Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband: (46 commits)
  IB/uverbs: Don't serialize with ib_uverbs_idr_mutex
  IB/mthca: Make all device methods truly reentrant
  IB/mthca: Fix memory leak on modify_qp error paths
  IB/uverbs: Factor out common idr code
  IB/uverbs: Don't decrement usecnt on error paths
  IB/uverbs: Release lock on error path
  IB/cm: Use address handle helpers
  IB/sa: Add ib_init_ah_from_path()
  IB: Add ib_init_ah_from_wc()
  IB/ucm: Get rid of duplicate P_Key parameter
  IB/srp: Factor out common request reset code
  IB/srp: Support SRP rev. 10 targets
  [SCSI] srp.h: Add I/O Class values
  IB/fmr: Use device's max_map_map_per_fmr attribute in FMR pool.
  IB/mthca: Fill in max_map_per_fmr device attribute
  IB/ipath: Add client reregister event generation
  IB/mthca: Add client reregister event generation
  IB: Move struct port_info from ipath to <rdma/ib_smi.h>
  IPoIB: Handle client reregister events
  IB: Add client reregister event type
  ...

Documentation/infiniband/ipoib.txt

 IP OVER INFINIBAND
 
   The ib_ipoib driver is an implementation of the IP over InfiniBand
-  protocol as specified by the latest Internet-Drafts issued by the
-  IETF ipoib working group.  It is a "native" implementation in the
-  sense of setting the interface type to ARPHRD_INFINIBAND and the
-  hardware address length to 20 (earlier proprietary implementations
+  protocol as specified by RFC 4391 and 4392, issued by the IETF ipoib
+  working group.  It is a "native" implementation in the sense of
+  setting the interface type to ARPHRD_INFINIBAND and the hardware
+  address length to 20 (earlier proprietary implementations
   masqueraded to the kernel as ethernet interfaces).
 
 Partitions and P_Keys
@@ -53,3 +53,7 @@ References
 
   IETF IP over InfiniBand (ipoib) Working Group
     http://ietf.org/html.charters/ipoib-charter.html
+  Transmission of IP over InfiniBand (IPoIB) (RFC 4391)
+    http://ietf.org/rfc/rfc4391.txt 
+  IP over InfiniBand (IPoIB) Architecture (RFC 4392)
+    http://ietf.org/rfc/rfc4392.txt 

drivers/infiniband/Kconfig

@@ -29,6 +29,11 @@ config INFINIBAND_USER_ACCESS
 	  libibverbs, libibcm and a hardware driver library from
 	  <http://www.openib.org>.
 
+config INFINIBAND_ADDR_TRANS
+	bool
+	depends on INFINIBAND && INET
+	default y
+
 source "drivers/infiniband/hw/mthca/Kconfig"
 source "drivers/infiniband/hw/ipath/Kconfig"
 

drivers/infiniband/core/Makefile

+infiniband-$(CONFIG_INFINIBAND_ADDR_TRANS)	:= ib_addr.o rdma_cm.o
+
 obj-$(CONFIG_INFINIBAND) +=		ib_core.o ib_mad.o ib_sa.o \
-					ib_cm.o
+					ib_cm.o $(infiniband-y)
 obj-$(CONFIG_INFINIBAND_USER_MAD) +=	ib_umad.o
 obj-$(CONFIG_INFINIBAND_USER_ACCESS) +=	ib_uverbs.o ib_ucm.o
 
@@ -12,8 +14,13 @@ ib_sa-y :=			sa_query.o
 
 ib_cm-y :=			cm.o
 
+rdma_cm-y :=			cma.o
+
+ib_addr-y :=			addr.o
+
 ib_umad-y :=			user_mad.o
 
 ib_ucm-y :=			ucm.o
 
-ib_uverbs-y :=			uverbs_main.o uverbs_cmd.o uverbs_mem.o
+ib_uverbs-y :=			uverbs_main.o uverbs_cmd.o uverbs_mem.o \
+				uverbs_marshall.o

