From b2cbae2c248776d81cc265ff7d48405b6a4cc463 Mon Sep 17 00:00:00 2001
From: Roland Dreier <roland@purestorage.com>
Date: Fri, 20 May 2011 11:46:11 -0700
Subject: RDMA: Add netlink infrastructure

Add basic RDMA netlink infrastructure that allows for registration of
RDMA clients for which data is to be exported and supplies message
construction callbacks.

Signed-off-by: Nir Muchtar <nirm@voltaire.com>

[ Reorganize a few things, add CONFIG_NET dependency.  - Roland ]

Signed-off-by: Roland Dreier <roland@purestorage.com>
---
 include/linux/netlink.h     |  1 +
 include/rdma/rdma_netlink.h | 64 +++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 65 insertions(+)
 create mode 100644 include/rdma/rdma_netlink.h

(limited to 'include')

diff --git a/include/linux/netlink.h b/include/linux/netlink.h
index 4c4ac3f3ce5..a9dd89552f9 100644
--- a/include/linux/netlink.h
+++ b/include/linux/netlink.h
@@ -24,6 +24,7 @@
 /* leave room for NETLINK_DM (DM Events) */
 #define NETLINK_SCSITRANSPORT	18	/* SCSI Transports */
 #define NETLINK_ECRYPTFS	19
+#define NETLINK_RDMA		20
 
 #define MAX_LINKS 32		
 
diff --git a/include/rdma/rdma_netlink.h b/include/rdma/rdma_netlink.h
new file mode 100644
index 00000000000..c983a193cec
--- /dev/null
+++ b/include/rdma/rdma_netlink.h
@@ -0,0 +1,64 @@
+#ifndef _RDMA_NETLINK_H
+#define _RDMA_NETLINK_H
+
+#define RDMA_NL_GET_CLIENT(type) ((type & (((1 << 6) - 1) << 10)) >> 10)
+#define RDMA_NL_GET_OP(type) (type & ((1 << 10) - 1))
+#define RDMA_NL_GET_TYPE(client, op) ((client << 10) + op)
+
+#ifdef __KERNEL__
+
+#include <linux/netlink.h>
+
+struct ibnl_client_cbs {
+	int (*dump)(struct sk_buff *skb, struct netlink_callback *nlcb);
+};
+
+int ibnl_init(void);
+void ibnl_cleanup(void);
+
+/**
+ * Add a a client to the list of IB netlink exporters.
+ * @index: Index of the added client
+ * @nops: Number of supported ops by the added client.
+ * @cb_table: A table for op->callback
+ *
+ * Returns 0 on success or a negative error code.
+ */
+int ibnl_add_client(int index, int nops,
+		    const struct ibnl_client_cbs cb_table[]);
+
+/**
+ * Remove a client from IB netlink.
+ * @index: Index of the removed IB client.
+ *
+ * Returns 0 on success or a negative error code.
+ */
+int ibnl_remove_client(int index);
+
+/**
+ * Put a new message in a supplied skb.
+ * @skb: The netlink skb.
+ * @nlh: Pointer to put the header of the new netlink message.
+ * @seq: The message sequence number.
+ * @len: The requested message length to allocate.
+ * @client: Calling IB netlink client.
+ * @op: message content op.
+ * Returns the allocated buffer on success and NULL on failure.
+ */
+void *ibnl_put_msg(struct sk_buff *skb, struct nlmsghdr **nlh, int seq,
+		   int len, int client, int op);
+/**
+ * Put a new attribute in a supplied skb.
+ * @skb: The netlink skb.
+ * @nlh: Header of the netlink message to append the attribute to.
+ * @len: The length of the attribute data.
+ * @data: The attribute data to put.
+ * @type: The attribute type.
+ * Returns the 0 and a negative error code on failure.
+ */
+int ibnl_put_attr(struct sk_buff *skb, struct nlmsghdr *nlh,
+		  int len, void *data, int type);
+
+#endif /* __KERNEL__ */
+
+#endif /* _RDMA_NETLINK_H */
-- 
cgit v1.2.3-70-g09d2


From 550e5ca77e96989c5e19f60e017205b2bcc615a5 Mon Sep 17 00:00:00 2001
From: Nir Muchtar <nirm@voltaire.com>
Date: Fri, 20 May 2011 11:46:11 -0700
Subject: RDMA/cma: Export enum cma_state in <rdma/rdma_cm.h>

Move cma.c's internal definition of enum cma_state to enum rdma_cm_state
in an exported header so that it can be exported via RDMA netlink.

Signed-off-by: Nir Muchtar <nirm@voltaire.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
---
 drivers/infiniband/core/cma.c | 170 ++++++++++++++++++++----------------------
 include/rdma/rdma_cm.h        |  14 ++++
 2 files changed, 94 insertions(+), 90 deletions(-)

(limited to 'include')

diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index 99dde874fbb..451d39e19cb 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -89,20 +89,6 @@ struct cma_device {
 	struct list_head	id_list;
 };
 
-enum cma_state {
-	CMA_IDLE,
-	CMA_ADDR_QUERY,
-	CMA_ADDR_RESOLVED,
-	CMA_ROUTE_QUERY,
-	CMA_ROUTE_RESOLVED,
-	CMA_CONNECT,
-	CMA_DISCONNECT,
-	CMA_ADDR_BOUND,
-	CMA_LISTEN,
-	CMA_DEVICE_REMOVAL,
-	CMA_DESTROYING
-};
-
 struct rdma_bind_list {
 	struct idr		*ps;
 	struct hlist_head	owners;
@@ -126,7 +112,7 @@ struct rdma_id_private {
 	struct list_head	mc_list;
 
 	int			internal_id;
-	enum cma_state		state;
+	enum rdma_cm_state	state;
 	spinlock_t		lock;
 	struct mutex		qp_mutex;
 
@@ -165,8 +151,8 @@ struct cma_multicast {
 struct cma_work {
 	struct work_struct	work;
 	struct rdma_id_private	*id;
-	enum cma_state		old_state;
-	enum cma_state		new_state;
+	enum rdma_cm_state	old_state;
+	enum rdma_cm_state	new_state;
 	struct rdma_cm_event	event;
 };
 
@@ -217,7 +203,7 @@ struct sdp_hah {
 #define CMA_VERSION 0x00
 #define SDP_MAJ_VERSION 0x2
 
-static int cma_comp(struct rdma_id_private *id_priv, enum cma_state comp)
+static int cma_comp(struct rdma_id_private *id_priv, enum rdma_cm_state comp)
 {
 	unsigned long flags;
 	int ret;
@@ -229,7 +215,7 @@ static int cma_comp(struct rdma_id_private *id_priv, enum cma_state comp)
 }
 
 static int cma_comp_exch(struct rdma_id_private *id_priv,
-			 enum cma_state comp, enum cma_state exch)
+			 enum rdma_cm_state comp, enum rdma_cm_state exch)
 {
 	unsigned long flags;
 	int ret;
@@ -241,11 +227,11 @@ static int cma_comp_exch(struct rdma_id_private *id_priv,
 	return ret;
 }
 
-static enum cma_state cma_exch(struct rdma_id_private *id_priv,
-			       enum cma_state exch)
+static enum rdma_cm_state cma_exch(struct rdma_id_private *id_priv,
+				   enum rdma_cm_state exch)
 {
 	unsigned long flags;
-	enum cma_state old;
+	enum rdma_cm_state old;
 
 	spin_lock_irqsave(&id_priv->lock, flags);
 	old = id_priv->state;
@@ -413,7 +399,7 @@ static void cma_deref_id(struct rdma_id_private *id_priv)
 }
 
 static int cma_disable_callback(struct rdma_id_private *id_priv,
-			      enum cma_state state)
+				enum rdma_cm_state state)
 {
 	mutex_lock(&id_priv->handler_mutex);
 	if (id_priv->state != state) {
@@ -437,7 +423,7 @@ struct rdma_cm_id *rdma_create_id(rdma_cm_event_handler event_handler,
 	if (!id_priv)
 		return ERR_PTR(-ENOMEM);
 
