Creation of Cybook 2416 (actually Gen4) repository

drivers/infiniband/core/Makefile

@@ -0,0 +1,32 @@
+infiniband-$(CONFIG_INFINIBAND_ADDR_TRANS)	:= ib_addr.o rdma_cm.o
+user_access-$(CONFIG_INFINIBAND_ADDR_TRANS)	:= rdma_ucm.o
+
+obj-$(CONFIG_INFINIBAND) +=		ib_core.o ib_mad.o ib_sa.o \
+					ib_cm.o iw_cm.o $(infiniband-y)
+obj-$(CONFIG_INFINIBAND_USER_MAD) +=	ib_umad.o
+obj-$(CONFIG_INFINIBAND_USER_ACCESS) +=	ib_uverbs.o ib_ucm.o \
+					$(user_access-y)
+
+ib_core-y :=			packer.o ud_header.o verbs.o sysfs.o \
+				device.o fmr_pool.o cache.o
+
+ib_mad-y :=			mad.o smi.o agent.o mad_rmpp.o
+
+ib_sa-y :=			sa_query.o multicast.o
+
+ib_cm-y :=			cm.o
+
+iw_cm-y :=			iwcm.o
+
+rdma_cm-y :=			cma.o
+
+rdma_ucm-y :=			ucma.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 \
+				uverbs_marshall.o

drivers/infiniband/core/addr.c

@@ -0,0 +1,391 @@
+/*
+ * 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 <net/netevent.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;
+	struct rdma_addr_client *client;
+	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(struct work_struct *work);
+
+static DEFINE_MUTEX(lock);
+static LIST_HEAD(req_list);
+static DECLARE_DELAYED_WORK(work, process_req);
+static struct workqueue_struct *addr_wq;
+
+void rdma_addr_register_client(struct rdma_addr_client *client)
+{
+	atomic_set(&client->refcount, 1);
+	init_completion(&client->comp);
+}
+EXPORT_SYMBOL(rdma_addr_register_client);
+
+static inline void put_client(struct rdma_addr_client *client)
+{
+	if (atomic_dec_and_test(&client->refcount))
+		complete(&client->comp);
+}
+
+void rdma_addr_unregister_client(struct rdma_addr_client *client)
+{
+	put_client(client);
+	wait_for_completion(&client->comp);
+}
+EXPORT_SYMBOL(rdma_addr_unregister_client);
+
+int rdma_copy_addr(struct rdma_dev_addr *dev_addr, struct net_device *dev,
+		     const unsigned char *dst_dev_addr)
+{
+	switch (dev->type) {
+	case ARPHRD_INFINIBAND:
+		dev_addr->dev_type = RDMA_NODE_IB_CA;
+		break;
+	case ARPHRD_ETHER:
+		dev_addr->dev_type = RDMA_NODE_RNIC;
+		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;
+}
+EXPORT_SYMBOL(rdma_copy_addr);
+
+int rdma_translate_ip(struct sockaddr *addr, struct rdma_dev_addr *dev_addr)
+{
+	struct net_device *dev;
+	__be32 ip = ((struct sockaddr_in *) addr)->sin_addr.s_addr;
+	int ret;
+
+	dev = ip_dev_find(ip);
+	if (!dev)
+		return -EADDRNOTAVAIL;
+
+	ret = rdma_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_eq(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) {
+		rdma_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 = rdma_copy_addr(addr, neigh->dev, neigh->ha);
+release:
+	neigh_release(neigh);
+put:
+	ip_rt_put(rt);
+out:
+	return ret;
+}
+
+static void process_req(struct work_struct *work)
+{
+	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 == -ENODATA) {
+			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_eq(jiffies, req->timeout))
+				req->status = -ETIMEDOUT;
+			else if (req->status == -ENODATA)
+				continue;
+		}
+		list_move_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);
+		put_client(req->client);
+		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;
+	__be32 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 = rdma_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 rdma_addr_client *client,
+		    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;
+	req->client = client;
+	atomic_inc(&client->refcount);
+
+	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;
+		atomic_dec(&client->refcount);
+		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_move(&req->list, &req_list);
+			set_timeout(req->timeout);
+			break;
+		}
+	}
+	mutex_unlock(&lock);
+}
+EXPORT_SYMBOL(rdma_addr_cancel);
+
+static int netevent_callback(struct notifier_block *self, unsigned long event,
+	void *ctx)
+{
+	if (event == NETEVENT_NEIGH_UPDATE) {
+		struct neighbour *neigh = ctx;
+
+		if (neigh->nud_state & NUD_VALID) {
+			set_timeout(jiffies);
+		}
+	}
+	return 0;
+}
+
+static struct notifier_block nb = {
+	.notifier_call = netevent_callback
+};
+
+static int addr_init(void)
+{
+	addr_wq = create_singlethread_workqueue("ib_addr");
+	if (!addr_wq)
+		return -ENOMEM;
+
+	register_netevent_notifier(&nb);
+	return 0;
+}
+
+static void addr_cleanup(void)
+{
+	unregister_netevent_notifier(&nb);
+	destroy_workqueue(addr_wq);
+}
+
+module_init(addr_init);
+module_exit(addr_cleanup);

drivers/infiniband/core/agent.c

@@ -0,0 +1,202 @@
+/*
+ * Copyright (c) 2004, 2005 Mellanox Technologies Ltd.  All rights reserved.
+ * Copyright (c) 2004, 2005 Infinicon Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Intel Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Topspin Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Voltaire Corporation.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. 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.
+ *
+ * $Id: agent.c,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#include <linux/slab.h>
+#include <linux/string.h>
+
+#include "agent.h"
+#include "smi.h"
+
+#define SPFX "ib_agent: "
+
+struct ib_agent_port_private {
+	struct list_head port_list;
+	struct ib_mad_agent *agent[2];
+};
+
+static DEFINE_SPINLOCK(ib_agent_port_list_lock);
+static LIST_HEAD(ib_agent_port_list);
+
+static struct ib_agent_port_private *
+__ib_get_agent_port(struct ib_device *device, int port_num)
+{
+	struct ib_agent_port_private *entry;
+
+	list_for_each_entry(entry, &ib_agent_port_list, port_list) {
+		if (entry->agent[0]->device == device &&
+		    entry->agent[0]->port_num == port_num)
+			return entry;
+	}
+	return NULL;
+}
+
+static struct ib_agent_port_private *
+ib_get_agent_port(struct ib_device *device, int port_num)
+{
+	struct ib_agent_port_private *entry;
+	unsigned long flags;
+
+	spin_lock_irqsave(&ib_agent_port_list_lock, flags);
+	entry = __ib_get_agent_port(device, port_num);
+	spin_unlock_irqrestore(&ib_agent_port_list_lock, flags);
+	return entry;
+}
+
+int agent_send_response(struct ib_mad *mad, struct ib_grh *grh,
+			struct ib_wc *wc, struct ib_device *device,
+			int port_num, int qpn)
+{
+	struct ib_agent_port_private *port_priv;
+	struct ib_mad_agent *agent;
+	struct ib_mad_send_buf *send_buf;
+	struct ib_ah *ah;
+	int ret;
+
+	port_priv = ib_get_agent_port(device, port_num);
+	if (!port_priv) {
+		printk(KERN_ERR SPFX "Unable to find port agent\n");
+		return -ENODEV;
+	}
+
+	agent = port_priv->agent[qpn];
+	ah = ib_create_ah_from_wc(agent->qp->pd, wc, grh, port_num);
+	if (IS_ERR(ah)) {
+		ret = PTR_ERR(ah);
+		printk(KERN_ERR SPFX "ib_create_ah_from_wc error:%d\n", ret);
+		return ret;
+	}
+
+	send_buf = ib_create_send_mad(agent, wc->src_qp, wc->pkey_index, 0,
+				      IB_MGMT_MAD_HDR, IB_MGMT_MAD_DATA,
+				      GFP_KERNEL);
+	if (IS_ERR(send_buf)) {
+		ret = PTR_ERR(send_buf);
+		printk(KERN_ERR SPFX "ib_create_send_mad error:%d\n", ret);
+		goto err1;
+	}
+
+	memcpy(send_buf->mad, mad, sizeof *mad);
+	send_buf->ah = ah;
+	if ((ret = ib_post_send_mad(send_buf, NULL))) {
+		printk(KERN_ERR SPFX "ib_post_send_mad error:%d\n", ret);
+		goto err2;
+	}
+	return 0;
+err2:
+	ib_free_send_mad(send_buf);
+err1:
+	ib_destroy_ah(ah);
+	return ret;
+}
+
+static void agent_send_handler(struct ib_mad_agent *mad_agent,
+			       struct ib_mad_send_wc *mad_send_wc)
+{
+	ib_destroy_ah(mad_send_wc->send_buf->ah);
+	ib_free_send_mad(mad_send_wc->send_buf);
+}
+
+int ib_agent_port_open(struct ib_device *device, int port_num)
+{
+	struct ib_agent_port_private *port_priv;
+	unsigned long flags;
+	int ret;
+
+	/* Create new device info */
+	port_priv = kzalloc(sizeof *port_priv, GFP_KERNEL);
+	if (!port_priv) {
+		printk(KERN_ERR SPFX "No memory for ib_agent_port_private\n");
+		ret = -ENOMEM;
+		goto error1;
+	}
+
+	/* Obtain send only MAD agent for SMI QP */
+	port_priv->agent[0] = ib_register_mad_agent(device, port_num,
+						    IB_QPT_SMI, NULL, 0,
+						    &agent_send_handler,
+						    NULL, NULL);
+	if (IS_ERR(port_priv->agent[0])) {
+		ret = PTR_ERR(port_priv->agent[0]);
+		goto error2;
+	}
+
+	/* Obtain send only MAD agent for GSI QP */
+	port_priv->agent[1] = ib_register_mad_agent(device, port_num,
+						    IB_QPT_GSI, NULL, 0,
+						    &agent_send_handler,
+						    NULL, NULL);
+	if (IS_ERR(port_priv->agent[1])) {
+		ret = PTR_ERR(port_priv->agent[1]);
+		goto error3;
+	}
+
+	spin_lock_irqsave(&ib_agent_port_list_lock, flags);
+	list_add_tail(&port_priv->port_list, &ib_agent_port_list);
+	spin_unlock_irqrestore(&ib_agent_port_list_lock, flags);
+
+	return 0;
+
+error3:
+	ib_unregister_mad_agent(port_priv->agent[0]);
+error2:
+	kfree(port_priv);
+error1:
+	return ret;
+}
+
+int ib_agent_port_close(struct ib_device *device, int port_num)
+{
+	struct ib_agent_port_private *port_priv;
+	unsigned long flags;
+
+	spin_lock_irqsave(&ib_agent_port_list_lock, flags);
+	port_priv = __ib_get_agent_port(device, port_num);
+	if (port_priv == NULL) {
+		spin_unlock_irqrestore(&ib_agent_port_list_lock, flags);
+		printk(KERN_ERR SPFX "Port %d not found\n", port_num);
+		return -ENODEV;
+	}
+	list_del(&port_priv->port_list);
+	spin_unlock_irqrestore(&ib_agent_port_list_lock, flags);
+
+	ib_unregister_mad_agent(port_priv->agent[1]);
+	ib_unregister_mad_agent(port_priv->agent[0]);
+	kfree(port_priv);
+	return 0;
+}

drivers/infiniband/core/agent.h

@@ -0,0 +1,53 @@
+/*
+ * Copyright (c) 2004 Mellanox Technologies Ltd.  All rights reserved.
+ * Copyright (c) 2004 Infinicon Corporation.  All rights reserved.
+ * Copyright (c) 2004 Intel Corporation.  All rights reserved.
+ * Copyright (c) 2004 Topspin Corporation.  All rights reserved.
+ * Copyright (c) 2004 Voltaire Corporation.  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.
+ *
+ * $Id: agent.h,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#ifndef __AGENT_H_
+#define __AGENT_H_
+
+#include <linux/err.h>
+#include <rdma/ib_mad.h>
+
+extern int ib_agent_port_open(struct ib_device *device, int port_num);
+
+extern int ib_agent_port_close(struct ib_device *device, int port_num);
+
+extern int agent_send_response(struct ib_mad *mad, struct ib_grh *grh,
+			       struct ib_wc *wc, struct ib_device *device,
+			       int port_num, int qpn);
+
+#endif	/* __AGENT_H_ */

drivers/infiniband/core/cache.c

@@ -0,0 +1,398 @@
+/*
+ * Copyright (c) 2004 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Intel Corporation. All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
+ * Copyright (c) 2005 Voltaire, Inc. 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.
+ *
+ * $Id: cache.c,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#include <linux/module.h>
+#include <linux/errno.h>
+#include <linux/slab.h>
+
+#include <rdma/ib_cache.h>
+
+#include "core_priv.h"
+
+struct ib_pkey_cache {
+	int             table_len;
+	u16             table[0];
+};
+
+struct ib_gid_cache {
+	int             table_len;
+	union ib_gid    table[0];
+};
+
+struct ib_update_work {
+	struct work_struct work;
+	struct ib_device  *device;
+	u8                 port_num;
+};
+
+static inline int start_port(struct ib_device *device)
+{
+	return (device->node_type == RDMA_NODE_IB_SWITCH) ? 0 : 1;
+}
+
+static inline int end_port(struct ib_device *device)
+{
+	return (device->node_type == RDMA_NODE_IB_SWITCH) ?
+		0 : device->phys_port_cnt;
+}
+
+int ib_get_cached_gid(struct ib_device *device,
+		      u8                port_num,
+		      int               index,
+		      union ib_gid     *gid)
+{
+	struct ib_gid_cache *cache;
+	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);
+
+	cache = device->cache.gid_cache[port_num - start_port(device)];
+
+	if (index < 0 || index >= cache->table_len)
+		ret = -EINVAL;
+	else
+		*gid = cache->table[index];
+
+	read_unlock_irqrestore(&device->cache.lock, flags);
+
+	return ret;
+}
+EXPORT_SYMBOL(ib_get_cached_gid);
+
+int ib_find_cached_gid(struct ib_device *device,
+		       union ib_gid	*gid,
+		       u8               *port_num,
+		       u16              *index)
+{
+	struct ib_gid_cache *cache;
+	unsigned long flags;
+	int p, i;
+	int ret = -ENOENT;
+
+	*port_num = -1;
+	if (index)
+		*index = -1;
+
+	read_lock_irqsave(&device->cache.lock, flags);
+
+	for (p = 0; p <= end_port(device) - start_port(device); ++p) {
+		cache = device->cache.gid_cache[p];
+		for (i = 0; i < cache->table_len; ++i) {
+			if (!memcmp(gid, &cache->table[i], sizeof *gid)) {
+				*port_num = p + start_port(device);
+				if (index)
+					*index = i;
+				ret = 0;
+				goto found;
+			}
+		}
+	}
+found:
+	read_unlock_irqrestore(&device->cache.lock, flags);
+
+	return ret;
+}
+EXPORT_SYMBOL(ib_find_cached_gid);
+
+int ib_get_cached_pkey(struct ib_device *device,
+		       u8                port_num,
+		       int               index,
+		       u16              *pkey)
+{
+	struct ib_pkey_cache *cache;
+	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);
+
+	cache = device->cache.pkey_cache[port_num - start_port(device)];
+
+	if (index < 0 || index >= cache->table_len)
+		ret = -EINVAL;
+	else
+		*pkey = cache->table[index];
+
+	read_unlock_irqrestore(&device->cache.lock, flags);
+
+	return ret;
+}
+EXPORT_SYMBOL(ib_get_cached_pkey);
+
+int ib_find_cached_pkey(struct ib_device *device,
+			u8                port_num,
+			u16               pkey,
+			u16              *index)
+{
+	struct ib_pkey_cache *cache;
+	unsigned long flags;
+	int i;
+	int ret = -ENOENT;
+
+	if (port_num < start_port(device) || port_num > end_port(device))
+		return -EINVAL;
+
+	read_lock_irqsave(&device->cache.lock, flags);
+
+	cache = device->cache.pkey_cache[port_num - start_port(device)];
+
+	*index = -1;
+
+	for (i = 0; i < cache->table_len; ++i)
+		if ((cache->table[i] & 0x7fff) == (pkey & 0x7fff)) {
+			*index = i;
+			ret = 0;
+			break;
+		}
+
+	read_unlock_irqrestore(&device->cache.lock, flags);
+
+	return ret;
+}
+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)
+{
+	struct ib_port_attr       *tprops = NULL;
+	struct ib_pkey_cache      *pkey_cache = NULL, *old_pkey_cache;
+	struct ib_gid_cache       *gid_cache = NULL, *old_gid_cache;
+	int                        i;
+	int                        ret;
+
+	tprops = kmalloc(sizeof *tprops, GFP_KERNEL);
+	if (!tprops)
+		return;
+
+	ret = ib_query_port(device, port, tprops);
+	if (ret) {
+		printk(KERN_WARNING "ib_query_port failed (%d) for %s\n",
+		       ret, device->name);
+		goto err;
+	}
+
+	pkey_cache = kmalloc(sizeof *pkey_cache + tprops->pkey_tbl_len *
+			     sizeof *pkey_cache->table, GFP_KERNEL);
+	if (!pkey_cache)
+		goto err;
+
+	pkey_cache->table_len = tprops->pkey_tbl_len;
+
+	gid_cache = kmalloc(sizeof *gid_cache + tprops->gid_tbl_len *
+			    sizeof *gid_cache->table, GFP_KERNEL);
+	if (!gid_cache)
+		goto err;
+
+	gid_cache->table_len = tprops->gid_tbl_len;
+
+	for (i = 0; i < pkey_cache->table_len; ++i) {
+		ret = ib_query_pkey(device, port, i, pkey_cache->table + i);
+		if (ret) {
+			printk(KERN_WARNING "ib_query_pkey failed (%d) for %s (index %d)\n",
+			       ret, device->name, i);
+			goto err;
+		}
+	}
+
+	for (i = 0; i < gid_cache->table_len; ++i) {
+		ret = ib_query_gid(device, port, i, gid_cache->table + i);
+		if (ret) {
+			printk(KERN_WARNING "ib_query_gid failed (%d) for %s (index %d)\n",
+			       ret, device->name, i);
+			goto err;
+		}
+	}
+
+	write_lock_irq(&device->cache.lock);
+
+	old_pkey_cache = device->cache.pkey_cache[port - start_port(device)];
+	old_gid_cache  = device->cache.gid_cache [port - start_port(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);
+	kfree(old_gid_cache);
+	kfree(tprops);
+	return;
+
+err:
+	kfree(pkey_cache);
+	kfree(gid_cache);
+	kfree(tprops);
+}
+
+static void ib_cache_task(struct work_struct *_work)
+{
+	struct ib_update_work *work =
+		container_of(_work, struct ib_update_work, work);
+
+	ib_cache_update(work->device, work->port_num);
+	kfree(work);
+}
+
+static void ib_cache_event(struct ib_event_handler *handler,
+			   struct ib_event *event)
+{
+	struct ib_update_work *work;
+
+	if (event->event == IB_EVENT_PORT_ERR    ||
+	    event->event == IB_EVENT_PORT_ACTIVE ||
+	    event->event == IB_EVENT_LID_CHANGE  ||
+	    event->event == IB_EVENT_PKEY_CHANGE ||
+	    event->event == IB_EVENT_SM_CHANGE   ||
+	    event->event == IB_EVENT_CLIENT_REREGISTER) {
+		work = kmalloc(sizeof *work, GFP_ATOMIC);
+		if (work) {
+			INIT_WORK(&work->work, ib_cache_task);
+			work->device   = event->device;
+			work->port_num = event->element.port_num;
+			schedule_work(&work->work);
+		}
+	}
+}
+
+static void ib_cache_setup_one(struct ib_device *device)
+{
+	int p;
+
+	rwlock_init(&device->cache.lock);
+
+	device->cache.pkey_cache =
+		kmalloc(sizeof *device->cache.pkey_cache *
+			(end_port(device) - start_port(device) + 1), GFP_KERNEL);
+	device->cache.gid_cache =
+		kmalloc(sizeof *device->cache.gid_cache *
+			(end_port(device) - start_port(device) + 1), GFP_KERNEL);
+
+	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;
+	}
+
+	for (p = 0; p <= end_port(device) - start_port(device); ++p) {
+		device->cache.pkey_cache[p] = NULL;
+		device->cache.gid_cache [p] = NULL;
+		ib_cache_update(device, p + start_port(device));
+	}
+
+	INIT_IB_EVENT_HANDLER(&device->cache.event_handler,
+			      device, ib_cache_event);
+	if (ib_register_event_handler(&device->cache.event_handler))
+		goto err_cache;
+
+	return;
+
+err_cache:
+	for (p = 0; p <= end_port(device) - start_port(device); ++p) {
+		kfree(device->cache.pkey_cache[p]);
+		kfree(device->cache.gid_cache[p]);
+	}
+
+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)
+{
+	int p;
+
+	ib_unregister_event_handler(&device->cache.event_handler);
+	flush_scheduled_work();
+
+	for (p = 0; p <= end_port(device) - start_port(device); ++p) {
+		kfree(device->cache.pkey_cache[p]);
+		kfree(device->cache.gid_cache[p]);
+	}
+
+	kfree(device->cache.pkey_cache);
+	kfree(device->cache.gid_cache);
+	kfree(device->cache.lmc_cache);
+}
+
+static struct ib_client cache_client = {
+	.name   = "cache",
+	.add    = ib_cache_setup_one,
+	.remove = ib_cache_cleanup_one
+};
+
+int __init ib_cache_setup(void)
+{
+	return ib_register_client(&cache_client);
+}
+
+void __exit ib_cache_cleanup(void)
+{
+	ib_unregister_client(&cache_client);
+}

drivers/infiniband/core/cm_msgs.h

@@ -0,0 +1,818 @@
+/*
+ * Copyright (c) 2004 Intel Corporation.  All rights reserved.
+ * Copyright (c) 2004 Topspin Corporation.  All rights reserved.
+ * Copyright (c) 2004 Voltaire Corporation.  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 the madirectory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use source and binary forms, with or
+ *     withmodification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retathe above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHWARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS THE
+ * SOFTWARE.
+ */
+#if !defined(CM_MSGS_H)
+#define CM_MSGS_H
+
+#include <rdma/ib_mad.h>
+
+/*
+ * Parameters to routines below should be in network-byte order, and values
+ * are returned in network-byte order.
+ */
+
+#define IB_CM_CLASS_VERSION	2 /* IB specification 1.2 */
+
+#define CM_REQ_ATTR_ID	    __constant_htons(0x0010)
+#define CM_MRA_ATTR_ID	    __constant_htons(0x0011)
+#define CM_REJ_ATTR_ID	    __constant_htons(0x0012)
+#define CM_REP_ATTR_ID	    __constant_htons(0x0013)
+#define CM_RTU_ATTR_ID	    __constant_htons(0x0014)
+#define CM_DREQ_ATTR_ID	    __constant_htons(0x0015)
+#define CM_DREP_ATTR_ID	    __constant_htons(0x0016)
+#define CM_SIDR_REQ_ATTR_ID __constant_htons(0x0017)
+#define CM_SIDR_REP_ATTR_ID __constant_htons(0x0018)
+#define CM_LAP_ATTR_ID      __constant_htons(0x0019)
+#define CM_APR_ATTR_ID      __constant_htons(0x001A)
+
+enum cm_msg_sequence {
+	CM_MSG_SEQUENCE_REQ,
+	CM_MSG_SEQUENCE_LAP,
+	CM_MSG_SEQUENCE_DREQ,
+	CM_MSG_SEQUENCE_SIDR
+};
+
+struct cm_req_msg {
+	struct ib_mad_hdr hdr;
+
+	__be32 local_comm_id;
+	__be32 rsvd4;
+	__be64 service_id;
+	__be64 local_ca_guid;
+	__be32 rsvd24;
+	__be32 local_qkey;
+	/* local QPN:24, responder resources:8 */
+	__be32 offset32;
+	/* local EECN:24, initiator depth:8 */
+	__be32 offset36;
+	/*
+	 * remote EECN:24, remote CM response timeout:5,
+	 * transport service type:2, end-to-end flow control:1
+	 */
+	__be32 offset40;
+	/* starting PSN:24, local CM response timeout:5, retry count:3 */
+	__be32 offset44;
+	__be16 pkey;
+	/* path MTU:4, RDC exists:1, RNR retry count:3. */
+	u8 offset50;
+	/* max CM Retries:4, SRQ:1, rsvd:3 */
+	u8 offset51;
+
+	__be16 primary_local_lid;
+	__be16 primary_remote_lid;
+	union ib_gid primary_local_gid;
+	union ib_gid primary_remote_gid;
+	/* flow label:20, rsvd:6, packet rate:6 */
+	__be32 primary_offset88;
+	u8 primary_traffic_class;
+	u8 primary_hop_limit;
+	/* SL:4, subnet local:1, rsvd:3 */
+	u8 primary_offset94;
+	/* local ACK timeout:5, rsvd:3 */
+	u8 primary_offset95;
+
+	__be16 alt_local_lid;
+	__be16 alt_remote_lid;
+	union ib_gid alt_local_gid;
+	union ib_gid alt_remote_gid;
+	/* flow label:20, rsvd:6, packet rate:6 */
+	__be32 alt_offset132;
+	u8 alt_traffic_class;
+	u8 alt_hop_limit;
+	/* SL:4, subnet local:1, rsvd:3 */
+	u8 alt_offset138;
+	/* local ACK timeout:5, rsvd:3 */
+	u8 alt_offset139;
+
+	u8 private_data[IB_CM_REQ_PRIVATE_DATA_SIZE];
+
+} __attribute__ ((packed));
+
+static inline __be32 cm_req_get_local_qpn(struct cm_req_msg *req_msg)
+{
+	return cpu_to_be32(be32_to_cpu(req_msg->offset32) >> 8);
+}
+
+static inline void cm_req_set_local_qpn(struct cm_req_msg *req_msg, __be32 qpn)
+{
+	req_msg->offset32 = cpu_to_be32((be32_to_cpu(qpn) << 8) |
+					 (be32_to_cpu(req_msg->offset32) &
+					  0x000000FF));
+}
+
+static inline u8 cm_req_get_resp_res(struct cm_req_msg *req_msg)
+{
+	return (u8) be32_to_cpu(req_msg->offset32);
+}
+
+static inline void cm_req_set_resp_res(struct cm_req_msg *req_msg, u8 resp_res)
+{
+	req_msg->offset32 = cpu_to_be32(resp_res |
+					(be32_to_cpu(req_msg->offset32) &
+					 0xFFFFFF00));
+}
+
+static inline u8 cm_req_get_init_depth(struct cm_req_msg *req_msg)
+{
+	return (u8) be32_to_cpu(req_msg->offset36);
+}
+
+static inline void cm_req_set_init_depth(struct cm_req_msg *req_msg,
+					 u8 init_depth)
+{
+	req_msg->offset36 = cpu_to_be32(init_depth |
+					(be32_to_cpu(req_msg->offset36) &
+					 0xFFFFFF00));
+}
+
+static inline u8 cm_req_get_remote_resp_timeout(struct cm_req_msg *req_msg)
+{
+	return (u8) ((be32_to_cpu(req_msg->offset40) & 0xF8) >> 3);
+}
+
+static inline void cm_req_set_remote_resp_timeout(struct cm_req_msg *req_msg,
+						  u8 resp_timeout)
+{
+	req_msg->offset40 = cpu_to_be32((resp_timeout << 3) |
+					 (be32_to_cpu(req_msg->offset40) &
+					  0xFFFFFF07));
+}
+
+static inline enum ib_qp_type cm_req_get_qp_type(struct cm_req_msg *req_msg)
+{
+	u8 transport_type = (u8) (be32_to_cpu(req_msg->offset40) & 0x06) >> 1;
+	switch(transport_type) {
+	case 0: return IB_QPT_RC;
+	case 1: return IB_QPT_UC;
+	default: return 0;
+	}
+}
+
+static inline void cm_req_set_qp_type(struct cm_req_msg *req_msg,
+				      enum ib_qp_type qp_type)
+{
+	switch(qp_type) {
+	case IB_QPT_UC:
+		req_msg->offset40 = cpu_to_be32((be32_to_cpu(
+						  req_msg->offset40) &
+						   0xFFFFFFF9) | 0x2);
+		break;
+	default:
+		req_msg->offset40 = cpu_to_be32(be32_to_cpu(
+						 req_msg->offset40) &
+						  0xFFFFFFF9);
+	}
+}
+
+static inline u8 cm_req_get_flow_ctrl(struct cm_req_msg *req_msg)
+{
+	return be32_to_cpu(req_msg->offset40) & 0x1;
+}
+
+static inline void cm_req_set_flow_ctrl(struct cm_req_msg *req_msg,
+					u8 flow_ctrl)
+{
+	req_msg->offset40 = cpu_to_be32((flow_ctrl & 0x1) |
+					 (be32_to_cpu(req_msg->offset40) &
+					  0xFFFFFFFE));
+}
+
+static inline __be32 cm_req_get_starting_psn(struct cm_req_msg *req_msg)
+{
+	return cpu_to_be32(be32_to_cpu(req_msg->offset44) >> 8);
+}
+
+static inline void cm_req_set_starting_psn(struct cm_req_msg *req_msg,
+					   __be32 starting_psn)
+{
+	req_msg->offset44 = cpu_to_be32((be32_to_cpu(starting_psn) << 8) |
+			    (be32_to_cpu(req_msg->offset44) & 0x000000FF));
+}
+
+static inline u8 cm_req_get_local_resp_timeout(struct cm_req_msg *req_msg)
+{
+	return (u8) ((be32_to_cpu(req_msg->offset44) & 0xF8) >> 3);
+}
+
+static inline void cm_req_set_local_resp_timeout(struct cm_req_msg *req_msg,
+						 u8 resp_timeout)
+{
+	req_msg->offset44 = cpu_to_be32((resp_timeout << 3) |
+			    (be32_to_cpu(req_msg->offset44) & 0xFFFFFF07));
+}
+
+static inline u8 cm_req_get_retry_count(struct cm_req_msg *req_msg)
+{
+	return (u8) (be32_to_cpu(req_msg->offset44) & 0x7);
+}
+
+static inline void cm_req_set_retry_count(struct cm_req_msg *req_msg,
+					  u8 retry_count)
+{
+	req_msg->offset44 = cpu_to_be32((retry_count & 0x7) |
+			    (be32_to_cpu(req_msg->offset44) & 0xFFFFFFF8));
+}
+
+static inline u8 cm_req_get_path_mtu(struct cm_req_msg *req_msg)
+{
+	return req_msg->offset50 >> 4;
+}
+
+static inline void cm_req_set_path_mtu(struct cm_req_msg *req_msg, u8 path_mtu)
+{
+	req_msg->offset50 = (u8) ((req_msg->offset50 & 0xF) | (path_mtu << 4));
+}
+
+static inline u8 cm_req_get_rnr_retry_count(struct cm_req_msg *req_msg)
+{
+	return req_msg->offset50 & 0x7;
+}
+
+static inline void cm_req_set_rnr_retry_count(struct cm_req_msg *req_msg,
+					      u8 rnr_retry_count)
+{
+	req_msg->offset50 = (u8) ((req_msg->offset50 & 0xF8) |
+				  (rnr_retry_count & 0x7));
+}
+
+static inline u8 cm_req_get_max_cm_retries(struct cm_req_msg *req_msg)
+{
+	return req_msg->offset51 >> 4;
+}
+
+static inline void cm_req_set_max_cm_retries(struct cm_req_msg *req_msg,
+					     u8 retries)
+{
+	req_msg->offset51 = (u8) ((req_msg->offset51 & 0xF) | (retries << 4));
+}
+
+static inline u8 cm_req_get_srq(struct cm_req_msg *req_msg)
+{
+	return (req_msg->offset51 & 0x8) >> 3;
+}
+
+static inline void cm_req_set_srq(struct cm_req_msg *req_msg, u8 srq)
+{
+	req_msg->offset51 = (u8) ((req_msg->offset51 & 0xF7) |
+				  ((srq & 0x1) << 3));
+}
+
+static inline __be32 cm_req_get_primary_flow_label(struct cm_req_msg *req_msg)
+{
+	return cpu_to_be32(be32_to_cpu(req_msg->primary_offset88) >> 12);
+}
+
+static inline void cm_req_set_primary_flow_label(struct cm_req_msg *req_msg,
+						 __be32 flow_label)
+{
+	req_msg->primary_offset88 = cpu_to_be32(
+				    (be32_to_cpu(req_msg->primary_offset88) &
+				     0x00000FFF) |
+				     (be32_to_cpu(flow_label) << 12));
+}
+
+static inline u8 cm_req_get_primary_packet_rate(struct cm_req_msg *req_msg)
+{
+	return (u8) (be32_to_cpu(req_msg->primary_offset88) & 0x3F);
+}
+
+static inline void cm_req_set_primary_packet_rate(struct cm_req_msg *req_msg,
+						  u8 rate)
+{
+	req_msg->primary_offset88 = cpu_to_be32(
+				    (be32_to_cpu(req_msg->primary_offset88) &
+				     0xFFFFFFC0) | (rate & 0x3F));
+}
+
+static inline u8 cm_req_get_primary_sl(struct cm_req_msg *req_msg)
+{
+	return (u8) (req_msg->primary_offset94 >> 4);
+}
+
+static inline void cm_req_set_primary_sl(struct cm_req_msg *req_msg, u8 sl)
+{
+	req_msg->primary_offset94 = (u8) ((req_msg->primary_offset94 & 0x0F) |
+					  (sl << 4));
+}
+
+static inline u8 cm_req_get_primary_subnet_local(struct cm_req_msg *req_msg)
+{
+	return (u8) ((req_msg->primary_offset94 & 0x08) >> 3);
+}
+
+static inline void cm_req_set_primary_subnet_local(struct cm_req_msg *req_msg,
+						   u8 subnet_local)
+{
+	req_msg->primary_offset94 = (u8) ((req_msg->primary_offset94 & 0xF7) |
+					  ((subnet_local & 0x1) << 3));
+}
+
+static inline u8 cm_req_get_primary_local_ack_timeout(struct cm_req_msg *req_msg)
+{
+	return (u8) (req_msg->primary_offset95 >> 3);
+}
+
+static inline void cm_req_set_primary_local_ack_timeout(struct cm_req_msg *req_msg,
+							u8 local_ack_timeout)
+{
+	req_msg->primary_offset95 = (u8) ((req_msg->primary_offset95 & 0x07) |
+					  (local_ack_timeout << 3));
+}
+
+static inline __be32 cm_req_get_alt_flow_label(struct cm_req_msg *req_msg)
+{
+	return cpu_to_be32(be32_to_cpu(req_msg->alt_offset132) >> 12);
+}
+
+static inline void cm_req_set_alt_flow_label(struct cm_req_msg *req_msg,
+					     __be32 flow_label)
+{
+	req_msg->alt_offset132 = cpu_to_be32(
+				 (be32_to_cpu(req_msg->alt_offset132) &
+				  0x00000FFF) |
+				  (be32_to_cpu(flow_label) << 12));
+}
+
+static inline u8 cm_req_get_alt_packet_rate(struct cm_req_msg *req_msg)
+{
+	return (u8) (be32_to_cpu(req_msg->alt_offset132) & 0x3F);
+}
+
+static inline void cm_req_set_alt_packet_rate(struct cm_req_msg *req_msg,
+					      u8 rate)
+{
+	req_msg->alt_offset132 = cpu_to_be32(
+				 (be32_to_cpu(req_msg->alt_offset132) &
+				  0xFFFFFFC0) | (rate & 0x3F));
+}
+
+static inline u8 cm_req_get_alt_sl(struct cm_req_msg *req_msg)
+{
+	return (u8) (req_msg->alt_offset138 >> 4);
+}
+
+static inline void cm_req_set_alt_sl(struct cm_req_msg *req_msg, u8 sl)
+{
+	req_msg->alt_offset138 = (u8) ((req_msg->alt_offset138 & 0x0F) |
+				       (sl << 4));
+}
+
+static inline u8 cm_req_get_alt_subnet_local(struct cm_req_msg *req_msg)
+{
+	return (u8) ((req_msg->alt_offset138 & 0x08) >> 3);
+}
+
+static inline void cm_req_set_alt_subnet_local(struct cm_req_msg *req_msg,
+					       u8 subnet_local)
+{
+	req_msg->alt_offset138 = (u8) ((req_msg->alt_offset138 & 0xF7) |
+				       ((subnet_local & 0x1) << 3));
+}
+
+static inline u8 cm_req_get_alt_local_ack_timeout(struct cm_req_msg *req_msg)
+{
+	return (u8) (req_msg->alt_offset139 >> 3);
+}
+
+static inline void cm_req_set_alt_local_ack_timeout(struct cm_req_msg *req_msg,
+						    u8 local_ack_timeout)
+{
+	req_msg->alt_offset139 = (u8) ((req_msg->alt_offset139 & 0x07) |
+				       (local_ack_timeout << 3));
+}
+
+/* Message REJected or MRAed */
+enum cm_msg_response {
+	CM_MSG_RESPONSE_REQ = 0x0,
+	CM_MSG_RESPONSE_REP = 0x1,
+	CM_MSG_RESPONSE_OTHER = 0x2
+};
+
+ struct cm_mra_msg {
+	struct ib_mad_hdr hdr;
+
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
+	/* message MRAed:2, rsvd:6 */
+	u8 offset8;
+	/* service timeout:5, rsvd:3 */
+	u8 offset9;
+
+	u8 private_data[IB_CM_MRA_PRIVATE_DATA_SIZE];
+
+} __attribute__ ((packed));
+
+static inline u8 cm_mra_get_msg_mraed(struct cm_mra_msg *mra_msg)
+{
+	return (u8) (mra_msg->offset8 >> 6);
+}
+
+static inline void cm_mra_set_msg_mraed(struct cm_mra_msg *mra_msg, u8 msg)
+{
+	mra_msg->offset8 = (u8) ((mra_msg->offset8 & 0x3F) | (msg << 6));
+}
+
+static inline u8 cm_mra_get_service_timeout(struct cm_mra_msg *mra_msg)
+{
+	return (u8) (mra_msg->offset9 >> 3);
+}
+
+static inline void cm_mra_set_service_timeout(struct cm_mra_msg *mra_msg,
+					      u8 service_timeout)
+{
+	mra_msg->offset9 = (u8) ((mra_msg->offset9 & 0x07) |
+				 (service_timeout << 3));
+}
+
+struct cm_rej_msg {
+	struct ib_mad_hdr hdr;
+
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
+	/* message REJected:2, rsvd:6 */
+	u8 offset8;
+	/* reject info length:7, rsvd:1. */
+	u8 offset9;
+	__be16 reason;
+	u8 ari[IB_CM_REJ_ARI_LENGTH];
+
+	u8 private_data[IB_CM_REJ_PRIVATE_DATA_SIZE];
+
+} __attribute__ ((packed));
+
+static inline u8 cm_rej_get_msg_rejected(struct cm_rej_msg *rej_msg)
+{
+	return (u8) (rej_msg->offset8 >> 6);
+}
+
+static inline void cm_rej_set_msg_rejected(struct cm_rej_msg *rej_msg, u8 msg)
+{
+	rej_msg->offset8 = (u8) ((rej_msg->offset8 & 0x3F) | (msg << 6));
+}
+
+static inline u8 cm_rej_get_reject_info_len(struct cm_rej_msg *rej_msg)
+{
+	return (u8) (rej_msg->offset9 >> 1);
+}
+
+static inline void cm_rej_set_reject_info_len(struct cm_rej_msg *rej_msg,
+					      u8 len)
+{
+	rej_msg->offset9 = (u8) ((rej_msg->offset9 & 0x1) | (len << 1));
+}
+
+struct cm_rep_msg {
+	struct ib_mad_hdr hdr;
+
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
+	__be32 local_qkey;
+	/* local QPN:24, rsvd:8 */
+	__be32 offset12;
+	/* local EECN:24, rsvd:8 */
+	__be32 offset16;
+	/* starting PSN:24 rsvd:8 */
+	__be32 offset20;
+	u8 resp_resources;
+	u8 initiator_depth;
+	/* target ACK delay:5, failover accepted:2, end-to-end flow control:1 */
+	u8 offset26;
+	/* RNR retry count:3, SRQ:1, rsvd:5 */
+	u8 offset27;
+	__be64 local_ca_guid;
+
+	u8 private_data[IB_CM_REP_PRIVATE_DATA_SIZE];
+
+} __attribute__ ((packed));
+
+static inline __be32 cm_rep_get_local_qpn(struct cm_rep_msg *rep_msg)
+{
+	return cpu_to_be32(be32_to_cpu(rep_msg->offset12) >> 8);
+}
+
+static inline void cm_rep_set_local_qpn(struct cm_rep_msg *rep_msg, __be32 qpn)
+{
+	rep_msg->offset12 = cpu_to_be32((be32_to_cpu(qpn) << 8) |
+			    (be32_to_cpu(rep_msg->offset12) & 0x000000FF));
+}
+
+static inline __be32 cm_rep_get_starting_psn(struct cm_rep_msg *rep_msg)
+{
+	return cpu_to_be32(be32_to_cpu(rep_msg->offset20) >> 8);
+}
+
+static inline void cm_rep_set_starting_psn(struct cm_rep_msg *rep_msg,
+					   __be32 starting_psn)
+{
+	rep_msg->offset20 = cpu_to_be32((be32_to_cpu(starting_psn) << 8) |
+			    (be32_to_cpu(rep_msg->offset20) & 0x000000FF));
+}
+
+static inline u8 cm_rep_get_target_ack_delay(struct cm_rep_msg *rep_msg)
+{
+	return (u8) (rep_msg->offset26 >> 3);
+}
+
+static inline void cm_rep_set_target_ack_delay(struct cm_rep_msg *rep_msg,
+					       u8 target_ack_delay)
+{
+	rep_msg->offset26 = (u8) ((rep_msg->offset26 & 0x07) |
+				  (target_ack_delay << 3));
+}
+
+static inline u8 cm_rep_get_failover(struct cm_rep_msg *rep_msg)
+{
+	return (u8) ((rep_msg->offset26 & 0x06) >> 1);
+}
+
+static inline void cm_rep_set_failover(struct cm_rep_msg *rep_msg, u8 failover)
+{
+	rep_msg->offset26 = (u8) ((rep_msg->offset26 & 0xF9) |
+				  ((failover & 0x3) << 1));
+}
+
+static inline u8 cm_rep_get_flow_ctrl(struct cm_rep_msg *rep_msg)
+{
+	return (u8) (rep_msg->offset26 & 0x01);
+}
+
+static inline void cm_rep_set_flow_ctrl(struct cm_rep_msg *rep_msg,
+					    u8 flow_ctrl)
+{
+	rep_msg->offset26 = (u8) ((rep_msg->offset26 & 0xFE) |
+				  (flow_ctrl & 0x1));
+}
+
+static inline u8 cm_rep_get_rnr_retry_count(struct cm_rep_msg *rep_msg)
+{
+	return (u8) (rep_msg->offset27 >> 5);
+}
+
+static inline void cm_rep_set_rnr_retry_count(struct cm_rep_msg *rep_msg,
+					      u8 rnr_retry_count)
+{
+	rep_msg->offset27 = (u8) ((rep_msg->offset27 & 0x1F) |
+				  (rnr_retry_count << 5));
+}
+
+static inline u8 cm_rep_get_srq(struct cm_rep_msg *rep_msg)
+{
+	return (u8) ((rep_msg->offset27 >> 4) & 0x1);
+}
+
+static inline void cm_rep_set_srq(struct cm_rep_msg *rep_msg, u8 srq)
+{
+	rep_msg->offset27 = (u8) ((rep_msg->offset27 & 0xEF) |
+				  ((srq & 0x1) << 4));
+}
+
+struct cm_rtu_msg {
+	struct ib_mad_hdr hdr;
+
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
+
+	u8 private_data[IB_CM_RTU_PRIVATE_DATA_SIZE];
+
+} __attribute__ ((packed));
+
+struct cm_dreq_msg {
+	struct ib_mad_hdr hdr;
+
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
+	/* remote QPN/EECN:24, rsvd:8 */
+	__be32 offset8;
+
+	u8 private_data[IB_CM_DREQ_PRIVATE_DATA_SIZE];
+
+} __attribute__ ((packed));
+
+static inline __be32 cm_dreq_get_remote_qpn(struct cm_dreq_msg *dreq_msg)
+{
+	return cpu_to_be32(be32_to_cpu(dreq_msg->offset8) >> 8);
+}
+
+static inline void cm_dreq_set_remote_qpn(struct cm_dreq_msg *dreq_msg, __be32 qpn)
+{
+	dreq_msg->offset8 = cpu_to_be32((be32_to_cpu(qpn) << 8) |
+			    (be32_to_cpu(dreq_msg->offset8) & 0x000000FF));
+}
+
+struct cm_drep_msg {
+	struct ib_mad_hdr hdr;
+
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
+
+	u8 private_data[IB_CM_DREP_PRIVATE_DATA_SIZE];
+
+} __attribute__ ((packed));
+
+struct cm_lap_msg {
+	struct ib_mad_hdr hdr;
+
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
+
+	__be32 rsvd8;
+	/* remote QPN/EECN:24, remote CM response timeout:5, rsvd:3 */
+	__be32 offset12;
+	__be32 rsvd16;
+
+	__be16 alt_local_lid;
+	__be16 alt_remote_lid;
+	union ib_gid alt_local_gid;
+	union ib_gid alt_remote_gid;
+	/* flow label:20, rsvd:4, traffic class:8 */
+	__be32 offset56;
+	u8 alt_hop_limit;
+	/* rsvd:2, packet rate:6 */
+	u8 offset61;
+	/* SL:4, subnet local:1, rsvd:3 */
+	u8 offset62;
+	/* local ACK timeout:5, rsvd:3 */
+	u8 offset63;
+
+	u8 private_data[IB_CM_LAP_PRIVATE_DATA_SIZE];
+} __attribute__  ((packed));
+
+static inline __be32 cm_lap_get_remote_qpn(struct cm_lap_msg *lap_msg)
+{
+	return cpu_to_be32(be32_to_cpu(lap_msg->offset12) >> 8);
+}
+
+static inline void cm_lap_set_remote_qpn(struct cm_lap_msg *lap_msg, __be32 qpn)
+{
+	lap_msg->offset12 = cpu_to_be32((be32_to_cpu(qpn) << 8) |
+					 (be32_to_cpu(lap_msg->offset12) &
+					  0x000000FF));
+}
+
+static inline u8 cm_lap_get_remote_resp_timeout(struct cm_lap_msg *lap_msg)
+{
+	return (u8) ((be32_to_cpu(lap_msg->offset12) & 0xF8) >> 3);
+}
+
+static inline void cm_lap_set_remote_resp_timeout(struct cm_lap_msg *lap_msg,
+						  u8 resp_timeout)
+{
+	lap_msg->offset12 = cpu_to_be32((resp_timeout << 3) |
+					 (be32_to_cpu(lap_msg->offset12) &
+					  0xFFFFFF07));
+}
+
+static inline __be32 cm_lap_get_flow_label(struct cm_lap_msg *lap_msg)
+{
+	return cpu_to_be32(be32_to_cpu(lap_msg->offset56) >> 12);
+}
+
+static inline void cm_lap_set_flow_label(struct cm_lap_msg *lap_msg,
+					 __be32 flow_label)
+{
+	lap_msg->offset56 = cpu_to_be32(
+				 (be32_to_cpu(lap_msg->offset56) & 0x00000FFF) |
+				 (be32_to_cpu(flow_label) << 12));
+}
+
+static inline u8 cm_lap_get_traffic_class(struct cm_lap_msg *lap_msg)
+{
+	return (u8) be32_to_cpu(lap_msg->offset56);
+}
+
+static inline void cm_lap_set_traffic_class(struct cm_lap_msg *lap_msg,
+					    u8 traffic_class)
+{
+	lap_msg->offset56 = cpu_to_be32(traffic_class |
+					 (be32_to_cpu(lap_msg->offset56) &
+					  0xFFFFFF00));
+}
+
+static inline u8 cm_lap_get_packet_rate(struct cm_lap_msg *lap_msg)
+{
+	return lap_msg->offset61 & 0x3F;
+}
+
+static inline void cm_lap_set_packet_rate(struct cm_lap_msg *lap_msg,
+					  u8 packet_rate)
+{
+	lap_msg->offset61 = (packet_rate & 0x3F) | (lap_msg->offset61 & 0xC0);
+}
+
+static inline u8 cm_lap_get_sl(struct cm_lap_msg *lap_msg)
+{
+	return lap_msg->offset62 >> 4;
+}
+
+static inline void cm_lap_set_sl(struct cm_lap_msg *lap_msg, u8 sl)
+{
+	lap_msg->offset62 = (sl << 4) | (lap_msg->offset62 & 0x0F);
+}
+
+static inline u8 cm_lap_get_subnet_local(struct cm_lap_msg *lap_msg)
+{
+	return (lap_msg->offset62 >> 3) & 0x1;
+}
+
+static inline void cm_lap_set_subnet_local(struct cm_lap_msg *lap_msg,
+					   u8 subnet_local)
+{
+	lap_msg->offset62 = ((subnet_local & 0x1) << 3) |
+			     (lap_msg->offset61 & 0xF7);
+}
+static inline u8 cm_lap_get_local_ack_timeout(struct cm_lap_msg *lap_msg)
+{
+	return lap_msg->offset63 >> 3;
+}
+
+static inline void cm_lap_set_local_ack_timeout(struct cm_lap_msg *lap_msg,
+						u8 local_ack_timeout)
+{
+	lap_msg->offset63 = (local_ack_timeout << 3) |
+			    (lap_msg->offset63 & 0x07);
+}
+
+struct cm_apr_msg {
+	struct ib_mad_hdr hdr;
+
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
+
+	u8 info_length;
+	u8 ap_status;
+	u8 info[IB_CM_APR_INFO_LENGTH];
+
+	u8 private_data[IB_CM_APR_PRIVATE_DATA_SIZE];
+} __attribute__ ((packed));
+
+struct cm_sidr_req_msg {
+	struct ib_mad_hdr hdr;
+
+	__be32 request_id;
+	__be16 pkey;
+	__be16 rsvd;
+	__be64 service_id;
+
+	u8 private_data[IB_CM_SIDR_REQ_PRIVATE_DATA_SIZE];
+} __attribute__ ((packed));
+
+struct cm_sidr_rep_msg {
+	struct ib_mad_hdr hdr;
+
+	__be32 request_id;
+	u8 status;
+	u8 info_length;
+	__be16 rsvd;
+	/* QPN:24, rsvd:8 */
+	__be32 offset8;
+	__be64 service_id;
+	__be32 qkey;
+	u8 info[IB_CM_SIDR_REP_INFO_LENGTH];
+
+	u8 private_data[IB_CM_SIDR_REP_PRIVATE_DATA_SIZE];
+} __attribute__ ((packed));
+
+static inline __be32 cm_sidr_rep_get_qpn(struct cm_sidr_rep_msg *sidr_rep_msg)
+{
+	return cpu_to_be32(be32_to_cpu(sidr_rep_msg->offset8) >> 8);
+}
+
+static inline void cm_sidr_rep_set_qpn(struct cm_sidr_rep_msg *sidr_rep_msg,
+				       __be32 qpn)
+{
+	sidr_rep_msg->offset8 = cpu_to_be32((be32_to_cpu(qpn) << 8) |
+					(be32_to_cpu(sidr_rep_msg->offset8) &
+					 0x000000FF));
+}
+
+#endif /* CM_MSGS_H */

