[SCSI] cnic: Add new Broadcom CNIC driver.

The CNIC driver controls BNX2 hardware rings and resources used by
iSCSI.  Most hardware resources for iSCSI are separate from those
used for ethernet networking.

iSCSI uses a separate MAC address and IP address.  The CNIC driver
creates a UIO interface to handle the non-offloaded packets such as
ARP, etc in userspace.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>

6 files changed, 3901 insertions, 0 deletions

diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index 214a92d1ef75..f3c4a3b910bb 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -2264,6 +2264,17 @@ config BNX2
 	  To compile this driver as a module, choose M here: the module
 	  will be called bnx2.  This is recommended.
 
+config CNIC
+	tristate "Broadcom CNIC support"
+	depends on BNX2
+	depends on UIO
+	help
+	  This driver supports offload features of Broadcom NetXtremeII
+	  gigabit Ethernet cards.
+
+	  To compile this driver as a module, choose M here: the module
+	  will be called cnic.  This is recommended.
+
 config SPIDER_NET
 	tristate "Spider Gigabit Ethernet driver"
 	depends on PCI && (PPC_IBM_CELL_BLADE || PPC_CELLEB)
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index 1fc4602a6ff2..e6f1f8c3f8d4 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -73,6 +73,7 @@ obj-$(CONFIG_STNIC) += stnic.o 8390.o
 obj-$(CONFIG_FEALNX) += fealnx.o
 obj-$(CONFIG_TIGON3) += tg3.o
 obj-$(CONFIG_BNX2) += bnx2.o
+obj-$(CONFIG_CNIC) += cnic.o
 obj-$(CONFIG_BNX2X) += bnx2x.o
 bnx2x-objs := bnx2x_main.o bnx2x_link.o
 spidernet-y += spider_net.o spider_net_ethtool.o
diff --git a/drivers/net/cnic.c b/drivers/net/cnic.c
new file mode 100644
index 000000000000..8d740376bbd2
--- /dev/null
+++ b/drivers/net/cnic.c
@@ -0,0 +1,2711 @@
+/* cnic.c: Broadcom CNIC core network driver.
+ *
+ * Copyright (c) 2006-2009 Broadcom Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation.
+ *
+ * Original skeleton written by: John(Zongxi) Chen (zongxi@broadcom.com)
+ * Modified and maintained by: Michael Chan <mchan@broadcom.com>
+ */
+
+#include <linux/module.h>
+
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/list.h>
+#include <linux/slab.h>
+#include <linux/pci.h>
+#include <linux/init.h>
+#include <linux/netdevice.h>
+#include <linux/uio_driver.h>
+#include <linux/in.h>
+#include <linux/dma-mapping.h>
+#include <linux/delay.h>
+#include <linux/ethtool.h>
+#include <linux/if_vlan.h>
+#if defined(CONFIG_VLAN_8021Q) || defined(CONFIG_VLAN_8021Q_MODULE)
+#define BCM_VLAN 1
+#endif
+#include <net/ip.h>
+#include <net/tcp.h>
+#include <net/route.h>
+#include <net/ipv6.h>
+#include <net/ip6_route.h>
+#include <scsi/iscsi_if.h>
+
+#include "cnic_if.h"
+#include "bnx2.h"
+#include "cnic.h"
+#include "cnic_defs.h"
+
+#define DRV_MODULE_NAME		"cnic"
+#define PFX DRV_MODULE_NAME	": "
+
+static char version[] __devinitdata =
+	"Broadcom NetXtreme II CNIC Driver " DRV_MODULE_NAME " v" CNIC_MODULE_VERSION " (" CNIC_MODULE_RELDATE ")\n";
+
+MODULE_AUTHOR("Michael Chan <mchan@broadcom.com> and John(Zongxi) "
+	      "Chen (zongxi@broadcom.com");
+MODULE_DESCRIPTION("Broadcom NetXtreme II CNIC Driver");
+MODULE_LICENSE("GPL");
+MODULE_VERSION(CNIC_MODULE_VERSION);
+
+static LIST_HEAD(cnic_dev_list);
+static DEFINE_RWLOCK(cnic_dev_lock);
+static DEFINE_MUTEX(cnic_lock);
+
+static struct cnic_ulp_ops *cnic_ulp_tbl[MAX_CNIC_ULP_TYPE];
+
+static int cnic_service_bnx2(void *, void *);
+static int cnic_ctl(void *, struct cnic_ctl_info *);
+
+static struct cnic_ops cnic_bnx2_ops = {
+	.cnic_owner	= THIS_MODULE,
+	.cnic_handler	= cnic_service_bnx2,
+	.cnic_ctl	= cnic_ctl,
+};
+
+static void cnic_shutdown_bnx2_rx_ring(struct cnic_dev *);