-	id_priv->state = CMA_IDLE;
+	id_priv->state = RDMA_CM_IDLE;
 	id_priv->id.context = context;
 	id_priv->id.event_handler = event_handler;
 	id_priv->id.ps = ps;
@@ -858,16 +844,16 @@ static void cma_cancel_listens(struct rdma_id_private *id_priv)
 }
 
 static void cma_cancel_operation(struct rdma_id_private *id_priv,
-				 enum cma_state state)
+				 enum rdma_cm_state state)
 {
 	switch (state) {
-	case CMA_ADDR_QUERY:
+	case RDMA_CM_ADDR_QUERY:
 		rdma_addr_cancel(&id_priv->id.route.addr.dev_addr);
 		break;
-	case CMA_ROUTE_QUERY:
+	case RDMA_CM_ROUTE_QUERY:
 		cma_cancel_route(id_priv);
 		break;
-	case CMA_LISTEN:
+	case RDMA_CM_LISTEN:
 		if (cma_any_addr((struct sockaddr *) &id_priv->id.route.addr.src_addr)
 				&& !id_priv->cma_dev)
 			cma_cancel_listens(id_priv);
@@ -918,10 +904,10 @@ static void cma_leave_mc_groups(struct rdma_id_private *id_priv)
 void rdma_destroy_id(struct rdma_cm_id *id)
 {
 	struct rdma_id_private *id_priv;
-	enum cma_state state;
+	enum rdma_cm_state state;
 
 	id_priv = container_of(id, struct rdma_id_private, id);
-	state = cma_exch(id_priv, CMA_DESTROYING);
+	state = cma_exch(id_priv, RDMA_CM_DESTROYING);
 	cma_cancel_operation(id_priv, state);
 
 	/*
@@ -1015,9 +1001,9 @@ static int cma_ib_handler(struct ib_cm_id *cm_id, struct ib_cm_event *ib_event)
 	int ret = 0;
 
 	if ((ib_event->event != IB_CM_TIMEWAIT_EXIT &&
-		cma_disable_callback(id_priv, CMA_CONNECT)) ||
+		cma_disable_callback(id_priv, RDMA_CM_CONNECT)) ||
 	    (ib_event->event == IB_CM_TIMEWAIT_EXIT &&
-		cma_disable_callback(id_priv, CMA_DISCONNECT)))
+		cma_disable_callback(id_priv, RDMA_CM_DISCONNECT)))
 		return 0;
 
 	memset(&event, 0, sizeof event);
@@ -1048,7 +1034,8 @@ static int cma_ib_handler(struct ib_cm_id *cm_id, struct ib_cm_event *ib_event)
 		event.status = -ETIMEDOUT; /* fall through */
 	case IB_CM_DREQ_RECEIVED:
 	case IB_CM_DREP_RECEIVED:
-		if (!cma_comp_exch(id_priv, CMA_CONNECT, CMA_DISCONNECT))
+		if (!cma_comp_exch(id_priv, RDMA_CM_CONNECT,
+				   RDMA_CM_DISCONNECT))
 			goto out;
 		event.event = RDMA_CM_EVENT_DISCONNECTED;
 		break;
@@ -1075,7 +1062,7 @@ static int cma_ib_handler(struct ib_cm_id *cm_id, struct ib_cm_event *ib_event)
 	if (ret) {
 		/* Destroy the CM ID by returning a non-zero value. */
 		id_priv->cm_id.ib = NULL;
-		cma_exch(id_priv, CMA_DESTROYING);
+		cma_exch(id_priv, RDMA_CM_DESTROYING);
 		mutex_unlock(&id_priv->handler_mutex);
 		rdma_destroy_id(&id_priv->id);
 		return ret;
@@ -1132,7 +1119,7 @@ static struct rdma_id_private *cma_new_conn_id(struct rdma_cm_id *listen_id,
 	rdma_addr_set_dgid(&rt->addr.dev_addr, &rt->path_rec[0].dgid);
 
 	id_priv = container_of(id, struct rdma_id_private, id);
-	id_priv->state = CMA_CONNECT;
+	id_priv->state = RDMA_CM_CONNECT;
 	return id_priv;
 
 destroy_id:
@@ -1172,7 +1159,7 @@ static struct rdma_id_private *cma_new_udp_id(struct rdma_cm_id *listen_id,
 	}
 
 	id_priv = container_of(id, struct rdma_id_private, id);
-	id_priv->state = CMA_CONNECT;
+	id_priv->state = RDMA_CM_CONNECT;
 	return id_priv;
 err:
 	rdma_destroy_id(id);
@@ -1201,7 +1188,7 @@ static int cma_req_handler(struct ib_cm_id *cm_id, struct ib_cm_event *ib_event)
 	int offset, ret;
 
 	listen_id = cm_id->context;
-	if (cma_disable_callback(listen_id, CMA_LISTEN))
+	if (cma_disable_callback(listen_id, RDMA_CM_LISTEN))
 		return -ECONNABORTED;
 
 	memset(&event, 0, sizeof event);
@@ -1243,7 +1230,7 @@ static int cma_req_handler(struct ib_cm_id *cm_id, struct ib_cm_event *ib_event)
 		 * while we're accessing the cm_id.
 		 */
 		mutex_lock(&lock);
-		if (cma_comp(conn_id, CMA_CONNECT) &&
+		if (cma_comp(conn_id, RDMA_CM_CONNECT) &&
 		    !cma_is_ud_ps(conn_id->id.ps))
 			ib_send_cm_mra(cm_id, CMA_CM_MRA_SETTING, NULL, 0);
 		mutex_unlock(&lock);
@@ -1257,7 +1244,7 @@ static int cma_req_handler(struct ib_cm_id *cm_id, struct ib_cm_event *ib_event)
 	conn_id->cm_id.ib = NULL;
 
 release_conn_id:
-	cma_exch(conn_id, CMA_DESTROYING);
+	cma_exch(conn_id, RDMA_CM_DESTROYING);
 	mutex_unlock(&conn_id->handler_mutex);
 	rdma_destroy_id(&conn_id->id);
 
@@ -1328,7 +1315,7 @@ static int cma_iw_handler(struct iw_cm_id *iw_id, struct iw_cm_event *iw_event)
 	struct sockaddr_in *sin;
 	int ret = 0;
 
-	if (cma_disable_callback(id_priv, CMA_CONNECT))
+	if (cma_disable_callback(id_priv, RDMA_CM_CONNECT))
 		return 0;
 
 	memset(&event, 0, sizeof event);
@@ -1371,7 +1358,7 @@ static int cma_iw_handler(struct iw_cm_id *iw_id, struct iw_cm_event *iw_event)
 	if (ret) {
 		/* Destroy the CM ID by returning a non-zero value. */
 		id_priv->cm_id.iw = NULL;
-		cma_exch(id_priv, CMA_DESTROYING);
+		cma_exch(id_priv, RDMA_CM_DESTROYING);
 		mutex_unlock(&id_priv->handler_mutex);
 		rdma_destroy_id(&id_priv->id);
 		return ret;
@@ -1393,7 +1380,7 @@ static int iw_conn_req_handler(struct iw_cm_id *cm_id,
 	struct ib_device_attr attr;
 
 	listen_id = cm_id->context;
-	if (cma_disable_callback(listen_id, CMA_LISTEN))
+	if (cma_disable_callback(listen_id, RDMA_CM_LISTEN))
 		return -ECONNABORTED;
 