drivers/infiniband/core/core_priv.h

@@ -0,0 +1,52 @@
+/*
+ * Copyright (c) 2004 Topspin Communications.  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.
+ *
+ * $Id: core_priv.h,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#ifndef _CORE_PRIV_H
+#define _CORE_PRIV_H
+
+#include <linux/list.h>
+#include <linux/spinlock.h>
+
+#include <rdma/ib_verbs.h>
+
+int  ib_device_register_sysfs(struct ib_device *device);
+void ib_device_unregister_sysfs(struct ib_device *device);
+
+int  ib_sysfs_setup(void);
+void ib_sysfs_cleanup(void);
+
+int  ib_cache_setup(void);
+void ib_cache_cleanup(void);
+
+#endif /* _CORE_PRIV_H */

drivers/infiniband/core/device.c

@@ -0,0 +1,619 @@
+/*
+ * Copyright (c) 2004 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. 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.
+ *
+ * $Id: device.c,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#include <linux/module.h>
+#include <linux/string.h>
+#include <linux/errno.h>
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/init.h>
+#include <linux/mutex.h>
+
+#include "core_priv.h"
+
+MODULE_AUTHOR("Roland Dreier");
+MODULE_DESCRIPTION("core kernel InfiniBand API");
+MODULE_LICENSE("Dual BSD/GPL");
+
+struct ib_client_data {
+	struct list_head  list;
+	struct ib_client *client;
+	void *            data;
+};
+
+static LIST_HEAD(device_list);
+static LIST_HEAD(client_list);
+
+/*
+ * device_mutex protects access to both device_list and client_list.
+ * There's no real point to using multiple locks or something fancier
+ * like an rwsem: we always access both lists, and we're always
+ * modifying one list or the other list.  In any case this is not a
+ * hot path so there's no point in trying to optimize.
+ */
+static DEFINE_MUTEX(device_mutex);
+
+static int ib_device_check_mandatory(struct ib_device *device)
+{
+#define IB_MANDATORY_FUNC(x) { offsetof(struct ib_device, x), #x }
+	static const struct {
+		size_t offset;
+		char  *name;
+	} mandatory_table[] = {
+		IB_MANDATORY_FUNC(query_device),
+		IB_MANDATORY_FUNC(query_port),
+		IB_MANDATORY_FUNC(query_pkey),
+		IB_MANDATORY_FUNC(query_gid),
+		IB_MANDATORY_FUNC(alloc_pd),
+		IB_MANDATORY_FUNC(dealloc_pd),
+		IB_MANDATORY_FUNC(create_ah),
+		IB_MANDATORY_FUNC(destroy_ah),
+		IB_MANDATORY_FUNC(create_qp),
+		IB_MANDATORY_FUNC(modify_qp),
+		IB_MANDATORY_FUNC(destroy_qp),
+		IB_MANDATORY_FUNC(post_send),
+		IB_MANDATORY_FUNC(post_recv),
+		IB_MANDATORY_FUNC(create_cq),
+		IB_MANDATORY_FUNC(destroy_cq),
+		IB_MANDATORY_FUNC(poll_cq),
+		IB_MANDATORY_FUNC(req_notify_cq),
+		IB_MANDATORY_FUNC(get_dma_mr),
+		IB_MANDATORY_FUNC(dereg_mr)
+	};
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(mandatory_table); ++i) {
+		if (!*(void **) ((void *) device + mandatory_table[i].offset)) {
+			printk(KERN_WARNING "Device %s is missing mandatory function %s\n",
+			       device->name, mandatory_table[i].name);
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+static struct ib_device *__ib_device_get_by_name(const char *name)
+{
+	struct ib_device *device;
+
+	list_for_each_entry(device, &device_list, core_list)
+		if (!strncmp(name, device->name, IB_DEVICE_NAME_MAX))
+			return device;
+
+	return NULL;
+}
+
+
+static int alloc_name(char *name)
+{
+	long *inuse;
+	char buf[IB_DEVICE_NAME_MAX];
+	struct ib_device *device;
+	int i;
+
+	inuse = (long *) get_zeroed_page(GFP_KERNEL);
+	if (!inuse)
+		return -ENOMEM;
+
+	list_for_each_entry(device, &device_list, core_list) {
+		if (!sscanf(device->name, name, &i))
+			continue;
+		if (i < 0 || i >= PAGE_SIZE * 8)
+			continue;
+		snprintf(buf, sizeof buf, name, i);
+		if (!strncmp(buf, device->name, IB_DEVICE_NAME_MAX))
+			set_bit(i, inuse);
+	}
+
+	i = find_first_zero_bit(inuse, PAGE_SIZE * 8);
+	free_page((unsigned long) inuse);
+	snprintf(buf, sizeof buf, name, i);
+
+	if (__ib_device_get_by_name(buf))
+		return -ENFILE;
+
+	strlcpy(name, buf, IB_DEVICE_NAME_MAX);
+	return 0;
+}
+
+/**
+ * ib_alloc_device - allocate an IB device struct
+ * @size:size of structure to allocate
+ *
+ * Low-level drivers should use ib_alloc_device() to allocate &struct
+ * ib_device.  @size is the size of the structure to be allocated,
+ * including any private data used by the low-level driver.
+ * ib_dealloc_device() must be used to free structures allocated with
+ * ib_alloc_device().
+ */
+struct ib_device *ib_alloc_device(size_t size)
+{
+	BUG_ON(size < sizeof (struct ib_device));
+
+	return kzalloc(size, GFP_KERNEL);
+}
+EXPORT_SYMBOL(ib_alloc_device);
+
+/**
+ * ib_dealloc_device - free an IB device struct
+ * @device:structure to free
+ *
+ * Free a structure allocated with ib_alloc_device().
+ */
+void ib_dealloc_device(struct ib_device *device)
+{
+	if (device->reg_state == IB_DEV_UNINITIALIZED) {
+		kfree(device);
+		return;
+	}
+
+	BUG_ON(device->reg_state != IB_DEV_UNREGISTERED);
+
+	ib_device_unregister_sysfs(device);
+}
+EXPORT_SYMBOL(ib_dealloc_device);
+
+static int add_client_context(struct ib_device *device, struct ib_client *client)
+{
+	struct ib_client_data *context;
+	unsigned long flags;
+
+	context = kmalloc(sizeof *context, GFP_KERNEL);
+	if (!context) {
+		printk(KERN_WARNING "Couldn't allocate client context for %s/%s\n",
+		       device->name, client->name);
+		return -ENOMEM;
+	}
+
+	context->client = client;
+	context->data   = NULL;
+
+	spin_lock_irqsave(&device->client_data_lock, flags);
+	list_add(&context->list, &device->client_data_list);
+	spin_unlock_irqrestore(&device->client_data_lock, flags);
+
+	return 0;
+}
+
+/**
+ * ib_register_device - Register an IB device with IB core
+ * @device:Device to register
+ *
+ * Low-level drivers use ib_register_device() to register their
+ * devices with the IB core.  All registered clients will receive a
+ * callback for each device that is added. @device must be allocated
+ * with ib_alloc_device().
+ */
+int ib_register_device(struct ib_device *device)
+{
+	int ret;
+
+	mutex_lock(&device_mutex);
+
+	if (strchr(device->name, '%')) {
+		ret = alloc_name(device->name);
+		if (ret)
+			goto out;
+	}
+
+	if (ib_device_check_mandatory(device)) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	INIT_LIST_HEAD(&device->event_handler_list);
+	INIT_LIST_HEAD(&device->client_data_list);
+	spin_lock_init(&device->event_handler_lock);
+	spin_lock_init(&device->client_data_lock);
+
+	ret = ib_device_register_sysfs(device);
+	if (ret) {
+		printk(KERN_WARNING "Couldn't register device %s with driver model\n",
+		       device->name);
+		goto out;
+	}
+
+	list_add_tail(&device->core_list, &device_list);
+
+	device->reg_state = IB_DEV_REGISTERED;
+
+	{
+		struct ib_client *client;
+
+		list_for_each_entry(client, &client_list, list)
+			if (client->add && !add_client_context(device, client))
+				client->add(device);
+	}
+
+ out:
+	mutex_unlock(&device_mutex);
+	return ret;
+}
+EXPORT_SYMBOL(ib_register_device);
+
+/**
+ * ib_unregister_device - Unregister an IB device
+ * @device:Device to unregister
+ *
+ * Unregister an IB device.  All clients will receive a remove callback.
+ */
+void ib_unregister_device(struct ib_device *device)
+{
+	struct ib_client *client;
+	struct ib_client_data *context, *tmp;
+	unsigned long flags;
+
+	mutex_lock(&device_mutex);
+
+	list_for_each_entry_reverse(client, &client_list, list)
+		if (client->remove)
+			client->remove(device);
+
+	list_del(&device->core_list);
+
+	mutex_unlock(&device_mutex);
+
+	spin_lock_irqsave(&device->client_data_lock, flags);
+	list_for_each_entry_safe(context, tmp, &device->client_data_list, list)
+		kfree(context);
+	spin_unlock_irqrestore(&device->client_data_lock, flags);
+
+	device->reg_state = IB_DEV_UNREGISTERED;
+}
+EXPORT_SYMBOL(ib_unregister_device);
+
+/**
+ * ib_register_client - Register an IB client
+ * @client:Client to register
+ *
+ * Upper level users of the IB drivers can use ib_register_client() to
+ * register callbacks for IB device addition and removal.  When an IB
+ * device is added, each registered client's add method will be called
+ * (in the order the clients were registered), and when a device is
+ * removed, each client's remove method will be called (in the reverse
+ * order that clients were registered).  In addition, when
+ * ib_register_client() is called, the client will receive an add
+ * callback for all devices already registered.
+ */
+int ib_register_client(struct ib_client *client)
+{
+	struct ib_device *device;
+
+	mutex_lock(&device_mutex);
+
+	list_add_tail(&client->list, &client_list);
+	list_for_each_entry(device, &device_list, core_list)
+		if (client->add && !add_client_context(device, client))
+			client->add(device);
+
+	mutex_unlock(&device_mutex);
+
+	return 0;
+}
+EXPORT_SYMBOL(ib_register_client);
+
+/**
+ * ib_unregister_client - Unregister an IB client
+ * @client:Client to unregister
+ *
+ * Upper level users use ib_unregister_client() to remove their client
+ * registration.  When ib_unregister_client() is called, the client
+ * will receive a remove callback for each IB device still registered.
+ */
+void ib_unregister_client(struct ib_client *client)
+{
+	struct ib_client_data *context, *tmp;
+	struct ib_device *device;
+	unsigned long flags;
+
+	mutex_lock(&device_mutex);
+
+	list_for_each_entry(device, &device_list, core_list) {
+		if (client->remove)
+			client->remove(device);
+
+		spin_lock_irqsave(&device->client_data_lock, flags);
+		list_for_each_entry_safe(context, tmp, &device->client_data_list, list)
+			if (context->client == client) {
+				list_del(&context->list);
+				kfree(context);
+			}
+		spin_unlock_irqrestore(&device->client_data_lock, flags);
+	}
+	list_del(&client->list);
+
+	mutex_unlock(&device_mutex);
+}
+EXPORT_SYMBOL(ib_unregister_client);
+
+/**
+ * ib_get_client_data - Get IB client context
+ * @device:Device to get context for
+ * @client:Client to get context for
+ *
+ * ib_get_client_data() returns client context set with
+ * ib_set_client_data().
+ */
+void *ib_get_client_data(struct ib_device *device, struct ib_client *client)
+{
+	struct ib_client_data *context;
+	void *ret = NULL;
+	unsigned long flags;
+
+	spin_lock_irqsave(&device->client_data_lock, flags);
+	list_for_each_entry(context, &device->client_data_list, list)
+		if (context->client == client) {
+			ret = context->data;
+			break;
+		}
+	spin_unlock_irqrestore(&device->client_data_lock, flags);
+
+	return ret;
+}
+EXPORT_SYMBOL(ib_get_client_data);
+
+/**
+ * ib_set_client_data - Set IB client context
+ * @device:Device to set context for
+ * @client:Client to set context for
+ * @data:Context to set
+ *
+ * ib_set_client_data() sets client context that can be retrieved with
+ * ib_get_client_data().
+ */
+void ib_set_client_data(struct ib_device *device, struct ib_client *client,
+			void *data)
+{
+	struct ib_client_data *context;
+	unsigned long flags;
+
+	spin_lock_irqsave(&device->client_data_lock, flags);
+	list_for_each_entry(context, &device->client_data_list, list)
+		if (context->client == client) {
+			context->data = data;
+			goto out;
+		}
+
+	printk(KERN_WARNING "No client context found for %s/%s\n",
+	       device->name, client->name);
+
+out:
+	spin_unlock_irqrestore(&device->client_data_lock, flags);
+}
+EXPORT_SYMBOL(ib_set_client_data);
+
+/**
+ * ib_register_event_handler - Register an IB event handler
+ * @event_handler:Handler to register
+ *
+ * ib_register_event_handler() registers an event handler that will be
+ * called back when asynchronous IB events occur (as defined in
+ * chapter 11 of the InfiniBand Architecture Specification).  This
+ * callback may occur in interrupt context.
+ */
+int ib_register_event_handler  (struct ib_event_handler *event_handler)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&event_handler->device->event_handler_lock, flags);
+	list_add_tail(&event_handler->list,
+		      &event_handler->device->event_handler_list);
+	spin_unlock_irqrestore(&event_handler->device->event_handler_lock, flags);
+
+	return 0;
+}
+EXPORT_SYMBOL(ib_register_event_handler);
+
+/**
+ * ib_unregister_event_handler - Unregister an event handler
+ * @event_handler:Handler to unregister
+ *
+ * Unregister an event handler registered with
+ * ib_register_event_handler().
+ */
+int ib_unregister_event_handler(struct ib_event_handler *event_handler)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&event_handler->device->event_handler_lock, flags);
+	list_del(&event_handler->list);
+	spin_unlock_irqrestore(&event_handler->device->event_handler_lock, flags);
+
+	return 0;
+}
+EXPORT_SYMBOL(ib_unregister_event_handler);
+
+/**
+ * ib_dispatch_event - Dispatch an asynchronous event
+ * @event:Event to dispatch
+ *
+ * Low-level drivers must call ib_dispatch_event() to dispatch the
+ * event to all registered event handlers when an asynchronous event
+ * occurs.
+ */
+void ib_dispatch_event(struct ib_event *event)
+{
+	unsigned long flags;
+	struct ib_event_handler *handler;
+
+	spin_lock_irqsave(&event->device->event_handler_lock, flags);
+
+	list_for_each_entry(handler, &event->device->event_handler_list, list)
+		handler->handler(handler, event);
+
+	spin_unlock_irqrestore(&event->device->event_handler_lock, flags);
+}
+EXPORT_SYMBOL(ib_dispatch_event);
+
+/**
+ * ib_query_device - Query IB device attributes
+ * @device:Device to query
+ * @device_attr:Device attributes
+ *
+ * ib_query_device() returns the attributes of a device through the
+ * @device_attr pointer.
+ */
+int ib_query_device(struct ib_device *device,
+		    struct ib_device_attr *device_attr)
+{
+	return device->query_device(device, device_attr);
+}
+EXPORT_SYMBOL(ib_query_device);
+
+/**
+ * ib_query_port - Query IB port attributes
+ * @device:Device to query
+ * @port_num:Port number to query
+ * @port_attr:Port attributes
+ *
+ * ib_query_port() returns the attributes of a port through the
+ * @port_attr pointer.
+ */
+int ib_query_port(struct ib_device *device,
+		  u8 port_num,
+		  struct ib_port_attr *port_attr)
+{
+	if (device->node_type == RDMA_NODE_IB_SWITCH) {
+		if (port_num)
+			return -EINVAL;
+	} else if (port_num < 1 || port_num > device->phys_port_cnt)
+		return -EINVAL;
+
+	return device->query_port(device, port_num, port_attr);
+}
+EXPORT_SYMBOL(ib_query_port);
+
+/**
+ * ib_query_gid - Get GID table entry
+ * @device:Device to query
+ * @port_num:Port number to query
+ * @index:GID table index to query
+ * @gid:Returned GID
+ *
+ * ib_query_gid() fetches the specified GID table entry.
+ */
+int ib_query_gid(struct ib_device *device,
+		 u8 port_num, int index, union ib_gid *gid)
+{
+	return device->query_gid(device, port_num, index, gid);
+}
+EXPORT_SYMBOL(ib_query_gid);
+
+/**
+ * ib_query_pkey - Get P_Key table entry
+ * @device:Device to query
+ * @port_num:Port number to query
+ * @index:P_Key table index to query
+ * @pkey:Returned P_Key
+ *
+ * ib_query_pkey() fetches the specified P_Key table entry.
+ */
+int ib_query_pkey(struct ib_device *device,
+		  u8 port_num, u16 index, u16 *pkey)
+{
+	return device->query_pkey(device, port_num, index, pkey);
+}
+EXPORT_SYMBOL(ib_query_pkey);
+
+/**
+ * ib_modify_device - Change IB device attributes
+ * @device:Device to modify
+ * @device_modify_mask:Mask of attributes to change
+ * @device_modify:New attribute values
+ *
+ * ib_modify_device() changes a device's attributes as specified by
+ * the @device_modify_mask and @device_modify structure.
+ */
+int ib_modify_device(struct ib_device *device,
+		     int device_modify_mask,
+		     struct ib_device_modify *device_modify)
+{
+	return device->modify_device(device, device_modify_mask,
+				     device_modify);
+}
+EXPORT_SYMBOL(ib_modify_device);
+
+/**
+ * ib_modify_port - Modifies the attributes for the specified port.
+ * @device: The device to modify.
+ * @port_num: The number of the port to modify.
+ * @port_modify_mask: Mask used to specify which attributes of the port
+ *   to change.
+ * @port_modify: New attribute values for the port.
+ *
+ * ib_modify_port() changes a port's attributes as specified by the
+ * @port_modify_mask and @port_modify structure.
+ */
+int ib_modify_port(struct ib_device *device,
+		   u8 port_num, int port_modify_mask,
+		   struct ib_port_modify *port_modify)
+{
+	if (device->node_type == RDMA_NODE_IB_SWITCH) {
+		if (port_num)
+			return -EINVAL;
+	} else if (port_num < 1 || port_num > device->phys_port_cnt)
+		return -EINVAL;
+
+	return device->modify_port(device, port_num, port_modify_mask,
+				   port_modify);
+}
+EXPORT_SYMBOL(ib_modify_port);
+
+static int __init ib_core_init(void)
+{
+	int ret;
+
+	ret = ib_sysfs_setup();
+	if (ret)
+		printk(KERN_WARNING "Couldn't create InfiniBand device class\n");
+
+	ret = ib_cache_setup();
+	if (ret) {
+		printk(KERN_WARNING "Couldn't set up InfiniBand P_Key/GID cache\n");
+		ib_sysfs_cleanup();
+	}
+
+	return ret;
+}
+
+static void __exit ib_core_cleanup(void)
+{
+	ib_cache_cleanup();
+	ib_sysfs_cleanup();
+}
+
+module_init(ib_core_init);
+module_exit(ib_core_cleanup);