+static void cnic_init_bnx2_tx_ring(struct cnic_dev *);
+static void cnic_init_bnx2_rx_ring(struct cnic_dev *);
+static int cnic_cm_set_pg(struct cnic_sock *);
+
+static int cnic_uio_open(struct uio_info *uinfo, struct inode *inode)
+{
+	struct cnic_dev *dev = uinfo->priv;
+	struct cnic_local *cp = dev->cnic_priv;
+
+	if (!capable(CAP_NET_ADMIN))
+		return -EPERM;
+
+	if (cp->uio_dev != -1)
+		return -EBUSY;
+
+	cp->uio_dev = iminor(inode);
+
+	cnic_shutdown_bnx2_rx_ring(dev);
+
+	cnic_init_bnx2_tx_ring(dev);
+	cnic_init_bnx2_rx_ring(dev);
+
+	return 0;
+}
+
+static int cnic_uio_close(struct uio_info *uinfo, struct inode *inode)
+{
+	struct cnic_dev *dev = uinfo->priv;
+	struct cnic_local *cp = dev->cnic_priv;
+
+	cp->uio_dev = -1;
+	return 0;
+}
+
+static inline void cnic_hold(struct cnic_dev *dev)
+{
+	atomic_inc(&dev->ref_count);
+}
+
+static inline void cnic_put(struct cnic_dev *dev)
+{
+	atomic_dec(&dev->ref_count);
+}
+
+static inline void csk_hold(struct cnic_sock *csk)
+{
+	atomic_inc(&csk->ref_count);
+}
+
+static inline void csk_put(struct cnic_sock *csk)
+{
+	atomic_dec(&csk->ref_count);
+}
+
+static struct cnic_dev *cnic_from_netdev(struct net_device *netdev)
+{
+	struct cnic_dev *cdev;
+
+	read_lock(&cnic_dev_lock);
+	list_for_each_entry(cdev, &cnic_dev_list, list) {
+		if (netdev == cdev->netdev) {
+			cnic_hold(cdev);
+			read_unlock(&cnic_dev_lock);
+			return cdev;
+		}
+	}
+	read_unlock(&cnic_dev_lock);
+	return NULL;
+}
+
+static void cnic_ctx_wr(struct cnic_dev *dev, u32 cid_addr, u32 off, u32 val)
+{
+	struct cnic_local *cp = dev->cnic_priv;
+	struct cnic_eth_dev *ethdev = cp->ethdev;
+	struct drv_ctl_info info;
+	struct drv_ctl_io *io = &info.data.io;
+
+	info.cmd = DRV_CTL_CTX_WR_CMD;
+	io->cid_addr = cid_addr;
+	io->offset = off;
+	io->data = val;
+	ethdev->drv_ctl(dev->netdev, &info);
+}
+
+static void cnic_reg_wr_ind(struct cnic_dev *dev, u32 off, u32 val)
+{
+	struct cnic_local *cp = dev->cnic_priv;
+	struct cnic_eth_dev *ethdev = cp->ethdev;
+	struct drv_ctl_info info;
+	struct drv_ctl_io *io = &info.data.io;
+
+	info.cmd = DRV_CTL_IO_WR_CMD;
+	io->offset = off;
+	io->data = val;
+	ethdev->drv_ctl(dev->netdev, &info);
+}
+
+static u32 cnic_reg_rd_ind(struct cnic_dev *dev, u32 off)
+{
+	struct cnic_local *cp = dev->cnic_priv;
+	struct cnic_eth_dev *ethdev = cp->ethdev;
+	struct drv_ctl_info info;
+	struct drv_ctl_io *io = &info.data.io;
+
+	info.cmd = DRV_CTL_IO_RD_CMD;
+	io->offset = off;
+	ethdev->drv_ctl(dev->netdev, &info);
+	return io->data;
+}
+
+static int cnic_in_use(struct cnic_sock *csk)
+{
+	return test_bit(SK_F_INUSE, &csk->flags);
+}
+
+static void cnic_kwq_completion(struct cnic_dev *dev, u32 count)
+{
+	struct cnic_local *cp = dev->cnic_priv;
+	struct cnic_eth_dev *ethdev = cp->ethdev;
+	struct drv_ctl_info info;
+
+	info.cmd = DRV_CTL_COMPLETION_CMD;
+	info.data.comp.comp_count = count;
+	ethdev->drv_ctl(dev->netdev, &info);
+}
+
+static int cnic_send_nlmsg(struct cnic_local *cp, u32 type,
+			   struct cnic_sock *csk)
+{
+	struct iscsi_path path_req;
+	char *buf = NULL;
+	u16 len = 0;
+	u32 msg_type = ISCSI_KEVENT_IF_DOWN;
+	struct cnic_ulp_ops *ulp_ops;
+
+	if (cp->uio_dev == -1)
+		return -ENODEV;
+
+	if (csk) {
+		len = sizeof(path_req);
+		buf = (char *) &path_req;
+		memset(&path_req, 0, len);
+
+		msg_type = ISCSI_KEVENT_PATH_REQ;
+		path_req.handle = (u64) csk->l5_cid;
+		if (test_bit(SK_F_IPV6, &csk->flags)) {
+			memcpy(&path_req.dst.v6_addr, &csk->dst_ip[0],
+			       sizeof(struct in6_addr));
+			path_req.ip_addr_len = 16;
+		} else {
+			memcpy(&path_req.dst.v4_addr, &csk->dst_ip[0],
+			       sizeof(struct in_addr));
+			path_req.ip_addr_len = 4;
+		}
