This commit was manufactured by cvs2svn to create branch 'vserver'.
[linux-2.6.git] / drivers / infiniband / core / user_mad.c
diff --git a/drivers/infiniband/core/user_mad.c b/drivers/infiniband/core/user_mad.c
new file mode 100644 (file)
index 0000000..54b4f33
--- /dev/null
@@ -0,0 +1,840 @@
+/*
+ * 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: user_mad.c 1389 2004-12-27 22:56:47Z roland $
+ */
+
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/device.h>
+#include <linux/err.h>
+#include <linux/fs.h>
+#include <linux/cdev.h>
+#include <linux/pci.h>
+#include <linux/dma-mapping.h>
+#include <linux/poll.h>
+#include <linux/rwsem.h>
+#include <linux/kref.h>
+
+#include <asm/uaccess.h>
+#include <asm/semaphore.h>
+
+#include <ib_mad.h>
+#include <ib_user_mad.h>
+
+MODULE_AUTHOR("Roland Dreier");
+MODULE_DESCRIPTION("InfiniBand userspace MAD packet access");
+MODULE_LICENSE("Dual BSD/GPL");
+
+enum {
+       IB_UMAD_MAX_PORTS  = 64,
+       IB_UMAD_MAX_AGENTS = 32,
+
+       IB_UMAD_MAJOR      = 231,
+       IB_UMAD_MINOR_BASE = 0
+};
+
+struct ib_umad_port {
+       int                    devnum;
+       struct cdev            dev;
+       struct class_device    class_dev;
+
+       int                    sm_devnum;
+       struct cdev            sm_dev;
+       struct class_device    sm_class_dev;
+       struct semaphore       sm_sem;
+
+       struct ib_device      *ib_dev;
+       struct ib_umad_device *umad_dev;
+       u8                     port_num;
+};
+
+struct ib_umad_device {
+       int                  start_port, end_port;
+       struct kref          ref;
+       struct ib_umad_port  port[0];
+};
+
+struct ib_umad_file {
+       struct ib_umad_port *port;
+       spinlock_t           recv_lock;
+       struct list_head     recv_list;
+       wait_queue_head_t    recv_wait;
+       struct rw_semaphore  agent_mutex;
+       struct ib_mad_agent *agent[IB_UMAD_MAX_AGENTS];
+       struct ib_mr        *mr[IB_UMAD_MAX_AGENTS];
+};
+
+struct ib_umad_packet {
+       struct ib_user_mad mad;
+       struct ib_ah      *ah;
+       struct list_head   list;
+       DECLARE_PCI_UNMAP_ADDR(mapping)
+};
+
+static const dev_t base_dev = MKDEV(IB_UMAD_MAJOR, IB_UMAD_MINOR_BASE);
+static spinlock_t map_lock;
+static DECLARE_BITMAP(dev_map, IB_UMAD_MAX_PORTS * 2);
+
+static void ib_umad_add_one(struct ib_device *device);
+static void ib_umad_remove_one(struct ib_device *device);
+
+static int queue_packet(struct ib_umad_file *file,
+                       struct ib_mad_agent *agent,
+                       struct ib_umad_packet *packet)
+{
+       int ret = 1;
+
+       down_read(&file->agent_mutex);
+       for (packet->mad.id = 0;
+            packet->mad.id < IB_UMAD_MAX_AGENTS;
+            packet->mad.id++)
+               if (agent == file->agent[packet->mad.id]) {
+                       spin_lock_irq(&file->recv_lock);
+                       list_add_tail(&packet->list, &file->recv_list);
+                       spin_unlock_irq(&file->recv_lock);
+                       wake_up_interruptible(&file->recv_wait);
+                       ret = 0;
+                       break;
+               }
+
+       up_read(&file->agent_mutex);
+
+       return ret;
+}
+
+static void send_handler(struct ib_mad_agent *agent,
+                        struct ib_mad_send_wc *send_wc)
+{
+       struct ib_umad_file *file = agent->context;
+       struct ib_umad_packet *packet =
+               (void *) (unsigned long) send_wc->wr_id;
+
+       dma_unmap_single(agent->device->dma_device,
+                        pci_unmap_addr(packet, mapping),
+                        sizeof packet->mad.data,
+                        DMA_TO_DEVICE);
+       ib_destroy_ah(packet->ah);
+
+       if (send_wc->status == IB_WC_RESP_TIMEOUT_ERR) {
+               packet->mad.status = ETIMEDOUT;
+
+               if (!queue_packet(file, agent, packet))
+                       return;
+       }
+
+       kfree(packet);
+}
+
+static void recv_handler(struct ib_mad_agent *agent,