drivers/infiniband/core/fmr_pool.c

@@ -0,0 +1,527 @@
+/*
+ * Copyright (c) 2004 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. 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.
+ *
+ * $Id: fmr_pool.c,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#include <linux/errno.h>
+#include <linux/spinlock.h>
+#include <linux/slab.h>
+#include <linux/jhash.h>
+#include <linux/kthread.h>
+
+#include <rdma/ib_fmr_pool.h>
+
+#include "core_priv.h"
+
+enum {
+	IB_FMR_MAX_REMAPS = 32,
+
+	IB_FMR_HASH_BITS  = 8,
+	IB_FMR_HASH_SIZE  = 1 << IB_FMR_HASH_BITS,
+	IB_FMR_HASH_MASK  = IB_FMR_HASH_SIZE - 1
+};
+
+/*
+ * 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 < 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
+ * either free_list or dirty_list.
+ *
+ * The cache_node member is used to link the FMR into a cache bucket
+ * (if caching is enabled).  This is independent of the reference
+ * count of the FMR.  When a valid FMR is released, its ref_count is
+ * decremented, and if ref_count reaches 0, the FMR is placed in
+ * either free_list or dirty_list as appropriate.  However, it is not
+ * removed from the cache and may be "revived" if a call to
+ * ib_fmr_register_physical() occurs before the FMR is remapped.  In
+ * this case we just increment the ref_count and remove the FMR from
+ * free_list/dirty_list.
+ *
+ * Before we remap an FMR from free_list, we remove it from the cache
+ * (to prevent another user from obtaining a stale FMR).  When an FMR
+ * is released, we add it to the tail of the free list, so that our
+ * cache eviction policy is "least recently used."
+ *
+ * All manipulation of ref_count, list and cache_node is protected by
+ * pool_lock to maintain consistency.
+ */
+
+struct ib_fmr_pool {
+	spinlock_t                pool_lock;
+
+	int                       pool_size;
+	int                       max_pages;
+	int			  max_remaps;
+	int                       dirty_watermark;
+	int                       dirty_len;
+	struct list_head          free_list;
+	struct list_head          dirty_list;
+	struct hlist_head        *cache_bucket;
+
+	void                     (*flush_function)(struct ib_fmr_pool *pool,
+						   void *              arg);
+	void                     *flush_arg;
+
+	struct task_struct       *thread;
+
+	atomic_t                  req_ser;
+	atomic_t                  flush_ser;
+
+	wait_queue_head_t         force_wait;
+};
+
+static inline u32 ib_fmr_hash(u64 first_page)
+{
+	return jhash_2words((u32) first_page, (u32) (first_page >> 32), 0) &
+		(IB_FMR_HASH_SIZE - 1);
+}
+
+/* Caller must hold pool_lock */
+static inline struct ib_pool_fmr *ib_fmr_cache_lookup(struct ib_fmr_pool *pool,
+						      u64 *page_list,
+						      int  page_list_len,
+						      u64  io_virtual_address)
+{
+	struct hlist_head *bucket;
+	struct ib_pool_fmr *fmr;
+	struct hlist_node *pos;
+
+	if (!pool->cache_bucket)
+		return NULL;
+
+	bucket = pool->cache_bucket + ib_fmr_hash(*page_list);
+
+	hlist_for_each_entry(fmr, pos, bucket, cache_node)
+		if (io_virtual_address == fmr->io_virtual_address &&
+		    page_list_len      == fmr->page_list_len      &&
+		    !memcmp(page_list, fmr->page_list,
+			    page_list_len * sizeof *page_list))
+			return fmr;
+
+	return NULL;
+}
+
+static void ib_fmr_batch_release(struct ib_fmr_pool *pool)
+{
+	int                 ret;
+	struct ib_pool_fmr *fmr;
+	LIST_HEAD(unmap_list);
+	LIST_HEAD(fmr_list);
+
+	spin_lock_irq(&pool->pool_lock);
+
+	list_for_each_entry(fmr, &pool->dirty_list, list) {
+		hlist_del_init(&fmr->cache_node);
+		fmr->remap_count = 0;
+		list_add_tail(&fmr->fmr->list, &fmr_list);
+
+#ifdef DEBUG
+		if (fmr->ref_count !=0) {
+			printk(KERN_WARNING "Unmapping FMR 0x%08x with ref count %d",
+			       fmr, fmr->ref_count);
+		}
+#endif
+	}
+
+	list_splice(&pool->dirty_list, &unmap_list);
+	INIT_LIST_HEAD(&pool->dirty_list);
+	pool->dirty_len = 0;
+
+	spin_unlock_irq(&pool->pool_lock);
+
+	if (list_empty(&unmap_list)) {
+		return;
+	}
+
+	ret = ib_unmap_fmr(&fmr_list);
+	if (ret)
+		printk(KERN_WARNING "ib_unmap_fmr returned %d", ret);
+
+	spin_lock_irq(&pool->pool_lock);
+	list_splice(&unmap_list, &pool->free_list);
+	spin_unlock_irq(&pool->pool_lock);
+}
+
+static int ib_fmr_cleanup_thread(void *pool_ptr)
+{
+	struct ib_fmr_pool *pool = pool_ptr;
+
+	do {
+		if (pool->dirty_len >= pool->dirty_watermark ||
+		    atomic_read(&pool->flush_ser) - atomic_read(&pool->req_ser) < 0) {
+			ib_fmr_batch_release(pool);
+
+			atomic_inc(&pool->flush_ser);
+			wake_up_interruptible(&pool->force_wait);
+
+			if (pool->flush_function)
+				pool->flush_function(pool, pool->flush_arg);
+		}
+
+		set_current_state(TASK_INTERRUPTIBLE);
+		if (pool->dirty_len < pool->dirty_watermark &&
+		    atomic_read(&pool->flush_ser) - atomic_read(&pool->req_ser) >= 0 &&
+		    !kthread_should_stop())
+			schedule();
+		__set_current_state(TASK_RUNNING);
+	} while (!kthread_should_stop());
+
+	return 0;
+}
+
+/**
+ * ib_create_fmr_pool - Create an FMR pool
+ * @pd:Protection domain for FMRs
+ * @params:FMR pool parameters
+ *
+ * Create a pool of FMRs.  Return value is pointer to new pool or
+ * error code if creation failed.
+ */
+struct ib_fmr_pool *ib_create_fmr_pool(struct ib_pd             *pd,
+				       struct ib_fmr_pool_param *params)
+{
+	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);
+
+	device = pd->device;
+	if (!device->alloc_fmr    || !device->dealloc_fmr  ||
+	    !device->map_phys_fmr || !device->unmap_fmr) {
+		printk(KERN_WARNING "Device %s does not support fast memory regions",
+		       device->name);
+		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");
+		return ERR_PTR(-ENOMEM);
+	}
+
+	pool->cache_bucket   = NULL;
+
+	pool->flush_function = params->flush_function;
+	pool->flush_arg      = params->flush_arg;
+
+	INIT_LIST_HEAD(&pool->free_list);
+	INIT_LIST_HEAD(&pool->dirty_list);
+
+	if (params->cache) {
+		pool->cache_bucket =
+			kmalloc(IB_FMR_HASH_SIZE * sizeof *pool->cache_bucket,
+				GFP_KERNEL);
+		if (!pool->cache_bucket) {
+			printk(KERN_WARNING "Failed to allocate cache in pool");
+			ret = -ENOMEM;
+			goto out_free_pool;
+		}
+
+		for (i = 0; i < IB_FMR_HASH_SIZE; ++i)
+			INIT_HLIST_HEAD(pool->cache_bucket + i);
+	}
+
+	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);
+	atomic_set(&pool->req_ser,   0);
+	atomic_set(&pool->flush_ser, 0);
+	init_waitqueue_head(&pool->force_wait);
+
+	pool->thread = kthread_create(ib_fmr_cleanup_thread,
+				      pool,
+				      "ib_fmr(%s)",
+				      device->name);
+	if (IS_ERR(pool->thread)) {
+		printk(KERN_WARNING "couldn't start cleanup thread");
+		ret = PTR_ERR(pool->thread);
+		goto out_free_pool;
+	}
+
+	{
+		struct ib_pool_fmr *fmr;
+		struct ib_fmr_attr fmr_attr = {
+			.max_pages  = params->max_pages_per_fmr,
+			.max_maps   = pool->max_remaps,
+			.page_shift = params->page_shift
+		};
+
+		for (i = 0; i < params->pool_size; ++i) {
+			fmr = kmalloc(sizeof *fmr + params->max_pages_per_fmr * sizeof (u64),
+				      GFP_KERNEL);
+			if (!fmr) {
+				printk(KERN_WARNING "failed to allocate fmr struct "
+				       "for FMR %d", i);
+				goto out_fail;
+			}
+
+			fmr->pool             = pool;
+			fmr->remap_count      = 0;
+			fmr->ref_count        = 0;
+			INIT_HLIST_NODE(&fmr->cache_node);
+
+			fmr->fmr = ib_alloc_fmr(pd, params->access, &fmr_attr);
+			if (IS_ERR(fmr->fmr)) {
+				printk(KERN_WARNING "fmr_create failed for FMR %d", i);
+				kfree(fmr);
+				goto out_fail;
+			}
+
+			list_add_tail(&fmr->list, &pool->free_list);
+			++pool->pool_size;
+		}
+	}
+
+	return pool;
+
+ out_free_pool:
+	kfree(pool->cache_bucket);
+	kfree(pool);
+
+	return ERR_PTR(ret);
+
+ out_fail:
+	ib_destroy_fmr_pool(pool);
+
+	return ERR_PTR(-ENOMEM);
+}
+EXPORT_SYMBOL(ib_create_fmr_pool);
+
+/**
+ * ib_destroy_fmr_pool - Free FMR pool
+ * @pool:FMR pool to free
+ *
+ * Destroy an FMR pool and free all associated resources.
+ */
+void ib_destroy_fmr_pool(struct ib_fmr_pool *pool)
+{
+	struct ib_pool_fmr *fmr;
+	struct ib_pool_fmr *tmp;
+	LIST_HEAD(fmr_list);
+	int                 i;
+
+	kthread_stop(pool->thread);
+	ib_fmr_batch_release(pool);
+
+	i = 0;
+	list_for_each_entry_safe(fmr, tmp, &pool->free_list, list) {
+		if (fmr->remap_count) {
+			INIT_LIST_HEAD(&fmr_list);
+			list_add_tail(&fmr->fmr->list, &fmr_list);
+			ib_unmap_fmr(&fmr_list);
+		}
+		ib_dealloc_fmr(fmr->fmr);
+		list_del(&fmr->list);
+		kfree(fmr);
+		++i;
+	}
+
+	if (i < pool->pool_size)
+		printk(KERN_WARNING "pool still has %d regions registered",
+		       pool->pool_size - i);
+
+	kfree(pool->cache_bucket);
+	kfree(pool);
+}
+EXPORT_SYMBOL(ib_destroy_fmr_pool);
+
+/**
+ * ib_flush_fmr_pool - Invalidate all unmapped FMRs
+ * @pool:FMR pool to flush
+ *
+ * Ensure that all unmapped FMRs are fully invalidated.
+ */
+int ib_flush_fmr_pool(struct ib_fmr_pool *pool)
+{
+	int serial = atomic_inc_return(&pool->req_ser);
+
+	wake_up_process(pool->thread);
+
+	if (wait_event_interruptible(pool->force_wait,
+				     atomic_read(&pool->flush_ser) - serial >= 0))
+		return -EINTR;
+
+	return 0;
+}
+EXPORT_SYMBOL(ib_flush_fmr_pool);
+
+/**
+ * ib_fmr_pool_map_phys -
+ * @pool:FMR pool to allocate FMR from
+ * @page_list:List of pages to map
+ * @list_len:Number of pages in @page_list
+ * @io_virtual_address:I/O virtual address for new FMR
+ *
+ * Map an FMR from an FMR pool.
+ */
+struct ib_pool_fmr *ib_fmr_pool_map_phys(struct ib_fmr_pool *pool_handle,
+					 u64                *page_list,
+					 int                 list_len,
+					 u64                 io_virtual_address)
+{
+	struct ib_fmr_pool *pool = pool_handle;
+	struct ib_pool_fmr *fmr;
+	unsigned long       flags;
+	int                 result;
+
+	if (list_len < 1 || list_len > pool->max_pages)
+		return ERR_PTR(-EINVAL);
+
+	spin_lock_irqsave(&pool->pool_lock, flags);
+	fmr = ib_fmr_cache_lookup(pool,
+				  page_list,
+				  list_len,
+				  io_virtual_address);
+	if (fmr) {
+		/* found in cache */
+		++fmr->ref_count;
+		if (fmr->ref_count == 1) {
+			list_del(&fmr->list);
+		}
+
+		spin_unlock_irqrestore(&pool->pool_lock, flags);
+
+		return fmr;
+	}
+
+	if (list_empty(&pool->free_list)) {
+		spin_unlock_irqrestore(&pool->pool_lock, flags);
+		return ERR_PTR(-EAGAIN);
+	}
+
+	fmr = list_entry(pool->free_list.next, struct ib_pool_fmr, list);
+	list_del(&fmr->list);
+	hlist_del_init(&fmr->cache_node);
+	spin_unlock_irqrestore(&pool->pool_lock, flags);
+
+	result = ib_map_phys_fmr(fmr->fmr, page_list, list_len,
+				 io_virtual_address);
+
+	if (result) {
+		spin_lock_irqsave(&pool->pool_lock, flags);
+		list_add(&fmr->list, &pool->free_list);
+		spin_unlock_irqrestore(&pool->pool_lock, flags);
+
+		printk(KERN_WARNING "fmr_map returns %d\n",
+		       result);
+
+		return ERR_PTR(result);
+	}
+
+	++fmr->remap_count;
+	fmr->ref_count = 1;
+
+	if (pool->cache_bucket) {
+		fmr->io_virtual_address = io_virtual_address;
+		fmr->page_list_len      = list_len;
+		memcpy(fmr->page_list, page_list, list_len * sizeof(*page_list));
+
+		spin_lock_irqsave(&pool->pool_lock, flags);
+		hlist_add_head(&fmr->cache_node,
+			       pool->cache_bucket + ib_fmr_hash(fmr->page_list[0]));
+		spin_unlock_irqrestore(&pool->pool_lock, flags);
+	}
+
+	return fmr;
+}
+EXPORT_SYMBOL(ib_fmr_pool_map_phys);
+
+/**
+ * ib_fmr_pool_unmap - Unmap FMR
+ * @fmr:FMR to unmap
+ *
+ * Unmap an FMR.  The FMR mapping may remain valid until the FMR is
+ * reused (or until ib_flush_fmr_pool() is called).
+ */
+int ib_fmr_pool_unmap(struct ib_pool_fmr *fmr)
+{
+	struct ib_fmr_pool *pool;
+	unsigned long flags;
+
+	pool = fmr->pool;
+
+	spin_lock_irqsave(&pool->pool_lock, flags);
+
+	--fmr->ref_count;
+	if (!fmr->ref_count) {
+		if (fmr->remap_count < pool->max_remaps) {
+			list_add_tail(&fmr->list, &pool->free_list);
+		} else {
+			list_add_tail(&fmr->list, &pool->dirty_list);
+			++pool->dirty_len;
+			wake_up_process(pool->thread);
+		}
+	}
+
+#ifdef DEBUG
+	if (fmr->ref_count < 0)
+		printk(KERN_WARNING "FMR %p has ref count %d < 0",
+		       fmr, fmr->ref_count);
+#endif
+
+	spin_unlock_irqrestore(&pool->pool_lock, flags);
+
+	return 0;
+}
+EXPORT_SYMBOL(ib_fmr_pool_unmap);

drivers/infiniband/core/iwcm.h

@@ -0,0 +1,62 @@
+/*
+ * Copyright (c) 2005 Network Appliance, Inc. All rights reserved.
+ * Copyright (c) 2005 Open Grid Computing, Inc. 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.
+ */
+#ifndef IWCM_H
+#define IWCM_H
+
+enum iw_cm_state {
+	IW_CM_STATE_IDLE,             /* unbound, inactive */
+	IW_CM_STATE_LISTEN,           /* listen waiting for connect */
+	IW_CM_STATE_CONN_RECV,        /* inbound waiting for user accept */
+	IW_CM_STATE_CONN_SENT,        /* outbound waiting for peer accept */
+	IW_CM_STATE_ESTABLISHED,      /* established */
+	IW_CM_STATE_CLOSING,	      /* disconnect */
+	IW_CM_STATE_DESTROYING        /* object being deleted */
+};
+
+struct iwcm_id_private {
+	struct iw_cm_id	id;
+	enum iw_cm_state state;
+	unsigned long flags;
+	struct ib_qp *qp;
+	struct completion destroy_comp;
+	wait_queue_head_t connect_wait;
+	struct list_head work_list;
+	spinlock_t lock;
+	atomic_t refcount;
+	struct list_head work_free_list;
+};
+
+#define IWCM_F_CALLBACK_DESTROY   1
+#define IWCM_F_CONNECT_WAIT       2
+
+#endif /* IWCM_H */

drivers/infiniband/core/mad_priv.h

@@ -0,0 +1,229 @@
+/*
+ * Copyright (c) 2004, 2005, Voltaire, Inc. All rights reserved.
+ * Copyright (c) 2005 Intel Corporation. All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. 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.
+ *
+ * $Id: mad_priv.h,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#ifndef __IB_MAD_PRIV_H__
+#define __IB_MAD_PRIV_H__
+
+#include <linux/completion.h>
+#include <linux/err.h>
+#include <linux/pci.h>
+#include <linux/workqueue.h>
+#include <rdma/ib_mad.h>
+#include <rdma/ib_smi.h>
+
+
+#define PFX "ib_mad: "
+
+#define IB_MAD_QPS_CORE		2 /* Always QP0 and QP1 as a minimum */
+
+/* QP and CQ parameters */
+#define IB_MAD_QP_SEND_SIZE	128
+#define IB_MAD_QP_RECV_SIZE	512
+#define IB_MAD_SEND_REQ_MAX_SG	2
+#define IB_MAD_RECV_REQ_MAX_SG	1
+
+#define IB_MAD_SEND_Q_PSN	0
+
+/* Registration table sizes */
+#define MAX_MGMT_CLASS		80
+#define MAX_MGMT_VERSION	8
+#define MAX_MGMT_OUI		8
+#define MAX_MGMT_VENDOR_RANGE2	(IB_MGMT_CLASS_VENDOR_RANGE2_END - \
+				IB_MGMT_CLASS_VENDOR_RANGE2_START + 1)
+
+struct ib_mad_list_head {
+	struct list_head list;
+	struct ib_mad_queue *mad_queue;
+};
+
+struct ib_mad_private_header {
+	struct ib_mad_list_head mad_list;
+	struct ib_mad_recv_wc recv_wc;
+	struct ib_wc wc;
+	u64 mapping;
+} __attribute__ ((packed));
+
+struct ib_mad_private {
+	struct ib_mad_private_header header;
+	struct ib_grh grh;
+	union {
+		struct ib_mad mad;
+		struct ib_rmpp_mad rmpp_mad;
+		struct ib_smp smp;
+	} mad;
+} __attribute__ ((packed));
+
+struct ib_rmpp_segment {
+	struct list_head list;
+	u32 num;
+	u8 data[0];
+};
+
+struct ib_mad_agent_private {
+	struct list_head agent_list;
+	struct ib_mad_agent agent;
+	struct ib_mad_reg_req *reg_req;
+	struct ib_mad_qp_info *qp_info;
+
+	spinlock_t lock;
+	struct list_head send_list;
+	struct list_head wait_list;
+	struct list_head done_list;
+	struct delayed_work timed_work;
+	unsigned long timeout;
+	struct list_head local_list;
+	struct work_struct local_work;
+	struct list_head rmpp_list;
+
+	atomic_t refcount;
+	struct completion comp;
+};
+
+struct ib_mad_snoop_private {
+	struct ib_mad_agent agent;
+	struct ib_mad_qp_info *qp_info;
+	int snoop_index;
+	int mad_snoop_flags;
+	atomic_t refcount;
+	struct completion comp;
+};
+
+struct ib_mad_send_wr_private {
+	struct ib_mad_list_head mad_list;
+	struct list_head agent_list;
+	struct ib_mad_agent_private *mad_agent_priv;
+	struct ib_mad_send_buf send_buf;
+	u64 header_mapping;
+	u64 payload_mapping;
+	struct ib_send_wr send_wr;
+	struct ib_sge sg_list[IB_MAD_SEND_REQ_MAX_SG];
+	__be64 tid;
+	unsigned long timeout;
+	int retries;
+	int retry;
+	int refcount;
+	enum ib_wc_status status;
+
+	/* RMPP control */
+	struct list_head rmpp_list;
+	struct ib_rmpp_segment *last_ack_seg;
+	struct ib_rmpp_segment *cur_seg;
+	int last_ack;
+	int seg_num;
+	int newwin;
+	int pad;
+};
+
+struct ib_mad_local_private {
+	struct list_head completion_list;
+	struct ib_mad_private *mad_priv;
+	struct ib_mad_agent_private *recv_mad_agent;
+	struct ib_mad_send_wr_private *mad_send_wr;
+};
+
+struct ib_mad_mgmt_method_table {
+	struct ib_mad_agent_private *agent[IB_MGMT_MAX_METHODS];
+};
+
+struct ib_mad_mgmt_class_table {
+	struct ib_mad_mgmt_method_table *method_table[MAX_MGMT_CLASS];
+};
+
+struct ib_mad_mgmt_vendor_class {
+	u8	oui[MAX_MGMT_OUI][3];
+	struct ib_mad_mgmt_method_table *method_table[MAX_MGMT_OUI];
+};
+
+struct ib_mad_mgmt_vendor_class_table {
+	struct ib_mad_mgmt_vendor_class *vendor_class[MAX_MGMT_VENDOR_RANGE2];
+};
+
+struct ib_mad_mgmt_version_table {
+	struct ib_mad_mgmt_class_table *class;
+	struct ib_mad_mgmt_vendor_class_table *vendor;
+};
+
+struct ib_mad_queue {
+	spinlock_t lock;
+	struct list_head list;
+	int count;
+	int max_active;
+	struct ib_mad_qp_info *qp_info;
+};
+
+struct ib_mad_qp_info {
+	struct ib_mad_port_private *port_priv;
+	struct ib_qp *qp;
+	struct ib_mad_queue send_queue;
+	struct ib_mad_queue recv_queue;
+	struct list_head overflow_list;
+	spinlock_t snoop_lock;
+	struct ib_mad_snoop_private **snoop_table;
+	int snoop_table_size;
+	atomic_t snoop_count;
+};
+
+struct ib_mad_port_private {
+	struct list_head port_list;
+	struct ib_device *device;
+	int port_num;
+	struct ib_cq *cq;
+	struct ib_pd *pd;
+	struct ib_mr *mr;
+
+	spinlock_t reg_lock;
+	struct ib_mad_mgmt_version_table version[MAX_MGMT_VERSION];
+	struct list_head agent_list;
+	struct workqueue_struct *wq;
+	struct work_struct work;
+	struct ib_mad_qp_info qp_info[IB_MAD_QPS_CORE];
+};
+
+int ib_send_mad(struct ib_mad_send_wr_private *mad_send_wr);
+
+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);
+
+void ib_mad_complete_send_wr(struct ib_mad_send_wr_private *mad_send_wr,
+			     struct ib_mad_send_wc *mad_send_wc);
+
+void ib_mark_mad_done(struct ib_mad_send_wr_private *mad_send_wr);
+
+void ib_reset_mad_timeout(struct ib_mad_send_wr_private *mad_send_wr,
+			  int timeout_ms);
+
+#endif	/* __IB_MAD_PRIV_H__ */