+		path_req.vlan_id = csk->vlan_id;
+		path_req.pmtu = csk->mtu;
+	}
+
+	rcu_read_lock();
+	ulp_ops = rcu_dereference(cp->ulp_ops[CNIC_ULP_ISCSI]);
+	if (ulp_ops)
+		ulp_ops->iscsi_nl_send_msg(cp->dev, msg_type, buf, len);
+	rcu_read_unlock();
+	return 0;
+}
+
+static int cnic_iscsi_nl_msg_recv(struct cnic_dev *dev, u32 msg_type,
+				  char *buf, u16 len)
+{
+	int rc = -EINVAL;
+
+	switch (msg_type) {
+	case ISCSI_UEVENT_PATH_UPDATE: {
+		struct cnic_local *cp;
+		u32 l5_cid;
+		struct cnic_sock *csk;
+		struct iscsi_path *path_resp;
+
+		if (len < sizeof(*path_resp))
+			break;
+
+		path_resp = (struct iscsi_path *) buf;
+		cp = dev->cnic_priv;
+		l5_cid = (u32) path_resp->handle;
+		if (l5_cid >= MAX_CM_SK_TBL_SZ)
+			break;
+
+		csk = &cp->csk_tbl[l5_cid];
+		csk_hold(csk);
+		if (cnic_in_use(csk)) {
+			memcpy(csk->ha, path_resp->mac_addr, 6);
+			if (test_bit(SK_F_IPV6, &csk->flags))
+				memcpy(&csk->src_ip[0], &path_resp->src.v6_addr,
+				       sizeof(struct in6_addr));
+			else
+				memcpy(&csk->src_ip[0], &path_resp->src.v4_addr,
+				       sizeof(struct in_addr));
+			if (is_valid_ether_addr(csk->ha))
+				cnic_cm_set_pg(csk);
+		}
+		csk_put(csk);
+		rc = 0;
+	}
+	}
+
+	return rc;
+}
+
+static int cnic_offld_prep(struct cnic_sock *csk)
+{
+	if (test_and_set_bit(SK_F_OFFLD_SCHED, &csk->flags))
+		return 0;
+
+	if (!test_bit(SK_F_CONNECT_START, &csk->flags)) {
+		clear_bit(SK_F_OFFLD_SCHED, &csk->flags);
+		return 0;
+	}
+
+	return 1;
+}
+
+static int cnic_close_prep(struct cnic_sock *csk)
+{
+	clear_bit(SK_F_CONNECT_START, &csk->flags);
+	smp_mb__after_clear_bit();
+
+	if (test_and_clear_bit(SK_F_OFFLD_COMPLETE, &csk->flags)) {
+		while (test_and_set_bit(SK_F_OFFLD_SCHED, &csk->flags))
+			msleep(1);
+
+		return 1;
+	}
+	return 0;
+}
+
+static int cnic_abort_prep(struct cnic_sock *csk)
+{
+	clear_bit(SK_F_CONNECT_START, &csk->flags);
+	smp_mb__after_clear_bit();
+
+	while (test_and_set_bit(SK_F_OFFLD_SCHED, &csk->flags))
+		msleep(1);
+
+	if (test_and_clear_bit(SK_F_OFFLD_COMPLETE, &csk->flags)) {
+		csk->state = L4_KCQE_OPCODE_VALUE_RESET_COMP;
+		return 1;
+	}
+
+	return 0;
+}
+
+int cnic_register_driver(int ulp_type, struct cnic_ulp_ops *ulp_ops)
+{
+	struct cnic_dev *dev;
+
+	if (ulp_type >= MAX_CNIC_ULP_TYPE) {
+		printk(KERN_ERR PFX "cnic_register_driver: Bad type %d\n",
+		       ulp_type);
+		return -EINVAL;
+	}
+	mutex_lock(&cnic_lock);
+	if (cnic_ulp_tbl[ulp_type]) {
+		printk(KERN_ERR PFX "cnic_register_driver: Type %d has already "
+				    "been registered\n", ulp_type);
+		mutex_unlock(&cnic_lock);
+		return -EBUSY;
+	}
+
+	read_lock(&cnic_dev_lock);
+	list_for_each_entry(dev, &cnic_dev_list, list) {
+		struct cnic_local *cp = dev->cnic_priv;
+
+		clear_bit(ULP_F_INIT, &cp->ulp_flags[ulp_type]);
+	}
+	read_unlock(&cnic_dev_lock);
+
+	rcu_assign_pointer(cnic_ulp_tbl[ulp_type], ulp_ops);
+	mutex_unlock(&cnic_lock);
+
+	/* Prevent race conditions with netdev_event */
+	rtnl_lock();
+	read_lock(&cnic_dev_lock);
+	list_for_each_entry(dev, &cnic_dev_list, list) {
+		struct cnic_local *cp = dev->cnic_priv;
+
+		if (!test_and_set_bit(ULP_F_INIT, &cp->ulp_flags[ulp_type]))
+			ulp_ops->cnic_init(dev);
+	}
+	read_unlock(&cnic_dev_lock);
+	rtnl_unlock();
+
+	return 0;
+}
+
+int cnic_unregister_driver(int ulp_type)
+{
+	struct cnic_dev *dev;
+
+	if (ulp_type >= MAX_CNIC_ULP_TYPE) {
+		printk(KERN_ERR PFX "cnic_unregister_driver: Bad type %d\n",
+		       ulp_type);
+		return -EINVAL;
+	}
+	mutex_lock(&cnic_lock);
+	if (!cnic_ulp_tbl[ulp_type]) {
+		printk(KERN_ERR PFX "cnic_unregister_driver: Type %d has not "
+				    "been registered\n", ulp_type);
+		goto out_unlock;
+	}
+	read_lock(&cnic_dev_lock);
