audit: stop an old auditd being starved out by a new auditd

Nothing prevents a new auditd starting up and replacing a valid
audit_pid when an old auditd is still running, effectively starving out
the old auditd since audit_pid no longer points to the old valid
auditd.

If no message to auditd has been attempted since auditd died
unnaturally or got killed, audit_pid will still indicate it is alive.
There isn't an easy way to detect if an old auditd is still running on
the existing audit_pid other than attempting to send a message to see
if it fails.  An -ECONNREFUSED almost certainly means it disappeared
and can be replaced.  Other errors are not so straightforward and may
indicate transient problems that will resolve themselves and the old
auditd will recover.  Yet others will likely need manual intervention
for which a new auditd will not solve the problem.

Send a new message type (AUDIT_REPLACE) to the old auditd containing a
u32 with the PID of the new auditd.  If the audit replace message
succeeds (or doesn't fail with certainty), fail to register the new
auditd and return an error (-EEXIST).

This is expected to make the patch preventing an old auditd orphaning a
new auditd redundant.

V3: Switch audit message type from 1000 to 1300 block.

Signed-off-by: Richard Guy Briggs <rgb@redhat.com>
Signed-off-by: Paul Moore <pmoore@redhat.com>

1 files changed, 15 insertions, 1 deletions

diff --git a/kernel/audit.c b/kernel/audit.c
index d6dd95c..2fd63d6 100644
--- a/kernel/audit.c
+++ b/kernel/audit.c
@@ -809,6 +809,16 @@ static int audit_set_feature(struct sk_buff *skb)
 	return 0;
 }
 
+static int audit_replace(pid_t pid)
+{
+	struct sk_buff *skb = audit_make_reply(0, 0, AUDIT_REPLACE, 0, 0,
+					       &pid, sizeof(pid));
+
+	if (!skb)
+		return -ENOMEM;
+	return netlink_unicast(audit_sock, skb, audit_nlk_portid, 0);
+}
+
 static int audit_receive_msg(struct sk_buff *skb, struct nlmsghdr *nlh)
 {
 	u32			seq;
@@ -870,9 +880,13 @@ static int audit_receive_msg(struct sk_buff *skb, struct nlmsghdr *nlh)
 		}
 		if (s.mask & AUDIT_STATUS_PID) {
 			int new_pid = s.pid;
+			pid_t requesting_pid = task_tgid_vnr(current);
 
-			if ((!new_pid) && (task_tgid_vnr(current) != audit_pid))
+			if ((!new_pid) && (requesting_pid != audit_pid))
 				return -EACCES;
+			if (audit_pid && new_pid &&
+			    audit_replace(requesting_pid) != -ECONNREFUSED)
+				return -EEXIST;
 			if (audit_enabled != AUDIT_OFF)
 				audit_log_config_change("audit_pid", new_pid, audit_pid, 1);
 			audit_pid = new_pid;