drivers/infiniband/core/addr.c

+/*
+ * Copyright (c) 2005 Voltaire Inc.  All rights reserved.
+ * Copyright (c) 2002-2005, Network Appliance, Inc. All rights reserved.
+ * Copyright (c) 1999-2005, Mellanox Technologies, Inc. All rights reserved.
+ * Copyright (c) 2005 Intel Corporation.  All rights reserved.
+ *
+ * This Software is licensed under one of the following licenses:
+ *
+ * 1) under the terms of the "Common Public License 1.0" a copy of which is
+ *    available from the Open Source Initiative, see
+ *    http://www.opensource.org/licenses/cpl.php.
+ *
+ * 2) under the terms of the "The BSD License" a copy of which is
+ *    available from the Open Source Initiative, see
+ *    http://www.opensource.org/licenses/bsd-license.php.
+ *
+ * 3) under the terms of the "GNU General Public License (GPL) Version 2" a
+ *    copy of which is available from the Open Source Initiative, see
+ *    http://www.opensource.org/licenses/gpl-license.php.
+ *
+ * Licensee has the right to choose one of the above licenses.
+ *
+ * Redistributions of source code must retain the above copyright
+ * notice and one of the license notices.
+ *
+ * Redistributions in binary form must reproduce both the above copyright
+ * notice, one of the license notices in the documentation
+ * and/or other materials provided with the distribution.
+ */
+
+#include <linux/mutex.h>
+#include <linux/inetdevice.h>
+#include <linux/workqueue.h>
+#include <linux/if_arp.h>
+#include <net/arp.h>
+#include <net/neighbour.h>
+#include <net/route.h>
+#include <rdma/ib_addr.h>
+
+MODULE_AUTHOR("Sean Hefty");
+MODULE_DESCRIPTION("IB Address Translation");
+MODULE_LICENSE("Dual BSD/GPL");
+
+struct addr_req {
+	struct list_head list;
+	struct sockaddr src_addr;
+	struct sockaddr dst_addr;
+	struct rdma_dev_addr *addr;
+	void *context;
+	void (*callback)(int status, struct sockaddr *src_addr,
+			 struct rdma_dev_addr *addr, void *context);
+	unsigned long timeout;
+	int status;
+};
+
+static void process_req(void *data);
+
+static DEFINE_MUTEX(lock);
+static LIST_HEAD(req_list);
+static DECLARE_WORK(work, process_req, NULL);
+static struct workqueue_struct *addr_wq;
+
+static int copy_addr(struct rdma_dev_addr *dev_addr, struct net_device *dev,
+		     unsigned char *dst_dev_addr)
+{
+	switch (dev->type) {
+	case ARPHRD_INFINIBAND:
+		dev_addr->dev_type = IB_NODE_CA;
+		break;
+	default:
+		return -EADDRNOTAVAIL;
+	}
+
+	memcpy(dev_addr->src_dev_addr, dev->dev_addr, MAX_ADDR_LEN);
+	memcpy(dev_addr->broadcast, dev->broadcast, MAX_ADDR_LEN);
+	if (dst_dev_addr)
+		memcpy(dev_addr->dst_dev_addr, dst_dev_addr, MAX_ADDR_LEN);
+	return 0;
+}
+
+int rdma_translate_ip(struct sockaddr *addr, struct rdma_dev_addr *dev_addr)
+{
+	struct net_device *dev;
+	u32 ip = ((struct sockaddr_in *) addr)->sin_addr.s_addr;
+	int ret;
+
+	dev = ip_dev_find(ip);
+	if (!dev)
+		return -EADDRNOTAVAIL;
+
+	ret = copy_addr(dev_addr, dev, NULL);
+	dev_put(dev);
+	return ret;
+}
+EXPORT_SYMBOL(rdma_translate_ip);
+
+static void set_timeout(unsigned long time)
+{
+	unsigned long delay;
+
+	cancel_delayed_work(&work);
+
+	delay = time - jiffies;
+	if ((long)delay <= 0)
+		delay = 1;
+
+	queue_delayed_work(addr_wq, &work, delay);
+}
+
+static void queue_req(struct addr_req *req)
+{
+	struct addr_req *temp_req;
+
+	mutex_lock(&lock);
+	list_for_each_entry_reverse(temp_req, &req_list, list) {
+		if (time_after(req->timeout, temp_req->timeout))
+			break;
+	}
+
+	list_add(&req->list, &temp_req->list);
+
+	if (req_list.next == &req->list)
+		set_timeout(req->timeout);
+	mutex_unlock(&lock);
+}
+
+static void addr_send_arp(struct sockaddr_in *dst_in)
+{
+	struct rtable *rt;
+	struct flowi fl;
+	u32 dst_ip = dst_in->sin_addr.s_addr;
+
+	memset(&fl, 0, sizeof fl);
+	fl.nl_u.ip4_u.daddr = dst_ip;
+	if (ip_route_output_key(&rt, &fl))
+		return;
+
+	arp_send(ARPOP_REQUEST, ETH_P_ARP, rt->rt_gateway, rt->idev->dev,
+		 rt->rt_src, NULL, rt->idev->dev->dev_addr, NULL);
+	ip_rt_put(rt);
+}
+
+static int addr_resolve_remote(struct sockaddr_in *src_in,
+			       struct sockaddr_in *dst_in,
+			       struct rdma_dev_addr *addr)
+{
+	u32 src_ip = src_in->sin_addr.s_addr;
+	u32 dst_ip = dst_in->sin_addr.s_addr;
+	struct flowi fl;
+	struct rtable *rt;
+	struct neighbour *neigh;
+	int ret;
+
+	memset(&fl, 0, sizeof fl);
+	fl.nl_u.ip4_u.daddr = dst_ip;
+	fl.nl_u.ip4_u.saddr = src_ip;
+	ret = ip_route_output_key(&rt, &fl);
+	if (ret)
+		goto out;
+
+	/* If the device does ARP internally, return 'done' */
+	if (rt->idev->dev->flags & IFF_NOARP) {
+		copy_addr(addr, rt->idev->dev, NULL);
+		goto put;
+	}
+
+	neigh = neigh_lookup(&arp_tbl, &rt->rt_gateway, rt->idev->dev);
+	if (!neigh) {
+		ret = -ENODATA;
+		goto put;
+	}
+
