Initial import

47 files changed, 18162 insertions, 0 deletions

diff --git a/net/rds/Kconfig b/net/rds/Kconfig
new file mode 100644
index 000000000..f2c670ba7
--- /dev/null
+++ b/net/rds/Kconfig
@@ -0,0 +1,28 @@
+
+config RDS
+	tristate "The RDS Protocol"
+	depends on INET
+	---help---
+	  The RDS (Reliable Datagram Sockets) protocol provides reliable,
+	  sequenced delivery of datagrams over Infiniband, iWARP,
+	  or TCP.
+
+config RDS_RDMA
+	tristate "RDS over Infiniband and iWARP"
+	depends on RDS && INFINIBAND && INFINIBAND_ADDR_TRANS
+	---help---
+	  Allow RDS to use Infiniband and iWARP as a transport.
+	  This transport supports RDMA operations.
+
+config RDS_TCP
+	tristate "RDS over TCP"
+	depends on RDS
+	---help---
+	  Allow RDS to use TCP as a transport.
+	  This transport does not support RDMA operations.
+
+config RDS_DEBUG
+        bool "RDS debugging messages"
+	depends on RDS
+        default n
+
diff --git a/net/rds/Makefile b/net/rds/Makefile
new file mode 100644
index 000000000..56d3f6023
--- /dev/null
+++ b/net/rds/Makefile
@@ -0,0 +1,19 @@
+obj-$(CONFIG_RDS) += rds.o
+rds-y :=	af_rds.o bind.o cong.o connection.o info.o message.o   \
+			recv.o send.o stats.o sysctl.o threads.o transport.o \
+			loop.o page.o rdma.o
+
+obj-$(CONFIG_RDS_RDMA) += rds_rdma.o
+rds_rdma-y :=	rdma_transport.o \
+			ib.o ib_cm.o ib_recv.o ib_ring.o ib_send.o ib_stats.o \
+			ib_sysctl.o ib_rdma.o \
+			iw.o iw_cm.o iw_recv.o iw_ring.o iw_send.o iw_stats.o \
+			iw_sysctl.o iw_rdma.o
+
+
+obj-$(CONFIG_RDS_TCP) += rds_tcp.o
+rds_tcp-y :=		tcp.o tcp_connect.o tcp_listen.o tcp_recv.o \
+			tcp_send.o tcp_stats.o
+
+ccflags-$(CONFIG_RDS_DEBUG)	:=	-DDEBUG
+
diff --git a/net/rds/af_rds.c b/net/rds/af_rds.c
new file mode 100644
index 000000000..10443377f
--- /dev/null
+++ b/net/rds/af_rds.c
@@ -0,0 +1,599 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/module.h>
+#include <linux/errno.h>
+#include <linux/kernel.h>
+#include <linux/gfp.h>
+#include <linux/in.h>
+#include <linux/poll.h>
+#include <net/sock.h>
+
+#include "rds.h"
+
+char *rds_str_array(char **array, size_t elements, size_t index)
+{
+	if ((index < elements) && array[index])
+		return array[index];
+	else
+		return "unknown";
+}
+EXPORT_SYMBOL(rds_str_array);
+
+/* this is just used for stats gathering :/ */
+static DEFINE_SPINLOCK(rds_sock_lock);
+static unsigned long rds_sock_count;
+static LIST_HEAD(rds_sock_list);
+DECLARE_WAIT_QUEUE_HEAD(rds_poll_waitq);
+
+/*
+ * This is called as the final descriptor referencing this socket is closed.
+ * We have to unbind the socket so that another socket can be bound to the
+ * address it was using.
+ *
+ * We have to be careful about racing with the incoming path.  sock_orphan()
+ * sets SOCK_DEAD and we use that as an indicator to the rx path that new
+ * messages shouldn't be queued.
+ */
+static int rds_release(struct socket *sock)
+{
+	struct sock *sk = sock->sk;
+	struct rds_sock *rs;
+
+	if (!sk)
+		goto out;
+
+	rs = rds_sk_to_rs(sk);
+
+	sock_orphan(sk);
+	/* Note - rds_clear_recv_queue grabs rs_recv_lock, so
+	 * that ensures the recv path has completed messing
+	 * with the socket. */
+	rds_clear_recv_queue(rs);
+	rds_cong_remove_socket(rs);
+
+	/*
+	 * the binding lookup hash uses rcu, we need to
+	 * make sure we synchronize_rcu before we free our
+	 * entry
+	 */
+	rds_remove_bound(rs);
+	synchronize_rcu();
+
+	rds_send_drop_to(rs, NULL);
+	rds_rdma_drop_keys(rs);
+	rds_notify_queue_get(rs, NULL);
+
+	spin_lock_bh(&rds_sock_lock);
+	list_del_init(&rs->rs_item);
+	rds_sock_count--;
+	spin_unlock_bh(&rds_sock_lock);
+
+	rds_trans_put(rs->rs_transport);
+
+	sock->sk = NULL;
+	sock_put(sk);
+out:
+	return 0;
+}
+
+/*
+ * Careful not to race with rds_release -> sock_orphan which clears sk_sleep.
+ * _bh() isn't OK here, we're called from interrupt handlers.  It's probably OK
+ * to wake the waitqueue after sk_sleep is clear as we hold a sock ref, but
+ * this seems more conservative.
+ * NB - normally, one would use sk_callback_lock for this, but we can
+ * get here from interrupts, whereas the network code grabs sk_callback_lock
+ * with _lock_bh only - so relying on sk_callback_lock introduces livelocks.
+ */
+void rds_wake_sk_sleep(struct rds_sock *rs)
+{
+	unsigned long flags;
+
+	read_lock_irqsave(&rs->rs_recv_lock, flags);
+	__rds_wake_sk_sleep(rds_rs_to_sk(rs));
+	read_unlock_irqrestore(&rs->rs_recv_lock, flags);
+}
+
+static int rds_getname(struct socket *sock, struct sockaddr *uaddr,
+		       int *uaddr_len, int peer)
+{
+	struct sockaddr_in *sin = (struct sockaddr_in *)uaddr;
+	struct rds_sock *rs = rds_sk_to_rs(sock->sk);
+
+	memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
+
+	/* racey, don't care */
+	if (peer) {
+		if (!rs->rs_conn_addr)
+			return -ENOTCONN;
+
+		sin->sin_port = rs->rs_conn_port;
+		sin->sin_addr.s_addr = rs->rs_conn_addr;
+	} else {
+		sin->sin_port = rs->rs_bound_port;
+		sin->sin_addr.s_addr = rs->rs_bound_addr;
+	}
+
+	sin->sin_family = AF_INET;
+
+	*uaddr_len = sizeof(*sin);
+	return 0;
+}
+
+/*
+ * RDS' poll is without a doubt the least intuitive part of the interface,
+ * as POLLIN and POLLOUT do not behave entirely as you would expect from
+ * a network protocol.
+ *
+ * POLLIN is asserted if
+ *  -	there is data on the receive queue.
+ *  -	to signal that a previously congested destination may have become
+ *	uncongested
+ *  -	A notification has been queued to the socket (this can be a congestion
+ *	update, or a RDMA completion).
+ *
+ * POLLOUT is asserted if there is room on the send queue. This does not mean
+ * however, that the next sendmsg() call will succeed. If the application tries
+ * to send to a congested destination, the system call may still fail (and
+ * return ENOBUFS).
+ */
+static unsigned int rds_poll(struct file *file, struct socket *sock,
+			     poll_table *wait)
+{
+	struct sock *sk = sock->sk;
+	struct rds_sock *rs = rds_sk_to_rs(sk);
+	unsigned int mask = 0;
+	unsigned long flags;
+
+	poll_wait(file, sk_sleep(sk), wait);
+
+	if (rs->rs_seen_congestion)
+		poll_wait(file, &rds_poll_waitq, wait);
+
+	read_lock_irqsave(&rs->rs_recv_lock, flags);
+	if (!rs->rs_cong_monitor) {
+		/* When a congestion map was updated, we signal POLLIN for
+		 * "historical" reasons. Applications can also poll for
+		 * WRBAND instead. */
+		if (rds_cong_updated_since(&rs->rs_cong_track))
+			mask |= (POLLIN | POLLRDNORM | POLLWRBAND);
+	} else {
+		spin_lock(&rs->rs_lock);
+		if (rs->rs_cong_notify)
+			mask |= (POLLIN | POLLRDNORM);
+		spin_unlock(&rs->rs_lock);
+	}
+	if (!list_empty(&rs->rs_recv_queue) ||
+	    !list_empty(&rs->rs_notify_queue))
+		mask |= (POLLIN | POLLRDNORM);
+	if (rs->rs_snd_bytes < rds_sk_sndbuf(rs))
+		mask |= (POLLOUT | POLLWRNORM);
+	read_unlock_irqrestore(&rs->rs_recv_lock, flags);
+
+	/* clear state any time we wake a seen-congested socket */
+	if (mask)
+		rs->rs_seen_congestion = 0;
+
+	return mask;
+}
+
+static int rds_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
+{
+	return -ENOIOCTLCMD;
+}
+
+static int rds_cancel_sent_to(struct rds_sock *rs, char __user *optval,
+			      int len)
+{
+	struct sockaddr_in sin;
+	int ret = 0;
+
+	/* racing with another thread binding seems ok here */
+	if (rs->rs_bound_addr == 0) {
+		ret = -ENOTCONN; /* XXX not a great errno */
+		goto out;
+	}
+
+	if (len < sizeof(struct sockaddr_in)) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	if (copy_from_user(&sin, optval, sizeof(sin))) {
+		ret = -EFAULT;
+		goto out;
+	}
+
+	rds_send_drop_to(rs, &sin);
+out:
+	return ret;
+}
+
+static int rds_set_bool_option(unsigned char *optvar, char __user *optval,
+			       int optlen)
+{
+	int value;
+
+	if (optlen < sizeof(int))
+		return -EINVAL;
+	if (get_user(value, (int __user *) optval))
+		return -EFAULT;
+	*optvar = !!value;
+	return 0;
+}
+
+static int rds_cong_monitor(struct rds_sock *rs, char __user *optval,
+			    int optlen)
+{
+	int ret;
+
+	ret = rds_set_bool_option(&rs->rs_cong_monitor, optval, optlen);
+	if (ret == 0) {
+		if (rs->rs_cong_monitor) {
+			rds_cong_add_socket(rs);
+		} else {
+			rds_cong_remove_socket(rs);
+			rs->rs_cong_mask = 0;
+			rs->rs_cong_notify = 0;
+		}
+	}
+	return ret;
+}
+
+static int rds_setsockopt(struct socket *sock, int level, int optname,
+			  char __user *optval, unsigned int optlen)
+{
+	struct rds_sock *rs = rds_sk_to_rs(sock->sk);
+	int ret;
+
+	if (level != SOL_RDS) {
+		ret = -ENOPROTOOPT;
+		goto out;
+	}
+
+	switch (optname) {
+	case RDS_CANCEL_SENT_TO:
+		ret = rds_cancel_sent_to(rs, optval, optlen);
+		break;
+	case RDS_GET_MR:
+		ret = rds_get_mr(rs, optval, optlen);
+		break;
+	case RDS_GET_MR_FOR_DEST:
+		ret = rds_get_mr_for_dest(rs, optval, optlen);
+		break;
+	case RDS_FREE_MR:
+		ret = rds_free_mr(rs, optval, optlen);
+		break;
+	case RDS_RECVERR:
+		ret = rds_set_bool_option(&rs->rs_recverr, optval, optlen);
+		break;
+	case RDS_CONG_MONITOR:
+		ret = rds_cong_monitor(rs, optval, optlen);
+		break;
+	default:
+		ret = -ENOPROTOOPT;
+	}
+out:
+	return ret;
+}
+
+static int rds_getsockopt(struct socket *sock, int level, int optname,
+			  char __user *optval, int __user *optlen)
+{
+	struct rds_sock *rs = rds_sk_to_rs(sock->sk);
+	int ret = -ENOPROTOOPT, len;
+
+	if (level != SOL_RDS)
+		goto out;
+
+	if (get_user(len, optlen)) {
+		ret = -EFAULT;
+		goto out;
+	}
+
+	switch (optname) {
+	case RDS_INFO_FIRST ... RDS_INFO_LAST:
+		ret = rds_info_getsockopt(sock, optname, optval,
+					  optlen);
+		break;
+
+	case RDS_RECVERR:
+		if (len < sizeof(int))
+			ret = -EINVAL;
+		else
+		if (put_user(rs->rs_recverr, (int __user *) optval) ||
+		    put_user(sizeof(int), optlen))
+			ret = -EFAULT;
+		else
+			ret = 0;
+		break;
+	default:
+		break;
+	}
+
+out:
+	return ret;
+
+}
+
+static int rds_connect(struct socket *sock, struct sockaddr *uaddr,
+		       int addr_len, int flags)
+{
+	struct sock *sk = sock->sk;
+	struct sockaddr_in *sin = (struct sockaddr_in *)uaddr;
+	struct rds_sock *rs = rds_sk_to_rs(sk);
+	int ret = 0;
+
+	lock_sock(sk);
+
+	if (addr_len != sizeof(struct sockaddr_in)) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	if (sin->sin_family != AF_INET) {
+		ret = -EAFNOSUPPORT;
+		goto out;
+	}
+
+	if (sin->sin_addr.s_addr == htonl(INADDR_ANY)) {
+		ret = -EDESTADDRREQ;
+		goto out;
+	}
+
+	rs->rs_conn_addr = sin->sin_addr.s_addr;
+	rs->rs_conn_port = sin->sin_port;
+
+out:
+	release_sock(sk);
+	return ret;
+}
+
+static struct proto rds_proto = {
+	.name	  = "RDS",
+	.owner	  = THIS_MODULE,
+	.obj_size = sizeof(struct rds_sock),
+};
+
+static const struct proto_ops rds_proto_ops = {
+	.family =	AF_RDS,
+	.owner =	THIS_MODULE,
+	.release =	rds_release,
+	.bind =		rds_bind,
+	.connect =	rds_connect,
+	.socketpair =	sock_no_socketpair,
+	.accept =	sock_no_accept,
+	.getname =	rds_getname,
+	.poll =		rds_poll,
+	.ioctl =	rds_ioctl,
+	.listen =	sock_no_listen,
+	.shutdown =	sock_no_shutdown,
+	.setsockopt =	rds_setsockopt,
+	.getsockopt =	rds_getsockopt,
+	.sendmsg =	rds_sendmsg,
+	.recvmsg =	rds_recvmsg,
+	.mmap =		sock_no_mmap,
+	.sendpage =	sock_no_sendpage,
+};
+
+static int __rds_create(struct socket *sock, struct sock *sk, int protocol)
+{
+	struct rds_sock *rs;
+
+	sock_init_data(sock, sk);
+	sock->ops		= &rds_proto_ops;
+	sk->sk_protocol		= protocol;
+
+	rs = rds_sk_to_rs(sk);
+	spin_lock_init(&rs->rs_lock);
+	rwlock_init(&rs->rs_recv_lock);
+	INIT_LIST_HEAD(&rs->rs_send_queue);
+	INIT_LIST_HEAD(&rs->rs_recv_queue);
+	INIT_LIST_HEAD(&rs->rs_notify_queue);
+	INIT_LIST_HEAD(&rs->rs_cong_list);
+	spin_lock_init(&rs->rs_rdma_lock);
+	rs->rs_rdma_keys = RB_ROOT;
+
+	spin_lock_bh(&rds_sock_lock);
+	list_add_tail(&rs->rs_item, &rds_sock_list);
+	rds_sock_count++;
+	spin_unlock_bh(&rds_sock_lock);
+
+	return 0;
+}
+
+static int rds_create(struct net *net, struct socket *sock, int protocol,
+		      int kern)
+{
+	struct sock *sk;
+
+	if (sock->type != SOCK_SEQPACKET || protocol)
+		return -ESOCKTNOSUPPORT;
+
+	sk = sk_alloc(net, AF_RDS, GFP_ATOMIC, &rds_proto);
+	if (!sk)
+		return -ENOMEM;
+
+	return __rds_create(sock, sk, protocol);
+}
+
+void rds_sock_addref(struct rds_sock *rs)
+{
+	sock_hold(rds_rs_to_sk(rs));
+}
+
+void rds_sock_put(struct rds_sock *rs)
+{
+	sock_put(rds_rs_to_sk(rs));
+}
+
+static const struct net_proto_family rds_family_ops = {
+	.family =	AF_RDS,
+	.create =	rds_create,
+	.owner	=	THIS_MODULE,
+};
+
+static void rds_sock_inc_info(struct socket *sock, unsigned int len,
+			      struct rds_info_iterator *iter,
+			      struct rds_info_lengths *lens)
+{
+	struct rds_sock *rs;
+	struct rds_incoming *inc;
+	unsigned int total = 0;
+
+	len /= sizeof(struct rds_info_message);
+
+	spin_lock_bh(&rds_sock_lock);
+
+	list_for_each_entry(rs, &rds_sock_list, rs_item) {
+		read_lock(&rs->rs_recv_lock);
+
+		/* XXX too lazy to maintain counts.. */
+		list_for_each_entry(inc, &rs->rs_recv_queue, i_item) {
+			total++;
+			if (total <= len)
+				rds_inc_info_copy(inc, iter, inc->i_saddr,
+						  rs->rs_bound_addr, 1);
+		}
+
+		read_unlock(&rs->rs_recv_lock);
+	}
+
+	spin_unlock_bh(&rds_sock_lock);
+
+	lens->nr = total;
+	lens->each = sizeof(struct rds_info_message);
+}
+
+static void rds_sock_info(struct socket *sock, unsigned int len,
+			  struct rds_info_iterator *iter,
+			  struct rds_info_lengths *lens)
+{
+	struct rds_info_socket sinfo;
+	struct rds_sock *rs;
+
+	len /= sizeof(struct rds_info_socket);
+
+	spin_lock_bh(&rds_sock_lock);
+
+	if (len < rds_sock_count)
+		goto out;
+
+	list_for_each_entry(rs, &rds_sock_list, rs_item) {
+		sinfo.sndbuf = rds_sk_sndbuf(rs);
+		sinfo.rcvbuf = rds_sk_rcvbuf(rs);
+		sinfo.bound_addr = rs->rs_bound_addr;
+		sinfo.connected_addr = rs->rs_conn_addr;
+		sinfo.bound_port = rs->rs_bound_port;
+		sinfo.connected_port = rs->rs_conn_port;
+		sinfo.inum = sock_i_ino(rds_rs_to_sk(rs));
+
+		rds_info_copy(iter, &sinfo, sizeof(sinfo));
+	}
+
+out:
+	lens->nr = rds_sock_count;
+	lens->each = sizeof(struct rds_info_socket);
+
+	spin_unlock_bh(&rds_sock_lock);
+}
+
+static void rds_exit(void)
+{
+	sock_unregister(rds_family_ops.family);
+	proto_unregister(&rds_proto);
+	rds_conn_exit();
+	rds_cong_exit();
+	rds_sysctl_exit();
+	rds_threads_exit();
+	rds_stats_exit();
+	rds_page_exit();
+	rds_info_deregister_func(RDS_INFO_SOCKETS, rds_sock_info);
+	rds_info_deregister_func(RDS_INFO_RECV_MESSAGES, rds_sock_inc_info);
+}
+module_exit(rds_exit);
+
+static int rds_init(void)
+{
+	int ret;
+
+	ret = rds_conn_init();
+	if (ret)
+		goto out;
+	ret = rds_threads_init();
+	if (ret)
+		goto out_conn;
+	ret = rds_sysctl_init();
+	if (ret)
+		goto out_threads;
+	ret = rds_stats_init();
+	if (ret)
+		goto out_sysctl;
+	ret = proto_register(&rds_proto, 1);
+	if (ret)
+		goto out_stats;
+	ret = sock_register(&rds_family_ops);
+	if (ret)
+		goto out_proto;
+
+	rds_info_register_func(RDS_INFO_SOCKETS, rds_sock_info);
+	rds_info_register_func(RDS_INFO_RECV_MESSAGES, rds_sock_inc_info);
+
+	goto out;
+
+out_proto:
+	proto_unregister(&rds_proto);
+out_stats:
+	rds_stats_exit();
+out_sysctl:
+	rds_sysctl_exit();
+out_threads:
+	rds_threads_exit();
+out_conn:
+	rds_conn_exit();
+	rds_cong_exit();
+	rds_page_exit();
+out:
+	return ret;
+}
+module_init(rds_init);
+
+#define DRV_VERSION     "4.0"
+#define DRV_RELDATE     "Feb 12, 2009"
+
+MODULE_AUTHOR("Oracle Corporation <rds-devel@oss.oracle.com>");
+MODULE_DESCRIPTION("RDS: Reliable Datagram Sockets"
+		   " v" DRV_VERSION " (" DRV_RELDATE ")");
+MODULE_VERSION(DRV_VERSION);
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_ALIAS_NETPROTO(PF_RDS);
diff --git a/net/rds/bind.c b/net/rds/bind.c
new file mode 100644
index 000000000..a2e6562da
--- /dev/null
+++ b/net/rds/bind.c
@@ -0,0 +1,203 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <net/sock.h>
+#include <linux/in.h>
+#include <linux/if_arp.h>
+#include <linux/jhash.h>
+#include <linux/ratelimit.h>
+#include "rds.h"
+
+#define BIND_HASH_SIZE 1024
+static struct hlist_head bind_hash_table[BIND_HASH_SIZE];
+static DEFINE_SPINLOCK(rds_bind_lock);
+
+static struct hlist_head *hash_to_bucket(__be32 addr, __be16 port)
+{
+	return bind_hash_table + (jhash_2words((u32)addr, (u32)port, 0) &
+				  (BIND_HASH_SIZE - 1));
+}
+
+static struct rds_sock *rds_bind_lookup(__be32 addr, __be16 port,
+					struct rds_sock *insert)
+{
+	struct rds_sock *rs;
+	struct hlist_head *head = hash_to_bucket(addr, port);
+	u64 cmp;
+	u64 needle = ((u64)be32_to_cpu(addr) << 32) | be16_to_cpu(port);
+
+	rcu_read_lock();
+	hlist_for_each_entry_rcu(rs, head, rs_bound_node) {
+		cmp = ((u64)be32_to_cpu(rs->rs_bound_addr) << 32) |
+		      be16_to_cpu(rs->rs_bound_port);
+
+		if (cmp == needle) {
+			rcu_read_unlock();
+			return rs;
+		}
+	}
+	rcu_read_unlock();
+
+	if (insert) {
+		/*
+		 * make sure our addr and port are set before
+		 * we are added to the list, other people
+		 * in rcu will find us as soon as the
+		 * hlist_add_head_rcu is done
+		 */
+		insert->rs_bound_addr = addr;
+		insert->rs_bound_port = port;
+		rds_sock_addref(insert);
+
+		hlist_add_head_rcu(&insert->rs_bound_node, head);
+	}
+	return NULL;
+}
+
+/*
+ * Return the rds_sock bound at the given local address.
+ *
+ * The rx path can race with rds_release.  We notice if rds_release() has
+ * marked this socket and don't return a rs ref to the rx path.
+ */
+struct rds_sock *rds_find_bound(__be32 addr, __be16 port)
+{
+	struct rds_sock *rs;
+
+	rs = rds_bind_lookup(addr, port, NULL);
+
+	if (rs && !sock_flag(rds_rs_to_sk(rs), SOCK_DEAD))
+		rds_sock_addref(rs);
+	else
+		rs = NULL;
+
+	rdsdebug("returning rs %p for %pI4:%u\n", rs, &addr,
+		ntohs(port));
+	return rs;
+}
+
+/* returns -ve errno or +ve port */
+static int rds_add_bound(struct rds_sock *rs, __be32 addr, __be16 *port)
+{
+	unsigned long flags;
+	int ret = -EADDRINUSE;
+	u16 rover, last;
+
+	if (*port != 0) {
+		rover = be16_to_cpu(*port);
+		last = rover;
+	} else {
+		rover = max_t(u16, prandom_u32(), 2);
+		last = rover - 1;
+	}
+
+	spin_lock_irqsave(&rds_bind_lock, flags);
+
+	do {
+		if (rover == 0)
+			rover++;
+		if (!rds_bind_lookup(addr, cpu_to_be16(rover), rs)) {
+			*port = rs->rs_bound_port;
+			ret = 0;
+			rdsdebug("rs %p binding to %pI4:%d\n",
+			  rs, &addr, (int)ntohs(*port));
+			break;
+		}
+	} while (rover++ != last);
+
+	spin_unlock_irqrestore(&rds_bind_lock, flags);
+
+	return ret;
+}
+
+void rds_remove_bound(struct rds_sock *rs)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&rds_bind_lock, flags);
+
+	if (rs->rs_bound_addr) {
+		rdsdebug("rs %p unbinding from %pI4:%d\n",
+		  rs, &rs->rs_bound_addr,
+		  ntohs(rs->rs_bound_port));
+
+		hlist_del_init_rcu(&rs->rs_bound_node);
+		rds_sock_put(rs);
+		rs->rs_bound_addr = 0;
+	}
+
+	spin_unlock_irqrestore(&rds_bind_lock, flags);
+}
+
+int rds_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
+{
+	struct sock *sk = sock->sk;
+	struct sockaddr_in *sin = (struct sockaddr_in *)uaddr;
+	struct rds_sock *rs = rds_sk_to_rs(sk);
+	struct rds_transport *trans;
+	int ret = 0;
+
+	lock_sock(sk);
+
+	if (addr_len != sizeof(struct sockaddr_in) ||
+	    sin->sin_family != AF_INET ||
+	    rs->rs_bound_addr ||
+	    sin->sin_addr.s_addr == htonl(INADDR_ANY)) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	ret = rds_add_bound(rs, sin->sin_addr.s_addr, &sin->sin_port);
+	if (ret)
+		goto out;
+
+	trans = rds_trans_get_preferred(sin->sin_addr.s_addr);
+	if (!trans) {
+		ret = -EADDRNOTAVAIL;
+		rds_remove_bound(rs);
+		printk_ratelimited(KERN_INFO "RDS: rds_bind() could not find a transport, "
+				"load rds_tcp or rds_rdma?\n");
+		goto out;
+	}
+
+	rs->rs_transport = trans;
+	ret = 0;
+
+out:
+	release_sock(sk);
+
+	/* we might have called rds_remove_bound on error */
+	if (ret)
+		synchronize_rcu();
+	return ret;
+}
diff --git a/net/rds/cong.c b/net/rds/cong.c
new file mode 100644
index 000000000..e6144b824
--- /dev/null
+++ b/net/rds/cong.c
@@ -0,0 +1,420 @@
+/*
+ * Copyright (c) 2007 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/slab.h>
+#include <linux/types.h>
+#include <linux/rbtree.h>
+#include <linux/bitops.h>
+#include <linux/export.h>
+
+#include "rds.h"
+
+/*
+ * This file implements the receive side of the unconventional congestion
+ * management in RDS.
+ *
+ * Messages waiting in the receive queue on the receiving socket are accounted
+ * against the sockets SO_RCVBUF option value.  Only the payload bytes in the
+ * message are accounted for.  If the number of bytes queued equals or exceeds
+ * rcvbuf then the socket is congested.  All sends attempted to this socket's
+ * address should return block or return -EWOULDBLOCK.
+ *
+ * Applications are expected to be reasonably tuned such that this situation
+ * very rarely occurs.  An application encountering this "back-pressure" is
+ * considered a bug.
+ *
+ * This is implemented by having each node maintain bitmaps which indicate
+ * which ports on bound addresses are congested.  As the bitmap changes it is
+ * sent through all the connections which terminate in the local address of the
+ * bitmap which changed.
+ *
+ * The bitmaps are allocated as connections are brought up.  This avoids
+ * allocation in the interrupt handling path which queues messages on sockets.
+ * The dense bitmaps let transports send the entire bitmap on any bitmap change
+ * reasonably efficiently.  This is much easier to implement than some
+ * finer-grained communication of per-port congestion.  The sender does a very
+ * inexpensive bit test to test if the port it's about to send to is congested
+ * or not.
+ */
+
+/*
+ * Interaction with poll is a tad tricky. We want all processes stuck in
+ * poll to wake up and check whether a congested destination became uncongested.
+ * The really sad thing is we have no idea which destinations the application
+ * wants to send to - we don't even know which rds_connections are involved.
+ * So until we implement a more flexible rds poll interface, we have to make
+ * do with this:
+ * We maintain a global counter that is incremented each time a congestion map
+ * update is received. Each rds socket tracks this value, and if rds_poll
+ * finds that the saved generation number is smaller than the global generation
+ * number, it wakes up the process.
+ */
+static atomic_t		rds_cong_generation = ATOMIC_INIT(0);
+
+/*
+ * Congestion monitoring
+ */
+static LIST_HEAD(rds_cong_monitor);
+static DEFINE_RWLOCK(rds_cong_monitor_lock);
+
+/*
+ * Yes, a global lock.  It's used so infrequently that it's worth keeping it
+ * global to simplify the locking.  It's only used in the following
+ * circumstances:
+ *
+ *  - on connection buildup to associate a conn with its maps
+ *  - on map changes to inform conns of a new map to send
+ *
+ *  It's sadly ordered under the socket callback lock and the connection lock.
+ *  Receive paths can mark ports congested from interrupt context so the
+ *  lock masks interrupts.
+ */
+static DEFINE_SPINLOCK(rds_cong_lock);
+static struct rb_root rds_cong_tree = RB_ROOT;
+
+static struct rds_cong_map *rds_cong_tree_walk(__be32 addr,
+					       struct rds_cong_map *insert)
+{
+	struct rb_node **p = &rds_cong_tree.rb_node;
+	struct rb_node *parent = NULL;
+	struct rds_cong_map *map;
+
+	while (*p) {
+		parent = *p;
+		map = rb_entry(parent, struct rds_cong_map, m_rb_node);
+
+		if (addr < map->m_addr)
+			p = &(*p)->rb_left;
+		else if (addr > map->m_addr)
+			p = &(*p)->rb_right;
+		else
+			return map;
+	}
+
+	if (insert) {
+		rb_link_node(&insert->m_rb_node, parent, p);
+		rb_insert_color(&insert->m_rb_node, &rds_cong_tree);
+	}
+	return NULL;
+}
+
+/*
+ * There is only ever one bitmap for any address.  Connections try and allocate
+ * these bitmaps in the process getting pointers to them.  The bitmaps are only
+ * ever freed as the module is removed after all connections have been freed.
+ */
+static struct rds_cong_map *rds_cong_from_addr(__be32 addr)
+{
+	struct rds_cong_map *map;
+	struct rds_cong_map *ret = NULL;
+	unsigned long zp;
+	unsigned long i;
+	unsigned long flags;
+
+	map = kzalloc(sizeof(struct rds_cong_map), GFP_KERNEL);
+	if (!map)
+		return NULL;
+
+	map->m_addr = addr;
+	init_waitqueue_head(&map->m_waitq);
+	INIT_LIST_HEAD(&map->m_conn_list);
+
+	for (i = 0; i < RDS_CONG_MAP_PAGES; i++) {
+		zp = get_zeroed_page(GFP_KERNEL);
+		if (zp == 0)
+			goto out;
+		map->m_page_addrs[i] = zp;
+	}
+
+	spin_lock_irqsave(&rds_cong_lock, flags);
+	ret = rds_cong_tree_walk(addr, map);
+	spin_unlock_irqrestore(&rds_cong_lock, flags);
+
+	if (!ret) {
+		ret = map;
+		map = NULL;
+	}
+
+out:
+	if (map) {
+		for (i = 0; i < RDS_CONG_MAP_PAGES && map->m_page_addrs[i]; i++)
+			free_page(map->m_page_addrs[i]);
+		kfree(map);
+	}
+
+	rdsdebug("map %p for addr %x\n", ret, be32_to_cpu(addr));
+
+	return ret;
+}
+
+/*
+ * Put the conn on its local map's list.  This is called when the conn is
+ * really added to the hash.  It's nested under the rds_conn_lock, sadly.
+ */
+void rds_cong_add_conn(struct rds_connection *conn)
+{
+	unsigned long flags;
+
+	rdsdebug("conn %p now on map %p\n", conn, conn->c_lcong);
+	spin_lock_irqsave(&rds_cong_lock, flags);
+	list_add_tail(&conn->c_map_item, &conn->c_lcong->m_conn_list);
+	spin_unlock_irqrestore(&rds_cong_lock, flags);
+}
+
+void rds_cong_remove_conn(struct rds_connection *conn)
+{
+	unsigned long flags;
+
+	rdsdebug("removing conn %p from map %p\n", conn, conn->c_lcong);
+	spin_lock_irqsave(&rds_cong_lock, flags);
+	list_del_init(&conn->c_map_item);
+	spin_unlock_irqrestore(&rds_cong_lock, flags);
+}
+
+int rds_cong_get_maps(struct rds_connection *conn)
+{
+	conn->c_lcong = rds_cong_from_addr(conn->c_laddr);
+	conn->c_fcong = rds_cong_from_addr(conn->c_faddr);
+
+	if (!(conn->c_lcong && conn->c_fcong))
+		return -ENOMEM;
+
+	return 0;
+}
+
+void rds_cong_queue_updates(struct rds_cong_map *map)
+{
+	struct rds_connection *conn;
+	unsigned long flags;
+
+	spin_lock_irqsave(&rds_cong_lock, flags);
+
+	list_for_each_entry(conn, &map->m_conn_list, c_map_item) {
+		if (!test_and_set_bit(0, &conn->c_map_queued)) {
+			rds_stats_inc(s_cong_update_queued);
+			/* We cannot inline the call to rds_send_xmit() here
+			 * for two reasons (both pertaining to a TCP transport):
+			 * 1. When we get here from the receive path, we
+			 *    are already holding the sock_lock (held by
+			 *    tcp_v4_rcv()). So inlining calls to
+			 *    tcp_setsockopt and/or tcp_sendmsg will deadlock
+			 *    when it tries to get the sock_lock())
+			 * 2. Interrupts are masked so that we can mark the
+			 *    the port congested from both send and recv paths.
+			 *    (See comment around declaration of rdc_cong_lock).
+			 *    An attempt to get the sock_lock() here will
+			 *    therefore trigger warnings.
+			 * Defer the xmit to rds_send_worker() instead.
+			 */
+			queue_delayed_work(rds_wq, &conn->c_send_w, 0);
+		}
+	}
+
+	spin_unlock_irqrestore(&rds_cong_lock, flags);
+}
+
+void rds_cong_map_updated(struct rds_cong_map *map, uint64_t portmask)
+{
+	rdsdebug("waking map %p for %pI4\n",
+	  map, &map->m_addr);
+	rds_stats_inc(s_cong_update_received);
+	atomic_inc(&rds_cong_generation);
+	if (waitqueue_active(&map->m_waitq))
+		wake_up(&map->m_waitq);
+	if (waitqueue_active(&rds_poll_waitq))
+		wake_up_all(&rds_poll_waitq);
+
+	if (portmask && !list_empty(&rds_cong_monitor)) {
+		unsigned long flags;
+		struct rds_sock *rs;
+
+		read_lock_irqsave(&rds_cong_monitor_lock, flags);
+		list_for_each_entry(rs, &rds_cong_monitor, rs_cong_list) {
+			spin_lock(&rs->rs_lock);
+			rs->rs_cong_notify |= (rs->rs_cong_mask & portmask);
+			rs->rs_cong_mask &= ~portmask;
+			spin_unlock(&rs->rs_lock);
+			if (rs->rs_cong_notify)
+				rds_wake_sk_sleep(rs);
+		}
+		read_unlock_irqrestore(&rds_cong_monitor_lock, flags);
+	}
+}
+EXPORT_SYMBOL_GPL(rds_cong_map_updated);
+
+int rds_cong_updated_since(unsigned long *recent)
+{
+	unsigned long gen = atomic_read(&rds_cong_generation);
+
+	if (likely(*recent == gen))
+		return 0;
+	*recent = gen;
+	return 1;
+}
+
+/*
+ * We're called under the locking that protects the sockets receive buffer
+ * consumption.  This makes it a lot easier for the caller to only call us
+ * when it knows that an existing set bit needs to be cleared, and vice versa.
+ * We can't block and we need to deal with concurrent sockets working against
+ * the same per-address map.
+ */
+void rds_cong_set_bit(struct rds_cong_map *map, __be16 port)
+{
+	unsigned long i;
+	unsigned long off;
+
+	rdsdebug("setting congestion for %pI4:%u in map %p\n",
+	  &map->m_addr, ntohs(port), map);
+
+	i = be16_to_cpu(port) / RDS_CONG_MAP_PAGE_BITS;
+	off = be16_to_cpu(port) % RDS_CONG_MAP_PAGE_BITS;
+
+	__set_bit_le(off, (void *)map->m_page_addrs[i]);
+}
+
+void rds_cong_clear_bit(struct rds_cong_map *map, __be16 port)
+{
+	unsigned long i;
+	unsigned long off;
+
+	rdsdebug("clearing congestion for %pI4:%u in map %p\n",
+	  &map->m_addr, ntohs(port), map);
+
+	i = be16_to_cpu(port) / RDS_CONG_MAP_PAGE_BITS;
+	off = be16_to_cpu(port) % RDS_CONG_MAP_PAGE_BITS;
+
+	__clear_bit_le(off, (void *)map->m_page_addrs[i]);
+}
+
+static int rds_cong_test_bit(struct rds_cong_map *map, __be16 port)
+{
+	unsigned long i;
+	unsigned long off;
+
+	i = be16_to_cpu(port) / RDS_CONG_MAP_PAGE_BITS;
+	off = be16_to_cpu(port) % RDS_CONG_MAP_PAGE_BITS;
+
+	return test_bit_le(off, (void *)map->m_page_addrs[i]);
+}
+
+void rds_cong_add_socket(struct rds_sock *rs)
+{
+	unsigned long flags;
+
+	write_lock_irqsave(&rds_cong_monitor_lock, flags);
+	if (list_empty(&rs->rs_cong_list))
+		list_add(&rs->rs_cong_list, &rds_cong_monitor);
+	write_unlock_irqrestore(&rds_cong_monitor_lock, flags);
+}
+
+void rds_cong_remove_socket(struct rds_sock *rs)
+{
+	unsigned long flags;
+	struct rds_cong_map *map;
+
+	write_lock_irqsave(&rds_cong_monitor_lock, flags);
+	list_del_init(&rs->rs_cong_list);
+	write_unlock_irqrestore(&rds_cong_monitor_lock, flags);
+
+	/* update congestion map for now-closed port */
+	spin_lock_irqsave(&rds_cong_lock, flags);
+	map = rds_cong_tree_walk(rs->rs_bound_addr, NULL);
+	spin_unlock_irqrestore(&rds_cong_lock, flags);
+
+	if (map && rds_cong_test_bit(map, rs->rs_bound_port)) {
+		rds_cong_clear_bit(map, rs->rs_bound_port);
+		rds_cong_queue_updates(map);
+	}
+}
+
+int rds_cong_wait(struct rds_cong_map *map, __be16 port, int nonblock,
+		  struct rds_sock *rs)
+{
+	if (!rds_cong_test_bit(map, port))
+		return 0;
+	if (nonblock) {
+		if (rs && rs->rs_cong_monitor) {
+			unsigned long flags;
+
+			/* It would have been nice to have an atomic set_bit on
+			 * a uint64_t. */
+			spin_lock_irqsave(&rs->rs_lock, flags);
+			rs->rs_cong_mask |= RDS_CONG_MONITOR_MASK(ntohs(port));
+			spin_unlock_irqrestore(&rs->rs_lock, flags);
+
+			/* Test again - a congestion update may have arrived in
+			 * the meantime. */
+			if (!rds_cong_test_bit(map, port))
+				return 0;
+		}
+		rds_stats_inc(s_cong_send_error);
+		return -ENOBUFS;
+	}
+
+	rds_stats_inc(s_cong_send_blocked);
+	rdsdebug("waiting on map %p for port %u\n", map, be16_to_cpu(port));
+
+	return wait_event_interruptible(map->m_waitq,
+					!rds_cong_test_bit(map, port));
+}
+
+void rds_cong_exit(void)
+{
+	struct rb_node *node;
+	struct rds_cong_map *map;
+	unsigned long i;
+
+	while ((node = rb_first(&rds_cong_tree))) {
+		map = rb_entry(node, struct rds_cong_map, m_rb_node);
+		rdsdebug("freeing map %p\n", map);
+		rb_erase(&map->m_rb_node, &rds_cong_tree);
+		for (i = 0; i < RDS_CONG_MAP_PAGES && map->m_page_addrs[i]; i++)
+			free_page(map->m_page_addrs[i]);
+		kfree(map);
+	}
+}
+
+/*
+ * Allocate a RDS message containing a congestion update.
+ */
+struct rds_message *rds_cong_update_alloc(struct rds_connection *conn)
+{
+	struct rds_cong_map *map = conn->c_lcong;
+	struct rds_message *rm;
+
+	rm = rds_message_map_pages(map->m_page_addrs, RDS_CONG_MAP_BYTES);
+	if (!IS_ERR(rm))
+		rm->m_inc.i_hdr.h_flags = RDS_FLAG_CONG_BITMAP;
+
+	return rm;
+}
diff --git a/net/rds/connection.c b/net/rds/connection.c
new file mode 100644
index 000000000..da6da57e5
--- /dev/null
+++ b/net/rds/connection.c
@@ -0,0 +1,591 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/list.h>
+#include <linux/slab.h>
+#include <linux/export.h>
+#include <net/inet_hashtables.h>
+
+#include "rds.h"
+#include "loop.h"
+
+#define RDS_CONNECTION_HASH_BITS 12
+#define RDS_CONNECTION_HASH_ENTRIES (1 << RDS_CONNECTION_HASH_BITS)
+#define RDS_CONNECTION_HASH_MASK (RDS_CONNECTION_HASH_ENTRIES - 1)
+
+/* converting this to RCU is a chore for another day.. */
+static DEFINE_SPINLOCK(rds_conn_lock);
+static unsigned long rds_conn_count;
+static struct hlist_head rds_conn_hash[RDS_CONNECTION_HASH_ENTRIES];
+static struct kmem_cache *rds_conn_slab;
+
+static struct hlist_head *rds_conn_bucket(__be32 laddr, __be32 faddr)
+{
+	static u32 rds_hash_secret __read_mostly;
+
+	unsigned long hash;
+
+	net_get_random_once(&rds_hash_secret, sizeof(rds_hash_secret));
+
+	/* Pass NULL, don't need struct net for hash */
+	hash = __inet_ehashfn(be32_to_cpu(laddr), 0,
+			      be32_to_cpu(faddr), 0,
+			      rds_hash_secret);
+	return &rds_conn_hash[hash & RDS_CONNECTION_HASH_MASK];
+}
+
+#define rds_conn_info_set(var, test, suffix) do {		\
+	if (test)						\
+		var |= RDS_INFO_CONNECTION_FLAG_##suffix;	\
+} while (0)
+
+/* rcu read lock must be held or the connection spinlock */
+static struct rds_connection *rds_conn_lookup(struct hlist_head *head,
+					      __be32 laddr, __be32 faddr,
+					      struct rds_transport *trans)
+{
+	struct rds_connection *conn, *ret = NULL;
+
+	hlist_for_each_entry_rcu(conn, head, c_hash_node) {
+		if (conn->c_faddr == faddr && conn->c_laddr == laddr &&
+				conn->c_trans == trans) {
+			ret = conn;
+			break;
+		}
+	}
+	rdsdebug("returning conn %p for %pI4 -> %pI4\n", ret,
+		 &laddr, &faddr);
+	return ret;
+}
+
+/*
+ * This is called by transports as they're bringing down a connection.
+ * It clears partial message state so that the transport can start sending
+ * and receiving over this connection again in the future.  It is up to
+ * the transport to have serialized this call with its send and recv.
+ */
+static void rds_conn_reset(struct rds_connection *conn)
+{
+	rdsdebug("connection %pI4 to %pI4 reset\n",
+	  &conn->c_laddr, &conn->c_faddr);
+
+	rds_stats_inc(s_conn_reset);
+	rds_send_reset(conn);
+	conn->c_flags = 0;
+
+	/* Do not clear next_rx_seq here, else we cannot distinguish
+	 * retransmitted packets from new packets, and will hand all
+	 * of them to the application. That is not consistent with the
+	 * reliability guarantees of RDS. */
+}
+
+/*
+ * There is only every one 'conn' for a given pair of addresses in the
+ * system at a time.  They contain messages to be retransmitted and so
+ * span the lifetime of the actual underlying transport connections.
+ *
+ * For now they are not garbage collected once they're created.  They
+ * are torn down as the module is removed, if ever.
+ */
+static struct rds_connection *__rds_conn_create(__be32 laddr, __be32 faddr,
+				       struct rds_transport *trans, gfp_t gfp,
+				       int is_outgoing)
+{
+	struct rds_connection *conn, *parent = NULL;
+	struct hlist_head *head = rds_conn_bucket(laddr, faddr);
+	struct rds_transport *loop_trans;
+	unsigned long flags;
+	int ret;
+	struct rds_transport *otrans = trans;
+
+	if (!is_outgoing && otrans->t_type == RDS_TRANS_TCP)
+		goto new_conn;
+	rcu_read_lock();
+	conn = rds_conn_lookup(head, laddr, faddr, trans);
+	if (conn && conn->c_loopback && conn->c_trans != &rds_loop_transport &&
+	    laddr == faddr && !is_outgoing) {
+		/* This is a looped back IB connection, and we're
+		 * called by the code handling the incoming connect.
+		 * We need a second connection object into which we
+		 * can stick the other QP. */
+		parent = conn;
+		conn = parent->c_passive;
+	}
+	rcu_read_unlock();
+	if (conn)
+		goto out;
+
+new_conn:
+	conn = kmem_cache_zalloc(rds_conn_slab, gfp);
+	if (!conn) {
+		conn = ERR_PTR(-ENOMEM);
+		goto out;
+	}
+
+	INIT_HLIST_NODE(&conn->c_hash_node);
+	conn->c_laddr = laddr;
+	conn->c_faddr = faddr;
+	spin_lock_init(&conn->c_lock);
+	conn->c_next_tx_seq = 1;
+
+	init_waitqueue_head(&conn->c_waitq);
+	INIT_LIST_HEAD(&conn->c_send_queue);
+	INIT_LIST_HEAD(&conn->c_retrans);
+
+	ret = rds_cong_get_maps(conn);
+	if (ret) {
+		kmem_cache_free(rds_conn_slab, conn);
+		conn = ERR_PTR(ret);
+		goto out;
+	}
+
+	/*
+	 * This is where a connection becomes loopback.  If *any* RDS sockets
+	 * can bind to the destination address then we'd rather the messages
+	 * flow through loopback rather than either transport.
+	 */
+	loop_trans = rds_trans_get_preferred(faddr);
+	if (loop_trans) {
+		rds_trans_put(loop_trans);
+		conn->c_loopback = 1;
+		if (is_outgoing && trans->t_prefer_loopback) {
+			/* "outgoing" connection - and the transport
+			 * says it wants the connection handled by the
+			 * loopback transport. This is what TCP does.
+			 */
+			trans = &rds_loop_transport;
+		}
+	}
+
+	conn->c_trans = trans;
+
+	ret = trans->conn_alloc(conn, gfp);
+	if (ret) {
+		kmem_cache_free(rds_conn_slab, conn);
+		conn = ERR_PTR(ret);
+		goto out;
+	}
+
+	atomic_set(&conn->c_state, RDS_CONN_DOWN);
+	conn->c_send_gen = 0;
+	conn->c_reconnect_jiffies = 0;
+	INIT_DELAYED_WORK(&conn->c_send_w, rds_send_worker);
+	INIT_DELAYED_WORK(&conn->c_recv_w, rds_recv_worker);
+	INIT_DELAYED_WORK(&conn->c_conn_w, rds_connect_worker);
+	INIT_WORK(&conn->c_down_w, rds_shutdown_worker);
+	mutex_init(&conn->c_cm_lock);
+	conn->c_flags = 0;
+
+	rdsdebug("allocated conn %p for %pI4 -> %pI4 over %s %s\n",
+	  conn, &laddr, &faddr,
+	  trans->t_name ? trans->t_name : "[unknown]",
+	  is_outgoing ? "(outgoing)" : "");
+
+	/*
+	 * Since we ran without holding the conn lock, someone could
+	 * have created the same conn (either normal or passive) in the
+	 * interim. We check while holding the lock. If we won, we complete
+	 * init and return our conn. If we lost, we rollback and return the
+	 * other one.
+	 */
+	spin_lock_irqsave(&rds_conn_lock, flags);
+	if (parent) {
+		/* Creating passive conn */
+		if (parent->c_passive) {
+			trans->conn_free(conn->c_transport_data);
+			kmem_cache_free(rds_conn_slab, conn);
+			conn = parent->c_passive;
+		} else {
+			parent->c_passive = conn;
+			rds_cong_add_conn(conn);
+			rds_conn_count++;
+		}
+	} else {
+		/* Creating normal conn */
+		struct rds_connection *found;
+
+		if (!is_outgoing && otrans->t_type == RDS_TRANS_TCP)
+			found = NULL;
+		else
+			found = rds_conn_lookup(head, laddr, faddr, trans);
+		if (found) {
+			trans->conn_free(conn->c_transport_data);
+			kmem_cache_free(rds_conn_slab, conn);
+			conn = found;
+		} else {
+			if ((is_outgoing && otrans->t_type == RDS_TRANS_TCP) ||
+			    (otrans->t_type != RDS_TRANS_TCP)) {
+				/* Only the active side should be added to
+				 * reconnect list for TCP.
+				 */
+				hlist_add_head_rcu(&conn->c_hash_node, head);
+			}
+			rds_cong_add_conn(conn);
+			rds_conn_count++;
+		}
+	}
+	spin_unlock_irqrestore(&rds_conn_lock, flags);
+
+out:
+	return conn;
+}
+
+struct rds_connection *rds_conn_create(__be32 laddr, __be32 faddr,
+				       struct rds_transport *trans, gfp_t gfp)
+{
+	return __rds_conn_create(laddr, faddr, trans, gfp, 0);
+}
+EXPORT_SYMBOL_GPL(rds_conn_create);
+
+struct rds_connection *rds_conn_create_outgoing(__be32 laddr, __be32 faddr,
+				       struct rds_transport *trans, gfp_t gfp)
+{
+	return __rds_conn_create(laddr, faddr, trans, gfp, 1);
+}
+EXPORT_SYMBOL_GPL(rds_conn_create_outgoing);
+
+void rds_conn_shutdown(struct rds_connection *conn)
+{
+	/* shut it down unless it's down already */
+	if (!rds_conn_transition(conn, RDS_CONN_DOWN, RDS_CONN_DOWN)) {
+		/*
+		 * Quiesce the connection mgmt handlers before we start tearing
+		 * things down. We don't hold the mutex for the entire
+		 * duration of the shutdown operation, else we may be
+		 * deadlocking with the CM handler. Instead, the CM event
+		 * handler is supposed to check for state DISCONNECTING
+		 */
+		mutex_lock(&conn->c_cm_lock);
+		if (!rds_conn_transition(conn, RDS_CONN_UP, RDS_CONN_DISCONNECTING)
+		 && !rds_conn_transition(conn, RDS_CONN_ERROR, RDS_CONN_DISCONNECTING)) {
+			rds_conn_error(conn, "shutdown called in state %d\n",
+					atomic_read(&conn->c_state));
+			mutex_unlock(&conn->c_cm_lock);
+			return;
+		}
+		mutex_unlock(&conn->c_cm_lock);
+
+		wait_event(conn->c_waitq,
+			   !test_bit(RDS_IN_XMIT, &conn->c_flags));
+
+		conn->c_trans->conn_shutdown(conn);
+		rds_conn_reset(conn);
+
+		if (!rds_conn_transition(conn, RDS_CONN_DISCONNECTING, RDS_CONN_DOWN)) {
+			/* This can happen - eg when we're in the middle of tearing
+			 * down the connection, and someone unloads the rds module.
+			 * Quite reproduceable with loopback connections.
+			 * Mostly harmless.
+			 */
+			rds_conn_error(conn,
+				"%s: failed to transition to state DOWN, "
+				"current state is %d\n",
+				__func__,
+				atomic_read(&conn->c_state));
+			return;
+		}
+	}
+
+	/* Then reconnect if it's still live.
+	 * The passive side of an IB loopback connection is never added
+	 * to the conn hash, so we never trigger a reconnect on this
+	 * conn - the reconnect is always triggered by the active peer. */
+	cancel_delayed_work_sync(&conn->c_conn_w);
+	rcu_read_lock();
+	if (!hlist_unhashed(&conn->c_hash_node)) {
+		rcu_read_unlock();
+		rds_queue_reconnect(conn);
+	} else {
+		rcu_read_unlock();
+	}
+}
+
+/*
+ * Stop and free a connection.
+ *
+ * This can only be used in very limited circumstances.  It assumes that once
+ * the conn has been shutdown that no one else is referencing the connection.
+ * We can only ensure this in the rmmod path in the current code.
+ */
+void rds_conn_destroy(struct rds_connection *conn)
+{
+	struct rds_message *rm, *rtmp;
+	unsigned long flags;
+
+	rdsdebug("freeing conn %p for %pI4 -> "
+		 "%pI4\n", conn, &conn->c_laddr,
+		 &conn->c_faddr);
+
+	/* Ensure conn will not be scheduled for reconnect */
+	spin_lock_irq(&rds_conn_lock);
+	hlist_del_init_rcu(&conn->c_hash_node);
+	spin_unlock_irq(&rds_conn_lock);
+	synchronize_rcu();
+
+	/* shut the connection down */
+	rds_conn_drop(conn);
+	flush_work(&conn->c_down_w);
+
+	/* make sure lingering queued work won't try to ref the conn */
+	cancel_delayed_work_sync(&conn->c_send_w);
+	cancel_delayed_work_sync(&conn->c_recv_w);
+
+	/* tear down queued messages */
+	list_for_each_entry_safe(rm, rtmp,
+				 &conn->c_send_queue,
+				 m_conn_item) {
+		list_del_init(&rm->m_conn_item);
+		BUG_ON(!list_empty(&rm->m_sock_item));
+		rds_message_put(rm);
+	}
+	if (conn->c_xmit_rm)
+		rds_message_put(conn->c_xmit_rm);
+
+	conn->c_trans->conn_free(conn->c_transport_data);
+
+	/*
+	 * The congestion maps aren't freed up here.  They're
+	 * freed by rds_cong_exit() after all the connections
+	 * have been freed.
+	 */
+	rds_cong_remove_conn(conn);
+
+	BUG_ON(!list_empty(&conn->c_retrans));
+	kmem_cache_free(rds_conn_slab, conn);
+
+	spin_lock_irqsave(&rds_conn_lock, flags);
+	rds_conn_count--;
+	spin_unlock_irqrestore(&rds_conn_lock, flags);
+}
+EXPORT_SYMBOL_GPL(rds_conn_destroy);
+
+static void rds_conn_message_info(struct socket *sock, unsigned int len,
+				  struct rds_info_iterator *iter,
+				  struct rds_info_lengths *lens,
+				  int want_send)
+{
+	struct hlist_head *head;
+	struct list_head *list;
+	struct rds_connection *conn;
+	struct rds_message *rm;
+	unsigned int total = 0;
+	unsigned long flags;
+	size_t i;
+
+	len /= sizeof(struct rds_info_message);
+
+	rcu_read_lock();
+
+	for (i = 0, head = rds_conn_hash; i < ARRAY_SIZE(rds_conn_hash);
+	     i++, head++) {
+		hlist_for_each_entry_rcu(conn, head, c_hash_node) {
+			if (want_send)
+				list = &conn->c_send_queue;
+			else
+				list = &conn->c_retrans;
+
+			spin_lock_irqsave(&conn->c_lock, flags);
+
+			/* XXX too lazy to maintain counts.. */
+			list_for_each_entry(rm, list, m_conn_item) {
+				total++;
+				if (total <= len)
+					rds_inc_info_copy(&rm->m_inc, iter,
+							  conn->c_laddr,
+							  conn->c_faddr, 0);
+			}
+
+			spin_unlock_irqrestore(&conn->c_lock, flags);
+		}
+	}
+	rcu_read_unlock();
+
+	lens->nr = total;
+	lens->each = sizeof(struct rds_info_message);
+}
+
+static void rds_conn_message_info_send(struct socket *sock, unsigned int len,
+				       struct rds_info_iterator *iter,
+				       struct rds_info_lengths *lens)
+{
+	rds_conn_message_info(sock, len, iter, lens, 1);
+}
+
+static void rds_conn_message_info_retrans(struct socket *sock,
+					  unsigned int len,
+					  struct rds_info_iterator *iter,
+					  struct rds_info_lengths *lens)
+{
+	rds_conn_message_info(sock, len, iter, lens, 0);
+}
+
+void rds_for_each_conn_info(struct socket *sock, unsigned int len,
+			  struct rds_info_iterator *iter,
+			  struct rds_info_lengths *lens,
+			  int (*visitor)(struct rds_connection *, void *),
+			  size_t item_len)
+{
+	uint64_t buffer[(item_len + 7) / 8];
+	struct hlist_head *head;
+	struct rds_connection *conn;
+	size_t i;
+
+	rcu_read_lock();
+
+	lens->nr = 0;
+	lens->each = item_len;
+
+	for (i = 0, head = rds_conn_hash; i < ARRAY_SIZE(rds_conn_hash);
+	     i++, head++) {
+		hlist_for_each_entry_rcu(conn, head, c_hash_node) {
+
+			/* XXX no c_lock usage.. */
+			if (!visitor(conn, buffer))
+				continue;
+
+			/* We copy as much as we can fit in the buffer,
+			 * but we count all items so that the caller
+			 * can resize the buffer. */
+			if (len >= item_len) {
+				rds_info_copy(iter, buffer, item_len);
+				len -= item_len;
+			}
+			lens->nr++;
+		}
+	}
+	rcu_read_unlock();
+}
+EXPORT_SYMBOL_GPL(rds_for_each_conn_info);
+
+static int rds_conn_info_visitor(struct rds_connection *conn,
+				  void *buffer)
+{
+	struct rds_info_connection *cinfo = buffer;
+
+	cinfo->next_tx_seq = conn->c_next_tx_seq;
+	cinfo->next_rx_seq = conn->c_next_rx_seq;
+	cinfo->laddr = conn->c_laddr;
+	cinfo->faddr = conn->c_faddr;
+	strncpy(cinfo->transport, conn->c_trans->t_name,
+		sizeof(cinfo->transport));
+	cinfo->flags = 0;
+
+	rds_conn_info_set(cinfo->flags, test_bit(RDS_IN_XMIT, &conn->c_flags),
+			  SENDING);
+	/* XXX Future: return the state rather than these funky bits */
+	rds_conn_info_set(cinfo->flags,
+			  atomic_read(&conn->c_state) == RDS_CONN_CONNECTING,
+			  CONNECTING);
+	rds_conn_info_set(cinfo->flags,
+			  atomic_read(&conn->c_state) == RDS_CONN_UP,
+			  CONNECTED);
+	return 1;
+}
+
+static void rds_conn_info(struct socket *sock, unsigned int len,
+			  struct rds_info_iterator *iter,
+			  struct rds_info_lengths *lens)
+{
+	rds_for_each_conn_info(sock, len, iter, lens,
+				rds_conn_info_visitor,
+				sizeof(struct rds_info_connection));
+}
+
+int rds_conn_init(void)
+{
+	rds_conn_slab = kmem_cache_create("rds_connection",
+					  sizeof(struct rds_connection),
+					  0, 0, NULL);
+	if (!rds_conn_slab)
+		return -ENOMEM;
+
+	rds_info_register_func(RDS_INFO_CONNECTIONS, rds_conn_info);
+	rds_info_register_func(RDS_INFO_SEND_MESSAGES,
+			       rds_conn_message_info_send);
+	rds_info_register_func(RDS_INFO_RETRANS_MESSAGES,
+			       rds_conn_message_info_retrans);
+
+	return 0;
+}
+
+void rds_conn_exit(void)
+{
+	rds_loop_exit();
+
+	WARN_ON(!hlist_empty(rds_conn_hash));
+
+	kmem_cache_destroy(rds_conn_slab);
+
+	rds_info_deregister_func(RDS_INFO_CONNECTIONS, rds_conn_info);
+	rds_info_deregister_func(RDS_INFO_SEND_MESSAGES,
+				 rds_conn_message_info_send);
+	rds_info_deregister_func(RDS_INFO_RETRANS_MESSAGES,
+				 rds_conn_message_info_retrans);
+}
+
+/*
+ * Force a disconnect
+ */
+void rds_conn_drop(struct rds_connection *conn)
+{
+	atomic_set(&conn->c_state, RDS_CONN_ERROR);
+	queue_work(rds_wq, &conn->c_down_w);
+}
+EXPORT_SYMBOL_GPL(rds_conn_drop);
+
+/*
+ * If the connection is down, trigger a connect. We may have scheduled a
+ * delayed reconnect however - in this case we should not interfere.
+ */
+void rds_conn_connect_if_down(struct rds_connection *conn)
+{
+	if (rds_conn_state(conn) == RDS_CONN_DOWN &&
+	    !test_and_set_bit(RDS_RECONNECT_PENDING, &conn->c_flags))
+		queue_delayed_work(rds_wq, &conn->c_conn_w, 0);
+}
+EXPORT_SYMBOL_GPL(rds_conn_connect_if_down);
+
+/*
+ * An error occurred on the connection
+ */
+void
+__rds_conn_error(struct rds_connection *conn, const char *fmt, ...)
+{
+	va_list ap;
+
+	va_start(ap, fmt);
+	vprintk(fmt, ap);
+	va_end(ap);
+
+	rds_conn_drop(conn);
+}
diff --git a/net/rds/ib.c b/net/rds/ib.c
new file mode 100644
index 000000000..ba2dffeff
--- /dev/null
+++ b/net/rds/ib.c
@@ -0,0 +1,435 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/in.h>
+#include <linux/if.h>
+#include <linux/netdevice.h>
+#include <linux/inetdevice.h>
+#include <linux/if_arp.h>
+#include <linux/delay.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+
+#include "rds.h"
+#include "ib.h"
+
+static unsigned int fmr_pool_size = RDS_FMR_POOL_SIZE;
+unsigned int fmr_message_size = RDS_FMR_SIZE + 1; /* +1 allows for unaligned MRs */
+unsigned int rds_ib_retry_count = RDS_IB_DEFAULT_RETRY_COUNT;
+
+module_param(fmr_pool_size, int, 0444);
+MODULE_PARM_DESC(fmr_pool_size, " Max number of fmr per HCA");
+module_param(fmr_message_size, int, 0444);
+MODULE_PARM_DESC(fmr_message_size, " Max size of a RDMA transfer");
+module_param(rds_ib_retry_count, int, 0444);
+MODULE_PARM_DESC(rds_ib_retry_count, " Number of hw retries before reporting an error");
+
+/*
+ * we have a clumsy combination of RCU and a rwsem protecting this list
+ * because it is used both in the get_mr fast path and while blocking in
+ * the FMR flushing path.
+ */
+DECLARE_RWSEM(rds_ib_devices_lock);
+struct list_head rds_ib_devices;
+
+/* NOTE: if also grabbing ibdev lock, grab this first */
+DEFINE_SPINLOCK(ib_nodev_conns_lock);
+LIST_HEAD(ib_nodev_conns);
+
+static void rds_ib_nodev_connect(void)
+{
+	struct rds_ib_connection *ic;
+
+	spin_lock(&ib_nodev_conns_lock);
+	list_for_each_entry(ic, &ib_nodev_conns, ib_node)
+		rds_conn_connect_if_down(ic->conn);
+	spin_unlock(&ib_nodev_conns_lock);
+}
+
+static void rds_ib_dev_shutdown(struct rds_ib_device *rds_ibdev)
+{
+	struct rds_ib_connection *ic;
+	unsigned long flags;
+
+	spin_lock_irqsave(&rds_ibdev->spinlock, flags);
+	list_for_each_entry(ic, &rds_ibdev->conn_list, ib_node)
+		rds_conn_drop(ic->conn);
+	spin_unlock_irqrestore(&rds_ibdev->spinlock, flags);
+}
+
+/*
+ * rds_ib_destroy_mr_pool() blocks on a few things and mrs drop references
+ * from interrupt context so we push freing off into a work struct in krdsd.
+ */
+static void rds_ib_dev_free(struct work_struct *work)
+{
+	struct rds_ib_ipaddr *i_ipaddr, *i_next;
+	struct rds_ib_device *rds_ibdev = container_of(work,
+					struct rds_ib_device, free_work);
+
+	if (rds_ibdev->mr_pool)
+		rds_ib_destroy_mr_pool(rds_ibdev->mr_pool);
+	if (rds_ibdev->mr)
+		ib_dereg_mr(rds_ibdev->mr);
+	if (rds_ibdev->pd)
+		ib_dealloc_pd(rds_ibdev->pd);
+
+	list_for_each_entry_safe(i_ipaddr, i_next, &rds_ibdev->ipaddr_list, list) {
+		list_del(&i_ipaddr->list);
+		kfree(i_ipaddr);
+	}
+
+	kfree(rds_ibdev);
+}
+
+void rds_ib_dev_put(struct rds_ib_device *rds_ibdev)
+{
+	BUG_ON(atomic_read(&rds_ibdev->refcount) <= 0);
+	if (atomic_dec_and_test(&rds_ibdev->refcount))
+		queue_work(rds_wq, &rds_ibdev->free_work);
+}
+
+static void rds_ib_add_one(struct ib_device *device)
+{
+	struct rds_ib_device *rds_ibdev;
+	struct ib_device_attr *dev_attr;
+
+	/* Only handle IB (no iWARP) devices */
+	if (device->node_type != RDMA_NODE_IB_CA)
+		return;
+
+	dev_attr = kmalloc(sizeof *dev_attr, GFP_KERNEL);
+	if (!dev_attr)
+		return;
+
+	if (ib_query_device(device, dev_attr)) {
+		rdsdebug("Query device failed for %s\n", device->name);
+		goto free_attr;
+	}
+
+	rds_ibdev = kzalloc_node(sizeof(struct rds_ib_device), GFP_KERNEL,
+				 ibdev_to_node(device));
+	if (!rds_ibdev)
+		goto free_attr;
+
+	spin_lock_init(&rds_ibdev->spinlock);
+	atomic_set(&rds_ibdev->refcount, 1);
+	INIT_WORK(&rds_ibdev->free_work, rds_ib_dev_free);
+
+	rds_ibdev->max_wrs = dev_attr->max_qp_wr;
+	rds_ibdev->max_sge = min(dev_attr->max_sge, RDS_IB_MAX_SGE);
+
+	rds_ibdev->fmr_max_remaps = dev_attr->max_map_per_fmr?: 32;
+	rds_ibdev->max_fmrs = dev_attr->max_fmr ?
+			min_t(unsigned int, dev_attr->max_fmr, fmr_pool_size) :
+			fmr_pool_size;
+
+	rds_ibdev->max_initiator_depth = dev_attr->max_qp_init_rd_atom;
+	rds_ibdev->max_responder_resources = dev_attr->max_qp_rd_atom;
+
+	rds_ibdev->dev = device;
+	rds_ibdev->pd = ib_alloc_pd(device);
+	if (IS_ERR(rds_ibdev->pd)) {
+		rds_ibdev->pd = NULL;
+		goto put_dev;
+	}
+
+	rds_ibdev->mr = ib_get_dma_mr(rds_ibdev->pd, IB_ACCESS_LOCAL_WRITE);
+	if (IS_ERR(rds_ibdev->mr)) {
+		rds_ibdev->mr = NULL;
+		goto put_dev;
+	}
+
+	rds_ibdev->mr_pool = rds_ib_create_mr_pool(rds_ibdev);
+	if (IS_ERR(rds_ibdev->mr_pool)) {
+		rds_ibdev->mr_pool = NULL;
+		goto put_dev;
+	}
+
+	INIT_LIST_HEAD(&rds_ibdev->ipaddr_list);
+	INIT_LIST_HEAD(&rds_ibdev->conn_list);
+
+	down_write(&rds_ib_devices_lock);
+	list_add_tail_rcu(&rds_ibdev->list, &rds_ib_devices);
+	up_write(&rds_ib_devices_lock);
+	atomic_inc(&rds_ibdev->refcount);
+
+	ib_set_client_data(device, &rds_ib_client, rds_ibdev);
+	atomic_inc(&rds_ibdev->refcount);
+
+	rds_ib_nodev_connect();
+
+put_dev:
+	rds_ib_dev_put(rds_ibdev);
+free_attr:
+	kfree(dev_attr);
+}
+
+/*
+ * New connections use this to find the device to associate with the
+ * connection.  It's not in the fast path so we're not concerned about the
+ * performance of the IB call.  (As of this writing, it uses an interrupt
+ * blocking spinlock to serialize walking a per-device list of all registered
+ * clients.)
+ *
+ * RCU is used to handle incoming connections racing with device teardown.
+ * Rather than use a lock to serialize removal from the client_data and
+ * getting a new reference, we use an RCU grace period.  The destruction
+ * path removes the device from client_data and then waits for all RCU
+ * readers to finish.
+ *
+ * A new connection can get NULL from this if its arriving on a
+ * device that is in the process of being removed.
+ */
+struct rds_ib_device *rds_ib_get_client_data(struct ib_device *device)
+{
+	struct rds_ib_device *rds_ibdev;
+
+	rcu_read_lock();
+	rds_ibdev = ib_get_client_data(device, &rds_ib_client);
+	if (rds_ibdev)
+		atomic_inc(&rds_ibdev->refcount);
+	rcu_read_unlock();
+	return rds_ibdev;
+}
+
+/*
+ * The IB stack is letting us know that a device is going away.  This can
+ * happen if the underlying HCA driver is removed or if PCI hotplug is removing
+ * the pci function, for example.
+ *
+ * This can be called at any time and can be racing with any other RDS path.
+ */
+static void rds_ib_remove_one(struct ib_device *device)
+{
+	struct rds_ib_device *rds_ibdev;
+
+	rds_ibdev = ib_get_client_data(device, &rds_ib_client);
+	if (!rds_ibdev)
+		return;
+
+	rds_ib_dev_shutdown(rds_ibdev);
+
+	/* stop connection attempts from getting a reference to this device. */
+	ib_set_client_data(device, &rds_ib_client, NULL);
+
+	down_write(&rds_ib_devices_lock);
+	list_del_rcu(&rds_ibdev->list);
+	up_write(&rds_ib_devices_lock);
+
+	/*
+	 * This synchronize rcu is waiting for readers of both the ib
+	 * client data and the devices list to finish before we drop
+	 * both of those references.
+	 */
+	synchronize_rcu();
+	rds_ib_dev_put(rds_ibdev);
+	rds_ib_dev_put(rds_ibdev);
+}
+
+struct ib_client rds_ib_client = {
+	.name   = "rds_ib",
+	.add    = rds_ib_add_one,
+	.remove = rds_ib_remove_one
+};
+
+static int rds_ib_conn_info_visitor(struct rds_connection *conn,
+				    void *buffer)
+{
+	struct rds_info_rdma_connection *iinfo = buffer;
+	struct rds_ib_connection *ic;
+
+	/* We will only ever look at IB transports */
+	if (conn->c_trans != &rds_ib_transport)
+		return 0;
+
+	iinfo->src_addr = conn->c_laddr;
+	iinfo->dst_addr = conn->c_faddr;
+
+	memset(&iinfo->src_gid, 0, sizeof(iinfo->src_gid));
+	memset(&iinfo->dst_gid, 0, sizeof(iinfo->dst_gid));
+	if (rds_conn_state(conn) == RDS_CONN_UP) {
+		struct rds_ib_device *rds_ibdev;
+		struct rdma_dev_addr *dev_addr;
+
+		ic = conn->c_transport_data;
+		dev_addr = &ic->i_cm_id->route.addr.dev_addr;
+
+		rdma_addr_get_sgid(dev_addr, (union ib_gid *) &iinfo->src_gid);
+		rdma_addr_get_dgid(dev_addr, (union ib_gid *) &iinfo->dst_gid);
+
+		rds_ibdev = ic->rds_ibdev;
+		iinfo->max_send_wr = ic->i_send_ring.w_nr;
+		iinfo->max_recv_wr = ic->i_recv_ring.w_nr;
+		iinfo->max_send_sge = rds_ibdev->max_sge;
+		rds_ib_get_mr_info(rds_ibdev, iinfo);
+	}
+	return 1;
+}
+
+static void rds_ib_ic_info(struct socket *sock, unsigned int len,
+			   struct rds_info_iterator *iter,
+			   struct rds_info_lengths *lens)
+{
+	rds_for_each_conn_info(sock, len, iter, lens,
+				rds_ib_conn_info_visitor,
+				sizeof(struct rds_info_rdma_connection));
+}
+
+
+/*
+ * Early RDS/IB was built to only bind to an address if there is an IPoIB
+ * device with that address set.
+ *
+ * If it were me, I'd advocate for something more flexible.  Sending and
+ * receiving should be device-agnostic.  Transports would try and maintain
+ * connections between peers who have messages queued.  Userspace would be
+ * allowed to influence which paths have priority.  We could call userspace
+ * asserting this policy "routing".
+ */
+static int rds_ib_laddr_check(__be32 addr)
+{
+	int ret;
+	struct rdma_cm_id *cm_id;
+	struct sockaddr_in sin;
+
+	/* Create a CMA ID and try to bind it. This catches both
+	 * IB and iWARP capable NICs.
+	 */
+	cm_id = rdma_create_id(NULL, NULL, RDMA_PS_TCP, IB_QPT_RC);
+	if (IS_ERR(cm_id))
+		return PTR_ERR(cm_id);
+
+	memset(&sin, 0, sizeof(sin));
+	sin.sin_family = AF_INET;
+	sin.sin_addr.s_addr = addr;
+
+	/* rdma_bind_addr will only succeed for IB & iWARP devices */
+	ret = rdma_bind_addr(cm_id, (struct sockaddr *)&sin);
+	/* due to this, we will claim to support iWARP devices unless we
+	   check node_type. */
+	if (ret || !cm_id->device ||
+	    cm_id->device->node_type != RDMA_NODE_IB_CA)
+		ret = -EADDRNOTAVAIL;
+
+	rdsdebug("addr %pI4 ret %d node type %d\n",
+		&addr, ret,
+		cm_id->device ? cm_id->device->node_type : -1);
+
+	rdma_destroy_id(cm_id);
+
+	return ret;
+}
+
+static void rds_ib_unregister_client(void)
+{
+	ib_unregister_client(&rds_ib_client);
+	/* wait for rds_ib_dev_free() to complete */
+	flush_workqueue(rds_wq);
+}
+
+void rds_ib_exit(void)
+{
+	rds_info_deregister_func(RDS_INFO_IB_CONNECTIONS, rds_ib_ic_info);
+	rds_ib_unregister_client();
+	rds_ib_destroy_nodev_conns();
+	rds_ib_sysctl_exit();
+	rds_ib_recv_exit();
+	rds_trans_unregister(&rds_ib_transport);
+}
+
+struct rds_transport rds_ib_transport = {
+	.laddr_check		= rds_ib_laddr_check,
+	.xmit_complete		= rds_ib_xmit_complete,
+	.xmit			= rds_ib_xmit,
+	.xmit_rdma		= rds_ib_xmit_rdma,
+	.xmit_atomic		= rds_ib_xmit_atomic,
+	.recv			= rds_ib_recv,
+	.conn_alloc		= rds_ib_conn_alloc,
+	.conn_free		= rds_ib_conn_free,
+	.conn_connect		= rds_ib_conn_connect,
+	.conn_shutdown		= rds_ib_conn_shutdown,
+	.inc_copy_to_user	= rds_ib_inc_copy_to_user,
+	.inc_free		= rds_ib_inc_free,
+	.cm_initiate_connect	= rds_ib_cm_initiate_connect,
+	.cm_handle_connect	= rds_ib_cm_handle_connect,
+	.cm_connect_complete	= rds_ib_cm_connect_complete,
+	.stats_info_copy	= rds_ib_stats_info_copy,
+	.exit			= rds_ib_exit,
+	.get_mr			= rds_ib_get_mr,
+	.sync_mr		= rds_ib_sync_mr,
+	.free_mr		= rds_ib_free_mr,
+	.flush_mrs		= rds_ib_flush_mrs,
+	.t_owner		= THIS_MODULE,
+	.t_name			= "infiniband",
+	.t_type			= RDS_TRANS_IB
+};
+
+int rds_ib_init(void)
+{
+	int ret;
+
+	INIT_LIST_HEAD(&rds_ib_devices);
+
+	ret = ib_register_client(&rds_ib_client);
+	if (ret)
+		goto out;
+
+	ret = rds_ib_sysctl_init();
+	if (ret)
+		goto out_ibreg;
+
+	ret = rds_ib_recv_init();
+	if (ret)
+		goto out_sysctl;
+
+	ret = rds_trans_register(&rds_ib_transport);
+	if (ret)
+		goto out_recv;
+
+	rds_info_register_func(RDS_INFO_IB_CONNECTIONS, rds_ib_ic_info);
+
+	goto out;
+
+out_recv:
+	rds_ib_recv_exit();
+out_sysctl:
+	rds_ib_sysctl_exit();
+out_ibreg:
+	rds_ib_unregister_client();
+out:
+	return ret;
+}
+
+MODULE_LICENSE("GPL");
+
diff --git a/net/rds/ib.h b/net/rds/ib.h
new file mode 100644
index 000000000..c36d71322
--- /dev/null
+++ b/net/rds/ib.h
@@ -0,0 +1,372 @@
+#ifndef _RDS_IB_H
+#define _RDS_IB_H
+
+#include <rdma/ib_verbs.h>
+#include <rdma/rdma_cm.h>
+#include <linux/interrupt.h>
+#include <linux/pci.h>
+#include <linux/slab.h>
+#include "rds.h"
+#include "rdma_transport.h"
+
+#define RDS_FMR_SIZE			256
+#define RDS_FMR_POOL_SIZE		8192
+
+#define RDS_IB_MAX_SGE			8
+#define RDS_IB_RECV_SGE 		2
+
+#define RDS_IB_DEFAULT_RECV_WR		1024
+#define RDS_IB_DEFAULT_SEND_WR		256
+
+#define RDS_IB_DEFAULT_RETRY_COUNT	2
+
+#define RDS_IB_SUPPORTED_PROTOCOLS	0x00000003	/* minor versions supported */
+
+#define RDS_IB_RECYCLE_BATCH_COUNT	32
+
+extern struct rw_semaphore rds_ib_devices_lock;
+extern struct list_head rds_ib_devices;
+
+/*
+ * IB posts RDS_FRAG_SIZE fragments of pages to the receive queues to
+ * try and minimize the amount of memory tied up both the device and
+ * socket receive queues.