drivers/infiniband/core/mad_rmpp.c

@@ -0,0 +1,944 @@
+/*
+ * Copyright (c) 2005 Intel Inc. All rights reserved.
+ * Copyright (c) 2005-2006 Voltaire, Inc. 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.
+ *
+ * $Id: mad_rmpp.c,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#include "mad_priv.h"
+#include "mad_rmpp.h"
+
+enum rmpp_state {
+	RMPP_STATE_ACTIVE,
+	RMPP_STATE_TIMEOUT,
+	RMPP_STATE_COMPLETE
+};
+
+struct mad_rmpp_recv {
+	struct ib_mad_agent_private *agent;
+	struct list_head list;
+	struct delayed_work timeout_work;
+	struct delayed_work cleanup_work;
+	struct completion comp;
+	enum rmpp_state state;
+	spinlock_t lock;
+	atomic_t refcount;
+
+	struct ib_ah *ah;
+	struct ib_mad_recv_wc *rmpp_wc;
+	struct ib_mad_recv_buf *cur_seg_buf;
+	int last_ack;
+	int seg_num;
+	int newwin;
+	int repwin;
+
+	__be64 tid;
+	u32 src_qp;
+	u16 slid;
+	u8 mgmt_class;
+	u8 class_version;
+	u8 method;
+};
+
+static inline void deref_rmpp_recv(struct mad_rmpp_recv *rmpp_recv)
+{
+	if (atomic_dec_and_test(&rmpp_recv->refcount))
+		complete(&rmpp_recv->comp);
+}
+
+static void destroy_rmpp_recv(struct mad_rmpp_recv *rmpp_recv)
+{
+	deref_rmpp_recv(rmpp_recv);
+	wait_for_completion(&rmpp_recv->comp);
+	ib_destroy_ah(rmpp_recv->ah);
+	kfree(rmpp_recv);
+}
+
+void ib_cancel_rmpp_recvs(struct ib_mad_agent_private *agent)
+{
+	struct mad_rmpp_recv *rmpp_recv, *temp_rmpp_recv;
+	unsigned long flags;
+
+	spin_lock_irqsave(&agent->lock, flags);
+	list_for_each_entry(rmpp_recv, &agent->rmpp_list, list) {
+		cancel_delayed_work(&rmpp_recv->timeout_work);
+		cancel_delayed_work(&rmpp_recv->cleanup_work);
+	}
+	spin_unlock_irqrestore(&agent->lock, flags);
+
+	flush_workqueue(agent->qp_info->port_priv->wq);
+
+	list_for_each_entry_safe(rmpp_recv, temp_rmpp_recv,
+				 &agent->rmpp_list, list) {
+		list_del(&rmpp_recv->list);
+		if (rmpp_recv->state != RMPP_STATE_COMPLETE)
+			ib_free_recv_mad(rmpp_recv->rmpp_wc);
+		destroy_rmpp_recv(rmpp_recv);
+	}
+}
+
+static void format_ack(struct ib_mad_send_buf *msg,
+		       struct ib_rmpp_mad *data,
+		       struct mad_rmpp_recv *rmpp_recv)
+{
+	struct ib_rmpp_mad *ack = msg->mad;
+	unsigned long flags;
+
+	memcpy(ack, &data->mad_hdr, msg->hdr_len);
+
+	ack->mad_hdr.method ^= IB_MGMT_METHOD_RESP;
+	ack->rmpp_hdr.rmpp_type = IB_MGMT_RMPP_TYPE_ACK;
+	ib_set_rmpp_flags(&ack->rmpp_hdr, IB_MGMT_RMPP_FLAG_ACTIVE);
+
+	spin_lock_irqsave(&rmpp_recv->lock, flags);
+	rmpp_recv->last_ack = rmpp_recv->seg_num;
+	ack->rmpp_hdr.seg_num = cpu_to_be32(rmpp_recv->seg_num);
+	ack->rmpp_hdr.paylen_newwin = cpu_to_be32(rmpp_recv->newwin);
+	spin_unlock_irqrestore(&rmpp_recv->lock, flags);
+}
+
+static void ack_recv(struct mad_rmpp_recv *rmpp_recv,
+		     struct ib_mad_recv_wc *recv_wc)
+{
+	struct ib_mad_send_buf *msg;
+	int ret, hdr_len;
+
+	hdr_len = ib_get_mad_data_offset(recv_wc->recv_buf.mad->mad_hdr.mgmt_class);
+	msg = ib_create_send_mad(&rmpp_recv->agent->agent, recv_wc->wc->src_qp,
+				 recv_wc->wc->pkey_index, 1, hdr_len,
+				 0, GFP_KERNEL);
+	if (!msg)
+		return;
+
+	format_ack(msg, (struct ib_rmpp_mad *) recv_wc->recv_buf.mad, rmpp_recv);
+	msg->ah = rmpp_recv->ah;
+	ret = ib_post_send_mad(msg, NULL);
+	if (ret)
+		ib_free_send_mad(msg);
+}
+
+static struct ib_mad_send_buf *alloc_response_msg(struct ib_mad_agent *agent,
+						  struct ib_mad_recv_wc *recv_wc)
+{
+	struct ib_mad_send_buf *msg;
+	struct ib_ah *ah;
+	int hdr_len;
+
+	ah = ib_create_ah_from_wc(agent->qp->pd, recv_wc->wc,
+				  recv_wc->recv_buf.grh, agent->port_num);
+	if (IS_ERR(ah))
+		return (void *) ah;
+
+	hdr_len = ib_get_mad_data_offset(recv_wc->recv_buf.mad->mad_hdr.mgmt_class);
+	msg = ib_create_send_mad(agent, recv_wc->wc->src_qp,
+				 recv_wc->wc->pkey_index, 1,
+				 hdr_len, 0, GFP_KERNEL);
+	if (IS_ERR(msg))
+		ib_destroy_ah(ah);
+	else
+		msg->ah = ah;
+
+	return msg;
+}
+
+static void ack_ds_ack(struct ib_mad_agent_private *agent,
+		       struct ib_mad_recv_wc *recv_wc)
+{
+	struct ib_mad_send_buf *msg;
+	struct ib_rmpp_mad *rmpp_mad;
+	int ret;
+
+	msg = alloc_response_msg(&agent->agent, recv_wc);
+	if (IS_ERR(msg))
+		return;
+
+	rmpp_mad = msg->mad;
+	memcpy(rmpp_mad, recv_wc->recv_buf.mad, msg->hdr_len);
+
+	rmpp_mad->mad_hdr.method ^= IB_MGMT_METHOD_RESP;
+	ib_set_rmpp_flags(&rmpp_mad->rmpp_hdr, IB_MGMT_RMPP_FLAG_ACTIVE);
+	rmpp_mad->rmpp_hdr.seg_num = 0;
+	rmpp_mad->rmpp_hdr.paylen_newwin = cpu_to_be32(1);
+
+	ret = ib_post_send_mad(msg, NULL);
+	if (ret) {
+		ib_destroy_ah(msg->ah);
+		ib_free_send_mad(msg);
+	}
+}
+
+void ib_rmpp_send_handler(struct ib_mad_send_wc *mad_send_wc)
+{
+	struct ib_rmpp_mad *rmpp_mad = mad_send_wc->send_buf->mad;
+
+	if (rmpp_mad->rmpp_hdr.rmpp_type != IB_MGMT_RMPP_TYPE_ACK)
+		ib_destroy_ah(mad_send_wc->send_buf->ah);
+	ib_free_send_mad(mad_send_wc->send_buf);
+}
+
+static void nack_recv(struct ib_mad_agent_private *agent,
+		      struct ib_mad_recv_wc *recv_wc, u8 rmpp_status)
+{
+	struct ib_mad_send_buf *msg;
+	struct ib_rmpp_mad *rmpp_mad;
+	int ret;
+
+	msg = alloc_response_msg(&agent->agent, recv_wc);
+	if (IS_ERR(msg))
+		return;
+
+	rmpp_mad = msg->mad;
+	memcpy(rmpp_mad, recv_wc->recv_buf.mad, msg->hdr_len);
+
+	rmpp_mad->mad_hdr.method ^= IB_MGMT_METHOD_RESP;
+	rmpp_mad->rmpp_hdr.rmpp_version = IB_MGMT_RMPP_VERSION;
+	rmpp_mad->rmpp_hdr.rmpp_type = IB_MGMT_RMPP_TYPE_ABORT;
+	ib_set_rmpp_flags(&rmpp_mad->rmpp_hdr, IB_MGMT_RMPP_FLAG_ACTIVE);
+	rmpp_mad->rmpp_hdr.rmpp_status = rmpp_status;
+	rmpp_mad->rmpp_hdr.seg_num = 0;
+	rmpp_mad->rmpp_hdr.paylen_newwin = 0;
+
+	ret = ib_post_send_mad(msg, NULL);
+	if (ret) {
+		ib_destroy_ah(msg->ah);
+		ib_free_send_mad(msg);
+	}
+}
+
+static void recv_timeout_handler(struct work_struct *work)
+{
+	struct mad_rmpp_recv *rmpp_recv =
+		container_of(work, struct mad_rmpp_recv, timeout_work.work);
+	struct ib_mad_recv_wc *rmpp_wc;
+	unsigned long flags;
+
+	spin_lock_irqsave(&rmpp_recv->agent->lock, flags);
+	if (rmpp_recv->state != RMPP_STATE_ACTIVE) {
+		spin_unlock_irqrestore(&rmpp_recv->agent->lock, flags);
+		return;
+	}
+	rmpp_recv->state = RMPP_STATE_TIMEOUT;
+	list_del(&rmpp_recv->list);
+	spin_unlock_irqrestore(&rmpp_recv->agent->lock, flags);
+
+	rmpp_wc = rmpp_recv->rmpp_wc;
+	nack_recv(rmpp_recv->agent, rmpp_wc, IB_MGMT_RMPP_STATUS_T2L);
+	destroy_rmpp_recv(rmpp_recv);
+	ib_free_recv_mad(rmpp_wc);
+}
+
+static void recv_cleanup_handler(struct work_struct *work)
+{
+	struct mad_rmpp_recv *rmpp_recv =
+		container_of(work, struct mad_rmpp_recv, cleanup_work.work);
+	unsigned long flags;
+
+	spin_lock_irqsave(&rmpp_recv->agent->lock, flags);
+	list_del(&rmpp_recv->list);
+	spin_unlock_irqrestore(&rmpp_recv->agent->lock, flags);
+	destroy_rmpp_recv(rmpp_recv);
+}
+
+static struct mad_rmpp_recv *
+create_rmpp_recv(struct ib_mad_agent_private *agent,
+		 struct ib_mad_recv_wc *mad_recv_wc)
+{
+	struct mad_rmpp_recv *rmpp_recv;
+	struct ib_mad_hdr *mad_hdr;
+
+	rmpp_recv = kmalloc(sizeof *rmpp_recv, GFP_KERNEL);
+	if (!rmpp_recv)
+		return NULL;
+
+	rmpp_recv->ah = ib_create_ah_from_wc(agent->agent.qp->pd,
+					     mad_recv_wc->wc,
+					     mad_recv_wc->recv_buf.grh,
+					     agent->agent.port_num);
+	if (IS_ERR(rmpp_recv->ah))
+		goto error;
+
+	rmpp_recv->agent = agent;
+	init_completion(&rmpp_recv->comp);
+	INIT_DELAYED_WORK(&rmpp_recv->timeout_work, recv_timeout_handler);
+	INIT_DELAYED_WORK(&rmpp_recv->cleanup_work, recv_cleanup_handler);
+	spin_lock_init(&rmpp_recv->lock);
+	rmpp_recv->state = RMPP_STATE_ACTIVE;
+	atomic_set(&rmpp_recv->refcount, 1);
+
+	rmpp_recv->rmpp_wc = mad_recv_wc;
+	rmpp_recv->cur_seg_buf = &mad_recv_wc->recv_buf;
+	rmpp_recv->newwin = 1;
+	rmpp_recv->seg_num = 1;
+	rmpp_recv->last_ack = 0;
+	rmpp_recv->repwin = 1;
+
+	mad_hdr = &mad_recv_wc->recv_buf.mad->mad_hdr;
+	rmpp_recv->tid = mad_hdr->tid;
+	rmpp_recv->src_qp = mad_recv_wc->wc->src_qp;
+	rmpp_recv->slid = mad_recv_wc->wc->slid;
+	rmpp_recv->mgmt_class = mad_hdr->mgmt_class;
+	rmpp_recv->class_version = mad_hdr->class_version;
+	rmpp_recv->method  = mad_hdr->method;
+	return rmpp_recv;
+
+error:	kfree(rmpp_recv);
+	return NULL;
+}
+
+static struct mad_rmpp_recv *
+find_rmpp_recv(struct ib_mad_agent_private *agent,
+	       struct ib_mad_recv_wc *mad_recv_wc)
+{
+	struct mad_rmpp_recv *rmpp_recv;
+	struct ib_mad_hdr *mad_hdr = &mad_recv_wc->recv_buf.mad->mad_hdr;
+
+	list_for_each_entry(rmpp_recv, &agent->rmpp_list, list) {
+		if (rmpp_recv->tid == mad_hdr->tid &&
+		    rmpp_recv->src_qp == mad_recv_wc->wc->src_qp &&
+		    rmpp_recv->slid == mad_recv_wc->wc->slid &&
+		    rmpp_recv->mgmt_class == mad_hdr->mgmt_class &&
+		    rmpp_recv->class_version == mad_hdr->class_version &&
+		    rmpp_recv->method == mad_hdr->method)
+			return rmpp_recv;
+	}
+	return NULL;
+}
+
+static struct mad_rmpp_recv *
+acquire_rmpp_recv(struct ib_mad_agent_private *agent,
+		  struct ib_mad_recv_wc *mad_recv_wc)
+{
+	struct mad_rmpp_recv *rmpp_recv;
+	unsigned long flags;
+
+	spin_lock_irqsave(&agent->lock, flags);
+	rmpp_recv = find_rmpp_recv(agent, mad_recv_wc);
+	if (rmpp_recv)
+		atomic_inc(&rmpp_recv->refcount);
+	spin_unlock_irqrestore(&agent->lock, flags);
+	return rmpp_recv;
+}
+
+static struct mad_rmpp_recv *
+insert_rmpp_recv(struct ib_mad_agent_private *agent,
+		 struct mad_rmpp_recv *rmpp_recv)
+{
+	struct mad_rmpp_recv *cur_rmpp_recv;
+
+	cur_rmpp_recv = find_rmpp_recv(agent, rmpp_recv->rmpp_wc);
+	if (!cur_rmpp_recv)
+		list_add_tail(&rmpp_recv->list, &agent->rmpp_list);
+
+	return cur_rmpp_recv;
+}
+
+static inline int get_last_flag(struct ib_mad_recv_buf *seg)
+{
+	struct ib_rmpp_mad *rmpp_mad;
+
+	rmpp_mad = (struct ib_rmpp_mad *) seg->mad;
+	return ib_get_rmpp_flags(&rmpp_mad->rmpp_hdr) & IB_MGMT_RMPP_FLAG_LAST;
+}
+
+static inline int get_seg_num(struct ib_mad_recv_buf *seg)
+{
+	struct ib_rmpp_mad *rmpp_mad;
+
+	rmpp_mad = (struct ib_rmpp_mad *) seg->mad;
+	return be32_to_cpu(rmpp_mad->rmpp_hdr.seg_num);
+}
+
+static inline struct ib_mad_recv_buf * get_next_seg(struct list_head *rmpp_list,
+						    struct ib_mad_recv_buf *seg)
+{
+	if (seg->list.next == rmpp_list)
+		return NULL;
+
+	return container_of(seg->list.next, struct ib_mad_recv_buf, list);
+}
+
+static inline int window_size(struct ib_mad_agent_private *agent)
+{
+	return max(agent->qp_info->recv_queue.max_active >> 3, 1);
+}
+
+static struct ib_mad_recv_buf * find_seg_location(struct list_head *rmpp_list,
+						  int seg_num)
+{
+	struct ib_mad_recv_buf *seg_buf;
+	int cur_seg_num;
+
+	list_for_each_entry_reverse(seg_buf, rmpp_list, list) {
+		cur_seg_num = get_seg_num(seg_buf);
+		if (seg_num > cur_seg_num)
+			return seg_buf;
+		if (seg_num == cur_seg_num)
+			break;
+	}
+	return NULL;
+}
+
+static void update_seg_num(struct mad_rmpp_recv *rmpp_recv,
+			   struct ib_mad_recv_buf *new_buf)
+{
+	struct list_head *rmpp_list = &rmpp_recv->rmpp_wc->rmpp_list;
+
+	while (new_buf && (get_seg_num(new_buf) == rmpp_recv->seg_num + 1)) {
+		rmpp_recv->cur_seg_buf = new_buf;
+		rmpp_recv->seg_num++;
+		new_buf = get_next_seg(rmpp_list, new_buf);
+	}
+}
+
+static inline int get_mad_len(struct mad_rmpp_recv *rmpp_recv)
+{
+	struct ib_rmpp_mad *rmpp_mad;
+	int hdr_size, data_size, pad;
+
+	rmpp_mad = (struct ib_rmpp_mad *)rmpp_recv->cur_seg_buf->mad;
+
+	hdr_size = ib_get_mad_data_offset(rmpp_mad->mad_hdr.mgmt_class);
+	data_size = sizeof(struct ib_rmpp_mad) - hdr_size;
+	pad = IB_MGMT_RMPP_DATA - be32_to_cpu(rmpp_mad->rmpp_hdr.paylen_newwin);
+	if (pad > IB_MGMT_RMPP_DATA || pad < 0)
+		pad = 0;
+
+	return hdr_size + rmpp_recv->seg_num * data_size - pad;
+}
+
+static struct ib_mad_recv_wc * complete_rmpp(struct mad_rmpp_recv *rmpp_recv)
+{
+	struct ib_mad_recv_wc *rmpp_wc;
+
+	ack_recv(rmpp_recv, rmpp_recv->rmpp_wc);
+	if (rmpp_recv->seg_num > 1)
+		cancel_delayed_work(&rmpp_recv->timeout_work);
+
+	rmpp_wc = rmpp_recv->rmpp_wc;
+	rmpp_wc->mad_len = get_mad_len(rmpp_recv);
+	/* 10 seconds until we can find the packet lifetime */
+	queue_delayed_work(rmpp_recv->agent->qp_info->port_priv->wq,
+			   &rmpp_recv->cleanup_work, msecs_to_jiffies(10000));
+	return rmpp_wc;
+}
+
+static struct ib_mad_recv_wc *
+continue_rmpp(struct ib_mad_agent_private *agent,
+	      struct ib_mad_recv_wc *mad_recv_wc)
+{
+	struct mad_rmpp_recv *rmpp_recv;
+	struct ib_mad_recv_buf *prev_buf;
+	struct ib_mad_recv_wc *done_wc;
+	int seg_num;
+	unsigned long flags;
+
+	rmpp_recv = acquire_rmpp_recv(agent, mad_recv_wc);
+	if (!rmpp_recv)
+		goto drop1;
+
+	seg_num = get_seg_num(&mad_recv_wc->recv_buf);
+
+	spin_lock_irqsave(&rmpp_recv->lock, flags);
+	if ((rmpp_recv->state == RMPP_STATE_TIMEOUT) ||
+	    (seg_num > rmpp_recv->newwin))
+		goto drop3;
+
+	if ((seg_num <= rmpp_recv->last_ack) ||
+	    (rmpp_recv->state == RMPP_STATE_COMPLETE)) {
+		spin_unlock_irqrestore(&rmpp_recv->lock, flags);
+		ack_recv(rmpp_recv, mad_recv_wc);
+		goto drop2;
+	}
+
+	prev_buf = find_seg_location(&rmpp_recv->rmpp_wc->rmpp_list, seg_num);
+	if (!prev_buf)
+		goto drop3;
+
+	done_wc = NULL;
+	list_add(&mad_recv_wc->recv_buf.list, &prev_buf->list);
+	if (rmpp_recv->cur_seg_buf == prev_buf) {
+		update_seg_num(rmpp_recv, &mad_recv_wc->recv_buf);
+		if (get_last_flag(rmpp_recv->cur_seg_buf)) {
+			rmpp_recv->state = RMPP_STATE_COMPLETE;
+			spin_unlock_irqrestore(&rmpp_recv->lock, flags);
+			done_wc = complete_rmpp(rmpp_recv);
+			goto out;
+		} else if (rmpp_recv->seg_num == rmpp_recv->newwin) {
+			rmpp_recv->newwin += window_size(agent);
+			spin_unlock_irqrestore(&rmpp_recv->lock, flags);
+			ack_recv(rmpp_recv, mad_recv_wc);
+			goto out;
+		}
+	}
+	spin_unlock_irqrestore(&rmpp_recv->lock, flags);
+out:
+	deref_rmpp_recv(rmpp_recv);
+	return done_wc;
+
+drop3:	spin_unlock_irqrestore(&rmpp_recv->lock, flags);
+drop2:	deref_rmpp_recv(rmpp_recv);
+drop1:	ib_free_recv_mad(mad_recv_wc);
+	return NULL;
+}
+
+static struct ib_mad_recv_wc *
+start_rmpp(struct ib_mad_agent_private *agent,
+	   struct ib_mad_recv_wc *mad_recv_wc)
+{
+	struct mad_rmpp_recv *rmpp_recv;
+	unsigned long flags;
+
+	rmpp_recv = create_rmpp_recv(agent, mad_recv_wc);
+	if (!rmpp_recv) {
+		ib_free_recv_mad(mad_recv_wc);
+		return NULL;
+	}
+
+	spin_lock_irqsave(&agent->lock, flags);
+	if (insert_rmpp_recv(agent, rmpp_recv)) {
+		spin_unlock_irqrestore(&agent->lock, flags);
+		/* duplicate first MAD */
+		destroy_rmpp_recv(rmpp_recv);
+		return continue_rmpp(agent, mad_recv_wc);
+	}
+	atomic_inc(&rmpp_recv->refcount);
+
+	if (get_last_flag(&mad_recv_wc->recv_buf)) {
+		rmpp_recv->state = RMPP_STATE_COMPLETE;
+		spin_unlock_irqrestore(&agent->lock, flags);
+		complete_rmpp(rmpp_recv);
+	} else {
+		spin_unlock_irqrestore(&agent->lock, flags);
+		/* 40 seconds until we can find the packet lifetimes */
+		queue_delayed_work(agent->qp_info->port_priv->wq,
+				   &rmpp_recv->timeout_work,
+				   msecs_to_jiffies(40000));
+		rmpp_recv->newwin += window_size(agent);
+		ack_recv(rmpp_recv, mad_recv_wc);
+		mad_recv_wc = NULL;
+	}
+	deref_rmpp_recv(rmpp_recv);
+	return mad_recv_wc;
+}
+
+static int send_next_seg(struct ib_mad_send_wr_private *mad_send_wr)
+{
+	struct ib_rmpp_mad *rmpp_mad;
+	int timeout;
+	u32 paylen = 0;
+
+	rmpp_mad = mad_send_wr->send_buf.mad;
+	ib_set_rmpp_flags(&rmpp_mad->rmpp_hdr, IB_MGMT_RMPP_FLAG_ACTIVE);
+	rmpp_mad->rmpp_hdr.seg_num = cpu_to_be32(++mad_send_wr->seg_num);
+
+	if (mad_send_wr->seg_num == 1) {
+		rmpp_mad->rmpp_hdr.rmpp_rtime_flags |= IB_MGMT_RMPP_FLAG_FIRST;
+		paylen = mad_send_wr->send_buf.seg_count * IB_MGMT_RMPP_DATA -
+			 mad_send_wr->pad;
+	}
+
+	if (mad_send_wr->seg_num == mad_send_wr->send_buf.seg_count) {
+		rmpp_mad->rmpp_hdr.rmpp_rtime_flags |= IB_MGMT_RMPP_FLAG_LAST;
+		paylen = IB_MGMT_RMPP_DATA - mad_send_wr->pad;
+	}
+	rmpp_mad->rmpp_hdr.paylen_newwin = cpu_to_be32(paylen);
+
+	/* 2 seconds for an ACK until we can find the packet lifetime */
+	timeout = mad_send_wr->send_buf.timeout_ms;
+	if (!timeout || timeout > 2000)
+		mad_send_wr->timeout = msecs_to_jiffies(2000);
+
+	return ib_send_mad(mad_send_wr);
+}
+
+static void abort_send(struct ib_mad_agent_private *agent,
+		       struct ib_mad_recv_wc *mad_recv_wc, u8 rmpp_status)
+{
+	struct ib_mad_send_wr_private *mad_send_wr;
+	struct ib_mad_send_wc wc;
+	unsigned long flags;
+
+	spin_lock_irqsave(&agent->lock, flags);
+	mad_send_wr = ib_find_send_mad(agent, mad_recv_wc);
+	if (!mad_send_wr)
+		goto out;	/* Unmatched send */
+
+	if ((mad_send_wr->last_ack == mad_send_wr->send_buf.seg_count) ||
+	    (!mad_send_wr->timeout) || (mad_send_wr->status != IB_WC_SUCCESS))
+		goto out;	/* Send is already done */
+
+	ib_mark_mad_done(mad_send_wr);
+	spin_unlock_irqrestore(&agent->lock, flags);
+
+	wc.status = IB_WC_REM_ABORT_ERR;
+	wc.vendor_err = rmpp_status;
+	wc.send_buf = &mad_send_wr->send_buf;
+	ib_mad_complete_send_wr(mad_send_wr, &wc);
+	return;
+out:
+	spin_unlock_irqrestore(&agent->lock, flags);
+}
+
+static inline void adjust_last_ack(struct ib_mad_send_wr_private *wr,
+				   int seg_num)
+{
+	struct list_head *list;
+
+	wr->last_ack = seg_num;
+	list = &wr->last_ack_seg->list;
+	list_for_each_entry(wr->last_ack_seg, list, list)
+		if (wr->last_ack_seg->num == seg_num)
+			break;
+}
+
+static void process_ds_ack(struct ib_mad_agent_private *agent,
+			   struct ib_mad_recv_wc *mad_recv_wc, int newwin)
+{
+	struct mad_rmpp_recv *rmpp_recv;
+
+	rmpp_recv = find_rmpp_recv(agent, mad_recv_wc);
+	if (rmpp_recv && rmpp_recv->state == RMPP_STATE_COMPLETE)
+		rmpp_recv->repwin = newwin;
+}
+
+static void process_rmpp_ack(struct ib_mad_agent_private *agent,
+			     struct ib_mad_recv_wc *mad_recv_wc)
+{
+	struct ib_mad_send_wr_private *mad_send_wr;
+	struct ib_rmpp_mad *rmpp_mad;
+	unsigned long flags;
+	int seg_num, newwin, ret;
+
+	rmpp_mad = (struct ib_rmpp_mad *)mad_recv_wc->recv_buf.mad;
+	if (rmpp_mad->rmpp_hdr.rmpp_status) {
+		abort_send(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_BAD_STATUS);
+		nack_recv(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_BAD_STATUS);
+		return;
+	}
+
+	seg_num = be32_to_cpu(rmpp_mad->rmpp_hdr.seg_num);
+	newwin = be32_to_cpu(rmpp_mad->rmpp_hdr.paylen_newwin);
+	if (newwin < seg_num) {
+		abort_send(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_W2S);
+		nack_recv(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_W2S);
+		return;
+	}
+
+	spin_lock_irqsave(&agent->lock, flags);
+	mad_send_wr = ib_find_send_mad(agent, mad_recv_wc);
+	if (!mad_send_wr) {
+		if (!seg_num)
+			process_ds_ack(agent, mad_recv_wc, newwin);
+		goto out;	/* Unmatched or DS RMPP ACK */
+	}
+
+	if ((mad_send_wr->last_ack == mad_send_wr->send_buf.seg_count) &&
+	    (mad_send_wr->timeout)) {
+		spin_unlock_irqrestore(&agent->lock, flags);
+		ack_ds_ack(agent, mad_recv_wc);
+		return;		/* Repeated ACK for DS RMPP transaction */
+	}
+
+	if ((mad_send_wr->last_ack == mad_send_wr->send_buf.seg_count) ||
+	    (!mad_send_wr->timeout) || (mad_send_wr->status != IB_WC_SUCCESS))
+		goto out;	/* Send is already done */
+
+	if (seg_num > mad_send_wr->send_buf.seg_count ||
+	    seg_num > mad_send_wr->newwin) {
+		spin_unlock_irqrestore(&agent->lock, flags);
+		abort_send(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_S2B);
+		nack_recv(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_S2B);
+		return;
+	}
+
+	if (newwin < mad_send_wr->newwin || seg_num < mad_send_wr->last_ack)
+		goto out;	/* Old ACK */
+
+	if (seg_num > mad_send_wr->last_ack) {
+		adjust_last_ack(mad_send_wr, seg_num);
+		mad_send_wr->retries = mad_send_wr->send_buf.retries;
+	}
+	mad_send_wr->newwin = newwin;
+	if (mad_send_wr->last_ack == mad_send_wr->send_buf.seg_count) {
+		/* If no response is expected, the ACK completes the send */
+		if (!mad_send_wr->send_buf.timeout_ms) {
+			struct ib_mad_send_wc wc;
+
+			ib_mark_mad_done(mad_send_wr);
+			spin_unlock_irqrestore(&agent->lock, flags);
+
+			wc.status = IB_WC_SUCCESS;
+			wc.vendor_err = 0;
+			wc.send_buf = &mad_send_wr->send_buf;
+			ib_mad_complete_send_wr(mad_send_wr, &wc);
+			return;
+		}
+		if (mad_send_wr->refcount == 1)
+			ib_reset_mad_timeout(mad_send_wr,
+					     mad_send_wr->send_buf.timeout_ms);
+		spin_unlock_irqrestore(&agent->lock, flags);
+		ack_ds_ack(agent, mad_recv_wc);
+		return;
+	} else if (mad_send_wr->refcount == 1 &&
+		   mad_send_wr->seg_num < mad_send_wr->newwin &&
+		   mad_send_wr->seg_num < mad_send_wr->send_buf.seg_count) {
+		/* Send failure will just result in a timeout/retry */
+		ret = send_next_seg(mad_send_wr);
+		if (ret)
+			goto out;
+
+		mad_send_wr->refcount++;
+		list_move_tail(&mad_send_wr->agent_list,
+			      &mad_send_wr->mad_agent_priv->send_list);
+	}
+out:
+	spin_unlock_irqrestore(&agent->lock, flags);
+}
+
+static struct ib_mad_recv_wc *
+process_rmpp_data(struct ib_mad_agent_private *agent,
+		  struct ib_mad_recv_wc *mad_recv_wc)
+{
+	struct ib_rmpp_hdr *rmpp_hdr;
+	u8 rmpp_status;
+
+	rmpp_hdr = &((struct ib_rmpp_mad *)mad_recv_wc->recv_buf.mad)->rmpp_hdr;
+
+	if (rmpp_hdr->rmpp_status) {
+		rmpp_status = IB_MGMT_RMPP_STATUS_BAD_STATUS;
+		goto bad;
+	}
+
+	if (rmpp_hdr->seg_num == __constant_htonl(1)) {
+		if (!(ib_get_rmpp_flags(rmpp_hdr) & IB_MGMT_RMPP_FLAG_FIRST)) {
+			rmpp_status = IB_MGMT_RMPP_STATUS_BAD_SEG;
+			goto bad;
+		}
+		return start_rmpp(agent, mad_recv_wc);
+	} else {
+		if (ib_get_rmpp_flags(rmpp_hdr) & IB_MGMT_RMPP_FLAG_FIRST) {
+			rmpp_status = IB_MGMT_RMPP_STATUS_BAD_SEG;
+			goto bad;
+		}
+		return continue_rmpp(agent, mad_recv_wc);
+	}
+bad:
+	nack_recv(agent, mad_recv_wc, rmpp_status);
+	ib_free_recv_mad(mad_recv_wc);
+	return NULL;
+}
+
+static void process_rmpp_stop(struct ib_mad_agent_private *agent,
+			      struct ib_mad_recv_wc *mad_recv_wc)
+{
+	struct ib_rmpp_mad *rmpp_mad;
+
+	rmpp_mad = (struct ib_rmpp_mad *)mad_recv_wc->recv_buf.mad;
+
+	if (rmpp_mad->rmpp_hdr.rmpp_status != IB_MGMT_RMPP_STATUS_RESX) {
+		abort_send(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_BAD_STATUS);
+		nack_recv(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_BAD_STATUS);
+	} else
+		abort_send(agent, mad_recv_wc, rmpp_mad->rmpp_hdr.rmpp_status);
+}
+
+static void process_rmpp_abort(struct ib_mad_agent_private *agent,
+			       struct ib_mad_recv_wc *mad_recv_wc)
+{
+	struct ib_rmpp_mad *rmpp_mad;
+
+	rmpp_mad = (struct ib_rmpp_mad *)mad_recv_wc->recv_buf.mad;
+
+	if (rmpp_mad->rmpp_hdr.rmpp_status < IB_MGMT_RMPP_STATUS_ABORT_MIN ||
+	    rmpp_mad->rmpp_hdr.rmpp_status > IB_MGMT_RMPP_STATUS_ABORT_MAX) {
+		abort_send(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_BAD_STATUS);
+		nack_recv(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_BAD_STATUS);
+	} else
+		abort_send(agent, mad_recv_wc, rmpp_mad->rmpp_hdr.rmpp_status);
+}
+
+struct ib_mad_recv_wc *
+ib_process_rmpp_recv_wc(struct ib_mad_agent_private *agent,
+			struct ib_mad_recv_wc *mad_recv_wc)
+{
+	struct ib_rmpp_mad *rmpp_mad;
+
+	rmpp_mad = (struct ib_rmpp_mad *)mad_recv_wc->recv_buf.mad;
+	if (!(rmpp_mad->rmpp_hdr.rmpp_rtime_flags & IB_MGMT_RMPP_FLAG_ACTIVE))
+		return mad_recv_wc;
+
+	if (rmpp_mad->rmpp_hdr.rmpp_version != IB_MGMT_RMPP_VERSION) {
+		abort_send(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_UNV);
+		nack_recv(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_UNV);
+		goto out;
+	}
+
+	switch (rmpp_mad->rmpp_hdr.rmpp_type) {
+	case IB_MGMT_RMPP_TYPE_DATA:
+		return process_rmpp_data(agent, mad_recv_wc);
+	case IB_MGMT_RMPP_TYPE_ACK:
+		process_rmpp_ack(agent, mad_recv_wc);
+		break;
+	case IB_MGMT_RMPP_TYPE_STOP:
+		process_rmpp_stop(agent, mad_recv_wc);
+		break;
+	case IB_MGMT_RMPP_TYPE_ABORT:
+		process_rmpp_abort(agent, mad_recv_wc);
+		break;
+	default:
+		abort_send(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_BADT);
+		nack_recv(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_BADT);
+		break;
+	}
+out:
+	ib_free_recv_mad(mad_recv_wc);
+	return NULL;
+}
+
+static int init_newwin(struct ib_mad_send_wr_private *mad_send_wr)
+{
+	struct ib_mad_agent_private *agent = mad_send_wr->mad_agent_priv;
+	struct ib_mad_hdr *mad_hdr = mad_send_wr->send_buf.mad;
+	struct mad_rmpp_recv *rmpp_recv;
+	struct ib_ah_attr ah_attr;
+	unsigned long flags;
+	int newwin = 1;
+
+	if (!(mad_hdr->method & IB_MGMT_METHOD_RESP))
+		goto out;
+
+	spin_lock_irqsave(&agent->lock, flags);
+	list_for_each_entry(rmpp_recv, &agent->rmpp_list, list) {
+		if (rmpp_recv->tid != mad_hdr->tid ||
+		    rmpp_recv->mgmt_class != mad_hdr->mgmt_class ||
+		    rmpp_recv->class_version != mad_hdr->class_version ||
+		    (rmpp_recv->method & IB_MGMT_METHOD_RESP))
+			continue;
+
+		if (ib_query_ah(mad_send_wr->send_buf.ah, &ah_attr))
+			continue;
+
+		if (rmpp_recv->slid == ah_attr.dlid) {
+			newwin = rmpp_recv->repwin;
+			break;
+		}
+	}
+	spin_unlock_irqrestore(&agent->lock, flags);
+out:
+	return newwin;
+}
+
+int ib_send_rmpp_mad(struct ib_mad_send_wr_private *mad_send_wr)
+{
+	struct ib_rmpp_mad *rmpp_mad;
+	int ret;
+
+	rmpp_mad = mad_send_wr->send_buf.mad;
+	if (!(ib_get_rmpp_flags(&rmpp_mad->rmpp_hdr) &
+	      IB_MGMT_RMPP_FLAG_ACTIVE))
+		return IB_RMPP_RESULT_UNHANDLED;
+
+	if (rmpp_mad->rmpp_hdr.rmpp_type != IB_MGMT_RMPP_TYPE_DATA) {
+		mad_send_wr->seg_num = 1;
+		return IB_RMPP_RESULT_INTERNAL;
+	}
+
+	mad_send_wr->newwin = init_newwin(mad_send_wr);
+
+	/* We need to wait for the final ACK even if there isn't a response */
+	mad_send_wr->refcount += (mad_send_wr->timeout == 0);
+	ret = send_next_seg(mad_send_wr);
+	if (!ret)
+		return IB_RMPP_RESULT_CONSUMED;
+	return ret;
+}
+
+int ib_process_rmpp_send_wc(struct ib_mad_send_wr_private *mad_send_wr,
+			    struct ib_mad_send_wc *mad_send_wc)
+{
+	struct ib_rmpp_mad *rmpp_mad;
+	int ret;
+
+	rmpp_mad = mad_send_wr->send_buf.mad;
+	if (!(ib_get_rmpp_flags(&rmpp_mad->rmpp_hdr) &
+	      IB_MGMT_RMPP_FLAG_ACTIVE))
+		return IB_RMPP_RESULT_UNHANDLED; /* RMPP not active */
+
+	if (rmpp_mad->rmpp_hdr.rmpp_type != IB_MGMT_RMPP_TYPE_DATA)
+		return IB_RMPP_RESULT_INTERNAL;	 /* ACK, STOP, or ABORT */
+
+	if (mad_send_wc->status != IB_WC_SUCCESS ||
+	    mad_send_wr->status != IB_WC_SUCCESS)
+		return IB_RMPP_RESULT_PROCESSED; /* Canceled or send error */
+
+	if (!mad_send_wr->timeout)
+		return IB_RMPP_RESULT_PROCESSED; /* Response received */
+
+	if (mad_send_wr->last_ack == mad_send_wr->send_buf.seg_count) {
+		mad_send_wr->timeout =
+			msecs_to_jiffies(mad_send_wr->send_buf.timeout_ms);
+		return IB_RMPP_RESULT_PROCESSED; /* Send done */
+	}
+
+	if (mad_send_wr->seg_num == mad_send_wr->newwin ||
+	    mad_send_wr->seg_num == mad_send_wr->send_buf.seg_count)
+		return IB_RMPP_RESULT_PROCESSED; /* Wait for ACK */
+
+	ret = send_next_seg(mad_send_wr);
+	if (ret) {
+		mad_send_wc->status = IB_WC_GENERAL_ERR;
+		return IB_RMPP_RESULT_PROCESSED;
+	}
+	return IB_RMPP_RESULT_CONSUMED;
+}
+
+int ib_retry_rmpp(struct ib_mad_send_wr_private *mad_send_wr)
+{
+	struct ib_rmpp_mad *rmpp_mad;
+	int ret;
+
+	rmpp_mad = mad_send_wr->send_buf.mad;
+	if (!(ib_get_rmpp_flags(&rmpp_mad->rmpp_hdr) &
+	      IB_MGMT_RMPP_FLAG_ACTIVE))
+		return IB_RMPP_RESULT_UNHANDLED; /* RMPP not active */
+
+	if (mad_send_wr->last_ack == mad_send_wr->send_buf.seg_count)
+		return IB_RMPP_RESULT_PROCESSED;
+
+	mad_send_wr->seg_num = mad_send_wr->last_ack;
+	mad_send_wr->cur_seg = mad_send_wr->last_ack_seg;
+
+	ret = send_next_seg(mad_send_wr);
+	if (ret)
+		return IB_RMPP_RESULT_PROCESSED;
+
+	return IB_RMPP_RESULT_CONSUMED;
+}

drivers/infiniband/core/mad_rmpp.h

@@ -0,0 +1,60 @@
+/*
+ * Copyright (c) 2005 Intel Inc. 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.
+ *
+ * $Id: mad_rmpp.h,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#ifndef __MAD_RMPP_H__
+#define __MAD_RMPP_H__
+
+enum {
+	IB_RMPP_RESULT_PROCESSED,
+	IB_RMPP_RESULT_CONSUMED,
+	IB_RMPP_RESULT_INTERNAL,
+	IB_RMPP_RESULT_UNHANDLED
+};
+
+int ib_send_rmpp_mad(struct ib_mad_send_wr_private *mad_send_wr);
+
+struct ib_mad_recv_wc *
+ib_process_rmpp_recv_wc(struct ib_mad_agent_private *agent,
+			struct ib_mad_recv_wc *mad_recv_wc);
+
+int ib_process_rmpp_send_wc(struct ib_mad_send_wr_private *mad_send_wr,
+			    struct ib_mad_send_wc *mad_send_wc);
+
+void ib_rmpp_send_handler(struct ib_mad_send_wc *mad_send_wc);
+
+void ib_cancel_rmpp_recvs(struct ib_mad_agent_private *agent);
+
+int ib_retry_rmpp(struct ib_mad_send_wr_private *mad_send_wr);
+
+#endif	/* __MAD_RMPP_H__ */

drivers/infiniband/core/multicast.c

@@ -0,0 +1,837 @@
+/*
+ * Copyright (c) 2006 Intel Corporation.  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/completion.h>
+#include <linux/dma-mapping.h>
+#include <linux/err.h>
+#include <linux/interrupt.h>
+#include <linux/pci.h>
+#include <linux/bitops.h>
+#include <linux/random.h>
+
+#include <rdma/ib_cache.h>
+#include "sa.h"
+
+static void mcast_add_one(struct ib_device *device);
+static void mcast_remove_one(struct ib_device *device);
+
+static struct ib_client mcast_client = {
+	.name   = "ib_multicast",
+	.add    = mcast_add_one,
+	.remove = mcast_remove_one
+};
+
+static struct ib_sa_client	sa_client;
+static struct workqueue_struct	*mcast_wq;
+static union ib_gid mgid0;
+
+struct mcast_device;
+
+struct mcast_port {
+	struct mcast_device	*dev;
+	spinlock_t		lock;
+	struct rb_root		table;
+	atomic_t		refcount;
+	struct completion	comp;
+	u8			port_num;
+};
+
+struct mcast_device {
+	struct ib_device	*device;
+	struct ib_event_handler	event_handler;
+	int			start_port;
+	int			end_port;
+	struct mcast_port	port[0];
+};
+
+enum mcast_state {
+	MCAST_IDLE,
+	MCAST_JOINING,
+	MCAST_MEMBER,
+	MCAST_BUSY,
+	MCAST_ERROR
+};
+
+struct mcast_member;
+
+struct mcast_group {
+	struct ib_sa_mcmember_rec rec;
+	struct rb_node		node;
+	struct mcast_port	*port;
+	spinlock_t		lock;
+	struct work_struct	work;
+	struct list_head	pending_list;
+	struct list_head	active_list;
+	struct mcast_member	*last_join;
+	int			members[3];
+	atomic_t		refcount;
+	enum mcast_state	state;
+	struct ib_sa_query	*query;
+	int			query_id;
+};
+
+struct mcast_member {
+	struct ib_sa_multicast	multicast;
+	struct ib_sa_client	*client;
+	struct mcast_group	*group;
+	struct list_head	list;
+	enum mcast_state	state;
+	atomic_t		refcount;
+	struct completion	comp;
+};
+
+static void join_handler(int status, struct ib_sa_mcmember_rec *rec,
+			 void *context);
+static void leave_handler(int status, struct ib_sa_mcmember_rec *rec,
+			  void *context);
+
+static struct mcast_group *mcast_find(struct mcast_port *port,
+				      union ib_gid *mgid)
+{
+	struct rb_node *node = port->table.rb_node;
+	struct mcast_group *group;
+	int ret;
+
+	while (node) {
+		group = rb_entry(node, struct mcast_group, node);
+		ret = memcmp(mgid->raw, group->rec.mgid.raw, sizeof *mgid);
+		if (!ret)
+			return group;
+
+		if (ret < 0)
+			node = node->rb_left;
+		else
+			node = node->rb_right;
+	}
+	return NULL;
+}
+
+static struct mcast_group *mcast_insert(struct mcast_port *port,
+					struct mcast_group *group,
+					int allow_duplicates)
+{
+	struct rb_node **link = &port->table.rb_node;
+	struct rb_node *parent = NULL;
+	struct mcast_group *cur_group;
+	int ret;
+
+	while (*link) {
+		parent = *link;
+		cur_group = rb_entry(parent, struct mcast_group, node);
+
+		ret = memcmp(group->rec.mgid.raw, cur_group->rec.mgid.raw,
+			     sizeof group->rec.mgid);
+		if (ret < 0)
+			link = &(*link)->rb_left;
+		else if (ret > 0)
+			link = &(*link)->rb_right;
+		else if (allow_duplicates)
+			link = &(*link)->rb_left;
+		else
+			return cur_group;
+	}
+	rb_link_node(&group->node, parent, link);
+	rb_insert_color(&group->node, &port->table);
+	return NULL;
+}
+
+static void deref_port(struct mcast_port *port)
+{
+	if (atomic_dec_and_test(&port->refcount))
+		complete(&port->comp);
+}
+
+static void release_group(struct mcast_group *group)
+{
+	struct mcast_port *port = group->port;
+	unsigned long flags;
+
+	spin_lock_irqsave(&port->lock, flags);
+	if (atomic_dec_and_test(&group->refcount)) {
+		rb_erase(&group->node, &port->table);
+		spin_unlock_irqrestore(&port->lock, flags);
+		kfree(group);
+		deref_port(port);
+	} else
+		spin_unlock_irqrestore(&port->lock, flags);
+}
+
+static void deref_member(struct mcast_member *member)
+{
+	if (atomic_dec_and_test(&member->refcount))
+		complete(&member->comp);
+}
+
+static void queue_join(struct mcast_member *member)
+{
+	struct mcast_group *group = member->group;
+	unsigned long flags;
+
+	spin_lock_irqsave(&group->lock, flags);
+	list_add(&member->list, &group->pending_list);
+	if (group->state == MCAST_IDLE) {
+		group->state = MCAST_BUSY;
+		atomic_inc(&group->refcount);
+		queue_work(mcast_wq, &group->work);
+	}
+	spin_unlock_irqrestore(&group->lock, flags);
+}
+
+/*
+ * A multicast group has three types of members: full member, non member, and
+ * send only member.  We need to keep track of the number of members of each
+ * type based on their join state.  Adjust the number of members the belong to
+ * the specified join states.
+ */
+static void adjust_membership(struct mcast_group *group, u8 join_state, int inc)
+{
+	int i;
+
+	for (i = 0; i < 3; i++, join_state >>= 1)
+		if (join_state & 0x1)
+			group->members[i] += inc;
+}
+
+/*
+ * If a multicast group has zero members left for a particular join state, but
+ * the group is still a member with the SA, we need to leave that join state.
+ * Determine which join states we still belong to, but that do not have any
+ * active members.
+ */
+static u8 get_leave_state(struct mcast_group *group)
+{
+	u8 leave_state = 0;
+	int i;
+
+	for (i = 0; i < 3; i++)
+		if (!group->members[i])
+			leave_state |= (0x1 << i);
+
+	return leave_state & group->rec.join_state;
+}
+
+static int check_selector(ib_sa_comp_mask comp_mask,
+			  ib_sa_comp_mask selector_mask,
+			  ib_sa_comp_mask value_mask,
+			  u8 selector, u8 src_value, u8 dst_value)
+{
+	int err;
+
+	if (!(comp_mask & selector_mask) || !(comp_mask & value_mask))
+		return 0;
+
+	switch (selector) {
+	case IB_SA_GT:
+		err = (src_value <= dst_value);
+		break;
+	case IB_SA_LT:
+		err = (src_value >= dst_value);
+		break;
+	case IB_SA_EQ:
+		err = (src_value != dst_value);
+		break;
+	default:
+		err = 0;
+		break;
+	}
+
+	return err;
+}
+
+static int cmp_rec(struct ib_sa_mcmember_rec *src,
+		   struct ib_sa_mcmember_rec *dst, ib_sa_comp_mask comp_mask)
+{
+	/* MGID must already match */
+
+	if (comp_mask & IB_SA_MCMEMBER_REC_PORT_GID &&
+	    memcmp(&src->port_gid, &dst->port_gid, sizeof src->port_gid))
+		return -EINVAL;
+	if (comp_mask & IB_SA_MCMEMBER_REC_QKEY && src->qkey != dst->qkey)
+		return -EINVAL;
+	if (comp_mask & IB_SA_MCMEMBER_REC_MLID && src->mlid != dst->mlid)
+		return -EINVAL;
+	if (check_selector(comp_mask, IB_SA_MCMEMBER_REC_MTU_SELECTOR,
+			   IB_SA_MCMEMBER_REC_MTU, dst->mtu_selector,
+			   src->mtu, dst->mtu))
+		return -EINVAL;
+	if (comp_mask & IB_SA_MCMEMBER_REC_TRAFFIC_CLASS &&
+	    src->traffic_class != dst->traffic_class)
+		return -EINVAL;
+	if (comp_mask & IB_SA_MCMEMBER_REC_PKEY && src->pkey != dst->pkey)
+		return -EINVAL;
+	if (check_selector(comp_mask, IB_SA_MCMEMBER_REC_RATE_SELECTOR,
+			   IB_SA_MCMEMBER_REC_RATE, dst->rate_selector,
+			   src->rate, dst->rate))
+		return -EINVAL;
+	if (check_selector(comp_mask,
+			   IB_SA_MCMEMBER_REC_PACKET_LIFE_TIME_SELECTOR,
+			   IB_SA_MCMEMBER_REC_PACKET_LIFE_TIME,
+			   dst->packet_life_time_selector,
+			   src->packet_life_time, dst->packet_life_time))
+		return -EINVAL;
+	if (comp_mask & IB_SA_MCMEMBER_REC_SL && src->sl != dst->sl)
+		return -EINVAL;
+	if (comp_mask & IB_SA_MCMEMBER_REC_FLOW_LABEL &&
+	    src->flow_label != dst->flow_label)
+		return -EINVAL;
+	if (comp_mask & IB_SA_MCMEMBER_REC_HOP_LIMIT &&
+	    src->hop_limit != dst->hop_limit)
+		return -EINVAL;
+	if (comp_mask & IB_SA_MCMEMBER_REC_SCOPE && src->scope != dst->scope)
+		return -EINVAL;
+
+	/* join_state checked separately, proxy_join ignored */
+
+	return 0;
+}
+
+static int send_join(struct mcast_group *group, struct mcast_member *member)
+{
+	struct mcast_port *port = group->port;
+	int ret;
+
+	group->last_join = member;
+	ret = ib_sa_mcmember_rec_query(&sa_client, port->dev->device,
+				       port->port_num, IB_MGMT_METHOD_SET,
+				       &member->multicast.rec,
+				       member->multicast.comp_mask,
+				       3000, GFP_KERNEL, join_handler, group,
+				       &group->query);
+	if (ret >= 0) {
+		group->query_id = ret;
+		ret = 0;
+	}
+	return ret;
+}
+
+static int send_leave(struct mcast_group *group, u8 leave_state)
+{
+	struct mcast_port *port = group->port;
+	struct ib_sa_mcmember_rec rec;
+	int ret;
+
+	rec = group->rec;
+	rec.join_state = leave_state;
+
+	ret = ib_sa_mcmember_rec_query(&sa_client, port->dev->device,
+				       port->port_num, IB_SA_METHOD_DELETE, &rec,
+				       IB_SA_MCMEMBER_REC_MGID     |
+				       IB_SA_MCMEMBER_REC_PORT_GID |
+				       IB_SA_MCMEMBER_REC_JOIN_STATE,
+				       3000, GFP_KERNEL, leave_handler,
+				       group, &group->query);
+	if (ret >= 0) {
+		group->query_id = ret;
+		ret = 0;
+	}
+	return ret;
+}
+
+static void join_group(struct mcast_group *group, struct mcast_member *member,
+		       u8 join_state)
+{
+	member->state = MCAST_MEMBER;
+	adjust_membership(group, join_state, 1);
+	group->rec.join_state |= join_state;
+	member->multicast.rec = group->rec;
+	member->multicast.rec.join_state = join_state;
+	list_move(&member->list, &group->active_list);
+}
+
+static int fail_join(struct mcast_group *group, struct mcast_member *member,
+		     int status)
+{
+	spin_lock_irq(&group->lock);
+	list_del_init(&member->list);
+	spin_unlock_irq(&group->lock);
+	return member->multicast.callback(status, &member->multicast);
+}
+
+static void process_group_error(struct mcast_group *group)
+{
+	struct mcast_member *member;
+	int ret;
+
+	spin_lock_irq(&group->lock);
+	while (!list_empty(&group->active_list)) {
+		member = list_entry(group->active_list.next,
+				    struct mcast_member, list);
+		atomic_inc(&member->refcount);
+		list_del_init(&member->list);
+		adjust_membership(group, member->multicast.rec.join_state, -1);
+		member->state = MCAST_ERROR;
+		spin_unlock_irq(&group->lock);
+
+		ret = member->multicast.callback(-ENETRESET,
+						 &member->multicast);
+		deref_member(member);
+		if (ret)
+			ib_sa_free_multicast(&member->multicast);
+		spin_lock_irq(&group->lock);
+	}
+
+	group->rec.join_state = 0;
+	group->state = MCAST_BUSY;
+	spin_unlock_irq(&group->lock);
+}
+
+static void mcast_work_handler(struct work_struct *work)
+{
+	struct mcast_group *group;
+	struct mcast_member *member;
+	struct ib_sa_multicast *multicast;
+	int status, ret;
+	u8 join_state;
+
+	group = container_of(work, typeof(*group), work);
+retest:
+	spin_lock_irq(&group->lock);
+	while (!list_empty(&group->pending_list) ||
+	       (group->state == MCAST_ERROR)) {
+
+		if (group->state == MCAST_ERROR) {
+			spin_unlock_irq(&group->lock);
+			process_group_error(group);
+			goto retest;
+		}
+
+		member = list_entry(group->pending_list.next,
+				    struct mcast_member, list);
+		multicast = &member->multicast;
+		join_state = multicast->rec.join_state;
+		atomic_inc(&member->refcount);
+
+		if (join_state == (group->rec.join_state & join_state)) {
+			status = cmp_rec(&group->rec, &multicast->rec,
+					 multicast->comp_mask);
+			if (!status)
+				join_group(group, member, join_state);
+			else
+				list_del_init(&member->list);
+			spin_unlock_irq(&group->lock);
+			ret = multicast->callback(status, multicast);
+		} else {
+			spin_unlock_irq(&group->lock);
+			status = send_join(group, member);
+			if (!status) {
+				deref_member(member);
+				return;
+			}
+			ret = fail_join(group, member, status);
+		}
+
+		deref_member(member);
+		if (ret)
+			ib_sa_free_multicast(&member->multicast);
+		spin_lock_irq(&group->lock);
+	}
+
+	join_state = get_leave_state(group);
+	if (join_state) {
+		group->rec.join_state &= ~join_state;
+		spin_unlock_irq(&group->lock);
+		if (send_leave(group, join_state))
+			goto retest;
+	} else {
+		group->state = MCAST_IDLE;
+		spin_unlock_irq(&group->lock);
+		release_group(group);
+	}
+}
+
+/*
+ * Fail a join request if it is still active - at the head of the pending queue.
+ */
+static void process_join_error(struct mcast_group *group, int status)
+{
+	struct mcast_member *member;
+	int ret;
+
+	spin_lock_irq(&group->lock);
+	member = list_entry(group->pending_list.next,
+			    struct mcast_member, list);
+	if (group->last_join == member) {
+		atomic_inc(&member->refcount);
+		list_del_init(&member->list);
+		spin_unlock_irq(&group->lock);
+		ret = member->multicast.callback(status, &member->multicast);
+		deref_member(member);
+		if (ret)
+			ib_sa_free_multicast(&member->multicast);
+	} else
+		spin_unlock_irq(&group->lock);
+}
+
+static void join_handler(int status, struct ib_sa_mcmember_rec *rec,
+			 void *context)
+{
+	struct mcast_group *group = context;
+
+	if (status)
+		process_join_error(group, status);
+	else {
+		spin_lock_irq(&group->port->lock);
+		group->rec = *rec;
+		if (!memcmp(&mgid0, &group->rec.mgid, sizeof mgid0)) {
+			rb_erase(&group->node, &group->port->table);
+			mcast_insert(group->port, group, 1);
+		}
+		spin_unlock_irq(&group->port->lock);
+	}
+	mcast_work_handler(&group->work);
+}
+
+static void leave_handler(int status, struct ib_sa_mcmember_rec *rec,
+			  void *context)
+{
+	struct mcast_group *group = context;
+
+	mcast_work_handler(&group->work);
+}
+
+static struct mcast_group *acquire_group(struct mcast_port *port,
+					 union ib_gid *mgid, gfp_t gfp_mask)
+{
+	struct mcast_group *group, *cur_group;
+	unsigned long flags;
+	int is_mgid0;
+
+	is_mgid0 = !memcmp(&mgid0, mgid, sizeof mgid0);
+	if (!is_mgid0) {
+		spin_lock_irqsave(&port->lock, flags);
+		group = mcast_find(port, mgid);
+		if (group)
+			goto found;
+		spin_unlock_irqrestore(&port->lock, flags);
+	}
+
+	group = kzalloc(sizeof *group, gfp_mask);
+	if (!group)
+		return NULL;
+
+	group->port = port;
+	group->rec.mgid = *mgid;
+	INIT_LIST_HEAD(&group->pending_list);
+	INIT_LIST_HEAD(&group->active_list);
+	INIT_WORK(&group->work, mcast_work_handler);
+	spin_lock_init(&group->lock);
+
+	spin_lock_irqsave(&port->lock, flags);
+	cur_group = mcast_insert(port, group, is_mgid0);
+	if (cur_group) {
+		kfree(group);
+		group = cur_group;
+	} else
+		atomic_inc(&port->refcount);
+found:
+	atomic_inc(&group->refcount);
+	spin_unlock_irqrestore(&port->lock, flags);
+	return group;
+}
+
+/*
+ * We serialize all join requests to a single group to make our lives much
+ * easier.  Otherwise, two users could try to join the same group
+ * simultaneously, with different configurations, one could leave while the
+ * join is in progress, etc., which makes locking around error recovery
+ * difficult.
+ */
+struct ib_sa_multicast *
+ib_sa_join_multicast(struct ib_sa_client *client,
+		     struct ib_device *device, u8 port_num,
+		     struct ib_sa_mcmember_rec *rec,
+		     ib_sa_comp_mask comp_mask, gfp_t gfp_mask,
+		     int (*callback)(int status,
+				     struct ib_sa_multicast *multicast),
+		     void *context)
+{
+	struct mcast_device *dev;
+	struct mcast_member *member;
+	struct ib_sa_multicast *multicast;
+	int ret;
+
+	dev = ib_get_client_data(device, &mcast_client);
+	if (!dev)
+		return ERR_PTR(-ENODEV);
+
+	member = kmalloc(sizeof *member, gfp_mask);
+	if (!member)
+		return ERR_PTR(-ENOMEM);
+
+	ib_sa_client_get(client);
+	member->client = client;
+	member->multicast.rec = *rec;
+	member->multicast.comp_mask = comp_mask;
+	member->multicast.callback = callback;
+	member->multicast.context = context;
+	init_completion(&member->comp);
+	atomic_set(&member->refcount, 1);
+	member->state = MCAST_JOINING;
+
+	member->group = acquire_group(&dev->port[port_num - dev->start_port],
+				      &rec->mgid, gfp_mask);
+	if (!member->group) {
+		ret = -ENOMEM;
+		goto err;
+	}
+
+	/*
+	 * The user will get the multicast structure in their callback.  They
+	 * could then free the multicast structure before we can return from
+	 * this routine.  So we save the pointer to return before queuing
+	 * any callback.
+	 */
+	multicast = &member->multicast;
+	queue_join(member);
+	return multicast;
+
+err:
+	ib_sa_client_put(client);
+	kfree(member);
+	return ERR_PTR(ret);
+}
+EXPORT_SYMBOL(ib_sa_join_multicast);
+
+void ib_sa_free_multicast(struct ib_sa_multicast *multicast)
+{
+	struct mcast_member *member;
+	struct mcast_group *group;
+
+	member = container_of(multicast, struct mcast_member, multicast);
+	group = member->group;
+
+	spin_lock_irq(&group->lock);
+	if (member->state == MCAST_MEMBER)
+		adjust_membership(group, multicast->rec.join_state, -1);
+
+	list_del_init(&member->list);
+
+	if (group->state == MCAST_IDLE) {
+		group->state = MCAST_BUSY;
+		spin_unlock_irq(&group->lock);
+		/* Continue to hold reference on group until callback */
+		queue_work(mcast_wq, &group->work);
+	} else {
+		spin_unlock_irq(&group->lock);
+		release_group(group);
+	}
+
+	deref_member(member);
+	wait_for_completion(&member->comp);
+	ib_sa_client_put(member->client);
+	kfree(member);
+}
+EXPORT_SYMBOL(ib_sa_free_multicast);
+
+int ib_sa_get_mcmember_rec(struct ib_device *device, u8 port_num,
+			   union ib_gid *mgid, struct ib_sa_mcmember_rec *rec)
+{
+	struct mcast_device *dev;
+	struct mcast_port *port;
+	struct mcast_group *group;
+	unsigned long flags;
+	int ret = 0;
+
+	dev = ib_get_client_data(device, &mcast_client);
+	if (!dev)
+		return -ENODEV;
+
+	port = &dev->port[port_num - dev->start_port];
+	spin_lock_irqsave(&port->lock, flags);
+	group = mcast_find(port, mgid);
+	if (group)
+		*rec = group->rec;
+	else
+		ret = -EADDRNOTAVAIL;
+	spin_unlock_irqrestore(&port->lock, flags);
+
+	return ret;
+}
+EXPORT_SYMBOL(ib_sa_get_mcmember_rec);
+
+int ib_init_ah_from_mcmember(struct ib_device *device, u8 port_num,
+			     struct ib_sa_mcmember_rec *rec,
+			     struct ib_ah_attr *ah_attr)
+{
+	int ret;
+	u16 gid_index;
+	u8 p;
+
+	ret = ib_find_cached_gid(device, &rec->port_gid, &p, &gid_index);
+	if (ret)
+		return ret;
+
+	memset(ah_attr, 0, sizeof *ah_attr);
+	ah_attr->dlid = be16_to_cpu(rec->mlid);
+	ah_attr->sl = rec->sl;
+	ah_attr->port_num = port_num;
+	ah_attr->static_rate = rec->rate;
+
+	ah_attr->ah_flags = IB_AH_GRH;
+	ah_attr->grh.dgid = rec->mgid;
+
+	ah_attr->grh.sgid_index = (u8) 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_mcmember);
+
+static void mcast_groups_lost(struct mcast_port *port)
+{
+	struct mcast_group *group;
+	struct rb_node *node;
+	unsigned long flags;
+
+	spin_lock_irqsave(&port->lock, flags);
+	for (node = rb_first(&port->table); node; node = rb_next(node)) {
+		group = rb_entry(node, struct mcast_group, node);
+		spin_lock(&group->lock);
+		if (group->state == MCAST_IDLE) {
+			atomic_inc(&group->refcount);
+			queue_work(mcast_wq, &group->work);
+		}
+		group->state = MCAST_ERROR;
+		spin_unlock(&group->lock);
+	}
+	spin_unlock_irqrestore(&port->lock, flags);
+}
+
+static void mcast_event_handler(struct ib_event_handler *handler,
+				struct ib_event *event)
+{
+	struct mcast_device *dev;
+
+	dev = container_of(handler, struct mcast_device, event_handler);
+
+	switch (event->event) {
+	case IB_EVENT_PORT_ERR:
+	case IB_EVENT_LID_CHANGE:
+	case IB_EVENT_SM_CHANGE:
+	case IB_EVENT_CLIENT_REREGISTER:
+		mcast_groups_lost(&dev->port[event->element.port_num -
+					     dev->start_port]);
+		break;
+	default:
+		break;
+	}
+}
+
+static void mcast_add_one(struct ib_device *device)
+{
+	struct mcast_device *dev;
+	struct mcast_port *port;
+	int i;
+
+	if (rdma_node_get_transport(device->node_type) != RDMA_TRANSPORT_IB)
+		return;
+
+	dev = kmalloc(sizeof *dev + device->phys_port_cnt * sizeof *port,
+		      GFP_KERNEL);
+	if (!dev)
+		return;
+
+	if (device->node_type == RDMA_NODE_IB_SWITCH)
+		dev->start_port = dev->end_port = 0;
+	else {
+		dev->start_port = 1;
+		dev->end_port = device->phys_port_cnt;
+	}
+
+	for (i = 0; i <= dev->end_port - dev->start_port; i++) {
+		port = &dev->port[i];
+		port->dev = dev;
+		port->port_num = dev->start_port + i;
+		spin_lock_init(&port->lock);
+		port->table = RB_ROOT;
+		init_completion(&port->comp);
+		atomic_set(&port->refcount, 1);
+	}
+
+	dev->device = device;
+	ib_set_client_data(device, &mcast_client, dev);
+
+	INIT_IB_EVENT_HANDLER(&dev->event_handler, device, mcast_event_handler);
+	ib_register_event_handler(&dev->event_handler);
+}
+
+static void mcast_remove_one(struct ib_device *device)
+{
+	struct mcast_device *dev;
+	struct mcast_port *port;
+	int i;
+
+	dev = ib_get_client_data(device, &mcast_client);
+	if (!dev)
+		return;
+
+	ib_unregister_event_handler(&dev->event_handler);
+	flush_workqueue(mcast_wq);
+
+	for (i = 0; i <= dev->end_port - dev->start_port; i++) {
+		port = &dev->port[i];
+		deref_port(port);
+		wait_for_completion(&port->comp);
+	}
+
+	kfree(dev);
+}
+
+int mcast_init(void)
+{
+	int ret;
+
+	mcast_wq = create_singlethread_workqueue("ib_mcast");
+	if (!mcast_wq)
+		return -ENOMEM;
+
+	ib_sa_register_client(&sa_client);
+
+	ret = ib_register_client(&mcast_client);
+	if (ret)
+		goto err;
+	return 0;
+
+err:
+	ib_sa_unregister_client(&sa_client);
+	destroy_workqueue(mcast_wq);
+	return ret;
+}
+
+void mcast_cleanup(void)
+{
+	ib_unregister_client(&mcast_client);
+	ib_sa_unregister_client(&sa_client);
+	destroy_workqueue(mcast_wq);
+}

