Hi Chuck,
On Tue, May 16, 2023 at 3:52 PM Chuck Lever <cel@xxxxxxxxxx> wrote:
>
> From: Chuck Lever <chuck.lever@xxxxxxxxxx>
>
> Use the new TLS handshake API to enable the SunRPC client code
> to request a TLS handshake. This implements support for RFC 9289.
>
> Signed-off-by: Chuck Lever <chuck.lever@xxxxxxxxxx>
> ---
> include/linux/sunrpc/xprtsock.h | 1
> net/sunrpc/xprtsock.c | 289 ++++++++++++++++++++++++++++++++++-----
> 2 files changed, 253 insertions(+), 37 deletions(-)
>
> diff --git a/include/linux/sunrpc/xprtsock.h b/include/linux/sunrpc/xprtsock.h
> index 574a6a5391ba..700a1e6c047c 100644
> --- a/include/linux/sunrpc/xprtsock.h
> +++ b/include/linux/sunrpc/xprtsock.h
> @@ -57,6 +57,7 @@ struct sock_xprt {
> struct work_struct error_worker;
> struct work_struct recv_worker;
> struct mutex recv_mutex;
> + struct completion handshake_done;
> struct sockaddr_storage srcaddr;
> unsigned short srcport;
> int xprt_err;
> diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
> index 7ea5984a52a3..686dd313f89f 100644
> --- a/net/sunrpc/xprtsock.c
> +++ b/net/sunrpc/xprtsock.c
> @@ -48,6 +48,7 @@
> #include <net/udp.h>
> #include <net/tcp.h>
> #include <net/tls.h>
> +#include <net/handshake.h>
>
> #include <linux/bvec.h>
> #include <linux/highmem.h>
> @@ -189,6 +190,11 @@ static struct ctl_table xs_tunables_table[] = {
> */
> #define XS_IDLE_DISC_TO (5U * 60 * HZ)
>
> +/*
> + * TLS handshake timeout.
> + */
> +#define XS_TLS_HANDSHAKE_TO (10U * HZ)
> +
> #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
> # undef RPC_DEBUG_DATA
> # define RPCDBG_FACILITY RPCDBG_TRANS
> @@ -1238,6 +1244,10 @@ static void xs_reset_transport(struct sock_xprt *transport)
> if (atomic_read(&transport->xprt.swapper))
> sk_clear_memalloc(sk);
>
> + /* XXX: Maybe also send a TLS Closure alert? */
> +
> + tls_handshake_cancel(sk);
> +
> kernel_sock_shutdown(sock, SHUT_RDWR);
>
> mutex_lock(&transport->recv_mutex);
> @@ -2411,60 +2421,266 @@ static void xs_tcp_setup_socket(struct work_struct *work)
> current_restore_flags(pflags, PF_MEMALLOC);
> }
>
> +/*
> + * Transfer the connected socket to @upper_transport, then mark that
> + * xprt CONNECTED.
> + */
> +static int xs_tls_finish_connecting(struct rpc_xprt *lower_xprt,
> + struct sock_xprt *upper_transport)
> +{
> + struct sock_xprt *lower_transport =
> + container_of(lower_xprt, struct sock_xprt, xprt);
> + struct rpc_xprt *upper_xprt = &upper_transport->xprt;
> +
> + if (!upper_transport->inet) {
> + struct socket *sock = lower_transport->sock;
> + struct sock *sk = sock->sk;
> +
> + /* Avoid temporary address, they are bad for long-lived
> + * connections such as NFS mounts.
> + * RFC4941, section 3.6 suggests that:
> + * Individual applications, which have specific
> + * knowledge about the normal duration of connections,
> + * MAY override this as appropriate.
> + */
> + if (xs_addr(upper_xprt)->sa_family == PF_INET6) {
> + ip6_sock_set_addr_preferences(sk,
> + IPV6_PREFER_SRC_PUBLIC);
> + }
> +
> + xs_tcp_set_socket_timeouts(upper_xprt, sock);
> + tcp_sock_set_nodelay(sk);
> +
> + lock_sock(sk);
> +
> + /*
> + * @sk is already connected, so it now has the RPC callbacks.