 	/* Create a new RDMA id for the new IW CM ID */
@@ -1406,7 +1393,7 @@ static int iw_conn_req_handler(struct iw_cm_id *cm_id,
 	}
 	conn_id = container_of(new_cm_id, struct rdma_id_private, id);
 	mutex_lock_nested(&conn_id->handler_mutex, SINGLE_DEPTH_NESTING);
-	conn_id->state = CMA_CONNECT;
+	conn_id->state = RDMA_CM_CONNECT;
 
 	dev = ip_dev_find(&init_net, iw_event->local_addr.sin_addr.s_addr);
 	if (!dev) {
@@ -1461,7 +1448,7 @@ static int iw_conn_req_handler(struct iw_cm_id *cm_id,
 	if (ret) {
 		/* User wants to destroy the CM ID */
 		conn_id->cm_id.iw = NULL;
-		cma_exch(conn_id, CMA_DESTROYING);
+		cma_exch(conn_id, RDMA_CM_DESTROYING);
 		mutex_unlock(&conn_id->handler_mutex);
 		cma_deref_id(conn_id);
 		rdma_destroy_id(&conn_id->id);
@@ -1554,7 +1541,7 @@ static void cma_listen_on_dev(struct rdma_id_private *id_priv,
 
 	dev_id_priv = container_of(id, struct rdma_id_private, id);
 
-	dev_id_priv->state = CMA_ADDR_BOUND;
+	dev_id_priv->state = RDMA_CM_ADDR_BOUND;
 	memcpy(&id->route.addr.src_addr, &id_priv->id.route.addr.src_addr,
 	       ip_addr_size((struct sockaddr *) &id_priv->id.route.addr.src_addr));
 
@@ -1601,8 +1588,8 @@ static void cma_query_handler(int status, struct ib_sa_path_rec *path_rec,
 		route->num_paths = 1;
 		*route->path_rec = *path_rec;
 	} else {
-		work->old_state = CMA_ROUTE_QUERY;
-		work->new_state = CMA_ADDR_RESOLVED;
+		work->old_state = RDMA_CM_ROUTE_QUERY;
+		work->new_state = RDMA_CM_ADDR_RESOLVED;
 		work->event.event = RDMA_CM_EVENT_ROUTE_ERROR;
 		work->event.status = status;
 	}
@@ -1660,7 +1647,7 @@ static void cma_work_handler(struct work_struct *_work)
 		goto out;
 
 	if (id_priv->id.event_handler(&id_priv->id, &work->event)) {
-		cma_exch(id_priv, CMA_DESTROYING);
+		cma_exch(id_priv, RDMA_CM_DESTROYING);
 		destroy = 1;
 	}
 out:
@@ -1678,12 +1665,12 @@ static void cma_ndev_work_handler(struct work_struct *_work)
 	int destroy = 0;
 
 	mutex_lock(&id_priv->handler_mutex);
-	if (id_priv->state == CMA_DESTROYING ||
-	    id_priv->state == CMA_DEVICE_REMOVAL)
+	if (id_priv->state == RDMA_CM_DESTROYING ||
+	    id_priv->state == RDMA_CM_DEVICE_REMOVAL)
 		goto out;
 
 	if (id_priv->id.event_handler(&id_priv->id, &work->event)) {
-		cma_exch(id_priv, CMA_DESTROYING);
+		cma_exch(id_priv, RDMA_CM_DESTROYING);
 		destroy = 1;
 	}
 
@@ -1707,8 +1694,8 @@ static int cma_resolve_ib_route(struct rdma_id_private *id_priv, int timeout_ms)
 
 	work->id = id_priv;
 	INIT_WORK(&work->work, cma_work_handler);
-	work->old_state = CMA_ROUTE_QUERY;
-	work->new_state = CMA_ROUTE_RESOLVED;
+	work->old_state = RDMA_CM_ROUTE_QUERY;
+	work->new_state = RDMA_CM_ROUTE_RESOLVED;
 	work->event.event = RDMA_CM_EVENT_ROUTE_RESOLVED;
 
 	route->path_rec = kmalloc(sizeof *route->path_rec, GFP_KERNEL);
@@ -1737,7 +1724,8 @@ int rdma_set_ib_paths(struct rdma_cm_id *id,
 	int ret;
 
 	id_priv = container_of(id, struct rdma_id_private, id);
-	if (!cma_comp_exch(id_priv, CMA_ADDR_RESOLVED, CMA_ROUTE_RESOLVED))
+	if (!cma_comp_exch(id_priv, RDMA_CM_ADDR_RESOLVED,
+			   RDMA_CM_ROUTE_RESOLVED))
 		return -EINVAL;
 
 	id->route.path_rec = kmemdup(path_rec, sizeof *path_rec * num_paths,
@@ -1750,7 +1738,7 @@ int rdma_set_ib_paths(struct rdma_cm_id *id,
 	id->route.num_paths = num_paths;
 	return 0;
 err:
-	cma_comp_exch(id_priv, CMA_ROUTE_RESOLVED, CMA_ADDR_RESOLVED);
+	cma_comp_exch(id_priv, RDMA_CM_ROUTE_RESOLVED, RDMA_CM_ADDR_RESOLVED);
 	return ret;
 }
 EXPORT_SYMBOL(rdma_set_ib_paths);
@@ -1765,8 +1753,8 @@ static int cma_resolve_iw_route(struct rdma_id_private *id_priv, int timeout_ms)
 
 	work->id = id_priv;
 	INIT_WORK(&work->work, cma_work_handler);
-	work->old_state = CMA_ROUTE_QUERY;
-	work->new_state = CMA_ROUTE_RESOLVED;
+	work->old_state = RDMA_CM_ROUTE_QUERY;
+	work->new_state = RDMA_CM_ROUTE_RESOLVED;
 	work->event.event = RDMA_CM_EVENT_ROUTE_RESOLVED;
 	queue_work(cma_wq, &work->work);
 	return 0;
@@ -1830,8 +1818,8 @@ static int cma_resolve_iboe_route(struct rdma_id_private *id_priv)
 		goto err2;
 	}
 
-	work->old_state = CMA_ROUTE_QUERY;
-	work->new_state = CMA_ROUTE_RESOLVED;
+	work->old_state = RDMA_CM_ROUTE_QUERY;
+	work->new_state = RDMA_CM_ROUTE_RESOLVED;
 	work->event.event = RDMA_CM_EVENT_ROUTE_RESOLVED;
 	work->event.status = 0;
 
@@ -1853,7 +1841,7 @@ int rdma_resolve_route(struct rdma_cm_id *id, int timeout_ms)
 	int ret;
 
 	id_priv = container_of(id, struct rdma_id_private, id);
-	if (!cma_comp_exch(id_priv, CMA_ADDR_RESOLVED, CMA_ROUTE_QUERY))
+	if (!cma_comp_exch(id_priv, RDMA_CM_ADDR_RESOLVED, RDMA_CM_ROUTE_QUERY))
 		return -EINVAL;
 
 	atomic_inc(&id_priv->refcount);
@@ -1882,7 +1870,7 @@ int rdma_resolve_route(struct rdma_cm_id *id, int timeout_ms)
 
 	return 0;
 err:
-	cma_comp_exch(id_priv, CMA_ROUTE_QUERY, CMA_ADDR_RESOLVED);
+	cma_comp_exch(id_priv, RDMA_CM_ROUTE_QUERY, RDMA_CM_ADDR_RESOLVED);
 	cma_deref_id(id_priv);
 	return ret;
 }
@@ -1941,14 +1929,16 @@ static void addr_handler(int status, struct sockaddr *src_addr,
 
 	memset(&event, 0, sizeof event);
 	mutex_lock(&id_priv->handler_mutex);
-	if (!cma_comp_exch(id_priv, CMA_ADDR_QUERY, CMA_ADDR_RESOLVED))
+	if (!cma_comp_exch(id_priv, RDMA_CM_ADDR_QUERY,
+			   RDMA_CM_ADDR_RESOLVED))
 		goto out;
 
 	if (!status && !id_priv->cma_dev)
 		status = cma_acquire_dev(id_priv);
 
 	if (status) {
-		if (!cma_comp_exch(id_priv, CMA_ADDR_RESOLVED, CMA_ADDR_BOUND))
+		if (!cma_comp_exch(id_priv, RDMA_CM_ADDR_RESOLVED,
+				   RDMA_CM_ADDR_BOUND))
 			goto out;
 		event.event = RDMA_CM_EVENT_ADDR_ERROR;
 		event.status = status;
@@ -1959,7 +1949,7 @@ static void addr_handler(int status, struct sockaddr *src_addr,
 	}
 