drivers/infiniband/core/packer.c

@@ -0,0 +1,204 @@
+/*
+ * Copyright (c) 2004 Topspin Corporation.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. 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.
+ *
+ * $Id: packer.c,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#include <linux/string.h>
+
+#include <rdma/ib_pack.h>
+
+static u64 value_read(int offset, int size, void *structure)
+{
+	switch (size) {
+	case 1: return                *(u8  *) (structure + offset);
+	case 2: return be16_to_cpup((__be16 *) (structure + offset));
+	case 4: return be32_to_cpup((__be32 *) (structure + offset));
+	case 8: return be64_to_cpup((__be64 *) (structure + offset));
+	default:
+		printk(KERN_WARNING "Field size %d bits not handled\n", size * 8);
+		return 0;
+	}
+}
+
+/**
+ * ib_pack - Pack a structure into a buffer
+ * @desc:Array of structure field descriptions
+ * @desc_len:Number of entries in @desc
+ * @structure:Structure to pack from
+ * @buf:Buffer to pack into
+ *
+ * ib_pack() packs a list of structure fields into a buffer,
+ * controlled by the array of fields in @desc.
+ */
+void ib_pack(const struct ib_field        *desc,
+	     int                           desc_len,
+	     void                         *structure,
+	     void                         *buf)
+{
+	int i;
+
+	for (i = 0; i < desc_len; ++i) {
+		if (desc[i].size_bits <= 32) {
+			int shift;
+			u32 val;
+			__be32 mask;
+			__be32 *addr;
+
+			shift = 32 - desc[i].offset_bits - desc[i].size_bits;
+			if (desc[i].struct_size_bytes)
+				val = value_read(desc[i].struct_offset_bytes,
+						 desc[i].struct_size_bytes,
+						 structure) << shift;
+			else
+				val = 0;
+
+			mask = cpu_to_be32(((1ull << desc[i].size_bits) - 1) << shift);
+			addr = (__be32 *) buf + desc[i].offset_words;
+			*addr = (*addr & ~mask) | (cpu_to_be32(val) & mask);
+		} else if (desc[i].size_bits <= 64) {
+			int shift;
+			u64 val;
+			__be64 mask;
+			__be64 *addr;
+
+			shift = 64 - desc[i].offset_bits - desc[i].size_bits;
+			if (desc[i].struct_size_bytes)
+				val = value_read(desc[i].struct_offset_bytes,
+						 desc[i].struct_size_bytes,
+						 structure) << shift;
+			else
+				val = 0;
+
+			mask = cpu_to_be64((~0ull >> (64 - desc[i].size_bits)) << shift);
+			addr = (__be64 *) ((__be32 *) buf + desc[i].offset_words);
+			*addr = (*addr & ~mask) | (cpu_to_be64(val) & mask);
+		} else {
+			if (desc[i].offset_bits % 8 ||
+			    desc[i].size_bits   % 8) {
+				printk(KERN_WARNING "Structure field %s of size %d "
+				       "bits is not byte-aligned\n",
+				       desc[i].field_name, desc[i].size_bits);
+			}
+
+			if (desc[i].struct_size_bytes)
+				memcpy(buf + desc[i].offset_words * 4 +
+				       desc[i].offset_bits / 8,
+				       structure + desc[i].struct_offset_bytes,
+				       desc[i].size_bits / 8);
+			else
+				memset(buf + desc[i].offset_words * 4 +
+				       desc[i].offset_bits / 8,
+				       0,
+				       desc[i].size_bits / 8);
+		}
+	}
+}
+EXPORT_SYMBOL(ib_pack);
+
+static void value_write(int offset, int size, u64 val, void *structure)
+{
+	switch (size * 8) {
+	case 8:  *(    u8 *) (structure + offset) = val; break;
+	case 16: *(__be16 *) (structure + offset) = cpu_to_be16(val); break;
+	case 32: *(__be32 *) (structure + offset) = cpu_to_be32(val); break;
+	case 64: *(__be64 *) (structure + offset) = cpu_to_be64(val); break;
+	default:
+		printk(KERN_WARNING "Field size %d bits not handled\n", size * 8);
+	}
+}
+
+/**
+ * ib_unpack - Unpack a buffer into a structure
+ * @desc:Array of structure field descriptions
+ * @desc_len:Number of entries in @desc
+ * @buf:Buffer to unpack from
+ * @structure:Structure to unpack into
+ *
+ * ib_pack() unpacks a list of structure fields from a buffer,
+ * controlled by the array of fields in @desc.
+ */
+void ib_unpack(const struct ib_field        *desc,
+	       int                           desc_len,
+	       void                         *buf,
+	       void                         *structure)
+{
+	int i;
+
+	for (i = 0; i < desc_len; ++i) {
+		if (!desc[i].struct_size_bytes)
+			continue;
+
+		if (desc[i].size_bits <= 32) {
+			int shift;
+			u32  val;
+			u32  mask;
+			__be32 *addr;
+
+			shift = 32 - desc[i].offset_bits - desc[i].size_bits;
+			mask = ((1ull << desc[i].size_bits) - 1) << shift;
+			addr = (__be32 *) buf + desc[i].offset_words;
+			val = (be32_to_cpup(addr) & mask) >> shift;
+			value_write(desc[i].struct_offset_bytes,
+				    desc[i].struct_size_bytes,
+				    val,
+				    structure);
+		} else if (desc[i].size_bits <= 64) {
+			int shift;
+			u64  val;
+			u64  mask;
+			__be64 *addr;
+
+			shift = 64 - desc[i].offset_bits - desc[i].size_bits;
+			mask = (~0ull >> (64 - desc[i].size_bits)) << shift;
+			addr = (__be64 *) buf + desc[i].offset_words;
+			val = (be64_to_cpup(addr) & mask) >> shift;
+			value_write(desc[i].struct_offset_bytes,
+				    desc[i].struct_size_bytes,
+				    val,
+				    structure);
+		} else {
+			if (desc[i].offset_bits % 8 ||
+			    desc[i].size_bits   % 8) {
+				printk(KERN_WARNING "Structure field %s of size %d "
+				       "bits is not byte-aligned\n",
+				       desc[i].field_name, desc[i].size_bits);
+			}
+
+			memcpy(structure + desc[i].struct_offset_bytes,
+			       buf + desc[i].offset_words * 4 +
+			       desc[i].offset_bits / 8,
+			       desc[i].size_bits / 8);
+		}
+	}
+}
+EXPORT_SYMBOL(ib_unpack);

drivers/infiniband/core/sa.h

@@ -0,0 +1,66 @@
+/*
+ * Copyright (c) 2004 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Voltaire, Inc.  All rights reserved.
+ * Copyright (c) 2006 Intel Corporation.  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.
+ */
+
+#ifndef SA_H
+#define SA_H
+
+#include <rdma/ib_sa.h>
+
+static inline void ib_sa_client_get(struct ib_sa_client *client)
+{
+	atomic_inc(&client->users);
+}
+
+static inline void ib_sa_client_put(struct ib_sa_client *client)
+{
+	if (atomic_dec_and_test(&client->users))
+		complete(&client->comp);
+}
+
+int ib_sa_mcmember_rec_query(struct ib_sa_client *client,
+			     struct ib_device *device, u8 port_num,
+			     u8 method,
+			     struct ib_sa_mcmember_rec *rec,
+			     ib_sa_comp_mask comp_mask,
+			     int timeout_ms, gfp_t gfp_mask,
+			     void (*callback)(int status,
+					      struct ib_sa_mcmember_rec *resp,
+					      void *context),
+			     void *context,
+			     struct ib_sa_query **sa_query);
+
+int mcast_init(void);
+void mcast_cleanup(void);
+
+#endif /* SA_H */

drivers/infiniband/core/smi.c

@@ -0,0 +1,235 @@
+/*
+ * Copyright (c) 2004, 2005 Mellanox Technologies Ltd.  All rights reserved.
+ * Copyright (c) 2004, 2005 Infinicon Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Intel Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Topspin Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Voltaire Corporation.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. 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.
+ *
+ * $Id: smi.c,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#include <rdma/ib_smi.h>
+#include "smi.h"
+
+/*
+ * Fixup a directed route SMP for sending
+ * Return 0 if the SMP should be discarded
+ */
+int smi_handle_dr_smp_send(struct ib_smp *smp,
+			   u8 node_type,
+			   int port_num)
+{
+	u8 hop_ptr, hop_cnt;
+
+	hop_ptr = smp->hop_ptr;
+	hop_cnt = smp->hop_cnt;
+
+	/* See section 14.2.2.2, Vol 1 IB spec */
+	if (!ib_get_smp_direction(smp)) {
+		/* C14-9:1 */
+		if (hop_cnt && hop_ptr == 0) {
+			smp->hop_ptr++;
+			return (smp->initial_path[smp->hop_ptr] ==
+				port_num);
+		}
+
+		/* C14-9:2 */
+		if (hop_ptr && hop_ptr < hop_cnt) {
+			if (node_type != RDMA_NODE_IB_SWITCH)
+				return 0;
+
+			/* smp->return_path set when received */
+			smp->hop_ptr++;
+			return (smp->initial_path[smp->hop_ptr] ==
+				port_num);
+		}
+
+		/* C14-9:3 -- We're at the end of the DR segment of path */
+		if (hop_ptr == hop_cnt) {
+			/* smp->return_path set when received */
+			smp->hop_ptr++;
+			return (node_type == RDMA_NODE_IB_SWITCH ||
+				smp->dr_dlid == IB_LID_PERMISSIVE);
+		}
+
+		/* C14-9:4 -- hop_ptr = hop_cnt + 1 -> give to SMA/SM */
+		/* C14-9:5 -- Fail unreasonable hop pointer */
+		return (hop_ptr == hop_cnt + 1);
+
+	} else {
+		/* C14-13:1 */
+		if (hop_cnt && hop_ptr == hop_cnt + 1) {
+			smp->hop_ptr--;
+			return (smp->return_path[smp->hop_ptr] ==
+				port_num);
+		}
+
+		/* C14-13:2 */
+		if (2 <= hop_ptr && hop_ptr <= hop_cnt) {
+			if (node_type != RDMA_NODE_IB_SWITCH)
+				return 0;
+
+			smp->hop_ptr--;
+			return (smp->return_path[smp->hop_ptr] ==
+				port_num);
+		}
+
+		/* C14-13:3 -- at the end of the DR segment of path */
+		if (hop_ptr == 1) {
+			smp->hop_ptr--;
+			/* C14-13:3 -- SMPs destined for SM shouldn't be here */
+			return (node_type == RDMA_NODE_IB_SWITCH ||
+				smp->dr_slid == IB_LID_PERMISSIVE);
+		}
+
+		/* C14-13:4 -- hop_ptr = 0 -> should have gone to SM */
+		if (hop_ptr == 0)
+			return 1;
+
+		/* C14-13:5 -- Check for unreasonable hop pointer */
+		return 0;
+	}
+}
+
+/*
+ * Adjust information for a received SMP
+ * Return 0 if the SMP should be dropped
+ */
+int smi_handle_dr_smp_recv(struct ib_smp *smp,
+			   u8 node_type,
+			   int port_num,
+			   int phys_port_cnt)
+{
+	u8 hop_ptr, hop_cnt;
+
+	hop_ptr = smp->hop_ptr;
+	hop_cnt = smp->hop_cnt;
+
+	/* See section 14.2.2.2, Vol 1 IB spec */
+	if (!ib_get_smp_direction(smp)) {
+		/* C14-9:1 -- sender should have incremented hop_ptr */
+		if (hop_cnt && hop_ptr == 0)
+			return 0;
+
+		/* C14-9:2 -- intermediate hop */
+		if (hop_ptr && hop_ptr < hop_cnt) {
+			if (node_type != RDMA_NODE_IB_SWITCH)
+				return 0;
+
+			smp->return_path[hop_ptr] = port_num;
+			/* smp->hop_ptr updated when sending */
+			return (smp->initial_path[hop_ptr+1] <= phys_port_cnt);
+		}
+
+		/* C14-9:3 -- We're at the end of the DR segment of path */
+		if (hop_ptr == hop_cnt) {
+			if (hop_cnt)
+				smp->return_path[hop_ptr] = port_num;
+			/* smp->hop_ptr updated when sending */
+
+			return (node_type == RDMA_NODE_IB_SWITCH ||
+				smp->dr_dlid == IB_LID_PERMISSIVE);
+		}
+
+		/* C14-9:4 -- hop_ptr = hop_cnt + 1 -> give to SMA/SM */
+		/* C14-9:5 -- fail unreasonable hop pointer */
+		return (hop_ptr == hop_cnt + 1);
+
+	} else {
+
+		/* C14-13:1 */
+		if (hop_cnt && hop_ptr == hop_cnt + 1) {
+			smp->hop_ptr--;
+			return (smp->return_path[smp->hop_ptr] ==
+				port_num);
+		}
+
+		/* C14-13:2 */
+		if (2 <= hop_ptr && hop_ptr <= hop_cnt) {
+			if (node_type != RDMA_NODE_IB_SWITCH)
+				return 0;
+
+			/* smp->hop_ptr updated when sending */
+			return (smp->return_path[hop_ptr-1] <= phys_port_cnt);
+		}
+
+		/* C14-13:3 -- We're at the end of the DR segment of path */
+		if (hop_ptr == 1) {
+			if (smp->dr_slid == IB_LID_PERMISSIVE) {
+				/* giving SMP to SM - update hop_ptr */
+				smp->hop_ptr--;
+				return 1;
+			}
+			/* smp->hop_ptr updated when sending */
+			return (node_type == RDMA_NODE_IB_SWITCH);
+		}
+
+		/* C14-13:4 -- hop_ptr = 0 -> give to SM */
+		/* C14-13:5 -- Check for unreasonable hop pointer */
+		return (hop_ptr == 0);
+	}
+}
+
+/*
+ * Return 1 if the received DR SMP should be forwarded to the send queue
+ * Return 0 if the SMP should be completed up the stack
+ */
+int smi_check_forward_dr_smp(struct ib_smp *smp)
+{
+	u8 hop_ptr, hop_cnt;
+
+	hop_ptr = smp->hop_ptr;
+	hop_cnt = smp->hop_cnt;
+
+	if (!ib_get_smp_direction(smp)) {
+		/* C14-9:2 -- intermediate hop */
+		if (hop_ptr && hop_ptr < hop_cnt)
+			return 1;
+
+		/* C14-9:3 -- at the end of the DR segment of path */
+		if (hop_ptr == hop_cnt)
+			return (smp->dr_dlid == IB_LID_PERMISSIVE);
+
+		/* C14-9:4 -- hop_ptr = hop_cnt + 1 -> give to SMA/SM */
+		if (hop_ptr == hop_cnt + 1)
+			return 1;
+	} else {
+		/* C14-13:2 */
+		if (2 <= hop_ptr && hop_ptr <= hop_cnt)
+			return 1;
+
+		/* C14-13:3 -- at the end of the DR segment of path */
+		if (hop_ptr == 1)
+			return (smp->dr_slid != IB_LID_PERMISSIVE);
+	}
+	return 0;
+}

drivers/infiniband/core/smi.h

@@ -0,0 +1,66 @@
+/*
+ * Copyright (c) 2004 Mellanox Technologies Ltd.  All rights reserved.
+ * Copyright (c) 2004 Infinicon Corporation.  All rights reserved.
+ * Copyright (c) 2004 Intel Corporation.  All rights reserved.
+ * Copyright (c) 2004 Topspin Corporation.  All rights reserved.
+ * Copyright (c) 2004 Voltaire Corporation.  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.
+ *
+ * $Id: smi.h,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#ifndef __SMI_H_
+#define __SMI_H_
+
+#include <rdma/ib_smi.h>
+
+int smi_handle_dr_smp_recv(struct ib_smp *smp,
+			   u8 node_type,
+			   int port_num,
+			   int phys_port_cnt);
+extern int smi_check_forward_dr_smp(struct ib_smp *smp);
+extern int smi_handle_dr_smp_send(struct ib_smp *smp,
+				  u8 node_type,
+				  int port_num);
+
+/*
+ * Return 1 if the SMP should be handled by the local SMA/SM via process_mad
+ */
+static inline int smi_check_local_smp(struct ib_smp *smp,
+				      struct ib_device *device)
+{
+	/* C14-9:3 -- We're at the end of the DR segment of path */
+	/* C14-9:4 -- Hop Pointer = Hop Count + 1 -> give to SMA/SM */
+	return ((device->process_mad &&
+		!ib_get_smp_direction(smp) &&
+		(smp->hop_ptr == smp->hop_cnt + 1)));
+}
+
+#endif	/* __SMI_H_ */