+	list_for_each_entry(dev, &cnic_dev_list, list) {
+		struct cnic_local *cp = dev->cnic_priv;
+
+		if (rcu_dereference(cp->ulp_ops[ulp_type])) {
+			printk(KERN_ERR PFX "cnic_unregister_driver: Type %d "
+			       "still has devices registered\n", ulp_type);
+			read_unlock(&cnic_dev_lock);
+			goto out_unlock;
+		}
+	}
+	read_unlock(&cnic_dev_lock);
+
+	rcu_assign_pointer(cnic_ulp_tbl[ulp_type], NULL);
+
+	mutex_unlock(&cnic_lock);
+	synchronize_rcu();
+	return 0;
+
+out_unlock:
+	mutex_unlock(&cnic_lock);
+	return -EINVAL;
+}
+
+static int cnic_start_hw(struct cnic_dev *);
+static void cnic_stop_hw(struct cnic_dev *);
+
+static int cnic_register_device(struct cnic_dev *dev, int ulp_type,
+				void *ulp_ctx)
+{
+	struct cnic_local *cp = dev->cnic_priv;
+	struct cnic_ulp_ops *ulp_ops;
+
+	if (ulp_type >= MAX_CNIC_ULP_TYPE) {
+		printk(KERN_ERR PFX "cnic_register_device: Bad type %d\n",
+		       ulp_type);
+		return -EINVAL;
+	}
+	mutex_lock(&cnic_lock);
+	if (cnic_ulp_tbl[ulp_type] == NULL) {
+		printk(KERN_ERR PFX "cnic_register_device: Driver with type %d "
+				    "has not been registered\n", ulp_type);
+		mutex_unlock(&cnic_lock);
+		return -EAGAIN;
+	}
+	if (rcu_dereference(cp->ulp_ops[ulp_type])) {
+		printk(KERN_ERR PFX "cnic_register_device: Type %d has already "
+		       "been registered to this device\n", ulp_type);
+		mutex_unlock(&cnic_lock);
+		return -EBUSY;
+	}
+
+	clear_bit(ULP_F_START, &cp->ulp_flags[ulp_type]);
+	cp->ulp_handle[ulp_type] = ulp_ctx;
+	ulp_ops = cnic_ulp_tbl[ulp_type];
+	rcu_assign_pointer(cp->ulp_ops[ulp_type], ulp_ops);
+	cnic_hold(dev);
+
+	if (test_bit(CNIC_F_CNIC_UP, &dev->flags))
+		if (!test_and_set_bit(ULP_F_START, &cp->ulp_flags[ulp_type]))
+			ulp_ops->cnic_start(cp->ulp_handle[ulp_type]);
+
+	mutex_unlock(&cnic_lock);
+
+	return 0;
+
+}
+EXPORT_SYMBOL(cnic_register_driver);
+
+static int cnic_unregister_device(struct cnic_dev *dev, int ulp_type)
+{
+	struct cnic_local *cp = dev->cnic_priv;
+
+	if (ulp_type >= MAX_CNIC_ULP_TYPE) {
+		printk(KERN_ERR PFX "cnic_unregister_device: Bad type %d\n",
+		       ulp_type);
+		return -EINVAL;
+	}
+	mutex_lock(&cnic_lock);
+	if (rcu_dereference(cp->ulp_ops[ulp_type])) {
+		rcu_assign_pointer(cp->ulp_ops[ulp_type], NULL);
+		cnic_put(dev);
+	} else {
+		printk(KERN_ERR PFX "cnic_unregister_device: device not "
+		       "registered to this ulp type %d\n", ulp_type);
+		mutex_unlock(&cnic_lock);
+		return -EINVAL;
+	}
+	mutex_unlock(&cnic_lock);
+
+	synchronize_rcu();
+
+	return 0;
+}
+EXPORT_SYMBOL(cnic_unregister_driver);
+
+static int cnic_init_id_tbl(struct cnic_id_tbl *id_tbl, u32 size, u32 start_id)
+{
+	id_tbl->start = start_id;
+	id_tbl->max = size;
+	id_tbl->next = 0;
+	spin_lock_init(&id_tbl->lock);
+	id_tbl->table = kzalloc(DIV_ROUND_UP(size, 32) * 4, GFP_KERNEL);
+	if (!id_tbl->table)
+		return -ENOMEM;
+
+	return 0;
+}
+
+static void cnic_free_id_tbl(struct cnic_id_tbl *id_tbl)
+{
+	kfree(id_tbl->table);
+	id_tbl->table = NULL;
+}
+
+static int cnic_alloc_id(struct cnic_id_tbl *id_tbl, u32 id)
+{
+	int ret = -1;
+
+	id -= id_tbl->start;
+	if (id >= id_tbl->max)
+		return ret;
+
+	spin_lock(&id_tbl->lock);
+	if (!test_bit(id, id_tbl->table)) {
+		set_bit(id, id_tbl->table);
+		ret = 0;
+	}
+	spin_unlock(&id_tbl->lock);
+	return ret;
+}
+
+/* Returns -1 if not successful */
+static u32 cnic_alloc_new_id(struct cnic_id_tbl *id_tbl)
+{
+	u32 id;
+
+	spin_lock(&id_tbl->lock);
+	id = find_next_zero_bit(id_tbl->table, id_tbl->max, id_tbl->next);
+	if (id >= id_tbl->max) {
+		id = -1;
+		if (id_tbl->next != 0) {
+			id = find_first_zero_bit(id_tbl->table, id_tbl->next);
+			if (id >= id_tbl->next)
+				id = -1;
+		}
+	}
+