+ */
+struct rds_page_frag {
+	struct list_head	f_item;
+	struct list_head	f_cache_entry;
+	struct scatterlist	f_sg;
+};
+
+struct rds_ib_incoming {
+	struct list_head	ii_frags;
+	struct list_head	ii_cache_entry;
+	struct rds_incoming	ii_inc;
+};
+
+struct rds_ib_cache_head {
+	struct list_head *first;
+	unsigned long count;
+};
+
+struct rds_ib_refill_cache {
+	struct rds_ib_cache_head __percpu *percpu;
+	struct list_head	 *xfer;
+	struct list_head	 *ready;
+};
+
+struct rds_ib_connect_private {
+	/* Add new fields at the end, and don't permute existing fields. */
+	__be32			dp_saddr;
+	__be32			dp_daddr;
+	u8			dp_protocol_major;
+	u8			dp_protocol_minor;
+	__be16			dp_protocol_minor_mask; /* bitmask */
+	__be32			dp_reserved1;
+	__be64			dp_ack_seq;
+	__be32			dp_credit;		/* non-zero enables flow ctl */
+};
+
+struct rds_ib_send_work {
+	void			*s_op;
+	struct ib_send_wr	s_wr;
+	struct ib_sge		s_sge[RDS_IB_MAX_SGE];
+	unsigned long		s_queued;
+};
+
+struct rds_ib_recv_work {
+	struct rds_ib_incoming 	*r_ibinc;
+	struct rds_page_frag	*r_frag;
+	struct ib_recv_wr	r_wr;
+	struct ib_sge		r_sge[2];
+};
+
+struct rds_ib_work_ring {
+	u32		w_nr;
+	u32		w_alloc_ptr;
+	u32		w_alloc_ctr;
+	u32		w_free_ptr;
+	atomic_t	w_free_ctr;
+};
+
+struct rds_ib_device;
+
+struct rds_ib_connection {
+
+	struct list_head	ib_node;
+	struct rds_ib_device	*rds_ibdev;
+	struct rds_connection	*conn;
+
+	/* alphabet soup, IBTA style */
+	struct rdma_cm_id	*i_cm_id;
+	struct ib_pd		*i_pd;
+	struct ib_mr		*i_mr;
+	struct ib_cq		*i_send_cq;
+	struct ib_cq		*i_recv_cq;
+
+	/* tx */
+	struct rds_ib_work_ring	i_send_ring;
+	struct rm_data_op	*i_data_op;
+	struct rds_header	*i_send_hdrs;
+	u64			i_send_hdrs_dma;
+	struct rds_ib_send_work *i_sends;
+	atomic_t		i_signaled_sends;
+
+	/* rx */
+	struct tasklet_struct	i_recv_tasklet;
+	struct mutex		i_recv_mutex;
+	struct rds_ib_work_ring	i_recv_ring;
+	struct rds_ib_incoming	*i_ibinc;
+	u32			i_recv_data_rem;
+	struct rds_header	*i_recv_hdrs;
+	u64			i_recv_hdrs_dma;
+	struct rds_ib_recv_work *i_recvs;
+	u64			i_ack_recv;	/* last ACK received */
+	struct rds_ib_refill_cache i_cache_incs;
+	struct rds_ib_refill_cache i_cache_frags;
+
+	/* sending acks */
+	unsigned long		i_ack_flags;
+#ifdef KERNEL_HAS_ATOMIC64
+	atomic64_t		i_ack_next;	/* next ACK to send */
+#else
+	spinlock_t		i_ack_lock;	/* protect i_ack_next */
+	u64			i_ack_next;	/* next ACK to send */
+#endif
+	struct rds_header	*i_ack;
+	struct ib_send_wr	i_ack_wr;
+	struct ib_sge		i_ack_sge;
+	u64			i_ack_dma;
+	unsigned long		i_ack_queued;
+
+	/* Flow control related information
+	 *
+	 * Our algorithm uses a pair variables that we need to access
+	 * atomically - one for the send credits, and one posted
+	 * recv credits we need to transfer to remote.
+	 * Rather than protect them using a slow spinlock, we put both into
+	 * a single atomic_t and update it using cmpxchg
+	 */
+	atomic_t		i_credits;
+
+	/* Protocol version specific information */
+	unsigned int		i_flowctl:1;	/* enable/disable flow ctl */
+
+	/* Batched completions */
+	unsigned int		i_unsignaled_wrs;
+};
+
+/* This assumes that atomic_t is at least 32 bits */
+#define IB_GET_SEND_CREDITS(v)	((v) & 0xffff)
+#define IB_GET_POST_CREDITS(v)	((v) >> 16)
+#define IB_SET_SEND_CREDITS(v)	((v) & 0xffff)
+#define IB_SET_POST_CREDITS(v)	((v) << 16)
+
+struct rds_ib_ipaddr {
+	struct list_head	list;
+	__be32			ipaddr;
+};
+
+struct rds_ib_device {
+	struct list_head	list;
+	struct list_head	ipaddr_list;
+	struct list_head	conn_list;
+	struct ib_device	*dev;
+	struct ib_pd		*pd;
+	struct ib_mr		*mr;
+	struct rds_ib_mr_pool	*mr_pool;
+	unsigned int		fmr_max_remaps;
+	unsigned int		max_fmrs;
+	int			max_sge;
+	unsigned int		max_wrs;
+	unsigned int		max_initiator_depth;
+	unsigned int		max_responder_resources;
+	spinlock_t		spinlock;	/* protect the above */
+	atomic_t		refcount;
+	struct work_struct	free_work;
+};
+
+#define ibdev_to_node(ibdev) dev_to_node(ibdev->dma_device)
+#define rdsibdev_to_node(rdsibdev) ibdev_to_node(rdsibdev->dev)
+
+/* bits for i_ack_flags */
+#define IB_ACK_IN_FLIGHT	0
+#define IB_ACK_REQUESTED	1
+
+/* Magic WR_ID for ACKs */
+#define RDS_IB_ACK_WR_ID	(~(u64) 0)
+
+struct rds_ib_statistics {
+	uint64_t	s_ib_connect_raced;
+	uint64_t	s_ib_listen_closed_stale;
+	uint64_t	s_ib_tx_cq_call;
+	uint64_t	s_ib_tx_cq_event;
+	uint64_t	s_ib_tx_ring_full;
+	uint64_t	s_ib_tx_throttle;
+	uint64_t	s_ib_tx_sg_mapping_failure;
+	uint64_t	s_ib_tx_stalled;
+	uint64_t	s_ib_tx_credit_updates;
+	uint64_t	s_ib_rx_cq_call;
+	uint64_t	s_ib_rx_cq_event;
+	uint64_t	s_ib_rx_ring_empty;
+	uint64_t	s_ib_rx_refill_from_cq;
+	uint64_t	s_ib_rx_refill_from_thread;
+	uint64_t	s_ib_rx_alloc_limit;
+	uint64_t	s_ib_rx_credit_updates;
+	uint64_t	s_ib_ack_sent;
+	uint64_t	s_ib_ack_send_failure;
+	uint64_t	s_ib_ack_send_delayed;
+	uint64_t	s_ib_ack_send_piggybacked;
+	uint64_t	s_ib_ack_received;
+	uint64_t	s_ib_rdma_mr_alloc;
+	uint64_t	s_ib_rdma_mr_free;
+	uint64_t	s_ib_rdma_mr_used;
+	uint64_t	s_ib_rdma_mr_pool_flush;
+	uint64_t	s_ib_rdma_mr_pool_wait;
+	uint64_t	s_ib_rdma_mr_pool_depleted;
+	uint64_t	s_ib_atomic_cswp;
+	uint64_t	s_ib_atomic_fadd;
+};
+
+extern struct workqueue_struct *rds_ib_wq;
+
+/*
+ * Fake ib_dma_sync_sg_for_{cpu,device} as long as ib_verbs.h
+ * doesn't define it.
+ */
+static inline void rds_ib_dma_sync_sg_for_cpu(struct ib_device *dev,
+		struct scatterlist *sg, unsigned int sg_dma_len, int direction)
+{
+	unsigned int i;
+
+	for (i = 0; i < sg_dma_len; ++i) {
+		ib_dma_sync_single_for_cpu(dev,
+				ib_sg_dma_address(dev, &sg[i]),
+				ib_sg_dma_len(dev, &sg[i]),
+				direction);
+	}
+}
+#define ib_dma_sync_sg_for_cpu	rds_ib_dma_sync_sg_for_cpu
+
+static inline void rds_ib_dma_sync_sg_for_device(struct ib_device *dev,
+		struct scatterlist *sg, unsigned int sg_dma_len, int direction)
+{
+	unsigned int i;
+
+	for (i = 0; i < sg_dma_len; ++i) {
+		ib_dma_sync_single_for_device(dev,
+				ib_sg_dma_address(dev, &sg[i]),
+				ib_sg_dma_len(dev, &sg[i]),
+				direction);
+	}
+}
+#define ib_dma_sync_sg_for_device	rds_ib_dma_sync_sg_for_device
+
+
+/* ib.c */
+extern struct rds_transport rds_ib_transport;
+struct rds_ib_device *rds_ib_get_client_data(struct ib_device *device);
+void rds_ib_dev_put(struct rds_ib_device *rds_ibdev);
+extern struct ib_client rds_ib_client;
+
+extern unsigned int fmr_message_size;
+extern unsigned int rds_ib_retry_count;
+
+extern spinlock_t ib_nodev_conns_lock;
+extern struct list_head ib_nodev_conns;
+
+/* ib_cm.c */
+int rds_ib_conn_alloc(struct rds_connection *conn, gfp_t gfp);
+void rds_ib_conn_free(void *arg);
+int rds_ib_conn_connect(struct rds_connection *conn);
+void rds_ib_conn_shutdown(struct rds_connection *conn);
+void rds_ib_state_change(struct sock *sk);
+int rds_ib_listen_init(void);
+void rds_ib_listen_stop(void);
+void __rds_ib_conn_error(struct rds_connection *conn, const char *, ...);
+int rds_ib_cm_handle_connect(struct rdma_cm_id *cm_id,
+			     struct rdma_cm_event *event);
+int rds_ib_cm_initiate_connect(struct rdma_cm_id *cm_id);
+void rds_ib_cm_connect_complete(struct rds_connection *conn,
+				struct rdma_cm_event *event);
+
+
+#define rds_ib_conn_error(conn, fmt...) \
+	__rds_ib_conn_error(conn, KERN_WARNING "RDS/IB: " fmt)
+
+/* ib_rdma.c */
+int rds_ib_update_ipaddr(struct rds_ib_device *rds_ibdev, __be32 ipaddr);
+void rds_ib_add_conn(struct rds_ib_device *rds_ibdev, struct rds_connection *conn);
+void rds_ib_remove_conn(struct rds_ib_device *rds_ibdev, struct rds_connection *conn);
+void rds_ib_destroy_nodev_conns(void);
+struct rds_ib_mr_pool *rds_ib_create_mr_pool(struct rds_ib_device *);
+void rds_ib_get_mr_info(struct rds_ib_device *rds_ibdev, struct rds_info_rdma_connection *iinfo);
+void rds_ib_destroy_mr_pool(struct rds_ib_mr_pool *);
+void *rds_ib_get_mr(struct scatterlist *sg, unsigned long nents,
+		    struct rds_sock *rs, u32 *key_ret);
+void rds_ib_sync_mr(void *trans_private, int dir);
+void rds_ib_free_mr(void *trans_private, int invalidate);
+void rds_ib_flush_mrs(void);
+
+/* ib_recv.c */
+int rds_ib_recv_init(void);
+void rds_ib_recv_exit(void);
+int rds_ib_recv(struct rds_connection *conn);
+int rds_ib_recv_alloc_caches(struct rds_ib_connection *ic);
+void rds_ib_recv_free_caches(struct rds_ib_connection *ic);
+void rds_ib_recv_refill(struct rds_connection *conn, int prefill);
+void rds_ib_inc_free(struct rds_incoming *inc);
+int rds_ib_inc_copy_to_user(struct rds_incoming *inc, struct iov_iter *to);
+void rds_ib_recv_cq_comp_handler(struct ib_cq *cq, void *context);
+void rds_ib_recv_tasklet_fn(unsigned long data);
+void rds_ib_recv_init_ring(struct rds_ib_connection *ic);
+void rds_ib_recv_clear_ring(struct rds_ib_connection *ic);
+void rds_ib_recv_init_ack(struct rds_ib_connection *ic);
+void rds_ib_attempt_ack(struct rds_ib_connection *ic);
+void rds_ib_ack_send_complete(struct rds_ib_connection *ic);
+u64 rds_ib_piggyb_ack(struct rds_ib_connection *ic);
+
+/* ib_ring.c */
+void rds_ib_ring_init(struct rds_ib_work_ring *ring, u32 nr);
+void rds_ib_ring_resize(struct rds_ib_work_ring *ring, u32 nr);
+u32 rds_ib_ring_alloc(struct rds_ib_work_ring *ring, u32 val, u32 *pos);
+void rds_ib_ring_free(struct rds_ib_work_ring *ring, u32 val);
+void rds_ib_ring_unalloc(struct rds_ib_work_ring *ring, u32 val);
+int rds_ib_ring_empty(struct rds_ib_work_ring *ring);
+int rds_ib_ring_low(struct rds_ib_work_ring *ring);
+u32 rds_ib_ring_oldest(struct rds_ib_work_ring *ring);
+u32 rds_ib_ring_completed(struct rds_ib_work_ring *ring, u32 wr_id, u32 oldest);
+extern wait_queue_head_t rds_ib_ring_empty_wait;
+
+/* ib_send.c */
+char *rds_ib_wc_status_str(enum ib_wc_status status);
+void rds_ib_xmit_complete(struct rds_connection *conn);
+int rds_ib_xmit(struct rds_connection *conn, struct rds_message *rm,
+		unsigned int hdr_off, unsigned int sg, unsigned int off);
+void rds_ib_send_cq_comp_handler(struct ib_cq *cq, void *context);
+void rds_ib_send_init_ring(struct rds_ib_connection *ic);
+void rds_ib_send_clear_ring(struct rds_ib_connection *ic);
+int rds_ib_xmit_rdma(struct rds_connection *conn, struct rm_rdma_op *op);
+void rds_ib_send_add_credits(struct rds_connection *conn, unsigned int credits);
+void rds_ib_advertise_credits(struct rds_connection *conn, unsigned int posted);
+int rds_ib_send_grab_credits(struct rds_ib_connection *ic, u32 wanted,
+			     u32 *adv_credits, int need_posted, int max_posted);
+int rds_ib_xmit_atomic(struct rds_connection *conn, struct rm_atomic_op *op);
+
+/* ib_stats.c */
+DECLARE_PER_CPU(struct rds_ib_statistics, rds_ib_stats);
+#define rds_ib_stats_inc(member) rds_stats_inc_which(rds_ib_stats, member)
+unsigned int rds_ib_stats_info_copy(struct rds_info_iterator *iter,
+				    unsigned int avail);
+
+/* ib_sysctl.c */
+int rds_ib_sysctl_init(void);
+void rds_ib_sysctl_exit(void);
+extern unsigned long rds_ib_sysctl_max_send_wr;
+extern unsigned long rds_ib_sysctl_max_recv_wr;
+extern unsigned long rds_ib_sysctl_max_unsig_wrs;
+extern unsigned long rds_ib_sysctl_max_unsig_bytes;
+extern unsigned long rds_ib_sysctl_max_recv_allocation;
+extern unsigned int rds_ib_sysctl_flow_control;
+
+#endif
diff --git a/net/rds/ib_cm.c b/net/rds/ib_cm.c
new file mode 100644
index 000000000..8a09ee7db
--- /dev/null
+++ b/net/rds/ib_cm.c
@@ -0,0 +1,838 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/in.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include <linux/ratelimit.h>
+
+#include "rds.h"
+#include "ib.h"
+
+static char *rds_ib_event_type_strings[] = {
+#define RDS_IB_EVENT_STRING(foo) \
+		[IB_EVENT_##foo] = __stringify(IB_EVENT_##foo)
+	RDS_IB_EVENT_STRING(CQ_ERR),
+	RDS_IB_EVENT_STRING(QP_FATAL),
+	RDS_IB_EVENT_STRING(QP_REQ_ERR),
+	RDS_IB_EVENT_STRING(QP_ACCESS_ERR),
+	RDS_IB_EVENT_STRING(COMM_EST),
+	RDS_IB_EVENT_STRING(SQ_DRAINED),
+	RDS_IB_EVENT_STRING(PATH_MIG),
+	RDS_IB_EVENT_STRING(PATH_MIG_ERR),
+	RDS_IB_EVENT_STRING(DEVICE_FATAL),
+	RDS_IB_EVENT_STRING(PORT_ACTIVE),
+	RDS_IB_EVENT_STRING(PORT_ERR),
+	RDS_IB_EVENT_STRING(LID_CHANGE),
+	RDS_IB_EVENT_STRING(PKEY_CHANGE),
+	RDS_IB_EVENT_STRING(SM_CHANGE),
+	RDS_IB_EVENT_STRING(SRQ_ERR),
+	RDS_IB_EVENT_STRING(SRQ_LIMIT_REACHED),
+	RDS_IB_EVENT_STRING(QP_LAST_WQE_REACHED),
+	RDS_IB_EVENT_STRING(CLIENT_REREGISTER),
+#undef RDS_IB_EVENT_STRING
+};
+
+static char *rds_ib_event_str(enum ib_event_type type)
+{
+	return rds_str_array(rds_ib_event_type_strings,
+			     ARRAY_SIZE(rds_ib_event_type_strings), type);
+};
+
+/*
+ * Set the selected protocol version
+ */
+static void rds_ib_set_protocol(struct rds_connection *conn, unsigned int version)
+{
+	conn->c_version = version;
+}
+
+/*
+ * Set up flow control
+ */
+static void rds_ib_set_flow_control(struct rds_connection *conn, u32 credits)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+
+	if (rds_ib_sysctl_flow_control && credits != 0) {
+		/* We're doing flow control */
+		ic->i_flowctl = 1;
+		rds_ib_send_add_credits(conn, credits);
+	} else {
+		ic->i_flowctl = 0;
+	}
+}
+
+/*
+ * Tune RNR behavior. Without flow control, we use a rather
+ * low timeout, but not the absolute minimum - this should
+ * be tunable.
+ *
+ * We already set the RNR retry count to 7 (which is the
+ * smallest infinite number :-) above.
+ * If flow control is off, we want to change this back to 0
+ * so that we learn quickly when our credit accounting is
+ * buggy.
+ *
+ * Caller passes in a qp_attr pointer - don't waste stack spacv
+ * by allocation this twice.
+ */
+static void
+rds_ib_tune_rnr(struct rds_ib_connection *ic, struct ib_qp_attr *attr)
+{
+	int ret;
+
+	attr->min_rnr_timer = IB_RNR_TIMER_000_32;
+	ret = ib_modify_qp(ic->i_cm_id->qp, attr, IB_QP_MIN_RNR_TIMER);
+	if (ret)
+		printk(KERN_NOTICE "ib_modify_qp(IB_QP_MIN_RNR_TIMER): err=%d\n", -ret);
+}
+
+/*
+ * Connection established.
+ * We get here for both outgoing and incoming connection.
+ */
+void rds_ib_cm_connect_complete(struct rds_connection *conn, struct rdma_cm_event *event)
+{
+	const struct rds_ib_connect_private *dp = NULL;
+	struct rds_ib_connection *ic = conn->c_transport_data;
+	struct ib_qp_attr qp_attr;
+	int err;
+
+	if (event->param.conn.private_data_len >= sizeof(*dp)) {
+		dp = event->param.conn.private_data;
+
+		/* make sure it isn't empty data */
+		if (dp->dp_protocol_major) {
+			rds_ib_set_protocol(conn,
+				RDS_PROTOCOL(dp->dp_protocol_major,
+				dp->dp_protocol_minor));
+			rds_ib_set_flow_control(conn, be32_to_cpu(dp->dp_credit));
+		}
+	}
+
+	if (conn->c_version < RDS_PROTOCOL(3,1)) {
+		printk(KERN_NOTICE "RDS/IB: Connection to %pI4 version %u.%u failed,"
+		       " no longer supported\n",
+		       &conn->c_faddr,
+		       RDS_PROTOCOL_MAJOR(conn->c_version),
+		       RDS_PROTOCOL_MINOR(conn->c_version));
+		rds_conn_destroy(conn);
+		return;
+	} else {
+		printk(KERN_NOTICE "RDS/IB: connected to %pI4 version %u.%u%s\n",
+		       &conn->c_faddr,
+		       RDS_PROTOCOL_MAJOR(conn->c_version),
+		       RDS_PROTOCOL_MINOR(conn->c_version),
+		       ic->i_flowctl ? ", flow control" : "");
+	}
+
+	/*
+	 * Init rings and fill recv. this needs to wait until protocol negotiation
+	 * is complete, since ring layout is different from 3.0 to 3.1.
+	 */
+	rds_ib_send_init_ring(ic);
+	rds_ib_recv_init_ring(ic);
+	/* Post receive buffers - as a side effect, this will update
+	 * the posted credit count. */
+	rds_ib_recv_refill(conn, 1);
+
+	/* Tune RNR behavior */
+	rds_ib_tune_rnr(ic, &qp_attr);
+
+	qp_attr.qp_state = IB_QPS_RTS;
+	err = ib_modify_qp(ic->i_cm_id->qp, &qp_attr, IB_QP_STATE);
+	if (err)
+		printk(KERN_NOTICE "ib_modify_qp(IB_QP_STATE, RTS): err=%d\n", err);
+
+	/* update ib_device with this local ipaddr */
+	err = rds_ib_update_ipaddr(ic->rds_ibdev, conn->c_laddr);
+	if (err)
+		printk(KERN_ERR "rds_ib_update_ipaddr failed (%d)\n",
+			err);
+
+	/* If the peer gave us the last packet it saw, process this as if
+	 * we had received a regular ACK. */
+	if (dp) {
+		/* dp structure start is not guaranteed to be 8 bytes aligned.
+		 * Since dp_ack_seq is 64-bit extended load operations can be
+		 * used so go through get_unaligned to avoid unaligned errors.
+		 */
+		__be64 dp_ack_seq = get_unaligned(&dp->dp_ack_seq);
+
+		if (dp_ack_seq)
+			rds_send_drop_acked(conn, be64_to_cpu(dp_ack_seq),
+					    NULL);
+	}
+
+	rds_connect_complete(conn);
+}
+
+static void rds_ib_cm_fill_conn_param(struct rds_connection *conn,
+			struct rdma_conn_param *conn_param,
+			struct rds_ib_connect_private *dp,
+			u32 protocol_version,
+			u32 max_responder_resources,
+			u32 max_initiator_depth)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+	struct rds_ib_device *rds_ibdev = ic->rds_ibdev;
+
+	memset(conn_param, 0, sizeof(struct rdma_conn_param));
+
+	conn_param->responder_resources =
+		min_t(u32, rds_ibdev->max_responder_resources, max_responder_resources);
+	conn_param->initiator_depth =
+		min_t(u32, rds_ibdev->max_initiator_depth, max_initiator_depth);
+	conn_param->retry_count = min_t(unsigned int, rds_ib_retry_count, 7);
+	conn_param->rnr_retry_count = 7;
+
+	if (dp) {
+		memset(dp, 0, sizeof(*dp));
+		dp->dp_saddr = conn->c_laddr;
+		dp->dp_daddr = conn->c_faddr;
+		dp->dp_protocol_major = RDS_PROTOCOL_MAJOR(protocol_version);
+		dp->dp_protocol_minor = RDS_PROTOCOL_MINOR(protocol_version);
+		dp->dp_protocol_minor_mask = cpu_to_be16(RDS_IB_SUPPORTED_PROTOCOLS);
+		dp->dp_ack_seq = rds_ib_piggyb_ack(ic);
+
+		/* Advertise flow control */
+		if (ic->i_flowctl) {
+			unsigned int credits;
+
+			credits = IB_GET_POST_CREDITS(atomic_read(&ic->i_credits));
+			dp->dp_credit = cpu_to_be32(credits);
+			atomic_sub(IB_SET_POST_CREDITS(credits), &ic->i_credits);
+		}
+
+		conn_param->private_data = dp;
+		conn_param->private_data_len = sizeof(*dp);
+	}
+}
+
+static void rds_ib_cq_event_handler(struct ib_event *event, void *data)
+{
+	rdsdebug("event %u (%s) data %p\n",
+		 event->event, rds_ib_event_str(event->event), data);
+}
+
+static void rds_ib_qp_event_handler(struct ib_event *event, void *data)
+{
+	struct rds_connection *conn = data;
+	struct rds_ib_connection *ic = conn->c_transport_data;
+
+	rdsdebug("conn %p ic %p event %u (%s)\n", conn, ic, event->event,
+		 rds_ib_event_str(event->event));
+
+	switch (event->event) {
+	case IB_EVENT_COMM_EST:
+		rdma_notify(ic->i_cm_id, IB_EVENT_COMM_EST);
+		break;
+	default:
+		rdsdebug("Fatal QP Event %u (%s) "
+			"- connection %pI4->%pI4, reconnecting\n",
+			event->event, rds_ib_event_str(event->event),
+			&conn->c_laddr, &conn->c_faddr);
+		rds_conn_drop(conn);
+		break;
+	}
+}
+
+/*
+ * This needs to be very careful to not leave IS_ERR pointers around for
+ * cleanup to trip over.
+ */
+static int rds_ib_setup_qp(struct rds_connection *conn)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+	struct ib_device *dev = ic->i_cm_id->device;
+	struct ib_qp_init_attr attr;
+	struct rds_ib_device *rds_ibdev;
+	int ret;
+
+	/*
+	 * It's normal to see a null device if an incoming connection races
+	 * with device removal, so we don't print a warning.
+	 */
+	rds_ibdev = rds_ib_get_client_data(dev);
+	if (!rds_ibdev)
+		return -EOPNOTSUPP;
+
+	/* add the conn now so that connection establishment has the dev */
+	rds_ib_add_conn(rds_ibdev, conn);
+
+	if (rds_ibdev->max_wrs < ic->i_send_ring.w_nr + 1)
+		rds_ib_ring_resize(&ic->i_send_ring, rds_ibdev->max_wrs - 1);
+	if (rds_ibdev->max_wrs < ic->i_recv_ring.w_nr + 1)
+		rds_ib_ring_resize(&ic->i_recv_ring, rds_ibdev->max_wrs - 1);
+
+	/* Protection domain and memory range */
+	ic->i_pd = rds_ibdev->pd;
+	ic->i_mr = rds_ibdev->mr;
+
+	ic->i_send_cq = ib_create_cq(dev, rds_ib_send_cq_comp_handler,
+				     rds_ib_cq_event_handler, conn,
+				     ic->i_send_ring.w_nr + 1, 0);
+	if (IS_ERR(ic->i_send_cq)) {
+		ret = PTR_ERR(ic->i_send_cq);
+		ic->i_send_cq = NULL;
+		rdsdebug("ib_create_cq send failed: %d\n", ret);
+		goto out;
+	}
+
+	ic->i_recv_cq = ib_create_cq(dev, rds_ib_recv_cq_comp_handler,
+				     rds_ib_cq_event_handler, conn,
+				     ic->i_recv_ring.w_nr, 0);
+	if (IS_ERR(ic->i_recv_cq)) {
+		ret = PTR_ERR(ic->i_recv_cq);
+		ic->i_recv_cq = NULL;
+		rdsdebug("ib_create_cq recv failed: %d\n", ret);
+		goto out;
+	}
+
+	ret = ib_req_notify_cq(ic->i_send_cq, IB_CQ_NEXT_COMP);
+	if (ret) {
+		rdsdebug("ib_req_notify_cq send failed: %d\n", ret);
+		goto out;
+	}
+
+	ret = ib_req_notify_cq(ic->i_recv_cq, IB_CQ_SOLICITED);
+	if (ret) {
+		rdsdebug("ib_req_notify_cq recv failed: %d\n", ret);
+		goto out;
+	}
+
+	/* XXX negotiate max send/recv with remote? */
+	memset(&attr, 0, sizeof(attr));
+	attr.event_handler = rds_ib_qp_event_handler;
+	attr.qp_context = conn;
+	/* + 1 to allow for the single ack message */
+	attr.cap.max_send_wr = ic->i_send_ring.w_nr + 1;
+	attr.cap.max_recv_wr = ic->i_recv_ring.w_nr + 1;
+	attr.cap.max_send_sge = rds_ibdev->max_sge;
+	attr.cap.max_recv_sge = RDS_IB_RECV_SGE;
+	attr.sq_sig_type = IB_SIGNAL_REQ_WR;
+	attr.qp_type = IB_QPT_RC;
+	attr.send_cq = ic->i_send_cq;
+	attr.recv_cq = ic->i_recv_cq;
+
+	/*
+	 * XXX this can fail if max_*_wr is too large?  Are we supposed
+	 * to back off until we get a value that the hardware can support?
+	 */
+	ret = rdma_create_qp(ic->i_cm_id, ic->i_pd, &attr);
+	if (ret) {
+		rdsdebug("rdma_create_qp failed: %d\n", ret);
+		goto out;
+	}
+
+	ic->i_send_hdrs = ib_dma_alloc_coherent(dev,
+					   ic->i_send_ring.w_nr *
+						sizeof(struct rds_header),
+					   &ic->i_send_hdrs_dma, GFP_KERNEL);
+	if (!ic->i_send_hdrs) {
+		ret = -ENOMEM;
+		rdsdebug("ib_dma_alloc_coherent send failed\n");
+		goto out;
+	}
+
+	ic->i_recv_hdrs = ib_dma_alloc_coherent(dev,
+					   ic->i_recv_ring.w_nr *
+						sizeof(struct rds_header),
+					   &ic->i_recv_hdrs_dma, GFP_KERNEL);
+	if (!ic->i_recv_hdrs) {
+		ret = -ENOMEM;
+		rdsdebug("ib_dma_alloc_coherent recv failed\n");
+		goto out;
+	}
+
+	ic->i_ack = ib_dma_alloc_coherent(dev, sizeof(struct rds_header),
+				       &ic->i_ack_dma, GFP_KERNEL);
+	if (!ic->i_ack) {
+		ret = -ENOMEM;
+		rdsdebug("ib_dma_alloc_coherent ack failed\n");
+		goto out;
+	}
+
+	ic->i_sends = vzalloc_node(ic->i_send_ring.w_nr * sizeof(struct rds_ib_send_work),
+				   ibdev_to_node(dev));
+	if (!ic->i_sends) {
+		ret = -ENOMEM;
+		rdsdebug("send allocation failed\n");
+		goto out;
+	}
+
+	ic->i_recvs = vzalloc_node(ic->i_recv_ring.w_nr * sizeof(struct rds_ib_recv_work),
+				   ibdev_to_node(dev));
+	if (!ic->i_recvs) {
+		ret = -ENOMEM;
+		rdsdebug("recv allocation failed\n");
+		goto out;
+	}
+
+	rds_ib_recv_init_ack(ic);
+
+	rdsdebug("conn %p pd %p mr %p cq %p %p\n", conn, ic->i_pd, ic->i_mr,
+		 ic->i_send_cq, ic->i_recv_cq);
+
+out:
+	rds_ib_dev_put(rds_ibdev);
+	return ret;
+}
+
+static u32 rds_ib_protocol_compatible(struct rdma_cm_event *event)
+{
+	const struct rds_ib_connect_private *dp = event->param.conn.private_data;
+	u16 common;
+	u32 version = 0;
+
+	/*
+	 * rdma_cm private data is odd - when there is any private data in the
+	 * request, we will be given a pretty large buffer without telling us the
+	 * original size. The only way to tell the difference is by looking at
+	 * the contents, which are initialized to zero.
+	 * If the protocol version fields aren't set, this is a connection attempt
+	 * from an older version. This could could be 3.0 or 2.0 - we can't tell.
+	 * We really should have changed this for OFED 1.3 :-(
+	 */
+
+	/* Be paranoid. RDS always has privdata */
+	if (!event->param.conn.private_data_len) {
+		printk(KERN_NOTICE "RDS incoming connection has no private data, "
+			"rejecting\n");
+		return 0;
+	}
+
+	/* Even if len is crap *now* I still want to check it. -ASG */
+	if (event->param.conn.private_data_len < sizeof (*dp) ||
+	    dp->dp_protocol_major == 0)
+		return RDS_PROTOCOL_3_0;
+
+	common = be16_to_cpu(dp->dp_protocol_minor_mask) & RDS_IB_SUPPORTED_PROTOCOLS;
+	if (dp->dp_protocol_major == 3 && common) {
+		version = RDS_PROTOCOL_3_0;
+		while ((common >>= 1) != 0)
+			version++;
+	} else
+		printk_ratelimited(KERN_NOTICE "RDS: Connection from %pI4 using incompatible protocol version %u.%u\n",
+				&dp->dp_saddr,
+				dp->dp_protocol_major,
+				dp->dp_protocol_minor);
+	return version;
+}
+
+int rds_ib_cm_handle_connect(struct rdma_cm_id *cm_id,
+				    struct rdma_cm_event *event)
+{
+	__be64 lguid = cm_id->route.path_rec->sgid.global.interface_id;
+	__be64 fguid = cm_id->route.path_rec->dgid.global.interface_id;
+	const struct rds_ib_connect_private *dp = event->param.conn.private_data;
+	struct rds_ib_connect_private dp_rep;
+	struct rds_connection *conn = NULL;
+	struct rds_ib_connection *ic = NULL;
+	struct rdma_conn_param conn_param;
+	u32 version;
+	int err = 1, destroy = 1;
+
+	/* Check whether the remote protocol version matches ours. */
+	version = rds_ib_protocol_compatible(event);
+	if (!version)
+		goto out;
+
+	rdsdebug("saddr %pI4 daddr %pI4 RDSv%u.%u lguid 0x%llx fguid "
+		 "0x%llx\n", &dp->dp_saddr, &dp->dp_daddr,
+		 RDS_PROTOCOL_MAJOR(version), RDS_PROTOCOL_MINOR(version),
+		 (unsigned long long)be64_to_cpu(lguid),
+		 (unsigned long long)be64_to_cpu(fguid));
+
+	conn = rds_conn_create(dp->dp_daddr, dp->dp_saddr, &rds_ib_transport,
+			       GFP_KERNEL);
+	if (IS_ERR(conn)) {
+		rdsdebug("rds_conn_create failed (%ld)\n", PTR_ERR(conn));
+		conn = NULL;
+		goto out;
+	}
+
+	/*
+	 * The connection request may occur while the
+	 * previous connection exist, e.g. in case of failover.
+	 * But as connections may be initiated simultaneously
+	 * by both hosts, we have a random backoff mechanism -
+	 * see the comment above rds_queue_reconnect()
+	 */
+	mutex_lock(&conn->c_cm_lock);
+	if (!rds_conn_transition(conn, RDS_CONN_DOWN, RDS_CONN_CONNECTING)) {
+		if (rds_conn_state(conn) == RDS_CONN_UP) {
+			rdsdebug("incoming connect while connecting\n");
+			rds_conn_drop(conn);
+			rds_ib_stats_inc(s_ib_listen_closed_stale);
+		} else
+		if (rds_conn_state(conn) == RDS_CONN_CONNECTING) {
+			/* Wait and see - our connect may still be succeeding */
+			rds_ib_stats_inc(s_ib_connect_raced);
+		}
+		goto out;
+	}
+
+	ic = conn->c_transport_data;
+
+	rds_ib_set_protocol(conn, version);
+	rds_ib_set_flow_control(conn, be32_to_cpu(dp->dp_credit));
+
+	/* If the peer gave us the last packet it saw, process this as if
+	 * we had received a regular ACK. */
+	if (dp->dp_ack_seq)
+		rds_send_drop_acked(conn, be64_to_cpu(dp->dp_ack_seq), NULL);
+
+	BUG_ON(cm_id->context);
+	BUG_ON(ic->i_cm_id);
+
+	ic->i_cm_id = cm_id;
+	cm_id->context = conn;
+
+	/* We got halfway through setting up the ib_connection, if we
+	 * fail now, we have to take the long route out of this mess. */
+	destroy = 0;
+
+	err = rds_ib_setup_qp(conn);
+	if (err) {
+		rds_ib_conn_error(conn, "rds_ib_setup_qp failed (%d)\n", err);
+		goto out;
+	}
+
+	rds_ib_cm_fill_conn_param(conn, &conn_param, &dp_rep, version,
+		event->param.conn.responder_resources,
+		event->param.conn.initiator_depth);
+
+	/* rdma_accept() calls rdma_reject() internally if it fails */
+	err = rdma_accept(cm_id, &conn_param);
+	if (err)
+		rds_ib_conn_error(conn, "rdma_accept failed (%d)\n", err);
+
+out:
+	if (conn)
+		mutex_unlock(&conn->c_cm_lock);
+	if (err)
+		rdma_reject(cm_id, NULL, 0);
+	return destroy;
+}
+
+
+int rds_ib_cm_initiate_connect(struct rdma_cm_id *cm_id)
+{
+	struct rds_connection *conn = cm_id->context;
+	struct rds_ib_connection *ic = conn->c_transport_data;
+	struct rdma_conn_param conn_param;
+	struct rds_ib_connect_private dp;
+	int ret;
+
+	/* If the peer doesn't do protocol negotiation, we must
+	 * default to RDSv3.0 */
+	rds_ib_set_protocol(conn, RDS_PROTOCOL_3_0);
+	ic->i_flowctl = rds_ib_sysctl_flow_control;	/* advertise flow control */
+
+	ret = rds_ib_setup_qp(conn);
+	if (ret) {
+		rds_ib_conn_error(conn, "rds_ib_setup_qp failed (%d)\n", ret);
+		goto out;
+	}
+
+	rds_ib_cm_fill_conn_param(conn, &conn_param, &dp, RDS_PROTOCOL_VERSION,
+		UINT_MAX, UINT_MAX);
+	ret = rdma_connect(cm_id, &conn_param);
+	if (ret)
+		rds_ib_conn_error(conn, "rdma_connect failed (%d)\n", ret);
+
+out:
+	/* Beware - returning non-zero tells the rdma_cm to destroy
+	 * the cm_id. We should certainly not do it as long as we still
+	 * "own" the cm_id. */
+	if (ret) {
+		if (ic->i_cm_id == cm_id)
+			ret = 0;
+	}
+	return ret;
+}
+
+int rds_ib_conn_connect(struct rds_connection *conn)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+	struct sockaddr_in src, dest;
+	int ret;
+
+	/* XXX I wonder what affect the port space has */
+	/* delegate cm event handler to rdma_transport */
+	ic->i_cm_id = rdma_create_id(rds_rdma_cm_event_handler, conn,
+				     RDMA_PS_TCP, IB_QPT_RC);
+	if (IS_ERR(ic->i_cm_id)) {
+		ret = PTR_ERR(ic->i_cm_id);
+		ic->i_cm_id = NULL;
+		rdsdebug("rdma_create_id() failed: %d\n", ret);
+		goto out;
+	}
+
+	rdsdebug("created cm id %p for conn %p\n", ic->i_cm_id, conn);
+
+	src.sin_family = AF_INET;
+	src.sin_addr.s_addr = (__force u32)conn->c_laddr;
+	src.sin_port = (__force u16)htons(0);
+
+	dest.sin_family = AF_INET;
+	dest.sin_addr.s_addr = (__force u32)conn->c_faddr;
+	dest.sin_port = (__force u16)htons(RDS_PORT);
+
+	ret = rdma_resolve_addr(ic->i_cm_id, (struct sockaddr *)&src,
+				(struct sockaddr *)&dest,
+				RDS_RDMA_RESOLVE_TIMEOUT_MS);
+	if (ret) {
+		rdsdebug("addr resolve failed for cm id %p: %d\n", ic->i_cm_id,
+			 ret);
+		rdma_destroy_id(ic->i_cm_id);
+		ic->i_cm_id = NULL;
+	}
+
+out:
+	return ret;
+}
+
+/*
+ * This is so careful about only cleaning up resources that were built up
+ * so that it can be called at any point during startup.  In fact it
+ * can be called multiple times for a given connection.
+ */
+void rds_ib_conn_shutdown(struct rds_connection *conn)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+	int err = 0;
+
+	rdsdebug("cm %p pd %p cq %p %p qp %p\n", ic->i_cm_id,
+		 ic->i_pd, ic->i_send_cq, ic->i_recv_cq,
+		 ic->i_cm_id ? ic->i_cm_id->qp : NULL);
+
+	if (ic->i_cm_id) {
+		struct ib_device *dev = ic->i_cm_id->device;
+
+		rdsdebug("disconnecting cm %p\n", ic->i_cm_id);
+		err = rdma_disconnect(ic->i_cm_id);
+		if (err) {
+			/* Actually this may happen quite frequently, when
+			 * an outgoing connect raced with an incoming connect.
+			 */
+			rdsdebug("failed to disconnect, cm: %p err %d\n",
+				ic->i_cm_id, err);
+		}
+
+		/*
+		 * We want to wait for tx and rx completion to finish
+		 * before we tear down the connection, but we have to be
+		 * careful not to get stuck waiting on a send ring that
+		 * only has unsignaled sends in it.  We've shutdown new
+		 * sends before getting here so by waiting for signaled
+		 * sends to complete we're ensured that there will be no
+		 * more tx processing.
+		 */
+		wait_event(rds_ib_ring_empty_wait,
+			   rds_ib_ring_empty(&ic->i_recv_ring) &&
+			   (atomic_read(&ic->i_signaled_sends) == 0));
+		tasklet_kill(&ic->i_recv_tasklet);
+
+		if (ic->i_send_hdrs)
+			ib_dma_free_coherent(dev,
+					   ic->i_send_ring.w_nr *
+						sizeof(struct rds_header),
+					   ic->i_send_hdrs,
+					   ic->i_send_hdrs_dma);
+
+		if (ic->i_recv_hdrs)
+			ib_dma_free_coherent(dev,
+					   ic->i_recv_ring.w_nr *
+						sizeof(struct rds_header),
+					   ic->i_recv_hdrs,
+					   ic->i_recv_hdrs_dma);
+
+		if (ic->i_ack)
+			ib_dma_free_coherent(dev, sizeof(struct rds_header),
+					     ic->i_ack, ic->i_ack_dma);
+
+		if (ic->i_sends)
+			rds_ib_send_clear_ring(ic);
+		if (ic->i_recvs)
+			rds_ib_recv_clear_ring(ic);
+
+		if (ic->i_cm_id->qp)
+			rdma_destroy_qp(ic->i_cm_id);
+		if (ic->i_send_cq)
+			ib_destroy_cq(ic->i_send_cq);
+		if (ic->i_recv_cq)
+			ib_destroy_cq(ic->i_recv_cq);
+		rdma_destroy_id(ic->i_cm_id);
+
+		/*
+		 * Move connection back to the nodev list.
+		 */
+		if (ic->rds_ibdev)
+			rds_ib_remove_conn(ic->rds_ibdev, conn);
+
+		ic->i_cm_id = NULL;
+		ic->i_pd = NULL;
+		ic->i_mr = NULL;
+		ic->i_send_cq = NULL;
+		ic->i_recv_cq = NULL;
+		ic->i_send_hdrs = NULL;
+		ic->i_recv_hdrs = NULL;
+		ic->i_ack = NULL;
+	}
+	BUG_ON(ic->rds_ibdev);
+
+	/* Clear pending transmit */
+	if (ic->i_data_op) {
+		struct rds_message *rm;
+
+		rm = container_of(ic->i_data_op, struct rds_message, data);
+		rds_message_put(rm);
+		ic->i_data_op = NULL;
+	}
+
+	/* Clear the ACK state */
+	clear_bit(IB_ACK_IN_FLIGHT, &ic->i_ack_flags);
+#ifdef KERNEL_HAS_ATOMIC64
+	atomic64_set(&ic->i_ack_next, 0);
+#else
+	ic->i_ack_next = 0;
+#endif
+	ic->i_ack_recv = 0;
+
+	/* Clear flow control state */
+	ic->i_flowctl = 0;
+	atomic_set(&ic->i_credits, 0);
+
+	rds_ib_ring_init(&ic->i_send_ring, rds_ib_sysctl_max_send_wr);
+	rds_ib_ring_init(&ic->i_recv_ring, rds_ib_sysctl_max_recv_wr);
+
+	if (ic->i_ibinc) {
+		rds_inc_put(&ic->i_ibinc->ii_inc);
+		ic->i_ibinc = NULL;
+	}
+
+	vfree(ic->i_sends);
+	ic->i_sends = NULL;
+	vfree(ic->i_recvs);
+	ic->i_recvs = NULL;
+}
+
+int rds_ib_conn_alloc(struct rds_connection *conn, gfp_t gfp)
+{
+	struct rds_ib_connection *ic;
+	unsigned long flags;
+	int ret;
+
+	/* XXX too lazy? */
+	ic = kzalloc(sizeof(struct rds_ib_connection), gfp);
+	if (!ic)
+		return -ENOMEM;
+
+	ret = rds_ib_recv_alloc_caches(ic);
+	if (ret) {
+		kfree(ic);
+		return ret;
+	}
+
+	INIT_LIST_HEAD(&ic->ib_node);
+	tasklet_init(&ic->i_recv_tasklet, rds_ib_recv_tasklet_fn,
+		     (unsigned long) ic);
+	mutex_init(&ic->i_recv_mutex);
+#ifndef KERNEL_HAS_ATOMIC64
+	spin_lock_init(&ic->i_ack_lock);
+#endif
+	atomic_set(&ic->i_signaled_sends, 0);
+
+	/*
+	 * rds_ib_conn_shutdown() waits for these to be emptied so they
+	 * must be initialized before it can be called.
+	 */
+	rds_ib_ring_init(&ic->i_send_ring, rds_ib_sysctl_max_send_wr);
+	rds_ib_ring_init(&ic->i_recv_ring, rds_ib_sysctl_max_recv_wr);
+
+	ic->conn = conn;
+	conn->c_transport_data = ic;
+
+	spin_lock_irqsave(&ib_nodev_conns_lock, flags);
+	list_add_tail(&ic->ib_node, &ib_nodev_conns);
+	spin_unlock_irqrestore(&ib_nodev_conns_lock, flags);
+
+
+	rdsdebug("conn %p conn ic %p\n", conn, conn->c_transport_data);
+	return 0;
+}
+
+/*
+ * Free a connection. Connection must be shut down and not set for reconnect.
+ */
+void rds_ib_conn_free(void *arg)
+{
+	struct rds_ib_connection *ic = arg;
+	spinlock_t	*lock_ptr;
+
+	rdsdebug("ic %p\n", ic);
+
+	/*
+	 * Conn is either on a dev's list or on the nodev list.
+	 * A race with shutdown() or connect() would cause problems
+	 * (since rds_ibdev would change) but that should never happen.
+	 */
+	lock_ptr = ic->rds_ibdev ? &ic->rds_ibdev->spinlock : &ib_nodev_conns_lock;
+
+	spin_lock_irq(lock_ptr);
+	list_del(&ic->ib_node);
+	spin_unlock_irq(lock_ptr);
+
+	rds_ib_recv_free_caches(ic);
+
+	kfree(ic);
+}
+
+
+/*
+ * An error occurred on the connection
+ */
+void
+__rds_ib_conn_error(struct rds_connection *conn, const char *fmt, ...)
+{
+	va_list ap;
+
+	rds_conn_drop(conn);
+
+	va_start(ap, fmt);
+	vprintk(fmt, ap);
+	va_end(ap);
+}
diff --git a/net/rds/ib_rdma.c b/net/rds/ib_rdma.c
new file mode 100644
index 000000000..273b8bff6
--- /dev/null
+++ b/net/rds/ib_rdma.c
@@ -0,0 +1,784 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/rculist.h>
+#include <linux/llist.h>
+
+#include "rds.h"
+#include "ib.h"
+
+static DEFINE_PER_CPU(unsigned long, clean_list_grace);
+#define CLEAN_LIST_BUSY_BIT 0
+
+/*
+ * This is stored as mr->r_trans_private.
+ */
+struct rds_ib_mr {
+	struct rds_ib_device	*device;
+	struct rds_ib_mr_pool	*pool;
+	struct ib_fmr		*fmr;
+
+	struct llist_node	llnode;
+
+	/* unmap_list is for freeing */
+	struct list_head	unmap_list;
+	unsigned int		remap_count;
+
+	struct scatterlist	*sg;
+	unsigned int		sg_len;
+	u64			*dma;
+	int			sg_dma_len;
+};
+
+/*
+ * Our own little FMR pool
+ */
+struct rds_ib_mr_pool {
+	struct mutex		flush_lock;		/* serialize fmr invalidate */
+	struct delayed_work	flush_worker;		/* flush worker */
+
+	atomic_t		item_count;		/* total # of MRs */
+	atomic_t		dirty_count;		/* # dirty of MRs */
+
+	struct llist_head	drop_list;		/* MRs that have reached their max_maps limit */
+	struct llist_head	free_list;		/* unused MRs */
+	struct llist_head	clean_list;		/* global unused & unamapped MRs */
+	wait_queue_head_t	flush_wait;
+
+	atomic_t		free_pinned;		/* memory pinned by free MRs */
+	unsigned long		max_items;
+	unsigned long		max_items_soft;
+	unsigned long		max_free_pinned;
+	struct ib_fmr_attr	fmr_attr;
+};
+
+static int rds_ib_flush_mr_pool(struct rds_ib_mr_pool *pool, int free_all, struct rds_ib_mr **);
+static void rds_ib_teardown_mr(struct rds_ib_mr *ibmr);
+static void rds_ib_mr_pool_flush_worker(struct work_struct *work);
+
+static struct rds_ib_device *rds_ib_get_device(__be32 ipaddr)
+{
+	struct rds_ib_device *rds_ibdev;
+	struct rds_ib_ipaddr *i_ipaddr;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(rds_ibdev, &rds_ib_devices, list) {
+		list_for_each_entry_rcu(i_ipaddr, &rds_ibdev->ipaddr_list, list) {
+			if (i_ipaddr->ipaddr == ipaddr) {
+				atomic_inc(&rds_ibdev->refcount);
+				rcu_read_unlock();
+				return rds_ibdev;
+			}
+		}
+	}
+	rcu_read_unlock();
+
+	return NULL;
+}
+
+static int rds_ib_add_ipaddr(struct rds_ib_device *rds_ibdev, __be32 ipaddr)
+{
+	struct rds_ib_ipaddr *i_ipaddr;
+
+	i_ipaddr = kmalloc(sizeof *i_ipaddr, GFP_KERNEL);
+	if (!i_ipaddr)
+		return -ENOMEM;
+
+	i_ipaddr->ipaddr = ipaddr;
+
+	spin_lock_irq(&rds_ibdev->spinlock);
+	list_add_tail_rcu(&i_ipaddr->list, &rds_ibdev->ipaddr_list);
+	spin_unlock_irq(&rds_ibdev->spinlock);
+
+	return 0;
+}
+
+static void rds_ib_remove_ipaddr(struct rds_ib_device *rds_ibdev, __be32 ipaddr)
+{
+	struct rds_ib_ipaddr *i_ipaddr;
+	struct rds_ib_ipaddr *to_free = NULL;
+
+
+	spin_lock_irq(&rds_ibdev->spinlock);
+	list_for_each_entry_rcu(i_ipaddr, &rds_ibdev->ipaddr_list, list) {
+		if (i_ipaddr->ipaddr == ipaddr) {
+			list_del_rcu(&i_ipaddr->list);
+			to_free = i_ipaddr;
+			break;
+		}
+	}
+	spin_unlock_irq(&rds_ibdev->spinlock);
+
+	if (to_free) {
+		synchronize_rcu();
+		kfree(to_free);
+	}
+}
+
+int rds_ib_update_ipaddr(struct rds_ib_device *rds_ibdev, __be32 ipaddr)
+{
+	struct rds_ib_device *rds_ibdev_old;
+
+	rds_ibdev_old = rds_ib_get_device(ipaddr);
+	if (rds_ibdev_old) {
+		rds_ib_remove_ipaddr(rds_ibdev_old, ipaddr);
+		rds_ib_dev_put(rds_ibdev_old);
+	}
+
+	return rds_ib_add_ipaddr(rds_ibdev, ipaddr);
+}
+
+void rds_ib_add_conn(struct rds_ib_device *rds_ibdev, struct rds_connection *conn)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+
+	/* conn was previously on the nodev_conns_list */
+	spin_lock_irq(&ib_nodev_conns_lock);
+	BUG_ON(list_empty(&ib_nodev_conns));
+	BUG_ON(list_empty(&ic->ib_node));
+	list_del(&ic->ib_node);
+
+	spin_lock(&rds_ibdev->spinlock);
+	list_add_tail(&ic->ib_node, &rds_ibdev->conn_list);
+	spin_unlock(&rds_ibdev->spinlock);
+	spin_unlock_irq(&ib_nodev_conns_lock);
+
+	ic->rds_ibdev = rds_ibdev;
+	atomic_inc(&rds_ibdev->refcount);
+}
+
+void rds_ib_remove_conn(struct rds_ib_device *rds_ibdev, struct rds_connection *conn)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+
+	/* place conn on nodev_conns_list */
+	spin_lock(&ib_nodev_conns_lock);
+
+	spin_lock_irq(&rds_ibdev->spinlock);
+	BUG_ON(list_empty(&ic->ib_node));
+	list_del(&ic->ib_node);
+	spin_unlock_irq(&rds_ibdev->spinlock);
+
+	list_add_tail(&ic->ib_node, &ib_nodev_conns);
+
+	spin_unlock(&ib_nodev_conns_lock);
+
+	ic->rds_ibdev = NULL;
+	rds_ib_dev_put(rds_ibdev);
+}
+
+void rds_ib_destroy_nodev_conns(void)
+{
+	struct rds_ib_connection *ic, *_ic;
+	LIST_HEAD(tmp_list);
+
+	/* avoid calling conn_destroy with irqs off */
+	spin_lock_irq(&ib_nodev_conns_lock);
+	list_splice(&ib_nodev_conns, &tmp_list);
+	spin_unlock_irq(&ib_nodev_conns_lock);
+
+	list_for_each_entry_safe(ic, _ic, &tmp_list, ib_node)
+		rds_conn_destroy(ic->conn);
+}
+
+struct rds_ib_mr_pool *rds_ib_create_mr_pool(struct rds_ib_device *rds_ibdev)
+{
+	struct rds_ib_mr_pool *pool;
+
+	pool = kzalloc(sizeof(*pool), GFP_KERNEL);
+	if (!pool)
+		return ERR_PTR(-ENOMEM);
+
+	init_llist_head(&pool->free_list);
+	init_llist_head(&pool->drop_list);
+	init_llist_head(&pool->clean_list);
+	mutex_init(&pool->flush_lock);
+	init_waitqueue_head(&pool->flush_wait);
+	INIT_DELAYED_WORK(&pool->flush_worker, rds_ib_mr_pool_flush_worker);
+
+	pool->fmr_attr.max_pages = fmr_message_size;
+	pool->fmr_attr.max_maps = rds_ibdev->fmr_max_remaps;
+	pool->fmr_attr.page_shift = PAGE_SHIFT;
+	pool->max_free_pinned = rds_ibdev->max_fmrs * fmr_message_size / 4;
+
+	/* We never allow more than max_items MRs to be allocated.
+	 * When we exceed more than max_items_soft, we start freeing
+	 * items more aggressively.
+	 * Make sure that max_items > max_items_soft > max_items / 2
+	 */
+	pool->max_items_soft = rds_ibdev->max_fmrs * 3 / 4;
+	pool->max_items = rds_ibdev->max_fmrs;
+
+	return pool;
+}
+
+void rds_ib_get_mr_info(struct rds_ib_device *rds_ibdev, struct rds_info_rdma_connection *iinfo)
+{
+	struct rds_ib_mr_pool *pool = rds_ibdev->mr_pool;
+
+	iinfo->rdma_mr_max = pool->max_items;
+	iinfo->rdma_mr_size = pool->fmr_attr.max_pages;
+}
+
+void rds_ib_destroy_mr_pool(struct rds_ib_mr_pool *pool)
+{
+	cancel_delayed_work_sync(&pool->flush_worker);
+	rds_ib_flush_mr_pool(pool, 1, NULL);
+	WARN_ON(atomic_read(&pool->item_count));
+	WARN_ON(atomic_read(&pool->free_pinned));
+	kfree(pool);
+}
+
+static inline struct rds_ib_mr *rds_ib_reuse_fmr(struct rds_ib_mr_pool *pool)
+{
+	struct rds_ib_mr *ibmr = NULL;
+	struct llist_node *ret;
+	unsigned long *flag;
+
+	preempt_disable();
+	flag = this_cpu_ptr(&clean_list_grace);
+	set_bit(CLEAN_LIST_BUSY_BIT, flag);
+	ret = llist_del_first(&pool->clean_list);
+	if (ret)
+		ibmr = llist_entry(ret, struct rds_ib_mr, llnode);
+
+	clear_bit(CLEAN_LIST_BUSY_BIT, flag);
+	preempt_enable();
+	return ibmr;
+}
+
+static inline void wait_clean_list_grace(void)
+{
+	int cpu;
+	unsigned long *flag;
+
+	for_each_online_cpu(cpu) {
+		flag = &per_cpu(clean_list_grace, cpu);
+		while (test_bit(CLEAN_LIST_BUSY_BIT, flag))
+			cpu_relax();
+	}
+}
+
+static struct rds_ib_mr *rds_ib_alloc_fmr(struct rds_ib_device *rds_ibdev)
+{
+	struct rds_ib_mr_pool *pool = rds_ibdev->mr_pool;
+	struct rds_ib_mr *ibmr = NULL;
+	int err = 0, iter = 0;
+
+	if (atomic_read(&pool->dirty_count) >= pool->max_items / 10)
+		schedule_delayed_work(&pool->flush_worker, 10);
+
+	while (1) {
+		ibmr = rds_ib_reuse_fmr(pool);
+		if (ibmr)
+			return ibmr;
+
+		/* No clean MRs - now we have the choice of either
+		 * allocating a fresh MR up to the limit imposed by the
+		 * driver, or flush any dirty unused MRs.
+		 * We try to avoid stalling in the send path if possible,
+		 * so we allocate as long as we're allowed to.
+		 *
+		 * We're fussy with enforcing the FMR limit, though. If the driver
+		 * tells us we can't use more than N fmrs, we shouldn't start
+		 * arguing with it */
+		if (atomic_inc_return(&pool->item_count) <= pool->max_items)
+			break;
+
+		atomic_dec(&pool->item_count);
+
+		if (++iter > 2) {
+			rds_ib_stats_inc(s_ib_rdma_mr_pool_depleted);
+			return ERR_PTR(-EAGAIN);
+		}
+
+		/* We do have some empty MRs. Flush them out. */
+		rds_ib_stats_inc(s_ib_rdma_mr_pool_wait);
+		rds_ib_flush_mr_pool(pool, 0, &ibmr);
+		if (ibmr)
+			return ibmr;
+	}
+
+	ibmr = kzalloc_node(sizeof(*ibmr), GFP_KERNEL, rdsibdev_to_node(rds_ibdev));
+	if (!ibmr) {
+		err = -ENOMEM;
+		goto out_no_cigar;
+	}
+
+	memset(ibmr, 0, sizeof(*ibmr));
+
+	ibmr->fmr = ib_alloc_fmr(rds_ibdev->pd,
+			(IB_ACCESS_LOCAL_WRITE |
+			 IB_ACCESS_REMOTE_READ |
+			 IB_ACCESS_REMOTE_WRITE|
+			 IB_ACCESS_REMOTE_ATOMIC),
+			&pool->fmr_attr);
+	if (IS_ERR(ibmr->fmr)) {
+		err = PTR_ERR(ibmr->fmr);
+		ibmr->fmr = NULL;
+		printk(KERN_WARNING "RDS/IB: ib_alloc_fmr failed (err=%d)\n", err);
+		goto out_no_cigar;
+	}
+
+	rds_ib_stats_inc(s_ib_rdma_mr_alloc);
+	return ibmr;
+
+out_no_cigar:
+	if (ibmr) {
+		if (ibmr->fmr)
+			ib_dealloc_fmr(ibmr->fmr);
+		kfree(ibmr);
+	}
+	atomic_dec(&pool->item_count);
+	return ERR_PTR(err);
+}
+
+static int rds_ib_map_fmr(struct rds_ib_device *rds_ibdev, struct rds_ib_mr *ibmr,
+	       struct scatterlist *sg, unsigned int nents)
+{
+	struct ib_device *dev = rds_ibdev->dev;
+	struct scatterlist *scat = sg;
+	u64 io_addr = 0;
+	u64 *dma_pages;
+	u32 len;
+	int page_cnt, sg_dma_len;
+	int i, j;
+	int ret;
+
+	sg_dma_len = ib_dma_map_sg(dev, sg, nents,
+				 DMA_BIDIRECTIONAL);
+	if (unlikely(!sg_dma_len)) {
+		printk(KERN_WARNING "RDS/IB: dma_map_sg failed!\n");
+		return -EBUSY;
+	}
+
+	len = 0;
+	page_cnt = 0;
+
+	for (i = 0; i < sg_dma_len; ++i) {
+		unsigned int dma_len = ib_sg_dma_len(dev, &scat[i]);
+		u64 dma_addr = ib_sg_dma_address(dev, &scat[i]);
+
+		if (dma_addr & ~PAGE_MASK) {
+			if (i > 0)
+				return -EINVAL;
+			else
+				++page_cnt;
+		}
+		if ((dma_addr + dma_len) & ~PAGE_MASK) {
+			if (i < sg_dma_len - 1)
+				return -EINVAL;
+			else
+				++page_cnt;
+		}
+
+		len += dma_len;
+	}
+
+	page_cnt += len >> PAGE_SHIFT;
+	if (page_cnt > fmr_message_size)
+		return -EINVAL;
+
+	dma_pages = kmalloc_node(sizeof(u64) * page_cnt, GFP_ATOMIC,
+				 rdsibdev_to_node(rds_ibdev));
+	if (!dma_pages)
+		return -ENOMEM;
+
+	page_cnt = 0;
+	for (i = 0; i < sg_dma_len; ++i) {
+		unsigned int dma_len = ib_sg_dma_len(dev, &scat[i]);
+		u64 dma_addr = ib_sg_dma_address(dev, &scat[i]);
+
+		for (j = 0; j < dma_len; j += PAGE_SIZE)
+			dma_pages[page_cnt++] =
+				(dma_addr & PAGE_MASK) + j;
+	}
+
+	ret = ib_map_phys_fmr(ibmr->fmr,
+				   dma_pages, page_cnt, io_addr);
+	if (ret)
+		goto out;
+
+	/* Success - we successfully remapped the MR, so we can
+	 * safely tear down the old mapping. */
+	rds_ib_teardown_mr(ibmr);
+
+	ibmr->sg = scat;
+	ibmr->sg_len = nents;
+	ibmr->sg_dma_len = sg_dma_len;
+	ibmr->remap_count++;
+
+	rds_ib_stats_inc(s_ib_rdma_mr_used);
+	ret = 0;
+
+out:
+	kfree(dma_pages);
+
+	return ret;
+}
+
+void rds_ib_sync_mr(void *trans_private, int direction)
+{
+	struct rds_ib_mr *ibmr = trans_private;
+	struct rds_ib_device *rds_ibdev = ibmr->device;
+
+	switch (direction) {
+	case DMA_FROM_DEVICE:
+		ib_dma_sync_sg_for_cpu(rds_ibdev->dev, ibmr->sg,
+			ibmr->sg_dma_len, DMA_BIDIRECTIONAL);
+		break;
+	case DMA_TO_DEVICE:
+		ib_dma_sync_sg_for_device(rds_ibdev->dev, ibmr->sg,
+			ibmr->sg_dma_len, DMA_BIDIRECTIONAL);
+		break;
+	}
+}
+
+static void __rds_ib_teardown_mr(struct rds_ib_mr *ibmr)
+{
+	struct rds_ib_device *rds_ibdev = ibmr->device;
+
+	if (ibmr->sg_dma_len) {
+		ib_dma_unmap_sg(rds_ibdev->dev,
+				ibmr->sg, ibmr->sg_len,
+				DMA_BIDIRECTIONAL);
+		ibmr->sg_dma_len = 0;
+	}
+
+	/* Release the s/g list */
+	if (ibmr->sg_len) {
+		unsigned int i;
+
+		for (i = 0; i < ibmr->sg_len; ++i) {
+			struct page *page = sg_page(&ibmr->sg[i]);
+
+			/* FIXME we need a way to tell a r/w MR
+			 * from a r/o MR */
+			BUG_ON(irqs_disabled());
+			set_page_dirty(page);
+			put_page(page);
+		}
+		kfree(ibmr->sg);
+
+		ibmr->sg = NULL;
+		ibmr->sg_len = 0;
+	}
+}
+
+static void rds_ib_teardown_mr(struct rds_ib_mr *ibmr)
+{
+	unsigned int pinned = ibmr->sg_len;
+
+	__rds_ib_teardown_mr(ibmr);
+	if (pinned) {
+		struct rds_ib_device *rds_ibdev = ibmr->device;
+		struct rds_ib_mr_pool *pool = rds_ibdev->mr_pool;
+
+		atomic_sub(pinned, &pool->free_pinned);
+	}
+}
+
+static inline unsigned int rds_ib_flush_goal(struct rds_ib_mr_pool *pool, int free_all)
+{
+	unsigned int item_count;
+
+	item_count = atomic_read(&pool->item_count);
+	if (free_all)
+		return item_count;
+
+	return 0;
+}
+
+/*
+ * given an llist of mrs, put them all into the list_head for more processing
+ */
+static void llist_append_to_list(struct llist_head *llist, struct list_head *list)
+{
+	struct rds_ib_mr *ibmr;
+	struct llist_node *node;
+	struct llist_node *next;
+
+	node = llist_del_all(llist);
+	while (node) {
+		next = node->next;
+		ibmr = llist_entry(node, struct rds_ib_mr, llnode);
+		list_add_tail(&ibmr->unmap_list, list);
+		node = next;
+	}
+}
+
+/*
+ * this takes a list head of mrs and turns it into linked llist nodes
+ * of clusters.  Each cluster has linked llist nodes of
+ * MR_CLUSTER_SIZE mrs that are ready for reuse.
+ */
+static void list_to_llist_nodes(struct rds_ib_mr_pool *pool,
+				struct list_head *list,
+				struct llist_node **nodes_head,
+				struct llist_node **nodes_tail)
+{
+	struct rds_ib_mr *ibmr;
+	struct llist_node *cur = NULL;
+	struct llist_node **next = nodes_head;
+
+	list_for_each_entry(ibmr, list, unmap_list) {
+		cur = &ibmr->llnode;
+		*next = cur;
+		next = &cur->next;
+	}
+	*next = NULL;
+	*nodes_tail = cur;
+}
+
+/*
+ * Flush our pool of MRs.
+ * At a minimum, all currently unused MRs are unmapped.
+ * If the number of MRs allocated exceeds the limit, we also try
+ * to free as many MRs as needed to get back to this limit.
+ */
+static int rds_ib_flush_mr_pool(struct rds_ib_mr_pool *pool,
+			        int free_all, struct rds_ib_mr **ibmr_ret)
+{
+	struct rds_ib_mr *ibmr, *next;
+	struct llist_node *clean_nodes;
+	struct llist_node *clean_tail;
+	LIST_HEAD(unmap_list);
+	LIST_HEAD(fmr_list);
+	unsigned long unpinned = 0;
+	unsigned int nfreed = 0, ncleaned = 0, free_goal;
+	int ret = 0;
+
+	rds_ib_stats_inc(s_ib_rdma_mr_pool_flush);
+
+	if (ibmr_ret) {
+		DEFINE_WAIT(wait);
+		while(!mutex_trylock(&pool->flush_lock)) {
+			ibmr = rds_ib_reuse_fmr(pool);
+			if (ibmr) {
+				*ibmr_ret = ibmr;
+				finish_wait(&pool->flush_wait, &wait);
+				goto out_nolock;
+			}
+
+			prepare_to_wait(&pool->flush_wait, &wait,
+					TASK_UNINTERRUPTIBLE);
+			if (llist_empty(&pool->clean_list))
+				schedule();
+
+			ibmr = rds_ib_reuse_fmr(pool);
+			if (ibmr) {
+				*ibmr_ret = ibmr;
+				finish_wait(&pool->flush_wait, &wait);
+				goto out_nolock;
+			}
+		}
+		finish_wait(&pool->flush_wait, &wait);
+	} else
+		mutex_lock(&pool->flush_lock);
+
+	if (ibmr_ret) {
+		ibmr = rds_ib_reuse_fmr(pool);
+		if (ibmr) {
+			*ibmr_ret = ibmr;
+			goto out;
+		}
+	}
+
+	/* Get the list of all MRs to be dropped. Ordering matters -
+	 * we want to put drop_list ahead of free_list.
+	 */
+	llist_append_to_list(&pool->drop_list, &unmap_list);
+	llist_append_to_list(&pool->free_list, &unmap_list);
+	if (free_all)
+		llist_append_to_list(&pool->clean_list, &unmap_list);
+
+	free_goal = rds_ib_flush_goal(pool, free_all);
+
+	if (list_empty(&unmap_list))
+		goto out;
+
+	/* String all ib_mr's onto one list and hand them to ib_unmap_fmr */
+	list_for_each_entry(ibmr, &unmap_list, unmap_list)
+		list_add(&ibmr->fmr->list, &fmr_list);
+
+	ret = ib_unmap_fmr(&fmr_list);
+	if (ret)
+		printk(KERN_WARNING "RDS/IB: ib_unmap_fmr failed (err=%d)\n", ret);
+
+	/* Now we can destroy the DMA mapping and unpin any pages */
+	list_for_each_entry_safe(ibmr, next, &unmap_list, unmap_list) {
+		unpinned += ibmr->sg_len;
+		__rds_ib_teardown_mr(ibmr);
+		if (nfreed < free_goal || ibmr->remap_count >= pool->fmr_attr.max_maps) {
+			rds_ib_stats_inc(s_ib_rdma_mr_free);
+			list_del(&ibmr->unmap_list);
+			ib_dealloc_fmr(ibmr->fmr);
+			kfree(ibmr);
+			nfreed++;
+		}
+		ncleaned++;
+	}
+
+	if (!list_empty(&unmap_list)) {
+		/* we have to make sure that none of the things we're about
+		 * to put on the clean list would race with other cpus trying
+		 * to pull items off.  The llist would explode if we managed to
+		 * remove something from the clean list and then add it back again
+		 * while another CPU was spinning on that same item in llist_del_first.
+		 *
+		 * This is pretty unlikely, but just in case  wait for an llist grace period
+		 * here before adding anything back into the clean list.
+		 */
+		wait_clean_list_grace();
+
+		list_to_llist_nodes(pool, &unmap_list, &clean_nodes, &clean_tail);
+		if (ibmr_ret)
+			*ibmr_ret = llist_entry(clean_nodes, struct rds_ib_mr, llnode);
+
+		/* more than one entry in llist nodes */
+		if (clean_nodes->next)
+			llist_add_batch(clean_nodes->next, clean_tail, &pool->clean_list);
+
+	}
+
+	atomic_sub(unpinned, &pool->free_pinned);
+	atomic_sub(ncleaned, &pool->dirty_count);
+	atomic_sub(nfreed, &pool->item_count);
+
+out:
+	mutex_unlock(&pool->flush_lock);
+	if (waitqueue_active(&pool->flush_wait))
+		wake_up(&pool->flush_wait);
+out_nolock:
+	return ret;
+}
+
+static void rds_ib_mr_pool_flush_worker(struct work_struct *work)
+{
+	struct rds_ib_mr_pool *pool = container_of(work, struct rds_ib_mr_pool, flush_worker.work);
+
+	rds_ib_flush_mr_pool(pool, 0, NULL);
+}
+
+void rds_ib_free_mr(void *trans_private, int invalidate)
+{
+	struct rds_ib_mr *ibmr = trans_private;
+	struct rds_ib_device *rds_ibdev = ibmr->device;
+	struct rds_ib_mr_pool *pool = rds_ibdev->mr_pool;
+
+	rdsdebug("RDS/IB: free_mr nents %u\n", ibmr->sg_len);
+
+	/* Return it to the pool's free list */
+	if (ibmr->remap_count >= pool->fmr_attr.max_maps)
+		llist_add(&ibmr->llnode, &pool->drop_list);
+	else
+		llist_add(&ibmr->llnode, &pool->free_list);
+
+	atomic_add(ibmr->sg_len, &pool->free_pinned);
+	atomic_inc(&pool->dirty_count);
+
+	/* If we've pinned too many pages, request a flush */
+	if (atomic_read(&pool->free_pinned) >= pool->max_free_pinned ||
+	    atomic_read(&pool->dirty_count) >= pool->max_items / 10)
+		schedule_delayed_work(&pool->flush_worker, 10);
+
+	if (invalidate) {
+		if (likely(!in_interrupt())) {
+			rds_ib_flush_mr_pool(pool, 0, NULL);
+		} else {
+			/* We get here if the user created a MR marked
+			 * as use_once and invalidate at the same time. */
+			schedule_delayed_work(&pool->flush_worker, 10);
+		}
+	}
+
+	rds_ib_dev_put(rds_ibdev);
+}
+
+void rds_ib_flush_mrs(void)
+{
+	struct rds_ib_device *rds_ibdev;
+
+	down_read(&rds_ib_devices_lock);
+	list_for_each_entry(rds_ibdev, &rds_ib_devices, list) {
+		struct rds_ib_mr_pool *pool = rds_ibdev->mr_pool;
+
+		if (pool)
+			rds_ib_flush_mr_pool(pool, 0, NULL);
+	}
+	up_read(&rds_ib_devices_lock);
+}
+
+void *rds_ib_get_mr(struct scatterlist *sg, unsigned long nents,
+		    struct rds_sock *rs, u32 *key_ret)
+{
+	struct rds_ib_device *rds_ibdev;
+	struct rds_ib_mr *ibmr = NULL;
+	int ret;
+
+	rds_ibdev = rds_ib_get_device(rs->rs_bound_addr);
+	if (!rds_ibdev) {
+		ret = -ENODEV;
+		goto out;
+	}
+
+	if (!rds_ibdev->mr_pool) {
+		ret = -ENODEV;
+		goto out;
+	}
+
+	ibmr = rds_ib_alloc_fmr(rds_ibdev);
+	if (IS_ERR(ibmr))
+		return ibmr;
+
+	ret = rds_ib_map_fmr(rds_ibdev, ibmr, sg, nents);
+	if (ret == 0)
+		*key_ret = ibmr->fmr->rkey;
+	else
+		printk(KERN_WARNING "RDS/IB: map_fmr failed (errno=%d)\n", ret);
+
+	ibmr->device = rds_ibdev;
+	rds_ibdev = NULL;
+
+ out:
+	if (ret) {
+		if (ibmr)
+			rds_ib_free_mr(ibmr, 0);
+		ibmr = ERR_PTR(ret);
+	}
+	if (rds_ibdev)
+		rds_ib_dev_put(rds_ibdev);
+	return ibmr;
+}
+
diff --git a/net/rds/ib_recv.c b/net/rds/ib_recv.c
new file mode 100644
index 000000000..1b981a4e4
--- /dev/null
+++ b/net/rds/ib_recv.c
@@ -0,0 +1,1064 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/pci.h>
+#include <linux/dma-mapping.h>
+#include <rdma/rdma_cm.h>
+
+#include "rds.h"
+#include "ib.h"
+
+static struct kmem_cache *rds_ib_incoming_slab;
+static struct kmem_cache *rds_ib_frag_slab;
+static atomic_t	rds_ib_allocation = ATOMIC_INIT(0);
+
+void rds_ib_recv_init_ring(struct rds_ib_connection *ic)
+{
+	struct rds_ib_recv_work *recv;
+	u32 i;
+
+	for (i = 0, recv = ic->i_recvs; i < ic->i_recv_ring.w_nr; i++, recv++) {
+		struct ib_sge *sge;
+
+		recv->r_ibinc = NULL;
+		recv->r_frag = NULL;
+
+		recv->r_wr.next = NULL;
+		recv->r_wr.wr_id = i;
+		recv->r_wr.sg_list = recv->r_sge;
+		recv->r_wr.num_sge = RDS_IB_RECV_SGE;
+
+		sge = &recv->r_sge[0];
+		sge->addr = ic->i_recv_hdrs_dma + (i * sizeof(struct rds_header));
+		sge->length = sizeof(struct rds_header);
+		sge->lkey = ic->i_mr->lkey;
+
+		sge = &recv->r_sge[1];
+		sge->addr = 0;
+		sge->length = RDS_FRAG_SIZE;
+		sge->lkey = ic->i_mr->lkey;
+	}
+}
+
+/*
+ * The entire 'from' list, including the from element itself, is put on
+ * to the tail of the 'to' list.
+ */
+static void list_splice_entire_tail(struct list_head *from,
+				    struct list_head *to)
+{
+	struct list_head *from_last = from->prev;
+
+	list_splice_tail(from_last, to);
+	list_add_tail(from_last, to);
+}
+
+static void rds_ib_cache_xfer_to_ready(struct rds_ib_refill_cache *cache)
+{
+	struct list_head *tmp;
+
+	tmp = xchg(&cache->xfer, NULL);
+	if (tmp) {
+		if (cache->ready)
+			list_splice_entire_tail(tmp, cache->ready);
+		else
+			cache->ready = tmp;
+	}
+}
+
+static int rds_ib_recv_alloc_cache(struct rds_ib_refill_cache *cache)
+{
+	struct rds_ib_cache_head *head;
+	int cpu;
+
+	cache->percpu = alloc_percpu(struct rds_ib_cache_head);
+	if (!cache->percpu)
+	       return -ENOMEM;
+
+	for_each_possible_cpu(cpu) {
+		head = per_cpu_ptr(cache->percpu, cpu);
+		head->first = NULL;
+		head->count = 0;
+	}
+	cache->xfer = NULL;
+	cache->ready = NULL;
+
+	return 0;
+}
+
+int rds_ib_recv_alloc_caches(struct rds_ib_connection *ic)
+{
+	int ret;
+
+	ret = rds_ib_recv_alloc_cache(&ic->i_cache_incs);
+	if (!ret) {
+		ret = rds_ib_recv_alloc_cache(&ic->i_cache_frags);
+		if (ret)
+			free_percpu(ic->i_cache_incs.percpu);
+	}
+
+	return ret;
+}
+
+static void rds_ib_cache_splice_all_lists(struct rds_ib_refill_cache *cache,
+					  struct list_head *caller_list)
+{
+	struct rds_ib_cache_head *head;
+	int cpu;
+
+	for_each_possible_cpu(cpu) {
+		head = per_cpu_ptr(cache->percpu, cpu);
+		if (head->first) {
+			list_splice_entire_tail(head->first, caller_list);
+			head->first = NULL;
+		}
+	}
+
+	if (cache->ready) {
+		list_splice_entire_tail(cache->ready, caller_list);
+		cache->ready = NULL;
+	}
+}
+
+void rds_ib_recv_free_caches(struct rds_ib_connection *ic)
+{
+	struct rds_ib_incoming *inc;
+	struct rds_ib_incoming *inc_tmp;
+	struct rds_page_frag *frag;
+	struct rds_page_frag *frag_tmp;
+	LIST_HEAD(list);
+
+	rds_ib_cache_xfer_to_ready(&ic->i_cache_incs);
+	rds_ib_cache_splice_all_lists(&ic->i_cache_incs, &list);
+	free_percpu(ic->i_cache_incs.percpu);
+
+	list_for_each_entry_safe(inc, inc_tmp, &list, ii_cache_entry) {
+		list_del(&inc->ii_cache_entry);
+		WARN_ON(!list_empty(&inc->ii_frags));
+		kmem_cache_free(rds_ib_incoming_slab, inc);
+	}
+
+	rds_ib_cache_xfer_to_ready(&ic->i_cache_frags);
+	rds_ib_cache_splice_all_lists(&ic->i_cache_frags, &list);
+	free_percpu(ic->i_cache_frags.percpu);
+
+	list_for_each_entry_safe(frag, frag_tmp, &list, f_cache_entry) {
+		list_del(&frag->f_cache_entry);
+		WARN_ON(!list_empty(&frag->f_item));
+		kmem_cache_free(rds_ib_frag_slab, frag);
+	}
+}
+
+/* fwd decl */
+static void rds_ib_recv_cache_put(struct list_head *new_item,
+				  struct rds_ib_refill_cache *cache);
+static struct list_head *rds_ib_recv_cache_get(struct rds_ib_refill_cache *cache);
+
+
+/* Recycle frag and attached recv buffer f_sg */
+static void rds_ib_frag_free(struct rds_ib_connection *ic,
+			     struct rds_page_frag *frag)
+{
+	rdsdebug("frag %p page %p\n", frag, sg_page(&frag->f_sg));
+
+	rds_ib_recv_cache_put(&frag->f_cache_entry, &ic->i_cache_frags);
+}
+
+/* Recycle inc after freeing attached frags */
+void rds_ib_inc_free(struct rds_incoming *inc)
+{
+	struct rds_ib_incoming *ibinc;
+	struct rds_page_frag *frag;
+	struct rds_page_frag *pos;
+	struct rds_ib_connection *ic = inc->i_conn->c_transport_data;
+
+	ibinc = container_of(inc, struct rds_ib_incoming, ii_inc);
+
+	/* Free attached frags */
+	list_for_each_entry_safe(frag, pos, &ibinc->ii_frags, f_item) {
+		list_del_init(&frag->f_item);
+		rds_ib_frag_free(ic, frag);
+	}
+	BUG_ON(!list_empty(&ibinc->ii_frags));
+
+	rdsdebug("freeing ibinc %p inc %p\n", ibinc, inc);
+	rds_ib_recv_cache_put(&ibinc->ii_cache_entry, &ic->i_cache_incs);
+}
+
+static void rds_ib_recv_clear_one(struct rds_ib_connection *ic,
+				  struct rds_ib_recv_work *recv)
+{
+	if (recv->r_ibinc) {
+		rds_inc_put(&recv->r_ibinc->ii_inc);
+		recv->r_ibinc = NULL;
+	}
+	if (recv->r_frag) {
+		ib_dma_unmap_sg(ic->i_cm_id->device, &recv->r_frag->f_sg, 1, DMA_FROM_DEVICE);
+		rds_ib_frag_free(ic, recv->r_frag);
+		recv->r_frag = NULL;
+	}
+}
+
+void rds_ib_recv_clear_ring(struct rds_ib_connection *ic)
+{
+	u32 i;
+
+	for (i = 0; i < ic->i_recv_ring.w_nr; i++)
+		rds_ib_recv_clear_one(ic, &ic->i_recvs[i]);
+}
+
+static struct rds_ib_incoming *rds_ib_refill_one_inc(struct rds_ib_connection *ic,
+						     gfp_t slab_mask)
+{
+	struct rds_ib_incoming *ibinc;
+	struct list_head *cache_item;
+	int avail_allocs;
+
+	cache_item = rds_ib_recv_cache_get(&ic->i_cache_incs);
+	if (cache_item) {
+		ibinc = container_of(cache_item, struct rds_ib_incoming, ii_cache_entry);
+	} else {
+		avail_allocs = atomic_add_unless(&rds_ib_allocation,
+						 1, rds_ib_sysctl_max_recv_allocation);
+		if (!avail_allocs) {
+			rds_ib_stats_inc(s_ib_rx_alloc_limit);
+			return NULL;
+		}
+		ibinc = kmem_cache_alloc(rds_ib_incoming_slab, slab_mask);
+		if (!ibinc) {
+			atomic_dec(&rds_ib_allocation);
+			return NULL;
+		}
+	}
+	INIT_LIST_HEAD(&ibinc->ii_frags);
+	rds_inc_init(&ibinc->ii_inc, ic->conn, ic->conn->c_faddr);
+
+	return ibinc;
+}
+
+static struct rds_page_frag *rds_ib_refill_one_frag(struct rds_ib_connection *ic,
+						    gfp_t slab_mask, gfp_t page_mask)
+{
+	struct rds_page_frag *frag;
+	struct list_head *cache_item;
+	int ret;
+
+	cache_item = rds_ib_recv_cache_get(&ic->i_cache_frags);
+	if (cache_item) {
+		frag = container_of(cache_item, struct rds_page_frag, f_cache_entry);
+	} else {
+		frag = kmem_cache_alloc(rds_ib_frag_slab, slab_mask);
+		if (!frag)
+			return NULL;
+
+		sg_init_table(&frag->f_sg, 1);
+		ret = rds_page_remainder_alloc(&frag->f_sg,
+					       RDS_FRAG_SIZE, page_mask);
+		if (ret) {
+			kmem_cache_free(rds_ib_frag_slab, frag);
+			return NULL;
+		}
+	}
+
+	INIT_LIST_HEAD(&frag->f_item);
+
+	return frag;
+}
+
+static int rds_ib_recv_refill_one(struct rds_connection *conn,
+				  struct rds_ib_recv_work *recv, int prefill)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+	struct ib_sge *sge;
+	int ret = -ENOMEM;
+	gfp_t slab_mask = GFP_NOWAIT;
+	gfp_t page_mask = GFP_NOWAIT;
+
+	if (prefill) {
+		slab_mask = GFP_KERNEL;
+		page_mask = GFP_HIGHUSER;
+	}
+
+	if (!ic->i_cache_incs.ready)
+		rds_ib_cache_xfer_to_ready(&ic->i_cache_incs);
+	if (!ic->i_cache_frags.ready)
+		rds_ib_cache_xfer_to_ready(&ic->i_cache_frags);
+
+	/*
+	 * ibinc was taken from recv if recv contained the start of a message.
