Re: [PATCH RFC 4/5] SUNRPC: Support TLS handshake in the server-side TCP socket code

[Chuck Lever III <chuck.lever@oracle.com>]

> On Mar 21, 2023, at 10:56 AM, Jeff Layton <jlayton@kernel.org> wrote:
> 
> On Tue, 2023-03-21 at 14:03 +0000, Chuck Lever III wrote:
>> 
>>> On Mar 21, 2023, at 7:43 AM, Jeff Layton <jlayton@kernel.org> wrote:
>>> 
>>> On Mon, 2023-03-20 at 10:24 -0400, Chuck Lever wrote:
>>>> From: Chuck Lever <chuck.lever@oracle.com>
>>>> 
>>>> This patch adds opportunitistic RPC-with-TLS to the Linux in-kernel
>>>> NFS server. If the client requests RPC-with-TLS and the user space
>>>> handshake agent is running, the server will set up a TLS session.
>>>> 
>>>> There are no policy settings yet. For example, the server cannot
>>>> yet require the use of RPC-with-TLS to access its data.
>>>> 
>>>> Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
>>>> ---
>>>> include/linux/sunrpc/svc_xprt.h |    5 ++
>>>> include/linux/sunrpc/svcsock.h  |    2 +
>>>> include/trace/events/sunrpc.h   |   16 ++++++-
>>>> net/sunrpc/svc_xprt.c           |    5 ++
>>>> net/sunrpc/svcauth_unix.c       |   11 ++++-
>>>> net/sunrpc/svcsock.c            |   91 +++++++++++++++++++++++++++++++++++++++
>>>> 6 files changed, 125 insertions(+), 5 deletions(-)
>>>> 
>>>> diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
>>>> index 775368802762..867479204840 100644
>>>> --- a/include/linux/sunrpc/svc_xprt.h
>>>> +++ b/include/linux/sunrpc/svc_xprt.h
>>>> @@ -27,7 +27,7 @@ struct svc_xprt_ops {
>>>> 	void		(*xpo_detach)(struct svc_xprt *);
>>>> 	void		(*xpo_free)(struct svc_xprt *);
>>>> 	void		(*xpo_kill_temp_xprt)(struct svc_xprt *);
>>>> -	void		(*xpo_start_tls)(struct svc_xprt *);
>>>> +	void		(*xpo_handshake)(struct svc_xprt *xprt);
>>>> };
>>>> 
>>>> struct svc_xprt_class {
>>>> @@ -70,6 +70,9 @@ struct svc_xprt {
>>>> #define XPT_LOCAL	12		/* connection from loopback interface */
>>>> #define XPT_KILL_TEMP   13		/* call xpo_kill_temp_xprt before closing */
>>>> #define XPT_CONG_CTRL	14		/* has congestion control */
>>>> +#define XPT_HANDSHAKE	15		/* xprt requests a handshake */
>>>> +#define XPT_TLS_SESSION	16		/* transport-layer security established */
>>>> +#define XPT_PEER_AUTH	17		/* peer has been authenticated */
>>>> 
>>>> 	struct svc_serv		*xpt_server;	/* service for transport */
>>>> 	atomic_t    	    	xpt_reserved;	/* space on outq that is rsvd */
>>>> diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
>>>> index bcc555c7ae9c..1175e1c38bac 100644
>>>> --- a/include/linux/sunrpc/svcsock.h
>>>> +++ b/include/linux/sunrpc/svcsock.h
>>>> @@ -38,6 +38,8 @@ struct svc_sock {
>>>> 	/* Number of queued send requests */
>>>> 	atomic_t		sk_sendqlen;
>>>> 
>>>> +	struct completion	sk_handshake_done;
>>>> +
>>>> 	struct page *		sk_pages[RPCSVC_MAXPAGES];	/* received data */
>>>> };
>>>> 
>>>> diff --git a/include/trace/events/sunrpc.h b/include/trace/events/sunrpc.h
>>>> index cf286a0a17d0..2667a8db4811 100644
>>>> --- a/include/trace/events/sunrpc.h
>>>> +++ b/include/trace/events/sunrpc.h
>>>> @@ -1948,7 +1948,10 @@ TRACE_EVENT(svc_stats_latency,
>>>> 		{ BIT(XPT_CACHE_AUTH),		"CACHE_AUTH" },		\
>>>> 		{ BIT(XPT_LOCAL),		"LOCAL" },		\
>>>> 		{ BIT(XPT_KILL_TEMP),		"KILL_TEMP" },		\
>>>> -		{ BIT(XPT_CONG_CTRL),		"CONG_CTRL" })
>>>> +		{ BIT(XPT_CONG_CTRL),		"CONG_CTRL" },		\
>>>> +		{ BIT(XPT_HANDSHAKE),		"HANDSHAKE" },		\
>>>> +		{ BIT(XPT_TLS_SESSION),		"TLS_SESSION" },	\
>>>> +		{ BIT(XPT_PEER_AUTH),		"PEER_AUTH" })
>>>> 
>>>> TRACE_EVENT(svc_xprt_create_err,
>>>> 	TP_PROTO(
>>>> @@ -2081,6 +2084,17 @@ DEFINE_SVC_XPRT_EVENT(close);