+	if (id < id_tbl->max) {
+		set_bit(id, id_tbl->table);
+		id_tbl->next = (id + 1) & (id_tbl->max - 1);
+		id += id_tbl->start;
+	}
+
+	spin_unlock(&id_tbl->lock);
+
+	return id;
+}
+
+static void cnic_free_id(struct cnic_id_tbl *id_tbl, u32 id)
+{
+	if (id == -1)
+		return;
+
+	id -= id_tbl->start;
+	if (id >= id_tbl->max)
+		return;
+
+	clear_bit(id, id_tbl->table);
+}
+
+static void cnic_free_dma(struct cnic_dev *dev, struct cnic_dma *dma)
+{
+	int i;
+
+	if (!dma->pg_arr)
+		return;
+
+	for (i = 0; i < dma->num_pages; i++) {
+		if (dma->pg_arr[i]) {
+			pci_free_consistent(dev->pcidev, BCM_PAGE_SIZE,
+					    dma->pg_arr[i], dma->pg_map_arr[i]);
+			dma->pg_arr[i] = NULL;
+		}
+	}
+	if (dma->pgtbl) {
+		pci_free_consistent(dev->pcidev, dma->pgtbl_size,
+				    dma->pgtbl, dma->pgtbl_map);
+		dma->pgtbl = NULL;
+	}
+	kfree(dma->pg_arr);
+	dma->pg_arr = NULL;
+	dma->num_pages = 0;
+}
+
+static void cnic_setup_page_tbl(struct cnic_dev *dev, struct cnic_dma *dma)
+{
+	int i;
+	u32 *page_table = dma->pgtbl;
+
+	for (i = 0; i < dma->num_pages; i++) {
+		/* Each entry needs to be in big endian format. */
+		*page_table = (u32) ((u64) dma->pg_map_arr[i] >> 32);
+		page_table++;
+		*page_table = (u32) dma->pg_map_arr[i];
+		page_table++;
+	}
+}
+
+static int cnic_alloc_dma(struct cnic_dev *dev, struct cnic_dma *dma,
+			  int pages, int use_pg_tbl)
+{
+	int i, size;
+	struct cnic_local *cp = dev->cnic_priv;
+
+	size = pages * (sizeof(void *) + sizeof(dma_addr_t));
+	dma->pg_arr = kzalloc(size, GFP_ATOMIC);
+	if (dma->pg_arr == NULL)
+		return -ENOMEM;
+
+	dma->pg_map_arr = (dma_addr_t *) (dma->pg_arr + pages);
+	dma->num_pages = pages;
+
+	for (i = 0; i < pages; i++) {
+		dma->pg_arr[i] = pci_alloc_consistent(dev->pcidev,
+						      BCM_PAGE_SIZE,
+						      &dma->pg_map_arr[i]);
+		if (dma->pg_arr[i] == NULL)
+			goto error;
+	}
+	if (!use_pg_tbl)
+		return 0;
+
+	dma->pgtbl_size = ((pages * 8) + BCM_PAGE_SIZE - 1) &
+			  ~(BCM_PAGE_SIZE - 1);
+	dma->pgtbl = pci_alloc_consistent(dev->pcidev, dma->pgtbl_size,
+					  &dma->pgtbl_map);
+	if (dma->pgtbl == NULL)
+		goto error;
+
+	cp->setup_pgtbl(dev, dma);
+
+	return 0;
+
+error:
+	cnic_free_dma(dev, dma);
+	return -ENOMEM;
+}
+
+static void cnic_free_resc(struct cnic_dev *dev)
+{
+	struct cnic_local *cp = dev->cnic_priv;
+	int i = 0;
+
+	if (cp->cnic_uinfo) {
+		cnic_send_nlmsg(cp, ISCSI_KEVENT_IF_DOWN, NULL);
+		while (cp->uio_dev != -1 && i < 15) {
+			msleep(100);
+			i++;
+		}
+		uio_unregister_device(cp->cnic_uinfo);
+		kfree(cp->cnic_uinfo);
+		cp->cnic_uinfo = NULL;
+	}
+
+	if (cp->l2_buf) {
+		pci_free_consistent(dev->pcidev, cp->l2_buf_size,
+				    cp->l2_buf, cp->l2_buf_map);
+		cp->l2_buf = NULL;
+	}
+
+	if (cp->l2_ring) {
+		pci_free_consistent(dev->pcidev, cp->l2_ring_size,
+				    cp->l2_ring, cp->l2_ring_map);
+		cp->l2_ring = NULL;
+	}
+
+	for (i = 0; i < cp->ctx_blks; i++) {
+		if (cp->ctx_arr[i].ctx) {
+			pci_free_consistent(dev->pcidev, cp->ctx_blk_size,
+					    cp->ctx_arr[i].ctx,
+					    cp->ctx_arr[i].mapping);
+			cp->ctx_arr[i].ctx = NULL;
+		}
+	}
+	kfree(cp->ctx_arr);
+	cp->ctx_arr = NULL;
+	cp->ctx_blks = 0;
+
+	cnic_free_dma(dev, &cp->gbl_buf_info);
+	cnic_free_dma(dev, &cp->conn_buf_info);
+	cnic_free_dma(dev, &cp->kwq_info);
+	cnic_free_dma(dev, &cp->kcq_info);
+	kfree(cp->iscsi_tbl);
+	cp->iscsi_tbl = NULL;
+	kfree(cp->ctx_tbl);
+	cp->ctx_tbl = NULL;
+
+	cnic_free_id_tbl(&cp->cid_tbl);
+}
+
+static int cnic_alloc_context(struct cnic_dev *dev)
+{
+	struct cnic_local *cp = dev->cnic_priv;
+
+	if (CHIP_NUM(cp) == CHIP_NUM_5709) {
+		int i, k, arr_size;
+
+		cp->ctx_blk_size = BCM_PAGE_SIZE;
+		cp->cids_per_blk = BCM_PAGE_SIZE / 128;