+                        struct ib_mad_recv_wc *mad_recv_wc)
+{
+       struct ib_umad_file *file = agent->context;
+       struct ib_umad_packet *packet;
+
+       if (mad_recv_wc->wc->status != IB_WC_SUCCESS)
+               goto out;
+
+       packet = kmalloc(sizeof *packet, GFP_KERNEL);
+       if (!packet)
+               goto out;
+
+       memset(packet, 0, sizeof *packet);
+
+       memcpy(packet->mad.data, mad_recv_wc->recv_buf.mad, sizeof packet->mad.data);
+       packet->mad.status        = 0;
+       packet->mad.qpn           = cpu_to_be32(mad_recv_wc->wc->src_qp);
+       packet->mad.lid           = cpu_to_be16(mad_recv_wc->wc->slid);
+       packet->mad.sl            = mad_recv_wc->wc->sl;
+       packet->mad.path_bits     = mad_recv_wc->wc->dlid_path_bits;
+       packet->mad.grh_present   = !!(mad_recv_wc->wc->wc_flags & IB_WC_GRH);
+       if (packet->mad.grh_present) {
+               /* XXX parse GRH */
+               packet->mad.gid_index     = 0;
+               packet->mad.hop_limit     = 0;
+               packet->mad.traffic_class = 0;
+               memset(packet->mad.gid, 0, 16);
+               packet->mad.flow_label    = 0;
+       }
+
+       if (queue_packet(file, agent, packet))
+               kfree(packet);
+
+out:
+       ib_free_recv_mad(mad_recv_wc);
+}
+
+static ssize_t ib_umad_read(struct file *filp, char __user *buf,
+                           size_t count, loff_t *pos)
+{
+       struct ib_umad_file *file = filp->private_data;
+       struct ib_umad_packet *packet;
+       ssize_t ret;
+
+       if (count < sizeof (struct ib_user_mad))
+               return -EINVAL;
+
+       spin_lock_irq(&file->recv_lock);
+
+       while (list_empty(&file->recv_list)) {
+               spin_unlock_irq(&file->recv_lock);
+
+               if (filp->f_flags & O_NONBLOCK)
+                       return -EAGAIN;
+
+               if (wait_event_interruptible(file->recv_wait,
+                                            !list_empty(&file->recv_list)))
+                       return -ERESTARTSYS;
+
+               spin_lock_irq(&file->recv_lock);
+       }
+
+       packet = list_entry(file->recv_list.next, struct ib_umad_packet, list);
+       list_del(&packet->list);
+
+       spin_unlock_irq(&file->recv_lock);
+
+       if (copy_to_user(buf, &packet->mad, sizeof packet->mad))
+               ret = -EFAULT;
+       else
+               ret = sizeof packet->mad;
+
+       kfree(packet);
+       return ret;
+}
+
+static ssize_t ib_umad_write(struct file *filp, const char __user *buf,
+                            size_t count, loff_t *pos)
+{
+       struct ib_umad_file *file = filp->private_data;
+       struct ib_umad_packet *packet;
+       struct ib_mad_agent *agent;
+       struct ib_ah_attr ah_attr;
+       struct ib_sge      gather_list;
+       struct ib_send_wr *bad_wr, wr = {
+               .opcode      = IB_WR_SEND,
+               .sg_list     = &gather_list,
+               .num_sge     = 1,
+               .send_flags  = IB_SEND_SIGNALED,
+       };
+       u8 method;
+       u64 *tid;
+       int ret;
+
+       if (count < sizeof (struct ib_user_mad))
+               return -EINVAL;
+
+       packet = kmalloc(sizeof *packet, GFP_KERNEL);
+       if (!packet)
+               return -ENOMEM;
+
+       if (copy_from_user(&packet->mad, buf, sizeof packet->mad)) {
+               kfree(packet);
+               return -EFAULT;
+       }
+
+       if (packet->mad.id < 0 || packet->mad.id >= IB_UMAD_MAX_AGENTS) {
+               ret = -EINVAL;
+               goto err;
+       }
+
+       down_read(&file->agent_mutex);
+
+       agent = file->agent[packet->mad.id];
+       if (!agent) {
+               ret = -EINVAL;
+               goto err_up;
+       }
+
+       /*
+        * If userspace is generating a request that will generate a
+        * response, we need to make sure the high-order part of the
+        * transaction ID matches the agent being used to send the
+        * MAD.