+	 * recvs that were continuations will still have this allocated.
+	 */
+	if (!recv->r_ibinc) {
+		recv->r_ibinc = rds_ib_refill_one_inc(ic, slab_mask);
+		if (!recv->r_ibinc)
+			goto out;
+	}
+
+	WARN_ON(recv->r_frag); /* leak! */
+	recv->r_frag = rds_ib_refill_one_frag(ic, slab_mask, page_mask);
+	if (!recv->r_frag)
+		goto out;
+
+	ret = ib_dma_map_sg(ic->i_cm_id->device, &recv->r_frag->f_sg,
+			    1, DMA_FROM_DEVICE);
+	WARN_ON(ret != 1);
+
+	sge = &recv->r_sge[0];
+	sge->addr = ic->i_recv_hdrs_dma + (recv - ic->i_recvs) * sizeof(struct rds_header);
+	sge->length = sizeof(struct rds_header);
+
+	sge = &recv->r_sge[1];
+	sge->addr = ib_sg_dma_address(ic->i_cm_id->device, &recv->r_frag->f_sg);
+	sge->length = ib_sg_dma_len(ic->i_cm_id->device, &recv->r_frag->f_sg);
+
+	ret = 0;
+out:
+	return ret;
+}
+
+/*
+ * This tries to allocate and post unused work requests after making sure that
+ * they have all the allocations they need to queue received fragments into
+ * sockets.
+ *
+ * -1 is returned if posting fails due to temporary resource exhaustion.
+ */
+void rds_ib_recv_refill(struct rds_connection *conn, int prefill)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+	struct rds_ib_recv_work *recv;
+	struct ib_recv_wr *failed_wr;
+	unsigned int posted = 0;
+	int ret = 0;
+	u32 pos;
+
+	while ((prefill || rds_conn_up(conn)) &&
+	       rds_ib_ring_alloc(&ic->i_recv_ring, 1, &pos)) {
+		if (pos >= ic->i_recv_ring.w_nr) {
+			printk(KERN_NOTICE "Argh - ring alloc returned pos=%u\n",
+					pos);
+			break;
+		}
+
+		recv = &ic->i_recvs[pos];
+		ret = rds_ib_recv_refill_one(conn, recv, prefill);
+		if (ret) {
+			break;
+		}
+
+		/* XXX when can this fail? */
+		ret = ib_post_recv(ic->i_cm_id->qp, &recv->r_wr, &failed_wr);
+		rdsdebug("recv %p ibinc %p page %p addr %lu ret %d\n", recv,
+			 recv->r_ibinc, sg_page(&recv->r_frag->f_sg),
+			 (long) ib_sg_dma_address(
+				ic->i_cm_id->device,
+				&recv->r_frag->f_sg),
+			ret);
+		if (ret) {
+			rds_ib_conn_error(conn, "recv post on "
+			       "%pI4 returned %d, disconnecting and "
+			       "reconnecting\n", &conn->c_faddr,
+			       ret);
+			break;
+		}
+
+		posted++;
+	}
+
+	/* We're doing flow control - update the window. */
+	if (ic->i_flowctl && posted)
+		rds_ib_advertise_credits(conn, posted);
+
+	if (ret)
+		rds_ib_ring_unalloc(&ic->i_recv_ring, 1);
+}
+
+/*
+ * We want to recycle several types of recv allocations, like incs and frags.
+ * To use this, the *_free() function passes in the ptr to a list_head within
+ * the recyclee, as well as the cache to put it on.
+ *
+ * First, we put the memory on a percpu list. When this reaches a certain size,
+ * We move it to an intermediate non-percpu list in a lockless manner, with some
+ * xchg/compxchg wizardry.
+ *
+ * N.B. Instead of a list_head as the anchor, we use a single pointer, which can
+ * be NULL and xchg'd. The list is actually empty when the pointer is NULL, and
+ * list_empty() will return true with one element is actually present.
+ */
+static void rds_ib_recv_cache_put(struct list_head *new_item,
+				 struct rds_ib_refill_cache *cache)
+{
+	unsigned long flags;
+	struct list_head *old, *chpfirst;
+
+	local_irq_save(flags);
+
+	chpfirst = __this_cpu_read(cache->percpu->first);
+	if (!chpfirst)
+		INIT_LIST_HEAD(new_item);
+	else /* put on front */
+		list_add_tail(new_item, chpfirst);
+
+	__this_cpu_write(cache->percpu->first, new_item);
+	__this_cpu_inc(cache->percpu->count);
+
+	if (__this_cpu_read(cache->percpu->count) < RDS_IB_RECYCLE_BATCH_COUNT)
+		goto end;
+
+	/*
+	 * Return our per-cpu first list to the cache's xfer by atomically
+	 * grabbing the current xfer list, appending it to our per-cpu list,
+	 * and then atomically returning that entire list back to the
+	 * cache's xfer list as long as it's still empty.
+	 */
+	do {
+		old = xchg(&cache->xfer, NULL);
+		if (old)
+			list_splice_entire_tail(old, chpfirst);
+		old = cmpxchg(&cache->xfer, NULL, chpfirst);
+	} while (old);
+
+
+	__this_cpu_write(cache->percpu->first, NULL);
+	__this_cpu_write(cache->percpu->count, 0);
+end:
+	local_irq_restore(flags);
+}
+
+static struct list_head *rds_ib_recv_cache_get(struct rds_ib_refill_cache *cache)
+{
+	struct list_head *head = cache->ready;
+
+	if (head) {
+		if (!list_empty(head)) {
+			cache->ready = head->next;
+			list_del_init(head);
+		} else
+			cache->ready = NULL;
+	}
+
+	return head;
+}
+
+int rds_ib_inc_copy_to_user(struct rds_incoming *inc, struct iov_iter *to)
+{
+	struct rds_ib_incoming *ibinc;
+	struct rds_page_frag *frag;
+	unsigned long to_copy;
+	unsigned long frag_off = 0;
+	int copied = 0;
+	int ret;
+	u32 len;
+
+	ibinc = container_of(inc, struct rds_ib_incoming, ii_inc);
+	frag = list_entry(ibinc->ii_frags.next, struct rds_page_frag, f_item);
+	len = be32_to_cpu(inc->i_hdr.h_len);
+
+	while (iov_iter_count(to) && copied < len) {
+		if (frag_off == RDS_FRAG_SIZE) {
+			frag = list_entry(frag->f_item.next,
+					  struct rds_page_frag, f_item);
+			frag_off = 0;
+		}
+		to_copy = min_t(unsigned long, iov_iter_count(to),
+				RDS_FRAG_SIZE - frag_off);
+		to_copy = min_t(unsigned long, to_copy, len - copied);
+
+		/* XXX needs + offset for multiple recvs per page */
+		rds_stats_add(s_copy_to_user, to_copy);
+		ret = copy_page_to_iter(sg_page(&frag->f_sg),
+					frag->f_sg.offset + frag_off,
+					to_copy,
+					to);
+		if (ret != to_copy)
+			return -EFAULT;
+
+		frag_off += to_copy;
+		copied += to_copy;
+	}
+
+	return copied;
+}
+
+/* ic starts out kzalloc()ed */
+void rds_ib_recv_init_ack(struct rds_ib_connection *ic)
+{
+	struct ib_send_wr *wr = &ic->i_ack_wr;
+	struct ib_sge *sge = &ic->i_ack_sge;
+
+	sge->addr = ic->i_ack_dma;
+	sge->length = sizeof(struct rds_header);
+	sge->lkey = ic->i_mr->lkey;
+
+	wr->sg_list = sge;
+	wr->num_sge = 1;
+	wr->opcode = IB_WR_SEND;
+	wr->wr_id = RDS_IB_ACK_WR_ID;
+	wr->send_flags = IB_SEND_SIGNALED | IB_SEND_SOLICITED;
+}
+
+/*
+ * You'd think that with reliable IB connections you wouldn't need to ack
+ * messages that have been received.  The problem is that IB hardware generates
+ * an ack message before it has DMAed the message into memory.  This creates a
+ * potential message loss if the HCA is disabled for any reason between when it
+ * sends the ack and before the message is DMAed and processed.  This is only a
+ * potential issue if another HCA is available for fail-over.
+ *
+ * When the remote host receives our ack they'll free the sent message from
+ * their send queue.  To decrease the latency of this we always send an ack
+ * immediately after we've received messages.
+ *
+ * For simplicity, we only have one ack in flight at a time.  This puts
+ * pressure on senders to have deep enough send queues to absorb the latency of
+ * a single ack frame being in flight.  This might not be good enough.
+ *
+ * This is implemented by have a long-lived send_wr and sge which point to a
+ * statically allocated ack frame.  This ack wr does not fall under the ring
+ * accounting that the tx and rx wrs do.  The QP attribute specifically makes
+ * room for it beyond the ring size.  Send completion notices its special
+ * wr_id and avoids working with the ring in that case.
+ */
+#ifndef KERNEL_HAS_ATOMIC64
+static void rds_ib_set_ack(struct rds_ib_connection *ic, u64 seq,
+				int ack_required)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&ic->i_ack_lock, flags);
+	ic->i_ack_next = seq;
+	if (ack_required)
+		set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
+	spin_unlock_irqrestore(&ic->i_ack_lock, flags);
+}
+
+static u64 rds_ib_get_ack(struct rds_ib_connection *ic)
+{
+	unsigned long flags;
+	u64 seq;
+
+	clear_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
+
+	spin_lock_irqsave(&ic->i_ack_lock, flags);
+	seq = ic->i_ack_next;
+	spin_unlock_irqrestore(&ic->i_ack_lock, flags);
+
+	return seq;
+}
+#else
+static void rds_ib_set_ack(struct rds_ib_connection *ic, u64 seq,
+				int ack_required)
+{
+	atomic64_set(&ic->i_ack_next, seq);
+	if (ack_required) {
+		smp_mb__before_atomic();
+		set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
+	}
+}
+
+static u64 rds_ib_get_ack(struct rds_ib_connection *ic)
+{
+	clear_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
+	smp_mb__after_atomic();
+
+	return atomic64_read(&ic->i_ack_next);
+}
+#endif
+
+
+static void rds_ib_send_ack(struct rds_ib_connection *ic, unsigned int adv_credits)
+{
+	struct rds_header *hdr = ic->i_ack;
+	struct ib_send_wr *failed_wr;
+	u64 seq;
+	int ret;
+
+	seq = rds_ib_get_ack(ic);
+
+	rdsdebug("send_ack: ic %p ack %llu\n", ic, (unsigned long long) seq);
+	rds_message_populate_header(hdr, 0, 0, 0);
+	hdr->h_ack = cpu_to_be64(seq);
+	hdr->h_credit = adv_credits;
+	rds_message_make_checksum(hdr);
+	ic->i_ack_queued = jiffies;
+
+	ret = ib_post_send(ic->i_cm_id->qp, &ic->i_ack_wr, &failed_wr);
+	if (unlikely(ret)) {
+		/* Failed to send. Release the WR, and
+		 * force another ACK.
+		 */
+		clear_bit(IB_ACK_IN_FLIGHT, &ic->i_ack_flags);
+		set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
+
+		rds_ib_stats_inc(s_ib_ack_send_failure);
+
+		rds_ib_conn_error(ic->conn, "sending ack failed\n");
+	} else
+		rds_ib_stats_inc(s_ib_ack_sent);
+}
+
+/*
+ * There are 3 ways of getting acknowledgements to the peer:
+ *  1.	We call rds_ib_attempt_ack from the recv completion handler
+ *	to send an ACK-only frame.
+ *	However, there can be only one such frame in the send queue
+ *	at any time, so we may have to postpone it.
+ *  2.	When another (data) packet is transmitted while there's
+ *	an ACK in the queue, we piggyback the ACK sequence number
+ *	on the data packet.
+ *  3.	If the ACK WR is done sending, we get called from the
+ *	send queue completion handler, and check whether there's
+ *	another ACK pending (postponed because the WR was on the
+ *	queue). If so, we transmit it.
+ *
+ * We maintain 2 variables:
+ *  -	i_ack_flags, which keeps track of whether the ACK WR
+ *	is currently in the send queue or not (IB_ACK_IN_FLIGHT)
+ *  -	i_ack_next, which is the last sequence number we received
+ *
+ * Potentially, send queue and receive queue handlers can run concurrently.
+ * It would be nice to not have to use a spinlock to synchronize things,
+ * but the one problem that rules this out is that 64bit updates are
+ * not atomic on all platforms. Things would be a lot simpler if
+ * we had atomic64 or maybe cmpxchg64 everywhere.
+ *
+ * Reconnecting complicates this picture just slightly. When we
+ * reconnect, we may be seeing duplicate packets. The peer
+ * is retransmitting them, because it hasn't seen an ACK for
+ * them. It is important that we ACK these.
+ *
+ * ACK mitigation adds a header flag "ACK_REQUIRED"; any packet with
+ * this flag set *MUST* be acknowledged immediately.
+ */
+
+/*
+ * When we get here, we're called from the recv queue handler.
+ * Check whether we ought to transmit an ACK.
+ */
+void rds_ib_attempt_ack(struct rds_ib_connection *ic)
+{
+	unsigned int adv_credits;
+
+	if (!test_bit(IB_ACK_REQUESTED, &ic->i_ack_flags))
+		return;
+
+	if (test_and_set_bit(IB_ACK_IN_FLIGHT, &ic->i_ack_flags)) {
+		rds_ib_stats_inc(s_ib_ack_send_delayed);
+		return;
+	}
+
+	/* Can we get a send credit? */
+	if (!rds_ib_send_grab_credits(ic, 1, &adv_credits, 0, RDS_MAX_ADV_CREDIT)) {
+		rds_ib_stats_inc(s_ib_tx_throttle);
+		clear_bit(IB_ACK_IN_FLIGHT, &ic->i_ack_flags);
+		return;
+	}
+
+	clear_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
+	rds_ib_send_ack(ic, adv_credits);
+}
+
+/*
+ * We get here from the send completion handler, when the
+ * adapter tells us the ACK frame was sent.
+ */
+void rds_ib_ack_send_complete(struct rds_ib_connection *ic)
+{
+	clear_bit(IB_ACK_IN_FLIGHT, &ic->i_ack_flags);
+	rds_ib_attempt_ack(ic);
+}
+
+/*
+ * This is called by the regular xmit code when it wants to piggyback
+ * an ACK on an outgoing frame.
+ */
+u64 rds_ib_piggyb_ack(struct rds_ib_connection *ic)
+{
+	if (test_and_clear_bit(IB_ACK_REQUESTED, &ic->i_ack_flags))
+		rds_ib_stats_inc(s_ib_ack_send_piggybacked);
+	return rds_ib_get_ack(ic);
+}
+
+/*
+ * It's kind of lame that we're copying from the posted receive pages into
+ * long-lived bitmaps.  We could have posted the bitmaps and rdma written into
+ * them.  But receiving new congestion bitmaps should be a *rare* event, so
+ * hopefully we won't need to invest that complexity in making it more
+ * efficient.  By copying we can share a simpler core with TCP which has to
+ * copy.
+ */
+static void rds_ib_cong_recv(struct rds_connection *conn,
+			      struct rds_ib_incoming *ibinc)
+{
+	struct rds_cong_map *map;
+	unsigned int map_off;
+	unsigned int map_page;
+	struct rds_page_frag *frag;
+	unsigned long frag_off;
+	unsigned long to_copy;
+	unsigned long copied;
+	uint64_t uncongested = 0;
+	void *addr;
+
+	/* catch completely corrupt packets */
+	if (be32_to_cpu(ibinc->ii_inc.i_hdr.h_len) != RDS_CONG_MAP_BYTES)
+		return;
+
+	map = conn->c_fcong;
+	map_page = 0;
+	map_off = 0;
+
+	frag = list_entry(ibinc->ii_frags.next, struct rds_page_frag, f_item);
+	frag_off = 0;
+
+	copied = 0;
+
+	while (copied < RDS_CONG_MAP_BYTES) {
+		uint64_t *src, *dst;
+		unsigned int k;
+
+		to_copy = min(RDS_FRAG_SIZE - frag_off, PAGE_SIZE - map_off);
+		BUG_ON(to_copy & 7); /* Must be 64bit aligned. */
+
+		addr = kmap_atomic(sg_page(&frag->f_sg));
+
+		src = addr + frag_off;
+		dst = (void *)map->m_page_addrs[map_page] + map_off;
+		for (k = 0; k < to_copy; k += 8) {
+			/* Record ports that became uncongested, ie
+			 * bits that changed from 0 to 1. */
+			uncongested |= ~(*src) & *dst;
+			*dst++ = *src++;
+		}
+		kunmap_atomic(addr);
+
+		copied += to_copy;
+
+		map_off += to_copy;
+		if (map_off == PAGE_SIZE) {
+			map_off = 0;
+			map_page++;
+		}
+
+		frag_off += to_copy;
+		if (frag_off == RDS_FRAG_SIZE) {
+			frag = list_entry(frag->f_item.next,
+					  struct rds_page_frag, f_item);
+			frag_off = 0;
+		}
+	}
+
+	/* the congestion map is in little endian order */
+	uncongested = le64_to_cpu(uncongested);
+
+	rds_cong_map_updated(map, uncongested);
+}
+
+/*
+ * Rings are posted with all the allocations they'll need to queue the
+ * incoming message to the receiving socket so this can't fail.
+ * All fragments start with a header, so we can make sure we're not receiving
+ * garbage, and we can tell a small 8 byte fragment from an ACK frame.
+ */
+struct rds_ib_ack_state {
+	u64		ack_next;
+	u64		ack_recv;
+	unsigned int	ack_required:1;
+	unsigned int	ack_next_valid:1;
+	unsigned int	ack_recv_valid:1;
+};
+
+static void rds_ib_process_recv(struct rds_connection *conn,
+				struct rds_ib_recv_work *recv, u32 data_len,
+				struct rds_ib_ack_state *state)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+	struct rds_ib_incoming *ibinc = ic->i_ibinc;
+	struct rds_header *ihdr, *hdr;
+
+	/* XXX shut down the connection if port 0,0 are seen? */
+
+	rdsdebug("ic %p ibinc %p recv %p byte len %u\n", ic, ibinc, recv,
+		 data_len);
+
+	if (data_len < sizeof(struct rds_header)) {
+		rds_ib_conn_error(conn, "incoming message "
+		       "from %pI4 didn't include a "
+		       "header, disconnecting and "
+		       "reconnecting\n",
+		       &conn->c_faddr);
+		return;
+	}
+	data_len -= sizeof(struct rds_header);
+
+	ihdr = &ic->i_recv_hdrs[recv - ic->i_recvs];
+
+	/* Validate the checksum. */
+	if (!rds_message_verify_checksum(ihdr)) {
+		rds_ib_conn_error(conn, "incoming message "
+		       "from %pI4 has corrupted header - "
+		       "forcing a reconnect\n",
+		       &conn->c_faddr);
+		rds_stats_inc(s_recv_drop_bad_checksum);
+		return;
+	}
+
+	/* Process the ACK sequence which comes with every packet */
+	state->ack_recv = be64_to_cpu(ihdr->h_ack);
+	state->ack_recv_valid = 1;
+
+	/* Process the credits update if there was one */
+	if (ihdr->h_credit)
+		rds_ib_send_add_credits(conn, ihdr->h_credit);
+
+	if (ihdr->h_sport == 0 && ihdr->h_dport == 0 && data_len == 0) {
+		/* This is an ACK-only packet. The fact that it gets
+		 * special treatment here is that historically, ACKs
+		 * were rather special beasts.
+		 */
+		rds_ib_stats_inc(s_ib_ack_received);
+
+		/*
+		 * Usually the frags make their way on to incs and are then freed as
+		 * the inc is freed.  We don't go that route, so we have to drop the
+		 * page ref ourselves.  We can't just leave the page on the recv
+		 * because that confuses the dma mapping of pages and each recv's use
+		 * of a partial page.
+		 *
+		 * FIXME: Fold this into the code path below.
+		 */
+		rds_ib_frag_free(ic, recv->r_frag);
+		recv->r_frag = NULL;
+		return;
+	}
+
+	/*
+	 * If we don't already have an inc on the connection then this
+	 * fragment has a header and starts a message.. copy its header
+	 * into the inc and save the inc so we can hang upcoming fragments
+	 * off its list.
+	 */
+	if (!ibinc) {
+		ibinc = recv->r_ibinc;
+		recv->r_ibinc = NULL;
+		ic->i_ibinc = ibinc;
+
+		hdr = &ibinc->ii_inc.i_hdr;
+		memcpy(hdr, ihdr, sizeof(*hdr));
+		ic->i_recv_data_rem = be32_to_cpu(hdr->h_len);
+
+		rdsdebug("ic %p ibinc %p rem %u flag 0x%x\n", ic, ibinc,
+			 ic->i_recv_data_rem, hdr->h_flags);
+	} else {
+		hdr = &ibinc->ii_inc.i_hdr;
+		/* We can't just use memcmp here; fragments of a
+		 * single message may carry different ACKs */
+		if (hdr->h_sequence != ihdr->h_sequence ||
+		    hdr->h_len != ihdr->h_len ||
+		    hdr->h_sport != ihdr->h_sport ||
+		    hdr->h_dport != ihdr->h_dport) {
+			rds_ib_conn_error(conn,
+				"fragment header mismatch; forcing reconnect\n");
+			return;
+		}
+	}
+
+	list_add_tail(&recv->r_frag->f_item, &ibinc->ii_frags);
+	recv->r_frag = NULL;
+
+	if (ic->i_recv_data_rem > RDS_FRAG_SIZE)
+		ic->i_recv_data_rem -= RDS_FRAG_SIZE;
+	else {
+		ic->i_recv_data_rem = 0;
+		ic->i_ibinc = NULL;
+
+		if (ibinc->ii_inc.i_hdr.h_flags == RDS_FLAG_CONG_BITMAP)
+			rds_ib_cong_recv(conn, ibinc);
+		else {
+			rds_recv_incoming(conn, conn->c_faddr, conn->c_laddr,
+					  &ibinc->ii_inc, GFP_ATOMIC);
+			state->ack_next = be64_to_cpu(hdr->h_sequence);
+			state->ack_next_valid = 1;
+		}
+
+		/* Evaluate the ACK_REQUIRED flag *after* we received
+		 * the complete frame, and after bumping the next_rx
+		 * sequence. */
+		if (hdr->h_flags & RDS_FLAG_ACK_REQUIRED) {
+			rds_stats_inc(s_recv_ack_required);
+			state->ack_required = 1;
+		}
+
+		rds_inc_put(&ibinc->ii_inc);
+	}
+}
+
+/*
+ * Plucking the oldest entry from the ring can be done concurrently with
+ * the thread refilling the ring.  Each ring operation is protected by
+ * spinlocks and the transient state of refilling doesn't change the
+ * recording of which entry is oldest.
+ *
+ * This relies on IB only calling one cq comp_handler for each cq so that
+ * there will only be one caller of rds_recv_incoming() per RDS connection.
+ */
+void rds_ib_recv_cq_comp_handler(struct ib_cq *cq, void *context)
+{
+	struct rds_connection *conn = context;
+	struct rds_ib_connection *ic = conn->c_transport_data;
+
+	rdsdebug("conn %p cq %p\n", conn, cq);
+
+	rds_ib_stats_inc(s_ib_rx_cq_call);
+
+	tasklet_schedule(&ic->i_recv_tasklet);
+}
+
+static inline void rds_poll_cq(struct rds_ib_connection *ic,
+			       struct rds_ib_ack_state *state)
+{
+	struct rds_connection *conn = ic->conn;
+	struct ib_wc wc;
+	struct rds_ib_recv_work *recv;
+
+	while (ib_poll_cq(ic->i_recv_cq, 1, &wc) > 0) {
+		rdsdebug("wc wr_id 0x%llx status %u (%s) byte_len %u imm_data %u\n",
+			 (unsigned long long)wc.wr_id, wc.status,
+			 rds_ib_wc_status_str(wc.status), wc.byte_len,
+			 be32_to_cpu(wc.ex.imm_data));
+		rds_ib_stats_inc(s_ib_rx_cq_event);
+
+		recv = &ic->i_recvs[rds_ib_ring_oldest(&ic->i_recv_ring)];
+
+		ib_dma_unmap_sg(ic->i_cm_id->device, &recv->r_frag->f_sg, 1, DMA_FROM_DEVICE);
+
+		/*
+		 * Also process recvs in connecting state because it is possible
+		 * to get a recv completion _before_ the rdmacm ESTABLISHED
+		 * event is processed.
+		 */
+		if (wc.status == IB_WC_SUCCESS) {
+			rds_ib_process_recv(conn, recv, wc.byte_len, state);
+		} else {
+			/* We expect errors as the qp is drained during shutdown */
+			if (rds_conn_up(conn) || rds_conn_connecting(conn))
+				rds_ib_conn_error(conn, "recv completion on %pI4 had "
+						  "status %u (%s), disconnecting and "
+						  "reconnecting\n", &conn->c_faddr,
+						  wc.status,
+						  rds_ib_wc_status_str(wc.status));
+		}
+
+		/*
+		 * It's very important that we only free this ring entry if we've truly
+		 * freed the resources allocated to the entry.  The refilling path can
+		 * leak if we don't.
+		 */
+		rds_ib_ring_free(&ic->i_recv_ring, 1);
+	}
+}
+
+void rds_ib_recv_tasklet_fn(unsigned long data)
+{
+	struct rds_ib_connection *ic = (struct rds_ib_connection *) data;
+	struct rds_connection *conn = ic->conn;
+	struct rds_ib_ack_state state = { 0, };
+
+	rds_poll_cq(ic, &state);
+	ib_req_notify_cq(ic->i_recv_cq, IB_CQ_SOLICITED);
+	rds_poll_cq(ic, &state);
+
+	if (state.ack_next_valid)
+		rds_ib_set_ack(ic, state.ack_next, state.ack_required);
+	if (state.ack_recv_valid && state.ack_recv > ic->i_ack_recv) {
+		rds_send_drop_acked(conn, state.ack_recv, NULL);
+		ic->i_ack_recv = state.ack_recv;
+	}
+	if (rds_conn_up(conn))
+		rds_ib_attempt_ack(ic);
+
+	/* If we ever end up with a really empty receive ring, we're
+	 * in deep trouble, as the sender will definitely see RNR
+	 * timeouts. */
+	if (rds_ib_ring_empty(&ic->i_recv_ring))
+		rds_ib_stats_inc(s_ib_rx_ring_empty);
+
+	if (rds_ib_ring_low(&ic->i_recv_ring))
+		rds_ib_recv_refill(conn, 0);
+}
+
+int rds_ib_recv(struct rds_connection *conn)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+	int ret = 0;
+
+	rdsdebug("conn %p\n", conn);
+	if (rds_conn_up(conn))
+		rds_ib_attempt_ack(ic);
+
+	return ret;
+}
+
+int rds_ib_recv_init(void)
+{
+	struct sysinfo si;
+	int ret = -ENOMEM;
+
+	/* Default to 30% of all available RAM for recv memory */
+	si_meminfo(&si);
+	rds_ib_sysctl_max_recv_allocation = si.totalram / 3 * PAGE_SIZE / RDS_FRAG_SIZE;
+
+	rds_ib_incoming_slab = kmem_cache_create("rds_ib_incoming",
+					sizeof(struct rds_ib_incoming),
+					0, SLAB_HWCACHE_ALIGN, NULL);
+	if (!rds_ib_incoming_slab)
+		goto out;
+
+	rds_ib_frag_slab = kmem_cache_create("rds_ib_frag",
+					sizeof(struct rds_page_frag),
+					0, SLAB_HWCACHE_ALIGN, NULL);
+	if (!rds_ib_frag_slab)
+		kmem_cache_destroy(rds_ib_incoming_slab);
+	else
+		ret = 0;
+out:
+	return ret;
+}
+
+void rds_ib_recv_exit(void)
+{
+	kmem_cache_destroy(rds_ib_incoming_slab);
+	kmem_cache_destroy(rds_ib_frag_slab);
+}
diff --git a/net/rds/ib_ring.c b/net/rds/ib_ring.c
new file mode 100644
index 000000000..ff97e8eda
--- /dev/null
+++ b/net/rds/ib_ring.c
@@ -0,0 +1,168 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+
+#include "rds.h"
+#include "ib.h"
+
+/*
+ * Locking for IB rings.
+ * We assume that allocation is always protected by a mutex
+ * in the caller (this is a valid assumption for the current
+ * implementation).
+ *
+ * Freeing always happens in an interrupt, and hence only
+ * races with allocations, but not with other free()s.
+ *
+ * The interaction between allocation and freeing is that
+ * the alloc code has to determine the number of free entries.
+ * To this end, we maintain two counters; an allocation counter
+ * and a free counter. Both are allowed to run freely, and wrap
+ * around.
+ * The number of used entries is always (alloc_ctr - free_ctr) % NR.
+ *
+ * The current implementation makes free_ctr atomic. When the
+ * caller finds an allocation fails, it should set an "alloc fail"
+ * bit and retry the allocation. The "alloc fail" bit essentially tells
+ * the CQ completion handlers to wake it up after freeing some
+ * more entries.
+ */
+
+/*
+ * This only happens on shutdown.
+ */
+DECLARE_WAIT_QUEUE_HEAD(rds_ib_ring_empty_wait);
+
+void rds_ib_ring_init(struct rds_ib_work_ring *ring, u32 nr)
+{
+	memset(ring, 0, sizeof(*ring));
+	ring->w_nr = nr;
+	rdsdebug("ring %p nr %u\n", ring, ring->w_nr);
+}
+
+static inline u32 __rds_ib_ring_used(struct rds_ib_work_ring *ring)
+{
+	u32 diff;
+
+	/* This assumes that atomic_t has at least as many bits as u32 */
+	diff = ring->w_alloc_ctr - (u32) atomic_read(&ring->w_free_ctr);
+	BUG_ON(diff > ring->w_nr);
+
+	return diff;
+}
+
+void rds_ib_ring_resize(struct rds_ib_work_ring *ring, u32 nr)
+{
+	/* We only ever get called from the connection setup code,
+	 * prior to creating the QP. */
+	BUG_ON(__rds_ib_ring_used(ring));
+	ring->w_nr = nr;
+}
+
+static int __rds_ib_ring_empty(struct rds_ib_work_ring *ring)
+{
+	return __rds_ib_ring_used(ring) == 0;
+}
+
+u32 rds_ib_ring_alloc(struct rds_ib_work_ring *ring, u32 val, u32 *pos)
+{
+	u32 ret = 0, avail;
+
+	avail = ring->w_nr - __rds_ib_ring_used(ring);
+
+	rdsdebug("ring %p val %u next %u free %u\n", ring, val,
+		 ring->w_alloc_ptr, avail);
+
+	if (val && avail) {
+		ret = min(val, avail);
+		*pos = ring->w_alloc_ptr;
+
+		ring->w_alloc_ptr = (ring->w_alloc_ptr + ret) % ring->w_nr;
+		ring->w_alloc_ctr += ret;
+	}
+
+	return ret;
+}
+
+void rds_ib_ring_free(struct rds_ib_work_ring *ring, u32 val)
+{
+	ring->w_free_ptr = (ring->w_free_ptr + val) % ring->w_nr;
+	atomic_add(val, &ring->w_free_ctr);
+
+	if (__rds_ib_ring_empty(ring) &&
+	    waitqueue_active(&rds_ib_ring_empty_wait))
+		wake_up(&rds_ib_ring_empty_wait);
+}
+
+void rds_ib_ring_unalloc(struct rds_ib_work_ring *ring, u32 val)
+{
+	ring->w_alloc_ptr = (ring->w_alloc_ptr - val) % ring->w_nr;
+	ring->w_alloc_ctr -= val;
+}
+
+int rds_ib_ring_empty(struct rds_ib_work_ring *ring)
+{
+	return __rds_ib_ring_empty(ring);
+}
+
+int rds_ib_ring_low(struct rds_ib_work_ring *ring)
+{
+	return __rds_ib_ring_used(ring) <= (ring->w_nr >> 1);
+}
+
+/*
+ * returns the oldest alloced ring entry.  This will be the next one
+ * freed.  This can't be called if there are none allocated.
+ */
+u32 rds_ib_ring_oldest(struct rds_ib_work_ring *ring)
+{
+	return ring->w_free_ptr;
+}
+
+/*
+ * returns the number of completed work requests.
+ */
+
+u32 rds_ib_ring_completed(struct rds_ib_work_ring *ring, u32 wr_id, u32 oldest)
+{
+	u32 ret;
+
+	if (oldest <= (unsigned long long)wr_id)
+		ret = (unsigned long long)wr_id - oldest + 1;
+	else
+		ret = ring->w_nr - oldest + (unsigned long long)wr_id + 1;
+
+	rdsdebug("ring %p ret %u wr_id %u oldest %u\n", ring, ret,
+		 wr_id, oldest);
+	return ret;
+}
diff --git a/net/rds/ib_send.c b/net/rds/ib_send.c
new file mode 100644
index 000000000..bd3825d38
--- /dev/null
+++ b/net/rds/ib_send.c
@@ -0,0 +1,1020 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/in.h>
+#include <linux/device.h>
+#include <linux/dmapool.h>
+#include <linux/ratelimit.h>
+
+#include "rds.h"
+#include "ib.h"
+
+static char *rds_ib_wc_status_strings[] = {
+#define RDS_IB_WC_STATUS_STR(foo) \
+		[IB_WC_##foo] = __stringify(IB_WC_##foo)
+	RDS_IB_WC_STATUS_STR(SUCCESS),
+	RDS_IB_WC_STATUS_STR(LOC_LEN_ERR),
+	RDS_IB_WC_STATUS_STR(LOC_QP_OP_ERR),
+	RDS_IB_WC_STATUS_STR(LOC_EEC_OP_ERR),
+	RDS_IB_WC_STATUS_STR(LOC_PROT_ERR),
+	RDS_IB_WC_STATUS_STR(WR_FLUSH_ERR),
+	RDS_IB_WC_STATUS_STR(MW_BIND_ERR),
+	RDS_IB_WC_STATUS_STR(BAD_RESP_ERR),
+	RDS_IB_WC_STATUS_STR(LOC_ACCESS_ERR),
+	RDS_IB_WC_STATUS_STR(REM_INV_REQ_ERR),
+	RDS_IB_WC_STATUS_STR(REM_ACCESS_ERR),
+	RDS_IB_WC_STATUS_STR(REM_OP_ERR),
+	RDS_IB_WC_STATUS_STR(RETRY_EXC_ERR),
+	RDS_IB_WC_STATUS_STR(RNR_RETRY_EXC_ERR),
+	RDS_IB_WC_STATUS_STR(LOC_RDD_VIOL_ERR),
+	RDS_IB_WC_STATUS_STR(REM_INV_RD_REQ_ERR),
+	RDS_IB_WC_STATUS_STR(REM_ABORT_ERR),
+	RDS_IB_WC_STATUS_STR(INV_EECN_ERR),
+	RDS_IB_WC_STATUS_STR(INV_EEC_STATE_ERR),
+	RDS_IB_WC_STATUS_STR(FATAL_ERR),
+	RDS_IB_WC_STATUS_STR(RESP_TIMEOUT_ERR),
+	RDS_IB_WC_STATUS_STR(GENERAL_ERR),
+#undef RDS_IB_WC_STATUS_STR
+};
+
+char *rds_ib_wc_status_str(enum ib_wc_status status)
+{
+	return rds_str_array(rds_ib_wc_status_strings,
+			     ARRAY_SIZE(rds_ib_wc_status_strings), status);
+}
+
+/*
+ * Convert IB-specific error message to RDS error message and call core
+ * completion handler.
+ */
+static void rds_ib_send_complete(struct rds_message *rm,
+				 int wc_status,
+				 void (*complete)(struct rds_message *rm, int status))
+{
+	int notify_status;
+
+	switch (wc_status) {
+	case IB_WC_WR_FLUSH_ERR:
+		return;
+
+	case IB_WC_SUCCESS:
+		notify_status = RDS_RDMA_SUCCESS;
+		break;
+
+	case IB_WC_REM_ACCESS_ERR:
+		notify_status = RDS_RDMA_REMOTE_ERROR;
+		break;
+
+	default:
+		notify_status = RDS_RDMA_OTHER_ERROR;
+		break;
+	}
+	complete(rm, notify_status);
+}
+
+static void rds_ib_send_unmap_data(struct rds_ib_connection *ic,
+				   struct rm_data_op *op,
+				   int wc_status)
+{
+	if (op->op_nents)
+		ib_dma_unmap_sg(ic->i_cm_id->device,
+				op->op_sg, op->op_nents,
+				DMA_TO_DEVICE);
+}
+
+static void rds_ib_send_unmap_rdma(struct rds_ib_connection *ic,
+				   struct rm_rdma_op *op,
+				   int wc_status)
+{
+	if (op->op_mapped) {
+		ib_dma_unmap_sg(ic->i_cm_id->device,
+				op->op_sg, op->op_nents,
+				op->op_write ? DMA_TO_DEVICE : DMA_FROM_DEVICE);
+		op->op_mapped = 0;
+	}
+
+	/* If the user asked for a completion notification on this
+	 * message, we can implement three different semantics:
+	 *  1.	Notify when we received the ACK on the RDS message
+	 *	that was queued with the RDMA. This provides reliable
+	 *	notification of RDMA status at the expense of a one-way
+	 *	packet delay.
+	 *  2.	Notify when the IB stack gives us the completion event for
+	 *	the RDMA operation.
+	 *  3.	Notify when the IB stack gives us the completion event for
+	 *	the accompanying RDS messages.
+	 * Here, we implement approach #3. To implement approach #2,
+	 * we would need to take an event for the rdma WR. To implement #1,
+	 * don't call rds_rdma_send_complete at all, and fall back to the notify
+	 * handling in the ACK processing code.
+	 *
+	 * Note: There's no need to explicitly sync any RDMA buffers using
+	 * ib_dma_sync_sg_for_cpu - the completion for the RDMA
+	 * operation itself unmapped the RDMA buffers, which takes care
+	 * of synching.
+	 */
+	rds_ib_send_complete(container_of(op, struct rds_message, rdma),
+			     wc_status, rds_rdma_send_complete);
+
+	if (op->op_write)
+		rds_stats_add(s_send_rdma_bytes, op->op_bytes);
+	else
+		rds_stats_add(s_recv_rdma_bytes, op->op_bytes);
+}
+
+static void rds_ib_send_unmap_atomic(struct rds_ib_connection *ic,
+				     struct rm_atomic_op *op,
+				     int wc_status)
+{
+	/* unmap atomic recvbuf */
+	if (op->op_mapped) {
+		ib_dma_unmap_sg(ic->i_cm_id->device, op->op_sg, 1,
+				DMA_FROM_DEVICE);
+		op->op_mapped = 0;
+	}
+
+	rds_ib_send_complete(container_of(op, struct rds_message, atomic),
+			     wc_status, rds_atomic_send_complete);
+
+	if (op->op_type == RDS_ATOMIC_TYPE_CSWP)
+		rds_ib_stats_inc(s_ib_atomic_cswp);
+	else
+		rds_ib_stats_inc(s_ib_atomic_fadd);
+}
+
+/*
+ * Unmap the resources associated with a struct send_work.
+ *
+ * Returns the rm for no good reason other than it is unobtainable
+ * other than by switching on wr.opcode, currently, and the caller,
+ * the event handler, needs it.
+ */
+static struct rds_message *rds_ib_send_unmap_op(struct rds_ib_connection *ic,
+						struct rds_ib_send_work *send,
+						int wc_status)
+{
+	struct rds_message *rm = NULL;
+
+	/* In the error case, wc.opcode sometimes contains garbage */
+	switch (send->s_wr.opcode) {
+	case IB_WR_SEND:
+		if (send->s_op) {
+			rm = container_of(send->s_op, struct rds_message, data);
+			rds_ib_send_unmap_data(ic, send->s_op, wc_status);
+		}
+		break;
+	case IB_WR_RDMA_WRITE:
+	case IB_WR_RDMA_READ:
+		if (send->s_op) {
+			rm = container_of(send->s_op, struct rds_message, rdma);
+			rds_ib_send_unmap_rdma(ic, send->s_op, wc_status);
+		}
+		break;
+	case IB_WR_ATOMIC_FETCH_AND_ADD:
+	case IB_WR_ATOMIC_CMP_AND_SWP:
+		if (send->s_op) {
+			rm = container_of(send->s_op, struct rds_message, atomic);
+			rds_ib_send_unmap_atomic(ic, send->s_op, wc_status);
+		}
+		break;
+	default:
+		printk_ratelimited(KERN_NOTICE
+			       "RDS/IB: %s: unexpected opcode 0x%x in WR!\n",
+			       __func__, send->s_wr.opcode);
+		break;
+	}
+
+	send->s_wr.opcode = 0xdead;
+
+	return rm;
+}
+
+void rds_ib_send_init_ring(struct rds_ib_connection *ic)
+{
+	struct rds_ib_send_work *send;
+	u32 i;
+
+	for (i = 0, send = ic->i_sends; i < ic->i_send_ring.w_nr; i++, send++) {
+		struct ib_sge *sge;
+
+		send->s_op = NULL;
+
+		send->s_wr.wr_id = i;
+		send->s_wr.sg_list = send->s_sge;
+		send->s_wr.ex.imm_data = 0;
+
+		sge = &send->s_sge[0];
+		sge->addr = ic->i_send_hdrs_dma + (i * sizeof(struct rds_header));
+		sge->length = sizeof(struct rds_header);
+		sge->lkey = ic->i_mr->lkey;
+
+		send->s_sge[1].lkey = ic->i_mr->lkey;
+	}
+}
+
+void rds_ib_send_clear_ring(struct rds_ib_connection *ic)
+{
+	struct rds_ib_send_work *send;
+	u32 i;
+
+	for (i = 0, send = ic->i_sends; i < ic->i_send_ring.w_nr; i++, send++) {
+		if (send->s_op && send->s_wr.opcode != 0xdead)
+			rds_ib_send_unmap_op(ic, send, IB_WC_WR_FLUSH_ERR);
+	}
+}
+
+/*
+ * The only fast path caller always has a non-zero nr, so we don't
+ * bother testing nr before performing the atomic sub.
+ */
+static void rds_ib_sub_signaled(struct rds_ib_connection *ic, int nr)
+{
+	if ((atomic_sub_return(nr, &ic->i_signaled_sends) == 0) &&
+	    waitqueue_active(&rds_ib_ring_empty_wait))
+		wake_up(&rds_ib_ring_empty_wait);
+	BUG_ON(atomic_read(&ic->i_signaled_sends) < 0);
+}
+
+/*
+ * The _oldest/_free ring operations here race cleanly with the alloc/unalloc
+ * operations performed in the send path.  As the sender allocs and potentially
+ * unallocs the next free entry in the ring it doesn't alter which is
+ * the next to be freed, which is what this is concerned with.
+ */
+void rds_ib_send_cq_comp_handler(struct ib_cq *cq, void *context)
+{
+	struct rds_connection *conn = context;
+	struct rds_ib_connection *ic = conn->c_transport_data;
+	struct rds_message *rm = NULL;
+	struct ib_wc wc;
+	struct rds_ib_send_work *send;
+	u32 completed;
+	u32 oldest;
+	u32 i = 0;
+	int ret;
+	int nr_sig = 0;
+
+	rdsdebug("cq %p conn %p\n", cq, conn);
+	rds_ib_stats_inc(s_ib_tx_cq_call);
+	ret = ib_req_notify_cq(cq, IB_CQ_NEXT_COMP);
+	if (ret)
+		rdsdebug("ib_req_notify_cq send failed: %d\n", ret);
+
+	while (ib_poll_cq(cq, 1, &wc) > 0) {
+		rdsdebug("wc wr_id 0x%llx status %u (%s) byte_len %u imm_data %u\n",
+			 (unsigned long long)wc.wr_id, wc.status,
+			 rds_ib_wc_status_str(wc.status), wc.byte_len,
+			 be32_to_cpu(wc.ex.imm_data));
+		rds_ib_stats_inc(s_ib_tx_cq_event);
+
+		if (wc.wr_id == RDS_IB_ACK_WR_ID) {
+			if (time_after(jiffies, ic->i_ack_queued + HZ/2))
+				rds_ib_stats_inc(s_ib_tx_stalled);
+			rds_ib_ack_send_complete(ic);
+			continue;
+		}
+
+		oldest = rds_ib_ring_oldest(&ic->i_send_ring);
+
+		completed = rds_ib_ring_completed(&ic->i_send_ring, wc.wr_id, oldest);
+
+		for (i = 0; i < completed; i++) {
+			send = &ic->i_sends[oldest];
+			if (send->s_wr.send_flags & IB_SEND_SIGNALED)
+				nr_sig++;
+
+			rm = rds_ib_send_unmap_op(ic, send, wc.status);
+
+			if (time_after(jiffies, send->s_queued + HZ/2))
+				rds_ib_stats_inc(s_ib_tx_stalled);
+
+			if (send->s_op) {
+				if (send->s_op == rm->m_final_op) {
+					/* If anyone waited for this message to get flushed out, wake
+					 * them up now */
+					rds_message_unmapped(rm);
+				}
+				rds_message_put(rm);
+				send->s_op = NULL;
+			}
+
+			oldest = (oldest + 1) % ic->i_send_ring.w_nr;
+		}
+
+		rds_ib_ring_free(&ic->i_send_ring, completed);
+		rds_ib_sub_signaled(ic, nr_sig);
+		nr_sig = 0;
+
+		if (test_and_clear_bit(RDS_LL_SEND_FULL, &conn->c_flags) ||
+		    test_bit(0, &conn->c_map_queued))
+			queue_delayed_work(rds_wq, &conn->c_send_w, 0);
+
+		/* We expect errors as the qp is drained during shutdown */
+		if (wc.status != IB_WC_SUCCESS && rds_conn_up(conn)) {
+			rds_ib_conn_error(conn, "send completion on %pI4 had status "
+					  "%u (%s), disconnecting and reconnecting\n",
+					  &conn->c_faddr, wc.status,
+					  rds_ib_wc_status_str(wc.status));
+		}
+	}
+}
+
+/*
+ * This is the main function for allocating credits when sending
+ * messages.
+ *
+ * Conceptually, we have two counters:
+ *  -	send credits: this tells us how many WRs we're allowed
+ *	to submit without overruning the receiver's queue. For
+ *	each SEND WR we post, we decrement this by one.
+ *
+ *  -	posted credits: this tells us how many WRs we recently
+ *	posted to the receive queue. This value is transferred
+ *	to the peer as a "credit update" in a RDS header field.
+ *	Every time we transmit credits to the peer, we subtract
+ *	the amount of transferred credits from this counter.
+ *
+ * It is essential that we avoid situations where both sides have
+ * exhausted their send credits, and are unable to send new credits
+ * to the peer. We achieve this by requiring that we send at least
+ * one credit update to the peer before exhausting our credits.
+ * When new credits arrive, we subtract one credit that is withheld
+ * until we've posted new buffers and are ready to transmit these
+ * credits (see rds_ib_send_add_credits below).
+ *
+ * The RDS send code is essentially single-threaded; rds_send_xmit
+ * sets RDS_IN_XMIT to ensure exclusive access to the send ring.
+ * However, the ACK sending code is independent and can race with
+ * message SENDs.
+ *
+ * In the send path, we need to update the counters for send credits
+ * and the counter of posted buffers atomically - when we use the
+ * last available credit, we cannot allow another thread to race us
+ * and grab the posted credits counter.  Hence, we have to use a
+ * spinlock to protect the credit counter, or use atomics.
+ *
+ * Spinlocks shared between the send and the receive path are bad,
+ * because they create unnecessary delays. An early implementation
+ * using a spinlock showed a 5% degradation in throughput at some
+ * loads.
+ *
+ * This implementation avoids spinlocks completely, putting both
+ * counters into a single atomic, and updating that atomic using
+ * atomic_add (in the receive path, when receiving fresh credits),
+ * and using atomic_cmpxchg when updating the two counters.
+ */
+int rds_ib_send_grab_credits(struct rds_ib_connection *ic,
+			     u32 wanted, u32 *adv_credits, int need_posted, int max_posted)
+{
+	unsigned int avail, posted, got = 0, advertise;
+	long oldval, newval;
+
+	*adv_credits = 0;
+	if (!ic->i_flowctl)
+		return wanted;
+
+try_again:
+	advertise = 0;
+	oldval = newval = atomic_read(&ic->i_credits);
+	posted = IB_GET_POST_CREDITS(oldval);
+	avail = IB_GET_SEND_CREDITS(oldval);
+
+	rdsdebug("wanted=%u credits=%u posted=%u\n",
+			wanted, avail, posted);
+
+	/* The last credit must be used to send a credit update. */
+	if (avail && !posted)
+		avail--;
+
+	if (avail < wanted) {
+		struct rds_connection *conn = ic->i_cm_id->context;
+
+		/* Oops, there aren't that many credits left! */
+		set_bit(RDS_LL_SEND_FULL, &conn->c_flags);
+		got = avail;
+	} else {
+		/* Sometimes you get what you want, lalala. */
+		got = wanted;
+	}
+	newval -= IB_SET_SEND_CREDITS(got);
+
+	/*
+	 * If need_posted is non-zero, then the caller wants
+	 * the posted regardless of whether any send credits are
+	 * available.
+	 */
+	if (posted && (got || need_posted)) {
+		advertise = min_t(unsigned int, posted, max_posted);
+		newval -= IB_SET_POST_CREDITS(advertise);
+	}
+
+	/* Finally bill everything */
+	if (atomic_cmpxchg(&ic->i_credits, oldval, newval) != oldval)
+		goto try_again;
+
+	*adv_credits = advertise;
+	return got;
+}
+
+void rds_ib_send_add_credits(struct rds_connection *conn, unsigned int credits)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+
+	if (credits == 0)
+		return;
+
+	rdsdebug("credits=%u current=%u%s\n",
+			credits,
+			IB_GET_SEND_CREDITS(atomic_read(&ic->i_credits)),
+			test_bit(RDS_LL_SEND_FULL, &conn->c_flags) ? ", ll_send_full" : "");
+
+	atomic_add(IB_SET_SEND_CREDITS(credits), &ic->i_credits);
+	if (test_and_clear_bit(RDS_LL_SEND_FULL, &conn->c_flags))
+		queue_delayed_work(rds_wq, &conn->c_send_w, 0);
+
+	WARN_ON(IB_GET_SEND_CREDITS(credits) >= 16384);
+
+	rds_ib_stats_inc(s_ib_rx_credit_updates);
+}
+
+void rds_ib_advertise_credits(struct rds_connection *conn, unsigned int posted)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+
+	if (posted == 0)
+		return;
+
+	atomic_add(IB_SET_POST_CREDITS(posted), &ic->i_credits);
+
+	/* Decide whether to send an update to the peer now.
+	 * If we would send a credit update for every single buffer we
+	 * post, we would end up with an ACK storm (ACK arrives,
+	 * consumes buffer, we refill the ring, send ACK to remote
+	 * advertising the newly posted buffer... ad inf)
+	 *
+	 * Performance pretty much depends on how often we send
+	 * credit updates - too frequent updates mean lots of ACKs.
+	 * Too infrequent updates, and the peer will run out of
+	 * credits and has to throttle.
+	 * For the time being, 16 seems to be a good compromise.
+	 */
+	if (IB_GET_POST_CREDITS(atomic_read(&ic->i_credits)) >= 16)
+		set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
+}
+
+static inline int rds_ib_set_wr_signal_state(struct rds_ib_connection *ic,
+					     struct rds_ib_send_work *send,
+					     bool notify)
+{
+	/*
+	 * We want to delay signaling completions just enough to get
+	 * the batching benefits but not so much that we create dead time
+	 * on the wire.
+	 */
+	if (ic->i_unsignaled_wrs-- == 0 || notify) {
+		ic->i_unsignaled_wrs = rds_ib_sysctl_max_unsig_wrs;
+		send->s_wr.send_flags |= IB_SEND_SIGNALED;
+		return 1;
+	}
+	return 0;
+}
+
+/*
+ * This can be called multiple times for a given message.  The first time
+ * we see a message we map its scatterlist into the IB device so that
+ * we can provide that mapped address to the IB scatter gather entries
+ * in the IB work requests.  We translate the scatterlist into a series
+ * of work requests that fragment the message.  These work requests complete
+ * in order so we pass ownership of the message to the completion handler
+ * once we send the final fragment.
+ *
+ * The RDS core uses the c_send_lock to only enter this function once
+ * per connection.  This makes sure that the tx ring alloc/unalloc pairs
+ * don't get out of sync and confuse the ring.
+ */
+int rds_ib_xmit(struct rds_connection *conn, struct rds_message *rm,
+		unsigned int hdr_off, unsigned int sg, unsigned int off)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+	struct ib_device *dev = ic->i_cm_id->device;
+	struct rds_ib_send_work *send = NULL;
+	struct rds_ib_send_work *first;
+	struct rds_ib_send_work *prev;
+	struct ib_send_wr *failed_wr;
+	struct scatterlist *scat;
+	u32 pos;
+	u32 i;
+	u32 work_alloc;
+	u32 credit_alloc = 0;
+	u32 posted;
+	u32 adv_credits = 0;
+	int send_flags = 0;
+	int bytes_sent = 0;
+	int ret;
+	int flow_controlled = 0;
+	int nr_sig = 0;
+
+	BUG_ON(off % RDS_FRAG_SIZE);
+	BUG_ON(hdr_off != 0 && hdr_off != sizeof(struct rds_header));
+
+	/* Do not send cong updates to IB loopback */
+	if (conn->c_loopback
+	    && rm->m_inc.i_hdr.h_flags & RDS_FLAG_CONG_BITMAP) {
+		rds_cong_map_updated(conn->c_fcong, ~(u64) 0);
+		scat = &rm->data.op_sg[sg];
+		ret = max_t(int, RDS_CONG_MAP_BYTES, scat->length);
+		return sizeof(struct rds_header) + ret;
+	}
+
+	/* FIXME we may overallocate here */
+	if (be32_to_cpu(rm->m_inc.i_hdr.h_len) == 0)
+		i = 1;
+	else
+		i = ceil(be32_to_cpu(rm->m_inc.i_hdr.h_len), RDS_FRAG_SIZE);
+
+	work_alloc = rds_ib_ring_alloc(&ic->i_send_ring, i, &pos);
+	if (work_alloc == 0) {
+		set_bit(RDS_LL_SEND_FULL, &conn->c_flags);
+		rds_ib_stats_inc(s_ib_tx_ring_full);
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	if (ic->i_flowctl) {
+		credit_alloc = rds_ib_send_grab_credits(ic, work_alloc, &posted, 0, RDS_MAX_ADV_CREDIT);
+		adv_credits += posted;
+		if (credit_alloc < work_alloc) {
+			rds_ib_ring_unalloc(&ic->i_send_ring, work_alloc - credit_alloc);
+			work_alloc = credit_alloc;
+			flow_controlled = 1;
+		}
+		if (work_alloc == 0) {
+			set_bit(RDS_LL_SEND_FULL, &conn->c_flags);
+			rds_ib_stats_inc(s_ib_tx_throttle);
+			ret = -ENOMEM;
+			goto out;
+		}
+	}
+
+	/* map the message the first time we see it */
+	if (!ic->i_data_op) {
+		if (rm->data.op_nents) {
+			rm->data.op_count = ib_dma_map_sg(dev,
+							  rm->data.op_sg,
+							  rm->data.op_nents,
+							  DMA_TO_DEVICE);
+			rdsdebug("ic %p mapping rm %p: %d\n", ic, rm, rm->data.op_count);
+			if (rm->data.op_count == 0) {
+				rds_ib_stats_inc(s_ib_tx_sg_mapping_failure);
+				rds_ib_ring_unalloc(&ic->i_send_ring, work_alloc);
+				ret = -ENOMEM; /* XXX ? */
+				goto out;
+			}
+		} else {
+			rm->data.op_count = 0;
+		}
+
+		rds_message_addref(rm);
+		ic->i_data_op = &rm->data;
+
+		/* Finalize the header */
+		if (test_bit(RDS_MSG_ACK_REQUIRED, &rm->m_flags))
+			rm->m_inc.i_hdr.h_flags |= RDS_FLAG_ACK_REQUIRED;
+		if (test_bit(RDS_MSG_RETRANSMITTED, &rm->m_flags))
+			rm->m_inc.i_hdr.h_flags |= RDS_FLAG_RETRANSMITTED;
+
+		/* If it has a RDMA op, tell the peer we did it. This is
+		 * used by the peer to release use-once RDMA MRs. */
+		if (rm->rdma.op_active) {
+			struct rds_ext_header_rdma ext_hdr;
+
+			ext_hdr.h_rdma_rkey = cpu_to_be32(rm->rdma.op_rkey);
+			rds_message_add_extension(&rm->m_inc.i_hdr,
+					RDS_EXTHDR_RDMA, &ext_hdr, sizeof(ext_hdr));
+		}
+		if (rm->m_rdma_cookie) {
+			rds_message_add_rdma_dest_extension(&rm->m_inc.i_hdr,
+					rds_rdma_cookie_key(rm->m_rdma_cookie),
+					rds_rdma_cookie_offset(rm->m_rdma_cookie));
+		}
+
+		/* Note - rds_ib_piggyb_ack clears the ACK_REQUIRED bit, so
+		 * we should not do this unless we have a chance of at least
+		 * sticking the header into the send ring. Which is why we
+		 * should call rds_ib_ring_alloc first. */
+		rm->m_inc.i_hdr.h_ack = cpu_to_be64(rds_ib_piggyb_ack(ic));
+		rds_message_make_checksum(&rm->m_inc.i_hdr);
+
+		/*
+		 * Update adv_credits since we reset the ACK_REQUIRED bit.
+		 */
+		if (ic->i_flowctl) {
+			rds_ib_send_grab_credits(ic, 0, &posted, 1, RDS_MAX_ADV_CREDIT - adv_credits);
+			adv_credits += posted;
+			BUG_ON(adv_credits > 255);
+		}
+	}
+
+	/* Sometimes you want to put a fence between an RDMA
+	 * READ and the following SEND.
+	 * We could either do this all the time
+	 * or when requested by the user. Right now, we let
+	 * the application choose.
+	 */
+	if (rm->rdma.op_active && rm->rdma.op_fence)
+		send_flags = IB_SEND_FENCE;
+
+	/* Each frag gets a header. Msgs may be 0 bytes */
+	send = &ic->i_sends[pos];
+	first = send;
+	prev = NULL;
+	scat = &ic->i_data_op->op_sg[sg];
+	i = 0;
+	do {
+		unsigned int len = 0;
+
+		/* Set up the header */
+		send->s_wr.send_flags = send_flags;
+		send->s_wr.opcode = IB_WR_SEND;
+		send->s_wr.num_sge = 1;
+		send->s_wr.next = NULL;
+		send->s_queued = jiffies;
+		send->s_op = NULL;
+
+		send->s_sge[0].addr = ic->i_send_hdrs_dma
+			+ (pos * sizeof(struct rds_header));
+		send->s_sge[0].length = sizeof(struct rds_header);
+
+		memcpy(&ic->i_send_hdrs[pos], &rm->m_inc.i_hdr, sizeof(struct rds_header));
+
+		/* Set up the data, if present */
+		if (i < work_alloc
+		    && scat != &rm->data.op_sg[rm->data.op_count]) {
+			len = min(RDS_FRAG_SIZE, ib_sg_dma_len(dev, scat) - off);
+			send->s_wr.num_sge = 2;
+
+			send->s_sge[1].addr = ib_sg_dma_address(dev, scat) + off;
+			send->s_sge[1].length = len;
+
+			bytes_sent += len;
+			off += len;
+			if (off == ib_sg_dma_len(dev, scat)) {
+				scat++;
+				off = 0;
+			}
+		}
+
+		rds_ib_set_wr_signal_state(ic, send, 0);
+
+		/*
+		 * Always signal the last one if we're stopping due to flow control.
+		 */
+		if (ic->i_flowctl && flow_controlled && i == (work_alloc-1))
+			send->s_wr.send_flags |= IB_SEND_SIGNALED | IB_SEND_SOLICITED;
+
+		if (send->s_wr.send_flags & IB_SEND_SIGNALED)
+			nr_sig++;
+
+		rdsdebug("send %p wr %p num_sge %u next %p\n", send,
+			 &send->s_wr, send->s_wr.num_sge, send->s_wr.next);
+
+		if (ic->i_flowctl && adv_credits) {
+			struct rds_header *hdr = &ic->i_send_hdrs[pos];
+
+			/* add credit and redo the header checksum */
+			hdr->h_credit = adv_credits;
+			rds_message_make_checksum(hdr);
+			adv_credits = 0;
+			rds_ib_stats_inc(s_ib_tx_credit_updates);
+		}
+
+		if (prev)
+			prev->s_wr.next = &send->s_wr;
+		prev = send;
+
+		pos = (pos + 1) % ic->i_send_ring.w_nr;
+		send = &ic->i_sends[pos];
+		i++;
+
+	} while (i < work_alloc
+		 && scat != &rm->data.op_sg[rm->data.op_count]);
+
+	/* Account the RDS header in the number of bytes we sent, but just once.
+	 * The caller has no concept of fragmentation. */
+	if (hdr_off == 0)
+		bytes_sent += sizeof(struct rds_header);
+
+	/* if we finished the message then send completion owns it */
+	if (scat == &rm->data.op_sg[rm->data.op_count]) {
+		prev->s_op = ic->i_data_op;
+		prev->s_wr.send_flags |= IB_SEND_SOLICITED;
+		ic->i_data_op = NULL;
+	}
+
+	/* Put back wrs & credits we didn't use */
+	if (i < work_alloc) {
+		rds_ib_ring_unalloc(&ic->i_send_ring, work_alloc - i);
+		work_alloc = i;
+	}
+	if (ic->i_flowctl && i < credit_alloc)
+		rds_ib_send_add_credits(conn, credit_alloc - i);
+
+	if (nr_sig)
+		atomic_add(nr_sig, &ic->i_signaled_sends);
+
+	/* XXX need to worry about failed_wr and partial sends. */
+	failed_wr = &first->s_wr;
+	ret = ib_post_send(ic->i_cm_id->qp, &first->s_wr, &failed_wr);
+	rdsdebug("ic %p first %p (wr %p) ret %d wr %p\n", ic,
+		 first, &first->s_wr, ret, failed_wr);
+	BUG_ON(failed_wr != &first->s_wr);
+	if (ret) {
+		printk(KERN_WARNING "RDS/IB: ib_post_send to %pI4 "
+		       "returned %d\n", &conn->c_faddr, ret);
+		rds_ib_ring_unalloc(&ic->i_send_ring, work_alloc);
+		rds_ib_sub_signaled(ic, nr_sig);
+		if (prev->s_op) {
+			ic->i_data_op = prev->s_op;
+			prev->s_op = NULL;
+		}
+
+		rds_ib_conn_error(ic->conn, "ib_post_send failed\n");
+		goto out;
+	}
+
+	ret = bytes_sent;
+out:
+	BUG_ON(adv_credits);
+	return ret;
+}
+
+/*
+ * Issue atomic operation.
+ * A simplified version of the rdma case, we always map 1 SG, and
+ * only 8 bytes, for the return value from the atomic operation.
+ */
+int rds_ib_xmit_atomic(struct rds_connection *conn, struct rm_atomic_op *op)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+	struct rds_ib_send_work *send = NULL;
+	struct ib_send_wr *failed_wr;
+	struct rds_ib_device *rds_ibdev;
+	u32 pos;
+	u32 work_alloc;
+	int ret;
+	int nr_sig = 0;
+
+	rds_ibdev = ib_get_client_data(ic->i_cm_id->device, &rds_ib_client);
+
+	work_alloc = rds_ib_ring_alloc(&ic->i_send_ring, 1, &pos);
+	if (work_alloc != 1) {
+		rds_ib_ring_unalloc(&ic->i_send_ring, work_alloc);
+		rds_ib_stats_inc(s_ib_tx_ring_full);
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	/* address of send request in ring */
+	send = &ic->i_sends[pos];
+	send->s_queued = jiffies;
+
+	if (op->op_type == RDS_ATOMIC_TYPE_CSWP) {
+		send->s_wr.opcode = IB_WR_MASKED_ATOMIC_CMP_AND_SWP;
+		send->s_wr.wr.atomic.compare_add = op->op_m_cswp.compare;
+		send->s_wr.wr.atomic.swap = op->op_m_cswp.swap;
+		send->s_wr.wr.atomic.compare_add_mask = op->op_m_cswp.compare_mask;
+		send->s_wr.wr.atomic.swap_mask = op->op_m_cswp.swap_mask;
+	} else { /* FADD */
+		send->s_wr.opcode = IB_WR_MASKED_ATOMIC_FETCH_AND_ADD;
+		send->s_wr.wr.atomic.compare_add = op->op_m_fadd.add;
+		send->s_wr.wr.atomic.swap = 0;
+		send->s_wr.wr.atomic.compare_add_mask = op->op_m_fadd.nocarry_mask;
+		send->s_wr.wr.atomic.swap_mask = 0;
+	}
+	nr_sig = rds_ib_set_wr_signal_state(ic, send, op->op_notify);
+	send->s_wr.num_sge = 1;
+	send->s_wr.next = NULL;
+	send->s_wr.wr.atomic.remote_addr = op->op_remote_addr;
+	send->s_wr.wr.atomic.rkey = op->op_rkey;
+	send->s_op = op;
+	rds_message_addref(container_of(send->s_op, struct rds_message, atomic));
+
+	/* map 8 byte retval buffer to the device */
+	ret = ib_dma_map_sg(ic->i_cm_id->device, op->op_sg, 1, DMA_FROM_DEVICE);
+	rdsdebug("ic %p mapping atomic op %p. mapped %d pg\n", ic, op, ret);
+	if (ret != 1) {
+		rds_ib_ring_unalloc(&ic->i_send_ring, work_alloc);
+		rds_ib_stats_inc(s_ib_tx_sg_mapping_failure);
+		ret = -ENOMEM; /* XXX ? */
+		goto out;
+	}
+
+	/* Convert our struct scatterlist to struct ib_sge */
+	send->s_sge[0].addr = ib_sg_dma_address(ic->i_cm_id->device, op->op_sg);
+	send->s_sge[0].length = ib_sg_dma_len(ic->i_cm_id->device, op->op_sg);
+	send->s_sge[0].lkey = ic->i_mr->lkey;
+
+	rdsdebug("rva %Lx rpa %Lx len %u\n", op->op_remote_addr,
+		 send->s_sge[0].addr, send->s_sge[0].length);
+
+	if (nr_sig)
+		atomic_add(nr_sig, &ic->i_signaled_sends);
+
+	failed_wr = &send->s_wr;
+	ret = ib_post_send(ic->i_cm_id->qp, &send->s_wr, &failed_wr);
+	rdsdebug("ic %p send %p (wr %p) ret %d wr %p\n", ic,
+		 send, &send->s_wr, ret, failed_wr);
+	BUG_ON(failed_wr != &send->s_wr);
+	if (ret) {
+		printk(KERN_WARNING "RDS/IB: atomic ib_post_send to %pI4 "
+		       "returned %d\n", &conn->c_faddr, ret);
+		rds_ib_ring_unalloc(&ic->i_send_ring, work_alloc);
+		rds_ib_sub_signaled(ic, nr_sig);
+		goto out;
+	}
+
+	if (unlikely(failed_wr != &send->s_wr)) {
+		printk(KERN_WARNING "RDS/IB: atomic ib_post_send() rc=%d, but failed_wqe updated!\n", ret);
+		BUG_ON(failed_wr != &send->s_wr);
+	}
+
+out:
+	return ret;
+}
+
+int rds_ib_xmit_rdma(struct rds_connection *conn, struct rm_rdma_op *op)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+	struct rds_ib_send_work *send = NULL;
+	struct rds_ib_send_work *first;
+	struct rds_ib_send_work *prev;
+	struct ib_send_wr *failed_wr;
+	struct scatterlist *scat;
+	unsigned long len;
+	u64 remote_addr = op->op_remote_addr;
+	u32 max_sge = ic->rds_ibdev->max_sge;
+	u32 pos;
+	u32 work_alloc;
+	u32 i;
+	u32 j;
+	int sent;
+	int ret;
+	int num_sge;
+	int nr_sig = 0;
+
+	/* map the op the first time we see it */
+	if (!op->op_mapped) {
+		op->op_count = ib_dma_map_sg(ic->i_cm_id->device,
+					     op->op_sg, op->op_nents, (op->op_write) ?
+					     DMA_TO_DEVICE : DMA_FROM_DEVICE);
+		rdsdebug("ic %p mapping op %p: %d\n", ic, op, op->op_count);
+		if (op->op_count == 0) {
+			rds_ib_stats_inc(s_ib_tx_sg_mapping_failure);
+			ret = -ENOMEM; /* XXX ? */
+			goto out;
+		}
+
+		op->op_mapped = 1;
+	}
+
+	/*
+	 * Instead of knowing how to return a partial rdma read/write we insist that there
+	 * be enough work requests to send the entire message.
+	 */
+	i = ceil(op->op_count, max_sge);
+
+	work_alloc = rds_ib_ring_alloc(&ic->i_send_ring, i, &pos);
+	if (work_alloc != i) {
+		rds_ib_ring_unalloc(&ic->i_send_ring, work_alloc);
+		rds_ib_stats_inc(s_ib_tx_ring_full);
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	send = &ic->i_sends[pos];
+	first = send;
+	prev = NULL;
+	scat = &op->op_sg[0];
+	sent = 0;
+	num_sge = op->op_count;
+
+	for (i = 0; i < work_alloc && scat != &op->op_sg[op->op_count]; i++) {
+		send->s_wr.send_flags = 0;
+		send->s_queued = jiffies;
+		send->s_op = NULL;
+
+		nr_sig += rds_ib_set_wr_signal_state(ic, send, op->op_notify);
+
+		send->s_wr.opcode = op->op_write ? IB_WR_RDMA_WRITE : IB_WR_RDMA_READ;
+		send->s_wr.wr.rdma.remote_addr = remote_addr;
+		send->s_wr.wr.rdma.rkey = op->op_rkey;
+
+		if (num_sge > max_sge) {
+			send->s_wr.num_sge = max_sge;
+			num_sge -= max_sge;
+		} else {
+			send->s_wr.num_sge = num_sge;
+		}
+
+		send->s_wr.next = NULL;
+
+		if (prev)
+			prev->s_wr.next = &send->s_wr;
+
+		for (j = 0; j < send->s_wr.num_sge && scat != &op->op_sg[op->op_count]; j++) {
+			len = ib_sg_dma_len(ic->i_cm_id->device, scat);
+			send->s_sge[j].addr =
+				 ib_sg_dma_address(ic->i_cm_id->device, scat);
+			send->s_sge[j].length = len;
+			send->s_sge[j].lkey = ic->i_mr->lkey;
+
+			sent += len;
+			rdsdebug("ic %p sent %d remote_addr %llu\n", ic, sent, remote_addr);
+
+			remote_addr += len;
+			scat++;
+		}
+
+		rdsdebug("send %p wr %p num_sge %u next %p\n", send,
+			&send->s_wr, send->s_wr.num_sge, send->s_wr.next);
+
+		prev = send;
+		if (++send == &ic->i_sends[ic->i_send_ring.w_nr])
+			send = ic->i_sends;
+	}
+
+	/* give a reference to the last op */
+	if (scat == &op->op_sg[op->op_count]) {
+		prev->s_op = op;
+		rds_message_addref(container_of(op, struct rds_message, rdma));
+	}
+
+	if (i < work_alloc) {
+		rds_ib_ring_unalloc(&ic->i_send_ring, work_alloc - i);
+		work_alloc = i;
+	}
+
+	if (nr_sig)
+		atomic_add(nr_sig, &ic->i_signaled_sends);
+
+	failed_wr = &first->s_wr;
+	ret = ib_post_send(ic->i_cm_id->qp, &first->s_wr, &failed_wr);
+	rdsdebug("ic %p first %p (wr %p) ret %d wr %p\n", ic,
+		 first, &first->s_wr, ret, failed_wr);
+	BUG_ON(failed_wr != &first->s_wr);
+	if (ret) {
+		printk(KERN_WARNING "RDS/IB: rdma ib_post_send to %pI4 "
+		       "returned %d\n", &conn->c_faddr, ret);
+		rds_ib_ring_unalloc(&ic->i_send_ring, work_alloc);
+		rds_ib_sub_signaled(ic, nr_sig);
+		goto out;
+	}
+
+	if (unlikely(failed_wr != &first->s_wr)) {
+		printk(KERN_WARNING "RDS/IB: ib_post_send() rc=%d, but failed_wqe updated!\n", ret);
+		BUG_ON(failed_wr != &first->s_wr);
+	}
+
+
+out:
+	return ret;
+}
+
+void rds_ib_xmit_complete(struct rds_connection *conn)
+{
+	struct rds_ib_connection *ic = conn->c_transport_data;
+
+	/* We may have a pending ACK or window update we were unable
+	 * to send previously (due to flow control). Try again. */
+	rds_ib_attempt_ack(ic);
+}
diff --git a/net/rds/ib_stats.c b/net/rds/ib_stats.c
new file mode 100644
index 000000000..2d5965d6e
--- /dev/null
+++ b/net/rds/ib_stats.c
@@ -0,0 +1,97 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/percpu.h>
+#include <linux/seq_file.h>
+#include <linux/proc_fs.h>
+
+#include "rds.h"
+#include "ib.h"
+
+DEFINE_PER_CPU_SHARED_ALIGNED(struct rds_ib_statistics, rds_ib_stats);
+
+static const char *const rds_ib_stat_names[] = {
+	"ib_connect_raced",
+	"ib_listen_closed_stale",
+	"ib_tx_cq_call",
+	"ib_tx_cq_event",
+	"ib_tx_ring_full",
+	"ib_tx_throttle",
+	"ib_tx_sg_mapping_failure",
+	"ib_tx_stalled",
+	"ib_tx_credit_updates",
+	"ib_rx_cq_call",
+	"ib_rx_cq_event",
+	"ib_rx_ring_empty",
+	"ib_rx_refill_from_cq",
+	"ib_rx_refill_from_thread",
+	"ib_rx_alloc_limit",
+	"ib_rx_credit_updates",
+	"ib_ack_sent",
+	"ib_ack_send_failure",
+	"ib_ack_send_delayed",
+	"ib_ack_send_piggybacked",
+	"ib_ack_received",
+	"ib_rdma_mr_alloc",
+	"ib_rdma_mr_free",
+	"ib_rdma_mr_used",
+	"ib_rdma_mr_pool_flush",
+	"ib_rdma_mr_pool_wait",
+	"ib_rdma_mr_pool_depleted",
+	"ib_atomic_cswp",
+	"ib_atomic_fadd",
+};
+
+unsigned int rds_ib_stats_info_copy(struct rds_info_iterator *iter,
+				    unsigned int avail)
+{
+	struct rds_ib_statistics stats = {0, };
+	uint64_t *src;
+	uint64_t *sum;
+	size_t i;
+	int cpu;
+
+	if (avail < ARRAY_SIZE(rds_ib_stat_names))
+		goto out;
+
+	for_each_online_cpu(cpu) {
+		src = (uint64_t *)&(per_cpu(rds_ib_stats, cpu));
+		sum = (uint64_t *)&stats;
+		for (i = 0; i < sizeof(stats) / sizeof(uint64_t); i++)
+			*(sum++) += *(src++);
+	}
+
+	rds_stats_info_copy(iter, (uint64_t *)&stats, rds_ib_stat_names,
+			    ARRAY_SIZE(rds_ib_stat_names));
+out:
+	return ARRAY_SIZE(rds_ib_stat_names);
+}
diff --git a/net/rds/ib_sysctl.c b/net/rds/ib_sysctl.c
new file mode 100644
index 000000000..e4e41b3af
--- /dev/null
+++ b/net/rds/ib_sysctl.c
@@ -0,0 +1,121 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/sysctl.h>
+#include <linux/proc_fs.h>
+
+#include "ib.h"
+
+static struct ctl_table_header *rds_ib_sysctl_hdr;
+
+unsigned long rds_ib_sysctl_max_send_wr = RDS_IB_DEFAULT_SEND_WR;
+unsigned long rds_ib_sysctl_max_recv_wr = RDS_IB_DEFAULT_RECV_WR;
+unsigned long rds_ib_sysctl_max_recv_allocation = (128 * 1024 * 1024) / RDS_FRAG_SIZE;
+static unsigned long rds_ib_sysctl_max_wr_min = 1;
+/* hardware will fail CQ creation long before this */
+static unsigned long rds_ib_sysctl_max_wr_max = (u32)~0;
+
+unsigned long rds_ib_sysctl_max_unsig_wrs = 16;
+static unsigned long rds_ib_sysctl_max_unsig_wr_min = 1;
+static unsigned long rds_ib_sysctl_max_unsig_wr_max = 64;
+
+/*
+ * This sysctl does nothing.
+ *
+ * Backwards compatibility with RDS 3.0 wire protocol
+ * disables initial FC credit exchange.
+ * If it's ever possible to drop 3.0 support,
+ * setting this to 1 and moving init/refill of send/recv
+ * rings from ib_cm_connect_complete() back into ib_setup_qp()
+ * will cause credits to be added before protocol negotiation.