+		arr_size = BNX2_MAX_CID / cp->cids_per_blk *
+			   sizeof(struct cnic_ctx);
+		cp->ctx_arr = kzalloc(arr_size, GFP_KERNEL);
+		if (cp->ctx_arr == NULL)
+			return -ENOMEM;
+
+		k = 0;
+		for (i = 0; i < 2; i++) {
+			u32 j, reg, off, lo, hi;
+
+			if (i == 0)
+				off = BNX2_PG_CTX_MAP;
+			else
+				off = BNX2_ISCSI_CTX_MAP;
+
+			reg = cnic_reg_rd_ind(dev, off);
+			lo = reg >> 16;
+			hi = reg & 0xffff;
+			for (j = lo; j < hi; j += cp->cids_per_blk, k++)
+				cp->ctx_arr[k].cid = j;
+		}
+
+		cp->ctx_blks = k;
+		if (cp->ctx_blks >= (BNX2_MAX_CID / cp->cids_per_blk)) {
+			cp->ctx_blks = 0;
+			return -ENOMEM;
+		}
+
+		for (i = 0; i < cp->ctx_blks; i++) {
+			cp->ctx_arr[i].ctx =
+				pci_alloc_consistent(dev->pcidev, BCM_PAGE_SIZE,
+						     &cp->ctx_arr[i].mapping);
+			if (cp->ctx_arr[i].ctx == NULL)
+				return -ENOMEM;
+		}
+	}
+	return 0;
+}
+
+static int cnic_alloc_bnx2_resc(struct cnic_dev *dev)
+{
+	struct cnic_local *cp = dev->cnic_priv;
+	struct uio_info *uinfo;
+	int ret;
+
+	ret = cnic_alloc_dma(dev, &cp->kwq_info, KWQ_PAGE_CNT, 1);
+	if (ret)
+		goto error;
+	cp->kwq = (struct kwqe **) cp->kwq_info.pg_arr;
+
+	ret = cnic_alloc_dma(dev, &cp->kcq_info, KCQ_PAGE_CNT, 1);
+	if (ret)
+		goto error;
+	cp->kcq = (struct kcqe **) cp->kcq_info.pg_arr;
+
+	ret = cnic_alloc_context(dev);
+	if (ret)
+		goto error;
+
+	cp->l2_ring_size = 2 * BCM_PAGE_SIZE;
+	cp->l2_ring = pci_alloc_consistent(dev->pcidev, cp->l2_ring_size,
+					   &cp->l2_ring_map);
+	if (!cp->l2_ring)
+		goto error;
+
+	cp->l2_buf_size = (cp->l2_rx_ring_size + 1) * cp->l2_single_buf_size;
+	cp->l2_buf_size = PAGE_ALIGN(cp->l2_buf_size);
+	cp->l2_buf = pci_alloc_consistent(dev->pcidev, cp->l2_buf_size,
+					   &cp->l2_buf_map);
+	if (!cp->l2_buf)
+		goto error;
+
+	uinfo = kzalloc(sizeof(*uinfo), GFP_ATOMIC);
+	if (!uinfo)
+		goto error;
+
+	uinfo->mem[0].addr = dev->netdev->base_addr;
+	uinfo->mem[0].internal_addr = dev->regview;
+	uinfo->mem[0].size = dev->netdev->mem_end - dev->netdev->mem_start;
+	uinfo->mem[0].memtype = UIO_MEM_PHYS;
+
+	uinfo->mem[1].addr = (unsigned long) cp->status_blk & PAGE_MASK;
+	if (cp->ethdev->drv_state & CNIC_DRV_STATE_USING_MSIX)
+		uinfo->mem[1].size = BNX2_SBLK_MSIX_ALIGN_SIZE * 9;
+	else
+		uinfo->mem[1].size = BNX2_SBLK_MSIX_ALIGN_SIZE;
+	uinfo->mem[1].memtype = UIO_MEM_LOGICAL;
+
+	uinfo->mem[2].addr = (unsigned long) cp->l2_ring;
+	uinfo->mem[2].size = cp->l2_ring_size;
+	uinfo->mem[2].memtype = UIO_MEM_LOGICAL;
+
+	uinfo->mem[3].addr = (unsigned long) cp->l2_buf;
+	uinfo->mem[3].size = cp->l2_buf_size;
+	uinfo->mem[3].memtype = UIO_MEM_LOGICAL;
+
+	uinfo->name = "bnx2_cnic";
+	uinfo->version = CNIC_MODULE_VERSION;
+	uinfo->irq = UIO_IRQ_CUSTOM;
+
+	uinfo->open = cnic_uio_open;
+	uinfo->release = cnic_uio_close;
+
+	uinfo->priv = dev;
+
+	ret = uio_register_device(&dev->pcidev->dev, uinfo);
+	if (ret) {
+		kfree(uinfo);
+		goto error;
+	}
+
+	cp->cnic_uinfo = uinfo;
+
+	return 0;
+
+error:
+	cnic_free_resc(dev);
+	return ret;
+}
+
+static inline u32 cnic_kwq_avail(struct cnic_local *cp)
+{
+	return cp->max_kwq_idx -
+		((cp->kwq_prod_idx - cp->kwq_con_idx) & cp->max_kwq_idx);
+}
+
+static int cnic_submit_bnx2_kwqes(struct cnic_dev *dev, struct kwqe *wqes[],
+				  u32 num_wqes)
+{
+	struct cnic_local *cp = dev->cnic_priv;
+	struct kwqe *prod_qe;
+	u16 prod, sw_prod, i;
+
+	if (!test_bit(CNIC_F_CNIC_UP, &dev->flags))
+		return -EAGAIN;		/* bnx2 is down */
+
+	spin_lock_bh(&cp->cnic_ulp_lock);
+	if (num_wqes > cnic_kwq_avail(cp) &&
+	    !(cp->cnic_local_flags & CNIC_LCL_FL_KWQ_INIT)) {
+		spin_unlock_bh(&cp->cnic_ulp_lock);
+		return -EAGAIN;
+	}
+
+	cp->cnic_local_flags &= ~CNIC_LCL_FL_KWQ_INIT;
+