>>>> DEFINE_SVC_XPRT_EVENT(detach);
>>>> DEFINE_SVC_XPRT_EVENT(free);
>>>> 
>>>> +#define DEFINE_SVC_TLS_EVENT(name) \
>>>> +	DEFINE_EVENT(svc_xprt_event, svc_tls_##name, \
>>>> +		TP_PROTO(const struct svc_xprt *xprt), \
>>>> +		TP_ARGS(xprt))
>>>> +
>>>> +DEFINE_SVC_TLS_EVENT(start);
>>>> +DEFINE_SVC_TLS_EVENT(upcall);
>>>> +DEFINE_SVC_TLS_EVENT(unavailable);
>>>> +DEFINE_SVC_TLS_EVENT(not_started);
>>>> +DEFINE_SVC_TLS_EVENT(timed_out);
>>>> +
>>>> TRACE_EVENT(svc_xprt_accept,
>>>> 	TP_PROTO(
>>>> 		const struct svc_xprt *xprt,
>>>> diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
>>>> index ba629297da4e..b68c04dbf876 100644
>>>> --- a/net/sunrpc/svc_xprt.c
>>>> +++ b/net/sunrpc/svc_xprt.c
>>>> @@ -427,7 +427,7 @@ static bool svc_xprt_ready(struct svc_xprt *xprt)
>>>> 
>>>> 	if (xpt_flags & BIT(XPT_BUSY))
>>>> 		return false;
>>>> -	if (xpt_flags & (BIT(XPT_CONN) | BIT(XPT_CLOSE)))
>>>> +	if (xpt_flags & (BIT(XPT_CONN) | BIT(XPT_CLOSE) | BIT(XPT_HANDSHAKE)))
>>>> 		return true;
>>>> 	if (xpt_flags & (BIT(XPT_DATA) | BIT(XPT_DEFERRED))) {
>>>> 		if (xprt->xpt_ops->xpo_has_wspace(xprt) &&
>>>> @@ -829,6 +829,9 @@ static int svc_handle_xprt(struct svc_rqst *rqstp, struct svc_xprt *xprt)
>>>> 			module_put(xprt->xpt_class->xcl_owner);
>>>> 		}
>>>> 		svc_xprt_received(xprt);
>>>> +	} else if (test_bit(XPT_HANDSHAKE, &xprt->xpt_flags)) {
>>>> +		xprt->xpt_ops->xpo_handshake(xprt);
>>>> +		svc_xprt_received(xprt);
>>>> 	} else if (svc_xprt_reserve_slot(rqstp, xprt)) {
>>>> 		/* XPT_DATA|XPT_DEFERRED case: */
>>>> 		dprintk("svc: server %p, pool %u, transport %p, inuse=%d\n",
>>>> diff --git a/net/sunrpc/svcauth_unix.c b/net/sunrpc/svcauth_unix.c
>>>> index 983c5891cb56..374995201df4 100644
>>>> --- a/net/sunrpc/svcauth_unix.c
>>>> +++ b/net/sunrpc/svcauth_unix.c
>>>> @@ -17,8 +17,9 @@
>>>> #include <net/ipv6.h>
>>>> #include <linux/kernel.h>
>>>> #include <linux/user_namespace.h>
>>>> -#define RPCDBG_FACILITY	RPCDBG_AUTH
>>>> +#include <trace/events/sunrpc.h>
>>>> 
>>>> +#define RPCDBG_FACILITY	RPCDBG_AUTH
>>>> 
>>>> #include "netns.h"
>>>> 
>>>> @@ -823,6 +824,7 @@ svcauth_tls_accept(struct svc_rqst *rqstp)
>>>> {
>>>> 	struct xdr_stream *xdr = &rqstp->rq_arg_stream;
>>>> 	struct svc_cred	*cred = &rqstp->rq_cred;
>>>> +	struct svc_xprt *xprt = rqstp->rq_xprt;
>>>> 	u32 flavor, len;
>>>> 	void *body;
>>>> 	__be32 *p;
>>>> @@ -856,14 +858,19 @@ svcauth_tls_accept(struct svc_rqst *rqstp)
>>>> 	if (cred->cr_group_info == NULL)
>>>> 		return SVC_CLOSE;
>>>> 
>>>> -	if (rqstp->rq_xprt->xpt_ops->xpo_start_tls) {
>>>> +	if (xprt->xpt_ops->xpo_handshake) {
>>>> 		p = xdr_reserve_space(&rqstp->rq_res_stream, XDR_UNIT * 2 + 8);
>>>> 		if (!p)
>>>> 			return SVC_CLOSE;
>>>> +		trace_svc_tls_start(xprt);
>>>> 		*p++ = rpc_auth_null;
>>>> 		*p++ = cpu_to_be32(8);
>>>> 		memcpy(p, "STARTTLS", 8);
>>>> +
>>>> +		set_bit(XPT_HANDSHAKE, &xprt->xpt_flags);
>>>> +		svc_xprt_enqueue(xprt);
>>>> 	} else {
>>>> +		trace_svc_tls_unavailable(xprt);
>>>> 		if (xdr_stream_encode_opaque_auth(&rqstp->rq_res_stream,
>>>> 						  RPC_AUTH_NULL, NULL, 0) < 0)
>>>> 			return SVC_CLOSE;
>>>> diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
>>>> index b6df73cb706a..16ba8d6ab20e 100644
>>>> --- a/net/sunrpc/svcsock.c
>>>> +++ b/net/sunrpc/svcsock.c
>>>> @@ -44,9 +44,11 @@
>>>> #include <net/tcp.h>
>>>> #include <net/tcp_states.h>
>>>> #include <net/tls.h>
>>>> +#include <net/handshake.h>
>>>> #include <linux/uaccess.h>
>>>> #include <linux/highmem.h>
>>>> #include <asm/ioctls.h>
>>>> +#include <linux/key.h>
>>>> 
>>>> #include <linux/sunrpc/types.h>
>>>> #include <linux/sunrpc/clnt.h>