+	if (!(neigh->nud_state & NUD_VALID)) {
+		ret = -ENODATA;
+		goto release;
+	}
+
+	if (!src_ip) {
+		src_in->sin_family = dst_in->sin_family;
+		src_in->sin_addr.s_addr = rt->rt_src;
+	}
+
+	ret = copy_addr(addr, neigh->dev, neigh->ha);
+release:
+	neigh_release(neigh);
+put:
+	ip_rt_put(rt);
+out:
+	return ret;
+}
+
+static void process_req(void *data)
+{
+	struct addr_req *req, *temp_req;
+	struct sockaddr_in *src_in, *dst_in;
+	struct list_head done_list;
+
+	INIT_LIST_HEAD(&done_list);
+
+	mutex_lock(&lock);
+	list_for_each_entry_safe(req, temp_req, &req_list, list) {
+		if (req->status) {
+			src_in = (struct sockaddr_in *) &req->src_addr;
+			dst_in = (struct sockaddr_in *) &req->dst_addr;
+			req->status = addr_resolve_remote(src_in, dst_in,
+							  req->addr);
+		}
+		if (req->status && time_after(jiffies, req->timeout))
+			req->status = -ETIMEDOUT;
+		else if (req->status == -ENODATA)
+			continue;
+
+		list_del(&req->list);
+		list_add_tail(&req->list, &done_list);
+	}
+
+	if (!list_empty(&req_list)) {
+		req = list_entry(req_list.next, struct addr_req, list);
+		set_timeout(req->timeout);
+	}
+	mutex_unlock(&lock);
+
+	list_for_each_entry_safe(req, temp_req, &done_list, list) {
+		list_del(&req->list);
+		req->callback(req->status, &req->src_addr, req->addr,
+			      req->context);
+		kfree(req);
+	}
+}
+
+static int addr_resolve_local(struct sockaddr_in *src_in,
+			      struct sockaddr_in *dst_in,
+			      struct rdma_dev_addr *addr)
+{
+	struct net_device *dev;
+	u32 src_ip = src_in->sin_addr.s_addr;
+	u32 dst_ip = dst_in->sin_addr.s_addr;
+	int ret;
+
+	dev = ip_dev_find(dst_ip);
+	if (!dev)
+		return -EADDRNOTAVAIL;
+
+	if (ZERONET(src_ip)) {
+		src_in->sin_family = dst_in->sin_family;
+		src_in->sin_addr.s_addr = dst_ip;
+		ret = copy_addr(addr, dev, dev->dev_addr);
+	} else if (LOOPBACK(src_ip)) {
+		ret = rdma_translate_ip((struct sockaddr *)dst_in, addr);
+		if (!ret)
+			memcpy(addr->dst_dev_addr, dev->dev_addr, MAX_ADDR_LEN);
+	} else {
+		ret = rdma_translate_ip((struct sockaddr *)src_in, addr);
+		if (!ret)
+			memcpy(addr->dst_dev_addr, dev->dev_addr, MAX_ADDR_LEN);
+	}
+
+	dev_put(dev);
+	return ret;
+}
+
+int rdma_resolve_ip(struct sockaddr *src_addr, struct sockaddr *dst_addr,
+		    struct rdma_dev_addr *addr, int timeout_ms,
+		    void (*callback)(int status, struct sockaddr *src_addr,
+				     struct rdma_dev_addr *addr, void *context),
+		    void *context)
+{
+	struct sockaddr_in *src_in, *dst_in;
+	struct addr_req *req;
+	int ret = 0;
+
+	req = kmalloc(sizeof *req, GFP_KERNEL);
+	if (!req)
+		return -ENOMEM;
+	memset(req, 0, sizeof *req);
+
+	if (src_addr)
+		memcpy(&req->src_addr, src_addr, ip_addr_size(src_addr));
+	memcpy(&req->dst_addr, dst_addr, ip_addr_size(dst_addr));
+	req->addr = addr;
+	req->callback = callback;
+	req->context = context;
+
+	src_in = (struct sockaddr_in *) &req->src_addr;
+	dst_in = (struct sockaddr_in *) &req->dst_addr;
+
+	req->status = addr_resolve_local(src_in, dst_in, addr);
+	if (req->status == -EADDRNOTAVAIL)
+		req->status = addr_resolve_remote(src_in, dst_in, addr);
+
+	switch (req->status) {
+	case 0:
+		req->timeout = jiffies;
+		queue_req(req);
+		break;
+	case -ENODATA:
+		req->timeout = msecs_to_jiffies(timeout_ms) + jiffies;
+		queue_req(req);
+		addr_send_arp(dst_in);
+		break;
+	default:
+		ret = req->status;
+		kfree(req);
+		break;
+	}
+	return ret;
+}
+EXPORT_SYMBOL(rdma_resolve_ip);
+
+void rdma_addr_cancel(struct rdma_dev_addr *addr)
+{
+	struct addr_req *req, *temp_req;
+
+	mutex_lock(&lock);
+	list_for_each_entry_safe(req, temp_req, &req_list, list) {
+		if (req->addr == addr) {
+			req->status = -ECANCELED;
+			req->timeout = jiffies;
+			list_del(&req->list);
+			list_add(&req->list, &req_list);
+			set_timeout(req->timeout);
+			break;
+		}
+	}
+	mutex_unlock(&lock);
+}
+EXPORT_SYMBOL(rdma_addr_cancel);
+
+static int addr_arp_recv(struct sk_buff *skb, struct net_device *dev,
+			 struct packet_type *pkt, struct net_device *orig_dev)
+{
+	struct arphdr *arp_hdr;
+
+	arp_hdr = (struct arphdr *) skb->nh.raw;
+
+	if (arp_hdr->ar_op == htons(ARPOP_REQUEST) ||
+	    arp_hdr->ar_op == htons(ARPOP_REPLY))
+		set_timeout(jiffies);
+
+	kfree_skb(skb);
+	return 0;
+}
+
+static struct packet_type addr_arp = {
+	.type           = __constant_htons(ETH_P_ARP),
+	.func           = addr_arp_recv,
+	.af_packet_priv = (void*) 1,
+};
+
+static int addr_init(void)
+{
+	addr_wq = create_singlethread_workqueue("ib_addr_wq");
+	if (!addr_wq)
+		return -ENOMEM;
+
+	dev_add_pack(&addr_arp);
+	return 0;
+}
+
+static void addr_cleanup(void)
+{
+	dev_remove_pack(&addr_arp);
+	destroy_workqueue(addr_wq);
+}
+
+module_init(addr_init);
+module_exit(addr_cleanup);