 	if (id_priv->id.event_handler(&id_priv->id, &event)) {
-		cma_exch(id_priv, CMA_DESTROYING);
+		cma_exch(id_priv, RDMA_CM_DESTROYING);
 		mutex_unlock(&id_priv->handler_mutex);
 		cma_deref_id(id_priv);
 		rdma_destroy_id(&id_priv->id);
@@ -2004,8 +1994,8 @@ static int cma_resolve_loopback(struct rdma_id_private *id_priv)
 
 	work->id = id_priv;
 	INIT_WORK(&work->work, cma_work_handler);
-	work->old_state = CMA_ADDR_QUERY;
-	work->new_state = CMA_ADDR_RESOLVED;
+	work->old_state = RDMA_CM_ADDR_QUERY;
+	work->new_state = RDMA_CM_ADDR_RESOLVED;
 	work->event.event = RDMA_CM_EVENT_ADDR_RESOLVED;
 	queue_work(cma_wq, &work->work);
 	return 0;
@@ -2034,13 +2024,13 @@ int rdma_resolve_addr(struct rdma_cm_id *id, struct sockaddr *src_addr,
 	int ret;
 
 	id_priv = container_of(id, struct rdma_id_private, id);
-	if (id_priv->state == CMA_IDLE) {
+	if (id_priv->state == RDMA_CM_IDLE) {
 		ret = cma_bind_addr(id, src_addr, dst_addr);
 		if (ret)
 			return ret;
 	}
 
-	if (!cma_comp_exch(id_priv, CMA_ADDR_BOUND, CMA_ADDR_QUERY))
+	if (!cma_comp_exch(id_priv, RDMA_CM_ADDR_BOUND, RDMA_CM_ADDR_QUERY))
 		return -EINVAL;
 
 	atomic_inc(&id_priv->refcount);
@@ -2056,7 +2046,7 @@ int rdma_resolve_addr(struct rdma_cm_id *id, struct sockaddr *src_addr,
 
 	return 0;
 err:
-	cma_comp_exch(id_priv, CMA_ADDR_QUERY, CMA_ADDR_BOUND);
+	cma_comp_exch(id_priv, RDMA_CM_ADDR_QUERY, RDMA_CM_ADDR_BOUND);
 	cma_deref_id(id_priv);
 	return ret;
 }
@@ -2070,7 +2060,7 @@ int rdma_set_reuseaddr(struct rdma_cm_id *id, int reuse)
 
 	id_priv = container_of(id, struct rdma_id_private, id);
 	spin_lock_irqsave(&id_priv->lock, flags);
-	if (id_priv->state == CMA_IDLE) {
+	if (id_priv->state == RDMA_CM_IDLE) {
 		id_priv->reuseaddr = reuse;
 		ret = 0;
 	} else {
@@ -2177,7 +2167,7 @@ static int cma_check_port(struct rdma_bind_list *bind_list,
 		if (id_priv == cur_id)
 			continue;
 
-		if ((cur_id->state == CMA_LISTEN) ||
+		if ((cur_id->state == RDMA_CM_LISTEN) ||
 		    !reuseaddr || !cur_id->reuseaddr) {
 			cur_addr = (struct sockaddr *) &cur_id->id.route.addr.src_addr;
 			if (cma_any_addr(cur_addr))
@@ -2280,14 +2270,14 @@ int rdma_listen(struct rdma_cm_id *id, int backlog)
 	int ret;
 
 	id_priv = container_of(id, struct rdma_id_private, id);
-	if (id_priv->state == CMA_IDLE) {
+	if (id_priv->state == RDMA_CM_IDLE) {
 		((struct sockaddr *) &id->route.addr.src_addr)->sa_family = AF_INET;
 		ret = rdma_bind_addr(id, (struct sockaddr *) &id->route.addr.src_addr);
 		if (ret)
 			return ret;
 	}
 
-	if (!cma_comp_exch(id_priv, CMA_ADDR_BOUND, CMA_LISTEN))
+	if (!cma_comp_exch(id_priv, RDMA_CM_ADDR_BOUND, RDMA_CM_LISTEN))
 		return -EINVAL;
 
 	if (id_priv->reuseaddr) {
@@ -2319,7 +2309,7 @@ int rdma_listen(struct rdma_cm_id *id, int backlog)
 	return 0;
 err:
 	id_priv->backlog = 0;
-	cma_comp_exch(id_priv, CMA_LISTEN, CMA_ADDR_BOUND);
+	cma_comp_exch(id_priv, RDMA_CM_LISTEN, RDMA_CM_ADDR_BOUND);
 	return ret;
 }
 EXPORT_SYMBOL(rdma_listen);
@@ -2333,7 +2323,7 @@ int rdma_bind_addr(struct rdma_cm_id *id, struct sockaddr *addr)
 		return -EAFNOSUPPORT;
 
 	id_priv = container_of(id, struct rdma_id_private, id);
-	if (!cma_comp_exch(id_priv, CMA_IDLE, CMA_ADDR_BOUND))
+	if (!cma_comp_exch(id_priv, RDMA_CM_IDLE, RDMA_CM_ADDR_BOUND))
 		return -EINVAL;
 
 	ret = cma_check_linklocal(&id->route.addr.dev_addr, addr);
@@ -2360,7 +2350,7 @@ err2:
 	if (id_priv->cma_dev)
 		cma_release_dev(id_priv);
 err1:
-	cma_comp_exch(id_priv, CMA_ADDR_BOUND, CMA_IDLE);
+	cma_comp_exch(id_priv, RDMA_CM_ADDR_BOUND, RDMA_CM_IDLE);
 	return ret;
 }
 EXPORT_SYMBOL(rdma_bind_addr);
@@ -2433,7 +2423,7 @@ static int cma_sidr_rep_handler(struct ib_cm_id *cm_id,
 	struct ib_cm_sidr_rep_event_param *rep = &ib_event->param.sidr_rep_rcvd;
 	int ret = 0;
 
-	if (cma_disable_callback(id_priv, CMA_CONNECT))
+	if (cma_disable_callback(id_priv, RDMA_CM_CONNECT))
 		return 0;
 
 	memset(&event, 0, sizeof event);
@@ -2479,7 +2469,7 @@ static int cma_sidr_rep_handler(struct ib_cm_id *cm_id,
 	if (ret) {
 		/* Destroy the CM ID by returning a non-zero value. */
 		id_priv->cm_id.ib = NULL;
-		cma_exch(id_priv, CMA_DESTROYING);
+		cma_exch(id_priv, RDMA_CM_DESTROYING);
 		mutex_unlock(&id_priv->handler_mutex);
 		rdma_destroy_id(&id_priv->id);
 		return ret;
@@ -2645,7 +2635,7 @@ int rdma_connect(struct rdma_cm_id *id, struct rdma_conn_param *conn_param)
 	int ret;
 
 	id_priv = container_of(id, struct rdma_id_private, id);
-	if (!cma_comp_exch(id_priv, CMA_ROUTE_RESOLVED, CMA_CONNECT))
+	if (!cma_comp_exch(id_priv, RDMA_CM_ROUTE_RESOLVED, RDMA_CM_CONNECT))
 		return -EINVAL;
 
 	if (!id->qp) {
@@ -2672,7 +2662,7 @@ int rdma_connect(struct rdma_cm_id *id, struct rdma_conn_param *conn_param)
 
 	return 0;
 err:
-	cma_comp_exch(id_priv, CMA_CONNECT, CMA_ROUTE_RESOLVED);
+	cma_comp_exch(id_priv, RDMA_CM_CONNECT, RDMA_CM_ROUTE_RESOLVED);
 	return ret;
 }
 EXPORT_SYMBOL(rdma_connect);
@@ -2758,7 +2748,7 @@ int rdma_accept(struct rdma_cm_id *id, struct rdma_conn_param *conn_param)
 	int ret;
 
 	id_priv = container_of(id, struct rdma_id_private, id);
-	if (!cma_comp(id_priv, CMA_CONNECT))
+	if (!cma_comp(id_priv, RDMA_CM_CONNECT))
 		return -EINVAL;
 