+        */
+       method = ((struct ib_mad_hdr *) packet->mad.data)->method;
+
+       if (!(method & IB_MGMT_METHOD_RESP)       &&
+           method != IB_MGMT_METHOD_TRAP_REPRESS &&
+           method != IB_MGMT_METHOD_SEND) {
+               tid = &((struct ib_mad_hdr *) packet->mad.data)->tid;
+               *tid = cpu_to_be64(((u64) agent->hi_tid) << 32 |
+                                  (be64_to_cpup(tid) & 0xffffffff));
+       }
+
+       memset(&ah_attr, 0, sizeof ah_attr);
+       ah_attr.dlid          = be16_to_cpu(packet->mad.lid);
+       ah_attr.sl            = packet->mad.sl;
+       ah_attr.src_path_bits = packet->mad.path_bits;
+       ah_attr.port_num      = file->port->port_num;
+       if (packet->mad.grh_present) {
+               ah_attr.ah_flags = IB_AH_GRH;
+               memcpy(ah_attr.grh.dgid.raw, packet->mad.gid, 16);
+               ah_attr.grh.flow_label     = packet->mad.flow_label;
+               ah_attr.grh.hop_limit      = packet->mad.hop_limit;
+               ah_attr.grh.traffic_class  = packet->mad.traffic_class;
+       }
+
+       packet->ah = ib_create_ah(agent->qp->pd, &ah_attr);
+       if (IS_ERR(packet->ah)) {
+               ret = PTR_ERR(packet->ah);
+               goto err_up;
+       }
+
+       gather_list.addr = dma_map_single(agent->device->dma_device,
+                                         packet->mad.data,
+                                         sizeof packet->mad.data,
+                                         DMA_TO_DEVICE);
+       gather_list.length = sizeof packet->mad.data;
+       gather_list.lkey   = file->mr[packet->mad.id]->lkey;
+       pci_unmap_addr_set(packet, mapping, gather_list.addr);
+
+       wr.wr.ud.mad_hdr     = (struct ib_mad_hdr *) packet->mad.data;
+       wr.wr.ud.ah          = packet->ah;
+       wr.wr.ud.remote_qpn  = be32_to_cpu(packet->mad.qpn);
+       wr.wr.ud.remote_qkey = be32_to_cpu(packet->mad.qkey);
+       wr.wr.ud.timeout_ms  = packet->mad.timeout_ms;
+
+       wr.wr_id            = (unsigned long) packet;
+
+       ret = ib_post_send_mad(agent, &wr, &bad_wr);
+       if (ret) {
+               dma_unmap_single(agent->device->dma_device,
+                                pci_unmap_addr(packet, mapping),
+                                sizeof packet->mad.data,
+                                DMA_TO_DEVICE);
+               goto err_up;
+       }
+
+       up_read(&file->agent_mutex);
+
+       return sizeof packet->mad;
+
+err_up:
+       up_read(&file->agent_mutex);
+
+err:
+       kfree(packet);
+       return ret;
+}
+
+static unsigned int ib_umad_poll(struct file *filp, struct poll_table_struct *wait)
+{
+       struct ib_umad_file *file = filp->private_data;
+
+       /* we will always be able to post a MAD send */
+       unsigned int mask = POLLOUT | POLLWRNORM;
+
+       poll_wait(filp, &file->recv_wait, wait);
+
+       if (!list_empty(&file->recv_list))
+               mask |= POLLIN | POLLRDNORM;
+
+       return mask;
+}
+
+static int ib_umad_reg_agent(struct ib_umad_file *file, unsigned long arg)
+{
+       struct ib_user_mad_reg_req ureq;
+       struct ib_mad_reg_req req;
+       struct ib_mad_agent *agent;
+       int agent_id;
+       int ret;
+
+       down_write(&file->agent_mutex);
+
+       if (copy_from_user(&ureq, (void __user *) arg, sizeof ureq)) {
+               ret = -EFAULT;
+               goto out;
+       }
+
