RE: [Resend PATCH] RDS: fix race condition when sending a message on unbound socket

[David Laight]

From: Santosh Shilimkar
> Sent: 24 November 2015 22:13
...
> Sasha's found a NULL pointer dereference in the RDS connection code when
> sending a message to an apparently unbound socket.  The problem is caused
> by the code checking if the socket is bound in rds_sendmsg(), which checks
> the rs_bound_addr field without taking a lock on the socket.  This opens a
> race where rs_bound_addr is temporarily set but where the transport is not
> in rds_bind(), leading to a NULL pointer dereference when trying to
> dereference 'trans' in __rds_conn_create().
> 
> Vegard wrote a reproducer for this issue, so kindly ask him to share if
> you're interested.
...
> diff --git a/net/rds/send.c b/net/rds/send.c
> index 827155c..c9cdb35 100644
> --- a/net/rds/send.c
> +++ b/net/rds/send.c
> @@ -1013,11 +1013,13 @@ int rds_sendmsg(struct socket *sock, struct msghdr *msg, size_t payload_len)
>  		release_sock(sk);

This is falling though into an unconditional lock_sock().
No need to unlock and relock immediately.

>  	}
> 
> -	/* racing with another thread binding seems ok here */
> +	lock_sock(sk);
>  	if (daddr == 0 || rs->rs_bound_addr == 0) {
> +		release_sock(sk);
>  		ret = -ENOTCONN; /* XXX not a great errno */
>  		goto out;
>  	}
> +	release_sock(sk);
> 

On the face of it the above looks somewhat dubious.
Locks usually tie together two action (eg a test and use of a value),
In this case you only have a test inside the lock.
That either means that the state can change after you release the lock
(ie rs->rs_bound_addr = 0 is executed somewhere), or you don't
really need the lock.

	David


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/