 	if (!id->qp && conn_param) {
@@ -2887,8 +2877,8 @@ static int cma_ib_mc_handler(int status, struct ib_sa_multicast *multicast)
 	int ret;
 
 	id_priv = mc->id_priv;
-	if (cma_disable_callback(id_priv, CMA_ADDR_BOUND) &&
-	    cma_disable_callback(id_priv, CMA_ADDR_RESOLVED))
+	if (cma_disable_callback(id_priv, RDMA_CM_ADDR_BOUND) &&
+	    cma_disable_callback(id_priv, RDMA_CM_ADDR_RESOLVED))
 		return 0;
 
 	mutex_lock(&id_priv->qp_mutex);
@@ -2912,7 +2902,7 @@ static int cma_ib_mc_handler(int status, struct ib_sa_multicast *multicast)
 
 	ret = id_priv->id.event_handler(&id_priv->id, &event);
 	if (ret) {
-		cma_exch(id_priv, CMA_DESTROYING);
+		cma_exch(id_priv, RDMA_CM_DESTROYING);
 		mutex_unlock(&id_priv->handler_mutex);
 		rdma_destroy_id(&id_priv->id);
 		return 0;
@@ -3095,8 +3085,8 @@ int rdma_join_multicast(struct rdma_cm_id *id, struct sockaddr *addr,
 	int ret;
 
 	id_priv = container_of(id, struct rdma_id_private, id);
-	if (!cma_comp(id_priv, CMA_ADDR_BOUND) &&
-	    !cma_comp(id_priv, CMA_ADDR_RESOLVED))
+	if (!cma_comp(id_priv, RDMA_CM_ADDR_BOUND) &&
+	    !cma_comp(id_priv, RDMA_CM_ADDR_RESOLVED))
 		return -EINVAL;
 
 	mc = kmalloc(sizeof *mc, GFP_KERNEL);
@@ -3261,19 +3251,19 @@ static void cma_add_one(struct ib_device *device)
 static int cma_remove_id_dev(struct rdma_id_private *id_priv)
 {
 	struct rdma_cm_event event;
-	enum cma_state state;
+	enum rdma_cm_state state;
 	int ret = 0;
 
 	/* Record that we want to remove the device */
-	state = cma_exch(id_priv, CMA_DEVICE_REMOVAL);
-	if (state == CMA_DESTROYING)
+	state = cma_exch(id_priv, RDMA_CM_DEVICE_REMOVAL);
+	if (state == RDMA_CM_DESTROYING)
 		return 0;
 
 	cma_cancel_operation(id_priv, state);
 	mutex_lock(&id_priv->handler_mutex);
 
 	/* Check for destruction from another callback. */
-	if (!cma_comp(id_priv, CMA_DEVICE_REMOVAL))
+	if (!cma_comp(id_priv, RDMA_CM_DEVICE_REMOVAL))
 		goto out;
 
 	memset(&event, 0, sizeof event);
diff --git a/include/rdma/rdma_cm.h b/include/rdma/rdma_cm.h
index 169f7a53fb0..d5b2265b7ce 100644
--- a/include/rdma/rdma_cm.h
+++ b/include/rdma/rdma_cm.h
@@ -111,6 +111,20 @@ struct rdma_cm_event {
 	} param;
 };
 
+enum rdma_cm_state {
+	RDMA_CM_IDLE,
+	RDMA_CM_ADDR_QUERY,
+	RDMA_CM_ADDR_RESOLVED,
+	RDMA_CM_ROUTE_QUERY,
+	RDMA_CM_ROUTE_RESOLVED,
+	RDMA_CM_CONNECT,
+	RDMA_CM_DISCONNECT,
+	RDMA_CM_ADDR_BOUND,
+	RDMA_CM_LISTEN,
+	RDMA_CM_DEVICE_REMOVAL,
+	RDMA_CM_DESTROYING
+};
+
 struct rdma_cm_id;
 
 /**
-- 
cgit v1.2.3-70-g09d2


From 9a7147b506ccae8552b0cf218b3c02982012eb4d Mon Sep 17 00:00:00 2001
From: Roland Dreier <roland@purestorage.com>
Date: Tue, 10 May 2011 12:36:01 -0700
Subject: RDMA: Update exported headers list

Various RDMA headers are intended to be exported to userspace, so add
them to the headers-y list.  Add a (strictly speaking, superfluous)
include of <linux/types.h> to avoid a headers_check warning.

Signed-off-by: Roland Dreier <roland@purestorage.com>
---
 include/rdma/Kbuild       | 5 +++++
 include/rdma/ib_user_cm.h | 1 +
 2 files changed, 6 insertions(+)

(limited to 'include')

diff --git a/include/rdma/Kbuild b/include/rdma/Kbuild
index e7c04321655..ea56f76c0c2 100644
--- a/include/rdma/Kbuild
+++ b/include/rdma/Kbuild
@@ -1 +1,6 @@
+header-y += ib_user_cm.h
 header-y += ib_user_mad.h
+header-y += ib_user_sa.h
+header-y += ib_user_verbs.h
+header-y += rdma_netlink.h
+header-y += rdma_user_cm.h
diff --git a/include/rdma/ib_user_cm.h b/include/rdma/ib_user_cm.h
index bd3d380781e..f79014aa28f 100644
--- a/include/rdma/ib_user_cm.h
+++ b/include/rdma/ib_user_cm.h
@@ -34,6 +34,7 @@
 #ifndef IB_USER_CM_H
 #define IB_USER_CM_H
 
+#include <linux/types.h>
 #include <rdma/ib_user_sa.h>
 
 #define IB_USER_CM_ABI_VERSION 5
-- 
cgit v1.2.3-70-g09d2


From b26f9b9949013fec31b23c426fc463164ae08891 Mon Sep 17 00:00:00 2001
From: Sean Hefty <sean.hefty@intel.com>
Date: Thu, 1 Apr 2010 17:08:41 +0000
Subject: RDMA/cma: Pass QP type into rdma_create_id()

The RDMA CM currently infers the QP type from the port space selected
by the user.  In the future (eg with RDMA_PS_IB or XRC), there may not
be a 1-1 correspondence between port space and QP type.  For netlink
export of RDMA CM state, we want to export the QP type to userspace,
so it is cleaner to explicitly associate a QP type to an ID.

Modify rdma_create_id() to allow the user to specify the QP type, and
use it to make our selections of datagram versus connected mode.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
---
 drivers/infiniband/core/cma.c            | 35 +++++++++++++++-----------------
 drivers/infiniband/core/ucma.c           | 27 ++++++++++++++++++++----
 drivers/infiniband/ulp/iser/iser_verbs.c |  2 +-
 include/rdma/rdma_cm.h                   |  5 ++++-
 net/9p/trans_rdma.c                      |  3 ++-
 net/rds/ib.c                             |  2 +-
 net/rds/ib_cm.c                          |  2 +-
 net/rds/iw.c                             |  2 +-
 net/rds/iw_cm.c                          |  2 +-
 net/rds/rdma_transport.c                 |  3 ++-
 net/sunrpc/xprtrdma/svc_rdma_transport.c |  3 ++-
 net/sunrpc/xprtrdma/verbs.c              |  2 +-
 12 files changed, 55 insertions(+), 33 deletions(-)

(limited to 'include')

diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index 451d39e19cb..44be1c9ed05 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -265,11 +265,6 @@ static inline void sdp_set_ip_ver(struct sdp_hh *hh, u8 ip_ver)
 	hh->ip_version = (ip_ver << 4) | (hh->ip_version & 0xF);
 }
 