+ */
+unsigned int rds_ib_sysctl_flow_control = 0;
+
+static struct ctl_table rds_ib_sysctl_table[] = {
+	{
+		.procname       = "max_send_wr",
+		.data		= &rds_ib_sysctl_max_send_wr,
+		.maxlen         = sizeof(unsigned long),
+		.mode           = 0644,
+		.proc_handler   = proc_doulongvec_minmax,
+		.extra1		= &rds_ib_sysctl_max_wr_min,
+		.extra2		= &rds_ib_sysctl_max_wr_max,
+	},
+	{
+		.procname       = "max_recv_wr",
+		.data		= &rds_ib_sysctl_max_recv_wr,
+		.maxlen         = sizeof(unsigned long),
+		.mode           = 0644,
+		.proc_handler   = proc_doulongvec_minmax,
+		.extra1		= &rds_ib_sysctl_max_wr_min,
+		.extra2		= &rds_ib_sysctl_max_wr_max,
+	},
+	{
+		.procname       = "max_unsignaled_wr",
+		.data		= &rds_ib_sysctl_max_unsig_wrs,
+		.maxlen         = sizeof(unsigned long),
+		.mode           = 0644,
+		.proc_handler   = proc_doulongvec_minmax,
+		.extra1		= &rds_ib_sysctl_max_unsig_wr_min,
+		.extra2		= &rds_ib_sysctl_max_unsig_wr_max,
+	},
+	{
+		.procname       = "max_recv_allocation",
+		.data		= &rds_ib_sysctl_max_recv_allocation,
+		.maxlen         = sizeof(unsigned long),
+		.mode           = 0644,
+		.proc_handler   = proc_doulongvec_minmax,
+	},
+	{
+		.procname	= "flow_control",
+		.data		= &rds_ib_sysctl_flow_control,
+		.maxlen		= sizeof(rds_ib_sysctl_flow_control),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec,
+	},
+	{ }
+};
+
+void rds_ib_sysctl_exit(void)
+{
+	if (rds_ib_sysctl_hdr)
+		unregister_net_sysctl_table(rds_ib_sysctl_hdr);
+}
+
+int rds_ib_sysctl_init(void)
+{
+	rds_ib_sysctl_hdr = register_net_sysctl(&init_net, "net/rds/ib", rds_ib_sysctl_table);
+	if (!rds_ib_sysctl_hdr)
+		return -ENOMEM;
+	return 0;
+}
diff --git a/net/rds/info.c b/net/rds/info.c
new file mode 100644
index 000000000..9a6b4f661
--- /dev/null
+++ b/net/rds/info.c
@@ -0,0 +1,243 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/percpu.h>
+#include <linux/seq_file.h>
+#include <linux/slab.h>
+#include <linux/proc_fs.h>
+#include <linux/export.h>
+
+#include "rds.h"
+
+/*
+ * This file implements a getsockopt() call which copies a set of fixed
+ * sized structs into a user-specified buffer as a means of providing
+ * read-only information about RDS.
+ *
+ * For a given information source there are a given number of fixed sized
+ * structs at a given time.  The structs are only copied if the user-specified
+ * buffer is big enough.  The destination pages that make up the buffer
+ * are pinned for the duration of the copy.
+ *
+ * This gives us the following benefits:
+ *
+ * - simple implementation, no copy "position" across multiple calls
+ * - consistent snapshot of an info source
+ * - atomic copy works well with whatever locking info source has
+ * - one portable tool to get rds info across implementations
+ * - long-lived tool can get info without allocating
+ *
+ * at the following costs:
+ *
+ * - info source copy must be pinned, may be "large"
+ */
+
+struct rds_info_iterator {
+	struct page **pages;
+	void *addr;
+	unsigned long offset;
+};
+
+static DEFINE_SPINLOCK(rds_info_lock);
+static rds_info_func rds_info_funcs[RDS_INFO_LAST - RDS_INFO_FIRST + 1];
+
+void rds_info_register_func(int optname, rds_info_func func)
+{
+	int offset = optname - RDS_INFO_FIRST;
+
+	BUG_ON(optname < RDS_INFO_FIRST || optname > RDS_INFO_LAST);
+
+	spin_lock(&rds_info_lock);
+	BUG_ON(rds_info_funcs[offset]);
+	rds_info_funcs[offset] = func;
+	spin_unlock(&rds_info_lock);
+}
+EXPORT_SYMBOL_GPL(rds_info_register_func);
+
+void rds_info_deregister_func(int optname, rds_info_func func)
+{
+	int offset = optname - RDS_INFO_FIRST;
+
+	BUG_ON(optname < RDS_INFO_FIRST || optname > RDS_INFO_LAST);
+
+	spin_lock(&rds_info_lock);
+	BUG_ON(rds_info_funcs[offset] != func);
+	rds_info_funcs[offset] = NULL;
+	spin_unlock(&rds_info_lock);
+}
+EXPORT_SYMBOL_GPL(rds_info_deregister_func);
+
+/*
+ * Typically we hold an atomic kmap across multiple rds_info_copy() calls
+ * because the kmap is so expensive.  This must be called before using blocking
+ * operations while holding the mapping and as the iterator is torn down.
+ */
+void rds_info_iter_unmap(struct rds_info_iterator *iter)
+{
+	if (iter->addr) {
+		kunmap_atomic(iter->addr);
+		iter->addr = NULL;
+	}
+}
+
+/*
+ * get_user_pages() called flush_dcache_page() on the pages for us.
+ */
+void rds_info_copy(struct rds_info_iterator *iter, void *data,
+		   unsigned long bytes)
+{
+	unsigned long this;
+
+	while (bytes) {
+		if (!iter->addr)
+			iter->addr = kmap_atomic(*iter->pages);
+
+		this = min(bytes, PAGE_SIZE - iter->offset);
+
+		rdsdebug("page %p addr %p offset %lu this %lu data %p "
+			  "bytes %lu\n", *iter->pages, iter->addr,
+			  iter->offset, this, data, bytes);
+
+		memcpy(iter->addr + iter->offset, data, this);
+
+		data += this;
+		bytes -= this;
+		iter->offset += this;
+
+		if (iter->offset == PAGE_SIZE) {
+			kunmap_atomic(iter->addr);
+			iter->addr = NULL;
+			iter->offset = 0;
+			iter->pages++;
+		}
+	}
+}
+EXPORT_SYMBOL_GPL(rds_info_copy);
+
+/*
+ * @optval points to the userspace buffer that the information snapshot
+ * will be copied into.
+ *
+ * @optlen on input is the size of the buffer in userspace.  @optlen
+ * on output is the size of the requested snapshot in bytes.
+ *
+ * This function returns -errno if there is a failure, particularly -ENOSPC
+ * if the given userspace buffer was not large enough to fit the snapshot.
+ * On success it returns the positive number of bytes of each array element
+ * in the snapshot.
+ */
+int rds_info_getsockopt(struct socket *sock, int optname, char __user *optval,
+			int __user *optlen)
+{
+	struct rds_info_iterator iter;
+	struct rds_info_lengths lens;
+	unsigned long nr_pages = 0;
+	unsigned long start;
+	unsigned long i;
+	rds_info_func func;
+	struct page **pages = NULL;
+	int ret;
+	int len;
+	int total;
+
+	if (get_user(len, optlen)) {
+		ret = -EFAULT;
+		goto out;
+	}
+
+	/* check for all kinds of wrapping and the like */
+	start = (unsigned long)optval;
+	if (len < 0 || len + PAGE_SIZE - 1 < len || start + len < start) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	/* a 0 len call is just trying to probe its length */
+	if (len == 0)
+		goto call_func;
+
+	nr_pages = (PAGE_ALIGN(start + len) - (start & PAGE_MASK))
+			>> PAGE_SHIFT;
+
+	pages = kmalloc(nr_pages * sizeof(struct page *), GFP_KERNEL);
+	if (!pages) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	ret = get_user_pages_fast(start, nr_pages, 1, pages);
+	if (ret != nr_pages) {
+		if (ret > 0)
+			nr_pages = ret;
+		else
+			nr_pages = 0;
+		ret = -EAGAIN; /* XXX ? */
+		goto out;
+	}
+
+	rdsdebug("len %d nr_pages %lu\n", len, nr_pages);
+
+call_func:
+	func = rds_info_funcs[optname - RDS_INFO_FIRST];
+	if (!func) {
+		ret = -ENOPROTOOPT;
+		goto out;
+	}
+
+	iter.pages = pages;
+	iter.addr = NULL;
+	iter.offset = start & (PAGE_SIZE - 1);
+
+	func(sock, len, &iter, &lens);
+	BUG_ON(lens.each == 0);
+
+	total = lens.nr * lens.each;
+
+	rds_info_iter_unmap(&iter);
+
+	if (total > len) {
+		len = total;
+		ret = -ENOSPC;
+	} else {
+		len = total;
+		ret = lens.each;
+	}
+
+	if (put_user(len, optlen))
+		ret = -EFAULT;
+
+out:
+	for (i = 0; pages && i < nr_pages; i++)
+		put_page(pages[i]);
+	kfree(pages);
+
+	return ret;
+}
diff --git a/net/rds/info.h b/net/rds/info.h
new file mode 100644
index 000000000..b6c052ca7
--- /dev/null
+++ b/net/rds/info.h
@@ -0,0 +1,30 @@
+#ifndef _RDS_INFO_H
+#define _RDS_INFO_H
+
+struct rds_info_lengths {
+	unsigned int	nr;
+	unsigned int	each;
+};
+
+struct rds_info_iterator;
+
+/*
+ * These functions must fill in the fields of @lens to reflect the size
+ * of the available info source.  If the snapshot fits in @len then it
+ * should be copied using @iter.  The caller will deduce if it was copied
+ * or not by comparing the lengths.
+ */
+typedef void (*rds_info_func)(struct socket *sock, unsigned int len,
+			      struct rds_info_iterator *iter,
+			      struct rds_info_lengths *lens);
+
+void rds_info_register_func(int optname, rds_info_func func);
+void rds_info_deregister_func(int optname, rds_info_func func);
+int rds_info_getsockopt(struct socket *sock, int optname, char __user *optval,
+			int __user *optlen);
+void rds_info_copy(struct rds_info_iterator *iter, void *data,
+		   unsigned long bytes);
+void rds_info_iter_unmap(struct rds_info_iterator *iter);
+
+
+#endif
diff --git a/net/rds/iw.c b/net/rds/iw.c
new file mode 100644
index 000000000..589935661
--- /dev/null
+++ b/net/rds/iw.c
@@ -0,0 +1,329 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/in.h>
+#include <linux/if.h>
+#include <linux/netdevice.h>
+#include <linux/inetdevice.h>
+#include <linux/if_arp.h>
+#include <linux/delay.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+
+#include "rds.h"
+#include "iw.h"
+
+unsigned int fastreg_pool_size = RDS_FASTREG_POOL_SIZE;
+unsigned int fastreg_message_size = RDS_FASTREG_SIZE + 1; /* +1 allows for unaligned MRs */
+
+module_param(fastreg_pool_size, int, 0444);
+MODULE_PARM_DESC(fastreg_pool_size, " Max number of fastreg MRs per device");
+module_param(fastreg_message_size, int, 0444);
+MODULE_PARM_DESC(fastreg_message_size, " Max size of a RDMA transfer (fastreg MRs)");
+
+struct list_head rds_iw_devices;
+
+/* NOTE: if also grabbing iwdev lock, grab this first */
+DEFINE_SPINLOCK(iw_nodev_conns_lock);
+LIST_HEAD(iw_nodev_conns);
+
+static void rds_iw_add_one(struct ib_device *device)
+{
+	struct rds_iw_device *rds_iwdev;
+	struct ib_device_attr *dev_attr;
+
+	/* Only handle iwarp devices */
+	if (device->node_type != RDMA_NODE_RNIC)
+		return;
+
+	dev_attr = kmalloc(sizeof *dev_attr, GFP_KERNEL);
+	if (!dev_attr)
+		return;
+
+	if (ib_query_device(device, dev_attr)) {
+		rdsdebug("Query device failed for %s\n", device->name);
+		goto free_attr;
+	}
+
+	rds_iwdev = kmalloc(sizeof *rds_iwdev, GFP_KERNEL);
+	if (!rds_iwdev)
+		goto free_attr;
+
+	spin_lock_init(&rds_iwdev->spinlock);
+
+	rds_iwdev->dma_local_lkey = !!(dev_attr->device_cap_flags & IB_DEVICE_LOCAL_DMA_LKEY);
+	rds_iwdev->max_wrs = dev_attr->max_qp_wr;
+	rds_iwdev->max_sge = min(dev_attr->max_sge, RDS_IW_MAX_SGE);
+
+	rds_iwdev->dev = device;
+	rds_iwdev->pd = ib_alloc_pd(device);
+	if (IS_ERR(rds_iwdev->pd))
+		goto free_dev;
+
+	if (!rds_iwdev->dma_local_lkey) {
+		rds_iwdev->mr = ib_get_dma_mr(rds_iwdev->pd,
+					IB_ACCESS_REMOTE_READ |
+					IB_ACCESS_REMOTE_WRITE |
+					IB_ACCESS_LOCAL_WRITE);
+		if (IS_ERR(rds_iwdev->mr))
+			goto err_pd;
+	} else
+		rds_iwdev->mr = NULL;
+
+	rds_iwdev->mr_pool = rds_iw_create_mr_pool(rds_iwdev);
+	if (IS_ERR(rds_iwdev->mr_pool)) {
+		rds_iwdev->mr_pool = NULL;
+		goto err_mr;
+	}
+
+	INIT_LIST_HEAD(&rds_iwdev->cm_id_list);
+	INIT_LIST_HEAD(&rds_iwdev->conn_list);
+	list_add_tail(&rds_iwdev->list, &rds_iw_devices);
+
+	ib_set_client_data(device, &rds_iw_client, rds_iwdev);
+
+	goto free_attr;
+
+err_mr:
+	if (rds_iwdev->mr)
+		ib_dereg_mr(rds_iwdev->mr);
+err_pd:
+	ib_dealloc_pd(rds_iwdev->pd);
+free_dev:
+	kfree(rds_iwdev);
+free_attr:
+	kfree(dev_attr);
+}
+
+static void rds_iw_remove_one(struct ib_device *device)
+{
+	struct rds_iw_device *rds_iwdev;
+	struct rds_iw_cm_id *i_cm_id, *next;
+
+	rds_iwdev = ib_get_client_data(device, &rds_iw_client);
+	if (!rds_iwdev)
+		return;
+
+	spin_lock_irq(&rds_iwdev->spinlock);
+	list_for_each_entry_safe(i_cm_id, next, &rds_iwdev->cm_id_list, list) {
+		list_del(&i_cm_id->list);
+		kfree(i_cm_id);
+	}
+	spin_unlock_irq(&rds_iwdev->spinlock);
+
+	rds_iw_destroy_conns(rds_iwdev);
+
+	if (rds_iwdev->mr_pool)
+		rds_iw_destroy_mr_pool(rds_iwdev->mr_pool);
+
+	if (rds_iwdev->mr)
+		ib_dereg_mr(rds_iwdev->mr);
+
+	while (ib_dealloc_pd(rds_iwdev->pd)) {
+		rdsdebug("Failed to dealloc pd %p\n", rds_iwdev->pd);
+		msleep(1);
+	}
+
+	list_del(&rds_iwdev->list);
+	kfree(rds_iwdev);
+}
+
+struct ib_client rds_iw_client = {
+	.name   = "rds_iw",
+	.add    = rds_iw_add_one,
+	.remove = rds_iw_remove_one
+};
+
+static int rds_iw_conn_info_visitor(struct rds_connection *conn,
+				    void *buffer)
+{
+	struct rds_info_rdma_connection *iinfo = buffer;
+	struct rds_iw_connection *ic;
+
+	/* We will only ever look at IB transports */
+	if (conn->c_trans != &rds_iw_transport)
+		return 0;
+
+	iinfo->src_addr = conn->c_laddr;
+	iinfo->dst_addr = conn->c_faddr;
+
+	memset(&iinfo->src_gid, 0, sizeof(iinfo->src_gid));
+	memset(&iinfo->dst_gid, 0, sizeof(iinfo->dst_gid));
+	if (rds_conn_state(conn) == RDS_CONN_UP) {
+		struct rds_iw_device *rds_iwdev;
+		struct rdma_dev_addr *dev_addr;
+
+		ic = conn->c_transport_data;
+		dev_addr = &ic->i_cm_id->route.addr.dev_addr;
+
+		rdma_addr_get_sgid(dev_addr, (union ib_gid *) &iinfo->src_gid);
+		rdma_addr_get_dgid(dev_addr, (union ib_gid *) &iinfo->dst_gid);
+
+		rds_iwdev = ib_get_client_data(ic->i_cm_id->device, &rds_iw_client);
+		iinfo->max_send_wr = ic->i_send_ring.w_nr;
+		iinfo->max_recv_wr = ic->i_recv_ring.w_nr;
+		iinfo->max_send_sge = rds_iwdev->max_sge;
+		rds_iw_get_mr_info(rds_iwdev, iinfo);
+	}
+	return 1;
+}
+
+static void rds_iw_ic_info(struct socket *sock, unsigned int len,
+			   struct rds_info_iterator *iter,
+			   struct rds_info_lengths *lens)
+{
+	rds_for_each_conn_info(sock, len, iter, lens,
+				rds_iw_conn_info_visitor,
+				sizeof(struct rds_info_rdma_connection));
+}
+
+
+/*
+ * Early RDS/IB was built to only bind to an address if there is an IPoIB
+ * device with that address set.
+ *
+ * If it were me, I'd advocate for something more flexible.  Sending and
+ * receiving should be device-agnostic.  Transports would try and maintain
+ * connections between peers who have messages queued.  Userspace would be
+ * allowed to influence which paths have priority.  We could call userspace
+ * asserting this policy "routing".
+ */
+static int rds_iw_laddr_check(__be32 addr)
+{
+	int ret;
+	struct rdma_cm_id *cm_id;
+	struct sockaddr_in sin;
+
+	/* Create a CMA ID and try to bind it. This catches both
+	 * IB and iWARP capable NICs.
+	 */
+	cm_id = rdma_create_id(NULL, NULL, RDMA_PS_TCP, IB_QPT_RC);
+	if (IS_ERR(cm_id))
+		return PTR_ERR(cm_id);
+
+	memset(&sin, 0, sizeof(sin));
+	sin.sin_family = AF_INET;
+	sin.sin_addr.s_addr = addr;
+
+	/* rdma_bind_addr will only succeed for IB & iWARP devices */
+	ret = rdma_bind_addr(cm_id, (struct sockaddr *)&sin);
+	/* due to this, we will claim to support IB devices unless we
+	   check node_type. */
+	if (ret || !cm_id->device ||
+	    cm_id->device->node_type != RDMA_NODE_RNIC)
+		ret = -EADDRNOTAVAIL;
+
+	rdsdebug("addr %pI4 ret %d node type %d\n",
+		&addr, ret,
+		cm_id->device ? cm_id->device->node_type : -1);
+
+	rdma_destroy_id(cm_id);
+
+	return ret;
+}
+
+void rds_iw_exit(void)
+{
+	rds_info_deregister_func(RDS_INFO_IWARP_CONNECTIONS, rds_iw_ic_info);
+	rds_iw_destroy_nodev_conns();
+	ib_unregister_client(&rds_iw_client);
+	rds_iw_sysctl_exit();
+	rds_iw_recv_exit();
+	rds_trans_unregister(&rds_iw_transport);
+}
+
+struct rds_transport rds_iw_transport = {
+	.laddr_check		= rds_iw_laddr_check,
+	.xmit_complete		= rds_iw_xmit_complete,
+	.xmit			= rds_iw_xmit,
+	.xmit_rdma		= rds_iw_xmit_rdma,
+	.recv			= rds_iw_recv,
+	.conn_alloc		= rds_iw_conn_alloc,
+	.conn_free		= rds_iw_conn_free,
+	.conn_connect		= rds_iw_conn_connect,
+	.conn_shutdown		= rds_iw_conn_shutdown,
+	.inc_copy_to_user	= rds_iw_inc_copy_to_user,
+	.inc_free		= rds_iw_inc_free,
+	.cm_initiate_connect	= rds_iw_cm_initiate_connect,
+	.cm_handle_connect	= rds_iw_cm_handle_connect,
+	.cm_connect_complete	= rds_iw_cm_connect_complete,
+	.stats_info_copy	= rds_iw_stats_info_copy,
+	.exit			= rds_iw_exit,
+	.get_mr			= rds_iw_get_mr,
+	.sync_mr		= rds_iw_sync_mr,
+	.free_mr		= rds_iw_free_mr,
+	.flush_mrs		= rds_iw_flush_mrs,
+	.t_owner		= THIS_MODULE,
+	.t_name			= "iwarp",
+	.t_type			= RDS_TRANS_IWARP,
+	.t_prefer_loopback	= 1,
+};
+
+int rds_iw_init(void)
+{
+	int ret;
+
+	INIT_LIST_HEAD(&rds_iw_devices);
+
+	ret = ib_register_client(&rds_iw_client);
+	if (ret)
+		goto out;
+
+	ret = rds_iw_sysctl_init();
+	if (ret)
+		goto out_ibreg;
+
+	ret = rds_iw_recv_init();
+	if (ret)
+		goto out_sysctl;
+
+	ret = rds_trans_register(&rds_iw_transport);
+	if (ret)
+		goto out_recv;
+
+	rds_info_register_func(RDS_INFO_IWARP_CONNECTIONS, rds_iw_ic_info);
+
+	goto out;
+
+out_recv:
+	rds_iw_recv_exit();
+out_sysctl:
+	rds_iw_sysctl_exit();
+out_ibreg:
+	ib_unregister_client(&rds_iw_client);
+out:
+	return ret;
+}
+
+MODULE_LICENSE("GPL");
+
diff --git a/net/rds/iw.h b/net/rds/iw.h
new file mode 100644
index 000000000..cbe6674e3
--- /dev/null
+++ b/net/rds/iw.h
@@ -0,0 +1,395 @@
+#ifndef _RDS_IW_H
+#define _RDS_IW_H
+
+#include <linux/interrupt.h>
+#include <rdma/ib_verbs.h>
+#include <rdma/rdma_cm.h>
+#include "rds.h"
+#include "rdma_transport.h"
+
+#define RDS_FASTREG_SIZE		20
+#define RDS_FASTREG_POOL_SIZE		2048
+
+#define RDS_IW_MAX_SGE			8
+#define RDS_IW_RECV_SGE 		2
+
+#define RDS_IW_DEFAULT_RECV_WR		1024
+#define RDS_IW_DEFAULT_SEND_WR		256
+
+#define RDS_IW_SUPPORTED_PROTOCOLS	0x00000003	/* minor versions supported */
+
+extern struct list_head rds_iw_devices;
+
+/*
+ * IB posts RDS_FRAG_SIZE fragments of pages to the receive queues to
+ * try and minimize the amount of memory tied up both the device and
+ * socket receive queues.
+ */
+/* page offset of the final full frag that fits in the page */
+#define RDS_PAGE_LAST_OFF (((PAGE_SIZE  / RDS_FRAG_SIZE) - 1) * RDS_FRAG_SIZE)
+struct rds_page_frag {
+	struct list_head	f_item;
+	struct page		*f_page;
+	unsigned long		f_offset;
+	dma_addr_t 		f_mapped;
+};
+
+struct rds_iw_incoming {
+	struct list_head	ii_frags;
+	struct rds_incoming	ii_inc;
+};
+
+struct rds_iw_connect_private {
+	/* Add new fields at the end, and don't permute existing fields. */
+	__be32			dp_saddr;
+	__be32			dp_daddr;
+	u8			dp_protocol_major;
+	u8			dp_protocol_minor;
+	__be16			dp_protocol_minor_mask; /* bitmask */
+	__be32			dp_reserved1;
+	__be64			dp_ack_seq;
+	__be32			dp_credit;		/* non-zero enables flow ctl */
+};
+
+struct rds_iw_scatterlist {
+	struct scatterlist	*list;
+	unsigned int		len;
+	int			dma_len;
+	unsigned int		dma_npages;
+	unsigned int		bytes;
+};
+
+struct rds_iw_mapping {
+	spinlock_t		m_lock;	/* protect the mapping struct */
+	struct list_head	m_list;
+	struct rds_iw_mr	*m_mr;
+	uint32_t		m_rkey;
+	struct rds_iw_scatterlist m_sg;
+};
+
+struct rds_iw_send_work {
+	struct rds_message	*s_rm;
+
+	/* We should really put these into a union: */
+	struct rm_rdma_op	*s_op;
+	struct rds_iw_mapping	*s_mapping;
+	struct ib_mr		*s_mr;
+	struct ib_fast_reg_page_list *s_page_list;
+	unsigned char		s_remap_count;
+
+	struct ib_send_wr	s_wr;
+	struct ib_sge		s_sge[RDS_IW_MAX_SGE];
+	unsigned long		s_queued;
+};
+
+struct rds_iw_recv_work {
+	struct rds_iw_incoming 	*r_iwinc;
+	struct rds_page_frag	*r_frag;
+	struct ib_recv_wr	r_wr;
+	struct ib_sge		r_sge[2];
+};
+
+struct rds_iw_work_ring {
+	u32		w_nr;
+	u32		w_alloc_ptr;
+	u32		w_alloc_ctr;
+	u32		w_free_ptr;
+	atomic_t	w_free_ctr;
+};
+
+struct rds_iw_device;
+
+struct rds_iw_connection {
+
+	struct list_head	iw_node;
+	struct rds_iw_device 	*rds_iwdev;
+	struct rds_connection	*conn;
+
+	/* alphabet soup, IBTA style */
+	struct rdma_cm_id	*i_cm_id;
+	struct ib_pd		*i_pd;
+	struct ib_mr		*i_mr;
+	struct ib_cq		*i_send_cq;
+	struct ib_cq		*i_recv_cq;
+
+	/* tx */
+	struct rds_iw_work_ring	i_send_ring;
+	struct rds_message	*i_rm;
+	struct rds_header	*i_send_hdrs;
+	u64			i_send_hdrs_dma;
+	struct rds_iw_send_work *i_sends;
+
+	/* rx */
+	struct tasklet_struct	i_recv_tasklet;
+	struct mutex		i_recv_mutex;
+	struct rds_iw_work_ring	i_recv_ring;
+	struct rds_iw_incoming	*i_iwinc;
+	u32			i_recv_data_rem;
+	struct rds_header	*i_recv_hdrs;
+	u64			i_recv_hdrs_dma;
+	struct rds_iw_recv_work *i_recvs;
+	struct rds_page_frag	i_frag;
+	u64			i_ack_recv;	/* last ACK received */
+
+	/* sending acks */
+	unsigned long		i_ack_flags;
+#ifdef KERNEL_HAS_ATOMIC64
+	atomic64_t		i_ack_next;	/* next ACK to send */
+#else
+	spinlock_t		i_ack_lock;	/* protect i_ack_next */
+	u64			i_ack_next;	/* next ACK to send */
+#endif
+	struct rds_header	*i_ack;
+	struct ib_send_wr	i_ack_wr;
+	struct ib_sge		i_ack_sge;
+	u64			i_ack_dma;
+	unsigned long		i_ack_queued;
+
+	/* Flow control related information
+	 *
+	 * Our algorithm uses a pair variables that we need to access
+	 * atomically - one for the send credits, and one posted
+	 * recv credits we need to transfer to remote.
+	 * Rather than protect them using a slow spinlock, we put both into
+	 * a single atomic_t and update it using cmpxchg
+	 */
+	atomic_t		i_credits;
+
+	/* Protocol version specific information */
+	unsigned int		i_flowctl:1;	/* enable/disable flow ctl */
+	unsigned int		i_dma_local_lkey:1;
+	unsigned int		i_fastreg_posted:1; /* fastreg posted on this connection */
+	/* Batched completions */
+	unsigned int		i_unsignaled_wrs;
+	long			i_unsignaled_bytes;
+};
+
+/* This assumes that atomic_t is at least 32 bits */
+#define IB_GET_SEND_CREDITS(v)	((v) & 0xffff)
+#define IB_GET_POST_CREDITS(v)	((v) >> 16)
+#define IB_SET_SEND_CREDITS(v)	((v) & 0xffff)
+#define IB_SET_POST_CREDITS(v)	((v) << 16)
+
+struct rds_iw_cm_id {
+	struct list_head	list;
+	struct rdma_cm_id	*cm_id;
+};
+
+struct rds_iw_device {
+	struct list_head	list;
+	struct list_head	cm_id_list;
+	struct list_head	conn_list;
+	struct ib_device	*dev;
+	struct ib_pd		*pd;
+	struct ib_mr		*mr;
+	struct rds_iw_mr_pool	*mr_pool;
+	int			max_sge;
+	unsigned int		max_wrs;
+	unsigned int		dma_local_lkey:1;
+	spinlock_t		spinlock;	/* protect the above */
+};
+
+/* bits for i_ack_flags */
+#define IB_ACK_IN_FLIGHT	0
+#define IB_ACK_REQUESTED	1
+
+/* Magic WR_ID for ACKs */
+#define RDS_IW_ACK_WR_ID	((u64)0xffffffffffffffffULL)
+#define RDS_IW_FAST_REG_WR_ID	((u64)0xefefefefefefefefULL)
+#define RDS_IW_LOCAL_INV_WR_ID	((u64)0xdfdfdfdfdfdfdfdfULL)
+
+struct rds_iw_statistics {
+	uint64_t	s_iw_connect_raced;
+	uint64_t	s_iw_listen_closed_stale;
+	uint64_t	s_iw_tx_cq_call;
+	uint64_t	s_iw_tx_cq_event;
+	uint64_t	s_iw_tx_ring_full;
+	uint64_t	s_iw_tx_throttle;
+	uint64_t	s_iw_tx_sg_mapping_failure;
+	uint64_t	s_iw_tx_stalled;
+	uint64_t	s_iw_tx_credit_updates;
+	uint64_t	s_iw_rx_cq_call;
+	uint64_t	s_iw_rx_cq_event;
+	uint64_t	s_iw_rx_ring_empty;
+	uint64_t	s_iw_rx_refill_from_cq;
+	uint64_t	s_iw_rx_refill_from_thread;
+	uint64_t	s_iw_rx_alloc_limit;
+	uint64_t	s_iw_rx_credit_updates;
+	uint64_t	s_iw_ack_sent;
+	uint64_t	s_iw_ack_send_failure;
+	uint64_t	s_iw_ack_send_delayed;
+	uint64_t	s_iw_ack_send_piggybacked;
+	uint64_t	s_iw_ack_received;
+	uint64_t	s_iw_rdma_mr_alloc;
+	uint64_t	s_iw_rdma_mr_free;
+	uint64_t	s_iw_rdma_mr_used;
+	uint64_t	s_iw_rdma_mr_pool_flush;
+	uint64_t	s_iw_rdma_mr_pool_wait;
+	uint64_t	s_iw_rdma_mr_pool_depleted;
+};
+
+extern struct workqueue_struct *rds_iw_wq;
+
+/*
+ * Fake ib_dma_sync_sg_for_{cpu,device} as long as ib_verbs.h
+ * doesn't define it.
+ */
+static inline void rds_iw_dma_sync_sg_for_cpu(struct ib_device *dev,
+		struct scatterlist *sg, unsigned int sg_dma_len, int direction)
+{
+	unsigned int i;
+
+	for (i = 0; i < sg_dma_len; ++i) {
+		ib_dma_sync_single_for_cpu(dev,
+				ib_sg_dma_address(dev, &sg[i]),
+				ib_sg_dma_len(dev, &sg[i]),
+				direction);
+	}
+}
+#define ib_dma_sync_sg_for_cpu	rds_iw_dma_sync_sg_for_cpu
+
+static inline void rds_iw_dma_sync_sg_for_device(struct ib_device *dev,
+		struct scatterlist *sg, unsigned int sg_dma_len, int direction)
+{
+	unsigned int i;
+
+	for (i = 0; i < sg_dma_len; ++i) {
+		ib_dma_sync_single_for_device(dev,
+				ib_sg_dma_address(dev, &sg[i]),
+				ib_sg_dma_len(dev, &sg[i]),
+				direction);
+	}
+}
+#define ib_dma_sync_sg_for_device	rds_iw_dma_sync_sg_for_device
+
+static inline u32 rds_iw_local_dma_lkey(struct rds_iw_connection *ic)
+{
+	return ic->i_dma_local_lkey ? ic->i_cm_id->device->local_dma_lkey : ic->i_mr->lkey;
+}
+
+/* ib.c */
+extern struct rds_transport rds_iw_transport;
+extern struct ib_client rds_iw_client;
+
+extern unsigned int fastreg_pool_size;
+extern unsigned int fastreg_message_size;
+
+extern spinlock_t iw_nodev_conns_lock;
+extern struct list_head iw_nodev_conns;
+
+/* ib_cm.c */
+int rds_iw_conn_alloc(struct rds_connection *conn, gfp_t gfp);
+void rds_iw_conn_free(void *arg);
+int rds_iw_conn_connect(struct rds_connection *conn);
+void rds_iw_conn_shutdown(struct rds_connection *conn);
+void rds_iw_state_change(struct sock *sk);
+int rds_iw_listen_init(void);
+void rds_iw_listen_stop(void);
+void __rds_iw_conn_error(struct rds_connection *conn, const char *, ...);
+int rds_iw_cm_handle_connect(struct rdma_cm_id *cm_id,
+			     struct rdma_cm_event *event);
+int rds_iw_cm_initiate_connect(struct rdma_cm_id *cm_id);
+void rds_iw_cm_connect_complete(struct rds_connection *conn,
+				struct rdma_cm_event *event);
+
+
+#define rds_iw_conn_error(conn, fmt...) \
+	__rds_iw_conn_error(conn, KERN_WARNING "RDS/IW: " fmt)
+
+/* ib_rdma.c */
+int rds_iw_update_cm_id(struct rds_iw_device *rds_iwdev, struct rdma_cm_id *cm_id);
+void rds_iw_add_conn(struct rds_iw_device *rds_iwdev, struct rds_connection *conn);
+void rds_iw_remove_conn(struct rds_iw_device *rds_iwdev, struct rds_connection *conn);
+void __rds_iw_destroy_conns(struct list_head *list, spinlock_t *list_lock);
+static inline void rds_iw_destroy_nodev_conns(void)
+{
+	__rds_iw_destroy_conns(&iw_nodev_conns, &iw_nodev_conns_lock);
+}
+static inline void rds_iw_destroy_conns(struct rds_iw_device *rds_iwdev)
+{
+	__rds_iw_destroy_conns(&rds_iwdev->conn_list, &rds_iwdev->spinlock);
+}
+struct rds_iw_mr_pool *rds_iw_create_mr_pool(struct rds_iw_device *);
+void rds_iw_get_mr_info(struct rds_iw_device *rds_iwdev, struct rds_info_rdma_connection *iinfo);
+void rds_iw_destroy_mr_pool(struct rds_iw_mr_pool *);
+void *rds_iw_get_mr(struct scatterlist *sg, unsigned long nents,
+		    struct rds_sock *rs, u32 *key_ret);
+void rds_iw_sync_mr(void *trans_private, int dir);
+void rds_iw_free_mr(void *trans_private, int invalidate);
+void rds_iw_flush_mrs(void);
+
+/* ib_recv.c */
+int rds_iw_recv_init(void);
+void rds_iw_recv_exit(void);
+int rds_iw_recv(struct rds_connection *conn);
+int rds_iw_recv_refill(struct rds_connection *conn, gfp_t kptr_gfp,
+		       gfp_t page_gfp, int prefill);
+void rds_iw_inc_free(struct rds_incoming *inc);
+int rds_iw_inc_copy_to_user(struct rds_incoming *inc, struct iov_iter *to);
+void rds_iw_recv_cq_comp_handler(struct ib_cq *cq, void *context);
+void rds_iw_recv_tasklet_fn(unsigned long data);
+void rds_iw_recv_init_ring(struct rds_iw_connection *ic);
+void rds_iw_recv_clear_ring(struct rds_iw_connection *ic);
+void rds_iw_recv_init_ack(struct rds_iw_connection *ic);
+void rds_iw_attempt_ack(struct rds_iw_connection *ic);
+void rds_iw_ack_send_complete(struct rds_iw_connection *ic);
+u64 rds_iw_piggyb_ack(struct rds_iw_connection *ic);
+
+/* ib_ring.c */
+void rds_iw_ring_init(struct rds_iw_work_ring *ring, u32 nr);
+void rds_iw_ring_resize(struct rds_iw_work_ring *ring, u32 nr);
+u32 rds_iw_ring_alloc(struct rds_iw_work_ring *ring, u32 val, u32 *pos);
+void rds_iw_ring_free(struct rds_iw_work_ring *ring, u32 val);
+void rds_iw_ring_unalloc(struct rds_iw_work_ring *ring, u32 val);
+int rds_iw_ring_empty(struct rds_iw_work_ring *ring);
+int rds_iw_ring_low(struct rds_iw_work_ring *ring);
+u32 rds_iw_ring_oldest(struct rds_iw_work_ring *ring);
+u32 rds_iw_ring_completed(struct rds_iw_work_ring *ring, u32 wr_id, u32 oldest);
+extern wait_queue_head_t rds_iw_ring_empty_wait;
+
+/* ib_send.c */
+void rds_iw_xmit_complete(struct rds_connection *conn);
+int rds_iw_xmit(struct rds_connection *conn, struct rds_message *rm,
+		unsigned int hdr_off, unsigned int sg, unsigned int off);
+void rds_iw_send_cq_comp_handler(struct ib_cq *cq, void *context);
+void rds_iw_send_init_ring(struct rds_iw_connection *ic);
+void rds_iw_send_clear_ring(struct rds_iw_connection *ic);
+int rds_iw_xmit_rdma(struct rds_connection *conn, struct rm_rdma_op *op);
+void rds_iw_send_add_credits(struct rds_connection *conn, unsigned int credits);
+void rds_iw_advertise_credits(struct rds_connection *conn, unsigned int posted);
+int rds_iw_send_grab_credits(struct rds_iw_connection *ic, u32 wanted,
+			     u32 *adv_credits, int need_posted, int max_posted);
+
+/* ib_stats.c */
+DECLARE_PER_CPU(struct rds_iw_statistics, rds_iw_stats);
+#define rds_iw_stats_inc(member) rds_stats_inc_which(rds_iw_stats, member)
+unsigned int rds_iw_stats_info_copy(struct rds_info_iterator *iter,
+				    unsigned int avail);
+
+/* ib_sysctl.c */
+int rds_iw_sysctl_init(void);
+void rds_iw_sysctl_exit(void);
+extern unsigned long rds_iw_sysctl_max_send_wr;
+extern unsigned long rds_iw_sysctl_max_recv_wr;
+extern unsigned long rds_iw_sysctl_max_unsig_wrs;
+extern unsigned long rds_iw_sysctl_max_unsig_bytes;
+extern unsigned long rds_iw_sysctl_max_recv_allocation;
+extern unsigned int rds_iw_sysctl_flow_control;
+
+/*
+ * Helper functions for getting/setting the header and data SGEs in
+ * RDS packets (not RDMA)
+ */
+static inline struct ib_sge *
+rds_iw_header_sge(struct rds_iw_connection *ic, struct ib_sge *sge)
+{
+	return &sge[0];
+}
+
+static inline struct ib_sge *
+rds_iw_data_sge(struct rds_iw_connection *ic, struct ib_sge *sge)
+{
+	return &sge[1];
+}
+
+#endif
diff --git a/net/rds/iw_cm.c b/net/rds/iw_cm.c
new file mode 100644
index 000000000..a6c2bea9f
--- /dev/null
+++ b/net/rds/iw_cm.c
@@ -0,0 +1,765 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/in.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include <linux/ratelimit.h>
+
+#include "rds.h"
+#include "iw.h"
+
+/*
+ * Set the selected protocol version
+ */
+static void rds_iw_set_protocol(struct rds_connection *conn, unsigned int version)
+{
+	conn->c_version = version;
+}
+
+/*
+ * Set up flow control
+ */
+static void rds_iw_set_flow_control(struct rds_connection *conn, u32 credits)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+
+	if (rds_iw_sysctl_flow_control && credits != 0) {
+		/* We're doing flow control */
+		ic->i_flowctl = 1;
+		rds_iw_send_add_credits(conn, credits);
+	} else {
+		ic->i_flowctl = 0;
+	}
+}
+
+/*
+ * Connection established.
+ * We get here for both outgoing and incoming connection.
+ */
+void rds_iw_cm_connect_complete(struct rds_connection *conn, struct rdma_cm_event *event)
+{
+	const struct rds_iw_connect_private *dp = NULL;
+	struct rds_iw_connection *ic = conn->c_transport_data;
+	struct rds_iw_device *rds_iwdev;
+	int err;
+
+	if (event->param.conn.private_data_len) {
+		dp = event->param.conn.private_data;
+
+		rds_iw_set_protocol(conn,
+				RDS_PROTOCOL(dp->dp_protocol_major,
+					dp->dp_protocol_minor));
+		rds_iw_set_flow_control(conn, be32_to_cpu(dp->dp_credit));
+	}
+
+	/* update ib_device with this local ipaddr & conn */
+	rds_iwdev = ib_get_client_data(ic->i_cm_id->device, &rds_iw_client);
+	err = rds_iw_update_cm_id(rds_iwdev, ic->i_cm_id);
+	if (err)
+		printk(KERN_ERR "rds_iw_update_ipaddr failed (%d)\n", err);
+	rds_iw_add_conn(rds_iwdev, conn);
+
+	/* If the peer gave us the last packet it saw, process this as if
+	 * we had received a regular ACK. */
+	if (dp && dp->dp_ack_seq)
+		rds_send_drop_acked(conn, be64_to_cpu(dp->dp_ack_seq), NULL);
+
+	printk(KERN_NOTICE "RDS/IW: connected to %pI4<->%pI4 version %u.%u%s\n",
+			&conn->c_laddr, &conn->c_faddr,
+			RDS_PROTOCOL_MAJOR(conn->c_version),
+			RDS_PROTOCOL_MINOR(conn->c_version),
+			ic->i_flowctl ? ", flow control" : "");
+
+	rds_connect_complete(conn);
+}
+
+static void rds_iw_cm_fill_conn_param(struct rds_connection *conn,
+			struct rdma_conn_param *conn_param,
+			struct rds_iw_connect_private *dp,
+			u32 protocol_version)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+
+	memset(conn_param, 0, sizeof(struct rdma_conn_param));
+	/* XXX tune these? */
+	conn_param->responder_resources = 1;
+	conn_param->initiator_depth = 1;
+
+	if (dp) {
+		memset(dp, 0, sizeof(*dp));
+		dp->dp_saddr = conn->c_laddr;
+		dp->dp_daddr = conn->c_faddr;
+		dp->dp_protocol_major = RDS_PROTOCOL_MAJOR(protocol_version);
+		dp->dp_protocol_minor = RDS_PROTOCOL_MINOR(protocol_version);
+		dp->dp_protocol_minor_mask = cpu_to_be16(RDS_IW_SUPPORTED_PROTOCOLS);
+		dp->dp_ack_seq = rds_iw_piggyb_ack(ic);
+
+		/* Advertise flow control */
+		if (ic->i_flowctl) {
+			unsigned int credits;
+
+			credits = IB_GET_POST_CREDITS(atomic_read(&ic->i_credits));
+			dp->dp_credit = cpu_to_be32(credits);
+			atomic_sub(IB_SET_POST_CREDITS(credits), &ic->i_credits);
+		}
+
+		conn_param->private_data = dp;
+		conn_param->private_data_len = sizeof(*dp);
+	}
+}
+
+static void rds_iw_cq_event_handler(struct ib_event *event, void *data)
+{
+	rdsdebug("event %u data %p\n", event->event, data);
+}
+
+static void rds_iw_qp_event_handler(struct ib_event *event, void *data)
+{
+	struct rds_connection *conn = data;
+	struct rds_iw_connection *ic = conn->c_transport_data;
+
+	rdsdebug("conn %p ic %p event %u\n", conn, ic, event->event);
+
+	switch (event->event) {
+	case IB_EVENT_COMM_EST:
+		rdma_notify(ic->i_cm_id, IB_EVENT_COMM_EST);
+		break;
+	case IB_EVENT_QP_REQ_ERR:
+	case IB_EVENT_QP_FATAL:
+	default:
+		rdsdebug("Fatal QP Event %u "
+			"- connection %pI4->%pI4, reconnecting\n",
+			event->event, &conn->c_laddr,
+			&conn->c_faddr);
+		rds_conn_drop(conn);
+		break;
+	}
+}
+
+/*
+ * Create a QP
+ */
+static int rds_iw_init_qp_attrs(struct ib_qp_init_attr *attr,
+		struct rds_iw_device *rds_iwdev,
+		struct rds_iw_work_ring *send_ring,
+		void (*send_cq_handler)(struct ib_cq *, void *),
+		struct rds_iw_work_ring *recv_ring,
+		void (*recv_cq_handler)(struct ib_cq *, void *),
+		void *context)
+{
+	struct ib_device *dev = rds_iwdev->dev;
+	unsigned int send_size, recv_size;
+	int ret;
+
+	/* The offset of 1 is to accommodate the additional ACK WR. */
+	send_size = min_t(unsigned int, rds_iwdev->max_wrs, rds_iw_sysctl_max_send_wr + 1);
+	recv_size = min_t(unsigned int, rds_iwdev->max_wrs, rds_iw_sysctl_max_recv_wr + 1);
+	rds_iw_ring_resize(send_ring, send_size - 1);
+	rds_iw_ring_resize(recv_ring, recv_size - 1);
+
+	memset(attr, 0, sizeof(*attr));
+	attr->event_handler = rds_iw_qp_event_handler;
+	attr->qp_context = context;
+	attr->cap.max_send_wr = send_size;
+	attr->cap.max_recv_wr = recv_size;
+	attr->cap.max_send_sge = rds_iwdev->max_sge;
+	attr->cap.max_recv_sge = RDS_IW_RECV_SGE;
+	attr->sq_sig_type = IB_SIGNAL_REQ_WR;
+	attr->qp_type = IB_QPT_RC;
+
+	attr->send_cq = ib_create_cq(dev, send_cq_handler,
+				     rds_iw_cq_event_handler,
+				     context, send_size, 0);
+	if (IS_ERR(attr->send_cq)) {
+		ret = PTR_ERR(attr->send_cq);
+		attr->send_cq = NULL;
+		rdsdebug("ib_create_cq send failed: %d\n", ret);
+		goto out;
+	}
+
+	attr->recv_cq = ib_create_cq(dev, recv_cq_handler,
+				     rds_iw_cq_event_handler,
+				     context, recv_size, 0);
+	if (IS_ERR(attr->recv_cq)) {
+		ret = PTR_ERR(attr->recv_cq);
+		attr->recv_cq = NULL;
+		rdsdebug("ib_create_cq send failed: %d\n", ret);
+		goto out;
+	}
+
+	ret = ib_req_notify_cq(attr->send_cq, IB_CQ_NEXT_COMP);
+	if (ret) {
+		rdsdebug("ib_req_notify_cq send failed: %d\n", ret);
+		goto out;
+	}
+
+	ret = ib_req_notify_cq(attr->recv_cq, IB_CQ_SOLICITED);
+	if (ret) {
+		rdsdebug("ib_req_notify_cq recv failed: %d\n", ret);
+		goto out;
+	}
+
+out:
+	if (ret) {
+		if (attr->send_cq)
+			ib_destroy_cq(attr->send_cq);
+		if (attr->recv_cq)
+			ib_destroy_cq(attr->recv_cq);
+	}
+	return ret;
+}
+
+/*
+ * This needs to be very careful to not leave IS_ERR pointers around for
+ * cleanup to trip over.
+ */
+static int rds_iw_setup_qp(struct rds_connection *conn)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+	struct ib_device *dev = ic->i_cm_id->device;
+	struct ib_qp_init_attr attr;
+	struct rds_iw_device *rds_iwdev;
+	int ret;
+
+	/* rds_iw_add_one creates a rds_iw_device object per IB device,
+	 * and allocates a protection domain, memory range and MR pool
+	 * for each.  If that fails for any reason, it will not register
+	 * the rds_iwdev at all.
+	 */
+	rds_iwdev = ib_get_client_data(dev, &rds_iw_client);
+	if (!rds_iwdev) {
+		printk_ratelimited(KERN_NOTICE "RDS/IW: No client_data for device %s\n",
+					dev->name);
+		return -EOPNOTSUPP;
+	}
+
+	/* Protection domain and memory range */
+	ic->i_pd = rds_iwdev->pd;
+	ic->i_mr = rds_iwdev->mr;
+
+	ret = rds_iw_init_qp_attrs(&attr, rds_iwdev,
+			&ic->i_send_ring, rds_iw_send_cq_comp_handler,
+			&ic->i_recv_ring, rds_iw_recv_cq_comp_handler,
+			conn);
+	if (ret < 0)
+		goto out;
+
+	ic->i_send_cq = attr.send_cq;
+	ic->i_recv_cq = attr.recv_cq;
+
+	/*
+	 * XXX this can fail if max_*_wr is too large?  Are we supposed
+	 * to back off until we get a value that the hardware can support?
+	 */
+	ret = rdma_create_qp(ic->i_cm_id, ic->i_pd, &attr);
+	if (ret) {
+		rdsdebug("rdma_create_qp failed: %d\n", ret);
+		goto out;
+	}
+
+	ic->i_send_hdrs = ib_dma_alloc_coherent(dev,
+					   ic->i_send_ring.w_nr *
+						sizeof(struct rds_header),
+					   &ic->i_send_hdrs_dma, GFP_KERNEL);
+	if (!ic->i_send_hdrs) {
+		ret = -ENOMEM;
+		rdsdebug("ib_dma_alloc_coherent send failed\n");
+		goto out;
+	}
+
+	ic->i_recv_hdrs = ib_dma_alloc_coherent(dev,
+					   ic->i_recv_ring.w_nr *
+						sizeof(struct rds_header),
+					   &ic->i_recv_hdrs_dma, GFP_KERNEL);
+	if (!ic->i_recv_hdrs) {
+		ret = -ENOMEM;
+		rdsdebug("ib_dma_alloc_coherent recv failed\n");
+		goto out;
+	}
+
+	ic->i_ack = ib_dma_alloc_coherent(dev, sizeof(struct rds_header),
+				       &ic->i_ack_dma, GFP_KERNEL);
+	if (!ic->i_ack) {
+		ret = -ENOMEM;
+		rdsdebug("ib_dma_alloc_coherent ack failed\n");
+		goto out;
+	}
+
+	ic->i_sends = vmalloc(ic->i_send_ring.w_nr * sizeof(struct rds_iw_send_work));
+	if (!ic->i_sends) {
+		ret = -ENOMEM;
+		rdsdebug("send allocation failed\n");
+		goto out;
+	}
+	rds_iw_send_init_ring(ic);
+
+	ic->i_recvs = vmalloc(ic->i_recv_ring.w_nr * sizeof(struct rds_iw_recv_work));
+	if (!ic->i_recvs) {
+		ret = -ENOMEM;
+		rdsdebug("recv allocation failed\n");
+		goto out;
+	}
+
+	rds_iw_recv_init_ring(ic);
+	rds_iw_recv_init_ack(ic);
+
+	/* Post receive buffers - as a side effect, this will update
+	 * the posted credit count. */
+	rds_iw_recv_refill(conn, GFP_KERNEL, GFP_HIGHUSER, 1);
+
+	rdsdebug("conn %p pd %p mr %p cq %p %p\n", conn, ic->i_pd, ic->i_mr,
+		 ic->i_send_cq, ic->i_recv_cq);
+
+out:
+	return ret;
+}
+
+static u32 rds_iw_protocol_compatible(const struct rds_iw_connect_private *dp)
+{
+	u16 common;
+	u32 version = 0;
+
+	/* rdma_cm private data is odd - when there is any private data in the
+	 * request, we will be given a pretty large buffer without telling us the
+	 * original size. The only way to tell the difference is by looking at
+	 * the contents, which are initialized to zero.
+	 * If the protocol version fields aren't set, this is a connection attempt
+	 * from an older version. This could could be 3.0 or 2.0 - we can't tell.
+	 * We really should have changed this for OFED 1.3 :-( */
+	if (dp->dp_protocol_major == 0)
+		return RDS_PROTOCOL_3_0;
+
+	common = be16_to_cpu(dp->dp_protocol_minor_mask) & RDS_IW_SUPPORTED_PROTOCOLS;
+	if (dp->dp_protocol_major == 3 && common) {
+		version = RDS_PROTOCOL_3_0;
+		while ((common >>= 1) != 0)
+			version++;
+	}
+	printk_ratelimited(KERN_NOTICE "RDS: Connection from %pI4 using "
+			"incompatible protocol version %u.%u\n",
+			&dp->dp_saddr,
+			dp->dp_protocol_major,
+			dp->dp_protocol_minor);
+	return version;
+}
+
+int rds_iw_cm_handle_connect(struct rdma_cm_id *cm_id,
+				    struct rdma_cm_event *event)
+{
+	const struct rds_iw_connect_private *dp = event->param.conn.private_data;
+	struct rds_iw_connect_private dp_rep;
+	struct rds_connection *conn = NULL;
+	struct rds_iw_connection *ic = NULL;
+	struct rdma_conn_param conn_param;
+	struct rds_iw_device *rds_iwdev;
+	u32 version;
+	int err, destroy = 1;
+
+	/* Check whether the remote protocol version matches ours. */
+	version = rds_iw_protocol_compatible(dp);
+	if (!version)
+		goto out;
+
+	rdsdebug("saddr %pI4 daddr %pI4 RDSv%u.%u\n",
+		 &dp->dp_saddr, &dp->dp_daddr,
+		 RDS_PROTOCOL_MAJOR(version), RDS_PROTOCOL_MINOR(version));
+
+	conn = rds_conn_create(dp->dp_daddr, dp->dp_saddr, &rds_iw_transport,
+			       GFP_KERNEL);
+	if (IS_ERR(conn)) {
+		rdsdebug("rds_conn_create failed (%ld)\n", PTR_ERR(conn));
+		conn = NULL;
+		goto out;
+	}
+
+	/*
+	 * The connection request may occur while the
+	 * previous connection exist, e.g. in case of failover.
+	 * But as connections may be initiated simultaneously
+	 * by both hosts, we have a random backoff mechanism -
+	 * see the comment above rds_queue_reconnect()
+	 */
+	mutex_lock(&conn->c_cm_lock);
+	if (!rds_conn_transition(conn, RDS_CONN_DOWN, RDS_CONN_CONNECTING)) {
+		if (rds_conn_state(conn) == RDS_CONN_UP) {
+			rdsdebug("incoming connect while connecting\n");
+			rds_conn_drop(conn);
+			rds_iw_stats_inc(s_iw_listen_closed_stale);
+		} else
+		if (rds_conn_state(conn) == RDS_CONN_CONNECTING) {
+			/* Wait and see - our connect may still be succeeding */
+			rds_iw_stats_inc(s_iw_connect_raced);
+		}
+		mutex_unlock(&conn->c_cm_lock);
+		goto out;
+	}
+
+	ic = conn->c_transport_data;
+
+	rds_iw_set_protocol(conn, version);
+	rds_iw_set_flow_control(conn, be32_to_cpu(dp->dp_credit));
+
+	/* If the peer gave us the last packet it saw, process this as if
+	 * we had received a regular ACK. */
+	if (dp->dp_ack_seq)
+		rds_send_drop_acked(conn, be64_to_cpu(dp->dp_ack_seq), NULL);
+
+	BUG_ON(cm_id->context);
+	BUG_ON(ic->i_cm_id);
+
+	ic->i_cm_id = cm_id;
+	cm_id->context = conn;
+
+	rds_iwdev = ib_get_client_data(cm_id->device, &rds_iw_client);
+	ic->i_dma_local_lkey = rds_iwdev->dma_local_lkey;
+
+	/* We got halfway through setting up the ib_connection, if we
+	 * fail now, we have to take the long route out of this mess. */
+	destroy = 0;
+
+	err = rds_iw_setup_qp(conn);
+	if (err) {
+		rds_iw_conn_error(conn, "rds_iw_setup_qp failed (%d)\n", err);
+		mutex_unlock(&conn->c_cm_lock);
+		goto out;
+	}
+
+	rds_iw_cm_fill_conn_param(conn, &conn_param, &dp_rep, version);
+
+	/* rdma_accept() calls rdma_reject() internally if it fails */
+	err = rdma_accept(cm_id, &conn_param);
+	mutex_unlock(&conn->c_cm_lock);
+	if (err) {
+		rds_iw_conn_error(conn, "rdma_accept failed (%d)\n", err);
+		goto out;
+	}
+
+	return 0;
+
+out:
+	rdma_reject(cm_id, NULL, 0);
+	return destroy;
+}
+
+
+int rds_iw_cm_initiate_connect(struct rdma_cm_id *cm_id)
+{
+	struct rds_connection *conn = cm_id->context;
+	struct rds_iw_connection *ic = conn->c_transport_data;
+	struct rdma_conn_param conn_param;
+	struct rds_iw_connect_private dp;
+	int ret;
+
+	/* If the peer doesn't do protocol negotiation, we must
+	 * default to RDSv3.0 */
+	rds_iw_set_protocol(conn, RDS_PROTOCOL_3_0);
+	ic->i_flowctl = rds_iw_sysctl_flow_control;	/* advertise flow control */
+
+	ret = rds_iw_setup_qp(conn);
+	if (ret) {
+		rds_iw_conn_error(conn, "rds_iw_setup_qp failed (%d)\n", ret);
+		goto out;
+	}
+
+	rds_iw_cm_fill_conn_param(conn, &conn_param, &dp, RDS_PROTOCOL_VERSION);
+
+	ret = rdma_connect(cm_id, &conn_param);
+	if (ret)
+		rds_iw_conn_error(conn, "rdma_connect failed (%d)\n", ret);
+
+out:
+	/* Beware - returning non-zero tells the rdma_cm to destroy
+	 * the cm_id. We should certainly not do it as long as we still
+	 * "own" the cm_id. */
+	if (ret) {
+		struct rds_iw_connection *ic = conn->c_transport_data;
+
+		if (ic->i_cm_id == cm_id)
+			ret = 0;
+	}
+	return ret;
+}
+
+int rds_iw_conn_connect(struct rds_connection *conn)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+	struct rds_iw_device *rds_iwdev;
+	struct sockaddr_in src, dest;
+	int ret;
+
+	/* XXX I wonder what affect the port space has */
+	/* delegate cm event handler to rdma_transport */
+	ic->i_cm_id = rdma_create_id(rds_rdma_cm_event_handler, conn,
+				     RDMA_PS_TCP, IB_QPT_RC);
+	if (IS_ERR(ic->i_cm_id)) {
+		ret = PTR_ERR(ic->i_cm_id);
+		ic->i_cm_id = NULL;
+		rdsdebug("rdma_create_id() failed: %d\n", ret);
+		goto out;
+	}
+
+	rdsdebug("created cm id %p for conn %p\n", ic->i_cm_id, conn);
+
+	src.sin_family = AF_INET;
+	src.sin_addr.s_addr = (__force u32)conn->c_laddr;
+	src.sin_port = (__force u16)htons(0);
+
+	/* First, bind to the local address and device. */
+	ret = rdma_bind_addr(ic->i_cm_id, (struct sockaddr *) &src);
+	if (ret) {
+		rdsdebug("rdma_bind_addr(%pI4) failed: %d\n",
+				&conn->c_laddr, ret);
+		rdma_destroy_id(ic->i_cm_id);
+		ic->i_cm_id = NULL;
+		goto out;
+	}
+
+	rds_iwdev = ib_get_client_data(ic->i_cm_id->device, &rds_iw_client);
+	ic->i_dma_local_lkey = rds_iwdev->dma_local_lkey;
+
+	dest.sin_family = AF_INET;
+	dest.sin_addr.s_addr = (__force u32)conn->c_faddr;
+	dest.sin_port = (__force u16)htons(RDS_PORT);
+
+	ret = rdma_resolve_addr(ic->i_cm_id, (struct sockaddr *)&src,
+				(struct sockaddr *)&dest,
+				RDS_RDMA_RESOLVE_TIMEOUT_MS);
+	if (ret) {
+		rdsdebug("addr resolve failed for cm id %p: %d\n", ic->i_cm_id,
+			 ret);
+		rdma_destroy_id(ic->i_cm_id);
+		ic->i_cm_id = NULL;
+	}
+
+out:
+	return ret;
+}
+
+/*
+ * This is so careful about only cleaning up resources that were built up
+ * so that it can be called at any point during startup.  In fact it
+ * can be called multiple times for a given connection.
+ */
+void rds_iw_conn_shutdown(struct rds_connection *conn)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+	int err = 0;
+	struct ib_qp_attr qp_attr;
+
+	rdsdebug("cm %p pd %p cq %p %p qp %p\n", ic->i_cm_id,
+		 ic->i_pd, ic->i_send_cq, ic->i_recv_cq,
+		 ic->i_cm_id ? ic->i_cm_id->qp : NULL);
+
+	if (ic->i_cm_id) {
+		struct ib_device *dev = ic->i_cm_id->device;
+
+		rdsdebug("disconnecting cm %p\n", ic->i_cm_id);
+		err = rdma_disconnect(ic->i_cm_id);
+		if (err) {
+			/* Actually this may happen quite frequently, when
+			 * an outgoing connect raced with an incoming connect.
+			 */
+			rdsdebug("failed to disconnect, cm: %p err %d\n",
+				 ic->i_cm_id, err);
+		}
+
+		if (ic->i_cm_id->qp) {
+			qp_attr.qp_state = IB_QPS_ERR;
+			ib_modify_qp(ic->i_cm_id->qp, &qp_attr, IB_QP_STATE);
+		}
+
+		wait_event(rds_iw_ring_empty_wait,
+			rds_iw_ring_empty(&ic->i_send_ring) &&
+			rds_iw_ring_empty(&ic->i_recv_ring));
+
+		if (ic->i_send_hdrs)
+			ib_dma_free_coherent(dev,
+					   ic->i_send_ring.w_nr *
+						sizeof(struct rds_header),
+					   ic->i_send_hdrs,
+					   ic->i_send_hdrs_dma);
+
+		if (ic->i_recv_hdrs)
+			ib_dma_free_coherent(dev,
+					   ic->i_recv_ring.w_nr *
+						sizeof(struct rds_header),
+					   ic->i_recv_hdrs,
+					   ic->i_recv_hdrs_dma);
+
+		if (ic->i_ack)
+			ib_dma_free_coherent(dev, sizeof(struct rds_header),
+					     ic->i_ack, ic->i_ack_dma);
+
+		if (ic->i_sends)
+			rds_iw_send_clear_ring(ic);
+		if (ic->i_recvs)
+			rds_iw_recv_clear_ring(ic);
+
+		if (ic->i_cm_id->qp)
+			rdma_destroy_qp(ic->i_cm_id);
+		if (ic->i_send_cq)
+			ib_destroy_cq(ic->i_send_cq);
+		if (ic->i_recv_cq)
+			ib_destroy_cq(ic->i_recv_cq);
+
+		/*
+		 * If associated with an rds_iw_device:
+		 * 	Move connection back to the nodev list.
+		 * 	Remove cm_id from the device cm_id list.
+		 */
+		if (ic->rds_iwdev)
+			rds_iw_remove_conn(ic->rds_iwdev, conn);
+
+		rdma_destroy_id(ic->i_cm_id);
+
+		ic->i_cm_id = NULL;
+		ic->i_pd = NULL;
+		ic->i_mr = NULL;
+		ic->i_send_cq = NULL;
+		ic->i_recv_cq = NULL;
+		ic->i_send_hdrs = NULL;
+		ic->i_recv_hdrs = NULL;
+		ic->i_ack = NULL;
+	}
+	BUG_ON(ic->rds_iwdev);
+
+	/* Clear pending transmit */
+	if (ic->i_rm) {
+		rds_message_put(ic->i_rm);
+		ic->i_rm = NULL;
+	}
+
+	/* Clear the ACK state */
+	clear_bit(IB_ACK_IN_FLIGHT, &ic->i_ack_flags);
+#ifdef KERNEL_HAS_ATOMIC64
+	atomic64_set(&ic->i_ack_next, 0);
+#else
+	ic->i_ack_next = 0;
+#endif
+	ic->i_ack_recv = 0;
+
+	/* Clear flow control state */
+	ic->i_flowctl = 0;
+	atomic_set(&ic->i_credits, 0);
+
+	rds_iw_ring_init(&ic->i_send_ring, rds_iw_sysctl_max_send_wr);
+	rds_iw_ring_init(&ic->i_recv_ring, rds_iw_sysctl_max_recv_wr);
+
+	if (ic->i_iwinc) {
+		rds_inc_put(&ic->i_iwinc->ii_inc);
+		ic->i_iwinc = NULL;
+	}
+
+	vfree(ic->i_sends);
+	ic->i_sends = NULL;
+	vfree(ic->i_recvs);
+	ic->i_recvs = NULL;
+	rdsdebug("shutdown complete\n");
+}
+
+int rds_iw_conn_alloc(struct rds_connection *conn, gfp_t gfp)
+{
+	struct rds_iw_connection *ic;
+	unsigned long flags;
+
+	/* XXX too lazy? */
+	ic = kzalloc(sizeof(struct rds_iw_connection), gfp);
+	if (!ic)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&ic->iw_node);
+	tasklet_init(&ic->i_recv_tasklet, rds_iw_recv_tasklet_fn,
+		     (unsigned long) ic);
+	mutex_init(&ic->i_recv_mutex);
+#ifndef KERNEL_HAS_ATOMIC64
+	spin_lock_init(&ic->i_ack_lock);
+#endif
+
+	/*
+	 * rds_iw_conn_shutdown() waits for these to be emptied so they
+	 * must be initialized before it can be called.
+	 */
+	rds_iw_ring_init(&ic->i_send_ring, rds_iw_sysctl_max_send_wr);
+	rds_iw_ring_init(&ic->i_recv_ring, rds_iw_sysctl_max_recv_wr);
+
+	ic->conn = conn;
+	conn->c_transport_data = ic;
+
+	spin_lock_irqsave(&iw_nodev_conns_lock, flags);
+	list_add_tail(&ic->iw_node, &iw_nodev_conns);
+	spin_unlock_irqrestore(&iw_nodev_conns_lock, flags);
+
+
+	rdsdebug("conn %p conn ic %p\n", conn, conn->c_transport_data);
+	return 0;
+}
+
+/*
+ * Free a connection. Connection must be shut down and not set for reconnect.
+ */
+void rds_iw_conn_free(void *arg)
+{
+	struct rds_iw_connection *ic = arg;
+	spinlock_t	*lock_ptr;
+
+	rdsdebug("ic %p\n", ic);
+
+	/*
+	 * Conn is either on a dev's list or on the nodev list.
+	 * A race with shutdown() or connect() would cause problems
+	 * (since rds_iwdev would change) but that should never happen.
+	 */
+	lock_ptr = ic->rds_iwdev ? &ic->rds_iwdev->spinlock : &iw_nodev_conns_lock;
+
+	spin_lock_irq(lock_ptr);
+	list_del(&ic->iw_node);
+	spin_unlock_irq(lock_ptr);
+
+	kfree(ic);
+}
+
+/*
+ * An error occurred on the connection
+ */
+void
+__rds_iw_conn_error(struct rds_connection *conn, const char *fmt, ...)
+{
+	va_list ap;
+
+	rds_conn_drop(conn);
+
+	va_start(ap, fmt);
+	vprintk(fmt, ap);
+	va_end(ap);
+}
diff --git a/net/rds/iw_rdma.c b/net/rds/iw_rdma.c
new file mode 100644
index 000000000..dba8d0864
--- /dev/null
+++ b/net/rds/iw_rdma.c
@@ -0,0 +1,875 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/ratelimit.h>
+
+#include "rds.h"
+#include "iw.h"
+
+
+/*
+ * This is stored as mr->r_trans_private.
+ */
+struct rds_iw_mr {
+	struct rds_iw_device	*device;
+	struct rds_iw_mr_pool	*pool;
+	struct rdma_cm_id	*cm_id;
+
+	struct ib_mr	*mr;
+	struct ib_fast_reg_page_list *page_list;
+
+	struct rds_iw_mapping	mapping;
+	unsigned char		remap_count;
+};
+
+/*
+ * Our own little MR pool
+ */
+struct rds_iw_mr_pool {
+	struct rds_iw_device	*device;		/* back ptr to the device that owns us */
+
+	struct mutex		flush_lock;		/* serialize fmr invalidate */
+	struct work_struct	flush_worker;		/* flush worker */
+
+	spinlock_t		list_lock;		/* protect variables below */
+	atomic_t		item_count;		/* total # of MRs */
+	atomic_t		dirty_count;		/* # dirty of MRs */
+	struct list_head	dirty_list;		/* dirty mappings */
+	struct list_head	clean_list;		/* unused & unamapped MRs */
+	atomic_t		free_pinned;		/* memory pinned by free MRs */
+	unsigned long		max_message_size;	/* in pages */
+	unsigned long		max_items;
+	unsigned long		max_items_soft;
+	unsigned long		max_free_pinned;
+	int			max_pages;
+};
+
+static int rds_iw_flush_mr_pool(struct rds_iw_mr_pool *pool, int free_all);
+static void rds_iw_mr_pool_flush_worker(struct work_struct *work);
+static int rds_iw_init_fastreg(struct rds_iw_mr_pool *pool, struct rds_iw_mr *ibmr);
+static int rds_iw_map_fastreg(struct rds_iw_mr_pool *pool,
+			  struct rds_iw_mr *ibmr,
+			  struct scatterlist *sg, unsigned int nents);
+static void rds_iw_free_fastreg(struct rds_iw_mr_pool *pool, struct rds_iw_mr *ibmr);
+static unsigned int rds_iw_unmap_fastreg_list(struct rds_iw_mr_pool *pool,
+			struct list_head *unmap_list,
+			struct list_head *kill_list,
+			int *unpinned);
+static void rds_iw_destroy_fastreg(struct rds_iw_mr_pool *pool, struct rds_iw_mr *ibmr);
+
+static int rds_iw_get_device(struct sockaddr_in *src, struct sockaddr_in *dst,
+			     struct rds_iw_device **rds_iwdev,
+			     struct rdma_cm_id **cm_id)
+{
+	struct rds_iw_device *iwdev;
+	struct rds_iw_cm_id *i_cm_id;
+
+	*rds_iwdev = NULL;
+	*cm_id = NULL;
+
+	list_for_each_entry(iwdev, &rds_iw_devices, list) {
+		spin_lock_irq(&iwdev->spinlock);
+		list_for_each_entry(i_cm_id, &iwdev->cm_id_list, list) {
+			struct sockaddr_in *src_addr, *dst_addr;
+
+			src_addr = (struct sockaddr_in *)&i_cm_id->cm_id->route.addr.src_addr;
+			dst_addr = (struct sockaddr_in *)&i_cm_id->cm_id->route.addr.dst_addr;
+
+			rdsdebug("local ipaddr = %x port %d, "
+				 "remote ipaddr = %x port %d"
+				 "..looking for %x port %d, "
+				 "remote ipaddr = %x port %d\n",
+				src_addr->sin_addr.s_addr,
+				src_addr->sin_port,
+				dst_addr->sin_addr.s_addr,
+				dst_addr->sin_port,
+				src->sin_addr.s_addr,
+				src->sin_port,
+				dst->sin_addr.s_addr,
+				dst->sin_port);
+#ifdef WORKING_TUPLE_DETECTION
+			if (src_addr->sin_addr.s_addr == src->sin_addr.s_addr &&
+			    src_addr->sin_port == src->sin_port &&
+			    dst_addr->sin_addr.s_addr == dst->sin_addr.s_addr &&
+			    dst_addr->sin_port == dst->sin_port) {
+#else
+			/* FIXME - needs to compare the local and remote
+			 * ipaddr/port tuple, but the ipaddr is the only
+			 * available information in the rds_sock (as the rest are
+			 * zero'ed.  It doesn't appear to be properly populated
+			 * during connection setup...
+			 */
+			if (src_addr->sin_addr.s_addr == src->sin_addr.s_addr) {
+#endif
+				spin_unlock_irq(&iwdev->spinlock);
+				*rds_iwdev = iwdev;
+				*cm_id = i_cm_id->cm_id;
+				return 0;
+			}
+		}
+		spin_unlock_irq(&iwdev->spinlock);
+	}
+
+	return 1;
+}
+
+static int rds_iw_add_cm_id(struct rds_iw_device *rds_iwdev, struct rdma_cm_id *cm_id)
+{
+	struct rds_iw_cm_id *i_cm_id;
+
+	i_cm_id = kmalloc(sizeof *i_cm_id, GFP_KERNEL);
+	if (!i_cm_id)
+		return -ENOMEM;
+
+	i_cm_id->cm_id = cm_id;
+
+	spin_lock_irq(&rds_iwdev->spinlock);
+	list_add_tail(&i_cm_id->list, &rds_iwdev->cm_id_list);
+	spin_unlock_irq(&rds_iwdev->spinlock);
+
+	return 0;
+}
+
+static void rds_iw_remove_cm_id(struct rds_iw_device *rds_iwdev,
+				struct rdma_cm_id *cm_id)
+{
+	struct rds_iw_cm_id *i_cm_id;
+
+	spin_lock_irq(&rds_iwdev->spinlock);
+	list_for_each_entry(i_cm_id, &rds_iwdev->cm_id_list, list) {
+		if (i_cm_id->cm_id == cm_id) {
+			list_del(&i_cm_id->list);
+			kfree(i_cm_id);
+			break;
+		}
+	}
+	spin_unlock_irq(&rds_iwdev->spinlock);
+}
+
+
+int rds_iw_update_cm_id(struct rds_iw_device *rds_iwdev, struct rdma_cm_id *cm_id)
+{
+	struct sockaddr_in *src_addr, *dst_addr;
+	struct rds_iw_device *rds_iwdev_old;
+	struct rdma_cm_id *pcm_id;
+	int rc;
+
+	src_addr = (struct sockaddr_in *)&cm_id->route.addr.src_addr;
+	dst_addr = (struct sockaddr_in *)&cm_id->route.addr.dst_addr;
+
+	rc = rds_iw_get_device(src_addr, dst_addr, &rds_iwdev_old, &pcm_id);
+	if (rc)
+		rds_iw_remove_cm_id(rds_iwdev, cm_id);
+
+	return rds_iw_add_cm_id(rds_iwdev, cm_id);
+}
+
+void rds_iw_add_conn(struct rds_iw_device *rds_iwdev, struct rds_connection *conn)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+
+	/* conn was previously on the nodev_conns_list */
+	spin_lock_irq(&iw_nodev_conns_lock);
+	BUG_ON(list_empty(&iw_nodev_conns));
+	BUG_ON(list_empty(&ic->iw_node));
+	list_del(&ic->iw_node);
+
+	spin_lock(&rds_iwdev->spinlock);
+	list_add_tail(&ic->iw_node, &rds_iwdev->conn_list);
+	spin_unlock(&rds_iwdev->spinlock);
+	spin_unlock_irq(&iw_nodev_conns_lock);
+
+	ic->rds_iwdev = rds_iwdev;
+}
+
+void rds_iw_remove_conn(struct rds_iw_device *rds_iwdev, struct rds_connection *conn)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+
+	/* place conn on nodev_conns_list */
+	spin_lock(&iw_nodev_conns_lock);
+
+	spin_lock_irq(&rds_iwdev->spinlock);
+	BUG_ON(list_empty(&ic->iw_node));
+	list_del(&ic->iw_node);
+	spin_unlock_irq(&rds_iwdev->spinlock);
+
+	list_add_tail(&ic->iw_node, &iw_nodev_conns);
+
+	spin_unlock(&iw_nodev_conns_lock);
+
+	rds_iw_remove_cm_id(ic->rds_iwdev, ic->i_cm_id);
+	ic->rds_iwdev = NULL;
+}
+
+void __rds_iw_destroy_conns(struct list_head *list, spinlock_t *list_lock)
+{
+	struct rds_iw_connection *ic, *_ic;
+	LIST_HEAD(tmp_list);
+
+	/* avoid calling conn_destroy with irqs off */
+	spin_lock_irq(list_lock);
+	list_splice(list, &tmp_list);
+	INIT_LIST_HEAD(list);
+	spin_unlock_irq(list_lock);
+
+	list_for_each_entry_safe(ic, _ic, &tmp_list, iw_node)
+		rds_conn_destroy(ic->conn);
+}
+
+static void rds_iw_set_scatterlist(struct rds_iw_scatterlist *sg,
+		struct scatterlist *list, unsigned int sg_len)
+{
+	sg->list = list;
+	sg->len = sg_len;
+	sg->dma_len = 0;
+	sg->dma_npages = 0;
+	sg->bytes = 0;
+}
+
+static u64 *rds_iw_map_scatterlist(struct rds_iw_device *rds_iwdev,
+			struct rds_iw_scatterlist *sg)
+{
+	struct ib_device *dev = rds_iwdev->dev;
+	u64 *dma_pages = NULL;
+	int i, j, ret;
+
+	WARN_ON(sg->dma_len);
+
+	sg->dma_len = ib_dma_map_sg(dev, sg->list, sg->len, DMA_BIDIRECTIONAL);
+	if (unlikely(!sg->dma_len)) {
+		printk(KERN_WARNING "RDS/IW: dma_map_sg failed!\n");
+		return ERR_PTR(-EBUSY);
+	}
+
+	sg->bytes = 0;
+	sg->dma_npages = 0;
+
+	ret = -EINVAL;
+	for (i = 0; i < sg->dma_len; ++i) {
+		unsigned int dma_len = ib_sg_dma_len(dev, &sg->list[i]);
+		u64 dma_addr = ib_sg_dma_address(dev, &sg->list[i]);
+		u64 end_addr;
+
+		sg->bytes += dma_len;
+
+		end_addr = dma_addr + dma_len;
+		if (dma_addr & PAGE_MASK) {
+			if (i > 0)
+				goto out_unmap;
+			dma_addr &= ~PAGE_MASK;
+		}
+		if (end_addr & PAGE_MASK) {
+			if (i < sg->dma_len - 1)
+				goto out_unmap;
+			end_addr = (end_addr + PAGE_MASK) & ~PAGE_MASK;
+		}
+
+		sg->dma_npages += (end_addr - dma_addr) >> PAGE_SHIFT;
+	}
+
+	/* Now gather the dma addrs into one list */
+	if (sg->dma_npages > fastreg_message_size)
+		goto out_unmap;
+
+	dma_pages = kmalloc(sizeof(u64) * sg->dma_npages, GFP_ATOMIC);
+	if (!dma_pages) {
+		ret = -ENOMEM;
+		goto out_unmap;
+	}
+
+	for (i = j = 0; i < sg->dma_len; ++i) {
+		unsigned int dma_len = ib_sg_dma_len(dev, &sg->list[i]);
+		u64 dma_addr = ib_sg_dma_address(dev, &sg->list[i]);
+		u64 end_addr;
+
+		end_addr = dma_addr + dma_len;
+		dma_addr &= ~PAGE_MASK;
+		for (; dma_addr < end_addr; dma_addr += PAGE_SIZE)
+			dma_pages[j++] = dma_addr;
+		BUG_ON(j > sg->dma_npages);
+	}
+
+	return dma_pages;
+
+out_unmap:
+	ib_dma_unmap_sg(rds_iwdev->dev, sg->list, sg->len, DMA_BIDIRECTIONAL);
+	sg->dma_len = 0;
+	kfree(dma_pages);
+	return ERR_PTR(ret);
+}
+
+
+struct rds_iw_mr_pool *rds_iw_create_mr_pool(struct rds_iw_device *rds_iwdev)
+{
+	struct rds_iw_mr_pool *pool;
+
+	pool = kzalloc(sizeof(*pool), GFP_KERNEL);
+	if (!pool) {
+		printk(KERN_WARNING "RDS/IW: rds_iw_create_mr_pool alloc error\n");
+		return ERR_PTR(-ENOMEM);
+	}
+
+	pool->device = rds_iwdev;
+	INIT_LIST_HEAD(&pool->dirty_list);
+	INIT_LIST_HEAD(&pool->clean_list);
+	mutex_init(&pool->flush_lock);
+	spin_lock_init(&pool->list_lock);
+	INIT_WORK(&pool->flush_worker, rds_iw_mr_pool_flush_worker);
+
+	pool->max_message_size = fastreg_message_size;
+	pool->max_items = fastreg_pool_size;
+	pool->max_free_pinned = pool->max_items * pool->max_message_size / 4;
+	pool->max_pages = fastreg_message_size;
+
+	/* We never allow more than max_items MRs to be allocated.
+	 * When we exceed more than max_items_soft, we start freeing
+	 * items more aggressively.