drivers/infiniband/core/cache.c

@@ -191,6 +191,24 @@ int ib_find_cached_pkey(struct ib_device *device,
 }
 EXPORT_SYMBOL(ib_find_cached_pkey);
 
+int ib_get_cached_lmc(struct ib_device *device,
+		      u8                port_num,
+		      u8                *lmc)
+{
+	unsigned long flags;
+	int ret = 0;
+
+	if (port_num < start_port(device) || port_num > end_port(device))
+		return -EINVAL;
+
+	read_lock_irqsave(&device->cache.lock, flags);
+	*lmc = device->cache.lmc_cache[port_num - start_port(device)];
+	read_unlock_irqrestore(&device->cache.lock, flags);
+
+	return ret;
+}
+EXPORT_SYMBOL(ib_get_cached_lmc);
+
 static void ib_cache_update(struct ib_device *device,
 			    u8                port)
 {
@@ -251,6 +269,8 @@ static void ib_cache_update(struct ib_device *device,
 	device->cache.pkey_cache[port - start_port(device)] = pkey_cache;
 	device->cache.gid_cache [port - start_port(device)] = gid_cache;
 
+	device->cache.lmc_cache[port - start_port(device)] = tprops->lmc;
+
 	write_unlock_irq(&device->cache.lock);
 
 	kfree(old_pkey_cache);
@@ -305,7 +325,13 @@ static void ib_cache_setup_one(struct ib_device *device)
 		kmalloc(sizeof *device->cache.gid_cache *
 			(end_port(device) - start_port(device) + 1), GFP_KERNEL);
 
-	if (!device->cache.pkey_cache || !device->cache.gid_cache) {
+	device->cache.lmc_cache = kmalloc(sizeof *device->cache.lmc_cache *
+					  (end_port(device) -
+					   start_port(device) + 1),
+					  GFP_KERNEL);
+
+	if (!device->cache.pkey_cache || !device->cache.gid_cache ||
+	    !device->cache.lmc_cache) {
 		printk(KERN_WARNING "Couldn't allocate cache "
 		       "for %s\n", device->name);
 		goto err;
@@ -333,6 +359,7 @@ static void ib_cache_setup_one(struct ib_device *device)
 err:
 	kfree(device->cache.pkey_cache);
 	kfree(device->cache.gid_cache);
+	kfree(device->cache.lmc_cache);
 }
 
 static void ib_cache_cleanup_one(struct ib_device *device)
@@ -349,6 +376,7 @@ static void ib_cache_cleanup_one(struct ib_device *device)
 
 	kfree(device->cache.pkey_cache);
 	kfree(device->cache.gid_cache);
+	kfree(device->cache.lmc_cache);
 }
 
 static struct ib_client cache_client = {

drivers/infiniband/core/cm.c

@@ -32,7 +32,7 @@
  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  *
- * $Id: cm.c 2821 2005-07-08 17:07:28Z sean.hefty $
+ * $Id: cm.c 4311 2005-12-05 18:42:01Z sean.hefty $
  */
 
 #include <linux/completion.h>
@@ -132,6 +132,7 @@ struct cm_id_private {
 	/* todo: use alternate port on send failure */
 	struct cm_av av;
 	struct cm_av alt_av;
+	struct ib_cm_compare_data *compare_data;
 
 	void *private_data;
 	__be64 tid;
@@ -253,23 +254,13 @@ static void cm_set_private_data(struct cm_id_private *cm_id_priv,
 	cm_id_priv->private_data_len = private_data_len;
 }
 
-static void cm_set_ah_attr(struct ib_ah_attr *ah_attr, u8 port_num,
-			   u16 dlid, u8 sl, u16 src_path_bits)
-{
-	memset(ah_attr, 0, sizeof ah_attr);
-	ah_attr->dlid = dlid;
-	ah_attr->sl = sl;
-	ah_attr->src_path_bits = src_path_bits;
-	ah_attr->port_num = port_num;
-}
-
-static void cm_init_av_for_response(struct cm_port *port,
-				    struct ib_wc *wc, struct cm_av *av)
+static void cm_init_av_for_response(struct cm_port *port, struct ib_wc *wc,
+				    struct ib_grh *grh, struct cm_av *av)
 {
 	av->port = port;
 	av->pkey_index = wc->pkey_index;
-	cm_set_ah_attr(&av->ah_attr, port->port_num, wc->slid,
-		       wc->sl, wc->dlid_path_bits);
+	ib_init_ah_from_wc(port->cm_dev->device, port->port_num, wc,
+			   grh, &av->ah_attr);
 }
 
 static int cm_init_av_by_path(struct ib_sa_path_rec *path, struct cm_av *av)
@@ -299,9 +290,8 @@ static int cm_init_av_by_path(struct ib_sa_path_rec *path, struct cm_av *av)
 		return ret;
 
 	av->port = port;
-	cm_set_ah_attr(&av->ah_attr, av->port->port_num,
-		       be16_to_cpu(path->dlid), path->sl,
-		       be16_to_cpu(path->slid) & 0x7F);
+	ib_init_ah_from_path(cm_dev->device, port->port_num, path,
+			     &av->ah_attr);
 	av->packet_life_time = path->packet_life_time;
 	return 0;
 }
@@ -357,6 +347,41 @@ static struct cm_id_private * cm_acquire_id(__be32 local_id, __be32 remote_id)
 	return cm_id_priv;
 }
 
+static void cm_mask_copy(u8 *dst, u8 *src, u8 *mask)
+{
+	int i;
+
+	for (i = 0; i < IB_CM_COMPARE_SIZE / sizeof(unsigned long); i++)
+		((unsigned long *) dst)[i] = ((unsigned long *) src)[i] &
+					     ((unsigned long *) mask)[i];
+}
+
+static int cm_compare_data(struct ib_cm_compare_data *src_data,
+			   struct ib_cm_compare_data *dst_data)
+{
+	u8 src[IB_CM_COMPARE_SIZE];
+	u8 dst[IB_CM_COMPARE_SIZE];
+
+	if (!src_data || !dst_data)
+		return 0;
+
+	cm_mask_copy(src, src_data->data, dst_data->mask);
+	cm_mask_copy(dst, dst_data->data, src_data->mask);
+	return memcmp(src, dst, IB_CM_COMPARE_SIZE);
+}
+
+static int cm_compare_private_data(u8 *private_data,
+				   struct ib_cm_compare_data *dst_data)
+{
+	u8 src[IB_CM_COMPARE_SIZE];
+
+	if (!dst_data)
+		return 0;
+
+	cm_mask_copy(src, private_data, dst_data->mask);
+	return memcmp(src, dst_data->data, IB_CM_COMPARE_SIZE);
+}
+
 static struct cm_id_private * cm_insert_listen(struct cm_id_private *cm_id_priv)
 {
 	struct rb_node **link = &cm.listen_service_table.rb_node;
@@ -364,14 +389,18 @@ static struct cm_id_private * cm_insert_listen(struct cm_id_private *cm_id_priv)
 	struct cm_id_private *cur_cm_id_priv;
 	__be64 service_id = cm_id_priv->id.service_id;
 	__be64 service_mask = cm_id_priv->id.service_mask;
+	int data_cmp;
 
