dynamic_key() has finished accessing k->hash_entry. */
Signature · 1 line
2.27.0
#1 lore raw

kernel/locking/lockdep.c void lockdep_unregister_key(struct lock_class_key *key) · line 6321

Quoted 6 lines · Waiman Long
kernel/locking/lockdep.c +1 -5
-	WARN_ON_ONCE(!found);
-	__lockdep_free_key_range(pf, key, 1);
-	call_rcu_zapped(pf);
-	graph_unlock();
-out_irq:
+	WARN_ON_ONCE(!found && debug_locks);

lockdep_unregister_key() should only be called for a registered key so I'd like to keep the WARN_ON_ONCE(!found) here instead of changing it into WARN_ON_ONCE(!found && debug_locks). Otherwise this patch looks good to me.

Thanks,

Bart.

author reviewer
#2 lore raw
Quoted 11 lines · Bart Van Assche
Quoted 6 lines · Waiman Long

-    WARN_ON_ONCE(!found);
-    __lockdep_free_key_range(pf, key, 1);
-    call_rcu_zapped(pf);
-    graph_unlock();
-out_irq:
+    WARN_ON_ONCE(!found && debug_locks);

lockdep_unregister_key() should only be called for a registered key so I'd like to keep the WARN_ON_ONCE(!found) here instead of changing it into WARN_ON_ONCE(!found && debug_locks). Otherwise this patch looks good to me.

The reason for this change is to handle the case that a lockdep_register_key()/lockdep_unregister_key() pair may be called after debug_locks is turned off. Without that change, we are going to get the warning in lockdep_unregister_key(). That is the motivation of your original commit 8b39adbee805 ("locking/lockdep: Make lockdep_unregister_key() honor 'debug_locks' again").

Cheers,
Longman

Reviewed-by
#3 lore raw
Quoted 9 lines · Waiman Long
Quoted 8 lines · Bart Van Assche
Quoted 6 lines · Waiman Long

-    WARN_ON_ONCE(!found);
-    __lockdep_free_key_range(pf, key, 1);
-    call_rcu_zapped(pf);
-    graph_unlock();
-out_irq:
+    WARN_ON_ONCE(!found && debug_locks);

Quoted 5 lines
Quoted 3 lines

lockdep_unregister_key() should only be called for a registered key so I'd like to keep the WARN_ON_ONCE(!found) here instead of changing it into WARN_ON_ONCE(!found && debug_locks). Otherwise this patch looks good to me.

The reason for this change is to handle the case that a lockdep_register_key()/lockdep_unregister_key() pair may be called after debug_locks is turned off. Without that change, we are going to get the warning in lockdep_unregister_key(). That is the motivation of your original commit 8b39adbee805 ("locking/lockdep: Make lockdep_unregister_key() honor 'debug_locks' again").

Ah, that's right. Hence:

merge bot Applied
#4 lore raw
In reply to Waiman Long
The following commit has been merged into the locking/core branch of tip:

The following commit has been merged into the locking/core branch of tip:

Commit-ID:     61cc4534b6550997c97a03759ab46b29d44c0017
Gitweb:        https://git.kernel.org/tip/61cc4534b6550997c97a03759ab46b29d44c0017
Author:        Waiman Long <[email protected]>
AuthorDate:    Sun, 02 Jan 2022 21:35:58 -05:00
Committer:     Peter Zijlstra <[email protected]>
CommitterDate: Tue, 25 Jan 2022 22:30:28 +01:00

locking/lockdep: Avoid potential access of invalid memory in lock_class

It was found that reading /proc/lockdep after a lockdep splat may potentially cause an access to freed memory if lockdep_unregister_key() is called after the splat but before access to /proc/lockdep [1]. This is due to the fact that graph_lock() call in lockdep_unregister_key() fails after the clearing of debug_locks by the splat process.

After lockdep_unregister_key() is called, the lock_name may be freed but the corresponding lock_class structure still have a reference to it. That invalid memory pointer will then be accessed when /proc/lockdep is read by a user and a use-after-free (UAF) error will be reported if KASAN is enabled.

To fix this problem, lockdep_unregister_key() is now modified to always search for a matching key irrespective of the debug_locks state and zap the corresponding lock class if a matching one is found.

[1] https://lore.kernel.org/lkml/[email protected]/

Fixes 8b39adbee805 ("locking/lockdep: Make lockdep_unregister_key() honor 'debug_locks' again")
---
 kernel/locking/lockdep.c | 24 +++++++++++++++---------
 1 file changed, 15 insertions(+), 9 deletions(-)
kernel/locking/lockdep.c +15 -9
index 89b3df5..2e6892e 100644
@@ -6287,7 +6287,13 @@ void lockdep_reset_lock(struct lockdep_map *lock)
 		lockdep_reset_lock_reg(lock);
 }
 
-/* Unregister a dynamically allocated key. */
+/*
+ * Unregister a dynamically allocated key.
+ *
+ * Unlike lockdep_register_key(), a search is always done to find a matching
+ * key irrespective of debug_locks to avoid potential invalid access to freed
+ * memory in lock_class entry.
+ */
 void lockdep_unregister_key(struct lock_class_key *key)
 {
 	struct hlist_head *hash_head = keyhashentry(key);
@@ -6302,10 +6308,8 @@ void lockdep_unregister_key(struct lock_class_key *key)
 		return;
 
 	raw_local_irq_save(flags);
-	if (!graph_lock())
-		goto out_irq;
+	lockdep_lock();
 
-	pf = get_pending_free();
 	hlist_for_each_entry_rcu(k, hash_head, hash_entry) {
 		if (k == key) {
 			hlist_del_rcu(&k->hash_entry);
@@ -6313,11 +6317,13 @@ void lockdep_unregister_key(struct lock_class_key *key)
 			break;
 		}
 	}
-	WARN_ON_ONCE(!found);
-	__lockdep_free_key_range(pf, key, 1);
-	call_rcu_zapped(pf);
-	graph_unlock();
-out_irq:
+	WARN_ON_ONCE(!found && debug_locks);
+	if (found) {
+		pf = get_pending_free();
+		__lockdep_free_key_range(pf, key, 1);
+		call_rcu_zapped(pf);
+	}
+	lockdep_unlock();
 	raw_local_irq_restore(flags);
 
 	/* Wait until is_dynamic_key() has finished accessing k->hash_entry. */