[PATCH] audit: Fix data race in audit_log_start()

From: wangchi <[email protected]>

audit_log_start() reads audit_queue.qlen via skb_queue_len() without
holding the queue lock or using READ_ONCE(), while kauditd writes to
this field via the skb_dequeue() → __skb_unlink() path with WRITE_ONCE()
protected by a spinlock. This constitutes a data race.

KCSAN reports the following conflicting access pattern:
==================================================================
BUG: KCSAN: data-race in audit_log_start / skb_dequeue

write (marked) to 0xffffffff8512ee20 of 4 bytes by task 661 on cpu 57:
 skb_dequeue+0x70/0xf0
 kauditd_send_queue+0x71/0x220
 kauditd_thread+0x1cb/0x430
 kthread+0x1c2/0x210
 ret_from_fork+0x162/0x1a0
 ret_from_fork_asm+0x1a/0x30

read to 0xffffffff8512ee20 of 4 bytes by task 36586 on cpu 1:
 audit_log_start+0x2a0/0x6b0
 audit_core_dumps+0x64/0xa0
 do_coredump+0x14b/0x1260
 get_signal+0xeb2/0xf70
 arch_do_signal_or_restart+0x41/0x170
 exit_to_user_mode_loop+0xa2/0x1c0
 do_syscall_64+0x1a3/0x1c0
 entry_SYSCALL_64_after_hwframe+0x76/0xe0

value changed: 0x00000001 -> 0x00000000
==================================================================

Resolve the race by switching to lockless helper skb_queue_len_lockless(),
which internally uses READ_ONCE() and properly pairs with the WRITE_ONCE()
write accesses already present on the writer side.

Fixes: 3197542482df ("audit: rework audit_log_start()")
Change-Id: I766cc99f6a567b898a08e1d68cd13bae5699e5b2
Signed-off-by: wangchi <[email protected]>
---
 kernel/audit.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/audit.c b/kernel/audit.c
index 34dc7cb246ff..e4c095017302 100644
--- a/kernel/audit.c
+++ b/kernel/audit.c
@@ -1933,7 +1933,7 @@ struct audit_buffer *audit_log_start(struct audit_context *ctx, gfp_t gfp_mask,
 		long stime = audit_backlog_wait_time;

 		while (audit_backlog_limit &&
-		       (skb_queue_len(&audit_queue) > audit_backlog_limit)) {
+			(skb_queue_len_lockless(&audit_queue) > audit_backlog_limit)) {
 			/* wake kauditd to try and flush the queue */
 			wake_up_interruptible(&kauditd_wait);

@@ -1953,7 +1953,7 @@ struct audit_buffer *audit_log_start(struct audit_context *ctx, gfp_t gfp_mask,
 			} else {
 				if (audit_rate_check() && printk_ratelimit())
 					pr_warn("audit_backlog=%d > audit_backlog_limit=%d\n",
-						skb_queue_len(&audit_queue),
+						skb_queue_len_lockless(&audit_queue),
 						audit_backlog_limit);
 				audit_log_lost("backlog limit exceeded");
 				return NULL;
--
2.25.1