+       if (ureq.qpn != 0 && ureq.qpn != 1) {
+               ret = -EINVAL;
+               goto out;
+       }
+
+       for (agent_id = 0; agent_id < IB_UMAD_MAX_AGENTS; ++agent_id)
+               if (!file->agent[agent_id])
+                       goto found;
+
+       ret = -ENOMEM;
+       goto out;
+
+found:
+       req.mgmt_class         = ureq.mgmt_class;
+       req.mgmt_class_version = ureq.mgmt_class_version;
+       memcpy(req.method_mask, ureq.method_mask, sizeof req.method_mask);
+       memcpy(req.oui,         ureq.oui,         sizeof req.oui);
+
+       agent = ib_register_mad_agent(file->port->ib_dev, file->port->port_num,
+                                     ureq.qpn ? IB_QPT_GSI : IB_QPT_SMI,
+                                     &req, 0, send_handler, recv_handler,
+                                     file);
+       if (IS_ERR(agent)) {
+               ret = PTR_ERR(agent);
+               goto out;
+       }
+
+       file->agent[agent_id] = agent;
+
+       file->mr[agent_id] = ib_get_dma_mr(agent->qp->pd, IB_ACCESS_LOCAL_WRITE);
+       if (IS_ERR(file->mr[agent_id])) {
+               ret = -ENOMEM;
+               goto err;
+       }
+
+       if (put_user(agent_id,
+                    (u32 __user *) (arg + offsetof(struct ib_user_mad_reg_req, id)))) {
+               ret = -EFAULT;
+               goto err_mr;
+       }
+
+       ret = 0;
+       goto out;
+
+err_mr:
+       ib_dereg_mr(file->mr[agent_id]);
+
+err:
+       file->agent[agent_id] = NULL;
+       ib_unregister_mad_agent(agent);
+
+out:
+       up_write(&file->agent_mutex);
+       return ret;
+}
+
+static int ib_umad_unreg_agent(struct ib_umad_file *file, unsigned long arg)
+{
+       u32 id;
+       int ret = 0;
+
+       down_write(&file->agent_mutex);
+
+       if (get_user(id, (u32 __user *) arg)) {
+               ret = -EFAULT;
+               goto out;
+       }
+
+       if (id < 0 || id >= IB_UMAD_MAX_AGENTS || !file->agent[id]) {
+               ret = -EINVAL;
+               goto out;
+       }
+
+       ib_dereg_mr(file->mr[id]);
+       ib_unregister_mad_agent(file->agent[id]);
+       file->agent[id] = NULL;
+
+out:
+       up_write(&file->agent_mutex);
+       return ret;
+}
+
+static long ib_umad_ioctl(struct file *filp,
+                        unsigned int cmd, unsigned long arg)
+{
+       switch (cmd) {
+       case IB_USER_MAD_REGISTER_AGENT:
+               return ib_umad_reg_agent(filp->private_data, arg);
+       case IB_USER_MAD_UNREGISTER_AGENT:
+               return ib_umad_unreg_agent(filp->private_data, arg);
+       default:
+               return -ENOIOCTLCMD;
+       }
+}
+
+static int ib_umad_open(struct inode *inode, struct file *filp)
+{
+       struct ib_umad_port *port =
+               container_of(inode->i_cdev, struct ib_umad_port, dev);
+       struct ib_umad_file *file;
+
+       file = kmalloc(sizeof *file, GFP_KERNEL);
+       if (!file)
+               return -ENOMEM;
+
+       memset(file, 0, sizeof *file);
+
+       spin_lock_init(&file->recv_lock);
+       init_rwsem(&file->agent_mutex);
+       INIT_LIST_HEAD(&file->recv_list);
+       init_waitqueue_head(&file->recv_wait);
+
+       file->port = port;
+       filp->private_data = file;
+
+       return 0;
+}
+
+static int ib_umad_close(struct inode *inode, struct file *filp)
+{
+       struct ib_umad_file *file = filp->private_data;
+       int i;
+