drivers/infiniband/core/sysfs.c

@@ -0,0 +1,776 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies Ltd.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. 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.
+ *
+ * $Id: sysfs.c,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#include "core_priv.h"
+
+#include <linux/slab.h>
+#include <linux/string.h>
+
+#include <rdma/ib_mad.h>
+
+struct ib_port {
+	struct kobject         kobj;
+	struct ib_device      *ibdev;
+	struct attribute_group gid_group;
+	struct attribute_group pkey_group;
+	u8                     port_num;
+};
+
+struct port_attribute {
+	struct attribute attr;
+	ssize_t (*show)(struct ib_port *, struct port_attribute *, char *buf);
+	ssize_t (*store)(struct ib_port *, struct port_attribute *,
+			 const char *buf, size_t count);
+};
+
+#define PORT_ATTR(_name, _mode, _show, _store) \
+struct port_attribute port_attr_##_name = __ATTR(_name, _mode, _show, _store)
+
+#define PORT_ATTR_RO(_name) \
+struct port_attribute port_attr_##_name = __ATTR_RO(_name)
+
+struct port_table_attribute {
+	struct port_attribute	attr;
+	char			name[8];
+	int			index;
+};
+
+static inline int ibdev_is_alive(const struct ib_device *dev)
+{
+	return dev->reg_state == IB_DEV_REGISTERED;
+}
+
+static ssize_t port_attr_show(struct kobject *kobj,
+			      struct attribute *attr, char *buf)
+{
+	struct port_attribute *port_attr =
+		container_of(attr, struct port_attribute, attr);
+	struct ib_port *p = container_of(kobj, struct ib_port, kobj);
+
+	if (!port_attr->show)
+		return -EIO;
+	if (!ibdev_is_alive(p->ibdev))
+		return -ENODEV;
+
+	return port_attr->show(p, port_attr, buf);
+}
+
+static struct sysfs_ops port_sysfs_ops = {
+	.show = port_attr_show
+};
+
+static ssize_t state_show(struct ib_port *p, struct port_attribute *unused,
+			  char *buf)
+{
+	struct ib_port_attr attr;
+	ssize_t ret;
+
+	static const char *state_name[] = {
+		[IB_PORT_NOP]		= "NOP",
+		[IB_PORT_DOWN]		= "DOWN",
+		[IB_PORT_INIT]		= "INIT",
+		[IB_PORT_ARMED]		= "ARMED",
+		[IB_PORT_ACTIVE]	= "ACTIVE",
+		[IB_PORT_ACTIVE_DEFER]	= "ACTIVE_DEFER"
+	};
+
+	ret = ib_query_port(p->ibdev, p->port_num, &attr);
+	if (ret)
+		return ret;
+
+	return sprintf(buf, "%d: %s\n", attr.state,
+		       attr.state >= 0 && attr.state < ARRAY_SIZE(state_name) ?
+		       state_name[attr.state] : "UNKNOWN");
+}
+
+static ssize_t lid_show(struct ib_port *p, struct port_attribute *unused,
+			char *buf)
+{
+	struct ib_port_attr attr;
+	ssize_t ret;
+
+	ret = ib_query_port(p->ibdev, p->port_num, &attr);
+	if (ret)
+		return ret;
+
+	return sprintf(buf, "0x%x\n", attr.lid);
+}
+
+static ssize_t lid_mask_count_show(struct ib_port *p,
+				   struct port_attribute *unused,
+				   char *buf)
+{
+	struct ib_port_attr attr;
+	ssize_t ret;
+
+	ret = ib_query_port(p->ibdev, p->port_num, &attr);
+	if (ret)
+		return ret;
+
+	return sprintf(buf, "%d\n", attr.lmc);
+}
+
+static ssize_t sm_lid_show(struct ib_port *p, struct port_attribute *unused,
+			   char *buf)
+{
+	struct ib_port_attr attr;
+	ssize_t ret;
+
+	ret = ib_query_port(p->ibdev, p->port_num, &attr);
+	if (ret)
+		return ret;
+
+	return sprintf(buf, "0x%x\n", attr.sm_lid);
+}
+
+static ssize_t sm_sl_show(struct ib_port *p, struct port_attribute *unused,
+			  char *buf)
+{
+	struct ib_port_attr attr;
+	ssize_t ret;
+
+	ret = ib_query_port(p->ibdev, p->port_num, &attr);
+	if (ret)
+		return ret;
+
+	return sprintf(buf, "%d\n", attr.sm_sl);
+}
+
+static ssize_t cap_mask_show(struct ib_port *p, struct port_attribute *unused,
+			     char *buf)
+{
+	struct ib_port_attr attr;
+	ssize_t ret;
+
+	ret = ib_query_port(p->ibdev, p->port_num, &attr);
+	if (ret)
+		return ret;
+
+	return sprintf(buf, "0x%08x\n", attr.port_cap_flags);
+}
+
+static ssize_t rate_show(struct ib_port *p, struct port_attribute *unused,
+			 char *buf)
+{
+	struct ib_port_attr attr;
+	char *speed = "";
+	int rate;
+	ssize_t ret;
+
+	ret = ib_query_port(p->ibdev, p->port_num, &attr);
+	if (ret)
+		return ret;
+
+	switch (attr.active_speed) {
+	case 2: speed = " DDR"; break;
+	case 4: speed = " QDR"; break;
+	}
+
+	rate = 25 * ib_width_enum_to_int(attr.active_width) * attr.active_speed;
+	if (rate < 0)
+		return -EINVAL;
+
+	return sprintf(buf, "%d%s Gb/sec (%dX%s)\n",
+		       rate / 10, rate % 10 ? ".5" : "",
+		       ib_width_enum_to_int(attr.active_width), speed);
+}
+
+static ssize_t phys_state_show(struct ib_port *p, struct port_attribute *unused,
+			       char *buf)
+{
+	struct ib_port_attr attr;
+
+	ssize_t ret;
+
+	ret = ib_query_port(p->ibdev, p->port_num, &attr);
+	if (ret)
+		return ret;
+
+	switch (attr.phys_state) {
+	case 1:  return sprintf(buf, "1: Sleep\n");
+	case 2:  return sprintf(buf, "2: Polling\n");
+	case 3:  return sprintf(buf, "3: Disabled\n");
+	case 4:  return sprintf(buf, "4: PortConfigurationTraining\n");
+	case 5:  return sprintf(buf, "5: LinkUp\n");
+	case 6:  return sprintf(buf, "6: LinkErrorRecovery\n");
+	case 7:  return sprintf(buf, "7: Phy Test\n");
+	default: return sprintf(buf, "%d: <unknown>\n", attr.phys_state);
+	}
+}
+
+static PORT_ATTR_RO(state);
+static PORT_ATTR_RO(lid);
+static PORT_ATTR_RO(lid_mask_count);
+static PORT_ATTR_RO(sm_lid);
+static PORT_ATTR_RO(sm_sl);
+static PORT_ATTR_RO(cap_mask);
+static PORT_ATTR_RO(rate);
+static PORT_ATTR_RO(phys_state);
+
+static struct attribute *port_default_attrs[] = {
+	&port_attr_state.attr,
+	&port_attr_lid.attr,
+	&port_attr_lid_mask_count.attr,
+	&port_attr_sm_lid.attr,
+	&port_attr_sm_sl.attr,
+	&port_attr_cap_mask.attr,
+	&port_attr_rate.attr,
+	&port_attr_phys_state.attr,
+	NULL
+};
+
+static ssize_t show_port_gid(struct ib_port *p, struct port_attribute *attr,
+			     char *buf)
+{
+	struct port_table_attribute *tab_attr =
+		container_of(attr, struct port_table_attribute, attr);
+	union ib_gid gid;
+	ssize_t ret;
+
+	ret = ib_query_gid(p->ibdev, p->port_num, tab_attr->index, &gid);
+	if (ret)
+		return ret;
+
+	return sprintf(buf, "%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x\n",
+		       be16_to_cpu(((__be16 *) gid.raw)[0]),
+		       be16_to_cpu(((__be16 *) gid.raw)[1]),
+		       be16_to_cpu(((__be16 *) gid.raw)[2]),
+		       be16_to_cpu(((__be16 *) gid.raw)[3]),
+		       be16_to_cpu(((__be16 *) gid.raw)[4]),
+		       be16_to_cpu(((__be16 *) gid.raw)[5]),
+		       be16_to_cpu(((__be16 *) gid.raw)[6]),
+		       be16_to_cpu(((__be16 *) gid.raw)[7]));
+}
+
+static ssize_t show_port_pkey(struct ib_port *p, struct port_attribute *attr,
+			      char *buf)
+{
+	struct port_table_attribute *tab_attr =
+		container_of(attr, struct port_table_attribute, attr);
+	u16 pkey;
+	ssize_t ret;
+
+	ret = ib_query_pkey(p->ibdev, p->port_num, tab_attr->index, &pkey);
+	if (ret)
+		return ret;
+
+	return sprintf(buf, "0x%04x\n", pkey);
+}
+
+#define PORT_PMA_ATTR(_name, _counter, _width, _offset)			\
+struct port_table_attribute port_pma_attr_##_name = {			\
+	.attr  = __ATTR(_name, S_IRUGO, show_pma_counter, NULL),	\
+	.index = (_offset) | ((_width) << 16) | ((_counter) << 24)	\
+}
+
+static ssize_t show_pma_counter(struct ib_port *p, struct port_attribute *attr,
+				char *buf)
+{
+	struct port_table_attribute *tab_attr =
+		container_of(attr, struct port_table_attribute, attr);
+	int offset = tab_attr->index & 0xffff;
+	int width  = (tab_attr->index >> 16) & 0xff;
+	struct ib_mad *in_mad  = NULL;
+	struct ib_mad *out_mad = NULL;
+	ssize_t ret;
+
+	if (!p->ibdev->process_mad)
+		return sprintf(buf, "N/A (no PMA)\n");
+
+	in_mad  = kzalloc(sizeof *in_mad, GFP_KERNEL);
+	out_mad = kmalloc(sizeof *in_mad, GFP_KERNEL);
+	if (!in_mad || !out_mad) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	in_mad->mad_hdr.base_version  = 1;
+	in_mad->mad_hdr.mgmt_class    = IB_MGMT_CLASS_PERF_MGMT;
+	in_mad->mad_hdr.class_version = 1;
+	in_mad->mad_hdr.method        = IB_MGMT_METHOD_GET;
+	in_mad->mad_hdr.attr_id       = cpu_to_be16(0x12); /* PortCounters */
+
+	in_mad->data[41] = p->port_num;	/* PortSelect field */
+
+	if ((p->ibdev->process_mad(p->ibdev, IB_MAD_IGNORE_MKEY,
+		 p->port_num, NULL, NULL, in_mad, out_mad) &
+	     (IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY)) !=
+	    (IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY)) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	switch (width) {
+	case 4:
+		ret = sprintf(buf, "%u\n", (out_mad->data[40 + offset / 8] >>
+					    (4 - (offset % 8))) & 0xf);
+		break;
+	case 8:
+		ret = sprintf(buf, "%u\n", out_mad->data[40 + offset / 8]);
+		break;
+	case 16:
+		ret = sprintf(buf, "%u\n",
+			      be16_to_cpup((__be16 *)(out_mad->data + 40 + offset / 8)));
+		break;
+	case 32:
+		ret = sprintf(buf, "%u\n",
+			      be32_to_cpup((__be32 *)(out_mad->data + 40 + offset / 8)));
+		break;
+	default:
+		ret = 0;
+	}
+
+out:
+	kfree(in_mad);
+	kfree(out_mad);
+
+	return ret;
+}
+
+static PORT_PMA_ATTR(symbol_error		    ,  0, 16,  32);
+static PORT_PMA_ATTR(link_error_recovery	    ,  1,  8,  48);
+static PORT_PMA_ATTR(link_downed		    ,  2,  8,  56);
+static PORT_PMA_ATTR(port_rcv_errors		    ,  3, 16,  64);
+static PORT_PMA_ATTR(port_rcv_remote_physical_errors,  4, 16,  80);
+static PORT_PMA_ATTR(port_rcv_switch_relay_errors   ,  5, 16,  96);
+static PORT_PMA_ATTR(port_xmit_discards		    ,  6, 16, 112);
+static PORT_PMA_ATTR(port_xmit_constraint_errors    ,  7,  8, 128);
+static PORT_PMA_ATTR(port_rcv_constraint_errors	    ,  8,  8, 136);
+static PORT_PMA_ATTR(local_link_integrity_errors    ,  9,  4, 152);
+static PORT_PMA_ATTR(excessive_buffer_overrun_errors, 10,  4, 156);
+static PORT_PMA_ATTR(VL15_dropped		    , 11, 16, 176);
+static PORT_PMA_ATTR(port_xmit_data		    , 12, 32, 192);
+static PORT_PMA_ATTR(port_rcv_data		    , 13, 32, 224);
+static PORT_PMA_ATTR(port_xmit_packets		    , 14, 32, 256);
+static PORT_PMA_ATTR(port_rcv_packets		    , 15, 32, 288);
+
+static struct attribute *pma_attrs[] = {
+	&port_pma_attr_symbol_error.attr.attr,
+	&port_pma_attr_link_error_recovery.attr.attr,
+	&port_pma_attr_link_downed.attr.attr,
+	&port_pma_attr_port_rcv_errors.attr.attr,
+	&port_pma_attr_port_rcv_remote_physical_errors.attr.attr,
+	&port_pma_attr_port_rcv_switch_relay_errors.attr.attr,
+	&port_pma_attr_port_xmit_discards.attr.attr,
+	&port_pma_attr_port_xmit_constraint_errors.attr.attr,
+	&port_pma_attr_port_rcv_constraint_errors.attr.attr,
+	&port_pma_attr_local_link_integrity_errors.attr.attr,
+	&port_pma_attr_excessive_buffer_overrun_errors.attr.attr,
+	&port_pma_attr_VL15_dropped.attr.attr,
+	&port_pma_attr_port_xmit_data.attr.attr,
+	&port_pma_attr_port_rcv_data.attr.attr,
+	&port_pma_attr_port_xmit_packets.attr.attr,
+	&port_pma_attr_port_rcv_packets.attr.attr,
+	NULL
+};
+
+static struct attribute_group pma_group = {
+	.name  = "counters",
+	.attrs  = pma_attrs
+};
+
+static void ib_port_release(struct kobject *kobj)
+{
+	struct ib_port *p = container_of(kobj, struct ib_port, kobj);
+	struct attribute *a;
+	int i;
+
+	for (i = 0; (a = p->gid_group.attrs[i]); ++i)
+		kfree(a);
+
+	kfree(p->gid_group.attrs);
+
+	for (i = 0; (a = p->pkey_group.attrs[i]); ++i)
+		kfree(a);
+
+	kfree(p->pkey_group.attrs);
+
+	kfree(p);
+}
+
+static struct kobj_type port_type = {
+	.release       = ib_port_release,
+	.sysfs_ops     = &port_sysfs_ops,
+	.default_attrs = port_default_attrs
+};
+
+static void ib_device_release(struct class_device *cdev)
+{
+	struct ib_device *dev = container_of(cdev, struct ib_device, class_dev);
+
+	kfree(dev);
+}
+
+static int ib_device_uevent(struct class_device *cdev, char **envp,
+			    int num_envp, char *buf, int size)
+{
+	struct ib_device *dev = container_of(cdev, struct ib_device, class_dev);
+	int i = 0, len = 0;
+
+	if (add_uevent_var(envp, num_envp, &i, buf, size, &len,
+			   "NAME=%s", dev->name))
+		return -ENOMEM;
+
+	/*
+	 * It would be nice to pass the node GUID with the event...
+	 */
+
+	envp[i] = NULL;
+	return 0;
+}
+
+static struct attribute **
+alloc_group_attrs(ssize_t (*show)(struct ib_port *,
+				  struct port_attribute *, char *buf),
+		  int len)
+{
+	struct attribute **tab_attr;
+	struct port_table_attribute *element;
+	int i;
+
+	tab_attr = kcalloc(1 + len, sizeof(struct attribute *), GFP_KERNEL);
+	if (!tab_attr)
+		return NULL;
+
+	for (i = 0; i < len; i++) {
+		element = kzalloc(sizeof(struct port_table_attribute),
+				  GFP_KERNEL);
+		if (!element)
+			goto err;
+
+		if (snprintf(element->name, sizeof(element->name),
+			     "%d", i) >= sizeof(element->name)) {
+			kfree(element);
+			goto err;
+		}
+
+		element->attr.attr.name  = element->name;
+		element->attr.attr.mode  = S_IRUGO;
+		element->attr.attr.owner = THIS_MODULE;
+		element->attr.show       = show;
+		element->index		 = i;
+
+		tab_attr[i] = &element->attr.attr;
+	}
+
+	return tab_attr;
+
+err:
+	while (--i >= 0)
+		kfree(tab_attr[i]);
+	kfree(tab_attr);
+	return NULL;
+}
+
+static int add_port(struct ib_device *device, int port_num)
+{
+	struct ib_port *p;
+	struct ib_port_attr attr;
+	int i;
+	int ret;
+
+	ret = ib_query_port(device, port_num, &attr);
+	if (ret)
+		return ret;
+
+	p = kzalloc(sizeof *p, GFP_KERNEL);
+	if (!p)
+		return -ENOMEM;
+
+	p->ibdev      = device;
+	p->port_num   = port_num;
+	p->kobj.ktype = &port_type;
+
+	p->kobj.parent = kobject_get(&device->ports_parent);
+	if (!p->kobj.parent) {
+		ret = -EBUSY;
+		goto err;
+	}
+
+	ret = kobject_set_name(&p->kobj, "%d", port_num);
+	if (ret)
+		goto err_put;
+
+	ret = kobject_register(&p->kobj);
+	if (ret)
+		goto err_put;
+
+	ret = sysfs_create_group(&p->kobj, &pma_group);
+	if (ret)
+		goto err_put;
+
+	p->gid_group.name  = "gids";
+	p->gid_group.attrs = alloc_group_attrs(show_port_gid, attr.gid_tbl_len);
+	if (!p->gid_group.attrs)
+		goto err_remove_pma;
+
+	ret = sysfs_create_group(&p->kobj, &p->gid_group);
+	if (ret)
+		goto err_free_gid;
+
+	p->pkey_group.name  = "pkeys";
+	p->pkey_group.attrs = alloc_group_attrs(show_port_pkey,
+						attr.pkey_tbl_len);
+	if (!p->pkey_group.attrs)
+		goto err_remove_gid;
+
+	ret = sysfs_create_group(&p->kobj, &p->pkey_group);
+	if (ret)
+		goto err_free_pkey;
+
+	list_add_tail(&p->kobj.entry, &device->port_list);
+
+	return 0;
+
+err_free_pkey:
+	for (i = 0; i < attr.pkey_tbl_len; ++i)
+		kfree(p->pkey_group.attrs[i]);
+
+	kfree(p->pkey_group.attrs);
+
+err_remove_gid:
+	sysfs_remove_group(&p->kobj, &p->gid_group);
+
+err_free_gid:
+	for (i = 0; i < attr.gid_tbl_len; ++i)
+		kfree(p->gid_group.attrs[i]);
+
+	kfree(p->gid_group.attrs);
+
+err_remove_pma:
+	sysfs_remove_group(&p->kobj, &pma_group);
+
+err_put:
+	kobject_put(&device->ports_parent);
+
+err:
+	kfree(p);
+	return ret;
+}
+
+static ssize_t show_node_type(struct class_device *cdev, char *buf)
+{
+	struct ib_device *dev = container_of(cdev, struct ib_device, class_dev);
+
+	if (!ibdev_is_alive(dev))
+		return -ENODEV;
+
+	switch (dev->node_type) {
+	case RDMA_NODE_IB_CA:	  return sprintf(buf, "%d: CA\n", dev->node_type);
+	case RDMA_NODE_RNIC:	  return sprintf(buf, "%d: RNIC\n", dev->node_type);
+	case RDMA_NODE_IB_SWITCH: return sprintf(buf, "%d: switch\n", dev->node_type);
+	case RDMA_NODE_IB_ROUTER: return sprintf(buf, "%d: router\n", dev->node_type);
+	default:		  return sprintf(buf, "%d: <unknown>\n", dev->node_type);
+	}
+}
+
+static ssize_t show_sys_image_guid(struct class_device *cdev, char *buf)
+{
+	struct ib_device *dev = container_of(cdev, struct ib_device, class_dev);
+	struct ib_device_attr attr;
+	ssize_t ret;
+
+	if (!ibdev_is_alive(dev))
+		return -ENODEV;
+
+	ret = ib_query_device(dev, &attr);
+	if (ret)
+		return ret;
+
+	return sprintf(buf, "%04x:%04x:%04x:%04x\n",
+		       be16_to_cpu(((__be16 *) &attr.sys_image_guid)[0]),
+		       be16_to_cpu(((__be16 *) &attr.sys_image_guid)[1]),
+		       be16_to_cpu(((__be16 *) &attr.sys_image_guid)[2]),
+		       be16_to_cpu(((__be16 *) &attr.sys_image_guid)[3]));
+}
+
+static ssize_t show_node_guid(struct class_device *cdev, char *buf)
+{
+	struct ib_device *dev = container_of(cdev, struct ib_device, class_dev);
+
+	if (!ibdev_is_alive(dev))
+		return -ENODEV;
+
+	return sprintf(buf, "%04x:%04x:%04x:%04x\n",
+		       be16_to_cpu(((__be16 *) &dev->node_guid)[0]),
+		       be16_to_cpu(((__be16 *) &dev->node_guid)[1]),
+		       be16_to_cpu(((__be16 *) &dev->node_guid)[2]),
+		       be16_to_cpu(((__be16 *) &dev->node_guid)[3]));
+}
+
+static ssize_t show_node_desc(struct class_device *cdev, char *buf)
+{
+	struct ib_device *dev = container_of(cdev, struct ib_device, class_dev);
+
+	return sprintf(buf, "%.64s\n", dev->node_desc);
+}
+
+static ssize_t set_node_desc(struct class_device *cdev, const char *buf,
+			      size_t count)
+{
+	struct ib_device *dev = container_of(cdev, struct ib_device, class_dev);
+	struct ib_device_modify desc = {};
+	int ret;
+
+	if (!dev->modify_device)
+		return -EIO;
+
+	memcpy(desc.node_desc, buf, min_t(int, count, 64));
+	ret = ib_modify_device(dev, IB_DEVICE_MODIFY_NODE_DESC, &desc);
+	if (ret)
+		return ret;
+
+	return count;
+}
+
+static CLASS_DEVICE_ATTR(node_type, S_IRUGO, show_node_type, NULL);
+static CLASS_DEVICE_ATTR(sys_image_guid, S_IRUGO, show_sys_image_guid, NULL);
+static CLASS_DEVICE_ATTR(node_guid, S_IRUGO, show_node_guid, NULL);
+static CLASS_DEVICE_ATTR(node_desc, S_IRUGO | S_IWUSR, show_node_desc,
+			 set_node_desc);
+
+static struct class_device_attribute *ib_class_attributes[] = {
+	&class_device_attr_node_type,
+	&class_device_attr_sys_image_guid,
+	&class_device_attr_node_guid,
+	&class_device_attr_node_desc
+};
+
+static struct class ib_class = {
+	.name    = "infiniband",
+	.release = ib_device_release,
+	.uevent = ib_device_uevent,
+};
+
+int ib_device_register_sysfs(struct ib_device *device)
+{
+	struct class_device *class_dev = &device->class_dev;
+	int ret;
+	int i;
+
+	class_dev->class      = &ib_class;
+	class_dev->class_data = device;
+	strlcpy(class_dev->class_id, device->name, BUS_ID_SIZE);
+
+	INIT_LIST_HEAD(&device->port_list);
+
+	ret = class_device_register(class_dev);
+	if (ret)
+		goto err;
+
+	for (i = 0; i < ARRAY_SIZE(ib_class_attributes); ++i) {
+		ret = class_device_create_file(class_dev, ib_class_attributes[i]);
+		if (ret)
+			goto err_unregister;
+	}
+
+	device->ports_parent.parent = kobject_get(&class_dev->kobj);
+	if (!device->ports_parent.parent) {
+		ret = -EBUSY;
+		goto err_unregister;
+	}
+	ret = kobject_set_name(&device->ports_parent, "ports");
+	if (ret)
+		goto err_put;
+	ret = kobject_register(&device->ports_parent);
+	if (ret)
+		goto err_put;
+
+	if (device->node_type == RDMA_NODE_IB_SWITCH) {
+		ret = add_port(device, 0);
+		if (ret)
+			goto err_put;
+	} else {
+		for (i = 1; i <= device->phys_port_cnt; ++i) {
+			ret = add_port(device, i);
+			if (ret)
+				goto err_put;
+		}
+	}
+
+	return 0;
+
+err_put:
+	{
+		struct kobject *p, *t;
+		struct ib_port *port;
+
+		list_for_each_entry_safe(p, t, &device->port_list, entry) {
+			list_del(&p->entry);
+			port = container_of(p, struct ib_port, kobj);
+			sysfs_remove_group(p, &pma_group);
+			sysfs_remove_group(p, &port->pkey_group);
+			sysfs_remove_group(p, &port->gid_group);
+			kobject_unregister(p);
+		}
+	}
+
+	kobject_put(&class_dev->kobj);
+
+err_unregister:
+	class_device_unregister(class_dev);
+
+err:
+	return ret;
+}
+
+void ib_device_unregister_sysfs(struct ib_device *device)
+{
+	struct kobject *p, *t;
+	struct ib_port *port;
+
+	list_for_each_entry_safe(p, t, &device->port_list, entry) {
+		list_del(&p->entry);
+		port = container_of(p, struct ib_port, kobj);
+		sysfs_remove_group(p, &pma_group);
+		sysfs_remove_group(p, &port->pkey_group);
+		sysfs_remove_group(p, &port->gid_group);
+		kobject_unregister(p);
+	}
+
+	kobject_unregister(&device->ports_parent);
+	class_device_unregister(&device->class_dev);
+}
+
+int ib_sysfs_setup(void)
+{
+	return class_register(&ib_class);
+}
+
+void ib_sysfs_cleanup(void)
+{
+	class_unregister(&ib_class);
+}

drivers/infiniband/core/ud_header.c

@@ -0,0 +1,367 @@
+/*
+ * Copyright (c) 2004 Topspin Corporation.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. 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.
+ *
+ * $Id: ud_header.c,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#include <linux/errno.h>
+#include <linux/string.h>
+
+#include <rdma/ib_pack.h>
+
+#define STRUCT_FIELD(header, field) \
+	.struct_offset_bytes = offsetof(struct ib_unpacked_ ## header, field),      \
+	.struct_size_bytes   = sizeof ((struct ib_unpacked_ ## header *) 0)->field, \
+	.field_name          = #header ":" #field
+
+static const struct ib_field lrh_table[]  = {
+	{ STRUCT_FIELD(lrh, virtual_lane),
+	  .offset_words = 0,
+	  .offset_bits  = 0,
+	  .size_bits    = 4 },
+	{ STRUCT_FIELD(lrh, link_version),
+	  .offset_words = 0,
+	  .offset_bits  = 4,
+	  .size_bits    = 4 },
+	{ STRUCT_FIELD(lrh, service_level),
+	  .offset_words = 0,
+	  .offset_bits  = 8,
+	  .size_bits    = 4 },
+	{ RESERVED,
+	  .offset_words = 0,
+	  .offset_bits  = 12,
+	  .size_bits    = 2 },
+	{ STRUCT_FIELD(lrh, link_next_header),
+	  .offset_words = 0,
+	  .offset_bits  = 14,
+	  .size_bits    = 2 },
+	{ STRUCT_FIELD(lrh, destination_lid),
+	  .offset_words = 0,
+	  .offset_bits  = 16,
+	  .size_bits    = 16 },
+	{ RESERVED,
+	  .offset_words = 1,
+	  .offset_bits  = 0,
+	  .size_bits    = 5 },
+	{ STRUCT_FIELD(lrh, packet_length),
+	  .offset_words = 1,
+	  .offset_bits  = 5,
+	  .size_bits    = 11 },
+	{ STRUCT_FIELD(lrh, source_lid),
+	  .offset_words = 1,
+	  .offset_bits  = 16,
+	  .size_bits    = 16 }
+};
+
+static const struct ib_field grh_table[]  = {
+	{ STRUCT_FIELD(grh, ip_version),
+	  .offset_words = 0,
+	  .offset_bits  = 0,
+	  .size_bits    = 4 },
+	{ STRUCT_FIELD(grh, traffic_class),
+	  .offset_words = 0,
+	  .offset_bits  = 4,
+	  .size_bits    = 8 },
+	{ STRUCT_FIELD(grh, flow_label),
+	  .offset_words = 0,
+	  .offset_bits  = 12,
+	  .size_bits    = 20 },
+	{ STRUCT_FIELD(grh, payload_length),
+	  .offset_words = 1,
+	  .offset_bits  = 0,
+	  .size_bits    = 16 },
+	{ STRUCT_FIELD(grh, next_header),
+	  .offset_words = 1,
+	  .offset_bits  = 16,
+	  .size_bits    = 8 },
+	{ STRUCT_FIELD(grh, hop_limit),
+	  .offset_words = 1,
+	  .offset_bits  = 24,
+	  .size_bits    = 8 },
+	{ STRUCT_FIELD(grh, source_gid),
+	  .offset_words = 2,
+	  .offset_bits  = 0,
+	  .size_bits    = 128 },
+	{ STRUCT_FIELD(grh, destination_gid),
+	  .offset_words = 6,
+	  .offset_bits  = 0,
+	  .size_bits    = 128 }
+};
+
+static const struct ib_field bth_table[]  = {
+	{ STRUCT_FIELD(bth, opcode),
+	  .offset_words = 0,
+	  .offset_bits  = 0,
+	  .size_bits    = 8 },
+	{ STRUCT_FIELD(bth, solicited_event),
+	  .offset_words = 0,
+	  .offset_bits  = 8,
+	  .size_bits    = 1 },
+	{ STRUCT_FIELD(bth, mig_req),
+	  .offset_words = 0,
+	  .offset_bits  = 9,
+	  .size_bits    = 1 },
+	{ STRUCT_FIELD(bth, pad_count),
+	  .offset_words = 0,
+	  .offset_bits  = 10,
+	  .size_bits    = 2 },
+	{ STRUCT_FIELD(bth, transport_header_version),
+	  .offset_words = 0,
+	  .offset_bits  = 12,
+	  .size_bits    = 4 },
+	{ STRUCT_FIELD(bth, pkey),
+	  .offset_words = 0,
+	  .offset_bits  = 16,
+	  .size_bits    = 16 },
+	{ RESERVED,
+	  .offset_words = 1,
+	  .offset_bits  = 0,
+	  .size_bits    = 8 },
+	{ STRUCT_FIELD(bth, destination_qpn),
+	  .offset_words = 1,
+	  .offset_bits  = 8,
+	  .size_bits    = 24 },
+	{ STRUCT_FIELD(bth, ack_req),
+	  .offset_words = 2,
+	  .offset_bits  = 0,
+	  .size_bits    = 1 },
+	{ RESERVED,
+	  .offset_words = 2,
+	  .offset_bits  = 1,
+	  .size_bits    = 7 },
+	{ STRUCT_FIELD(bth, psn),
+	  .offset_words = 2,
+	  .offset_bits  = 8,
+	  .size_bits    = 24 }
+};
+
+static const struct ib_field deth_table[] = {
+	{ STRUCT_FIELD(deth, qkey),
+	  .offset_words = 0,
+	  .offset_bits  = 0,
+	  .size_bits    = 32 },
+	{ RESERVED,
+	  .offset_words = 1,
+	  .offset_bits  = 0,
+	  .size_bits    = 8 },
+	{ STRUCT_FIELD(deth, source_qpn),
+	  .offset_words = 1,
+	  .offset_bits  = 8,
+	  .size_bits    = 24 }
+};
+
+/**
+ * ib_ud_header_init - Initialize UD header structure
+ * @payload_bytes:Length of packet payload
+ * @grh_present:GRH flag (if non-zero, GRH will be included)
+ * @header:Structure to initialize
+ *
+ * ib_ud_header_init() initializes the lrh.link_version, lrh.link_next_header,
+ * lrh.packet_length, grh.ip_version, grh.payload_length,
+ * grh.next_header, bth.opcode, bth.pad_count and
+ * bth.transport_header_version fields of a &struct ib_ud_header given
+ * the payload length and whether a GRH will be included.
+ */
+void ib_ud_header_init(int     		    payload_bytes,
+		       int    		    grh_present,
+		       struct ib_ud_header *header)
+{
+	int header_len;
+	u16 packet_length;
+
+	memset(header, 0, sizeof *header);
+
+	header_len =
+		IB_LRH_BYTES  +
+		IB_BTH_BYTES  +
+		IB_DETH_BYTES;
+	if (grh_present) {
+		header_len += IB_GRH_BYTES;
+	}
+
+	header->lrh.link_version     = 0;
+	header->lrh.link_next_header =
+		grh_present ? IB_LNH_IBA_GLOBAL : IB_LNH_IBA_LOCAL;
+	packet_length		     = (IB_LRH_BYTES     +
+					IB_BTH_BYTES     +
+					IB_DETH_BYTES    +
+					payload_bytes    +
+					4                + /* ICRC     */
+					3) / 4;            /* round up */
+
+	header->grh_present          = grh_present;
+	if (grh_present) {
+		packet_length		   += IB_GRH_BYTES / 4;
+		header->grh.ip_version      = 6;
+		header->grh.payload_length  =
+			cpu_to_be16((IB_BTH_BYTES     +
+				     IB_DETH_BYTES    +
+				     payload_bytes    +
+				     4                + /* ICRC     */
+				     3) & ~3);          /* round up */
+		header->grh.next_header     = 0x1b;
+	}
+
+	header->lrh.packet_length = cpu_to_be16(packet_length);
+
+	if (header->immediate_present)
+		header->bth.opcode           = IB_OPCODE_UD_SEND_ONLY_WITH_IMMEDIATE;
+	else
+		header->bth.opcode           = IB_OPCODE_UD_SEND_ONLY;
+	header->bth.pad_count                = (4 - payload_bytes) & 3;
+	header->bth.transport_header_version = 0;
+}
+EXPORT_SYMBOL(ib_ud_header_init);
+
+/**
+ * ib_ud_header_pack - Pack UD header struct into wire format
+ * @header:UD header struct
+ * @buf:Buffer to pack into
+ *
+ * ib_ud_header_pack() packs the UD header structure @header into wire
+ * format in the buffer @buf.
+ */
+int ib_ud_header_pack(struct ib_ud_header *header,
+		      void                *buf)
+{
+	int len = 0;
+
+	ib_pack(lrh_table, ARRAY_SIZE(lrh_table),
+		&header->lrh, buf);
+	len += IB_LRH_BYTES;
+
+	if (header->grh_present) {
+		ib_pack(grh_table, ARRAY_SIZE(grh_table),
+			&header->grh, buf + len);
+		len += IB_GRH_BYTES;
+	}
+
+	ib_pack(bth_table, ARRAY_SIZE(bth_table),
+		&header->bth, buf + len);
+	len += IB_BTH_BYTES;
+
+	ib_pack(deth_table, ARRAY_SIZE(deth_table),
+		&header->deth, buf + len);
+	len += IB_DETH_BYTES;
+
+	if (header->immediate_present) {
+		memcpy(buf + len, &header->immediate_data, sizeof header->immediate_data);
+		len += sizeof header->immediate_data;
+	}
+
+	return len;
+}
+EXPORT_SYMBOL(ib_ud_header_pack);
+
+/**
+ * ib_ud_header_unpack - Unpack UD header struct from wire format
+ * @header:UD header struct
+ * @buf:Buffer to pack into
+ *
+ * ib_ud_header_pack() unpacks the UD header structure @header from wire
+ * format in the buffer @buf.
+ */
+int ib_ud_header_unpack(void                *buf,
+			struct ib_ud_header *header)
+{
+	ib_unpack(lrh_table, ARRAY_SIZE(lrh_table),
+		  buf, &header->lrh);
+	buf += IB_LRH_BYTES;
+
+	if (header->lrh.link_version != 0) {
+		printk(KERN_WARNING "Invalid LRH.link_version %d\n",
+		       header->lrh.link_version);
+		return -EINVAL;
+	}
+
+	switch (header->lrh.link_next_header) {
+	case IB_LNH_IBA_LOCAL:
+		header->grh_present = 0;
+		break;
+
+	case IB_LNH_IBA_GLOBAL:
+		header->grh_present = 1;
+		ib_unpack(grh_table, ARRAY_SIZE(grh_table),
+			  buf, &header->grh);
+		buf += IB_GRH_BYTES;
+
+		if (header->grh.ip_version != 6) {
+			printk(KERN_WARNING "Invalid GRH.ip_version %d\n",
+			       header->grh.ip_version);
+			return -EINVAL;
+		}
+		if (header->grh.next_header != 0x1b) {
+			printk(KERN_WARNING "Invalid GRH.next_header 0x%02x\n",
+			       header->grh.next_header);
+			return -EINVAL;
+		}
+		break;
+
+	default:
+		printk(KERN_WARNING "Invalid LRH.link_next_header %d\n",
+		       header->lrh.link_next_header);
+		return -EINVAL;
+	}
+
+	ib_unpack(bth_table, ARRAY_SIZE(bth_table),
+		  buf, &header->bth);
+	buf += IB_BTH_BYTES;
+
+	switch (header->bth.opcode) {
+	case IB_OPCODE_UD_SEND_ONLY:
+		header->immediate_present = 0;
+		break;
+	case IB_OPCODE_UD_SEND_ONLY_WITH_IMMEDIATE:
+		header->immediate_present = 1;
+		break;
+	default:
+		printk(KERN_WARNING "Invalid BTH.opcode 0x%02x\n",
+		       header->bth.opcode);
+		return -EINVAL;
+	}
+
+	if (header->bth.transport_header_version != 0) {
+		printk(KERN_WARNING "Invalid BTH.transport_header_version %d\n",
+		       header->bth.transport_header_version);
+		return -EINVAL;
+	}
+
+	ib_unpack(deth_table, ARRAY_SIZE(deth_table),
+		  buf, &header->deth);
+	buf += IB_DETH_BYTES;
+
+	if (header->immediate_present)
+		memcpy(&header->immediate_data, buf, sizeof header->immediate_data);
+
+	return 0;
+}
+EXPORT_SYMBOL(ib_ud_header_unpack);

drivers/infiniband/core/uverbs.h

@@ -0,0 +1,205 @@
+/*
+ * Copyright (c) 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005, 2006 Cisco Systems.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2005 Voltaire, Inc. All rights reserved.
+ * Copyright (c) 2005 PathScale, Inc. 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.
+ *
+ * $Id: uverbs.h,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#ifndef UVERBS_H
+#define UVERBS_H
+
+#include <linux/kref.h>
+#include <linux/idr.h>
+#include <linux/mutex.h>
+#include <linux/completion.h>
+
+#include <rdma/ib_verbs.h>
+#include <rdma/ib_user_verbs.h>
+
+/*
+ * Our lifetime rules for these structs are the following:
+ *
+ * struct ib_uverbs_device: One reference is held by the module and
+ * released in ib_uverbs_remove_one().  Another reference is taken by
+ * ib_uverbs_open() each time the character special file is opened,
+ * and released in ib_uverbs_release_file() when the file is released.
+ *
+ * struct ib_uverbs_file: One reference is held by the VFS and
+ * released when the file is closed.  Another reference is taken when
+ * an asynchronous event queue file is created and released when the
+ * event file is closed.
+ *
+ * struct ib_uverbs_event_file: One reference is held by the VFS and
+ * released when the file is closed.  For asynchronous event files,
+ * another reference is held by the corresponding main context file
+ * and released when that file is closed.  For completion event files,
+ * a reference is taken when a CQ is created that uses the file, and
+ * released when the CQ is destroyed.
+ */
+
+struct ib_uverbs_device {
+	struct kref				ref;
+	struct completion			comp;
+	int					devnum;
+	struct cdev			       *dev;
+	struct class_device		       *class_dev;
+	struct ib_device		       *ib_dev;
+	int					num_comp_vectors;
+};
+
+struct ib_uverbs_event_file {
+	struct kref				ref;
+	struct file			       *file;
+	struct ib_uverbs_file		       *uverbs_file;
+	spinlock_t				lock;
+	int					is_async;
+	wait_queue_head_t			poll_wait;
+	struct fasync_struct		       *async_queue;
+	struct list_head			event_list;
+};
+
+struct ib_uverbs_file {
+	struct kref				ref;
+	struct mutex				mutex;
+	struct ib_uverbs_device		       *device;
+	struct ib_ucontext		       *ucontext;
+	struct ib_event_handler			event_handler;
+	struct ib_uverbs_event_file	       *async_file;
+};
+
+struct ib_uverbs_event {
+	union {
+		struct ib_uverbs_async_event_desc	async;
+		struct ib_uverbs_comp_event_desc	comp;
+	}					desc;
+	struct list_head			list;
+	struct list_head			obj_list;
+	u32				       *counter;
+};
+
+struct ib_uverbs_mcast_entry {
+	struct list_head	list;
+	union ib_gid 		gid;
+	u16 			lid;
+};
+
+struct ib_uevent_object {
+	struct ib_uobject	uobject;
+	struct list_head	event_list;
+	u32			events_reported;
+};
+
+struct ib_uqp_object {
+	struct ib_uevent_object	uevent;
+	struct list_head 	mcast_list;
+};
+
+struct ib_ucq_object {
+	struct ib_uobject	uobject;
+	struct ib_uverbs_file  *uverbs_file;
+	struct list_head	comp_list;
+	struct list_head	async_list;
+	u32			comp_events_reported;
+	u32			async_events_reported;
+};
+
+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;
+extern struct idr ib_uverbs_ah_idr;
+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);
+struct ib_uverbs_event_file *ib_uverbs_lookup_comp_file(int fd);
+
+void ib_uverbs_release_ucq(struct ib_uverbs_file *file,
+			   struct ib_uverbs_event_file *ev_file,
+			   struct ib_ucq_object *uobj);
+void ib_uverbs_release_uevent(struct ib_uverbs_file *file,
+			      struct ib_uevent_object *uobj);
+
+void ib_uverbs_comp_handler(struct ib_cq *cq, void *cq_context);
+void ib_uverbs_cq_event_handler(struct ib_event *event, void *context_ptr);
+void ib_uverbs_qp_event_handler(struct ib_event *event, void *context_ptr);
+void ib_uverbs_srq_event_handler(struct ib_event *event, void *context_ptr);
+void ib_uverbs_event_handler(struct ib_event_handler *handler,
+			     struct ib_event *event);
+
+int ib_umem_get(struct ib_device *dev, struct ib_umem *mem,
+		void *addr, size_t size, int write);
+void ib_umem_release(struct ib_device *dev, struct ib_umem *umem);
+void ib_umem_release_on_close(struct ib_device *dev, struct ib_umem *umem);
+
+#define IB_UVERBS_DECLARE_CMD(name)					\
+	ssize_t ib_uverbs_##name(struct ib_uverbs_file *file,		\
+				 const char __user *buf, int in_len,	\
+				 int out_len)
+
+IB_UVERBS_DECLARE_CMD(get_context);
+IB_UVERBS_DECLARE_CMD(query_device);
+IB_UVERBS_DECLARE_CMD(query_port);
+IB_UVERBS_DECLARE_CMD(alloc_pd);
+IB_UVERBS_DECLARE_CMD(dealloc_pd);
+IB_UVERBS_DECLARE_CMD(reg_mr);
+IB_UVERBS_DECLARE_CMD(dereg_mr);
+IB_UVERBS_DECLARE_CMD(create_comp_channel);
+IB_UVERBS_DECLARE_CMD(create_cq);
+IB_UVERBS_DECLARE_CMD(resize_cq);
+IB_UVERBS_DECLARE_CMD(poll_cq);
+IB_UVERBS_DECLARE_CMD(req_notify_cq);
+IB_UVERBS_DECLARE_CMD(destroy_cq);
+IB_UVERBS_DECLARE_CMD(create_qp);
+IB_UVERBS_DECLARE_CMD(query_qp);
+IB_UVERBS_DECLARE_CMD(modify_qp);
+IB_UVERBS_DECLARE_CMD(destroy_qp);
+IB_UVERBS_DECLARE_CMD(post_send);
+IB_UVERBS_DECLARE_CMD(post_recv);
+IB_UVERBS_DECLARE_CMD(post_srq_recv);
+IB_UVERBS_DECLARE_CMD(create_ah);
+IB_UVERBS_DECLARE_CMD(destroy_ah);
+IB_UVERBS_DECLARE_CMD(attach_mcast);
+IB_UVERBS_DECLARE_CMD(detach_mcast);
+IB_UVERBS_DECLARE_CMD(create_srq);
+IB_UVERBS_DECLARE_CMD(modify_srq);
+IB_UVERBS_DECLARE_CMD(query_srq);
+IB_UVERBS_DECLARE_CMD(destroy_srq);
+
+#endif /* UVERBS_H */

