rds: zerocopy Tx support.

If the MSG_ZEROCOPY flag is specified with rds_sendmsg(), and,
if the SO_ZEROCOPY socket option has been set on the PF_RDS socket,
application pages sent down with rds_sendmsg() are pinned.

The pinning uses the accounting infrastructure added by
Commit a91dbff551a6 ("sock: ulimit on MSG_ZEROCOPY pages")

The payload bytes in the message may not be modified for the
duration that the message has been pinned. A multi-threaded
application using this infrastructure may thus need to be notified
about send-completion so that it can free/reuse the buffers
passed to rds_sendmsg(). Notification of send-completion will
identify each message-buffer by a cookie that the application
must specify as ancillary data to rds_sendmsg().
The ancillary data in this case has cmsg_level == SOL_RDS
and cmsg_type == RDS_CMSG_ZCOPY_COOKIE.

Signed-off-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 38 insertions, 6 deletions

diff --git a/net/rds/send.c b/net/rds/send.c
index e8f3ff471b15..028ab598ac1b 100644
--- a/net/rds/send.c
+++ b/net/rds/send.c
@@ -875,12 +875,13 @@ out:
  * rds_message is getting to be quite complicated, and we'd like to allocate
  * it all in one go. This figures out how big it needs to be up front.
  */
-static int rds_rm_size(struct msghdr *msg, int data_len)
+static int rds_rm_size(struct msghdr *msg, int num_sgs)
 {
 	struct cmsghdr *cmsg;
 	int size = 0;
 	int cmsg_groups = 0;
 	int retval;
+	bool zcopy_cookie = false;
 
 	for_each_cmsghdr(cmsg, msg) {
 		if (!CMSG_OK(msg, cmsg))
@@ -899,6 +900,8 @@ static int rds_rm_size(struct msghdr *msg, int data_len)
 
 			break;
 
+		case RDS_CMSG_ZCOPY_COOKIE:
+			zcopy_cookie = true;
 		case RDS_CMSG_RDMA_DEST:
 		case RDS_CMSG_RDMA_MAP:
 			cmsg_groups |= 2;
@@ -919,7 +922,10 @@ static int rds_rm_size(struct msghdr *msg, int data_len)
 
 	}
 
-	size += ceil(data_len, PAGE_SIZE) * sizeof(struct scatterlist);
+	if ((msg->msg_flags & MSG_ZEROCOPY) && !zcopy_cookie)
+		return -EINVAL;
+
+	size += num_sgs * sizeof(struct scatterlist);
 
 	/* Ensure (DEST, MAP) are never used with (ARGS, ATOMIC) */
 	if (cmsg_groups == 3)
@@ -928,6 +934,18 @@ static int rds_rm_size(struct msghdr *msg, int data_len)
 	return size;
 }
 
+static int rds_cmsg_zcopy(struct rds_sock *rs, struct rds_message *rm,
+			  struct cmsghdr *cmsg)
+{
+	u32 *cookie;
+
+	if (cmsg->cmsg_len < CMSG_LEN(sizeof(*cookie)))
+		return -EINVAL;
+	cookie = CMSG_DATA(cmsg);
+	rm->data.op_mmp_znotifier->z_cookie = *cookie;
+	return 0;
+}
+
 static int rds_cmsg_send(struct rds_sock *rs, struct rds_message *rm,
 			 struct msghdr *msg, int *allocated_mr)
 {
@@ -970,6 +988,10 @@ static int rds_cmsg_send(struct rds_sock *rs, struct rds_message *rm,
 			ret = rds_cmsg_atomic(rs, rm, cmsg);
 			break;
 
+		case RDS_CMSG_ZCOPY_COOKIE:
+			ret = rds_cmsg_zcopy(rs, rm, cmsg);
+			break;
+
 		default:
 			return -EINVAL;
 		}
@@ -1040,10 +1062,13 @@ int rds_sendmsg(struct socket *sock, struct msghdr *msg, size_t payload_len)
 	long timeo = sock_sndtimeo(sk, nonblock);
 	struct rds_conn_path *cpath;
 	size_t total_payload_len = payload_len, rdma_payload_len = 0;
+	bool zcopy = ((msg->msg_flags & MSG_ZEROCOPY) &&
+		      sock_flag(rds_rs_to_sk(rs), SOCK_ZEROCOPY));
+	int num_sgs = ceil(payload_len, PAGE_SIZE);
 
 	/* Mirror Linux UDP mirror of BSD error message compatibility */
 	/* XXX: Perhaps MSG_MORE someday */
-	if (msg->msg_flags & ~(MSG_DONTWAIT | MSG_CMSG_COMPAT)) {
+	if (msg->msg_flags & ~(MSG_DONTWAIT | MSG_CMSG_COMPAT | MSG_ZEROCOPY)) {
 		ret = -EOPNOTSUPP;
 		goto out;
 	}
@@ -1087,8 +1112,15 @@ int rds_sendmsg(struct socket *sock, struct msghdr *msg, size_t payload_len)
 		goto out;
 	}
 
+	if (zcopy) {
+		if (rs->rs_transport->t_type != RDS_TRANS_TCP) {
+			ret = -EOPNOTSUPP;
+			goto out;
+		}
+		num_sgs = iov_iter_npages(&msg->msg_iter, INT_MAX);
+	}
 	/* size of rm including all sgs */
-	ret = rds_rm_size(msg, payload_len);
+	ret = rds_rm_size(msg, num_sgs);
 	if (ret < 0)
 		goto out;
 
@@ -1100,12 +1132,12 @@ int rds_sendmsg(struct socket *sock, struct msghdr *msg, size_t payload_len)
 
 	/* Attach data to the rm */
 	if (payload_len) {
-		rm->data.op_sg = rds_message_alloc_sgs(rm, ceil(payload_len, PAGE_SIZE));
+		rm->data.op_sg = rds_message_alloc_sgs(rm, num_sgs);
 		if (!rm->data.op_sg) {
 			ret = -ENOMEM;
 			goto out;
 		}
-		ret = rds_message_copy_from_user(rm, &msg->msg_iter);
+		ret = rds_message_copy_from_user(rm, &msg->msg_iter, zcopy);
 		if (ret)
 			goto out;
 	}