> + * Reach into @lower_transport to save the original ones.
> + */
> + upper_transport->old_data_ready = lower_transport->old_data_ready;
> + upper_transport->old_state_change = lower_transport->old_state_change;
> + upper_transport->old_write_space = lower_transport->old_write_space;
> + upper_transport->old_error_report = lower_transport->old_error_report;
> + sk->sk_user_data = upper_xprt;
> +
> + /* socket options */
> + sock_reset_flag(sk, SOCK_LINGER);
> +
> + xprt_clear_connected(upper_xprt);
> +
> + upper_transport->sock = sock;
> + upper_transport->inet = sk;
> + upper_transport->file = lower_transport->file;
> +
> + release_sock(sk);
> +
> + /* Reset lower_transport before shutting down its clnt */
> + mutex_lock(&lower_transport->recv_mutex);
> + lower_transport->inet = NULL;
> + lower_transport->sock = NULL;
> + lower_transport->file = NULL;
> +
> + xprt_clear_connected(lower_xprt);
> + xs_sock_reset_connection_flags(lower_xprt);
> + xs_stream_reset_connect(lower_transport);
> + mutex_unlock(&lower_transport->recv_mutex);
> + }
> +
> + if (!xprt_bound(upper_xprt))
> + return -ENOTCONN;
> +
> + xs_set_memalloc(upper_xprt);
> +
> + if (!xprt_test_and_set_connected(upper_xprt)) {
> + upper_xprt->connect_cookie++;
> + clear_bit(XPRT_SOCK_CONNECTING, &upper_transport->sock_state);
> + xprt_clear_connecting(upper_xprt);
> +
> + upper_xprt->stat.connect_count++;
> + upper_xprt->stat.connect_time += (long)jiffies -
> + upper_xprt->stat.connect_start;
> + xs_run_error_worker(upper_transport, XPRT_SOCK_WAKE_PENDING);
> + }
> + return 0;
> +}
> +
> /**
> - * xs_tls_connect - establish a TLS session on a socket
> - * @work: queued work item
> + * xs_tls_handshake_done - TLS handshake completion handler
> + * @data: address of xprt to wake
> + * @status: status of handshake
> + * @peerid: serial number of key containing the remote's identity
> *
> */
> -static void xs_tls_connect(struct work_struct *work)
> +static void xs_tls_handshake_done(void *data, int status, key_serial_t peerid)
> {
> - struct sock_xprt *transport =
> - container_of(work, struct sock_xprt, connect_worker.work);
> - struct rpc_clnt *clnt;
> + struct rpc_xprt *lower_xprt = data;
> + struct sock_xprt *lower_transport =
> + container_of(lower_xprt, struct sock_xprt, xprt);
>
> - clnt = transport->clnt;
> - transport->clnt = NULL;
> - if (IS_ERR(clnt))
> - goto out_unlock;
> + lower_transport->xprt_err = status ? -EACCES : 0;
> + complete(&lower_transport->handshake_done);
> + xprt_put(lower_xprt);
> +}
>
> - xs_tcp_setup_socket(work);
> +static int xs_tls_handshake_sync(struct rpc_xprt *lower_xprt, struct xprtsec_parms *xprtsec)
> +{
> + struct sock_xprt *lower_transport =
> + container_of(lower_xprt, struct sock_xprt, xprt);
> + struct tls_handshake_args args = {
> + .ta_sock = lower_transport->sock,
> + .ta_done = xs_tls_handshake_done,
> + .ta_data = xprt_get(lower_xprt),
> + .ta_peername = lower_xprt->servername,
This part isn't compiling for me on v6.4-rc2:
net/sunrpc/xprtsock.c:2538:4: error: field designator 'ta_peername'
does not refer to any field in type 'struct tls_handshake_args'
.ta_peername = lower_xprt->servername,
^
1 error generated.