+       for (i = 0; i < IB_UMAD_MAX_AGENTS; ++i)
+               if (file->agent[i]) {
+                       ib_dereg_mr(file->mr[i]);
+                       ib_unregister_mad_agent(file->agent[i]);
+               }
+
+       kfree(file);
+
+       return 0;
+}
+
+static struct file_operations umad_fops = {
+       .owner          = THIS_MODULE,
+       .read           = ib_umad_read,
+       .write          = ib_umad_write,
+       .poll           = ib_umad_poll,
+       .unlocked_ioctl = ib_umad_ioctl,
+       .compat_ioctl   = ib_umad_ioctl,
+       .open           = ib_umad_open,
+       .release        = ib_umad_close
+};
+
+static int ib_umad_sm_open(struct inode *inode, struct file *filp)
+{
+       struct ib_umad_port *port =
+               container_of(inode->i_cdev, struct ib_umad_port, sm_dev);
+       struct ib_port_modify props = {
+               .set_port_cap_mask = IB_PORT_SM
+       };
+       int ret;
+
+       if (filp->f_flags & O_NONBLOCK) {
+               if (down_trylock(&port->sm_sem))
+                       return -EAGAIN;
+       } else {
+               if (down_interruptible(&port->sm_sem))
+                       return -ERESTARTSYS;
+       }
+
+       ret = ib_modify_port(port->ib_dev, port->port_num, 0, &props);
+       if (ret) {
+               up(&port->sm_sem);
+               return ret;
+       }
+
+       filp->private_data = port;
+
+       return 0;
+}
+
+static int ib_umad_sm_close(struct inode *inode, struct file *filp)
+{
+       struct ib_umad_port *port = filp->private_data;
+       struct ib_port_modify props = {
+               .clr_port_cap_mask = IB_PORT_SM
+       };
+       int ret;
+
+       ret = ib_modify_port(port->ib_dev, port->port_num, 0, &props);
+       up(&port->sm_sem);
+
+       return ret;
+}
+
+static struct file_operations umad_sm_fops = {
+       .owner   = THIS_MODULE,
+       .open    = ib_umad_sm_open,
+       .release = ib_umad_sm_close
+};
+
+static struct ib_client umad_client = {
+       .name   = "umad",
+       .add    = ib_umad_add_one,
+       .remove = ib_umad_remove_one
+};
+
+static ssize_t show_dev(struct class_device *class_dev, char *buf)
+{
+       struct ib_umad_port *port = class_get_devdata(class_dev);
+
+       if (class_dev == &port->class_dev)
+               return print_dev_t(buf, port->dev.dev);
+       else
+               return print_dev_t(buf, port->sm_dev.dev);
+}
+static CLASS_DEVICE_ATTR(dev, S_IRUGO, show_dev, NULL);
+
+static ssize_t show_ibdev(struct class_device *class_dev, char *buf)
+{
+       struct ib_umad_port *port = class_get_devdata(class_dev);
+
+       return sprintf(buf, "%s\n", port->ib_dev->name);
+}
+static CLASS_DEVICE_ATTR(ibdev, S_IRUGO, show_ibdev, NULL);
+
+static ssize_t show_port(struct class_device *class_dev, char *buf)
+{
+       struct ib_umad_port *port = class_get_devdata(class_dev);
+
+       return sprintf(buf, "%d\n", port->port_num);
+}
+static CLASS_DEVICE_ATTR(port, S_IRUGO, show_port, NULL);
+
+static void ib_umad_release_dev(struct kref *ref)
+{
+       struct ib_umad_device *dev =
+               container_of(ref, struct ib_umad_device, ref);
+
+       kfree(dev);
+}
+
+static void ib_umad_release_port(struct class_device *class_dev)
+{
+       struct ib_umad_port *port = class_get_devdata(class_dev);
+
+       if (class_dev == &port->class_dev) {
+               cdev_del(&port->dev);