 	while (*link) {
 		parent = *link;
 		cur_cm_id_priv = rb_entry(parent, struct cm_id_private,
 					  service_node);
+		data_cmp = cm_compare_data(cm_id_priv->compare_data,
+					   cur_cm_id_priv->compare_data);
 		if ((cur_cm_id_priv->id.service_mask & service_id) ==
 		    (service_mask & cur_cm_id_priv->id.service_id) &&
-		    (cm_id_priv->id.device == cur_cm_id_priv->id.device))
+		    (cm_id_priv->id.device == cur_cm_id_priv->id.device) &&
+		    !data_cmp)
 			return cur_cm_id_priv;
 
 		if (cm_id_priv->id.device < cur_cm_id_priv->id.device)
@@ -380,6 +409,10 @@ static struct cm_id_private * cm_insert_listen(struct cm_id_private *cm_id_priv)
 			link = &(*link)->rb_right;
 		else if (service_id < cur_cm_id_priv->id.service_id)
 			link = &(*link)->rb_left;
+		else if (service_id > cur_cm_id_priv->id.service_id)
+			link = &(*link)->rb_right;
+		else if (data_cmp < 0)
+			link = &(*link)->rb_left;
 		else
 			link = &(*link)->rb_right;
 	}
@@ -389,16 +422,20 @@ static struct cm_id_private * cm_insert_listen(struct cm_id_private *cm_id_priv)
 }
 
 static struct cm_id_private * cm_find_listen(struct ib_device *device,
-					     __be64 service_id)
+					     __be64 service_id,
+					     u8 *private_data)
 {
 	struct rb_node *node = cm.listen_service_table.rb_node;
 	struct cm_id_private *cm_id_priv;
+	int data_cmp;
 
 	while (node) {
 		cm_id_priv = rb_entry(node, struct cm_id_private, service_node);
+		data_cmp = cm_compare_private_data(private_data,
+						   cm_id_priv->compare_data);
 		if ((cm_id_priv->id.service_mask & service_id) ==
 		     cm_id_priv->id.service_id &&
-		    (cm_id_priv->id.device == device))
+		    (cm_id_priv->id.device == device) && !data_cmp)
 			return cm_id_priv;
 
 		if (device < cm_id_priv->id.device)
@@ -407,6 +444,10 @@ static struct cm_id_private * cm_find_listen(struct ib_device *device,
 			node = node->rb_right;
 		else if (service_id < cm_id_priv->id.service_id)
 			node = node->rb_left;
+		else if (service_id > cm_id_priv->id.service_id)
+			node = node->rb_right;
+		else if (data_cmp < 0)
+			node = node->rb_left;
 		else
 			node = node->rb_right;
 	}
@@ -730,15 +771,14 @@ void ib_destroy_cm_id(struct ib_cm_id *cm_id)
 	wait_for_completion(&cm_id_priv->comp);
 	while ((work = cm_dequeue_work(cm_id_priv)) != NULL)
 		cm_free_work(work);
-	if (cm_id_priv->private_data && cm_id_priv->private_data_len)
-		kfree(cm_id_priv->private_data);
+	kfree(cm_id_priv->compare_data);
+	kfree(cm_id_priv->private_data);
 	kfree(cm_id_priv);
 }
 EXPORT_SYMBOL(ib_destroy_cm_id);
 
-int ib_cm_listen(struct ib_cm_id *cm_id,
-		 __be64 service_id,
-		 __be64 service_mask)
+int ib_cm_listen(struct ib_cm_id *cm_id, __be64 service_id, __be64 service_mask,
+		 struct ib_cm_compare_data *compare_data)
 {
 	struct cm_id_private *cm_id_priv, *cur_cm_id_priv;
 	unsigned long flags;
@@ -752,7 +792,19 @@ int ib_cm_listen(struct ib_cm_id *cm_id,
 		return -EINVAL;
 
 	cm_id_priv = container_of(cm_id, struct cm_id_private, id);
-	BUG_ON(cm_id->state != IB_CM_IDLE);
+	if (cm_id->state != IB_CM_IDLE)
+		return -EINVAL;
+
+	if (compare_data) {
+		cm_id_priv->compare_data = kzalloc(sizeof *compare_data,
+						   GFP_KERNEL);
+		if (!cm_id_priv->compare_data)
+			return -ENOMEM;
+		cm_mask_copy(cm_id_priv->compare_data->data,
+			     compare_data->data, compare_data->mask);
+		memcpy(cm_id_priv->compare_data->mask, compare_data->mask,
+		       IB_CM_COMPARE_SIZE);
+	}
 
 	cm_id->state = IB_CM_LISTEN;
 
@@ -769,6 +821,8 @@ int ib_cm_listen(struct ib_cm_id *cm_id,
 
 	if (cur_cm_id_priv) {
 		cm_id->state = IB_CM_IDLE;
+		kfree(cm_id_priv->compare_data);
+		cm_id_priv->compare_data = NULL;
 		ret = -EBUSY;
 	}
 	return ret;
@@ -1241,7 +1295,8 @@ static struct cm_id_private * cm_match_req(struct cm_work *work,
 
 	/* Find matching listen request. */
 	listen_cm_id_priv = cm_find_listen(cm_id_priv->id.device,
-					   req_msg->service_id);
+					   req_msg->service_id,
+					   req_msg->private_data);
 	if (!listen_cm_id_priv) {
 		spin_unlock_irqrestore(&cm.lock, flags);
 		cm_issue_rej(work->port, work->mad_recv_wc,
@@ -1276,6 +1331,7 @@ static int cm_req_handler(struct cm_work *work)
 	cm_id_priv = container_of(cm_id, struct cm_id_private, id);
 	cm_id_priv->id.remote_id = req_msg->local_comm_id;
 	cm_init_av_for_response(work->port, work->mad_recv_wc->wc,
+				work->mad_recv_wc->recv_buf.grh,
 				&cm_id_priv->av);
 	cm_id_priv->timewait_info = cm_create_timewait_info(cm_id_priv->
 							    id.local_id);
@@ -2549,7 +2605,7 @@ static void cm_format_sidr_req(struct cm_sidr_req_msg *sidr_req_msg,
 	cm_format_mad_hdr(&sidr_req_msg->hdr, CM_SIDR_REQ_ATTR_ID,
 			  cm_form_tid(cm_id_priv, CM_MSG_SEQUENCE_SIDR));
 	sidr_req_msg->request_id = cm_id_priv->id.local_id;
-	sidr_req_msg->pkey = cpu_to_be16(param->pkey);
+	sidr_req_msg->pkey = cpu_to_be16(param->path->pkey);
 	sidr_req_msg->service_id = param->service_id;
 