>>>> @@ -64,6 +66,7 @@
>>>> 
>>>> #define RPCDBG_FACILITY	RPCDBG_SVCXPRT
>>>> 
>>>> +#define SVC_HANDSHAKE_TO	(20U * HZ)
>>>> 
>>>> static struct svc_sock *svc_setup_socket(struct svc_serv *, struct socket *,
>>>> 					 int flags);
>>>> @@ -360,6 +363,8 @@ static void svc_data_ready(struct sock *sk)
>>>> 		rmb();
>>>> 		svsk->sk_odata(sk);
>>>> 		trace_svcsock_data_ready(&svsk->sk_xprt, 0);
>>>> +		if (test_bit(XPT_HANDSHAKE, &svsk->sk_xprt.xpt_flags))
>>>> +			return;
>>>> 		if (!test_and_set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags))
>>>> 			svc_xprt_enqueue(&svsk->sk_xprt);
>>>> 	}
>>>> @@ -397,6 +402,89 @@ static void svc_tcp_kill_temp_xprt(struct svc_xprt *xprt)
>>>> 	sock_no_linger(svsk->sk_sock->sk);
>>>> }
>>>> 
>>>> +/**
>>>> + * svc_tcp_handshake_done - Handshake completion handler
>>>> + * @data: address of xprt to wake
>>>> + * @status: status of handshake
>>>> + * @peerid: serial number of key containing the remote peer's identity
>>>> + *
>>>> + * If a security policy is specified as an export option, we don't
>>>> + * have a specific export here to check. So we set a "TLS session
>>>> + * is present" flag on the xprt and let an upper layer enforce local
>>>> + * security policy.
>>>> + */
>>>> +static void svc_tcp_handshake_done(void *data, int status, key_serial_t peerid)
>>>> +{
>>>> +	struct svc_xprt *xprt = data;
>>>> +	struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt);
>>>> +
>>>> +	if (!status) {
>>>> +		if (peerid != TLS_NO_PEERID)
>>>> +			set_bit(XPT_PEER_AUTH, &xprt->xpt_flags);
>>>> +		set_bit(XPT_TLS_SESSION, &xprt->xpt_flags);
>>>> +	}
>>>> +	clear_bit(XPT_HANDSHAKE, &xprt->xpt_flags);
>>>> +	complete_all(&svsk->sk_handshake_done);
>>>> +}
>>>> +
>>>> +/**
>>>> + * svc_tcp_handshake - Perform a transport-layer security handshake
>>>> + * @xprt: connected transport endpoint
>>>> + *
>>>> + */
>>>> +static void svc_tcp_handshake(struct svc_xprt *xprt)
>>>> +{
>>>> +	struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt);
>>>> +	struct tls_handshake_args args = {
>>>> +		.ta_sock	= svsk->sk_sock,
>>>> +		.ta_done	= svc_tcp_handshake_done,
>>>> +		.ta_data	= xprt,
>>>> +	};
>>>> +	int ret;
>>>> +
>>>> +	trace_svc_tls_upcall(xprt);
>>>> +
>>>> +	clear_bit(XPT_TLS_SESSION, &xprt->xpt_flags);
>>>> +	init_completion(&svsk->sk_handshake_done);
>>>> +	smp_wmb();
>>>> +
>>>> +	ret = tls_server_hello_x509(&args, GFP_KERNEL);
>>>> +	if (ret) {
>>>> +		trace_svc_tls_not_started(xprt);
>>>> +		goto out_failed;
>>>> +	}
>>>> +
>>>> +	ret = wait_for_completion_interruptible_timeout(&svsk->sk_handshake_done,
>>>> +							SVC_HANDSHAKE_TO);
>>> 
>>> Just curious: is this 20s timeout mandated by the spec?
>> 
>> The spec doesn't mandate a timeout. I simply wanted
>> to guarantee forward progress.
>> 
>> 
>>> It seems like a long time to block a kernel thread if so.
>> 
>> It's about the same as the client side timeout, fwiw.
>> 
>> 
>>> Do we need to be concerned
>>> with DoS attacks here? Could a client initiate handshakes and then stop
>>> communicating, forcing the server to tie up threads with these 20s
>>> waits?
>> 
>> I think a malicious client can do all kinds of similar things
>> already. Do you have a particular timeout value in mind, or
>> is there some other mechanism we can use to better bullet-
>> proof this aspect of the handshake? I'm open to suggestion.
>> 
> 
> I don't have any suggestions, just trying to speculate about ways this
> could break. The only alternative I could see would be to defer the
> connection somehow until the reply comes in so that the thread can do
> other stuff in the meantime.