+               clear_bit(port->devnum, dev_map);
+       } else {
+               cdev_del(&port->sm_dev);
+               clear_bit(port->sm_devnum, dev_map);
+       }
+
+       kref_put(&port->umad_dev->ref, ib_umad_release_dev);
+}
+
+static struct class umad_class = {
+       .name    = "infiniband_mad",
+       .release = ib_umad_release_port
+};
+
+static ssize_t show_abi_version(struct class *class, char *buf)
+{
+       return sprintf(buf, "%d\n", IB_USER_MAD_ABI_VERSION);
+}
+static CLASS_ATTR(abi_version, S_IRUGO, show_abi_version, NULL);
+
+static int ib_umad_init_port(struct ib_device *device, int port_num,
+                            struct ib_umad_port *port)
+{
+       spin_lock(&map_lock);
+       port->devnum = find_first_zero_bit(dev_map, IB_UMAD_MAX_PORTS);
+       if (port->devnum >= IB_UMAD_MAX_PORTS) {
+               spin_unlock(&map_lock);
+               return -1;
+       }
+       port->sm_devnum = find_next_zero_bit(dev_map, IB_UMAD_MAX_PORTS * 2, IB_UMAD_MAX_PORTS);
+       if (port->sm_devnum >= IB_UMAD_MAX_PORTS * 2) {
+               spin_unlock(&map_lock);
+               return -1;
+       }
+       set_bit(port->devnum, dev_map);
+       set_bit(port->sm_devnum, dev_map);
+       spin_unlock(&map_lock);
+
+       port->ib_dev   = device;
+       port->port_num = port_num;
+       init_MUTEX(&port->sm_sem);
+
+       cdev_init(&port->dev, &umad_fops);
+       port->dev.owner = THIS_MODULE;
+       kobject_set_name(&port->dev.kobj, "umad%d", port->devnum);
+       if (cdev_add(&port->dev, base_dev + port->devnum, 1))
+               return -1;
+
+       port->class_dev.class = &umad_class;
+       port->class_dev.dev   = device->dma_device;
+
+       snprintf(port->class_dev.class_id, BUS_ID_SIZE, "umad%d", port->devnum);
+
+       if (class_device_register(&port->class_dev))
+               goto err_cdev;
+
+       class_set_devdata(&port->class_dev, port);
+       kref_get(&port->umad_dev->ref);
+
+       if (class_device_create_file(&port->class_dev, &class_device_attr_dev))
+               goto err_class;
+       if (class_device_create_file(&port->class_dev, &class_device_attr_ibdev))
+               goto err_class;
+       if (class_device_create_file(&port->class_dev, &class_device_attr_port))
+               goto err_class;
+
+       cdev_init(&port->sm_dev, &umad_sm_fops);
+       port->sm_dev.owner = THIS_MODULE;
+       kobject_set_name(&port->dev.kobj, "issm%d", port->sm_devnum - IB_UMAD_MAX_PORTS);
+       if (cdev_add(&port->sm_dev, base_dev + port->sm_devnum, 1))
+               return -1;
+
+       port->sm_class_dev.class = &umad_class;
+       port->sm_class_dev.dev   = device->dma_device;
+
+       snprintf(port->sm_class_dev.class_id, BUS_ID_SIZE, "issm%d", port->sm_devnum - IB_UMAD_MAX_PORTS);
+
+       if (class_device_register(&port->sm_class_dev))
+               goto err_sm_cdev;
+
+       class_set_devdata(&port->sm_class_dev, port);
+       kref_get(&port->umad_dev->ref);
+
+       if (class_device_create_file(&port->sm_class_dev, &class_device_attr_dev))
+               goto err_sm_class;
+       if (class_device_create_file(&port->sm_class_dev, &class_device_attr_ibdev))
+               goto err_sm_class;
+       if (class_device_create_file(&port->sm_class_dev, &class_device_attr_port))
+               goto err_sm_class;
+
+       return 0;
+
+err_sm_class:
+       class_device_unregister(&port->sm_class_dev);