+	prod = cp->kwq_prod_idx;
+	sw_prod = prod & MAX_KWQ_IDX;
+	for (i = 0; i < num_wqes; i++) {
+		prod_qe = &cp->kwq[KWQ_PG(sw_prod)][KWQ_IDX(sw_prod)];
+		memcpy(prod_qe, wqes[i], sizeof(struct kwqe));
+		prod++;
+		sw_prod = prod & MAX_KWQ_IDX;
+	}
+	cp->kwq_prod_idx = prod;
+
+	CNIC_WR16(dev, cp->kwq_io_addr, cp->kwq_prod_idx);
+
+	spin_unlock_bh(&cp->cnic_ulp_lock);
+	return 0;
+}
+
+static void service_kcqes(struct cnic_dev *dev, int num_cqes)
+{
+	struct cnic_local *cp = dev->cnic_priv;
+	int i, j;
+
+	i = 0;
+	j = 1;
+	while (num_cqes) {
+		struct cnic_ulp_ops *ulp_ops;
+		int ulp_type;
+		u32 kcqe_op_flag = cp->completed_kcq[i]->kcqe_op_flag;
+		u32 kcqe_layer = kcqe_op_flag & KCQE_FLAGS_LAYER_MASK;
+
+		if (unlikely(kcqe_op_flag & KCQE_RAMROD_COMPLETION))
+			cnic_kwq_completion(dev, 1);
+
+		while (j < num_cqes) {
+			u32 next_op = cp->completed_kcq[i + j]->kcqe_op_flag;
+
+			if ((next_op & KCQE_FLAGS_LAYER_MASK) != kcqe_layer)
+				break;
+
+			if (unlikely(next_op & KCQE_RAMROD_COMPLETION))
+				cnic_kwq_completion(dev, 1);
+			j++;
+		}
+
+		if (kcqe_layer == KCQE_FLAGS_LAYER_MASK_L5_RDMA)
+			ulp_type = CNIC_ULP_RDMA;
+		else if (kcqe_layer == KCQE_FLAGS_LAYER_MASK_L5_ISCSI)
+			ulp_type = CNIC_ULP_ISCSI;
+		else if (kcqe_layer == KCQE_FLAGS_LAYER_MASK_L4)
+			ulp_type = CNIC_ULP_L4;
+		else if (kcqe_layer == KCQE_FLAGS_LAYER_MASK_L2)
+			goto end;
+		else {
+			printk(KERN_ERR PFX "%s: Unknown type of KCQE(0x%x)\n",
+			       dev->netdev->name, kcqe_op_flag);
+			goto end;
+		}
+
+		rcu_read_lock();
+		ulp_ops = rcu_dereference(cp->ulp_ops[ulp_type]);
+		if (likely(ulp_ops)) {
+			ulp_ops->indicate_kcqes(cp->ulp_handle[ulp_type],
+						  cp->completed_kcq + i, j);
+		}
+		rcu_read_unlock();
+end:
+		num_cqes -= j;
+		i += j;
+		j = 1;
+	}
+	return;
+}
+
+static u16 cnic_bnx2_next_idx(u16 idx)
+{
+	return idx + 1;
+}
+
+static u16 cnic_bnx2_hw_idx(u16 idx)
+{
+	return idx;
+}
+
+static int cnic_get_kcqes(struct cnic_dev *dev, u16 hw_prod, u16 *sw_prod)
+{
+	struct cnic_local *cp = dev->cnic_priv;
+	u16 i, ri, last;
+	struct kcqe *kcqe;
+	int kcqe_cnt = 0, last_cnt = 0;
+
+	i = ri = last = *sw_prod;
+	ri &= MAX_KCQ_IDX;
+
+	while ((i != hw_prod) && (kcqe_cnt < MAX_COMPLETED_KCQE)) {
+		kcqe = &cp->kcq[KCQ_PG(ri)][KCQ_IDX(ri)];
+		cp->completed_kcq[kcqe_cnt++] = kcqe;
+		i = cp->next_idx(i);
+		ri = i & MAX_KCQ_IDX;
+		if (likely(!(kcqe->kcqe_op_flag & KCQE_FLAGS_NEXT))) {
+			last_cnt = kcqe_cnt;
+			last = i;
+		}
+	}
+
+	*sw_prod = last;
+	return last_cnt;
+}
+
+static void cnic_chk_bnx2_pkt_rings(struct cnic_local *cp)
+{
+	u16 rx_cons = *cp->rx_cons_ptr;
+	u16 tx_cons = *cp->tx_cons_ptr;
+
+	if (cp->tx_cons != tx_cons || cp->rx_cons != rx_cons) {
+		cp->tx_cons = tx_cons;
+		cp->rx_cons = rx_cons;
+		uio_event_notify(cp->cnic_uinfo);
+	}
+}
+
+static int cnic_service_bnx2(void *data, void *status_blk)
+{
+	struct cnic_dev *dev = data;
+	struct status_block *sblk = status_blk;
+	struct cnic_local *cp = dev->cnic_priv;
+	u32 status_idx = sblk->status_idx;
+	u16 hw_prod, sw_prod;
+	int kcqe_cnt;
+
+	if (unlikely(!test_bit(CNIC_F_CNIC_UP, &dev->flags)))
+		return status_idx;
+
+	cp->kwq_con_idx = *cp->kwq_con_idx_ptr;
+
+	hw_prod = sblk->status_completion_producer_index;
+	sw_prod = cp->kcq_prod_idx;
+	while (sw_prod != hw_prod) {
+		kcqe_cnt = cnic_get_kcqes(dev, hw_prod, &sw_prod);
+		if (kcqe_cnt == 0)
+			goto done;
+
+		service_kcqes(dev, kcqe_cnt);
+
+		/* Tell compiler that status_blk fields can change. */
+		barrier();
+		if (status_idx != sblk->status_idx) {
+			status_idx = sblk->status_idx;
+			cp->kwq_con_idx = *cp->kwq_con_idx_ptr;
+			hw_prod = sblk->status_completion_producer_index;
+		} else