Am I missing a patch, or did this struct get changed somewhere along the line?
Anna
> + };
> + struct sock *sk = lower_transport->inet;
> + int rc;
>
> - rpc_shutdown_client(clnt);
> + init_completion(&lower_transport->handshake_done);
> + set_bit(XPRT_SOCK_IGNORE_RECV, &lower_transport->sock_state);
>
> -out_unlock:
> - return;
> + lower_transport->xprt_err = -ETIMEDOUT;
> + switch (xprtsec->policy) {
> + case RPC_XPRTSEC_TLS_ANON:
> + rc = tls_client_hello_anon(&args, GFP_KERNEL);
> + if (rc)
> + goto out_put_xprt;
> + break;
> + case RPC_XPRTSEC_TLS_X509:
> + args.ta_my_cert = xprtsec->cert_serial;
> + args.ta_my_privkey = xprtsec->privkey_serial;
> + rc = tls_client_hello_x509(&args, GFP_KERNEL);
> + if (rc)
> + goto out_put_xprt;
> + break;
> + default:
> + rc = -EACCES;
> + goto out_put_xprt;
> + }
> +
> + rc = wait_for_completion_interruptible_timeout(&lower_transport->handshake_done,
> + XS_TLS_HANDSHAKE_TO);
> + if (rc <= 0) {
> + if (!tls_handshake_cancel(sk)) {
> + if (rc == 0)
> + rc = -ETIMEDOUT;
> + goto out_put_xprt;
> + }
> + }
> +
> + rc = lower_transport->xprt_err;
> +
> +out:
> + xs_stream_reset_connect(lower_transport);
> + clear_bit(XPRT_SOCK_IGNORE_RECV, &lower_transport->sock_state);
> + return rc;
> +
> +out_put_xprt:
> + xprt_put(lower_xprt);
> + goto out;
> }
>
> -static void xs_set_transport_clnt(struct rpc_clnt *clnt, struct rpc_xprt *xprt)
> +/**
> + * xs_tls_connect - establish a TLS session on a socket
> + * @work: queued work item
> + *
> + * For RPC-with-TLS, there is a two-stage connection process.
> + *
> + * The "upper-layer xprt" is visible to the RPC consumer. Once it has
> + * been marked connected, the consumer knows that a TCP connection and
> + * a TLS session have been established.
> + *
> + * A "lower-layer xprt", created in this function, handles the mechanics
> + * of connecting the TCP socket, performing the RPC_AUTH_TLS probe, and
> + * then driving the TLS handshake. Once all that is complete, the upper
> + * layer xprt is marked connected.
> + */
> +static void xs_tls_connect(struct work_struct *work)
> {
> - struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
> + struct sock_xprt *upper_transport =
> + container_of(work, struct sock_xprt, connect_worker.work);
> + struct rpc_clnt *upper_clnt = upper_transport->clnt;
> + struct rpc_xprt *upper_xprt = &upper_transport->xprt;
> struct rpc_create_args args = {
> - .net = xprt->xprt_net,
> - .protocol = xprt->prot,
> - .address = (struct sockaddr *)&xprt->addr,
> - .addrsize = xprt->addrlen,
> - .timeout = clnt->cl_timeout,
> - .servername = xprt->servername,
> - .nodename = clnt->cl_nodename,
> - .program = clnt->cl_program,
> - .prognumber = clnt->cl_prog,
> - .version = clnt->cl_vers,
> + .net = upper_xprt->xprt_net,
> + .protocol = upper_xprt->prot,
> + .address = (struct sockaddr *)&upper_xprt->addr,
> + .addrsize = upper_xprt->addrlen,
> + .timeout = upper_clnt->cl_timeout,
> + .servername = upper_xprt->servername,
