[PATCH RFC 2/5] SUNRPC: Recognize control messages in server-side TCP socket code

[Chuck Lever <cel@kernel.org>]

From: Chuck Lever <chuck.lever@oracle.com>

To support kTLS, the server-side TCP socket code needs to watch for
CMSGs, just like on the client side.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
 include/trace/events/sunrpc.h |   26 ++++++++++++++++++++++
 net/sunrpc/svcsock.c          |   49 +++++++++++++++++++++++++++++++++++++++--
 2 files changed, 73 insertions(+), 2 deletions(-)

diff --git a/include/trace/events/sunrpc.h b/include/trace/events/sunrpc.h
index 280a31be3806..cf286a0a17d0 100644
--- a/include/trace/events/sunrpc.h
+++ b/include/trace/events/sunrpc.h
@@ -2370,6 +2370,32 @@ DECLARE_EVENT_CLASS(svcsock_accept_class,
 DEFINE_ACCEPT_EVENT(accept);
 DEFINE_ACCEPT_EVENT(getpeername);
 
+TRACE_EVENT(svcsock_tls_ctype,
+	TP_PROTO(
+		const struct svc_xprt *xprt,
+		unsigned char ctype
+	),
+
+	TP_ARGS(xprt, ctype),
+
+	TP_STRUCT__entry(
+		SVC_XPRT_ENDPOINT_FIELDS(xprt)
+
+		__field(unsigned long, ctype)
+	),
+
+	TP_fast_assign(
+		SVC_XPRT_ENDPOINT_ASSIGNMENTS(xprt);
+
+		__entry->ctype = ctype;
+	),
+
+	TP_printk(SVC_XPRT_ENDPOINT_FORMAT " %s",
+		SVC_XPRT_ENDPOINT_VARARGS,
+		rpc_show_tls_content_type(__entry->ctype)
+	)
+);
+
 DECLARE_EVENT_CLASS(cache_event,
 	TP_PROTO(
 		const struct cache_detail *cd,
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 28deb4acc392..91a62bea6999 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -43,6 +43,7 @@
 #include <net/udp.h>
 #include <net/tcp.h>
 #include <net/tcp_states.h>
+#include <net/tls.h>
 #include <linux/uaccess.h>
 #include <linux/highmem.h>
 #include <asm/ioctls.h>
@@ -216,6 +217,50 @@ static int svc_one_sock_name(struct svc_sock *svsk, char *buf, int remaining)
 	return len;
 }
 
+static int
+svc_tcp_sock_process_cmsg(struct svc_sock *svsk, struct msghdr *msg,
+			  struct cmsghdr *cmsg, int ret)
+{
+	if (cmsg->cmsg_level == SOL_TLS &&
+	    cmsg->cmsg_type == TLS_GET_RECORD_TYPE) {
+		u8 content_type = *((u8 *)CMSG_DATA(cmsg));
+
+		trace_svcsock_tls_ctype(&svsk->sk_xprt, content_type);
+		switch (content_type) {
+		case TLS_RECORD_TYPE_DATA:
+			/* TLS sets EOR at the end of each application data
+			 * record, even though there might be more frames
+			 * waiting to be decrypted.
+			 */
+			msg->msg_flags &= ~MSG_EOR;
+			break;
+		case TLS_RECORD_TYPE_ALERT:
+			ret = -ENOTCONN;
+			break;
+		default:
+			ret = -EAGAIN;
+		}
+	}
+	return ret;
+}
+
+static int
+svc_tcp_sock_recv_cmsg(struct svc_sock *svsk, struct msghdr *msg)
+{
+	union {
+		struct cmsghdr	cmsg;
+		u8		buf[CMSG_SPACE(sizeof(u8))];
+	} u;
+	int ret;
+
+	msg->msg_control = &u;
+	msg->msg_controllen = sizeof(u);
+	ret = sock_recvmsg(svsk->sk_sock, msg, MSG_DONTWAIT);
+	if (unlikely(msg->msg_controllen != sizeof(u)))
+		ret = svc_tcp_sock_process_cmsg(svsk, msg, &u.cmsg, ret);
+	return ret;
+}
+
 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
 static void svc_flush_bvec(const struct bio_vec *bvec, size_t size, size_t seek)
 {
@@ -263,7 +308,7 @@ static ssize_t svc_tcp_read_msg(struct svc_rqst *rqstp, size_t buflen,
 		iov_iter_advance(&msg.msg_iter, seek);
 		buflen -= seek;
 	}
-	len = sock_recvmsg(svsk->sk_sock, &msg, MSG_DONTWAIT);
+	len = svc_tcp_sock_recv_cmsg(svsk, &msg);
 	if (len > 0)
 		svc_flush_bvec(bvec, len, seek);
 
@@ -877,7 +922,7 @@ static ssize_t svc_tcp_read_marker(struct svc_sock *svsk,
 		iov.iov_base = ((char *)&svsk->sk_marker) + svsk->sk_tcplen;
 		iov.iov_len  = want;
 		iov_iter_kvec(&msg.msg_iter, ITER_DEST, &iov, 1, want);
-		len = sock_recvmsg(svsk->sk_sock, &msg, MSG_DONTWAIT);
+		len = svc_tcp_sock_recv_cmsg(svsk, &msg);
 		if (len < 0)
 			return len;
 		svsk->sk_tcplen += len;