-static inline int cma_is_ud_ps(enum rdma_port_space ps)
-{
-	return (ps == RDMA_PS_UDP || ps == RDMA_PS_IPOIB);
-}
-
 static void cma_attach_to_dev(struct rdma_id_private *id_priv,
 			      struct cma_device *cma_dev)
 {
@@ -415,7 +410,8 @@ static int cma_has_cm_dev(struct rdma_id_private *id_priv)
 }
 
 struct rdma_cm_id *rdma_create_id(rdma_cm_event_handler event_handler,
-				  void *context, enum rdma_port_space ps)
+				  void *context, enum rdma_port_space ps,
+				  enum ib_qp_type qp_type)
 {
 	struct rdma_id_private *id_priv;
 
@@ -427,6 +423,7 @@ struct rdma_cm_id *rdma_create_id(rdma_cm_event_handler event_handler,
 	id_priv->id.context = context;
 	id_priv->id.event_handler = event_handler;
 	id_priv->id.ps = ps;
+	id_priv->id.qp_type = qp_type;
 	spin_lock_init(&id_priv->lock);
 	mutex_init(&id_priv->qp_mutex);
 	init_completion(&id_priv->comp);
@@ -494,7 +491,7 @@ int rdma_create_qp(struct rdma_cm_id *id, struct ib_pd *pd,
 	if (IS_ERR(qp))
 		return PTR_ERR(qp);
 
-	if (cma_is_ud_ps(id_priv->id.ps))
+	if (id->qp_type == IB_QPT_UD)
 		ret = cma_init_ud_qp(id_priv, qp);
 	else
 		ret = cma_init_conn_qp(id_priv, qp);
@@ -622,7 +619,7 @@ static int cma_ib_init_qp_attr(struct rdma_id_private *id_priv,
 	qp_attr->port_num = id_priv->id.port_num;
 	*qp_attr_mask = IB_QP_STATE | IB_QP_PKEY_INDEX | IB_QP_PORT;
 
-	if (cma_is_ud_ps(id_priv->id.ps)) {
+	if (id_priv->id.qp_type == IB_QPT_UD) {
 		ret = cma_set_qkey(id_priv);
 		if (ret)
 			return ret;
@@ -645,7 +642,7 @@ int rdma_init_qp_attr(struct rdma_cm_id *id, struct ib_qp_attr *qp_attr,
 	id_priv = container_of(id, struct rdma_id_private, id);
 	switch (rdma_node_get_transport(id_priv->id.device->node_type)) {
 	case RDMA_TRANSPORT_IB:
-		if (!id_priv->cm_id.ib || cma_is_ud_ps(id_priv->id.ps))
+		if (!id_priv->cm_id.ib || (id_priv->id.qp_type == IB_QPT_UD))
 			ret = cma_ib_init_qp_attr(id_priv, qp_attr, qp_attr_mask);
 		else
 			ret = ib_cm_init_qp_attr(id_priv->cm_id.ib, qp_attr,
@@ -1088,7 +1085,7 @@ static struct rdma_id_private *cma_new_conn_id(struct rdma_cm_id *listen_id,
 		goto err;
 
 	id = rdma_create_id(listen_id->event_handler, listen_id->context,
-			    listen_id->ps);
+			    listen_id->ps, ib_event->param.req_rcvd.qp_type);
 	if (IS_ERR(id))
 		goto err;
 
@@ -1139,7 +1136,7 @@ static struct rdma_id_private *cma_new_udp_id(struct rdma_cm_id *listen_id,
 	int ret;
 
 	id = rdma_create_id(listen_id->event_handler, listen_id->context,
-			    listen_id->ps);
+			    listen_id->ps, IB_QPT_UD);
 	if (IS_ERR(id))
 		return NULL;
 
@@ -1194,7 +1191,7 @@ static int cma_req_handler(struct ib_cm_id *cm_id, struct ib_cm_event *ib_event)
 	memset(&event, 0, sizeof event);
 	offset = cma_user_data_offset(listen_id->id.ps);
 	event.event = RDMA_CM_EVENT_CONNECT_REQUEST;
-	if (cma_is_ud_ps(listen_id->id.ps)) {
+	if (listen_id->id.qp_type == IB_QPT_UD) {
 		conn_id = cma_new_udp_id(&listen_id->id, ib_event);
 		event.param.ud.private_data = ib_event->private_data + offset;
 		event.param.ud.private_data_len =
@@ -1230,8 +1227,7 @@ static int cma_req_handler(struct ib_cm_id *cm_id, struct ib_cm_event *ib_event)
 		 * while we're accessing the cm_id.
 		 */
 		mutex_lock(&lock);
-		if (cma_comp(conn_id, RDMA_CM_CONNECT) &&
-		    !cma_is_ud_ps(conn_id->id.ps))
+		if (cma_comp(conn_id, RDMA_CM_CONNECT) && (conn_id->id.qp_type != IB_QPT_UD))
 			ib_send_cm_mra(cm_id, CMA_CM_MRA_SETTING, NULL, 0);
 		mutex_unlock(&lock);
 		mutex_unlock(&conn_id->handler_mutex);
@@ -1386,7 +1382,7 @@ static int iw_conn_req_handler(struct iw_cm_id *cm_id,
 	/* Create a new RDMA id for the new IW CM ID */
 	new_cm_id = rdma_create_id(listen_id->id.event_handler,
 				   listen_id->id.context,
-				   RDMA_PS_TCP);
+				   RDMA_PS_TCP, IB_QPT_RC);
 	if (IS_ERR(new_cm_id)) {
 		ret = -ENOMEM;
 		goto out;
@@ -1535,7 +1531,8 @@ static void cma_listen_on_dev(struct rdma_id_private *id_priv,
 	struct rdma_cm_id *id;
 	int ret;
 
-	id = rdma_create_id(cma_listen_handler, id_priv, id_priv->id.ps);
+	id = rdma_create_id(cma_listen_handler, id_priv, id_priv->id.ps,
+			    id_priv->id.qp_type);
 	if (IS_ERR(id))
 		return;
 
@@ -2645,7 +2642,7 @@ int rdma_connect(struct rdma_cm_id *id, struct rdma_conn_param *conn_param)
 
 	switch (rdma_node_get_transport(id->device->node_type)) {
 	case RDMA_TRANSPORT_IB:
-		if (cma_is_ud_ps(id->ps))
+		if (id->qp_type == IB_QPT_UD)
 			ret = cma_resolve_ib_udp(id_priv, conn_param);
 		else
 			ret = cma_connect_ib(id_priv, conn_param);
@@ -2758,7 +2755,7 @@ int rdma_accept(struct rdma_cm_id *id, struct rdma_conn_param *conn_param)
 
 	switch (rdma_node_get_transport(id->device->node_type)) {
 	case RDMA_TRANSPORT_IB:
-		if (cma_is_ud_ps(id->ps))
+		if (id->qp_type == IB_QPT_UD)
 			ret = cma_send_sidr_rep(id_priv, IB_SIDR_SUCCESS,
 						conn_param->private_data,
 						conn_param->private_data_len);
@@ -2819,7 +2816,7 @@ int rdma_reject(struct rdma_cm_id *id, const void *private_data,
 
 	switch (rdma_node_get_transport(id->device->node_type)) {
 	case RDMA_TRANSPORT_IB:
-		if (cma_is_ud_ps(id->ps))
+		if (id->qp_type == IB_QPT_UD)
 			ret = cma_send_sidr_rep(id_priv, IB_SIDR_REJECT,
 						private_data, private_data_len);
 		else
diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c
index b3fa798525b..7109d5d23ba 100644
--- a/drivers/infiniband/core/ucma.c
+++ b/drivers/infiniband/core/ucma.c
@@ -367,13 +367,28 @@ done:
 	return ret;
 }
 
-static ssize_t ucma_create_id(struct ucma_file *file,
-				const char __user *inbuf,
-				int in_len, int out_len)
+static int ucma_get_qp_type(struct rdma_ucm_create_id *cmd, enum ib_qp_type *qp_type)
+{
+	switch (cmd->ps) {
+	case RDMA_PS_TCP:
+		*qp_type = IB_QPT_RC;
+		return 0;
+	case RDMA_PS_UDP:
+	case RDMA_PS_IPOIB:
+		*qp_type = IB_QPT_UD;
+		return 0;
+	default:
+		return -EINVAL;
+	}
+}
+
+static ssize_t ucma_create_id(struct ucma_file *file, const char __user *inbuf,
+			      int in_len, int out_len)
 {
 	struct rdma_ucm_create_id cmd;
 	struct rdma_ucm_create_id_resp resp;
 	struct ucma_context *ctx;
+	enum ib_qp_type qp_type;
 	int ret;
 
 	if (out_len < sizeof(resp))
@@ -382,6 +397,10 @@ static ssize_t ucma_create_id(struct ucma_file *file,
 	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
 		return -EFAULT;
 