> + .nodename = upper_clnt->cl_nodename,
> + .program = upper_clnt->cl_program,
> + .prognumber = upper_clnt->cl_prog,
> + .version = upper_clnt->cl_vers,
> .authflavor = RPC_AUTH_TLS,
> - .cred = clnt->cl_cred,
> + .cred = upper_clnt->cl_cred,
> .xprtsec = {
> .policy = RPC_XPRTSEC_NONE,
> },
> - .flags = RPC_CLNT_CREATE_NOPING,
> };
> + unsigned int pflags = current->flags;
> + struct rpc_clnt *lower_clnt;
> + struct rpc_xprt *lower_xprt;
> + int status;
>
> - switch (xprt->xprtsec.policy) {
> - case RPC_XPRTSEC_TLS_ANON:
> - case RPC_XPRTSEC_TLS_X509:
> - transport->clnt = rpc_create(&args);
> - break;
> - default:
> - transport->clnt = ERR_PTR(-ENOTCONN);
> + if (atomic_read(&upper_xprt->swapper))
> + current->flags |= PF_MEMALLOC;
> +
> + xs_stream_start_connect(upper_transport);
> +
> + /* This implicitly sends an RPC_AUTH_TLS probe */
> + lower_clnt = rpc_create(&args);
> + if (IS_ERR(lower_clnt)) {
> + clear_bit(XPRT_SOCK_CONNECTING, &upper_transport->sock_state);
> + xprt_clear_connecting(upper_xprt);
> + xprt_wake_pending_tasks(upper_xprt, PTR_ERR(lower_clnt));
> + smp_mb__before_atomic();
> + xs_run_error_worker(upper_transport, XPRT_SOCK_WAKE_PENDING);
> + goto out_unlock;
> }
> +
> + /* RPC_AUTH_TLS probe was successful. Try a TLS handshake on
> + * the lower xprt.
> + */
> + rcu_read_lock();
> + lower_xprt = rcu_dereference(lower_clnt->cl_xprt);
> + rcu_read_unlock();
> + status = xs_tls_handshake_sync(lower_xprt, &upper_xprt->xprtsec);
> + if (status)
> + goto out_close;
> +
> + status = xs_tls_finish_connecting(lower_xprt, upper_transport);
> + if (status)
> + goto out_close;
> +
> + trace_rpc_socket_connect(upper_xprt, upper_transport->sock, 0);
> + if (!xprt_test_and_set_connected(upper_xprt)) {
> + upper_xprt->connect_cookie++;
> + clear_bit(XPRT_SOCK_CONNECTING, &upper_transport->sock_state);
> + xprt_clear_connecting(upper_xprt);
> +
> + upper_xprt->stat.connect_count++;
> + upper_xprt->stat.connect_time += (long)jiffies -
> + upper_xprt->stat.connect_start;
> + xs_run_error_worker(upper_transport, XPRT_SOCK_WAKE_PENDING);
> + }
> + rpc_shutdown_client(lower_clnt);
> +
> +out_unlock:
> + current_restore_flags(pflags, PF_MEMALLOC);
> + upper_transport->clnt = NULL;
> + xprt_unlock_connect(upper_xprt, upper_transport);
> + return;
> +
> +out_close:
> + rpc_shutdown_client(lower_clnt);
> +
> + /* xprt_force_disconnect() wakes tasks with a fixed tk_status code.
> + * Wake them first here to ensure they get our tk_status code.
> + */
> + xprt_wake_pending_tasks(upper_xprt, status);
> + xs_tcp_force_close(upper_xprt);
> + xprt_clear_connecting(upper_xprt);
> + goto out_unlock;
> }
>
> /**
> @@ -2498,8 +2714,7 @@ static void xs_connect(struct rpc_xprt *xprt, struct rpc_task *task)
> } else
> dprintk("RPC: xs_connect scheduled xprt %p\n", xprt);
>
> - xs_set_transport_clnt(task->tk_client, xprt);
> -
> + transport->clnt = task->tk_client;
> queue_delayed_work(xprtiod_workqueue,
> &transport->connect_worker,
> delay);
>
>
