audit: bugfix for infinite loop when flush the hold queue

hulk inclusion
category: bugfix
bugzilla: 186105, https://gitee.com/openeuler/kernel/issues/I4RGWS?from=project-issue


CVE: NA

-----------------------------------------------------------------

When we add "audit=1" to the cmdline, if we keep the audit_hold_queue
non-empty, flush the hold queue will fall into an infinite loop. So we
need to fix it by stoping flush the hold queue when netlink abnormal.

Fixes: 3413ddc9 ("audit: improve robustness of the audit queue handling")
Signed-off-by: Cui GaoSheng <cuigaosheng1@huawei.com>
Reviewed-by: Xiu Jianfeng <xiujianfeng@huawei.com>
Reviewed-by: weiyang wang <wangweiyang2@huawei.com>
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>

kernel/audit.c

@@ -740,6 +740,8 @@ static int kauditd_send_queue(struct sock *sk, u32 portid,
 		if (!sk) {
 			if (err_hook)
 				(*err_hook)(skb);
+			if (queue == &audit_hold_queue)
+				goto out;
 			continue;
 		}
 
@@ -756,6 +758,8 @@ static int kauditd_send_queue(struct sock *sk, u32 portid,
 					(*err_hook)(skb);
 				if (rc == -EAGAIN)
 					rc = 0;
+				if (queue == &audit_hold_queue)
+					goto out;
 				/* continue to drain the queue */
 				continue;
 			} else
@@ -767,6 +771,7 @@ static int kauditd_send_queue(struct sock *sk, u32 portid,
 		}
 	}
 
+out:
 	return (rc >= 0 ? 0 : rc);
 }