+	ret = ucma_get_qp_type(&cmd, &qp_type);
+	if (ret)
+		return ret;
+
 	mutex_lock(&file->mut);
 	ctx = ucma_alloc_ctx(file);
 	mutex_unlock(&file->mut);
@@ -389,7 +408,7 @@ static ssize_t ucma_create_id(struct ucma_file *file,
 		return -ENOMEM;
 
 	ctx->uid = cmd.uid;
-	ctx->cm_id = rdma_create_id(ucma_event_handler, ctx, cmd.ps);
+	ctx->cm_id = rdma_create_id(ucma_event_handler, ctx, cmd.ps, qp_type);
 	if (IS_ERR(ctx->cm_id)) {
 		ret = PTR_ERR(ctx->cm_id);
 		goto err1;
diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c
index 9876865732f..ede1475bee0 100644
--- a/drivers/infiniband/ulp/iser/iser_verbs.c
+++ b/drivers/infiniband/ulp/iser/iser_verbs.c
@@ -548,7 +548,7 @@ int iser_connect(struct iser_conn   *ib_conn,
 	iser_conn_get(ib_conn); /* ref ib conn's cma id */
 	ib_conn->cma_id = rdma_create_id(iser_cma_handler,
 					     (void *)ib_conn,
-					     RDMA_PS_TCP);
+					     RDMA_PS_TCP, IB_QPT_RC);
 	if (IS_ERR(ib_conn->cma_id)) {
 		err = PTR_ERR(ib_conn->cma_id);
 		iser_err("rdma_create_id failed: %d\n", err);
diff --git a/include/rdma/rdma_cm.h b/include/rdma/rdma_cm.h
index d5b2265b7ce..26977c149c4 100644
--- a/include/rdma/rdma_cm.h
+++ b/include/rdma/rdma_cm.h
@@ -144,6 +144,7 @@ struct rdma_cm_id {
 	rdma_cm_event_handler	 event_handler;
 	struct rdma_route	 route;
 	enum rdma_port_space	 ps;
+	enum ib_qp_type		 qp_type;
 	u8			 port_num;
 };
 
@@ -154,9 +155,11 @@ struct rdma_cm_id {
  *   returned rdma_id.
  * @context: User specified context associated with the id.
  * @ps: RDMA port space.
+ * @qp_type: type of queue pair associated with the id.
  */
 struct rdma_cm_id *rdma_create_id(rdma_cm_event_handler event_handler,
-				  void *context, enum rdma_port_space ps);
+				  void *context, enum rdma_port_space ps,
+				  enum ib_qp_type qp_type);
 
 /**
   * rdma_destroy_id - Destroys an RDMA identifier.
diff --git a/net/9p/trans_rdma.c b/net/9p/trans_rdma.c
index 150e0c4bbf4..3640e83eef8 100644
--- a/net/9p/trans_rdma.c
+++ b/net/9p/trans_rdma.c
@@ -591,7 +591,8 @@ rdma_create_trans(struct p9_client *client, const char *addr, char *args)
 		return -ENOMEM;
 
 	/* Create the RDMA CM ID */
-	rdma->cm_id = rdma_create_id(p9_cm_event_handler, client, RDMA_PS_TCP);
+	rdma->cm_id = rdma_create_id(p9_cm_event_handler, client, RDMA_PS_TCP,
+				     IB_QPT_RC);
 	if (IS_ERR(rdma->cm_id))
 		goto error;
 
diff --git a/net/rds/ib.c b/net/rds/ib.c
index cce19f95c62..3b83086bcc3 100644
--- a/net/rds/ib.c
+++ b/net/rds/ib.c
@@ -325,7 +325,7 @@ static int rds_ib_laddr_check(__be32 addr)
 	/* 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);
+	cm_id = rdma_create_id(NULL, NULL, RDMA_PS_TCP, IB_QPT_RC);
 	if (IS_ERR(cm_id))
 		return PTR_ERR(cm_id);
 
diff --git a/net/rds/ib_cm.c b/net/rds/ib_cm.c
index ee369d201a6..fd453dd5124 100644
--- a/net/rds/ib_cm.c
+++ b/net/rds/ib_cm.c
@@ -587,7 +587,7 @@ int rds_ib_conn_connect(struct rds_connection *conn)
 	/* 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);
+				     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;
diff --git a/net/rds/iw.c b/net/rds/iw.c
index 5a9676fe594..f7474844f09 100644
--- a/net/rds/iw.c
+++ b/net/rds/iw.c
@@ -226,7 +226,7 @@ static int rds_iw_laddr_check(__be32 addr)
 	/* 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);
+	cm_id = rdma_create_id(NULL, NULL, RDMA_PS_TCP, IB_QPT_RC);
 	if (IS_ERR(cm_id))
 		return PTR_ERR(cm_id);
 
diff --git a/net/rds/iw_cm.c b/net/rds/iw_cm.c
index 3a60a15d1b4..c12db66f24c 100644
--- a/net/rds/iw_cm.c
+++ b/net/rds/iw_cm.c
@@ -522,7 +522,7 @@ int rds_iw_conn_connect(struct rds_connection *conn)
 	/* 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);
+				     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;
diff --git a/net/rds/rdma_transport.c b/net/rds/rdma_transport.c
index 4195a053982..f8760e1b668 100644
--- a/net/rds/rdma_transport.c
+++ b/net/rds/rdma_transport.c
@@ -158,7 +158,8 @@ static int rds_rdma_listen_init(void)
 	struct rdma_cm_id *cm_id;
 	int ret;
 
-	cm_id = rdma_create_id(rds_rdma_cm_event_handler, NULL, RDMA_PS_TCP);
+	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, "
diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c
index 1a10dcd999e..afff475642c 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_transport.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c
@@ -695,7 +695,8 @@ static struct svc_xprt *svc_rdma_create(struct svc_serv *serv,
 		return ERR_PTR(-ENOMEM);
 	xprt = &cma_xprt->sc_xprt;
 
-	listen_id = rdma_create_id(rdma_listen_handler, cma_xprt, RDMA_PS_TCP);
+	listen_id = rdma_create_id(rdma_listen_handler, cma_xprt, RDMA_PS_TCP,
+				   IB_QPT_RC);
 	if (IS_ERR(listen_id)) {
 		ret = PTR_ERR(listen_id);
 		dprintk("svcrdma: rdma_create_id failed = %d\n", ret);
diff --git a/net/sunrpc/xprtrdma/verbs.c b/net/sunrpc/xprtrdma/verbs.c
index d4297dc43dc..80f8da344df 100644
--- a/net/sunrpc/xprtrdma/verbs.c
+++ b/net/sunrpc/xprtrdma/verbs.c
@@ -387,7 +387,7 @@ rpcrdma_create_id(struct rpcrdma_xprt *xprt,
 
 	init_completion(&ia->ri_done);
 
-	id = rdma_create_id(rpcrdma_conn_upcall, xprt, RDMA_PS_TCP);
+	id = rdma_create_id(rpcrdma_conn_upcall, xprt, RDMA_PS_TCP, IB_QPT_RC);
 	if (IS_ERR(id)) {
 		rc = PTR_ERR(id);
 		dprintk("RPC:       %s: rdma_create_id() failed %i\n",
-- 
cgit v1.2.3-70-g09d2


From 753f618ae03a699cc30044d64eea84165016dc6e Mon Sep 17 00:00:00 2001
From: Nir Muchtar <nirm@voltaire.com>
Date: Mon, 3 Jan 2011 15:33:53 +0000
Subject: RDMA/cma: Add support for netlink statistics export

Add callbacks and data types for statistics export of all current
devices/ids.  The schema for RDMA CM is a series of netlink messages.
Each one contains an rdma_cm_stat struct.  Additionally, two netlink
attributes are created for the addresses for each message (if
applicable).