 	if (param->private_data && param->private_data_len)
@@ -2641,6 +2697,7 @@ static int cm_sidr_req_handler(struct cm_work *work)
 	cm_id_priv->av.dgid.global.subnet_prefix = cpu_to_be64(wc->slid);
 	cm_id_priv->av.dgid.global.interface_id = 0;
 	cm_init_av_for_response(work->port, work->mad_recv_wc->wc,
+				work->mad_recv_wc->recv_buf.grh,
 				&cm_id_priv->av);
 	cm_id_priv->id.remote_id = sidr_req_msg->request_id;
 	cm_id_priv->id.state = IB_CM_SIDR_REQ_RCVD;
@@ -2654,7 +2711,8 @@ static int cm_sidr_req_handler(struct cm_work *work)
 		goto out; /* Duplicate message. */
 	}
 	cur_cm_id_priv = cm_find_listen(cm_id->device,
-					sidr_req_msg->service_id);
+					sidr_req_msg->service_id,
+					sidr_req_msg->private_data);
 	if (!cur_cm_id_priv) {
 		rb_erase(&cm_id_priv->sidr_id_node, &cm.remote_sidr_table);
 		spin_unlock_irqrestore(&cm.lock, flags);
@@ -3291,7 +3349,6 @@ static int __init ib_cm_init(void)
 
 static void __exit ib_cm_cleanup(void)
 {
-	flush_workqueue(cm.wq);
 	destroy_workqueue(cm.wq);
 	ib_unregister_client(&cm_client);
 	idr_destroy(&cm.local_id_table);

drivers/infiniband/core/fmr_pool.c

@@ -54,7 +54,7 @@ enum {
 /*
  * If an FMR is not in use, then the list member will point to either
  * its pool's free_list (if the FMR can be mapped again; that is,
- * remap_count < IB_FMR_MAX_REMAPS) or its pool's dirty_list (if the
+ * remap_count < pool->max_remaps) or its pool's dirty_list (if the
  * FMR needs to be unmapped before being remapped).  In either of
  * these cases it is a bug if the ref_count is not 0.  In other words,
  * if ref_count is > 0, then the list member must not be linked into
@@ -84,6 +84,7 @@ struct ib_fmr_pool {
 
 	int                       pool_size;
 	int                       max_pages;
+	int			  max_remaps;
 	int                       dirty_watermark;
 	int                       dirty_len;
 	struct list_head          free_list;
@@ -214,8 +215,10 @@ struct ib_fmr_pool *ib_create_fmr_pool(struct ib_pd             *pd,
 {
 	struct ib_device   *device;
 	struct ib_fmr_pool *pool;
+	struct ib_device_attr *attr;
 	int i;
 	int ret;
+	int max_remaps;
 
 	if (!params)
 		return ERR_PTR(-EINVAL);
@@ -228,6 +231,26 @@ struct ib_fmr_pool *ib_create_fmr_pool(struct ib_pd             *pd,
 		return ERR_PTR(-ENOSYS);
 	}
 
+	attr = kmalloc(sizeof *attr, GFP_KERNEL);
+	if (!attr) {
+		printk(KERN_WARNING "couldn't allocate device attr struct");
+		return ERR_PTR(-ENOMEM);
+	}
+
+	ret = ib_query_device(device, attr);
+	if (ret) {
+		printk(KERN_WARNING "couldn't query device");
+		kfree(attr);
+		return ERR_PTR(ret);
+	}
+
+	if (!attr->max_map_per_fmr)
+		max_remaps = IB_FMR_MAX_REMAPS;
+	else
+		max_remaps = attr->max_map_per_fmr;
+
+	kfree(attr);
+
 	pool = kmalloc(sizeof *pool, GFP_KERNEL);
 	if (!pool) {
 		printk(KERN_WARNING "couldn't allocate pool struct");
@@ -258,6 +281,7 @@ struct ib_fmr_pool *ib_create_fmr_pool(struct ib_pd             *pd,
 
 	pool->pool_size       = 0;
 	pool->max_pages       = params->max_pages_per_fmr;
+	pool->max_remaps      = max_remaps;
 	pool->dirty_watermark = params->dirty_watermark;
 	pool->dirty_len       = 0;
 	spin_lock_init(&pool->pool_lock);
@@ -279,7 +303,7 @@ struct ib_fmr_pool *ib_create_fmr_pool(struct ib_pd             *pd,
 		struct ib_pool_fmr *fmr;
 		struct ib_fmr_attr attr = {
 			.max_pages  = params->max_pages_per_fmr,
-			.max_maps   = IB_FMR_MAX_REMAPS,
+			.max_maps   = pool->max_remaps,
 			.page_shift = params->page_shift
 		};
 
@@ -489,7 +513,7 @@ int ib_fmr_pool_unmap(struct ib_pool_fmr *fmr)
 
 	--fmr->ref_count;
 	if (!fmr->ref_count) {
-		if (fmr->remap_count < IB_FMR_MAX_REMAPS) {
+		if (fmr->remap_count < pool->max_remaps) {
 			list_add_tail(&fmr->list, &pool->free_list);
 		} else {
 			list_add_tail(&fmr->list, &pool->dirty_list);

drivers/infiniband/core/mad.c

@@ -34,6 +34,7 @@
  * $Id: mad.c 5596 2006-03-03 01:00:07Z sean.hefty $
  */
 #include <linux/dma-mapping.h>
+#include <rdma/ib_cache.h>
 
 #include "mad_priv.h"
 #include "mad_rmpp.h"
@@ -45,8 +46,7 @@ MODULE_DESCRIPTION("kernel IB MAD API");
 MODULE_AUTHOR("Hal Rosenstock");
 MODULE_AUTHOR("Sean Hefty");
 