+	 * Make sure that max_items > max_items_soft > max_items / 2
+	 */
+	pool->max_items_soft = pool->max_items * 3 / 4;
+
+	return pool;
+}
+
+void rds_iw_get_mr_info(struct rds_iw_device *rds_iwdev, struct rds_info_rdma_connection *iinfo)
+{
+	struct rds_iw_mr_pool *pool = rds_iwdev->mr_pool;
+
+	iinfo->rdma_mr_max = pool->max_items;
+	iinfo->rdma_mr_size = pool->max_pages;
+}
+
+void rds_iw_destroy_mr_pool(struct rds_iw_mr_pool *pool)
+{
+	flush_workqueue(rds_wq);
+	rds_iw_flush_mr_pool(pool, 1);
+	BUG_ON(atomic_read(&pool->item_count));
+	BUG_ON(atomic_read(&pool->free_pinned));
+	kfree(pool);
+}
+
+static inline struct rds_iw_mr *rds_iw_reuse_fmr(struct rds_iw_mr_pool *pool)
+{
+	struct rds_iw_mr *ibmr = NULL;
+	unsigned long flags;
+
+	spin_lock_irqsave(&pool->list_lock, flags);
+	if (!list_empty(&pool->clean_list)) {
+		ibmr = list_entry(pool->clean_list.next, struct rds_iw_mr, mapping.m_list);
+		list_del_init(&ibmr->mapping.m_list);
+	}
+	spin_unlock_irqrestore(&pool->list_lock, flags);
+
+	return ibmr;
+}
+
+static struct rds_iw_mr *rds_iw_alloc_mr(struct rds_iw_device *rds_iwdev)
+{
+	struct rds_iw_mr_pool *pool = rds_iwdev->mr_pool;
+	struct rds_iw_mr *ibmr = NULL;
+	int err = 0, iter = 0;
+
+	while (1) {
+		ibmr = rds_iw_reuse_fmr(pool);
+		if (ibmr)
+			return ibmr;
+
+		/* No clean MRs - now we have the choice of either
+		 * allocating a fresh MR up to the limit imposed by the
+		 * driver, or flush any dirty unused MRs.
+		 * We try to avoid stalling in the send path if possible,
+		 * so we allocate as long as we're allowed to.
+		 *
+		 * We're fussy with enforcing the FMR limit, though. If the driver
+		 * tells us we can't use more than N fmrs, we shouldn't start
+		 * arguing with it */
+		if (atomic_inc_return(&pool->item_count) <= pool->max_items)
+			break;
+
+		atomic_dec(&pool->item_count);
+
+		if (++iter > 2) {
+			rds_iw_stats_inc(s_iw_rdma_mr_pool_depleted);
+			return ERR_PTR(-EAGAIN);
+		}
+
+		/* We do have some empty MRs. Flush them out. */
+		rds_iw_stats_inc(s_iw_rdma_mr_pool_wait);
+		rds_iw_flush_mr_pool(pool, 0);
+	}
+
+	ibmr = kzalloc(sizeof(*ibmr), GFP_KERNEL);
+	if (!ibmr) {
+		err = -ENOMEM;
+		goto out_no_cigar;
+	}
+
+	spin_lock_init(&ibmr->mapping.m_lock);
+	INIT_LIST_HEAD(&ibmr->mapping.m_list);
+	ibmr->mapping.m_mr = ibmr;
+
+	err = rds_iw_init_fastreg(pool, ibmr);
+	if (err)
+		goto out_no_cigar;
+
+	rds_iw_stats_inc(s_iw_rdma_mr_alloc);
+	return ibmr;
+
+out_no_cigar:
+	if (ibmr) {
+		rds_iw_destroy_fastreg(pool, ibmr);
+		kfree(ibmr);
+	}
+	atomic_dec(&pool->item_count);
+	return ERR_PTR(err);
+}
+
+void rds_iw_sync_mr(void *trans_private, int direction)
+{
+	struct rds_iw_mr *ibmr = trans_private;
+	struct rds_iw_device *rds_iwdev = ibmr->device;
+
+	switch (direction) {
+	case DMA_FROM_DEVICE:
+		ib_dma_sync_sg_for_cpu(rds_iwdev->dev, ibmr->mapping.m_sg.list,
+			ibmr->mapping.m_sg.dma_len, DMA_BIDIRECTIONAL);
+		break;
+	case DMA_TO_DEVICE:
+		ib_dma_sync_sg_for_device(rds_iwdev->dev, ibmr->mapping.m_sg.list,
+			ibmr->mapping.m_sg.dma_len, DMA_BIDIRECTIONAL);
+		break;
+	}
+}
+
+/*
+ * Flush our pool of MRs.
+ * At a minimum, all currently unused MRs are unmapped.
+ * If the number of MRs allocated exceeds the limit, we also try
+ * to free as many MRs as needed to get back to this limit.
+ */
+static int rds_iw_flush_mr_pool(struct rds_iw_mr_pool *pool, int free_all)
+{
+	struct rds_iw_mr *ibmr, *next;
+	LIST_HEAD(unmap_list);
+	LIST_HEAD(kill_list);
+	unsigned long flags;
+	unsigned int nfreed = 0, ncleaned = 0, unpinned = 0;
+	int ret = 0;
+
+	rds_iw_stats_inc(s_iw_rdma_mr_pool_flush);
+
+	mutex_lock(&pool->flush_lock);
+
+	spin_lock_irqsave(&pool->list_lock, flags);
+	/* Get the list of all mappings to be destroyed */
+	list_splice_init(&pool->dirty_list, &unmap_list);
+	if (free_all)
+		list_splice_init(&pool->clean_list, &kill_list);
+	spin_unlock_irqrestore(&pool->list_lock, flags);
+
+	/* Batched invalidate of dirty MRs.
+	 * For FMR based MRs, the mappings on the unmap list are
+	 * actually members of an ibmr (ibmr->mapping). They either
+	 * migrate to the kill_list, or have been cleaned and should be
+	 * moved to the clean_list.
+	 * For fastregs, they will be dynamically allocated, and
+	 * will be destroyed by the unmap function.
+	 */
+	if (!list_empty(&unmap_list)) {
+		ncleaned = rds_iw_unmap_fastreg_list(pool, &unmap_list,
+						     &kill_list, &unpinned);
+		/* If we've been asked to destroy all MRs, move those
+		 * that were simply cleaned to the kill list */
+		if (free_all)
+			list_splice_init(&unmap_list, &kill_list);
+	}
+
+	/* Destroy any MRs that are past their best before date */
+	list_for_each_entry_safe(ibmr, next, &kill_list, mapping.m_list) {
+		rds_iw_stats_inc(s_iw_rdma_mr_free);
+		list_del(&ibmr->mapping.m_list);
+		rds_iw_destroy_fastreg(pool, ibmr);
+		kfree(ibmr);
+		nfreed++;
+	}
+
+	/* Anything that remains are laundered ibmrs, which we can add
+	 * back to the clean list. */
+	if (!list_empty(&unmap_list)) {
+		spin_lock_irqsave(&pool->list_lock, flags);
+		list_splice(&unmap_list, &pool->clean_list);
+		spin_unlock_irqrestore(&pool->list_lock, flags);
+	}
+
+	atomic_sub(unpinned, &pool->free_pinned);
+	atomic_sub(ncleaned, &pool->dirty_count);
+	atomic_sub(nfreed, &pool->item_count);
+
+	mutex_unlock(&pool->flush_lock);
+	return ret;
+}
+
+static void rds_iw_mr_pool_flush_worker(struct work_struct *work)
+{
+	struct rds_iw_mr_pool *pool = container_of(work, struct rds_iw_mr_pool, flush_worker);
+
+	rds_iw_flush_mr_pool(pool, 0);
+}
+
+void rds_iw_free_mr(void *trans_private, int invalidate)
+{
+	struct rds_iw_mr *ibmr = trans_private;
+	struct rds_iw_mr_pool *pool = ibmr->device->mr_pool;
+
+	rdsdebug("RDS/IW: free_mr nents %u\n", ibmr->mapping.m_sg.len);
+	if (!pool)
+		return;
+
+	/* Return it to the pool's free list */
+	rds_iw_free_fastreg(pool, ibmr);
+
+	/* If we've pinned too many pages, request a flush */
+	if (atomic_read(&pool->free_pinned) >= pool->max_free_pinned ||
+	    atomic_read(&pool->dirty_count) >= pool->max_items / 10)
+		queue_work(rds_wq, &pool->flush_worker);
+
+	if (invalidate) {
+		if (likely(!in_interrupt())) {
+			rds_iw_flush_mr_pool(pool, 0);
+		} else {
+			/* We get here if the user created a MR marked
+			 * as use_once and invalidate at the same time. */
+			queue_work(rds_wq, &pool->flush_worker);
+		}
+	}
+}
+
+void rds_iw_flush_mrs(void)
+{
+	struct rds_iw_device *rds_iwdev;
+
+	list_for_each_entry(rds_iwdev, &rds_iw_devices, list) {
+		struct rds_iw_mr_pool *pool = rds_iwdev->mr_pool;
+
+		if (pool)
+			rds_iw_flush_mr_pool(pool, 0);
+	}
+}
+
+void *rds_iw_get_mr(struct scatterlist *sg, unsigned long nents,
+		    struct rds_sock *rs, u32 *key_ret)
+{
+	struct rds_iw_device *rds_iwdev;
+	struct rds_iw_mr *ibmr = NULL;
+	struct rdma_cm_id *cm_id;
+	struct sockaddr_in src = {
+		.sin_addr.s_addr = rs->rs_bound_addr,
+		.sin_port = rs->rs_bound_port,
+	};
+	struct sockaddr_in dst = {
+		.sin_addr.s_addr = rs->rs_conn_addr,
+		.sin_port = rs->rs_conn_port,
+	};
+	int ret;
+
+	ret = rds_iw_get_device(&src, &dst, &rds_iwdev, &cm_id);
+	if (ret || !cm_id) {
+		ret = -ENODEV;
+		goto out;
+	}
+
+	if (!rds_iwdev->mr_pool) {
+		ret = -ENODEV;
+		goto out;
+	}
+
+	ibmr = rds_iw_alloc_mr(rds_iwdev);
+	if (IS_ERR(ibmr))
+		return ibmr;
+
+	ibmr->cm_id = cm_id;
+	ibmr->device = rds_iwdev;
+
+	ret = rds_iw_map_fastreg(rds_iwdev->mr_pool, ibmr, sg, nents);
+	if (ret == 0)
+		*key_ret = ibmr->mr->rkey;
+	else
+		printk(KERN_WARNING "RDS/IW: failed to map mr (errno=%d)\n", ret);
+
+out:
+	if (ret) {
+		if (ibmr)
+			rds_iw_free_mr(ibmr, 0);
+		ibmr = ERR_PTR(ret);
+	}
+	return ibmr;
+}
+
+/*
+ * iWARP fastreg handling
+ *
+ * The life cycle of a fastreg registration is a bit different from
+ * FMRs.
+ * The idea behind fastreg is to have one MR, to which we bind different
+ * mappings over time. To avoid stalling on the expensive map and invalidate
+ * operations, these operations are pipelined on the same send queue on
+ * which we want to send the message containing the r_key.
+ *
+ * This creates a bit of a problem for us, as we do not have the destination
+ * IP in GET_MR, so the connection must be setup prior to the GET_MR call for
+ * RDMA to be correctly setup.  If a fastreg request is present, rds_iw_xmit
+ * will try to queue a LOCAL_INV (if needed) and a FAST_REG_MR work request
+ * before queuing the SEND. When completions for these arrive, they are
+ * dispatched to the MR has a bit set showing that RDMa can be performed.
+ *
+ * There is another interesting aspect that's related to invalidation.
+ * The application can request that a mapping is invalidated in FREE_MR.
+ * The expectation there is that this invalidation step includes ALL
+ * PREVIOUSLY FREED MRs.
+ */
+static int rds_iw_init_fastreg(struct rds_iw_mr_pool *pool,
+				struct rds_iw_mr *ibmr)
+{
+	struct rds_iw_device *rds_iwdev = pool->device;
+	struct ib_fast_reg_page_list *page_list = NULL;
+	struct ib_mr *mr;
+	int err;
+
+	mr = ib_alloc_fast_reg_mr(rds_iwdev->pd, pool->max_message_size);
+	if (IS_ERR(mr)) {
+		err = PTR_ERR(mr);
+
+		printk(KERN_WARNING "RDS/IW: ib_alloc_fast_reg_mr failed (err=%d)\n", err);
+		return err;
+	}
+
+	/* FIXME - this is overkill, but mapping->m_sg.dma_len/mapping->m_sg.dma_npages
+	 * is not filled in.
+	 */
+	page_list = ib_alloc_fast_reg_page_list(rds_iwdev->dev, pool->max_message_size);
+	if (IS_ERR(page_list)) {
+		err = PTR_ERR(page_list);
+
+		printk(KERN_WARNING "RDS/IW: ib_alloc_fast_reg_page_list failed (err=%d)\n", err);
+		ib_dereg_mr(mr);
+		return err;
+	}
+
+	ibmr->page_list = page_list;
+	ibmr->mr = mr;
+	return 0;
+}
+
+static int rds_iw_rdma_build_fastreg(struct rds_iw_mapping *mapping)
+{
+	struct rds_iw_mr *ibmr = mapping->m_mr;
+	struct ib_send_wr f_wr, *failed_wr;
+	int ret;
+
+	/*
+	 * Perform a WR for the fast_reg_mr. Each individual page
+	 * in the sg list is added to the fast reg page list and placed
+	 * inside the fast_reg_mr WR.  The key used is a rolling 8bit
+	 * counter, which should guarantee uniqueness.
+	 */
+	ib_update_fast_reg_key(ibmr->mr, ibmr->remap_count++);
+	mapping->m_rkey = ibmr->mr->rkey;
+
+	memset(&f_wr, 0, sizeof(f_wr));
+	f_wr.wr_id = RDS_IW_FAST_REG_WR_ID;
+	f_wr.opcode = IB_WR_FAST_REG_MR;
+	f_wr.wr.fast_reg.length = mapping->m_sg.bytes;
+	f_wr.wr.fast_reg.rkey = mapping->m_rkey;
+	f_wr.wr.fast_reg.page_list = ibmr->page_list;
+	f_wr.wr.fast_reg.page_list_len = mapping->m_sg.dma_len;
+	f_wr.wr.fast_reg.page_shift = PAGE_SHIFT;
+	f_wr.wr.fast_reg.access_flags = IB_ACCESS_LOCAL_WRITE |
+				IB_ACCESS_REMOTE_READ |
+				IB_ACCESS_REMOTE_WRITE;
+	f_wr.wr.fast_reg.iova_start = 0;
+	f_wr.send_flags = IB_SEND_SIGNALED;
+
+	failed_wr = &f_wr;
+	ret = ib_post_send(ibmr->cm_id->qp, &f_wr, &failed_wr);
+	BUG_ON(failed_wr != &f_wr);
+	if (ret)
+		printk_ratelimited(KERN_WARNING "RDS/IW: %s:%d ib_post_send returned %d\n",
+			__func__, __LINE__, ret);
+	return ret;
+}
+
+static int rds_iw_rdma_fastreg_inv(struct rds_iw_mr *ibmr)
+{
+	struct ib_send_wr s_wr, *failed_wr;
+	int ret = 0;
+
+	if (!ibmr->cm_id->qp || !ibmr->mr)
+		goto out;
+
+	memset(&s_wr, 0, sizeof(s_wr));
+	s_wr.wr_id = RDS_IW_LOCAL_INV_WR_ID;
+	s_wr.opcode = IB_WR_LOCAL_INV;
+	s_wr.ex.invalidate_rkey = ibmr->mr->rkey;
+	s_wr.send_flags = IB_SEND_SIGNALED;
+
+	failed_wr = &s_wr;
+	ret = ib_post_send(ibmr->cm_id->qp, &s_wr, &failed_wr);
+	if (ret) {
+		printk_ratelimited(KERN_WARNING "RDS/IW: %s:%d ib_post_send returned %d\n",
+			__func__, __LINE__, ret);
+		goto out;
+	}
+out:
+	return ret;
+}
+
+static int rds_iw_map_fastreg(struct rds_iw_mr_pool *pool,
+			struct rds_iw_mr *ibmr,
+			struct scatterlist *sg,
+			unsigned int sg_len)
+{
+	struct rds_iw_device *rds_iwdev = pool->device;
+	struct rds_iw_mapping *mapping = &ibmr->mapping;
+	u64 *dma_pages;
+	int i, ret = 0;
+
+	rds_iw_set_scatterlist(&mapping->m_sg, sg, sg_len);
+
+	dma_pages = rds_iw_map_scatterlist(rds_iwdev, &mapping->m_sg);
+	if (IS_ERR(dma_pages)) {
+		ret = PTR_ERR(dma_pages);
+		dma_pages = NULL;
+		goto out;
+	}
+
+	if (mapping->m_sg.dma_len > pool->max_message_size) {
+		ret = -EMSGSIZE;
+		goto out;
+	}
+
+	for (i = 0; i < mapping->m_sg.dma_npages; ++i)
+		ibmr->page_list->page_list[i] = dma_pages[i];
+
+	ret = rds_iw_rdma_build_fastreg(mapping);
+	if (ret)
+		goto out;
+
+	rds_iw_stats_inc(s_iw_rdma_mr_used);
+
+out:
+	kfree(dma_pages);
+
+	return ret;
+}
+
+/*
+ * "Free" a fastreg MR.
+ */
+static void rds_iw_free_fastreg(struct rds_iw_mr_pool *pool,
+		struct rds_iw_mr *ibmr)
+{
+	unsigned long flags;
+	int ret;
+
+	if (!ibmr->mapping.m_sg.dma_len)
+		return;
+
+	ret = rds_iw_rdma_fastreg_inv(ibmr);
+	if (ret)
+		return;
+
+	/* Try to post the LOCAL_INV WR to the queue. */
+	spin_lock_irqsave(&pool->list_lock, flags);
+
+	list_add_tail(&ibmr->mapping.m_list, &pool->dirty_list);
+	atomic_add(ibmr->mapping.m_sg.len, &pool->free_pinned);
+	atomic_inc(&pool->dirty_count);
+
+	spin_unlock_irqrestore(&pool->list_lock, flags);
+}
+
+static unsigned int rds_iw_unmap_fastreg_list(struct rds_iw_mr_pool *pool,
+				struct list_head *unmap_list,
+				struct list_head *kill_list,
+				int *unpinned)
+{
+	struct rds_iw_mapping *mapping, *next;
+	unsigned int ncleaned = 0;
+	LIST_HEAD(laundered);
+
+	/* Batched invalidation of fastreg MRs.
+	 * Why do we do it this way, even though we could pipeline unmap
+	 * and remap? The reason is the application semantics - when the
+	 * application requests an invalidation of MRs, it expects all
+	 * previously released R_Keys to become invalid.
+	 *
+	 * If we implement MR reuse naively, we risk memory corruption
+	 * (this has actually been observed). So the default behavior
+	 * requires that a MR goes through an explicit unmap operation before
+	 * we can reuse it again.
+	 *
+	 * We could probably improve on this a little, by allowing immediate
+	 * reuse of a MR on the same socket (eg you could add small
+	 * cache of unused MRs to strct rds_socket - GET_MR could grab one
+	 * of these without requiring an explicit invalidate).
+	 */
+	while (!list_empty(unmap_list)) {
+		unsigned long flags;
+
+		spin_lock_irqsave(&pool->list_lock, flags);
+		list_for_each_entry_safe(mapping, next, unmap_list, m_list) {
+			*unpinned += mapping->m_sg.len;
+			list_move(&mapping->m_list, &laundered);
+			ncleaned++;
+		}
+		spin_unlock_irqrestore(&pool->list_lock, flags);
+	}
+
+	/* Move all laundered mappings back to the unmap list.
+	 * We do not kill any WRs right now - it doesn't seem the
+	 * fastreg API has a max_remap limit. */
+	list_splice_init(&laundered, unmap_list);
+
+	return ncleaned;
+}
+
+static void rds_iw_destroy_fastreg(struct rds_iw_mr_pool *pool,
+		struct rds_iw_mr *ibmr)
+{
+	if (ibmr->page_list)
+		ib_free_fast_reg_page_list(ibmr->page_list);
+	if (ibmr->mr)
+		ib_dereg_mr(ibmr->mr);
+}
diff --git a/net/rds/iw_recv.c b/net/rds/iw_recv.c
new file mode 100644
index 000000000..a66d1794b
--- /dev/null
+++ b/net/rds/iw_recv.c
@@ -0,0 +1,904 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/pci.h>
+#include <linux/dma-mapping.h>
+#include <rdma/rdma_cm.h>
+
+#include "rds.h"
+#include "iw.h"
+
+static struct kmem_cache *rds_iw_incoming_slab;
+static struct kmem_cache *rds_iw_frag_slab;
+static atomic_t	rds_iw_allocation = ATOMIC_INIT(0);
+
+static void rds_iw_frag_drop_page(struct rds_page_frag *frag)
+{
+	rdsdebug("frag %p page %p\n", frag, frag->f_page);
+	__free_page(frag->f_page);
+	frag->f_page = NULL;
+}
+
+static void rds_iw_frag_free(struct rds_page_frag *frag)
+{
+	rdsdebug("frag %p page %p\n", frag, frag->f_page);
+	BUG_ON(frag->f_page);
+	kmem_cache_free(rds_iw_frag_slab, frag);
+}
+
+/*
+ * We map a page at a time.  Its fragments are posted in order.  This
+ * is called in fragment order as the fragments get send completion events.
+ * Only the last frag in the page performs the unmapping.
+ *
+ * It's OK for ring cleanup to call this in whatever order it likes because
+ * DMA is not in flight and so we can unmap while other ring entries still
+ * hold page references in their frags.
+ */
+static void rds_iw_recv_unmap_page(struct rds_iw_connection *ic,
+				   struct rds_iw_recv_work *recv)
+{
+	struct rds_page_frag *frag = recv->r_frag;
+
+	rdsdebug("recv %p frag %p page %p\n", recv, frag, frag->f_page);
+	if (frag->f_mapped)
+		ib_dma_unmap_page(ic->i_cm_id->device,
+			       frag->f_mapped,
+			       RDS_FRAG_SIZE, DMA_FROM_DEVICE);
+	frag->f_mapped = 0;
+}
+
+void rds_iw_recv_init_ring(struct rds_iw_connection *ic)
+{
+	struct rds_iw_recv_work *recv;
+	u32 i;
+
+	for (i = 0, recv = ic->i_recvs; i < ic->i_recv_ring.w_nr; i++, recv++) {
+		struct ib_sge *sge;
+
+		recv->r_iwinc = NULL;
+		recv->r_frag = NULL;
+
+		recv->r_wr.next = NULL;
+		recv->r_wr.wr_id = i;
+		recv->r_wr.sg_list = recv->r_sge;
+		recv->r_wr.num_sge = RDS_IW_RECV_SGE;
+
+		sge = rds_iw_data_sge(ic, recv->r_sge);
+		sge->addr = 0;
+		sge->length = RDS_FRAG_SIZE;
+		sge->lkey = 0;
+
+		sge = rds_iw_header_sge(ic, recv->r_sge);
+		sge->addr = ic->i_recv_hdrs_dma + (i * sizeof(struct rds_header));
+		sge->length = sizeof(struct rds_header);
+		sge->lkey = 0;
+	}
+}
+
+static void rds_iw_recv_clear_one(struct rds_iw_connection *ic,
+				  struct rds_iw_recv_work *recv)
+{
+	if (recv->r_iwinc) {
+		rds_inc_put(&recv->r_iwinc->ii_inc);
+		recv->r_iwinc = NULL;
+	}
+	if (recv->r_frag) {
+		rds_iw_recv_unmap_page(ic, recv);
+		if (recv->r_frag->f_page)
+			rds_iw_frag_drop_page(recv->r_frag);
+		rds_iw_frag_free(recv->r_frag);
+		recv->r_frag = NULL;
+	}
+}
+
+void rds_iw_recv_clear_ring(struct rds_iw_connection *ic)
+{
+	u32 i;
+
+	for (i = 0; i < ic->i_recv_ring.w_nr; i++)
+		rds_iw_recv_clear_one(ic, &ic->i_recvs[i]);
+
+	if (ic->i_frag.f_page)
+		rds_iw_frag_drop_page(&ic->i_frag);
+}
+
+static int rds_iw_recv_refill_one(struct rds_connection *conn,
+				  struct rds_iw_recv_work *recv,
+				  gfp_t kptr_gfp, gfp_t page_gfp)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+	dma_addr_t dma_addr;
+	struct ib_sge *sge;
+	int ret = -ENOMEM;
+
+	if (!recv->r_iwinc) {
+		if (!atomic_add_unless(&rds_iw_allocation, 1, rds_iw_sysctl_max_recv_allocation)) {
+			rds_iw_stats_inc(s_iw_rx_alloc_limit);
+			goto out;
+		}
+		recv->r_iwinc = kmem_cache_alloc(rds_iw_incoming_slab,
+						 kptr_gfp);
+		if (!recv->r_iwinc) {
+			atomic_dec(&rds_iw_allocation);
+			goto out;
+		}
+		INIT_LIST_HEAD(&recv->r_iwinc->ii_frags);
+		rds_inc_init(&recv->r_iwinc->ii_inc, conn, conn->c_faddr);
+	}
+
+	if (!recv->r_frag) {
+		recv->r_frag = kmem_cache_alloc(rds_iw_frag_slab, kptr_gfp);
+		if (!recv->r_frag)
+			goto out;
+		INIT_LIST_HEAD(&recv->r_frag->f_item);
+		recv->r_frag->f_page = NULL;
+	}
+
+	if (!ic->i_frag.f_page) {
+		ic->i_frag.f_page = alloc_page(page_gfp);
+		if (!ic->i_frag.f_page)
+			goto out;
+		ic->i_frag.f_offset = 0;
+	}
+
+	dma_addr = ib_dma_map_page(ic->i_cm_id->device,
+				  ic->i_frag.f_page,
+				  ic->i_frag.f_offset,
+				  RDS_FRAG_SIZE,
+				  DMA_FROM_DEVICE);
+	if (ib_dma_mapping_error(ic->i_cm_id->device, dma_addr))
+		goto out;
+
+	/*
+	 * Once we get the RDS_PAGE_LAST_OFF frag then rds_iw_frag_unmap()
+	 * must be called on this recv.  This happens as completions hit
+	 * in order or on connection shutdown.
+	 */
+	recv->r_frag->f_page = ic->i_frag.f_page;
+	recv->r_frag->f_offset = ic->i_frag.f_offset;
+	recv->r_frag->f_mapped = dma_addr;
+
+	sge = rds_iw_data_sge(ic, recv->r_sge);
+	sge->addr = dma_addr;
+	sge->length = RDS_FRAG_SIZE;
+
+	sge = rds_iw_header_sge(ic, recv->r_sge);
+	sge->addr = ic->i_recv_hdrs_dma + (recv - ic->i_recvs) * sizeof(struct rds_header);
+	sge->length = sizeof(struct rds_header);
+
+	get_page(recv->r_frag->f_page);
+
+	if (ic->i_frag.f_offset < RDS_PAGE_LAST_OFF) {
+		ic->i_frag.f_offset += RDS_FRAG_SIZE;
+	} else {
+		put_page(ic->i_frag.f_page);
+		ic->i_frag.f_page = NULL;
+		ic->i_frag.f_offset = 0;
+	}
+
+	ret = 0;
+out:
+	return ret;
+}
+
+/*
+ * This tries to allocate and post unused work requests after making sure that
+ * they have all the allocations they need to queue received fragments into
+ * sockets.  The i_recv_mutex is held here so that ring_alloc and _unalloc
+ * pairs don't go unmatched.
+ *
+ * -1 is returned if posting fails due to temporary resource exhaustion.
+ */
+int rds_iw_recv_refill(struct rds_connection *conn, gfp_t kptr_gfp,
+		       gfp_t page_gfp, int prefill)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+	struct rds_iw_recv_work *recv;
+	struct ib_recv_wr *failed_wr;
+	unsigned int posted = 0;
+	int ret = 0;
+	u32 pos;
+
+	while ((prefill || rds_conn_up(conn)) &&
+	       rds_iw_ring_alloc(&ic->i_recv_ring, 1, &pos)) {
+		if (pos >= ic->i_recv_ring.w_nr) {
+			printk(KERN_NOTICE "Argh - ring alloc returned pos=%u\n",
+					pos);
+			ret = -EINVAL;
+			break;
+		}
+
+		recv = &ic->i_recvs[pos];
+		ret = rds_iw_recv_refill_one(conn, recv, kptr_gfp, page_gfp);
+		if (ret) {
+			ret = -1;
+			break;
+		}
+
+		/* XXX when can this fail? */
+		ret = ib_post_recv(ic->i_cm_id->qp, &recv->r_wr, &failed_wr);
+		rdsdebug("recv %p iwinc %p page %p addr %lu ret %d\n", recv,
+			 recv->r_iwinc, recv->r_frag->f_page,
+			 (long) recv->r_frag->f_mapped, ret);
+		if (ret) {
+			rds_iw_conn_error(conn, "recv post on "
+			       "%pI4 returned %d, disconnecting and "
+			       "reconnecting\n", &conn->c_faddr,
+			       ret);
+			ret = -1;
+			break;
+		}
+
+		posted++;
+	}
+
+	/* We're doing flow control - update the window. */
+	if (ic->i_flowctl && posted)
+		rds_iw_advertise_credits(conn, posted);
+
+	if (ret)
+		rds_iw_ring_unalloc(&ic->i_recv_ring, 1);
+	return ret;
+}
+
+static void rds_iw_inc_purge(struct rds_incoming *inc)
+{
+	struct rds_iw_incoming *iwinc;
+	struct rds_page_frag *frag;
+	struct rds_page_frag *pos;
+
+	iwinc = container_of(inc, struct rds_iw_incoming, ii_inc);
+	rdsdebug("purging iwinc %p inc %p\n", iwinc, inc);
+
+	list_for_each_entry_safe(frag, pos, &iwinc->ii_frags, f_item) {
+		list_del_init(&frag->f_item);
+		rds_iw_frag_drop_page(frag);
+		rds_iw_frag_free(frag);
+	}
+}
+
+void rds_iw_inc_free(struct rds_incoming *inc)
+{
+	struct rds_iw_incoming *iwinc;
+
+	iwinc = container_of(inc, struct rds_iw_incoming, ii_inc);
+
+	rds_iw_inc_purge(inc);
+	rdsdebug("freeing iwinc %p inc %p\n", iwinc, inc);
+	BUG_ON(!list_empty(&iwinc->ii_frags));
+	kmem_cache_free(rds_iw_incoming_slab, iwinc);
+	atomic_dec(&rds_iw_allocation);
+	BUG_ON(atomic_read(&rds_iw_allocation) < 0);
+}
+
+int rds_iw_inc_copy_to_user(struct rds_incoming *inc, struct iov_iter *to)
+{
+	struct rds_iw_incoming *iwinc;
+	struct rds_page_frag *frag;
+	unsigned long to_copy;
+	unsigned long frag_off = 0;
+	int copied = 0;
+	int ret;
+	u32 len;
+
+	iwinc = container_of(inc, struct rds_iw_incoming, ii_inc);
+	frag = list_entry(iwinc->ii_frags.next, struct rds_page_frag, f_item);
+	len = be32_to_cpu(inc->i_hdr.h_len);
+
+	while (iov_iter_count(to) && copied < len) {
+		if (frag_off == RDS_FRAG_SIZE) {
+			frag = list_entry(frag->f_item.next,
+					  struct rds_page_frag, f_item);
+			frag_off = 0;
+		}
+		to_copy = min_t(unsigned long, iov_iter_count(to),
+				RDS_FRAG_SIZE - frag_off);
+		to_copy = min_t(unsigned long, to_copy, len - copied);
+
+		/* XXX needs + offset for multiple recvs per page */
+		rds_stats_add(s_copy_to_user, to_copy);
+		ret = copy_page_to_iter(frag->f_page,
+					frag->f_offset + frag_off,
+					to_copy,
+					to);
+		if (ret != to_copy)
+			return -EFAULT;
+
+		frag_off += to_copy;
+		copied += to_copy;
+	}
+
+	return copied;
+}
+
+/* ic starts out kzalloc()ed */
+void rds_iw_recv_init_ack(struct rds_iw_connection *ic)
+{
+	struct ib_send_wr *wr = &ic->i_ack_wr;
+	struct ib_sge *sge = &ic->i_ack_sge;
+
+	sge->addr = ic->i_ack_dma;
+	sge->length = sizeof(struct rds_header);
+	sge->lkey = rds_iw_local_dma_lkey(ic);
+
+	wr->sg_list = sge;
+	wr->num_sge = 1;
+	wr->opcode = IB_WR_SEND;
+	wr->wr_id = RDS_IW_ACK_WR_ID;
+	wr->send_flags = IB_SEND_SIGNALED | IB_SEND_SOLICITED;
+}
+
+/*
+ * You'd think that with reliable IB connections you wouldn't need to ack
+ * messages that have been received.  The problem is that IB hardware generates
+ * an ack message before it has DMAed the message into memory.  This creates a
+ * potential message loss if the HCA is disabled for any reason between when it
+ * sends the ack and before the message is DMAed and processed.  This is only a
+ * potential issue if another HCA is available for fail-over.
+ *
+ * When the remote host receives our ack they'll free the sent message from
+ * their send queue.  To decrease the latency of this we always send an ack
+ * immediately after we've received messages.
+ *
+ * For simplicity, we only have one ack in flight at a time.  This puts
+ * pressure on senders to have deep enough send queues to absorb the latency of
+ * a single ack frame being in flight.  This might not be good enough.
+ *
+ * This is implemented by have a long-lived send_wr and sge which point to a
+ * statically allocated ack frame.  This ack wr does not fall under the ring
+ * accounting that the tx and rx wrs do.  The QP attribute specifically makes
+ * room for it beyond the ring size.  Send completion notices its special
+ * wr_id and avoids working with the ring in that case.
+ */
+#ifndef KERNEL_HAS_ATOMIC64
+static void rds_iw_set_ack(struct rds_iw_connection *ic, u64 seq,
+				int ack_required)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&ic->i_ack_lock, flags);
+	ic->i_ack_next = seq;
+	if (ack_required)
+		set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
+	spin_unlock_irqrestore(&ic->i_ack_lock, flags);
+}
+
+static u64 rds_iw_get_ack(struct rds_iw_connection *ic)
+{
+	unsigned long flags;
+	u64 seq;
+
+	clear_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
+
+	spin_lock_irqsave(&ic->i_ack_lock, flags);
+	seq = ic->i_ack_next;
+	spin_unlock_irqrestore(&ic->i_ack_lock, flags);
+
+	return seq;
+}
+#else
+static void rds_iw_set_ack(struct rds_iw_connection *ic, u64 seq,
+				int ack_required)
+{
+	atomic64_set(&ic->i_ack_next, seq);
+	if (ack_required) {
+		smp_mb__before_atomic();
+		set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
+	}
+}
+
+static u64 rds_iw_get_ack(struct rds_iw_connection *ic)
+{
+	clear_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
+	smp_mb__after_atomic();
+
+	return atomic64_read(&ic->i_ack_next);
+}
+#endif
+
+
+static void rds_iw_send_ack(struct rds_iw_connection *ic, unsigned int adv_credits)
+{
+	struct rds_header *hdr = ic->i_ack;
+	struct ib_send_wr *failed_wr;
+	u64 seq;
+	int ret;
+
+	seq = rds_iw_get_ack(ic);
+
+	rdsdebug("send_ack: ic %p ack %llu\n", ic, (unsigned long long) seq);
+	rds_message_populate_header(hdr, 0, 0, 0);
+	hdr->h_ack = cpu_to_be64(seq);
+	hdr->h_credit = adv_credits;
+	rds_message_make_checksum(hdr);
+	ic->i_ack_queued = jiffies;
+
+	ret = ib_post_send(ic->i_cm_id->qp, &ic->i_ack_wr, &failed_wr);
+	if (unlikely(ret)) {
+		/* Failed to send. Release the WR, and
+		 * force another ACK.
+		 */
+		clear_bit(IB_ACK_IN_FLIGHT, &ic->i_ack_flags);
+		set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
+
+		rds_iw_stats_inc(s_iw_ack_send_failure);
+
+		rds_iw_conn_error(ic->conn, "sending ack failed\n");
+	} else
+		rds_iw_stats_inc(s_iw_ack_sent);
+}
+
+/*
+ * There are 3 ways of getting acknowledgements to the peer:
+ *  1.	We call rds_iw_attempt_ack from the recv completion handler
+ *	to send an ACK-only frame.
+ *	However, there can be only one such frame in the send queue
+ *	at any time, so we may have to postpone it.
+ *  2.	When another (data) packet is transmitted while there's
+ *	an ACK in the queue, we piggyback the ACK sequence number
+ *	on the data packet.
+ *  3.	If the ACK WR is done sending, we get called from the
+ *	send queue completion handler, and check whether there's
+ *	another ACK pending (postponed because the WR was on the
+ *	queue). If so, we transmit it.
+ *
+ * We maintain 2 variables:
+ *  -	i_ack_flags, which keeps track of whether the ACK WR
+ *	is currently in the send queue or not (IB_ACK_IN_FLIGHT)
+ *  -	i_ack_next, which is the last sequence number we received
+ *
+ * Potentially, send queue and receive queue handlers can run concurrently.
+ * It would be nice to not have to use a spinlock to synchronize things,
+ * but the one problem that rules this out is that 64bit updates are
+ * not atomic on all platforms. Things would be a lot simpler if
+ * we had atomic64 or maybe cmpxchg64 everywhere.
+ *
+ * Reconnecting complicates this picture just slightly. When we
+ * reconnect, we may be seeing duplicate packets. The peer
+ * is retransmitting them, because it hasn't seen an ACK for
+ * them. It is important that we ACK these.
+ *
+ * ACK mitigation adds a header flag "ACK_REQUIRED"; any packet with
+ * this flag set *MUST* be acknowledged immediately.
+ */
+
+/*
+ * When we get here, we're called from the recv queue handler.
+ * Check whether we ought to transmit an ACK.
+ */
+void rds_iw_attempt_ack(struct rds_iw_connection *ic)
+{
+	unsigned int adv_credits;
+
+	if (!test_bit(IB_ACK_REQUESTED, &ic->i_ack_flags))
+		return;
+
+	if (test_and_set_bit(IB_ACK_IN_FLIGHT, &ic->i_ack_flags)) {
+		rds_iw_stats_inc(s_iw_ack_send_delayed);
+		return;
+	}
+
+	/* Can we get a send credit? */
+	if (!rds_iw_send_grab_credits(ic, 1, &adv_credits, 0, RDS_MAX_ADV_CREDIT)) {
+		rds_iw_stats_inc(s_iw_tx_throttle);
+		clear_bit(IB_ACK_IN_FLIGHT, &ic->i_ack_flags);
+		return;
+	}
+
+	clear_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
+	rds_iw_send_ack(ic, adv_credits);
+}
+
+/*
+ * We get here from the send completion handler, when the
+ * adapter tells us the ACK frame was sent.
+ */
+void rds_iw_ack_send_complete(struct rds_iw_connection *ic)
+{
+	clear_bit(IB_ACK_IN_FLIGHT, &ic->i_ack_flags);
+	rds_iw_attempt_ack(ic);
+}
+
+/*
+ * This is called by the regular xmit code when it wants to piggyback
+ * an ACK on an outgoing frame.
+ */
+u64 rds_iw_piggyb_ack(struct rds_iw_connection *ic)
+{
+	if (test_and_clear_bit(IB_ACK_REQUESTED, &ic->i_ack_flags))
+		rds_iw_stats_inc(s_iw_ack_send_piggybacked);
+	return rds_iw_get_ack(ic);
+}
+
+/*
+ * It's kind of lame that we're copying from the posted receive pages into
+ * long-lived bitmaps.  We could have posted the bitmaps and rdma written into
+ * them.  But receiving new congestion bitmaps should be a *rare* event, so
+ * hopefully we won't need to invest that complexity in making it more
+ * efficient.  By copying we can share a simpler core with TCP which has to
+ * copy.
+ */
+static void rds_iw_cong_recv(struct rds_connection *conn,
+			      struct rds_iw_incoming *iwinc)
+{
+	struct rds_cong_map *map;
+	unsigned int map_off;
+	unsigned int map_page;
+	struct rds_page_frag *frag;
+	unsigned long frag_off;
+	unsigned long to_copy;
+	unsigned long copied;
+	uint64_t uncongested = 0;
+	void *addr;
+
+	/* catch completely corrupt packets */
+	if (be32_to_cpu(iwinc->ii_inc.i_hdr.h_len) != RDS_CONG_MAP_BYTES)
+		return;
+
+	map = conn->c_fcong;
+	map_page = 0;
+	map_off = 0;
+
+	frag = list_entry(iwinc->ii_frags.next, struct rds_page_frag, f_item);
+	frag_off = 0;
+
+	copied = 0;
+
+	while (copied < RDS_CONG_MAP_BYTES) {
+		uint64_t *src, *dst;
+		unsigned int k;
+
+		to_copy = min(RDS_FRAG_SIZE - frag_off, PAGE_SIZE - map_off);
+		BUG_ON(to_copy & 7); /* Must be 64bit aligned. */
+
+		addr = kmap_atomic(frag->f_page);
+
+		src = addr + frag_off;
+		dst = (void *)map->m_page_addrs[map_page] + map_off;
+		for (k = 0; k < to_copy; k += 8) {
+			/* Record ports that became uncongested, ie
+			 * bits that changed from 0 to 1. */
+			uncongested |= ~(*src) & *dst;
+			*dst++ = *src++;
+		}
+		kunmap_atomic(addr);
+
+		copied += to_copy;
+
+		map_off += to_copy;
+		if (map_off == PAGE_SIZE) {
+			map_off = 0;
+			map_page++;
+		}
+
+		frag_off += to_copy;
+		if (frag_off == RDS_FRAG_SIZE) {
+			frag = list_entry(frag->f_item.next,
+					  struct rds_page_frag, f_item);
+			frag_off = 0;
+		}
+	}
+
+	/* the congestion map is in little endian order */
+	uncongested = le64_to_cpu(uncongested);
+
+	rds_cong_map_updated(map, uncongested);
+}
+
+/*
+ * Rings are posted with all the allocations they'll need to queue the
+ * incoming message to the receiving socket so this can't fail.
+ * All fragments start with a header, so we can make sure we're not receiving
+ * garbage, and we can tell a small 8 byte fragment from an ACK frame.
+ */
+struct rds_iw_ack_state {
+	u64		ack_next;
+	u64		ack_recv;
+	unsigned int	ack_required:1;
+	unsigned int	ack_next_valid:1;
+	unsigned int	ack_recv_valid:1;
+};
+
+static void rds_iw_process_recv(struct rds_connection *conn,
+				struct rds_iw_recv_work *recv, u32 byte_len,
+				struct rds_iw_ack_state *state)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+	struct rds_iw_incoming *iwinc = ic->i_iwinc;
+	struct rds_header *ihdr, *hdr;
+
+	/* XXX shut down the connection if port 0,0 are seen? */
+
+	rdsdebug("ic %p iwinc %p recv %p byte len %u\n", ic, iwinc, recv,
+		 byte_len);
+
+	if (byte_len < sizeof(struct rds_header)) {
+		rds_iw_conn_error(conn, "incoming message "
+		       "from %pI4 didn't include a "
+		       "header, disconnecting and "
+		       "reconnecting\n",
+		       &conn->c_faddr);
+		return;
+	}
+	byte_len -= sizeof(struct rds_header);
+
+	ihdr = &ic->i_recv_hdrs[recv - ic->i_recvs];
+
+	/* Validate the checksum. */
+	if (!rds_message_verify_checksum(ihdr)) {
+		rds_iw_conn_error(conn, "incoming message "
+		       "from %pI4 has corrupted header - "
+		       "forcing a reconnect\n",
+		       &conn->c_faddr);
+		rds_stats_inc(s_recv_drop_bad_checksum);
+		return;
+	}
+
+	/* Process the ACK sequence which comes with every packet */
+	state->ack_recv = be64_to_cpu(ihdr->h_ack);
+	state->ack_recv_valid = 1;
+
+	/* Process the credits update if there was one */
+	if (ihdr->h_credit)
+		rds_iw_send_add_credits(conn, ihdr->h_credit);
+
+	if (ihdr->h_sport == 0 && ihdr->h_dport == 0 && byte_len == 0) {
+		/* This is an ACK-only packet. The fact that it gets
+		 * special treatment here is that historically, ACKs
+		 * were rather special beasts.
+		 */
+		rds_iw_stats_inc(s_iw_ack_received);
+
+		/*
+		 * Usually the frags make their way on to incs and are then freed as
+		 * the inc is freed.  We don't go that route, so we have to drop the
+		 * page ref ourselves.  We can't just leave the page on the recv
+		 * because that confuses the dma mapping of pages and each recv's use
+		 * of a partial page.  We can leave the frag, though, it will be
+		 * reused.
+		 *
+		 * FIXME: Fold this into the code path below.
+		 */
+		rds_iw_frag_drop_page(recv->r_frag);
+		return;
+	}
+
+	/*
+	 * If we don't already have an inc on the connection then this
+	 * fragment has a header and starts a message.. copy its header
+	 * into the inc and save the inc so we can hang upcoming fragments
+	 * off its list.
+	 */
+	if (!iwinc) {
+		iwinc = recv->r_iwinc;
+		recv->r_iwinc = NULL;
+		ic->i_iwinc = iwinc;
+
+		hdr = &iwinc->ii_inc.i_hdr;
+		memcpy(hdr, ihdr, sizeof(*hdr));
+		ic->i_recv_data_rem = be32_to_cpu(hdr->h_len);
+
+		rdsdebug("ic %p iwinc %p rem %u flag 0x%x\n", ic, iwinc,
+			 ic->i_recv_data_rem, hdr->h_flags);
+	} else {
+		hdr = &iwinc->ii_inc.i_hdr;
+		/* We can't just use memcmp here; fragments of a
+		 * single message may carry different ACKs */
+		if (hdr->h_sequence != ihdr->h_sequence ||
+		    hdr->h_len != ihdr->h_len ||
+		    hdr->h_sport != ihdr->h_sport ||
+		    hdr->h_dport != ihdr->h_dport) {
+			rds_iw_conn_error(conn,
+				"fragment header mismatch; forcing reconnect\n");
+			return;
+		}
+	}
+
+	list_add_tail(&recv->r_frag->f_item, &iwinc->ii_frags);
+	recv->r_frag = NULL;
+
+	if (ic->i_recv_data_rem > RDS_FRAG_SIZE)
+		ic->i_recv_data_rem -= RDS_FRAG_SIZE;
+	else {
+		ic->i_recv_data_rem = 0;
+		ic->i_iwinc = NULL;
+
+		if (iwinc->ii_inc.i_hdr.h_flags == RDS_FLAG_CONG_BITMAP)
+			rds_iw_cong_recv(conn, iwinc);
+		else {
+			rds_recv_incoming(conn, conn->c_faddr, conn->c_laddr,
+					  &iwinc->ii_inc, GFP_ATOMIC);
+			state->ack_next = be64_to_cpu(hdr->h_sequence);
+			state->ack_next_valid = 1;
+		}
+
+		/* Evaluate the ACK_REQUIRED flag *after* we received
+		 * the complete frame, and after bumping the next_rx
+		 * sequence. */
+		if (hdr->h_flags & RDS_FLAG_ACK_REQUIRED) {
+			rds_stats_inc(s_recv_ack_required);
+			state->ack_required = 1;
+		}
+
+		rds_inc_put(&iwinc->ii_inc);
+	}
+}
+
+/*
+ * Plucking the oldest entry from the ring can be done concurrently with
+ * the thread refilling the ring.  Each ring operation is protected by
+ * spinlocks and the transient state of refilling doesn't change the
+ * recording of which entry is oldest.
+ *
+ * This relies on IB only calling one cq comp_handler for each cq so that
+ * there will only be one caller of rds_recv_incoming() per RDS connection.
+ */
+void rds_iw_recv_cq_comp_handler(struct ib_cq *cq, void *context)
+{
+	struct rds_connection *conn = context;
+	struct rds_iw_connection *ic = conn->c_transport_data;
+
+	rdsdebug("conn %p cq %p\n", conn, cq);
+
+	rds_iw_stats_inc(s_iw_rx_cq_call);
+
+	tasklet_schedule(&ic->i_recv_tasklet);
+}
+
+static inline void rds_poll_cq(struct rds_iw_connection *ic,
+			       struct rds_iw_ack_state *state)
+{
+	struct rds_connection *conn = ic->conn;
+	struct ib_wc wc;
+	struct rds_iw_recv_work *recv;
+
+	while (ib_poll_cq(ic->i_recv_cq, 1, &wc) > 0) {
+		rdsdebug("wc wr_id 0x%llx status %u byte_len %u imm_data %u\n",
+			 (unsigned long long)wc.wr_id, wc.status, wc.byte_len,
+			 be32_to_cpu(wc.ex.imm_data));
+		rds_iw_stats_inc(s_iw_rx_cq_event);
+
+		recv = &ic->i_recvs[rds_iw_ring_oldest(&ic->i_recv_ring)];
+
+		rds_iw_recv_unmap_page(ic, recv);
+
+		/*
+		 * Also process recvs in connecting state because it is possible
+		 * to get a recv completion _before_ the rdmacm ESTABLISHED
+		 * event is processed.
+		 */
+		if (rds_conn_up(conn) || rds_conn_connecting(conn)) {
+			/* We expect errors as the qp is drained during shutdown */
+			if (wc.status == IB_WC_SUCCESS) {
+				rds_iw_process_recv(conn, recv, wc.byte_len, state);
+			} else {
+				rds_iw_conn_error(conn, "recv completion on "
+				       "%pI4 had status %u, disconnecting and "
+				       "reconnecting\n", &conn->c_faddr,
+				       wc.status);
+			}
+		}
+
+		rds_iw_ring_free(&ic->i_recv_ring, 1);
+	}
+}
+
+void rds_iw_recv_tasklet_fn(unsigned long data)
+{
+	struct rds_iw_connection *ic = (struct rds_iw_connection *) data;
+	struct rds_connection *conn = ic->conn;
+	struct rds_iw_ack_state state = { 0, };
+
+	rds_poll_cq(ic, &state);
+	ib_req_notify_cq(ic->i_recv_cq, IB_CQ_SOLICITED);
+	rds_poll_cq(ic, &state);
+
+	if (state.ack_next_valid)
+		rds_iw_set_ack(ic, state.ack_next, state.ack_required);
+	if (state.ack_recv_valid && state.ack_recv > ic->i_ack_recv) {
+		rds_send_drop_acked(conn, state.ack_recv, NULL);
+		ic->i_ack_recv = state.ack_recv;
+	}
+	if (rds_conn_up(conn))
+		rds_iw_attempt_ack(ic);
+
+	/* If we ever end up with a really empty receive ring, we're
+	 * in deep trouble, as the sender will definitely see RNR
+	 * timeouts. */
+	if (rds_iw_ring_empty(&ic->i_recv_ring))
+		rds_iw_stats_inc(s_iw_rx_ring_empty);
+
+	/*
+	 * If the ring is running low, then schedule the thread to refill.
+	 */
+	if (rds_iw_ring_low(&ic->i_recv_ring))
+		queue_delayed_work(rds_wq, &conn->c_recv_w, 0);
+}
+
+int rds_iw_recv(struct rds_connection *conn)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+	int ret = 0;
+
+	rdsdebug("conn %p\n", conn);
+
+	/*
+	 * If we get a temporary posting failure in this context then
+	 * we're really low and we want the caller to back off for a bit.
+	 */
+	mutex_lock(&ic->i_recv_mutex);
+	if (rds_iw_recv_refill(conn, GFP_KERNEL, GFP_HIGHUSER, 0))
+		ret = -ENOMEM;
+	else
+		rds_iw_stats_inc(s_iw_rx_refill_from_thread);
+	mutex_unlock(&ic->i_recv_mutex);
+
+	if (rds_conn_up(conn))
+		rds_iw_attempt_ack(ic);
+
+	return ret;
+}
+
+int rds_iw_recv_init(void)
+{
+	struct sysinfo si;
+	int ret = -ENOMEM;
+
+	/* Default to 30% of all available RAM for recv memory */
+	si_meminfo(&si);
+	rds_iw_sysctl_max_recv_allocation = si.totalram / 3 * PAGE_SIZE / RDS_FRAG_SIZE;
+
+	rds_iw_incoming_slab = kmem_cache_create("rds_iw_incoming",
+					sizeof(struct rds_iw_incoming),
+					0, 0, NULL);
+	if (!rds_iw_incoming_slab)
+		goto out;
+
+	rds_iw_frag_slab = kmem_cache_create("rds_iw_frag",
+					sizeof(struct rds_page_frag),
+					0, 0, NULL);
+	if (!rds_iw_frag_slab)
+		kmem_cache_destroy(rds_iw_incoming_slab);
+	else
+		ret = 0;
+out:
+	return ret;
+}
+
+void rds_iw_recv_exit(void)
+{
+	kmem_cache_destroy(rds_iw_incoming_slab);
+	kmem_cache_destroy(rds_iw_frag_slab);
+}
diff --git a/net/rds/iw_ring.c b/net/rds/iw_ring.c
new file mode 100644
index 000000000..da8e3b63f
--- /dev/null
+++ b/net/rds/iw_ring.c
@@ -0,0 +1,169 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+
+#include "rds.h"
+#include "iw.h"
+
+/*
+ * Locking for IB rings.
+ * We assume that allocation is always protected by a mutex
+ * in the caller (this is a valid assumption for the current
+ * implementation).
+ *
+ * Freeing always happens in an interrupt, and hence only
+ * races with allocations, but not with other free()s.
+ *
+ * The interaction between allocation and freeing is that
+ * the alloc code has to determine the number of free entries.
+ * To this end, we maintain two counters; an allocation counter
+ * and a free counter. Both are allowed to run freely, and wrap
+ * around.
+ * The number of used entries is always (alloc_ctr - free_ctr) % NR.
+ *
+ * The current implementation makes free_ctr atomic. When the
+ * caller finds an allocation fails, it should set an "alloc fail"
+ * bit and retry the allocation. The "alloc fail" bit essentially tells
+ * the CQ completion handlers to wake it up after freeing some
+ * more entries.
+ */
+
+/*
+ * This only happens on shutdown.
+ */
+DECLARE_WAIT_QUEUE_HEAD(rds_iw_ring_empty_wait);
+
+void rds_iw_ring_init(struct rds_iw_work_ring *ring, u32 nr)
+{
+	memset(ring, 0, sizeof(*ring));
+	ring->w_nr = nr;
+	rdsdebug("ring %p nr %u\n", ring, ring->w_nr);
+}
+
+static inline u32 __rds_iw_ring_used(struct rds_iw_work_ring *ring)
+{
+	u32 diff;
+
+	/* This assumes that atomic_t has at least as many bits as u32 */
+	diff = ring->w_alloc_ctr - (u32) atomic_read(&ring->w_free_ctr);
+	BUG_ON(diff > ring->w_nr);
+
+	return diff;
+}
+
+void rds_iw_ring_resize(struct rds_iw_work_ring *ring, u32 nr)
+{
+	/* We only ever get called from the connection setup code,
+	 * prior to creating the QP. */
+	BUG_ON(__rds_iw_ring_used(ring));
+	ring->w_nr = nr;
+}
+
+static int __rds_iw_ring_empty(struct rds_iw_work_ring *ring)
+{
+	return __rds_iw_ring_used(ring) == 0;
+}
+
+u32 rds_iw_ring_alloc(struct rds_iw_work_ring *ring, u32 val, u32 *pos)
+{
+	u32 ret = 0, avail;
+
+	avail = ring->w_nr - __rds_iw_ring_used(ring);
+
+	rdsdebug("ring %p val %u next %u free %u\n", ring, val,
+		 ring->w_alloc_ptr, avail);
+
+	if (val && avail) {
+		ret = min(val, avail);
+		*pos = ring->w_alloc_ptr;
+
+		ring->w_alloc_ptr = (ring->w_alloc_ptr + ret) % ring->w_nr;
+		ring->w_alloc_ctr += ret;
+	}
+
+	return ret;
+}
+
+void rds_iw_ring_free(struct rds_iw_work_ring *ring, u32 val)
+{
+	ring->w_free_ptr = (ring->w_free_ptr + val) % ring->w_nr;
+	atomic_add(val, &ring->w_free_ctr);
+
+	if (__rds_iw_ring_empty(ring) &&
+	    waitqueue_active(&rds_iw_ring_empty_wait))
+		wake_up(&rds_iw_ring_empty_wait);
+}
+
+void rds_iw_ring_unalloc(struct rds_iw_work_ring *ring, u32 val)
+{
+	ring->w_alloc_ptr = (ring->w_alloc_ptr - val) % ring->w_nr;
+	ring->w_alloc_ctr -= val;
+}
+
+int rds_iw_ring_empty(struct rds_iw_work_ring *ring)
+{
+	return __rds_iw_ring_empty(ring);
+}
+
+int rds_iw_ring_low(struct rds_iw_work_ring *ring)
+{
+	return __rds_iw_ring_used(ring) <= (ring->w_nr >> 1);
+}
+
+
+/*
+ * returns the oldest alloced ring entry.  This will be the next one
+ * freed.  This can't be called if there are none allocated.
+ */
+u32 rds_iw_ring_oldest(struct rds_iw_work_ring *ring)
+{
+	return ring->w_free_ptr;
+}
+
+/*
+ * returns the number of completed work requests.
+ */
+
+u32 rds_iw_ring_completed(struct rds_iw_work_ring *ring, u32 wr_id, u32 oldest)
+{
+	u32 ret;
+
+	if (oldest <= (unsigned long long)wr_id)
+		ret = (unsigned long long)wr_id - oldest + 1;
+	else
+		ret = ring->w_nr - oldest + (unsigned long long)wr_id + 1;
+
+	rdsdebug("ring %p ret %u wr_id %u oldest %u\n", ring, ret,
+		 wr_id, oldest);
+	return ret;
+}
diff --git a/net/rds/iw_send.c b/net/rds/iw_send.c
new file mode 100644
index 000000000..13834780a
--- /dev/null
+++ b/net/rds/iw_send.c
@@ -0,0 +1,974 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/in.h>
+#include <linux/device.h>
+#include <linux/dmapool.h>
+#include <linux/ratelimit.h>
+
+#include "rds.h"
+#include "iw.h"
+
+static void rds_iw_send_rdma_complete(struct rds_message *rm,
+				      int wc_status)
+{
+	int notify_status;
+
+	switch (wc_status) {
+	case IB_WC_WR_FLUSH_ERR:
+		return;
+
+	case IB_WC_SUCCESS:
+		notify_status = RDS_RDMA_SUCCESS;
+		break;
+
+	case IB_WC_REM_ACCESS_ERR:
+		notify_status = RDS_RDMA_REMOTE_ERROR;
+		break;
+
+	default:
+		notify_status = RDS_RDMA_OTHER_ERROR;
+		break;
+	}
+	rds_rdma_send_complete(rm, notify_status);
+}
+
+static void rds_iw_send_unmap_rdma(struct rds_iw_connection *ic,
+				   struct rm_rdma_op *op)
+{
+	if (op->op_mapped) {
+		ib_dma_unmap_sg(ic->i_cm_id->device,
+			op->op_sg, op->op_nents,
+			op->op_write ? DMA_TO_DEVICE : DMA_FROM_DEVICE);
+		op->op_mapped = 0;
+	}
+}
+
+static void rds_iw_send_unmap_rm(struct rds_iw_connection *ic,
+			  struct rds_iw_send_work *send,
+			  int wc_status)
+{
+	struct rds_message *rm = send->s_rm;
+
+	rdsdebug("ic %p send %p rm %p\n", ic, send, rm);
+
+	ib_dma_unmap_sg(ic->i_cm_id->device,
+		     rm->data.op_sg, rm->data.op_nents,
+		     DMA_TO_DEVICE);
+
+	if (rm->rdma.op_active) {
+		rds_iw_send_unmap_rdma(ic, &rm->rdma);
+
+		/* If the user asked for a completion notification on this
+		 * message, we can implement three different semantics:
+		 *  1.	Notify when we received the ACK on the RDS message
+		 *	that was queued with the RDMA. This provides reliable
+		 *	notification of RDMA status at the expense of a one-way
+		 *	packet delay.
+		 *  2.	Notify when the IB stack gives us the completion event for
+		 *	the RDMA operation.
+		 *  3.	Notify when the IB stack gives us the completion event for
+		 *	the accompanying RDS messages.
+		 * Here, we implement approach #3. To implement approach #2,
+		 * call rds_rdma_send_complete from the cq_handler. To implement #1,
+		 * don't call rds_rdma_send_complete at all, and fall back to the notify
+		 * handling in the ACK processing code.
+		 *
+		 * Note: There's no need to explicitly sync any RDMA buffers using
+		 * ib_dma_sync_sg_for_cpu - the completion for the RDMA
+		 * operation itself unmapped the RDMA buffers, which takes care
+		 * of synching.
+		 */
+		rds_iw_send_rdma_complete(rm, wc_status);
+
+		if (rm->rdma.op_write)
+			rds_stats_add(s_send_rdma_bytes, rm->rdma.op_bytes);
+		else
+			rds_stats_add(s_recv_rdma_bytes, rm->rdma.op_bytes);
+	}
+
+	/* If anyone waited for this message to get flushed out, wake
+	 * them up now */
+	rds_message_unmapped(rm);
+
+	rds_message_put(rm);
+	send->s_rm = NULL;
+}
+
+void rds_iw_send_init_ring(struct rds_iw_connection *ic)
+{
+	struct rds_iw_send_work *send;
+	u32 i;
+
+	for (i = 0, send = ic->i_sends; i < ic->i_send_ring.w_nr; i++, send++) {
+		struct ib_sge *sge;
+
+		send->s_rm = NULL;
+		send->s_op = NULL;
+		send->s_mapping = NULL;
+
+		send->s_wr.next = NULL;
+		send->s_wr.wr_id = i;
+		send->s_wr.sg_list = send->s_sge;
+		send->s_wr.num_sge = 1;
+		send->s_wr.opcode = IB_WR_SEND;
+		send->s_wr.send_flags = 0;
+		send->s_wr.ex.imm_data = 0;
+
+		sge = rds_iw_data_sge(ic, send->s_sge);
+		sge->lkey = 0;
+
+		sge = rds_iw_header_sge(ic, send->s_sge);
+		sge->addr = ic->i_send_hdrs_dma + (i * sizeof(struct rds_header));
+		sge->length = sizeof(struct rds_header);
+		sge->lkey = 0;
+
+		send->s_mr = ib_alloc_fast_reg_mr(ic->i_pd, fastreg_message_size);
+		if (IS_ERR(send->s_mr)) {
+			printk(KERN_WARNING "RDS/IW: ib_alloc_fast_reg_mr failed\n");
+			break;
+		}
+
+		send->s_page_list = ib_alloc_fast_reg_page_list(
+			ic->i_cm_id->device, fastreg_message_size);
+		if (IS_ERR(send->s_page_list)) {
+			printk(KERN_WARNING "RDS/IW: ib_alloc_fast_reg_page_list failed\n");
+			break;
+		}
+	}
+}
+
+void rds_iw_send_clear_ring(struct rds_iw_connection *ic)
+{
+	struct rds_iw_send_work *send;
+	u32 i;
+
+	for (i = 0, send = ic->i_sends; i < ic->i_send_ring.w_nr; i++, send++) {
+		BUG_ON(!send->s_mr);
+		ib_dereg_mr(send->s_mr);
+		BUG_ON(!send->s_page_list);
+		ib_free_fast_reg_page_list(send->s_page_list);
+		if (send->s_wr.opcode == 0xdead)
+			continue;
+		if (send->s_rm)
+			rds_iw_send_unmap_rm(ic, send, IB_WC_WR_FLUSH_ERR);
+		if (send->s_op)
+			rds_iw_send_unmap_rdma(ic, send->s_op);
+	}
+}
+
+/*
+ * The _oldest/_free ring operations here race cleanly with the alloc/unalloc
+ * operations performed in the send path.  As the sender allocs and potentially
+ * unallocs the next free entry in the ring it doesn't alter which is
+ * the next to be freed, which is what this is concerned with.
+ */
+void rds_iw_send_cq_comp_handler(struct ib_cq *cq, void *context)
+{
+	struct rds_connection *conn = context;
+	struct rds_iw_connection *ic = conn->c_transport_data;
+	struct ib_wc wc;
+	struct rds_iw_send_work *send;
+	u32 completed;
+	u32 oldest;
+	u32 i;
+	int ret;
+
+	rdsdebug("cq %p conn %p\n", cq, conn);
+	rds_iw_stats_inc(s_iw_tx_cq_call);
+	ret = ib_req_notify_cq(cq, IB_CQ_NEXT_COMP);
+	if (ret)
+		rdsdebug("ib_req_notify_cq send failed: %d\n", ret);
+
+	while (ib_poll_cq(cq, 1, &wc) > 0) {
+		rdsdebug("wc wr_id 0x%llx status %u byte_len %u imm_data %u\n",
+			 (unsigned long long)wc.wr_id, wc.status, wc.byte_len,
+			 be32_to_cpu(wc.ex.imm_data));
+		rds_iw_stats_inc(s_iw_tx_cq_event);
+
+		if (wc.status != IB_WC_SUCCESS) {
+			printk(KERN_ERR "WC Error:  status = %d opcode = %d\n", wc.status, wc.opcode);
+			break;
+		}
+
+		if (wc.opcode == IB_WC_LOCAL_INV && wc.wr_id == RDS_IW_LOCAL_INV_WR_ID) {
+			ic->i_fastreg_posted = 0;
+			continue;
+		}
+
+		if (wc.opcode == IB_WC_FAST_REG_MR && wc.wr_id == RDS_IW_FAST_REG_WR_ID) {
+			ic->i_fastreg_posted = 1;
+			continue;
+		}
+
+		if (wc.wr_id == RDS_IW_ACK_WR_ID) {
+			if (time_after(jiffies, ic->i_ack_queued + HZ/2))
+				rds_iw_stats_inc(s_iw_tx_stalled);
+			rds_iw_ack_send_complete(ic);
+			continue;
+		}
+
+		oldest = rds_iw_ring_oldest(&ic->i_send_ring);
+
+		completed = rds_iw_ring_completed(&ic->i_send_ring, wc.wr_id, oldest);
+
+		for (i = 0; i < completed; i++) {
+			send = &ic->i_sends[oldest];
+
+			/* In the error case, wc.opcode sometimes contains garbage */
+			switch (send->s_wr.opcode) {
+			case IB_WR_SEND:
+				if (send->s_rm)
+					rds_iw_send_unmap_rm(ic, send, wc.status);
+				break;
+			case IB_WR_FAST_REG_MR:
+			case IB_WR_RDMA_WRITE:
+			case IB_WR_RDMA_READ:
+			case IB_WR_RDMA_READ_WITH_INV:
+				/* Nothing to be done - the SG list will be unmapped
+				 * when the SEND completes. */
+				break;
+			default:
+				printk_ratelimited(KERN_NOTICE
+						"RDS/IW: %s: unexpected opcode 0x%x in WR!\n",
+						__func__, send->s_wr.opcode);
+				break;
+			}
+
+			send->s_wr.opcode = 0xdead;
+			send->s_wr.num_sge = 1;
+			if (time_after(jiffies, send->s_queued + HZ/2))
+				rds_iw_stats_inc(s_iw_tx_stalled);
+
+			/* If a RDMA operation produced an error, signal this right
+			 * away. If we don't, the subsequent SEND that goes with this
+			 * RDMA will be canceled with ERR_WFLUSH, and the application
+			 * never learn that the RDMA failed. */
+			if (unlikely(wc.status == IB_WC_REM_ACCESS_ERR && send->s_op)) {
+				struct rds_message *rm;
+
+				rm = rds_send_get_message(conn, send->s_op);
+				if (rm)
+					rds_iw_send_rdma_complete(rm, wc.status);
+			}
+
+			oldest = (oldest + 1) % ic->i_send_ring.w_nr;
+		}
+
+		rds_iw_ring_free(&ic->i_send_ring, completed);
+
+		if (test_and_clear_bit(RDS_LL_SEND_FULL, &conn->c_flags) ||
+		    test_bit(0, &conn->c_map_queued))
+			queue_delayed_work(rds_wq, &conn->c_send_w, 0);
+
+		/* We expect errors as the qp is drained during shutdown */
+		if (wc.status != IB_WC_SUCCESS && rds_conn_up(conn)) {
+			rds_iw_conn_error(conn,
+				"send completion on %pI4 "
+				"had status %u, disconnecting and reconnecting\n",
+				&conn->c_faddr, wc.status);
+		}
+	}
+}
+
+/*
+ * This is the main function for allocating credits when sending
+ * messages.
+ *
+ * Conceptually, we have two counters:
+ *  -	send credits: this tells us how many WRs we're allowed
+ *	to submit without overruning the receiver's queue. For
+ *	each SEND WR we post, we decrement this by one.
+ *
+ *  -	posted credits: this tells us how many WRs we recently
+ *	posted to the receive queue. This value is transferred
+ *	to the peer as a "credit update" in a RDS header field.
+ *	Every time we transmit credits to the peer, we subtract
+ *	the amount of transferred credits from this counter.
+ *
+ * It is essential that we avoid situations where both sides have
+ * exhausted their send credits, and are unable to send new credits
+ * to the peer. We achieve this by requiring that we send at least
+ * one credit update to the peer before exhausting our credits.
+ * When new credits arrive, we subtract one credit that is withheld
+ * until we've posted new buffers and are ready to transmit these
+ * credits (see rds_iw_send_add_credits below).
+ *
+ * The RDS send code is essentially single-threaded; rds_send_xmit
+ * grabs c_send_lock to ensure exclusive access to the send ring.
+ * However, the ACK sending code is independent and can race with
+ * message SENDs.
+ *
+ * In the send path, we need to update the counters for send credits
+ * and the counter of posted buffers atomically - when we use the
+ * last available credit, we cannot allow another thread to race us
+ * and grab the posted credits counter.  Hence, we have to use a
+ * spinlock to protect the credit counter, or use atomics.
+ *
+ * Spinlocks shared between the send and the receive path are bad,
+ * because they create unnecessary delays. An early implementation
+ * using a spinlock showed a 5% degradation in throughput at some
+ * loads.
+ *
+ * This implementation avoids spinlocks completely, putting both
+ * counters into a single atomic, and updating that atomic using
+ * atomic_add (in the receive path, when receiving fresh credits),
+ * and using atomic_cmpxchg when updating the two counters.
+ */
+int rds_iw_send_grab_credits(struct rds_iw_connection *ic,
+			     u32 wanted, u32 *adv_credits, int need_posted, int max_posted)
+{
+	unsigned int avail, posted, got = 0, advertise;
+	long oldval, newval;
+
+	*adv_credits = 0;
+	if (!ic->i_flowctl)
+		return wanted;
+
+try_again:
+	advertise = 0;
+	oldval = newval = atomic_read(&ic->i_credits);
+	posted = IB_GET_POST_CREDITS(oldval);
+	avail = IB_GET_SEND_CREDITS(oldval);
+
+	rdsdebug("wanted=%u credits=%u posted=%u\n",
+			wanted, avail, posted);
+
+	/* The last credit must be used to send a credit update. */
+	if (avail && !posted)
+		avail--;
+
+	if (avail < wanted) {
+		struct rds_connection *conn = ic->i_cm_id->context;
+
+		/* Oops, there aren't that many credits left! */
+		set_bit(RDS_LL_SEND_FULL, &conn->c_flags);
+		got = avail;
+	} else {
+		/* Sometimes you get what you want, lalala. */
+		got = wanted;
+	}
+	newval -= IB_SET_SEND_CREDITS(got);
+
+	/*
+	 * If need_posted is non-zero, then the caller wants
+	 * the posted regardless of whether any send credits are
+	 * available.
+	 */
+	if (posted && (got || need_posted)) {
+		advertise = min_t(unsigned int, posted, max_posted);
+		newval -= IB_SET_POST_CREDITS(advertise);
+	}
+
+	/* Finally bill everything */
+	if (atomic_cmpxchg(&ic->i_credits, oldval, newval) != oldval)
+		goto try_again;
+
+	*adv_credits = advertise;
+	return got;
+}
+
+void rds_iw_send_add_credits(struct rds_connection *conn, unsigned int credits)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+
+	if (credits == 0)
+		return;
+
+	rdsdebug("credits=%u current=%u%s\n",
+			credits,
+			IB_GET_SEND_CREDITS(atomic_read(&ic->i_credits)),
+			test_bit(RDS_LL_SEND_FULL, &conn->c_flags) ? ", ll_send_full" : "");
+
+	atomic_add(IB_SET_SEND_CREDITS(credits), &ic->i_credits);
+	if (test_and_clear_bit(RDS_LL_SEND_FULL, &conn->c_flags))
+		queue_delayed_work(rds_wq, &conn->c_send_w, 0);
+
+	WARN_ON(IB_GET_SEND_CREDITS(credits) >= 16384);
+
+	rds_iw_stats_inc(s_iw_rx_credit_updates);
+}
+
+void rds_iw_advertise_credits(struct rds_connection *conn, unsigned int posted)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+
+	if (posted == 0)
+		return;
+
+	atomic_add(IB_SET_POST_CREDITS(posted), &ic->i_credits);
+
+	/* Decide whether to send an update to the peer now.
+	 * If we would send a credit update for every single buffer we
+	 * post, we would end up with an ACK storm (ACK arrives,
+	 * consumes buffer, we refill the ring, send ACK to remote
+	 * advertising the newly posted buffer... ad inf)
+	 *
+	 * Performance pretty much depends on how often we send
+	 * credit updates - too frequent updates mean lots of ACKs.
+	 * Too infrequent updates, and the peer will run out of
+	 * credits and has to throttle.
+	 * For the time being, 16 seems to be a good compromise.
+	 */
+	if (IB_GET_POST_CREDITS(atomic_read(&ic->i_credits)) >= 16)
+		set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
+}
+
+static inline void
+rds_iw_xmit_populate_wr(struct rds_iw_connection *ic,
+		struct rds_iw_send_work *send, unsigned int pos,
+		unsigned long buffer, unsigned int length,
+		int send_flags)
+{
+	struct ib_sge *sge;
+
+	WARN_ON(pos != send - ic->i_sends);
+
+	send->s_wr.send_flags = send_flags;
+	send->s_wr.opcode = IB_WR_SEND;
+	send->s_wr.num_sge = 2;
+	send->s_wr.next = NULL;
+	send->s_queued = jiffies;
+	send->s_op = NULL;
+
+	if (length != 0) {
+		sge = rds_iw_data_sge(ic, send->s_sge);
+		sge->addr = buffer;
+		sge->length = length;
+		sge->lkey = rds_iw_local_dma_lkey(ic);
+
+		sge = rds_iw_header_sge(ic, send->s_sge);
+	} else {
+		/* We're sending a packet with no payload. There is only
+		 * one SGE */
+		send->s_wr.num_sge = 1;
+		sge = &send->s_sge[0];
+	}
+
+	sge->addr = ic->i_send_hdrs_dma + (pos * sizeof(struct rds_header));
+	sge->length = sizeof(struct rds_header);
+	sge->lkey = rds_iw_local_dma_lkey(ic);
+}
+
+/*
+ * This can be called multiple times for a given message.  The first time
+ * we see a message we map its scatterlist into the IB device so that
+ * we can provide that mapped address to the IB scatter gather entries
+ * in the IB work requests.  We translate the scatterlist into a series
+ * of work requests that fragment the message.  These work requests complete
+ * in order so we pass ownership of the message to the completion handler
+ * once we send the final fragment.
+ *
+ * The RDS core uses the c_send_lock to only enter this function once
+ * per connection.  This makes sure that the tx ring alloc/unalloc pairs
+ * don't get out of sync and confuse the ring.
+ */
+int rds_iw_xmit(struct rds_connection *conn, struct rds_message *rm,
+		unsigned int hdr_off, unsigned int sg, unsigned int off)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+	struct ib_device *dev = ic->i_cm_id->device;
+	struct rds_iw_send_work *send = NULL;
+	struct rds_iw_send_work *first;
+	struct rds_iw_send_work *prev;
+	struct ib_send_wr *failed_wr;
+	struct scatterlist *scat;
+	u32 pos;
+	u32 i;
+	u32 work_alloc;
+	u32 credit_alloc;
+	u32 posted;
+	u32 adv_credits = 0;
+	int send_flags = 0;
+	int sent;
+	int ret;
+	int flow_controlled = 0;
+
+	BUG_ON(off % RDS_FRAG_SIZE);
+	BUG_ON(hdr_off != 0 && hdr_off != sizeof(struct rds_header));
+
+	/* Fastreg support */
+	if (rds_rdma_cookie_key(rm->m_rdma_cookie) && !ic->i_fastreg_posted) {
+		ret = -EAGAIN;
+		goto out;
+	}
+
+	/* FIXME we may overallocate here */
+	if (be32_to_cpu(rm->m_inc.i_hdr.h_len) == 0)
+		i = 1;
+	else
+		i = ceil(be32_to_cpu(rm->m_inc.i_hdr.h_len), RDS_FRAG_SIZE);
+
+	work_alloc = rds_iw_ring_alloc(&ic->i_send_ring, i, &pos);
+	if (work_alloc == 0) {
+		set_bit(RDS_LL_SEND_FULL, &conn->c_flags);
+		rds_iw_stats_inc(s_iw_tx_ring_full);
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	credit_alloc = work_alloc;
+	if (ic->i_flowctl) {
+		credit_alloc = rds_iw_send_grab_credits(ic, work_alloc, &posted, 0, RDS_MAX_ADV_CREDIT);
+		adv_credits += posted;
+		if (credit_alloc < work_alloc) {
+			rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc - credit_alloc);
+			work_alloc = credit_alloc;
+			flow_controlled++;
+		}
+		if (work_alloc == 0) {
+			set_bit(RDS_LL_SEND_FULL, &conn->c_flags);
+			rds_iw_stats_inc(s_iw_tx_throttle);
+			ret = -ENOMEM;
+			goto out;
+		}
+	}
+
+	/* map the message the first time we see it */
+	if (!ic->i_rm) {
+		/*
+		printk(KERN_NOTICE "rds_iw_xmit prep msg dport=%u flags=0x%x len=%d\n",
+				be16_to_cpu(rm->m_inc.i_hdr.h_dport),
+				rm->m_inc.i_hdr.h_flags,
+				be32_to_cpu(rm->m_inc.i_hdr.h_len));
+		   */
+		if (rm->data.op_nents) {
+			rm->data.op_count = ib_dma_map_sg(dev,
+							  rm->data.op_sg,
+							  rm->data.op_nents,
+							  DMA_TO_DEVICE);
+			rdsdebug("ic %p mapping rm %p: %d\n", ic, rm, rm->data.op_count);
+			if (rm->data.op_count == 0) {
+				rds_iw_stats_inc(s_iw_tx_sg_mapping_failure);
+				rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
+				ret = -ENOMEM; /* XXX ? */
+				goto out;
+			}
+		} else {
+			rm->data.op_count = 0;
+		}
+
+		ic->i_unsignaled_wrs = rds_iw_sysctl_max_unsig_wrs;
+		ic->i_unsignaled_bytes = rds_iw_sysctl_max_unsig_bytes;
+		rds_message_addref(rm);
+		ic->i_rm = rm;
+
+		/* Finalize the header */
+		if (test_bit(RDS_MSG_ACK_REQUIRED, &rm->m_flags))
+			rm->m_inc.i_hdr.h_flags |= RDS_FLAG_ACK_REQUIRED;
+		if (test_bit(RDS_MSG_RETRANSMITTED, &rm->m_flags))
+			rm->m_inc.i_hdr.h_flags |= RDS_FLAG_RETRANSMITTED;
+
+		/* If it has a RDMA op, tell the peer we did it. This is
+		 * used by the peer to release use-once RDMA MRs. */
+		if (rm->rdma.op_active) {
+			struct rds_ext_header_rdma ext_hdr;
+
+			ext_hdr.h_rdma_rkey = cpu_to_be32(rm->rdma.op_rkey);
+			rds_message_add_extension(&rm->m_inc.i_hdr,
+					RDS_EXTHDR_RDMA, &ext_hdr, sizeof(ext_hdr));
+		}
+		if (rm->m_rdma_cookie) {
+			rds_message_add_rdma_dest_extension(&rm->m_inc.i_hdr,
+					rds_rdma_cookie_key(rm->m_rdma_cookie),
+					rds_rdma_cookie_offset(rm->m_rdma_cookie));
+		}
+
+		/* Note - rds_iw_piggyb_ack clears the ACK_REQUIRED bit, so
+		 * we should not do this unless we have a chance of at least
+		 * sticking the header into the send ring. Which is why we
+		 * should call rds_iw_ring_alloc first. */
+		rm->m_inc.i_hdr.h_ack = cpu_to_be64(rds_iw_piggyb_ack(ic));
+		rds_message_make_checksum(&rm->m_inc.i_hdr);
+
+		/*
+		 * Update adv_credits since we reset the ACK_REQUIRED bit.