Their types used are:
RDMA_NL_RDMA_CM_ATTR_SRC_ADDR (The source address for this ID)
RDMA_NL_RDMA_CM_ATTR_DST_ADDR (The destination address for this ID)
sockaddr_* structs are encapsulated within these attributes.

In other words, every transaction contains a series of messages like:

-------message 1-------
struct rdma_cm_id_stats {
       __u32 qp_num;
       __u32 bound_dev_if;
       __u32 port_space;
       __s32 pid;
       __u8 cm_state;
       __u8 node_type;
       __u8 port_num;
       __u8 reserved;
}
RDMA_NL_RDMA_CM_ATTR_SRC_ADDR attribute - contains the source address
RDMA_NL_RDMA_CM_ATTR_DST_ADDR attribute - contains the destination address
-------end 1-------
-------message 2-------
struct rdma_cm_id_stats
RDMA_NL_RDMA_CM_ATTR_SRC_ADDR attribute
RDMA_NL_RDMA_CM_ATTR_DST_ADDR attribute
-------end 2-------

Signed-off-by: Nir Muchtar <nirm@voltaire.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
---
 drivers/infiniband/core/cma.c | 99 +++++++++++++++++++++++++++++++++++++++++++
 include/rdma/rdma_netlink.h   | 28 ++++++++++++
 2 files changed, 127 insertions(+)

(limited to 'include')

diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index 44be1c9ed05..5dd9acb92c3 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -47,6 +47,7 @@
 
 #include <rdma/rdma_cm.h>
 #include <rdma/rdma_cm_ib.h>
+#include <rdma/rdma_netlink.h>
 #include <rdma/ib_cache.h>
 #include <rdma/ib_cm.h>
 #include <rdma/ib_sa.h>
@@ -3315,6 +3316,99 @@ static void cma_remove_one(struct ib_device *device)
 	kfree(cma_dev);
 }
 
+static int cma_get_id_stats(struct sk_buff *skb, struct netlink_callback *cb)
+{
+	struct nlmsghdr *nlh;
+	struct rdma_cm_id_stats *id_stats;
+	struct rdma_id_private *id_priv;
+	struct rdma_cm_id *id = NULL;
+	struct cma_device *cma_dev;
+	int i_dev = 0, i_id = 0;
+
+	/*
+	 * We export all of the IDs as a sequence of messages.  Each
+	 * ID gets its own netlink message.
+	 */
+	mutex_lock(&lock);
+
+	list_for_each_entry(cma_dev, &dev_list, list) {
+		if (i_dev < cb->args[0]) {
+			i_dev++;
+			continue;
+		}
+
+		i_id = 0;
+		list_for_each_entry(id_priv, &cma_dev->id_list, list) {
+			if (i_id < cb->args[1]) {
+				i_id++;
+				continue;
+			}
+
+			id_stats = ibnl_put_msg(skb, &nlh, cb->nlh->nlmsg_seq,
+						sizeof *id_stats, RDMA_NL_RDMA_CM,
+						RDMA_NL_RDMA_CM_ID_STATS);
+			if (!id_stats)
+				goto out;
+
+			memset(id_stats, 0, sizeof *id_stats);
+			id = &id_priv->id;
+			id_stats->node_type = id->route.addr.dev_addr.dev_type;
+			id_stats->port_num = id->port_num;
+			id_stats->bound_dev_if =
+				id->route.addr.dev_addr.bound_dev_if;
+
+			if (id->route.addr.src_addr.ss_family == AF_INET) {
+				if (ibnl_put_attr(skb, nlh,
+						  sizeof(struct sockaddr_in),
+						  &id->route.addr.src_addr,
+						  RDMA_NL_RDMA_CM_ATTR_SRC_ADDR)) {
+					goto out;
+				}
+				if (ibnl_put_attr(skb, nlh,
+						  sizeof(struct sockaddr_in),
+						  &id->route.addr.dst_addr,
+						  RDMA_NL_RDMA_CM_ATTR_DST_ADDR)) {
+					goto out;
+				}
+			} else if (id->route.addr.src_addr.ss_family == AF_INET6) {
+				if (ibnl_put_attr(skb, nlh,
+						  sizeof(struct sockaddr_in6),
+						  &id->route.addr.src_addr,
+						  RDMA_NL_RDMA_CM_ATTR_SRC_ADDR)) {
+					goto out;
+				}
+				if (ibnl_put_attr(skb, nlh,
+						  sizeof(struct sockaddr_in6),
+						  &id->route.addr.dst_addr,
+						  RDMA_NL_RDMA_CM_ATTR_DST_ADDR)) {
+					goto out;
+				}
+			}
+
+			id_stats->port_space	= id->ps;
+			id_stats->cm_state	= id_priv->state;
+			id_stats->qp_num	= id_priv->qp_num;
+			id_stats->qp_type	= id->qp_type;
+
+			i_id++;
+		}
+
+		cb->args[1] = 0;
+		i_dev++;
+	}
+
+out:
+	mutex_unlock(&lock);
+	cb->args[0] = i_dev;
+	cb->args[1] = i_id;
+
+	return skb->len;
+}
+
+static const struct ibnl_client_cbs cma_cb_table[] = {
+	[RDMA_NL_RDMA_CM_ID_STATS] = { .dump = cma_get_id_stats },
+};
+
 static int __init cma_init(void)
 {
 	int ret;
@@ -3330,6 +3424,10 @@ static int __init cma_init(void)
 	ret = ib_register_client(&cma_client);
 	if (ret)
 		goto err;
+
+	if (ibnl_add_client(RDMA_NL_RDMA_CM, RDMA_NL_RDMA_CM_NUM_OPS, cma_cb_table))
+		printk(KERN_WARNING "RDMA CMA: failed to add netlink callback\n");
+
 	return 0;
 
 err:
@@ -3342,6 +3440,7 @@ err:
 
 static void __exit cma_cleanup(void)
 {
+	ibnl_remove_client(RDMA_NL_RDMA_CM);
 	ib_unregister_client(&cma_client);
 	unregister_netdevice_notifier(&cma_nb);
 	rdma_addr_unregister_client(&addr_client);
diff --git a/include/rdma/rdma_netlink.h b/include/rdma/rdma_netlink.h
index c983a193cec..3c5363ab867 100644
--- a/include/rdma/rdma_netlink.h
+++ b/include/rdma/rdma_netlink.h
@@ -1,10 +1,38 @@
 #ifndef _RDMA_NETLINK_H
 #define _RDMA_NETLINK_H
 
+#include <linux/types.h>
+
+enum {
+	RDMA_NL_RDMA_CM = 1
+};
+
 #define RDMA_NL_GET_CLIENT(type) ((type & (((1 << 6) - 1) << 10)) >> 10)
 #define RDMA_NL_GET_OP(type) (type & ((1 << 10) - 1))
 #define RDMA_NL_GET_TYPE(client, op) ((client << 10) + op)
 
+enum {
+	RDMA_NL_RDMA_CM_ID_STATS = 0,
+	RDMA_NL_RDMA_CM_NUM_OPS
+};
+
+enum {
+	RDMA_NL_RDMA_CM_ATTR_SRC_ADDR = 1,
+	RDMA_NL_RDMA_CM_ATTR_DST_ADDR,
+	RDMA_NL_RDMA_CM_NUM_ATTR,
+};
+
+struct rdma_cm_id_stats {
+	__u32	qp_num;
+	__u32	bound_dev_if;
+	__u32	port_space;
+	__s32	pid;
+	__u8	cm_state;
+	__u8	node_type;
+	__u8	port_num;
+	__u8	qp_type;
+};
+
 #ifdef __KERNEL__
 
 #include <linux/netlink.h>
-- 
cgit v1.2.3-70-g09d2