drivers/infiniband/core/uverbs_main.c

@@ -0,0 +1,920 @@
+/*
+ * Copyright (c) 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005, 2006 Cisco Systems.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2005 Voltaire, Inc. All rights reserved.
+ * Copyright (c) 2005 PathScale, Inc. 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.
+ *
+ * $Id: uverbs_main.c,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/device.h>
+#include <linux/err.h>
+#include <linux/fs.h>
+#include <linux/poll.h>
+#include <linux/file.h>
+#include <linux/mount.h>
+#include <linux/cdev.h>
+
+#include <asm/uaccess.h>
+
+#include "uverbs.h"
+
+MODULE_AUTHOR("Roland Dreier");
+MODULE_DESCRIPTION("InfiniBand userspace verbs access");
+MODULE_LICENSE("Dual BSD/GPL");
+
+#define INFINIBANDEVENTFS_MAGIC	0x49426576	/* "IBev" */
+
+enum {
+	IB_UVERBS_MAJOR       = 231,
+	IB_UVERBS_BASE_MINOR  = 192,
+	IB_UVERBS_MAX_DEVICES = 32
+};
+
+#define IB_UVERBS_BASE_DEV	MKDEV(IB_UVERBS_MAJOR, IB_UVERBS_BASE_MINOR)
+
+static struct class *uverbs_class;
+
+DEFINE_SPINLOCK(ib_uverbs_idr_lock);
+DEFINE_IDR(ib_uverbs_pd_idr);
+DEFINE_IDR(ib_uverbs_mr_idr);
+DEFINE_IDR(ib_uverbs_mw_idr);
+DEFINE_IDR(ib_uverbs_ah_idr);
+DEFINE_IDR(ib_uverbs_cq_idr);
+DEFINE_IDR(ib_uverbs_qp_idr);
+DEFINE_IDR(ib_uverbs_srq_idr);
+
+static spinlock_t map_lock;
+static struct ib_uverbs_device *dev_table[IB_UVERBS_MAX_DEVICES];
+static DECLARE_BITMAP(dev_map, IB_UVERBS_MAX_DEVICES);
+
+static ssize_t (*uverbs_cmd_table[])(struct ib_uverbs_file *file,
+				     const char __user *buf, int in_len,
+				     int out_len) = {
+	[IB_USER_VERBS_CMD_GET_CONTEXT]   	= ib_uverbs_get_context,
+	[IB_USER_VERBS_CMD_QUERY_DEVICE]  	= ib_uverbs_query_device,
+	[IB_USER_VERBS_CMD_QUERY_PORT]    	= ib_uverbs_query_port,
+	[IB_USER_VERBS_CMD_ALLOC_PD]      	= ib_uverbs_alloc_pd,
+	[IB_USER_VERBS_CMD_DEALLOC_PD]    	= ib_uverbs_dealloc_pd,
+	[IB_USER_VERBS_CMD_REG_MR]        	= ib_uverbs_reg_mr,
+	[IB_USER_VERBS_CMD_DEREG_MR]      	= ib_uverbs_dereg_mr,
+	[IB_USER_VERBS_CMD_CREATE_COMP_CHANNEL] = ib_uverbs_create_comp_channel,
+	[IB_USER_VERBS_CMD_CREATE_CQ]     	= ib_uverbs_create_cq,
+	[IB_USER_VERBS_CMD_RESIZE_CQ]     	= ib_uverbs_resize_cq,
+	[IB_USER_VERBS_CMD_POLL_CQ]     	= ib_uverbs_poll_cq,
+	[IB_USER_VERBS_CMD_REQ_NOTIFY_CQ]     	= ib_uverbs_req_notify_cq,
+	[IB_USER_VERBS_CMD_DESTROY_CQ]    	= ib_uverbs_destroy_cq,
+	[IB_USER_VERBS_CMD_CREATE_QP]     	= ib_uverbs_create_qp,
+	[IB_USER_VERBS_CMD_QUERY_QP]     	= ib_uverbs_query_qp,
+	[IB_USER_VERBS_CMD_MODIFY_QP]     	= ib_uverbs_modify_qp,
+	[IB_USER_VERBS_CMD_DESTROY_QP]    	= ib_uverbs_destroy_qp,
+	[IB_USER_VERBS_CMD_POST_SEND]    	= ib_uverbs_post_send,
+	[IB_USER_VERBS_CMD_POST_RECV]    	= ib_uverbs_post_recv,
+	[IB_USER_VERBS_CMD_POST_SRQ_RECV]    	= ib_uverbs_post_srq_recv,
+	[IB_USER_VERBS_CMD_CREATE_AH]    	= ib_uverbs_create_ah,
+	[IB_USER_VERBS_CMD_DESTROY_AH]    	= ib_uverbs_destroy_ah,
+	[IB_USER_VERBS_CMD_ATTACH_MCAST]  	= ib_uverbs_attach_mcast,
+	[IB_USER_VERBS_CMD_DETACH_MCAST]  	= ib_uverbs_detach_mcast,
+	[IB_USER_VERBS_CMD_CREATE_SRQ]    	= ib_uverbs_create_srq,
+	[IB_USER_VERBS_CMD_MODIFY_SRQ]    	= ib_uverbs_modify_srq,
+	[IB_USER_VERBS_CMD_QUERY_SRQ]     	= ib_uverbs_query_srq,
+	[IB_USER_VERBS_CMD_DESTROY_SRQ]   	= ib_uverbs_destroy_srq,
+};
+
+static struct vfsmount *uverbs_event_mnt;
+
+static void ib_uverbs_add_one(struct ib_device *device);
+static void ib_uverbs_remove_one(struct ib_device *device);
+
+static void ib_uverbs_release_dev(struct kref *ref)
+{
+	struct ib_uverbs_device *dev =
+		container_of(ref, struct ib_uverbs_device, ref);
+
+	complete(&dev->comp);
+}
+
+void ib_uverbs_release_ucq(struct ib_uverbs_file *file,
+			  struct ib_uverbs_event_file *ev_file,
+			  struct ib_ucq_object *uobj)
+{
+	struct ib_uverbs_event *evt, *tmp;
+
+	if (ev_file) {
+		spin_lock_irq(&ev_file->lock);
+		list_for_each_entry_safe(evt, tmp, &uobj->comp_list, obj_list) {
+			list_del(&evt->list);
+			kfree(evt);
+		}
+		spin_unlock_irq(&ev_file->lock);
+
+		kref_put(&ev_file->ref, ib_uverbs_release_event_file);
+	}
+
+	spin_lock_irq(&file->async_file->lock);
+	list_for_each_entry_safe(evt, tmp, &uobj->async_list, obj_list) {
+		list_del(&evt->list);
+		kfree(evt);
+	}
+	spin_unlock_irq(&file->async_file->lock);
+}
+
+void ib_uverbs_release_uevent(struct ib_uverbs_file *file,
+			      struct ib_uevent_object *uobj)
+{
+	struct ib_uverbs_event *evt, *tmp;
+
+	spin_lock_irq(&file->async_file->lock);
+	list_for_each_entry_safe(evt, tmp, &uobj->event_list, obj_list) {
+		list_del(&evt->list);
+		kfree(evt);
+	}
+	spin_unlock_irq(&file->async_file->lock);
+}
+
+static void ib_uverbs_detach_umcast(struct ib_qp *qp,
+				    struct ib_uqp_object *uobj)
+{
+	struct ib_uverbs_mcast_entry *mcast, *tmp;
+
+	list_for_each_entry_safe(mcast, tmp, &uobj->mcast_list, list) {
+		ib_detach_mcast(qp, &mcast->gid, mcast->lid);
+		list_del(&mcast->list);
+		kfree(mcast);
+	}
+}
+
+static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file,
+				      struct ib_ucontext *context)
+{
+	struct ib_uobject *uobj, *tmp;
+
+	if (!context)
+		return 0;
+
+	list_for_each_entry_safe(uobj, tmp, &context->ah_list, list) {
+		struct ib_ah *ah = uobj->object;
+
+		idr_remove_uobj(&ib_uverbs_ah_idr, uobj);
+		ib_destroy_ah(ah);
+		kfree(uobj);
+	}
+
+	list_for_each_entry_safe(uobj, tmp, &context->qp_list, list) {
+		struct ib_qp *qp = uobj->object;
+		struct ib_uqp_object *uqp =
+			container_of(uobj, struct ib_uqp_object, uevent.uobject);
+
+		idr_remove_uobj(&ib_uverbs_qp_idr, uobj);
+		ib_uverbs_detach_umcast(qp, uqp);
+		ib_destroy_qp(qp);
+		ib_uverbs_release_uevent(file, &uqp->uevent);
+		kfree(uqp);
+	}
+
+	list_for_each_entry_safe(uobj, tmp, &context->cq_list, list) {
+		struct ib_cq *cq = uobj->object;
+		struct ib_uverbs_event_file *ev_file = cq->cq_context;
+		struct ib_ucq_object *ucq =
+			container_of(uobj, struct ib_ucq_object, uobject);
+
+		idr_remove_uobj(&ib_uverbs_cq_idr, uobj);
+		ib_destroy_cq(cq);
+		ib_uverbs_release_ucq(file, ev_file, ucq);
+		kfree(ucq);
+	}
+
+	list_for_each_entry_safe(uobj, tmp, &context->srq_list, list) {
+		struct ib_srq *srq = uobj->object;
+		struct ib_uevent_object *uevent =
+			container_of(uobj, struct ib_uevent_object, uobject);
+
+		idr_remove_uobj(&ib_uverbs_srq_idr, uobj);
+		ib_destroy_srq(srq);
+		ib_uverbs_release_uevent(file, uevent);
+		kfree(uevent);
+	}
+
+	/* XXX Free MWs */
+
+	list_for_each_entry_safe(uobj, tmp, &context->mr_list, list) {
+		struct ib_mr *mr = uobj->object;
+		struct ib_device *mrdev = mr->device;
+		struct ib_umem_object *memobj;
+
+		idr_remove_uobj(&ib_uverbs_mr_idr, uobj);
+		ib_dereg_mr(mr);
+
+		memobj = container_of(uobj, struct ib_umem_object, uobject);
+		ib_umem_release_on_close(mrdev, &memobj->umem);
+
+		kfree(memobj);
+	}
+
+	list_for_each_entry_safe(uobj, tmp, &context->pd_list, list) {
+		struct ib_pd *pd = uobj->object;
+
+		idr_remove_uobj(&ib_uverbs_pd_idr, uobj);
+		ib_dealloc_pd(pd);
+		kfree(uobj);
+	}
+
+	return context->device->dealloc_ucontext(context);
+}
+
+static void ib_uverbs_release_file(struct kref *ref)
+{
+	struct ib_uverbs_file *file =
+		container_of(ref, struct ib_uverbs_file, ref);
+
+	module_put(file->device->ib_dev->owner);
+	kref_put(&file->device->ref, ib_uverbs_release_dev);
+
+	kfree(file);
+}
+
+static ssize_t ib_uverbs_event_read(struct file *filp, char __user *buf,
+				    size_t count, loff_t *pos)
+{
+	struct ib_uverbs_event_file *file = filp->private_data;
+	struct ib_uverbs_event *event;
+	int eventsz;
+	int ret = 0;
+
+	spin_lock_irq(&file->lock);
+
+	while (list_empty(&file->event_list)) {
+		spin_unlock_irq(&file->lock);
+
+		if (filp->f_flags & O_NONBLOCK)
+			return -EAGAIN;
+
+		if (wait_event_interruptible(file->poll_wait,
+					     !list_empty(&file->event_list)))
+			return -ERESTARTSYS;
+
+		spin_lock_irq(&file->lock);
+	}
+
+	event = list_entry(file->event_list.next, struct ib_uverbs_event, list);
+
+	if (file->is_async)
+		eventsz = sizeof (struct ib_uverbs_async_event_desc);
+	else
+		eventsz = sizeof (struct ib_uverbs_comp_event_desc);
+
+	if (eventsz > count) {
+		ret   = -EINVAL;
+		event = NULL;
+	} else {
+		list_del(file->event_list.next);
+		if (event->counter) {
+			++(*event->counter);
+			list_del(&event->obj_list);
+		}
+	}
+
+	spin_unlock_irq(&file->lock);
+
+	if (event) {
+		if (copy_to_user(buf, event, eventsz))
+			ret = -EFAULT;
+		else
+			ret = eventsz;
+	}
+
+	kfree(event);
+
+	return ret;
+}
+
+static unsigned int ib_uverbs_event_poll(struct file *filp,
+					 struct poll_table_struct *wait)
+{
+	unsigned int pollflags = 0;
+	struct ib_uverbs_event_file *file = filp->private_data;
+
+	poll_wait(filp, &file->poll_wait, wait);
+
+	spin_lock_irq(&file->lock);
+	if (!list_empty(&file->event_list))
+		pollflags = POLLIN | POLLRDNORM;
+	spin_unlock_irq(&file->lock);
+
+	return pollflags;
+}
+
+void ib_uverbs_release_event_file(struct kref *ref)
+{
+	struct ib_uverbs_event_file *file =
+		container_of(ref, struct ib_uverbs_event_file, ref);
+
+	kfree(file);
+}
+
+static int ib_uverbs_event_fasync(int fd, struct file *filp, int on)
+{
+	struct ib_uverbs_event_file *file = filp->private_data;
+
+	return fasync_helper(fd, filp, on, &file->async_queue);
+}
+
+static int ib_uverbs_event_close(struct inode *inode, struct file *filp)
+{
+	struct ib_uverbs_event_file *file = filp->private_data;
+	struct ib_uverbs_event *entry, *tmp;
+
+	spin_lock_irq(&file->lock);
+	file->file = NULL;
+	list_for_each_entry_safe(entry, tmp, &file->event_list, list) {
+		if (entry->counter)
+			list_del(&entry->obj_list);
+		kfree(entry);
+	}
+	spin_unlock_irq(&file->lock);
+
+	ib_uverbs_event_fasync(-1, filp, 0);
+
+	if (file->is_async) {
+		ib_unregister_event_handler(&file->uverbs_file->event_handler);
+		kref_put(&file->uverbs_file->ref, ib_uverbs_release_file);
+	}
+	kref_put(&file->ref, ib_uverbs_release_event_file);
+
+	return 0;
+}
+
+static const struct file_operations uverbs_event_fops = {
+	.owner	 = THIS_MODULE,
+	.read 	 = ib_uverbs_event_read,
+	.poll    = ib_uverbs_event_poll,
+	.release = ib_uverbs_event_close,
+	.fasync  = ib_uverbs_event_fasync
+};
+
+void ib_uverbs_comp_handler(struct ib_cq *cq, void *cq_context)
+{
+	struct ib_uverbs_event_file    *file = cq_context;
+	struct ib_ucq_object	       *uobj;
+	struct ib_uverbs_event	       *entry;
+	unsigned long			flags;
+
+	if (!file)
+		return;
+
+	spin_lock_irqsave(&file->lock, flags);
+	if (!file->file) {
+		spin_unlock_irqrestore(&file->lock, flags);
+		return;
+	}
+
+	entry = kmalloc(sizeof *entry, GFP_ATOMIC);
+	if (!entry) {
+		spin_unlock_irqrestore(&file->lock, flags);
+		return;
+	}
+
+	uobj = container_of(cq->uobject, struct ib_ucq_object, uobject);
+
+	entry->desc.comp.cq_handle = cq->uobject->user_handle;
+	entry->counter		   = &uobj->comp_events_reported;
+
+	list_add_tail(&entry->list, &file->event_list);
+	list_add_tail(&entry->obj_list, &uobj->comp_list);
+	spin_unlock_irqrestore(&file->lock, flags);
+
+	wake_up_interruptible(&file->poll_wait);
+	kill_fasync(&file->async_queue, SIGIO, POLL_IN);
+}
+
+static void ib_uverbs_async_handler(struct ib_uverbs_file *file,
+				    __u64 element, __u64 event,
+				    struct list_head *obj_list,
+				    u32 *counter)
+{
+	struct ib_uverbs_event *entry;
+	unsigned long flags;
+
+	spin_lock_irqsave(&file->async_file->lock, flags);
+	if (!file->async_file->file) {
+		spin_unlock_irqrestore(&file->async_file->lock, flags);
+		return;
+	}
+
+	entry = kmalloc(sizeof *entry, GFP_ATOMIC);
+	if (!entry) {
+		spin_unlock_irqrestore(&file->async_file->lock, flags);
+		return;
+	}
+
+	entry->desc.async.element    = element;
+	entry->desc.async.event_type = event;
+	entry->counter               = counter;
+
+	list_add_tail(&entry->list, &file->async_file->event_list);
+	if (obj_list)
+		list_add_tail(&entry->obj_list, obj_list);
+	spin_unlock_irqrestore(&file->async_file->lock, flags);
+
+	wake_up_interruptible(&file->async_file->poll_wait);
+	kill_fasync(&file->async_file->async_queue, SIGIO, POLL_IN);
+}
+
+void ib_uverbs_cq_event_handler(struct ib_event *event, void *context_ptr)
+{
+	struct ib_ucq_object *uobj = container_of(event->element.cq->uobject,
+						  struct ib_ucq_object, uobject);
+
+	ib_uverbs_async_handler(uobj->uverbs_file, uobj->uobject.user_handle,
+				event->event, &uobj->async_list,
+				&uobj->async_events_reported);
+}
+
+void ib_uverbs_qp_event_handler(struct ib_event *event, void *context_ptr)
+{
+	struct ib_uevent_object *uobj;
+
+	uobj = container_of(event->element.qp->uobject,
+			    struct ib_uevent_object, uobject);
+
+	ib_uverbs_async_handler(context_ptr, uobj->uobject.user_handle,
+				event->event, &uobj->event_list,
+				&uobj->events_reported);
+}
+
+void ib_uverbs_srq_event_handler(struct ib_event *event, void *context_ptr)
+{
+	struct ib_uevent_object *uobj;
+
+	uobj = container_of(event->element.srq->uobject,
+			    struct ib_uevent_object, uobject);
+
+	ib_uverbs_async_handler(context_ptr, uobj->uobject.user_handle,
+				event->event, &uobj->event_list,
+				&uobj->events_reported);
+}
+
+void ib_uverbs_event_handler(struct ib_event_handler *handler,
+			     struct ib_event *event)
+{
+	struct ib_uverbs_file *file =
+		container_of(handler, struct ib_uverbs_file, event_handler);
+
+	ib_uverbs_async_handler(file, event->element.port_num, event->event,
+				NULL, NULL);
+}
+
+struct file *ib_uverbs_alloc_event_file(struct ib_uverbs_file *uverbs_file,
+					int is_async, int *fd)
+{
+	struct ib_uverbs_event_file *ev_file;
+	struct file *filp;
+	int ret;
+
+	ev_file = kmalloc(sizeof *ev_file, GFP_KERNEL);
+	if (!ev_file)
+		return ERR_PTR(-ENOMEM);
+
+	kref_init(&ev_file->ref);
+	spin_lock_init(&ev_file->lock);
+	INIT_LIST_HEAD(&ev_file->event_list);
+	init_waitqueue_head(&ev_file->poll_wait);
+	ev_file->uverbs_file = uverbs_file;
+	ev_file->async_queue = NULL;
+	ev_file->is_async    = is_async;
+
+	*fd = get_unused_fd();
+	if (*fd < 0) {
+		ret = *fd;
+		goto err;
+	}
+
+	filp = get_empty_filp();
+	if (!filp) {
+		ret = -ENFILE;
+		goto err_fd;
+	}
+
+	ev_file->file      = filp;
+
+	/*
+	 * fops_get() can't fail here, because we're coming from a
+	 * system call on a uverbs file, which will already have a
+	 * module reference.
+	 */
+	filp->f_op 	   = fops_get(&uverbs_event_fops);
+	filp->f_path.mnt 	   = mntget(uverbs_event_mnt);
+	filp->f_path.dentry 	   = dget(uverbs_event_mnt->mnt_root);
+	filp->f_mapping    = filp->f_path.dentry->d_inode->i_mapping;
+	filp->f_flags      = O_RDONLY;
+	filp->f_mode       = FMODE_READ;
+	filp->private_data = ev_file;
+
+	return filp;
+
+err_fd:
+	put_unused_fd(*fd);
+
+err:
+	kfree(ev_file);
+	return ERR_PTR(ret);
+}
+
+/*
+ * Look up a completion event file by FD.  If lookup is successful,
+ * takes a ref to the event file struct that it returns; if
+ * unsuccessful, returns NULL.
+ */
+struct ib_uverbs_event_file *ib_uverbs_lookup_comp_file(int fd)
+{
+	struct ib_uverbs_event_file *ev_file = NULL;
+	struct file *filp;
+
+	filp = fget(fd);
+	if (!filp)
+		return NULL;
+
+	if (filp->f_op != &uverbs_event_fops)
+		goto out;
+
+	ev_file = filp->private_data;
+	if (ev_file->is_async) {
+		ev_file = NULL;
+		goto out;
+	}
+
+	kref_get(&ev_file->ref);
+
+out:
+	fput(filp);
+	return ev_file;
+}
+
+static ssize_t ib_uverbs_write(struct file *filp, const char __user *buf,
+			     size_t count, loff_t *pos)
+{
+	struct ib_uverbs_file *file = filp->private_data;
+	struct ib_uverbs_cmd_hdr hdr;
+
+	if (count < sizeof hdr)
+		return -EINVAL;
+
+	if (copy_from_user(&hdr, buf, sizeof hdr))
+		return -EFAULT;
+
+	if (hdr.in_words * 4 != count)
+		return -EINVAL;
+
+	if (hdr.command < 0				||
+	    hdr.command >= ARRAY_SIZE(uverbs_cmd_table) ||
+	    !uverbs_cmd_table[hdr.command]		||
+	    !(file->device->ib_dev->uverbs_cmd_mask & (1ull << hdr.command)))
+		return -EINVAL;
+
+	if (!file->ucontext &&
+	    hdr.command != IB_USER_VERBS_CMD_GET_CONTEXT)
+		return -EINVAL;
+
+	return uverbs_cmd_table[hdr.command](file, buf + sizeof hdr,
+					     hdr.in_words * 4, hdr.out_words * 4);
+}
+
+static int ib_uverbs_mmap(struct file *filp, struct vm_area_struct *vma)
+{
+	struct ib_uverbs_file *file = filp->private_data;
+
+	if (!file->ucontext)
+		return -ENODEV;
+	else
+		return file->device->ib_dev->mmap(file->ucontext, vma);
+}
+
+static int ib_uverbs_open(struct inode *inode, struct file *filp)
+{
+	struct ib_uverbs_device *dev;
+	struct ib_uverbs_file *file;
+	int ret;
+
+	spin_lock(&map_lock);
+	dev = dev_table[iminor(inode) - IB_UVERBS_BASE_MINOR];
+	if (dev)
+		kref_get(&dev->ref);
+	spin_unlock(&map_lock);
+
+	if (!dev)
+		return -ENXIO;
+
+	if (!try_module_get(dev->ib_dev->owner)) {
+		ret = -ENODEV;
+		goto err;
+	}
+
+	file = kmalloc(sizeof *file, GFP_KERNEL);
+	if (!file) {
+		ret = -ENOMEM;
+		goto err_module;
+	}
+
+	file->device	 = dev;
+	file->ucontext	 = NULL;
+	file->async_file = NULL;
+	kref_init(&file->ref);
+	mutex_init(&file->mutex);
+
+	filp->private_data = file;
+
+	return 0;
+
+err_module:
+	module_put(dev->ib_dev->owner);
+
+err:
+	kref_put(&dev->ref, ib_uverbs_release_dev);
+
+	return ret;
+}
+
+static int ib_uverbs_close(struct inode *inode, struct file *filp)
+{
+	struct ib_uverbs_file *file = filp->private_data;
+
+	ib_uverbs_cleanup_ucontext(file, file->ucontext);
+
+	if (file->async_file)
+		kref_put(&file->async_file->ref, ib_uverbs_release_event_file);
+
+	kref_put(&file->ref, ib_uverbs_release_file);
+
+	return 0;
+}
+
+static const struct file_operations uverbs_fops = {
+	.owner 	 = THIS_MODULE,
+	.write 	 = ib_uverbs_write,
+	.open 	 = ib_uverbs_open,
+	.release = ib_uverbs_close
+};
+
+static const struct file_operations uverbs_mmap_fops = {
+	.owner 	 = THIS_MODULE,
+	.write 	 = ib_uverbs_write,
+	.mmap    = ib_uverbs_mmap,
+	.open 	 = ib_uverbs_open,
+	.release = ib_uverbs_close
+};
+
+static struct ib_client uverbs_client = {
+	.name   = "uverbs",
+	.add    = ib_uverbs_add_one,
+	.remove = ib_uverbs_remove_one
+};
+
+static ssize_t show_ibdev(struct class_device *class_dev, char *buf)
+{
+	struct ib_uverbs_device *dev = class_get_devdata(class_dev);
+
+	if (!dev)
+		return -ENODEV;
+
+	return sprintf(buf, "%s\n", dev->ib_dev->name);
+}
+static CLASS_DEVICE_ATTR(ibdev, S_IRUGO, show_ibdev, NULL);
+
+static ssize_t show_dev_abi_version(struct class_device *class_dev, char *buf)
+{
+	struct ib_uverbs_device *dev = class_get_devdata(class_dev);
+
+	if (!dev)
+		return -ENODEV;
+
+	return sprintf(buf, "%d\n", dev->ib_dev->uverbs_abi_ver);
+}
+static CLASS_DEVICE_ATTR(abi_version, S_IRUGO, show_dev_abi_version, NULL);
+
+static ssize_t show_abi_version(struct class *class, char *buf)
+{
+	return sprintf(buf, "%d\n", IB_USER_VERBS_ABI_VERSION);
+}
+static CLASS_ATTR(abi_version, S_IRUGO, show_abi_version, NULL);
+
+static void ib_uverbs_add_one(struct ib_device *device)
+{
+	struct ib_uverbs_device *uverbs_dev;
+
+	if (!device->alloc_ucontext)
+		return;
+
+	uverbs_dev = kzalloc(sizeof *uverbs_dev, GFP_KERNEL);
+	if (!uverbs_dev)
+		return;
+
+	kref_init(&uverbs_dev->ref);
+	init_completion(&uverbs_dev->comp);
+
+	spin_lock(&map_lock);
+	uverbs_dev->devnum = find_first_zero_bit(dev_map, IB_UVERBS_MAX_DEVICES);
+	if (uverbs_dev->devnum >= IB_UVERBS_MAX_DEVICES) {
+		spin_unlock(&map_lock);
+		goto err;
+	}
+	set_bit(uverbs_dev->devnum, dev_map);
+	spin_unlock(&map_lock);
+
+	uverbs_dev->ib_dev           = device;
+	uverbs_dev->num_comp_vectors = 1;
+
+	uverbs_dev->dev = cdev_alloc();
+	if (!uverbs_dev->dev)
+		goto err;
+	uverbs_dev->dev->owner = THIS_MODULE;
+	uverbs_dev->dev->ops = device->mmap ? &uverbs_mmap_fops : &uverbs_fops;
+	kobject_set_name(&uverbs_dev->dev->kobj, "uverbs%d", uverbs_dev->devnum);
+	if (cdev_add(uverbs_dev->dev, IB_UVERBS_BASE_DEV + uverbs_dev->devnum, 1))
+		goto err_cdev;
+
+	uverbs_dev->class_dev = class_device_create(uverbs_class, NULL,
+						    uverbs_dev->dev->dev,
+						    device->dma_device,
+						    "uverbs%d", uverbs_dev->devnum);
+	if (IS_ERR(uverbs_dev->class_dev))
+		goto err_cdev;
+
+	class_set_devdata(uverbs_dev->class_dev, uverbs_dev);
+
+	if (class_device_create_file(uverbs_dev->class_dev, &class_device_attr_ibdev))
+		goto err_class;
+	if (class_device_create_file(uverbs_dev->class_dev, &class_device_attr_abi_version))
+		goto err_class;
+
+	spin_lock(&map_lock);
+	dev_table[uverbs_dev->devnum] = uverbs_dev;
+	spin_unlock(&map_lock);
+
+	ib_set_client_data(device, &uverbs_client, uverbs_dev);
+
+	return;
+
+err_class:
+	class_device_destroy(uverbs_class, uverbs_dev->dev->dev);
+
+err_cdev:
+	cdev_del(uverbs_dev->dev);
+	clear_bit(uverbs_dev->devnum, dev_map);
+
+err:
+	kref_put(&uverbs_dev->ref, ib_uverbs_release_dev);
+	wait_for_completion(&uverbs_dev->comp);
+	kfree(uverbs_dev);
+	return;
+}
+
+static void ib_uverbs_remove_one(struct ib_device *device)
+{
+	struct ib_uverbs_device *uverbs_dev = ib_get_client_data(device, &uverbs_client);
+
+	if (!uverbs_dev)
+		return;
+
+	class_set_devdata(uverbs_dev->class_dev, NULL);
+	class_device_destroy(uverbs_class, uverbs_dev->dev->dev);
+	cdev_del(uverbs_dev->dev);
+
+	spin_lock(&map_lock);
+	dev_table[uverbs_dev->devnum] = NULL;
+	spin_unlock(&map_lock);
+
+	clear_bit(uverbs_dev->devnum, dev_map);
+
+	kref_put(&uverbs_dev->ref, ib_uverbs_release_dev);
+	wait_for_completion(&uverbs_dev->comp);
+	kfree(uverbs_dev);
+}
+
+static int uverbs_event_get_sb(struct file_system_type *fs_type, int flags,
+			       const char *dev_name, void *data,
+			       struct vfsmount *mnt)
+{
+	return get_sb_pseudo(fs_type, "infinibandevent:", NULL,
+			     INFINIBANDEVENTFS_MAGIC, mnt);
+}
+
+static struct file_system_type uverbs_event_fs = {
+	/* No owner field so module can be unloaded */
+	.name    = "infinibandeventfs",
+	.get_sb  = uverbs_event_get_sb,
+	.kill_sb = kill_litter_super
+};
+
+static int __init ib_uverbs_init(void)
+{
+	int ret;
+
+	spin_lock_init(&map_lock);
+
+	ret = register_chrdev_region(IB_UVERBS_BASE_DEV, IB_UVERBS_MAX_DEVICES,
+				     "infiniband_verbs");
+	if (ret) {
+		printk(KERN_ERR "user_verbs: couldn't register device number\n");
+		goto out;
+	}
+
+	uverbs_class = class_create(THIS_MODULE, "infiniband_verbs");
+	if (IS_ERR(uverbs_class)) {
+		ret = PTR_ERR(uverbs_class);
+		printk(KERN_ERR "user_verbs: couldn't create class infiniband_verbs\n");
+		goto out_chrdev;
+	}
+
+	ret = class_create_file(uverbs_class, &class_attr_abi_version);
+	if (ret) {
+		printk(KERN_ERR "user_verbs: couldn't create abi_version attribute\n");
+		goto out_class;
+	}
+
+	ret = register_filesystem(&uverbs_event_fs);
+	if (ret) {
+		printk(KERN_ERR "user_verbs: couldn't register infinibandeventfs\n");
+		goto out_class;
+	}
+
+	uverbs_event_mnt = kern_mount(&uverbs_event_fs);
+	if (IS_ERR(uverbs_event_mnt)) {
+		ret = PTR_ERR(uverbs_event_mnt);
+		printk(KERN_ERR "user_verbs: couldn't mount infinibandeventfs\n");
+		goto out_fs;
+	}
+
+	ret = ib_register_client(&uverbs_client);
+	if (ret) {
+		printk(KERN_ERR "user_verbs: couldn't register client\n");
+		goto out_mnt;
+	}
+
+	return 0;
+
+out_mnt:
+	mntput(uverbs_event_mnt);
+
+out_fs:
+	unregister_filesystem(&uverbs_event_fs);
+
+out_class:
+	class_destroy(uverbs_class);
+
+out_chrdev:
+	unregister_chrdev_region(IB_UVERBS_BASE_DEV, IB_UVERBS_MAX_DEVICES);
+
+out:
+	return ret;
+}
+
+static void __exit ib_uverbs_cleanup(void)
+{
+	ib_unregister_client(&uverbs_client);
+	mntput(uverbs_event_mnt);
+	unregister_filesystem(&uverbs_event_fs);
+	class_destroy(uverbs_class);
+	unregister_chrdev_region(IB_UVERBS_BASE_DEV, IB_UVERBS_MAX_DEVICES);
+	flush_scheduled_work();
+	idr_destroy(&ib_uverbs_pd_idr);
+	idr_destroy(&ib_uverbs_mr_idr);
+	idr_destroy(&ib_uverbs_mw_idr);
+	idr_destroy(&ib_uverbs_ah_idr);
+	idr_destroy(&ib_uverbs_cq_idr);
+	idr_destroy(&ib_uverbs_qp_idr);
+	idr_destroy(&ib_uverbs_srq_idr);
+}
+
+module_init(ib_uverbs_init);
+module_exit(ib_uverbs_cleanup);

drivers/infiniband/core/uverbs_marshall.c

@@ -0,0 +1,139 @@
+/*
+ * Copyright (c) 2005 Intel Corporation.  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 <rdma/ib_marshall.h>
+
+void ib_copy_ah_attr_to_user(struct ib_uverbs_ah_attr *dst,
+			     struct ib_ah_attr *src)
+{
+	memcpy(dst->grh.dgid, src->grh.dgid.raw, sizeof src->grh.dgid);
+	dst->grh.flow_label        = src->grh.flow_label;
+	dst->grh.sgid_index        = src->grh.sgid_index;
+	dst->grh.hop_limit         = src->grh.hop_limit;
+	dst->grh.traffic_class     = src->grh.traffic_class;
+	dst->dlid 	    	   = src->dlid;
+	dst->sl   	    	   = src->sl;
+	dst->src_path_bits 	   = src->src_path_bits;
+	dst->static_rate   	   = src->static_rate;
+	dst->is_global             = src->ah_flags & IB_AH_GRH ? 1 : 0;
+	dst->port_num 	    	   = src->port_num;
+}
+EXPORT_SYMBOL(ib_copy_ah_attr_to_user);
+
+void ib_copy_qp_attr_to_user(struct ib_uverbs_qp_attr *dst,
+			     struct ib_qp_attr *src)
+{
+	dst->cur_qp_state	= src->cur_qp_state;
+	dst->path_mtu		= src->path_mtu;
+	dst->path_mig_state	= src->path_mig_state;
+	dst->qkey		= src->qkey;
+	dst->rq_psn		= src->rq_psn;
+	dst->sq_psn		= src->sq_psn;
+	dst->dest_qp_num	= src->dest_qp_num;
+	dst->qp_access_flags	= src->qp_access_flags;
+
+	dst->max_send_wr	= src->cap.max_send_wr;
+	dst->max_recv_wr	= src->cap.max_recv_wr;
+	dst->max_send_sge	= src->cap.max_send_sge;
+	dst->max_recv_sge	= src->cap.max_recv_sge;
+	dst->max_inline_data	= src->cap.max_inline_data;
+
+	ib_copy_ah_attr_to_user(&dst->ah_attr, &src->ah_attr);
+	ib_copy_ah_attr_to_user(&dst->alt_ah_attr, &src->alt_ah_attr);
+
+	dst->pkey_index		= src->pkey_index;
+	dst->alt_pkey_index	= src->alt_pkey_index;
+	dst->en_sqd_async_notify = src->en_sqd_async_notify;
+	dst->sq_draining	= src->sq_draining;
+	dst->max_rd_atomic	= src->max_rd_atomic;
+	dst->max_dest_rd_atomic	= src->max_dest_rd_atomic;
+	dst->min_rnr_timer	= src->min_rnr_timer;
+	dst->port_num		= src->port_num;
+	dst->timeout		= src->timeout;
+	dst->retry_cnt		= src->retry_cnt;
+	dst->rnr_retry		= src->rnr_retry;
+	dst->alt_port_num	= src->alt_port_num;
+	dst->alt_timeout	= src->alt_timeout;
+}
+EXPORT_SYMBOL(ib_copy_qp_attr_to_user);
+
+void ib_copy_path_rec_to_user(struct ib_user_path_rec *dst,
+			      struct ib_sa_path_rec *src)
+{
+	memcpy(dst->dgid, src->dgid.raw, sizeof src->dgid);
+	memcpy(dst->sgid, src->sgid.raw, sizeof src->sgid);
+
+	dst->dlid		= src->dlid;
+	dst->slid		= src->slid;
+	dst->raw_traffic	= src->raw_traffic;
+	dst->flow_label		= src->flow_label;
+	dst->hop_limit		= src->hop_limit;
+	dst->traffic_class	= src->traffic_class;
+	dst->reversible		= src->reversible;
+	dst->numb_path		= src->numb_path;
+	dst->pkey		= src->pkey;
+	dst->sl			= src->sl;
+	dst->mtu_selector	= src->mtu_selector;
+	dst->mtu		= src->mtu;
+	dst->rate_selector	= src->rate_selector;
+	dst->rate		= src->rate;
+	dst->packet_life_time	= src->packet_life_time;
+	dst->preference		= src->preference;
+	dst->packet_life_time_selector = src->packet_life_time_selector;
+}
+EXPORT_SYMBOL(ib_copy_path_rec_to_user);
+
+void ib_copy_path_rec_from_user(struct ib_sa_path_rec *dst,
+				struct ib_user_path_rec *src)
+{
+	memcpy(dst->dgid.raw, src->dgid, sizeof dst->dgid);
+	memcpy(dst->sgid.raw, src->sgid, sizeof dst->sgid);
+
+	dst->dlid		= src->dlid;
+	dst->slid		= src->slid;
+	dst->raw_traffic	= src->raw_traffic;
+	dst->flow_label		= src->flow_label;
+	dst->hop_limit		= src->hop_limit;
+	dst->traffic_class	= src->traffic_class;
+	dst->reversible		= src->reversible;
+	dst->numb_path		= src->numb_path;
+	dst->pkey		= src->pkey;
+	dst->sl			= src->sl;
+	dst->mtu_selector	= src->mtu_selector;
+	dst->mtu		= src->mtu;
+	dst->rate_selector	= src->rate_selector;
+	dst->rate		= src->rate;
+	dst->packet_life_time	= src->packet_life_time;
+	dst->preference		= src->preference;
+	dst->packet_life_time_selector = src->packet_life_time_selector;
+}
+EXPORT_SYMBOL(ib_copy_path_rec_from_user);

