netdev - Re: [PATCH net] rxrpc: Fix lack of conn cleanup when local endpoint is cleaned up

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Sat, 24 Aug 2019 14:35:33 -0700 (PDT)
From:   David Miller <davem@...emloft.net>
To:     dhowells@...hat.com
Cc:     netdev@...r.kernel.org, marc.dionne@...istor.com,
        linux-afs@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH net] rxrpc: Fix lack of conn cleanup when local
 endpoint is cleaned up

From: David Howells <dhowells@...hat.com>
Date: Thu, 22 Aug 2019 13:26:38 +0100

> +	spin_lock(&rxnet->client_conn_cache_lock);
> +	nr_active = rxnet->nr_active_client_conns;
> +
> +	list_for_each_entry_safe(conn, tmp, &rxnet->idle_client_conns,
> +				 cache_link) {
> +		if (conn->params.local == local) {
> +			ASSERTCMP(conn->cache_state, ==, RXRPC_CONN_CLIENT_IDLE);
> +
> +			trace_rxrpc_client(conn, -1, rxrpc_client_discard);
> +			if (!test_and_clear_bit(RXRPC_CONN_EXPOSED, &conn->flags))
> +				BUG();
> +			conn->cache_state = RXRPC_CONN_CLIENT_INACTIVE;
> +			list_move(&conn->cache_link, &graveyard);
> +			nr_active--;
> +		}
> +	}
> +
> +	rxnet->nr_active_client_conns = nr_active;
> +	spin_unlock(&rxnet->client_conn_cache_lock);
> +	ASSERTCMP(nr_active, >=, 0);
> +
> +	spin_lock(&rxnet->client_conn_cache_lock);
> +	while (!list_empty(&graveyard)) {
> +		conn = list_entry(graveyard.next,
> +				  struct rxrpc_connection, cache_link);
> +		list_del_init(&conn->cache_link);
> +		spin_unlock(&rxnet->client_conn_cache_lock);
> +
> +		rxrpc_put_connection(conn);
> +
> +		spin_lock(&rxnet->client_conn_cache_lock);
> +	}
> +	spin_unlock(&rxnet->client_conn_cache_lock);
> +
> +	_leave(" [culled]");

Once you've removed the entries from the globally visible idle_client_comms
list, and put them on the local garbage list, they cannot be seen in any way
by external threads of control outside of this function.

Therefore, you don't need to take the client_conn_cache_lock at all in the
second while loop.