+		 */
+		rds_iw_send_grab_credits(ic, 0, &posted, 1, RDS_MAX_ADV_CREDIT - adv_credits);
+		adv_credits += posted;
+		BUG_ON(adv_credits > 255);
+	}
+
+	send = &ic->i_sends[pos];
+	first = send;
+	prev = NULL;
+	scat = &rm->data.op_sg[sg];
+	sent = 0;
+	i = 0;
+
+	/* Sometimes you want to put a fence between an RDMA
+	 * READ and the following SEND.
+	 * We could either do this all the time
+	 * or when requested by the user. Right now, we let
+	 * the application choose.
+	 */
+	if (rm->rdma.op_active && rm->rdma.op_fence)
+		send_flags = IB_SEND_FENCE;
+
+	/*
+	 * We could be copying the header into the unused tail of the page.
+	 * That would need to be changed in the future when those pages might
+	 * be mapped userspace pages or page cache pages.  So instead we always
+	 * use a second sge and our long-lived ring of mapped headers.  We send
+	 * the header after the data so that the data payload can be aligned on
+	 * the receiver.
+	 */
+
+	/* handle a 0-len message */
+	if (be32_to_cpu(rm->m_inc.i_hdr.h_len) == 0) {
+		rds_iw_xmit_populate_wr(ic, send, pos, 0, 0, send_flags);
+		goto add_header;
+	}
+
+	/* if there's data reference it with a chain of work reqs */
+	for (; i < work_alloc && scat != &rm->data.op_sg[rm->data.op_count]; i++) {
+		unsigned int len;
+
+		send = &ic->i_sends[pos];
+
+		len = min(RDS_FRAG_SIZE, ib_sg_dma_len(dev, scat) - off);
+		rds_iw_xmit_populate_wr(ic, send, pos,
+				ib_sg_dma_address(dev, scat) + off, len,
+				send_flags);
+
+		/*
+		 * We want to delay signaling completions just enough to get
+		 * the batching benefits but not so much that we create dead time
+		 * on the wire.
+		 */
+		if (ic->i_unsignaled_wrs-- == 0) {
+			ic->i_unsignaled_wrs = rds_iw_sysctl_max_unsig_wrs;
+			send->s_wr.send_flags |= IB_SEND_SIGNALED | IB_SEND_SOLICITED;
+		}
+
+		ic->i_unsignaled_bytes -= len;
+		if (ic->i_unsignaled_bytes <= 0) {
+			ic->i_unsignaled_bytes = rds_iw_sysctl_max_unsig_bytes;
+			send->s_wr.send_flags |= IB_SEND_SIGNALED | IB_SEND_SOLICITED;
+		}
+
+		/*
+		 * Always signal the last one if we're stopping due to flow control.
+		 */
+		if (flow_controlled && i == (work_alloc-1))
+			send->s_wr.send_flags |= IB_SEND_SIGNALED | IB_SEND_SOLICITED;
+
+		rdsdebug("send %p wr %p num_sge %u next %p\n", send,
+			 &send->s_wr, send->s_wr.num_sge, send->s_wr.next);
+
+		sent += len;
+		off += len;
+		if (off == ib_sg_dma_len(dev, scat)) {
+			scat++;
+			off = 0;
+		}
+
+add_header:
+		/* Tack on the header after the data. The header SGE should already
+		 * have been set up to point to the right header buffer. */
+		memcpy(&ic->i_send_hdrs[pos], &rm->m_inc.i_hdr, sizeof(struct rds_header));
+
+		if (0) {
+			struct rds_header *hdr = &ic->i_send_hdrs[pos];
+
+			printk(KERN_NOTICE "send WR dport=%u flags=0x%x len=%d\n",
+				be16_to_cpu(hdr->h_dport),
+				hdr->h_flags,
+				be32_to_cpu(hdr->h_len));
+		}
+		if (adv_credits) {
+			struct rds_header *hdr = &ic->i_send_hdrs[pos];
+
+			/* add credit and redo the header checksum */
+			hdr->h_credit = adv_credits;
+			rds_message_make_checksum(hdr);
+			adv_credits = 0;
+			rds_iw_stats_inc(s_iw_tx_credit_updates);
+		}
+
+		if (prev)
+			prev->s_wr.next = &send->s_wr;
+		prev = send;
+
+		pos = (pos + 1) % ic->i_send_ring.w_nr;
+	}
+
+	/* Account the RDS header in the number of bytes we sent, but just once.
+	 * The caller has no concept of fragmentation. */
+	if (hdr_off == 0)
+		sent += sizeof(struct rds_header);
+
+	/* if we finished the message then send completion owns it */
+	if (scat == &rm->data.op_sg[rm->data.op_count]) {
+		prev->s_rm = ic->i_rm;
+		prev->s_wr.send_flags |= IB_SEND_SIGNALED | IB_SEND_SOLICITED;
+		ic->i_rm = NULL;
+	}
+
+	if (i < work_alloc) {
+		rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc - i);
+		work_alloc = i;
+	}
+	if (ic->i_flowctl && i < credit_alloc)
+		rds_iw_send_add_credits(conn, credit_alloc - i);
+
+	/* XXX need to worry about failed_wr and partial sends. */
+	failed_wr = &first->s_wr;
+	ret = ib_post_send(ic->i_cm_id->qp, &first->s_wr, &failed_wr);
+	rdsdebug("ic %p first %p (wr %p) ret %d wr %p\n", ic,
+		 first, &first->s_wr, ret, failed_wr);
+	BUG_ON(failed_wr != &first->s_wr);
+	if (ret) {
+		printk(KERN_WARNING "RDS/IW: ib_post_send to %pI4 "
+		       "returned %d\n", &conn->c_faddr, ret);
+		rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
+		if (prev->s_rm) {
+			ic->i_rm = prev->s_rm;
+			prev->s_rm = NULL;
+		}
+		goto out;
+	}
+
+	ret = sent;
+out:
+	BUG_ON(adv_credits);
+	return ret;
+}
+
+static void rds_iw_build_send_fastreg(struct rds_iw_device *rds_iwdev, struct rds_iw_connection *ic, struct rds_iw_send_work *send, int nent, int len, u64 sg_addr)
+{
+	BUG_ON(nent > send->s_page_list->max_page_list_len);
+	/*
+	 * Perform a WR for the fast_reg_mr. Each individual page
+	 * in the sg list is added to the fast reg page list and placed
+	 * inside the fast_reg_mr WR.
+	 */
+	send->s_wr.opcode = IB_WR_FAST_REG_MR;
+	send->s_wr.wr.fast_reg.length = len;
+	send->s_wr.wr.fast_reg.rkey = send->s_mr->rkey;
+	send->s_wr.wr.fast_reg.page_list = send->s_page_list;
+	send->s_wr.wr.fast_reg.page_list_len = nent;
+	send->s_wr.wr.fast_reg.page_shift = PAGE_SHIFT;
+	send->s_wr.wr.fast_reg.access_flags = IB_ACCESS_REMOTE_WRITE;
+	send->s_wr.wr.fast_reg.iova_start = sg_addr;
+
+	ib_update_fast_reg_key(send->s_mr, send->s_remap_count++);
+}
+
+int rds_iw_xmit_rdma(struct rds_connection *conn, struct rm_rdma_op *op)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+	struct rds_iw_send_work *send = NULL;
+	struct rds_iw_send_work *first;
+	struct rds_iw_send_work *prev;
+	struct ib_send_wr *failed_wr;
+	struct rds_iw_device *rds_iwdev;
+	struct scatterlist *scat;
+	unsigned long len;
+	u64 remote_addr = op->op_remote_addr;
+	u32 pos, fr_pos;
+	u32 work_alloc;
+	u32 i;
+	u32 j;
+	int sent;
+	int ret;
+	int num_sge;
+
+	rds_iwdev = ib_get_client_data(ic->i_cm_id->device, &rds_iw_client);
+
+	/* map the message the first time we see it */
+	if (!op->op_mapped) {
+		op->op_count = ib_dma_map_sg(ic->i_cm_id->device,
+					     op->op_sg, op->op_nents, (op->op_write) ?
+					     DMA_TO_DEVICE : DMA_FROM_DEVICE);
+		rdsdebug("ic %p mapping op %p: %d\n", ic, op, op->op_count);
+		if (op->op_count == 0) {
+			rds_iw_stats_inc(s_iw_tx_sg_mapping_failure);
+			ret = -ENOMEM; /* XXX ? */
+			goto out;
+		}
+
+		op->op_mapped = 1;
+	}
+
+	if (!op->op_write) {
+		/* Alloc space on the send queue for the fastreg */
+		work_alloc = rds_iw_ring_alloc(&ic->i_send_ring, 1, &fr_pos);
+		if (work_alloc != 1) {
+			rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
+			rds_iw_stats_inc(s_iw_tx_ring_full);
+			ret = -ENOMEM;
+			goto out;
+		}
+	}
+
+	/*
+	 * Instead of knowing how to return a partial rdma read/write we insist that there
+	 * be enough work requests to send the entire message.
+	 */
+	i = ceil(op->op_count, rds_iwdev->max_sge);
+
+	work_alloc = rds_iw_ring_alloc(&ic->i_send_ring, i, &pos);
+	if (work_alloc != i) {
+		rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
+		rds_iw_stats_inc(s_iw_tx_ring_full);
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	send = &ic->i_sends[pos];
+	if (!op->op_write) {
+		first = prev = &ic->i_sends[fr_pos];
+	} else {
+		first = send;
+		prev = NULL;
+	}
+	scat = &op->op_sg[0];
+	sent = 0;
+	num_sge = op->op_count;
+
+	for (i = 0; i < work_alloc && scat != &op->op_sg[op->op_count]; i++) {
+		send->s_wr.send_flags = 0;
+		send->s_queued = jiffies;
+
+		/*
+		 * We want to delay signaling completions just enough to get
+		 * the batching benefits but not so much that we create dead time on the wire.
+		 */
+		if (ic->i_unsignaled_wrs-- == 0) {
+			ic->i_unsignaled_wrs = rds_iw_sysctl_max_unsig_wrs;
+			send->s_wr.send_flags = IB_SEND_SIGNALED;
+		}
+
+		/* To avoid the need to have the plumbing to invalidate the fastreg_mr used
+		 * for local access after RDS is finished with it, using
+		 * IB_WR_RDMA_READ_WITH_INV will invalidate it after the read has completed.
+		 */
+		if (op->op_write)
+			send->s_wr.opcode = IB_WR_RDMA_WRITE;
+		else
+			send->s_wr.opcode = IB_WR_RDMA_READ_WITH_INV;
+
+		send->s_wr.wr.rdma.remote_addr = remote_addr;
+		send->s_wr.wr.rdma.rkey = op->op_rkey;
+		send->s_op = op;
+
+		if (num_sge > rds_iwdev->max_sge) {
+			send->s_wr.num_sge = rds_iwdev->max_sge;
+			num_sge -= rds_iwdev->max_sge;
+		} else
+			send->s_wr.num_sge = num_sge;
+
+		send->s_wr.next = NULL;
+
+		if (prev)
+			prev->s_wr.next = &send->s_wr;
+
+		for (j = 0; j < send->s_wr.num_sge && scat != &op->op_sg[op->op_count]; j++) {
+			len = ib_sg_dma_len(ic->i_cm_id->device, scat);
+
+			if (send->s_wr.opcode == IB_WR_RDMA_READ_WITH_INV)
+				send->s_page_list->page_list[j] = ib_sg_dma_address(ic->i_cm_id->device, scat);
+			else {
+				send->s_sge[j].addr = ib_sg_dma_address(ic->i_cm_id->device, scat);
+				send->s_sge[j].length = len;
+				send->s_sge[j].lkey = rds_iw_local_dma_lkey(ic);
+			}
+
+			sent += len;
+			rdsdebug("ic %p sent %d remote_addr %llu\n", ic, sent, remote_addr);
+			remote_addr += len;
+
+			scat++;
+		}
+
+		if (send->s_wr.opcode == IB_WR_RDMA_READ_WITH_INV) {
+			send->s_wr.num_sge = 1;
+			send->s_sge[0].addr = conn->c_xmit_rm->m_rs->rs_user_addr;
+			send->s_sge[0].length = conn->c_xmit_rm->m_rs->rs_user_bytes;
+			send->s_sge[0].lkey = ic->i_sends[fr_pos].s_mr->lkey;
+		}
+
+		rdsdebug("send %p wr %p num_sge %u next %p\n", send,
+			&send->s_wr, send->s_wr.num_sge, send->s_wr.next);
+
+		prev = send;
+		if (++send == &ic->i_sends[ic->i_send_ring.w_nr])
+			send = ic->i_sends;
+	}
+
+	/* if we finished the message then send completion owns it */
+	if (scat == &op->op_sg[op->op_count])
+		first->s_wr.send_flags = IB_SEND_SIGNALED;
+
+	if (i < work_alloc) {
+		rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc - i);
+		work_alloc = i;
+	}
+
+	/* On iWARP, local memory access by a remote system (ie, RDMA Read) is not
+	 * recommended.  Putting the lkey on the wire is a security hole, as it can
+	 * allow for memory access to all of memory on the remote system.  Some
+	 * adapters do not allow using the lkey for this at all.  To bypass this use a
+	 * fastreg_mr (or possibly a dma_mr)
+	 */
+	if (!op->op_write) {
+		rds_iw_build_send_fastreg(rds_iwdev, ic, &ic->i_sends[fr_pos],
+			op->op_count, sent, conn->c_xmit_rm->m_rs->rs_user_addr);
+		work_alloc++;
+	}
+
+	failed_wr = &first->s_wr;
+	ret = ib_post_send(ic->i_cm_id->qp, &first->s_wr, &failed_wr);
+	rdsdebug("ic %p first %p (wr %p) ret %d wr %p\n", ic,
+		 first, &first->s_wr, ret, failed_wr);
+	BUG_ON(failed_wr != &first->s_wr);
+	if (ret) {
+		printk(KERN_WARNING "RDS/IW: rdma ib_post_send to %pI4 "
+		       "returned %d\n", &conn->c_faddr, ret);
+		rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
+		goto out;
+	}
+
+out:
+	return ret;
+}
+
+void rds_iw_xmit_complete(struct rds_connection *conn)
+{
+	struct rds_iw_connection *ic = conn->c_transport_data;
+
+	/* We may have a pending ACK or window update we were unable
+	 * to send previously (due to flow control). Try again. */
+	rds_iw_attempt_ack(ic);
+}
diff --git a/net/rds/iw_stats.c b/net/rds/iw_stats.c
new file mode 100644
index 000000000..5fe67f6a1
--- /dev/null
+++ b/net/rds/iw_stats.c
@@ -0,0 +1,95 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/percpu.h>
+#include <linux/seq_file.h>
+#include <linux/proc_fs.h>
+
+#include "rds.h"
+#include "iw.h"
+
+DEFINE_PER_CPU_SHARED_ALIGNED(struct rds_iw_statistics, rds_iw_stats);
+
+static const char *const rds_iw_stat_names[] = {
+	"iw_connect_raced",
+	"iw_listen_closed_stale",
+	"iw_tx_cq_call",
+	"iw_tx_cq_event",
+	"iw_tx_ring_full",
+	"iw_tx_throttle",
+	"iw_tx_sg_mapping_failure",
+	"iw_tx_stalled",
+	"iw_tx_credit_updates",
+	"iw_rx_cq_call",
+	"iw_rx_cq_event",
+	"iw_rx_ring_empty",
+	"iw_rx_refill_from_cq",
+	"iw_rx_refill_from_thread",
+	"iw_rx_alloc_limit",
+	"iw_rx_credit_updates",
+	"iw_ack_sent",
+	"iw_ack_send_failure",
+	"iw_ack_send_delayed",
+	"iw_ack_send_piggybacked",
+	"iw_ack_received",
+	"iw_rdma_mr_alloc",
+	"iw_rdma_mr_free",
+	"iw_rdma_mr_used",
+	"iw_rdma_mr_pool_flush",
+	"iw_rdma_mr_pool_wait",
+	"iw_rdma_mr_pool_depleted",
+};
+
+unsigned int rds_iw_stats_info_copy(struct rds_info_iterator *iter,
+				    unsigned int avail)
+{
+	struct rds_iw_statistics stats = {0, };
+	uint64_t *src;
+	uint64_t *sum;
+	size_t i;
+	int cpu;
+
+	if (avail < ARRAY_SIZE(rds_iw_stat_names))
+		goto out;
+
+	for_each_online_cpu(cpu) {
+		src = (uint64_t *)&(per_cpu(rds_iw_stats, cpu));
+		sum = (uint64_t *)&stats;
+		for (i = 0; i < sizeof(stats) / sizeof(uint64_t); i++)
+			*(sum++) += *(src++);
+	}
+
+	rds_stats_info_copy(iter, (uint64_t *)&stats, rds_iw_stat_names,
+			    ARRAY_SIZE(rds_iw_stat_names));
+out:
+	return ARRAY_SIZE(rds_iw_stat_names);
+}
diff --git a/net/rds/iw_sysctl.c b/net/rds/iw_sysctl.c
new file mode 100644
index 000000000..139239d2c
--- /dev/null
+++ b/net/rds/iw_sysctl.c
@@ -0,0 +1,123 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/sysctl.h>
+#include <linux/proc_fs.h>
+
+#include "iw.h"
+
+static struct ctl_table_header *rds_iw_sysctl_hdr;
+
+unsigned long rds_iw_sysctl_max_send_wr = RDS_IW_DEFAULT_SEND_WR;
+unsigned long rds_iw_sysctl_max_recv_wr = RDS_IW_DEFAULT_RECV_WR;
+unsigned long rds_iw_sysctl_max_recv_allocation = (128 * 1024 * 1024) / RDS_FRAG_SIZE;
+static unsigned long rds_iw_sysctl_max_wr_min = 1;
+/* hardware will fail CQ creation long before this */
+static unsigned long rds_iw_sysctl_max_wr_max = (u32)~0;
+
+unsigned long rds_iw_sysctl_max_unsig_wrs = 16;
+static unsigned long rds_iw_sysctl_max_unsig_wr_min = 1;
+static unsigned long rds_iw_sysctl_max_unsig_wr_max = 64;
+
+unsigned long rds_iw_sysctl_max_unsig_bytes = (16 << 20);
+static unsigned long rds_iw_sysctl_max_unsig_bytes_min = 1;
+static unsigned long rds_iw_sysctl_max_unsig_bytes_max = ~0UL;
+
+unsigned int rds_iw_sysctl_flow_control = 1;
+
+static struct ctl_table rds_iw_sysctl_table[] = {
+	{
+		.procname       = "max_send_wr",
+		.data		= &rds_iw_sysctl_max_send_wr,
+		.maxlen         = sizeof(unsigned long),
+		.mode           = 0644,
+		.proc_handler   = proc_doulongvec_minmax,
+		.extra1		= &rds_iw_sysctl_max_wr_min,
+		.extra2		= &rds_iw_sysctl_max_wr_max,
+	},
+	{
+		.procname       = "max_recv_wr",
+		.data		= &rds_iw_sysctl_max_recv_wr,
+		.maxlen         = sizeof(unsigned long),
+		.mode           = 0644,
+		.proc_handler   = proc_doulongvec_minmax,
+		.extra1		= &rds_iw_sysctl_max_wr_min,
+		.extra2		= &rds_iw_sysctl_max_wr_max,
+	},
+	{
+		.procname       = "max_unsignaled_wr",
+		.data		= &rds_iw_sysctl_max_unsig_wrs,
+		.maxlen         = sizeof(unsigned long),
+		.mode           = 0644,
+		.proc_handler   = proc_doulongvec_minmax,
+		.extra1		= &rds_iw_sysctl_max_unsig_wr_min,
+		.extra2		= &rds_iw_sysctl_max_unsig_wr_max,
+	},
+	{
+		.procname       = "max_unsignaled_bytes",
+		.data		= &rds_iw_sysctl_max_unsig_bytes,
+		.maxlen         = sizeof(unsigned long),
+		.mode           = 0644,
+		.proc_handler   = proc_doulongvec_minmax,
+		.extra1		= &rds_iw_sysctl_max_unsig_bytes_min,
+		.extra2		= &rds_iw_sysctl_max_unsig_bytes_max,
+	},
+	{
+		.procname       = "max_recv_allocation",
+		.data		= &rds_iw_sysctl_max_recv_allocation,
+		.maxlen         = sizeof(unsigned long),
+		.mode           = 0644,
+		.proc_handler   = proc_doulongvec_minmax,
+	},
+	{
+		.procname	= "flow_control",
+		.data		= &rds_iw_sysctl_flow_control,
+		.maxlen		= sizeof(rds_iw_sysctl_flow_control),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec,
+	},
+	{ }
+};
+
+void rds_iw_sysctl_exit(void)
+{
+	unregister_net_sysctl_table(rds_iw_sysctl_hdr);
+}
+
+int rds_iw_sysctl_init(void)
+{
+	rds_iw_sysctl_hdr = register_net_sysctl(&init_net, "net/rds/iw", rds_iw_sysctl_table);
+	if (!rds_iw_sysctl_hdr)
+		return -ENOMEM;
+	return 0;
+}
diff --git a/net/rds/loop.c b/net/rds/loop.c
new file mode 100644
index 000000000..6b12b6854
--- /dev/null
+++ b/net/rds/loop.c
@@ -0,0 +1,194 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/in.h>
+
+#include "rds.h"
+#include "loop.h"
+
+static DEFINE_SPINLOCK(loop_conns_lock);
+static LIST_HEAD(loop_conns);
+
+/*
+ * This 'loopback' transport is a special case for flows that originate
+ * and terminate on the same machine.
+ *
+ * Connection build-up notices if the destination address is thought of
+ * as a local address by a transport.  At that time it decides to use the
+ * loopback transport instead of the bound transport of the sending socket.
+ *
+ * The loopback transport's sending path just hands the sent rds_message
+ * straight to the receiving path via an embedded rds_incoming.
+ */
+
+/*
+ * Usually a message transits both the sender and receiver's conns as it
+ * flows to the receiver.  In the loopback case, though, the receive path
+ * is handed the sending conn so the sense of the addresses is reversed.
+ */
+static int rds_loop_xmit(struct rds_connection *conn, struct rds_message *rm,
+			 unsigned int hdr_off, unsigned int sg,
+			 unsigned int off)
+{
+	struct scatterlist *sgp = &rm->data.op_sg[sg];
+	int ret = sizeof(struct rds_header) +
+			be32_to_cpu(rm->m_inc.i_hdr.h_len);
+
+	/* Do not send cong updates to loopback */
+	if (rm->m_inc.i_hdr.h_flags & RDS_FLAG_CONG_BITMAP) {
+		rds_cong_map_updated(conn->c_fcong, ~(u64) 0);
+		ret = min_t(int, ret, sgp->length - conn->c_xmit_data_off);
+		goto out;
+	}
+
+	BUG_ON(hdr_off || sg || off);
+
+	rds_inc_init(&rm->m_inc, conn, conn->c_laddr);
+	/* For the embedded inc. Matching put is in loop_inc_free() */
+	rds_message_addref(rm);
+
+	rds_recv_incoming(conn, conn->c_laddr, conn->c_faddr, &rm->m_inc,
+			  GFP_KERNEL);
+
+	rds_send_drop_acked(conn, be64_to_cpu(rm->m_inc.i_hdr.h_sequence),
+			    NULL);
+
+	rds_inc_put(&rm->m_inc);
+out:
+	return ret;
+}
+
+/*
+ * See rds_loop_xmit(). Since our inc is embedded in the rm, we
+ * make sure the rm lives at least until the inc is done.
+ */
+static void rds_loop_inc_free(struct rds_incoming *inc)
+{
+        struct rds_message *rm = container_of(inc, struct rds_message, m_inc);
+        rds_message_put(rm);
+}
+
+/* we need to at least give the thread something to succeed */
+static int rds_loop_recv(struct rds_connection *conn)
+{
+	return 0;
+}
+
+struct rds_loop_connection {
+	struct list_head loop_node;
+	struct rds_connection *conn;
+};
+
+/*
+ * Even the loopback transport needs to keep track of its connections,
+ * so it can call rds_conn_destroy() on them on exit. N.B. there are
+ * 1+ loopback addresses (127.*.*.*) so it's not a bug to have
+ * multiple loopback conns allocated, although rather useless.
+ */
+static int rds_loop_conn_alloc(struct rds_connection *conn, gfp_t gfp)
+{
+	struct rds_loop_connection *lc;
+	unsigned long flags;
+
+	lc = kzalloc(sizeof(struct rds_loop_connection), gfp);
+	if (!lc)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&lc->loop_node);
+	lc->conn = conn;
+	conn->c_transport_data = lc;
+
+	spin_lock_irqsave(&loop_conns_lock, flags);
+	list_add_tail(&lc->loop_node, &loop_conns);
+	spin_unlock_irqrestore(&loop_conns_lock, flags);
+
+	return 0;
+}
+
+static void rds_loop_conn_free(void *arg)
+{
+	struct rds_loop_connection *lc = arg;
+	unsigned long flags;
+
+	rdsdebug("lc %p\n", lc);
+	spin_lock_irqsave(&loop_conns_lock, flags);
+	list_del(&lc->loop_node);
+	spin_unlock_irqrestore(&loop_conns_lock, flags);
+	kfree(lc);
+}
+
+static int rds_loop_conn_connect(struct rds_connection *conn)
+{
+	rds_connect_complete(conn);
+	return 0;
+}
+
+static void rds_loop_conn_shutdown(struct rds_connection *conn)
+{
+}
+
+void rds_loop_exit(void)
+{
+	struct rds_loop_connection *lc, *_lc;
+	LIST_HEAD(tmp_list);
+
+	/* avoid calling conn_destroy with irqs off */
+	spin_lock_irq(&loop_conns_lock);
+	list_splice(&loop_conns, &tmp_list);
+	INIT_LIST_HEAD(&loop_conns);
+	spin_unlock_irq(&loop_conns_lock);
+
+	list_for_each_entry_safe(lc, _lc, &tmp_list, loop_node) {
+		WARN_ON(lc->conn->c_passive);
+		rds_conn_destroy(lc->conn);
+	}
+}
+
+/*
+ * This is missing .xmit_* because loop doesn't go through generic
+ * rds_send_xmit() and doesn't call rds_recv_incoming().  .listen_stop and
+ * .laddr_check are missing because transport.c doesn't iterate over
+ * rds_loop_transport.
+ */
+struct rds_transport rds_loop_transport = {
+	.xmit			= rds_loop_xmit,
+	.recv			= rds_loop_recv,
+	.conn_alloc		= rds_loop_conn_alloc,
+	.conn_free		= rds_loop_conn_free,
+	.conn_connect		= rds_loop_conn_connect,
+	.conn_shutdown		= rds_loop_conn_shutdown,
+	.inc_copy_to_user	= rds_message_inc_copy_to_user,
+	.inc_free		= rds_loop_inc_free,
+	.t_name			= "loopback",
+};
diff --git a/net/rds/loop.h b/net/rds/loop.h
new file mode 100644
index 000000000..f32b0939a
--- /dev/null
+++ b/net/rds/loop.h
@@ -0,0 +1,9 @@
+#ifndef _RDS_LOOP_H
+#define _RDS_LOOP_H
+
+/* loop.c */
+extern struct rds_transport rds_loop_transport;
+
+void rds_loop_exit(void);
+
+#endif
diff --git a/net/rds/message.c b/net/rds/message.c
new file mode 100644
index 000000000..756c73729
--- /dev/null
+++ b/net/rds/message.c
@@ -0,0 +1,366 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/export.h>
+
+#include "rds.h"
+
+static unsigned int	rds_exthdr_size[__RDS_EXTHDR_MAX] = {
+[RDS_EXTHDR_NONE]	= 0,
+[RDS_EXTHDR_VERSION]	= sizeof(struct rds_ext_header_version),
+[RDS_EXTHDR_RDMA]	= sizeof(struct rds_ext_header_rdma),
+[RDS_EXTHDR_RDMA_DEST]	= sizeof(struct rds_ext_header_rdma_dest),
+};
+
+
+void rds_message_addref(struct rds_message *rm)
+{
+	rdsdebug("addref rm %p ref %d\n", rm, atomic_read(&rm->m_refcount));
+	atomic_inc(&rm->m_refcount);
+}
+EXPORT_SYMBOL_GPL(rds_message_addref);
+
+/*
+ * This relies on dma_map_sg() not touching sg[].page during merging.
+ */
+static void rds_message_purge(struct rds_message *rm)
+{
+	unsigned long i;
+
+	if (unlikely(test_bit(RDS_MSG_PAGEVEC, &rm->m_flags)))
+		return;
+
+	for (i = 0; i < rm->data.op_nents; i++) {
+		rdsdebug("putting data page %p\n", (void *)sg_page(&rm->data.op_sg[i]));
+		/* XXX will have to put_page for page refs */
+		__free_page(sg_page(&rm->data.op_sg[i]));
+	}
+	rm->data.op_nents = 0;
+
+	if (rm->rdma.op_active)
+		rds_rdma_free_op(&rm->rdma);
+	if (rm->rdma.op_rdma_mr)
+		rds_mr_put(rm->rdma.op_rdma_mr);
+
+	if (rm->atomic.op_active)
+		rds_atomic_free_op(&rm->atomic);
+	if (rm->atomic.op_rdma_mr)
+		rds_mr_put(rm->atomic.op_rdma_mr);
+}
+
+void rds_message_put(struct rds_message *rm)
+{
+	rdsdebug("put rm %p ref %d\n", rm, atomic_read(&rm->m_refcount));
+	WARN(!atomic_read(&rm->m_refcount), "danger refcount zero on %p\n", rm);
+	if (atomic_dec_and_test(&rm->m_refcount)) {
+		BUG_ON(!list_empty(&rm->m_sock_item));
+		BUG_ON(!list_empty(&rm->m_conn_item));
+		rds_message_purge(rm);
+
+		kfree(rm);
+	}
+}
+EXPORT_SYMBOL_GPL(rds_message_put);
+
+void rds_message_populate_header(struct rds_header *hdr, __be16 sport,
+				 __be16 dport, u64 seq)
+{
+	hdr->h_flags = 0;
+	hdr->h_sport = sport;
+	hdr->h_dport = dport;
+	hdr->h_sequence = cpu_to_be64(seq);
+	hdr->h_exthdr[0] = RDS_EXTHDR_NONE;
+}
+EXPORT_SYMBOL_GPL(rds_message_populate_header);
+
+int rds_message_add_extension(struct rds_header *hdr, unsigned int type,
+			      const void *data, unsigned int len)
+{
+	unsigned int ext_len = sizeof(u8) + len;
+	unsigned char *dst;
+
+	/* For now, refuse to add more than one extension header */
+	if (hdr->h_exthdr[0] != RDS_EXTHDR_NONE)
+		return 0;
+
+	if (type >= __RDS_EXTHDR_MAX || len != rds_exthdr_size[type])
+		return 0;
+
+	if (ext_len >= RDS_HEADER_EXT_SPACE)
+		return 0;
+	dst = hdr->h_exthdr;
+
+	*dst++ = type;
+	memcpy(dst, data, len);
+
+	dst[len] = RDS_EXTHDR_NONE;
+	return 1;
+}
+EXPORT_SYMBOL_GPL(rds_message_add_extension);
+
+/*
+ * If a message has extension headers, retrieve them here.
+ * Call like this:
+ *
+ * unsigned int pos = 0;
+ *
+ * while (1) {
+ *	buflen = sizeof(buffer);
+ *	type = rds_message_next_extension(hdr, &pos, buffer, &buflen);
+ *	if (type == RDS_EXTHDR_NONE)
+ *		break;
+ *	...
+ * }
+ */
+int rds_message_next_extension(struct rds_header *hdr,
+		unsigned int *pos, void *buf, unsigned int *buflen)
+{
+	unsigned int offset, ext_type, ext_len;
+	u8 *src = hdr->h_exthdr;
+
+	offset = *pos;
+	if (offset >= RDS_HEADER_EXT_SPACE)
+		goto none;
+
+	/* Get the extension type and length. For now, the
+	 * length is implied by the extension type. */
+	ext_type = src[offset++];
+
+	if (ext_type == RDS_EXTHDR_NONE || ext_type >= __RDS_EXTHDR_MAX)
+		goto none;
+	ext_len = rds_exthdr_size[ext_type];
+	if (offset + ext_len > RDS_HEADER_EXT_SPACE)
+		goto none;
+
+	*pos = offset + ext_len;
+	if (ext_len < *buflen)
+		*buflen = ext_len;
+	memcpy(buf, src + offset, *buflen);
+	return ext_type;
+
+none:
+	*pos = RDS_HEADER_EXT_SPACE;
+	*buflen = 0;
+	return RDS_EXTHDR_NONE;
+}
+
+int rds_message_add_rdma_dest_extension(struct rds_header *hdr, u32 r_key, u32 offset)
+{
+	struct rds_ext_header_rdma_dest ext_hdr;
+
+	ext_hdr.h_rdma_rkey = cpu_to_be32(r_key);
+	ext_hdr.h_rdma_offset = cpu_to_be32(offset);
+	return rds_message_add_extension(hdr, RDS_EXTHDR_RDMA_DEST, &ext_hdr, sizeof(ext_hdr));
+}
+EXPORT_SYMBOL_GPL(rds_message_add_rdma_dest_extension);
+
+/*
+ * Each rds_message is allocated with extra space for the scatterlist entries
+ * rds ops will need. This is to minimize memory allocation count. Then, each rds op
+ * can grab SGs when initializing its part of the rds_message.
+ */
+struct rds_message *rds_message_alloc(unsigned int extra_len, gfp_t gfp)
+{
+	struct rds_message *rm;
+
+	if (extra_len > KMALLOC_MAX_SIZE - sizeof(struct rds_message))
+		return NULL;
+
+	rm = kzalloc(sizeof(struct rds_message) + extra_len, gfp);
+	if (!rm)
+		goto out;
+
+	rm->m_used_sgs = 0;
+	rm->m_total_sgs = extra_len / sizeof(struct scatterlist);
+
+	atomic_set(&rm->m_refcount, 1);
+	INIT_LIST_HEAD(&rm->m_sock_item);
+	INIT_LIST_HEAD(&rm->m_conn_item);
+	spin_lock_init(&rm->m_rs_lock);
+	init_waitqueue_head(&rm->m_flush_wait);
+
+out:
+	return rm;
+}
+
+/*
+ * RDS ops use this to grab SG entries from the rm's sg pool.
+ */
+struct scatterlist *rds_message_alloc_sgs(struct rds_message *rm, int nents)
+{
+	struct scatterlist *sg_first = (struct scatterlist *) &rm[1];
+	struct scatterlist *sg_ret;
+
+	WARN_ON(rm->m_used_sgs + nents > rm->m_total_sgs);
+	WARN_ON(!nents);
+
+	if (rm->m_used_sgs + nents > rm->m_total_sgs)
+		return NULL;
+
+	sg_ret = &sg_first[rm->m_used_sgs];
+	sg_init_table(sg_ret, nents);
+	rm->m_used_sgs += nents;
+
+	return sg_ret;
+}
+
+struct rds_message *rds_message_map_pages(unsigned long *page_addrs, unsigned int total_len)
+{
+	struct rds_message *rm;
+	unsigned int i;
+	int num_sgs = ceil(total_len, PAGE_SIZE);
+	int extra_bytes = num_sgs * sizeof(struct scatterlist);
+
+	rm = rds_message_alloc(extra_bytes, GFP_NOWAIT);
+	if (!rm)
+		return ERR_PTR(-ENOMEM);
+
+	set_bit(RDS_MSG_PAGEVEC, &rm->m_flags);
+	rm->m_inc.i_hdr.h_len = cpu_to_be32(total_len);
+	rm->data.op_nents = ceil(total_len, PAGE_SIZE);
+	rm->data.op_sg = rds_message_alloc_sgs(rm, num_sgs);
+	if (!rm->data.op_sg) {
+		rds_message_put(rm);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	for (i = 0; i < rm->data.op_nents; ++i) {
+		sg_set_page(&rm->data.op_sg[i],
+				virt_to_page(page_addrs[i]),
+				PAGE_SIZE, 0);
+	}
+
+	return rm;
+}
+
+int rds_message_copy_from_user(struct rds_message *rm, struct iov_iter *from)
+{
+	unsigned long to_copy, nbytes;
+	unsigned long sg_off;
+	struct scatterlist *sg;
+	int ret = 0;
+
+	rm->m_inc.i_hdr.h_len = cpu_to_be32(iov_iter_count(from));
+
+	/*
+	 * now allocate and copy in the data payload.
+	 */
+	sg = rm->data.op_sg;
+	sg_off = 0; /* Dear gcc, sg->page will be null from kzalloc. */
+
+	while (iov_iter_count(from)) {
+		if (!sg_page(sg)) {
+			ret = rds_page_remainder_alloc(sg, iov_iter_count(from),
+						       GFP_HIGHUSER);
+			if (ret)
+				return ret;
+			rm->data.op_nents++;
+			sg_off = 0;
+		}
+
+		to_copy = min_t(unsigned long, iov_iter_count(from),
+				sg->length - sg_off);
+
+		rds_stats_add(s_copy_from_user, to_copy);
+		nbytes = copy_page_from_iter(sg_page(sg), sg->offset + sg_off,
+					     to_copy, from);
+		if (nbytes != to_copy)
+			return -EFAULT;
+
+		sg_off += to_copy;
+
+		if (sg_off == sg->length)
+			sg++;
+	}
+
+	return ret;
+}
+
+int rds_message_inc_copy_to_user(struct rds_incoming *inc, struct iov_iter *to)
+{
+	struct rds_message *rm;
+	struct scatterlist *sg;
+	unsigned long to_copy;
+	unsigned long vec_off;
+	int copied;
+	int ret;
+	u32 len;
+
+	rm = container_of(inc, struct rds_message, m_inc);
+	len = be32_to_cpu(rm->m_inc.i_hdr.h_len);
+
+	sg = rm->data.op_sg;
+	vec_off = 0;
+	copied = 0;
+
+	while (iov_iter_count(to) && copied < len) {
+		to_copy = min_t(unsigned long, iov_iter_count(to),
+				sg->length - vec_off);
+		to_copy = min_t(unsigned long, to_copy, len - copied);
+
+		rds_stats_add(s_copy_to_user, to_copy);
+		ret = copy_page_to_iter(sg_page(sg), sg->offset + vec_off,
+					to_copy, to);
+		if (ret != to_copy)
+			return -EFAULT;
+
+		vec_off += to_copy;
+		copied += to_copy;
+
+		if (vec_off == sg->length) {
+			vec_off = 0;
+			sg++;
+		}
+	}
+
+	return copied;
+}
+
+/*
+ * If the message is still on the send queue, wait until the transport
+ * is done with it. This is particularly important for RDMA operations.
+ */
+void rds_message_wait(struct rds_message *rm)
+{
+	wait_event_interruptible(rm->m_flush_wait,
+			!test_bit(RDS_MSG_MAPPED, &rm->m_flags));
+}
+
+void rds_message_unmapped(struct rds_message *rm)
+{
+	clear_bit(RDS_MSG_MAPPED, &rm->m_flags);
+	wake_up_interruptible(&rm->m_flush_wait);
+}
+EXPORT_SYMBOL_GPL(rds_message_unmapped);
+
diff --git a/net/rds/page.c b/net/rds/page.c
new file mode 100644
index 000000000..9005a2c92
--- /dev/null
+++ b/net/rds/page.c
@@ -0,0 +1,215 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/highmem.h>
+#include <linux/gfp.h>
+#include <linux/cpu.h>
+#include <linux/export.h>
+
+#include "rds.h"
+
+struct rds_page_remainder {
+	struct page	*r_page;
+	unsigned long	r_offset;
+};
+
+static DEFINE_PER_CPU_SHARED_ALIGNED(struct rds_page_remainder,
+				     rds_page_remainders);
+
+/*
+ * returns 0 on success or -errno on failure.
+ *
+ * We don't have to worry about flush_dcache_page() as this only works
+ * with private pages.  If, say, we were to do directed receive to pinned
+ * user pages we'd have to worry more about cache coherence.  (Though
+ * the flush_dcache_page() in get_user_pages() would probably be enough).
+ */
+int rds_page_copy_user(struct page *page, unsigned long offset,
+		       void __user *ptr, unsigned long bytes,
+		       int to_user)
+{
+	unsigned long ret;
+	void *addr;
+
+	addr = kmap(page);
+	if (to_user) {
+		rds_stats_add(s_copy_to_user, bytes);
+		ret = copy_to_user(ptr, addr + offset, bytes);
+	} else {
+		rds_stats_add(s_copy_from_user, bytes);
+		ret = copy_from_user(addr + offset, ptr, bytes);
+	}
+	kunmap(page);
+
+	return ret ? -EFAULT : 0;
+}
+EXPORT_SYMBOL_GPL(rds_page_copy_user);
+
+/**
+ * rds_page_remainder_alloc - build up regions of a message.
+ *
+ * @scat: Scatter list for message
+ * @bytes: the number of bytes needed.
+ * @gfp: the waiting behaviour of the allocation
+ *
+ * @gfp is always ored with __GFP_HIGHMEM.  Callers must be prepared to
+ * kmap the pages, etc.
+ *
+ * If @bytes is at least a full page then this just returns a page from
+ * alloc_page().
+ *
+ * If @bytes is a partial page then this stores the unused region of the
+ * page in a per-cpu structure.  Future partial-page allocations may be
+ * satisfied from that cached region.  This lets us waste less memory on
+ * small allocations with minimal complexity.  It works because the transmit
+ * path passes read-only page regions down to devices.  They hold a page
+ * reference until they are done with the region.
+ */
+int rds_page_remainder_alloc(struct scatterlist *scat, unsigned long bytes,
+			     gfp_t gfp)
+{
+	struct rds_page_remainder *rem;
+	unsigned long flags;
+	struct page *page;
+	int ret;
+
+	gfp |= __GFP_HIGHMEM;
+
+	/* jump straight to allocation if we're trying for a huge page */
+	if (bytes >= PAGE_SIZE) {
+		page = alloc_page(gfp);
+		if (!page) {
+			ret = -ENOMEM;
+		} else {
+			sg_set_page(scat, page, PAGE_SIZE, 0);
+			ret = 0;
+		}
+		goto out;
+	}
+
+	rem = &per_cpu(rds_page_remainders, get_cpu());
+	local_irq_save(flags);
+
+	while (1) {
+		/* avoid a tiny region getting stuck by tossing it */
+		if (rem->r_page && bytes > (PAGE_SIZE - rem->r_offset)) {
+			rds_stats_inc(s_page_remainder_miss);
+			__free_page(rem->r_page);
+			rem->r_page = NULL;
+		}
+
+		/* hand out a fragment from the cached page */
+		if (rem->r_page && bytes <= (PAGE_SIZE - rem->r_offset)) {
+			sg_set_page(scat, rem->r_page, bytes, rem->r_offset);
+			get_page(sg_page(scat));
+
+			if (rem->r_offset != 0)
+				rds_stats_inc(s_page_remainder_hit);
+
+			rem->r_offset += bytes;
+			if (rem->r_offset == PAGE_SIZE) {
+				__free_page(rem->r_page);
+				rem->r_page = NULL;
+			}
+			ret = 0;
+			break;
+		}
+
+		/* alloc if there is nothing for us to use */
+		local_irq_restore(flags);
+		put_cpu();
+
+		page = alloc_page(gfp);
+
+		rem = &per_cpu(rds_page_remainders, get_cpu());
+		local_irq_save(flags);
+
+		if (!page) {
+			ret = -ENOMEM;
+			break;
+		}
+
+		/* did someone race to fill the remainder before us? */
+		if (rem->r_page) {
+			__free_page(page);
+			continue;
+		}
+
+		/* otherwise install our page and loop around to alloc */
+		rem->r_page = page;
+		rem->r_offset = 0;
+	}
+
+	local_irq_restore(flags);
+	put_cpu();
+out:
+	rdsdebug("bytes %lu ret %d %p %u %u\n", bytes, ret,
+		 ret ? NULL : sg_page(scat), ret ? 0 : scat->offset,
+		 ret ? 0 : scat->length);
+	return ret;
+}
+EXPORT_SYMBOL_GPL(rds_page_remainder_alloc);
+
+static int rds_page_remainder_cpu_notify(struct notifier_block *self,
+					 unsigned long action, void *hcpu)
+{
+	struct rds_page_remainder *rem;
+	long cpu = (long)hcpu;
+
+	rem = &per_cpu(rds_page_remainders, cpu);
+
+	rdsdebug("cpu %ld action 0x%lx\n", cpu, action);
+
+	switch (action) {
+	case CPU_DEAD:
+		if (rem->r_page)
+			__free_page(rem->r_page);
+		rem->r_page = NULL;
+		break;
+	}
+
+	return 0;
+}
+
+static struct notifier_block rds_page_remainder_nb = {
+	.notifier_call = rds_page_remainder_cpu_notify,
+};
+
+void rds_page_exit(void)
+{
+	int i;
+
+	for_each_possible_cpu(i)
+		rds_page_remainder_cpu_notify(&rds_page_remainder_nb,
+					      (unsigned long)CPU_DEAD,
+					      (void *)(long)i);
+}
diff --git a/net/rds/rdma.c b/net/rds/rdma.c
new file mode 100644
index 000000000..40084d843
--- /dev/null
+++ b/net/rds/rdma.c
@@ -0,0 +1,859 @@
+/*
+ * Copyright (c) 2007 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/pagemap.h>
+#include <linux/slab.h>
+#include <linux/rbtree.h>
+#include <linux/dma-mapping.h> /* for DMA_*_DEVICE */
+
+#include "rds.h"
+
+/*
+ * XXX
+ *  - build with sparse
+ *  - should we limit the size of a mr region?  let transport return failure?
+ *  - should we detect duplicate keys on a socket?  hmm.
+ *  - an rdma is an mlock, apply rlimit?
+ */
+
+/*
+ * get the number of pages by looking at the page indices that the start and
+ * end addresses fall in.
+ *
+ * Returns 0 if the vec is invalid.  It is invalid if the number of bytes
+ * causes the address to wrap or overflows an unsigned int.  This comes
+ * from being stored in the 'length' member of 'struct scatterlist'.
+ */
+static unsigned int rds_pages_in_vec(struct rds_iovec *vec)
+{
+	if ((vec->addr + vec->bytes <= vec->addr) ||
+	    (vec->bytes > (u64)UINT_MAX))
+		return 0;
+
+	return ((vec->addr + vec->bytes + PAGE_SIZE - 1) >> PAGE_SHIFT) -
+		(vec->addr >> PAGE_SHIFT);
+}
+
+static struct rds_mr *rds_mr_tree_walk(struct rb_root *root, u64 key,
+				       struct rds_mr *insert)
+{
+	struct rb_node **p = &root->rb_node;
+	struct rb_node *parent = NULL;
+	struct rds_mr *mr;
+
+	while (*p) {
+		parent = *p;
+		mr = rb_entry(parent, struct rds_mr, r_rb_node);
+
+		if (key < mr->r_key)
+			p = &(*p)->rb_left;
+		else if (key > mr->r_key)
+			p = &(*p)->rb_right;
+		else
+			return mr;
+	}
+
+	if (insert) {
+		rb_link_node(&insert->r_rb_node, parent, p);
+		rb_insert_color(&insert->r_rb_node, root);
+		atomic_inc(&insert->r_refcount);
+	}
+	return NULL;
+}
+
+/*
+ * Destroy the transport-specific part of a MR.
+ */
+static void rds_destroy_mr(struct rds_mr *mr)
+{
+	struct rds_sock *rs = mr->r_sock;
+	void *trans_private = NULL;
+	unsigned long flags;
+
+	rdsdebug("RDS: destroy mr key is %x refcnt %u\n",
+			mr->r_key, atomic_read(&mr->r_refcount));
+
+	if (test_and_set_bit(RDS_MR_DEAD, &mr->r_state))
+		return;
+
+	spin_lock_irqsave(&rs->rs_rdma_lock, flags);
+	if (!RB_EMPTY_NODE(&mr->r_rb_node))
+		rb_erase(&mr->r_rb_node, &rs->rs_rdma_keys);
+	trans_private = mr->r_trans_private;
+	mr->r_trans_private = NULL;
+	spin_unlock_irqrestore(&rs->rs_rdma_lock, flags);
+
+	if (trans_private)
+		mr->r_trans->free_mr(trans_private, mr->r_invalidate);
+}
+
+void __rds_put_mr_final(struct rds_mr *mr)
+{
+	rds_destroy_mr(mr);
+	kfree(mr);
+}
+
+/*
+ * By the time this is called we can't have any more ioctls called on
+ * the socket so we don't need to worry about racing with others.
+ */
+void rds_rdma_drop_keys(struct rds_sock *rs)
+{
+	struct rds_mr *mr;
+	struct rb_node *node;
+	unsigned long flags;
+
+	/* Release any MRs associated with this socket */
+	spin_lock_irqsave(&rs->rs_rdma_lock, flags);
+	while ((node = rb_first(&rs->rs_rdma_keys))) {
+		mr = container_of(node, struct rds_mr, r_rb_node);
+		if (mr->r_trans == rs->rs_transport)
+			mr->r_invalidate = 0;
+		rb_erase(&mr->r_rb_node, &rs->rs_rdma_keys);
+		RB_CLEAR_NODE(&mr->r_rb_node);
+		spin_unlock_irqrestore(&rs->rs_rdma_lock, flags);
+		rds_destroy_mr(mr);
+		rds_mr_put(mr);
+		spin_lock_irqsave(&rs->rs_rdma_lock, flags);
+	}
+	spin_unlock_irqrestore(&rs->rs_rdma_lock, flags);
+
+	if (rs->rs_transport && rs->rs_transport->flush_mrs)
+		rs->rs_transport->flush_mrs();
+}
+
+/*
+ * Helper function to pin user pages.
+ */
+static int rds_pin_pages(unsigned long user_addr, unsigned int nr_pages,
+			struct page **pages, int write)
+{
+	int ret;
+
+	ret = get_user_pages_fast(user_addr, nr_pages, write, pages);
+
+	if (ret >= 0 && ret < nr_pages) {
+		while (ret--)
+			put_page(pages[ret]);
+		ret = -EFAULT;
+	}
+
+	return ret;
+}
+
+static int __rds_rdma_map(struct rds_sock *rs, struct rds_get_mr_args *args,
+				u64 *cookie_ret, struct rds_mr **mr_ret)
+{
+	struct rds_mr *mr = NULL, *found;
+	unsigned int nr_pages;
+	struct page **pages = NULL;
+	struct scatterlist *sg;
+	void *trans_private;
+	unsigned long flags;
+	rds_rdma_cookie_t cookie;
+	unsigned int nents;
+	long i;
+	int ret;
+
+	if (rs->rs_bound_addr == 0) {
+		ret = -ENOTCONN; /* XXX not a great errno */
+		goto out;
+	}
+
+	if (!rs->rs_transport->get_mr) {
+		ret = -EOPNOTSUPP;
+		goto out;
+	}
+
+	nr_pages = rds_pages_in_vec(&args->vec);
+	if (nr_pages == 0) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	rdsdebug("RDS: get_mr addr %llx len %llu nr_pages %u\n",
+		args->vec.addr, args->vec.bytes, nr_pages);
+
+	/* XXX clamp nr_pages to limit the size of this alloc? */
+	pages = kcalloc(nr_pages, sizeof(struct page *), GFP_KERNEL);
+	if (!pages) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	mr = kzalloc(sizeof(struct rds_mr), GFP_KERNEL);
+	if (!mr) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	atomic_set(&mr->r_refcount, 1);
+	RB_CLEAR_NODE(&mr->r_rb_node);
+	mr->r_trans = rs->rs_transport;
+	mr->r_sock = rs;
+
+	if (args->flags & RDS_RDMA_USE_ONCE)
+		mr->r_use_once = 1;
+	if (args->flags & RDS_RDMA_INVALIDATE)
+		mr->r_invalidate = 1;
+	if (args->flags & RDS_RDMA_READWRITE)
+		mr->r_write = 1;
+
+	/*
+	 * Pin the pages that make up the user buffer and transfer the page
+	 * pointers to the mr's sg array.  We check to see if we've mapped
+	 * the whole region after transferring the partial page references
+	 * to the sg array so that we can have one page ref cleanup path.
+	 *
+	 * For now we have no flag that tells us whether the mapping is
+	 * r/o or r/w. We need to assume r/w, or we'll do a lot of RDMA to
+	 * the zero page.
+	 */
+	ret = rds_pin_pages(args->vec.addr, nr_pages, pages, 1);
+	if (ret < 0)
+		goto out;
+
+	nents = ret;
+	sg = kcalloc(nents, sizeof(*sg), GFP_KERNEL);
+	if (!sg) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	WARN_ON(!nents);
+	sg_init_table(sg, nents);
+
+	/* Stick all pages into the scatterlist */
+	for (i = 0 ; i < nents; i++)
+		sg_set_page(&sg[i], pages[i], PAGE_SIZE, 0);
+
+	rdsdebug("RDS: trans_private nents is %u\n", nents);
+
+	/* Obtain a transport specific MR. If this succeeds, the
+	 * s/g list is now owned by the MR.
+	 * Note that dma_map() implies that pending writes are
+	 * flushed to RAM, so no dma_sync is needed here. */
+	trans_private = rs->rs_transport->get_mr(sg, nents, rs,
+						 &mr->r_key);
+
+	if (IS_ERR(trans_private)) {
+		for (i = 0 ; i < nents; i++)
+			put_page(sg_page(&sg[i]));
+		kfree(sg);
+		ret = PTR_ERR(trans_private);
+		goto out;
+	}
+
+	mr->r_trans_private = trans_private;
+
+	rdsdebug("RDS: get_mr put_user key is %x cookie_addr %p\n",
+	       mr->r_key, (void *)(unsigned long) args->cookie_addr);
+
+	/* The user may pass us an unaligned address, but we can only
+	 * map page aligned regions. So we keep the offset, and build
+	 * a 64bit cookie containing <R_Key, offset> and pass that
+	 * around. */
+	cookie = rds_rdma_make_cookie(mr->r_key, args->vec.addr & ~PAGE_MASK);
+	if (cookie_ret)
+		*cookie_ret = cookie;
+
+	if (args->cookie_addr && put_user(cookie, (u64 __user *)(unsigned long) args->cookie_addr)) {
+		ret = -EFAULT;
+		goto out;
+	}
+
+	/* Inserting the new MR into the rbtree bumps its
+	 * reference count. */
+	spin_lock_irqsave(&rs->rs_rdma_lock, flags);
+	found = rds_mr_tree_walk(&rs->rs_rdma_keys, mr->r_key, mr);
+	spin_unlock_irqrestore(&rs->rs_rdma_lock, flags);
+
+	BUG_ON(found && found != mr);
+
+	rdsdebug("RDS: get_mr key is %x\n", mr->r_key);
+	if (mr_ret) {
+		atomic_inc(&mr->r_refcount);
+		*mr_ret = mr;
+	}
+
+	ret = 0;
+out:
+	kfree(pages);
+	if (mr)
+		rds_mr_put(mr);
+	return ret;
+}
+
+int rds_get_mr(struct rds_sock *rs, char __user *optval, int optlen)
+{
+	struct rds_get_mr_args args;
+
+	if (optlen != sizeof(struct rds_get_mr_args))
+		return -EINVAL;
+
+	if (copy_from_user(&args, (struct rds_get_mr_args __user *)optval,
+			   sizeof(struct rds_get_mr_args)))
+		return -EFAULT;
+
+	return __rds_rdma_map(rs, &args, NULL, NULL);
+}
+
+int rds_get_mr_for_dest(struct rds_sock *rs, char __user *optval, int optlen)
+{
+	struct rds_get_mr_for_dest_args args;
+	struct rds_get_mr_args new_args;
+
+	if (optlen != sizeof(struct rds_get_mr_for_dest_args))
+		return -EINVAL;
+
+	if (copy_from_user(&args, (struct rds_get_mr_for_dest_args __user *)optval,
+			   sizeof(struct rds_get_mr_for_dest_args)))
+		return -EFAULT;
+
+	/*
+	 * Initially, just behave like get_mr().
+	 * TODO: Implement get_mr as wrapper around this
+	 *	 and deprecate it.
+	 */
+	new_args.vec = args.vec;
+	new_args.cookie_addr = args.cookie_addr;
+	new_args.flags = args.flags;
+
+	return __rds_rdma_map(rs, &new_args, NULL, NULL);
+}
+
+/*
+ * Free the MR indicated by the given R_Key
+ */
+int rds_free_mr(struct rds_sock *rs, char __user *optval, int optlen)
+{
+	struct rds_free_mr_args args;
+	struct rds_mr *mr;
+	unsigned long flags;
+
+	if (optlen != sizeof(struct rds_free_mr_args))
+		return -EINVAL;
+
+	if (copy_from_user(&args, (struct rds_free_mr_args __user *)optval,
+			   sizeof(struct rds_free_mr_args)))
+		return -EFAULT;
+
+	/* Special case - a null cookie means flush all unused MRs */
+	if (args.cookie == 0) {
+		if (!rs->rs_transport || !rs->rs_transport->flush_mrs)
+			return -EINVAL;
+		rs->rs_transport->flush_mrs();
+		return 0;
+	}
+
+	/* Look up the MR given its R_key and remove it from the rbtree
+	 * so nobody else finds it.
+	 * This should also prevent races with rds_rdma_unuse.
+	 */
+	spin_lock_irqsave(&rs->rs_rdma_lock, flags);
+	mr = rds_mr_tree_walk(&rs->rs_rdma_keys, rds_rdma_cookie_key(args.cookie), NULL);
+	if (mr) {
+		rb_erase(&mr->r_rb_node, &rs->rs_rdma_keys);
+		RB_CLEAR_NODE(&mr->r_rb_node);
+		if (args.flags & RDS_RDMA_INVALIDATE)
+			mr->r_invalidate = 1;
+	}
+	spin_unlock_irqrestore(&rs->rs_rdma_lock, flags);
+
+	if (!mr)
+		return -EINVAL;
+
+	/*
+	 * call rds_destroy_mr() ourselves so that we're sure it's done by the time
+	 * we return.  If we let rds_mr_put() do it it might not happen until
+	 * someone else drops their ref.
+	 */
+	rds_destroy_mr(mr);
+	rds_mr_put(mr);
+	return 0;
+}
+
+/*
+ * This is called when we receive an extension header that
+ * tells us this MR was used. It allows us to implement
+ * use_once semantics
+ */
+void rds_rdma_unuse(struct rds_sock *rs, u32 r_key, int force)
+{
+	struct rds_mr *mr;
+	unsigned long flags;
+	int zot_me = 0;
+
+	spin_lock_irqsave(&rs->rs_rdma_lock, flags);
+	mr = rds_mr_tree_walk(&rs->rs_rdma_keys, r_key, NULL);
+	if (!mr) {
+		printk(KERN_ERR "rds: trying to unuse MR with unknown r_key %u!\n", r_key);
+		spin_unlock_irqrestore(&rs->rs_rdma_lock, flags);
+		return;
+	}
+
+	if (mr->r_use_once || force) {
+		rb_erase(&mr->r_rb_node, &rs->rs_rdma_keys);
+		RB_CLEAR_NODE(&mr->r_rb_node);
+		zot_me = 1;
+	}
+	spin_unlock_irqrestore(&rs->rs_rdma_lock, flags);
+
+	/* May have to issue a dma_sync on this memory region.
+	 * Note we could avoid this if the operation was a RDMA READ,
+	 * but at this point we can't tell. */
+	if (mr->r_trans->sync_mr)
+		mr->r_trans->sync_mr(mr->r_trans_private, DMA_FROM_DEVICE);
+
+	/* If the MR was marked as invalidate, this will
+	 * trigger an async flush. */
+	if (zot_me)
+		rds_destroy_mr(mr);
+	rds_mr_put(mr);
+}
+
+void rds_rdma_free_op(struct rm_rdma_op *ro)
+{
+	unsigned int i;
+
+	for (i = 0; i < ro->op_nents; i++) {
+		struct page *page = sg_page(&ro->op_sg[i]);
+
+		/* Mark page dirty if it was possibly modified, which
+		 * is the case for a RDMA_READ which copies from remote
+		 * to local memory */
+		if (!ro->op_write) {
+			BUG_ON(irqs_disabled());
+			set_page_dirty(page);
+		}
+		put_page(page);
+	}
+
+	kfree(ro->op_notifier);
+	ro->op_notifier = NULL;
+	ro->op_active = 0;
+}
+
+void rds_atomic_free_op(struct rm_atomic_op *ao)
+{
+	struct page *page = sg_page(ao->op_sg);
+
+	/* Mark page dirty if it was possibly modified, which
+	 * is the case for a RDMA_READ which copies from remote
+	 * to local memory */
+	set_page_dirty(page);
+	put_page(page);
+
+	kfree(ao->op_notifier);
+	ao->op_notifier = NULL;
+	ao->op_active = 0;
+}
+
+
+/*
+ * Count the number of pages needed to describe an incoming iovec array.
+ */
+static int rds_rdma_pages(struct rds_iovec iov[], int nr_iovecs)
+{
+	int tot_pages = 0;
+	unsigned int nr_pages;
+	unsigned int i;
+
+	/* figure out the number of pages in the vector */
+	for (i = 0; i < nr_iovecs; i++) {
+		nr_pages = rds_pages_in_vec(&iov[i]);
+		if (nr_pages == 0)
+			return -EINVAL;
+
+		tot_pages += nr_pages;
+
+		/*
+		 * nr_pages for one entry is limited to (UINT_MAX>>PAGE_SHIFT)+1,
+		 * so tot_pages cannot overflow without first going negative.
+		 */
+		if (tot_pages < 0)
+			return -EINVAL;
+	}
+
+	return tot_pages;
+}
+
+int rds_rdma_extra_size(struct rds_rdma_args *args)
+{
+	struct rds_iovec vec;
+	struct rds_iovec __user *local_vec;
+	int tot_pages = 0;
+	unsigned int nr_pages;
+	unsigned int i;
+
+	local_vec = (struct rds_iovec __user *)(unsigned long) args->local_vec_addr;
+
+	/* figure out the number of pages in the vector */
+	for (i = 0; i < args->nr_local; i++) {
+		if (copy_from_user(&vec, &local_vec[i],
+				   sizeof(struct rds_iovec)))
+			return -EFAULT;
+
+		nr_pages = rds_pages_in_vec(&vec);
+		if (nr_pages == 0)
+			return -EINVAL;
+
+		tot_pages += nr_pages;
+
+		/*
+		 * nr_pages for one entry is limited to (UINT_MAX>>PAGE_SHIFT)+1,
+		 * so tot_pages cannot overflow without first going negative.
+		 */
+		if (tot_pages < 0)
+			return -EINVAL;
+	}
+
+	return tot_pages * sizeof(struct scatterlist);
+}
+
+/*
+ * The application asks for a RDMA transfer.
+ * Extract all arguments and set up the rdma_op
+ */
+int rds_cmsg_rdma_args(struct rds_sock *rs, struct rds_message *rm,
+			  struct cmsghdr *cmsg)
+{
+	struct rds_rdma_args *args;
+	struct rm_rdma_op *op = &rm->rdma;
+	int nr_pages;
+	unsigned int nr_bytes;
+	struct page **pages = NULL;
+	struct rds_iovec iovstack[UIO_FASTIOV], *iovs = iovstack;
+	int iov_size;
+	unsigned int i, j;
+	int ret = 0;
+
+	if (cmsg->cmsg_len < CMSG_LEN(sizeof(struct rds_rdma_args))
+	    || rm->rdma.op_active)
+		return -EINVAL;
+
+	args = CMSG_DATA(cmsg);
+
+	if (rs->rs_bound_addr == 0) {
+		ret = -ENOTCONN; /* XXX not a great errno */
+		goto out_ret;
+	}
+
+	if (args->nr_local > UIO_MAXIOV) {
+		ret = -EMSGSIZE;
+		goto out_ret;
+	}
+
+	/* Check whether to allocate the iovec area */
+	iov_size = args->nr_local * sizeof(struct rds_iovec);
+	if (args->nr_local > UIO_FASTIOV) {
+		iovs = sock_kmalloc(rds_rs_to_sk(rs), iov_size, GFP_KERNEL);
+		if (!iovs) {
+			ret = -ENOMEM;
+			goto out_ret;
+		}
+	}
+
+	if (copy_from_user(iovs, (struct rds_iovec __user *)(unsigned long) args->local_vec_addr, iov_size)) {
+		ret = -EFAULT;
+		goto out;
+	}
+
+	nr_pages = rds_rdma_pages(iovs, args->nr_local);
+	if (nr_pages < 0) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	pages = kcalloc(nr_pages, sizeof(struct page *), GFP_KERNEL);
+	if (!pages) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	op->op_write = !!(args->flags & RDS_RDMA_READWRITE);
+	op->op_fence = !!(args->flags & RDS_RDMA_FENCE);
+	op->op_notify = !!(args->flags & RDS_RDMA_NOTIFY_ME);
+	op->op_silent = !!(args->flags & RDS_RDMA_SILENT);
+	op->op_active = 1;
+	op->op_recverr = rs->rs_recverr;
+	WARN_ON(!nr_pages);
+	op->op_sg = rds_message_alloc_sgs(rm, nr_pages);
+	if (!op->op_sg) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	if (op->op_notify || op->op_recverr) {
+		/* We allocate an uninitialized notifier here, because
+		 * we don't want to do that in the completion handler. We
+		 * would have to use GFP_ATOMIC there, and don't want to deal
+		 * with failed allocations.
+		 */
+		op->op_notifier = kmalloc(sizeof(struct rds_notifier), GFP_KERNEL);
+		if (!op->op_notifier) {
+			ret = -ENOMEM;
+			goto out;
+		}
+		op->op_notifier->n_user_token = args->user_token;
+		op->op_notifier->n_status = RDS_RDMA_SUCCESS;
+	}
+
+	/* The cookie contains the R_Key of the remote memory region, and
+	 * optionally an offset into it. This is how we implement RDMA into
+	 * unaligned memory.
+	 * When setting up the RDMA, we need to add that offset to the
+	 * destination address (which is really an offset into the MR)
+	 * FIXME: We may want to move this into ib_rdma.c
+	 */
+	op->op_rkey = rds_rdma_cookie_key(args->cookie);
+	op->op_remote_addr = args->remote_vec.addr + rds_rdma_cookie_offset(args->cookie);
+
+	nr_bytes = 0;
+
+	rdsdebug("RDS: rdma prepare nr_local %llu rva %llx rkey %x\n",
+	       (unsigned long long)args->nr_local,
+	       (unsigned long long)args->remote_vec.addr,
+	       op->op_rkey);
+
+	for (i = 0; i < args->nr_local; i++) {
+		struct rds_iovec *iov = &iovs[i];
+		/* don't need to check, rds_rdma_pages() verified nr will be +nonzero */
+		unsigned int nr = rds_pages_in_vec(iov);
+
+		rs->rs_user_addr = iov->addr;
+		rs->rs_user_bytes = iov->bytes;
+
+		/* If it's a WRITE operation, we want to pin the pages for reading.
+		 * If it's a READ operation, we need to pin the pages for writing.
+		 */
+		ret = rds_pin_pages(iov->addr, nr, pages, !op->op_write);
+		if (ret < 0)
+			goto out;
+
+		rdsdebug("RDS: nr_bytes %u nr %u iov->bytes %llu iov->addr %llx\n",
+			 nr_bytes, nr, iov->bytes, iov->addr);
+
+		nr_bytes += iov->bytes;
+
+		for (j = 0; j < nr; j++) {
+			unsigned int offset = iov->addr & ~PAGE_MASK;
+			struct scatterlist *sg;
+
+			sg = &op->op_sg[op->op_nents + j];
+			sg_set_page(sg, pages[j],
+					min_t(unsigned int, iov->bytes, PAGE_SIZE - offset),
+					offset);
+
+			rdsdebug("RDS: sg->offset %x sg->len %x iov->addr %llx iov->bytes %llu\n",
+			       sg->offset, sg->length, iov->addr, iov->bytes);
+
+			iov->addr += sg->length;
+			iov->bytes -= sg->length;
+		}
+
+		op->op_nents += nr;
+	}
+
+	if (nr_bytes > args->remote_vec.bytes) {
+		rdsdebug("RDS nr_bytes %u remote_bytes %u do not match\n",
+				nr_bytes,
+				(unsigned int) args->remote_vec.bytes);
+		ret = -EINVAL;
+		goto out;
+	}
+	op->op_bytes = nr_bytes;
+
+out:
+	if (iovs != iovstack)
+		sock_kfree_s(rds_rs_to_sk(rs), iovs, iov_size);
+	kfree(pages);
+out_ret:
+	if (ret)
+		rds_rdma_free_op(op);
+	else
+		rds_stats_inc(s_send_rdma);
+
+	return ret;
+}
+
+/*
+ * The application wants us to pass an RDMA destination (aka MR)
+ * to the remote
+ */
+int rds_cmsg_rdma_dest(struct rds_sock *rs, struct rds_message *rm,
+			  struct cmsghdr *cmsg)
+{
+	unsigned long flags;
+	struct rds_mr *mr;
+	u32 r_key;
+	int err = 0;
+
+	if (cmsg->cmsg_len < CMSG_LEN(sizeof(rds_rdma_cookie_t)) ||
+	    rm->m_rdma_cookie != 0)
+		return -EINVAL;
+
+	memcpy(&rm->m_rdma_cookie, CMSG_DATA(cmsg), sizeof(rm->m_rdma_cookie));
+
+	/* We are reusing a previously mapped MR here. Most likely, the
+	 * application has written to the buffer, so we need to explicitly
+	 * flush those writes to RAM. Otherwise the HCA may not see them
+	 * when doing a DMA from that buffer.
+	 */
+	r_key = rds_rdma_cookie_key(rm->m_rdma_cookie);
+
+	spin_lock_irqsave(&rs->rs_rdma_lock, flags);
+	mr = rds_mr_tree_walk(&rs->rs_rdma_keys, r_key, NULL);
+	if (!mr)
+		err = -EINVAL;	/* invalid r_key */
+	else
+		atomic_inc(&mr->r_refcount);
+	spin_unlock_irqrestore(&rs->rs_rdma_lock, flags);
+
+	if (mr) {
+		mr->r_trans->sync_mr(mr->r_trans_private, DMA_TO_DEVICE);
+		rm->rdma.op_rdma_mr = mr;
+	}
+	return err;
+}
+
+/*
+ * The application passes us an address range it wants to enable RDMA
+ * to/from. We map the area, and save the <R_Key,offset> pair
+ * in rm->m_rdma_cookie. This causes it to be sent along to the peer
+ * in an extension header.
+ */
+int rds_cmsg_rdma_map(struct rds_sock *rs, struct rds_message *rm,
+			  struct cmsghdr *cmsg)
+{
+	if (cmsg->cmsg_len < CMSG_LEN(sizeof(struct rds_get_mr_args)) ||
+	    rm->m_rdma_cookie != 0)
+		return -EINVAL;
+
+	return __rds_rdma_map(rs, CMSG_DATA(cmsg), &rm->m_rdma_cookie, &rm->rdma.op_rdma_mr);
+}
+
+/*
+ * Fill in rds_message for an atomic request.
+ */
+int rds_cmsg_atomic(struct rds_sock *rs, struct rds_message *rm,
+		    struct cmsghdr *cmsg)
+{
+	struct page *page = NULL;
+	struct rds_atomic_args *args;
+	int ret = 0;
+
+	if (cmsg->cmsg_len < CMSG_LEN(sizeof(struct rds_atomic_args))
+	 || rm->atomic.op_active)
+		return -EINVAL;
+
+	args = CMSG_DATA(cmsg);
+
+	/* Nonmasked & masked cmsg ops converted to masked hw ops */
+	switch (cmsg->cmsg_type) {
+	case RDS_CMSG_ATOMIC_FADD:
+		rm->atomic.op_type = RDS_ATOMIC_TYPE_FADD;
+		rm->atomic.op_m_fadd.add = args->fadd.add;
+		rm->atomic.op_m_fadd.nocarry_mask = 0;
+		break;
+	case RDS_CMSG_MASKED_ATOMIC_FADD:
+		rm->atomic.op_type = RDS_ATOMIC_TYPE_FADD;
+		rm->atomic.op_m_fadd.add = args->m_fadd.add;
+		rm->atomic.op_m_fadd.nocarry_mask = args->m_fadd.nocarry_mask;
+		break;
+	case RDS_CMSG_ATOMIC_CSWP:
+		rm->atomic.op_type = RDS_ATOMIC_TYPE_CSWP;
+		rm->atomic.op_m_cswp.compare = args->cswp.compare;
+		rm->atomic.op_m_cswp.swap = args->cswp.swap;
+		rm->atomic.op_m_cswp.compare_mask = ~0;
+		rm->atomic.op_m_cswp.swap_mask = ~0;
+		break;
+	case RDS_CMSG_MASKED_ATOMIC_CSWP:
+		rm->atomic.op_type = RDS_ATOMIC_TYPE_CSWP;
+		rm->atomic.op_m_cswp.compare = args->m_cswp.compare;
+		rm->atomic.op_m_cswp.swap = args->m_cswp.swap;
+		rm->atomic.op_m_cswp.compare_mask = args->m_cswp.compare_mask;
+		rm->atomic.op_m_cswp.swap_mask = args->m_cswp.swap_mask;
+		break;
+	default:
+		BUG(); /* should never happen */
+	}
+
+	rm->atomic.op_notify = !!(args->flags & RDS_RDMA_NOTIFY_ME);
+	rm->atomic.op_silent = !!(args->flags & RDS_RDMA_SILENT);
+	rm->atomic.op_active = 1;
+	rm->atomic.op_recverr = rs->rs_recverr;
+	rm->atomic.op_sg = rds_message_alloc_sgs(rm, 1);
+	if (!rm->atomic.op_sg) {
+		ret = -ENOMEM;
+		goto err;
+	}
+
+	/* verify 8 byte-aligned */
+	if (args->local_addr & 0x7) {
+		ret = -EFAULT;
+		goto err;
+	}
+
+	ret = rds_pin_pages(args->local_addr, 1, &page, 1);
+	if (ret != 1)
+		goto err;
+	ret = 0;
+
+	sg_set_page(rm->atomic.op_sg, page, 8, offset_in_page(args->local_addr));
+
+	if (rm->atomic.op_notify || rm->atomic.op_recverr) {
+		/* We allocate an uninitialized notifier here, because
+		 * we don't want to do that in the completion handler. We
+		 * would have to use GFP_ATOMIC there, and don't want to deal
+		 * with failed allocations.
+		 */
+		rm->atomic.op_notifier = kmalloc(sizeof(*rm->atomic.op_notifier), GFP_KERNEL);
+		if (!rm->atomic.op_notifier) {
+			ret = -ENOMEM;
+			goto err;
+		}
+
+		rm->atomic.op_notifier->n_user_token = args->user_token;
+		rm->atomic.op_notifier->n_status = RDS_RDMA_SUCCESS;
+	}
+
+	rm->atomic.op_rkey = rds_rdma_cookie_key(args->cookie);
+	rm->atomic.op_remote_addr = args->remote_addr + rds_rdma_cookie_offset(args->cookie);
+
+	return ret;
+err:
+	if (page)
+		put_page(page);
+	kfree(rm->atomic.op_notifier);
+
+	return ret;
+}
diff --git a/net/rds/rdma_transport.c b/net/rds/rdma_transport.c
new file mode 100644
index 000000000..6cd9d1dea
--- /dev/null
+++ b/net/rds/rdma_transport.c
@@ -0,0 +1,251 @@
+/*
+ * Copyright (c) 2009 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/module.h>
+#include <rdma/rdma_cm.h>
+
+#include "rdma_transport.h"
+
+static struct rdma_cm_id *rds_rdma_listen_id;
+
+static char *rds_cm_event_strings[] = {
+#define RDS_CM_EVENT_STRING(foo) \
+		[RDMA_CM_EVENT_##foo] = __stringify(RDMA_CM_EVENT_##foo)
+	RDS_CM_EVENT_STRING(ADDR_RESOLVED),
+	RDS_CM_EVENT_STRING(ADDR_ERROR),
+	RDS_CM_EVENT_STRING(ROUTE_RESOLVED),
+	RDS_CM_EVENT_STRING(ROUTE_ERROR),
+	RDS_CM_EVENT_STRING(CONNECT_REQUEST),
+	RDS_CM_EVENT_STRING(CONNECT_RESPONSE),
+	RDS_CM_EVENT_STRING(CONNECT_ERROR),
+	RDS_CM_EVENT_STRING(UNREACHABLE),
+	RDS_CM_EVENT_STRING(REJECTED),
+	RDS_CM_EVENT_STRING(ESTABLISHED),
+	RDS_CM_EVENT_STRING(DISCONNECTED),
+	RDS_CM_EVENT_STRING(DEVICE_REMOVAL),
+	RDS_CM_EVENT_STRING(MULTICAST_JOIN),
+	RDS_CM_EVENT_STRING(MULTICAST_ERROR),
+	RDS_CM_EVENT_STRING(ADDR_CHANGE),
+	RDS_CM_EVENT_STRING(TIMEWAIT_EXIT),
+#undef RDS_CM_EVENT_STRING
+};
+
+static char *rds_cm_event_str(enum rdma_cm_event_type type)
+{
+	return rds_str_array(rds_cm_event_strings,
+			     ARRAY_SIZE(rds_cm_event_strings), type);
+};
+
+int rds_rdma_cm_event_handler(struct rdma_cm_id *cm_id,
+			      struct rdma_cm_event *event)
+{
+	/* this can be null in the listening path */
+	struct rds_connection *conn = cm_id->context;
+	struct rds_transport *trans;
+	int ret = 0;
+
+	rdsdebug("conn %p id %p handling event %u (%s)\n", conn, cm_id,
+		 event->event, rds_cm_event_str(event->event));
+
+	if (cm_id->device->node_type == RDMA_NODE_RNIC)
+		trans = &rds_iw_transport;
+	else
+		trans = &rds_ib_transport;
+
+	/* Prevent shutdown from tearing down the connection
+	 * while we're executing. */
+	if (conn) {
+		mutex_lock(&conn->c_cm_lock);
+
+		/* If the connection is being shut down, bail out
+		 * right away. We return 0 so cm_id doesn't get
+		 * destroyed prematurely */
+		if (rds_conn_state(conn) == RDS_CONN_DISCONNECTING) {
+			/* Reject incoming connections while we're tearing
+			 * down an existing one. */
+			if (event->event == RDMA_CM_EVENT_CONNECT_REQUEST)
+				ret = 1;
+			goto out;
+		}
+	}
+
+	switch (event->event) {
+	case RDMA_CM_EVENT_CONNECT_REQUEST:
+		ret = trans->cm_handle_connect(cm_id, event);
+		break;
+
+	case RDMA_CM_EVENT_ADDR_RESOLVED:
+		/* XXX do we need to clean up if this fails? */
+		ret = rdma_resolve_route(cm_id,
+					 RDS_RDMA_RESOLVE_TIMEOUT_MS);
+		break;
+
+	case RDMA_CM_EVENT_ROUTE_RESOLVED:
+		/* XXX worry about racing with listen acceptance */
+		ret = trans->cm_initiate_connect(cm_id);
+		break;
+
+	case RDMA_CM_EVENT_ESTABLISHED:
+		trans->cm_connect_complete(conn, event);
+		break;
+
+	case RDMA_CM_EVENT_ADDR_ERROR:
+	case RDMA_CM_EVENT_ROUTE_ERROR:
+	case RDMA_CM_EVENT_CONNECT_ERROR:
+	case RDMA_CM_EVENT_UNREACHABLE:
+	case RDMA_CM_EVENT_REJECTED:
+	case RDMA_CM_EVENT_DEVICE_REMOVAL:
+	case RDMA_CM_EVENT_ADDR_CHANGE:
+		if (conn)
+			rds_conn_drop(conn);
+		break;
+
+	case RDMA_CM_EVENT_DISCONNECTED:
+		rdsdebug("DISCONNECT event - dropping connection "
+			"%pI4->%pI4\n", &conn->c_laddr,
+			 &conn->c_faddr);
+		rds_conn_drop(conn);
+		break;
+
+	default:
+		/* things like device disconnect? */
+		printk(KERN_ERR "RDS: unknown event %u (%s)!\n",
+		       event->event, rds_cm_event_str(event->event));
+		break;
+	}
+
+out:
+	if (conn)
+		mutex_unlock(&conn->c_cm_lock);
+
+	rdsdebug("id %p event %u (%s) handling ret %d\n", cm_id, event->event,
+		 rds_cm_event_str(event->event), ret);
+
+	return ret;
+}
+
+static int rds_rdma_listen_init(void)
+{
+	struct sockaddr_in sin;
+	struct rdma_cm_id *cm_id;
+	int ret;
+
+	cm_id = rdma_create_id(rds_rdma_cm_event_handler, NULL, RDMA_PS_TCP,
+			       IB_QPT_RC);
+	if (IS_ERR(cm_id)) {
+		ret = PTR_ERR(cm_id);
+		printk(KERN_ERR "RDS/RDMA: failed to setup listener, "
+		       "rdma_create_id() returned %d\n", ret);
+		return ret;
+	}
+
+	sin.sin_family = AF_INET;
+	sin.sin_addr.s_addr = (__force u32)htonl(INADDR_ANY);
+	sin.sin_port = (__force u16)htons(RDS_PORT);
+
+	/*
+	 * XXX I bet this binds the cm_id to a device.  If we want to support
+	 * fail-over we'll have to take this into consideration.