+			break;
+	}
+
+done:
+	CNIC_WR16(dev, cp->kcq_io_addr, sw_prod);
+
+	cp->kcq_prod_idx = sw_prod;
+
+	cnic_chk_bnx2_pkt_rings(cp);
+	return status_idx;
+}
+
+static void cnic_service_bnx2_msix(unsigned long data)
+{
+	struct cnic_dev *dev = (struct cnic_dev *) data;
+	struct cnic_local *cp = dev->cnic_priv;
+	struct status_block_msix *status_blk = cp->bnx2_status_blk;
+	u32 status_idx = status_blk->status_idx;
+	u16 hw_prod, sw_prod;
+	int kcqe_cnt;
+
+	cp->kwq_con_idx = status_blk->status_cmd_consumer_index;
+
+	hw_prod = status_blk->status_completion_producer_index;
+	sw_prod = cp->kcq_prod_idx;
+	while (sw_prod != hw_prod) {
+		kcqe_cnt = cnic_get_kcqes(dev, hw_prod, &sw_prod);
+		if (kcqe_cnt == 0)
+			goto done;
+
+		service_kcqes(dev, kcqe_cnt);
+
+		/* Tell compiler that status_blk fields can change. */
+		barrier();
+		if (status_idx != status_blk->status_idx) {
+			status_idx = status_blk->status_idx;
+			cp->kwq_con_idx = status_blk->status_cmd_consumer_index;
+			hw_prod = status_blk->status_completion_producer_index;
+		} else
+			break;
+	}
+
+done:
+	CNIC_WR16(dev, cp->kcq_io_addr, sw_prod);
+	cp->kcq_prod_idx = sw_prod;
+
+	cnic_chk_bnx2_pkt_rings(cp);
+
+	cp->last_status_idx = status_idx;
+	CNIC_WR(dev, BNX2_PCICFG_INT_ACK_CMD, cp->int_num |
+		BNX2_PCICFG_INT_ACK_CMD_INDEX_VALID | cp->last_status_idx);
+}
+
+static irqreturn_t cnic_irq(int irq, void *dev_instance)
+{
+	struct cnic_dev *dev = dev_instance;
+	struct cnic_local *cp = dev->cnic_priv;
+	u16 prod = cp->kcq_prod_idx & MAX_KCQ_IDX;
+
+	if (cp->ack_int)
+		cp->ack_int(dev);
+
+	prefetch(cp->status_blk);
+	prefetch(&cp->kcq[KCQ_PG(prod)][KCQ_IDX(prod)]);
+
+	if (likely(test_bit(CNIC_F_CNIC_UP, &dev->flags)))
+		tasklet_schedule(&cp->cnic_irq_task);
+
+	return IRQ_HANDLED;
+}
+
+static void cnic_ulp_stop(struct cnic_dev *dev)
+{
+	struct cnic_local *cp = dev->cnic_priv;
+	int if_type;
+
+	rcu_read_lock();
+	for (if_type = 0; if_type < MAX_CNIC_ULP_TYPE; if_type++) {
+		struct cnic_ulp_ops *ulp_ops;
+
+		ulp_ops = rcu_dereference(cp->ulp_ops[if_type]);
+		if (!ulp_ops)
+			continue;
+
+		if (test_and_clear_bit(ULP_F_START, &cp->ulp_flags[if_type]))
+			ulp_ops->cnic_stop(cp->ulp_handle[if_type]);
+	}
+	rcu_read_unlock();
+}
+
+static void cnic_ulp_start(struct cnic_dev *dev)
+{
+	struct cnic_local *cp = dev->cnic_priv;
+	int if_type;
+
+	rcu_read_lock();
+	for (if_type = 0; if_type < MAX_CNIC_ULP_TYPE; if_type++) {
+		struct cnic_ulp_ops *ulp_ops;
+
+		ulp_ops = rcu_dereference(cp->ulp_ops[if_type]);
+		if (!ulp_ops || !ulp_ops->cnic_start)
+			continue;
+
+		if (!test_and_set_bit(ULP_F_START, &cp->ulp_flags[if_type]))
+			ulp_ops->cnic_start(cp->ulp_handle[if_type]);
+	}
+	rcu_read_unlock();
+}
+
+static int cnic_ctl(void *data, struct cnic_ctl_info *info)
+{
+	struct cnic_dev *dev = data;
+
+	switch (info->cmd) {
+	case CNIC_CTL_STOP_CMD:
+		cnic_hold(dev);
+		mutex_lock(&cnic_lock);
+
+		cnic_ulp_stop(dev);
+		cnic_stop_hw(dev);
+
+		mutex_unlock(&cnic_lock);
+		cnic_put(dev);
+		break;
+	case CNIC_CTL_START_CMD:
+		cnic_hold(dev);
+		mutex_lock(&cnic_lock);
+
+		if (!cnic_start_hw(dev))
+			cnic_ulp_start(dev);
+
+		mutex_unlock(&cnic_lock);
+		cnic_put(dev);
+		break;
+	default:
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static void cnic_ulp_init(struct cnic_dev *dev)
+{
+	int i;
+	struct cnic_local *cp = dev->cnic_priv;
+
+	rcu_read_lock();
+	for (i = 0; i < MAX_CNIC_ULP_TYPE_EXT; i++) {
+		struct cnic_ulp_ops *ulp_ops;
+
+		ulp_ops = rcu_dereference(cnic_ulp_tbl[i]);
+		if (!ulp_ops || !ulp_ops->cnic_init)
+			continue;
+
+		if (!test_and_set_bit(ULP_F_INIT, &cp->ulp_flags[i]))
+			ulp_ops->cnic_init(dev);
+
+	}
+	rcu_read_unlock();