[PATCH] audit: annotate the lockless read of the audit control lock owner

[Babanpreet Singh]

audit_ctl_owner_current() reads the lock owner without taking the lock
and KCSAN complains about it. Only the current task can store itself as
owner, so the check can't give a wrong answer. Mark the accesses with
READ_ONCE() and WRITE_ONCE().

Reported-by: syzbot+39b8fea0641107a1ccee@xxxxxxxxxxxxxxxxxxxxxxxxx
Closes: https://syzkaller.appspot.com/bug?extid=39b8fea0641107a1ccee
Assisted-by: Claude:claude-opus-5-5
Signed-off-by: Babanpreet Singh <bbnpreetsingh@xxxxxxxxx>
---
Paul, you looked at this race in 2022 and asked whether READ_ONCE()
was worth it and what it would cost:
https://lore.kernel.org/all/CAHC9VhTXNPWBDRoPcz-Jw=f+NNAEhxbh-ySc56CUd-ZbuboW5w@xxxxxxxxxxxxxx/
Compile tested only (gcc and the KCSAN instrumentation diff); I could
not reproduce the race in QEMU.

 kernel/audit.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/kernel/audit.c b/kernel/audit.c
index 631d6d4a23cfd..f8c43ac6d5b82 100644
--- a/kernel/audit.c
+++ b/kernel/audit.c
@@ -247,7 +247,7 @@ int auditd_test_task(struct task_struct *task)
 void audit_ctl_lock(void)
 {
 	mutex_lock(&audit_cmd_mutex.lock);
-	audit_cmd_mutex.owner = current;
+	WRITE_ONCE(audit_cmd_mutex.owner, current);
 }
 
 /**
@@ -255,7 +255,7 @@ void audit_ctl_lock(void)
  */
 void audit_ctl_unlock(void)
 {
-	audit_cmd_mutex.owner = NULL;
+	WRITE_ONCE(audit_cmd_mutex.owner, NULL);
 	mutex_unlock(&audit_cmd_mutex.lock);
 }
 
@@ -268,7 +268,12 @@ void audit_ctl_unlock(void)
  */
 static bool audit_ctl_owner_current(void)
 {
-	return (current == audit_cmd_mutex.owner);
+	/*
+	 * Lockless read: the owner can only equal current if current set
+	 * it, so another task changing the owner concurrently can never
+	 * make this return the wrong answer.
+	 */
+	return (current == READ_ONCE(audit_cmd_mutex.owner));
 }
 
 /**

base-commit: a8bdcf944504980635c2069b4a4cfcf677b09bbb
-- 
2.43.0