-
-kmem_cache_t *ib_mad_cache;
+static kmem_cache_t *ib_mad_cache;
 
 static struct list_head ib_mad_port_list;
 static u32 ib_mad_client_id = 0;
@@ -1673,20 +1673,21 @@ static inline int rcv_has_same_class(struct ib_mad_send_wr_private *wr,
 		rwc->recv_buf.mad->mad_hdr.mgmt_class;
 }
 
-static inline int rcv_has_same_gid(struct ib_mad_send_wr_private *wr,
+static inline int rcv_has_same_gid(struct ib_mad_agent_private *mad_agent_priv,
+				   struct ib_mad_send_wr_private *wr,
 				   struct ib_mad_recv_wc *rwc )
 {
 	struct ib_ah_attr attr;
 	u8 send_resp, rcv_resp;
+	union ib_gid sgid;
+	struct ib_device *device = mad_agent_priv->agent.device;
+	u8 port_num = mad_agent_priv->agent.port_num;
+	u8 lmc;
 
 	send_resp = ((struct ib_mad *)(wr->send_buf.mad))->
 		     mad_hdr.method & IB_MGMT_METHOD_RESP;
 	rcv_resp = rwc->recv_buf.mad->mad_hdr.method & IB_MGMT_METHOD_RESP;
 
-	if (!send_resp && rcv_resp)
-		/* is request/response. GID/LIDs are both local (same). */
-		return 1;
-
 	if (send_resp == rcv_resp)
 		/* both requests, or both responses. GIDs different */
 		return 0;
@@ -1695,48 +1696,78 @@ static inline int rcv_has_same_gid(struct ib_mad_send_wr_private *wr,
 		/* Assume not equal, to avoid false positives. */
 		return 0;
 
-	if (!(attr.ah_flags & IB_AH_GRH) && !(rwc->wc->wc_flags & IB_WC_GRH))
-		return attr.dlid == rwc->wc->slid;
-	else if ((attr.ah_flags & IB_AH_GRH) &&
-		 (rwc->wc->wc_flags & IB_WC_GRH))
-		return memcmp(attr.grh.dgid.raw,
-			      rwc->recv_buf.grh->sgid.raw, 16) == 0;
-	else
+	if (!!(attr.ah_flags & IB_AH_GRH) !=
+	    !!(rwc->wc->wc_flags & IB_WC_GRH))
 		/* one has GID, other does not.  Assume different */
 		return 0;
+
+	if (!send_resp && rcv_resp) {
+		/* is request/response. */
+		if (!(attr.ah_flags & IB_AH_GRH)) {
+			if (ib_get_cached_lmc(device, port_num, &lmc))
+				return 0;
+			return (!lmc || !((attr.src_path_bits ^
+					   rwc->wc->dlid_path_bits) &
+					  ((1 << lmc) - 1)));
+		} else {
+			if (ib_get_cached_gid(device, port_num,
+					      attr.grh.sgid_index, &sgid))
+				return 0;
+			return !memcmp(sgid.raw, rwc->recv_buf.grh->dgid.raw,
+				       16);
+		}
+	}
+
+	if (!(attr.ah_flags & IB_AH_GRH))
+		return attr.dlid == rwc->wc->slid;
+	else
+		return !memcmp(attr.grh.dgid.raw, rwc->recv_buf.grh->sgid.raw,
+			       16);
+}
+
+static inline int is_direct(u8 class)
+{
+	return (class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE);
 }
+
 struct ib_mad_send_wr_private*
 ib_find_send_mad(struct ib_mad_agent_private *mad_agent_priv,
-		 struct ib_mad_recv_wc *mad_recv_wc)
+		 struct ib_mad_recv_wc *wc)
 {
-	struct ib_mad_send_wr_private *mad_send_wr;
+	struct ib_mad_send_wr_private *wr;
 	struct ib_mad *mad;
 
-	mad = (struct ib_mad *)mad_recv_wc->recv_buf.mad;
+	mad = (struct ib_mad *)wc->recv_buf.mad;
 
-	list_for_each_entry(mad_send_wr, &mad_agent_priv->wait_list,
-			    agent_list) {
-		if ((mad_send_wr->tid == mad->mad_hdr.tid) &&
-		    rcv_has_same_class(mad_send_wr, mad_recv_wc) &&
-		    rcv_has_same_gid(mad_send_wr, mad_recv_wc))
-			return mad_send_wr;
+	list_for_each_entry(wr, &mad_agent_priv->wait_list, agent_list) {
+		if ((wr->tid == mad->mad_hdr.tid) &&
+		    rcv_has_same_class(wr, wc) &&
+		    /*
+		     * Don't check GID for direct routed MADs.
+		     * These might have permissive LIDs.
+		     */
+		    (is_direct(wc->recv_buf.mad->mad_hdr.mgmt_class) ||
+		     rcv_has_same_gid(mad_agent_priv, wr, wc)))
+			return wr;
 	}
 
 	/*
 	 * It's possible to receive the response before we've
 	 * been notified that the send has completed
 	 */
-	list_for_each_entry(mad_send_wr, &mad_agent_priv->send_list,
-			    agent_list) {
-		if (is_data_mad(mad_agent_priv, mad_send_wr->send_buf.mad) &&
-		    mad_send_wr->tid == mad->mad_hdr.tid &&
-		    mad_send_wr->timeout &&
-		    rcv_has_same_class(mad_send_wr, mad_recv_wc) &&
-		    rcv_has_same_gid(mad_send_wr, mad_recv_wc)) {
+	list_for_each_entry(wr, &mad_agent_priv->send_list, agent_list) {
+		if (is_data_mad(mad_agent_priv, wr->send_buf.mad) &&
+		    wr->tid == mad->mad_hdr.tid &&
+		    wr->timeout &&
+		    rcv_has_same_class(wr, wc) &&
+		    /*
+		     * Don't check GID for direct routed MADs.
+		     * These might have permissive LIDs.
+		     */
+		    (is_direct(wc->recv_buf.mad->mad_hdr.mgmt_class) ||
+		     rcv_has_same_gid(mad_agent_priv, wr, wc)))
 			/* Verify request has not been canceled */
-			return (mad_send_wr->status == IB_WC_SUCCESS) ?
-				mad_send_wr : NULL;
-		}
+			return (wr->status == IB_WC_SUCCESS) ? wr : NULL;
 	}
 	return NULL;
 }

drivers/infiniband/core/mad_priv.h

@@ -212,8 +212,6 @@ struct ib_mad_port_private {
 	struct ib_mad_qp_info qp_info[IB_MAD_QPS_CORE];
 };
 