drivers/infiniband/core/uverbs_mem.c

@@ -0,0 +1,225 @@
+/*
+ * Copyright (c) 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Cisco Systems.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. 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.
+ *
+ * $Id: uverbs_mem.c,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#include <linux/mm.h>
+#include <linux/dma-mapping.h>
+
+#include "uverbs.h"
+
+struct ib_umem_account_work {
+	struct work_struct work;
+	struct mm_struct  *mm;
+	unsigned long      diff;
+};
+
+
+static void __ib_umem_release(struct ib_device *dev, struct ib_umem *umem, int dirty)
+{
+	struct ib_umem_chunk *chunk, *tmp;
+	int i;
+
+	list_for_each_entry_safe(chunk, tmp, &umem->chunk_list, list) {
+		ib_dma_unmap_sg(dev, chunk->page_list,
+				chunk->nents, DMA_BIDIRECTIONAL);
+		for (i = 0; i < chunk->nents; ++i) {
+			if (umem->writable && dirty)
+				set_page_dirty_lock(chunk->page_list[i].page);
+			put_page(chunk->page_list[i].page);
+		}
+
+		kfree(chunk);
+	}
+}
+
+int ib_umem_get(struct ib_device *dev, struct ib_umem *mem,
+		void *addr, size_t size, int write)
+{
+	struct page **page_list;
+	struct ib_umem_chunk *chunk;
+	unsigned long locked;
+	unsigned long lock_limit;
+	unsigned long cur_base;
+	unsigned long npages;
+	int ret = 0;
+	int off;
+	int i;
+
+	if (!can_do_mlock())
+		return -EPERM;
+
+	page_list = (struct page **) __get_free_page(GFP_KERNEL);
+	if (!page_list)
+		return -ENOMEM;
+
+	mem->user_base = (unsigned long) addr;
+	mem->length    = size;
+	mem->offset    = (unsigned long) addr & ~PAGE_MASK;
+	mem->page_size = PAGE_SIZE;
+	mem->writable  = write;
+
+	INIT_LIST_HEAD(&mem->chunk_list);
+
+	npages = PAGE_ALIGN(size + mem->offset) >> PAGE_SHIFT;
+
+	down_write(&current->mm->mmap_sem);
+
+	locked     = npages + current->mm->locked_vm;
+	lock_limit = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur >> PAGE_SHIFT;
+
+	if ((locked > lock_limit) && !capable(CAP_IPC_LOCK)) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	cur_base = (unsigned long) addr & PAGE_MASK;
+
+	while (npages) {
+		ret = get_user_pages(current, current->mm, cur_base,
+				     min_t(int, npages,
+					   PAGE_SIZE / sizeof (struct page *)),
+				     1, !write, page_list, NULL);
+
+		if (ret < 0)
+			goto out;
+
+		cur_base += ret * PAGE_SIZE;
+		npages   -= ret;
+
+		off = 0;
+
+		while (ret) {
+			chunk = kmalloc(sizeof *chunk + sizeof (struct scatterlist) *
+					min_t(int, ret, IB_UMEM_MAX_PAGE_CHUNK),
+					GFP_KERNEL);
+			if (!chunk) {
+				ret = -ENOMEM;
+				goto out;
+			}
+
+			chunk->nents = min_t(int, ret, IB_UMEM_MAX_PAGE_CHUNK);
+			for (i = 0; i < chunk->nents; ++i) {
+				chunk->page_list[i].page   = page_list[i + off];
+				chunk->page_list[i].offset = 0;
+				chunk->page_list[i].length = PAGE_SIZE;
+			}
+
+			chunk->nmap = ib_dma_map_sg(dev,
+						    &chunk->page_list[0],
+						    chunk->nents,
+						    DMA_BIDIRECTIONAL);
+			if (chunk->nmap <= 0) {
+				for (i = 0; i < chunk->nents; ++i)
+					put_page(chunk->page_list[i].page);
+				kfree(chunk);
+
+				ret = -ENOMEM;
+				goto out;
+			}
+
+			ret -= chunk->nents;
+			off += chunk->nents;
+			list_add_tail(&chunk->list, &mem->chunk_list);
+		}
+
+		ret = 0;
+	}
+
+out:
+	if (ret < 0)
+		__ib_umem_release(dev, mem, 0);
+	else
+		current->mm->locked_vm = locked;
+
+	up_write(&current->mm->mmap_sem);
+	free_page((unsigned long) page_list);
+
+	return ret;
+}
+
+void ib_umem_release(struct ib_device *dev, struct ib_umem *umem)
+{
+	__ib_umem_release(dev, umem, 1);
+
+	down_write(&current->mm->mmap_sem);
+	current->mm->locked_vm -=
+		PAGE_ALIGN(umem->length + umem->offset) >> PAGE_SHIFT;
+	up_write(&current->mm->mmap_sem);
+}
+
+static void ib_umem_account(struct work_struct *_work)
+{
+	struct ib_umem_account_work *work =
+		container_of(_work, struct ib_umem_account_work, work);
+
+	down_write(&work->mm->mmap_sem);
+	work->mm->locked_vm -= work->diff;
+	up_write(&work->mm->mmap_sem);
+	mmput(work->mm);
+	kfree(work);
+}
+
+void ib_umem_release_on_close(struct ib_device *dev, struct ib_umem *umem)
+{
+	struct ib_umem_account_work *work;
+	struct mm_struct *mm;
+
+	__ib_umem_release(dev, umem, 1);
+
+	mm = get_task_mm(current);
+	if (!mm)
+		return;
+
+	/*
+	 * We may be called with the mm's mmap_sem already held.  This
+	 * can happen when a userspace munmap() is the call that drops
+	 * the last reference to our file and calls our release
+	 * method.  If there are memory regions to destroy, we'll end
+	 * up here and not be able to take the mmap_sem.  Therefore we
+	 * defer the vm_locked accounting to the system workqueue.
+	 */
+
+	work = kmalloc(sizeof *work, GFP_KERNEL);
+	if (!work) {
+		mmput(mm);
+		return;
+	}
+
+	INIT_WORK(&work->work, ib_umem_account);
+	work->mm   = mm;
+	work->diff = PAGE_ALIGN(umem->length + umem->offset) >> PAGE_SHIFT;
+
+	schedule_work(&work->work);
+}

drivers/infiniband/core/verbs.c

@@ -0,0 +1,851 @@
+/*
+ * Copyright (c) 2004 Mellanox Technologies Ltd.  All rights reserved.
+ * Copyright (c) 2004 Infinicon Corporation.  All rights reserved.
+ * Copyright (c) 2004 Intel Corporation.  All rights reserved.
+ * Copyright (c) 2004 Topspin Corporation.  All rights reserved.
+ * Copyright (c) 2004 Voltaire Corporation.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
+ * Copyright (c) 2005, 2006 Cisco Systems.  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.
+ *
+ * $Id: verbs.c,v 1.1.1.1 2007/06/12 07:27:10 eyryu Exp $
+ */
+
+#include <linux/errno.h>
+#include <linux/err.h>
+#include <linux/string.h>
+
+#include <rdma/ib_verbs.h>
+#include <rdma/ib_cache.h>
+
+int ib_rate_to_mult(enum ib_rate rate)
+{
+	switch (rate) {
+	case IB_RATE_2_5_GBPS: return  1;
+	case IB_RATE_5_GBPS:   return  2;
+	case IB_RATE_10_GBPS:  return  4;
+	case IB_RATE_20_GBPS:  return  8;
+	case IB_RATE_30_GBPS:  return 12;
+	case IB_RATE_40_GBPS:  return 16;
+	case IB_RATE_60_GBPS:  return 24;
+	case IB_RATE_80_GBPS:  return 32;
+	case IB_RATE_120_GBPS: return 48;
+	default:	       return -1;
+	}
+}
+EXPORT_SYMBOL(ib_rate_to_mult);
+
+enum ib_rate mult_to_ib_rate(int mult)
+{
+	switch (mult) {
+	case 1:  return IB_RATE_2_5_GBPS;
+	case 2:  return IB_RATE_5_GBPS;
+	case 4:  return IB_RATE_10_GBPS;
+	case 8:  return IB_RATE_20_GBPS;
+	case 12: return IB_RATE_30_GBPS;
+	case 16: return IB_RATE_40_GBPS;
+	case 24: return IB_RATE_60_GBPS;
+	case 32: return IB_RATE_80_GBPS;
+	case 48: return IB_RATE_120_GBPS;
+	default: return IB_RATE_PORT_CURRENT;
+	}
+}
+EXPORT_SYMBOL(mult_to_ib_rate);
+
+enum rdma_transport_type
+rdma_node_get_transport(enum rdma_node_type node_type)
+{
+	switch (node_type) {
+	case RDMA_NODE_IB_CA:
+	case RDMA_NODE_IB_SWITCH:
+	case RDMA_NODE_IB_ROUTER:
+		return RDMA_TRANSPORT_IB;
+	case RDMA_NODE_RNIC:
+		return RDMA_TRANSPORT_IWARP;
+	default:
+		BUG();
+		return 0;
+	}
+}
+EXPORT_SYMBOL(rdma_node_get_transport);
+
+/* Protection domains */
+
+struct ib_pd *ib_alloc_pd(struct ib_device *device)
+{
+	struct ib_pd *pd;
+
+	pd = device->alloc_pd(device, NULL, NULL);
+
+	if (!IS_ERR(pd)) {
+		pd->device  = device;
+		pd->uobject = NULL;
+		atomic_set(&pd->usecnt, 0);
+	}
+
+	return pd;
+}
+EXPORT_SYMBOL(ib_alloc_pd);
+
+int ib_dealloc_pd(struct ib_pd *pd)
+{
+	if (atomic_read(&pd->usecnt))
+		return -EBUSY;
+
+	return pd->device->dealloc_pd(pd);
+}
+EXPORT_SYMBOL(ib_dealloc_pd);
+
+/* Address handles */
+
+struct ib_ah *ib_create_ah(struct ib_pd *pd, struct ib_ah_attr *ah_attr)
+{
+	struct ib_ah *ah;
+
+	ah = pd->device->create_ah(pd, ah_attr);
+
+	if (!IS_ERR(ah)) {
+		ah->device  = pd->device;
+		ah->pd      = pd;
+		ah->uobject = NULL;
+		atomic_inc(&pd->usecnt);
+	}
+
+	return ah;
+}
+EXPORT_SYMBOL(ib_create_ah);
+
+int ib_init_ah_from_wc(struct ib_device *device, u8 port_num, struct ib_wc *wc,
+		       struct ib_grh *grh, struct ib_ah_attr *ah_attr)
+{
+	u32 flow_class;
+	u16 gid_index;
+	int ret;
+
+	memset(ah_attr, 0, sizeof *ah_attr);
+	ah_attr->dlid = wc->slid;
+	ah_attr->sl = wc->sl;
+	ah_attr->src_path_bits = wc->dlid_path_bits;
+	ah_attr->port_num = port_num;
+
+	if (wc->wc_flags & IB_WC_GRH) {
+		ah_attr->ah_flags = IB_AH_GRH;
+		ah_attr->grh.dgid = grh->sgid;
+
+		ret = ib_find_cached_gid(device, &grh->dgid, &port_num,
+					 &gid_index);
+		if (ret)
+			return ret;
+
+		ah_attr->grh.sgid_index = (u8) gid_index;
+		flow_class = be32_to_cpu(grh->version_tclass_flow);
+		ah_attr->grh.flow_label = flow_class & 0xFFFFF;
+		ah_attr->grh.hop_limit = 0xFF;
+		ah_attr->grh.traffic_class = (flow_class >> 20) & 0xFF;
+	}
+	return 0;
+}
+EXPORT_SYMBOL(ib_init_ah_from_wc);
+
+struct ib_ah *ib_create_ah_from_wc(struct ib_pd *pd, struct ib_wc *wc,
+				   struct ib_grh *grh, u8 port_num)
+{
+	struct ib_ah_attr ah_attr;
+	int ret;
+
+	ret = ib_init_ah_from_wc(pd->device, port_num, wc, grh, &ah_attr);
+	if (ret)
+		return ERR_PTR(ret);
+
+	return ib_create_ah(pd, &ah_attr);
+}
+EXPORT_SYMBOL(ib_create_ah_from_wc);
+
+int ib_modify_ah(struct ib_ah *ah, struct ib_ah_attr *ah_attr)
+{
+	return ah->device->modify_ah ?
+		ah->device->modify_ah(ah, ah_attr) :
+		-ENOSYS;
+}
+EXPORT_SYMBOL(ib_modify_ah);
+
+int ib_query_ah(struct ib_ah *ah, struct ib_ah_attr *ah_attr)
+{
+	return ah->device->query_ah ?
+		ah->device->query_ah(ah, ah_attr) :
+		-ENOSYS;
+}
+EXPORT_SYMBOL(ib_query_ah);
+
+int ib_destroy_ah(struct ib_ah *ah)
+{
+	struct ib_pd *pd;
+	int ret;
+
+	pd = ah->pd;
+	ret = ah->device->destroy_ah(ah);
+	if (!ret)
+		atomic_dec(&pd->usecnt);
+
+	return ret;
+}
+EXPORT_SYMBOL(ib_destroy_ah);
+
+/* Shared receive queues */
+
+struct ib_srq *ib_create_srq(struct ib_pd *pd,
+			     struct ib_srq_init_attr *srq_init_attr)
+{
+	struct ib_srq *srq;
+
+	if (!pd->device->create_srq)
+		return ERR_PTR(-ENOSYS);
+
+	srq = pd->device->create_srq(pd, srq_init_attr, NULL);
+
+	if (!IS_ERR(srq)) {
+		srq->device    	   = pd->device;
+		srq->pd        	   = pd;
+		srq->uobject       = NULL;
+		srq->event_handler = srq_init_attr->event_handler;
+		srq->srq_context   = srq_init_attr->srq_context;
+		atomic_inc(&pd->usecnt);
+		atomic_set(&srq->usecnt, 0);
+	}
+
+	return srq;
+}
+EXPORT_SYMBOL(ib_create_srq);
+
+int ib_modify_srq(struct ib_srq *srq,
+		  struct ib_srq_attr *srq_attr,
+		  enum ib_srq_attr_mask srq_attr_mask)
+{
+	return srq->device->modify_srq(srq, srq_attr, srq_attr_mask, NULL);
+}
+EXPORT_SYMBOL(ib_modify_srq);
+
+int ib_query_srq(struct ib_srq *srq,
+		 struct ib_srq_attr *srq_attr)
+{
+	return srq->device->query_srq ?
+		srq->device->query_srq(srq, srq_attr) : -ENOSYS;
+}
+EXPORT_SYMBOL(ib_query_srq);
+
+int ib_destroy_srq(struct ib_srq *srq)
+{
+	struct ib_pd *pd;
+	int ret;
+
+	if (atomic_read(&srq->usecnt))
+		return -EBUSY;
+
+	pd = srq->pd;
+
+	ret = srq->device->destroy_srq(srq);
+	if (!ret)
+		atomic_dec(&pd->usecnt);
+
+	return ret;
+}
+EXPORT_SYMBOL(ib_destroy_srq);
+
+/* Queue pairs */
+
+struct ib_qp *ib_create_qp(struct ib_pd *pd,
+			   struct ib_qp_init_attr *qp_init_attr)
+{
+	struct ib_qp *qp;
+
+	qp = pd->device->create_qp(pd, qp_init_attr, NULL);
+
+	if (!IS_ERR(qp)) {
+		qp->device     	  = pd->device;
+		qp->pd         	  = pd;
+		qp->send_cq    	  = qp_init_attr->send_cq;
+		qp->recv_cq    	  = qp_init_attr->recv_cq;
+		qp->srq	       	  = qp_init_attr->srq;
+		qp->uobject       = NULL;
+		qp->event_handler = qp_init_attr->event_handler;
+		qp->qp_context    = qp_init_attr->qp_context;
+		qp->qp_type	  = qp_init_attr->qp_type;
+		atomic_inc(&pd->usecnt);
+		atomic_inc(&qp_init_attr->send_cq->usecnt);
+		atomic_inc(&qp_init_attr->recv_cq->usecnt);
+		if (qp_init_attr->srq)
+			atomic_inc(&qp_init_attr->srq->usecnt);
+	}
+
+	return qp;
+}
+EXPORT_SYMBOL(ib_create_qp);
+
+static const struct {
+	int			valid;
+	enum ib_qp_attr_mask	req_param[IB_QPT_RAW_ETY + 1];
+	enum ib_qp_attr_mask	opt_param[IB_QPT_RAW_ETY + 1];
+} qp_state_table[IB_QPS_ERR + 1][IB_QPS_ERR + 1] = {
+	[IB_QPS_RESET] = {
+		[IB_QPS_RESET] = { .valid = 1 },
+		[IB_QPS_ERR]   = { .valid = 1 },
+		[IB_QPS_INIT]  = {
+			.valid = 1,
+			.req_param = {
+				[IB_QPT_UD]  = (IB_QP_PKEY_INDEX		|
+						IB_QP_PORT			|
+						IB_QP_QKEY),
+				[IB_QPT_UC]  = (IB_QP_PKEY_INDEX		|
+						IB_QP_PORT			|
+						IB_QP_ACCESS_FLAGS),
+				[IB_QPT_RC]  = (IB_QP_PKEY_INDEX		|
+						IB_QP_PORT			|
+						IB_QP_ACCESS_FLAGS),
+				[IB_QPT_SMI] = (IB_QP_PKEY_INDEX		|
+						IB_QP_QKEY),
+				[IB_QPT_GSI] = (IB_QP_PKEY_INDEX		|
+						IB_QP_QKEY),
+			}
+		},
+	},
+	[IB_QPS_INIT]  = {
+		[IB_QPS_RESET] = { .valid = 1 },
+		[IB_QPS_ERR] =   { .valid = 1 },
+		[IB_QPS_INIT]  = {
+			.valid = 1,
+			.opt_param = {
+				[IB_QPT_UD]  = (IB_QP_PKEY_INDEX		|
+						IB_QP_PORT			|
+						IB_QP_QKEY),
+				[IB_QPT_UC]  = (IB_QP_PKEY_INDEX		|
+						IB_QP_PORT			|
+						IB_QP_ACCESS_FLAGS),
+				[IB_QPT_RC]  = (IB_QP_PKEY_INDEX		|
+						IB_QP_PORT			|
+						IB_QP_ACCESS_FLAGS),
+				[IB_QPT_SMI] = (IB_QP_PKEY_INDEX		|
+						IB_QP_QKEY),
+				[IB_QPT_GSI] = (IB_QP_PKEY_INDEX		|
+						IB_QP_QKEY),
+			}
+		},
+		[IB_QPS_RTR]   = {
+			.valid = 1,
+			.req_param = {
+				[IB_QPT_UC]  = (IB_QP_AV			|
+						IB_QP_PATH_MTU			|
+						IB_QP_DEST_QPN			|
+						IB_QP_RQ_PSN),
+				[IB_QPT_RC]  = (IB_QP_AV			|
+						IB_QP_PATH_MTU			|
+						IB_QP_DEST_QPN			|
+						IB_QP_RQ_PSN			|
+						IB_QP_MAX_DEST_RD_ATOMIC	|
+						IB_QP_MIN_RNR_TIMER),
+			},
+			.opt_param = {
+				 [IB_QPT_UD]  = (IB_QP_PKEY_INDEX		|
+						 IB_QP_QKEY),
+				 [IB_QPT_UC]  = (IB_QP_ALT_PATH			|
+						 IB_QP_ACCESS_FLAGS		|
+						 IB_QP_PKEY_INDEX),
+				 [IB_QPT_RC]  = (IB_QP_ALT_PATH			|
+						 IB_QP_ACCESS_FLAGS		|
+						 IB_QP_PKEY_INDEX),
+				 [IB_QPT_SMI] = (IB_QP_PKEY_INDEX		|
+						 IB_QP_QKEY),
+				 [IB_QPT_GSI] = (IB_QP_PKEY_INDEX		|
+						 IB_QP_QKEY),
+			 }
+		}
+	},
+	[IB_QPS_RTR]   = {
+		[IB_QPS_RESET] = { .valid = 1 },
+		[IB_QPS_ERR] =   { .valid = 1 },
+		[IB_QPS_RTS]   = {
+			.valid = 1,
+			.req_param = {
+				[IB_QPT_UD]  = IB_QP_SQ_PSN,
+				[IB_QPT_UC]  = IB_QP_SQ_PSN,
+				[IB_QPT_RC]  = (IB_QP_TIMEOUT			|
+						IB_QP_RETRY_CNT			|
+						IB_QP_RNR_RETRY			|
+						IB_QP_SQ_PSN			|
+						IB_QP_MAX_QP_RD_ATOMIC),
+				[IB_QPT_SMI] = IB_QP_SQ_PSN,
+				[IB_QPT_GSI] = IB_QP_SQ_PSN,
+			},
+			.opt_param = {
+				 [IB_QPT_UD]  = (IB_QP_CUR_STATE		|
+						 IB_QP_QKEY),
+				 [IB_QPT_UC]  = (IB_QP_CUR_STATE		|
+						 IB_QP_ALT_PATH			|
+						 IB_QP_ACCESS_FLAGS		|
+						 IB_QP_PATH_MIG_STATE),
+				 [IB_QPT_RC]  = (IB_QP_CUR_STATE		|
+						 IB_QP_ALT_PATH			|
+						 IB_QP_ACCESS_FLAGS		|
+						 IB_QP_MIN_RNR_TIMER		|
+						 IB_QP_PATH_MIG_STATE),
+				 [IB_QPT_SMI] = (IB_QP_CUR_STATE		|
+						 IB_QP_QKEY),
+				 [IB_QPT_GSI] = (IB_QP_CUR_STATE		|
+						 IB_QP_QKEY),
+			 }
+		}
+	},
+	[IB_QPS_RTS]   = {
+		[IB_QPS_RESET] = { .valid = 1 },
+		[IB_QPS_ERR] =   { .valid = 1 },
+		[IB_QPS_RTS]   = {
+			.valid = 1,
+			.opt_param = {
+				[IB_QPT_UD]  = (IB_QP_CUR_STATE			|
+						IB_QP_QKEY),
+				[IB_QPT_UC]  = (IB_QP_CUR_STATE			|
+						IB_QP_ACCESS_FLAGS		|
+						IB_QP_ALT_PATH			|
+						IB_QP_PATH_MIG_STATE),
+				[IB_QPT_RC]  = (IB_QP_CUR_STATE			|
+						IB_QP_ACCESS_FLAGS		|
+						IB_QP_ALT_PATH			|
+						IB_QP_PATH_MIG_STATE		|
+						IB_QP_MIN_RNR_TIMER),
+				[IB_QPT_SMI] = (IB_QP_CUR_STATE			|
+						IB_QP_QKEY),
+				[IB_QPT_GSI] = (IB_QP_CUR_STATE			|
+						IB_QP_QKEY),
+			}
+		},
+		[IB_QPS_SQD]   = {
+			.valid = 1,
+			.opt_param = {
+				[IB_QPT_UD]  = IB_QP_EN_SQD_ASYNC_NOTIFY,
+				[IB_QPT_UC]  = IB_QP_EN_SQD_ASYNC_NOTIFY,
+				[IB_QPT_RC]  = IB_QP_EN_SQD_ASYNC_NOTIFY,
+				[IB_QPT_SMI] = IB_QP_EN_SQD_ASYNC_NOTIFY,
+				[IB_QPT_GSI] = IB_QP_EN_SQD_ASYNC_NOTIFY
+			}
+		},
+	},
+	[IB_QPS_SQD]   = {
+		[IB_QPS_RESET] = { .valid = 1 },
+		[IB_QPS_ERR] =   { .valid = 1 },
+		[IB_QPS_RTS]   = {
+			.valid = 1,
+			.opt_param = {
+				[IB_QPT_UD]  = (IB_QP_CUR_STATE			|
+						IB_QP_QKEY),
+				[IB_QPT_UC]  = (IB_QP_CUR_STATE			|
+						IB_QP_ALT_PATH			|
+						IB_QP_ACCESS_FLAGS		|
+						IB_QP_PATH_MIG_STATE),
+				[IB_QPT_RC]  = (IB_QP_CUR_STATE			|
+						IB_QP_ALT_PATH			|
+						IB_QP_ACCESS_FLAGS		|
+						IB_QP_MIN_RNR_TIMER		|
+						IB_QP_PATH_MIG_STATE),
+				[IB_QPT_SMI] = (IB_QP_CUR_STATE			|
+						IB_QP_QKEY),
+				[IB_QPT_GSI] = (IB_QP_CUR_STATE			|
+						IB_QP_QKEY),
+			}
+		},
+		[IB_QPS_SQD]   = {
+			.valid = 1,
+			.opt_param = {
+				[IB_QPT_UD]  = (IB_QP_PKEY_INDEX		|
+						IB_QP_QKEY),
+				[IB_QPT_UC]  = (IB_QP_AV			|
+						IB_QP_ALT_PATH			|
+						IB_QP_ACCESS_FLAGS		|
+						IB_QP_PKEY_INDEX		|
+						IB_QP_PATH_MIG_STATE),
+				[IB_QPT_RC]  = (IB_QP_PORT			|
+						IB_QP_AV			|
+						IB_QP_TIMEOUT			|
+						IB_QP_RETRY_CNT			|
+						IB_QP_RNR_RETRY			|
+						IB_QP_MAX_QP_RD_ATOMIC		|
+						IB_QP_MAX_DEST_RD_ATOMIC	|
+						IB_QP_ALT_PATH			|
+						IB_QP_ACCESS_FLAGS		|
+						IB_QP_PKEY_INDEX		|
+						IB_QP_MIN_RNR_TIMER		|
+						IB_QP_PATH_MIG_STATE),
+				[IB_QPT_SMI] = (IB_QP_PKEY_INDEX		|
+						IB_QP_QKEY),
+				[IB_QPT_GSI] = (IB_QP_PKEY_INDEX		|
+						IB_QP_QKEY),
+			}
+		}
+	},
+	[IB_QPS_SQE]   = {
+		[IB_QPS_RESET] = { .valid = 1 },
+		[IB_QPS_ERR] =   { .valid = 1 },
+		[IB_QPS_RTS]   = {
+			.valid = 1,
+			.opt_param = {
+				[IB_QPT_UD]  = (IB_QP_CUR_STATE			|
+						IB_QP_QKEY),
+				[IB_QPT_UC]  = (IB_QP_CUR_STATE			|
+						IB_QP_ACCESS_FLAGS),
+				[IB_QPT_SMI] = (IB_QP_CUR_STATE			|
+						IB_QP_QKEY),
+				[IB_QPT_GSI] = (IB_QP_CUR_STATE			|
+						IB_QP_QKEY),
+			}
+		}
+	},
+	[IB_QPS_ERR] = {
+		[IB_QPS_RESET] = { .valid = 1 },
+		[IB_QPS_ERR] =   { .valid = 1 }
+	}
+};
+
+int ib_modify_qp_is_ok(enum ib_qp_state cur_state, enum ib_qp_state next_state,
+		       enum ib_qp_type type, enum ib_qp_attr_mask mask)
+{
+	enum ib_qp_attr_mask req_param, opt_param;
+
+	if (cur_state  < 0 || cur_state  > IB_QPS_ERR ||
+	    next_state < 0 || next_state > IB_QPS_ERR)
+		return 0;
+
+	if (mask & IB_QP_CUR_STATE  &&
+	    cur_state != IB_QPS_RTR && cur_state != IB_QPS_RTS &&
+	    cur_state != IB_QPS_SQD && cur_state != IB_QPS_SQE)
+		return 0;
+
+	if (!qp_state_table[cur_state][next_state].valid)
+		return 0;
+
+	req_param = qp_state_table[cur_state][next_state].req_param[type];
+	opt_param = qp_state_table[cur_state][next_state].opt_param[type];
+
+	if ((mask & req_param) != req_param)
+		return 0;
+
+	if (mask & ~(req_param | opt_param | IB_QP_STATE))
+		return 0;
+
+	return 1;
+}
+EXPORT_SYMBOL(ib_modify_qp_is_ok);
+
+int ib_modify_qp(struct ib_qp *qp,
+		 struct ib_qp_attr *qp_attr,
+		 int qp_attr_mask)
+{
+	return qp->device->modify_qp(qp, qp_attr, qp_attr_mask, NULL);
+}
+EXPORT_SYMBOL(ib_modify_qp);
+
+int ib_query_qp(struct ib_qp *qp,
+		struct ib_qp_attr *qp_attr,
+		int qp_attr_mask,
+		struct ib_qp_init_attr *qp_init_attr)
+{
+	return qp->device->query_qp ?
+		qp->device->query_qp(qp, qp_attr, qp_attr_mask, qp_init_attr) :
+		-ENOSYS;
+}
+EXPORT_SYMBOL(ib_query_qp);
+
+int ib_destroy_qp(struct ib_qp *qp)
+{
+	struct ib_pd *pd;
+	struct ib_cq *scq, *rcq;
+	struct ib_srq *srq;
+	int ret;
+
+	pd  = qp->pd;
+	scq = qp->send_cq;
+	rcq = qp->recv_cq;
+	srq = qp->srq;
+
+	ret = qp->device->destroy_qp(qp);
+	if (!ret) {
+		atomic_dec(&pd->usecnt);
+		atomic_dec(&scq->usecnt);
+		atomic_dec(&rcq->usecnt);
+		if (srq)
+			atomic_dec(&srq->usecnt);
+	}
+
+	return ret;
+}
+EXPORT_SYMBOL(ib_destroy_qp);
+
+/* Completion queues */
+
+struct ib_cq *ib_create_cq(struct ib_device *device,
+			   ib_comp_handler comp_handler,
+			   void (*event_handler)(struct ib_event *, void *),
+			   void *cq_context, int cqe)
+{
+	struct ib_cq *cq;
+
+	cq = device->create_cq(device, cqe, NULL, NULL);
+
+	if (!IS_ERR(cq)) {
+		cq->device        = device;
+		cq->uobject       = NULL;
+		cq->comp_handler  = comp_handler;
+		cq->event_handler = event_handler;
+		cq->cq_context    = cq_context;
+		atomic_set(&cq->usecnt, 0);
+	}
+
+	return cq;
+}
+EXPORT_SYMBOL(ib_create_cq);
+
+int ib_destroy_cq(struct ib_cq *cq)
+{
+	if (atomic_read(&cq->usecnt))
+		return -EBUSY;
+
+	return cq->device->destroy_cq(cq);
+}
+EXPORT_SYMBOL(ib_destroy_cq);
+
+int ib_resize_cq(struct ib_cq *cq, int cqe)
+{
+	return cq->device->resize_cq ?
+		cq->device->resize_cq(cq, cqe, NULL) : -ENOSYS;
+}
+EXPORT_SYMBOL(ib_resize_cq);
+
+/* Memory regions */
+
+struct ib_mr *ib_get_dma_mr(struct ib_pd *pd, int mr_access_flags)
+{
+	struct ib_mr *mr;
+
+	mr = pd->device->get_dma_mr(pd, mr_access_flags);
+
+	if (!IS_ERR(mr)) {
+		mr->device  = pd->device;
+		mr->pd      = pd;
+		mr->uobject = NULL;
+		atomic_inc(&pd->usecnt);
+		atomic_set(&mr->usecnt, 0);
+	}
+
+	return mr;
+}
+EXPORT_SYMBOL(ib_get_dma_mr);
+
+struct ib_mr *ib_reg_phys_mr(struct ib_pd *pd,
+			     struct ib_phys_buf *phys_buf_array,
+			     int num_phys_buf,
+			     int mr_access_flags,
+			     u64 *iova_start)
+{
+	struct ib_mr *mr;
+
+	mr = pd->device->reg_phys_mr(pd, phys_buf_array, num_phys_buf,
+				     mr_access_flags, iova_start);
+
+	if (!IS_ERR(mr)) {
+		mr->device  = pd->device;
+		mr->pd      = pd;
+		mr->uobject = NULL;
+		atomic_inc(&pd->usecnt);
+		atomic_set(&mr->usecnt, 0);
+	}
+
+	return mr;
+}
+EXPORT_SYMBOL(ib_reg_phys_mr);
+
+int ib_rereg_phys_mr(struct ib_mr *mr,
+		     int mr_rereg_mask,
+		     struct ib_pd *pd,
+		     struct ib_phys_buf *phys_buf_array,
+		     int num_phys_buf,
+		     int mr_access_flags,
+		     u64 *iova_start)
+{
+	struct ib_pd *old_pd;
+	int ret;
+
+	if (!mr->device->rereg_phys_mr)
+		return -ENOSYS;
+
+	if (atomic_read(&mr->usecnt))
+		return -EBUSY;
+
+	old_pd = mr->pd;
+
+	ret = mr->device->rereg_phys_mr(mr, mr_rereg_mask, pd,
+					phys_buf_array, num_phys_buf,
+					mr_access_flags, iova_start);
+
+	if (!ret && (mr_rereg_mask & IB_MR_REREG_PD)) {
+		atomic_dec(&old_pd->usecnt);
+		atomic_inc(&pd->usecnt);
+	}
+
+	return ret;
+}
+EXPORT_SYMBOL(ib_rereg_phys_mr);
+
+int ib_query_mr(struct ib_mr *mr, struct ib_mr_attr *mr_attr)
+{
+	return mr->device->query_mr ?
+		mr->device->query_mr(mr, mr_attr) : -ENOSYS;
+}
+EXPORT_SYMBOL(ib_query_mr);
+
+int ib_dereg_mr(struct ib_mr *mr)
+{
+	struct ib_pd *pd;
+	int ret;
+
+	if (atomic_read(&mr->usecnt))
+		return -EBUSY;
+
+	pd = mr->pd;
+	ret = mr->device->dereg_mr(mr);
+	if (!ret)
+		atomic_dec(&pd->usecnt);
+
+	return ret;
+}
+EXPORT_SYMBOL(ib_dereg_mr);
+
+/* Memory windows */
+
+struct ib_mw *ib_alloc_mw(struct ib_pd *pd)
+{
+	struct ib_mw *mw;
+
+	if (!pd->device->alloc_mw)
+		return ERR_PTR(-ENOSYS);
+
+	mw = pd->device->alloc_mw(pd);
+	if (!IS_ERR(mw)) {
+		mw->device  = pd->device;
+		mw->pd      = pd;
+		mw->uobject = NULL;
+		atomic_inc(&pd->usecnt);
+	}
+
+	return mw;
+}
+EXPORT_SYMBOL(ib_alloc_mw);
+
+int ib_dealloc_mw(struct ib_mw *mw)
+{
+	struct ib_pd *pd;
+	int ret;
+
+	pd = mw->pd;
+	ret = mw->device->dealloc_mw(mw);
+	if (!ret)
+		atomic_dec(&pd->usecnt);
+
+	return ret;
+}
+EXPORT_SYMBOL(ib_dealloc_mw);
+
+/* "Fast" memory regions */
+
+struct ib_fmr *ib_alloc_fmr(struct ib_pd *pd,
+			    int mr_access_flags,
+			    struct ib_fmr_attr *fmr_attr)
+{
+	struct ib_fmr *fmr;
+
+	if (!pd->device->alloc_fmr)
+		return ERR_PTR(-ENOSYS);
+
+	fmr = pd->device->alloc_fmr(pd, mr_access_flags, fmr_attr);
+	if (!IS_ERR(fmr)) {
+		fmr->device = pd->device;
+		fmr->pd     = pd;
+		atomic_inc(&pd->usecnt);
+	}
+
+	return fmr;
+}
+EXPORT_SYMBOL(ib_alloc_fmr);
+
+int ib_unmap_fmr(struct list_head *fmr_list)
+{
+	struct ib_fmr *fmr;
+
+	if (list_empty(fmr_list))
+		return 0;
+
+	fmr = list_entry(fmr_list->next, struct ib_fmr, list);
+	return fmr->device->unmap_fmr(fmr_list);
+}
+EXPORT_SYMBOL(ib_unmap_fmr);
+
+int ib_dealloc_fmr(struct ib_fmr *fmr)
+{
+	struct ib_pd *pd;
+	int ret;
+
+	pd = fmr->pd;
+	ret = fmr->device->dealloc_fmr(fmr);
+	if (!ret)
+		atomic_dec(&pd->usecnt);
+
+	return ret;
+}
+EXPORT_SYMBOL(ib_dealloc_fmr);
+
+/* Multicast groups */
+
+int ib_attach_mcast(struct ib_qp *qp, union ib_gid *gid, u16 lid)
+{
+	if (!qp->device->attach_mcast)
+		return -ENOSYS;
+	if (gid->raw[0] != 0xff || qp->qp_type != IB_QPT_UD)
+		return -EINVAL;
+
+	return qp->device->attach_mcast(qp, gid, lid);
+}
+EXPORT_SYMBOL(ib_attach_mcast);
+
+int ib_detach_mcast(struct ib_qp *qp, union ib_gid *gid, u16 lid)
+{
+	if (!qp->device->detach_mcast)
+		return -ENOSYS;
+	if (gid->raw[0] != 0xff || qp->qp_type != IB_QPT_UD)
+		return -EINVAL;
+
+	return qp->device->detach_mcast(qp, gid, lid);
+}
+EXPORT_SYMBOL(ib_detach_mcast);