The server has a deferral mechanism, funnily enough. ;-) But
we could also just use a system workqueue.

Note that the handshake upcall mechanism itself has a limit
on the concurrent number of pending handshakes it will allow.
That might be enough to prevent a DoS.


> That's something we can potentially add
> later if we decide it's necessary though.

I shortened the timeout and added a comment suggesting this
as future work.


>>>> +	if (ret <= 0) {
>>>> +		if (tls_handshake_cancel(svsk->sk_sock)) {
>>>> +			trace_svc_tls_timed_out(xprt);
>>>> +			goto out_close;
>>>> +		}
>>>> +	}
>>>> +
>>>> +	if (!test_bit(XPT_TLS_SESSION, &xprt->xpt_flags)) {
>>>> +		trace_svc_tls_unavailable(xprt);
>>>> +		goto out_close;
>>>> +	}
>>>> +
>>>> +	/*
>>>> +	 * Mark the transport ready in case the remote sent RPC
>>>> +	 * traffic before the kernel received the handshake
>>>> +	 * completion downcall.
>>>> +	 */
>>>> +	set_bit(XPT_DATA, &xprt->xpt_flags);
>>>> +	svc_xprt_enqueue(xprt);
>>>> +	return;
>>>> +
>>>> +out_close:
>>>> +	set_bit(XPT_CLOSE, &xprt->xpt_flags);
>>>> +out_failed:
>>>> +	clear_bit(XPT_HANDSHAKE, &xprt->xpt_flags);
>>>> +	set_bit(XPT_DATA, &xprt->xpt_flags);
>>>> +	svc_xprt_enqueue(xprt);
>>>> +}
>>>> +
>>>> /*
>>>> * See net/ipv6/ip_sockglue.c : ip_cmsg_recv_pktinfo
>>>> */
>>>> @@ -1260,6 +1348,7 @@ static const struct svc_xprt_ops svc_tcp_ops = {
>>>> 	.xpo_has_wspace = svc_tcp_has_wspace,
>>>> 	.xpo_accept = svc_tcp_accept,
>>>> 	.xpo_kill_temp_xprt = svc_tcp_kill_temp_xprt,
>>>> +	.xpo_handshake = svc_tcp_handshake,
>>>> };
>>>> 
>>>> static struct svc_xprt_class svc_tcp_class = {
>>>> @@ -1584,6 +1673,8 @@ static void svc_sock_free(struct svc_xprt *xprt)
>>>> {
>>>> 	struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt);
>>>> 
>>>> +	tls_handshake_cancel(svsk->sk_sock);
>>>> +
>>>> 	if (svsk->sk_sock->file)
>>>> 		sockfd_put(svsk->sk_sock);
>>>> 	else
>>>> 
>>>> 
>>> 
>>> -- 
>>> Jeff Layton <jlayton@kernel.org>
>> 
>> --
>> Chuck Lever
>> 
>> 
> 
> -- 
> Jeff Layton <jlayton@kernel.org>

--
Chuck Lever