-extern kmem_cache_t *ib_mad_cache;
-
 int ib_send_mad(struct ib_mad_send_wr_private *mad_send_wr);
 
 struct ib_mad_send_wr_private *

drivers/infiniband/core/sa_query.c

@@ -47,6 +47,7 @@
 
 #include <rdma/ib_pack.h>
 #include <rdma/ib_sa.h>
+#include <rdma/ib_cache.h>
 
 MODULE_AUTHOR("Roland Dreier");
 MODULE_DESCRIPTION("InfiniBand subnet administration query support");
@@ -441,6 +442,36 @@ void ib_sa_cancel_query(int id, struct ib_sa_query *query)
 }
 EXPORT_SYMBOL(ib_sa_cancel_query);
 
+int ib_init_ah_from_path(struct ib_device *device, u8 port_num,
+			 struct ib_sa_path_rec *rec, struct ib_ah_attr *ah_attr)
+{
+	int ret;
+	u16 gid_index;
+
+	memset(ah_attr, 0, sizeof *ah_attr);
+	ah_attr->dlid = be16_to_cpu(rec->dlid);
+	ah_attr->sl = rec->sl;
+	ah_attr->src_path_bits = be16_to_cpu(rec->slid) & 0x7f;
+	ah_attr->port_num = port_num;
+
+	if (rec->hop_limit > 1) {
+		ah_attr->ah_flags = IB_AH_GRH;
+		ah_attr->grh.dgid = rec->dgid;
+
+		ret = ib_find_cached_gid(device, &rec->sgid, &port_num,
+					 &gid_index);
+		if (ret)
+			return ret;
+
+		ah_attr->grh.sgid_index    = gid_index;
+		ah_attr->grh.flow_label    = be32_to_cpu(rec->flow_label);
+		ah_attr->grh.hop_limit     = rec->hop_limit;
+		ah_attr->grh.traffic_class = rec->traffic_class;
+	}
+	return 0;
+}
+EXPORT_SYMBOL(ib_init_ah_from_path);
+
 static void init_mad(struct ib_sa_mad *mad, struct ib_mad_agent *agent)
 {
 	unsigned long flags;

drivers/infiniband/core/uverbs.h

@@ -132,7 +132,7 @@ struct ib_ucq_object {
 	u32			async_events_reported;
 };
 
-extern struct mutex ib_uverbs_idr_mutex;
+extern spinlock_t ib_uverbs_idr_lock;
 extern struct idr ib_uverbs_pd_idr;
 extern struct idr ib_uverbs_mr_idr;
 extern struct idr ib_uverbs_mw_idr;
@@ -141,6 +141,8 @@ extern struct idr ib_uverbs_cq_idr;
 extern struct idr ib_uverbs_qp_idr;
 extern struct idr ib_uverbs_srq_idr;
 
+void idr_remove_uobj(struct idr *idp, struct ib_uobject *uobj);
+
 struct file *ib_uverbs_alloc_event_file(struct ib_uverbs_file *uverbs_file,
 					int is_async, int *fd);
 void ib_uverbs_release_event_file(struct kref *ref);

drivers/infiniband/hw/mthca/mthca_cmd.c

@@ -174,7 +174,6 @@ enum {
 
 struct mthca_cmd_context {
 	struct completion done;
-	struct timer_list timer;
 	int               result;
 	int               next;
 	u64               out_param;
@@ -362,15 +361,6 @@ void mthca_cmd_event(struct mthca_dev *dev,
 	complete(&context->done);
 }
 
-static void event_timeout(unsigned long context_ptr)
-{
-	struct mthca_cmd_context *context =
-		(struct mthca_cmd_context *) context_ptr;
-
-	context->result = -EBUSY;
-	complete(&context->done);
-}
-
 static int mthca_cmd_wait(struct mthca_dev *dev,
 			  u64 in_param,
 			  u64 *out_param,
@@ -401,11 +391,10 @@ static int mthca_cmd_wait(struct mthca_dev *dev,
 	if (err)
 		goto out;
 
-	context->timer.expires  = jiffies + timeout;
-	add_timer(&context->timer);
-
-	wait_for_completion(&context->done);
-	del_timer_sync(&context->timer);
+	if (!wait_for_completion_timeout(&context->done, timeout)) {
+		err = -EBUSY;
+		goto out;
+	}
 
 	err = context->result;
 	if (err)
@@ -535,10 +524,6 @@ int mthca_cmd_use_events(struct mthca_dev *dev)
 	for (i = 0; i < dev->cmd.max_cmds; ++i) {
 		dev->cmd.context[i].token = i;
 		dev->cmd.context[i].next = i + 1;
-		init_timer(&dev->cmd.context[i].timer);
-		dev->cmd.context[i].timer.data     =
-			(unsigned long) &dev->cmd.context[i];
-		dev->cmd.context[i].timer.function = event_timeout;
 	}
 
 	dev->cmd.context[dev->cmd.max_cmds - 1].next = -1;

drivers/infiniband/hw/mthca/mthca_eq.c

@@ -695,10 +695,6 @@ static void mthca_unmap_reg(struct mthca_dev *dev, unsigned long offset,
 
 static int __devinit mthca_map_eq_regs(struct mthca_dev *dev)
 {
-	unsigned long mthca_base;
-
-	mthca_base = pci_resource_start(dev->pdev, 0);
-
 	if (mthca_is_memfree(dev)) {
 		/*
 		 * We assume that the EQ arm and EQ set CI registers