+
+err_sm_cdev:
+       cdev_del(&port->sm_dev);
+
+err_class:
+       class_device_unregister(&port->class_dev);
+
+err_cdev:
+       cdev_del(&port->dev);
+       clear_bit(port->devnum, dev_map);
+
+       return -1;
+}
+
+static void ib_umad_add_one(struct ib_device *device)
+{
+       struct ib_umad_device *umad_dev;
+       int s, e, i;
+
+       if (device->node_type == IB_NODE_SWITCH)
+               s = e = 0;
+       else {
+               s = 1;
+               e = device->phys_port_cnt;
+       }
+
+       umad_dev = kmalloc(sizeof *umad_dev +
+                          (e - s + 1) * sizeof (struct ib_umad_port),
+                          GFP_KERNEL);
+       if (!umad_dev)
+               return;
+
+       memset(umad_dev, 0, sizeof *umad_dev +
+              (e - s + 1) * sizeof (struct ib_umad_port));
+
+       kref_init(&umad_dev->ref);
+
+       umad_dev->start_port = s;
+       umad_dev->end_port   = e;
+
+       for (i = s; i <= e; ++i) {
+               umad_dev->port[i - s].umad_dev = umad_dev;
+
+               if (ib_umad_init_port(device, i, &umad_dev->port[i - s]))
+                       goto err;
+       }
+
+       ib_set_client_data(device, &umad_client, umad_dev);
+
+       return;
+
+err:
+       while (--i >= s) {
+               class_device_unregister(&umad_dev->port[i - s].class_dev);
+               class_device_unregister(&umad_dev->port[i - s].sm_class_dev);
+       }
+
+       kref_put(&umad_dev->ref, ib_umad_release_dev);
+}
+
+static void ib_umad_remove_one(struct ib_device *device)
+{
+       struct ib_umad_device *umad_dev = ib_get_client_data(device, &umad_client);
+       int i;
+
+       if (!umad_dev)
+               return;
+
+       for (i = 0; i <= umad_dev->end_port - umad_dev->start_port; ++i) {
+               class_device_unregister(&umad_dev->port[i].class_dev);
+               class_device_unregister(&umad_dev->port[i].sm_class_dev);
+       }
+
+       kref_put(&umad_dev->ref, ib_umad_release_dev);
+}
+
+static int __init ib_umad_init(void)
+{
+       int ret;
+
+       spin_lock_init(&map_lock);
+
+       ret = register_chrdev_region(base_dev, IB_UMAD_MAX_PORTS * 2,
+                                    "infiniband_mad");
+       if (ret) {
+               printk(KERN_ERR "user_mad: couldn't register device number\n");
+               goto out;
+       }
+
+       ret = class_register(&umad_class);
+       if (ret) {
+               printk(KERN_ERR "user_mad: couldn't create class infiniband_mad\n");
+               goto out_chrdev;
+       }
+
+       ret = class_create_file(&umad_class, &class_attr_abi_version);
+       if (ret) {
+               printk(KERN_ERR "user_mad: couldn't create abi_version attribute\n");
+               goto out_class;
+       }
+
+       ret = ib_register_client(&umad_client);
+       if (ret) {
+               printk(KERN_ERR "user_mad: couldn't register ib_umad client\n");
+               goto out_class;
+       }
+
+       return 0;
+
+out_class:
+       class_unregister(&umad_class);
+
+out_chrdev:
+       unregister_chrdev_region(base_dev, IB_UMAD_MAX_PORTS * 2);
+
+out:
+       return ret;
+}
+
+static void __exit ib_umad_cleanup(void)
+{
+       ib_unregister_client(&umad_client);
+       class_unregister(&umad_class);
+       unregister_chrdev_region(base_dev, IB_UMAD_MAX_PORTS * 2);
+}
+
+module_init(ib_umad_init);
+module_exit(ib_umad_cleanup);