+	 */
+	ret = rdma_bind_addr(cm_id, (struct sockaddr *)&sin);
+	if (ret) {
+		printk(KERN_ERR "RDS/RDMA: failed to setup listener, "
+		       "rdma_bind_addr() returned %d\n", ret);
+		goto out;
+	}
+
+	ret = rdma_listen(cm_id, 128);
+	if (ret) {
+		printk(KERN_ERR "RDS/RDMA: failed to setup listener, "
+		       "rdma_listen() returned %d\n", ret);
+		goto out;
+	}
+
+	rdsdebug("cm %p listening on port %u\n", cm_id, RDS_PORT);
+
+	rds_rdma_listen_id = cm_id;
+	cm_id = NULL;
+out:
+	if (cm_id)
+		rdma_destroy_id(cm_id);
+	return ret;
+}
+
+static void rds_rdma_listen_stop(void)
+{
+	if (rds_rdma_listen_id) {
+		rdsdebug("cm %p\n", rds_rdma_listen_id);
+		rdma_destroy_id(rds_rdma_listen_id);
+		rds_rdma_listen_id = NULL;
+	}
+}
+
+static int rds_rdma_init(void)
+{
+	int ret;
+
+	ret = rds_rdma_listen_init();
+	if (ret)
+		goto out;
+
+	ret = rds_iw_init();
+	if (ret)
+		goto err_iw_init;
+
+	ret = rds_ib_init();
+	if (ret)
+		goto err_ib_init;
+
+	goto out;
+
+err_ib_init:
+	rds_iw_exit();
+err_iw_init:
+	rds_rdma_listen_stop();
+out:
+	return ret;
+}
+module_init(rds_rdma_init);
+
+static void rds_rdma_exit(void)
+{
+	/* stop listening first to ensure no new connections are attempted */
+	rds_rdma_listen_stop();
+	rds_ib_exit();
+	rds_iw_exit();
+}
+module_exit(rds_rdma_exit);
+
+MODULE_AUTHOR("Oracle Corporation <rds-devel@oss.oracle.com>");
+MODULE_DESCRIPTION("RDS: IB/iWARP transport");
+MODULE_LICENSE("Dual BSD/GPL");
+
diff --git a/net/rds/rdma_transport.h b/net/rds/rdma_transport.h
new file mode 100644
index 000000000..faba4e382
--- /dev/null
+++ b/net/rds/rdma_transport.h
@@ -0,0 +1,24 @@
+#ifndef _RDMA_TRANSPORT_H
+#define _RDMA_TRANSPORT_H
+
+#include <rdma/ib_verbs.h>
+#include <rdma/rdma_cm.h>
+#include "rds.h"
+
+#define RDS_RDMA_RESOLVE_TIMEOUT_MS     5000
+
+int rds_rdma_conn_connect(struct rds_connection *conn);
+int rds_rdma_cm_event_handler(struct rdma_cm_id *cm_id,
+			      struct rdma_cm_event *event);
+
+/* from ib.c */
+extern struct rds_transport rds_ib_transport;
+int rds_ib_init(void);
+void rds_ib_exit(void);
+
+/* from iw.c */
+extern struct rds_transport rds_iw_transport;
+int rds_iw_init(void);
+void rds_iw_exit(void);
+
+#endif
diff --git a/net/rds/rds.h b/net/rds/rds.h
new file mode 100644
index 000000000..0d41155a2
--- /dev/null
+++ b/net/rds/rds.h
@@ -0,0 +1,809 @@
+#ifndef _RDS_RDS_H
+#define _RDS_RDS_H
+
+#include <net/sock.h>
+#include <linux/scatterlist.h>
+#include <linux/highmem.h>
+#include <rdma/rdma_cm.h>
+#include <linux/mutex.h>
+#include <linux/rds.h>
+
+#include "info.h"
+
+/*
+ * RDS Network protocol version
+ */
+#define RDS_PROTOCOL_3_0	0x0300
+#define RDS_PROTOCOL_3_1	0x0301
+#define RDS_PROTOCOL_VERSION	RDS_PROTOCOL_3_1
+#define RDS_PROTOCOL_MAJOR(v)	((v) >> 8)
+#define RDS_PROTOCOL_MINOR(v)	((v) & 255)
+#define RDS_PROTOCOL(maj, min)	(((maj) << 8) | min)
+
+/*
+ * XXX randomly chosen, but at least seems to be unused:
+ * #               18464-18768 Unassigned
+ * We should do better.  We want a reserved port to discourage unpriv'ed
+ * userspace from listening.
+ */
+#define RDS_PORT	18634
+
+#ifdef ATOMIC64_INIT
+#define KERNEL_HAS_ATOMIC64
+#endif
+
+#ifdef DEBUG
+#define rdsdebug(fmt, args...) pr_debug("%s(): " fmt, __func__ , ##args)
+#else
+/* sigh, pr_debug() causes unused variable warnings */
+static inline __printf(1, 2)
+void rdsdebug(char *fmt, ...)
+{
+}
+#endif
+
+/* XXX is there one of these somewhere? */
+#define ceil(x, y) \
+	({ unsigned long __x = (x), __y = (y); (__x + __y - 1) / __y; })
+
+#define RDS_FRAG_SHIFT	12
+#define RDS_FRAG_SIZE	((unsigned int)(1 << RDS_FRAG_SHIFT))
+
+#define RDS_CONG_MAP_BYTES	(65536 / 8)
+#define RDS_CONG_MAP_PAGES	(PAGE_ALIGN(RDS_CONG_MAP_BYTES) / PAGE_SIZE)
+#define RDS_CONG_MAP_PAGE_BITS	(PAGE_SIZE * 8)
+
+struct rds_cong_map {
+	struct rb_node		m_rb_node;
+	__be32			m_addr;
+	wait_queue_head_t	m_waitq;
+	struct list_head	m_conn_list;
+	unsigned long		m_page_addrs[RDS_CONG_MAP_PAGES];
+};
+
+
+/*
+ * This is how we will track the connection state:
+ * A connection is always in one of the following
+ * states. Updates to the state are atomic and imply
+ * a memory barrier.
+ */
+enum {
+	RDS_CONN_DOWN = 0,
+	RDS_CONN_CONNECTING,
+	RDS_CONN_DISCONNECTING,
+	RDS_CONN_UP,
+	RDS_CONN_ERROR,
+};
+
+/* Bits for c_flags */
+#define RDS_LL_SEND_FULL	0
+#define RDS_RECONNECT_PENDING	1
+#define RDS_IN_XMIT		2
+
+struct rds_connection {
+	struct hlist_node	c_hash_node;
+	__be32			c_laddr;
+	__be32			c_faddr;
+	unsigned int		c_loopback:1;
+	struct rds_connection	*c_passive;
+
+	struct rds_cong_map	*c_lcong;
+	struct rds_cong_map	*c_fcong;
+
+	struct rds_message	*c_xmit_rm;
+	unsigned long		c_xmit_sg;
+	unsigned int		c_xmit_hdr_off;
+	unsigned int		c_xmit_data_off;
+	unsigned int		c_xmit_atomic_sent;
+	unsigned int		c_xmit_rdma_sent;
+	unsigned int		c_xmit_data_sent;
+
+	spinlock_t		c_lock;		/* protect msg queues */
+	u64			c_next_tx_seq;
+	struct list_head	c_send_queue;
+	struct list_head	c_retrans;
+
+	u64			c_next_rx_seq;
+
+	struct rds_transport	*c_trans;
+	void			*c_transport_data;
+
+	atomic_t		c_state;
+	unsigned long		c_send_gen;
+	unsigned long		c_flags;
+	unsigned long		c_reconnect_jiffies;
+	struct delayed_work	c_send_w;
+	struct delayed_work	c_recv_w;
+	struct delayed_work	c_conn_w;
+	struct work_struct	c_down_w;
+	struct mutex		c_cm_lock;	/* protect conn state & cm */
+	wait_queue_head_t	c_waitq;
+
+	struct list_head	c_map_item;
+	unsigned long		c_map_queued;
+
+	unsigned int		c_unacked_packets;
+	unsigned int		c_unacked_bytes;
+
+	/* Protocol version */
+	unsigned int		c_version;
+};
+
+#define RDS_FLAG_CONG_BITMAP	0x01
+#define RDS_FLAG_ACK_REQUIRED	0x02
+#define RDS_FLAG_RETRANSMITTED	0x04
+#define RDS_MAX_ADV_CREDIT	255
+
+/*
+ * Maximum space available for extension headers.
+ */
+#define RDS_HEADER_EXT_SPACE	16
+
+struct rds_header {
+	__be64	h_sequence;
+	__be64	h_ack;
+	__be32	h_len;
+	__be16	h_sport;
+	__be16	h_dport;
+	u8	h_flags;
+	u8	h_credit;
+	u8	h_padding[4];
+	__sum16	h_csum;
+
+	u8	h_exthdr[RDS_HEADER_EXT_SPACE];
+};
+
+/*
+ * Reserved - indicates end of extensions
+ */
+#define RDS_EXTHDR_NONE		0
+
+/*
+ * This extension header is included in the very
+ * first message that is sent on a new connection,
+ * and identifies the protocol level. This will help
+ * rolling updates if a future change requires breaking
+ * the protocol.
+ * NB: This is no longer true for IB, where we do a version
+ * negotiation during the connection setup phase (protocol
+ * version information is included in the RDMA CM private data).
+ */
+#define RDS_EXTHDR_VERSION	1
+struct rds_ext_header_version {
+	__be32			h_version;
+};
+
+/*
+ * This extension header is included in the RDS message
+ * chasing an RDMA operation.
+ */
+#define RDS_EXTHDR_RDMA		2
+struct rds_ext_header_rdma {
+	__be32			h_rdma_rkey;
+};
+
+/*
+ * This extension header tells the peer about the
+ * destination <R_Key,offset> of the requested RDMA
+ * operation.
+ */
+#define RDS_EXTHDR_RDMA_DEST	3
+struct rds_ext_header_rdma_dest {
+	__be32			h_rdma_rkey;
+	__be32			h_rdma_offset;
+};
+
+#define __RDS_EXTHDR_MAX	16 /* for now */
+
+struct rds_incoming {
+	atomic_t		i_refcount;
+	struct list_head	i_item;
+	struct rds_connection	*i_conn;
+	struct rds_header	i_hdr;
+	unsigned long		i_rx_jiffies;
+	__be32			i_saddr;
+
+	rds_rdma_cookie_t	i_rdma_cookie;
+};
+
+struct rds_mr {
+	struct rb_node		r_rb_node;
+	atomic_t		r_refcount;
+	u32			r_key;
+
+	/* A copy of the creation flags */
+	unsigned int		r_use_once:1;
+	unsigned int		r_invalidate:1;
+	unsigned int		r_write:1;
+
+	/* This is for RDS_MR_DEAD.
+	 * It would be nice & consistent to make this part of the above
+	 * bit field here, but we need to use test_and_set_bit.
+	 */
+	unsigned long		r_state;
+	struct rds_sock		*r_sock; /* back pointer to the socket that owns us */
+	struct rds_transport	*r_trans;
+	void			*r_trans_private;
+};
+
+/* Flags for mr->r_state */
+#define RDS_MR_DEAD		0
+
+static inline rds_rdma_cookie_t rds_rdma_make_cookie(u32 r_key, u32 offset)
+{
+	return r_key | (((u64) offset) << 32);
+}
+
+static inline u32 rds_rdma_cookie_key(rds_rdma_cookie_t cookie)
+{
+	return cookie;
+}
+
+static inline u32 rds_rdma_cookie_offset(rds_rdma_cookie_t cookie)
+{
+	return cookie >> 32;
+}
+
+/* atomic operation types */
+#define RDS_ATOMIC_TYPE_CSWP		0
+#define RDS_ATOMIC_TYPE_FADD		1
+
+/*
+ * m_sock_item and m_conn_item are on lists that are serialized under
+ * conn->c_lock.  m_sock_item has additional meaning in that once it is empty
+ * the message will not be put back on the retransmit list after being sent.
+ * messages that are canceled while being sent rely on this.
+ *
+ * m_inc is used by loopback so that it can pass an incoming message straight
+ * back up into the rx path.  It embeds a wire header which is also used by
+ * the send path, which is kind of awkward.
+ *
+ * m_sock_item indicates the message's presence on a socket's send or receive
+ * queue.  m_rs will point to that socket.
+ *
+ * m_daddr is used by cancellation to prune messages to a given destination.
+ *
+ * The RDS_MSG_ON_SOCK and RDS_MSG_ON_CONN flags are used to avoid lock
+ * nesting.  As paths iterate over messages on a sock, or conn, they must
+ * also lock the conn, or sock, to remove the message from those lists too.
+ * Testing the flag to determine if the message is still on the lists lets
+ * us avoid testing the list_head directly.  That means each path can use
+ * the message's list_head to keep it on a local list while juggling locks
+ * without confusing the other path.
+ *
+ * m_ack_seq is an optional field set by transports who need a different
+ * sequence number range to invalidate.  They can use this in a callback
+ * that they pass to rds_send_drop_acked() to see if each message has been
+ * acked.  The HAS_ACK_SEQ flag can be used to detect messages which haven't
+ * had ack_seq set yet.
+ */
+#define RDS_MSG_ON_SOCK		1
+#define RDS_MSG_ON_CONN		2
+#define RDS_MSG_HAS_ACK_SEQ	3
+#define RDS_MSG_ACK_REQUIRED	4
+#define RDS_MSG_RETRANSMITTED	5
+#define RDS_MSG_MAPPED		6
+#define RDS_MSG_PAGEVEC		7
+
+struct rds_message {
+	atomic_t		m_refcount;
+	struct list_head	m_sock_item;
+	struct list_head	m_conn_item;
+	struct rds_incoming	m_inc;
+	u64			m_ack_seq;
+	__be32			m_daddr;
+	unsigned long		m_flags;
+
+	/* Never access m_rs without holding m_rs_lock.
+	 * Lock nesting is
+	 *  rm->m_rs_lock
+	 *   -> rs->rs_lock
+	 */
+	spinlock_t		m_rs_lock;
+	wait_queue_head_t	m_flush_wait;
+
+	struct rds_sock		*m_rs;
+
+	/* cookie to send to remote, in rds header */
+	rds_rdma_cookie_t	m_rdma_cookie;
+
+	unsigned int		m_used_sgs;
+	unsigned int		m_total_sgs;
+
+	void			*m_final_op;
+
+	struct {
+		struct rm_atomic_op {
+			int			op_type;
+			union {
+				struct {
+					uint64_t	compare;
+					uint64_t	swap;
+					uint64_t	compare_mask;
+					uint64_t	swap_mask;
+				} op_m_cswp;
+				struct {
+					uint64_t	add;
+					uint64_t	nocarry_mask;
+				} op_m_fadd;
+			};
+
+			u32			op_rkey;
+			u64			op_remote_addr;
+			unsigned int		op_notify:1;
+			unsigned int		op_recverr:1;
+			unsigned int		op_mapped:1;
+			unsigned int		op_silent:1;
+			unsigned int		op_active:1;
+			struct scatterlist	*op_sg;
+			struct rds_notifier	*op_notifier;
+
+			struct rds_mr		*op_rdma_mr;
+		} atomic;
+		struct rm_rdma_op {
+			u32			op_rkey;
+			u64			op_remote_addr;
+			unsigned int		op_write:1;
+			unsigned int		op_fence:1;
+			unsigned int		op_notify:1;
+			unsigned int		op_recverr:1;
+			unsigned int		op_mapped:1;
+			unsigned int		op_silent:1;
+			unsigned int		op_active:1;
+			unsigned int		op_bytes;
+			unsigned int		op_nents;
+			unsigned int		op_count;
+			struct scatterlist	*op_sg;
+			struct rds_notifier	*op_notifier;
+
+			struct rds_mr		*op_rdma_mr;
+		} rdma;
+		struct rm_data_op {
+			unsigned int		op_active:1;
+			unsigned int		op_nents;
+			unsigned int		op_count;
+			struct scatterlist	*op_sg;
+		} data;
+	};
+};
+
+/*
+ * The RDS notifier is used (optionally) to tell the application about
+ * completed RDMA operations. Rather than keeping the whole rds message
+ * around on the queue, we allocate a small notifier that is put on the
+ * socket's notifier_list. Notifications are delivered to the application
+ * through control messages.
+ */
+struct rds_notifier {
+	struct list_head	n_list;
+	uint64_t		n_user_token;
+	int			n_status;
+};
+
+/**
+ * struct rds_transport -  transport specific behavioural hooks
+ *
+ * @xmit: .xmit is called by rds_send_xmit() to tell the transport to send
+ *        part of a message.  The caller serializes on the send_sem so this
+ *        doesn't need to be reentrant for a given conn.  The header must be
+ *        sent before the data payload.  .xmit must be prepared to send a
+ *        message with no data payload.  .xmit should return the number of
+ *        bytes that were sent down the connection, including header bytes.
+ *        Returning 0 tells the caller that it doesn't need to perform any
+ *        additional work now.  This is usually the case when the transport has
+ *        filled the sending queue for its connection and will handle
+ *        triggering the rds thread to continue the send when space becomes
+ *        available.  Returning -EAGAIN tells the caller to retry the send
+ *        immediately.  Returning -ENOMEM tells the caller to retry the send at
+ *        some point in the future.
+ *
+ * @conn_shutdown: conn_shutdown stops traffic on the given connection.  Once
+ *                 it returns the connection can not call rds_recv_incoming().
+ *                 This will only be called once after conn_connect returns
+ *                 non-zero success and will The caller serializes this with
+ *                 the send and connecting paths (xmit_* and conn_*).  The
+ *                 transport is responsible for other serialization, including
+ *                 rds_recv_incoming().  This is called in process context but
+ *                 should try hard not to block.
+ */
+
+#define RDS_TRANS_IB	0
+#define RDS_TRANS_IWARP	1
+#define RDS_TRANS_TCP	2
+#define RDS_TRANS_COUNT	3
+
+struct rds_transport {
+	char			t_name[TRANSNAMSIZ];
+	struct list_head	t_item;
+	struct module		*t_owner;
+	unsigned int		t_prefer_loopback:1;
+	unsigned int		t_type;
+
+	int (*laddr_check)(__be32 addr);
+	int (*conn_alloc)(struct rds_connection *conn, gfp_t gfp);
+	void (*conn_free)(void *data);
+	int (*conn_connect)(struct rds_connection *conn);
+	void (*conn_shutdown)(struct rds_connection *conn);
+	void (*xmit_prepare)(struct rds_connection *conn);
+	void (*xmit_complete)(struct rds_connection *conn);
+	int (*xmit)(struct rds_connection *conn, struct rds_message *rm,
+		    unsigned int hdr_off, unsigned int sg, unsigned int off);
+	int (*xmit_rdma)(struct rds_connection *conn, struct rm_rdma_op *op);
+	int (*xmit_atomic)(struct rds_connection *conn, struct rm_atomic_op *op);
+	int (*recv)(struct rds_connection *conn);
+	int (*inc_copy_to_user)(struct rds_incoming *inc, struct iov_iter *to);
+	void (*inc_free)(struct rds_incoming *inc);
+
+	int (*cm_handle_connect)(struct rdma_cm_id *cm_id,
+				 struct rdma_cm_event *event);
+	int (*cm_initiate_connect)(struct rdma_cm_id *cm_id);
+	void (*cm_connect_complete)(struct rds_connection *conn,
+				    struct rdma_cm_event *event);
+
+	unsigned int (*stats_info_copy)(struct rds_info_iterator *iter,
+					unsigned int avail);
+	void (*exit)(void);
+	void *(*get_mr)(struct scatterlist *sg, unsigned long nr_sg,
+			struct rds_sock *rs, u32 *key_ret);
+	void (*sync_mr)(void *trans_private, int direction);
+	void (*free_mr)(void *trans_private, int invalidate);
+	void (*flush_mrs)(void);
+};
+
+struct rds_sock {
+	struct sock		rs_sk;
+
+	u64			rs_user_addr;
+	u64			rs_user_bytes;
+
+	/*
+	 * bound_addr used for both incoming and outgoing, no INADDR_ANY
+	 * support.
+	 */
+	struct hlist_node	rs_bound_node;
+	__be32			rs_bound_addr;
+	__be32			rs_conn_addr;
+	__be16			rs_bound_port;
+	__be16			rs_conn_port;
+	struct rds_transport    *rs_transport;
+
+	/*
+	 * rds_sendmsg caches the conn it used the last time around.
+	 * This helps avoid costly lookups.
+	 */
+	struct rds_connection	*rs_conn;
+
+	/* flag indicating we were congested or not */
+	int			rs_congested;
+	/* seen congestion (ENOBUFS) when sending? */
+	int			rs_seen_congestion;
+
+	/* rs_lock protects all these adjacent members before the newline */
+	spinlock_t		rs_lock;
+	struct list_head	rs_send_queue;
+	u32			rs_snd_bytes;
+	int			rs_rcv_bytes;
+	struct list_head	rs_notify_queue;	/* currently used for failed RDMAs */
+
+	/* Congestion wake_up. If rs_cong_monitor is set, we use cong_mask
+	 * to decide whether the application should be woken up.
+	 * If not set, we use rs_cong_track to find out whether a cong map
+	 * update arrived.
+	 */
+	uint64_t		rs_cong_mask;
+	uint64_t		rs_cong_notify;
+	struct list_head	rs_cong_list;
+	unsigned long		rs_cong_track;
+
+	/*
+	 * rs_recv_lock protects the receive queue, and is
+	 * used to serialize with rds_release.
+	 */
+	rwlock_t		rs_recv_lock;
+	struct list_head	rs_recv_queue;
+
+	/* just for stats reporting */
+	struct list_head	rs_item;
+
+	/* these have their own lock */
+	spinlock_t		rs_rdma_lock;
+	struct rb_root		rs_rdma_keys;
+
+	/* Socket options - in case there will be more */
+	unsigned char		rs_recverr,
+				rs_cong_monitor;
+};
+
+static inline struct rds_sock *rds_sk_to_rs(const struct sock *sk)
+{
+	return container_of(sk, struct rds_sock, rs_sk);
+}
+static inline struct sock *rds_rs_to_sk(struct rds_sock *rs)
+{
+	return &rs->rs_sk;
+}
+
+/*
+ * The stack assigns sk_sndbuf and sk_rcvbuf to twice the specified value
+ * to account for overhead.  We don't account for overhead, we just apply
+ * the number of payload bytes to the specified value.
+ */
+static inline int rds_sk_sndbuf(struct rds_sock *rs)
+{
+	return rds_rs_to_sk(rs)->sk_sndbuf / 2;
+}
+static inline int rds_sk_rcvbuf(struct rds_sock *rs)
+{
+	return rds_rs_to_sk(rs)->sk_rcvbuf / 2;
+}
+
+struct rds_statistics {
+	uint64_t	s_conn_reset;
+	uint64_t	s_recv_drop_bad_checksum;
+	uint64_t	s_recv_drop_old_seq;
+	uint64_t	s_recv_drop_no_sock;
+	uint64_t	s_recv_drop_dead_sock;
+	uint64_t	s_recv_deliver_raced;
+	uint64_t	s_recv_delivered;
+	uint64_t	s_recv_queued;
+	uint64_t	s_recv_immediate_retry;
+	uint64_t	s_recv_delayed_retry;
+	uint64_t	s_recv_ack_required;
+	uint64_t	s_recv_rdma_bytes;
+	uint64_t	s_recv_ping;
+	uint64_t	s_send_queue_empty;
+	uint64_t	s_send_queue_full;
+	uint64_t	s_send_lock_contention;
+	uint64_t	s_send_lock_queue_raced;
+	uint64_t	s_send_immediate_retry;
+	uint64_t	s_send_delayed_retry;
+	uint64_t	s_send_drop_acked;
+	uint64_t	s_send_ack_required;
+	uint64_t	s_send_queued;
+	uint64_t	s_send_rdma;
+	uint64_t	s_send_rdma_bytes;
+	uint64_t	s_send_pong;
+	uint64_t	s_page_remainder_hit;
+	uint64_t	s_page_remainder_miss;
+	uint64_t	s_copy_to_user;
+	uint64_t	s_copy_from_user;
+	uint64_t	s_cong_update_queued;
+	uint64_t	s_cong_update_received;
+	uint64_t	s_cong_send_error;
+	uint64_t	s_cong_send_blocked;
+};
+
+/* af_rds.c */
+char *rds_str_array(char **array, size_t elements, size_t index);
+void rds_sock_addref(struct rds_sock *rs);
+void rds_sock_put(struct rds_sock *rs);
+void rds_wake_sk_sleep(struct rds_sock *rs);
+static inline void __rds_wake_sk_sleep(struct sock *sk)
+{
+	wait_queue_head_t *waitq = sk_sleep(sk);
+
+	if (!sock_flag(sk, SOCK_DEAD) && waitq)
+		wake_up(waitq);
+}
+extern wait_queue_head_t rds_poll_waitq;
+
+
+/* bind.c */
+int rds_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len);
+void rds_remove_bound(struct rds_sock *rs);
+struct rds_sock *rds_find_bound(__be32 addr, __be16 port);
+
+/* cong.c */
+int rds_cong_get_maps(struct rds_connection *conn);
+void rds_cong_add_conn(struct rds_connection *conn);
+void rds_cong_remove_conn(struct rds_connection *conn);
+void rds_cong_set_bit(struct rds_cong_map *map, __be16 port);
+void rds_cong_clear_bit(struct rds_cong_map *map, __be16 port);
+int rds_cong_wait(struct rds_cong_map *map, __be16 port, int nonblock, struct rds_sock *rs);
+void rds_cong_queue_updates(struct rds_cong_map *map);
+void rds_cong_map_updated(struct rds_cong_map *map, uint64_t);
+int rds_cong_updated_since(unsigned long *recent);
+void rds_cong_add_socket(struct rds_sock *);
+void rds_cong_remove_socket(struct rds_sock *);
+void rds_cong_exit(void);
+struct rds_message *rds_cong_update_alloc(struct rds_connection *conn);
+
+/* conn.c */
+int rds_conn_init(void);
+void rds_conn_exit(void);
+struct rds_connection *rds_conn_create(__be32 laddr, __be32 faddr,
+				       struct rds_transport *trans, gfp_t gfp);
+struct rds_connection *rds_conn_create_outgoing(__be32 laddr, __be32 faddr,
+			       struct rds_transport *trans, gfp_t gfp);
+void rds_conn_shutdown(struct rds_connection *conn);
+void rds_conn_destroy(struct rds_connection *conn);
+void rds_conn_drop(struct rds_connection *conn);
+void rds_conn_connect_if_down(struct rds_connection *conn);
+void rds_for_each_conn_info(struct socket *sock, unsigned int len,
+			  struct rds_info_iterator *iter,
+			  struct rds_info_lengths *lens,
+			  int (*visitor)(struct rds_connection *, void *),
+			  size_t item_len);
+__printf(2, 3)
+void __rds_conn_error(struct rds_connection *conn, const char *, ...);
+#define rds_conn_error(conn, fmt...) \
+	__rds_conn_error(conn, KERN_WARNING "RDS: " fmt)
+
+static inline int
+rds_conn_transition(struct rds_connection *conn, int old, int new)
+{
+	return atomic_cmpxchg(&conn->c_state, old, new) == old;
+}
+
+static inline int
+rds_conn_state(struct rds_connection *conn)
+{
+	return atomic_read(&conn->c_state);
+}
+
+static inline int
+rds_conn_up(struct rds_connection *conn)
+{
+	return atomic_read(&conn->c_state) == RDS_CONN_UP;
+}
+
+static inline int
+rds_conn_connecting(struct rds_connection *conn)
+{
+	return atomic_read(&conn->c_state) == RDS_CONN_CONNECTING;
+}
+
+/* message.c */
+struct rds_message *rds_message_alloc(unsigned int nents, gfp_t gfp);
+struct scatterlist *rds_message_alloc_sgs(struct rds_message *rm, int nents);
+int rds_message_copy_from_user(struct rds_message *rm, struct iov_iter *from);
+struct rds_message *rds_message_map_pages(unsigned long *page_addrs, unsigned int total_len);
+void rds_message_populate_header(struct rds_header *hdr, __be16 sport,
+				 __be16 dport, u64 seq);
+int rds_message_add_extension(struct rds_header *hdr,
+			      unsigned int type, const void *data, unsigned int len);
+int rds_message_next_extension(struct rds_header *hdr,
+			       unsigned int *pos, void *buf, unsigned int *buflen);
+int rds_message_add_rdma_dest_extension(struct rds_header *hdr, u32 r_key, u32 offset);
+int rds_message_inc_copy_to_user(struct rds_incoming *inc, struct iov_iter *to);
+void rds_message_inc_free(struct rds_incoming *inc);
+void rds_message_addref(struct rds_message *rm);
+void rds_message_put(struct rds_message *rm);
+void rds_message_wait(struct rds_message *rm);
+void rds_message_unmapped(struct rds_message *rm);
+
+static inline void rds_message_make_checksum(struct rds_header *hdr)
+{
+	hdr->h_csum = 0;
+	hdr->h_csum = ip_fast_csum((void *) hdr, sizeof(*hdr) >> 2);
+}
+
+static inline int rds_message_verify_checksum(const struct rds_header *hdr)
+{
+	return !hdr->h_csum || ip_fast_csum((void *) hdr, sizeof(*hdr) >> 2) == 0;
+}
+
+
+/* page.c */
+int rds_page_remainder_alloc(struct scatterlist *scat, unsigned long bytes,
+			     gfp_t gfp);
+int rds_page_copy_user(struct page *page, unsigned long offset,
+		       void __user *ptr, unsigned long bytes,
+		       int to_user);
+#define rds_page_copy_to_user(page, offset, ptr, bytes) \
+	rds_page_copy_user(page, offset, ptr, bytes, 1)
+#define rds_page_copy_from_user(page, offset, ptr, bytes) \
+	rds_page_copy_user(page, offset, ptr, bytes, 0)
+void rds_page_exit(void);
+
+/* recv.c */
+void rds_inc_init(struct rds_incoming *inc, struct rds_connection *conn,
+		  __be32 saddr);
+void rds_inc_put(struct rds_incoming *inc);
+void rds_recv_incoming(struct rds_connection *conn, __be32 saddr, __be32 daddr,
+		       struct rds_incoming *inc, gfp_t gfp);
+int rds_recvmsg(struct socket *sock, struct msghdr *msg, size_t size,
+		int msg_flags);
+void rds_clear_recv_queue(struct rds_sock *rs);
+int rds_notify_queue_get(struct rds_sock *rs, struct msghdr *msg);
+void rds_inc_info_copy(struct rds_incoming *inc,
+		       struct rds_info_iterator *iter,
+		       __be32 saddr, __be32 daddr, int flip);
+
+/* send.c */
+int rds_sendmsg(struct socket *sock, struct msghdr *msg, size_t payload_len);
+void rds_send_reset(struct rds_connection *conn);
+int rds_send_xmit(struct rds_connection *conn);
+struct sockaddr_in;
+void rds_send_drop_to(struct rds_sock *rs, struct sockaddr_in *dest);
+typedef int (*is_acked_func)(struct rds_message *rm, uint64_t ack);
+void rds_send_drop_acked(struct rds_connection *conn, u64 ack,
+			 is_acked_func is_acked);
+int rds_send_pong(struct rds_connection *conn, __be16 dport);
+struct rds_message *rds_send_get_message(struct rds_connection *,
+					 struct rm_rdma_op *);
+
+/* rdma.c */
+void rds_rdma_unuse(struct rds_sock *rs, u32 r_key, int force);
+int rds_get_mr(struct rds_sock *rs, char __user *optval, int optlen);
+int rds_get_mr_for_dest(struct rds_sock *rs, char __user *optval, int optlen);
+int rds_free_mr(struct rds_sock *rs, char __user *optval, int optlen);
+void rds_rdma_drop_keys(struct rds_sock *rs);
+int rds_rdma_extra_size(struct rds_rdma_args *args);
+int rds_cmsg_rdma_args(struct rds_sock *rs, struct rds_message *rm,
+			  struct cmsghdr *cmsg);
+int rds_cmsg_rdma_dest(struct rds_sock *rs, struct rds_message *rm,
+			  struct cmsghdr *cmsg);
+int rds_cmsg_rdma_args(struct rds_sock *rs, struct rds_message *rm,
+			  struct cmsghdr *cmsg);
+int rds_cmsg_rdma_map(struct rds_sock *rs, struct rds_message *rm,
+			  struct cmsghdr *cmsg);
+void rds_rdma_free_op(struct rm_rdma_op *ro);
+void rds_atomic_free_op(struct rm_atomic_op *ao);
+void rds_rdma_send_complete(struct rds_message *rm, int wc_status);
+void rds_atomic_send_complete(struct rds_message *rm, int wc_status);
+int rds_cmsg_atomic(struct rds_sock *rs, struct rds_message *rm,
+		    struct cmsghdr *cmsg);
+
+void __rds_put_mr_final(struct rds_mr *mr);
+static inline void rds_mr_put(struct rds_mr *mr)
+{
+	if (atomic_dec_and_test(&mr->r_refcount))
+		__rds_put_mr_final(mr);
+}
+
+/* stats.c */
+DECLARE_PER_CPU_SHARED_ALIGNED(struct rds_statistics, rds_stats);
+#define rds_stats_inc_which(which, member) do {		\
+	per_cpu(which, get_cpu()).member++;		\
+	put_cpu();					\
+} while (0)
+#define rds_stats_inc(member) rds_stats_inc_which(rds_stats, member)
+#define rds_stats_add_which(which, member, count) do {		\
+	per_cpu(which, get_cpu()).member += count;	\
+	put_cpu();					\
+} while (0)
+#define rds_stats_add(member, count) rds_stats_add_which(rds_stats, member, count)
+int rds_stats_init(void);
+void rds_stats_exit(void);
+void rds_stats_info_copy(struct rds_info_iterator *iter,
+			 uint64_t *values, const char *const *names,
+			 size_t nr);
+
+/* sysctl.c */
+int rds_sysctl_init(void);
+void rds_sysctl_exit(void);
+extern unsigned long rds_sysctl_sndbuf_min;
+extern unsigned long rds_sysctl_sndbuf_default;
+extern unsigned long rds_sysctl_sndbuf_max;
+extern unsigned long rds_sysctl_reconnect_min_jiffies;
+extern unsigned long rds_sysctl_reconnect_max_jiffies;
+extern unsigned int  rds_sysctl_max_unacked_packets;
+extern unsigned int  rds_sysctl_max_unacked_bytes;
+extern unsigned int  rds_sysctl_ping_enable;
+extern unsigned long rds_sysctl_trace_flags;
+extern unsigned int  rds_sysctl_trace_level;
+
+/* threads.c */
+int rds_threads_init(void);
+void rds_threads_exit(void);
+extern struct workqueue_struct *rds_wq;
+void rds_queue_reconnect(struct rds_connection *conn);
+void rds_connect_worker(struct work_struct *);
+void rds_shutdown_worker(struct work_struct *);
+void rds_send_worker(struct work_struct *);
+void rds_recv_worker(struct work_struct *);
+void rds_connect_complete(struct rds_connection *conn);
+
+/* transport.c */
+int rds_trans_register(struct rds_transport *trans);
+void rds_trans_unregister(struct rds_transport *trans);
+struct rds_transport *rds_trans_get_preferred(__be32 addr);
+void rds_trans_put(struct rds_transport *trans);
+unsigned int rds_trans_stats_info_copy(struct rds_info_iterator *iter,
+				       unsigned int avail);
+int rds_trans_init(void);
+void rds_trans_exit(void);
+
+#endif
diff --git a/net/rds/recv.c b/net/rds/recv.c
new file mode 100644
index 000000000..a00462b0d
--- /dev/null
+++ b/net/rds/recv.c
@@ -0,0 +1,549 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <net/sock.h>
+#include <linux/in.h>
+#include <linux/export.h>
+
+#include "rds.h"
+
+void rds_inc_init(struct rds_incoming *inc, struct rds_connection *conn,
+		  __be32 saddr)
+{
+	atomic_set(&inc->i_refcount, 1);
+	INIT_LIST_HEAD(&inc->i_item);
+	inc->i_conn = conn;
+	inc->i_saddr = saddr;
+	inc->i_rdma_cookie = 0;
+}
+EXPORT_SYMBOL_GPL(rds_inc_init);
+
+static void rds_inc_addref(struct rds_incoming *inc)
+{
+	rdsdebug("addref inc %p ref %d\n", inc, atomic_read(&inc->i_refcount));
+	atomic_inc(&inc->i_refcount);
+}
+
+void rds_inc_put(struct rds_incoming *inc)
+{
+	rdsdebug("put inc %p ref %d\n", inc, atomic_read(&inc->i_refcount));
+	if (atomic_dec_and_test(&inc->i_refcount)) {
+		BUG_ON(!list_empty(&inc->i_item));
+
+		inc->i_conn->c_trans->inc_free(inc);
+	}
+}
+EXPORT_SYMBOL_GPL(rds_inc_put);
+
+static void rds_recv_rcvbuf_delta(struct rds_sock *rs, struct sock *sk,
+				  struct rds_cong_map *map,
+				  int delta, __be16 port)
+{
+	int now_congested;
+
+	if (delta == 0)
+		return;
+
+	rs->rs_rcv_bytes += delta;
+	now_congested = rs->rs_rcv_bytes > rds_sk_rcvbuf(rs);
+
+	rdsdebug("rs %p (%pI4:%u) recv bytes %d buf %d "
+	  "now_cong %d delta %d\n",
+	  rs, &rs->rs_bound_addr,
+	  ntohs(rs->rs_bound_port), rs->rs_rcv_bytes,
+	  rds_sk_rcvbuf(rs), now_congested, delta);
+
+	/* wasn't -> am congested */
+	if (!rs->rs_congested && now_congested) {
+		rs->rs_congested = 1;
+		rds_cong_set_bit(map, port);
+		rds_cong_queue_updates(map);
+	}
+	/* was -> aren't congested */
+	/* Require more free space before reporting uncongested to prevent
+	   bouncing cong/uncong state too often */
+	else if (rs->rs_congested && (rs->rs_rcv_bytes < (rds_sk_rcvbuf(rs)/2))) {
+		rs->rs_congested = 0;
+		rds_cong_clear_bit(map, port);
+		rds_cong_queue_updates(map);
+	}
+
+	/* do nothing if no change in cong state */
+}
+
+/*
+ * Process all extension headers that come with this message.
+ */
+static void rds_recv_incoming_exthdrs(struct rds_incoming *inc, struct rds_sock *rs)
+{
+	struct rds_header *hdr = &inc->i_hdr;
+	unsigned int pos = 0, type, len;
+	union {
+		struct rds_ext_header_version version;
+		struct rds_ext_header_rdma rdma;
+		struct rds_ext_header_rdma_dest rdma_dest;
+	} buffer;
+
+	while (1) {
+		len = sizeof(buffer);
+		type = rds_message_next_extension(hdr, &pos, &buffer, &len);
+		if (type == RDS_EXTHDR_NONE)
+			break;
+		/* Process extension header here */
+		switch (type) {
+		case RDS_EXTHDR_RDMA:
+			rds_rdma_unuse(rs, be32_to_cpu(buffer.rdma.h_rdma_rkey), 0);
+			break;
+
+		case RDS_EXTHDR_RDMA_DEST:
+			/* We ignore the size for now. We could stash it
+			 * somewhere and use it for error checking. */
+			inc->i_rdma_cookie = rds_rdma_make_cookie(
+					be32_to_cpu(buffer.rdma_dest.h_rdma_rkey),
+					be32_to_cpu(buffer.rdma_dest.h_rdma_offset));
+
+			break;
+		}
+	}
+}
+
+/*
+ * The transport must make sure that this is serialized against other
+ * rx and conn reset on this specific conn.
+ *
+ * We currently assert that only one fragmented message will be sent
+ * down a connection at a time.  This lets us reassemble in the conn
+ * instead of per-flow which means that we don't have to go digging through
+ * flows to tear down partial reassembly progress on conn failure and
+ * we save flow lookup and locking for each frag arrival.  It does mean
+ * that small messages will wait behind large ones.  Fragmenting at all
+ * is only to reduce the memory consumption of pre-posted buffers.
+ *
+ * The caller passes in saddr and daddr instead of us getting it from the
+ * conn.  This lets loopback, who only has one conn for both directions,
+ * tell us which roles the addrs in the conn are playing for this message.
+ */
+void rds_recv_incoming(struct rds_connection *conn, __be32 saddr, __be32 daddr,
+		       struct rds_incoming *inc, gfp_t gfp)
+{
+	struct rds_sock *rs = NULL;
+	struct sock *sk;
+	unsigned long flags;
+
+	inc->i_conn = conn;
+	inc->i_rx_jiffies = jiffies;
+
+	rdsdebug("conn %p next %llu inc %p seq %llu len %u sport %u dport %u "
+		 "flags 0x%x rx_jiffies %lu\n", conn,
+		 (unsigned long long)conn->c_next_rx_seq,
+		 inc,
+		 (unsigned long long)be64_to_cpu(inc->i_hdr.h_sequence),
+		 be32_to_cpu(inc->i_hdr.h_len),
+		 be16_to_cpu(inc->i_hdr.h_sport),
+		 be16_to_cpu(inc->i_hdr.h_dport),
+		 inc->i_hdr.h_flags,
+		 inc->i_rx_jiffies);
+
+	/*
+	 * Sequence numbers should only increase.  Messages get their
+	 * sequence number as they're queued in a sending conn.  They
+	 * can be dropped, though, if the sending socket is closed before
+	 * they hit the wire.  So sequence numbers can skip forward
+	 * under normal operation.  They can also drop back in the conn
+	 * failover case as previously sent messages are resent down the
+	 * new instance of a conn.  We drop those, otherwise we have
+	 * to assume that the next valid seq does not come after a
+	 * hole in the fragment stream.
+	 *
+	 * The headers don't give us a way to realize if fragments of
+	 * a message have been dropped.  We assume that frags that arrive
+	 * to a flow are part of the current message on the flow that is
+	 * being reassembled.  This means that senders can't drop messages
+	 * from the sending conn until all their frags are sent.
+	 *
+	 * XXX we could spend more on the wire to get more robust failure
+	 * detection, arguably worth it to avoid data corruption.
+	 */
+	if (be64_to_cpu(inc->i_hdr.h_sequence) < conn->c_next_rx_seq &&
+	    (inc->i_hdr.h_flags & RDS_FLAG_RETRANSMITTED)) {
+		rds_stats_inc(s_recv_drop_old_seq);
+		goto out;
+	}
+	conn->c_next_rx_seq = be64_to_cpu(inc->i_hdr.h_sequence) + 1;
+
+	if (rds_sysctl_ping_enable && inc->i_hdr.h_dport == 0) {
+		rds_stats_inc(s_recv_ping);
+		rds_send_pong(conn, inc->i_hdr.h_sport);
+		goto out;
+	}
+
+	rs = rds_find_bound(daddr, inc->i_hdr.h_dport);
+	if (!rs) {
+		rds_stats_inc(s_recv_drop_no_sock);
+		goto out;
+	}
+
+	/* Process extension headers */
+	rds_recv_incoming_exthdrs(inc, rs);
+
+	/* We can be racing with rds_release() which marks the socket dead. */
+	sk = rds_rs_to_sk(rs);
+
+	/* serialize with rds_release -> sock_orphan */
+	write_lock_irqsave(&rs->rs_recv_lock, flags);
+	if (!sock_flag(sk, SOCK_DEAD)) {
+		rdsdebug("adding inc %p to rs %p's recv queue\n", inc, rs);
+		rds_stats_inc(s_recv_queued);
+		rds_recv_rcvbuf_delta(rs, sk, inc->i_conn->c_lcong,
+				      be32_to_cpu(inc->i_hdr.h_len),
+				      inc->i_hdr.h_dport);
+		rds_inc_addref(inc);
+		list_add_tail(&inc->i_item, &rs->rs_recv_queue);
+		__rds_wake_sk_sleep(sk);
+	} else {
+		rds_stats_inc(s_recv_drop_dead_sock);
+	}
+	write_unlock_irqrestore(&rs->rs_recv_lock, flags);
+
+out:
+	if (rs)
+		rds_sock_put(rs);
+}
+EXPORT_SYMBOL_GPL(rds_recv_incoming);
+
+/*
+ * be very careful here.  This is being called as the condition in
+ * wait_event_*() needs to cope with being called many times.
+ */
+static int rds_next_incoming(struct rds_sock *rs, struct rds_incoming **inc)
+{
+	unsigned long flags;
+
+	if (!*inc) {
+		read_lock_irqsave(&rs->rs_recv_lock, flags);
+		if (!list_empty(&rs->rs_recv_queue)) {
+			*inc = list_entry(rs->rs_recv_queue.next,
+					  struct rds_incoming,
+					  i_item);
+			rds_inc_addref(*inc);
+		}
+		read_unlock_irqrestore(&rs->rs_recv_lock, flags);
+	}
+
+	return *inc != NULL;
+}
+
+static int rds_still_queued(struct rds_sock *rs, struct rds_incoming *inc,
+			    int drop)
+{
+	struct sock *sk = rds_rs_to_sk(rs);
+	int ret = 0;
+	unsigned long flags;
+
+	write_lock_irqsave(&rs->rs_recv_lock, flags);
+	if (!list_empty(&inc->i_item)) {
+		ret = 1;
+		if (drop) {
+			/* XXX make sure this i_conn is reliable */
+			rds_recv_rcvbuf_delta(rs, sk, inc->i_conn->c_lcong,
+					      -be32_to_cpu(inc->i_hdr.h_len),
+					      inc->i_hdr.h_dport);
+			list_del_init(&inc->i_item);
+			rds_inc_put(inc);
+		}
+	}
+	write_unlock_irqrestore(&rs->rs_recv_lock, flags);
+
+	rdsdebug("inc %p rs %p still %d dropped %d\n", inc, rs, ret, drop);
+	return ret;
+}
+
+/*
+ * Pull errors off the error queue.
+ * If msghdr is NULL, we will just purge the error queue.
+ */
+int rds_notify_queue_get(struct rds_sock *rs, struct msghdr *msghdr)
+{
+	struct rds_notifier *notifier;
+	struct rds_rdma_notify cmsg = { 0 }; /* fill holes with zero */
+	unsigned int count = 0, max_messages = ~0U;
+	unsigned long flags;
+	LIST_HEAD(copy);
+	int err = 0;
+
+
+	/* put_cmsg copies to user space and thus may sleep. We can't do this
+	 * with rs_lock held, so first grab as many notifications as we can stuff
+	 * in the user provided cmsg buffer. We don't try to copy more, to avoid
+	 * losing notifications - except when the buffer is so small that it wouldn't
+	 * even hold a single notification. Then we give him as much of this single
+	 * msg as we can squeeze in, and set MSG_CTRUNC.
+	 */
+	if (msghdr) {
+		max_messages = msghdr->msg_controllen / CMSG_SPACE(sizeof(cmsg));
+		if (!max_messages)
+			max_messages = 1;
+	}
+
+	spin_lock_irqsave(&rs->rs_lock, flags);
+	while (!list_empty(&rs->rs_notify_queue) && count < max_messages) {
+		notifier = list_entry(rs->rs_notify_queue.next,
+				struct rds_notifier, n_list);
+		list_move(&notifier->n_list, &copy);
+		count++;
+	}
+	spin_unlock_irqrestore(&rs->rs_lock, flags);
+
+	if (!count)
+		return 0;
+
+	while (!list_empty(&copy)) {
+		notifier = list_entry(copy.next, struct rds_notifier, n_list);
+
+		if (msghdr) {
+			cmsg.user_token = notifier->n_user_token;
+			cmsg.status = notifier->n_status;
+
+			err = put_cmsg(msghdr, SOL_RDS, RDS_CMSG_RDMA_STATUS,
+				       sizeof(cmsg), &cmsg);
+			if (err)
+				break;
+		}
+
+		list_del_init(&notifier->n_list);
+		kfree(notifier);
+	}
+
+	/* If we bailed out because of an error in put_cmsg,
+	 * we may be left with one or more notifications that we
+	 * didn't process. Return them to the head of the list. */
+	if (!list_empty(&copy)) {
+		spin_lock_irqsave(&rs->rs_lock, flags);
+		list_splice(&copy, &rs->rs_notify_queue);
+		spin_unlock_irqrestore(&rs->rs_lock, flags);
+	}
+
+	return err;
+}
+
+/*
+ * Queue a congestion notification
+ */
+static int rds_notify_cong(struct rds_sock *rs, struct msghdr *msghdr)
+{
+	uint64_t notify = rs->rs_cong_notify;
+	unsigned long flags;
+	int err;
+
+	err = put_cmsg(msghdr, SOL_RDS, RDS_CMSG_CONG_UPDATE,
+			sizeof(notify), &notify);
+	if (err)
+		return err;
+
+	spin_lock_irqsave(&rs->rs_lock, flags);
+	rs->rs_cong_notify &= ~notify;
+	spin_unlock_irqrestore(&rs->rs_lock, flags);
+
+	return 0;
+}
+
+/*
+ * Receive any control messages.
+ */
+static int rds_cmsg_recv(struct rds_incoming *inc, struct msghdr *msg)
+{
+	int ret = 0;
+
+	if (inc->i_rdma_cookie) {
+		ret = put_cmsg(msg, SOL_RDS, RDS_CMSG_RDMA_DEST,
+				sizeof(inc->i_rdma_cookie), &inc->i_rdma_cookie);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+int rds_recvmsg(struct socket *sock, struct msghdr *msg, size_t size,
+		int msg_flags)
+{
+	struct sock *sk = sock->sk;
+	struct rds_sock *rs = rds_sk_to_rs(sk);
+	long timeo;
+	int ret = 0, nonblock = msg_flags & MSG_DONTWAIT;
+	DECLARE_SOCKADDR(struct sockaddr_in *, sin, msg->msg_name);
+	struct rds_incoming *inc = NULL;
+
+	/* udp_recvmsg()->sock_recvtimeo() gets away without locking too.. */
+	timeo = sock_rcvtimeo(sk, nonblock);
+
+	rdsdebug("size %zu flags 0x%x timeo %ld\n", size, msg_flags, timeo);
+
+	if (msg_flags & MSG_OOB)
+		goto out;
+
+	while (1) {
+		struct iov_iter save;
+		/* If there are pending notifications, do those - and nothing else */
+		if (!list_empty(&rs->rs_notify_queue)) {
+			ret = rds_notify_queue_get(rs, msg);
+			break;
+		}
+
+		if (rs->rs_cong_notify) {
+			ret = rds_notify_cong(rs, msg);
+			break;
+		}
+
+		if (!rds_next_incoming(rs, &inc)) {
+			if (nonblock) {
+				ret = -EAGAIN;
+				break;
+			}
+
+			timeo = wait_event_interruptible_timeout(*sk_sleep(sk),
+					(!list_empty(&rs->rs_notify_queue) ||
+					 rs->rs_cong_notify ||
+					 rds_next_incoming(rs, &inc)), timeo);
+			rdsdebug("recvmsg woke inc %p timeo %ld\n", inc,
+				 timeo);
+			if (timeo > 0 || timeo == MAX_SCHEDULE_TIMEOUT)
+				continue;
+
+			ret = timeo;
+			if (ret == 0)
+				ret = -ETIMEDOUT;
+			break;
+		}
+
+		rdsdebug("copying inc %p from %pI4:%u to user\n", inc,
+			 &inc->i_conn->c_faddr,
+			 ntohs(inc->i_hdr.h_sport));
+		save = msg->msg_iter;
+		ret = inc->i_conn->c_trans->inc_copy_to_user(inc, &msg->msg_iter);
+		if (ret < 0)
+			break;
+
+		/*
+		 * if the message we just copied isn't at the head of the
+		 * recv queue then someone else raced us to return it, try
+		 * to get the next message.
+		 */
+		if (!rds_still_queued(rs, inc, !(msg_flags & MSG_PEEK))) {
+			rds_inc_put(inc);
+			inc = NULL;
+			rds_stats_inc(s_recv_deliver_raced);
+			msg->msg_iter = save;
+			continue;
+		}
+
+		if (ret < be32_to_cpu(inc->i_hdr.h_len)) {
+			if (msg_flags & MSG_TRUNC)
+				ret = be32_to_cpu(inc->i_hdr.h_len);
+			msg->msg_flags |= MSG_TRUNC;
+		}
+
+		if (rds_cmsg_recv(inc, msg)) {
+			ret = -EFAULT;
+			goto out;
+		}
+
+		rds_stats_inc(s_recv_delivered);
+
+		if (sin) {
+			sin->sin_family = AF_INET;
+			sin->sin_port = inc->i_hdr.h_sport;
+			sin->sin_addr.s_addr = inc->i_saddr;
+			memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
+			msg->msg_namelen = sizeof(*sin);
+		}
+		break;
+	}
+
+	if (inc)
+		rds_inc_put(inc);
+
+out:
+	return ret;
+}
+
+/*
+ * The socket is being shut down and we're asked to drop messages that were
+ * queued for recvmsg.  The caller has unbound the socket so the receive path
+ * won't queue any more incoming fragments or messages on the socket.
+ */
+void rds_clear_recv_queue(struct rds_sock *rs)
+{
+	struct sock *sk = rds_rs_to_sk(rs);
+	struct rds_incoming *inc, *tmp;
+	unsigned long flags;
+
+	write_lock_irqsave(&rs->rs_recv_lock, flags);
+	list_for_each_entry_safe(inc, tmp, &rs->rs_recv_queue, i_item) {
+		rds_recv_rcvbuf_delta(rs, sk, inc->i_conn->c_lcong,
+				      -be32_to_cpu(inc->i_hdr.h_len),
+				      inc->i_hdr.h_dport);
+		list_del_init(&inc->i_item);
+		rds_inc_put(inc);
+	}
+	write_unlock_irqrestore(&rs->rs_recv_lock, flags);
+}
+
+/*
+ * inc->i_saddr isn't used here because it is only set in the receive
+ * path.
+ */
+void rds_inc_info_copy(struct rds_incoming *inc,
+		       struct rds_info_iterator *iter,
+		       __be32 saddr, __be32 daddr, int flip)
+{
+	struct rds_info_message minfo;
+
+	minfo.seq = be64_to_cpu(inc->i_hdr.h_sequence);
+	minfo.len = be32_to_cpu(inc->i_hdr.h_len);
+
+	if (flip) {
+		minfo.laddr = daddr;
+		minfo.faddr = saddr;
+		minfo.lport = inc->i_hdr.h_dport;
+		minfo.fport = inc->i_hdr.h_sport;
+	} else {
+		minfo.laddr = saddr;
+		minfo.faddr = daddr;
+		minfo.lport = inc->i_hdr.h_sport;
+		minfo.fport = inc->i_hdr.h_dport;
+	}
+
+	rds_info_copy(iter, &minfo, sizeof(minfo));
+}
diff --git a/net/rds/send.c b/net/rds/send.c
new file mode 100644
index 000000000..e9430f537
--- /dev/null
+++ b/net/rds/send.c
@@ -0,0 +1,1165 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/moduleparam.h>
+#include <linux/gfp.h>
+#include <net/sock.h>
+#include <linux/in.h>
+#include <linux/list.h>
+#include <linux/ratelimit.h>
+#include <linux/export.h>
+
+#include "rds.h"
+
+/* When transmitting messages in rds_send_xmit, we need to emerge from
+ * time to time and briefly release the CPU. Otherwise the softlock watchdog
+ * will kick our shin.
+ * Also, it seems fairer to not let one busy connection stall all the
+ * others.
+ *
+ * send_batch_count is the number of times we'll loop in send_xmit. Setting
+ * it to 0 will restore the old behavior (where we looped until we had
+ * drained the queue).
+ */
+static int send_batch_count = 64;
+module_param(send_batch_count, int, 0444);
+MODULE_PARM_DESC(send_batch_count, " batch factor when working the send queue");
+
+static void rds_send_remove_from_sock(struct list_head *messages, int status);
+
+/*
+ * Reset the send state.  Callers must ensure that this doesn't race with
+ * rds_send_xmit().
+ */
+void rds_send_reset(struct rds_connection *conn)
+{
+	struct rds_message *rm, *tmp;
+	unsigned long flags;
+
+	if (conn->c_xmit_rm) {
+		rm = conn->c_xmit_rm;
+		conn->c_xmit_rm = NULL;
+		/* Tell the user the RDMA op is no longer mapped by the
+		 * transport. This isn't entirely true (it's flushed out
+		 * independently) but as the connection is down, there's
+		 * no ongoing RDMA to/from that memory */
+		rds_message_unmapped(rm);
+		rds_message_put(rm);
+	}
+
+	conn->c_xmit_sg = 0;
+	conn->c_xmit_hdr_off = 0;
+	conn->c_xmit_data_off = 0;
+	conn->c_xmit_atomic_sent = 0;
+	conn->c_xmit_rdma_sent = 0;
+	conn->c_xmit_data_sent = 0;
+
+	conn->c_map_queued = 0;
+
+	conn->c_unacked_packets = rds_sysctl_max_unacked_packets;
+	conn->c_unacked_bytes = rds_sysctl_max_unacked_bytes;
+
+	/* Mark messages as retransmissions, and move them to the send q */
+	spin_lock_irqsave(&conn->c_lock, flags);
+	list_for_each_entry_safe(rm, tmp, &conn->c_retrans, m_conn_item) {
+		set_bit(RDS_MSG_ACK_REQUIRED, &rm->m_flags);
+		set_bit(RDS_MSG_RETRANSMITTED, &rm->m_flags);
+	}
+	list_splice_init(&conn->c_retrans, &conn->c_send_queue);
+	spin_unlock_irqrestore(&conn->c_lock, flags);
+}
+
+static int acquire_in_xmit(struct rds_connection *conn)
+{
+	return test_and_set_bit(RDS_IN_XMIT, &conn->c_flags) == 0;
+}
+
+static void release_in_xmit(struct rds_connection *conn)
+{
+	clear_bit(RDS_IN_XMIT, &conn->c_flags);
+	smp_mb__after_atomic();
+	/*
+	 * We don't use wait_on_bit()/wake_up_bit() because our waking is in a
+	 * hot path and finding waiters is very rare.  We don't want to walk
+	 * the system-wide hashed waitqueue buckets in the fast path only to
+	 * almost never find waiters.
+	 */
+	if (waitqueue_active(&conn->c_waitq))
+		wake_up_all(&conn->c_waitq);
+}
+
+/*
+ * We're making the conscious trade-off here to only send one message
+ * down the connection at a time.
+ *   Pro:
+ *      - tx queueing is a simple fifo list
+ *   	- reassembly is optional and easily done by transports per conn
+ *      - no per flow rx lookup at all, straight to the socket
+ *   	- less per-frag memory and wire overhead
+ *   Con:
+ *      - queued acks can be delayed behind large messages
+ *   Depends:
+ *      - small message latency is higher behind queued large messages
+ *      - large message latency isn't starved by intervening small sends
+ */
+int rds_send_xmit(struct rds_connection *conn)
+{
+	struct rds_message *rm;
+	unsigned long flags;
+	unsigned int tmp;
+	struct scatterlist *sg;
+	int ret = 0;
+	LIST_HEAD(to_be_dropped);
+	int batch_count;
+	unsigned long send_gen = 0;
+
+restart:
+	batch_count = 0;
+
+	/*
+	 * sendmsg calls here after having queued its message on the send
+	 * queue.  We only have one task feeding the connection at a time.  If
+	 * another thread is already feeding the queue then we back off.  This
+	 * avoids blocking the caller and trading per-connection data between
+	 * caches per message.
+	 */
+	if (!acquire_in_xmit(conn)) {
+		rds_stats_inc(s_send_lock_contention);
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	/*
+	 * we record the send generation after doing the xmit acquire.
+	 * if someone else manages to jump in and do some work, we'll use
+	 * this to avoid a goto restart farther down.
+	 *
+	 * The acquire_in_xmit() check above ensures that only one
+	 * caller can increment c_send_gen at any time.
+	 */
+	conn->c_send_gen++;
+	send_gen = conn->c_send_gen;
+
+	/*
+	 * rds_conn_shutdown() sets the conn state and then tests RDS_IN_XMIT,
+	 * we do the opposite to avoid races.
+	 */
+	if (!rds_conn_up(conn)) {
+		release_in_xmit(conn);
+		ret = 0;
+		goto out;
+	}
+
+	if (conn->c_trans->xmit_prepare)
+		conn->c_trans->xmit_prepare(conn);
+
+	/*
+	 * spin trying to push headers and data down the connection until
+	 * the connection doesn't make forward progress.
+	 */
+	while (1) {
+
+		rm = conn->c_xmit_rm;
+
+		/*
+		 * If between sending messages, we can send a pending congestion
+		 * map update.
+		 */
+		if (!rm && test_and_clear_bit(0, &conn->c_map_queued)) {
+			rm = rds_cong_update_alloc(conn);
+			if (IS_ERR(rm)) {
+				ret = PTR_ERR(rm);
+				break;
+			}
+			rm->data.op_active = 1;
+
+			conn->c_xmit_rm = rm;
+		}
+
+		/*
+		 * If not already working on one, grab the next message.
+		 *
+		 * c_xmit_rm holds a ref while we're sending this message down
+		 * the connction.  We can use this ref while holding the
+		 * send_sem.. rds_send_reset() is serialized with it.
+		 */
+		if (!rm) {
+			unsigned int len;
+
+			batch_count++;
+
+			/* we want to process as big a batch as we can, but
+			 * we also want to avoid softlockups.  If we've been
+			 * through a lot of messages, lets back off and see
+			 * if anyone else jumps in
+			 */
+			if (batch_count >= 1024)
+				goto over_batch;
+
+			spin_lock_irqsave(&conn->c_lock, flags);
+
+			if (!list_empty(&conn->c_send_queue)) {
+				rm = list_entry(conn->c_send_queue.next,
+						struct rds_message,
+						m_conn_item);
+				rds_message_addref(rm);
+
+				/*
+				 * Move the message from the send queue to the retransmit
+				 * list right away.
+				 */
+				list_move_tail(&rm->m_conn_item, &conn->c_retrans);
+			}
+
+			spin_unlock_irqrestore(&conn->c_lock, flags);
+
+			if (!rm)
+				break;
+
+			/* Unfortunately, the way Infiniband deals with
+			 * RDMA to a bad MR key is by moving the entire
+			 * queue pair to error state. We cold possibly
+			 * recover from that, but right now we drop the
+			 * connection.
+			 * Therefore, we never retransmit messages with RDMA ops.
+			 */
+			if (rm->rdma.op_active &&
+			    test_bit(RDS_MSG_RETRANSMITTED, &rm->m_flags)) {
+				spin_lock_irqsave(&conn->c_lock, flags);
+				if (test_and_clear_bit(RDS_MSG_ON_CONN, &rm->m_flags))
+					list_move(&rm->m_conn_item, &to_be_dropped);
+				spin_unlock_irqrestore(&conn->c_lock, flags);
+				continue;
+			}
+
+			/* Require an ACK every once in a while */
+			len = ntohl(rm->m_inc.i_hdr.h_len);
+			if (conn->c_unacked_packets == 0 ||
+			    conn->c_unacked_bytes < len) {
+				__set_bit(RDS_MSG_ACK_REQUIRED, &rm->m_flags);
+
+				conn->c_unacked_packets = rds_sysctl_max_unacked_packets;
+				conn->c_unacked_bytes = rds_sysctl_max_unacked_bytes;
+				rds_stats_inc(s_send_ack_required);
+			} else {
+				conn->c_unacked_bytes -= len;
+				conn->c_unacked_packets--;
+			}
+
+			conn->c_xmit_rm = rm;
+		}
+
+		/* The transport either sends the whole rdma or none of it */
+		if (rm->rdma.op_active && !conn->c_xmit_rdma_sent) {
+			rm->m_final_op = &rm->rdma;
+			ret = conn->c_trans->xmit_rdma(conn, &rm->rdma);
+			if (ret)
+				break;
+			conn->c_xmit_rdma_sent = 1;
+
+			/* The transport owns the mapped memory for now.
+			 * You can't unmap it while it's on the send queue */
+			set_bit(RDS_MSG_MAPPED, &rm->m_flags);
+		}
+
+		if (rm->atomic.op_active && !conn->c_xmit_atomic_sent) {
+			rm->m_final_op = &rm->atomic;
+			ret = conn->c_trans->xmit_atomic(conn, &rm->atomic);
+			if (ret)
+				break;
+			conn->c_xmit_atomic_sent = 1;
+
+			/* The transport owns the mapped memory for now.
+			 * You can't unmap it while it's on the send queue */
+			set_bit(RDS_MSG_MAPPED, &rm->m_flags);
+		}
+
+		/*
+		 * A number of cases require an RDS header to be sent
+		 * even if there is no data.
+		 * We permit 0-byte sends; rds-ping depends on this.
+		 * However, if there are exclusively attached silent ops,
+		 * we skip the hdr/data send, to enable silent operation.
+		 */
+		if (rm->data.op_nents == 0) {
+			int ops_present;
+			int all_ops_are_silent = 1;
+
+			ops_present = (rm->atomic.op_active || rm->rdma.op_active);
+			if (rm->atomic.op_active && !rm->atomic.op_silent)
+				all_ops_are_silent = 0;
+			if (rm->rdma.op_active && !rm->rdma.op_silent)
+				all_ops_are_silent = 0;
+
+			if (ops_present && all_ops_are_silent
+			    && !rm->m_rdma_cookie)
+				rm->data.op_active = 0;
+		}
+
+		if (rm->data.op_active && !conn->c_xmit_data_sent) {
+			rm->m_final_op = &rm->data;
+			ret = conn->c_trans->xmit(conn, rm,
+						  conn->c_xmit_hdr_off,
+						  conn->c_xmit_sg,
+						  conn->c_xmit_data_off);
+			if (ret <= 0)
+				break;
+
+			if (conn->c_xmit_hdr_off < sizeof(struct rds_header)) {
+				tmp = min_t(int, ret,
+					    sizeof(struct rds_header) -
+					    conn->c_xmit_hdr_off);
+				conn->c_xmit_hdr_off += tmp;
+				ret -= tmp;
+			}
+
+			sg = &rm->data.op_sg[conn->c_xmit_sg];
+			while (ret) {
+				tmp = min_t(int, ret, sg->length -
+						      conn->c_xmit_data_off);
+				conn->c_xmit_data_off += tmp;
+				ret -= tmp;
+				if (conn->c_xmit_data_off == sg->length) {
+					conn->c_xmit_data_off = 0;
+					sg++;
+					conn->c_xmit_sg++;
+					BUG_ON(ret != 0 &&
+					       conn->c_xmit_sg == rm->data.op_nents);
+				}
+			}
+
+			if (conn->c_xmit_hdr_off == sizeof(struct rds_header) &&
+			    (conn->c_xmit_sg == rm->data.op_nents))
+				conn->c_xmit_data_sent = 1;
+		}
+
+		/*
+		 * A rm will only take multiple times through this loop
+		 * if there is a data op. Thus, if the data is sent (or there was
+		 * none), then we're done with the rm.
+		 */
+		if (!rm->data.op_active || conn->c_xmit_data_sent) {
+			conn->c_xmit_rm = NULL;
+			conn->c_xmit_sg = 0;
+			conn->c_xmit_hdr_off = 0;
+			conn->c_xmit_data_off = 0;
+			conn->c_xmit_rdma_sent = 0;
+			conn->c_xmit_atomic_sent = 0;
+			conn->c_xmit_data_sent = 0;
+
+			rds_message_put(rm);
+		}
+	}
+
+over_batch:
+	if (conn->c_trans->xmit_complete)
+		conn->c_trans->xmit_complete(conn);
+	release_in_xmit(conn);
+
+	/* Nuke any messages we decided not to retransmit. */
+	if (!list_empty(&to_be_dropped)) {
+		/* irqs on here, so we can put(), unlike above */
+		list_for_each_entry(rm, &to_be_dropped, m_conn_item)
+			rds_message_put(rm);
+		rds_send_remove_from_sock(&to_be_dropped, RDS_RDMA_DROPPED);
+	}
+
+	/*
+	 * Other senders can queue a message after we last test the send queue
+	 * but before we clear RDS_IN_XMIT.  In that case they'd back off and
+	 * not try and send their newly queued message.  We need to check the
+	 * send queue after having cleared RDS_IN_XMIT so that their message
+	 * doesn't get stuck on the send queue.
+	 *
+	 * If the transport cannot continue (i.e ret != 0), then it must
+	 * call us when more room is available, such as from the tx
+	 * completion handler.
+	 *
+	 * We have an extra generation check here so that if someone manages
+	 * to jump in after our release_in_xmit, we'll see that they have done
+	 * some work and we will skip our goto
+	 */
+	if (ret == 0) {
+		smp_mb();
+		if (!list_empty(&conn->c_send_queue) &&
+		    send_gen == conn->c_send_gen) {
+			rds_stats_inc(s_send_lock_queue_raced);
+			goto restart;
+		}
+	}
+out:
+	return ret;
+}
+
+static void rds_send_sndbuf_remove(struct rds_sock *rs, struct rds_message *rm)
+{
+	u32 len = be32_to_cpu(rm->m_inc.i_hdr.h_len);
+
+	assert_spin_locked(&rs->rs_lock);
+
+	BUG_ON(rs->rs_snd_bytes < len);
+	rs->rs_snd_bytes -= len;
+
+	if (rs->rs_snd_bytes == 0)
+		rds_stats_inc(s_send_queue_empty);
+}
+
+static inline int rds_send_is_acked(struct rds_message *rm, u64 ack,
+				    is_acked_func is_acked)
+{
+	if (is_acked)
+		return is_acked(rm, ack);
+	return be64_to_cpu(rm->m_inc.i_hdr.h_sequence) <= ack;
+}
+
+/*
+ * This is pretty similar to what happens below in the ACK
+ * handling code - except that we call here as soon as we get
+ * the IB send completion on the RDMA op and the accompanying
+ * message.
+ */
+void rds_rdma_send_complete(struct rds_message *rm, int status)
+{
+	struct rds_sock *rs = NULL;
+	struct rm_rdma_op *ro;
+	struct rds_notifier *notifier;
+	unsigned long flags;
+
+	spin_lock_irqsave(&rm->m_rs_lock, flags);
+
+	ro = &rm->rdma;
+	if (test_bit(RDS_MSG_ON_SOCK, &rm->m_flags) &&
+	    ro->op_active && ro->op_notify && ro->op_notifier) {
+		notifier = ro->op_notifier;
+		rs = rm->m_rs;
+		sock_hold(rds_rs_to_sk(rs));
+
+		notifier->n_status = status;
+		spin_lock(&rs->rs_lock);
+		list_add_tail(&notifier->n_list, &rs->rs_notify_queue);
+		spin_unlock(&rs->rs_lock);
+
+		ro->op_notifier = NULL;
+	}
+
+	spin_unlock_irqrestore(&rm->m_rs_lock, flags);
+
+	if (rs) {
+		rds_wake_sk_sleep(rs);
+		sock_put(rds_rs_to_sk(rs));
+	}
+}
+EXPORT_SYMBOL_GPL(rds_rdma_send_complete);
+
+/*
+ * Just like above, except looks at atomic op
+ */
+void rds_atomic_send_complete(struct rds_message *rm, int status)
+{
+	struct rds_sock *rs = NULL;
+	struct rm_atomic_op *ao;
+	struct rds_notifier *notifier;
+	unsigned long flags;
+
+	spin_lock_irqsave(&rm->m_rs_lock, flags);
+
+	ao = &rm->atomic;
+	if (test_bit(RDS_MSG_ON_SOCK, &rm->m_flags)
+	    && ao->op_active && ao->op_notify && ao->op_notifier) {
+		notifier = ao->op_notifier;
+		rs = rm->m_rs;
+		sock_hold(rds_rs_to_sk(rs));
+
+		notifier->n_status = status;
+		spin_lock(&rs->rs_lock);
+		list_add_tail(&notifier->n_list, &rs->rs_notify_queue);
+		spin_unlock(&rs->rs_lock);
+
+		ao->op_notifier = NULL;
+	}
+
+	spin_unlock_irqrestore(&rm->m_rs_lock, flags);
+
+	if (rs) {
+		rds_wake_sk_sleep(rs);
+		sock_put(rds_rs_to_sk(rs));
+	}
+}
+EXPORT_SYMBOL_GPL(rds_atomic_send_complete);
+
+/*
+ * This is the same as rds_rdma_send_complete except we
+ * don't do any locking - we have all the ingredients (message,
+ * socket, socket lock) and can just move the notifier.
+ */
+static inline void
+__rds_send_complete(struct rds_sock *rs, struct rds_message *rm, int status)
+{
+	struct rm_rdma_op *ro;
+	struct rm_atomic_op *ao;
+
+	ro = &rm->rdma;
+	if (ro->op_active && ro->op_notify && ro->op_notifier) {
+		ro->op_notifier->n_status = status;
+		list_add_tail(&ro->op_notifier->n_list, &rs->rs_notify_queue);
+		ro->op_notifier = NULL;
+	}
+
+	ao = &rm->atomic;
+	if (ao->op_active && ao->op_notify && ao->op_notifier) {
+		ao->op_notifier->n_status = status;
+		list_add_tail(&ao->op_notifier->n_list, &rs->rs_notify_queue);
+		ao->op_notifier = NULL;
+	}
+
+	/* No need to wake the app - caller does this */
+}
+
+/*
+ * This is called from the IB send completion when we detect
+ * a RDMA operation that failed with remote access error.
+ * So speed is not an issue here.
+ */
+struct rds_message *rds_send_get_message(struct rds_connection *conn,
+					 struct rm_rdma_op *op)
+{
+	struct rds_message *rm, *tmp, *found = NULL;
+	unsigned long flags;
+
+	spin_lock_irqsave(&conn->c_lock, flags);
+
+	list_for_each_entry_safe(rm, tmp, &conn->c_retrans, m_conn_item) {
+		if (&rm->rdma == op) {
+			atomic_inc(&rm->m_refcount);
+			found = rm;
+			goto out;
+		}
+	}
+
+	list_for_each_entry_safe(rm, tmp, &conn->c_send_queue, m_conn_item) {
+		if (&rm->rdma == op) {
+			atomic_inc(&rm->m_refcount);
+			found = rm;
+			break;
+		}
+	}
+
+out:
+	spin_unlock_irqrestore(&conn->c_lock, flags);
+
+	return found;
+}
+EXPORT_SYMBOL_GPL(rds_send_get_message);
+
+/*
+ * This removes messages from the socket's list if they're on it.  The list
+ * argument must be private to the caller, we must be able to modify it
+ * without locks.  The messages must have a reference held for their
+ * position on the list.  This function will drop that reference after
+ * removing the messages from the 'messages' list regardless of if it found
+ * the messages on the socket list or not.
+ */
+static void rds_send_remove_from_sock(struct list_head *messages, int status)
+{
+	unsigned long flags;
+	struct rds_sock *rs = NULL;
+	struct rds_message *rm;
+
+	while (!list_empty(messages)) {
+		int was_on_sock = 0;
+
+		rm = list_entry(messages->next, struct rds_message,
+				m_conn_item);
+		list_del_init(&rm->m_conn_item);
+
+		/*
+		 * If we see this flag cleared then we're *sure* that someone
+		 * else beat us to removing it from the sock.  If we race
+		 * with their flag update we'll get the lock and then really
+		 * see that the flag has been cleared.
+		 *
+		 * The message spinlock makes sure nobody clears rm->m_rs
+		 * while we're messing with it. It does not prevent the
+		 * message from being removed from the socket, though.
+		 */
+		spin_lock_irqsave(&rm->m_rs_lock, flags);
+		if (!test_bit(RDS_MSG_ON_SOCK, &rm->m_flags))
+			goto unlock_and_drop;
+
+		if (rs != rm->m_rs) {
+			if (rs) {
+				rds_wake_sk_sleep(rs);
+				sock_put(rds_rs_to_sk(rs));
+			}
+			rs = rm->m_rs;
+			if (rs)
+				sock_hold(rds_rs_to_sk(rs));
+		}
+		if (!rs)
+			goto unlock_and_drop;
+		spin_lock(&rs->rs_lock);
+
+		if (test_and_clear_bit(RDS_MSG_ON_SOCK, &rm->m_flags)) {
+			struct rm_rdma_op *ro = &rm->rdma;
+			struct rds_notifier *notifier;
+
+			list_del_init(&rm->m_sock_item);
+			rds_send_sndbuf_remove(rs, rm);
+
+			if (ro->op_active && ro->op_notifier &&
+			       (ro->op_notify || (ro->op_recverr && status))) {
+				notifier = ro->op_notifier;
+				list_add_tail(&notifier->n_list,
+						&rs->rs_notify_queue);
+				if (!notifier->n_status)
+					notifier->n_status = status;
+				rm->rdma.op_notifier = NULL;
+			}
+			was_on_sock = 1;
+			rm->m_rs = NULL;
+		}
+		spin_unlock(&rs->rs_lock);
+
+unlock_and_drop:
+		spin_unlock_irqrestore(&rm->m_rs_lock, flags);
+		rds_message_put(rm);
+		if (was_on_sock)
+			rds_message_put(rm);
+	}
+
+	if (rs) {
+		rds_wake_sk_sleep(rs);
+		sock_put(rds_rs_to_sk(rs));
+	}
+}
+
+/*
+ * Transports call here when they've determined that the receiver queued
+ * messages up to, and including, the given sequence number.  Messages are
+ * moved to the retrans queue when rds_send_xmit picks them off the send
+ * queue. This means that in the TCP case, the message may not have been
+ * assigned the m_ack_seq yet - but that's fine as long as tcp_is_acked
+ * checks the RDS_MSG_HAS_ACK_SEQ bit.
+ */
+void rds_send_drop_acked(struct rds_connection *conn, u64 ack,
+			 is_acked_func is_acked)
+{
+	struct rds_message *rm, *tmp;
+	unsigned long flags;
+	LIST_HEAD(list);
+
+	spin_lock_irqsave(&conn->c_lock, flags);
+
+	list_for_each_entry_safe(rm, tmp, &conn->c_retrans, m_conn_item) {
+		if (!rds_send_is_acked(rm, ack, is_acked))
+			break;
+
+		list_move(&rm->m_conn_item, &list);
+		clear_bit(RDS_MSG_ON_CONN, &rm->m_flags);
+	}
+
+	/* order flag updates with spin locks */
+	if (!list_empty(&list))
+		smp_mb__after_atomic();
+
+	spin_unlock_irqrestore(&conn->c_lock, flags);
+
+	/* now remove the messages from the sock list as needed */
+	rds_send_remove_from_sock(&list, RDS_RDMA_SUCCESS);
+}
+EXPORT_SYMBOL_GPL(rds_send_drop_acked);
+
+void rds_send_drop_to(struct rds_sock *rs, struct sockaddr_in *dest)
+{
+	struct rds_message *rm, *tmp;
+	struct rds_connection *conn;
+	unsigned long flags;
+	LIST_HEAD(list);
+
+	/* get all the messages we're dropping under the rs lock */
+	spin_lock_irqsave(&rs->rs_lock, flags);
+
+	list_for_each_entry_safe(rm, tmp, &rs->rs_send_queue, m_sock_item) {
+		if (dest && (dest->sin_addr.s_addr != rm->m_daddr ||
+			     dest->sin_port != rm->m_inc.i_hdr.h_dport))
+			continue;
+
+		list_move(&rm->m_sock_item, &list);
+		rds_send_sndbuf_remove(rs, rm);
+		clear_bit(RDS_MSG_ON_SOCK, &rm->m_flags);
+	}
+
+	/* order flag updates with the rs lock */
+	smp_mb__after_atomic();
+
+	spin_unlock_irqrestore(&rs->rs_lock, flags);
+
+	if (list_empty(&list))
+		return;
+
+	/* Remove the messages from the conn */
+	list_for_each_entry(rm, &list, m_sock_item) {
+
+		conn = rm->m_inc.i_conn;
+
+		spin_lock_irqsave(&conn->c_lock, flags);
+		/*
+		 * Maybe someone else beat us to removing rm from the conn.
+		 * If we race with their flag update we'll get the lock and
+		 * then really see that the flag has been cleared.
+		 */
+		if (!test_and_clear_bit(RDS_MSG_ON_CONN, &rm->m_flags)) {
+			spin_unlock_irqrestore(&conn->c_lock, flags);
+			spin_lock_irqsave(&rm->m_rs_lock, flags);
+			rm->m_rs = NULL;
+			spin_unlock_irqrestore(&rm->m_rs_lock, flags);
+			continue;
+		}
+		list_del_init(&rm->m_conn_item);
+		spin_unlock_irqrestore(&conn->c_lock, flags);
+
+		/*
+		 * Couldn't grab m_rs_lock in top loop (lock ordering),
+		 * but we can now.
+		 */
+		spin_lock_irqsave(&rm->m_rs_lock, flags);
+
+		spin_lock(&rs->rs_lock);
+		__rds_send_complete(rs, rm, RDS_RDMA_CANCELED);
+		spin_unlock(&rs->rs_lock);
+
+		rm->m_rs = NULL;
+		spin_unlock_irqrestore(&rm->m_rs_lock, flags);
+
+		rds_message_put(rm);
+	}
+
+	rds_wake_sk_sleep(rs);
+
+	while (!list_empty(&list)) {
+		rm = list_entry(list.next, struct rds_message, m_sock_item);
+		list_del_init(&rm->m_sock_item);
+
+		rds_message_wait(rm);
+		rds_message_put(rm);
+	}
+}
+
+/*
+ * we only want this to fire once so we use the callers 'queued'.  It's
+ * possible that another thread can race with us and remove the
+ * message from the flow with RDS_CANCEL_SENT_TO.
+ */
+static int rds_send_queue_rm(struct rds_sock *rs, struct rds_connection *conn,
+			     struct rds_message *rm, __be16 sport,
+			     __be16 dport, int *queued)
+{
+	unsigned long flags;
+	u32 len;
+
+	if (*queued)
+		goto out;
+
+	len = be32_to_cpu(rm->m_inc.i_hdr.h_len);
+
+	/* this is the only place which holds both the socket's rs_lock
+	 * and the connection's c_lock */
+	spin_lock_irqsave(&rs->rs_lock, flags);
+
+	/*
+	 * If there is a little space in sndbuf, we don't queue anything,
+	 * and userspace gets -EAGAIN. But poll() indicates there's send
+	 * room. This can lead to bad behavior (spinning) if snd_bytes isn't
+	 * freed up by incoming acks. So we check the *old* value of
+	 * rs_snd_bytes here to allow the last msg to exceed the buffer,
+	 * and poll() now knows no more data can be sent.
+	 */
+	if (rs->rs_snd_bytes < rds_sk_sndbuf(rs)) {
+		rs->rs_snd_bytes += len;
+
+		/* let recv side know we are close to send space exhaustion.
+		 * This is probably not the optimal way to do it, as this
+		 * means we set the flag on *all* messages as soon as our
+		 * throughput hits a certain threshold.
+		 */
+		if (rs->rs_snd_bytes >= rds_sk_sndbuf(rs) / 2)
+			__set_bit(RDS_MSG_ACK_REQUIRED, &rm->m_flags);
+
+		list_add_tail(&rm->m_sock_item, &rs->rs_send_queue);
+		set_bit(RDS_MSG_ON_SOCK, &rm->m_flags);
+		rds_message_addref(rm);
+		rm->m_rs = rs;
+
+		/* The code ordering is a little weird, but we're
+		   trying to minimize the time we hold c_lock */
+		rds_message_populate_header(&rm->m_inc.i_hdr, sport, dport, 0);
+		rm->m_inc.i_conn = conn;
+		rds_message_addref(rm);
+
+		spin_lock(&conn->c_lock);
+		rm->m_inc.i_hdr.h_sequence = cpu_to_be64(conn->c_next_tx_seq++);
+		list_add_tail(&rm->m_conn_item, &conn->c_send_queue);
+		set_bit(RDS_MSG_ON_CONN, &rm->m_flags);
+		spin_unlock(&conn->c_lock);
+
+		rdsdebug("queued msg %p len %d, rs %p bytes %d seq %llu\n",
+			 rm, len, rs, rs->rs_snd_bytes,
+			 (unsigned long long)be64_to_cpu(rm->m_inc.i_hdr.h_sequence));
+
+		*queued = 1;
+	}
+
+	spin_unlock_irqrestore(&rs->rs_lock, flags);
+out:
+	return *queued;
+}
+
+/*
+ * 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)
+{
+	struct cmsghdr *cmsg;
+	int size = 0;
+	int cmsg_groups = 0;
+	int retval;
+
+	for_each_cmsghdr(cmsg, msg) {
+		if (!CMSG_OK(msg, cmsg))
+			return -EINVAL;
+
+		if (cmsg->cmsg_level != SOL_RDS)
+			continue;
+
+		switch (cmsg->cmsg_type) {
+		case RDS_CMSG_RDMA_ARGS:
+			cmsg_groups |= 1;
+			retval = rds_rdma_extra_size(CMSG_DATA(cmsg));
+			if (retval < 0)
+				return retval;
+			size += retval;
+
+			break;
+
+		case RDS_CMSG_RDMA_DEST:
+		case RDS_CMSG_RDMA_MAP:
+			cmsg_groups |= 2;
+			/* these are valid but do no add any size */
+			break;
+
+		case RDS_CMSG_ATOMIC_CSWP:
+		case RDS_CMSG_ATOMIC_FADD:
+		case RDS_CMSG_MASKED_ATOMIC_CSWP:
+		case RDS_CMSG_MASKED_ATOMIC_FADD:
+			cmsg_groups |= 1;
+			size += sizeof(struct scatterlist);
+			break;
+
+		default:
+			return -EINVAL;
+		}
+
+	}
+
+	size += ceil(data_len, PAGE_SIZE) * sizeof(struct scatterlist);
+
+	/* Ensure (DEST, MAP) are never used with (ARGS, ATOMIC) */
+	if (cmsg_groups == 3)
+		return -EINVAL;
+
+	return size;
+}
+
+static int rds_cmsg_send(struct rds_sock *rs, struct rds_message *rm,
+			 struct msghdr *msg, int *allocated_mr)
+{
+	struct cmsghdr *cmsg;
+	int ret = 0;
+
+	for_each_cmsghdr(cmsg, msg) {
+		if (!CMSG_OK(msg, cmsg))
+			return -EINVAL;
+
+		if (cmsg->cmsg_level != SOL_RDS)
+			continue;
+
+		/* As a side effect, RDMA_DEST and RDMA_MAP will set
+		 * rm->rdma.m_rdma_cookie and rm->rdma.m_rdma_mr.
+		 */
+		switch (cmsg->cmsg_type) {
+		case RDS_CMSG_RDMA_ARGS:
+			ret = rds_cmsg_rdma_args(rs, rm, cmsg);
+			break;
+
+		case RDS_CMSG_RDMA_DEST:
+			ret = rds_cmsg_rdma_dest(rs, rm, cmsg);
+			break;
+
+		case RDS_CMSG_RDMA_MAP:
+			ret = rds_cmsg_rdma_map(rs, rm, cmsg);
+			if (!ret)
+				*allocated_mr = 1;
+			break;
+		case RDS_CMSG_ATOMIC_CSWP:
+		case RDS_CMSG_ATOMIC_FADD:
+		case RDS_CMSG_MASKED_ATOMIC_CSWP:
+		case RDS_CMSG_MASKED_ATOMIC_FADD:
+			ret = rds_cmsg_atomic(rs, rm, cmsg);
+			break;
+
+		default:
+			return -EINVAL;
+		}
+
+		if (ret)
+			break;
+	}
+
+	return ret;
+}
+
+int rds_sendmsg(struct socket *sock, struct msghdr *msg, size_t payload_len)
+{
+	struct sock *sk = sock->sk;
+	struct rds_sock *rs = rds_sk_to_rs(sk);
+	DECLARE_SOCKADDR(struct sockaddr_in *, usin, msg->msg_name);
+	__be32 daddr;
+	__be16 dport;
+	struct rds_message *rm = NULL;
+	struct rds_connection *conn;
+	int ret = 0;
+	int queued = 0, allocated_mr = 0;
+	int nonblock = msg->msg_flags & MSG_DONTWAIT;
+	long timeo = sock_sndtimeo(sk, nonblock);
+
+	/* Mirror Linux UDP mirror of BSD error message compatibility */
+	/* XXX: Perhaps MSG_MORE someday */
+	if (msg->msg_flags & ~(MSG_DONTWAIT | MSG_CMSG_COMPAT)) {
+		ret = -EOPNOTSUPP;
+		goto out;
+	}
+
+	if (msg->msg_namelen) {
+		/* XXX fail non-unicast destination IPs? */
+		if (msg->msg_namelen < sizeof(*usin) || usin->sin_family != AF_INET) {
+			ret = -EINVAL;
+			goto out;
+		}
+		daddr = usin->sin_addr.s_addr;
+		dport = usin->sin_port;
+	} else {
+		/* We only care about consistency with ->connect() */
+		lock_sock(sk);
+		daddr = rs->rs_conn_addr;
+		dport = rs->rs_conn_port;
+		release_sock(sk);
+	}
+
+	/* racing with another thread binding seems ok here */
+	if (daddr == 0 || rs->rs_bound_addr == 0) {
+		ret = -ENOTCONN; /* XXX not a great errno */
+		goto out;
+	}
+
+	/* size of rm including all sgs */
+	ret = rds_rm_size(msg, payload_len);
+	if (ret < 0)
+		goto out;
+
+	rm = rds_message_alloc(ret, GFP_KERNEL);
+	if (!rm) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	/* Attach data to the rm */
+	if (payload_len) {
+		rm->data.op_sg = rds_message_alloc_sgs(rm, ceil(payload_len, PAGE_SIZE));
+		if (!rm->data.op_sg) {
+			ret = -ENOMEM;
+			goto out;
+		}
+		ret = rds_message_copy_from_user(rm, &msg->msg_iter);
+		if (ret)
+			goto out;
+	}
+	rm->data.op_active = 1;
+
+	rm->m_daddr = daddr;
+
+	/* rds_conn_create has a spinlock that runs with IRQ off.
+	 * Caching the conn in the socket helps a lot. */
+	if (rs->rs_conn && rs->rs_conn->c_faddr == daddr)
+		conn = rs->rs_conn;
+	else {
+		conn = rds_conn_create_outgoing(rs->rs_bound_addr, daddr,
+					rs->rs_transport,
+					sock->sk->sk_allocation);
+		if (IS_ERR(conn)) {
+			ret = PTR_ERR(conn);
+			goto out;
+		}
+		rs->rs_conn = conn;
+	}
+
+	/* Parse any control messages the user may have included. */
+	ret = rds_cmsg_send(rs, rm, msg, &allocated_mr);
+	if (ret)
+		goto out;
+
+	if (rm->rdma.op_active && !conn->c_trans->xmit_rdma) {
+		printk_ratelimited(KERN_NOTICE "rdma_op %p conn xmit_rdma %p\n",
+			       &rm->rdma, conn->c_trans->xmit_rdma);
+		ret = -EOPNOTSUPP;
+		goto out;
+	}
+
+	if (rm->atomic.op_active && !conn->c_trans->xmit_atomic) {
+		printk_ratelimited(KERN_NOTICE "atomic_op %p conn xmit_atomic %p\n",
+			       &rm->atomic, conn->c_trans->xmit_atomic);
+		ret = -EOPNOTSUPP;
+		goto out;
+	}
+
+	rds_conn_connect_if_down(conn);
+
+	ret = rds_cong_wait(conn->c_fcong, dport, nonblock, rs);
+	if (ret) {
+		rs->rs_seen_congestion = 1;
+		goto out;
+	}
+
+	while (!rds_send_queue_rm(rs, conn, rm, rs->rs_bound_port,
+				  dport, &queued)) {
+		rds_stats_inc(s_send_queue_full);
+		/* XXX make sure this is reasonable */
+		if (payload_len > rds_sk_sndbuf(rs)) {
+			ret = -EMSGSIZE;
+			goto out;
+		}
+		if (nonblock) {
+			ret = -EAGAIN;
+			goto out;
+		}
+
+		timeo = wait_event_interruptible_timeout(*sk_sleep(sk),
+					rds_send_queue_rm(rs, conn, rm,
+							  rs->rs_bound_port,
+							  dport,
+							  &queued),
+					timeo);
+		rdsdebug("sendmsg woke queued %d timeo %ld\n", queued, timeo);
+		if (timeo > 0 || timeo == MAX_SCHEDULE_TIMEOUT)
+			continue;
+
+		ret = timeo;
+		if (ret == 0)
+			ret = -ETIMEDOUT;
+		goto out;
+	}
+
+	/*
+	 * By now we've committed to the send.  We reuse rds_send_worker()
+	 * to retry sends in the rds thread if the transport asks us to.
+	 */
+	rds_stats_inc(s_send_queued);
+
+	if (!test_bit(RDS_LL_SEND_FULL, &conn->c_flags))
+		rds_send_xmit(conn);
+
+	rds_message_put(rm);
+	return payload_len;
+
+out:
+	/* If the user included a RDMA_MAP cmsg, we allocated a MR on the fly.
+	 * If the sendmsg goes through, we keep the MR. If it fails with EAGAIN
+	 * or in any other way, we need to destroy the MR again */
+	if (allocated_mr)
+		rds_rdma_unuse(rs, rds_rdma_cookie_key(rm->m_rdma_cookie), 1);
+
+	if (rm)
+		rds_message_put(rm);
+	return ret;
+}
+
+/*
+ * Reply to a ping packet.
+ */
+int
+rds_send_pong(struct rds_connection *conn, __be16 dport)
+{
+	struct rds_message *rm;
+	unsigned long flags;
+	int ret = 0;
+
+	rm = rds_message_alloc(0, GFP_ATOMIC);
+	if (!rm) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	rm->m_daddr = conn->c_faddr;
+	rm->data.op_active = 1;
+
+	rds_conn_connect_if_down(conn);
+
+	ret = rds_cong_wait(conn->c_fcong, dport, 1, NULL);
+	if (ret)
+		goto out;
+
+	spin_lock_irqsave(&conn->c_lock, flags);
+	list_add_tail(&rm->m_conn_item, &conn->c_send_queue);
+	set_bit(RDS_MSG_ON_CONN, &rm->m_flags);
+	rds_message_addref(rm);
+	rm->m_inc.i_conn = conn;
+
+	rds_message_populate_header(&rm->m_inc.i_hdr, 0, dport,
+				    conn->c_next_tx_seq);
+	conn->c_next_tx_seq++;
+	spin_unlock_irqrestore(&conn->c_lock, flags);
+
+	rds_stats_inc(s_send_queued);
+	rds_stats_inc(s_send_pong);
+
+	if (!test_bit(RDS_LL_SEND_FULL, &conn->c_flags))
+		queue_delayed_work(rds_wq, &conn->c_send_w, 0);
+
+	rds_message_put(rm);
+	return 0;
+
+out:
+	if (rm)
+		rds_message_put(rm);
+	return ret;
+}
diff --git a/net/rds/stats.c b/net/rds/stats.c
new file mode 100644
index 000000000..73be187d3
--- /dev/null
+++ b/net/rds/stats.c
@@ -0,0 +1,152 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/percpu.h>
+#include <linux/seq_file.h>
+#include <linux/proc_fs.h>
+#include <linux/export.h>
+
+#include "rds.h"
+
+DEFINE_PER_CPU_SHARED_ALIGNED(struct rds_statistics, rds_stats);
+EXPORT_PER_CPU_SYMBOL_GPL(rds_stats);
+
+/* :.,$s/unsigned long\>.*\<s_\(.*\);/"\1",/g */
+
+static const char *const rds_stat_names[] = {
+	"conn_reset",
+	"recv_drop_bad_checksum",
+	"recv_drop_old_seq",
+	"recv_drop_no_sock",
+	"recv_drop_dead_sock",
+	"recv_deliver_raced",
+	"recv_delivered",
+	"recv_queued",
+	"recv_immediate_retry",
+	"recv_delayed_retry",
+	"recv_ack_required",
+	"recv_rdma_bytes",
+	"recv_ping",
+	"send_queue_empty",
+	"send_queue_full",
+	"send_lock_contention",
+	"send_lock_queue_raced",
+	"send_immediate_retry",
+	"send_delayed_retry",
+	"send_drop_acked",
+	"send_ack_required",
+	"send_queued",
+	"send_rdma",
+	"send_rdma_bytes",
+	"send_pong",
+	"page_remainder_hit",
+	"page_remainder_miss",
+	"copy_to_user",
+	"copy_from_user",
+	"cong_update_queued",
+	"cong_update_received",
+	"cong_send_error",
+	"cong_send_blocked",
+};
+
+void rds_stats_info_copy(struct rds_info_iterator *iter,
+			 uint64_t *values, const char *const *names, size_t nr)
+{
+	struct rds_info_counter ctr;
+	size_t i;
+
+	for (i = 0; i < nr; i++) {
+		BUG_ON(strlen(names[i]) >= sizeof(ctr.name));
+		strncpy(ctr.name, names[i], sizeof(ctr.name) - 1);
+		ctr.name[sizeof(ctr.name) - 1] = '\0';
+		ctr.value = values[i];
+
+		rds_info_copy(iter, &ctr, sizeof(ctr));
+	}
+}
+EXPORT_SYMBOL_GPL(rds_stats_info_copy);
+
+/*
+ * This gives global counters across all the transports.  The strings
+ * are copied in so that the tool doesn't need knowledge of the specific
+ * stats that we're exporting.  Some are pretty implementation dependent
+ * and may change over time.  That doesn't stop them from being useful.
+ *
+ * This is the only function in the chain that knows about the byte granular
+ * length in userspace.  It converts it to number of stat entries that the
+ * rest of the functions operate in.
+ */
+static void rds_stats_info(struct socket *sock, unsigned int len,
+			   struct rds_info_iterator *iter,
+			   struct rds_info_lengths *lens)
+{
+	struct rds_statistics stats = {0, };
+	uint64_t *src;
+	uint64_t *sum;
+	size_t i;
+	int cpu;
+	unsigned int avail;
+
+	avail = len / sizeof(struct rds_info_counter);
+
+	if (avail < ARRAY_SIZE(rds_stat_names)) {
+		avail = 0;
+		goto trans;
+	}
+
+	for_each_online_cpu(cpu) {
+		src = (uint64_t *)&(per_cpu(rds_stats, cpu));
+		sum = (uint64_t *)&stats;
+		for (i = 0; i < sizeof(stats) / sizeof(uint64_t); i++)
+			*(sum++) += *(src++);
+	}
+
+	rds_stats_info_copy(iter, (uint64_t *)&stats, rds_stat_names,
+			    ARRAY_SIZE(rds_stat_names));
+	avail -= ARRAY_SIZE(rds_stat_names);
+
+trans:
+	lens->each = sizeof(struct rds_info_counter);
+	lens->nr = rds_trans_stats_info_copy(iter, avail) +
+		   ARRAY_SIZE(rds_stat_names);
+}
+
+void rds_stats_exit(void)
+{
+	rds_info_deregister_func(RDS_INFO_COUNTERS, rds_stats_info);
+}
+
+int rds_stats_init(void)
+{
+	rds_info_register_func(RDS_INFO_COUNTERS, rds_stats_info);
+	return 0;
+}
diff --git a/net/rds/sysctl.c b/net/rds/sysctl.c
new file mode 100644
index 000000000..c173f69e1
--- /dev/null
+++ b/net/rds/sysctl.c
@@ -0,0 +1,109 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/sysctl.h>
+#include <linux/proc_fs.h>
+
+#include "rds.h"
+
+static struct ctl_table_header *rds_sysctl_reg_table;
+
+static unsigned long rds_sysctl_reconnect_min = 1;
+static unsigned long rds_sysctl_reconnect_max = ~0UL;
+
+unsigned long rds_sysctl_reconnect_min_jiffies;
+unsigned long rds_sysctl_reconnect_max_jiffies = HZ;
+
+unsigned int  rds_sysctl_max_unacked_packets = 8;
+unsigned int  rds_sysctl_max_unacked_bytes = (16 << 20);
+
+unsigned int rds_sysctl_ping_enable = 1;
+
+static struct ctl_table rds_sysctl_rds_table[] = {
+	{
+		.procname       = "reconnect_min_delay_ms",
+		.data		= &rds_sysctl_reconnect_min_jiffies,
+		.maxlen         = sizeof(unsigned long),
+		.mode           = 0644,
+		.proc_handler   = proc_doulongvec_ms_jiffies_minmax,
+		.extra1		= &rds_sysctl_reconnect_min,
+		.extra2		= &rds_sysctl_reconnect_max_jiffies,
+	},
+	{
+		.procname       = "reconnect_max_delay_ms",
+		.data		= &rds_sysctl_reconnect_max_jiffies,
+		.maxlen         = sizeof(unsigned long),
+		.mode           = 0644,
+		.proc_handler   = proc_doulongvec_ms_jiffies_minmax,
+		.extra1		= &rds_sysctl_reconnect_min_jiffies,
+		.extra2		= &rds_sysctl_reconnect_max,
+	},
+	{
+		.procname	= "max_unacked_packets",
+		.data		= &rds_sysctl_max_unacked_packets,
+		.maxlen         = sizeof(int),
+		.mode           = 0644,
+		.proc_handler   = proc_dointvec,
+	},
+	{
+		.procname	= "max_unacked_bytes",
+		.data		= &rds_sysctl_max_unacked_bytes,
+		.maxlen         = sizeof(int),
+		.mode           = 0644,
+		.proc_handler   = proc_dointvec,
+	},
+	{
+		.procname	= "ping_enable",
+		.data		= &rds_sysctl_ping_enable,
+		.maxlen         = sizeof(int),
+		.mode           = 0644,
+		.proc_handler   = proc_dointvec,
+	},
+	{ }
+};
+
+void rds_sysctl_exit(void)
+{
+	unregister_net_sysctl_table(rds_sysctl_reg_table);
+}
+
+int rds_sysctl_init(void)
+{
+	rds_sysctl_reconnect_min = msecs_to_jiffies(1);
+	rds_sysctl_reconnect_min_jiffies = rds_sysctl_reconnect_min;
+
+	rds_sysctl_reg_table = register_net_sysctl(&init_net,"net/rds", rds_sysctl_rds_table);
+	if (!rds_sysctl_reg_table)
+		return -ENOMEM;
+	return 0;
+}
diff --git a/net/rds/tcp.c b/net/rds/tcp.c
new file mode 100644
index 000000000..edac9ef2b
--- /dev/null
+++ b/net/rds/tcp.c
@@ -0,0 +1,326 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/in.h>
+#include <linux/module.h>
+#include <net/tcp.h>
+
+#include "rds.h"
+#include "tcp.h"
+
+/* only for info exporting */
+static DEFINE_SPINLOCK(rds_tcp_tc_list_lock);
+static LIST_HEAD(rds_tcp_tc_list);
+static unsigned int rds_tcp_tc_count;
+
+/* Track rds_tcp_connection structs so they can be cleaned up */
+static DEFINE_SPINLOCK(rds_tcp_conn_lock);
+static LIST_HEAD(rds_tcp_conn_list);
+
+static struct kmem_cache *rds_tcp_conn_slab;
+
+#define RDS_TCP_DEFAULT_BUFSIZE (128 * 1024)
+
+/* doing it this way avoids calling tcp_sk() */
+void rds_tcp_nonagle(struct socket *sock)
+{
+	mm_segment_t oldfs = get_fs();
+	int val = 1;
+
+	set_fs(KERNEL_DS);
+	sock->ops->setsockopt(sock, SOL_TCP, TCP_NODELAY, (char __user *)&val,
+			      sizeof(val));
+	set_fs(oldfs);
+}
+
+void rds_tcp_tune(struct socket *sock)
+{
+	struct sock *sk = sock->sk;
+
+	rds_tcp_nonagle(sock);
+
+	/*
+	 * We're trying to saturate gigabit with the default,
+	 * see svc_sock_setbufsize().
+	 */
+	lock_sock(sk);
+	sk->sk_sndbuf = RDS_TCP_DEFAULT_BUFSIZE;
+	sk->sk_rcvbuf = RDS_TCP_DEFAULT_BUFSIZE;
+	sk->sk_userlocks |= SOCK_SNDBUF_LOCK|SOCK_RCVBUF_LOCK;
+	release_sock(sk);
+}
+
+u32 rds_tcp_snd_nxt(struct rds_tcp_connection *tc)
+{
+	return tcp_sk(tc->t_sock->sk)->snd_nxt;
+}
+
+u32 rds_tcp_snd_una(struct rds_tcp_connection *tc)
+{
+	return tcp_sk(tc->t_sock->sk)->snd_una;
+}
+
+void rds_tcp_restore_callbacks(struct socket *sock,
+			       struct rds_tcp_connection *tc)
+{
+	rdsdebug("restoring sock %p callbacks from tc %p\n", sock, tc);
+	write_lock_bh(&sock->sk->sk_callback_lock);
+
+	/* done under the callback_lock to serialize with write_space */
+	spin_lock(&rds_tcp_tc_list_lock);
+	list_del_init(&tc->t_list_item);
+	rds_tcp_tc_count--;
+	spin_unlock(&rds_tcp_tc_list_lock);
+
+	tc->t_sock = NULL;
+
+	sock->sk->sk_write_space = tc->t_orig_write_space;
+	sock->sk->sk_data_ready = tc->t_orig_data_ready;
+	sock->sk->sk_state_change = tc->t_orig_state_change;
+	sock->sk->sk_user_data = NULL;
+
+	write_unlock_bh(&sock->sk->sk_callback_lock);
+}
+
+/*
+ * This is the only path that sets tc->t_sock.  Send and receive trust that
+ * it is set.  The RDS_CONN_CONNECTED bit protects those paths from being
+ * called while it isn't set.
+ */
+void rds_tcp_set_callbacks(struct socket *sock, struct rds_connection *conn)
+{
+	struct rds_tcp_connection *tc = conn->c_transport_data;
+
+	rdsdebug("setting sock %p callbacks to tc %p\n", sock, tc);
+	write_lock_bh(&sock->sk->sk_callback_lock);
+
+	/* done under the callback_lock to serialize with write_space */
+	spin_lock(&rds_tcp_tc_list_lock);
+	list_add_tail(&tc->t_list_item, &rds_tcp_tc_list);
+	rds_tcp_tc_count++;
+	spin_unlock(&rds_tcp_tc_list_lock);
+
+	/* accepted sockets need our listen data ready undone */
+	if (sock->sk->sk_data_ready == rds_tcp_listen_data_ready)
+		sock->sk->sk_data_ready = sock->sk->sk_user_data;
+
+	tc->t_sock = sock;
+	tc->conn = conn;
+	tc->t_orig_data_ready = sock->sk->sk_data_ready;
+	tc->t_orig_write_space = sock->sk->sk_write_space;
+	tc->t_orig_state_change = sock->sk->sk_state_change;
+
+	sock->sk->sk_user_data = conn;
+	sock->sk->sk_data_ready = rds_tcp_data_ready;
+	sock->sk->sk_write_space = rds_tcp_write_space;
+	sock->sk->sk_state_change = rds_tcp_state_change;
+
+	write_unlock_bh(&sock->sk->sk_callback_lock);
+}
+
+static void rds_tcp_tc_info(struct socket *sock, unsigned int len,
+			    struct rds_info_iterator *iter,
+			    struct rds_info_lengths *lens)
+{
+	struct rds_info_tcp_socket tsinfo;
+	struct rds_tcp_connection *tc;
+	unsigned long flags;
+	struct sockaddr_in sin;
+	int sinlen;
+
+	spin_lock_irqsave(&rds_tcp_tc_list_lock, flags);
+
+	if (len / sizeof(tsinfo) < rds_tcp_tc_count)
+		goto out;
+
+	list_for_each_entry(tc, &rds_tcp_tc_list, t_list_item) {
+
+		sock->ops->getname(sock, (struct sockaddr *)&sin, &sinlen, 0);
+		tsinfo.local_addr = sin.sin_addr.s_addr;
+		tsinfo.local_port = sin.sin_port;
+		sock->ops->getname(sock, (struct sockaddr *)&sin, &sinlen, 1);
+		tsinfo.peer_addr = sin.sin_addr.s_addr;
+		tsinfo.peer_port = sin.sin_port;
+
+		tsinfo.hdr_rem = tc->t_tinc_hdr_rem;
+		tsinfo.data_rem = tc->t_tinc_data_rem;
+		tsinfo.last_sent_nxt = tc->t_last_sent_nxt;
+		tsinfo.last_expected_una = tc->t_last_expected_una;
+		tsinfo.last_seen_una = tc->t_last_seen_una;
+
+		rds_info_copy(iter, &tsinfo, sizeof(tsinfo));
+	}
+
+out:
+	lens->nr = rds_tcp_tc_count;
+	lens->each = sizeof(tsinfo);
+
+	spin_unlock_irqrestore(&rds_tcp_tc_list_lock, flags);
+}
+
+static int rds_tcp_laddr_check(__be32 addr)
+{
+	if (inet_addr_type(&init_net, addr) == RTN_LOCAL)
+		return 0;
+	return -EADDRNOTAVAIL;
+}
+
+static int rds_tcp_conn_alloc(struct rds_connection *conn, gfp_t gfp)
+{
+	struct rds_tcp_connection *tc;
+
+	tc = kmem_cache_alloc(rds_tcp_conn_slab, gfp);
+	if (!tc)
+		return -ENOMEM;
+
+	tc->t_sock = NULL;
+	tc->t_tinc = NULL;
+	tc->t_tinc_hdr_rem = sizeof(struct rds_header);
+	tc->t_tinc_data_rem = 0;
+
+	conn->c_transport_data = tc;
+
+	spin_lock_irq(&rds_tcp_conn_lock);
+	list_add_tail(&tc->t_tcp_node, &rds_tcp_conn_list);
+	spin_unlock_irq(&rds_tcp_conn_lock);
+
+	rdsdebug("alloced tc %p\n", conn->c_transport_data);
+	return 0;
+}
+
+static void rds_tcp_conn_free(void *arg)
+{
+	struct rds_tcp_connection *tc = arg;
+	unsigned long flags;
+	rdsdebug("freeing tc %p\n", tc);
+
+	spin_lock_irqsave(&rds_tcp_conn_lock, flags);
+	list_del(&tc->t_tcp_node);
+	spin_unlock_irqrestore(&rds_tcp_conn_lock, flags);
+
+	kmem_cache_free(rds_tcp_conn_slab, tc);
+}
+
+static void rds_tcp_destroy_conns(void)
+{
+	struct rds_tcp_connection *tc, *_tc;
+	LIST_HEAD(tmp_list);
+
+	/* avoid calling conn_destroy with irqs off */
+	spin_lock_irq(&rds_tcp_conn_lock);
+	list_splice(&rds_tcp_conn_list, &tmp_list);
+	INIT_LIST_HEAD(&rds_tcp_conn_list);
+	spin_unlock_irq(&rds_tcp_conn_lock);
+
+	list_for_each_entry_safe(tc, _tc, &tmp_list, t_tcp_node) {
+		if (tc->conn->c_passive)
+			rds_conn_destroy(tc->conn->c_passive);
+		rds_conn_destroy(tc->conn);
+	}
+}
+
+static void rds_tcp_exit(void)
+{
+	rds_info_deregister_func(RDS_INFO_TCP_SOCKETS, rds_tcp_tc_info);
+	rds_tcp_listen_stop();
+	rds_tcp_destroy_conns();
+	rds_trans_unregister(&rds_tcp_transport);
+	rds_tcp_recv_exit();
+	kmem_cache_destroy(rds_tcp_conn_slab);
+}
+module_exit(rds_tcp_exit);
+
+struct rds_transport rds_tcp_transport = {
+	.laddr_check		= rds_tcp_laddr_check,
+	.xmit_prepare		= rds_tcp_xmit_prepare,
+	.xmit_complete		= rds_tcp_xmit_complete,
+	.xmit			= rds_tcp_xmit,
+	.recv			= rds_tcp_recv,
+	.conn_alloc		= rds_tcp_conn_alloc,
+	.conn_free		= rds_tcp_conn_free,
+	.conn_connect		= rds_tcp_conn_connect,
+	.conn_shutdown		= rds_tcp_conn_shutdown,
+	.inc_copy_to_user	= rds_tcp_inc_copy_to_user,
+	.inc_free		= rds_tcp_inc_free,
+	.stats_info_copy	= rds_tcp_stats_info_copy,
+	.exit			= rds_tcp_exit,
+	.t_owner		= THIS_MODULE,
+	.t_name			= "tcp",
+	.t_type			= RDS_TRANS_TCP,
+	.t_prefer_loopback	= 1,
+};
+
+static int rds_tcp_init(void)
+{
+	int ret;
+
+	rds_tcp_conn_slab = kmem_cache_create("rds_tcp_connection",
+					      sizeof(struct rds_tcp_connection),
+					      0, 0, NULL);
+	if (!rds_tcp_conn_slab) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	ret = rds_tcp_recv_init();
+	if (ret)
+		goto out_slab;
+
+	ret = rds_trans_register(&rds_tcp_transport);
+	if (ret)
+		goto out_recv;
+
+	ret = rds_tcp_listen_init();
+	if (ret)
+		goto out_register;
+
+	rds_info_register_func(RDS_INFO_TCP_SOCKETS, rds_tcp_tc_info);
+
+	goto out;
+
+out_register:
+	rds_trans_unregister(&rds_tcp_transport);
+out_recv:
+	rds_tcp_recv_exit();
+out_slab:
+	kmem_cache_destroy(rds_tcp_conn_slab);
+out:
+	return ret;
+}
+module_init(rds_tcp_init);
+
+MODULE_AUTHOR("Oracle Corporation <rds-devel@oss.oracle.com>");
+MODULE_DESCRIPTION("RDS: TCP transport");
+MODULE_LICENSE("Dual BSD/GPL");
+
diff --git a/net/rds/tcp.h b/net/rds/tcp.h
new file mode 100644
index 000000000..0dbdd3716
--- /dev/null
+++ b/net/rds/tcp.h
@@ -0,0 +1,87 @@
+#ifndef _RDS_TCP_H
+#define _RDS_TCP_H
+
+#define RDS_TCP_PORT	16385
+
+struct rds_tcp_incoming {
+	struct rds_incoming	ti_inc;
+	struct sk_buff_head	ti_skb_list;
+};
+
+struct rds_tcp_connection {
+
+	struct list_head	t_tcp_node;
+	struct rds_connection   *conn;
+	struct socket		*t_sock;
+	void			*t_orig_write_space;
+	void			*t_orig_data_ready;
+	void			*t_orig_state_change;
+
+	struct rds_tcp_incoming	*t_tinc;
+	size_t			t_tinc_hdr_rem;
+	size_t			t_tinc_data_rem;
+
+	/* XXX error report? */
+	struct work_struct	t_conn_w;
+	struct work_struct	t_send_w;
+	struct work_struct	t_down_w;
+	struct work_struct	t_recv_w;
+
+	/* for info exporting only */
+	struct list_head	t_list_item;
+	u32			t_last_sent_nxt;
+	u32			t_last_expected_una;
+	u32			t_last_seen_una;
+};
+
+struct rds_tcp_statistics {
+	uint64_t	s_tcp_data_ready_calls;
+	uint64_t	s_tcp_write_space_calls;
+	uint64_t	s_tcp_sndbuf_full;
+	uint64_t	s_tcp_connect_raced;
+	uint64_t	s_tcp_listen_closed_stale;
+};
+
+/* tcp.c */
+void rds_tcp_tune(struct socket *sock);
+void rds_tcp_nonagle(struct socket *sock);
+void rds_tcp_set_callbacks(struct socket *sock, struct rds_connection *conn);
+void rds_tcp_restore_callbacks(struct socket *sock,
+			       struct rds_tcp_connection *tc);
+u32 rds_tcp_snd_nxt(struct rds_tcp_connection *tc);
+u32 rds_tcp_snd_una(struct rds_tcp_connection *tc);
+u64 rds_tcp_map_seq(struct rds_tcp_connection *tc, u32 seq);
+extern struct rds_transport rds_tcp_transport;
+
+/* tcp_connect.c */
+int rds_tcp_conn_connect(struct rds_connection *conn);
+void rds_tcp_conn_shutdown(struct rds_connection *conn);
+void rds_tcp_state_change(struct sock *sk);
+
+/* tcp_listen.c */
+int rds_tcp_listen_init(void);
+void rds_tcp_listen_stop(void);
+void rds_tcp_listen_data_ready(struct sock *sk);
+
+/* tcp_recv.c */
+int rds_tcp_recv_init(void);
+void rds_tcp_recv_exit(void);
+void rds_tcp_data_ready(struct sock *sk);
+int rds_tcp_recv(struct rds_connection *conn);
+void rds_tcp_inc_free(struct rds_incoming *inc);
+int rds_tcp_inc_copy_to_user(struct rds_incoming *inc, struct iov_iter *to);
+
+/* tcp_send.c */
+void rds_tcp_xmit_prepare(struct rds_connection *conn);
+void rds_tcp_xmit_complete(struct rds_connection *conn);
+int rds_tcp_xmit(struct rds_connection *conn, struct rds_message *rm,
+	         unsigned int hdr_off, unsigned int sg, unsigned int off);
+void rds_tcp_write_space(struct sock *sk);
+
+/* tcp_stats.c */
+DECLARE_PER_CPU(struct rds_tcp_statistics, rds_tcp_stats);
+#define rds_tcp_stats_inc(member) rds_stats_inc_which(rds_tcp_stats, member)
+unsigned int rds_tcp_stats_info_copy(struct rds_info_iterator *iter,
+				     unsigned int avail);
+
+#endif
diff --git a/net/rds/tcp_connect.c b/net/rds/tcp_connect.c
new file mode 100644
index 000000000..973109c7b
--- /dev/null
+++ b/net/rds/tcp_connect.c
@@ -0,0 +1,156 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/in.h>
+#include <net/tcp.h>
+
+#include "rds.h"
+#include "tcp.h"
+
+void rds_tcp_state_change(struct sock *sk)
+{
+	void (*state_change)(struct sock *sk);
+	struct rds_connection *conn;
+	struct rds_tcp_connection *tc;
+
+	read_lock(&sk->sk_callback_lock);
+	conn = sk->sk_user_data;
+	if (!conn) {
+		state_change = sk->sk_state_change;
+		goto out;
+	}
+	tc = conn->c_transport_data;
+	state_change = tc->t_orig_state_change;
+
+	rdsdebug("sock %p state_change to %d\n", tc->t_sock, sk->sk_state);
+
+	switch(sk->sk_state) {
+		/* ignore connecting sockets as they make progress */
+		case TCP_SYN_SENT:
+		case TCP_SYN_RECV:
+			break;
+		case TCP_ESTABLISHED:
+			rds_connect_complete(conn);
+			break;
+		case TCP_CLOSE_WAIT:
+		case TCP_CLOSE:
+			rds_conn_drop(conn);
+		default:
+			break;
+	}
+out:
+	read_unlock(&sk->sk_callback_lock);
+	state_change(sk);
+}
+
+int rds_tcp_conn_connect(struct rds_connection *conn)
+{
+	struct socket *sock = NULL;
+	struct sockaddr_in src, dest;
+	int ret;
+
+	ret = sock_create(PF_INET, SOCK_STREAM, IPPROTO_TCP, &sock);
+	if (ret < 0)
+		goto out;
+
+	rds_tcp_tune(sock);
+
+	src.sin_family = AF_INET;
+	src.sin_addr.s_addr = (__force u32)conn->c_laddr;
+	src.sin_port = (__force u16)htons(0);
+
+	ret = sock->ops->bind(sock, (struct sockaddr *)&src, sizeof(src));
+	if (ret) {
+		rdsdebug("bind failed with %d at address %pI4\n",
+			 ret, &conn->c_laddr);
+		goto out;
+	}
+
+	dest.sin_family = AF_INET;
+	dest.sin_addr.s_addr = (__force u32)conn->c_faddr;
+	dest.sin_port = (__force u16)htons(RDS_TCP_PORT);
+
+	/*
+	 * once we call connect() we can start getting callbacks and they
+	 * own the socket
+	 */
+	rds_tcp_set_callbacks(sock, conn);
+	ret = sock->ops->connect(sock, (struct sockaddr *)&dest, sizeof(dest),
+				 O_NONBLOCK);
+
+	rdsdebug("connect to address %pI4 returned %d\n", &conn->c_faddr, ret);
+	if (ret == -EINPROGRESS)
+		ret = 0;
+	if (ret == 0)
+		sock = NULL;
+	else
+		rds_tcp_restore_callbacks(sock, conn->c_transport_data);
+
+out:
+	if (sock)
+		sock_release(sock);
+	return ret;
+}
+
+/*
+ * Before killing the tcp socket this needs to serialize with callbacks.  The
+ * caller has already grabbed the sending sem so we're serialized with other
+ * senders.
+ *
+ * TCP calls the callbacks with the sock lock so we hold it while we reset the
+ * callbacks to those set by TCP.  Our callbacks won't execute again once we
+ * hold the sock lock.
+ */
+void rds_tcp_conn_shutdown(struct rds_connection *conn)
+{
+	struct rds_tcp_connection *tc = conn->c_transport_data;
+	struct socket *sock = tc->t_sock;
+
+	rdsdebug("shutting down conn %p tc %p sock %p\n", conn, tc, sock);
+
+	if (sock) {
+		sock->ops->shutdown(sock, RCV_SHUTDOWN | SEND_SHUTDOWN);
+		lock_sock(sock->sk);
+		rds_tcp_restore_callbacks(sock, tc); /* tc->tc_sock = NULL */
+
+		release_sock(sock->sk);
+		sock_release(sock);
+	}
+
+	if (tc->t_tinc) {
+		rds_inc_put(&tc->t_tinc->ti_inc);
+		tc->t_tinc = NULL;
+	}
+	tc->t_tinc_hdr_rem = sizeof(struct rds_header);
+	tc->t_tinc_data_rem = 0;
+}
diff --git a/net/rds/tcp_listen.c b/net/rds/tcp_listen.c
new file mode 100644
index 000000000..0da49e344
--- /dev/null
+++ b/net/rds/tcp_listen.c
@@ -0,0 +1,246 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/gfp.h>
+#include <linux/in.h>
+#include <net/tcp.h>
+
+#include "rds.h"
+#include "tcp.h"
+
+/*
+ * cheesy, but simple..
+ */
+static void rds_tcp_accept_worker(struct work_struct *work);
+static DECLARE_WORK(rds_tcp_listen_work, rds_tcp_accept_worker);
+static struct socket *rds_tcp_listen_sock;
+
+static int rds_tcp_keepalive(struct socket *sock)
+{
+	/* values below based on xs_udp_default_timeout */
+	int keepidle = 5; /* send a probe 'keepidle' secs after last data */
+	int keepcnt = 5; /* number of unack'ed probes before declaring dead */
+	int keepalive = 1;
+	int ret = 0;
+
+	ret = kernel_setsockopt(sock, SOL_SOCKET, SO_KEEPALIVE,
+				(char *)&keepalive, sizeof(keepalive));
+	if (ret < 0)
+		goto bail;
+
+	ret = kernel_setsockopt(sock, IPPROTO_TCP, TCP_KEEPCNT,
+				(char *)&keepcnt, sizeof(keepcnt));
+	if (ret < 0)
+		goto bail;
+
+	ret = kernel_setsockopt(sock, IPPROTO_TCP, TCP_KEEPIDLE,
+				(char *)&keepidle, sizeof(keepidle));
+	if (ret < 0)
+		goto bail;
+
+	/* KEEPINTVL is the interval between successive probes. We follow
+	 * the model in xs_tcp_finish_connecting() and re-use keepidle.
+	 */
+	ret = kernel_setsockopt(sock, IPPROTO_TCP, TCP_KEEPINTVL,
+				(char *)&keepidle, sizeof(keepidle));
+bail:
+	return ret;
+}
+
+static int rds_tcp_accept_one(struct socket *sock)
+{
+	struct socket *new_sock = NULL;
+	struct rds_connection *conn;
+	int ret;
+	struct inet_sock *inet;
+	struct rds_tcp_connection *rs_tcp;
+
+	ret = sock_create_lite(sock->sk->sk_family, sock->sk->sk_type,
+			       sock->sk->sk_protocol, &new_sock);
+	if (ret)
+		goto out;
+
+	new_sock->type = sock->type;
+	new_sock->ops = sock->ops;
+	ret = sock->ops->accept(sock, new_sock, O_NONBLOCK);
+	if (ret < 0)
+		goto out;
+
+	ret = rds_tcp_keepalive(new_sock);
+	if (ret < 0)
+		goto out;
+
+	rds_tcp_tune(new_sock);
+
+	inet = inet_sk(new_sock->sk);
+
+	rdsdebug("accepted tcp %pI4:%u -> %pI4:%u\n",
+		 &inet->inet_saddr, ntohs(inet->inet_sport),
+		 &inet->inet_daddr, ntohs(inet->inet_dport));
+
+	conn = rds_conn_create(inet->inet_saddr, inet->inet_daddr,
+			       &rds_tcp_transport, GFP_KERNEL);
+	if (IS_ERR(conn)) {
+		ret = PTR_ERR(conn);
+		goto out;
+	}
+	/* An incoming SYN request came in, and TCP just accepted it.
+	 * We always create a new conn for listen side of TCP, and do not
+	 * add it to the c_hash_list.
+	 *
+	 * If the client reboots, this conn will need to be cleaned up.
+	 * rds_tcp_state_change() will do that cleanup
+	 */
+	rs_tcp = (struct rds_tcp_connection *)conn->c_transport_data;
+	WARN_ON(!rs_tcp || rs_tcp->t_sock);
+
+	/*
+	 * see the comment above rds_queue_delayed_reconnect()
+	 */
+	if (!rds_conn_transition(conn, RDS_CONN_DOWN, RDS_CONN_CONNECTING)) {
+		if (rds_conn_state(conn) == RDS_CONN_UP)
+			rds_tcp_stats_inc(s_tcp_listen_closed_stale);
+		else
+			rds_tcp_stats_inc(s_tcp_connect_raced);
+		rds_conn_drop(conn);
+		ret = 0;
+		goto out;
+	}
+
+	rds_tcp_set_callbacks(new_sock, conn);
+	rds_connect_complete(conn);
+	new_sock = NULL;
+	ret = 0;
+
+out:
+	if (new_sock)
+		sock_release(new_sock);
+	return ret;
+}
+
+static void rds_tcp_accept_worker(struct work_struct *work)
+{
+	while (rds_tcp_accept_one(rds_tcp_listen_sock) == 0)
+		cond_resched();
+}
+
+void rds_tcp_listen_data_ready(struct sock *sk)
+{
+	void (*ready)(struct sock *sk);
+
+	rdsdebug("listen data ready sk %p\n", sk);
+
+	read_lock(&sk->sk_callback_lock);
+	ready = sk->sk_user_data;
+	if (!ready) { /* check for teardown race */
+		ready = sk->sk_data_ready;
+		goto out;
+	}
+
+	/*
+	 * ->sk_data_ready is also called for a newly established child socket
+	 * before it has been accepted and the accepter has set up their
+	 * data_ready.. we only want to queue listen work for our listening
+	 * socket
+	 */
+	if (sk->sk_state == TCP_LISTEN)
+		queue_work(rds_wq, &rds_tcp_listen_work);
+
+out:
+	read_unlock(&sk->sk_callback_lock);
+	ready(sk);
+}
+
+int rds_tcp_listen_init(void)
+{
+	struct sockaddr_in sin;
+	struct socket *sock = NULL;
+	int ret;
+
+	ret = sock_create(PF_INET, SOCK_STREAM, IPPROTO_TCP, &sock);
+	if (ret < 0)
+		goto out;
+
+	sock->sk->sk_reuse = SK_CAN_REUSE;
+	rds_tcp_nonagle(sock);
+
+	write_lock_bh(&sock->sk->sk_callback_lock);
+	sock->sk->sk_user_data = sock->sk->sk_data_ready;
+	sock->sk->sk_data_ready = rds_tcp_listen_data_ready;
+	write_unlock_bh(&sock->sk->sk_callback_lock);
+
+	sin.sin_family = PF_INET;
+	sin.sin_addr.s_addr = (__force u32)htonl(INADDR_ANY);
+	sin.sin_port = (__force u16)htons(RDS_TCP_PORT);
+
+	ret = sock->ops->bind(sock, (struct sockaddr *)&sin, sizeof(sin));
+	if (ret < 0)
+		goto out;
+
+	ret = sock->ops->listen(sock, 64);
+	if (ret < 0)
+		goto out;
+
+	rds_tcp_listen_sock = sock;
+	sock = NULL;
+out:
+	if (sock)
+		sock_release(sock);
+	return ret;
+}
+
+void rds_tcp_listen_stop(void)
+{
+	struct socket *sock = rds_tcp_listen_sock;
+	struct sock *sk;
+
+	if (!sock)
+		return;
+
+	sk = sock->sk;
+
+	/* serialize with and prevent further callbacks */
+	lock_sock(sk);
+	write_lock_bh(&sk->sk_callback_lock);
+	if (sk->sk_user_data) {
+		sk->sk_data_ready = sk->sk_user_data;
+		sk->sk_user_data = NULL;
+	}
+	write_unlock_bh(&sk->sk_callback_lock);
+	release_sock(sk);
+
+	/* wait for accepts to stop and close the socket */
+	flush_workqueue(rds_wq);
+	sock_release(sock);
+	rds_tcp_listen_sock = NULL;
+}
diff --git a/net/rds/tcp_recv.c b/net/rds/tcp_recv.c
new file mode 100644
index 000000000..fbc5ef88b
--- /dev/null
+++ b/net/rds/tcp_recv.c
@@ -0,0 +1,336 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <net/tcp.h>
+
+#include "rds.h"
+#include "tcp.h"
+
+static struct kmem_cache *rds_tcp_incoming_slab;
+
+static void rds_tcp_inc_purge(struct rds_incoming *inc)
+{
+	struct rds_tcp_incoming *tinc;
+	tinc = container_of(inc, struct rds_tcp_incoming, ti_inc);
+	rdsdebug("purging tinc %p inc %p\n", tinc, inc);
+	skb_queue_purge(&tinc->ti_skb_list);
+}
+
+void rds_tcp_inc_free(struct rds_incoming *inc)
+{
+	struct rds_tcp_incoming *tinc;
+	tinc = container_of(inc, struct rds_tcp_incoming, ti_inc);
+	rds_tcp_inc_purge(inc);
+	rdsdebug("freeing tinc %p inc %p\n", tinc, inc);
+	kmem_cache_free(rds_tcp_incoming_slab, tinc);
+}
+
+/*
+ * this is pretty lame, but, whatever.
+ */
+int rds_tcp_inc_copy_to_user(struct rds_incoming *inc, struct iov_iter *to)
+{
+	struct rds_tcp_incoming *tinc;
+	struct sk_buff *skb;
+	int ret = 0;
+
+	if (!iov_iter_count(to))
+		goto out;
+
+	tinc = container_of(inc, struct rds_tcp_incoming, ti_inc);
+
+	skb_queue_walk(&tinc->ti_skb_list, skb) {
+		unsigned long to_copy, skb_off;
+		for (skb_off = 0; skb_off < skb->len; skb_off += to_copy) {
+			to_copy = iov_iter_count(to);
+			to_copy = min(to_copy, skb->len - skb_off);
+
+			if (skb_copy_datagram_iter(skb, skb_off, to, to_copy))
+				return -EFAULT;
+
+			rds_stats_add(s_copy_to_user, to_copy);
+			ret += to_copy;
+
+			if (!iov_iter_count(to))
+				goto out;
+		}
+	}
+out:
+	return ret;
+}
+
+/*
+ * We have a series of skbs that have fragmented pieces of the congestion
+ * bitmap.  They must add up to the exact size of the congestion bitmap.  We
+ * use the skb helpers to copy those into the pages that make up the in-memory
+ * congestion bitmap for the remote address of this connection.  We then tell
+ * the congestion core that the bitmap has been changed so that it can wake up
+ * sleepers.
+ *
+ * This is racing with sending paths which are using test_bit to see if the
+ * bitmap indicates that their recipient is congested.
+ */
+
+static void rds_tcp_cong_recv(struct rds_connection *conn,
+			      struct rds_tcp_incoming *tinc)
+{
+	struct sk_buff *skb;
+	unsigned int to_copy, skb_off;
+	unsigned int map_off;
+	unsigned int map_page;
+	struct rds_cong_map *map;
+	int ret;
+
+	/* catch completely corrupt packets */
+	if (be32_to_cpu(tinc->ti_inc.i_hdr.h_len) != RDS_CONG_MAP_BYTES)
+		return;
+
+	map_page = 0;
+	map_off = 0;
+	map = conn->c_fcong;
+
+	skb_queue_walk(&tinc->ti_skb_list, skb) {
+		skb_off = 0;
+		while (skb_off < skb->len) {
+			to_copy = min_t(unsigned int, PAGE_SIZE - map_off,
+					skb->len - skb_off);
+
+			BUG_ON(map_page >= RDS_CONG_MAP_PAGES);
+
+			/* only returns 0 or -error */
+			ret = skb_copy_bits(skb, skb_off,
+				(void *)map->m_page_addrs[map_page] + map_off,
+				to_copy);
+			BUG_ON(ret != 0);
+
+			skb_off += to_copy;
+			map_off += to_copy;
+			if (map_off == PAGE_SIZE) {
+				map_off = 0;
+				map_page++;
+			}
+		}
+	}
+
+	rds_cong_map_updated(map, ~(u64) 0);
+}
+
+struct rds_tcp_desc_arg {
+	struct rds_connection *conn;
+	gfp_t gfp;
+};
+
+static int rds_tcp_data_recv(read_descriptor_t *desc, struct sk_buff *skb,
+			     unsigned int offset, size_t len)
+{
+	struct rds_tcp_desc_arg *arg = desc->arg.data;
+	struct rds_connection *conn = arg->conn;
+	struct rds_tcp_connection *tc = conn->c_transport_data;
+	struct rds_tcp_incoming *tinc = tc->t_tinc;
+	struct sk_buff *clone;
+	size_t left = len, to_copy;
+
+	rdsdebug("tcp data tc %p skb %p offset %u len %zu\n", tc, skb, offset,
+		 len);
+
+	/*
+	 * tcp_read_sock() interprets partial progress as an indication to stop
+	 * processing.
+	 */
+	while (left) {
+		if (!tinc) {
+			tinc = kmem_cache_alloc(rds_tcp_incoming_slab,
+					        arg->gfp);
+			if (!tinc) {
+				desc->error = -ENOMEM;
+				goto out;
+			}
+			tc->t_tinc = tinc;
+			rdsdebug("alloced tinc %p\n", tinc);
+			rds_inc_init(&tinc->ti_inc, conn, conn->c_faddr);
+			/*
+			 * XXX * we might be able to use the __ variants when
+			 * we've already serialized at a higher level.
+			 */
+			skb_queue_head_init(&tinc->ti_skb_list);
+		}
+
+		if (left && tc->t_tinc_hdr_rem) {
+			to_copy = min(tc->t_tinc_hdr_rem, left);
+			rdsdebug("copying %zu header from skb %p\n", to_copy,
+				 skb);
+			skb_copy_bits(skb, offset,
+				      (char *)&tinc->ti_inc.i_hdr +
+						sizeof(struct rds_header) -
+						tc->t_tinc_hdr_rem,
+				      to_copy);
+			tc->t_tinc_hdr_rem -= to_copy;
+			left -= to_copy;
+			offset += to_copy;
+
+			if (tc->t_tinc_hdr_rem == 0) {
+				/* could be 0 for a 0 len message */
+				tc->t_tinc_data_rem =
+					be32_to_cpu(tinc->ti_inc.i_hdr.h_len);
+			}
+		}
+
+		if (left && tc->t_tinc_data_rem) {
+			clone = skb_clone(skb, arg->gfp);
+			if (!clone) {
+				desc->error = -ENOMEM;
+				goto out;
+			}
+
+			to_copy = min(tc->t_tinc_data_rem, left);
+			pskb_pull(clone, offset);
+			pskb_trim(clone, to_copy);
+			skb_queue_tail(&tinc->ti_skb_list, clone);
+
+			rdsdebug("skb %p data %p len %d off %u to_copy %zu -> "
+				 "clone %p data %p len %d\n",
+				 skb, skb->data, skb->len, offset, to_copy,
+				 clone, clone->data, clone->len);
+
+			tc->t_tinc_data_rem -= to_copy;
+			left -= to_copy;
+			offset += to_copy;
+		}
+
+		if (tc->t_tinc_hdr_rem == 0 && tc->t_tinc_data_rem == 0) {
+			if (tinc->ti_inc.i_hdr.h_flags == RDS_FLAG_CONG_BITMAP)
+				rds_tcp_cong_recv(conn, tinc);
+			else
+				rds_recv_incoming(conn, conn->c_faddr,
+						  conn->c_laddr, &tinc->ti_inc,
+						  arg->gfp);
+
+			tc->t_tinc_hdr_rem = sizeof(struct rds_header);
+			tc->t_tinc_data_rem = 0;
+			tc->t_tinc = NULL;
+			rds_inc_put(&tinc->ti_inc);
+			tinc = NULL;
+		}
+	}
+out:
+	rdsdebug("returning len %zu left %zu skb len %d rx queue depth %d\n",
+		 len, left, skb->len,
+		 skb_queue_len(&tc->t_sock->sk->sk_receive_queue));
+	return len - left;
+}
+
+/* the caller has to hold the sock lock */
+static int rds_tcp_read_sock(struct rds_connection *conn, gfp_t gfp)
+{
+	struct rds_tcp_connection *tc = conn->c_transport_data;
+	struct socket *sock = tc->t_sock;
+	read_descriptor_t desc;
+	struct rds_tcp_desc_arg arg;
+
+	/* It's like glib in the kernel! */
+	arg.conn = conn;
+	arg.gfp = gfp;
+	desc.arg.data = &arg;
+	desc.error = 0;
+	desc.count = 1; /* give more than one skb per call */
+
+	tcp_read_sock(sock->sk, &desc, rds_tcp_data_recv);
+	rdsdebug("tcp_read_sock for tc %p gfp 0x%x returned %d\n", tc, gfp,
+		 desc.error);
+
+	return desc.error;
+}
+
+/*
+ * We hold the sock lock to serialize our rds_tcp_recv->tcp_read_sock from
+ * data_ready.
+ *
+ * if we fail to allocate we're in trouble.. blindly wait some time before
+ * trying again to see if the VM can free up something for us.
+ */
+int rds_tcp_recv(struct rds_connection *conn)
+{
+	struct rds_tcp_connection *tc = conn->c_transport_data;
+	struct socket *sock = tc->t_sock;
+	int ret = 0;
+
+	rdsdebug("recv worker conn %p tc %p sock %p\n", conn, tc, sock);
+
+	lock_sock(sock->sk);
+	ret = rds_tcp_read_sock(conn, GFP_KERNEL);
+	release_sock(sock->sk);
+
+	return ret;
+}
+
+void rds_tcp_data_ready(struct sock *sk)
+{
+	void (*ready)(struct sock *sk);
+	struct rds_connection *conn;
+	struct rds_tcp_connection *tc;
+
+	rdsdebug("data ready sk %p\n", sk);
+
+	read_lock(&sk->sk_callback_lock);
+	conn = sk->sk_user_data;
+	if (!conn) { /* check for teardown race */
+		ready = sk->sk_data_ready;
+		goto out;
+	}
+
+	tc = conn->c_transport_data;
+	ready = tc->t_orig_data_ready;
+	rds_tcp_stats_inc(s_tcp_data_ready_calls);
+
+	if (rds_tcp_read_sock(conn, GFP_ATOMIC) == -ENOMEM)
+		queue_delayed_work(rds_wq, &conn->c_recv_w, 0);
+out:
+	read_unlock(&sk->sk_callback_lock);
+	ready(sk);
+}
+
+int rds_tcp_recv_init(void)
+{
+	rds_tcp_incoming_slab = kmem_cache_create("rds_tcp_incoming",
+					sizeof(struct rds_tcp_incoming),
+					0, 0, NULL);
+	if (!rds_tcp_incoming_slab)
+		return -ENOMEM;
+	return 0;
+}
+
+void rds_tcp_recv_exit(void)
+{
+	kmem_cache_destroy(rds_tcp_incoming_slab);
+}
diff --git a/net/rds/tcp_send.c b/net/rds/tcp_send.c
new file mode 100644
index 000000000..53b17ca0d
--- /dev/null
+++ b/net/rds/tcp_send.c
@@ -0,0 +1,215 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/in.h>
+#include <net/tcp.h>
+
+#include "rds.h"
+#include "tcp.h"
+
+static void rds_tcp_cork(struct socket *sock, int val)
+{
+	mm_segment_t oldfs;
+
+	oldfs = get_fs();
+	set_fs(KERNEL_DS);
+	sock->ops->setsockopt(sock, SOL_TCP, TCP_CORK, (char __user *)&val,
+			      sizeof(val));
+	set_fs(oldfs);
+}
+
+void rds_tcp_xmit_prepare(struct rds_connection *conn)
+{
+	struct rds_tcp_connection *tc = conn->c_transport_data;
+
+	rds_tcp_cork(tc->t_sock, 1);
+}
+
+void rds_tcp_xmit_complete(struct rds_connection *conn)
+{
+	struct rds_tcp_connection *tc = conn->c_transport_data;
+
+	rds_tcp_cork(tc->t_sock, 0);
+}
+
+/* the core send_sem serializes this with other xmit and shutdown */
+static int rds_tcp_sendmsg(struct socket *sock, void *data, unsigned int len)
+{
+	struct kvec vec = {
+                .iov_base = data,
+                .iov_len = len,
+	};
+        struct msghdr msg = {
+                .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL,
+        };
+
+	return kernel_sendmsg(sock, &msg, &vec, 1, vec.iov_len);
+}
+
+/* the core send_sem serializes this with other xmit and shutdown */
+int rds_tcp_xmit(struct rds_connection *conn, struct rds_message *rm,
+	         unsigned int hdr_off, unsigned int sg, unsigned int off)
+{
+	struct rds_tcp_connection *tc = conn->c_transport_data;
+	int done = 0;
+	int ret = 0;
+
+	if (hdr_off == 0) {
+		/*
+		 * m_ack_seq is set to the sequence number of the last byte of
+		 * header and data.  see rds_tcp_is_acked().
+		 */
+		tc->t_last_sent_nxt = rds_tcp_snd_nxt(tc);
+		rm->m_ack_seq = tc->t_last_sent_nxt +
+				sizeof(struct rds_header) +
+				be32_to_cpu(rm->m_inc.i_hdr.h_len) - 1;
+		smp_mb__before_atomic();
+		set_bit(RDS_MSG_HAS_ACK_SEQ, &rm->m_flags);
+		tc->t_last_expected_una = rm->m_ack_seq + 1;
+
+		rdsdebug("rm %p tcp nxt %u ack_seq %llu\n",
+			 rm, rds_tcp_snd_nxt(tc),
+			 (unsigned long long)rm->m_ack_seq);
+	}
+
+	if (hdr_off < sizeof(struct rds_header)) {
+		/* see rds_tcp_write_space() */
+		set_bit(SOCK_NOSPACE, &tc->t_sock->sk->sk_socket->flags);
+
+		ret = rds_tcp_sendmsg(tc->t_sock,
+				      (void *)&rm->m_inc.i_hdr + hdr_off,
+				      sizeof(rm->m_inc.i_hdr) - hdr_off);
+		if (ret < 0)
+			goto out;
+		done += ret;
+		if (hdr_off + done != sizeof(struct rds_header))
+			goto out;
+	}
+
+	while (sg < rm->data.op_nents) {
+		ret = tc->t_sock->ops->sendpage(tc->t_sock,
+						sg_page(&rm->data.op_sg[sg]),
+						rm->data.op_sg[sg].offset + off,
+						rm->data.op_sg[sg].length - off,
+						MSG_DONTWAIT|MSG_NOSIGNAL);
+		rdsdebug("tcp sendpage %p:%u:%u ret %d\n", (void *)sg_page(&rm->data.op_sg[sg]),
+			 rm->data.op_sg[sg].offset + off, rm->data.op_sg[sg].length - off,
+			 ret);
+		if (ret <= 0)
+			break;
+
+		off += ret;
+		done += ret;
+		if (off == rm->data.op_sg[sg].length) {
+			off = 0;
+			sg++;
+		}
+	}
+
+out:
+	if (ret <= 0) {
+		/* write_space will hit after EAGAIN, all else fatal */
+		if (ret == -EAGAIN) {
+			rds_tcp_stats_inc(s_tcp_sndbuf_full);
+			ret = 0;
+		} else {
+			printk(KERN_WARNING "RDS/tcp: send to %pI4 "
+			       "returned %d, disconnecting and reconnecting\n",
+			       &conn->c_faddr, ret);
+			rds_conn_drop(conn);
+		}
+	}
+	if (done == 0)
+		done = ret;
+	return done;
+}
+
+/*
+ * rm->m_ack_seq is set to the tcp sequence number that corresponds to the
+ * last byte of the message, including the header.  This means that the
+ * entire message has been received if rm->m_ack_seq is "before" the next
+ * unacked byte of the TCP sequence space.  We have to do very careful
+ * wrapping 32bit comparisons here.
+ */
+static int rds_tcp_is_acked(struct rds_message *rm, uint64_t ack)
+{
+	if (!test_bit(RDS_MSG_HAS_ACK_SEQ, &rm->m_flags))
+		return 0;
+	return (__s32)((u32)rm->m_ack_seq - (u32)ack) < 0;
+}
+
+void rds_tcp_write_space(struct sock *sk)
+{
+	void (*write_space)(struct sock *sk);
+	struct rds_connection *conn;
+	struct rds_tcp_connection *tc;
+
+	read_lock(&sk->sk_callback_lock);
+	conn = sk->sk_user_data;
+	if (!conn) {
+		write_space = sk->sk_write_space;
+		goto out;
+	}
+
+	tc = conn->c_transport_data;
+	rdsdebug("write_space for tc %p\n", tc);
+	write_space = tc->t_orig_write_space;
+	rds_tcp_stats_inc(s_tcp_write_space_calls);
+
+	rdsdebug("tcp una %u\n", rds_tcp_snd_una(tc));
+	tc->t_last_seen_una = rds_tcp_snd_una(tc);
+	rds_send_drop_acked(conn, rds_tcp_snd_una(tc), rds_tcp_is_acked);
+
+        if ((atomic_read(&sk->sk_wmem_alloc) << 1) <= sk->sk_sndbuf)
+		queue_delayed_work(rds_wq, &conn->c_send_w, 0);
+
+out:
+	read_unlock(&sk->sk_callback_lock);
+
+	/*
+	 * write_space is only called when data leaves tcp's send queue if
+	 * SOCK_NOSPACE is set.  We set SOCK_NOSPACE every time we put
+	 * data in tcp's send queue because we use write_space to parse the
+	 * sequence numbers and notice that rds messages have been fully
+	 * received.
+	 *
+	 * tcp's write_space clears SOCK_NOSPACE if the send queue has more
+	 * than a certain amount of space. So we need to set it again *after*
+	 * we call tcp's write_space or else we might only get called on the
+	 * first of a series of incoming tcp acks.
+	 */
+	write_space(sk);
+
+	if (sk->sk_socket)
+		set_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
+}
diff --git a/net/rds/tcp_stats.c b/net/rds/tcp_stats.c
new file mode 100644
index 000000000..f8a7954f1
--- /dev/null
+++ b/net/rds/tcp_stats.c
@@ -0,0 +1,74 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/percpu.h>
+#include <linux/seq_file.h>
+#include <linux/proc_fs.h>
+
+#include "rds.h"
+#include "tcp.h"
+
+DEFINE_PER_CPU(struct rds_tcp_statistics, rds_tcp_stats)
+	____cacheline_aligned;
+
+static const char * const rds_tcp_stat_names[] = {
+	"tcp_data_ready_calls",
+	"tcp_write_space_calls",
+	"tcp_sndbuf_full",
+	"tcp_connect_raced",
+	"tcp_listen_closed_stale",
+};
+
+unsigned int rds_tcp_stats_info_copy(struct rds_info_iterator *iter,
+				     unsigned int avail)
+{
+	struct rds_tcp_statistics stats = {0, };
+	uint64_t *src;
+	uint64_t *sum;
+	size_t i;
+	int cpu;
+
+	if (avail < ARRAY_SIZE(rds_tcp_stat_names))
+		goto out;
+
+	for_each_online_cpu(cpu) {
+		src = (uint64_t *)&(per_cpu(rds_tcp_stats, cpu));
+		sum = (uint64_t *)&stats;
+		for (i = 0; i < sizeof(stats) / sizeof(uint64_t); i++)
+			*(sum++) += *(src++);
+	}
+
+	rds_stats_info_copy(iter, (uint64_t *)&stats, rds_tcp_stat_names,
+			    ARRAY_SIZE(rds_tcp_stat_names));
+out:
+	return ARRAY_SIZE(rds_tcp_stat_names);
+}
diff --git a/net/rds/threads.c b/net/rds/threads.c
new file mode 100644
index 000000000..dc2402e87
--- /dev/null
+++ b/net/rds/threads.c
@@ -0,0 +1,222 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/random.h>
+#include <linux/export.h>
+
+#include "rds.h"
+
+/*
+ * All of connection management is simplified by serializing it through
+ * work queues that execute in a connection managing thread.
+ *
+ * TCP wants to send acks through sendpage() in response to data_ready(),
+ * but it needs a process context to do so.
+ *
+ * The receive paths need to allocate but can't drop packets (!) so we have
+ * a thread around to block allocating if the receive fast path sees an
+ * allocation failure.
+ */
+
+/* Grand Unified Theory of connection life cycle:
+ * At any point in time, the connection can be in one of these states:
+ * DOWN, CONNECTING, UP, DISCONNECTING, ERROR
+ *
+ * The following transitions are possible:
+ *  ANY		  -> ERROR
+ *  UP		  -> DISCONNECTING
+ *  ERROR	  -> DISCONNECTING
+ *  DISCONNECTING -> DOWN
+ *  DOWN	  -> CONNECTING
+ *  CONNECTING	  -> UP
+ *
+ * Transition to state DISCONNECTING/DOWN:
+ *  -	Inside the shutdown worker; synchronizes with xmit path
+ *	through RDS_IN_XMIT, and with connection management callbacks
+ *	via c_cm_lock.
+ *
+ *	For receive callbacks, we rely on the underlying transport
+ *	(TCP, IB/RDMA) to provide the necessary synchronisation.
+ */
+struct workqueue_struct *rds_wq;
+EXPORT_SYMBOL_GPL(rds_wq);
+
+void rds_connect_complete(struct rds_connection *conn)
+{
+	if (!rds_conn_transition(conn, RDS_CONN_CONNECTING, RDS_CONN_UP)) {
+		printk(KERN_WARNING "%s: Cannot transition to state UP, "
+				"current state is %d\n",
+				__func__,
+				atomic_read(&conn->c_state));
+		rds_conn_drop(conn);
+		return;
+	}
+
+	rdsdebug("conn %p for %pI4 to %pI4 complete\n",
+	  conn, &conn->c_laddr, &conn->c_faddr);
+
+	conn->c_reconnect_jiffies = 0;
+	set_bit(0, &conn->c_map_queued);
+	queue_delayed_work(rds_wq, &conn->c_send_w, 0);
+	queue_delayed_work(rds_wq, &conn->c_recv_w, 0);
+}
+EXPORT_SYMBOL_GPL(rds_connect_complete);
+
+/*
+ * This random exponential backoff is relied on to eventually resolve racing
+ * connects.
+ *
+ * If connect attempts race then both parties drop both connections and come
+ * here to wait for a random amount of time before trying again.  Eventually
+ * the backoff range will be so much greater than the time it takes to
+ * establish a connection that one of the pair will establish the connection
+ * before the other's random delay fires.
+ *
+ * Connection attempts that arrive while a connection is already established
+ * are also considered to be racing connects.  This lets a connection from
+ * a rebooted machine replace an existing stale connection before the transport
+ * notices that the connection has failed.
+ *
+ * We should *always* start with a random backoff; otherwise a broken connection
+ * will always take several iterations to be re-established.
+ */
+void rds_queue_reconnect(struct rds_connection *conn)
+{
+	unsigned long rand;
+
+	rdsdebug("conn %p for %pI4 to %pI4 reconnect jiffies %lu\n",
+	  conn, &conn->c_laddr, &conn->c_faddr,
+	  conn->c_reconnect_jiffies);
+
+	set_bit(RDS_RECONNECT_PENDING, &conn->c_flags);
+	if (conn->c_reconnect_jiffies == 0) {
+		conn->c_reconnect_jiffies = rds_sysctl_reconnect_min_jiffies;
+		queue_delayed_work(rds_wq, &conn->c_conn_w, 0);
+		return;
+	}
+
+	get_random_bytes(&rand, sizeof(rand));
+	rdsdebug("%lu delay %lu ceil conn %p for %pI4 -> %pI4\n",
+		 rand % conn->c_reconnect_jiffies, conn->c_reconnect_jiffies,
+		 conn, &conn->c_laddr, &conn->c_faddr);
+	queue_delayed_work(rds_wq, &conn->c_conn_w,
+			   rand % conn->c_reconnect_jiffies);
+
+	conn->c_reconnect_jiffies = min(conn->c_reconnect_jiffies * 2,
+					rds_sysctl_reconnect_max_jiffies);
+}
+
+void rds_connect_worker(struct work_struct *work)
+{
+	struct rds_connection *conn = container_of(work, struct rds_connection, c_conn_w.work);
+	int ret;
+
+	clear_bit(RDS_RECONNECT_PENDING, &conn->c_flags);
+	if (rds_conn_transition(conn, RDS_CONN_DOWN, RDS_CONN_CONNECTING)) {
+		ret = conn->c_trans->conn_connect(conn);
+		rdsdebug("conn %p for %pI4 to %pI4 dispatched, ret %d\n",
+			conn, &conn->c_laddr, &conn->c_faddr, ret);
+
+		if (ret) {
+			if (rds_conn_transition(conn, RDS_CONN_CONNECTING, RDS_CONN_DOWN))
+				rds_queue_reconnect(conn);
+			else
+				rds_conn_error(conn, "RDS: connect failed\n");
+		}
+	}
+}
+
+void rds_send_worker(struct work_struct *work)
+{
+	struct rds_connection *conn = container_of(work, struct rds_connection, c_send_w.work);
+	int ret;
+
+	if (rds_conn_state(conn) == RDS_CONN_UP) {
+		ret = rds_send_xmit(conn);
+		rdsdebug("conn %p ret %d\n", conn, ret);
+		switch (ret) {
+		case -EAGAIN:
+			rds_stats_inc(s_send_immediate_retry);
+			queue_delayed_work(rds_wq, &conn->c_send_w, 0);
+			break;
+		case -ENOMEM:
+			rds_stats_inc(s_send_delayed_retry);
+			queue_delayed_work(rds_wq, &conn->c_send_w, 2);
+		default:
+			break;
+		}
+	}
+}
+
+void rds_recv_worker(struct work_struct *work)
+{
+	struct rds_connection *conn = container_of(work, struct rds_connection, c_recv_w.work);
+	int ret;
+
+	if (rds_conn_state(conn) == RDS_CONN_UP) {
+		ret = conn->c_trans->recv(conn);
+		rdsdebug("conn %p ret %d\n", conn, ret);
+		switch (ret) {
+		case -EAGAIN:
+			rds_stats_inc(s_recv_immediate_retry);
+			queue_delayed_work(rds_wq, &conn->c_recv_w, 0);
+			break;
+		case -ENOMEM:
+			rds_stats_inc(s_recv_delayed_retry);
+			queue_delayed_work(rds_wq, &conn->c_recv_w, 2);
+		default:
+			break;
+		}
+	}
+}
+
+void rds_shutdown_worker(struct work_struct *work)
+{
+	struct rds_connection *conn = container_of(work, struct rds_connection, c_down_w);
+
+	rds_conn_shutdown(conn);
+}
+
+void rds_threads_exit(void)
+{
+	destroy_workqueue(rds_wq);
+}
+
+int rds_threads_init(void)
+{
+	rds_wq = create_singlethread_workqueue("krdsd");
+	if (!rds_wq)
+		return -ENOMEM;
+
+	return 0;
+}
diff --git a/net/rds/transport.c b/net/rds/transport.c
new file mode 100644
index 000000000..7f2ac4fec
--- /dev/null
+++ b/net/rds/transport.c
@@ -0,0 +1,137 @@
+/*
+ * Copyright (c) 2006 Oracle.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/in.h>
+
+#include "rds.h"
+#include "loop.h"
+
+static struct rds_transport *transports[RDS_TRANS_COUNT];
+static DECLARE_RWSEM(rds_trans_sem);
+
+int rds_trans_register(struct rds_transport *trans)
+{
+	BUG_ON(strlen(trans->t_name) + 1 > TRANSNAMSIZ);
+
+	down_write(&rds_trans_sem);
+
+	if (transports[trans->t_type])
+		printk(KERN_ERR "RDS Transport type %d already registered\n",
+			trans->t_type);
+	else {
+		transports[trans->t_type] = trans;
+		printk(KERN_INFO "Registered RDS/%s transport\n", trans->t_name);
+	}
+
+	up_write(&rds_trans_sem);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(rds_trans_register);
+
+void rds_trans_unregister(struct rds_transport *trans)
+{
+	down_write(&rds_trans_sem);
+
+	transports[trans->t_type] = NULL;
+	printk(KERN_INFO "Unregistered RDS/%s transport\n", trans->t_name);
+
+	up_write(&rds_trans_sem);
+}
+EXPORT_SYMBOL_GPL(rds_trans_unregister);
+
+void rds_trans_put(struct rds_transport *trans)
+{
+	if (trans && trans->t_owner)
+		module_put(trans->t_owner);
+}
+
+struct rds_transport *rds_trans_get_preferred(__be32 addr)
+{
+	struct rds_transport *ret = NULL;
+	struct rds_transport *trans;
+	unsigned int i;
+
+	if (IN_LOOPBACK(ntohl(addr)))
+		return &rds_loop_transport;
+
+	down_read(&rds_trans_sem);
+	for (i = 0; i < RDS_TRANS_COUNT; i++) {
+		trans = transports[i];
+
+		if (trans && (trans->laddr_check(addr) == 0) &&
+		    (!trans->t_owner || try_module_get(trans->t_owner))) {
+			ret = trans;
+			break;
+		}
+	}
+	up_read(&rds_trans_sem);
+
+	return ret;
+}
+
+/*
+ * This returns the number of stats entries in the snapshot and only
+ * copies them using the iter if there is enough space for them.  The
+ * caller passes in the global stats so that we can size and copy while
+ * holding the lock.
+ */
+unsigned int rds_trans_stats_info_copy(struct rds_info_iterator *iter,
+				       unsigned int avail)
+
+{
+	struct rds_transport *trans;
+	unsigned int total = 0;
+	unsigned int part;
+	int i;
+
+	rds_info_iter_unmap(iter);
+	down_read(&rds_trans_sem);
+
+	for (i = 0; i < RDS_TRANS_COUNT; i++)
+	{
+		trans = transports[i];
+		if (!trans || !trans->stats_info_copy)
+			continue;
+
+		part = trans->stats_info_copy(iter, avail);
+		avail -= min(avail, part);
+		total += part;
+	}
+
+	up_read(&rds_trans_sem);
+
+	return total;
+}
+